{ "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": 258.5494079589844, "learning_rate": 1.5803336259877089e-07, "loss": 0.7026621341705322, "step": 10 }, { "epoch": 0.0005268304889974745, "grad_norm": 24.36787223815918, "learning_rate": 3.336259877085163e-07, "loss": 0.3358674764633179, "step": 20 }, { "epoch": 0.0007902457334962118, "grad_norm": 15.10773754119873, "learning_rate": 5.092186128182617e-07, "loss": 0.2469015121459961, "step": 30 }, { "epoch": 0.001053660977994949, "grad_norm": 10.90369987487793, "learning_rate": 6.848112379280071e-07, "loss": 0.22429945468902587, "step": 40 }, { "epoch": 0.0013170762224936862, "grad_norm": 17.52402114868164, "learning_rate": 8.604038630377525e-07, "loss": 0.2419422149658203, "step": 50 }, { "epoch": 0.0015804914669924235, "grad_norm": 8.74491024017334, "learning_rate": 1.035996488147498e-06, "loss": 0.22556061744689943, "step": 60 }, { "epoch": 0.0018439067114911607, "grad_norm": 12.159110069274902, "learning_rate": 1.2115891132572432e-06, "loss": 0.2161879539489746, "step": 70 }, { "epoch": 0.002107321955989898, "grad_norm": 6.291670322418213, "learning_rate": 1.3871817383669885e-06, "loss": 0.21901860237121581, "step": 80 }, { "epoch": 0.002370737200488635, "grad_norm": 6.634461879730225, "learning_rate": 1.5627743634767342e-06, "loss": 0.20751218795776366, "step": 90 }, { "epoch": 0.0026341524449873723, "grad_norm": 11.333536148071289, "learning_rate": 1.7383669885864795e-06, "loss": 0.2040024757385254, "step": 100 }, { "epoch": 0.00289756768948611, "grad_norm": 7.536369323730469, "learning_rate": 1.913959613696225e-06, "loss": 0.20576441287994385, "step": 110 }, { "epoch": 0.003160982933984847, "grad_norm": 10.027281761169434, "learning_rate": 2.08955223880597e-06, "loss": 0.2232499122619629, "step": 120 }, { "epoch": 0.003424398178483584, "grad_norm": 7.196838855743408, "learning_rate": 2.265144863915716e-06, "loss": 0.1924229860305786, "step": 130 }, { "epoch": 0.0036878134229823213, "grad_norm": 7.655050754547119, "learning_rate": 2.440737489025461e-06, "loss": 0.20641629695892333, "step": 140 }, { "epoch": 0.0039512286674810585, "grad_norm": 4.6251325607299805, "learning_rate": 2.6163301141352064e-06, "loss": 0.20491809844970704, "step": 150 }, { "epoch": 0.004214643911979796, "grad_norm": 7.507219314575195, "learning_rate": 2.791922739244952e-06, "loss": 0.2211151123046875, "step": 160 }, { "epoch": 0.004478059156478534, "grad_norm": 6.760087966918945, "learning_rate": 2.9675153643546974e-06, "loss": 0.21572668552398683, "step": 170 }, { "epoch": 0.00474147440097727, "grad_norm": 4.089375972747803, "learning_rate": 3.1431079894644427e-06, "loss": 0.21705842018127441, "step": 180 }, { "epoch": 0.005004889645476008, "grad_norm": 5.2463250160217285, "learning_rate": 3.318700614574188e-06, "loss": 0.20389366149902344, "step": 190 }, { "epoch": 0.005268304889974745, "grad_norm": 8.303729057312012, "learning_rate": 3.4942932396839337e-06, "loss": 0.21362195014953614, "step": 200 }, { "epoch": 0.005531720134473482, "grad_norm": 11.161720275878906, "learning_rate": 3.669885864793679e-06, "loss": 0.208734393119812, "step": 210 }, { "epoch": 0.00579513537897222, "grad_norm": 15.054389953613281, "learning_rate": 3.845478489903424e-06, "loss": 0.19894249439239503, "step": 220 }, { "epoch": 0.0060585506234709565, "grad_norm": 11.53581428527832, "learning_rate": 4.0210711150131695e-06, "loss": 0.21477060317993163, "step": 230 }, { "epoch": 0.006321965867969694, "grad_norm": 10.09111499786377, "learning_rate": 4.196663740122916e-06, "loss": 0.2184680223464966, "step": 240 }, { "epoch": 0.006585381112468432, "grad_norm": 5.9035868644714355, "learning_rate": 4.37225636523266e-06, "loss": 0.1760698676109314, "step": 250 }, { "epoch": 0.006848796356967168, "grad_norm": 4.83136510848999, "learning_rate": 4.547848990342406e-06, "loss": 0.20621585845947266, "step": 260 }, { "epoch": 0.007112211601465906, "grad_norm": 5.634449481964111, "learning_rate": 4.7234416154521515e-06, "loss": 0.18691493272781373, "step": 270 }, { "epoch": 0.007375626845964643, "grad_norm": 6.650679588317871, "learning_rate": 4.899034240561897e-06, "loss": 0.1941311240196228, "step": 280 }, { "epoch": 0.00763904209046338, "grad_norm": 5.619701385498047, "learning_rate": 5.074626865671642e-06, "loss": 0.19179294109344483, "step": 290 }, { "epoch": 0.007902457334962117, "grad_norm": 3.892063617706299, "learning_rate": 5.250219490781387e-06, "loss": 0.1930668592453003, "step": 300 }, { "epoch": 0.008165872579460855, "grad_norm": 4.611456871032715, "learning_rate": 5.425812115891133e-06, "loss": 0.18274096250534058, "step": 310 }, { "epoch": 0.008429287823959592, "grad_norm": 7.958328723907471, "learning_rate": 5.601404741000879e-06, "loss": 0.1879866123199463, "step": 320 }, { "epoch": 0.008692703068458329, "grad_norm": 3.1397433280944824, "learning_rate": 5.776997366110624e-06, "loss": 0.20520176887512206, "step": 330 }, { "epoch": 0.008956118312957067, "grad_norm": 7.029052257537842, "learning_rate": 5.9525899912203685e-06, "loss": 0.18854966163635253, "step": 340 }, { "epoch": 0.009219533557455804, "grad_norm": 2.6229405403137207, "learning_rate": 6.128182616330115e-06, "loss": 0.21463043689727784, "step": 350 }, { "epoch": 0.00948294880195454, "grad_norm": 3.5317466259002686, "learning_rate": 6.30377524143986e-06, "loss": 0.19853221178054808, "step": 360 }, { "epoch": 0.00974636404645328, "grad_norm": 2.889594078063965, "learning_rate": 6.479367866549605e-06, "loss": 0.20160720348358155, "step": 370 }, { "epoch": 0.010009779290952016, "grad_norm": 4.1464972496032715, "learning_rate": 6.654960491659351e-06, "loss": 0.20273165702819823, "step": 380 }, { "epoch": 0.010273194535450753, "grad_norm": 9.08263874053955, "learning_rate": 6.830553116769097e-06, "loss": 0.1945396065711975, "step": 390 }, { "epoch": 0.01053660977994949, "grad_norm": 6.832367897033691, "learning_rate": 7.006145741878841e-06, "loss": 0.18575795888900756, "step": 400 }, { "epoch": 0.010800025024448228, "grad_norm": 4.737504005432129, "learning_rate": 7.181738366988587e-06, "loss": 0.1839202880859375, "step": 410 }, { "epoch": 0.011063440268946964, "grad_norm": 4.670442581176758, "learning_rate": 7.3573309920983325e-06, "loss": 0.18817282915115358, "step": 420 }, { "epoch": 0.011326855513445701, "grad_norm": 10.22765827178955, "learning_rate": 7.532923617208078e-06, "loss": 0.17572619915008544, "step": 430 }, { "epoch": 0.01159027075794444, "grad_norm": 3.1663076877593994, "learning_rate": 7.708516242317823e-06, "loss": 0.18496015071868896, "step": 440 }, { "epoch": 0.011853686002443176, "grad_norm": 6.171747207641602, "learning_rate": 7.88410886742757e-06, "loss": 0.19756603240966797, "step": 450 }, { "epoch": 0.012117101246941913, "grad_norm": 9.790752410888672, "learning_rate": 8.059701492537314e-06, "loss": 0.18902413845062255, "step": 460 }, { "epoch": 0.012380516491440651, "grad_norm": 3.3440237045288086, "learning_rate": 8.23529411764706e-06, "loss": 0.1827101230621338, "step": 470 }, { "epoch": 0.012643931735939388, "grad_norm": 4.69491720199585, "learning_rate": 8.410886742756806e-06, "loss": 0.1857290744781494, "step": 480 }, { "epoch": 0.012907346980438125, "grad_norm": 5.252322673797607, "learning_rate": 8.58647936786655e-06, "loss": 0.19154565334320067, "step": 490 }, { "epoch": 0.013170762224936863, "grad_norm": 5.606055736541748, "learning_rate": 8.762071992976295e-06, "loss": 0.2152474880218506, "step": 500 }, { "epoch": 0.0134341774694356, "grad_norm": 5.707082748413086, "learning_rate": 8.937664618086041e-06, "loss": 0.20666353702545165, "step": 510 }, { "epoch": 0.013697592713934337, "grad_norm": 4.102421760559082, "learning_rate": 9.113257243195787e-06, "loss": 0.19506688117980958, "step": 520 }, { "epoch": 0.013961007958433075, "grad_norm": 5.240752696990967, "learning_rate": 9.288849868305532e-06, "loss": 0.18570051193237305, "step": 530 }, { "epoch": 0.014224423202931812, "grad_norm": 5.712233543395996, "learning_rate": 9.464442493415278e-06, "loss": 0.18633685111999512, "step": 540 }, { "epoch": 0.014487838447430549, "grad_norm": 3.05686354637146, "learning_rate": 9.640035118525024e-06, "loss": 0.18120217323303223, "step": 550 }, { "epoch": 0.014751253691929285, "grad_norm": 3.590808868408203, "learning_rate": 9.815627743634768e-06, "loss": 0.17575221061706542, "step": 560 }, { "epoch": 0.015014668936428024, "grad_norm": 2.754572629928589, "learning_rate": 9.991220368744513e-06, "loss": 0.1911107301712036, "step": 570 }, { "epoch": 0.01527808418092676, "grad_norm": 4.6930389404296875, "learning_rate": 1.0166812993854259e-05, "loss": 0.19126155376434326, "step": 580 }, { "epoch": 0.015541499425425497, "grad_norm": 6.437513828277588, "learning_rate": 1.0342405618964005e-05, "loss": 0.182111394405365, "step": 590 }, { "epoch": 0.015804914669924234, "grad_norm": 7.598888397216797, "learning_rate": 1.0517998244073751e-05, "loss": 0.18775372505187987, "step": 600 }, { "epoch": 0.016068329914422974, "grad_norm": 3.659184217453003, "learning_rate": 1.0693590869183494e-05, "loss": 0.19088712930679322, "step": 610 }, { "epoch": 0.01633174515892171, "grad_norm": 4.501362323760986, "learning_rate": 1.086918349429324e-05, "loss": 0.17987891435623168, "step": 620 }, { "epoch": 0.016595160403420448, "grad_norm": 7.172569274902344, "learning_rate": 1.1044776119402986e-05, "loss": 0.20523407459259033, "step": 630 }, { "epoch": 0.016858575647919184, "grad_norm": 6.790720462799072, "learning_rate": 1.122036874451273e-05, "loss": 0.1815967321395874, "step": 640 }, { "epoch": 0.01712199089241792, "grad_norm": 4.399389743804932, "learning_rate": 1.1395961369622477e-05, "loss": 0.18036503791809083, "step": 650 }, { "epoch": 0.017385406136916658, "grad_norm": 3.314239263534546, "learning_rate": 1.1571553994732223e-05, "loss": 0.2014004707336426, "step": 660 }, { "epoch": 0.017648821381415394, "grad_norm": 3.6589560508728027, "learning_rate": 1.1747146619841967e-05, "loss": 0.19469548463821412, "step": 670 }, { "epoch": 0.017912236625914135, "grad_norm": 3.532658576965332, "learning_rate": 1.1922739244951713e-05, "loss": 0.18091588020324706, "step": 680 }, { "epoch": 0.01817565187041287, "grad_norm": 6.121266841888428, "learning_rate": 1.209833187006146e-05, "loss": 0.19926671981811522, "step": 690 }, { "epoch": 0.018439067114911608, "grad_norm": 3.1881725788116455, "learning_rate": 1.2273924495171204e-05, "loss": 0.1933217763900757, "step": 700 }, { "epoch": 0.018702482359410345, "grad_norm": 5.833216190338135, "learning_rate": 1.244951712028095e-05, "loss": 0.1806323766708374, "step": 710 }, { "epoch": 0.01896589760390908, "grad_norm": 4.54648494720459, "learning_rate": 1.2625109745390696e-05, "loss": 0.18420839309692383, "step": 720 }, { "epoch": 0.019229312848407818, "grad_norm": 2.4707279205322266, "learning_rate": 1.2800702370500439e-05, "loss": 0.1842769742012024, "step": 730 }, { "epoch": 0.01949272809290656, "grad_norm": 6.141325950622559, "learning_rate": 1.2976294995610185e-05, "loss": 0.19099496603012084, "step": 740 }, { "epoch": 0.019756143337405295, "grad_norm": 3.566481351852417, "learning_rate": 1.315188762071993e-05, "loss": 0.20596930980682374, "step": 750 }, { "epoch": 0.02001955858190403, "grad_norm": 4.855190753936768, "learning_rate": 1.3327480245829676e-05, "loss": 0.1901150107383728, "step": 760 }, { "epoch": 0.02028297382640277, "grad_norm": 6.5758795738220215, "learning_rate": 1.3503072870939422e-05, "loss": 0.20587270259857177, "step": 770 }, { "epoch": 0.020546389070901505, "grad_norm": 2.101954698562622, "learning_rate": 1.3678665496049166e-05, "loss": 0.19062024354934692, "step": 780 }, { "epoch": 0.020809804315400242, "grad_norm": 5.183767795562744, "learning_rate": 1.3854258121158912e-05, "loss": 0.18205790519714354, "step": 790 }, { "epoch": 0.02107321955989898, "grad_norm": 11.50570297241211, "learning_rate": 1.4029850746268658e-05, "loss": 0.20664286613464355, "step": 800 }, { "epoch": 0.02133663480439772, "grad_norm": 4.387182235717773, "learning_rate": 1.4205443371378403e-05, "loss": 0.18969212770462035, "step": 810 }, { "epoch": 0.021600050048896455, "grad_norm": 3.376422882080078, "learning_rate": 1.4381035996488149e-05, "loss": 0.1810009479522705, "step": 820 }, { "epoch": 0.021863465293395192, "grad_norm": 1.4434841871261597, "learning_rate": 1.4556628621597895e-05, "loss": 0.18399991989135742, "step": 830 }, { "epoch": 0.02212688053789393, "grad_norm": 3.72482967376709, "learning_rate": 1.4732221246707638e-05, "loss": 0.18574364185333253, "step": 840 }, { "epoch": 0.022390295782392666, "grad_norm": 6.217770099639893, "learning_rate": 1.4907813871817384e-05, "loss": 0.18929582834243774, "step": 850 }, { "epoch": 0.022653711026891402, "grad_norm": 3.7733728885650635, "learning_rate": 1.508340649692713e-05, "loss": 0.1805187225341797, "step": 860 }, { "epoch": 0.022917126271390142, "grad_norm": 6.72900390625, "learning_rate": 1.5258999122036876e-05, "loss": 0.19487558603286742, "step": 870 }, { "epoch": 0.02318054151588888, "grad_norm": 3.2060248851776123, "learning_rate": 1.5434591747146622e-05, "loss": 0.2049407958984375, "step": 880 }, { "epoch": 0.023443956760387616, "grad_norm": 4.229701042175293, "learning_rate": 1.561018437225637e-05, "loss": 0.18241866827011108, "step": 890 }, { "epoch": 0.023707372004886353, "grad_norm": 4.210204601287842, "learning_rate": 1.578577699736611e-05, "loss": 0.17710986137390136, "step": 900 }, { "epoch": 0.02397078724938509, "grad_norm": 2.7853219509124756, "learning_rate": 1.5961369622475857e-05, "loss": 0.1940379858016968, "step": 910 }, { "epoch": 0.024234202493883826, "grad_norm": 8.393997192382812, "learning_rate": 1.6136962247585604e-05, "loss": 0.19514433145523072, "step": 920 }, { "epoch": 0.024497617738382566, "grad_norm": 2.9138898849487305, "learning_rate": 1.6312554872695346e-05, "loss": 0.19327244758605958, "step": 930 }, { "epoch": 0.024761032982881303, "grad_norm": 2.3137898445129395, "learning_rate": 1.6488147497805093e-05, "loss": 0.18281105756759644, "step": 940 }, { "epoch": 0.02502444822738004, "grad_norm": 3.003206968307495, "learning_rate": 1.666374012291484e-05, "loss": 0.17923564910888673, "step": 950 }, { "epoch": 0.025287863471878776, "grad_norm": 3.7879319190979004, "learning_rate": 1.6839332748024585e-05, "loss": 0.18448681831359864, "step": 960 }, { "epoch": 0.025551278716377513, "grad_norm": 2.152418851852417, "learning_rate": 1.701492537313433e-05, "loss": 0.17487335205078125, "step": 970 }, { "epoch": 0.02581469396087625, "grad_norm": 7.408229351043701, "learning_rate": 1.7190517998244077e-05, "loss": 0.20368638038635253, "step": 980 }, { "epoch": 0.026078109205374986, "grad_norm": 2.4159469604492188, "learning_rate": 1.736611062335382e-05, "loss": 0.17888660430908204, "step": 990 }, { "epoch": 0.026341524449873727, "grad_norm": 5.224660873413086, "learning_rate": 1.7541703248463566e-05, "loss": 0.19062639474868776, "step": 1000 }, { "epoch": 0.026604939694372463, "grad_norm": 2.645724058151245, "learning_rate": 1.771729587357331e-05, "loss": 0.19382178783416748, "step": 1010 }, { "epoch": 0.0268683549388712, "grad_norm": 4.211667537689209, "learning_rate": 1.7892888498683055e-05, "loss": 0.1925441026687622, "step": 1020 }, { "epoch": 0.027131770183369937, "grad_norm": 2.226701259613037, "learning_rate": 1.80684811237928e-05, "loss": 0.17627108097076416, "step": 1030 }, { "epoch": 0.027395185427868674, "grad_norm": 2.7954087257385254, "learning_rate": 1.8244073748902547e-05, "loss": 0.18700058460235597, "step": 1040 }, { "epoch": 0.02765860067236741, "grad_norm": 1.964969277381897, "learning_rate": 1.8419666374012293e-05, "loss": 0.180810546875, "step": 1050 }, { "epoch": 0.02792201591686615, "grad_norm": 3.448383331298828, "learning_rate": 1.859525899912204e-05, "loss": 0.1997987747192383, "step": 1060 }, { "epoch": 0.028185431161364887, "grad_norm": 2.3492472171783447, "learning_rate": 1.8770851624231782e-05, "loss": 0.1730423927307129, "step": 1070 }, { "epoch": 0.028448846405863624, "grad_norm": 1.6688950061798096, "learning_rate": 1.8946444249341528e-05, "loss": 0.19617725610733033, "step": 1080 }, { "epoch": 0.02871226165036236, "grad_norm": 5.571728229522705, "learning_rate": 1.9122036874451274e-05, "loss": 0.19413117170333863, "step": 1090 }, { "epoch": 0.028975676894861097, "grad_norm": 2.7966716289520264, "learning_rate": 1.929762949956102e-05, "loss": 0.1844877004623413, "step": 1100 }, { "epoch": 0.029239092139359834, "grad_norm": 3.4907538890838623, "learning_rate": 1.9473222124670767e-05, "loss": 0.1918797016143799, "step": 1110 }, { "epoch": 0.02950250738385857, "grad_norm": 4.151276588439941, "learning_rate": 1.9648814749780513e-05, "loss": 0.1754427433013916, "step": 1120 }, { "epoch": 0.02976592262835731, "grad_norm": 4.583204746246338, "learning_rate": 1.9824407374890256e-05, "loss": 0.1856753945350647, "step": 1130 }, { "epoch": 0.030029337872856048, "grad_norm": 1.847219705581665, "learning_rate": 2e-05, "loss": 0.18943570852279662, "step": 1140 }, { "epoch": 0.030292753117354784, "grad_norm": 6.929484844207764, "learning_rate": 1.9999996360784282e-05, "loss": 0.1833400845527649, "step": 1150 }, { "epoch": 0.03055616836185352, "grad_norm": 3.677119016647339, "learning_rate": 1.9999985443139778e-05, "loss": 0.1743283748626709, "step": 1160 }, { "epoch": 0.030819583606352258, "grad_norm": 3.4307937622070312, "learning_rate": 1.999996724707443e-05, "loss": 0.1905799150466919, "step": 1170 }, { "epoch": 0.031082998850850994, "grad_norm": 2.0864408016204834, "learning_rate": 1.9999941772601482e-05, "loss": 0.177330482006073, "step": 1180 }, { "epoch": 0.03134641409534973, "grad_norm": 3.996828079223633, "learning_rate": 1.9999909019739482e-05, "loss": 0.1842091917991638, "step": 1190 }, { "epoch": 0.03160982933984847, "grad_norm": 3.210365056991577, "learning_rate": 1.9999868988512265e-05, "loss": 0.17470746040344237, "step": 1200 }, { "epoch": 0.031873244584347205, "grad_norm": 8.872448921203613, "learning_rate": 1.9999821678948964e-05, "loss": 0.17257288694381714, "step": 1210 }, { "epoch": 0.03213665982884595, "grad_norm": 5.979714393615723, "learning_rate": 1.999976709108402e-05, "loss": 0.18720695972442628, "step": 1220 }, { "epoch": 0.032400075073344685, "grad_norm": 6.96621036529541, "learning_rate": 1.9999705224957156e-05, "loss": 0.2011380672454834, "step": 1230 }, { "epoch": 0.03266349031784342, "grad_norm": 11.58321475982666, "learning_rate": 1.9999636080613406e-05, "loss": 0.19556007385253907, "step": 1240 }, { "epoch": 0.03292690556234216, "grad_norm": 4.83900260925293, "learning_rate": 1.9999559658103097e-05, "loss": 0.19261517524719238, "step": 1250 }, { "epoch": 0.033190320806840895, "grad_norm": 5.751323699951172, "learning_rate": 1.9999475957481848e-05, "loss": 0.19263167381286622, "step": 1260 }, { "epoch": 0.03345373605133963, "grad_norm": 1.4598523378372192, "learning_rate": 1.999938497881059e-05, "loss": 0.19707214832305908, "step": 1270 }, { "epoch": 0.03371715129583837, "grad_norm": 2.5506598949432373, "learning_rate": 1.999928672215553e-05, "loss": 0.18191065788269042, "step": 1280 }, { "epoch": 0.033980566540337105, "grad_norm": 1.7294237613677979, "learning_rate": 1.9999181187588187e-05, "loss": 0.19501736164093017, "step": 1290 }, { "epoch": 0.03424398178483584, "grad_norm": 4.6684980392456055, "learning_rate": 1.9999068375185372e-05, "loss": 0.1780603289604187, "step": 1300 }, { "epoch": 0.03450739702933458, "grad_norm": 6.918583393096924, "learning_rate": 1.99989482850292e-05, "loss": 0.18058443069458008, "step": 1310 }, { "epoch": 0.034770812273833315, "grad_norm": 1.8445689678192139, "learning_rate": 1.9998820917207076e-05, "loss": 0.17636876106262206, "step": 1320 }, { "epoch": 0.03503422751833205, "grad_norm": 1.609938621520996, "learning_rate": 1.99986862718117e-05, "loss": 0.18162616491317748, "step": 1330 }, { "epoch": 0.03529764276283079, "grad_norm": 4.488358497619629, "learning_rate": 1.999854434894108e-05, "loss": 0.1871998906135559, "step": 1340 }, { "epoch": 0.03556105800732953, "grad_norm": 3.3179218769073486, "learning_rate": 1.9998395148698505e-05, "loss": 0.18248987197875977, "step": 1350 }, { "epoch": 0.03582447325182827, "grad_norm": 7.8547515869140625, "learning_rate": 1.9998238671192575e-05, "loss": 0.19143036603927613, "step": 1360 }, { "epoch": 0.036087888496327006, "grad_norm": 4.28023099899292, "learning_rate": 1.999807491653718e-05, "loss": 0.18874034881591797, "step": 1370 }, { "epoch": 0.03635130374082574, "grad_norm": 3.570460081100464, "learning_rate": 1.9997903884851503e-05, "loss": 0.1830179810523987, "step": 1380 }, { "epoch": 0.03661471898532448, "grad_norm": 2.0661065578460693, "learning_rate": 1.999772557626004e-05, "loss": 0.1824434518814087, "step": 1390 }, { "epoch": 0.036878134229823216, "grad_norm": 3.376643657684326, "learning_rate": 1.9997539990892555e-05, "loss": 0.16956480741500854, "step": 1400 }, { "epoch": 0.03714154947432195, "grad_norm": 3.1696624755859375, "learning_rate": 1.9997347128884138e-05, "loss": 0.179532253742218, "step": 1410 }, { "epoch": 0.03740496471882069, "grad_norm": 2.2696433067321777, "learning_rate": 1.9997146990375162e-05, "loss": 0.19119040966033934, "step": 1420 }, { "epoch": 0.037668379963319426, "grad_norm": 3.780425548553467, "learning_rate": 1.999693957551129e-05, "loss": 0.17809926271438598, "step": 1430 }, { "epoch": 0.03793179520781816, "grad_norm": 4.621781826019287, "learning_rate": 1.999672488444349e-05, "loss": 0.17357277870178223, "step": 1440 }, { "epoch": 0.0381952104523169, "grad_norm": 3.1671087741851807, "learning_rate": 1.999650291732802e-05, "loss": 0.19275472164154053, "step": 1450 }, { "epoch": 0.038458625696815636, "grad_norm": 2.26438307762146, "learning_rate": 1.9996273674326446e-05, "loss": 0.1807342529296875, "step": 1460 }, { "epoch": 0.03872204094131437, "grad_norm": 2.609966516494751, "learning_rate": 1.9996037155605615e-05, "loss": 0.17559963464736938, "step": 1470 }, { "epoch": 0.03898545618581312, "grad_norm": 3.859421730041504, "learning_rate": 1.9995793361337676e-05, "loss": 0.18137001991271973, "step": 1480 }, { "epoch": 0.03924887143031185, "grad_norm": 5.253348350524902, "learning_rate": 1.9995542291700075e-05, "loss": 0.18196004629135132, "step": 1490 }, { "epoch": 0.03951228667481059, "grad_norm": 1.4769941568374634, "learning_rate": 1.9995283946875548e-05, "loss": 0.1728345274925232, "step": 1500 }, { "epoch": 0.03977570191930933, "grad_norm": 5.539405345916748, "learning_rate": 1.999501832705213e-05, "loss": 0.1693485975265503, "step": 1510 }, { "epoch": 0.04003911716380806, "grad_norm": 1.4773509502410889, "learning_rate": 1.9994745432423154e-05, "loss": 0.1684037446975708, "step": 1520 }, { "epoch": 0.0403025324083068, "grad_norm": 4.749035835266113, "learning_rate": 1.9994465263187242e-05, "loss": 0.16785643100738526, "step": 1530 }, { "epoch": 0.04056594765280554, "grad_norm": 5.830752372741699, "learning_rate": 1.9994177819548313e-05, "loss": 0.19773929119110106, "step": 1540 }, { "epoch": 0.040829362897304274, "grad_norm": 4.104715347290039, "learning_rate": 1.999388310171558e-05, "loss": 0.1910729169845581, "step": 1550 }, { "epoch": 0.04109277814180301, "grad_norm": 4.796066761016846, "learning_rate": 1.9993581109903557e-05, "loss": 0.16582725048065186, "step": 1560 }, { "epoch": 0.04135619338630175, "grad_norm": 7.825209140777588, "learning_rate": 1.999327184433204e-05, "loss": 0.18026400804519654, "step": 1570 }, { "epoch": 0.041619608630800484, "grad_norm": 2.6738674640655518, "learning_rate": 1.9992955305226127e-05, "loss": 0.1760789632797241, "step": 1580 }, { "epoch": 0.04188302387529922, "grad_norm": 2.193556308746338, "learning_rate": 1.9992631492816214e-05, "loss": 0.1784481167793274, "step": 1590 }, { "epoch": 0.04214643911979796, "grad_norm": 2.4423370361328125, "learning_rate": 1.9992300407337977e-05, "loss": 0.1917218804359436, "step": 1600 }, { "epoch": 0.0424098543642967, "grad_norm": 7.086067199707031, "learning_rate": 1.9991962049032403e-05, "loss": 0.17045328617095948, "step": 1610 }, { "epoch": 0.04267326960879544, "grad_norm": 6.226284503936768, "learning_rate": 1.9991616418145757e-05, "loss": 0.17921960353851318, "step": 1620 }, { "epoch": 0.042936684853294174, "grad_norm": 2.6485233306884766, "learning_rate": 1.9991263514929606e-05, "loss": 0.1642894744873047, "step": 1630 }, { "epoch": 0.04320010009779291, "grad_norm": 5.3581461906433105, "learning_rate": 1.999090333964081e-05, "loss": 0.18506579399108886, "step": 1640 }, { "epoch": 0.04346351534229165, "grad_norm": 1.9961304664611816, "learning_rate": 1.9990535892541524e-05, "loss": 0.16968700885772706, "step": 1650 }, { "epoch": 0.043726930586790384, "grad_norm": 4.116445064544678, "learning_rate": 1.999016117389918e-05, "loss": 0.19753968715667725, "step": 1660 }, { "epoch": 0.04399034583128912, "grad_norm": 4.736153602600098, "learning_rate": 1.998977918398653e-05, "loss": 0.19476613998413086, "step": 1670 }, { "epoch": 0.04425376107578786, "grad_norm": 1.445264220237732, "learning_rate": 1.9989389923081587e-05, "loss": 0.1730787754058838, "step": 1680 }, { "epoch": 0.044517176320286594, "grad_norm": 3.5059876441955566, "learning_rate": 1.998899339146768e-05, "loss": 0.17655656337738038, "step": 1690 }, { "epoch": 0.04478059156478533, "grad_norm": 1.6106339693069458, "learning_rate": 1.9988589589433423e-05, "loss": 0.18839809894561768, "step": 1700 }, { "epoch": 0.04504400680928407, "grad_norm": 4.304480075836182, "learning_rate": 1.9988178517272717e-05, "loss": 0.1861241102218628, "step": 1710 }, { "epoch": 0.045307422053782805, "grad_norm": 2.427762269973755, "learning_rate": 1.998776017528476e-05, "loss": 0.18289952278137206, "step": 1720 }, { "epoch": 0.04557083729828155, "grad_norm": 4.614663124084473, "learning_rate": 1.9987334563774038e-05, "loss": 0.1882009267807007, "step": 1730 }, { "epoch": 0.045834252542780285, "grad_norm": 4.298706531524658, "learning_rate": 1.998690168305033e-05, "loss": 0.17877914905548095, "step": 1740 }, { "epoch": 0.04609766778727902, "grad_norm": 1.7473368644714355, "learning_rate": 1.9986461533428708e-05, "loss": 0.1828122019767761, "step": 1750 }, { "epoch": 0.04636108303177776, "grad_norm": 2.70392107963562, "learning_rate": 1.9986014115229525e-05, "loss": 0.17625017166137696, "step": 1760 }, { "epoch": 0.046624498276276495, "grad_norm": 1.3157813549041748, "learning_rate": 1.998555942877844e-05, "loss": 0.18333003520965577, "step": 1770 }, { "epoch": 0.04688791352077523, "grad_norm": 2.9048821926116943, "learning_rate": 1.9985097474406387e-05, "loss": 0.18014311790466309, "step": 1780 }, { "epoch": 0.04715132876527397, "grad_norm": 2.592315435409546, "learning_rate": 1.99846282524496e-05, "loss": 0.1690608024597168, "step": 1790 }, { "epoch": 0.047414744009772705, "grad_norm": 4.200170516967773, "learning_rate": 1.9984151763249598e-05, "loss": 0.18071997165679932, "step": 1800 }, { "epoch": 0.04767815925427144, "grad_norm": 2.5111443996429443, "learning_rate": 1.9983668007153185e-05, "loss": 0.18945413827896118, "step": 1810 }, { "epoch": 0.04794157449877018, "grad_norm": 3.2767677307128906, "learning_rate": 1.998317698451247e-05, "loss": 0.1769859790802002, "step": 1820 }, { "epoch": 0.048204989743268915, "grad_norm": 2.4303133487701416, "learning_rate": 1.9982678695684833e-05, "loss": 0.17796556949615477, "step": 1830 }, { "epoch": 0.04846840498776765, "grad_norm": 4.410314559936523, "learning_rate": 1.998217314103295e-05, "loss": 0.17561975717544556, "step": 1840 }, { "epoch": 0.04873182023226639, "grad_norm": 1.451420545578003, "learning_rate": 1.998166032092479e-05, "loss": 0.18310314416885376, "step": 1850 }, { "epoch": 0.04899523547676513, "grad_norm": 1.6260236501693726, "learning_rate": 1.99811402357336e-05, "loss": 0.1645521640777588, "step": 1860 }, { "epoch": 0.04925865072126387, "grad_norm": 2.8479056358337402, "learning_rate": 1.9980612885837925e-05, "loss": 0.18152329921722413, "step": 1870 }, { "epoch": 0.049522065965762606, "grad_norm": 3.434023857116699, "learning_rate": 1.998007827162159e-05, "loss": 0.17120943069458008, "step": 1880 }, { "epoch": 0.04978548121026134, "grad_norm": 4.058463096618652, "learning_rate": 1.9979536393473715e-05, "loss": 0.1836883544921875, "step": 1890 }, { "epoch": 0.05004889645476008, "grad_norm": 2.857686758041382, "learning_rate": 1.9978987251788696e-05, "loss": 0.18008556365966796, "step": 1900 }, { "epoch": 0.050312311699258816, "grad_norm": 4.272885799407959, "learning_rate": 1.997843084696623e-05, "loss": 0.18092842102050782, "step": 1910 }, { "epoch": 0.05057572694375755, "grad_norm": 4.974207878112793, "learning_rate": 1.9977867179411282e-05, "loss": 0.1544112205505371, "step": 1920 }, { "epoch": 0.05083914218825629, "grad_norm": 2.884653329849243, "learning_rate": 1.997729624953412e-05, "loss": 0.18403953313827515, "step": 1930 }, { "epoch": 0.051102557432755026, "grad_norm": 1.7507340908050537, "learning_rate": 1.997671805775029e-05, "loss": 0.1789767861366272, "step": 1940 }, { "epoch": 0.05136597267725376, "grad_norm": 3.490316152572632, "learning_rate": 1.9976132604480625e-05, "loss": 0.1749184250831604, "step": 1950 }, { "epoch": 0.0516293879217525, "grad_norm": 3.8869993686676025, "learning_rate": 1.997553989015125e-05, "loss": 0.19040322303771973, "step": 1960 }, { "epoch": 0.051892803166251236, "grad_norm": 4.155693531036377, "learning_rate": 1.9974939915193557e-05, "loss": 0.17040687799453735, "step": 1970 }, { "epoch": 0.05215621841074997, "grad_norm": 1.6790688037872314, "learning_rate": 1.9974332680044237e-05, "loss": 0.18179289102554322, "step": 1980 }, { "epoch": 0.05241963365524872, "grad_norm": 1.6242108345031738, "learning_rate": 1.9973718185145262e-05, "loss": 0.1661834239959717, "step": 1990 }, { "epoch": 0.05268304889974745, "grad_norm": 4.162831783294678, "learning_rate": 1.997309643094389e-05, "loss": 0.1747814178466797, "step": 2000 }, { "epoch": 0.05294646414424619, "grad_norm": 1.384961485862732, "learning_rate": 1.997246741789266e-05, "loss": 0.17896897792816163, "step": 2010 }, { "epoch": 0.05320987938874493, "grad_norm": 4.968136310577393, "learning_rate": 1.99718311464494e-05, "loss": 0.18545992374420167, "step": 2020 }, { "epoch": 0.05347329463324366, "grad_norm": 3.244908332824707, "learning_rate": 1.9971187617077204e-05, "loss": 0.1739330291748047, "step": 2030 }, { "epoch": 0.0537367098777424, "grad_norm": 2.5963361263275146, "learning_rate": 1.9970536830244466e-05, "loss": 0.17251211404800415, "step": 2040 }, { "epoch": 0.05400012512224114, "grad_norm": 2.023715019226074, "learning_rate": 1.9969878786424863e-05, "loss": 0.17024842500686646, "step": 2050 }, { "epoch": 0.054263540366739874, "grad_norm": 1.5900593996047974, "learning_rate": 1.9969213486097335e-05, "loss": 0.18348467350006104, "step": 2060 }, { "epoch": 0.05452695561123861, "grad_norm": 4.949227809906006, "learning_rate": 1.996854092974613e-05, "loss": 0.1835564374923706, "step": 2070 }, { "epoch": 0.05479037085573735, "grad_norm": 3.683088779449463, "learning_rate": 1.9967861117860754e-05, "loss": 0.17205485105514526, "step": 2080 }, { "epoch": 0.055053786100236084, "grad_norm": 1.9927631616592407, "learning_rate": 1.9967174050936008e-05, "loss": 0.17923445701599122, "step": 2090 }, { "epoch": 0.05531720134473482, "grad_norm": 2.545353651046753, "learning_rate": 1.9966479729471966e-05, "loss": 0.18440502882003784, "step": 2100 }, { "epoch": 0.05558061658923356, "grad_norm": 4.966377258300781, "learning_rate": 1.9965778153973986e-05, "loss": 0.1657766342163086, "step": 2110 }, { "epoch": 0.0558440318337323, "grad_norm": 2.377627372741699, "learning_rate": 1.9965069324952703e-05, "loss": 0.1782365083694458, "step": 2120 }, { "epoch": 0.05610744707823104, "grad_norm": 5.926941394805908, "learning_rate": 1.9964353242924038e-05, "loss": 0.17844765186309813, "step": 2130 }, { "epoch": 0.056370862322729774, "grad_norm": 2.0028176307678223, "learning_rate": 1.9963629908409184e-05, "loss": 0.17593191862106322, "step": 2140 }, { "epoch": 0.05663427756722851, "grad_norm": 2.4002223014831543, "learning_rate": 1.9962899321934616e-05, "loss": 0.16249873638153076, "step": 2150 }, { "epoch": 0.05689769281172725, "grad_norm": 1.9579808712005615, "learning_rate": 1.9962161484032085e-05, "loss": 0.16547211408615112, "step": 2160 }, { "epoch": 0.057161108056225984, "grad_norm": 2.7050206661224365, "learning_rate": 1.9961416395238623e-05, "loss": 0.15595681667327882, "step": 2170 }, { "epoch": 0.05742452330072472, "grad_norm": 2.6615474224090576, "learning_rate": 1.9960664056096534e-05, "loss": 0.18808884620666505, "step": 2180 }, { "epoch": 0.05768793854522346, "grad_norm": 7.249985694885254, "learning_rate": 1.9959904467153408e-05, "loss": 0.18617500066757203, "step": 2190 }, { "epoch": 0.057951353789722194, "grad_norm": 1.2595937252044678, "learning_rate": 1.99591376289621e-05, "loss": 0.16368980407714845, "step": 2200 }, { "epoch": 0.05821476903422093, "grad_norm": 3.7606801986694336, "learning_rate": 1.9958363542080758e-05, "loss": 0.17077407836914063, "step": 2210 }, { "epoch": 0.05847818427871967, "grad_norm": 1.8336760997772217, "learning_rate": 1.9957582207072784e-05, "loss": 0.16321921348571777, "step": 2220 }, { "epoch": 0.058741599523218405, "grad_norm": 2.8326961994171143, "learning_rate": 1.9956793624506875e-05, "loss": 0.1778048276901245, "step": 2230 }, { "epoch": 0.05900501476771714, "grad_norm": 5.458047866821289, "learning_rate": 1.9955997794956994e-05, "loss": 0.16475337743759155, "step": 2240 }, { "epoch": 0.059268430012215885, "grad_norm": 3.6259820461273193, "learning_rate": 1.995519471900238e-05, "loss": 0.17252004146575928, "step": 2250 }, { "epoch": 0.05953184525671462, "grad_norm": 2.6252448558807373, "learning_rate": 1.9954384397227545e-05, "loss": 0.17881488800048828, "step": 2260 }, { "epoch": 0.05979526050121336, "grad_norm": 3.9076385498046875, "learning_rate": 1.9953566830222276e-05, "loss": 0.17018630504608154, "step": 2270 }, { "epoch": 0.060058675745712095, "grad_norm": 2.9033734798431396, "learning_rate": 1.9952742018581634e-05, "loss": 0.17954621315002442, "step": 2280 }, { "epoch": 0.06032209099021083, "grad_norm": 3.6916890144348145, "learning_rate": 1.9951909962905956e-05, "loss": 0.17365128993988038, "step": 2290 }, { "epoch": 0.06058550623470957, "grad_norm": 4.976506233215332, "learning_rate": 1.9951070663800843e-05, "loss": 0.1733959436416626, "step": 2300 }, { "epoch": 0.060848921479208305, "grad_norm": 1.9410442113876343, "learning_rate": 1.995022412187718e-05, "loss": 0.16992549896240233, "step": 2310 }, { "epoch": 0.06111233672370704, "grad_norm": 3.6085565090179443, "learning_rate": 1.9949370337751105e-05, "loss": 0.17807869911193847, "step": 2320 }, { "epoch": 0.06137575196820578, "grad_norm": 3.8716084957122803, "learning_rate": 1.994850931204405e-05, "loss": 0.17094887495040895, "step": 2330 }, { "epoch": 0.061639167212704515, "grad_norm": 7.494531154632568, "learning_rate": 1.99476410453827e-05, "loss": 0.18423082828521728, "step": 2340 }, { "epoch": 0.06190258245720325, "grad_norm": 4.898642063140869, "learning_rate": 1.994676553839902e-05, "loss": 0.1892184019088745, "step": 2350 }, { "epoch": 0.06216599770170199, "grad_norm": 2.67594313621521, "learning_rate": 1.9945882791730246e-05, "loss": 0.18599352836608887, "step": 2360 }, { "epoch": 0.06242941294620073, "grad_norm": 4.25570821762085, "learning_rate": 1.994499280601887e-05, "loss": 0.1649078369140625, "step": 2370 }, { "epoch": 0.06269282819069946, "grad_norm": 3.1538801193237305, "learning_rate": 1.994409558191267e-05, "loss": 0.16146588325500488, "step": 2380 }, { "epoch": 0.0629562434351982, "grad_norm": 5.567690849304199, "learning_rate": 1.9943191120064672e-05, "loss": 0.18166754245758057, "step": 2390 }, { "epoch": 0.06321965867969694, "grad_norm": 2.4334769248962402, "learning_rate": 1.99422794211332e-05, "loss": 0.16374545097351073, "step": 2400 }, { "epoch": 0.06348307392419568, "grad_norm": 1.3590749502182007, "learning_rate": 1.9941360485781814e-05, "loss": 0.17545441389083863, "step": 2410 }, { "epoch": 0.06374648916869441, "grad_norm": 1.4375967979431152, "learning_rate": 1.9940434314679363e-05, "loss": 0.16284945011138915, "step": 2420 }, { "epoch": 0.06400990441319315, "grad_norm": 2.259549617767334, "learning_rate": 1.9939500908499947e-05, "loss": 0.16767736673355102, "step": 2430 }, { "epoch": 0.0642733196576919, "grad_norm": 7.7775678634643555, "learning_rate": 1.9938560267922948e-05, "loss": 0.18161977529525758, "step": 2440 }, { "epoch": 0.06453673490219063, "grad_norm": 2.012789011001587, "learning_rate": 1.9937612393632995e-05, "loss": 0.15819730758666992, "step": 2450 }, { "epoch": 0.06480015014668937, "grad_norm": 2.912641763687134, "learning_rate": 1.9936657286320003e-05, "loss": 0.16389341354370118, "step": 2460 }, { "epoch": 0.0650635653911881, "grad_norm": 2.673279285430908, "learning_rate": 1.993569494667913e-05, "loss": 0.17740426063537598, "step": 2470 }, { "epoch": 0.06532698063568684, "grad_norm": 4.238875389099121, "learning_rate": 1.9934725375410812e-05, "loss": 0.18156671524047852, "step": 2480 }, { "epoch": 0.06559039588018557, "grad_norm": 2.0430192947387695, "learning_rate": 1.9933748573220745e-05, "loss": 0.1791893482208252, "step": 2490 }, { "epoch": 0.06585381112468432, "grad_norm": 4.585734844207764, "learning_rate": 1.993276454081989e-05, "loss": 0.17079017162322999, "step": 2500 }, { "epoch": 0.06611722636918305, "grad_norm": 2.2382588386535645, "learning_rate": 1.9931773278924467e-05, "loss": 0.1800147771835327, "step": 2510 }, { "epoch": 0.06638064161368179, "grad_norm": 3.7071409225463867, "learning_rate": 1.9930774788255957e-05, "loss": 0.1830217123031616, "step": 2520 }, { "epoch": 0.06664405685818052, "grad_norm": 2.2259862422943115, "learning_rate": 1.992976906954111e-05, "loss": 0.18647122383117676, "step": 2530 }, { "epoch": 0.06690747210267926, "grad_norm": 3.0462238788604736, "learning_rate": 1.992875612351192e-05, "loss": 0.17233673334121705, "step": 2540 }, { "epoch": 0.067170887347178, "grad_norm": 1.0444905757904053, "learning_rate": 1.9927735950905663e-05, "loss": 0.15813591480255126, "step": 2550 }, { "epoch": 0.06743430259167674, "grad_norm": 0.9914074540138245, "learning_rate": 1.9926708552464865e-05, "loss": 0.18611812591552734, "step": 2560 }, { "epoch": 0.06769771783617548, "grad_norm": 5.989922523498535, "learning_rate": 1.9925673928937307e-05, "loss": 0.18290371894836427, "step": 2570 }, { "epoch": 0.06796113308067421, "grad_norm": 2.9222300052642822, "learning_rate": 1.9924632081076032e-05, "loss": 0.17300392389297486, "step": 2580 }, { "epoch": 0.06822454832517295, "grad_norm": 1.6255520582199097, "learning_rate": 1.9923583009639344e-05, "loss": 0.18728654384613036, "step": 2590 }, { "epoch": 0.06848796356967168, "grad_norm": 3.6903510093688965, "learning_rate": 1.9922526715390802e-05, "loss": 0.1758969783782959, "step": 2600 }, { "epoch": 0.06875137881417043, "grad_norm": 4.133755683898926, "learning_rate": 1.9921463199099222e-05, "loss": 0.17395939826965331, "step": 2610 }, { "epoch": 0.06901479405866916, "grad_norm": 1.8804093599319458, "learning_rate": 1.992039246153868e-05, "loss": 0.1822284460067749, "step": 2620 }, { "epoch": 0.0692782093031679, "grad_norm": 4.366495609283447, "learning_rate": 1.99193145034885e-05, "loss": 0.16699020862579345, "step": 2630 }, { "epoch": 0.06954162454766663, "grad_norm": 40.970558166503906, "learning_rate": 1.9918229325733267e-05, "loss": 1.6216285705566407, "step": 2640 }, { "epoch": 0.06980503979216537, "grad_norm": 2.911933183670044, "learning_rate": 1.9917136929062824e-05, "loss": 1.43115234375, "step": 2650 }, { "epoch": 0.0700684550366641, "grad_norm": 3.2500436305999756, "learning_rate": 1.991603731427226e-05, "loss": 0.215869140625, "step": 2660 }, { "epoch": 0.07033187028116285, "grad_norm": 3.1116034984588623, "learning_rate": 1.9914930482161928e-05, "loss": 0.20029737949371337, "step": 2670 }, { "epoch": 0.07059528552566158, "grad_norm": 1.7858082056045532, "learning_rate": 1.9913816433537425e-05, "loss": 0.17998937368392945, "step": 2680 }, { "epoch": 0.07085870077016032, "grad_norm": 1.9605040550231934, "learning_rate": 1.99126951692096e-05, "loss": 0.2029872179031372, "step": 2690 }, { "epoch": 0.07112211601465906, "grad_norm": 4.031591892242432, "learning_rate": 1.991156668999456e-05, "loss": 0.2077608585357666, "step": 2700 }, { "epoch": 0.0713855312591578, "grad_norm": 2.325939655303955, "learning_rate": 1.9910430996713668e-05, "loss": 0.178499436378479, "step": 2710 }, { "epoch": 0.07164894650365654, "grad_norm": 3.8449032306671143, "learning_rate": 1.9909288090193518e-05, "loss": 0.19340702295303344, "step": 2720 }, { "epoch": 0.07191236174815527, "grad_norm": 1.9329261779785156, "learning_rate": 1.9908137971265977e-05, "loss": 0.18916661739349366, "step": 2730 }, { "epoch": 0.07217577699265401, "grad_norm": 2.2261054515838623, "learning_rate": 1.9906980640768146e-05, "loss": 0.17411012649536134, "step": 2740 }, { "epoch": 0.07243919223715274, "grad_norm": 2.6896958351135254, "learning_rate": 1.9905816099542382e-05, "loss": 0.17388646602630614, "step": 2750 }, { "epoch": 0.07270260748165149, "grad_norm": 4.654409408569336, "learning_rate": 1.9904644348436282e-05, "loss": 0.18504722118377687, "step": 2760 }, { "epoch": 0.07296602272615021, "grad_norm": 2.4094605445861816, "learning_rate": 1.990346538830271e-05, "loss": 0.17755537033081054, "step": 2770 }, { "epoch": 0.07322943797064896, "grad_norm": 3.809527635574341, "learning_rate": 1.9902279219999752e-05, "loss": 0.17706785202026368, "step": 2780 }, { "epoch": 0.07349285321514769, "grad_norm": 4.955341815948486, "learning_rate": 1.9901085844390757e-05, "loss": 0.17934825420379638, "step": 2790 }, { "epoch": 0.07375626845964643, "grad_norm": 7.148049354553223, "learning_rate": 1.9899885262344316e-05, "loss": 0.17972722053527831, "step": 2800 }, { "epoch": 0.07401968370414516, "grad_norm": 2.7793192863464355, "learning_rate": 1.9898677474734266e-05, "loss": 0.18006224632263185, "step": 2810 }, { "epoch": 0.0742830989486439, "grad_norm": 3.0571916103363037, "learning_rate": 1.9897462482439684e-05, "loss": 0.1761543035507202, "step": 2820 }, { "epoch": 0.07454651419314265, "grad_norm": 3.655093193054199, "learning_rate": 1.9896240286344892e-05, "loss": 0.18790814876556397, "step": 2830 }, { "epoch": 0.07480992943764138, "grad_norm": 3.340087413787842, "learning_rate": 1.989501088733946e-05, "loss": 0.17855364084243774, "step": 2840 }, { "epoch": 0.07507334468214012, "grad_norm": 2.1285271644592285, "learning_rate": 1.9893774286318198e-05, "loss": 0.17807376384735107, "step": 2850 }, { "epoch": 0.07533675992663885, "grad_norm": 7.163172245025635, "learning_rate": 1.9892530484181155e-05, "loss": 0.18757336139678954, "step": 2860 }, { "epoch": 0.0756001751711376, "grad_norm": 3.9823410511016846, "learning_rate": 1.989127948183363e-05, "loss": 0.170552396774292, "step": 2870 }, { "epoch": 0.07586359041563633, "grad_norm": 3.685450553894043, "learning_rate": 1.9890021280186148e-05, "loss": 0.19075794219970704, "step": 2880 }, { "epoch": 0.07612700566013507, "grad_norm": 1.975399136543274, "learning_rate": 1.9888755880154485e-05, "loss": 0.16604335308074952, "step": 2890 }, { "epoch": 0.0763904209046338, "grad_norm": 4.934932231903076, "learning_rate": 1.988748328265966e-05, "loss": 0.15102682113647461, "step": 2900 }, { "epoch": 0.07665383614913254, "grad_norm": 2.8079192638397217, "learning_rate": 1.9886203488627918e-05, "loss": 0.18284976482391357, "step": 2910 }, { "epoch": 0.07691725139363127, "grad_norm": 4.950689792633057, "learning_rate": 1.988491649899075e-05, "loss": 0.17645611763000488, "step": 2920 }, { "epoch": 0.07718066663813002, "grad_norm": 4.967823028564453, "learning_rate": 1.988362231468488e-05, "loss": 0.1677510142326355, "step": 2930 }, { "epoch": 0.07744408188262875, "grad_norm": 1.7485764026641846, "learning_rate": 1.9882320936652278e-05, "loss": 0.1873460054397583, "step": 2940 }, { "epoch": 0.07770749712712749, "grad_norm": 3.148735284805298, "learning_rate": 1.9881012365840138e-05, "loss": 0.1752831220626831, "step": 2950 }, { "epoch": 0.07797091237162623, "grad_norm": 2.4079582691192627, "learning_rate": 1.987969660320089e-05, "loss": 0.1770440936088562, "step": 2960 }, { "epoch": 0.07823432761612496, "grad_norm": 3.234750986099243, "learning_rate": 1.9878373649692213e-05, "loss": 0.18718385696411133, "step": 2970 }, { "epoch": 0.0784977428606237, "grad_norm": 1.2746676206588745, "learning_rate": 1.9877043506277004e-05, "loss": 0.18140406608581544, "step": 2980 }, { "epoch": 0.07876115810512244, "grad_norm": 4.2238078117370605, "learning_rate": 1.9875706173923396e-05, "loss": 0.17327748537063598, "step": 2990 }, { "epoch": 0.07902457334962118, "grad_norm": 2.159634590148926, "learning_rate": 1.9874361653604765e-05, "loss": 0.16466054916381836, "step": 3000 }, { "epoch": 0.07928798859411991, "grad_norm": 2.4309141635894775, "learning_rate": 1.9873009946299702e-05, "loss": 0.18613797426223755, "step": 3010 }, { "epoch": 0.07955140383861865, "grad_norm": 3.9900870323181152, "learning_rate": 1.9871651052992046e-05, "loss": 0.164133620262146, "step": 3020 }, { "epoch": 0.07981481908311738, "grad_norm": 4.18991756439209, "learning_rate": 1.987028497467085e-05, "loss": 0.15075888633728027, "step": 3030 }, { "epoch": 0.08007823432761613, "grad_norm": 2.3080949783325195, "learning_rate": 1.986891171233041e-05, "loss": 0.16066547632217407, "step": 3040 }, { "epoch": 0.08034164957211486, "grad_norm": 4.233375549316406, "learning_rate": 1.9867531266970247e-05, "loss": 0.17523720264434814, "step": 3050 }, { "epoch": 0.0806050648166136, "grad_norm": 1.6291990280151367, "learning_rate": 1.9866143639595107e-05, "loss": 0.17919942140579223, "step": 3060 }, { "epoch": 0.08086848006111233, "grad_norm": 1.9843177795410156, "learning_rate": 1.9864748831214964e-05, "loss": 0.16186938285827637, "step": 3070 }, { "epoch": 0.08113189530561107, "grad_norm": 2.487389326095581, "learning_rate": 1.986334684284502e-05, "loss": 0.1855526089668274, "step": 3080 }, { "epoch": 0.08139531055010982, "grad_norm": 4.981988906860352, "learning_rate": 1.98619376755057e-05, "loss": 0.17050676345825194, "step": 3090 }, { "epoch": 0.08165872579460855, "grad_norm": 1.663664698600769, "learning_rate": 1.986052133022266e-05, "loss": 0.19078845977783204, "step": 3100 }, { "epoch": 0.08192214103910729, "grad_norm": 4.502136707305908, "learning_rate": 1.985909780802678e-05, "loss": 0.1691272735595703, "step": 3110 }, { "epoch": 0.08218555628360602, "grad_norm": 3.1019911766052246, "learning_rate": 1.985766710995416e-05, "loss": 0.1697216510772705, "step": 3120 }, { "epoch": 0.08244897152810476, "grad_norm": 6.9822916984558105, "learning_rate": 1.9856229237046115e-05, "loss": 0.17920457124710082, "step": 3130 }, { "epoch": 0.0827123867726035, "grad_norm": 2.4769084453582764, "learning_rate": 1.9854784190349203e-05, "loss": 0.16911983489990234, "step": 3140 }, { "epoch": 0.08297580201710224, "grad_norm": 6.559800624847412, "learning_rate": 1.985333197091518e-05, "loss": 0.17824146747589112, "step": 3150 }, { "epoch": 0.08323921726160097, "grad_norm": 5.821385860443115, "learning_rate": 1.9851872579801044e-05, "loss": 0.17066729068756104, "step": 3160 }, { "epoch": 0.08350263250609971, "grad_norm": 1.2309391498565674, "learning_rate": 1.9850406018068997e-05, "loss": 0.16355862617492675, "step": 3170 }, { "epoch": 0.08376604775059844, "grad_norm": 3.973520040512085, "learning_rate": 1.984893228678647e-05, "loss": 0.15969277620315553, "step": 3180 }, { "epoch": 0.08402946299509718, "grad_norm": 3.6637420654296875, "learning_rate": 1.9847451387026103e-05, "loss": 0.1678025245666504, "step": 3190 }, { "epoch": 0.08429287823959591, "grad_norm": 4.729883193969727, "learning_rate": 1.9845963319865758e-05, "loss": 0.1806029796600342, "step": 3200 }, { "epoch": 0.08455629348409466, "grad_norm": 2.749605178833008, "learning_rate": 1.9844468086388523e-05, "loss": 0.16283800601959228, "step": 3210 }, { "epoch": 0.0848197087285934, "grad_norm": 5.675039768218994, "learning_rate": 1.9842965687682686e-05, "loss": 0.1616047739982605, "step": 3220 }, { "epoch": 0.08508312397309213, "grad_norm": 2.598233461380005, "learning_rate": 1.9841456124841763e-05, "loss": 0.1753537654876709, "step": 3230 }, { "epoch": 0.08534653921759087, "grad_norm": 4.184293746948242, "learning_rate": 1.983993939896447e-05, "loss": 0.16005247831344604, "step": 3240 }, { "epoch": 0.0856099544620896, "grad_norm": 2.395411252975464, "learning_rate": 1.9838415511154755e-05, "loss": 0.18136755228042603, "step": 3250 }, { "epoch": 0.08587336970658835, "grad_norm": 3.916792392730713, "learning_rate": 1.9836884462521768e-05, "loss": 0.15952765941619873, "step": 3260 }, { "epoch": 0.08613678495108708, "grad_norm": 1.7144755125045776, "learning_rate": 1.9835346254179865e-05, "loss": 0.16965481042861938, "step": 3270 }, { "epoch": 0.08640020019558582, "grad_norm": 2.092229127883911, "learning_rate": 1.9833800887248628e-05, "loss": 0.17597393989562987, "step": 3280 }, { "epoch": 0.08666361544008455, "grad_norm": 4.647211074829102, "learning_rate": 1.983224836285284e-05, "loss": 0.17522543668746948, "step": 3290 }, { "epoch": 0.0869270306845833, "grad_norm": 5.068032264709473, "learning_rate": 1.9830688682122494e-05, "loss": 0.1847475290298462, "step": 3300 }, { "epoch": 0.08719044592908202, "grad_norm": 5.291374206542969, "learning_rate": 1.982912184619279e-05, "loss": 0.16535460948944092, "step": 3310 }, { "epoch": 0.08745386117358077, "grad_norm": 7.786944389343262, "learning_rate": 1.9827547856204147e-05, "loss": 0.16262664794921874, "step": 3320 }, { "epoch": 0.0877172764180795, "grad_norm": 3.2643558979034424, "learning_rate": 1.9825966713302172e-05, "loss": 0.16811026334762574, "step": 3330 }, { "epoch": 0.08798069166257824, "grad_norm": 3.914182662963867, "learning_rate": 1.98243784186377e-05, "loss": 0.16510536670684814, "step": 3340 }, { "epoch": 0.08824410690707699, "grad_norm": 3.3157384395599365, "learning_rate": 1.982278297336675e-05, "loss": 0.17148419618606567, "step": 3350 }, { "epoch": 0.08850752215157572, "grad_norm": 1.752765417098999, "learning_rate": 1.9821180378650563e-05, "loss": 0.1749314546585083, "step": 3360 }, { "epoch": 0.08877093739607446, "grad_norm": 4.425608158111572, "learning_rate": 1.9819570635655577e-05, "loss": 0.17113746404647828, "step": 3370 }, { "epoch": 0.08903435264057319, "grad_norm": 4.71959924697876, "learning_rate": 1.981795374555343e-05, "loss": 0.16317377090454102, "step": 3380 }, { "epoch": 0.08929776788507193, "grad_norm": 2.4054806232452393, "learning_rate": 1.981632970952096e-05, "loss": 0.1622145652770996, "step": 3390 }, { "epoch": 0.08956118312957066, "grad_norm": 1.7097535133361816, "learning_rate": 1.9814698528740222e-05, "loss": 0.1701422929763794, "step": 3400 }, { "epoch": 0.0898245983740694, "grad_norm": 3.4253084659576416, "learning_rate": 1.9813060204398448e-05, "loss": 0.16646794080734253, "step": 3410 }, { "epoch": 0.09008801361856814, "grad_norm": 2.6755495071411133, "learning_rate": 1.9811414737688088e-05, "loss": 0.17155914306640624, "step": 3420 }, { "epoch": 0.09035142886306688, "grad_norm": 1.913999080657959, "learning_rate": 1.9809762129806777e-05, "loss": 0.172481906414032, "step": 3430 }, { "epoch": 0.09061484410756561, "grad_norm": 2.5129635334014893, "learning_rate": 1.9808102381957364e-05, "loss": 0.17470791339874267, "step": 3440 }, { "epoch": 0.09087825935206435, "grad_norm": 3.1081159114837646, "learning_rate": 1.980643549534788e-05, "loss": 0.16773662567138672, "step": 3450 }, { "epoch": 0.0911416745965631, "grad_norm": 1.6166002750396729, "learning_rate": 1.980476147119155e-05, "loss": 0.15200254917144776, "step": 3460 }, { "epoch": 0.09140508984106183, "grad_norm": 1.6495922803878784, "learning_rate": 1.980308031070682e-05, "loss": 0.16635050773620605, "step": 3470 }, { "epoch": 0.09166850508556057, "grad_norm": 2.3104543685913086, "learning_rate": 1.9801392015117286e-05, "loss": 0.18132328987121582, "step": 3480 }, { "epoch": 0.0919319203300593, "grad_norm": 4.137028694152832, "learning_rate": 1.9799696585651782e-05, "loss": 0.15755207538604737, "step": 3490 }, { "epoch": 0.09219533557455804, "grad_norm": 1.7372796535491943, "learning_rate": 1.9797994023544307e-05, "loss": 0.17775838375091552, "step": 3500 }, { "epoch": 0.09245875081905677, "grad_norm": 1.2958176136016846, "learning_rate": 1.979628433003406e-05, "loss": 0.16738407611846923, "step": 3510 }, { "epoch": 0.09272216606355552, "grad_norm": 1.3560142517089844, "learning_rate": 1.9794567506365432e-05, "loss": 0.16316637992858887, "step": 3520 }, { "epoch": 0.09298558130805425, "grad_norm": 2.5291733741760254, "learning_rate": 1.9792843553788e-05, "loss": 0.16402239799499513, "step": 3530 }, { "epoch": 0.09324899655255299, "grad_norm": 2.2112607955932617, "learning_rate": 1.9791112473556528e-05, "loss": 0.1748220682144165, "step": 3540 }, { "epoch": 0.09351241179705172, "grad_norm": 1.8012775182724, "learning_rate": 1.978937426693097e-05, "loss": 0.17038497924804688, "step": 3550 }, { "epoch": 0.09377582704155046, "grad_norm": 2.194432020187378, "learning_rate": 1.9787628935176476e-05, "loss": 0.17311197519302368, "step": 3560 }, { "epoch": 0.0940392422860492, "grad_norm": 3.6434409618377686, "learning_rate": 1.9785876479563367e-05, "loss": 0.1741226315498352, "step": 3570 }, { "epoch": 0.09430265753054794, "grad_norm": 1.2375397682189941, "learning_rate": 1.978411690136716e-05, "loss": 0.17406488656997682, "step": 3580 }, { "epoch": 0.09456607277504668, "grad_norm": 2.157252311706543, "learning_rate": 1.9782350201868547e-05, "loss": 0.1665569543838501, "step": 3590 }, { "epoch": 0.09482948801954541, "grad_norm": 3.9698572158813477, "learning_rate": 1.978057638235341e-05, "loss": 0.17062528133392335, "step": 3600 }, { "epoch": 0.09509290326404415, "grad_norm": 2.7293968200683594, "learning_rate": 1.9778795444112812e-05, "loss": 0.1716085910797119, "step": 3610 }, { "epoch": 0.09535631850854288, "grad_norm": 5.102773666381836, "learning_rate": 1.9777007388443e-05, "loss": 0.16079416275024414, "step": 3620 }, { "epoch": 0.09561973375304163, "grad_norm": 1.377591609954834, "learning_rate": 1.9775212216645393e-05, "loss": 0.15675275325775145, "step": 3630 }, { "epoch": 0.09588314899754036, "grad_norm": 1.245958924293518, "learning_rate": 1.977340993002659e-05, "loss": 0.17969844341278077, "step": 3640 }, { "epoch": 0.0961465642420391, "grad_norm": 4.2353196144104, "learning_rate": 1.977160052989839e-05, "loss": 0.17932512760162353, "step": 3650 }, { "epoch": 0.09640997948653783, "grad_norm": 2.405181884765625, "learning_rate": 1.9769784017577734e-05, "loss": 0.16092146635055543, "step": 3660 }, { "epoch": 0.09667339473103657, "grad_norm": 9.04055118560791, "learning_rate": 1.976796039438677e-05, "loss": 0.18250563144683837, "step": 3670 }, { "epoch": 0.0969368099755353, "grad_norm": 5.899242401123047, "learning_rate": 1.9766129661652803e-05, "loss": 0.18091106414794922, "step": 3680 }, { "epoch": 0.09720022522003405, "grad_norm": 4.7462615966796875, "learning_rate": 1.9764291820708324e-05, "loss": 0.18074597120285035, "step": 3690 }, { "epoch": 0.09746364046453278, "grad_norm": 3.636507272720337, "learning_rate": 1.9762446872890984e-05, "loss": 0.16563973426818848, "step": 3700 }, { "epoch": 0.09772705570903152, "grad_norm": 2.274927854537964, "learning_rate": 1.976059481954363e-05, "loss": 0.18046171665191652, "step": 3710 }, { "epoch": 0.09799047095353026, "grad_norm": 1.392961859703064, "learning_rate": 1.9758735662014253e-05, "loss": 0.1657285213470459, "step": 3720 }, { "epoch": 0.098253886198029, "grad_norm": 2.6139304637908936, "learning_rate": 1.975686940165604e-05, "loss": 0.15866241455078126, "step": 3730 }, { "epoch": 0.09851730144252774, "grad_norm": 1.7217931747436523, "learning_rate": 1.9754996039827322e-05, "loss": 0.15727269649505615, "step": 3740 }, { "epoch": 0.09878071668702647, "grad_norm": 2.938109874725342, "learning_rate": 1.9753115577891623e-05, "loss": 0.16420741081237794, "step": 3750 }, { "epoch": 0.09904413193152521, "grad_norm": 1.2894885540008545, "learning_rate": 1.9751228017217623e-05, "loss": 0.15680664777755737, "step": 3760 }, { "epoch": 0.09930754717602394, "grad_norm": 2.5356783866882324, "learning_rate": 1.9749333359179166e-05, "loss": 0.15749037265777588, "step": 3770 }, { "epoch": 0.09957096242052269, "grad_norm": 3.032097816467285, "learning_rate": 1.974743160515527e-05, "loss": 0.16330833435058595, "step": 3780 }, { "epoch": 0.09983437766502141, "grad_norm": 3.3442776203155518, "learning_rate": 1.974552275653011e-05, "loss": 0.16863176822662354, "step": 3790 }, { "epoch": 0.10009779290952016, "grad_norm": 3.316377878189087, "learning_rate": 1.974360681469303e-05, "loss": 0.17165509462356568, "step": 3800 }, { "epoch": 0.10036120815401889, "grad_norm": 2.6527059078216553, "learning_rate": 1.9741683781038538e-05, "loss": 0.1694910407066345, "step": 3810 }, { "epoch": 0.10062462339851763, "grad_norm": 4.87701416015625, "learning_rate": 1.9739753656966294e-05, "loss": 0.16928658485412598, "step": 3820 }, { "epoch": 0.10088803864301636, "grad_norm": 2.733722686767578, "learning_rate": 1.9737816443881134e-05, "loss": 0.17243874073028564, "step": 3830 }, { "epoch": 0.1011514538875151, "grad_norm": 3.2734267711639404, "learning_rate": 1.973587214319304e-05, "loss": 0.16187753677368164, "step": 3840 }, { "epoch": 0.10141486913201385, "grad_norm": 1.5686461925506592, "learning_rate": 1.9733920756317157e-05, "loss": 0.1661075234413147, "step": 3850 }, { "epoch": 0.10167828437651258, "grad_norm": 1.671432614326477, "learning_rate": 1.973196228467379e-05, "loss": 0.16093600988388063, "step": 3860 }, { "epoch": 0.10194169962101132, "grad_norm": 1.977136254310608, "learning_rate": 1.9729996729688403e-05, "loss": 0.17692548036575317, "step": 3870 }, { "epoch": 0.10220511486551005, "grad_norm": 3.1447701454162598, "learning_rate": 1.9728024092791606e-05, "loss": 0.17668371200561522, "step": 3880 }, { "epoch": 0.1024685301100088, "grad_norm": 1.2613054513931274, "learning_rate": 1.9726044375419172e-05, "loss": 0.1734908938407898, "step": 3890 }, { "epoch": 0.10273194535450753, "grad_norm": 4.237308502197266, "learning_rate": 1.9724057579012026e-05, "loss": 0.16630399227142334, "step": 3900 }, { "epoch": 0.10299536059900627, "grad_norm": 2.2400238513946533, "learning_rate": 1.972206370501624e-05, "loss": 0.17254867553710937, "step": 3910 }, { "epoch": 0.103258775843505, "grad_norm": 3.9512953758239746, "learning_rate": 1.9720062754883048e-05, "loss": 0.17251510620117189, "step": 3920 }, { "epoch": 0.10352219108800374, "grad_norm": 2.048676013946533, "learning_rate": 1.971805473006882e-05, "loss": 0.1662726879119873, "step": 3930 }, { "epoch": 0.10378560633250247, "grad_norm": 2.684033155441284, "learning_rate": 1.9716039632035086e-05, "loss": 0.1604898452758789, "step": 3940 }, { "epoch": 0.10404902157700122, "grad_norm": 1.5240614414215088, "learning_rate": 1.9714017462248523e-05, "loss": 0.19755614995956422, "step": 3950 }, { "epoch": 0.10431243682149995, "grad_norm": 3.4563236236572266, "learning_rate": 1.9711988222180958e-05, "loss": 0.1829979181289673, "step": 3960 }, { "epoch": 0.10457585206599869, "grad_norm": 5.001173973083496, "learning_rate": 1.9709951913309347e-05, "loss": 0.18318467140197753, "step": 3970 }, { "epoch": 0.10483926731049743, "grad_norm": 2.2194933891296387, "learning_rate": 1.9707908537115816e-05, "loss": 0.17466239929199218, "step": 3980 }, { "epoch": 0.10510268255499616, "grad_norm": 1.9840394258499146, "learning_rate": 1.9705858095087613e-05, "loss": 0.1735537528991699, "step": 3990 }, { "epoch": 0.1053660977994949, "grad_norm": 2.423297643661499, "learning_rate": 1.9703800588717144e-05, "loss": 0.1590242862701416, "step": 4000 }, { "epoch": 0.10562951304399364, "grad_norm": 2.6404106616973877, "learning_rate": 1.9701736019501948e-05, "loss": 0.17462489604949952, "step": 4010 }, { "epoch": 0.10589292828849238, "grad_norm": 3.7710623741149902, "learning_rate": 1.969966438894471e-05, "loss": 0.16213595867156982, "step": 4020 }, { "epoch": 0.10615634353299111, "grad_norm": 2.8006043434143066, "learning_rate": 1.969758569855325e-05, "loss": 0.17167317867279053, "step": 4030 }, { "epoch": 0.10641975877748985, "grad_norm": 1.5705488920211792, "learning_rate": 1.969549994984053e-05, "loss": 0.1797211766242981, "step": 4040 }, { "epoch": 0.10668317402198858, "grad_norm": 3.8317694664001465, "learning_rate": 1.9693407144324642e-05, "loss": 0.1515257716178894, "step": 4050 }, { "epoch": 0.10694658926648733, "grad_norm": 3.341564893722534, "learning_rate": 1.9691307283528834e-05, "loss": 0.17537102699279786, "step": 4060 }, { "epoch": 0.10721000451098606, "grad_norm": 2.1940057277679443, "learning_rate": 1.968920036898146e-05, "loss": 0.17219345569610595, "step": 4070 }, { "epoch": 0.1074734197554848, "grad_norm": 2.5527126789093018, "learning_rate": 1.9687086402216033e-05, "loss": 0.1596426010131836, "step": 4080 }, { "epoch": 0.10773683499998353, "grad_norm": 1.3674613237380981, "learning_rate": 1.968496538477118e-05, "loss": 0.16047794818878175, "step": 4090 }, { "epoch": 0.10800025024448227, "grad_norm": 3.1536667346954346, "learning_rate": 1.9682837318190683e-05, "loss": 0.18563060760498046, "step": 4100 }, { "epoch": 0.10826366548898102, "grad_norm": 1.7109092473983765, "learning_rate": 1.968070220402343e-05, "loss": 0.14628292322158815, "step": 4110 }, { "epoch": 0.10852708073347975, "grad_norm": 1.1364326477050781, "learning_rate": 1.967856004382345e-05, "loss": 0.17049882411956788, "step": 4120 }, { "epoch": 0.10879049597797849, "grad_norm": 4.935472011566162, "learning_rate": 1.96764108391499e-05, "loss": 0.16335859298706054, "step": 4130 }, { "epoch": 0.10905391122247722, "grad_norm": 6.264809608459473, "learning_rate": 1.967425459156707e-05, "loss": 0.174141001701355, "step": 4140 }, { "epoch": 0.10931732646697596, "grad_norm": 1.5625287294387817, "learning_rate": 1.9672091302644363e-05, "loss": 0.16708141565322876, "step": 4150 }, { "epoch": 0.1095807417114747, "grad_norm": 5.640592575073242, "learning_rate": 1.966992097395632e-05, "loss": 0.17460529804229735, "step": 4160 }, { "epoch": 0.10984415695597344, "grad_norm": 2.007657051086426, "learning_rate": 1.9667743607082593e-05, "loss": 0.16625492572784423, "step": 4170 }, { "epoch": 0.11010757220047217, "grad_norm": 2.723097085952759, "learning_rate": 1.9665559203607965e-05, "loss": 0.16456278562545776, "step": 4180 }, { "epoch": 0.11037098744497091, "grad_norm": 1.736526370048523, "learning_rate": 1.9663367765122346e-05, "loss": 0.1693955898284912, "step": 4190 }, { "epoch": 0.11063440268946964, "grad_norm": 0.9378103613853455, "learning_rate": 1.966116929322075e-05, "loss": 0.17070314884185792, "step": 4200 }, { "epoch": 0.11089781793396838, "grad_norm": 2.7641422748565674, "learning_rate": 1.9658963789503328e-05, "loss": 0.16373908519744873, "step": 4210 }, { "epoch": 0.11116123317846711, "grad_norm": 1.891454815864563, "learning_rate": 1.9656751255575336e-05, "loss": 0.15315334796905516, "step": 4220 }, { "epoch": 0.11142464842296586, "grad_norm": 3.619718074798584, "learning_rate": 1.965453169304715e-05, "loss": 0.1720930814743042, "step": 4230 }, { "epoch": 0.1116880636674646, "grad_norm": 3.7225708961486816, "learning_rate": 1.9652305103534265e-05, "loss": 0.1745893955230713, "step": 4240 }, { "epoch": 0.11195147891196333, "grad_norm": 2.531562089920044, "learning_rate": 1.9650071488657293e-05, "loss": 0.16153115034103394, "step": 4250 }, { "epoch": 0.11221489415646207, "grad_norm": 1.6234118938446045, "learning_rate": 1.964783085004195e-05, "loss": 0.16619924306869507, "step": 4260 }, { "epoch": 0.1124783094009608, "grad_norm": 2.443927526473999, "learning_rate": 1.9645583189319074e-05, "loss": 0.1744059443473816, "step": 4270 }, { "epoch": 0.11274172464545955, "grad_norm": 5.106007099151611, "learning_rate": 1.9643328508124605e-05, "loss": 0.1728104829788208, "step": 4280 }, { "epoch": 0.11300513988995828, "grad_norm": 3.7937073707580566, "learning_rate": 1.9641066808099597e-05, "loss": 0.17519863843917846, "step": 4290 }, { "epoch": 0.11326855513445702, "grad_norm": 1.1012943983078003, "learning_rate": 1.963879809089022e-05, "loss": 0.1511667013168335, "step": 4300 }, { "epoch": 0.11353197037895575, "grad_norm": 5.187806606292725, "learning_rate": 1.9636522358147737e-05, "loss": 0.17602872848510742, "step": 4310 }, { "epoch": 0.1137953856234545, "grad_norm": 2.976146697998047, "learning_rate": 1.9634239611528527e-05, "loss": 0.1725107192993164, "step": 4320 }, { "epoch": 0.11405880086795323, "grad_norm": 1.9714077711105347, "learning_rate": 1.9631949852694073e-05, "loss": 0.17308764457702636, "step": 4330 }, { "epoch": 0.11432221611245197, "grad_norm": 2.735720157623291, "learning_rate": 1.9629653083310956e-05, "loss": 0.1773531198501587, "step": 4340 }, { "epoch": 0.1145856313569507, "grad_norm": 2.817998170852661, "learning_rate": 1.9627349305050867e-05, "loss": 0.1801053285598755, "step": 4350 }, { "epoch": 0.11484904660144944, "grad_norm": 4.15476131439209, "learning_rate": 1.9625038519590596e-05, "loss": 0.1796336889266968, "step": 4360 }, { "epoch": 0.11511246184594819, "grad_norm": 6.434434413909912, "learning_rate": 1.962272072861203e-05, "loss": 0.17991493940353392, "step": 4370 }, { "epoch": 0.11537587709044692, "grad_norm": 2.2948594093322754, "learning_rate": 1.9620395933802158e-05, "loss": 0.18441362380981446, "step": 4380 }, { "epoch": 0.11563929233494566, "grad_norm": 1.716952919960022, "learning_rate": 1.9618064136853066e-05, "loss": 0.16117144823074342, "step": 4390 }, { "epoch": 0.11590270757944439, "grad_norm": 3.094327211380005, "learning_rate": 1.9615725339461936e-05, "loss": 0.1718725562095642, "step": 4400 }, { "epoch": 0.11616612282394313, "grad_norm": 4.401452541351318, "learning_rate": 1.961337954333105e-05, "loss": 0.16403237581253052, "step": 4410 }, { "epoch": 0.11642953806844186, "grad_norm": 1.9396082162857056, "learning_rate": 1.961102675016777e-05, "loss": 0.16996617317199708, "step": 4420 }, { "epoch": 0.1166929533129406, "grad_norm": 3.168165683746338, "learning_rate": 1.960866696168457e-05, "loss": 0.15161622762680055, "step": 4430 }, { "epoch": 0.11695636855743934, "grad_norm": 3.8417882919311523, "learning_rate": 1.9606300179599004e-05, "loss": 0.15698529481887818, "step": 4440 }, { "epoch": 0.11721978380193808, "grad_norm": 2.6910195350646973, "learning_rate": 1.9603926405633712e-05, "loss": 0.17263619899749755, "step": 4450 }, { "epoch": 0.11748319904643681, "grad_norm": 1.7374567985534668, "learning_rate": 1.9601545641516434e-05, "loss": 0.17747807502746582, "step": 4460 }, { "epoch": 0.11774661429093555, "grad_norm": 3.3197102546691895, "learning_rate": 1.9599157888979997e-05, "loss": 0.16663045883178712, "step": 4470 }, { "epoch": 0.11801002953543428, "grad_norm": 1.6749063730239868, "learning_rate": 1.95967631497623e-05, "loss": 0.16504361629486083, "step": 4480 }, { "epoch": 0.11827344477993303, "grad_norm": 2.8819522857666016, "learning_rate": 1.9594361425606348e-05, "loss": 0.17393248081207274, "step": 4490 }, { "epoch": 0.11853686002443177, "grad_norm": 1.4097082614898682, "learning_rate": 1.959195271826021e-05, "loss": 0.16466038227081298, "step": 4500 }, { "epoch": 0.1188002752689305, "grad_norm": 4.058746814727783, "learning_rate": 1.9589537029477052e-05, "loss": 0.1562976837158203, "step": 4510 }, { "epoch": 0.11906369051342924, "grad_norm": 1.5180714130401611, "learning_rate": 1.9587114361015118e-05, "loss": 0.17397809028625488, "step": 4520 }, { "epoch": 0.11932710575792797, "grad_norm": 3.6789464950561523, "learning_rate": 1.958468471463773e-05, "loss": 0.16938257217407227, "step": 4530 }, { "epoch": 0.11959052100242672, "grad_norm": 3.823451280593872, "learning_rate": 1.9582248092113286e-05, "loss": 0.16858978271484376, "step": 4540 }, { "epoch": 0.11985393624692545, "grad_norm": 1.799371361732483, "learning_rate": 1.9579804495215268e-05, "loss": 0.15974942445755005, "step": 4550 }, { "epoch": 0.12011735149142419, "grad_norm": 1.364443302154541, "learning_rate": 1.9577353925722228e-05, "loss": 0.17236313819885254, "step": 4560 }, { "epoch": 0.12038076673592292, "grad_norm": 1.917953610420227, "learning_rate": 1.9574896385417803e-05, "loss": 0.17606046199798583, "step": 4570 }, { "epoch": 0.12064418198042166, "grad_norm": 1.899935245513916, "learning_rate": 1.957243187609069e-05, "loss": 0.17075786590576172, "step": 4580 }, { "epoch": 0.1209075972249204, "grad_norm": 2.75644850730896, "learning_rate": 1.9569960399534668e-05, "loss": 0.16586816310882568, "step": 4590 }, { "epoch": 0.12117101246941914, "grad_norm": 3.664863348007202, "learning_rate": 1.9567481957548585e-05, "loss": 0.17859050035476684, "step": 4600 }, { "epoch": 0.12143442771391787, "grad_norm": 4.368910789489746, "learning_rate": 1.9564996551936354e-05, "loss": 0.17064111232757567, "step": 4610 }, { "epoch": 0.12169784295841661, "grad_norm": 2.755300521850586, "learning_rate": 1.9562504184506966e-05, "loss": 0.16125308275222777, "step": 4620 }, { "epoch": 0.12196125820291535, "grad_norm": 2.968672752380371, "learning_rate": 1.9560004857074472e-05, "loss": 0.16630752086639405, "step": 4630 }, { "epoch": 0.12222467344741408, "grad_norm": 2.0359013080596924, "learning_rate": 1.955749857145799e-05, "loss": 0.1621622323989868, "step": 4640 }, { "epoch": 0.12248808869191283, "grad_norm": 2.3700320720672607, "learning_rate": 1.95549853294817e-05, "loss": 0.15967659950256347, "step": 4650 }, { "epoch": 0.12275150393641156, "grad_norm": 1.3813985586166382, "learning_rate": 1.955246513297485e-05, "loss": 0.17847855091094972, "step": 4660 }, { "epoch": 0.1230149191809103, "grad_norm": 1.694366216659546, "learning_rate": 1.9549937983771748e-05, "loss": 0.17692224979400634, "step": 4670 }, { "epoch": 0.12327833442540903, "grad_norm": 6.663546085357666, "learning_rate": 1.954740388371176e-05, "loss": 0.15727486610412597, "step": 4680 }, { "epoch": 0.12354174966990777, "grad_norm": 1.0934396982192993, "learning_rate": 1.9544862834639316e-05, "loss": 0.1624971628189087, "step": 4690 }, { "epoch": 0.1238051649144065, "grad_norm": 1.4697751998901367, "learning_rate": 1.9542314838403907e-05, "loss": 0.15474092960357666, "step": 4700 }, { "epoch": 0.12406858015890525, "grad_norm": 3.140092372894287, "learning_rate": 1.9539759896860064e-05, "loss": 0.1716754913330078, "step": 4710 }, { "epoch": 0.12433199540340398, "grad_norm": 4.630941390991211, "learning_rate": 1.9537198011867383e-05, "loss": 0.16436944007873536, "step": 4720 }, { "epoch": 0.12459541064790272, "grad_norm": 4.397977352142334, "learning_rate": 1.9534629185290524e-05, "loss": 0.1791585922241211, "step": 4730 }, { "epoch": 0.12485882589240146, "grad_norm": 2.7380034923553467, "learning_rate": 1.9532053418999186e-05, "loss": 0.17261731624603271, "step": 4740 }, { "epoch": 0.1251222411369002, "grad_norm": 3.922076463699341, "learning_rate": 1.952947071486812e-05, "loss": 0.18300895690917968, "step": 4750 }, { "epoch": 0.12538565638139892, "grad_norm": 2.7415411472320557, "learning_rate": 1.952688107477713e-05, "loss": 0.1580664873123169, "step": 4760 }, { "epoch": 0.12564907162589767, "grad_norm": 2.46980619430542, "learning_rate": 1.9524284500611076e-05, "loss": 0.17997801303863525, "step": 4770 }, { "epoch": 0.1259124868703964, "grad_norm": 2.3771560192108154, "learning_rate": 1.9521680994259845e-05, "loss": 0.17066535949707032, "step": 4780 }, { "epoch": 0.12617590211489516, "grad_norm": 2.5633952617645264, "learning_rate": 1.9519070557618386e-05, "loss": 0.1714002251625061, "step": 4790 }, { "epoch": 0.12643931735939387, "grad_norm": 1.4678040742874146, "learning_rate": 1.9516453192586686e-05, "loss": 0.1675640344619751, "step": 4800 }, { "epoch": 0.12670273260389261, "grad_norm": 2.1174073219299316, "learning_rate": 1.951382890106978e-05, "loss": 0.16410930156707765, "step": 4810 }, { "epoch": 0.12696614784839136, "grad_norm": 1.3770277500152588, "learning_rate": 1.951119768497774e-05, "loss": 0.18307852745056152, "step": 4820 }, { "epoch": 0.1272295630928901, "grad_norm": 2.019474506378174, "learning_rate": 1.9508559546225672e-05, "loss": 0.15929291248321534, "step": 4830 }, { "epoch": 0.12749297833738882, "grad_norm": 1.31596839427948, "learning_rate": 1.9505914486733735e-05, "loss": 0.16462233066558837, "step": 4840 }, { "epoch": 0.12775639358188756, "grad_norm": 1.2437026500701904, "learning_rate": 1.9503262508427115e-05, "loss": 0.17023252248764037, "step": 4850 }, { "epoch": 0.1280198088263863, "grad_norm": 2.4953904151916504, "learning_rate": 1.9500603613236034e-05, "loss": 0.16080286502838134, "step": 4860 }, { "epoch": 0.12828322407088505, "grad_norm": 1.2260075807571411, "learning_rate": 1.9497937803095756e-05, "loss": 0.17252910137176514, "step": 4870 }, { "epoch": 0.1285466393153838, "grad_norm": 4.724473476409912, "learning_rate": 1.9495265079946565e-05, "loss": 0.1647987961769104, "step": 4880 }, { "epoch": 0.1288100545598825, "grad_norm": 2.828563690185547, "learning_rate": 1.9492585445733788e-05, "loss": 0.1805588960647583, "step": 4890 }, { "epoch": 0.12907346980438125, "grad_norm": 4.332505226135254, "learning_rate": 1.9489898902407783e-05, "loss": 0.1749863862991333, "step": 4900 }, { "epoch": 0.12933688504888, "grad_norm": 3.754615068435669, "learning_rate": 1.9487205451923925e-05, "loss": 0.1706026554107666, "step": 4910 }, { "epoch": 0.12960030029337874, "grad_norm": 6.2855119705200195, "learning_rate": 1.9484505096242628e-05, "loss": 0.17561652660369872, "step": 4920 }, { "epoch": 0.12986371553787746, "grad_norm": 5.205380439758301, "learning_rate": 1.9481797837329323e-05, "loss": 0.16374801397323607, "step": 4930 }, { "epoch": 0.1301271307823762, "grad_norm": 1.6639430522918701, "learning_rate": 1.9479083677154474e-05, "loss": 0.16021920442581178, "step": 4940 }, { "epoch": 0.13039054602687494, "grad_norm": 1.002833366394043, "learning_rate": 1.9476362617693564e-05, "loss": 0.16798183917999268, "step": 4950 }, { "epoch": 0.1306539612713737, "grad_norm": 1.3035612106323242, "learning_rate": 1.9473634660927092e-05, "loss": 0.1746903419494629, "step": 4960 }, { "epoch": 0.1309173765158724, "grad_norm": 5.660372734069824, "learning_rate": 1.9470899808840585e-05, "loss": 0.17651717662811278, "step": 4970 }, { "epoch": 0.13118079176037115, "grad_norm": 2.307795286178589, "learning_rate": 1.9468158063424593e-05, "loss": 0.17581586837768554, "step": 4980 }, { "epoch": 0.1314442070048699, "grad_norm": 3.8826780319213867, "learning_rate": 1.946540942667467e-05, "loss": 0.16087687015533447, "step": 4990 }, { "epoch": 0.13170762224936863, "grad_norm": 0.9635523557662964, "learning_rate": 1.9462653900591393e-05, "loss": 0.1723293900489807, "step": 5000 }, { "epoch": 0.13197103749386738, "grad_norm": 2.5468146800994873, "learning_rate": 1.9459891487180353e-05, "loss": 0.1681812286376953, "step": 5010 }, { "epoch": 0.1322344527383661, "grad_norm": 1.1652864217758179, "learning_rate": 1.9457122188452156e-05, "loss": 0.15470050573348998, "step": 5020 }, { "epoch": 0.13249786798286484, "grad_norm": 4.119149208068848, "learning_rate": 1.9454346006422414e-05, "loss": 0.1811077356338501, "step": 5030 }, { "epoch": 0.13276128322736358, "grad_norm": 1.5191199779510498, "learning_rate": 1.9451562943111753e-05, "loss": 0.17443296909332276, "step": 5040 }, { "epoch": 0.13302469847186232, "grad_norm": 1.9579966068267822, "learning_rate": 1.944877300054581e-05, "loss": 0.1791300058364868, "step": 5050 }, { "epoch": 0.13328811371636104, "grad_norm": 1.784654974937439, "learning_rate": 1.944597618075522e-05, "loss": 0.16384587287902833, "step": 5060 }, { "epoch": 0.13355152896085978, "grad_norm": 1.756259799003601, "learning_rate": 1.944317248577563e-05, "loss": 0.15236151218414307, "step": 5070 }, { "epoch": 0.13381494420535853, "grad_norm": 2.4345130920410156, "learning_rate": 1.9440361917647693e-05, "loss": 0.1738606572151184, "step": 5080 }, { "epoch": 0.13407835944985727, "grad_norm": 1.7232661247253418, "learning_rate": 1.943754447841706e-05, "loss": 0.1531437873840332, "step": 5090 }, { "epoch": 0.134341774694356, "grad_norm": 3.7084317207336426, "learning_rate": 1.9434720170134387e-05, "loss": 0.16825637817382813, "step": 5100 }, { "epoch": 0.13460518993885473, "grad_norm": 2.845515727996826, "learning_rate": 1.9431888994855324e-05, "loss": 0.17757031917572022, "step": 5110 }, { "epoch": 0.13486860518335347, "grad_norm": 3.403524398803711, "learning_rate": 1.9429050954640528e-05, "loss": 0.1656688332557678, "step": 5120 }, { "epoch": 0.13513202042785222, "grad_norm": 4.169882774353027, "learning_rate": 1.942620605155564e-05, "loss": 0.16367547512054442, "step": 5130 }, { "epoch": 0.13539543567235096, "grad_norm": 1.3590176105499268, "learning_rate": 1.9423354287671307e-05, "loss": 0.17250083684921264, "step": 5140 }, { "epoch": 0.13565885091684968, "grad_norm": 2.1340010166168213, "learning_rate": 1.9420495665063166e-05, "loss": 0.16274117231369017, "step": 5150 }, { "epoch": 0.13592226616134842, "grad_norm": 1.4497123956680298, "learning_rate": 1.9417630185811846e-05, "loss": 0.1647856593132019, "step": 5160 }, { "epoch": 0.13618568140584716, "grad_norm": 2.862107515335083, "learning_rate": 1.9414757852002962e-05, "loss": 0.18299010992050171, "step": 5170 }, { "epoch": 0.1364490966503459, "grad_norm": 2.8362410068511963, "learning_rate": 1.941187866572713e-05, "loss": 0.16643636226654052, "step": 5180 }, { "epoch": 0.13671251189484462, "grad_norm": 1.1611926555633545, "learning_rate": 1.940899262907994e-05, "loss": 0.1715112805366516, "step": 5190 }, { "epoch": 0.13697592713934337, "grad_norm": 2.940190076828003, "learning_rate": 1.940609974416198e-05, "loss": 0.167623770236969, "step": 5200 }, { "epoch": 0.1372393423838421, "grad_norm": 7.544941425323486, "learning_rate": 1.940320001307881e-05, "loss": 0.17074329853057862, "step": 5210 }, { "epoch": 0.13750275762834085, "grad_norm": 3.0669169425964355, "learning_rate": 1.940029343794098e-05, "loss": 0.17722322940826415, "step": 5220 }, { "epoch": 0.13776617287283957, "grad_norm": 1.315445899963379, "learning_rate": 1.9397380020864024e-05, "loss": 0.17317724227905273, "step": 5230 }, { "epoch": 0.13802958811733831, "grad_norm": 4.046011924743652, "learning_rate": 1.939445976396845e-05, "loss": 0.167444109916687, "step": 5240 }, { "epoch": 0.13829300336183706, "grad_norm": 1.9847060441970825, "learning_rate": 1.939153266937975e-05, "loss": 0.1726990222930908, "step": 5250 }, { "epoch": 0.1385564186063358, "grad_norm": 2.7359514236450195, "learning_rate": 1.938859873922839e-05, "loss": 0.17543762922286987, "step": 5260 }, { "epoch": 0.13881983385083455, "grad_norm": 2.2651283740997314, "learning_rate": 1.9385657975649807e-05, "loss": 0.16991454362869263, "step": 5270 }, { "epoch": 0.13908324909533326, "grad_norm": 1.5934628248214722, "learning_rate": 1.9382710380784418e-05, "loss": 0.16259206533432008, "step": 5280 }, { "epoch": 0.139346664339832, "grad_norm": 1.7886736392974854, "learning_rate": 1.937975595677761e-05, "loss": 0.16409413814544677, "step": 5290 }, { "epoch": 0.13961007958433075, "grad_norm": 2.799224615097046, "learning_rate": 1.9376794705779735e-05, "loss": 0.16521065235137938, "step": 5300 }, { "epoch": 0.1398734948288295, "grad_norm": 1.217618465423584, "learning_rate": 1.937382662994613e-05, "loss": 0.16728705167770386, "step": 5310 }, { "epoch": 0.1401369100733282, "grad_norm": 2.2884161472320557, "learning_rate": 1.9370851731437077e-05, "loss": 0.16949846744537353, "step": 5320 }, { "epoch": 0.14040032531782695, "grad_norm": 6.244661808013916, "learning_rate": 1.9367870012417844e-05, "loss": 0.17415452003479004, "step": 5330 }, { "epoch": 0.1406637405623257, "grad_norm": 1.135058045387268, "learning_rate": 1.936488147505865e-05, "loss": 0.1723649501800537, "step": 5340 }, { "epoch": 0.14092715580682444, "grad_norm": 3.7817881107330322, "learning_rate": 1.9361886121534686e-05, "loss": 0.1619960069656372, "step": 5350 }, { "epoch": 0.14119057105132315, "grad_norm": 1.3398606777191162, "learning_rate": 1.93588839540261e-05, "loss": 0.1680556893348694, "step": 5360 }, { "epoch": 0.1414539862958219, "grad_norm": 1.9170235395431519, "learning_rate": 1.9355874974717988e-05, "loss": 0.1630922555923462, "step": 5370 }, { "epoch": 0.14171740154032064, "grad_norm": 1.562071681022644, "learning_rate": 1.9352859185800424e-05, "loss": 0.15708152055740357, "step": 5380 }, { "epoch": 0.14198081678481939, "grad_norm": 1.2009748220443726, "learning_rate": 1.934983658946843e-05, "loss": 0.17363381385803223, "step": 5390 }, { "epoch": 0.14224423202931813, "grad_norm": 2.6904850006103516, "learning_rate": 1.9346807187921983e-05, "loss": 0.16279950141906738, "step": 5400 }, { "epoch": 0.14250764727381685, "grad_norm": 6.496646881103516, "learning_rate": 1.9343770983366004e-05, "loss": 0.16125898361206054, "step": 5410 }, { "epoch": 0.1427710625183156, "grad_norm": 1.8062407970428467, "learning_rate": 1.9340727978010377e-05, "loss": 0.17494983673095704, "step": 5420 }, { "epoch": 0.14303447776281433, "grad_norm": 2.3372535705566406, "learning_rate": 1.933767817406994e-05, "loss": 0.1709280252456665, "step": 5430 }, { "epoch": 0.14329789300731308, "grad_norm": 3.4164671897888184, "learning_rate": 1.9334621573764462e-05, "loss": 0.1679758310317993, "step": 5440 }, { "epoch": 0.1435613082518118, "grad_norm": 2.3904166221618652, "learning_rate": 1.9331558179318676e-05, "loss": 0.16845502853393554, "step": 5450 }, { "epoch": 0.14382472349631054, "grad_norm": 3.350797653198242, "learning_rate": 1.9328487992962247e-05, "loss": 0.17083832025527954, "step": 5460 }, { "epoch": 0.14408813874080928, "grad_norm": 2.467726469039917, "learning_rate": 1.9325411016929793e-05, "loss": 0.1721234440803528, "step": 5470 }, { "epoch": 0.14435155398530802, "grad_norm": 1.0883634090423584, "learning_rate": 1.9322327253460867e-05, "loss": 0.16955227851867677, "step": 5480 }, { "epoch": 0.14461496922980674, "grad_norm": 2.748567581176758, "learning_rate": 1.931923670479997e-05, "loss": 0.1588674545288086, "step": 5490 }, { "epoch": 0.14487838447430548, "grad_norm": 1.244309663772583, "learning_rate": 1.9316139373196533e-05, "loss": 0.1657163381576538, "step": 5500 }, { "epoch": 0.14514179971880423, "grad_norm": 2.607403039932251, "learning_rate": 1.9313035260904923e-05, "loss": 0.17295286655426026, "step": 5510 }, { "epoch": 0.14540521496330297, "grad_norm": 1.8630870580673218, "learning_rate": 1.9309924370184455e-05, "loss": 0.17081022262573242, "step": 5520 }, { "epoch": 0.1456686302078017, "grad_norm": 4.203898906707764, "learning_rate": 1.9306806703299367e-05, "loss": 0.16988623142242432, "step": 5530 }, { "epoch": 0.14593204545230043, "grad_norm": 2.3426015377044678, "learning_rate": 1.9303682262518828e-05, "loss": 0.17603561878204346, "step": 5540 }, { "epoch": 0.14619546069679917, "grad_norm": 2.0784380435943604, "learning_rate": 1.9300551050116946e-05, "loss": 0.16761014461517335, "step": 5550 }, { "epoch": 0.14645887594129792, "grad_norm": 3.3396308422088623, "learning_rate": 1.9297413068372745e-05, "loss": 0.1641085624694824, "step": 5560 }, { "epoch": 0.14672229118579666, "grad_norm": 4.122959613800049, "learning_rate": 1.9294268319570195e-05, "loss": 0.1632613182067871, "step": 5570 }, { "epoch": 0.14698570643029538, "grad_norm": 3.348757743835449, "learning_rate": 1.9291116805998168e-05, "loss": 0.18270522356033325, "step": 5580 }, { "epoch": 0.14724912167479412, "grad_norm": 1.2395871877670288, "learning_rate": 1.9287958529950477e-05, "loss": 0.16460738182067872, "step": 5590 }, { "epoch": 0.14751253691929286, "grad_norm": 3.166311264038086, "learning_rate": 1.9284793493725852e-05, "loss": 0.1722532868385315, "step": 5600 }, { "epoch": 0.1477759521637916, "grad_norm": 4.346956253051758, "learning_rate": 1.9281621699627943e-05, "loss": 0.17357189655303956, "step": 5610 }, { "epoch": 0.14803936740829032, "grad_norm": 1.8267149925231934, "learning_rate": 1.927844314996532e-05, "loss": 0.1562258243560791, "step": 5620 }, { "epoch": 0.14830278265278907, "grad_norm": 3.4600636959075928, "learning_rate": 1.9275257847051464e-05, "loss": 0.1645805835723877, "step": 5630 }, { "epoch": 0.1485661978972878, "grad_norm": 4.831007480621338, "learning_rate": 1.9272065793204778e-05, "loss": 0.16408077478408814, "step": 5640 }, { "epoch": 0.14882961314178655, "grad_norm": 4.824771881103516, "learning_rate": 1.9268866990748578e-05, "loss": 0.1650339722633362, "step": 5650 }, { "epoch": 0.1490930283862853, "grad_norm": 2.1462271213531494, "learning_rate": 1.9265661442011087e-05, "loss": 0.18420054912567138, "step": 5660 }, { "epoch": 0.149356443630784, "grad_norm": 7.353031158447266, "learning_rate": 1.9262449149325445e-05, "loss": 0.18497102260589598, "step": 5670 }, { "epoch": 0.14961985887528276, "grad_norm": 0.8760023713111877, "learning_rate": 1.92592301150297e-05, "loss": 0.1646502733230591, "step": 5680 }, { "epoch": 0.1498832741197815, "grad_norm": 1.6395844221115112, "learning_rate": 1.9256004341466795e-05, "loss": 0.16418051719665527, "step": 5690 }, { "epoch": 0.15014668936428024, "grad_norm": 3.8214783668518066, "learning_rate": 1.9252771830984594e-05, "loss": 0.15435153245925903, "step": 5700 }, { "epoch": 0.15041010460877896, "grad_norm": 7.499322891235352, "learning_rate": 1.9249532585935855e-05, "loss": 0.16541242599487305, "step": 5710 }, { "epoch": 0.1506735198532777, "grad_norm": 6.288152694702148, "learning_rate": 1.9246286608678243e-05, "loss": 0.16635048389434814, "step": 5720 }, { "epoch": 0.15093693509777645, "grad_norm": 3.4717178344726562, "learning_rate": 1.9243033901574316e-05, "loss": 0.17292351722717286, "step": 5730 }, { "epoch": 0.1512003503422752, "grad_norm": 5.838156223297119, "learning_rate": 1.923977446699154e-05, "loss": 0.1772749662399292, "step": 5740 }, { "epoch": 0.1514637655867739, "grad_norm": 5.276182651519775, "learning_rate": 1.923650830730227e-05, "loss": 0.16207895278930665, "step": 5750 }, { "epoch": 0.15172718083127265, "grad_norm": 4.675018787384033, "learning_rate": 1.923323542488375e-05, "loss": 0.1630285859107971, "step": 5760 }, { "epoch": 0.1519905960757714, "grad_norm": 1.0229389667510986, "learning_rate": 1.9229955822118138e-05, "loss": 0.16536058187484742, "step": 5770 }, { "epoch": 0.15225401132027014, "grad_norm": 2.326789617538452, "learning_rate": 1.9226669501392465e-05, "loss": 0.16546292304992677, "step": 5780 }, { "epoch": 0.15251742656476888, "grad_norm": 2.10243821144104, "learning_rate": 1.9223376465098657e-05, "loss": 0.1615067958831787, "step": 5790 }, { "epoch": 0.1527808418092676, "grad_norm": 8.608362197875977, "learning_rate": 1.9220076715633526e-05, "loss": 0.17150568962097168, "step": 5800 }, { "epoch": 0.15304425705376634, "grad_norm": 4.207789897918701, "learning_rate": 1.921677025539877e-05, "loss": 0.16899211406707765, "step": 5810 }, { "epoch": 0.15330767229826509, "grad_norm": 1.1099334955215454, "learning_rate": 1.921345708680098e-05, "loss": 0.16218099594116211, "step": 5820 }, { "epoch": 0.15357108754276383, "grad_norm": 2.0995967388153076, "learning_rate": 1.921013721225162e-05, "loss": 0.16815612316131592, "step": 5830 }, { "epoch": 0.15383450278726254, "grad_norm": 4.759561538696289, "learning_rate": 1.9206810634167036e-05, "loss": 0.1559740662574768, "step": 5840 }, { "epoch": 0.1540979180317613, "grad_norm": 2.2096645832061768, "learning_rate": 1.920347735496846e-05, "loss": 0.15946033000946044, "step": 5850 }, { "epoch": 0.15436133327626003, "grad_norm": 1.2605345249176025, "learning_rate": 1.9200137377081993e-05, "loss": 0.16562517881393432, "step": 5860 }, { "epoch": 0.15462474852075878, "grad_norm": 1.972827434539795, "learning_rate": 1.919679070293861e-05, "loss": 0.1636871099472046, "step": 5870 }, { "epoch": 0.1548881637652575, "grad_norm": 3.7144620418548584, "learning_rate": 1.9193437334974178e-05, "loss": 0.15798656940460204, "step": 5880 }, { "epoch": 0.15515157900975624, "grad_norm": 4.763118743896484, "learning_rate": 1.9190077275629408e-05, "loss": 0.15902175903320312, "step": 5890 }, { "epoch": 0.15541499425425498, "grad_norm": 0.9978688359260559, "learning_rate": 1.9186710527349907e-05, "loss": 0.16240688562393188, "step": 5900 }, { "epoch": 0.15567840949875372, "grad_norm": 5.250338077545166, "learning_rate": 1.9183337092586137e-05, "loss": 0.16301229000091552, "step": 5910 }, { "epoch": 0.15594182474325247, "grad_norm": 1.8666775226593018, "learning_rate": 1.9179956973793422e-05, "loss": 0.1643809676170349, "step": 5920 }, { "epoch": 0.15620523998775118, "grad_norm": 1.0240880250930786, "learning_rate": 1.917657017343197e-05, "loss": 0.17567923069000244, "step": 5930 }, { "epoch": 0.15646865523224993, "grad_norm": 1.5523172616958618, "learning_rate": 1.9173176693966835e-05, "loss": 0.172146213054657, "step": 5940 }, { "epoch": 0.15673207047674867, "grad_norm": 3.0212604999542236, "learning_rate": 1.9169776537867936e-05, "loss": 0.1709882378578186, "step": 5950 }, { "epoch": 0.1569954857212474, "grad_norm": 2.4823358058929443, "learning_rate": 1.9166369707610055e-05, "loss": 0.1613013505935669, "step": 5960 }, { "epoch": 0.15725890096574613, "grad_norm": 2.0636720657348633, "learning_rate": 1.9162956205672833e-05, "loss": 0.17039374113082886, "step": 5970 }, { "epoch": 0.15752231621024487, "grad_norm": 1.2038991451263428, "learning_rate": 1.9159536034540757e-05, "loss": 0.16704468727111815, "step": 5980 }, { "epoch": 0.15778573145474362, "grad_norm": 2.4136595726013184, "learning_rate": 1.9156109196703177e-05, "loss": 0.16952641010284425, "step": 5990 }, { "epoch": 0.15804914669924236, "grad_norm": 1.6050562858581543, "learning_rate": 1.91526756946543e-05, "loss": 0.16955113410949707, "step": 6000 }, { "epoch": 0.15831256194374108, "grad_norm": 6.0465593338012695, "learning_rate": 1.9149235530893172e-05, "loss": 0.1659196972846985, "step": 6010 }, { "epoch": 0.15857597718823982, "grad_norm": 2.69038462638855, "learning_rate": 1.9145788707923692e-05, "loss": 0.17757880687713623, "step": 6020 }, { "epoch": 0.15883939243273856, "grad_norm": 1.4113954305648804, "learning_rate": 1.914233522825461e-05, "loss": 0.16365833282470704, "step": 6030 }, { "epoch": 0.1591028076772373, "grad_norm": 2.93098783493042, "learning_rate": 1.913887509439951e-05, "loss": 0.16749119758605957, "step": 6040 }, { "epoch": 0.15936622292173605, "grad_norm": 4.533895492553711, "learning_rate": 1.9135408308876834e-05, "loss": 0.1684796094894409, "step": 6050 }, { "epoch": 0.15962963816623477, "grad_norm": 2.096957206726074, "learning_rate": 1.913193487420986e-05, "loss": 0.1571810483932495, "step": 6060 }, { "epoch": 0.1598930534107335, "grad_norm": 1.3056650161743164, "learning_rate": 1.9128454792926694e-05, "loss": 0.16199294328689576, "step": 6070 }, { "epoch": 0.16015646865523225, "grad_norm": 4.694887161254883, "learning_rate": 1.9124968067560296e-05, "loss": 0.16734968423843383, "step": 6080 }, { "epoch": 0.160419883899731, "grad_norm": 5.076507568359375, "learning_rate": 1.9121474700648454e-05, "loss": 0.16640456914901733, "step": 6090 }, { "epoch": 0.1606832991442297, "grad_norm": 3.4485409259796143, "learning_rate": 1.9117974694733795e-05, "loss": 0.16535412073135375, "step": 6100 }, { "epoch": 0.16094671438872846, "grad_norm": 2.3967959880828857, "learning_rate": 1.9114468052363764e-05, "loss": 0.17386319637298583, "step": 6110 }, { "epoch": 0.1612101296332272, "grad_norm": 1.271723747253418, "learning_rate": 1.9110954776090656e-05, "loss": 0.17437417507171632, "step": 6120 }, { "epoch": 0.16147354487772594, "grad_norm": 3.0450174808502197, "learning_rate": 1.910743486847158e-05, "loss": 0.15961699485778807, "step": 6130 }, { "epoch": 0.16173696012222466, "grad_norm": 1.6759530305862427, "learning_rate": 1.9103908332068483e-05, "loss": 0.1741499662399292, "step": 6140 }, { "epoch": 0.1620003753667234, "grad_norm": 3.514509677886963, "learning_rate": 1.910037516944812e-05, "loss": 0.1535240888595581, "step": 6150 }, { "epoch": 0.16226379061122215, "grad_norm": 3.7204718589782715, "learning_rate": 1.909683538318209e-05, "loss": 0.1588594913482666, "step": 6160 }, { "epoch": 0.1625272058557209, "grad_norm": 3.0800704956054688, "learning_rate": 1.9093288975846797e-05, "loss": 0.15705790519714355, "step": 6170 }, { "epoch": 0.16279062110021963, "grad_norm": 2.1745901107788086, "learning_rate": 1.9089735950023463e-05, "loss": 0.1671543836593628, "step": 6180 }, { "epoch": 0.16305403634471835, "grad_norm": 0.8811078071594238, "learning_rate": 1.9086176308298147e-05, "loss": 0.16637351512908935, "step": 6190 }, { "epoch": 0.1633174515892171, "grad_norm": 7.8882551193237305, "learning_rate": 1.90826100532617e-05, "loss": 0.17948020696640016, "step": 6200 }, { "epoch": 0.16358086683371584, "grad_norm": 2.865544319152832, "learning_rate": 1.90790371875098e-05, "loss": 0.16675518751144408, "step": 6210 }, { "epoch": 0.16384428207821458, "grad_norm": 1.265760898590088, "learning_rate": 1.9075457713642934e-05, "loss": 0.17425847053527832, "step": 6220 }, { "epoch": 0.1641076973227133, "grad_norm": 1.4732950925827026, "learning_rate": 1.9071871634266392e-05, "loss": 0.15977346897125244, "step": 6230 }, { "epoch": 0.16437111256721204, "grad_norm": 1.5458706617355347, "learning_rate": 1.9068278951990282e-05, "loss": 0.15775830745697023, "step": 6240 }, { "epoch": 0.16463452781171078, "grad_norm": 5.577322483062744, "learning_rate": 1.9064679669429515e-05, "loss": 0.16669113636016847, "step": 6250 }, { "epoch": 0.16489794305620953, "grad_norm": 2.76534104347229, "learning_rate": 1.9061073789203798e-05, "loss": 0.16864652633666993, "step": 6260 }, { "epoch": 0.16516135830070824, "grad_norm": 1.309943437576294, "learning_rate": 1.9057461313937646e-05, "loss": 0.18125952482223512, "step": 6270 }, { "epoch": 0.165424773545207, "grad_norm": 2.682126998901367, "learning_rate": 1.905384224626038e-05, "loss": 0.1760707139968872, "step": 6280 }, { "epoch": 0.16568818878970573, "grad_norm": 4.395334720611572, "learning_rate": 1.9050216588806113e-05, "loss": 0.16801165342330932, "step": 6290 }, { "epoch": 0.16595160403420448, "grad_norm": 3.912828207015991, "learning_rate": 1.9046584344213754e-05, "loss": 0.1537899374961853, "step": 6300 }, { "epoch": 0.16621501927870322, "grad_norm": 2.629648208618164, "learning_rate": 1.9042945515127004e-05, "loss": 0.156956684589386, "step": 6310 }, { "epoch": 0.16647843452320193, "grad_norm": 2.2210047245025635, "learning_rate": 1.9039300104194366e-05, "loss": 0.15356192588806153, "step": 6320 }, { "epoch": 0.16674184976770068, "grad_norm": 3.366645097732544, "learning_rate": 1.903564811406912e-05, "loss": 0.17151384353637694, "step": 6330 }, { "epoch": 0.16700526501219942, "grad_norm": 1.9660518169403076, "learning_rate": 1.9031989547409347e-05, "loss": 0.15803225040435792, "step": 6340 }, { "epoch": 0.16726868025669817, "grad_norm": 1.0356597900390625, "learning_rate": 1.9028324406877905e-05, "loss": 0.1524649143218994, "step": 6350 }, { "epoch": 0.16753209550119688, "grad_norm": 2.006226062774658, "learning_rate": 1.9024652695142447e-05, "loss": 0.1473541259765625, "step": 6360 }, { "epoch": 0.16779551074569563, "grad_norm": 1.420190691947937, "learning_rate": 1.90209744148754e-05, "loss": 0.16662172079086304, "step": 6370 }, { "epoch": 0.16805892599019437, "grad_norm": 3.3143184185028076, "learning_rate": 1.901728956875398e-05, "loss": 0.15683668851852417, "step": 6380 }, { "epoch": 0.1683223412346931, "grad_norm": 2.2023041248321533, "learning_rate": 1.9013598159460168e-05, "loss": 0.15049574375152588, "step": 6390 }, { "epoch": 0.16858575647919183, "grad_norm": 5.163723468780518, "learning_rate": 1.900990018968074e-05, "loss": 0.15863507986068726, "step": 6400 }, { "epoch": 0.16884917172369057, "grad_norm": 2.3111484050750732, "learning_rate": 1.900619566210723e-05, "loss": 0.1488100051879883, "step": 6410 }, { "epoch": 0.16911258696818932, "grad_norm": 1.0933449268341064, "learning_rate": 1.900248457943596e-05, "loss": 0.15391716957092286, "step": 6420 }, { "epoch": 0.16937600221268806, "grad_norm": 4.957375526428223, "learning_rate": 1.899876694436801e-05, "loss": 0.16747313737869263, "step": 6430 }, { "epoch": 0.1696394174571868, "grad_norm": 2.2285051345825195, "learning_rate": 1.899504275960924e-05, "loss": 0.16374616622924804, "step": 6440 }, { "epoch": 0.16990283270168552, "grad_norm": 4.741170406341553, "learning_rate": 1.899131202787027e-05, "loss": 0.16888251304626464, "step": 6450 }, { "epoch": 0.17016624794618426, "grad_norm": 1.5322903394699097, "learning_rate": 1.8987574751866488e-05, "loss": 0.18181968927383424, "step": 6460 }, { "epoch": 0.170429663190683, "grad_norm": 6.601553916931152, "learning_rate": 1.8983830934318044e-05, "loss": 0.17504862546920777, "step": 6470 }, { "epoch": 0.17069307843518175, "grad_norm": 4.164116382598877, "learning_rate": 1.8980080577949848e-05, "loss": 0.17404441833496093, "step": 6480 }, { "epoch": 0.17095649367968047, "grad_norm": 4.737561225891113, "learning_rate": 1.8976323685491577e-05, "loss": 0.1752851724624634, "step": 6490 }, { "epoch": 0.1712199089241792, "grad_norm": 2.9165279865264893, "learning_rate": 1.8972560259677658e-05, "loss": 0.17901648283004762, "step": 6500 }, { "epoch": 0.17148332416867795, "grad_norm": 6.996109485626221, "learning_rate": 1.896879030324727e-05, "loss": 0.16466951370239258, "step": 6510 }, { "epoch": 0.1717467394131767, "grad_norm": 1.863609790802002, "learning_rate": 1.8965013818944357e-05, "loss": 0.17026636600494385, "step": 6520 }, { "epoch": 0.1720101546576754, "grad_norm": 2.1932358741760254, "learning_rate": 1.89612308095176e-05, "loss": 0.1689756393432617, "step": 6530 }, { "epoch": 0.17227356990217416, "grad_norm": 2.4492695331573486, "learning_rate": 1.8957441277720438e-05, "loss": 0.17321479320526123, "step": 6540 }, { "epoch": 0.1725369851466729, "grad_norm": 1.176882028579712, "learning_rate": 1.8953645226311056e-05, "loss": 0.15681705474853516, "step": 6550 }, { "epoch": 0.17280040039117164, "grad_norm": 1.6415830850601196, "learning_rate": 1.894984265805239e-05, "loss": 0.17232871055603027, "step": 6560 }, { "epoch": 0.1730638156356704, "grad_norm": 1.6631789207458496, "learning_rate": 1.8946033575712104e-05, "loss": 0.1560147762298584, "step": 6570 }, { "epoch": 0.1733272308801691, "grad_norm": 1.1470301151275635, "learning_rate": 1.894221798206262e-05, "loss": 0.16622209548950195, "step": 6580 }, { "epoch": 0.17359064612466785, "grad_norm": 3.531216859817505, "learning_rate": 1.8938395879881086e-05, "loss": 0.1711844801902771, "step": 6590 }, { "epoch": 0.1738540613691666, "grad_norm": 1.5197385549545288, "learning_rate": 1.8934567271949397e-05, "loss": 0.17712302207946778, "step": 6600 }, { "epoch": 0.17411747661366533, "grad_norm": 2.1603918075561523, "learning_rate": 1.8930732161054176e-05, "loss": 0.15516217947006225, "step": 6610 }, { "epoch": 0.17438089185816405, "grad_norm": 4.891586780548096, "learning_rate": 1.8926890549986786e-05, "loss": 0.1632138729095459, "step": 6620 }, { "epoch": 0.1746443071026628, "grad_norm": 1.7992910146713257, "learning_rate": 1.8923042441543313e-05, "loss": 0.16551499366760253, "step": 6630 }, { "epoch": 0.17490772234716154, "grad_norm": 5.790839195251465, "learning_rate": 1.891918783852458e-05, "loss": 0.1729496955871582, "step": 6640 }, { "epoch": 0.17517113759166028, "grad_norm": 1.205173134803772, "learning_rate": 1.8915326743736134e-05, "loss": 0.14551281929016113, "step": 6650 }, { "epoch": 0.175434552836159, "grad_norm": 5.556744575500488, "learning_rate": 1.8911459159988238e-05, "loss": 0.1747755527496338, "step": 6660 }, { "epoch": 0.17569796808065774, "grad_norm": 2.3823609352111816, "learning_rate": 1.8907585090095897e-05, "loss": 0.16585732698440553, "step": 6670 }, { "epoch": 0.17596138332515648, "grad_norm": 4.007838249206543, "learning_rate": 1.890370453687882e-05, "loss": 0.16486895084381104, "step": 6680 }, { "epoch": 0.17622479856965523, "grad_norm": 4.651484966278076, "learning_rate": 1.889981750316144e-05, "loss": 0.1717075824737549, "step": 6690 }, { "epoch": 0.17648821381415397, "grad_norm": 1.4820616245269775, "learning_rate": 1.8895923991772913e-05, "loss": 0.16815788745880128, "step": 6700 }, { "epoch": 0.1767516290586527, "grad_norm": 1.2984338998794556, "learning_rate": 1.8892024005547104e-05, "loss": 0.1782602548599243, "step": 6710 }, { "epoch": 0.17701504430315143, "grad_norm": 5.1518449783325195, "learning_rate": 1.8888117547322586e-05, "loss": 0.15981152057647705, "step": 6720 }, { "epoch": 0.17727845954765017, "grad_norm": 2.176992416381836, "learning_rate": 1.888420461994265e-05, "loss": 0.1671803116798401, "step": 6730 }, { "epoch": 0.17754187479214892, "grad_norm": 4.293085098266602, "learning_rate": 1.88802852262553e-05, "loss": 0.16184558868408203, "step": 6740 }, { "epoch": 0.17780529003664763, "grad_norm": 3.9315717220306396, "learning_rate": 1.887635936911323e-05, "loss": 0.17729830741882324, "step": 6750 }, { "epoch": 0.17806870528114638, "grad_norm": 1.4514373540878296, "learning_rate": 1.8872427051373852e-05, "loss": 0.1660056471824646, "step": 6760 }, { "epoch": 0.17833212052564512, "grad_norm": 1.4763824939727783, "learning_rate": 1.886848827589928e-05, "loss": 0.1606285572052002, "step": 6770 }, { "epoch": 0.17859553577014387, "grad_norm": 2.770839214324951, "learning_rate": 1.8864543045556318e-05, "loss": 0.17145537137985228, "step": 6780 }, { "epoch": 0.1788589510146426, "grad_norm": 7.222318649291992, "learning_rate": 1.8860591363216482e-05, "loss": 0.17399613857269286, "step": 6790 }, { "epoch": 0.17912236625914132, "grad_norm": 1.7482091188430786, "learning_rate": 1.8856633231755972e-05, "loss": 0.1677156686782837, "step": 6800 }, { "epoch": 0.17938578150364007, "grad_norm": 1.7039234638214111, "learning_rate": 1.8852668654055686e-05, "loss": 0.1682279109954834, "step": 6810 }, { "epoch": 0.1796491967481388, "grad_norm": 2.845111131668091, "learning_rate": 1.884869763300122e-05, "loss": 0.16683454513549806, "step": 6820 }, { "epoch": 0.17991261199263756, "grad_norm": 1.9796345233917236, "learning_rate": 1.8844720171482854e-05, "loss": 0.15973427295684814, "step": 6830 }, { "epoch": 0.18017602723713627, "grad_norm": 2.467048168182373, "learning_rate": 1.884073627239555e-05, "loss": 0.16721432209014891, "step": 6840 }, { "epoch": 0.18043944248163502, "grad_norm": 3.8423800468444824, "learning_rate": 1.8836745938638963e-05, "loss": 0.1693597197532654, "step": 6850 }, { "epoch": 0.18070285772613376, "grad_norm": 1.564354419708252, "learning_rate": 1.8832749173117435e-05, "loss": 0.17012400627136232, "step": 6860 }, { "epoch": 0.1809662729706325, "grad_norm": 3.046687602996826, "learning_rate": 1.8828745978739982e-05, "loss": 0.16263189315795898, "step": 6870 }, { "epoch": 0.18122968821513122, "grad_norm": 0.8588241934776306, "learning_rate": 1.88247363584203e-05, "loss": 0.1680288553237915, "step": 6880 }, { "epoch": 0.18149310345962996, "grad_norm": 4.3904266357421875, "learning_rate": 1.8820720315076767e-05, "loss": 0.1639777898788452, "step": 6890 }, { "epoch": 0.1817565187041287, "grad_norm": 2.614952564239502, "learning_rate": 1.8816697851632425e-05, "loss": 0.1756368398666382, "step": 6900 }, { "epoch": 0.18201993394862745, "grad_norm": 2.1330161094665527, "learning_rate": 1.8812668971015006e-05, "loss": 0.17072802782058716, "step": 6910 }, { "epoch": 0.1822833491931262, "grad_norm": 1.8449633121490479, "learning_rate": 1.88086336761569e-05, "loss": 0.1645757555961609, "step": 6920 }, { "epoch": 0.1825467644376249, "grad_norm": 1.1175785064697266, "learning_rate": 1.8804591969995167e-05, "loss": 0.1686408281326294, "step": 6930 }, { "epoch": 0.18281017968212365, "grad_norm": 2.7015438079833984, "learning_rate": 1.8800543855471535e-05, "loss": 0.16428084373474122, "step": 6940 }, { "epoch": 0.1830735949266224, "grad_norm": 2.045989513397217, "learning_rate": 1.87964893355324e-05, "loss": 0.1600802183151245, "step": 6950 }, { "epoch": 0.18333701017112114, "grad_norm": 4.082156658172607, "learning_rate": 1.879242841312881e-05, "loss": 0.16440305709838868, "step": 6960 }, { "epoch": 0.18360042541561986, "grad_norm": 5.050792694091797, "learning_rate": 1.878836109121648e-05, "loss": 0.16436634063720704, "step": 6970 }, { "epoch": 0.1838638406601186, "grad_norm": 2.2529029846191406, "learning_rate": 1.8784287372755795e-05, "loss": 0.17805129289627075, "step": 6980 }, { "epoch": 0.18412725590461734, "grad_norm": 5.111148834228516, "learning_rate": 1.8780207260711767e-05, "loss": 0.17226040363311768, "step": 6990 }, { "epoch": 0.1843906711491161, "grad_norm": 2.583089828491211, "learning_rate": 1.8776120758054087e-05, "loss": 0.17198605537414552, "step": 7000 }, { "epoch": 0.1846540863936148, "grad_norm": 3.499967098236084, "learning_rate": 1.8772027867757082e-05, "loss": 0.16614365577697754, "step": 7010 }, { "epoch": 0.18491750163811355, "grad_norm": 5.248245716094971, "learning_rate": 1.8767928592799742e-05, "loss": 0.1624068021774292, "step": 7020 }, { "epoch": 0.1851809168826123, "grad_norm": 4.674092769622803, "learning_rate": 1.876382293616569e-05, "loss": 0.1803666353225708, "step": 7030 }, { "epoch": 0.18544433212711103, "grad_norm": 2.8748278617858887, "learning_rate": 1.8759710900843202e-05, "loss": 0.21097030639648437, "step": 7040 }, { "epoch": 0.18570774737160978, "grad_norm": 7.278939247131348, "learning_rate": 1.8755592489825193e-05, "loss": 0.18995391130447387, "step": 7050 }, { "epoch": 0.1859711626161085, "grad_norm": 2.659989833831787, "learning_rate": 1.8751467706109224e-05, "loss": 0.19684903621673583, "step": 7060 }, { "epoch": 0.18623457786060724, "grad_norm": 2.9802424907684326, "learning_rate": 1.8747336552697483e-05, "loss": 0.1760891079902649, "step": 7070 }, { "epoch": 0.18649799310510598, "grad_norm": 2.326528787612915, "learning_rate": 1.874319903259681e-05, "loss": 0.1654604196548462, "step": 7080 }, { "epoch": 0.18676140834960472, "grad_norm": 3.42419171333313, "learning_rate": 1.873905514881867e-05, "loss": 0.18177261352539062, "step": 7090 }, { "epoch": 0.18702482359410344, "grad_norm": 5.611527919769287, "learning_rate": 1.873490490437915e-05, "loss": 0.16119472980499266, "step": 7100 }, { "epoch": 0.18728823883860218, "grad_norm": 4.592707633972168, "learning_rate": 1.8730748302298993e-05, "loss": 0.1767200708389282, "step": 7110 }, { "epoch": 0.18755165408310093, "grad_norm": 2.373608350753784, "learning_rate": 1.8726585345603544e-05, "loss": 0.19712188243865966, "step": 7120 }, { "epoch": 0.18781506932759967, "grad_norm": 1.3141475915908813, "learning_rate": 1.8722416037322778e-05, "loss": 0.17659077644348145, "step": 7130 }, { "epoch": 0.1880784845720984, "grad_norm": 2.6782824993133545, "learning_rate": 1.871824038049131e-05, "loss": 0.1829690456390381, "step": 7140 }, { "epoch": 0.18834189981659713, "grad_norm": 1.63167142868042, "learning_rate": 1.871405837814835e-05, "loss": 0.16516244411468506, "step": 7150 }, { "epoch": 0.18860531506109587, "grad_norm": 2.5763537883758545, "learning_rate": 1.8709870033337753e-05, "loss": 0.18662575483322144, "step": 7160 }, { "epoch": 0.18886873030559462, "grad_norm": 5.820642948150635, "learning_rate": 1.8705675349107964e-05, "loss": 0.16892417669296264, "step": 7170 }, { "epoch": 0.18913214555009336, "grad_norm": 5.2289042472839355, "learning_rate": 1.870147432851206e-05, "loss": 0.1670227527618408, "step": 7180 }, { "epoch": 0.18939556079459208, "grad_norm": 3.6817381381988525, "learning_rate": 1.8697266974607732e-05, "loss": 0.16935240030288695, "step": 7190 }, { "epoch": 0.18965897603909082, "grad_norm": 1.8282876014709473, "learning_rate": 1.8693053290457264e-05, "loss": 0.15586115121841432, "step": 7200 }, { "epoch": 0.18992239128358956, "grad_norm": 4.511528015136719, "learning_rate": 1.868883327912756e-05, "loss": 0.1727290391921997, "step": 7210 }, { "epoch": 0.1901858065280883, "grad_norm": 1.9607263803482056, "learning_rate": 1.868460694369013e-05, "loss": 0.17223796844482422, "step": 7220 }, { "epoch": 0.19044922177258702, "grad_norm": 1.6740351915359497, "learning_rate": 1.868037428722108e-05, "loss": 0.1681333899497986, "step": 7230 }, { "epoch": 0.19071263701708577, "grad_norm": 2.9509968757629395, "learning_rate": 1.867613531280112e-05, "loss": 0.16073808670043946, "step": 7240 }, { "epoch": 0.1909760522615845, "grad_norm": 2.1239049434661865, "learning_rate": 1.867189002351556e-05, "loss": 0.16127709150314332, "step": 7250 }, { "epoch": 0.19123946750608325, "grad_norm": 4.496032238006592, "learning_rate": 1.8667638422454306e-05, "loss": 0.1679969072341919, "step": 7260 }, { "epoch": 0.19150288275058197, "grad_norm": 3.453192710876465, "learning_rate": 1.8663380512711852e-05, "loss": 0.1619948387145996, "step": 7270 }, { "epoch": 0.19176629799508071, "grad_norm": 1.6147561073303223, "learning_rate": 1.865911629738729e-05, "loss": 0.17010650634765626, "step": 7280 }, { "epoch": 0.19202971323957946, "grad_norm": 2.042144775390625, "learning_rate": 1.86548457795843e-05, "loss": 0.17494421005249022, "step": 7290 }, { "epoch": 0.1922931284840782, "grad_norm": 2.0798842906951904, "learning_rate": 1.8650568962411154e-05, "loss": 0.16411374807357787, "step": 7300 }, { "epoch": 0.19255654372857695, "grad_norm": 2.243671178817749, "learning_rate": 1.8646285848980693e-05, "loss": 0.15799417495727539, "step": 7310 }, { "epoch": 0.19281995897307566, "grad_norm": 7.6539154052734375, "learning_rate": 1.8641996442410365e-05, "loss": 0.15058610439300538, "step": 7320 }, { "epoch": 0.1930833742175744, "grad_norm": 0.8980183005332947, "learning_rate": 1.8637700745822176e-05, "loss": 0.17131319046020507, "step": 7330 }, { "epoch": 0.19334678946207315, "grad_norm": 1.0666052103042603, "learning_rate": 1.8633398762342722e-05, "loss": 0.16221153736114502, "step": 7340 }, { "epoch": 0.1936102047065719, "grad_norm": 4.088079929351807, "learning_rate": 1.8629090495103175e-05, "loss": 0.16774516105651854, "step": 7350 }, { "epoch": 0.1938736199510706, "grad_norm": 2.302339792251587, "learning_rate": 1.8624775947239272e-05, "loss": 0.14571820497512816, "step": 7360 }, { "epoch": 0.19413703519556935, "grad_norm": 4.165560245513916, "learning_rate": 1.862045512189133e-05, "loss": 0.1630127787590027, "step": 7370 }, { "epoch": 0.1944004504400681, "grad_norm": 3.193629741668701, "learning_rate": 1.8616128022204233e-05, "loss": 0.16239585876464843, "step": 7380 }, { "epoch": 0.19466386568456684, "grad_norm": 1.9364898204803467, "learning_rate": 1.8611794651327432e-05, "loss": 0.1697068452835083, "step": 7390 }, { "epoch": 0.19492728092906556, "grad_norm": 1.0055220127105713, "learning_rate": 1.8607455012414936e-05, "loss": 0.15749528408050537, "step": 7400 }, { "epoch": 0.1951906961735643, "grad_norm": 1.3633909225463867, "learning_rate": 1.8603109108625327e-05, "loss": 0.16487329006195067, "step": 7410 }, { "epoch": 0.19545411141806304, "grad_norm": 3.2677676677703857, "learning_rate": 1.859875694312174e-05, "loss": 0.14989620447158813, "step": 7420 }, { "epoch": 0.19571752666256179, "grad_norm": 1.011766791343689, "learning_rate": 1.8594398519071867e-05, "loss": 0.17113375663757324, "step": 7430 }, { "epoch": 0.19598094190706053, "grad_norm": 2.0004847049713135, "learning_rate": 1.8590033839647955e-05, "loss": 0.17015955448150635, "step": 7440 }, { "epoch": 0.19624435715155925, "grad_norm": 1.2696068286895752, "learning_rate": 1.8585662908026814e-05, "loss": 0.155293607711792, "step": 7450 }, { "epoch": 0.196507772396058, "grad_norm": 1.3070279359817505, "learning_rate": 1.8581285727389787e-05, "loss": 0.16239447593688966, "step": 7460 }, { "epoch": 0.19677118764055673, "grad_norm": 5.392116546630859, "learning_rate": 1.857690230092278e-05, "loss": 0.16823854446411132, "step": 7470 }, { "epoch": 0.19703460288505548, "grad_norm": 2.5561437606811523, "learning_rate": 1.8572512631816243e-05, "loss": 0.15853567123413087, "step": 7480 }, { "epoch": 0.1972980181295542, "grad_norm": 1.8729355335235596, "learning_rate": 1.856811672326516e-05, "loss": 0.15561175346374512, "step": 7490 }, { "epoch": 0.19756143337405294, "grad_norm": 2.3347041606903076, "learning_rate": 1.8563714578469064e-05, "loss": 0.169126296043396, "step": 7500 }, { "epoch": 0.19782484861855168, "grad_norm": 4.865407466888428, "learning_rate": 1.8559306200632036e-05, "loss": 0.1629047155380249, "step": 7510 }, { "epoch": 0.19808826386305042, "grad_norm": 2.0954182147979736, "learning_rate": 1.8554891592962666e-05, "loss": 0.14928641319274902, "step": 7520 }, { "epoch": 0.19835167910754914, "grad_norm": 2.8538010120391846, "learning_rate": 1.8550470758674112e-05, "loss": 0.15893958806991576, "step": 7530 }, { "epoch": 0.19861509435204788, "grad_norm": 1.3183573484420776, "learning_rate": 1.8546043700984036e-05, "loss": 0.1622536897659302, "step": 7540 }, { "epoch": 0.19887850959654663, "grad_norm": 1.6269279718399048, "learning_rate": 1.8541610423114647e-05, "loss": 0.17128093242645265, "step": 7550 }, { "epoch": 0.19914192484104537, "grad_norm": 2.849456310272217, "learning_rate": 1.853717092829268e-05, "loss": 0.16845288276672363, "step": 7560 }, { "epoch": 0.1994053400855441, "grad_norm": 2.577843189239502, "learning_rate": 1.853272521974938e-05, "loss": 0.17209984064102174, "step": 7570 }, { "epoch": 0.19966875533004283, "grad_norm": 1.5756548643112183, "learning_rate": 1.8528273300720538e-05, "loss": 0.15944375991821289, "step": 7580 }, { "epoch": 0.19993217057454157, "grad_norm": 1.97694993019104, "learning_rate": 1.8523815174446444e-05, "loss": 0.1614782452583313, "step": 7590 }, { "epoch": 0.20019558581904032, "grad_norm": 0.865290105342865, "learning_rate": 1.851935084417192e-05, "loss": 0.1544059157371521, "step": 7600 }, { "epoch": 0.20045900106353906, "grad_norm": 2.20296573638916, "learning_rate": 1.8514880313146295e-05, "loss": 0.17597074508666993, "step": 7610 }, { "epoch": 0.20072241630803778, "grad_norm": 1.9151873588562012, "learning_rate": 1.8510403584623414e-05, "loss": 0.16932079792022706, "step": 7620 }, { "epoch": 0.20098583155253652, "grad_norm": 6.755666255950928, "learning_rate": 1.850592066186163e-05, "loss": 0.15634685754776, "step": 7630 }, { "epoch": 0.20124924679703526, "grad_norm": 2.9476845264434814, "learning_rate": 1.850143154812382e-05, "loss": 0.15576770305633544, "step": 7640 }, { "epoch": 0.201512662041534, "grad_norm": 2.0119848251342773, "learning_rate": 1.849693624667734e-05, "loss": 0.1581106662750244, "step": 7650 }, { "epoch": 0.20177607728603272, "grad_norm": 2.5852901935577393, "learning_rate": 1.8492434760794072e-05, "loss": 0.16946741342544555, "step": 7660 }, { "epoch": 0.20203949253053147, "grad_norm": 1.8606983423233032, "learning_rate": 1.848792709375039e-05, "loss": 0.16043777465820314, "step": 7670 }, { "epoch": 0.2023029077750302, "grad_norm": 0.8206872940063477, "learning_rate": 1.848341324882717e-05, "loss": 0.1492548942565918, "step": 7680 }, { "epoch": 0.20256632301952895, "grad_norm": 0.8807576894760132, "learning_rate": 1.8478893229309784e-05, "loss": 0.167093026638031, "step": 7690 }, { "epoch": 0.2028297382640277, "grad_norm": 1.4192417860031128, "learning_rate": 1.847436703848809e-05, "loss": 0.17023615837097167, "step": 7700 }, { "epoch": 0.20309315350852641, "grad_norm": 3.3878228664398193, "learning_rate": 1.8469834679656452e-05, "loss": 0.17230358123779296, "step": 7710 }, { "epoch": 0.20335656875302516, "grad_norm": 3.113363027572632, "learning_rate": 1.8465296156113715e-05, "loss": 0.1614341616630554, "step": 7720 }, { "epoch": 0.2036199839975239, "grad_norm": 3.0346970558166504, "learning_rate": 1.8460751471163212e-05, "loss": 0.17043224573135377, "step": 7730 }, { "epoch": 0.20388339924202264, "grad_norm": 1.855709195137024, "learning_rate": 1.845620062811276e-05, "loss": 0.16707150936126708, "step": 7740 }, { "epoch": 0.20414681448652136, "grad_norm": 3.180055856704712, "learning_rate": 1.8451643630274664e-05, "loss": 0.15887982845306398, "step": 7750 }, { "epoch": 0.2044102297310201, "grad_norm": 2.192854166030884, "learning_rate": 1.8447080480965694e-05, "loss": 0.17054371833801268, "step": 7760 }, { "epoch": 0.20467364497551885, "grad_norm": 2.495220899581909, "learning_rate": 1.844251118350711e-05, "loss": 0.15964164733886718, "step": 7770 }, { "epoch": 0.2049370602200176, "grad_norm": 1.4784324169158936, "learning_rate": 1.8437935741224652e-05, "loss": 0.16575998067855835, "step": 7780 }, { "epoch": 0.2052004754645163, "grad_norm": 1.889410376548767, "learning_rate": 1.8433354157448513e-05, "loss": 0.16511085033416747, "step": 7790 }, { "epoch": 0.20546389070901505, "grad_norm": 4.894556999206543, "learning_rate": 1.8428766435513373e-05, "loss": 0.16572890281677247, "step": 7800 }, { "epoch": 0.2057273059535138, "grad_norm": 5.919103145599365, "learning_rate": 1.8424172578758374e-05, "loss": 0.16392452716827394, "step": 7810 }, { "epoch": 0.20599072119801254, "grad_norm": 4.01084041595459, "learning_rate": 1.841957259052712e-05, "loss": 0.15389950275421144, "step": 7820 }, { "epoch": 0.20625413644251128, "grad_norm": 1.4713680744171143, "learning_rate": 1.8414966474167688e-05, "loss": 0.16840438842773436, "step": 7830 }, { "epoch": 0.20651755168701, "grad_norm": 5.15854024887085, "learning_rate": 1.8410354233032604e-05, "loss": 0.15194060802459716, "step": 7840 }, { "epoch": 0.20678096693150874, "grad_norm": 1.3724641799926758, "learning_rate": 1.840573587047885e-05, "loss": 0.16198667287826538, "step": 7850 }, { "epoch": 0.20704438217600749, "grad_norm": 1.5902847051620483, "learning_rate": 1.8401111389867882e-05, "loss": 0.17209094762802124, "step": 7860 }, { "epoch": 0.20730779742050623, "grad_norm": 3.7435860633850098, "learning_rate": 1.8396480794565583e-05, "loss": 0.1607959032058716, "step": 7870 }, { "epoch": 0.20757121266500494, "grad_norm": 3.052342653274536, "learning_rate": 1.839184408794231e-05, "loss": 0.16280531883239746, "step": 7880 }, { "epoch": 0.2078346279095037, "grad_norm": 1.997627854347229, "learning_rate": 1.8387201273372855e-05, "loss": 0.16262807846069335, "step": 7890 }, { "epoch": 0.20809804315400243, "grad_norm": 5.7960309982299805, "learning_rate": 1.838255235423646e-05, "loss": 0.15122652053833008, "step": 7900 }, { "epoch": 0.20836145839850118, "grad_norm": 2.3266141414642334, "learning_rate": 1.8377897333916808e-05, "loss": 0.16003901958465577, "step": 7910 }, { "epoch": 0.2086248736429999, "grad_norm": 6.148773193359375, "learning_rate": 1.837323621580202e-05, "loss": 0.15379596948623658, "step": 7920 }, { "epoch": 0.20888828888749864, "grad_norm": 4.489358425140381, "learning_rate": 1.8368569003284662e-05, "loss": 0.15821530818939208, "step": 7930 }, { "epoch": 0.20915170413199738, "grad_norm": 1.725313425064087, "learning_rate": 1.8363895699761736e-05, "loss": 0.16715964078903198, "step": 7940 }, { "epoch": 0.20941511937649612, "grad_norm": 6.693430423736572, "learning_rate": 1.8359216308634668e-05, "loss": 0.16134485006332397, "step": 7950 }, { "epoch": 0.20967853462099487, "grad_norm": 0.9763492941856384, "learning_rate": 1.8354530833309323e-05, "loss": 0.16330621242523194, "step": 7960 }, { "epoch": 0.20994194986549358, "grad_norm": 3.515533447265625, "learning_rate": 1.8349839277195993e-05, "loss": 0.1539737343788147, "step": 7970 }, { "epoch": 0.21020536510999233, "grad_norm": 2.5364975929260254, "learning_rate": 1.8345141643709394e-05, "loss": 0.16240112781524657, "step": 7980 }, { "epoch": 0.21046878035449107, "grad_norm": 3.9959020614624023, "learning_rate": 1.8340437936268667e-05, "loss": 0.16309607028961182, "step": 7990 }, { "epoch": 0.2107321955989898, "grad_norm": 4.428228855133057, "learning_rate": 1.8335728158297374e-05, "loss": 0.1705467700958252, "step": 8000 }, { "epoch": 0.21099561084348853, "grad_norm": 1.071292519569397, "learning_rate": 1.833101231322349e-05, "loss": 0.16448044776916504, "step": 8010 }, { "epoch": 0.21125902608798727, "grad_norm": 3.8537795543670654, "learning_rate": 1.832629040447942e-05, "loss": 0.16650757789611817, "step": 8020 }, { "epoch": 0.21152244133248602, "grad_norm": 2.4984142780303955, "learning_rate": 1.8321562435501963e-05, "loss": 0.17768845558166504, "step": 8030 }, { "epoch": 0.21178585657698476, "grad_norm": 3.83296799659729, "learning_rate": 1.831682840973234e-05, "loss": 0.16121954917907716, "step": 8040 }, { "epoch": 0.21204927182148348, "grad_norm": 2.2065391540527344, "learning_rate": 1.8312088330616188e-05, "loss": 0.15920827388763428, "step": 8050 }, { "epoch": 0.21231268706598222, "grad_norm": 2.02602219581604, "learning_rate": 1.8307342201603533e-05, "loss": 0.14968457221984863, "step": 8060 }, { "epoch": 0.21257610231048096, "grad_norm": 1.378604769706726, "learning_rate": 1.830259002614881e-05, "loss": 0.14141228199005126, "step": 8070 }, { "epoch": 0.2128395175549797, "grad_norm": 1.9179811477661133, "learning_rate": 1.829783180771087e-05, "loss": 0.16994504928588866, "step": 8080 }, { "epoch": 0.21310293279947845, "grad_norm": 1.573759913444519, "learning_rate": 1.8293067549752937e-05, "loss": 0.16119494438171386, "step": 8090 }, { "epoch": 0.21336634804397717, "grad_norm": 1.8149417638778687, "learning_rate": 1.8288297255742646e-05, "loss": 0.1807626962661743, "step": 8100 }, { "epoch": 0.2136297632884759, "grad_norm": 3.14925217628479, "learning_rate": 1.828352092915203e-05, "loss": 0.168023681640625, "step": 8110 }, { "epoch": 0.21389317853297465, "grad_norm": 0.9630879163742065, "learning_rate": 1.8278738573457496e-05, "loss": 0.15091509819030763, "step": 8120 }, { "epoch": 0.2141565937774734, "grad_norm": 0.7821836471557617, "learning_rate": 1.8273950192139854e-05, "loss": 0.16353490352630615, "step": 8130 }, { "epoch": 0.2144200090219721, "grad_norm": 3.4238839149475098, "learning_rate": 1.826915578868429e-05, "loss": 0.16664153337478638, "step": 8140 }, { "epoch": 0.21468342426647086, "grad_norm": 3.610527753829956, "learning_rate": 1.8264355366580387e-05, "loss": 0.1638600468635559, "step": 8150 }, { "epoch": 0.2149468395109696, "grad_norm": 1.6107364892959595, "learning_rate": 1.825954892932209e-05, "loss": 0.1538951277732849, "step": 8160 }, { "epoch": 0.21521025475546834, "grad_norm": 2.0129730701446533, "learning_rate": 1.8254736480407738e-05, "loss": 0.15785839557647705, "step": 8170 }, { "epoch": 0.21547366999996706, "grad_norm": 2.3647446632385254, "learning_rate": 1.8249918023340036e-05, "loss": 0.1561770796775818, "step": 8180 }, { "epoch": 0.2157370852444658, "grad_norm": 5.446483612060547, "learning_rate": 1.8245093561626063e-05, "loss": 0.16768028736114501, "step": 8190 }, { "epoch": 0.21600050048896455, "grad_norm": 5.458536148071289, "learning_rate": 1.824026309877727e-05, "loss": 0.1745610475540161, "step": 8200 }, { "epoch": 0.2162639157334633, "grad_norm": 4.08908224105835, "learning_rate": 1.8235426638309483e-05, "loss": 0.17395291328430176, "step": 8210 }, { "epoch": 0.21652733097796203, "grad_norm": 1.3845990896224976, "learning_rate": 1.823058418374288e-05, "loss": 0.1723832368850708, "step": 8220 }, { "epoch": 0.21679074622246075, "grad_norm": 2.009977340698242, "learning_rate": 1.822573573860201e-05, "loss": 0.15581518411636353, "step": 8230 }, { "epoch": 0.2170541614669595, "grad_norm": 1.021185040473938, "learning_rate": 1.8220881306415783e-05, "loss": 0.15295180082321166, "step": 8240 }, { "epoch": 0.21731757671145824, "grad_norm": 1.6888254880905151, "learning_rate": 1.821602089071746e-05, "loss": 0.1581830620765686, "step": 8250 }, { "epoch": 0.21758099195595698, "grad_norm": 1.5942339897155762, "learning_rate": 1.8211154495044666e-05, "loss": 0.16804072856903077, "step": 8260 }, { "epoch": 0.2178444072004557, "grad_norm": 1.1106088161468506, "learning_rate": 1.8206282122939368e-05, "loss": 0.14857172966003418, "step": 8270 }, { "epoch": 0.21810782244495444, "grad_norm": 2.313356399536133, "learning_rate": 1.8201403777947896e-05, "loss": 0.1536933660507202, "step": 8280 }, { "epoch": 0.21837123768945318, "grad_norm": 1.264906406402588, "learning_rate": 1.8196519463620912e-05, "loss": 0.16141271591186523, "step": 8290 }, { "epoch": 0.21863465293395193, "grad_norm": 2.3241682052612305, "learning_rate": 1.8191629183513435e-05, "loss": 0.15700697898864746, "step": 8300 }, { "epoch": 0.21889806817845064, "grad_norm": 1.2825320959091187, "learning_rate": 1.8186732941184824e-05, "loss": 0.16410326957702637, "step": 8310 }, { "epoch": 0.2191614834229494, "grad_norm": 1.6750715970993042, "learning_rate": 1.818183074019877e-05, "loss": 0.16290085315704345, "step": 8320 }, { "epoch": 0.21942489866744813, "grad_norm": 2.5448312759399414, "learning_rate": 1.8176922584123308e-05, "loss": 0.1531542420387268, "step": 8330 }, { "epoch": 0.21968831391194688, "grad_norm": 4.3327250480651855, "learning_rate": 1.817200847653081e-05, "loss": 0.18224371671676637, "step": 8340 }, { "epoch": 0.21995172915644562, "grad_norm": 1.7740013599395752, "learning_rate": 1.8167088420997975e-05, "loss": 0.1755634069442749, "step": 8350 }, { "epoch": 0.22021514440094433, "grad_norm": 2.181528091430664, "learning_rate": 1.816216242110582e-05, "loss": 0.16393498182296753, "step": 8360 }, { "epoch": 0.22047855964544308, "grad_norm": 1.1649916172027588, "learning_rate": 1.815723048043972e-05, "loss": 0.15106732845306398, "step": 8370 }, { "epoch": 0.22074197488994182, "grad_norm": 1.7986702919006348, "learning_rate": 1.8152292602589336e-05, "loss": 0.16529239416122438, "step": 8380 }, { "epoch": 0.22100539013444057, "grad_norm": 0.9336449503898621, "learning_rate": 1.8147348791148678e-05, "loss": 0.17406890392303467, "step": 8390 }, { "epoch": 0.22126880537893928, "grad_norm": 0.6829098463058472, "learning_rate": 1.8142399049716065e-05, "loss": 0.15464972257614135, "step": 8400 }, { "epoch": 0.22153222062343803, "grad_norm": 2.613943338394165, "learning_rate": 1.8137443381894126e-05, "loss": 0.1736659288406372, "step": 8410 }, { "epoch": 0.22179563586793677, "grad_norm": 1.3014777898788452, "learning_rate": 1.813248179128982e-05, "loss": 0.15308001041412353, "step": 8420 }, { "epoch": 0.2220590511124355, "grad_norm": 1.6485323905944824, "learning_rate": 1.81275142815144e-05, "loss": 0.17042776346206664, "step": 8430 }, { "epoch": 0.22232246635693423, "grad_norm": 2.7369368076324463, "learning_rate": 1.812254085618343e-05, "loss": 0.17015035152435304, "step": 8440 }, { "epoch": 0.22258588160143297, "grad_norm": 3.462459087371826, "learning_rate": 1.8117561518916792e-05, "loss": 0.16322412490844726, "step": 8450 }, { "epoch": 0.22284929684593172, "grad_norm": 2.6772983074188232, "learning_rate": 1.811257627333866e-05, "loss": 0.16365425586700438, "step": 8460 }, { "epoch": 0.22311271209043046, "grad_norm": 1.018692135810852, "learning_rate": 1.810758512307751e-05, "loss": 0.15944724082946776, "step": 8470 }, { "epoch": 0.2233761273349292, "grad_norm": 5.075756072998047, "learning_rate": 1.8102588071766116e-05, "loss": 0.1628314733505249, "step": 8480 }, { "epoch": 0.22363954257942792, "grad_norm": 3.675539970397949, "learning_rate": 1.8097585123041547e-05, "loss": 0.1567941427230835, "step": 8490 }, { "epoch": 0.22390295782392666, "grad_norm": 4.205177307128906, "learning_rate": 1.809257628054517e-05, "loss": 0.15695706605911255, "step": 8500 }, { "epoch": 0.2241663730684254, "grad_norm": 1.7804229259490967, "learning_rate": 1.808756154792263e-05, "loss": 0.16098370552062988, "step": 8510 }, { "epoch": 0.22442978831292415, "grad_norm": 3.64386248588562, "learning_rate": 1.8082540928823866e-05, "loss": 0.17046539783477782, "step": 8520 }, { "epoch": 0.22469320355742287, "grad_norm": 1.1410773992538452, "learning_rate": 1.8077514426903107e-05, "loss": 0.16370209455490112, "step": 8530 }, { "epoch": 0.2249566188019216, "grad_norm": 3.0749733448028564, "learning_rate": 1.8072482045818852e-05, "loss": 0.14837101697921753, "step": 8540 }, { "epoch": 0.22522003404642035, "grad_norm": 2.7820868492126465, "learning_rate": 1.806744378923389e-05, "loss": 0.16463356018066405, "step": 8550 }, { "epoch": 0.2254834492909191, "grad_norm": 2.0203561782836914, "learning_rate": 1.8062399660815277e-05, "loss": 0.16238459348678588, "step": 8560 }, { "epoch": 0.2257468645354178, "grad_norm": 0.7987009286880493, "learning_rate": 1.8057349664234348e-05, "loss": 0.15177173614501954, "step": 8570 }, { "epoch": 0.22601027977991656, "grad_norm": 2.354477882385254, "learning_rate": 1.8052293803166714e-05, "loss": 0.16242661476135253, "step": 8580 }, { "epoch": 0.2262736950244153, "grad_norm": 3.2814409732818604, "learning_rate": 1.8047232081292242e-05, "loss": 0.16261953115463257, "step": 8590 }, { "epoch": 0.22653711026891404, "grad_norm": 1.8655952215194702, "learning_rate": 1.8042164502295074e-05, "loss": 0.1677911877632141, "step": 8600 }, { "epoch": 0.2268005255134128, "grad_norm": 0.8071324229240417, "learning_rate": 1.8037091069863615e-05, "loss": 0.160125470161438, "step": 8610 }, { "epoch": 0.2270639407579115, "grad_norm": 2.163273572921753, "learning_rate": 1.8032011787690523e-05, "loss": 0.15370912551879884, "step": 8620 }, { "epoch": 0.22732735600241025, "grad_norm": 3.2958908081054688, "learning_rate": 1.8026926659472727e-05, "loss": 0.15182051658630372, "step": 8630 }, { "epoch": 0.227590771246909, "grad_norm": 2.024746894836426, "learning_rate": 1.802183568891139e-05, "loss": 0.1628546953201294, "step": 8640 }, { "epoch": 0.22785418649140773, "grad_norm": 2.420069932937622, "learning_rate": 1.801673887971195e-05, "loss": 0.1664818048477173, "step": 8650 }, { "epoch": 0.22811760173590645, "grad_norm": 1.5036851167678833, "learning_rate": 1.801163623558408e-05, "loss": 0.1666705846786499, "step": 8660 }, { "epoch": 0.2283810169804052, "grad_norm": 1.2828360795974731, "learning_rate": 1.800652776024171e-05, "loss": 0.15661541223526002, "step": 8670 }, { "epoch": 0.22864443222490394, "grad_norm": 2.4194955825805664, "learning_rate": 1.8001413457403005e-05, "loss": 0.16349000930786134, "step": 8680 }, { "epoch": 0.22890784746940268, "grad_norm": 2.4754958152770996, "learning_rate": 1.7996293330790374e-05, "loss": 0.16099815368652343, "step": 8690 }, { "epoch": 0.2291712627139014, "grad_norm": 1.3669472932815552, "learning_rate": 1.7991167384130466e-05, "loss": 0.15984094142913818, "step": 8700 }, { "epoch": 0.22943467795840014, "grad_norm": 2.1582024097442627, "learning_rate": 1.798603562115417e-05, "loss": 0.15655250549316407, "step": 8710 }, { "epoch": 0.22969809320289888, "grad_norm": 1.3744041919708252, "learning_rate": 1.79808980455966e-05, "loss": 0.15316004753112794, "step": 8720 }, { "epoch": 0.22996150844739763, "grad_norm": 1.0100109577178955, "learning_rate": 1.7975754661197106e-05, "loss": 0.15934855937957765, "step": 8730 }, { "epoch": 0.23022492369189637, "grad_norm": 2.3449339866638184, "learning_rate": 1.797060547169927e-05, "loss": 0.16223411560058593, "step": 8740 }, { "epoch": 0.2304883389363951, "grad_norm": 1.784799337387085, "learning_rate": 1.796545048085089e-05, "loss": 0.1610018253326416, "step": 8750 }, { "epoch": 0.23075175418089383, "grad_norm": 1.9664961099624634, "learning_rate": 1.796028969240399e-05, "loss": 0.1617911696434021, "step": 8760 }, { "epoch": 0.23101516942539257, "grad_norm": 1.7211217880249023, "learning_rate": 1.7955123110114815e-05, "loss": 0.164029860496521, "step": 8770 }, { "epoch": 0.23127858466989132, "grad_norm": 1.2064893245697021, "learning_rate": 1.794995073774383e-05, "loss": 0.14954013824462892, "step": 8780 }, { "epoch": 0.23154199991439003, "grad_norm": 0.8653826713562012, "learning_rate": 1.7944772579055705e-05, "loss": 0.1660940170288086, "step": 8790 }, { "epoch": 0.23180541515888878, "grad_norm": 2.2353200912475586, "learning_rate": 1.793958863781933e-05, "loss": 0.1560220956802368, "step": 8800 }, { "epoch": 0.23206883040338752, "grad_norm": 3.105337619781494, "learning_rate": 1.7934398917807802e-05, "loss": 0.18914644718170165, "step": 8810 }, { "epoch": 0.23233224564788627, "grad_norm": 1.5907995700836182, "learning_rate": 1.7929203422798426e-05, "loss": 0.15878183841705323, "step": 8820 }, { "epoch": 0.23259566089238498, "grad_norm": 1.2332062721252441, "learning_rate": 1.79240021565727e-05, "loss": 0.1568005084991455, "step": 8830 }, { "epoch": 0.23285907613688372, "grad_norm": 7.446102619171143, "learning_rate": 1.7918795122916337e-05, "loss": 0.16466379165649414, "step": 8840 }, { "epoch": 0.23312249138138247, "grad_norm": 0.9547043442726135, "learning_rate": 1.7913582325619236e-05, "loss": 0.16558750867843627, "step": 8850 }, { "epoch": 0.2333859066258812, "grad_norm": 4.7721357345581055, "learning_rate": 1.7908363768475498e-05, "loss": 0.16606947183609008, "step": 8860 }, { "epoch": 0.23364932187037996, "grad_norm": 2.0960066318511963, "learning_rate": 1.7903139455283412e-05, "loss": 0.1635076403617859, "step": 8870 }, { "epoch": 0.23391273711487867, "grad_norm": 4.8676934242248535, "learning_rate": 1.7897909389845462e-05, "loss": 0.1601066470146179, "step": 8880 }, { "epoch": 0.23417615235937742, "grad_norm": 6.350419521331787, "learning_rate": 1.789267357596831e-05, "loss": 0.14980695247650147, "step": 8890 }, { "epoch": 0.23443956760387616, "grad_norm": 1.8976478576660156, "learning_rate": 1.7887432017462812e-05, "loss": 0.17265095710754394, "step": 8900 }, { "epoch": 0.2347029828483749, "grad_norm": 5.35283088684082, "learning_rate": 1.7882184718144002e-05, "loss": 0.1677072286605835, "step": 8910 }, { "epoch": 0.23496639809287362, "grad_norm": 2.9542768001556396, "learning_rate": 1.7876931681831087e-05, "loss": 0.1610551357269287, "step": 8920 }, { "epoch": 0.23522981333737236, "grad_norm": 1.9859586954116821, "learning_rate": 1.787167291234745e-05, "loss": 0.15537066459655763, "step": 8930 }, { "epoch": 0.2354932285818711, "grad_norm": 6.846118450164795, "learning_rate": 1.7866408413520662e-05, "loss": 0.15538939237594604, "step": 8940 }, { "epoch": 0.23575664382636985, "grad_norm": 2.1352996826171875, "learning_rate": 1.786113818918244e-05, "loss": 0.17079986333847047, "step": 8950 }, { "epoch": 0.23602005907086857, "grad_norm": 4.303771018981934, "learning_rate": 1.7855862243168684e-05, "loss": 0.15477923154830933, "step": 8960 }, { "epoch": 0.2362834743153673, "grad_norm": 2.977372169494629, "learning_rate": 1.785058057931946e-05, "loss": 0.16519565582275392, "step": 8970 }, { "epoch": 0.23654688955986605, "grad_norm": 3.213517665863037, "learning_rate": 1.7845293201478985e-05, "loss": 0.17775216102600097, "step": 8980 }, { "epoch": 0.2368103048043648, "grad_norm": 1.5466820001602173, "learning_rate": 1.7840000113495647e-05, "loss": 0.1613213300704956, "step": 8990 }, { "epoch": 0.23707372004886354, "grad_norm": 6.239384174346924, "learning_rate": 1.7834701319221978e-05, "loss": 0.16352636814117433, "step": 9000 }, { "epoch": 0.23733713529336226, "grad_norm": 1.822892427444458, "learning_rate": 1.782939682251467e-05, "loss": 0.16078196763992308, "step": 9010 }, { "epoch": 0.237600550537861, "grad_norm": 2.787990093231201, "learning_rate": 1.7824086627234565e-05, "loss": 0.15946011543273925, "step": 9020 }, { "epoch": 0.23786396578235974, "grad_norm": 5.311291694641113, "learning_rate": 1.7818770737246653e-05, "loss": 0.16355724334716798, "step": 9030 }, { "epoch": 0.2381273810268585, "grad_norm": 1.1583698987960815, "learning_rate": 1.7813449156420072e-05, "loss": 0.1382601499557495, "step": 9040 }, { "epoch": 0.2383907962713572, "grad_norm": 2.5530636310577393, "learning_rate": 1.7808121888628093e-05, "loss": 0.16405898332595825, "step": 9050 }, { "epoch": 0.23865421151585595, "grad_norm": 1.531480073928833, "learning_rate": 1.780278893774813e-05, "loss": 0.15784218311309814, "step": 9060 }, { "epoch": 0.2389176267603547, "grad_norm": 2.1450676918029785, "learning_rate": 1.7797450307661733e-05, "loss": 0.15881617069244386, "step": 9070 }, { "epoch": 0.23918104200485343, "grad_norm": 3.81913685798645, "learning_rate": 1.77921060022546e-05, "loss": 0.1604081630706787, "step": 9080 }, { "epoch": 0.23944445724935215, "grad_norm": 1.9348931312561035, "learning_rate": 1.7786756025416535e-05, "loss": 0.16837451457977295, "step": 9090 }, { "epoch": 0.2397078724938509, "grad_norm": 1.6844196319580078, "learning_rate": 1.7781400381041484e-05, "loss": 0.16427018642425537, "step": 9100 }, { "epoch": 0.23997128773834964, "grad_norm": 1.1727327108383179, "learning_rate": 1.777603907302752e-05, "loss": 0.15101703405380248, "step": 9110 }, { "epoch": 0.24023470298284838, "grad_norm": 1.9588807821273804, "learning_rate": 1.7770672105276833e-05, "loss": 0.1449112892150879, "step": 9120 }, { "epoch": 0.24049811822734712, "grad_norm": 3.27424693107605, "learning_rate": 1.776529948169573e-05, "loss": 0.16287606954574585, "step": 9130 }, { "epoch": 0.24076153347184584, "grad_norm": 3.0163488388061523, "learning_rate": 1.775992120619464e-05, "loss": 0.16827895641326904, "step": 9140 }, { "epoch": 0.24102494871634458, "grad_norm": 1.8187462091445923, "learning_rate": 1.7754537282688104e-05, "loss": 0.1624929666519165, "step": 9150 }, { "epoch": 0.24128836396084333, "grad_norm": 4.260072231292725, "learning_rate": 1.7749147715094777e-05, "loss": 0.1541570544242859, "step": 9160 }, { "epoch": 0.24155177920534207, "grad_norm": 0.8767756223678589, "learning_rate": 1.7743752507337414e-05, "loss": 0.16993422508239747, "step": 9170 }, { "epoch": 0.2418151944498408, "grad_norm": 3.881415367126465, "learning_rate": 1.7738351663342887e-05, "loss": 0.1681411862373352, "step": 9180 }, { "epoch": 0.24207860969433953, "grad_norm": 1.7987306118011475, "learning_rate": 1.7732945187042155e-05, "loss": 0.15435802936553955, "step": 9190 }, { "epoch": 0.24234202493883827, "grad_norm": 2.0025625228881836, "learning_rate": 1.7727533082370288e-05, "loss": 0.1446176528930664, "step": 9200 }, { "epoch": 0.24260544018333702, "grad_norm": 0.9883690476417542, "learning_rate": 1.772211535326645e-05, "loss": 0.17719157934188842, "step": 9210 }, { "epoch": 0.24286885542783573, "grad_norm": 2.498627185821533, "learning_rate": 1.7716692003673898e-05, "loss": 0.1549053430557251, "step": 9220 }, { "epoch": 0.24313227067233448, "grad_norm": 2.5401811599731445, "learning_rate": 1.7711263037539978e-05, "loss": 0.17411792278289795, "step": 9230 }, { "epoch": 0.24339568591683322, "grad_norm": 3.5713226795196533, "learning_rate": 1.770582845881613e-05, "loss": 0.16309481859207153, "step": 9240 }, { "epoch": 0.24365910116133196, "grad_norm": 1.326696515083313, "learning_rate": 1.7700388271457872e-05, "loss": 0.15828161239624022, "step": 9250 }, { "epoch": 0.2439225164058307, "grad_norm": 1.0580599308013916, "learning_rate": 1.7694942479424803e-05, "loss": 0.1589298129081726, "step": 9260 }, { "epoch": 0.24418593165032942, "grad_norm": 1.5410759449005127, "learning_rate": 1.768949108668061e-05, "loss": 0.15093696117401123, "step": 9270 }, { "epoch": 0.24444934689482817, "grad_norm": 1.2539401054382324, "learning_rate": 1.7684034097193055e-05, "loss": 0.15048615932464598, "step": 9280 }, { "epoch": 0.2447127621393269, "grad_norm": 5.068118572235107, "learning_rate": 1.7678571514933966e-05, "loss": 0.1590583562850952, "step": 9290 }, { "epoch": 0.24497617738382566, "grad_norm": 3.98771595954895, "learning_rate": 1.7673103343879242e-05, "loss": 0.1546589493751526, "step": 9300 }, { "epoch": 0.24523959262832437, "grad_norm": 2.6484882831573486, "learning_rate": 1.7667629588008862e-05, "loss": 0.15887119770050048, "step": 9310 }, { "epoch": 0.24550300787282311, "grad_norm": 1.6623294353485107, "learning_rate": 1.766215025130686e-05, "loss": 0.15384013652801515, "step": 9320 }, { "epoch": 0.24576642311732186, "grad_norm": 0.9852460622787476, "learning_rate": 1.7656665337761327e-05, "loss": 0.15523699522018433, "step": 9330 }, { "epoch": 0.2460298383618206, "grad_norm": 1.4262453317642212, "learning_rate": 1.7651174851364427e-05, "loss": 0.15548343658447267, "step": 9340 }, { "epoch": 0.24629325360631932, "grad_norm": 7.312509059906006, "learning_rate": 1.764567879611237e-05, "loss": 0.15748167037963867, "step": 9350 }, { "epoch": 0.24655666885081806, "grad_norm": 3.9893600940704346, "learning_rate": 1.7640177176005425e-05, "loss": 0.15498136281967162, "step": 9360 }, { "epoch": 0.2468200840953168, "grad_norm": 0.8755735754966736, "learning_rate": 1.7634669995047904e-05, "loss": 0.1716411828994751, "step": 9370 }, { "epoch": 0.24708349933981555, "grad_norm": 2.0170817375183105, "learning_rate": 1.7629157257248175e-05, "loss": 0.15246298313140869, "step": 9380 }, { "epoch": 0.2473469145843143, "grad_norm": 1.155087947845459, "learning_rate": 1.762363896661864e-05, "loss": 0.15378528833389282, "step": 9390 }, { "epoch": 0.247610329828813, "grad_norm": 4.494060039520264, "learning_rate": 1.761811512717576e-05, "loss": 0.1590442180633545, "step": 9400 }, { "epoch": 0.24787374507331175, "grad_norm": 1.3278852701187134, "learning_rate": 1.761258574294001e-05, "loss": 0.13836575746536256, "step": 9410 }, { "epoch": 0.2481371603178105, "grad_norm": 1.9113118648529053, "learning_rate": 1.760705081793593e-05, "loss": 0.1547330141067505, "step": 9420 }, { "epoch": 0.24840057556230924, "grad_norm": 4.874924659729004, "learning_rate": 1.760151035619206e-05, "loss": 0.17244716882705688, "step": 9430 }, { "epoch": 0.24866399080680796, "grad_norm": 0.9104582071304321, "learning_rate": 1.7595964361741e-05, "loss": 0.15058138370513915, "step": 9440 }, { "epoch": 0.2489274060513067, "grad_norm": 4.454753398895264, "learning_rate": 1.759041283861936e-05, "loss": 0.16414661407470704, "step": 9450 }, { "epoch": 0.24919082129580544, "grad_norm": 1.074864387512207, "learning_rate": 1.7584855790867773e-05, "loss": 0.16030123233795165, "step": 9460 }, { "epoch": 0.2494542365403042, "grad_norm": 1.5185661315917969, "learning_rate": 1.7579293222530907e-05, "loss": 0.16684322357177733, "step": 9470 }, { "epoch": 0.24971765178480293, "grad_norm": 0.8529759049415588, "learning_rate": 1.7573725137657432e-05, "loss": 0.16122859716415405, "step": 9480 }, { "epoch": 0.24998106702930165, "grad_norm": 2.7000362873077393, "learning_rate": 1.756815154030005e-05, "loss": 0.1567996025085449, "step": 9490 }, { "epoch": 0.2502444822738004, "grad_norm": 0.699916660785675, "learning_rate": 1.756257243451545e-05, "loss": 0.14968756437301636, "step": 9500 }, { "epoch": 0.2505078975182991, "grad_norm": 0.7811856865882874, "learning_rate": 1.7556987824364363e-05, "loss": 0.15362317562103273, "step": 9510 }, { "epoch": 0.25077131276279785, "grad_norm": 2.4824469089508057, "learning_rate": 1.75513977139115e-05, "loss": 0.15646660327911377, "step": 9520 }, { "epoch": 0.2510347280072966, "grad_norm": 5.742457389831543, "learning_rate": 1.7545802107225585e-05, "loss": 0.16845314502716063, "step": 9530 }, { "epoch": 0.25129814325179534, "grad_norm": 2.8151063919067383, "learning_rate": 1.754020100837934e-05, "loss": 0.16599757671356202, "step": 9540 }, { "epoch": 0.2515615584962941, "grad_norm": 2.042320489883423, "learning_rate": 1.753459442144949e-05, "loss": 0.17022705078125, "step": 9550 }, { "epoch": 0.2518249737407928, "grad_norm": 5.3042192459106445, "learning_rate": 1.7528982350516754e-05, "loss": 0.15827046632766723, "step": 9560 }, { "epoch": 0.25208838898529157, "grad_norm": 2.8801214694976807, "learning_rate": 1.7523364799665832e-05, "loss": 0.15745408535003663, "step": 9570 }, { "epoch": 0.2523518042297903, "grad_norm": 2.807886838912964, "learning_rate": 1.7517741772985424e-05, "loss": 0.16317267417907716, "step": 9580 }, { "epoch": 0.252615219474289, "grad_norm": 0.9256696701049805, "learning_rate": 1.751211327456821e-05, "loss": 0.16246535778045654, "step": 9590 }, { "epoch": 0.25287863471878774, "grad_norm": 4.98044490814209, "learning_rate": 1.7506479308510855e-05, "loss": 0.15570768117904663, "step": 9600 }, { "epoch": 0.2531420499632865, "grad_norm": 4.768322467803955, "learning_rate": 1.7500839878914004e-05, "loss": 0.15859371423721313, "step": 9610 }, { "epoch": 0.25340546520778523, "grad_norm": 2.4477169513702393, "learning_rate": 1.749519498988227e-05, "loss": 0.17221422195434571, "step": 9620 }, { "epoch": 0.253668880452284, "grad_norm": 3.977327585220337, "learning_rate": 1.748954464552426e-05, "loss": 0.15461130142211915, "step": 9630 }, { "epoch": 0.2539322956967827, "grad_norm": 1.2401930093765259, "learning_rate": 1.7483888849952525e-05, "loss": 0.16556849479675292, "step": 9640 }, { "epoch": 0.25419571094128146, "grad_norm": 3.002312660217285, "learning_rate": 1.7478227607283604e-05, "loss": 0.15401817560195924, "step": 9650 }, { "epoch": 0.2544591261857802, "grad_norm": 2.001497983932495, "learning_rate": 1.7472560921637993e-05, "loss": 0.16117751598358154, "step": 9660 }, { "epoch": 0.25472254143027895, "grad_norm": 5.921236038208008, "learning_rate": 1.746688879714015e-05, "loss": 0.15972357988357544, "step": 9670 }, { "epoch": 0.25498595667477764, "grad_norm": 0.9933639168739319, "learning_rate": 1.746121123791849e-05, "loss": 0.15508182048797609, "step": 9680 }, { "epoch": 0.2552493719192764, "grad_norm": 4.4232048988342285, "learning_rate": 1.7455528248105385e-05, "loss": 0.17247583866119384, "step": 9690 }, { "epoch": 0.2555127871637751, "grad_norm": 1.016816258430481, "learning_rate": 1.7449839831837166e-05, "loss": 0.1709409236907959, "step": 9700 }, { "epoch": 0.25577620240827387, "grad_norm": 2.3460609912872314, "learning_rate": 1.7444145993254105e-05, "loss": 0.13898379802703859, "step": 9710 }, { "epoch": 0.2560396176527726, "grad_norm": 1.080133080482483, "learning_rate": 1.7438446736500418e-05, "loss": 0.1738959312438965, "step": 9720 }, { "epoch": 0.25630303289727135, "grad_norm": 3.7530157566070557, "learning_rate": 1.7432742065724276e-05, "loss": 0.16329047679901124, "step": 9730 }, { "epoch": 0.2565664481417701, "grad_norm": 3.8227269649505615, "learning_rate": 1.7427031985077785e-05, "loss": 0.16140151023864746, "step": 9740 }, { "epoch": 0.25682986338626884, "grad_norm": 2.8809683322906494, "learning_rate": 1.7421316498716986e-05, "loss": 0.14520742893218994, "step": 9750 }, { "epoch": 0.2570932786307676, "grad_norm": 3.5507254600524902, "learning_rate": 1.741559561080185e-05, "loss": 0.14741086959838867, "step": 9760 }, { "epoch": 0.2573566938752663, "grad_norm": 1.7972016334533691, "learning_rate": 1.74098693254963e-05, "loss": 0.13928165435791015, "step": 9770 }, { "epoch": 0.257620109119765, "grad_norm": 1.5250829458236694, "learning_rate": 1.7404137646968162e-05, "loss": 0.1630263090133667, "step": 9780 }, { "epoch": 0.25788352436426376, "grad_norm": 2.2166237831115723, "learning_rate": 1.7398400579389204e-05, "loss": 0.14721161127090454, "step": 9790 }, { "epoch": 0.2581469396087625, "grad_norm": 3.91241192817688, "learning_rate": 1.739265812693511e-05, "loss": 0.16540660858154296, "step": 9800 }, { "epoch": 0.25841035485326125, "grad_norm": 3.897055149078369, "learning_rate": 1.7386910293785485e-05, "loss": 0.1545685648918152, "step": 9810 }, { "epoch": 0.25867377009776, "grad_norm": 2.928481101989746, "learning_rate": 1.738115708412385e-05, "loss": 0.15302250385284424, "step": 9820 }, { "epoch": 0.25893718534225874, "grad_norm": 2.064818859100342, "learning_rate": 1.737539850213764e-05, "loss": 0.16262331008911132, "step": 9830 }, { "epoch": 0.2592006005867575, "grad_norm": 2.507858991622925, "learning_rate": 1.7369634552018196e-05, "loss": 0.15317468643188475, "step": 9840 }, { "epoch": 0.25946401583125617, "grad_norm": 1.463276982307434, "learning_rate": 1.7363865237960773e-05, "loss": 0.1568153977394104, "step": 9850 }, { "epoch": 0.2597274310757549, "grad_norm": 2.670776128768921, "learning_rate": 1.7358090564164524e-05, "loss": 0.15294055938720702, "step": 9860 }, { "epoch": 0.25999084632025365, "grad_norm": 1.828546166419983, "learning_rate": 1.735231053483251e-05, "loss": 0.1624011993408203, "step": 9870 }, { "epoch": 0.2602542615647524, "grad_norm": 0.9152469635009766, "learning_rate": 1.734652515417168e-05, "loss": 0.17207159996032714, "step": 9880 }, { "epoch": 0.26051767680925114, "grad_norm": 1.383907437324524, "learning_rate": 1.7340734426392894e-05, "loss": 0.15778086185455323, "step": 9890 }, { "epoch": 0.2607810920537499, "grad_norm": 2.0351905822753906, "learning_rate": 1.7334938355710876e-05, "loss": 0.1632934808731079, "step": 9900 }, { "epoch": 0.26104450729824863, "grad_norm": 1.7019355297088623, "learning_rate": 1.7329136946344272e-05, "loss": 0.16357917785644532, "step": 9910 }, { "epoch": 0.2613079225427474, "grad_norm": 1.5452851057052612, "learning_rate": 1.7323330202515592e-05, "loss": 0.16083266735076904, "step": 9920 }, { "epoch": 0.2615713377872461, "grad_norm": 4.568751811981201, "learning_rate": 1.731751812845124e-05, "loss": 0.17630295753479003, "step": 9930 }, { "epoch": 0.2618347530317448, "grad_norm": 1.7185641527175903, "learning_rate": 1.7311700728381486e-05, "loss": 0.15100091695785522, "step": 9940 }, { "epoch": 0.26209816827624355, "grad_norm": 3.3731424808502197, "learning_rate": 1.7305878006540483e-05, "loss": 0.16083929538726807, "step": 9950 }, { "epoch": 0.2623615835207423, "grad_norm": 1.8167479038238525, "learning_rate": 1.7300049967166273e-05, "loss": 0.15495245456695556, "step": 9960 }, { "epoch": 0.26262499876524104, "grad_norm": 1.4870405197143555, "learning_rate": 1.729421661450074e-05, "loss": 0.14823269844055176, "step": 9970 }, { "epoch": 0.2628884140097398, "grad_norm": 2.0741069316864014, "learning_rate": 1.728837795278966e-05, "loss": 0.14875742197036743, "step": 9980 }, { "epoch": 0.2631518292542385, "grad_norm": 3.630723714828491, "learning_rate": 1.7282533986282656e-05, "loss": 0.16729702949523925, "step": 9990 }, { "epoch": 0.26341524449873727, "grad_norm": 11.076537132263184, "learning_rate": 1.7276684719233222e-05, "loss": 0.1777113199234009, "step": 10000 }, { "epoch": 0.263678659743236, "grad_norm": 2.0820112228393555, "learning_rate": 1.7270830155898704e-05, "loss": 0.18023130893707276, "step": 10010 }, { "epoch": 0.26394207498773475, "grad_norm": 1.5838850736618042, "learning_rate": 1.726497030054031e-05, "loss": 0.16350300312042237, "step": 10020 }, { "epoch": 0.26420549023223344, "grad_norm": 0.7173135280609131, "learning_rate": 1.72591051574231e-05, "loss": 0.15490198135375977, "step": 10030 }, { "epoch": 0.2644689054767322, "grad_norm": 2.597554922103882, "learning_rate": 1.7253234730815964e-05, "loss": 0.1686321496963501, "step": 10040 }, { "epoch": 0.26473232072123093, "grad_norm": 0.848781168460846, "learning_rate": 1.724735902499166e-05, "loss": 0.16521143913269043, "step": 10050 }, { "epoch": 0.2649957359657297, "grad_norm": 0.7671242356300354, "learning_rate": 1.7241478044226782e-05, "loss": 0.1631824016571045, "step": 10060 }, { "epoch": 0.2652591512102284, "grad_norm": 1.2165613174438477, "learning_rate": 1.723559179280176e-05, "loss": 0.15884581804275513, "step": 10070 }, { "epoch": 0.26552256645472716, "grad_norm": 2.068286895751953, "learning_rate": 1.722970027500086e-05, "loss": 0.16869674921035765, "step": 10080 }, { "epoch": 0.2657859816992259, "grad_norm": 3.699312925338745, "learning_rate": 1.7223803495112185e-05, "loss": 0.16106823682785035, "step": 10090 }, { "epoch": 0.26604939694372465, "grad_norm": 1.186835527420044, "learning_rate": 1.7217901457427667e-05, "loss": 0.1457056760787964, "step": 10100 }, { "epoch": 0.26631281218822334, "grad_norm": 4.065095901489258, "learning_rate": 1.7211994166243058e-05, "loss": 0.16350131034851073, "step": 10110 }, { "epoch": 0.2665762274327221, "grad_norm": 1.383087396621704, "learning_rate": 1.7206081625857943e-05, "loss": 0.16764931678771972, "step": 10120 }, { "epoch": 0.2668396426772208, "grad_norm": 3.3859035968780518, "learning_rate": 1.7200163840575728e-05, "loss": 0.164072847366333, "step": 10130 }, { "epoch": 0.26710305792171957, "grad_norm": 3.2696011066436768, "learning_rate": 1.7194240814703626e-05, "loss": 0.1616593599319458, "step": 10140 }, { "epoch": 0.2673664731662183, "grad_norm": 1.8864432573318481, "learning_rate": 1.7188312552552672e-05, "loss": 0.16077418327331544, "step": 10150 }, { "epoch": 0.26762988841071705, "grad_norm": 3.399134635925293, "learning_rate": 1.718237905843771e-05, "loss": 0.15946967601776124, "step": 10160 }, { "epoch": 0.2678933036552158, "grad_norm": 2.8217761516571045, "learning_rate": 1.71764403366774e-05, "loss": 0.16016331911087037, "step": 10170 }, { "epoch": 0.26815671889971454, "grad_norm": 4.409322738647461, "learning_rate": 1.7170496391594194e-05, "loss": 0.15434545278549194, "step": 10180 }, { "epoch": 0.2684201341442133, "grad_norm": 1.3694844245910645, "learning_rate": 1.716454722751435e-05, "loss": 0.15575380325317384, "step": 10190 }, { "epoch": 0.268683549388712, "grad_norm": 1.625174641609192, "learning_rate": 1.715859284876793e-05, "loss": 0.16137052774429322, "step": 10200 }, { "epoch": 0.2689469646332107, "grad_norm": 0.8492891788482666, "learning_rate": 1.7152633259688788e-05, "loss": 0.1565261721611023, "step": 10210 }, { "epoch": 0.26921037987770946, "grad_norm": 6.200183391571045, "learning_rate": 1.714666846461457e-05, "loss": 0.1565410852432251, "step": 10220 }, { "epoch": 0.2694737951222082, "grad_norm": 2.3805649280548096, "learning_rate": 1.7140698467886707e-05, "loss": 0.15452203750610352, "step": 10230 }, { "epoch": 0.26973721036670695, "grad_norm": 1.4319738149642944, "learning_rate": 1.713472327385042e-05, "loss": 0.15409393310546876, "step": 10240 }, { "epoch": 0.2700006256112057, "grad_norm": 2.808387041091919, "learning_rate": 1.7128742886854723e-05, "loss": 0.14627771377563475, "step": 10250 }, { "epoch": 0.27026404085570443, "grad_norm": 2.580510377883911, "learning_rate": 1.712275731125239e-05, "loss": 0.16169955730438232, "step": 10260 }, { "epoch": 0.2705274561002032, "grad_norm": 7.338343620300293, "learning_rate": 1.7116766551399986e-05, "loss": 0.16136958599090576, "step": 10270 }, { "epoch": 0.2707908713447019, "grad_norm": 1.1913007497787476, "learning_rate": 1.711077061165784e-05, "loss": 0.16001569032669066, "step": 10280 }, { "epoch": 0.2710542865892006, "grad_norm": 4.615795135498047, "learning_rate": 1.710476949639006e-05, "loss": 0.16886864900588988, "step": 10290 }, { "epoch": 0.27131770183369935, "grad_norm": 3.703213930130005, "learning_rate": 1.7098763209964514e-05, "loss": 0.15819602012634276, "step": 10300 }, { "epoch": 0.2715811170781981, "grad_norm": 1.838748812675476, "learning_rate": 1.7092751756752838e-05, "loss": 0.15289804935455323, "step": 10310 }, { "epoch": 0.27184453232269684, "grad_norm": 1.7823153734207153, "learning_rate": 1.7086735141130423e-05, "loss": 0.16462527513504027, "step": 10320 }, { "epoch": 0.2721079475671956, "grad_norm": 4.1352972984313965, "learning_rate": 1.7080713367476428e-05, "loss": 0.15538666248321534, "step": 10330 }, { "epoch": 0.27237136281169433, "grad_norm": 0.9100702404975891, "learning_rate": 1.7074686440173752e-05, "loss": 0.163918399810791, "step": 10340 }, { "epoch": 0.2726347780561931, "grad_norm": 2.388869047164917, "learning_rate": 1.7068654363609058e-05, "loss": 0.1499672055244446, "step": 10350 }, { "epoch": 0.2728981933006918, "grad_norm": 2.5830633640289307, "learning_rate": 1.7062617142172754e-05, "loss": 0.1751251459121704, "step": 10360 }, { "epoch": 0.2731616085451905, "grad_norm": 4.386563301086426, "learning_rate": 1.7056574780258986e-05, "loss": 0.16410412788391113, "step": 10370 }, { "epoch": 0.27342502378968925, "grad_norm": 3.073807716369629, "learning_rate": 1.705052728226564e-05, "loss": 0.15158367156982422, "step": 10380 }, { "epoch": 0.273688439034188, "grad_norm": 1.7759929895401, "learning_rate": 1.704447465259436e-05, "loss": 0.15631163120269775, "step": 10390 }, { "epoch": 0.27395185427868674, "grad_norm": 3.4994685649871826, "learning_rate": 1.7038416895650504e-05, "loss": 0.15987308025360109, "step": 10400 }, { "epoch": 0.2742152695231855, "grad_norm": 2.602501392364502, "learning_rate": 1.7032354015843162e-05, "loss": 0.16555479764938355, "step": 10410 }, { "epoch": 0.2744786847676842, "grad_norm": 5.801226615905762, "learning_rate": 1.7026286017585168e-05, "loss": 0.15378559827804567, "step": 10420 }, { "epoch": 0.27474210001218297, "grad_norm": 1.916266679763794, "learning_rate": 1.7020212905293072e-05, "loss": 0.16329150199890136, "step": 10430 }, { "epoch": 0.2750055152566817, "grad_norm": 1.525369644165039, "learning_rate": 1.7014134683387147e-05, "loss": 0.1690939784049988, "step": 10440 }, { "epoch": 0.27526893050118045, "grad_norm": 1.7643418312072754, "learning_rate": 1.7008051356291382e-05, "loss": 0.15270731449127198, "step": 10450 }, { "epoch": 0.27553234574567914, "grad_norm": 1.920602560043335, "learning_rate": 1.7001962928433488e-05, "loss": 0.16699979305267335, "step": 10460 }, { "epoch": 0.2757957609901779, "grad_norm": 2.5080466270446777, "learning_rate": 1.699586940424488e-05, "loss": 0.17522438764572143, "step": 10470 }, { "epoch": 0.27605917623467663, "grad_norm": 1.4007408618927002, "learning_rate": 1.6989770788160697e-05, "loss": 0.16394779682159424, "step": 10480 }, { "epoch": 0.2763225914791754, "grad_norm": 4.495804309844971, "learning_rate": 1.6983667084619768e-05, "loss": 0.16255930662155152, "step": 10490 }, { "epoch": 0.2765860067236741, "grad_norm": 3.802591562271118, "learning_rate": 1.6977558298064636e-05, "loss": 0.16244508028030397, "step": 10500 }, { "epoch": 0.27684942196817286, "grad_norm": 2.7910852432250977, "learning_rate": 1.6971444432941533e-05, "loss": 0.1475851058959961, "step": 10510 }, { "epoch": 0.2771128372126716, "grad_norm": 1.7701960802078247, "learning_rate": 1.69653254937004e-05, "loss": 0.16310672760009765, "step": 10520 }, { "epoch": 0.27737625245717035, "grad_norm": 1.283119559288025, "learning_rate": 1.6959201484794867e-05, "loss": 0.15813026428222657, "step": 10530 }, { "epoch": 0.2776396677016691, "grad_norm": 2.072066307067871, "learning_rate": 1.6953072410682243e-05, "loss": 0.16705491542816162, "step": 10540 }, { "epoch": 0.2779030829461678, "grad_norm": 4.808235168457031, "learning_rate": 1.694693827582354e-05, "loss": 0.16395436525344848, "step": 10550 }, { "epoch": 0.2781664981906665, "grad_norm": 2.7582685947418213, "learning_rate": 1.694079908468344e-05, "loss": 0.15509830713272094, "step": 10560 }, { "epoch": 0.27842991343516527, "grad_norm": 2.1014819145202637, "learning_rate": 1.6934654841730318e-05, "loss": 0.16608026027679443, "step": 10570 }, { "epoch": 0.278693328679664, "grad_norm": 1.773779034614563, "learning_rate": 1.6928505551436215e-05, "loss": 0.16824166774749755, "step": 10580 }, { "epoch": 0.27895674392416275, "grad_norm": 4.155906677246094, "learning_rate": 1.6922351218276852e-05, "loss": 0.16647238731384278, "step": 10590 }, { "epoch": 0.2792201591686615, "grad_norm": 3.481778860092163, "learning_rate": 1.6916191846731615e-05, "loss": 0.15333452224731445, "step": 10600 }, { "epoch": 0.27948357441316024, "grad_norm": 5.008841514587402, "learning_rate": 1.6910027441283562e-05, "loss": 0.1720367670059204, "step": 10610 }, { "epoch": 0.279746989657659, "grad_norm": 3.1764724254608154, "learning_rate": 1.6903858006419413e-05, "loss": 0.1763388156890869, "step": 10620 }, { "epoch": 0.2800104049021577, "grad_norm": 2.3889262676239014, "learning_rate": 1.689768354662955e-05, "loss": 0.1588399052619934, "step": 10630 }, { "epoch": 0.2802738201466564, "grad_norm": 4.093148708343506, "learning_rate": 1.689150406640801e-05, "loss": 0.16256248950958252, "step": 10640 }, { "epoch": 0.28053723539115516, "grad_norm": 1.7264376878738403, "learning_rate": 1.6885319570252485e-05, "loss": 0.16111867427825927, "step": 10650 }, { "epoch": 0.2808006506356539, "grad_norm": 4.349819183349609, "learning_rate": 1.687913006266432e-05, "loss": 0.15827925205230714, "step": 10660 }, { "epoch": 0.28106406588015265, "grad_norm": 3.8866806030273438, "learning_rate": 1.6872935548148503e-05, "loss": 0.16570439338684081, "step": 10670 }, { "epoch": 0.2813274811246514, "grad_norm": 4.116086959838867, "learning_rate": 1.686673603121367e-05, "loss": 0.15983837842941284, "step": 10680 }, { "epoch": 0.28159089636915013, "grad_norm": 5.411882400512695, "learning_rate": 1.6860531516372102e-05, "loss": 0.16482877731323242, "step": 10690 }, { "epoch": 0.2818543116136489, "grad_norm": 2.9221436977386475, "learning_rate": 1.6854322008139704e-05, "loss": 0.15832579135894775, "step": 10700 }, { "epoch": 0.2821177268581476, "grad_norm": 1.0492572784423828, "learning_rate": 1.6848107511036026e-05, "loss": 0.16055817604064943, "step": 10710 }, { "epoch": 0.2823811421026463, "grad_norm": 2.3545992374420166, "learning_rate": 1.6841888029584255e-05, "loss": 0.14273910522460936, "step": 10720 }, { "epoch": 0.28264455734714505, "grad_norm": 1.8865089416503906, "learning_rate": 1.6835663568311187e-05, "loss": 0.14650142192840576, "step": 10730 }, { "epoch": 0.2829079725916438, "grad_norm": 2.8795833587646484, "learning_rate": 1.682943413174726e-05, "loss": 0.15772312879562378, "step": 10740 }, { "epoch": 0.28317138783614254, "grad_norm": 1.4322642087936401, "learning_rate": 1.6823199724426523e-05, "loss": 0.16657676696777343, "step": 10750 }, { "epoch": 0.2834348030806413, "grad_norm": 2.0002310276031494, "learning_rate": 1.6816960350886653e-05, "loss": 0.16989047527313234, "step": 10760 }, { "epoch": 0.28369821832514003, "grad_norm": 4.291472434997559, "learning_rate": 1.681071601566893e-05, "loss": 0.16872618198394776, "step": 10770 }, { "epoch": 0.28396163356963877, "grad_norm": 1.3855549097061157, "learning_rate": 1.6804466723318248e-05, "loss": 0.16596055030822754, "step": 10780 }, { "epoch": 0.2842250488141375, "grad_norm": 4.836543083190918, "learning_rate": 1.6798212478383118e-05, "loss": 0.16381547451019288, "step": 10790 }, { "epoch": 0.28448846405863626, "grad_norm": 4.094981670379639, "learning_rate": 1.6791953285415645e-05, "loss": 0.1588038682937622, "step": 10800 }, { "epoch": 0.28475187930313495, "grad_norm": 2.3578500747680664, "learning_rate": 1.678568914897154e-05, "loss": 0.16371567249298097, "step": 10810 }, { "epoch": 0.2850152945476337, "grad_norm": 5.771791934967041, "learning_rate": 1.6779420073610116e-05, "loss": 0.15383846759796144, "step": 10820 }, { "epoch": 0.28527870979213243, "grad_norm": 6.648291110992432, "learning_rate": 1.6773146063894273e-05, "loss": 0.1786036252975464, "step": 10830 }, { "epoch": 0.2855421250366312, "grad_norm": 6.923746109008789, "learning_rate": 1.6766867124390505e-05, "loss": 0.1639509081840515, "step": 10840 }, { "epoch": 0.2858055402811299, "grad_norm": 3.695939540863037, "learning_rate": 1.6760583259668897e-05, "loss": 0.16150848865509032, "step": 10850 }, { "epoch": 0.28606895552562867, "grad_norm": 1.576560378074646, "learning_rate": 1.6754294474303116e-05, "loss": 0.15893261432647704, "step": 10860 }, { "epoch": 0.2863323707701274, "grad_norm": 1.4878910779953003, "learning_rate": 1.674800077287041e-05, "loss": 0.16546232700347902, "step": 10870 }, { "epoch": 0.28659578601462615, "grad_norm": 2.8307135105133057, "learning_rate": 1.6741702159951613e-05, "loss": 0.16323212385177613, "step": 10880 }, { "epoch": 0.2868592012591249, "grad_norm": 1.6742745637893677, "learning_rate": 1.6735398640131117e-05, "loss": 0.16431151628494262, "step": 10890 }, { "epoch": 0.2871226165036236, "grad_norm": 1.6062328815460205, "learning_rate": 1.6729090217996904e-05, "loss": 0.1649436116218567, "step": 10900 }, { "epoch": 0.28738603174812233, "grad_norm": 1.986021637916565, "learning_rate": 1.6722776898140513e-05, "loss": 0.1560852646827698, "step": 10910 }, { "epoch": 0.28764944699262107, "grad_norm": 7.019567489624023, "learning_rate": 1.6716458685157047e-05, "loss": 0.16632803678512573, "step": 10920 }, { "epoch": 0.2879128622371198, "grad_norm": 1.7575145959854126, "learning_rate": 1.671013558364518e-05, "loss": 0.1558562159538269, "step": 10930 }, { "epoch": 0.28817627748161856, "grad_norm": 7.400558948516846, "learning_rate": 1.6703807598207132e-05, "loss": 0.157819926738739, "step": 10940 }, { "epoch": 0.2884396927261173, "grad_norm": 2.207105875015259, "learning_rate": 1.669747473344869e-05, "loss": 0.15089609622955322, "step": 10950 }, { "epoch": 0.28870310797061605, "grad_norm": 3.780820369720459, "learning_rate": 1.6691136993979183e-05, "loss": 0.14185709953308107, "step": 10960 }, { "epoch": 0.2889665232151148, "grad_norm": 2.224398136138916, "learning_rate": 1.6684794384411494e-05, "loss": 0.15644978284835814, "step": 10970 }, { "epoch": 0.2892299384596135, "grad_norm": 1.3716332912445068, "learning_rate": 1.667844690936204e-05, "loss": 0.16419031620025634, "step": 10980 }, { "epoch": 0.2894933537041122, "grad_norm": 2.7148420810699463, "learning_rate": 1.6672094573450798e-05, "loss": 0.1745321273803711, "step": 10990 }, { "epoch": 0.28975676894861097, "grad_norm": 2.478494167327881, "learning_rate": 1.6665737381301263e-05, "loss": 0.16610963344573976, "step": 11000 }, { "epoch": 0.2900201841931097, "grad_norm": 1.3090566396713257, "learning_rate": 1.665937533754048e-05, "loss": 0.15552325248718263, "step": 11010 }, { "epoch": 0.29028359943760845, "grad_norm": 2.194580316543579, "learning_rate": 1.6653008446799014e-05, "loss": 0.1628174066543579, "step": 11020 }, { "epoch": 0.2905470146821072, "grad_norm": 0.8884150385856628, "learning_rate": 1.6646636713710967e-05, "loss": 0.14837026596069336, "step": 11030 }, { "epoch": 0.29081042992660594, "grad_norm": 3.149526596069336, "learning_rate": 1.664026014291396e-05, "loss": 0.1669802188873291, "step": 11040 }, { "epoch": 0.2910738451711047, "grad_norm": 4.107648849487305, "learning_rate": 1.6633878739049135e-05, "loss": 0.15684006214141846, "step": 11050 }, { "epoch": 0.2913372604156034, "grad_norm": 3.4728708267211914, "learning_rate": 1.6627492506761155e-05, "loss": 0.1639029264450073, "step": 11060 }, { "epoch": 0.2916006756601021, "grad_norm": 1.7191388607025146, "learning_rate": 1.6621101450698193e-05, "loss": 0.16856825351715088, "step": 11070 }, { "epoch": 0.29186409090460086, "grad_norm": 2.933284044265747, "learning_rate": 1.6614705575511937e-05, "loss": 0.16771326065063477, "step": 11080 }, { "epoch": 0.2921275061490996, "grad_norm": 4.066222190856934, "learning_rate": 1.660830488585758e-05, "loss": 0.1561591148376465, "step": 11090 }, { "epoch": 0.29239092139359835, "grad_norm": 2.209397792816162, "learning_rate": 1.660189938639382e-05, "loss": 0.15462486743927, "step": 11100 }, { "epoch": 0.2926543366380971, "grad_norm": 1.3880767822265625, "learning_rate": 1.6595489081782858e-05, "loss": 0.1616670608520508, "step": 11110 }, { "epoch": 0.29291775188259583, "grad_norm": 1.6143261194229126, "learning_rate": 1.6589073976690387e-05, "loss": 0.1686730146408081, "step": 11120 }, { "epoch": 0.2931811671270946, "grad_norm": 3.589939832687378, "learning_rate": 1.65826540757856e-05, "loss": 0.1568726658821106, "step": 11130 }, { "epoch": 0.2934445823715933, "grad_norm": 1.768761157989502, "learning_rate": 1.6576229383741177e-05, "loss": 0.16637134552001953, "step": 11140 }, { "epoch": 0.29370799761609206, "grad_norm": 2.976274013519287, "learning_rate": 1.6569799905233285e-05, "loss": 0.15895651578903197, "step": 11150 }, { "epoch": 0.29397141286059075, "grad_norm": 1.7120509147644043, "learning_rate": 1.6563365644941578e-05, "loss": 0.1609356164932251, "step": 11160 }, { "epoch": 0.2942348281050895, "grad_norm": 2.496975898742676, "learning_rate": 1.6556926607549186e-05, "loss": 0.15306599140167237, "step": 11170 }, { "epoch": 0.29449824334958824, "grad_norm": 3.9029624462127686, "learning_rate": 1.655048279774272e-05, "loss": 0.1650075674057007, "step": 11180 }, { "epoch": 0.294761658594087, "grad_norm": 3.1993370056152344, "learning_rate": 1.6544034220212262e-05, "loss": 0.15786528587341309, "step": 11190 }, { "epoch": 0.2950250738385857, "grad_norm": 1.483444333076477, "learning_rate": 1.6537580879651364e-05, "loss": 0.15552499294281005, "step": 11200 }, { "epoch": 0.29528848908308447, "grad_norm": 1.4783592224121094, "learning_rate": 1.6531122780757046e-05, "loss": 0.1659780502319336, "step": 11210 }, { "epoch": 0.2955519043275832, "grad_norm": 5.345235824584961, "learning_rate": 1.6524659928229792e-05, "loss": 0.16753364801406861, "step": 11220 }, { "epoch": 0.29581531957208196, "grad_norm": 6.206290245056152, "learning_rate": 1.6518192326773546e-05, "loss": 0.16535592079162598, "step": 11230 }, { "epoch": 0.29607873481658065, "grad_norm": 3.1296305656433105, "learning_rate": 1.6511719981095703e-05, "loss": 0.17163039445877076, "step": 11240 }, { "epoch": 0.2963421500610794, "grad_norm": 4.1353278160095215, "learning_rate": 1.6505242895907122e-05, "loss": 0.16447994709014893, "step": 11250 }, { "epoch": 0.29660556530557813, "grad_norm": 3.502366542816162, "learning_rate": 1.64987610759221e-05, "loss": 0.16262385845184327, "step": 11260 }, { "epoch": 0.2968689805500769, "grad_norm": 4.42378568649292, "learning_rate": 1.6492274525858386e-05, "loss": 0.1602785110473633, "step": 11270 }, { "epoch": 0.2971323957945756, "grad_norm": 1.835993766784668, "learning_rate": 1.648578325043717e-05, "loss": 0.16241788864135742, "step": 11280 }, { "epoch": 0.29739581103907436, "grad_norm": 0.7924180626869202, "learning_rate": 1.6479287254383084e-05, "loss": 0.16942483186721802, "step": 11290 }, { "epoch": 0.2976592262835731, "grad_norm": 2.008800745010376, "learning_rate": 1.6472786542424196e-05, "loss": 0.14841234683990479, "step": 11300 }, { "epoch": 0.29792264152807185, "grad_norm": 4.722722053527832, "learning_rate": 1.6466281119292004e-05, "loss": 0.16634178161621094, "step": 11310 }, { "epoch": 0.2981860567725706, "grad_norm": 2.40830397605896, "learning_rate": 1.6459770989721433e-05, "loss": 0.16720659732818605, "step": 11320 }, { "epoch": 0.2984494720170693, "grad_norm": 3.1877503395080566, "learning_rate": 1.6453256158450835e-05, "loss": 0.15803160667419433, "step": 11330 }, { "epoch": 0.298712887261568, "grad_norm": 1.4349254369735718, "learning_rate": 1.644673663022199e-05, "loss": 0.1630706548690796, "step": 11340 }, { "epoch": 0.29897630250606677, "grad_norm": 3.2877790927886963, "learning_rate": 1.6440212409780088e-05, "loss": 0.16172118186950685, "step": 11350 }, { "epoch": 0.2992397177505655, "grad_norm": 1.2220301628112793, "learning_rate": 1.643368350187374e-05, "loss": 0.17258858680725098, "step": 11360 }, { "epoch": 0.29950313299506426, "grad_norm": 2.9151597023010254, "learning_rate": 1.6427149911254973e-05, "loss": 0.15740426778793334, "step": 11370 }, { "epoch": 0.299766548239563, "grad_norm": 2.3563950061798096, "learning_rate": 1.64206116426792e-05, "loss": 0.14992947578430177, "step": 11380 }, { "epoch": 0.30002996348406175, "grad_norm": 1.6129889488220215, "learning_rate": 1.641406870090527e-05, "loss": 0.14965425729751586, "step": 11390 }, { "epoch": 0.3002933787285605, "grad_norm": 2.4202077388763428, "learning_rate": 1.640752109069541e-05, "loss": 0.1494004487991333, "step": 11400 }, { "epoch": 0.30055679397305923, "grad_norm": 0.8249907493591309, "learning_rate": 1.640096881681526e-05, "loss": 0.16353873014450074, "step": 11410 }, { "epoch": 0.3008202092175579, "grad_norm": 1.859605312347412, "learning_rate": 1.6394411884033844e-05, "loss": 0.16566561460494994, "step": 11420 }, { "epoch": 0.30108362446205666, "grad_norm": 1.841460108757019, "learning_rate": 1.638785029712358e-05, "loss": 0.15565207004547119, "step": 11430 }, { "epoch": 0.3013470397065554, "grad_norm": 2.729956865310669, "learning_rate": 1.638128406086027e-05, "loss": 0.1689429759979248, "step": 11440 }, { "epoch": 0.30161045495105415, "grad_norm": 1.577181339263916, "learning_rate": 1.6374713180023114e-05, "loss": 0.15535104274749756, "step": 11450 }, { "epoch": 0.3018738701955529, "grad_norm": 1.1323635578155518, "learning_rate": 1.6368137659394673e-05, "loss": 0.1634555697441101, "step": 11460 }, { "epoch": 0.30213728544005164, "grad_norm": 4.3703460693359375, "learning_rate": 1.63615575037609e-05, "loss": 0.1795522689819336, "step": 11470 }, { "epoch": 0.3024007006845504, "grad_norm": 5.13347864151001, "learning_rate": 1.6354972717911115e-05, "loss": 0.14439539909362792, "step": 11480 }, { "epoch": 0.3026641159290491, "grad_norm": 3.9094934463500977, "learning_rate": 1.634838330663801e-05, "loss": 0.15809285640716553, "step": 11490 }, { "epoch": 0.3029275311735478, "grad_norm": 2.5500543117523193, "learning_rate": 1.6341789274737644e-05, "loss": 0.15822601318359375, "step": 11500 }, { "epoch": 0.30319094641804656, "grad_norm": 3.0966455936431885, "learning_rate": 1.6335190627009432e-05, "loss": 0.1716955542564392, "step": 11510 }, { "epoch": 0.3034543616625453, "grad_norm": 6.167402267456055, "learning_rate": 1.632858736825616e-05, "loss": 0.15499656200408934, "step": 11520 }, { "epoch": 0.30371777690704405, "grad_norm": 1.8247241973876953, "learning_rate": 1.632197950328396e-05, "loss": 0.15386179685592652, "step": 11530 }, { "epoch": 0.3039811921515428, "grad_norm": 4.2076897621154785, "learning_rate": 1.631536703690233e-05, "loss": 0.16298623085021974, "step": 11540 }, { "epoch": 0.30424460739604153, "grad_norm": 2.373699188232422, "learning_rate": 1.6308749973924096e-05, "loss": 0.1614643692970276, "step": 11550 }, { "epoch": 0.3045080226405403, "grad_norm": 2.10640811920166, "learning_rate": 1.6302128319165454e-05, "loss": 0.1705669641494751, "step": 11560 }, { "epoch": 0.304771437885039, "grad_norm": 1.8461230993270874, "learning_rate": 1.629550207744592e-05, "loss": 0.17101314067840576, "step": 11570 }, { "epoch": 0.30503485312953776, "grad_norm": 1.2124334573745728, "learning_rate": 1.6288871253588366e-05, "loss": 0.16561563014984132, "step": 11580 }, { "epoch": 0.30529826837403645, "grad_norm": 4.241819858551025, "learning_rate": 1.6282235852418983e-05, "loss": 0.17119933366775514, "step": 11590 }, { "epoch": 0.3055616836185352, "grad_norm": 2.399413585662842, "learning_rate": 1.6275595878767315e-05, "loss": 0.1609861135482788, "step": 11600 }, { "epoch": 0.30582509886303394, "grad_norm": 1.7344646453857422, "learning_rate": 1.6268951337466212e-05, "loss": 0.1591123342514038, "step": 11610 }, { "epoch": 0.3060885141075327, "grad_norm": 0.9423506855964661, "learning_rate": 1.626230223335186e-05, "loss": 0.17393941879272462, "step": 11620 }, { "epoch": 0.3063519293520314, "grad_norm": 1.6361570358276367, "learning_rate": 1.6255648571263762e-05, "loss": 0.16754275560379028, "step": 11630 }, { "epoch": 0.30661534459653017, "grad_norm": 4.321982383728027, "learning_rate": 1.6248990356044742e-05, "loss": 0.17104319334030152, "step": 11640 }, { "epoch": 0.3068787598410289, "grad_norm": 3.025998115539551, "learning_rate": 1.6242327592540942e-05, "loss": 0.15358778238296508, "step": 11650 }, { "epoch": 0.30714217508552766, "grad_norm": 1.4970589876174927, "learning_rate": 1.62356602856018e-05, "loss": 0.1610827088356018, "step": 11660 }, { "epoch": 0.3074055903300264, "grad_norm": 2.2927379608154297, "learning_rate": 1.6228988440080068e-05, "loss": 0.15118367671966554, "step": 11670 }, { "epoch": 0.3076690055745251, "grad_norm": 3.745365858078003, "learning_rate": 1.6222312060831814e-05, "loss": 0.148390531539917, "step": 11680 }, { "epoch": 0.30793242081902383, "grad_norm": 4.223616600036621, "learning_rate": 1.6215631152716387e-05, "loss": 0.16949806213378907, "step": 11690 }, { "epoch": 0.3081958360635226, "grad_norm": 2.0019824504852295, "learning_rate": 1.620894572059644e-05, "loss": 0.16564682722091675, "step": 11700 }, { "epoch": 0.3084592513080213, "grad_norm": 2.0329275131225586, "learning_rate": 1.6202255769337924e-05, "loss": 0.14533140659332275, "step": 11710 }, { "epoch": 0.30872266655252006, "grad_norm": 1.614762306213379, "learning_rate": 1.619556130381007e-05, "loss": 0.16447663307189941, "step": 11720 }, { "epoch": 0.3089860817970188, "grad_norm": 2.7015912532806396, "learning_rate": 1.61888623288854e-05, "loss": 0.17527570724487304, "step": 11730 }, { "epoch": 0.30924949704151755, "grad_norm": 2.2428197860717773, "learning_rate": 1.6182158849439718e-05, "loss": 0.14922010898590088, "step": 11740 }, { "epoch": 0.3095129122860163, "grad_norm": 2.734147787094116, "learning_rate": 1.61754508703521e-05, "loss": 0.16183679103851317, "step": 11750 }, { "epoch": 0.309776327530515, "grad_norm": 1.9051953554153442, "learning_rate": 1.6168738396504912e-05, "loss": 0.1681393265724182, "step": 11760 }, { "epoch": 0.3100397427750137, "grad_norm": 4.172499179840088, "learning_rate": 1.6162021432783773e-05, "loss": 0.15992714166641236, "step": 11770 }, { "epoch": 0.31030315801951247, "grad_norm": 2.8348045349121094, "learning_rate": 1.6155299984077584e-05, "loss": 0.15581147670745848, "step": 11780 }, { "epoch": 0.3105665732640112, "grad_norm": 1.2615724802017212, "learning_rate": 1.6148574055278507e-05, "loss": 0.13811025619506836, "step": 11790 }, { "epoch": 0.31082998850850996, "grad_norm": 2.314166784286499, "learning_rate": 1.6141843651281955e-05, "loss": 0.15315773487091064, "step": 11800 }, { "epoch": 0.3110934037530087, "grad_norm": 3.2651774883270264, "learning_rate": 1.6135108776986615e-05, "loss": 0.16868001222610474, "step": 11810 }, { "epoch": 0.31135681899750745, "grad_norm": 1.7288589477539062, "learning_rate": 1.612836943729442e-05, "loss": 0.14558889865875244, "step": 11820 }, { "epoch": 0.3116202342420062, "grad_norm": 2.423826217651367, "learning_rate": 1.6121625637110546e-05, "loss": 0.16968448162078859, "step": 11830 }, { "epoch": 0.31188364948650493, "grad_norm": 2.3068108558654785, "learning_rate": 1.6114877381343422e-05, "loss": 0.15538171529769898, "step": 11840 }, { "epoch": 0.3121470647310036, "grad_norm": 1.7929956912994385, "learning_rate": 1.610812467490472e-05, "loss": 0.17351195812225342, "step": 11850 }, { "epoch": 0.31241047997550236, "grad_norm": 2.6744048595428467, "learning_rate": 1.6101367522709356e-05, "loss": 0.15390434265136718, "step": 11860 }, { "epoch": 0.3126738952200011, "grad_norm": 3.47819447517395, "learning_rate": 1.6094605929675476e-05, "loss": 0.1524508476257324, "step": 11870 }, { "epoch": 0.31293731046449985, "grad_norm": 2.102189302444458, "learning_rate": 1.608783990072445e-05, "loss": 0.15964123010635375, "step": 11880 }, { "epoch": 0.3132007257089986, "grad_norm": 3.8845088481903076, "learning_rate": 1.6081069440780895e-05, "loss": 0.1592250108718872, "step": 11890 }, { "epoch": 0.31346414095349734, "grad_norm": 2.8978257179260254, "learning_rate": 1.607429455477264e-05, "loss": 0.14248621463775635, "step": 11900 }, { "epoch": 0.3137275561979961, "grad_norm": 1.2561264038085938, "learning_rate": 1.606751524763075e-05, "loss": 0.1456430196762085, "step": 11910 }, { "epoch": 0.3139909714424948, "grad_norm": 0.9297100901603699, "learning_rate": 1.606073152428948e-05, "loss": 0.15934667587280274, "step": 11920 }, { "epoch": 0.31425438668699357, "grad_norm": 3.0345137119293213, "learning_rate": 1.6053943389686323e-05, "loss": 0.15551772117614746, "step": 11930 }, { "epoch": 0.31451780193149226, "grad_norm": 2.327972888946533, "learning_rate": 1.604715084876198e-05, "loss": 0.15545198917388917, "step": 11940 }, { "epoch": 0.314781217175991, "grad_norm": 1.6620959043502808, "learning_rate": 1.6040353906460353e-05, "loss": 0.16576023101806642, "step": 11950 }, { "epoch": 0.31504463242048975, "grad_norm": 2.6404035091400146, "learning_rate": 1.6033552567728553e-05, "loss": 0.15008461475372314, "step": 11960 }, { "epoch": 0.3153080476649885, "grad_norm": 1.435070514678955, "learning_rate": 1.602674683751688e-05, "loss": 0.1646263360977173, "step": 11970 }, { "epoch": 0.31557146290948723, "grad_norm": 2.2753145694732666, "learning_rate": 1.6019936720778845e-05, "loss": 0.15797555446624756, "step": 11980 }, { "epoch": 0.315834878153986, "grad_norm": 2.388885736465454, "learning_rate": 1.601312222247114e-05, "loss": 0.158156156539917, "step": 11990 }, { "epoch": 0.3160982933984847, "grad_norm": 1.7996796369552612, "learning_rate": 1.6006303347553655e-05, "loss": 0.16361836194992066, "step": 12000 }, { "epoch": 0.31636170864298346, "grad_norm": 1.3037877082824707, "learning_rate": 1.5999480100989458e-05, "loss": 0.16221292018890382, "step": 12010 }, { "epoch": 0.31662512388748215, "grad_norm": 2.3339784145355225, "learning_rate": 1.5992652487744806e-05, "loss": 0.16223665475845336, "step": 12020 }, { "epoch": 0.3168885391319809, "grad_norm": 1.2668161392211914, "learning_rate": 1.5985820512789128e-05, "loss": 0.14765703678131104, "step": 12030 }, { "epoch": 0.31715195437647964, "grad_norm": 2.4096381664276123, "learning_rate": 1.5978984181095034e-05, "loss": 0.16710798740386962, "step": 12040 }, { "epoch": 0.3174153696209784, "grad_norm": 2.064910411834717, "learning_rate": 1.5972143497638294e-05, "loss": 0.16326072216033935, "step": 12050 }, { "epoch": 0.3176787848654771, "grad_norm": 2.3017289638519287, "learning_rate": 1.5965298467397855e-05, "loss": 0.15485696792602538, "step": 12060 }, { "epoch": 0.31794220010997587, "grad_norm": 1.6842786073684692, "learning_rate": 1.595844909535583e-05, "loss": 0.15627577304840087, "step": 12070 }, { "epoch": 0.3182056153544746, "grad_norm": 1.6097005605697632, "learning_rate": 1.595159538649748e-05, "loss": 0.15717239379882814, "step": 12080 }, { "epoch": 0.31846903059897336, "grad_norm": 3.961068630218506, "learning_rate": 1.5944737345811238e-05, "loss": 0.1624805212020874, "step": 12090 }, { "epoch": 0.3187324458434721, "grad_norm": 0.9142292737960815, "learning_rate": 1.5937874978288676e-05, "loss": 0.15565998554229737, "step": 12100 }, { "epoch": 0.3189958610879708, "grad_norm": 1.3366121053695679, "learning_rate": 1.5931008288924523e-05, "loss": 0.15465998649597168, "step": 12110 }, { "epoch": 0.31925927633246953, "grad_norm": 1.2931430339813232, "learning_rate": 1.5924137282716653e-05, "loss": 0.1516583204269409, "step": 12120 }, { "epoch": 0.3195226915769683, "grad_norm": 1.7568047046661377, "learning_rate": 1.591726196466608e-05, "loss": 0.16917104721069337, "step": 12130 }, { "epoch": 0.319786106821467, "grad_norm": 0.7370215058326721, "learning_rate": 1.5910382339776953e-05, "loss": 0.15607624053955077, "step": 12140 }, { "epoch": 0.32004952206596576, "grad_norm": 3.4593095779418945, "learning_rate": 1.5903498413056568e-05, "loss": 0.16011019945144653, "step": 12150 }, { "epoch": 0.3203129373104645, "grad_norm": 1.9487980604171753, "learning_rate": 1.5896610189515335e-05, "loss": 0.15653544664382935, "step": 12160 }, { "epoch": 0.32057635255496325, "grad_norm": 2.5409181118011475, "learning_rate": 1.588971767416681e-05, "loss": 0.16376049518585206, "step": 12170 }, { "epoch": 0.320839767799462, "grad_norm": 5.777863502502441, "learning_rate": 1.5882820872027654e-05, "loss": 0.16144341230392456, "step": 12180 }, { "epoch": 0.32110318304396074, "grad_norm": 0.8028665781021118, "learning_rate": 1.5875919788117662e-05, "loss": 0.16041166782379152, "step": 12190 }, { "epoch": 0.3213665982884594, "grad_norm": 1.030518889427185, "learning_rate": 1.586901442745974e-05, "loss": 0.16365211009979247, "step": 12200 }, { "epoch": 0.32163001353295817, "grad_norm": 2.6058382987976074, "learning_rate": 1.5862104795079903e-05, "loss": 0.1438262104988098, "step": 12210 }, { "epoch": 0.3218934287774569, "grad_norm": 1.732832431793213, "learning_rate": 1.5855190896007287e-05, "loss": 0.17418526411056517, "step": 12220 }, { "epoch": 0.32215684402195566, "grad_norm": 0.833755373954773, "learning_rate": 1.584827273527412e-05, "loss": 0.13889455795288086, "step": 12230 }, { "epoch": 0.3224202592664544, "grad_norm": 4.720329761505127, "learning_rate": 1.5841350317915745e-05, "loss": 0.15570368766784667, "step": 12240 }, { "epoch": 0.32268367451095314, "grad_norm": 1.2756150960922241, "learning_rate": 1.5834423648970588e-05, "loss": 0.1686248779296875, "step": 12250 }, { "epoch": 0.3229470897554519, "grad_norm": 3.0401723384857178, "learning_rate": 1.5827492733480183e-05, "loss": 0.16401468515396117, "step": 12260 }, { "epoch": 0.32321050499995063, "grad_norm": 2.5745770931243896, "learning_rate": 1.5820557576489145e-05, "loss": 0.14252951145172119, "step": 12270 }, { "epoch": 0.3234739202444493, "grad_norm": 0.9074382185935974, "learning_rate": 1.581361818304518e-05, "loss": 0.150801944732666, "step": 12280 }, { "epoch": 0.32373733548894806, "grad_norm": 1.486183524131775, "learning_rate": 1.5806674558199088e-05, "loss": 0.16367946863174437, "step": 12290 }, { "epoch": 0.3240007507334468, "grad_norm": 1.1924463510513306, "learning_rate": 1.5799726707004728e-05, "loss": 0.16685655117034912, "step": 12300 }, { "epoch": 0.32426416597794555, "grad_norm": 1.2315759658813477, "learning_rate": 1.5792774634519046e-05, "loss": 0.17318971157073976, "step": 12310 }, { "epoch": 0.3245275812224443, "grad_norm": 3.864956855773926, "learning_rate": 1.5785818345802068e-05, "loss": 0.1587338328361511, "step": 12320 }, { "epoch": 0.32479099646694304, "grad_norm": 1.2385776042938232, "learning_rate": 1.5778857845916874e-05, "loss": 0.15934497117996216, "step": 12330 }, { "epoch": 0.3250544117114418, "grad_norm": 1.7594445943832397, "learning_rate": 1.577189313992962e-05, "loss": 0.15253794193267822, "step": 12340 }, { "epoch": 0.3253178269559405, "grad_norm": 0.9924459457397461, "learning_rate": 1.5764924232909518e-05, "loss": 0.16712689399719238, "step": 12350 }, { "epoch": 0.32558124220043927, "grad_norm": 1.2778974771499634, "learning_rate": 1.5757951129928837e-05, "loss": 0.15383925437927246, "step": 12360 }, { "epoch": 0.32584465744493796, "grad_norm": 1.2864617109298706, "learning_rate": 1.5750973836062907e-05, "loss": 0.15076887607574463, "step": 12370 }, { "epoch": 0.3261080726894367, "grad_norm": 3.490239381790161, "learning_rate": 1.57439923563901e-05, "loss": 0.1721884250640869, "step": 12380 }, { "epoch": 0.32637148793393544, "grad_norm": 1.9297094345092773, "learning_rate": 1.573700669599184e-05, "loss": 0.15293848514556885, "step": 12390 }, { "epoch": 0.3266349031784342, "grad_norm": 2.378236770629883, "learning_rate": 1.573001685995259e-05, "loss": 0.1454631805419922, "step": 12400 }, { "epoch": 0.32689831842293293, "grad_norm": 4.187576770782471, "learning_rate": 1.572302285335985e-05, "loss": 0.16944684982299804, "step": 12410 }, { "epoch": 0.3271617336674317, "grad_norm": 2.146456003189087, "learning_rate": 1.5716024681304177e-05, "loss": 0.1557131290435791, "step": 12420 }, { "epoch": 0.3274251489119304, "grad_norm": 5.434981346130371, "learning_rate": 1.5709022348879124e-05, "loss": 0.15820029973983765, "step": 12430 }, { "epoch": 0.32768856415642916, "grad_norm": 2.033608913421631, "learning_rate": 1.5702015861181295e-05, "loss": 0.16097573041915894, "step": 12440 }, { "epoch": 0.3279519794009279, "grad_norm": 5.073653697967529, "learning_rate": 1.569500522331032e-05, "loss": 0.16276190280914307, "step": 12450 }, { "epoch": 0.3282153946454266, "grad_norm": 6.319255828857422, "learning_rate": 1.568799044036884e-05, "loss": 0.15394243001937866, "step": 12460 }, { "epoch": 0.32847880988992534, "grad_norm": 2.492699146270752, "learning_rate": 1.5680971517462516e-05, "loss": 0.14418224096298218, "step": 12470 }, { "epoch": 0.3287422251344241, "grad_norm": 5.766873836517334, "learning_rate": 1.5673948459700024e-05, "loss": 0.1394218921661377, "step": 12480 }, { "epoch": 0.3290056403789228, "grad_norm": 3.7925162315368652, "learning_rate": 1.5666921272193046e-05, "loss": 0.1521831750869751, "step": 12490 }, { "epoch": 0.32926905562342157, "grad_norm": 2.652134656906128, "learning_rate": 1.565988996005628e-05, "loss": 0.16195998191833497, "step": 12500 }, { "epoch": 0.3295324708679203, "grad_norm": 2.45097017288208, "learning_rate": 1.5652854528407407e-05, "loss": 0.15374367237091063, "step": 12510 }, { "epoch": 0.32979588611241906, "grad_norm": 1.7623239755630493, "learning_rate": 1.5645814982367125e-05, "loss": 0.15855705738067627, "step": 12520 }, { "epoch": 0.3300593013569178, "grad_norm": 2.0425469875335693, "learning_rate": 1.5638771327059117e-05, "loss": 0.16682995557785035, "step": 12530 }, { "epoch": 0.3303227166014165, "grad_norm": 3.8511199951171875, "learning_rate": 1.563172356761006e-05, "loss": 0.159567391872406, "step": 12540 }, { "epoch": 0.33058613184591523, "grad_norm": 0.9944903254508972, "learning_rate": 1.5624671709149618e-05, "loss": 0.1496408462524414, "step": 12550 }, { "epoch": 0.330849547090414, "grad_norm": 3.117342233657837, "learning_rate": 1.561761575681044e-05, "loss": 0.15251923799514772, "step": 12560 }, { "epoch": 0.3311129623349127, "grad_norm": 2.3128104209899902, "learning_rate": 1.5610555715728146e-05, "loss": 0.1513433575630188, "step": 12570 }, { "epoch": 0.33137637757941146, "grad_norm": 2.787945508956909, "learning_rate": 1.560349159104134e-05, "loss": 0.1608109712600708, "step": 12580 }, { "epoch": 0.3316397928239102, "grad_norm": 3.4443869590759277, "learning_rate": 1.55964233878916e-05, "loss": 0.15388593673706055, "step": 12590 }, { "epoch": 0.33190320806840895, "grad_norm": 1.2923465967178345, "learning_rate": 1.5589351111423474e-05, "loss": 0.15083856582641603, "step": 12600 }, { "epoch": 0.3321666233129077, "grad_norm": 2.383901834487915, "learning_rate": 1.5582274766784457e-05, "loss": 0.17322616577148436, "step": 12610 }, { "epoch": 0.33243003855740644, "grad_norm": 2.2239770889282227, "learning_rate": 1.557519435912503e-05, "loss": 0.14317835569381715, "step": 12620 }, { "epoch": 0.3326934538019051, "grad_norm": 2.637990951538086, "learning_rate": 1.556810989359861e-05, "loss": 0.15693941116333007, "step": 12630 }, { "epoch": 0.33295686904640387, "grad_norm": 4.352642059326172, "learning_rate": 1.5561021375361583e-05, "loss": 0.1590863823890686, "step": 12640 }, { "epoch": 0.3332202842909026, "grad_norm": 1.8099098205566406, "learning_rate": 1.5553928809573275e-05, "loss": 0.15741755962371826, "step": 12650 }, { "epoch": 0.33348369953540136, "grad_norm": 1.1883820295333862, "learning_rate": 1.5546832201395964e-05, "loss": 0.15507175922393798, "step": 12660 }, { "epoch": 0.3337471147799001, "grad_norm": 2.883622646331787, "learning_rate": 1.5539731555994864e-05, "loss": 0.15319105386734008, "step": 12670 }, { "epoch": 0.33401053002439884, "grad_norm": 2.4051787853240967, "learning_rate": 1.5532626878538137e-05, "loss": 0.15606212615966797, "step": 12680 }, { "epoch": 0.3342739452688976, "grad_norm": 3.901456117630005, "learning_rate": 1.5525518174196867e-05, "loss": 0.15831183195114135, "step": 12690 }, { "epoch": 0.33453736051339633, "grad_norm": 2.7622764110565186, "learning_rate": 1.551840544814508e-05, "loss": 0.14675657749176024, "step": 12700 }, { "epoch": 0.3348007757578951, "grad_norm": 1.4960697889328003, "learning_rate": 1.551128870555972e-05, "loss": 0.1630489706993103, "step": 12710 }, { "epoch": 0.33506419100239376, "grad_norm": 2.000282049179077, "learning_rate": 1.550416795162067e-05, "loss": 0.15433270931243898, "step": 12720 }, { "epoch": 0.3353276062468925, "grad_norm": 3.0130889415740967, "learning_rate": 1.549704319151071e-05, "loss": 0.15676023960113525, "step": 12730 }, { "epoch": 0.33559102149139125, "grad_norm": 1.530669093132019, "learning_rate": 1.548991443041555e-05, "loss": 0.15625331401824952, "step": 12740 }, { "epoch": 0.33585443673589, "grad_norm": 3.2316300868988037, "learning_rate": 1.5482781673523816e-05, "loss": 0.16359055042266846, "step": 12750 }, { "epoch": 0.33611785198038874, "grad_norm": 3.0508041381835938, "learning_rate": 1.5475644926027035e-05, "loss": 0.14686300754547119, "step": 12760 }, { "epoch": 0.3363812672248875, "grad_norm": 4.68299674987793, "learning_rate": 1.5468504193119636e-05, "loss": 0.17509009838104247, "step": 12770 }, { "epoch": 0.3366446824693862, "grad_norm": 3.683105230331421, "learning_rate": 1.5461359479998953e-05, "loss": 0.15415937900543214, "step": 12780 }, { "epoch": 0.33690809771388497, "grad_norm": 2.762730836868286, "learning_rate": 1.5454210791865213e-05, "loss": 0.157653284072876, "step": 12790 }, { "epoch": 0.33717151295838366, "grad_norm": 2.7830536365509033, "learning_rate": 1.5447058133921552e-05, "loss": 0.17430522441864013, "step": 12800 }, { "epoch": 0.3374349282028824, "grad_norm": 2.8838114738464355, "learning_rate": 1.543990151137397e-05, "loss": 0.1562286138534546, "step": 12810 }, { "epoch": 0.33769834344738114, "grad_norm": 1.70232355594635, "learning_rate": 1.5432740929431372e-05, "loss": 0.1455442190170288, "step": 12820 }, { "epoch": 0.3379617586918799, "grad_norm": 3.929051637649536, "learning_rate": 1.5425576393305537e-05, "loss": 0.15198957920074463, "step": 12830 }, { "epoch": 0.33822517393637863, "grad_norm": 1.3719321489334106, "learning_rate": 1.5418407908211125e-05, "loss": 0.15464996099472045, "step": 12840 }, { "epoch": 0.3384885891808774, "grad_norm": 2.5590133666992188, "learning_rate": 1.5411235479365664e-05, "loss": 0.15163860321044922, "step": 12850 }, { "epoch": 0.3387520044253761, "grad_norm": 1.6788889169692993, "learning_rate": 1.5404059111989565e-05, "loss": 0.16663382053375245, "step": 12860 }, { "epoch": 0.33901541966987486, "grad_norm": 3.0411064624786377, "learning_rate": 1.5396878811306088e-05, "loss": 0.15618540048599244, "step": 12870 }, { "epoch": 0.3392788349143736, "grad_norm": 1.549117088317871, "learning_rate": 1.5389694582541376e-05, "loss": 0.15877372026443481, "step": 12880 }, { "epoch": 0.3395422501588723, "grad_norm": 1.040724515914917, "learning_rate": 1.5382506430924414e-05, "loss": 0.15791335105895996, "step": 12890 }, { "epoch": 0.33980566540337104, "grad_norm": 3.589566230773926, "learning_rate": 1.5375314361687053e-05, "loss": 0.1504300355911255, "step": 12900 }, { "epoch": 0.3400690806478698, "grad_norm": 3.6981382369995117, "learning_rate": 1.5368118380063984e-05, "loss": 0.16207464933395385, "step": 12910 }, { "epoch": 0.3403324958923685, "grad_norm": 3.1001906394958496, "learning_rate": 1.536091849129276e-05, "loss": 0.15955662727355957, "step": 12920 }, { "epoch": 0.34059591113686727, "grad_norm": 1.008660078048706, "learning_rate": 1.5353714700613768e-05, "loss": 0.15381213426589965, "step": 12930 }, { "epoch": 0.340859326381366, "grad_norm": 4.447527885437012, "learning_rate": 1.5346507013270237e-05, "loss": 0.14654723405838013, "step": 12940 }, { "epoch": 0.34112274162586476, "grad_norm": 4.571254730224609, "learning_rate": 1.5339295434508238e-05, "loss": 0.1485970973968506, "step": 12950 }, { "epoch": 0.3413861568703635, "grad_norm": 3.3915746212005615, "learning_rate": 1.533207996957666e-05, "loss": 0.1564070463180542, "step": 12960 }, { "epoch": 0.34164957211486224, "grad_norm": 1.466364860534668, "learning_rate": 1.5324860623727236e-05, "loss": 0.1589105248451233, "step": 12970 }, { "epoch": 0.34191298735936093, "grad_norm": 2.3881568908691406, "learning_rate": 1.5317637402214515e-05, "loss": 0.15390123128890992, "step": 12980 }, { "epoch": 0.3421764026038597, "grad_norm": 2.2594292163848877, "learning_rate": 1.5310410310295876e-05, "loss": 0.14801232814788817, "step": 12990 }, { "epoch": 0.3424398178483584, "grad_norm": 2.6686270236968994, "learning_rate": 1.5303179353231494e-05, "loss": 0.16260597705841065, "step": 13000 }, { "epoch": 0.34270323309285716, "grad_norm": 3.512061357498169, "learning_rate": 1.529594453628439e-05, "loss": 0.15737838745117189, "step": 13010 }, { "epoch": 0.3429666483373559, "grad_norm": 3.1925265789031982, "learning_rate": 1.5288705864720356e-05, "loss": 0.1587701916694641, "step": 13020 }, { "epoch": 0.34323006358185465, "grad_norm": 3.5191726684570312, "learning_rate": 1.528146334380803e-05, "loss": 0.18409376144409179, "step": 13030 }, { "epoch": 0.3434934788263534, "grad_norm": 2.4383716583251953, "learning_rate": 1.5274216978818813e-05, "loss": 0.1541909694671631, "step": 13040 }, { "epoch": 0.34375689407085214, "grad_norm": 5.711147785186768, "learning_rate": 1.526696677502693e-05, "loss": 0.15480339527130127, "step": 13050 }, { "epoch": 0.3440203093153508, "grad_norm": 4.418207168579102, "learning_rate": 1.525971273770939e-05, "loss": 0.15626920461654664, "step": 13060 }, { "epoch": 0.34428372455984957, "grad_norm": 4.175559043884277, "learning_rate": 1.5252454872146003e-05, "loss": 0.15471420288085938, "step": 13070 }, { "epoch": 0.3445471398043483, "grad_norm": 4.193700790405273, "learning_rate": 1.5245193183619347e-05, "loss": 0.16440666913986207, "step": 13080 }, { "epoch": 0.34481055504884706, "grad_norm": 0.7208430767059326, "learning_rate": 1.5237927677414797e-05, "loss": 0.16074087619781494, "step": 13090 }, { "epoch": 0.3450739702933458, "grad_norm": 1.7487246990203857, "learning_rate": 1.5230658358820493e-05, "loss": 0.1513054370880127, "step": 13100 }, { "epoch": 0.34533738553784454, "grad_norm": 2.51452374458313, "learning_rate": 1.5223385233127372e-05, "loss": 0.1448679208755493, "step": 13110 }, { "epoch": 0.3456008007823433, "grad_norm": 4.2671003341674805, "learning_rate": 1.5216108305629123e-05, "loss": 0.1512455701828003, "step": 13120 }, { "epoch": 0.34586421602684203, "grad_norm": 3.520005702972412, "learning_rate": 1.5208827581622205e-05, "loss": 0.17305948734283447, "step": 13130 }, { "epoch": 0.3461276312713408, "grad_norm": 0.7060644626617432, "learning_rate": 1.5201543066405844e-05, "loss": 0.1475168228149414, "step": 13140 }, { "epoch": 0.34639104651583946, "grad_norm": 2.1247406005859375, "learning_rate": 1.5194254765282025e-05, "loss": 0.16476812362670898, "step": 13150 }, { "epoch": 0.3466544617603382, "grad_norm": 2.900681495666504, "learning_rate": 1.5186962683555493e-05, "loss": 0.158073353767395, "step": 13160 }, { "epoch": 0.34691787700483695, "grad_norm": 0.7967590093612671, "learning_rate": 1.517966682653373e-05, "loss": 0.1477302074432373, "step": 13170 }, { "epoch": 0.3471812922493357, "grad_norm": 4.04049825668335, "learning_rate": 1.5172367199526985e-05, "loss": 0.15378986597061156, "step": 13180 }, { "epoch": 0.34744470749383444, "grad_norm": 3.6853220462799072, "learning_rate": 1.5165063807848234e-05, "loss": 0.163934588432312, "step": 13190 }, { "epoch": 0.3477081227383332, "grad_norm": 4.728620529174805, "learning_rate": 1.5157756656813207e-05, "loss": 0.16947389841079713, "step": 13200 }, { "epoch": 0.3479715379828319, "grad_norm": 2.8881349563598633, "learning_rate": 1.5150445751740356e-05, "loss": 0.15014164447784423, "step": 13210 }, { "epoch": 0.34823495322733067, "grad_norm": 1.5933194160461426, "learning_rate": 1.5143131097950882e-05, "loss": 0.14749406576156615, "step": 13220 }, { "epoch": 0.3484983684718294, "grad_norm": 2.5747294425964355, "learning_rate": 1.5135812700768702e-05, "loss": 0.16826587915420532, "step": 13230 }, { "epoch": 0.3487617837163281, "grad_norm": 1.0085186958312988, "learning_rate": 1.512849056552046e-05, "loss": 0.14585365056991578, "step": 13240 }, { "epoch": 0.34902519896082684, "grad_norm": 2.48184871673584, "learning_rate": 1.512116469753552e-05, "loss": 0.15839531421661376, "step": 13250 }, { "epoch": 0.3492886142053256, "grad_norm": 3.954153060913086, "learning_rate": 1.5113835102145968e-05, "loss": 0.15383121967315674, "step": 13260 }, { "epoch": 0.34955202944982433, "grad_norm": 1.8645808696746826, "learning_rate": 1.51065017846866e-05, "loss": 0.15207326412200928, "step": 13270 }, { "epoch": 0.3498154446943231, "grad_norm": 2.4316766262054443, "learning_rate": 1.509916475049492e-05, "loss": 0.1452833294868469, "step": 13280 }, { "epoch": 0.3500788599388218, "grad_norm": 5.203769683837891, "learning_rate": 1.5091824004911137e-05, "loss": 0.18288179636001586, "step": 13290 }, { "epoch": 0.35034227518332056, "grad_norm": 3.6142828464508057, "learning_rate": 1.5084479553278164e-05, "loss": 0.15225850343704223, "step": 13300 }, { "epoch": 0.3506056904278193, "grad_norm": 1.9940658807754517, "learning_rate": 1.5077131400941609e-05, "loss": 0.14709794521331787, "step": 13310 }, { "epoch": 0.350869105672318, "grad_norm": 0.9904831051826477, "learning_rate": 1.5069779553249775e-05, "loss": 0.16154839992523193, "step": 13320 }, { "epoch": 0.35113252091681674, "grad_norm": 2.7305588722229004, "learning_rate": 1.5062424015553653e-05, "loss": 0.17563393115997314, "step": 13330 }, { "epoch": 0.3513959361613155, "grad_norm": 3.6411561965942383, "learning_rate": 1.505506479320692e-05, "loss": 0.1476583957672119, "step": 13340 }, { "epoch": 0.3516593514058142, "grad_norm": 1.8469700813293457, "learning_rate": 1.5047701891565938e-05, "loss": 0.15750393867492676, "step": 13350 }, { "epoch": 0.35192276665031297, "grad_norm": 1.7784860134124756, "learning_rate": 1.504033531598974e-05, "loss": 0.14798624515533448, "step": 13360 }, { "epoch": 0.3521861818948117, "grad_norm": 3.7305212020874023, "learning_rate": 1.5032965071840045e-05, "loss": 0.1563820481300354, "step": 13370 }, { "epoch": 0.35244959713931046, "grad_norm": 0.8277287483215332, "learning_rate": 1.5025591164481226e-05, "loss": 0.17212202548980712, "step": 13380 }, { "epoch": 0.3527130123838092, "grad_norm": 4.830750942230225, "learning_rate": 1.5018213599280339e-05, "loss": 0.15483288764953612, "step": 13390 }, { "epoch": 0.35297642762830794, "grad_norm": 2.7403528690338135, "learning_rate": 1.5010832381607085e-05, "loss": 0.15780625343322754, "step": 13400 }, { "epoch": 0.35323984287280663, "grad_norm": 5.731367588043213, "learning_rate": 1.500344751683384e-05, "loss": 0.1576176881790161, "step": 13410 }, { "epoch": 0.3535032581173054, "grad_norm": 5.118147373199463, "learning_rate": 1.4996059010335626e-05, "loss": 0.16167192459106444, "step": 13420 }, { "epoch": 0.3537666733618041, "grad_norm": 1.9963892698287964, "learning_rate": 1.4988666867490117e-05, "loss": 0.1467756986618042, "step": 13430 }, { "epoch": 0.35403008860630286, "grad_norm": 5.975164890289307, "learning_rate": 1.4981271093677632e-05, "loss": 0.14846646785736084, "step": 13440 }, { "epoch": 0.3542935038508016, "grad_norm": 4.952840805053711, "learning_rate": 1.4973871694281133e-05, "loss": 0.16628026962280273, "step": 13450 }, { "epoch": 0.35455691909530035, "grad_norm": 3.000861883163452, "learning_rate": 1.4966468674686226e-05, "loss": 0.15074121952056885, "step": 13460 }, { "epoch": 0.3548203343397991, "grad_norm": 5.035278797149658, "learning_rate": 1.4959062040281143e-05, "loss": 0.15497480630874633, "step": 13470 }, { "epoch": 0.35508374958429784, "grad_norm": 2.6768972873687744, "learning_rate": 1.4951651796456758e-05, "loss": 0.15993027687072753, "step": 13480 }, { "epoch": 0.3553471648287966, "grad_norm": 0.6191369295120239, "learning_rate": 1.4944237948606561e-05, "loss": 0.14909098148345948, "step": 13490 }, { "epoch": 0.35561058007329527, "grad_norm": 1.4712839126586914, "learning_rate": 1.4936820502126676e-05, "loss": 0.145619535446167, "step": 13500 }, { "epoch": 0.355873995317794, "grad_norm": 3.501053810119629, "learning_rate": 1.4929399462415832e-05, "loss": 0.1551842212677002, "step": 13510 }, { "epoch": 0.35613741056229276, "grad_norm": 2.6036622524261475, "learning_rate": 1.492197483487539e-05, "loss": 0.15389113426208495, "step": 13520 }, { "epoch": 0.3564008258067915, "grad_norm": 2.7931056022644043, "learning_rate": 1.491454662490931e-05, "loss": 0.1578327775001526, "step": 13530 }, { "epoch": 0.35666424105129024, "grad_norm": 1.210008978843689, "learning_rate": 1.4907114837924168e-05, "loss": 0.15020571947097777, "step": 13540 }, { "epoch": 0.356927656295789, "grad_norm": 1.3263297080993652, "learning_rate": 1.4899679479329133e-05, "loss": 0.1485185742378235, "step": 13550 }, { "epoch": 0.35719107154028773, "grad_norm": 3.904026746749878, "learning_rate": 1.4892240554535981e-05, "loss": 0.14909703731536866, "step": 13560 }, { "epoch": 0.3574544867847865, "grad_norm": 1.00082528591156, "learning_rate": 1.4884798068959085e-05, "loss": 0.1490505814552307, "step": 13570 }, { "epoch": 0.3577179020292852, "grad_norm": 1.7562328577041626, "learning_rate": 1.4877352028015408e-05, "loss": 0.1574547052383423, "step": 13580 }, { "epoch": 0.3579813172737839, "grad_norm": 1.7152397632598877, "learning_rate": 1.4869902437124499e-05, "loss": 0.16316239833831786, "step": 13590 }, { "epoch": 0.35824473251828265, "grad_norm": 2.486084461212158, "learning_rate": 1.486244930170849e-05, "loss": 0.15855484008789061, "step": 13600 }, { "epoch": 0.3585081477627814, "grad_norm": 1.1699907779693604, "learning_rate": 1.4854992627192096e-05, "loss": 0.1595973253250122, "step": 13610 }, { "epoch": 0.35877156300728014, "grad_norm": 3.5880208015441895, "learning_rate": 1.4847532419002602e-05, "loss": 0.15127887725830078, "step": 13620 }, { "epoch": 0.3590349782517789, "grad_norm": 0.9334840774536133, "learning_rate": 1.4840068682569877e-05, "loss": 0.1614994525909424, "step": 13630 }, { "epoch": 0.3592983934962776, "grad_norm": 4.105827331542969, "learning_rate": 1.4832601423326346e-05, "loss": 0.13425490856170655, "step": 13640 }, { "epoch": 0.35956180874077637, "grad_norm": 1.8240792751312256, "learning_rate": 1.4825130646707002e-05, "loss": 0.13875250816345214, "step": 13650 }, { "epoch": 0.3598252239852751, "grad_norm": 2.8327982425689697, "learning_rate": 1.4817656358149398e-05, "loss": 0.15917975902557374, "step": 13660 }, { "epoch": 0.3600886392297738, "grad_norm": 1.3292466402053833, "learning_rate": 1.4810178563093648e-05, "loss": 0.16083253622055055, "step": 13670 }, { "epoch": 0.36035205447427254, "grad_norm": 1.5185973644256592, "learning_rate": 1.4802697266982411e-05, "loss": 0.1603541135787964, "step": 13680 }, { "epoch": 0.3606154697187713, "grad_norm": 3.6771650314331055, "learning_rate": 1.4795212475260896e-05, "loss": 0.1562427759170532, "step": 13690 }, { "epoch": 0.36087888496327003, "grad_norm": 1.558714747428894, "learning_rate": 1.4787724193376858e-05, "loss": 0.14789848327636718, "step": 13700 }, { "epoch": 0.3611423002077688, "grad_norm": 4.7190752029418945, "learning_rate": 1.4780232426780592e-05, "loss": 0.15097179412841796, "step": 13710 }, { "epoch": 0.3614057154522675, "grad_norm": 1.3779897689819336, "learning_rate": 1.477273718092493e-05, "loss": 0.15766135454177857, "step": 13720 }, { "epoch": 0.36166913069676626, "grad_norm": 1.570532202720642, "learning_rate": 1.4765238461265235e-05, "loss": 0.1563927173614502, "step": 13730 }, { "epoch": 0.361932545941265, "grad_norm": 3.4850306510925293, "learning_rate": 1.4757736273259396e-05, "loss": 0.1673651933670044, "step": 13740 }, { "epoch": 0.36219596118576375, "grad_norm": 2.2478458881378174, "learning_rate": 1.4750230622367831e-05, "loss": 0.15416955947875977, "step": 13750 }, { "epoch": 0.36245937643026244, "grad_norm": 4.723495006561279, "learning_rate": 1.474272151405348e-05, "loss": 0.15046384334564208, "step": 13760 }, { "epoch": 0.3627227916747612, "grad_norm": 2.239496946334839, "learning_rate": 1.473520895378179e-05, "loss": 0.16150949001312256, "step": 13770 }, { "epoch": 0.3629862069192599, "grad_norm": 2.9631705284118652, "learning_rate": 1.472769294702073e-05, "loss": 0.17091062068939208, "step": 13780 }, { "epoch": 0.36324962216375867, "grad_norm": 5.536523818969727, "learning_rate": 1.4720173499240771e-05, "loss": 0.1562723159790039, "step": 13790 }, { "epoch": 0.3635130374082574, "grad_norm": 3.7149553298950195, "learning_rate": 1.4712650615914899e-05, "loss": 0.15723769664764403, "step": 13800 }, { "epoch": 0.36377645265275615, "grad_norm": 3.218623399734497, "learning_rate": 1.4705124302518581e-05, "loss": 0.16516518592834473, "step": 13810 }, { "epoch": 0.3640398678972549, "grad_norm": 3.750060796737671, "learning_rate": 1.4697594564529805e-05, "loss": 0.16350119113922118, "step": 13820 }, { "epoch": 0.36430328314175364, "grad_norm": 1.141168236732483, "learning_rate": 1.4690061407429028e-05, "loss": 0.1495755672454834, "step": 13830 }, { "epoch": 0.3645666983862524, "grad_norm": 1.6256650686264038, "learning_rate": 1.468252483669922e-05, "loss": 0.16765048503875732, "step": 13840 }, { "epoch": 0.3648301136307511, "grad_norm": 1.7256685495376587, "learning_rate": 1.4674984857825808e-05, "loss": 0.1674511432647705, "step": 13850 }, { "epoch": 0.3650935288752498, "grad_norm": 4.256455421447754, "learning_rate": 1.4667441476296723e-05, "loss": 0.16066058874130248, "step": 13860 }, { "epoch": 0.36535694411974856, "grad_norm": 1.7516167163848877, "learning_rate": 1.4659894697602358e-05, "loss": 0.15649149417877198, "step": 13870 }, { "epoch": 0.3656203593642473, "grad_norm": 3.859694480895996, "learning_rate": 1.465234452723559e-05, "loss": 0.15465049743652343, "step": 13880 }, { "epoch": 0.36588377460874605, "grad_norm": 3.1839685440063477, "learning_rate": 1.4644790970691755e-05, "loss": 0.1470196008682251, "step": 13890 }, { "epoch": 0.3661471898532448, "grad_norm": 2.765091896057129, "learning_rate": 1.4637234033468658e-05, "loss": 0.15427899360656738, "step": 13900 }, { "epoch": 0.36641060509774354, "grad_norm": 2.2436978816986084, "learning_rate": 1.4629673721066566e-05, "loss": 0.14553892612457275, "step": 13910 }, { "epoch": 0.3666740203422423, "grad_norm": 4.2316389083862305, "learning_rate": 1.4622110038988196e-05, "loss": 0.1605063796043396, "step": 13920 }, { "epoch": 0.36693743558674097, "grad_norm": 1.3559348583221436, "learning_rate": 1.4614542992738727e-05, "loss": 0.1618654251098633, "step": 13930 }, { "epoch": 0.3672008508312397, "grad_norm": 0.9106554985046387, "learning_rate": 1.4606972587825779e-05, "loss": 0.16210724115371705, "step": 13940 }, { "epoch": 0.36746426607573845, "grad_norm": 3.860856771469116, "learning_rate": 1.4599398829759419e-05, "loss": 0.1566847324371338, "step": 13950 }, { "epoch": 0.3677276813202372, "grad_norm": 2.794706344604492, "learning_rate": 1.4591821724052155e-05, "loss": 0.14849469661712647, "step": 13960 }, { "epoch": 0.36799109656473594, "grad_norm": 1.6010122299194336, "learning_rate": 1.4584241276218935e-05, "loss": 0.16697168350219727, "step": 13970 }, { "epoch": 0.3682545118092347, "grad_norm": 0.871196985244751, "learning_rate": 1.4576657491777134e-05, "loss": 0.15453532934188843, "step": 13980 }, { "epoch": 0.36851792705373343, "grad_norm": 3.2850492000579834, "learning_rate": 1.4569070376246552e-05, "loss": 0.16447165012359619, "step": 13990 }, { "epoch": 0.3687813422982322, "grad_norm": 1.1078507900238037, "learning_rate": 1.4561479935149426e-05, "loss": 0.15780943632125854, "step": 14000 }, { "epoch": 0.3690447575427309, "grad_norm": 2.8454017639160156, "learning_rate": 1.4553886174010405e-05, "loss": 0.17040677070617677, "step": 14010 }, { "epoch": 0.3693081727872296, "grad_norm": 0.7846514582633972, "learning_rate": 1.4546289098356556e-05, "loss": 0.14442946910858154, "step": 14020 }, { "epoch": 0.36957158803172835, "grad_norm": 1.1199849843978882, "learning_rate": 1.4538688713717357e-05, "loss": 0.1505346417427063, "step": 14030 }, { "epoch": 0.3698350032762271, "grad_norm": 5.050404071807861, "learning_rate": 1.4531085025624694e-05, "loss": 0.1570458173751831, "step": 14040 }, { "epoch": 0.37009841852072584, "grad_norm": 1.3123372793197632, "learning_rate": 1.4523478039612863e-05, "loss": 0.17294027805328369, "step": 14050 }, { "epoch": 0.3703618337652246, "grad_norm": 2.1569607257843018, "learning_rate": 1.4515867761218558e-05, "loss": 0.15910665988922118, "step": 14060 }, { "epoch": 0.3706252490097233, "grad_norm": 2.679267168045044, "learning_rate": 1.4508254195980861e-05, "loss": 0.1499685287475586, "step": 14070 }, { "epoch": 0.37088866425422207, "grad_norm": 1.2824442386627197, "learning_rate": 1.450063734944126e-05, "loss": 0.15964646339416505, "step": 14080 }, { "epoch": 0.3711520794987208, "grad_norm": 1.2714837789535522, "learning_rate": 1.449301722714362e-05, "loss": 0.15890177488327026, "step": 14090 }, { "epoch": 0.37141549474321955, "grad_norm": 1.7515004873275757, "learning_rate": 1.4485393834634202e-05, "loss": 0.15760133266448975, "step": 14100 }, { "epoch": 0.37167890998771824, "grad_norm": 2.8757009506225586, "learning_rate": 1.447776717746163e-05, "loss": 0.15064315795898436, "step": 14110 }, { "epoch": 0.371942325232217, "grad_norm": 1.8756635189056396, "learning_rate": 1.4470137261176922e-05, "loss": 0.1532139301300049, "step": 14120 }, { "epoch": 0.37220574047671573, "grad_norm": 6.082047939300537, "learning_rate": 1.4462504091333454e-05, "loss": 0.1567413330078125, "step": 14130 }, { "epoch": 0.3724691557212145, "grad_norm": 1.8303635120391846, "learning_rate": 1.4454867673486983e-05, "loss": 0.15669186115264894, "step": 14140 }, { "epoch": 0.3727325709657132, "grad_norm": 3.2554080486297607, "learning_rate": 1.4447228013195616e-05, "loss": 0.1548445224761963, "step": 14150 }, { "epoch": 0.37299598621021196, "grad_norm": 1.8733655214309692, "learning_rate": 1.4439585116019835e-05, "loss": 0.15747406482696533, "step": 14160 }, { "epoch": 0.3732594014547107, "grad_norm": 3.3358824253082275, "learning_rate": 1.4431938987522463e-05, "loss": 0.16382310390472413, "step": 14170 }, { "epoch": 0.37352281669920945, "grad_norm": 2.0066051483154297, "learning_rate": 1.4424289633268689e-05, "loss": 0.16608269214630128, "step": 14180 }, { "epoch": 0.37378623194370814, "grad_norm": 1.4993326663970947, "learning_rate": 1.4416637058826038e-05, "loss": 0.14693477153778076, "step": 14190 }, { "epoch": 0.3740496471882069, "grad_norm": 3.699746608734131, "learning_rate": 1.4408981269764386e-05, "loss": 0.15984396934509276, "step": 14200 }, { "epoch": 0.3743130624327056, "grad_norm": 4.365082740783691, "learning_rate": 1.4401322271655947e-05, "loss": 0.15340790748596192, "step": 14210 }, { "epoch": 0.37457647767720437, "grad_norm": 2.1216940879821777, "learning_rate": 1.4393660070075267e-05, "loss": 0.16332361698150635, "step": 14220 }, { "epoch": 0.3748398929217031, "grad_norm": 4.988616943359375, "learning_rate": 1.4385994670599233e-05, "loss": 0.15384645462036134, "step": 14230 }, { "epoch": 0.37510330816620185, "grad_norm": 2.1278722286224365, "learning_rate": 1.4378326078807048e-05, "loss": 0.154416823387146, "step": 14240 }, { "epoch": 0.3753667234107006, "grad_norm": 0.7625327706336975, "learning_rate": 1.4370654300280246e-05, "loss": 0.16041359901428223, "step": 14250 }, { "epoch": 0.37563013865519934, "grad_norm": 2.319284200668335, "learning_rate": 1.4362979340602677e-05, "loss": 0.15661613941192626, "step": 14260 }, { "epoch": 0.3758935538996981, "grad_norm": 1.5435218811035156, "learning_rate": 1.4355301205360513e-05, "loss": 0.1590883493423462, "step": 14270 }, { "epoch": 0.3761569691441968, "grad_norm": 1.8966283798217773, "learning_rate": 1.4347619900142223e-05, "loss": 0.17667946815490723, "step": 14280 }, { "epoch": 0.3764203843886955, "grad_norm": 3.7790472507476807, "learning_rate": 1.43399354305386e-05, "loss": 0.14672544002532958, "step": 14290 }, { "epoch": 0.37668379963319426, "grad_norm": 5.709134578704834, "learning_rate": 1.4332247802142729e-05, "loss": 0.17135469913482665, "step": 14300 }, { "epoch": 0.376947214877693, "grad_norm": 1.7261080741882324, "learning_rate": 1.4324557020550001e-05, "loss": 0.15495681762695312, "step": 14310 }, { "epoch": 0.37721063012219175, "grad_norm": 1.3902658224105835, "learning_rate": 1.4316863091358093e-05, "loss": 0.15425329208374022, "step": 14320 }, { "epoch": 0.3774740453666905, "grad_norm": 1.7450416088104248, "learning_rate": 1.4309166020166984e-05, "loss": 0.1367850661277771, "step": 14330 }, { "epoch": 0.37773746061118924, "grad_norm": 1.1896568536758423, "learning_rate": 1.430146581257893e-05, "loss": 0.15383696556091309, "step": 14340 }, { "epoch": 0.378000875855688, "grad_norm": 2.399277448654175, "learning_rate": 1.4293762474198478e-05, "loss": 0.1480440616607666, "step": 14350 }, { "epoch": 0.3782642911001867, "grad_norm": 2.7361326217651367, "learning_rate": 1.428605601063245e-05, "loss": 0.15665643215179442, "step": 14360 }, { "epoch": 0.3785277063446854, "grad_norm": 1.77297842502594, "learning_rate": 1.427834642748994e-05, "loss": 0.15664370059967042, "step": 14370 }, { "epoch": 0.37879112158918415, "grad_norm": 2.1223278045654297, "learning_rate": 1.4270633730382315e-05, "loss": 0.15858845710754393, "step": 14380 }, { "epoch": 0.3790545368336829, "grad_norm": 3.130255699157715, "learning_rate": 1.4262917924923212e-05, "loss": 0.14708689451217652, "step": 14390 }, { "epoch": 0.37931795207818164, "grad_norm": 2.2686219215393066, "learning_rate": 1.4255199016728527e-05, "loss": 0.1584317207336426, "step": 14400 }, { "epoch": 0.3795813673226804, "grad_norm": 1.6433639526367188, "learning_rate": 1.4247477011416409e-05, "loss": 0.14748880863189698, "step": 14410 }, { "epoch": 0.37984478256717913, "grad_norm": 1.4716888666152954, "learning_rate": 1.4239751914607274e-05, "loss": 0.1618494510650635, "step": 14420 }, { "epoch": 0.3801081978116779, "grad_norm": 3.717939615249634, "learning_rate": 1.4232023731923772e-05, "loss": 0.13844015598297119, "step": 14430 }, { "epoch": 0.3803716130561766, "grad_norm": 2.5835957527160645, "learning_rate": 1.422429246899082e-05, "loss": 0.16677572727203369, "step": 14440 }, { "epoch": 0.3806350283006753, "grad_norm": 2.86434006690979, "learning_rate": 1.4216558131435551e-05, "loss": 0.15580954551696777, "step": 14450 }, { "epoch": 0.38089844354517405, "grad_norm": 2.0215024948120117, "learning_rate": 1.420882072488736e-05, "loss": 0.15311020612716675, "step": 14460 }, { "epoch": 0.3811618587896728, "grad_norm": 2.7362427711486816, "learning_rate": 1.420108025497786e-05, "loss": 0.17461414337158204, "step": 14470 }, { "epoch": 0.38142527403417154, "grad_norm": 1.9275941848754883, "learning_rate": 1.41933367273409e-05, "loss": 0.15505471229553222, "step": 14480 }, { "epoch": 0.3816886892786703, "grad_norm": 1.371342658996582, "learning_rate": 1.4185590147612555e-05, "loss": 0.15430819988250732, "step": 14490 }, { "epoch": 0.381952104523169, "grad_norm": 7.058064937591553, "learning_rate": 1.4177840521431117e-05, "loss": 0.15375725030899048, "step": 14500 }, { "epoch": 0.38221551976766777, "grad_norm": 5.9320549964904785, "learning_rate": 1.4170087854437102e-05, "loss": 0.14558796882629393, "step": 14510 }, { "epoch": 0.3824789350121665, "grad_norm": 2.383286714553833, "learning_rate": 1.4162332152273233e-05, "loss": 0.17081704139709472, "step": 14520 }, { "epoch": 0.38274235025666525, "grad_norm": 2.7128396034240723, "learning_rate": 1.4154573420584448e-05, "loss": 0.15744285583496093, "step": 14530 }, { "epoch": 0.38300576550116394, "grad_norm": 1.0562629699707031, "learning_rate": 1.414681166501788e-05, "loss": 0.15875465869903566, "step": 14540 }, { "epoch": 0.3832691807456627, "grad_norm": 1.1499518156051636, "learning_rate": 1.4139046891222877e-05, "loss": 0.16710405349731444, "step": 14550 }, { "epoch": 0.38353259599016143, "grad_norm": 2.0028839111328125, "learning_rate": 1.4131279104850968e-05, "loss": 0.17713561058044433, "step": 14560 }, { "epoch": 0.3837960112346602, "grad_norm": 1.1660892963409424, "learning_rate": 1.4123508311555894e-05, "loss": 0.14895374774932862, "step": 14570 }, { "epoch": 0.3840594264791589, "grad_norm": 1.8991953134536743, "learning_rate": 1.4115734516993562e-05, "loss": 0.1618351936340332, "step": 14580 }, { "epoch": 0.38432284172365766, "grad_norm": 6.593647480010986, "learning_rate": 1.4107957726822084e-05, "loss": 0.15811924934387206, "step": 14590 }, { "epoch": 0.3845862569681564, "grad_norm": 4.334743499755859, "learning_rate": 1.4100177946701738e-05, "loss": 0.16228703260421753, "step": 14600 }, { "epoch": 0.38484967221265515, "grad_norm": 1.2624168395996094, "learning_rate": 1.4092395182294989e-05, "loss": 0.15746757984161378, "step": 14610 }, { "epoch": 0.3851130874571539, "grad_norm": 1.5944483280181885, "learning_rate": 1.4084609439266464e-05, "loss": 0.15852782726287842, "step": 14620 }, { "epoch": 0.3853765027016526, "grad_norm": 1.7336452007293701, "learning_rate": 1.4076820723282962e-05, "loss": 0.1497088313102722, "step": 14630 }, { "epoch": 0.3856399179461513, "grad_norm": 0.8710024356842041, "learning_rate": 1.4069029040013448e-05, "loss": 0.1643545389175415, "step": 14640 }, { "epoch": 0.38590333319065007, "grad_norm": 2.4682390689849854, "learning_rate": 1.4061234395129049e-05, "loss": 0.15749199390411378, "step": 14650 }, { "epoch": 0.3861667484351488, "grad_norm": 3.499019145965576, "learning_rate": 1.405343679430304e-05, "loss": 0.1574141263961792, "step": 14660 }, { "epoch": 0.38643016367964755, "grad_norm": 1.95785391330719, "learning_rate": 1.4045636243210852e-05, "loss": 0.1653212308883667, "step": 14670 }, { "epoch": 0.3866935789241463, "grad_norm": 1.91001296043396, "learning_rate": 1.4037832747530063e-05, "loss": 0.14822671413421631, "step": 14680 }, { "epoch": 0.38695699416864504, "grad_norm": 1.4499162435531616, "learning_rate": 1.403002631294039e-05, "loss": 0.15451838970184326, "step": 14690 }, { "epoch": 0.3872204094131438, "grad_norm": 1.7529876232147217, "learning_rate": 1.40222169451237e-05, "loss": 0.14531986713409423, "step": 14700 }, { "epoch": 0.3874838246576425, "grad_norm": 1.3009899854660034, "learning_rate": 1.4014404649763983e-05, "loss": 0.1561952829360962, "step": 14710 }, { "epoch": 0.3877472399021412, "grad_norm": 4.093192100524902, "learning_rate": 1.4006589432547367e-05, "loss": 0.1503729462623596, "step": 14720 }, { "epoch": 0.38801065514663996, "grad_norm": 3.1991803646087646, "learning_rate": 1.3998771299162099e-05, "loss": 0.16112723350524902, "step": 14730 }, { "epoch": 0.3882740703911387, "grad_norm": 2.5105605125427246, "learning_rate": 1.3990950255298565e-05, "loss": 0.1582539916038513, "step": 14740 }, { "epoch": 0.38853748563563745, "grad_norm": 1.6898441314697266, "learning_rate": 1.3983126306649243e-05, "loss": 0.17234054803848267, "step": 14750 }, { "epoch": 0.3888009008801362, "grad_norm": 4.500149726867676, "learning_rate": 1.3975299458908753e-05, "loss": 0.15038704872131348, "step": 14760 }, { "epoch": 0.38906431612463493, "grad_norm": 2.4195761680603027, "learning_rate": 1.3967469717773807e-05, "loss": 0.16156454086303712, "step": 14770 }, { "epoch": 0.3893277313691337, "grad_norm": 3.6395013332366943, "learning_rate": 1.395963708894323e-05, "loss": 0.14805705547332765, "step": 14780 }, { "epoch": 0.3895911466136324, "grad_norm": 4.630706787109375, "learning_rate": 1.3951801578117946e-05, "loss": 0.1572324275970459, "step": 14790 }, { "epoch": 0.3898545618581311, "grad_norm": 1.1041996479034424, "learning_rate": 1.3943963191000976e-05, "loss": 0.1494309902191162, "step": 14800 }, { "epoch": 0.39011797710262985, "grad_norm": 3.7352774143218994, "learning_rate": 1.393612193329744e-05, "loss": 0.16731475591659545, "step": 14810 }, { "epoch": 0.3903813923471286, "grad_norm": 5.800961971282959, "learning_rate": 1.3928277810714542e-05, "loss": 0.15586072206497192, "step": 14820 }, { "epoch": 0.39064480759162734, "grad_norm": 3.7819576263427734, "learning_rate": 1.3920430828961576e-05, "loss": 0.14672372341156006, "step": 14830 }, { "epoch": 0.3909082228361261, "grad_norm": 1.430116057395935, "learning_rate": 1.3912580993749907e-05, "loss": 0.15300720930099487, "step": 14840 }, { "epoch": 0.39117163808062483, "grad_norm": 1.8206298351287842, "learning_rate": 1.3904728310792988e-05, "loss": 0.15164531469345094, "step": 14850 }, { "epoch": 0.39143505332512357, "grad_norm": 4.060937881469727, "learning_rate": 1.3896872785806342e-05, "loss": 0.15742716789245606, "step": 14860 }, { "epoch": 0.3916984685696223, "grad_norm": 4.129009246826172, "learning_rate": 1.3889014424507557e-05, "loss": 0.14495474100112915, "step": 14870 }, { "epoch": 0.39196188381412106, "grad_norm": 1.6618010997772217, "learning_rate": 1.3881153232616284e-05, "loss": 0.15090125799179077, "step": 14880 }, { "epoch": 0.39222529905861975, "grad_norm": 1.9863559007644653, "learning_rate": 1.3873289215854247e-05, "loss": 0.15215522050857544, "step": 14890 }, { "epoch": 0.3924887143031185, "grad_norm": 4.043011665344238, "learning_rate": 1.3865422379945205e-05, "loss": 0.1427147626876831, "step": 14900 }, { "epoch": 0.39275212954761723, "grad_norm": 1.6915533542633057, "learning_rate": 1.3857552730614989e-05, "loss": 0.1695636749267578, "step": 14910 }, { "epoch": 0.393015544792116, "grad_norm": 1.0186645984649658, "learning_rate": 1.3849680273591468e-05, "loss": 0.15714812278747559, "step": 14920 }, { "epoch": 0.3932789600366147, "grad_norm": 2.211946964263916, "learning_rate": 1.3841805014604553e-05, "loss": 0.1613412618637085, "step": 14930 }, { "epoch": 0.39354237528111347, "grad_norm": 4.101125717163086, "learning_rate": 1.3833926959386199e-05, "loss": 0.15748703479766846, "step": 14940 }, { "epoch": 0.3938057905256122, "grad_norm": 3.376776933670044, "learning_rate": 1.3826046113670395e-05, "loss": 0.15617982149124146, "step": 14950 }, { "epoch": 0.39406920577011095, "grad_norm": 1.0805249214172363, "learning_rate": 1.3818162483193159e-05, "loss": 0.15115951299667357, "step": 14960 }, { "epoch": 0.39433262101460964, "grad_norm": 0.9107287526130676, "learning_rate": 1.3810276073692537e-05, "loss": 0.14599530696868895, "step": 14970 }, { "epoch": 0.3945960362591084, "grad_norm": 1.295396089553833, "learning_rate": 1.3802386890908602e-05, "loss": 0.15293856859207153, "step": 14980 }, { "epoch": 0.39485945150360713, "grad_norm": 3.18869948387146, "learning_rate": 1.3794494940583437e-05, "loss": 0.15812582969665528, "step": 14990 }, { "epoch": 0.39512286674810587, "grad_norm": 1.9056696891784668, "learning_rate": 1.3786600228461147e-05, "loss": 0.15470926761627196, "step": 15000 }, { "epoch": 0.3953862819926046, "grad_norm": 5.299422264099121, "learning_rate": 1.377870276028784e-05, "loss": 0.1510596752166748, "step": 15010 }, { "epoch": 0.39564969723710336, "grad_norm": 1.9737203121185303, "learning_rate": 1.377080254181164e-05, "loss": 0.1634632468223572, "step": 15020 }, { "epoch": 0.3959131124816021, "grad_norm": 1.6660012006759644, "learning_rate": 1.3762899578782662e-05, "loss": 0.16386860609054565, "step": 15030 }, { "epoch": 0.39617652772610085, "grad_norm": 5.63769006729126, "learning_rate": 1.375499387695303e-05, "loss": 0.1550290107727051, "step": 15040 }, { "epoch": 0.3964399429705996, "grad_norm": 0.96148282289505, "learning_rate": 1.3747085442076844e-05, "loss": 0.15292292833328247, "step": 15050 }, { "epoch": 0.3967033582150983, "grad_norm": 0.8603999614715576, "learning_rate": 1.3739174279910214e-05, "loss": 0.15051679611206054, "step": 15060 }, { "epoch": 0.396966773459597, "grad_norm": 2.1576180458068848, "learning_rate": 1.3731260396211219e-05, "loss": 0.14113444089889526, "step": 15070 }, { "epoch": 0.39723018870409577, "grad_norm": 4.316129684448242, "learning_rate": 1.372334379673993e-05, "loss": 0.15351996421813965, "step": 15080 }, { "epoch": 0.3974936039485945, "grad_norm": 1.0873920917510986, "learning_rate": 1.3715424487258386e-05, "loss": 0.14669286012649535, "step": 15090 }, { "epoch": 0.39775701919309325, "grad_norm": 3.482088327407837, "learning_rate": 1.37075024735306e-05, "loss": 0.1663944363594055, "step": 15100 }, { "epoch": 0.398020434437592, "grad_norm": 2.172525644302368, "learning_rate": 1.3699577761322563e-05, "loss": 0.1605351448059082, "step": 15110 }, { "epoch": 0.39828384968209074, "grad_norm": 1.6151652336120605, "learning_rate": 1.3691650356402212e-05, "loss": 0.16177027225494384, "step": 15120 }, { "epoch": 0.3985472649265895, "grad_norm": 1.5747263431549072, "learning_rate": 1.3683720264539469e-05, "loss": 0.14106380939483643, "step": 15130 }, { "epoch": 0.3988106801710882, "grad_norm": 2.880281925201416, "learning_rate": 1.3675787491506184e-05, "loss": 0.1532272458076477, "step": 15140 }, { "epoch": 0.3990740954155869, "grad_norm": 1.2327467203140259, "learning_rate": 1.3667852043076173e-05, "loss": 0.15322619676589966, "step": 15150 }, { "epoch": 0.39933751066008566, "grad_norm": 3.5417397022247314, "learning_rate": 1.3659913925025203e-05, "loss": 0.16063895225524902, "step": 15160 }, { "epoch": 0.3996009259045844, "grad_norm": 4.052223205566406, "learning_rate": 1.3651973143130979e-05, "loss": 0.14889791011810302, "step": 15170 }, { "epoch": 0.39986434114908315, "grad_norm": 2.043214797973633, "learning_rate": 1.3644029703173139e-05, "loss": 0.14334354400634766, "step": 15180 }, { "epoch": 0.4001277563935819, "grad_norm": 1.578662395477295, "learning_rate": 1.3636083610933262e-05, "loss": 0.16035242080688478, "step": 15190 }, { "epoch": 0.40039117163808063, "grad_norm": 4.191829204559326, "learning_rate": 1.3628134872194861e-05, "loss": 0.15531364679336548, "step": 15200 }, { "epoch": 0.4006545868825794, "grad_norm": 1.8379265069961548, "learning_rate": 1.362018349274337e-05, "loss": 0.15206551551818848, "step": 15210 }, { "epoch": 0.4009180021270781, "grad_norm": 3.577890396118164, "learning_rate": 1.3612229478366144e-05, "loss": 0.1439568042755127, "step": 15220 }, { "epoch": 0.4011814173715768, "grad_norm": 2.436844825744629, "learning_rate": 1.360427283485246e-05, "loss": 0.15504860877990723, "step": 15230 }, { "epoch": 0.40144483261607555, "grad_norm": 2.1455514430999756, "learning_rate": 1.3596313567993504e-05, "loss": 0.15996718406677246, "step": 15240 }, { "epoch": 0.4017082478605743, "grad_norm": 1.499210000038147, "learning_rate": 1.3588351683582374e-05, "loss": 0.1453237295150757, "step": 15250 }, { "epoch": 0.40197166310507304, "grad_norm": 1.8014490604400635, "learning_rate": 1.3580387187414077e-05, "loss": 0.14692981243133546, "step": 15260 }, { "epoch": 0.4022350783495718, "grad_norm": 3.56420636177063, "learning_rate": 1.3572420085285512e-05, "loss": 0.1432950258255005, "step": 15270 }, { "epoch": 0.4024984935940705, "grad_norm": 2.459829568862915, "learning_rate": 1.3564450382995482e-05, "loss": 0.1540936589241028, "step": 15280 }, { "epoch": 0.40276190883856927, "grad_norm": 2.187544345855713, "learning_rate": 1.3556478086344678e-05, "loss": 0.14696338176727294, "step": 15290 }, { "epoch": 0.403025324083068, "grad_norm": 4.84259557723999, "learning_rate": 1.3548503201135687e-05, "loss": 0.1349020004272461, "step": 15300 }, { "epoch": 0.40328873932756676, "grad_norm": 1.5697189569473267, "learning_rate": 1.3540525733172966e-05, "loss": 0.15532852411270143, "step": 15310 }, { "epoch": 0.40355215457206545, "grad_norm": 0.9824891090393066, "learning_rate": 1.353254568826287e-05, "loss": 0.15511349439620972, "step": 15320 }, { "epoch": 0.4038155698165642, "grad_norm": 2.506747245788574, "learning_rate": 1.3524563072213612e-05, "loss": 0.15569844245910644, "step": 15330 }, { "epoch": 0.40407898506106293, "grad_norm": 3.6191420555114746, "learning_rate": 1.3516577890835292e-05, "loss": 0.15601141452789308, "step": 15340 }, { "epoch": 0.4043424003055617, "grad_norm": 1.2923489809036255, "learning_rate": 1.350859014993986e-05, "loss": 0.15636414289474487, "step": 15350 }, { "epoch": 0.4046058155500604, "grad_norm": 1.4317861795425415, "learning_rate": 1.3500599855341146e-05, "loss": 0.14919557571411132, "step": 15360 }, { "epoch": 0.40486923079455917, "grad_norm": 4.155268669128418, "learning_rate": 1.3492607012854824e-05, "loss": 0.16622923612594603, "step": 15370 }, { "epoch": 0.4051326460390579, "grad_norm": 1.5362519025802612, "learning_rate": 1.3484611628298438e-05, "loss": 0.13963549137115477, "step": 15380 }, { "epoch": 0.40539606128355665, "grad_norm": 4.795967102050781, "learning_rate": 1.3476613707491372e-05, "loss": 0.14530866146087645, "step": 15390 }, { "epoch": 0.4056594765280554, "grad_norm": 0.8311429619789124, "learning_rate": 1.3468613256254852e-05, "loss": 0.14315896034240722, "step": 15400 }, { "epoch": 0.4059228917725541, "grad_norm": 1.901777744293213, "learning_rate": 1.3460610280411957e-05, "loss": 0.15960079431533813, "step": 15410 }, { "epoch": 0.40618630701705283, "grad_norm": 2.4899866580963135, "learning_rate": 1.3452604785787595e-05, "loss": 0.15219497680664062, "step": 15420 }, { "epoch": 0.40644972226155157, "grad_norm": 2.9760968685150146, "learning_rate": 1.3444596778208516e-05, "loss": 0.15970592498779296, "step": 15430 }, { "epoch": 0.4067131375060503, "grad_norm": 2.82486629486084, "learning_rate": 1.3436586263503286e-05, "loss": 0.1513447642326355, "step": 15440 }, { "epoch": 0.40697655275054906, "grad_norm": 2.8225409984588623, "learning_rate": 1.3428573247502307e-05, "loss": 0.16389126777648927, "step": 15450 }, { "epoch": 0.4072399679950478, "grad_norm": 5.190708637237549, "learning_rate": 1.3420557736037796e-05, "loss": 0.15395805835723878, "step": 15460 }, { "epoch": 0.40750338323954655, "grad_norm": 1.4344755411148071, "learning_rate": 1.3412539734943792e-05, "loss": 0.13245464563369752, "step": 15470 }, { "epoch": 0.4077667984840453, "grad_norm": 1.2630995512008667, "learning_rate": 1.3404519250056139e-05, "loss": 0.14394068717956543, "step": 15480 }, { "epoch": 0.408030213728544, "grad_norm": 2.7855348587036133, "learning_rate": 1.3396496287212491e-05, "loss": 0.14656604528427125, "step": 15490 }, { "epoch": 0.4082936289730427, "grad_norm": 3.8313868045806885, "learning_rate": 1.3388470852252307e-05, "loss": 0.1575998306274414, "step": 15500 }, { "epoch": 0.40855704421754147, "grad_norm": 1.3506258726119995, "learning_rate": 1.3380442951016846e-05, "loss": 0.1400783896446228, "step": 15510 }, { "epoch": 0.4088204594620402, "grad_norm": 3.388718605041504, "learning_rate": 1.3372412589349164e-05, "loss": 0.15263025760650634, "step": 15520 }, { "epoch": 0.40908387470653895, "grad_norm": 2.59322452545166, "learning_rate": 1.3364379773094099e-05, "loss": 0.15172516107559203, "step": 15530 }, { "epoch": 0.4093472899510377, "grad_norm": 1.940091609954834, "learning_rate": 1.335634450809828e-05, "loss": 0.15710825920104982, "step": 15540 }, { "epoch": 0.40961070519553644, "grad_norm": 1.2303601503372192, "learning_rate": 1.3348306800210127e-05, "loss": 0.15783886909484862, "step": 15550 }, { "epoch": 0.4098741204400352, "grad_norm": 1.9361932277679443, "learning_rate": 1.3340266655279828e-05, "loss": 0.15087361335754396, "step": 15560 }, { "epoch": 0.4101375356845339, "grad_norm": 1.1080657243728638, "learning_rate": 1.3332224079159342e-05, "loss": 0.14708350896835326, "step": 15570 }, { "epoch": 0.4104009509290326, "grad_norm": 0.951759934425354, "learning_rate": 1.3324179077702408e-05, "loss": 0.15268995761871337, "step": 15580 }, { "epoch": 0.41066436617353136, "grad_norm": 4.562122344970703, "learning_rate": 1.3316131656764523e-05, "loss": 0.14031150341033935, "step": 15590 }, { "epoch": 0.4109277814180301, "grad_norm": 1.5263406038284302, "learning_rate": 1.3308081822202953e-05, "loss": 0.1565114974975586, "step": 15600 }, { "epoch": 0.41119119666252885, "grad_norm": 2.518218517303467, "learning_rate": 1.3300029579876705e-05, "loss": 0.1620193362236023, "step": 15610 }, { "epoch": 0.4114546119070276, "grad_norm": 1.602419137954712, "learning_rate": 1.3291974935646554e-05, "loss": 0.15068093538284302, "step": 15620 }, { "epoch": 0.41171802715152633, "grad_norm": 1.104636549949646, "learning_rate": 1.3283917895375018e-05, "loss": 0.1456826448440552, "step": 15630 }, { "epoch": 0.4119814423960251, "grad_norm": 1.2524794340133667, "learning_rate": 1.327585846492636e-05, "loss": 0.14515149593353271, "step": 15640 }, { "epoch": 0.4122448576405238, "grad_norm": 3.665956735610962, "learning_rate": 1.3267796650166576e-05, "loss": 0.1619308590888977, "step": 15650 }, { "epoch": 0.41250827288502256, "grad_norm": 1.1174601316452026, "learning_rate": 1.3259732456963406e-05, "loss": 0.15986865758895874, "step": 15660 }, { "epoch": 0.41277168812952125, "grad_norm": 3.631497621536255, "learning_rate": 1.3251665891186313e-05, "loss": 0.1460919499397278, "step": 15670 }, { "epoch": 0.41303510337402, "grad_norm": 1.7975772619247437, "learning_rate": 1.32435969587065e-05, "loss": 0.1502147674560547, "step": 15680 }, { "epoch": 0.41329851861851874, "grad_norm": 1.7753989696502686, "learning_rate": 1.323552566539688e-05, "loss": 0.14832831621170045, "step": 15690 }, { "epoch": 0.4135619338630175, "grad_norm": 5.392041206359863, "learning_rate": 1.3227452017132085e-05, "loss": 0.1568540334701538, "step": 15700 }, { "epoch": 0.4138253491075162, "grad_norm": 4.565306663513184, "learning_rate": 1.321937601978847e-05, "loss": 0.15076777935028077, "step": 15710 }, { "epoch": 0.41408876435201497, "grad_norm": 1.8857014179229736, "learning_rate": 1.3211297679244093e-05, "loss": 0.158722448348999, "step": 15720 }, { "epoch": 0.4143521795965137, "grad_norm": 2.5602376461029053, "learning_rate": 1.3203217001378716e-05, "loss": 0.15953531265258789, "step": 15730 }, { "epoch": 0.41461559484101246, "grad_norm": 4.098326683044434, "learning_rate": 1.3195133992073808e-05, "loss": 0.14931092262268067, "step": 15740 }, { "epoch": 0.41487901008551115, "grad_norm": 2.950946569442749, "learning_rate": 1.318704865721253e-05, "loss": 0.1533435344696045, "step": 15750 }, { "epoch": 0.4151424253300099, "grad_norm": 2.237382650375366, "learning_rate": 1.3178961002679738e-05, "loss": 0.15668736696243285, "step": 15760 }, { "epoch": 0.41540584057450863, "grad_norm": 2.4699108600616455, "learning_rate": 1.317087103436198e-05, "loss": 0.15825290679931642, "step": 15770 }, { "epoch": 0.4156692558190074, "grad_norm": 1.6673147678375244, "learning_rate": 1.3162778758147477e-05, "loss": 0.15374574661254883, "step": 15780 }, { "epoch": 0.4159326710635061, "grad_norm": 1.8918758630752563, "learning_rate": 1.315468417992614e-05, "loss": 0.16324350833892823, "step": 15790 }, { "epoch": 0.41619608630800486, "grad_norm": 1.1534171104431152, "learning_rate": 1.3146587305589552e-05, "loss": 0.1575068235397339, "step": 15800 }, { "epoch": 0.4164595015525036, "grad_norm": 1.4766664505004883, "learning_rate": 1.3138488141030967e-05, "loss": 0.14655203819274903, "step": 15810 }, { "epoch": 0.41672291679700235, "grad_norm": 5.544238090515137, "learning_rate": 1.313038669214531e-05, "loss": 0.1608677864074707, "step": 15820 }, { "epoch": 0.4169863320415011, "grad_norm": 1.3896723985671997, "learning_rate": 1.312228296482916e-05, "loss": 0.15141500234603883, "step": 15830 }, { "epoch": 0.4172497472859998, "grad_norm": 1.5144537687301636, "learning_rate": 1.311417696498076e-05, "loss": 0.1521978497505188, "step": 15840 }, { "epoch": 0.4175131625304985, "grad_norm": 2.3484325408935547, "learning_rate": 1.310606869850001e-05, "loss": 0.13584163188934326, "step": 15850 }, { "epoch": 0.41777657777499727, "grad_norm": 1.5679718255996704, "learning_rate": 1.3097958171288453e-05, "loss": 0.14886956214904784, "step": 15860 }, { "epoch": 0.418039993019496, "grad_norm": 5.557840347290039, "learning_rate": 1.3089845389249282e-05, "loss": 0.148595130443573, "step": 15870 }, { "epoch": 0.41830340826399476, "grad_norm": 1.7717138528823853, "learning_rate": 1.308173035828733e-05, "loss": 0.15241363048553466, "step": 15880 }, { "epoch": 0.4185668235084935, "grad_norm": 1.5234938859939575, "learning_rate": 1.3073613084309065e-05, "loss": 0.14910242557525635, "step": 15890 }, { "epoch": 0.41883023875299225, "grad_norm": 4.543954849243164, "learning_rate": 1.3065493573222594e-05, "loss": 0.16383492946624756, "step": 15900 }, { "epoch": 0.419093653997491, "grad_norm": 1.1505203247070312, "learning_rate": 1.305737183093764e-05, "loss": 0.1514114499092102, "step": 15910 }, { "epoch": 0.41935706924198973, "grad_norm": 1.7117079496383667, "learning_rate": 1.3049247863365562e-05, "loss": 0.14880661964416503, "step": 15920 }, { "epoch": 0.4196204844864884, "grad_norm": 1.8665906190872192, "learning_rate": 1.3041121676419333e-05, "loss": 0.14779183864593506, "step": 15930 }, { "epoch": 0.41988389973098716, "grad_norm": 1.9916080236434937, "learning_rate": 1.3032993276013546e-05, "loss": 0.14795541763305664, "step": 15940 }, { "epoch": 0.4201473149754859, "grad_norm": 1.8008073568344116, "learning_rate": 1.3024862668064393e-05, "loss": 0.1451387882232666, "step": 15950 }, { "epoch": 0.42041073021998465, "grad_norm": 1.3196710348129272, "learning_rate": 1.3016729858489689e-05, "loss": 0.15886299610137938, "step": 15960 }, { "epoch": 0.4206741454644834, "grad_norm": 1.7665629386901855, "learning_rate": 1.300859485320884e-05, "loss": 0.15813443660736085, "step": 15970 }, { "epoch": 0.42093756070898214, "grad_norm": 3.241222620010376, "learning_rate": 1.3000457658142858e-05, "loss": 0.15802288055419922, "step": 15980 }, { "epoch": 0.4212009759534809, "grad_norm": 2.712970733642578, "learning_rate": 1.299231827921434e-05, "loss": 0.15506337881088256, "step": 15990 }, { "epoch": 0.4214643911979796, "grad_norm": 1.5885953903198242, "learning_rate": 1.2984176722347478e-05, "loss": 0.1496671676635742, "step": 16000 }, { "epoch": 0.4217278064424783, "grad_norm": 1.7316136360168457, "learning_rate": 1.2976032993468049e-05, "loss": 0.14996217489242553, "step": 16010 }, { "epoch": 0.42199122168697706, "grad_norm": 4.041691303253174, "learning_rate": 1.296788709850341e-05, "loss": 0.1359616756439209, "step": 16020 }, { "epoch": 0.4222546369314758, "grad_norm": 2.2151780128479004, "learning_rate": 1.29597390433825e-05, "loss": 0.1350279211997986, "step": 16030 }, { "epoch": 0.42251805217597455, "grad_norm": 2.8426289558410645, "learning_rate": 1.2951588834035813e-05, "loss": 0.14434511661529542, "step": 16040 }, { "epoch": 0.4227814674204733, "grad_norm": 1.4796301126480103, "learning_rate": 1.2943436476395435e-05, "loss": 0.15976780652999878, "step": 16050 }, { "epoch": 0.42304488266497203, "grad_norm": 0.9206482172012329, "learning_rate": 1.2935281976394998e-05, "loss": 0.15785937309265136, "step": 16060 }, { "epoch": 0.4233082979094708, "grad_norm": 3.477235794067383, "learning_rate": 1.2927125339969701e-05, "loss": 0.14422284364700316, "step": 16070 }, { "epoch": 0.4235717131539695, "grad_norm": 1.8465417623519897, "learning_rate": 1.2918966573056291e-05, "loss": 0.14408543109893798, "step": 16080 }, { "epoch": 0.42383512839846826, "grad_norm": 2.011043071746826, "learning_rate": 1.2910805681593078e-05, "loss": 0.1629809856414795, "step": 16090 }, { "epoch": 0.42409854364296695, "grad_norm": 1.2926278114318848, "learning_rate": 1.2902642671519903e-05, "loss": 0.1482851505279541, "step": 16100 }, { "epoch": 0.4243619588874657, "grad_norm": 1.0406163930892944, "learning_rate": 1.2894477548778164e-05, "loss": 0.1596511721611023, "step": 16110 }, { "epoch": 0.42462537413196444, "grad_norm": 1.3566416501998901, "learning_rate": 1.2886310319310787e-05, "loss": 0.1550499677658081, "step": 16120 }, { "epoch": 0.4248887893764632, "grad_norm": 2.389435052871704, "learning_rate": 1.2878140989062233e-05, "loss": 0.15986433029174804, "step": 16130 }, { "epoch": 0.4251522046209619, "grad_norm": 2.75429368019104, "learning_rate": 1.2869969563978493e-05, "loss": 0.15871202945709229, "step": 16140 }, { "epoch": 0.42541561986546067, "grad_norm": 3.7886457443237305, "learning_rate": 1.2861796050007082e-05, "loss": 0.1480776309967041, "step": 16150 }, { "epoch": 0.4256790351099594, "grad_norm": 1.695283055305481, "learning_rate": 1.2853620453097039e-05, "loss": 0.16909905672073364, "step": 16160 }, { "epoch": 0.42594245035445816, "grad_norm": 1.3267077207565308, "learning_rate": 1.2845442779198916e-05, "loss": 0.14822362661361693, "step": 16170 }, { "epoch": 0.4262058655989569, "grad_norm": 0.7791911363601685, "learning_rate": 1.2837263034264773e-05, "loss": 0.14012109041213988, "step": 16180 }, { "epoch": 0.4264692808434556, "grad_norm": 1.6815156936645508, "learning_rate": 1.2829081224248183e-05, "loss": 0.15312891006469725, "step": 16190 }, { "epoch": 0.42673269608795433, "grad_norm": 1.672685146331787, "learning_rate": 1.2820897355104224e-05, "loss": 0.15905375480651857, "step": 16200 }, { "epoch": 0.4269961113324531, "grad_norm": 4.124423980712891, "learning_rate": 1.2812711432789464e-05, "loss": 0.15411341190338135, "step": 16210 }, { "epoch": 0.4272595265769518, "grad_norm": 2.7128889560699463, "learning_rate": 1.2804523463261975e-05, "loss": 0.14127422571182252, "step": 16220 }, { "epoch": 0.42752294182145056, "grad_norm": 2.5476455688476562, "learning_rate": 1.2796333452481309e-05, "loss": 0.1590774178504944, "step": 16230 }, { "epoch": 0.4277863570659493, "grad_norm": 3.7668893337249756, "learning_rate": 1.2788141406408516e-05, "loss": 0.15852417945861816, "step": 16240 }, { "epoch": 0.42804977231044805, "grad_norm": 1.2041887044906616, "learning_rate": 1.2779947331006113e-05, "loss": 0.14868990182876587, "step": 16250 }, { "epoch": 0.4283131875549468, "grad_norm": 2.5385513305664062, "learning_rate": 1.2771751232238107e-05, "loss": 0.13811070919036866, "step": 16260 }, { "epoch": 0.42857660279944554, "grad_norm": 5.650758266448975, "learning_rate": 1.2763553116069969e-05, "loss": 0.15170156955718994, "step": 16270 }, { "epoch": 0.4288400180439442, "grad_norm": 1.0695912837982178, "learning_rate": 1.2755352988468646e-05, "loss": 0.1380717396736145, "step": 16280 }, { "epoch": 0.42910343328844297, "grad_norm": 2.584158420562744, "learning_rate": 1.2747150855402541e-05, "loss": 0.15550354719161988, "step": 16290 }, { "epoch": 0.4293668485329417, "grad_norm": 1.7701966762542725, "learning_rate": 1.2738946722841519e-05, "loss": 0.14550490379333497, "step": 16300 }, { "epoch": 0.42963026377744046, "grad_norm": 4.326870918273926, "learning_rate": 1.2730740596756908e-05, "loss": 0.13987202644348146, "step": 16310 }, { "epoch": 0.4298936790219392, "grad_norm": 4.786603927612305, "learning_rate": 1.2722532483121475e-05, "loss": 0.1546424388885498, "step": 16320 }, { "epoch": 0.43015709426643794, "grad_norm": 1.6808862686157227, "learning_rate": 1.271432238790944e-05, "loss": 0.1436220407485962, "step": 16330 }, { "epoch": 0.4304205095109367, "grad_norm": 2.9612581729888916, "learning_rate": 1.2706110317096463e-05, "loss": 0.14307485818862914, "step": 16340 }, { "epoch": 0.43068392475543543, "grad_norm": 4.76164436340332, "learning_rate": 1.269789627665965e-05, "loss": 0.157615327835083, "step": 16350 }, { "epoch": 0.4309473399999341, "grad_norm": 1.6333705186843872, "learning_rate": 1.2689680272577527e-05, "loss": 0.1664123058319092, "step": 16360 }, { "epoch": 0.43121075524443286, "grad_norm": 2.732064962387085, "learning_rate": 1.2681462310830061e-05, "loss": 0.1486262321472168, "step": 16370 }, { "epoch": 0.4314741704889316, "grad_norm": 2.519484043121338, "learning_rate": 1.2673242397398635e-05, "loss": 0.15321121215820313, "step": 16380 }, { "epoch": 0.43173758573343035, "grad_norm": 1.107654094696045, "learning_rate": 1.266502053826606e-05, "loss": 0.15550007820129394, "step": 16390 }, { "epoch": 0.4320010009779291, "grad_norm": 3.546283483505249, "learning_rate": 1.2656796739416557e-05, "loss": 0.15806305408477783, "step": 16400 }, { "epoch": 0.43226441622242784, "grad_norm": 1.8306317329406738, "learning_rate": 1.2648571006835764e-05, "loss": 0.13577303886413575, "step": 16410 }, { "epoch": 0.4325278314669266, "grad_norm": 2.3030757904052734, "learning_rate": 1.2640343346510724e-05, "loss": 0.15092291831970214, "step": 16420 }, { "epoch": 0.4327912467114253, "grad_norm": 1.2321237325668335, "learning_rate": 1.2632113764429884e-05, "loss": 0.15341951847076415, "step": 16430 }, { "epoch": 0.43305466195592407, "grad_norm": 2.01873517036438, "learning_rate": 1.2623882266583083e-05, "loss": 0.15529465675354004, "step": 16440 }, { "epoch": 0.43331807720042276, "grad_norm": 2.625983953475952, "learning_rate": 1.2615648858961564e-05, "loss": 0.14592926502227782, "step": 16450 }, { "epoch": 0.4335814924449215, "grad_norm": 3.1897857189178467, "learning_rate": 1.260741354755796e-05, "loss": 0.1549884557723999, "step": 16460 }, { "epoch": 0.43384490768942024, "grad_norm": 1.1135693788528442, "learning_rate": 1.2599176338366282e-05, "loss": 0.16219625473022461, "step": 16470 }, { "epoch": 0.434108322933919, "grad_norm": 3.3730714321136475, "learning_rate": 1.2590937237381927e-05, "loss": 0.14137471914291383, "step": 16480 }, { "epoch": 0.43437173817841773, "grad_norm": 1.523758888244629, "learning_rate": 1.2582696250601667e-05, "loss": 0.15228524208068847, "step": 16490 }, { "epoch": 0.4346351534229165, "grad_norm": 2.9039032459259033, "learning_rate": 1.257445338402365e-05, "loss": 0.15231372117996217, "step": 16500 }, { "epoch": 0.4348985686674152, "grad_norm": 1.7519501447677612, "learning_rate": 1.256620864364739e-05, "loss": 0.1589857816696167, "step": 16510 }, { "epoch": 0.43516198391191396, "grad_norm": 1.098525047302246, "learning_rate": 1.255796203547376e-05, "loss": 0.14709537029266356, "step": 16520 }, { "epoch": 0.4354253991564127, "grad_norm": 1.6011368036270142, "learning_rate": 1.2549713565505001e-05, "loss": 0.14676520824432374, "step": 16530 }, { "epoch": 0.4356888144009114, "grad_norm": 1.7880967855453491, "learning_rate": 1.254146323974471e-05, "loss": 0.1365620255470276, "step": 16540 }, { "epoch": 0.43595222964541014, "grad_norm": 2.745271921157837, "learning_rate": 1.253321106419782e-05, "loss": 0.1371903657913208, "step": 16550 }, { "epoch": 0.4362156448899089, "grad_norm": 0.8839582204818726, "learning_rate": 1.2524957044870627e-05, "loss": 0.147552490234375, "step": 16560 }, { "epoch": 0.4364790601344076, "grad_norm": 1.404618263244629, "learning_rate": 1.2516701187770761e-05, "loss": 0.15086331367492675, "step": 16570 }, { "epoch": 0.43674247537890637, "grad_norm": 2.3689496517181396, "learning_rate": 1.250844349890719e-05, "loss": 0.1492406964302063, "step": 16580 }, { "epoch": 0.4370058906234051, "grad_norm": 1.65335214138031, "learning_rate": 1.250018398429022e-05, "loss": 0.15691442489624025, "step": 16590 }, { "epoch": 0.43726930586790386, "grad_norm": 4.260436058044434, "learning_rate": 1.2491922649931475e-05, "loss": 0.16122760772705078, "step": 16600 }, { "epoch": 0.4375327211124026, "grad_norm": 4.580180644989014, "learning_rate": 1.2483659501843916e-05, "loss": 0.15541067123413085, "step": 16610 }, { "epoch": 0.4377961363569013, "grad_norm": 0.6729968190193176, "learning_rate": 1.2475394546041822e-05, "loss": 0.1423407554626465, "step": 16620 }, { "epoch": 0.43805955160140003, "grad_norm": 2.184600353240967, "learning_rate": 1.2467127788540774e-05, "loss": 0.13905742168426513, "step": 16630 }, { "epoch": 0.4383229668458988, "grad_norm": 1.1764639616012573, "learning_rate": 1.245885923535768e-05, "loss": 0.15017271041870117, "step": 16640 }, { "epoch": 0.4385863820903975, "grad_norm": 0.8532541990280151, "learning_rate": 1.2450588892510756e-05, "loss": 0.1537682294845581, "step": 16650 }, { "epoch": 0.43884979733489626, "grad_norm": 1.0473750829696655, "learning_rate": 1.2442316766019502e-05, "loss": 0.15995895862579346, "step": 16660 }, { "epoch": 0.439113212579395, "grad_norm": 2.5840160846710205, "learning_rate": 1.2434042861904743e-05, "loss": 0.14813891649246216, "step": 16670 }, { "epoch": 0.43937662782389375, "grad_norm": 2.0361719131469727, "learning_rate": 1.2425767186188569e-05, "loss": 0.14267356395721437, "step": 16680 }, { "epoch": 0.4396400430683925, "grad_norm": 2.4636311531066895, "learning_rate": 1.2417489744894381e-05, "loss": 0.14081921577453613, "step": 16690 }, { "epoch": 0.43990345831289124, "grad_norm": 2.76338791847229, "learning_rate": 1.2409210544046859e-05, "loss": 0.14067955017089845, "step": 16700 }, { "epoch": 0.4401668735573899, "grad_norm": 2.9716148376464844, "learning_rate": 1.2400929589671962e-05, "loss": 0.15376908779144288, "step": 16710 }, { "epoch": 0.44043028880188867, "grad_norm": 3.828425407409668, "learning_rate": 1.2392646887796924e-05, "loss": 0.1589216947555542, "step": 16720 }, { "epoch": 0.4406937040463874, "grad_norm": 0.9986746907234192, "learning_rate": 1.2384362444450253e-05, "loss": 0.15103663206100465, "step": 16730 }, { "epoch": 0.44095711929088616, "grad_norm": 1.1920894384384155, "learning_rate": 1.2376076265661722e-05, "loss": 0.16067532300949097, "step": 16740 }, { "epoch": 0.4412205345353849, "grad_norm": 3.0548219680786133, "learning_rate": 1.2367788357462375e-05, "loss": 0.15166118144989013, "step": 16750 }, { "epoch": 0.44148394977988364, "grad_norm": 3.5365023612976074, "learning_rate": 1.2359498725884509e-05, "loss": 0.14785091876983641, "step": 16760 }, { "epoch": 0.4417473650243824, "grad_norm": 1.7984591722488403, "learning_rate": 1.2351207376961666e-05, "loss": 0.15136566162109374, "step": 16770 }, { "epoch": 0.44201078026888113, "grad_norm": 2.3292462825775146, "learning_rate": 1.234291431672866e-05, "loss": 0.1511287808418274, "step": 16780 }, { "epoch": 0.4422741955133799, "grad_norm": 1.8954432010650635, "learning_rate": 1.2334619551221531e-05, "loss": 0.144712495803833, "step": 16790 }, { "epoch": 0.44253761075787856, "grad_norm": 2.366711139678955, "learning_rate": 1.232632308647757e-05, "loss": 0.1579180121421814, "step": 16800 }, { "epoch": 0.4428010260023773, "grad_norm": 4.1478986740112305, "learning_rate": 1.2318024928535296e-05, "loss": 0.1495264768600464, "step": 16810 }, { "epoch": 0.44306444124687605, "grad_norm": 1.5250355005264282, "learning_rate": 1.2309725083434477e-05, "loss": 0.15140910148620607, "step": 16820 }, { "epoch": 0.4433278564913748, "grad_norm": 0.9479109048843384, "learning_rate": 1.2301423557216088e-05, "loss": 0.15858159065246583, "step": 16830 }, { "epoch": 0.44359127173587354, "grad_norm": 2.373446226119995, "learning_rate": 1.2293120355922346e-05, "loss": 0.140635347366333, "step": 16840 }, { "epoch": 0.4438546869803723, "grad_norm": 1.0205456018447876, "learning_rate": 1.2284815485596671e-05, "loss": 0.13459579944610595, "step": 16850 }, { "epoch": 0.444118102224871, "grad_norm": 2.618030309677124, "learning_rate": 1.2276508952283712e-05, "loss": 0.14534547328948974, "step": 16860 }, { "epoch": 0.44438151746936977, "grad_norm": 3.9646451473236084, "learning_rate": 1.2268200762029319e-05, "loss": 0.1449613094329834, "step": 16870 }, { "epoch": 0.44464493271386846, "grad_norm": 1.398246169090271, "learning_rate": 1.2259890920880555e-05, "loss": 0.15144954919815062, "step": 16880 }, { "epoch": 0.4449083479583672, "grad_norm": 3.513338804244995, "learning_rate": 1.2251579434885679e-05, "loss": 0.14747343063354493, "step": 16890 }, { "epoch": 0.44517176320286594, "grad_norm": 1.2358038425445557, "learning_rate": 1.2243266310094146e-05, "loss": 0.143426775932312, "step": 16900 }, { "epoch": 0.4454351784473647, "grad_norm": 2.825491428375244, "learning_rate": 1.2234951552556613e-05, "loss": 0.15707589387893678, "step": 16910 }, { "epoch": 0.44569859369186343, "grad_norm": 1.9718925952911377, "learning_rate": 1.2226635168324916e-05, "loss": 0.1485974073410034, "step": 16920 }, { "epoch": 0.4459620089363622, "grad_norm": 1.86538565158844, "learning_rate": 1.2218317163452078e-05, "loss": 0.13546431064605713, "step": 16930 }, { "epoch": 0.4462254241808609, "grad_norm": 3.2278940677642822, "learning_rate": 1.2209997543992301e-05, "loss": 0.14734671115875245, "step": 16940 }, { "epoch": 0.44648883942535966, "grad_norm": 5.505903244018555, "learning_rate": 1.2201676316000964e-05, "loss": 0.1548520803451538, "step": 16950 }, { "epoch": 0.4467522546698584, "grad_norm": 1.583135962486267, "learning_rate": 1.2193353485534614e-05, "loss": 0.1464773893356323, "step": 16960 }, { "epoch": 0.4470156699143571, "grad_norm": 1.1160922050476074, "learning_rate": 1.2185029058650972e-05, "loss": 0.14453625679016113, "step": 16970 }, { "epoch": 0.44727908515885584, "grad_norm": 2.33665132522583, "learning_rate": 1.2176703041408908e-05, "loss": 0.14881917238235473, "step": 16980 }, { "epoch": 0.4475425004033546, "grad_norm": 4.112253189086914, "learning_rate": 1.2168375439868464e-05, "loss": 0.15010364055633546, "step": 16990 }, { "epoch": 0.4478059156478533, "grad_norm": 1.5828948020935059, "learning_rate": 1.2160046260090816e-05, "loss": 0.1409573435783386, "step": 17000 }, { "epoch": 0.44806933089235207, "grad_norm": 3.2934083938598633, "learning_rate": 1.2151715508138313e-05, "loss": 0.14017102718353272, "step": 17010 }, { "epoch": 0.4483327461368508, "grad_norm": 3.120731830596924, "learning_rate": 1.2143383190074429e-05, "loss": 0.14271165132522584, "step": 17020 }, { "epoch": 0.44859616138134956, "grad_norm": 4.580853462219238, "learning_rate": 1.2135049311963787e-05, "loss": 0.15783565044403075, "step": 17030 }, { "epoch": 0.4488595766258483, "grad_norm": 1.251008152961731, "learning_rate": 1.2126713879872138e-05, "loss": 0.15406513214111328, "step": 17040 }, { "epoch": 0.44912299187034704, "grad_norm": 1.3726242780685425, "learning_rate": 1.2118376899866375e-05, "loss": 0.15764734745025635, "step": 17050 }, { "epoch": 0.44938640711484573, "grad_norm": 6.1211934089660645, "learning_rate": 1.2110038378014512e-05, "loss": 0.14602112770080566, "step": 17060 }, { "epoch": 0.4496498223593445, "grad_norm": 1.8132952451705933, "learning_rate": 1.210169832038568e-05, "loss": 0.14377207756042482, "step": 17070 }, { "epoch": 0.4499132376038432, "grad_norm": 1.3873728513717651, "learning_rate": 1.2093356733050137e-05, "loss": 0.13456614017486573, "step": 17080 }, { "epoch": 0.45017665284834196, "grad_norm": 4.477853298187256, "learning_rate": 1.2085013622079248e-05, "loss": 0.16123108863830565, "step": 17090 }, { "epoch": 0.4504400680928407, "grad_norm": 2.1596829891204834, "learning_rate": 1.2076668993545491e-05, "loss": 0.14471949338912965, "step": 17100 }, { "epoch": 0.45070348333733945, "grad_norm": 5.360065937042236, "learning_rate": 1.206832285352244e-05, "loss": 0.1643162965774536, "step": 17110 }, { "epoch": 0.4509668985818382, "grad_norm": 1.9877153635025024, "learning_rate": 1.2059975208084787e-05, "loss": 0.1514907717704773, "step": 17120 }, { "epoch": 0.45123031382633694, "grad_norm": 3.861067533493042, "learning_rate": 1.2051626063308299e-05, "loss": 0.14777736663818358, "step": 17130 }, { "epoch": 0.4514937290708356, "grad_norm": 1.2028014659881592, "learning_rate": 1.2043275425269854e-05, "loss": 0.1391960859298706, "step": 17140 }, { "epoch": 0.45175714431533437, "grad_norm": 2.2473199367523193, "learning_rate": 1.2034923300047391e-05, "loss": 0.1494433045387268, "step": 17150 }, { "epoch": 0.4520205595598331, "grad_norm": 3.3419487476348877, "learning_rate": 1.2026569693719963e-05, "loss": 0.14053165912628174, "step": 17160 }, { "epoch": 0.45228397480433186, "grad_norm": 1.2570676803588867, "learning_rate": 1.2018214612367673e-05, "loss": 0.14447324275970458, "step": 17170 }, { "epoch": 0.4525473900488306, "grad_norm": 1.3334836959838867, "learning_rate": 1.200985806207172e-05, "loss": 0.1426562786102295, "step": 17180 }, { "epoch": 0.45281080529332934, "grad_norm": 2.935546636581421, "learning_rate": 1.2001500048914354e-05, "loss": 0.13811299800872803, "step": 17190 }, { "epoch": 0.4530742205378281, "grad_norm": 1.2067712545394897, "learning_rate": 1.1993140578978898e-05, "loss": 0.15081963539123536, "step": 17200 }, { "epoch": 0.45333763578232683, "grad_norm": 1.1049250364303589, "learning_rate": 1.1984779658349743e-05, "loss": 0.14595334529876708, "step": 17210 }, { "epoch": 0.4536010510268256, "grad_norm": 2.1348016262054443, "learning_rate": 1.1976417293112318e-05, "loss": 0.13773831129074096, "step": 17220 }, { "epoch": 0.45386446627132426, "grad_norm": 2.270219087600708, "learning_rate": 1.1968053489353122e-05, "loss": 0.13995764255523682, "step": 17230 }, { "epoch": 0.454127881515823, "grad_norm": 3.504542350769043, "learning_rate": 1.195968825315968e-05, "loss": 0.1525423526763916, "step": 17240 }, { "epoch": 0.45439129676032175, "grad_norm": 1.6672719717025757, "learning_rate": 1.1951321590620583e-05, "loss": 0.15233614444732665, "step": 17250 }, { "epoch": 0.4546547120048205, "grad_norm": 3.536714553833008, "learning_rate": 1.1942953507825446e-05, "loss": 0.14974691867828369, "step": 17260 }, { "epoch": 0.45491812724931924, "grad_norm": 2.105053663253784, "learning_rate": 1.1934584010864921e-05, "loss": 0.1486322045326233, "step": 17270 }, { "epoch": 0.455181542493818, "grad_norm": 1.0727455615997314, "learning_rate": 1.1926213105830683e-05, "loss": 0.14923336505889892, "step": 17280 }, { "epoch": 0.4554449577383167, "grad_norm": 2.0237762928009033, "learning_rate": 1.1917840798815448e-05, "loss": 0.15838017463684081, "step": 17290 }, { "epoch": 0.45570837298281547, "grad_norm": 2.4039223194122314, "learning_rate": 1.1909467095912932e-05, "loss": 0.1590579032897949, "step": 17300 }, { "epoch": 0.4559717882273142, "grad_norm": 4.218881607055664, "learning_rate": 1.1901092003217885e-05, "loss": 0.1460996150970459, "step": 17310 }, { "epoch": 0.4562352034718129, "grad_norm": 2.1617608070373535, "learning_rate": 1.1892715526826058e-05, "loss": 0.15459835529327393, "step": 17320 }, { "epoch": 0.45649861871631164, "grad_norm": 0.9742910861968994, "learning_rate": 1.1884337672834212e-05, "loss": 0.14198665618896483, "step": 17330 }, { "epoch": 0.4567620339608104, "grad_norm": 2.067317485809326, "learning_rate": 1.1875958447340108e-05, "loss": 0.13575422763824463, "step": 17340 }, { "epoch": 0.45702544920530913, "grad_norm": 4.921116352081299, "learning_rate": 1.1867577856442514e-05, "loss": 0.14759972095489501, "step": 17350 }, { "epoch": 0.4572888644498079, "grad_norm": 2.7339723110198975, "learning_rate": 1.1859195906241182e-05, "loss": 0.14463484287261963, "step": 17360 }, { "epoch": 0.4575522796943066, "grad_norm": 7.373263835906982, "learning_rate": 1.1850812602836851e-05, "loss": 0.15630874633789063, "step": 17370 }, { "epoch": 0.45781569493880536, "grad_norm": 1.3663195371627808, "learning_rate": 1.1842427952331263e-05, "loss": 0.14452669620513917, "step": 17380 }, { "epoch": 0.4580791101833041, "grad_norm": 3.2784483432769775, "learning_rate": 1.183404196082712e-05, "loss": 0.15899996757507323, "step": 17390 }, { "epoch": 0.4583425254278028, "grad_norm": 2.514662027359009, "learning_rate": 1.1825654634428111e-05, "loss": 0.14027132987976074, "step": 17400 }, { "epoch": 0.45860594067230154, "grad_norm": 3.60013747215271, "learning_rate": 1.1817265979238892e-05, "loss": 0.1542532205581665, "step": 17410 }, { "epoch": 0.4588693559168003, "grad_norm": 1.278398036956787, "learning_rate": 1.1808876001365093e-05, "loss": 0.15356092453002929, "step": 17420 }, { "epoch": 0.459132771161299, "grad_norm": 5.018589496612549, "learning_rate": 1.1800484706913294e-05, "loss": 0.15330156087875366, "step": 17430 }, { "epoch": 0.45939618640579777, "grad_norm": 2.661078453063965, "learning_rate": 1.1792092101991052e-05, "loss": 0.14476736783981323, "step": 17440 }, { "epoch": 0.4596596016502965, "grad_norm": 5.198332786560059, "learning_rate": 1.1783698192706856e-05, "loss": 0.15171046257019044, "step": 17450 }, { "epoch": 0.45992301689479526, "grad_norm": 2.1296629905700684, "learning_rate": 1.1775302985170161e-05, "loss": 0.15213580131530763, "step": 17460 }, { "epoch": 0.460186432139294, "grad_norm": 1.807611107826233, "learning_rate": 1.1766906485491361e-05, "loss": 0.15824491977691652, "step": 17470 }, { "epoch": 0.46044984738379274, "grad_norm": 2.2700228691101074, "learning_rate": 1.175850869978179e-05, "loss": 0.15731542110443114, "step": 17480 }, { "epoch": 0.46071326262829143, "grad_norm": 1.3783830404281616, "learning_rate": 1.1750109634153719e-05, "loss": 0.1372069835662842, "step": 17490 }, { "epoch": 0.4609766778727902, "grad_norm": 4.675701141357422, "learning_rate": 1.1741709294720346e-05, "loss": 0.13481054306030274, "step": 17500 }, { "epoch": 0.4612400931172889, "grad_norm": 3.779128313064575, "learning_rate": 1.173330768759581e-05, "loss": 0.14416314363479615, "step": 17510 }, { "epoch": 0.46150350836178766, "grad_norm": 1.9421943426132202, "learning_rate": 1.1724904818895158e-05, "loss": 0.14255284070968627, "step": 17520 }, { "epoch": 0.4617669236062864, "grad_norm": 1.3147510290145874, "learning_rate": 1.1716500694734359e-05, "loss": 0.1569844126701355, "step": 17530 }, { "epoch": 0.46203033885078515, "grad_norm": 2.3971195220947266, "learning_rate": 1.1708095321230295e-05, "loss": 0.14553686380386352, "step": 17540 }, { "epoch": 0.4622937540952839, "grad_norm": 2.05509090423584, "learning_rate": 1.1699688704500765e-05, "loss": 0.13719685077667237, "step": 17550 }, { "epoch": 0.46255716933978264, "grad_norm": 2.11871337890625, "learning_rate": 1.1691280850664463e-05, "loss": 0.14599391222000122, "step": 17560 }, { "epoch": 0.4628205845842814, "grad_norm": 1.9210566282272339, "learning_rate": 1.1682871765840996e-05, "loss": 0.14794089794158935, "step": 17570 }, { "epoch": 0.46308399982878007, "grad_norm": 2.4164342880249023, "learning_rate": 1.1674461456150845e-05, "loss": 0.1528932809829712, "step": 17580 }, { "epoch": 0.4633474150732788, "grad_norm": 4.166483402252197, "learning_rate": 1.1666049927715408e-05, "loss": 0.14749835729598998, "step": 17590 }, { "epoch": 0.46361083031777756, "grad_norm": 1.9893544912338257, "learning_rate": 1.1657637186656952e-05, "loss": 0.13949657678604127, "step": 17600 }, { "epoch": 0.4638742455622763, "grad_norm": 1.3753890991210938, "learning_rate": 1.1649223239098639e-05, "loss": 0.15692858695983886, "step": 17610 }, { "epoch": 0.46413766080677504, "grad_norm": 2.8728229999542236, "learning_rate": 1.1640808091164495e-05, "loss": 0.1440461754798889, "step": 17620 }, { "epoch": 0.4644010760512738, "grad_norm": 8.885788917541504, "learning_rate": 1.1632391748979435e-05, "loss": 0.14077636003494262, "step": 17630 }, { "epoch": 0.46466449129577253, "grad_norm": 3.3379156589508057, "learning_rate": 1.162397421866923e-05, "loss": 0.15194165706634521, "step": 17640 }, { "epoch": 0.4649279065402713, "grad_norm": 5.551271438598633, "learning_rate": 1.1615555506360526e-05, "loss": 0.15804413557052613, "step": 17650 }, { "epoch": 0.46519132178476996, "grad_norm": 1.4148998260498047, "learning_rate": 1.1607135618180823e-05, "loss": 0.16067948341369628, "step": 17660 }, { "epoch": 0.4654547370292687, "grad_norm": 2.4548497200012207, "learning_rate": 1.1598714560258477e-05, "loss": 0.15272111892700196, "step": 17670 }, { "epoch": 0.46571815227376745, "grad_norm": 3.054800271987915, "learning_rate": 1.1590292338722702e-05, "loss": 0.157637357711792, "step": 17680 }, { "epoch": 0.4659815675182662, "grad_norm": 3.559147834777832, "learning_rate": 1.1581868959703553e-05, "loss": 0.1398775100708008, "step": 17690 }, { "epoch": 0.46624498276276494, "grad_norm": 4.0827436447143555, "learning_rate": 1.1573444429331926e-05, "loss": 0.14088406562805175, "step": 17700 }, { "epoch": 0.4665083980072637, "grad_norm": 1.1348927021026611, "learning_rate": 1.1565018753739559e-05, "loss": 0.14896018505096437, "step": 17710 }, { "epoch": 0.4667718132517624, "grad_norm": 1.1768771409988403, "learning_rate": 1.1556591939059024e-05, "loss": 0.14079153537750244, "step": 17720 }, { "epoch": 0.46703522849626117, "grad_norm": 5.544724464416504, "learning_rate": 1.1548163991423716e-05, "loss": 0.14295668601989747, "step": 17730 }, { "epoch": 0.4672986437407599, "grad_norm": 1.88199782371521, "learning_rate": 1.1539734916967864e-05, "loss": 0.14649496078491211, "step": 17740 }, { "epoch": 0.4675620589852586, "grad_norm": 1.4817789793014526, "learning_rate": 1.1531304721826506e-05, "loss": 0.13858602046966553, "step": 17750 }, { "epoch": 0.46782547422975734, "grad_norm": 2.995091438293457, "learning_rate": 1.1522873412135509e-05, "loss": 0.14381009340286255, "step": 17760 }, { "epoch": 0.4680888894742561, "grad_norm": 1.8309831619262695, "learning_rate": 1.1514440994031536e-05, "loss": 0.13227418661117554, "step": 17770 }, { "epoch": 0.46835230471875483, "grad_norm": 1.4755866527557373, "learning_rate": 1.1506007473652075e-05, "loss": 0.15096597671508788, "step": 17780 }, { "epoch": 0.4686157199632536, "grad_norm": 4.924942493438721, "learning_rate": 1.1497572857135397e-05, "loss": 0.14888689517974854, "step": 17790 }, { "epoch": 0.4688791352077523, "grad_norm": 3.443530321121216, "learning_rate": 1.1489137150620582e-05, "loss": 0.16082797050476075, "step": 17800 }, { "epoch": 0.46914255045225106, "grad_norm": 1.5364506244659424, "learning_rate": 1.1480700360247504e-05, "loss": 0.1544039487838745, "step": 17810 }, { "epoch": 0.4694059656967498, "grad_norm": 3.3682072162628174, "learning_rate": 1.147226249215682e-05, "loss": 0.1571389317512512, "step": 17820 }, { "epoch": 0.46966938094124855, "grad_norm": 1.7392265796661377, "learning_rate": 1.1463823552489977e-05, "loss": 0.14893791675567628, "step": 17830 }, { "epoch": 0.46993279618574724, "grad_norm": 2.7322871685028076, "learning_rate": 1.1455383547389193e-05, "loss": 0.14399304389953613, "step": 17840 }, { "epoch": 0.470196211430246, "grad_norm": 2.6648569107055664, "learning_rate": 1.1446942482997479e-05, "loss": 0.13883116245269775, "step": 17850 }, { "epoch": 0.4704596266747447, "grad_norm": 1.6584136486053467, "learning_rate": 1.1438500365458593e-05, "loss": 0.13720910549163817, "step": 17860 }, { "epoch": 0.47072304191924347, "grad_norm": 2.165189504623413, "learning_rate": 1.1430057200917084e-05, "loss": 0.1521741509437561, "step": 17870 }, { "epoch": 0.4709864571637422, "grad_norm": 2.3159923553466797, "learning_rate": 1.1421612995518243e-05, "loss": 0.14369475841522217, "step": 17880 }, { "epoch": 0.47124987240824096, "grad_norm": 2.066889524459839, "learning_rate": 1.1413167755408132e-05, "loss": 0.1518293023109436, "step": 17890 }, { "epoch": 0.4715132876527397, "grad_norm": 2.309795618057251, "learning_rate": 1.1404721486733558e-05, "loss": 0.15224547386169435, "step": 17900 }, { "epoch": 0.47177670289723844, "grad_norm": 2.0100133419036865, "learning_rate": 1.1396274195642085e-05, "loss": 0.13798768520355226, "step": 17910 }, { "epoch": 0.47204011814173713, "grad_norm": 1.8168342113494873, "learning_rate": 1.1387825888282006e-05, "loss": 0.15514146089553832, "step": 17920 }, { "epoch": 0.4723035333862359, "grad_norm": 1.6169188022613525, "learning_rate": 1.1379376570802373e-05, "loss": 0.1338021993637085, "step": 17930 }, { "epoch": 0.4725669486307346, "grad_norm": 1.2004348039627075, "learning_rate": 1.137092624935296e-05, "loss": 0.144059956073761, "step": 17940 }, { "epoch": 0.47283036387523336, "grad_norm": 5.080802917480469, "learning_rate": 1.1362474930084278e-05, "loss": 0.15527124404907228, "step": 17950 }, { "epoch": 0.4730937791197321, "grad_norm": 1.0158066749572754, "learning_rate": 1.1354022619147557e-05, "loss": 0.13757712841033937, "step": 17960 }, { "epoch": 0.47335719436423085, "grad_norm": 2.3437864780426025, "learning_rate": 1.1345569322694756e-05, "loss": 0.1446049690246582, "step": 17970 }, { "epoch": 0.4736206096087296, "grad_norm": 4.523947715759277, "learning_rate": 1.133711504687855e-05, "loss": 0.15159082412719727, "step": 17980 }, { "epoch": 0.47388402485322834, "grad_norm": 1.5513962507247925, "learning_rate": 1.1328659797852325e-05, "loss": 0.1407231330871582, "step": 17990 }, { "epoch": 0.4741474400977271, "grad_norm": 3.2631137371063232, "learning_rate": 1.1320203581770177e-05, "loss": 0.15893909931182862, "step": 18000 }, { "epoch": 0.47441085534222577, "grad_norm": 6.170321464538574, "learning_rate": 1.1311746404786903e-05, "loss": 0.15394902229309082, "step": 18010 }, { "epoch": 0.4746742705867245, "grad_norm": 4.100283622741699, "learning_rate": 1.1303288273058002e-05, "loss": 0.14365382194519044, "step": 18020 }, { "epoch": 0.47493768583122326, "grad_norm": 1.2624465227127075, "learning_rate": 1.1294829192739665e-05, "loss": 0.14673657417297364, "step": 18030 }, { "epoch": 0.475201101075722, "grad_norm": 1.1331801414489746, "learning_rate": 1.1286369169988782e-05, "loss": 0.15330471992492675, "step": 18040 }, { "epoch": 0.47546451632022074, "grad_norm": 2.3366100788116455, "learning_rate": 1.1277908210962916e-05, "loss": 0.15688188076019288, "step": 18050 }, { "epoch": 0.4757279315647195, "grad_norm": 2.543347120285034, "learning_rate": 1.126944632182032e-05, "loss": 0.14684727191925048, "step": 18060 }, { "epoch": 0.47599134680921823, "grad_norm": 2.1307129859924316, "learning_rate": 1.1260983508719922e-05, "loss": 0.13786942958831788, "step": 18070 }, { "epoch": 0.476254762053717, "grad_norm": 3.0553300380706787, "learning_rate": 1.1252519777821324e-05, "loss": 0.1510005235671997, "step": 18080 }, { "epoch": 0.4765181772982157, "grad_norm": 3.894409656524658, "learning_rate": 1.1244055135284795e-05, "loss": 0.14362910985946656, "step": 18090 }, { "epoch": 0.4767815925427144, "grad_norm": 2.029390811920166, "learning_rate": 1.123558958727126e-05, "loss": 0.15089504718780516, "step": 18100 }, { "epoch": 0.47704500778721315, "grad_norm": 2.83849835395813, "learning_rate": 1.1227123139942322e-05, "loss": 0.1412869691848755, "step": 18110 }, { "epoch": 0.4773084230317119, "grad_norm": 1.640663981437683, "learning_rate": 1.1218655799460216e-05, "loss": 0.14860334396362304, "step": 18120 }, { "epoch": 0.47757183827621064, "grad_norm": 2.3339850902557373, "learning_rate": 1.1210187571987845e-05, "loss": 0.14167208671569825, "step": 18130 }, { "epoch": 0.4778352535207094, "grad_norm": 5.730471134185791, "learning_rate": 1.1201718463688745e-05, "loss": 0.16032683849334717, "step": 18140 }, { "epoch": 0.4780986687652081, "grad_norm": 2.116832733154297, "learning_rate": 1.1193248480727101e-05, "loss": 0.14836082458496094, "step": 18150 }, { "epoch": 0.47836208400970687, "grad_norm": 1.8777830600738525, "learning_rate": 1.118477762926773e-05, "loss": 0.13482969999313354, "step": 18160 }, { "epoch": 0.4786254992542056, "grad_norm": 2.302612066268921, "learning_rate": 1.117630591547609e-05, "loss": 0.15602471828460693, "step": 18170 }, { "epoch": 0.4788889144987043, "grad_norm": 2.109049081802368, "learning_rate": 1.1167833345518248e-05, "loss": 0.13850171566009523, "step": 18180 }, { "epoch": 0.47915232974320304, "grad_norm": 2.36637544631958, "learning_rate": 1.1159359925560918e-05, "loss": 0.14497255086898803, "step": 18190 }, { "epoch": 0.4794157449877018, "grad_norm": 4.575222492218018, "learning_rate": 1.115088566177141e-05, "loss": 0.1491103172302246, "step": 18200 }, { "epoch": 0.47967916023220053, "grad_norm": 1.9209293127059937, "learning_rate": 1.1142410560317669e-05, "loss": 0.15002882480621338, "step": 18210 }, { "epoch": 0.4799425754766993, "grad_norm": 1.6482740640640259, "learning_rate": 1.113393462736823e-05, "loss": 0.16031256914138795, "step": 18220 }, { "epoch": 0.480205990721198, "grad_norm": 1.5206444263458252, "learning_rate": 1.1125457869092251e-05, "loss": 0.1503826856613159, "step": 18230 }, { "epoch": 0.48046940596569676, "grad_norm": 2.2383522987365723, "learning_rate": 1.1116980291659474e-05, "loss": 0.1475675106048584, "step": 18240 }, { "epoch": 0.4807328212101955, "grad_norm": 2.502188205718994, "learning_rate": 1.1108501901240254e-05, "loss": 0.13911420106887817, "step": 18250 }, { "epoch": 0.48099623645469425, "grad_norm": 2.417482614517212, "learning_rate": 1.1100022704005524e-05, "loss": 0.14703365564346313, "step": 18260 }, { "epoch": 0.48125965169919294, "grad_norm": 5.655045032501221, "learning_rate": 1.109154270612681e-05, "loss": 0.1377067446708679, "step": 18270 }, { "epoch": 0.4815230669436917, "grad_norm": 2.1430959701538086, "learning_rate": 1.1083061913776224e-05, "loss": 0.1363696575164795, "step": 18280 }, { "epoch": 0.4817864821881904, "grad_norm": 2.5723588466644287, "learning_rate": 1.1074580333126448e-05, "loss": 0.1414236068725586, "step": 18290 }, { "epoch": 0.48204989743268917, "grad_norm": 1.8552695512771606, "learning_rate": 1.1066097970350747e-05, "loss": 0.14337577819824218, "step": 18300 }, { "epoch": 0.4823133126771879, "grad_norm": 0.9435719847679138, "learning_rate": 1.1057614831622942e-05, "loss": 0.1397321939468384, "step": 18310 }, { "epoch": 0.48257672792168665, "grad_norm": 0.9288102984428406, "learning_rate": 1.1049130923117437e-05, "loss": 0.14546693563461305, "step": 18320 }, { "epoch": 0.4828401431661854, "grad_norm": 1.2725909948349, "learning_rate": 1.1040646251009181e-05, "loss": 0.15397319793701172, "step": 18330 }, { "epoch": 0.48310355841068414, "grad_norm": 1.83091402053833, "learning_rate": 1.1032160821473692e-05, "loss": 0.1477686882019043, "step": 18340 }, { "epoch": 0.4833669736551829, "grad_norm": 4.157585620880127, "learning_rate": 1.1023674640687019e-05, "loss": 0.1523931860923767, "step": 18350 }, { "epoch": 0.4836303888996816, "grad_norm": 1.252753496170044, "learning_rate": 1.1015187714825782e-05, "loss": 0.14579737186431885, "step": 18360 }, { "epoch": 0.4838938041441803, "grad_norm": 0.9111543893814087, "learning_rate": 1.1006700050067122e-05, "loss": 0.13515433073043823, "step": 18370 }, { "epoch": 0.48415721938867906, "grad_norm": 2.8161065578460693, "learning_rate": 1.0998211652588737e-05, "loss": 0.1372082471847534, "step": 18380 }, { "epoch": 0.4844206346331778, "grad_norm": 1.9992785453796387, "learning_rate": 1.0989722528568845e-05, "loss": 0.15202436447143555, "step": 18390 }, { "epoch": 0.48468404987767655, "grad_norm": 3.328155517578125, "learning_rate": 1.0981232684186191e-05, "loss": 0.15179575681686402, "step": 18400 }, { "epoch": 0.4849474651221753, "grad_norm": 2.4213621616363525, "learning_rate": 1.0972742125620062e-05, "loss": 0.1501847267150879, "step": 18410 }, { "epoch": 0.48521088036667404, "grad_norm": 2.999082326889038, "learning_rate": 1.0964250859050245e-05, "loss": 0.13955732583999633, "step": 18420 }, { "epoch": 0.4854742956111728, "grad_norm": 1.5570768117904663, "learning_rate": 1.0955758890657051e-05, "loss": 0.14017401933670043, "step": 18430 }, { "epoch": 0.48573771085567147, "grad_norm": 3.9058170318603516, "learning_rate": 1.0947266226621298e-05, "loss": 0.12146221399307251, "step": 18440 }, { "epoch": 0.4860011261001702, "grad_norm": 2.1198720932006836, "learning_rate": 1.0938772873124321e-05, "loss": 0.14485933780670165, "step": 18450 }, { "epoch": 0.48626454134466895, "grad_norm": 1.7134709358215332, "learning_rate": 1.0930278836347943e-05, "loss": 0.13421876430511476, "step": 18460 }, { "epoch": 0.4865279565891677, "grad_norm": 0.9848700761795044, "learning_rate": 1.0921784122474495e-05, "loss": 0.1458763837814331, "step": 18470 }, { "epoch": 0.48679137183366644, "grad_norm": 3.7551612854003906, "learning_rate": 1.0913288737686789e-05, "loss": 0.15676584243774414, "step": 18480 }, { "epoch": 0.4870547870781652, "grad_norm": 2.0695769786834717, "learning_rate": 1.090479268816814e-05, "loss": 0.15045397281646727, "step": 18490 }, { "epoch": 0.48731820232266393, "grad_norm": 3.630159378051758, "learning_rate": 1.0896295980102334e-05, "loss": 0.15532593727111815, "step": 18500 }, { "epoch": 0.4875816175671627, "grad_norm": 1.2266324758529663, "learning_rate": 1.0887798619673648e-05, "loss": 0.13843460083007814, "step": 18510 }, { "epoch": 0.4878450328116614, "grad_norm": 0.8463799953460693, "learning_rate": 1.087930061306682e-05, "loss": 0.14807486534118652, "step": 18520 }, { "epoch": 0.4881084480561601, "grad_norm": 1.0933573246002197, "learning_rate": 1.0870801966467073e-05, "loss": 0.14626641273498536, "step": 18530 }, { "epoch": 0.48837186330065885, "grad_norm": 1.7691917419433594, "learning_rate": 1.0862302686060082e-05, "loss": 0.13612698316574096, "step": 18540 }, { "epoch": 0.4886352785451576, "grad_norm": 3.9748218059539795, "learning_rate": 1.0853802778031995e-05, "loss": 0.15033811330795288, "step": 18550 }, { "epoch": 0.48889869378965634, "grad_norm": 4.655544757843018, "learning_rate": 1.0845302248569411e-05, "loss": 0.14544403553009033, "step": 18560 }, { "epoch": 0.4891621090341551, "grad_norm": 3.16888165473938, "learning_rate": 1.0836801103859377e-05, "loss": 0.1433086633682251, "step": 18570 }, { "epoch": 0.4894255242786538, "grad_norm": 1.5390186309814453, "learning_rate": 1.0828299350089401e-05, "loss": 0.13862212896347045, "step": 18580 }, { "epoch": 0.48968893952315257, "grad_norm": 1.6056312322616577, "learning_rate": 1.0819796993447422e-05, "loss": 0.1408369779586792, "step": 18590 }, { "epoch": 0.4899523547676513, "grad_norm": 2.3824329376220703, "learning_rate": 1.081129404012182e-05, "loss": 0.13813529014587403, "step": 18600 }, { "epoch": 0.49021577001215005, "grad_norm": 2.3874411582946777, "learning_rate": 1.0802790496301414e-05, "loss": 0.1450487732887268, "step": 18610 }, { "epoch": 0.49047918525664874, "grad_norm": 4.02972936630249, "learning_rate": 1.079428636817545e-05, "loss": 0.14368932247161864, "step": 18620 }, { "epoch": 0.4907426005011475, "grad_norm": 0.9112387895584106, "learning_rate": 1.0785781661933596e-05, "loss": 0.14039552211761475, "step": 18630 }, { "epoch": 0.49100601574564623, "grad_norm": 1.163683533668518, "learning_rate": 1.0777276383765953e-05, "loss": 0.1401686668395996, "step": 18640 }, { "epoch": 0.491269430990145, "grad_norm": 2.9921579360961914, "learning_rate": 1.0768770539863017e-05, "loss": 0.15701106786727906, "step": 18650 }, { "epoch": 0.4915328462346437, "grad_norm": 0.9480744004249573, "learning_rate": 1.076026413641572e-05, "loss": 0.14472403526306152, "step": 18660 }, { "epoch": 0.49179626147914246, "grad_norm": 2.574169635772705, "learning_rate": 1.0751757179615378e-05, "loss": 0.15405720472335815, "step": 18670 }, { "epoch": 0.4920596767236412, "grad_norm": 3.5489704608917236, "learning_rate": 1.0743249675653733e-05, "loss": 0.14481441974639891, "step": 18680 }, { "epoch": 0.49232309196813995, "grad_norm": 1.774916410446167, "learning_rate": 1.0734741630722904e-05, "loss": 0.15206804275512695, "step": 18690 }, { "epoch": 0.49258650721263864, "grad_norm": 0.7406677007675171, "learning_rate": 1.0726233051015415e-05, "loss": 0.14030168056488038, "step": 18700 }, { "epoch": 0.4928499224571374, "grad_norm": 3.162558078765869, "learning_rate": 1.0717723942724182e-05, "loss": 0.13239411115646363, "step": 18710 }, { "epoch": 0.4931133377016361, "grad_norm": 2.7945303916931152, "learning_rate": 1.07092143120425e-05, "loss": 0.1367664337158203, "step": 18720 }, { "epoch": 0.49337675294613487, "grad_norm": 5.040863513946533, "learning_rate": 1.0700704165164042e-05, "loss": 0.1370453119277954, "step": 18730 }, { "epoch": 0.4936401681906336, "grad_norm": 3.0018107891082764, "learning_rate": 1.069219350828286e-05, "loss": 0.15621776580810548, "step": 18740 }, { "epoch": 0.49390358343513235, "grad_norm": 1.0343565940856934, "learning_rate": 1.068368234759338e-05, "loss": 0.1383526086807251, "step": 18750 }, { "epoch": 0.4941669986796311, "grad_norm": 1.666869044303894, "learning_rate": 1.0675170689290393e-05, "loss": 0.14720029830932618, "step": 18760 }, { "epoch": 0.49443041392412984, "grad_norm": 3.7313947677612305, "learning_rate": 1.0666658539569048e-05, "loss": 0.14441156387329102, "step": 18770 }, { "epoch": 0.4946938291686286, "grad_norm": 3.1963863372802734, "learning_rate": 1.0658145904624852e-05, "loss": 0.1596766471862793, "step": 18780 }, { "epoch": 0.4949572444131273, "grad_norm": 2.4291727542877197, "learning_rate": 1.0649632790653677e-05, "loss": 0.14320449829101561, "step": 18790 }, { "epoch": 0.495220659657626, "grad_norm": 0.9660411477088928, "learning_rate": 1.0641119203851724e-05, "loss": 0.15097883939743043, "step": 18800 }, { "epoch": 0.49548407490212476, "grad_norm": 1.692743182182312, "learning_rate": 1.0632605150415561e-05, "loss": 0.15515012741088868, "step": 18810 }, { "epoch": 0.4957474901466235, "grad_norm": 1.944303274154663, "learning_rate": 1.062409063654207e-05, "loss": 0.1541278123855591, "step": 18820 }, { "epoch": 0.49601090539112225, "grad_norm": 6.930377960205078, "learning_rate": 1.0615575668428494e-05, "loss": 0.1420016407966614, "step": 18830 }, { "epoch": 0.496274320635621, "grad_norm": 1.4707825183868408, "learning_rate": 1.0607060252272383e-05, "loss": 0.13906221389770507, "step": 18840 }, { "epoch": 0.49653773588011973, "grad_norm": 2.7428958415985107, "learning_rate": 1.0598544394271633e-05, "loss": 0.12578129768371582, "step": 18850 }, { "epoch": 0.4968011511246185, "grad_norm": 1.8248592615127563, "learning_rate": 1.0590028100624448e-05, "loss": 0.1468092679977417, "step": 18860 }, { "epoch": 0.4970645663691172, "grad_norm": 4.5354108810424805, "learning_rate": 1.0581511377529354e-05, "loss": 0.14954516887664795, "step": 18870 }, { "epoch": 0.4973279816136159, "grad_norm": 1.31220543384552, "learning_rate": 1.0572994231185194e-05, "loss": 0.15840787887573243, "step": 18880 }, { "epoch": 0.49759139685811465, "grad_norm": 1.3878289461135864, "learning_rate": 1.056447666779111e-05, "loss": 0.14459002017974854, "step": 18890 }, { "epoch": 0.4978548121026134, "grad_norm": 1.5772650241851807, "learning_rate": 1.0555958693546552e-05, "loss": 0.1396350622177124, "step": 18900 }, { "epoch": 0.49811822734711214, "grad_norm": 2.04256534576416, "learning_rate": 1.0547440314651274e-05, "loss": 0.13109123706817627, "step": 18910 }, { "epoch": 0.4983816425916109, "grad_norm": 2.4299051761627197, "learning_rate": 1.0538921537305315e-05, "loss": 0.1592763304710388, "step": 18920 }, { "epoch": 0.49864505783610963, "grad_norm": 3.2909131050109863, "learning_rate": 1.0530402367709007e-05, "loss": 0.14597971439361573, "step": 18930 }, { "epoch": 0.4989084730806084, "grad_norm": 3.0859320163726807, "learning_rate": 1.0521882812062978e-05, "loss": 0.1400775671005249, "step": 18940 }, { "epoch": 0.4991718883251071, "grad_norm": 1.6463959217071533, "learning_rate": 1.0513362876568118e-05, "loss": 0.14259912967681884, "step": 18950 }, { "epoch": 0.49943530356960586, "grad_norm": 3.6318020820617676, "learning_rate": 1.0504842567425611e-05, "loss": 0.14226444959640502, "step": 18960 }, { "epoch": 0.49969871881410455, "grad_norm": 1.8448221683502197, "learning_rate": 1.0496321890836901e-05, "loss": 0.143051815032959, "step": 18970 }, { "epoch": 0.4999621340586033, "grad_norm": 2.1617791652679443, "learning_rate": 1.0487800853003706e-05, "loss": 0.14899779558181764, "step": 18980 }, { "epoch": 0.500225549303102, "grad_norm": 4.530691623687744, "learning_rate": 1.0479279460128005e-05, "loss": 0.1476006031036377, "step": 18990 }, { "epoch": 0.5004889645476008, "grad_norm": 2.877302408218384, "learning_rate": 1.0470757718412035e-05, "loss": 0.13958754539489746, "step": 19000 }, { "epoch": 0.5007523797920995, "grad_norm": 1.8374227285385132, "learning_rate": 1.0462235634058287e-05, "loss": 0.15079221725463868, "step": 19010 }, { "epoch": 0.5010157950365982, "grad_norm": 1.439542531967163, "learning_rate": 1.0453713213269503e-05, "loss": 0.150976824760437, "step": 19020 }, { "epoch": 0.501279210281097, "grad_norm": 1.7746859788894653, "learning_rate": 1.0445190462248667e-05, "loss": 0.13989601135253907, "step": 19030 }, { "epoch": 0.5015426255255957, "grad_norm": 2.9650728702545166, "learning_rate": 1.0436667387199004e-05, "loss": 0.14730911254882811, "step": 19040 }, { "epoch": 0.5018060407700945, "grad_norm": 2.0239903926849365, "learning_rate": 1.0428143994323978e-05, "loss": 0.1490662932395935, "step": 19050 }, { "epoch": 0.5020694560145932, "grad_norm": 2.4599382877349854, "learning_rate": 1.041962028982728e-05, "loss": 0.1413433313369751, "step": 19060 }, { "epoch": 0.502332871259092, "grad_norm": 4.36340856552124, "learning_rate": 1.0411096279912835e-05, "loss": 0.15781066417694092, "step": 19070 }, { "epoch": 0.5025962865035907, "grad_norm": 1.829224944114685, "learning_rate": 1.0402571970784778e-05, "loss": 0.14360556602478028, "step": 19080 }, { "epoch": 0.5028597017480895, "grad_norm": 1.785922646522522, "learning_rate": 1.0394047368647473e-05, "loss": 0.13318568468093872, "step": 19090 }, { "epoch": 0.5031231169925882, "grad_norm": 2.330540895462036, "learning_rate": 1.0385522479705493e-05, "loss": 0.1388378143310547, "step": 19100 }, { "epoch": 0.5033865322370868, "grad_norm": 1.2405176162719727, "learning_rate": 1.0376997310163622e-05, "loss": 0.15570054054260254, "step": 19110 }, { "epoch": 0.5036499474815856, "grad_norm": 1.9298192262649536, "learning_rate": 1.0368471866226837e-05, "loss": 0.13578122854232788, "step": 19120 }, { "epoch": 0.5039133627260843, "grad_norm": 6.882496356964111, "learning_rate": 1.0359946154100336e-05, "loss": 0.1396587610244751, "step": 19130 }, { "epoch": 0.5041767779705831, "grad_norm": 1.9090723991394043, "learning_rate": 1.035142017998949e-05, "loss": 0.1449378490447998, "step": 19140 }, { "epoch": 0.5044401932150818, "grad_norm": 1.4806382656097412, "learning_rate": 1.0342893950099879e-05, "loss": 0.14414381980895996, "step": 19150 }, { "epoch": 0.5047036084595806, "grad_norm": 1.4933698177337646, "learning_rate": 1.0334367470637257e-05, "loss": 0.1431250810623169, "step": 19160 }, { "epoch": 0.5049670237040793, "grad_norm": 1.0945110321044922, "learning_rate": 1.0325840747807562e-05, "loss": 0.1430830717086792, "step": 19170 }, { "epoch": 0.505230438948578, "grad_norm": 2.0467112064361572, "learning_rate": 1.0317313787816915e-05, "loss": 0.15359052419662475, "step": 19180 }, { "epoch": 0.5054938541930768, "grad_norm": 1.9916324615478516, "learning_rate": 1.0308786596871606e-05, "loss": 0.1494274139404297, "step": 19190 }, { "epoch": 0.5057572694375755, "grad_norm": 2.265256881713867, "learning_rate": 1.0300259181178086e-05, "loss": 0.1426830530166626, "step": 19200 }, { "epoch": 0.5060206846820743, "grad_norm": 1.3190499544143677, "learning_rate": 1.0291731546942977e-05, "loss": 0.1341904878616333, "step": 19210 }, { "epoch": 0.506284099926573, "grad_norm": 2.0611183643341064, "learning_rate": 1.0283203700373065e-05, "loss": 0.13465754985809325, "step": 19220 }, { "epoch": 0.5065475151710718, "grad_norm": 1.063719391822815, "learning_rate": 1.0274675647675281e-05, "loss": 0.15168821811676025, "step": 19230 }, { "epoch": 0.5068109304155705, "grad_norm": 4.437199115753174, "learning_rate": 1.0266147395056712e-05, "loss": 0.14859936237335206, "step": 19240 }, { "epoch": 0.5070743456600693, "grad_norm": 2.891488552093506, "learning_rate": 1.0257618948724583e-05, "loss": 0.14451045989990235, "step": 19250 }, { "epoch": 0.507337760904568, "grad_norm": 2.4472646713256836, "learning_rate": 1.024909031488627e-05, "loss": 0.15330677032470702, "step": 19260 }, { "epoch": 0.5076011761490666, "grad_norm": 2.2949016094207764, "learning_rate": 1.0240561499749278e-05, "loss": 0.12943739891052247, "step": 19270 }, { "epoch": 0.5078645913935654, "grad_norm": 1.3652379512786865, "learning_rate": 1.023203250952125e-05, "loss": 0.14109336137771605, "step": 19280 }, { "epoch": 0.5081280066380641, "grad_norm": 2.1728549003601074, "learning_rate": 1.0223503350409949e-05, "loss": 0.13586180210113524, "step": 19290 }, { "epoch": 0.5083914218825629, "grad_norm": 0.9183894395828247, "learning_rate": 1.0214974028623268e-05, "loss": 0.15509185791015626, "step": 19300 }, { "epoch": 0.5086548371270616, "grad_norm": 2.4412894248962402, "learning_rate": 1.0206444550369212e-05, "loss": 0.15622735023498535, "step": 19310 }, { "epoch": 0.5089182523715604, "grad_norm": 1.8034844398498535, "learning_rate": 1.0197914921855909e-05, "loss": 0.14280904531478883, "step": 19320 }, { "epoch": 0.5091816676160591, "grad_norm": 2.5021297931671143, "learning_rate": 1.0189385149291586e-05, "loss": 0.14737656116485595, "step": 19330 }, { "epoch": 0.5094450828605579, "grad_norm": 1.2100032567977905, "learning_rate": 1.018085523888458e-05, "loss": 0.14166258573532103, "step": 19340 }, { "epoch": 0.5097084981050566, "grad_norm": 4.326999664306641, "learning_rate": 1.0172325196843328e-05, "loss": 0.14304933547973633, "step": 19350 }, { "epoch": 0.5099719133495553, "grad_norm": 2.1712071895599365, "learning_rate": 1.0163795029376362e-05, "loss": 0.13532955646514894, "step": 19360 }, { "epoch": 0.5102353285940541, "grad_norm": 1.942169427871704, "learning_rate": 1.015526474269231e-05, "loss": 0.14597718715667723, "step": 19370 }, { "epoch": 0.5104987438385528, "grad_norm": 3.969367027282715, "learning_rate": 1.0146734342999875e-05, "loss": 0.14305517673492432, "step": 19380 }, { "epoch": 0.5107621590830516, "grad_norm": 2.4457695484161377, "learning_rate": 1.0138203836507859e-05, "loss": 0.15396249294281006, "step": 19390 }, { "epoch": 0.5110255743275502, "grad_norm": 1.9056061506271362, "learning_rate": 1.0129673229425125e-05, "loss": 0.14798109531402587, "step": 19400 }, { "epoch": 0.511288989572049, "grad_norm": 3.062471389770508, "learning_rate": 1.0121142527960624e-05, "loss": 0.12861759662628175, "step": 19410 }, { "epoch": 0.5115524048165477, "grad_norm": 3.0674540996551514, "learning_rate": 1.0112611738323362e-05, "loss": 0.1364983558654785, "step": 19420 }, { "epoch": 0.5118158200610465, "grad_norm": 1.7413078546524048, "learning_rate": 1.0104080866722419e-05, "loss": 0.1280942440032959, "step": 19430 }, { "epoch": 0.5120792353055452, "grad_norm": 3.068617343902588, "learning_rate": 1.0095549919366928e-05, "loss": 0.13585844039916992, "step": 19440 }, { "epoch": 0.5123426505500439, "grad_norm": 1.8148273229599, "learning_rate": 1.0087018902466086e-05, "loss": 0.14900847673416137, "step": 19450 }, { "epoch": 0.5126060657945427, "grad_norm": 3.641519546508789, "learning_rate": 1.0078487822229132e-05, "loss": 0.15328660011291503, "step": 19460 }, { "epoch": 0.5128694810390414, "grad_norm": 1.178701639175415, "learning_rate": 1.0069956684865354e-05, "loss": 0.1474743127822876, "step": 19470 }, { "epoch": 0.5131328962835402, "grad_norm": 2.7857069969177246, "learning_rate": 1.0061425496584084e-05, "loss": 0.1395440459251404, "step": 19480 }, { "epoch": 0.5133963115280389, "grad_norm": 3.7836391925811768, "learning_rate": 1.0052894263594687e-05, "loss": 0.12500126361846925, "step": 19490 }, { "epoch": 0.5136597267725377, "grad_norm": 1.3032761812210083, "learning_rate": 1.0044362992106563e-05, "loss": 0.13729171752929686, "step": 19500 }, { "epoch": 0.5139231420170364, "grad_norm": 1.447482705116272, "learning_rate": 1.0035831688329139e-05, "loss": 0.1466564416885376, "step": 19510 }, { "epoch": 0.5141865572615352, "grad_norm": 2.371460199356079, "learning_rate": 1.0027300358471865e-05, "loss": 0.14258430004119874, "step": 19520 }, { "epoch": 0.5144499725060339, "grad_norm": 3.244363307952881, "learning_rate": 1.0018769008744214e-05, "loss": 0.15259206295013428, "step": 19530 }, { "epoch": 0.5147133877505325, "grad_norm": 1.161818027496338, "learning_rate": 1.0010237645355669e-05, "loss": 0.1414320945739746, "step": 19540 }, { "epoch": 0.5149768029950313, "grad_norm": 2.2434935569763184, "learning_rate": 1.0001706274515724e-05, "loss": 0.15284926891326905, "step": 19550 }, { "epoch": 0.51524021823953, "grad_norm": 3.1185154914855957, "learning_rate": 9.993174902433877e-06, "loss": 0.13693528175354003, "step": 19560 }, { "epoch": 0.5155036334840288, "grad_norm": 1.63671875, "learning_rate": 9.984643535319632e-06, "loss": 0.13399280309677125, "step": 19570 }, { "epoch": 0.5157670487285275, "grad_norm": 3.9208056926727295, "learning_rate": 9.976112179382483e-06, "loss": 0.1448366403579712, "step": 19580 }, { "epoch": 0.5160304639730263, "grad_norm": 1.5778132677078247, "learning_rate": 9.967580840831922e-06, "loss": 0.14918574094772338, "step": 19590 }, { "epoch": 0.516293879217525, "grad_norm": 1.7432140111923218, "learning_rate": 9.959049525877425e-06, "loss": 0.14858012199401854, "step": 19600 }, { "epoch": 0.5165572944620238, "grad_norm": 2.910454511642456, "learning_rate": 9.950518240728446e-06, "loss": 0.1485252022743225, "step": 19610 }, { "epoch": 0.5168207097065225, "grad_norm": 1.1572527885437012, "learning_rate": 9.94198699159443e-06, "loss": 0.14072619676589965, "step": 19620 }, { "epoch": 0.5170841249510212, "grad_norm": 1.9035941362380981, "learning_rate": 9.93345578468478e-06, "loss": 0.151932692527771, "step": 19630 }, { "epoch": 0.51734754019552, "grad_norm": 2.8937606811523438, "learning_rate": 9.924924626208884e-06, "loss": 0.1379567265510559, "step": 19640 }, { "epoch": 0.5176109554400187, "grad_norm": 1.1433424949645996, "learning_rate": 9.916393522376087e-06, "loss": 0.14346754550933838, "step": 19650 }, { "epoch": 0.5178743706845175, "grad_norm": 3.066723585128784, "learning_rate": 9.907862479395684e-06, "loss": 0.1433386445045471, "step": 19660 }, { "epoch": 0.5181377859290162, "grad_norm": 2.5192995071411133, "learning_rate": 9.899331503476951e-06, "loss": 0.1511707067489624, "step": 19670 }, { "epoch": 0.518401201173515, "grad_norm": 1.9752411842346191, "learning_rate": 9.890800600829086e-06, "loss": 0.13249289989471436, "step": 19680 }, { "epoch": 0.5186646164180136, "grad_norm": 3.5836427211761475, "learning_rate": 9.882269777661261e-06, "loss": 0.13308706283569335, "step": 19690 }, { "epoch": 0.5189280316625123, "grad_norm": 1.884109616279602, "learning_rate": 9.873739040182569e-06, "loss": 0.13505433797836303, "step": 19700 }, { "epoch": 0.5191914469070111, "grad_norm": 2.9205708503723145, "learning_rate": 9.86520839460205e-06, "loss": 0.1570219874382019, "step": 19710 }, { "epoch": 0.5194548621515098, "grad_norm": 3.708847999572754, "learning_rate": 9.856677847128678e-06, "loss": 0.13415827751159667, "step": 19720 }, { "epoch": 0.5197182773960086, "grad_norm": 1.9226890802383423, "learning_rate": 9.848147403971353e-06, "loss": 0.14094550609588624, "step": 19730 }, { "epoch": 0.5199816926405073, "grad_norm": 3.182772636413574, "learning_rate": 9.839617071338897e-06, "loss": 0.14321680068969728, "step": 19740 }, { "epoch": 0.5202451078850061, "grad_norm": 2.0991501808166504, "learning_rate": 9.831086855440056e-06, "loss": 0.15823917388916015, "step": 19750 }, { "epoch": 0.5205085231295048, "grad_norm": 1.838134765625, "learning_rate": 9.82255676248349e-06, "loss": 0.13090238571166993, "step": 19760 }, { "epoch": 0.5207719383740036, "grad_norm": 4.315074443817139, "learning_rate": 9.814026798677767e-06, "loss": 0.12775037288665772, "step": 19770 }, { "epoch": 0.5210353536185023, "grad_norm": 1.2956249713897705, "learning_rate": 9.805496970231367e-06, "loss": 0.13575408458709717, "step": 19780 }, { "epoch": 0.521298768863001, "grad_norm": 0.9616620540618896, "learning_rate": 9.796967283352657e-06, "loss": 0.12710577249526978, "step": 19790 }, { "epoch": 0.5215621841074998, "grad_norm": 1.6322020292282104, "learning_rate": 9.788437744249925e-06, "loss": 0.13878605365753174, "step": 19800 }, { "epoch": 0.5218255993519985, "grad_norm": 3.0845389366149902, "learning_rate": 9.779908359131327e-06, "loss": 0.13309013843536377, "step": 19810 }, { "epoch": 0.5220890145964973, "grad_norm": 2.040384292602539, "learning_rate": 9.77137913420492e-06, "loss": 0.13082793951034546, "step": 19820 }, { "epoch": 0.522352429840996, "grad_norm": 1.7820383310317993, "learning_rate": 9.762850075678644e-06, "loss": 0.15095739364624022, "step": 19830 }, { "epoch": 0.5226158450854947, "grad_norm": 3.8938162326812744, "learning_rate": 9.754321189760312e-06, "loss": 0.13851616382598878, "step": 19840 }, { "epoch": 0.5228792603299934, "grad_norm": 0.9731853604316711, "learning_rate": 9.74579248265762e-06, "loss": 0.13685617446899415, "step": 19850 }, { "epoch": 0.5231426755744922, "grad_norm": 1.0896964073181152, "learning_rate": 9.737263960578127e-06, "loss": 0.1365896701812744, "step": 19860 }, { "epoch": 0.5234060908189909, "grad_norm": 1.6294726133346558, "learning_rate": 9.728735629729256e-06, "loss": 0.14156389236450195, "step": 19870 }, { "epoch": 0.5236695060634896, "grad_norm": 2.3582801818847656, "learning_rate": 9.7202074963183e-06, "loss": 0.1458229660987854, "step": 19880 }, { "epoch": 0.5239329213079884, "grad_norm": 1.713363766670227, "learning_rate": 9.711679566552396e-06, "loss": 0.13623062372207642, "step": 19890 }, { "epoch": 0.5241963365524871, "grad_norm": 2.0497047901153564, "learning_rate": 9.703151846638545e-06, "loss": 0.14276248216629028, "step": 19900 }, { "epoch": 0.5244597517969859, "grad_norm": 1.515616774559021, "learning_rate": 9.694624342783591e-06, "loss": 0.14772706031799315, "step": 19910 }, { "epoch": 0.5247231670414846, "grad_norm": 2.0964062213897705, "learning_rate": 9.686097061194209e-06, "loss": 0.15493139028549194, "step": 19920 }, { "epoch": 0.5249865822859834, "grad_norm": 2.426100969314575, "learning_rate": 9.677570008076933e-06, "loss": 0.14313595294952391, "step": 19930 }, { "epoch": 0.5252499975304821, "grad_norm": 3.255297899246216, "learning_rate": 9.669043189638115e-06, "loss": 0.14481489658355712, "step": 19940 }, { "epoch": 0.5255134127749809, "grad_norm": 2.060208320617676, "learning_rate": 9.66051661208394e-06, "loss": 0.15363869667053223, "step": 19950 }, { "epoch": 0.5257768280194796, "grad_norm": 1.7070585489273071, "learning_rate": 9.651990281620423e-06, "loss": 0.1334742546081543, "step": 19960 }, { "epoch": 0.5260402432639782, "grad_norm": 5.527339458465576, "learning_rate": 9.643464204453392e-06, "loss": 0.13966741561889648, "step": 19970 }, { "epoch": 0.526303658508477, "grad_norm": 4.368741035461426, "learning_rate": 9.634938386788497e-06, "loss": 0.15704786777496338, "step": 19980 }, { "epoch": 0.5265670737529757, "grad_norm": 1.7991729974746704, "learning_rate": 9.626412834831193e-06, "loss": 0.1507539391517639, "step": 19990 }, { "epoch": 0.5268304889974745, "grad_norm": 3.5358710289001465, "learning_rate": 9.617887554786746e-06, "loss": 0.13025400638580323, "step": 20000 }, { "epoch": 0.5270939042419732, "grad_norm": 3.4207184314727783, "learning_rate": 9.609362552860225e-06, "loss": 0.13684492111206054, "step": 20010 }, { "epoch": 0.527357319486472, "grad_norm": 3.0097854137420654, "learning_rate": 9.600837835256486e-06, "loss": 0.1439067840576172, "step": 20020 }, { "epoch": 0.5276207347309707, "grad_norm": 2.737870216369629, "learning_rate": 9.592313408180199e-06, "loss": 0.12593917846679686, "step": 20030 }, { "epoch": 0.5278841499754695, "grad_norm": 6.492725849151611, "learning_rate": 9.583789277835798e-06, "loss": 0.15252790451049805, "step": 20040 }, { "epoch": 0.5281475652199682, "grad_norm": 2.1028079986572266, "learning_rate": 9.575265450427517e-06, "loss": 0.1465463638305664, "step": 20050 }, { "epoch": 0.5284109804644669, "grad_norm": 1.5958685874938965, "learning_rate": 9.566741932159369e-06, "loss": 0.1521294355392456, "step": 20060 }, { "epoch": 0.5286743957089657, "grad_norm": 1.744401454925537, "learning_rate": 9.558218729235132e-06, "loss": 0.1494441270828247, "step": 20070 }, { "epoch": 0.5289378109534644, "grad_norm": 4.307605266571045, "learning_rate": 9.549695847858363e-06, "loss": 0.14231044054031372, "step": 20080 }, { "epoch": 0.5292012261979632, "grad_norm": 2.3925182819366455, "learning_rate": 9.541173294232387e-06, "loss": 0.14872173070907593, "step": 20090 }, { "epoch": 0.5294646414424619, "grad_norm": 2.429720878601074, "learning_rate": 9.53265107456028e-06, "loss": 0.1426640510559082, "step": 20100 }, { "epoch": 0.5297280566869607, "grad_norm": 1.005024790763855, "learning_rate": 9.524129195044886e-06, "loss": 0.12891921997070313, "step": 20110 }, { "epoch": 0.5299914719314593, "grad_norm": 4.840083599090576, "learning_rate": 9.515607661888799e-06, "loss": 0.1452496886253357, "step": 20120 }, { "epoch": 0.5302548871759581, "grad_norm": 2.735975980758667, "learning_rate": 9.507086481294345e-06, "loss": 0.15587737560272216, "step": 20130 }, { "epoch": 0.5305183024204568, "grad_norm": 2.4257280826568604, "learning_rate": 9.498565659463624e-06, "loss": 0.13558619022369384, "step": 20140 }, { "epoch": 0.5307817176649555, "grad_norm": 1.525987982749939, "learning_rate": 9.490045202598442e-06, "loss": 0.13221316337585448, "step": 20150 }, { "epoch": 0.5310451329094543, "grad_norm": 1.4957202672958374, "learning_rate": 9.48152511690037e-06, "loss": 0.14250972270965576, "step": 20160 }, { "epoch": 0.531308548153953, "grad_norm": 3.312649726867676, "learning_rate": 9.473005408570685e-06, "loss": 0.1390053629875183, "step": 20170 }, { "epoch": 0.5315719633984518, "grad_norm": 1.931809425354004, "learning_rate": 9.464486083810397e-06, "loss": 0.13470146656036378, "step": 20180 }, { "epoch": 0.5318353786429505, "grad_norm": 2.305032253265381, "learning_rate": 9.455967148820243e-06, "loss": 0.14128929376602173, "step": 20190 }, { "epoch": 0.5320987938874493, "grad_norm": 4.207945823669434, "learning_rate": 9.44744860980067e-06, "loss": 0.14647326469421387, "step": 20200 }, { "epoch": 0.532362209131948, "grad_norm": 1.5441632270812988, "learning_rate": 9.438930472951838e-06, "loss": 0.12561535835266113, "step": 20210 }, { "epoch": 0.5326256243764467, "grad_norm": 2.1747395992279053, "learning_rate": 9.430412744473615e-06, "loss": 0.1300243616104126, "step": 20220 }, { "epoch": 0.5328890396209455, "grad_norm": 2.4596188068389893, "learning_rate": 9.421895430565568e-06, "loss": 0.13951549530029297, "step": 20230 }, { "epoch": 0.5331524548654442, "grad_norm": 2.6195950508117676, "learning_rate": 9.413378537426972e-06, "loss": 0.13880953788757325, "step": 20240 }, { "epoch": 0.533415870109943, "grad_norm": 1.567142367362976, "learning_rate": 9.404862071256785e-06, "loss": 0.13469678163528442, "step": 20250 }, { "epoch": 0.5336792853544416, "grad_norm": 2.4696285724639893, "learning_rate": 9.396346038253654e-06, "loss": 0.15341503620147706, "step": 20260 }, { "epoch": 0.5339427005989404, "grad_norm": 4.067445755004883, "learning_rate": 9.387830444615926e-06, "loss": 0.13147879838943483, "step": 20270 }, { "epoch": 0.5342061158434391, "grad_norm": 6.151753902435303, "learning_rate": 9.379315296541611e-06, "loss": 0.1341244697570801, "step": 20280 }, { "epoch": 0.5344695310879379, "grad_norm": 1.6839271783828735, "learning_rate": 9.370800600228399e-06, "loss": 0.14775084257125853, "step": 20290 }, { "epoch": 0.5347329463324366, "grad_norm": 0.9472609758377075, "learning_rate": 9.362286361873658e-06, "loss": 0.144197678565979, "step": 20300 }, { "epoch": 0.5349963615769353, "grad_norm": 2.7751684188842773, "learning_rate": 9.353772587674413e-06, "loss": 0.14369821548461914, "step": 20310 }, { "epoch": 0.5352597768214341, "grad_norm": 2.0206878185272217, "learning_rate": 9.345259283827362e-06, "loss": 0.1320814847946167, "step": 20320 }, { "epoch": 0.5355231920659328, "grad_norm": 2.85801362991333, "learning_rate": 9.336746456528853e-06, "loss": 0.13794963359832763, "step": 20330 }, { "epoch": 0.5357866073104316, "grad_norm": 1.694788932800293, "learning_rate": 9.328234111974885e-06, "loss": 0.1308119297027588, "step": 20340 }, { "epoch": 0.5360500225549303, "grad_norm": 1.277082920074463, "learning_rate": 9.319722256361117e-06, "loss": 0.1424214243888855, "step": 20350 }, { "epoch": 0.5363134377994291, "grad_norm": 5.7032318115234375, "learning_rate": 9.311210895882835e-06, "loss": 0.1433882713317871, "step": 20360 }, { "epoch": 0.5365768530439278, "grad_norm": 2.4404399394989014, "learning_rate": 9.302700036734985e-06, "loss": 0.1334259033203125, "step": 20370 }, { "epoch": 0.5368402682884266, "grad_norm": 2.2041234970092773, "learning_rate": 9.294189685112133e-06, "loss": 0.14298372268676757, "step": 20380 }, { "epoch": 0.5371036835329253, "grad_norm": 2.2140085697174072, "learning_rate": 9.285679847208474e-06, "loss": 0.12571728229522705, "step": 20390 }, { "epoch": 0.537367098777424, "grad_norm": 1.346773386001587, "learning_rate": 9.277170529217847e-06, "loss": 0.14446187019348145, "step": 20400 }, { "epoch": 0.5376305140219227, "grad_norm": 0.6893863081932068, "learning_rate": 9.268661737333691e-06, "loss": 0.13953860998153686, "step": 20410 }, { "epoch": 0.5378939292664214, "grad_norm": 1.7103896141052246, "learning_rate": 9.260153477749072e-06, "loss": 0.14413391351699828, "step": 20420 }, { "epoch": 0.5381573445109202, "grad_norm": 2.5010430812835693, "learning_rate": 9.251645756656676e-06, "loss": 0.14595623016357423, "step": 20430 }, { "epoch": 0.5384207597554189, "grad_norm": 2.2957675457000732, "learning_rate": 9.243138580248782e-06, "loss": 0.1324528694152832, "step": 20440 }, { "epoch": 0.5386841749999177, "grad_norm": 4.654347896575928, "learning_rate": 9.234631954717285e-06, "loss": 0.13996150493621826, "step": 20450 }, { "epoch": 0.5389475902444164, "grad_norm": 2.207487106323242, "learning_rate": 9.226125886253673e-06, "loss": 0.14405362606048583, "step": 20460 }, { "epoch": 0.5392110054889152, "grad_norm": 1.219172477722168, "learning_rate": 9.217620381049026e-06, "loss": 0.14160555601119995, "step": 20470 }, { "epoch": 0.5394744207334139, "grad_norm": 1.1816847324371338, "learning_rate": 9.209115445294022e-06, "loss": 0.12993841171264647, "step": 20480 }, { "epoch": 0.5397378359779126, "grad_norm": 1.310904860496521, "learning_rate": 9.200611085178917e-06, "loss": 0.1336743116378784, "step": 20490 }, { "epoch": 0.5400012512224114, "grad_norm": 2.5935375690460205, "learning_rate": 9.192107306893556e-06, "loss": 0.13781027793884276, "step": 20500 }, { "epoch": 0.5402646664669101, "grad_norm": 4.917055130004883, "learning_rate": 9.183604116627354e-06, "loss": 0.14664100408554076, "step": 20510 }, { "epoch": 0.5405280817114089, "grad_norm": 1.6418333053588867, "learning_rate": 9.175101520569293e-06, "loss": 0.14214735031127929, "step": 20520 }, { "epoch": 0.5407914969559076, "grad_norm": 2.504624366760254, "learning_rate": 9.166599524907937e-06, "loss": 0.13684476613998414, "step": 20530 }, { "epoch": 0.5410549122004064, "grad_norm": 2.8817665576934814, "learning_rate": 9.158098135831404e-06, "loss": 0.1427687644958496, "step": 20540 }, { "epoch": 0.541318327444905, "grad_norm": 1.7038347721099854, "learning_rate": 9.149597359527368e-06, "loss": 0.13542078733444213, "step": 20550 }, { "epoch": 0.5415817426894038, "grad_norm": 1.6327331066131592, "learning_rate": 9.141097202183063e-06, "loss": 0.13803386688232422, "step": 20560 }, { "epoch": 0.5418451579339025, "grad_norm": 1.779976487159729, "learning_rate": 9.13259766998527e-06, "loss": 0.13511860370635986, "step": 20570 }, { "epoch": 0.5421085731784012, "grad_norm": 3.4013803005218506, "learning_rate": 9.124098769120318e-06, "loss": 0.13688390254974364, "step": 20580 }, { "epoch": 0.5423719884229, "grad_norm": 1.871234655380249, "learning_rate": 9.115600505774072e-06, "loss": 0.14776952266693116, "step": 20590 }, { "epoch": 0.5426354036673987, "grad_norm": 1.7054516077041626, "learning_rate": 9.10710288613193e-06, "loss": 0.14255952835083008, "step": 20600 }, { "epoch": 0.5428988189118975, "grad_norm": 1.6279609203338623, "learning_rate": 9.098605916378837e-06, "loss": 0.13723148107528688, "step": 20610 }, { "epoch": 0.5431622341563962, "grad_norm": 3.079514741897583, "learning_rate": 9.090109602699239e-06, "loss": 0.13253659009933472, "step": 20620 }, { "epoch": 0.543425649400895, "grad_norm": 3.75347638130188, "learning_rate": 9.081613951277135e-06, "loss": 0.13552671670913696, "step": 20630 }, { "epoch": 0.5436890646453937, "grad_norm": 1.437095284461975, "learning_rate": 9.07311896829602e-06, "loss": 0.13920776844024657, "step": 20640 }, { "epoch": 0.5439524798898925, "grad_norm": 1.2724261283874512, "learning_rate": 9.064624659938902e-06, "loss": 0.12410078048706055, "step": 20650 }, { "epoch": 0.5442158951343912, "grad_norm": 2.1637656688690186, "learning_rate": 9.056131032388318e-06, "loss": 0.1435115933418274, "step": 20660 }, { "epoch": 0.5444793103788899, "grad_norm": 3.802889585494995, "learning_rate": 9.047638091826291e-06, "loss": 0.1384589910507202, "step": 20670 }, { "epoch": 0.5447427256233887, "grad_norm": 1.2388966083526611, "learning_rate": 9.039145844434344e-06, "loss": 0.146861469745636, "step": 20680 }, { "epoch": 0.5450061408678873, "grad_norm": 2.6652705669403076, "learning_rate": 9.030654296393512e-06, "loss": 0.1333676815032959, "step": 20690 }, { "epoch": 0.5452695561123861, "grad_norm": 1.7313876152038574, "learning_rate": 9.022163453884299e-06, "loss": 0.13155664205551149, "step": 20700 }, { "epoch": 0.5455329713568848, "grad_norm": 1.5233296155929565, "learning_rate": 9.013673323086716e-06, "loss": 0.1379331588745117, "step": 20710 }, { "epoch": 0.5457963866013836, "grad_norm": 3.0633530616760254, "learning_rate": 9.005183910180243e-06, "loss": 0.13422744274139403, "step": 20720 }, { "epoch": 0.5460598018458823, "grad_norm": 2.701253890991211, "learning_rate": 8.996695221343834e-06, "loss": 0.14021923542022705, "step": 20730 }, { "epoch": 0.546323217090381, "grad_norm": 2.39233136177063, "learning_rate": 8.988207262755936e-06, "loss": 0.13506557941436767, "step": 20740 }, { "epoch": 0.5465866323348798, "grad_norm": 1.5222059488296509, "learning_rate": 8.979720040594438e-06, "loss": 0.14977662563323973, "step": 20750 }, { "epoch": 0.5468500475793785, "grad_norm": 2.2700679302215576, "learning_rate": 8.971233561036718e-06, "loss": 0.13245413303375245, "step": 20760 }, { "epoch": 0.5471134628238773, "grad_norm": 1.7118301391601562, "learning_rate": 8.962747830259598e-06, "loss": 0.13001879453659057, "step": 20770 }, { "epoch": 0.547376878068376, "grad_norm": 2.47261118888855, "learning_rate": 8.954262854439354e-06, "loss": 0.14689772129058837, "step": 20780 }, { "epoch": 0.5476402933128748, "grad_norm": 2.2229983806610107, "learning_rate": 8.945778639751723e-06, "loss": 0.14228541851043702, "step": 20790 }, { "epoch": 0.5479037085573735, "grad_norm": 3.9080893993377686, "learning_rate": 8.937295192371882e-06, "loss": 0.1301426410675049, "step": 20800 }, { "epoch": 0.5481671238018723, "grad_norm": 1.5563583374023438, "learning_rate": 8.928812518474445e-06, "loss": 0.11943185329437256, "step": 20810 }, { "epoch": 0.548430539046371, "grad_norm": 1.4850322008132935, "learning_rate": 8.920330624233477e-06, "loss": 0.12431204319000244, "step": 20820 }, { "epoch": 0.5486939542908696, "grad_norm": 2.4076828956604004, "learning_rate": 8.911849515822457e-06, "loss": 0.13916056156158446, "step": 20830 }, { "epoch": 0.5489573695353684, "grad_norm": 1.7373899221420288, "learning_rate": 8.903369199414311e-06, "loss": 0.14403114318847657, "step": 20840 }, { "epoch": 0.5492207847798671, "grad_norm": 1.964308500289917, "learning_rate": 8.894889681181374e-06, "loss": 0.13852062225341796, "step": 20850 }, { "epoch": 0.5494842000243659, "grad_norm": 2.695859670639038, "learning_rate": 8.886410967295401e-06, "loss": 0.14522913694381714, "step": 20860 }, { "epoch": 0.5497476152688646, "grad_norm": 1.5462347269058228, "learning_rate": 8.877933063927576e-06, "loss": 0.12990777492523192, "step": 20870 }, { "epoch": 0.5500110305133634, "grad_norm": 1.7083221673965454, "learning_rate": 8.869455977248473e-06, "loss": 0.13523728847503663, "step": 20880 }, { "epoch": 0.5502744457578621, "grad_norm": 3.8993594646453857, "learning_rate": 8.860979713428085e-06, "loss": 0.12209963798522949, "step": 20890 }, { "epoch": 0.5505378610023609, "grad_norm": 2.6215360164642334, "learning_rate": 8.852504278635806e-06, "loss": 0.13721753358840943, "step": 20900 }, { "epoch": 0.5508012762468596, "grad_norm": 3.71848201751709, "learning_rate": 8.844029679040418e-06, "loss": 0.13244296312332154, "step": 20910 }, { "epoch": 0.5510646914913583, "grad_norm": 2.349755048751831, "learning_rate": 8.835555920810104e-06, "loss": 0.13381270170211793, "step": 20920 }, { "epoch": 0.5513281067358571, "grad_norm": 1.7520688772201538, "learning_rate": 8.827083010112429e-06, "loss": 0.12486550807952881, "step": 20930 }, { "epoch": 0.5515915219803558, "grad_norm": 1.4994884729385376, "learning_rate": 8.818610953114342e-06, "loss": 0.12256667613983155, "step": 20940 }, { "epoch": 0.5518549372248546, "grad_norm": 1.634819507598877, "learning_rate": 8.810139755982173e-06, "loss": 0.12724267244338988, "step": 20950 }, { "epoch": 0.5521183524693533, "grad_norm": 1.9570564031600952, "learning_rate": 8.801669424881624e-06, "loss": 0.1397400140762329, "step": 20960 }, { "epoch": 0.5523817677138521, "grad_norm": 1.8105111122131348, "learning_rate": 8.79319996597777e-06, "loss": 0.12985800504684447, "step": 20970 }, { "epoch": 0.5526451829583507, "grad_norm": 3.690967321395874, "learning_rate": 8.784731385435049e-06, "loss": 0.12930243015289306, "step": 20980 }, { "epoch": 0.5529085982028495, "grad_norm": 0.9009004235267639, "learning_rate": 8.77626368941725e-06, "loss": 0.12512786388397218, "step": 20990 }, { "epoch": 0.5531720134473482, "grad_norm": 1.9260778427124023, "learning_rate": 8.767796884087539e-06, "loss": 0.13574541807174684, "step": 21000 }, { "epoch": 0.5534354286918469, "grad_norm": 2.7627618312835693, "learning_rate": 8.759330975608418e-06, "loss": 0.13887290954589843, "step": 21010 }, { "epoch": 0.5536988439363457, "grad_norm": 1.4234466552734375, "learning_rate": 8.750865970141736e-06, "loss": 0.13324692249298095, "step": 21020 }, { "epoch": 0.5539622591808444, "grad_norm": 1.7505556344985962, "learning_rate": 8.742401873848696e-06, "loss": 0.13882414102554322, "step": 21030 }, { "epoch": 0.5542256744253432, "grad_norm": 2.204536199569702, "learning_rate": 8.733938692889827e-06, "loss": 0.13670998811721802, "step": 21040 }, { "epoch": 0.5544890896698419, "grad_norm": 1.2205959558486938, "learning_rate": 8.725476433425003e-06, "loss": 0.13541903495788574, "step": 21050 }, { "epoch": 0.5547525049143407, "grad_norm": 2.996575117111206, "learning_rate": 8.717015101613417e-06, "loss": 0.144240665435791, "step": 21060 }, { "epoch": 0.5550159201588394, "grad_norm": 1.8576351404190063, "learning_rate": 8.708554703613592e-06, "loss": 0.13405399322509765, "step": 21070 }, { "epoch": 0.5552793354033382, "grad_norm": 1.8189034461975098, "learning_rate": 8.700095245583374e-06, "loss": 0.121501886844635, "step": 21080 }, { "epoch": 0.5555427506478369, "grad_norm": 1.5240919589996338, "learning_rate": 8.691636733679915e-06, "loss": 0.12885910272598267, "step": 21090 }, { "epoch": 0.5558061658923356, "grad_norm": 1.6599313020706177, "learning_rate": 8.683179174059691e-06, "loss": 0.1432579517364502, "step": 21100 }, { "epoch": 0.5560695811368344, "grad_norm": 1.598676323890686, "learning_rate": 8.67472257287848e-06, "loss": 0.13572684526443482, "step": 21110 }, { "epoch": 0.556332996381333, "grad_norm": 3.569300651550293, "learning_rate": 8.666266936291354e-06, "loss": 0.13952603340148925, "step": 21120 }, { "epoch": 0.5565964116258318, "grad_norm": 2.9468841552734375, "learning_rate": 8.657812270452695e-06, "loss": 0.15000243186950685, "step": 21130 }, { "epoch": 0.5568598268703305, "grad_norm": 3.762007236480713, "learning_rate": 8.649358581516171e-06, "loss": 0.1346367120742798, "step": 21140 }, { "epoch": 0.5571232421148293, "grad_norm": 1.491865634918213, "learning_rate": 8.640905875634744e-06, "loss": 0.13483109474182128, "step": 21150 }, { "epoch": 0.557386657359328, "grad_norm": 2.3163275718688965, "learning_rate": 8.632454158960656e-06, "loss": 0.1315129280090332, "step": 21160 }, { "epoch": 0.5576500726038268, "grad_norm": 2.482107400894165, "learning_rate": 8.624003437645431e-06, "loss": 0.1261651635169983, "step": 21170 }, { "epoch": 0.5579134878483255, "grad_norm": 4.217422962188721, "learning_rate": 8.615553717839873e-06, "loss": 0.1268307685852051, "step": 21180 }, { "epoch": 0.5581769030928242, "grad_norm": 2.7843425273895264, "learning_rate": 8.607105005694048e-06, "loss": 0.14523530006408691, "step": 21190 }, { "epoch": 0.558440318337323, "grad_norm": 2.2467827796936035, "learning_rate": 8.598657307357293e-06, "loss": 0.1436774969100952, "step": 21200 }, { "epoch": 0.5587037335818217, "grad_norm": 1.3802474737167358, "learning_rate": 8.590210628978213e-06, "loss": 0.13431090116500854, "step": 21210 }, { "epoch": 0.5589671488263205, "grad_norm": 3.460470676422119, "learning_rate": 8.581764976704653e-06, "loss": 0.13997108936309816, "step": 21220 }, { "epoch": 0.5592305640708192, "grad_norm": 1.7431894540786743, "learning_rate": 8.573320356683739e-06, "loss": 0.13763797283172607, "step": 21230 }, { "epoch": 0.559493979315318, "grad_norm": 1.275162935256958, "learning_rate": 8.564876775061819e-06, "loss": 0.12579375505447388, "step": 21240 }, { "epoch": 0.5597573945598167, "grad_norm": 1.6554234027862549, "learning_rate": 8.556434237984496e-06, "loss": 0.11662452220916748, "step": 21250 }, { "epoch": 0.5600208098043153, "grad_norm": 3.2810404300689697, "learning_rate": 8.547992751596617e-06, "loss": 0.14599788188934326, "step": 21260 }, { "epoch": 0.5602842250488141, "grad_norm": 2.6898770332336426, "learning_rate": 8.53955232204226e-06, "loss": 0.14062852859497071, "step": 21270 }, { "epoch": 0.5605476402933128, "grad_norm": 2.6010029315948486, "learning_rate": 8.531112955464728e-06, "loss": 0.1371389150619507, "step": 21280 }, { "epoch": 0.5608110555378116, "grad_norm": 1.0112862586975098, "learning_rate": 8.522674658006564e-06, "loss": 0.12638274431228638, "step": 21290 }, { "epoch": 0.5610744707823103, "grad_norm": 2.570094585418701, "learning_rate": 8.514237435809517e-06, "loss": 0.12863929271698, "step": 21300 }, { "epoch": 0.5613378860268091, "grad_norm": 1.0241663455963135, "learning_rate": 8.505801295014569e-06, "loss": 0.12450752258300782, "step": 21310 }, { "epoch": 0.5616013012713078, "grad_norm": 2.9933981895446777, "learning_rate": 8.497366241761907e-06, "loss": 0.1356739282608032, "step": 21320 }, { "epoch": 0.5618647165158066, "grad_norm": 1.0668023824691772, "learning_rate": 8.488932282190914e-06, "loss": 0.13528215885162354, "step": 21330 }, { "epoch": 0.5621281317603053, "grad_norm": 0.9224807024002075, "learning_rate": 8.480499422440207e-06, "loss": 0.12589950561523439, "step": 21340 }, { "epoch": 0.562391547004804, "grad_norm": 3.3678581714630127, "learning_rate": 8.472067668647571e-06, "loss": 0.1328420639038086, "step": 21350 }, { "epoch": 0.5626549622493028, "grad_norm": 2.1752207279205322, "learning_rate": 8.463637026950013e-06, "loss": 0.1361228346824646, "step": 21360 }, { "epoch": 0.5629183774938015, "grad_norm": 1.982703685760498, "learning_rate": 8.455207503483708e-06, "loss": 0.13075792789459229, "step": 21370 }, { "epoch": 0.5631817927383003, "grad_norm": 5.0988850593566895, "learning_rate": 8.446779104384028e-06, "loss": 0.13817754983901978, "step": 21380 }, { "epoch": 0.563445207982799, "grad_norm": 2.49631667137146, "learning_rate": 8.43835183578553e-06, "loss": 0.1291806697845459, "step": 21390 }, { "epoch": 0.5637086232272978, "grad_norm": 2.068906307220459, "learning_rate": 8.429925703821941e-06, "loss": 0.14527509212493897, "step": 21400 }, { "epoch": 0.5639720384717964, "grad_norm": 1.717454195022583, "learning_rate": 8.421500714626164e-06, "loss": 0.13891602754592897, "step": 21410 }, { "epoch": 0.5642354537162952, "grad_norm": 3.264132499694824, "learning_rate": 8.413076874330266e-06, "loss": 0.1421574354171753, "step": 21420 }, { "epoch": 0.5644988689607939, "grad_norm": 1.7208442687988281, "learning_rate": 8.404654189065486e-06, "loss": 0.1311122417449951, "step": 21430 }, { "epoch": 0.5647622842052926, "grad_norm": 1.781158685684204, "learning_rate": 8.396232664962218e-06, "loss": 0.15053308010101318, "step": 21440 }, { "epoch": 0.5650256994497914, "grad_norm": 0.9758434295654297, "learning_rate": 8.387812308150011e-06, "loss": 0.11429564952850342, "step": 21450 }, { "epoch": 0.5652891146942901, "grad_norm": 1.696702480316162, "learning_rate": 8.379393124757554e-06, "loss": 0.14811789989471436, "step": 21460 }, { "epoch": 0.5655525299387889, "grad_norm": 1.7230501174926758, "learning_rate": 8.370975120912708e-06, "loss": 0.13873705863952637, "step": 21470 }, { "epoch": 0.5658159451832876, "grad_norm": 4.2364983558654785, "learning_rate": 8.362558302742447e-06, "loss": 0.12262637615203857, "step": 21480 }, { "epoch": 0.5660793604277864, "grad_norm": 4.2179484367370605, "learning_rate": 8.354142676372898e-06, "loss": 0.13950045108795167, "step": 21490 }, { "epoch": 0.5663427756722851, "grad_norm": 1.0744121074676514, "learning_rate": 8.345728247929319e-06, "loss": 0.1270014762878418, "step": 21500 }, { "epoch": 0.5666061909167839, "grad_norm": 2.6036217212677, "learning_rate": 8.33731502353609e-06, "loss": 0.12983253002166747, "step": 21510 }, { "epoch": 0.5668696061612826, "grad_norm": 3.6810662746429443, "learning_rate": 8.328903009316724e-06, "loss": 0.1572199583053589, "step": 21520 }, { "epoch": 0.5671330214057813, "grad_norm": 2.1212527751922607, "learning_rate": 8.320492211393844e-06, "loss": 0.1307546615600586, "step": 21530 }, { "epoch": 0.5673964366502801, "grad_norm": 4.20723295211792, "learning_rate": 8.312082635889193e-06, "loss": 0.14129332304000855, "step": 21540 }, { "epoch": 0.5676598518947787, "grad_norm": 1.5558438301086426, "learning_rate": 8.303674288923621e-06, "loss": 0.14019598960876464, "step": 21550 }, { "epoch": 0.5679232671392775, "grad_norm": 2.5294575691223145, "learning_rate": 8.29526717661709e-06, "loss": 0.1380619764328003, "step": 21560 }, { "epoch": 0.5681866823837762, "grad_norm": 3.475032091140747, "learning_rate": 8.286861305088655e-06, "loss": 0.13213907480239867, "step": 21570 }, { "epoch": 0.568450097628275, "grad_norm": 1.4168968200683594, "learning_rate": 8.278456680456476e-06, "loss": 0.13359609842300416, "step": 21580 }, { "epoch": 0.5687135128727737, "grad_norm": 1.3734369277954102, "learning_rate": 8.270053308837791e-06, "loss": 0.1305910110473633, "step": 21590 }, { "epoch": 0.5689769281172725, "grad_norm": 1.988825798034668, "learning_rate": 8.261651196348953e-06, "loss": 0.12698397636413575, "step": 21600 }, { "epoch": 0.5692403433617712, "grad_norm": 2.9962759017944336, "learning_rate": 8.253250349105369e-06, "loss": 0.13172144889831544, "step": 21610 }, { "epoch": 0.5695037586062699, "grad_norm": 1.6765131950378418, "learning_rate": 8.24485077322154e-06, "loss": 0.1268153190612793, "step": 21620 }, { "epoch": 0.5697671738507687, "grad_norm": 1.3748316764831543, "learning_rate": 8.236452474811046e-06, "loss": 0.13660798072814942, "step": 21630 }, { "epoch": 0.5700305890952674, "grad_norm": 1.6525979042053223, "learning_rate": 8.228055459986521e-06, "loss": 0.132004714012146, "step": 21640 }, { "epoch": 0.5702940043397662, "grad_norm": 2.5119376182556152, "learning_rate": 8.219659734859686e-06, "loss": 0.12212507724761963, "step": 21650 }, { "epoch": 0.5705574195842649, "grad_norm": 1.948258399963379, "learning_rate": 8.211265305541304e-06, "loss": 0.13806021213531494, "step": 21660 }, { "epoch": 0.5708208348287637, "grad_norm": 2.9654831886291504, "learning_rate": 8.202872178141204e-06, "loss": 0.14145060777664184, "step": 21670 }, { "epoch": 0.5710842500732624, "grad_norm": 6.976021766662598, "learning_rate": 8.19448035876827e-06, "loss": 0.1243404507637024, "step": 21680 }, { "epoch": 0.5713476653177612, "grad_norm": 2.31480073928833, "learning_rate": 8.186089853530426e-06, "loss": 0.1317291021347046, "step": 21690 }, { "epoch": 0.5716110805622598, "grad_norm": 5.333384037017822, "learning_rate": 8.177700668534646e-06, "loss": 0.13723477125167846, "step": 21700 }, { "epoch": 0.5718744958067585, "grad_norm": 1.3732331991195679, "learning_rate": 8.169312809886944e-06, "loss": 0.13107869625091553, "step": 21710 }, { "epoch": 0.5721379110512573, "grad_norm": 1.6664506196975708, "learning_rate": 8.160926283692355e-06, "loss": 0.13021547794342042, "step": 21720 }, { "epoch": 0.572401326295756, "grad_norm": 2.0196831226348877, "learning_rate": 8.152541096054965e-06, "loss": 0.12055550813674927, "step": 21730 }, { "epoch": 0.5726647415402548, "grad_norm": 1.6117527484893799, "learning_rate": 8.144157253077871e-06, "loss": 0.1408841133117676, "step": 21740 }, { "epoch": 0.5729281567847535, "grad_norm": 1.5789297819137573, "learning_rate": 8.135774760863193e-06, "loss": 0.14153783321380614, "step": 21750 }, { "epoch": 0.5731915720292523, "grad_norm": 1.3914684057235718, "learning_rate": 8.127393625512077e-06, "loss": 0.13274455070495605, "step": 21760 }, { "epoch": 0.573454987273751, "grad_norm": 2.294029951095581, "learning_rate": 8.11901385312467e-06, "loss": 0.13297765254974364, "step": 21770 }, { "epoch": 0.5737184025182498, "grad_norm": 1.5755064487457275, "learning_rate": 8.110635449800132e-06, "loss": 0.13997561931610109, "step": 21780 }, { "epoch": 0.5739818177627485, "grad_norm": 2.263707399368286, "learning_rate": 8.10225842163663e-06, "loss": 0.14283397197723388, "step": 21790 }, { "epoch": 0.5742452330072472, "grad_norm": 2.038067102432251, "learning_rate": 8.093882774731322e-06, "loss": 0.14396408796310425, "step": 21800 }, { "epoch": 0.574508648251746, "grad_norm": 3.1226837635040283, "learning_rate": 8.085508515180373e-06, "loss": 0.13280773162841797, "step": 21810 }, { "epoch": 0.5747720634962447, "grad_norm": 2.103642225265503, "learning_rate": 8.077135649078917e-06, "loss": 0.12823407649993895, "step": 21820 }, { "epoch": 0.5750354787407435, "grad_norm": 3.072145462036133, "learning_rate": 8.068764182521102e-06, "loss": 0.14114899635314943, "step": 21830 }, { "epoch": 0.5752988939852421, "grad_norm": 2.726248264312744, "learning_rate": 8.060394121600033e-06, "loss": 0.1266741394996643, "step": 21840 }, { "epoch": 0.5755623092297409, "grad_norm": 1.3653043508529663, "learning_rate": 8.052025472407802e-06, "loss": 0.13029171228408815, "step": 21850 }, { "epoch": 0.5758257244742396, "grad_norm": 1.573460340499878, "learning_rate": 8.043658241035477e-06, "loss": 0.14384572505950927, "step": 21860 }, { "epoch": 0.5760891397187383, "grad_norm": 2.0971953868865967, "learning_rate": 8.035292433573088e-06, "loss": 0.12660635709762574, "step": 21870 }, { "epoch": 0.5763525549632371, "grad_norm": 2.22607421875, "learning_rate": 8.026928056109628e-06, "loss": 0.12761958837509155, "step": 21880 }, { "epoch": 0.5766159702077358, "grad_norm": 2.656078338623047, "learning_rate": 8.018565114733056e-06, "loss": 0.14641647338867186, "step": 21890 }, { "epoch": 0.5768793854522346, "grad_norm": 1.369968056678772, "learning_rate": 8.010203615530279e-06, "loss": 0.1302046775817871, "step": 21900 }, { "epoch": 0.5771428006967333, "grad_norm": 2.9192147254943848, "learning_rate": 8.00184356458716e-06, "loss": 0.13446836471557616, "step": 21910 }, { "epoch": 0.5774062159412321, "grad_norm": 2.417100429534912, "learning_rate": 7.993484967988504e-06, "loss": 0.13434629440307616, "step": 21920 }, { "epoch": 0.5776696311857308, "grad_norm": 2.229515314102173, "learning_rate": 7.98512783181805e-06, "loss": 0.11976934671401977, "step": 21930 }, { "epoch": 0.5779330464302296, "grad_norm": 1.1558529138565063, "learning_rate": 7.976772162158498e-06, "loss": 0.1461641550064087, "step": 21940 }, { "epoch": 0.5781964616747283, "grad_norm": 1.5080711841583252, "learning_rate": 7.968417965091449e-06, "loss": 0.12300753593444824, "step": 21950 }, { "epoch": 0.578459876919227, "grad_norm": 4.257937431335449, "learning_rate": 7.960065246697456e-06, "loss": 0.13064727783203126, "step": 21960 }, { "epoch": 0.5787232921637258, "grad_norm": 1.554740071296692, "learning_rate": 7.951714013055988e-06, "loss": 0.1447994589805603, "step": 21970 }, { "epoch": 0.5789867074082244, "grad_norm": 2.639625310897827, "learning_rate": 7.94336427024543e-06, "loss": 0.13162858486175538, "step": 21980 }, { "epoch": 0.5792501226527232, "grad_norm": 2.724452257156372, "learning_rate": 7.93501602434309e-06, "loss": 0.13339933156967163, "step": 21990 }, { "epoch": 0.5795135378972219, "grad_norm": 1.6660155057907104, "learning_rate": 7.926669281425177e-06, "loss": 0.13221426010131837, "step": 22000 }, { "epoch": 0.5797769531417207, "grad_norm": 2.4266209602355957, "learning_rate": 7.91832404756681e-06, "loss": 0.1462024927139282, "step": 22010 }, { "epoch": 0.5800403683862194, "grad_norm": 2.2683935165405273, "learning_rate": 7.909980328842015e-06, "loss": 0.13477596044540405, "step": 22020 }, { "epoch": 0.5803037836307182, "grad_norm": 1.6021836996078491, "learning_rate": 7.901638131323705e-06, "loss": 0.12568423748016358, "step": 22030 }, { "epoch": 0.5805671988752169, "grad_norm": 1.2983052730560303, "learning_rate": 7.893297461083697e-06, "loss": 0.13331068754196168, "step": 22040 }, { "epoch": 0.5808306141197156, "grad_norm": 2.44528865814209, "learning_rate": 7.884958324192689e-06, "loss": 0.13883352279663086, "step": 22050 }, { "epoch": 0.5810940293642144, "grad_norm": 2.068587303161621, "learning_rate": 7.876620726720257e-06, "loss": 0.15122495889663695, "step": 22060 }, { "epoch": 0.5813574446087131, "grad_norm": 1.9183564186096191, "learning_rate": 7.868284674734877e-06, "loss": 0.13895649909973146, "step": 22070 }, { "epoch": 0.5816208598532119, "grad_norm": 0.9674125909805298, "learning_rate": 7.859950174303879e-06, "loss": 0.12874352931976318, "step": 22080 }, { "epoch": 0.5818842750977106, "grad_norm": 1.179215908050537, "learning_rate": 7.851617231493472e-06, "loss": 0.13507623672485353, "step": 22090 }, { "epoch": 0.5821476903422094, "grad_norm": 3.0959692001342773, "learning_rate": 7.843285852368734e-06, "loss": 0.13502118587493897, "step": 22100 }, { "epoch": 0.5824111055867081, "grad_norm": 1.002777338027954, "learning_rate": 7.8349560429936e-06, "loss": 0.12935606241226197, "step": 22110 }, { "epoch": 0.5826745208312069, "grad_norm": 1.799494743347168, "learning_rate": 7.826627809430867e-06, "loss": 0.14205071926116944, "step": 22120 }, { "epoch": 0.5829379360757055, "grad_norm": 1.3846181631088257, "learning_rate": 7.818301157742182e-06, "loss": 0.13322194814682006, "step": 22130 }, { "epoch": 0.5832013513202042, "grad_norm": 2.892735719680786, "learning_rate": 7.809976093988039e-06, "loss": 0.13259851932525635, "step": 22140 }, { "epoch": 0.583464766564703, "grad_norm": 3.055375814437866, "learning_rate": 7.801652624227782e-06, "loss": 0.14138441085815429, "step": 22150 }, { "epoch": 0.5837281818092017, "grad_norm": 2.7275454998016357, "learning_rate": 7.79333075451959e-06, "loss": 0.12432999610900879, "step": 22160 }, { "epoch": 0.5839915970537005, "grad_norm": 2.0654332637786865, "learning_rate": 7.785010490920477e-06, "loss": 0.13254821300506592, "step": 22170 }, { "epoch": 0.5842550122981992, "grad_norm": 3.737649440765381, "learning_rate": 7.776691839486297e-06, "loss": 0.12285923957824707, "step": 22180 }, { "epoch": 0.584518427542698, "grad_norm": 2.602126359939575, "learning_rate": 7.76837480627171e-06, "loss": 0.13526443243026734, "step": 22190 }, { "epoch": 0.5847818427871967, "grad_norm": 3.9457457065582275, "learning_rate": 7.760059397330225e-06, "loss": 0.1255089521408081, "step": 22200 }, { "epoch": 0.5850452580316955, "grad_norm": 4.102071762084961, "learning_rate": 7.75174561871415e-06, "loss": 0.134657883644104, "step": 22210 }, { "epoch": 0.5853086732761942, "grad_norm": 2.464287757873535, "learning_rate": 7.743433476474608e-06, "loss": 0.12928030490875245, "step": 22220 }, { "epoch": 0.5855720885206929, "grad_norm": 3.2074170112609863, "learning_rate": 7.73512297666154e-06, "loss": 0.12923240661621094, "step": 22230 }, { "epoch": 0.5858355037651917, "grad_norm": 1.6819660663604736, "learning_rate": 7.726814125323683e-06, "loss": 0.12271614074707031, "step": 22240 }, { "epoch": 0.5860989190096904, "grad_norm": 3.4959909915924072, "learning_rate": 7.718506928508583e-06, "loss": 0.12391782999038696, "step": 22250 }, { "epoch": 0.5863623342541892, "grad_norm": 2.444336414337158, "learning_rate": 7.71020139226257e-06, "loss": 0.12972710132598878, "step": 22260 }, { "epoch": 0.5866257494986878, "grad_norm": 1.1250463724136353, "learning_rate": 7.701897522630772e-06, "loss": 0.1266411066055298, "step": 22270 }, { "epoch": 0.5868891647431866, "grad_norm": 2.22271466255188, "learning_rate": 7.693595325657109e-06, "loss": 0.13538984060287476, "step": 22280 }, { "epoch": 0.5871525799876853, "grad_norm": 1.1386778354644775, "learning_rate": 7.685294807384273e-06, "loss": 0.12602272033691406, "step": 22290 }, { "epoch": 0.5874159952321841, "grad_norm": 2.7016735076904297, "learning_rate": 7.676995973853743e-06, "loss": 0.12903739213943483, "step": 22300 }, { "epoch": 0.5876794104766828, "grad_norm": 2.6092140674591064, "learning_rate": 7.668698831105768e-06, "loss": 0.12244329452514649, "step": 22310 }, { "epoch": 0.5879428257211815, "grad_norm": 1.429179072380066, "learning_rate": 7.66040338517936e-06, "loss": 0.13877177238464355, "step": 22320 }, { "epoch": 0.5882062409656803, "grad_norm": 1.050300121307373, "learning_rate": 7.652109642112308e-06, "loss": 0.12955511808395387, "step": 22330 }, { "epoch": 0.588469656210179, "grad_norm": 0.9848669767379761, "learning_rate": 7.643817607941157e-06, "loss": 0.11701879501342774, "step": 22340 }, { "epoch": 0.5887330714546778, "grad_norm": 2.3699893951416016, "learning_rate": 7.635527288701201e-06, "loss": 0.12662012577056886, "step": 22350 }, { "epoch": 0.5889964866991765, "grad_norm": 2.154675006866455, "learning_rate": 7.6272386904265005e-06, "loss": 0.13250333070755005, "step": 22360 }, { "epoch": 0.5892599019436753, "grad_norm": 1.4766792058944702, "learning_rate": 7.618951819149848e-06, "loss": 0.12422857284545899, "step": 22370 }, { "epoch": 0.589523317188174, "grad_norm": 1.2888741493225098, "learning_rate": 7.610666680902791e-06, "loss": 0.11054558753967285, "step": 22380 }, { "epoch": 0.5897867324326727, "grad_norm": 1.674411416053772, "learning_rate": 7.602383281715608e-06, "loss": 0.12144479751586915, "step": 22390 }, { "epoch": 0.5900501476771715, "grad_norm": 4.308157920837402, "learning_rate": 7.594101627617312e-06, "loss": 0.11798145771026611, "step": 22400 }, { "epoch": 0.5903135629216701, "grad_norm": 3.5383477210998535, "learning_rate": 7.585821724635656e-06, "loss": 0.1293273687362671, "step": 22410 }, { "epoch": 0.5905769781661689, "grad_norm": 3.409527063369751, "learning_rate": 7.577543578797097e-06, "loss": 0.12702080011367797, "step": 22420 }, { "epoch": 0.5908403934106676, "grad_norm": 4.944250583648682, "learning_rate": 7.569267196126844e-06, "loss": 0.12376648187637329, "step": 22430 }, { "epoch": 0.5911038086551664, "grad_norm": 1.8449451923370361, "learning_rate": 7.560992582648793e-06, "loss": 0.14197192192077637, "step": 22440 }, { "epoch": 0.5913672238996651, "grad_norm": 1.2160606384277344, "learning_rate": 7.552719744385566e-06, "loss": 0.13026580810546876, "step": 22450 }, { "epoch": 0.5916306391441639, "grad_norm": 1.1920984983444214, "learning_rate": 7.544448687358495e-06, "loss": 0.13362468481063844, "step": 22460 }, { "epoch": 0.5918940543886626, "grad_norm": 1.3950470685958862, "learning_rate": 7.536179417587612e-06, "loss": 0.13613946437835694, "step": 22470 }, { "epoch": 0.5921574696331613, "grad_norm": 2.7421517372131348, "learning_rate": 7.527911941091644e-06, "loss": 0.13333919048309326, "step": 22480 }, { "epoch": 0.5924208848776601, "grad_norm": 2.4044578075408936, "learning_rate": 7.519646263888023e-06, "loss": 0.1381608247756958, "step": 22490 }, { "epoch": 0.5926843001221588, "grad_norm": 2.231717109680176, "learning_rate": 7.511382391992861e-06, "loss": 0.12622437477111817, "step": 22500 }, { "epoch": 0.5929477153666576, "grad_norm": 2.669846773147583, "learning_rate": 7.503120331420963e-06, "loss": 0.1282397747039795, "step": 22510 }, { "epoch": 0.5932111306111563, "grad_norm": 1.2787513732910156, "learning_rate": 7.494860088185815e-06, "loss": 0.13110486268997193, "step": 22520 }, { "epoch": 0.5934745458556551, "grad_norm": 1.2213499546051025, "learning_rate": 7.48660166829957e-06, "loss": 0.12953357696533202, "step": 22530 }, { "epoch": 0.5937379611001538, "grad_norm": 1.901336431503296, "learning_rate": 7.478345077773075e-06, "loss": 0.12274940013885498, "step": 22540 }, { "epoch": 0.5940013763446526, "grad_norm": 2.3535995483398438, "learning_rate": 7.470090322615822e-06, "loss": 0.1379769563674927, "step": 22550 }, { "epoch": 0.5942647915891512, "grad_norm": 2.9160940647125244, "learning_rate": 7.4618374088359805e-06, "loss": 0.14046499729156495, "step": 22560 }, { "epoch": 0.5945282068336499, "grad_norm": 1.8997191190719604, "learning_rate": 7.45358634244038e-06, "loss": 0.12255038022994995, "step": 22570 }, { "epoch": 0.5947916220781487, "grad_norm": 1.8416856527328491, "learning_rate": 7.4453371294345e-06, "loss": 0.12868170738220214, "step": 22580 }, { "epoch": 0.5950550373226474, "grad_norm": 1.5144555568695068, "learning_rate": 7.437089775822477e-06, "loss": 0.14018398523330688, "step": 22590 }, { "epoch": 0.5953184525671462, "grad_norm": 3.9352827072143555, "learning_rate": 7.4288442876070865e-06, "loss": 0.13016098737716675, "step": 22600 }, { "epoch": 0.5955818678116449, "grad_norm": 1.6848182678222656, "learning_rate": 7.420600670789753e-06, "loss": 0.13005930185317993, "step": 22610 }, { "epoch": 0.5958452830561437, "grad_norm": 1.6781424283981323, "learning_rate": 7.412358931370536e-06, "loss": 0.13717880249023437, "step": 22620 }, { "epoch": 0.5961086983006424, "grad_norm": 0.952142596244812, "learning_rate": 7.404119075348129e-06, "loss": 0.12668538093566895, "step": 22630 }, { "epoch": 0.5963721135451412, "grad_norm": 1.3663328886032104, "learning_rate": 7.3958811087198555e-06, "loss": 0.12903627157211303, "step": 22640 }, { "epoch": 0.5966355287896399, "grad_norm": 2.356342077255249, "learning_rate": 7.3876450374816644e-06, "loss": 0.11533797979354858, "step": 22650 }, { "epoch": 0.5968989440341386, "grad_norm": 2.6189188957214355, "learning_rate": 7.379410867628115e-06, "loss": 0.1267960786819458, "step": 22660 }, { "epoch": 0.5971623592786374, "grad_norm": 1.8363300561904907, "learning_rate": 7.371178605152405e-06, "loss": 0.1328721046447754, "step": 22670 }, { "epoch": 0.597425774523136, "grad_norm": 1.9647793769836426, "learning_rate": 7.362948256046321e-06, "loss": 0.11610579490661621, "step": 22680 }, { "epoch": 0.5976891897676349, "grad_norm": 2.6293046474456787, "learning_rate": 7.354719826300266e-06, "loss": 0.13595972061157227, "step": 22690 }, { "epoch": 0.5979526050121335, "grad_norm": 5.447819232940674, "learning_rate": 7.346493321903251e-06, "loss": 0.11964955329895019, "step": 22700 }, { "epoch": 0.5982160202566323, "grad_norm": 2.404522657394409, "learning_rate": 7.338268748842876e-06, "loss": 0.1304348349571228, "step": 22710 }, { "epoch": 0.598479435501131, "grad_norm": 4.745080947875977, "learning_rate": 7.330046113105345e-06, "loss": 0.13891628980636597, "step": 22720 }, { "epoch": 0.5987428507456298, "grad_norm": 2.6571879386901855, "learning_rate": 7.321825420675444e-06, "loss": 0.11599383354187012, "step": 22730 }, { "epoch": 0.5990062659901285, "grad_norm": 1.6935169696807861, "learning_rate": 7.3136066775365455e-06, "loss": 0.11173630952835083, "step": 22740 }, { "epoch": 0.5992696812346272, "grad_norm": 1.9744653701782227, "learning_rate": 7.305389889670611e-06, "loss": 0.11639724969863892, "step": 22750 }, { "epoch": 0.599533096479126, "grad_norm": 4.749563694000244, "learning_rate": 7.2971750630581675e-06, "loss": 0.14217092990875244, "step": 22760 }, { "epoch": 0.5997965117236247, "grad_norm": 1.7598284482955933, "learning_rate": 7.2889622036783245e-06, "loss": 0.1256051778793335, "step": 22770 }, { "epoch": 0.6000599269681235, "grad_norm": 2.135709524154663, "learning_rate": 7.280751317508756e-06, "loss": 0.1280606746673584, "step": 22780 }, { "epoch": 0.6003233422126222, "grad_norm": 2.7593917846679688, "learning_rate": 7.272542410525691e-06, "loss": 0.12494006156921386, "step": 22790 }, { "epoch": 0.600586757457121, "grad_norm": 4.911579132080078, "learning_rate": 7.2643354887039395e-06, "loss": 0.13776047229766847, "step": 22800 }, { "epoch": 0.6008501727016197, "grad_norm": 2.0371925830841064, "learning_rate": 7.256130558016843e-06, "loss": 0.12653855085372925, "step": 22810 }, { "epoch": 0.6011135879461185, "grad_norm": 2.234006404876709, "learning_rate": 7.247927624436304e-06, "loss": 0.1279069423675537, "step": 22820 }, { "epoch": 0.6013770031906172, "grad_norm": 4.83394193649292, "learning_rate": 7.239726693932779e-06, "loss": 0.13580352067947388, "step": 22830 }, { "epoch": 0.6016404184351158, "grad_norm": 1.0002046823501587, "learning_rate": 7.2315277724752495e-06, "loss": 0.10965046882629395, "step": 22840 }, { "epoch": 0.6019038336796146, "grad_norm": 2.2983877658843994, "learning_rate": 7.223330866031253e-06, "loss": 0.12598795890808107, "step": 22850 }, { "epoch": 0.6021672489241133, "grad_norm": 1.9720898866653442, "learning_rate": 7.215135980566847e-06, "loss": 0.12876539230346679, "step": 22860 }, { "epoch": 0.6024306641686121, "grad_norm": 2.838481903076172, "learning_rate": 7.2069431220466215e-06, "loss": 0.11956233978271484, "step": 22870 }, { "epoch": 0.6026940794131108, "grad_norm": 2.3618435859680176, "learning_rate": 7.198752296433697e-06, "loss": 0.12448790073394775, "step": 22880 }, { "epoch": 0.6029574946576096, "grad_norm": 1.229212999343872, "learning_rate": 7.190563509689705e-06, "loss": 0.12335752248764038, "step": 22890 }, { "epoch": 0.6032209099021083, "grad_norm": 2.2420313358306885, "learning_rate": 7.182376767774803e-06, "loss": 0.12442591190338134, "step": 22900 }, { "epoch": 0.603484325146607, "grad_norm": 2.784210205078125, "learning_rate": 7.174192076647653e-06, "loss": 0.13121044635772705, "step": 22910 }, { "epoch": 0.6037477403911058, "grad_norm": 1.2931396961212158, "learning_rate": 7.166009442265423e-06, "loss": 0.11269650459289551, "step": 22920 }, { "epoch": 0.6040111556356045, "grad_norm": 1.814039945602417, "learning_rate": 7.15782887058379e-06, "loss": 0.11696739196777343, "step": 22930 }, { "epoch": 0.6042745708801033, "grad_norm": 2.9440503120422363, "learning_rate": 7.149650367556926e-06, "loss": 0.12217247486114502, "step": 22940 }, { "epoch": 0.604537986124602, "grad_norm": 1.6457151174545288, "learning_rate": 7.141473939137498e-06, "loss": 0.12115293741226196, "step": 22950 }, { "epoch": 0.6048014013691008, "grad_norm": 2.036334753036499, "learning_rate": 7.133299591276665e-06, "loss": 0.13493063449859619, "step": 22960 }, { "epoch": 0.6050648166135995, "grad_norm": 4.74530029296875, "learning_rate": 7.12512732992407e-06, "loss": 0.12255135774612427, "step": 22970 }, { "epoch": 0.6053282318580983, "grad_norm": 2.3480782508850098, "learning_rate": 7.1169571610278375e-06, "loss": 0.13819921016693115, "step": 22980 }, { "epoch": 0.6055916471025969, "grad_norm": 2.0502991676330566, "learning_rate": 7.108789090534569e-06, "loss": 0.12352769374847412, "step": 22990 }, { "epoch": 0.6058550623470956, "grad_norm": 0.9066327810287476, "learning_rate": 7.100623124389334e-06, "loss": 0.11939864158630371, "step": 23000 }, { "epoch": 0.6061184775915944, "grad_norm": 2.4121155738830566, "learning_rate": 7.092459268535685e-06, "loss": 0.12477411031723022, "step": 23010 }, { "epoch": 0.6063818928360931, "grad_norm": 2.26101016998291, "learning_rate": 7.084297528915616e-06, "loss": 0.1380523681640625, "step": 23020 }, { "epoch": 0.6066453080805919, "grad_norm": 2.2820870876312256, "learning_rate": 7.076137911469603e-06, "loss": 0.12343144416809082, "step": 23030 }, { "epoch": 0.6069087233250906, "grad_norm": 2.6523516178131104, "learning_rate": 7.067980422136565e-06, "loss": 0.1240123987197876, "step": 23040 }, { "epoch": 0.6071721385695894, "grad_norm": 3.5411691665649414, "learning_rate": 7.059825066853872e-06, "loss": 0.13334929943084717, "step": 23050 }, { "epoch": 0.6074355538140881, "grad_norm": 2.1686716079711914, "learning_rate": 7.0516718515573445e-06, "loss": 0.13185633420944215, "step": 23060 }, { "epoch": 0.6076989690585869, "grad_norm": 1.0651129484176636, "learning_rate": 7.043520782181248e-06, "loss": 0.12312207221984864, "step": 23070 }, { "epoch": 0.6079623843030856, "grad_norm": 1.2360291481018066, "learning_rate": 7.035371864658276e-06, "loss": 0.13471639156341553, "step": 23080 }, { "epoch": 0.6082257995475843, "grad_norm": 4.331872463226318, "learning_rate": 7.0272251049195684e-06, "loss": 0.12393448352813721, "step": 23090 }, { "epoch": 0.6084892147920831, "grad_norm": 1.568347454071045, "learning_rate": 7.019080508894683e-06, "loss": 0.13014755249023438, "step": 23100 }, { "epoch": 0.6087526300365818, "grad_norm": 1.398984670639038, "learning_rate": 7.010938082511614e-06, "loss": 0.12751630544662476, "step": 23110 }, { "epoch": 0.6090160452810806, "grad_norm": 1.293954849243164, "learning_rate": 7.0027978316967705e-06, "loss": 0.11616168022155762, "step": 23120 }, { "epoch": 0.6092794605255792, "grad_norm": 1.8594557046890259, "learning_rate": 6.994659762374967e-06, "loss": 0.12739017009735107, "step": 23130 }, { "epoch": 0.609542875770078, "grad_norm": 2.4192233085632324, "learning_rate": 6.986523880469458e-06, "loss": 0.13551883697509765, "step": 23140 }, { "epoch": 0.6098062910145767, "grad_norm": 1.9973586797714233, "learning_rate": 6.978390191901878e-06, "loss": 0.14110360145568848, "step": 23150 }, { "epoch": 0.6100697062590755, "grad_norm": 3.5355358123779297, "learning_rate": 6.970258702592277e-06, "loss": 0.1320729970932007, "step": 23160 }, { "epoch": 0.6103331215035742, "grad_norm": 2.4816653728485107, "learning_rate": 6.962129418459109e-06, "loss": 0.12177754640579223, "step": 23170 }, { "epoch": 0.6105965367480729, "grad_norm": 1.3389955759048462, "learning_rate": 6.954002345419214e-06, "loss": 0.11925629377365113, "step": 23180 }, { "epoch": 0.6108599519925717, "grad_norm": 2.850316286087036, "learning_rate": 6.945877489387829e-06, "loss": 0.1391072988510132, "step": 23190 }, { "epoch": 0.6111233672370704, "grad_norm": 1.347436547279358, "learning_rate": 6.9377548562785735e-06, "loss": 0.1355818033218384, "step": 23200 }, { "epoch": 0.6113867824815692, "grad_norm": 2.5686228275299072, "learning_rate": 6.929634452003447e-06, "loss": 0.1317095637321472, "step": 23210 }, { "epoch": 0.6116501977260679, "grad_norm": 2.565027952194214, "learning_rate": 6.921516282472837e-06, "loss": 0.13256385326385497, "step": 23220 }, { "epoch": 0.6119136129705667, "grad_norm": 2.8809404373168945, "learning_rate": 6.91340035359549e-06, "loss": 0.13174269199371338, "step": 23230 }, { "epoch": 0.6121770282150654, "grad_norm": 1.8195396661758423, "learning_rate": 6.905286671278536e-06, "loss": 0.13109210729599, "step": 23240 }, { "epoch": 0.6124404434595642, "grad_norm": 3.0484118461608887, "learning_rate": 6.897175241427462e-06, "loss": 0.1197645902633667, "step": 23250 }, { "epoch": 0.6127038587040629, "grad_norm": 2.280795097351074, "learning_rate": 6.889066069946107e-06, "loss": 0.1284419536590576, "step": 23260 }, { "epoch": 0.6129672739485615, "grad_norm": 2.4106802940368652, "learning_rate": 6.880959162736691e-06, "loss": 0.1329984188079834, "step": 23270 }, { "epoch": 0.6132306891930603, "grad_norm": 1.794170618057251, "learning_rate": 6.872854525699761e-06, "loss": 0.13378055095672609, "step": 23280 }, { "epoch": 0.613494104437559, "grad_norm": 3.4788529872894287, "learning_rate": 6.8647521647342205e-06, "loss": 0.12131696939468384, "step": 23290 }, { "epoch": 0.6137575196820578, "grad_norm": 1.5342469215393066, "learning_rate": 6.856652085737323e-06, "loss": 0.12837914228439332, "step": 23300 }, { "epoch": 0.6140209349265565, "grad_norm": 1.2043019533157349, "learning_rate": 6.848554294604652e-06, "loss": 0.116176438331604, "step": 23310 }, { "epoch": 0.6142843501710553, "grad_norm": 1.4071837663650513, "learning_rate": 6.8404587972301316e-06, "loss": 0.11795487403869628, "step": 23320 }, { "epoch": 0.614547765415554, "grad_norm": 1.72222101688385, "learning_rate": 6.8323655995060135e-06, "loss": 0.13982458114624025, "step": 23330 }, { "epoch": 0.6148111806600528, "grad_norm": 2.3728790283203125, "learning_rate": 6.824274707322874e-06, "loss": 0.13157637119293214, "step": 23340 }, { "epoch": 0.6150745959045515, "grad_norm": 2.882805109024048, "learning_rate": 6.816186126569615e-06, "loss": 0.13189712762832642, "step": 23350 }, { "epoch": 0.6153380111490502, "grad_norm": 5.999914646148682, "learning_rate": 6.808099863133453e-06, "loss": 0.13808579444885255, "step": 23360 }, { "epoch": 0.615601426393549, "grad_norm": 2.5893337726593018, "learning_rate": 6.800015922899923e-06, "loss": 0.12749876976013183, "step": 23370 }, { "epoch": 0.6158648416380477, "grad_norm": 1.6816071271896362, "learning_rate": 6.7919343117528655e-06, "loss": 0.12903517484664917, "step": 23380 }, { "epoch": 0.6161282568825465, "grad_norm": 4.7601423263549805, "learning_rate": 6.783855035574418e-06, "loss": 0.13068649768829346, "step": 23390 }, { "epoch": 0.6163916721270452, "grad_norm": 2.159823179244995, "learning_rate": 6.775778100245038e-06, "loss": 0.13690558671951295, "step": 23400 }, { "epoch": 0.616655087371544, "grad_norm": 2.8388426303863525, "learning_rate": 6.76770351164346e-06, "loss": 0.12586286067962646, "step": 23410 }, { "epoch": 0.6169185026160426, "grad_norm": 4.979775428771973, "learning_rate": 6.759631275646715e-06, "loss": 0.1266077995300293, "step": 23420 }, { "epoch": 0.6171819178605413, "grad_norm": 3.1789674758911133, "learning_rate": 6.751561398130133e-06, "loss": 0.1270599603652954, "step": 23430 }, { "epoch": 0.6174453331050401, "grad_norm": 2.0713438987731934, "learning_rate": 6.743493884967312e-06, "loss": 0.12590211629867554, "step": 23440 }, { "epoch": 0.6177087483495388, "grad_norm": 3.0916459560394287, "learning_rate": 6.735428742030141e-06, "loss": 0.12631548643112184, "step": 23450 }, { "epoch": 0.6179721635940376, "grad_norm": 1.43771493434906, "learning_rate": 6.7273659751887775e-06, "loss": 0.12638585567474364, "step": 23460 }, { "epoch": 0.6182355788385363, "grad_norm": 1.9423599243164062, "learning_rate": 6.719305590311649e-06, "loss": 0.12447209358215332, "step": 23470 }, { "epoch": 0.6184989940830351, "grad_norm": 2.702345371246338, "learning_rate": 6.7112475932654554e-06, "loss": 0.12252316474914551, "step": 23480 }, { "epoch": 0.6187624093275338, "grad_norm": 1.6822645664215088, "learning_rate": 6.70319198991515e-06, "loss": 0.12156713008880615, "step": 23490 }, { "epoch": 0.6190258245720326, "grad_norm": 2.170996904373169, "learning_rate": 6.695138786123952e-06, "loss": 0.11593495607376099, "step": 23500 }, { "epoch": 0.6192892398165313, "grad_norm": 1.698368787765503, "learning_rate": 6.687087987753331e-06, "loss": 0.1273486614227295, "step": 23510 }, { "epoch": 0.61955265506103, "grad_norm": 2.569575071334839, "learning_rate": 6.679039600663001e-06, "loss": 0.13265880346298217, "step": 23520 }, { "epoch": 0.6198160703055288, "grad_norm": 1.3788694143295288, "learning_rate": 6.670993630710929e-06, "loss": 0.1223791241645813, "step": 23530 }, { "epoch": 0.6200794855500275, "grad_norm": 3.0434763431549072, "learning_rate": 6.662950083753319e-06, "loss": 0.11394158601760865, "step": 23540 }, { "epoch": 0.6203429007945263, "grad_norm": 1.6219500303268433, "learning_rate": 6.654908965644609e-06, "loss": 0.133498477935791, "step": 23550 }, { "epoch": 0.6206063160390249, "grad_norm": 3.6617209911346436, "learning_rate": 6.646870282237474e-06, "loss": 0.11833910942077637, "step": 23560 }, { "epoch": 0.6208697312835237, "grad_norm": 1.5186281204223633, "learning_rate": 6.6388340393828125e-06, "loss": 0.1331470012664795, "step": 23570 }, { "epoch": 0.6211331465280224, "grad_norm": 3.244636297225952, "learning_rate": 6.6308002429297516e-06, "loss": 0.11852875947952271, "step": 23580 }, { "epoch": 0.6213965617725212, "grad_norm": 1.4751616716384888, "learning_rate": 6.622768898725634e-06, "loss": 0.11435632705688477, "step": 23590 }, { "epoch": 0.6216599770170199, "grad_norm": 3.393868923187256, "learning_rate": 6.614740012616017e-06, "loss": 0.13412563800811766, "step": 23600 }, { "epoch": 0.6219233922615186, "grad_norm": 4.446054935455322, "learning_rate": 6.606713590444673e-06, "loss": 0.1293182849884033, "step": 23610 }, { "epoch": 0.6221868075060174, "grad_norm": 4.027760982513428, "learning_rate": 6.598689638053577e-06, "loss": 0.12046589851379394, "step": 23620 }, { "epoch": 0.6224502227505161, "grad_norm": 1.9127986431121826, "learning_rate": 6.590668161282903e-06, "loss": 0.12246872186660766, "step": 23630 }, { "epoch": 0.6227136379950149, "grad_norm": 3.3416380882263184, "learning_rate": 6.582649165971037e-06, "loss": 0.1229668378829956, "step": 23640 }, { "epoch": 0.6229770532395136, "grad_norm": 1.3431938886642456, "learning_rate": 6.574632657954542e-06, "loss": 0.11861591339111328, "step": 23650 }, { "epoch": 0.6232404684840124, "grad_norm": 1.4579219818115234, "learning_rate": 6.5666186430681835e-06, "loss": 0.11781656742095947, "step": 23660 }, { "epoch": 0.6235038837285111, "grad_norm": 2.299748182296753, "learning_rate": 6.558607127144905e-06, "loss": 0.11574631929397583, "step": 23670 }, { "epoch": 0.6237672989730099, "grad_norm": 2.6402978897094727, "learning_rate": 6.550598116015832e-06, "loss": 0.11817092895507812, "step": 23680 }, { "epoch": 0.6240307142175086, "grad_norm": 6.471603870391846, "learning_rate": 6.54259161551027e-06, "loss": 0.13782968521118164, "step": 23690 }, { "epoch": 0.6242941294620072, "grad_norm": 1.6988328695297241, "learning_rate": 6.534587631455696e-06, "loss": 0.12424542903900146, "step": 23700 }, { "epoch": 0.624557544706506, "grad_norm": 1.562666416168213, "learning_rate": 6.5265861696777556e-06, "loss": 0.13313446044921876, "step": 23710 }, { "epoch": 0.6248209599510047, "grad_norm": 2.8029518127441406, "learning_rate": 6.518587236000259e-06, "loss": 0.12291786670684815, "step": 23720 }, { "epoch": 0.6250843751955035, "grad_norm": 2.6074962615966797, "learning_rate": 6.510590836245166e-06, "loss": 0.11369004249572753, "step": 23730 }, { "epoch": 0.6253477904400022, "grad_norm": 1.8760346174240112, "learning_rate": 6.502596976232616e-06, "loss": 0.12283486127853394, "step": 23740 }, { "epoch": 0.625611205684501, "grad_norm": 4.062148094177246, "learning_rate": 6.494605661780874e-06, "loss": 0.1292288064956665, "step": 23750 }, { "epoch": 0.6258746209289997, "grad_norm": 2.0715274810791016, "learning_rate": 6.486616898706365e-06, "loss": 0.12285134792327881, "step": 23760 }, { "epoch": 0.6261380361734985, "grad_norm": 2.0520331859588623, "learning_rate": 6.47863069282366e-06, "loss": 0.12106752395629883, "step": 23770 }, { "epoch": 0.6264014514179972, "grad_norm": 1.7122641801834106, "learning_rate": 6.47064704994546e-06, "loss": 0.11837937831878662, "step": 23780 }, { "epoch": 0.6266648666624959, "grad_norm": 2.7230892181396484, "learning_rate": 6.462665975882607e-06, "loss": 0.1203126311302185, "step": 23790 }, { "epoch": 0.6269282819069947, "grad_norm": 3.5612950325012207, "learning_rate": 6.454687476444071e-06, "loss": 0.1210473895072937, "step": 23800 }, { "epoch": 0.6271916971514934, "grad_norm": 2.175149440765381, "learning_rate": 6.446711557436944e-06, "loss": 0.10729343891143799, "step": 23810 }, { "epoch": 0.6274551123959922, "grad_norm": 1.5041693449020386, "learning_rate": 6.438738224666452e-06, "loss": 0.12369651794433593, "step": 23820 }, { "epoch": 0.6277185276404909, "grad_norm": 1.6684211492538452, "learning_rate": 6.4307674839359225e-06, "loss": 0.13294976949691772, "step": 23830 }, { "epoch": 0.6279819428849897, "grad_norm": 2.152061700820923, "learning_rate": 6.42279934104681e-06, "loss": 0.12373347282409668, "step": 23840 }, { "epoch": 0.6282453581294883, "grad_norm": 2.3555407524108887, "learning_rate": 6.4148338017986725e-06, "loss": 0.11967902183532715, "step": 23850 }, { "epoch": 0.6285087733739871, "grad_norm": 1.6072096824645996, "learning_rate": 6.406870871989166e-06, "loss": 0.1141202688217163, "step": 23860 }, { "epoch": 0.6287721886184858, "grad_norm": 1.8735636472702026, "learning_rate": 6.398910557414064e-06, "loss": 0.12568098306655884, "step": 23870 }, { "epoch": 0.6290356038629845, "grad_norm": 2.370623826980591, "learning_rate": 6.390952863867224e-06, "loss": 0.11535359621047973, "step": 23880 }, { "epoch": 0.6292990191074833, "grad_norm": 4.551270008087158, "learning_rate": 6.382997797140591e-06, "loss": 0.1263345956802368, "step": 23890 }, { "epoch": 0.629562434351982, "grad_norm": 1.3994423151016235, "learning_rate": 6.375045363024217e-06, "loss": 0.12388112545013427, "step": 23900 }, { "epoch": 0.6298258495964808, "grad_norm": 2.4847350120544434, "learning_rate": 6.367095567306219e-06, "loss": 0.1363063335418701, "step": 23910 }, { "epoch": 0.6300892648409795, "grad_norm": 1.2663614749908447, "learning_rate": 6.359148415772805e-06, "loss": 0.10347590446472169, "step": 23920 }, { "epoch": 0.6303526800854783, "grad_norm": 2.80417799949646, "learning_rate": 6.351203914208254e-06, "loss": 0.13042981624603273, "step": 23930 }, { "epoch": 0.630616095329977, "grad_norm": 2.4920356273651123, "learning_rate": 6.343262068394914e-06, "loss": 0.12825319766998292, "step": 23940 }, { "epoch": 0.6308795105744757, "grad_norm": 3.577592611312866, "learning_rate": 6.335322884113208e-06, "loss": 0.13463151454925537, "step": 23950 }, { "epoch": 0.6311429258189745, "grad_norm": 2.056514024734497, "learning_rate": 6.3273863671416146e-06, "loss": 0.12022595405578614, "step": 23960 }, { "epoch": 0.6314063410634732, "grad_norm": 4.644936561584473, "learning_rate": 6.319452523256673e-06, "loss": 0.12246954441070557, "step": 23970 }, { "epoch": 0.631669756307972, "grad_norm": 2.9015145301818848, "learning_rate": 6.31152135823298e-06, "loss": 0.13074686527252197, "step": 23980 }, { "epoch": 0.6319331715524706, "grad_norm": 1.051571249961853, "learning_rate": 6.303592877843171e-06, "loss": 0.1266113519668579, "step": 23990 }, { "epoch": 0.6321965867969694, "grad_norm": 2.9092159271240234, "learning_rate": 6.295667087857948e-06, "loss": 0.12300969362258911, "step": 24000 }, { "epoch": 0.6324600020414681, "grad_norm": 5.285294055938721, "learning_rate": 6.287743994046035e-06, "loss": 0.11695022583007812, "step": 24010 }, { "epoch": 0.6327234172859669, "grad_norm": 5.236830234527588, "learning_rate": 6.279823602174202e-06, "loss": 0.12130875587463379, "step": 24020 }, { "epoch": 0.6329868325304656, "grad_norm": 2.8691070079803467, "learning_rate": 6.271905918007255e-06, "loss": 0.12564951181411743, "step": 24030 }, { "epoch": 0.6332502477749643, "grad_norm": 1.8033490180969238, "learning_rate": 6.263990947308022e-06, "loss": 0.1330721139907837, "step": 24040 }, { "epoch": 0.6335136630194631, "grad_norm": 1.6029472351074219, "learning_rate": 6.2560786958373645e-06, "loss": 0.13339834213256835, "step": 24050 }, { "epoch": 0.6337770782639618, "grad_norm": 1.1759430170059204, "learning_rate": 6.248169169354159e-06, "loss": 0.1231231451034546, "step": 24060 }, { "epoch": 0.6340404935084606, "grad_norm": 2.040376901626587, "learning_rate": 6.240262373615298e-06, "loss": 0.11958739757537842, "step": 24070 }, { "epoch": 0.6343039087529593, "grad_norm": 1.6724627017974854, "learning_rate": 6.232358314375693e-06, "loss": 0.11719021797180176, "step": 24080 }, { "epoch": 0.6345673239974581, "grad_norm": 2.8318963050842285, "learning_rate": 6.224456997388254e-06, "loss": 0.11722817420959472, "step": 24090 }, { "epoch": 0.6348307392419568, "grad_norm": 2.6055798530578613, "learning_rate": 6.2165584284039046e-06, "loss": 0.12114338874816895, "step": 24100 }, { "epoch": 0.6350941544864556, "grad_norm": 2.8081679344177246, "learning_rate": 6.208662613171566e-06, "loss": 0.1242711067199707, "step": 24110 }, { "epoch": 0.6353575697309543, "grad_norm": 3.938969373703003, "learning_rate": 6.200769557438144e-06, "loss": 0.12206132411956787, "step": 24120 }, { "epoch": 0.6356209849754529, "grad_norm": 1.267844319343567, "learning_rate": 6.192879266948554e-06, "loss": 0.1224173665046692, "step": 24130 }, { "epoch": 0.6358844002199517, "grad_norm": 1.4853806495666504, "learning_rate": 6.184991747445686e-06, "loss": 0.12444579601287842, "step": 24140 }, { "epoch": 0.6361478154644504, "grad_norm": 2.8907158374786377, "learning_rate": 6.1771070046704165e-06, "loss": 0.1161499261856079, "step": 24150 }, { "epoch": 0.6364112307089492, "grad_norm": 1.792475938796997, "learning_rate": 6.169225044361604e-06, "loss": 0.12884924411773682, "step": 24160 }, { "epoch": 0.6366746459534479, "grad_norm": 3.6411683559417725, "learning_rate": 6.161345872256075e-06, "loss": 0.1327136516571045, "step": 24170 }, { "epoch": 0.6369380611979467, "grad_norm": 1.950809121131897, "learning_rate": 6.153469494088638e-06, "loss": 0.12154829502105713, "step": 24180 }, { "epoch": 0.6372014764424454, "grad_norm": 2.1783664226531982, "learning_rate": 6.1455959155920555e-06, "loss": 0.11906989812850952, "step": 24190 }, { "epoch": 0.6374648916869442, "grad_norm": 1.9816340208053589, "learning_rate": 6.1377251424970555e-06, "loss": 0.11065722703933716, "step": 24200 }, { "epoch": 0.6377283069314429, "grad_norm": 2.7789134979248047, "learning_rate": 6.1298571805323344e-06, "loss": 0.11500167846679688, "step": 24210 }, { "epoch": 0.6379917221759416, "grad_norm": 2.2984514236450195, "learning_rate": 6.121992035424529e-06, "loss": 0.1106907844543457, "step": 24220 }, { "epoch": 0.6382551374204404, "grad_norm": 2.2290685176849365, "learning_rate": 6.114129712898227e-06, "loss": 0.13314719200134278, "step": 24230 }, { "epoch": 0.6385185526649391, "grad_norm": 2.399221658706665, "learning_rate": 6.106270218675975e-06, "loss": 0.11631650924682617, "step": 24240 }, { "epoch": 0.6387819679094379, "grad_norm": 2.3177783489227295, "learning_rate": 6.098413558478244e-06, "loss": 0.1236376404762268, "step": 24250 }, { "epoch": 0.6390453831539366, "grad_norm": 1.5042330026626587, "learning_rate": 6.090559738023456e-06, "loss": 0.12559905052185058, "step": 24260 }, { "epoch": 0.6393087983984354, "grad_norm": 2.337920904159546, "learning_rate": 6.0827087630279605e-06, "loss": 0.11997114419937134, "step": 24270 }, { "epoch": 0.639572213642934, "grad_norm": 1.8905036449432373, "learning_rate": 6.074860639206028e-06, "loss": 0.1214374303817749, "step": 24280 }, { "epoch": 0.6398356288874328, "grad_norm": 1.7872533798217773, "learning_rate": 6.067015372269872e-06, "loss": 0.11555410623550415, "step": 24290 }, { "epoch": 0.6400990441319315, "grad_norm": 2.307758092880249, "learning_rate": 6.059172967929609e-06, "loss": 0.12027500867843628, "step": 24300 }, { "epoch": 0.6403624593764302, "grad_norm": 1.7789331674575806, "learning_rate": 6.0513334318932825e-06, "loss": 0.11524627208709717, "step": 24310 }, { "epoch": 0.640625874620929, "grad_norm": 1.5150800943374634, "learning_rate": 6.043496769866848e-06, "loss": 0.1257576584815979, "step": 24320 }, { "epoch": 0.6408892898654277, "grad_norm": 1.7254152297973633, "learning_rate": 6.035662987554154e-06, "loss": 0.11536033153533935, "step": 24330 }, { "epoch": 0.6411527051099265, "grad_norm": 2.5077502727508545, "learning_rate": 6.027832090656982e-06, "loss": 0.11932727098464965, "step": 24340 }, { "epoch": 0.6414161203544252, "grad_norm": 1.7078330516815186, "learning_rate": 6.020004084874984e-06, "loss": 0.11982719898223877, "step": 24350 }, { "epoch": 0.641679535598924, "grad_norm": 1.6057512760162354, "learning_rate": 6.012178975905725e-06, "loss": 0.11603846549987792, "step": 24360 }, { "epoch": 0.6419429508434227, "grad_norm": 3.993624687194824, "learning_rate": 6.0043567694446566e-06, "loss": 0.10495501756668091, "step": 24370 }, { "epoch": 0.6422063660879215, "grad_norm": 3.70585560798645, "learning_rate": 5.9965374711851175e-06, "loss": 0.11851446628570557, "step": 24380 }, { "epoch": 0.6424697813324202, "grad_norm": 1.6720988750457764, "learning_rate": 5.9887210868183325e-06, "loss": 0.1276315689086914, "step": 24390 }, { "epoch": 0.6427331965769189, "grad_norm": 1.7475216388702393, "learning_rate": 5.9809076220334024e-06, "loss": 0.12524104118347168, "step": 24400 }, { "epoch": 0.6429966118214177, "grad_norm": 2.2409250736236572, "learning_rate": 5.973097082517301e-06, "loss": 0.12284747362136841, "step": 24410 }, { "epoch": 0.6432600270659163, "grad_norm": 4.8946967124938965, "learning_rate": 5.965289473954879e-06, "loss": 0.11414949893951416, "step": 24420 }, { "epoch": 0.6435234423104151, "grad_norm": 3.141935348510742, "learning_rate": 5.95748480202885e-06, "loss": 0.12137355804443359, "step": 24430 }, { "epoch": 0.6437868575549138, "grad_norm": 3.3144924640655518, "learning_rate": 5.949683072419792e-06, "loss": 0.11561769247055054, "step": 24440 }, { "epoch": 0.6440502727994126, "grad_norm": 2.870086431503296, "learning_rate": 5.941884290806142e-06, "loss": 0.11908930540084839, "step": 24450 }, { "epoch": 0.6443136880439113, "grad_norm": 2.4738075733184814, "learning_rate": 5.934088462864181e-06, "loss": 0.1222222089767456, "step": 24460 }, { "epoch": 0.6445771032884101, "grad_norm": 2.0880982875823975, "learning_rate": 5.926295594268059e-06, "loss": 0.12181336879730224, "step": 24470 }, { "epoch": 0.6448405185329088, "grad_norm": 2.7661633491516113, "learning_rate": 5.918505690689759e-06, "loss": 0.12092170715332032, "step": 24480 }, { "epoch": 0.6451039337774075, "grad_norm": 3.5697638988494873, "learning_rate": 5.910718757799105e-06, "loss": 0.13029205799102783, "step": 24490 }, { "epoch": 0.6453673490219063, "grad_norm": 1.6735531091690063, "learning_rate": 5.902934801263766e-06, "loss": 0.12665126323699952, "step": 24500 }, { "epoch": 0.645630764266405, "grad_norm": 1.5593141317367554, "learning_rate": 5.89515382674924e-06, "loss": 0.11091468334197999, "step": 24510 }, { "epoch": 0.6458941795109038, "grad_norm": 1.9528224468231201, "learning_rate": 5.88737583991886e-06, "loss": 0.13578345775604247, "step": 24520 }, { "epoch": 0.6461575947554025, "grad_norm": 2.3952558040618896, "learning_rate": 5.879600846433777e-06, "loss": 0.11530869007110596, "step": 24530 }, { "epoch": 0.6464210099999013, "grad_norm": 2.3278138637542725, "learning_rate": 5.871828851952965e-06, "loss": 0.12276225090026856, "step": 24540 }, { "epoch": 0.6466844252444, "grad_norm": 2.8133885860443115, "learning_rate": 5.864059862133221e-06, "loss": 0.10975878238677979, "step": 24550 }, { "epoch": 0.6469478404888986, "grad_norm": 1.4985510110855103, "learning_rate": 5.856293882629147e-06, "loss": 0.11101841926574707, "step": 24560 }, { "epoch": 0.6472112557333974, "grad_norm": 2.668635606765747, "learning_rate": 5.848530919093162e-06, "loss": 0.12137398719787598, "step": 24570 }, { "epoch": 0.6474746709778961, "grad_norm": 2.184644937515259, "learning_rate": 5.840770977175485e-06, "loss": 0.11761245727539063, "step": 24580 }, { "epoch": 0.6477380862223949, "grad_norm": 2.8135440349578857, "learning_rate": 5.833014062524129e-06, "loss": 0.1165842056274414, "step": 24590 }, { "epoch": 0.6480015014668936, "grad_norm": 1.5824942588806152, "learning_rate": 5.825260180784923e-06, "loss": 0.11304153203964233, "step": 24600 }, { "epoch": 0.6482649167113924, "grad_norm": 2.5701305866241455, "learning_rate": 5.817509337601475e-06, "loss": 0.12482750415802002, "step": 24610 }, { "epoch": 0.6485283319558911, "grad_norm": 1.7992264032363892, "learning_rate": 5.8097615386151686e-06, "loss": 0.11791853904724121, "step": 24620 }, { "epoch": 0.6487917472003899, "grad_norm": 3.1548452377319336, "learning_rate": 5.8020167894652035e-06, "loss": 0.12034391164779663, "step": 24630 }, { "epoch": 0.6490551624448886, "grad_norm": 2.3261232376098633, "learning_rate": 5.79427509578853e-06, "loss": 0.12056100368499756, "step": 24640 }, { "epoch": 0.6493185776893873, "grad_norm": 2.5284652709960938, "learning_rate": 5.786536463219899e-06, "loss": 0.12481971979141235, "step": 24650 }, { "epoch": 0.6495819929338861, "grad_norm": 1.577832818031311, "learning_rate": 5.778800897391812e-06, "loss": 0.10775902271270751, "step": 24660 }, { "epoch": 0.6498454081783848, "grad_norm": 1.1896721124649048, "learning_rate": 5.771068403934545e-06, "loss": 0.12645281553268434, "step": 24670 }, { "epoch": 0.6501088234228836, "grad_norm": 2.2477049827575684, "learning_rate": 5.76333898847615e-06, "loss": 0.11388319730758667, "step": 24680 }, { "epoch": 0.6503722386673823, "grad_norm": 1.494344711303711, "learning_rate": 5.755612656642424e-06, "loss": 0.11645935773849488, "step": 24690 }, { "epoch": 0.650635653911881, "grad_norm": 2.425726890563965, "learning_rate": 5.747889414056925e-06, "loss": 0.11727994680404663, "step": 24700 }, { "epoch": 0.6508990691563797, "grad_norm": 2.5093295574188232, "learning_rate": 5.740169266340963e-06, "loss": 0.11726016998291015, "step": 24710 }, { "epoch": 0.6511624844008785, "grad_norm": 1.6273459196090698, "learning_rate": 5.732452219113589e-06, "loss": 0.11948788166046143, "step": 24720 }, { "epoch": 0.6514258996453772, "grad_norm": 3.7475669384002686, "learning_rate": 5.724738277991611e-06, "loss": 0.11895780563354492, "step": 24730 }, { "epoch": 0.6516893148898759, "grad_norm": 2.4605050086975098, "learning_rate": 5.717027448589568e-06, "loss": 0.12915313243865967, "step": 24740 }, { "epoch": 0.6519527301343747, "grad_norm": 1.0920137166976929, "learning_rate": 5.709319736519725e-06, "loss": 0.10998222827911378, "step": 24750 }, { "epoch": 0.6522161453788734, "grad_norm": 1.6856118440628052, "learning_rate": 5.701615147392096e-06, "loss": 0.12686764001846312, "step": 24760 }, { "epoch": 0.6524795606233722, "grad_norm": 1.5113343000411987, "learning_rate": 5.6939136868144075e-06, "loss": 0.12247774600982667, "step": 24770 }, { "epoch": 0.6527429758678709, "grad_norm": 1.449407696723938, "learning_rate": 5.686215360392127e-06, "loss": 0.114306640625, "step": 24780 }, { "epoch": 0.6530063911123697, "grad_norm": 7.382358074188232, "learning_rate": 5.678520173728414e-06, "loss": 0.11968895196914672, "step": 24790 }, { "epoch": 0.6532698063568684, "grad_norm": 2.963205099105835, "learning_rate": 5.67082813242416e-06, "loss": 0.11502240896224976, "step": 24800 }, { "epoch": 0.6535332216013672, "grad_norm": 1.2680944204330444, "learning_rate": 5.663139242077971e-06, "loss": 0.1161266565322876, "step": 24810 }, { "epoch": 0.6537966368458659, "grad_norm": 1.7871346473693848, "learning_rate": 5.6554535082861496e-06, "loss": 0.1238161563873291, "step": 24820 }, { "epoch": 0.6540600520903646, "grad_norm": 3.415555477142334, "learning_rate": 5.647770936642705e-06, "loss": 0.11426963806152343, "step": 24830 }, { "epoch": 0.6543234673348634, "grad_norm": 2.5621180534362793, "learning_rate": 5.6400915327393435e-06, "loss": 0.1209221363067627, "step": 24840 }, { "epoch": 0.654586882579362, "grad_norm": 1.410569429397583, "learning_rate": 5.632415302165464e-06, "loss": 0.11599645614624024, "step": 24850 }, { "epoch": 0.6548502978238608, "grad_norm": 1.966328740119934, "learning_rate": 5.624742250508164e-06, "loss": 0.11131763458251953, "step": 24860 }, { "epoch": 0.6551137130683595, "grad_norm": 1.5224119424819946, "learning_rate": 5.617072383352222e-06, "loss": 0.12179665565490723, "step": 24870 }, { "epoch": 0.6553771283128583, "grad_norm": 2.803194999694824, "learning_rate": 5.6094057062800886e-06, "loss": 0.10979762077331542, "step": 24880 }, { "epoch": 0.655640543557357, "grad_norm": 1.9688349962234497, "learning_rate": 5.601742224871911e-06, "loss": 0.1286189317703247, "step": 24890 }, { "epoch": 0.6559039588018558, "grad_norm": 4.431042194366455, "learning_rate": 5.594081944705496e-06, "loss": 0.12369145154953003, "step": 24900 }, { "epoch": 0.6561673740463545, "grad_norm": 2.9437167644500732, "learning_rate": 5.586424871356337e-06, "loss": 0.11994303464889526, "step": 24910 }, { "epoch": 0.6564307892908532, "grad_norm": 2.0829317569732666, "learning_rate": 5.578771010397572e-06, "loss": 0.12063679695129395, "step": 24920 }, { "epoch": 0.656694204535352, "grad_norm": 1.7094805240631104, "learning_rate": 5.571120367400008e-06, "loss": 0.11964969635009766, "step": 24930 }, { "epoch": 0.6569576197798507, "grad_norm": 5.042670726776123, "learning_rate": 5.563472947932123e-06, "loss": 0.1170515775680542, "step": 24940 }, { "epoch": 0.6572210350243495, "grad_norm": 3.0328636169433594, "learning_rate": 5.555828757560035e-06, "loss": 0.11796975135803223, "step": 24950 }, { "epoch": 0.6574844502688482, "grad_norm": 1.5710797309875488, "learning_rate": 5.548187801847513e-06, "loss": 0.11976811885833741, "step": 24960 }, { "epoch": 0.657747865513347, "grad_norm": 1.646046757698059, "learning_rate": 5.5405500863559795e-06, "loss": 0.11978057622909546, "step": 24970 }, { "epoch": 0.6580112807578457, "grad_norm": 3.7972490787506104, "learning_rate": 5.532915616644484e-06, "loss": 0.1175163745880127, "step": 24980 }, { "epoch": 0.6582746960023445, "grad_norm": 2.3794965744018555, "learning_rate": 5.525284398269733e-06, "loss": 0.11051833629608154, "step": 24990 }, { "epoch": 0.6585381112468431, "grad_norm": 2.0640201568603516, "learning_rate": 5.517656436786059e-06, "loss": 0.11815519332885742, "step": 25000 }, { "epoch": 0.6588015264913418, "grad_norm": 3.4546854496002197, "learning_rate": 5.510031737745405e-06, "loss": 0.11525294780731202, "step": 25010 }, { "epoch": 0.6590649417358406, "grad_norm": 2.054624557495117, "learning_rate": 5.50241030669737e-06, "loss": 0.11955959796905517, "step": 25020 }, { "epoch": 0.6593283569803393, "grad_norm": 2.0678956508636475, "learning_rate": 5.4947921491891565e-06, "loss": 0.12197389602661132, "step": 25030 }, { "epoch": 0.6595917722248381, "grad_norm": 1.4096959829330444, "learning_rate": 5.487177270765588e-06, "loss": 0.11492838859558105, "step": 25040 }, { "epoch": 0.6598551874693368, "grad_norm": 2.44714093208313, "learning_rate": 5.479565676969101e-06, "loss": 0.1135443925857544, "step": 25050 }, { "epoch": 0.6601186027138356, "grad_norm": 1.3787596225738525, "learning_rate": 5.47195737333974e-06, "loss": 0.11119561195373535, "step": 25060 }, { "epoch": 0.6603820179583343, "grad_norm": 1.121188759803772, "learning_rate": 5.464352365415162e-06, "loss": 0.11828372478485108, "step": 25070 }, { "epoch": 0.660645433202833, "grad_norm": 1.499832272529602, "learning_rate": 5.456750658730616e-06, "loss": 0.1207000732421875, "step": 25080 }, { "epoch": 0.6609088484473318, "grad_norm": 1.693893551826477, "learning_rate": 5.449152258818955e-06, "loss": 0.11478798389434815, "step": 25090 }, { "epoch": 0.6611722636918305, "grad_norm": 3.810499906539917, "learning_rate": 5.4415571712106184e-06, "loss": 0.11828222274780273, "step": 25100 }, { "epoch": 0.6614356789363293, "grad_norm": 3.4356253147125244, "learning_rate": 5.433965401433639e-06, "loss": 0.11260800361633301, "step": 25110 }, { "epoch": 0.661699094180828, "grad_norm": 2.2860991954803467, "learning_rate": 5.426376955013641e-06, "loss": 0.11896748542785644, "step": 25120 }, { "epoch": 0.6619625094253268, "grad_norm": 2.677682638168335, "learning_rate": 5.4187918374738154e-06, "loss": 0.12769637107849122, "step": 25130 }, { "epoch": 0.6622259246698254, "grad_norm": 1.6739890575408936, "learning_rate": 5.4112100543349365e-06, "loss": 0.10784530639648438, "step": 25140 }, { "epoch": 0.6624893399143242, "grad_norm": 3.2245097160339355, "learning_rate": 5.403631611115362e-06, "loss": 0.12139253616333008, "step": 25150 }, { "epoch": 0.6627527551588229, "grad_norm": 2.9010350704193115, "learning_rate": 5.396056513331005e-06, "loss": 0.11179491281509399, "step": 25160 }, { "epoch": 0.6630161704033216, "grad_norm": 2.414754629135132, "learning_rate": 5.388484766495348e-06, "loss": 0.1239896535873413, "step": 25170 }, { "epoch": 0.6632795856478204, "grad_norm": 2.468529462814331, "learning_rate": 5.3809163761194364e-06, "loss": 0.11825611591339111, "step": 25180 }, { "epoch": 0.6635430008923191, "grad_norm": 2.0856387615203857, "learning_rate": 5.373351347711865e-06, "loss": 0.10369453430175782, "step": 25190 }, { "epoch": 0.6638064161368179, "grad_norm": 3.5789151191711426, "learning_rate": 5.365789686778797e-06, "loss": 0.13340792655944825, "step": 25200 }, { "epoch": 0.6640698313813166, "grad_norm": 2.5787417888641357, "learning_rate": 5.358231398823937e-06, "loss": 0.10948007106781006, "step": 25210 }, { "epoch": 0.6643332466258154, "grad_norm": 2.334721088409424, "learning_rate": 5.350676489348521e-06, "loss": 0.11454999446868896, "step": 25220 }, { "epoch": 0.6645966618703141, "grad_norm": 2.722600221633911, "learning_rate": 5.3431249638513475e-06, "loss": 0.1262824058532715, "step": 25230 }, { "epoch": 0.6648600771148129, "grad_norm": 1.6340686082839966, "learning_rate": 5.335576827828737e-06, "loss": 0.1137197494506836, "step": 25240 }, { "epoch": 0.6651234923593116, "grad_norm": 3.8923168182373047, "learning_rate": 5.328032086774557e-06, "loss": 0.11118323802947998, "step": 25250 }, { "epoch": 0.6653869076038103, "grad_norm": 3.6737704277038574, "learning_rate": 5.320490746180189e-06, "loss": 0.11865394115447998, "step": 25260 }, { "epoch": 0.665650322848309, "grad_norm": 2.548515558242798, "learning_rate": 5.312952811534542e-06, "loss": 0.11637837886810302, "step": 25270 }, { "epoch": 0.6659137380928077, "grad_norm": 1.8531414270401, "learning_rate": 5.305418288324058e-06, "loss": 0.11143181324005128, "step": 25280 }, { "epoch": 0.6661771533373065, "grad_norm": 2.2816050052642822, "learning_rate": 5.297887182032686e-06, "loss": 0.12244322299957275, "step": 25290 }, { "epoch": 0.6664405685818052, "grad_norm": 2.2860357761383057, "learning_rate": 5.29035949814189e-06, "loss": 0.11146941184997558, "step": 25300 }, { "epoch": 0.666703983826304, "grad_norm": 2.51468825340271, "learning_rate": 5.2828352421306415e-06, "loss": 0.11611036062240601, "step": 25310 }, { "epoch": 0.6669673990708027, "grad_norm": 2.0146689414978027, "learning_rate": 5.275314419475417e-06, "loss": 0.10453264713287354, "step": 25320 }, { "epoch": 0.6672308143153015, "grad_norm": 2.3493971824645996, "learning_rate": 5.267797035650202e-06, "loss": 0.10969479084014892, "step": 25330 }, { "epoch": 0.6674942295598002, "grad_norm": 1.601420521736145, "learning_rate": 5.2602830961264704e-06, "loss": 0.12559876441955567, "step": 25340 }, { "epoch": 0.6677576448042989, "grad_norm": 2.262242317199707, "learning_rate": 5.252772606373186e-06, "loss": 0.10920019149780273, "step": 25350 }, { "epoch": 0.6680210600487977, "grad_norm": 1.4780393838882446, "learning_rate": 5.245265571856812e-06, "loss": 0.1069488763809204, "step": 25360 }, { "epoch": 0.6682844752932964, "grad_norm": 3.4813148975372314, "learning_rate": 5.237761998041291e-06, "loss": 0.1168360948562622, "step": 25370 }, { "epoch": 0.6685478905377952, "grad_norm": 1.571789264678955, "learning_rate": 5.230261890388053e-06, "loss": 0.12604223489761351, "step": 25380 }, { "epoch": 0.6688113057822939, "grad_norm": 1.537702202796936, "learning_rate": 5.2227652543559924e-06, "loss": 0.11230732202529907, "step": 25390 }, { "epoch": 0.6690747210267927, "grad_norm": 3.5204265117645264, "learning_rate": 5.215272095401483e-06, "loss": 0.11061277389526367, "step": 25400 }, { "epoch": 0.6693381362712914, "grad_norm": 1.4039283990859985, "learning_rate": 5.207782418978378e-06, "loss": 0.11109435558319092, "step": 25410 }, { "epoch": 0.6696015515157901, "grad_norm": 2.623809576034546, "learning_rate": 5.200296230537982e-06, "loss": 0.11381160020828247, "step": 25420 }, { "epoch": 0.6698649667602888, "grad_norm": 1.7297829389572144, "learning_rate": 5.192813535529066e-06, "loss": 0.12026016712188721, "step": 25430 }, { "epoch": 0.6701283820047875, "grad_norm": 1.966206669807434, "learning_rate": 5.1853343393978585e-06, "loss": 0.11382014751434326, "step": 25440 }, { "epoch": 0.6703917972492863, "grad_norm": 1.8303940296173096, "learning_rate": 5.177858647588037e-06, "loss": 0.11834148168563843, "step": 25450 }, { "epoch": 0.670655212493785, "grad_norm": 1.982202172279358, "learning_rate": 5.170386465540741e-06, "loss": 0.10382379293441772, "step": 25460 }, { "epoch": 0.6709186277382838, "grad_norm": 2.472532272338867, "learning_rate": 5.162917798694546e-06, "loss": 0.11100116968154908, "step": 25470 }, { "epoch": 0.6711820429827825, "grad_norm": 2.2825958728790283, "learning_rate": 5.15545265248546e-06, "loss": 0.12265999317169189, "step": 25480 }, { "epoch": 0.6714454582272813, "grad_norm": 1.4536925554275513, "learning_rate": 5.147991032346948e-06, "loss": 0.11486150026321411, "step": 25490 }, { "epoch": 0.67170887347178, "grad_norm": 2.1157212257385254, "learning_rate": 5.140532943709893e-06, "loss": 0.12735416889190673, "step": 25500 }, { "epoch": 0.6719722887162788, "grad_norm": 1.7859280109405518, "learning_rate": 5.133078392002623e-06, "loss": 0.114312744140625, "step": 25510 }, { "epoch": 0.6722357039607775, "grad_norm": 2.2118170261383057, "learning_rate": 5.125627382650873e-06, "loss": 0.113094162940979, "step": 25520 }, { "epoch": 0.6724991192052762, "grad_norm": 4.843816757202148, "learning_rate": 5.118179921077808e-06, "loss": 0.1236585259437561, "step": 25530 }, { "epoch": 0.672762534449775, "grad_norm": 2.3188164234161377, "learning_rate": 5.1107360127040206e-06, "loss": 0.1160537838935852, "step": 25540 }, { "epoch": 0.6730259496942737, "grad_norm": 2.706326484680176, "learning_rate": 5.103295662947504e-06, "loss": 0.1112107515335083, "step": 25550 }, { "epoch": 0.6732893649387724, "grad_norm": 4.728794574737549, "learning_rate": 5.095858877223665e-06, "loss": 0.1079744815826416, "step": 25560 }, { "epoch": 0.6735527801832711, "grad_norm": 1.78413724899292, "learning_rate": 5.088425660945319e-06, "loss": 0.11646134853363037, "step": 25570 }, { "epoch": 0.6738161954277699, "grad_norm": 1.8524399995803833, "learning_rate": 5.080996019522675e-06, "loss": 0.1064866065979004, "step": 25580 }, { "epoch": 0.6740796106722686, "grad_norm": 5.926259994506836, "learning_rate": 5.073569958363356e-06, "loss": 0.12080183029174804, "step": 25590 }, { "epoch": 0.6743430259167673, "grad_norm": 1.6905959844589233, "learning_rate": 5.06614748287237e-06, "loss": 0.10661547183990479, "step": 25600 }, { "epoch": 0.6746064411612661, "grad_norm": 3.5447640419006348, "learning_rate": 5.058728598452103e-06, "loss": 0.12492809295654297, "step": 25610 }, { "epoch": 0.6748698564057648, "grad_norm": 2.291562557220459, "learning_rate": 5.05131331050235e-06, "loss": 0.11009521484375, "step": 25620 }, { "epoch": 0.6751332716502636, "grad_norm": 1.708545446395874, "learning_rate": 5.043901624420273e-06, "loss": 0.1229319453239441, "step": 25630 }, { "epoch": 0.6753966868947623, "grad_norm": 3.0083673000335693, "learning_rate": 5.036493545600421e-06, "loss": 0.10803743600845336, "step": 25640 }, { "epoch": 0.6756601021392611, "grad_norm": 1.5470904111862183, "learning_rate": 5.029089079434708e-06, "loss": 0.10902447700500488, "step": 25650 }, { "epoch": 0.6759235173837598, "grad_norm": 2.77051043510437, "learning_rate": 5.021688231312423e-06, "loss": 0.10021008253097534, "step": 25660 }, { "epoch": 0.6761869326282586, "grad_norm": 1.8082724809646606, "learning_rate": 5.014291006620229e-06, "loss": 0.11955833435058594, "step": 25670 }, { "epoch": 0.6764503478727573, "grad_norm": 1.6793193817138672, "learning_rate": 5.006897410742142e-06, "loss": 0.11157927513122559, "step": 25680 }, { "epoch": 0.676713763117256, "grad_norm": 1.4958515167236328, "learning_rate": 4.999507449059542e-06, "loss": 0.11300013065338135, "step": 25690 }, { "epoch": 0.6769771783617548, "grad_norm": 1.837912917137146, "learning_rate": 4.992121126951159e-06, "loss": 0.11242778301239013, "step": 25700 }, { "epoch": 0.6772405936062534, "grad_norm": 2.3307759761810303, "learning_rate": 4.984738449793074e-06, "loss": 0.11034756898880005, "step": 25710 }, { "epoch": 0.6775040088507522, "grad_norm": 2.8678016662597656, "learning_rate": 4.977359422958731e-06, "loss": 0.11619025468826294, "step": 25720 }, { "epoch": 0.6777674240952509, "grad_norm": 2.0487396717071533, "learning_rate": 4.96998405181889e-06, "loss": 0.12367427349090576, "step": 25730 }, { "epoch": 0.6780308393397497, "grad_norm": 2.1244378089904785, "learning_rate": 4.962612341741666e-06, "loss": 0.0984379529953003, "step": 25740 }, { "epoch": 0.6782942545842484, "grad_norm": 1.9664230346679688, "learning_rate": 4.955244298092514e-06, "loss": 0.1153019905090332, "step": 25750 }, { "epoch": 0.6785576698287472, "grad_norm": 1.909746766090393, "learning_rate": 4.947879926234212e-06, "loss": 0.11907379627227783, "step": 25760 }, { "epoch": 0.6788210850732459, "grad_norm": 4.050475597381592, "learning_rate": 4.940519231526867e-06, "loss": 0.1173654317855835, "step": 25770 }, { "epoch": 0.6790845003177446, "grad_norm": 1.412060022354126, "learning_rate": 4.933162219327909e-06, "loss": 0.10771496295928955, "step": 25780 }, { "epoch": 0.6793479155622434, "grad_norm": 1.7718515396118164, "learning_rate": 4.925808894992087e-06, "loss": 0.10594402551651001, "step": 25790 }, { "epoch": 0.6796113308067421, "grad_norm": 1.5528955459594727, "learning_rate": 4.918459263871474e-06, "loss": 0.10989512205123901, "step": 25800 }, { "epoch": 0.6798747460512409, "grad_norm": 2.5067479610443115, "learning_rate": 4.911113331315448e-06, "loss": 0.10358650684356689, "step": 25810 }, { "epoch": 0.6801381612957396, "grad_norm": 4.807336807250977, "learning_rate": 4.903771102670686e-06, "loss": 0.11676914691925049, "step": 25820 }, { "epoch": 0.6804015765402384, "grad_norm": 2.0037999153137207, "learning_rate": 4.896432583281189e-06, "loss": 0.1092990756034851, "step": 25830 }, { "epoch": 0.680664991784737, "grad_norm": 1.5205477476119995, "learning_rate": 4.88909777848824e-06, "loss": 0.1083332657814026, "step": 25840 }, { "epoch": 0.6809284070292358, "grad_norm": 2.6204023361206055, "learning_rate": 4.881766693630439e-06, "loss": 0.12026776075363159, "step": 25850 }, { "epoch": 0.6811918222737345, "grad_norm": 2.6768360137939453, "learning_rate": 4.874439334043654e-06, "loss": 0.11223104000091552, "step": 25860 }, { "epoch": 0.6814552375182332, "grad_norm": 1.5643634796142578, "learning_rate": 4.8671157050610525e-06, "loss": 0.10896027088165283, "step": 25870 }, { "epoch": 0.681718652762732, "grad_norm": 2.106860399246216, "learning_rate": 4.859795812013099e-06, "loss": 0.11673811674118043, "step": 25880 }, { "epoch": 0.6819820680072307, "grad_norm": 2.124135732650757, "learning_rate": 4.852479660227518e-06, "loss": 0.10993530750274658, "step": 25890 }, { "epoch": 0.6822454832517295, "grad_norm": 3.1845624446868896, "learning_rate": 4.8451672550293236e-06, "loss": 0.10299911499023437, "step": 25900 }, { "epoch": 0.6825088984962282, "grad_norm": 3.7828867435455322, "learning_rate": 4.837858601740799e-06, "loss": 0.11351691484451294, "step": 25910 }, { "epoch": 0.682772313740727, "grad_norm": 2.5129075050354004, "learning_rate": 4.830553705681496e-06, "loss": 0.10441291332244873, "step": 25920 }, { "epoch": 0.6830357289852257, "grad_norm": 2.038973331451416, "learning_rate": 4.823252572168236e-06, "loss": 0.11174668073654175, "step": 25930 }, { "epoch": 0.6832991442297245, "grad_norm": 0.9718108177185059, "learning_rate": 4.815955206515104e-06, "loss": 0.11601905822753907, "step": 25940 }, { "epoch": 0.6835625594742232, "grad_norm": 3.588756799697876, "learning_rate": 4.808661614033422e-06, "loss": 0.10768392086029052, "step": 25950 }, { "epoch": 0.6838259747187219, "grad_norm": 1.3711668252944946, "learning_rate": 4.801371800031796e-06, "loss": 0.11300935745239257, "step": 25960 }, { "epoch": 0.6840893899632207, "grad_norm": 2.4680328369140625, "learning_rate": 4.794085769816057e-06, "loss": 0.09444608688354492, "step": 25970 }, { "epoch": 0.6843528052077194, "grad_norm": 1.6629494428634644, "learning_rate": 4.786803528689304e-06, "loss": 0.12011275291442872, "step": 25980 }, { "epoch": 0.6846162204522181, "grad_norm": 3.761106491088867, "learning_rate": 4.779525081951857e-06, "loss": 0.1185636043548584, "step": 25990 }, { "epoch": 0.6848796356967168, "grad_norm": 2.6102402210235596, "learning_rate": 4.772250434901281e-06, "loss": 0.1128774642944336, "step": 26000 }, { "epoch": 0.6851430509412156, "grad_norm": 3.2058799266815186, "learning_rate": 4.7649795928323875e-06, "loss": 0.10687670707702637, "step": 26010 }, { "epoch": 0.6854064661857143, "grad_norm": 1.8319517374038696, "learning_rate": 4.757712561037203e-06, "loss": 0.1098482370376587, "step": 26020 }, { "epoch": 0.6856698814302131, "grad_norm": 4.6830267906188965, "learning_rate": 4.750449344804989e-06, "loss": 0.1078522801399231, "step": 26030 }, { "epoch": 0.6859332966747118, "grad_norm": 6.519756317138672, "learning_rate": 4.743189949422227e-06, "loss": 0.1347496509552002, "step": 26040 }, { "epoch": 0.6861967119192105, "grad_norm": 1.6672495603561401, "learning_rate": 4.735934380172613e-06, "loss": 0.10709600448608399, "step": 26050 }, { "epoch": 0.6864601271637093, "grad_norm": 2.285676956176758, "learning_rate": 4.728682642337072e-06, "loss": 0.11508550643920898, "step": 26060 }, { "epoch": 0.686723542408208, "grad_norm": 1.8676486015319824, "learning_rate": 4.7214347411937325e-06, "loss": 0.11443108320236206, "step": 26070 }, { "epoch": 0.6869869576527068, "grad_norm": 1.7470051050186157, "learning_rate": 4.714190682017918e-06, "loss": 0.1116481065750122, "step": 26080 }, { "epoch": 0.6872503728972055, "grad_norm": 1.8507020473480225, "learning_rate": 4.706950470082177e-06, "loss": 0.11057091951370239, "step": 26090 }, { "epoch": 0.6875137881417043, "grad_norm": 2.7176599502563477, "learning_rate": 4.699714110656242e-06, "loss": 0.10336344242095948, "step": 26100 }, { "epoch": 0.687777203386203, "grad_norm": 1.3760986328125, "learning_rate": 4.6924816090070575e-06, "loss": 0.10731718540191651, "step": 26110 }, { "epoch": 0.6880406186307017, "grad_norm": 2.9415032863616943, "learning_rate": 4.685252970398742e-06, "loss": 0.12132539749145507, "step": 26120 }, { "epoch": 0.6883040338752004, "grad_norm": 1.7272549867630005, "learning_rate": 4.678028200092607e-06, "loss": 0.10890870094299317, "step": 26130 }, { "epoch": 0.6885674491196991, "grad_norm": 1.6030473709106445, "learning_rate": 4.670807303347161e-06, "loss": 0.11088777780532837, "step": 26140 }, { "epoch": 0.6888308643641979, "grad_norm": 5.130202293395996, "learning_rate": 4.663590285418082e-06, "loss": 0.10697674751281738, "step": 26150 }, { "epoch": 0.6890942796086966, "grad_norm": 3.70847487449646, "learning_rate": 4.656377151558224e-06, "loss": 0.09501540660858154, "step": 26160 }, { "epoch": 0.6893576948531954, "grad_norm": 2.608171224594116, "learning_rate": 4.64916790701762e-06, "loss": 0.12284754514694214, "step": 26170 }, { "epoch": 0.6896211100976941, "grad_norm": 2.6243324279785156, "learning_rate": 4.641962557043462e-06, "loss": 0.1071729063987732, "step": 26180 }, { "epoch": 0.6898845253421929, "grad_norm": 2.36163067817688, "learning_rate": 4.634761106880127e-06, "loss": 0.10428638458251953, "step": 26190 }, { "epoch": 0.6901479405866916, "grad_norm": 2.6709790229797363, "learning_rate": 4.627563561769136e-06, "loss": 0.10762346982955932, "step": 26200 }, { "epoch": 0.6904113558311903, "grad_norm": 2.286667585372925, "learning_rate": 4.620369926949164e-06, "loss": 0.1008458137512207, "step": 26210 }, { "epoch": 0.6906747710756891, "grad_norm": 1.8623954057693481, "learning_rate": 4.613180207656061e-06, "loss": 0.09718794226646424, "step": 26220 }, { "epoch": 0.6909381863201878, "grad_norm": 1.6280382871627808, "learning_rate": 4.60599440912281e-06, "loss": 0.11523665189743042, "step": 26230 }, { "epoch": 0.6912016015646866, "grad_norm": 3.003394842147827, "learning_rate": 4.598812536579545e-06, "loss": 0.10372766256332397, "step": 26240 }, { "epoch": 0.6914650168091853, "grad_norm": 1.688461184501648, "learning_rate": 4.591634595253541e-06, "loss": 0.09567297101020814, "step": 26250 }, { "epoch": 0.6917284320536841, "grad_norm": 1.673598289489746, "learning_rate": 4.584460590369214e-06, "loss": 0.1156511902809143, "step": 26260 }, { "epoch": 0.6919918472981827, "grad_norm": 2.8600656986236572, "learning_rate": 4.577290527148118e-06, "loss": 0.10903410911560059, "step": 26270 }, { "epoch": 0.6922552625426815, "grad_norm": 2.0687880516052246, "learning_rate": 4.5701244108089325e-06, "loss": 0.11387733221054078, "step": 26280 }, { "epoch": 0.6925186777871802, "grad_norm": 1.8671419620513916, "learning_rate": 4.562962246567466e-06, "loss": 0.12150198221206665, "step": 26290 }, { "epoch": 0.6927820930316789, "grad_norm": 3.040314197540283, "learning_rate": 4.555804039636649e-06, "loss": 0.11843760013580322, "step": 26300 }, { "epoch": 0.6930455082761777, "grad_norm": 3.6089799404144287, "learning_rate": 4.548649795226532e-06, "loss": 0.10564470291137695, "step": 26310 }, { "epoch": 0.6933089235206764, "grad_norm": 1.751103162765503, "learning_rate": 4.541499518544292e-06, "loss": 0.11270220279693603, "step": 26320 }, { "epoch": 0.6935723387651752, "grad_norm": 2.406740427017212, "learning_rate": 4.5343532147941985e-06, "loss": 0.10529781579971313, "step": 26330 }, { "epoch": 0.6938357540096739, "grad_norm": 1.8397706747055054, "learning_rate": 4.527210889177639e-06, "loss": 0.11629277467727661, "step": 26340 }, { "epoch": 0.6940991692541727, "grad_norm": 3.705477476119995, "learning_rate": 4.520072546893115e-06, "loss": 0.12390015125274659, "step": 26350 }, { "epoch": 0.6943625844986714, "grad_norm": 1.538050889968872, "learning_rate": 4.512938193136216e-06, "loss": 0.10387393236160278, "step": 26360 }, { "epoch": 0.6946259997431702, "grad_norm": 1.4947136640548706, "learning_rate": 4.5058078330996315e-06, "loss": 0.10245908498764038, "step": 26370 }, { "epoch": 0.6948894149876689, "grad_norm": 2.4869723320007324, "learning_rate": 4.498681471973145e-06, "loss": 0.1054995059967041, "step": 26380 }, { "epoch": 0.6951528302321676, "grad_norm": 2.6159348487854004, "learning_rate": 4.491559114943628e-06, "loss": 0.10261356830596924, "step": 26390 }, { "epoch": 0.6954162454766664, "grad_norm": 2.2148749828338623, "learning_rate": 4.484440767195044e-06, "loss": 0.11877810955047607, "step": 26400 }, { "epoch": 0.695679660721165, "grad_norm": 2.109316825866699, "learning_rate": 4.477326433908435e-06, "loss": 0.1141460657119751, "step": 26410 }, { "epoch": 0.6959430759656638, "grad_norm": 1.1671078205108643, "learning_rate": 4.4702161202619085e-06, "loss": 0.11018766164779663, "step": 26420 }, { "epoch": 0.6962064912101625, "grad_norm": 1.7567870616912842, "learning_rate": 4.463109831430668e-06, "loss": 0.10249439477920533, "step": 26430 }, { "epoch": 0.6964699064546613, "grad_norm": 2.5082521438598633, "learning_rate": 4.456007572586971e-06, "loss": 0.10876874923706055, "step": 26440 }, { "epoch": 0.69673332169916, "grad_norm": 2.6851418018341064, "learning_rate": 4.448909348900159e-06, "loss": 0.10283267498016357, "step": 26450 }, { "epoch": 0.6969967369436588, "grad_norm": 2.741248846054077, "learning_rate": 4.441815165536612e-06, "loss": 0.12708450555801393, "step": 26460 }, { "epoch": 0.6972601521881575, "grad_norm": 2.036430597305298, "learning_rate": 4.434725027659786e-06, "loss": 0.10831692218780517, "step": 26470 }, { "epoch": 0.6975235674326562, "grad_norm": 2.6070327758789062, "learning_rate": 4.427638940430192e-06, "loss": 0.10359854698181152, "step": 26480 }, { "epoch": 0.697786982677155, "grad_norm": 2.603768825531006, "learning_rate": 4.420556909005392e-06, "loss": 0.10942862033843995, "step": 26490 }, { "epoch": 0.6980503979216537, "grad_norm": 3.4628260135650635, "learning_rate": 4.41347893853999e-06, "loss": 0.11875522136688232, "step": 26500 }, { "epoch": 0.6983138131661525, "grad_norm": 3.7588295936584473, "learning_rate": 4.40640503418564e-06, "loss": 0.09560293555259705, "step": 26510 }, { "epoch": 0.6985772284106512, "grad_norm": 1.5322169065475464, "learning_rate": 4.39933520109103e-06, "loss": 0.10003459453582764, "step": 26520 }, { "epoch": 0.69884064365515, "grad_norm": 4.34731388092041, "learning_rate": 4.3922694444018985e-06, "loss": 0.1100502371788025, "step": 26530 }, { "epoch": 0.6991040588996487, "grad_norm": 2.6584513187408447, "learning_rate": 4.385207769261007e-06, "loss": 0.10541074275970459, "step": 26540 }, { "epoch": 0.6993674741441475, "grad_norm": 3.451059103012085, "learning_rate": 4.378150180808136e-06, "loss": 0.1118274450302124, "step": 26550 }, { "epoch": 0.6996308893886461, "grad_norm": 2.923142671585083, "learning_rate": 4.371096684180114e-06, "loss": 0.10897352695465087, "step": 26560 }, { "epoch": 0.6998943046331448, "grad_norm": 2.1459457874298096, "learning_rate": 4.364047284510774e-06, "loss": 0.08762995004653931, "step": 26570 }, { "epoch": 0.7001577198776436, "grad_norm": 1.6761478185653687, "learning_rate": 4.357001986930982e-06, "loss": 0.10137320756912231, "step": 26580 }, { "epoch": 0.7004211351221423, "grad_norm": 3.9369893074035645, "learning_rate": 4.349960796568601e-06, "loss": 0.10692728757858276, "step": 26590 }, { "epoch": 0.7006845503666411, "grad_norm": 2.7496981620788574, "learning_rate": 4.342923718548512e-06, "loss": 0.11157698631286621, "step": 26600 }, { "epoch": 0.7009479656111398, "grad_norm": 1.826000452041626, "learning_rate": 4.335890757992611e-06, "loss": 0.10309978723526, "step": 26610 }, { "epoch": 0.7012113808556386, "grad_norm": 2.197232723236084, "learning_rate": 4.328861920019787e-06, "loss": 0.09792444109916687, "step": 26620 }, { "epoch": 0.7014747961001373, "grad_norm": 4.093085289001465, "learning_rate": 4.321837209745931e-06, "loss": 0.10706264972686767, "step": 26630 }, { "epoch": 0.701738211344636, "grad_norm": 1.798120141029358, "learning_rate": 4.3148166322839305e-06, "loss": 0.10322338342666626, "step": 26640 }, { "epoch": 0.7020016265891348, "grad_norm": 4.006990432739258, "learning_rate": 4.3078001927436605e-06, "loss": 0.09989851713180542, "step": 26650 }, { "epoch": 0.7022650418336335, "grad_norm": 2.0668838024139404, "learning_rate": 4.300787896231997e-06, "loss": 0.10585267543792724, "step": 26660 }, { "epoch": 0.7025284570781323, "grad_norm": 3.031540632247925, "learning_rate": 4.293779747852791e-06, "loss": 0.1073683500289917, "step": 26670 }, { "epoch": 0.702791872322631, "grad_norm": 3.630627155303955, "learning_rate": 4.286775752706865e-06, "loss": 0.099798583984375, "step": 26680 }, { "epoch": 0.7030552875671298, "grad_norm": 3.397545337677002, "learning_rate": 4.279775915892038e-06, "loss": 0.1158372163772583, "step": 26690 }, { "epoch": 0.7033187028116284, "grad_norm": 1.7300623655319214, "learning_rate": 4.2727802425030885e-06, "loss": 0.09644834995269776, "step": 26700 }, { "epoch": 0.7035821180561272, "grad_norm": 1.8725776672363281, "learning_rate": 4.265788737631779e-06, "loss": 0.09923970699310303, "step": 26710 }, { "epoch": 0.7038455333006259, "grad_norm": 1.833725929260254, "learning_rate": 4.258801406366817e-06, "loss": 0.1120455265045166, "step": 26720 }, { "epoch": 0.7041089485451246, "grad_norm": 4.407054424285889, "learning_rate": 4.251818253793885e-06, "loss": 0.11577441692352294, "step": 26730 }, { "epoch": 0.7043723637896234, "grad_norm": 2.104567289352417, "learning_rate": 4.244839284995627e-06, "loss": 0.10928330421447754, "step": 26740 }, { "epoch": 0.7046357790341221, "grad_norm": 2.766056776046753, "learning_rate": 4.237864505051638e-06, "loss": 0.11325064897537232, "step": 26750 }, { "epoch": 0.7048991942786209, "grad_norm": 1.9505723714828491, "learning_rate": 4.230893919038461e-06, "loss": 0.11502447128295898, "step": 26760 }, { "epoch": 0.7051626095231196, "grad_norm": 1.837382435798645, "learning_rate": 4.223927532029592e-06, "loss": 0.10040793418884278, "step": 26770 }, { "epoch": 0.7054260247676184, "grad_norm": 3.091242551803589, "learning_rate": 4.21696534909546e-06, "loss": 0.11493597030639649, "step": 26780 }, { "epoch": 0.7056894400121171, "grad_norm": 1.8446335792541504, "learning_rate": 4.2100073753034535e-06, "loss": 0.10170681476593017, "step": 26790 }, { "epoch": 0.7059528552566159, "grad_norm": 2.354661703109741, "learning_rate": 4.203053615717884e-06, "loss": 0.10344594717025757, "step": 26800 }, { "epoch": 0.7062162705011146, "grad_norm": 1.8727854490280151, "learning_rate": 4.196104075399986e-06, "loss": 0.11436095237731933, "step": 26810 }, { "epoch": 0.7064796857456133, "grad_norm": 2.0742132663726807, "learning_rate": 4.1891587594079485e-06, "loss": 0.12265142202377319, "step": 26820 }, { "epoch": 0.7067431009901121, "grad_norm": 2.761561632156372, "learning_rate": 4.182217672796867e-06, "loss": 0.11507759094238282, "step": 26830 }, { "epoch": 0.7070065162346107, "grad_norm": 1.775908350944519, "learning_rate": 4.175280820618763e-06, "loss": 0.10333763360977173, "step": 26840 }, { "epoch": 0.7072699314791095, "grad_norm": 3.362190008163452, "learning_rate": 4.168348207922577e-06, "loss": 0.09132924675941467, "step": 26850 }, { "epoch": 0.7075333467236082, "grad_norm": 1.9238423109054565, "learning_rate": 4.16141983975416e-06, "loss": 0.11037780046463012, "step": 26860 }, { "epoch": 0.707796761968107, "grad_norm": 2.958519697189331, "learning_rate": 4.154495721156287e-06, "loss": 0.10638842582702637, "step": 26870 }, { "epoch": 0.7080601772126057, "grad_norm": 3.244874954223633, "learning_rate": 4.147575857168623e-06, "loss": 0.101504647731781, "step": 26880 }, { "epoch": 0.7083235924571045, "grad_norm": 1.9491446018218994, "learning_rate": 4.140660252827746e-06, "loss": 0.09535064697265624, "step": 26890 }, { "epoch": 0.7085870077016032, "grad_norm": 3.9388427734375, "learning_rate": 4.13374891316713e-06, "loss": 0.09107176065444947, "step": 26900 }, { "epoch": 0.7088504229461019, "grad_norm": 1.7729077339172363, "learning_rate": 4.126841843217144e-06, "loss": 0.10465823411941529, "step": 26910 }, { "epoch": 0.7091138381906007, "grad_norm": 3.9375319480895996, "learning_rate": 4.1199390480050605e-06, "loss": 0.11019625663757324, "step": 26920 }, { "epoch": 0.7093772534350994, "grad_norm": 2.8095226287841797, "learning_rate": 4.113040532555021e-06, "loss": 0.10564770698547363, "step": 26930 }, { "epoch": 0.7096406686795982, "grad_norm": 2.2378504276275635, "learning_rate": 4.106146301888065e-06, "loss": 0.10676084756851197, "step": 26940 }, { "epoch": 0.7099040839240969, "grad_norm": 2.1854135990142822, "learning_rate": 4.0992563610221155e-06, "loss": 0.1032336950302124, "step": 26950 }, { "epoch": 0.7101674991685957, "grad_norm": 1.9057834148406982, "learning_rate": 4.092370714971966e-06, "loss": 0.11042447090148926, "step": 26960 }, { "epoch": 0.7104309144130944, "grad_norm": 2.7837133407592773, "learning_rate": 4.085489368749287e-06, "loss": 0.11235288381576539, "step": 26970 }, { "epoch": 0.7106943296575932, "grad_norm": 2.0422568321228027, "learning_rate": 4.07861232736262e-06, "loss": 0.11118950843811035, "step": 26980 }, { "epoch": 0.7109577449020918, "grad_norm": 4.611517906188965, "learning_rate": 4.0717395958173665e-06, "loss": 0.10482017993927002, "step": 26990 }, { "epoch": 0.7112211601465905, "grad_norm": 4.5284833908081055, "learning_rate": 4.064871179115807e-06, "loss": 0.10067911148071289, "step": 27000 }, { "epoch": 0.7114845753910893, "grad_norm": 1.985154628753662, "learning_rate": 4.0580070822570694e-06, "loss": 0.0989345908164978, "step": 27010 }, { "epoch": 0.711747990635588, "grad_norm": 2.2363314628601074, "learning_rate": 4.051147310237129e-06, "loss": 0.09981286525726318, "step": 27020 }, { "epoch": 0.7120114058800868, "grad_norm": 1.8462885618209839, "learning_rate": 4.044291868048837e-06, "loss": 0.09835363030433655, "step": 27030 }, { "epoch": 0.7122748211245855, "grad_norm": 2.0583853721618652, "learning_rate": 4.0374407606818684e-06, "loss": 0.11059472560882569, "step": 27040 }, { "epoch": 0.7125382363690843, "grad_norm": 3.8702096939086914, "learning_rate": 4.03059399312277e-06, "loss": 0.12388089895248414, "step": 27050 }, { "epoch": 0.712801651613583, "grad_norm": 2.411036252975464, "learning_rate": 4.0237515703549015e-06, "loss": 0.10318982601165771, "step": 27060 }, { "epoch": 0.7130650668580818, "grad_norm": 1.50007164478302, "learning_rate": 4.0169134973584726e-06, "loss": 0.0943223237991333, "step": 27070 }, { "epoch": 0.7133284821025805, "grad_norm": 1.5609984397888184, "learning_rate": 4.010079779110537e-06, "loss": 0.10512318611145019, "step": 27080 }, { "epoch": 0.7135918973470792, "grad_norm": 1.7264964580535889, "learning_rate": 4.003250420584966e-06, "loss": 0.11352152824401855, "step": 27090 }, { "epoch": 0.713855312591578, "grad_norm": 2.377685546875, "learning_rate": 3.996425426752459e-06, "loss": 0.11565855741500855, "step": 27100 }, { "epoch": 0.7141187278360767, "grad_norm": 2.1571733951568604, "learning_rate": 3.989604802580545e-06, "loss": 0.11022926568984985, "step": 27110 }, { "epoch": 0.7143821430805755, "grad_norm": 1.8727221488952637, "learning_rate": 3.982788553033561e-06, "loss": 0.1014124870300293, "step": 27120 }, { "epoch": 0.7146455583250741, "grad_norm": 2.207387924194336, "learning_rate": 3.975976683072676e-06, "loss": 0.10558619499206542, "step": 27130 }, { "epoch": 0.714908973569573, "grad_norm": 3.0038630962371826, "learning_rate": 3.9691691976558636e-06, "loss": 0.11020160913467407, "step": 27140 }, { "epoch": 0.7151723888140716, "grad_norm": 2.1746699810028076, "learning_rate": 3.9623661017378946e-06, "loss": 0.10838826894760131, "step": 27150 }, { "epoch": 0.7154358040585704, "grad_norm": 2.5181069374084473, "learning_rate": 3.955567400270365e-06, "loss": 0.11167012453079224, "step": 27160 }, { "epoch": 0.7156992193030691, "grad_norm": 2.7249135971069336, "learning_rate": 3.948773098201657e-06, "loss": 0.09874458312988281, "step": 27170 }, { "epoch": 0.7159626345475678, "grad_norm": 2.7252395153045654, "learning_rate": 3.941983200476969e-06, "loss": 0.10429879426956176, "step": 27180 }, { "epoch": 0.7162260497920666, "grad_norm": 2.5084426403045654, "learning_rate": 3.935197712038267e-06, "loss": 0.10067574977874756, "step": 27190 }, { "epoch": 0.7164894650365653, "grad_norm": 1.6001451015472412, "learning_rate": 3.928416637824326e-06, "loss": 0.09831972122192383, "step": 27200 }, { "epoch": 0.7167528802810641, "grad_norm": 2.8537638187408447, "learning_rate": 3.921639982770709e-06, "loss": 0.10756325721740723, "step": 27210 }, { "epoch": 0.7170162955255628, "grad_norm": 1.7120542526245117, "learning_rate": 3.914867751809756e-06, "loss": 0.10527528524398803, "step": 27220 }, { "epoch": 0.7172797107700616, "grad_norm": 3.1364378929138184, "learning_rate": 3.908099949870588e-06, "loss": 0.10279562473297119, "step": 27230 }, { "epoch": 0.7175431260145603, "grad_norm": 1.9472075700759888, "learning_rate": 3.901336581879102e-06, "loss": 0.09489362239837647, "step": 27240 }, { "epoch": 0.717806541259059, "grad_norm": 1.9846357107162476, "learning_rate": 3.894577652757968e-06, "loss": 0.11159543991088867, "step": 27250 }, { "epoch": 0.7180699565035578, "grad_norm": 2.0579304695129395, "learning_rate": 3.887823167426633e-06, "loss": 0.10172348022460938, "step": 27260 }, { "epoch": 0.7183333717480564, "grad_norm": 1.523954153060913, "learning_rate": 3.8810731308013005e-06, "loss": 0.10222802162170411, "step": 27270 }, { "epoch": 0.7185967869925552, "grad_norm": 1.935898780822754, "learning_rate": 3.874327547794931e-06, "loss": 0.10258688926696777, "step": 27280 }, { "epoch": 0.7188602022370539, "grad_norm": 2.284811019897461, "learning_rate": 3.8675864233172575e-06, "loss": 0.09862067699432372, "step": 27290 }, { "epoch": 0.7191236174815527, "grad_norm": 1.654184341430664, "learning_rate": 3.860849762274759e-06, "loss": 0.09131355881690979, "step": 27300 }, { "epoch": 0.7193870327260514, "grad_norm": 2.1364238262176514, "learning_rate": 3.854117569570674e-06, "loss": 0.11025781631469726, "step": 27310 }, { "epoch": 0.7196504479705502, "grad_norm": 2.0328431129455566, "learning_rate": 3.8473898501049745e-06, "loss": 0.10525705814361572, "step": 27320 }, { "epoch": 0.7199138632150489, "grad_norm": 2.0853655338287354, "learning_rate": 3.840666608774385e-06, "loss": 0.09704537391662597, "step": 27330 }, { "epoch": 0.7201772784595476, "grad_norm": 3.0440421104431152, "learning_rate": 3.833947850472378e-06, "loss": 0.09809238314628602, "step": 27340 }, { "epoch": 0.7204406937040464, "grad_norm": 3.270444631576538, "learning_rate": 3.827233580089152e-06, "loss": 0.12020424604415894, "step": 27350 }, { "epoch": 0.7207041089485451, "grad_norm": 3.136650323867798, "learning_rate": 3.82052380251164e-06, "loss": 0.10439175367355347, "step": 27360 }, { "epoch": 0.7209675241930439, "grad_norm": 1.978072166442871, "learning_rate": 3.8138185226235114e-06, "loss": 0.09824099540710449, "step": 27370 }, { "epoch": 0.7212309394375426, "grad_norm": 3.2845752239227295, "learning_rate": 3.8071177453051523e-06, "loss": 0.11362314224243164, "step": 27380 }, { "epoch": 0.7214943546820414, "grad_norm": 3.798970937728882, "learning_rate": 3.8004214754336853e-06, "loss": 0.11389795541763306, "step": 27390 }, { "epoch": 0.7217577699265401, "grad_norm": 2.177180051803589, "learning_rate": 3.7937297178829458e-06, "loss": 0.09834338426589966, "step": 27400 }, { "epoch": 0.7220211851710389, "grad_norm": 2.810804843902588, "learning_rate": 3.7870424775234703e-06, "loss": 0.09848037958145142, "step": 27410 }, { "epoch": 0.7222846004155375, "grad_norm": 1.9583642482757568, "learning_rate": 3.780359759222534e-06, "loss": 0.11349470615386963, "step": 27420 }, { "epoch": 0.7225480156600362, "grad_norm": 2.1705527305603027, "learning_rate": 3.773681567844103e-06, "loss": 0.10757471323013305, "step": 27430 }, { "epoch": 0.722811430904535, "grad_norm": 1.948034644126892, "learning_rate": 3.767007908248852e-06, "loss": 0.10248229503631592, "step": 27440 }, { "epoch": 0.7230748461490337, "grad_norm": 2.803314447402954, "learning_rate": 3.7603387852941608e-06, "loss": 0.09176271557807922, "step": 27450 }, { "epoch": 0.7233382613935325, "grad_norm": 3.1383426189422607, "learning_rate": 3.753674203834099e-06, "loss": 0.09791352748870849, "step": 27460 }, { "epoch": 0.7236016766380312, "grad_norm": 4.494500637054443, "learning_rate": 3.747014168719445e-06, "loss": 0.10006207227706909, "step": 27470 }, { "epoch": 0.72386509188253, "grad_norm": 2.6118383407592773, "learning_rate": 3.740358684797656e-06, "loss": 0.10187606811523438, "step": 27480 }, { "epoch": 0.7241285071270287, "grad_norm": 1.806121826171875, "learning_rate": 3.7337077569128797e-06, "loss": 0.09872116446495056, "step": 27490 }, { "epoch": 0.7243919223715275, "grad_norm": 2.505248785018921, "learning_rate": 3.7270613899059495e-06, "loss": 0.09095984697341919, "step": 27500 }, { "epoch": 0.7246553376160262, "grad_norm": 2.1066105365753174, "learning_rate": 3.7204195886143736e-06, "loss": 0.09599379301071168, "step": 27510 }, { "epoch": 0.7249187528605249, "grad_norm": 2.559454917907715, "learning_rate": 3.713782357872352e-06, "loss": 0.09213841557502747, "step": 27520 }, { "epoch": 0.7251821681050237, "grad_norm": 3.342690944671631, "learning_rate": 3.7071497025107373e-06, "loss": 0.10784438848495484, "step": 27530 }, { "epoch": 0.7254455833495224, "grad_norm": 1.7525923252105713, "learning_rate": 3.7005216273570623e-06, "loss": 0.09429584741592408, "step": 27540 }, { "epoch": 0.7257089985940212, "grad_norm": 2.3867366313934326, "learning_rate": 3.693898137235532e-06, "loss": 0.09777287244796753, "step": 27550 }, { "epoch": 0.7259724138385198, "grad_norm": 2.678814172744751, "learning_rate": 3.6872792369670074e-06, "loss": 0.10305534601211548, "step": 27560 }, { "epoch": 0.7262358290830186, "grad_norm": 1.735451579093933, "learning_rate": 3.6806649313690066e-06, "loss": 0.1044346809387207, "step": 27570 }, { "epoch": 0.7264992443275173, "grad_norm": 1.8942759037017822, "learning_rate": 3.6740552252557093e-06, "loss": 0.10093930959701539, "step": 27580 }, { "epoch": 0.7267626595720161, "grad_norm": 1.7637079954147339, "learning_rate": 3.667450123437939e-06, "loss": 0.09987549185752868, "step": 27590 }, { "epoch": 0.7270260748165148, "grad_norm": 2.6085503101348877, "learning_rate": 3.6608496307231823e-06, "loss": 0.09890617728233338, "step": 27600 }, { "epoch": 0.7272894900610135, "grad_norm": 2.3406713008880615, "learning_rate": 3.6542537519155617e-06, "loss": 0.0895750105381012, "step": 27610 }, { "epoch": 0.7275529053055123, "grad_norm": 2.482203960418701, "learning_rate": 3.647662491815833e-06, "loss": 0.1050060510635376, "step": 27620 }, { "epoch": 0.727816320550011, "grad_norm": 2.5373268127441406, "learning_rate": 3.641075855221409e-06, "loss": 0.10911686420440674, "step": 27630 }, { "epoch": 0.7280797357945098, "grad_norm": 2.0523602962493896, "learning_rate": 3.634493846926321e-06, "loss": 0.11191375255584717, "step": 27640 }, { "epoch": 0.7283431510390085, "grad_norm": 2.2569093704223633, "learning_rate": 3.62791647172125e-06, "loss": 0.10336625576019287, "step": 27650 }, { "epoch": 0.7286065662835073, "grad_norm": 4.238804340362549, "learning_rate": 3.621343734393482e-06, "loss": 0.10115303993225097, "step": 27660 }, { "epoch": 0.728869981528006, "grad_norm": 2.548898696899414, "learning_rate": 3.6147756397269397e-06, "loss": 0.10967919826507569, "step": 27670 }, { "epoch": 0.7291333967725048, "grad_norm": 2.7172608375549316, "learning_rate": 3.6082121925021717e-06, "loss": 0.10179730653762817, "step": 27680 }, { "epoch": 0.7293968120170035, "grad_norm": 1.7818399667739868, "learning_rate": 3.601653397496335e-06, "loss": 0.0921276867389679, "step": 27690 }, { "epoch": 0.7296602272615021, "grad_norm": 2.3252193927764893, "learning_rate": 3.5950992594832044e-06, "loss": 0.11031739711761475, "step": 27700 }, { "epoch": 0.729923642506001, "grad_norm": 1.746782898902893, "learning_rate": 3.588549783233164e-06, "loss": 0.09344148635864258, "step": 27710 }, { "epoch": 0.7301870577504996, "grad_norm": 2.782689094543457, "learning_rate": 3.582004973513201e-06, "loss": 0.10025155544281006, "step": 27720 }, { "epoch": 0.7304504729949984, "grad_norm": 3.817610025405884, "learning_rate": 3.575464835086917e-06, "loss": 0.11272754669189453, "step": 27730 }, { "epoch": 0.7307138882394971, "grad_norm": 5.085299968719482, "learning_rate": 3.568929372714508e-06, "loss": 0.1231967568397522, "step": 27740 }, { "epoch": 0.7309773034839959, "grad_norm": 4.17976188659668, "learning_rate": 3.562398591152755e-06, "loss": 0.10660080909729004, "step": 27750 }, { "epoch": 0.7312407187284946, "grad_norm": 2.008932590484619, "learning_rate": 3.5558724951550515e-06, "loss": 0.09929621815681458, "step": 27760 }, { "epoch": 0.7315041339729933, "grad_norm": 1.757002353668213, "learning_rate": 3.549351089471367e-06, "loss": 0.09614676237106323, "step": 27770 }, { "epoch": 0.7317675492174921, "grad_norm": 2.830120325088501, "learning_rate": 3.54283437884827e-06, "loss": 0.10072418451309204, "step": 27780 }, { "epoch": 0.7320309644619908, "grad_norm": 2.2668919563293457, "learning_rate": 3.536322368028895e-06, "loss": 0.11051071882247925, "step": 27790 }, { "epoch": 0.7322943797064896, "grad_norm": 2.5806825160980225, "learning_rate": 3.5298150617529635e-06, "loss": 0.10498062372207642, "step": 27800 }, { "epoch": 0.7325577949509883, "grad_norm": 2.299799919128418, "learning_rate": 3.5233124647567817e-06, "loss": 0.10894465446472168, "step": 27810 }, { "epoch": 0.7328212101954871, "grad_norm": 1.9912693500518799, "learning_rate": 3.516814581773216e-06, "loss": 0.09823567271232606, "step": 27820 }, { "epoch": 0.7330846254399858, "grad_norm": 1.7839984893798828, "learning_rate": 3.5103214175317056e-06, "loss": 0.10189203023910523, "step": 27830 }, { "epoch": 0.7333480406844846, "grad_norm": 2.1594011783599854, "learning_rate": 3.5038329767582567e-06, "loss": 0.1066053867340088, "step": 27840 }, { "epoch": 0.7336114559289832, "grad_norm": 2.3731753826141357, "learning_rate": 3.4973492641754315e-06, "loss": 0.10183262825012207, "step": 27850 }, { "epoch": 0.7338748711734819, "grad_norm": 2.133366823196411, "learning_rate": 3.4908702845023643e-06, "loss": 0.0982647955417633, "step": 27860 }, { "epoch": 0.7341382864179807, "grad_norm": 2.1219191551208496, "learning_rate": 3.484396042454734e-06, "loss": 0.10256855487823487, "step": 27870 }, { "epoch": 0.7344017016624794, "grad_norm": 2.467501640319824, "learning_rate": 3.477926542744766e-06, "loss": 0.10075763463974, "step": 27880 }, { "epoch": 0.7346651169069782, "grad_norm": 3.5061516761779785, "learning_rate": 3.4714617900812485e-06, "loss": 0.10494542121887207, "step": 27890 }, { "epoch": 0.7349285321514769, "grad_norm": 3.2682909965515137, "learning_rate": 3.465001789169502e-06, "loss": 0.10331914424896241, "step": 27900 }, { "epoch": 0.7351919473959757, "grad_norm": 2.1470370292663574, "learning_rate": 3.458546544711404e-06, "loss": 0.10262191295623779, "step": 27910 }, { "epoch": 0.7354553626404744, "grad_norm": 2.05138897895813, "learning_rate": 3.45209606140535e-06, "loss": 0.09278889894485473, "step": 27920 }, { "epoch": 0.7357187778849732, "grad_norm": 2.6783061027526855, "learning_rate": 3.4456503439462797e-06, "loss": 0.10044195652008056, "step": 27930 }, { "epoch": 0.7359821931294719, "grad_norm": 3.294482946395874, "learning_rate": 3.4392093970256703e-06, "loss": 0.12145287990570068, "step": 27940 }, { "epoch": 0.7362456083739706, "grad_norm": 1.8555399179458618, "learning_rate": 3.4327732253315195e-06, "loss": 0.10919106006622314, "step": 27950 }, { "epoch": 0.7365090236184694, "grad_norm": 3.063446044921875, "learning_rate": 3.4263418335483477e-06, "loss": 0.10706827640533448, "step": 27960 }, { "epoch": 0.7367724388629681, "grad_norm": 2.8431131839752197, "learning_rate": 3.419915226357203e-06, "loss": 0.09683443307876587, "step": 27970 }, { "epoch": 0.7370358541074669, "grad_norm": 2.9204607009887695, "learning_rate": 3.413493408435642e-06, "loss": 0.10789331197738647, "step": 27980 }, { "epoch": 0.7372992693519655, "grad_norm": 2.3793022632598877, "learning_rate": 3.407076384457747e-06, "loss": 0.09342894554138184, "step": 27990 }, { "epoch": 0.7375626845964643, "grad_norm": 1.5921000242233276, "learning_rate": 3.400664159094106e-06, "loss": 0.08595030307769776, "step": 28000 }, { "epoch": 0.737826099840963, "grad_norm": 2.212064027786255, "learning_rate": 3.394256737011803e-06, "loss": 0.09668026566505432, "step": 28010 }, { "epoch": 0.7380895150854618, "grad_norm": 2.154963493347168, "learning_rate": 3.3878541228744455e-06, "loss": 0.08793253898620605, "step": 28020 }, { "epoch": 0.7383529303299605, "grad_norm": 1.60637366771698, "learning_rate": 3.38145632134213e-06, "loss": 0.09386801719665527, "step": 28030 }, { "epoch": 0.7386163455744592, "grad_norm": 3.5444517135620117, "learning_rate": 3.3750633370714527e-06, "loss": 0.09795426726341247, "step": 28040 }, { "epoch": 0.738879760818958, "grad_norm": 1.3432629108428955, "learning_rate": 3.368675174715501e-06, "loss": 0.1017157793045044, "step": 28050 }, { "epoch": 0.7391431760634567, "grad_norm": 3.02139949798584, "learning_rate": 3.3622918389238544e-06, "loss": 0.10340157747268677, "step": 28060 }, { "epoch": 0.7394065913079555, "grad_norm": 2.474731922149658, "learning_rate": 3.3559133343425865e-06, "loss": 0.10694707632064819, "step": 28070 }, { "epoch": 0.7396700065524542, "grad_norm": 2.786886215209961, "learning_rate": 3.3495396656142442e-06, "loss": 0.09945442080497742, "step": 28080 }, { "epoch": 0.739933421796953, "grad_norm": 2.106501340866089, "learning_rate": 3.343170837377858e-06, "loss": 0.09386416673660278, "step": 28090 }, { "epoch": 0.7401968370414517, "grad_norm": 2.766957998275757, "learning_rate": 3.336806854268936e-06, "loss": 0.10468993186950684, "step": 28100 }, { "epoch": 0.7404602522859505, "grad_norm": 2.5274133682250977, "learning_rate": 3.3304477209194574e-06, "loss": 0.10388832092285157, "step": 28110 }, { "epoch": 0.7407236675304492, "grad_norm": 3.1056859493255615, "learning_rate": 3.3240934419578828e-06, "loss": 0.09965682029724121, "step": 28120 }, { "epoch": 0.7409870827749478, "grad_norm": 1.6219466924667358, "learning_rate": 3.3177440220091207e-06, "loss": 0.10188593864440917, "step": 28130 }, { "epoch": 0.7412504980194466, "grad_norm": 2.0092713832855225, "learning_rate": 3.3113994656945537e-06, "loss": 0.09447439908981323, "step": 28140 }, { "epoch": 0.7415139132639453, "grad_norm": 4.52925443649292, "learning_rate": 3.3050597776320293e-06, "loss": 0.0935451865196228, "step": 28150 }, { "epoch": 0.7417773285084441, "grad_norm": 2.6169466972351074, "learning_rate": 3.298724962435842e-06, "loss": 0.09391886591911316, "step": 28160 }, { "epoch": 0.7420407437529428, "grad_norm": 4.0969648361206055, "learning_rate": 3.292395024716746e-06, "loss": 0.10677224397659302, "step": 28170 }, { "epoch": 0.7423041589974416, "grad_norm": 2.4028255939483643, "learning_rate": 3.2860699690819407e-06, "loss": 0.09383483529090882, "step": 28180 }, { "epoch": 0.7425675742419403, "grad_norm": 2.902477264404297, "learning_rate": 3.2797498001350737e-06, "loss": 0.09085143208503724, "step": 28190 }, { "epoch": 0.7428309894864391, "grad_norm": 1.9157549142837524, "learning_rate": 3.27343452247624e-06, "loss": 0.08822721242904663, "step": 28200 }, { "epoch": 0.7430944047309378, "grad_norm": 2.1931116580963135, "learning_rate": 3.267124140701974e-06, "loss": 0.10451792478561402, "step": 28210 }, { "epoch": 0.7433578199754365, "grad_norm": 1.769761323928833, "learning_rate": 3.2608186594052314e-06, "loss": 0.09247677326202393, "step": 28220 }, { "epoch": 0.7436212352199353, "grad_norm": 2.413564443588257, "learning_rate": 3.2545180831754264e-06, "loss": 0.10546669960021973, "step": 28230 }, { "epoch": 0.743884650464434, "grad_norm": 2.748979330062866, "learning_rate": 3.248222416598381e-06, "loss": 0.09430636167526245, "step": 28240 }, { "epoch": 0.7441480657089328, "grad_norm": 1.8320872783660889, "learning_rate": 3.2419316642563646e-06, "loss": 0.09588855504989624, "step": 28250 }, { "epoch": 0.7444114809534315, "grad_norm": 2.591697931289673, "learning_rate": 3.235645830728047e-06, "loss": 0.10449882745742797, "step": 28260 }, { "epoch": 0.7446748961979303, "grad_norm": 1.976649284362793, "learning_rate": 3.2293649205885293e-06, "loss": 0.08560896515846253, "step": 28270 }, { "epoch": 0.744938311442429, "grad_norm": 2.1190361976623535, "learning_rate": 3.2230889384093355e-06, "loss": 0.08355957269668579, "step": 28280 }, { "epoch": 0.7452017266869276, "grad_norm": 2.6404058933258057, "learning_rate": 3.2168178887583947e-06, "loss": 0.104206120967865, "step": 28290 }, { "epoch": 0.7454651419314264, "grad_norm": 5.493001937866211, "learning_rate": 3.210551776200044e-06, "loss": 0.0933218777179718, "step": 28300 }, { "epoch": 0.7457285571759251, "grad_norm": 2.026585578918457, "learning_rate": 3.204290605295033e-06, "loss": 0.10020768642425537, "step": 28310 }, { "epoch": 0.7459919724204239, "grad_norm": 3.0464818477630615, "learning_rate": 3.1980343806005077e-06, "loss": 0.0981270432472229, "step": 28320 }, { "epoch": 0.7462553876649226, "grad_norm": 2.390859365463257, "learning_rate": 3.1917831066700255e-06, "loss": 0.10742367506027221, "step": 28330 }, { "epoch": 0.7465188029094214, "grad_norm": 2.0110936164855957, "learning_rate": 3.1855367880535336e-06, "loss": 0.10969111919403077, "step": 28340 }, { "epoch": 0.7467822181539201, "grad_norm": 2.123685598373413, "learning_rate": 3.179295429297361e-06, "loss": 0.10457582473754883, "step": 28350 }, { "epoch": 0.7470456333984189, "grad_norm": 1.978973150253296, "learning_rate": 3.173059034944248e-06, "loss": 0.09839736819267272, "step": 28360 }, { "epoch": 0.7473090486429176, "grad_norm": 2.0881593227386475, "learning_rate": 3.1668276095333063e-06, "loss": 0.0992547631263733, "step": 28370 }, { "epoch": 0.7475724638874163, "grad_norm": 1.6047779321670532, "learning_rate": 3.160601157600045e-06, "loss": 0.10460590124130249, "step": 28380 }, { "epoch": 0.7478358791319151, "grad_norm": 2.226555347442627, "learning_rate": 3.154379683676334e-06, "loss": 0.08303243517875672, "step": 28390 }, { "epoch": 0.7480992943764138, "grad_norm": 3.1835081577301025, "learning_rate": 3.148163192290432e-06, "loss": 0.09979709386825561, "step": 28400 }, { "epoch": 0.7483627096209126, "grad_norm": 2.9995081424713135, "learning_rate": 3.1419516879669744e-06, "loss": 0.09587408304214477, "step": 28410 }, { "epoch": 0.7486261248654112, "grad_norm": 1.9188891649246216, "learning_rate": 3.1357451752269595e-06, "loss": 0.09649604558944702, "step": 28420 }, { "epoch": 0.74888954010991, "grad_norm": 3.011249542236328, "learning_rate": 3.1295436585877558e-06, "loss": 0.1129106879234314, "step": 28430 }, { "epoch": 0.7491529553544087, "grad_norm": 6.939969062805176, "learning_rate": 3.1233471425630947e-06, "loss": 0.10281457901000976, "step": 28440 }, { "epoch": 0.7494163705989075, "grad_norm": 2.2458865642547607, "learning_rate": 3.1171556316630626e-06, "loss": 0.08872330784797669, "step": 28450 }, { "epoch": 0.7496797858434062, "grad_norm": 2.0734472274780273, "learning_rate": 3.110969130394117e-06, "loss": 0.08176556825637818, "step": 28460 }, { "epoch": 0.7499432010879049, "grad_norm": 2.0350544452667236, "learning_rate": 3.1047876432590596e-06, "loss": 0.10205944776535034, "step": 28470 }, { "epoch": 0.7502066163324037, "grad_norm": 2.671984910964966, "learning_rate": 3.0986111747570335e-06, "loss": 0.09553874135017396, "step": 28480 }, { "epoch": 0.7504700315769024, "grad_norm": 2.6679515838623047, "learning_rate": 3.092439729383548e-06, "loss": 0.10842065811157227, "step": 28490 }, { "epoch": 0.7507334468214012, "grad_norm": 1.819542407989502, "learning_rate": 3.0862733116304435e-06, "loss": 0.09263455867767334, "step": 28500 }, { "epoch": 0.7509968620658999, "grad_norm": 3.2785470485687256, "learning_rate": 3.0801119259859115e-06, "loss": 0.10728580951690674, "step": 28510 }, { "epoch": 0.7512602773103987, "grad_norm": 3.1553173065185547, "learning_rate": 3.0739555769344666e-06, "loss": 0.08091068267822266, "step": 28520 }, { "epoch": 0.7515236925548974, "grad_norm": 3.485546827316284, "learning_rate": 3.067804268956964e-06, "loss": 0.10621962547302247, "step": 28530 }, { "epoch": 0.7517871077993962, "grad_norm": 2.9485819339752197, "learning_rate": 3.0616580065305977e-06, "loss": 0.09227765798568725, "step": 28540 }, { "epoch": 0.7520505230438949, "grad_norm": 1.7040555477142334, "learning_rate": 3.055516794128881e-06, "loss": 0.09956996440887451, "step": 28550 }, { "epoch": 0.7523139382883935, "grad_norm": 4.2626824378967285, "learning_rate": 3.0493806362216516e-06, "loss": 0.09763548374176026, "step": 28560 }, { "epoch": 0.7525773535328923, "grad_norm": 3.0209834575653076, "learning_rate": 3.0432495372750692e-06, "loss": 0.09299440383911133, "step": 28570 }, { "epoch": 0.752840768777391, "grad_norm": 2.9840471744537354, "learning_rate": 3.037123501751611e-06, "loss": 0.09338844418525696, "step": 28580 }, { "epoch": 0.7531041840218898, "grad_norm": 3.1554367542266846, "learning_rate": 3.0310025341100757e-06, "loss": 0.09847873449325562, "step": 28590 }, { "epoch": 0.7533675992663885, "grad_norm": 3.1454548835754395, "learning_rate": 3.0248866388055674e-06, "loss": 0.10472718477249146, "step": 28600 }, { "epoch": 0.7536310145108873, "grad_norm": 1.5822937488555908, "learning_rate": 3.018775820289488e-06, "loss": 0.07633579969406128, "step": 28610 }, { "epoch": 0.753894429755386, "grad_norm": 2.140939712524414, "learning_rate": 3.0126700830095668e-06, "loss": 0.09136924147605896, "step": 28620 }, { "epoch": 0.7541578449998848, "grad_norm": 2.990619421005249, "learning_rate": 3.006569431409817e-06, "loss": 0.08655252456665039, "step": 28630 }, { "epoch": 0.7544212602443835, "grad_norm": 2.742727279663086, "learning_rate": 3.000473869930558e-06, "loss": 0.09310863018035889, "step": 28640 }, { "epoch": 0.7546846754888822, "grad_norm": 2.9281086921691895, "learning_rate": 2.9943834030084006e-06, "loss": 0.10970838069915771, "step": 28650 }, { "epoch": 0.754948090733381, "grad_norm": 2.348353385925293, "learning_rate": 2.988298035076249e-06, "loss": 0.09133574962615967, "step": 28660 }, { "epoch": 0.7552115059778797, "grad_norm": 2.5512325763702393, "learning_rate": 2.9822177705633005e-06, "loss": 0.08868701457977295, "step": 28670 }, { "epoch": 0.7554749212223785, "grad_norm": 2.8413760662078857, "learning_rate": 2.976142613895031e-06, "loss": 0.09564074277877807, "step": 28680 }, { "epoch": 0.7557383364668772, "grad_norm": 2.2070534229278564, "learning_rate": 2.970072569493202e-06, "loss": 0.08654633760452271, "step": 28690 }, { "epoch": 0.756001751711376, "grad_norm": 2.447084665298462, "learning_rate": 2.9640076417758546e-06, "loss": 0.09433935880661011, "step": 28700 }, { "epoch": 0.7562651669558746, "grad_norm": 3.5959064960479736, "learning_rate": 2.9579478351573e-06, "loss": 0.09279980659484863, "step": 28710 }, { "epoch": 0.7565285822003734, "grad_norm": 2.6907219886779785, "learning_rate": 2.9518931540481377e-06, "loss": 0.09703127145767212, "step": 28720 }, { "epoch": 0.7567919974448721, "grad_norm": 1.2882415056228638, "learning_rate": 2.945843602855215e-06, "loss": 0.10106698274612427, "step": 28730 }, { "epoch": 0.7570554126893708, "grad_norm": 2.566296100616455, "learning_rate": 2.9397991859816588e-06, "loss": 0.10895864963531494, "step": 28740 }, { "epoch": 0.7573188279338696, "grad_norm": 2.6259870529174805, "learning_rate": 2.9337599078268597e-06, "loss": 0.09466239213943481, "step": 28750 }, { "epoch": 0.7575822431783683, "grad_norm": 2.4503023624420166, "learning_rate": 2.927725772786464e-06, "loss": 0.09702770709991455, "step": 28760 }, { "epoch": 0.7578456584228671, "grad_norm": 2.397489547729492, "learning_rate": 2.9216967852523747e-06, "loss": 0.09596680402755738, "step": 28770 }, { "epoch": 0.7581090736673658, "grad_norm": 3.508759021759033, "learning_rate": 2.915672949612749e-06, "loss": 0.09282307028770446, "step": 28780 }, { "epoch": 0.7583724889118646, "grad_norm": 1.8979169130325317, "learning_rate": 2.9096542702519904e-06, "loss": 0.0938810408115387, "step": 28790 }, { "epoch": 0.7586359041563633, "grad_norm": 2.7121872901916504, "learning_rate": 2.9036407515507613e-06, "loss": 0.08442033529281616, "step": 28800 }, { "epoch": 0.758899319400862, "grad_norm": 2.1328201293945312, "learning_rate": 2.89763239788596e-06, "loss": 0.09118715524673462, "step": 28810 }, { "epoch": 0.7591627346453608, "grad_norm": 2.2630295753479004, "learning_rate": 2.8916292136307147e-06, "loss": 0.09399818778038024, "step": 28820 }, { "epoch": 0.7594261498898595, "grad_norm": 1.7647418975830078, "learning_rate": 2.885631203154412e-06, "loss": 0.10753552913665772, "step": 28830 }, { "epoch": 0.7596895651343583, "grad_norm": 2.8345916271209717, "learning_rate": 2.879638370822656e-06, "loss": 0.10428094863891602, "step": 28840 }, { "epoch": 0.759952980378857, "grad_norm": 2.144148588180542, "learning_rate": 2.8736507209972996e-06, "loss": 0.0906878113746643, "step": 28850 }, { "epoch": 0.7602163956233557, "grad_norm": 3.1901183128356934, "learning_rate": 2.867668258036401e-06, "loss": 0.09294580221176148, "step": 28860 }, { "epoch": 0.7604798108678544, "grad_norm": 2.3735923767089844, "learning_rate": 2.8616909862942565e-06, "loss": 0.09900845289230346, "step": 28870 }, { "epoch": 0.7607432261123532, "grad_norm": 1.687879204750061, "learning_rate": 2.855718910121389e-06, "loss": 0.08793826699256897, "step": 28880 }, { "epoch": 0.7610066413568519, "grad_norm": 1.8836381435394287, "learning_rate": 2.8497520338645278e-06, "loss": 0.0767088770866394, "step": 28890 }, { "epoch": 0.7612700566013506, "grad_norm": 2.214472532272339, "learning_rate": 2.843790361866625e-06, "loss": 0.08874666690826416, "step": 28900 }, { "epoch": 0.7615334718458494, "grad_norm": 2.8661532402038574, "learning_rate": 2.837833898466842e-06, "loss": 0.08589435815811157, "step": 28910 }, { "epoch": 0.7617968870903481, "grad_norm": 3.2364132404327393, "learning_rate": 2.831882648000547e-06, "loss": 0.08863229155540467, "step": 28920 }, { "epoch": 0.7620603023348469, "grad_norm": 2.3490757942199707, "learning_rate": 2.8259366147993217e-06, "loss": 0.09601744413375854, "step": 28930 }, { "epoch": 0.7623237175793456, "grad_norm": 2.1961212158203125, "learning_rate": 2.819995803190948e-06, "loss": 0.09832123517990113, "step": 28940 }, { "epoch": 0.7625871328238444, "grad_norm": 4.62362813949585, "learning_rate": 2.8140602174993927e-06, "loss": 0.09241945743560791, "step": 28950 }, { "epoch": 0.7628505480683431, "grad_norm": 3.0954999923706055, "learning_rate": 2.8081298620448427e-06, "loss": 0.09429466724395752, "step": 28960 }, { "epoch": 0.7631139633128419, "grad_norm": 1.7147774696350098, "learning_rate": 2.8022047411436604e-06, "loss": 0.08244146704673767, "step": 28970 }, { "epoch": 0.7633773785573406, "grad_norm": 3.3695449829101562, "learning_rate": 2.7962848591084122e-06, "loss": 0.10852025747299195, "step": 28980 }, { "epoch": 0.7636407938018392, "grad_norm": 1.8923660516738892, "learning_rate": 2.7903702202478355e-06, "loss": 0.08753031492233276, "step": 28990 }, { "epoch": 0.763904209046338, "grad_norm": 4.60612678527832, "learning_rate": 2.7844608288668586e-06, "loss": 0.08406691551208496, "step": 29000 }, { "epoch": 0.7641676242908367, "grad_norm": 2.1912004947662354, "learning_rate": 2.7785566892665993e-06, "loss": 0.09659311771392823, "step": 29010 }, { "epoch": 0.7644310395353355, "grad_norm": 2.5284523963928223, "learning_rate": 2.77265780574434e-06, "loss": 0.09953500032424926, "step": 29020 }, { "epoch": 0.7646944547798342, "grad_norm": 2.218496799468994, "learning_rate": 2.766764182593544e-06, "loss": 0.09423129558563233, "step": 29030 }, { "epoch": 0.764957870024333, "grad_norm": 2.630033254623413, "learning_rate": 2.760875824103846e-06, "loss": 0.0920904517173767, "step": 29040 }, { "epoch": 0.7652212852688317, "grad_norm": 2.140984535217285, "learning_rate": 2.7549927345610407e-06, "loss": 0.0846093773841858, "step": 29050 }, { "epoch": 0.7654847005133305, "grad_norm": 2.926020622253418, "learning_rate": 2.749114918247102e-06, "loss": 0.09687070846557617, "step": 29060 }, { "epoch": 0.7657481157578292, "grad_norm": 1.93439781665802, "learning_rate": 2.7432423794401598e-06, "loss": 0.08891530632972718, "step": 29070 }, { "epoch": 0.7660115310023279, "grad_norm": 3.1132757663726807, "learning_rate": 2.7373751224144905e-06, "loss": 0.09711871147155762, "step": 29080 }, { "epoch": 0.7662749462468267, "grad_norm": 2.532039165496826, "learning_rate": 2.731513151440547e-06, "loss": 0.09756840467453003, "step": 29090 }, { "epoch": 0.7665383614913254, "grad_norm": 1.5524587631225586, "learning_rate": 2.7256564707849165e-06, "loss": 0.09318320155143738, "step": 29100 }, { "epoch": 0.7668017767358242, "grad_norm": 3.4899935722351074, "learning_rate": 2.719805084710356e-06, "loss": 0.10532212257385254, "step": 29110 }, { "epoch": 0.7670651919803229, "grad_norm": 1.9352490901947021, "learning_rate": 2.7139589974757454e-06, "loss": 0.08483395576477051, "step": 29120 }, { "epoch": 0.7673286072248217, "grad_norm": 2.6994922161102295, "learning_rate": 2.7081182133361207e-06, "loss": 0.089570814371109, "step": 29130 }, { "epoch": 0.7675920224693203, "grad_norm": 1.8842612504959106, "learning_rate": 2.7022827365426606e-06, "loss": 0.09112989902496338, "step": 29140 }, { "epoch": 0.7678554377138191, "grad_norm": 1.9989778995513916, "learning_rate": 2.696452571342676e-06, "loss": 0.09307501316070557, "step": 29150 }, { "epoch": 0.7681188529583178, "grad_norm": 2.377436637878418, "learning_rate": 2.6906277219796116e-06, "loss": 0.07965896129608155, "step": 29160 }, { "epoch": 0.7683822682028165, "grad_norm": 2.867647409439087, "learning_rate": 2.6848081926930447e-06, "loss": 0.08079749941825867, "step": 29170 }, { "epoch": 0.7686456834473153, "grad_norm": 2.0541000366210938, "learning_rate": 2.6789939877186753e-06, "loss": 0.09014179110527039, "step": 29180 }, { "epoch": 0.768909098691814, "grad_norm": 2.0953352451324463, "learning_rate": 2.67318511128834e-06, "loss": 0.08508173823356628, "step": 29190 }, { "epoch": 0.7691725139363128, "grad_norm": 3.921510934829712, "learning_rate": 2.66738156762999e-06, "loss": 0.0791674256324768, "step": 29200 }, { "epoch": 0.7694359291808115, "grad_norm": 3.5270185470581055, "learning_rate": 2.6615833609676846e-06, "loss": 0.09084223508834839, "step": 29210 }, { "epoch": 0.7696993444253103, "grad_norm": 2.5304408073425293, "learning_rate": 2.6557904955216185e-06, "loss": 0.10241789817810058, "step": 29220 }, { "epoch": 0.769962759669809, "grad_norm": 1.5457338094711304, "learning_rate": 2.650002975508086e-06, "loss": 0.08680304288864135, "step": 29230 }, { "epoch": 0.7702261749143078, "grad_norm": 3.4740285873413086, "learning_rate": 2.6442208051394936e-06, "loss": 0.08968276977539062, "step": 29240 }, { "epoch": 0.7704895901588065, "grad_norm": 1.9929414987564087, "learning_rate": 2.638443988624355e-06, "loss": 0.0889630675315857, "step": 29250 }, { "epoch": 0.7707530054033052, "grad_norm": 1.9364831447601318, "learning_rate": 2.632672530167283e-06, "loss": 0.09208944439888, "step": 29260 }, { "epoch": 0.771016420647804, "grad_norm": 2.357314109802246, "learning_rate": 2.6269064339689986e-06, "loss": 0.08987574577331543, "step": 29270 }, { "epoch": 0.7712798358923026, "grad_norm": 1.933821678161621, "learning_rate": 2.6211457042263157e-06, "loss": 0.08520848751068115, "step": 29280 }, { "epoch": 0.7715432511368014, "grad_norm": 2.6606876850128174, "learning_rate": 2.6153903451321384e-06, "loss": 0.09121593236923217, "step": 29290 }, { "epoch": 0.7718066663813001, "grad_norm": 2.6754186153411865, "learning_rate": 2.6096403608754685e-06, "loss": 0.08025048971176148, "step": 29300 }, { "epoch": 0.7720700816257989, "grad_norm": 3.5581741333007812, "learning_rate": 2.603895755641387e-06, "loss": 0.10184316635131836, "step": 29310 }, { "epoch": 0.7723334968702976, "grad_norm": 1.5301134586334229, "learning_rate": 2.598156533611077e-06, "loss": 0.09203006029129028, "step": 29320 }, { "epoch": 0.7725969121147963, "grad_norm": 3.0415236949920654, "learning_rate": 2.592422698961781e-06, "loss": 0.0862890362739563, "step": 29330 }, { "epoch": 0.7728603273592951, "grad_norm": 2.8997414112091064, "learning_rate": 2.586694255866831e-06, "loss": 0.10128507614135743, "step": 29340 }, { "epoch": 0.7731237426037938, "grad_norm": 2.555791139602661, "learning_rate": 2.580971208495642e-06, "loss": 0.08747572898864746, "step": 29350 }, { "epoch": 0.7733871578482926, "grad_norm": 3.6633896827697754, "learning_rate": 2.575253561013692e-06, "loss": 0.08505126237869262, "step": 29360 }, { "epoch": 0.7736505730927913, "grad_norm": 3.020815849304199, "learning_rate": 2.56954131758253e-06, "loss": 0.07821706533432007, "step": 29370 }, { "epoch": 0.7739139883372901, "grad_norm": 3.708372116088867, "learning_rate": 2.5638344823597762e-06, "loss": 0.08755351901054383, "step": 29380 }, { "epoch": 0.7741774035817888, "grad_norm": 2.0453524589538574, "learning_rate": 2.558133059499105e-06, "loss": 0.0872275710105896, "step": 29390 }, { "epoch": 0.7744408188262876, "grad_norm": 2.1815953254699707, "learning_rate": 2.552437053150266e-06, "loss": 0.08082455396652222, "step": 29400 }, { "epoch": 0.7747042340707863, "grad_norm": 2.1302521228790283, "learning_rate": 2.546746467459058e-06, "loss": 0.08727036714553833, "step": 29410 }, { "epoch": 0.774967649315285, "grad_norm": 3.180314779281616, "learning_rate": 2.541061306567325e-06, "loss": 0.10407878160476684, "step": 29420 }, { "epoch": 0.7752310645597837, "grad_norm": 2.763427972793579, "learning_rate": 2.5353815746129827e-06, "loss": 0.08496575355529785, "step": 29430 }, { "epoch": 0.7754944798042824, "grad_norm": 2.910874843597412, "learning_rate": 2.529707275729979e-06, "loss": 0.09237818121910095, "step": 29440 }, { "epoch": 0.7757578950487812, "grad_norm": 3.6308939456939697, "learning_rate": 2.524038414048321e-06, "loss": 0.09484554529190063, "step": 29450 }, { "epoch": 0.7760213102932799, "grad_norm": 2.922049045562744, "learning_rate": 2.518374993694043e-06, "loss": 0.08626904487609863, "step": 29460 }, { "epoch": 0.7762847255377787, "grad_norm": 2.3749148845672607, "learning_rate": 2.5127170187892257e-06, "loss": 0.08837567567825318, "step": 29470 }, { "epoch": 0.7765481407822774, "grad_norm": 2.7694058418273926, "learning_rate": 2.507064493451995e-06, "loss": 0.08501144647598266, "step": 29480 }, { "epoch": 0.7768115560267762, "grad_norm": 2.0266053676605225, "learning_rate": 2.501417421796498e-06, "loss": 0.09618663787841797, "step": 29490 }, { "epoch": 0.7770749712712749, "grad_norm": 2.5935680866241455, "learning_rate": 2.4957758079329173e-06, "loss": 0.09920796751976013, "step": 29500 }, { "epoch": 0.7773383865157736, "grad_norm": 2.0130650997161865, "learning_rate": 2.490139655967465e-06, "loss": 0.09187167882919312, "step": 29510 }, { "epoch": 0.7776018017602724, "grad_norm": 2.348369836807251, "learning_rate": 2.4845089700023694e-06, "loss": 0.09310215711593628, "step": 29520 }, { "epoch": 0.7778652170047711, "grad_norm": 2.5182199478149414, "learning_rate": 2.4788837541358957e-06, "loss": 0.08300119042396545, "step": 29530 }, { "epoch": 0.7781286322492699, "grad_norm": 2.9279797077178955, "learning_rate": 2.4732640124623152e-06, "loss": 0.0924929916858673, "step": 29540 }, { "epoch": 0.7783920474937686, "grad_norm": 2.4477994441986084, "learning_rate": 2.467649749071913e-06, "loss": 0.08381636142730713, "step": 29550 }, { "epoch": 0.7786554627382674, "grad_norm": 2.974266529083252, "learning_rate": 2.4620409680509993e-06, "loss": 0.09056227207183838, "step": 29560 }, { "epoch": 0.778918877982766, "grad_norm": 2.226238250732422, "learning_rate": 2.4564376734818818e-06, "loss": 0.0906804084777832, "step": 29570 }, { "epoch": 0.7791822932272648, "grad_norm": 2.3909101486206055, "learning_rate": 2.450839869442887e-06, "loss": 0.08458212614059449, "step": 29580 }, { "epoch": 0.7794457084717635, "grad_norm": 1.9600927829742432, "learning_rate": 2.4452475600083323e-06, "loss": 0.07496619820594788, "step": 29590 }, { "epoch": 0.7797091237162622, "grad_norm": 2.423837661743164, "learning_rate": 2.4396607492485393e-06, "loss": 0.08756452202796935, "step": 29600 }, { "epoch": 0.779972538960761, "grad_norm": 3.6921050548553467, "learning_rate": 2.4340794412298364e-06, "loss": 0.10088367462158203, "step": 29610 }, { "epoch": 0.7802359542052597, "grad_norm": 2.8824121952056885, "learning_rate": 2.428503640014539e-06, "loss": 0.09398627281188965, "step": 29620 }, { "epoch": 0.7804993694497585, "grad_norm": 1.8444591760635376, "learning_rate": 2.4229333496609532e-06, "loss": 0.09271512031555176, "step": 29630 }, { "epoch": 0.7807627846942572, "grad_norm": 3.245485305786133, "learning_rate": 2.4173685742233787e-06, "loss": 0.08380706906318665, "step": 29640 }, { "epoch": 0.781026199938756, "grad_norm": 2.7868289947509766, "learning_rate": 2.411809317752095e-06, "loss": 0.08659094572067261, "step": 29650 }, { "epoch": 0.7812896151832547, "grad_norm": 2.785828113555908, "learning_rate": 2.4062555842933753e-06, "loss": 0.08500564098358154, "step": 29660 }, { "epoch": 0.7815530304277535, "grad_norm": 2.237974166870117, "learning_rate": 2.4007073778894665e-06, "loss": 0.08731939792633056, "step": 29670 }, { "epoch": 0.7818164456722522, "grad_norm": 2.11653470993042, "learning_rate": 2.395164702578583e-06, "loss": 0.08875954747200013, "step": 29680 }, { "epoch": 0.7820798609167509, "grad_norm": 3.959003448486328, "learning_rate": 2.3896275623949327e-06, "loss": 0.09177401065826415, "step": 29690 }, { "epoch": 0.7823432761612497, "grad_norm": 3.2279117107391357, "learning_rate": 2.38409596136868e-06, "loss": 0.08197893500328064, "step": 29700 }, { "epoch": 0.7826066914057483, "grad_norm": 2.14776873588562, "learning_rate": 2.378569903525969e-06, "loss": 0.09043680429458618, "step": 29710 }, { "epoch": 0.7828701066502471, "grad_norm": 2.691437005996704, "learning_rate": 2.3730493928888963e-06, "loss": 0.09785207509994506, "step": 29720 }, { "epoch": 0.7831335218947458, "grad_norm": 2.6185035705566406, "learning_rate": 2.367534433475527e-06, "loss": 0.09478012919425964, "step": 29730 }, { "epoch": 0.7833969371392446, "grad_norm": 2.45581316947937, "learning_rate": 2.3620250292998925e-06, "loss": 0.0932729721069336, "step": 29740 }, { "epoch": 0.7836603523837433, "grad_norm": 2.381981372833252, "learning_rate": 2.356521184371973e-06, "loss": 0.08716692924499511, "step": 29750 }, { "epoch": 0.7839237676282421, "grad_norm": 2.4502158164978027, "learning_rate": 2.3510229026977018e-06, "loss": 0.08373565673828125, "step": 29760 }, { "epoch": 0.7841871828727408, "grad_norm": 2.1655447483062744, "learning_rate": 2.345530188278967e-06, "loss": 0.08180286884307861, "step": 29770 }, { "epoch": 0.7844505981172395, "grad_norm": 1.8963894844055176, "learning_rate": 2.3400430451136e-06, "loss": 0.09449170231819153, "step": 29780 }, { "epoch": 0.7847140133617383, "grad_norm": 4.73817777633667, "learning_rate": 2.334561477195387e-06, "loss": 0.08169646263122558, "step": 29790 }, { "epoch": 0.784977428606237, "grad_norm": 3.0039288997650146, "learning_rate": 2.329085488514049e-06, "loss": 0.08390746116638184, "step": 29800 }, { "epoch": 0.7852408438507358, "grad_norm": 2.673894166946411, "learning_rate": 2.3236150830552374e-06, "loss": 0.08037449717521668, "step": 29810 }, { "epoch": 0.7855042590952345, "grad_norm": 2.4538917541503906, "learning_rate": 2.3181502648005593e-06, "loss": 0.08786808252334595, "step": 29820 }, { "epoch": 0.7857676743397333, "grad_norm": 2.6540868282318115, "learning_rate": 2.3126910377275415e-06, "loss": 0.07732715606689453, "step": 29830 }, { "epoch": 0.786031089584232, "grad_norm": 3.927133798599243, "learning_rate": 2.3072374058096447e-06, "loss": 0.11513460874557495, "step": 29840 }, { "epoch": 0.7862945048287308, "grad_norm": 2.134880542755127, "learning_rate": 2.301789373016259e-06, "loss": 0.08772221207618713, "step": 29850 }, { "epoch": 0.7865579200732294, "grad_norm": 3.695307970046997, "learning_rate": 2.296346943312693e-06, "loss": 0.0940642774105072, "step": 29860 }, { "epoch": 0.7868213353177281, "grad_norm": 2.7273318767547607, "learning_rate": 2.290910120660188e-06, "loss": 0.0951572060585022, "step": 29870 }, { "epoch": 0.7870847505622269, "grad_norm": 2.304422616958618, "learning_rate": 2.285478909015896e-06, "loss": 0.07854439616203308, "step": 29880 }, { "epoch": 0.7873481658067256, "grad_norm": 2.5479419231414795, "learning_rate": 2.2800533123328874e-06, "loss": 0.103392493724823, "step": 29890 }, { "epoch": 0.7876115810512244, "grad_norm": 2.8074777126312256, "learning_rate": 2.2746333345601435e-06, "loss": 0.08892914056777954, "step": 29900 }, { "epoch": 0.7878749962957231, "grad_norm": 1.8659031391143799, "learning_rate": 2.269218979642558e-06, "loss": 0.0832866370677948, "step": 29910 }, { "epoch": 0.7881384115402219, "grad_norm": 2.519178628921509, "learning_rate": 2.2638102515209382e-06, "loss": 0.09150443077087403, "step": 29920 }, { "epoch": 0.7884018267847206, "grad_norm": 4.058499813079834, "learning_rate": 2.258407154131982e-06, "loss": 0.09199479818344117, "step": 29930 }, { "epoch": 0.7886652420292193, "grad_norm": 2.1130099296569824, "learning_rate": 2.2530096914082965e-06, "loss": 0.09281176924705506, "step": 29940 }, { "epoch": 0.7889286572737181, "grad_norm": 2.269911289215088, "learning_rate": 2.2476178672783943e-06, "loss": 0.0877065122127533, "step": 29950 }, { "epoch": 0.7891920725182168, "grad_norm": 2.6930084228515625, "learning_rate": 2.242231685666675e-06, "loss": 0.08590928316116334, "step": 29960 }, { "epoch": 0.7894554877627156, "grad_norm": 2.7782742977142334, "learning_rate": 2.2368511504934333e-06, "loss": 0.08608969449996948, "step": 29970 }, { "epoch": 0.7897189030072143, "grad_norm": 2.3896477222442627, "learning_rate": 2.2314762656748544e-06, "loss": 0.08569257259368897, "step": 29980 }, { "epoch": 0.7899823182517131, "grad_norm": 1.4516273736953735, "learning_rate": 2.2261070351230107e-06, "loss": 0.08829410076141357, "step": 29990 }, { "epoch": 0.7902457334962117, "grad_norm": 2.6263279914855957, "learning_rate": 2.2207434627458626e-06, "loss": 0.08615537881851196, "step": 30000 }, { "epoch": 0.7905091487407105, "grad_norm": 2.901116132736206, "learning_rate": 2.2153855524472513e-06, "loss": 0.0907454252243042, "step": 30010 }, { "epoch": 0.7907725639852092, "grad_norm": 1.9600228071212769, "learning_rate": 2.2100333081268853e-06, "loss": 0.0864433228969574, "step": 30020 }, { "epoch": 0.7910359792297079, "grad_norm": 3.071591854095459, "learning_rate": 2.204686733680369e-06, "loss": 0.08392963409423829, "step": 30030 }, { "epoch": 0.7912993944742067, "grad_norm": 2.7054455280303955, "learning_rate": 2.1993458329991635e-06, "loss": 0.08720777034759522, "step": 30040 }, { "epoch": 0.7915628097187054, "grad_norm": 2.7217955589294434, "learning_rate": 2.1940106099706137e-06, "loss": 0.08970911502838134, "step": 30050 }, { "epoch": 0.7918262249632042, "grad_norm": 1.810126543045044, "learning_rate": 2.1886810684779204e-06, "loss": 0.08763549923896789, "step": 30060 }, { "epoch": 0.7920896402077029, "grad_norm": 2.5805320739746094, "learning_rate": 2.18335721240015e-06, "loss": 0.08594081401824952, "step": 30070 }, { "epoch": 0.7923530554522017, "grad_norm": 3.483250379562378, "learning_rate": 2.1780390456122424e-06, "loss": 0.07706838846206665, "step": 30080 }, { "epoch": 0.7926164706967004, "grad_norm": 2.477193832397461, "learning_rate": 2.1727265719849853e-06, "loss": 0.09225273132324219, "step": 30090 }, { "epoch": 0.7928798859411992, "grad_norm": 1.7633944749832153, "learning_rate": 2.1674197953850274e-06, "loss": 0.08523052930831909, "step": 30100 }, { "epoch": 0.7931433011856979, "grad_norm": 3.2719123363494873, "learning_rate": 2.162118719674867e-06, "loss": 0.09826046228408813, "step": 30110 }, { "epoch": 0.7934067164301966, "grad_norm": 3.6637346744537354, "learning_rate": 2.1568233487128556e-06, "loss": 0.07591392993927001, "step": 30120 }, { "epoch": 0.7936701316746954, "grad_norm": 3.048534631729126, "learning_rate": 2.1515336863531976e-06, "loss": 0.09183664321899414, "step": 30130 }, { "epoch": 0.793933546919194, "grad_norm": 2.877507448196411, "learning_rate": 2.1462497364459356e-06, "loss": 0.08337746858596802, "step": 30140 }, { "epoch": 0.7941969621636928, "grad_norm": 2.3191170692443848, "learning_rate": 2.1409715028369504e-06, "loss": 0.0826757311820984, "step": 30150 }, { "epoch": 0.7944603774081915, "grad_norm": 2.446415662765503, "learning_rate": 2.135698989367976e-06, "loss": 0.08664758801460266, "step": 30160 }, { "epoch": 0.7947237926526903, "grad_norm": 2.3971171379089355, "learning_rate": 2.1304321998765685e-06, "loss": 0.07773652076721191, "step": 30170 }, { "epoch": 0.794987207897189, "grad_norm": 2.118086338043213, "learning_rate": 2.1251711381961336e-06, "loss": 0.08699110746383668, "step": 30180 }, { "epoch": 0.7952506231416878, "grad_norm": 3.4169137477874756, "learning_rate": 2.11991580815589e-06, "loss": 0.08706842660903931, "step": 30190 }, { "epoch": 0.7955140383861865, "grad_norm": 2.363497018814087, "learning_rate": 2.1146662135808948e-06, "loss": 0.08999662399291992, "step": 30200 }, { "epoch": 0.7957774536306852, "grad_norm": 2.9783506393432617, "learning_rate": 2.1094223582920346e-06, "loss": 0.08576791286468506, "step": 30210 }, { "epoch": 0.796040868875184, "grad_norm": 3.4433071613311768, "learning_rate": 2.1041842461060103e-06, "loss": 0.08500398397445678, "step": 30220 }, { "epoch": 0.7963042841196827, "grad_norm": 3.1248700618743896, "learning_rate": 2.098951880835346e-06, "loss": 0.09013777375221252, "step": 30230 }, { "epoch": 0.7965676993641815, "grad_norm": 2.8061108589172363, "learning_rate": 2.093725266288382e-06, "loss": 0.09180600643157959, "step": 30240 }, { "epoch": 0.7968311146086802, "grad_norm": 3.4006361961364746, "learning_rate": 2.088504406269274e-06, "loss": 0.0885395348072052, "step": 30250 }, { "epoch": 0.797094529853179, "grad_norm": 1.7919023036956787, "learning_rate": 2.083289304577992e-06, "loss": 0.07750924825668334, "step": 30260 }, { "epoch": 0.7973579450976777, "grad_norm": 2.4459080696105957, "learning_rate": 2.0780799650103123e-06, "loss": 0.0835038661956787, "step": 30270 }, { "epoch": 0.7976213603421765, "grad_norm": 4.031065940856934, "learning_rate": 2.0728763913578097e-06, "loss": 0.09352859258651733, "step": 30280 }, { "epoch": 0.7978847755866751, "grad_norm": 2.9771738052368164, "learning_rate": 2.0676785874078773e-06, "loss": 0.08051111698150634, "step": 30290 }, { "epoch": 0.7981481908311738, "grad_norm": 2.640281915664673, "learning_rate": 2.062486556943699e-06, "loss": 0.09099133610725403, "step": 30300 }, { "epoch": 0.7984116060756726, "grad_norm": 2.963686943054199, "learning_rate": 2.0573003037442575e-06, "loss": 0.09336358308792114, "step": 30310 }, { "epoch": 0.7986750213201713, "grad_norm": 2.7313249111175537, "learning_rate": 2.0521198315843315e-06, "loss": 0.08780993223190307, "step": 30320 }, { "epoch": 0.7989384365646701, "grad_norm": 3.1690754890441895, "learning_rate": 2.046945144234491e-06, "loss": 0.08576956987380982, "step": 30330 }, { "epoch": 0.7992018518091688, "grad_norm": 3.218993902206421, "learning_rate": 2.0417762454611012e-06, "loss": 0.09719876050949097, "step": 30340 }, { "epoch": 0.7994652670536676, "grad_norm": 2.720952272415161, "learning_rate": 2.0366131390263055e-06, "loss": 0.08743389844894409, "step": 30350 }, { "epoch": 0.7997286822981663, "grad_norm": 2.945373773574829, "learning_rate": 2.0314558286880383e-06, "loss": 0.08432192206382752, "step": 30360 }, { "epoch": 0.7999920975426651, "grad_norm": 3.994567394256592, "learning_rate": 2.026304318200011e-06, "loss": 0.08806929588317872, "step": 30370 }, { "epoch": 0.8002555127871638, "grad_norm": 3.0750808715820312, "learning_rate": 2.021158611311712e-06, "loss": 0.08321398496627808, "step": 30380 }, { "epoch": 0.8005189280316625, "grad_norm": 1.678617000579834, "learning_rate": 2.016018711768416e-06, "loss": 0.0879716157913208, "step": 30390 }, { "epoch": 0.8007823432761613, "grad_norm": 2.5323140621185303, "learning_rate": 2.010884623311162e-06, "loss": 0.09323452115058899, "step": 30400 }, { "epoch": 0.80104575852066, "grad_norm": 2.1512506008148193, "learning_rate": 2.005756349676753e-06, "loss": 0.07968565225601196, "step": 30410 }, { "epoch": 0.8013091737651588, "grad_norm": 2.654207468032837, "learning_rate": 2.0006338945977777e-06, "loss": 0.07940917015075684, "step": 30420 }, { "epoch": 0.8015725890096574, "grad_norm": 3.4031822681427, "learning_rate": 1.9955172618025743e-06, "loss": 0.08615797758102417, "step": 30430 }, { "epoch": 0.8018360042541562, "grad_norm": 2.654376983642578, "learning_rate": 1.9904064550152515e-06, "loss": 0.07742431163787841, "step": 30440 }, { "epoch": 0.8020994194986549, "grad_norm": 2.074766159057617, "learning_rate": 1.9853014779556734e-06, "loss": 0.0806098997592926, "step": 30450 }, { "epoch": 0.8023628347431536, "grad_norm": 2.9500184059143066, "learning_rate": 1.9802023343394606e-06, "loss": 0.08233503103256226, "step": 30460 }, { "epoch": 0.8026262499876524, "grad_norm": 3.1071200370788574, "learning_rate": 1.975109027877995e-06, "loss": 0.08520928025245667, "step": 30470 }, { "epoch": 0.8028896652321511, "grad_norm": 2.094205617904663, "learning_rate": 1.9700215622784026e-06, "loss": 0.07271157503128052, "step": 30480 }, { "epoch": 0.8031530804766499, "grad_norm": 5.078820705413818, "learning_rate": 1.96493994124356e-06, "loss": 0.09022271633148193, "step": 30490 }, { "epoch": 0.8034164957211486, "grad_norm": 2.121095895767212, "learning_rate": 1.9598641684720897e-06, "loss": 0.0761498212814331, "step": 30500 }, { "epoch": 0.8036799109656474, "grad_norm": 1.88832688331604, "learning_rate": 1.954794247658356e-06, "loss": 0.08678195476531983, "step": 30510 }, { "epoch": 0.8039433262101461, "grad_norm": 2.0162553787231445, "learning_rate": 1.949730182492473e-06, "loss": 0.07694587111473083, "step": 30520 }, { "epoch": 0.8042067414546449, "grad_norm": 3.861743927001953, "learning_rate": 1.9446719766602793e-06, "loss": 0.09080779552459717, "step": 30530 }, { "epoch": 0.8044701566991436, "grad_norm": 5.262766361236572, "learning_rate": 1.9396196338433547e-06, "loss": 0.082577383518219, "step": 30540 }, { "epoch": 0.8047335719436423, "grad_norm": 3.1703126430511475, "learning_rate": 1.9345731577190187e-06, "loss": 0.0822169840335846, "step": 30550 }, { "epoch": 0.804996987188141, "grad_norm": 3.254270553588867, "learning_rate": 1.929532551960309e-06, "loss": 0.08513719439506531, "step": 30560 }, { "epoch": 0.8052604024326397, "grad_norm": 2.9540443420410156, "learning_rate": 1.9244978202359974e-06, "loss": 0.06990282535552979, "step": 30570 }, { "epoch": 0.8055238176771385, "grad_norm": 3.1470656394958496, "learning_rate": 1.9194689662105803e-06, "loss": 0.07910778522491455, "step": 30580 }, { "epoch": 0.8057872329216372, "grad_norm": 2.693213939666748, "learning_rate": 1.9144459935442705e-06, "loss": 0.08685904741287231, "step": 30590 }, { "epoch": 0.806050648166136, "grad_norm": 2.43536376953125, "learning_rate": 1.9094289058930095e-06, "loss": 0.07715029120445252, "step": 30600 }, { "epoch": 0.8063140634106347, "grad_norm": 2.455214262008667, "learning_rate": 1.9044177069084502e-06, "loss": 0.07656545042991639, "step": 30610 }, { "epoch": 0.8065774786551335, "grad_norm": 2.823262929916382, "learning_rate": 1.899412400237951e-06, "loss": 0.0822932243347168, "step": 30620 }, { "epoch": 0.8068408938996322, "grad_norm": 2.544018030166626, "learning_rate": 1.8944129895246e-06, "loss": 0.08946459293365479, "step": 30630 }, { "epoch": 0.8071043091441309, "grad_norm": 3.0296802520751953, "learning_rate": 1.8894194784071761e-06, "loss": 0.08379071950912476, "step": 30640 }, { "epoch": 0.8073677243886297, "grad_norm": 2.5744240283966064, "learning_rate": 1.8844318705201815e-06, "loss": 0.0783932626247406, "step": 30650 }, { "epoch": 0.8076311396331284, "grad_norm": 2.404564619064331, "learning_rate": 1.879450169493805e-06, "loss": 0.08090419173240662, "step": 30660 }, { "epoch": 0.8078945548776272, "grad_norm": 2.2859694957733154, "learning_rate": 1.874474378953942e-06, "loss": 0.09044898748397827, "step": 30670 }, { "epoch": 0.8081579701221259, "grad_norm": 4.3050537109375, "learning_rate": 1.8695045025221936e-06, "loss": 0.07754315137863159, "step": 30680 }, { "epoch": 0.8084213853666247, "grad_norm": 2.9403786659240723, "learning_rate": 1.864540543815848e-06, "loss": 0.07737046480178833, "step": 30690 }, { "epoch": 0.8086848006111234, "grad_norm": 5.132088661193848, "learning_rate": 1.859582506447889e-06, "loss": 0.09047596454620362, "step": 30700 }, { "epoch": 0.8089482158556222, "grad_norm": 3.2016756534576416, "learning_rate": 1.8546303940269883e-06, "loss": 0.07931430339813232, "step": 30710 }, { "epoch": 0.8092116311001208, "grad_norm": 2.792496681213379, "learning_rate": 1.8496842101575053e-06, "loss": 0.07870762348175049, "step": 30720 }, { "epoch": 0.8094750463446195, "grad_norm": 2.8154919147491455, "learning_rate": 1.844743958439491e-06, "loss": 0.0938068985939026, "step": 30730 }, { "epoch": 0.8097384615891183, "grad_norm": 2.148768901824951, "learning_rate": 1.839809642468674e-06, "loss": 0.08462405204772949, "step": 30740 }, { "epoch": 0.810001876833617, "grad_norm": 2.321540594100952, "learning_rate": 1.8348812658364557e-06, "loss": 0.09137141704559326, "step": 30750 }, { "epoch": 0.8102652920781158, "grad_norm": 2.2075185775756836, "learning_rate": 1.8299588321299266e-06, "loss": 0.08152340650558472, "step": 30760 }, { "epoch": 0.8105287073226145, "grad_norm": 3.6624882221221924, "learning_rate": 1.8250423449318434e-06, "loss": 0.08253830671310425, "step": 30770 }, { "epoch": 0.8107921225671133, "grad_norm": 2.540477752685547, "learning_rate": 1.8201318078206443e-06, "loss": 0.08873430490493775, "step": 30780 }, { "epoch": 0.811055537811612, "grad_norm": 2.589667797088623, "learning_rate": 1.8152272243704216e-06, "loss": 0.09692566990852355, "step": 30790 }, { "epoch": 0.8113189530561108, "grad_norm": 2.778604507446289, "learning_rate": 1.8103285981509432e-06, "loss": 0.08577510118484497, "step": 30800 }, { "epoch": 0.8115823683006095, "grad_norm": 2.296704053878784, "learning_rate": 1.8054359327276473e-06, "loss": 0.08391463756561279, "step": 30810 }, { "epoch": 0.8118457835451082, "grad_norm": 2.123708963394165, "learning_rate": 1.8005492316616224e-06, "loss": 0.08061193227767945, "step": 30820 }, { "epoch": 0.812109198789607, "grad_norm": 2.507856845855713, "learning_rate": 1.7956684985096217e-06, "loss": 0.07538530230522156, "step": 30830 }, { "epoch": 0.8123726140341057, "grad_norm": 2.537041425704956, "learning_rate": 1.7907937368240525e-06, "loss": 0.0818285584449768, "step": 30840 }, { "epoch": 0.8126360292786045, "grad_norm": 4.513879299163818, "learning_rate": 1.7859249501529742e-06, "loss": 0.09056147336959838, "step": 30850 }, { "epoch": 0.8128994445231031, "grad_norm": 2.719139814376831, "learning_rate": 1.7810621420401054e-06, "loss": 0.08299989700317383, "step": 30860 }, { "epoch": 0.8131628597676019, "grad_norm": 2.3398799896240234, "learning_rate": 1.776205316024806e-06, "loss": 0.0836468756198883, "step": 30870 }, { "epoch": 0.8134262750121006, "grad_norm": 2.4988341331481934, "learning_rate": 1.7713544756420785e-06, "loss": 0.07525048255920411, "step": 30880 }, { "epoch": 0.8136896902565994, "grad_norm": 2.391540765762329, "learning_rate": 1.76650962442258e-06, "loss": 0.07815425992012023, "step": 30890 }, { "epoch": 0.8139531055010981, "grad_norm": 3.4006645679473877, "learning_rate": 1.761670765892599e-06, "loss": 0.08644297122955322, "step": 30900 }, { "epoch": 0.8142165207455968, "grad_norm": 2.764188766479492, "learning_rate": 1.7568379035740669e-06, "loss": 0.08736618757247924, "step": 30910 }, { "epoch": 0.8144799359900956, "grad_norm": 2.2514235973358154, "learning_rate": 1.7520110409845493e-06, "loss": 0.07925164699554443, "step": 30920 }, { "epoch": 0.8147433512345943, "grad_norm": 4.826331615447998, "learning_rate": 1.7471901816372416e-06, "loss": 0.08249950408935547, "step": 30930 }, { "epoch": 0.8150067664790931, "grad_norm": 2.1894338130950928, "learning_rate": 1.7423753290409784e-06, "loss": 0.08701100945472717, "step": 30940 }, { "epoch": 0.8152701817235918, "grad_norm": 3.931910991668701, "learning_rate": 1.7375664867002151e-06, "loss": 0.08073727488517761, "step": 30950 }, { "epoch": 0.8155335969680906, "grad_norm": 1.5167498588562012, "learning_rate": 1.7327636581150343e-06, "loss": 0.08502700924873352, "step": 30960 }, { "epoch": 0.8157970122125893, "grad_norm": 3.0197250843048096, "learning_rate": 1.7279668467811427e-06, "loss": 0.07694605588912964, "step": 30970 }, { "epoch": 0.816060427457088, "grad_norm": 2.638261318206787, "learning_rate": 1.7231760561898625e-06, "loss": 0.0797653079032898, "step": 30980 }, { "epoch": 0.8163238427015868, "grad_norm": 2.298715829849243, "learning_rate": 1.718391289828144e-06, "loss": 0.07382727861404419, "step": 30990 }, { "epoch": 0.8165872579460854, "grad_norm": 2.918879508972168, "learning_rate": 1.7136125511785462e-06, "loss": 0.07430043220520019, "step": 31000 }, { "epoch": 0.8168506731905842, "grad_norm": 2.7076807022094727, "learning_rate": 1.7088398437192344e-06, "loss": 0.08923771381378173, "step": 31010 }, { "epoch": 0.8171140884350829, "grad_norm": 2.025507688522339, "learning_rate": 1.7040731709239967e-06, "loss": 0.08071700334548951, "step": 31020 }, { "epoch": 0.8173775036795817, "grad_norm": 2.629371166229248, "learning_rate": 1.6993125362622231e-06, "loss": 0.07993229031562805, "step": 31030 }, { "epoch": 0.8176409189240804, "grad_norm": 3.5331363677978516, "learning_rate": 1.6945579431989067e-06, "loss": 0.07219305634498596, "step": 31040 }, { "epoch": 0.8179043341685792, "grad_norm": 2.6024935245513916, "learning_rate": 1.689809395194647e-06, "loss": 0.08059090375900269, "step": 31050 }, { "epoch": 0.8181677494130779, "grad_norm": 3.818247079849243, "learning_rate": 1.6850668957056393e-06, "loss": 0.08856091499328614, "step": 31060 }, { "epoch": 0.8184311646575766, "grad_norm": 3.364556074142456, "learning_rate": 1.6803304481836835e-06, "loss": 0.0779933512210846, "step": 31070 }, { "epoch": 0.8186945799020754, "grad_norm": 4.275763988494873, "learning_rate": 1.6756000560761687e-06, "loss": 0.06894055008888245, "step": 31080 }, { "epoch": 0.8189579951465741, "grad_norm": 3.0347630977630615, "learning_rate": 1.6708757228260786e-06, "loss": 0.08087823390960694, "step": 31090 }, { "epoch": 0.8192214103910729, "grad_norm": 2.159158229827881, "learning_rate": 1.6661574518719859e-06, "loss": 0.07661340236663819, "step": 31100 }, { "epoch": 0.8194848256355716, "grad_norm": 3.410316228866577, "learning_rate": 1.6614452466480502e-06, "loss": 0.08058860898017883, "step": 31110 }, { "epoch": 0.8197482408800704, "grad_norm": 2.899606704711914, "learning_rate": 1.6567391105840247e-06, "loss": 0.08042541146278381, "step": 31120 }, { "epoch": 0.820011656124569, "grad_norm": 2.215858221054077, "learning_rate": 1.6520390471052306e-06, "loss": 0.07632079720497131, "step": 31130 }, { "epoch": 0.8202750713690679, "grad_norm": 1.8580610752105713, "learning_rate": 1.647345059632578e-06, "loss": 0.06908929347991943, "step": 31140 }, { "epoch": 0.8205384866135665, "grad_norm": 3.87904953956604, "learning_rate": 1.6426571515825563e-06, "loss": 0.08505501747131347, "step": 31150 }, { "epoch": 0.8208019018580652, "grad_norm": 2.911346912384033, "learning_rate": 1.6379753263672272e-06, "loss": 0.0798795223236084, "step": 31160 }, { "epoch": 0.821065317102564, "grad_norm": 2.5251474380493164, "learning_rate": 1.6332995873942237e-06, "loss": 0.0744126558303833, "step": 31170 }, { "epoch": 0.8213287323470627, "grad_norm": 2.631186008453369, "learning_rate": 1.6286299380667504e-06, "loss": 0.06859964728355408, "step": 31180 }, { "epoch": 0.8215921475915615, "grad_norm": 4.14045524597168, "learning_rate": 1.6239663817835783e-06, "loss": 0.0827595591545105, "step": 31190 }, { "epoch": 0.8218555628360602, "grad_norm": 2.9398086071014404, "learning_rate": 1.6193089219390478e-06, "loss": 0.08216267824172974, "step": 31200 }, { "epoch": 0.822118978080559, "grad_norm": 2.5466420650482178, "learning_rate": 1.6146575619230597e-06, "loss": 0.08513419032096863, "step": 31210 }, { "epoch": 0.8223823933250577, "grad_norm": 2.62937068939209, "learning_rate": 1.610012305121068e-06, "loss": 0.07074662446975707, "step": 31220 }, { "epoch": 0.8226458085695565, "grad_norm": 2.5852315425872803, "learning_rate": 1.6053731549140982e-06, "loss": 0.09282787442207337, "step": 31230 }, { "epoch": 0.8229092238140552, "grad_norm": 2.476212739944458, "learning_rate": 1.6007401146787195e-06, "loss": 0.08488115072250366, "step": 31240 }, { "epoch": 0.8231726390585539, "grad_norm": 2.1105473041534424, "learning_rate": 1.596113187787064e-06, "loss": 0.07595652341842651, "step": 31250 }, { "epoch": 0.8234360543030527, "grad_norm": 2.391359567642212, "learning_rate": 1.5914923776068025e-06, "loss": 0.07861894369125366, "step": 31260 }, { "epoch": 0.8236994695475514, "grad_norm": 2.4380757808685303, "learning_rate": 1.5868776875011604e-06, "loss": 0.06981359124183655, "step": 31270 }, { "epoch": 0.8239628847920502, "grad_norm": 3.5578994750976562, "learning_rate": 1.582269120828912e-06, "loss": 0.09207847714424133, "step": 31280 }, { "epoch": 0.8242263000365488, "grad_norm": 2.136059045791626, "learning_rate": 1.5776666809443697e-06, "loss": 0.07107574343681336, "step": 31290 }, { "epoch": 0.8244897152810476, "grad_norm": 2.4012224674224854, "learning_rate": 1.5730703711973883e-06, "loss": 0.07558913230895996, "step": 31300 }, { "epoch": 0.8247531305255463, "grad_norm": 2.043140411376953, "learning_rate": 1.5684801949333584e-06, "loss": 0.07676472663879394, "step": 31310 }, { "epoch": 0.8250165457700451, "grad_norm": 2.868488311767578, "learning_rate": 1.5638961554932064e-06, "loss": 0.08698805570602416, "step": 31320 }, { "epoch": 0.8252799610145438, "grad_norm": 2.92977237701416, "learning_rate": 1.5593182562134002e-06, "loss": 0.08215725421905518, "step": 31330 }, { "epoch": 0.8255433762590425, "grad_norm": 3.3169336318969727, "learning_rate": 1.554746500425931e-06, "loss": 0.0734907865524292, "step": 31340 }, { "epoch": 0.8258067915035413, "grad_norm": 2.296592950820923, "learning_rate": 1.5501808914583139e-06, "loss": 0.07749072313308716, "step": 31350 }, { "epoch": 0.82607020674804, "grad_norm": 2.4353582859039307, "learning_rate": 1.5456214326336017e-06, "loss": 0.08032748699188233, "step": 31360 }, { "epoch": 0.8263336219925388, "grad_norm": 2.336390256881714, "learning_rate": 1.5410681272703643e-06, "loss": 0.07691059112548829, "step": 31370 }, { "epoch": 0.8265970372370375, "grad_norm": 3.1841673851013184, "learning_rate": 1.536520978682693e-06, "loss": 0.07302982807159424, "step": 31380 }, { "epoch": 0.8268604524815363, "grad_norm": 2.0210633277893066, "learning_rate": 1.531979990180199e-06, "loss": 0.07113553285598755, "step": 31390 }, { "epoch": 0.827123867726035, "grad_norm": 2.5620195865631104, "learning_rate": 1.5274451650680066e-06, "loss": 0.08590147495269776, "step": 31400 }, { "epoch": 0.8273872829705338, "grad_norm": 2.4031801223754883, "learning_rate": 1.5229165066467632e-06, "loss": 0.07585285902023316, "step": 31410 }, { "epoch": 0.8276506982150325, "grad_norm": 1.8299903869628906, "learning_rate": 1.5183940182126189e-06, "loss": 0.07332787513732911, "step": 31420 }, { "epoch": 0.8279141134595311, "grad_norm": 2.3855857849121094, "learning_rate": 1.513877703057235e-06, "loss": 0.0693997859954834, "step": 31430 }, { "epoch": 0.8281775287040299, "grad_norm": 2.423375368118286, "learning_rate": 1.509367564467782e-06, "loss": 0.07384347915649414, "step": 31440 }, { "epoch": 0.8284409439485286, "grad_norm": 2.839118003845215, "learning_rate": 1.5048636057269294e-06, "loss": 0.08622643947601319, "step": 31450 }, { "epoch": 0.8287043591930274, "grad_norm": 2.613755464553833, "learning_rate": 1.50036583011286e-06, "loss": 0.07213656902313233, "step": 31460 }, { "epoch": 0.8289677744375261, "grad_norm": 2.142578125, "learning_rate": 1.4958742408992454e-06, "loss": 0.07515689134597778, "step": 31470 }, { "epoch": 0.8292311896820249, "grad_norm": 2.345771074295044, "learning_rate": 1.4913888413552536e-06, "loss": 0.06564115285873413, "step": 31480 }, { "epoch": 0.8294946049265236, "grad_norm": 2.7693045139312744, "learning_rate": 1.486909634745557e-06, "loss": 0.0764570951461792, "step": 31490 }, { "epoch": 0.8297580201710223, "grad_norm": 2.2076003551483154, "learning_rate": 1.4824366243303156e-06, "loss": 0.07197414636611939, "step": 31500 }, { "epoch": 0.8300214354155211, "grad_norm": 2.403532028198242, "learning_rate": 1.4779698133651766e-06, "loss": 0.07789269089698792, "step": 31510 }, { "epoch": 0.8302848506600198, "grad_norm": 3.5407114028930664, "learning_rate": 1.47350920510128e-06, "loss": 0.07363196611404418, "step": 31520 }, { "epoch": 0.8305482659045186, "grad_norm": 3.04248046875, "learning_rate": 1.469054802785246e-06, "loss": 0.0852239489555359, "step": 31530 }, { "epoch": 0.8308116811490173, "grad_norm": 3.095937490463257, "learning_rate": 1.4646066096591838e-06, "loss": 0.08555875420570373, "step": 31540 }, { "epoch": 0.8310750963935161, "grad_norm": 3.0640997886657715, "learning_rate": 1.4601646289606809e-06, "loss": 0.07994820475578308, "step": 31550 }, { "epoch": 0.8313385116380148, "grad_norm": 2.604940176010132, "learning_rate": 1.4557288639227995e-06, "loss": 0.06747153401374817, "step": 31560 }, { "epoch": 0.8316019268825136, "grad_norm": 3.3648035526275635, "learning_rate": 1.451299317774083e-06, "loss": 0.07508964538574218, "step": 31570 }, { "epoch": 0.8318653421270122, "grad_norm": 3.034281015396118, "learning_rate": 1.4468759937385423e-06, "loss": 0.06892810463905334, "step": 31580 }, { "epoch": 0.8321287573715109, "grad_norm": 2.5103142261505127, "learning_rate": 1.4424588950356688e-06, "loss": 0.07576513886451722, "step": 31590 }, { "epoch": 0.8323921726160097, "grad_norm": 2.8474769592285156, "learning_rate": 1.4380480248804185e-06, "loss": 0.10034613609313965, "step": 31600 }, { "epoch": 0.8326555878605084, "grad_norm": 4.289809703826904, "learning_rate": 1.4336433864832033e-06, "loss": 0.0870785653591156, "step": 31610 }, { "epoch": 0.8329190031050072, "grad_norm": 3.237309217453003, "learning_rate": 1.4292449830499189e-06, "loss": 0.07913011312484741, "step": 31620 }, { "epoch": 0.8331824183495059, "grad_norm": 3.0890893936157227, "learning_rate": 1.4248528177819098e-06, "loss": 0.0774227499961853, "step": 31630 }, { "epoch": 0.8334458335940047, "grad_norm": 3.7110888957977295, "learning_rate": 1.4204668938759835e-06, "loss": 0.07211633920669555, "step": 31640 }, { "epoch": 0.8337092488385034, "grad_norm": 3.6271610260009766, "learning_rate": 1.416087214524403e-06, "loss": 0.07707797288894654, "step": 31650 }, { "epoch": 0.8339726640830022, "grad_norm": 3.7420458793640137, "learning_rate": 1.4117137829148885e-06, "loss": 0.08244045376777649, "step": 31660 }, { "epoch": 0.8342360793275009, "grad_norm": 1.8564256429672241, "learning_rate": 1.4073466022306148e-06, "loss": 0.0716295063495636, "step": 31670 }, { "epoch": 0.8344994945719996, "grad_norm": 3.6229424476623535, "learning_rate": 1.4029856756502026e-06, "loss": 0.08521056771278382, "step": 31680 }, { "epoch": 0.8347629098164984, "grad_norm": 2.9741287231445312, "learning_rate": 1.3986310063477215e-06, "loss": 0.08365689516067505, "step": 31690 }, { "epoch": 0.835026325060997, "grad_norm": 2.4078640937805176, "learning_rate": 1.3942825974926888e-06, "loss": 0.08176717162132263, "step": 31700 }, { "epoch": 0.8352897403054959, "grad_norm": 1.6723438501358032, "learning_rate": 1.3899404522500626e-06, "loss": 0.07886381149291992, "step": 31710 }, { "epoch": 0.8355531555499945, "grad_norm": 2.424121379852295, "learning_rate": 1.385604573780248e-06, "loss": 0.06928369998931885, "step": 31720 }, { "epoch": 0.8358165707944933, "grad_norm": 2.432534694671631, "learning_rate": 1.38127496523908e-06, "loss": 0.08582148551940919, "step": 31730 }, { "epoch": 0.836079986038992, "grad_norm": 2.307237386703491, "learning_rate": 1.376951629777833e-06, "loss": 0.07592177391052246, "step": 31740 }, { "epoch": 0.8363434012834908, "grad_norm": 2.152128219604492, "learning_rate": 1.372634570543222e-06, "loss": 0.07915083169937134, "step": 31750 }, { "epoch": 0.8366068165279895, "grad_norm": 2.407851457595825, "learning_rate": 1.3683237906773873e-06, "loss": 0.08328071832656861, "step": 31760 }, { "epoch": 0.8368702317724882, "grad_norm": 2.4233829975128174, "learning_rate": 1.3640192933178997e-06, "loss": 0.07624136209487915, "step": 31770 }, { "epoch": 0.837133647016987, "grad_norm": 2.654672145843506, "learning_rate": 1.359721081597758e-06, "loss": 0.08158130645751953, "step": 31780 }, { "epoch": 0.8373970622614857, "grad_norm": 2.137094497680664, "learning_rate": 1.355429158645384e-06, "loss": 0.07758219242095947, "step": 31790 }, { "epoch": 0.8376604775059845, "grad_norm": 2.0634090900421143, "learning_rate": 1.3511435275846297e-06, "loss": 0.09059802293777466, "step": 31800 }, { "epoch": 0.8379238927504832, "grad_norm": 3.6578803062438965, "learning_rate": 1.3468641915347613e-06, "loss": 0.07989699840545654, "step": 31810 }, { "epoch": 0.838187307994982, "grad_norm": 3.437450408935547, "learning_rate": 1.3425911536104573e-06, "loss": 0.07572497129440307, "step": 31820 }, { "epoch": 0.8384507232394807, "grad_norm": 2.6099154949188232, "learning_rate": 1.3383244169218268e-06, "loss": 0.07069239616394044, "step": 31830 }, { "epoch": 0.8387141384839795, "grad_norm": 3.4807393550872803, "learning_rate": 1.3340639845743797e-06, "loss": 0.08829448223114014, "step": 31840 }, { "epoch": 0.8389775537284782, "grad_norm": 2.827934980392456, "learning_rate": 1.32980985966905e-06, "loss": 0.08687688112258911, "step": 31850 }, { "epoch": 0.8392409689729768, "grad_norm": 3.7298364639282227, "learning_rate": 1.3255620453021656e-06, "loss": 0.07962723970413207, "step": 31860 }, { "epoch": 0.8395043842174756, "grad_norm": 3.058321475982666, "learning_rate": 1.3213205445654688e-06, "loss": 0.07420924901962281, "step": 31870 }, { "epoch": 0.8397677994619743, "grad_norm": 2.1176533699035645, "learning_rate": 1.3170853605461132e-06, "loss": 0.07457836270332337, "step": 31880 }, { "epoch": 0.8400312147064731, "grad_norm": 3.0203163623809814, "learning_rate": 1.3128564963266444e-06, "loss": 0.08041719198226929, "step": 31890 }, { "epoch": 0.8402946299509718, "grad_norm": 3.4591832160949707, "learning_rate": 1.3086339549850125e-06, "loss": 0.08158789277076721, "step": 31900 }, { "epoch": 0.8405580451954706, "grad_norm": 2.7237014770507812, "learning_rate": 1.3044177395945657e-06, "loss": 0.0721343219280243, "step": 31910 }, { "epoch": 0.8408214604399693, "grad_norm": 1.8415241241455078, "learning_rate": 1.3002078532240447e-06, "loss": 0.08380599617958069, "step": 31920 }, { "epoch": 0.8410848756844681, "grad_norm": 2.6546661853790283, "learning_rate": 1.296004298937592e-06, "loss": 0.07872028350830078, "step": 31930 }, { "epoch": 0.8413482909289668, "grad_norm": 2.766887664794922, "learning_rate": 1.2918070797947346e-06, "loss": 0.05872756242752075, "step": 31940 }, { "epoch": 0.8416117061734655, "grad_norm": 2.0296881198883057, "learning_rate": 1.2876161988503843e-06, "loss": 0.07991015315055847, "step": 31950 }, { "epoch": 0.8418751214179643, "grad_norm": 3.219862222671509, "learning_rate": 1.2834316591548502e-06, "loss": 0.07881994247436523, "step": 31960 }, { "epoch": 0.842138536662463, "grad_norm": 3.5728538036346436, "learning_rate": 1.279253463753819e-06, "loss": 0.08005270957946778, "step": 31970 }, { "epoch": 0.8424019519069618, "grad_norm": 3.1919608116149902, "learning_rate": 1.2750816156883627e-06, "loss": 0.0800094723701477, "step": 31980 }, { "epoch": 0.8426653671514605, "grad_norm": 2.9237425327301025, "learning_rate": 1.2709161179949303e-06, "loss": 0.0719909906387329, "step": 31990 }, { "epoch": 0.8429287823959593, "grad_norm": 2.294107675552368, "learning_rate": 1.266756973705351e-06, "loss": 0.07485975623130799, "step": 32000 }, { "epoch": 0.8431921976404579, "grad_norm": 2.946704149246216, "learning_rate": 1.2626041858468318e-06, "loss": 0.07967197895050049, "step": 32010 }, { "epoch": 0.8434556128849566, "grad_norm": 3.922823429107666, "learning_rate": 1.2584577574419499e-06, "loss": 0.06714965701103211, "step": 32020 }, { "epoch": 0.8437190281294554, "grad_norm": 3.1492321491241455, "learning_rate": 1.2543176915086542e-06, "loss": 0.0855346143245697, "step": 32030 }, { "epoch": 0.8439824433739541, "grad_norm": 3.3104634284973145, "learning_rate": 1.2501839910602632e-06, "loss": 0.07454878091812134, "step": 32040 }, { "epoch": 0.8442458586184529, "grad_norm": 2.8406167030334473, "learning_rate": 1.24605665910546e-06, "loss": 0.08480717539787293, "step": 32050 }, { "epoch": 0.8445092738629516, "grad_norm": 2.3607027530670166, "learning_rate": 1.2419356986483e-06, "loss": 0.071519535779953, "step": 32060 }, { "epoch": 0.8447726891074504, "grad_norm": 2.2984251976013184, "learning_rate": 1.2378211126881956e-06, "loss": 0.07435768246650695, "step": 32070 }, { "epoch": 0.8450361043519491, "grad_norm": 3.555785655975342, "learning_rate": 1.2337129042199147e-06, "loss": 0.0747345507144928, "step": 32080 }, { "epoch": 0.8452995195964479, "grad_norm": 2.720135450363159, "learning_rate": 1.2296110762335933e-06, "loss": 0.06779175400733947, "step": 32090 }, { "epoch": 0.8455629348409466, "grad_norm": 2.664417028427124, "learning_rate": 1.2255156317147165e-06, "loss": 0.07179308533668519, "step": 32100 }, { "epoch": 0.8458263500854453, "grad_norm": 2.3127641677856445, "learning_rate": 1.2214265736441277e-06, "loss": 0.07628662586212158, "step": 32110 }, { "epoch": 0.8460897653299441, "grad_norm": 3.3037662506103516, "learning_rate": 1.2173439049980174e-06, "loss": 0.07470242381095886, "step": 32120 }, { "epoch": 0.8463531805744428, "grad_norm": 2.815647840499878, "learning_rate": 1.2132676287479273e-06, "loss": 0.06733776926994324, "step": 32130 }, { "epoch": 0.8466165958189416, "grad_norm": 3.617751121520996, "learning_rate": 1.209197747860752e-06, "loss": 0.0788641095161438, "step": 32140 }, { "epoch": 0.8468800110634402, "grad_norm": 2.682581901550293, "learning_rate": 1.2051342652987218e-06, "loss": 0.07412976026535034, "step": 32150 }, { "epoch": 0.847143426307939, "grad_norm": 3.636383533477783, "learning_rate": 1.2010771840194168e-06, "loss": 0.07233170866966247, "step": 32160 }, { "epoch": 0.8474068415524377, "grad_norm": 2.817136287689209, "learning_rate": 1.1970265069757548e-06, "loss": 0.0799074113368988, "step": 32170 }, { "epoch": 0.8476702567969365, "grad_norm": 2.857727289199829, "learning_rate": 1.1929822371159915e-06, "loss": 0.07596198320388795, "step": 32180 }, { "epoch": 0.8479336720414352, "grad_norm": 2.153958797454834, "learning_rate": 1.188944377383725e-06, "loss": 0.07810769081115723, "step": 32190 }, { "epoch": 0.8481970872859339, "grad_norm": 1.4313136339187622, "learning_rate": 1.1849129307178841e-06, "loss": 0.07334248423576355, "step": 32200 }, { "epoch": 0.8484605025304327, "grad_norm": 2.6865580081939697, "learning_rate": 1.180887900052724e-06, "loss": 0.07066527605056763, "step": 32210 }, { "epoch": 0.8487239177749314, "grad_norm": 3.959764003753662, "learning_rate": 1.17686928831784e-06, "loss": 0.08197015523910522, "step": 32220 }, { "epoch": 0.8489873330194302, "grad_norm": 2.2939136028289795, "learning_rate": 1.1728570984381505e-06, "loss": 0.08295502662658691, "step": 32230 }, { "epoch": 0.8492507482639289, "grad_norm": 3.2396838665008545, "learning_rate": 1.1688513333339003e-06, "loss": 0.08414174318313598, "step": 32240 }, { "epoch": 0.8495141635084277, "grad_norm": 2.5473392009735107, "learning_rate": 1.1648519959206584e-06, "loss": 0.07467130422592164, "step": 32250 }, { "epoch": 0.8497775787529264, "grad_norm": 4.2792463302612305, "learning_rate": 1.1608590891093118e-06, "loss": 0.07301737666130066, "step": 32260 }, { "epoch": 0.8500409939974252, "grad_norm": 3.3999643325805664, "learning_rate": 1.156872615806075e-06, "loss": 0.07685568332672119, "step": 32270 }, { "epoch": 0.8503044092419239, "grad_norm": 2.8921515941619873, "learning_rate": 1.1528925789124734e-06, "loss": 0.06868394017219544, "step": 32280 }, { "epoch": 0.8505678244864225, "grad_norm": 2.1199822425842285, "learning_rate": 1.1489189813253499e-06, "loss": 0.07982059717178344, "step": 32290 }, { "epoch": 0.8508312397309213, "grad_norm": 3.0946640968322754, "learning_rate": 1.1449518259368596e-06, "loss": 0.07156882882118225, "step": 32300 }, { "epoch": 0.85109465497542, "grad_norm": 3.0856423377990723, "learning_rate": 1.1409911156344677e-06, "loss": 0.07336493730545043, "step": 32310 }, { "epoch": 0.8513580702199188, "grad_norm": 2.7530407905578613, "learning_rate": 1.1370368533009557e-06, "loss": 0.06889991760253907, "step": 32320 }, { "epoch": 0.8516214854644175, "grad_norm": 2.9834237098693848, "learning_rate": 1.1330890418144014e-06, "loss": 0.06588118076324463, "step": 32330 }, { "epoch": 0.8518849007089163, "grad_norm": 5.059815406799316, "learning_rate": 1.1291476840481908e-06, "loss": 0.07939664721488952, "step": 32340 }, { "epoch": 0.852148315953415, "grad_norm": 2.8766307830810547, "learning_rate": 1.125212782871019e-06, "loss": 0.07311562895774841, "step": 32350 }, { "epoch": 0.8524117311979138, "grad_norm": 3.4781382083892822, "learning_rate": 1.1212843411468755e-06, "loss": 0.07507489323616028, "step": 32360 }, { "epoch": 0.8526751464424125, "grad_norm": 4.22849178314209, "learning_rate": 1.1173623617350472e-06, "loss": 0.07204790115356445, "step": 32370 }, { "epoch": 0.8529385616869112, "grad_norm": 3.0287911891937256, "learning_rate": 1.1134468474901227e-06, "loss": 0.07306666374206543, "step": 32380 }, { "epoch": 0.85320197693141, "grad_norm": 3.3242549896240234, "learning_rate": 1.1095378012619796e-06, "loss": 0.07144333124160766, "step": 32390 }, { "epoch": 0.8534653921759087, "grad_norm": 2.0499024391174316, "learning_rate": 1.1056352258957924e-06, "loss": 0.06988563537597656, "step": 32400 }, { "epoch": 0.8537288074204075, "grad_norm": 2.9381160736083984, "learning_rate": 1.1017391242320263e-06, "loss": 0.06605163216590881, "step": 32410 }, { "epoch": 0.8539922226649062, "grad_norm": 3.4681272506713867, "learning_rate": 1.0978494991064248e-06, "loss": 0.07857808470726013, "step": 32420 }, { "epoch": 0.854255637909405, "grad_norm": 2.376750946044922, "learning_rate": 1.0939663533500311e-06, "loss": 0.0639781892299652, "step": 32430 }, { "epoch": 0.8545190531539036, "grad_norm": 2.45635724067688, "learning_rate": 1.0900896897891634e-06, "loss": 0.0791463553905487, "step": 32440 }, { "epoch": 0.8547824683984024, "grad_norm": 3.033156633377075, "learning_rate": 1.0862195112454289e-06, "loss": 0.08042479753494262, "step": 32450 }, { "epoch": 0.8550458836429011, "grad_norm": 2.662323236465454, "learning_rate": 1.0823558205357054e-06, "loss": 0.06874158382415771, "step": 32460 }, { "epoch": 0.8553092988873998, "grad_norm": 3.353233575820923, "learning_rate": 1.0784986204721537e-06, "loss": 0.0773589015007019, "step": 32470 }, { "epoch": 0.8555727141318986, "grad_norm": 3.6894309520721436, "learning_rate": 1.0746479138622134e-06, "loss": 0.07428136467933655, "step": 32480 }, { "epoch": 0.8558361293763973, "grad_norm": 3.3276314735412598, "learning_rate": 1.070803703508595e-06, "loss": 0.07699465155601501, "step": 32490 }, { "epoch": 0.8560995446208961, "grad_norm": 2.4953818321228027, "learning_rate": 1.066965992209279e-06, "loss": 0.0642586588859558, "step": 32500 }, { "epoch": 0.8563629598653948, "grad_norm": 3.214146137237549, "learning_rate": 1.0631347827575179e-06, "loss": 0.0679144024848938, "step": 32510 }, { "epoch": 0.8566263751098936, "grad_norm": 4.75848913192749, "learning_rate": 1.0593100779418286e-06, "loss": 0.07684834003448486, "step": 32520 }, { "epoch": 0.8568897903543923, "grad_norm": 3.857362985610962, "learning_rate": 1.0554918805460013e-06, "loss": 0.08826611638069153, "step": 32530 }, { "epoch": 0.8571532055988911, "grad_norm": 2.4975743293762207, "learning_rate": 1.051680193349084e-06, "loss": 0.06730028986930847, "step": 32540 }, { "epoch": 0.8574166208433898, "grad_norm": 1.9754375219345093, "learning_rate": 1.0478750191253817e-06, "loss": 0.07581037282943726, "step": 32550 }, { "epoch": 0.8576800360878885, "grad_norm": 2.7509713172912598, "learning_rate": 1.0440763606444694e-06, "loss": 0.06983497142791747, "step": 32560 }, { "epoch": 0.8579434513323873, "grad_norm": 2.438183069229126, "learning_rate": 1.0402842206711728e-06, "loss": 0.08304935693740845, "step": 32570 }, { "epoch": 0.8582068665768859, "grad_norm": 4.41560697555542, "learning_rate": 1.0364986019655765e-06, "loss": 0.07726389765739441, "step": 32580 }, { "epoch": 0.8584702818213847, "grad_norm": 3.5782546997070312, "learning_rate": 1.0327195072830153e-06, "loss": 0.08008349537849427, "step": 32590 }, { "epoch": 0.8587336970658834, "grad_norm": 3.208392858505249, "learning_rate": 1.028946939374076e-06, "loss": 0.07979797720909118, "step": 32600 }, { "epoch": 0.8589971123103822, "grad_norm": 3.6212551593780518, "learning_rate": 1.0251809009846003e-06, "loss": 0.072353994846344, "step": 32610 }, { "epoch": 0.8592605275548809, "grad_norm": 2.364478349685669, "learning_rate": 1.0214213948556718e-06, "loss": 0.06307721138000488, "step": 32620 }, { "epoch": 0.8595239427993796, "grad_norm": 2.933319091796875, "learning_rate": 1.0176684237236212e-06, "loss": 0.08033392429351807, "step": 32630 }, { "epoch": 0.8597873580438784, "grad_norm": 2.989347457885742, "learning_rate": 1.0139219903200227e-06, "loss": 0.08320461511611939, "step": 32640 }, { "epoch": 0.8600507732883771, "grad_norm": 3.1189169883728027, "learning_rate": 1.0101820973716891e-06, "loss": 0.07587476968765258, "step": 32650 }, { "epoch": 0.8603141885328759, "grad_norm": 3.4196536540985107, "learning_rate": 1.0064487476006812e-06, "loss": 0.07188872694969177, "step": 32660 }, { "epoch": 0.8605776037773746, "grad_norm": 2.318359375, "learning_rate": 1.0027219437242907e-06, "loss": 0.07469532489776612, "step": 32670 }, { "epoch": 0.8608410190218734, "grad_norm": 3.4026944637298584, "learning_rate": 9.990016884550425e-07, "loss": 0.0699702799320221, "step": 32680 }, { "epoch": 0.8611044342663721, "grad_norm": 2.84324049949646, "learning_rate": 9.952879845007024e-07, "loss": 0.06667369604110718, "step": 32690 }, { "epoch": 0.8613678495108709, "grad_norm": 3.607266664505005, "learning_rate": 9.915808345642642e-07, "loss": 0.06946250796318054, "step": 32700 }, { "epoch": 0.8616312647553696, "grad_norm": 3.4601194858551025, "learning_rate": 9.878802413439514e-07, "loss": 0.07183845639228821, "step": 32710 }, { "epoch": 0.8618946799998682, "grad_norm": 3.4299354553222656, "learning_rate": 9.841862075332143e-07, "loss": 0.07802043557167053, "step": 32720 }, { "epoch": 0.862158095244367, "grad_norm": 1.9996025562286377, "learning_rate": 9.804987358207274e-07, "loss": 0.0642418920993805, "step": 32730 }, { "epoch": 0.8624215104888657, "grad_norm": 2.3252217769622803, "learning_rate": 9.768178288903985e-07, "loss": 0.06986652016639709, "step": 32740 }, { "epoch": 0.8626849257333645, "grad_norm": 3.5909101963043213, "learning_rate": 9.731434894213465e-07, "loss": 0.07807788848876954, "step": 32750 }, { "epoch": 0.8629483409778632, "grad_norm": 2.7963268756866455, "learning_rate": 9.694757200879146e-07, "loss": 0.07899802923202515, "step": 32760 }, { "epoch": 0.863211756222362, "grad_norm": 2.5288727283477783, "learning_rate": 9.658145235596628e-07, "loss": 0.07504598498344421, "step": 32770 }, { "epoch": 0.8634751714668607, "grad_norm": 3.9462859630584717, "learning_rate": 9.621599025013672e-07, "loss": 0.08048299551010132, "step": 32780 }, { "epoch": 0.8637385867113595, "grad_norm": 3.3194799423217773, "learning_rate": 9.585118595730226e-07, "loss": 0.0754865825176239, "step": 32790 }, { "epoch": 0.8640020019558582, "grad_norm": 2.8680734634399414, "learning_rate": 9.548703974298311e-07, "loss": 0.07574642300605774, "step": 32800 }, { "epoch": 0.8642654172003569, "grad_norm": 3.9168930053710938, "learning_rate": 9.512355187222e-07, "loss": 0.07066202163696289, "step": 32810 }, { "epoch": 0.8645288324448557, "grad_norm": 2.2803614139556885, "learning_rate": 9.47607226095758e-07, "loss": 0.07808729410171508, "step": 32820 }, { "epoch": 0.8647922476893544, "grad_norm": 3.5153517723083496, "learning_rate": 9.439855221913297e-07, "loss": 0.0724899172782898, "step": 32830 }, { "epoch": 0.8650556629338532, "grad_norm": 3.361056089401245, "learning_rate": 9.403704096449484e-07, "loss": 0.074041748046875, "step": 32840 }, { "epoch": 0.8653190781783519, "grad_norm": 3.3473711013793945, "learning_rate": 9.367618910878484e-07, "loss": 0.08084893226623535, "step": 32850 }, { "epoch": 0.8655824934228507, "grad_norm": 3.438089609146118, "learning_rate": 9.331599691464644e-07, "loss": 0.08615338802337646, "step": 32860 }, { "epoch": 0.8658459086673493, "grad_norm": 2.6230509281158447, "learning_rate": 9.295646464424323e-07, "loss": 0.06694923043251037, "step": 32870 }, { "epoch": 0.8661093239118481, "grad_norm": 3.5781607627868652, "learning_rate": 9.259759255925827e-07, "loss": 0.07004954814910888, "step": 32880 }, { "epoch": 0.8663727391563468, "grad_norm": 3.8044698238372803, "learning_rate": 9.223938092089413e-07, "loss": 0.0748827576637268, "step": 32890 }, { "epoch": 0.8666361544008455, "grad_norm": 3.648385763168335, "learning_rate": 9.188182998987261e-07, "loss": 0.06922022104263306, "step": 32900 }, { "epoch": 0.8668995696453443, "grad_norm": 2.4185309410095215, "learning_rate": 9.152494002643464e-07, "loss": 0.06684056520462037, "step": 32910 }, { "epoch": 0.867162984889843, "grad_norm": 3.6789352893829346, "learning_rate": 9.116871129034066e-07, "loss": 0.07406472563743591, "step": 32920 }, { "epoch": 0.8674264001343418, "grad_norm": 2.3124654293060303, "learning_rate": 9.081314404086872e-07, "loss": 0.06990896463394165, "step": 32930 }, { "epoch": 0.8676898153788405, "grad_norm": 2.3642146587371826, "learning_rate": 9.045823853681602e-07, "loss": 0.07337081432342529, "step": 32940 }, { "epoch": 0.8679532306233393, "grad_norm": 2.313554048538208, "learning_rate": 9.010399503649836e-07, "loss": 0.06669379472732544, "step": 32950 }, { "epoch": 0.868216645867838, "grad_norm": 2.930901288986206, "learning_rate": 8.975041379774951e-07, "loss": 0.0856128454208374, "step": 32960 }, { "epoch": 0.8684800611123368, "grad_norm": 3.4119811058044434, "learning_rate": 8.939749507792094e-07, "loss": 0.08219394683837891, "step": 32970 }, { "epoch": 0.8687434763568355, "grad_norm": 3.1517527103424072, "learning_rate": 8.904523913388218e-07, "loss": 0.06213920116424561, "step": 32980 }, { "epoch": 0.8690068916013342, "grad_norm": 2.621286392211914, "learning_rate": 8.869364622202015e-07, "loss": 0.06817446947097779, "step": 32990 }, { "epoch": 0.869270306845833, "grad_norm": 3.2995989322662354, "learning_rate": 8.834271659823967e-07, "loss": 0.07792847156524658, "step": 33000 }, { "epoch": 0.8695337220903316, "grad_norm": 2.5079538822174072, "learning_rate": 8.799245051796246e-07, "loss": 0.0718591868877411, "step": 33010 }, { "epoch": 0.8697971373348304, "grad_norm": 2.775405168533325, "learning_rate": 8.764284823612667e-07, "loss": 0.07840514183044434, "step": 33020 }, { "epoch": 0.8700605525793291, "grad_norm": 2.5182790756225586, "learning_rate": 8.729391000718867e-07, "loss": 0.08306746482849121, "step": 33030 }, { "epoch": 0.8703239678238279, "grad_norm": 2.425018787384033, "learning_rate": 8.694563608512041e-07, "loss": 0.07540307044982911, "step": 33040 }, { "epoch": 0.8705873830683266, "grad_norm": 2.9030232429504395, "learning_rate": 8.659802672341078e-07, "loss": 0.0717666506767273, "step": 33050 }, { "epoch": 0.8708507983128254, "grad_norm": 2.601390838623047, "learning_rate": 8.625108217506472e-07, "loss": 0.06940147876739503, "step": 33060 }, { "epoch": 0.8711142135573241, "grad_norm": 3.1758368015289307, "learning_rate": 8.590480269260337e-07, "loss": 0.06856483221054077, "step": 33070 }, { "epoch": 0.8713776288018228, "grad_norm": 3.401982307434082, "learning_rate": 8.555918852806421e-07, "loss": 0.07502508163452148, "step": 33080 }, { "epoch": 0.8716410440463216, "grad_norm": 2.8658010959625244, "learning_rate": 8.521423993300005e-07, "loss": 0.08143857717514039, "step": 33090 }, { "epoch": 0.8719044592908203, "grad_norm": 3.332559585571289, "learning_rate": 8.486995715847923e-07, "loss": 0.06530752182006835, "step": 33100 }, { "epoch": 0.8721678745353191, "grad_norm": 3.729301929473877, "learning_rate": 8.452634045508557e-07, "loss": 0.07090425491333008, "step": 33110 }, { "epoch": 0.8724312897798178, "grad_norm": 3.2632954120635986, "learning_rate": 8.41833900729182e-07, "loss": 0.07650619745254517, "step": 33120 }, { "epoch": 0.8726947050243166, "grad_norm": 2.8977363109588623, "learning_rate": 8.384110626159126e-07, "loss": 0.07774307131767273, "step": 33130 }, { "epoch": 0.8729581202688153, "grad_norm": 2.386173963546753, "learning_rate": 8.349948927023388e-07, "loss": 0.06905542612075806, "step": 33140 }, { "epoch": 0.8732215355133139, "grad_norm": 2.8985536098480225, "learning_rate": 8.315853934748908e-07, "loss": 0.0811755895614624, "step": 33150 }, { "epoch": 0.8734849507578127, "grad_norm": 2.7298312187194824, "learning_rate": 8.281825674151533e-07, "loss": 0.07457337379455567, "step": 33160 }, { "epoch": 0.8737483660023114, "grad_norm": 2.0452640056610107, "learning_rate": 8.247864169998499e-07, "loss": 0.06374762058258057, "step": 33170 }, { "epoch": 0.8740117812468102, "grad_norm": 2.99220609664917, "learning_rate": 8.213969447008441e-07, "loss": 0.06894598603248596, "step": 33180 }, { "epoch": 0.8742751964913089, "grad_norm": 3.3707077503204346, "learning_rate": 8.180141529851404e-07, "loss": 0.07539301514625549, "step": 33190 }, { "epoch": 0.8745386117358077, "grad_norm": 3.2059578895568848, "learning_rate": 8.146380443148794e-07, "loss": 0.06829622387886047, "step": 33200 }, { "epoch": 0.8748020269803064, "grad_norm": 4.515559673309326, "learning_rate": 8.11268621147342e-07, "loss": 0.07628743648529053, "step": 33210 }, { "epoch": 0.8750654422248052, "grad_norm": 2.8323380947113037, "learning_rate": 8.079058859349364e-07, "loss": 0.0798027515411377, "step": 33220 }, { "epoch": 0.8753288574693039, "grad_norm": 2.560957670211792, "learning_rate": 8.045498411252095e-07, "loss": 0.07174152135848999, "step": 33230 }, { "epoch": 0.8755922727138026, "grad_norm": 4.660841464996338, "learning_rate": 8.012004891608327e-07, "loss": 0.07500030398368836, "step": 33240 }, { "epoch": 0.8758556879583014, "grad_norm": 2.784461736679077, "learning_rate": 7.978578324796072e-07, "loss": 0.07294382452964783, "step": 33250 }, { "epoch": 0.8761191032028001, "grad_norm": 3.064830780029297, "learning_rate": 7.945218735144689e-07, "loss": 0.06593897938728333, "step": 33260 }, { "epoch": 0.8763825184472989, "grad_norm": 2.7747554779052734, "learning_rate": 7.91192614693469e-07, "loss": 0.06587000489234925, "step": 33270 }, { "epoch": 0.8766459336917976, "grad_norm": 2.8992388248443604, "learning_rate": 7.878700584397836e-07, "loss": 0.06681081652641296, "step": 33280 }, { "epoch": 0.8769093489362964, "grad_norm": 3.999875545501709, "learning_rate": 7.84554207171716e-07, "loss": 0.06380292773246765, "step": 33290 }, { "epoch": 0.877172764180795, "grad_norm": 3.1548070907592773, "learning_rate": 7.812450633026847e-07, "loss": 0.06231154203414917, "step": 33300 }, { "epoch": 0.8774361794252938, "grad_norm": 2.9885826110839844, "learning_rate": 7.779426292412273e-07, "loss": 0.06868510842323303, "step": 33310 }, { "epoch": 0.8776995946697925, "grad_norm": 2.9547629356384277, "learning_rate": 7.746469073909979e-07, "loss": 0.06733212471008301, "step": 33320 }, { "epoch": 0.8779630099142912, "grad_norm": 2.7990283966064453, "learning_rate": 7.713579001507632e-07, "loss": 0.08172173500061035, "step": 33330 }, { "epoch": 0.87822642515879, "grad_norm": 2.469426393508911, "learning_rate": 7.680756099144072e-07, "loss": 0.07236613035202026, "step": 33340 }, { "epoch": 0.8784898404032887, "grad_norm": 3.0244252681732178, "learning_rate": 7.648000390709221e-07, "loss": 0.07958743572235108, "step": 33350 }, { "epoch": 0.8787532556477875, "grad_norm": 2.566596508026123, "learning_rate": 7.615311900044087e-07, "loss": 0.06390266418457032, "step": 33360 }, { "epoch": 0.8790166708922862, "grad_norm": 4.94434118270874, "learning_rate": 7.582690650940771e-07, "loss": 0.06655697822570801, "step": 33370 }, { "epoch": 0.879280086136785, "grad_norm": 2.966273307800293, "learning_rate": 7.550136667142393e-07, "loss": 0.06958682537078857, "step": 33380 }, { "epoch": 0.8795435013812837, "grad_norm": 2.667632818222046, "learning_rate": 7.517649972343199e-07, "loss": 0.06810912489891052, "step": 33390 }, { "epoch": 0.8798069166257825, "grad_norm": 2.4089410305023193, "learning_rate": 7.485230590188397e-07, "loss": 0.06016781330108643, "step": 33400 }, { "epoch": 0.8800703318702812, "grad_norm": 3.676494598388672, "learning_rate": 7.452878544274167e-07, "loss": 0.07624058723449707, "step": 33410 }, { "epoch": 0.8803337471147799, "grad_norm": 3.0725598335266113, "learning_rate": 7.420593858147773e-07, "loss": 0.06724740862846375, "step": 33420 }, { "epoch": 0.8805971623592787, "grad_norm": 2.7400333881378174, "learning_rate": 7.388376555307386e-07, "loss": 0.07760597467422485, "step": 33430 }, { "epoch": 0.8808605776037773, "grad_norm": 2.5694940090179443, "learning_rate": 7.356226659202137e-07, "loss": 0.07361149787902832, "step": 33440 }, { "epoch": 0.8811239928482761, "grad_norm": 2.4736170768737793, "learning_rate": 7.324144193232119e-07, "loss": 0.06509525775909424, "step": 33450 }, { "epoch": 0.8813874080927748, "grad_norm": 2.7572147846221924, "learning_rate": 7.29212918074833e-07, "loss": 0.06430892944335938, "step": 33460 }, { "epoch": 0.8816508233372736, "grad_norm": 2.8935787677764893, "learning_rate": 7.260181645052689e-07, "loss": 0.06922895908355713, "step": 33470 }, { "epoch": 0.8819142385817723, "grad_norm": 2.770494222640991, "learning_rate": 7.228301609397981e-07, "loss": 0.08088369965553284, "step": 33480 }, { "epoch": 0.8821776538262711, "grad_norm": 3.4571874141693115, "learning_rate": 7.196489096987869e-07, "loss": 0.079948091506958, "step": 33490 }, { "epoch": 0.8824410690707698, "grad_norm": 3.807070255279541, "learning_rate": 7.164744130976886e-07, "loss": 0.05441897511482239, "step": 33500 }, { "epoch": 0.8827044843152685, "grad_norm": 2.9099819660186768, "learning_rate": 7.133066734470351e-07, "loss": 0.07750643491744995, "step": 33510 }, { "epoch": 0.8829678995597673, "grad_norm": 3.3996591567993164, "learning_rate": 7.101456930524509e-07, "loss": 0.07054178714752198, "step": 33520 }, { "epoch": 0.883231314804266, "grad_norm": 2.9759418964385986, "learning_rate": 7.069914742146267e-07, "loss": 0.07596681118011475, "step": 33530 }, { "epoch": 0.8834947300487648, "grad_norm": 3.0630104541778564, "learning_rate": 7.038440192293416e-07, "loss": 0.06693424582481385, "step": 33540 }, { "epoch": 0.8837581452932635, "grad_norm": 2.7666029930114746, "learning_rate": 7.007033303874511e-07, "loss": 0.07410350441932678, "step": 33550 }, { "epoch": 0.8840215605377623, "grad_norm": 2.712813138961792, "learning_rate": 6.97569409974882e-07, "loss": 0.06801515817642212, "step": 33560 }, { "epoch": 0.884284975782261, "grad_norm": 4.315465450286865, "learning_rate": 6.944422602726375e-07, "loss": 0.05725525021553039, "step": 33570 }, { "epoch": 0.8845483910267598, "grad_norm": 2.3640921115875244, "learning_rate": 6.913218835567925e-07, "loss": 0.06908199191093445, "step": 33580 }, { "epoch": 0.8848118062712584, "grad_norm": 3.9279863834381104, "learning_rate": 6.882082820984892e-07, "loss": 0.06713714599609374, "step": 33590 }, { "epoch": 0.8850752215157571, "grad_norm": 2.9798614978790283, "learning_rate": 6.851014581639437e-07, "loss": 0.07080844044685364, "step": 33600 }, { "epoch": 0.8853386367602559, "grad_norm": 1.9928306341171265, "learning_rate": 6.820014140144382e-07, "loss": 0.06226721405982971, "step": 33610 }, { "epoch": 0.8856020520047546, "grad_norm": 3.2183823585510254, "learning_rate": 6.789081519063134e-07, "loss": 0.07229365110397339, "step": 33620 }, { "epoch": 0.8858654672492534, "grad_norm": 3.3109781742095947, "learning_rate": 6.758216740909829e-07, "loss": 0.08030664920806885, "step": 33630 }, { "epoch": 0.8861288824937521, "grad_norm": 3.5089664459228516, "learning_rate": 6.727419828149173e-07, "loss": 0.07495964765548706, "step": 33640 }, { "epoch": 0.8863922977382509, "grad_norm": 2.635664939880371, "learning_rate": 6.696690803196482e-07, "loss": 0.07335529327392579, "step": 33650 }, { "epoch": 0.8866557129827496, "grad_norm": 3.6699953079223633, "learning_rate": 6.666029688417674e-07, "loss": 0.06913385987281799, "step": 33660 }, { "epoch": 0.8869191282272483, "grad_norm": 1.7606924772262573, "learning_rate": 6.635436506129223e-07, "loss": 0.06278157234191895, "step": 33670 }, { "epoch": 0.8871825434717471, "grad_norm": 4.316891670227051, "learning_rate": 6.60491127859818e-07, "loss": 0.06737094521522521, "step": 33680 }, { "epoch": 0.8874459587162458, "grad_norm": 2.798102855682373, "learning_rate": 6.57445402804211e-07, "loss": 0.07180758714675903, "step": 33690 }, { "epoch": 0.8877093739607446, "grad_norm": 4.004476070404053, "learning_rate": 6.544064776629123e-07, "loss": 0.06530025601387024, "step": 33700 }, { "epoch": 0.8879727892052433, "grad_norm": 2.6998648643493652, "learning_rate": 6.513743546477824e-07, "loss": 0.056210416555404666, "step": 33710 }, { "epoch": 0.888236204449742, "grad_norm": 3.0208747386932373, "learning_rate": 6.483490359657296e-07, "loss": 0.07242718935012818, "step": 33720 }, { "epoch": 0.8884996196942407, "grad_norm": 2.423265218734741, "learning_rate": 6.453305238187147e-07, "loss": 0.0590227484703064, "step": 33730 }, { "epoch": 0.8887630349387395, "grad_norm": 2.981734275817871, "learning_rate": 6.423188204037412e-07, "loss": 0.07109463810920716, "step": 33740 }, { "epoch": 0.8890264501832382, "grad_norm": 3.6583681106567383, "learning_rate": 6.393139279128513e-07, "loss": 0.07007819414138794, "step": 33750 }, { "epoch": 0.8892898654277369, "grad_norm": 2.641749143600464, "learning_rate": 6.3631584853314e-07, "loss": 0.06810947656631469, "step": 33760 }, { "epoch": 0.8895532806722357, "grad_norm": 5.702767372131348, "learning_rate": 6.333245844467384e-07, "loss": 0.06707994937896729, "step": 33770 }, { "epoch": 0.8898166959167344, "grad_norm": 2.944956064224243, "learning_rate": 6.303401378308172e-07, "loss": 0.0649865746498108, "step": 33780 }, { "epoch": 0.8900801111612332, "grad_norm": 3.2795064449310303, "learning_rate": 6.273625108575843e-07, "loss": 0.06581043601036071, "step": 33790 }, { "epoch": 0.8903435264057319, "grad_norm": 2.3983020782470703, "learning_rate": 6.243917056942849e-07, "loss": 0.0638850212097168, "step": 33800 }, { "epoch": 0.8906069416502307, "grad_norm": 3.6216540336608887, "learning_rate": 6.214277245032019e-07, "loss": 0.06669334173202515, "step": 33810 }, { "epoch": 0.8908703568947294, "grad_norm": 2.609646797180176, "learning_rate": 6.184705694416471e-07, "loss": 0.06867830753326416, "step": 33820 }, { "epoch": 0.8911337721392282, "grad_norm": 2.813908100128174, "learning_rate": 6.155202426619655e-07, "loss": 0.06556869745254516, "step": 33830 }, { "epoch": 0.8913971873837269, "grad_norm": 3.4948315620422363, "learning_rate": 6.125767463115317e-07, "loss": 0.07150853872299194, "step": 33840 }, { "epoch": 0.8916606026282256, "grad_norm": 3.334435224533081, "learning_rate": 6.096400825327487e-07, "loss": 0.08023866415023803, "step": 33850 }, { "epoch": 0.8919240178727244, "grad_norm": 2.511286735534668, "learning_rate": 6.067102534630498e-07, "loss": 0.06771400570869446, "step": 33860 }, { "epoch": 0.892187433117223, "grad_norm": 3.1262454986572266, "learning_rate": 6.037872612348905e-07, "loss": 0.07396710515022278, "step": 33870 }, { "epoch": 0.8924508483617218, "grad_norm": 2.9600107669830322, "learning_rate": 6.008711079757467e-07, "loss": 0.06258314251899719, "step": 33880 }, { "epoch": 0.8927142636062205, "grad_norm": 2.9168734550476074, "learning_rate": 5.979617958081241e-07, "loss": 0.06708050966262817, "step": 33890 }, { "epoch": 0.8929776788507193, "grad_norm": 2.04771089553833, "learning_rate": 5.950593268495464e-07, "loss": 0.06274181008338928, "step": 33900 }, { "epoch": 0.893241094095218, "grad_norm": 2.1335015296936035, "learning_rate": 5.921637032125538e-07, "loss": 0.06191791892051697, "step": 33910 }, { "epoch": 0.8935045093397168, "grad_norm": 3.1331570148468018, "learning_rate": 5.89274927004706e-07, "loss": 0.06401211023330688, "step": 33920 }, { "epoch": 0.8937679245842155, "grad_norm": 2.456979513168335, "learning_rate": 5.863930003285779e-07, "loss": 0.05751658082008362, "step": 33930 }, { "epoch": 0.8940313398287142, "grad_norm": 2.6631906032562256, "learning_rate": 5.835179252817635e-07, "loss": 0.062140464782714844, "step": 33940 }, { "epoch": 0.894294755073213, "grad_norm": 2.8506836891174316, "learning_rate": 5.806497039568648e-07, "loss": 0.06735683679580688, "step": 33950 }, { "epoch": 0.8945581703177117, "grad_norm": 1.5776642560958862, "learning_rate": 5.777883384414973e-07, "loss": 0.06483604311943054, "step": 33960 }, { "epoch": 0.8948215855622105, "grad_norm": 2.7705576419830322, "learning_rate": 5.749338308182851e-07, "loss": 0.059918975830078124, "step": 33970 }, { "epoch": 0.8950850008067092, "grad_norm": 4.581395149230957, "learning_rate": 5.720861831648617e-07, "loss": 0.06958953738212585, "step": 33980 }, { "epoch": 0.895348416051208, "grad_norm": 2.683781623840332, "learning_rate": 5.692453975538703e-07, "loss": 0.06656657457351685, "step": 33990 }, { "epoch": 0.8956118312957067, "grad_norm": 2.2806172370910645, "learning_rate": 5.664114760529571e-07, "loss": 0.060177838802337645, "step": 34000 }, { "epoch": 0.8958752465402054, "grad_norm": 2.9025614261627197, "learning_rate": 5.635844207247687e-07, "loss": 0.0675632119178772, "step": 34010 }, { "epoch": 0.8961386617847041, "grad_norm": 3.399123191833496, "learning_rate": 5.607642336269615e-07, "loss": 0.05553511381149292, "step": 34020 }, { "epoch": 0.8964020770292028, "grad_norm": 3.2239863872528076, "learning_rate": 5.579509168121888e-07, "loss": 0.06159287691116333, "step": 34030 }, { "epoch": 0.8966654922737016, "grad_norm": 3.672731876373291, "learning_rate": 5.551444723281029e-07, "loss": 0.06465942859649658, "step": 34040 }, { "epoch": 0.8969289075182003, "grad_norm": 4.237736701965332, "learning_rate": 5.523449022173566e-07, "loss": 0.08307678699493408, "step": 34050 }, { "epoch": 0.8971923227626991, "grad_norm": 2.5882039070129395, "learning_rate": 5.495522085175942e-07, "loss": 0.06726406812667847, "step": 34060 }, { "epoch": 0.8974557380071978, "grad_norm": 3.1376700401306152, "learning_rate": 5.467663932614642e-07, "loss": 0.07016782164573669, "step": 34070 }, { "epoch": 0.8977191532516966, "grad_norm": 2.753380537033081, "learning_rate": 5.439874584766014e-07, "loss": 0.07298843264579773, "step": 34080 }, { "epoch": 0.8979825684961953, "grad_norm": 3.404411554336548, "learning_rate": 5.412154061856301e-07, "loss": 0.05262005925178528, "step": 34090 }, { "epoch": 0.8982459837406941, "grad_norm": 2.5851171016693115, "learning_rate": 5.38450238406174e-07, "loss": 0.0800599992275238, "step": 34100 }, { "epoch": 0.8985093989851928, "grad_norm": 2.789374351501465, "learning_rate": 5.356919571508401e-07, "loss": 0.07844670414924622, "step": 34110 }, { "epoch": 0.8987728142296915, "grad_norm": 3.9418654441833496, "learning_rate": 5.329405644272268e-07, "loss": 0.07148008942604064, "step": 34120 }, { "epoch": 0.8990362294741903, "grad_norm": 3.488544225692749, "learning_rate": 5.301960622379132e-07, "loss": 0.06487627029418945, "step": 34130 }, { "epoch": 0.899299644718689, "grad_norm": 2.7656826972961426, "learning_rate": 5.27458452580466e-07, "loss": 0.06651309728622437, "step": 34140 }, { "epoch": 0.8995630599631877, "grad_norm": 2.7786519527435303, "learning_rate": 5.24727737447438e-07, "loss": 0.059243416786193846, "step": 34150 }, { "epoch": 0.8998264752076864, "grad_norm": 4.502134799957275, "learning_rate": 5.220039188263615e-07, "loss": 0.07351697087287903, "step": 34160 }, { "epoch": 0.9000898904521852, "grad_norm": 3.9922306537628174, "learning_rate": 5.192869986997484e-07, "loss": 0.07171061635017395, "step": 34170 }, { "epoch": 0.9003533056966839, "grad_norm": 2.74399995803833, "learning_rate": 5.165769790450903e-07, "loss": 0.06711770296096801, "step": 34180 }, { "epoch": 0.9006167209411826, "grad_norm": 3.415722131729126, "learning_rate": 5.138738618348548e-07, "loss": 0.07246404290199279, "step": 34190 }, { "epoch": 0.9008801361856814, "grad_norm": 3.0980727672576904, "learning_rate": 5.111776490364906e-07, "loss": 0.0771128237247467, "step": 34200 }, { "epoch": 0.9011435514301801, "grad_norm": 2.900808095932007, "learning_rate": 5.084883426124165e-07, "loss": 0.05247111320495605, "step": 34210 }, { "epoch": 0.9014069666746789, "grad_norm": 3.418433904647827, "learning_rate": 5.058059445200237e-07, "loss": 0.07066994309425353, "step": 34220 }, { "epoch": 0.9016703819191776, "grad_norm": 2.3561339378356934, "learning_rate": 5.031304567116802e-07, "loss": 0.07301044464111328, "step": 34230 }, { "epoch": 0.9019337971636764, "grad_norm": 4.399892807006836, "learning_rate": 5.004618811347195e-07, "loss": 0.07047061920166016, "step": 34240 }, { "epoch": 0.9021972124081751, "grad_norm": 3.55444598197937, "learning_rate": 4.978002197314469e-07, "loss": 0.06913584470748901, "step": 34250 }, { "epoch": 0.9024606276526739, "grad_norm": 2.589284896850586, "learning_rate": 4.951454744391338e-07, "loss": 0.07341957688331605, "step": 34260 }, { "epoch": 0.9027240428971726, "grad_norm": 2.4250681400299072, "learning_rate": 4.924976471900178e-07, "loss": 0.05832265615463257, "step": 34270 }, { "epoch": 0.9029874581416713, "grad_norm": 4.96615743637085, "learning_rate": 4.89856739911303e-07, "loss": 0.06664260029792786, "step": 34280 }, { "epoch": 0.90325087338617, "grad_norm": 3.6920433044433594, "learning_rate": 4.872227545251562e-07, "loss": 0.0685396671295166, "step": 34290 }, { "epoch": 0.9035142886306687, "grad_norm": 3.308919668197632, "learning_rate": 4.84595692948705e-07, "loss": 0.062111294269561766, "step": 34300 }, { "epoch": 0.9037777038751675, "grad_norm": 3.697450876235962, "learning_rate": 4.819755570940377e-07, "loss": 0.07377087473869323, "step": 34310 }, { "epoch": 0.9040411191196662, "grad_norm": 3.294534206390381, "learning_rate": 4.793623488682008e-07, "loss": 0.06684002876281739, "step": 34320 }, { "epoch": 0.904304534364165, "grad_norm": 6.550315856933594, "learning_rate": 4.7675607017320323e-07, "loss": 0.06573799252510071, "step": 34330 }, { "epoch": 0.9045679496086637, "grad_norm": 6.270641803741455, "learning_rate": 4.741567229060062e-07, "loss": 0.060546493530273436, "step": 34340 }, { "epoch": 0.9048313648531625, "grad_norm": 3.2333054542541504, "learning_rate": 4.715643089585242e-07, "loss": 0.06963454484939575, "step": 34350 }, { "epoch": 0.9050947800976612, "grad_norm": 2.8914146423339844, "learning_rate": 4.6897883021763013e-07, "loss": 0.05667920112609863, "step": 34360 }, { "epoch": 0.9053581953421599, "grad_norm": 3.321272134780884, "learning_rate": 4.664002885651475e-07, "loss": 0.06662573814392089, "step": 34370 }, { "epoch": 0.9056216105866587, "grad_norm": 3.4973583221435547, "learning_rate": 4.638286858778496e-07, "loss": 0.06962118148803711, "step": 34380 }, { "epoch": 0.9058850258311574, "grad_norm": 2.708573341369629, "learning_rate": 4.612640240274591e-07, "loss": 0.06993463039398193, "step": 34390 }, { "epoch": 0.9061484410756562, "grad_norm": 3.4472262859344482, "learning_rate": 4.587063048806462e-07, "loss": 0.0660646378993988, "step": 34400 }, { "epoch": 0.9064118563201549, "grad_norm": 2.4332125186920166, "learning_rate": 4.561555302990328e-07, "loss": 0.07137627005577088, "step": 34410 }, { "epoch": 0.9066752715646537, "grad_norm": 2.4246745109558105, "learning_rate": 4.536117021391806e-07, "loss": 0.07046207785606384, "step": 34420 }, { "epoch": 0.9069386868091524, "grad_norm": 3.6197385787963867, "learning_rate": 4.5107482225259735e-07, "loss": 0.06669858694076539, "step": 34430 }, { "epoch": 0.9072021020536511, "grad_norm": 3.4592044353485107, "learning_rate": 4.4854489248573494e-07, "loss": 0.07422381639480591, "step": 34440 }, { "epoch": 0.9074655172981498, "grad_norm": 3.6516060829162598, "learning_rate": 4.460219146799838e-07, "loss": 0.06241422295570374, "step": 34450 }, { "epoch": 0.9077289325426485, "grad_norm": 3.4071602821350098, "learning_rate": 4.4350589067167717e-07, "loss": 0.06552016735076904, "step": 34460 }, { "epoch": 0.9079923477871473, "grad_norm": 3.9214041233062744, "learning_rate": 4.409968222920868e-07, "loss": 0.085662579536438, "step": 34470 }, { "epoch": 0.908255763031646, "grad_norm": 3.3029510974884033, "learning_rate": 4.384947113674165e-07, "loss": 0.06616451740264892, "step": 34480 }, { "epoch": 0.9085191782761448, "grad_norm": 2.169626474380493, "learning_rate": 4.35999559718816e-07, "loss": 0.06089569330215454, "step": 34490 }, { "epoch": 0.9087825935206435, "grad_norm": 3.0607211589813232, "learning_rate": 4.3351136916236156e-07, "loss": 0.06911986470222473, "step": 34500 }, { "epoch": 0.9090460087651423, "grad_norm": 4.981792449951172, "learning_rate": 4.3103014150906675e-07, "loss": 0.05948582887649536, "step": 34510 }, { "epoch": 0.909309424009641, "grad_norm": 3.6763217449188232, "learning_rate": 4.285558785648747e-07, "loss": 0.07182815074920654, "step": 34520 }, { "epoch": 0.9095728392541398, "grad_norm": 2.194852590560913, "learning_rate": 4.260885821306604e-07, "loss": 0.051758098602294925, "step": 34530 }, { "epoch": 0.9098362544986385, "grad_norm": 2.836747646331787, "learning_rate": 4.236282540022307e-07, "loss": 0.06697534322738648, "step": 34540 }, { "epoch": 0.9100996697431372, "grad_norm": 3.211219310760498, "learning_rate": 4.2117489597031767e-07, "loss": 0.07155795693397522, "step": 34550 }, { "epoch": 0.910363084987636, "grad_norm": 3.844904899597168, "learning_rate": 4.1872850982058067e-07, "loss": 0.06796789169311523, "step": 34560 }, { "epoch": 0.9106265002321347, "grad_norm": 3.302058219909668, "learning_rate": 4.1628909733360534e-07, "loss": 0.06816842555999755, "step": 34570 }, { "epoch": 0.9108899154766334, "grad_norm": 2.330777406692505, "learning_rate": 4.1385666028490036e-07, "loss": 0.07303138375282288, "step": 34580 }, { "epoch": 0.9111533307211321, "grad_norm": 3.7076096534729004, "learning_rate": 4.114312004449006e-07, "loss": 0.05936057567596435, "step": 34590 }, { "epoch": 0.9114167459656309, "grad_norm": 3.4135234355926514, "learning_rate": 4.090127195789606e-07, "loss": 0.06765905022621155, "step": 34600 }, { "epoch": 0.9116801612101296, "grad_norm": 2.921660900115967, "learning_rate": 4.066012194473512e-07, "loss": 0.0720510482788086, "step": 34610 }, { "epoch": 0.9119435764546284, "grad_norm": 2.658021926879883, "learning_rate": 4.041967018052706e-07, "loss": 0.07212410569190979, "step": 34620 }, { "epoch": 0.9122069916991271, "grad_norm": 2.5927419662475586, "learning_rate": 4.0179916840282776e-07, "loss": 0.060499167442321776, "step": 34630 }, { "epoch": 0.9124704069436258, "grad_norm": 2.840946674346924, "learning_rate": 3.9940862098505117e-07, "loss": 0.06429675817489625, "step": 34640 }, { "epoch": 0.9127338221881246, "grad_norm": 4.114678382873535, "learning_rate": 3.970250612918858e-07, "loss": 0.07651277780532836, "step": 34650 }, { "epoch": 0.9129972374326233, "grad_norm": 3.27508544921875, "learning_rate": 3.946484910581871e-07, "loss": 0.06486624479293823, "step": 34660 }, { "epoch": 0.9132606526771221, "grad_norm": 3.9032092094421387, "learning_rate": 3.9227891201372827e-07, "loss": 0.06455878019332886, "step": 34670 }, { "epoch": 0.9135240679216208, "grad_norm": 3.216555595397949, "learning_rate": 3.899163258831895e-07, "loss": 0.07500683069229126, "step": 34680 }, { "epoch": 0.9137874831661196, "grad_norm": 2.1663460731506348, "learning_rate": 3.8756073438616203e-07, "loss": 0.06462491154670716, "step": 34690 }, { "epoch": 0.9140508984106183, "grad_norm": 2.458247184753418, "learning_rate": 3.8521213923714775e-07, "loss": 0.07104578614234924, "step": 34700 }, { "epoch": 0.914314313655117, "grad_norm": 2.7819652557373047, "learning_rate": 3.8287054214555694e-07, "loss": 0.05853608250617981, "step": 34710 }, { "epoch": 0.9145777288996157, "grad_norm": 2.965559959411621, "learning_rate": 3.805359448157031e-07, "loss": 0.06336586475372315, "step": 34720 }, { "epoch": 0.9148411441441144, "grad_norm": 3.4099690914154053, "learning_rate": 3.78208348946808e-07, "loss": 0.061061149835586546, "step": 34730 }, { "epoch": 0.9151045593886132, "grad_norm": 2.174481153488159, "learning_rate": 3.7588775623299433e-07, "loss": 0.06471932530403138, "step": 34740 }, { "epoch": 0.9153679746331119, "grad_norm": 2.8823471069335938, "learning_rate": 3.735741683632932e-07, "loss": 0.061433160305023195, "step": 34750 }, { "epoch": 0.9156313898776107, "grad_norm": 2.9284534454345703, "learning_rate": 3.712675870216309e-07, "loss": 0.06620916128158569, "step": 34760 }, { "epoch": 0.9158948051221094, "grad_norm": 3.550694704055786, "learning_rate": 3.689680138868379e-07, "loss": 0.062067800760269166, "step": 34770 }, { "epoch": 0.9161582203666082, "grad_norm": 2.852551221847534, "learning_rate": 3.66675450632642e-07, "loss": 0.06298303604125977, "step": 34780 }, { "epoch": 0.9164216356111069, "grad_norm": 3.396662712097168, "learning_rate": 3.643898989276684e-07, "loss": 0.0714790940284729, "step": 34790 }, { "epoch": 0.9166850508556056, "grad_norm": 4.9980926513671875, "learning_rate": 3.6211136043544425e-07, "loss": 0.06943727731704712, "step": 34800 }, { "epoch": 0.9169484661001044, "grad_norm": 2.571525812149048, "learning_rate": 3.598398368143863e-07, "loss": 0.0724039375782013, "step": 34810 }, { "epoch": 0.9172118813446031, "grad_norm": 2.651531219482422, "learning_rate": 3.575753297178053e-07, "loss": 0.06137374639511108, "step": 34820 }, { "epoch": 0.9174752965891019, "grad_norm": 2.841427803039551, "learning_rate": 3.553178407939095e-07, "loss": 0.06070883274078369, "step": 34830 }, { "epoch": 0.9177387118336006, "grad_norm": 2.0778913497924805, "learning_rate": 3.5306737168579685e-07, "loss": 0.06099547147750854, "step": 34840 }, { "epoch": 0.9180021270780994, "grad_norm": 3.5067005157470703, "learning_rate": 3.5082392403145373e-07, "loss": 0.06863335967063904, "step": 34850 }, { "epoch": 0.918265542322598, "grad_norm": 2.414606809616089, "learning_rate": 3.485874994637606e-07, "loss": 0.06724667549133301, "step": 34860 }, { "epoch": 0.9185289575670968, "grad_norm": 3.35115122795105, "learning_rate": 3.463580996104821e-07, "loss": 0.06820938587188721, "step": 34870 }, { "epoch": 0.9187923728115955, "grad_norm": 2.8392510414123535, "learning_rate": 3.4413572609427235e-07, "loss": 0.06619869470596314, "step": 34880 }, { "epoch": 0.9190557880560942, "grad_norm": 4.156548976898193, "learning_rate": 3.4192038053267095e-07, "loss": 0.0746225357055664, "step": 34890 }, { "epoch": 0.919319203300593, "grad_norm": 3.6641926765441895, "learning_rate": 3.3971206453810133e-07, "loss": 0.059934604167938235, "step": 34900 }, { "epoch": 0.9195826185450917, "grad_norm": 2.5493860244750977, "learning_rate": 3.3751077971787115e-07, "loss": 0.05617091655731201, "step": 34910 }, { "epoch": 0.9198460337895905, "grad_norm": 3.6379706859588623, "learning_rate": 3.353165276741699e-07, "loss": 0.06659682989120483, "step": 34920 }, { "epoch": 0.9201094490340892, "grad_norm": 3.103461503982544, "learning_rate": 3.331293100040711e-07, "loss": 0.06614254713058472, "step": 34930 }, { "epoch": 0.920372864278588, "grad_norm": 4.025054931640625, "learning_rate": 3.3094912829952477e-07, "loss": 0.0697891116142273, "step": 34940 }, { "epoch": 0.9206362795230867, "grad_norm": 3.6292612552642822, "learning_rate": 3.2877598414736033e-07, "loss": 0.06088302135467529, "step": 34950 }, { "epoch": 0.9208996947675855, "grad_norm": 3.2276840209960938, "learning_rate": 3.266098791292871e-07, "loss": 0.0739856481552124, "step": 34960 }, { "epoch": 0.9211631100120842, "grad_norm": 2.9250845909118652, "learning_rate": 3.244508148218895e-07, "loss": 0.05788544416427612, "step": 34970 }, { "epoch": 0.9214265252565829, "grad_norm": 2.6494016647338867, "learning_rate": 3.22298792796627e-07, "loss": 0.06768461465835571, "step": 34980 }, { "epoch": 0.9216899405010817, "grad_norm": 2.946368455886841, "learning_rate": 3.2015381461983577e-07, "loss": 0.07040110230445862, "step": 34990 }, { "epoch": 0.9219533557455803, "grad_norm": 2.946453094482422, "learning_rate": 3.1801588185272016e-07, "loss": 0.05062521696090698, "step": 35000 }, { "epoch": 0.9222167709900791, "grad_norm": 2.794009208679199, "learning_rate": 3.1588499605136435e-07, "loss": 0.06689966917037964, "step": 35010 }, { "epoch": 0.9224801862345778, "grad_norm": 3.4992125034332275, "learning_rate": 3.1376115876671553e-07, "loss": 0.07450852394104004, "step": 35020 }, { "epoch": 0.9227436014790766, "grad_norm": 1.9453727006912231, "learning_rate": 3.1164437154459494e-07, "loss": 0.05880904197692871, "step": 35030 }, { "epoch": 0.9230070167235753, "grad_norm": 5.2709808349609375, "learning_rate": 3.0953463592569343e-07, "loss": 0.06500686407089233, "step": 35040 }, { "epoch": 0.9232704319680741, "grad_norm": 2.998124599456787, "learning_rate": 3.074319534455639e-07, "loss": 0.07283071279525757, "step": 35050 }, { "epoch": 0.9235338472125728, "grad_norm": 2.862246036529541, "learning_rate": 3.053363256346331e-07, "loss": 0.06483680009841919, "step": 35060 }, { "epoch": 0.9237972624570715, "grad_norm": 5.585199356079102, "learning_rate": 3.0324775401818884e-07, "loss": 0.05879566669464111, "step": 35070 }, { "epoch": 0.9240606777015703, "grad_norm": 4.149601459503174, "learning_rate": 3.0116624011637953e-07, "loss": 0.07577759623527527, "step": 35080 }, { "epoch": 0.924324092946069, "grad_norm": 3.236816167831421, "learning_rate": 2.9909178544422656e-07, "loss": 0.06814301013946533, "step": 35090 }, { "epoch": 0.9245875081905678, "grad_norm": 2.7934131622314453, "learning_rate": 2.9702439151160334e-07, "loss": 0.0669049859046936, "step": 35100 }, { "epoch": 0.9248509234350665, "grad_norm": 4.362020492553711, "learning_rate": 2.949640598232517e-07, "loss": 0.05945996642112732, "step": 35110 }, { "epoch": 0.9251143386795653, "grad_norm": 4.1862030029296875, "learning_rate": 2.9291079187876656e-07, "loss": 0.0710377812385559, "step": 35120 }, { "epoch": 0.925377753924064, "grad_norm": 2.7156708240509033, "learning_rate": 2.9086458917260585e-07, "loss": 0.062160050868988036, "step": 35130 }, { "epoch": 0.9256411691685628, "grad_norm": 3.34242844581604, "learning_rate": 2.8882545319408706e-07, "loss": 0.05506402254104614, "step": 35140 }, { "epoch": 0.9259045844130614, "grad_norm": 3.718435764312744, "learning_rate": 2.8679338542737745e-07, "loss": 0.07129460573196411, "step": 35150 }, { "epoch": 0.9261679996575601, "grad_norm": 4.234333515167236, "learning_rate": 2.8476838735150613e-07, "loss": 0.06324107646942138, "step": 35160 }, { "epoch": 0.9264314149020589, "grad_norm": 3.8922274112701416, "learning_rate": 2.8275046044035304e-07, "loss": 0.06187015175819397, "step": 35170 }, { "epoch": 0.9266948301465576, "grad_norm": 3.620448112487793, "learning_rate": 2.8073960616265215e-07, "loss": 0.06987126469612122, "step": 35180 }, { "epoch": 0.9269582453910564, "grad_norm": 4.918920040130615, "learning_rate": 2.7873582598199056e-07, "loss": 0.06673462390899658, "step": 35190 }, { "epoch": 0.9272216606355551, "grad_norm": 2.8469221591949463, "learning_rate": 2.7673912135680714e-07, "loss": 0.05462604761123657, "step": 35200 }, { "epoch": 0.9274850758800539, "grad_norm": 3.4576547145843506, "learning_rate": 2.7474949374038717e-07, "loss": 0.06771667003631592, "step": 35210 }, { "epoch": 0.9277484911245526, "grad_norm": 1.6881390810012817, "learning_rate": 2.7276694458086894e-07, "loss": 0.07095270156860352, "step": 35220 }, { "epoch": 0.9280119063690514, "grad_norm": 3.0208003520965576, "learning_rate": 2.7079147532123704e-07, "loss": 0.06828190088272094, "step": 35230 }, { "epoch": 0.9282753216135501, "grad_norm": 3.037111520767212, "learning_rate": 2.688230873993225e-07, "loss": 0.046409237384796145, "step": 35240 }, { "epoch": 0.9285387368580488, "grad_norm": 2.2825560569763184, "learning_rate": 2.6686178224780367e-07, "loss": 0.060014545917510986, "step": 35250 }, { "epoch": 0.9288021521025476, "grad_norm": 3.0610392093658447, "learning_rate": 2.6490756129420313e-07, "loss": 0.06580082774162292, "step": 35260 }, { "epoch": 0.9290655673470463, "grad_norm": 3.3262665271759033, "learning_rate": 2.6296042596088643e-07, "loss": 0.06081411838531494, "step": 35270 }, { "epoch": 0.9293289825915451, "grad_norm": 3.828575611114502, "learning_rate": 2.610203776650655e-07, "loss": 0.06917884349822997, "step": 35280 }, { "epoch": 0.9295923978360437, "grad_norm": 3.6280529499053955, "learning_rate": 2.5908741781878744e-07, "loss": 0.0654872179031372, "step": 35290 }, { "epoch": 0.9298558130805425, "grad_norm": 3.249727725982666, "learning_rate": 2.5716154782894573e-07, "loss": 0.05768862962722778, "step": 35300 }, { "epoch": 0.9301192283250412, "grad_norm": 3.8758578300476074, "learning_rate": 2.552427690972703e-07, "loss": 0.05914404392242432, "step": 35310 }, { "epoch": 0.9303826435695399, "grad_norm": 3.8234362602233887, "learning_rate": 2.5333108302033394e-07, "loss": 0.06953497529029846, "step": 35320 }, { "epoch": 0.9306460588140387, "grad_norm": 4.039280414581299, "learning_rate": 2.514264909895414e-07, "loss": 0.0614018976688385, "step": 35330 }, { "epoch": 0.9309094740585374, "grad_norm": 4.1669793128967285, "learning_rate": 2.495289943911361e-07, "loss": 0.07517010569572449, "step": 35340 }, { "epoch": 0.9311728893030362, "grad_norm": 3.239549160003662, "learning_rate": 2.476385946062021e-07, "loss": 0.06539897918701172, "step": 35350 }, { "epoch": 0.9314363045475349, "grad_norm": 3.582869529724121, "learning_rate": 2.4575529301065104e-07, "loss": 0.06148406267166138, "step": 35360 }, { "epoch": 0.9316997197920337, "grad_norm": 3.748199939727783, "learning_rate": 2.43879090975232e-07, "loss": 0.07415874600410462, "step": 35370 }, { "epoch": 0.9319631350365324, "grad_norm": 5.360779285430908, "learning_rate": 2.4200998986552484e-07, "loss": 0.06405571699142457, "step": 35380 }, { "epoch": 0.9322265502810312, "grad_norm": 3.9248595237731934, "learning_rate": 2.401479910419413e-07, "loss": 0.06239662766456604, "step": 35390 }, { "epoch": 0.9324899655255299, "grad_norm": 3.154788017272949, "learning_rate": 2.3829309585972738e-07, "loss": 0.07319233417510987, "step": 35400 }, { "epoch": 0.9327533807700286, "grad_norm": 3.3457484245300293, "learning_rate": 2.364453056689553e-07, "loss": 0.06260472536087036, "step": 35410 }, { "epoch": 0.9330167960145274, "grad_norm": 4.129461288452148, "learning_rate": 2.3460462181452258e-07, "loss": 0.06408477425575257, "step": 35420 }, { "epoch": 0.933280211259026, "grad_norm": 2.882317066192627, "learning_rate": 2.3277104563616205e-07, "loss": 0.0696748971939087, "step": 35430 }, { "epoch": 0.9335436265035248, "grad_norm": 3.289205551147461, "learning_rate": 2.3094457846842943e-07, "loss": 0.06763666868209839, "step": 35440 }, { "epoch": 0.9338070417480235, "grad_norm": 4.425334453582764, "learning_rate": 2.291252216407047e-07, "loss": 0.05761308670043945, "step": 35450 }, { "epoch": 0.9340704569925223, "grad_norm": 3.006817579269409, "learning_rate": 2.273129764771953e-07, "loss": 0.06177629232406616, "step": 35460 }, { "epoch": 0.934333872237021, "grad_norm": 3.8035807609558105, "learning_rate": 2.2550784429693052e-07, "loss": 0.06681252717971801, "step": 35470 }, { "epoch": 0.9345972874815198, "grad_norm": 2.556056499481201, "learning_rate": 2.2370982641376494e-07, "loss": 0.04824299812316894, "step": 35480 }, { "epoch": 0.9348607027260185, "grad_norm": 4.988071918487549, "learning_rate": 2.2191892413637173e-07, "loss": 0.06883771419525146, "step": 35490 }, { "epoch": 0.9351241179705172, "grad_norm": 2.707794189453125, "learning_rate": 2.2013513876824933e-07, "loss": 0.061020249128341676, "step": 35500 }, { "epoch": 0.935387533215016, "grad_norm": 2.4449210166931152, "learning_rate": 2.183584716077114e-07, "loss": 0.0618930459022522, "step": 35510 }, { "epoch": 0.9356509484595147, "grad_norm": 2.4225828647613525, "learning_rate": 2.1658892394789244e-07, "loss": 0.06079896688461304, "step": 35520 }, { "epoch": 0.9359143637040135, "grad_norm": 3.604703187942505, "learning_rate": 2.1482649707674775e-07, "loss": 0.056290817260742185, "step": 35530 }, { "epoch": 0.9361777789485122, "grad_norm": 3.5604753494262695, "learning_rate": 2.13071192277049e-07, "loss": 0.05984348654747009, "step": 35540 }, { "epoch": 0.936441194193011, "grad_norm": 2.4270474910736084, "learning_rate": 2.1132301082637752e-07, "loss": 0.06051046848297119, "step": 35550 }, { "epoch": 0.9367046094375097, "grad_norm": 3.2172441482543945, "learning_rate": 2.0958195399713888e-07, "loss": 0.06525610685348511, "step": 35560 }, { "epoch": 0.9369680246820085, "grad_norm": 2.9475784301757812, "learning_rate": 2.0784802305654938e-07, "loss": 0.061692464351654056, "step": 35570 }, { "epoch": 0.9372314399265071, "grad_norm": 3.7206618785858154, "learning_rate": 2.061212192666373e-07, "loss": 0.06715022325515747, "step": 35580 }, { "epoch": 0.9374948551710058, "grad_norm": 3.2202534675598145, "learning_rate": 2.0440154388424616e-07, "loss": 0.06224389672279358, "step": 35590 }, { "epoch": 0.9377582704155046, "grad_norm": 2.9608423709869385, "learning_rate": 2.0268899816102916e-07, "loss": 0.05771682262420654, "step": 35600 }, { "epoch": 0.9380216856600033, "grad_norm": 3.9707295894622803, "learning_rate": 2.0098358334345147e-07, "loss": 0.06465097665786743, "step": 35610 }, { "epoch": 0.9382851009045021, "grad_norm": 2.755300760269165, "learning_rate": 1.9928530067278684e-07, "loss": 0.056502151489257815, "step": 35620 }, { "epoch": 0.9385485161490008, "grad_norm": 3.4549598693847656, "learning_rate": 1.9759415138511984e-07, "loss": 0.06469356417655944, "step": 35630 }, { "epoch": 0.9388119313934996, "grad_norm": 4.016645908355713, "learning_rate": 1.959101367113414e-07, "loss": 0.06417992115020751, "step": 35640 }, { "epoch": 0.9390753466379983, "grad_norm": 2.490912914276123, "learning_rate": 1.9423325787714774e-07, "loss": 0.06370960474014283, "step": 35650 }, { "epoch": 0.9393387618824971, "grad_norm": 2.477259397506714, "learning_rate": 1.9256351610304814e-07, "loss": 0.06411116123199463, "step": 35660 }, { "epoch": 0.9396021771269958, "grad_norm": 3.0000438690185547, "learning_rate": 1.909009126043504e-07, "loss": 0.05828585028648377, "step": 35670 }, { "epoch": 0.9398655923714945, "grad_norm": 2.038930892944336, "learning_rate": 1.8924544859116656e-07, "loss": 0.06776336431503296, "step": 35680 }, { "epoch": 0.9401290076159933, "grad_norm": 3.0112216472625732, "learning_rate": 1.8759712526841943e-07, "loss": 0.06551603078842164, "step": 35690 }, { "epoch": 0.940392422860492, "grad_norm": 3.8171463012695312, "learning_rate": 1.8595594383582605e-07, "loss": 0.07150283455848694, "step": 35700 }, { "epoch": 0.9406558381049908, "grad_norm": 3.3044631481170654, "learning_rate": 1.8432190548791197e-07, "loss": 0.0702254056930542, "step": 35710 }, { "epoch": 0.9409192533494894, "grad_norm": 2.331223249435425, "learning_rate": 1.8269501141399915e-07, "loss": 0.06010733842849732, "step": 35720 }, { "epoch": 0.9411826685939882, "grad_norm": 3.1220593452453613, "learning_rate": 1.8107526279821041e-07, "loss": 0.07000739574432373, "step": 35730 }, { "epoch": 0.9414460838384869, "grad_norm": 3.4708974361419678, "learning_rate": 1.794626608194705e-07, "loss": 0.06597225069999695, "step": 35740 }, { "epoch": 0.9417094990829857, "grad_norm": 2.359861135482788, "learning_rate": 1.7785720665150054e-07, "loss": 0.061911338567733766, "step": 35750 }, { "epoch": 0.9419729143274844, "grad_norm": 4.016693115234375, "learning_rate": 1.7625890146281798e-07, "loss": 0.06420623660087585, "step": 35760 }, { "epoch": 0.9422363295719831, "grad_norm": 2.3122940063476562, "learning_rate": 1.7466774641674012e-07, "loss": 0.039618858695030214, "step": 35770 }, { "epoch": 0.9424997448164819, "grad_norm": 2.96588134765625, "learning_rate": 1.7308374267137606e-07, "loss": 0.06671398878097534, "step": 35780 }, { "epoch": 0.9427631600609806, "grad_norm": 3.4092154502868652, "learning_rate": 1.715068913796336e-07, "loss": 0.0662351667881012, "step": 35790 }, { "epoch": 0.9430265753054794, "grad_norm": 3.343649387359619, "learning_rate": 1.6993719368921468e-07, "loss": 0.065952467918396, "step": 35800 }, { "epoch": 0.9432899905499781, "grad_norm": 3.7595322132110596, "learning_rate": 1.683746507426087e-07, "loss": 0.05176767110824585, "step": 35810 }, { "epoch": 0.9435534057944769, "grad_norm": 2.8251609802246094, "learning_rate": 1.66819263677106e-07, "loss": 0.06183411478996277, "step": 35820 }, { "epoch": 0.9438168210389756, "grad_norm": 3.992302417755127, "learning_rate": 1.6527103362478202e-07, "loss": 0.06956695318222046, "step": 35830 }, { "epoch": 0.9440802362834743, "grad_norm": 2.4836747646331787, "learning_rate": 1.6372996171250544e-07, "loss": 0.06692380905151367, "step": 35840 }, { "epoch": 0.9443436515279731, "grad_norm": 3.065761089324951, "learning_rate": 1.6219604906193677e-07, "loss": 0.05390572547912598, "step": 35850 }, { "epoch": 0.9446070667724717, "grad_norm": 3.9247779846191406, "learning_rate": 1.606692967895207e-07, "loss": 0.06416180729866028, "step": 35860 }, { "epoch": 0.9448704820169705, "grad_norm": 3.3264148235321045, "learning_rate": 1.5914970600649615e-07, "loss": 0.06878658533096313, "step": 35870 }, { "epoch": 0.9451338972614692, "grad_norm": 2.5051918029785156, "learning_rate": 1.5763727781888726e-07, "loss": 0.04452327191829682, "step": 35880 }, { "epoch": 0.945397312505968, "grad_norm": 2.7474563121795654, "learning_rate": 1.5613201332750127e-07, "loss": 0.05224467515945434, "step": 35890 }, { "epoch": 0.9456607277504667, "grad_norm": 3.84110951423645, "learning_rate": 1.5463391362793735e-07, "loss": 0.052927166223526, "step": 35900 }, { "epoch": 0.9459241429949655, "grad_norm": 2.773376941680908, "learning_rate": 1.5314297981057658e-07, "loss": 0.06437426805496216, "step": 35910 }, { "epoch": 0.9461875582394642, "grad_norm": 1.844032645225525, "learning_rate": 1.5165921296058427e-07, "loss": 0.05415863990783691, "step": 35920 }, { "epoch": 0.9464509734839629, "grad_norm": 2.0917375087738037, "learning_rate": 1.501826141579099e-07, "loss": 0.0542169451713562, "step": 35930 }, { "epoch": 0.9467143887284617, "grad_norm": 2.1053998470306396, "learning_rate": 1.48713184477286e-07, "loss": 0.05710204839706421, "step": 35940 }, { "epoch": 0.9469778039729604, "grad_norm": 6.389218330383301, "learning_rate": 1.4725092498822814e-07, "loss": 0.07080699205398559, "step": 35950 }, { "epoch": 0.9472412192174592, "grad_norm": 2.710676431655884, "learning_rate": 1.457958367550305e-07, "loss": 0.06482576131820679, "step": 35960 }, { "epoch": 0.9475046344619579, "grad_norm": 2.3109018802642822, "learning_rate": 1.443479208367693e-07, "loss": 0.05304688811302185, "step": 35970 }, { "epoch": 0.9477680497064567, "grad_norm": 2.9469337463378906, "learning_rate": 1.4290717828730037e-07, "loss": 0.056238210201263426, "step": 35980 }, { "epoch": 0.9480314649509554, "grad_norm": 3.0757200717926025, "learning_rate": 1.4147361015525717e-07, "loss": 0.06683888435363769, "step": 35990 }, { "epoch": 0.9482948801954542, "grad_norm": 2.6698098182678223, "learning_rate": 1.4004721748405392e-07, "loss": 0.0759120225906372, "step": 36000 }, { "epoch": 0.9485582954399528, "grad_norm": 3.099714517593384, "learning_rate": 1.3862800131188237e-07, "loss": 0.06745811700820922, "step": 36010 }, { "epoch": 0.9488217106844515, "grad_norm": 3.8056981563568115, "learning_rate": 1.37215962671704e-07, "loss": 0.06095402240753174, "step": 36020 }, { "epoch": 0.9490851259289503, "grad_norm": 4.0516557693481445, "learning_rate": 1.3581110259126674e-07, "loss": 0.06737704277038574, "step": 36030 }, { "epoch": 0.949348541173449, "grad_norm": 2.370896816253662, "learning_rate": 1.3441342209308595e-07, "loss": 0.062187522649765015, "step": 36040 }, { "epoch": 0.9496119564179478, "grad_norm": 6.022850036621094, "learning_rate": 1.3302292219445457e-07, "loss": 0.06757278442382812, "step": 36050 }, { "epoch": 0.9498753716624465, "grad_norm": 3.0562589168548584, "learning_rate": 1.3163960390743746e-07, "loss": 0.05824088454246521, "step": 36060 }, { "epoch": 0.9501387869069453, "grad_norm": 2.840195894241333, "learning_rate": 1.302634682388726e-07, "loss": 0.06105105876922608, "step": 36070 }, { "epoch": 0.950402202151444, "grad_norm": 1.9746220111846924, "learning_rate": 1.2889451619037318e-07, "loss": 0.05463370680809021, "step": 36080 }, { "epoch": 0.9506656173959428, "grad_norm": 4.332595348358154, "learning_rate": 1.2753274875832112e-07, "loss": 0.06307222247123719, "step": 36090 }, { "epoch": 0.9509290326404415, "grad_norm": 1.9003493785858154, "learning_rate": 1.2617816693386907e-07, "loss": 0.05740306973457336, "step": 36100 }, { "epoch": 0.9511924478849402, "grad_norm": 2.701573133468628, "learning_rate": 1.2483077170293954e-07, "loss": 0.07299237251281739, "step": 36110 }, { "epoch": 0.951455863129439, "grad_norm": 2.5535805225372314, "learning_rate": 1.2349056404622472e-07, "loss": 0.05116230249404907, "step": 36120 }, { "epoch": 0.9517192783739377, "grad_norm": 2.774998426437378, "learning_rate": 1.2215754493918763e-07, "loss": 0.06638706326484681, "step": 36130 }, { "epoch": 0.9519826936184365, "grad_norm": 2.6995925903320312, "learning_rate": 1.208317153520555e-07, "loss": 0.06534093618392944, "step": 36140 }, { "epoch": 0.9522461088629351, "grad_norm": 3.209724187850952, "learning_rate": 1.195130762498231e-07, "loss": 0.06456388235092163, "step": 36150 }, { "epoch": 0.952509524107434, "grad_norm": 3.4315178394317627, "learning_rate": 1.1820162859225493e-07, "loss": 0.05848100185394287, "step": 36160 }, { "epoch": 0.9527729393519326, "grad_norm": 3.709329605102539, "learning_rate": 1.1689737333387852e-07, "loss": 0.05848492383956909, "step": 36170 }, { "epoch": 0.9530363545964314, "grad_norm": 4.891172409057617, "learning_rate": 1.1560031142398787e-07, "loss": 0.07288170456886292, "step": 36180 }, { "epoch": 0.9532997698409301, "grad_norm": 3.7905526161193848, "learning_rate": 1.1431044380663891e-07, "loss": 0.07299690246582032, "step": 36190 }, { "epoch": 0.9535631850854288, "grad_norm": 3.2312161922454834, "learning_rate": 1.1302777142065402e-07, "loss": 0.06000280976295471, "step": 36200 }, { "epoch": 0.9538266003299276, "grad_norm": 2.653223752975464, "learning_rate": 1.117522951996175e-07, "loss": 0.07204906940460205, "step": 36210 }, { "epoch": 0.9540900155744263, "grad_norm": 4.423652648925781, "learning_rate": 1.1048401607187565e-07, "loss": 0.063446044921875, "step": 36220 }, { "epoch": 0.9543534308189251, "grad_norm": 3.1638143062591553, "learning_rate": 1.0922293496053671e-07, "loss": 0.07644930481910706, "step": 36230 }, { "epoch": 0.9546168460634238, "grad_norm": 4.558931827545166, "learning_rate": 1.0796905278347092e-07, "loss": 0.06429603695869446, "step": 36240 }, { "epoch": 0.9548802613079226, "grad_norm": 3.4021217823028564, "learning_rate": 1.0672237045330491e-07, "loss": 0.06209831237792969, "step": 36250 }, { "epoch": 0.9551436765524213, "grad_norm": 3.7995922565460205, "learning_rate": 1.0548288887743174e-07, "loss": 0.06248587965965271, "step": 36260 }, { "epoch": 0.9554070917969201, "grad_norm": 4.490288734436035, "learning_rate": 1.0425060895799865e-07, "loss": 0.05750571489334107, "step": 36270 }, { "epoch": 0.9556705070414188, "grad_norm": 4.544161319732666, "learning_rate": 1.0302553159190932e-07, "loss": 0.07212551832199096, "step": 36280 }, { "epoch": 0.9559339222859174, "grad_norm": 2.6906330585479736, "learning_rate": 1.0180765767083045e-07, "loss": 0.06251028776168824, "step": 36290 }, { "epoch": 0.9561973375304162, "grad_norm": 3.065214157104492, "learning_rate": 1.00596988081183e-07, "loss": 0.058473408222198486, "step": 36300 }, { "epoch": 0.9564607527749149, "grad_norm": 3.1617796421051025, "learning_rate": 9.939352370414435e-08, "loss": 0.05471808910369873, "step": 36310 }, { "epoch": 0.9567241680194137, "grad_norm": 3.205885410308838, "learning_rate": 9.819726541564711e-08, "loss": 0.06076980829238891, "step": 36320 }, { "epoch": 0.9569875832639124, "grad_norm": 4.369968891143799, "learning_rate": 9.700821408638039e-08, "loss": 0.06605891585350036, "step": 36330 }, { "epoch": 0.9572509985084112, "grad_norm": 2.738532304763794, "learning_rate": 9.582637058178635e-08, "loss": 0.06150341033935547, "step": 36340 }, { "epoch": 0.9575144137529099, "grad_norm": 2.8357796669006348, "learning_rate": 9.465173576206355e-08, "loss": 0.06208198070526123, "step": 36350 }, { "epoch": 0.9577778289974086, "grad_norm": 3.110534191131592, "learning_rate": 9.348431048215923e-08, "loss": 0.07007046937942504, "step": 36360 }, { "epoch": 0.9580412442419074, "grad_norm": 3.7319934368133545, "learning_rate": 9.232409559177701e-08, "loss": 0.05710186362266541, "step": 36370 }, { "epoch": 0.9583046594864061, "grad_norm": 2.8039822578430176, "learning_rate": 9.117109193537254e-08, "loss": 0.07532470822334289, "step": 36380 }, { "epoch": 0.9585680747309049, "grad_norm": 2.576531410217285, "learning_rate": 9.002530035215007e-08, "loss": 0.06182205677032471, "step": 36390 }, { "epoch": 0.9588314899754036, "grad_norm": 3.7339987754821777, "learning_rate": 8.888672167606693e-08, "loss": 0.0667090117931366, "step": 36400 }, { "epoch": 0.9590949052199024, "grad_norm": 2.8098626136779785, "learning_rate": 8.775535673582913e-08, "loss": 0.05308043360710144, "step": 36410 }, { "epoch": 0.9593583204644011, "grad_norm": 2.661515235900879, "learning_rate": 8.663120635489241e-08, "loss": 0.062312227487564084, "step": 36420 }, { "epoch": 0.9596217357088999, "grad_norm": 3.9191224575042725, "learning_rate": 8.55142713514645e-08, "loss": 0.06717383861541748, "step": 36430 }, { "epoch": 0.9598851509533985, "grad_norm": 3.0809268951416016, "learning_rate": 8.440455253849622e-08, "loss": 0.06245568990707397, "step": 36440 }, { "epoch": 0.9601485661978972, "grad_norm": 3.3924214839935303, "learning_rate": 8.330205072369035e-08, "loss": 0.061897313594818114, "step": 36450 }, { "epoch": 0.960411981442396, "grad_norm": 3.1932637691497803, "learning_rate": 8.22067667094939e-08, "loss": 0.06780104637145996, "step": 36460 }, { "epoch": 0.9606753966868947, "grad_norm": 2.559570789337158, "learning_rate": 8.11187012931025e-08, "loss": 0.06601017117500305, "step": 36470 }, { "epoch": 0.9609388119313935, "grad_norm": 3.7208118438720703, "learning_rate": 8.003785526645824e-08, "loss": 0.06418870687484741, "step": 36480 }, { "epoch": 0.9612022271758922, "grad_norm": 4.233975410461426, "learning_rate": 7.896422941624626e-08, "loss": 0.061167746782302856, "step": 36490 }, { "epoch": 0.961465642420391, "grad_norm": 3.8591222763061523, "learning_rate": 7.789782452389705e-08, "loss": 0.07432438135147094, "step": 36500 }, { "epoch": 0.9617290576648897, "grad_norm": 3.6547117233276367, "learning_rate": 7.683864136558749e-08, "loss": 0.05547650456428528, "step": 36510 }, { "epoch": 0.9619924729093885, "grad_norm": 4.109374523162842, "learning_rate": 7.578668071223649e-08, "loss": 0.061409497261047365, "step": 36520 }, { "epoch": 0.9622558881538872, "grad_norm": 3.4943735599517822, "learning_rate": 7.474194332950602e-08, "loss": 0.06395503878593445, "step": 36530 }, { "epoch": 0.9625193033983859, "grad_norm": 3.579702377319336, "learning_rate": 7.370442997780114e-08, "loss": 0.058521497249603274, "step": 36540 }, { "epoch": 0.9627827186428847, "grad_norm": 2.5064849853515625, "learning_rate": 7.267414141227002e-08, "loss": 0.05352417230606079, "step": 36550 }, { "epoch": 0.9630461338873834, "grad_norm": 2.3022725582122803, "learning_rate": 7.165107838279838e-08, "loss": 0.0684069812297821, "step": 36560 }, { "epoch": 0.9633095491318822, "grad_norm": 2.080159902572632, "learning_rate": 7.063524163401836e-08, "loss": 0.07052454948425294, "step": 36570 }, { "epoch": 0.9635729643763808, "grad_norm": 3.960526704788208, "learning_rate": 6.962663190529851e-08, "loss": 0.056590044498443605, "step": 36580 }, { "epoch": 0.9638363796208796, "grad_norm": 2.8208744525909424, "learning_rate": 6.862524993074825e-08, "loss": 0.05303820371627808, "step": 36590 }, { "epoch": 0.9640997948653783, "grad_norm": 3.95232892036438, "learning_rate": 6.763109643921683e-08, "loss": 0.05927501916885376, "step": 36600 }, { "epoch": 0.9643632101098771, "grad_norm": 3.3001320362091064, "learning_rate": 6.664417215429319e-08, "loss": 0.07824054956436158, "step": 36610 }, { "epoch": 0.9646266253543758, "grad_norm": 2.683145046234131, "learning_rate": 6.566447779430163e-08, "loss": 0.06134311556816101, "step": 36620 }, { "epoch": 0.9648900405988745, "grad_norm": 2.7040867805480957, "learning_rate": 6.469201407230729e-08, "loss": 0.060402727127075194, "step": 36630 }, { "epoch": 0.9651534558433733, "grad_norm": 3.596644878387451, "learning_rate": 6.372678169611068e-08, "loss": 0.05915899276733398, "step": 36640 }, { "epoch": 0.965416871087872, "grad_norm": 3.2616140842437744, "learning_rate": 6.276878136824982e-08, "loss": 0.056795501708984376, "step": 36650 }, { "epoch": 0.9656802863323708, "grad_norm": 2.9140939712524414, "learning_rate": 6.181801378599805e-08, "loss": 0.05959173440933228, "step": 36660 }, { "epoch": 0.9659437015768695, "grad_norm": 2.7612690925598145, "learning_rate": 6.087447964136517e-08, "loss": 0.06299986839294433, "step": 36670 }, { "epoch": 0.9662071168213683, "grad_norm": 3.6207146644592285, "learning_rate": 5.993817962109627e-08, "loss": 0.0732944130897522, "step": 36680 }, { "epoch": 0.966470532065867, "grad_norm": 2.7682149410247803, "learning_rate": 5.9009114406671787e-08, "loss": 0.06204707622528076, "step": 36690 }, { "epoch": 0.9667339473103658, "grad_norm": 4.292416572570801, "learning_rate": 5.8087284674304136e-08, "loss": 0.059627437591552736, "step": 36700 }, { "epoch": 0.9669973625548645, "grad_norm": 3.616943120956421, "learning_rate": 5.7172691094942165e-08, "loss": 0.05935654640197754, "step": 36710 }, { "epoch": 0.9672607777993631, "grad_norm": 3.398334264755249, "learning_rate": 5.626533433426451e-08, "loss": 0.07134142518043518, "step": 36720 }, { "epoch": 0.967524193043862, "grad_norm": 3.797056198120117, "learning_rate": 5.5365215052686216e-08, "loss": 0.06341452598571777, "step": 36730 }, { "epoch": 0.9677876082883606, "grad_norm": 2.0214180946350098, "learning_rate": 5.4472333905352115e-08, "loss": 0.05105733871459961, "step": 36740 }, { "epoch": 0.9680510235328594, "grad_norm": 2.8581695556640625, "learning_rate": 5.3586691542140135e-08, "loss": 0.05502415895462036, "step": 36750 }, { "epoch": 0.9683144387773581, "grad_norm": 2.9842963218688965, "learning_rate": 5.2708288607659084e-08, "loss": 0.05389357805252075, "step": 36760 }, { "epoch": 0.9685778540218569, "grad_norm": 3.465888500213623, "learning_rate": 5.183712574124755e-08, "loss": 0.05505635142326355, "step": 36770 }, { "epoch": 0.9688412692663556, "grad_norm": 4.46764612197876, "learning_rate": 5.0973203576976104e-08, "loss": 0.07001177668571472, "step": 36780 }, { "epoch": 0.9691046845108544, "grad_norm": 3.788167953491211, "learning_rate": 5.011652274364509e-08, "loss": 0.06213077902793884, "step": 36790 }, { "epoch": 0.9693680997553531, "grad_norm": 3.165309190750122, "learning_rate": 4.926708386478352e-08, "loss": 0.05217143297195435, "step": 36800 }, { "epoch": 0.9696315149998518, "grad_norm": 4.695135593414307, "learning_rate": 4.8424887558649045e-08, "loss": 0.060825669765472413, "step": 36810 }, { "epoch": 0.9698949302443506, "grad_norm": 2.4620320796966553, "learning_rate": 4.7589934438228017e-08, "loss": 0.06930739283561707, "step": 36820 }, { "epoch": 0.9701583454888493, "grad_norm": 2.7158539295196533, "learning_rate": 4.676222511123762e-08, "loss": 0.07136245965957641, "step": 36830 }, { "epoch": 0.9704217607333481, "grad_norm": 4.940596103668213, "learning_rate": 4.59417601801182e-08, "loss": 0.06947525143623352, "step": 36840 }, { "epoch": 0.9706851759778468, "grad_norm": 3.742870569229126, "learning_rate": 4.512854024203983e-08, "loss": 0.07622558474540711, "step": 36850 }, { "epoch": 0.9709485912223456, "grad_norm": 1.4418046474456787, "learning_rate": 4.432256588890016e-08, "loss": 0.060635578632354734, "step": 36860 }, { "epoch": 0.9712120064668442, "grad_norm": 3.852719306945801, "learning_rate": 4.352383770732216e-08, "loss": 0.06284514665603638, "step": 36870 }, { "epoch": 0.9714754217113429, "grad_norm": 4.052299976348877, "learning_rate": 4.273235627865302e-08, "loss": 0.060453379154205324, "step": 36880 }, { "epoch": 0.9717388369558417, "grad_norm": 2.8200626373291016, "learning_rate": 4.19481221789686e-08, "loss": 0.0647871732711792, "step": 36890 }, { "epoch": 0.9720022522003404, "grad_norm": 3.987096071243286, "learning_rate": 4.117113597906675e-08, "loss": 0.06181532144546509, "step": 36900 }, { "epoch": 0.9722656674448392, "grad_norm": 4.309356689453125, "learning_rate": 4.040139824447398e-08, "loss": 0.06137176752090454, "step": 36910 }, { "epoch": 0.9725290826893379, "grad_norm": 5.364724159240723, "learning_rate": 3.963890953543659e-08, "loss": 0.0625640869140625, "step": 36920 }, { "epoch": 0.9727924979338367, "grad_norm": 3.687328577041626, "learning_rate": 3.88836704069262e-08, "loss": 0.06041463613510132, "step": 36930 }, { "epoch": 0.9730559131783354, "grad_norm": 5.084613800048828, "learning_rate": 3.813568140864088e-08, "loss": 0.07049065232276916, "step": 36940 }, { "epoch": 0.9733193284228342, "grad_norm": 2.9805638790130615, "learning_rate": 3.739494308499736e-08, "loss": 0.0681851863861084, "step": 36950 }, { "epoch": 0.9735827436673329, "grad_norm": 2.8583309650421143, "learning_rate": 3.66614559751377e-08, "loss": 0.06560817956924439, "step": 36960 }, { "epoch": 0.9738461589118316, "grad_norm": 4.065157890319824, "learning_rate": 3.593522061292598e-08, "loss": 0.046027582883834836, "step": 36970 }, { "epoch": 0.9741095741563304, "grad_norm": 4.239371299743652, "learning_rate": 3.521623752694603e-08, "loss": 0.06662788391113281, "step": 36980 }, { "epoch": 0.9743729894008291, "grad_norm": 2.827139377593994, "learning_rate": 3.4504507240507027e-08, "loss": 0.06225603222846985, "step": 36990 }, { "epoch": 0.9746364046453279, "grad_norm": 2.53149676322937, "learning_rate": 3.380003027163459e-08, "loss": 0.048912644386291504, "step": 37000 }, { "epoch": 0.9748998198898265, "grad_norm": 3.5596821308135986, "learning_rate": 3.310280713307856e-08, "loss": 0.055740964412689206, "step": 37010 }, { "epoch": 0.9751632351343253, "grad_norm": 4.813565254211426, "learning_rate": 3.241283833230746e-08, "loss": 0.05871249437332153, "step": 37020 }, { "epoch": 0.975426650378824, "grad_norm": 3.883119821548462, "learning_rate": 3.1730124371511794e-08, "loss": 0.05805238485336304, "step": 37030 }, { "epoch": 0.9756900656233228, "grad_norm": 3.065087080001831, "learning_rate": 3.1054665747599635e-08, "loss": 0.05188207626342774, "step": 37040 }, { "epoch": 0.9759534808678215, "grad_norm": 3.2943618297576904, "learning_rate": 3.038646295219771e-08, "loss": 0.0792877197265625, "step": 37050 }, { "epoch": 0.9762168961123202, "grad_norm": 3.0581490993499756, "learning_rate": 2.972551647165367e-08, "loss": 0.07103570699691772, "step": 37060 }, { "epoch": 0.976480311356819, "grad_norm": 3.407069444656372, "learning_rate": 2.9071826787033797e-08, "loss": 0.0685268759727478, "step": 37070 }, { "epoch": 0.9767437266013177, "grad_norm": 2.9463186264038086, "learning_rate": 2.8425394374120842e-08, "loss": 0.0510295033454895, "step": 37080 }, { "epoch": 0.9770071418458165, "grad_norm": 3.4702494144439697, "learning_rate": 2.7786219703415108e-08, "loss": 0.05799933671951294, "step": 37090 }, { "epoch": 0.9772705570903152, "grad_norm": 2.942330837249756, "learning_rate": 2.715430324013779e-08, "loss": 0.0472880631685257, "step": 37100 }, { "epoch": 0.977533972334814, "grad_norm": 3.8896145820617676, "learning_rate": 2.6529645444223206e-08, "loss": 0.062497389316558835, "step": 37110 }, { "epoch": 0.9777973875793127, "grad_norm": 2.745476245880127, "learning_rate": 2.5912246770324333e-08, "loss": 0.05214165449142456, "step": 37120 }, { "epoch": 0.9780608028238115, "grad_norm": 4.1486310958862305, "learning_rate": 2.5302107667811716e-08, "loss": 0.06928523778915405, "step": 37130 }, { "epoch": 0.9783242180683102, "grad_norm": 4.101527214050293, "learning_rate": 2.469922858076901e-08, "loss": 0.05779673457145691, "step": 37140 }, { "epoch": 0.9785876333128088, "grad_norm": 5.907912731170654, "learning_rate": 2.410360994799965e-08, "loss": 0.0684014916419983, "step": 37150 }, { "epoch": 0.9788510485573076, "grad_norm": 3.4073240756988525, "learning_rate": 2.351525220301909e-08, "loss": 0.06044005155563355, "step": 37160 }, { "epoch": 0.9791144638018063, "grad_norm": 3.030017852783203, "learning_rate": 2.293415577406144e-08, "loss": 0.0704139530658722, "step": 37170 }, { "epoch": 0.9793778790463051, "grad_norm": 3.783949613571167, "learning_rate": 2.2360321084071712e-08, "loss": 0.05154076814651489, "step": 37180 }, { "epoch": 0.9796412942908038, "grad_norm": 2.309088945388794, "learning_rate": 2.179374855071137e-08, "loss": 0.05949966311454773, "step": 37190 }, { "epoch": 0.9799047095353026, "grad_norm": 2.5515754222869873, "learning_rate": 2.123443858635832e-08, "loss": 0.0600774884223938, "step": 37200 }, { "epoch": 0.9801681247798013, "grad_norm": 2.109328508377075, "learning_rate": 2.068239159810137e-08, "loss": 0.061071574687957764, "step": 37210 }, { "epoch": 0.9804315400243001, "grad_norm": 4.727712154388428, "learning_rate": 2.0137607987743558e-08, "loss": 0.05730457305908203, "step": 37220 }, { "epoch": 0.9806949552687988, "grad_norm": 4.376340389251709, "learning_rate": 1.9600088151803255e-08, "loss": 0.06149517297744751, "step": 37230 }, { "epoch": 0.9809583705132975, "grad_norm": 4.225915908813477, "learning_rate": 1.906983248150973e-08, "loss": 0.04975322186946869, "step": 37240 }, { "epoch": 0.9812217857577963, "grad_norm": 2.6586833000183105, "learning_rate": 1.854684136280538e-08, "loss": 0.059248697757720944, "step": 37250 }, { "epoch": 0.981485201002295, "grad_norm": 5.076155662536621, "learning_rate": 1.8031115176346815e-08, "loss": 0.0664952278137207, "step": 37260 }, { "epoch": 0.9817486162467938, "grad_norm": 3.132823944091797, "learning_rate": 1.752265429750044e-08, "loss": 0.06743626594543457, "step": 37270 }, { "epoch": 0.9820120314912925, "grad_norm": 4.7221856117248535, "learning_rate": 1.7021459096348004e-08, "loss": 0.06445568799972534, "step": 37280 }, { "epoch": 0.9822754467357913, "grad_norm": 3.079085111618042, "learning_rate": 1.652752993767992e-08, "loss": 0.058270198106765744, "step": 37290 }, { "epoch": 0.98253886198029, "grad_norm": 4.210781097412109, "learning_rate": 1.604086718099973e-08, "loss": 0.05973916053771973, "step": 37300 }, { "epoch": 0.9828022772247887, "grad_norm": 3.6348624229431152, "learning_rate": 1.556147118051965e-08, "loss": 0.061738085746765134, "step": 37310 }, { "epoch": 0.9830656924692874, "grad_norm": 3.10733962059021, "learning_rate": 1.508934228516723e-08, "loss": 0.060544437170028685, "step": 37320 }, { "epoch": 0.9833291077137861, "grad_norm": 3.6002562046051025, "learning_rate": 1.4624480838576483e-08, "loss": 0.06462849378585815, "step": 37330 }, { "epoch": 0.9835925229582849, "grad_norm": 2.0158212184906006, "learning_rate": 1.4166887179094535e-08, "loss": 0.05490938425064087, "step": 37340 }, { "epoch": 0.9838559382027836, "grad_norm": 3.6057186126708984, "learning_rate": 1.3716561639777192e-08, "loss": 0.06757857799530029, "step": 37350 }, { "epoch": 0.9841193534472824, "grad_norm": 3.1502859592437744, "learning_rate": 1.3273504548392269e-08, "loss": 0.05457271337509155, "step": 37360 }, { "epoch": 0.9843827686917811, "grad_norm": 3.421194076538086, "learning_rate": 1.2837716227414032e-08, "loss": 0.050599581003189086, "step": 37370 }, { "epoch": 0.9846461839362799, "grad_norm": 2.648801326751709, "learning_rate": 1.240919699402765e-08, "loss": 0.05905076265335083, "step": 37380 }, { "epoch": 0.9849095991807786, "grad_norm": 3.1793055534362793, "learning_rate": 1.1987947160130297e-08, "loss": 0.06113569736480713, "step": 37390 }, { "epoch": 0.9851730144252773, "grad_norm": 3.174959659576416, "learning_rate": 1.1573967032324495e-08, "loss": 0.0634928822517395, "step": 37400 }, { "epoch": 0.9854364296697761, "grad_norm": 2.7273948192596436, "learning_rate": 1.1167256911922552e-08, "loss": 0.05582513809204102, "step": 37410 }, { "epoch": 0.9856998449142748, "grad_norm": 3.75221848487854, "learning_rate": 1.0767817094945454e-08, "loss": 0.06376756429672241, "step": 37420 }, { "epoch": 0.9859632601587736, "grad_norm": 4.900347709655762, "learning_rate": 1.0375647872123974e-08, "loss": 0.06101369857788086, "step": 37430 }, { "epoch": 0.9862266754032722, "grad_norm": 4.794971466064453, "learning_rate": 9.990749528894227e-09, "loss": 0.06328186988830567, "step": 37440 }, { "epoch": 0.986490090647771, "grad_norm": 2.4704575538635254, "learning_rate": 9.613122345403236e-09, "loss": 0.06032406091690064, "step": 37450 }, { "epoch": 0.9867535058922697, "grad_norm": 3.2431576251983643, "learning_rate": 9.24276659650336e-09, "loss": 0.05777112245559692, "step": 37460 }, { "epoch": 0.9870169211367685, "grad_norm": 3.270052671432495, "learning_rate": 8.879682551755643e-09, "loss": 0.062217038869857785, "step": 37470 }, { "epoch": 0.9872803363812672, "grad_norm": 2.3695640563964844, "learning_rate": 8.523870475429797e-09, "loss": 0.05576260089874267, "step": 37480 }, { "epoch": 0.9875437516257659, "grad_norm": 2.224086046218872, "learning_rate": 8.175330626498668e-09, "loss": 0.05544992089271546, "step": 37490 }, { "epoch": 0.9878071668702647, "grad_norm": 4.216085433959961, "learning_rate": 7.834063258645997e-09, "loss": 0.07060191631317139, "step": 37500 }, { "epoch": 0.9880705821147634, "grad_norm": 3.5357675552368164, "learning_rate": 7.500068620260869e-09, "loss": 0.06102967858314514, "step": 37510 }, { "epoch": 0.9883339973592622, "grad_norm": 3.086509943008423, "learning_rate": 7.173346954438831e-09, "loss": 0.05892540812492371, "step": 37520 }, { "epoch": 0.9885974126037609, "grad_norm": 3.5890142917633057, "learning_rate": 6.853898498982992e-09, "loss": 0.06210651397705078, "step": 37530 }, { "epoch": 0.9888608278482597, "grad_norm": 3.6876044273376465, "learning_rate": 6.541723486399587e-09, "loss": 0.05739049911499024, "step": 37540 }, { "epoch": 0.9891242430927584, "grad_norm": 3.5036025047302246, "learning_rate": 6.236822143904642e-09, "loss": 0.058118689060211184, "step": 37550 }, { "epoch": 0.9893876583372572, "grad_norm": 2.9350717067718506, "learning_rate": 5.939194693418415e-09, "loss": 0.05445452928543091, "step": 37560 }, { "epoch": 0.9896510735817559, "grad_norm": 4.112997055053711, "learning_rate": 5.648841351566514e-09, "loss": 0.06418215632438659, "step": 37570 }, { "epoch": 0.9899144888262545, "grad_norm": 4.24894905090332, "learning_rate": 5.365762329681002e-09, "loss": 0.05628114938735962, "step": 37580 }, { "epoch": 0.9901779040707533, "grad_norm": 2.41082763671875, "learning_rate": 5.089957833798176e-09, "loss": 0.06576555371284484, "step": 37590 }, { "epoch": 0.990441319315252, "grad_norm": 2.5604183673858643, "learning_rate": 4.8214280646619035e-09, "loss": 0.060748052597045896, "step": 37600 }, { "epoch": 0.9907047345597508, "grad_norm": 3.58798885345459, "learning_rate": 4.560173217718067e-09, "loss": 0.05806952714920044, "step": 37610 }, { "epoch": 0.9909681498042495, "grad_norm": 4.627460479736328, "learning_rate": 4.3061934831212235e-09, "loss": 0.05500376224517822, "step": 37620 }, { "epoch": 0.9912315650487483, "grad_norm": 2.907646656036377, "learning_rate": 4.0594890457268384e-09, "loss": 0.05651489496231079, "step": 37630 }, { "epoch": 0.991494980293247, "grad_norm": 4.411056041717529, "learning_rate": 3.820060085097943e-09, "loss": 0.059450531005859376, "step": 37640 }, { "epoch": 0.9917583955377458, "grad_norm": 6.279633522033691, "learning_rate": 3.587906775501804e-09, "loss": 0.06400758028030396, "step": 37650 }, { "epoch": 0.9920218107822445, "grad_norm": 1.991619348526001, "learning_rate": 3.363029285907704e-09, "loss": 0.05556267499923706, "step": 37660 }, { "epoch": 0.9922852260267432, "grad_norm": 2.5423238277435303, "learning_rate": 3.145427779993604e-09, "loss": 0.06810675859451294, "step": 37670 }, { "epoch": 0.992548641271242, "grad_norm": 3.280219793319702, "learning_rate": 2.935102416138369e-09, "loss": 0.06914938688278198, "step": 37680 }, { "epoch": 0.9928120565157407, "grad_norm": 3.1502554416656494, "learning_rate": 2.732053347425101e-09, "loss": 0.05881795883178711, "step": 37690 }, { "epoch": 0.9930754717602395, "grad_norm": 2.6038618087768555, "learning_rate": 2.536280721642248e-09, "loss": 0.0670561134815216, "step": 37700 }, { "epoch": 0.9933388870047382, "grad_norm": 3.749451160430908, "learning_rate": 2.3477846812813843e-09, "loss": 0.05416772365570068, "step": 37710 }, { "epoch": 0.993602302249237, "grad_norm": 3.3911755084991455, "learning_rate": 2.16656536353832e-09, "loss": 0.057948839664459226, "step": 37720 }, { "epoch": 0.9938657174937356, "grad_norm": 3.7741217613220215, "learning_rate": 1.9926229003119912e-09, "loss": 0.058225595951080324, "step": 37730 }, { "epoch": 0.9941291327382344, "grad_norm": 4.5007758140563965, "learning_rate": 1.8259574182055707e-09, "loss": 0.06272299289703369, "step": 37740 }, { "epoch": 0.9943925479827331, "grad_norm": 3.127060651779175, "learning_rate": 1.6665690385253564e-09, "loss": 0.05427701473236084, "step": 37750 }, { "epoch": 0.9946559632272318, "grad_norm": 4.05221700668335, "learning_rate": 1.5144578772807728e-09, "loss": 0.06853688955307007, "step": 37760 }, { "epoch": 0.9949193784717306, "grad_norm": 1.91194748878479, "learning_rate": 1.36962404518437e-09, "loss": 0.06747026443481445, "step": 37770 }, { "epoch": 0.9951827937162293, "grad_norm": 3.3678252696990967, "learning_rate": 1.232067647654045e-09, "loss": 0.06186845302581787, "step": 37780 }, { "epoch": 0.9954462089607281, "grad_norm": 3.6414027214050293, "learning_rate": 1.1017887848085995e-09, "loss": 0.06260403394699096, "step": 37790 }, { "epoch": 0.9957096242052268, "grad_norm": 2.238715887069702, "learning_rate": 9.78787551469962e-10, "loss": 0.05996856689453125, "step": 37800 }, { "epoch": 0.9959730394497256, "grad_norm": 3.988309383392334, "learning_rate": 8.630640371642963e-10, "loss": 0.0719624102115631, "step": 37810 }, { "epoch": 0.9962364546942243, "grad_norm": 3.6875298023223877, "learning_rate": 7.546183261197826e-10, "loss": 0.06594178676605225, "step": 37820 }, { "epoch": 0.9964998699387231, "grad_norm": 4.322195529937744, "learning_rate": 6.534504972688372e-10, "loss": 0.06954535245895385, "step": 37830 }, { "epoch": 0.9967632851832218, "grad_norm": 2.842454433441162, "learning_rate": 5.595606242447815e-10, "loss": 0.07532417178153991, "step": 37840 }, { "epoch": 0.9970267004277205, "grad_norm": 3.1752262115478516, "learning_rate": 4.729487753851736e-10, "loss": 0.05914503335952759, "step": 37850 }, { "epoch": 0.9972901156722193, "grad_norm": 3.7538788318634033, "learning_rate": 3.9361501372958686e-10, "loss": 0.06440417766571045, "step": 37860 }, { "epoch": 0.997553530916718, "grad_norm": 2.951054096221924, "learning_rate": 3.2155939702072093e-10, "loss": 0.06024588942527771, "step": 37870 }, { "epoch": 0.9978169461612167, "grad_norm": 2.5781307220458984, "learning_rate": 2.5678197770329094e-10, "loss": 0.06591175794601441, "step": 37880 }, { "epoch": 0.9980803614057154, "grad_norm": 3.2005767822265625, "learning_rate": 1.9928280292624835e-10, "loss": 0.06420115232467652, "step": 37890 }, { "epoch": 0.9983437766502142, "grad_norm": 3.929208517074585, "learning_rate": 1.4906191453833985e-10, "loss": 0.06423643827438355, "step": 37900 }, { "epoch": 0.9986071918947129, "grad_norm": 2.826920986175537, "learning_rate": 1.061193490936585e-10, "loss": 0.055414891242980956, "step": 37910 }, { "epoch": 0.9988706071392117, "grad_norm": 3.047431707382202, "learning_rate": 7.045513784831315e-11, "loss": 0.0588281512260437, "step": 37920 }, { "epoch": 0.9991340223837104, "grad_norm": 2.4550559520721436, "learning_rate": 4.206930675820786e-11, "loss": 0.054300183057785036, "step": 37930 }, { "epoch": 0.9993974376282091, "grad_norm": 2.4658539295196533, "learning_rate": 2.0961876485703358e-11, "loss": 0.04999401569366455, "step": 37940 }, { "epoch": 0.9996608528727079, "grad_norm": 1.9776800870895386, "learning_rate": 7.132862392955631e-12, "loss": 0.06333435773849487, "step": 37950 }, { "epoch": 0.9999242681172066, "grad_norm": 4.018247127532959, "learning_rate": 5.822745441363964e-13, "loss": 0.059854382276535036, "step": 37960 }, { "epoch": 1.0, "step": 37963, "total_flos": 3.8840012014027276e+20, "train_loss": 0.13153531528544013, "train_runtime": 586157.7991, "train_samples_per_second": 4.145, "train_steps_per_second": 0.065 } ], "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": 3.8840012014027276e+20, "train_batch_size": 1, "trial_name": null, "trial_params": null }