{ "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": 1222.0948486328125, "learning_rate": 1.5803336259877089e-07, "loss": 2.3519577026367187, "step": 10 }, { "epoch": 0.0005268304889974745, "grad_norm": 56.56095504760742, "learning_rate": 3.336259877085163e-07, "loss": 0.6700016975402832, "step": 20 }, { "epoch": 0.0007902457334962118, "grad_norm": 32.644588470458984, "learning_rate": 5.092186128182617e-07, "loss": 0.29873156547546387, "step": 30 }, { "epoch": 0.001053660977994949, "grad_norm": 14.587905883789062, "learning_rate": 6.848112379280071e-07, "loss": 0.24937758445739747, "step": 40 }, { "epoch": 0.0013170762224936862, "grad_norm": 20.448841094970703, "learning_rate": 8.604038630377525e-07, "loss": 0.25665714740753176, "step": 50 }, { "epoch": 0.0015804914669924235, "grad_norm": 14.717493057250977, "learning_rate": 1.035996488147498e-06, "loss": 0.2369797945022583, "step": 60 }, { "epoch": 0.0018439067114911607, "grad_norm": 11.747084617614746, "learning_rate": 1.2115891132572432e-06, "loss": 0.22146809101104736, "step": 70 }, { "epoch": 0.002107321955989898, "grad_norm": 9.72093677520752, "learning_rate": 1.3871817383669885e-06, "loss": 0.23249568939208984, "step": 80 }, { "epoch": 0.002370737200488635, "grad_norm": 13.551506042480469, "learning_rate": 1.5627743634767342e-06, "loss": 0.21590170860290528, "step": 90 }, { "epoch": 0.0026341524449873723, "grad_norm": 12.352090835571289, "learning_rate": 1.7383669885864795e-06, "loss": 0.2140030860900879, "step": 100 }, { "epoch": 0.00289756768948611, "grad_norm": 10.572576522827148, "learning_rate": 1.913959613696225e-06, "loss": 0.216910457611084, "step": 110 }, { "epoch": 0.003160982933984847, "grad_norm": 17.833526611328125, "learning_rate": 2.08955223880597e-06, "loss": 0.23264720439910888, "step": 120 }, { "epoch": 0.003424398178483584, "grad_norm": 10.585830688476562, "learning_rate": 2.265144863915716e-06, "loss": 0.20175528526306152, "step": 130 }, { "epoch": 0.0036878134229823213, "grad_norm": 15.713763236999512, "learning_rate": 2.440737489025461e-06, "loss": 0.21480393409729004, "step": 140 }, { "epoch": 0.0039512286674810585, "grad_norm": 10.168322563171387, "learning_rate": 2.6163301141352064e-06, "loss": 0.2171527624130249, "step": 150 }, { "epoch": 0.004214643911979796, "grad_norm": 18.469451904296875, "learning_rate": 2.791922739244952e-06, "loss": 0.2334719181060791, "step": 160 }, { "epoch": 0.004478059156478534, "grad_norm": 12.276195526123047, "learning_rate": 2.9675153643546974e-06, "loss": 0.23141922950744628, "step": 170 }, { "epoch": 0.00474147440097727, "grad_norm": 10.59259033203125, "learning_rate": 3.1431079894644427e-06, "loss": 0.23035364151000975, "step": 180 }, { "epoch": 0.005004889645476008, "grad_norm": 7.403031826019287, "learning_rate": 3.318700614574188e-06, "loss": 0.21332101821899413, "step": 190 }, { "epoch": 0.005268304889974745, "grad_norm": 12.61300277709961, "learning_rate": 3.4942932396839337e-06, "loss": 0.22472186088562013, "step": 200 }, { "epoch": 0.005531720134473482, "grad_norm": 13.484363555908203, "learning_rate": 3.669885864793679e-06, "loss": 0.2135566234588623, "step": 210 }, { "epoch": 0.00579513537897222, "grad_norm": 19.459684371948242, "learning_rate": 3.845478489903424e-06, "loss": 0.20437860488891602, "step": 220 }, { "epoch": 0.0060585506234709565, "grad_norm": 11.809774398803711, "learning_rate": 4.0210711150131695e-06, "loss": 0.21282684803009033, "step": 230 }, { "epoch": 0.006321965867969694, "grad_norm": 13.085128784179688, "learning_rate": 4.196663740122916e-06, "loss": 0.22220582962036134, "step": 240 }, { "epoch": 0.006585381112468432, "grad_norm": 15.69548511505127, "learning_rate": 4.37225636523266e-06, "loss": 0.1772662878036499, "step": 250 }, { "epoch": 0.006848796356967168, "grad_norm": 14.16679859161377, "learning_rate": 4.547848990342406e-06, "loss": 0.2156728744506836, "step": 260 }, { "epoch": 0.007112211601465906, "grad_norm": 6.038997650146484, "learning_rate": 4.7234416154521515e-06, "loss": 0.1933818817138672, "step": 270 }, { "epoch": 0.007375626845964643, "grad_norm": 13.32958698272705, "learning_rate": 4.899034240561897e-06, "loss": 0.19964178800582885, "step": 280 }, { "epoch": 0.00763904209046338, "grad_norm": 8.862226486206055, "learning_rate": 5.074626865671642e-06, "loss": 0.19048032760620118, "step": 290 }, { "epoch": 0.007902457334962117, "grad_norm": 10.271728515625, "learning_rate": 5.250219490781387e-06, "loss": 0.19905507564544678, "step": 300 }, { "epoch": 0.008165872579460855, "grad_norm": 8.858388900756836, "learning_rate": 5.425812115891133e-06, "loss": 0.19294438362121583, "step": 310 }, { "epoch": 0.008429287823959592, "grad_norm": 11.858563423156738, "learning_rate": 5.601404741000879e-06, "loss": 0.2001544713973999, "step": 320 }, { "epoch": 0.008692703068458329, "grad_norm": 6.768753528594971, "learning_rate": 5.776997366110624e-06, "loss": 0.20624189376831054, "step": 330 }, { "epoch": 0.008956118312957067, "grad_norm": 13.317550659179688, "learning_rate": 5.9525899912203685e-06, "loss": 0.19141422510147094, "step": 340 }, { "epoch": 0.009219533557455804, "grad_norm": 9.766947746276855, "learning_rate": 6.128182616330115e-06, "loss": 0.21354334354400634, "step": 350 }, { "epoch": 0.00948294880195454, "grad_norm": 9.64574909210205, "learning_rate": 6.30377524143986e-06, "loss": 0.20117881298065185, "step": 360 }, { "epoch": 0.00974636404645328, "grad_norm": 7.200226306915283, "learning_rate": 6.479367866549605e-06, "loss": 0.2057814121246338, "step": 370 }, { "epoch": 0.010009779290952016, "grad_norm": 5.936819076538086, "learning_rate": 6.654960491659351e-06, "loss": 0.20612454414367676, "step": 380 }, { "epoch": 0.010273194535450753, "grad_norm": 12.837786674499512, "learning_rate": 6.830553116769097e-06, "loss": 0.19961905479431152, "step": 390 }, { "epoch": 0.01053660977994949, "grad_norm": 6.926418304443359, "learning_rate": 7.006145741878841e-06, "loss": 0.19582968950271606, "step": 400 }, { "epoch": 0.010800025024448228, "grad_norm": 4.692302227020264, "learning_rate": 7.181738366988587e-06, "loss": 0.1881798028945923, "step": 410 }, { "epoch": 0.011063440268946964, "grad_norm": 5.540403366088867, "learning_rate": 7.3573309920983325e-06, "loss": 0.19867100715637206, "step": 420 }, { "epoch": 0.011326855513445701, "grad_norm": 12.043661117553711, "learning_rate": 7.532923617208078e-06, "loss": 0.1786741852760315, "step": 430 }, { "epoch": 0.01159027075794444, "grad_norm": 4.973073482513428, "learning_rate": 7.708516242317823e-06, "loss": 0.19011027812957765, "step": 440 }, { "epoch": 0.011853686002443176, "grad_norm": 6.3872857093811035, "learning_rate": 7.88410886742757e-06, "loss": 0.1955338716506958, "step": 450 }, { "epoch": 0.012117101246941913, "grad_norm": 7.31369161605835, "learning_rate": 8.059701492537314e-06, "loss": 0.1935097098350525, "step": 460 }, { "epoch": 0.012380516491440651, "grad_norm": 5.8278584480285645, "learning_rate": 8.23529411764706e-06, "loss": 0.1809788942337036, "step": 470 }, { "epoch": 0.012643931735939388, "grad_norm": 5.682644844055176, "learning_rate": 8.410886742756806e-06, "loss": 0.1938350558280945, "step": 480 }, { "epoch": 0.012907346980438125, "grad_norm": 8.25541877746582, "learning_rate": 8.58647936786655e-06, "loss": 0.19185057878494263, "step": 490 }, { "epoch": 0.013170762224936863, "grad_norm": 13.724642753601074, "learning_rate": 8.762071992976295e-06, "loss": 0.20768375396728517, "step": 500 }, { "epoch": 0.0134341774694356, "grad_norm": 9.45821762084961, "learning_rate": 8.937664618086041e-06, "loss": 0.2102895736694336, "step": 510 }, { "epoch": 0.013697592713934337, "grad_norm": 11.051301956176758, "learning_rate": 9.113257243195787e-06, "loss": 0.20626819133758545, "step": 520 }, { "epoch": 0.013961007958433075, "grad_norm": 6.5893378257751465, "learning_rate": 9.288849868305532e-06, "loss": 0.19969332218170166, "step": 530 }, { "epoch": 0.014224423202931812, "grad_norm": 10.432629585266113, "learning_rate": 9.464442493415278e-06, "loss": 0.1959909439086914, "step": 540 }, { "epoch": 0.014487838447430549, "grad_norm": 3.2753069400787354, "learning_rate": 9.640035118525024e-06, "loss": 0.18904026746749877, "step": 550 }, { "epoch": 0.014751253691929285, "grad_norm": 5.6973795890808105, "learning_rate": 9.815627743634768e-06, "loss": 0.1823084235191345, "step": 560 }, { "epoch": 0.015014668936428024, "grad_norm": 5.830636978149414, "learning_rate": 9.991220368744513e-06, "loss": 0.19191257953643798, "step": 570 }, { "epoch": 0.01527808418092676, "grad_norm": 6.434067726135254, "learning_rate": 1.0166812993854259e-05, "loss": 0.19569530487060546, "step": 580 }, { "epoch": 0.015541499425425497, "grad_norm": 8.419146537780762, "learning_rate": 1.0342405618964005e-05, "loss": 0.18251742124557496, "step": 590 }, { "epoch": 0.015804914669924234, "grad_norm": 9.552521705627441, "learning_rate": 1.0517998244073751e-05, "loss": 0.18704030513763428, "step": 600 }, { "epoch": 0.016068329914422974, "grad_norm": 5.5397114753723145, "learning_rate": 1.0693590869183494e-05, "loss": 0.1930368423461914, "step": 610 }, { "epoch": 0.01633174515892171, "grad_norm": 5.1964240074157715, "learning_rate": 1.086918349429324e-05, "loss": 0.1817394495010376, "step": 620 }, { "epoch": 0.016595160403420448, "grad_norm": 9.948412895202637, "learning_rate": 1.1044776119402986e-05, "loss": 0.20809812545776368, "step": 630 }, { "epoch": 0.016858575647919184, "grad_norm": 7.335473537445068, "learning_rate": 1.122036874451273e-05, "loss": 0.1852678656578064, "step": 640 }, { "epoch": 0.01712199089241792, "grad_norm": 6.941799163818359, "learning_rate": 1.1395961369622477e-05, "loss": 0.19018338918685912, "step": 650 }, { "epoch": 0.017385406136916658, "grad_norm": 4.946405410766602, "learning_rate": 1.1571553994732223e-05, "loss": 0.20967531204223633, "step": 660 }, { "epoch": 0.017648821381415394, "grad_norm": 5.624059677124023, "learning_rate": 1.1747146619841967e-05, "loss": 0.19646921157836914, "step": 670 }, { "epoch": 0.017912236625914135, "grad_norm": 4.788023948669434, "learning_rate": 1.1922739244951713e-05, "loss": 0.1881968855857849, "step": 680 }, { "epoch": 0.01817565187041287, "grad_norm": 9.263679504394531, "learning_rate": 1.209833187006146e-05, "loss": 0.20204803943634034, "step": 690 }, { "epoch": 0.018439067114911608, "grad_norm": 6.563736438751221, "learning_rate": 1.2273924495171204e-05, "loss": 0.19841067790985106, "step": 700 }, { "epoch": 0.018702482359410345, "grad_norm": 7.775269508361816, "learning_rate": 1.244951712028095e-05, "loss": 0.18217201232910157, "step": 710 }, { "epoch": 0.01896589760390908, "grad_norm": 7.282246112823486, "learning_rate": 1.2625109745390696e-05, "loss": 0.18927026987075807, "step": 720 }, { "epoch": 0.019229312848407818, "grad_norm": 4.685034275054932, "learning_rate": 1.2800702370500439e-05, "loss": 0.1851009488105774, "step": 730 }, { "epoch": 0.01949272809290656, "grad_norm": 7.93074369430542, "learning_rate": 1.2976294995610185e-05, "loss": 0.19374492168426513, "step": 740 }, { "epoch": 0.019756143337405295, "grad_norm": 6.322512149810791, "learning_rate": 1.315188762071993e-05, "loss": 0.19962947368621825, "step": 750 }, { "epoch": 0.02001955858190403, "grad_norm": 7.48881196975708, "learning_rate": 1.3327480245829676e-05, "loss": 0.19475264549255372, "step": 760 }, { "epoch": 0.02028297382640277, "grad_norm": 8.127150535583496, "learning_rate": 1.3503072870939422e-05, "loss": 0.20944764614105224, "step": 770 }, { "epoch": 0.020546389070901505, "grad_norm": 4.533668041229248, "learning_rate": 1.3678665496049166e-05, "loss": 0.19578347206115723, "step": 780 }, { "epoch": 0.020809804315400242, "grad_norm": 4.146142482757568, "learning_rate": 1.3854258121158912e-05, "loss": 0.18484915494918824, "step": 790 }, { "epoch": 0.02107321955989898, "grad_norm": 14.369434356689453, "learning_rate": 1.4029850746268658e-05, "loss": 0.20894804000854492, "step": 800 }, { "epoch": 0.02133663480439772, "grad_norm": 5.428366661071777, "learning_rate": 1.4205443371378403e-05, "loss": 0.18694268465042113, "step": 810 }, { "epoch": 0.021600050048896455, "grad_norm": 3.9447431564331055, "learning_rate": 1.4381035996488149e-05, "loss": 0.18180232048034667, "step": 820 }, { "epoch": 0.021863465293395192, "grad_norm": 3.147404193878174, "learning_rate": 1.4556628621597895e-05, "loss": 0.18536533117294313, "step": 830 }, { "epoch": 0.02212688053789393, "grad_norm": 4.5036115646362305, "learning_rate": 1.4732221246707638e-05, "loss": 0.18947126865386962, "step": 840 }, { "epoch": 0.022390295782392666, "grad_norm": 5.684880256652832, "learning_rate": 1.4907813871817384e-05, "loss": 0.19054064750671387, "step": 850 }, { "epoch": 0.022653711026891402, "grad_norm": 3.831749439239502, "learning_rate": 1.508340649692713e-05, "loss": 0.18225693702697754, "step": 860 }, { "epoch": 0.022917126271390142, "grad_norm": 5.397392272949219, "learning_rate": 1.5258999122036876e-05, "loss": 0.19177463054656982, "step": 870 }, { "epoch": 0.02318054151588888, "grad_norm": 4.571076393127441, "learning_rate": 1.5434591747146622e-05, "loss": 0.20141208171844482, "step": 880 }, { "epoch": 0.023443956760387616, "grad_norm": 5.584746360778809, "learning_rate": 1.561018437225637e-05, "loss": 0.1815434455871582, "step": 890 }, { "epoch": 0.023707372004886353, "grad_norm": 3.8923778533935547, "learning_rate": 1.578577699736611e-05, "loss": 0.17717095613479614, "step": 900 }, { "epoch": 0.02397078724938509, "grad_norm": 2.5127642154693604, "learning_rate": 1.5961369622475857e-05, "loss": 0.1898900032043457, "step": 910 }, { "epoch": 0.024234202493883826, "grad_norm": 10.7523775100708, "learning_rate": 1.6136962247585604e-05, "loss": 0.19681832790374756, "step": 920 }, { "epoch": 0.024497617738382566, "grad_norm": 2.9549574851989746, "learning_rate": 1.6312554872695346e-05, "loss": 0.19814596176147461, "step": 930 }, { "epoch": 0.024761032982881303, "grad_norm": 3.572303533554077, "learning_rate": 1.6488147497805093e-05, "loss": 0.18656278848648072, "step": 940 }, { "epoch": 0.02502444822738004, "grad_norm": 3.406682014465332, "learning_rate": 1.666374012291484e-05, "loss": 0.1728109359741211, "step": 950 }, { "epoch": 0.025287863471878776, "grad_norm": 5.742796897888184, "learning_rate": 1.6839332748024585e-05, "loss": 0.18306879997253417, "step": 960 }, { "epoch": 0.025551278716377513, "grad_norm": 2.7379143238067627, "learning_rate": 1.701492537313433e-05, "loss": 0.1693957805633545, "step": 970 }, { "epoch": 0.02581469396087625, "grad_norm": 8.872781753540039, "learning_rate": 1.7190517998244077e-05, "loss": 0.20078740119934083, "step": 980 }, { "epoch": 0.026078109205374986, "grad_norm": 3.22904896736145, "learning_rate": 1.736611062335382e-05, "loss": 0.18435258865356446, "step": 990 }, { "epoch": 0.026341524449873727, "grad_norm": 7.3408379554748535, "learning_rate": 1.7541703248463566e-05, "loss": 0.19584569931030274, "step": 1000 }, { "epoch": 0.026604939694372463, "grad_norm": 5.127617359161377, "learning_rate": 1.771729587357331e-05, "loss": 0.19794195890426636, "step": 1010 }, { "epoch": 0.0268683549388712, "grad_norm": 6.323272228240967, "learning_rate": 1.7892888498683055e-05, "loss": 0.196799898147583, "step": 1020 }, { "epoch": 0.027131770183369937, "grad_norm": 3.177222728729248, "learning_rate": 1.80684811237928e-05, "loss": 0.18211159706115723, "step": 1030 }, { "epoch": 0.027395185427868674, "grad_norm": 3.2568020820617676, "learning_rate": 1.8244073748902547e-05, "loss": 0.18934638500213624, "step": 1040 }, { "epoch": 0.02765860067236741, "grad_norm": 4.00691556930542, "learning_rate": 1.8419666374012293e-05, "loss": 0.17810218334197997, "step": 1050 }, { "epoch": 0.02792201591686615, "grad_norm": 4.806098461151123, "learning_rate": 1.859525899912204e-05, "loss": 0.1959786057472229, "step": 1060 }, { "epoch": 0.028185431161364887, "grad_norm": 3.7079885005950928, "learning_rate": 1.8770851624231782e-05, "loss": 0.18069349527359008, "step": 1070 }, { "epoch": 0.028448846405863624, "grad_norm": 2.5978479385375977, "learning_rate": 1.8946444249341528e-05, "loss": 0.2076857089996338, "step": 1080 }, { "epoch": 0.02871226165036236, "grad_norm": 6.979891777038574, "learning_rate": 1.9122036874451274e-05, "loss": 0.1991156220436096, "step": 1090 }, { "epoch": 0.028975676894861097, "grad_norm": 4.067086696624756, "learning_rate": 1.929762949956102e-05, "loss": 0.1842670202255249, "step": 1100 }, { "epoch": 0.029239092139359834, "grad_norm": 3.2875936031341553, "learning_rate": 1.9473222124670767e-05, "loss": 0.19707705974578857, "step": 1110 }, { "epoch": 0.02950250738385857, "grad_norm": 5.949685096740723, "learning_rate": 1.9648814749780513e-05, "loss": 0.18080635070800782, "step": 1120 }, { "epoch": 0.02976592262835731, "grad_norm": 5.34707498550415, "learning_rate": 1.9824407374890256e-05, "loss": 0.1832728624343872, "step": 1130 }, { "epoch": 0.030029337872856048, "grad_norm": 3.1346850395202637, "learning_rate": 2e-05, "loss": 0.18595972061157226, "step": 1140 }, { "epoch": 0.030292753117354784, "grad_norm": 7.9842095375061035, "learning_rate": 1.9999996360784282e-05, "loss": 0.18741233348846437, "step": 1150 }, { "epoch": 0.03055616836185352, "grad_norm": 4.95793342590332, "learning_rate": 1.9999985443139778e-05, "loss": 0.1803184986114502, "step": 1160 }, { "epoch": 0.030819583606352258, "grad_norm": 3.345271110534668, "learning_rate": 1.999996724707443e-05, "loss": 0.18603978157043458, "step": 1170 }, { "epoch": 0.031082998850850994, "grad_norm": 4.337471961975098, "learning_rate": 1.9999941772601482e-05, "loss": 0.1772492527961731, "step": 1180 }, { "epoch": 0.03134641409534973, "grad_norm": 3.580561876296997, "learning_rate": 1.9999909019739482e-05, "loss": 0.18614537715911866, "step": 1190 }, { "epoch": 0.03160982933984847, "grad_norm": 4.757112503051758, "learning_rate": 1.9999868988512265e-05, "loss": 0.17301172018051147, "step": 1200 }, { "epoch": 0.031873244584347205, "grad_norm": 12.863966941833496, "learning_rate": 1.9999821678948964e-05, "loss": 0.17854983806610109, "step": 1210 }, { "epoch": 0.03213665982884595, "grad_norm": 5.420206546783447, "learning_rate": 1.999976709108402e-05, "loss": 0.18923416137695312, "step": 1220 }, { "epoch": 0.032400075073344685, "grad_norm": 6.902030944824219, "learning_rate": 1.9999705224957156e-05, "loss": 0.20016658306121826, "step": 1230 }, { "epoch": 0.03266349031784342, "grad_norm": 13.476462364196777, "learning_rate": 1.9999636080613406e-05, "loss": 0.19344722032546996, "step": 1240 }, { "epoch": 0.03292690556234216, "grad_norm": 5.676305294036865, "learning_rate": 1.9999559658103097e-05, "loss": 0.1957757830619812, "step": 1250 }, { "epoch": 0.033190320806840895, "grad_norm": 6.2624921798706055, "learning_rate": 1.9999475957481848e-05, "loss": 0.19853267669677735, "step": 1260 }, { "epoch": 0.03345373605133963, "grad_norm": 2.2870848178863525, "learning_rate": 1.999938497881059e-05, "loss": 0.1948472023010254, "step": 1270 }, { "epoch": 0.03371715129583837, "grad_norm": 3.760174036026001, "learning_rate": 1.999928672215553e-05, "loss": 0.18161869049072266, "step": 1280 }, { "epoch": 0.033980566540337105, "grad_norm": 2.66616153717041, "learning_rate": 1.9999181187588187e-05, "loss": 0.19087595939636232, "step": 1290 }, { "epoch": 0.03424398178483584, "grad_norm": 5.11327600479126, "learning_rate": 1.9999068375185372e-05, "loss": 0.18265395164489745, "step": 1300 }, { "epoch": 0.03450739702933458, "grad_norm": 8.652410507202148, "learning_rate": 1.99989482850292e-05, "loss": 0.18561105728149413, "step": 1310 }, { "epoch": 0.034770812273833315, "grad_norm": 2.579620361328125, "learning_rate": 1.9998820917207076e-05, "loss": 0.1833183526992798, "step": 1320 }, { "epoch": 0.03503422751833205, "grad_norm": 3.370988130569458, "learning_rate": 1.99986862718117e-05, "loss": 0.1849505066871643, "step": 1330 }, { "epoch": 0.03529764276283079, "grad_norm": 4.192021369934082, "learning_rate": 1.999854434894108e-05, "loss": 0.18968818187713624, "step": 1340 }, { "epoch": 0.03556105800732953, "grad_norm": 4.798720836639404, "learning_rate": 1.9998395148698505e-05, "loss": 0.18668742179870607, "step": 1350 }, { "epoch": 0.03582447325182827, "grad_norm": 11.151535987854004, "learning_rate": 1.9998238671192575e-05, "loss": 0.18769582509994506, "step": 1360 }, { "epoch": 0.036087888496327006, "grad_norm": 5.416773796081543, "learning_rate": 1.999807491653718e-05, "loss": 0.18964158296585082, "step": 1370 }, { "epoch": 0.03635130374082574, "grad_norm": 5.400311470031738, "learning_rate": 1.9997903884851503e-05, "loss": 0.1779132604598999, "step": 1380 }, { "epoch": 0.03661471898532448, "grad_norm": 2.6964402198791504, "learning_rate": 1.999772557626004e-05, "loss": 0.18046587705612183, "step": 1390 }, { "epoch": 0.036878134229823216, "grad_norm": 3.6984410285949707, "learning_rate": 1.9997539990892555e-05, "loss": 0.17246553897857667, "step": 1400 }, { "epoch": 0.03714154947432195, "grad_norm": 4.086616039276123, "learning_rate": 1.9997347128884138e-05, "loss": 0.1794880747795105, "step": 1410 }, { "epoch": 0.03740496471882069, "grad_norm": 2.1450629234313965, "learning_rate": 1.9997146990375162e-05, "loss": 0.1892001986503601, "step": 1420 }, { "epoch": 0.037668379963319426, "grad_norm": 4.67707633972168, "learning_rate": 1.999693957551129e-05, "loss": 0.17885711193084716, "step": 1430 }, { "epoch": 0.03793179520781816, "grad_norm": 4.924267768859863, "learning_rate": 1.999672488444349e-05, "loss": 0.17529442310333251, "step": 1440 }, { "epoch": 0.0381952104523169, "grad_norm": 4.899992942810059, "learning_rate": 1.999650291732802e-05, "loss": 0.1924641728401184, "step": 1450 }, { "epoch": 0.038458625696815636, "grad_norm": 2.0006463527679443, "learning_rate": 1.9996273674326446e-05, "loss": 0.18616669178009032, "step": 1460 }, { "epoch": 0.03872204094131437, "grad_norm": 3.5930655002593994, "learning_rate": 1.9996037155605615e-05, "loss": 0.1733822464942932, "step": 1470 }, { "epoch": 0.03898545618581312, "grad_norm": 3.709883689880371, "learning_rate": 1.9995793361337676e-05, "loss": 0.18384379148483276, "step": 1480 }, { "epoch": 0.03924887143031185, "grad_norm": 4.92201566696167, "learning_rate": 1.9995542291700075e-05, "loss": 0.1833254337310791, "step": 1490 }, { "epoch": 0.03951228667481059, "grad_norm": 1.8115609884262085, "learning_rate": 1.9995283946875548e-05, "loss": 0.17687654495239258, "step": 1500 }, { "epoch": 0.03977570191930933, "grad_norm": 6.1263580322265625, "learning_rate": 1.999501832705213e-05, "loss": 0.16596851348876954, "step": 1510 }, { "epoch": 0.04003911716380806, "grad_norm": 1.784846544265747, "learning_rate": 1.9994745432423154e-05, "loss": 0.17170164585113526, "step": 1520 }, { "epoch": 0.0403025324083068, "grad_norm": 4.241473197937012, "learning_rate": 1.9994465263187242e-05, "loss": 0.16500792503356934, "step": 1530 }, { "epoch": 0.04056594765280554, "grad_norm": 4.777111530303955, "learning_rate": 1.9994177819548313e-05, "loss": 0.19689801931381226, "step": 1540 }, { "epoch": 0.040829362897304274, "grad_norm": 3.56050443649292, "learning_rate": 1.999388310171558e-05, "loss": 0.18798370361328126, "step": 1550 }, { "epoch": 0.04109277814180301, "grad_norm": 5.652228832244873, "learning_rate": 1.9993581109903557e-05, "loss": 0.16813764572143555, "step": 1560 }, { "epoch": 0.04135619338630175, "grad_norm": 7.1295366287231445, "learning_rate": 1.999327184433204e-05, "loss": 0.17892696857452392, "step": 1570 }, { "epoch": 0.041619608630800484, "grad_norm": 2.825451612472534, "learning_rate": 1.9992955305226127e-05, "loss": 0.17450989484786988, "step": 1580 }, { "epoch": 0.04188302387529922, "grad_norm": 2.7440011501312256, "learning_rate": 1.9992631492816214e-05, "loss": 0.18088998794555664, "step": 1590 }, { "epoch": 0.04214643911979796, "grad_norm": 2.3184196949005127, "learning_rate": 1.9992300407337977e-05, "loss": 0.19379336833953859, "step": 1600 }, { "epoch": 0.0424098543642967, "grad_norm": 9.222993850708008, "learning_rate": 1.9991962049032403e-05, "loss": 0.17197897434234619, "step": 1610 }, { "epoch": 0.04267326960879544, "grad_norm": 7.191405773162842, "learning_rate": 1.9991616418145757e-05, "loss": 0.17909488677978516, "step": 1620 }, { "epoch": 0.042936684853294174, "grad_norm": 3.3159122467041016, "learning_rate": 1.9991263514929606e-05, "loss": 0.16889517307281493, "step": 1630 }, { "epoch": 0.04320010009779291, "grad_norm": 7.969412326812744, "learning_rate": 1.999090333964081e-05, "loss": 0.18881509304046631, "step": 1640 }, { "epoch": 0.04346351534229165, "grad_norm": 1.9606205224990845, "learning_rate": 1.9990535892541524e-05, "loss": 0.17080923318862914, "step": 1650 }, { "epoch": 0.043726930586790384, "grad_norm": 4.505372524261475, "learning_rate": 1.999016117389918e-05, "loss": 0.19822311401367188, "step": 1660 }, { "epoch": 0.04399034583128912, "grad_norm": 4.982966423034668, "learning_rate": 1.998977918398653e-05, "loss": 0.19067201614379883, "step": 1670 }, { "epoch": 0.04425376107578786, "grad_norm": 2.4486217498779297, "learning_rate": 1.9989389923081587e-05, "loss": 0.1702754855155945, "step": 1680 }, { "epoch": 0.044517176320286594, "grad_norm": 2.909409761428833, "learning_rate": 1.998899339146768e-05, "loss": 0.1709128737449646, "step": 1690 }, { "epoch": 0.04478059156478533, "grad_norm": 1.9277448654174805, "learning_rate": 1.9988589589433423e-05, "loss": 0.18273216485977173, "step": 1700 }, { "epoch": 0.04504400680928407, "grad_norm": 4.281924247741699, "learning_rate": 1.9988178517272717e-05, "loss": 0.18537048101425171, "step": 1710 }, { "epoch": 0.045307422053782805, "grad_norm": 3.2392818927764893, "learning_rate": 1.998776017528476e-05, "loss": 0.18133695125579835, "step": 1720 }, { "epoch": 0.04557083729828155, "grad_norm": 4.290902137756348, "learning_rate": 1.9987334563774038e-05, "loss": 0.18306388854980468, "step": 1730 }, { "epoch": 0.045834252542780285, "grad_norm": 5.442599773406982, "learning_rate": 1.998690168305033e-05, "loss": 0.1740411639213562, "step": 1740 }, { "epoch": 0.04609766778727902, "grad_norm": 1.7645630836486816, "learning_rate": 1.9986461533428708e-05, "loss": 0.1884701132774353, "step": 1750 }, { "epoch": 0.04636108303177776, "grad_norm": 3.483386516571045, "learning_rate": 1.9986014115229525e-05, "loss": 0.18106117248535156, "step": 1760 }, { "epoch": 0.046624498276276495, "grad_norm": 2.172147750854492, "learning_rate": 1.998555942877844e-05, "loss": 0.1869133710861206, "step": 1770 }, { "epoch": 0.04688791352077523, "grad_norm": 3.8447344303131104, "learning_rate": 1.9985097474406387e-05, "loss": 0.18182940483093263, "step": 1780 }, { "epoch": 0.04715132876527397, "grad_norm": 2.925926923751831, "learning_rate": 1.99846282524496e-05, "loss": 0.16939241886138917, "step": 1790 }, { "epoch": 0.047414744009772705, "grad_norm": 4.405815601348877, "learning_rate": 1.9984151763249598e-05, "loss": 0.18325445652008057, "step": 1800 }, { "epoch": 0.04767815925427144, "grad_norm": 2.362884759902954, "learning_rate": 1.9983668007153185e-05, "loss": 0.18583736419677735, "step": 1810 }, { "epoch": 0.04794157449877018, "grad_norm": 2.785599946975708, "learning_rate": 1.998317698451247e-05, "loss": 0.1737363815307617, "step": 1820 }, { "epoch": 0.048204989743268915, "grad_norm": 2.610322952270508, "learning_rate": 1.9982678695684833e-05, "loss": 0.17511745691299438, "step": 1830 }, { "epoch": 0.04846840498776765, "grad_norm": 2.7770256996154785, "learning_rate": 1.998217314103295e-05, "loss": 0.18166340589523317, "step": 1840 }, { "epoch": 0.04873182023226639, "grad_norm": 2.428842067718506, "learning_rate": 1.998166032092479e-05, "loss": 0.18520731925964357, "step": 1850 }, { "epoch": 0.04899523547676513, "grad_norm": 2.283881425857544, "learning_rate": 1.99811402357336e-05, "loss": 0.16577301025390626, "step": 1860 }, { "epoch": 0.04925865072126387, "grad_norm": 4.023209095001221, "learning_rate": 1.9980612885837925e-05, "loss": 0.1839050054550171, "step": 1870 }, { "epoch": 0.049522065965762606, "grad_norm": 2.419445276260376, "learning_rate": 1.998007827162159e-05, "loss": 0.17573541402816772, "step": 1880 }, { "epoch": 0.04978548121026134, "grad_norm": 5.434495449066162, "learning_rate": 1.9979536393473715e-05, "loss": 0.1801037073135376, "step": 1890 }, { "epoch": 0.05004889645476008, "grad_norm": 3.780832529067993, "learning_rate": 1.9978987251788696e-05, "loss": 0.1790006160736084, "step": 1900 }, { "epoch": 0.050312311699258816, "grad_norm": 5.2016825675964355, "learning_rate": 1.997843084696623e-05, "loss": 0.18657957315444945, "step": 1910 }, { "epoch": 0.05057572694375755, "grad_norm": 6.112628936767578, "learning_rate": 1.9977867179411282e-05, "loss": 0.16174499988555907, "step": 1920 }, { "epoch": 0.05083914218825629, "grad_norm": 3.8047409057617188, "learning_rate": 1.997729624953412e-05, "loss": 0.1849309802055359, "step": 1930 }, { "epoch": 0.051102557432755026, "grad_norm": 2.294853925704956, "learning_rate": 1.997671805775029e-05, "loss": 0.17933177947998047, "step": 1940 }, { "epoch": 0.05136597267725376, "grad_norm": 4.224855899810791, "learning_rate": 1.9976132604480625e-05, "loss": 0.18102643489837647, "step": 1950 }, { "epoch": 0.0516293879217525, "grad_norm": 3.1496810913085938, "learning_rate": 1.997553989015125e-05, "loss": 0.19317302703857422, "step": 1960 }, { "epoch": 0.051892803166251236, "grad_norm": 3.9058940410614014, "learning_rate": 1.9974939915193557e-05, "loss": 0.17491204738616944, "step": 1970 }, { "epoch": 0.05215621841074997, "grad_norm": 2.5774526596069336, "learning_rate": 1.9974332680044237e-05, "loss": 0.1813138484954834, "step": 1980 }, { "epoch": 0.05241963365524872, "grad_norm": 2.181058406829834, "learning_rate": 1.9973718185145262e-05, "loss": 0.16504065990447997, "step": 1990 }, { "epoch": 0.05268304889974745, "grad_norm": 5.973459243774414, "learning_rate": 1.997309643094389e-05, "loss": 0.18015506267547607, "step": 2000 }, { "epoch": 0.05294646414424619, "grad_norm": 1.8023885488510132, "learning_rate": 1.997246741789266e-05, "loss": 0.1767021894454956, "step": 2010 }, { "epoch": 0.05320987938874493, "grad_norm": 4.574063301086426, "learning_rate": 1.99718311464494e-05, "loss": 0.1899261713027954, "step": 2020 }, { "epoch": 0.05347329463324366, "grad_norm": 3.4117138385772705, "learning_rate": 1.9971187617077204e-05, "loss": 0.1831165075302124, "step": 2030 }, { "epoch": 0.0537367098777424, "grad_norm": 3.8727126121520996, "learning_rate": 1.9970536830244466e-05, "loss": 0.18381407260894775, "step": 2040 }, { "epoch": 0.05400012512224114, "grad_norm": 1.8427610397338867, "learning_rate": 1.9969878786424863e-05, "loss": 0.1744077682495117, "step": 2050 }, { "epoch": 0.054263540366739874, "grad_norm": 2.2908947467803955, "learning_rate": 1.9969213486097335e-05, "loss": 0.18389899730682374, "step": 2060 }, { "epoch": 0.05452695561123861, "grad_norm": 5.329391956329346, "learning_rate": 1.996854092974613e-05, "loss": 0.19043388366699218, "step": 2070 }, { "epoch": 0.05479037085573735, "grad_norm": 4.699202060699463, "learning_rate": 1.9967861117860754e-05, "loss": 0.17316067218780518, "step": 2080 }, { "epoch": 0.055053786100236084, "grad_norm": 1.9592821598052979, "learning_rate": 1.9967174050936008e-05, "loss": 0.17926033735275268, "step": 2090 }, { "epoch": 0.05531720134473482, "grad_norm": 2.4083690643310547, "learning_rate": 1.9966479729471966e-05, "loss": 0.19171918630599977, "step": 2100 }, { "epoch": 0.05558061658923356, "grad_norm": 5.225124835968018, "learning_rate": 1.9965778153973986e-05, "loss": 0.16785417795181273, "step": 2110 }, { "epoch": 0.0558440318337323, "grad_norm": 3.0696468353271484, "learning_rate": 1.9965069324952703e-05, "loss": 0.17760568857192993, "step": 2120 }, { "epoch": 0.05610744707823104, "grad_norm": 6.293588161468506, "learning_rate": 1.9964353242924038e-05, "loss": 0.1786039352416992, "step": 2130 }, { "epoch": 0.056370862322729774, "grad_norm": 2.1680946350097656, "learning_rate": 1.9963629908409184e-05, "loss": 0.17985448837280274, "step": 2140 }, { "epoch": 0.05663427756722851, "grad_norm": 2.2423644065856934, "learning_rate": 1.9962899321934616e-05, "loss": 0.16538684368133544, "step": 2150 }, { "epoch": 0.05689769281172725, "grad_norm": 2.691453695297241, "learning_rate": 1.9962161484032085e-05, "loss": 0.16381483078002929, "step": 2160 }, { "epoch": 0.057161108056225984, "grad_norm": 3.0502705574035645, "learning_rate": 1.9961416395238623e-05, "loss": 0.15508612394332885, "step": 2170 }, { "epoch": 0.05742452330072472, "grad_norm": 4.203874588012695, "learning_rate": 1.9960664056096534e-05, "loss": 0.19700683355331422, "step": 2180 }, { "epoch": 0.05768793854522346, "grad_norm": 8.176562309265137, "learning_rate": 1.9959904467153408e-05, "loss": 0.18920013904571534, "step": 2190 }, { "epoch": 0.057951353789722194, "grad_norm": 1.50550377368927, "learning_rate": 1.99591376289621e-05, "loss": 0.1680821180343628, "step": 2200 }, { "epoch": 0.05821476903422093, "grad_norm": 3.606895685195923, "learning_rate": 1.9958363542080758e-05, "loss": 0.17289118766784667, "step": 2210 }, { "epoch": 0.05847818427871967, "grad_norm": 2.4037699699401855, "learning_rate": 1.9957582207072784e-05, "loss": 0.16795046329498292, "step": 2220 }, { "epoch": 0.058741599523218405, "grad_norm": 4.020079612731934, "learning_rate": 1.9956793624506875e-05, "loss": 0.18110791444778443, "step": 2230 }, { "epoch": 0.05900501476771714, "grad_norm": 5.8807291984558105, "learning_rate": 1.9955997794956994e-05, "loss": 0.16692888736724854, "step": 2240 }, { "epoch": 0.059268430012215885, "grad_norm": 4.712007522583008, "learning_rate": 1.995519471900238e-05, "loss": 0.1733164072036743, "step": 2250 }, { "epoch": 0.05953184525671462, "grad_norm": 2.2142868041992188, "learning_rate": 1.9954384397227545e-05, "loss": 0.17442538738250732, "step": 2260 }, { "epoch": 0.05979526050121336, "grad_norm": 5.314425468444824, "learning_rate": 1.9953566830222276e-05, "loss": 0.17016592025756835, "step": 2270 }, { "epoch": 0.060058675745712095, "grad_norm": 3.5268280506134033, "learning_rate": 1.9952742018581634e-05, "loss": 0.17992749214172363, "step": 2280 }, { "epoch": 0.06032209099021083, "grad_norm": 3.850910186767578, "learning_rate": 1.9951909962905956e-05, "loss": 0.17061890363693238, "step": 2290 }, { "epoch": 0.06058550623470957, "grad_norm": 6.055057525634766, "learning_rate": 1.9951070663800843e-05, "loss": 0.16783838272094725, "step": 2300 }, { "epoch": 0.060848921479208305, "grad_norm": 1.7355868816375732, "learning_rate": 1.995022412187718e-05, "loss": 0.16899465322494506, "step": 2310 }, { "epoch": 0.06111233672370704, "grad_norm": 3.89322566986084, "learning_rate": 1.9949370337751105e-05, "loss": 0.17713162899017335, "step": 2320 }, { "epoch": 0.06137575196820578, "grad_norm": 3.6141517162323, "learning_rate": 1.994850931204405e-05, "loss": 0.17213895320892333, "step": 2330 }, { "epoch": 0.061639167212704515, "grad_norm": 7.712164402008057, "learning_rate": 1.99476410453827e-05, "loss": 0.18736391067504882, "step": 2340 }, { "epoch": 0.06190258245720325, "grad_norm": 4.759211540222168, "learning_rate": 1.994676553839902e-05, "loss": 0.18477323055267333, "step": 2350 }, { "epoch": 0.06216599770170199, "grad_norm": 3.493088483810425, "learning_rate": 1.9945882791730246e-05, "loss": 0.18695447444915772, "step": 2360 }, { "epoch": 0.06242941294620073, "grad_norm": 4.4327778816223145, "learning_rate": 1.994499280601887e-05, "loss": 0.17105319499969482, "step": 2370 }, { "epoch": 0.06269282819069946, "grad_norm": 4.168586730957031, "learning_rate": 1.994409558191267e-05, "loss": 0.165738046169281, "step": 2380 }, { "epoch": 0.0629562434351982, "grad_norm": 6.160863876342773, "learning_rate": 1.9943191120064672e-05, "loss": 0.18477983474731446, "step": 2390 }, { "epoch": 0.06321965867969694, "grad_norm": 2.920475482940674, "learning_rate": 1.99422794211332e-05, "loss": 0.1701115608215332, "step": 2400 }, { "epoch": 0.06348307392419568, "grad_norm": 5.345788478851318, "learning_rate": 1.9941360485781814e-05, "loss": 0.17670942544937135, "step": 2410 }, { "epoch": 0.06374648916869441, "grad_norm": 2.233072280883789, "learning_rate": 1.9940434314679363e-05, "loss": 0.1710216522216797, "step": 2420 }, { "epoch": 0.06400990441319315, "grad_norm": 2.754680871963501, "learning_rate": 1.9939500908499947e-05, "loss": 0.17032406330108643, "step": 2430 }, { "epoch": 0.0642733196576919, "grad_norm": 8.24351692199707, "learning_rate": 1.9938560267922948e-05, "loss": 0.18030173778533937, "step": 2440 }, { "epoch": 0.06453673490219063, "grad_norm": 2.435267210006714, "learning_rate": 1.9937612393632995e-05, "loss": 0.15470414161682128, "step": 2450 }, { "epoch": 0.06480015014668937, "grad_norm": 2.9143879413604736, "learning_rate": 1.9936657286320003e-05, "loss": 0.15969743728637695, "step": 2460 }, { "epoch": 0.0650635653911881, "grad_norm": 2.649341344833374, "learning_rate": 1.993569494667913e-05, "loss": 0.17602367401123048, "step": 2470 }, { "epoch": 0.06532698063568684, "grad_norm": 4.446321964263916, "learning_rate": 1.9934725375410812e-05, "loss": 0.18866920471191406, "step": 2480 }, { "epoch": 0.06559039588018557, "grad_norm": 2.4203391075134277, "learning_rate": 1.9933748573220745e-05, "loss": 0.18160107135772705, "step": 2490 }, { "epoch": 0.06585381112468432, "grad_norm": 5.421919345855713, "learning_rate": 1.993276454081989e-05, "loss": 0.17109525203704834, "step": 2500 }, { "epoch": 0.06611722636918305, "grad_norm": 3.009272336959839, "learning_rate": 1.9931773278924467e-05, "loss": 0.18315014839172364, "step": 2510 }, { "epoch": 0.06638064161368179, "grad_norm": 3.0740206241607666, "learning_rate": 1.9930774788255957e-05, "loss": 0.18382885456085205, "step": 2520 }, { "epoch": 0.06664405685818052, "grad_norm": 2.4829816818237305, "learning_rate": 1.992976906954111e-05, "loss": 0.18483729362487794, "step": 2530 }, { "epoch": 0.06690747210267926, "grad_norm": 2.3147408962249756, "learning_rate": 1.992875612351192e-05, "loss": 0.17436102628707886, "step": 2540 }, { "epoch": 0.067170887347178, "grad_norm": 1.1114705801010132, "learning_rate": 1.9927735950905663e-05, "loss": 0.16056630611419678, "step": 2550 }, { "epoch": 0.06743430259167674, "grad_norm": 1.3798120021820068, "learning_rate": 1.9926708552464865e-05, "loss": 0.1951879858970642, "step": 2560 }, { "epoch": 0.06769771783617548, "grad_norm": 6.1253743171691895, "learning_rate": 1.9925673928937307e-05, "loss": 0.1824423551559448, "step": 2570 }, { "epoch": 0.06796113308067421, "grad_norm": 2.9715278148651123, "learning_rate": 1.9924632081076032e-05, "loss": 0.17028138637542725, "step": 2580 }, { "epoch": 0.06822454832517295, "grad_norm": 2.1775808334350586, "learning_rate": 1.9923583009639344e-05, "loss": 0.19221780300140381, "step": 2590 }, { "epoch": 0.06848796356967168, "grad_norm": 3.4760570526123047, "learning_rate": 1.9922526715390802e-05, "loss": 0.1733992576599121, "step": 2600 }, { "epoch": 0.06875137881417043, "grad_norm": 4.540346145629883, "learning_rate": 1.9921463199099222e-05, "loss": 0.17315328121185303, "step": 2610 }, { "epoch": 0.06901479405866916, "grad_norm": 2.1484668254852295, "learning_rate": 1.992039246153868e-05, "loss": 0.18548152446746827, "step": 2620 }, { "epoch": 0.0692782093031679, "grad_norm": 4.389924049377441, "learning_rate": 1.99193145034885e-05, "loss": 0.16876236200332642, "step": 2630 }, { "epoch": 0.06954162454766663, "grad_norm": 1.501552939414978, "learning_rate": 1.9918229325733267e-05, "loss": 0.16910407543182374, "step": 2640 }, { "epoch": 0.06980503979216537, "grad_norm": 1.7707685232162476, "learning_rate": 1.9917136929062824e-05, "loss": 0.17019779682159425, "step": 2650 }, { "epoch": 0.0700684550366641, "grad_norm": 4.278175354003906, "learning_rate": 1.991603731427226e-05, "loss": 0.1723952293395996, "step": 2660 }, { "epoch": 0.07033187028116285, "grad_norm": 6.3203654289245605, "learning_rate": 1.9914930482161928e-05, "loss": 0.16744835376739503, "step": 2670 }, { "epoch": 0.07059528552566158, "grad_norm": 2.759131669998169, "learning_rate": 1.9913816433537425e-05, "loss": 0.1594109296798706, "step": 2680 }, { "epoch": 0.07085870077016032, "grad_norm": 3.2137975692749023, "learning_rate": 1.99126951692096e-05, "loss": 0.17708659172058105, "step": 2690 }, { "epoch": 0.07112211601465906, "grad_norm": 3.270878791809082, "learning_rate": 1.991156668999456e-05, "loss": 0.1791637897491455, "step": 2700 }, { "epoch": 0.0713855312591578, "grad_norm": 2.2882707118988037, "learning_rate": 1.9910430996713668e-05, "loss": 0.1647800087928772, "step": 2710 }, { "epoch": 0.07164894650365654, "grad_norm": 3.8899245262145996, "learning_rate": 1.9909288090193518e-05, "loss": 0.181083607673645, "step": 2720 }, { "epoch": 0.07191236174815527, "grad_norm": 1.813941478729248, "learning_rate": 1.9908137971265977e-05, "loss": 0.18577444553375244, "step": 2730 }, { "epoch": 0.07217577699265401, "grad_norm": 2.6823368072509766, "learning_rate": 1.9906980640768146e-05, "loss": 0.17710893154144286, "step": 2740 }, { "epoch": 0.07243919223715274, "grad_norm": 4.8263702392578125, "learning_rate": 1.9905816099542382e-05, "loss": 0.1723855257034302, "step": 2750 }, { "epoch": 0.07270260748165149, "grad_norm": 4.082388877868652, "learning_rate": 1.9904644348436282e-05, "loss": 0.1748762845993042, "step": 2760 }, { "epoch": 0.07296602272615021, "grad_norm": 2.7506601810455322, "learning_rate": 1.990346538830271e-05, "loss": 0.17155351638793945, "step": 2770 }, { "epoch": 0.07322943797064896, "grad_norm": 2.685300827026367, "learning_rate": 1.9902279219999752e-05, "loss": 0.1649477481842041, "step": 2780 }, { "epoch": 0.07349285321514769, "grad_norm": 4.830761432647705, "learning_rate": 1.9901085844390757e-05, "loss": 0.16968194246292115, "step": 2790 }, { "epoch": 0.07375626845964643, "grad_norm": 7.105067729949951, "learning_rate": 1.9899885262344316e-05, "loss": 0.18046623468399048, "step": 2800 }, { "epoch": 0.07401968370414516, "grad_norm": 2.9547553062438965, "learning_rate": 1.9898677474734266e-05, "loss": 0.17206363677978515, "step": 2810 }, { "epoch": 0.0742830989486439, "grad_norm": 2.9760494232177734, "learning_rate": 1.9897462482439684e-05, "loss": 0.16697492599487304, "step": 2820 }, { "epoch": 0.07454651419314265, "grad_norm": 4.790968894958496, "learning_rate": 1.9896240286344892e-05, "loss": 0.18875933885574342, "step": 2830 }, { "epoch": 0.07480992943764138, "grad_norm": 3.63333797454834, "learning_rate": 1.989501088733946e-05, "loss": 0.17478711605072023, "step": 2840 }, { "epoch": 0.07507334468214012, "grad_norm": 1.67331063747406, "learning_rate": 1.9893774286318198e-05, "loss": 0.17434098720550537, "step": 2850 }, { "epoch": 0.07533675992663885, "grad_norm": 7.925144195556641, "learning_rate": 1.9892530484181155e-05, "loss": 0.1792346715927124, "step": 2860 }, { "epoch": 0.0756001751711376, "grad_norm": 5.535426616668701, "learning_rate": 1.989127948183363e-05, "loss": 0.1687788724899292, "step": 2870 }, { "epoch": 0.07586359041563633, "grad_norm": 3.4412753582000732, "learning_rate": 1.9890021280186148e-05, "loss": 0.1884918212890625, "step": 2880 }, { "epoch": 0.07612700566013507, "grad_norm": 4.3645548820495605, "learning_rate": 1.9888755880154485e-05, "loss": 0.1582123041152954, "step": 2890 }, { "epoch": 0.0763904209046338, "grad_norm": 4.105087757110596, "learning_rate": 1.988748328265966e-05, "loss": 0.15279684066772461, "step": 2900 }, { "epoch": 0.07665383614913254, "grad_norm": 2.394240140914917, "learning_rate": 1.9886203488627918e-05, "loss": 0.1868112564086914, "step": 2910 }, { "epoch": 0.07691725139363127, "grad_norm": 4.139469623565674, "learning_rate": 1.988491649899075e-05, "loss": 0.17514727115631104, "step": 2920 }, { "epoch": 0.07718066663813002, "grad_norm": 4.841442108154297, "learning_rate": 1.988362231468488e-05, "loss": 0.16254594326019287, "step": 2930 }, { "epoch": 0.07744408188262875, "grad_norm": 1.664572834968567, "learning_rate": 1.9882320936652278e-05, "loss": 0.17915699481964112, "step": 2940 }, { "epoch": 0.07770749712712749, "grad_norm": 2.8251044750213623, "learning_rate": 1.9881012365840138e-05, "loss": 0.16991913318634033, "step": 2950 }, { "epoch": 0.07797091237162623, "grad_norm": 2.4157230854034424, "learning_rate": 1.987969660320089e-05, "loss": 0.17600449323654174, "step": 2960 }, { "epoch": 0.07823432761612496, "grad_norm": 1.8973982334136963, "learning_rate": 1.9878373649692213e-05, "loss": 0.18735899925231933, "step": 2970 }, { "epoch": 0.0784977428606237, "grad_norm": 2.1171324253082275, "learning_rate": 1.9877043506277004e-05, "loss": 0.18300366401672363, "step": 2980 }, { "epoch": 0.07876115810512244, "grad_norm": 3.6777334213256836, "learning_rate": 1.9875706173923396e-05, "loss": 0.17323405742645265, "step": 2990 }, { "epoch": 0.07902457334962118, "grad_norm": 2.5737922191619873, "learning_rate": 1.9874361653604765e-05, "loss": 0.16038062572479247, "step": 3000 }, { "epoch": 0.07928798859411991, "grad_norm": 3.463996410369873, "learning_rate": 1.9873009946299702e-05, "loss": 0.18302521705627442, "step": 3010 }, { "epoch": 0.07955140383861865, "grad_norm": 3.541024923324585, "learning_rate": 1.9871651052992046e-05, "loss": 0.1644590139389038, "step": 3020 }, { "epoch": 0.07981481908311738, "grad_norm": 23.722578048706055, "learning_rate": 1.987028497467085e-05, "loss": 0.1561272382736206, "step": 3030 }, { "epoch": 0.08007823432761613, "grad_norm": 28.584827423095703, "learning_rate": 1.986891171233041e-05, "loss": 2.7920976638793946, "step": 3040 }, { "epoch": 0.08034164957211486, "grad_norm": 4.7452006340026855, "learning_rate": 1.9867531266970247e-05, "loss": 0.3303677558898926, "step": 3050 }, { "epoch": 0.0806050648166136, "grad_norm": 1.2053817510604858, "learning_rate": 1.9866143639595107e-05, "loss": 0.24787664413452148, "step": 3060 }, { "epoch": 0.08086848006111233, "grad_norm": 2.489018440246582, "learning_rate": 1.9864748831214964e-05, "loss": 0.2344205856323242, "step": 3070 }, { "epoch": 0.08113189530561107, "grad_norm": 9.228951454162598, "learning_rate": 1.986334684284502e-05, "loss": 0.24973616600036622, "step": 3080 }, { "epoch": 0.08139531055010982, "grad_norm": 5.184675216674805, "learning_rate": 1.98619376755057e-05, "loss": 0.232269811630249, "step": 3090 }, { "epoch": 0.08165872579460855, "grad_norm": 2.4573352336883545, "learning_rate": 1.986052133022266e-05, "loss": 0.23588459491729735, "step": 3100 }, { "epoch": 0.08192214103910729, "grad_norm": 5.99935245513916, "learning_rate": 1.985909780802678e-05, "loss": 0.21793489456176757, "step": 3110 }, { "epoch": 0.08218555628360602, "grad_norm": 2.7376482486724854, "learning_rate": 1.985766710995416e-05, "loss": 0.21219420433044434, "step": 3120 }, { "epoch": 0.08244897152810476, "grad_norm": 10.990240097045898, "learning_rate": 1.9856229237046115e-05, "loss": 0.22406415939331054, "step": 3130 }, { "epoch": 0.0827123867726035, "grad_norm": 2.894727945327759, "learning_rate": 1.9854784190349203e-05, "loss": 0.22075304985046387, "step": 3140 }, { "epoch": 0.08297580201710224, "grad_norm": 1.9598814249038696, "learning_rate": 1.985333197091518e-05, "loss": 0.20621285438537598, "step": 3150 }, { "epoch": 0.08323921726160097, "grad_norm": 6.447747707366943, "learning_rate": 1.9851872579801044e-05, "loss": 0.19751291275024413, "step": 3160 }, { "epoch": 0.08350263250609971, "grad_norm": 5.041415691375732, "learning_rate": 1.9850406018068997e-05, "loss": 0.1875074863433838, "step": 3170 }, { "epoch": 0.08376604775059844, "grad_norm": 1.6489659547805786, "learning_rate": 1.984893228678647e-05, "loss": 0.18590322732925416, "step": 3180 }, { "epoch": 0.08402946299509718, "grad_norm": 3.009755849838257, "learning_rate": 1.9847451387026103e-05, "loss": 0.19208320379257202, "step": 3190 }, { "epoch": 0.08429287823959591, "grad_norm": 3.951791286468506, "learning_rate": 1.9845963319865758e-05, "loss": 0.20527005195617676, "step": 3200 }, { "epoch": 0.08455629348409466, "grad_norm": 1.579649567604065, "learning_rate": 1.9844468086388523e-05, "loss": 0.1827871322631836, "step": 3210 }, { "epoch": 0.0848197087285934, "grad_norm": 3.826097011566162, "learning_rate": 1.9842965687682686e-05, "loss": 0.18613487482070923, "step": 3220 }, { "epoch": 0.08508312397309213, "grad_norm": 2.51517915725708, "learning_rate": 1.9841456124841763e-05, "loss": 0.2000732421875, "step": 3230 }, { "epoch": 0.08534653921759087, "grad_norm": 1.6248602867126465, "learning_rate": 1.983993939896447e-05, "loss": 0.1770808458328247, "step": 3240 }, { "epoch": 0.0856099544620896, "grad_norm": 1.9973970651626587, "learning_rate": 1.9838415511154755e-05, "loss": 0.20615105628967284, "step": 3250 }, { "epoch": 0.08587336970658835, "grad_norm": 1.9068961143493652, "learning_rate": 1.9836884462521768e-05, "loss": 0.1818971037864685, "step": 3260 }, { "epoch": 0.08613678495108708, "grad_norm": 3.0783579349517822, "learning_rate": 1.9835346254179865e-05, "loss": 0.18954063653945924, "step": 3270 }, { "epoch": 0.08640020019558582, "grad_norm": 6.42568302154541, "learning_rate": 1.9833800887248628e-05, "loss": 0.19929811954498292, "step": 3280 }, { "epoch": 0.08666361544008455, "grad_norm": 7.794387340545654, "learning_rate": 1.983224836285284e-05, "loss": 0.20740840435028077, "step": 3290 }, { "epoch": 0.0869270306845833, "grad_norm": 7.942866802215576, "learning_rate": 1.9830688682122494e-05, "loss": 0.21026244163513183, "step": 3300 }, { "epoch": 0.08719044592908202, "grad_norm": 1.275241494178772, "learning_rate": 1.982912184619279e-05, "loss": 0.19614675045013427, "step": 3310 }, { "epoch": 0.08745386117358077, "grad_norm": 8.339273452758789, "learning_rate": 1.9827547856204147e-05, "loss": 0.18812782764434816, "step": 3320 }, { "epoch": 0.0877172764180795, "grad_norm": 4.23816442489624, "learning_rate": 1.9825966713302172e-05, "loss": 0.19103646278381348, "step": 3330 }, { "epoch": 0.08798069166257824, "grad_norm": 3.5992443561553955, "learning_rate": 1.98243784186377e-05, "loss": 0.18873465061187744, "step": 3340 }, { "epoch": 0.08824410690707699, "grad_norm": 3.003166437149048, "learning_rate": 1.982278297336675e-05, "loss": 0.19645187854766846, "step": 3350 }, { "epoch": 0.08850752215157572, "grad_norm": 3.091383457183838, "learning_rate": 1.9821180378650563e-05, "loss": 0.2010653257369995, "step": 3360 }, { "epoch": 0.08877093739607446, "grad_norm": 3.816319465637207, "learning_rate": 1.9819570635655577e-05, "loss": 0.19104039669036865, "step": 3370 }, { "epoch": 0.08903435264057319, "grad_norm": 4.380512714385986, "learning_rate": 1.981795374555343e-05, "loss": 0.18695712089538574, "step": 3380 }, { "epoch": 0.08929776788507193, "grad_norm": 2.90265154838562, "learning_rate": 1.981632970952096e-05, "loss": 0.17370958328247071, "step": 3390 }, { "epoch": 0.08956118312957066, "grad_norm": 1.915366768836975, "learning_rate": 1.9814698528740222e-05, "loss": 0.19185254573822022, "step": 3400 }, { "epoch": 0.0898245983740694, "grad_norm": 5.130300045013428, "learning_rate": 1.9813060204398448e-05, "loss": 0.19022200107574463, "step": 3410 }, { "epoch": 0.09008801361856814, "grad_norm": 2.5253820419311523, "learning_rate": 1.9811414737688088e-05, "loss": 0.1927955150604248, "step": 3420 }, { "epoch": 0.09035142886306688, "grad_norm": 2.8772919178009033, "learning_rate": 1.9809762129806777e-05, "loss": 0.19529761075973512, "step": 3430 }, { "epoch": 0.09061484410756561, "grad_norm": 4.046281337738037, "learning_rate": 1.9808102381957364e-05, "loss": 0.1932288885116577, "step": 3440 }, { "epoch": 0.09087825935206435, "grad_norm": 2.1114866733551025, "learning_rate": 1.980643549534788e-05, "loss": 0.187571382522583, "step": 3450 }, { "epoch": 0.0911416745965631, "grad_norm": 1.0399141311645508, "learning_rate": 1.980476147119155e-05, "loss": 0.17468459606170655, "step": 3460 }, { "epoch": 0.09140508984106183, "grad_norm": 1.578346610069275, "learning_rate": 1.980308031070682e-05, "loss": 0.17306201457977294, "step": 3470 }, { "epoch": 0.09166850508556057, "grad_norm": 2.079922676086426, "learning_rate": 1.9801392015117286e-05, "loss": 0.19301185607910157, "step": 3480 }, { "epoch": 0.0919319203300593, "grad_norm": 4.277310848236084, "learning_rate": 1.9799696585651782e-05, "loss": 0.1675293564796448, "step": 3490 }, { "epoch": 0.09219533557455804, "grad_norm": 1.8822944164276123, "learning_rate": 1.9797994023544307e-05, "loss": 0.1965293526649475, "step": 3500 }, { "epoch": 0.09245875081905677, "grad_norm": 1.7470754384994507, "learning_rate": 1.979628433003406e-05, "loss": 0.1806775450706482, "step": 3510 }, { "epoch": 0.09272216606355552, "grad_norm": 2.121969223022461, "learning_rate": 1.9794567506365432e-05, "loss": 0.1824957013130188, "step": 3520 }, { "epoch": 0.09298558130805425, "grad_norm": 1.9035098552703857, "learning_rate": 1.9792843553788e-05, "loss": 0.17550466060638428, "step": 3530 }, { "epoch": 0.09324899655255299, "grad_norm": 2.955170154571533, "learning_rate": 1.9791112473556528e-05, "loss": 0.18684909343719483, "step": 3540 }, { "epoch": 0.09351241179705172, "grad_norm": 3.4219257831573486, "learning_rate": 1.978937426693097e-05, "loss": 0.18127481937408446, "step": 3550 }, { "epoch": 0.09377582704155046, "grad_norm": 2.4930479526519775, "learning_rate": 1.9787628935176476e-05, "loss": 0.18282269239425658, "step": 3560 }, { "epoch": 0.0940392422860492, "grad_norm": 4.7687153816223145, "learning_rate": 1.9785876479563367e-05, "loss": 0.18722889423370362, "step": 3570 }, { "epoch": 0.09430265753054794, "grad_norm": 3.4208271503448486, "learning_rate": 1.978411690136716e-05, "loss": 0.182647967338562, "step": 3580 }, { "epoch": 0.09456607277504668, "grad_norm": 4.212703704833984, "learning_rate": 1.9782350201868547e-05, "loss": 0.17892252206802367, "step": 3590 }, { "epoch": 0.09482948801954541, "grad_norm": 2.8523216247558594, "learning_rate": 1.978057638235341e-05, "loss": 0.17533000707626342, "step": 3600 }, { "epoch": 0.09509290326404415, "grad_norm": 3.7378389835357666, "learning_rate": 1.9778795444112812e-05, "loss": 0.18288695812225342, "step": 3610 }, { "epoch": 0.09535631850854288, "grad_norm": 7.939394950866699, "learning_rate": 1.9777007388443e-05, "loss": 0.17423038482666015, "step": 3620 }, { "epoch": 0.09561973375304163, "grad_norm": 3.628375768661499, "learning_rate": 1.9775212216645393e-05, "loss": 0.17674527168273926, "step": 3630 }, { "epoch": 0.09588314899754036, "grad_norm": 2.0059709548950195, "learning_rate": 1.977340993002659e-05, "loss": 0.19062167406082153, "step": 3640 }, { "epoch": 0.0961465642420391, "grad_norm": 3.462477207183838, "learning_rate": 1.977160052989839e-05, "loss": 0.190860116481781, "step": 3650 }, { "epoch": 0.09640997948653783, "grad_norm": 2.364314317703247, "learning_rate": 1.9769784017577734e-05, "loss": 0.17107152938842773, "step": 3660 }, { "epoch": 0.09667339473103657, "grad_norm": 9.497424125671387, "learning_rate": 1.976796039438677e-05, "loss": 0.19047200679779053, "step": 3670 }, { "epoch": 0.0969368099755353, "grad_norm": 7.422232151031494, "learning_rate": 1.9766129661652803e-05, "loss": 0.19752004146575927, "step": 3680 }, { "epoch": 0.09720022522003405, "grad_norm": 3.809910297393799, "learning_rate": 1.9764291820708324e-05, "loss": 0.18852641582489013, "step": 3690 }, { "epoch": 0.09746364046453278, "grad_norm": 4.315698623657227, "learning_rate": 1.9762446872890984e-05, "loss": 0.17786502838134766, "step": 3700 }, { "epoch": 0.09772705570903152, "grad_norm": 2.6497087478637695, "learning_rate": 1.976059481954363e-05, "loss": 0.18953579664230347, "step": 3710 }, { "epoch": 0.09799047095353026, "grad_norm": 1.8291451930999756, "learning_rate": 1.9758735662014253e-05, "loss": 0.16662790775299072, "step": 3720 }, { "epoch": 0.098253886198029, "grad_norm": 2.479123115539551, "learning_rate": 1.975686940165604e-05, "loss": 0.16723244190216063, "step": 3730 }, { "epoch": 0.09851730144252774, "grad_norm": 2.5788750648498535, "learning_rate": 1.9754996039827322e-05, "loss": 0.16323435306549072, "step": 3740 }, { "epoch": 0.09878071668702647, "grad_norm": 2.3780648708343506, "learning_rate": 1.9753115577891623e-05, "loss": 0.17154557704925538, "step": 3750 }, { "epoch": 0.09904413193152521, "grad_norm": 1.3977123498916626, "learning_rate": 1.9751228017217623e-05, "loss": 0.1650092601776123, "step": 3760 }, { "epoch": 0.09930754717602394, "grad_norm": 2.7307846546173096, "learning_rate": 1.9749333359179166e-05, "loss": 0.17222151756286622, "step": 3770 }, { "epoch": 0.09957096242052269, "grad_norm": 2.580350875854492, "learning_rate": 1.974743160515527e-05, "loss": 0.17034196853637695, "step": 3780 }, { "epoch": 0.09983437766502141, "grad_norm": 3.503479242324829, "learning_rate": 1.974552275653011e-05, "loss": 0.17879910469055177, "step": 3790 }, { "epoch": 0.10009779290952016, "grad_norm": 3.279223918914795, "learning_rate": 1.974360681469303e-05, "loss": 0.17830684185028076, "step": 3800 }, { "epoch": 0.10036120815401889, "grad_norm": 2.142148017883301, "learning_rate": 1.9741683781038538e-05, "loss": 0.17638206481933594, "step": 3810 }, { "epoch": 0.10062462339851763, "grad_norm": 3.6368846893310547, "learning_rate": 1.9739753656966294e-05, "loss": 0.17462124824523925, "step": 3820 }, { "epoch": 0.10088803864301636, "grad_norm": 3.247462272644043, "learning_rate": 1.9737816443881134e-05, "loss": 0.1868587017059326, "step": 3830 }, { "epoch": 0.1011514538875151, "grad_norm": 2.9462051391601562, "learning_rate": 1.973587214319304e-05, "loss": 0.16580522060394287, "step": 3840 }, { "epoch": 0.10141486913201385, "grad_norm": 1.8859926462173462, "learning_rate": 1.9733920756317157e-05, "loss": 0.17364673614501952, "step": 3850 }, { "epoch": 0.10167828437651258, "grad_norm": 2.073545455932617, "learning_rate": 1.973196228467379e-05, "loss": 0.17359886169433594, "step": 3860 }, { "epoch": 0.10194169962101132, "grad_norm": 2.855222463607788, "learning_rate": 1.9729996729688403e-05, "loss": 0.17968978881835937, "step": 3870 }, { "epoch": 0.10220511486551005, "grad_norm": 2.4405477046966553, "learning_rate": 1.9728024092791606e-05, "loss": 0.17973893880844116, "step": 3880 }, { "epoch": 0.1024685301100088, "grad_norm": 1.9924967288970947, "learning_rate": 1.9726044375419172e-05, "loss": 0.17884178161621095, "step": 3890 }, { "epoch": 0.10273194535450753, "grad_norm": 5.056756973266602, "learning_rate": 1.9724057579012026e-05, "loss": 0.17908120155334473, "step": 3900 }, { "epoch": 0.10299536059900627, "grad_norm": 3.4916493892669678, "learning_rate": 1.972206370501624e-05, "loss": 0.18018542528152465, "step": 3910 }, { "epoch": 0.103258775843505, "grad_norm": 3.281074285507202, "learning_rate": 1.9720062754883048e-05, "loss": 0.17581722736358643, "step": 3920 }, { "epoch": 0.10352219108800374, "grad_norm": 1.312873363494873, "learning_rate": 1.971805473006882e-05, "loss": 0.16717392206192017, "step": 3930 }, { "epoch": 0.10378560633250247, "grad_norm": 3.037243127822876, "learning_rate": 1.9716039632035086e-05, "loss": 0.16684823036193847, "step": 3940 }, { "epoch": 0.10404902157700122, "grad_norm": 2.096386432647705, "learning_rate": 1.9714017462248523e-05, "loss": 0.19852685928344727, "step": 3950 }, { "epoch": 0.10431243682149995, "grad_norm": 3.1457369327545166, "learning_rate": 1.9711988222180958e-05, "loss": 0.18713669776916503, "step": 3960 }, { "epoch": 0.10457585206599869, "grad_norm": 4.247480869293213, "learning_rate": 1.9709951913309347e-05, "loss": 0.1823624849319458, "step": 3970 }, { "epoch": 0.10483926731049743, "grad_norm": 2.1507174968719482, "learning_rate": 1.9707908537115816e-05, "loss": 0.17958511114120485, "step": 3980 }, { "epoch": 0.10510268255499616, "grad_norm": 2.9655656814575195, "learning_rate": 1.9705858095087613e-05, "loss": 0.18231666088104248, "step": 3990 }, { "epoch": 0.1053660977994949, "grad_norm": 2.70013165473938, "learning_rate": 1.9703800588717144e-05, "loss": 0.16808843612670898, "step": 4000 }, { "epoch": 0.10562951304399364, "grad_norm": 2.706859827041626, "learning_rate": 1.9701736019501948e-05, "loss": 0.18307673931121826, "step": 4010 }, { "epoch": 0.10589292828849238, "grad_norm": 3.3193695545196533, "learning_rate": 1.969966438894471e-05, "loss": 0.16947674751281738, "step": 4020 }, { "epoch": 0.10615634353299111, "grad_norm": 2.007354974746704, "learning_rate": 1.969758569855325e-05, "loss": 0.17301636934280396, "step": 4030 }, { "epoch": 0.10641975877748985, "grad_norm": 1.3479503393173218, "learning_rate": 1.969549994984053e-05, "loss": 0.181119966506958, "step": 4040 }, { "epoch": 0.10668317402198858, "grad_norm": 4.362004280090332, "learning_rate": 1.9693407144324642e-05, "loss": 0.1536334276199341, "step": 4050 }, { "epoch": 0.10694658926648733, "grad_norm": 3.614828586578369, "learning_rate": 1.9691307283528834e-05, "loss": 0.18264180421829224, "step": 4060 }, { "epoch": 0.10721000451098606, "grad_norm": 2.7675445079803467, "learning_rate": 1.968920036898146e-05, "loss": 0.17776143550872803, "step": 4070 }, { "epoch": 0.1074734197554848, "grad_norm": 2.912280559539795, "learning_rate": 1.9687086402216033e-05, "loss": 0.16595890522003173, "step": 4080 }, { "epoch": 0.10773683499998353, "grad_norm": 2.1039044857025146, "learning_rate": 1.968496538477118e-05, "loss": 0.1696398973464966, "step": 4090 }, { "epoch": 0.10800025024448227, "grad_norm": 4.308491230010986, "learning_rate": 1.9682837318190683e-05, "loss": 0.1928565263748169, "step": 4100 }, { "epoch": 0.10826366548898102, "grad_norm": 3.094573497772217, "learning_rate": 1.968070220402343e-05, "loss": 0.15756561756134033, "step": 4110 }, { "epoch": 0.10852708073347975, "grad_norm": 1.5563037395477295, "learning_rate": 1.967856004382345e-05, "loss": 0.17865440845489503, "step": 4120 }, { "epoch": 0.10879049597797849, "grad_norm": 4.7517852783203125, "learning_rate": 1.96764108391499e-05, "loss": 0.16742019653320311, "step": 4130 }, { "epoch": 0.10905391122247722, "grad_norm": 6.385738849639893, "learning_rate": 1.967425459156707e-05, "loss": 0.17958521842956543, "step": 4140 }, { "epoch": 0.10931732646697596, "grad_norm": 2.336958646774292, "learning_rate": 1.9672091302644363e-05, "loss": 0.17486822605133057, "step": 4150 }, { "epoch": 0.1095807417114747, "grad_norm": 5.942532062530518, "learning_rate": 1.966992097395632e-05, "loss": 0.1727718949317932, "step": 4160 }, { "epoch": 0.10984415695597344, "grad_norm": 1.612680435180664, "learning_rate": 1.9667743607082593e-05, "loss": 0.1733916997909546, "step": 4170 }, { "epoch": 0.11010757220047217, "grad_norm": 3.050499677658081, "learning_rate": 1.9665559203607965e-05, "loss": 0.16527857780456542, "step": 4180 }, { "epoch": 0.11037098744497091, "grad_norm": 1.7830233573913574, "learning_rate": 1.9663367765122346e-05, "loss": 0.17223713397979737, "step": 4190 }, { "epoch": 0.11063440268946964, "grad_norm": 1.5147689580917358, "learning_rate": 1.966116929322075e-05, "loss": 0.17958757877349854, "step": 4200 }, { "epoch": 0.11089781793396838, "grad_norm": 2.80940318107605, "learning_rate": 1.9658963789503328e-05, "loss": 0.16654845476150512, "step": 4210 }, { "epoch": 0.11116123317846711, "grad_norm": 1.4891940355300903, "learning_rate": 1.9656751255575336e-05, "loss": 0.15116056203842163, "step": 4220 }, { "epoch": 0.11142464842296586, "grad_norm": 3.428494453430176, "learning_rate": 1.965453169304715e-05, "loss": 0.17632181644439698, "step": 4230 }, { "epoch": 0.1116880636674646, "grad_norm": 4.294515132904053, "learning_rate": 1.9652305103534265e-05, "loss": 0.1780168294906616, "step": 4240 }, { "epoch": 0.11195147891196333, "grad_norm": 2.8248586654663086, "learning_rate": 1.9650071488657293e-05, "loss": 0.16454813480377198, "step": 4250 }, { "epoch": 0.11221489415646207, "grad_norm": 2.1240148544311523, "learning_rate": 1.964783085004195e-05, "loss": 0.1735195755958557, "step": 4260 }, { "epoch": 0.1124783094009608, "grad_norm": 2.040858507156372, "learning_rate": 1.9645583189319074e-05, "loss": 0.17776906490325928, "step": 4270 }, { "epoch": 0.11274172464545955, "grad_norm": 5.763367176055908, "learning_rate": 1.9643328508124605e-05, "loss": 0.18349777460098265, "step": 4280 }, { "epoch": 0.11300513988995828, "grad_norm": 3.6410651206970215, "learning_rate": 1.9641066808099597e-05, "loss": 0.1810505509376526, "step": 4290 }, { "epoch": 0.11326855513445702, "grad_norm": 1.5642269849777222, "learning_rate": 1.963879809089022e-05, "loss": 0.1587612271308899, "step": 4300 }, { "epoch": 0.11353197037895575, "grad_norm": 3.7351226806640625, "learning_rate": 1.9636522358147737e-05, "loss": 0.17723655700683594, "step": 4310 }, { "epoch": 0.1137953856234545, "grad_norm": 3.388965606689453, "learning_rate": 1.9634239611528527e-05, "loss": 0.17037689685821533, "step": 4320 }, { "epoch": 0.11405880086795323, "grad_norm": 2.234239339828491, "learning_rate": 1.9631949852694073e-05, "loss": 0.17847261428833008, "step": 4330 }, { "epoch": 0.11432221611245197, "grad_norm": 3.6162946224212646, "learning_rate": 1.9629653083310956e-05, "loss": 0.17918612957000732, "step": 4340 }, { "epoch": 0.1145856313569507, "grad_norm": 2.9755825996398926, "learning_rate": 1.9627349305050867e-05, "loss": 0.1823677659034729, "step": 4350 }, { "epoch": 0.11484904660144944, "grad_norm": 3.5854387283325195, "learning_rate": 1.9625038519590596e-05, "loss": 0.1768319845199585, "step": 4360 }, { "epoch": 0.11511246184594819, "grad_norm": 6.810959815979004, "learning_rate": 1.962272072861203e-05, "loss": 0.1851728916168213, "step": 4370 }, { "epoch": 0.11537587709044692, "grad_norm": 4.011751174926758, "learning_rate": 1.9620395933802158e-05, "loss": 0.18775969743728638, "step": 4380 }, { "epoch": 0.11563929233494566, "grad_norm": 1.8826040029525757, "learning_rate": 1.9618064136853066e-05, "loss": 0.16199989318847657, "step": 4390 }, { "epoch": 0.11590270757944439, "grad_norm": 2.267582654953003, "learning_rate": 1.9615725339461936e-05, "loss": 0.17262115478515624, "step": 4400 }, { "epoch": 0.11616612282394313, "grad_norm": 4.011380195617676, "learning_rate": 1.961337954333105e-05, "loss": 0.16350414752960205, "step": 4410 }, { "epoch": 0.11642953806844186, "grad_norm": 1.8212988376617432, "learning_rate": 1.961102675016777e-05, "loss": 0.17136261463165284, "step": 4420 }, { "epoch": 0.1166929533129406, "grad_norm": 4.152067184448242, "learning_rate": 1.960866696168457e-05, "loss": 0.15729837417602538, "step": 4430 }, { "epoch": 0.11695636855743934, "grad_norm": 4.1282758712768555, "learning_rate": 1.9606300179599004e-05, "loss": 0.16427383422851563, "step": 4440 }, { "epoch": 0.11721978380193808, "grad_norm": 3.0474400520324707, "learning_rate": 1.9603926405633712e-05, "loss": 0.1788862705230713, "step": 4450 }, { "epoch": 0.11748319904643681, "grad_norm": 2.4819204807281494, "learning_rate": 1.9601545641516434e-05, "loss": 0.17976126670837403, "step": 4460 }, { "epoch": 0.11774661429093555, "grad_norm": 3.4965476989746094, "learning_rate": 1.9599157888979997e-05, "loss": 0.17068238258361818, "step": 4470 }, { "epoch": 0.11801002953543428, "grad_norm": 2.0634891986846924, "learning_rate": 1.95967631497623e-05, "loss": 0.16580960750579835, "step": 4480 }, { "epoch": 0.11827344477993303, "grad_norm": 2.885200023651123, "learning_rate": 1.9594361425606348e-05, "loss": 0.1767939805984497, "step": 4490 }, { "epoch": 0.11853686002443177, "grad_norm": 1.74416184425354, "learning_rate": 1.959195271826021e-05, "loss": 0.16694288253784179, "step": 4500 }, { "epoch": 0.1188002752689305, "grad_norm": 3.237255096435547, "learning_rate": 1.9589537029477052e-05, "loss": 0.15850431919097902, "step": 4510 }, { "epoch": 0.11906369051342924, "grad_norm": 1.8473687171936035, "learning_rate": 1.9587114361015118e-05, "loss": 0.17620463371276857, "step": 4520 }, { "epoch": 0.11932710575792797, "grad_norm": 2.9187183380126953, "learning_rate": 1.958468471463773e-05, "loss": 0.16990748643875123, "step": 4530 }, { "epoch": 0.11959052100242672, "grad_norm": 2.6570849418640137, "learning_rate": 1.9582248092113286e-05, "loss": 0.1718679666519165, "step": 4540 }, { "epoch": 0.11985393624692545, "grad_norm": 1.6148037910461426, "learning_rate": 1.9579804495215268e-05, "loss": 0.16461594104766847, "step": 4550 }, { "epoch": 0.12011735149142419, "grad_norm": 1.3862676620483398, "learning_rate": 1.9577353925722228e-05, "loss": 0.17266640663146973, "step": 4560 }, { "epoch": 0.12038076673592292, "grad_norm": 1.6196484565734863, "learning_rate": 1.9574896385417803e-05, "loss": 0.17861151695251465, "step": 4570 }, { "epoch": 0.12064418198042166, "grad_norm": 1.8473608493804932, "learning_rate": 1.957243187609069e-05, "loss": 0.1711553454399109, "step": 4580 }, { "epoch": 0.1209075972249204, "grad_norm": 2.9736320972442627, "learning_rate": 1.9569960399534668e-05, "loss": 0.1648787498474121, "step": 4590 }, { "epoch": 0.12117101246941914, "grad_norm": 3.7043371200561523, "learning_rate": 1.9567481957548585e-05, "loss": 0.18146886825561523, "step": 4600 }, { "epoch": 0.12143442771391787, "grad_norm": 4.365318298339844, "learning_rate": 1.9564996551936354e-05, "loss": 0.1750759959220886, "step": 4610 }, { "epoch": 0.12169784295841661, "grad_norm": 3.5019760131835938, "learning_rate": 1.9562504184506966e-05, "loss": 0.16540852785110474, "step": 4620 }, { "epoch": 0.12196125820291535, "grad_norm": 3.5301899909973145, "learning_rate": 1.9560004857074472e-05, "loss": 0.16508703231811522, "step": 4630 }, { "epoch": 0.12222467344741408, "grad_norm": 2.1799139976501465, "learning_rate": 1.955749857145799e-05, "loss": 0.16656038761138917, "step": 4640 }, { "epoch": 0.12248808869191283, "grad_norm": 3.1042277812957764, "learning_rate": 1.95549853294817e-05, "loss": 0.1682344913482666, "step": 4650 }, { "epoch": 0.12275150393641156, "grad_norm": 1.2100898027420044, "learning_rate": 1.955246513297485e-05, "loss": 0.18183069229125975, "step": 4660 }, { "epoch": 0.1230149191809103, "grad_norm": 1.6057660579681396, "learning_rate": 1.9549937983771748e-05, "loss": 0.1796944856643677, "step": 4670 }, { "epoch": 0.12327833442540903, "grad_norm": 6.869030952453613, "learning_rate": 1.954740388371176e-05, "loss": 0.1605294704437256, "step": 4680 }, { "epoch": 0.12354174966990777, "grad_norm": 1.8781287670135498, "learning_rate": 1.9544862834639316e-05, "loss": 0.168251371383667, "step": 4690 }, { "epoch": 0.1238051649144065, "grad_norm": 2.168802499771118, "learning_rate": 1.9542314838403907e-05, "loss": 0.15914785861968994, "step": 4700 }, { "epoch": 0.12406858015890525, "grad_norm": 2.9190616607666016, "learning_rate": 1.9539759896860064e-05, "loss": 0.17726086378097533, "step": 4710 }, { "epoch": 0.12433199540340398, "grad_norm": 4.609450817108154, "learning_rate": 1.9537198011867383e-05, "loss": 0.1734035611152649, "step": 4720 }, { "epoch": 0.12459541064790272, "grad_norm": 3.9282195568084717, "learning_rate": 1.9534629185290524e-05, "loss": 0.17904093265533447, "step": 4730 }, { "epoch": 0.12485882589240146, "grad_norm": 2.621140480041504, "learning_rate": 1.9532053418999186e-05, "loss": 0.17410523891448976, "step": 4740 }, { "epoch": 0.1251222411369002, "grad_norm": 3.414299488067627, "learning_rate": 1.952947071486812e-05, "loss": 0.182973313331604, "step": 4750 }, { "epoch": 0.12538565638139892, "grad_norm": 3.229123592376709, "learning_rate": 1.952688107477713e-05, "loss": 0.15886776447296141, "step": 4760 }, { "epoch": 0.12564907162589767, "grad_norm": 2.0591588020324707, "learning_rate": 1.9524284500611076e-05, "loss": 0.18390076160430907, "step": 4770 }, { "epoch": 0.1259124868703964, "grad_norm": 2.1930994987487793, "learning_rate": 1.9521680994259845e-05, "loss": 0.17561812400817872, "step": 4780 }, { "epoch": 0.12617590211489516, "grad_norm": 2.0849344730377197, "learning_rate": 1.9519070557618386e-05, "loss": 0.17753183841705322, "step": 4790 }, { "epoch": 0.12643931735939387, "grad_norm": 1.7865592241287231, "learning_rate": 1.9516453192586686e-05, "loss": 0.16928551197052003, "step": 4800 }, { "epoch": 0.12670273260389261, "grad_norm": 1.8782027959823608, "learning_rate": 1.951382890106978e-05, "loss": 0.16900138854980468, "step": 4810 }, { "epoch": 0.12696614784839136, "grad_norm": 2.0131118297576904, "learning_rate": 1.951119768497774e-05, "loss": 0.1863102674484253, "step": 4820 }, { "epoch": 0.1272295630928901, "grad_norm": 1.9706189632415771, "learning_rate": 1.9508559546225672e-05, "loss": 0.163653564453125, "step": 4830 }, { "epoch": 0.12749297833738882, "grad_norm": 1.531347632408142, "learning_rate": 1.9505914486733735e-05, "loss": 0.16948320865631103, "step": 4840 }, { "epoch": 0.12775639358188756, "grad_norm": 1.6593226194381714, "learning_rate": 1.9503262508427115e-05, "loss": 0.17748838663101196, "step": 4850 }, { "epoch": 0.1280198088263863, "grad_norm": 3.103731870651245, "learning_rate": 1.9500603613236034e-05, "loss": 0.1613692045211792, "step": 4860 }, { "epoch": 0.12828322407088505, "grad_norm": 1.5908207893371582, "learning_rate": 1.9497937803095756e-05, "loss": 0.17474265098571778, "step": 4870 }, { "epoch": 0.1285466393153838, "grad_norm": 4.766870021820068, "learning_rate": 1.9495265079946565e-05, "loss": 0.17146872282028197, "step": 4880 }, { "epoch": 0.1288100545598825, "grad_norm": 2.525085687637329, "learning_rate": 1.9492585445733788e-05, "loss": 0.18177573680877684, "step": 4890 }, { "epoch": 0.12907346980438125, "grad_norm": 3.6531484127044678, "learning_rate": 1.9489898902407783e-05, "loss": 0.1770370125770569, "step": 4900 }, { "epoch": 0.12933688504888, "grad_norm": 4.435078144073486, "learning_rate": 1.9487205451923925e-05, "loss": 0.17170412540435792, "step": 4910 }, { "epoch": 0.12960030029337874, "grad_norm": 6.519932746887207, "learning_rate": 1.9484505096242628e-05, "loss": 0.1778235673904419, "step": 4920 }, { "epoch": 0.12986371553787746, "grad_norm": 5.265796184539795, "learning_rate": 1.9481797837329323e-05, "loss": 0.16493582725524902, "step": 4930 }, { "epoch": 0.1301271307823762, "grad_norm": 1.58909010887146, "learning_rate": 1.9479083677154474e-05, "loss": 0.168040132522583, "step": 4940 }, { "epoch": 0.13039054602687494, "grad_norm": 1.2800170183181763, "learning_rate": 1.9476362617693564e-05, "loss": 0.17903897762298585, "step": 4950 }, { "epoch": 0.1306539612713737, "grad_norm": 1.252787470817566, "learning_rate": 1.9473634660927092e-05, "loss": 0.18359582424163817, "step": 4960 }, { "epoch": 0.1309173765158724, "grad_norm": 5.529157638549805, "learning_rate": 1.9470899808840585e-05, "loss": 0.17962098121643066, "step": 4970 }, { "epoch": 0.13118079176037115, "grad_norm": 2.485341787338257, "learning_rate": 1.9468158063424593e-05, "loss": 0.18268868923187256, "step": 4980 }, { "epoch": 0.1314442070048699, "grad_norm": 3.7003633975982666, "learning_rate": 1.946540942667467e-05, "loss": 0.16505537033081055, "step": 4990 }, { "epoch": 0.13170762224936863, "grad_norm": 1.0040724277496338, "learning_rate": 1.9462653900591393e-05, "loss": 0.1740810513496399, "step": 5000 }, { "epoch": 0.13197103749386738, "grad_norm": 2.7724456787109375, "learning_rate": 1.9459891487180353e-05, "loss": 0.1674664258956909, "step": 5010 }, { "epoch": 0.1322344527383661, "grad_norm": 2.5798914432525635, "learning_rate": 1.9457122188452156e-05, "loss": 0.1587741494178772, "step": 5020 }, { "epoch": 0.13249786798286484, "grad_norm": 3.924309253692627, "learning_rate": 1.9454346006422414e-05, "loss": 0.166922926902771, "step": 5030 }, { "epoch": 0.13276128322736358, "grad_norm": 1.1995837688446045, "learning_rate": 1.9451562943111753e-05, "loss": 0.1827164888381958, "step": 5040 }, { "epoch": 0.13302469847186232, "grad_norm": 2.3906779289245605, "learning_rate": 1.944877300054581e-05, "loss": 0.18534679412841798, "step": 5050 }, { "epoch": 0.13328811371636104, "grad_norm": 2.211793899536133, "learning_rate": 1.944597618075522e-05, "loss": 0.16675864458084105, "step": 5060 }, { "epoch": 0.13355152896085978, "grad_norm": 2.4652037620544434, "learning_rate": 1.944317248577563e-05, "loss": 0.15974626541137696, "step": 5070 }, { "epoch": 0.13381494420535853, "grad_norm": 2.4540975093841553, "learning_rate": 1.9440361917647693e-05, "loss": 0.17763781547546387, "step": 5080 }, { "epoch": 0.13407835944985727, "grad_norm": 1.936125636100769, "learning_rate": 1.943754447841706e-05, "loss": 0.16102080345153807, "step": 5090 }, { "epoch": 0.134341774694356, "grad_norm": 3.6814067363739014, "learning_rate": 1.9434720170134387e-05, "loss": 0.16724681854248047, "step": 5100 }, { "epoch": 0.13460518993885473, "grad_norm": 2.9391884803771973, "learning_rate": 1.9431888994855324e-05, "loss": 0.18537368774414062, "step": 5110 }, { "epoch": 0.13486860518335347, "grad_norm": 3.905693769454956, "learning_rate": 1.9429050954640528e-05, "loss": 0.16798501014709472, "step": 5120 }, { "epoch": 0.13513202042785222, "grad_norm": 3.644754409790039, "learning_rate": 1.942620605155564e-05, "loss": 0.16750814914703369, "step": 5130 }, { "epoch": 0.13539543567235096, "grad_norm": 1.0424598455429077, "learning_rate": 1.9423354287671307e-05, "loss": 0.17416996955871583, "step": 5140 }, { "epoch": 0.13565885091684968, "grad_norm": 2.2378482818603516, "learning_rate": 1.9420495665063166e-05, "loss": 0.16586008071899414, "step": 5150 }, { "epoch": 0.13592226616134842, "grad_norm": 1.2759547233581543, "learning_rate": 1.9417630185811846e-05, "loss": 0.15992746353149415, "step": 5160 }, { "epoch": 0.13618568140584716, "grad_norm": 3.715003252029419, "learning_rate": 1.9414757852002962e-05, "loss": 0.1835693120956421, "step": 5170 }, { "epoch": 0.1364490966503459, "grad_norm": 2.7125706672668457, "learning_rate": 1.941187866572713e-05, "loss": 0.16812194585800172, "step": 5180 }, { "epoch": 0.13671251189484462, "grad_norm": 1.2327361106872559, "learning_rate": 1.940899262907994e-05, "loss": 0.17667936086654662, "step": 5190 }, { "epoch": 0.13697592713934337, "grad_norm": 2.2778711318969727, "learning_rate": 1.940609974416198e-05, "loss": 0.17652361392974852, "step": 5200 }, { "epoch": 0.1372393423838421, "grad_norm": 7.442572116851807, "learning_rate": 1.940320001307881e-05, "loss": 0.1699751138687134, "step": 5210 }, { "epoch": 0.13750275762834085, "grad_norm": 3.031425952911377, "learning_rate": 1.940029343794098e-05, "loss": 0.18121869564056398, "step": 5220 }, { "epoch": 0.13776617287283957, "grad_norm": 1.3786126375198364, "learning_rate": 1.9397380020864024e-05, "loss": 0.17372168302536012, "step": 5230 }, { "epoch": 0.13802958811733831, "grad_norm": 3.18849515914917, "learning_rate": 1.939445976396845e-05, "loss": 0.16586316823959352, "step": 5240 }, { "epoch": 0.13829300336183706, "grad_norm": 2.211639642715454, "learning_rate": 1.939153266937975e-05, "loss": 0.17195456027984618, "step": 5250 }, { "epoch": 0.1385564186063358, "grad_norm": 1.4709794521331787, "learning_rate": 1.938859873922839e-05, "loss": 0.17981343269348143, "step": 5260 }, { "epoch": 0.13881983385083455, "grad_norm": 2.3116302490234375, "learning_rate": 1.9385657975649807e-05, "loss": 0.17279331684112548, "step": 5270 }, { "epoch": 0.13908324909533326, "grad_norm": 2.034252643585205, "learning_rate": 1.9382710380784418e-05, "loss": 0.16595940589904784, "step": 5280 }, { "epoch": 0.139346664339832, "grad_norm": 1.5613701343536377, "learning_rate": 1.937975595677761e-05, "loss": 0.1637969970703125, "step": 5290 }, { "epoch": 0.13961007958433075, "grad_norm": 2.0771608352661133, "learning_rate": 1.9376794705779735e-05, "loss": 0.16534454822540284, "step": 5300 }, { "epoch": 0.1398734948288295, "grad_norm": 1.7306203842163086, "learning_rate": 1.937382662994613e-05, "loss": 0.1704167604446411, "step": 5310 }, { "epoch": 0.1401369100733282, "grad_norm": 2.231848955154419, "learning_rate": 1.9370851731437077e-05, "loss": 0.17299402952194215, "step": 5320 }, { "epoch": 0.14040032531782695, "grad_norm": 6.544114112854004, "learning_rate": 1.9367870012417844e-05, "loss": 0.1774529218673706, "step": 5330 }, { "epoch": 0.1406637405623257, "grad_norm": 1.1603378057479858, "learning_rate": 1.936488147505865e-05, "loss": 0.1728719711303711, "step": 5340 }, { "epoch": 0.14092715580682444, "grad_norm": 3.405526876449585, "learning_rate": 1.9361886121534686e-05, "loss": 0.16777267456054687, "step": 5350 }, { "epoch": 0.14119057105132315, "grad_norm": 2.025038719177246, "learning_rate": 1.93588839540261e-05, "loss": 0.17135350704193114, "step": 5360 }, { "epoch": 0.1414539862958219, "grad_norm": 2.36775541305542, "learning_rate": 1.9355874974717988e-05, "loss": 0.16582388877868653, "step": 5370 }, { "epoch": 0.14171740154032064, "grad_norm": 1.6991667747497559, "learning_rate": 1.9352859185800424e-05, "loss": 0.15699193477630616, "step": 5380 }, { "epoch": 0.14198081678481939, "grad_norm": 2.421307325363159, "learning_rate": 1.934983658946843e-05, "loss": 0.1724412202835083, "step": 5390 }, { "epoch": 0.14224423202931813, "grad_norm": 2.954306125640869, "learning_rate": 1.9346807187921983e-05, "loss": 0.16638995409011842, "step": 5400 }, { "epoch": 0.14250764727381685, "grad_norm": 5.191941738128662, "learning_rate": 1.9343770983366004e-05, "loss": 0.16390657424926758, "step": 5410 }, { "epoch": 0.1427710625183156, "grad_norm": 2.526332139968872, "learning_rate": 1.9340727978010377e-05, "loss": 0.17583848237991334, "step": 5420 }, { "epoch": 0.14303447776281433, "grad_norm": 2.349304437637329, "learning_rate": 1.933767817406994e-05, "loss": 0.1671211004257202, "step": 5430 }, { "epoch": 0.14329789300731308, "grad_norm": 2.9710516929626465, "learning_rate": 1.9334621573764462e-05, "loss": 0.17310160398483276, "step": 5440 }, { "epoch": 0.1435613082518118, "grad_norm": 2.69673228263855, "learning_rate": 1.9331558179318676e-05, "loss": 0.17209534645080565, "step": 5450 }, { "epoch": 0.14382472349631054, "grad_norm": 2.449807643890381, "learning_rate": 1.9328487992962247e-05, "loss": 0.17541890144348143, "step": 5460 }, { "epoch": 0.14408813874080928, "grad_norm": 2.8995983600616455, "learning_rate": 1.9325411016929793e-05, "loss": 0.17622849941253663, "step": 5470 }, { "epoch": 0.14435155398530802, "grad_norm": 1.2148582935333252, "learning_rate": 1.9322327253460867e-05, "loss": 0.17149591445922852, "step": 5480 }, { "epoch": 0.14461496922980674, "grad_norm": 1.892899751663208, "learning_rate": 1.931923670479997e-05, "loss": 0.1627707839012146, "step": 5490 }, { "epoch": 0.14487838447430548, "grad_norm": 1.1573002338409424, "learning_rate": 1.9316139373196533e-05, "loss": 0.17386655807495116, "step": 5500 }, { "epoch": 0.14514179971880423, "grad_norm": 3.00343656539917, "learning_rate": 1.9313035260904923e-05, "loss": 0.1772059440612793, "step": 5510 }, { "epoch": 0.14540521496330297, "grad_norm": 2.185471296310425, "learning_rate": 1.9309924370184455e-05, "loss": 0.17125160694122316, "step": 5520 }, { "epoch": 0.1456686302078017, "grad_norm": 4.679265975952148, "learning_rate": 1.9306806703299367e-05, "loss": 0.17756716012954712, "step": 5530 }, { "epoch": 0.14593204545230043, "grad_norm": 2.5976669788360596, "learning_rate": 1.9303682262518828e-05, "loss": 0.18183822631835939, "step": 5540 }, { "epoch": 0.14619546069679917, "grad_norm": 2.1735470294952393, "learning_rate": 1.9300551050116946e-05, "loss": 0.1719763994216919, "step": 5550 }, { "epoch": 0.14645887594129792, "grad_norm": 3.3181090354919434, "learning_rate": 1.9297413068372745e-05, "loss": 0.16088545322418213, "step": 5560 }, { "epoch": 0.14672229118579666, "grad_norm": 3.8710477352142334, "learning_rate": 1.9294268319570195e-05, "loss": 0.16260890960693358, "step": 5570 }, { "epoch": 0.14698570643029538, "grad_norm": 3.31966495513916, "learning_rate": 1.9291116805998168e-05, "loss": 0.17805628776550292, "step": 5580 }, { "epoch": 0.14724912167479412, "grad_norm": 1.2650567293167114, "learning_rate": 1.9287958529950477e-05, "loss": 0.17166742086410522, "step": 5590 }, { "epoch": 0.14751253691929286, "grad_norm": 3.4613895416259766, "learning_rate": 1.9284793493725852e-05, "loss": 0.1771928310394287, "step": 5600 }, { "epoch": 0.1477759521637916, "grad_norm": 3.836705446243286, "learning_rate": 1.9281621699627943e-05, "loss": 0.1692814350128174, "step": 5610 }, { "epoch": 0.14803936740829032, "grad_norm": 1.164014220237732, "learning_rate": 1.927844314996532e-05, "loss": 0.15804444551467894, "step": 5620 }, { "epoch": 0.14830278265278907, "grad_norm": 3.9719417095184326, "learning_rate": 1.9275257847051464e-05, "loss": 0.16703232526779174, "step": 5630 }, { "epoch": 0.1485661978972878, "grad_norm": 4.595858573913574, "learning_rate": 1.9272065793204778e-05, "loss": 0.16799925565719603, "step": 5640 }, { "epoch": 0.14882961314178655, "grad_norm": 5.105083465576172, "learning_rate": 1.9268866990748578e-05, "loss": 0.16730031967163086, "step": 5650 }, { "epoch": 0.1490930283862853, "grad_norm": 1.4815399646759033, "learning_rate": 1.9265661442011087e-05, "loss": 0.18868820667266845, "step": 5660 }, { "epoch": 0.149356443630784, "grad_norm": 6.319389820098877, "learning_rate": 1.9262449149325445e-05, "loss": 0.18458069562911988, "step": 5670 }, { "epoch": 0.14961985887528276, "grad_norm": 1.2408415079116821, "learning_rate": 1.92592301150297e-05, "loss": 0.16641920804977417, "step": 5680 }, { "epoch": 0.1498832741197815, "grad_norm": 1.8543950319290161, "learning_rate": 1.9256004341466795e-05, "loss": 0.16615359783172606, "step": 5690 }, { "epoch": 0.15014668936428024, "grad_norm": 4.299649238586426, "learning_rate": 1.9252771830984594e-05, "loss": 0.1534864902496338, "step": 5700 }, { "epoch": 0.15041010460877896, "grad_norm": 6.193323135375977, "learning_rate": 1.9249532585935855e-05, "loss": 0.17149753570556642, "step": 5710 }, { "epoch": 0.1506735198532777, "grad_norm": 6.316836357116699, "learning_rate": 1.9246286608678243e-05, "loss": 0.16778836250305176, "step": 5720 }, { "epoch": 0.15093693509777645, "grad_norm": 2.8649961948394775, "learning_rate": 1.9243033901574316e-05, "loss": 0.17136616706848146, "step": 5730 }, { "epoch": 0.1512003503422752, "grad_norm": 6.314356803894043, "learning_rate": 1.923977446699154e-05, "loss": 0.17610604763031007, "step": 5740 }, { "epoch": 0.1514637655867739, "grad_norm": 4.412478923797607, "learning_rate": 1.923650830730227e-05, "loss": 0.16842265129089357, "step": 5750 }, { "epoch": 0.15172718083127265, "grad_norm": 4.505606651306152, "learning_rate": 1.923323542488375e-05, "loss": 0.1640387535095215, "step": 5760 }, { "epoch": 0.1519905960757714, "grad_norm": 1.559616208076477, "learning_rate": 1.9229955822118138e-05, "loss": 0.1675552487373352, "step": 5770 }, { "epoch": 0.15225401132027014, "grad_norm": 2.9577205181121826, "learning_rate": 1.9226669501392465e-05, "loss": 0.1699594497680664, "step": 5780 }, { "epoch": 0.15251742656476888, "grad_norm": 1.892250418663025, "learning_rate": 1.9223376465098657e-05, "loss": 0.16279611587524415, "step": 5790 }, { "epoch": 0.1527808418092676, "grad_norm": 8.280914306640625, "learning_rate": 1.9220076715633526e-05, "loss": 0.16990970373153685, "step": 5800 }, { "epoch": 0.15304425705376634, "grad_norm": 3.648590326309204, "learning_rate": 1.921677025539877e-05, "loss": 0.16657552719116211, "step": 5810 }, { "epoch": 0.15330767229826509, "grad_norm": 1.5436463356018066, "learning_rate": 1.921345708680098e-05, "loss": 0.1608739137649536, "step": 5820 }, { "epoch": 0.15357108754276383, "grad_norm": 2.3433873653411865, "learning_rate": 1.921013721225162e-05, "loss": 0.1710874319076538, "step": 5830 }, { "epoch": 0.15383450278726254, "grad_norm": 4.17683744430542, "learning_rate": 1.9206810634167036e-05, "loss": 0.1545642137527466, "step": 5840 }, { "epoch": 0.1540979180317613, "grad_norm": 1.6678367853164673, "learning_rate": 1.920347735496846e-05, "loss": 0.1622430443763733, "step": 5850 }, { "epoch": 0.15436133327626003, "grad_norm": 1.5314058065414429, "learning_rate": 1.9200137377081993e-05, "loss": 0.16901836395263672, "step": 5860 }, { "epoch": 0.15462474852075878, "grad_norm": 1.4772472381591797, "learning_rate": 1.919679070293861e-05, "loss": 0.15832581520080566, "step": 5870 }, { "epoch": 0.1548881637652575, "grad_norm": 3.7396438121795654, "learning_rate": 1.9193437334974178e-05, "loss": 0.15812952518463136, "step": 5880 }, { "epoch": 0.15515157900975624, "grad_norm": 5.364645004272461, "learning_rate": 1.9190077275629408e-05, "loss": 0.16294536590576172, "step": 5890 }, { "epoch": 0.15541499425425498, "grad_norm": 1.226230502128601, "learning_rate": 1.9186710527349907e-05, "loss": 0.16501951217651367, "step": 5900 }, { "epoch": 0.15567840949875372, "grad_norm": 5.60057258605957, "learning_rate": 1.9183337092586137e-05, "loss": 0.17108590602874757, "step": 5910 }, { "epoch": 0.15594182474325247, "grad_norm": 1.3026608228683472, "learning_rate": 1.9179956973793422e-05, "loss": 0.16908713579177856, "step": 5920 }, { "epoch": 0.15620523998775118, "grad_norm": 1.3746027946472168, "learning_rate": 1.917657017343197e-05, "loss": 0.16499861478805541, "step": 5930 }, { "epoch": 0.15646865523224993, "grad_norm": 2.173626661300659, "learning_rate": 1.9173176693966835e-05, "loss": 0.17672333717346192, "step": 5940 }, { "epoch": 0.15673207047674867, "grad_norm": 3.447535514831543, "learning_rate": 1.9169776537867936e-05, "loss": 0.1734776258468628, "step": 5950 }, { "epoch": 0.1569954857212474, "grad_norm": 2.8676111698150635, "learning_rate": 1.9166369707610055e-05, "loss": 0.16631195545196534, "step": 5960 }, { "epoch": 0.15725890096574613, "grad_norm": 1.6580955982208252, "learning_rate": 1.9162956205672833e-05, "loss": 0.17541799545288086, "step": 5970 }, { "epoch": 0.15752231621024487, "grad_norm": 0.9106699824333191, "learning_rate": 1.9159536034540757e-05, "loss": 0.16560558080673218, "step": 5980 }, { "epoch": 0.15778573145474362, "grad_norm": 2.641714572906494, "learning_rate": 1.9156109196703177e-05, "loss": 0.1706124424934387, "step": 5990 }, { "epoch": 0.15804914669924236, "grad_norm": 1.8202080726623535, "learning_rate": 1.91526756946543e-05, "loss": 0.17435910701751708, "step": 6000 }, { "epoch": 0.15831256194374108, "grad_norm": 6.195431709289551, "learning_rate": 1.9149235530893172e-05, "loss": 0.17253327369689941, "step": 6010 }, { "epoch": 0.15857597718823982, "grad_norm": 2.8922672271728516, "learning_rate": 1.9145788707923692e-05, "loss": 0.18391237258911133, "step": 6020 }, { "epoch": 0.15883939243273856, "grad_norm": 2.248342514038086, "learning_rate": 1.914233522825461e-05, "loss": 0.16499848365783693, "step": 6030 }, { "epoch": 0.1591028076772373, "grad_norm": 3.3761746883392334, "learning_rate": 1.913887509439951e-05, "loss": 0.1674929976463318, "step": 6040 }, { "epoch": 0.15936622292173605, "grad_norm": 4.503216743469238, "learning_rate": 1.9135408308876834e-05, "loss": 0.1715470314025879, "step": 6050 }, { "epoch": 0.15962963816623477, "grad_norm": 2.400670289993286, "learning_rate": 1.913193487420986e-05, "loss": 0.1604616641998291, "step": 6060 }, { "epoch": 0.1598930534107335, "grad_norm": 2.343841314315796, "learning_rate": 1.9128454792926694e-05, "loss": 0.16970880031585694, "step": 6070 }, { "epoch": 0.16015646865523225, "grad_norm": 4.445486545562744, "learning_rate": 1.9124968067560296e-05, "loss": 0.1711619257926941, "step": 6080 }, { "epoch": 0.160419883899731, "grad_norm": 4.289151191711426, "learning_rate": 1.9121474700648454e-05, "loss": 0.16968673467636108, "step": 6090 }, { "epoch": 0.1606832991442297, "grad_norm": 3.1308329105377197, "learning_rate": 1.9117974694733795e-05, "loss": 0.17475130558013915, "step": 6100 }, { "epoch": 0.16094671438872846, "grad_norm": 2.693610191345215, "learning_rate": 1.9114468052363764e-05, "loss": 0.17302005290985106, "step": 6110 }, { "epoch": 0.1612101296332272, "grad_norm": 2.294874906539917, "learning_rate": 1.9110954776090656e-05, "loss": 0.1739039659500122, "step": 6120 }, { "epoch": 0.16147354487772594, "grad_norm": 3.3213512897491455, "learning_rate": 1.910743486847158e-05, "loss": 0.1628331184387207, "step": 6130 }, { "epoch": 0.16173696012222466, "grad_norm": 1.830768346786499, "learning_rate": 1.9103908332068483e-05, "loss": 0.1778757691383362, "step": 6140 }, { "epoch": 0.1620003753667234, "grad_norm": 4.066507816314697, "learning_rate": 1.910037516944812e-05, "loss": 0.15867701768875123, "step": 6150 }, { "epoch": 0.16226379061122215, "grad_norm": 2.5460917949676514, "learning_rate": 1.909683538318209e-05, "loss": 0.1665799140930176, "step": 6160 }, { "epoch": 0.1625272058557209, "grad_norm": 3.157618999481201, "learning_rate": 1.9093288975846797e-05, "loss": 0.16264684200286866, "step": 6170 }, { "epoch": 0.16279062110021963, "grad_norm": 2.272037982940674, "learning_rate": 1.9089735950023463e-05, "loss": 0.18245644569396974, "step": 6180 }, { "epoch": 0.16305403634471835, "grad_norm": 1.7914032936096191, "learning_rate": 1.9086176308298147e-05, "loss": 0.1706324338912964, "step": 6190 }, { "epoch": 0.1633174515892171, "grad_norm": 8.686756134033203, "learning_rate": 1.90826100532617e-05, "loss": 0.18479440212249756, "step": 6200 }, { "epoch": 0.16358086683371584, "grad_norm": 3.328874349594116, "learning_rate": 1.90790371875098e-05, "loss": 0.17243902683258056, "step": 6210 }, { "epoch": 0.16384428207821458, "grad_norm": 1.350071907043457, "learning_rate": 1.9075457713642934e-05, "loss": 0.1761566162109375, "step": 6220 }, { "epoch": 0.1641076973227133, "grad_norm": 1.8652448654174805, "learning_rate": 1.9071871634266392e-05, "loss": 0.16634671688079833, "step": 6230 }, { "epoch": 0.16437111256721204, "grad_norm": 1.912419319152832, "learning_rate": 1.9068278951990282e-05, "loss": 0.15880290269851685, "step": 6240 }, { "epoch": 0.16463452781171078, "grad_norm": 5.499303340911865, "learning_rate": 1.9064679669429515e-05, "loss": 0.17226836681365967, "step": 6250 }, { "epoch": 0.16489794305620953, "grad_norm": 2.3715922832489014, "learning_rate": 1.9061073789203798e-05, "loss": 0.16967489719390869, "step": 6260 }, { "epoch": 0.16516135830070824, "grad_norm": 0.9351702928543091, "learning_rate": 1.9057461313937646e-05, "loss": 0.1798255681991577, "step": 6270 }, { "epoch": 0.165424773545207, "grad_norm": 1.8363850116729736, "learning_rate": 1.905384224626038e-05, "loss": 0.18178093433380127, "step": 6280 }, { "epoch": 0.16568818878970573, "grad_norm": 3.7483270168304443, "learning_rate": 1.9050216588806113e-05, "loss": 0.17316328287124633, "step": 6290 }, { "epoch": 0.16595160403420448, "grad_norm": 4.045187950134277, "learning_rate": 1.9046584344213754e-05, "loss": 0.16284570693969727, "step": 6300 }, { "epoch": 0.16621501927870322, "grad_norm": 2.8518712520599365, "learning_rate": 1.9042945515127004e-05, "loss": 0.16173110008239747, "step": 6310 }, { "epoch": 0.16647843452320193, "grad_norm": 1.9285919666290283, "learning_rate": 1.9039300104194366e-05, "loss": 0.15872421264648437, "step": 6320 }, { "epoch": 0.16674184976770068, "grad_norm": 2.8352231979370117, "learning_rate": 1.903564811406912e-05, "loss": 0.1724141240119934, "step": 6330 }, { "epoch": 0.16700526501219942, "grad_norm": 2.6719415187835693, "learning_rate": 1.9031989547409347e-05, "loss": 0.157776403427124, "step": 6340 }, { "epoch": 0.16726868025669817, "grad_norm": 1.2758421897888184, "learning_rate": 1.9028324406877905e-05, "loss": 0.15539517402648925, "step": 6350 }, { "epoch": 0.16753209550119688, "grad_norm": 1.7081279754638672, "learning_rate": 1.9024652695142447e-05, "loss": 0.15525586605072023, "step": 6360 }, { "epoch": 0.16779551074569563, "grad_norm": 2.116184711456299, "learning_rate": 1.90209744148754e-05, "loss": 0.16932883262634277, "step": 6370 }, { "epoch": 0.16805892599019437, "grad_norm": 4.640033721923828, "learning_rate": 1.901728956875398e-05, "loss": 0.1654934763908386, "step": 6380 }, { "epoch": 0.1683223412346931, "grad_norm": 1.4270668029785156, "learning_rate": 1.9013598159460168e-05, "loss": 0.15640132427215575, "step": 6390 }, { "epoch": 0.16858575647919183, "grad_norm": 5.337968349456787, "learning_rate": 1.900990018968074e-05, "loss": 0.16485788822174072, "step": 6400 }, { "epoch": 0.16884917172369057, "grad_norm": 1.7192708253860474, "learning_rate": 1.900619566210723e-05, "loss": 0.15020540952682496, "step": 6410 }, { "epoch": 0.16911258696818932, "grad_norm": 1.334838628768921, "learning_rate": 1.900248457943596e-05, "loss": 0.1583791971206665, "step": 6420 }, { "epoch": 0.16937600221268806, "grad_norm": 4.092357158660889, "learning_rate": 1.899876694436801e-05, "loss": 0.16993744373321534, "step": 6430 }, { "epoch": 0.1696394174571868, "grad_norm": 1.8147633075714111, "learning_rate": 1.899504275960924e-05, "loss": 0.16153521537780763, "step": 6440 }, { "epoch": 0.16990283270168552, "grad_norm": 4.099387168884277, "learning_rate": 1.899131202787027e-05, "loss": 0.17236267328262328, "step": 6450 }, { "epoch": 0.17016624794618426, "grad_norm": 1.4710395336151123, "learning_rate": 1.8987574751866488e-05, "loss": 0.18834834098815917, "step": 6460 }, { "epoch": 0.170429663190683, "grad_norm": 5.762836456298828, "learning_rate": 1.8983830934318044e-05, "loss": 0.1798737645149231, "step": 6470 }, { "epoch": 0.17069307843518175, "grad_norm": 4.204227447509766, "learning_rate": 1.8980080577949848e-05, "loss": 0.17450743913650513, "step": 6480 }, { "epoch": 0.17095649367968047, "grad_norm": 4.1486663818359375, "learning_rate": 1.8976323685491577e-05, "loss": 0.17158024311065673, "step": 6490 }, { "epoch": 0.1712199089241792, "grad_norm": 3.2643206119537354, "learning_rate": 1.8972560259677658e-05, "loss": 0.17105766534805297, "step": 6500 }, { "epoch": 0.17148332416867795, "grad_norm": 6.278489589691162, "learning_rate": 1.896879030324727e-05, "loss": 0.16140648126602172, "step": 6510 }, { "epoch": 0.1717467394131767, "grad_norm": 1.705124855041504, "learning_rate": 1.8965013818944357e-05, "loss": 0.17306268215179443, "step": 6520 }, { "epoch": 0.1720101546576754, "grad_norm": 1.801294207572937, "learning_rate": 1.89612308095176e-05, "loss": 0.17171010971069336, "step": 6530 }, { "epoch": 0.17227356990217416, "grad_norm": 2.1223320960998535, "learning_rate": 1.8957441277720438e-05, "loss": 0.17910101413726806, "step": 6540 }, { "epoch": 0.1725369851466729, "grad_norm": 1.5529301166534424, "learning_rate": 1.8953645226311056e-05, "loss": 0.15945274829864503, "step": 6550 }, { "epoch": 0.17280040039117164, "grad_norm": 2.069559335708618, "learning_rate": 1.894984265805239e-05, "loss": 0.17833086252212524, "step": 6560 }, { "epoch": 0.1730638156356704, "grad_norm": 1.996840000152588, "learning_rate": 1.8946033575712104e-05, "loss": 0.15828256607055663, "step": 6570 }, { "epoch": 0.1733272308801691, "grad_norm": 1.6921437978744507, "learning_rate": 1.894221798206262e-05, "loss": 0.16661245822906495, "step": 6580 }, { "epoch": 0.17359064612466785, "grad_norm": 3.7046592235565186, "learning_rate": 1.8938395879881086e-05, "loss": 0.17444067001342772, "step": 6590 }, { "epoch": 0.1738540613691666, "grad_norm": 2.0411226749420166, "learning_rate": 1.8934567271949397e-05, "loss": 0.18265844583511354, "step": 6600 }, { "epoch": 0.17411747661366533, "grad_norm": 1.9445009231567383, "learning_rate": 1.8930732161054176e-05, "loss": 0.16466751098632812, "step": 6610 }, { "epoch": 0.17438089185816405, "grad_norm": 4.3247857093811035, "learning_rate": 1.8926890549986786e-05, "loss": 0.16920498609542847, "step": 6620 }, { "epoch": 0.1746443071026628, "grad_norm": 3.10579514503479, "learning_rate": 1.8923042441543313e-05, "loss": 0.16620644330978393, "step": 6630 }, { "epoch": 0.17490772234716154, "grad_norm": 5.572936534881592, "learning_rate": 1.891918783852458e-05, "loss": 0.1724491834640503, "step": 6640 }, { "epoch": 0.17517113759166028, "grad_norm": 2.422200918197632, "learning_rate": 1.8915326743736134e-05, "loss": 0.15038318634033204, "step": 6650 }, { "epoch": 0.175434552836159, "grad_norm": 5.0948357582092285, "learning_rate": 1.8911459159988238e-05, "loss": 0.17589011192321777, "step": 6660 }, { "epoch": 0.17569796808065774, "grad_norm": 1.5022882223129272, "learning_rate": 1.8907585090095897e-05, "loss": 0.17167057991027831, "step": 6670 }, { "epoch": 0.17596138332515648, "grad_norm": 4.588441848754883, "learning_rate": 1.890370453687882e-05, "loss": 0.17737482786178588, "step": 6680 }, { "epoch": 0.17622479856965523, "grad_norm": 2.9673049449920654, "learning_rate": 1.889981750316144e-05, "loss": 0.17688963413238526, "step": 6690 }, { "epoch": 0.17648821381415397, "grad_norm": 2.034874200820923, "learning_rate": 1.8895923991772913e-05, "loss": 0.17158868312835693, "step": 6700 }, { "epoch": 0.1767516290586527, "grad_norm": 2.105823278427124, "learning_rate": 1.8892024005547104e-05, "loss": 0.18040387630462645, "step": 6710 }, { "epoch": 0.17701504430315143, "grad_norm": 5.721667766571045, "learning_rate": 1.8888117547322586e-05, "loss": 0.16441752910614013, "step": 6720 }, { "epoch": 0.17727845954765017, "grad_norm": 2.05898380279541, "learning_rate": 1.888420461994265e-05, "loss": 0.16779735088348388, "step": 6730 }, { "epoch": 0.17754187479214892, "grad_norm": 3.8968136310577393, "learning_rate": 1.88802852262553e-05, "loss": 0.163329553604126, "step": 6740 }, { "epoch": 0.17780529003664763, "grad_norm": 4.253916263580322, "learning_rate": 1.887635936911323e-05, "loss": 0.18034402132034302, "step": 6750 }, { "epoch": 0.17806870528114638, "grad_norm": 1.7128275632858276, "learning_rate": 1.8872427051373852e-05, "loss": 0.1708505392074585, "step": 6760 }, { "epoch": 0.17833212052564512, "grad_norm": 2.3046090602874756, "learning_rate": 1.886848827589928e-05, "loss": 0.16632399559020997, "step": 6770 }, { "epoch": 0.17859553577014387, "grad_norm": 3.082343578338623, "learning_rate": 1.8864543045556318e-05, "loss": 0.17454488277435304, "step": 6780 }, { "epoch": 0.1788589510146426, "grad_norm": 6.28101110458374, "learning_rate": 1.8860591363216482e-05, "loss": 0.17623682022094728, "step": 6790 }, { "epoch": 0.17912236625914132, "grad_norm": 2.2027626037597656, "learning_rate": 1.8856633231755972e-05, "loss": 0.17497538328170775, "step": 6800 }, { "epoch": 0.17938578150364007, "grad_norm": 2.0437092781066895, "learning_rate": 1.8852668654055686e-05, "loss": 0.1660524845123291, "step": 6810 }, { "epoch": 0.1796491967481388, "grad_norm": 3.140660524368286, "learning_rate": 1.884869763300122e-05, "loss": 0.17055776119232177, "step": 6820 }, { "epoch": 0.17991261199263756, "grad_norm": 2.134204149246216, "learning_rate": 1.8844720171482854e-05, "loss": 0.16167261600494384, "step": 6830 }, { "epoch": 0.18017602723713627, "grad_norm": 2.426995277404785, "learning_rate": 1.884073627239555e-05, "loss": 0.169888174533844, "step": 6840 }, { "epoch": 0.18043944248163502, "grad_norm": 4.118577003479004, "learning_rate": 1.8836745938638963e-05, "loss": 0.17239440679550172, "step": 6850 }, { "epoch": 0.18070285772613376, "grad_norm": 1.6490929126739502, "learning_rate": 1.8832749173117435e-05, "loss": 0.17026309967041015, "step": 6860 }, { "epoch": 0.1809662729706325, "grad_norm": 3.1386053562164307, "learning_rate": 1.8828745978739982e-05, "loss": 0.1703119993209839, "step": 6870 }, { "epoch": 0.18122968821513122, "grad_norm": 0.8857545256614685, "learning_rate": 1.88247363584203e-05, "loss": 0.16679558753967286, "step": 6880 }, { "epoch": 0.18149310345962996, "grad_norm": 3.765375852584839, "learning_rate": 1.8820720315076767e-05, "loss": 0.16913124322891235, "step": 6890 }, { "epoch": 0.1817565187041287, "grad_norm": 2.190749168395996, "learning_rate": 1.8816697851632425e-05, "loss": 0.17666095495224, "step": 6900 }, { "epoch": 0.18201993394862745, "grad_norm": 2.5285422801971436, "learning_rate": 1.8812668971015006e-05, "loss": 0.16909213066101075, "step": 6910 }, { "epoch": 0.1822833491931262, "grad_norm": 1.4958933591842651, "learning_rate": 1.88086336761569e-05, "loss": 0.1671827554702759, "step": 6920 }, { "epoch": 0.1825467644376249, "grad_norm": 1.327950119972229, "learning_rate": 1.8804591969995167e-05, "loss": 0.17159262895584107, "step": 6930 }, { "epoch": 0.18281017968212365, "grad_norm": 2.856576442718506, "learning_rate": 1.8800543855471535e-05, "loss": 0.17030508518218995, "step": 6940 }, { "epoch": 0.1830735949266224, "grad_norm": 2.3181302547454834, "learning_rate": 1.87964893355324e-05, "loss": 0.1628416061401367, "step": 6950 }, { "epoch": 0.18333701017112114, "grad_norm": 3.3454272747039795, "learning_rate": 1.879242841312881e-05, "loss": 0.16824691295623778, "step": 6960 }, { "epoch": 0.18360042541561986, "grad_norm": 5.219283580780029, "learning_rate": 1.878836109121648e-05, "loss": 0.16939153671264648, "step": 6970 }, { "epoch": 0.1838638406601186, "grad_norm": 2.2742817401885986, "learning_rate": 1.8784287372755795e-05, "loss": 0.18172891139984132, "step": 6980 }, { "epoch": 0.18412725590461734, "grad_norm": 5.322461128234863, "learning_rate": 1.8780207260711767e-05, "loss": 0.17077877521514892, "step": 6990 }, { "epoch": 0.1843906711491161, "grad_norm": 2.0238912105560303, "learning_rate": 1.8776120758054087e-05, "loss": 0.17099454402923583, "step": 7000 }, { "epoch": 0.1846540863936148, "grad_norm": 3.1083180904388428, "learning_rate": 1.8772027867757082e-05, "loss": 0.16998847723007202, "step": 7010 }, { "epoch": 0.18491750163811355, "grad_norm": 4.520308971405029, "learning_rate": 1.8767928592799742e-05, "loss": 0.16870694160461425, "step": 7020 }, { "epoch": 0.1851809168826123, "grad_norm": 3.872560739517212, "learning_rate": 1.876382293616569e-05, "loss": 0.1816251516342163, "step": 7030 }, { "epoch": 0.18544433212711103, "grad_norm": 2.407565116882324, "learning_rate": 1.8759710900843202e-05, "loss": 0.16687875986099243, "step": 7040 }, { "epoch": 0.18570774737160978, "grad_norm": 1.815317153930664, "learning_rate": 1.8755592489825193e-05, "loss": 0.15534011125564576, "step": 7050 }, { "epoch": 0.1859711626161085, "grad_norm": 1.8710746765136719, "learning_rate": 1.8751467706109224e-05, "loss": 0.1791010856628418, "step": 7060 }, { "epoch": 0.18623457786060724, "grad_norm": 4.819329738616943, "learning_rate": 1.8747336552697483e-05, "loss": 0.16611595153808595, "step": 7070 }, { "epoch": 0.18649799310510598, "grad_norm": 1.4968774318695068, "learning_rate": 1.874319903259681e-05, "loss": 0.1628337860107422, "step": 7080 }, { "epoch": 0.18676140834960472, "grad_norm": 1.4801664352416992, "learning_rate": 1.873905514881867e-05, "loss": 0.18200368881225587, "step": 7090 }, { "epoch": 0.18702482359410344, "grad_norm": 4.7115678787231445, "learning_rate": 1.873490490437915e-05, "loss": 0.16423263549804687, "step": 7100 }, { "epoch": 0.18728823883860218, "grad_norm": 5.169195175170898, "learning_rate": 1.8730748302298993e-05, "loss": 0.1709810972213745, "step": 7110 }, { "epoch": 0.18755165408310093, "grad_norm": 2.5847325325012207, "learning_rate": 1.8726585345603544e-05, "loss": 0.19629859924316406, "step": 7120 }, { "epoch": 0.18781506932759967, "grad_norm": 1.2578965425491333, "learning_rate": 1.8722416037322778e-05, "loss": 0.16987218856811523, "step": 7130 }, { "epoch": 0.1880784845720984, "grad_norm": 2.509981155395508, "learning_rate": 1.871824038049131e-05, "loss": 0.18485674858093262, "step": 7140 }, { "epoch": 0.18834189981659713, "grad_norm": 2.8872334957122803, "learning_rate": 1.871405837814835e-05, "loss": 0.16677299737930298, "step": 7150 }, { "epoch": 0.18860531506109587, "grad_norm": 1.9060386419296265, "learning_rate": 1.8709870033337753e-05, "loss": 0.1820969581604004, "step": 7160 }, { "epoch": 0.18886873030559462, "grad_norm": 4.710076332092285, "learning_rate": 1.8705675349107964e-05, "loss": 0.16595271825790406, "step": 7170 }, { "epoch": 0.18913214555009336, "grad_norm": 3.39457106590271, "learning_rate": 1.870147432851206e-05, "loss": 0.1682062864303589, "step": 7180 }, { "epoch": 0.18939556079459208, "grad_norm": 3.764185905456543, "learning_rate": 1.8697266974607732e-05, "loss": 0.1759105443954468, "step": 7190 }, { "epoch": 0.18965897603909082, "grad_norm": 2.6150403022766113, "learning_rate": 1.8693053290457264e-05, "loss": 0.16495842933654786, "step": 7200 }, { "epoch": 0.18992239128358956, "grad_norm": 3.2689499855041504, "learning_rate": 1.868883327912756e-05, "loss": 0.17251185178756714, "step": 7210 }, { "epoch": 0.1901858065280883, "grad_norm": 1.294586181640625, "learning_rate": 1.868460694369013e-05, "loss": 0.17698278427124023, "step": 7220 }, { "epoch": 0.19044922177258702, "grad_norm": 1.808193325996399, "learning_rate": 1.868037428722108e-05, "loss": 0.16999614238739014, "step": 7230 }, { "epoch": 0.19071263701708577, "grad_norm": 2.846940040588379, "learning_rate": 1.867613531280112e-05, "loss": 0.16914936304092407, "step": 7240 }, { "epoch": 0.1909760522615845, "grad_norm": 2.1703858375549316, "learning_rate": 1.867189002351556e-05, "loss": 0.16598134040832518, "step": 7250 }, { "epoch": 0.19123946750608325, "grad_norm": 4.924707889556885, "learning_rate": 1.8667638422454306e-05, "loss": 0.17067334651947022, "step": 7260 }, { "epoch": 0.19150288275058197, "grad_norm": 2.7115631103515625, "learning_rate": 1.8663380512711852e-05, "loss": 0.17158885002136232, "step": 7270 }, { "epoch": 0.19176629799508071, "grad_norm": 2.2688255310058594, "learning_rate": 1.865911629738729e-05, "loss": 0.166017746925354, "step": 7280 }, { "epoch": 0.19202971323957946, "grad_norm": 2.6186532974243164, "learning_rate": 1.86548457795843e-05, "loss": 0.17738898992538452, "step": 7290 }, { "epoch": 0.1922931284840782, "grad_norm": 1.6590479612350464, "learning_rate": 1.8650568962411154e-05, "loss": 0.1706554651260376, "step": 7300 }, { "epoch": 0.19255654372857695, "grad_norm": 3.065463066101074, "learning_rate": 1.8646285848980693e-05, "loss": 0.1604873299598694, "step": 7310 }, { "epoch": 0.19281995897307566, "grad_norm": 7.578081130981445, "learning_rate": 1.8641996442410365e-05, "loss": 0.15506317615509033, "step": 7320 }, { "epoch": 0.1930833742175744, "grad_norm": 1.213533878326416, "learning_rate": 1.8637700745822176e-05, "loss": 0.17450158596038817, "step": 7330 }, { "epoch": 0.19334678946207315, "grad_norm": 0.7251972556114197, "learning_rate": 1.8633398762342722e-05, "loss": 0.16905813217163085, "step": 7340 }, { "epoch": 0.1936102047065719, "grad_norm": 3.2628347873687744, "learning_rate": 1.8629090495103175e-05, "loss": 0.16542177200317382, "step": 7350 }, { "epoch": 0.1938736199510706, "grad_norm": 2.7949180603027344, "learning_rate": 1.8624775947239272e-05, "loss": 0.14994461536407472, "step": 7360 }, { "epoch": 0.19413703519556935, "grad_norm": 3.2019636631011963, "learning_rate": 1.862045512189133e-05, "loss": 0.16356490850448607, "step": 7370 }, { "epoch": 0.1944004504400681, "grad_norm": 3.0409035682678223, "learning_rate": 1.8616128022204233e-05, "loss": 0.17048518657684325, "step": 7380 }, { "epoch": 0.19466386568456684, "grad_norm": 1.6410999298095703, "learning_rate": 1.8611794651327432e-05, "loss": 0.17243160009384156, "step": 7390 }, { "epoch": 0.19492728092906556, "grad_norm": 1.6028625965118408, "learning_rate": 1.8607455012414936e-05, "loss": 0.1588508129119873, "step": 7400 }, { "epoch": 0.1951906961735643, "grad_norm": 1.2579373121261597, "learning_rate": 1.8603109108625327e-05, "loss": 0.1656240701675415, "step": 7410 }, { "epoch": 0.19545411141806304, "grad_norm": 3.154045581817627, "learning_rate": 1.859875694312174e-05, "loss": 0.15902515649795532, "step": 7420 }, { "epoch": 0.19571752666256179, "grad_norm": 1.1880967617034912, "learning_rate": 1.8594398519071867e-05, "loss": 0.17755728960037231, "step": 7430 }, { "epoch": 0.19598094190706053, "grad_norm": 1.4015402793884277, "learning_rate": 1.8590033839647955e-05, "loss": 0.17489659786224365, "step": 7440 }, { "epoch": 0.19624435715155925, "grad_norm": 1.3492811918258667, "learning_rate": 1.8585662908026814e-05, "loss": 0.1613961338996887, "step": 7450 }, { "epoch": 0.196507772396058, "grad_norm": 1.2078641653060913, "learning_rate": 1.8581285727389787e-05, "loss": 0.166686749458313, "step": 7460 }, { "epoch": 0.19677118764055673, "grad_norm": 5.156308650970459, "learning_rate": 1.857690230092278e-05, "loss": 0.18014998435974122, "step": 7470 }, { "epoch": 0.19703460288505548, "grad_norm": 2.1922101974487305, "learning_rate": 1.8572512631816243e-05, "loss": 0.16732826232910156, "step": 7480 }, { "epoch": 0.1972980181295542, "grad_norm": 1.5574091672897339, "learning_rate": 1.856811672326516e-05, "loss": 0.16051845550537108, "step": 7490 }, { "epoch": 0.19756143337405294, "grad_norm": 2.1057159900665283, "learning_rate": 1.8563714578469064e-05, "loss": 0.1761303424835205, "step": 7500 }, { "epoch": 0.19782484861855168, "grad_norm": 4.13985013961792, "learning_rate": 1.8559306200632036e-05, "loss": 0.17418509721755981, "step": 7510 }, { "epoch": 0.19808826386305042, "grad_norm": 1.7249618768692017, "learning_rate": 1.8554891592962666e-05, "loss": 0.15851471424102784, "step": 7520 }, { "epoch": 0.19835167910754914, "grad_norm": 2.1624081134796143, "learning_rate": 1.8550470758674112e-05, "loss": 0.1663556456565857, "step": 7530 }, { "epoch": 0.19861509435204788, "grad_norm": 1.553000807762146, "learning_rate": 1.8546043700984036e-05, "loss": 0.16653939485549926, "step": 7540 }, { "epoch": 0.19887850959654663, "grad_norm": 1.9871368408203125, "learning_rate": 1.8541610423114647e-05, "loss": 0.1730506181716919, "step": 7550 }, { "epoch": 0.19914192484104537, "grad_norm": 2.3889684677124023, "learning_rate": 1.853717092829268e-05, "loss": 0.1733371138572693, "step": 7560 }, { "epoch": 0.1994053400855441, "grad_norm": 1.2184563875198364, "learning_rate": 1.853272521974938e-05, "loss": 0.17358769178390504, "step": 7570 }, { "epoch": 0.19966875533004283, "grad_norm": 1.47380793094635, "learning_rate": 1.8528273300720538e-05, "loss": 0.1694783926010132, "step": 7580 }, { "epoch": 0.19993217057454157, "grad_norm": 1.4913792610168457, "learning_rate": 1.8523815174446444e-05, "loss": 0.1675785779953003, "step": 7590 }, { "epoch": 0.20019558581904032, "grad_norm": 1.1210178136825562, "learning_rate": 1.851935084417192e-05, "loss": 0.16176989078521728, "step": 7600 }, { "epoch": 0.20045900106353906, "grad_norm": 1.7860515117645264, "learning_rate": 1.8514880313146295e-05, "loss": 0.18065106868743896, "step": 7610 }, { "epoch": 0.20072241630803778, "grad_norm": 2.1595287322998047, "learning_rate": 1.8510403584623414e-05, "loss": 0.17859983444213867, "step": 7620 }, { "epoch": 0.20098583155253652, "grad_norm": 6.141869068145752, "learning_rate": 1.850592066186163e-05, "loss": 0.16513538360595703, "step": 7630 }, { "epoch": 0.20124924679703526, "grad_norm": 2.916069746017456, "learning_rate": 1.850143154812382e-05, "loss": 0.16009286642074586, "step": 7640 }, { "epoch": 0.201512662041534, "grad_norm": 1.8909670114517212, "learning_rate": 1.849693624667734e-05, "loss": 0.1623300552368164, "step": 7650 }, { "epoch": 0.20177607728603272, "grad_norm": 1.7283434867858887, "learning_rate": 1.8492434760794072e-05, "loss": 0.17345850467681884, "step": 7660 }, { "epoch": 0.20203949253053147, "grad_norm": 1.906062364578247, "learning_rate": 1.848792709375039e-05, "loss": 0.16811028718948365, "step": 7670 }, { "epoch": 0.2023029077750302, "grad_norm": 1.2785416841506958, "learning_rate": 1.848341324882717e-05, "loss": 0.15248665809631348, "step": 7680 }, { "epoch": 0.20256632301952895, "grad_norm": 0.8517978191375732, "learning_rate": 1.8478893229309784e-05, "loss": 0.17577085494995118, "step": 7690 }, { "epoch": 0.2028297382640277, "grad_norm": 1.9601638317108154, "learning_rate": 1.847436703848809e-05, "loss": 0.1745567202568054, "step": 7700 }, { "epoch": 0.20309315350852641, "grad_norm": 3.59157133102417, "learning_rate": 1.8469834679656452e-05, "loss": 0.17949577569961547, "step": 7710 }, { "epoch": 0.20335656875302516, "grad_norm": 2.1916379928588867, "learning_rate": 1.8465296156113715e-05, "loss": 0.16837507486343384, "step": 7720 }, { "epoch": 0.2036199839975239, "grad_norm": 2.859231472015381, "learning_rate": 1.8460751471163212e-05, "loss": 0.18054780960083008, "step": 7730 }, { "epoch": 0.20388339924202264, "grad_norm": 2.502269744873047, "learning_rate": 1.845620062811276e-05, "loss": 0.17746856212615966, "step": 7740 }, { "epoch": 0.20414681448652136, "grad_norm": 3.5052852630615234, "learning_rate": 1.8451643630274664e-05, "loss": 0.16831854581832886, "step": 7750 }, { "epoch": 0.2044102297310201, "grad_norm": 1.8573521375656128, "learning_rate": 1.8447080480965694e-05, "loss": 0.17167856693267822, "step": 7760 }, { "epoch": 0.20467364497551885, "grad_norm": 3.0610623359680176, "learning_rate": 1.844251118350711e-05, "loss": 0.1654578924179077, "step": 7770 }, { "epoch": 0.2049370602200176, "grad_norm": 1.8958430290222168, "learning_rate": 1.8437935741224652e-05, "loss": 0.17650134563446046, "step": 7780 }, { "epoch": 0.2052004754645163, "grad_norm": 1.5732953548431396, "learning_rate": 1.8433354157448513e-05, "loss": 0.1704841136932373, "step": 7790 }, { "epoch": 0.20546389070901505, "grad_norm": 4.600805759429932, "learning_rate": 1.8428766435513373e-05, "loss": 0.17308875322341918, "step": 7800 }, { "epoch": 0.2057273059535138, "grad_norm": 6.324937343597412, "learning_rate": 1.8424172578758374e-05, "loss": 0.17094852924346923, "step": 7810 }, { "epoch": 0.20599072119801254, "grad_norm": 4.09133768081665, "learning_rate": 1.841957259052712e-05, "loss": 0.1605220317840576, "step": 7820 }, { "epoch": 0.20625413644251128, "grad_norm": 1.807431697845459, "learning_rate": 1.8414966474167688e-05, "loss": 0.17546494007110597, "step": 7830 }, { "epoch": 0.20651755168701, "grad_norm": 5.808052062988281, "learning_rate": 1.8410354233032604e-05, "loss": 0.1582520365715027, "step": 7840 }, { "epoch": 0.20678096693150874, "grad_norm": 1.1619890928268433, "learning_rate": 1.840573587047885e-05, "loss": 0.16894900798797607, "step": 7850 }, { "epoch": 0.20704438217600749, "grad_norm": 1.3188138008117676, "learning_rate": 1.8401111389867882e-05, "loss": 0.180841064453125, "step": 7860 }, { "epoch": 0.20730779742050623, "grad_norm": 3.5619540214538574, "learning_rate": 1.8396480794565583e-05, "loss": 0.1648564338684082, "step": 7870 }, { "epoch": 0.20757121266500494, "grad_norm": 2.56322979927063, "learning_rate": 1.839184408794231e-05, "loss": 0.1723032832145691, "step": 7880 }, { "epoch": 0.2078346279095037, "grad_norm": 1.4702250957489014, "learning_rate": 1.8387201273372855e-05, "loss": 0.16760246753692626, "step": 7890 }, { "epoch": 0.20809804315400243, "grad_norm": 5.092892646789551, "learning_rate": 1.838255235423646e-05, "loss": 0.1567755937576294, "step": 7900 }, { "epoch": 0.20836145839850118, "grad_norm": 1.8203986883163452, "learning_rate": 1.8377897333916808e-05, "loss": 0.1641380786895752, "step": 7910 }, { "epoch": 0.2086248736429999, "grad_norm": 5.310081958770752, "learning_rate": 1.837323621580202e-05, "loss": 0.15717867612838746, "step": 7920 }, { "epoch": 0.20888828888749864, "grad_norm": 4.2085747718811035, "learning_rate": 1.8368569003284662e-05, "loss": 0.16271182298660278, "step": 7930 }, { "epoch": 0.20915170413199738, "grad_norm": 2.358267068862915, "learning_rate": 1.8363895699761736e-05, "loss": 0.17364985942840577, "step": 7940 }, { "epoch": 0.20941511937649612, "grad_norm": 5.847020149230957, "learning_rate": 1.8359216308634668e-05, "loss": 0.16563925743103028, "step": 7950 }, { "epoch": 0.20967853462099487, "grad_norm": 0.798214316368103, "learning_rate": 1.8354530833309323e-05, "loss": 0.16987030506134032, "step": 7960 }, { "epoch": 0.20994194986549358, "grad_norm": 3.048307180404663, "learning_rate": 1.8349839277195993e-05, "loss": 0.15956900119781495, "step": 7970 }, { "epoch": 0.21020536510999233, "grad_norm": 2.39316463470459, "learning_rate": 1.8345141643709394e-05, "loss": 0.1690322756767273, "step": 7980 }, { "epoch": 0.21046878035449107, "grad_norm": 3.321821451187134, "learning_rate": 1.8340437936268667e-05, "loss": 0.1732789993286133, "step": 7990 }, { "epoch": 0.2107321955989898, "grad_norm": 4.378774166107178, "learning_rate": 1.8335728158297374e-05, "loss": 0.17706646919250488, "step": 8000 }, { "epoch": 0.21099561084348853, "grad_norm": 1.0986887216567993, "learning_rate": 1.833101231322349e-05, "loss": 0.17701363563537598, "step": 8010 }, { "epoch": 0.21125902608798727, "grad_norm": 4.139017105102539, "learning_rate": 1.832629040447942e-05, "loss": 0.18133198022842406, "step": 8020 }, { "epoch": 0.21152244133248602, "grad_norm": 1.7217844724655151, "learning_rate": 1.8321562435501963e-05, "loss": 0.1858844518661499, "step": 8030 }, { "epoch": 0.21178585657698476, "grad_norm": 2.747563362121582, "learning_rate": 1.831682840973234e-05, "loss": 0.1629952073097229, "step": 8040 }, { "epoch": 0.21204927182148348, "grad_norm": 1.6161788702011108, "learning_rate": 1.8312088330616188e-05, "loss": 0.17265627384185792, "step": 8050 }, { "epoch": 0.21231268706598222, "grad_norm": 1.9221843481063843, "learning_rate": 1.8307342201603533e-05, "loss": 0.15791378021240235, "step": 8060 }, { "epoch": 0.21257610231048096, "grad_norm": 2.1965527534484863, "learning_rate": 1.830259002614881e-05, "loss": 0.15694574117660523, "step": 8070 }, { "epoch": 0.2128395175549797, "grad_norm": 0.9037136435508728, "learning_rate": 1.829783180771087e-05, "loss": 0.17699925899505614, "step": 8080 }, { "epoch": 0.21310293279947845, "grad_norm": 1.8319586515426636, "learning_rate": 1.8293067549752937e-05, "loss": 0.1703951597213745, "step": 8090 }, { "epoch": 0.21336634804397717, "grad_norm": 1.102665901184082, "learning_rate": 1.8288297255742646e-05, "loss": 0.18982118368148804, "step": 8100 }, { "epoch": 0.2136297632884759, "grad_norm": 3.4327757358551025, "learning_rate": 1.828352092915203e-05, "loss": 0.1710628867149353, "step": 8110 }, { "epoch": 0.21389317853297465, "grad_norm": 1.4730970859527588, "learning_rate": 1.8278738573457496e-05, "loss": 0.15348416566848755, "step": 8120 }, { "epoch": 0.2141565937774734, "grad_norm": 1.1160558462142944, "learning_rate": 1.8273950192139854e-05, "loss": 0.16850967407226564, "step": 8130 }, { "epoch": 0.2144200090219721, "grad_norm": 4.285676002502441, "learning_rate": 1.826915578868429e-05, "loss": 0.17628827095031738, "step": 8140 }, { "epoch": 0.21468342426647086, "grad_norm": 3.5875189304351807, "learning_rate": 1.8264355366580387e-05, "loss": 0.16531918048858643, "step": 8150 }, { "epoch": 0.2149468395109696, "grad_norm": 1.0301213264465332, "learning_rate": 1.825954892932209e-05, "loss": 0.16193549633026122, "step": 8160 }, { "epoch": 0.21521025475546834, "grad_norm": 1.5795708894729614, "learning_rate": 1.8254736480407738e-05, "loss": 0.1626498818397522, "step": 8170 }, { "epoch": 0.21547366999996706, "grad_norm": 1.8823148012161255, "learning_rate": 1.8249918023340036e-05, "loss": 0.15947791337966918, "step": 8180 }, { "epoch": 0.2157370852444658, "grad_norm": 3.916524648666382, "learning_rate": 1.8245093561626063e-05, "loss": 0.16956694126129152, "step": 8190 }, { "epoch": 0.21600050048896455, "grad_norm": 4.349510669708252, "learning_rate": 1.824026309877727e-05, "loss": 0.17732396125793456, "step": 8200 }, { "epoch": 0.2162639157334633, "grad_norm": 3.5279691219329834, "learning_rate": 1.8235426638309483e-05, "loss": 0.17857892513275148, "step": 8210 }, { "epoch": 0.21652733097796203, "grad_norm": 1.498414397239685, "learning_rate": 1.823058418374288e-05, "loss": 0.17476658821105956, "step": 8220 }, { "epoch": 0.21679074622246075, "grad_norm": 2.0657331943511963, "learning_rate": 1.822573573860201e-05, "loss": 0.15963902473449706, "step": 8230 }, { "epoch": 0.2170541614669595, "grad_norm": 0.9068095684051514, "learning_rate": 1.8220881306415783e-05, "loss": 0.16444916725158693, "step": 8240 }, { "epoch": 0.21731757671145824, "grad_norm": 1.3565354347229004, "learning_rate": 1.821602089071746e-05, "loss": 0.1584041714668274, "step": 8250 }, { "epoch": 0.21758099195595698, "grad_norm": 1.5323035717010498, "learning_rate": 1.8211154495044666e-05, "loss": 0.1724594473838806, "step": 8260 }, { "epoch": 0.2178444072004557, "grad_norm": 1.1768689155578613, "learning_rate": 1.8206282122939368e-05, "loss": 0.15612294673919677, "step": 8270 }, { "epoch": 0.21810782244495444, "grad_norm": 2.4337174892425537, "learning_rate": 1.8201403777947896e-05, "loss": 0.16237227916717528, "step": 8280 }, { "epoch": 0.21837123768945318, "grad_norm": 0.7439559102058411, "learning_rate": 1.8196519463620912e-05, "loss": 0.16947522163391113, "step": 8290 }, { "epoch": 0.21863465293395193, "grad_norm": 2.0236780643463135, "learning_rate": 1.8191629183513435e-05, "loss": 0.1640971302986145, "step": 8300 }, { "epoch": 0.21889806817845064, "grad_norm": 0.8326374888420105, "learning_rate": 1.8186732941184824e-05, "loss": 0.1637672781944275, "step": 8310 }, { "epoch": 0.2191614834229494, "grad_norm": 1.796667218208313, "learning_rate": 1.818183074019877e-05, "loss": 0.16751667261123657, "step": 8320 }, { "epoch": 0.21942489866744813, "grad_norm": 2.8410205841064453, "learning_rate": 1.8176922584123308e-05, "loss": 0.16054805517196655, "step": 8330 }, { "epoch": 0.21968831391194688, "grad_norm": 3.7460551261901855, "learning_rate": 1.817200847653081e-05, "loss": 0.18631840944290162, "step": 8340 }, { "epoch": 0.21995172915644562, "grad_norm": 1.0255998373031616, "learning_rate": 1.8167088420997975e-05, "loss": 0.18312858343124389, "step": 8350 }, { "epoch": 0.22021514440094433, "grad_norm": 2.1522016525268555, "learning_rate": 1.816216242110582e-05, "loss": 0.1767812490463257, "step": 8360 }, { "epoch": 0.22047855964544308, "grad_norm": 1.7517930269241333, "learning_rate": 1.815723048043972e-05, "loss": 0.16181308031082153, "step": 8370 }, { "epoch": 0.22074197488994182, "grad_norm": 1.961943507194519, "learning_rate": 1.8152292602589336e-05, "loss": 0.17094357013702394, "step": 8380 }, { "epoch": 0.22100539013444057, "grad_norm": 1.2619025707244873, "learning_rate": 1.8147348791148678e-05, "loss": 0.17886215448379517, "step": 8390 }, { "epoch": 0.22126880537893928, "grad_norm": 0.8030175566673279, "learning_rate": 1.8142399049716065e-05, "loss": 0.1637798070907593, "step": 8400 }, { "epoch": 0.22153222062343803, "grad_norm": 2.235788106918335, "learning_rate": 1.8137443381894126e-05, "loss": 0.1826714277267456, "step": 8410 }, { "epoch": 0.22179563586793677, "grad_norm": 0.8370832204818726, "learning_rate": 1.813248179128982e-05, "loss": 0.16154632568359376, "step": 8420 }, { "epoch": 0.2220590511124355, "grad_norm": 0.788213312625885, "learning_rate": 1.81275142815144e-05, "loss": 0.17518948316574096, "step": 8430 }, { "epoch": 0.22232246635693423, "grad_norm": 3.6824328899383545, "learning_rate": 1.812254085618343e-05, "loss": 0.17220652103424072, "step": 8440 }, { "epoch": 0.22258588160143297, "grad_norm": 3.760133981704712, "learning_rate": 1.8117561518916792e-05, "loss": 0.1680700659751892, "step": 8450 }, { "epoch": 0.22284929684593172, "grad_norm": 2.5600826740264893, "learning_rate": 1.811257627333866e-05, "loss": 0.17138888835906982, "step": 8460 }, { "epoch": 0.22311271209043046, "grad_norm": 0.7360926270484924, "learning_rate": 1.810758512307751e-05, "loss": 0.16549522876739503, "step": 8470 }, { "epoch": 0.2233761273349292, "grad_norm": 4.991323947906494, "learning_rate": 1.8102588071766116e-05, "loss": 0.17228050231933595, "step": 8480 }, { "epoch": 0.22363954257942792, "grad_norm": 2.6040077209472656, "learning_rate": 1.8097585123041547e-05, "loss": 0.16345622539520263, "step": 8490 }, { "epoch": 0.22390295782392666, "grad_norm": 3.353663682937622, "learning_rate": 1.809257628054517e-05, "loss": 0.16786420345306396, "step": 8500 }, { "epoch": 0.2241663730684254, "grad_norm": 1.633655071258545, "learning_rate": 1.808756154792263e-05, "loss": 0.17182614803314208, "step": 8510 }, { "epoch": 0.22442978831292415, "grad_norm": 3.54319167137146, "learning_rate": 1.8082540928823866e-05, "loss": 0.17249183654785155, "step": 8520 }, { "epoch": 0.22469320355742287, "grad_norm": 0.9022901654243469, "learning_rate": 1.8077514426903107e-05, "loss": 0.17170660495758056, "step": 8530 }, { "epoch": 0.2249566188019216, "grad_norm": 2.675499439239502, "learning_rate": 1.8072482045818852e-05, "loss": 0.15625488758087158, "step": 8540 }, { "epoch": 0.22522003404642035, "grad_norm": 2.4010210037231445, "learning_rate": 1.806744378923389e-05, "loss": 0.16569451093673707, "step": 8550 }, { "epoch": 0.2254834492909191, "grad_norm": 2.042875051498413, "learning_rate": 1.8062399660815277e-05, "loss": 0.17066457271575927, "step": 8560 }, { "epoch": 0.2257468645354178, "grad_norm": 1.0451970100402832, "learning_rate": 1.8057349664234348e-05, "loss": 0.15659682750701903, "step": 8570 }, { "epoch": 0.22601027977991656, "grad_norm": 2.003408908843994, "learning_rate": 1.8052293803166714e-05, "loss": 0.17222245931625366, "step": 8580 }, { "epoch": 0.2262736950244153, "grad_norm": 1.9228003025054932, "learning_rate": 1.8047232081292242e-05, "loss": 0.16385233402252197, "step": 8590 }, { "epoch": 0.22653711026891404, "grad_norm": 1.7854118347167969, "learning_rate": 1.8042164502295074e-05, "loss": 0.18269015550613404, "step": 8600 }, { "epoch": 0.2268005255134128, "grad_norm": 0.882986307144165, "learning_rate": 1.8037091069863615e-05, "loss": 0.16489899158477783, "step": 8610 }, { "epoch": 0.2270639407579115, "grad_norm": 2.0755813121795654, "learning_rate": 1.8032011787690523e-05, "loss": 0.1630818009376526, "step": 8620 }, { "epoch": 0.22732735600241025, "grad_norm": 3.4779562950134277, "learning_rate": 1.8026926659472727e-05, "loss": 0.15938680171966552, "step": 8630 }, { "epoch": 0.227590771246909, "grad_norm": 1.5413002967834473, "learning_rate": 1.802183568891139e-05, "loss": 0.16981608867645265, "step": 8640 }, { "epoch": 0.22785418649140773, "grad_norm": 2.81174898147583, "learning_rate": 1.801673887971195e-05, "loss": 0.1784644365310669, "step": 8650 }, { "epoch": 0.22811760173590645, "grad_norm": 1.8718162775039673, "learning_rate": 1.801163623558408e-05, "loss": 0.16923441886901855, "step": 8660 }, { "epoch": 0.2283810169804052, "grad_norm": 1.530125379562378, "learning_rate": 1.800652776024171e-05, "loss": 0.16267595291137696, "step": 8670 }, { "epoch": 0.22864443222490394, "grad_norm": 2.272157907485962, "learning_rate": 1.8001413457403005e-05, "loss": 0.1689075231552124, "step": 8680 }, { "epoch": 0.22890784746940268, "grad_norm": 1.9084858894348145, "learning_rate": 1.7996293330790374e-05, "loss": 0.1693948268890381, "step": 8690 }, { "epoch": 0.2291712627139014, "grad_norm": 1.205635666847229, "learning_rate": 1.7991167384130466e-05, "loss": 0.16304999589920044, "step": 8700 }, { "epoch": 0.22943467795840014, "grad_norm": 1.7590794563293457, "learning_rate": 1.798603562115417e-05, "loss": 0.1658156156539917, "step": 8710 }, { "epoch": 0.22969809320289888, "grad_norm": 1.7716587781906128, "learning_rate": 1.79808980455966e-05, "loss": 0.16502032279968262, "step": 8720 }, { "epoch": 0.22996150844739763, "grad_norm": 1.0306954383850098, "learning_rate": 1.7975754661197106e-05, "loss": 0.16331005096435547, "step": 8730 }, { "epoch": 0.23022492369189637, "grad_norm": 1.9684290885925293, "learning_rate": 1.797060547169927e-05, "loss": 0.16990772485733033, "step": 8740 }, { "epoch": 0.2304883389363951, "grad_norm": 1.9737530946731567, "learning_rate": 1.796545048085089e-05, "loss": 0.16810165643692015, "step": 8750 }, { "epoch": 0.23075175418089383, "grad_norm": 1.2646992206573486, "learning_rate": 1.796028969240399e-05, "loss": 0.16663039922714235, "step": 8760 }, { "epoch": 0.23101516942539257, "grad_norm": 1.6061220169067383, "learning_rate": 1.7955123110114815e-05, "loss": 0.16965718269348146, "step": 8770 }, { "epoch": 0.23127858466989132, "grad_norm": 1.0871518850326538, "learning_rate": 1.794995073774383e-05, "loss": 0.15506887435913086, "step": 8780 }, { "epoch": 0.23154199991439003, "grad_norm": 0.7651626467704773, "learning_rate": 1.7944772579055705e-05, "loss": 0.1719715714454651, "step": 8790 }, { "epoch": 0.23180541515888878, "grad_norm": 4.068540096282959, "learning_rate": 1.793958863781933e-05, "loss": 0.16795437335968016, "step": 8800 }, { "epoch": 0.23206883040338752, "grad_norm": 2.624683141708374, "learning_rate": 1.7934398917807802e-05, "loss": 0.19373276233673095, "step": 8810 }, { "epoch": 0.23233224564788627, "grad_norm": 1.5921249389648438, "learning_rate": 1.7929203422798426e-05, "loss": 0.16160991191864013, "step": 8820 }, { "epoch": 0.23259566089238498, "grad_norm": 1.9465101957321167, "learning_rate": 1.79240021565727e-05, "loss": 0.1634509325027466, "step": 8830 }, { "epoch": 0.23285907613688372, "grad_norm": 5.72005558013916, "learning_rate": 1.7918795122916337e-05, "loss": 0.17032263278961182, "step": 8840 }, { "epoch": 0.23312249138138247, "grad_norm": 0.8429386615753174, "learning_rate": 1.7913582325619236e-05, "loss": 0.171757709980011, "step": 8850 }, { "epoch": 0.2333859066258812, "grad_norm": 3.7584540843963623, "learning_rate": 1.7908363768475498e-05, "loss": 0.17103884220123292, "step": 8860 }, { "epoch": 0.23364932187037996, "grad_norm": 2.578233480453491, "learning_rate": 1.7903139455283412e-05, "loss": 0.1715204358100891, "step": 8870 }, { "epoch": 0.23391273711487867, "grad_norm": 4.824211597442627, "learning_rate": 1.7897909389845462e-05, "loss": 0.1721007466316223, "step": 8880 }, { "epoch": 0.23417615235937742, "grad_norm": 6.093372821807861, "learning_rate": 1.789267357596831e-05, "loss": 0.16257503032684326, "step": 8890 }, { "epoch": 0.23443956760387616, "grad_norm": 1.9143023490905762, "learning_rate": 1.7887432017462812e-05, "loss": 0.17467330694198607, "step": 8900 }, { "epoch": 0.2347029828483749, "grad_norm": 5.524956703186035, "learning_rate": 1.7882184718144002e-05, "loss": 0.17833387851715088, "step": 8910 }, { "epoch": 0.23496639809287362, "grad_norm": 3.4731881618499756, "learning_rate": 1.7876931681831087e-05, "loss": 0.16882646083831787, "step": 8920 }, { "epoch": 0.23522981333737236, "grad_norm": 1.6157647371292114, "learning_rate": 1.787167291234745e-05, "loss": 0.16738191843032837, "step": 8930 }, { "epoch": 0.2354932285818711, "grad_norm": 6.206955432891846, "learning_rate": 1.7866408413520662e-05, "loss": 0.16089837551116942, "step": 8940 }, { "epoch": 0.23575664382636985, "grad_norm": 1.8435863256454468, "learning_rate": 1.786113818918244e-05, "loss": 0.1770928382873535, "step": 8950 }, { "epoch": 0.23602005907086857, "grad_norm": 3.8322701454162598, "learning_rate": 1.7855862243168684e-05, "loss": 0.15710458755493165, "step": 8960 }, { "epoch": 0.2362834743153673, "grad_norm": 2.01676869392395, "learning_rate": 1.785058057931946e-05, "loss": 0.16996997594833374, "step": 8970 }, { "epoch": 0.23654688955986605, "grad_norm": 2.9885246753692627, "learning_rate": 1.7845293201478985e-05, "loss": 0.181387197971344, "step": 8980 }, { "epoch": 0.2368103048043648, "grad_norm": 1.2926081418991089, "learning_rate": 1.7840000113495647e-05, "loss": 0.17049378156661987, "step": 8990 }, { "epoch": 0.23707372004886354, "grad_norm": 6.073495864868164, "learning_rate": 1.7834701319221978e-05, "loss": 0.1721758484840393, "step": 9000 }, { "epoch": 0.23733713529336226, "grad_norm": 1.56259286403656, "learning_rate": 1.782939682251467e-05, "loss": 0.16776742935180664, "step": 9010 }, { "epoch": 0.237600550537861, "grad_norm": 2.518516778945923, "learning_rate": 1.7824086627234565e-05, "loss": 0.17066075801849365, "step": 9020 }, { "epoch": 0.23786396578235974, "grad_norm": 4.587469577789307, "learning_rate": 1.7818770737246653e-05, "loss": 0.17038898468017577, "step": 9030 }, { "epoch": 0.2381273810268585, "grad_norm": 1.2567633390426636, "learning_rate": 1.7813449156420072e-05, "loss": 0.1432661533355713, "step": 9040 }, { "epoch": 0.2383907962713572, "grad_norm": 2.357633113861084, "learning_rate": 1.7808121888628093e-05, "loss": 0.1717435598373413, "step": 9050 }, { "epoch": 0.23865421151585595, "grad_norm": 1.465989351272583, "learning_rate": 1.780278893774813e-05, "loss": 0.1672459363937378, "step": 9060 }, { "epoch": 0.2389176267603547, "grad_norm": 2.3341777324676514, "learning_rate": 1.7797450307661733e-05, "loss": 0.16577794551849365, "step": 9070 }, { "epoch": 0.23918104200485343, "grad_norm": 3.3832483291625977, "learning_rate": 1.77921060022546e-05, "loss": 0.1674659490585327, "step": 9080 }, { "epoch": 0.23944445724935215, "grad_norm": 1.5483368635177612, "learning_rate": 1.7786756025416535e-05, "loss": 0.17533764839172364, "step": 9090 }, { "epoch": 0.2397078724938509, "grad_norm": 2.0270514488220215, "learning_rate": 1.7781400381041484e-05, "loss": 0.1680479884147644, "step": 9100 }, { "epoch": 0.23997128773834964, "grad_norm": 1.1625314950942993, "learning_rate": 1.777603907302752e-05, "loss": 0.16522928476333618, "step": 9110 }, { "epoch": 0.24023470298284838, "grad_norm": 2.130340337753296, "learning_rate": 1.7770672105276833e-05, "loss": 0.1510825514793396, "step": 9120 }, { "epoch": 0.24049811822734712, "grad_norm": 2.2340962886810303, "learning_rate": 1.776529948169573e-05, "loss": 0.17037367820739746, "step": 9130 }, { "epoch": 0.24076153347184584, "grad_norm": 1.9374818801879883, "learning_rate": 1.775992120619464e-05, "loss": 0.17572952508926393, "step": 9140 }, { "epoch": 0.24102494871634458, "grad_norm": 1.5641545057296753, "learning_rate": 1.7754537282688104e-05, "loss": 0.16501060724258423, "step": 9150 }, { "epoch": 0.24128836396084333, "grad_norm": 3.1594552993774414, "learning_rate": 1.7749147715094777e-05, "loss": 0.1561112642288208, "step": 9160 }, { "epoch": 0.24155177920534207, "grad_norm": 1.4782109260559082, "learning_rate": 1.7743752507337414e-05, "loss": 0.177468204498291, "step": 9170 }, { "epoch": 0.2418151944498408, "grad_norm": 3.0976295471191406, "learning_rate": 1.7738351663342887e-05, "loss": 0.17096641063690185, "step": 9180 }, { "epoch": 0.24207860969433953, "grad_norm": 1.6709790229797363, "learning_rate": 1.7732945187042155e-05, "loss": 0.16008195877075196, "step": 9190 }, { "epoch": 0.24234202493883827, "grad_norm": 1.7195671796798706, "learning_rate": 1.7727533082370288e-05, "loss": 0.14919466972351075, "step": 9200 }, { "epoch": 0.24260544018333702, "grad_norm": 1.0383801460266113, "learning_rate": 1.772211535326645e-05, "loss": 0.18376386165618896, "step": 9210 }, { "epoch": 0.24286885542783573, "grad_norm": 2.6052052974700928, "learning_rate": 1.7716692003673898e-05, "loss": 0.15817463397979736, "step": 9220 }, { "epoch": 0.24313227067233448, "grad_norm": 2.6222047805786133, "learning_rate": 1.7711263037539978e-05, "loss": 0.18346331119537354, "step": 9230 }, { "epoch": 0.24339568591683322, "grad_norm": 3.529115915298462, "learning_rate": 1.770582845881613e-05, "loss": 0.17128732204437255, "step": 9240 }, { "epoch": 0.24365910116133196, "grad_norm": 1.2127983570098877, "learning_rate": 1.7700388271457872e-05, "loss": 0.16625311374664306, "step": 9250 }, { "epoch": 0.2439225164058307, "grad_norm": 1.0232747793197632, "learning_rate": 1.7694942479424803e-05, "loss": 0.1659519076347351, "step": 9260 }, { "epoch": 0.24418593165032942, "grad_norm": 2.381772518157959, "learning_rate": 1.768949108668061e-05, "loss": 0.15629289150238038, "step": 9270 }, { "epoch": 0.24444934689482817, "grad_norm": 1.0578994750976562, "learning_rate": 1.7684034097193055e-05, "loss": 0.15308015346527098, "step": 9280 }, { "epoch": 0.2447127621393269, "grad_norm": 4.632020473480225, "learning_rate": 1.7678571514933966e-05, "loss": 0.16751242876052858, "step": 9290 }, { "epoch": 0.24497617738382566, "grad_norm": 4.177544593811035, "learning_rate": 1.7673103343879242e-05, "loss": 0.16513726711273194, "step": 9300 }, { "epoch": 0.24523959262832437, "grad_norm": 1.8288462162017822, "learning_rate": 1.7667629588008862e-05, "loss": 0.15825300216674804, "step": 9310 }, { "epoch": 0.24550300787282311, "grad_norm": 1.186229944229126, "learning_rate": 1.766215025130686e-05, "loss": 0.16069140434265136, "step": 9320 }, { "epoch": 0.24576642311732186, "grad_norm": 1.0694271326065063, "learning_rate": 1.7656665337761327e-05, "loss": 0.16384910345077514, "step": 9330 }, { "epoch": 0.2460298383618206, "grad_norm": 1.1911624670028687, "learning_rate": 1.7651174851364427e-05, "loss": 0.16345407962799072, "step": 9340 }, { "epoch": 0.24629325360631932, "grad_norm": 6.816508769989014, "learning_rate": 1.764567879611237e-05, "loss": 0.1661921977996826, "step": 9350 }, { "epoch": 0.24655666885081806, "grad_norm": 3.312504291534424, "learning_rate": 1.7640177176005425e-05, "loss": 0.1621119499206543, "step": 9360 }, { "epoch": 0.2468200840953168, "grad_norm": 1.1982641220092773, "learning_rate": 1.7634669995047904e-05, "loss": 0.17733452320098878, "step": 9370 }, { "epoch": 0.24708349933981555, "grad_norm": 1.874787449836731, "learning_rate": 1.7629157257248175e-05, "loss": 0.15752644538879396, "step": 9380 }, { "epoch": 0.2473469145843143, "grad_norm": 1.1695481538772583, "learning_rate": 1.762363896661864e-05, "loss": 0.16198962926864624, "step": 9390 }, { "epoch": 0.247610329828813, "grad_norm": 4.119010925292969, "learning_rate": 1.761811512717576e-05, "loss": 0.16779084205627443, "step": 9400 }, { "epoch": 0.24787374507331175, "grad_norm": 1.3103113174438477, "learning_rate": 1.761258574294001e-05, "loss": 0.14933569431304933, "step": 9410 }, { "epoch": 0.2481371603178105, "grad_norm": 2.178499221801758, "learning_rate": 1.760705081793593e-05, "loss": 0.16521581411361694, "step": 9420 }, { "epoch": 0.24840057556230924, "grad_norm": 3.82889723777771, "learning_rate": 1.760151035619206e-05, "loss": 0.17594034671783448, "step": 9430 }, { "epoch": 0.24866399080680796, "grad_norm": 0.9229384064674377, "learning_rate": 1.7595964361741e-05, "loss": 0.16105185747146605, "step": 9440 }, { "epoch": 0.2489274060513067, "grad_norm": 4.086063861846924, "learning_rate": 1.759041283861936e-05, "loss": 0.1695645809173584, "step": 9450 }, { "epoch": 0.24919082129580544, "grad_norm": 1.647241473197937, "learning_rate": 1.7584855790867773e-05, "loss": 0.16659388542175294, "step": 9460 }, { "epoch": 0.2494542365403042, "grad_norm": 1.3214895725250244, "learning_rate": 1.7579293222530907e-05, "loss": 0.17211365699768066, "step": 9470 }, { "epoch": 0.24971765178480293, "grad_norm": 2.0166406631469727, "learning_rate": 1.7573725137657432e-05, "loss": 0.16315600872039795, "step": 9480 }, { "epoch": 0.24998106702930165, "grad_norm": 2.7264046669006348, "learning_rate": 1.756815154030005e-05, "loss": 0.16033079624176025, "step": 9490 }, { "epoch": 0.2502444822738004, "grad_norm": 0.9574469327926636, "learning_rate": 1.756257243451545e-05, "loss": 0.15452346801757813, "step": 9500 }, { "epoch": 0.2505078975182991, "grad_norm": 1.380598545074463, "learning_rate": 1.7556987824364363e-05, "loss": 0.16524865627288818, "step": 9510 }, { "epoch": 0.25077131276279785, "grad_norm": 2.3946871757507324, "learning_rate": 1.75513977139115e-05, "loss": 0.16539311408996582, "step": 9520 }, { "epoch": 0.2510347280072966, "grad_norm": 5.426186561584473, "learning_rate": 1.7545802107225585e-05, "loss": 0.17423610687255858, "step": 9530 }, { "epoch": 0.25129814325179534, "grad_norm": 1.0943360328674316, "learning_rate": 1.754020100837934e-05, "loss": 0.17193944454193116, "step": 9540 }, { "epoch": 0.2515615584962941, "grad_norm": 1.6234475374221802, "learning_rate": 1.753459442144949e-05, "loss": 0.17230201959609986, "step": 9550 }, { "epoch": 0.2518249737407928, "grad_norm": 4.0484938621521, "learning_rate": 1.7528982350516754e-05, "loss": 0.1651078939437866, "step": 9560 }, { "epoch": 0.25208838898529157, "grad_norm": 2.0297536849975586, "learning_rate": 1.7523364799665832e-05, "loss": 0.1604618549346924, "step": 9570 }, { "epoch": 0.2523518042297903, "grad_norm": 2.2787113189697266, "learning_rate": 1.7517741772985424e-05, "loss": 0.16765229701995848, "step": 9580 }, { "epoch": 0.252615219474289, "grad_norm": 0.9259963035583496, "learning_rate": 1.751211327456821e-05, "loss": 0.16891340017318726, "step": 9590 }, { "epoch": 0.25287863471878774, "grad_norm": 4.48685359954834, "learning_rate": 1.7506479308510855e-05, "loss": 0.17027032375335693, "step": 9600 }, { "epoch": 0.2531420499632865, "grad_norm": 4.649251937866211, "learning_rate": 1.7500839878914004e-05, "loss": 0.16042076349258422, "step": 9610 }, { "epoch": 0.25340546520778523, "grad_norm": 2.0527420043945312, "learning_rate": 1.749519498988227e-05, "loss": 0.17904973030090332, "step": 9620 }, { "epoch": 0.253668880452284, "grad_norm": 4.270792484283447, "learning_rate": 1.748954464552426e-05, "loss": 0.1680680990219116, "step": 9630 }, { "epoch": 0.2539322956967827, "grad_norm": 0.9698401093482971, "learning_rate": 1.7483888849952525e-05, "loss": 0.1742458462715149, "step": 9640 }, { "epoch": 0.25419571094128146, "grad_norm": 2.8758797645568848, "learning_rate": 1.7478227607283604e-05, "loss": 0.15941661596298218, "step": 9650 }, { "epoch": 0.2544591261857802, "grad_norm": 1.312247633934021, "learning_rate": 1.7472560921637993e-05, "loss": 0.16757116317749024, "step": 9660 }, { "epoch": 0.25472254143027895, "grad_norm": 4.787957668304443, "learning_rate": 1.746688879714015e-05, "loss": 0.16725537776947022, "step": 9670 }, { "epoch": 0.25498595667477764, "grad_norm": 1.1892850399017334, "learning_rate": 1.746121123791849e-05, "loss": 0.16907689571380616, "step": 9680 }, { "epoch": 0.2552493719192764, "grad_norm": 4.947500228881836, "learning_rate": 1.7455528248105385e-05, "loss": 0.17784414291381836, "step": 9690 }, { "epoch": 0.2555127871637751, "grad_norm": 1.4554306268692017, "learning_rate": 1.7449839831837166e-05, "loss": 0.17530922889709472, "step": 9700 }, { "epoch": 0.25577620240827387, "grad_norm": 1.960382103919983, "learning_rate": 1.7444145993254105e-05, "loss": 0.15095317363739014, "step": 9710 }, { "epoch": 0.2560396176527726, "grad_norm": 0.6459771394729614, "learning_rate": 1.7438446736500418e-05, "loss": 0.17640599012374877, "step": 9720 }, { "epoch": 0.25630303289727135, "grad_norm": 3.330103874206543, "learning_rate": 1.7432742065724276e-05, "loss": 0.16739974021911622, "step": 9730 }, { "epoch": 0.2565664481417701, "grad_norm": 3.131443500518799, "learning_rate": 1.7427031985077785e-05, "loss": 0.1670603036880493, "step": 9740 }, { "epoch": 0.25682986338626884, "grad_norm": 2.5550453662872314, "learning_rate": 1.7421316498716986e-05, "loss": 0.14717557430267333, "step": 9750 }, { "epoch": 0.2570932786307676, "grad_norm": 3.53070068359375, "learning_rate": 1.741559561080185e-05, "loss": 0.154929780960083, "step": 9760 }, { "epoch": 0.2573566938752663, "grad_norm": 2.1708788871765137, "learning_rate": 1.74098693254963e-05, "loss": 0.1439017176628113, "step": 9770 }, { "epoch": 0.257620109119765, "grad_norm": 1.5836647748947144, "learning_rate": 1.7404137646968162e-05, "loss": 0.16277723312377929, "step": 9780 }, { "epoch": 0.25788352436426376, "grad_norm": 1.7132405042648315, "learning_rate": 1.7398400579389204e-05, "loss": 0.15093826055526732, "step": 9790 }, { "epoch": 0.2581469396087625, "grad_norm": 2.9093592166900635, "learning_rate": 1.739265812693511e-05, "loss": 0.17131611108779907, "step": 9800 }, { "epoch": 0.25841035485326125, "grad_norm": 2.9501261711120605, "learning_rate": 1.7386910293785485e-05, "loss": 0.16204776763916015, "step": 9810 }, { "epoch": 0.25867377009776, "grad_norm": 3.028329610824585, "learning_rate": 1.738115708412385e-05, "loss": 0.1518233299255371, "step": 9820 }, { "epoch": 0.25893718534225874, "grad_norm": 1.7527090311050415, "learning_rate": 1.737539850213764e-05, "loss": 0.17185790538787843, "step": 9830 }, { "epoch": 0.2592006005867575, "grad_norm": 2.2706334590911865, "learning_rate": 1.7369634552018196e-05, "loss": 0.16264610290527343, "step": 9840 }, { "epoch": 0.25946401583125617, "grad_norm": 2.0018460750579834, "learning_rate": 1.7363865237960773e-05, "loss": 0.160880184173584, "step": 9850 }, { "epoch": 0.2597274310757549, "grad_norm": 2.494598388671875, "learning_rate": 1.7358090564164524e-05, "loss": 0.15832366943359374, "step": 9860 }, { "epoch": 0.25999084632025365, "grad_norm": 1.9412529468536377, "learning_rate": 1.735231053483251e-05, "loss": 0.1636672019958496, "step": 9870 }, { "epoch": 0.2602542615647524, "grad_norm": 0.7912483811378479, "learning_rate": 1.734652515417168e-05, "loss": 0.17658305168151855, "step": 9880 }, { "epoch": 0.26051767680925114, "grad_norm": 1.4129468202590942, "learning_rate": 1.7340734426392894e-05, "loss": 0.164047908782959, "step": 9890 }, { "epoch": 0.2607810920537499, "grad_norm": 2.489975690841675, "learning_rate": 1.7334938355710876e-05, "loss": 0.17179975509643555, "step": 9900 }, { "epoch": 0.26104450729824863, "grad_norm": 1.4137898683547974, "learning_rate": 1.7329136946344272e-05, "loss": 0.17372653484344483, "step": 9910 }, { "epoch": 0.2613079225427474, "grad_norm": 1.1850107908248901, "learning_rate": 1.7323330202515592e-05, "loss": 0.16095125675201416, "step": 9920 }, { "epoch": 0.2615713377872461, "grad_norm": 3.270859718322754, "learning_rate": 1.731751812845124e-05, "loss": 0.18236262798309327, "step": 9930 }, { "epoch": 0.2618347530317448, "grad_norm": 2.0402326583862305, "learning_rate": 1.7311700728381486e-05, "loss": 0.15771994590759278, "step": 9940 }, { "epoch": 0.26209816827624355, "grad_norm": 2.8205556869506836, "learning_rate": 1.7305878006540483e-05, "loss": 0.16296851634979248, "step": 9950 }, { "epoch": 0.2623615835207423, "grad_norm": 1.533987283706665, "learning_rate": 1.7300049967166273e-05, "loss": 0.16257330179214477, "step": 9960 }, { "epoch": 0.26262499876524104, "grad_norm": 1.5356184244155884, "learning_rate": 1.729421661450074e-05, "loss": 0.156392240524292, "step": 9970 }, { "epoch": 0.2628884140097398, "grad_norm": 2.157177209854126, "learning_rate": 1.728837795278966e-05, "loss": 0.1612519860267639, "step": 9980 }, { "epoch": 0.2631518292542385, "grad_norm": 3.792165994644165, "learning_rate": 1.7282533986282656e-05, "loss": 0.17110512256622315, "step": 9990 }, { "epoch": 0.26341524449873727, "grad_norm": 9.662402153015137, "learning_rate": 1.7276684719233222e-05, "loss": 0.18094757795333863, "step": 10000 }, { "epoch": 0.263678659743236, "grad_norm": 2.463265895843506, "learning_rate": 1.7270830155898704e-05, "loss": 0.1861622452735901, "step": 10010 }, { "epoch": 0.26394207498773475, "grad_norm": 1.0251437425613403, "learning_rate": 1.726497030054031e-05, "loss": 0.16782608032226562, "step": 10020 }, { "epoch": 0.26420549023223344, "grad_norm": 0.9864535927772522, "learning_rate": 1.72591051574231e-05, "loss": 0.15789003372192384, "step": 10030 }, { "epoch": 0.2644689054767322, "grad_norm": 2.324613094329834, "learning_rate": 1.7253234730815964e-05, "loss": 0.17872658967971802, "step": 10040 }, { "epoch": 0.26473232072123093, "grad_norm": 1.3559833765029907, "learning_rate": 1.724735902499166e-05, "loss": 0.17475378513336182, "step": 10050 }, { "epoch": 0.2649957359657297, "grad_norm": 0.9344319701194763, "learning_rate": 1.7241478044226782e-05, "loss": 0.16647578477859498, "step": 10060 }, { "epoch": 0.2652591512102284, "grad_norm": 1.2698625326156616, "learning_rate": 1.723559179280176e-05, "loss": 0.16791088581085206, "step": 10070 }, { "epoch": 0.26552256645472716, "grad_norm": 2.06083345413208, "learning_rate": 1.722970027500086e-05, "loss": 0.17773219347000122, "step": 10080 }, { "epoch": 0.2657859816992259, "grad_norm": 4.017472267150879, "learning_rate": 1.7223803495112185e-05, "loss": 0.16584922075271608, "step": 10090 }, { "epoch": 0.26604939694372465, "grad_norm": 1.3952457904815674, "learning_rate": 1.7217901457427667e-05, "loss": 0.1523995041847229, "step": 10100 }, { "epoch": 0.26631281218822334, "grad_norm": 4.141232967376709, "learning_rate": 1.7211994166243058e-05, "loss": 0.17194281816482543, "step": 10110 }, { "epoch": 0.2665762274327221, "grad_norm": 1.740613579750061, "learning_rate": 1.7206081625857943e-05, "loss": 0.17476009130477904, "step": 10120 }, { "epoch": 0.2668396426772208, "grad_norm": 3.2740638256073, "learning_rate": 1.7200163840575728e-05, "loss": 0.16696653366088868, "step": 10130 }, { "epoch": 0.26710305792171957, "grad_norm": 2.773113965988159, "learning_rate": 1.7194240814703626e-05, "loss": 0.16969016790390015, "step": 10140 }, { "epoch": 0.2673664731662183, "grad_norm": 1.559554934501648, "learning_rate": 1.7188312552552672e-05, "loss": 0.17069804668426514, "step": 10150 }, { "epoch": 0.26762988841071705, "grad_norm": 3.1862192153930664, "learning_rate": 1.718237905843771e-05, "loss": 0.16690871715545655, "step": 10160 }, { "epoch": 0.2678933036552158, "grad_norm": 3.125011682510376, "learning_rate": 1.71764403366774e-05, "loss": 0.17268149852752684, "step": 10170 }, { "epoch": 0.26815671889971454, "grad_norm": 3.386645555496216, "learning_rate": 1.7170496391594194e-05, "loss": 0.15923668146133424, "step": 10180 }, { "epoch": 0.2684201341442133, "grad_norm": 1.8616093397140503, "learning_rate": 1.716454722751435e-05, "loss": 0.16415295600891114, "step": 10190 }, { "epoch": 0.268683549388712, "grad_norm": 2.35764741897583, "learning_rate": 1.715859284876793e-05, "loss": 0.16594960689544677, "step": 10200 }, { "epoch": 0.2689469646332107, "grad_norm": 0.8774339556694031, "learning_rate": 1.7152633259688788e-05, "loss": 0.15964117050170898, "step": 10210 }, { "epoch": 0.26921037987770946, "grad_norm": 5.7230224609375, "learning_rate": 1.714666846461457e-05, "loss": 0.16373023986816407, "step": 10220 }, { "epoch": 0.2694737951222082, "grad_norm": 0.9557089805603027, "learning_rate": 1.7140698467886707e-05, "loss": 0.15815045833587646, "step": 10230 }, { "epoch": 0.26973721036670695, "grad_norm": 1.223046064376831, "learning_rate": 1.713472327385042e-05, "loss": 0.16388497352600098, "step": 10240 }, { "epoch": 0.2700006256112057, "grad_norm": 2.4007933139801025, "learning_rate": 1.7128742886854723e-05, "loss": 0.15350841283798217, "step": 10250 }, { "epoch": 0.27026404085570443, "grad_norm": 2.2684929370880127, "learning_rate": 1.712275731125239e-05, "loss": 0.1734294056892395, "step": 10260 }, { "epoch": 0.2705274561002032, "grad_norm": 6.932127952575684, "learning_rate": 1.7116766551399986e-05, "loss": 0.16599090099334718, "step": 10270 }, { "epoch": 0.2707908713447019, "grad_norm": 1.6701183319091797, "learning_rate": 1.711077061165784e-05, "loss": 0.16645511388778686, "step": 10280 }, { "epoch": 0.2710542865892006, "grad_norm": 4.436915397644043, "learning_rate": 1.710476949639006e-05, "loss": 0.17370054721832276, "step": 10290 }, { "epoch": 0.27131770183369935, "grad_norm": 2.3751323223114014, "learning_rate": 1.7098763209964514e-05, "loss": 0.15643608570098877, "step": 10300 }, { "epoch": 0.2715811170781981, "grad_norm": 2.3903591632843018, "learning_rate": 1.7092751756752838e-05, "loss": 0.15645062923431396, "step": 10310 }, { "epoch": 0.27184453232269684, "grad_norm": 1.7601816654205322, "learning_rate": 1.7086735141130423e-05, "loss": 0.17030799388885498, "step": 10320 }, { "epoch": 0.2721079475671956, "grad_norm": 3.421687602996826, "learning_rate": 1.7080713367476428e-05, "loss": 0.16177115440368653, "step": 10330 }, { "epoch": 0.27237136281169433, "grad_norm": 0.6314657330513, "learning_rate": 1.7074686440173752e-05, "loss": 0.17154257297515868, "step": 10340 }, { "epoch": 0.2726347780561931, "grad_norm": 2.2178354263305664, "learning_rate": 1.7068654363609058e-05, "loss": 0.15417032241821288, "step": 10350 }, { "epoch": 0.2728981933006918, "grad_norm": 2.045485496520996, "learning_rate": 1.7062617142172754e-05, "loss": 0.17708985805511473, "step": 10360 }, { "epoch": 0.2731616085451905, "grad_norm": 2.5167553424835205, "learning_rate": 1.7056574780258986e-05, "loss": 0.1660929560661316, "step": 10370 }, { "epoch": 0.27342502378968925, "grad_norm": 2.0877315998077393, "learning_rate": 1.705052728226564e-05, "loss": 0.15708762407302856, "step": 10380 }, { "epoch": 0.273688439034188, "grad_norm": 1.8743776082992554, "learning_rate": 1.704447465259436e-05, "loss": 0.1594928503036499, "step": 10390 }, { "epoch": 0.27395185427868674, "grad_norm": 3.1458418369293213, "learning_rate": 1.7038416895650504e-05, "loss": 0.16506197452545165, "step": 10400 }, { "epoch": 0.2742152695231855, "grad_norm": 2.578489065170288, "learning_rate": 1.7032354015843162e-05, "loss": 0.17167023420333863, "step": 10410 }, { "epoch": 0.2744786847676842, "grad_norm": 4.831424236297607, "learning_rate": 1.7026286017585168e-05, "loss": 0.15907952785491944, "step": 10420 }, { "epoch": 0.27474210001218297, "grad_norm": 1.7243436574935913, "learning_rate": 1.7020212905293072e-05, "loss": 0.17183616161346435, "step": 10430 }, { "epoch": 0.2750055152566817, "grad_norm": 1.524453043937683, "learning_rate": 1.7014134683387147e-05, "loss": 0.16989479064941407, "step": 10440 }, { "epoch": 0.27526893050118045, "grad_norm": 1.5447585582733154, "learning_rate": 1.7008051356291382e-05, "loss": 0.15674877166748047, "step": 10450 }, { "epoch": 0.27553234574567914, "grad_norm": 1.6481207609176636, "learning_rate": 1.7001962928433488e-05, "loss": 0.18048523664474486, "step": 10460 }, { "epoch": 0.2757957609901779, "grad_norm": 2.52713942527771, "learning_rate": 1.699586940424488e-05, "loss": 0.17879793643951417, "step": 10470 }, { "epoch": 0.27605917623467663, "grad_norm": 1.4268134832382202, "learning_rate": 1.6989770788160697e-05, "loss": 0.1709039807319641, "step": 10480 }, { "epoch": 0.2763225914791754, "grad_norm": 4.5503339767456055, "learning_rate": 1.6983667084619768e-05, "loss": 0.1748952865600586, "step": 10490 }, { "epoch": 0.2765860067236741, "grad_norm": 2.258347749710083, "learning_rate": 1.6977558298064636e-05, "loss": 0.16356425285339354, "step": 10500 }, { "epoch": 0.27684942196817286, "grad_norm": 2.6058590412139893, "learning_rate": 1.6971444432941533e-05, "loss": 0.15933059453964232, "step": 10510 }, { "epoch": 0.2771128372126716, "grad_norm": 1.553292155265808, "learning_rate": 1.69653254937004e-05, "loss": 0.16664440631866456, "step": 10520 }, { "epoch": 0.27737625245717035, "grad_norm": 1.4316277503967285, "learning_rate": 1.6959201484794867e-05, "loss": 0.16223397254943847, "step": 10530 }, { "epoch": 0.2776396677016691, "grad_norm": 2.2402100563049316, "learning_rate": 1.6953072410682243e-05, "loss": 0.17723779678344725, "step": 10540 }, { "epoch": 0.2779030829461678, "grad_norm": 3.7819724082946777, "learning_rate": 1.694693827582354e-05, "loss": 0.17454977035522462, "step": 10550 }, { "epoch": 0.2781664981906665, "grad_norm": 2.408108711242676, "learning_rate": 1.694079908468344e-05, "loss": 0.15973111391067504, "step": 10560 }, { "epoch": 0.27842991343516527, "grad_norm": 2.7237308025360107, "learning_rate": 1.6934654841730318e-05, "loss": 0.16996054649353026, "step": 10570 }, { "epoch": 0.278693328679664, "grad_norm": 1.4300963878631592, "learning_rate": 1.6928505551436215e-05, "loss": 0.17241983413696288, "step": 10580 }, { "epoch": 0.27895674392416275, "grad_norm": 2.8705921173095703, "learning_rate": 1.6922351218276852e-05, "loss": 0.16380054950714112, "step": 10590 }, { "epoch": 0.2792201591686615, "grad_norm": 3.316056489944458, "learning_rate": 1.6916191846731615e-05, "loss": 0.15659732818603517, "step": 10600 }, { "epoch": 0.27948357441316024, "grad_norm": 3.062025308609009, "learning_rate": 1.6910027441283562e-05, "loss": 0.16965744495391846, "step": 10610 }, { "epoch": 0.279746989657659, "grad_norm": 2.6317310333251953, "learning_rate": 1.6903858006419413e-05, "loss": 0.17889437675476075, "step": 10620 }, { "epoch": 0.2800104049021577, "grad_norm": 1.8021204471588135, "learning_rate": 1.689768354662955e-05, "loss": 0.16995518207550048, "step": 10630 }, { "epoch": 0.2802738201466564, "grad_norm": 2.2987492084503174, "learning_rate": 1.689150406640801e-05, "loss": 0.16517696380615235, "step": 10640 }, { "epoch": 0.28053723539115516, "grad_norm": 1.594343900680542, "learning_rate": 1.6885319570252485e-05, "loss": 0.16751401424407958, "step": 10650 }, { "epoch": 0.2808006506356539, "grad_norm": 3.488431453704834, "learning_rate": 1.687913006266432e-05, "loss": 0.1582413673400879, "step": 10660 }, { "epoch": 0.28106406588015265, "grad_norm": 3.4193761348724365, "learning_rate": 1.6872935548148503e-05, "loss": 0.16991901397705078, "step": 10670 }, { "epoch": 0.2813274811246514, "grad_norm": 3.3362181186676025, "learning_rate": 1.686673603121367e-05, "loss": 0.1610870599746704, "step": 10680 }, { "epoch": 0.28159089636915013, "grad_norm": 3.7614426612854004, "learning_rate": 1.6860531516372102e-05, "loss": 0.17177704572677613, "step": 10690 }, { "epoch": 0.2818543116136489, "grad_norm": 1.3844425678253174, "learning_rate": 1.6854322008139704e-05, "loss": 0.16159870624542236, "step": 10700 }, { "epoch": 0.2821177268581476, "grad_norm": 0.8879409432411194, "learning_rate": 1.6848107511036026e-05, "loss": 0.16951370239257812, "step": 10710 }, { "epoch": 0.2823811421026463, "grad_norm": 2.518427848815918, "learning_rate": 1.6841888029584255e-05, "loss": 0.14959726333618165, "step": 10720 }, { "epoch": 0.28264455734714505, "grad_norm": 1.9380420446395874, "learning_rate": 1.6835663568311187e-05, "loss": 0.1480758547782898, "step": 10730 }, { "epoch": 0.2829079725916438, "grad_norm": 1.5657206773757935, "learning_rate": 1.682943413174726e-05, "loss": 0.16793931722640992, "step": 10740 }, { "epoch": 0.28317138783614254, "grad_norm": 1.5475335121154785, "learning_rate": 1.6823199724426523e-05, "loss": 0.17735841274261474, "step": 10750 }, { "epoch": 0.2834348030806413, "grad_norm": 1.612341046333313, "learning_rate": 1.6816960350886653e-05, "loss": 0.1746675968170166, "step": 10760 }, { "epoch": 0.28369821832514003, "grad_norm": 3.204740047454834, "learning_rate": 1.681071601566893e-05, "loss": 0.17530468702316285, "step": 10770 }, { "epoch": 0.28396163356963877, "grad_norm": 1.0662633180618286, "learning_rate": 1.6804466723318248e-05, "loss": 0.17045823335647584, "step": 10780 }, { "epoch": 0.2842250488141375, "grad_norm": 3.981703758239746, "learning_rate": 1.6798212478383118e-05, "loss": 0.16545217037200927, "step": 10790 }, { "epoch": 0.28448846405863626, "grad_norm": 4.008540153503418, "learning_rate": 1.6791953285415645e-05, "loss": 0.16588287353515624, "step": 10800 }, { "epoch": 0.28475187930313495, "grad_norm": 2.4966659545898438, "learning_rate": 1.678568914897154e-05, "loss": 0.16658425331115723, "step": 10810 }, { "epoch": 0.2850152945476337, "grad_norm": 5.874054908752441, "learning_rate": 1.6779420073610116e-05, "loss": 0.16624821424484254, "step": 10820 }, { "epoch": 0.28527870979213243, "grad_norm": 6.891036510467529, "learning_rate": 1.6773146063894273e-05, "loss": 0.18483786582946776, "step": 10830 }, { "epoch": 0.2855421250366312, "grad_norm": 5.735870361328125, "learning_rate": 1.6766867124390505e-05, "loss": 0.16784790754318238, "step": 10840 }, { "epoch": 0.2858055402811299, "grad_norm": 2.910490036010742, "learning_rate": 1.6760583259668897e-05, "loss": 0.16464723348617555, "step": 10850 }, { "epoch": 0.28606895552562867, "grad_norm": 1.466589331626892, "learning_rate": 1.6754294474303116e-05, "loss": 0.16626901626586915, "step": 10860 }, { "epoch": 0.2863323707701274, "grad_norm": 0.9261763095855713, "learning_rate": 1.674800077287041e-05, "loss": 0.17164807319641112, "step": 10870 }, { "epoch": 0.28659578601462615, "grad_norm": 3.3685269355773926, "learning_rate": 1.6741702159951613e-05, "loss": 0.17568838596343994, "step": 10880 }, { "epoch": 0.2868592012591249, "grad_norm": 1.5928702354431152, "learning_rate": 1.6735398640131117e-05, "loss": 0.17363083362579346, "step": 10890 }, { "epoch": 0.2871226165036236, "grad_norm": 1.6628206968307495, "learning_rate": 1.6729090217996904e-05, "loss": 0.16835625171661378, "step": 10900 }, { "epoch": 0.28738603174812233, "grad_norm": 1.7735601663589478, "learning_rate": 1.6722776898140513e-05, "loss": 0.1591097593307495, "step": 10910 }, { "epoch": 0.28764944699262107, "grad_norm": 5.368422508239746, "learning_rate": 1.6716458685157047e-05, "loss": 0.1684829354286194, "step": 10920 }, { "epoch": 0.2879128622371198, "grad_norm": 2.396552324295044, "learning_rate": 1.671013558364518e-05, "loss": 0.1620041847229004, "step": 10930 }, { "epoch": 0.28817627748161856, "grad_norm": 6.948822498321533, "learning_rate": 1.6703807598207132e-05, "loss": 0.1658502221107483, "step": 10940 }, { "epoch": 0.2884396927261173, "grad_norm": 1.2832229137420654, "learning_rate": 1.669747473344869e-05, "loss": 0.15943857431411743, "step": 10950 }, { "epoch": 0.28870310797061605, "grad_norm": 3.720066547393799, "learning_rate": 1.6691136993979183e-05, "loss": 0.151045823097229, "step": 10960 }, { "epoch": 0.2889665232151148, "grad_norm": 1.5854593515396118, "learning_rate": 1.6684794384411494e-05, "loss": 0.15304117202758788, "step": 10970 }, { "epoch": 0.2892299384596135, "grad_norm": 1.6544697284698486, "learning_rate": 1.667844690936204e-05, "loss": 0.1650315761566162, "step": 10980 }, { "epoch": 0.2894933537041122, "grad_norm": 2.7223968505859375, "learning_rate": 1.6672094573450798e-05, "loss": 0.18572217226028442, "step": 10990 }, { "epoch": 0.28975676894861097, "grad_norm": 1.1482141017913818, "learning_rate": 1.6665737381301263e-05, "loss": 0.1702268362045288, "step": 11000 }, { "epoch": 0.2900201841931097, "grad_norm": 1.4379786252975464, "learning_rate": 1.665937533754048e-05, "loss": 0.1564648389816284, "step": 11010 }, { "epoch": 0.29028359943760845, "grad_norm": 1.1702214479446411, "learning_rate": 1.6653008446799014e-05, "loss": 0.1701132535934448, "step": 11020 }, { "epoch": 0.2905470146821072, "grad_norm": 0.8808426260948181, "learning_rate": 1.6646636713710967e-05, "loss": 0.15764005184173585, "step": 11030 }, { "epoch": 0.29081042992660594, "grad_norm": 3.570423126220703, "learning_rate": 1.664026014291396e-05, "loss": 0.1714759111404419, "step": 11040 }, { "epoch": 0.2910738451711047, "grad_norm": 4.1624016761779785, "learning_rate": 1.6633878739049135e-05, "loss": 0.16614744663238526, "step": 11050 }, { "epoch": 0.2913372604156034, "grad_norm": 2.552229881286621, "learning_rate": 1.6627492506761155e-05, "loss": 0.1662541151046753, "step": 11060 }, { "epoch": 0.2916006756601021, "grad_norm": 1.6524015665054321, "learning_rate": 1.6621101450698193e-05, "loss": 0.17416653633117676, "step": 11070 }, { "epoch": 0.29186409090460086, "grad_norm": 3.1373860836029053, "learning_rate": 1.6614705575511937e-05, "loss": 0.1756231188774109, "step": 11080 }, { "epoch": 0.2921275061490996, "grad_norm": 3.0223309993743896, "learning_rate": 1.660830488585758e-05, "loss": 0.15916659832000732, "step": 11090 }, { "epoch": 0.29239092139359835, "grad_norm": 1.880226492881775, "learning_rate": 1.660189938639382e-05, "loss": 0.15877662897109984, "step": 11100 }, { "epoch": 0.2926543366380971, "grad_norm": 1.1866347789764404, "learning_rate": 1.6595489081782858e-05, "loss": 0.16774023771286012, "step": 11110 }, { "epoch": 0.29291775188259583, "grad_norm": 0.9645969271659851, "learning_rate": 1.6589073976690387e-05, "loss": 0.17319231033325194, "step": 11120 }, { "epoch": 0.2931811671270946, "grad_norm": 2.8338592052459717, "learning_rate": 1.65826540757856e-05, "loss": 0.1642634630203247, "step": 11130 }, { "epoch": 0.2934445823715933, "grad_norm": 1.4441312551498413, "learning_rate": 1.6576229383741177e-05, "loss": 0.16907191276550293, "step": 11140 }, { "epoch": 0.29370799761609206, "grad_norm": 1.637696385383606, "learning_rate": 1.6569799905233285e-05, "loss": 0.16196587085723876, "step": 11150 }, { "epoch": 0.29397141286059075, "grad_norm": 1.636794090270996, "learning_rate": 1.6563365644941578e-05, "loss": 0.1649841547012329, "step": 11160 }, { "epoch": 0.2942348281050895, "grad_norm": 3.1871559619903564, "learning_rate": 1.6556926607549186e-05, "loss": 0.16223729848861695, "step": 11170 }, { "epoch": 0.29449824334958824, "grad_norm": 2.739184617996216, "learning_rate": 1.655048279774272e-05, "loss": 0.17253656387329103, "step": 11180 }, { "epoch": 0.294761658594087, "grad_norm": 2.1970651149749756, "learning_rate": 1.6544034220212262e-05, "loss": 0.1640394449234009, "step": 11190 }, { "epoch": 0.2950250738385857, "grad_norm": 0.9840145707130432, "learning_rate": 1.6537580879651364e-05, "loss": 0.16038234233856202, "step": 11200 }, { "epoch": 0.29528848908308447, "grad_norm": 2.120908260345459, "learning_rate": 1.6531122780757046e-05, "loss": 0.17105038166046144, "step": 11210 }, { "epoch": 0.2955519043275832, "grad_norm": 3.872670888900757, "learning_rate": 1.6524659928229792e-05, "loss": 0.17874484062194823, "step": 11220 }, { "epoch": 0.29581531957208196, "grad_norm": 4.961794853210449, "learning_rate": 1.6518192326773546e-05, "loss": 0.17859985828399658, "step": 11230 }, { "epoch": 0.29607873481658065, "grad_norm": 2.520573139190674, "learning_rate": 1.6511719981095703e-05, "loss": 0.17820777893066406, "step": 11240 }, { "epoch": 0.2963421500610794, "grad_norm": 3.295724630355835, "learning_rate": 1.6505242895907122e-05, "loss": 0.16529159545898436, "step": 11250 }, { "epoch": 0.29660556530557813, "grad_norm": 2.9249987602233887, "learning_rate": 1.64987610759221e-05, "loss": 0.16650917530059814, "step": 11260 }, { "epoch": 0.2968689805500769, "grad_norm": 1.8429811000823975, "learning_rate": 1.6492274525858386e-05, "loss": 0.16741855144500734, "step": 11270 }, { "epoch": 0.2971323957945756, "grad_norm": 1.7344787120819092, "learning_rate": 1.648578325043717e-05, "loss": 0.16229608058929443, "step": 11280 }, { "epoch": 0.29739581103907436, "grad_norm": 0.7857820987701416, "learning_rate": 1.6479287254383084e-05, "loss": 0.17534786462783813, "step": 11290 }, { "epoch": 0.2976592262835731, "grad_norm": 2.227970600128174, "learning_rate": 1.6472786542424196e-05, "loss": 0.15018335580825806, "step": 11300 }, { "epoch": 0.29792264152807185, "grad_norm": 2.7598910331726074, "learning_rate": 1.6466281119292004e-05, "loss": 0.17270386219024658, "step": 11310 }, { "epoch": 0.2981860567725706, "grad_norm": 1.8089276552200317, "learning_rate": 1.6459770989721433e-05, "loss": 0.17201974391937255, "step": 11320 }, { "epoch": 0.2984494720170693, "grad_norm": 3.8647379875183105, "learning_rate": 1.6453256158450835e-05, "loss": 0.16401461362838746, "step": 11330 }, { "epoch": 0.298712887261568, "grad_norm": 1.1882694959640503, "learning_rate": 1.644673663022199e-05, "loss": 0.16705315113067626, "step": 11340 }, { "epoch": 0.29897630250606677, "grad_norm": 2.8321380615234375, "learning_rate": 1.6440212409780088e-05, "loss": 0.16427735090255738, "step": 11350 }, { "epoch": 0.2992397177505655, "grad_norm": 0.9519067406654358, "learning_rate": 1.643368350187374e-05, "loss": 0.1765391707420349, "step": 11360 }, { "epoch": 0.29950313299506426, "grad_norm": 2.8897175788879395, "learning_rate": 1.6427149911254973e-05, "loss": 0.159765625, "step": 11370 }, { "epoch": 0.299766548239563, "grad_norm": 2.0516347885131836, "learning_rate": 1.64206116426792e-05, "loss": 0.15683337450027465, "step": 11380 }, { "epoch": 0.30002996348406175, "grad_norm": 2.2750277519226074, "learning_rate": 1.641406870090527e-05, "loss": 0.1641973853111267, "step": 11390 }, { "epoch": 0.3002933787285605, "grad_norm": 1.4649454355239868, "learning_rate": 1.640752109069541e-05, "loss": 0.16863577365875243, "step": 11400 }, { "epoch": 0.30055679397305923, "grad_norm": 0.8883000612258911, "learning_rate": 1.640096881681526e-05, "loss": 0.1679812788963318, "step": 11410 }, { "epoch": 0.3008202092175579, "grad_norm": 2.660835027694702, "learning_rate": 1.6394411884033844e-05, "loss": 0.17192364931106568, "step": 11420 }, { "epoch": 0.30108362446205666, "grad_norm": 1.13938570022583, "learning_rate": 1.638785029712358e-05, "loss": 0.1633521795272827, "step": 11430 }, { "epoch": 0.3013470397065554, "grad_norm": 2.7785675525665283, "learning_rate": 1.638128406086027e-05, "loss": 0.1753305435180664, "step": 11440 }, { "epoch": 0.30161045495105415, "grad_norm": 1.1789990663528442, "learning_rate": 1.6374713180023114e-05, "loss": 0.16012831926345825, "step": 11450 }, { "epoch": 0.3018738701955529, "grad_norm": 1.1380728483200073, "learning_rate": 1.6368137659394673e-05, "loss": 0.16983939409255983, "step": 11460 }, { "epoch": 0.30213728544005164, "grad_norm": 3.4474782943725586, "learning_rate": 1.63615575037609e-05, "loss": 0.1802208185195923, "step": 11470 }, { "epoch": 0.3024007006845504, "grad_norm": 5.435559272766113, "learning_rate": 1.6354972717911115e-05, "loss": 0.14973697662353516, "step": 11480 }, { "epoch": 0.3026641159290491, "grad_norm": 4.010959148406982, "learning_rate": 1.634838330663801e-05, "loss": 0.1659401535987854, "step": 11490 }, { "epoch": 0.3029275311735478, "grad_norm": 2.209268569946289, "learning_rate": 1.6341789274737644e-05, "loss": 0.16649816036224366, "step": 11500 }, { "epoch": 0.30319094641804656, "grad_norm": 1.8336012363433838, "learning_rate": 1.6335190627009432e-05, "loss": 0.17964648008346557, "step": 11510 }, { "epoch": 0.3034543616625453, "grad_norm": 4.587643623352051, "learning_rate": 1.632858736825616e-05, "loss": 0.16033486127853394, "step": 11520 }, { "epoch": 0.30371777690704405, "grad_norm": 1.6194956302642822, "learning_rate": 1.632197950328396e-05, "loss": 0.15624479055404664, "step": 11530 }, { "epoch": 0.3039811921515428, "grad_norm": 3.0278844833374023, "learning_rate": 1.631536703690233e-05, "loss": 0.16515434980392457, "step": 11540 }, { "epoch": 0.30424460739604153, "grad_norm": 1.973886251449585, "learning_rate": 1.6308749973924096e-05, "loss": 0.16819705963134765, "step": 11550 }, { "epoch": 0.3045080226405403, "grad_norm": 1.6494299173355103, "learning_rate": 1.6302128319165454e-05, "loss": 0.17658408880233764, "step": 11560 }, { "epoch": 0.304771437885039, "grad_norm": 2.307959794998169, "learning_rate": 1.629550207744592e-05, "loss": 0.17409952878952026, "step": 11570 }, { "epoch": 0.30503485312953776, "grad_norm": 1.1694875955581665, "learning_rate": 1.6288871253588366e-05, "loss": 0.16727542877197266, "step": 11580 }, { "epoch": 0.30529826837403645, "grad_norm": 4.605657577514648, "learning_rate": 1.6282235852418983e-05, "loss": 0.17723958492279052, "step": 11590 }, { "epoch": 0.3055616836185352, "grad_norm": 2.1601459980010986, "learning_rate": 1.6275595878767315e-05, "loss": 0.16574196815490722, "step": 11600 }, { "epoch": 0.30582509886303394, "grad_norm": 1.7819843292236328, "learning_rate": 1.6268951337466212e-05, "loss": 0.16439999341964723, "step": 11610 }, { "epoch": 0.3060885141075327, "grad_norm": 1.297773003578186, "learning_rate": 1.626230223335186e-05, "loss": 0.17566540241241455, "step": 11620 }, { "epoch": 0.3063519293520314, "grad_norm": 1.2043445110321045, "learning_rate": 1.6255648571263762e-05, "loss": 0.17310492992401122, "step": 11630 }, { "epoch": 0.30661534459653017, "grad_norm": 3.3443987369537354, "learning_rate": 1.6248990356044742e-05, "loss": 0.17639741897583008, "step": 11640 }, { "epoch": 0.3068787598410289, "grad_norm": 2.0069499015808105, "learning_rate": 1.6242327592540942e-05, "loss": 0.15682785511016845, "step": 11650 }, { "epoch": 0.30714217508552766, "grad_norm": 1.4340468645095825, "learning_rate": 1.62356602856018e-05, "loss": 0.16861940622329713, "step": 11660 }, { "epoch": 0.3074055903300264, "grad_norm": 1.569976568222046, "learning_rate": 1.6228988440080068e-05, "loss": 0.16197431087493896, "step": 11670 }, { "epoch": 0.3076690055745251, "grad_norm": 3.0879485607147217, "learning_rate": 1.6222312060831814e-05, "loss": 0.1553718090057373, "step": 11680 }, { "epoch": 0.30793242081902383, "grad_norm": 3.3235435485839844, "learning_rate": 1.6215631152716387e-05, "loss": 0.1709826946258545, "step": 11690 }, { "epoch": 0.3081958360635226, "grad_norm": 2.0740602016448975, "learning_rate": 1.620894572059644e-05, "loss": 0.17799601554870606, "step": 11700 }, { "epoch": 0.3084592513080213, "grad_norm": 1.546289324760437, "learning_rate": 1.6202255769337924e-05, "loss": 0.15421252250671386, "step": 11710 }, { "epoch": 0.30872266655252006, "grad_norm": 1.43673837184906, "learning_rate": 1.619556130381007e-05, "loss": 0.16762034893035888, "step": 11720 }, { "epoch": 0.3089860817970188, "grad_norm": 2.0670626163482666, "learning_rate": 1.61888623288854e-05, "loss": 0.17886171340942383, "step": 11730 }, { "epoch": 0.30924949704151755, "grad_norm": 2.517134666442871, "learning_rate": 1.6182158849439718e-05, "loss": 0.15904793739318848, "step": 11740 }, { "epoch": 0.3095129122860163, "grad_norm": 1.938016653060913, "learning_rate": 1.61754508703521e-05, "loss": 0.1656235098838806, "step": 11750 }, { "epoch": 0.309776327530515, "grad_norm": 1.8861347436904907, "learning_rate": 1.6168738396504912e-05, "loss": 0.17167586088180542, "step": 11760 }, { "epoch": 0.3100397427750137, "grad_norm": 3.413760185241699, "learning_rate": 1.6162021432783773e-05, "loss": 0.16690291166305543, "step": 11770 }, { "epoch": 0.31030315801951247, "grad_norm": 2.9643819332122803, "learning_rate": 1.6155299984077584e-05, "loss": 0.16369855403900146, "step": 11780 }, { "epoch": 0.3105665732640112, "grad_norm": 0.9605616331100464, "learning_rate": 1.6148574055278507e-05, "loss": 0.13977476358413696, "step": 11790 }, { "epoch": 0.31082998850850996, "grad_norm": 2.3137426376342773, "learning_rate": 1.6141843651281955e-05, "loss": 0.15686485767364503, "step": 11800 }, { "epoch": 0.3110934037530087, "grad_norm": 2.598907232284546, "learning_rate": 1.6135108776986615e-05, "loss": 0.17445592880249022, "step": 11810 }, { "epoch": 0.31135681899750745, "grad_norm": 1.3369842767715454, "learning_rate": 1.612836943729442e-05, "loss": 0.14936201572418212, "step": 11820 }, { "epoch": 0.3116202342420062, "grad_norm": 1.8779966831207275, "learning_rate": 1.6121625637110546e-05, "loss": 0.1712444543838501, "step": 11830 }, { "epoch": 0.31188364948650493, "grad_norm": 1.5209234952926636, "learning_rate": 1.6114877381343422e-05, "loss": 0.1596410632133484, "step": 11840 }, { "epoch": 0.3121470647310036, "grad_norm": 1.0573532581329346, "learning_rate": 1.610812467490472e-05, "loss": 0.17529237270355225, "step": 11850 }, { "epoch": 0.31241047997550236, "grad_norm": 2.8698391914367676, "learning_rate": 1.6101367522709356e-05, "loss": 0.15633950233459473, "step": 11860 }, { "epoch": 0.3126738952200011, "grad_norm": 2.259204387664795, "learning_rate": 1.6094605929675476e-05, "loss": 0.15159796476364135, "step": 11870 }, { "epoch": 0.31293731046449985, "grad_norm": 1.825177788734436, "learning_rate": 1.608783990072445e-05, "loss": 0.16404392719268798, "step": 11880 }, { "epoch": 0.3132007257089986, "grad_norm": 4.292430400848389, "learning_rate": 1.6081069440780895e-05, "loss": 0.17048494815826415, "step": 11890 }, { "epoch": 0.31346414095349734, "grad_norm": 2.3494980335235596, "learning_rate": 1.607429455477264e-05, "loss": 0.15285369157791137, "step": 11900 }, { "epoch": 0.3137275561979961, "grad_norm": 0.8025835156440735, "learning_rate": 1.606751524763075e-05, "loss": 0.15025805234909057, "step": 11910 }, { "epoch": 0.3139909714424948, "grad_norm": 0.9215836524963379, "learning_rate": 1.606073152428948e-05, "loss": 0.16244964599609374, "step": 11920 }, { "epoch": 0.31425438668699357, "grad_norm": 2.39138126373291, "learning_rate": 1.6053943389686323e-05, "loss": 0.16149390935897828, "step": 11930 }, { "epoch": 0.31451780193149226, "grad_norm": 2.5323266983032227, "learning_rate": 1.604715084876198e-05, "loss": 0.16290384531021118, "step": 11940 }, { "epoch": 0.314781217175991, "grad_norm": 1.984793782234192, "learning_rate": 1.6040353906460353e-05, "loss": 0.17320513725280762, "step": 11950 }, { "epoch": 0.31504463242048975, "grad_norm": 1.6017067432403564, "learning_rate": 1.6033552567728553e-05, "loss": 0.15437251329421997, "step": 11960 }, { "epoch": 0.3153080476649885, "grad_norm": 1.1393630504608154, "learning_rate": 1.602674683751688e-05, "loss": 0.16981351375579834, "step": 11970 }, { "epoch": 0.31557146290948723, "grad_norm": 2.0415287017822266, "learning_rate": 1.6019936720778845e-05, "loss": 0.16682512760162355, "step": 11980 }, { "epoch": 0.315834878153986, "grad_norm": 2.176375389099121, "learning_rate": 1.601312222247114e-05, "loss": 0.16368227005004882, "step": 11990 }, { "epoch": 0.3160982933984847, "grad_norm": 1.6590533256530762, "learning_rate": 1.6006303347553655e-05, "loss": 0.17204928398132324, "step": 12000 }, { "epoch": 0.31636170864298346, "grad_norm": 1.5428056716918945, "learning_rate": 1.5999480100989458e-05, "loss": 0.16829333305358887, "step": 12010 }, { "epoch": 0.31662512388748215, "grad_norm": 2.9498519897460938, "learning_rate": 1.5992652487744806e-05, "loss": 0.16917670965194703, "step": 12020 }, { "epoch": 0.3168885391319809, "grad_norm": 1.0504101514816284, "learning_rate": 1.5985820512789128e-05, "loss": 0.1550315260887146, "step": 12030 }, { "epoch": 0.31715195437647964, "grad_norm": 1.9375141859054565, "learning_rate": 1.5978984181095034e-05, "loss": 0.1718536138534546, "step": 12040 }, { "epoch": 0.3174153696209784, "grad_norm": 1.198528528213501, "learning_rate": 1.5972143497638294e-05, "loss": 0.1683253288269043, "step": 12050 }, { "epoch": 0.3176787848654771, "grad_norm": 1.793599247932434, "learning_rate": 1.5965298467397855e-05, "loss": 0.15358256101608275, "step": 12060 }, { "epoch": 0.31794220010997587, "grad_norm": 1.1564805507659912, "learning_rate": 1.595844909535583e-05, "loss": 0.15445624589920043, "step": 12070 }, { "epoch": 0.3182056153544746, "grad_norm": 0.8654870390892029, "learning_rate": 1.595159538649748e-05, "loss": 0.16674211025238037, "step": 12080 }, { "epoch": 0.31846903059897336, "grad_norm": 4.093878746032715, "learning_rate": 1.5944737345811238e-05, "loss": 0.1729595422744751, "step": 12090 }, { "epoch": 0.3187324458434721, "grad_norm": 0.6815077662467957, "learning_rate": 1.5937874978288676e-05, "loss": 0.15999040603637696, "step": 12100 }, { "epoch": 0.3189958610879708, "grad_norm": 1.0719542503356934, "learning_rate": 1.5931008288924523e-05, "loss": 0.15814632177352905, "step": 12110 }, { "epoch": 0.31925927633246953, "grad_norm": 1.1723014116287231, "learning_rate": 1.5924137282716653e-05, "loss": 0.15637754201889037, "step": 12120 }, { "epoch": 0.3195226915769683, "grad_norm": 1.5247602462768555, "learning_rate": 1.591726196466608e-05, "loss": 0.17007763385772706, "step": 12130 }, { "epoch": 0.319786106821467, "grad_norm": 1.355601191520691, "learning_rate": 1.5910382339776953e-05, "loss": 0.16177010536193848, "step": 12140 }, { "epoch": 0.32004952206596576, "grad_norm": 3.1556859016418457, "learning_rate": 1.5903498413056568e-05, "loss": 0.16580312252044677, "step": 12150 }, { "epoch": 0.3203129373104645, "grad_norm": 2.0224921703338623, "learning_rate": 1.5896610189515335e-05, "loss": 0.1628035306930542, "step": 12160 }, { "epoch": 0.32057635255496325, "grad_norm": 2.036090135574341, "learning_rate": 1.588971767416681e-05, "loss": 0.1642577290534973, "step": 12170 }, { "epoch": 0.320839767799462, "grad_norm": 4.854579925537109, "learning_rate": 1.5882820872027654e-05, "loss": 0.16824254989624024, "step": 12180 }, { "epoch": 0.32110318304396074, "grad_norm": 0.7630724310874939, "learning_rate": 1.5875919788117662e-05, "loss": 0.16828675270080568, "step": 12190 }, { "epoch": 0.3213665982884594, "grad_norm": 1.0165163278579712, "learning_rate": 1.586901442745974e-05, "loss": 0.1643362045288086, "step": 12200 }, { "epoch": 0.32163001353295817, "grad_norm": 2.2256600856781006, "learning_rate": 1.5862104795079903e-05, "loss": 0.15278377532958984, "step": 12210 }, { "epoch": 0.3218934287774569, "grad_norm": 1.0451489686965942, "learning_rate": 1.5855190896007287e-05, "loss": 0.18001521825790406, "step": 12220 }, { "epoch": 0.32215684402195566, "grad_norm": 0.9287758469581604, "learning_rate": 1.584827273527412e-05, "loss": 0.14663949012756347, "step": 12230 }, { "epoch": 0.3224202592664544, "grad_norm": 3.875478982925415, "learning_rate": 1.5841350317915745e-05, "loss": 0.16286567449569703, "step": 12240 }, { "epoch": 0.32268367451095314, "grad_norm": 0.9049936532974243, "learning_rate": 1.5834423648970588e-05, "loss": 0.1700582504272461, "step": 12250 }, { "epoch": 0.3229470897554519, "grad_norm": 3.1734657287597656, "learning_rate": 1.5827492733480183e-05, "loss": 0.17678778171539306, "step": 12260 }, { "epoch": 0.32321050499995063, "grad_norm": 2.2438361644744873, "learning_rate": 1.5820557576489145e-05, "loss": 0.14856247901916503, "step": 12270 }, { "epoch": 0.3234739202444493, "grad_norm": 1.4331705570220947, "learning_rate": 1.581361818304518e-05, "loss": 0.15927562713623047, "step": 12280 }, { "epoch": 0.32373733548894806, "grad_norm": 1.139947533607483, "learning_rate": 1.5806674558199088e-05, "loss": 0.17220215797424315, "step": 12290 }, { "epoch": 0.3240007507334468, "grad_norm": 1.2826852798461914, "learning_rate": 1.5799726707004728e-05, "loss": 0.17166146039962768, "step": 12300 }, { "epoch": 0.32426416597794555, "grad_norm": 1.8443913459777832, "learning_rate": 1.5792774634519046e-05, "loss": 0.17622745037078857, "step": 12310 }, { "epoch": 0.3245275812224443, "grad_norm": 3.021904945373535, "learning_rate": 1.5785818345802068e-05, "loss": 0.16900023221969604, "step": 12320 }, { "epoch": 0.32479099646694304, "grad_norm": 0.9715909361839294, "learning_rate": 1.5778857845916874e-05, "loss": 0.16198484897613524, "step": 12330 }, { "epoch": 0.3250544117114418, "grad_norm": 1.8678393363952637, "learning_rate": 1.577189313992962e-05, "loss": 0.15906171798706054, "step": 12340 }, { "epoch": 0.3253178269559405, "grad_norm": 1.0725959539413452, "learning_rate": 1.5764924232909518e-05, "loss": 0.17369647026062013, "step": 12350 }, { "epoch": 0.32558124220043927, "grad_norm": 0.7044297456741333, "learning_rate": 1.5757951129928837e-05, "loss": 0.16584162712097167, "step": 12360 }, { "epoch": 0.32584465744493796, "grad_norm": 0.9937907457351685, "learning_rate": 1.5750973836062907e-05, "loss": 0.15703574419021607, "step": 12370 }, { "epoch": 0.3261080726894367, "grad_norm": 3.6677258014678955, "learning_rate": 1.57439923563901e-05, "loss": 0.17479686737060546, "step": 12380 }, { "epoch": 0.32637148793393544, "grad_norm": 1.8211106061935425, "learning_rate": 1.573700669599184e-05, "loss": 0.16125848293304443, "step": 12390 }, { "epoch": 0.3266349031784342, "grad_norm": 1.1857714653015137, "learning_rate": 1.573001685995259e-05, "loss": 0.14981864690780639, "step": 12400 }, { "epoch": 0.32689831842293293, "grad_norm": 3.948453187942505, "learning_rate": 1.572302285335985e-05, "loss": 0.17513349056243896, "step": 12410 }, { "epoch": 0.3271617336674317, "grad_norm": 1.9335097074508667, "learning_rate": 1.5716024681304177e-05, "loss": 0.15942329168319702, "step": 12420 }, { "epoch": 0.3274251489119304, "grad_norm": 2.7765161991119385, "learning_rate": 1.5709022348879124e-05, "loss": 0.1634984493255615, "step": 12430 }, { "epoch": 0.32768856415642916, "grad_norm": 1.479443907737732, "learning_rate": 1.5702015861181295e-05, "loss": 0.16092126369476317, "step": 12440 }, { "epoch": 0.3279519794009279, "grad_norm": 4.188726902008057, "learning_rate": 1.569500522331032e-05, "loss": 0.1673589825630188, "step": 12450 }, { "epoch": 0.3282153946454266, "grad_norm": 5.234379768371582, "learning_rate": 1.568799044036884e-05, "loss": 0.15941561460494996, "step": 12460 }, { "epoch": 0.32847880988992534, "grad_norm": 1.5132001638412476, "learning_rate": 1.5680971517462516e-05, "loss": 0.1476387619972229, "step": 12470 }, { "epoch": 0.3287422251344241, "grad_norm": 3.5965285301208496, "learning_rate": 1.5673948459700024e-05, "loss": 0.1487952709197998, "step": 12480 }, { "epoch": 0.3290056403789228, "grad_norm": 1.821197271347046, "learning_rate": 1.5666921272193046e-05, "loss": 0.16506495475769042, "step": 12490 }, { "epoch": 0.32926905562342157, "grad_norm": 2.3445663452148438, "learning_rate": 1.565988996005628e-05, "loss": 0.1679394245147705, "step": 12500 }, { "epoch": 0.3295324708679203, "grad_norm": 1.73589289188385, "learning_rate": 1.5652854528407407e-05, "loss": 0.15383129119873046, "step": 12510 }, { "epoch": 0.32979588611241906, "grad_norm": 1.248824954032898, "learning_rate": 1.5645814982367125e-05, "loss": 0.1637810468673706, "step": 12520 }, { "epoch": 0.3300593013569178, "grad_norm": 0.7604377865791321, "learning_rate": 1.5638771327059117e-05, "loss": 0.16638951301574706, "step": 12530 }, { "epoch": 0.3303227166014165, "grad_norm": 2.597781181335449, "learning_rate": 1.563172356761006e-05, "loss": 0.16201989650726317, "step": 12540 }, { "epoch": 0.33058613184591523, "grad_norm": 1.1569896936416626, "learning_rate": 1.5624671709149618e-05, "loss": 0.15293629169464112, "step": 12550 }, { "epoch": 0.330849547090414, "grad_norm": 1.6777772903442383, "learning_rate": 1.561761575681044e-05, "loss": 0.15311598777770996, "step": 12560 }, { "epoch": 0.3311129623349127, "grad_norm": 2.036975860595703, "learning_rate": 1.5610555715728146e-05, "loss": 0.15445423126220703, "step": 12570 }, { "epoch": 0.33137637757941146, "grad_norm": 2.2033958435058594, "learning_rate": 1.560349159104134e-05, "loss": 0.16276795864105226, "step": 12580 }, { "epoch": 0.3316397928239102, "grad_norm": 2.744297504425049, "learning_rate": 1.55964233878916e-05, "loss": 0.16350626945495605, "step": 12590 }, { "epoch": 0.33190320806840895, "grad_norm": 0.6225466132164001, "learning_rate": 1.5589351111423474e-05, "loss": 0.16014235019683837, "step": 12600 }, { "epoch": 0.3321666233129077, "grad_norm": 2.2482848167419434, "learning_rate": 1.5582274766784457e-05, "loss": 0.17888777256011962, "step": 12610 }, { "epoch": 0.33243003855740644, "grad_norm": 2.0722544193267822, "learning_rate": 1.557519435912503e-05, "loss": 0.14517589807510375, "step": 12620 }, { "epoch": 0.3326934538019051, "grad_norm": 1.9607303142547607, "learning_rate": 1.556810989359861e-05, "loss": 0.16401169300079346, "step": 12630 }, { "epoch": 0.33295686904640387, "grad_norm": 3.1998708248138428, "learning_rate": 1.5561021375361583e-05, "loss": 0.16806215047836304, "step": 12640 }, { "epoch": 0.3332202842909026, "grad_norm": 2.160473108291626, "learning_rate": 1.5553928809573275e-05, "loss": 0.16415324211120605, "step": 12650 }, { "epoch": 0.33348369953540136, "grad_norm": 1.381251335144043, "learning_rate": 1.5546832201395964e-05, "loss": 0.15883655548095704, "step": 12660 }, { "epoch": 0.3337471147799001, "grad_norm": 1.996829628944397, "learning_rate": 1.5539731555994864e-05, "loss": 0.15967044830322266, "step": 12670 }, { "epoch": 0.33401053002439884, "grad_norm": 2.22542142868042, "learning_rate": 1.5532626878538137e-05, "loss": 0.15729440450668336, "step": 12680 }, { "epoch": 0.3342739452688976, "grad_norm": 3.184938430786133, "learning_rate": 1.5525518174196867e-05, "loss": 0.15604205131530763, "step": 12690 }, { "epoch": 0.33453736051339633, "grad_norm": 2.35019588470459, "learning_rate": 1.551840544814508e-05, "loss": 0.15363293886184692, "step": 12700 }, { "epoch": 0.3348007757578951, "grad_norm": 1.4125927686691284, "learning_rate": 1.551128870555972e-05, "loss": 0.16526037454605103, "step": 12710 }, { "epoch": 0.33506419100239376, "grad_norm": 1.667455792427063, "learning_rate": 1.550416795162067e-05, "loss": 0.16446597576141359, "step": 12720 }, { "epoch": 0.3353276062468925, "grad_norm": 2.7749440670013428, "learning_rate": 1.549704319151071e-05, "loss": 0.16128824949264525, "step": 12730 }, { "epoch": 0.33559102149139125, "grad_norm": 1.2819366455078125, "learning_rate": 1.548991443041555e-05, "loss": 0.16379565000534058, "step": 12740 }, { "epoch": 0.33585443673589, "grad_norm": 2.918961763381958, "learning_rate": 1.5482781673523816e-05, "loss": 0.1673366069793701, "step": 12750 }, { "epoch": 0.33611785198038874, "grad_norm": 2.411257266998291, "learning_rate": 1.5475644926027035e-05, "loss": 0.1538607120513916, "step": 12760 }, { "epoch": 0.3363812672248875, "grad_norm": 3.2800776958465576, "learning_rate": 1.5468504193119636e-05, "loss": 0.1764507293701172, "step": 12770 }, { "epoch": 0.3366446824693862, "grad_norm": 3.464249610900879, "learning_rate": 1.5461359479998953e-05, "loss": 0.15841460227966309, "step": 12780 }, { "epoch": 0.33690809771388497, "grad_norm": 1.627739667892456, "learning_rate": 1.5454210791865213e-05, "loss": 0.16153721809387206, "step": 12790 }, { "epoch": 0.33717151295838366, "grad_norm": 2.762716770172119, "learning_rate": 1.5447058133921552e-05, "loss": 0.1775266170501709, "step": 12800 }, { "epoch": 0.3374349282028824, "grad_norm": 2.648456573486328, "learning_rate": 1.543990151137397e-05, "loss": 0.16653507947921753, "step": 12810 }, { "epoch": 0.33769834344738114, "grad_norm": 1.3806065320968628, "learning_rate": 1.5432740929431372e-05, "loss": 0.14568220376968383, "step": 12820 }, { "epoch": 0.3379617586918799, "grad_norm": 2.9027576446533203, "learning_rate": 1.5425576393305537e-05, "loss": 0.15896351337432862, "step": 12830 }, { "epoch": 0.33822517393637863, "grad_norm": 1.0109001398086548, "learning_rate": 1.5418407908211125e-05, "loss": 0.15789370536804198, "step": 12840 }, { "epoch": 0.3384885891808774, "grad_norm": 2.457829236984253, "learning_rate": 1.5411235479365664e-05, "loss": 0.15394221544265746, "step": 12850 }, { "epoch": 0.3387520044253761, "grad_norm": 1.8230059146881104, "learning_rate": 1.5404059111989565e-05, "loss": 0.1683773159980774, "step": 12860 }, { "epoch": 0.33901541966987486, "grad_norm": 2.4747097492218018, "learning_rate": 1.5396878811306088e-05, "loss": 0.16061921119689943, "step": 12870 }, { "epoch": 0.3392788349143736, "grad_norm": 1.2559101581573486, "learning_rate": 1.5389694582541376e-05, "loss": 0.1655982494354248, "step": 12880 }, { "epoch": 0.3395422501588723, "grad_norm": 1.1445618867874146, "learning_rate": 1.5382506430924414e-05, "loss": 0.16106950044631957, "step": 12890 }, { "epoch": 0.33980566540337104, "grad_norm": 2.027930736541748, "learning_rate": 1.5375314361687053e-05, "loss": 0.15863230228424072, "step": 12900 }, { "epoch": 0.3400690806478698, "grad_norm": 3.5349793434143066, "learning_rate": 1.5368118380063984e-05, "loss": 0.16313310861587524, "step": 12910 }, { "epoch": 0.3403324958923685, "grad_norm": 2.804553747177124, "learning_rate": 1.536091849129276e-05, "loss": 0.16684311628341675, "step": 12920 }, { "epoch": 0.34059591113686727, "grad_norm": 1.791702151298523, "learning_rate": 1.5353714700613768e-05, "loss": 0.16419000625610353, "step": 12930 }, { "epoch": 0.340859326381366, "grad_norm": 4.138360977172852, "learning_rate": 1.5346507013270237e-05, "loss": 0.15577497482299804, "step": 12940 }, { "epoch": 0.34112274162586476, "grad_norm": 3.3401150703430176, "learning_rate": 1.5339295434508238e-05, "loss": 0.1570298194885254, "step": 12950 }, { "epoch": 0.3413861568703635, "grad_norm": 1.5823715925216675, "learning_rate": 1.533207996957666e-05, "loss": 0.15922951698303223, "step": 12960 }, { "epoch": 0.34164957211486224, "grad_norm": 1.6699458360671997, "learning_rate": 1.5324860623727236e-05, "loss": 0.15910356044769286, "step": 12970 }, { "epoch": 0.34191298735936093, "grad_norm": 2.853590250015259, "learning_rate": 1.5317637402214515e-05, "loss": 0.16053136587142944, "step": 12980 }, { "epoch": 0.3421764026038597, "grad_norm": 1.5882517099380493, "learning_rate": 1.5310410310295876e-05, "loss": 0.15166324377059937, "step": 12990 }, { "epoch": 0.3424398178483584, "grad_norm": 2.7369863986968994, "learning_rate": 1.5303179353231494e-05, "loss": 0.1678456425666809, "step": 13000 }, { "epoch": 0.34270323309285716, "grad_norm": 3.5271031856536865, "learning_rate": 1.529594453628439e-05, "loss": 0.15972318649291992, "step": 13010 }, { "epoch": 0.3429666483373559, "grad_norm": 2.009366273880005, "learning_rate": 1.5288705864720356e-05, "loss": 0.1642901659011841, "step": 13020 }, { "epoch": 0.34323006358185465, "grad_norm": 3.1165311336517334, "learning_rate": 1.528146334380803e-05, "loss": 0.18637347221374512, "step": 13030 }, { "epoch": 0.3434934788263534, "grad_norm": 1.993100643157959, "learning_rate": 1.5274216978818813e-05, "loss": 0.1593569278717041, "step": 13040 }, { "epoch": 0.34375689407085214, "grad_norm": 4.6266069412231445, "learning_rate": 1.526696677502693e-05, "loss": 0.15747077465057374, "step": 13050 }, { "epoch": 0.3440203093153508, "grad_norm": 3.264514207839966, "learning_rate": 1.525971273770939e-05, "loss": 0.1600273370742798, "step": 13060 }, { "epoch": 0.34428372455984957, "grad_norm": 2.6868135929107666, "learning_rate": 1.5252454872146003e-05, "loss": 0.1626504421234131, "step": 13070 }, { "epoch": 0.3445471398043483, "grad_norm": 2.869832992553711, "learning_rate": 1.5245193183619347e-05, "loss": 0.16262203454971313, "step": 13080 }, { "epoch": 0.34481055504884706, "grad_norm": 1.0974141359329224, "learning_rate": 1.5237927677414797e-05, "loss": 0.1652994394302368, "step": 13090 }, { "epoch": 0.3450739702933458, "grad_norm": 0.6971778869628906, "learning_rate": 1.5230658358820493e-05, "loss": 0.16043579578399658, "step": 13100 }, { "epoch": 0.34533738553784454, "grad_norm": 3.2558205127716064, "learning_rate": 1.5223385233127372e-05, "loss": 0.15510764122009277, "step": 13110 }, { "epoch": 0.3456008007823433, "grad_norm": 3.2548434734344482, "learning_rate": 1.5216108305629123e-05, "loss": 0.16091943979263307, "step": 13120 }, { "epoch": 0.34586421602684203, "grad_norm": 2.598045587539673, "learning_rate": 1.5208827581622205e-05, "loss": 0.1823554515838623, "step": 13130 }, { "epoch": 0.3461276312713408, "grad_norm": 1.3126760721206665, "learning_rate": 1.5201543066405844e-05, "loss": 0.15154241323471068, "step": 13140 }, { "epoch": 0.34639104651583946, "grad_norm": 1.7761560678482056, "learning_rate": 1.5194254765282025e-05, "loss": 0.16843410730361938, "step": 13150 }, { "epoch": 0.3466544617603382, "grad_norm": 1.6874388456344604, "learning_rate": 1.5186962683555493e-05, "loss": 0.15685877799987794, "step": 13160 }, { "epoch": 0.34691787700483695, "grad_norm": 1.121598720550537, "learning_rate": 1.517966682653373e-05, "loss": 0.14566428661346437, "step": 13170 }, { "epoch": 0.3471812922493357, "grad_norm": 3.1395325660705566, "learning_rate": 1.5172367199526985e-05, "loss": 0.1567463159561157, "step": 13180 }, { "epoch": 0.34744470749383444, "grad_norm": 3.190075159072876, "learning_rate": 1.5165063807848234e-05, "loss": 0.17015154361724855, "step": 13190 }, { "epoch": 0.3477081227383332, "grad_norm": 3.7017555236816406, "learning_rate": 1.5157756656813207e-05, "loss": 0.17296922206878662, "step": 13200 }, { "epoch": 0.3479715379828319, "grad_norm": 2.1365673542022705, "learning_rate": 1.5150445751740356e-05, "loss": 0.15181565284729004, "step": 13210 }, { "epoch": 0.34823495322733067, "grad_norm": 2.070613384246826, "learning_rate": 1.5143131097950882e-05, "loss": 0.1557554006576538, "step": 13220 }, { "epoch": 0.3484983684718294, "grad_norm": 2.295597791671753, "learning_rate": 1.5135812700768702e-05, "loss": 0.1716722369194031, "step": 13230 }, { "epoch": 0.3487617837163281, "grad_norm": 1.6590375900268555, "learning_rate": 1.512849056552046e-05, "loss": 0.1520382881164551, "step": 13240 }, { "epoch": 0.34902519896082684, "grad_norm": 2.218820095062256, "learning_rate": 1.512116469753552e-05, "loss": 0.15825960636138917, "step": 13250 }, { "epoch": 0.3492886142053256, "grad_norm": 2.8653712272644043, "learning_rate": 1.5113835102145968e-05, "loss": 0.1571236252784729, "step": 13260 }, { "epoch": 0.34955202944982433, "grad_norm": 1.7162284851074219, "learning_rate": 1.51065017846866e-05, "loss": 0.15420835018157958, "step": 13270 }, { "epoch": 0.3498154446943231, "grad_norm": 2.1327812671661377, "learning_rate": 1.509916475049492e-05, "loss": 0.15234239101409913, "step": 13280 }, { "epoch": 0.3500788599388218, "grad_norm": 4.448373317718506, "learning_rate": 1.5091824004911137e-05, "loss": 0.18508446216583252, "step": 13290 }, { "epoch": 0.35034227518332056, "grad_norm": 4.022204875946045, "learning_rate": 1.5084479553278164e-05, "loss": 0.1603198766708374, "step": 13300 }, { "epoch": 0.3506056904278193, "grad_norm": 2.4868273735046387, "learning_rate": 1.5077131400941609e-05, "loss": 0.1543164372444153, "step": 13310 }, { "epoch": 0.350869105672318, "grad_norm": 1.5387120246887207, "learning_rate": 1.5069779553249775e-05, "loss": 0.15823899507522582, "step": 13320 }, { "epoch": 0.35113252091681674, "grad_norm": 2.397981643676758, "learning_rate": 1.5062424015553653e-05, "loss": 0.17788903713226317, "step": 13330 }, { "epoch": 0.3513959361613155, "grad_norm": 2.9094808101654053, "learning_rate": 1.505506479320692e-05, "loss": 0.15300402641296387, "step": 13340 }, { "epoch": 0.3516593514058142, "grad_norm": 2.083972215652466, "learning_rate": 1.5047701891565938e-05, "loss": 0.16350606679916382, "step": 13350 }, { "epoch": 0.35192276665031297, "grad_norm": 1.1281750202178955, "learning_rate": 1.504033531598974e-05, "loss": 0.1524580955505371, "step": 13360 }, { "epoch": 0.3521861818948117, "grad_norm": 2.5901906490325928, "learning_rate": 1.5032965071840045e-05, "loss": 0.16463065147399902, "step": 13370 }, { "epoch": 0.35244959713931046, "grad_norm": 0.7893030047416687, "learning_rate": 1.5025591164481226e-05, "loss": 0.1752345085144043, "step": 13380 }, { "epoch": 0.3527130123838092, "grad_norm": 3.5874977111816406, "learning_rate": 1.5018213599280339e-05, "loss": 0.16132285594940185, "step": 13390 }, { "epoch": 0.35297642762830794, "grad_norm": 1.4714226722717285, "learning_rate": 1.5010832381607085e-05, "loss": 0.1644591808319092, "step": 13400 }, { "epoch": 0.35323984287280663, "grad_norm": 3.8437304496765137, "learning_rate": 1.500344751683384e-05, "loss": 0.16465480327606202, "step": 13410 }, { "epoch": 0.3535032581173054, "grad_norm": 5.167119979858398, "learning_rate": 1.4996059010335626e-05, "loss": 0.16809617280960082, "step": 13420 }, { "epoch": 0.3537666733618041, "grad_norm": 1.724470853805542, "learning_rate": 1.4988666867490117e-05, "loss": 0.15287526845932006, "step": 13430 }, { "epoch": 0.35403008860630286, "grad_norm": 5.089141845703125, "learning_rate": 1.4981271093677632e-05, "loss": 0.15647463798522948, "step": 13440 }, { "epoch": 0.3542935038508016, "grad_norm": 3.5242934226989746, "learning_rate": 1.4973871694281133e-05, "loss": 0.17033278942108154, "step": 13450 }, { "epoch": 0.35455691909530035, "grad_norm": 2.2527706623077393, "learning_rate": 1.4966468674686226e-05, "loss": 0.15055999755859376, "step": 13460 }, { "epoch": 0.3548203343397991, "grad_norm": 3.3796448707580566, "learning_rate": 1.4959062040281143e-05, "loss": 0.1660386562347412, "step": 13470 }, { "epoch": 0.35508374958429784, "grad_norm": 2.1424055099487305, "learning_rate": 1.4951651796456758e-05, "loss": 0.16657302379608155, "step": 13480 }, { "epoch": 0.3553471648287966, "grad_norm": 1.0358284711837769, "learning_rate": 1.4944237948606561e-05, "loss": 0.14878427982330322, "step": 13490 }, { "epoch": 0.35561058007329527, "grad_norm": 1.4037983417510986, "learning_rate": 1.4936820502126676e-05, "loss": 0.14650522470474242, "step": 13500 }, { "epoch": 0.355873995317794, "grad_norm": 3.091266393661499, "learning_rate": 1.4929399462415832e-05, "loss": 0.16045340299606323, "step": 13510 }, { "epoch": 0.35613741056229276, "grad_norm": 1.289367437362671, "learning_rate": 1.492197483487539e-05, "loss": 0.15986852645874022, "step": 13520 }, { "epoch": 0.3564008258067915, "grad_norm": 2.930661201477051, "learning_rate": 1.491454662490931e-05, "loss": 0.16319742202758789, "step": 13530 }, { "epoch": 0.35666424105129024, "grad_norm": 1.931762456893921, "learning_rate": 1.4907114837924168e-05, "loss": 0.1587587356567383, "step": 13540 }, { "epoch": 0.356927656295789, "grad_norm": 1.0475772619247437, "learning_rate": 1.4899679479329133e-05, "loss": 0.15687528848648072, "step": 13550 }, { "epoch": 0.35719107154028773, "grad_norm": 2.545457601547241, "learning_rate": 1.4892240554535981e-05, "loss": 0.15108500719070433, "step": 13560 }, { "epoch": 0.3574544867847865, "grad_norm": 0.8717813491821289, "learning_rate": 1.4884798068959085e-05, "loss": 0.1512166142463684, "step": 13570 }, { "epoch": 0.3577179020292852, "grad_norm": 1.341284155845642, "learning_rate": 1.4877352028015408e-05, "loss": 0.16443872451782227, "step": 13580 }, { "epoch": 0.3579813172737839, "grad_norm": 1.6681209802627563, "learning_rate": 1.4869902437124499e-05, "loss": 0.17020646333694459, "step": 13590 }, { "epoch": 0.35824473251828265, "grad_norm": 2.019318103790283, "learning_rate": 1.486244930170849e-05, "loss": 0.162303626537323, "step": 13600 }, { "epoch": 0.3585081477627814, "grad_norm": 2.1186182498931885, "learning_rate": 1.4854992627192096e-05, "loss": 0.16450183391571044, "step": 13610 }, { "epoch": 0.35877156300728014, "grad_norm": 3.4905714988708496, "learning_rate": 1.4847532419002602e-05, "loss": 0.1560025691986084, "step": 13620 }, { "epoch": 0.3590349782517789, "grad_norm": 0.760114312171936, "learning_rate": 1.4840068682569877e-05, "loss": 0.16143943071365358, "step": 13630 }, { "epoch": 0.3592983934962776, "grad_norm": 3.399817943572998, "learning_rate": 1.4832601423326346e-05, "loss": 0.13558695316314698, "step": 13640 }, { "epoch": 0.35956180874077637, "grad_norm": 1.9682550430297852, "learning_rate": 1.4825130646707002e-05, "loss": 0.13999924659729004, "step": 13650 }, { "epoch": 0.3598252239852751, "grad_norm": 3.139939308166504, "learning_rate": 1.4817656358149398e-05, "loss": 0.16865631341934204, "step": 13660 }, { "epoch": 0.3600886392297738, "grad_norm": 1.5956394672393799, "learning_rate": 1.4810178563093648e-05, "loss": 0.16565704345703125, "step": 13670 }, { "epoch": 0.36035205447427254, "grad_norm": 1.351992130279541, "learning_rate": 1.4802697266982411e-05, "loss": 0.1689929962158203, "step": 13680 }, { "epoch": 0.3606154697187713, "grad_norm": 2.5865750312805176, "learning_rate": 1.4795212475260896e-05, "loss": 0.16066808700561525, "step": 13690 }, { "epoch": 0.36087888496327003, "grad_norm": 1.4158074855804443, "learning_rate": 1.4787724193376858e-05, "loss": 0.14977107048034669, "step": 13700 }, { "epoch": 0.3611423002077688, "grad_norm": 4.021091461181641, "learning_rate": 1.4780232426780592e-05, "loss": 0.15779869556427, "step": 13710 }, { "epoch": 0.3614057154522675, "grad_norm": 1.9469753503799438, "learning_rate": 1.477273718092493e-05, "loss": 0.164007306098938, "step": 13720 }, { "epoch": 0.36166913069676626, "grad_norm": 1.4598522186279297, "learning_rate": 1.4765238461265235e-05, "loss": 0.16362442970275878, "step": 13730 }, { "epoch": 0.361932545941265, "grad_norm": 2.4680113792419434, "learning_rate": 1.4757736273259396e-05, "loss": 0.174299955368042, "step": 13740 }, { "epoch": 0.36219596118576375, "grad_norm": 1.6664282083511353, "learning_rate": 1.4750230622367831e-05, "loss": 0.1565871000289917, "step": 13750 }, { "epoch": 0.36245937643026244, "grad_norm": 5.093985557556152, "learning_rate": 1.474272151405348e-05, "loss": 0.15878005027770997, "step": 13760 }, { "epoch": 0.3627227916747612, "grad_norm": 1.6441926956176758, "learning_rate": 1.473520895378179e-05, "loss": 0.15884637832641602, "step": 13770 }, { "epoch": 0.3629862069192599, "grad_norm": 2.4103050231933594, "learning_rate": 1.472769294702073e-05, "loss": 0.1736898422241211, "step": 13780 }, { "epoch": 0.36324962216375867, "grad_norm": 4.705120086669922, "learning_rate": 1.4720173499240771e-05, "loss": 0.16004103422164917, "step": 13790 }, { "epoch": 0.3635130374082574, "grad_norm": 3.0650811195373535, "learning_rate": 1.4712650615914899e-05, "loss": 0.16469888687133788, "step": 13800 }, { "epoch": 0.36377645265275615, "grad_norm": 3.5502824783325195, "learning_rate": 1.4705124302518581e-05, "loss": 0.17754319906234742, "step": 13810 }, { "epoch": 0.3640398678972549, "grad_norm": 2.818157911300659, "learning_rate": 1.4697594564529805e-05, "loss": 0.16947882175445556, "step": 13820 }, { "epoch": 0.36430328314175364, "grad_norm": 1.6498527526855469, "learning_rate": 1.4690061407429028e-05, "loss": 0.15735697746276855, "step": 13830 }, { "epoch": 0.3645666983862524, "grad_norm": 1.2570174932479858, "learning_rate": 1.468252483669922e-05, "loss": 0.16694036722183228, "step": 13840 }, { "epoch": 0.3648301136307511, "grad_norm": 1.397150993347168, "learning_rate": 1.4674984857825808e-05, "loss": 0.16796432733535765, "step": 13850 }, { "epoch": 0.3650935288752498, "grad_norm": 3.247556447982788, "learning_rate": 1.4667441476296723e-05, "loss": 0.16585893630981446, "step": 13860 }, { "epoch": 0.36535694411974856, "grad_norm": 1.5549577474594116, "learning_rate": 1.4659894697602358e-05, "loss": 0.16174814701080323, "step": 13870 }, { "epoch": 0.3656203593642473, "grad_norm": 2.9744927883148193, "learning_rate": 1.465234452723559e-05, "loss": 0.1613584041595459, "step": 13880 }, { "epoch": 0.36588377460874605, "grad_norm": 2.5032029151916504, "learning_rate": 1.4644790970691755e-05, "loss": 0.15492992401123046, "step": 13890 }, { "epoch": 0.3661471898532448, "grad_norm": 2.190046787261963, "learning_rate": 1.4637234033468658e-05, "loss": 0.15627636909484863, "step": 13900 }, { "epoch": 0.36641060509774354, "grad_norm": 0.6320352554321289, "learning_rate": 1.4629673721066566e-05, "loss": 0.1597401261329651, "step": 13910 }, { "epoch": 0.3666740203422423, "grad_norm": 2.2658379077911377, "learning_rate": 1.4622110038988196e-05, "loss": 0.16727907657623292, "step": 13920 }, { "epoch": 0.36693743558674097, "grad_norm": 1.1726195812225342, "learning_rate": 1.4614542992738727e-05, "loss": 0.1698577284812927, "step": 13930 }, { "epoch": 0.3672008508312397, "grad_norm": 0.9888374209403992, "learning_rate": 1.4606972587825779e-05, "loss": 0.16544891595840455, "step": 13940 }, { "epoch": 0.36746426607573845, "grad_norm": 2.9786651134490967, "learning_rate": 1.4599398829759419e-05, "loss": 0.15621554851531982, "step": 13950 }, { "epoch": 0.3677276813202372, "grad_norm": 1.6982179880142212, "learning_rate": 1.4591821724052155e-05, "loss": 0.1490892291069031, "step": 13960 }, { "epoch": 0.36799109656473594, "grad_norm": 1.2809933423995972, "learning_rate": 1.4584241276218935e-05, "loss": 0.1678941249847412, "step": 13970 }, { "epoch": 0.3682545118092347, "grad_norm": 0.7000133991241455, "learning_rate": 1.4576657491777134e-05, "loss": 0.16132240295410155, "step": 13980 }, { "epoch": 0.36851792705373343, "grad_norm": 1.6672176122665405, "learning_rate": 1.4569070376246552e-05, "loss": 0.1637367606163025, "step": 13990 }, { "epoch": 0.3687813422982322, "grad_norm": 1.0611776113510132, "learning_rate": 1.4561479935149426e-05, "loss": 0.16353108882904052, "step": 14000 }, { "epoch": 0.3690447575427309, "grad_norm": 2.1859302520751953, "learning_rate": 1.4553886174010405e-05, "loss": 0.1701628088951111, "step": 14010 }, { "epoch": 0.3693081727872296, "grad_norm": 0.8339923024177551, "learning_rate": 1.4546289098356556e-05, "loss": 0.14804115295410156, "step": 14020 }, { "epoch": 0.36957158803172835, "grad_norm": 1.1490659713745117, "learning_rate": 1.4538688713717357e-05, "loss": 0.15027709007263185, "step": 14030 }, { "epoch": 0.3698350032762271, "grad_norm": 4.158492088317871, "learning_rate": 1.4531085025624694e-05, "loss": 0.1630620002746582, "step": 14040 }, { "epoch": 0.37009841852072584, "grad_norm": 0.8702024221420288, "learning_rate": 1.4523478039612863e-05, "loss": 0.17348761558532716, "step": 14050 }, { "epoch": 0.3703618337652246, "grad_norm": 1.8486634492874146, "learning_rate": 1.4515867761218558e-05, "loss": 0.1610948324203491, "step": 14060 }, { "epoch": 0.3706252490097233, "grad_norm": 2.1503372192382812, "learning_rate": 1.4508254195980861e-05, "loss": 0.15387767553329468, "step": 14070 }, { "epoch": 0.37088866425422207, "grad_norm": 1.200584053993225, "learning_rate": 1.450063734944126e-05, "loss": 0.16964712142944335, "step": 14080 }, { "epoch": 0.3711520794987208, "grad_norm": 1.408305287361145, "learning_rate": 1.449301722714362e-05, "loss": 0.16058638095855712, "step": 14090 }, { "epoch": 0.37141549474321955, "grad_norm": 1.4790067672729492, "learning_rate": 1.4485393834634202e-05, "loss": 0.15912286043167115, "step": 14100 }, { "epoch": 0.37167890998771824, "grad_norm": 2.476807117462158, "learning_rate": 1.447776717746163e-05, "loss": 0.15865724086761473, "step": 14110 }, { "epoch": 0.371942325232217, "grad_norm": 1.6286611557006836, "learning_rate": 1.4470137261176922e-05, "loss": 0.16535521745681764, "step": 14120 }, { "epoch": 0.37220574047671573, "grad_norm": 4.834408760070801, "learning_rate": 1.4462504091333454e-05, "loss": 0.1672847032546997, "step": 14130 }, { "epoch": 0.3724691557212145, "grad_norm": 1.7228273153305054, "learning_rate": 1.4454867673486983e-05, "loss": 0.15746675729751586, "step": 14140 }, { "epoch": 0.3727325709657132, "grad_norm": 2.3935463428497314, "learning_rate": 1.4447228013195616e-05, "loss": 0.1582711696624756, "step": 14150 }, { "epoch": 0.37299598621021196, "grad_norm": 1.2783974409103394, "learning_rate": 1.4439585116019835e-05, "loss": 0.16370561122894287, "step": 14160 }, { "epoch": 0.3732594014547107, "grad_norm": 2.507098436355591, "learning_rate": 1.4431938987522463e-05, "loss": 0.17095613479614258, "step": 14170 }, { "epoch": 0.37352281669920945, "grad_norm": 2.4468483924865723, "learning_rate": 1.4424289633268689e-05, "loss": 0.16964751482009888, "step": 14180 }, { "epoch": 0.37378623194370814, "grad_norm": 1.5342522859573364, "learning_rate": 1.4416637058826038e-05, "loss": 0.15073689222335815, "step": 14190 }, { "epoch": 0.3740496471882069, "grad_norm": 3.1413066387176514, "learning_rate": 1.4408981269764386e-05, "loss": 0.16064417362213135, "step": 14200 }, { "epoch": 0.3743130624327056, "grad_norm": 3.473292827606201, "learning_rate": 1.4401322271655947e-05, "loss": 0.154327392578125, "step": 14210 }, { "epoch": 0.37457647767720437, "grad_norm": 1.143991470336914, "learning_rate": 1.4393660070075267e-05, "loss": 0.16477541923522948, "step": 14220 }, { "epoch": 0.3748398929217031, "grad_norm": 3.4347336292266846, "learning_rate": 1.4385994670599233e-05, "loss": 0.1563987135887146, "step": 14230 }, { "epoch": 0.37510330816620185, "grad_norm": 1.664065957069397, "learning_rate": 1.4378326078807048e-05, "loss": 0.15738754272460936, "step": 14240 }, { "epoch": 0.3753667234107006, "grad_norm": 0.6678531169891357, "learning_rate": 1.4370654300280246e-05, "loss": 0.16624982357025148, "step": 14250 }, { "epoch": 0.37563013865519934, "grad_norm": 1.8956183195114136, "learning_rate": 1.4362979340602677e-05, "loss": 0.16856122016906738, "step": 14260 }, { "epoch": 0.3758935538996981, "grad_norm": 1.0687062740325928, "learning_rate": 1.4355301205360513e-05, "loss": 0.1654583692550659, "step": 14270 }, { "epoch": 0.3761569691441968, "grad_norm": 0.8567618131637573, "learning_rate": 1.4347619900142223e-05, "loss": 0.18261922597885133, "step": 14280 }, { "epoch": 0.3764203843886955, "grad_norm": 2.644319534301758, "learning_rate": 1.43399354305386e-05, "loss": 0.14619865417480468, "step": 14290 }, { "epoch": 0.37668379963319426, "grad_norm": 5.0101823806762695, "learning_rate": 1.4332247802142729e-05, "loss": 0.17979981899261474, "step": 14300 }, { "epoch": 0.376947214877693, "grad_norm": 0.8599063158035278, "learning_rate": 1.4324557020550001e-05, "loss": 0.16115405559539794, "step": 14310 }, { "epoch": 0.37721063012219175, "grad_norm": 1.3125149011611938, "learning_rate": 1.4316863091358093e-05, "loss": 0.15699079036712646, "step": 14320 }, { "epoch": 0.3774740453666905, "grad_norm": 1.430658221244812, "learning_rate": 1.4309166020166984e-05, "loss": 0.1426365613937378, "step": 14330 }, { "epoch": 0.37773746061118924, "grad_norm": 1.171064019203186, "learning_rate": 1.430146581257893e-05, "loss": 0.1578814744949341, "step": 14340 }, { "epoch": 0.378000875855688, "grad_norm": 3.0393030643463135, "learning_rate": 1.4293762474198478e-05, "loss": 0.15770294666290283, "step": 14350 }, { "epoch": 0.3782642911001867, "grad_norm": 2.3620078563690186, "learning_rate": 1.428605601063245e-05, "loss": 0.15714809894561768, "step": 14360 }, { "epoch": 0.3785277063446854, "grad_norm": 1.508978247642517, "learning_rate": 1.427834642748994e-05, "loss": 0.15609703063964844, "step": 14370 }, { "epoch": 0.37879112158918415, "grad_norm": 1.4067661762237549, "learning_rate": 1.4270633730382315e-05, "loss": 0.16570128202438356, "step": 14380 }, { "epoch": 0.3790545368336829, "grad_norm": 3.218079090118408, "learning_rate": 1.4262917924923212e-05, "loss": 0.15144460201263427, "step": 14390 }, { "epoch": 0.37931795207818164, "grad_norm": 2.3295094966888428, "learning_rate": 1.4255199016728527e-05, "loss": 0.1589252233505249, "step": 14400 }, { "epoch": 0.3795813673226804, "grad_norm": 258.0946350097656, "learning_rate": 1.4247477011416409e-05, "loss": 0.45247249603271483, "step": 14410 }, { "epoch": 0.37984478256717913, "grad_norm": 1.1835099458694458, "learning_rate": 1.4239751914607274e-05, "loss": 0.17500343322753906, "step": 14420 }, { "epoch": 0.3801081978116779, "grad_norm": 3.674506187438965, "learning_rate": 1.4232023731923772e-05, "loss": 0.14928234815597535, "step": 14430 }, { "epoch": 0.3803716130561766, "grad_norm": 0.9253420829772949, "learning_rate": 1.422429246899082e-05, "loss": 0.17087066173553467, "step": 14440 }, { "epoch": 0.3806350283006753, "grad_norm": 2.170544385910034, "learning_rate": 1.4216558131435551e-05, "loss": 0.15705521106719972, "step": 14450 }, { "epoch": 0.38089844354517405, "grad_norm": 1.4342777729034424, "learning_rate": 1.420882072488736e-05, "loss": 0.15809000730514527, "step": 14460 }, { "epoch": 0.3811618587896728, "grad_norm": 3.682626247406006, "learning_rate": 1.420108025497786e-05, "loss": 0.1690453290939331, "step": 14470 }, { "epoch": 0.38142527403417154, "grad_norm": 1.1393108367919922, "learning_rate": 1.41933367273409e-05, "loss": 0.15913679599761962, "step": 14480 }, { "epoch": 0.3816886892786703, "grad_norm": 1.518021821975708, "learning_rate": 1.4185590147612555e-05, "loss": 0.16025235652923583, "step": 14490 }, { "epoch": 0.381952104523169, "grad_norm": 4.006959438323975, "learning_rate": 1.4177840521431117e-05, "loss": 0.15303401947021483, "step": 14500 }, { "epoch": 0.38221551976766777, "grad_norm": 5.361456394195557, "learning_rate": 1.4170087854437102e-05, "loss": 0.14715441465377807, "step": 14510 }, { "epoch": 0.3824789350121665, "grad_norm": 2.9036991596221924, "learning_rate": 1.4162332152273233e-05, "loss": 0.17217705249786378, "step": 14520 }, { "epoch": 0.38274235025666525, "grad_norm": 3.1034581661224365, "learning_rate": 1.4154573420584448e-05, "loss": 0.15572348833084107, "step": 14530 }, { "epoch": 0.38300576550116394, "grad_norm": 1.282259464263916, "learning_rate": 1.414681166501788e-05, "loss": 0.15965726375579833, "step": 14540 }, { "epoch": 0.3832691807456627, "grad_norm": 1.0059623718261719, "learning_rate": 1.4139046891222877e-05, "loss": 0.16898428201675414, "step": 14550 }, { "epoch": 0.38353259599016143, "grad_norm": 1.368445634841919, "learning_rate": 1.4131279104850968e-05, "loss": 0.18199900388717652, "step": 14560 }, { "epoch": 0.3837960112346602, "grad_norm": 1.227403998374939, "learning_rate": 1.4123508311555894e-05, "loss": 0.15673208236694336, "step": 14570 }, { "epoch": 0.3840594264791589, "grad_norm": 1.7012523412704468, "learning_rate": 1.4115734516993562e-05, "loss": 0.1723322868347168, "step": 14580 }, { "epoch": 0.38432284172365766, "grad_norm": 4.948211193084717, "learning_rate": 1.4107957726822084e-05, "loss": 0.1628418207168579, "step": 14590 }, { "epoch": 0.3845862569681564, "grad_norm": 3.140411138534546, "learning_rate": 1.4100177946701738e-05, "loss": 0.16641961336135863, "step": 14600 }, { "epoch": 0.38484967221265515, "grad_norm": 1.0507076978683472, "learning_rate": 1.4092395182294989e-05, "loss": 0.1585771083831787, "step": 14610 }, { "epoch": 0.3851130874571539, "grad_norm": 1.8507930040359497, "learning_rate": 1.4084609439266464e-05, "loss": 0.16215171813964843, "step": 14620 }, { "epoch": 0.3853765027016526, "grad_norm": 1.8338567018508911, "learning_rate": 1.4076820723282962e-05, "loss": 0.15502206087112427, "step": 14630 }, { "epoch": 0.3856399179461513, "grad_norm": 1.707466959953308, "learning_rate": 1.4069029040013448e-05, "loss": 0.17360570430755615, "step": 14640 }, { "epoch": 0.38590333319065007, "grad_norm": 1.9835909605026245, "learning_rate": 1.4061234395129049e-05, "loss": 0.16454769372940065, "step": 14650 }, { "epoch": 0.3861667484351488, "grad_norm": 2.0270962715148926, "learning_rate": 1.405343679430304e-05, "loss": 0.157112193107605, "step": 14660 }, { "epoch": 0.38643016367964755, "grad_norm": 2.0894534587860107, "learning_rate": 1.4045636243210852e-05, "loss": 0.16364270448684692, "step": 14670 }, { "epoch": 0.3866935789241463, "grad_norm": 1.4398894309997559, "learning_rate": 1.4037832747530063e-05, "loss": 0.15105621814727782, "step": 14680 }, { "epoch": 0.38695699416864504, "grad_norm": 1.772464394569397, "learning_rate": 1.403002631294039e-05, "loss": 0.15247184038162231, "step": 14690 }, { "epoch": 0.3872204094131438, "grad_norm": 2.2886688709259033, "learning_rate": 1.40222169451237e-05, "loss": 0.15062260627746582, "step": 14700 }, { "epoch": 0.3874838246576425, "grad_norm": 1.8002984523773193, "learning_rate": 1.4014404649763983e-05, "loss": 0.1612554669380188, "step": 14710 }, { "epoch": 0.3877472399021412, "grad_norm": 2.6874983310699463, "learning_rate": 1.4006589432547367e-05, "loss": 0.16163210868835448, "step": 14720 }, { "epoch": 0.38801065514663996, "grad_norm": 3.229072332382202, "learning_rate": 1.3998771299162099e-05, "loss": 0.16106843948364258, "step": 14730 }, { "epoch": 0.3882740703911387, "grad_norm": 2.3598477840423584, "learning_rate": 1.3990950255298565e-05, "loss": 0.15754120349884032, "step": 14740 }, { "epoch": 0.38853748563563745, "grad_norm": 1.2223684787750244, "learning_rate": 1.3983126306649243e-05, "loss": 0.1722850799560547, "step": 14750 }, { "epoch": 0.3888009008801362, "grad_norm": 3.4291553497314453, "learning_rate": 1.3975299458908753e-05, "loss": 0.15089843273162842, "step": 14760 }, { "epoch": 0.38906431612463493, "grad_norm": 2.4585399627685547, "learning_rate": 1.3967469717773807e-05, "loss": 0.16426119804382325, "step": 14770 }, { "epoch": 0.3893277313691337, "grad_norm": 2.933173656463623, "learning_rate": 1.395963708894323e-05, "loss": 0.1489073634147644, "step": 14780 }, { "epoch": 0.3895911466136324, "grad_norm": 3.88875675201416, "learning_rate": 1.3951801578117946e-05, "loss": 0.1631092071533203, "step": 14790 }, { "epoch": 0.3898545618581311, "grad_norm": 1.477650761604309, "learning_rate": 1.3943963191000976e-05, "loss": 0.15655603408813476, "step": 14800 }, { "epoch": 0.39011797710262985, "grad_norm": 2.059464693069458, "learning_rate": 1.393612193329744e-05, "loss": 0.1646730899810791, "step": 14810 }, { "epoch": 0.3903813923471286, "grad_norm": 4.727813720703125, "learning_rate": 1.3928277810714542e-05, "loss": 0.15897603034973146, "step": 14820 }, { "epoch": 0.39064480759162734, "grad_norm": 3.2612833976745605, "learning_rate": 1.3920430828961576e-05, "loss": 0.15469725131988527, "step": 14830 }, { "epoch": 0.3909082228361261, "grad_norm": 1.5628430843353271, "learning_rate": 1.3912580993749907e-05, "loss": 0.15268006324768066, "step": 14840 }, { "epoch": 0.39117163808062483, "grad_norm": 2.1253840923309326, "learning_rate": 1.3904728310792988e-05, "loss": 0.1598591089248657, "step": 14850 }, { "epoch": 0.39143505332512357, "grad_norm": 3.1208159923553467, "learning_rate": 1.3896872785806342e-05, "loss": 0.15533146858215333, "step": 14860 }, { "epoch": 0.3916984685696223, "grad_norm": 4.066123008728027, "learning_rate": 1.3889014424507557e-05, "loss": 0.14902329444885254, "step": 14870 }, { "epoch": 0.39196188381412106, "grad_norm": 1.6864590644836426, "learning_rate": 1.3881153232616284e-05, "loss": 0.15356595516204835, "step": 14880 }, { "epoch": 0.39222529905861975, "grad_norm": 1.7028852701187134, "learning_rate": 1.3873289215854247e-05, "loss": 0.15473037958145142, "step": 14890 }, { "epoch": 0.3924887143031185, "grad_norm": 2.2599151134490967, "learning_rate": 1.3865422379945205e-05, "loss": 0.140410315990448, "step": 14900 }, { "epoch": 0.39275212954761723, "grad_norm": 1.386487603187561, "learning_rate": 1.3857552730614989e-05, "loss": 0.16866681575775147, "step": 14910 }, { "epoch": 0.393015544792116, "grad_norm": 1.5485247373580933, "learning_rate": 1.3849680273591468e-05, "loss": 0.15432244539260864, "step": 14920 }, { "epoch": 0.3932789600366147, "grad_norm": 1.4610096216201782, "learning_rate": 1.3841805014604553e-05, "loss": 0.1735697031021118, "step": 14930 }, { "epoch": 0.39354237528111347, "grad_norm": 3.936833143234253, "learning_rate": 1.3833926959386199e-05, "loss": 0.16180727481842042, "step": 14940 }, { "epoch": 0.3938057905256122, "grad_norm": 3.2332775592803955, "learning_rate": 1.3826046113670395e-05, "loss": 0.16393299102783204, "step": 14950 }, { "epoch": 0.39406920577011095, "grad_norm": 1.547184705734253, "learning_rate": 1.3818162483193159e-05, "loss": 0.1599513292312622, "step": 14960 }, { "epoch": 0.39433262101460964, "grad_norm": 1.1914112567901611, "learning_rate": 1.3810276073692537e-05, "loss": 0.14430402517318724, "step": 14970 }, { "epoch": 0.3945960362591084, "grad_norm": 1.05608069896698, "learning_rate": 1.3802386890908602e-05, "loss": 0.1532583236694336, "step": 14980 }, { "epoch": 0.39485945150360713, "grad_norm": 2.339719533920288, "learning_rate": 1.3794494940583437e-05, "loss": 0.16175962686538697, "step": 14990 }, { "epoch": 0.39512286674810587, "grad_norm": 1.4726624488830566, "learning_rate": 1.3786600228461147e-05, "loss": 0.15936880111694335, "step": 15000 }, { "epoch": 0.3953862819926046, "grad_norm": 3.1126046180725098, "learning_rate": 1.377870276028784e-05, "loss": 0.1552330732345581, "step": 15010 }, { "epoch": 0.39564969723710336, "grad_norm": 1.8616976737976074, "learning_rate": 1.377080254181164e-05, "loss": 0.16712779998779298, "step": 15020 }, { "epoch": 0.3959131124816021, "grad_norm": 1.1375939846038818, "learning_rate": 1.3762899578782662e-05, "loss": 0.16148959398269652, "step": 15030 }, { "epoch": 0.39617652772610085, "grad_norm": 3.726802349090576, "learning_rate": 1.375499387695303e-05, "loss": 0.15516880750656128, "step": 15040 }, { "epoch": 0.3964399429705996, "grad_norm": 1.609256625175476, "learning_rate": 1.3747085442076844e-05, "loss": 0.15336871147155762, "step": 15050 }, { "epoch": 0.3967033582150983, "grad_norm": 1.0346674919128418, "learning_rate": 1.3739174279910214e-05, "loss": 0.15490858554840087, "step": 15060 }, { "epoch": 0.396966773459597, "grad_norm": 1.814074158668518, "learning_rate": 1.3731260396211219e-05, "loss": 0.14047739505767823, "step": 15070 }, { "epoch": 0.39723018870409577, "grad_norm": 4.398323059082031, "learning_rate": 1.372334379673993e-05, "loss": 0.1576615810394287, "step": 15080 }, { "epoch": 0.3974936039485945, "grad_norm": 1.0333209037780762, "learning_rate": 1.3715424487258386e-05, "loss": 0.15380159616470337, "step": 15090 }, { "epoch": 0.39775701919309325, "grad_norm": 1.5495966672897339, "learning_rate": 1.37075024735306e-05, "loss": 0.17000943422317505, "step": 15100 }, { "epoch": 0.398020434437592, "grad_norm": 2.444828987121582, "learning_rate": 1.3699577761322563e-05, "loss": 0.15729379653930664, "step": 15110 }, { "epoch": 0.39828384968209074, "grad_norm": 1.6755802631378174, "learning_rate": 1.3691650356402212e-05, "loss": 0.1656790256500244, "step": 15120 }, { "epoch": 0.3985472649265895, "grad_norm": 1.3463246822357178, "learning_rate": 1.3683720264539469e-05, "loss": 0.14073143005371094, "step": 15130 }, { "epoch": 0.3988106801710882, "grad_norm": 1.9016344547271729, "learning_rate": 1.3675787491506184e-05, "loss": 0.15013134479522705, "step": 15140 }, { "epoch": 0.3990740954155869, "grad_norm": 0.9839884042739868, "learning_rate": 1.3667852043076173e-05, "loss": 0.14874560832977296, "step": 15150 }, { "epoch": 0.39933751066008566, "grad_norm": 3.531522035598755, "learning_rate": 1.3659913925025203e-05, "loss": 0.1657315492630005, "step": 15160 }, { "epoch": 0.3996009259045844, "grad_norm": 3.979154348373413, "learning_rate": 1.3651973143130979e-05, "loss": 0.15080522298812865, "step": 15170 }, { "epoch": 0.39986434114908315, "grad_norm": 2.0310568809509277, "learning_rate": 1.3644029703173139e-05, "loss": 0.14776322841644288, "step": 15180 }, { "epoch": 0.4001277563935819, "grad_norm": 1.4997801780700684, "learning_rate": 1.3636083610933262e-05, "loss": 0.1580416202545166, "step": 15190 }, { "epoch": 0.40039117163808063, "grad_norm": 2.8284435272216797, "learning_rate": 1.3628134872194861e-05, "loss": 0.15753769874572754, "step": 15200 }, { "epoch": 0.4006545868825794, "grad_norm": 0.9221677184104919, "learning_rate": 1.362018349274337e-05, "loss": 0.14885252714157104, "step": 15210 }, { "epoch": 0.4009180021270781, "grad_norm": 2.902886152267456, "learning_rate": 1.3612229478366144e-05, "loss": 0.14972057342529296, "step": 15220 }, { "epoch": 0.4011814173715768, "grad_norm": 3.6704702377319336, "learning_rate": 1.360427283485246e-05, "loss": 0.16149468421936036, "step": 15230 }, { "epoch": 0.40144483261607555, "grad_norm": 0.8449860215187073, "learning_rate": 1.3596313567993504e-05, "loss": 0.16396572589874267, "step": 15240 }, { "epoch": 0.4017082478605743, "grad_norm": 0.7099227905273438, "learning_rate": 1.3588351683582374e-05, "loss": 0.1483350157737732, "step": 15250 }, { "epoch": 0.40197166310507304, "grad_norm": 1.922419548034668, "learning_rate": 1.3580387187414077e-05, "loss": 0.14881529808044433, "step": 15260 }, { "epoch": 0.4022350783495718, "grad_norm": 2.0435967445373535, "learning_rate": 1.3572420085285512e-05, "loss": 0.14907667636871338, "step": 15270 }, { "epoch": 0.4024984935940705, "grad_norm": 2.370396375656128, "learning_rate": 1.3564450382995482e-05, "loss": 0.15713901519775392, "step": 15280 }, { "epoch": 0.40276190883856927, "grad_norm": 1.2747790813446045, "learning_rate": 1.3556478086344678e-05, "loss": 0.14656463861465455, "step": 15290 }, { "epoch": 0.403025324083068, "grad_norm": 2.8674535751342773, "learning_rate": 1.3548503201135687e-05, "loss": 0.13284735679626464, "step": 15300 }, { "epoch": 0.40328873932756676, "grad_norm": 2.263246774673462, "learning_rate": 1.3540525733172966e-05, "loss": 0.16349029541015625, "step": 15310 }, { "epoch": 0.40355215457206545, "grad_norm": 1.731724739074707, "learning_rate": 1.353254568826287e-05, "loss": 0.1564262628555298, "step": 15320 }, { "epoch": 0.4038155698165642, "grad_norm": 1.3334931135177612, "learning_rate": 1.3524563072213612e-05, "loss": 0.1590888023376465, "step": 15330 }, { "epoch": 0.40407898506106293, "grad_norm": 1.9963860511779785, "learning_rate": 1.3516577890835292e-05, "loss": 0.1557229280471802, "step": 15340 }, { "epoch": 0.4043424003055617, "grad_norm": 1.5001667737960815, "learning_rate": 1.350859014993986e-05, "loss": 0.15979363918304443, "step": 15350 }, { "epoch": 0.4046058155500604, "grad_norm": 1.369351863861084, "learning_rate": 1.3500599855341146e-05, "loss": 0.1521340250968933, "step": 15360 }, { "epoch": 0.40486923079455917, "grad_norm": 2.084557056427002, "learning_rate": 1.3492607012854824e-05, "loss": 0.16356745958328248, "step": 15370 }, { "epoch": 0.4051326460390579, "grad_norm": 1.494112253189087, "learning_rate": 1.3484611628298438e-05, "loss": 0.13845330476760864, "step": 15380 }, { "epoch": 0.40539606128355665, "grad_norm": 4.1488189697265625, "learning_rate": 1.3476613707491372e-05, "loss": 0.1549588441848755, "step": 15390 }, { "epoch": 0.4056594765280554, "grad_norm": 1.0269775390625, "learning_rate": 1.3468613256254852e-05, "loss": 0.1461463212966919, "step": 15400 }, { "epoch": 0.4059228917725541, "grad_norm": 1.1385730504989624, "learning_rate": 1.3460610280411957e-05, "loss": 0.16253145933151245, "step": 15410 }, { "epoch": 0.40618630701705283, "grad_norm": 2.308112144470215, "learning_rate": 1.3452604785787595e-05, "loss": 0.15360026359558104, "step": 15420 }, { "epoch": 0.40644972226155157, "grad_norm": 2.361760139465332, "learning_rate": 1.3444596778208516e-05, "loss": 0.16123514175415038, "step": 15430 }, { "epoch": 0.4067131375060503, "grad_norm": 1.9772143363952637, "learning_rate": 1.3436586263503286e-05, "loss": 0.15388458967208862, "step": 15440 }, { "epoch": 0.40697655275054906, "grad_norm": 1.9585380554199219, "learning_rate": 1.3428573247502307e-05, "loss": 0.1657373547554016, "step": 15450 }, { "epoch": 0.4072399679950478, "grad_norm": 4.317541599273682, "learning_rate": 1.3420557736037796e-05, "loss": 0.15769777297973633, "step": 15460 }, { "epoch": 0.40750338323954655, "grad_norm": 1.4029672145843506, "learning_rate": 1.3412539734943792e-05, "loss": 0.13722939491271974, "step": 15470 }, { "epoch": 0.4077667984840453, "grad_norm": 1.0600261688232422, "learning_rate": 1.3404519250056139e-05, "loss": 0.1507514476776123, "step": 15480 }, { "epoch": 0.408030213728544, "grad_norm": 2.183746576309204, "learning_rate": 1.3396496287212491e-05, "loss": 0.14772436618804932, "step": 15490 }, { "epoch": 0.4082936289730427, "grad_norm": 2.743912696838379, "learning_rate": 1.3388470852252307e-05, "loss": 0.16028909683227538, "step": 15500 }, { "epoch": 0.40855704421754147, "grad_norm": 1.7925910949707031, "learning_rate": 1.3380442951016846e-05, "loss": 0.14300792217254638, "step": 15510 }, { "epoch": 0.4088204594620402, "grad_norm": 2.7721145153045654, "learning_rate": 1.3372412589349164e-05, "loss": 0.1553818941116333, "step": 15520 }, { "epoch": 0.40908387470653895, "grad_norm": 2.2580835819244385, "learning_rate": 1.3364379773094099e-05, "loss": 0.16050641536712645, "step": 15530 }, { "epoch": 0.4093472899510377, "grad_norm": 1.517825722694397, "learning_rate": 1.335634450809828e-05, "loss": 0.1589105248451233, "step": 15540 }, { "epoch": 0.40961070519553644, "grad_norm": 1.3187273740768433, "learning_rate": 1.3348306800210127e-05, "loss": 0.15825140476226807, "step": 15550 }, { "epoch": 0.4098741204400352, "grad_norm": 1.5363178253173828, "learning_rate": 1.3340266655279828e-05, "loss": 0.1540547013282776, "step": 15560 }, { "epoch": 0.4101375356845339, "grad_norm": 1.1576497554779053, "learning_rate": 1.3332224079159342e-05, "loss": 0.15715334415435792, "step": 15570 }, { "epoch": 0.4104009509290326, "grad_norm": 0.7747457027435303, "learning_rate": 1.3324179077702408e-05, "loss": 0.15561435222625733, "step": 15580 }, { "epoch": 0.41066436617353136, "grad_norm": 4.118073463439941, "learning_rate": 1.3316131656764523e-05, "loss": 0.13673267364501954, "step": 15590 }, { "epoch": 0.4109277814180301, "grad_norm": 1.8171545267105103, "learning_rate": 1.3308081822202953e-05, "loss": 0.16314237117767333, "step": 15600 }, { "epoch": 0.41119119666252885, "grad_norm": 2.2092185020446777, "learning_rate": 1.3300029579876705e-05, "loss": 0.1659896492958069, "step": 15610 }, { "epoch": 0.4114546119070276, "grad_norm": 1.0958943367004395, "learning_rate": 1.3291974935646554e-05, "loss": 0.15250880718231202, "step": 15620 }, { "epoch": 0.41171802715152633, "grad_norm": 0.9746562838554382, "learning_rate": 1.3283917895375018e-05, "loss": 0.14739325046539306, "step": 15630 }, { "epoch": 0.4119814423960251, "grad_norm": 1.5897372961044312, "learning_rate": 1.327585846492636e-05, "loss": 0.14322038888931274, "step": 15640 }, { "epoch": 0.4122448576405238, "grad_norm": 2.4600863456726074, "learning_rate": 1.3267796650166576e-05, "loss": 0.1604902148246765, "step": 15650 }, { "epoch": 0.41250827288502256, "grad_norm": 0.934002161026001, "learning_rate": 1.3259732456963406e-05, "loss": 0.15738061666488648, "step": 15660 }, { "epoch": 0.41277168812952125, "grad_norm": 2.5999176502227783, "learning_rate": 1.3251665891186313e-05, "loss": 0.1525482177734375, "step": 15670 }, { "epoch": 0.41303510337402, "grad_norm": 1.4940656423568726, "learning_rate": 1.32435969587065e-05, "loss": 0.14969490766525267, "step": 15680 }, { "epoch": 0.41329851861851874, "grad_norm": 1.0396983623504639, "learning_rate": 1.323552566539688e-05, "loss": 0.15518062114715575, "step": 15690 }, { "epoch": 0.4135619338630175, "grad_norm": 3.847123622894287, "learning_rate": 1.3227452017132085e-05, "loss": 0.15496978759765626, "step": 15700 }, { "epoch": 0.4138253491075162, "grad_norm": 3.629978895187378, "learning_rate": 1.321937601978847e-05, "loss": 0.15425710678100585, "step": 15710 }, { "epoch": 0.41408876435201497, "grad_norm": 2.8705928325653076, "learning_rate": 1.3211297679244093e-05, "loss": 0.1595106840133667, "step": 15720 }, { "epoch": 0.4143521795965137, "grad_norm": 2.088203191757202, "learning_rate": 1.3203217001378716e-05, "loss": 0.1662042260169983, "step": 15730 }, { "epoch": 0.41461559484101246, "grad_norm": 2.6319375038146973, "learning_rate": 1.3195133992073808e-05, "loss": 0.15411105155944824, "step": 15740 }, { "epoch": 0.41487901008551115, "grad_norm": 2.4527335166931152, "learning_rate": 1.318704865721253e-05, "loss": 0.14951118230819702, "step": 15750 }, { "epoch": 0.4151424253300099, "grad_norm": 2.6788928508758545, "learning_rate": 1.3178961002679738e-05, "loss": 0.15774575471878052, "step": 15760 }, { "epoch": 0.41540584057450863, "grad_norm": 2.533646821975708, "learning_rate": 1.317087103436198e-05, "loss": 0.16922245025634766, "step": 15770 }, { "epoch": 0.4156692558190074, "grad_norm": 1.3864277601242065, "learning_rate": 1.3162778758147477e-05, "loss": 0.15826280117034913, "step": 15780 }, { "epoch": 0.4159326710635061, "grad_norm": 1.4617489576339722, "learning_rate": 1.315468417992614e-05, "loss": 0.16288834810256958, "step": 15790 }, { "epoch": 0.41619608630800486, "grad_norm": 1.4129775762557983, "learning_rate": 1.3146587305589552e-05, "loss": 0.1571655750274658, "step": 15800 }, { "epoch": 0.4164595015525036, "grad_norm": 1.2778468132019043, "learning_rate": 1.3138488141030967e-05, "loss": 0.14710841178894044, "step": 15810 }, { "epoch": 0.41672291679700235, "grad_norm": 3.7519264221191406, "learning_rate": 1.313038669214531e-05, "loss": 0.15905776023864746, "step": 15820 }, { "epoch": 0.4169863320415011, "grad_norm": 0.9647161364555359, "learning_rate": 1.312228296482916e-05, "loss": 0.16102735996246337, "step": 15830 }, { "epoch": 0.4172497472859998, "grad_norm": 0.7667250037193298, "learning_rate": 1.311417696498076e-05, "loss": 0.1525890827178955, "step": 15840 }, { "epoch": 0.4175131625304985, "grad_norm": 1.5678218603134155, "learning_rate": 1.310606869850001e-05, "loss": 0.13802632093429565, "step": 15850 }, { "epoch": 0.41777657777499727, "grad_norm": 0.9613314867019653, "learning_rate": 1.3097958171288453e-05, "loss": 0.15221437215805053, "step": 15860 }, { "epoch": 0.418039993019496, "grad_norm": 5.010793685913086, "learning_rate": 1.3089845389249282e-05, "loss": 0.15328794717788696, "step": 15870 }, { "epoch": 0.41830340826399476, "grad_norm": 1.5990492105484009, "learning_rate": 1.308173035828733e-05, "loss": 0.1594420313835144, "step": 15880 }, { "epoch": 0.4185668235084935, "grad_norm": 1.7168344259262085, "learning_rate": 1.3073613084309065e-05, "loss": 0.1574240207672119, "step": 15890 }, { "epoch": 0.41883023875299225, "grad_norm": 2.9633078575134277, "learning_rate": 1.3065493573222594e-05, "loss": 0.16617171764373778, "step": 15900 }, { "epoch": 0.419093653997491, "grad_norm": 1.2596832513809204, "learning_rate": 1.305737183093764e-05, "loss": 0.1537771701812744, "step": 15910 }, { "epoch": 0.41935706924198973, "grad_norm": 1.6739434003829956, "learning_rate": 1.3049247863365562e-05, "loss": 0.15421160459518432, "step": 15920 }, { "epoch": 0.4196204844864884, "grad_norm": 1.234724760055542, "learning_rate": 1.3041121676419333e-05, "loss": 0.15118536949157715, "step": 15930 }, { "epoch": 0.41988389973098716, "grad_norm": 1.3341141939163208, "learning_rate": 1.3032993276013546e-05, "loss": 0.1540053367614746, "step": 15940 }, { "epoch": 0.4201473149754859, "grad_norm": 0.8176248669624329, "learning_rate": 1.3024862668064393e-05, "loss": 0.1471743106842041, "step": 15950 }, { "epoch": 0.42041073021998465, "grad_norm": 1.1468281745910645, "learning_rate": 1.3016729858489689e-05, "loss": 0.1648090124130249, "step": 15960 }, { "epoch": 0.4206741454644834, "grad_norm": 1.092151165008545, "learning_rate": 1.300859485320884e-05, "loss": 0.16025390625, "step": 15970 }, { "epoch": 0.42093756070898214, "grad_norm": 2.792149782180786, "learning_rate": 1.3000457658142858e-05, "loss": 0.16141164302825928, "step": 15980 }, { "epoch": 0.4212009759534809, "grad_norm": 2.1879098415374756, "learning_rate": 1.299231827921434e-05, "loss": 0.15910725593566893, "step": 15990 }, { "epoch": 0.4214643911979796, "grad_norm": 1.4352707862854004, "learning_rate": 1.2984176722347478e-05, "loss": 0.1510244607925415, "step": 16000 }, { "epoch": 0.4217278064424783, "grad_norm": 1.3080593347549438, "learning_rate": 1.2976032993468049e-05, "loss": 0.15394012928009032, "step": 16010 }, { "epoch": 0.42199122168697706, "grad_norm": 3.4411914348602295, "learning_rate": 1.296788709850341e-05, "loss": 0.1401626944541931, "step": 16020 }, { "epoch": 0.4222546369314758, "grad_norm": 1.6342495679855347, "learning_rate": 1.29597390433825e-05, "loss": 0.13755732774734497, "step": 16030 }, { "epoch": 0.42251805217597455, "grad_norm": 1.9505252838134766, "learning_rate": 1.2951588834035813e-05, "loss": 0.1568729043006897, "step": 16040 }, { "epoch": 0.4227814674204733, "grad_norm": 1.6271623373031616, "learning_rate": 1.2943436476395435e-05, "loss": 0.1640044331550598, "step": 16050 }, { "epoch": 0.42304488266497203, "grad_norm": 1.098147988319397, "learning_rate": 1.2935281976394998e-05, "loss": 0.1586430311203003, "step": 16060 }, { "epoch": 0.4233082979094708, "grad_norm": 2.333918333053589, "learning_rate": 1.2927125339969701e-05, "loss": 0.1520540714263916, "step": 16070 }, { "epoch": 0.4235717131539695, "grad_norm": 1.5542707443237305, "learning_rate": 1.2918966573056291e-05, "loss": 0.14246711730957032, "step": 16080 }, { "epoch": 0.42383512839846826, "grad_norm": 2.547348737716675, "learning_rate": 1.2910805681593078e-05, "loss": 0.1646599292755127, "step": 16090 }, { "epoch": 0.42409854364296695, "grad_norm": 0.8342078328132629, "learning_rate": 1.2902642671519903e-05, "loss": 0.1490323305130005, "step": 16100 }, { "epoch": 0.4243619588874657, "grad_norm": 1.069466471672058, "learning_rate": 1.2894477548778164e-05, "loss": 0.16436704397201538, "step": 16110 }, { "epoch": 0.42462537413196444, "grad_norm": 0.8931851387023926, "learning_rate": 1.2886310319310787e-05, "loss": 0.1539754867553711, "step": 16120 }, { "epoch": 0.4248887893764632, "grad_norm": 2.5316250324249268, "learning_rate": 1.2878140989062233e-05, "loss": 0.16133629083633422, "step": 16130 }, { "epoch": 0.4251522046209619, "grad_norm": 1.3136898279190063, "learning_rate": 1.2869969563978493e-05, "loss": 0.16274101734161378, "step": 16140 }, { "epoch": 0.42541561986546067, "grad_norm": 2.459261655807495, "learning_rate": 1.2861796050007082e-05, "loss": 0.1510681390762329, "step": 16150 }, { "epoch": 0.4256790351099594, "grad_norm": 1.9380046129226685, "learning_rate": 1.2853620453097039e-05, "loss": 0.16797888278961182, "step": 16160 }, { "epoch": 0.42594245035445816, "grad_norm": 0.8184524774551392, "learning_rate": 1.2845442779198916e-05, "loss": 0.14595274925231932, "step": 16170 }, { "epoch": 0.4262058655989569, "grad_norm": 0.6704291701316833, "learning_rate": 1.2837263034264773e-05, "loss": 0.14300341606140138, "step": 16180 }, { "epoch": 0.4264692808434556, "grad_norm": 1.3706048727035522, "learning_rate": 1.2829081224248183e-05, "loss": 0.15042531490325928, "step": 16190 }, { "epoch": 0.42673269608795433, "grad_norm": 2.152024269104004, "learning_rate": 1.2820897355104224e-05, "loss": 0.16412477493286132, "step": 16200 }, { "epoch": 0.4269961113324531, "grad_norm": 2.1982626914978027, "learning_rate": 1.2812711432789464e-05, "loss": 0.1532210350036621, "step": 16210 }, { "epoch": 0.4272595265769518, "grad_norm": 1.3960427045822144, "learning_rate": 1.2804523463261975e-05, "loss": 0.14129047393798827, "step": 16220 }, { "epoch": 0.42752294182145056, "grad_norm": 1.4890245199203491, "learning_rate": 1.2796333452481309e-05, "loss": 0.15915040969848632, "step": 16230 }, { "epoch": 0.4277863570659493, "grad_norm": 2.8152265548706055, "learning_rate": 1.2788141406408516e-05, "loss": 0.15883623361587523, "step": 16240 }, { "epoch": 0.42804977231044805, "grad_norm": 0.9891716837882996, "learning_rate": 1.2779947331006113e-05, "loss": 0.14837379455566407, "step": 16250 }, { "epoch": 0.4283131875549468, "grad_norm": 2.013589382171631, "learning_rate": 1.2771751232238107e-05, "loss": 0.14349472522735596, "step": 16260 }, { "epoch": 0.42857660279944554, "grad_norm": 4.149853229522705, "learning_rate": 1.2763553116069969e-05, "loss": 0.15102829933166503, "step": 16270 }, { "epoch": 0.4288400180439442, "grad_norm": 1.407572865486145, "learning_rate": 1.2755352988468646e-05, "loss": 0.13931517601013182, "step": 16280 }, { "epoch": 0.42910343328844297, "grad_norm": 2.255560874938965, "learning_rate": 1.2747150855402541e-05, "loss": 0.15986242294311523, "step": 16290 }, { "epoch": 0.4293668485329417, "grad_norm": 1.1836609840393066, "learning_rate": 1.2738946722841519e-05, "loss": 0.14679101705551148, "step": 16300 }, { "epoch": 0.42963026377744046, "grad_norm": 3.501180410385132, "learning_rate": 1.2730740596756908e-05, "loss": 0.14036480188369752, "step": 16310 }, { "epoch": 0.4298936790219392, "grad_norm": 3.070729970932007, "learning_rate": 1.2722532483121475e-05, "loss": 0.16050688028335572, "step": 16320 }, { "epoch": 0.43015709426643794, "grad_norm": 1.4617079496383667, "learning_rate": 1.271432238790944e-05, "loss": 0.14965178966522216, "step": 16330 }, { "epoch": 0.4304205095109367, "grad_norm": 2.7742016315460205, "learning_rate": 1.2706110317096463e-05, "loss": 0.1476892352104187, "step": 16340 }, { "epoch": 0.43068392475543543, "grad_norm": 3.599971055984497, "learning_rate": 1.269789627665965e-05, "loss": 0.16447643041610718, "step": 16350 }, { "epoch": 0.4309473399999341, "grad_norm": 1.9928582906723022, "learning_rate": 1.2689680272577527e-05, "loss": 0.17222787141799928, "step": 16360 }, { "epoch": 0.43121075524443286, "grad_norm": 1.9023810625076294, "learning_rate": 1.2681462310830061e-05, "loss": 0.15792570114135743, "step": 16370 }, { "epoch": 0.4314741704889316, "grad_norm": 1.9388517141342163, "learning_rate": 1.2673242397398635e-05, "loss": 0.15353498458862305, "step": 16380 }, { "epoch": 0.43173758573343035, "grad_norm": 0.840318500995636, "learning_rate": 1.266502053826606e-05, "loss": 0.1579735279083252, "step": 16390 }, { "epoch": 0.4320010009779291, "grad_norm": 3.224992275238037, "learning_rate": 1.2656796739416557e-05, "loss": 0.16110695600509645, "step": 16400 }, { "epoch": 0.43226441622242784, "grad_norm": 1.6280087232589722, "learning_rate": 1.2648571006835764e-05, "loss": 0.14642176628112794, "step": 16410 }, { "epoch": 0.4325278314669266, "grad_norm": 1.3453718423843384, "learning_rate": 1.2640343346510724e-05, "loss": 0.15483520030975342, "step": 16420 }, { "epoch": 0.4327912467114253, "grad_norm": 0.733833372592926, "learning_rate": 1.2632113764429884e-05, "loss": 0.1586972951889038, "step": 16430 }, { "epoch": 0.43305466195592407, "grad_norm": 1.7953180074691772, "learning_rate": 1.2623882266583083e-05, "loss": 0.15607495307922364, "step": 16440 }, { "epoch": 0.43331807720042276, "grad_norm": 1.4227256774902344, "learning_rate": 1.2615648858961564e-05, "loss": 0.1504669427871704, "step": 16450 }, { "epoch": 0.4335814924449215, "grad_norm": 2.6498939990997314, "learning_rate": 1.260741354755796e-05, "loss": 0.15226055383682252, "step": 16460 }, { "epoch": 0.43384490768942024, "grad_norm": 0.785519003868103, "learning_rate": 1.2599176338366282e-05, "loss": 0.17093273401260375, "step": 16470 }, { "epoch": 0.434108322933919, "grad_norm": 2.7746400833129883, "learning_rate": 1.2590937237381927e-05, "loss": 0.1448938012123108, "step": 16480 }, { "epoch": 0.43437173817841773, "grad_norm": 1.1745799779891968, "learning_rate": 1.2582696250601667e-05, "loss": 0.1595979928970337, "step": 16490 }, { "epoch": 0.4346351534229165, "grad_norm": 2.1279296875, "learning_rate": 1.257445338402365e-05, "loss": 0.15233910083770752, "step": 16500 }, { "epoch": 0.4348985686674152, "grad_norm": 0.9658538699150085, "learning_rate": 1.256620864364739e-05, "loss": 0.16380343437194825, "step": 16510 }, { "epoch": 0.43516198391191396, "grad_norm": 1.3396894931793213, "learning_rate": 1.255796203547376e-05, "loss": 0.15208616256713867, "step": 16520 }, { "epoch": 0.4354253991564127, "grad_norm": 1.2936817407608032, "learning_rate": 1.2549713565505001e-05, "loss": 0.14982306957244873, "step": 16530 }, { "epoch": 0.4356888144009114, "grad_norm": 1.7822065353393555, "learning_rate": 1.254146323974471e-05, "loss": 0.14759708642959596, "step": 16540 }, { "epoch": 0.43595222964541014, "grad_norm": 1.6857446432113647, "learning_rate": 1.253321106419782e-05, "loss": 0.14177157878875732, "step": 16550 }, { "epoch": 0.4362156448899089, "grad_norm": 0.8702084422111511, "learning_rate": 1.2524957044870627e-05, "loss": 0.15132949352264405, "step": 16560 }, { "epoch": 0.4364790601344076, "grad_norm": 1.044737458229065, "learning_rate": 1.2516701187770761e-05, "loss": 0.15256140232086182, "step": 16570 }, { "epoch": 0.43674247537890637, "grad_norm": 2.1775054931640625, "learning_rate": 1.250844349890719e-05, "loss": 0.1530524730682373, "step": 16580 }, { "epoch": 0.4370058906234051, "grad_norm": 1.3586595058441162, "learning_rate": 1.250018398429022e-05, "loss": 0.1616924524307251, "step": 16590 }, { "epoch": 0.43726930586790386, "grad_norm": 3.5294134616851807, "learning_rate": 1.2491922649931475e-05, "loss": 0.1633819818496704, "step": 16600 }, { "epoch": 0.4375327211124026, "grad_norm": 3.1379551887512207, "learning_rate": 1.2483659501843916e-05, "loss": 0.16031503677368164, "step": 16610 }, { "epoch": 0.4377961363569013, "grad_norm": 1.0025608539581299, "learning_rate": 1.2475394546041822e-05, "loss": 0.14959181547164918, "step": 16620 }, { "epoch": 0.43805955160140003, "grad_norm": 1.792577862739563, "learning_rate": 1.2467127788540774e-05, "loss": 0.13923563957214355, "step": 16630 }, { "epoch": 0.4383229668458988, "grad_norm": 1.6120823621749878, "learning_rate": 1.245885923535768e-05, "loss": 0.15544087886810304, "step": 16640 }, { "epoch": 0.4385863820903975, "grad_norm": 1.0096666812896729, "learning_rate": 1.2450588892510756e-05, "loss": 0.15784783363342286, "step": 16650 }, { "epoch": 0.43884979733489626, "grad_norm": 0.843782365322113, "learning_rate": 1.2442316766019502e-05, "loss": 0.16103837490081788, "step": 16660 }, { "epoch": 0.439113212579395, "grad_norm": 2.0222620964050293, "learning_rate": 1.2434042861904743e-05, "loss": 0.15182267427444457, "step": 16670 }, { "epoch": 0.43937662782389375, "grad_norm": 2.5627706050872803, "learning_rate": 1.2425767186188569e-05, "loss": 0.1500542640686035, "step": 16680 }, { "epoch": 0.4396400430683925, "grad_norm": 2.2375450134277344, "learning_rate": 1.2417489744894381e-05, "loss": 0.14976048469543457, "step": 16690 }, { "epoch": 0.43990345831289124, "grad_norm": 2.93941593170166, "learning_rate": 1.2409210544046859e-05, "loss": 0.1452588438987732, "step": 16700 }, { "epoch": 0.4401668735573899, "grad_norm": 2.584886074066162, "learning_rate": 1.2400929589671962e-05, "loss": 0.15288645029067993, "step": 16710 }, { "epoch": 0.44043028880188867, "grad_norm": 3.514571189880371, "learning_rate": 1.2392646887796924e-05, "loss": 0.1682556986808777, "step": 16720 }, { "epoch": 0.4406937040463874, "grad_norm": 0.6813005805015564, "learning_rate": 1.2384362444450253e-05, "loss": 0.149363112449646, "step": 16730 }, { "epoch": 0.44095711929088616, "grad_norm": 0.6856755614280701, "learning_rate": 1.2376076265661722e-05, "loss": 0.16377458572387696, "step": 16740 }, { "epoch": 0.4412205345353849, "grad_norm": 1.7517497539520264, "learning_rate": 1.2367788357462375e-05, "loss": 0.16089198589324952, "step": 16750 }, { "epoch": 0.44148394977988364, "grad_norm": 2.283170223236084, "learning_rate": 1.2359498725884509e-05, "loss": 0.15390570163726808, "step": 16760 }, { "epoch": 0.4417473650243824, "grad_norm": 1.8384474515914917, "learning_rate": 1.2351207376961666e-05, "loss": 0.155115807056427, "step": 16770 }, { "epoch": 0.44201078026888113, "grad_norm": 2.2271714210510254, "learning_rate": 1.234291431672866e-05, "loss": 0.16156879663467408, "step": 16780 }, { "epoch": 0.4422741955133799, "grad_norm": 1.3978924751281738, "learning_rate": 1.2334619551221531e-05, "loss": 0.14523510932922362, "step": 16790 }, { "epoch": 0.44253761075787856, "grad_norm": 1.613507628440857, "learning_rate": 1.232632308647757e-05, "loss": 0.16035012006759644, "step": 16800 }, { "epoch": 0.4428010260023773, "grad_norm": 2.8048362731933594, "learning_rate": 1.2318024928535296e-05, "loss": 0.15323270559310914, "step": 16810 }, { "epoch": 0.44306444124687605, "grad_norm": 1.8448997735977173, "learning_rate": 1.2309725083434477e-05, "loss": 0.15517882108688355, "step": 16820 }, { "epoch": 0.4433278564913748, "grad_norm": 0.9704386591911316, "learning_rate": 1.2301423557216088e-05, "loss": 0.1558102250099182, "step": 16830 }, { "epoch": 0.44359127173587354, "grad_norm": 2.4688303470611572, "learning_rate": 1.2293120355922346e-05, "loss": 0.14799141883850098, "step": 16840 }, { "epoch": 0.4438546869803723, "grad_norm": 0.6211060881614685, "learning_rate": 1.2284815485596671e-05, "loss": 0.13563356399536133, "step": 16850 }, { "epoch": 0.444118102224871, "grad_norm": 2.3960068225860596, "learning_rate": 1.2276508952283712e-05, "loss": 0.15039277076721191, "step": 16860 }, { "epoch": 0.44438151746936977, "grad_norm": 2.7696499824523926, "learning_rate": 1.2268200762029319e-05, "loss": 0.14846346378326417, "step": 16870 }, { "epoch": 0.44464493271386846, "grad_norm": 1.3154606819152832, "learning_rate": 1.2259890920880555e-05, "loss": 0.1540623903274536, "step": 16880 }, { "epoch": 0.4449083479583672, "grad_norm": 2.5087063312530518, "learning_rate": 1.2251579434885679e-05, "loss": 0.15963618755340575, "step": 16890 }, { "epoch": 0.44517176320286594, "grad_norm": 1.2330876588821411, "learning_rate": 1.2243266310094146e-05, "loss": 0.14884456396102905, "step": 16900 }, { "epoch": 0.4454351784473647, "grad_norm": 1.6888554096221924, "learning_rate": 1.2234951552556613e-05, "loss": 0.15220367908477783, "step": 16910 }, { "epoch": 0.44569859369186343, "grad_norm": 1.3105894327163696, "learning_rate": 1.2226635168324916e-05, "loss": 0.15127904415130616, "step": 16920 }, { "epoch": 0.4459620089363622, "grad_norm": 2.0725796222686768, "learning_rate": 1.2218317163452078e-05, "loss": 0.14357392787933348, "step": 16930 }, { "epoch": 0.4462254241808609, "grad_norm": 2.207306385040283, "learning_rate": 1.2209997543992301e-05, "loss": 0.14926642179489136, "step": 16940 }, { "epoch": 0.44648883942535966, "grad_norm": 4.336031436920166, "learning_rate": 1.2201676316000964e-05, "loss": 0.15915724039077758, "step": 16950 }, { "epoch": 0.4467522546698584, "grad_norm": 1.5278472900390625, "learning_rate": 1.2193353485534614e-05, "loss": 0.1512095093727112, "step": 16960 }, { "epoch": 0.4470156699143571, "grad_norm": 0.975974977016449, "learning_rate": 1.2185029058650972e-05, "loss": 0.14823853969573975, "step": 16970 }, { "epoch": 0.44727908515885584, "grad_norm": 1.4518502950668335, "learning_rate": 1.2176703041408908e-05, "loss": 0.15170979499816895, "step": 16980 }, { "epoch": 0.4475425004033546, "grad_norm": 3.5906829833984375, "learning_rate": 1.2168375439868464e-05, "loss": 0.15460413694381714, "step": 16990 }, { "epoch": 0.4478059156478533, "grad_norm": 0.6094316840171814, "learning_rate": 1.2160046260090816e-05, "loss": 0.14290096759796142, "step": 17000 }, { "epoch": 0.44806933089235207, "grad_norm": 2.4098026752471924, "learning_rate": 1.2151715508138313e-05, "loss": 0.14526081085205078, "step": 17010 }, { "epoch": 0.4483327461368508, "grad_norm": 3.0537002086639404, "learning_rate": 1.2143383190074429e-05, "loss": 0.15323188304901122, "step": 17020 }, { "epoch": 0.44859616138134956, "grad_norm": 3.506859302520752, "learning_rate": 1.2135049311963787e-05, "loss": 0.15954464673995972, "step": 17030 }, { "epoch": 0.4488595766258483, "grad_norm": 1.0281262397766113, "learning_rate": 1.2126713879872138e-05, "loss": 0.15523561239242553, "step": 17040 }, { "epoch": 0.44912299187034704, "grad_norm": 1.040500521659851, "learning_rate": 1.2118376899866375e-05, "loss": 0.1658540725708008, "step": 17050 }, { "epoch": 0.44938640711484573, "grad_norm": 4.428962230682373, "learning_rate": 1.2110038378014512e-05, "loss": 0.1528918743133545, "step": 17060 }, { "epoch": 0.4496498223593445, "grad_norm": 1.5440220832824707, "learning_rate": 1.210169832038568e-05, "loss": 0.14743118286132811, "step": 17070 }, { "epoch": 0.4499132376038432, "grad_norm": 1.1829328536987305, "learning_rate": 1.2093356733050137e-05, "loss": 0.14337663650512694, "step": 17080 }, { "epoch": 0.45017665284834196, "grad_norm": 4.742575168609619, "learning_rate": 1.2085013622079248e-05, "loss": 0.16484321355819703, "step": 17090 }, { "epoch": 0.4504400680928407, "grad_norm": 2.406095266342163, "learning_rate": 1.2076668993545491e-05, "loss": 0.1508827805519104, "step": 17100 }, { "epoch": 0.45070348333733945, "grad_norm": 3.477672815322876, "learning_rate": 1.206832285352244e-05, "loss": 0.1681915521621704, "step": 17110 }, { "epoch": 0.4509668985818382, "grad_norm": 1.6720026731491089, "learning_rate": 1.2059975208084787e-05, "loss": 0.15151877403259278, "step": 17120 }, { "epoch": 0.45123031382633694, "grad_norm": 3.4418601989746094, "learning_rate": 1.2051626063308299e-05, "loss": 0.1463625431060791, "step": 17130 }, { "epoch": 0.4514937290708356, "grad_norm": 1.3110085725784302, "learning_rate": 1.2043275425269854e-05, "loss": 0.15457770824432374, "step": 17140 }, { "epoch": 0.45175714431533437, "grad_norm": 1.6228764057159424, "learning_rate": 1.2034923300047391e-05, "loss": 0.15203118324279785, "step": 17150 }, { "epoch": 0.4520205595598331, "grad_norm": 2.663299560546875, "learning_rate": 1.2026569693719963e-05, "loss": 0.1463995933532715, "step": 17160 }, { "epoch": 0.45228397480433186, "grad_norm": 1.377143144607544, "learning_rate": 1.2018214612367673e-05, "loss": 0.1466726303100586, "step": 17170 }, { "epoch": 0.4525473900488306, "grad_norm": 1.1237938404083252, "learning_rate": 1.200985806207172e-05, "loss": 0.15160555839538575, "step": 17180 }, { "epoch": 0.45281080529332934, "grad_norm": 3.5162315368652344, "learning_rate": 1.2001500048914354e-05, "loss": 0.13842417001724244, "step": 17190 }, { "epoch": 0.4530742205378281, "grad_norm": 0.6841759085655212, "learning_rate": 1.1993140578978898e-05, "loss": 0.15774205923080445, "step": 17200 }, { "epoch": 0.45333763578232683, "grad_norm": 0.9330343008041382, "learning_rate": 1.1984779658349743e-05, "loss": 0.1553006052970886, "step": 17210 }, { "epoch": 0.4536010510268256, "grad_norm": 2.4203109741210938, "learning_rate": 1.1976417293112318e-05, "loss": 0.14428286552429198, "step": 17220 }, { "epoch": 0.45386446627132426, "grad_norm": 1.5743408203125, "learning_rate": 1.1968053489353122e-05, "loss": 0.1421884059906006, "step": 17230 }, { "epoch": 0.454127881515823, "grad_norm": 2.0298633575439453, "learning_rate": 1.195968825315968e-05, "loss": 0.15771849155426027, "step": 17240 }, { "epoch": 0.45439129676032175, "grad_norm": 2.1472530364990234, "learning_rate": 1.1951321590620583e-05, "loss": 0.15609338283538818, "step": 17250 }, { "epoch": 0.4546547120048205, "grad_norm": 3.2225918769836426, "learning_rate": 1.1942953507825446e-05, "loss": 0.15740573406219482, "step": 17260 }, { "epoch": 0.45491812724931924, "grad_norm": 1.8985354900360107, "learning_rate": 1.1934584010864921e-05, "loss": 0.15575696229934693, "step": 17270 }, { "epoch": 0.455181542493818, "grad_norm": 1.3761630058288574, "learning_rate": 1.1926213105830683e-05, "loss": 0.15058035850524903, "step": 17280 }, { "epoch": 0.4554449577383167, "grad_norm": 1.9935176372528076, "learning_rate": 1.1917840798815448e-05, "loss": 0.16145122051239014, "step": 17290 }, { "epoch": 0.45570837298281547, "grad_norm": 2.7763302326202393, "learning_rate": 1.1909467095912932e-05, "loss": 0.1642608165740967, "step": 17300 }, { "epoch": 0.4559717882273142, "grad_norm": 3.616373300552368, "learning_rate": 1.1901092003217885e-05, "loss": 0.15474588871002198, "step": 17310 }, { "epoch": 0.4562352034718129, "grad_norm": 2.072603464126587, "learning_rate": 1.1892715526826058e-05, "loss": 0.15555708408355712, "step": 17320 }, { "epoch": 0.45649861871631164, "grad_norm": 0.7418891191482544, "learning_rate": 1.1884337672834212e-05, "loss": 0.14445044994354247, "step": 17330 }, { "epoch": 0.4567620339608104, "grad_norm": 1.719552755355835, "learning_rate": 1.1875958447340108e-05, "loss": 0.14371089935302733, "step": 17340 }, { "epoch": 0.45702544920530913, "grad_norm": 4.1286468505859375, "learning_rate": 1.1867577856442514e-05, "loss": 0.14792757034301757, "step": 17350 }, { "epoch": 0.4572888644498079, "grad_norm": 1.6863950490951538, "learning_rate": 1.1859195906241182e-05, "loss": 0.15424163341522218, "step": 17360 }, { "epoch": 0.4575522796943066, "grad_norm": 5.526644229888916, "learning_rate": 1.1850812602836851e-05, "loss": 0.15839929580688478, "step": 17370 }, { "epoch": 0.45781569493880536, "grad_norm": 1.293627381324768, "learning_rate": 1.1842427952331263e-05, "loss": 0.14752079248428346, "step": 17380 }, { "epoch": 0.4580791101833041, "grad_norm": 2.4576900005340576, "learning_rate": 1.183404196082712e-05, "loss": 0.1625518560409546, "step": 17390 }, { "epoch": 0.4583425254278028, "grad_norm": 1.7171216011047363, "learning_rate": 1.1825654634428111e-05, "loss": 0.14800361394882203, "step": 17400 }, { "epoch": 0.45860594067230154, "grad_norm": 4.142831802368164, "learning_rate": 1.1817265979238892e-05, "loss": 0.15765713453292846, "step": 17410 }, { "epoch": 0.4588693559168003, "grad_norm": 0.7545953989028931, "learning_rate": 1.1808876001365093e-05, "loss": 0.1666802167892456, "step": 17420 }, { "epoch": 0.459132771161299, "grad_norm": 4.254464149475098, "learning_rate": 1.1800484706913294e-05, "loss": 0.1576477289199829, "step": 17430 }, { "epoch": 0.45939618640579777, "grad_norm": 2.8825042247772217, "learning_rate": 1.1792092101991052e-05, "loss": 0.15233781337738037, "step": 17440 }, { "epoch": 0.4596596016502965, "grad_norm": 4.003962516784668, "learning_rate": 1.1783698192706856e-05, "loss": 0.15821213722229005, "step": 17450 }, { "epoch": 0.45992301689479526, "grad_norm": 2.2560958862304688, "learning_rate": 1.1775302985170161e-05, "loss": 0.15542783737182617, "step": 17460 }, { "epoch": 0.460186432139294, "grad_norm": 1.5766375064849854, "learning_rate": 1.1766906485491361e-05, "loss": 0.1602041244506836, "step": 17470 }, { "epoch": 0.46044984738379274, "grad_norm": 1.4053819179534912, "learning_rate": 1.175850869978179e-05, "loss": 0.16311244964599608, "step": 17480 }, { "epoch": 0.46071326262829143, "grad_norm": 1.016176700592041, "learning_rate": 1.1750109634153719e-05, "loss": 0.14313979148864747, "step": 17490 }, { "epoch": 0.4609766778727902, "grad_norm": 3.3100876808166504, "learning_rate": 1.1741709294720346e-05, "loss": 0.1410536527633667, "step": 17500 }, { "epoch": 0.4612400931172889, "grad_norm": 3.2843706607818604, "learning_rate": 1.173330768759581e-05, "loss": 0.14986555576324462, "step": 17510 }, { "epoch": 0.46150350836178766, "grad_norm": 2.445256233215332, "learning_rate": 1.1724904818895158e-05, "loss": 0.14183114767074584, "step": 17520 }, { "epoch": 0.4617669236062864, "grad_norm": 1.900673747062683, "learning_rate": 1.1716500694734359e-05, "loss": 0.16144129037857055, "step": 17530 }, { "epoch": 0.46203033885078515, "grad_norm": 1.3088933229446411, "learning_rate": 1.1708095321230295e-05, "loss": 0.14926098585128783, "step": 17540 }, { "epoch": 0.4622937540952839, "grad_norm": 1.7969093322753906, "learning_rate": 1.1699688704500765e-05, "loss": 0.14005894660949708, "step": 17550 }, { "epoch": 0.46255716933978264, "grad_norm": 2.091113567352295, "learning_rate": 1.1691280850664463e-05, "loss": 0.15307278633117677, "step": 17560 }, { "epoch": 0.4628205845842814, "grad_norm": 1.7107869386672974, "learning_rate": 1.1682871765840996e-05, "loss": 0.15209950208663942, "step": 17570 }, { "epoch": 0.46308399982878007, "grad_norm": 2.5239040851593018, "learning_rate": 1.1674461456150845e-05, "loss": 0.1541449785232544, "step": 17580 }, { "epoch": 0.4633474150732788, "grad_norm": 3.7186641693115234, "learning_rate": 1.1666049927715408e-05, "loss": 0.15303763151168823, "step": 17590 }, { "epoch": 0.46361083031777756, "grad_norm": 1.4574060440063477, "learning_rate": 1.1657637186656952e-05, "loss": 0.1382211446762085, "step": 17600 }, { "epoch": 0.4638742455622763, "grad_norm": 1.10658597946167, "learning_rate": 1.1649223239098639e-05, "loss": 0.1593904972076416, "step": 17610 }, { "epoch": 0.46413766080677504, "grad_norm": 2.680229425430298, "learning_rate": 1.1640808091164495e-05, "loss": 0.15100271701812745, "step": 17620 }, { "epoch": 0.4644010760512738, "grad_norm": 2.8750386238098145, "learning_rate": 1.1632391748979435e-05, "loss": 0.1471303939819336, "step": 17630 }, { "epoch": 0.46466449129577253, "grad_norm": 1.0483840703964233, "learning_rate": 1.162397421866923e-05, "loss": 0.15784497261047364, "step": 17640 }, { "epoch": 0.4649279065402713, "grad_norm": 3.2711267471313477, "learning_rate": 1.1615555506360526e-05, "loss": 0.15180368423461915, "step": 17650 }, { "epoch": 0.46519132178476996, "grad_norm": 2.2520737648010254, "learning_rate": 1.1607135618180823e-05, "loss": 0.16434746980667114, "step": 17660 }, { "epoch": 0.4654547370292687, "grad_norm": 1.671573281288147, "learning_rate": 1.1598714560258477e-05, "loss": 0.14850420951843263, "step": 17670 }, { "epoch": 0.46571815227376745, "grad_norm": 2.0916104316711426, "learning_rate": 1.1590292338722702e-05, "loss": 0.15687665939331055, "step": 17680 }, { "epoch": 0.4659815675182662, "grad_norm": 1.3735636472702026, "learning_rate": 1.1581868959703553e-05, "loss": 0.14345301389694215, "step": 17690 }, { "epoch": 0.46624498276276494, "grad_norm": 3.9386298656463623, "learning_rate": 1.1573444429331926e-05, "loss": 0.14569568634033203, "step": 17700 }, { "epoch": 0.4665083980072637, "grad_norm": 1.335404396057129, "learning_rate": 1.1565018753739559e-05, "loss": 0.150396466255188, "step": 17710 }, { "epoch": 0.4667718132517624, "grad_norm": 1.6026393175125122, "learning_rate": 1.1556591939059024e-05, "loss": 0.15038795471191407, "step": 17720 }, { "epoch": 0.46703522849626117, "grad_norm": 3.409907102584839, "learning_rate": 1.1548163991423716e-05, "loss": 0.148020339012146, "step": 17730 }, { "epoch": 0.4672986437407599, "grad_norm": 1.6803288459777832, "learning_rate": 1.1539734916967864e-05, "loss": 0.151039457321167, "step": 17740 }, { "epoch": 0.4675620589852586, "grad_norm": 1.1101127862930298, "learning_rate": 1.1531304721826506e-05, "loss": 0.14011588096618652, "step": 17750 }, { "epoch": 0.46782547422975734, "grad_norm": 2.5106821060180664, "learning_rate": 1.1522873412135509e-05, "loss": 0.1470681071281433, "step": 17760 }, { "epoch": 0.4680888894742561, "grad_norm": 1.4112448692321777, "learning_rate": 1.1514440994031536e-05, "loss": 0.1403361439704895, "step": 17770 }, { "epoch": 0.46835230471875483, "grad_norm": 1.252746343612671, "learning_rate": 1.1506007473652075e-05, "loss": 0.1554647445678711, "step": 17780 }, { "epoch": 0.4686157199632536, "grad_norm": 4.849606513977051, "learning_rate": 1.1497572857135397e-05, "loss": 0.15244724750518798, "step": 17790 }, { "epoch": 0.4688791352077523, "grad_norm": 3.4741158485412598, "learning_rate": 1.1489137150620582e-05, "loss": 0.16825127601623535, "step": 17800 }, { "epoch": 0.46914255045225106, "grad_norm": 1.2704617977142334, "learning_rate": 1.1480700360247504e-05, "loss": 0.15788123607635499, "step": 17810 }, { "epoch": 0.4694059656967498, "grad_norm": 1.8402987718582153, "learning_rate": 1.147226249215682e-05, "loss": 0.15683989524841307, "step": 17820 }, { "epoch": 0.46966938094124855, "grad_norm": 1.2682667970657349, "learning_rate": 1.1463823552489977e-05, "loss": 0.1555035352706909, "step": 17830 }, { "epoch": 0.46993279618574724, "grad_norm": 2.2926342487335205, "learning_rate": 1.1455383547389193e-05, "loss": 0.1440472960472107, "step": 17840 }, { "epoch": 0.470196211430246, "grad_norm": 1.4886294603347778, "learning_rate": 1.1446942482997479e-05, "loss": 0.13965657949447632, "step": 17850 }, { "epoch": 0.4704596266747447, "grad_norm": 1.1946781873703003, "learning_rate": 1.1438500365458593e-05, "loss": 0.14270827770233155, "step": 17860 }, { "epoch": 0.47072304191924347, "grad_norm": 1.507462978363037, "learning_rate": 1.1430057200917084e-05, "loss": 0.15369625091552735, "step": 17870 }, { "epoch": 0.4709864571637422, "grad_norm": 2.540231943130493, "learning_rate": 1.1421612995518243e-05, "loss": 0.14642118215560912, "step": 17880 }, { "epoch": 0.47124987240824096, "grad_norm": 2.0531575679779053, "learning_rate": 1.1413167755408132e-05, "loss": 0.16015431880950928, "step": 17890 }, { "epoch": 0.4715132876527397, "grad_norm": 2.4448423385620117, "learning_rate": 1.1404721486733558e-05, "loss": 0.15582799911499023, "step": 17900 }, { "epoch": 0.47177670289723844, "grad_norm": 2.101017475128174, "learning_rate": 1.1396274195642085e-05, "loss": 0.14349815845489503, "step": 17910 }, { "epoch": 0.47204011814173713, "grad_norm": 1.5290327072143555, "learning_rate": 1.1387825888282006e-05, "loss": 0.15570768117904663, "step": 17920 }, { "epoch": 0.4723035333862359, "grad_norm": 1.6524473428726196, "learning_rate": 1.1379376570802373e-05, "loss": 0.1417905330657959, "step": 17930 }, { "epoch": 0.4725669486307346, "grad_norm": 1.3382726907730103, "learning_rate": 1.137092624935296e-05, "loss": 0.1486682653427124, "step": 17940 }, { "epoch": 0.47283036387523336, "grad_norm": 3.1110129356384277, "learning_rate": 1.1362474930084278e-05, "loss": 0.15991647243499757, "step": 17950 }, { "epoch": 0.4730937791197321, "grad_norm": 0.9287854433059692, "learning_rate": 1.1354022619147557e-05, "loss": 0.14234864711761475, "step": 17960 }, { "epoch": 0.47335719436423085, "grad_norm": 1.36696457862854, "learning_rate": 1.1345569322694756e-05, "loss": 0.15225754976272582, "step": 17970 }, { "epoch": 0.4736206096087296, "grad_norm": 2.917004346847534, "learning_rate": 1.133711504687855e-05, "loss": 0.16267759799957277, "step": 17980 }, { "epoch": 0.47388402485322834, "grad_norm": 1.446702241897583, "learning_rate": 1.1328659797852325e-05, "loss": 0.145277738571167, "step": 17990 }, { "epoch": 0.4741474400977271, "grad_norm": 2.7443671226501465, "learning_rate": 1.1320203581770177e-05, "loss": 0.1581435203552246, "step": 18000 }, { "epoch": 0.47441085534222577, "grad_norm": 4.902108669281006, "learning_rate": 1.1311746404786903e-05, "loss": 0.15877606868743896, "step": 18010 }, { "epoch": 0.4746742705867245, "grad_norm": 3.2380619049072266, "learning_rate": 1.1303288273058002e-05, "loss": 0.14769461154937744, "step": 18020 }, { "epoch": 0.47493768583122326, "grad_norm": 2.823505401611328, "learning_rate": 1.1294829192739665e-05, "loss": 0.14721492528915406, "step": 18030 }, { "epoch": 0.475201101075722, "grad_norm": 0.9288403391838074, "learning_rate": 1.1286369169988782e-05, "loss": 0.15723743438720703, "step": 18040 }, { "epoch": 0.47546451632022074, "grad_norm": 1.2161459922790527, "learning_rate": 1.1277908210962916e-05, "loss": 0.15905768871307374, "step": 18050 }, { "epoch": 0.4757279315647195, "grad_norm": 2.5864694118499756, "learning_rate": 1.126944632182032e-05, "loss": 0.15601871013641358, "step": 18060 }, { "epoch": 0.47599134680921823, "grad_norm": 1.5415196418762207, "learning_rate": 1.1260983508719922e-05, "loss": 0.13970420360565186, "step": 18070 }, { "epoch": 0.476254762053717, "grad_norm": 2.0478675365448, "learning_rate": 1.1252519777821324e-05, "loss": 0.1522207736968994, "step": 18080 }, { "epoch": 0.4765181772982157, "grad_norm": 2.790027141571045, "learning_rate": 1.1244055135284795e-05, "loss": 0.14770933389663696, "step": 18090 }, { "epoch": 0.4767815925427144, "grad_norm": 1.6300212144851685, "learning_rate": 1.123558958727126e-05, "loss": 0.1545116901397705, "step": 18100 }, { "epoch": 0.47704500778721315, "grad_norm": 2.4011518955230713, "learning_rate": 1.1227123139942322e-05, "loss": 0.1472862720489502, "step": 18110 }, { "epoch": 0.4773084230317119, "grad_norm": 1.1298655271530151, "learning_rate": 1.1218655799460216e-05, "loss": 0.15558470487594606, "step": 18120 }, { "epoch": 0.47757183827621064, "grad_norm": 1.3881374597549438, "learning_rate": 1.1210187571987845e-05, "loss": 0.15257219076156617, "step": 18130 }, { "epoch": 0.4778352535207094, "grad_norm": 4.600710868835449, "learning_rate": 1.1201718463688745e-05, "loss": 0.16030758619308472, "step": 18140 }, { "epoch": 0.4780986687652081, "grad_norm": 1.686179280281067, "learning_rate": 1.1193248480727101e-05, "loss": 0.15296366214752197, "step": 18150 }, { "epoch": 0.47836208400970687, "grad_norm": 2.1518571376800537, "learning_rate": 1.118477762926773e-05, "loss": 0.14202895164489746, "step": 18160 }, { "epoch": 0.4786254992542056, "grad_norm": 2.586888074874878, "learning_rate": 1.117630591547609e-05, "loss": 0.15335314273834227, "step": 18170 }, { "epoch": 0.4788889144987043, "grad_norm": 1.7034664154052734, "learning_rate": 1.1167833345518248e-05, "loss": 0.14540750980377198, "step": 18180 }, { "epoch": 0.47915232974320304, "grad_norm": 1.839914321899414, "learning_rate": 1.1159359925560918e-05, "loss": 0.1530157208442688, "step": 18190 }, { "epoch": 0.4794157449877018, "grad_norm": 3.7272439002990723, "learning_rate": 1.115088566177141e-05, "loss": 0.1544398307800293, "step": 18200 }, { "epoch": 0.47967916023220053, "grad_norm": 1.8212954998016357, "learning_rate": 1.1142410560317669e-05, "loss": 0.14970364570617675, "step": 18210 }, { "epoch": 0.4799425754766993, "grad_norm": 1.4447938203811646, "learning_rate": 1.113393462736823e-05, "loss": 0.16209276914596557, "step": 18220 }, { "epoch": 0.480205990721198, "grad_norm": 0.7580074071884155, "learning_rate": 1.1125457869092251e-05, "loss": 0.15339884757995606, "step": 18230 }, { "epoch": 0.48046940596569676, "grad_norm": 1.2526637315750122, "learning_rate": 1.1116980291659474e-05, "loss": 0.1490097999572754, "step": 18240 }, { "epoch": 0.4807328212101955, "grad_norm": 1.885706901550293, "learning_rate": 1.1108501901240254e-05, "loss": 0.13883180618286134, "step": 18250 }, { "epoch": 0.48099623645469425, "grad_norm": 2.4262638092041016, "learning_rate": 1.1100022704005524e-05, "loss": 0.1515493869781494, "step": 18260 }, { "epoch": 0.48125965169919294, "grad_norm": 3.81673002243042, "learning_rate": 1.109154270612681e-05, "loss": 0.14946138858795166, "step": 18270 }, { "epoch": 0.4815230669436917, "grad_norm": 1.129412055015564, "learning_rate": 1.1083061913776224e-05, "loss": 0.14321932792663575, "step": 18280 }, { "epoch": 0.4817864821881904, "grad_norm": 2.3499903678894043, "learning_rate": 1.1074580333126448e-05, "loss": 0.15298417806625367, "step": 18290 }, { "epoch": 0.48204989743268917, "grad_norm": 1.577411413192749, "learning_rate": 1.1066097970350747e-05, "loss": 0.15274157524108886, "step": 18300 }, { "epoch": 0.4823133126771879, "grad_norm": 0.9240908026695251, "learning_rate": 1.1057614831622942e-05, "loss": 0.14244149923324584, "step": 18310 }, { "epoch": 0.48257672792168665, "grad_norm": 0.9216356873512268, "learning_rate": 1.1049130923117437e-05, "loss": 0.15077028274536133, "step": 18320 }, { "epoch": 0.4828401431661854, "grad_norm": 1.164516568183899, "learning_rate": 1.1040646251009181e-05, "loss": 0.15937352180480957, "step": 18330 }, { "epoch": 0.48310355841068414, "grad_norm": 1.2764029502868652, "learning_rate": 1.1032160821473692e-05, "loss": 0.1548241138458252, "step": 18340 }, { "epoch": 0.4833669736551829, "grad_norm": 2.6449694633483887, "learning_rate": 1.1023674640687019e-05, "loss": 0.15503935813903807, "step": 18350 }, { "epoch": 0.4836303888996816, "grad_norm": 0.9908109903335571, "learning_rate": 1.1015187714825782e-05, "loss": 0.1488044500350952, "step": 18360 }, { "epoch": 0.4838938041441803, "grad_norm": 0.8591862916946411, "learning_rate": 1.1006700050067122e-05, "loss": 0.1367645263671875, "step": 18370 }, { "epoch": 0.48415721938867906, "grad_norm": 2.303893804550171, "learning_rate": 1.0998211652588737e-05, "loss": 0.14292932748794557, "step": 18380 }, { "epoch": 0.4844206346331778, "grad_norm": 2.8925344944000244, "learning_rate": 1.0989722528568845e-05, "loss": 0.15510153770446777, "step": 18390 }, { "epoch": 0.48468404987767655, "grad_norm": 2.5006797313690186, "learning_rate": 1.0981232684186191e-05, "loss": 0.1551844596862793, "step": 18400 }, { "epoch": 0.4849474651221753, "grad_norm": 1.5194085836410522, "learning_rate": 1.0972742125620062e-05, "loss": 0.14827566146850585, "step": 18410 }, { "epoch": 0.48521088036667404, "grad_norm": 1.594976544380188, "learning_rate": 1.0964250859050245e-05, "loss": 0.1440058469772339, "step": 18420 }, { "epoch": 0.4854742956111728, "grad_norm": 1.1030200719833374, "learning_rate": 1.0955758890657051e-05, "loss": 0.14295148849487305, "step": 18430 }, { "epoch": 0.48573771085567147, "grad_norm": 2.5781991481781006, "learning_rate": 1.0947266226621298e-05, "loss": 0.12838118076324462, "step": 18440 }, { "epoch": 0.4860011261001702, "grad_norm": 0.8339633941650391, "learning_rate": 1.0938772873124321e-05, "loss": 0.16099655628204346, "step": 18450 }, { "epoch": 0.48626454134466895, "grad_norm": 1.0165364742279053, "learning_rate": 1.0930278836347943e-05, "loss": 0.13798564672470093, "step": 18460 }, { "epoch": 0.4865279565891677, "grad_norm": 0.9736217856407166, "learning_rate": 1.0921784122474495e-05, "loss": 0.14660873413085937, "step": 18470 }, { "epoch": 0.48679137183366644, "grad_norm": 2.88090181350708, "learning_rate": 1.0913288737686789e-05, "loss": 0.16135270595550538, "step": 18480 }, { "epoch": 0.4870547870781652, "grad_norm": 1.8157131671905518, "learning_rate": 1.090479268816814e-05, "loss": 0.15486624240875244, "step": 18490 }, { "epoch": 0.48731820232266393, "grad_norm": 3.0950112342834473, "learning_rate": 1.0896295980102334e-05, "loss": 0.15982117652893066, "step": 18500 }, { "epoch": 0.4875816175671627, "grad_norm": 1.000649094581604, "learning_rate": 1.0887798619673648e-05, "loss": 0.14463115930557252, "step": 18510 }, { "epoch": 0.4878450328116614, "grad_norm": 1.124606966972351, "learning_rate": 1.087930061306682e-05, "loss": 0.14509378671646117, "step": 18520 }, { "epoch": 0.4881084480561601, "grad_norm": 0.9597430229187012, "learning_rate": 1.0870801966467073e-05, "loss": 0.15526185035705567, "step": 18530 }, { "epoch": 0.48837186330065885, "grad_norm": 1.6738057136535645, "learning_rate": 1.0862302686060082e-05, "loss": 0.14601948261260986, "step": 18540 }, { "epoch": 0.4886352785451576, "grad_norm": 2.608100175857544, "learning_rate": 1.0853802778031995e-05, "loss": 0.14954397678375245, "step": 18550 }, { "epoch": 0.48889869378965634, "grad_norm": 3.871042251586914, "learning_rate": 1.0845302248569411e-05, "loss": 0.15190478563308715, "step": 18560 }, { "epoch": 0.4891621090341551, "grad_norm": 1.9043421745300293, "learning_rate": 1.0836801103859377e-05, "loss": 0.14714841842651366, "step": 18570 }, { "epoch": 0.4894255242786538, "grad_norm": 1.3079254627227783, "learning_rate": 1.0828299350089401e-05, "loss": 0.1386800765991211, "step": 18580 }, { "epoch": 0.48968893952315257, "grad_norm": 1.5812489986419678, "learning_rate": 1.0819796993447422e-05, "loss": 0.14838705062866211, "step": 18590 }, { "epoch": 0.4899523547676513, "grad_norm": 1.9352118968963623, "learning_rate": 1.081129404012182e-05, "loss": 0.14702658653259276, "step": 18600 }, { "epoch": 0.49021577001215005, "grad_norm": 2.005181074142456, "learning_rate": 1.0802790496301414e-05, "loss": 0.15306098461151124, "step": 18610 }, { "epoch": 0.49047918525664874, "grad_norm": 2.9405152797698975, "learning_rate": 1.079428636817545e-05, "loss": 0.1526167631149292, "step": 18620 }, { "epoch": 0.4907426005011475, "grad_norm": 0.6744325757026672, "learning_rate": 1.0785781661933596e-05, "loss": 0.14646056890487671, "step": 18630 }, { "epoch": 0.49100601574564623, "grad_norm": 0.7166680097579956, "learning_rate": 1.0777276383765953e-05, "loss": 0.14582784175872804, "step": 18640 }, { "epoch": 0.491269430990145, "grad_norm": 1.7080188989639282, "learning_rate": 1.0768770539863017e-05, "loss": 0.15878746509552003, "step": 18650 }, { "epoch": 0.4915328462346437, "grad_norm": 1.1072964668273926, "learning_rate": 1.076026413641572e-05, "loss": 0.15205140113830568, "step": 18660 }, { "epoch": 0.49179626147914246, "grad_norm": 2.4944565296173096, "learning_rate": 1.0751757179615378e-05, "loss": 0.1599825382232666, "step": 18670 }, { "epoch": 0.4920596767236412, "grad_norm": 3.445615768432617, "learning_rate": 1.0743249675653733e-05, "loss": 0.1552867889404297, "step": 18680 }, { "epoch": 0.49232309196813995, "grad_norm": 1.2225470542907715, "learning_rate": 1.0734741630722904e-05, "loss": 0.15288994312286378, "step": 18690 }, { "epoch": 0.49258650721263864, "grad_norm": 0.8631060123443604, "learning_rate": 1.0726233051015415e-05, "loss": 0.14850203990936278, "step": 18700 }, { "epoch": 0.4928499224571374, "grad_norm": 2.358422040939331, "learning_rate": 1.0717723942724182e-05, "loss": 0.13346234560012818, "step": 18710 }, { "epoch": 0.4931133377016361, "grad_norm": 2.020928144454956, "learning_rate": 1.07092143120425e-05, "loss": 0.14582184553146363, "step": 18720 }, { "epoch": 0.49337675294613487, "grad_norm": 4.209255218505859, "learning_rate": 1.0700704165164042e-05, "loss": 0.14487390518188475, "step": 18730 }, { "epoch": 0.4936401681906336, "grad_norm": 2.2352218627929688, "learning_rate": 1.069219350828286e-05, "loss": 0.15623615980148314, "step": 18740 }, { "epoch": 0.49390358343513235, "grad_norm": 0.9942696690559387, "learning_rate": 1.068368234759338e-05, "loss": 0.1441200017929077, "step": 18750 }, { "epoch": 0.4941669986796311, "grad_norm": 1.756912350654602, "learning_rate": 1.0675170689290393e-05, "loss": 0.14915850162506103, "step": 18760 }, { "epoch": 0.49443041392412984, "grad_norm": 3.1119956970214844, "learning_rate": 1.0666658539569048e-05, "loss": 0.151054310798645, "step": 18770 }, { "epoch": 0.4946938291686286, "grad_norm": 3.4602925777435303, "learning_rate": 1.0658145904624852e-05, "loss": 0.16227080821990966, "step": 18780 }, { "epoch": 0.4949572444131273, "grad_norm": 2.6251614093780518, "learning_rate": 1.0649632790653677e-05, "loss": 0.15145540237426758, "step": 18790 }, { "epoch": 0.495220659657626, "grad_norm": 1.0728830099105835, "learning_rate": 1.0641119203851724e-05, "loss": 0.15538651943206788, "step": 18800 }, { "epoch": 0.49548407490212476, "grad_norm": 1.6357367038726807, "learning_rate": 1.0632605150415561e-05, "loss": 0.15939040184020997, "step": 18810 }, { "epoch": 0.4957474901466235, "grad_norm": 1.8496853113174438, "learning_rate": 1.062409063654207e-05, "loss": 0.15912678241729736, "step": 18820 }, { "epoch": 0.49601090539112225, "grad_norm": 6.232728004455566, "learning_rate": 1.0615575668428494e-05, "loss": 0.1456735372543335, "step": 18830 }, { "epoch": 0.496274320635621, "grad_norm": 1.1766221523284912, "learning_rate": 1.0607060252272383e-05, "loss": 0.1514783501625061, "step": 18840 }, { "epoch": 0.49653773588011973, "grad_norm": 3.01419734954834, "learning_rate": 1.0598544394271633e-05, "loss": 0.1354604721069336, "step": 18850 }, { "epoch": 0.4968011511246185, "grad_norm": 1.2777668237686157, "learning_rate": 1.0590028100624448e-05, "loss": 0.15235435962677002, "step": 18860 }, { "epoch": 0.4970645663691172, "grad_norm": 3.2191736698150635, "learning_rate": 1.0581511377529354e-05, "loss": 0.15945425033569335, "step": 18870 }, { "epoch": 0.4973279816136159, "grad_norm": 1.0032774209976196, "learning_rate": 1.0572994231185194e-05, "loss": 0.16291286945343017, "step": 18880 }, { "epoch": 0.49759139685811465, "grad_norm": 1.215927004814148, "learning_rate": 1.056447666779111e-05, "loss": 0.1517862558364868, "step": 18890 }, { "epoch": 0.4978548121026134, "grad_norm": 1.0105834007263184, "learning_rate": 1.0555958693546552e-05, "loss": 0.14797980785369874, "step": 18900 }, { "epoch": 0.49811822734711214, "grad_norm": 1.5870792865753174, "learning_rate": 1.0547440314651274e-05, "loss": 0.13289071321487428, "step": 18910 }, { "epoch": 0.4983816425916109, "grad_norm": 1.4807571172714233, "learning_rate": 1.0538921537305315e-05, "loss": 0.15948340892791749, "step": 18920 }, { "epoch": 0.49864505783610963, "grad_norm": 2.7839701175689697, "learning_rate": 1.0530402367709007e-05, "loss": 0.15361036062240602, "step": 18930 }, { "epoch": 0.4989084730806084, "grad_norm": 2.1132898330688477, "learning_rate": 1.0521882812062978e-05, "loss": 0.1501086473464966, "step": 18940 }, { "epoch": 0.4991718883251071, "grad_norm": 1.4434572458267212, "learning_rate": 1.0513362876568118e-05, "loss": 0.1470585584640503, "step": 18950 }, { "epoch": 0.49943530356960586, "grad_norm": 3.100132942199707, "learning_rate": 1.0504842567425611e-05, "loss": 0.15921907424926757, "step": 18960 }, { "epoch": 0.49969871881410455, "grad_norm": 0.9632322192192078, "learning_rate": 1.0496321890836901e-05, "loss": 0.15081676244735717, "step": 18970 }, { "epoch": 0.4999621340586033, "grad_norm": 2.255331039428711, "learning_rate": 1.0487800853003706e-05, "loss": 0.15424529314041138, "step": 18980 }, { "epoch": 0.500225549303102, "grad_norm": 3.1794614791870117, "learning_rate": 1.0479279460128005e-05, "loss": 0.15123333930969238, "step": 18990 }, { "epoch": 0.5004889645476008, "grad_norm": 2.185394048690796, "learning_rate": 1.0470757718412035e-05, "loss": 0.13997280597686768, "step": 19000 }, { "epoch": 0.5007523797920995, "grad_norm": 2.014730930328369, "learning_rate": 1.0462235634058287e-05, "loss": 0.14649763107299804, "step": 19010 }, { "epoch": 0.5010157950365982, "grad_norm": 1.3337178230285645, "learning_rate": 1.0453713213269503e-05, "loss": 0.1506080746650696, "step": 19020 }, { "epoch": 0.501279210281097, "grad_norm": 1.4715526103973389, "learning_rate": 1.0445190462248667e-05, "loss": 0.14619245529174804, "step": 19030 }, { "epoch": 0.5015426255255957, "grad_norm": 3.0909969806671143, "learning_rate": 1.0436667387199004e-05, "loss": 0.15015841722488404, "step": 19040 }, { "epoch": 0.5018060407700945, "grad_norm": 2.348781108856201, "learning_rate": 1.0428143994323978e-05, "loss": 0.15770962238311767, "step": 19050 }, { "epoch": 0.5020694560145932, "grad_norm": 1.85309636592865, "learning_rate": 1.041962028982728e-05, "loss": 0.1471175193786621, "step": 19060 }, { "epoch": 0.502332871259092, "grad_norm": 2.870142698287964, "learning_rate": 1.0411096279912835e-05, "loss": 0.15971564054489135, "step": 19070 }, { "epoch": 0.5025962865035907, "grad_norm": 1.7979763746261597, "learning_rate": 1.0402571970784778e-05, "loss": 0.1504366159439087, "step": 19080 }, { "epoch": 0.5028597017480895, "grad_norm": 1.2969671487808228, "learning_rate": 1.0394047368647473e-05, "loss": 0.13969731330871582, "step": 19090 }, { "epoch": 0.5031231169925882, "grad_norm": 2.1090168952941895, "learning_rate": 1.0385522479705493e-05, "loss": 0.1435520887374878, "step": 19100 }, { "epoch": 0.5033865322370868, "grad_norm": 0.7242745757102966, "learning_rate": 1.0376997310163622e-05, "loss": 0.15452883243560792, "step": 19110 }, { "epoch": 0.5036499474815856, "grad_norm": 1.3539619445800781, "learning_rate": 1.0368471866226837e-05, "loss": 0.14657604694366455, "step": 19120 }, { "epoch": 0.5039133627260843, "grad_norm": 4.487184047698975, "learning_rate": 1.0359946154100336e-05, "loss": 0.14804697036743164, "step": 19130 }, { "epoch": 0.5041767779705831, "grad_norm": 1.2895219326019287, "learning_rate": 1.035142017998949e-05, "loss": 0.15176663398742676, "step": 19140 }, { "epoch": 0.5044401932150818, "grad_norm": 1.211864709854126, "learning_rate": 1.0342893950099879e-05, "loss": 0.15475401878356934, "step": 19150 }, { "epoch": 0.5047036084595806, "grad_norm": 0.9599676132202148, "learning_rate": 1.0334367470637257e-05, "loss": 0.14747564792633056, "step": 19160 }, { "epoch": 0.5049670237040793, "grad_norm": 1.3695443868637085, "learning_rate": 1.0325840747807562e-05, "loss": 0.14363133907318115, "step": 19170 }, { "epoch": 0.505230438948578, "grad_norm": 1.7358707189559937, "learning_rate": 1.0317313787816915e-05, "loss": 0.1553389310836792, "step": 19180 }, { "epoch": 0.5054938541930768, "grad_norm": 1.8831380605697632, "learning_rate": 1.0308786596871606e-05, "loss": 0.1552414894104004, "step": 19190 }, { "epoch": 0.5057572694375755, "grad_norm": 1.276721715927124, "learning_rate": 1.0300259181178086e-05, "loss": 0.15004448890686034, "step": 19200 }, { "epoch": 0.5060206846820743, "grad_norm": 0.8425965905189514, "learning_rate": 1.0291731546942977e-05, "loss": 0.14197835922241211, "step": 19210 }, { "epoch": 0.506284099926573, "grad_norm": 1.612291693687439, "learning_rate": 1.0283203700373065e-05, "loss": 0.14290275573730468, "step": 19220 }, { "epoch": 0.5065475151710718, "grad_norm": 0.8125320672988892, "learning_rate": 1.0274675647675281e-05, "loss": 0.15694935321807862, "step": 19230 }, { "epoch": 0.5068109304155705, "grad_norm": 3.088165760040283, "learning_rate": 1.0266147395056712e-05, "loss": 0.1532548189163208, "step": 19240 }, { "epoch": 0.5070743456600693, "grad_norm": 2.7321746349334717, "learning_rate": 1.0257618948724583e-05, "loss": 0.14903726577758789, "step": 19250 }, { "epoch": 0.507337760904568, "grad_norm": 1.4009515047073364, "learning_rate": 1.024909031488627e-05, "loss": 0.15740323066711426, "step": 19260 }, { "epoch": 0.5076011761490666, "grad_norm": 2.0425827503204346, "learning_rate": 1.0240561499749278e-05, "loss": 0.1403442144393921, "step": 19270 }, { "epoch": 0.5078645913935654, "grad_norm": 0.7092601656913757, "learning_rate": 1.023203250952125e-05, "loss": 0.14584548473358155, "step": 19280 }, { "epoch": 0.5081280066380641, "grad_norm": 2.100027322769165, "learning_rate": 1.0223503350409949e-05, "loss": 0.14343366622924805, "step": 19290 }, { "epoch": 0.5083914218825629, "grad_norm": 0.8560479283332825, "learning_rate": 1.0214974028623268e-05, "loss": 0.15864832401275636, "step": 19300 }, { "epoch": 0.5086548371270616, "grad_norm": 1.5887647867202759, "learning_rate": 1.0206444550369212e-05, "loss": 0.16364021301269532, "step": 19310 }, { "epoch": 0.5089182523715604, "grad_norm": 1.1724826097488403, "learning_rate": 1.0197914921855909e-05, "loss": 0.14727271795272828, "step": 19320 }, { "epoch": 0.5091816676160591, "grad_norm": 2.1660549640655518, "learning_rate": 1.0189385149291586e-05, "loss": 0.1535084843635559, "step": 19330 }, { "epoch": 0.5094450828605579, "grad_norm": 1.715639591217041, "learning_rate": 1.018085523888458e-05, "loss": 0.14609036445617676, "step": 19340 }, { "epoch": 0.5097084981050566, "grad_norm": 3.3488383293151855, "learning_rate": 1.0172325196843328e-05, "loss": 0.14747707843780516, "step": 19350 }, { "epoch": 0.5099719133495553, "grad_norm": 1.2911272048950195, "learning_rate": 1.0163795029376362e-05, "loss": 0.14055546522140502, "step": 19360 }, { "epoch": 0.5102353285940541, "grad_norm": 2.3314247131347656, "learning_rate": 1.015526474269231e-05, "loss": 0.155441951751709, "step": 19370 }, { "epoch": 0.5104987438385528, "grad_norm": 2.511197566986084, "learning_rate": 1.0146734342999875e-05, "loss": 0.14266684055328369, "step": 19380 }, { "epoch": 0.5107621590830516, "grad_norm": 1.9773536920547485, "learning_rate": 1.0138203836507859e-05, "loss": 0.15748621225357057, "step": 19390 }, { "epoch": 0.5110255743275502, "grad_norm": 1.7699416875839233, "learning_rate": 1.0129673229425125e-05, "loss": 0.15399568080902098, "step": 19400 }, { "epoch": 0.511288989572049, "grad_norm": 2.345905303955078, "learning_rate": 1.0121142527960624e-05, "loss": 0.1375579357147217, "step": 19410 }, { "epoch": 0.5115524048165477, "grad_norm": 2.351602792739868, "learning_rate": 1.0112611738323362e-05, "loss": 0.13347805738449098, "step": 19420 }, { "epoch": 0.5118158200610465, "grad_norm": 2.335505723953247, "learning_rate": 1.0104080866722419e-05, "loss": 0.13947478532791138, "step": 19430 }, { "epoch": 0.5120792353055452, "grad_norm": 2.355656862258911, "learning_rate": 1.0095549919366928e-05, "loss": 0.13625020980834962, "step": 19440 }, { "epoch": 0.5123426505500439, "grad_norm": 2.462373733520508, "learning_rate": 1.0087018902466086e-05, "loss": 0.16152634620666503, "step": 19450 }, { "epoch": 0.5126060657945427, "grad_norm": 3.204869508743286, "learning_rate": 1.0078487822229132e-05, "loss": 0.15583112239837646, "step": 19460 }, { "epoch": 0.5128694810390414, "grad_norm": 1.034657597541809, "learning_rate": 1.0069956684865354e-05, "loss": 0.1558009386062622, "step": 19470 }, { "epoch": 0.5131328962835402, "grad_norm": 1.80286705493927, "learning_rate": 1.0061425496584084e-05, "loss": 0.1443479061126709, "step": 19480 }, { "epoch": 0.5133963115280389, "grad_norm": 3.3885152339935303, "learning_rate": 1.0052894263594687e-05, "loss": 0.13322566747665404, "step": 19490 }, { "epoch": 0.5136597267725377, "grad_norm": 1.2497249841690063, "learning_rate": 1.0044362992106563e-05, "loss": 0.14308847188949586, "step": 19500 }, { "epoch": 0.5139231420170364, "grad_norm": 1.243523120880127, "learning_rate": 1.0035831688329139e-05, "loss": 0.15359795093536377, "step": 19510 }, { "epoch": 0.5141865572615352, "grad_norm": 1.4485580921173096, "learning_rate": 1.0027300358471865e-05, "loss": 0.14852778911590575, "step": 19520 }, { "epoch": 0.5144499725060339, "grad_norm": 2.474640369415283, "learning_rate": 1.0018769008744214e-05, "loss": 0.15495946407318115, "step": 19530 }, { "epoch": 0.5147133877505325, "grad_norm": 1.2248371839523315, "learning_rate": 1.0010237645355669e-05, "loss": 0.14852662086486818, "step": 19540 }, { "epoch": 0.5149768029950313, "grad_norm": 1.6051722764968872, "learning_rate": 1.0001706274515724e-05, "loss": 0.15607304573059083, "step": 19550 }, { "epoch": 0.51524021823953, "grad_norm": 2.4197323322296143, "learning_rate": 9.993174902433877e-06, "loss": 0.14092416763305665, "step": 19560 }, { "epoch": 0.5155036334840288, "grad_norm": 0.9579778909683228, "learning_rate": 9.984643535319632e-06, "loss": 0.14858040809631348, "step": 19570 }, { "epoch": 0.5157670487285275, "grad_norm": 3.360670328140259, "learning_rate": 9.976112179382483e-06, "loss": 0.15862569808959961, "step": 19580 }, { "epoch": 0.5160304639730263, "grad_norm": 1.1658297777175903, "learning_rate": 9.967580840831922e-06, "loss": 0.15765079259872436, "step": 19590 }, { "epoch": 0.516293879217525, "grad_norm": 1.380677580833435, "learning_rate": 9.959049525877425e-06, "loss": 0.15280426740646363, "step": 19600 }, { "epoch": 0.5165572944620238, "grad_norm": 2.661062717437744, "learning_rate": 9.950518240728446e-06, "loss": 0.15941776037216188, "step": 19610 }, { "epoch": 0.5168207097065225, "grad_norm": 0.8821897506713867, "learning_rate": 9.94198699159443e-06, "loss": 0.14647622108459474, "step": 19620 }, { "epoch": 0.5170841249510212, "grad_norm": 1.4190675020217896, "learning_rate": 9.93345578468478e-06, "loss": 0.1535135269165039, "step": 19630 }, { "epoch": 0.51734754019552, "grad_norm": 2.118621349334717, "learning_rate": 9.924924626208884e-06, "loss": 0.1386951208114624, "step": 19640 }, { "epoch": 0.5176109554400187, "grad_norm": 0.9796146750450134, "learning_rate": 9.916393522376087e-06, "loss": 0.14829704761505128, "step": 19650 }, { "epoch": 0.5178743706845175, "grad_norm": 2.780444622039795, "learning_rate": 9.907862479395684e-06, "loss": 0.15220954418182372, "step": 19660 }, { "epoch": 0.5181377859290162, "grad_norm": 2.379035234451294, "learning_rate": 9.899331503476951e-06, "loss": 0.1538402795791626, "step": 19670 }, { "epoch": 0.518401201173515, "grad_norm": 1.5190812349319458, "learning_rate": 9.890800600829086e-06, "loss": 0.1388224482536316, "step": 19680 }, { "epoch": 0.5186646164180136, "grad_norm": 3.117465019226074, "learning_rate": 9.882269777661261e-06, "loss": 0.13884470462799073, "step": 19690 }, { "epoch": 0.5189280316625123, "grad_norm": 1.6039189100265503, "learning_rate": 9.873739040182569e-06, "loss": 0.14424961805343628, "step": 19700 }, { "epoch": 0.5191914469070111, "grad_norm": 1.247118353843689, "learning_rate": 9.86520839460205e-06, "loss": 0.16376758813858033, "step": 19710 }, { "epoch": 0.5194548621515098, "grad_norm": 3.6311373710632324, "learning_rate": 9.856677847128678e-06, "loss": 0.1381059169769287, "step": 19720 }, { "epoch": 0.5197182773960086, "grad_norm": 1.297121524810791, "learning_rate": 9.848147403971353e-06, "loss": 0.14613240957260132, "step": 19730 }, { "epoch": 0.5199816926405073, "grad_norm": 4.107046604156494, "learning_rate": 9.839617071338897e-06, "loss": 0.14943904876708985, "step": 19740 }, { "epoch": 0.5202451078850061, "grad_norm": 2.444190263748169, "learning_rate": 9.831086855440056e-06, "loss": 0.15718945264816284, "step": 19750 }, { "epoch": 0.5205085231295048, "grad_norm": 1.9561671018600464, "learning_rate": 9.82255676248349e-06, "loss": 0.13698807954788209, "step": 19760 }, { "epoch": 0.5207719383740036, "grad_norm": 2.473585367202759, "learning_rate": 9.814026798677767e-06, "loss": 0.14195082187652588, "step": 19770 }, { "epoch": 0.5210353536185023, "grad_norm": 0.8344389200210571, "learning_rate": 9.805496970231367e-06, "loss": 0.13964931964874266, "step": 19780 }, { "epoch": 0.521298768863001, "grad_norm": 1.5922402143478394, "learning_rate": 9.796967283352657e-06, "loss": 0.13810564279556276, "step": 19790 }, { "epoch": 0.5215621841074998, "grad_norm": 1.447973370552063, "learning_rate": 9.788437744249925e-06, "loss": 0.14386470317840577, "step": 19800 }, { "epoch": 0.5218255993519985, "grad_norm": 3.0764379501342773, "learning_rate": 9.779908359131327e-06, "loss": 0.1407092332839966, "step": 19810 }, { "epoch": 0.5220890145964973, "grad_norm": 1.4171935319900513, "learning_rate": 9.77137913420492e-06, "loss": 0.1419020414352417, "step": 19820 }, { "epoch": 0.522352429840996, "grad_norm": 1.6122891902923584, "learning_rate": 9.762850075678644e-06, "loss": 0.15948596000671386, "step": 19830 }, { "epoch": 0.5226158450854947, "grad_norm": 3.0366289615631104, "learning_rate": 9.754321189760312e-06, "loss": 0.139486563205719, "step": 19840 }, { "epoch": 0.5228792603299934, "grad_norm": 0.842744767665863, "learning_rate": 9.74579248265762e-06, "loss": 0.13982185125350952, "step": 19850 }, { "epoch": 0.5231426755744922, "grad_norm": 0.8723801374435425, "learning_rate": 9.737263960578127e-06, "loss": 0.14841666221618652, "step": 19860 }, { "epoch": 0.5234060908189909, "grad_norm": 1.3968539237976074, "learning_rate": 9.728735629729256e-06, "loss": 0.1507830500602722, "step": 19870 }, { "epoch": 0.5236695060634896, "grad_norm": 1.9818618297576904, "learning_rate": 9.7202074963183e-06, "loss": 0.1483245611190796, "step": 19880 }, { "epoch": 0.5239329213079884, "grad_norm": 1.590447187423706, "learning_rate": 9.711679566552396e-06, "loss": 0.14182058572769166, "step": 19890 }, { "epoch": 0.5241963365524871, "grad_norm": 1.8121058940887451, "learning_rate": 9.703151846638545e-06, "loss": 0.1481227993965149, "step": 19900 }, { "epoch": 0.5244597517969859, "grad_norm": 1.651023507118225, "learning_rate": 9.694624342783591e-06, "loss": 0.15161309242248536, "step": 19910 }, { "epoch": 0.5247231670414846, "grad_norm": 1.8157117366790771, "learning_rate": 9.686097061194209e-06, "loss": 0.16490273475646972, "step": 19920 }, { "epoch": 0.5249865822859834, "grad_norm": 2.1731362342834473, "learning_rate": 9.677570008076933e-06, "loss": 0.14911468029022218, "step": 19930 }, { "epoch": 0.5252499975304821, "grad_norm": 1.728745937347412, "learning_rate": 9.669043189638115e-06, "loss": 0.1477292776107788, "step": 19940 }, { "epoch": 0.5255134127749809, "grad_norm": 1.574718713760376, "learning_rate": 9.66051661208394e-06, "loss": 0.15902770757675172, "step": 19950 }, { "epoch": 0.5257768280194796, "grad_norm": 1.708188533782959, "learning_rate": 9.651990281620423e-06, "loss": 0.14575982093811035, "step": 19960 }, { "epoch": 0.5260402432639782, "grad_norm": 4.31252384185791, "learning_rate": 9.643464204453392e-06, "loss": 0.1464163541793823, "step": 19970 }, { "epoch": 0.526303658508477, "grad_norm": 3.074385404586792, "learning_rate": 9.634938386788497e-06, "loss": 0.16108343601226807, "step": 19980 }, { "epoch": 0.5265670737529757, "grad_norm": 2.0484418869018555, "learning_rate": 9.626412834831193e-06, "loss": 0.15793508291244507, "step": 19990 }, { "epoch": 0.5268304889974745, "grad_norm": 2.7779951095581055, "learning_rate": 9.617887554786746e-06, "loss": 0.13806238174438476, "step": 20000 }, { "epoch": 0.5270939042419732, "grad_norm": 2.2121970653533936, "learning_rate": 9.609362552860225e-06, "loss": 0.13657737970352174, "step": 20010 }, { "epoch": 0.527357319486472, "grad_norm": 3.0842013359069824, "learning_rate": 9.600837835256486e-06, "loss": 0.15003643035888672, "step": 20020 }, { "epoch": 0.5276207347309707, "grad_norm": 2.051593542098999, "learning_rate": 9.592313408180199e-06, "loss": 0.13344225883483887, "step": 20030 }, { "epoch": 0.5278841499754695, "grad_norm": 5.574258804321289, "learning_rate": 9.583789277835798e-06, "loss": 0.16205579042434692, "step": 20040 }, { "epoch": 0.5281475652199682, "grad_norm": 1.5833200216293335, "learning_rate": 9.575265450427517e-06, "loss": 0.15293368101119995, "step": 20050 }, { "epoch": 0.5284109804644669, "grad_norm": 1.2875908613204956, "learning_rate": 9.566741932159369e-06, "loss": 0.16040222644805907, "step": 20060 }, { "epoch": 0.5286743957089657, "grad_norm": 1.6021705865859985, "learning_rate": 9.558218729235132e-06, "loss": 0.15769338607788086, "step": 20070 }, { "epoch": 0.5289378109534644, "grad_norm": 3.7839298248291016, "learning_rate": 9.549695847858363e-06, "loss": 0.14805419445037843, "step": 20080 }, { "epoch": 0.5292012261979632, "grad_norm": 1.442306399345398, "learning_rate": 9.541173294232387e-06, "loss": 0.15459967851638795, "step": 20090 }, { "epoch": 0.5294646414424619, "grad_norm": 1.6324033737182617, "learning_rate": 9.53265107456028e-06, "loss": 0.1527867555618286, "step": 20100 }, { "epoch": 0.5297280566869607, "grad_norm": 0.9612331986427307, "learning_rate": 9.524129195044886e-06, "loss": 0.13519047498703002, "step": 20110 }, { "epoch": 0.5299914719314593, "grad_norm": 2.9374566078186035, "learning_rate": 9.515607661888799e-06, "loss": 0.1507028102874756, "step": 20120 }, { "epoch": 0.5302548871759581, "grad_norm": 2.5068650245666504, "learning_rate": 9.507086481294345e-06, "loss": 0.1665672779083252, "step": 20130 }, { "epoch": 0.5305183024204568, "grad_norm": 0.8812257647514343, "learning_rate": 9.498565659463624e-06, "loss": 0.14424571990966797, "step": 20140 }, { "epoch": 0.5307817176649555, "grad_norm": 0.8438196182250977, "learning_rate": 9.490045202598442e-06, "loss": 0.13818280696868895, "step": 20150 }, { "epoch": 0.5310451329094543, "grad_norm": 1.0332552194595337, "learning_rate": 9.48152511690037e-06, "loss": 0.14557839632034303, "step": 20160 }, { "epoch": 0.531308548153953, "grad_norm": 2.3029367923736572, "learning_rate": 9.473005408570685e-06, "loss": 0.14195234775543214, "step": 20170 }, { "epoch": 0.5315719633984518, "grad_norm": 1.6547695398330688, "learning_rate": 9.464486083810397e-06, "loss": 0.13649988174438477, "step": 20180 }, { "epoch": 0.5318353786429505, "grad_norm": 2.3410074710845947, "learning_rate": 9.455967148820243e-06, "loss": 0.14423476457595824, "step": 20190 }, { "epoch": 0.5320987938874493, "grad_norm": 3.593230724334717, "learning_rate": 9.44744860980067e-06, "loss": 0.15048544406890868, "step": 20200 }, { "epoch": 0.532362209131948, "grad_norm": 1.3884204626083374, "learning_rate": 9.438930472951838e-06, "loss": 0.1362478494644165, "step": 20210 }, { "epoch": 0.5326256243764467, "grad_norm": 1.6536999940872192, "learning_rate": 9.430412744473615e-06, "loss": 0.1413721561431885, "step": 20220 }, { "epoch": 0.5328890396209455, "grad_norm": 2.059605598449707, "learning_rate": 9.421895430565568e-06, "loss": 0.14804893732070923, "step": 20230 }, { "epoch": 0.5331524548654442, "grad_norm": 1.8926386833190918, "learning_rate": 9.413378537426972e-06, "loss": 0.14420204162597655, "step": 20240 }, { "epoch": 0.533415870109943, "grad_norm": 1.0535145998001099, "learning_rate": 9.404862071256785e-06, "loss": 0.1443215250968933, "step": 20250 }, { "epoch": 0.5336792853544416, "grad_norm": 2.0967800617218018, "learning_rate": 9.396346038253654e-06, "loss": 0.1617643117904663, "step": 20260 }, { "epoch": 0.5339427005989404, "grad_norm": 4.282349586486816, "learning_rate": 9.387830444615926e-06, "loss": 0.13865216970443725, "step": 20270 }, { "epoch": 0.5342061158434391, "grad_norm": 4.782562732696533, "learning_rate": 9.379315296541611e-06, "loss": 0.13940919637680055, "step": 20280 }, { "epoch": 0.5344695310879379, "grad_norm": 2.016904830932617, "learning_rate": 9.370800600228399e-06, "loss": 0.15352187156677247, "step": 20290 }, { "epoch": 0.5347329463324366, "grad_norm": 1.0069741010665894, "learning_rate": 9.362286361873658e-06, "loss": 0.1500488042831421, "step": 20300 }, { "epoch": 0.5349963615769353, "grad_norm": 2.001856565475464, "learning_rate": 9.353772587674413e-06, "loss": 0.14746170043945311, "step": 20310 }, { "epoch": 0.5352597768214341, "grad_norm": 1.786384105682373, "learning_rate": 9.345259283827362e-06, "loss": 0.1414815068244934, "step": 20320 }, { "epoch": 0.5355231920659328, "grad_norm": 2.0832901000976562, "learning_rate": 9.336746456528853e-06, "loss": 0.14520881175994874, "step": 20330 }, { "epoch": 0.5357866073104316, "grad_norm": 0.9742457866668701, "learning_rate": 9.328234111974885e-06, "loss": 0.1382880687713623, "step": 20340 }, { "epoch": 0.5360500225549303, "grad_norm": 1.1480334997177124, "learning_rate": 9.319722256361117e-06, "loss": 0.16234784126281737, "step": 20350 }, { "epoch": 0.5363134377994291, "grad_norm": 4.288468837738037, "learning_rate": 9.311210895882835e-06, "loss": 0.1436067819595337, "step": 20360 }, { "epoch": 0.5365768530439278, "grad_norm": 1.9411201477050781, "learning_rate": 9.302700036734985e-06, "loss": 0.14275434017181396, "step": 20370 }, { "epoch": 0.5368402682884266, "grad_norm": 1.3491886854171753, "learning_rate": 9.294189685112133e-06, "loss": 0.14588290452957153, "step": 20380 }, { "epoch": 0.5371036835329253, "grad_norm": 1.731461763381958, "learning_rate": 9.285679847208474e-06, "loss": 0.13174926042556762, "step": 20390 }, { "epoch": 0.537367098777424, "grad_norm": 1.61678946018219, "learning_rate": 9.277170529217847e-06, "loss": 0.15051127672195436, "step": 20400 }, { "epoch": 0.5376305140219227, "grad_norm": 0.8735066652297974, "learning_rate": 9.268661737333691e-06, "loss": 0.14237666130065918, "step": 20410 }, { "epoch": 0.5378939292664214, "grad_norm": 1.7820262908935547, "learning_rate": 9.260153477749072e-06, "loss": 0.1460167407989502, "step": 20420 }, { "epoch": 0.5381573445109202, "grad_norm": 2.596994161605835, "learning_rate": 9.251645756656676e-06, "loss": 0.15734362602233887, "step": 20430 }, { "epoch": 0.5384207597554189, "grad_norm": 1.941490888595581, "learning_rate": 9.243138580248782e-06, "loss": 0.14289690256118776, "step": 20440 }, { "epoch": 0.5386841749999177, "grad_norm": 4.553114891052246, "learning_rate": 9.234631954717285e-06, "loss": 0.15179227590560912, "step": 20450 }, { "epoch": 0.5389475902444164, "grad_norm": 2.4885194301605225, "learning_rate": 9.226125886253673e-06, "loss": 0.15191010236740113, "step": 20460 }, { "epoch": 0.5392110054889152, "grad_norm": 0.7556988596916199, "learning_rate": 9.217620381049026e-06, "loss": 0.14736534357070924, "step": 20470 }, { "epoch": 0.5394744207334139, "grad_norm": 1.0965614318847656, "learning_rate": 9.209115445294022e-06, "loss": 0.14010233879089357, "step": 20480 }, { "epoch": 0.5397378359779126, "grad_norm": 0.852715253829956, "learning_rate": 9.200611085178917e-06, "loss": 0.1398270845413208, "step": 20490 }, { "epoch": 0.5400012512224114, "grad_norm": 1.5547642707824707, "learning_rate": 9.192107306893556e-06, "loss": 0.1483771562576294, "step": 20500 }, { "epoch": 0.5402646664669101, "grad_norm": 3.89907169342041, "learning_rate": 9.183604116627354e-06, "loss": 0.15803873538970947, "step": 20510 }, { "epoch": 0.5405280817114089, "grad_norm": 1.7138044834136963, "learning_rate": 9.175101520569293e-06, "loss": 0.15344412326812745, "step": 20520 }, { "epoch": 0.5407914969559076, "grad_norm": 3.1048967838287354, "learning_rate": 9.166599524907937e-06, "loss": 0.1466521978378296, "step": 20530 }, { "epoch": 0.5410549122004064, "grad_norm": 3.2263617515563965, "learning_rate": 9.158098135831404e-06, "loss": 0.14947271347045898, "step": 20540 }, { "epoch": 0.541318327444905, "grad_norm": 2.382995128631592, "learning_rate": 9.149597359527368e-06, "loss": 0.15096124410629272, "step": 20550 }, { "epoch": 0.5415817426894038, "grad_norm": 1.3728573322296143, "learning_rate": 9.141097202183063e-06, "loss": 0.15010135173797606, "step": 20560 }, { "epoch": 0.5418451579339025, "grad_norm": 1.4263091087341309, "learning_rate": 9.13259766998527e-06, "loss": 0.14216145277023315, "step": 20570 }, { "epoch": 0.5421085731784012, "grad_norm": 2.1668832302093506, "learning_rate": 9.124098769120318e-06, "loss": 0.14291670322418212, "step": 20580 }, { "epoch": 0.5423719884229, "grad_norm": 0.83204185962677, "learning_rate": 9.115600505774072e-06, "loss": 0.1531731367111206, "step": 20590 }, { "epoch": 0.5426354036673987, "grad_norm": 1.29835844039917, "learning_rate": 9.10710288613193e-06, "loss": 0.1510285973548889, "step": 20600 }, { "epoch": 0.5428988189118975, "grad_norm": 1.0487946271896362, "learning_rate": 9.098605916378837e-06, "loss": 0.14494857788085938, "step": 20610 }, { "epoch": 0.5431622341563962, "grad_norm": 2.2586169242858887, "learning_rate": 9.090109602699239e-06, "loss": 0.1410732626914978, "step": 20620 }, { "epoch": 0.543425649400895, "grad_norm": 2.5841708183288574, "learning_rate": 9.081613951277135e-06, "loss": 0.14325637817382814, "step": 20630 }, { "epoch": 0.5436890646453937, "grad_norm": 1.6573786735534668, "learning_rate": 9.07311896829602e-06, "loss": 0.144321072101593, "step": 20640 }, { "epoch": 0.5439524798898925, "grad_norm": 1.6532015800476074, "learning_rate": 9.064624659938902e-06, "loss": 0.13687453269958497, "step": 20650 }, { "epoch": 0.5442158951343912, "grad_norm": 1.8089046478271484, "learning_rate": 9.056131032388318e-06, "loss": 0.14754185676574708, "step": 20660 }, { "epoch": 0.5444793103788899, "grad_norm": 3.4435834884643555, "learning_rate": 9.047638091826291e-06, "loss": 0.14827349185943603, "step": 20670 }, { "epoch": 0.5447427256233887, "grad_norm": 0.7993977069854736, "learning_rate": 9.039145844434344e-06, "loss": 0.15196317434310913, "step": 20680 }, { "epoch": 0.5450061408678873, "grad_norm": 2.633631467819214, "learning_rate": 9.030654296393512e-06, "loss": 0.14118051528930664, "step": 20690 }, { "epoch": 0.5452695561123861, "grad_norm": 1.459303617477417, "learning_rate": 9.022163453884299e-06, "loss": 0.13886044025421143, "step": 20700 }, { "epoch": 0.5455329713568848, "grad_norm": 0.9729973077774048, "learning_rate": 9.013673323086716e-06, "loss": 0.1470130443572998, "step": 20710 }, { "epoch": 0.5457963866013836, "grad_norm": 2.3470396995544434, "learning_rate": 9.005183910180243e-06, "loss": 0.14343658685684205, "step": 20720 }, { "epoch": 0.5460598018458823, "grad_norm": 1.8304154872894287, "learning_rate": 8.996695221343834e-06, "loss": 0.14400068521499634, "step": 20730 }, { "epoch": 0.546323217090381, "grad_norm": 1.6546202898025513, "learning_rate": 8.988207262755936e-06, "loss": 0.1470433473587036, "step": 20740 }, { "epoch": 0.5465866323348798, "grad_norm": 1.5395190715789795, "learning_rate": 8.979720040594438e-06, "loss": 0.16131197214126586, "step": 20750 }, { "epoch": 0.5468500475793785, "grad_norm": 3.239316701889038, "learning_rate": 8.971233561036718e-06, "loss": 0.13821531534194947, "step": 20760 }, { "epoch": 0.5471134628238773, "grad_norm": 1.276380181312561, "learning_rate": 8.962747830259598e-06, "loss": 0.144502854347229, "step": 20770 }, { "epoch": 0.547376878068376, "grad_norm": 1.5047119855880737, "learning_rate": 8.954262854439354e-06, "loss": 0.1532581090927124, "step": 20780 }, { "epoch": 0.5476402933128748, "grad_norm": 1.5382119417190552, "learning_rate": 8.945778639751723e-06, "loss": 0.14867240190505981, "step": 20790 }, { "epoch": 0.5479037085573735, "grad_norm": 3.802170991897583, "learning_rate": 8.937295192371882e-06, "loss": 0.13765184879302977, "step": 20800 }, { "epoch": 0.5481671238018723, "grad_norm": 1.020039677619934, "learning_rate": 8.928812518474445e-06, "loss": 0.12552324533462525, "step": 20810 }, { "epoch": 0.548430539046371, "grad_norm": 1.7955783605575562, "learning_rate": 8.920330624233477e-06, "loss": 0.12870399951934813, "step": 20820 }, { "epoch": 0.5486939542908696, "grad_norm": 1.3040871620178223, "learning_rate": 8.911849515822457e-06, "loss": 0.14758349657058717, "step": 20830 }, { "epoch": 0.5489573695353684, "grad_norm": 1.2498328685760498, "learning_rate": 8.903369199414311e-06, "loss": 0.15090444087982177, "step": 20840 }, { "epoch": 0.5492207847798671, "grad_norm": 1.2460306882858276, "learning_rate": 8.894889681181374e-06, "loss": 0.14661232233047486, "step": 20850 }, { "epoch": 0.5494842000243659, "grad_norm": 3.0775554180145264, "learning_rate": 8.886410967295401e-06, "loss": 0.1494502305984497, "step": 20860 }, { "epoch": 0.5497476152688646, "grad_norm": 1.7206737995147705, "learning_rate": 8.877933063927576e-06, "loss": 0.1303120255470276, "step": 20870 }, { "epoch": 0.5500110305133634, "grad_norm": 1.6315948963165283, "learning_rate": 8.869455977248473e-06, "loss": 0.1440571904182434, "step": 20880 }, { "epoch": 0.5502744457578621, "grad_norm": 2.493156671524048, "learning_rate": 8.860979713428085e-06, "loss": 0.13010702133178711, "step": 20890 }, { "epoch": 0.5505378610023609, "grad_norm": 2.2492499351501465, "learning_rate": 8.852504278635806e-06, "loss": 0.14493780136108397, "step": 20900 }, { "epoch": 0.5508012762468596, "grad_norm": 2.094176769256592, "learning_rate": 8.844029679040418e-06, "loss": 0.14410717487335206, "step": 20910 }, { "epoch": 0.5510646914913583, "grad_norm": 2.008723735809326, "learning_rate": 8.835555920810104e-06, "loss": 0.13646814823150635, "step": 20920 }, { "epoch": 0.5513281067358571, "grad_norm": 1.1772255897521973, "learning_rate": 8.827083010112429e-06, "loss": 0.12543367147445678, "step": 20930 }, { "epoch": 0.5515915219803558, "grad_norm": 1.1870588064193726, "learning_rate": 8.818610953114342e-06, "loss": 0.13578561544418336, "step": 20940 }, { "epoch": 0.5518549372248546, "grad_norm": 2.101116895675659, "learning_rate": 8.810139755982173e-06, "loss": 0.13448797464370726, "step": 20950 }, { "epoch": 0.5521183524693533, "grad_norm": 2.0837466716766357, "learning_rate": 8.801669424881624e-06, "loss": 0.150975239276886, "step": 20960 }, { "epoch": 0.5523817677138521, "grad_norm": 1.0012232065200806, "learning_rate": 8.79319996597777e-06, "loss": 0.14093093872070311, "step": 20970 }, { "epoch": 0.5526451829583507, "grad_norm": 2.4256057739257812, "learning_rate": 8.784731385435049e-06, "loss": 0.1394744634628296, "step": 20980 }, { "epoch": 0.5529085982028495, "grad_norm": 1.2908728122711182, "learning_rate": 8.77626368941725e-06, "loss": 0.12495323419570922, "step": 20990 }, { "epoch": 0.5531720134473482, "grad_norm": 1.8849323987960815, "learning_rate": 8.767796884087539e-06, "loss": 0.145155131816864, "step": 21000 }, { "epoch": 0.5534354286918469, "grad_norm": 3.1568658351898193, "learning_rate": 8.759330975608418e-06, "loss": 0.14910073280334474, "step": 21010 }, { "epoch": 0.5536988439363457, "grad_norm": 2.025113821029663, "learning_rate": 8.750865970141736e-06, "loss": 0.14222159385681152, "step": 21020 }, { "epoch": 0.5539622591808444, "grad_norm": 1.3859155178070068, "learning_rate": 8.742401873848696e-06, "loss": 0.1402755618095398, "step": 21030 }, { "epoch": 0.5542256744253432, "grad_norm": 2.350101947784424, "learning_rate": 8.733938692889827e-06, "loss": 0.14422743320465087, "step": 21040 }, { "epoch": 0.5544890896698419, "grad_norm": 1.3044174909591675, "learning_rate": 8.725476433425003e-06, "loss": 0.14186592102050782, "step": 21050 }, { "epoch": 0.5547525049143407, "grad_norm": 2.741549491882324, "learning_rate": 8.717015101613417e-06, "loss": 0.14771076440811157, "step": 21060 }, { "epoch": 0.5550159201588394, "grad_norm": 1.0144208669662476, "learning_rate": 8.708554703613592e-06, "loss": 0.13903751373291015, "step": 21070 }, { "epoch": 0.5552793354033382, "grad_norm": 1.175955057144165, "learning_rate": 8.700095245583374e-06, "loss": 0.1261897087097168, "step": 21080 }, { "epoch": 0.5555427506478369, "grad_norm": 1.6117303371429443, "learning_rate": 8.691636733679915e-06, "loss": 0.14848096370697023, "step": 21090 }, { "epoch": 0.5558061658923356, "grad_norm": 1.2325843572616577, "learning_rate": 8.683179174059691e-06, "loss": 0.14985344409942628, "step": 21100 }, { "epoch": 0.5560695811368344, "grad_norm": 1.2673490047454834, "learning_rate": 8.67472257287848e-06, "loss": 0.1377278208732605, "step": 21110 }, { "epoch": 0.556332996381333, "grad_norm": 2.1423354148864746, "learning_rate": 8.666266936291354e-06, "loss": 0.14707562923431397, "step": 21120 }, { "epoch": 0.5565964116258318, "grad_norm": 1.8813751935958862, "learning_rate": 8.657812270452695e-06, "loss": 0.1502697467803955, "step": 21130 }, { "epoch": 0.5568598268703305, "grad_norm": 2.4911046028137207, "learning_rate": 8.649358581516171e-06, "loss": 0.15161219835281373, "step": 21140 }, { "epoch": 0.5571232421148293, "grad_norm": 1.310139775276184, "learning_rate": 8.640905875634744e-06, "loss": 0.14432966709136963, "step": 21150 }, { "epoch": 0.557386657359328, "grad_norm": 2.1932532787323, "learning_rate": 8.632454158960656e-06, "loss": 0.14147670269012452, "step": 21160 }, { "epoch": 0.5576500726038268, "grad_norm": 2.0616180896759033, "learning_rate": 8.624003437645431e-06, "loss": 0.12359380722045898, "step": 21170 }, { "epoch": 0.5579134878483255, "grad_norm": 3.60184907913208, "learning_rate": 8.615553717839873e-06, "loss": 0.12976629734039308, "step": 21180 }, { "epoch": 0.5581769030928242, "grad_norm": 1.872583031654358, "learning_rate": 8.607105005694048e-06, "loss": 0.15238466262817382, "step": 21190 }, { "epoch": 0.558440318337323, "grad_norm": 1.5576529502868652, "learning_rate": 8.598657307357293e-06, "loss": 0.14107991456985475, "step": 21200 }, { "epoch": 0.5587037335818217, "grad_norm": 1.2961856126785278, "learning_rate": 8.590210628978213e-06, "loss": 0.13829371929168702, "step": 21210 }, { "epoch": 0.5589671488263205, "grad_norm": 2.2723796367645264, "learning_rate": 8.581764976704653e-06, "loss": 0.14567232131958008, "step": 21220 }, { "epoch": 0.5592305640708192, "grad_norm": 1.4037529230117798, "learning_rate": 8.573320356683739e-06, "loss": 0.14631651639938353, "step": 21230 }, { "epoch": 0.559493979315318, "grad_norm": 1.5114437341690063, "learning_rate": 8.564876775061819e-06, "loss": 0.13065664768218993, "step": 21240 }, { "epoch": 0.5597573945598167, "grad_norm": 2.0957131385803223, "learning_rate": 8.556434237984496e-06, "loss": 0.12676538228988649, "step": 21250 }, { "epoch": 0.5600208098043153, "grad_norm": 3.0347232818603516, "learning_rate": 8.547992751596617e-06, "loss": 0.1532711386680603, "step": 21260 }, { "epoch": 0.5602842250488141, "grad_norm": 1.8465825319290161, "learning_rate": 8.53955232204226e-06, "loss": 0.1465422034263611, "step": 21270 }, { "epoch": 0.5605476402933128, "grad_norm": 1.71163010597229, "learning_rate": 8.531112955464728e-06, "loss": 0.13994369506835938, "step": 21280 }, { "epoch": 0.5608110555378116, "grad_norm": 1.0622494220733643, "learning_rate": 8.522674658006564e-06, "loss": 0.13588943481445312, "step": 21290 }, { "epoch": 0.5610744707823103, "grad_norm": 1.7054866552352905, "learning_rate": 8.514237435809517e-06, "loss": 0.13373805284500123, "step": 21300 }, { "epoch": 0.5613378860268091, "grad_norm": 1.1773606538772583, "learning_rate": 8.505801295014569e-06, "loss": 0.13322513103485106, "step": 21310 }, { "epoch": 0.5616013012713078, "grad_norm": 2.87135910987854, "learning_rate": 8.497366241761907e-06, "loss": 0.14530980587005615, "step": 21320 }, { "epoch": 0.5618647165158066, "grad_norm": 1.0300897359848022, "learning_rate": 8.488932282190914e-06, "loss": 0.14291737079620362, "step": 21330 }, { "epoch": 0.5621281317603053, "grad_norm": 0.8155497908592224, "learning_rate": 8.480499422440207e-06, "loss": 0.13424286842346192, "step": 21340 }, { "epoch": 0.562391547004804, "grad_norm": 2.219208002090454, "learning_rate": 8.472067668647571e-06, "loss": 0.14322822093963622, "step": 21350 }, { "epoch": 0.5626549622493028, "grad_norm": 1.6711620092391968, "learning_rate": 8.463637026950013e-06, "loss": 0.14273319244384766, "step": 21360 }, { "epoch": 0.5629183774938015, "grad_norm": 1.4090702533721924, "learning_rate": 8.455207503483708e-06, "loss": 0.1370985507965088, "step": 21370 }, { "epoch": 0.5631817927383003, "grad_norm": 4.723635196685791, "learning_rate": 8.446779104384028e-06, "loss": 0.14657609462738036, "step": 21380 }, { "epoch": 0.563445207982799, "grad_norm": 2.209191083908081, "learning_rate": 8.43835183578553e-06, "loss": 0.13343958854675292, "step": 21390 }, { "epoch": 0.5637086232272978, "grad_norm": 0.941810667514801, "learning_rate": 8.429925703821941e-06, "loss": 0.1558711290359497, "step": 21400 }, { "epoch": 0.5639720384717964, "grad_norm": 2.583458662033081, "learning_rate": 8.421500714626164e-06, "loss": 0.15044102668762208, "step": 21410 }, { "epoch": 0.5642354537162952, "grad_norm": 1.4460488557815552, "learning_rate": 8.413076874330266e-06, "loss": 0.14620141983032225, "step": 21420 }, { "epoch": 0.5644988689607939, "grad_norm": 1.143908977508545, "learning_rate": 8.404654189065486e-06, "loss": 0.1382364511489868, "step": 21430 }, { "epoch": 0.5647622842052926, "grad_norm": 1.2126270532608032, "learning_rate": 8.396232664962218e-06, "loss": 0.15484755039215087, "step": 21440 }, { "epoch": 0.5650256994497914, "grad_norm": 1.0481163263320923, "learning_rate": 8.387812308150011e-06, "loss": 0.12720870971679688, "step": 21450 }, { "epoch": 0.5652891146942901, "grad_norm": 1.241355061531067, "learning_rate": 8.379393124757554e-06, "loss": 0.16106779575347902, "step": 21460 }, { "epoch": 0.5655525299387889, "grad_norm": 1.3420491218566895, "learning_rate": 8.370975120912708e-06, "loss": 0.14853293895721437, "step": 21470 }, { "epoch": 0.5658159451832876, "grad_norm": 3.8669097423553467, "learning_rate": 8.362558302742447e-06, "loss": 0.1342007040977478, "step": 21480 }, { "epoch": 0.5660793604277864, "grad_norm": 3.6232404708862305, "learning_rate": 8.354142676372898e-06, "loss": 0.14056577682495117, "step": 21490 }, { "epoch": 0.5663427756722851, "grad_norm": 1.0239332914352417, "learning_rate": 8.345728247929319e-06, "loss": 0.1302180051803589, "step": 21500 }, { "epoch": 0.5666061909167839, "grad_norm": 1.6669012308120728, "learning_rate": 8.33731502353609e-06, "loss": 0.1382071256637573, "step": 21510 }, { "epoch": 0.5668696061612826, "grad_norm": 2.174027442932129, "learning_rate": 8.328903009316724e-06, "loss": 0.15902810096740722, "step": 21520 }, { "epoch": 0.5671330214057813, "grad_norm": 1.419447422027588, "learning_rate": 8.320492211393844e-06, "loss": 0.1419313669204712, "step": 21530 }, { "epoch": 0.5673964366502801, "grad_norm": 3.366384506225586, "learning_rate": 8.312082635889193e-06, "loss": 0.1495208978652954, "step": 21540 }, { "epoch": 0.5676598518947787, "grad_norm": 1.3991185426712036, "learning_rate": 8.303674288923621e-06, "loss": 0.14812039136886596, "step": 21550 }, { "epoch": 0.5679232671392775, "grad_norm": 2.0307765007019043, "learning_rate": 8.29526717661709e-06, "loss": 0.14384055137634277, "step": 21560 }, { "epoch": 0.5681866823837762, "grad_norm": 1.464534044265747, "learning_rate": 8.286861305088655e-06, "loss": 0.13400115966796874, "step": 21570 }, { "epoch": 0.568450097628275, "grad_norm": 1.1892354488372803, "learning_rate": 8.278456680456476e-06, "loss": 0.13692812919616698, "step": 21580 }, { "epoch": 0.5687135128727737, "grad_norm": 1.5851631164550781, "learning_rate": 8.270053308837791e-06, "loss": 0.13368771076202393, "step": 21590 }, { "epoch": 0.5689769281172725, "grad_norm": 1.9877800941467285, "learning_rate": 8.261651196348953e-06, "loss": 0.13937277793884278, "step": 21600 }, { "epoch": 0.5692403433617712, "grad_norm": 2.655430316925049, "learning_rate": 8.253250349105369e-06, "loss": 0.1396688461303711, "step": 21610 }, { "epoch": 0.5695037586062699, "grad_norm": 2.0296547412872314, "learning_rate": 8.24485077322154e-06, "loss": 0.14097836017608642, "step": 21620 }, { "epoch": 0.5697671738507687, "grad_norm": 1.8066022396087646, "learning_rate": 8.236452474811046e-06, "loss": 0.1415086269378662, "step": 21630 }, { "epoch": 0.5700305890952674, "grad_norm": 1.6056108474731445, "learning_rate": 8.228055459986521e-06, "loss": 0.14477320909500122, "step": 21640 }, { "epoch": 0.5702940043397662, "grad_norm": 2.341719150543213, "learning_rate": 8.219659734859686e-06, "loss": 0.13230206966400146, "step": 21650 }, { "epoch": 0.5705574195842649, "grad_norm": 1.8944026231765747, "learning_rate": 8.211265305541304e-06, "loss": 0.1418304920196533, "step": 21660 }, { "epoch": 0.5708208348287637, "grad_norm": 2.347529649734497, "learning_rate": 8.202872178141204e-06, "loss": 0.14604220390319825, "step": 21670 }, { "epoch": 0.5710842500732624, "grad_norm": 5.682767391204834, "learning_rate": 8.19448035876827e-06, "loss": 0.13072391748428344, "step": 21680 }, { "epoch": 0.5713476653177612, "grad_norm": 1.3736798763275146, "learning_rate": 8.186089853530426e-06, "loss": 0.1421207070350647, "step": 21690 }, { "epoch": 0.5716110805622598, "grad_norm": 4.923738479614258, "learning_rate": 8.177700668534646e-06, "loss": 0.1521954894065857, "step": 21700 }, { "epoch": 0.5718744958067585, "grad_norm": 1.5560204982757568, "learning_rate": 8.169312809886944e-06, "loss": 0.13542914390563965, "step": 21710 }, { "epoch": 0.5721379110512573, "grad_norm": 2.306309461593628, "learning_rate": 8.160926283692355e-06, "loss": 0.14358242750167846, "step": 21720 }, { "epoch": 0.572401326295756, "grad_norm": 2.7621967792510986, "learning_rate": 8.152541096054965e-06, "loss": 0.13436051607131957, "step": 21730 }, { "epoch": 0.5726647415402548, "grad_norm": 1.9179672002792358, "learning_rate": 8.144157253077871e-06, "loss": 0.14611129760742186, "step": 21740 }, { "epoch": 0.5729281567847535, "grad_norm": 1.443507432937622, "learning_rate": 8.135774760863193e-06, "loss": 0.14620288610458373, "step": 21750 }, { "epoch": 0.5731915720292523, "grad_norm": 1.078504204750061, "learning_rate": 8.127393625512077e-06, "loss": 0.1385321855545044, "step": 21760 }, { "epoch": 0.573454987273751, "grad_norm": 1.7375528812408447, "learning_rate": 8.11901385312467e-06, "loss": 0.14332112073898315, "step": 21770 }, { "epoch": 0.5737184025182498, "grad_norm": 1.4810868501663208, "learning_rate": 8.110635449800132e-06, "loss": 0.14761834144592284, "step": 21780 }, { "epoch": 0.5739818177627485, "grad_norm": 1.6788854598999023, "learning_rate": 8.10225842163663e-06, "loss": 0.13913891315460206, "step": 21790 }, { "epoch": 0.5742452330072472, "grad_norm": 1.5407063961029053, "learning_rate": 8.093882774731322e-06, "loss": 0.151542329788208, "step": 21800 }, { "epoch": 0.574508648251746, "grad_norm": 3.3598172664642334, "learning_rate": 8.085508515180373e-06, "loss": 0.13980729579925538, "step": 21810 }, { "epoch": 0.5747720634962447, "grad_norm": 1.9632338285446167, "learning_rate": 8.077135649078917e-06, "loss": 0.13263394832611083, "step": 21820 }, { "epoch": 0.5750354787407435, "grad_norm": 2.1725924015045166, "learning_rate": 8.068764182521102e-06, "loss": 0.14601659774780273, "step": 21830 }, { "epoch": 0.5752988939852421, "grad_norm": 1.4798129796981812, "learning_rate": 8.060394121600033e-06, "loss": 0.13821232318878174, "step": 21840 }, { "epoch": 0.5755623092297409, "grad_norm": 1.2875155210494995, "learning_rate": 8.052025472407802e-06, "loss": 0.13840768337249756, "step": 21850 }, { "epoch": 0.5758257244742396, "grad_norm": 1.6894282102584839, "learning_rate": 8.043658241035477e-06, "loss": 0.15289692878723143, "step": 21860 }, { "epoch": 0.5760891397187383, "grad_norm": 0.9349255561828613, "learning_rate": 8.035292433573088e-06, "loss": 0.1324984312057495, "step": 21870 }, { "epoch": 0.5763525549632371, "grad_norm": 2.359450578689575, "learning_rate": 8.026928056109628e-06, "loss": 0.13507336378097534, "step": 21880 }, { "epoch": 0.5766159702077358, "grad_norm": 2.948235511779785, "learning_rate": 8.018565114733056e-06, "loss": 0.1527742028236389, "step": 21890 }, { "epoch": 0.5768793854522346, "grad_norm": 1.0627684593200684, "learning_rate": 8.010203615530279e-06, "loss": 0.1404071569442749, "step": 21900 }, { "epoch": 0.5771428006967333, "grad_norm": 2.3313255310058594, "learning_rate": 8.00184356458716e-06, "loss": 0.1396845579147339, "step": 21910 }, { "epoch": 0.5774062159412321, "grad_norm": 1.6960331201553345, "learning_rate": 7.993484967988504e-06, "loss": 0.14025511741638183, "step": 21920 }, { "epoch": 0.5776696311857308, "grad_norm": 1.568557620048523, "learning_rate": 7.98512783181805e-06, "loss": 0.1297624707221985, "step": 21930 }, { "epoch": 0.5779330464302296, "grad_norm": 1.5068798065185547, "learning_rate": 7.976772162158498e-06, "loss": 0.15565838813781738, "step": 21940 }, { "epoch": 0.5781964616747283, "grad_norm": 1.5136886835098267, "learning_rate": 7.968417965091449e-06, "loss": 0.12988755702972413, "step": 21950 }, { "epoch": 0.578459876919227, "grad_norm": 3.153916835784912, "learning_rate": 7.960065246697456e-06, "loss": 0.14077308177947997, "step": 21960 }, { "epoch": 0.5787232921637258, "grad_norm": 1.2417161464691162, "learning_rate": 7.951714013055988e-06, "loss": 0.14312522411346434, "step": 21970 }, { "epoch": 0.5789867074082244, "grad_norm": 2.0061445236206055, "learning_rate": 7.94336427024543e-06, "loss": 0.1466947078704834, "step": 21980 }, { "epoch": 0.5792501226527232, "grad_norm": 1.9077224731445312, "learning_rate": 7.93501602434309e-06, "loss": 0.1396080732345581, "step": 21990 }, { "epoch": 0.5795135378972219, "grad_norm": 1.361834168434143, "learning_rate": 7.926669281425177e-06, "loss": 0.1391435146331787, "step": 22000 }, { "epoch": 0.5797769531417207, "grad_norm": 1.5236406326293945, "learning_rate": 7.91832404756681e-06, "loss": 0.15454932451248168, "step": 22010 }, { "epoch": 0.5800403683862194, "grad_norm": 2.698582649230957, "learning_rate": 7.909980328842015e-06, "loss": 0.14285314083099365, "step": 22020 }, { "epoch": 0.5803037836307182, "grad_norm": 1.3058586120605469, "learning_rate": 7.901638131323705e-06, "loss": 0.13784148693084716, "step": 22030 }, { "epoch": 0.5805671988752169, "grad_norm": 1.7351971864700317, "learning_rate": 7.893297461083697e-06, "loss": 0.14177780151367186, "step": 22040 }, { "epoch": 0.5808306141197156, "grad_norm": 1.5919466018676758, "learning_rate": 7.884958324192689e-06, "loss": 0.14988539218902588, "step": 22050 }, { "epoch": 0.5810940293642144, "grad_norm": 2.2328805923461914, "learning_rate": 7.876620726720257e-06, "loss": 0.1564191460609436, "step": 22060 }, { "epoch": 0.5813574446087131, "grad_norm": 1.2887156009674072, "learning_rate": 7.868284674734877e-06, "loss": 0.14550423622131348, "step": 22070 }, { "epoch": 0.5816208598532119, "grad_norm": 1.0460065603256226, "learning_rate": 7.859950174303879e-06, "loss": 0.14624447822570802, "step": 22080 }, { "epoch": 0.5818842750977106, "grad_norm": 1.7024633884429932, "learning_rate": 7.851617231493472e-06, "loss": 0.14328664541244507, "step": 22090 }, { "epoch": 0.5821476903422094, "grad_norm": 3.0462679862976074, "learning_rate": 7.843285852368734e-06, "loss": 0.1417796492576599, "step": 22100 }, { "epoch": 0.5824111055867081, "grad_norm": 1.016140341758728, "learning_rate": 7.8349560429936e-06, "loss": 0.13689427375793456, "step": 22110 }, { "epoch": 0.5826745208312069, "grad_norm": 2.479987621307373, "learning_rate": 7.826627809430867e-06, "loss": 0.14591479301452637, "step": 22120 }, { "epoch": 0.5829379360757055, "grad_norm": 1.5599220991134644, "learning_rate": 7.818301157742182e-06, "loss": 0.14115071296691895, "step": 22130 }, { "epoch": 0.5832013513202042, "grad_norm": 2.38405442237854, "learning_rate": 7.809976093988039e-06, "loss": 0.1453394412994385, "step": 22140 }, { "epoch": 0.583464766564703, "grad_norm": 2.0131008625030518, "learning_rate": 7.801652624227782e-06, "loss": 0.14816884994506835, "step": 22150 }, { "epoch": 0.5837281818092017, "grad_norm": 1.9740056991577148, "learning_rate": 7.79333075451959e-06, "loss": 0.13593578338623047, "step": 22160 }, { "epoch": 0.5839915970537005, "grad_norm": 2.3036744594573975, "learning_rate": 7.785010490920477e-06, "loss": 0.14054646492004394, "step": 22170 }, { "epoch": 0.5842550122981992, "grad_norm": 3.332252025604248, "learning_rate": 7.776691839486297e-06, "loss": 0.13426132202148439, "step": 22180 }, { "epoch": 0.584518427542698, "grad_norm": 1.6995183229446411, "learning_rate": 7.76837480627171e-06, "loss": 0.14515866041183473, "step": 22190 }, { "epoch": 0.5847818427871967, "grad_norm": 2.5111470222473145, "learning_rate": 7.760059397330225e-06, "loss": 0.13093550205230714, "step": 22200 }, { "epoch": 0.5850452580316955, "grad_norm": 2.2028727531433105, "learning_rate": 7.75174561871415e-06, "loss": 0.137522554397583, "step": 22210 }, { "epoch": 0.5853086732761942, "grad_norm": 1.2737462520599365, "learning_rate": 7.743433476474608e-06, "loss": 0.13734430074691772, "step": 22220 }, { "epoch": 0.5855720885206929, "grad_norm": 2.9930458068847656, "learning_rate": 7.73512297666154e-06, "loss": 0.1351541519165039, "step": 22230 }, { "epoch": 0.5858355037651917, "grad_norm": 1.20274817943573, "learning_rate": 7.726814125323683e-06, "loss": 0.1317118525505066, "step": 22240 }, { "epoch": 0.5860989190096904, "grad_norm": 3.215444564819336, "learning_rate": 7.718506928508583e-06, "loss": 0.13690128326416015, "step": 22250 }, { "epoch": 0.5863623342541892, "grad_norm": 2.234407901763916, "learning_rate": 7.71020139226257e-06, "loss": 0.1321993350982666, "step": 22260 }, { "epoch": 0.5866257494986878, "grad_norm": 1.5209683179855347, "learning_rate": 7.701897522630772e-06, "loss": 0.1389221429824829, "step": 22270 }, { "epoch": 0.5868891647431866, "grad_norm": 1.7848658561706543, "learning_rate": 7.693595325657109e-06, "loss": 0.14440956115722656, "step": 22280 }, { "epoch": 0.5871525799876853, "grad_norm": 1.1000940799713135, "learning_rate": 7.685294807384273e-06, "loss": 0.13652210235595702, "step": 22290 }, { "epoch": 0.5874159952321841, "grad_norm": 1.9608829021453857, "learning_rate": 7.676995973853743e-06, "loss": 0.13127510547637938, "step": 22300 }, { "epoch": 0.5876794104766828, "grad_norm": 1.678728461265564, "learning_rate": 7.668698831105768e-06, "loss": 0.12787717580795288, "step": 22310 }, { "epoch": 0.5879428257211815, "grad_norm": 1.6505913734436035, "learning_rate": 7.66040338517936e-06, "loss": 0.15045320987701416, "step": 22320 }, { "epoch": 0.5882062409656803, "grad_norm": 1.2271833419799805, "learning_rate": 7.652109642112308e-06, "loss": 0.1416846990585327, "step": 22330 }, { "epoch": 0.588469656210179, "grad_norm": 0.9187130331993103, "learning_rate": 7.643817607941157e-06, "loss": 0.12472922801971435, "step": 22340 }, { "epoch": 0.5887330714546778, "grad_norm": 1.7777491807937622, "learning_rate": 7.635527288701201e-06, "loss": 0.13109701871871948, "step": 22350 }, { "epoch": 0.5889964866991765, "grad_norm": 2.3057756423950195, "learning_rate": 7.6272386904265005e-06, "loss": 0.1417819380760193, "step": 22360 }, { "epoch": 0.5892599019436753, "grad_norm": 1.3261091709136963, "learning_rate": 7.618951819149848e-06, "loss": 0.12894318103790284, "step": 22370 }, { "epoch": 0.589523317188174, "grad_norm": 1.9180238246917725, "learning_rate": 7.610666680902791e-06, "loss": 0.1162794589996338, "step": 22380 }, { "epoch": 0.5897867324326727, "grad_norm": 1.0280283689498901, "learning_rate": 7.602383281715608e-06, "loss": 0.12699910402297973, "step": 22390 }, { "epoch": 0.5900501476771715, "grad_norm": 3.5772705078125, "learning_rate": 7.594101627617312e-06, "loss": 0.13075460195541383, "step": 22400 }, { "epoch": 0.5903135629216701, "grad_norm": 3.136131525039673, "learning_rate": 7.585821724635656e-06, "loss": 0.14569852352142335, "step": 22410 }, { "epoch": 0.5905769781661689, "grad_norm": 2.243835687637329, "learning_rate": 7.577543578797097e-06, "loss": 0.13479584455490112, "step": 22420 }, { "epoch": 0.5908403934106676, "grad_norm": 3.8731164932250977, "learning_rate": 7.569267196126844e-06, "loss": 0.13311846256256105, "step": 22430 }, { "epoch": 0.5911038086551664, "grad_norm": 1.450075626373291, "learning_rate": 7.560992582648793e-06, "loss": 0.1502191424369812, "step": 22440 }, { "epoch": 0.5913672238996651, "grad_norm": 0.7981085181236267, "learning_rate": 7.552719744385566e-06, "loss": 0.1389840602874756, "step": 22450 }, { "epoch": 0.5916306391441639, "grad_norm": 1.2468152046203613, "learning_rate": 7.544448687358495e-06, "loss": 0.14516721963882445, "step": 22460 }, { "epoch": 0.5918940543886626, "grad_norm": 0.8845928311347961, "learning_rate": 7.536179417587612e-06, "loss": 0.14512939453125, "step": 22470 }, { "epoch": 0.5921574696331613, "grad_norm": 2.053637981414795, "learning_rate": 7.527911941091644e-06, "loss": 0.1372859239578247, "step": 22480 }, { "epoch": 0.5924208848776601, "grad_norm": 2.028012275695801, "learning_rate": 7.519646263888023e-06, "loss": 0.1443265914916992, "step": 22490 }, { "epoch": 0.5926843001221588, "grad_norm": 1.3210680484771729, "learning_rate": 7.511382391992861e-06, "loss": 0.13392438888549804, "step": 22500 }, { "epoch": 0.5929477153666576, "grad_norm": 2.427366256713867, "learning_rate": 7.503120331420963e-06, "loss": 0.13758856058120728, "step": 22510 }, { "epoch": 0.5932111306111563, "grad_norm": 1.3613890409469604, "learning_rate": 7.494860088185815e-06, "loss": 0.1360732674598694, "step": 22520 }, { "epoch": 0.5934745458556551, "grad_norm": 1.0122222900390625, "learning_rate": 7.48660166829957e-06, "loss": 0.13666763305664062, "step": 22530 }, { "epoch": 0.5937379611001538, "grad_norm": 1.6932742595672607, "learning_rate": 7.478345077773075e-06, "loss": 0.13772120475769042, "step": 22540 }, { "epoch": 0.5940013763446526, "grad_norm": 2.48172926902771, "learning_rate": 7.470090322615822e-06, "loss": 0.1488971710205078, "step": 22550 }, { "epoch": 0.5942647915891512, "grad_norm": 3.407700777053833, "learning_rate": 7.4618374088359805e-06, "loss": 0.14080616235733032, "step": 22560 }, { "epoch": 0.5945282068336499, "grad_norm": 1.2014280557632446, "learning_rate": 7.45358634244038e-06, "loss": 0.12837806940078736, "step": 22570 }, { "epoch": 0.5947916220781487, "grad_norm": 2.556623935699463, "learning_rate": 7.4453371294345e-06, "loss": 0.1372031092643738, "step": 22580 }, { "epoch": 0.5950550373226474, "grad_norm": 1.6595507860183716, "learning_rate": 7.437089775822477e-06, "loss": 0.15442910194396972, "step": 22590 }, { "epoch": 0.5953184525671462, "grad_norm": 2.923326015472412, "learning_rate": 7.4288442876070865e-06, "loss": 0.14357154369354247, "step": 22600 }, { "epoch": 0.5955818678116449, "grad_norm": 2.073225975036621, "learning_rate": 7.420600670789753e-06, "loss": 0.14219048023223876, "step": 22610 }, { "epoch": 0.5958452830561437, "grad_norm": 1.5507218837738037, "learning_rate": 7.412358931370536e-06, "loss": 0.1405065178871155, "step": 22620 }, { "epoch": 0.5961086983006424, "grad_norm": 1.355630874633789, "learning_rate": 7.404119075348129e-06, "loss": 0.13446329832077025, "step": 22630 }, { "epoch": 0.5963721135451412, "grad_norm": 1.1937286853790283, "learning_rate": 7.3958811087198555e-06, "loss": 0.14013926982879638, "step": 22640 }, { "epoch": 0.5966355287896399, "grad_norm": 1.3555660247802734, "learning_rate": 7.3876450374816644e-06, "loss": 0.1272792100906372, "step": 22650 }, { "epoch": 0.5968989440341386, "grad_norm": 2.622286319732666, "learning_rate": 7.379410867628115e-06, "loss": 0.14103918075561522, "step": 22660 }, { "epoch": 0.5971623592786374, "grad_norm": 1.642360806465149, "learning_rate": 7.371178605152405e-06, "loss": 0.13884241580963136, "step": 22670 }, { "epoch": 0.597425774523136, "grad_norm": 1.9344121217727661, "learning_rate": 7.362948256046321e-06, "loss": 0.12724034786224364, "step": 22680 }, { "epoch": 0.5976891897676349, "grad_norm": 1.4742131233215332, "learning_rate": 7.354719826300266e-06, "loss": 0.14378056526184083, "step": 22690 }, { "epoch": 0.5979526050121335, "grad_norm": 3.579472064971924, "learning_rate": 7.346493321903251e-06, "loss": 0.1261759042739868, "step": 22700 }, { "epoch": 0.5982160202566323, "grad_norm": 2.1435060501098633, "learning_rate": 7.338268748842876e-06, "loss": 0.13784875869750976, "step": 22710 }, { "epoch": 0.598479435501131, "grad_norm": 3.8056864738464355, "learning_rate": 7.330046113105345e-06, "loss": 0.14773708581924438, "step": 22720 }, { "epoch": 0.5987428507456298, "grad_norm": 2.24951171875, "learning_rate": 7.321825420675444e-06, "loss": 0.12282736301422119, "step": 22730 }, { "epoch": 0.5990062659901285, "grad_norm": 1.5886726379394531, "learning_rate": 7.3136066775365455e-06, "loss": 0.12909483909606934, "step": 22740 }, { "epoch": 0.5992696812346272, "grad_norm": 1.4664946794509888, "learning_rate": 7.305389889670611e-06, "loss": 0.121628737449646, "step": 22750 }, { "epoch": 0.599533096479126, "grad_norm": 2.9848926067352295, "learning_rate": 7.2971750630581675e-06, "loss": 0.14677770137786866, "step": 22760 }, { "epoch": 0.5997965117236247, "grad_norm": 1.3805354833602905, "learning_rate": 7.2889622036783245e-06, "loss": 0.1291604995727539, "step": 22770 }, { "epoch": 0.6000599269681235, "grad_norm": 1.5942620038986206, "learning_rate": 7.280751317508756e-06, "loss": 0.13438335657119752, "step": 22780 }, { "epoch": 0.6003233422126222, "grad_norm": 1.8209117650985718, "learning_rate": 7.272542410525691e-06, "loss": 0.13644211292266845, "step": 22790 }, { "epoch": 0.600586757457121, "grad_norm": 3.8738183975219727, "learning_rate": 7.2643354887039395e-06, "loss": 0.14868764877319335, "step": 22800 }, { "epoch": 0.6008501727016197, "grad_norm": 2.6522278785705566, "learning_rate": 7.256130558016843e-06, "loss": 0.12848386764526368, "step": 22810 }, { "epoch": 0.6011135879461185, "grad_norm": 1.5552077293395996, "learning_rate": 7.247927624436304e-06, "loss": 0.13535420894622802, "step": 22820 }, { "epoch": 0.6013770031906172, "grad_norm": 3.6822032928466797, "learning_rate": 7.239726693932779e-06, "loss": 0.13787930011749266, "step": 22830 }, { "epoch": 0.6016404184351158, "grad_norm": 1.3158931732177734, "learning_rate": 7.2315277724752495e-06, "loss": 0.1236430287361145, "step": 22840 }, { "epoch": 0.6019038336796146, "grad_norm": 2.1411850452423096, "learning_rate": 7.223330866031253e-06, "loss": 0.14061403274536133, "step": 22850 }, { "epoch": 0.6021672489241133, "grad_norm": 1.5964853763580322, "learning_rate": 7.215135980566847e-06, "loss": 0.13146544694900514, "step": 22860 }, { "epoch": 0.6024306641686121, "grad_norm": 2.2085070610046387, "learning_rate": 7.2069431220466215e-06, "loss": 0.13685338497161864, "step": 22870 }, { "epoch": 0.6026940794131108, "grad_norm": 2.1121301651000977, "learning_rate": 7.198752296433697e-06, "loss": 0.13300530910491942, "step": 22880 }, { "epoch": 0.6029574946576096, "grad_norm": 0.9606817960739136, "learning_rate": 7.190563509689705e-06, "loss": 0.13394988775253297, "step": 22890 }, { "epoch": 0.6032209099021083, "grad_norm": 2.0468955039978027, "learning_rate": 7.182376767774803e-06, "loss": 0.13396542072296141, "step": 22900 }, { "epoch": 0.603484325146607, "grad_norm": 2.4099888801574707, "learning_rate": 7.174192076647653e-06, "loss": 0.13754119873046874, "step": 22910 }, { "epoch": 0.6037477403911058, "grad_norm": 0.9950894117355347, "learning_rate": 7.166009442265423e-06, "loss": 0.12917752265930177, "step": 22920 }, { "epoch": 0.6040111556356045, "grad_norm": 1.9289265871047974, "learning_rate": 7.15782887058379e-06, "loss": 0.12259975671768189, "step": 22930 }, { "epoch": 0.6042745708801033, "grad_norm": 2.570810317993164, "learning_rate": 7.149650367556926e-06, "loss": 0.13116365671157837, "step": 22940 }, { "epoch": 0.604537986124602, "grad_norm": 1.3623111248016357, "learning_rate": 7.141473939137498e-06, "loss": 0.13227441310882568, "step": 22950 }, { "epoch": 0.6048014013691008, "grad_norm": 1.8089680671691895, "learning_rate": 7.133299591276665e-06, "loss": 0.14200212955474853, "step": 22960 }, { "epoch": 0.6050648166135995, "grad_norm": 3.6160335540771484, "learning_rate": 7.12512732992407e-06, "loss": 0.1306261420249939, "step": 22970 }, { "epoch": 0.6053282318580983, "grad_norm": 2.1783838272094727, "learning_rate": 7.1169571610278375e-06, "loss": 0.14969260692596437, "step": 22980 }, { "epoch": 0.6055916471025969, "grad_norm": 1.7283658981323242, "learning_rate": 7.108789090534569e-06, "loss": 0.12638287544250487, "step": 22990 }, { "epoch": 0.6058550623470956, "grad_norm": 1.5811183452606201, "learning_rate": 7.100623124389334e-06, "loss": 0.13283437490463257, "step": 23000 }, { "epoch": 0.6061184775915944, "grad_norm": 1.821526050567627, "learning_rate": 7.092459268535685e-06, "loss": 0.1330440640449524, "step": 23010 }, { "epoch": 0.6063818928360931, "grad_norm": 2.4612200260162354, "learning_rate": 7.084297528915616e-06, "loss": 0.14278435707092285, "step": 23020 }, { "epoch": 0.6066453080805919, "grad_norm": 2.246642827987671, "learning_rate": 7.076137911469603e-06, "loss": 0.13101328611373902, "step": 23030 }, { "epoch": 0.6069087233250906, "grad_norm": 1.5381102561950684, "learning_rate": 7.067980422136565e-06, "loss": 0.13301801681518555, "step": 23040 }, { "epoch": 0.6071721385695894, "grad_norm": 2.9287185668945312, "learning_rate": 7.059825066853872e-06, "loss": 0.14615418910980224, "step": 23050 }, { "epoch": 0.6074355538140881, "grad_norm": 1.536143183708191, "learning_rate": 7.0516718515573445e-06, "loss": 0.13719289302825927, "step": 23060 }, { "epoch": 0.6076989690585869, "grad_norm": 1.3159434795379639, "learning_rate": 7.043520782181248e-06, "loss": 0.12906830310821532, "step": 23070 }, { "epoch": 0.6079623843030856, "grad_norm": 1.5194520950317383, "learning_rate": 7.035371864658276e-06, "loss": 0.13766695261001588, "step": 23080 }, { "epoch": 0.6082257995475843, "grad_norm": 2.8551788330078125, "learning_rate": 7.0272251049195684e-06, "loss": 0.1278231143951416, "step": 23090 }, { "epoch": 0.6084892147920831, "grad_norm": 1.3098939657211304, "learning_rate": 7.019080508894683e-06, "loss": 0.14169464111328126, "step": 23100 }, { "epoch": 0.6087526300365818, "grad_norm": 1.4177049398422241, "learning_rate": 7.010938082511614e-06, "loss": 0.1300576448440552, "step": 23110 }, { "epoch": 0.6090160452810806, "grad_norm": 1.0407593250274658, "learning_rate": 7.0027978316967705e-06, "loss": 0.12088308334350586, "step": 23120 }, { "epoch": 0.6092794605255792, "grad_norm": 1.7285243272781372, "learning_rate": 6.994659762374967e-06, "loss": 0.13834679126739502, "step": 23130 }, { "epoch": 0.609542875770078, "grad_norm": 2.054643154144287, "learning_rate": 6.986523880469458e-06, "loss": 0.14034926891326904, "step": 23140 }, { "epoch": 0.6098062910145767, "grad_norm": 1.6518820524215698, "learning_rate": 6.978390191901878e-06, "loss": 0.1524626851081848, "step": 23150 }, { "epoch": 0.6100697062590755, "grad_norm": 3.1637284755706787, "learning_rate": 6.970258702592277e-06, "loss": 0.14723507165908814, "step": 23160 }, { "epoch": 0.6103331215035742, "grad_norm": 1.7525873184204102, "learning_rate": 6.962129418459109e-06, "loss": 0.13325049877166747, "step": 23170 }, { "epoch": 0.6105965367480729, "grad_norm": 1.1153079271316528, "learning_rate": 6.954002345419214e-06, "loss": 0.13512058258056642, "step": 23180 }, { "epoch": 0.6108599519925717, "grad_norm": 1.54268479347229, "learning_rate": 6.945877489387829e-06, "loss": 0.14808454513549804, "step": 23190 }, { "epoch": 0.6111233672370704, "grad_norm": 1.0965834856033325, "learning_rate": 6.9377548562785735e-06, "loss": 0.14037853479385376, "step": 23200 }, { "epoch": 0.6113867824815692, "grad_norm": 2.896287679672241, "learning_rate": 6.929634452003447e-06, "loss": 0.13868200778961182, "step": 23210 }, { "epoch": 0.6116501977260679, "grad_norm": 2.224553346633911, "learning_rate": 6.921516282472837e-06, "loss": 0.14316998720169066, "step": 23220 }, { "epoch": 0.6119136129705667, "grad_norm": 1.713212013244629, "learning_rate": 6.91340035359549e-06, "loss": 0.14898806810379028, "step": 23230 }, { "epoch": 0.6121770282150654, "grad_norm": 2.2148144245147705, "learning_rate": 6.905286671278536e-06, "loss": 0.14626084566116332, "step": 23240 }, { "epoch": 0.6124404434595642, "grad_norm": 2.471961736679077, "learning_rate": 6.897175241427462e-06, "loss": 0.1323764681816101, "step": 23250 }, { "epoch": 0.6127038587040629, "grad_norm": 1.2745777368545532, "learning_rate": 6.889066069946107e-06, "loss": 0.14306488037109374, "step": 23260 }, { "epoch": 0.6129672739485615, "grad_norm": 1.7173922061920166, "learning_rate": 6.880959162736691e-06, "loss": 0.13600170612335205, "step": 23270 }, { "epoch": 0.6132306891930603, "grad_norm": 2.6325559616088867, "learning_rate": 6.872854525699761e-06, "loss": 0.14241223335266112, "step": 23280 }, { "epoch": 0.613494104437559, "grad_norm": 2.3235156536102295, "learning_rate": 6.8647521647342205e-06, "loss": 0.1327200174331665, "step": 23290 }, { "epoch": 0.6137575196820578, "grad_norm": 1.6949186325073242, "learning_rate": 6.856652085737323e-06, "loss": 0.13710912466049194, "step": 23300 }, { "epoch": 0.6140209349265565, "grad_norm": 1.3266109228134155, "learning_rate": 6.848554294604652e-06, "loss": 0.13030214309692384, "step": 23310 }, { "epoch": 0.6142843501710553, "grad_norm": 1.2555327415466309, "learning_rate": 6.8404587972301316e-06, "loss": 0.1300485372543335, "step": 23320 }, { "epoch": 0.614547765415554, "grad_norm": 2.3682243824005127, "learning_rate": 6.8323655995060135e-06, "loss": 0.14855685234069824, "step": 23330 }, { "epoch": 0.6148111806600528, "grad_norm": 1.836645483970642, "learning_rate": 6.824274707322874e-06, "loss": 0.14299777746200562, "step": 23340 }, { "epoch": 0.6150745959045515, "grad_norm": 2.7066051959991455, "learning_rate": 6.816186126569615e-06, "loss": 0.14256632328033447, "step": 23350 }, { "epoch": 0.6153380111490502, "grad_norm": 4.03224515914917, "learning_rate": 6.808099863133453e-06, "loss": 0.14379985332489015, "step": 23360 }, { "epoch": 0.615601426393549, "grad_norm": 1.3842401504516602, "learning_rate": 6.800015922899923e-06, "loss": 0.14149465560913085, "step": 23370 }, { "epoch": 0.6158648416380477, "grad_norm": 1.5259644985198975, "learning_rate": 6.7919343117528655e-06, "loss": 0.13309569358825685, "step": 23380 }, { "epoch": 0.6161282568825465, "grad_norm": 3.115347146987915, "learning_rate": 6.783855035574418e-06, "loss": 0.1406582236289978, "step": 23390 }, { "epoch": 0.6163916721270452, "grad_norm": 1.0316205024719238, "learning_rate": 6.775778100245038e-06, "loss": 0.1415982127189636, "step": 23400 }, { "epoch": 0.616655087371544, "grad_norm": 3.5068910121917725, "learning_rate": 6.76770351164346e-06, "loss": 0.13311314582824707, "step": 23410 }, { "epoch": 0.6169185026160426, "grad_norm": 4.354078769683838, "learning_rate": 6.759631275646715e-06, "loss": 0.13312662839889527, "step": 23420 }, { "epoch": 0.6171819178605413, "grad_norm": 3.005330801010132, "learning_rate": 6.751561398130133e-06, "loss": 0.13902747631072998, "step": 23430 }, { "epoch": 0.6174453331050401, "grad_norm": 2.4590654373168945, "learning_rate": 6.743493884967312e-06, "loss": 0.13506897687911987, "step": 23440 }, { "epoch": 0.6177087483495388, "grad_norm": 2.6364471912384033, "learning_rate": 6.735428742030141e-06, "loss": 0.12325876951217651, "step": 23450 }, { "epoch": 0.6179721635940376, "grad_norm": 1.9513821601867676, "learning_rate": 6.7273659751887775e-06, "loss": 0.14014732837677002, "step": 23460 }, { "epoch": 0.6182355788385363, "grad_norm": 2.106910228729248, "learning_rate": 6.719305590311649e-06, "loss": 0.13900833129882811, "step": 23470 }, { "epoch": 0.6184989940830351, "grad_norm": 2.1737310886383057, "learning_rate": 6.7112475932654554e-06, "loss": 0.13346797227859497, "step": 23480 }, { "epoch": 0.6187624093275338, "grad_norm": 2.141326904296875, "learning_rate": 6.70319198991515e-06, "loss": 0.13533480167388917, "step": 23490 }, { "epoch": 0.6190258245720326, "grad_norm": 2.2818427085876465, "learning_rate": 6.695138786123952e-06, "loss": 0.12738900184631347, "step": 23500 }, { "epoch": 0.6192892398165313, "grad_norm": 1.758859634399414, "learning_rate": 6.687087987753331e-06, "loss": 0.12968196868896484, "step": 23510 }, { "epoch": 0.61955265506103, "grad_norm": 2.1958203315734863, "learning_rate": 6.679039600663001e-06, "loss": 0.14438673257827758, "step": 23520 }, { "epoch": 0.6198160703055288, "grad_norm": 1.3194913864135742, "learning_rate": 6.670993630710929e-06, "loss": 0.13432841300964354, "step": 23530 }, { "epoch": 0.6200794855500275, "grad_norm": 2.9537153244018555, "learning_rate": 6.662950083753319e-06, "loss": 0.12557411193847656, "step": 23540 }, { "epoch": 0.6203429007945263, "grad_norm": 1.8925057649612427, "learning_rate": 6.654908965644609e-06, "loss": 0.14791357517242432, "step": 23550 }, { "epoch": 0.6206063160390249, "grad_norm": 2.4424819946289062, "learning_rate": 6.646870282237474e-06, "loss": 0.1305706024169922, "step": 23560 }, { "epoch": 0.6208697312835237, "grad_norm": 1.1741371154785156, "learning_rate": 6.6388340393828125e-06, "loss": 0.14553802013397216, "step": 23570 }, { "epoch": 0.6211331465280224, "grad_norm": 2.025547504425049, "learning_rate": 6.6308002429297516e-06, "loss": 0.1302317976951599, "step": 23580 }, { "epoch": 0.6213965617725212, "grad_norm": 1.6406062841415405, "learning_rate": 6.622768898725634e-06, "loss": 0.12102110385894775, "step": 23590 }, { "epoch": 0.6216599770170199, "grad_norm": 2.5114545822143555, "learning_rate": 6.614740012616017e-06, "loss": 0.14331157207489015, "step": 23600 }, { "epoch": 0.6219233922615186, "grad_norm": 3.50799298286438, "learning_rate": 6.606713590444673e-06, "loss": 0.13226501941680907, "step": 23610 }, { "epoch": 0.6221868075060174, "grad_norm": 3.262235164642334, "learning_rate": 6.598689638053577e-06, "loss": 0.13176226615905762, "step": 23620 }, { "epoch": 0.6224502227505161, "grad_norm": 1.7546546459197998, "learning_rate": 6.590668161282903e-06, "loss": 0.13263158798217772, "step": 23630 }, { "epoch": 0.6227136379950149, "grad_norm": 1.983640432357788, "learning_rate": 6.582649165971037e-06, "loss": 0.1336248278617859, "step": 23640 }, { "epoch": 0.6229770532395136, "grad_norm": 1.0711021423339844, "learning_rate": 6.574632657954542e-06, "loss": 0.12607941627502442, "step": 23650 }, { "epoch": 0.6232404684840124, "grad_norm": 1.8643027544021606, "learning_rate": 6.5666186430681835e-06, "loss": 0.13193429708480836, "step": 23660 }, { "epoch": 0.6235038837285111, "grad_norm": 2.4540374279022217, "learning_rate": 6.558607127144905e-06, "loss": 0.12190977334976197, "step": 23670 }, { "epoch": 0.6237672989730099, "grad_norm": 1.7889375686645508, "learning_rate": 6.550598116015832e-06, "loss": 0.13258901834487916, "step": 23680 }, { "epoch": 0.6240307142175086, "grad_norm": 5.424026966094971, "learning_rate": 6.54259161551027e-06, "loss": 0.14460155963897706, "step": 23690 }, { "epoch": 0.6242941294620072, "grad_norm": 1.6094707250595093, "learning_rate": 6.534587631455696e-06, "loss": 0.13167259693145753, "step": 23700 }, { "epoch": 0.624557544706506, "grad_norm": 1.3782246112823486, "learning_rate": 6.5265861696777556e-06, "loss": 0.14363408088684082, "step": 23710 }, { "epoch": 0.6248209599510047, "grad_norm": 2.778296947479248, "learning_rate": 6.518587236000259e-06, "loss": 0.13469027280807494, "step": 23720 }, { "epoch": 0.6250843751955035, "grad_norm": 2.6629040241241455, "learning_rate": 6.510590836245166e-06, "loss": 0.12325677871704102, "step": 23730 }, { "epoch": 0.6253477904400022, "grad_norm": 1.9555919170379639, "learning_rate": 6.502596976232616e-06, "loss": 0.13497759103775026, "step": 23740 }, { "epoch": 0.625611205684501, "grad_norm": 2.6523778438568115, "learning_rate": 6.494605661780874e-06, "loss": 0.13491945266723632, "step": 23750 }, { "epoch": 0.6258746209289997, "grad_norm": 1.6180658340454102, "learning_rate": 6.486616898706365e-06, "loss": 0.13064122200012207, "step": 23760 }, { "epoch": 0.6261380361734985, "grad_norm": 1.2653387784957886, "learning_rate": 6.47863069282366e-06, "loss": 0.1304025411605835, "step": 23770 }, { "epoch": 0.6264014514179972, "grad_norm": 1.8806467056274414, "learning_rate": 6.47064704994546e-06, "loss": 0.13124511241912842, "step": 23780 }, { "epoch": 0.6266648666624959, "grad_norm": 1.2443821430206299, "learning_rate": 6.462665975882607e-06, "loss": 0.13148884773254393, "step": 23790 }, { "epoch": 0.6269282819069947, "grad_norm": 2.170485496520996, "learning_rate": 6.454687476444071e-06, "loss": 0.133293080329895, "step": 23800 }, { "epoch": 0.6271916971514934, "grad_norm": 1.5137957334518433, "learning_rate": 6.446711557436944e-06, "loss": 0.11665506362915039, "step": 23810 }, { "epoch": 0.6274551123959922, "grad_norm": 1.4680763483047485, "learning_rate": 6.438738224666452e-06, "loss": 0.13703539371490478, "step": 23820 }, { "epoch": 0.6277185276404909, "grad_norm": 1.8164074420928955, "learning_rate": 6.4307674839359225e-06, "loss": 0.13738350868225097, "step": 23830 }, { "epoch": 0.6279819428849897, "grad_norm": 2.447295665740967, "learning_rate": 6.42279934104681e-06, "loss": 0.1322417140007019, "step": 23840 }, { "epoch": 0.6282453581294883, "grad_norm": 2.087611198425293, "learning_rate": 6.4148338017986725e-06, "loss": 0.1296220302581787, "step": 23850 }, { "epoch": 0.6285087733739871, "grad_norm": 1.546128273010254, "learning_rate": 6.406870871989166e-06, "loss": 0.1194232702255249, "step": 23860 }, { "epoch": 0.6287721886184858, "grad_norm": 1.9357444047927856, "learning_rate": 6.398910557414064e-06, "loss": 0.1392584800720215, "step": 23870 }, { "epoch": 0.6290356038629845, "grad_norm": 2.3190932273864746, "learning_rate": 6.390952863867224e-06, "loss": 0.12028157711029053, "step": 23880 }, { "epoch": 0.6292990191074833, "grad_norm": 2.8212623596191406, "learning_rate": 6.382997797140591e-06, "loss": 0.1369701147079468, "step": 23890 }, { "epoch": 0.629562434351982, "grad_norm": 1.3269035816192627, "learning_rate": 6.375045363024217e-06, "loss": 0.13087592124938965, "step": 23900 }, { "epoch": 0.6298258495964808, "grad_norm": 2.8857808113098145, "learning_rate": 6.367095567306219e-06, "loss": 0.14141921997070311, "step": 23910 }, { "epoch": 0.6300892648409795, "grad_norm": 1.5036447048187256, "learning_rate": 6.359148415772805e-06, "loss": 0.11441586017608643, "step": 23920 }, { "epoch": 0.6303526800854783, "grad_norm": 3.587218761444092, "learning_rate": 6.351203914208254e-06, "loss": 0.14222766160964967, "step": 23930 }, { "epoch": 0.630616095329977, "grad_norm": 1.4589827060699463, "learning_rate": 6.343262068394914e-06, "loss": 0.13629329204559326, "step": 23940 }, { "epoch": 0.6308795105744757, "grad_norm": 2.339580535888672, "learning_rate": 6.335322884113208e-06, "loss": 0.1434022903442383, "step": 23950 }, { "epoch": 0.6311429258189745, "grad_norm": 2.4862654209136963, "learning_rate": 6.3273863671416146e-06, "loss": 0.13416773080825806, "step": 23960 }, { "epoch": 0.6314063410634732, "grad_norm": 3.1866683959960938, "learning_rate": 6.319452523256673e-06, "loss": 0.129569149017334, "step": 23970 }, { "epoch": 0.631669756307972, "grad_norm": 1.4239470958709717, "learning_rate": 6.31152135823298e-06, "loss": 0.13721191883087158, "step": 23980 }, { "epoch": 0.6319331715524706, "grad_norm": 0.9657994508743286, "learning_rate": 6.303592877843171e-06, "loss": 0.12756915092468263, "step": 23990 }, { "epoch": 0.6321965867969694, "grad_norm": 2.2617743015289307, "learning_rate": 6.295667087857948e-06, "loss": 0.1295264720916748, "step": 24000 }, { "epoch": 0.6324600020414681, "grad_norm": 3.2446699142456055, "learning_rate": 6.287743994046035e-06, "loss": 0.1245896577835083, "step": 24010 }, { "epoch": 0.6327234172859669, "grad_norm": 4.831447601318359, "learning_rate": 6.279823602174202e-06, "loss": 0.13242266178131104, "step": 24020 }, { "epoch": 0.6329868325304656, "grad_norm": 3.067967653274536, "learning_rate": 6.271905918007255e-06, "loss": 0.13720451593399047, "step": 24030 }, { "epoch": 0.6332502477749643, "grad_norm": 1.5573335886001587, "learning_rate": 6.263990947308022e-06, "loss": 0.13948116302490235, "step": 24040 }, { "epoch": 0.6335136630194631, "grad_norm": 1.304072618484497, "learning_rate": 6.2560786958373645e-06, "loss": 0.14146754741668702, "step": 24050 }, { "epoch": 0.6337770782639618, "grad_norm": 1.4448329210281372, "learning_rate": 6.248169169354159e-06, "loss": 0.13449311256408691, "step": 24060 }, { "epoch": 0.6340404935084606, "grad_norm": 1.8153716325759888, "learning_rate": 6.240262373615298e-06, "loss": 0.12981717586517333, "step": 24070 }, { "epoch": 0.6343039087529593, "grad_norm": 1.2793073654174805, "learning_rate": 6.232358314375693e-06, "loss": 0.12354384660720825, "step": 24080 }, { "epoch": 0.6345673239974581, "grad_norm": 2.0464980602264404, "learning_rate": 6.224456997388254e-06, "loss": 0.12045921087265014, "step": 24090 }, { "epoch": 0.6348307392419568, "grad_norm": 1.4478384256362915, "learning_rate": 6.2165584284039046e-06, "loss": 0.12762746810913086, "step": 24100 }, { "epoch": 0.6350941544864556, "grad_norm": 3.2348761558532715, "learning_rate": 6.208662613171566e-06, "loss": 0.13088676929473878, "step": 24110 }, { "epoch": 0.6353575697309543, "grad_norm": 3.4676523208618164, "learning_rate": 6.200769557438144e-06, "loss": 0.13449214696884154, "step": 24120 }, { "epoch": 0.6356209849754529, "grad_norm": 1.2244386672973633, "learning_rate": 6.192879266948554e-06, "loss": 0.13229589462280272, "step": 24130 }, { "epoch": 0.6358844002199517, "grad_norm": 2.1439852714538574, "learning_rate": 6.184991747445686e-06, "loss": 0.13118430376052856, "step": 24140 }, { "epoch": 0.6361478154644504, "grad_norm": 2.7168312072753906, "learning_rate": 6.1771070046704165e-06, "loss": 0.12395350933074951, "step": 24150 }, { "epoch": 0.6364112307089492, "grad_norm": 1.5067243576049805, "learning_rate": 6.169225044361604e-06, "loss": 0.1385481595993042, "step": 24160 }, { "epoch": 0.6366746459534479, "grad_norm": 4.100353717803955, "learning_rate": 6.161345872256075e-06, "loss": 0.1395488500595093, "step": 24170 }, { "epoch": 0.6369380611979467, "grad_norm": 1.7457060813903809, "learning_rate": 6.153469494088638e-06, "loss": 0.1322873830795288, "step": 24180 }, { "epoch": 0.6372014764424454, "grad_norm": 2.3779172897338867, "learning_rate": 6.1455959155920555e-06, "loss": 0.12720897197723388, "step": 24190 }, { "epoch": 0.6374648916869442, "grad_norm": 2.158050775527954, "learning_rate": 6.1377251424970555e-06, "loss": 0.12479932308197021, "step": 24200 }, { "epoch": 0.6377283069314429, "grad_norm": 2.429170608520508, "learning_rate": 6.1298571805323344e-06, "loss": 0.12935752868652345, "step": 24210 }, { "epoch": 0.6379917221759416, "grad_norm": 2.0747268199920654, "learning_rate": 6.121992035424529e-06, "loss": 0.12024786472320556, "step": 24220 }, { "epoch": 0.6382551374204404, "grad_norm": 1.3240482807159424, "learning_rate": 6.114129712898227e-06, "loss": 0.13565165996551515, "step": 24230 }, { "epoch": 0.6385185526649391, "grad_norm": 1.668627142906189, "learning_rate": 6.106270218675975e-06, "loss": 0.1308264136314392, "step": 24240 }, { "epoch": 0.6387819679094379, "grad_norm": 1.9308044910430908, "learning_rate": 6.098413558478244e-06, "loss": 0.13586515188217163, "step": 24250 }, { "epoch": 0.6390453831539366, "grad_norm": 1.366990327835083, "learning_rate": 6.090559738023456e-06, "loss": 0.12462676763534546, "step": 24260 }, { "epoch": 0.6393087983984354, "grad_norm": 1.7612029314041138, "learning_rate": 6.0827087630279605e-06, "loss": 0.1311100721359253, "step": 24270 }, { "epoch": 0.639572213642934, "grad_norm": 1.7062410116195679, "learning_rate": 6.074860639206028e-06, "loss": 0.12695069313049318, "step": 24280 }, { "epoch": 0.6398356288874328, "grad_norm": 1.2856471538543701, "learning_rate": 6.067015372269872e-06, "loss": 0.12629104852676393, "step": 24290 }, { "epoch": 0.6400990441319315, "grad_norm": 1.6726995706558228, "learning_rate": 6.059172967929609e-06, "loss": 0.1269448757171631, "step": 24300 }, { "epoch": 0.6403624593764302, "grad_norm": 1.4109594821929932, "learning_rate": 6.0513334318932825e-06, "loss": 0.12444555759429932, "step": 24310 }, { "epoch": 0.640625874620929, "grad_norm": 1.3218586444854736, "learning_rate": 6.043496769866848e-06, "loss": 0.14012696743011474, "step": 24320 }, { "epoch": 0.6408892898654277, "grad_norm": 1.5134843587875366, "learning_rate": 6.035662987554154e-06, "loss": 0.12399135828018189, "step": 24330 }, { "epoch": 0.6411527051099265, "grad_norm": 1.6340813636779785, "learning_rate": 6.027832090656982e-06, "loss": 0.1316126823425293, "step": 24340 }, { "epoch": 0.6414161203544252, "grad_norm": 1.7409120798110962, "learning_rate": 6.020004084874984e-06, "loss": 0.13388981819152831, "step": 24350 }, { "epoch": 0.641679535598924, "grad_norm": 1.623937964439392, "learning_rate": 6.012178975905725e-06, "loss": 0.12719976902008057, "step": 24360 }, { "epoch": 0.6419429508434227, "grad_norm": 2.6205780506134033, "learning_rate": 6.0043567694446566e-06, "loss": 0.1164353370666504, "step": 24370 }, { "epoch": 0.6422063660879215, "grad_norm": 2.2199440002441406, "learning_rate": 5.9965374711851175e-06, "loss": 0.1253509044647217, "step": 24380 }, { "epoch": 0.6424697813324202, "grad_norm": 1.3395609855651855, "learning_rate": 5.9887210868183325e-06, "loss": 0.13419992923736573, "step": 24390 }, { "epoch": 0.6427331965769189, "grad_norm": 1.3170294761657715, "learning_rate": 5.9809076220334024e-06, "loss": 0.13342374563217163, "step": 24400 }, { "epoch": 0.6429966118214177, "grad_norm": 2.310504198074341, "learning_rate": 5.973097082517301e-06, "loss": 0.13266849517822266, "step": 24410 }, { "epoch": 0.6432600270659163, "grad_norm": 3.5624446868896484, "learning_rate": 5.965289473954879e-06, "loss": 0.12006151676177979, "step": 24420 }, { "epoch": 0.6435234423104151, "grad_norm": 1.8615652322769165, "learning_rate": 5.95748480202885e-06, "loss": 0.13880407810211182, "step": 24430 }, { "epoch": 0.6437868575549138, "grad_norm": 2.5235579013824463, "learning_rate": 5.949683072419792e-06, "loss": 0.12730271816253663, "step": 24440 }, { "epoch": 0.6440502727994126, "grad_norm": 2.400424003601074, "learning_rate": 5.941884290806142e-06, "loss": 0.12767372131347657, "step": 24450 }, { "epoch": 0.6443136880439113, "grad_norm": 2.2394840717315674, "learning_rate": 5.934088462864181e-06, "loss": 0.1368882179260254, "step": 24460 }, { "epoch": 0.6445771032884101, "grad_norm": 2.886436939239502, "learning_rate": 5.926295594268059e-06, "loss": 0.13658559322357178, "step": 24470 }, { "epoch": 0.6448405185329088, "grad_norm": 2.307382106781006, "learning_rate": 5.918505690689759e-06, "loss": 0.12928464412689208, "step": 24480 }, { "epoch": 0.6451039337774075, "grad_norm": 2.660278081893921, "learning_rate": 5.910718757799105e-06, "loss": 0.1377729892730713, "step": 24490 }, { "epoch": 0.6453673490219063, "grad_norm": 2.093390703201294, "learning_rate": 5.902934801263766e-06, "loss": 0.14395923614501954, "step": 24500 }, { "epoch": 0.645630764266405, "grad_norm": 1.2309894561767578, "learning_rate": 5.89515382674924e-06, "loss": 0.1246678352355957, "step": 24510 }, { "epoch": 0.6458941795109038, "grad_norm": 1.8627684116363525, "learning_rate": 5.88737583991886e-06, "loss": 0.14668872356414794, "step": 24520 }, { "epoch": 0.6461575947554025, "grad_norm": 2.1053733825683594, "learning_rate": 5.879600846433777e-06, "loss": 0.12865524291992186, "step": 24530 }, { "epoch": 0.6464210099999013, "grad_norm": 2.443143606185913, "learning_rate": 5.871828851952965e-06, "loss": 0.13343188762664795, "step": 24540 }, { "epoch": 0.6466844252444, "grad_norm": 2.5137405395507812, "learning_rate": 5.864059862133221e-06, "loss": 0.12984349727630615, "step": 24550 }, { "epoch": 0.6469478404888986, "grad_norm": 1.093026876449585, "learning_rate": 5.856293882629147e-06, "loss": 0.116237473487854, "step": 24560 }, { "epoch": 0.6472112557333974, "grad_norm": 1.959536075592041, "learning_rate": 5.848530919093162e-06, "loss": 0.13346190452575685, "step": 24570 }, { "epoch": 0.6474746709778961, "grad_norm": 2.0060014724731445, "learning_rate": 5.840770977175485e-06, "loss": 0.12806546688079834, "step": 24580 }, { "epoch": 0.6477380862223949, "grad_norm": 2.573974609375, "learning_rate": 5.833014062524129e-06, "loss": 0.12706689834594725, "step": 24590 }, { "epoch": 0.6480015014668936, "grad_norm": 1.9685213565826416, "learning_rate": 5.825260180784923e-06, "loss": 0.1305103063583374, "step": 24600 }, { "epoch": 0.6482649167113924, "grad_norm": 2.0548195838928223, "learning_rate": 5.817509337601475e-06, "loss": 0.13225802183151245, "step": 24610 }, { "epoch": 0.6485283319558911, "grad_norm": 2.9862656593322754, "learning_rate": 5.8097615386151686e-06, "loss": 0.13601422309875488, "step": 24620 }, { "epoch": 0.6487917472003899, "grad_norm": 2.445239782333374, "learning_rate": 5.8020167894652035e-06, "loss": 0.12475478649139404, "step": 24630 }, { "epoch": 0.6490551624448886, "grad_norm": 1.8642785549163818, "learning_rate": 5.79427509578853e-06, "loss": 0.12881593704223632, "step": 24640 }, { "epoch": 0.6493185776893873, "grad_norm": 2.620107889175415, "learning_rate": 5.786536463219899e-06, "loss": 0.1348121404647827, "step": 24650 }, { "epoch": 0.6495819929338861, "grad_norm": 1.5753095149993896, "learning_rate": 5.778800897391812e-06, "loss": 0.12267103195190429, "step": 24660 }, { "epoch": 0.6498454081783848, "grad_norm": 1.4537851810455322, "learning_rate": 5.771068403934545e-06, "loss": 0.13852349519729615, "step": 24670 }, { "epoch": 0.6501088234228836, "grad_norm": 2.636244535446167, "learning_rate": 5.76333898847615e-06, "loss": 0.12331650257110596, "step": 24680 }, { "epoch": 0.6503722386673823, "grad_norm": 1.8543598651885986, "learning_rate": 5.755612656642424e-06, "loss": 0.12746315002441405, "step": 24690 }, { "epoch": 0.650635653911881, "grad_norm": 2.1532487869262695, "learning_rate": 5.747889414056925e-06, "loss": 0.1266958475112915, "step": 24700 }, { "epoch": 0.6508990691563797, "grad_norm": 2.159271001815796, "learning_rate": 5.740169266340963e-06, "loss": 0.12645018100738525, "step": 24710 }, { "epoch": 0.6511624844008785, "grad_norm": 1.336330771446228, "learning_rate": 5.732452219113589e-06, "loss": 0.13402295112609863, "step": 24720 }, { "epoch": 0.6514258996453772, "grad_norm": 3.0489697456359863, "learning_rate": 5.724738277991611e-06, "loss": 0.13174808025360107, "step": 24730 }, { "epoch": 0.6516893148898759, "grad_norm": 1.5266603231430054, "learning_rate": 5.717027448589568e-06, "loss": 0.1389584183692932, "step": 24740 }, { "epoch": 0.6519527301343747, "grad_norm": 1.2988907098770142, "learning_rate": 5.709319736519725e-06, "loss": 0.11785199642181396, "step": 24750 }, { "epoch": 0.6522161453788734, "grad_norm": 1.3463072776794434, "learning_rate": 5.701615147392096e-06, "loss": 0.1411191701889038, "step": 24760 }, { "epoch": 0.6524795606233722, "grad_norm": 1.6514453887939453, "learning_rate": 5.6939136868144075e-06, "loss": 0.13104233741760254, "step": 24770 }, { "epoch": 0.6527429758678709, "grad_norm": 1.6986024379730225, "learning_rate": 5.686215360392127e-06, "loss": 0.1234686017036438, "step": 24780 }, { "epoch": 0.6530063911123697, "grad_norm": 4.7255449295043945, "learning_rate": 5.678520173728414e-06, "loss": 0.12461962699890136, "step": 24790 }, { "epoch": 0.6532698063568684, "grad_norm": 1.9910571575164795, "learning_rate": 5.67082813242416e-06, "loss": 0.13504679203033448, "step": 24800 }, { "epoch": 0.6535332216013672, "grad_norm": 1.2628834247589111, "learning_rate": 5.663139242077971e-06, "loss": 0.13406074047088623, "step": 24810 }, { "epoch": 0.6537966368458659, "grad_norm": 1.7861618995666504, "learning_rate": 5.6554535082861496e-06, "loss": 0.14265209436416626, "step": 24820 }, { "epoch": 0.6540600520903646, "grad_norm": 1.9199743270874023, "learning_rate": 5.647770936642705e-06, "loss": 0.12243030071258545, "step": 24830 }, { "epoch": 0.6543234673348634, "grad_norm": 2.468273162841797, "learning_rate": 5.6400915327393435e-06, "loss": 0.13025959730148315, "step": 24840 }, { "epoch": 0.654586882579362, "grad_norm": 1.3651357889175415, "learning_rate": 5.632415302165464e-06, "loss": 0.13164572715759276, "step": 24850 }, { "epoch": 0.6548502978238608, "grad_norm": 2.4329030513763428, "learning_rate": 5.624742250508164e-06, "loss": 0.12352888584136963, "step": 24860 }, { "epoch": 0.6551137130683595, "grad_norm": 1.5882138013839722, "learning_rate": 5.617072383352222e-06, "loss": 0.13032468557357788, "step": 24870 }, { "epoch": 0.6553771283128583, "grad_norm": 2.012160062789917, "learning_rate": 5.6094057062800886e-06, "loss": 0.11882902383804321, "step": 24880 }, { "epoch": 0.655640543557357, "grad_norm": 1.2780117988586426, "learning_rate": 5.601742224871911e-06, "loss": 0.1401231050491333, "step": 24890 }, { "epoch": 0.6559039588018558, "grad_norm": 2.7542340755462646, "learning_rate": 5.594081944705496e-06, "loss": 0.13081231117248535, "step": 24900 }, { "epoch": 0.6561673740463545, "grad_norm": 2.011812210083008, "learning_rate": 5.586424871356337e-06, "loss": 0.12610366344451904, "step": 24910 }, { "epoch": 0.6564307892908532, "grad_norm": 1.735095739364624, "learning_rate": 5.578771010397572e-06, "loss": 0.1306707501411438, "step": 24920 }, { "epoch": 0.656694204535352, "grad_norm": 1.5062230825424194, "learning_rate": 5.571120367400008e-06, "loss": 0.13661863803863525, "step": 24930 }, { "epoch": 0.6569576197798507, "grad_norm": 4.458468437194824, "learning_rate": 5.563472947932123e-06, "loss": 0.12536828517913817, "step": 24940 }, { "epoch": 0.6572210350243495, "grad_norm": 1.866593837738037, "learning_rate": 5.555828757560035e-06, "loss": 0.12481102943420411, "step": 24950 }, { "epoch": 0.6574844502688482, "grad_norm": 1.7339571714401245, "learning_rate": 5.548187801847513e-06, "loss": 0.13650138378143312, "step": 24960 }, { "epoch": 0.657747865513347, "grad_norm": 1.3340421915054321, "learning_rate": 5.5405500863559795e-06, "loss": 0.13193936347961427, "step": 24970 }, { "epoch": 0.6580112807578457, "grad_norm": 2.165153980255127, "learning_rate": 5.532915616644484e-06, "loss": 0.12607823610305785, "step": 24980 }, { "epoch": 0.6582746960023445, "grad_norm": 1.3781907558441162, "learning_rate": 5.525284398269733e-06, "loss": 0.11604745388031006, "step": 24990 }, { "epoch": 0.6585381112468431, "grad_norm": 2.237640142440796, "learning_rate": 5.517656436786059e-06, "loss": 0.12796638011932374, "step": 25000 }, { "epoch": 0.6588015264913418, "grad_norm": 3.4568850994110107, "learning_rate": 5.510031737745405e-06, "loss": 0.12945657968521118, "step": 25010 }, { "epoch": 0.6590649417358406, "grad_norm": 1.4982258081436157, "learning_rate": 5.50241030669737e-06, "loss": 0.1300530791282654, "step": 25020 }, { "epoch": 0.6593283569803393, "grad_norm": 1.4327235221862793, "learning_rate": 5.4947921491891565e-06, "loss": 0.126871919631958, "step": 25030 }, { "epoch": 0.6595917722248381, "grad_norm": 1.1393970251083374, "learning_rate": 5.487177270765588e-06, "loss": 0.13019275665283203, "step": 25040 }, { "epoch": 0.6598551874693368, "grad_norm": 1.7901639938354492, "learning_rate": 5.479565676969101e-06, "loss": 0.1235609769821167, "step": 25050 }, { "epoch": 0.6601186027138356, "grad_norm": 1.3930466175079346, "learning_rate": 5.47195737333974e-06, "loss": 0.12125270366668701, "step": 25060 }, { "epoch": 0.6603820179583343, "grad_norm": 1.3740041255950928, "learning_rate": 5.464352365415162e-06, "loss": 0.12466232776641846, "step": 25070 }, { "epoch": 0.660645433202833, "grad_norm": 1.9611543416976929, "learning_rate": 5.456750658730616e-06, "loss": 0.13712122440338134, "step": 25080 }, { "epoch": 0.6609088484473318, "grad_norm": 1.5871446132659912, "learning_rate": 5.449152258818955e-06, "loss": 0.13182642459869384, "step": 25090 }, { "epoch": 0.6611722636918305, "grad_norm": 3.0277998447418213, "learning_rate": 5.4415571712106184e-06, "loss": 0.13397729396820068, "step": 25100 }, { "epoch": 0.6614356789363293, "grad_norm": 1.781994342803955, "learning_rate": 5.433965401433639e-06, "loss": 0.11753547191619873, "step": 25110 }, { "epoch": 0.661699094180828, "grad_norm": 2.0219457149505615, "learning_rate": 5.426376955013641e-06, "loss": 0.1295809268951416, "step": 25120 }, { "epoch": 0.6619625094253268, "grad_norm": 1.9544864892959595, "learning_rate": 5.4187918374738154e-06, "loss": 0.13811752796173096, "step": 25130 }, { "epoch": 0.6622259246698254, "grad_norm": 1.0879260301589966, "learning_rate": 5.4112100543349365e-06, "loss": 0.12401814460754394, "step": 25140 }, { "epoch": 0.6624893399143242, "grad_norm": 2.481025218963623, "learning_rate": 5.403631611115362e-06, "loss": 0.13092324733734131, "step": 25150 }, { "epoch": 0.6627527551588229, "grad_norm": 2.2066502571105957, "learning_rate": 5.396056513331005e-06, "loss": 0.126028311252594, "step": 25160 }, { "epoch": 0.6630161704033216, "grad_norm": 2.097850799560547, "learning_rate": 5.388484766495348e-06, "loss": 0.13947641849517822, "step": 25170 }, { "epoch": 0.6632795856478204, "grad_norm": 2.9836785793304443, "learning_rate": 5.3809163761194364e-06, "loss": 0.137762451171875, "step": 25180 }, { "epoch": 0.6635430008923191, "grad_norm": 1.1449602842330933, "learning_rate": 5.373351347711865e-06, "loss": 0.10769634246826172, "step": 25190 }, { "epoch": 0.6638064161368179, "grad_norm": 2.49053955078125, "learning_rate": 5.365789686778797e-06, "loss": 0.13664284944534302, "step": 25200 }, { "epoch": 0.6640698313813166, "grad_norm": 2.471658706665039, "learning_rate": 5.358231398823937e-06, "loss": 0.11819164752960205, "step": 25210 }, { "epoch": 0.6643332466258154, "grad_norm": 2.225069522857666, "learning_rate": 5.350676489348521e-06, "loss": 0.12350388765335082, "step": 25220 }, { "epoch": 0.6645966618703141, "grad_norm": 1.965186595916748, "learning_rate": 5.3431249638513475e-06, "loss": 0.14150989055633545, "step": 25230 }, { "epoch": 0.6648600771148129, "grad_norm": 1.5043431520462036, "learning_rate": 5.335576827828737e-06, "loss": 0.1264769673347473, "step": 25240 }, { "epoch": 0.6651234923593116, "grad_norm": 2.176565408706665, "learning_rate": 5.328032086774557e-06, "loss": 0.12204139232635498, "step": 25250 }, { "epoch": 0.6653869076038103, "grad_norm": 2.386937141418457, "learning_rate": 5.320490746180189e-06, "loss": 0.12067791223526, "step": 25260 }, { "epoch": 0.665650322848309, "grad_norm": 1.8121510744094849, "learning_rate": 5.312952811534542e-06, "loss": 0.12915782928466796, "step": 25270 }, { "epoch": 0.6659137380928077, "grad_norm": 1.4561764001846313, "learning_rate": 5.305418288324058e-06, "loss": 0.1265083909034729, "step": 25280 }, { "epoch": 0.6661771533373065, "grad_norm": 2.0944669246673584, "learning_rate": 5.297887182032686e-06, "loss": 0.12736833095550537, "step": 25290 }, { "epoch": 0.6664405685818052, "grad_norm": 2.1389613151550293, "learning_rate": 5.29035949814189e-06, "loss": 0.11665053367614746, "step": 25300 }, { "epoch": 0.666703983826304, "grad_norm": 2.6053309440612793, "learning_rate": 5.2828352421306415e-06, "loss": 0.12919648885726928, "step": 25310 }, { "epoch": 0.6669673990708027, "grad_norm": 1.9850714206695557, "learning_rate": 5.275314419475417e-06, "loss": 0.11512933969497681, "step": 25320 }, { "epoch": 0.6672308143153015, "grad_norm": 2.5950536727905273, "learning_rate": 5.267797035650202e-06, "loss": 0.11939346790313721, "step": 25330 }, { "epoch": 0.6674942295598002, "grad_norm": 1.4475500583648682, "learning_rate": 5.2602830961264704e-06, "loss": 0.13259328603744508, "step": 25340 }, { "epoch": 0.6677576448042989, "grad_norm": 2.9021992683410645, "learning_rate": 5.252772606373186e-06, "loss": 0.1310909628868103, "step": 25350 }, { "epoch": 0.6680210600487977, "grad_norm": 1.8734848499298096, "learning_rate": 5.245265571856812e-06, "loss": 0.11653389930725097, "step": 25360 }, { "epoch": 0.6682844752932964, "grad_norm": 2.008847713470459, "learning_rate": 5.237761998041291e-06, "loss": 0.13375200033187867, "step": 25370 }, { "epoch": 0.6685478905377952, "grad_norm": 1.6617369651794434, "learning_rate": 5.230261890388053e-06, "loss": 0.13722680807113646, "step": 25380 }, { "epoch": 0.6688113057822939, "grad_norm": 1.5283138751983643, "learning_rate": 5.2227652543559924e-06, "loss": 0.12367751598358154, "step": 25390 }, { "epoch": 0.6690747210267927, "grad_norm": 2.745687246322632, "learning_rate": 5.215272095401483e-06, "loss": 0.1259669065475464, "step": 25400 }, { "epoch": 0.6693381362712914, "grad_norm": 1.637078046798706, "learning_rate": 5.207782418978378e-06, "loss": 0.12629094123840331, "step": 25410 }, { "epoch": 0.6696015515157901, "grad_norm": 1.4034409523010254, "learning_rate": 5.200296230537982e-06, "loss": 0.12661452293395997, "step": 25420 }, { "epoch": 0.6698649667602888, "grad_norm": 1.6262989044189453, "learning_rate": 5.192813535529066e-06, "loss": 0.13413641452789307, "step": 25430 }, { "epoch": 0.6701283820047875, "grad_norm": 1.7855287790298462, "learning_rate": 5.1853343393978585e-06, "loss": 0.12054316997528076, "step": 25440 }, { "epoch": 0.6703917972492863, "grad_norm": 2.185941457748413, "learning_rate": 5.177858647588037e-06, "loss": 0.1275627374649048, "step": 25450 }, { "epoch": 0.670655212493785, "grad_norm": 2.11429500579834, "learning_rate": 5.170386465540741e-06, "loss": 0.12068359851837158, "step": 25460 }, { "epoch": 0.6709186277382838, "grad_norm": 1.8045390844345093, "learning_rate": 5.162917798694546e-06, "loss": 0.1265416622161865, "step": 25470 }, { "epoch": 0.6711820429827825, "grad_norm": 1.9909175634384155, "learning_rate": 5.15545265248546e-06, "loss": 0.13539612293243408, "step": 25480 }, { "epoch": 0.6714454582272813, "grad_norm": 1.6995505094528198, "learning_rate": 5.147991032346948e-06, "loss": 0.11081185340881347, "step": 25490 }, { "epoch": 0.67170887347178, "grad_norm": 1.6613028049468994, "learning_rate": 5.140532943709893e-06, "loss": 0.13140588998794556, "step": 25500 }, { "epoch": 0.6719722887162788, "grad_norm": 1.9568207263946533, "learning_rate": 5.133078392002623e-06, "loss": 0.12900121212005616, "step": 25510 }, { "epoch": 0.6722357039607775, "grad_norm": 1.8648637533187866, "learning_rate": 5.125627382650873e-06, "loss": 0.12296264171600342, "step": 25520 }, { "epoch": 0.6724991192052762, "grad_norm": 3.5467073917388916, "learning_rate": 5.118179921077808e-06, "loss": 0.13576477766036987, "step": 25530 }, { "epoch": 0.672762534449775, "grad_norm": 2.036374568939209, "learning_rate": 5.1107360127040206e-06, "loss": 0.13080509901046752, "step": 25540 }, { "epoch": 0.6730259496942737, "grad_norm": 2.063298463821411, "learning_rate": 5.103295662947504e-06, "loss": 0.12557960748672486, "step": 25550 }, { "epoch": 0.6732893649387724, "grad_norm": 4.359191417694092, "learning_rate": 5.095858877223665e-06, "loss": 0.11703747510910034, "step": 25560 }, { "epoch": 0.6735527801832711, "grad_norm": 1.5920201539993286, "learning_rate": 5.088425660945319e-06, "loss": 0.12311210632324218, "step": 25570 }, { "epoch": 0.6738161954277699, "grad_norm": 1.7187516689300537, "learning_rate": 5.080996019522675e-06, "loss": 0.119242262840271, "step": 25580 }, { "epoch": 0.6740796106722686, "grad_norm": 4.3557634353637695, "learning_rate": 5.073569958363356e-06, "loss": 0.13516607284545898, "step": 25590 }, { "epoch": 0.6743430259167673, "grad_norm": 1.4164161682128906, "learning_rate": 5.06614748287237e-06, "loss": 0.11265807151794434, "step": 25600 }, { "epoch": 0.6746064411612661, "grad_norm": 1.9293524026870728, "learning_rate": 5.058728598452103e-06, "loss": 0.13572036027908324, "step": 25610 }, { "epoch": 0.6748698564057648, "grad_norm": 1.4970288276672363, "learning_rate": 5.05131331050235e-06, "loss": 0.11595454216003417, "step": 25620 }, { "epoch": 0.6751332716502636, "grad_norm": 1.404593586921692, "learning_rate": 5.043901624420273e-06, "loss": 0.1308261752128601, "step": 25630 }, { "epoch": 0.6753966868947623, "grad_norm": 3.1665353775024414, "learning_rate": 5.036493545600421e-06, "loss": 0.11732860803604125, "step": 25640 }, { "epoch": 0.6756601021392611, "grad_norm": 1.6482093334197998, "learning_rate": 5.029089079434708e-06, "loss": 0.11948127746582031, "step": 25650 }, { "epoch": 0.6759235173837598, "grad_norm": 2.525040626525879, "learning_rate": 5.021688231312423e-06, "loss": 0.11172350645065307, "step": 25660 }, { "epoch": 0.6761869326282586, "grad_norm": 1.280282735824585, "learning_rate": 5.014291006620229e-06, "loss": 0.12283248901367187, "step": 25670 }, { "epoch": 0.6764503478727573, "grad_norm": 1.5789624452590942, "learning_rate": 5.006897410742142e-06, "loss": 0.1235762357711792, "step": 25680 }, { "epoch": 0.676713763117256, "grad_norm": 1.2596571445465088, "learning_rate": 4.999507449059542e-06, "loss": 0.1177514672279358, "step": 25690 }, { "epoch": 0.6769771783617548, "grad_norm": 1.8212652206420898, "learning_rate": 4.992121126951159e-06, "loss": 0.1338188648223877, "step": 25700 }, { "epoch": 0.6772405936062534, "grad_norm": 1.7249115705490112, "learning_rate": 4.984738449793074e-06, "loss": 0.1260267972946167, "step": 25710 }, { "epoch": 0.6775040088507522, "grad_norm": 2.138927459716797, "learning_rate": 4.977359422958731e-06, "loss": 0.12675334215164186, "step": 25720 }, { "epoch": 0.6777674240952509, "grad_norm": 1.4222583770751953, "learning_rate": 4.96998405181889e-06, "loss": 0.1307663917541504, "step": 25730 }, { "epoch": 0.6780308393397497, "grad_norm": 1.754241704940796, "learning_rate": 4.962612341741666e-06, "loss": 0.1086653470993042, "step": 25740 }, { "epoch": 0.6782942545842484, "grad_norm": 1.7567720413208008, "learning_rate": 4.955244298092514e-06, "loss": 0.12436672449111938, "step": 25750 }, { "epoch": 0.6785576698287472, "grad_norm": 1.5771957635879517, "learning_rate": 4.947879926234212e-06, "loss": 0.13228044509887696, "step": 25760 }, { "epoch": 0.6788210850732459, "grad_norm": 3.056753635406494, "learning_rate": 4.940519231526867e-06, "loss": 0.13154469728469848, "step": 25770 }, { "epoch": 0.6790845003177446, "grad_norm": 1.3839590549468994, "learning_rate": 4.933162219327909e-06, "loss": 0.1172297477722168, "step": 25780 }, { "epoch": 0.6793479155622434, "grad_norm": 1.6796401739120483, "learning_rate": 4.925808894992087e-06, "loss": 0.12140449285507202, "step": 25790 }, { "epoch": 0.6796113308067421, "grad_norm": 1.9257500171661377, "learning_rate": 4.918459263871474e-06, "loss": 0.12366465330123902, "step": 25800 }, { "epoch": 0.6798747460512409, "grad_norm": 2.26816987991333, "learning_rate": 4.911113331315448e-06, "loss": 0.11358758211135864, "step": 25810 }, { "epoch": 0.6801381612957396, "grad_norm": 4.651535987854004, "learning_rate": 4.903771102670686e-06, "loss": 0.12529826164245605, "step": 25820 }, { "epoch": 0.6804015765402384, "grad_norm": 1.72089421749115, "learning_rate": 4.896432583281189e-06, "loss": 0.1203508734703064, "step": 25830 }, { "epoch": 0.680664991784737, "grad_norm": 2.2176575660705566, "learning_rate": 4.88909777848824e-06, "loss": 0.1255162239074707, "step": 25840 }, { "epoch": 0.6809284070292358, "grad_norm": 1.7549612522125244, "learning_rate": 4.881766693630439e-06, "loss": 0.1316195011138916, "step": 25850 }, { "epoch": 0.6811918222737345, "grad_norm": 1.8761966228485107, "learning_rate": 4.874439334043654e-06, "loss": 0.1229283094406128, "step": 25860 }, { "epoch": 0.6814552375182332, "grad_norm": 1.457851529121399, "learning_rate": 4.8671157050610525e-06, "loss": 0.12962268590927123, "step": 25870 }, { "epoch": 0.681718652762732, "grad_norm": 2.5484132766723633, "learning_rate": 4.859795812013099e-06, "loss": 0.12990341186523438, "step": 25880 }, { "epoch": 0.6819820680072307, "grad_norm": 1.5767356157302856, "learning_rate": 4.852479660227518e-06, "loss": 0.12391210794448852, "step": 25890 }, { "epoch": 0.6822454832517295, "grad_norm": 2.7414870262145996, "learning_rate": 4.8451672550293236e-06, "loss": 0.11233649253845215, "step": 25900 }, { "epoch": 0.6825088984962282, "grad_norm": 2.3302276134490967, "learning_rate": 4.837858601740799e-06, "loss": 0.1204877257347107, "step": 25910 }, { "epoch": 0.682772313740727, "grad_norm": 3.0352540016174316, "learning_rate": 4.830553705681496e-06, "loss": 0.1213829517364502, "step": 25920 }, { "epoch": 0.6830357289852257, "grad_norm": 1.8734699487686157, "learning_rate": 4.823252572168236e-06, "loss": 0.1239163875579834, "step": 25930 }, { "epoch": 0.6832991442297245, "grad_norm": 1.2328468561172485, "learning_rate": 4.815955206515104e-06, "loss": 0.11766842603683472, "step": 25940 }, { "epoch": 0.6835625594742232, "grad_norm": 2.967926025390625, "learning_rate": 4.808661614033422e-06, "loss": 0.13179595470428468, "step": 25950 }, { "epoch": 0.6838259747187219, "grad_norm": 1.294344425201416, "learning_rate": 4.801371800031796e-06, "loss": 0.12739224433898927, "step": 25960 }, { "epoch": 0.6840893899632207, "grad_norm": 2.003908157348633, "learning_rate": 4.794085769816057e-06, "loss": 0.10330522060394287, "step": 25970 }, { "epoch": 0.6843528052077194, "grad_norm": 1.608959674835205, "learning_rate": 4.786803528689304e-06, "loss": 0.1335829257965088, "step": 25980 }, { "epoch": 0.6846162204522181, "grad_norm": 3.23142147064209, "learning_rate": 4.779525081951857e-06, "loss": 0.13077845573425292, "step": 25990 }, { "epoch": 0.6848796356967168, "grad_norm": 1.7600892782211304, "learning_rate": 4.772250434901281e-06, "loss": 0.12317354679107666, "step": 26000 }, { "epoch": 0.6851430509412156, "grad_norm": 2.1127707958221436, "learning_rate": 4.7649795928323875e-06, "loss": 0.11472207307815552, "step": 26010 }, { "epoch": 0.6854064661857143, "grad_norm": 1.500970482826233, "learning_rate": 4.757712561037203e-06, "loss": 0.12222540378570557, "step": 26020 }, { "epoch": 0.6856698814302131, "grad_norm": 2.355318546295166, "learning_rate": 4.750449344804989e-06, "loss": 0.11857664585113525, "step": 26030 }, { "epoch": 0.6859332966747118, "grad_norm": 4.913366794586182, "learning_rate": 4.743189949422227e-06, "loss": 0.13991940021514893, "step": 26040 }, { "epoch": 0.6861967119192105, "grad_norm": 1.969147801399231, "learning_rate": 4.735934380172613e-06, "loss": 0.11878516674041747, "step": 26050 }, { "epoch": 0.6864601271637093, "grad_norm": 2.0877227783203125, "learning_rate": 4.728682642337072e-06, "loss": 0.12298836708068847, "step": 26060 }, { "epoch": 0.686723542408208, "grad_norm": 1.1420387029647827, "learning_rate": 4.7214347411937325e-06, "loss": 0.12285630702972412, "step": 26070 }, { "epoch": 0.6869869576527068, "grad_norm": 1.155465006828308, "learning_rate": 4.714190682017918e-06, "loss": 0.1228485345840454, "step": 26080 }, { "epoch": 0.6872503728972055, "grad_norm": 2.004664421081543, "learning_rate": 4.706950470082177e-06, "loss": 0.12048528194427491, "step": 26090 }, { "epoch": 0.6875137881417043, "grad_norm": 2.3042054176330566, "learning_rate": 4.699714110656242e-06, "loss": 0.1134720802307129, "step": 26100 }, { "epoch": 0.687777203386203, "grad_norm": 2.856109380722046, "learning_rate": 4.6924816090070575e-06, "loss": 0.1216476321220398, "step": 26110 }, { "epoch": 0.6880406186307017, "grad_norm": 2.685292959213257, "learning_rate": 4.685252970398742e-06, "loss": 0.12939616441726684, "step": 26120 }, { "epoch": 0.6883040338752004, "grad_norm": 1.4079540967941284, "learning_rate": 4.678028200092607e-06, "loss": 0.12574300765991211, "step": 26130 }, { "epoch": 0.6885674491196991, "grad_norm": 1.4030389785766602, "learning_rate": 4.670807303347161e-06, "loss": 0.12580063343048095, "step": 26140 }, { "epoch": 0.6888308643641979, "grad_norm": 2.139693021774292, "learning_rate": 4.663590285418082e-06, "loss": 0.12005487680435181, "step": 26150 }, { "epoch": 0.6890942796086966, "grad_norm": 3.0167086124420166, "learning_rate": 4.656377151558224e-06, "loss": 0.11046824455261231, "step": 26160 }, { "epoch": 0.6893576948531954, "grad_norm": 1.9827876091003418, "learning_rate": 4.64916790701762e-06, "loss": 0.1379615545272827, "step": 26170 }, { "epoch": 0.6896211100976941, "grad_norm": 2.575227737426758, "learning_rate": 4.641962557043462e-06, "loss": 0.11227961778640747, "step": 26180 }, { "epoch": 0.6898845253421929, "grad_norm": 2.8510305881500244, "learning_rate": 4.634761106880127e-06, "loss": 0.12001692056655884, "step": 26190 }, { "epoch": 0.6901479405866916, "grad_norm": 2.7223093509674072, "learning_rate": 4.627563561769136e-06, "loss": 0.12134416103363037, "step": 26200 }, { "epoch": 0.6904113558311903, "grad_norm": 1.7502139806747437, "learning_rate": 4.620369926949164e-06, "loss": 0.11709914207458497, "step": 26210 }, { "epoch": 0.6906747710756891, "grad_norm": 2.075230360031128, "learning_rate": 4.613180207656061e-06, "loss": 0.1058706521987915, "step": 26220 }, { "epoch": 0.6909381863201878, "grad_norm": 1.5123090744018555, "learning_rate": 4.60599440912281e-06, "loss": 0.12172591686248779, "step": 26230 }, { "epoch": 0.6912016015646866, "grad_norm": 2.52662992477417, "learning_rate": 4.598812536579545e-06, "loss": 0.11612820625305176, "step": 26240 }, { "epoch": 0.6914650168091853, "grad_norm": 1.7782092094421387, "learning_rate": 4.591634595253541e-06, "loss": 0.11321591138839722, "step": 26250 }, { "epoch": 0.6917284320536841, "grad_norm": 1.4248547554016113, "learning_rate": 4.584460590369214e-06, "loss": 0.13409662246704102, "step": 26260 }, { "epoch": 0.6919918472981827, "grad_norm": 1.67210054397583, "learning_rate": 4.577290527148118e-06, "loss": 0.12261521816253662, "step": 26270 }, { "epoch": 0.6922552625426815, "grad_norm": 1.9297171831130981, "learning_rate": 4.5701244108089325e-06, "loss": 0.1248747706413269, "step": 26280 }, { "epoch": 0.6925186777871802, "grad_norm": 1.5500236749649048, "learning_rate": 4.562962246567466e-06, "loss": 0.13318222761154175, "step": 26290 }, { "epoch": 0.6927820930316789, "grad_norm": 2.9978997707366943, "learning_rate": 4.555804039636649e-06, "loss": 0.12905218601226806, "step": 26300 }, { "epoch": 0.6930455082761777, "grad_norm": 4.690601348876953, "learning_rate": 4.548649795226532e-06, "loss": 0.12290036678314209, "step": 26310 }, { "epoch": 0.6933089235206764, "grad_norm": 1.5590612888336182, "learning_rate": 4.541499518544292e-06, "loss": 0.1223304033279419, "step": 26320 }, { "epoch": 0.6935723387651752, "grad_norm": 3.086876392364502, "learning_rate": 4.5343532147941985e-06, "loss": 0.11616771221160889, "step": 26330 }, { "epoch": 0.6938357540096739, "grad_norm": 2.297889471054077, "learning_rate": 4.527210889177639e-06, "loss": 0.1373889207839966, "step": 26340 }, { "epoch": 0.6940991692541727, "grad_norm": 3.859842300415039, "learning_rate": 4.520072546893115e-06, "loss": 0.13419479131698608, "step": 26350 }, { "epoch": 0.6943625844986714, "grad_norm": 1.7768802642822266, "learning_rate": 4.512938193136216e-06, "loss": 0.11878393888473511, "step": 26360 }, { "epoch": 0.6946259997431702, "grad_norm": 1.3206181526184082, "learning_rate": 4.5058078330996315e-06, "loss": 0.1198842167854309, "step": 26370 }, { "epoch": 0.6948894149876689, "grad_norm": 2.190000295639038, "learning_rate": 4.498681471973145e-06, "loss": 0.120261549949646, "step": 26380 }, { "epoch": 0.6951528302321676, "grad_norm": 2.0979511737823486, "learning_rate": 4.491559114943628e-06, "loss": 0.11332600116729737, "step": 26390 }, { "epoch": 0.6954162454766664, "grad_norm": 2.0393307209014893, "learning_rate": 4.484440767195044e-06, "loss": 0.13478963375091552, "step": 26400 }, { "epoch": 0.695679660721165, "grad_norm": 1.324168086051941, "learning_rate": 4.477326433908435e-06, "loss": 0.12447428703308105, "step": 26410 }, { "epoch": 0.6959430759656638, "grad_norm": 1.0555838346481323, "learning_rate": 4.4702161202619085e-06, "loss": 0.12055637836456298, "step": 26420 }, { "epoch": 0.6962064912101625, "grad_norm": 1.5911062955856323, "learning_rate": 4.463109831430668e-06, "loss": 0.1213998556137085, "step": 26430 }, { "epoch": 0.6964699064546613, "grad_norm": 2.1988425254821777, "learning_rate": 4.456007572586971e-06, "loss": 0.1221612811088562, "step": 26440 }, { "epoch": 0.69673332169916, "grad_norm": 3.1018199920654297, "learning_rate": 4.448909348900159e-06, "loss": 0.12259836196899414, "step": 26450 }, { "epoch": 0.6969967369436588, "grad_norm": 2.683924436569214, "learning_rate": 4.441815165536612e-06, "loss": 0.13543896675109862, "step": 26460 }, { "epoch": 0.6972601521881575, "grad_norm": 2.051347494125366, "learning_rate": 4.434725027659786e-06, "loss": 0.12135701179504395, "step": 26470 }, { "epoch": 0.6975235674326562, "grad_norm": 2.2480764389038086, "learning_rate": 4.427638940430192e-06, "loss": 0.11239570379257202, "step": 26480 }, { "epoch": 0.697786982677155, "grad_norm": 2.0730552673339844, "learning_rate": 4.420556909005392e-06, "loss": 0.12687418460845948, "step": 26490 }, { "epoch": 0.6980503979216537, "grad_norm": 3.8925397396087646, "learning_rate": 4.41347893853999e-06, "loss": 0.1330235481262207, "step": 26500 }, { "epoch": 0.6983138131661525, "grad_norm": 2.878728151321411, "learning_rate": 4.40640503418564e-06, "loss": 0.11185575723648071, "step": 26510 }, { "epoch": 0.6985772284106512, "grad_norm": 1.4713603258132935, "learning_rate": 4.39933520109103e-06, "loss": 0.10709481239318848, "step": 26520 }, { "epoch": 0.69884064365515, "grad_norm": 3.2316958904266357, "learning_rate": 4.3922694444018985e-06, "loss": 0.11402443647384644, "step": 26530 }, { "epoch": 0.6991040588996487, "grad_norm": 2.436147928237915, "learning_rate": 4.385207769261007e-06, "loss": 0.11515792608261108, "step": 26540 }, { "epoch": 0.6993674741441475, "grad_norm": 3.7004361152648926, "learning_rate": 4.378150180808136e-06, "loss": 0.12449809312820434, "step": 26550 }, { "epoch": 0.6996308893886461, "grad_norm": 1.8889561891555786, "learning_rate": 4.371096684180114e-06, "loss": 0.11892021894454956, "step": 26560 }, { "epoch": 0.6998943046331448, "grad_norm": 2.8461780548095703, "learning_rate": 4.364047284510774e-06, "loss": 0.09949966669082641, "step": 26570 }, { "epoch": 0.7001577198776436, "grad_norm": 1.9564266204833984, "learning_rate": 4.357001986930982e-06, "loss": 0.11947216987609863, "step": 26580 }, { "epoch": 0.7004211351221423, "grad_norm": 2.841643810272217, "learning_rate": 4.349960796568601e-06, "loss": 0.11944887638092042, "step": 26590 }, { "epoch": 0.7006845503666411, "grad_norm": 1.9505993127822876, "learning_rate": 4.342923718548512e-06, "loss": 0.12187380790710449, "step": 26600 }, { "epoch": 0.7009479656111398, "grad_norm": 1.672797441482544, "learning_rate": 4.335890757992611e-06, "loss": 0.11446785926818848, "step": 26610 }, { "epoch": 0.7012113808556386, "grad_norm": 1.784255862236023, "learning_rate": 4.328861920019787e-06, "loss": 0.10584161281585694, "step": 26620 }, { "epoch": 0.7014747961001373, "grad_norm": 4.046283721923828, "learning_rate": 4.321837209745931e-06, "loss": 0.12324796915054322, "step": 26630 }, { "epoch": 0.701738211344636, "grad_norm": 1.8300449848175049, "learning_rate": 4.3148166322839305e-06, "loss": 0.11865694522857666, "step": 26640 }, { "epoch": 0.7020016265891348, "grad_norm": 2.207509756088257, "learning_rate": 4.3078001927436605e-06, "loss": 0.11210267543792725, "step": 26650 }, { "epoch": 0.7022650418336335, "grad_norm": 1.7588509321212769, "learning_rate": 4.300787896231997e-06, "loss": 0.1143575668334961, "step": 26660 }, { "epoch": 0.7025284570781323, "grad_norm": 2.697012186050415, "learning_rate": 4.293779747852791e-06, "loss": 0.11260255575180053, "step": 26670 }, { "epoch": 0.702791872322631, "grad_norm": 3.624657392501831, "learning_rate": 4.286775752706865e-06, "loss": 0.12030225992202759, "step": 26680 }, { "epoch": 0.7030552875671298, "grad_norm": 2.595905303955078, "learning_rate": 4.279775915892038e-06, "loss": 0.12674369812011718, "step": 26690 }, { "epoch": 0.7033187028116284, "grad_norm": 1.49913489818573, "learning_rate": 4.2727802425030885e-06, "loss": 0.10415291786193848, "step": 26700 }, { "epoch": 0.7035821180561272, "grad_norm": 1.5260857343673706, "learning_rate": 4.265788737631779e-06, "loss": 0.11000747680664062, "step": 26710 }, { "epoch": 0.7038455333006259, "grad_norm": 1.3679254055023193, "learning_rate": 4.258801406366817e-06, "loss": 0.12300143241882325, "step": 26720 }, { "epoch": 0.7041089485451246, "grad_norm": 3.466766595840454, "learning_rate": 4.251818253793885e-06, "loss": 0.1191406488418579, "step": 26730 }, { "epoch": 0.7043723637896234, "grad_norm": 1.8511537313461304, "learning_rate": 4.244839284995627e-06, "loss": 0.1243741750717163, "step": 26740 }, { "epoch": 0.7046357790341221, "grad_norm": 2.178011655807495, "learning_rate": 4.237864505051638e-06, "loss": 0.12644134759902953, "step": 26750 }, { "epoch": 0.7048991942786209, "grad_norm": 2.280947685241699, "learning_rate": 4.230893919038461e-06, "loss": 0.12442110776901245, "step": 26760 }, { "epoch": 0.7051626095231196, "grad_norm": 2.3759615421295166, "learning_rate": 4.223927532029592e-06, "loss": 0.10959227085113525, "step": 26770 }, { "epoch": 0.7054260247676184, "grad_norm": 1.7773149013519287, "learning_rate": 4.21696534909546e-06, "loss": 0.13171743154525756, "step": 26780 }, { "epoch": 0.7056894400121171, "grad_norm": 1.8794255256652832, "learning_rate": 4.2100073753034535e-06, "loss": 0.11737065315246582, "step": 26790 }, { "epoch": 0.7059528552566159, "grad_norm": 1.9533369541168213, "learning_rate": 4.203053615717884e-06, "loss": 0.11770172119140625, "step": 26800 }, { "epoch": 0.7062162705011146, "grad_norm": 2.0200417041778564, "learning_rate": 4.196104075399986e-06, "loss": 0.12926700115203857, "step": 26810 }, { "epoch": 0.7064796857456133, "grad_norm": 1.8926966190338135, "learning_rate": 4.1891587594079485e-06, "loss": 0.12637197971343994, "step": 26820 }, { "epoch": 0.7067431009901121, "grad_norm": 2.4674925804138184, "learning_rate": 4.182217672796867e-06, "loss": 0.12594866752624512, "step": 26830 }, { "epoch": 0.7070065162346107, "grad_norm": 1.5826257467269897, "learning_rate": 4.175280820618763e-06, "loss": 0.116143798828125, "step": 26840 }, { "epoch": 0.7072699314791095, "grad_norm": 1.7604154348373413, "learning_rate": 4.168348207922577e-06, "loss": 0.1105802059173584, "step": 26850 }, { "epoch": 0.7075333467236082, "grad_norm": 2.0610592365264893, "learning_rate": 4.16141983975416e-06, "loss": 0.12238141298294067, "step": 26860 }, { "epoch": 0.707796761968107, "grad_norm": 3.1139044761657715, "learning_rate": 4.154495721156287e-06, "loss": 0.12805941104888915, "step": 26870 }, { "epoch": 0.7080601772126057, "grad_norm": 3.5406298637390137, "learning_rate": 4.147575857168623e-06, "loss": 0.12184069156646729, "step": 26880 }, { "epoch": 0.7083235924571045, "grad_norm": 1.2096083164215088, "learning_rate": 4.140660252827746e-06, "loss": 0.10384467840194703, "step": 26890 }, { "epoch": 0.7085870077016032, "grad_norm": 2.2088680267333984, "learning_rate": 4.13374891316713e-06, "loss": 0.11279842853546143, "step": 26900 }, { "epoch": 0.7088504229461019, "grad_norm": 1.3123871088027954, "learning_rate": 4.126841843217144e-06, "loss": 0.1136695146560669, "step": 26910 }, { "epoch": 0.7091138381906007, "grad_norm": 2.7104930877685547, "learning_rate": 4.1199390480050605e-06, "loss": 0.12076718807220459, "step": 26920 }, { "epoch": 0.7093772534350994, "grad_norm": 2.315394163131714, "learning_rate": 4.113040532555021e-06, "loss": 0.11268256902694702, "step": 26930 }, { "epoch": 0.7096406686795982, "grad_norm": 2.45595383644104, "learning_rate": 4.106146301888065e-06, "loss": 0.1212533950805664, "step": 26940 }, { "epoch": 0.7099040839240969, "grad_norm": 1.8824836015701294, "learning_rate": 4.0992563610221155e-06, "loss": 0.11253548860549926, "step": 26950 }, { "epoch": 0.7101674991685957, "grad_norm": 2.1704001426696777, "learning_rate": 4.092370714971966e-06, "loss": 0.1165687084197998, "step": 26960 }, { "epoch": 0.7104309144130944, "grad_norm": 2.707366943359375, "learning_rate": 4.085489368749287e-06, "loss": 0.12274092435836792, "step": 26970 }, { "epoch": 0.7106943296575932, "grad_norm": 2.1692728996276855, "learning_rate": 4.07861232736262e-06, "loss": 0.11850094795227051, "step": 26980 }, { "epoch": 0.7109577449020918, "grad_norm": 4.003645420074463, "learning_rate": 4.0717395958173665e-06, "loss": 0.11962003707885742, "step": 26990 }, { "epoch": 0.7112211601465905, "grad_norm": 3.863560438156128, "learning_rate": 4.064871179115807e-06, "loss": 0.11641445159912109, "step": 27000 }, { "epoch": 0.7114845753910893, "grad_norm": 2.163660764694214, "learning_rate": 4.0580070822570694e-06, "loss": 0.12070533037185668, "step": 27010 }, { "epoch": 0.711747990635588, "grad_norm": 1.346962809562683, "learning_rate": 4.051147310237129e-06, "loss": 0.1212616205215454, "step": 27020 }, { "epoch": 0.7120114058800868, "grad_norm": 1.6262186765670776, "learning_rate": 4.044291868048837e-06, "loss": 0.1105048656463623, "step": 27030 }, { "epoch": 0.7122748211245855, "grad_norm": 1.9795993566513062, "learning_rate": 4.0374407606818684e-06, "loss": 0.11527537107467652, "step": 27040 }, { "epoch": 0.7125382363690843, "grad_norm": 2.741405725479126, "learning_rate": 4.03059399312277e-06, "loss": 0.14091284275054933, "step": 27050 }, { "epoch": 0.712801651613583, "grad_norm": 2.771688222885132, "learning_rate": 4.0237515703549015e-06, "loss": 0.11555664539337158, "step": 27060 }, { "epoch": 0.7130650668580818, "grad_norm": 1.567628026008606, "learning_rate": 4.0169134973584726e-06, "loss": 0.10475755929946899, "step": 27070 }, { "epoch": 0.7133284821025805, "grad_norm": 1.721662998199463, "learning_rate": 4.010079779110537e-06, "loss": 0.12148547172546387, "step": 27080 }, { "epoch": 0.7135918973470792, "grad_norm": 1.563997745513916, "learning_rate": 4.003250420584966e-06, "loss": 0.13332996368408204, "step": 27090 }, { "epoch": 0.713855312591578, "grad_norm": 2.5253167152404785, "learning_rate": 3.996425426752459e-06, "loss": 0.13299083709716797, "step": 27100 }, { "epoch": 0.7141187278360767, "grad_norm": 2.495619773864746, "learning_rate": 3.989604802580545e-06, "loss": 0.12435919046401978, "step": 27110 }, { "epoch": 0.7143821430805755, "grad_norm": 1.858680248260498, "learning_rate": 3.982788553033561e-06, "loss": 0.11276297569274903, "step": 27120 }, { "epoch": 0.7146455583250741, "grad_norm": 2.4426095485687256, "learning_rate": 3.975976683072676e-06, "loss": 0.1298250675201416, "step": 27130 }, { "epoch": 0.714908973569573, "grad_norm": 1.9641033411026, "learning_rate": 3.9691691976558636e-06, "loss": 0.1255307674407959, "step": 27140 }, { "epoch": 0.7151723888140716, "grad_norm": 2.0114283561706543, "learning_rate": 3.9623661017378946e-06, "loss": 0.1229059100151062, "step": 27150 }, { "epoch": 0.7154358040585704, "grad_norm": 1.9771918058395386, "learning_rate": 3.955567400270365e-06, "loss": 0.1227752685546875, "step": 27160 }, { "epoch": 0.7156992193030691, "grad_norm": 2.0700504779815674, "learning_rate": 3.948773098201657e-06, "loss": 0.11443538665771484, "step": 27170 }, { "epoch": 0.7159626345475678, "grad_norm": 3.7748279571533203, "learning_rate": 3.941983200476969e-06, "loss": 0.11670223474502564, "step": 27180 }, { "epoch": 0.7162260497920666, "grad_norm": 2.204925060272217, "learning_rate": 3.935197712038267e-06, "loss": 0.11112852096557617, "step": 27190 }, { "epoch": 0.7164894650365653, "grad_norm": 1.4707427024841309, "learning_rate": 3.928416637824326e-06, "loss": 0.10862786769866943, "step": 27200 }, { "epoch": 0.7167528802810641, "grad_norm": 2.5636868476867676, "learning_rate": 3.921639982770709e-06, "loss": 0.12526813745498658, "step": 27210 }, { "epoch": 0.7170162955255628, "grad_norm": 1.3294296264648438, "learning_rate": 3.914867751809756e-06, "loss": 0.11803065538406372, "step": 27220 }, { "epoch": 0.7172797107700616, "grad_norm": 1.9625568389892578, "learning_rate": 3.908099949870588e-06, "loss": 0.12536382675170898, "step": 27230 }, { "epoch": 0.7175431260145603, "grad_norm": 2.313377618789673, "learning_rate": 3.901336581879102e-06, "loss": 0.11405071020126342, "step": 27240 }, { "epoch": 0.717806541259059, "grad_norm": 2.6272518634796143, "learning_rate": 3.894577652757968e-06, "loss": 0.12662557363510132, "step": 27250 }, { "epoch": 0.7180699565035578, "grad_norm": 1.4458234310150146, "learning_rate": 3.887823167426633e-06, "loss": 0.12098197937011719, "step": 27260 }, { "epoch": 0.7183333717480564, "grad_norm": 1.288330078125, "learning_rate": 3.8810731308013005e-06, "loss": 0.115581476688385, "step": 27270 }, { "epoch": 0.7185967869925552, "grad_norm": 1.6680947542190552, "learning_rate": 3.874327547794931e-06, "loss": 0.11370306015014649, "step": 27280 }, { "epoch": 0.7188602022370539, "grad_norm": 2.5179343223571777, "learning_rate": 3.8675864233172575e-06, "loss": 0.11925253868103028, "step": 27290 }, { "epoch": 0.7191236174815527, "grad_norm": 1.6462855339050293, "learning_rate": 3.860849762274759e-06, "loss": 0.11048314571380616, "step": 27300 }, { "epoch": 0.7193870327260514, "grad_norm": 1.7800883054733276, "learning_rate": 3.854117569570674e-06, "loss": 0.12192096710205078, "step": 27310 }, { "epoch": 0.7196504479705502, "grad_norm": 2.0920636653900146, "learning_rate": 3.8473898501049745e-06, "loss": 0.11608432531356812, "step": 27320 }, { "epoch": 0.7199138632150489, "grad_norm": 1.6187185049057007, "learning_rate": 3.840666608774385e-06, "loss": 0.11553473472595215, "step": 27330 }, { "epoch": 0.7201772784595476, "grad_norm": 2.7899701595306396, "learning_rate": 3.833947850472378e-06, "loss": 0.11555697917938232, "step": 27340 }, { "epoch": 0.7204406937040464, "grad_norm": 3.393972396850586, "learning_rate": 3.827233580089152e-06, "loss": 0.1346518039703369, "step": 27350 }, { "epoch": 0.7207041089485451, "grad_norm": 3.6094186305999756, "learning_rate": 3.82052380251164e-06, "loss": 0.12047668695449829, "step": 27360 }, { "epoch": 0.7209675241930439, "grad_norm": 1.7362536191940308, "learning_rate": 3.8138185226235114e-06, "loss": 0.10929501056671143, "step": 27370 }, { "epoch": 0.7212309394375426, "grad_norm": 2.8358283042907715, "learning_rate": 3.8071177453051523e-06, "loss": 0.12383179664611817, "step": 27380 }, { "epoch": 0.7214943546820414, "grad_norm": 3.0693860054016113, "learning_rate": 3.8004214754336853e-06, "loss": 0.11303399801254273, "step": 27390 }, { "epoch": 0.7217577699265401, "grad_norm": 1.7572977542877197, "learning_rate": 3.7937297178829458e-06, "loss": 0.10806598663330078, "step": 27400 }, { "epoch": 0.7220211851710389, "grad_norm": 2.6146721839904785, "learning_rate": 3.7870424775234703e-06, "loss": 0.11070849895477294, "step": 27410 }, { "epoch": 0.7222846004155375, "grad_norm": 1.4014419317245483, "learning_rate": 3.780359759222534e-06, "loss": 0.12298357486724854, "step": 27420 }, { "epoch": 0.7225480156600362, "grad_norm": 1.5138787031173706, "learning_rate": 3.773681567844103e-06, "loss": 0.116546368598938, "step": 27430 }, { "epoch": 0.722811430904535, "grad_norm": 1.7205500602722168, "learning_rate": 3.767007908248852e-06, "loss": 0.11358397006988526, "step": 27440 }, { "epoch": 0.7230748461490337, "grad_norm": 2.3584439754486084, "learning_rate": 3.7603387852941608e-06, "loss": 0.11013567447662354, "step": 27450 }, { "epoch": 0.7233382613935325, "grad_norm": 2.1929821968078613, "learning_rate": 3.753674203834099e-06, "loss": 0.10614191293716431, "step": 27460 }, { "epoch": 0.7236016766380312, "grad_norm": 2.929100275039673, "learning_rate": 3.747014168719445e-06, "loss": 0.10846130847930908, "step": 27470 }, { "epoch": 0.72386509188253, "grad_norm": 2.356372833251953, "learning_rate": 3.740358684797656e-06, "loss": 0.11826541423797607, "step": 27480 }, { "epoch": 0.7241285071270287, "grad_norm": 1.7445732355117798, "learning_rate": 3.7337077569128797e-06, "loss": 0.11199514865875244, "step": 27490 }, { "epoch": 0.7243919223715275, "grad_norm": 2.3352043628692627, "learning_rate": 3.7270613899059495e-06, "loss": 0.10780893564224243, "step": 27500 }, { "epoch": 0.7246553376160262, "grad_norm": 1.8524807691574097, "learning_rate": 3.7204195886143736e-06, "loss": 0.10712571144104004, "step": 27510 }, { "epoch": 0.7249187528605249, "grad_norm": 1.9022974967956543, "learning_rate": 3.713782357872352e-06, "loss": 0.11252042055130004, "step": 27520 }, { "epoch": 0.7251821681050237, "grad_norm": 1.9835642576217651, "learning_rate": 3.7071497025107373e-06, "loss": 0.12621251344680787, "step": 27530 }, { "epoch": 0.7254455833495224, "grad_norm": 1.7052032947540283, "learning_rate": 3.7005216273570623e-06, "loss": 0.10758278369903565, "step": 27540 }, { "epoch": 0.7257089985940212, "grad_norm": 2.597579002380371, "learning_rate": 3.693898137235532e-06, "loss": 0.10602824687957764, "step": 27550 }, { "epoch": 0.7259724138385198, "grad_norm": 2.2232394218444824, "learning_rate": 3.6872792369670074e-06, "loss": 0.11248877048492431, "step": 27560 }, { "epoch": 0.7262358290830186, "grad_norm": 1.404557704925537, "learning_rate": 3.6806649313690066e-06, "loss": 0.1133711576461792, "step": 27570 }, { "epoch": 0.7264992443275173, "grad_norm": 1.4242339134216309, "learning_rate": 3.6740552252557093e-06, "loss": 0.1188056230545044, "step": 27580 }, { "epoch": 0.7267626595720161, "grad_norm": 1.6227198839187622, "learning_rate": 3.667450123437939e-06, "loss": 0.1125436544418335, "step": 27590 }, { "epoch": 0.7270260748165148, "grad_norm": 1.9072630405426025, "learning_rate": 3.6608496307231823e-06, "loss": 0.118526291847229, "step": 27600 }, { "epoch": 0.7272894900610135, "grad_norm": 2.3706412315368652, "learning_rate": 3.6542537519155617e-06, "loss": 0.10182870626449585, "step": 27610 }, { "epoch": 0.7275529053055123, "grad_norm": 2.278379201889038, "learning_rate": 3.647662491815833e-06, "loss": 0.12256121635437012, "step": 27620 }, { "epoch": 0.727816320550011, "grad_norm": 2.675626277923584, "learning_rate": 3.641075855221409e-06, "loss": 0.12630798816680908, "step": 27630 }, { "epoch": 0.7280797357945098, "grad_norm": 1.5246511697769165, "learning_rate": 3.634493846926321e-06, "loss": 0.1224137783050537, "step": 27640 }, { "epoch": 0.7283431510390085, "grad_norm": 3.0286026000976562, "learning_rate": 3.62791647172125e-06, "loss": 0.12669923305511474, "step": 27650 }, { "epoch": 0.7286065662835073, "grad_norm": 2.118055820465088, "learning_rate": 3.621343734393482e-06, "loss": 0.11801023483276367, "step": 27660 }, { "epoch": 0.728869981528006, "grad_norm": 1.968205213546753, "learning_rate": 3.6147756397269397e-06, "loss": 0.12332457304000854, "step": 27670 }, { "epoch": 0.7291333967725048, "grad_norm": 2.719196319580078, "learning_rate": 3.6082121925021717e-06, "loss": 0.12135218381881714, "step": 27680 }, { "epoch": 0.7293968120170035, "grad_norm": 2.2485721111297607, "learning_rate": 3.601653397496335e-06, "loss": 0.1091046690940857, "step": 27690 }, { "epoch": 0.7296602272615021, "grad_norm": 1.948586344718933, "learning_rate": 3.5950992594832044e-06, "loss": 0.12350910902023315, "step": 27700 }, { "epoch": 0.729923642506001, "grad_norm": 1.7178637981414795, "learning_rate": 3.588549783233164e-06, "loss": 0.10531158447265625, "step": 27710 }, { "epoch": 0.7301870577504996, "grad_norm": 2.250720739364624, "learning_rate": 3.582004973513201e-06, "loss": 0.1128436803817749, "step": 27720 }, { "epoch": 0.7304504729949984, "grad_norm": 3.2477407455444336, "learning_rate": 3.575464835086917e-06, "loss": 0.1211483359336853, "step": 27730 }, { "epoch": 0.7307138882394971, "grad_norm": 2.99295711517334, "learning_rate": 3.568929372714508e-06, "loss": 0.13150265216827392, "step": 27740 }, { "epoch": 0.7309773034839959, "grad_norm": 3.2500460147857666, "learning_rate": 3.562398591152755e-06, "loss": 0.11531850099563598, "step": 27750 }, { "epoch": 0.7312407187284946, "grad_norm": 1.8901963233947754, "learning_rate": 3.5558724951550515e-06, "loss": 0.10648658275604247, "step": 27760 }, { "epoch": 0.7315041339729933, "grad_norm": 1.831478476524353, "learning_rate": 3.549351089471367e-06, "loss": 0.11071878671646118, "step": 27770 }, { "epoch": 0.7317675492174921, "grad_norm": 2.807220458984375, "learning_rate": 3.54283437884827e-06, "loss": 0.12118263244628906, "step": 27780 }, { "epoch": 0.7320309644619908, "grad_norm": 2.0990328788757324, "learning_rate": 3.536322368028895e-06, "loss": 0.11289765834808349, "step": 27790 }, { "epoch": 0.7322943797064896, "grad_norm": 2.314908504486084, "learning_rate": 3.5298150617529635e-06, "loss": 0.12917989492416382, "step": 27800 }, { "epoch": 0.7325577949509883, "grad_norm": 2.0253612995147705, "learning_rate": 3.5233124647567817e-06, "loss": 0.12453632354736328, "step": 27810 }, { "epoch": 0.7328212101954871, "grad_norm": 1.8231160640716553, "learning_rate": 3.516814581773216e-06, "loss": 0.1124267339706421, "step": 27820 }, { "epoch": 0.7330846254399858, "grad_norm": 2.1331138610839844, "learning_rate": 3.5103214175317056e-06, "loss": 0.11760637760162354, "step": 27830 }, { "epoch": 0.7333480406844846, "grad_norm": 1.5577598810195923, "learning_rate": 3.5038329767582567e-06, "loss": 0.11863607168197632, "step": 27840 }, { "epoch": 0.7336114559289832, "grad_norm": 2.3442904949188232, "learning_rate": 3.4973492641754315e-06, "loss": 0.11014220714569092, "step": 27850 }, { "epoch": 0.7338748711734819, "grad_norm": 2.243004322052002, "learning_rate": 3.4908702845023643e-06, "loss": 0.11050511598587036, "step": 27860 }, { "epoch": 0.7341382864179807, "grad_norm": 2.2083382606506348, "learning_rate": 3.484396042454734e-06, "loss": 0.1243295669555664, "step": 27870 }, { "epoch": 0.7344017016624794, "grad_norm": 2.244218111038208, "learning_rate": 3.477926542744766e-06, "loss": 0.11931651830673218, "step": 27880 }, { "epoch": 0.7346651169069782, "grad_norm": 2.534398317337036, "learning_rate": 3.4714617900812485e-06, "loss": 0.12131642103195191, "step": 27890 }, { "epoch": 0.7349285321514769, "grad_norm": 2.5917186737060547, "learning_rate": 3.465001789169502e-06, "loss": 0.11736705303192138, "step": 27900 }, { "epoch": 0.7351919473959757, "grad_norm": 1.9126397371292114, "learning_rate": 3.458546544711404e-06, "loss": 0.12139555215835571, "step": 27910 }, { "epoch": 0.7354553626404744, "grad_norm": 3.0155961513519287, "learning_rate": 3.45209606140535e-06, "loss": 0.1145552635192871, "step": 27920 }, { "epoch": 0.7357187778849732, "grad_norm": 2.0598807334899902, "learning_rate": 3.4456503439462797e-06, "loss": 0.10822411775588989, "step": 27930 }, { "epoch": 0.7359821931294719, "grad_norm": 2.036349296569824, "learning_rate": 3.4392093970256703e-06, "loss": 0.13367736339569092, "step": 27940 }, { "epoch": 0.7362456083739706, "grad_norm": 2.053581476211548, "learning_rate": 3.4327732253315195e-06, "loss": 0.12264009714126586, "step": 27950 }, { "epoch": 0.7365090236184694, "grad_norm": 2.1044247150421143, "learning_rate": 3.4263418335483477e-06, "loss": 0.12280154228210449, "step": 27960 }, { "epoch": 0.7367724388629681, "grad_norm": 1.7402856349945068, "learning_rate": 3.419915226357203e-06, "loss": 0.1080585241317749, "step": 27970 }, { "epoch": 0.7370358541074669, "grad_norm": 1.8868837356567383, "learning_rate": 3.413493408435642e-06, "loss": 0.11546366214752198, "step": 27980 }, { "epoch": 0.7372992693519655, "grad_norm": 2.456786632537842, "learning_rate": 3.407076384457747e-06, "loss": 0.10587148666381836, "step": 27990 }, { "epoch": 0.7375626845964643, "grad_norm": 1.9367892742156982, "learning_rate": 3.400664159094106e-06, "loss": 0.09794518947601319, "step": 28000 }, { "epoch": 0.737826099840963, "grad_norm": 2.192988872528076, "learning_rate": 3.394256737011803e-06, "loss": 0.10850362777709961, "step": 28010 }, { "epoch": 0.7380895150854618, "grad_norm": 2.085209846496582, "learning_rate": 3.3878541228744455e-06, "loss": 0.10066531896591187, "step": 28020 }, { "epoch": 0.7383529303299605, "grad_norm": 2.2153897285461426, "learning_rate": 3.38145632134213e-06, "loss": 0.10513443946838379, "step": 28030 }, { "epoch": 0.7386163455744592, "grad_norm": 2.530094623565674, "learning_rate": 3.3750633370714527e-06, "loss": 0.10987223386764526, "step": 28040 }, { "epoch": 0.738879760818958, "grad_norm": 1.7985490560531616, "learning_rate": 3.368675174715501e-06, "loss": 0.11599688529968262, "step": 28050 }, { "epoch": 0.7391431760634567, "grad_norm": 3.0821075439453125, "learning_rate": 3.3622918389238544e-06, "loss": 0.11874337196350097, "step": 28060 }, { "epoch": 0.7394065913079555, "grad_norm": 1.526536464691162, "learning_rate": 3.3559133343425865e-06, "loss": 0.12499873638153076, "step": 28070 }, { "epoch": 0.7396700065524542, "grad_norm": 2.5006885528564453, "learning_rate": 3.3495396656142442e-06, "loss": 0.10613948106765747, "step": 28080 }, { "epoch": 0.739933421796953, "grad_norm": 2.2185797691345215, "learning_rate": 3.343170837377858e-06, "loss": 0.11789040565490723, "step": 28090 }, { "epoch": 0.7401968370414517, "grad_norm": 2.1205127239227295, "learning_rate": 3.336806854268936e-06, "loss": 0.11327084302902221, "step": 28100 }, { "epoch": 0.7404602522859505, "grad_norm": 2.5446362495422363, "learning_rate": 3.3304477209194574e-06, "loss": 0.11496033668518066, "step": 28110 }, { "epoch": 0.7407236675304492, "grad_norm": 1.7289506196975708, "learning_rate": 3.3240934419578828e-06, "loss": 0.11442646980285645, "step": 28120 }, { "epoch": 0.7409870827749478, "grad_norm": 1.3397984504699707, "learning_rate": 3.3177440220091207e-06, "loss": 0.10788416862487793, "step": 28130 }, { "epoch": 0.7412504980194466, "grad_norm": 1.6654504537582397, "learning_rate": 3.3113994656945537e-06, "loss": 0.10414924621582031, "step": 28140 }, { "epoch": 0.7415139132639453, "grad_norm": 3.677963972091675, "learning_rate": 3.3050597776320293e-06, "loss": 0.10424157381057739, "step": 28150 }, { "epoch": 0.7417773285084441, "grad_norm": 3.393810272216797, "learning_rate": 3.298724962435842e-06, "loss": 0.10942113399505615, "step": 28160 }, { "epoch": 0.7420407437529428, "grad_norm": 3.4336397647857666, "learning_rate": 3.292395024716746e-06, "loss": 0.11839437484741211, "step": 28170 }, { "epoch": 0.7423041589974416, "grad_norm": 1.9884452819824219, "learning_rate": 3.2860699690819407e-06, "loss": 0.10929865837097168, "step": 28180 }, { "epoch": 0.7425675742419403, "grad_norm": 2.453486204147339, "learning_rate": 3.2797498001350737e-06, "loss": 0.10702412128448487, "step": 28190 }, { "epoch": 0.7428309894864391, "grad_norm": 2.0664947032928467, "learning_rate": 3.27343452247624e-06, "loss": 0.10897016525268555, "step": 28200 }, { "epoch": 0.7430944047309378, "grad_norm": 2.164720296859741, "learning_rate": 3.267124140701974e-06, "loss": 0.11899876594543457, "step": 28210 }, { "epoch": 0.7433578199754365, "grad_norm": 1.6056231260299683, "learning_rate": 3.2608186594052314e-06, "loss": 0.10381662845611572, "step": 28220 }, { "epoch": 0.7436212352199353, "grad_norm": 2.2105679512023926, "learning_rate": 3.2545180831754264e-06, "loss": 0.11788208484649658, "step": 28230 }, { "epoch": 0.743884650464434, "grad_norm": 2.2219417095184326, "learning_rate": 3.248222416598381e-06, "loss": 0.09927573204040527, "step": 28240 }, { "epoch": 0.7441480657089328, "grad_norm": 1.8088035583496094, "learning_rate": 3.2419316642563646e-06, "loss": 0.10469032526016235, "step": 28250 }, { "epoch": 0.7444114809534315, "grad_norm": 2.280301570892334, "learning_rate": 3.235645830728047e-06, "loss": 0.11744141578674316, "step": 28260 }, { "epoch": 0.7446748961979303, "grad_norm": 1.821919322013855, "learning_rate": 3.2293649205885293e-06, "loss": 0.10649539232254028, "step": 28270 }, { "epoch": 0.744938311442429, "grad_norm": 1.2409635782241821, "learning_rate": 3.2230889384093355e-06, "loss": 0.09943976998329163, "step": 28280 }, { "epoch": 0.7452017266869276, "grad_norm": 2.7484848499298096, "learning_rate": 3.2168178887583947e-06, "loss": 0.10585129261016846, "step": 28290 }, { "epoch": 0.7454651419314264, "grad_norm": 3.679080009460449, "learning_rate": 3.210551776200044e-06, "loss": 0.10324451923370362, "step": 28300 }, { "epoch": 0.7457285571759251, "grad_norm": 2.2537872791290283, "learning_rate": 3.204290605295033e-06, "loss": 0.12003159523010254, "step": 28310 }, { "epoch": 0.7459919724204239, "grad_norm": 2.3055381774902344, "learning_rate": 3.1980343806005077e-06, "loss": 0.11096010208129883, "step": 28320 }, { "epoch": 0.7462553876649226, "grad_norm": 2.1758503913879395, "learning_rate": 3.1917831066700255e-06, "loss": 0.1179729700088501, "step": 28330 }, { "epoch": 0.7465188029094214, "grad_norm": 2.2692935466766357, "learning_rate": 3.1855367880535336e-06, "loss": 0.11761171817779541, "step": 28340 }, { "epoch": 0.7467822181539201, "grad_norm": 2.1746551990509033, "learning_rate": 3.179295429297361e-06, "loss": 0.11871297359466552, "step": 28350 }, { "epoch": 0.7470456333984189, "grad_norm": 2.014078378677368, "learning_rate": 3.173059034944248e-06, "loss": 0.10909202098846435, "step": 28360 }, { "epoch": 0.7473090486429176, "grad_norm": 2.1482443809509277, "learning_rate": 3.1668276095333063e-06, "loss": 0.103645658493042, "step": 28370 }, { "epoch": 0.7475724638874163, "grad_norm": 1.6698490381240845, "learning_rate": 3.160601157600045e-06, "loss": 0.11248242855072021, "step": 28380 }, { "epoch": 0.7478358791319151, "grad_norm": 2.592087745666504, "learning_rate": 3.154379683676334e-06, "loss": 0.10077571868896484, "step": 28390 }, { "epoch": 0.7480992943764138, "grad_norm": 3.2662007808685303, "learning_rate": 3.148163192290432e-06, "loss": 0.11509755849838257, "step": 28400 }, { "epoch": 0.7483627096209126, "grad_norm": 2.8067641258239746, "learning_rate": 3.1419516879669744e-06, "loss": 0.1133871078491211, "step": 28410 }, { "epoch": 0.7486261248654112, "grad_norm": 1.7784823179244995, "learning_rate": 3.1357451752269595e-06, "loss": 0.11528398990631103, "step": 28420 }, { "epoch": 0.74888954010991, "grad_norm": 1.8799751996994019, "learning_rate": 3.1295436585877558e-06, "loss": 0.12388651371002198, "step": 28430 }, { "epoch": 0.7491529553544087, "grad_norm": 5.563723564147949, "learning_rate": 3.1233471425630947e-06, "loss": 0.11502344608306884, "step": 28440 }, { "epoch": 0.7494163705989075, "grad_norm": 2.370453119277954, "learning_rate": 3.1171556316630626e-06, "loss": 0.10498316287994384, "step": 28450 }, { "epoch": 0.7496797858434062, "grad_norm": 3.140679359436035, "learning_rate": 3.110969130394117e-06, "loss": 0.09898802042007446, "step": 28460 }, { "epoch": 0.7499432010879049, "grad_norm": 2.3055944442749023, "learning_rate": 3.1047876432590596e-06, "loss": 0.12175242900848389, "step": 28470 }, { "epoch": 0.7502066163324037, "grad_norm": 2.562967300415039, "learning_rate": 3.0986111747570335e-06, "loss": 0.11032843589782715, "step": 28480 }, { "epoch": 0.7504700315769024, "grad_norm": 1.9785845279693604, "learning_rate": 3.092439729383548e-06, "loss": 0.11814179420471191, "step": 28490 }, { "epoch": 0.7507334468214012, "grad_norm": 1.9586827754974365, "learning_rate": 3.0862733116304435e-06, "loss": 0.10442861318588256, "step": 28500 }, { "epoch": 0.7509968620658999, "grad_norm": 2.2788727283477783, "learning_rate": 3.0801119259859115e-06, "loss": 0.12541611194610597, "step": 28510 }, { "epoch": 0.7512602773103987, "grad_norm": 3.058168888092041, "learning_rate": 3.0739555769344666e-06, "loss": 0.09994184374809265, "step": 28520 }, { "epoch": 0.7515236925548974, "grad_norm": 2.262439012527466, "learning_rate": 3.067804268956964e-06, "loss": 0.11877281665802002, "step": 28530 }, { "epoch": 0.7517871077993962, "grad_norm": 2.5879604816436768, "learning_rate": 3.0616580065305977e-06, "loss": 0.10679495334625244, "step": 28540 }, { "epoch": 0.7520505230438949, "grad_norm": 1.7994904518127441, "learning_rate": 3.055516794128881e-06, "loss": 0.1144054651260376, "step": 28550 }, { "epoch": 0.7523139382883935, "grad_norm": 2.734403133392334, "learning_rate": 3.0493806362216516e-06, "loss": 0.10753995180130005, "step": 28560 }, { "epoch": 0.7525773535328923, "grad_norm": 2.265207052230835, "learning_rate": 3.0432495372750692e-06, "loss": 0.11054960489273072, "step": 28570 }, { "epoch": 0.752840768777391, "grad_norm": 2.9545016288757324, "learning_rate": 3.037123501751611e-06, "loss": 0.10941143035888672, "step": 28580 }, { "epoch": 0.7531041840218898, "grad_norm": 3.328138828277588, "learning_rate": 3.0310025341100757e-06, "loss": 0.11379318237304688, "step": 28590 }, { "epoch": 0.7533675992663885, "grad_norm": 2.367344379425049, "learning_rate": 3.0248866388055674e-06, "loss": 0.11472923755645752, "step": 28600 }, { "epoch": 0.7536310145108873, "grad_norm": 1.8372243642807007, "learning_rate": 3.018775820289488e-06, "loss": 0.0897216558456421, "step": 28610 }, { "epoch": 0.753894429755386, "grad_norm": 3.060023069381714, "learning_rate": 3.0126700830095668e-06, "loss": 0.11148929595947266, "step": 28620 }, { "epoch": 0.7541578449998848, "grad_norm": 3.1776576042175293, "learning_rate": 3.006569431409817e-06, "loss": 0.10884199142456055, "step": 28630 }, { "epoch": 0.7544212602443835, "grad_norm": 3.078367233276367, "learning_rate": 3.000473869930558e-06, "loss": 0.10646247863769531, "step": 28640 }, { "epoch": 0.7546846754888822, "grad_norm": 2.112919569015503, "learning_rate": 2.9943834030084006e-06, "loss": 0.12011075019836426, "step": 28650 }, { "epoch": 0.754948090733381, "grad_norm": 2.2859175205230713, "learning_rate": 2.988298035076249e-06, "loss": 0.10654613971710206, "step": 28660 }, { "epoch": 0.7552115059778797, "grad_norm": 2.453521966934204, "learning_rate": 2.9822177705633005e-06, "loss": 0.10952181816101074, "step": 28670 }, { "epoch": 0.7554749212223785, "grad_norm": 2.651029586791992, "learning_rate": 2.976142613895031e-06, "loss": 0.1030811071395874, "step": 28680 }, { "epoch": 0.7557383364668772, "grad_norm": 2.2515978813171387, "learning_rate": 2.970072569493202e-06, "loss": 0.10310401916503906, "step": 28690 }, { "epoch": 0.756001751711376, "grad_norm": 2.8322718143463135, "learning_rate": 2.9640076417758546e-06, "loss": 0.1129384994506836, "step": 28700 }, { "epoch": 0.7562651669558746, "grad_norm": 2.5777883529663086, "learning_rate": 2.9579478351573e-06, "loss": 0.10470199584960938, "step": 28710 }, { "epoch": 0.7565285822003734, "grad_norm": 2.7050254344940186, "learning_rate": 2.9518931540481377e-06, "loss": 0.117061185836792, "step": 28720 }, { "epoch": 0.7567919974448721, "grad_norm": 1.4342206716537476, "learning_rate": 2.945843602855215e-06, "loss": 0.11578727960586548, "step": 28730 }, { "epoch": 0.7570554126893708, "grad_norm": 2.233332633972168, "learning_rate": 2.9397991859816588e-06, "loss": 0.11420664787292481, "step": 28740 }, { "epoch": 0.7573188279338696, "grad_norm": 3.0826869010925293, "learning_rate": 2.9337599078268597e-06, "loss": 0.10941345691680908, "step": 28750 }, { "epoch": 0.7575822431783683, "grad_norm": 2.0440633296966553, "learning_rate": 2.927725772786464e-06, "loss": 0.1157524824142456, "step": 28760 }, { "epoch": 0.7578456584228671, "grad_norm": 1.4750139713287354, "learning_rate": 2.9216967852523747e-06, "loss": 0.10182628631591797, "step": 28770 }, { "epoch": 0.7581090736673658, "grad_norm": 2.6155569553375244, "learning_rate": 2.915672949612749e-06, "loss": 0.10501797199249267, "step": 28780 }, { "epoch": 0.7583724889118646, "grad_norm": 1.551931619644165, "learning_rate": 2.9096542702519904e-06, "loss": 0.10752742290496826, "step": 28790 }, { "epoch": 0.7586359041563633, "grad_norm": 1.9885140657424927, "learning_rate": 2.9036407515507613e-06, "loss": 0.1070057511329651, "step": 28800 }, { "epoch": 0.758899319400862, "grad_norm": 1.8082555532455444, "learning_rate": 2.89763239788596e-06, "loss": 0.10740631818771362, "step": 28810 }, { "epoch": 0.7591627346453608, "grad_norm": 3.335322856903076, "learning_rate": 2.8916292136307147e-06, "loss": 0.11901791095733642, "step": 28820 }, { "epoch": 0.7594261498898595, "grad_norm": 1.8789993524551392, "learning_rate": 2.885631203154412e-06, "loss": 0.11852951049804687, "step": 28830 }, { "epoch": 0.7596895651343583, "grad_norm": 2.2818901538848877, "learning_rate": 2.879638370822656e-06, "loss": 0.10937255620956421, "step": 28840 }, { "epoch": 0.759952980378857, "grad_norm": 1.8024121522903442, "learning_rate": 2.8736507209972996e-06, "loss": 0.10154335498809815, "step": 28850 }, { "epoch": 0.7602163956233557, "grad_norm": 2.4638307094573975, "learning_rate": 2.867668258036401e-06, "loss": 0.11236562728881835, "step": 28860 }, { "epoch": 0.7604798108678544, "grad_norm": 1.8957860469818115, "learning_rate": 2.8616909862942565e-06, "loss": 0.10292593240737916, "step": 28870 }, { "epoch": 0.7607432261123532, "grad_norm": 1.9937160015106201, "learning_rate": 2.855718910121389e-06, "loss": 0.10082902908325195, "step": 28880 }, { "epoch": 0.7610066413568519, "grad_norm": 2.0824923515319824, "learning_rate": 2.8497520338645278e-06, "loss": 0.09269967079162597, "step": 28890 }, { "epoch": 0.7612700566013506, "grad_norm": 1.9078203439712524, "learning_rate": 2.843790361866625e-06, "loss": 0.10159785747528076, "step": 28900 }, { "epoch": 0.7615334718458494, "grad_norm": 1.8565196990966797, "learning_rate": 2.837833898466842e-06, "loss": 0.10365339517593383, "step": 28910 }, { "epoch": 0.7617968870903481, "grad_norm": 1.917731523513794, "learning_rate": 2.831882648000547e-06, "loss": 0.09631645679473877, "step": 28920 }, { "epoch": 0.7620603023348469, "grad_norm": 2.4358203411102295, "learning_rate": 2.8259366147993217e-06, "loss": 0.10645520687103271, "step": 28930 }, { "epoch": 0.7623237175793456, "grad_norm": 2.117281675338745, "learning_rate": 2.819995803190948e-06, "loss": 0.11624157428741455, "step": 28940 }, { "epoch": 0.7625871328238444, "grad_norm": 2.9173381328582764, "learning_rate": 2.8140602174993927e-06, "loss": 0.1044117569923401, "step": 28950 }, { "epoch": 0.7628505480683431, "grad_norm": 2.432147979736328, "learning_rate": 2.8081298620448427e-06, "loss": 0.10331262350082397, "step": 28960 }, { "epoch": 0.7631139633128419, "grad_norm": 1.8033791780471802, "learning_rate": 2.8022047411436604e-06, "loss": 0.09767295718193054, "step": 28970 }, { "epoch": 0.7633773785573406, "grad_norm": 2.8374624252319336, "learning_rate": 2.7962848591084122e-06, "loss": 0.11555124521255493, "step": 28980 }, { "epoch": 0.7636407938018392, "grad_norm": 2.569871187210083, "learning_rate": 2.7903702202478355e-06, "loss": 0.10605533123016357, "step": 28990 }, { "epoch": 0.763904209046338, "grad_norm": 3.6279850006103516, "learning_rate": 2.7844608288668586e-06, "loss": 0.10336604118347167, "step": 29000 }, { "epoch": 0.7641676242908367, "grad_norm": 2.294569730758667, "learning_rate": 2.7785566892665993e-06, "loss": 0.11322555541992188, "step": 29010 }, { "epoch": 0.7644310395353355, "grad_norm": 2.633347749710083, "learning_rate": 2.77265780574434e-06, "loss": 0.11894333362579346, "step": 29020 }, { "epoch": 0.7646944547798342, "grad_norm": 2.2696759700775146, "learning_rate": 2.766764182593544e-06, "loss": 0.11281648874282837, "step": 29030 }, { "epoch": 0.764957870024333, "grad_norm": 1.9161415100097656, "learning_rate": 2.760875824103846e-06, "loss": 0.10670158863067628, "step": 29040 }, { "epoch": 0.7652212852688317, "grad_norm": 2.354365587234497, "learning_rate": 2.7549927345610407e-06, "loss": 0.09990816712379455, "step": 29050 }, { "epoch": 0.7654847005133305, "grad_norm": 2.5430383682250977, "learning_rate": 2.749114918247102e-06, "loss": 0.10768753290176392, "step": 29060 }, { "epoch": 0.7657481157578292, "grad_norm": 1.7969497442245483, "learning_rate": 2.7432423794401598e-06, "loss": 0.09662961959838867, "step": 29070 }, { "epoch": 0.7660115310023279, "grad_norm": 2.1538186073303223, "learning_rate": 2.7373751224144905e-06, "loss": 0.10527024269104004, "step": 29080 }, { "epoch": 0.7662749462468267, "grad_norm": 2.3817179203033447, "learning_rate": 2.731513151440547e-06, "loss": 0.10490823984146118, "step": 29090 }, { "epoch": 0.7665383614913254, "grad_norm": 1.5772098302841187, "learning_rate": 2.7256564707849165e-06, "loss": 0.10078487396240235, "step": 29100 }, { "epoch": 0.7668017767358242, "grad_norm": 3.236734390258789, "learning_rate": 2.719805084710356e-06, "loss": 0.11921880245208741, "step": 29110 }, { "epoch": 0.7670651919803229, "grad_norm": 2.253817558288574, "learning_rate": 2.7139589974757454e-06, "loss": 0.10074288845062256, "step": 29120 }, { "epoch": 0.7673286072248217, "grad_norm": 2.322199583053589, "learning_rate": 2.7081182133361207e-06, "loss": 0.10466890335083008, "step": 29130 }, { "epoch": 0.7675920224693203, "grad_norm": 2.06842041015625, "learning_rate": 2.7022827365426606e-06, "loss": 0.1130098819732666, "step": 29140 }, { "epoch": 0.7678554377138191, "grad_norm": 2.103964328765869, "learning_rate": 2.696452571342676e-06, "loss": 0.1065864086151123, "step": 29150 }, { "epoch": 0.7681188529583178, "grad_norm": 1.91415274143219, "learning_rate": 2.6906277219796116e-06, "loss": 0.08992280960083007, "step": 29160 }, { "epoch": 0.7683822682028165, "grad_norm": 2.256945848464966, "learning_rate": 2.6848081926930447e-06, "loss": 0.08926403522491455, "step": 29170 }, { "epoch": 0.7686456834473153, "grad_norm": 2.393296003341675, "learning_rate": 2.6789939877186753e-06, "loss": 0.10914722681045533, "step": 29180 }, { "epoch": 0.768909098691814, "grad_norm": 2.5433509349823, "learning_rate": 2.67318511128834e-06, "loss": 0.10067393779754638, "step": 29190 }, { "epoch": 0.7691725139363128, "grad_norm": 2.426996946334839, "learning_rate": 2.66738156762999e-06, "loss": 0.09349538087844848, "step": 29200 }, { "epoch": 0.7694359291808115, "grad_norm": 2.9691498279571533, "learning_rate": 2.6615833609676846e-06, "loss": 0.10951046943664551, "step": 29210 }, { "epoch": 0.7696993444253103, "grad_norm": 2.2311081886291504, "learning_rate": 2.6557904955216185e-06, "loss": 0.10442800521850586, "step": 29220 }, { "epoch": 0.769962759669809, "grad_norm": 1.4801541566848755, "learning_rate": 2.650002975508086e-06, "loss": 0.10166184902191162, "step": 29230 }, { "epoch": 0.7702261749143078, "grad_norm": 2.7999072074890137, "learning_rate": 2.6442208051394936e-06, "loss": 0.09818661212921143, "step": 29240 }, { "epoch": 0.7704895901588065, "grad_norm": 2.1366374492645264, "learning_rate": 2.638443988624355e-06, "loss": 0.09817240238189698, "step": 29250 }, { "epoch": 0.7707530054033052, "grad_norm": 2.466740131378174, "learning_rate": 2.632672530167283e-06, "loss": 0.09027492403984069, "step": 29260 }, { "epoch": 0.771016420647804, "grad_norm": 2.1648755073547363, "learning_rate": 2.6269064339689986e-06, "loss": 0.09660470485687256, "step": 29270 }, { "epoch": 0.7712798358923026, "grad_norm": 1.85687255859375, "learning_rate": 2.6211457042263157e-06, "loss": 0.10579928159713745, "step": 29280 }, { "epoch": 0.7715432511368014, "grad_norm": 1.6613315343856812, "learning_rate": 2.6153903451321384e-06, "loss": 0.09909712076187134, "step": 29290 }, { "epoch": 0.7718066663813001, "grad_norm": 2.6907761096954346, "learning_rate": 2.6096403608754685e-06, "loss": 0.09743176102638244, "step": 29300 }, { "epoch": 0.7720700816257989, "grad_norm": 2.334192991256714, "learning_rate": 2.603895755641387e-06, "loss": 0.11983089447021485, "step": 29310 }, { "epoch": 0.7723334968702976, "grad_norm": 1.7232203483581543, "learning_rate": 2.598156533611077e-06, "loss": 0.11296803951263427, "step": 29320 }, { "epoch": 0.7725969121147963, "grad_norm": 3.0892560482025146, "learning_rate": 2.592422698961781e-06, "loss": 0.09281902313232422, "step": 29330 }, { "epoch": 0.7728603273592951, "grad_norm": 2.2857203483581543, "learning_rate": 2.586694255866831e-06, "loss": 0.11521414518356324, "step": 29340 }, { "epoch": 0.7731237426037938, "grad_norm": 2.3574721813201904, "learning_rate": 2.580971208495642e-06, "loss": 0.0984583854675293, "step": 29350 }, { "epoch": 0.7733871578482926, "grad_norm": 2.548250436782837, "learning_rate": 2.575253561013692e-06, "loss": 0.10493714809417724, "step": 29360 }, { "epoch": 0.7736505730927913, "grad_norm": 2.0756683349609375, "learning_rate": 2.56954131758253e-06, "loss": 0.09304795265197754, "step": 29370 }, { "epoch": 0.7739139883372901, "grad_norm": 3.397902250289917, "learning_rate": 2.5638344823597762e-06, "loss": 0.10647079944610596, "step": 29380 }, { "epoch": 0.7741774035817888, "grad_norm": 2.1572113037109375, "learning_rate": 2.558133059499105e-06, "loss": 0.10502135753631592, "step": 29390 }, { "epoch": 0.7744408188262876, "grad_norm": 3.4421093463897705, "learning_rate": 2.552437053150266e-06, "loss": 0.09973953366279602, "step": 29400 }, { "epoch": 0.7747042340707863, "grad_norm": 1.8648494482040405, "learning_rate": 2.546746467459058e-06, "loss": 0.1004898190498352, "step": 29410 }, { "epoch": 0.774967649315285, "grad_norm": 3.152540445327759, "learning_rate": 2.541061306567325e-06, "loss": 0.11221575736999512, "step": 29420 }, { "epoch": 0.7752310645597837, "grad_norm": 2.276336669921875, "learning_rate": 2.5353815746129827e-06, "loss": 0.10240216255187988, "step": 29430 }, { "epoch": 0.7754944798042824, "grad_norm": 3.0723798274993896, "learning_rate": 2.529707275729979e-06, "loss": 0.10296146869659424, "step": 29440 }, { "epoch": 0.7757578950487812, "grad_norm": 2.5932273864746094, "learning_rate": 2.524038414048321e-06, "loss": 0.11459414958953858, "step": 29450 }, { "epoch": 0.7760213102932799, "grad_norm": 2.483304023742676, "learning_rate": 2.518374993694043e-06, "loss": 0.09289824962615967, "step": 29460 }, { "epoch": 0.7762847255377787, "grad_norm": 2.1038191318511963, "learning_rate": 2.5127170187892257e-06, "loss": 0.10543749332427979, "step": 29470 }, { "epoch": 0.7765481407822774, "grad_norm": 3.356661558151245, "learning_rate": 2.507064493451995e-06, "loss": 0.09911500215530396, "step": 29480 }, { "epoch": 0.7768115560267762, "grad_norm": 2.2051491737365723, "learning_rate": 2.501417421796498e-06, "loss": 0.10474684238433837, "step": 29490 }, { "epoch": 0.7770749712712749, "grad_norm": 2.3848066329956055, "learning_rate": 2.4957758079329173e-06, "loss": 0.11320087909698487, "step": 29500 }, { "epoch": 0.7773383865157736, "grad_norm": 2.3427114486694336, "learning_rate": 2.490139655967465e-06, "loss": 0.10522465705871582, "step": 29510 }, { "epoch": 0.7776018017602724, "grad_norm": 1.728965401649475, "learning_rate": 2.4845089700023694e-06, "loss": 0.10859668254852295, "step": 29520 }, { "epoch": 0.7778652170047711, "grad_norm": 1.8779040575027466, "learning_rate": 2.4788837541358957e-06, "loss": 0.10251134634017944, "step": 29530 }, { "epoch": 0.7781286322492699, "grad_norm": 2.5605504512786865, "learning_rate": 2.4732640124623152e-06, "loss": 0.10952075719833373, "step": 29540 }, { "epoch": 0.7783920474937686, "grad_norm": 2.1050984859466553, "learning_rate": 2.467649749071913e-06, "loss": 0.1006511926651001, "step": 29550 }, { "epoch": 0.7786554627382674, "grad_norm": 2.4800891876220703, "learning_rate": 2.4620409680509993e-06, "loss": 0.10938359498977661, "step": 29560 }, { "epoch": 0.778918877982766, "grad_norm": 2.0667474269866943, "learning_rate": 2.4564376734818818e-06, "loss": 0.11585335731506348, "step": 29570 }, { "epoch": 0.7791822932272648, "grad_norm": 1.6377058029174805, "learning_rate": 2.450839869442887e-06, "loss": 0.09669789671897888, "step": 29580 }, { "epoch": 0.7794457084717635, "grad_norm": 3.1848881244659424, "learning_rate": 2.4452475600083323e-06, "loss": 0.09341498613357543, "step": 29590 }, { "epoch": 0.7797091237162622, "grad_norm": 2.4510300159454346, "learning_rate": 2.4396607492485393e-06, "loss": 0.10355596542358399, "step": 29600 }, { "epoch": 0.779972538960761, "grad_norm": 2.253654718399048, "learning_rate": 2.4340794412298364e-06, "loss": 0.10747592449188233, "step": 29610 }, { "epoch": 0.7802359542052597, "grad_norm": 2.5044219493865967, "learning_rate": 2.428503640014539e-06, "loss": 0.10158523321151733, "step": 29620 }, { "epoch": 0.7804993694497585, "grad_norm": 2.347909688949585, "learning_rate": 2.4229333496609532e-06, "loss": 0.1008861780166626, "step": 29630 }, { "epoch": 0.7807627846942572, "grad_norm": 2.8540687561035156, "learning_rate": 2.4173685742233787e-06, "loss": 0.10130012035369873, "step": 29640 }, { "epoch": 0.781026199938756, "grad_norm": 2.638204574584961, "learning_rate": 2.411809317752095e-06, "loss": 0.09844475984573364, "step": 29650 }, { "epoch": 0.7812896151832547, "grad_norm": 2.2879343032836914, "learning_rate": 2.4062555842933753e-06, "loss": 0.09960451126098632, "step": 29660 }, { "epoch": 0.7815530304277535, "grad_norm": 2.1119816303253174, "learning_rate": 2.4007073778894665e-06, "loss": 0.10135726928710938, "step": 29670 }, { "epoch": 0.7818164456722522, "grad_norm": 1.8421070575714111, "learning_rate": 2.395164702578583e-06, "loss": 0.09862233400344848, "step": 29680 }, { "epoch": 0.7820798609167509, "grad_norm": 3.1185109615325928, "learning_rate": 2.3896275623949327e-06, "loss": 0.10991867780685424, "step": 29690 }, { "epoch": 0.7823432761612497, "grad_norm": 2.7894489765167236, "learning_rate": 2.38409596136868e-06, "loss": 0.10749517679214478, "step": 29700 }, { "epoch": 0.7826066914057483, "grad_norm": 2.1750597953796387, "learning_rate": 2.378569903525969e-06, "loss": 0.10405347347259522, "step": 29710 }, { "epoch": 0.7828701066502471, "grad_norm": 3.245936632156372, "learning_rate": 2.3730493928888963e-06, "loss": 0.11765267848968505, "step": 29720 }, { "epoch": 0.7831335218947458, "grad_norm": 2.837980270385742, "learning_rate": 2.367534433475527e-06, "loss": 0.10647499561309814, "step": 29730 }, { "epoch": 0.7833969371392446, "grad_norm": 2.2806596755981445, "learning_rate": 2.3620250292998925e-06, "loss": 0.10911874771118164, "step": 29740 }, { "epoch": 0.7836603523837433, "grad_norm": 1.945346474647522, "learning_rate": 2.356521184371973e-06, "loss": 0.10843048095703126, "step": 29750 }, { "epoch": 0.7839237676282421, "grad_norm": 1.867738962173462, "learning_rate": 2.3510229026977018e-06, "loss": 0.10736335515975952, "step": 29760 }, { "epoch": 0.7841871828727408, "grad_norm": 2.578725814819336, "learning_rate": 2.345530188278967e-06, "loss": 0.10529301166534424, "step": 29770 }, { "epoch": 0.7844505981172395, "grad_norm": 1.3504459857940674, "learning_rate": 2.3400430451136e-06, "loss": 0.09962930679321289, "step": 29780 }, { "epoch": 0.7847140133617383, "grad_norm": 4.058687686920166, "learning_rate": 2.334561477195387e-06, "loss": 0.09443398714065551, "step": 29790 }, { "epoch": 0.784977428606237, "grad_norm": 2.028885841369629, "learning_rate": 2.329085488514049e-06, "loss": 0.09641462564468384, "step": 29800 }, { "epoch": 0.7852408438507358, "grad_norm": 2.2670085430145264, "learning_rate": 2.3236150830552374e-06, "loss": 0.09186230897903443, "step": 29810 }, { "epoch": 0.7855042590952345, "grad_norm": 3.04617977142334, "learning_rate": 2.3181502648005593e-06, "loss": 0.1032490849494934, "step": 29820 }, { "epoch": 0.7857676743397333, "grad_norm": 2.629058599472046, "learning_rate": 2.3126910377275415e-06, "loss": 0.09937612414360046, "step": 29830 }, { "epoch": 0.786031089584232, "grad_norm": 3.1131460666656494, "learning_rate": 2.3072374058096447e-06, "loss": 0.11701743602752686, "step": 29840 }, { "epoch": 0.7862945048287308, "grad_norm": 2.0987603664398193, "learning_rate": 2.301789373016259e-06, "loss": 0.0995549738407135, "step": 29850 }, { "epoch": 0.7865579200732294, "grad_norm": 2.7862391471862793, "learning_rate": 2.296346943312693e-06, "loss": 0.09856040477752685, "step": 29860 }, { "epoch": 0.7868213353177281, "grad_norm": 2.9672324657440186, "learning_rate": 2.290910120660188e-06, "loss": 0.10205627679824829, "step": 29870 }, { "epoch": 0.7870847505622269, "grad_norm": 2.364295482635498, "learning_rate": 2.285478909015896e-06, "loss": 0.10159573554992676, "step": 29880 }, { "epoch": 0.7873481658067256, "grad_norm": 2.511728525161743, "learning_rate": 2.2800533123328874e-06, "loss": 0.11940038204193115, "step": 29890 }, { "epoch": 0.7876115810512244, "grad_norm": 2.583460569381714, "learning_rate": 2.2746333345601435e-06, "loss": 0.10552797317504883, "step": 29900 }, { "epoch": 0.7878749962957231, "grad_norm": 1.9071189165115356, "learning_rate": 2.269218979642558e-06, "loss": 0.09247616529464722, "step": 29910 }, { "epoch": 0.7881384115402219, "grad_norm": 2.445531129837036, "learning_rate": 2.2638102515209382e-06, "loss": 0.10525949001312256, "step": 29920 }, { "epoch": 0.7884018267847206, "grad_norm": 3.044832706451416, "learning_rate": 2.258407154131982e-06, "loss": 0.09553251266479493, "step": 29930 }, { "epoch": 0.7886652420292193, "grad_norm": 2.8922007083892822, "learning_rate": 2.2530096914082965e-06, "loss": 0.11348795890808105, "step": 29940 }, { "epoch": 0.7889286572737181, "grad_norm": 2.296851634979248, "learning_rate": 2.2476178672783943e-06, "loss": 0.09607595205307007, "step": 29950 }, { "epoch": 0.7891920725182168, "grad_norm": 2.4066855907440186, "learning_rate": 2.242231685666675e-06, "loss": 0.09603742361068726, "step": 29960 }, { "epoch": 0.7894554877627156, "grad_norm": 3.042358875274658, "learning_rate": 2.2368511504934333e-06, "loss": 0.10509712696075439, "step": 29970 }, { "epoch": 0.7897189030072143, "grad_norm": 2.1364428997039795, "learning_rate": 2.2314762656748544e-06, "loss": 0.09688753485679627, "step": 29980 }, { "epoch": 0.7899823182517131, "grad_norm": 1.6757506132125854, "learning_rate": 2.2261070351230107e-06, "loss": 0.09571580290794372, "step": 29990 }, { "epoch": 0.7902457334962117, "grad_norm": 2.4869821071624756, "learning_rate": 2.2207434627458626e-06, "loss": 0.10761899948120117, "step": 30000 }, { "epoch": 0.7905091487407105, "grad_norm": 2.857881784439087, "learning_rate": 2.2153855524472513e-06, "loss": 0.10840505361557007, "step": 30010 }, { "epoch": 0.7907725639852092, "grad_norm": 1.5608867406845093, "learning_rate": 2.2100333081268853e-06, "loss": 0.1013002872467041, "step": 30020 }, { "epoch": 0.7910359792297079, "grad_norm": 2.291374444961548, "learning_rate": 2.204686733680369e-06, "loss": 0.10108568668365478, "step": 30030 }, { "epoch": 0.7912993944742067, "grad_norm": 2.862267255783081, "learning_rate": 2.1993458329991635e-06, "loss": 0.09866518378257752, "step": 30040 }, { "epoch": 0.7915628097187054, "grad_norm": 2.087688446044922, "learning_rate": 2.1940106099706137e-06, "loss": 0.10358455181121826, "step": 30050 }, { "epoch": 0.7918262249632042, "grad_norm": 1.8563437461853027, "learning_rate": 2.1886810684779204e-06, "loss": 0.09823676943778992, "step": 30060 }, { "epoch": 0.7920896402077029, "grad_norm": 2.31445050239563, "learning_rate": 2.18335721240015e-06, "loss": 0.10432848930358887, "step": 30070 }, { "epoch": 0.7923530554522017, "grad_norm": 2.446629762649536, "learning_rate": 2.1780390456122424e-06, "loss": 0.08685063123703003, "step": 30080 }, { "epoch": 0.7926164706967004, "grad_norm": 2.573331594467163, "learning_rate": 2.1727265719849853e-06, "loss": 0.10624498128890991, "step": 30090 }, { "epoch": 0.7928798859411992, "grad_norm": 2.0450994968414307, "learning_rate": 2.1674197953850274e-06, "loss": 0.09271222949028016, "step": 30100 }, { "epoch": 0.7931433011856979, "grad_norm": 2.7545065879821777, "learning_rate": 2.162118719674867e-06, "loss": 0.10201321840286255, "step": 30110 }, { "epoch": 0.7934067164301966, "grad_norm": 3.335209846496582, "learning_rate": 2.1568233487128556e-06, "loss": 0.09800377488136292, "step": 30120 }, { "epoch": 0.7936701316746954, "grad_norm": 3.284773588180542, "learning_rate": 2.1515336863531976e-06, "loss": 0.10626838207244874, "step": 30130 }, { "epoch": 0.793933546919194, "grad_norm": 3.0645370483398438, "learning_rate": 2.1462497364459356e-06, "loss": 0.10171151161193848, "step": 30140 }, { "epoch": 0.7941969621636928, "grad_norm": 2.5944314002990723, "learning_rate": 2.1409715028369504e-06, "loss": 0.09841280579566955, "step": 30150 }, { "epoch": 0.7944603774081915, "grad_norm": 2.392913579940796, "learning_rate": 2.135698989367976e-06, "loss": 0.10815870761871338, "step": 30160 }, { "epoch": 0.7947237926526903, "grad_norm": 2.8018341064453125, "learning_rate": 2.1304321998765685e-06, "loss": 0.09201432466506958, "step": 30170 }, { "epoch": 0.794987207897189, "grad_norm": 2.254417896270752, "learning_rate": 2.1251711381961336e-06, "loss": 0.10851731300354003, "step": 30180 }, { "epoch": 0.7952506231416878, "grad_norm": 2.9403531551361084, "learning_rate": 2.11991580815589e-06, "loss": 0.10299762487411498, "step": 30190 }, { "epoch": 0.7955140383861865, "grad_norm": 2.574395179748535, "learning_rate": 2.1146662135808948e-06, "loss": 0.10639439821243286, "step": 30200 }, { "epoch": 0.7957774536306852, "grad_norm": 2.6679325103759766, "learning_rate": 2.1094223582920346e-06, "loss": 0.10151200294494629, "step": 30210 }, { "epoch": 0.796040868875184, "grad_norm": 3.0083439350128174, "learning_rate": 2.1041842461060103e-06, "loss": 0.09836913347244262, "step": 30220 }, { "epoch": 0.7963042841196827, "grad_norm": 2.050929307937622, "learning_rate": 2.098951880835346e-06, "loss": 0.10264995098114013, "step": 30230 }, { "epoch": 0.7965676993641815, "grad_norm": 2.4675402641296387, "learning_rate": 2.093725266288382e-06, "loss": 0.10289419889450073, "step": 30240 }, { "epoch": 0.7968311146086802, "grad_norm": 3.122805118560791, "learning_rate": 2.088504406269274e-06, "loss": 0.10898675918579101, "step": 30250 }, { "epoch": 0.797094529853179, "grad_norm": 2.586874485015869, "learning_rate": 2.083289304577992e-06, "loss": 0.09371882677078247, "step": 30260 }, { "epoch": 0.7973579450976777, "grad_norm": 2.638899803161621, "learning_rate": 2.0780799650103123e-06, "loss": 0.09408811926841736, "step": 30270 }, { "epoch": 0.7976213603421765, "grad_norm": 3.344984531402588, "learning_rate": 2.0728763913578097e-06, "loss": 0.10041320323944092, "step": 30280 }, { "epoch": 0.7978847755866751, "grad_norm": 2.6091227531433105, "learning_rate": 2.0676785874078773e-06, "loss": 0.09842783212661743, "step": 30290 }, { "epoch": 0.7981481908311738, "grad_norm": 1.9696067571640015, "learning_rate": 2.062486556943699e-06, "loss": 0.09739567637443543, "step": 30300 }, { "epoch": 0.7984116060756726, "grad_norm": 2.2240428924560547, "learning_rate": 2.0573003037442575e-06, "loss": 0.10173752307891845, "step": 30310 }, { "epoch": 0.7986750213201713, "grad_norm": 2.4756503105163574, "learning_rate": 2.0521198315843315e-06, "loss": 0.0985413908958435, "step": 30320 }, { "epoch": 0.7989384365646701, "grad_norm": 2.6751832962036133, "learning_rate": 2.046945144234491e-06, "loss": 0.10618581771850585, "step": 30330 }, { "epoch": 0.7992018518091688, "grad_norm": 2.0327963829040527, "learning_rate": 2.0417762454611012e-06, "loss": 0.10950655937194824, "step": 30340 }, { "epoch": 0.7994652670536676, "grad_norm": 2.4006338119506836, "learning_rate": 2.0366131390263055e-06, "loss": 0.09871245622634887, "step": 30350 }, { "epoch": 0.7997286822981663, "grad_norm": 2.297597885131836, "learning_rate": 2.0314558286880383e-06, "loss": 0.09787940382957458, "step": 30360 }, { "epoch": 0.7999920975426651, "grad_norm": 3.538442611694336, "learning_rate": 2.026304318200011e-06, "loss": 0.10412187576293945, "step": 30370 }, { "epoch": 0.8002555127871638, "grad_norm": 2.3468470573425293, "learning_rate": 2.021158611311712e-06, "loss": 0.09965409636497498, "step": 30380 }, { "epoch": 0.8005189280316625, "grad_norm": 3.0809741020202637, "learning_rate": 2.016018711768416e-06, "loss": 0.10355597734451294, "step": 30390 }, { "epoch": 0.8007823432761613, "grad_norm": 2.352421283721924, "learning_rate": 2.010884623311162e-06, "loss": 0.1107520341873169, "step": 30400 }, { "epoch": 0.80104575852066, "grad_norm": 3.0261306762695312, "learning_rate": 2.005756349676753e-06, "loss": 0.10732653141021728, "step": 30410 }, { "epoch": 0.8013091737651588, "grad_norm": 3.0863161087036133, "learning_rate": 2.0006338945977777e-06, "loss": 0.0960336446762085, "step": 30420 }, { "epoch": 0.8015725890096574, "grad_norm": 2.843794822692871, "learning_rate": 1.9955172618025743e-06, "loss": 0.0935747742652893, "step": 30430 }, { "epoch": 0.8018360042541562, "grad_norm": 2.3931283950805664, "learning_rate": 1.9904064550152515e-06, "loss": 0.08653953075408935, "step": 30440 }, { "epoch": 0.8020994194986549, "grad_norm": 1.9594041109085083, "learning_rate": 1.9853014779556734e-06, "loss": 0.09595672488212585, "step": 30450 }, { "epoch": 0.8023628347431536, "grad_norm": 2.4230663776397705, "learning_rate": 1.9802023343394606e-06, "loss": 0.09608579874038696, "step": 30460 }, { "epoch": 0.8026262499876524, "grad_norm": 3.179110050201416, "learning_rate": 1.975109027877995e-06, "loss": 0.10201476812362671, "step": 30470 }, { "epoch": 0.8028896652321511, "grad_norm": 3.5405194759368896, "learning_rate": 1.9700215622784026e-06, "loss": 0.08601321578025818, "step": 30480 }, { "epoch": 0.8031530804766499, "grad_norm": 3.8413710594177246, "learning_rate": 1.96493994124356e-06, "loss": 0.10111880302429199, "step": 30490 }, { "epoch": 0.8034164957211486, "grad_norm": 2.395873785018921, "learning_rate": 1.9598641684720897e-06, "loss": 0.08863002061843872, "step": 30500 }, { "epoch": 0.8036799109656474, "grad_norm": 2.5068862438201904, "learning_rate": 1.954794247658356e-06, "loss": 0.10834617614746093, "step": 30510 }, { "epoch": 0.8039433262101461, "grad_norm": 1.7936779260635376, "learning_rate": 1.949730182492473e-06, "loss": 0.09266990423202515, "step": 30520 }, { "epoch": 0.8042067414546449, "grad_norm": 2.3381783962249756, "learning_rate": 1.9446719766602793e-06, "loss": 0.09551510810852051, "step": 30530 }, { "epoch": 0.8044701566991436, "grad_norm": 4.907716751098633, "learning_rate": 1.9396196338433547e-06, "loss": 0.10248389244079589, "step": 30540 }, { "epoch": 0.8047335719436423, "grad_norm": 3.1806817054748535, "learning_rate": 1.9345731577190187e-06, "loss": 0.09794178009033203, "step": 30550 }, { "epoch": 0.804996987188141, "grad_norm": 2.9237232208251953, "learning_rate": 1.929532551960309e-06, "loss": 0.10022220611572266, "step": 30560 }, { "epoch": 0.8052604024326397, "grad_norm": 2.226301670074463, "learning_rate": 1.9244978202359974e-06, "loss": 0.08280019164085388, "step": 30570 }, { "epoch": 0.8055238176771385, "grad_norm": 2.3634235858917236, "learning_rate": 1.9194689662105803e-06, "loss": 0.0862764596939087, "step": 30580 }, { "epoch": 0.8057872329216372, "grad_norm": 3.0692973136901855, "learning_rate": 1.9144459935442705e-06, "loss": 0.10585191249847412, "step": 30590 }, { "epoch": 0.806050648166136, "grad_norm": 2.078564405441284, "learning_rate": 1.9094289058930095e-06, "loss": 0.095831298828125, "step": 30600 }, { "epoch": 0.8063140634106347, "grad_norm": 2.248180627822876, "learning_rate": 1.9044177069084502e-06, "loss": 0.09428093433380128, "step": 30610 }, { "epoch": 0.8065774786551335, "grad_norm": 2.868706464767456, "learning_rate": 1.899412400237951e-06, "loss": 0.09897406697273255, "step": 30620 }, { "epoch": 0.8068408938996322, "grad_norm": 2.8816494941711426, "learning_rate": 1.8944129895246e-06, "loss": 0.09990842938423157, "step": 30630 }, { "epoch": 0.8071043091441309, "grad_norm": 2.5211827754974365, "learning_rate": 1.8894194784071761e-06, "loss": 0.09709778428077698, "step": 30640 }, { "epoch": 0.8073677243886297, "grad_norm": 2.743467330932617, "learning_rate": 1.8844318705201815e-06, "loss": 0.10167313814163208, "step": 30650 }, { "epoch": 0.8076311396331284, "grad_norm": 2.6645095348358154, "learning_rate": 1.879450169493805e-06, "loss": 0.09646999835968018, "step": 30660 }, { "epoch": 0.8078945548776272, "grad_norm": 2.4814369678497314, "learning_rate": 1.874474378953942e-06, "loss": 0.10594594478607178, "step": 30670 }, { "epoch": 0.8081579701221259, "grad_norm": 3.0946226119995117, "learning_rate": 1.8695045025221936e-06, "loss": 0.09078269004821778, "step": 30680 }, { "epoch": 0.8084213853666247, "grad_norm": 2.339773416519165, "learning_rate": 1.864540543815848e-06, "loss": 0.08643577694892883, "step": 30690 }, { "epoch": 0.8086848006111234, "grad_norm": 4.152156829833984, "learning_rate": 1.859582506447889e-06, "loss": 0.10582339763641357, "step": 30700 }, { "epoch": 0.8089482158556222, "grad_norm": 2.9115519523620605, "learning_rate": 1.8546303940269883e-06, "loss": 0.09866341948509216, "step": 30710 }, { "epoch": 0.8092116311001208, "grad_norm": 2.5475013256073, "learning_rate": 1.8496842101575053e-06, "loss": 0.08904224038124084, "step": 30720 }, { "epoch": 0.8094750463446195, "grad_norm": 3.980696439743042, "learning_rate": 1.844743958439491e-06, "loss": 0.10812768936157227, "step": 30730 }, { "epoch": 0.8097384615891183, "grad_norm": 2.1719236373901367, "learning_rate": 1.839809642468674e-06, "loss": 0.09089086055755616, "step": 30740 }, { "epoch": 0.810001876833617, "grad_norm": 2.1164262294769287, "learning_rate": 1.8348812658364557e-06, "loss": 0.11146339178085327, "step": 30750 }, { "epoch": 0.8102652920781158, "grad_norm": 2.0041279792785645, "learning_rate": 1.8299588321299266e-06, "loss": 0.08861572742462158, "step": 30760 }, { "epoch": 0.8105287073226145, "grad_norm": 2.4304537773132324, "learning_rate": 1.8250423449318434e-06, "loss": 0.10155043601989747, "step": 30770 }, { "epoch": 0.8107921225671133, "grad_norm": 2.6284854412078857, "learning_rate": 1.8201318078206443e-06, "loss": 0.10375233888626098, "step": 30780 }, { "epoch": 0.811055537811612, "grad_norm": 2.0524144172668457, "learning_rate": 1.8152272243704216e-06, "loss": 0.11281903982162475, "step": 30790 }, { "epoch": 0.8113189530561108, "grad_norm": 2.297621726989746, "learning_rate": 1.8103285981509432e-06, "loss": 0.10382499694824218, "step": 30800 }, { "epoch": 0.8115823683006095, "grad_norm": 2.2444655895233154, "learning_rate": 1.8054359327276473e-06, "loss": 0.09537651538848876, "step": 30810 }, { "epoch": 0.8118457835451082, "grad_norm": 2.4107964038848877, "learning_rate": 1.8005492316616224e-06, "loss": 0.10228395462036133, "step": 30820 }, { "epoch": 0.812109198789607, "grad_norm": 2.5273852348327637, "learning_rate": 1.7956684985096217e-06, "loss": 0.08936657905578613, "step": 30830 }, { "epoch": 0.8123726140341057, "grad_norm": 3.3991477489471436, "learning_rate": 1.7907937368240525e-06, "loss": 0.09051579236984253, "step": 30840 }, { "epoch": 0.8126360292786045, "grad_norm": 3.0197556018829346, "learning_rate": 1.7859249501529742e-06, "loss": 0.10738208293914794, "step": 30850 }, { "epoch": 0.8128994445231031, "grad_norm": 2.8534059524536133, "learning_rate": 1.7810621420401054e-06, "loss": 0.10212757587432861, "step": 30860 }, { "epoch": 0.8131628597676019, "grad_norm": 2.1791276931762695, "learning_rate": 1.776205316024806e-06, "loss": 0.09426709413528442, "step": 30870 }, { "epoch": 0.8134262750121006, "grad_norm": 1.9910576343536377, "learning_rate": 1.7713544756420785e-06, "loss": 0.09418888092041015, "step": 30880 }, { "epoch": 0.8136896902565994, "grad_norm": 2.43086576461792, "learning_rate": 1.76650962442258e-06, "loss": 0.10285725593566894, "step": 30890 }, { "epoch": 0.8139531055010981, "grad_norm": 4.227409362792969, "learning_rate": 1.761670765892599e-06, "loss": 0.10386795997619629, "step": 30900 }, { "epoch": 0.8142165207455968, "grad_norm": 2.5395941734313965, "learning_rate": 1.7568379035740669e-06, "loss": 0.09774562120437622, "step": 30910 }, { "epoch": 0.8144799359900956, "grad_norm": 2.738994836807251, "learning_rate": 1.7520110409845493e-06, "loss": 0.09285587072372437, "step": 30920 }, { "epoch": 0.8147433512345943, "grad_norm": 4.55165433883667, "learning_rate": 1.7471901816372416e-06, "loss": 0.09486839175224304, "step": 30930 }, { "epoch": 0.8150067664790931, "grad_norm": 2.1373038291931152, "learning_rate": 1.7423753290409784e-06, "loss": 0.09066016674041748, "step": 30940 }, { "epoch": 0.8152701817235918, "grad_norm": 3.117457389831543, "learning_rate": 1.7375664867002151e-06, "loss": 0.09342095851898194, "step": 30950 }, { "epoch": 0.8155335969680906, "grad_norm": 1.830337643623352, "learning_rate": 1.7327636581150343e-06, "loss": 0.09484783411026002, "step": 30960 }, { "epoch": 0.8157970122125893, "grad_norm": 2.6915721893310547, "learning_rate": 1.7279668467811427e-06, "loss": 0.09830444455146789, "step": 30970 }, { "epoch": 0.816060427457088, "grad_norm": 2.9246022701263428, "learning_rate": 1.7231760561898625e-06, "loss": 0.09713563919067383, "step": 30980 }, { "epoch": 0.8163238427015868, "grad_norm": 2.894904136657715, "learning_rate": 1.718391289828144e-06, "loss": 0.08002921342849731, "step": 30990 }, { "epoch": 0.8165872579460854, "grad_norm": 3.4762189388275146, "learning_rate": 1.7136125511785462e-06, "loss": 0.08948060870170593, "step": 31000 }, { "epoch": 0.8168506731905842, "grad_norm": 2.3498125076293945, "learning_rate": 1.7088398437192344e-06, "loss": 0.10672527551651001, "step": 31010 }, { "epoch": 0.8171140884350829, "grad_norm": 2.2244081497192383, "learning_rate": 1.7040731709239967e-06, "loss": 0.10066291093826293, "step": 31020 }, { "epoch": 0.8173775036795817, "grad_norm": 2.7793843746185303, "learning_rate": 1.6993125362622231e-06, "loss": 0.0979849100112915, "step": 31030 }, { "epoch": 0.8176409189240804, "grad_norm": 2.3649961948394775, "learning_rate": 1.6945579431989067e-06, "loss": 0.0941652774810791, "step": 31040 }, { "epoch": 0.8179043341685792, "grad_norm": 1.987760305404663, "learning_rate": 1.689809395194647e-06, "loss": 0.08241910934448242, "step": 31050 }, { "epoch": 0.8181677494130779, "grad_norm": 3.7616117000579834, "learning_rate": 1.6850668957056393e-06, "loss": 0.1077894926071167, "step": 31060 }, { "epoch": 0.8184311646575766, "grad_norm": 3.2787516117095947, "learning_rate": 1.6803304481836835e-06, "loss": 0.09642963409423828, "step": 31070 }, { "epoch": 0.8186945799020754, "grad_norm": 3.4077577590942383, "learning_rate": 1.6756000560761687e-06, "loss": 0.08659828901290893, "step": 31080 }, { "epoch": 0.8189579951465741, "grad_norm": 2.0871922969818115, "learning_rate": 1.6708757228260786e-06, "loss": 0.09538092613220214, "step": 31090 }, { "epoch": 0.8192214103910729, "grad_norm": 2.0642812252044678, "learning_rate": 1.6661574518719859e-06, "loss": 0.08698605298995972, "step": 31100 }, { "epoch": 0.8194848256355716, "grad_norm": 3.0015039443969727, "learning_rate": 1.6614452466480502e-06, "loss": 0.09203532338142395, "step": 31110 }, { "epoch": 0.8197482408800704, "grad_norm": 2.4529013633728027, "learning_rate": 1.6567391105840247e-06, "loss": 0.08689934015274048, "step": 31120 }, { "epoch": 0.820011656124569, "grad_norm": 2.5096912384033203, "learning_rate": 1.6520390471052306e-06, "loss": 0.0933087944984436, "step": 31130 }, { "epoch": 0.8202750713690679, "grad_norm": 2.102799415588379, "learning_rate": 1.647345059632578e-06, "loss": 0.07490870952606202, "step": 31140 }, { "epoch": 0.8205384866135665, "grad_norm": 2.478741407394409, "learning_rate": 1.6426571515825563e-06, "loss": 0.09692553281784058, "step": 31150 }, { "epoch": 0.8208019018580652, "grad_norm": 2.7815725803375244, "learning_rate": 1.6379753263672272e-06, "loss": 0.09287257790565491, "step": 31160 }, { "epoch": 0.821065317102564, "grad_norm": 3.459646224975586, "learning_rate": 1.6332995873942237e-06, "loss": 0.0885515809059143, "step": 31170 }, { "epoch": 0.8213287323470627, "grad_norm": 3.0951526165008545, "learning_rate": 1.6286299380667504e-06, "loss": 0.09339772462844849, "step": 31180 }, { "epoch": 0.8215921475915615, "grad_norm": 3.6410279273986816, "learning_rate": 1.6239663817835783e-06, "loss": 0.09201655387878419, "step": 31190 }, { "epoch": 0.8218555628360602, "grad_norm": 2.8296027183532715, "learning_rate": 1.6193089219390478e-06, "loss": 0.09622573852539062, "step": 31200 }, { "epoch": 0.822118978080559, "grad_norm": 2.3281657695770264, "learning_rate": 1.6146575619230597e-06, "loss": 0.09542523622512818, "step": 31210 }, { "epoch": 0.8223823933250577, "grad_norm": 2.5819571018218994, "learning_rate": 1.610012305121068e-06, "loss": 0.0890629231929779, "step": 31220 }, { "epoch": 0.8226458085695565, "grad_norm": 2.534057140350342, "learning_rate": 1.6053731549140982e-06, "loss": 0.10453605651855469, "step": 31230 }, { "epoch": 0.8229092238140552, "grad_norm": 2.7559938430786133, "learning_rate": 1.6007401146787195e-06, "loss": 0.10017104148864746, "step": 31240 }, { "epoch": 0.8231726390585539, "grad_norm": 2.034790515899658, "learning_rate": 1.596113187787064e-06, "loss": 0.08866162896156311, "step": 31250 }, { "epoch": 0.8234360543030527, "grad_norm": 3.0805184841156006, "learning_rate": 1.5914923776068025e-06, "loss": 0.09727307558059692, "step": 31260 }, { "epoch": 0.8236994695475514, "grad_norm": 2.464045286178589, "learning_rate": 1.5868776875011604e-06, "loss": 0.08792357444763184, "step": 31270 }, { "epoch": 0.8239628847920502, "grad_norm": 3.8989758491516113, "learning_rate": 1.582269120828912e-06, "loss": 0.10847678184509277, "step": 31280 }, { "epoch": 0.8242263000365488, "grad_norm": 2.3710880279541016, "learning_rate": 1.5776666809443697e-06, "loss": 0.07940750122070313, "step": 31290 }, { "epoch": 0.8244897152810476, "grad_norm": 2.2270874977111816, "learning_rate": 1.5730703711973883e-06, "loss": 0.09298474788665771, "step": 31300 }, { "epoch": 0.8247531305255463, "grad_norm": 2.3417305946350098, "learning_rate": 1.5684801949333584e-06, "loss": 0.09298722743988037, "step": 31310 }, { "epoch": 0.8250165457700451, "grad_norm": 2.444632053375244, "learning_rate": 1.5638961554932064e-06, "loss": 0.10411978960037231, "step": 31320 }, { "epoch": 0.8252799610145438, "grad_norm": 2.585911273956299, "learning_rate": 1.5593182562134002e-06, "loss": 0.10083966255187989, "step": 31330 }, { "epoch": 0.8255433762590425, "grad_norm": 2.5457499027252197, "learning_rate": 1.554746500425931e-06, "loss": 0.090892493724823, "step": 31340 }, { "epoch": 0.8258067915035413, "grad_norm": 1.728645920753479, "learning_rate": 1.5501808914583139e-06, "loss": 0.09070125818252564, "step": 31350 }, { "epoch": 0.82607020674804, "grad_norm": 2.352275848388672, "learning_rate": 1.5456214326336017e-06, "loss": 0.09215530157089233, "step": 31360 }, { "epoch": 0.8263336219925388, "grad_norm": 2.2035036087036133, "learning_rate": 1.5410681272703643e-06, "loss": 0.08808825016021729, "step": 31370 }, { "epoch": 0.8265970372370375, "grad_norm": 3.3652148246765137, "learning_rate": 1.536520978682693e-06, "loss": 0.09698827266693115, "step": 31380 }, { "epoch": 0.8268604524815363, "grad_norm": 2.276203155517578, "learning_rate": 1.531979990180199e-06, "loss": 0.08429746031761169, "step": 31390 }, { "epoch": 0.827123867726035, "grad_norm": 2.603691577911377, "learning_rate": 1.5274451650680066e-06, "loss": 0.10550189018249512, "step": 31400 }, { "epoch": 0.8273872829705338, "grad_norm": 3.0391976833343506, "learning_rate": 1.5229165066467632e-06, "loss": 0.0988371193408966, "step": 31410 }, { "epoch": 0.8276506982150325, "grad_norm": 2.063725709915161, "learning_rate": 1.5183940182126189e-06, "loss": 0.09098328351974487, "step": 31420 }, { "epoch": 0.8279141134595311, "grad_norm": 2.476106882095337, "learning_rate": 1.513877703057235e-06, "loss": 0.09415099024772644, "step": 31430 }, { "epoch": 0.8281775287040299, "grad_norm": 2.1580634117126465, "learning_rate": 1.509367564467782e-06, "loss": 0.09145324230194092, "step": 31440 }, { "epoch": 0.8284409439485286, "grad_norm": 3.262197256088257, "learning_rate": 1.5048636057269294e-06, "loss": 0.09776331186294555, "step": 31450 }, { "epoch": 0.8287043591930274, "grad_norm": 2.831303596496582, "learning_rate": 1.50036583011286e-06, "loss": 0.09955040216445923, "step": 31460 }, { "epoch": 0.8289677744375261, "grad_norm": 2.0637359619140625, "learning_rate": 1.4958742408992454e-06, "loss": 0.09139820337295532, "step": 31470 }, { "epoch": 0.8292311896820249, "grad_norm": 2.036017894744873, "learning_rate": 1.4913888413552536e-06, "loss": 0.08535237312316894, "step": 31480 }, { "epoch": 0.8294946049265236, "grad_norm": 3.20422101020813, "learning_rate": 1.486909634745557e-06, "loss": 0.08529722690582275, "step": 31490 }, { "epoch": 0.8297580201710223, "grad_norm": 2.3104121685028076, "learning_rate": 1.4824366243303156e-06, "loss": 0.08662387132644653, "step": 31500 }, { "epoch": 0.8300214354155211, "grad_norm": 2.1646652221679688, "learning_rate": 1.4779698133651766e-06, "loss": 0.09531365633010865, "step": 31510 }, { "epoch": 0.8302848506600198, "grad_norm": 2.7675859928131104, "learning_rate": 1.47350920510128e-06, "loss": 0.09713966846466064, "step": 31520 }, { "epoch": 0.8305482659045186, "grad_norm": 3.3948557376861572, "learning_rate": 1.469054802785246e-06, "loss": 0.09569525718688965, "step": 31530 }, { "epoch": 0.8308116811490173, "grad_norm": 3.6092159748077393, "learning_rate": 1.4646066096591838e-06, "loss": 0.09507790803909302, "step": 31540 }, { "epoch": 0.8310750963935161, "grad_norm": 3.075101137161255, "learning_rate": 1.4601646289606809e-06, "loss": 0.09998743534088135, "step": 31550 }, { "epoch": 0.8313385116380148, "grad_norm": 2.9922666549682617, "learning_rate": 1.4557288639227995e-06, "loss": 0.08000229597091675, "step": 31560 }, { "epoch": 0.8316019268825136, "grad_norm": 2.5776848793029785, "learning_rate": 1.451299317774083e-06, "loss": 0.08807108402252198, "step": 31570 }, { "epoch": 0.8318653421270122, "grad_norm": 2.8142130374908447, "learning_rate": 1.4468759937385423e-06, "loss": 0.08307018280029296, "step": 31580 }, { "epoch": 0.8321287573715109, "grad_norm": 2.614319086074829, "learning_rate": 1.4424588950356688e-06, "loss": 0.09174758195877075, "step": 31590 }, { "epoch": 0.8323921726160097, "grad_norm": 2.7027266025543213, "learning_rate": 1.4380480248804185e-06, "loss": 0.11584388017654419, "step": 31600 }, { "epoch": 0.8326555878605084, "grad_norm": 3.740129232406616, "learning_rate": 1.4336433864832033e-06, "loss": 0.10076655149459839, "step": 31610 }, { "epoch": 0.8329190031050072, "grad_norm": 3.0298213958740234, "learning_rate": 1.4292449830499189e-06, "loss": 0.09049751162528992, "step": 31620 }, { "epoch": 0.8331824183495059, "grad_norm": 2.243001937866211, "learning_rate": 1.4248528177819098e-06, "loss": 0.09158796668052674, "step": 31630 }, { "epoch": 0.8334458335940047, "grad_norm": 3.5461277961730957, "learning_rate": 1.4204668938759835e-06, "loss": 0.08796547651290894, "step": 31640 }, { "epoch": 0.8337092488385034, "grad_norm": 2.8989877700805664, "learning_rate": 1.416087214524403e-06, "loss": 0.09824891090393066, "step": 31650 }, { "epoch": 0.8339726640830022, "grad_norm": 2.799724817276001, "learning_rate": 1.4117137829148885e-06, "loss": 0.08414973020553589, "step": 31660 }, { "epoch": 0.8342360793275009, "grad_norm": 1.921341061592102, "learning_rate": 1.4073466022306148e-06, "loss": 0.08686314821243286, "step": 31670 }, { "epoch": 0.8344994945719996, "grad_norm": 2.6184468269348145, "learning_rate": 1.4029856756502026e-06, "loss": 0.09252899289131164, "step": 31680 }, { "epoch": 0.8347629098164984, "grad_norm": 3.201317310333252, "learning_rate": 1.3986310063477215e-06, "loss": 0.09729156494140626, "step": 31690 }, { "epoch": 0.835026325060997, "grad_norm": 2.6880311965942383, "learning_rate": 1.3942825974926888e-06, "loss": 0.09880802035331726, "step": 31700 }, { "epoch": 0.8352897403054959, "grad_norm": 2.0088493824005127, "learning_rate": 1.3899404522500626e-06, "loss": 0.09689077138900756, "step": 31710 }, { "epoch": 0.8355531555499945, "grad_norm": 2.187307834625244, "learning_rate": 1.385604573780248e-06, "loss": 0.07924439311027527, "step": 31720 }, { "epoch": 0.8358165707944933, "grad_norm": 2.568347692489624, "learning_rate": 1.38127496523908e-06, "loss": 0.10539534091949462, "step": 31730 }, { "epoch": 0.836079986038992, "grad_norm": 2.3447234630584717, "learning_rate": 1.376951629777833e-06, "loss": 0.09251841306686401, "step": 31740 }, { "epoch": 0.8363434012834908, "grad_norm": 2.2706544399261475, "learning_rate": 1.372634570543222e-06, "loss": 0.09897315502166748, "step": 31750 }, { "epoch": 0.8366068165279895, "grad_norm": 1.9222424030303955, "learning_rate": 1.3683237906773873e-06, "loss": 0.08837059140205383, "step": 31760 }, { "epoch": 0.8368702317724882, "grad_norm": 1.8603436946868896, "learning_rate": 1.3640192933178997e-06, "loss": 0.09027116894721984, "step": 31770 }, { "epoch": 0.837133647016987, "grad_norm": 2.943298101425171, "learning_rate": 1.359721081597758e-06, "loss": 0.09466145634651184, "step": 31780 }, { "epoch": 0.8373970622614857, "grad_norm": 2.3467488288879395, "learning_rate": 1.355429158645384e-06, "loss": 0.09216447472572327, "step": 31790 }, { "epoch": 0.8376604775059845, "grad_norm": 2.9414145946502686, "learning_rate": 1.3511435275846297e-06, "loss": 0.10846641063690185, "step": 31800 }, { "epoch": 0.8379238927504832, "grad_norm": 3.8172607421875, "learning_rate": 1.3468641915347613e-06, "loss": 0.0958518147468567, "step": 31810 }, { "epoch": 0.838187307994982, "grad_norm": 3.151538848876953, "learning_rate": 1.3425911536104573e-06, "loss": 0.085825514793396, "step": 31820 }, { "epoch": 0.8384507232394807, "grad_norm": 2.043036699295044, "learning_rate": 1.3383244169218268e-06, "loss": 0.08491257429122925, "step": 31830 }, { "epoch": 0.8387141384839795, "grad_norm": 3.5143566131591797, "learning_rate": 1.3340639845743797e-06, "loss": 0.10637490749359131, "step": 31840 }, { "epoch": 0.8389775537284782, "grad_norm": 3.369982957839966, "learning_rate": 1.32980985966905e-06, "loss": 0.10065252780914306, "step": 31850 }, { "epoch": 0.8392409689729768, "grad_norm": 3.1288812160491943, "learning_rate": 1.3255620453021656e-06, "loss": 0.09662045240402221, "step": 31860 }, { "epoch": 0.8395043842174756, "grad_norm": 3.069850206375122, "learning_rate": 1.3213205445654688e-06, "loss": 0.0963205873966217, "step": 31870 }, { "epoch": 0.8397677994619743, "grad_norm": 2.8296048641204834, "learning_rate": 1.3170853605461132e-06, "loss": 0.09412869215011596, "step": 31880 }, { "epoch": 0.8400312147064731, "grad_norm": 3.0377068519592285, "learning_rate": 1.3128564963266444e-06, "loss": 0.10329768657684327, "step": 31890 }, { "epoch": 0.8402946299509718, "grad_norm": 2.4232866764068604, "learning_rate": 1.3086339549850125e-06, "loss": 0.0925719678401947, "step": 31900 }, { "epoch": 0.8405580451954706, "grad_norm": 2.3965601921081543, "learning_rate": 1.3044177395945657e-06, "loss": 0.08683871030807495, "step": 31910 }, { "epoch": 0.8408214604399693, "grad_norm": 2.1395440101623535, "learning_rate": 1.3002078532240447e-06, "loss": 0.09970546364784241, "step": 31920 }, { "epoch": 0.8410848756844681, "grad_norm": 2.106067419052124, "learning_rate": 1.296004298937592e-06, "loss": 0.08915809392929078, "step": 31930 }, { "epoch": 0.8413482909289668, "grad_norm": 2.7812604904174805, "learning_rate": 1.2918070797947346e-06, "loss": 0.07749372720718384, "step": 31940 }, { "epoch": 0.8416117061734655, "grad_norm": 2.5115103721618652, "learning_rate": 1.2876161988503843e-06, "loss": 0.10428810119628906, "step": 31950 }, { "epoch": 0.8418751214179643, "grad_norm": 2.3500421047210693, "learning_rate": 1.2834316591548502e-06, "loss": 0.08797168135643005, "step": 31960 }, { "epoch": 0.842138536662463, "grad_norm": 3.381831407546997, "learning_rate": 1.279253463753819e-06, "loss": 0.09051263928413392, "step": 31970 }, { "epoch": 0.8424019519069618, "grad_norm": 3.07833194732666, "learning_rate": 1.2750816156883627e-06, "loss": 0.08954388499259949, "step": 31980 }, { "epoch": 0.8426653671514605, "grad_norm": 2.722583293914795, "learning_rate": 1.2709161179949303e-06, "loss": 0.08398844003677368, "step": 31990 }, { "epoch": 0.8429287823959593, "grad_norm": 2.4658799171447754, "learning_rate": 1.266756973705351e-06, "loss": 0.09751396775245666, "step": 32000 }, { "epoch": 0.8431921976404579, "grad_norm": 2.4769816398620605, "learning_rate": 1.2626041858468318e-06, "loss": 0.094350266456604, "step": 32010 }, { "epoch": 0.8434556128849566, "grad_norm": 3.7852659225463867, "learning_rate": 1.2584577574419499e-06, "loss": 0.08941997289657592, "step": 32020 }, { "epoch": 0.8437190281294554, "grad_norm": 2.679804563522339, "learning_rate": 1.2543176915086542e-06, "loss": 0.09915469288825988, "step": 32030 }, { "epoch": 0.8439824433739541, "grad_norm": 3.950805425643921, "learning_rate": 1.2501839910602632e-06, "loss": 0.08649638891220093, "step": 32040 }, { "epoch": 0.8442458586184529, "grad_norm": 3.1717381477355957, "learning_rate": 1.24605665910546e-06, "loss": 0.09840685725212098, "step": 32050 }, { "epoch": 0.8445092738629516, "grad_norm": 2.830350637435913, "learning_rate": 1.2419356986483e-06, "loss": 0.09246244430541992, "step": 32060 }, { "epoch": 0.8447726891074504, "grad_norm": 2.619256019592285, "learning_rate": 1.2378211126881956e-06, "loss": 0.08771970272064208, "step": 32070 }, { "epoch": 0.8450361043519491, "grad_norm": 3.602458953857422, "learning_rate": 1.2337129042199147e-06, "loss": 0.0785430371761322, "step": 32080 }, { "epoch": 0.8452995195964479, "grad_norm": 2.352810859680176, "learning_rate": 1.2296110762335933e-06, "loss": 0.08224709033966064, "step": 32090 }, { "epoch": 0.8455629348409466, "grad_norm": 2.7197721004486084, "learning_rate": 1.2255156317147165e-06, "loss": 0.08867591619491577, "step": 32100 }, { "epoch": 0.8458263500854453, "grad_norm": 3.315081834793091, "learning_rate": 1.2214265736441277e-06, "loss": 0.09439079761505127, "step": 32110 }, { "epoch": 0.8460897653299441, "grad_norm": 4.1434783935546875, "learning_rate": 1.2173439049980174e-06, "loss": 0.09573618173599244, "step": 32120 }, { "epoch": 0.8463531805744428, "grad_norm": 3.240906000137329, "learning_rate": 1.2132676287479273e-06, "loss": 0.08306522369384765, "step": 32130 }, { "epoch": 0.8466165958189416, "grad_norm": 3.0027520656585693, "learning_rate": 1.209197747860752e-06, "loss": 0.08844602704048157, "step": 32140 }, { "epoch": 0.8468800110634402, "grad_norm": 3.121194839477539, "learning_rate": 1.2051342652987218e-06, "loss": 0.09060277938842773, "step": 32150 }, { "epoch": 0.847143426307939, "grad_norm": 3.3218421936035156, "learning_rate": 1.2010771840194168e-06, "loss": 0.08614819049835205, "step": 32160 }, { "epoch": 0.8474068415524377, "grad_norm": 3.201578140258789, "learning_rate": 1.1970265069757548e-06, "loss": 0.09462710618972778, "step": 32170 }, { "epoch": 0.8476702567969365, "grad_norm": 2.758577823638916, "learning_rate": 1.1929822371159915e-06, "loss": 0.09486951828002929, "step": 32180 }, { "epoch": 0.8479336720414352, "grad_norm": 2.8519110679626465, "learning_rate": 1.188944377383725e-06, "loss": 0.09094016551971436, "step": 32190 }, { "epoch": 0.8481970872859339, "grad_norm": 2.3336007595062256, "learning_rate": 1.1849129307178841e-06, "loss": 0.09027502536773682, "step": 32200 }, { "epoch": 0.8484605025304327, "grad_norm": 2.641582489013672, "learning_rate": 1.180887900052724e-06, "loss": 0.08529087901115417, "step": 32210 }, { "epoch": 0.8487239177749314, "grad_norm": 3.259018659591675, "learning_rate": 1.17686928831784e-06, "loss": 0.09849525094032288, "step": 32220 }, { "epoch": 0.8489873330194302, "grad_norm": 2.2554421424865723, "learning_rate": 1.1728570984381505e-06, "loss": 0.10396763086318969, "step": 32230 }, { "epoch": 0.8492507482639289, "grad_norm": 2.743765115737915, "learning_rate": 1.1688513333339003e-06, "loss": 0.09535440802574158, "step": 32240 }, { "epoch": 0.8495141635084277, "grad_norm": 2.3280506134033203, "learning_rate": 1.1648519959206584e-06, "loss": 0.08113361597061157, "step": 32250 }, { "epoch": 0.8497775787529264, "grad_norm": 2.901970624923706, "learning_rate": 1.1608590891093118e-06, "loss": 0.07513526678085328, "step": 32260 }, { "epoch": 0.8500409939974252, "grad_norm": 3.805258274078369, "learning_rate": 1.156872615806075e-06, "loss": 0.09249999523162841, "step": 32270 }, { "epoch": 0.8503044092419239, "grad_norm": 2.2241711616516113, "learning_rate": 1.1528925789124734e-06, "loss": 0.08473803400993347, "step": 32280 }, { "epoch": 0.8505678244864225, "grad_norm": 2.9970738887786865, "learning_rate": 1.1489189813253499e-06, "loss": 0.09487800002098083, "step": 32290 }, { "epoch": 0.8508312397309213, "grad_norm": 3.258272171020508, "learning_rate": 1.1449518259368596e-06, "loss": 0.08526232242584228, "step": 32300 }, { "epoch": 0.85109465497542, "grad_norm": 2.5618741512298584, "learning_rate": 1.1409911156344677e-06, "loss": 0.08937000632286071, "step": 32310 }, { "epoch": 0.8513580702199188, "grad_norm": 2.6608293056488037, "learning_rate": 1.1370368533009557e-06, "loss": 0.08330026268959045, "step": 32320 }, { "epoch": 0.8516214854644175, "grad_norm": 2.454956531524658, "learning_rate": 1.1330890418144014e-06, "loss": 0.08216527700424195, "step": 32330 }, { "epoch": 0.8518849007089163, "grad_norm": 4.227522373199463, "learning_rate": 1.1291476840481908e-06, "loss": 0.09793332815170289, "step": 32340 }, { "epoch": 0.852148315953415, "grad_norm": 3.046685218811035, "learning_rate": 1.125212782871019e-06, "loss": 0.09203503131866456, "step": 32350 }, { "epoch": 0.8524117311979138, "grad_norm": 2.3183887004852295, "learning_rate": 1.1212843411468755e-06, "loss": 0.08685963153839112, "step": 32360 }, { "epoch": 0.8526751464424125, "grad_norm": 3.7299699783325195, "learning_rate": 1.1173623617350472e-06, "loss": 0.08451573848724366, "step": 32370 }, { "epoch": 0.8529385616869112, "grad_norm": 2.43953537940979, "learning_rate": 1.1134468474901227e-06, "loss": 0.092566579580307, "step": 32380 }, { "epoch": 0.85320197693141, "grad_norm": 2.890434980392456, "learning_rate": 1.1095378012619796e-06, "loss": 0.08471944332122802, "step": 32390 }, { "epoch": 0.8534653921759087, "grad_norm": 2.2957026958465576, "learning_rate": 1.1056352258957924e-06, "loss": 0.08526875972747802, "step": 32400 }, { "epoch": 0.8537288074204075, "grad_norm": 3.2105636596679688, "learning_rate": 1.1017391242320263e-06, "loss": 0.07909715175628662, "step": 32410 }, { "epoch": 0.8539922226649062, "grad_norm": 3.073667526245117, "learning_rate": 1.0978494991064248e-06, "loss": 0.09309277534484864, "step": 32420 }, { "epoch": 0.854255637909405, "grad_norm": 2.7713539600372314, "learning_rate": 1.0939663533500311e-06, "loss": 0.09165113568305969, "step": 32430 }, { "epoch": 0.8545190531539036, "grad_norm": 2.0865566730499268, "learning_rate": 1.0900896897891634e-06, "loss": 0.08766847848892212, "step": 32440 }, { "epoch": 0.8547824683984024, "grad_norm": 3.547109365463257, "learning_rate": 1.0862195112454289e-06, "loss": 0.08916077017784119, "step": 32450 }, { "epoch": 0.8550458836429011, "grad_norm": 1.868564248085022, "learning_rate": 1.0823558205357054e-06, "loss": 0.07824475765228271, "step": 32460 }, { "epoch": 0.8553092988873998, "grad_norm": 3.138070583343506, "learning_rate": 1.0784986204721537e-06, "loss": 0.0924800157546997, "step": 32470 }, { "epoch": 0.8555727141318986, "grad_norm": 4.602108001708984, "learning_rate": 1.0746479138622134e-06, "loss": 0.09036647081375122, "step": 32480 }, { "epoch": 0.8558361293763973, "grad_norm": 4.089128017425537, "learning_rate": 1.070803703508595e-06, "loss": 0.09178963899612427, "step": 32490 }, { "epoch": 0.8560995446208961, "grad_norm": 3.183924913406372, "learning_rate": 1.066965992209279e-06, "loss": 0.08206756114959717, "step": 32500 }, { "epoch": 0.8563629598653948, "grad_norm": 3.005079984664917, "learning_rate": 1.0631347827575179e-06, "loss": 0.0867100179195404, "step": 32510 }, { "epoch": 0.8566263751098936, "grad_norm": 3.83398175239563, "learning_rate": 1.0593100779418286e-06, "loss": 0.08517922759056092, "step": 32520 }, { "epoch": 0.8568897903543923, "grad_norm": 3.2933239936828613, "learning_rate": 1.0554918805460013e-06, "loss": 0.10249640941619872, "step": 32530 }, { "epoch": 0.8571532055988911, "grad_norm": 3.275099515914917, "learning_rate": 1.051680193349084e-06, "loss": 0.08297280073165894, "step": 32540 }, { "epoch": 0.8574166208433898, "grad_norm": 2.6533119678497314, "learning_rate": 1.0478750191253817e-06, "loss": 0.08793706893920898, "step": 32550 }, { "epoch": 0.8576800360878885, "grad_norm": 3.3969833850860596, "learning_rate": 1.0440763606444694e-06, "loss": 0.0836532473564148, "step": 32560 }, { "epoch": 0.8579434513323873, "grad_norm": 2.816520929336548, "learning_rate": 1.0402842206711728e-06, "loss": 0.10299391746520996, "step": 32570 }, { "epoch": 0.8582068665768859, "grad_norm": 4.19797945022583, "learning_rate": 1.0364986019655765e-06, "loss": 0.08938438296318055, "step": 32580 }, { "epoch": 0.8584702818213847, "grad_norm": 3.4998795986175537, "learning_rate": 1.0327195072830153e-06, "loss": 0.09490979313850403, "step": 32590 }, { "epoch": 0.8587336970658834, "grad_norm": 2.7188382148742676, "learning_rate": 1.028946939374076e-06, "loss": 0.10289099216461181, "step": 32600 }, { "epoch": 0.8589971123103822, "grad_norm": 3.746230363845825, "learning_rate": 1.0251809009846003e-06, "loss": 0.0901245892047882, "step": 32610 }, { "epoch": 0.8592605275548809, "grad_norm": 2.0887339115142822, "learning_rate": 1.0214213948556718e-06, "loss": 0.07304342985153198, "step": 32620 }, { "epoch": 0.8595239427993796, "grad_norm": 2.6326181888580322, "learning_rate": 1.0176684237236212e-06, "loss": 0.09382811784744263, "step": 32630 }, { "epoch": 0.8597873580438784, "grad_norm": 2.637924909591675, "learning_rate": 1.0139219903200227e-06, "loss": 0.08879799842834472, "step": 32640 }, { "epoch": 0.8600507732883771, "grad_norm": 2.807480812072754, "learning_rate": 1.0101820973716891e-06, "loss": 0.09452091455459595, "step": 32650 }, { "epoch": 0.8603141885328759, "grad_norm": 3.830073833465576, "learning_rate": 1.0064487476006812e-06, "loss": 0.08214370012283326, "step": 32660 }, { "epoch": 0.8605776037773746, "grad_norm": 3.4015252590179443, "learning_rate": 1.0027219437242907e-06, "loss": 0.09494649171829224, "step": 32670 }, { "epoch": 0.8608410190218734, "grad_norm": 3.5806620121002197, "learning_rate": 9.990016884550425e-07, "loss": 0.09274395108222962, "step": 32680 }, { "epoch": 0.8611044342663721, "grad_norm": 3.296393871307373, "learning_rate": 9.952879845007024e-07, "loss": 0.08576748371124268, "step": 32690 }, { "epoch": 0.8613678495108709, "grad_norm": 4.538103103637695, "learning_rate": 9.915808345642642e-07, "loss": 0.08663167357444763, "step": 32700 }, { "epoch": 0.8616312647553696, "grad_norm": 3.0867528915405273, "learning_rate": 9.878802413439514e-07, "loss": 0.08874956369400025, "step": 32710 }, { "epoch": 0.8618946799998682, "grad_norm": 3.160236120223999, "learning_rate": 9.841862075332143e-07, "loss": 0.0861254334449768, "step": 32720 }, { "epoch": 0.862158095244367, "grad_norm": 2.04128360748291, "learning_rate": 9.804987358207274e-07, "loss": 0.07846826314926147, "step": 32730 }, { "epoch": 0.8624215104888657, "grad_norm": 2.4094319343566895, "learning_rate": 9.768178288903985e-07, "loss": 0.08859136700630188, "step": 32740 }, { "epoch": 0.8626849257333645, "grad_norm": 2.745574951171875, "learning_rate": 9.731434894213465e-07, "loss": 0.08816262483596801, "step": 32750 }, { "epoch": 0.8629483409778632, "grad_norm": 2.7950148582458496, "learning_rate": 9.694757200879146e-07, "loss": 0.09601414799690247, "step": 32760 }, { "epoch": 0.863211756222362, "grad_norm": 3.03022837638855, "learning_rate": 9.658145235596628e-07, "loss": 0.0889272391796112, "step": 32770 }, { "epoch": 0.8634751714668607, "grad_norm": 3.494351625442505, "learning_rate": 9.621599025013672e-07, "loss": 0.09237630963325501, "step": 32780 }, { "epoch": 0.8637385867113595, "grad_norm": 3.01875376701355, "learning_rate": 9.585118595730226e-07, "loss": 0.10077033042907715, "step": 32790 }, { "epoch": 0.8640020019558582, "grad_norm": 3.4191527366638184, "learning_rate": 9.548703974298311e-07, "loss": 0.09456490874290466, "step": 32800 }, { "epoch": 0.8642654172003569, "grad_norm": 3.578895092010498, "learning_rate": 9.512355187222e-07, "loss": 0.09126850962638855, "step": 32810 }, { "epoch": 0.8645288324448557, "grad_norm": 2.125699520111084, "learning_rate": 9.47607226095758e-07, "loss": 0.0885597050189972, "step": 32820 }, { "epoch": 0.8647922476893544, "grad_norm": 4.101398468017578, "learning_rate": 9.439855221913297e-07, "loss": 0.09448999166488647, "step": 32830 }, { "epoch": 0.8650556629338532, "grad_norm": 3.3444647789001465, "learning_rate": 9.403704096449484e-07, "loss": 0.0926922082901001, "step": 32840 }, { "epoch": 0.8653190781783519, "grad_norm": 3.691126585006714, "learning_rate": 9.367618910878484e-07, "loss": 0.09707042574882507, "step": 32850 }, { "epoch": 0.8655824934228507, "grad_norm": 3.1530892848968506, "learning_rate": 9.331599691464644e-07, "loss": 0.09490395188331605, "step": 32860 }, { "epoch": 0.8658459086673493, "grad_norm": 2.3055613040924072, "learning_rate": 9.295646464424323e-07, "loss": 0.08329107761383056, "step": 32870 }, { "epoch": 0.8661093239118481, "grad_norm": 4.039424419403076, "learning_rate": 9.259759255925827e-07, "loss": 0.08386080861091613, "step": 32880 }, { "epoch": 0.8663727391563468, "grad_norm": 2.942166566848755, "learning_rate": 9.223938092089413e-07, "loss": 0.08463520407676697, "step": 32890 }, { "epoch": 0.8666361544008455, "grad_norm": 2.47821044921875, "learning_rate": 9.188182998987261e-07, "loss": 0.08591043949127197, "step": 32900 }, { "epoch": 0.8668995696453443, "grad_norm": 3.032508611679077, "learning_rate": 9.152494002643464e-07, "loss": 0.08442703485488892, "step": 32910 }, { "epoch": 0.867162984889843, "grad_norm": 3.00931715965271, "learning_rate": 9.116871129034066e-07, "loss": 0.08556808829307556, "step": 32920 }, { "epoch": 0.8674264001343418, "grad_norm": 2.784231185913086, "learning_rate": 9.081314404086872e-07, "loss": 0.08147199153900146, "step": 32930 }, { "epoch": 0.8676898153788405, "grad_norm": 2.7261788845062256, "learning_rate": 9.045823853681602e-07, "loss": 0.08856332302093506, "step": 32940 }, { "epoch": 0.8679532306233393, "grad_norm": 2.931943416595459, "learning_rate": 9.010399503649836e-07, "loss": 0.08383693695068359, "step": 32950 }, { "epoch": 0.868216645867838, "grad_norm": 3.285297155380249, "learning_rate": 8.975041379774951e-07, "loss": 0.09230266809463501, "step": 32960 }, { "epoch": 0.8684800611123368, "grad_norm": 3.4607696533203125, "learning_rate": 8.939749507792094e-07, "loss": 0.09397191405296326, "step": 32970 }, { "epoch": 0.8687434763568355, "grad_norm": 2.8619349002838135, "learning_rate": 8.904523913388218e-07, "loss": 0.07095471024513245, "step": 32980 }, { "epoch": 0.8690068916013342, "grad_norm": 2.2179036140441895, "learning_rate": 8.869364622202015e-07, "loss": 0.08215484023094177, "step": 32990 }, { "epoch": 0.869270306845833, "grad_norm": 3.068981409072876, "learning_rate": 8.834271659823967e-07, "loss": 0.09096800088882447, "step": 33000 }, { "epoch": 0.8695337220903316, "grad_norm": 2.681337356567383, "learning_rate": 8.799245051796246e-07, "loss": 0.090767502784729, "step": 33010 }, { "epoch": 0.8697971373348304, "grad_norm": 2.6497504711151123, "learning_rate": 8.764284823612667e-07, "loss": 0.0959240198135376, "step": 33020 }, { "epoch": 0.8700605525793291, "grad_norm": 3.059906482696533, "learning_rate": 8.729391000718867e-07, "loss": 0.09785443544387817, "step": 33030 }, { "epoch": 0.8703239678238279, "grad_norm": 2.0296592712402344, "learning_rate": 8.694563608512041e-07, "loss": 0.09433220624923706, "step": 33040 }, { "epoch": 0.8705873830683266, "grad_norm": 3.117859125137329, "learning_rate": 8.659802672341078e-07, "loss": 0.08896069526672364, "step": 33050 }, { "epoch": 0.8708507983128254, "grad_norm": 2.433976411819458, "learning_rate": 8.625108217506472e-07, "loss": 0.07890144586563111, "step": 33060 }, { "epoch": 0.8711142135573241, "grad_norm": 3.8501663208007812, "learning_rate": 8.590480269260337e-07, "loss": 0.08671849966049194, "step": 33070 }, { "epoch": 0.8713776288018228, "grad_norm": 3.1162827014923096, "learning_rate": 8.555918852806421e-07, "loss": 0.08594321012496949, "step": 33080 }, { "epoch": 0.8716410440463216, "grad_norm": 2.461860179901123, "learning_rate": 8.521423993300005e-07, "loss": 0.09044568538665772, "step": 33090 }, { "epoch": 0.8719044592908203, "grad_norm": 3.3381519317626953, "learning_rate": 8.486995715847923e-07, "loss": 0.09297910332679749, "step": 33100 }, { "epoch": 0.8721678745353191, "grad_norm": 3.1674349308013916, "learning_rate": 8.452634045508557e-07, "loss": 0.0813959002494812, "step": 33110 }, { "epoch": 0.8724312897798178, "grad_norm": 3.134422779083252, "learning_rate": 8.41833900729182e-07, "loss": 0.08712018132209778, "step": 33120 }, { "epoch": 0.8726947050243166, "grad_norm": 3.069056510925293, "learning_rate": 8.384110626159126e-07, "loss": 0.09002451300621032, "step": 33130 }, { "epoch": 0.8729581202688153, "grad_norm": 2.242553949356079, "learning_rate": 8.349948927023388e-07, "loss": 0.07746073007583618, "step": 33140 }, { "epoch": 0.8732215355133139, "grad_norm": 2.8032143115997314, "learning_rate": 8.315853934748908e-07, "loss": 0.09639966487884521, "step": 33150 }, { "epoch": 0.8734849507578127, "grad_norm": 2.6767566204071045, "learning_rate": 8.281825674151533e-07, "loss": 0.08478000164031982, "step": 33160 }, { "epoch": 0.8737483660023114, "grad_norm": 2.281797409057617, "learning_rate": 8.247864169998499e-07, "loss": 0.08280776143074035, "step": 33170 }, { "epoch": 0.8740117812468102, "grad_norm": 2.781633138656616, "learning_rate": 8.213969447008441e-07, "loss": 0.07649248838424683, "step": 33180 }, { "epoch": 0.8742751964913089, "grad_norm": 3.078033924102783, "learning_rate": 8.180141529851404e-07, "loss": 0.0886139452457428, "step": 33190 }, { "epoch": 0.8745386117358077, "grad_norm": 2.9448914527893066, "learning_rate": 8.146380443148794e-07, "loss": 0.08840880990028381, "step": 33200 }, { "epoch": 0.8748020269803064, "grad_norm": 3.7573511600494385, "learning_rate": 8.11268621147342e-07, "loss": 0.08516093492507934, "step": 33210 }, { "epoch": 0.8750654422248052, "grad_norm": 2.684739828109741, "learning_rate": 8.079058859349364e-07, "loss": 0.0878604531288147, "step": 33220 }, { "epoch": 0.8753288574693039, "grad_norm": 3.5581090450286865, "learning_rate": 8.045498411252095e-07, "loss": 0.09067517518997192, "step": 33230 }, { "epoch": 0.8755922727138026, "grad_norm": 3.2967748641967773, "learning_rate": 8.012004891608327e-07, "loss": 0.09628018140792846, "step": 33240 }, { "epoch": 0.8758556879583014, "grad_norm": 2.8928306102752686, "learning_rate": 7.978578324796072e-07, "loss": 0.08442550301551818, "step": 33250 }, { "epoch": 0.8761191032028001, "grad_norm": 2.5066611766815186, "learning_rate": 7.945218735144689e-07, "loss": 0.08676971197128296, "step": 33260 }, { "epoch": 0.8763825184472989, "grad_norm": 2.690941333770752, "learning_rate": 7.91192614693469e-07, "loss": 0.08213620185852051, "step": 33270 }, { "epoch": 0.8766459336917976, "grad_norm": 3.846646308898926, "learning_rate": 7.878700584397836e-07, "loss": 0.08713524341583252, "step": 33280 }, { "epoch": 0.8769093489362964, "grad_norm": 3.7962002754211426, "learning_rate": 7.84554207171716e-07, "loss": 0.0740441083908081, "step": 33290 }, { "epoch": 0.877172764180795, "grad_norm": 3.0979597568511963, "learning_rate": 7.812450633026847e-07, "loss": 0.08278085589408875, "step": 33300 }, { "epoch": 0.8774361794252938, "grad_norm": 3.3459603786468506, "learning_rate": 7.779426292412273e-07, "loss": 0.07940917015075684, "step": 33310 }, { "epoch": 0.8776995946697925, "grad_norm": 2.334589958190918, "learning_rate": 7.746469073909979e-07, "loss": 0.07651271224021912, "step": 33320 }, { "epoch": 0.8779630099142912, "grad_norm": 2.876112699508667, "learning_rate": 7.713579001507632e-07, "loss": 0.09641004800796509, "step": 33330 }, { "epoch": 0.87822642515879, "grad_norm": 2.5397872924804688, "learning_rate": 7.680756099144072e-07, "loss": 0.0870010495185852, "step": 33340 }, { "epoch": 0.8784898404032887, "grad_norm": 3.396958827972412, "learning_rate": 7.648000390709221e-07, "loss": 0.09644205570220947, "step": 33350 }, { "epoch": 0.8787532556477875, "grad_norm": 2.8059639930725098, "learning_rate": 7.615311900044087e-07, "loss": 0.07516561150550842, "step": 33360 }, { "epoch": 0.8790166708922862, "grad_norm": 3.3274779319763184, "learning_rate": 7.582690650940771e-07, "loss": 0.0820818543434143, "step": 33370 }, { "epoch": 0.879280086136785, "grad_norm": 2.875657081604004, "learning_rate": 7.550136667142393e-07, "loss": 0.08755146265029908, "step": 33380 }, { "epoch": 0.8795435013812837, "grad_norm": 3.144047498703003, "learning_rate": 7.517649972343199e-07, "loss": 0.08334254622459411, "step": 33390 }, { "epoch": 0.8798069166257825, "grad_norm": 2.4048049449920654, "learning_rate": 7.485230590188397e-07, "loss": 0.07335070371627808, "step": 33400 }, { "epoch": 0.8800703318702812, "grad_norm": 2.992647647857666, "learning_rate": 7.452878544274167e-07, "loss": 0.08995810747146607, "step": 33410 }, { "epoch": 0.8803337471147799, "grad_norm": 2.754671812057495, "learning_rate": 7.420593858147773e-07, "loss": 0.08626447916030884, "step": 33420 }, { "epoch": 0.8805971623592787, "grad_norm": 3.5021579265594482, "learning_rate": 7.388376555307386e-07, "loss": 0.09876896142959594, "step": 33430 }, { "epoch": 0.8808605776037773, "grad_norm": 3.036665201187134, "learning_rate": 7.356226659202137e-07, "loss": 0.08726202249526978, "step": 33440 }, { "epoch": 0.8811239928482761, "grad_norm": 2.666141986846924, "learning_rate": 7.324144193232119e-07, "loss": 0.07788995504379273, "step": 33450 }, { "epoch": 0.8813874080927748, "grad_norm": 3.184176445007324, "learning_rate": 7.29212918074833e-07, "loss": 0.07752854824066162, "step": 33460 }, { "epoch": 0.8816508233372736, "grad_norm": 2.7578272819519043, "learning_rate": 7.260181645052689e-07, "loss": 0.08853319883346558, "step": 33470 }, { "epoch": 0.8819142385817723, "grad_norm": 2.8890461921691895, "learning_rate": 7.228301609397981e-07, "loss": 0.08927289843559265, "step": 33480 }, { "epoch": 0.8821776538262711, "grad_norm": 3.0644898414611816, "learning_rate": 7.196489096987869e-07, "loss": 0.09086288213729858, "step": 33490 }, { "epoch": 0.8824410690707698, "grad_norm": 2.787330150604248, "learning_rate": 7.164744130976886e-07, "loss": 0.07253751158714294, "step": 33500 }, { "epoch": 0.8827044843152685, "grad_norm": 3.2939209938049316, "learning_rate": 7.133066734470351e-07, "loss": 0.09588460922241211, "step": 33510 }, { "epoch": 0.8829678995597673, "grad_norm": 3.5020885467529297, "learning_rate": 7.101456930524509e-07, "loss": 0.09322354793548585, "step": 33520 }, { "epoch": 0.883231314804266, "grad_norm": 3.1098473072052, "learning_rate": 7.069914742146267e-07, "loss": 0.08596744537353515, "step": 33530 }, { "epoch": 0.8834947300487648, "grad_norm": 3.146132469177246, "learning_rate": 7.038440192293416e-07, "loss": 0.08728258609771729, "step": 33540 }, { "epoch": 0.8837581452932635, "grad_norm": 2.77651309967041, "learning_rate": 7.007033303874511e-07, "loss": 0.08041763901710511, "step": 33550 }, { "epoch": 0.8840215605377623, "grad_norm": 2.3987905979156494, "learning_rate": 6.97569409974882e-07, "loss": 0.09235653877258301, "step": 33560 }, { "epoch": 0.884284975782261, "grad_norm": 2.7175116539001465, "learning_rate": 6.944422602726375e-07, "loss": 0.0793921709060669, "step": 33570 }, { "epoch": 0.8845483910267598, "grad_norm": 2.9221811294555664, "learning_rate": 6.913218835567925e-07, "loss": 0.09670501351356506, "step": 33580 }, { "epoch": 0.8848118062712584, "grad_norm": 3.564230442047119, "learning_rate": 6.882082820984892e-07, "loss": 0.08537364602088929, "step": 33590 }, { "epoch": 0.8850752215157571, "grad_norm": 2.39494252204895, "learning_rate": 6.851014581639437e-07, "loss": 0.08222534060478211, "step": 33600 }, { "epoch": 0.8853386367602559, "grad_norm": 3.1132254600524902, "learning_rate": 6.820014140144382e-07, "loss": 0.08291353583335877, "step": 33610 }, { "epoch": 0.8856020520047546, "grad_norm": 3.306553840637207, "learning_rate": 6.789081519063134e-07, "loss": 0.08937951326370239, "step": 33620 }, { "epoch": 0.8858654672492534, "grad_norm": 2.9116692543029785, "learning_rate": 6.758216740909829e-07, "loss": 0.08971294164657592, "step": 33630 }, { "epoch": 0.8861288824937521, "grad_norm": 3.232734203338623, "learning_rate": 6.727419828149173e-07, "loss": 0.09107747077941894, "step": 33640 }, { "epoch": 0.8863922977382509, "grad_norm": 3.4487111568450928, "learning_rate": 6.696690803196482e-07, "loss": 0.08913617134094239, "step": 33650 }, { "epoch": 0.8866557129827496, "grad_norm": 2.7919974327087402, "learning_rate": 6.666029688417674e-07, "loss": 0.08768656253814697, "step": 33660 }, { "epoch": 0.8869191282272483, "grad_norm": 2.3810722827911377, "learning_rate": 6.635436506129223e-07, "loss": 0.08093607425689697, "step": 33670 }, { "epoch": 0.8871825434717471, "grad_norm": 3.6794416904449463, "learning_rate": 6.60491127859818e-07, "loss": 0.07881078720092774, "step": 33680 }, { "epoch": 0.8874459587162458, "grad_norm": 2.933333396911621, "learning_rate": 6.57445402804211e-07, "loss": 0.08011915683746337, "step": 33690 }, { "epoch": 0.8877093739607446, "grad_norm": 3.9122297763824463, "learning_rate": 6.544064776629123e-07, "loss": 0.08153476119041443, "step": 33700 }, { "epoch": 0.8879727892052433, "grad_norm": 2.8276820182800293, "learning_rate": 6.513743546477824e-07, "loss": 0.07481906414031983, "step": 33710 }, { "epoch": 0.888236204449742, "grad_norm": 3.0341475009918213, "learning_rate": 6.483490359657296e-07, "loss": 0.08863773345947265, "step": 33720 }, { "epoch": 0.8884996196942407, "grad_norm": 2.5881025791168213, "learning_rate": 6.453305238187147e-07, "loss": 0.06797686815261841, "step": 33730 }, { "epoch": 0.8887630349387395, "grad_norm": 3.020376205444336, "learning_rate": 6.423188204037412e-07, "loss": 0.08453619480133057, "step": 33740 }, { "epoch": 0.8890264501832382, "grad_norm": 2.7573869228363037, "learning_rate": 6.393139279128513e-07, "loss": 0.08108209371566773, "step": 33750 }, { "epoch": 0.8892898654277369, "grad_norm": 2.8041722774505615, "learning_rate": 6.3631584853314e-07, "loss": 0.0907333493232727, "step": 33760 }, { "epoch": 0.8895532806722357, "grad_norm": 3.8382208347320557, "learning_rate": 6.333245844467384e-07, "loss": 0.08969994783401489, "step": 33770 }, { "epoch": 0.8898166959167344, "grad_norm": 2.4679818153381348, "learning_rate": 6.303401378308172e-07, "loss": 0.0769578218460083, "step": 33780 }, { "epoch": 0.8900801111612332, "grad_norm": 2.752073049545288, "learning_rate": 6.273625108575843e-07, "loss": 0.07736952900886536, "step": 33790 }, { "epoch": 0.8903435264057319, "grad_norm": 2.7410783767700195, "learning_rate": 6.243917056942849e-07, "loss": 0.08307559490203857, "step": 33800 }, { "epoch": 0.8906069416502307, "grad_norm": 3.41556978225708, "learning_rate": 6.214277245032019e-07, "loss": 0.08365075588226319, "step": 33810 }, { "epoch": 0.8908703568947294, "grad_norm": 3.2725491523742676, "learning_rate": 6.184705694416471e-07, "loss": 0.0914037585258484, "step": 33820 }, { "epoch": 0.8911337721392282, "grad_norm": 3.2522621154785156, "learning_rate": 6.155202426619655e-07, "loss": 0.08546156883239746, "step": 33830 }, { "epoch": 0.8913971873837269, "grad_norm": 2.7522754669189453, "learning_rate": 6.125767463115317e-07, "loss": 0.08739117980003357, "step": 33840 }, { "epoch": 0.8916606026282256, "grad_norm": 3.6819605827331543, "learning_rate": 6.096400825327487e-07, "loss": 0.08961586952209473, "step": 33850 }, { "epoch": 0.8919240178727244, "grad_norm": 3.2891170978546143, "learning_rate": 6.067102534630498e-07, "loss": 0.08795164823532105, "step": 33860 }, { "epoch": 0.892187433117223, "grad_norm": 3.2816922664642334, "learning_rate": 6.037872612348905e-07, "loss": 0.08558709025382996, "step": 33870 }, { "epoch": 0.8924508483617218, "grad_norm": 2.7946255207061768, "learning_rate": 6.008711079757467e-07, "loss": 0.08377814292907715, "step": 33880 }, { "epoch": 0.8927142636062205, "grad_norm": 3.360410451889038, "learning_rate": 5.979617958081241e-07, "loss": 0.07692353129386902, "step": 33890 }, { "epoch": 0.8929776788507193, "grad_norm": 2.481738328933716, "learning_rate": 5.950593268495464e-07, "loss": 0.07357560396194458, "step": 33900 }, { "epoch": 0.893241094095218, "grad_norm": 2.625570774078369, "learning_rate": 5.921637032125538e-07, "loss": 0.08008660078048706, "step": 33910 }, { "epoch": 0.8935045093397168, "grad_norm": 3.133161783218384, "learning_rate": 5.89274927004706e-07, "loss": 0.08599036931991577, "step": 33920 }, { "epoch": 0.8937679245842155, "grad_norm": 2.8041534423828125, "learning_rate": 5.863930003285779e-07, "loss": 0.07634978890419006, "step": 33930 }, { "epoch": 0.8940313398287142, "grad_norm": 3.692877769470215, "learning_rate": 5.835179252817635e-07, "loss": 0.08418000936508178, "step": 33940 }, { "epoch": 0.894294755073213, "grad_norm": 2.951289176940918, "learning_rate": 5.806497039568648e-07, "loss": 0.0797906756401062, "step": 33950 }, { "epoch": 0.8945581703177117, "grad_norm": 2.6239917278289795, "learning_rate": 5.777883384414973e-07, "loss": 0.08445168733596801, "step": 33960 }, { "epoch": 0.8948215855622105, "grad_norm": 2.527585029602051, "learning_rate": 5.749338308182851e-07, "loss": 0.07630161643028259, "step": 33970 }, { "epoch": 0.8950850008067092, "grad_norm": 3.1803040504455566, "learning_rate": 5.720861831648617e-07, "loss": 0.09056280255317688, "step": 33980 }, { "epoch": 0.895348416051208, "grad_norm": 3.0964272022247314, "learning_rate": 5.692453975538703e-07, "loss": 0.08000656962394714, "step": 33990 }, { "epoch": 0.8956118312957067, "grad_norm": 2.457335948944092, "learning_rate": 5.664114760529571e-07, "loss": 0.07810680866241455, "step": 34000 }, { "epoch": 0.8958752465402054, "grad_norm": 3.132589340209961, "learning_rate": 5.635844207247687e-07, "loss": 0.08018959760665893, "step": 34010 }, { "epoch": 0.8961386617847041, "grad_norm": 2.9026055335998535, "learning_rate": 5.607642336269615e-07, "loss": 0.07397165298461914, "step": 34020 }, { "epoch": 0.8964020770292028, "grad_norm": 3.61257004737854, "learning_rate": 5.579509168121888e-07, "loss": 0.07792788743972778, "step": 34030 }, { "epoch": 0.8966654922737016, "grad_norm": 2.541200876235962, "learning_rate": 5.551444723281029e-07, "loss": 0.07085214853286743, "step": 34040 }, { "epoch": 0.8969289075182003, "grad_norm": 3.256208896636963, "learning_rate": 5.523449022173566e-07, "loss": 0.10056209564208984, "step": 34050 }, { "epoch": 0.8971923227626991, "grad_norm": 2.391845703125, "learning_rate": 5.495522085175942e-07, "loss": 0.08055959939956665, "step": 34060 }, { "epoch": 0.8974557380071978, "grad_norm": 3.4492390155792236, "learning_rate": 5.467663932614642e-07, "loss": 0.07581804394721985, "step": 34070 }, { "epoch": 0.8977191532516966, "grad_norm": 2.77935528755188, "learning_rate": 5.439874584766014e-07, "loss": 0.08346217274665832, "step": 34080 }, { "epoch": 0.8979825684961953, "grad_norm": 2.772278070449829, "learning_rate": 5.412154061856301e-07, "loss": 0.06946210265159607, "step": 34090 }, { "epoch": 0.8982459837406941, "grad_norm": 2.5098533630371094, "learning_rate": 5.38450238406174e-07, "loss": 0.08919641971588135, "step": 34100 }, { "epoch": 0.8985093989851928, "grad_norm": 3.1419103145599365, "learning_rate": 5.356919571508401e-07, "loss": 0.09015671610832214, "step": 34110 }, { "epoch": 0.8987728142296915, "grad_norm": 3.18878173828125, "learning_rate": 5.329405644272268e-07, "loss": 0.08499984741210938, "step": 34120 }, { "epoch": 0.8990362294741903, "grad_norm": 3.5465776920318604, "learning_rate": 5.301960622379132e-07, "loss": 0.08590351343154908, "step": 34130 }, { "epoch": 0.899299644718689, "grad_norm": 3.6779630184173584, "learning_rate": 5.27458452580466e-07, "loss": 0.08827905058860779, "step": 34140 }, { "epoch": 0.8995630599631877, "grad_norm": 2.984116315841675, "learning_rate": 5.24727737447438e-07, "loss": 0.08496482372283935, "step": 34150 }, { "epoch": 0.8998264752076864, "grad_norm": 2.8965799808502197, "learning_rate": 5.220039188263615e-07, "loss": 0.08636751770973206, "step": 34160 }, { "epoch": 0.9000898904521852, "grad_norm": 2.5290913581848145, "learning_rate": 5.192869986997484e-07, "loss": 0.08264815211296081, "step": 34170 }, { "epoch": 0.9003533056966839, "grad_norm": 2.9389967918395996, "learning_rate": 5.165769790450903e-07, "loss": 0.08412392139434814, "step": 34180 }, { "epoch": 0.9006167209411826, "grad_norm": 2.7227704524993896, "learning_rate": 5.138738618348548e-07, "loss": 0.07915840148925782, "step": 34190 }, { "epoch": 0.9008801361856814, "grad_norm": 3.1728675365448, "learning_rate": 5.111776490364906e-07, "loss": 0.09402227401733398, "step": 34200 }, { "epoch": 0.9011435514301801, "grad_norm": 3.379725456237793, "learning_rate": 5.084883426124165e-07, "loss": 0.05978962779045105, "step": 34210 }, { "epoch": 0.9014069666746789, "grad_norm": 2.853121280670166, "learning_rate": 5.058059445200237e-07, "loss": 0.08197707533836365, "step": 34220 }, { "epoch": 0.9016703819191776, "grad_norm": 2.088646650314331, "learning_rate": 5.031304567116802e-07, "loss": 0.09373457431793213, "step": 34230 }, { "epoch": 0.9019337971636764, "grad_norm": 4.495093822479248, "learning_rate": 5.004618811347195e-07, "loss": 0.09369348287582398, "step": 34240 }, { "epoch": 0.9021972124081751, "grad_norm": 2.9299745559692383, "learning_rate": 4.978002197314469e-07, "loss": 0.08506298065185547, "step": 34250 }, { "epoch": 0.9024606276526739, "grad_norm": 2.909101724624634, "learning_rate": 4.951454744391338e-07, "loss": 0.08683397173881531, "step": 34260 }, { "epoch": 0.9027240428971726, "grad_norm": 4.071627140045166, "learning_rate": 4.924976471900178e-07, "loss": 0.07268698811531067, "step": 34270 }, { "epoch": 0.9029874581416713, "grad_norm": 4.837248802185059, "learning_rate": 4.89856739911303e-07, "loss": 0.08231371641159058, "step": 34280 }, { "epoch": 0.90325087338617, "grad_norm": 2.835005760192871, "learning_rate": 4.872227545251562e-07, "loss": 0.08061109781265259, "step": 34290 }, { "epoch": 0.9035142886306687, "grad_norm": 3.2916836738586426, "learning_rate": 4.84595692948705e-07, "loss": 0.07663966417312622, "step": 34300 }, { "epoch": 0.9037777038751675, "grad_norm": 3.5217995643615723, "learning_rate": 4.819755570940377e-07, "loss": 0.08970426321029663, "step": 34310 }, { "epoch": 0.9040411191196662, "grad_norm": 2.5489604473114014, "learning_rate": 4.793623488682008e-07, "loss": 0.08374741077423095, "step": 34320 }, { "epoch": 0.904304534364165, "grad_norm": 3.715205669403076, "learning_rate": 4.7675607017320323e-07, "loss": 0.08321235179901124, "step": 34330 }, { "epoch": 0.9045679496086637, "grad_norm": 4.6329874992370605, "learning_rate": 4.741567229060062e-07, "loss": 0.07811980843544006, "step": 34340 }, { "epoch": 0.9048313648531625, "grad_norm": 3.5473499298095703, "learning_rate": 4.715643089585242e-07, "loss": 0.08798832893371582, "step": 34350 }, { "epoch": 0.9050947800976612, "grad_norm": 2.634737968444824, "learning_rate": 4.6897883021763013e-07, "loss": 0.07634804248809815, "step": 34360 }, { "epoch": 0.9053581953421599, "grad_norm": 3.4456276893615723, "learning_rate": 4.664002885651475e-07, "loss": 0.08323029279708863, "step": 34370 }, { "epoch": 0.9056216105866587, "grad_norm": 4.768863677978516, "learning_rate": 4.638286858778496e-07, "loss": 0.08934440016746521, "step": 34380 }, { "epoch": 0.9058850258311574, "grad_norm": 2.7553205490112305, "learning_rate": 4.612640240274591e-07, "loss": 0.08833858966827393, "step": 34390 }, { "epoch": 0.9061484410756562, "grad_norm": 3.0182816982269287, "learning_rate": 4.587063048806462e-07, "loss": 0.08840490579605102, "step": 34400 }, { "epoch": 0.9064118563201549, "grad_norm": 2.5007545948028564, "learning_rate": 4.561555302990328e-07, "loss": 0.09500161409378052, "step": 34410 }, { "epoch": 0.9066752715646537, "grad_norm": 3.0879483222961426, "learning_rate": 4.536117021391806e-07, "loss": 0.09044214487075805, "step": 34420 }, { "epoch": 0.9069386868091524, "grad_norm": 2.3675100803375244, "learning_rate": 4.5107482225259735e-07, "loss": 0.08319075107574463, "step": 34430 }, { "epoch": 0.9072021020536511, "grad_norm": 2.995887279510498, "learning_rate": 4.4854489248573494e-07, "loss": 0.0800423264503479, "step": 34440 }, { "epoch": 0.9074655172981498, "grad_norm": 4.095870494842529, "learning_rate": 4.460219146799838e-07, "loss": 0.07161707282066346, "step": 34450 }, { "epoch": 0.9077289325426485, "grad_norm": 3.194533586502075, "learning_rate": 4.4350589067167717e-07, "loss": 0.07714236974716186, "step": 34460 }, { "epoch": 0.9079923477871473, "grad_norm": 3.780094623565674, "learning_rate": 4.409968222920868e-07, "loss": 0.09492408037185669, "step": 34470 }, { "epoch": 0.908255763031646, "grad_norm": 3.299259901046753, "learning_rate": 4.384947113674165e-07, "loss": 0.08574960827827453, "step": 34480 }, { "epoch": 0.9085191782761448, "grad_norm": 3.151876211166382, "learning_rate": 4.35999559718816e-07, "loss": 0.07740657329559326, "step": 34490 }, { "epoch": 0.9087825935206435, "grad_norm": 2.830949306488037, "learning_rate": 4.3351136916236156e-07, "loss": 0.08408417701721191, "step": 34500 }, { "epoch": 0.9090460087651423, "grad_norm": 3.3693463802337646, "learning_rate": 4.3103014150906675e-07, "loss": 0.07756191492080688, "step": 34510 }, { "epoch": 0.909309424009641, "grad_norm": 3.2284927368164062, "learning_rate": 4.285558785648747e-07, "loss": 0.09111039638519287, "step": 34520 }, { "epoch": 0.9095728392541398, "grad_norm": 2.8843977451324463, "learning_rate": 4.260885821306604e-07, "loss": 0.07495027780532837, "step": 34530 }, { "epoch": 0.9098362544986385, "grad_norm": 2.9775350093841553, "learning_rate": 4.236282540022307e-07, "loss": 0.08439353108406067, "step": 34540 }, { "epoch": 0.9100996697431372, "grad_norm": 2.970065116882324, "learning_rate": 4.2117489597031767e-07, "loss": 0.09068306684494018, "step": 34550 }, { "epoch": 0.910363084987636, "grad_norm": 4.231962203979492, "learning_rate": 4.1872850982058067e-07, "loss": 0.08502748608589172, "step": 34560 }, { "epoch": 0.9106265002321347, "grad_norm": 3.1401355266571045, "learning_rate": 4.1628909733360534e-07, "loss": 0.0797230362892151, "step": 34570 }, { "epoch": 0.9108899154766334, "grad_norm": 3.0751259326934814, "learning_rate": 4.1385666028490036e-07, "loss": 0.09498467445373535, "step": 34580 }, { "epoch": 0.9111533307211321, "grad_norm": 3.887324810028076, "learning_rate": 4.114312004449006e-07, "loss": 0.0782845675945282, "step": 34590 }, { "epoch": 0.9114167459656309, "grad_norm": 3.3501195907592773, "learning_rate": 4.090127195789606e-07, "loss": 0.08011427521705627, "step": 34600 }, { "epoch": 0.9116801612101296, "grad_norm": 3.3038244247436523, "learning_rate": 4.066012194473512e-07, "loss": 0.08885693550109863, "step": 34610 }, { "epoch": 0.9119435764546284, "grad_norm": 2.869598388671875, "learning_rate": 4.041967018052706e-07, "loss": 0.08450930714607238, "step": 34620 }, { "epoch": 0.9122069916991271, "grad_norm": 2.9801859855651855, "learning_rate": 4.0179916840282776e-07, "loss": 0.0773149847984314, "step": 34630 }, { "epoch": 0.9124704069436258, "grad_norm": 3.7568233013153076, "learning_rate": 3.9940862098505117e-07, "loss": 0.08449708223342896, "step": 34640 }, { "epoch": 0.9127338221881246, "grad_norm": 4.103240966796875, "learning_rate": 3.970250612918858e-07, "loss": 0.09254777431488037, "step": 34650 }, { "epoch": 0.9129972374326233, "grad_norm": 2.5941004753112793, "learning_rate": 3.946484910581871e-07, "loss": 0.0840369164943695, "step": 34660 }, { "epoch": 0.9132606526771221, "grad_norm": 2.5737810134887695, "learning_rate": 3.9227891201372827e-07, "loss": 0.07466940879821778, "step": 34670 }, { "epoch": 0.9135240679216208, "grad_norm": 3.6814124584198, "learning_rate": 3.899163258831895e-07, "loss": 0.08560547828674317, "step": 34680 }, { "epoch": 0.9137874831661196, "grad_norm": 2.593897819519043, "learning_rate": 3.8756073438616203e-07, "loss": 0.08113186359405518, "step": 34690 }, { "epoch": 0.9140508984106183, "grad_norm": 3.38010573387146, "learning_rate": 3.8521213923714775e-07, "loss": 0.09731302857398987, "step": 34700 }, { "epoch": 0.914314313655117, "grad_norm": 2.75032901763916, "learning_rate": 3.8287054214555694e-07, "loss": 0.07562723159790039, "step": 34710 }, { "epoch": 0.9145777288996157, "grad_norm": 2.6390507221221924, "learning_rate": 3.805359448157031e-07, "loss": 0.08483003377914429, "step": 34720 }, { "epoch": 0.9148411441441144, "grad_norm": 3.0011744499206543, "learning_rate": 3.78208348946808e-07, "loss": 0.0740580439567566, "step": 34730 }, { "epoch": 0.9151045593886132, "grad_norm": 2.976027250289917, "learning_rate": 3.7588775623299433e-07, "loss": 0.07825124859809876, "step": 34740 }, { "epoch": 0.9153679746331119, "grad_norm": 3.6837470531463623, "learning_rate": 3.735741683632932e-07, "loss": 0.08064918518066407, "step": 34750 }, { "epoch": 0.9156313898776107, "grad_norm": 2.065453290939331, "learning_rate": 3.712675870216309e-07, "loss": 0.07743239402770996, "step": 34760 }, { "epoch": 0.9158948051221094, "grad_norm": 3.443582773208618, "learning_rate": 3.689680138868379e-07, "loss": 0.07999435663223267, "step": 34770 }, { "epoch": 0.9161582203666082, "grad_norm": 2.838343381881714, "learning_rate": 3.66675450632642e-07, "loss": 0.07394101619720458, "step": 34780 }, { "epoch": 0.9164216356111069, "grad_norm": 3.6612930297851562, "learning_rate": 3.643898989276684e-07, "loss": 0.09041569232940674, "step": 34790 }, { "epoch": 0.9166850508556056, "grad_norm": 3.4560117721557617, "learning_rate": 3.6211136043544425e-07, "loss": 0.0816217303276062, "step": 34800 }, { "epoch": 0.9169484661001044, "grad_norm": 3.692101240158081, "learning_rate": 3.598398368143863e-07, "loss": 0.09841799139976501, "step": 34810 }, { "epoch": 0.9172118813446031, "grad_norm": 2.316560745239258, "learning_rate": 3.575753297178053e-07, "loss": 0.07427357435226441, "step": 34820 }, { "epoch": 0.9174752965891019, "grad_norm": 2.1827151775360107, "learning_rate": 3.553178407939095e-07, "loss": 0.07793729305267334, "step": 34830 }, { "epoch": 0.9177387118336006, "grad_norm": 2.248440742492676, "learning_rate": 3.5306737168579685e-07, "loss": 0.07882912158966064, "step": 34840 }, { "epoch": 0.9180021270780994, "grad_norm": 2.9543967247009277, "learning_rate": 3.5082392403145373e-07, "loss": 0.08337749242782592, "step": 34850 }, { "epoch": 0.918265542322598, "grad_norm": 3.656569242477417, "learning_rate": 3.485874994637606e-07, "loss": 0.08617987632751464, "step": 34860 }, { "epoch": 0.9185289575670968, "grad_norm": 1.8510258197784424, "learning_rate": 3.463580996104821e-07, "loss": 0.08752210140228271, "step": 34870 }, { "epoch": 0.9187923728115955, "grad_norm": 3.6778621673583984, "learning_rate": 3.4413572609427235e-07, "loss": 0.09384311437606811, "step": 34880 }, { "epoch": 0.9190557880560942, "grad_norm": 4.008219242095947, "learning_rate": 3.4192038053267095e-07, "loss": 0.08774001002311707, "step": 34890 }, { "epoch": 0.919319203300593, "grad_norm": 2.2508435249328613, "learning_rate": 3.3971206453810133e-07, "loss": 0.07645740509033203, "step": 34900 }, { "epoch": 0.9195826185450917, "grad_norm": 3.475588083267212, "learning_rate": 3.3751077971787115e-07, "loss": 0.07567692995071411, "step": 34910 }, { "epoch": 0.9198460337895905, "grad_norm": 3.0590109825134277, "learning_rate": 3.353165276741699e-07, "loss": 0.07985925674438477, "step": 34920 }, { "epoch": 0.9201094490340892, "grad_norm": 3.7350475788116455, "learning_rate": 3.331293100040711e-07, "loss": 0.08623093366622925, "step": 34930 }, { "epoch": 0.920372864278588, "grad_norm": 3.599019765853882, "learning_rate": 3.3094912829952477e-07, "loss": 0.08508193492889404, "step": 34940 }, { "epoch": 0.9206362795230867, "grad_norm": 3.436642646789551, "learning_rate": 3.2877598414736033e-07, "loss": 0.07748895287513732, "step": 34950 }, { "epoch": 0.9208996947675855, "grad_norm": 3.7414088249206543, "learning_rate": 3.266098791292871e-07, "loss": 0.08814102411270142, "step": 34960 }, { "epoch": 0.9211631100120842, "grad_norm": 2.7337820529937744, "learning_rate": 3.244508148218895e-07, "loss": 0.06783693432807922, "step": 34970 }, { "epoch": 0.9214265252565829, "grad_norm": 3.2204673290252686, "learning_rate": 3.22298792796627e-07, "loss": 0.07956069707870483, "step": 34980 }, { "epoch": 0.9216899405010817, "grad_norm": 2.8537404537200928, "learning_rate": 3.2015381461983577e-07, "loss": 0.0871503233909607, "step": 34990 }, { "epoch": 0.9219533557455803, "grad_norm": 2.445773124694824, "learning_rate": 3.1801588185272016e-07, "loss": 0.07559666633605958, "step": 35000 }, { "epoch": 0.9222167709900791, "grad_norm": 2.318601131439209, "learning_rate": 3.1588499605136435e-07, "loss": 0.08312547206878662, "step": 35010 }, { "epoch": 0.9224801862345778, "grad_norm": 3.1960971355438232, "learning_rate": 3.1376115876671553e-07, "loss": 0.08110404014587402, "step": 35020 }, { "epoch": 0.9227436014790766, "grad_norm": 2.4968442916870117, "learning_rate": 3.1164437154459494e-07, "loss": 0.07325662970542908, "step": 35030 }, { "epoch": 0.9230070167235753, "grad_norm": 4.785655975341797, "learning_rate": 3.0953463592569343e-07, "loss": 0.07734774351119995, "step": 35040 }, { "epoch": 0.9232704319680741, "grad_norm": 3.50876522064209, "learning_rate": 3.074319534455639e-07, "loss": 0.09267466068267823, "step": 35050 }, { "epoch": 0.9235338472125728, "grad_norm": 2.74977970123291, "learning_rate": 3.053363256346331e-07, "loss": 0.08663660287857056, "step": 35060 }, { "epoch": 0.9237972624570715, "grad_norm": 4.7891926765441895, "learning_rate": 3.0324775401818884e-07, "loss": 0.0738097071647644, "step": 35070 }, { "epoch": 0.9240606777015703, "grad_norm": 3.7174181938171387, "learning_rate": 3.0116624011637953e-07, "loss": 0.0889184057712555, "step": 35080 }, { "epoch": 0.924324092946069, "grad_norm": 3.5469188690185547, "learning_rate": 2.9909178544422656e-07, "loss": 0.0783422827720642, "step": 35090 }, { "epoch": 0.9245875081905678, "grad_norm": 3.5101959705352783, "learning_rate": 2.9702439151160334e-07, "loss": 0.08336013555526733, "step": 35100 }, { "epoch": 0.9248509234350665, "grad_norm": 3.603929042816162, "learning_rate": 2.949640598232517e-07, "loss": 0.07868963479995728, "step": 35110 }, { "epoch": 0.9251143386795653, "grad_norm": 4.3262834548950195, "learning_rate": 2.9291079187876656e-07, "loss": 0.09047509431838989, "step": 35120 }, { "epoch": 0.925377753924064, "grad_norm": 3.2962331771850586, "learning_rate": 2.9086458917260585e-07, "loss": 0.07409002184867859, "step": 35130 }, { "epoch": 0.9256411691685628, "grad_norm": 3.583031177520752, "learning_rate": 2.8882545319408706e-07, "loss": 0.07494138479232788, "step": 35140 }, { "epoch": 0.9259045844130614, "grad_norm": 3.5523126125335693, "learning_rate": 2.8679338542737745e-07, "loss": 0.09071100950241089, "step": 35150 }, { "epoch": 0.9261679996575601, "grad_norm": 5.00605583190918, "learning_rate": 2.8476838735150613e-07, "loss": 0.07615122795104981, "step": 35160 }, { "epoch": 0.9264314149020589, "grad_norm": 4.086753845214844, "learning_rate": 2.8275046044035304e-07, "loss": 0.08188893795013427, "step": 35170 }, { "epoch": 0.9266948301465576, "grad_norm": 2.5618607997894287, "learning_rate": 2.8073960616265215e-07, "loss": 0.09452409744262695, "step": 35180 }, { "epoch": 0.9269582453910564, "grad_norm": 4.112534046173096, "learning_rate": 2.7873582598199056e-07, "loss": 0.08350568413734435, "step": 35190 }, { "epoch": 0.9272216606355551, "grad_norm": 3.4115407466888428, "learning_rate": 2.7673912135680714e-07, "loss": 0.07036318182945252, "step": 35200 }, { "epoch": 0.9274850758800539, "grad_norm": 2.6013436317443848, "learning_rate": 2.7474949374038717e-07, "loss": 0.07776384353637696, "step": 35210 }, { "epoch": 0.9277484911245526, "grad_norm": 2.3910279273986816, "learning_rate": 2.7276694458086894e-07, "loss": 0.08375919461250306, "step": 35220 }, { "epoch": 0.9280119063690514, "grad_norm": 3.501655101776123, "learning_rate": 2.7079147532123704e-07, "loss": 0.08309696912765503, "step": 35230 }, { "epoch": 0.9282753216135501, "grad_norm": 2.413550853729248, "learning_rate": 2.688230873993225e-07, "loss": 0.0622582733631134, "step": 35240 }, { "epoch": 0.9285387368580488, "grad_norm": 3.015953779220581, "learning_rate": 2.6686178224780367e-07, "loss": 0.08221194744110108, "step": 35250 }, { "epoch": 0.9288021521025476, "grad_norm": 3.0519542694091797, "learning_rate": 2.6490756129420313e-07, "loss": 0.08313764333724975, "step": 35260 }, { "epoch": 0.9290655673470463, "grad_norm": 2.8214621543884277, "learning_rate": 2.6296042596088643e-07, "loss": 0.07934315204620361, "step": 35270 }, { "epoch": 0.9293289825915451, "grad_norm": 3.00990629196167, "learning_rate": 2.610203776650655e-07, "loss": 0.07830279469490051, "step": 35280 }, { "epoch": 0.9295923978360437, "grad_norm": 3.5542664527893066, "learning_rate": 2.5908741781878744e-07, "loss": 0.07232429981231689, "step": 35290 }, { "epoch": 0.9298558130805425, "grad_norm": 3.2503645420074463, "learning_rate": 2.5716154782894573e-07, "loss": 0.08090128898620605, "step": 35300 }, { "epoch": 0.9301192283250412, "grad_norm": 3.9143800735473633, "learning_rate": 2.552427690972703e-07, "loss": 0.07269743084907532, "step": 35310 }, { "epoch": 0.9303826435695399, "grad_norm": 3.6442084312438965, "learning_rate": 2.5333108302033394e-07, "loss": 0.08488568663597107, "step": 35320 }, { "epoch": 0.9306460588140387, "grad_norm": 3.366866111755371, "learning_rate": 2.514264909895414e-07, "loss": 0.08343087434768677, "step": 35330 }, { "epoch": 0.9309094740585374, "grad_norm": 2.9351391792297363, "learning_rate": 2.495289943911361e-07, "loss": 0.08709936141967774, "step": 35340 }, { "epoch": 0.9311728893030362, "grad_norm": 3.40505313873291, "learning_rate": 2.476385946062021e-07, "loss": 0.07748746871948242, "step": 35350 }, { "epoch": 0.9314363045475349, "grad_norm": 3.953016757965088, "learning_rate": 2.4575529301065104e-07, "loss": 0.08122819066047668, "step": 35360 }, { "epoch": 0.9316997197920337, "grad_norm": 2.8683505058288574, "learning_rate": 2.43879090975232e-07, "loss": 0.07541593313217163, "step": 35370 }, { "epoch": 0.9319631350365324, "grad_norm": 4.135806560516357, "learning_rate": 2.4200998986552484e-07, "loss": 0.08498759269714355, "step": 35380 }, { "epoch": 0.9322265502810312, "grad_norm": 4.13145112991333, "learning_rate": 2.401479910419413e-07, "loss": 0.08368885517120361, "step": 35390 }, { "epoch": 0.9324899655255299, "grad_norm": 2.9485042095184326, "learning_rate": 2.3829309585972738e-07, "loss": 0.09085201025009156, "step": 35400 }, { "epoch": 0.9327533807700286, "grad_norm": 3.9642374515533447, "learning_rate": 2.364453056689553e-07, "loss": 0.08170690536499023, "step": 35410 }, { "epoch": 0.9330167960145274, "grad_norm": 2.9607913494110107, "learning_rate": 2.3460462181452258e-07, "loss": 0.08204178810119629, "step": 35420 }, { "epoch": 0.933280211259026, "grad_norm": 2.706430196762085, "learning_rate": 2.3277104563616205e-07, "loss": 0.08400626182556152, "step": 35430 }, { "epoch": 0.9335436265035248, "grad_norm": 3.235340118408203, "learning_rate": 2.3094457846842943e-07, "loss": 0.07907242774963379, "step": 35440 }, { "epoch": 0.9338070417480235, "grad_norm": 3.937851667404175, "learning_rate": 2.291252216407047e-07, "loss": 0.06949504017829895, "step": 35450 }, { "epoch": 0.9340704569925223, "grad_norm": 3.1254231929779053, "learning_rate": 2.273129764771953e-07, "loss": 0.07254213690757752, "step": 35460 }, { "epoch": 0.934333872237021, "grad_norm": 3.157832622528076, "learning_rate": 2.2550784429693052e-07, "loss": 0.0800349235534668, "step": 35470 }, { "epoch": 0.9345972874815198, "grad_norm": 2.955381155014038, "learning_rate": 2.2370982641376494e-07, "loss": 0.06383780241012574, "step": 35480 }, { "epoch": 0.9348607027260185, "grad_norm": 4.692790985107422, "learning_rate": 2.2191892413637173e-07, "loss": 0.08500412702560425, "step": 35490 }, { "epoch": 0.9351241179705172, "grad_norm": 4.18523645401001, "learning_rate": 2.2013513876824933e-07, "loss": 0.07956756353378296, "step": 35500 }, { "epoch": 0.935387533215016, "grad_norm": 2.8149187564849854, "learning_rate": 2.183584716077114e-07, "loss": 0.07912111878395081, "step": 35510 }, { "epoch": 0.9356509484595147, "grad_norm": 3.1418068408966064, "learning_rate": 2.1658892394789244e-07, "loss": 0.07894062995910645, "step": 35520 }, { "epoch": 0.9359143637040135, "grad_norm": 3.366915702819824, "learning_rate": 2.1482649707674775e-07, "loss": 0.06938014030456544, "step": 35530 }, { "epoch": 0.9361777789485122, "grad_norm": 2.676084518432617, "learning_rate": 2.13071192277049e-07, "loss": 0.07317404747009278, "step": 35540 }, { "epoch": 0.936441194193011, "grad_norm": 2.8033039569854736, "learning_rate": 2.1132301082637752e-07, "loss": 0.07515240907669067, "step": 35550 }, { "epoch": 0.9367046094375097, "grad_norm": 2.5725975036621094, "learning_rate": 2.0958195399713888e-07, "loss": 0.08555128574371337, "step": 35560 }, { "epoch": 0.9369680246820085, "grad_norm": 2.7338767051696777, "learning_rate": 2.0784802305654938e-07, "loss": 0.07074893712997436, "step": 35570 }, { "epoch": 0.9372314399265071, "grad_norm": 4.528617858886719, "learning_rate": 2.061212192666373e-07, "loss": 0.08275331258773803, "step": 35580 }, { "epoch": 0.9374948551710058, "grad_norm": 2.9656572341918945, "learning_rate": 2.0440154388424616e-07, "loss": 0.07426000833511352, "step": 35590 }, { "epoch": 0.9377582704155046, "grad_norm": 2.4706530570983887, "learning_rate": 2.0268899816102916e-07, "loss": 0.0767543375492096, "step": 35600 }, { "epoch": 0.9380216856600033, "grad_norm": 3.607004404067993, "learning_rate": 2.0098358334345147e-07, "loss": 0.07651098370552063, "step": 35610 }, { "epoch": 0.9382851009045021, "grad_norm": 3.915017604827881, "learning_rate": 1.9928530067278684e-07, "loss": 0.07720277905464172, "step": 35620 }, { "epoch": 0.9385485161490008, "grad_norm": 2.4975039958953857, "learning_rate": 1.9759415138511984e-07, "loss": 0.07928810715675354, "step": 35630 }, { "epoch": 0.9388119313934996, "grad_norm": 3.5872607231140137, "learning_rate": 1.959101367113414e-07, "loss": 0.08782591819763183, "step": 35640 }, { "epoch": 0.9390753466379983, "grad_norm": 2.9706006050109863, "learning_rate": 1.9423325787714774e-07, "loss": 0.08240129947662353, "step": 35650 }, { "epoch": 0.9393387618824971, "grad_norm": 2.965994119644165, "learning_rate": 1.9256351610304814e-07, "loss": 0.08036993741989136, "step": 35660 }, { "epoch": 0.9396021771269958, "grad_norm": 2.2217862606048584, "learning_rate": 1.909009126043504e-07, "loss": 0.06960595846176147, "step": 35670 }, { "epoch": 0.9398655923714945, "grad_norm": 3.4050607681274414, "learning_rate": 1.8924544859116656e-07, "loss": 0.08491497635841369, "step": 35680 }, { "epoch": 0.9401290076159933, "grad_norm": 3.582850933074951, "learning_rate": 1.8759712526841943e-07, "loss": 0.08921940326690674, "step": 35690 }, { "epoch": 0.940392422860492, "grad_norm": 3.5856471061706543, "learning_rate": 1.8595594383582605e-07, "loss": 0.08938060998916626, "step": 35700 }, { "epoch": 0.9406558381049908, "grad_norm": 3.0847437381744385, "learning_rate": 1.8432190548791197e-07, "loss": 0.08968163728713989, "step": 35710 }, { "epoch": 0.9409192533494894, "grad_norm": 2.349480390548706, "learning_rate": 1.8269501141399915e-07, "loss": 0.07612475156784057, "step": 35720 }, { "epoch": 0.9411826685939882, "grad_norm": 3.2389116287231445, "learning_rate": 1.8107526279821041e-07, "loss": 0.08071170449256897, "step": 35730 }, { "epoch": 0.9414460838384869, "grad_norm": 2.2877414226531982, "learning_rate": 1.794626608194705e-07, "loss": 0.08866195678710938, "step": 35740 }, { "epoch": 0.9417094990829857, "grad_norm": 2.613938808441162, "learning_rate": 1.7785720665150054e-07, "loss": 0.08242596387863159, "step": 35750 }, { "epoch": 0.9419729143274844, "grad_norm": 3.9819695949554443, "learning_rate": 1.7625890146281798e-07, "loss": 0.08412038087844849, "step": 35760 }, { "epoch": 0.9422363295719831, "grad_norm": 2.7375199794769287, "learning_rate": 1.7466774641674012e-07, "loss": 0.06916233301162719, "step": 35770 }, { "epoch": 0.9424997448164819, "grad_norm": 3.434941291809082, "learning_rate": 1.7308374267137606e-07, "loss": 0.07885555028915406, "step": 35780 }, { "epoch": 0.9427631600609806, "grad_norm": 2.524991512298584, "learning_rate": 1.715068913796336e-07, "loss": 0.08663789033889771, "step": 35790 }, { "epoch": 0.9430265753054794, "grad_norm": 3.6918535232543945, "learning_rate": 1.6993719368921468e-07, "loss": 0.08979403972625732, "step": 35800 }, { "epoch": 0.9432899905499781, "grad_norm": 4.595870494842529, "learning_rate": 1.683746507426087e-07, "loss": 0.07836392521858215, "step": 35810 }, { "epoch": 0.9435534057944769, "grad_norm": 3.379242181777954, "learning_rate": 1.66819263677106e-07, "loss": 0.08136743307113647, "step": 35820 }, { "epoch": 0.9438168210389756, "grad_norm": 3.3317744731903076, "learning_rate": 1.6527103362478202e-07, "loss": 0.08300837278366088, "step": 35830 }, { "epoch": 0.9440802362834743, "grad_norm": 2.7947933673858643, "learning_rate": 1.6372996171250544e-07, "loss": 0.09148207902908326, "step": 35840 }, { "epoch": 0.9443436515279731, "grad_norm": 3.210326910018921, "learning_rate": 1.6219604906193677e-07, "loss": 0.07560021877288818, "step": 35850 }, { "epoch": 0.9446070667724717, "grad_norm": 3.0919759273529053, "learning_rate": 1.606692967895207e-07, "loss": 0.0854874610900879, "step": 35860 }, { "epoch": 0.9448704820169705, "grad_norm": 3.476323366165161, "learning_rate": 1.5914970600649615e-07, "loss": 0.09278051853179932, "step": 35870 }, { "epoch": 0.9451338972614692, "grad_norm": 3.3188445568084717, "learning_rate": 1.5763727781888726e-07, "loss": 0.062486135959625246, "step": 35880 }, { "epoch": 0.945397312505968, "grad_norm": 3.0215492248535156, "learning_rate": 1.5613201332750127e-07, "loss": 0.07964460849761963, "step": 35890 }, { "epoch": 0.9456607277504667, "grad_norm": 3.5957584381103516, "learning_rate": 1.5463391362793735e-07, "loss": 0.07604031562805176, "step": 35900 }, { "epoch": 0.9459241429949655, "grad_norm": 3.4259071350097656, "learning_rate": 1.5314297981057658e-07, "loss": 0.0787634789943695, "step": 35910 }, { "epoch": 0.9461875582394642, "grad_norm": 3.652707576751709, "learning_rate": 1.5165921296058427e-07, "loss": 0.08308886885643005, "step": 35920 }, { "epoch": 0.9464509734839629, "grad_norm": 2.1251630783081055, "learning_rate": 1.501826141579099e-07, "loss": 0.06699660420417786, "step": 35930 }, { "epoch": 0.9467143887284617, "grad_norm": 2.535583734512329, "learning_rate": 1.48713184477286e-07, "loss": 0.08439853191375732, "step": 35940 }, { "epoch": 0.9469778039729604, "grad_norm": 4.060014247894287, "learning_rate": 1.4725092498822814e-07, "loss": 0.07825191617012024, "step": 35950 }, { "epoch": 0.9472412192174592, "grad_norm": 2.9363644123077393, "learning_rate": 1.457958367550305e-07, "loss": 0.0805823028087616, "step": 35960 }, { "epoch": 0.9475046344619579, "grad_norm": 2.493426561355591, "learning_rate": 1.443479208367693e-07, "loss": 0.06927481889724732, "step": 35970 }, { "epoch": 0.9477680497064567, "grad_norm": 3.0853662490844727, "learning_rate": 1.4290717828730037e-07, "loss": 0.07386460304260253, "step": 35980 }, { "epoch": 0.9480314649509554, "grad_norm": 3.3300957679748535, "learning_rate": 1.4147361015525717e-07, "loss": 0.08119441866874695, "step": 35990 }, { "epoch": 0.9482948801954542, "grad_norm": 4.1796979904174805, "learning_rate": 1.4004721748405392e-07, "loss": 0.09469932317733765, "step": 36000 }, { "epoch": 0.9485582954399528, "grad_norm": 2.9659616947174072, "learning_rate": 1.3862800131188237e-07, "loss": 0.08346711397171021, "step": 36010 }, { "epoch": 0.9488217106844515, "grad_norm": 3.255330801010132, "learning_rate": 1.37215962671704e-07, "loss": 0.08812445402145386, "step": 36020 }, { "epoch": 0.9490851259289503, "grad_norm": 3.2507567405700684, "learning_rate": 1.3581110259126674e-07, "loss": 0.08273928165435791, "step": 36030 }, { "epoch": 0.949348541173449, "grad_norm": 2.187594175338745, "learning_rate": 1.3441342209308595e-07, "loss": 0.07958150506019593, "step": 36040 }, { "epoch": 0.9496119564179478, "grad_norm": 3.909101724624634, "learning_rate": 1.3302292219445457e-07, "loss": 0.08248295783996581, "step": 36050 }, { "epoch": 0.9498753716624465, "grad_norm": 2.4768636226654053, "learning_rate": 1.3163960390743746e-07, "loss": 0.07507244348526002, "step": 36060 }, { "epoch": 0.9501387869069453, "grad_norm": 4.2101335525512695, "learning_rate": 1.302634682388726e-07, "loss": 0.08072606325149537, "step": 36070 }, { "epoch": 0.950402202151444, "grad_norm": 2.5153541564941406, "learning_rate": 1.2889451619037318e-07, "loss": 0.08055658340454101, "step": 36080 }, { "epoch": 0.9506656173959428, "grad_norm": 3.6651196479797363, "learning_rate": 1.2753274875832112e-07, "loss": 0.07899879217147827, "step": 36090 }, { "epoch": 0.9509290326404415, "grad_norm": 2.706493854522705, "learning_rate": 1.2617816693386907e-07, "loss": 0.073237144947052, "step": 36100 }, { "epoch": 0.9511924478849402, "grad_norm": 3.1319477558135986, "learning_rate": 1.2483077170293954e-07, "loss": 0.08563274145126343, "step": 36110 }, { "epoch": 0.951455863129439, "grad_norm": 3.1964781284332275, "learning_rate": 1.2349056404622472e-07, "loss": 0.06879477500915528, "step": 36120 }, { "epoch": 0.9517192783739377, "grad_norm": 3.5721960067749023, "learning_rate": 1.2215754493918763e-07, "loss": 0.08271613121032714, "step": 36130 }, { "epoch": 0.9519826936184365, "grad_norm": 3.2468454837799072, "learning_rate": 1.208317153520555e-07, "loss": 0.08411765694618226, "step": 36140 }, { "epoch": 0.9522461088629351, "grad_norm": 2.6432578563690186, "learning_rate": 1.195130762498231e-07, "loss": 0.07579632997512817, "step": 36150 }, { "epoch": 0.952509524107434, "grad_norm": 3.326054811477661, "learning_rate": 1.1820162859225493e-07, "loss": 0.07939767837524414, "step": 36160 }, { "epoch": 0.9527729393519326, "grad_norm": 2.0526316165924072, "learning_rate": 1.1689737333387852e-07, "loss": 0.07079216241836547, "step": 36170 }, { "epoch": 0.9530363545964314, "grad_norm": 4.17957067489624, "learning_rate": 1.1560031142398787e-07, "loss": 0.07876816987991334, "step": 36180 }, { "epoch": 0.9532997698409301, "grad_norm": 3.725442409515381, "learning_rate": 1.1431044380663891e-07, "loss": 0.09070421457290649, "step": 36190 }, { "epoch": 0.9535631850854288, "grad_norm": 2.7366597652435303, "learning_rate": 1.1302777142065402e-07, "loss": 0.08014842867851257, "step": 36200 }, { "epoch": 0.9538266003299276, "grad_norm": 3.1129438877105713, "learning_rate": 1.117522951996175e-07, "loss": 0.08476539850234985, "step": 36210 }, { "epoch": 0.9540900155744263, "grad_norm": 3.4977097511291504, "learning_rate": 1.1048401607187565e-07, "loss": 0.07974229454994201, "step": 36220 }, { "epoch": 0.9543534308189251, "grad_norm": 3.0557701587677, "learning_rate": 1.0922293496053671e-07, "loss": 0.09451369047164918, "step": 36230 }, { "epoch": 0.9546168460634238, "grad_norm": 3.7162129878997803, "learning_rate": 1.0796905278347092e-07, "loss": 0.08324459791183472, "step": 36240 }, { "epoch": 0.9548802613079226, "grad_norm": 3.7535510063171387, "learning_rate": 1.0672237045330491e-07, "loss": 0.08393813371658325, "step": 36250 }, { "epoch": 0.9551436765524213, "grad_norm": 3.7519075870513916, "learning_rate": 1.0548288887743174e-07, "loss": 0.07734280228614807, "step": 36260 }, { "epoch": 0.9554070917969201, "grad_norm": 3.8494131565093994, "learning_rate": 1.0425060895799865e-07, "loss": 0.07349618673324584, "step": 36270 }, { "epoch": 0.9556705070414188, "grad_norm": 4.274208068847656, "learning_rate": 1.0302553159190932e-07, "loss": 0.08525245189666748, "step": 36280 }, { "epoch": 0.9559339222859174, "grad_norm": 3.1997056007385254, "learning_rate": 1.0180765767083045e-07, "loss": 0.07654430866241455, "step": 36290 }, { "epoch": 0.9561973375304162, "grad_norm": 2.9393422603607178, "learning_rate": 1.00596988081183e-07, "loss": 0.07420247197151184, "step": 36300 }, { "epoch": 0.9564607527749149, "grad_norm": 3.876213788986206, "learning_rate": 9.939352370414435e-08, "loss": 0.07689143419265747, "step": 36310 }, { "epoch": 0.9567241680194137, "grad_norm": 2.3190653324127197, "learning_rate": 9.819726541564711e-08, "loss": 0.07759740352630615, "step": 36320 }, { "epoch": 0.9569875832639124, "grad_norm": 3.1869349479675293, "learning_rate": 9.700821408638039e-08, "loss": 0.08332625627517701, "step": 36330 }, { "epoch": 0.9572509985084112, "grad_norm": 2.3300840854644775, "learning_rate": 9.582637058178635e-08, "loss": 0.0735745906829834, "step": 36340 }, { "epoch": 0.9575144137529099, "grad_norm": 2.965456008911133, "learning_rate": 9.465173576206355e-08, "loss": 0.07679799795150757, "step": 36350 }, { "epoch": 0.9577778289974086, "grad_norm": 4.271644592285156, "learning_rate": 9.348431048215923e-08, "loss": 0.08464287519454956, "step": 36360 }, { "epoch": 0.9580412442419074, "grad_norm": 3.1043190956115723, "learning_rate": 9.232409559177701e-08, "loss": 0.08798786401748657, "step": 36370 }, { "epoch": 0.9583046594864061, "grad_norm": 3.124770402908325, "learning_rate": 9.117109193537254e-08, "loss": 0.08738539218902588, "step": 36380 }, { "epoch": 0.9585680747309049, "grad_norm": 3.1574716567993164, "learning_rate": 9.002530035215007e-08, "loss": 0.08069019317626953, "step": 36390 }, { "epoch": 0.9588314899754036, "grad_norm": 3.536794424057007, "learning_rate": 8.888672167606693e-08, "loss": 0.07905368804931641, "step": 36400 }, { "epoch": 0.9590949052199024, "grad_norm": 3.580777406692505, "learning_rate": 8.775535673582913e-08, "loss": 0.06812957525253296, "step": 36410 }, { "epoch": 0.9593583204644011, "grad_norm": 2.9038054943084717, "learning_rate": 8.663120635489241e-08, "loss": 0.07751784324645997, "step": 36420 }, { "epoch": 0.9596217357088999, "grad_norm": 3.545236587524414, "learning_rate": 8.55142713514645e-08, "loss": 0.08519411087036133, "step": 36430 }, { "epoch": 0.9598851509533985, "grad_norm": 2.952738046646118, "learning_rate": 8.440455253849622e-08, "loss": 0.07661999464035034, "step": 36440 }, { "epoch": 0.9601485661978972, "grad_norm": 2.692084789276123, "learning_rate": 8.330205072369035e-08, "loss": 0.0825497567653656, "step": 36450 }, { "epoch": 0.960411981442396, "grad_norm": 3.246901750564575, "learning_rate": 8.22067667094939e-08, "loss": 0.09295592308044434, "step": 36460 }, { "epoch": 0.9606753966868947, "grad_norm": 3.106320858001709, "learning_rate": 8.11187012931025e-08, "loss": 0.07901166081428528, "step": 36470 }, { "epoch": 0.9609388119313935, "grad_norm": 3.7284998893737793, "learning_rate": 8.003785526645824e-08, "loss": 0.08314721584320069, "step": 36480 }, { "epoch": 0.9612022271758922, "grad_norm": 3.984807252883911, "learning_rate": 7.896422941624626e-08, "loss": 0.07955197095870972, "step": 36490 }, { "epoch": 0.961465642420391, "grad_norm": 3.2311160564422607, "learning_rate": 7.789782452389705e-08, "loss": 0.08459025621414185, "step": 36500 }, { "epoch": 0.9617290576648897, "grad_norm": 2.7406513690948486, "learning_rate": 7.683864136558749e-08, "loss": 0.07868072986602784, "step": 36510 }, { "epoch": 0.9619924729093885, "grad_norm": 4.703701496124268, "learning_rate": 7.578668071223649e-08, "loss": 0.07841401100158692, "step": 36520 }, { "epoch": 0.9622558881538872, "grad_norm": 2.8782529830932617, "learning_rate": 7.474194332950602e-08, "loss": 0.077199786901474, "step": 36530 }, { "epoch": 0.9625193033983859, "grad_norm": 3.2128305435180664, "learning_rate": 7.370442997780114e-08, "loss": 0.08038398027420043, "step": 36540 }, { "epoch": 0.9627827186428847, "grad_norm": 2.5832183361053467, "learning_rate": 7.267414141227002e-08, "loss": 0.06866539716720581, "step": 36550 }, { "epoch": 0.9630461338873834, "grad_norm": 3.358004093170166, "learning_rate": 7.165107838279838e-08, "loss": 0.08164038062095642, "step": 36560 }, { "epoch": 0.9633095491318822, "grad_norm": 2.77791428565979, "learning_rate": 7.063524163401836e-08, "loss": 0.08126288056373596, "step": 36570 }, { "epoch": 0.9635729643763808, "grad_norm": 3.6106138229370117, "learning_rate": 6.962663190529851e-08, "loss": 0.0809722363948822, "step": 36580 }, { "epoch": 0.9638363796208796, "grad_norm": 3.3295812606811523, "learning_rate": 6.862524993074825e-08, "loss": 0.06950827836990356, "step": 36590 }, { "epoch": 0.9640997948653783, "grad_norm": 4.151965141296387, "learning_rate": 6.763109643921683e-08, "loss": 0.07208040952682496, "step": 36600 }, { "epoch": 0.9643632101098771, "grad_norm": 3.4100072383880615, "learning_rate": 6.664417215429319e-08, "loss": 0.09160200357437134, "step": 36610 }, { "epoch": 0.9646266253543758, "grad_norm": 3.525944232940674, "learning_rate": 6.566447779430163e-08, "loss": 0.08668269515037537, "step": 36620 }, { "epoch": 0.9648900405988745, "grad_norm": 2.9546639919281006, "learning_rate": 6.469201407230729e-08, "loss": 0.07734536528587341, "step": 36630 }, { "epoch": 0.9651534558433733, "grad_norm": 3.4185919761657715, "learning_rate": 6.372678169611068e-08, "loss": 0.07392004132270813, "step": 36640 }, { "epoch": 0.965416871087872, "grad_norm": 3.6929147243499756, "learning_rate": 6.276878136824982e-08, "loss": 0.07712298631668091, "step": 36650 }, { "epoch": 0.9656802863323708, "grad_norm": 2.791720390319824, "learning_rate": 6.181801378599805e-08, "loss": 0.07653757333755493, "step": 36660 }, { "epoch": 0.9659437015768695, "grad_norm": 3.2582476139068604, "learning_rate": 6.087447964136517e-08, "loss": 0.07638850212097167, "step": 36670 }, { "epoch": 0.9662071168213683, "grad_norm": 4.1653337478637695, "learning_rate": 5.993817962109627e-08, "loss": 0.09356833696365356, "step": 36680 }, { "epoch": 0.966470532065867, "grad_norm": 3.7259178161621094, "learning_rate": 5.9009114406671787e-08, "loss": 0.08345742225646972, "step": 36690 }, { "epoch": 0.9667339473103658, "grad_norm": 4.181189060211182, "learning_rate": 5.8087284674304136e-08, "loss": 0.07576520442962646, "step": 36700 }, { "epoch": 0.9669973625548645, "grad_norm": 2.901170015335083, "learning_rate": 5.7172691094942165e-08, "loss": 0.06923032999038696, "step": 36710 }, { "epoch": 0.9672607777993631, "grad_norm": 4.368630886077881, "learning_rate": 5.626533433426451e-08, "loss": 0.08662058115005493, "step": 36720 }, { "epoch": 0.967524193043862, "grad_norm": 3.0518579483032227, "learning_rate": 5.5365215052686216e-08, "loss": 0.07936562895774842, "step": 36730 }, { "epoch": 0.9677876082883606, "grad_norm": 3.8211121559143066, "learning_rate": 5.4472333905352115e-08, "loss": 0.08023069500923156, "step": 36740 }, { "epoch": 0.9680510235328594, "grad_norm": 3.1296002864837646, "learning_rate": 5.3586691542140135e-08, "loss": 0.07046308517456054, "step": 36750 }, { "epoch": 0.9683144387773581, "grad_norm": 3.0232062339782715, "learning_rate": 5.2708288607659084e-08, "loss": 0.07212632894515991, "step": 36760 }, { "epoch": 0.9685778540218569, "grad_norm": 3.0256876945495605, "learning_rate": 5.183712574124755e-08, "loss": 0.07511810064315796, "step": 36770 }, { "epoch": 0.9688412692663556, "grad_norm": 3.7894859313964844, "learning_rate": 5.0973203576976104e-08, "loss": 0.082280832529068, "step": 36780 }, { "epoch": 0.9691046845108544, "grad_norm": 2.6824374198913574, "learning_rate": 5.011652274364509e-08, "loss": 0.07485482692718506, "step": 36790 }, { "epoch": 0.9693680997553531, "grad_norm": 3.513300895690918, "learning_rate": 4.926708386478352e-08, "loss": 0.0702333390712738, "step": 36800 }, { "epoch": 0.9696315149998518, "grad_norm": 3.601963758468628, "learning_rate": 4.8424887558649045e-08, "loss": 0.07556544542312622, "step": 36810 }, { "epoch": 0.9698949302443506, "grad_norm": 2.312626361846924, "learning_rate": 4.7589934438228017e-08, "loss": 0.08257483243942261, "step": 36820 }, { "epoch": 0.9701583454888493, "grad_norm": 2.9478111267089844, "learning_rate": 4.676222511123762e-08, "loss": 0.08224466443061829, "step": 36830 }, { "epoch": 0.9704217607333481, "grad_norm": 4.928873538970947, "learning_rate": 4.59417601801182e-08, "loss": 0.08147554397583008, "step": 36840 }, { "epoch": 0.9706851759778468, "grad_norm": 2.947183132171631, "learning_rate": 4.512854024203983e-08, "loss": 0.09569700956344604, "step": 36850 }, { "epoch": 0.9709485912223456, "grad_norm": 1.9414899349212646, "learning_rate": 4.432256588890016e-08, "loss": 0.0798923909664154, "step": 36860 }, { "epoch": 0.9712120064668442, "grad_norm": 3.5305824279785156, "learning_rate": 4.352383770732216e-08, "loss": 0.07402817010879517, "step": 36870 }, { "epoch": 0.9714754217113429, "grad_norm": 3.7674367427825928, "learning_rate": 4.273235627865302e-08, "loss": 0.0700104832649231, "step": 36880 }, { "epoch": 0.9717388369558417, "grad_norm": 2.799612045288086, "learning_rate": 4.19481221789686e-08, "loss": 0.0758358359336853, "step": 36890 }, { "epoch": 0.9720022522003404, "grad_norm": 3.647714614868164, "learning_rate": 4.117113597906675e-08, "loss": 0.07419056892395019, "step": 36900 }, { "epoch": 0.9722656674448392, "grad_norm": 3.179960012435913, "learning_rate": 4.040139824447398e-08, "loss": 0.07962167263031006, "step": 36910 }, { "epoch": 0.9725290826893379, "grad_norm": 3.8715782165527344, "learning_rate": 3.963890953543659e-08, "loss": 0.0727982223033905, "step": 36920 }, { "epoch": 0.9727924979338367, "grad_norm": 4.500765323638916, "learning_rate": 3.88836704069262e-08, "loss": 0.0717010498046875, "step": 36930 }, { "epoch": 0.9730559131783354, "grad_norm": 3.508429527282715, "learning_rate": 3.813568140864088e-08, "loss": 0.08526763916015626, "step": 36940 }, { "epoch": 0.9733193284228342, "grad_norm": 3.344327688217163, "learning_rate": 3.739494308499736e-08, "loss": 0.0982741117477417, "step": 36950 }, { "epoch": 0.9735827436673329, "grad_norm": 3.0001180171966553, "learning_rate": 3.66614559751377e-08, "loss": 0.08676490783691407, "step": 36960 }, { "epoch": 0.9738461589118316, "grad_norm": 3.3189477920532227, "learning_rate": 3.593522061292598e-08, "loss": 0.05739823579788208, "step": 36970 }, { "epoch": 0.9741095741563304, "grad_norm": 4.595390319824219, "learning_rate": 3.521623752694603e-08, "loss": 0.08681726455688477, "step": 36980 }, { "epoch": 0.9743729894008291, "grad_norm": 2.662384033203125, "learning_rate": 3.4504507240507027e-08, "loss": 0.07425220012664795, "step": 36990 }, { "epoch": 0.9746364046453279, "grad_norm": 3.124876022338867, "learning_rate": 3.380003027163459e-08, "loss": 0.07612388730049133, "step": 37000 }, { "epoch": 0.9748998198898265, "grad_norm": 3.627427339553833, "learning_rate": 3.310280713307856e-08, "loss": 0.07537530660629273, "step": 37010 }, { "epoch": 0.9751632351343253, "grad_norm": 3.9098846912384033, "learning_rate": 3.241283833230746e-08, "loss": 0.07303674817085266, "step": 37020 }, { "epoch": 0.975426650378824, "grad_norm": 3.2383813858032227, "learning_rate": 3.1730124371511794e-08, "loss": 0.07097781896591186, "step": 37030 }, { "epoch": 0.9756900656233228, "grad_norm": 3.512845039367676, "learning_rate": 3.1054665747599635e-08, "loss": 0.06088153123855591, "step": 37040 }, { "epoch": 0.9759534808678215, "grad_norm": 3.947493553161621, "learning_rate": 3.038646295219771e-08, "loss": 0.1008114218711853, "step": 37050 }, { "epoch": 0.9762168961123202, "grad_norm": 3.1637635231018066, "learning_rate": 2.972551647165367e-08, "loss": 0.0829900324344635, "step": 37060 }, { "epoch": 0.976480311356819, "grad_norm": 3.347137689590454, "learning_rate": 2.9071826787033797e-08, "loss": 0.08734087347984314, "step": 37070 }, { "epoch": 0.9767437266013177, "grad_norm": 3.0308544635772705, "learning_rate": 2.8425394374120842e-08, "loss": 0.07665975093841552, "step": 37080 }, { "epoch": 0.9770071418458165, "grad_norm": 3.363190174102783, "learning_rate": 2.7786219703415108e-08, "loss": 0.07550719380378723, "step": 37090 }, { "epoch": 0.9772705570903152, "grad_norm": 2.2930634021759033, "learning_rate": 2.715430324013779e-08, "loss": 0.060484069585800174, "step": 37100 }, { "epoch": 0.977533972334814, "grad_norm": 3.6044762134552, "learning_rate": 2.6529645444223206e-08, "loss": 0.08935660123825073, "step": 37110 }, { "epoch": 0.9777973875793127, "grad_norm": 3.4969990253448486, "learning_rate": 2.5912246770324333e-08, "loss": 0.08136753439903259, "step": 37120 }, { "epoch": 0.9780608028238115, "grad_norm": 3.7425410747528076, "learning_rate": 2.5302107667811716e-08, "loss": 0.0812181830406189, "step": 37130 }, { "epoch": 0.9783242180683102, "grad_norm": 3.616577386856079, "learning_rate": 2.469922858076901e-08, "loss": 0.08431209325790405, "step": 37140 }, { "epoch": 0.9785876333128088, "grad_norm": 5.54459810256958, "learning_rate": 2.410360994799965e-08, "loss": 0.08989672660827637, "step": 37150 }, { "epoch": 0.9788510485573076, "grad_norm": 3.278508424758911, "learning_rate": 2.351525220301909e-08, "loss": 0.08542128801345825, "step": 37160 }, { "epoch": 0.9791144638018063, "grad_norm": 2.9900476932525635, "learning_rate": 2.293415577406144e-08, "loss": 0.09053856134414673, "step": 37170 }, { "epoch": 0.9793778790463051, "grad_norm": 3.34546160697937, "learning_rate": 2.2360321084071712e-08, "loss": 0.0636170744895935, "step": 37180 }, { "epoch": 0.9796412942908038, "grad_norm": 3.35791277885437, "learning_rate": 2.179374855071137e-08, "loss": 0.08241952657699585, "step": 37190 }, { "epoch": 0.9799047095353026, "grad_norm": 3.3200953006744385, "learning_rate": 2.123443858635832e-08, "loss": 0.08143237233161926, "step": 37200 }, { "epoch": 0.9801681247798013, "grad_norm": 3.085994005203247, "learning_rate": 2.068239159810137e-08, "loss": 0.08010343313217164, "step": 37210 }, { "epoch": 0.9804315400243001, "grad_norm": 4.877513408660889, "learning_rate": 2.0137607987743558e-08, "loss": 0.0712773084640503, "step": 37220 }, { "epoch": 0.9806949552687988, "grad_norm": 3.1044914722442627, "learning_rate": 1.9600088151803255e-08, "loss": 0.06795154809951783, "step": 37230 }, { "epoch": 0.9809583705132975, "grad_norm": 3.900892734527588, "learning_rate": 1.906983248150973e-08, "loss": 0.068202805519104, "step": 37240 }, { "epoch": 0.9812217857577963, "grad_norm": 2.255739212036133, "learning_rate": 1.854684136280538e-08, "loss": 0.07756006717681885, "step": 37250 }, { "epoch": 0.981485201002295, "grad_norm": 4.424480438232422, "learning_rate": 1.8031115176346815e-08, "loss": 0.08052337765693665, "step": 37260 }, { "epoch": 0.9817486162467938, "grad_norm": 3.1798274517059326, "learning_rate": 1.752265429750044e-08, "loss": 0.09057775139808655, "step": 37270 }, { "epoch": 0.9820120314912925, "grad_norm": 4.4971537590026855, "learning_rate": 1.7021459096348004e-08, "loss": 0.08075262308120727, "step": 37280 }, { "epoch": 0.9822754467357913, "grad_norm": 3.6185412406921387, "learning_rate": 1.652752993767992e-08, "loss": 0.0763546347618103, "step": 37290 }, { "epoch": 0.98253886198029, "grad_norm": 2.893226146697998, "learning_rate": 1.604086718099973e-08, "loss": 0.07810492515563965, "step": 37300 }, { "epoch": 0.9828022772247887, "grad_norm": 2.9331471920013428, "learning_rate": 1.556147118051965e-08, "loss": 0.0737647533416748, "step": 37310 }, { "epoch": 0.9830656924692874, "grad_norm": 3.0597543716430664, "learning_rate": 1.508934228516723e-08, "loss": 0.08252391219139099, "step": 37320 }, { "epoch": 0.9833291077137861, "grad_norm": 3.8699281215667725, "learning_rate": 1.4624480838576483e-08, "loss": 0.08341807126998901, "step": 37330 }, { "epoch": 0.9835925229582849, "grad_norm": 2.948063850402832, "learning_rate": 1.4166887179094535e-08, "loss": 0.06418104767799378, "step": 37340 }, { "epoch": 0.9838559382027836, "grad_norm": 3.1125924587249756, "learning_rate": 1.3716561639777192e-08, "loss": 0.07187076210975647, "step": 37350 }, { "epoch": 0.9841193534472824, "grad_norm": 3.183595895767212, "learning_rate": 1.3273504548392269e-08, "loss": 0.07569841742515564, "step": 37360 }, { "epoch": 0.9843827686917811, "grad_norm": 3.215208053588867, "learning_rate": 1.2837716227414032e-08, "loss": 0.06660377979278564, "step": 37370 }, { "epoch": 0.9846461839362799, "grad_norm": 2.7004621028900146, "learning_rate": 1.240919699402765e-08, "loss": 0.07177968621253968, "step": 37380 }, { "epoch": 0.9849095991807786, "grad_norm": 3.2651877403259277, "learning_rate": 1.1987947160130297e-08, "loss": 0.07940088510513306, "step": 37390 }, { "epoch": 0.9851730144252773, "grad_norm": 2.4362688064575195, "learning_rate": 1.1573967032324495e-08, "loss": 0.08306593894958496, "step": 37400 }, { "epoch": 0.9854364296697761, "grad_norm": 3.588179349899292, "learning_rate": 1.1167256911922552e-08, "loss": 0.07781577706336976, "step": 37410 }, { "epoch": 0.9856998449142748, "grad_norm": 3.144303560256958, "learning_rate": 1.0767817094945454e-08, "loss": 0.07923663854598999, "step": 37420 }, { "epoch": 0.9859632601587736, "grad_norm": 3.261697769165039, "learning_rate": 1.0375647872123974e-08, "loss": 0.07122613191604614, "step": 37430 }, { "epoch": 0.9862266754032722, "grad_norm": 4.631298542022705, "learning_rate": 9.990749528894227e-09, "loss": 0.08086256980895996, "step": 37440 }, { "epoch": 0.986490090647771, "grad_norm": 3.520907163619995, "learning_rate": 9.613122345403236e-09, "loss": 0.08582785129547119, "step": 37450 }, { "epoch": 0.9867535058922697, "grad_norm": 3.6395130157470703, "learning_rate": 9.24276659650336e-09, "loss": 0.0736070692539215, "step": 37460 }, { "epoch": 0.9870169211367685, "grad_norm": 3.301961660385132, "learning_rate": 8.879682551755643e-09, "loss": 0.08850346803665161, "step": 37470 }, { "epoch": 0.9872803363812672, "grad_norm": 3.173818349838257, "learning_rate": 8.523870475429797e-09, "loss": 0.07176629304885865, "step": 37480 }, { "epoch": 0.9875437516257659, "grad_norm": 3.430192708969116, "learning_rate": 8.175330626498668e-09, "loss": 0.0765186071395874, "step": 37490 }, { "epoch": 0.9878071668702647, "grad_norm": 3.262955665588379, "learning_rate": 7.834063258645997e-09, "loss": 0.07930638790130615, "step": 37500 }, { "epoch": 0.9880705821147634, "grad_norm": 2.945673704147339, "learning_rate": 7.500068620260869e-09, "loss": 0.0764473557472229, "step": 37510 }, { "epoch": 0.9883339973592622, "grad_norm": 3.3923799991607666, "learning_rate": 7.173346954438831e-09, "loss": 0.0764809489250183, "step": 37520 }, { "epoch": 0.9885974126037609, "grad_norm": 3.432147264480591, "learning_rate": 6.853898498982992e-09, "loss": 0.07773237824440002, "step": 37530 }, { "epoch": 0.9888608278482597, "grad_norm": 3.2720041275024414, "learning_rate": 6.541723486399587e-09, "loss": 0.07817720174789429, "step": 37540 }, { "epoch": 0.9891242430927584, "grad_norm": 3.31453275680542, "learning_rate": 6.236822143904642e-09, "loss": 0.07847856283187866, "step": 37550 }, { "epoch": 0.9893876583372572, "grad_norm": 3.8542683124542236, "learning_rate": 5.939194693418415e-09, "loss": 0.07383745908737183, "step": 37560 }, { "epoch": 0.9896510735817559, "grad_norm": 4.651130199432373, "learning_rate": 5.648841351566514e-09, "loss": 0.07695866823196411, "step": 37570 }, { "epoch": 0.9899144888262545, "grad_norm": 3.89056134223938, "learning_rate": 5.365762329681002e-09, "loss": 0.07416090369224548, "step": 37580 }, { "epoch": 0.9901779040707533, "grad_norm": 2.158252239227295, "learning_rate": 5.089957833798176e-09, "loss": 0.07602699398994446, "step": 37590 }, { "epoch": 0.990441319315252, "grad_norm": 2.760948896408081, "learning_rate": 4.8214280646619035e-09, "loss": 0.07971548438072204, "step": 37600 }, { "epoch": 0.9907047345597508, "grad_norm": 2.9066269397735596, "learning_rate": 4.560173217718067e-09, "loss": 0.0768587350845337, "step": 37610 }, { "epoch": 0.9909681498042495, "grad_norm": 4.147697448730469, "learning_rate": 4.3061934831212235e-09, "loss": 0.06948434114456177, "step": 37620 }, { "epoch": 0.9912315650487483, "grad_norm": 2.5103397369384766, "learning_rate": 4.0594890457268384e-09, "loss": 0.07324331998825073, "step": 37630 }, { "epoch": 0.991494980293247, "grad_norm": 4.118902206420898, "learning_rate": 3.820060085097943e-09, "loss": 0.07873862981796265, "step": 37640 }, { "epoch": 0.9917583955377458, "grad_norm": 4.904651165008545, "learning_rate": 3.587906775501804e-09, "loss": 0.08098738193511963, "step": 37650 }, { "epoch": 0.9920218107822445, "grad_norm": 2.6761322021484375, "learning_rate": 3.363029285907704e-09, "loss": 0.06993563771247864, "step": 37660 }, { "epoch": 0.9922852260267432, "grad_norm": 2.9400861263275146, "learning_rate": 3.145427779993604e-09, "loss": 0.08477054834365845, "step": 37670 }, { "epoch": 0.992548641271242, "grad_norm": 2.625338554382324, "learning_rate": 2.935102416138369e-09, "loss": 0.08911079168319702, "step": 37680 }, { "epoch": 0.9928120565157407, "grad_norm": 2.4013583660125732, "learning_rate": 2.732053347425101e-09, "loss": 0.07075082659721374, "step": 37690 }, { "epoch": 0.9930754717602395, "grad_norm": 3.2991507053375244, "learning_rate": 2.536280721642248e-09, "loss": 0.08366369009017945, "step": 37700 }, { "epoch": 0.9933388870047382, "grad_norm": 3.3153536319732666, "learning_rate": 2.3477846812813843e-09, "loss": 0.07743412256240845, "step": 37710 }, { "epoch": 0.993602302249237, "grad_norm": 4.039905071258545, "learning_rate": 2.16656536353832e-09, "loss": 0.07678064107894897, "step": 37720 }, { "epoch": 0.9938657174937356, "grad_norm": 3.651860475540161, "learning_rate": 1.9926229003119912e-09, "loss": 0.07498173713684082, "step": 37730 }, { "epoch": 0.9941291327382344, "grad_norm": 3.541916608810425, "learning_rate": 1.8259574182055707e-09, "loss": 0.07635020017623902, "step": 37740 }, { "epoch": 0.9943925479827331, "grad_norm": 2.8652400970458984, "learning_rate": 1.6665690385253564e-09, "loss": 0.07228919863700867, "step": 37750 }, { "epoch": 0.9946559632272318, "grad_norm": 2.7861557006835938, "learning_rate": 1.5144578772807728e-09, "loss": 0.0788446307182312, "step": 37760 }, { "epoch": 0.9949193784717306, "grad_norm": 2.349595785140991, "learning_rate": 1.36962404518437e-09, "loss": 0.08906376361846924, "step": 37770 }, { "epoch": 0.9951827937162293, "grad_norm": 3.74971342086792, "learning_rate": 1.232067647654045e-09, "loss": 0.0762283205986023, "step": 37780 }, { "epoch": 0.9954462089607281, "grad_norm": 4.034066200256348, "learning_rate": 1.1017887848085995e-09, "loss": 0.08260319232940674, "step": 37790 }, { "epoch": 0.9957096242052268, "grad_norm": 2.7533040046691895, "learning_rate": 9.78787551469962e-10, "loss": 0.0735251784324646, "step": 37800 }, { "epoch": 0.9959730394497256, "grad_norm": 3.590857982635498, "learning_rate": 8.630640371642963e-10, "loss": 0.07326527833938598, "step": 37810 }, { "epoch": 0.9962364546942243, "grad_norm": 3.6786584854125977, "learning_rate": 7.546183261197826e-10, "loss": 0.07715892791748047, "step": 37820 }, { "epoch": 0.9964998699387231, "grad_norm": 3.3959007263183594, "learning_rate": 6.534504972688372e-10, "loss": 0.07502488493919372, "step": 37830 }, { "epoch": 0.9967632851832218, "grad_norm": 3.0592262744903564, "learning_rate": 5.595606242447815e-10, "loss": 0.0936536431312561, "step": 37840 }, { "epoch": 0.9970267004277205, "grad_norm": 2.99973726272583, "learning_rate": 4.729487753851736e-10, "loss": 0.07042630910873413, "step": 37850 }, { "epoch": 0.9972901156722193, "grad_norm": 4.111169338226318, "learning_rate": 3.9361501372958686e-10, "loss": 0.07580758333206176, "step": 37860 }, { "epoch": 0.997553530916718, "grad_norm": 4.139433860778809, "learning_rate": 3.2155939702072093e-10, "loss": 0.09232652187347412, "step": 37870 }, { "epoch": 0.9978169461612167, "grad_norm": 2.61824369430542, "learning_rate": 2.5678197770329094e-10, "loss": 0.08896642923355103, "step": 37880 }, { "epoch": 0.9980803614057154, "grad_norm": 3.5672225952148438, "learning_rate": 1.9928280292624835e-10, "loss": 0.08291504383087159, "step": 37890 }, { "epoch": 0.9983437766502142, "grad_norm": 3.055878162384033, "learning_rate": 1.4906191453833985e-10, "loss": 0.07394527196884156, "step": 37900 }, { "epoch": 0.9986071918947129, "grad_norm": 2.499595880508423, "learning_rate": 1.061193490936585e-10, "loss": 0.06773900389671325, "step": 37910 }, { "epoch": 0.9988706071392117, "grad_norm": 3.7744088172912598, "learning_rate": 7.045513784831315e-11, "loss": 0.07089773416519166, "step": 37920 }, { "epoch": 0.9991340223837104, "grad_norm": 2.782545566558838, "learning_rate": 4.206930675820786e-11, "loss": 0.06456289887428283, "step": 37930 }, { "epoch": 0.9993974376282091, "grad_norm": 2.664315938949585, "learning_rate": 2.0961876485703358e-11, "loss": 0.07547470331192016, "step": 37940 }, { "epoch": 0.9996608528727079, "grad_norm": 2.354475259780884, "learning_rate": 7.132862392955631e-12, "loss": 0.07975548505783081, "step": 37950 }, { "epoch": 0.9999242681172066, "grad_norm": 3.767303943634033, "learning_rate": 5.822745441363964e-13, "loss": 0.07791775465011597, "step": 37960 }, { "epoch": 1.0, "step": 37963, "total_flos": 2.0205788732727283e+20, "train_loss": 0.14081798901884623, "train_runtime": 355201.6781, "train_samples_per_second": 6.84, "train_steps_per_second": 0.107 } ], "logging_steps": 10, "max_steps": 37963, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 1000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 2.0205788732727283e+20, "train_batch_size": 1, "trial_name": null, "trial_params": null }