{ "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": 1213.853759765625, "learning_rate": 1.5803336259877089e-07, "loss": 2.3447616577148436, "step": 10 }, { "epoch": 0.0005268304889974745, "grad_norm": 56.11354064941406, "learning_rate": 3.336259877085163e-07, "loss": 0.6658806800842285, "step": 20 }, { "epoch": 0.0007902457334962118, "grad_norm": 32.74578094482422, "learning_rate": 5.092186128182617e-07, "loss": 0.2987356662750244, "step": 30 }, { "epoch": 0.001053660977994949, "grad_norm": 14.561054229736328, "learning_rate": 6.848112379280071e-07, "loss": 0.24938445091247557, "step": 40 }, { "epoch": 0.0013170762224936862, "grad_norm": 20.70635414123535, "learning_rate": 8.604038630377525e-07, "loss": 0.2558992147445679, "step": 50 }, { "epoch": 0.0015804914669924235, "grad_norm": 14.778559684753418, "learning_rate": 1.035996488147498e-06, "loss": 0.23694872856140137, "step": 60 }, { "epoch": 0.0018439067114911607, "grad_norm": 11.012696266174316, "learning_rate": 1.2115891132572432e-06, "loss": 0.22169573307037355, "step": 70 }, { "epoch": 0.002107321955989898, "grad_norm": 9.719491004943848, "learning_rate": 1.3871817383669885e-06, "loss": 0.23246588706970214, "step": 80 }, { "epoch": 0.002370737200488635, "grad_norm": 13.835502624511719, "learning_rate": 1.5627743634767342e-06, "loss": 0.2154603958129883, "step": 90 }, { "epoch": 0.0026341524449873723, "grad_norm": 12.435432434082031, "learning_rate": 1.7383669885864795e-06, "loss": 0.21393170356750488, "step": 100 }, { "epoch": 0.00289756768948611, "grad_norm": 10.624913215637207, "learning_rate": 1.913959613696225e-06, "loss": 0.21830177307128906, "step": 110 }, { "epoch": 0.003160982933984847, "grad_norm": 17.651155471801758, "learning_rate": 2.08955223880597e-06, "loss": 0.23218910694122313, "step": 120 }, { "epoch": 0.003424398178483584, "grad_norm": 10.375974655151367, "learning_rate": 2.265144863915716e-06, "loss": 0.2023460865020752, "step": 130 }, { "epoch": 0.0036878134229823213, "grad_norm": 15.337003707885742, "learning_rate": 2.440737489025461e-06, "loss": 0.2147892475128174, "step": 140 }, { "epoch": 0.0039512286674810585, "grad_norm": 10.703423500061035, "learning_rate": 2.6163301141352064e-06, "loss": 0.21736392974853516, "step": 150 }, { "epoch": 0.004214643911979796, "grad_norm": 17.63776206970215, "learning_rate": 2.791922739244952e-06, "loss": 0.23426344394683837, "step": 160 }, { "epoch": 0.004478059156478534, "grad_norm": 11.715472221374512, "learning_rate": 2.9675153643546974e-06, "loss": 0.23157052993774413, "step": 170 }, { "epoch": 0.00474147440097727, "grad_norm": 10.725784301757812, "learning_rate": 3.1431079894644427e-06, "loss": 0.23005118370056152, "step": 180 }, { "epoch": 0.005004889645476008, "grad_norm": 7.275376796722412, "learning_rate": 3.318700614574188e-06, "loss": 0.21270663738250734, "step": 190 }, { "epoch": 0.005268304889974745, "grad_norm": 12.43260669708252, "learning_rate": 3.4942932396839337e-06, "loss": 0.2250192642211914, "step": 200 }, { "epoch": 0.005531720134473482, "grad_norm": 13.314983367919922, "learning_rate": 3.669885864793679e-06, "loss": 0.21406316757202148, "step": 210 }, { "epoch": 0.00579513537897222, "grad_norm": 19.876806259155273, "learning_rate": 3.845478489903424e-06, "loss": 0.2051485538482666, "step": 220 }, { "epoch": 0.0060585506234709565, "grad_norm": 11.659071922302246, "learning_rate": 4.0210711150131695e-06, "loss": 0.21260828971862794, "step": 230 }, { "epoch": 0.006321965867969694, "grad_norm": 13.595244407653809, "learning_rate": 4.196663740122916e-06, "loss": 0.22221150398254394, "step": 240 }, { "epoch": 0.006585381112468432, "grad_norm": 14.938165664672852, "learning_rate": 4.37225636523266e-06, "loss": 0.17944687604904175, "step": 250 }, { "epoch": 0.006848796356967168, "grad_norm": 13.919473648071289, "learning_rate": 4.547848990342406e-06, "loss": 0.215757155418396, "step": 260 }, { "epoch": 0.007112211601465906, "grad_norm": 5.823989391326904, "learning_rate": 4.7234416154521515e-06, "loss": 0.19328951835632324, "step": 270 }, { "epoch": 0.007375626845964643, "grad_norm": 12.386701583862305, "learning_rate": 4.899034240561897e-06, "loss": 0.19956209659576415, "step": 280 }, { "epoch": 0.00763904209046338, "grad_norm": 9.185185432434082, "learning_rate": 5.074626865671642e-06, "loss": 0.1913324236869812, "step": 290 }, { "epoch": 0.007902457334962117, "grad_norm": 9.66295337677002, "learning_rate": 5.250219490781387e-06, "loss": 0.19735319614410402, "step": 300 }, { "epoch": 0.008165872579460855, "grad_norm": 8.514077186584473, "learning_rate": 5.425812115891133e-06, "loss": 0.19270682334899902, "step": 310 }, { "epoch": 0.008429287823959592, "grad_norm": 12.31623363494873, "learning_rate": 5.601404741000879e-06, "loss": 0.20009355545043944, "step": 320 }, { "epoch": 0.008692703068458329, "grad_norm": 7.190951347351074, "learning_rate": 5.776997366110624e-06, "loss": 0.20620157718658447, "step": 330 }, { "epoch": 0.008956118312957067, "grad_norm": 13.58059024810791, "learning_rate": 5.9525899912203685e-06, "loss": 0.189217209815979, "step": 340 }, { "epoch": 0.009219533557455804, "grad_norm": 9.881999015808105, "learning_rate": 6.128182616330115e-06, "loss": 0.21762690544128419, "step": 350 }, { "epoch": 0.00948294880195454, "grad_norm": 10.044063568115234, "learning_rate": 6.30377524143986e-06, "loss": 0.2002377986907959, "step": 360 }, { "epoch": 0.00974636404645328, "grad_norm": 7.230624198913574, "learning_rate": 6.479367866549605e-06, "loss": 0.20482277870178223, "step": 370 }, { "epoch": 0.010009779290952016, "grad_norm": 5.98422384262085, "learning_rate": 6.654960491659351e-06, "loss": 0.2053527593612671, "step": 380 }, { "epoch": 0.010273194535450753, "grad_norm": 12.446501731872559, "learning_rate": 6.830553116769097e-06, "loss": 0.19890973567962647, "step": 390 }, { "epoch": 0.01053660977994949, "grad_norm": 6.469152927398682, "learning_rate": 7.006145741878841e-06, "loss": 0.1954428434371948, "step": 400 }, { "epoch": 0.010800025024448228, "grad_norm": 4.233999729156494, "learning_rate": 7.181738366988587e-06, "loss": 0.1923543930053711, "step": 410 }, { "epoch": 0.011063440268946964, "grad_norm": 7.163558006286621, "learning_rate": 7.3573309920983325e-06, "loss": 0.1974882960319519, "step": 420 }, { "epoch": 0.011326855513445701, "grad_norm": 11.31024169921875, "learning_rate": 7.532923617208078e-06, "loss": 0.18022000789642334, "step": 430 }, { "epoch": 0.01159027075794444, "grad_norm": 4.744543075561523, "learning_rate": 7.708516242317823e-06, "loss": 0.18819494247436525, "step": 440 }, { "epoch": 0.011853686002443176, "grad_norm": 7.333344459533691, "learning_rate": 7.88410886742757e-06, "loss": 0.19481472969055175, "step": 450 }, { "epoch": 0.012117101246941913, "grad_norm": 8.321006774902344, "learning_rate": 8.059701492537314e-06, "loss": 0.19232465028762818, "step": 460 }, { "epoch": 0.012380516491440651, "grad_norm": 5.622868061065674, "learning_rate": 8.23529411764706e-06, "loss": 0.18018527030944825, "step": 470 }, { "epoch": 0.012643931735939388, "grad_norm": 6.201679229736328, "learning_rate": 8.410886742756806e-06, "loss": 0.19390923976898194, "step": 480 }, { "epoch": 0.012907346980438125, "grad_norm": 7.933331489562988, "learning_rate": 8.58647936786655e-06, "loss": 0.19396061897277833, "step": 490 }, { "epoch": 0.013170762224936863, "grad_norm": 13.89413070678711, "learning_rate": 8.762071992976295e-06, "loss": 0.20955476760864258, "step": 500 }, { "epoch": 0.0134341774694356, "grad_norm": 7.629615783691406, "learning_rate": 8.937664618086041e-06, "loss": 0.20989909172058105, "step": 510 }, { "epoch": 0.013697592713934337, "grad_norm": 10.688849449157715, "learning_rate": 9.113257243195787e-06, "loss": 0.20512137413024903, "step": 520 }, { "epoch": 0.013961007958433075, "grad_norm": 5.66908597946167, "learning_rate": 9.288849868305532e-06, "loss": 0.19818532466888428, "step": 530 }, { "epoch": 0.014224423202931812, "grad_norm": 11.467631340026855, "learning_rate": 9.464442493415278e-06, "loss": 0.19799351692199707, "step": 540 }, { "epoch": 0.014487838447430549, "grad_norm": 2.7566654682159424, "learning_rate": 9.640035118525024e-06, "loss": 0.18927077054977418, "step": 550 }, { "epoch": 0.014751253691929285, "grad_norm": 5.6378254890441895, "learning_rate": 9.815627743634768e-06, "loss": 0.18606919050216675, "step": 560 }, { "epoch": 0.015014668936428024, "grad_norm": 5.247837066650391, "learning_rate": 9.991220368744513e-06, "loss": 0.1957606554031372, "step": 570 }, { "epoch": 0.01527808418092676, "grad_norm": 6.268489360809326, "learning_rate": 1.0166812993854259e-05, "loss": 0.19652947187423705, "step": 580 }, { "epoch": 0.015541499425425497, "grad_norm": 7.421579360961914, "learning_rate": 1.0342405618964005e-05, "loss": 0.1817408800125122, "step": 590 }, { "epoch": 0.015804914669924234, "grad_norm": 9.6539888381958, "learning_rate": 1.0517998244073751e-05, "loss": 0.19063999652862548, "step": 600 }, { "epoch": 0.016068329914422974, "grad_norm": 6.174042224884033, "learning_rate": 1.0693590869183494e-05, "loss": 0.19408154487609863, "step": 610 }, { "epoch": 0.01633174515892171, "grad_norm": 4.681012153625488, "learning_rate": 1.086918349429324e-05, "loss": 0.18212637901306153, "step": 620 }, { "epoch": 0.016595160403420448, "grad_norm": 10.34129524230957, "learning_rate": 1.1044776119402986e-05, "loss": 0.20589234828948974, "step": 630 }, { "epoch": 0.016858575647919184, "grad_norm": 7.0870041847229, "learning_rate": 1.122036874451273e-05, "loss": 0.184228253364563, "step": 640 }, { "epoch": 0.01712199089241792, "grad_norm": 6.96578311920166, "learning_rate": 1.1395961369622477e-05, "loss": 0.19321576356887818, "step": 650 }, { "epoch": 0.017385406136916658, "grad_norm": 5.053447723388672, "learning_rate": 1.1571553994732223e-05, "loss": 0.21489789485931396, "step": 660 }, { "epoch": 0.017648821381415394, "grad_norm": 5.832465648651123, "learning_rate": 1.1747146619841967e-05, "loss": 0.19735374450683593, "step": 670 }, { "epoch": 0.017912236625914135, "grad_norm": 4.49865198135376, "learning_rate": 1.1922739244951713e-05, "loss": 0.18788537979125977, "step": 680 }, { "epoch": 0.01817565187041287, "grad_norm": 7.63986349105835, "learning_rate": 1.209833187006146e-05, "loss": 0.20276927947998047, "step": 690 }, { "epoch": 0.018439067114911608, "grad_norm": 6.861851215362549, "learning_rate": 1.2273924495171204e-05, "loss": 0.19753156900405883, "step": 700 }, { "epoch": 0.018702482359410345, "grad_norm": 7.664979457855225, "learning_rate": 1.244951712028095e-05, "loss": 0.18222628831863402, "step": 710 }, { "epoch": 0.01896589760390908, "grad_norm": 7.98739767074585, "learning_rate": 1.2625109745390696e-05, "loss": 0.1874994993209839, "step": 720 }, { "epoch": 0.019229312848407818, "grad_norm": 4.063989639282227, "learning_rate": 1.2800702370500439e-05, "loss": 0.1873386859893799, "step": 730 }, { "epoch": 0.01949272809290656, "grad_norm": 7.797830581665039, "learning_rate": 1.2976294995610185e-05, "loss": 0.19260737895965577, "step": 740 }, { "epoch": 0.019756143337405295, "grad_norm": 6.3047590255737305, "learning_rate": 1.315188762071993e-05, "loss": 0.20644373893737794, "step": 750 }, { "epoch": 0.02001955858190403, "grad_norm": 5.755674362182617, "learning_rate": 1.3327480245829676e-05, "loss": 0.1950676202774048, "step": 760 }, { "epoch": 0.02028297382640277, "grad_norm": 8.321192741394043, "learning_rate": 1.3503072870939422e-05, "loss": 0.20606229305267335, "step": 770 }, { "epoch": 0.020546389070901505, "grad_norm": 3.934008836746216, "learning_rate": 1.3678665496049166e-05, "loss": 0.19398720264434816, "step": 780 }, { "epoch": 0.020809804315400242, "grad_norm": 4.474685192108154, "learning_rate": 1.3854258121158912e-05, "loss": 0.18364189863204955, "step": 790 }, { "epoch": 0.02107321955989898, "grad_norm": 14.472000122070312, "learning_rate": 1.4029850746268658e-05, "loss": 0.2094477891921997, "step": 800 }, { "epoch": 0.02133663480439772, "grad_norm": 4.927107810974121, "learning_rate": 1.4205443371378403e-05, "loss": 0.18549418449401855, "step": 810 }, { "epoch": 0.021600050048896455, "grad_norm": 4.542661190032959, "learning_rate": 1.4381035996488149e-05, "loss": 0.18159523010253906, "step": 820 }, { "epoch": 0.021863465293395192, "grad_norm": 3.155791997909546, "learning_rate": 1.4556628621597895e-05, "loss": 0.19228725433349608, "step": 830 }, { "epoch": 0.02212688053789393, "grad_norm": 4.353564262390137, "learning_rate": 1.4732221246707638e-05, "loss": 0.18702208995819092, "step": 840 }, { "epoch": 0.022390295782392666, "grad_norm": 5.855139255523682, "learning_rate": 1.4907813871817384e-05, "loss": 0.1885390043258667, "step": 850 }, { "epoch": 0.022653711026891402, "grad_norm": 3.2592806816101074, "learning_rate": 1.508340649692713e-05, "loss": 0.1821555972099304, "step": 860 }, { "epoch": 0.022917126271390142, "grad_norm": 6.030641078948975, "learning_rate": 1.5258999122036876e-05, "loss": 0.18880798816680908, "step": 870 }, { "epoch": 0.02318054151588888, "grad_norm": 4.193276405334473, "learning_rate": 1.5434591747146622e-05, "loss": 0.19863847494125367, "step": 880 }, { "epoch": 0.023443956760387616, "grad_norm": 6.0843071937561035, "learning_rate": 1.561018437225637e-05, "loss": 0.18098020553588867, "step": 890 }, { "epoch": 0.023707372004886353, "grad_norm": 5.233428955078125, "learning_rate": 1.578577699736611e-05, "loss": 0.1761097192764282, "step": 900 }, { "epoch": 0.02397078724938509, "grad_norm": 2.444744348526001, "learning_rate": 1.5961369622475857e-05, "loss": 0.19454495906829833, "step": 910 }, { "epoch": 0.024234202493883826, "grad_norm": 10.244645118713379, "learning_rate": 1.6136962247585604e-05, "loss": 0.19052202701568605, "step": 920 }, { "epoch": 0.024497617738382566, "grad_norm": 2.9832234382629395, "learning_rate": 1.6312554872695346e-05, "loss": 0.19734046459197999, "step": 930 }, { "epoch": 0.024761032982881303, "grad_norm": 3.473438262939453, "learning_rate": 1.6488147497805093e-05, "loss": 0.18573708534240724, "step": 940 }, { "epoch": 0.02502444822738004, "grad_norm": 3.3913497924804688, "learning_rate": 1.666374012291484e-05, "loss": 0.17256749868392945, "step": 950 }, { "epoch": 0.025287863471878776, "grad_norm": 4.64911413192749, "learning_rate": 1.6839332748024585e-05, "loss": 0.18427510261535646, "step": 960 }, { "epoch": 0.025551278716377513, "grad_norm": 2.568333625793457, "learning_rate": 1.701492537313433e-05, "loss": 0.17302186489105226, "step": 970 }, { "epoch": 0.02581469396087625, "grad_norm": 8.548494338989258, "learning_rate": 1.7190517998244077e-05, "loss": 0.19803428649902344, "step": 980 }, { "epoch": 0.026078109205374986, "grad_norm": 4.628727912902832, "learning_rate": 1.736611062335382e-05, "loss": 0.18425811529159547, "step": 990 }, { "epoch": 0.026341524449873727, "grad_norm": 5.913219928741455, "learning_rate": 1.7541703248463566e-05, "loss": 0.19183614253997802, "step": 1000 }, { "epoch": 0.026604939694372463, "grad_norm": 4.534789085388184, "learning_rate": 1.771729587357331e-05, "loss": 0.19450198411941527, "step": 1010 }, { "epoch": 0.0268683549388712, "grad_norm": 5.626791000366211, "learning_rate": 1.7892888498683055e-05, "loss": 0.19665119647979737, "step": 1020 }, { "epoch": 0.027131770183369937, "grad_norm": 3.325604200363159, "learning_rate": 1.80684811237928e-05, "loss": 0.17775366306304932, "step": 1030 }, { "epoch": 0.027395185427868674, "grad_norm": 3.685610771179199, "learning_rate": 1.8244073748902547e-05, "loss": 0.1893642544746399, "step": 1040 }, { "epoch": 0.02765860067236741, "grad_norm": 3.9268224239349365, "learning_rate": 1.8419666374012293e-05, "loss": 0.1820056200027466, "step": 1050 }, { "epoch": 0.02792201591686615, "grad_norm": 4.456267356872559, "learning_rate": 1.859525899912204e-05, "loss": 0.19983551502227784, "step": 1060 }, { "epoch": 0.028185431161364887, "grad_norm": 3.9226627349853516, "learning_rate": 1.8770851624231782e-05, "loss": 0.17890586853027343, "step": 1070 }, { "epoch": 0.028448846405863624, "grad_norm": 2.509916305541992, "learning_rate": 1.8946444249341528e-05, "loss": 0.2011794090270996, "step": 1080 }, { "epoch": 0.02871226165036236, "grad_norm": 6.960899829864502, "learning_rate": 1.9122036874451274e-05, "loss": 0.20108499526977539, "step": 1090 }, { "epoch": 0.028975676894861097, "grad_norm": 2.3062973022460938, "learning_rate": 1.929762949956102e-05, "loss": 0.18049287796020508, "step": 1100 }, { "epoch": 0.029239092139359834, "grad_norm": 4.6892805099487305, "learning_rate": 1.9473222124670767e-05, "loss": 0.19263887405395508, "step": 1110 }, { "epoch": 0.02950250738385857, "grad_norm": 4.693334102630615, "learning_rate": 1.9648814749780513e-05, "loss": 0.1748729944229126, "step": 1120 }, { "epoch": 0.02976592262835731, "grad_norm": 5.994380950927734, "learning_rate": 1.9824407374890256e-05, "loss": 0.18314557075500487, "step": 1130 }, { "epoch": 0.030029337872856048, "grad_norm": 3.029182195663452, "learning_rate": 2e-05, "loss": 0.18897643089294433, "step": 1140 }, { "epoch": 0.030292753117354784, "grad_norm": 8.279314041137695, "learning_rate": 1.9999996360784282e-05, "loss": 0.1883655905723572, "step": 1150 }, { "epoch": 0.03055616836185352, "grad_norm": 4.16698694229126, "learning_rate": 1.9999985443139778e-05, "loss": 0.18790898323059083, "step": 1160 }, { "epoch": 0.030819583606352258, "grad_norm": 3.299469232559204, "learning_rate": 1.999996724707443e-05, "loss": 0.1873364567756653, "step": 1170 }, { "epoch": 0.031082998850850994, "grad_norm": 3.6636407375335693, "learning_rate": 1.9999941772601482e-05, "loss": 0.18061861991882325, "step": 1180 }, { "epoch": 0.03134641409534973, "grad_norm": 3.777423858642578, "learning_rate": 1.9999909019739482e-05, "loss": 0.18225769996643065, "step": 1190 }, { "epoch": 0.03160982933984847, "grad_norm": 5.451104164123535, "learning_rate": 1.9999868988512265e-05, "loss": 0.17332603931427001, "step": 1200 }, { "epoch": 0.031873244584347205, "grad_norm": 13.987212181091309, "learning_rate": 1.9999821678948964e-05, "loss": 0.17774345874786376, "step": 1210 }, { "epoch": 0.03213665982884595, "grad_norm": 5.783534526824951, "learning_rate": 1.999976709108402e-05, "loss": 0.18999624252319336, "step": 1220 }, { "epoch": 0.032400075073344685, "grad_norm": 6.749079704284668, "learning_rate": 1.9999705224957156e-05, "loss": 0.19959557056427002, "step": 1230 }, { "epoch": 0.03266349031784342, "grad_norm": 12.250381469726562, "learning_rate": 1.9999636080613406e-05, "loss": 0.1927339792251587, "step": 1240 }, { "epoch": 0.03292690556234216, "grad_norm": 5.235373497009277, "learning_rate": 1.9999559658103097e-05, "loss": 0.19090607166290283, "step": 1250 }, { "epoch": 0.033190320806840895, "grad_norm": 5.766205310821533, "learning_rate": 1.9999475957481848e-05, "loss": 0.19557932615280152, "step": 1260 }, { "epoch": 0.03345373605133963, "grad_norm": 1.800480604171753, "learning_rate": 1.999938497881059e-05, "loss": 0.19995884895324706, "step": 1270 }, { "epoch": 0.03371715129583837, "grad_norm": 3.016634702682495, "learning_rate": 1.999928672215553e-05, "loss": 0.1851441502571106, "step": 1280 }, { "epoch": 0.033980566540337105, "grad_norm": 2.221158981323242, "learning_rate": 1.9999181187588187e-05, "loss": 0.19578487873077394, "step": 1290 }, { "epoch": 0.03424398178483584, "grad_norm": 3.3184025287628174, "learning_rate": 1.9999068375185372e-05, "loss": 0.17817424535751342, "step": 1300 }, { "epoch": 0.03450739702933458, "grad_norm": 8.12085247039795, "learning_rate": 1.99989482850292e-05, "loss": 0.18440167903900145, "step": 1310 }, { "epoch": 0.034770812273833315, "grad_norm": 2.410979747772217, "learning_rate": 1.9998820917207076e-05, "loss": 0.17832350730895996, "step": 1320 }, { "epoch": 0.03503422751833205, "grad_norm": 3.170703649520874, "learning_rate": 1.99986862718117e-05, "loss": 0.1825875997543335, "step": 1330 }, { "epoch": 0.03529764276283079, "grad_norm": 4.95659875869751, "learning_rate": 1.999854434894108e-05, "loss": 0.18885856866836548, "step": 1340 }, { "epoch": 0.03556105800732953, "grad_norm": 3.4830992221832275, "learning_rate": 1.9998395148698505e-05, "loss": 0.18464934825897217, "step": 1350 }, { "epoch": 0.03582447325182827, "grad_norm": 10.097084045410156, "learning_rate": 1.9998238671192575e-05, "loss": 0.1849837303161621, "step": 1360 }, { "epoch": 0.036087888496327006, "grad_norm": 4.518742561340332, "learning_rate": 1.999807491653718e-05, "loss": 0.18392249345779418, "step": 1370 }, { "epoch": 0.03635130374082574, "grad_norm": 4.247766971588135, "learning_rate": 1.9997903884851503e-05, "loss": 0.1806490421295166, "step": 1380 }, { "epoch": 0.03661471898532448, "grad_norm": 3.1550521850585938, "learning_rate": 1.999772557626004e-05, "loss": 0.17513761520385743, "step": 1390 }, { "epoch": 0.036878134229823216, "grad_norm": 3.375330924987793, "learning_rate": 1.9997539990892555e-05, "loss": 0.16807618141174316, "step": 1400 }, { "epoch": 0.03714154947432195, "grad_norm": 5.480328559875488, "learning_rate": 1.9997347128884138e-05, "loss": 0.18065738677978516, "step": 1410 }, { "epoch": 0.03740496471882069, "grad_norm": 2.3564560413360596, "learning_rate": 1.9997146990375162e-05, "loss": 0.18996598720550537, "step": 1420 }, { "epoch": 0.037668379963319426, "grad_norm": 4.318639278411865, "learning_rate": 1.999693957551129e-05, "loss": 0.17786145210266113, "step": 1430 }, { "epoch": 0.03793179520781816, "grad_norm": 5.585061073303223, "learning_rate": 1.999672488444349e-05, "loss": 0.173331356048584, "step": 1440 }, { "epoch": 0.0381952104523169, "grad_norm": 4.483724594116211, "learning_rate": 1.999650291732802e-05, "loss": 0.1946944236755371, "step": 1450 }, { "epoch": 0.038458625696815636, "grad_norm": 2.2546629905700684, "learning_rate": 1.9996273674326446e-05, "loss": 0.18732209205627443, "step": 1460 }, { "epoch": 0.03872204094131437, "grad_norm": 2.473327159881592, "learning_rate": 1.9996037155605615e-05, "loss": 0.17481160163879395, "step": 1470 }, { "epoch": 0.03898545618581312, "grad_norm": 4.0760626792907715, "learning_rate": 1.9995793361337676e-05, "loss": 0.1811249852180481, "step": 1480 }, { "epoch": 0.03924887143031185, "grad_norm": 5.340445518493652, "learning_rate": 1.9995542291700075e-05, "loss": 0.1867058515548706, "step": 1490 }, { "epoch": 0.03951228667481059, "grad_norm": 3.857970952987671, "learning_rate": 1.9995283946875548e-05, "loss": 0.17352583408355712, "step": 1500 }, { "epoch": 0.03977570191930933, "grad_norm": 5.456820964813232, "learning_rate": 1.999501832705213e-05, "loss": 0.1639024019241333, "step": 1510 }, { "epoch": 0.04003911716380806, "grad_norm": 2.0374743938446045, "learning_rate": 1.9994745432423154e-05, "loss": 0.1717957615852356, "step": 1520 }, { "epoch": 0.0403025324083068, "grad_norm": 4.362387657165527, "learning_rate": 1.9994465263187242e-05, "loss": 0.1657761812210083, "step": 1530 }, { "epoch": 0.04056594765280554, "grad_norm": 5.010385990142822, "learning_rate": 1.9994177819548313e-05, "loss": 0.19325563907623292, "step": 1540 }, { "epoch": 0.040829362897304274, "grad_norm": 3.443777561187744, "learning_rate": 1.999388310171558e-05, "loss": 0.18690500259399415, "step": 1550 }, { "epoch": 0.04109277814180301, "grad_norm": 4.828578948974609, "learning_rate": 1.9993581109903557e-05, "loss": 0.1676480531692505, "step": 1560 }, { "epoch": 0.04135619338630175, "grad_norm": 7.6464457511901855, "learning_rate": 1.999327184433204e-05, "loss": 0.18241802453994752, "step": 1570 }, { "epoch": 0.041619608630800484, "grad_norm": 2.758058547973633, "learning_rate": 1.9992955305226127e-05, "loss": 0.17461131811141967, "step": 1580 }, { "epoch": 0.04188302387529922, "grad_norm": 2.0377120971679688, "learning_rate": 1.9992631492816214e-05, "loss": 0.1844525933265686, "step": 1590 }, { "epoch": 0.04214643911979796, "grad_norm": 2.7191989421844482, "learning_rate": 1.9992300407337977e-05, "loss": 0.19686455726623536, "step": 1600 }, { "epoch": 0.0424098543642967, "grad_norm": 8.818717002868652, "learning_rate": 1.9991962049032403e-05, "loss": 0.17312088012695312, "step": 1610 }, { "epoch": 0.04267326960879544, "grad_norm": 7.503368377685547, "learning_rate": 1.9991616418145757e-05, "loss": 0.17926650047302245, "step": 1620 }, { "epoch": 0.042936684853294174, "grad_norm": 3.215280532836914, "learning_rate": 1.9991263514929606e-05, "loss": 0.1707767128944397, "step": 1630 }, { "epoch": 0.04320010009779291, "grad_norm": 7.286446571350098, "learning_rate": 1.999090333964081e-05, "loss": 0.18864471912384034, "step": 1640 }, { "epoch": 0.04346351534229165, "grad_norm": 2.3259575366973877, "learning_rate": 1.9990535892541524e-05, "loss": 0.17362875938415528, "step": 1650 }, { "epoch": 0.043726930586790384, "grad_norm": 4.09296989440918, "learning_rate": 1.999016117389918e-05, "loss": 0.19721035957336425, "step": 1660 }, { "epoch": 0.04399034583128912, "grad_norm": 4.512185096740723, "learning_rate": 1.998977918398653e-05, "loss": 0.18918107748031615, "step": 1670 }, { "epoch": 0.04425376107578786, "grad_norm": 2.5933544635772705, "learning_rate": 1.9989389923081587e-05, "loss": 0.16829257011413573, "step": 1680 }, { "epoch": 0.044517176320286594, "grad_norm": 2.9502041339874268, "learning_rate": 1.998899339146768e-05, "loss": 0.17024123668670654, "step": 1690 }, { "epoch": 0.04478059156478533, "grad_norm": 2.2272887229919434, "learning_rate": 1.9988589589433423e-05, "loss": 0.1875399351119995, "step": 1700 }, { "epoch": 0.04504400680928407, "grad_norm": 5.079029560089111, "learning_rate": 1.9988178517272717e-05, "loss": 0.18371498584747314, "step": 1710 }, { "epoch": 0.045307422053782805, "grad_norm": 2.818305492401123, "learning_rate": 1.998776017528476e-05, "loss": 0.18279367685317993, "step": 1720 }, { "epoch": 0.04557083729828155, "grad_norm": 4.5685858726501465, "learning_rate": 1.9987334563774038e-05, "loss": 0.18833543062210084, "step": 1730 }, { "epoch": 0.045834252542780285, "grad_norm": 6.7735676765441895, "learning_rate": 1.998690168305033e-05, "loss": 0.1772863745689392, "step": 1740 }, { "epoch": 0.04609766778727902, "grad_norm": 2.2192695140838623, "learning_rate": 1.9986461533428708e-05, "loss": 0.1923602342605591, "step": 1750 }, { "epoch": 0.04636108303177776, "grad_norm": 4.248669147491455, "learning_rate": 1.9986014115229525e-05, "loss": 0.1891528606414795, "step": 1760 }, { "epoch": 0.046624498276276495, "grad_norm": 2.019760847091675, "learning_rate": 1.998555942877844e-05, "loss": 0.18973848819732667, "step": 1770 }, { "epoch": 0.04688791352077523, "grad_norm": 3.632713794708252, "learning_rate": 1.9985097474406387e-05, "loss": 0.1785494327545166, "step": 1780 }, { "epoch": 0.04715132876527397, "grad_norm": 2.2787046432495117, "learning_rate": 1.99846282524496e-05, "loss": 0.16764893531799316, "step": 1790 }, { "epoch": 0.047414744009772705, "grad_norm": 4.70776891708374, "learning_rate": 1.9984151763249598e-05, "loss": 0.17963919639587403, "step": 1800 }, { "epoch": 0.04767815925427144, "grad_norm": 2.926903247833252, "learning_rate": 1.9983668007153185e-05, "loss": 0.18661699295043946, "step": 1810 }, { "epoch": 0.04794157449877018, "grad_norm": 2.6325504779815674, "learning_rate": 1.998317698451247e-05, "loss": 0.1757776379585266, "step": 1820 }, { "epoch": 0.048204989743268915, "grad_norm": 2.641555070877075, "learning_rate": 1.9982678695684833e-05, "loss": 0.1776546597480774, "step": 1830 }, { "epoch": 0.04846840498776765, "grad_norm": 2.13979434967041, "learning_rate": 1.998217314103295e-05, "loss": 0.18397865295410157, "step": 1840 }, { "epoch": 0.04873182023226639, "grad_norm": 2.386775493621826, "learning_rate": 1.998166032092479e-05, "loss": 0.1841289520263672, "step": 1850 }, { "epoch": 0.04899523547676513, "grad_norm": 2.1079599857330322, "learning_rate": 1.99811402357336e-05, "loss": 0.16743483543395996, "step": 1860 }, { "epoch": 0.04925865072126387, "grad_norm": 3.953768014907837, "learning_rate": 1.9980612885837925e-05, "loss": 0.1867035746574402, "step": 1870 }, { "epoch": 0.049522065965762606, "grad_norm": 2.310353994369507, "learning_rate": 1.998007827162159e-05, "loss": 0.17433682680130005, "step": 1880 }, { "epoch": 0.04978548121026134, "grad_norm": 4.993059158325195, "learning_rate": 1.9979536393473715e-05, "loss": 0.18066301345825195, "step": 1890 }, { "epoch": 0.05004889645476008, "grad_norm": 3.4257237911224365, "learning_rate": 1.9978987251788696e-05, "loss": 0.17976889610290528, "step": 1900 }, { "epoch": 0.050312311699258816, "grad_norm": 4.569813251495361, "learning_rate": 1.997843084696623e-05, "loss": 0.184104585647583, "step": 1910 }, { "epoch": 0.05057572694375755, "grad_norm": 5.255149841308594, "learning_rate": 1.9977867179411282e-05, "loss": 0.15691514015197755, "step": 1920 }, { "epoch": 0.05083914218825629, "grad_norm": 3.0875601768493652, "learning_rate": 1.997729624953412e-05, "loss": 0.1828230619430542, "step": 1930 }, { "epoch": 0.051102557432755026, "grad_norm": 1.8998608589172363, "learning_rate": 1.997671805775029e-05, "loss": 0.18237528800964356, "step": 1940 }, { "epoch": 0.05136597267725376, "grad_norm": 4.775116920471191, "learning_rate": 1.9976132604480625e-05, "loss": 0.1786854386329651, "step": 1950 }, { "epoch": 0.0516293879217525, "grad_norm": 3.480067491531372, "learning_rate": 1.997553989015125e-05, "loss": 0.18951246738433838, "step": 1960 }, { "epoch": 0.051892803166251236, "grad_norm": 4.14471960067749, "learning_rate": 1.9974939915193557e-05, "loss": 0.1749230742454529, "step": 1970 }, { "epoch": 0.05215621841074997, "grad_norm": 1.9808193445205688, "learning_rate": 1.9974332680044237e-05, "loss": 0.18287246227264403, "step": 1980 }, { "epoch": 0.05241963365524872, "grad_norm": 2.116323947906494, "learning_rate": 1.9973718185145262e-05, "loss": 0.16723763942718506, "step": 1990 }, { "epoch": 0.05268304889974745, "grad_norm": 5.416147708892822, "learning_rate": 1.997309643094389e-05, "loss": 0.18068946599960328, "step": 2000 }, { "epoch": 0.05294646414424619, "grad_norm": 0.990854799747467, "learning_rate": 1.997246741789266e-05, "loss": 0.17617056369781495, "step": 2010 }, { "epoch": 0.05320987938874493, "grad_norm": 3.6900405883789062, "learning_rate": 1.99718311464494e-05, "loss": 0.1899439811706543, "step": 2020 }, { "epoch": 0.05347329463324366, "grad_norm": 2.7243618965148926, "learning_rate": 1.9971187617077204e-05, "loss": 0.1794942855834961, "step": 2030 }, { "epoch": 0.0537367098777424, "grad_norm": 3.1074182987213135, "learning_rate": 1.9970536830244466e-05, "loss": 0.17568942308425903, "step": 2040 }, { "epoch": 0.05400012512224114, "grad_norm": 1.9364631175994873, "learning_rate": 1.9969878786424863e-05, "loss": 0.16960790157318115, "step": 2050 }, { "epoch": 0.054263540366739874, "grad_norm": 2.275014877319336, "learning_rate": 1.9969213486097335e-05, "loss": 0.18705902099609376, "step": 2060 }, { "epoch": 0.05452695561123861, "grad_norm": 5.539295673370361, "learning_rate": 1.996854092974613e-05, "loss": 0.18483871221542358, "step": 2070 }, { "epoch": 0.05479037085573735, "grad_norm": 4.470524787902832, "learning_rate": 1.9967861117860754e-05, "loss": 0.16967298984527587, "step": 2080 }, { "epoch": 0.055053786100236084, "grad_norm": 2.4366872310638428, "learning_rate": 1.9967174050936008e-05, "loss": 0.1789120078086853, "step": 2090 }, { "epoch": 0.05531720134473482, "grad_norm": 2.0359723567962646, "learning_rate": 1.9966479729471966e-05, "loss": 0.18926808834075928, "step": 2100 }, { "epoch": 0.05558061658923356, "grad_norm": 5.003600597381592, "learning_rate": 1.9965778153973986e-05, "loss": 0.16884214878082277, "step": 2110 }, { "epoch": 0.0558440318337323, "grad_norm": 2.5042009353637695, "learning_rate": 1.9965069324952703e-05, "loss": 0.17595038414001465, "step": 2120 }, { "epoch": 0.05610744707823104, "grad_norm": 6.560398101806641, "learning_rate": 1.9964353242924038e-05, "loss": 0.18250911235809325, "step": 2130 }, { "epoch": 0.056370862322729774, "grad_norm": 2.504173994064331, "learning_rate": 1.9963629908409184e-05, "loss": 0.1776634693145752, "step": 2140 }, { "epoch": 0.05663427756722851, "grad_norm": 2.4536001682281494, "learning_rate": 1.9962899321934616e-05, "loss": 0.16339056491851806, "step": 2150 }, { "epoch": 0.05689769281172725, "grad_norm": 2.769205093383789, "learning_rate": 1.9962161484032085e-05, "loss": 0.15990519523620605, "step": 2160 }, { "epoch": 0.057161108056225984, "grad_norm": 3.114741325378418, "learning_rate": 1.9961416395238623e-05, "loss": 0.15324027538299562, "step": 2170 }, { "epoch": 0.05742452330072472, "grad_norm": 2.765859842300415, "learning_rate": 1.9960664056096534e-05, "loss": 0.19383933544158935, "step": 2180 }, { "epoch": 0.05768793854522346, "grad_norm": 7.737806797027588, "learning_rate": 1.9959904467153408e-05, "loss": 0.19174882173538207, "step": 2190 }, { "epoch": 0.057951353789722194, "grad_norm": 1.8162436485290527, "learning_rate": 1.99591376289621e-05, "loss": 0.16712512969970703, "step": 2200 }, { "epoch": 0.05821476903422093, "grad_norm": 3.9195947647094727, "learning_rate": 1.9958363542080758e-05, "loss": 0.17442030906677247, "step": 2210 }, { "epoch": 0.05847818427871967, "grad_norm": 3.0483009815216064, "learning_rate": 1.9957582207072784e-05, "loss": 0.1668136954307556, "step": 2220 }, { "epoch": 0.058741599523218405, "grad_norm": 4.592332363128662, "learning_rate": 1.9956793624506875e-05, "loss": 0.17963871955871583, "step": 2230 }, { "epoch": 0.05900501476771714, "grad_norm": 5.714788913726807, "learning_rate": 1.9955997794956994e-05, "loss": 0.16540656089782715, "step": 2240 }, { "epoch": 0.059268430012215885, "grad_norm": 4.5298333168029785, "learning_rate": 1.995519471900238e-05, "loss": 0.1789555788040161, "step": 2250 }, { "epoch": 0.05953184525671462, "grad_norm": 2.182401180267334, "learning_rate": 1.9954384397227545e-05, "loss": 0.1744832754135132, "step": 2260 }, { "epoch": 0.05979526050121336, "grad_norm": 5.092346668243408, "learning_rate": 1.9953566830222276e-05, "loss": 0.1687922239303589, "step": 2270 }, { "epoch": 0.060058675745712095, "grad_norm": 3.96846079826355, "learning_rate": 1.9952742018581634e-05, "loss": 0.17963365316390992, "step": 2280 }, { "epoch": 0.06032209099021083, "grad_norm": 3.5978212356567383, "learning_rate": 1.9951909962905956e-05, "loss": 0.16627718210220338, "step": 2290 }, { "epoch": 0.06058550623470957, "grad_norm": 5.0072407722473145, "learning_rate": 1.9951070663800843e-05, "loss": 0.16901774406433107, "step": 2300 }, { "epoch": 0.060848921479208305, "grad_norm": 1.8899542093276978, "learning_rate": 1.995022412187718e-05, "loss": 0.16940083503723144, "step": 2310 }, { "epoch": 0.06111233672370704, "grad_norm": 4.134581089019775, "learning_rate": 1.9949370337751105e-05, "loss": 0.17897541522979737, "step": 2320 }, { "epoch": 0.06137575196820578, "grad_norm": 3.6975278854370117, "learning_rate": 1.994850931204405e-05, "loss": 0.17508645057678224, "step": 2330 }, { "epoch": 0.061639167212704515, "grad_norm": 7.860805034637451, "learning_rate": 1.99476410453827e-05, "loss": 0.18218538761138917, "step": 2340 }, { "epoch": 0.06190258245720325, "grad_norm": 5.773592472076416, "learning_rate": 1.994676553839902e-05, "loss": 0.18750149011611938, "step": 2350 }, { "epoch": 0.06216599770170199, "grad_norm": 3.5945394039154053, "learning_rate": 1.9945882791730246e-05, "loss": 0.18385186195373535, "step": 2360 }, { "epoch": 0.06242941294620073, "grad_norm": 5.445847988128662, "learning_rate": 1.994499280601887e-05, "loss": 0.17006127834320067, "step": 2370 }, { "epoch": 0.06269282819069946, "grad_norm": 4.305008888244629, "learning_rate": 1.994409558191267e-05, "loss": 0.16458984613418579, "step": 2380 }, { "epoch": 0.0629562434351982, "grad_norm": 5.459083557128906, "learning_rate": 1.9943191120064672e-05, "loss": 0.1809215545654297, "step": 2390 }, { "epoch": 0.06321965867969694, "grad_norm": 2.806043863296509, "learning_rate": 1.99422794211332e-05, "loss": 0.16609575748443603, "step": 2400 }, { "epoch": 0.06348307392419568, "grad_norm": 15.248547554016113, "learning_rate": 1.9941360485781814e-05, "loss": 0.17418453693389893, "step": 2410 }, { "epoch": 0.06374648916869441, "grad_norm": 2.375356674194336, "learning_rate": 1.9940434314679363e-05, "loss": 0.17240833044052123, "step": 2420 }, { "epoch": 0.06400990441319315, "grad_norm": 3.497227430343628, "learning_rate": 1.9939500908499947e-05, "loss": 0.17027150392532348, "step": 2430 }, { "epoch": 0.0642733196576919, "grad_norm": 8.57918643951416, "learning_rate": 1.9938560267922948e-05, "loss": 0.18742706775665283, "step": 2440 }, { "epoch": 0.06453673490219063, "grad_norm": 2.4534220695495605, "learning_rate": 1.9937612393632995e-05, "loss": 0.15978662967681884, "step": 2450 }, { "epoch": 0.06480015014668937, "grad_norm": 2.396292209625244, "learning_rate": 1.9936657286320003e-05, "loss": 0.16072057485580443, "step": 2460 }, { "epoch": 0.0650635653911881, "grad_norm": 2.4367740154266357, "learning_rate": 1.993569494667913e-05, "loss": 0.1768656611442566, "step": 2470 }, { "epoch": 0.06532698063568684, "grad_norm": 3.9994521141052246, "learning_rate": 1.9934725375410812e-05, "loss": 0.1847400665283203, "step": 2480 }, { "epoch": 0.06559039588018557, "grad_norm": 2.0416882038116455, "learning_rate": 1.9933748573220745e-05, "loss": 0.1834241271018982, "step": 2490 }, { "epoch": 0.06585381112468432, "grad_norm": 5.564525127410889, "learning_rate": 1.993276454081989e-05, "loss": 0.1700831890106201, "step": 2500 }, { "epoch": 0.06611722636918305, "grad_norm": 3.658693552017212, "learning_rate": 1.9931773278924467e-05, "loss": 0.1774335026741028, "step": 2510 }, { "epoch": 0.06638064161368179, "grad_norm": 3.581808567047119, "learning_rate": 1.9930774788255957e-05, "loss": 0.1809831142425537, "step": 2520 }, { "epoch": 0.06664405685818052, "grad_norm": 2.124931812286377, "learning_rate": 1.992976906954111e-05, "loss": 0.18324557542800904, "step": 2530 }, { "epoch": 0.06690747210267926, "grad_norm": 2.8153276443481445, "learning_rate": 1.992875612351192e-05, "loss": 0.16865072250366211, "step": 2540 }, { "epoch": 0.067170887347178, "grad_norm": 1.2017066478729248, "learning_rate": 1.9927735950905663e-05, "loss": 0.15883004665374756, "step": 2550 }, { "epoch": 0.06743430259167674, "grad_norm": 1.2406625747680664, "learning_rate": 1.9926708552464865e-05, "loss": 0.19160606861114501, "step": 2560 }, { "epoch": 0.06769771783617548, "grad_norm": 5.941833972930908, "learning_rate": 1.9925673928937307e-05, "loss": 0.18246570825576783, "step": 2570 }, { "epoch": 0.06796113308067421, "grad_norm": 2.528409719467163, "learning_rate": 1.9924632081076032e-05, "loss": 0.17777807712554933, "step": 2580 }, { "epoch": 0.06822454832517295, "grad_norm": 2.4376869201660156, "learning_rate": 1.9923583009639344e-05, "loss": 0.1887561559677124, "step": 2590 }, { "epoch": 0.06848796356967168, "grad_norm": 3.958491086959839, "learning_rate": 1.9922526715390802e-05, "loss": 0.17350794076919557, "step": 2600 }, { "epoch": 0.06875137881417043, "grad_norm": 4.694818496704102, "learning_rate": 1.9921463199099222e-05, "loss": 0.1769506335258484, "step": 2610 }, { "epoch": 0.06901479405866916, "grad_norm": 2.20530104637146, "learning_rate": 1.992039246153868e-05, "loss": 0.18529040813446046, "step": 2620 }, { "epoch": 0.0692782093031679, "grad_norm": 4.683708667755127, "learning_rate": 1.99193145034885e-05, "loss": 0.1675152063369751, "step": 2630 }, { "epoch": 0.06954162454766663, "grad_norm": 2.07639479637146, "learning_rate": 1.9918229325733267e-05, "loss": 0.17177026271820067, "step": 2640 }, { "epoch": 0.06980503979216537, "grad_norm": 1.360260009765625, "learning_rate": 1.9917136929062824e-05, "loss": 0.17395167350769042, "step": 2650 }, { "epoch": 0.0700684550366641, "grad_norm": 4.556027889251709, "learning_rate": 1.991603731427226e-05, "loss": 0.1741271734237671, "step": 2660 }, { "epoch": 0.07033187028116285, "grad_norm": 6.990688800811768, "learning_rate": 1.9914930482161928e-05, "loss": 0.17015174627304078, "step": 2670 }, { "epoch": 0.07059528552566158, "grad_norm": 2.772233486175537, "learning_rate": 1.9913816433537425e-05, "loss": 0.1593541145324707, "step": 2680 }, { "epoch": 0.07085870077016032, "grad_norm": 3.284693956375122, "learning_rate": 1.99126951692096e-05, "loss": 0.1752576470375061, "step": 2690 }, { "epoch": 0.07112211601465906, "grad_norm": 2.5792171955108643, "learning_rate": 1.991156668999456e-05, "loss": 0.1866455554962158, "step": 2700 }, { "epoch": 0.0713855312591578, "grad_norm": 2.276620388031006, "learning_rate": 1.9910430996713668e-05, "loss": 0.16820745468139647, "step": 2710 }, { "epoch": 0.07164894650365654, "grad_norm": 4.047654151916504, "learning_rate": 1.9909288090193518e-05, "loss": 0.18410875797271728, "step": 2720 }, { "epoch": 0.07191236174815527, "grad_norm": 2.015934467315674, "learning_rate": 1.9908137971265977e-05, "loss": 0.18779525756835938, "step": 2730 }, { "epoch": 0.07217577699265401, "grad_norm": 3.3355324268341064, "learning_rate": 1.9906980640768146e-05, "loss": 0.17896010875701904, "step": 2740 }, { "epoch": 0.07243919223715274, "grad_norm": 5.11514949798584, "learning_rate": 1.9905816099542382e-05, "loss": 0.17484050989151, "step": 2750 }, { "epoch": 0.07270260748165149, "grad_norm": 4.03486442565918, "learning_rate": 1.9904644348436282e-05, "loss": 0.1748443603515625, "step": 2760 }, { "epoch": 0.07296602272615021, "grad_norm": 3.616081714630127, "learning_rate": 1.990346538830271e-05, "loss": 0.16918439865112306, "step": 2770 }, { "epoch": 0.07322943797064896, "grad_norm": 3.5150482654571533, "learning_rate": 1.9902279219999752e-05, "loss": 0.16574454307556152, "step": 2780 }, { "epoch": 0.07349285321514769, "grad_norm": 4.69978141784668, "learning_rate": 1.9901085844390757e-05, "loss": 0.1701863169670105, "step": 2790 }, { "epoch": 0.07375626845964643, "grad_norm": 6.424964427947998, "learning_rate": 1.9899885262344316e-05, "loss": 0.17672109603881836, "step": 2800 }, { "epoch": 0.07401968370414516, "grad_norm": 3.57743501663208, "learning_rate": 1.9898677474734266e-05, "loss": 0.16964842081069947, "step": 2810 }, { "epoch": 0.0742830989486439, "grad_norm": 2.7270941734313965, "learning_rate": 1.9897462482439684e-05, "loss": 0.1675887107849121, "step": 2820 }, { "epoch": 0.07454651419314265, "grad_norm": 5.135262489318848, "learning_rate": 1.9896240286344892e-05, "loss": 0.19278861284255983, "step": 2830 }, { "epoch": 0.07480992943764138, "grad_norm": 4.415496826171875, "learning_rate": 1.989501088733946e-05, "loss": 0.17518354654312135, "step": 2840 }, { "epoch": 0.07507334468214012, "grad_norm": 2.5478262901306152, "learning_rate": 1.9893774286318198e-05, "loss": 0.175394868850708, "step": 2850 }, { "epoch": 0.07533675992663885, "grad_norm": 9.249306678771973, "learning_rate": 1.9892530484181155e-05, "loss": 0.1855395555496216, "step": 2860 }, { "epoch": 0.0756001751711376, "grad_norm": 3.3610517978668213, "learning_rate": 1.989127948183363e-05, "loss": 0.17591773271560668, "step": 2870 }, { "epoch": 0.07586359041563633, "grad_norm": 4.558925151824951, "learning_rate": 1.9890021280186148e-05, "loss": 0.19274908304214478, "step": 2880 }, { "epoch": 0.07612700566013507, "grad_norm": 2.2421505451202393, "learning_rate": 1.9888755880154485e-05, "loss": 0.16077778339385987, "step": 2890 }, { "epoch": 0.0763904209046338, "grad_norm": 4.637321949005127, "learning_rate": 1.988748328265966e-05, "loss": 0.15136784315109253, "step": 2900 }, { "epoch": 0.07665383614913254, "grad_norm": 2.2762718200683594, "learning_rate": 1.9886203488627918e-05, "loss": 0.18226230144500732, "step": 2910 }, { "epoch": 0.07691725139363127, "grad_norm": 3.7826597690582275, "learning_rate": 1.988491649899075e-05, "loss": 0.17562289237976075, "step": 2920 }, { "epoch": 0.07718066663813002, "grad_norm": 5.500918865203857, "learning_rate": 1.988362231468488e-05, "loss": 0.16804947853088378, "step": 2930 }, { "epoch": 0.07744408188262875, "grad_norm": 1.9142930507659912, "learning_rate": 1.9882320936652278e-05, "loss": 0.17871756553649903, "step": 2940 }, { "epoch": 0.07770749712712749, "grad_norm": 2.129884719848633, "learning_rate": 1.9881012365840138e-05, "loss": 0.1694040298461914, "step": 2950 }, { "epoch": 0.07797091237162623, "grad_norm": 2.2781362533569336, "learning_rate": 1.987969660320089e-05, "loss": 0.17015597820281983, "step": 2960 }, { "epoch": 0.07823432761612496, "grad_norm": 2.504615545272827, "learning_rate": 1.9878373649692213e-05, "loss": 0.18511658906936646, "step": 2970 }, { "epoch": 0.0784977428606237, "grad_norm": 1.9262624979019165, "learning_rate": 1.9877043506277004e-05, "loss": 0.18343596458435057, "step": 2980 }, { "epoch": 0.07876115810512244, "grad_norm": 4.18161153793335, "learning_rate": 1.9875706173923396e-05, "loss": 0.16998767852783203, "step": 2990 }, { "epoch": 0.07902457334962118, "grad_norm": 3.188481330871582, "learning_rate": 1.9874361653604765e-05, "loss": 0.15898195505142212, "step": 3000 }, { "epoch": 0.07928798859411991, "grad_norm": 3.4452617168426514, "learning_rate": 1.9873009946299702e-05, "loss": 0.18394674062728883, "step": 3010 }, { "epoch": 0.07955140383861865, "grad_norm": 4.188591480255127, "learning_rate": 1.9871651052992046e-05, "loss": 0.1591788649559021, "step": 3020 }, { "epoch": 0.07981481908311738, "grad_norm": 4.648423671722412, "learning_rate": 1.987028497467085e-05, "loss": 0.15908188819885255, "step": 3030 }, { "epoch": 0.08007823432761613, "grad_norm": 2.544194221496582, "learning_rate": 1.986891171233041e-05, "loss": 0.16518454551696776, "step": 3040 }, { "epoch": 0.08034164957211486, "grad_norm": 4.835196018218994, "learning_rate": 1.9867531266970247e-05, "loss": 0.1783139705657959, "step": 3050 }, { "epoch": 0.0806050648166136, "grad_norm": 1.3829095363616943, "learning_rate": 1.9866143639595107e-05, "loss": 0.1787714719772339, "step": 3060 }, { "epoch": 0.08086848006111233, "grad_norm": 3.209656000137329, "learning_rate": 1.9864748831214964e-05, "loss": 0.16357862949371338, "step": 3070 }, { "epoch": 0.08113189530561107, "grad_norm": 2.9779372215270996, "learning_rate": 1.986334684284502e-05, "loss": 0.18394919633865356, "step": 3080 }, { "epoch": 0.08139531055010982, "grad_norm": 5.177148342132568, "learning_rate": 1.98619376755057e-05, "loss": 0.17536817789077758, "step": 3090 }, { "epoch": 0.08165872579460855, "grad_norm": 1.7103697061538696, "learning_rate": 1.986052133022266e-05, "loss": 0.18967607021331787, "step": 3100 }, { "epoch": 0.08192214103910729, "grad_norm": 3.6065218448638916, "learning_rate": 1.985909780802678e-05, "loss": 0.16490695476531983, "step": 3110 }, { "epoch": 0.08218555628360602, "grad_norm": 2.844053030014038, "learning_rate": 1.985766710995416e-05, "loss": 0.16936759948730468, "step": 3120 }, { "epoch": 0.08244897152810476, "grad_norm": 7.438908100128174, "learning_rate": 1.9856229237046115e-05, "loss": 0.17896740436553954, "step": 3130 }, { "epoch": 0.0827123867726035, "grad_norm": 3.4769020080566406, "learning_rate": 1.9854784190349203e-05, "loss": 0.1733659029006958, "step": 3140 }, { "epoch": 0.08297580201710224, "grad_norm": 7.00600004196167, "learning_rate": 1.985333197091518e-05, "loss": 0.18061583042144774, "step": 3150 }, { "epoch": 0.08323921726160097, "grad_norm": 6.197855472564697, "learning_rate": 1.9851872579801044e-05, "loss": 0.17196264266967773, "step": 3160 }, { "epoch": 0.08350263250609971, "grad_norm": 1.1776789426803589, "learning_rate": 1.9850406018068997e-05, "loss": 0.16351239681243895, "step": 3170 }, { "epoch": 0.08376604775059844, "grad_norm": 4.085625171661377, "learning_rate": 1.984893228678647e-05, "loss": 0.15729930400848388, "step": 3180 }, { "epoch": 0.08402946299509718, "grad_norm": 3.416560173034668, "learning_rate": 1.9847451387026103e-05, "loss": 0.16842066049575805, "step": 3190 }, { "epoch": 0.08429287823959591, "grad_norm": 4.39070463180542, "learning_rate": 1.9845963319865758e-05, "loss": 0.1798922061920166, "step": 3200 }, { "epoch": 0.08455629348409466, "grad_norm": 2.465787172317505, "learning_rate": 1.9844468086388523e-05, "loss": 0.16168004274368286, "step": 3210 }, { "epoch": 0.0848197087285934, "grad_norm": 5.6526875495910645, "learning_rate": 1.9842965687682686e-05, "loss": 0.16240211725234985, "step": 3220 }, { "epoch": 0.08508312397309213, "grad_norm": 3.0739715099334717, "learning_rate": 1.9841456124841763e-05, "loss": 0.18028011322021484, "step": 3230 }, { "epoch": 0.08534653921759087, "grad_norm": 3.5990395545959473, "learning_rate": 1.983993939896447e-05, "loss": 0.16247918605804443, "step": 3240 }, { "epoch": 0.0856099544620896, "grad_norm": 3.1962015628814697, "learning_rate": 1.9838415511154755e-05, "loss": 0.18511825799942017, "step": 3250 }, { "epoch": 0.08587336970658835, "grad_norm": 4.571393966674805, "learning_rate": 1.9836884462521768e-05, "loss": 0.16345113515853882, "step": 3260 }, { "epoch": 0.08613678495108708, "grad_norm": 1.6657286882400513, "learning_rate": 1.9835346254179865e-05, "loss": 0.17652188539505004, "step": 3270 }, { "epoch": 0.08640020019558582, "grad_norm": 2.1104304790496826, "learning_rate": 1.9833800887248628e-05, "loss": 0.17684825658798217, "step": 3280 }, { "epoch": 0.08666361544008455, "grad_norm": 5.127033710479736, "learning_rate": 1.983224836285284e-05, "loss": 0.17143778800964354, "step": 3290 }, { "epoch": 0.0869270306845833, "grad_norm": 5.995574474334717, "learning_rate": 1.9830688682122494e-05, "loss": 0.1829805016517639, "step": 3300 }, { "epoch": 0.08719044592908202, "grad_norm": 4.853494167327881, "learning_rate": 1.982912184619279e-05, "loss": 0.17200614213943483, "step": 3310 }, { "epoch": 0.08745386117358077, "grad_norm": 7.984829902648926, "learning_rate": 1.9827547856204147e-05, "loss": 0.16227188110351562, "step": 3320 }, { "epoch": 0.0877172764180795, "grad_norm": 3.7791049480438232, "learning_rate": 1.9825966713302172e-05, "loss": 0.17087780237197875, "step": 3330 }, { "epoch": 0.08798069166257824, "grad_norm": 3.5410101413726807, "learning_rate": 1.98243784186377e-05, "loss": 0.16737122535705568, "step": 3340 }, { "epoch": 0.08824410690707699, "grad_norm": 3.8058021068573, "learning_rate": 1.982278297336675e-05, "loss": 0.17722208499908448, "step": 3350 }, { "epoch": 0.08850752215157572, "grad_norm": 1.6982210874557495, "learning_rate": 1.9821180378650563e-05, "loss": 0.17792577743530275, "step": 3360 }, { "epoch": 0.08877093739607446, "grad_norm": 4.606264591217041, "learning_rate": 1.9819570635655577e-05, "loss": 0.17522087097167968, "step": 3370 }, { "epoch": 0.08903435264057319, "grad_norm": 5.5729756355285645, "learning_rate": 1.981795374555343e-05, "loss": 0.16213397979736327, "step": 3380 }, { "epoch": 0.08929776788507193, "grad_norm": 2.2267351150512695, "learning_rate": 1.981632970952096e-05, "loss": 0.166667377948761, "step": 3390 }, { "epoch": 0.08956118312957066, "grad_norm": 1.9641575813293457, "learning_rate": 1.9814698528740222e-05, "loss": 0.17222795486450196, "step": 3400 }, { "epoch": 0.0898245983740694, "grad_norm": 4.566251277923584, "learning_rate": 1.9813060204398448e-05, "loss": 0.16780471801757812, "step": 3410 }, { "epoch": 0.09008801361856814, "grad_norm": 2.8531575202941895, "learning_rate": 1.9811414737688088e-05, "loss": 0.17125431299209595, "step": 3420 }, { "epoch": 0.09035142886306688, "grad_norm": 2.503525495529175, "learning_rate": 1.9809762129806777e-05, "loss": 0.1721203565597534, "step": 3430 }, { "epoch": 0.09061484410756561, "grad_norm": 3.773655891418457, "learning_rate": 1.9808102381957364e-05, "loss": 0.17802207469940184, "step": 3440 }, { "epoch": 0.09087825935206435, "grad_norm": 3.4262375831604004, "learning_rate": 1.980643549534788e-05, "loss": 0.1673513650894165, "step": 3450 }, { "epoch": 0.0911416745965631, "grad_norm": 1.8866558074951172, "learning_rate": 1.980476147119155e-05, "loss": 0.15911438465118408, "step": 3460 }, { "epoch": 0.09140508984106183, "grad_norm": 1.1810684204101562, "learning_rate": 1.980308031070682e-05, "loss": 0.16938039064407348, "step": 3470 }, { "epoch": 0.09166850508556057, "grad_norm": 2.9523589611053467, "learning_rate": 1.9801392015117286e-05, "loss": 0.18596019744873046, "step": 3480 }, { "epoch": 0.0919319203300593, "grad_norm": 3.7992656230926514, "learning_rate": 1.9799696585651782e-05, "loss": 0.15685501098632812, "step": 3490 }, { "epoch": 0.09219533557455804, "grad_norm": 1.1968823671340942, "learning_rate": 1.9797994023544307e-05, "loss": 0.18120188713073732, "step": 3500 }, { "epoch": 0.09245875081905677, "grad_norm": 1.9265352487564087, "learning_rate": 1.979628433003406e-05, "loss": 0.1737501621246338, "step": 3510 }, { "epoch": 0.09272216606355552, "grad_norm": 1.6019021272659302, "learning_rate": 1.9794567506365432e-05, "loss": 0.17019884586334227, "step": 3520 }, { "epoch": 0.09298558130805425, "grad_norm": 3.6709628105163574, "learning_rate": 1.9792843553788e-05, "loss": 0.16490205526351928, "step": 3530 }, { "epoch": 0.09324899655255299, "grad_norm": 3.021487236022949, "learning_rate": 1.9791112473556528e-05, "loss": 0.17663896083831787, "step": 3540 }, { "epoch": 0.09351241179705172, "grad_norm": 1.9023574590682983, "learning_rate": 1.978937426693097e-05, "loss": 0.1718762159347534, "step": 3550 }, { "epoch": 0.09377582704155046, "grad_norm": 2.368898868560791, "learning_rate": 1.9787628935176476e-05, "loss": 0.1721566915512085, "step": 3560 }, { "epoch": 0.0940392422860492, "grad_norm": 4.2968854904174805, "learning_rate": 1.9785876479563367e-05, "loss": 0.17789971828460693, "step": 3570 }, { "epoch": 0.09430265753054794, "grad_norm": 1.6244983673095703, "learning_rate": 1.978411690136716e-05, "loss": 0.1741708993911743, "step": 3580 }, { "epoch": 0.09456607277504668, "grad_norm": 2.748901605606079, "learning_rate": 1.9782350201868547e-05, "loss": 0.16877100467681885, "step": 3590 }, { "epoch": 0.09482948801954541, "grad_norm": 4.2705464363098145, "learning_rate": 1.978057638235341e-05, "loss": 0.16645052433013915, "step": 3600 }, { "epoch": 0.09509290326404415, "grad_norm": 3.2782468795776367, "learning_rate": 1.9778795444112812e-05, "loss": 0.16818060874938964, "step": 3610 }, { "epoch": 0.09535631850854288, "grad_norm": 6.536076545715332, "learning_rate": 1.9777007388443e-05, "loss": 0.16492848396301268, "step": 3620 }, { "epoch": 0.09561973375304163, "grad_norm": 2.4057869911193848, "learning_rate": 1.9775212216645393e-05, "loss": 0.15858101844787598, "step": 3630 }, { "epoch": 0.09588314899754036, "grad_norm": 1.2258137464523315, "learning_rate": 1.977340993002659e-05, "loss": 0.17911961078643798, "step": 3640 }, { "epoch": 0.0961465642420391, "grad_norm": 4.498673915863037, "learning_rate": 1.977160052989839e-05, "loss": 0.17694683074951173, "step": 3650 }, { "epoch": 0.09640997948653783, "grad_norm": 3.265545606613159, "learning_rate": 1.9769784017577734e-05, "loss": 0.1590961217880249, "step": 3660 }, { "epoch": 0.09667339473103657, "grad_norm": 9.40554428100586, "learning_rate": 1.976796039438677e-05, "loss": 0.18539254665374755, "step": 3670 }, { "epoch": 0.0969368099755353, "grad_norm": 5.099091529846191, "learning_rate": 1.9766129661652803e-05, "loss": 0.18205788135528564, "step": 3680 }, { "epoch": 0.09720022522003405, "grad_norm": 5.382017612457275, "learning_rate": 1.9764291820708324e-05, "loss": 0.18011595010757447, "step": 3690 }, { "epoch": 0.09746364046453278, "grad_norm": 2.5891661643981934, "learning_rate": 1.9762446872890984e-05, "loss": 0.16343882083892822, "step": 3700 }, { "epoch": 0.09772705570903152, "grad_norm": 2.9732563495635986, "learning_rate": 1.976059481954363e-05, "loss": 0.17642593383789062, "step": 3710 }, { "epoch": 0.09799047095353026, "grad_norm": 1.3045353889465332, "learning_rate": 1.9758735662014253e-05, "loss": 0.16141705513000487, "step": 3720 }, { "epoch": 0.098253886198029, "grad_norm": 3.4105288982391357, "learning_rate": 1.975686940165604e-05, "loss": 0.1597197413444519, "step": 3730 }, { "epoch": 0.09851730144252774, "grad_norm": 2.3756208419799805, "learning_rate": 1.9754996039827322e-05, "loss": 0.15719451904296874, "step": 3740 }, { "epoch": 0.09878071668702647, "grad_norm": 2.65950870513916, "learning_rate": 1.9753115577891623e-05, "loss": 0.16199564933776855, "step": 3750 }, { "epoch": 0.09904413193152521, "grad_norm": 1.9366179704666138, "learning_rate": 1.9751228017217623e-05, "loss": 0.15806516408920288, "step": 3760 }, { "epoch": 0.09930754717602394, "grad_norm": 2.0314812660217285, "learning_rate": 1.9749333359179166e-05, "loss": 0.1636293888092041, "step": 3770 }, { "epoch": 0.09957096242052269, "grad_norm": 2.4928383827209473, "learning_rate": 1.974743160515527e-05, "loss": 0.16500955820083618, "step": 3780 }, { "epoch": 0.09983437766502141, "grad_norm": 3.04986310005188, "learning_rate": 1.974552275653011e-05, "loss": 0.1725706100463867, "step": 3790 }, { "epoch": 0.10009779290952016, "grad_norm": 3.11575984954834, "learning_rate": 1.974360681469303e-05, "loss": 0.17040719985961914, "step": 3800 }, { "epoch": 0.10036120815401889, "grad_norm": 2.4536232948303223, "learning_rate": 1.9741683781038538e-05, "loss": 0.16992148160934448, "step": 3810 }, { "epoch": 0.10062462339851763, "grad_norm": 3.8847615718841553, "learning_rate": 1.9739753656966294e-05, "loss": 0.17242614030838013, "step": 3820 }, { "epoch": 0.10088803864301636, "grad_norm": 2.950798749923706, "learning_rate": 1.9737816443881134e-05, "loss": 0.17825216054916382, "step": 3830 }, { "epoch": 0.1011514538875151, "grad_norm": 4.090005874633789, "learning_rate": 1.973587214319304e-05, "loss": 0.16333072185516356, "step": 3840 }, { "epoch": 0.10141486913201385, "grad_norm": 2.0875000953674316, "learning_rate": 1.9733920756317157e-05, "loss": 0.17109159231185914, "step": 3850 }, { "epoch": 0.10167828437651258, "grad_norm": 2.2715654373168945, "learning_rate": 1.973196228467379e-05, "loss": 0.16274991035461425, "step": 3860 }, { "epoch": 0.10194169962101132, "grad_norm": 2.2898213863372803, "learning_rate": 1.9729996729688403e-05, "loss": 0.17771718502044678, "step": 3870 }, { "epoch": 0.10220511486551005, "grad_norm": 2.9507739543914795, "learning_rate": 1.9728024092791606e-05, "loss": 0.17696831226348878, "step": 3880 }, { "epoch": 0.1024685301100088, "grad_norm": 1.4754111766815186, "learning_rate": 1.9726044375419172e-05, "loss": 0.17042477130889894, "step": 3890 }, { "epoch": 0.10273194535450753, "grad_norm": 5.40613317489624, "learning_rate": 1.9724057579012026e-05, "loss": 0.1723585844039917, "step": 3900 }, { "epoch": 0.10299536059900627, "grad_norm": 2.37760329246521, "learning_rate": 1.972206370501624e-05, "loss": 0.17405056953430176, "step": 3910 }, { "epoch": 0.103258775843505, "grad_norm": 3.0686140060424805, "learning_rate": 1.9720062754883048e-05, "loss": 0.17442970275878905, "step": 3920 }, { "epoch": 0.10352219108800374, "grad_norm": 1.7428640127182007, "learning_rate": 1.971805473006882e-05, "loss": 0.16731137037277222, "step": 3930 }, { "epoch": 0.10378560633250247, "grad_norm": 2.5261166095733643, "learning_rate": 1.9716039632035086e-05, "loss": 0.16226847171783448, "step": 3940 }, { "epoch": 0.10404902157700122, "grad_norm": 1.864688754081726, "learning_rate": 1.9714017462248523e-05, "loss": 0.19696928262710572, "step": 3950 }, { "epoch": 0.10431243682149995, "grad_norm": 2.6549181938171387, "learning_rate": 1.9711988222180958e-05, "loss": 0.18856461048126222, "step": 3960 }, { "epoch": 0.10457585206599869, "grad_norm": 4.839781284332275, "learning_rate": 1.9709951913309347e-05, "loss": 0.17953338623046874, "step": 3970 }, { "epoch": 0.10483926731049743, "grad_norm": 2.10209584236145, "learning_rate": 1.9707908537115816e-05, "loss": 0.17297787666320802, "step": 3980 }, { "epoch": 0.10510268255499616, "grad_norm": 2.8248238563537598, "learning_rate": 1.9705858095087613e-05, "loss": 0.17693523168563843, "step": 3990 }, { "epoch": 0.1053660977994949, "grad_norm": 2.4041404724121094, "learning_rate": 1.9703800588717144e-05, "loss": 0.15789952278137206, "step": 4000 }, { "epoch": 0.10562951304399364, "grad_norm": 2.4523766040802, "learning_rate": 1.9701736019501948e-05, "loss": 0.1802968978881836, "step": 4010 }, { "epoch": 0.10589292828849238, "grad_norm": 3.753265380859375, "learning_rate": 1.969966438894471e-05, "loss": 0.16564335823059081, "step": 4020 }, { "epoch": 0.10615634353299111, "grad_norm": 2.1847662925720215, "learning_rate": 1.969758569855325e-05, "loss": 0.17150951623916627, "step": 4030 }, { "epoch": 0.10641975877748985, "grad_norm": 1.0530521869659424, "learning_rate": 1.969549994984053e-05, "loss": 0.18100328445434571, "step": 4040 }, { "epoch": 0.10668317402198858, "grad_norm": 4.437838077545166, "learning_rate": 1.9693407144324642e-05, "loss": 0.15388827323913573, "step": 4050 }, { "epoch": 0.10694658926648733, "grad_norm": 3.8867928981781006, "learning_rate": 1.9691307283528834e-05, "loss": 0.18169565200805665, "step": 4060 }, { "epoch": 0.10721000451098606, "grad_norm": 2.8830039501190186, "learning_rate": 1.968920036898146e-05, "loss": 0.17617979049682617, "step": 4070 }, { "epoch": 0.1074734197554848, "grad_norm": 2.779534101486206, "learning_rate": 1.9687086402216033e-05, "loss": 0.1653350830078125, "step": 4080 }, { "epoch": 0.10773683499998353, "grad_norm": 2.1574740409851074, "learning_rate": 1.968496538477118e-05, "loss": 0.1651676893234253, "step": 4090 }, { "epoch": 0.10800025024448227, "grad_norm": 3.5880961418151855, "learning_rate": 1.9682837318190683e-05, "loss": 0.18387491703033448, "step": 4100 }, { "epoch": 0.10826366548898102, "grad_norm": 3.2538559436798096, "learning_rate": 1.968070220402343e-05, "loss": 0.14626176357269288, "step": 4110 }, { "epoch": 0.10852708073347975, "grad_norm": 0.9968183040618896, "learning_rate": 1.967856004382345e-05, "loss": 0.17565104961395264, "step": 4120 }, { "epoch": 0.10879049597797849, "grad_norm": 4.250136852264404, "learning_rate": 1.96764108391499e-05, "loss": 0.16469523906707764, "step": 4130 }, { "epoch": 0.10905391122247722, "grad_norm": 6.090249538421631, "learning_rate": 1.967425459156707e-05, "loss": 0.17661898136138915, "step": 4140 }, { "epoch": 0.10931732646697596, "grad_norm": 1.7353490591049194, "learning_rate": 1.9672091302644363e-05, "loss": 0.17251813411712646, "step": 4150 }, { "epoch": 0.1095807417114747, "grad_norm": 5.175922393798828, "learning_rate": 1.966992097395632e-05, "loss": 0.17210166454315184, "step": 4160 }, { "epoch": 0.10984415695597344, "grad_norm": 1.9386160373687744, "learning_rate": 1.9667743607082593e-05, "loss": 0.1676517367362976, "step": 4170 }, { "epoch": 0.11010757220047217, "grad_norm": 2.6791059970855713, "learning_rate": 1.9665559203607965e-05, "loss": 0.1628250241279602, "step": 4180 }, { "epoch": 0.11037098744497091, "grad_norm": 1.679293155670166, "learning_rate": 1.9663367765122346e-05, "loss": 0.1707902431488037, "step": 4190 }, { "epoch": 0.11063440268946964, "grad_norm": 1.2598458528518677, "learning_rate": 1.966116929322075e-05, "loss": 0.17590756416320802, "step": 4200 }, { "epoch": 0.11089781793396838, "grad_norm": 3.1790931224823, "learning_rate": 1.9658963789503328e-05, "loss": 0.16079995632171631, "step": 4210 }, { "epoch": 0.11116123317846711, "grad_norm": 1.4302773475646973, "learning_rate": 1.9656751255575336e-05, "loss": 0.15136511325836183, "step": 4220 }, { "epoch": 0.11142464842296586, "grad_norm": 3.2195093631744385, "learning_rate": 1.965453169304715e-05, "loss": 0.17317715883255005, "step": 4230 }, { "epoch": 0.1116880636674646, "grad_norm": 3.5319864749908447, "learning_rate": 1.9652305103534265e-05, "loss": 0.1743646740913391, "step": 4240 }, { "epoch": 0.11195147891196333, "grad_norm": 2.8592166900634766, "learning_rate": 1.9650071488657293e-05, "loss": 0.16283891201019288, "step": 4250 }, { "epoch": 0.11221489415646207, "grad_norm": 2.4355392456054688, "learning_rate": 1.964783085004195e-05, "loss": 0.16786746978759765, "step": 4260 }, { "epoch": 0.1124783094009608, "grad_norm": 1.6558443307876587, "learning_rate": 1.9645583189319074e-05, "loss": 0.17752788066864014, "step": 4270 }, { "epoch": 0.11274172464545955, "grad_norm": 5.259485721588135, "learning_rate": 1.9643328508124605e-05, "loss": 0.17741150856018068, "step": 4280 }, { "epoch": 0.11300513988995828, "grad_norm": 4.173855304718018, "learning_rate": 1.9641066808099597e-05, "loss": 0.17621996402740478, "step": 4290 }, { "epoch": 0.11326855513445702, "grad_norm": 1.511672019958496, "learning_rate": 1.963879809089022e-05, "loss": 0.15456709861755372, "step": 4300 }, { "epoch": 0.11353197037895575, "grad_norm": 3.9766323566436768, "learning_rate": 1.9636522358147737e-05, "loss": 0.1749556303024292, "step": 4310 }, { "epoch": 0.1137953856234545, "grad_norm": 3.454371213912964, "learning_rate": 1.9634239611528527e-05, "loss": 0.1724604368209839, "step": 4320 }, { "epoch": 0.11405880086795323, "grad_norm": 1.8875199556350708, "learning_rate": 1.9631949852694073e-05, "loss": 0.17612724304199218, "step": 4330 }, { "epoch": 0.11432221611245197, "grad_norm": 3.3896474838256836, "learning_rate": 1.9629653083310956e-05, "loss": 0.17624471187591553, "step": 4340 }, { "epoch": 0.1145856313569507, "grad_norm": 3.1960549354553223, "learning_rate": 1.9627349305050867e-05, "loss": 0.18193154335021972, "step": 4350 }, { "epoch": 0.11484904660144944, "grad_norm": 3.844573974609375, "learning_rate": 1.9625038519590596e-05, "loss": 0.17437388896942138, "step": 4360 }, { "epoch": 0.11511246184594819, "grad_norm": 6.795212745666504, "learning_rate": 1.962272072861203e-05, "loss": 0.18093856573104858, "step": 4370 }, { "epoch": 0.11537587709044692, "grad_norm": 3.197979211807251, "learning_rate": 1.9620395933802158e-05, "loss": 0.18678053617477416, "step": 4380 }, { "epoch": 0.11563929233494566, "grad_norm": 2.113936185836792, "learning_rate": 1.9618064136853066e-05, "loss": 0.15992101430892944, "step": 4390 }, { "epoch": 0.11590270757944439, "grad_norm": 83.32921600341797, "learning_rate": 1.9615725339461936e-05, "loss": 1.8751018524169922, "step": 4400 }, { "epoch": 0.11616612282394313, "grad_norm": 5.212477684020996, "learning_rate": 1.961337954333105e-05, "loss": 0.29957876205444334, "step": 4410 }, { "epoch": 0.11642953806844186, "grad_norm": 56.681884765625, "learning_rate": 1.961102675016777e-05, "loss": 1.8302486419677735, "step": 4420 }, { "epoch": 0.1166929533129406, "grad_norm": 3.809683322906494, "learning_rate": 1.960866696168457e-05, "loss": 0.7570222377777099, "step": 4430 }, { "epoch": 0.11695636855743934, "grad_norm": 7.275449275970459, "learning_rate": 1.9606300179599004e-05, "loss": 0.25081400871276854, "step": 4440 }, { "epoch": 0.11721978380193808, "grad_norm": 2.8946375846862793, "learning_rate": 1.9603926405633712e-05, "loss": 0.24654991626739503, "step": 4450 }, { "epoch": 0.11748319904643681, "grad_norm": 3.73775053024292, "learning_rate": 1.9601545641516434e-05, "loss": 0.2500543355941772, "step": 4460 }, { "epoch": 0.11774661429093555, "grad_norm": 7.582231044769287, "learning_rate": 1.9599157888979997e-05, "loss": 0.24706850051879883, "step": 4470 }, { "epoch": 0.11801002953543428, "grad_norm": 4.275675296783447, "learning_rate": 1.95967631497623e-05, "loss": 0.23651280403137206, "step": 4480 }, { "epoch": 0.11827344477993303, "grad_norm": 2.980259656906128, "learning_rate": 1.9594361425606348e-05, "loss": 0.24500999450683594, "step": 4490 }, { "epoch": 0.11853686002443177, "grad_norm": 0.3290567398071289, "learning_rate": 1.959195271826021e-05, "loss": 0.2379545211791992, "step": 4500 }, { "epoch": 0.1188002752689305, "grad_norm": 2.39420485496521, "learning_rate": 1.9589537029477052e-05, "loss": 0.21550312042236328, "step": 4510 }, { "epoch": 0.11906369051342924, "grad_norm": 3.505429267883301, "learning_rate": 1.9587114361015118e-05, "loss": 0.23494415283203124, "step": 4520 }, { "epoch": 0.11932710575792797, "grad_norm": 2.3663010597229004, "learning_rate": 1.958468471463773e-05, "loss": 0.23454227447509765, "step": 4530 }, { "epoch": 0.11959052100242672, "grad_norm": 3.049983501434326, "learning_rate": 1.9582248092113286e-05, "loss": 0.2383573293685913, "step": 4540 }, { "epoch": 0.11985393624692545, "grad_norm": 6.30569314956665, "learning_rate": 1.9579804495215268e-05, "loss": 0.2209705352783203, "step": 4550 }, { "epoch": 0.12011735149142419, "grad_norm": 1.8287336826324463, "learning_rate": 1.9577353925722228e-05, "loss": 0.2226668119430542, "step": 4560 }, { "epoch": 0.12038076673592292, "grad_norm": 2.757720470428467, "learning_rate": 1.9574896385417803e-05, "loss": 0.22914953231811525, "step": 4570 }, { "epoch": 0.12064418198042166, "grad_norm": 3.0288197994232178, "learning_rate": 1.957243187609069e-05, "loss": 0.23006138801574708, "step": 4580 }, { "epoch": 0.1209075972249204, "grad_norm": 2.4438023567199707, "learning_rate": 1.9569960399534668e-05, "loss": 0.21682019233703614, "step": 4590 }, { "epoch": 0.12117101246941914, "grad_norm": 0.9146013259887695, "learning_rate": 1.9567481957548585e-05, "loss": 0.23229119777679444, "step": 4600 }, { "epoch": 0.12143442771391787, "grad_norm": 2.784860849380493, "learning_rate": 1.9564996551936354e-05, "loss": 0.22362179756164552, "step": 4610 }, { "epoch": 0.12169784295841661, "grad_norm": 3.759202718734741, "learning_rate": 1.9562504184506966e-05, "loss": 0.2156538963317871, "step": 4620 }, { "epoch": 0.12196125820291535, "grad_norm": 3.7127883434295654, "learning_rate": 1.9560004857074472e-05, "loss": 0.21273226737976075, "step": 4630 }, { "epoch": 0.12222467344741408, "grad_norm": 2.18203067779541, "learning_rate": 1.955749857145799e-05, "loss": 0.19864935874938966, "step": 4640 }, { "epoch": 0.12248808869191283, "grad_norm": 9.18358039855957, "learning_rate": 1.95549853294817e-05, "loss": 0.20105843544006347, "step": 4650 }, { "epoch": 0.12275150393641156, "grad_norm": 1.6339999437332153, "learning_rate": 1.955246513297485e-05, "loss": 0.22276363372802735, "step": 4660 }, { "epoch": 0.1230149191809103, "grad_norm": 1.403900146484375, "learning_rate": 1.9549937983771748e-05, "loss": 0.21804614067077638, "step": 4670 }, { "epoch": 0.12327833442540903, "grad_norm": 4.010140419006348, "learning_rate": 1.954740388371176e-05, "loss": 0.1993324875831604, "step": 4680 }, { "epoch": 0.12354174966990777, "grad_norm": 3.6140193939208984, "learning_rate": 1.9544862834639316e-05, "loss": 0.19990270137786864, "step": 4690 }, { "epoch": 0.1238051649144065, "grad_norm": 1.7118303775787354, "learning_rate": 1.9542314838403907e-05, "loss": 0.19296932220458984, "step": 4700 }, { "epoch": 0.12406858015890525, "grad_norm": 3.7402403354644775, "learning_rate": 1.9539759896860064e-05, "loss": 0.2162031888961792, "step": 4710 }, { "epoch": 0.12433199540340398, "grad_norm": 5.0142292976379395, "learning_rate": 1.9537198011867383e-05, "loss": 0.21493678092956542, "step": 4720 }, { "epoch": 0.12459541064790272, "grad_norm": 1.2460545301437378, "learning_rate": 1.9534629185290524e-05, "loss": 0.1992203950881958, "step": 4730 }, { "epoch": 0.12485882589240146, "grad_norm": 2.302189588546753, "learning_rate": 1.9532053418999186e-05, "loss": 0.2056973934173584, "step": 4740 }, { "epoch": 0.1251222411369002, "grad_norm": 4.569289684295654, "learning_rate": 1.952947071486812e-05, "loss": 0.2213512420654297, "step": 4750 }, { "epoch": 0.12538565638139892, "grad_norm": 5.815361499786377, "learning_rate": 1.952688107477713e-05, "loss": 0.18342888355255127, "step": 4760 }, { "epoch": 0.12564907162589767, "grad_norm": 1.668448567390442, "learning_rate": 1.9524284500611076e-05, "loss": 0.21031036376953124, "step": 4770 }, { "epoch": 0.1259124868703964, "grad_norm": 1.4336483478546143, "learning_rate": 1.9521680994259845e-05, "loss": 0.20485496520996094, "step": 4780 }, { "epoch": 0.12617590211489516, "grad_norm": 1.3994982242584229, "learning_rate": 1.9519070557618386e-05, "loss": 0.20014123916625975, "step": 4790 }, { "epoch": 0.12643931735939387, "grad_norm": 1.100663423538208, "learning_rate": 1.9516453192586686e-05, "loss": 0.18996620178222656, "step": 4800 }, { "epoch": 0.12670273260389261, "grad_norm": 1.9734052419662476, "learning_rate": 1.951382890106978e-05, "loss": 0.1923414349555969, "step": 4810 }, { "epoch": 0.12696614784839136, "grad_norm": 2.014249801635742, "learning_rate": 1.951119768497774e-05, "loss": 0.2092797040939331, "step": 4820 }, { "epoch": 0.1272295630928901, "grad_norm": 1.4278851747512817, "learning_rate": 1.9508559546225672e-05, "loss": 0.18575786352157592, "step": 4830 }, { "epoch": 0.12749297833738882, "grad_norm": 1.105285882949829, "learning_rate": 1.9505914486733735e-05, "loss": 0.18650593757629394, "step": 4840 }, { "epoch": 0.12775639358188756, "grad_norm": 1.5348461866378784, "learning_rate": 1.9503262508427115e-05, "loss": 0.2068106174468994, "step": 4850 }, { "epoch": 0.1280198088263863, "grad_norm": 2.350780963897705, "learning_rate": 1.9500603613236034e-05, "loss": 0.18836781978607178, "step": 4860 }, { "epoch": 0.12828322407088505, "grad_norm": 1.0185545682907104, "learning_rate": 1.9497937803095756e-05, "loss": 0.2014880895614624, "step": 4870 }, { "epoch": 0.1285466393153838, "grad_norm": 4.4832305908203125, "learning_rate": 1.9495265079946565e-05, "loss": 0.19417835474014283, "step": 4880 }, { "epoch": 0.1288100545598825, "grad_norm": 4.804571628570557, "learning_rate": 1.9492585445733788e-05, "loss": 0.2123647689819336, "step": 4890 }, { "epoch": 0.12907346980438125, "grad_norm": 3.429396390914917, "learning_rate": 1.9489898902407783e-05, "loss": 0.19851138591766357, "step": 4900 }, { "epoch": 0.12933688504888, "grad_norm": 4.897866725921631, "learning_rate": 1.9487205451923925e-05, "loss": 0.1902082681655884, "step": 4910 }, { "epoch": 0.12960030029337874, "grad_norm": 6.311420440673828, "learning_rate": 1.9484505096242628e-05, "loss": 0.20587334632873536, "step": 4920 }, { "epoch": 0.12986371553787746, "grad_norm": 4.72446870803833, "learning_rate": 1.9481797837329323e-05, "loss": 0.19118587970733641, "step": 4930 }, { "epoch": 0.1301271307823762, "grad_norm": 1.934660792350769, "learning_rate": 1.9479083677154474e-05, "loss": 0.18851950168609619, "step": 4940 }, { "epoch": 0.13039054602687494, "grad_norm": 1.2106409072875977, "learning_rate": 1.9476362617693564e-05, "loss": 0.19410953521728516, "step": 4950 }, { "epoch": 0.1306539612713737, "grad_norm": 1.134537696838379, "learning_rate": 1.9473634660927092e-05, "loss": 0.20282492637634278, "step": 4960 }, { "epoch": 0.1309173765158724, "grad_norm": 4.569996356964111, "learning_rate": 1.9470899808840585e-05, "loss": 0.1979243278503418, "step": 4970 }, { "epoch": 0.13118079176037115, "grad_norm": 1.7284327745437622, "learning_rate": 1.9468158063424593e-05, "loss": 0.20043132305145264, "step": 4980 }, { "epoch": 0.1314442070048699, "grad_norm": 1.6507734060287476, "learning_rate": 1.946540942667467e-05, "loss": 0.18219213485717772, "step": 4990 }, { "epoch": 0.13170762224936863, "grad_norm": 1.1909054517745972, "learning_rate": 1.9462653900591393e-05, "loss": 0.19384970664978027, "step": 5000 }, { "epoch": 0.13197103749386738, "grad_norm": 1.95111882686615, "learning_rate": 1.9459891487180353e-05, "loss": 0.19176805019378662, "step": 5010 }, { "epoch": 0.1322344527383661, "grad_norm": 1.1380332708358765, "learning_rate": 1.9457122188452156e-05, "loss": 0.17808163166046143, "step": 5020 }, { "epoch": 0.13249786798286484, "grad_norm": 4.1339874267578125, "learning_rate": 1.9454346006422414e-05, "loss": 0.18872616291046143, "step": 5030 }, { "epoch": 0.13276128322736358, "grad_norm": 0.7271566987037659, "learning_rate": 1.9451562943111753e-05, "loss": 0.19522219896316528, "step": 5040 }, { "epoch": 0.13302469847186232, "grad_norm": 1.5698851346969604, "learning_rate": 1.944877300054581e-05, "loss": 0.19346814155578612, "step": 5050 }, { "epoch": 0.13328811371636104, "grad_norm": 1.657689094543457, "learning_rate": 1.944597618075522e-05, "loss": 0.18854622840881347, "step": 5060 }, { "epoch": 0.13355152896085978, "grad_norm": 1.7233623266220093, "learning_rate": 1.944317248577563e-05, "loss": 0.18283939361572266, "step": 5070 }, { "epoch": 0.13381494420535853, "grad_norm": 1.8631333112716675, "learning_rate": 1.9440361917647693e-05, "loss": 0.19703885316848754, "step": 5080 }, { "epoch": 0.13407835944985727, "grad_norm": 1.7237244844436646, "learning_rate": 1.943754447841706e-05, "loss": 0.17790876626968383, "step": 5090 }, { "epoch": 0.134341774694356, "grad_norm": 3.226841449737549, "learning_rate": 1.9434720170134387e-05, "loss": 0.18813579082489013, "step": 5100 }, { "epoch": 0.13460518993885473, "grad_norm": 3.189938545227051, "learning_rate": 1.9431888994855324e-05, "loss": 0.1939726710319519, "step": 5110 }, { "epoch": 0.13486860518335347, "grad_norm": 3.925098419189453, "learning_rate": 1.9429050954640528e-05, "loss": 0.1829402208328247, "step": 5120 }, { "epoch": 0.13513202042785222, "grad_norm": 4.0518903732299805, "learning_rate": 1.942620605155564e-05, "loss": 0.1804246187210083, "step": 5130 }, { "epoch": 0.13539543567235096, "grad_norm": 1.4754866361618042, "learning_rate": 1.9423354287671307e-05, "loss": 0.1886010527610779, "step": 5140 }, { "epoch": 0.13565885091684968, "grad_norm": 2.0051679611206055, "learning_rate": 1.9420495665063166e-05, "loss": 0.190614914894104, "step": 5150 }, { "epoch": 0.13592226616134842, "grad_norm": 1.7766417264938354, "learning_rate": 1.9417630185811846e-05, "loss": 0.18588171005249024, "step": 5160 }, { "epoch": 0.13618568140584716, "grad_norm": 1.8210434913635254, "learning_rate": 1.9414757852002962e-05, "loss": 0.2037893295288086, "step": 5170 }, { "epoch": 0.1364490966503459, "grad_norm": 2.8244569301605225, "learning_rate": 1.941187866572713e-05, "loss": 0.18657609224319457, "step": 5180 }, { "epoch": 0.13671251189484462, "grad_norm": 1.2905091047286987, "learning_rate": 1.940899262907994e-05, "loss": 0.19393532276153563, "step": 5190 }, { "epoch": 0.13697592713934337, "grad_norm": 2.689478874206543, "learning_rate": 1.940609974416198e-05, "loss": 0.1910707950592041, "step": 5200 }, { "epoch": 0.1372393423838421, "grad_norm": 5.991103649139404, "learning_rate": 1.940320001307881e-05, "loss": 0.18704516887664796, "step": 5210 }, { "epoch": 0.13750275762834085, "grad_norm": 1.8027046918869019, "learning_rate": 1.940029343794098e-05, "loss": 0.19424235820770264, "step": 5220 }, { "epoch": 0.13776617287283957, "grad_norm": 1.5420407056808472, "learning_rate": 1.9397380020864024e-05, "loss": 0.19744768142700195, "step": 5230 }, { "epoch": 0.13802958811733831, "grad_norm": 4.077568054199219, "learning_rate": 1.939445976396845e-05, "loss": 0.1780909776687622, "step": 5240 }, { "epoch": 0.13829300336183706, "grad_norm": 2.765721559524536, "learning_rate": 1.939153266937975e-05, "loss": 0.1868884563446045, "step": 5250 }, { "epoch": 0.1385564186063358, "grad_norm": 2.0679304599761963, "learning_rate": 1.938859873922839e-05, "loss": 0.19270967245101928, "step": 5260 }, { "epoch": 0.13881983385083455, "grad_norm": 2.5818660259246826, "learning_rate": 1.9385657975649807e-05, "loss": 0.189180326461792, "step": 5270 }, { "epoch": 0.13908324909533326, "grad_norm": 1.953590750694275, "learning_rate": 1.9382710380784418e-05, "loss": 0.1819969892501831, "step": 5280 }, { "epoch": 0.139346664339832, "grad_norm": 1.465858817100525, "learning_rate": 1.937975595677761e-05, "loss": 0.18006339073181152, "step": 5290 }, { "epoch": 0.13961007958433075, "grad_norm": 2.904454469680786, "learning_rate": 1.9376794705779735e-05, "loss": 0.18562541007995606, "step": 5300 }, { "epoch": 0.1398734948288295, "grad_norm": 1.6855322122573853, "learning_rate": 1.937382662994613e-05, "loss": 0.18644968271255494, "step": 5310 }, { "epoch": 0.1401369100733282, "grad_norm": 2.7684643268585205, "learning_rate": 1.9370851731437077e-05, "loss": 0.19018877744674684, "step": 5320 }, { "epoch": 0.14040032531782695, "grad_norm": 5.336484909057617, "learning_rate": 1.9367870012417844e-05, "loss": 0.19358155727386475, "step": 5330 }, { "epoch": 0.1406637405623257, "grad_norm": 1.0045199394226074, "learning_rate": 1.936488147505865e-05, "loss": 0.18811397552490233, "step": 5340 }, { "epoch": 0.14092715580682444, "grad_norm": 3.559382677078247, "learning_rate": 1.9361886121534686e-05, "loss": 0.17774267196655275, "step": 5350 }, { "epoch": 0.14119057105132315, "grad_norm": 0.9747342467308044, "learning_rate": 1.93588839540261e-05, "loss": 0.19243173599243163, "step": 5360 }, { "epoch": 0.1414539862958219, "grad_norm": 1.2484478950500488, "learning_rate": 1.9355874974717988e-05, "loss": 0.18183441162109376, "step": 5370 }, { "epoch": 0.14171740154032064, "grad_norm": 1.9931073188781738, "learning_rate": 1.9352859185800424e-05, "loss": 0.18046661615371704, "step": 5380 }, { "epoch": 0.14198081678481939, "grad_norm": 1.4336895942687988, "learning_rate": 1.934983658946843e-05, "loss": 0.19155486822128295, "step": 5390 }, { "epoch": 0.14224423202931813, "grad_norm": 2.263335704803467, "learning_rate": 1.9346807187921983e-05, "loss": 0.1825003981590271, "step": 5400 }, { "epoch": 0.14250764727381685, "grad_norm": 4.389558792114258, "learning_rate": 1.9343770983366004e-05, "loss": 0.17619857788085938, "step": 5410 }, { "epoch": 0.1427710625183156, "grad_norm": 1.2548726797103882, "learning_rate": 1.9340727978010377e-05, "loss": 0.19399516582489013, "step": 5420 }, { "epoch": 0.14303447776281433, "grad_norm": 2.5202221870422363, "learning_rate": 1.933767817406994e-05, "loss": 0.18469767570495604, "step": 5430 }, { "epoch": 0.14329789300731308, "grad_norm": 2.6601669788360596, "learning_rate": 1.9334621573764462e-05, "loss": 0.18758847713470458, "step": 5440 }, { "epoch": 0.1435613082518118, "grad_norm": 1.8423303365707397, "learning_rate": 1.9331558179318676e-05, "loss": 0.18331973552703856, "step": 5450 }, { "epoch": 0.14382472349631054, "grad_norm": 2.209519863128662, "learning_rate": 1.9328487992962247e-05, "loss": 0.18815521001815796, "step": 5460 }, { "epoch": 0.14408813874080928, "grad_norm": 2.4590461254119873, "learning_rate": 1.9325411016929793e-05, "loss": 0.191814124584198, "step": 5470 }, { "epoch": 0.14435155398530802, "grad_norm": 0.804036021232605, "learning_rate": 1.9322327253460867e-05, "loss": 0.18531278371810914, "step": 5480 }, { "epoch": 0.14461496922980674, "grad_norm": 1.3719685077667236, "learning_rate": 1.931923670479997e-05, "loss": 0.18164094686508178, "step": 5490 }, { "epoch": 0.14487838447430548, "grad_norm": 0.7092767357826233, "learning_rate": 1.9316139373196533e-05, "loss": 0.18780651092529296, "step": 5500 }, { "epoch": 0.14514179971880423, "grad_norm": 2.4112930297851562, "learning_rate": 1.9313035260904923e-05, "loss": 0.1969895601272583, "step": 5510 }, { "epoch": 0.14540521496330297, "grad_norm": 2.607775926589966, "learning_rate": 1.9309924370184455e-05, "loss": 0.18638501167297364, "step": 5520 }, { "epoch": 0.1456686302078017, "grad_norm": 2.4811131954193115, "learning_rate": 1.9306806703299367e-05, "loss": 0.18753352165222167, "step": 5530 }, { "epoch": 0.14593204545230043, "grad_norm": 2.1052956581115723, "learning_rate": 1.9303682262518828e-05, "loss": 0.19534837007522582, "step": 5540 }, { "epoch": 0.14619546069679917, "grad_norm": 2.0487418174743652, "learning_rate": 1.9300551050116946e-05, "loss": 0.18584376573562622, "step": 5550 }, { "epoch": 0.14645887594129792, "grad_norm": 1.989164113998413, "learning_rate": 1.9297413068372745e-05, "loss": 0.17337388992309571, "step": 5560 }, { "epoch": 0.14672229118579666, "grad_norm": 4.228849411010742, "learning_rate": 1.9294268319570195e-05, "loss": 0.18138408660888672, "step": 5570 }, { "epoch": 0.14698570643029538, "grad_norm": 3.622770071029663, "learning_rate": 1.9291116805998168e-05, "loss": 0.19710860252380372, "step": 5580 }, { "epoch": 0.14724912167479412, "grad_norm": 1.39974844455719, "learning_rate": 1.9287958529950477e-05, "loss": 0.1829051375389099, "step": 5590 }, { "epoch": 0.14751253691929286, "grad_norm": 4.022392749786377, "learning_rate": 1.9284793493725852e-05, "loss": 0.18536820411682128, "step": 5600 }, { "epoch": 0.1477759521637916, "grad_norm": 3.8661515712738037, "learning_rate": 1.9281621699627943e-05, "loss": 0.1893622636795044, "step": 5610 }, { "epoch": 0.14803936740829032, "grad_norm": 1.8001810312271118, "learning_rate": 1.927844314996532e-05, "loss": 0.17649043798446656, "step": 5620 }, { "epoch": 0.14830278265278907, "grad_norm": 3.5022809505462646, "learning_rate": 1.9275257847051464e-05, "loss": 0.18123152256011962, "step": 5630 }, { "epoch": 0.1485661978972878, "grad_norm": 4.624876499176025, "learning_rate": 1.9272065793204778e-05, "loss": 0.182723069190979, "step": 5640 }, { "epoch": 0.14882961314178655, "grad_norm": 4.8767194747924805, "learning_rate": 1.9268866990748578e-05, "loss": 0.17553372383117677, "step": 5650 }, { "epoch": 0.1490930283862853, "grad_norm": 1.316911220550537, "learning_rate": 1.9265661442011087e-05, "loss": 0.20113584995269776, "step": 5660 }, { "epoch": 0.149356443630784, "grad_norm": 6.758270263671875, "learning_rate": 1.9262449149325445e-05, "loss": 0.1974040150642395, "step": 5670 }, { "epoch": 0.14961985887528276, "grad_norm": 1.52607262134552, "learning_rate": 1.92592301150297e-05, "loss": 0.183089280128479, "step": 5680 }, { "epoch": 0.1498832741197815, "grad_norm": 2.0351786613464355, "learning_rate": 1.9256004341466795e-05, "loss": 0.18253921270370482, "step": 5690 }, { "epoch": 0.15014668936428024, "grad_norm": 3.2275197505950928, "learning_rate": 1.9252771830984594e-05, "loss": 0.16165759563446044, "step": 5700 }, { "epoch": 0.15041010460877896, "grad_norm": 6.539508819580078, "learning_rate": 1.9249532585935855e-05, "loss": 0.17770451307296753, "step": 5710 }, { "epoch": 0.1506735198532777, "grad_norm": 5.593748569488525, "learning_rate": 1.9246286608678243e-05, "loss": 0.1795358657836914, "step": 5720 }, { "epoch": 0.15093693509777645, "grad_norm": 2.451145887374878, "learning_rate": 1.9243033901574316e-05, "loss": 0.18597848415374757, "step": 5730 }, { "epoch": 0.1512003503422752, "grad_norm": 5.047183990478516, "learning_rate": 1.923977446699154e-05, "loss": 0.18749245405197143, "step": 5740 }, { "epoch": 0.1514637655867739, "grad_norm": 4.306195259094238, "learning_rate": 1.923650830730227e-05, "loss": 0.17773075103759767, "step": 5750 }, { "epoch": 0.15172718083127265, "grad_norm": 4.240781784057617, "learning_rate": 1.923323542488375e-05, "loss": 0.17640169858932495, "step": 5760 }, { "epoch": 0.1519905960757714, "grad_norm": 1.7581286430358887, "learning_rate": 1.9229955822118138e-05, "loss": 0.18100602626800538, "step": 5770 }, { "epoch": 0.15225401132027014, "grad_norm": 2.747093915939331, "learning_rate": 1.9226669501392465e-05, "loss": 0.1855805516242981, "step": 5780 }, { "epoch": 0.15251742656476888, "grad_norm": 2.268267869949341, "learning_rate": 1.9223376465098657e-05, "loss": 0.16977410316467284, "step": 5790 }, { "epoch": 0.1527808418092676, "grad_norm": 7.323516845703125, "learning_rate": 1.9220076715633526e-05, "loss": 0.1856072187423706, "step": 5800 }, { "epoch": 0.15304425705376634, "grad_norm": 3.027644157409668, "learning_rate": 1.921677025539877e-05, "loss": 0.17609140872955323, "step": 5810 }, { "epoch": 0.15330767229826509, "grad_norm": 1.2249512672424316, "learning_rate": 1.921345708680098e-05, "loss": 0.17574974298477172, "step": 5820 }, { "epoch": 0.15357108754276383, "grad_norm": 2.003034830093384, "learning_rate": 1.921013721225162e-05, "loss": 0.18006489276885987, "step": 5830 }, { "epoch": 0.15383450278726254, "grad_norm": 3.6668026447296143, "learning_rate": 1.9206810634167036e-05, "loss": 0.16834326982498168, "step": 5840 }, { "epoch": 0.1540979180317613, "grad_norm": 1.5234532356262207, "learning_rate": 1.920347735496846e-05, "loss": 0.1768726110458374, "step": 5850 }, { "epoch": 0.15436133327626003, "grad_norm": 1.2704390287399292, "learning_rate": 1.9200137377081993e-05, "loss": 0.178072988986969, "step": 5860 }, { "epoch": 0.15462474852075878, "grad_norm": 1.4171069860458374, "learning_rate": 1.919679070293861e-05, "loss": 0.17433125972747804, "step": 5870 }, { "epoch": 0.1548881637652575, "grad_norm": 3.7823970317840576, "learning_rate": 1.9193437334974178e-05, "loss": 0.1742643117904663, "step": 5880 }, { "epoch": 0.15515157900975624, "grad_norm": 4.977006912231445, "learning_rate": 1.9190077275629408e-05, "loss": 0.1720792531967163, "step": 5890 }, { "epoch": 0.15541499425425498, "grad_norm": 1.5386306047439575, "learning_rate": 1.9186710527349907e-05, "loss": 0.176568865776062, "step": 5900 }, { "epoch": 0.15567840949875372, "grad_norm": 5.220499038696289, "learning_rate": 1.9183337092586137e-05, "loss": 0.18229477405548095, "step": 5910 }, { "epoch": 0.15594182474325247, "grad_norm": 1.6675987243652344, "learning_rate": 1.9179956973793422e-05, "loss": 0.1825932264328003, "step": 5920 }, { "epoch": 0.15620523998775118, "grad_norm": 0.6900759935379028, "learning_rate": 1.917657017343197e-05, "loss": 0.18583159446716307, "step": 5930 }, { "epoch": 0.15646865523224993, "grad_norm": 2.293212652206421, "learning_rate": 1.9173176693966835e-05, "loss": 0.1930854082107544, "step": 5940 }, { "epoch": 0.15673207047674867, "grad_norm": 2.522359609603882, "learning_rate": 1.9169776537867936e-05, "loss": 0.1915360450744629, "step": 5950 }, { "epoch": 0.1569954857212474, "grad_norm": 2.816586971282959, "learning_rate": 1.9166369707610055e-05, "loss": 0.1755748510360718, "step": 5960 }, { "epoch": 0.15725890096574613, "grad_norm": 1.7987374067306519, "learning_rate": 1.9162956205672833e-05, "loss": 0.18636915683746338, "step": 5970 }, { "epoch": 0.15752231621024487, "grad_norm": 1.2744853496551514, "learning_rate": 1.9159536034540757e-05, "loss": 0.18030662536621095, "step": 5980 }, { "epoch": 0.15778573145474362, "grad_norm": 2.9156880378723145, "learning_rate": 1.9156109196703177e-05, "loss": 0.17945438623428345, "step": 5990 }, { "epoch": 0.15804914669924236, "grad_norm": 1.3471322059631348, "learning_rate": 1.91526756946543e-05, "loss": 0.19220865964889527, "step": 6000 }, { "epoch": 0.15831256194374108, "grad_norm": 6.660000324249268, "learning_rate": 1.9149235530893172e-05, "loss": 0.1858498215675354, "step": 6010 }, { "epoch": 0.15857597718823982, "grad_norm": 2.0262272357940674, "learning_rate": 1.9145788707923692e-05, "loss": 0.1904280662536621, "step": 6020 }, { "epoch": 0.15883939243273856, "grad_norm": 2.211155891418457, "learning_rate": 1.914233522825461e-05, "loss": 0.17845650911331176, "step": 6030 }, { "epoch": 0.1591028076772373, "grad_norm": 2.9801058769226074, "learning_rate": 1.913887509439951e-05, "loss": 0.17622783184051513, "step": 6040 }, { "epoch": 0.15936622292173605, "grad_norm": 3.882659673690796, "learning_rate": 1.9135408308876834e-05, "loss": 0.18288249969482423, "step": 6050 }, { "epoch": 0.15962963816623477, "grad_norm": 1.8242113590240479, "learning_rate": 1.913193487420986e-05, "loss": 0.1820768356323242, "step": 6060 }, { "epoch": 0.1598930534107335, "grad_norm": 1.4216312170028687, "learning_rate": 1.9128454792926694e-05, "loss": 0.1740484356880188, "step": 6070 }, { "epoch": 0.16015646865523225, "grad_norm": 3.4184253215789795, "learning_rate": 1.9124968067560296e-05, "loss": 0.18592777252197265, "step": 6080 }, { "epoch": 0.160419883899731, "grad_norm": 4.000383377075195, "learning_rate": 1.9121474700648454e-05, "loss": 0.1814777970314026, "step": 6090 }, { "epoch": 0.1606832991442297, "grad_norm": 3.044079065322876, "learning_rate": 1.9117974694733795e-05, "loss": 0.1774237871170044, "step": 6100 }, { "epoch": 0.16094671438872846, "grad_norm": 2.2713191509246826, "learning_rate": 1.9114468052363764e-05, "loss": 0.18940006494522094, "step": 6110 }, { "epoch": 0.1612101296332272, "grad_norm": 1.767372488975525, "learning_rate": 1.9110954776090656e-05, "loss": 0.18555872440338134, "step": 6120 }, { "epoch": 0.16147354487772594, "grad_norm": 2.275505781173706, "learning_rate": 1.910743486847158e-05, "loss": 0.1724867820739746, "step": 6130 }, { "epoch": 0.16173696012222466, "grad_norm": 1.733323574066162, "learning_rate": 1.9103908332068483e-05, "loss": 0.18716111183166503, "step": 6140 }, { "epoch": 0.1620003753667234, "grad_norm": 3.6215696334838867, "learning_rate": 1.910037516944812e-05, "loss": 0.1676236867904663, "step": 6150 }, { "epoch": 0.16226379061122215, "grad_norm": 3.5598037242889404, "learning_rate": 1.909683538318209e-05, "loss": 0.17729365825653076, "step": 6160 }, { "epoch": 0.1625272058557209, "grad_norm": 3.402039051055908, "learning_rate": 1.9093288975846797e-05, "loss": 0.17903318405151367, "step": 6170 }, { "epoch": 0.16279062110021963, "grad_norm": 1.9035298824310303, "learning_rate": 1.9089735950023463e-05, "loss": 0.19040203094482422, "step": 6180 }, { "epoch": 0.16305403634471835, "grad_norm": 0.6519178748130798, "learning_rate": 1.9086176308298147e-05, "loss": 0.17967305183410645, "step": 6190 }, { "epoch": 0.1633174515892171, "grad_norm": 8.392556190490723, "learning_rate": 1.90826100532617e-05, "loss": 0.19041550159454346, "step": 6200 }, { "epoch": 0.16358086683371584, "grad_norm": 3.348349094390869, "learning_rate": 1.90790371875098e-05, "loss": 0.1862289309501648, "step": 6210 }, { "epoch": 0.16384428207821458, "grad_norm": 1.0912079811096191, "learning_rate": 1.9075457713642934e-05, "loss": 0.1865411162376404, "step": 6220 }, { "epoch": 0.1641076973227133, "grad_norm": 1.523605465888977, "learning_rate": 1.9071871634266392e-05, "loss": 0.17627744674682616, "step": 6230 }, { "epoch": 0.16437111256721204, "grad_norm": 0.972046434879303, "learning_rate": 1.9068278951990282e-05, "loss": 0.16807688474655152, "step": 6240 }, { "epoch": 0.16463452781171078, "grad_norm": 5.047930717468262, "learning_rate": 1.9064679669429515e-05, "loss": 0.1838808536529541, "step": 6250 }, { "epoch": 0.16489794305620953, "grad_norm": 3.0882692337036133, "learning_rate": 1.9061073789203798e-05, "loss": 0.17740026712417603, "step": 6260 }, { "epoch": 0.16516135830070824, "grad_norm": 1.2378528118133545, "learning_rate": 1.9057461313937646e-05, "loss": 0.19275953769683837, "step": 6270 }, { "epoch": 0.165424773545207, "grad_norm": 1.8307173252105713, "learning_rate": 1.905384224626038e-05, "loss": 0.19018245935440065, "step": 6280 }, { "epoch": 0.16568818878970573, "grad_norm": 3.809479236602783, "learning_rate": 1.9050216588806113e-05, "loss": 0.18106496334075928, "step": 6290 }, { "epoch": 0.16595160403420448, "grad_norm": 4.884668350219727, "learning_rate": 1.9046584344213754e-05, "loss": 0.17564244270324708, "step": 6300 }, { "epoch": 0.16621501927870322, "grad_norm": 2.5494234561920166, "learning_rate": 1.9042945515127004e-05, "loss": 0.18096710443496705, "step": 6310 }, { "epoch": 0.16647843452320193, "grad_norm": 2.5470170974731445, "learning_rate": 1.9039300104194366e-05, "loss": 0.17255386114120483, "step": 6320 }, { "epoch": 0.16674184976770068, "grad_norm": 2.5803580284118652, "learning_rate": 1.903564811406912e-05, "loss": 0.18161146640777587, "step": 6330 }, { "epoch": 0.16700526501219942, "grad_norm": 2.6573147773742676, "learning_rate": 1.9031989547409347e-05, "loss": 0.16941677331924437, "step": 6340 }, { "epoch": 0.16726868025669817, "grad_norm": 1.093811273574829, "learning_rate": 1.9028324406877905e-05, "loss": 0.16744914054870605, "step": 6350 }, { "epoch": 0.16753209550119688, "grad_norm": 2.7999320030212402, "learning_rate": 1.9024652695142447e-05, "loss": 0.16521579027175903, "step": 6360 }, { "epoch": 0.16779551074569563, "grad_norm": 2.0227015018463135, "learning_rate": 1.90209744148754e-05, "loss": 0.17245750427246093, "step": 6370 }, { "epoch": 0.16805892599019437, "grad_norm": 3.177515983581543, "learning_rate": 1.901728956875398e-05, "loss": 0.16800882816314697, "step": 6380 }, { "epoch": 0.1683223412346931, "grad_norm": 1.7765941619873047, "learning_rate": 1.9013598159460168e-05, "loss": 0.15568060874938966, "step": 6390 }, { "epoch": 0.16858575647919183, "grad_norm": 3.797597885131836, "learning_rate": 1.900990018968074e-05, "loss": 0.1756840467453003, "step": 6400 }, { "epoch": 0.16884917172369057, "grad_norm": 1.5509068965911865, "learning_rate": 1.900619566210723e-05, "loss": 0.1598289966583252, "step": 6410 }, { "epoch": 0.16911258696818932, "grad_norm": 1.544917345046997, "learning_rate": 1.900248457943596e-05, "loss": 0.17402288913726807, "step": 6420 }, { "epoch": 0.16937600221268806, "grad_norm": 4.8092732429504395, "learning_rate": 1.899876694436801e-05, "loss": 0.1804758906364441, "step": 6430 }, { "epoch": 0.1696394174571868, "grad_norm": 1.8425848484039307, "learning_rate": 1.899504275960924e-05, "loss": 0.18042454719543458, "step": 6440 }, { "epoch": 0.16990283270168552, "grad_norm": 3.4565205574035645, "learning_rate": 1.899131202787027e-05, "loss": 0.17908041477203368, "step": 6450 }, { "epoch": 0.17016624794618426, "grad_norm": 1.981529712677002, "learning_rate": 1.8987574751866488e-05, "loss": 0.19641530513763428, "step": 6460 }, { "epoch": 0.170429663190683, "grad_norm": 5.034128665924072, "learning_rate": 1.8983830934318044e-05, "loss": 0.19203541278839112, "step": 6470 }, { "epoch": 0.17069307843518175, "grad_norm": 4.7060933113098145, "learning_rate": 1.8980080577949848e-05, "loss": 0.18257532119750977, "step": 6480 }, { "epoch": 0.17095649367968047, "grad_norm": 3.520148515701294, "learning_rate": 1.8976323685491577e-05, "loss": 0.18209165334701538, "step": 6490 }, { "epoch": 0.1712199089241792, "grad_norm": 2.3610925674438477, "learning_rate": 1.8972560259677658e-05, "loss": 0.1799578309059143, "step": 6500 }, { "epoch": 0.17148332416867795, "grad_norm": 6.306294918060303, "learning_rate": 1.896879030324727e-05, "loss": 0.1773930788040161, "step": 6510 }, { "epoch": 0.1717467394131767, "grad_norm": 0.7151606678962708, "learning_rate": 1.8965013818944357e-05, "loss": 0.18359014987945557, "step": 6520 }, { "epoch": 0.1720101546576754, "grad_norm": 1.757624626159668, "learning_rate": 1.89612308095176e-05, "loss": 0.18723965883255006, "step": 6530 }, { "epoch": 0.17227356990217416, "grad_norm": 1.9149487018585205, "learning_rate": 1.8957441277720438e-05, "loss": 0.18344099521636964, "step": 6540 }, { "epoch": 0.1725369851466729, "grad_norm": 1.6695787906646729, "learning_rate": 1.8953645226311056e-05, "loss": 0.1705393075942993, "step": 6550 }, { "epoch": 0.17280040039117164, "grad_norm": 1.428368091583252, "learning_rate": 1.894984265805239e-05, "loss": 0.19564977884292603, "step": 6560 }, { "epoch": 0.1730638156356704, "grad_norm": 2.1702752113342285, "learning_rate": 1.8946033575712104e-05, "loss": 0.1715014696121216, "step": 6570 }, { "epoch": 0.1733272308801691, "grad_norm": 1.6893748044967651, "learning_rate": 1.894221798206262e-05, "loss": 0.18304336071014404, "step": 6580 }, { "epoch": 0.17359064612466785, "grad_norm": 3.724792242050171, "learning_rate": 1.8938395879881086e-05, "loss": 0.1852871298789978, "step": 6590 }, { "epoch": 0.1738540613691666, "grad_norm": 1.8329954147338867, "learning_rate": 1.8934567271949397e-05, "loss": 0.1903809428215027, "step": 6600 }, { "epoch": 0.17411747661366533, "grad_norm": 1.7412115335464478, "learning_rate": 1.8930732161054176e-05, "loss": 0.16736655235290526, "step": 6610 }, { "epoch": 0.17438089185816405, "grad_norm": 4.217787265777588, "learning_rate": 1.8926890549986786e-05, "loss": 0.1769413471221924, "step": 6620 }, { "epoch": 0.1746443071026628, "grad_norm": 2.1061654090881348, "learning_rate": 1.8923042441543313e-05, "loss": 0.1781160831451416, "step": 6630 }, { "epoch": 0.17490772234716154, "grad_norm": 5.357895374298096, "learning_rate": 1.891918783852458e-05, "loss": 0.18956550359725952, "step": 6640 }, { "epoch": 0.17517113759166028, "grad_norm": 0.9909026622772217, "learning_rate": 1.8915326743736134e-05, "loss": 0.16525330543518066, "step": 6650 }, { "epoch": 0.175434552836159, "grad_norm": 4.065378189086914, "learning_rate": 1.8911459159988238e-05, "loss": 0.18399231433868407, "step": 6660 }, { "epoch": 0.17569796808065774, "grad_norm": 1.0658223628997803, "learning_rate": 1.8907585090095897e-05, "loss": 0.17222397327423095, "step": 6670 }, { "epoch": 0.17596138332515648, "grad_norm": 3.162492036819458, "learning_rate": 1.890370453687882e-05, "loss": 0.1776159882545471, "step": 6680 }, { "epoch": 0.17622479856965523, "grad_norm": 3.4341626167297363, "learning_rate": 1.889981750316144e-05, "loss": 0.182609486579895, "step": 6690 }, { "epoch": 0.17648821381415397, "grad_norm": 1.7721443176269531, "learning_rate": 1.8895923991772913e-05, "loss": 0.1835877537727356, "step": 6700 }, { "epoch": 0.1767516290586527, "grad_norm": 1.5589373111724854, "learning_rate": 1.8892024005547104e-05, "loss": 0.18838950395584106, "step": 6710 }, { "epoch": 0.17701504430315143, "grad_norm": 4.726339340209961, "learning_rate": 1.8888117547322586e-05, "loss": 0.17104849815368653, "step": 6720 }, { "epoch": 0.17727845954765017, "grad_norm": 2.15175724029541, "learning_rate": 1.888420461994265e-05, "loss": 0.18151514530181884, "step": 6730 }, { "epoch": 0.17754187479214892, "grad_norm": 3.933379650115967, "learning_rate": 1.88802852262553e-05, "loss": 0.17075040340423583, "step": 6740 }, { "epoch": 0.17780529003664763, "grad_norm": 2.1856303215026855, "learning_rate": 1.887635936911323e-05, "loss": 0.19206292629241944, "step": 6750 }, { "epoch": 0.17806870528114638, "grad_norm": 1.225058674812317, "learning_rate": 1.8872427051373852e-05, "loss": 0.17918946743011474, "step": 6760 }, { "epoch": 0.17833212052564512, "grad_norm": 1.421595811843872, "learning_rate": 1.886848827589928e-05, "loss": 0.1764788031578064, "step": 6770 }, { "epoch": 0.17859553577014387, "grad_norm": 2.862948417663574, "learning_rate": 1.8864543045556318e-05, "loss": 0.18490734100341796, "step": 6780 }, { "epoch": 0.1788589510146426, "grad_norm": 6.220715045928955, "learning_rate": 1.8860591363216482e-05, "loss": 0.18557120561599733, "step": 6790 }, { "epoch": 0.17912236625914132, "grad_norm": 2.4098293781280518, "learning_rate": 1.8856633231755972e-05, "loss": 0.18087855577468873, "step": 6800 }, { "epoch": 0.17938578150364007, "grad_norm": 1.6451056003570557, "learning_rate": 1.8852668654055686e-05, "loss": 0.17363669872283935, "step": 6810 }, { "epoch": 0.1796491967481388, "grad_norm": 2.2308499813079834, "learning_rate": 1.884869763300122e-05, "loss": 0.17653357982635498, "step": 6820 }, { "epoch": 0.17991261199263756, "grad_norm": 1.534700870513916, "learning_rate": 1.8844720171482854e-05, "loss": 0.16458797454833984, "step": 6830 }, { "epoch": 0.18017602723713627, "grad_norm": 2.3104069232940674, "learning_rate": 1.884073627239555e-05, "loss": 0.17637635469436647, "step": 6840 }, { "epoch": 0.18043944248163502, "grad_norm": 3.438866138458252, "learning_rate": 1.8836745938638963e-05, "loss": 0.1796119451522827, "step": 6850 }, { "epoch": 0.18070285772613376, "grad_norm": 2.344696044921875, "learning_rate": 1.8832749173117435e-05, "loss": 0.1791142463684082, "step": 6860 }, { "epoch": 0.1809662729706325, "grad_norm": 2.7142066955566406, "learning_rate": 1.8828745978739982e-05, "loss": 0.17556840181350708, "step": 6870 }, { "epoch": 0.18122968821513122, "grad_norm": 0.8444920778274536, "learning_rate": 1.88247363584203e-05, "loss": 0.1717628002166748, "step": 6880 }, { "epoch": 0.18149310345962996, "grad_norm": 4.265467166900635, "learning_rate": 1.8820720315076767e-05, "loss": 0.1761951446533203, "step": 6890 }, { "epoch": 0.1817565187041287, "grad_norm": 1.903084397315979, "learning_rate": 1.8816697851632425e-05, "loss": 0.18645294904708862, "step": 6900 }, { "epoch": 0.18201993394862745, "grad_norm": 1.917852759361267, "learning_rate": 1.8812668971015006e-05, "loss": 0.18230605125427246, "step": 6910 }, { "epoch": 0.1822833491931262, "grad_norm": 1.3042776584625244, "learning_rate": 1.88086336761569e-05, "loss": 0.17382968664169313, "step": 6920 }, { "epoch": 0.1825467644376249, "grad_norm": 1.1588703393936157, "learning_rate": 1.8804591969995167e-05, "loss": 0.17784734964370727, "step": 6930 }, { "epoch": 0.18281017968212365, "grad_norm": 2.0347914695739746, "learning_rate": 1.8800543855471535e-05, "loss": 0.17824349403381348, "step": 6940 }, { "epoch": 0.1830735949266224, "grad_norm": 1.3494000434875488, "learning_rate": 1.87964893355324e-05, "loss": 0.1694621443748474, "step": 6950 }, { "epoch": 0.18333701017112114, "grad_norm": 2.983165979385376, "learning_rate": 1.879242841312881e-05, "loss": 0.17224109172821045, "step": 6960 }, { "epoch": 0.18360042541561986, "grad_norm": 4.531947612762451, "learning_rate": 1.878836109121648e-05, "loss": 0.17845913171768188, "step": 6970 }, { "epoch": 0.1838638406601186, "grad_norm": 2.466287612915039, "learning_rate": 1.8784287372755795e-05, "loss": 0.18998174667358397, "step": 6980 }, { "epoch": 0.18412725590461734, "grad_norm": 4.873678684234619, "learning_rate": 1.8780207260711767e-05, "loss": 0.19118729829788209, "step": 6990 }, { "epoch": 0.1843906711491161, "grad_norm": 1.9291900396347046, "learning_rate": 1.8776120758054087e-05, "loss": 0.18442778587341307, "step": 7000 }, { "epoch": 0.1846540863936148, "grad_norm": 3.45621657371521, "learning_rate": 1.8772027867757082e-05, "loss": 0.17692042589187623, "step": 7010 }, { "epoch": 0.18491750163811355, "grad_norm": 4.4261064529418945, "learning_rate": 1.8767928592799742e-05, "loss": 0.18038636445999146, "step": 7020 }, { "epoch": 0.1851809168826123, "grad_norm": 3.1848490238189697, "learning_rate": 1.876382293616569e-05, "loss": 0.1862688183784485, "step": 7030 }, { "epoch": 0.18544433212711103, "grad_norm": 1.4918112754821777, "learning_rate": 1.8759710900843202e-05, "loss": 0.178133761882782, "step": 7040 }, { "epoch": 0.18570774737160978, "grad_norm": 2.5588228702545166, "learning_rate": 1.8755592489825193e-05, "loss": 0.1606306552886963, "step": 7050 }, { "epoch": 0.1859711626161085, "grad_norm": 2.1087827682495117, "learning_rate": 1.8751467706109224e-05, "loss": 0.18530614376068116, "step": 7060 }, { "epoch": 0.18623457786060724, "grad_norm": 3.890367269515991, "learning_rate": 1.8747336552697483e-05, "loss": 0.17664597034454346, "step": 7070 }, { "epoch": 0.18649799310510598, "grad_norm": 1.2323077917099, "learning_rate": 1.874319903259681e-05, "loss": 0.16759703159332276, "step": 7080 }, { "epoch": 0.18676140834960472, "grad_norm": 1.6826958656311035, "learning_rate": 1.873905514881867e-05, "loss": 0.1901644468307495, "step": 7090 }, { "epoch": 0.18702482359410344, "grad_norm": 3.808974504470825, "learning_rate": 1.873490490437915e-05, "loss": 0.16703429222106933, "step": 7100 }, { "epoch": 0.18728823883860218, "grad_norm": 5.240165710449219, "learning_rate": 1.8730748302298993e-05, "loss": 0.18087146282196045, "step": 7110 }, { "epoch": 0.18755165408310093, "grad_norm": 2.90761661529541, "learning_rate": 1.8726585345603544e-05, "loss": 0.20561087131500244, "step": 7120 }, { "epoch": 0.18781506932759967, "grad_norm": 1.4283521175384521, "learning_rate": 1.8722416037322778e-05, "loss": 0.18256471157073975, "step": 7130 }, { "epoch": 0.1880784845720984, "grad_norm": 2.5559353828430176, "learning_rate": 1.871824038049131e-05, "loss": 0.18867577314376832, "step": 7140 }, { "epoch": 0.18834189981659713, "grad_norm": 1.6678025722503662, "learning_rate": 1.871405837814835e-05, "loss": 0.16874442100524903, "step": 7150 }, { "epoch": 0.18860531506109587, "grad_norm": 2.1880834102630615, "learning_rate": 1.8709870033337753e-05, "loss": 0.18929708003997803, "step": 7160 }, { "epoch": 0.18886873030559462, "grad_norm": 5.31240701675415, "learning_rate": 1.8705675349107964e-05, "loss": 0.1765273094177246, "step": 7170 }, { "epoch": 0.18913214555009336, "grad_norm": 3.584062337875366, "learning_rate": 1.870147432851206e-05, "loss": 0.1704852819442749, "step": 7180 }, { "epoch": 0.18939556079459208, "grad_norm": 3.8750803470611572, "learning_rate": 1.8697266974607732e-05, "loss": 0.18532745838165282, "step": 7190 }, { "epoch": 0.18965897603909082, "grad_norm": 2.430366277694702, "learning_rate": 1.8693053290457264e-05, "loss": 0.1687709331512451, "step": 7200 }, { "epoch": 0.18992239128358956, "grad_norm": 2.8951399326324463, "learning_rate": 1.868883327912756e-05, "loss": 0.18028119802474976, "step": 7210 }, { "epoch": 0.1901858065280883, "grad_norm": 1.145390510559082, "learning_rate": 1.868460694369013e-05, "loss": 0.17825764417648315, "step": 7220 }, { "epoch": 0.19044922177258702, "grad_norm": 1.5743963718414307, "learning_rate": 1.868037428722108e-05, "loss": 0.1769658923149109, "step": 7230 }, { "epoch": 0.19071263701708577, "grad_norm": 2.3714754581451416, "learning_rate": 1.867613531280112e-05, "loss": 0.17446699142456054, "step": 7240 }, { "epoch": 0.1909760522615845, "grad_norm": 1.5688337087631226, "learning_rate": 1.867189002351556e-05, "loss": 0.17505919933319092, "step": 7250 }, { "epoch": 0.19123946750608325, "grad_norm": 4.70933198928833, "learning_rate": 1.8667638422454306e-05, "loss": 0.17648229598999024, "step": 7260 }, { "epoch": 0.19150288275058197, "grad_norm": 3.1561050415039062, "learning_rate": 1.8663380512711852e-05, "loss": 0.17478333711624144, "step": 7270 }, { "epoch": 0.19176629799508071, "grad_norm": 1.8535821437835693, "learning_rate": 1.865911629738729e-05, "loss": 0.17577826976776123, "step": 7280 }, { "epoch": 0.19202971323957946, "grad_norm": 1.777890682220459, "learning_rate": 1.86548457795843e-05, "loss": 0.18091144561767578, "step": 7290 }, { "epoch": 0.1922931284840782, "grad_norm": 1.6042468547821045, "learning_rate": 1.8650568962411154e-05, "loss": 0.17748132944107056, "step": 7300 }, { "epoch": 0.19255654372857695, "grad_norm": 2.904414653778076, "learning_rate": 1.8646285848980693e-05, "loss": 0.16744959354400635, "step": 7310 }, { "epoch": 0.19281995897307566, "grad_norm": 5.994386672973633, "learning_rate": 1.8641996442410365e-05, "loss": 0.15679068565368653, "step": 7320 }, { "epoch": 0.1930833742175744, "grad_norm": 1.082806944847107, "learning_rate": 1.8637700745822176e-05, "loss": 0.1785661220550537, "step": 7330 }, { "epoch": 0.19334678946207315, "grad_norm": 0.5502932667732239, "learning_rate": 1.8633398762342722e-05, "loss": 0.17923312187194823, "step": 7340 }, { "epoch": 0.1936102047065719, "grad_norm": 3.407576560974121, "learning_rate": 1.8629090495103175e-05, "loss": 0.17114322185516356, "step": 7350 }, { "epoch": 0.1938736199510706, "grad_norm": 3.206166982650757, "learning_rate": 1.8624775947239272e-05, "loss": 0.15250256061553955, "step": 7360 }, { "epoch": 0.19413703519556935, "grad_norm": 3.955223798751831, "learning_rate": 1.862045512189133e-05, "loss": 0.1728896379470825, "step": 7370 }, { "epoch": 0.1944004504400681, "grad_norm": 3.4924798011779785, "learning_rate": 1.8616128022204233e-05, "loss": 0.1753464460372925, "step": 7380 }, { "epoch": 0.19466386568456684, "grad_norm": 1.490264654159546, "learning_rate": 1.8611794651327432e-05, "loss": 0.17799506187438965, "step": 7390 }, { "epoch": 0.19492728092906556, "grad_norm": 1.4260743856430054, "learning_rate": 1.8607455012414936e-05, "loss": 0.162539803981781, "step": 7400 }, { "epoch": 0.1951906961735643, "grad_norm": 1.9768517017364502, "learning_rate": 1.8603109108625327e-05, "loss": 0.17241146564483642, "step": 7410 }, { "epoch": 0.19545411141806304, "grad_norm": 1.9129469394683838, "learning_rate": 1.859875694312174e-05, "loss": 0.16263829469680785, "step": 7420 }, { "epoch": 0.19571752666256179, "grad_norm": 1.1310105323791504, "learning_rate": 1.8594398519071867e-05, "loss": 0.18411765098571778, "step": 7430 }, { "epoch": 0.19598094190706053, "grad_norm": 1.402269959449768, "learning_rate": 1.8590033839647955e-05, "loss": 0.1835169792175293, "step": 7440 }, { "epoch": 0.19624435715155925, "grad_norm": 0.9120914936065674, "learning_rate": 1.8585662908026814e-05, "loss": 0.16708855628967284, "step": 7450 }, { "epoch": 0.196507772396058, "grad_norm": 1.2544386386871338, "learning_rate": 1.8581285727389787e-05, "loss": 0.1730677843093872, "step": 7460 }, { "epoch": 0.19677118764055673, "grad_norm": 5.016749382019043, "learning_rate": 1.857690230092278e-05, "loss": 0.18054637908935547, "step": 7470 }, { "epoch": 0.19703460288505548, "grad_norm": 1.5839722156524658, "learning_rate": 1.8572512631816243e-05, "loss": 0.16625056266784669, "step": 7480 }, { "epoch": 0.1972980181295542, "grad_norm": 1.2694367170333862, "learning_rate": 1.856811672326516e-05, "loss": 0.15948513746261597, "step": 7490 }, { "epoch": 0.19756143337405294, "grad_norm": 1.5905879735946655, "learning_rate": 1.8563714578469064e-05, "loss": 0.1803716540336609, "step": 7500 }, { "epoch": 0.19782484861855168, "grad_norm": 3.6282401084899902, "learning_rate": 1.8559306200632036e-05, "loss": 0.17157657146453859, "step": 7510 }, { "epoch": 0.19808826386305042, "grad_norm": 1.3506537675857544, "learning_rate": 1.8554891592962666e-05, "loss": 0.1555756688117981, "step": 7520 }, { "epoch": 0.19835167910754914, "grad_norm": 2.279569149017334, "learning_rate": 1.8550470758674112e-05, "loss": 0.166595983505249, "step": 7530 }, { "epoch": 0.19861509435204788, "grad_norm": 1.2613201141357422, "learning_rate": 1.8546043700984036e-05, "loss": 0.17138307094573973, "step": 7540 }, { "epoch": 0.19887850959654663, "grad_norm": 2.193166732788086, "learning_rate": 1.8541610423114647e-05, "loss": 0.18183976411819458, "step": 7550 }, { "epoch": 0.19914192484104537, "grad_norm": 1.3997734785079956, "learning_rate": 1.853717092829268e-05, "loss": 0.17834208011627198, "step": 7560 }, { "epoch": 0.1994053400855441, "grad_norm": 1.3891154527664185, "learning_rate": 1.853272521974938e-05, "loss": 0.1785561203956604, "step": 7570 }, { "epoch": 0.19966875533004283, "grad_norm": 1.2774021625518799, "learning_rate": 1.8528273300720538e-05, "loss": 0.1664695143699646, "step": 7580 }, { "epoch": 0.19993217057454157, "grad_norm": 1.2531236410140991, "learning_rate": 1.8523815174446444e-05, "loss": 0.17169777154922486, "step": 7590 }, { "epoch": 0.20019558581904032, "grad_norm": 0.8074085116386414, "learning_rate": 1.851935084417192e-05, "loss": 0.16495543718338013, "step": 7600 }, { "epoch": 0.20045900106353906, "grad_norm": 1.8421509265899658, "learning_rate": 1.8514880313146295e-05, "loss": 0.1861811399459839, "step": 7610 }, { "epoch": 0.20072241630803778, "grad_norm": 1.6243897676467896, "learning_rate": 1.8510403584623414e-05, "loss": 0.17641656398773192, "step": 7620 }, { "epoch": 0.20098583155253652, "grad_norm": 5.773672580718994, "learning_rate": 1.850592066186163e-05, "loss": 0.16626558303833008, "step": 7630 }, { "epoch": 0.20124924679703526, "grad_norm": 3.3634841442108154, "learning_rate": 1.850143154812382e-05, "loss": 0.16601054668426513, "step": 7640 }, { "epoch": 0.201512662041534, "grad_norm": 1.9282995462417603, "learning_rate": 1.849693624667734e-05, "loss": 0.16700419187545776, "step": 7650 }, { "epoch": 0.20177607728603272, "grad_norm": 1.7158163785934448, "learning_rate": 1.8492434760794072e-05, "loss": 0.17530906200408936, "step": 7660 }, { "epoch": 0.20203949253053147, "grad_norm": 1.77261483669281, "learning_rate": 1.848792709375039e-05, "loss": 0.16649234294891357, "step": 7670 }, { "epoch": 0.2023029077750302, "grad_norm": 0.9272355437278748, "learning_rate": 1.848341324882717e-05, "loss": 0.16091517210006714, "step": 7680 }, { "epoch": 0.20256632301952895, "grad_norm": 0.8453692197799683, "learning_rate": 1.8478893229309784e-05, "loss": 0.17420539855957032, "step": 7690 }, { "epoch": 0.2028297382640277, "grad_norm": 0.9777116775512695, "learning_rate": 1.847436703848809e-05, "loss": 0.17061185836791992, "step": 7700 }, { "epoch": 0.20309315350852641, "grad_norm": 4.17047643661499, "learning_rate": 1.8469834679656452e-05, "loss": 0.17967405319213867, "step": 7710 }, { "epoch": 0.20335656875302516, "grad_norm": 2.154564619064331, "learning_rate": 1.8465296156113715e-05, "loss": 0.16533370018005372, "step": 7720 }, { "epoch": 0.2036199839975239, "grad_norm": 1.8516314029693604, "learning_rate": 1.8460751471163212e-05, "loss": 0.1806187868118286, "step": 7730 }, { "epoch": 0.20388339924202264, "grad_norm": 1.905804991722107, "learning_rate": 1.845620062811276e-05, "loss": 0.17741782665252687, "step": 7740 }, { "epoch": 0.20414681448652136, "grad_norm": 3.481450080871582, "learning_rate": 1.8451643630274664e-05, "loss": 0.16591814756393433, "step": 7750 }, { "epoch": 0.2044102297310201, "grad_norm": 1.9053852558135986, "learning_rate": 1.8447080480965694e-05, "loss": 0.17131128311157226, "step": 7760 }, { "epoch": 0.20467364497551885, "grad_norm": 2.64622163772583, "learning_rate": 1.844251118350711e-05, "loss": 0.16887354850769043, "step": 7770 }, { "epoch": 0.2049370602200176, "grad_norm": 2.2803473472595215, "learning_rate": 1.8437935741224652e-05, "loss": 0.18199517726898193, "step": 7780 }, { "epoch": 0.2052004754645163, "grad_norm": 1.5717597007751465, "learning_rate": 1.8433354157448513e-05, "loss": 0.17459672689437866, "step": 7790 }, { "epoch": 0.20546389070901505, "grad_norm": 3.6453068256378174, "learning_rate": 1.8428766435513373e-05, "loss": 0.17498908042907715, "step": 7800 }, { "epoch": 0.2057273059535138, "grad_norm": 5.805761337280273, "learning_rate": 1.8424172578758374e-05, "loss": 0.1704413414001465, "step": 7810 }, { "epoch": 0.20599072119801254, "grad_norm": 3.4142537117004395, "learning_rate": 1.841957259052712e-05, "loss": 0.16838133335113525, "step": 7820 }, { "epoch": 0.20625413644251128, "grad_norm": 1.308903455734253, "learning_rate": 1.8414966474167688e-05, "loss": 0.17753676176071168, "step": 7830 }, { "epoch": 0.20651755168701, "grad_norm": 5.771010875701904, "learning_rate": 1.8410354233032604e-05, "loss": 0.16408174037933348, "step": 7840 }, { "epoch": 0.20678096693150874, "grad_norm": 2.1630449295043945, "learning_rate": 1.840573587047885e-05, "loss": 0.17130223512649537, "step": 7850 }, { "epoch": 0.20704438217600749, "grad_norm": 1.616127848625183, "learning_rate": 1.8401111389867882e-05, "loss": 0.17861260175704957, "step": 7860 }, { "epoch": 0.20730779742050623, "grad_norm": 2.6703407764434814, "learning_rate": 1.8396480794565583e-05, "loss": 0.16779441833496095, "step": 7870 }, { "epoch": 0.20757121266500494, "grad_norm": 3.3459627628326416, "learning_rate": 1.839184408794231e-05, "loss": 0.17538464069366455, "step": 7880 }, { "epoch": 0.2078346279095037, "grad_norm": 1.559653639793396, "learning_rate": 1.8387201273372855e-05, "loss": 0.1665958881378174, "step": 7890 }, { "epoch": 0.20809804315400243, "grad_norm": 4.664608001708984, "learning_rate": 1.838255235423646e-05, "loss": 0.15895390510559082, "step": 7900 }, { "epoch": 0.20836145839850118, "grad_norm": 2.0746819972991943, "learning_rate": 1.8377897333916808e-05, "loss": 0.16998980045318604, "step": 7910 }, { "epoch": 0.2086248736429999, "grad_norm": 6.224613189697266, "learning_rate": 1.837323621580202e-05, "loss": 0.16112093925476073, "step": 7920 }, { "epoch": 0.20888828888749864, "grad_norm": 4.352329730987549, "learning_rate": 1.8368569003284662e-05, "loss": 0.16780250072479247, "step": 7930 }, { "epoch": 0.20915170413199738, "grad_norm": 2.0574586391448975, "learning_rate": 1.8363895699761736e-05, "loss": 0.17062046527862548, "step": 7940 }, { "epoch": 0.20941511937649612, "grad_norm": 5.149139404296875, "learning_rate": 1.8359216308634668e-05, "loss": 0.16920344829559325, "step": 7950 }, { "epoch": 0.20967853462099487, "grad_norm": 0.7540169954299927, "learning_rate": 1.8354530833309323e-05, "loss": 0.17527155876159667, "step": 7960 }, { "epoch": 0.20994194986549358, "grad_norm": 3.226954460144043, "learning_rate": 1.8349839277195993e-05, "loss": 0.1582091212272644, "step": 7970 }, { "epoch": 0.21020536510999233, "grad_norm": 2.3239126205444336, "learning_rate": 1.8345141643709394e-05, "loss": 0.16758508682250978, "step": 7980 }, { "epoch": 0.21046878035449107, "grad_norm": 2.078674793243408, "learning_rate": 1.8340437936268667e-05, "loss": 0.17403109073638917, "step": 7990 }, { "epoch": 0.2107321955989898, "grad_norm": 4.156668663024902, "learning_rate": 1.8335728158297374e-05, "loss": 0.1811628222465515, "step": 8000 }, { "epoch": 0.21099561084348853, "grad_norm": 0.6621637940406799, "learning_rate": 1.833101231322349e-05, "loss": 0.1757247567176819, "step": 8010 }, { "epoch": 0.21125902608798727, "grad_norm": 3.0950112342834473, "learning_rate": 1.832629040447942e-05, "loss": 0.17991694211959838, "step": 8020 }, { "epoch": 0.21152244133248602, "grad_norm": 1.7668782472610474, "learning_rate": 1.8321562435501963e-05, "loss": 0.1901179790496826, "step": 8030 }, { "epoch": 0.21178585657698476, "grad_norm": 2.7912442684173584, "learning_rate": 1.831682840973234e-05, "loss": 0.16844961643218995, "step": 8040 }, { "epoch": 0.21204927182148348, "grad_norm": 1.8175852298736572, "learning_rate": 1.8312088330616188e-05, "loss": 0.1730794906616211, "step": 8050 }, { "epoch": 0.21231268706598222, "grad_norm": 1.1191699504852295, "learning_rate": 1.8307342201603533e-05, "loss": 0.16093212366104126, "step": 8060 }, { "epoch": 0.21257610231048096, "grad_norm": 1.558603048324585, "learning_rate": 1.830259002614881e-05, "loss": 0.15285627841949462, "step": 8070 }, { "epoch": 0.2128395175549797, "grad_norm": 0.8829283714294434, "learning_rate": 1.829783180771087e-05, "loss": 0.17920248508453368, "step": 8080 }, { "epoch": 0.21310293279947845, "grad_norm": 2.097639322280884, "learning_rate": 1.8293067549752937e-05, "loss": 0.17404958009719848, "step": 8090 }, { "epoch": 0.21336634804397717, "grad_norm": 1.0718662738800049, "learning_rate": 1.8288297255742646e-05, "loss": 0.19177900552749633, "step": 8100 }, { "epoch": 0.2136297632884759, "grad_norm": 3.121816396713257, "learning_rate": 1.828352092915203e-05, "loss": 0.17738124132156372, "step": 8110 }, { "epoch": 0.21389317853297465, "grad_norm": 0.6742199659347534, "learning_rate": 1.8278738573457496e-05, "loss": 0.1548532247543335, "step": 8120 }, { "epoch": 0.2141565937774734, "grad_norm": 1.306997299194336, "learning_rate": 1.8273950192139854e-05, "loss": 0.1647806406021118, "step": 8130 }, { "epoch": 0.2144200090219721, "grad_norm": 4.177762985229492, "learning_rate": 1.826915578868429e-05, "loss": 0.18046531677246094, "step": 8140 }, { "epoch": 0.21468342426647086, "grad_norm": 3.234393358230591, "learning_rate": 1.8264355366580387e-05, "loss": 0.1679924726486206, "step": 8150 }, { "epoch": 0.2149468395109696, "grad_norm": 0.9188328385353088, "learning_rate": 1.825954892932209e-05, "loss": 0.16430363655090333, "step": 8160 }, { "epoch": 0.21521025475546834, "grad_norm": 1.9639832973480225, "learning_rate": 1.8254736480407738e-05, "loss": 0.16494343280792237, "step": 8170 }, { "epoch": 0.21547366999996706, "grad_norm": 1.3393019437789917, "learning_rate": 1.8249918023340036e-05, "loss": 0.1635584592819214, "step": 8180 }, { "epoch": 0.2157370852444658, "grad_norm": 4.124630451202393, "learning_rate": 1.8245093561626063e-05, "loss": 0.17528971433639526, "step": 8190 }, { "epoch": 0.21600050048896455, "grad_norm": 3.5196664333343506, "learning_rate": 1.824026309877727e-05, "loss": 0.1816201090812683, "step": 8200 }, { "epoch": 0.2162639157334633, "grad_norm": 3.8262627124786377, "learning_rate": 1.8235426638309483e-05, "loss": 0.17898149490356446, "step": 8210 }, { "epoch": 0.21652733097796203, "grad_norm": 1.2265589237213135, "learning_rate": 1.823058418374288e-05, "loss": 0.1753918170928955, "step": 8220 }, { "epoch": 0.21679074622246075, "grad_norm": 1.6316043138504028, "learning_rate": 1.822573573860201e-05, "loss": 0.16264519691467286, "step": 8230 }, { "epoch": 0.2170541614669595, "grad_norm": 0.7442600131034851, "learning_rate": 1.8220881306415783e-05, "loss": 0.16485304832458497, "step": 8240 }, { "epoch": 0.21731757671145824, "grad_norm": 1.5907267332077026, "learning_rate": 1.821602089071746e-05, "loss": 0.1585137963294983, "step": 8250 }, { "epoch": 0.21758099195595698, "grad_norm": 1.7398148775100708, "learning_rate": 1.8211154495044666e-05, "loss": 0.17783811092376708, "step": 8260 }, { "epoch": 0.2178444072004557, "grad_norm": 1.0184011459350586, "learning_rate": 1.8206282122939368e-05, "loss": 0.16027967929840087, "step": 8270 }, { "epoch": 0.21810782244495444, "grad_norm": 2.5547149181365967, "learning_rate": 1.8201403777947896e-05, "loss": 0.16030657291412354, "step": 8280 }, { "epoch": 0.21837123768945318, "grad_norm": 0.7442046403884888, "learning_rate": 1.8196519463620912e-05, "loss": 0.1714733600616455, "step": 8290 }, { "epoch": 0.21863465293395193, "grad_norm": 1.8921680450439453, "learning_rate": 1.8191629183513435e-05, "loss": 0.16230124235153198, "step": 8300 }, { "epoch": 0.21889806817845064, "grad_norm": 0.7958590984344482, "learning_rate": 1.8186732941184824e-05, "loss": 0.16609933376312255, "step": 8310 }, { "epoch": 0.2191614834229494, "grad_norm": 2.1584560871124268, "learning_rate": 1.818183074019877e-05, "loss": 0.16997039318084717, "step": 8320 }, { "epoch": 0.21942489866744813, "grad_norm": 2.5897159576416016, "learning_rate": 1.8176922584123308e-05, "loss": 0.1619265556335449, "step": 8330 }, { "epoch": 0.21968831391194688, "grad_norm": 3.5465550422668457, "learning_rate": 1.817200847653081e-05, "loss": 0.18482214212417603, "step": 8340 }, { "epoch": 0.21995172915644562, "grad_norm": 1.4426968097686768, "learning_rate": 1.8167088420997975e-05, "loss": 0.1826837420463562, "step": 8350 }, { "epoch": 0.22021514440094433, "grad_norm": 2.5950539112091064, "learning_rate": 1.816216242110582e-05, "loss": 0.17640061378479005, "step": 8360 }, { "epoch": 0.22047855964544308, "grad_norm": 1.5350110530853271, "learning_rate": 1.815723048043972e-05, "loss": 0.16160162687301635, "step": 8370 }, { "epoch": 0.22074197488994182, "grad_norm": 1.8456950187683105, "learning_rate": 1.8152292602589336e-05, "loss": 0.167228627204895, "step": 8380 }, { "epoch": 0.22100539013444057, "grad_norm": 0.7883056402206421, "learning_rate": 1.8147348791148678e-05, "loss": 0.1818044066429138, "step": 8390 }, { "epoch": 0.22126880537893928, "grad_norm": 0.9891846776008606, "learning_rate": 1.8142399049716065e-05, "loss": 0.16165895462036134, "step": 8400 }, { "epoch": 0.22153222062343803, "grad_norm": 2.0356295108795166, "learning_rate": 1.8137443381894126e-05, "loss": 0.18389130830764772, "step": 8410 }, { "epoch": 0.22179563586793677, "grad_norm": 0.9215615391731262, "learning_rate": 1.813248179128982e-05, "loss": 0.16275219917297362, "step": 8420 }, { "epoch": 0.2220590511124355, "grad_norm": 1.037361741065979, "learning_rate": 1.81275142815144e-05, "loss": 0.17517465353012085, "step": 8430 }, { "epoch": 0.22232246635693423, "grad_norm": 3.445150375366211, "learning_rate": 1.812254085618343e-05, "loss": 0.17520109415054322, "step": 8440 }, { "epoch": 0.22258588160143297, "grad_norm": 3.609097480773926, "learning_rate": 1.8117561518916792e-05, "loss": 0.16599847078323365, "step": 8450 }, { "epoch": 0.22284929684593172, "grad_norm": 2.0366194248199463, "learning_rate": 1.811257627333866e-05, "loss": 0.16538599729537964, "step": 8460 }, { "epoch": 0.22311271209043046, "grad_norm": 0.7842395901679993, "learning_rate": 1.810758512307751e-05, "loss": 0.16986621618270875, "step": 8470 }, { "epoch": 0.2233761273349292, "grad_norm": 4.399916172027588, "learning_rate": 1.8102588071766116e-05, "loss": 0.17613527774810792, "step": 8480 }, { "epoch": 0.22363954257942792, "grad_norm": 2.0540618896484375, "learning_rate": 1.8097585123041547e-05, "loss": 0.16546733379364015, "step": 8490 }, { "epoch": 0.22390295782392666, "grad_norm": 2.762781858444214, "learning_rate": 1.809257628054517e-05, "loss": 0.16611781120300292, "step": 8500 }, { "epoch": 0.2241663730684254, "grad_norm": 1.5640153884887695, "learning_rate": 1.808756154792263e-05, "loss": 0.1721196174621582, "step": 8510 }, { "epoch": 0.22442978831292415, "grad_norm": 3.073481321334839, "learning_rate": 1.8082540928823866e-05, "loss": 0.17447092533111572, "step": 8520 }, { "epoch": 0.22469320355742287, "grad_norm": 1.129050612449646, "learning_rate": 1.8077514426903107e-05, "loss": 0.16736114025115967, "step": 8530 }, { "epoch": 0.2249566188019216, "grad_norm": 2.1508867740631104, "learning_rate": 1.8072482045818852e-05, "loss": 0.1555781126022339, "step": 8540 }, { "epoch": 0.22522003404642035, "grad_norm": 2.7976982593536377, "learning_rate": 1.806744378923389e-05, "loss": 0.17083005905151366, "step": 8550 }, { "epoch": 0.2254834492909191, "grad_norm": 2.3208539485931396, "learning_rate": 1.8062399660815277e-05, "loss": 0.1732415199279785, "step": 8560 }, { "epoch": 0.2257468645354178, "grad_norm": 1.1212955713272095, "learning_rate": 1.8057349664234348e-05, "loss": 0.16085344552993774, "step": 8570 }, { "epoch": 0.22601027977991656, "grad_norm": 1.7316492795944214, "learning_rate": 1.8052293803166714e-05, "loss": 0.1726798892021179, "step": 8580 }, { "epoch": 0.2262736950244153, "grad_norm": 1.3166216611862183, "learning_rate": 1.8047232081292242e-05, "loss": 0.1585933804512024, "step": 8590 }, { "epoch": 0.22653711026891404, "grad_norm": 1.789621114730835, "learning_rate": 1.8042164502295074e-05, "loss": 0.18410954475402833, "step": 8600 }, { "epoch": 0.2268005255134128, "grad_norm": 0.8585262894630432, "learning_rate": 1.8037091069863615e-05, "loss": 0.1661034941673279, "step": 8610 }, { "epoch": 0.2270639407579115, "grad_norm": 2.4551267623901367, "learning_rate": 1.8032011787690523e-05, "loss": 0.16096916198730468, "step": 8620 }, { "epoch": 0.22732735600241025, "grad_norm": 2.8925764560699463, "learning_rate": 1.8026926659472727e-05, "loss": 0.1601356029510498, "step": 8630 }, { "epoch": 0.227590771246909, "grad_norm": 1.3881791830062866, "learning_rate": 1.802183568891139e-05, "loss": 0.17179462909698487, "step": 8640 }, { "epoch": 0.22785418649140773, "grad_norm": 2.5808005332946777, "learning_rate": 1.801673887971195e-05, "loss": 0.1789689302444458, "step": 8650 }, { "epoch": 0.22811760173590645, "grad_norm": 1.472485899925232, "learning_rate": 1.801163623558408e-05, "loss": 0.1715448260307312, "step": 8660 }, { "epoch": 0.2283810169804052, "grad_norm": 1.055450201034546, "learning_rate": 1.800652776024171e-05, "loss": 0.16966888904571534, "step": 8670 }, { "epoch": 0.22864443222490394, "grad_norm": 2.0093460083007812, "learning_rate": 1.8001413457403005e-05, "loss": 0.16870408058166503, "step": 8680 }, { "epoch": 0.22890784746940268, "grad_norm": 2.209728479385376, "learning_rate": 1.7996293330790374e-05, "loss": 0.17196786403656006, "step": 8690 }, { "epoch": 0.2291712627139014, "grad_norm": 1.0473684072494507, "learning_rate": 1.7991167384130466e-05, "loss": 0.16634092330932618, "step": 8700 }, { "epoch": 0.22943467795840014, "grad_norm": 1.33102285861969, "learning_rate": 1.798603562115417e-05, "loss": 0.16636266708374023, "step": 8710 }, { "epoch": 0.22969809320289888, "grad_norm": 1.1791363954544067, "learning_rate": 1.79808980455966e-05, "loss": 0.16383336782455443, "step": 8720 }, { "epoch": 0.22996150844739763, "grad_norm": 1.122530221939087, "learning_rate": 1.7975754661197106e-05, "loss": 0.1688985824584961, "step": 8730 }, { "epoch": 0.23022492369189637, "grad_norm": 1.7277724742889404, "learning_rate": 1.797060547169927e-05, "loss": 0.17150311470031737, "step": 8740 }, { "epoch": 0.2304883389363951, "grad_norm": 1.9370720386505127, "learning_rate": 1.796545048085089e-05, "loss": 0.16769204139709473, "step": 8750 }, { "epoch": 0.23075175418089383, "grad_norm": 1.301882266998291, "learning_rate": 1.796028969240399e-05, "loss": 0.1683528184890747, "step": 8760 }, { "epoch": 0.23101516942539257, "grad_norm": 1.3902573585510254, "learning_rate": 1.7955123110114815e-05, "loss": 0.17253830432891845, "step": 8770 }, { "epoch": 0.23127858466989132, "grad_norm": 1.3033952713012695, "learning_rate": 1.794995073774383e-05, "loss": 0.16027021408081055, "step": 8780 }, { "epoch": 0.23154199991439003, "grad_norm": 0.971534013748169, "learning_rate": 1.7944772579055705e-05, "loss": 0.17855416536331176, "step": 8790 }, { "epoch": 0.23180541515888878, "grad_norm": 3.3209939002990723, "learning_rate": 1.793958863781933e-05, "loss": 0.1659646987915039, "step": 8800 }, { "epoch": 0.23206883040338752, "grad_norm": 2.5961179733276367, "learning_rate": 1.7934398917807802e-05, "loss": 0.19653483629226684, "step": 8810 }, { "epoch": 0.23233224564788627, "grad_norm": 1.4196664094924927, "learning_rate": 1.7929203422798426e-05, "loss": 0.1637791872024536, "step": 8820 }, { "epoch": 0.23259566089238498, "grad_norm": 1.4622281789779663, "learning_rate": 1.79240021565727e-05, "loss": 0.1582207202911377, "step": 8830 }, { "epoch": 0.23285907613688372, "grad_norm": 6.3339314460754395, "learning_rate": 1.7918795122916337e-05, "loss": 0.16570701599121093, "step": 8840 }, { "epoch": 0.23312249138138247, "grad_norm": 1.4018630981445312, "learning_rate": 1.7913582325619236e-05, "loss": 0.17137608528137208, "step": 8850 }, { "epoch": 0.2333859066258812, "grad_norm": 3.455718517303467, "learning_rate": 1.7908363768475498e-05, "loss": 0.17441163063049317, "step": 8860 }, { "epoch": 0.23364932187037996, "grad_norm": 2.348400115966797, "learning_rate": 1.7903139455283412e-05, "loss": 0.17155711650848388, "step": 8870 }, { "epoch": 0.23391273711487867, "grad_norm": 5.272608757019043, "learning_rate": 1.7897909389845462e-05, "loss": 0.17202417850494384, "step": 8880 }, { "epoch": 0.23417615235937742, "grad_norm": 5.240749359130859, "learning_rate": 1.789267357596831e-05, "loss": 0.16425819396972657, "step": 8890 }, { "epoch": 0.23443956760387616, "grad_norm": 1.1603704690933228, "learning_rate": 1.7887432017462812e-05, "loss": 0.1758279323577881, "step": 8900 }, { "epoch": 0.2347029828483749, "grad_norm": 4.817521095275879, "learning_rate": 1.7882184718144002e-05, "loss": 0.17808241844177247, "step": 8910 }, { "epoch": 0.23496639809287362, "grad_norm": 3.2170019149780273, "learning_rate": 1.7876931681831087e-05, "loss": 0.17218095064163208, "step": 8920 }, { "epoch": 0.23522981333737236, "grad_norm": 1.5848686695098877, "learning_rate": 1.787167291234745e-05, "loss": 0.166497802734375, "step": 8930 }, { "epoch": 0.2354932285818711, "grad_norm": 6.035172462463379, "learning_rate": 1.7866408413520662e-05, "loss": 0.1605350375175476, "step": 8940 }, { "epoch": 0.23575664382636985, "grad_norm": 2.055878162384033, "learning_rate": 1.786113818918244e-05, "loss": 0.18303200006484985, "step": 8950 }, { "epoch": 0.23602005907086857, "grad_norm": 3.612069845199585, "learning_rate": 1.7855862243168684e-05, "loss": 0.161812686920166, "step": 8960 }, { "epoch": 0.2362834743153673, "grad_norm": 2.5722765922546387, "learning_rate": 1.785058057931946e-05, "loss": 0.17696616649627686, "step": 8970 }, { "epoch": 0.23654688955986605, "grad_norm": 2.760464906692505, "learning_rate": 1.7845293201478985e-05, "loss": 0.18766918182373046, "step": 8980 }, { "epoch": 0.2368103048043648, "grad_norm": 0.9650112390518188, "learning_rate": 1.7840000113495647e-05, "loss": 0.1719886064529419, "step": 8990 }, { "epoch": 0.23707372004886354, "grad_norm": 5.3266921043396, "learning_rate": 1.7834701319221978e-05, "loss": 0.16940996646881104, "step": 9000 }, { "epoch": 0.23733713529336226, "grad_norm": 1.8363152742385864, "learning_rate": 1.782939682251467e-05, "loss": 0.17262870073318481, "step": 9010 }, { "epoch": 0.237600550537861, "grad_norm": 2.8939313888549805, "learning_rate": 1.7824086627234565e-05, "loss": 0.1671499490737915, "step": 9020 }, { "epoch": 0.23786396578235974, "grad_norm": 4.427179336547852, "learning_rate": 1.7818770737246653e-05, "loss": 0.17268772125244142, "step": 9030 }, { "epoch": 0.2381273810268585, "grad_norm": 0.9220793843269348, "learning_rate": 1.7813449156420072e-05, "loss": 0.14996596574783325, "step": 9040 }, { "epoch": 0.2383907962713572, "grad_norm": 1.9942970275878906, "learning_rate": 1.7808121888628093e-05, "loss": 0.17314144372940063, "step": 9050 }, { "epoch": 0.23865421151585595, "grad_norm": 1.4299087524414062, "learning_rate": 1.780278893774813e-05, "loss": 0.16504154205322266, "step": 9060 }, { "epoch": 0.2389176267603547, "grad_norm": 2.47707462310791, "learning_rate": 1.7797450307661733e-05, "loss": 0.16623027324676515, "step": 9070 }, { "epoch": 0.23918104200485343, "grad_norm": 3.3927221298217773, "learning_rate": 1.77921060022546e-05, "loss": 0.1655115604400635, "step": 9080 }, { "epoch": 0.23944445724935215, "grad_norm": 1.6047048568725586, "learning_rate": 1.7786756025416535e-05, "loss": 0.17488632202148438, "step": 9090 }, { "epoch": 0.2397078724938509, "grad_norm": 1.7465187311172485, "learning_rate": 1.7781400381041484e-05, "loss": 0.17150264978408813, "step": 9100 }, { "epoch": 0.23997128773834964, "grad_norm": 1.5738862752914429, "learning_rate": 1.777603907302752e-05, "loss": 0.16510356664657594, "step": 9110 }, { "epoch": 0.24023470298284838, "grad_norm": 1.4007190465927124, "learning_rate": 1.7770672105276833e-05, "loss": 0.15256062746047974, "step": 9120 }, { "epoch": 0.24049811822734712, "grad_norm": 2.826395273208618, "learning_rate": 1.776529948169573e-05, "loss": 0.16575222015380858, "step": 9130 }, { "epoch": 0.24076153347184584, "grad_norm": 1.7570511102676392, "learning_rate": 1.775992120619464e-05, "loss": 0.17636556625366212, "step": 9140 }, { "epoch": 0.24102494871634458, "grad_norm": 1.3200613260269165, "learning_rate": 1.7754537282688104e-05, "loss": 0.16396206617355347, "step": 9150 }, { "epoch": 0.24128836396084333, "grad_norm": 3.0087194442749023, "learning_rate": 1.7749147715094777e-05, "loss": 0.16225023269653321, "step": 9160 }, { "epoch": 0.24155177920534207, "grad_norm": 0.8839743137359619, "learning_rate": 1.7743752507337414e-05, "loss": 0.17751678228378295, "step": 9170 }, { "epoch": 0.2418151944498408, "grad_norm": 2.9908218383789062, "learning_rate": 1.7738351663342887e-05, "loss": 0.17295455932617188, "step": 9180 }, { "epoch": 0.24207860969433953, "grad_norm": 1.5697158575057983, "learning_rate": 1.7732945187042155e-05, "loss": 0.15962339639663697, "step": 9190 }, { "epoch": 0.24234202493883827, "grad_norm": 1.2140413522720337, "learning_rate": 1.7727533082370288e-05, "loss": 0.15723097324371338, "step": 9200 }, { "epoch": 0.24260544018333702, "grad_norm": 1.3200987577438354, "learning_rate": 1.772211535326645e-05, "loss": 0.1830831289291382, "step": 9210 }, { "epoch": 0.24286885542783573, "grad_norm": 2.257219076156616, "learning_rate": 1.7716692003673898e-05, "loss": 0.159094500541687, "step": 9220 }, { "epoch": 0.24313227067233448, "grad_norm": 3.1759414672851562, "learning_rate": 1.7711263037539978e-05, "loss": 0.1847006320953369, "step": 9230 }, { "epoch": 0.24339568591683322, "grad_norm": 3.274716377258301, "learning_rate": 1.770582845881613e-05, "loss": 0.17125356197357178, "step": 9240 }, { "epoch": 0.24365910116133196, "grad_norm": 1.1672812700271606, "learning_rate": 1.7700388271457872e-05, "loss": 0.16980671882629395, "step": 9250 }, { "epoch": 0.2439225164058307, "grad_norm": 0.7783506512641907, "learning_rate": 1.7694942479424803e-05, "loss": 0.16746120452880858, "step": 9260 }, { "epoch": 0.24418593165032942, "grad_norm": 1.6577924489974976, "learning_rate": 1.768949108668061e-05, "loss": 0.15681886672973633, "step": 9270 }, { "epoch": 0.24444934689482817, "grad_norm": 1.4289088249206543, "learning_rate": 1.7684034097193055e-05, "loss": 0.15725910663604736, "step": 9280 }, { "epoch": 0.2447127621393269, "grad_norm": 4.178260803222656, "learning_rate": 1.7678571514933966e-05, "loss": 0.17422304153442383, "step": 9290 }, { "epoch": 0.24497617738382566, "grad_norm": 3.2299821376800537, "learning_rate": 1.7673103343879242e-05, "loss": 0.16337347030639648, "step": 9300 }, { "epoch": 0.24523959262832437, "grad_norm": 1.9450066089630127, "learning_rate": 1.7667629588008862e-05, "loss": 0.1590017557144165, "step": 9310 }, { "epoch": 0.24550300787282311, "grad_norm": 1.0313754081726074, "learning_rate": 1.766215025130686e-05, "loss": 0.15866444110870362, "step": 9320 }, { "epoch": 0.24576642311732186, "grad_norm": 1.0061222314834595, "learning_rate": 1.7656665337761327e-05, "loss": 0.16547775268554688, "step": 9330 }, { "epoch": 0.2460298383618206, "grad_norm": 1.11082923412323, "learning_rate": 1.7651174851364427e-05, "loss": 0.1635960578918457, "step": 9340 }, { "epoch": 0.24629325360631932, "grad_norm": 6.710975170135498, "learning_rate": 1.764567879611237e-05, "loss": 0.1686002492904663, "step": 9350 }, { "epoch": 0.24655666885081806, "grad_norm": 3.165215253829956, "learning_rate": 1.7640177176005425e-05, "loss": 0.16412581205368043, "step": 9360 }, { "epoch": 0.2468200840953168, "grad_norm": 1.3674025535583496, "learning_rate": 1.7634669995047904e-05, "loss": 0.18153434991836548, "step": 9370 }, { "epoch": 0.24708349933981555, "grad_norm": 1.9669017791748047, "learning_rate": 1.7629157257248175e-05, "loss": 0.15959157943725585, "step": 9380 }, { "epoch": 0.2473469145843143, "grad_norm": 1.0165603160858154, "learning_rate": 1.762363896661864e-05, "loss": 0.16267757415771483, "step": 9390 }, { "epoch": 0.247610329828813, "grad_norm": 4.412399768829346, "learning_rate": 1.761811512717576e-05, "loss": 0.16770037412643432, "step": 9400 }, { "epoch": 0.24787374507331175, "grad_norm": 1.6328822374343872, "learning_rate": 1.761258574294001e-05, "loss": 0.1492958188056946, "step": 9410 }, { "epoch": 0.2481371603178105, "grad_norm": 1.8419150114059448, "learning_rate": 1.760705081793593e-05, "loss": 0.16460981369018554, "step": 9420 }, { "epoch": 0.24840057556230924, "grad_norm": 3.7884559631347656, "learning_rate": 1.760151035619206e-05, "loss": 0.17848763465881348, "step": 9430 }, { "epoch": 0.24866399080680796, "grad_norm": 1.1245852708816528, "learning_rate": 1.7595964361741e-05, "loss": 0.16498491764068604, "step": 9440 }, { "epoch": 0.2489274060513067, "grad_norm": 3.312364101409912, "learning_rate": 1.759041283861936e-05, "loss": 0.17213313579559325, "step": 9450 }, { "epoch": 0.24919082129580544, "grad_norm": 0.8783972859382629, "learning_rate": 1.7584855790867773e-05, "loss": 0.1637056827545166, "step": 9460 }, { "epoch": 0.2494542365403042, "grad_norm": 1.0045013427734375, "learning_rate": 1.7579293222530907e-05, "loss": 0.1707191586494446, "step": 9470 }, { "epoch": 0.24971765178480293, "grad_norm": 1.2508455514907837, "learning_rate": 1.7573725137657432e-05, "loss": 0.16513433456420898, "step": 9480 }, { "epoch": 0.24998106702930165, "grad_norm": 2.997368097305298, "learning_rate": 1.756815154030005e-05, "loss": 0.16477596759796143, "step": 9490 }, { "epoch": 0.2502444822738004, "grad_norm": 0.510978639125824, "learning_rate": 1.756257243451545e-05, "loss": 0.15883625745773317, "step": 9500 }, { "epoch": 0.2505078975182991, "grad_norm": 1.077164649963379, "learning_rate": 1.7556987824364363e-05, "loss": 0.16225591897964478, "step": 9510 }, { "epoch": 0.25077131276279785, "grad_norm": 2.460937976837158, "learning_rate": 1.75513977139115e-05, "loss": 0.16436704397201538, "step": 9520 }, { "epoch": 0.2510347280072966, "grad_norm": 4.657648086547852, "learning_rate": 1.7545802107225585e-05, "loss": 0.1768571138381958, "step": 9530 }, { "epoch": 0.25129814325179534, "grad_norm": 1.5831584930419922, "learning_rate": 1.754020100837934e-05, "loss": 0.16987011432647706, "step": 9540 }, { "epoch": 0.2515615584962941, "grad_norm": 1.2210910320281982, "learning_rate": 1.753459442144949e-05, "loss": 0.17678617238998412, "step": 9550 }, { "epoch": 0.2518249737407928, "grad_norm": 4.004513740539551, "learning_rate": 1.7528982350516754e-05, "loss": 0.1637563705444336, "step": 9560 }, { "epoch": 0.25208838898529157, "grad_norm": 2.0207552909851074, "learning_rate": 1.7523364799665832e-05, "loss": 0.16130634546279907, "step": 9570 }, { "epoch": 0.2523518042297903, "grad_norm": 2.3277392387390137, "learning_rate": 1.7517741772985424e-05, "loss": 0.1711869716644287, "step": 9580 }, { "epoch": 0.252615219474289, "grad_norm": 0.7696321606636047, "learning_rate": 1.751211327456821e-05, "loss": 0.16950148344039917, "step": 9590 }, { "epoch": 0.25287863471878774, "grad_norm": 4.733989238739014, "learning_rate": 1.7506479308510855e-05, "loss": 0.1673688769340515, "step": 9600 }, { "epoch": 0.2531420499632865, "grad_norm": 3.9727280139923096, "learning_rate": 1.7500839878914004e-05, "loss": 0.16299231052398683, "step": 9610 }, { "epoch": 0.25340546520778523, "grad_norm": 2.810917377471924, "learning_rate": 1.749519498988227e-05, "loss": 0.177673864364624, "step": 9620 }, { "epoch": 0.253668880452284, "grad_norm": 4.048280239105225, "learning_rate": 1.748954464552426e-05, "loss": 0.1617470145225525, "step": 9630 }, { "epoch": 0.2539322956967827, "grad_norm": 1.1418534517288208, "learning_rate": 1.7483888849952525e-05, "loss": 0.1752826452255249, "step": 9640 }, { "epoch": 0.25419571094128146, "grad_norm": 2.037736177444458, "learning_rate": 1.7478227607283604e-05, "loss": 0.15677525997161865, "step": 9650 }, { "epoch": 0.2544591261857802, "grad_norm": 0.8223567605018616, "learning_rate": 1.7472560921637993e-05, "loss": 0.1708451509475708, "step": 9660 }, { "epoch": 0.25472254143027895, "grad_norm": 4.377071857452393, "learning_rate": 1.746688879714015e-05, "loss": 0.1653681755065918, "step": 9670 }, { "epoch": 0.25498595667477764, "grad_norm": 0.848404049873352, "learning_rate": 1.746121123791849e-05, "loss": 0.16294776201248168, "step": 9680 }, { "epoch": 0.2552493719192764, "grad_norm": 3.824845314025879, "learning_rate": 1.7455528248105385e-05, "loss": 0.17940263748168944, "step": 9690 }, { "epoch": 0.2555127871637751, "grad_norm": 1.2720822095870972, "learning_rate": 1.7449839831837166e-05, "loss": 0.17231266498565673, "step": 9700 }, { "epoch": 0.25577620240827387, "grad_norm": 1.7779508829116821, "learning_rate": 1.7444145993254105e-05, "loss": 0.14654734134674072, "step": 9710 }, { "epoch": 0.2560396176527726, "grad_norm": 0.8414924144744873, "learning_rate": 1.7438446736500418e-05, "loss": 0.18227559328079224, "step": 9720 }, { "epoch": 0.25630303289727135, "grad_norm": 2.786273956298828, "learning_rate": 1.7432742065724276e-05, "loss": 0.16732425689697267, "step": 9730 }, { "epoch": 0.2565664481417701, "grad_norm": 2.525033950805664, "learning_rate": 1.7427031985077785e-05, "loss": 0.17112348079681397, "step": 9740 }, { "epoch": 0.25682986338626884, "grad_norm": 1.4743099212646484, "learning_rate": 1.7421316498716986e-05, "loss": 0.14785475730895997, "step": 9750 }, { "epoch": 0.2570932786307676, "grad_norm": 3.525759220123291, "learning_rate": 1.741559561080185e-05, "loss": 0.15681469440460205, "step": 9760 }, { "epoch": 0.2573566938752663, "grad_norm": 1.514054775238037, "learning_rate": 1.74098693254963e-05, "loss": 0.14014346599578859, "step": 9770 }, { "epoch": 0.257620109119765, "grad_norm": 1.568541407585144, "learning_rate": 1.7404137646968162e-05, "loss": 0.16313583850860597, "step": 9780 }, { "epoch": 0.25788352436426376, "grad_norm": 2.591898202896118, "learning_rate": 1.7398400579389204e-05, "loss": 0.15333327054977416, "step": 9790 }, { "epoch": 0.2581469396087625, "grad_norm": 3.1672284603118896, "learning_rate": 1.739265812693511e-05, "loss": 0.1704309105873108, "step": 9800 }, { "epoch": 0.25841035485326125, "grad_norm": 2.9400928020477295, "learning_rate": 1.7386910293785485e-05, "loss": 0.15977680683135986, "step": 9810 }, { "epoch": 0.25867377009776, "grad_norm": 2.387079954147339, "learning_rate": 1.738115708412385e-05, "loss": 0.15688204765319824, "step": 9820 }, { "epoch": 0.25893718534225874, "grad_norm": 1.621545672416687, "learning_rate": 1.737539850213764e-05, "loss": 0.16624432802200317, "step": 9830 }, { "epoch": 0.2592006005867575, "grad_norm": 1.6322239637374878, "learning_rate": 1.7369634552018196e-05, "loss": 0.16625096797943115, "step": 9840 }, { "epoch": 0.25946401583125617, "grad_norm": 2.097468852996826, "learning_rate": 1.7363865237960773e-05, "loss": 0.16877232789993285, "step": 9850 }, { "epoch": 0.2597274310757549, "grad_norm": 2.3330469131469727, "learning_rate": 1.7358090564164524e-05, "loss": 0.16030220985412597, "step": 9860 }, { "epoch": 0.25999084632025365, "grad_norm": 0.9654650092124939, "learning_rate": 1.735231053483251e-05, "loss": 0.17085236310958862, "step": 9870 }, { "epoch": 0.2602542615647524, "grad_norm": 0.7602819800376892, "learning_rate": 1.734652515417168e-05, "loss": 0.17308129072189332, "step": 9880 }, { "epoch": 0.26051767680925114, "grad_norm": 1.014905571937561, "learning_rate": 1.7340734426392894e-05, "loss": 0.1676319122314453, "step": 9890 }, { "epoch": 0.2607810920537499, "grad_norm": 2.4526937007904053, "learning_rate": 1.7334938355710876e-05, "loss": 0.176541006565094, "step": 9900 }, { "epoch": 0.26104450729824863, "grad_norm": 1.8713016510009766, "learning_rate": 1.7329136946344272e-05, "loss": 0.17023603916168212, "step": 9910 }, { "epoch": 0.2613079225427474, "grad_norm": 1.3603559732437134, "learning_rate": 1.7323330202515592e-05, "loss": 0.16435494422912597, "step": 9920 }, { "epoch": 0.2615713377872461, "grad_norm": 2.757355213165283, "learning_rate": 1.731751812845124e-05, "loss": 0.18296506404876708, "step": 9930 }, { "epoch": 0.2618347530317448, "grad_norm": 0.7415362000465393, "learning_rate": 1.7311700728381486e-05, "loss": 0.1618773341178894, "step": 9940 }, { "epoch": 0.26209816827624355, "grad_norm": 2.8806328773498535, "learning_rate": 1.7305878006540483e-05, "loss": 0.16831471920013427, "step": 9950 }, { "epoch": 0.2623615835207423, "grad_norm": 1.9759142398834229, "learning_rate": 1.7300049967166273e-05, "loss": 0.17108780145645142, "step": 9960 }, { "epoch": 0.26262499876524104, "grad_norm": 1.3531625270843506, "learning_rate": 1.729421661450074e-05, "loss": 0.16224803924560546, "step": 9970 }, { "epoch": 0.2628884140097398, "grad_norm": 2.2369844913482666, "learning_rate": 1.728837795278966e-05, "loss": 0.1642148733139038, "step": 9980 }, { "epoch": 0.2631518292542385, "grad_norm": 3.7866296768188477, "learning_rate": 1.7282533986282656e-05, "loss": 0.16824781894683838, "step": 9990 }, { "epoch": 0.26341524449873727, "grad_norm": 8.550382614135742, "learning_rate": 1.7276684719233222e-05, "loss": 0.18029210567474366, "step": 10000 }, { "epoch": 0.263678659743236, "grad_norm": 2.824876070022583, "learning_rate": 1.7270830155898704e-05, "loss": 0.186208176612854, "step": 10010 }, { "epoch": 0.26394207498773475, "grad_norm": 1.403643012046814, "learning_rate": 1.726497030054031e-05, "loss": 0.17231040000915526, "step": 10020 }, { "epoch": 0.26420549023223344, "grad_norm": 1.0046170949935913, "learning_rate": 1.72591051574231e-05, "loss": 0.16417150497436522, "step": 10030 }, { "epoch": 0.2644689054767322, "grad_norm": 1.5648057460784912, "learning_rate": 1.7253234730815964e-05, "loss": 0.17795683145523072, "step": 10040 }, { "epoch": 0.26473232072123093, "grad_norm": 1.6092174053192139, "learning_rate": 1.724735902499166e-05, "loss": 0.1732587218284607, "step": 10050 }, { "epoch": 0.2649957359657297, "grad_norm": 1.0660184621810913, "learning_rate": 1.7241478044226782e-05, "loss": 0.1711615204811096, "step": 10060 }, { "epoch": 0.2652591512102284, "grad_norm": 1.290484070777893, "learning_rate": 1.723559179280176e-05, "loss": 0.16897355318069457, "step": 10070 }, { "epoch": 0.26552256645472716, "grad_norm": 1.7811274528503418, "learning_rate": 1.722970027500086e-05, "loss": 0.18032667636871338, "step": 10080 }, { "epoch": 0.2657859816992259, "grad_norm": 3.206664800643921, "learning_rate": 1.7223803495112185e-05, "loss": 0.1641620874404907, "step": 10090 }, { "epoch": 0.26604939694372465, "grad_norm": 1.153622031211853, "learning_rate": 1.7217901457427667e-05, "loss": 0.15750811100006104, "step": 10100 }, { "epoch": 0.26631281218822334, "grad_norm": 2.5803945064544678, "learning_rate": 1.7211994166243058e-05, "loss": 0.17133831977844238, "step": 10110 }, { "epoch": 0.2665762274327221, "grad_norm": 1.14617919921875, "learning_rate": 1.7206081625857943e-05, "loss": 0.1734297513961792, "step": 10120 }, { "epoch": 0.2668396426772208, "grad_norm": 2.9049439430236816, "learning_rate": 1.7200163840575728e-05, "loss": 0.16900765895843506, "step": 10130 }, { "epoch": 0.26710305792171957, "grad_norm": 2.961559772491455, "learning_rate": 1.7194240814703626e-05, "loss": 0.16533963680267333, "step": 10140 }, { "epoch": 0.2673664731662183, "grad_norm": 1.2749592065811157, "learning_rate": 1.7188312552552672e-05, "loss": 0.16905884742736815, "step": 10150 }, { "epoch": 0.26762988841071705, "grad_norm": 3.510164260864258, "learning_rate": 1.718237905843771e-05, "loss": 0.16585731506347656, "step": 10160 }, { "epoch": 0.2678933036552158, "grad_norm": 2.756028413772583, "learning_rate": 1.71764403366774e-05, "loss": 0.17004435062408446, "step": 10170 }, { "epoch": 0.26815671889971454, "grad_norm": 3.420605421066284, "learning_rate": 1.7170496391594194e-05, "loss": 0.1619368314743042, "step": 10180 }, { "epoch": 0.2684201341442133, "grad_norm": 1.3713849782943726, "learning_rate": 1.716454722751435e-05, "loss": 0.1604318618774414, "step": 10190 }, { "epoch": 0.268683549388712, "grad_norm": 2.022181749343872, "learning_rate": 1.715859284876793e-05, "loss": 0.16777069568634034, "step": 10200 }, { "epoch": 0.2689469646332107, "grad_norm": 0.7530263662338257, "learning_rate": 1.7152633259688788e-05, "loss": 0.1598554730415344, "step": 10210 }, { "epoch": 0.26921037987770946, "grad_norm": 5.402015209197998, "learning_rate": 1.714666846461457e-05, "loss": 0.1675887942314148, "step": 10220 }, { "epoch": 0.2694737951222082, "grad_norm": 1.7832890748977661, "learning_rate": 1.7140698467886707e-05, "loss": 0.15580607652664186, "step": 10230 }, { "epoch": 0.26973721036670695, "grad_norm": 0.9303087592124939, "learning_rate": 1.713472327385042e-05, "loss": 0.1571975588798523, "step": 10240 }, { "epoch": 0.2700006256112057, "grad_norm": 1.3812167644500732, "learning_rate": 1.7128742886854723e-05, "loss": 0.15094258785247802, "step": 10250 }, { "epoch": 0.27026404085570443, "grad_norm": 1.6795560121536255, "learning_rate": 1.712275731125239e-05, "loss": 0.1702711582183838, "step": 10260 }, { "epoch": 0.2705274561002032, "grad_norm": 5.952234745025635, "learning_rate": 1.7116766551399986e-05, "loss": 0.16068058013916015, "step": 10270 }, { "epoch": 0.2707908713447019, "grad_norm": 1.8487671613693237, "learning_rate": 1.711077061165784e-05, "loss": 0.1610506772994995, "step": 10280 }, { "epoch": 0.2710542865892006, "grad_norm": 4.687558650970459, "learning_rate": 1.710476949639006e-05, "loss": 0.17489041090011598, "step": 10290 }, { "epoch": 0.27131770183369935, "grad_norm": 2.449228048324585, "learning_rate": 1.7098763209964514e-05, "loss": 0.1553255081176758, "step": 10300 }, { "epoch": 0.2715811170781981, "grad_norm": 1.8026924133300781, "learning_rate": 1.7092751756752838e-05, "loss": 0.15568486452102662, "step": 10310 }, { "epoch": 0.27184453232269684, "grad_norm": 1.4915302991867065, "learning_rate": 1.7086735141130423e-05, "loss": 0.16699554920196533, "step": 10320 }, { "epoch": 0.2721079475671956, "grad_norm": 3.103924036026001, "learning_rate": 1.7080713367476428e-05, "loss": 0.16433117389678956, "step": 10330 }, { "epoch": 0.27237136281169433, "grad_norm": 0.6680436730384827, "learning_rate": 1.7074686440173752e-05, "loss": 0.1703255295753479, "step": 10340 }, { "epoch": 0.2726347780561931, "grad_norm": 2.0610246658325195, "learning_rate": 1.7068654363609058e-05, "loss": 0.15826716423034667, "step": 10350 }, { "epoch": 0.2728981933006918, "grad_norm": 2.514132499694824, "learning_rate": 1.7062617142172754e-05, "loss": 0.17867127656936646, "step": 10360 }, { "epoch": 0.2731616085451905, "grad_norm": 2.29616641998291, "learning_rate": 1.7056574780258986e-05, "loss": 0.16741286516189574, "step": 10370 }, { "epoch": 0.27342502378968925, "grad_norm": 2.212372303009033, "learning_rate": 1.705052728226564e-05, "loss": 0.1604917526245117, "step": 10380 }, { "epoch": 0.273688439034188, "grad_norm": 1.437878966331482, "learning_rate": 1.704447465259436e-05, "loss": 0.1598968505859375, "step": 10390 }, { "epoch": 0.27395185427868674, "grad_norm": 3.5340049266815186, "learning_rate": 1.7038416895650504e-05, "loss": 0.1627725839614868, "step": 10400 }, { "epoch": 0.2742152695231855, "grad_norm": 2.2031426429748535, "learning_rate": 1.7032354015843162e-05, "loss": 0.1664196252822876, "step": 10410 }, { "epoch": 0.2744786847676842, "grad_norm": 4.664189338684082, "learning_rate": 1.7026286017585168e-05, "loss": 0.15792784690856934, "step": 10420 }, { "epoch": 0.27474210001218297, "grad_norm": 1.7119652032852173, "learning_rate": 1.7020212905293072e-05, "loss": 0.17192305326461793, "step": 10430 }, { "epoch": 0.2750055152566817, "grad_norm": 1.2959072589874268, "learning_rate": 1.7014134683387147e-05, "loss": 0.17592802047729492, "step": 10440 }, { "epoch": 0.27526893050118045, "grad_norm": 1.1155436038970947, "learning_rate": 1.7008051356291382e-05, "loss": 0.15636602640151978, "step": 10450 }, { "epoch": 0.27553234574567914, "grad_norm": 1.7900842428207397, "learning_rate": 1.7001962928433488e-05, "loss": 0.1728128671646118, "step": 10460 }, { "epoch": 0.2757957609901779, "grad_norm": 2.2117645740509033, "learning_rate": 1.699586940424488e-05, "loss": 0.179030704498291, "step": 10470 }, { "epoch": 0.27605917623467663, "grad_norm": 1.8062154054641724, "learning_rate": 1.6989770788160697e-05, "loss": 0.1714838981628418, "step": 10480 }, { "epoch": 0.2763225914791754, "grad_norm": 4.309415817260742, "learning_rate": 1.6983667084619768e-05, "loss": 0.1706938624382019, "step": 10490 }, { "epoch": 0.2765860067236741, "grad_norm": 1.9639493227005005, "learning_rate": 1.6977558298064636e-05, "loss": 0.16234419345855713, "step": 10500 }, { "epoch": 0.27684942196817286, "grad_norm": 2.2997846603393555, "learning_rate": 1.6971444432941533e-05, "loss": 0.16184593439102174, "step": 10510 }, { "epoch": 0.2771128372126716, "grad_norm": 1.2648286819458008, "learning_rate": 1.69653254937004e-05, "loss": 0.1699685573577881, "step": 10520 }, { "epoch": 0.27737625245717035, "grad_norm": 1.786576509475708, "learning_rate": 1.6959201484794867e-05, "loss": 0.16413400173187256, "step": 10530 }, { "epoch": 0.2776396677016691, "grad_norm": 1.6035809516906738, "learning_rate": 1.6953072410682243e-05, "loss": 0.17708425521850585, "step": 10540 }, { "epoch": 0.2779030829461678, "grad_norm": 3.673635244369507, "learning_rate": 1.694693827582354e-05, "loss": 0.1768030643463135, "step": 10550 }, { "epoch": 0.2781664981906665, "grad_norm": 2.563678741455078, "learning_rate": 1.694079908468344e-05, "loss": 0.15974587202072144, "step": 10560 }, { "epoch": 0.27842991343516527, "grad_norm": 2.0756354331970215, "learning_rate": 1.6934654841730318e-05, "loss": 0.17032742500305176, "step": 10570 }, { "epoch": 0.278693328679664, "grad_norm": 1.5182968378067017, "learning_rate": 1.6928505551436215e-05, "loss": 0.17414435148239135, "step": 10580 }, { "epoch": 0.27895674392416275, "grad_norm": 2.44348406791687, "learning_rate": 1.6922351218276852e-05, "loss": 0.1669877886772156, "step": 10590 }, { "epoch": 0.2792201591686615, "grad_norm": 3.799182415008545, "learning_rate": 1.6916191846731615e-05, "loss": 0.15896365642547608, "step": 10600 }, { "epoch": 0.27948357441316024, "grad_norm": 2.79146146774292, "learning_rate": 1.6910027441283562e-05, "loss": 0.16927802562713623, "step": 10610 }, { "epoch": 0.279746989657659, "grad_norm": 1.7954540252685547, "learning_rate": 1.6903858006419413e-05, "loss": 0.17887290716171264, "step": 10620 }, { "epoch": 0.2800104049021577, "grad_norm": 1.7590101957321167, "learning_rate": 1.689768354662955e-05, "loss": 0.16316648721694946, "step": 10630 }, { "epoch": 0.2802738201466564, "grad_norm": 2.71625018119812, "learning_rate": 1.689150406640801e-05, "loss": 0.1665618896484375, "step": 10640 }, { "epoch": 0.28053723539115516, "grad_norm": 1.4049134254455566, "learning_rate": 1.6885319570252485e-05, "loss": 0.1657789707183838, "step": 10650 }, { "epoch": 0.2808006506356539, "grad_norm": 3.0362610816955566, "learning_rate": 1.687913006266432e-05, "loss": 0.16010285615921022, "step": 10660 }, { "epoch": 0.28106406588015265, "grad_norm": 2.1369919776916504, "learning_rate": 1.6872935548148503e-05, "loss": 0.17048206329345703, "step": 10670 }, { "epoch": 0.2813274811246514, "grad_norm": 3.6855506896972656, "learning_rate": 1.686673603121367e-05, "loss": 0.16412019729614258, "step": 10680 }, { "epoch": 0.28159089636915013, "grad_norm": 3.8612983226776123, "learning_rate": 1.6860531516372102e-05, "loss": 0.1761609435081482, "step": 10690 }, { "epoch": 0.2818543116136489, "grad_norm": 1.559732437133789, "learning_rate": 1.6854322008139704e-05, "loss": 0.16393356323242186, "step": 10700 }, { "epoch": 0.2821177268581476, "grad_norm": 0.8180396556854248, "learning_rate": 1.6848107511036026e-05, "loss": 0.16722190380096436, "step": 10710 }, { "epoch": 0.2823811421026463, "grad_norm": 1.339619755744934, "learning_rate": 1.6841888029584255e-05, "loss": 0.15153220891952515, "step": 10720 }, { "epoch": 0.28264455734714505, "grad_norm": 1.3027136325836182, "learning_rate": 1.6835663568311187e-05, "loss": 0.1520042061805725, "step": 10730 }, { "epoch": 0.2829079725916438, "grad_norm": 1.8169755935668945, "learning_rate": 1.682943413174726e-05, "loss": 0.16777443885803223, "step": 10740 }, { "epoch": 0.28317138783614254, "grad_norm": 1.4904578924179077, "learning_rate": 1.6823199724426523e-05, "loss": 0.17828013896942138, "step": 10750 }, { "epoch": 0.2834348030806413, "grad_norm": 1.0392476320266724, "learning_rate": 1.6816960350886653e-05, "loss": 0.16877200603485107, "step": 10760 }, { "epoch": 0.28369821832514003, "grad_norm": 3.439833641052246, "learning_rate": 1.681071601566893e-05, "loss": 0.17150074243545532, "step": 10770 }, { "epoch": 0.28396163356963877, "grad_norm": 1.1540278196334839, "learning_rate": 1.6804466723318248e-05, "loss": 0.1744608163833618, "step": 10780 }, { "epoch": 0.2842250488141375, "grad_norm": 3.3582961559295654, "learning_rate": 1.6798212478383118e-05, "loss": 0.16593236923217775, "step": 10790 }, { "epoch": 0.28448846405863626, "grad_norm": 2.4194812774658203, "learning_rate": 1.6791953285415645e-05, "loss": 0.15987823009490967, "step": 10800 }, { "epoch": 0.28475187930313495, "grad_norm": 1.859545350074768, "learning_rate": 1.678568914897154e-05, "loss": 0.1654893636703491, "step": 10810 }, { "epoch": 0.2850152945476337, "grad_norm": 5.322012424468994, "learning_rate": 1.6779420073610116e-05, "loss": 0.1605435013771057, "step": 10820 }, { "epoch": 0.28527870979213243, "grad_norm": 5.7419915199279785, "learning_rate": 1.6773146063894273e-05, "loss": 0.18907468318939208, "step": 10830 }, { "epoch": 0.2855421250366312, "grad_norm": 4.820532321929932, "learning_rate": 1.6766867124390505e-05, "loss": 0.1668742537498474, "step": 10840 }, { "epoch": 0.2858055402811299, "grad_norm": 3.2031331062316895, "learning_rate": 1.6760583259668897e-05, "loss": 0.16596896648406984, "step": 10850 }, { "epoch": 0.28606895552562867, "grad_norm": 1.1901782751083374, "learning_rate": 1.6754294474303116e-05, "loss": 0.16645950078964233, "step": 10860 }, { "epoch": 0.2863323707701274, "grad_norm": 0.8006712794303894, "learning_rate": 1.674800077287041e-05, "loss": 0.16930205821990968, "step": 10870 }, { "epoch": 0.28659578601462615, "grad_norm": 2.363497734069824, "learning_rate": 1.6741702159951613e-05, "loss": 0.17185215950012206, "step": 10880 }, { "epoch": 0.2868592012591249, "grad_norm": 1.816776156425476, "learning_rate": 1.6735398640131117e-05, "loss": 0.16824926137924195, "step": 10890 }, { "epoch": 0.2871226165036236, "grad_norm": 1.6242551803588867, "learning_rate": 1.6729090217996904e-05, "loss": 0.17429921627044678, "step": 10900 }, { "epoch": 0.28738603174812233, "grad_norm": 1.8999963998794556, "learning_rate": 1.6722776898140513e-05, "loss": 0.1635087251663208, "step": 10910 }, { "epoch": 0.28764944699262107, "grad_norm": 5.588191509246826, "learning_rate": 1.6716458685157047e-05, "loss": 0.16625847816467285, "step": 10920 }, { "epoch": 0.2879128622371198, "grad_norm": 1.9257102012634277, "learning_rate": 1.671013558364518e-05, "loss": 0.16400996446609498, "step": 10930 }, { "epoch": 0.28817627748161856, "grad_norm": 6.210118293762207, "learning_rate": 1.6703807598207132e-05, "loss": 0.16818206310272216, "step": 10940 }, { "epoch": 0.2884396927261173, "grad_norm": 1.8685710430145264, "learning_rate": 1.669747473344869e-05, "loss": 0.15555001497268678, "step": 10950 }, { "epoch": 0.28870310797061605, "grad_norm": 2.776845693588257, "learning_rate": 1.6691136993979183e-05, "loss": 0.1470228314399719, "step": 10960 }, { "epoch": 0.2889665232151148, "grad_norm": 2.1199870109558105, "learning_rate": 1.6684794384411494e-05, "loss": 0.15825462341308594, "step": 10970 }, { "epoch": 0.2892299384596135, "grad_norm": 1.331329584121704, "learning_rate": 1.667844690936204e-05, "loss": 0.16797292232513428, "step": 10980 }, { "epoch": 0.2894933537041122, "grad_norm": 1.659630537033081, "learning_rate": 1.6672094573450798e-05, "loss": 0.17908960580825806, "step": 10990 }, { "epoch": 0.28975676894861097, "grad_norm": 2.06862473487854, "learning_rate": 1.6665737381301263e-05, "loss": 0.1711701989173889, "step": 11000 }, { "epoch": 0.2900201841931097, "grad_norm": 1.2005459070205688, "learning_rate": 1.665937533754048e-05, "loss": 0.15768969058990479, "step": 11010 }, { "epoch": 0.29028359943760845, "grad_norm": 1.2409340143203735, "learning_rate": 1.6653008446799014e-05, "loss": 0.17093360424041748, "step": 11020 }, { "epoch": 0.2905470146821072, "grad_norm": 0.8020076751708984, "learning_rate": 1.6646636713710967e-05, "loss": 0.16246814727783204, "step": 11030 }, { "epoch": 0.29081042992660594, "grad_norm": 2.5145821571350098, "learning_rate": 1.664026014291396e-05, "loss": 0.17655856609344484, "step": 11040 }, { "epoch": 0.2910738451711047, "grad_norm": 3.7938034534454346, "learning_rate": 1.6633878739049135e-05, "loss": 0.1655851721763611, "step": 11050 }, { "epoch": 0.2913372604156034, "grad_norm": 2.930690288543701, "learning_rate": 1.6627492506761155e-05, "loss": 0.17132968902587892, "step": 11060 }, { "epoch": 0.2916006756601021, "grad_norm": 1.4207184314727783, "learning_rate": 1.6621101450698193e-05, "loss": 0.1765220046043396, "step": 11070 }, { "epoch": 0.29186409090460086, "grad_norm": 2.5115535259246826, "learning_rate": 1.6614705575511937e-05, "loss": 0.17428739070892335, "step": 11080 }, { "epoch": 0.2921275061490996, "grad_norm": 3.210080146789551, "learning_rate": 1.660830488585758e-05, "loss": 0.15946345329284667, "step": 11090 }, { "epoch": 0.29239092139359835, "grad_norm": 2.396756410598755, "learning_rate": 1.660189938639382e-05, "loss": 0.1574809432029724, "step": 11100 }, { "epoch": 0.2926543366380971, "grad_norm": 1.0108145475387573, "learning_rate": 1.6595489081782858e-05, "loss": 0.16758954524993896, "step": 11110 }, { "epoch": 0.29291775188259583, "grad_norm": 1.0475053787231445, "learning_rate": 1.6589073976690387e-05, "loss": 0.17478532791137696, "step": 11120 }, { "epoch": 0.2931811671270946, "grad_norm": 2.9519259929656982, "learning_rate": 1.65826540757856e-05, "loss": 0.16305146217346192, "step": 11130 }, { "epoch": 0.2934445823715933, "grad_norm": 1.8846590518951416, "learning_rate": 1.6576229383741177e-05, "loss": 0.17024316787719726, "step": 11140 }, { "epoch": 0.29370799761609206, "grad_norm": 1.8895740509033203, "learning_rate": 1.6569799905233285e-05, "loss": 0.16390528678894042, "step": 11150 }, { "epoch": 0.29397141286059075, "grad_norm": 1.8632429838180542, "learning_rate": 1.6563365644941578e-05, "loss": 0.16520599126815796, "step": 11160 }, { "epoch": 0.2942348281050895, "grad_norm": 3.2854316234588623, "learning_rate": 1.6556926607549186e-05, "loss": 0.15674692392349243, "step": 11170 }, { "epoch": 0.29449824334958824, "grad_norm": 2.737569570541382, "learning_rate": 1.655048279774272e-05, "loss": 0.1726733684539795, "step": 11180 }, { "epoch": 0.294761658594087, "grad_norm": 3.2571802139282227, "learning_rate": 1.6544034220212262e-05, "loss": 0.1628953218460083, "step": 11190 }, { "epoch": 0.2950250738385857, "grad_norm": 1.2649500370025635, "learning_rate": 1.6537580879651364e-05, "loss": 0.15725045204162597, "step": 11200 }, { "epoch": 0.29528848908308447, "grad_norm": 1.6368577480316162, "learning_rate": 1.6531122780757046e-05, "loss": 0.1652848720550537, "step": 11210 }, { "epoch": 0.2955519043275832, "grad_norm": 4.036079406738281, "learning_rate": 1.6524659928229792e-05, "loss": 0.172975754737854, "step": 11220 }, { "epoch": 0.29581531957208196, "grad_norm": 5.236570358276367, "learning_rate": 1.6518192326773546e-05, "loss": 0.17746150493621826, "step": 11230 }, { "epoch": 0.29607873481658065, "grad_norm": 2.556795597076416, "learning_rate": 1.6511719981095703e-05, "loss": 0.17682780027389527, "step": 11240 }, { "epoch": 0.2963421500610794, "grad_norm": 3.5474283695220947, "learning_rate": 1.6505242895907122e-05, "loss": 0.1669124484062195, "step": 11250 }, { "epoch": 0.29660556530557813, "grad_norm": 2.484407901763916, "learning_rate": 1.64987610759221e-05, "loss": 0.16871249675750732, "step": 11260 }, { "epoch": 0.2968689805500769, "grad_norm": 1.8198829889297485, "learning_rate": 1.6492274525858386e-05, "loss": 0.16701982021331788, "step": 11270 }, { "epoch": 0.2971323957945756, "grad_norm": 1.3986209630966187, "learning_rate": 1.648578325043717e-05, "loss": 0.16403663158416748, "step": 11280 }, { "epoch": 0.29739581103907436, "grad_norm": 0.6029418110847473, "learning_rate": 1.6479287254383084e-05, "loss": 0.17614636421203614, "step": 11290 }, { "epoch": 0.2976592262835731, "grad_norm": 1.3963044881820679, "learning_rate": 1.6472786542424196e-05, "loss": 0.151849102973938, "step": 11300 }, { "epoch": 0.29792264152807185, "grad_norm": 3.408853054046631, "learning_rate": 1.6466281119292004e-05, "loss": 0.17257609367370605, "step": 11310 }, { "epoch": 0.2981860567725706, "grad_norm": 1.5788133144378662, "learning_rate": 1.6459770989721433e-05, "loss": 0.17445533275604247, "step": 11320 }, { "epoch": 0.2984494720170693, "grad_norm": 3.1210248470306396, "learning_rate": 1.6453256158450835e-05, "loss": 0.16449990272521972, "step": 11330 }, { "epoch": 0.298712887261568, "grad_norm": 1.3135395050048828, "learning_rate": 1.644673663022199e-05, "loss": 0.16964504718780518, "step": 11340 }, { "epoch": 0.29897630250606677, "grad_norm": 2.8665809631347656, "learning_rate": 1.6440212409780088e-05, "loss": 0.16665889024734498, "step": 11350 }, { "epoch": 0.2992397177505655, "grad_norm": 1.0680831670761108, "learning_rate": 1.643368350187374e-05, "loss": 0.1776324987411499, "step": 11360 }, { "epoch": 0.29950313299506426, "grad_norm": 2.850450038909912, "learning_rate": 1.6427149911254973e-05, "loss": 0.16272256374359131, "step": 11370 }, { "epoch": 0.299766548239563, "grad_norm": 1.7019052505493164, "learning_rate": 1.64206116426792e-05, "loss": 0.15396790504455565, "step": 11380 }, { "epoch": 0.30002996348406175, "grad_norm": 1.124341607093811, "learning_rate": 1.641406870090527e-05, "loss": 0.15562903881072998, "step": 11390 }, { "epoch": 0.3002933787285605, "grad_norm": 1.6657263040542603, "learning_rate": 1.640752109069541e-05, "loss": 0.16325914859771729, "step": 11400 }, { "epoch": 0.30055679397305923, "grad_norm": 0.8648161292076111, "learning_rate": 1.640096881681526e-05, "loss": 0.1685037851333618, "step": 11410 }, { "epoch": 0.3008202092175579, "grad_norm": 2.3382558822631836, "learning_rate": 1.6394411884033844e-05, "loss": 0.1744473934173584, "step": 11420 }, { "epoch": 0.30108362446205666, "grad_norm": 0.987130880355835, "learning_rate": 1.638785029712358e-05, "loss": 0.16195476055145264, "step": 11430 }, { "epoch": 0.3013470397065554, "grad_norm": 2.4652490615844727, "learning_rate": 1.638128406086027e-05, "loss": 0.17083916664123536, "step": 11440 }, { "epoch": 0.30161045495105415, "grad_norm": 1.130215048789978, "learning_rate": 1.6374713180023114e-05, "loss": 0.16074044704437257, "step": 11450 }, { "epoch": 0.3018738701955529, "grad_norm": 1.0418635606765747, "learning_rate": 1.6368137659394673e-05, "loss": 0.16950345039367676, "step": 11460 }, { "epoch": 0.30213728544005164, "grad_norm": 2.8226728439331055, "learning_rate": 1.63615575037609e-05, "loss": 0.18549768924713134, "step": 11470 }, { "epoch": 0.3024007006845504, "grad_norm": 4.28086519241333, "learning_rate": 1.6354972717911115e-05, "loss": 0.14237611293792723, "step": 11480 }, { "epoch": 0.3026641159290491, "grad_norm": 2.9390690326690674, "learning_rate": 1.634838330663801e-05, "loss": 0.16435774564743041, "step": 11490 }, { "epoch": 0.3029275311735478, "grad_norm": 1.8204584121704102, "learning_rate": 1.6341789274737644e-05, "loss": 0.16708576679229736, "step": 11500 }, { "epoch": 0.30319094641804656, "grad_norm": 2.497471809387207, "learning_rate": 1.6335190627009432e-05, "loss": 0.17515612840652467, "step": 11510 }, { "epoch": 0.3034543616625453, "grad_norm": 4.9359002113342285, "learning_rate": 1.632858736825616e-05, "loss": 0.1653974771499634, "step": 11520 }, { "epoch": 0.30371777690704405, "grad_norm": 1.6993952989578247, "learning_rate": 1.632197950328396e-05, "loss": 0.15518550872802733, "step": 11530 }, { "epoch": 0.3039811921515428, "grad_norm": 2.490797996520996, "learning_rate": 1.631536703690233e-05, "loss": 0.16186363697052003, "step": 11540 }, { "epoch": 0.30424460739604153, "grad_norm": 1.7758580446243286, "learning_rate": 1.6308749973924096e-05, "loss": 0.16590945720672606, "step": 11550 }, { "epoch": 0.3045080226405403, "grad_norm": 1.6008228063583374, "learning_rate": 1.6302128319165454e-05, "loss": 0.1823681116104126, "step": 11560 }, { "epoch": 0.304771437885039, "grad_norm": 2.0452115535736084, "learning_rate": 1.629550207744592e-05, "loss": 0.17513371706008912, "step": 11570 }, { "epoch": 0.30503485312953776, "grad_norm": 1.1745566129684448, "learning_rate": 1.6288871253588366e-05, "loss": 0.1733898401260376, "step": 11580 }, { "epoch": 0.30529826837403645, "grad_norm": 3.6585302352905273, "learning_rate": 1.6282235852418983e-05, "loss": 0.17676918506622313, "step": 11590 }, { "epoch": 0.3055616836185352, "grad_norm": 1.6300427913665771, "learning_rate": 1.6275595878767315e-05, "loss": 0.16325322389602662, "step": 11600 }, { "epoch": 0.30582509886303394, "grad_norm": 1.1414796113967896, "learning_rate": 1.6268951337466212e-05, "loss": 0.1597089409828186, "step": 11610 }, { "epoch": 0.3060885141075327, "grad_norm": 1.4206650257110596, "learning_rate": 1.626230223335186e-05, "loss": 0.17675745487213135, "step": 11620 }, { "epoch": 0.3063519293520314, "grad_norm": 1.0690362453460693, "learning_rate": 1.6255648571263762e-05, "loss": 0.16793133020401002, "step": 11630 }, { "epoch": 0.30661534459653017, "grad_norm": 3.0551340579986572, "learning_rate": 1.6248990356044742e-05, "loss": 0.17352118492126464, "step": 11640 }, { "epoch": 0.3068787598410289, "grad_norm": 1.9039993286132812, "learning_rate": 1.6242327592540942e-05, "loss": 0.15912392139434814, "step": 11650 }, { "epoch": 0.30714217508552766, "grad_norm": 2.0017707347869873, "learning_rate": 1.62356602856018e-05, "loss": 0.16824097633361818, "step": 11660 }, { "epoch": 0.3074055903300264, "grad_norm": 1.6546647548675537, "learning_rate": 1.6228988440080068e-05, "loss": 0.16062049865722655, "step": 11670 }, { "epoch": 0.3076690055745251, "grad_norm": 3.221575975418091, "learning_rate": 1.6222312060831814e-05, "loss": 0.15515775680541993, "step": 11680 }, { "epoch": 0.30793242081902383, "grad_norm": 3.144634485244751, "learning_rate": 1.6215631152716387e-05, "loss": 0.17091705799102783, "step": 11690 }, { "epoch": 0.3081958360635226, "grad_norm": 1.8530101776123047, "learning_rate": 1.620894572059644e-05, "loss": 0.17098743915557862, "step": 11700 }, { "epoch": 0.3084592513080213, "grad_norm": 1.4347213506698608, "learning_rate": 1.6202255769337924e-05, "loss": 0.15205006599426268, "step": 11710 }, { "epoch": 0.30872266655252006, "grad_norm": 1.466942310333252, "learning_rate": 1.619556130381007e-05, "loss": 0.15827696323394774, "step": 11720 }, { "epoch": 0.3089860817970188, "grad_norm": 1.7039711475372314, "learning_rate": 1.61888623288854e-05, "loss": 0.17623426914215087, "step": 11730 }, { "epoch": 0.30924949704151755, "grad_norm": 2.5133533477783203, "learning_rate": 1.6182158849439718e-05, "loss": 0.157742178440094, "step": 11740 }, { "epoch": 0.3095129122860163, "grad_norm": 1.637947916984558, "learning_rate": 1.61754508703521e-05, "loss": 0.15952446460723876, "step": 11750 }, { "epoch": 0.309776327530515, "grad_norm": 2.215831995010376, "learning_rate": 1.6168738396504912e-05, "loss": 0.17531020641326905, "step": 11760 }, { "epoch": 0.3100397427750137, "grad_norm": 3.5027079582214355, "learning_rate": 1.6162021432783773e-05, "loss": 0.16849623918533324, "step": 11770 }, { "epoch": 0.31030315801951247, "grad_norm": 2.6626877784729004, "learning_rate": 1.6155299984077584e-05, "loss": 0.16416921615600585, "step": 11780 }, { "epoch": 0.3105665732640112, "grad_norm": 1.0923207998275757, "learning_rate": 1.6148574055278507e-05, "loss": 0.14240487813949584, "step": 11790 }, { "epoch": 0.31082998850850996, "grad_norm": 1.3902711868286133, "learning_rate": 1.6141843651281955e-05, "loss": 0.16195428371429443, "step": 11800 }, { "epoch": 0.3110934037530087, "grad_norm": 2.387554883956909, "learning_rate": 1.6135108776986615e-05, "loss": 0.1731163501739502, "step": 11810 }, { "epoch": 0.31135681899750745, "grad_norm": 0.7760036587715149, "learning_rate": 1.612836943729442e-05, "loss": 0.14770042896270752, "step": 11820 }, { "epoch": 0.3116202342420062, "grad_norm": 1.5641987323760986, "learning_rate": 1.6121625637110546e-05, "loss": 0.1710381865501404, "step": 11830 }, { "epoch": 0.31188364948650493, "grad_norm": 2.2985777854919434, "learning_rate": 1.6114877381343422e-05, "loss": 0.1627763032913208, "step": 11840 }, { "epoch": 0.3121470647310036, "grad_norm": 1.0705338716506958, "learning_rate": 1.610812467490472e-05, "loss": 0.1776867389678955, "step": 11850 }, { "epoch": 0.31241047997550236, "grad_norm": 2.56355357170105, "learning_rate": 1.6101367522709356e-05, "loss": 0.15788700580596923, "step": 11860 }, { "epoch": 0.3126738952200011, "grad_norm": 2.139309883117676, "learning_rate": 1.6094605929675476e-05, "loss": 0.15457645654678345, "step": 11870 }, { "epoch": 0.31293731046449985, "grad_norm": 1.4678841829299927, "learning_rate": 1.608783990072445e-05, "loss": 0.16344494819641114, "step": 11880 }, { "epoch": 0.3132007257089986, "grad_norm": 4.133192539215088, "learning_rate": 1.6081069440780895e-05, "loss": 0.1659551739692688, "step": 11890 }, { "epoch": 0.31346414095349734, "grad_norm": 2.315136194229126, "learning_rate": 1.607429455477264e-05, "loss": 0.15172414779663085, "step": 11900 }, { "epoch": 0.3137275561979961, "grad_norm": 1.2632192373275757, "learning_rate": 1.606751524763075e-05, "loss": 0.15311622619628906, "step": 11910 }, { "epoch": 0.3139909714424948, "grad_norm": 1.1550730466842651, "learning_rate": 1.606073152428948e-05, "loss": 0.16057791709899902, "step": 11920 }, { "epoch": 0.31425438668699357, "grad_norm": 2.113792896270752, "learning_rate": 1.6053943389686323e-05, "loss": 0.16079840660095215, "step": 11930 }, { "epoch": 0.31451780193149226, "grad_norm": 1.7749292850494385, "learning_rate": 1.604715084876198e-05, "loss": 0.1591684937477112, "step": 11940 }, { "epoch": 0.314781217175991, "grad_norm": 1.582933783531189, "learning_rate": 1.6040353906460353e-05, "loss": 0.17450454235076904, "step": 11950 }, { "epoch": 0.31504463242048975, "grad_norm": 1.7867708206176758, "learning_rate": 1.6033552567728553e-05, "loss": 0.15277966260910034, "step": 11960 }, { "epoch": 0.3153080476649885, "grad_norm": 1.744115948677063, "learning_rate": 1.602674683751688e-05, "loss": 0.16870300769805907, "step": 11970 }, { "epoch": 0.31557146290948723, "grad_norm": 1.477709174156189, "learning_rate": 1.6019936720778845e-05, "loss": 0.16802330017089845, "step": 11980 }, { "epoch": 0.315834878153986, "grad_norm": 1.7514704465866089, "learning_rate": 1.601312222247114e-05, "loss": 0.16253381967544556, "step": 11990 }, { "epoch": 0.3160982933984847, "grad_norm": 2.9821584224700928, "learning_rate": 1.6006303347553655e-05, "loss": 0.17100119590759277, "step": 12000 }, { "epoch": 0.31636170864298346, "grad_norm": 1.3010458946228027, "learning_rate": 1.5999480100989458e-05, "loss": 0.1697869658470154, "step": 12010 }, { "epoch": 0.31662512388748215, "grad_norm": 1.9456349611282349, "learning_rate": 1.5992652487744806e-05, "loss": 0.17075555324554442, "step": 12020 }, { "epoch": 0.3168885391319809, "grad_norm": 1.1838383674621582, "learning_rate": 1.5985820512789128e-05, "loss": 0.15731865167617798, "step": 12030 }, { "epoch": 0.31715195437647964, "grad_norm": 1.877752661705017, "learning_rate": 1.5978984181095034e-05, "loss": 0.16972534656524657, "step": 12040 }, { "epoch": 0.3174153696209784, "grad_norm": 1.6617549657821655, "learning_rate": 1.5972143497638294e-05, "loss": 0.1668969750404358, "step": 12050 }, { "epoch": 0.3176787848654771, "grad_norm": 2.0259597301483154, "learning_rate": 1.5965298467397855e-05, "loss": 0.1587581753730774, "step": 12060 }, { "epoch": 0.31794220010997587, "grad_norm": 1.0706363916397095, "learning_rate": 1.595844909535583e-05, "loss": 0.16039059162139893, "step": 12070 }, { "epoch": 0.3182056153544746, "grad_norm": 1.1597133874893188, "learning_rate": 1.595159538649748e-05, "loss": 0.17027621269226073, "step": 12080 }, { "epoch": 0.31846903059897336, "grad_norm": 3.5902516841888428, "learning_rate": 1.5944737345811238e-05, "loss": 0.17017631530761718, "step": 12090 }, { "epoch": 0.3187324458434721, "grad_norm": 0.7048720717430115, "learning_rate": 1.5937874978288676e-05, "loss": 0.1548607349395752, "step": 12100 }, { "epoch": 0.3189958610879708, "grad_norm": 1.6134787797927856, "learning_rate": 1.5931008288924523e-05, "loss": 0.16250534057617189, "step": 12110 }, { "epoch": 0.31925927633246953, "grad_norm": 0.9561542272567749, "learning_rate": 1.5924137282716653e-05, "loss": 0.16167998313903809, "step": 12120 }, { "epoch": 0.3195226915769683, "grad_norm": 1.5887871980667114, "learning_rate": 1.591726196466608e-05, "loss": 0.17217962741851806, "step": 12130 }, { "epoch": 0.319786106821467, "grad_norm": 0.9081801176071167, "learning_rate": 1.5910382339776953e-05, "loss": 0.16163625717163085, "step": 12140 }, { "epoch": 0.32004952206596576, "grad_norm": 3.2551591396331787, "learning_rate": 1.5903498413056568e-05, "loss": 0.16830387115478515, "step": 12150 }, { "epoch": 0.3203129373104645, "grad_norm": 1.5970406532287598, "learning_rate": 1.5896610189515335e-05, "loss": 0.16034440994262694, "step": 12160 }, { "epoch": 0.32057635255496325, "grad_norm": 1.5852446556091309, "learning_rate": 1.588971767416681e-05, "loss": 0.16747443675994872, "step": 12170 }, { "epoch": 0.320839767799462, "grad_norm": 4.376384258270264, "learning_rate": 1.5882820872027654e-05, "loss": 0.16053156852722167, "step": 12180 }, { "epoch": 0.32110318304396074, "grad_norm": 1.2627575397491455, "learning_rate": 1.5875919788117662e-05, "loss": 0.16815330982208251, "step": 12190 }, { "epoch": 0.3213665982884594, "grad_norm": 0.8394989967346191, "learning_rate": 1.586901442745974e-05, "loss": 0.16623694896698, "step": 12200 }, { "epoch": 0.32163001353295817, "grad_norm": 2.1672866344451904, "learning_rate": 1.5862104795079903e-05, "loss": 0.1500806450843811, "step": 12210 }, { "epoch": 0.3218934287774569, "grad_norm": 1.270696997642517, "learning_rate": 1.5855190896007287e-05, "loss": 0.18245718479156495, "step": 12220 }, { "epoch": 0.32215684402195566, "grad_norm": 0.6335862874984741, "learning_rate": 1.584827273527412e-05, "loss": 0.1484852910041809, "step": 12230 }, { "epoch": 0.3224202592664544, "grad_norm": 3.430446147918701, "learning_rate": 1.5841350317915745e-05, "loss": 0.16364083290100098, "step": 12240 }, { "epoch": 0.32268367451095314, "grad_norm": 1.3682293891906738, "learning_rate": 1.5834423648970588e-05, "loss": 0.1692550539970398, "step": 12250 }, { "epoch": 0.3229470897554519, "grad_norm": 3.2113330364227295, "learning_rate": 1.5827492733480183e-05, "loss": 0.1730958938598633, "step": 12260 }, { "epoch": 0.32321050499995063, "grad_norm": 2.291295289993286, "learning_rate": 1.5820557576489145e-05, "loss": 0.1499802589416504, "step": 12270 }, { "epoch": 0.3234739202444493, "grad_norm": 1.013949990272522, "learning_rate": 1.581361818304518e-05, "loss": 0.15953283309936522, "step": 12280 }, { "epoch": 0.32373733548894806, "grad_norm": 2.3317532539367676, "learning_rate": 1.5806674558199088e-05, "loss": 0.17801681756973267, "step": 12290 }, { "epoch": 0.3240007507334468, "grad_norm": 0.7801991105079651, "learning_rate": 1.5799726707004728e-05, "loss": 0.17480344772338868, "step": 12300 }, { "epoch": 0.32426416597794555, "grad_norm": 1.1115602254867554, "learning_rate": 1.5792774634519046e-05, "loss": 0.17321211099624634, "step": 12310 }, { "epoch": 0.3245275812224443, "grad_norm": 2.7116270065307617, "learning_rate": 1.5785818345802068e-05, "loss": 0.1607472538948059, "step": 12320 }, { "epoch": 0.32479099646694304, "grad_norm": 1.1925809383392334, "learning_rate": 1.5778857845916874e-05, "loss": 0.16188032627105714, "step": 12330 }, { "epoch": 0.3250544117114418, "grad_norm": 1.6102639436721802, "learning_rate": 1.577189313992962e-05, "loss": 0.1609344244003296, "step": 12340 }, { "epoch": 0.3253178269559405, "grad_norm": 1.1843045949935913, "learning_rate": 1.5764924232909518e-05, "loss": 0.1771610975265503, "step": 12350 }, { "epoch": 0.32558124220043927, "grad_norm": 1.7483949661254883, "learning_rate": 1.5757951129928837e-05, "loss": 0.16500872373580933, "step": 12360 }, { "epoch": 0.32584465744493796, "grad_norm": 1.0358778238296509, "learning_rate": 1.5750973836062907e-05, "loss": 0.1554909586906433, "step": 12370 }, { "epoch": 0.3261080726894367, "grad_norm": 2.4128024578094482, "learning_rate": 1.57439923563901e-05, "loss": 0.17331202030181886, "step": 12380 }, { "epoch": 0.32637148793393544, "grad_norm": 1.5040613412857056, "learning_rate": 1.573700669599184e-05, "loss": 0.16054275035858154, "step": 12390 }, { "epoch": 0.3266349031784342, "grad_norm": 1.5733978748321533, "learning_rate": 1.573001685995259e-05, "loss": 0.1530306339263916, "step": 12400 }, { "epoch": 0.32689831842293293, "grad_norm": 3.041506767272949, "learning_rate": 1.572302285335985e-05, "loss": 0.17457146644592286, "step": 12410 }, { "epoch": 0.3271617336674317, "grad_norm": 1.916473150253296, "learning_rate": 1.5716024681304177e-05, "loss": 0.15922027826309204, "step": 12420 }, { "epoch": 0.3274251489119304, "grad_norm": 3.5456485748291016, "learning_rate": 1.5709022348879124e-05, "loss": 0.16094062328338624, "step": 12430 }, { "epoch": 0.32768856415642916, "grad_norm": 1.5280117988586426, "learning_rate": 1.5702015861181295e-05, "loss": 0.15914995670318605, "step": 12440 }, { "epoch": 0.3279519794009279, "grad_norm": 3.9758334159851074, "learning_rate": 1.569500522331032e-05, "loss": 0.16459481716156005, "step": 12450 }, { "epoch": 0.3282153946454266, "grad_norm": 4.795502185821533, "learning_rate": 1.568799044036884e-05, "loss": 0.16111247539520263, "step": 12460 }, { "epoch": 0.32847880988992534, "grad_norm": 1.4023115634918213, "learning_rate": 1.5680971517462516e-05, "loss": 0.14674601554870606, "step": 12470 }, { "epoch": 0.3287422251344241, "grad_norm": 3.7419052124023438, "learning_rate": 1.5673948459700024e-05, "loss": 0.14484684467315673, "step": 12480 }, { "epoch": 0.3290056403789228, "grad_norm": 1.986222267150879, "learning_rate": 1.5666921272193046e-05, "loss": 0.15754536390304566, "step": 12490 }, { "epoch": 0.32926905562342157, "grad_norm": 2.031851053237915, "learning_rate": 1.565988996005628e-05, "loss": 0.16663858890533448, "step": 12500 }, { "epoch": 0.3295324708679203, "grad_norm": 1.59164559841156, "learning_rate": 1.5652854528407407e-05, "loss": 0.15937893390655516, "step": 12510 }, { "epoch": 0.32979588611241906, "grad_norm": 1.7793153524398804, "learning_rate": 1.5645814982367125e-05, "loss": 0.1641155242919922, "step": 12520 }, { "epoch": 0.3300593013569178, "grad_norm": 0.978807806968689, "learning_rate": 1.5638771327059117e-05, "loss": 0.16813082695007325, "step": 12530 }, { "epoch": 0.3303227166014165, "grad_norm": 2.89106822013855, "learning_rate": 1.563172356761006e-05, "loss": 0.1647047758102417, "step": 12540 }, { "epoch": 0.33058613184591523, "grad_norm": 1.2329272031784058, "learning_rate": 1.5624671709149618e-05, "loss": 0.14967553615570067, "step": 12550 }, { "epoch": 0.330849547090414, "grad_norm": 2.3704752922058105, "learning_rate": 1.561761575681044e-05, "loss": 0.1506453275680542, "step": 12560 }, { "epoch": 0.3311129623349127, "grad_norm": 2.0915417671203613, "learning_rate": 1.5610555715728146e-05, "loss": 0.1519816994667053, "step": 12570 }, { "epoch": 0.33137637757941146, "grad_norm": 2.0226426124572754, "learning_rate": 1.560349159104134e-05, "loss": 0.16372265815734863, "step": 12580 }, { "epoch": 0.3316397928239102, "grad_norm": 3.076500177383423, "learning_rate": 1.55964233878916e-05, "loss": 0.1591806411743164, "step": 12590 }, { "epoch": 0.33190320806840895, "grad_norm": 0.7007569670677185, "learning_rate": 1.5589351111423474e-05, "loss": 0.15369076728820802, "step": 12600 }, { "epoch": 0.3321666233129077, "grad_norm": 2.268784523010254, "learning_rate": 1.5582274766784457e-05, "loss": 0.17538976669311523, "step": 12610 }, { "epoch": 0.33243003855740644, "grad_norm": 1.9468588829040527, "learning_rate": 1.557519435912503e-05, "loss": 0.14952313899993896, "step": 12620 }, { "epoch": 0.3326934538019051, "grad_norm": 1.8621524572372437, "learning_rate": 1.556810989359861e-05, "loss": 0.16187973022460939, "step": 12630 }, { "epoch": 0.33295686904640387, "grad_norm": 2.86262583732605, "learning_rate": 1.5561021375361583e-05, "loss": 0.16644887924194335, "step": 12640 }, { "epoch": 0.3332202842909026, "grad_norm": 1.1067125797271729, "learning_rate": 1.5553928809573275e-05, "loss": 0.1608344316482544, "step": 12650 }, { "epoch": 0.33348369953540136, "grad_norm": 1.3339133262634277, "learning_rate": 1.5546832201395964e-05, "loss": 0.1570656418800354, "step": 12660 }, { "epoch": 0.3337471147799001, "grad_norm": 2.356454849243164, "learning_rate": 1.5539731555994864e-05, "loss": 0.1566924571990967, "step": 12670 }, { "epoch": 0.33401053002439884, "grad_norm": 1.1750408411026, "learning_rate": 1.5532626878538137e-05, "loss": 0.16225959062576295, "step": 12680 }, { "epoch": 0.3342739452688976, "grad_norm": 2.987916946411133, "learning_rate": 1.5525518174196867e-05, "loss": 0.1608634352684021, "step": 12690 }, { "epoch": 0.33453736051339633, "grad_norm": 1.753652572631836, "learning_rate": 1.551840544814508e-05, "loss": 0.14962459802627565, "step": 12700 }, { "epoch": 0.3348007757578951, "grad_norm": 1.061826467514038, "learning_rate": 1.551128870555972e-05, "loss": 0.16480599641799926, "step": 12710 }, { "epoch": 0.33506419100239376, "grad_norm": 1.8700740337371826, "learning_rate": 1.550416795162067e-05, "loss": 0.1650984525680542, "step": 12720 }, { "epoch": 0.3353276062468925, "grad_norm": 2.4280784130096436, "learning_rate": 1.549704319151071e-05, "loss": 0.16467635631561278, "step": 12730 }, { "epoch": 0.33559102149139125, "grad_norm": 0.8484900593757629, "learning_rate": 1.548991443041555e-05, "loss": 0.16005182266235352, "step": 12740 }, { "epoch": 0.33585443673589, "grad_norm": 2.789339065551758, "learning_rate": 1.5482781673523816e-05, "loss": 0.16615614891052247, "step": 12750 }, { "epoch": 0.33611785198038874, "grad_norm": 2.1268556118011475, "learning_rate": 1.5475644926027035e-05, "loss": 0.15029048919677734, "step": 12760 }, { "epoch": 0.3363812672248875, "grad_norm": 3.2354047298431396, "learning_rate": 1.5468504193119636e-05, "loss": 0.17754714488983153, "step": 12770 }, { "epoch": 0.3366446824693862, "grad_norm": 2.8607547283172607, "learning_rate": 1.5461359479998953e-05, "loss": 0.16064789295196533, "step": 12780 }, { "epoch": 0.33690809771388497, "grad_norm": 1.6103706359863281, "learning_rate": 1.5454210791865213e-05, "loss": 0.16285221576690673, "step": 12790 }, { "epoch": 0.33717151295838366, "grad_norm": 2.80947208404541, "learning_rate": 1.5447058133921552e-05, "loss": 0.17358009815216063, "step": 12800 }, { "epoch": 0.3374349282028824, "grad_norm": 1.847052812576294, "learning_rate": 1.543990151137397e-05, "loss": 0.16634509563446045, "step": 12810 }, { "epoch": 0.33769834344738114, "grad_norm": 1.5002447366714478, "learning_rate": 1.5432740929431372e-05, "loss": 0.1483629584312439, "step": 12820 }, { "epoch": 0.3379617586918799, "grad_norm": 2.3965811729431152, "learning_rate": 1.5425576393305537e-05, "loss": 0.1584348201751709, "step": 12830 }, { "epoch": 0.33822517393637863, "grad_norm": 1.2846416234970093, "learning_rate": 1.5418407908211125e-05, "loss": 0.1616022229194641, "step": 12840 }, { "epoch": 0.3384885891808774, "grad_norm": 2.3215558528900146, "learning_rate": 1.5411235479365664e-05, "loss": 0.1551844835281372, "step": 12850 }, { "epoch": 0.3387520044253761, "grad_norm": 1.6935652494430542, "learning_rate": 1.5404059111989565e-05, "loss": 0.16582695245742798, "step": 12860 }, { "epoch": 0.33901541966987486, "grad_norm": 2.9369685649871826, "learning_rate": 1.5396878811306088e-05, "loss": 0.16598910093307495, "step": 12870 }, { "epoch": 0.3392788349143736, "grad_norm": 1.2712229490280151, "learning_rate": 1.5389694582541376e-05, "loss": 0.16777031421661376, "step": 12880 }, { "epoch": 0.3395422501588723, "grad_norm": 0.841659665107727, "learning_rate": 1.5382506430924414e-05, "loss": 0.1599994421005249, "step": 12890 }, { "epoch": 0.33980566540337104, "grad_norm": 1.6790391206741333, "learning_rate": 1.5375314361687053e-05, "loss": 0.15987569093704224, "step": 12900 }, { "epoch": 0.3400690806478698, "grad_norm": 3.389899253845215, "learning_rate": 1.5368118380063984e-05, "loss": 0.16848618984222413, "step": 12910 }, { "epoch": 0.3403324958923685, "grad_norm": 3.010955333709717, "learning_rate": 1.536091849129276e-05, "loss": 0.16758220195770263, "step": 12920 }, { "epoch": 0.34059591113686727, "grad_norm": 1.1346596479415894, "learning_rate": 1.5353714700613768e-05, "loss": 0.16292734146118165, "step": 12930 }, { "epoch": 0.340859326381366, "grad_norm": 3.7672877311706543, "learning_rate": 1.5346507013270237e-05, "loss": 0.1536705255508423, "step": 12940 }, { "epoch": 0.34112274162586476, "grad_norm": 3.4688546657562256, "learning_rate": 1.5339295434508238e-05, "loss": 0.15294407606124877, "step": 12950 }, { "epoch": 0.3413861568703635, "grad_norm": 1.4511734247207642, "learning_rate": 1.533207996957666e-05, "loss": 0.1551366329193115, "step": 12960 }, { "epoch": 0.34164957211486224, "grad_norm": 1.8830393552780151, "learning_rate": 1.5324860623727236e-05, "loss": 0.16245791912078858, "step": 12970 }, { "epoch": 0.34191298735936093, "grad_norm": 2.493340015411377, "learning_rate": 1.5317637402214515e-05, "loss": 0.1609959840774536, "step": 12980 }, { "epoch": 0.3421764026038597, "grad_norm": 2.042330741882324, "learning_rate": 1.5310410310295876e-05, "loss": 0.15141602754592895, "step": 12990 }, { "epoch": 0.3424398178483584, "grad_norm": 2.2834219932556152, "learning_rate": 1.5303179353231494e-05, "loss": 0.16092402935028077, "step": 13000 }, { "epoch": 0.34270323309285716, "grad_norm": 3.473893404006958, "learning_rate": 1.529594453628439e-05, "loss": 0.15998351573944092, "step": 13010 }, { "epoch": 0.3429666483373559, "grad_norm": 2.057950735092163, "learning_rate": 1.5288705864720356e-05, "loss": 0.16499075889587403, "step": 13020 }, { "epoch": 0.34323006358185465, "grad_norm": 3.009828805923462, "learning_rate": 1.528146334380803e-05, "loss": 0.18798869848251343, "step": 13030 }, { "epoch": 0.3434934788263534, "grad_norm": 2.159287691116333, "learning_rate": 1.5274216978818813e-05, "loss": 0.15959391593933106, "step": 13040 }, { "epoch": 0.34375689407085214, "grad_norm": 5.048702239990234, "learning_rate": 1.526696677502693e-05, "loss": 0.15725278854370117, "step": 13050 }, { "epoch": 0.3440203093153508, "grad_norm": 2.7990164756774902, "learning_rate": 1.525971273770939e-05, "loss": 0.1598353385925293, "step": 13060 }, { "epoch": 0.34428372455984957, "grad_norm": 2.841128349304199, "learning_rate": 1.5252454872146003e-05, "loss": 0.1610758662223816, "step": 13070 }, { "epoch": 0.3445471398043483, "grad_norm": 2.5701260566711426, "learning_rate": 1.5245193183619347e-05, "loss": 0.1640993118286133, "step": 13080 }, { "epoch": 0.34481055504884706, "grad_norm": 0.780337393283844, "learning_rate": 1.5237927677414797e-05, "loss": 0.15783545970916749, "step": 13090 }, { "epoch": 0.3450739702933458, "grad_norm": 0.8684700727462769, "learning_rate": 1.5230658358820493e-05, "loss": 0.15619157552719115, "step": 13100 }, { "epoch": 0.34533738553784454, "grad_norm": 1.986222267150879, "learning_rate": 1.5223385233127372e-05, "loss": 0.15197412967681884, "step": 13110 }, { "epoch": 0.3456008007823433, "grad_norm": 3.2464942932128906, "learning_rate": 1.5216108305629123e-05, "loss": 0.1547295093536377, "step": 13120 }, { "epoch": 0.34586421602684203, "grad_norm": 2.981682777404785, "learning_rate": 1.5208827581622205e-05, "loss": 0.17661104202270508, "step": 13130 }, { "epoch": 0.3461276312713408, "grad_norm": 1.3145333528518677, "learning_rate": 1.5201543066405844e-05, "loss": 0.15438836812973022, "step": 13140 }, { "epoch": 0.34639104651583946, "grad_norm": 1.675646424293518, "learning_rate": 1.5194254765282025e-05, "loss": 0.16747381687164306, "step": 13150 }, { "epoch": 0.3466544617603382, "grad_norm": 1.40717351436615, "learning_rate": 1.5186962683555493e-05, "loss": 0.1592262029647827, "step": 13160 }, { "epoch": 0.34691787700483695, "grad_norm": 1.0149074792861938, "learning_rate": 1.517966682653373e-05, "loss": 0.14794304370880126, "step": 13170 }, { "epoch": 0.3471812922493357, "grad_norm": 3.0287933349609375, "learning_rate": 1.5172367199526985e-05, "loss": 0.16306759119033815, "step": 13180 }, { "epoch": 0.34744470749383444, "grad_norm": 2.444000482559204, "learning_rate": 1.5165063807848234e-05, "loss": 0.16980907917022706, "step": 13190 }, { "epoch": 0.3477081227383332, "grad_norm": 3.4584012031555176, "learning_rate": 1.5157756656813207e-05, "loss": 0.1720575213432312, "step": 13200 }, { "epoch": 0.3479715379828319, "grad_norm": 2.7019195556640625, "learning_rate": 1.5150445751740356e-05, "loss": 0.15381661653518677, "step": 13210 }, { "epoch": 0.34823495322733067, "grad_norm": 1.2897852659225464, "learning_rate": 1.5143131097950882e-05, "loss": 0.15318506956100464, "step": 13220 }, { "epoch": 0.3484983684718294, "grad_norm": 2.1714766025543213, "learning_rate": 1.5135812700768702e-05, "loss": 0.17357442378997803, "step": 13230 }, { "epoch": 0.3487617837163281, "grad_norm": 1.0076520442962646, "learning_rate": 1.512849056552046e-05, "loss": 0.15145435333251953, "step": 13240 }, { "epoch": 0.34902519896082684, "grad_norm": 2.1412487030029297, "learning_rate": 1.512116469753552e-05, "loss": 0.15723694562911988, "step": 13250 }, { "epoch": 0.3492886142053256, "grad_norm": 2.888231039047241, "learning_rate": 1.5113835102145968e-05, "loss": 0.1566460132598877, "step": 13260 }, { "epoch": 0.34955202944982433, "grad_norm": 1.9512666463851929, "learning_rate": 1.51065017846866e-05, "loss": 0.16052062511444093, "step": 13270 }, { "epoch": 0.3498154446943231, "grad_norm": 1.4869699478149414, "learning_rate": 1.509916475049492e-05, "loss": 0.15348443984985352, "step": 13280 }, { "epoch": 0.3500788599388218, "grad_norm": 4.0496506690979, "learning_rate": 1.5091824004911137e-05, "loss": 0.18549782037734985, "step": 13290 }, { "epoch": 0.35034227518332056, "grad_norm": 3.675184488296509, "learning_rate": 1.5084479553278164e-05, "loss": 0.16187057495117188, "step": 13300 }, { "epoch": 0.3506056904278193, "grad_norm": 1.7936882972717285, "learning_rate": 1.5077131400941609e-05, "loss": 0.15224459171295165, "step": 13310 }, { "epoch": 0.350869105672318, "grad_norm": 0.9686471223831177, "learning_rate": 1.5069779553249775e-05, "loss": 0.1527955174446106, "step": 13320 }, { "epoch": 0.35113252091681674, "grad_norm": 1.556260585784912, "learning_rate": 1.5062424015553653e-05, "loss": 0.1790785551071167, "step": 13330 }, { "epoch": 0.3513959361613155, "grad_norm": 1.6453629732131958, "learning_rate": 1.505506479320692e-05, "loss": 0.14949235916137696, "step": 13340 }, { "epoch": 0.3516593514058142, "grad_norm": 1.6664364337921143, "learning_rate": 1.5047701891565938e-05, "loss": 0.16434990167617797, "step": 13350 }, { "epoch": 0.35192276665031297, "grad_norm": 1.178947925567627, "learning_rate": 1.504033531598974e-05, "loss": 0.1581302762031555, "step": 13360 }, { "epoch": 0.3521861818948117, "grad_norm": 2.550812244415283, "learning_rate": 1.5032965071840045e-05, "loss": 0.16299725770950318, "step": 13370 }, { "epoch": 0.35244959713931046, "grad_norm": 0.6258136630058289, "learning_rate": 1.5025591164481226e-05, "loss": 0.17651286125183105, "step": 13380 }, { "epoch": 0.3527130123838092, "grad_norm": 3.208214521408081, "learning_rate": 1.5018213599280339e-05, "loss": 0.16464560031890868, "step": 13390 }, { "epoch": 0.35297642762830794, "grad_norm": 1.7496134042739868, "learning_rate": 1.5010832381607085e-05, "loss": 0.15972195863723754, "step": 13400 }, { "epoch": 0.35323984287280663, "grad_norm": 3.709364414215088, "learning_rate": 1.500344751683384e-05, "loss": 0.16708455085754395, "step": 13410 }, { "epoch": 0.3535032581173054, "grad_norm": 4.0033183097839355, "learning_rate": 1.4996059010335626e-05, "loss": 0.16794264316558838, "step": 13420 }, { "epoch": 0.3537666733618041, "grad_norm": 1.475917935371399, "learning_rate": 1.4988666867490117e-05, "loss": 0.15090568065643312, "step": 13430 }, { "epoch": 0.35403008860630286, "grad_norm": 4.346846103668213, "learning_rate": 1.4981271093677632e-05, "loss": 0.1577092170715332, "step": 13440 }, { "epoch": 0.3542935038508016, "grad_norm": 2.8634018898010254, "learning_rate": 1.4973871694281133e-05, "loss": 0.1657116413116455, "step": 13450 }, { "epoch": 0.35455691909530035, "grad_norm": 2.776815891265869, "learning_rate": 1.4966468674686226e-05, "loss": 0.15446724891662597, "step": 13460 }, { "epoch": 0.3548203343397991, "grad_norm": 3.305234670639038, "learning_rate": 1.4959062040281143e-05, "loss": 0.16078269481658936, "step": 13470 }, { "epoch": 0.35508374958429784, "grad_norm": 1.9433318376541138, "learning_rate": 1.4951651796456758e-05, "loss": 0.16352198123931885, "step": 13480 }, { "epoch": 0.3553471648287966, "grad_norm": 1.1535645723342896, "learning_rate": 1.4944237948606561e-05, "loss": 0.15399281978607177, "step": 13490 }, { "epoch": 0.35561058007329527, "grad_norm": 1.0361511707305908, "learning_rate": 1.4936820502126676e-05, "loss": 0.14732692241668702, "step": 13500 }, { "epoch": 0.355873995317794, "grad_norm": 2.2341055870056152, "learning_rate": 1.4929399462415832e-05, "loss": 0.15774080753326417, "step": 13510 }, { "epoch": 0.35613741056229276, "grad_norm": 1.3424617052078247, "learning_rate": 1.492197483487539e-05, "loss": 0.1559865355491638, "step": 13520 }, { "epoch": 0.3564008258067915, "grad_norm": 2.6288154125213623, "learning_rate": 1.491454662490931e-05, "loss": 0.16691796779632567, "step": 13530 }, { "epoch": 0.35666424105129024, "grad_norm": 1.7737241983413696, "learning_rate": 1.4907114837924168e-05, "loss": 0.1577976942062378, "step": 13540 }, { "epoch": 0.356927656295789, "grad_norm": 0.7902471423149109, "learning_rate": 1.4899679479329133e-05, "loss": 0.15238602161407472, "step": 13550 }, { "epoch": 0.35719107154028773, "grad_norm": 3.517662763595581, "learning_rate": 1.4892240554535981e-05, "loss": 0.1526522159576416, "step": 13560 }, { "epoch": 0.3574544867847865, "grad_norm": 0.8081567883491516, "learning_rate": 1.4884798068959085e-05, "loss": 0.15120327472686768, "step": 13570 }, { "epoch": 0.3577179020292852, "grad_norm": 1.0499403476715088, "learning_rate": 1.4877352028015408e-05, "loss": 0.1597208023071289, "step": 13580 }, { "epoch": 0.3579813172737839, "grad_norm": 1.4289036989212036, "learning_rate": 1.4869902437124499e-05, "loss": 0.16865940093994142, "step": 13590 }, { "epoch": 0.35824473251828265, "grad_norm": 2.0047402381896973, "learning_rate": 1.486244930170849e-05, "loss": 0.1593583941459656, "step": 13600 }, { "epoch": 0.3585081477627814, "grad_norm": 2.1193652153015137, "learning_rate": 1.4854992627192096e-05, "loss": 0.16536051034927368, "step": 13610 }, { "epoch": 0.35877156300728014, "grad_norm": 3.7147343158721924, "learning_rate": 1.4847532419002602e-05, "loss": 0.15875542163848877, "step": 13620 }, { "epoch": 0.3590349782517789, "grad_norm": 0.7716376185417175, "learning_rate": 1.4840068682569877e-05, "loss": 0.16922330856323242, "step": 13630 }, { "epoch": 0.3592983934962776, "grad_norm": 2.796682357788086, "learning_rate": 1.4832601423326346e-05, "loss": 0.13906619548797608, "step": 13640 }, { "epoch": 0.35956180874077637, "grad_norm": 1.2022768259048462, "learning_rate": 1.4825130646707002e-05, "loss": 0.13839188814163209, "step": 13650 }, { "epoch": 0.3598252239852751, "grad_norm": 2.555393695831299, "learning_rate": 1.4817656358149398e-05, "loss": 0.16463189125061034, "step": 13660 }, { "epoch": 0.3600886392297738, "grad_norm": 0.9329903721809387, "learning_rate": 1.4810178563093648e-05, "loss": 0.16276812553405762, "step": 13670 }, { "epoch": 0.36035205447427254, "grad_norm": 1.419155478477478, "learning_rate": 1.4802697266982411e-05, "loss": 0.1689749240875244, "step": 13680 }, { "epoch": 0.3606154697187713, "grad_norm": 2.5536837577819824, "learning_rate": 1.4795212475260896e-05, "loss": 0.15813233852386474, "step": 13690 }, { "epoch": 0.36087888496327003, "grad_norm": 1.5055019855499268, "learning_rate": 1.4787724193376858e-05, "loss": 0.1554615616798401, "step": 13700 }, { "epoch": 0.3611423002077688, "grad_norm": 3.834707021713257, "learning_rate": 1.4780232426780592e-05, "loss": 0.1556493043899536, "step": 13710 }, { "epoch": 0.3614057154522675, "grad_norm": 1.2424087524414062, "learning_rate": 1.477273718092493e-05, "loss": 0.16134587526321412, "step": 13720 }, { "epoch": 0.36166913069676626, "grad_norm": 1.3306254148483276, "learning_rate": 1.4765238461265235e-05, "loss": 0.1614288568496704, "step": 13730 }, { "epoch": 0.361932545941265, "grad_norm": 2.7573094367980957, "learning_rate": 1.4757736273259396e-05, "loss": 0.17377245426177979, "step": 13740 }, { "epoch": 0.36219596118576375, "grad_norm": 2.157956123352051, "learning_rate": 1.4750230622367831e-05, "loss": 0.1551492691040039, "step": 13750 }, { "epoch": 0.36245937643026244, "grad_norm": 3.9151782989501953, "learning_rate": 1.474272151405348e-05, "loss": 0.15541268587112428, "step": 13760 }, { "epoch": 0.3627227916747612, "grad_norm": 1.9270018339157104, "learning_rate": 1.473520895378179e-05, "loss": 0.16688116788864135, "step": 13770 }, { "epoch": 0.3629862069192599, "grad_norm": 2.3966753482818604, "learning_rate": 1.472769294702073e-05, "loss": 0.17488651275634765, "step": 13780 }, { "epoch": 0.36324962216375867, "grad_norm": 4.181122303009033, "learning_rate": 1.4720173499240771e-05, "loss": 0.16579897403717042, "step": 13790 }, { "epoch": 0.3635130374082574, "grad_norm": 2.4726991653442383, "learning_rate": 1.4712650615914899e-05, "loss": 0.1608533501625061, "step": 13800 }, { "epoch": 0.36377645265275615, "grad_norm": 2.9288089275360107, "learning_rate": 1.4705124302518581e-05, "loss": 0.1758953332901001, "step": 13810 }, { "epoch": 0.3640398678972549, "grad_norm": 3.1712698936462402, "learning_rate": 1.4697594564529805e-05, "loss": 0.1721956968307495, "step": 13820 }, { "epoch": 0.36430328314175364, "grad_norm": 1.2789112329483032, "learning_rate": 1.4690061407429028e-05, "loss": 0.15439612865448, "step": 13830 }, { "epoch": 0.3645666983862524, "grad_norm": 1.2428339719772339, "learning_rate": 1.468252483669922e-05, "loss": 0.16959123611450194, "step": 13840 }, { "epoch": 0.3648301136307511, "grad_norm": 1.4412213563919067, "learning_rate": 1.4674984857825808e-05, "loss": 0.17516423463821412, "step": 13850 }, { "epoch": 0.3650935288752498, "grad_norm": 2.478947401046753, "learning_rate": 1.4667441476296723e-05, "loss": 0.16587460041046143, "step": 13860 }, { "epoch": 0.36535694411974856, "grad_norm": 1.242981195449829, "learning_rate": 1.4659894697602358e-05, "loss": 0.15998352766036988, "step": 13870 }, { "epoch": 0.3656203593642473, "grad_norm": 3.11322021484375, "learning_rate": 1.465234452723559e-05, "loss": 0.15779421329498292, "step": 13880 }, { "epoch": 0.36588377460874605, "grad_norm": 1.828975796699524, "learning_rate": 1.4644790970691755e-05, "loss": 0.15190746784210205, "step": 13890 }, { "epoch": 0.3661471898532448, "grad_norm": 2.2903480529785156, "learning_rate": 1.4637234033468658e-05, "loss": 0.15627124309539794, "step": 13900 }, { "epoch": 0.36641060509774354, "grad_norm": 0.718008816242218, "learning_rate": 1.4629673721066566e-05, "loss": 0.15835541486740112, "step": 13910 }, { "epoch": 0.3666740203422423, "grad_norm": 3.111137866973877, "learning_rate": 1.4622110038988196e-05, "loss": 0.16323221921920777, "step": 13920 }, { "epoch": 0.36693743558674097, "grad_norm": 1.018926978111267, "learning_rate": 1.4614542992738727e-05, "loss": 0.17072926759719848, "step": 13930 }, { "epoch": 0.3672008508312397, "grad_norm": 0.8419188857078552, "learning_rate": 1.4606972587825779e-05, "loss": 0.16902432441711426, "step": 13940 }, { "epoch": 0.36746426607573845, "grad_norm": 2.6583316326141357, "learning_rate": 1.4599398829759419e-05, "loss": 0.15961642265319825, "step": 13950 }, { "epoch": 0.3677276813202372, "grad_norm": 1.7969391345977783, "learning_rate": 1.4591821724052155e-05, "loss": 0.15223778486251832, "step": 13960 }, { "epoch": 0.36799109656473594, "grad_norm": 1.5195804834365845, "learning_rate": 1.4584241276218935e-05, "loss": 0.16970887184143066, "step": 13970 }, { "epoch": 0.3682545118092347, "grad_norm": 0.7372853755950928, "learning_rate": 1.4576657491777134e-05, "loss": 0.16130306720733642, "step": 13980 }, { "epoch": 0.36851792705373343, "grad_norm": 1.9860111474990845, "learning_rate": 1.4569070376246552e-05, "loss": 0.16585664749145507, "step": 13990 }, { "epoch": 0.3687813422982322, "grad_norm": 0.9306544661521912, "learning_rate": 1.4561479935149426e-05, "loss": 0.16253440380096434, "step": 14000 }, { "epoch": 0.3690447575427309, "grad_norm": 2.0542361736297607, "learning_rate": 1.4553886174010405e-05, "loss": 0.1713175058364868, "step": 14010 }, { "epoch": 0.3693081727872296, "grad_norm": 0.7722748517990112, "learning_rate": 1.4546289098356556e-05, "loss": 0.14716097116470336, "step": 14020 }, { "epoch": 0.36957158803172835, "grad_norm": 1.42759370803833, "learning_rate": 1.4538688713717357e-05, "loss": 0.14907078742980956, "step": 14030 }, { "epoch": 0.3698350032762271, "grad_norm": 3.562758207321167, "learning_rate": 1.4531085025624694e-05, "loss": 0.1613990545272827, "step": 14040 }, { "epoch": 0.37009841852072584, "grad_norm": 1.0599431991577148, "learning_rate": 1.4523478039612863e-05, "loss": 0.1774560332298279, "step": 14050 }, { "epoch": 0.3703618337652246, "grad_norm": 1.6705571413040161, "learning_rate": 1.4515867761218558e-05, "loss": 0.16094647645950316, "step": 14060 }, { "epoch": 0.3706252490097233, "grad_norm": 1.5989559888839722, "learning_rate": 1.4508254195980861e-05, "loss": 0.14968886375427246, "step": 14070 }, { "epoch": 0.37088866425422207, "grad_norm": 1.117005705833435, "learning_rate": 1.450063734944126e-05, "loss": 0.16151833534240723, "step": 14080 }, { "epoch": 0.3711520794987208, "grad_norm": 1.4636526107788086, "learning_rate": 1.449301722714362e-05, "loss": 0.15900840759277343, "step": 14090 }, { "epoch": 0.37141549474321955, "grad_norm": 1.2198823690414429, "learning_rate": 1.4485393834634202e-05, "loss": 0.15766713619232178, "step": 14100 }, { "epoch": 0.37167890998771824, "grad_norm": 2.449146032333374, "learning_rate": 1.447776717746163e-05, "loss": 0.157135272026062, "step": 14110 }, { "epoch": 0.371942325232217, "grad_norm": 1.6207059621810913, "learning_rate": 1.4470137261176922e-05, "loss": 0.16371628046035766, "step": 14120 }, { "epoch": 0.37220574047671573, "grad_norm": 4.701324462890625, "learning_rate": 1.4462504091333454e-05, "loss": 0.15985503196716308, "step": 14130 }, { "epoch": 0.3724691557212145, "grad_norm": 1.8411732912063599, "learning_rate": 1.4454867673486983e-05, "loss": 0.1608067750930786, "step": 14140 }, { "epoch": 0.3727325709657132, "grad_norm": 1.7759637832641602, "learning_rate": 1.4447228013195616e-05, "loss": 0.16208053827285768, "step": 14150 }, { "epoch": 0.37299598621021196, "grad_norm": 1.3489598035812378, "learning_rate": 1.4439585116019835e-05, "loss": 0.16315631866455077, "step": 14160 }, { "epoch": 0.3732594014547107, "grad_norm": 2.4325788021087646, "learning_rate": 1.4431938987522463e-05, "loss": 0.16706695556640624, "step": 14170 }, { "epoch": 0.37352281669920945, "grad_norm": 1.6445215940475464, "learning_rate": 1.4424289633268689e-05, "loss": 0.16949899196624757, "step": 14180 }, { "epoch": 0.37378623194370814, "grad_norm": 1.331618309020996, "learning_rate": 1.4416637058826038e-05, "loss": 0.15119926929473876, "step": 14190 }, { "epoch": 0.3740496471882069, "grad_norm": 2.8081588745117188, "learning_rate": 1.4408981269764386e-05, "loss": 0.1623462438583374, "step": 14200 }, { "epoch": 0.3743130624327056, "grad_norm": 3.1727077960968018, "learning_rate": 1.4401322271655947e-05, "loss": 0.16036157608032225, "step": 14210 }, { "epoch": 0.37457647767720437, "grad_norm": 1.3883464336395264, "learning_rate": 1.4393660070075267e-05, "loss": 0.170221209526062, "step": 14220 }, { "epoch": 0.3748398929217031, "grad_norm": 3.3960399627685547, "learning_rate": 1.4385994670599233e-05, "loss": 0.15690181255340577, "step": 14230 }, { "epoch": 0.37510330816620185, "grad_norm": 1.1421246528625488, "learning_rate": 1.4378326078807048e-05, "loss": 0.15846357345581055, "step": 14240 }, { "epoch": 0.3753667234107006, "grad_norm": 0.8839032649993896, "learning_rate": 1.4370654300280246e-05, "loss": 0.16031999588012696, "step": 14250 }, { "epoch": 0.37563013865519934, "grad_norm": 1.884786605834961, "learning_rate": 1.4362979340602677e-05, "loss": 0.1613199830055237, "step": 14260 }, { "epoch": 0.3758935538996981, "grad_norm": 1.1782702207565308, "learning_rate": 1.4355301205360513e-05, "loss": 0.16225073337554932, "step": 14270 }, { "epoch": 0.3761569691441968, "grad_norm": 1.253873586654663, "learning_rate": 1.4347619900142223e-05, "loss": 0.1786879539489746, "step": 14280 }, { "epoch": 0.3764203843886955, "grad_norm": 2.2402007579803467, "learning_rate": 1.43399354305386e-05, "loss": 0.1463927745819092, "step": 14290 }, { "epoch": 0.37668379963319426, "grad_norm": 4.476430892944336, "learning_rate": 1.4332247802142729e-05, "loss": 0.17276132106781006, "step": 14300 }, { "epoch": 0.376947214877693, "grad_norm": 1.3495967388153076, "learning_rate": 1.4324557020550001e-05, "loss": 0.16111526489257813, "step": 14310 }, { "epoch": 0.37721063012219175, "grad_norm": 1.422609806060791, "learning_rate": 1.4316863091358093e-05, "loss": 0.15804381370544435, "step": 14320 }, { "epoch": 0.3774740453666905, "grad_norm": 1.1812386512756348, "learning_rate": 1.4309166020166984e-05, "loss": 0.1414476752281189, "step": 14330 }, { "epoch": 0.37773746061118924, "grad_norm": 0.9673908948898315, "learning_rate": 1.430146581257893e-05, "loss": 0.15545213222503662, "step": 14340 }, { "epoch": 0.378000875855688, "grad_norm": 3.174424886703491, "learning_rate": 1.4293762474198478e-05, "loss": 0.15755176544189453, "step": 14350 }, { "epoch": 0.3782642911001867, "grad_norm": 2.1479854583740234, "learning_rate": 1.428605601063245e-05, "loss": 0.16120536327362062, "step": 14360 }, { "epoch": 0.3785277063446854, "grad_norm": 1.2040653228759766, "learning_rate": 1.427834642748994e-05, "loss": 0.15789166688919068, "step": 14370 }, { "epoch": 0.37879112158918415, "grad_norm": 1.9878863096237183, "learning_rate": 1.4270633730382315e-05, "loss": 0.16886527538299562, "step": 14380 }, { "epoch": 0.3790545368336829, "grad_norm": 3.096762180328369, "learning_rate": 1.4262917924923212e-05, "loss": 0.14850401878356934, "step": 14390 }, { "epoch": 0.37931795207818164, "grad_norm": 1.865491271018982, "learning_rate": 1.4255199016728527e-05, "loss": 0.16015427112579345, "step": 14400 }, { "epoch": 0.3795813673226804, "grad_norm": 1.883485198020935, "learning_rate": 1.4247477011416409e-05, "loss": 0.15821332931518556, "step": 14410 }, { "epoch": 0.37984478256717913, "grad_norm": 0.6862416863441467, "learning_rate": 1.4239751914607274e-05, "loss": 0.16944698095321656, "step": 14420 }, { "epoch": 0.3801081978116779, "grad_norm": 3.483422040939331, "learning_rate": 1.4232023731923772e-05, "loss": 0.14734039306640626, "step": 14430 }, { "epoch": 0.3803716130561766, "grad_norm": 1.3103821277618408, "learning_rate": 1.422429246899082e-05, "loss": 0.16423335075378417, "step": 14440 }, { "epoch": 0.3806350283006753, "grad_norm": 2.2878856658935547, "learning_rate": 1.4216558131435551e-05, "loss": 0.16180727481842042, "step": 14450 }, { "epoch": 0.38089844354517405, "grad_norm": 1.7234885692596436, "learning_rate": 1.420882072488736e-05, "loss": 0.15575332641601564, "step": 14460 }, { "epoch": 0.3811618587896728, "grad_norm": 3.1115453243255615, "learning_rate": 1.420108025497786e-05, "loss": 0.1703550100326538, "step": 14470 }, { "epoch": 0.38142527403417154, "grad_norm": 1.3233168125152588, "learning_rate": 1.41933367273409e-05, "loss": 0.16219501495361327, "step": 14480 }, { "epoch": 0.3816886892786703, "grad_norm": 1.394536018371582, "learning_rate": 1.4185590147612555e-05, "loss": 0.1552955150604248, "step": 14490 }, { "epoch": 0.381952104523169, "grad_norm": 4.205181121826172, "learning_rate": 1.4177840521431117e-05, "loss": 0.15404160022735597, "step": 14500 }, { "epoch": 0.38221551976766777, "grad_norm": 5.701910972595215, "learning_rate": 1.4170087854437102e-05, "loss": 0.1497516632080078, "step": 14510 }, { "epoch": 0.3824789350121665, "grad_norm": 2.640321731567383, "learning_rate": 1.4162332152273233e-05, "loss": 0.17152905464172363, "step": 14520 }, { "epoch": 0.38274235025666525, "grad_norm": 2.345750093460083, "learning_rate": 1.4154573420584448e-05, "loss": 0.15545103549957276, "step": 14530 }, { "epoch": 0.38300576550116394, "grad_norm": 0.9004387855529785, "learning_rate": 1.414681166501788e-05, "loss": 0.15894801616668702, "step": 14540 }, { "epoch": 0.3832691807456627, "grad_norm": 0.7988437414169312, "learning_rate": 1.4139046891222877e-05, "loss": 0.17499198913574218, "step": 14550 }, { "epoch": 0.38353259599016143, "grad_norm": 1.5994335412979126, "learning_rate": 1.4131279104850968e-05, "loss": 0.18732330799102784, "step": 14560 }, { "epoch": 0.3837960112346602, "grad_norm": 0.9367303252220154, "learning_rate": 1.4123508311555894e-05, "loss": 0.15417041778564453, "step": 14570 }, { "epoch": 0.3840594264791589, "grad_norm": 1.6919058561325073, "learning_rate": 1.4115734516993562e-05, "loss": 0.17015045881271362, "step": 14580 }, { "epoch": 0.38432284172365766, "grad_norm": 4.726294994354248, "learning_rate": 1.4107957726822084e-05, "loss": 0.1664164423942566, "step": 14590 }, { "epoch": 0.3845862569681564, "grad_norm": 3.4013102054595947, "learning_rate": 1.4100177946701738e-05, "loss": 0.16563065052032472, "step": 14600 }, { "epoch": 0.38484967221265515, "grad_norm": 0.9509757161140442, "learning_rate": 1.4092395182294989e-05, "loss": 0.1594771146774292, "step": 14610 }, { "epoch": 0.3851130874571539, "grad_norm": 1.8797061443328857, "learning_rate": 1.4084609439266464e-05, "loss": 0.16621053218841553, "step": 14620 }, { "epoch": 0.3853765027016526, "grad_norm": 1.7033262252807617, "learning_rate": 1.4076820723282962e-05, "loss": 0.15437595844268798, "step": 14630 }, { "epoch": 0.3856399179461513, "grad_norm": 0.8112645149230957, "learning_rate": 1.4069029040013448e-05, "loss": 0.17035001516342163, "step": 14640 }, { "epoch": 0.38590333319065007, "grad_norm": 2.047931671142578, "learning_rate": 1.4061234395129049e-05, "loss": 0.1679358720779419, "step": 14650 }, { "epoch": 0.3861667484351488, "grad_norm": 2.046647071838379, "learning_rate": 1.405343679430304e-05, "loss": 0.16062276363372802, "step": 14660 }, { "epoch": 0.38643016367964755, "grad_norm": 1.9033626317977905, "learning_rate": 1.4045636243210852e-05, "loss": 0.16262547969818114, "step": 14670 }, { "epoch": 0.3866935789241463, "grad_norm": 1.239335060119629, "learning_rate": 1.4037832747530063e-05, "loss": 0.14746005535125734, "step": 14680 }, { "epoch": 0.38695699416864504, "grad_norm": 1.7865574359893799, "learning_rate": 1.403002631294039e-05, "loss": 0.15417972803115845, "step": 14690 }, { "epoch": 0.3872204094131438, "grad_norm": 1.7072341442108154, "learning_rate": 1.40222169451237e-05, "loss": 0.1479562282562256, "step": 14700 }, { "epoch": 0.3874838246576425, "grad_norm": 1.1938254833221436, "learning_rate": 1.4014404649763983e-05, "loss": 0.16444621086120606, "step": 14710 }, { "epoch": 0.3877472399021412, "grad_norm": 2.7734837532043457, "learning_rate": 1.4006589432547367e-05, "loss": 0.16089264154434205, "step": 14720 }, { "epoch": 0.38801065514663996, "grad_norm": 2.9442484378814697, "learning_rate": 1.3998771299162099e-05, "loss": 0.1620124101638794, "step": 14730 }, { "epoch": 0.3882740703911387, "grad_norm": 2.5206167697906494, "learning_rate": 1.3990950255298565e-05, "loss": 0.15979864597320556, "step": 14740 }, { "epoch": 0.38853748563563745, "grad_norm": 0.9935599565505981, "learning_rate": 1.3983126306649243e-05, "loss": 0.17705743312835692, "step": 14750 }, { "epoch": 0.3888009008801362, "grad_norm": 3.6090593338012695, "learning_rate": 1.3975299458908753e-05, "loss": 0.15628496408462525, "step": 14760 }, { "epoch": 0.38906431612463493, "grad_norm": 2.1162614822387695, "learning_rate": 1.3967469717773807e-05, "loss": 0.1625460147857666, "step": 14770 }, { "epoch": 0.3893277313691337, "grad_norm": 6.249377250671387, "learning_rate": 1.395963708894323e-05, "loss": 0.5233012199401855, "step": 14780 }, { "epoch": 0.3895911466136324, "grad_norm": 3.2792742252349854, "learning_rate": 1.3951801578117946e-05, "loss": 0.18635501861572265, "step": 14790 }, { "epoch": 0.3898545618581311, "grad_norm": 1.2690891027450562, "learning_rate": 1.3943963191000976e-05, "loss": 0.15697798728942872, "step": 14800 }, { "epoch": 0.39011797710262985, "grad_norm": 2.12166690826416, "learning_rate": 1.393612193329744e-05, "loss": 0.17152836322784423, "step": 14810 }, { "epoch": 0.3903813923471286, "grad_norm": 4.944075584411621, "learning_rate": 1.3928277810714542e-05, "loss": 0.1664700388908386, "step": 14820 }, { "epoch": 0.39064480759162734, "grad_norm": 3.693774461746216, "learning_rate": 1.3920430828961576e-05, "loss": 0.15618646144866943, "step": 14830 }, { "epoch": 0.3909082228361261, "grad_norm": 1.1345560550689697, "learning_rate": 1.3912580993749907e-05, "loss": 0.15532442331314086, "step": 14840 }, { "epoch": 0.39117163808062483, "grad_norm": 1.8452640771865845, "learning_rate": 1.3904728310792988e-05, "loss": 0.16046221256256105, "step": 14850 }, { "epoch": 0.39143505332512357, "grad_norm": 2.5159356594085693, "learning_rate": 1.3896872785806342e-05, "loss": 0.15543869733810425, "step": 14860 }, { "epoch": 0.3916984685696223, "grad_norm": 3.510531187057495, "learning_rate": 1.3889014424507557e-05, "loss": 0.14779195785522461, "step": 14870 }, { "epoch": 0.39196188381412106, "grad_norm": 1.7881743907928467, "learning_rate": 1.3881153232616284e-05, "loss": 0.15519537925720214, "step": 14880 }, { "epoch": 0.39222529905861975, "grad_norm": 1.1526639461517334, "learning_rate": 1.3873289215854247e-05, "loss": 0.15681636333465576, "step": 14890 }, { "epoch": 0.3924887143031185, "grad_norm": 2.2586863040924072, "learning_rate": 1.3865422379945205e-05, "loss": 0.14522287845611573, "step": 14900 }, { "epoch": 0.39275212954761723, "grad_norm": 1.4921613931655884, "learning_rate": 1.3857552730614989e-05, "loss": 0.1736075758934021, "step": 14910 }, { "epoch": 0.393015544792116, "grad_norm": 1.472511649131775, "learning_rate": 1.3849680273591468e-05, "loss": 0.15887938737869262, "step": 14920 }, { "epoch": 0.3932789600366147, "grad_norm": 1.8973714113235474, "learning_rate": 1.3841805014604553e-05, "loss": 0.16593608856201172, "step": 14930 }, { "epoch": 0.39354237528111347, "grad_norm": 2.8743979930877686, "learning_rate": 1.3833926959386199e-05, "loss": 0.15673911571502686, "step": 14940 }, { "epoch": 0.3938057905256122, "grad_norm": 2.5067861080169678, "learning_rate": 1.3826046113670395e-05, "loss": 0.16356658935546875, "step": 14950 }, { "epoch": 0.39406920577011095, "grad_norm": 1.2482409477233887, "learning_rate": 1.3818162483193159e-05, "loss": 0.1585782766342163, "step": 14960 }, { "epoch": 0.39433262101460964, "grad_norm": 1.2160186767578125, "learning_rate": 1.3810276073692537e-05, "loss": 0.14771537780761718, "step": 14970 }, { "epoch": 0.3945960362591084, "grad_norm": 0.7986568808555603, "learning_rate": 1.3802386890908602e-05, "loss": 0.15551316738128662, "step": 14980 }, { "epoch": 0.39485945150360713, "grad_norm": 2.060133695602417, "learning_rate": 1.3794494940583437e-05, "loss": 0.16337940692901612, "step": 14990 }, { "epoch": 0.39512286674810587, "grad_norm": 1.5474284887313843, "learning_rate": 1.3786600228461147e-05, "loss": 0.16450998783111573, "step": 15000 }, { "epoch": 0.3953862819926046, "grad_norm": 3.0316262245178223, "learning_rate": 1.377870276028784e-05, "loss": 0.15378892421722412, "step": 15010 }, { "epoch": 0.39564969723710336, "grad_norm": 1.3050819635391235, "learning_rate": 1.377080254181164e-05, "loss": 0.16629414558410643, "step": 15020 }, { "epoch": 0.3959131124816021, "grad_norm": 0.8801980018615723, "learning_rate": 1.3762899578782662e-05, "loss": 0.16574654579162598, "step": 15030 }, { "epoch": 0.39617652772610085, "grad_norm": 3.733158588409424, "learning_rate": 1.375499387695303e-05, "loss": 0.15849156379699708, "step": 15040 }, { "epoch": 0.3964399429705996, "grad_norm": 1.8074469566345215, "learning_rate": 1.3747085442076844e-05, "loss": 0.15735604763031005, "step": 15050 }, { "epoch": 0.3967033582150983, "grad_norm": 1.1832683086395264, "learning_rate": 1.3739174279910214e-05, "loss": 0.15410058498382567, "step": 15060 }, { "epoch": 0.396966773459597, "grad_norm": 1.4862561225891113, "learning_rate": 1.3731260396211219e-05, "loss": 0.14496339559555055, "step": 15070 }, { "epoch": 0.39723018870409577, "grad_norm": 3.6839849948883057, "learning_rate": 1.372334379673993e-05, "loss": 0.158633029460907, "step": 15080 }, { "epoch": 0.3974936039485945, "grad_norm": 0.8244784474372864, "learning_rate": 1.3715424487258386e-05, "loss": 0.1489630699157715, "step": 15090 }, { "epoch": 0.39775701919309325, "grad_norm": 1.6457266807556152, "learning_rate": 1.37075024735306e-05, "loss": 0.1716590404510498, "step": 15100 }, { "epoch": 0.398020434437592, "grad_norm": 1.8655197620391846, "learning_rate": 1.3699577761322563e-05, "loss": 0.15789899826049805, "step": 15110 }, { "epoch": 0.39828384968209074, "grad_norm": 1.7799829244613647, "learning_rate": 1.3691650356402212e-05, "loss": 0.16418282985687255, "step": 15120 }, { "epoch": 0.3985472649265895, "grad_norm": 1.5859270095825195, "learning_rate": 1.3683720264539469e-05, "loss": 0.14068005084991456, "step": 15130 }, { "epoch": 0.3988106801710882, "grad_norm": 1.39181649684906, "learning_rate": 1.3675787491506184e-05, "loss": 0.15249185562133788, "step": 15140 }, { "epoch": 0.3990740954155869, "grad_norm": 1.1956344842910767, "learning_rate": 1.3667852043076173e-05, "loss": 0.15785958766937255, "step": 15150 }, { "epoch": 0.39933751066008566, "grad_norm": 2.8401620388031006, "learning_rate": 1.3659913925025203e-05, "loss": 0.16308016777038575, "step": 15160 }, { "epoch": 0.3996009259045844, "grad_norm": 3.7313144207000732, "learning_rate": 1.3651973143130979e-05, "loss": 0.15335096120834352, "step": 15170 }, { "epoch": 0.39986434114908315, "grad_norm": 1.9621413946151733, "learning_rate": 1.3644029703173139e-05, "loss": 0.14948539733886718, "step": 15180 }, { "epoch": 0.4001277563935819, "grad_norm": 1.3092213869094849, "learning_rate": 1.3636083610933262e-05, "loss": 0.15736993551254272, "step": 15190 }, { "epoch": 0.40039117163808063, "grad_norm": 2.5942392349243164, "learning_rate": 1.3628134872194861e-05, "loss": 0.15852231979370118, "step": 15200 }, { "epoch": 0.4006545868825794, "grad_norm": 1.0970028638839722, "learning_rate": 1.362018349274337e-05, "loss": 0.14888436794281007, "step": 15210 }, { "epoch": 0.4009180021270781, "grad_norm": 2.255272626876831, "learning_rate": 1.3612229478366144e-05, "loss": 0.14890704154968262, "step": 15220 }, { "epoch": 0.4011814173715768, "grad_norm": 2.7750236988067627, "learning_rate": 1.360427283485246e-05, "loss": 0.3491981029510498, "step": 15230 }, { "epoch": 0.40144483261607555, "grad_norm": 0.9086284041404724, "learning_rate": 1.3596313567993504e-05, "loss": 0.35700254440307616, "step": 15240 }, { "epoch": 0.4017082478605743, "grad_norm": 0.8711146116256714, "learning_rate": 1.3588351683582374e-05, "loss": 0.15909340381622314, "step": 15250 }, { "epoch": 0.40197166310507304, "grad_norm": 2.2629048824310303, "learning_rate": 1.3580387187414077e-05, "loss": 0.15350466966629028, "step": 15260 }, { "epoch": 0.4022350783495718, "grad_norm": 2.152318000793457, "learning_rate": 1.3572420085285512e-05, "loss": 0.15170912742614745, "step": 15270 }, { "epoch": 0.4024984935940705, "grad_norm": 2.0383951663970947, "learning_rate": 1.3564450382995482e-05, "loss": 0.16151621341705322, "step": 15280 }, { "epoch": 0.40276190883856927, "grad_norm": 0.973297119140625, "learning_rate": 1.3556478086344678e-05, "loss": 0.1477547287940979, "step": 15290 }, { "epoch": 0.403025324083068, "grad_norm": 3.2883570194244385, "learning_rate": 1.3548503201135687e-05, "loss": 0.13876054286956788, "step": 15300 }, { "epoch": 0.40328873932756676, "grad_norm": 1.5632144212722778, "learning_rate": 1.3540525733172966e-05, "loss": 0.16547088623046874, "step": 15310 }, { "epoch": 0.40355215457206545, "grad_norm": 1.9724067449569702, "learning_rate": 1.353254568826287e-05, "loss": 0.15679097175598145, "step": 15320 }, { "epoch": 0.4038155698165642, "grad_norm": 1.4363526105880737, "learning_rate": 1.3524563072213612e-05, "loss": 0.16056860685348512, "step": 15330 }, { "epoch": 0.40407898506106293, "grad_norm": 2.041565418243408, "learning_rate": 1.3516577890835292e-05, "loss": 0.15464388132095336, "step": 15340 }, { "epoch": 0.4043424003055617, "grad_norm": 1.3943575620651245, "learning_rate": 1.350859014993986e-05, "loss": 0.16275719404220582, "step": 15350 }, { "epoch": 0.4046058155500604, "grad_norm": 1.2282586097717285, "learning_rate": 1.3500599855341146e-05, "loss": 0.14965800046920777, "step": 15360 }, { "epoch": 0.40486923079455917, "grad_norm": 2.3869800567626953, "learning_rate": 1.3492607012854824e-05, "loss": 0.1602623224258423, "step": 15370 }, { "epoch": 0.4051326460390579, "grad_norm": 1.339966058731079, "learning_rate": 1.3484611628298438e-05, "loss": 0.14100385904312135, "step": 15380 }, { "epoch": 0.40539606128355665, "grad_norm": 3.6482748985290527, "learning_rate": 1.3476613707491372e-05, "loss": 0.15127038955688477, "step": 15390 }, { "epoch": 0.4056594765280554, "grad_norm": 0.7006930708885193, "learning_rate": 1.3468613256254852e-05, "loss": 0.14484143257141113, "step": 15400 }, { "epoch": 0.4059228917725541, "grad_norm": 0.7088536620140076, "learning_rate": 1.3460610280411957e-05, "loss": 0.1606515884399414, "step": 15410 }, { "epoch": 0.40618630701705283, "grad_norm": 1.634874701499939, "learning_rate": 1.3452604785787595e-05, "loss": 0.15322043895721435, "step": 15420 }, { "epoch": 0.40644972226155157, "grad_norm": 2.3096957206726074, "learning_rate": 1.3444596778208516e-05, "loss": 0.1599031686782837, "step": 15430 }, { "epoch": 0.4067131375060503, "grad_norm": 2.443615674972534, "learning_rate": 1.3436586263503286e-05, "loss": 0.14971667528152466, "step": 15440 }, { "epoch": 0.40697655275054906, "grad_norm": 1.970433235168457, "learning_rate": 1.3428573247502307e-05, "loss": 0.16775765419006347, "step": 15450 }, { "epoch": 0.4072399679950478, "grad_norm": 3.9891257286071777, "learning_rate": 1.3420557736037796e-05, "loss": 0.15326448678970336, "step": 15460 }, { "epoch": 0.40750338323954655, "grad_norm": 0.9568434357643127, "learning_rate": 1.3412539734943792e-05, "loss": 0.13835556507110597, "step": 15470 }, { "epoch": 0.4077667984840453, "grad_norm": 1.1259909868240356, "learning_rate": 1.3404519250056139e-05, "loss": 0.14399116039276122, "step": 15480 }, { "epoch": 0.408030213728544, "grad_norm": 2.444887161254883, "learning_rate": 1.3396496287212491e-05, "loss": 0.1528860807418823, "step": 15490 }, { "epoch": 0.4082936289730427, "grad_norm": 3.0641016960144043, "learning_rate": 1.3388470852252307e-05, "loss": 0.15545694828033446, "step": 15500 }, { "epoch": 0.40855704421754147, "grad_norm": 1.5791088342666626, "learning_rate": 1.3380442951016846e-05, "loss": 0.1448586702346802, "step": 15510 }, { "epoch": 0.4088204594620402, "grad_norm": 2.5434417724609375, "learning_rate": 1.3372412589349164e-05, "loss": 0.15230677127838135, "step": 15520 }, { "epoch": 0.40908387470653895, "grad_norm": 2.17634654045105, "learning_rate": 1.3364379773094099e-05, "loss": 0.15511126518249513, "step": 15530 }, { "epoch": 0.4093472899510377, "grad_norm": 1.6157563924789429, "learning_rate": 1.335634450809828e-05, "loss": 0.16486989259719848, "step": 15540 }, { "epoch": 0.40961070519553644, "grad_norm": 1.0103594064712524, "learning_rate": 1.3348306800210127e-05, "loss": 0.1556219696998596, "step": 15550 }, { "epoch": 0.4098741204400352, "grad_norm": 1.5967402458190918, "learning_rate": 1.3340266655279828e-05, "loss": 0.15059397220611573, "step": 15560 }, { "epoch": 0.4101375356845339, "grad_norm": 0.9035849571228027, "learning_rate": 1.3332224079159342e-05, "loss": 0.1500466585159302, "step": 15570 }, { "epoch": 0.4104009509290326, "grad_norm": 0.6845632791519165, "learning_rate": 1.3324179077702408e-05, "loss": 0.15461220741271972, "step": 15580 }, { "epoch": 0.41066436617353136, "grad_norm": 3.8803627490997314, "learning_rate": 1.3316131656764523e-05, "loss": 0.13735017776489258, "step": 15590 }, { "epoch": 0.4109277814180301, "grad_norm": 1.8745982646942139, "learning_rate": 1.3308081822202953e-05, "loss": 0.15935821533203126, "step": 15600 }, { "epoch": 0.41119119666252885, "grad_norm": 1.3692634105682373, "learning_rate": 1.3300029579876705e-05, "loss": 0.1620649814605713, "step": 15610 }, { "epoch": 0.4114546119070276, "grad_norm": 1.1543495655059814, "learning_rate": 1.3291974935646554e-05, "loss": 0.14815449714660645, "step": 15620 }, { "epoch": 0.41171802715152633, "grad_norm": 1.2199329137802124, "learning_rate": 1.3283917895375018e-05, "loss": 0.14624087810516356, "step": 15630 }, { "epoch": 0.4119814423960251, "grad_norm": 1.6117390394210815, "learning_rate": 1.327585846492636e-05, "loss": 0.14290101528167726, "step": 15640 }, { "epoch": 0.4122448576405238, "grad_norm": 2.7908506393432617, "learning_rate": 1.3267796650166576e-05, "loss": 0.16364755630493164, "step": 15650 }, { "epoch": 0.41250827288502256, "grad_norm": 1.030150294303894, "learning_rate": 1.3259732456963406e-05, "loss": 0.16272857189178466, "step": 15660 }, { "epoch": 0.41277168812952125, "grad_norm": 3.12846040725708, "learning_rate": 1.3251665891186313e-05, "loss": 0.15006940364837645, "step": 15670 }, { "epoch": 0.41303510337402, "grad_norm": 1.376662254333496, "learning_rate": 1.32435969587065e-05, "loss": 0.14920709133148194, "step": 15680 }, { "epoch": 0.41329851861851874, "grad_norm": 1.0743873119354248, "learning_rate": 1.323552566539688e-05, "loss": 0.15409982204437256, "step": 15690 }, { "epoch": 0.4135619338630175, "grad_norm": 2.3537981510162354, "learning_rate": 1.3227452017132085e-05, "loss": 0.150724458694458, "step": 15700 }, { "epoch": 0.4138253491075162, "grad_norm": 3.039405584335327, "learning_rate": 1.321937601978847e-05, "loss": 0.14992226362228395, "step": 15710 }, { "epoch": 0.41408876435201497, "grad_norm": 1.239086627960205, "learning_rate": 1.3211297679244093e-05, "loss": 0.1588166832923889, "step": 15720 }, { "epoch": 0.4143521795965137, "grad_norm": 2.206144332885742, "learning_rate": 1.3203217001378716e-05, "loss": 0.16569199562072753, "step": 15730 }, { "epoch": 0.41461559484101246, "grad_norm": 2.136653423309326, "learning_rate": 1.3195133992073808e-05, "loss": 0.14890867471694946, "step": 15740 }, { "epoch": 0.41487901008551115, "grad_norm": 2.3060193061828613, "learning_rate": 1.318704865721253e-05, "loss": 0.15257501602172852, "step": 15750 }, { "epoch": 0.4151424253300099, "grad_norm": 2.6170287132263184, "learning_rate": 1.3178961002679738e-05, "loss": 0.15653600692749023, "step": 15760 }, { "epoch": 0.41540584057450863, "grad_norm": 1.658083200454712, "learning_rate": 1.317087103436198e-05, "loss": 0.16172442436218262, "step": 15770 }, { "epoch": 0.4156692558190074, "grad_norm": 1.2473167181015015, "learning_rate": 1.3162778758147477e-05, "loss": 0.1523059129714966, "step": 15780 }, { "epoch": 0.4159326710635061, "grad_norm": 1.0130254030227661, "learning_rate": 1.315468417992614e-05, "loss": 0.15999965667724608, "step": 15790 }, { "epoch": 0.41619608630800486, "grad_norm": 1.3975350856781006, "learning_rate": 1.3146587305589552e-05, "loss": 0.1566930890083313, "step": 15800 }, { "epoch": 0.4164595015525036, "grad_norm": 1.9974117279052734, "learning_rate": 1.3138488141030967e-05, "loss": 0.14558095932006837, "step": 15810 }, { "epoch": 0.41672291679700235, "grad_norm": 2.685502052307129, "learning_rate": 1.313038669214531e-05, "loss": 0.1621572732925415, "step": 15820 }, { "epoch": 0.4169863320415011, "grad_norm": 0.8956100940704346, "learning_rate": 1.312228296482916e-05, "loss": 0.15572171211242675, "step": 15830 }, { "epoch": 0.4172497472859998, "grad_norm": 0.7769653797149658, "learning_rate": 1.311417696498076e-05, "loss": 0.1580382227897644, "step": 15840 }, { "epoch": 0.4175131625304985, "grad_norm": 1.2159281969070435, "learning_rate": 1.310606869850001e-05, "loss": 0.13666920661926268, "step": 15850 }, { "epoch": 0.41777657777499727, "grad_norm": 0.716962993144989, "learning_rate": 1.3097958171288453e-05, "loss": 0.14947797060012818, "step": 15860 }, { "epoch": 0.418039993019496, "grad_norm": 3.7638862133026123, "learning_rate": 1.3089845389249282e-05, "loss": 0.151194167137146, "step": 15870 }, { "epoch": 0.41830340826399476, "grad_norm": 0.9653654098510742, "learning_rate": 1.308173035828733e-05, "loss": 0.1569131135940552, "step": 15880 }, { "epoch": 0.4185668235084935, "grad_norm": 1.8172125816345215, "learning_rate": 1.3073613084309065e-05, "loss": 0.15747512578964235, "step": 15890 }, { "epoch": 0.41883023875299225, "grad_norm": 3.109323740005493, "learning_rate": 1.3065493573222594e-05, "loss": 0.16121134757995606, "step": 15900 }, { "epoch": 0.419093653997491, "grad_norm": 1.054032325744629, "learning_rate": 1.305737183093764e-05, "loss": 0.15359201431274414, "step": 15910 }, { "epoch": 0.41935706924198973, "grad_norm": 1.6783521175384521, "learning_rate": 1.3049247863365562e-05, "loss": 0.14671862125396729, "step": 15920 }, { "epoch": 0.4196204844864884, "grad_norm": 1.4131574630737305, "learning_rate": 1.3041121676419333e-05, "loss": 0.15311973094940184, "step": 15930 }, { "epoch": 0.41988389973098716, "grad_norm": 1.2586801052093506, "learning_rate": 1.3032993276013546e-05, "loss": 0.15254380702972412, "step": 15940 }, { "epoch": 0.4201473149754859, "grad_norm": 0.7001785635948181, "learning_rate": 1.3024862668064393e-05, "loss": 0.14546722173690796, "step": 15950 }, { "epoch": 0.42041073021998465, "grad_norm": 0.8735871911048889, "learning_rate": 1.3016729858489689e-05, "loss": 0.16555492877960204, "step": 15960 }, { "epoch": 0.4206741454644834, "grad_norm": 1.0758672952651978, "learning_rate": 1.300859485320884e-05, "loss": 0.16066136360168456, "step": 15970 }, { "epoch": 0.42093756070898214, "grad_norm": 2.544052839279175, "learning_rate": 1.3000457658142858e-05, "loss": 0.1599881887435913, "step": 15980 }, { "epoch": 0.4212009759534809, "grad_norm": 2.209754228591919, "learning_rate": 1.299231827921434e-05, "loss": 0.1555259943008423, "step": 15990 }, { "epoch": 0.4214643911979796, "grad_norm": 1.4943021535873413, "learning_rate": 1.2984176722347478e-05, "loss": 0.1504265546798706, "step": 16000 }, { "epoch": 0.4217278064424783, "grad_norm": 1.1458823680877686, "learning_rate": 1.2976032993468049e-05, "loss": 0.15208785533905028, "step": 16010 }, { "epoch": 0.42199122168697706, "grad_norm": 3.459364891052246, "learning_rate": 1.296788709850341e-05, "loss": 0.14168908596038818, "step": 16020 }, { "epoch": 0.4222546369314758, "grad_norm": 1.8981494903564453, "learning_rate": 1.29597390433825e-05, "loss": 0.14104118347167968, "step": 16030 }, { "epoch": 0.42251805217597455, "grad_norm": 1.704345464706421, "learning_rate": 1.2951588834035813e-05, "loss": 0.14607744216918944, "step": 16040 }, { "epoch": 0.4227814674204733, "grad_norm": 1.5819803476333618, "learning_rate": 1.2943436476395435e-05, "loss": 0.1622147560119629, "step": 16050 }, { "epoch": 0.42304488266497203, "grad_norm": 1.0661872625350952, "learning_rate": 1.2935281976394998e-05, "loss": 0.16063667535781861, "step": 16060 }, { "epoch": 0.4233082979094708, "grad_norm": 2.1556382179260254, "learning_rate": 1.2927125339969701e-05, "loss": 0.15253210067749023, "step": 16070 }, { "epoch": 0.4235717131539695, "grad_norm": 1.4034380912780762, "learning_rate": 1.2918966573056291e-05, "loss": 0.1447349190711975, "step": 16080 }, { "epoch": 0.42383512839846826, "grad_norm": 2.6522274017333984, "learning_rate": 1.2910805681593078e-05, "loss": 0.16005275249481202, "step": 16090 }, { "epoch": 0.42409854364296695, "grad_norm": 0.7686362266540527, "learning_rate": 1.2902642671519903e-05, "loss": 0.1473139762878418, "step": 16100 }, { "epoch": 0.4243619588874657, "grad_norm": 0.9847095608711243, "learning_rate": 1.2894477548778164e-05, "loss": 0.15727920532226564, "step": 16110 }, { "epoch": 0.42462537413196444, "grad_norm": 1.4461596012115479, "learning_rate": 1.2886310319310787e-05, "loss": 0.15324963331222535, "step": 16120 }, { "epoch": 0.4248887893764632, "grad_norm": 2.3647217750549316, "learning_rate": 1.2878140989062233e-05, "loss": 0.15715872049331664, "step": 16130 }, { "epoch": 0.4251522046209619, "grad_norm": 0.9733963012695312, "learning_rate": 1.2869969563978493e-05, "loss": 0.15666437149047852, "step": 16140 }, { "epoch": 0.42541561986546067, "grad_norm": 1.4700831174850464, "learning_rate": 1.2861796050007082e-05, "loss": 0.1480933666229248, "step": 16150 }, { "epoch": 0.4256790351099594, "grad_norm": 2.294638156890869, "learning_rate": 1.2853620453097039e-05, "loss": 0.16832102537155152, "step": 16160 }, { "epoch": 0.42594245035445816, "grad_norm": 1.0040233135223389, "learning_rate": 1.2845442779198916e-05, "loss": 0.1501576542854309, "step": 16170 }, { "epoch": 0.4262058655989569, "grad_norm": 0.5397549867630005, "learning_rate": 1.2837263034264773e-05, "loss": 0.14186930656433105, "step": 16180 }, { "epoch": 0.4264692808434556, "grad_norm": 1.3898855447769165, "learning_rate": 1.2829081224248183e-05, "loss": 0.15606018304824829, "step": 16190 }, { "epoch": 0.42673269608795433, "grad_norm": 1.6465195417404175, "learning_rate": 1.2820897355104224e-05, "loss": 0.15902764797210694, "step": 16200 }, { "epoch": 0.4269961113324531, "grad_norm": 1.5563993453979492, "learning_rate": 1.2812711432789464e-05, "loss": 0.15710607767105103, "step": 16210 }, { "epoch": 0.4272595265769518, "grad_norm": 2.0028510093688965, "learning_rate": 1.2804523463261975e-05, "loss": 0.13808599710464478, "step": 16220 }, { "epoch": 0.42752294182145056, "grad_norm": 1.109779953956604, "learning_rate": 1.2796333452481309e-05, "loss": 0.15857243537902832, "step": 16230 }, { "epoch": 0.4277863570659493, "grad_norm": 2.6579349040985107, "learning_rate": 1.2788141406408516e-05, "loss": 0.15383198261260986, "step": 16240 }, { "epoch": 0.42804977231044805, "grad_norm": 0.8993719220161438, "learning_rate": 1.2779947331006113e-05, "loss": 0.14722439050674438, "step": 16250 }, { "epoch": 0.4283131875549468, "grad_norm": 1.4726622104644775, "learning_rate": 1.2771751232238107e-05, "loss": 0.14281692504882812, "step": 16260 }, { "epoch": 0.42857660279944554, "grad_norm": 3.880511999130249, "learning_rate": 1.2763553116069969e-05, "loss": 0.1523958921432495, "step": 16270 }, { "epoch": 0.4288400180439442, "grad_norm": 1.4510000944137573, "learning_rate": 1.2755352988468646e-05, "loss": 0.14460785388946534, "step": 16280 }, { "epoch": 0.42910343328844297, "grad_norm": 1.8473707437515259, "learning_rate": 1.2747150855402541e-05, "loss": 0.15954279899597168, "step": 16290 }, { "epoch": 0.4293668485329417, "grad_norm": 0.76918625831604, "learning_rate": 1.2738946722841519e-05, "loss": 0.14919662475585938, "step": 16300 }, { "epoch": 0.42963026377744046, "grad_norm": 3.1272106170654297, "learning_rate": 1.2730740596756908e-05, "loss": 0.1395730495452881, "step": 16310 }, { "epoch": 0.4298936790219392, "grad_norm": 3.502929925918579, "learning_rate": 1.2722532483121475e-05, "loss": 0.15978851318359374, "step": 16320 }, { "epoch": 0.43015709426643794, "grad_norm": 1.2762552499771118, "learning_rate": 1.271432238790944e-05, "loss": 0.15154104232788085, "step": 16330 }, { "epoch": 0.4304205095109367, "grad_norm": 2.3651134967803955, "learning_rate": 1.2706110317096463e-05, "loss": 0.14618200063705444, "step": 16340 }, { "epoch": 0.43068392475543543, "grad_norm": 3.3647220134735107, "learning_rate": 1.269789627665965e-05, "loss": 0.1587885618209839, "step": 16350 }, { "epoch": 0.4309473399999341, "grad_norm": 1.433227300643921, "learning_rate": 1.2689680272577527e-05, "loss": 0.17181553840637206, "step": 16360 }, { "epoch": 0.43121075524443286, "grad_norm": 1.5660550594329834, "learning_rate": 1.2681462310830061e-05, "loss": 0.15452113151550292, "step": 16370 }, { "epoch": 0.4314741704889316, "grad_norm": 2.8285818099975586, "learning_rate": 1.2673242397398635e-05, "loss": 0.15511705875396728, "step": 16380 }, { "epoch": 0.43173758573343035, "grad_norm": 0.9234334230422974, "learning_rate": 1.266502053826606e-05, "loss": 0.15792561769485475, "step": 16390 }, { "epoch": 0.4320010009779291, "grad_norm": 3.9509897232055664, "learning_rate": 1.2656796739416557e-05, "loss": 0.1611285090446472, "step": 16400 }, { "epoch": 0.43226441622242784, "grad_norm": 1.3773064613342285, "learning_rate": 1.2648571006835764e-05, "loss": 0.13784401416778563, "step": 16410 }, { "epoch": 0.4325278314669266, "grad_norm": 0.9811415672302246, "learning_rate": 1.2640343346510724e-05, "loss": 0.15049262046813966, "step": 16420 }, { "epoch": 0.4327912467114253, "grad_norm": 0.9049400091171265, "learning_rate": 1.2632113764429884e-05, "loss": 0.15479564666748047, "step": 16430 }, { "epoch": 0.43305466195592407, "grad_norm": 1.385136365890503, "learning_rate": 1.2623882266583083e-05, "loss": 0.15418190956115724, "step": 16440 }, { "epoch": 0.43331807720042276, "grad_norm": 1.2683683633804321, "learning_rate": 1.2615648858961564e-05, "loss": 0.14406729936599733, "step": 16450 }, { "epoch": 0.4335814924449215, "grad_norm": 2.33311128616333, "learning_rate": 1.260741354755796e-05, "loss": 0.15452361106872559, "step": 16460 }, { "epoch": 0.43384490768942024, "grad_norm": 1.1179512739181519, "learning_rate": 1.2599176338366282e-05, "loss": 0.16845713853836058, "step": 16470 }, { "epoch": 0.434108322933919, "grad_norm": 2.3448660373687744, "learning_rate": 1.2590937237381927e-05, "loss": 0.14067296981811522, "step": 16480 }, { "epoch": 0.43437173817841773, "grad_norm": 1.3552653789520264, "learning_rate": 1.2582696250601667e-05, "loss": 0.15790326595306398, "step": 16490 }, { "epoch": 0.4346351534229165, "grad_norm": 1.5646400451660156, "learning_rate": 1.257445338402365e-05, "loss": 0.15263011455535888, "step": 16500 }, { "epoch": 0.4348985686674152, "grad_norm": 1.1546330451965332, "learning_rate": 1.256620864364739e-05, "loss": 0.15825531482696534, "step": 16510 }, { "epoch": 0.43516198391191396, "grad_norm": 1.2310433387756348, "learning_rate": 1.255796203547376e-05, "loss": 0.15041877031326295, "step": 16520 }, { "epoch": 0.4354253991564127, "grad_norm": 1.3590569496154785, "learning_rate": 1.2549713565505001e-05, "loss": 0.14612404108047486, "step": 16530 }, { "epoch": 0.4356888144009114, "grad_norm": 1.6311297416687012, "learning_rate": 1.254146323974471e-05, "loss": 0.14534382820129393, "step": 16540 }, { "epoch": 0.43595222964541014, "grad_norm": 1.8028194904327393, "learning_rate": 1.253321106419782e-05, "loss": 0.1380316972732544, "step": 16550 }, { "epoch": 0.4362156448899089, "grad_norm": 0.580466091632843, "learning_rate": 1.2524957044870627e-05, "loss": 0.15049206018447875, "step": 16560 }, { "epoch": 0.4364790601344076, "grad_norm": 1.0908881425857544, "learning_rate": 1.2516701187770761e-05, "loss": 0.15209172964096068, "step": 16570 }, { "epoch": 0.43674247537890637, "grad_norm": 1.4474984407424927, "learning_rate": 1.250844349890719e-05, "loss": 0.15100861787796022, "step": 16580 }, { "epoch": 0.4370058906234051, "grad_norm": 1.4980125427246094, "learning_rate": 1.250018398429022e-05, "loss": 0.16545181274414061, "step": 16590 }, { "epoch": 0.43726930586790386, "grad_norm": 3.2384467124938965, "learning_rate": 1.2491922649931475e-05, "loss": 0.1662168025970459, "step": 16600 }, { "epoch": 0.4375327211124026, "grad_norm": 2.7831263542175293, "learning_rate": 1.2483659501843916e-05, "loss": 0.15607577562332153, "step": 16610 }, { "epoch": 0.4377961363569013, "grad_norm": 0.6706334352493286, "learning_rate": 1.2475394546041822e-05, "loss": 0.14548947811126708, "step": 16620 }, { "epoch": 0.43805955160140003, "grad_norm": 1.6207035779953003, "learning_rate": 1.2467127788540774e-05, "loss": 0.13917461633682252, "step": 16630 }, { "epoch": 0.4383229668458988, "grad_norm": 1.5896637439727783, "learning_rate": 1.245885923535768e-05, "loss": 0.15649881362915039, "step": 16640 }, { "epoch": 0.4385863820903975, "grad_norm": 0.8670057058334351, "learning_rate": 1.2450588892510756e-05, "loss": 0.15223747491836548, "step": 16650 }, { "epoch": 0.43884979733489626, "grad_norm": 0.7708163857460022, "learning_rate": 1.2442316766019502e-05, "loss": 0.15726079940795898, "step": 16660 }, { "epoch": 0.439113212579395, "grad_norm": 1.7659831047058105, "learning_rate": 1.2434042861904743e-05, "loss": 0.1494495153427124, "step": 16670 }, { "epoch": 0.43937662782389375, "grad_norm": 2.3283751010894775, "learning_rate": 1.2425767186188569e-05, "loss": 0.14093434810638428, "step": 16680 }, { "epoch": 0.4396400430683925, "grad_norm": 2.7675864696502686, "learning_rate": 1.2417489744894381e-05, "loss": 0.14512486457824708, "step": 16690 }, { "epoch": 0.43990345831289124, "grad_norm": 3.4286694526672363, "learning_rate": 1.2409210544046859e-05, "loss": 0.14730582237243653, "step": 16700 }, { "epoch": 0.4401668735573899, "grad_norm": 1.7851117849349976, "learning_rate": 1.2400929589671962e-05, "loss": 0.15006959438323975, "step": 16710 }, { "epoch": 0.44043028880188867, "grad_norm": 3.865478515625, "learning_rate": 1.2392646887796924e-05, "loss": 0.16390817165374755, "step": 16720 }, { "epoch": 0.4406937040463874, "grad_norm": 0.8991650342941284, "learning_rate": 1.2384362444450253e-05, "loss": 0.15120633840560913, "step": 16730 }, { "epoch": 0.44095711929088616, "grad_norm": 0.695600688457489, "learning_rate": 1.2376076265661722e-05, "loss": 0.16102967262268067, "step": 16740 }, { "epoch": 0.4412205345353849, "grad_norm": 1.3506460189819336, "learning_rate": 1.2367788357462375e-05, "loss": 0.15616745948791505, "step": 16750 }, { "epoch": 0.44148394977988364, "grad_norm": 1.938599944114685, "learning_rate": 1.2359498725884509e-05, "loss": 0.15270363092422484, "step": 16760 }, { "epoch": 0.4417473650243824, "grad_norm": 1.9863756895065308, "learning_rate": 1.2351207376961666e-05, "loss": 0.15392929315567017, "step": 16770 }, { "epoch": 0.44201078026888113, "grad_norm": 2.3821306228637695, "learning_rate": 1.234291431672866e-05, "loss": 0.16022813320159912, "step": 16780 }, { "epoch": 0.4422741955133799, "grad_norm": 1.2398598194122314, "learning_rate": 1.2334619551221531e-05, "loss": 0.14218297004699706, "step": 16790 }, { "epoch": 0.44253761075787856, "grad_norm": 1.4395051002502441, "learning_rate": 1.232632308647757e-05, "loss": 0.15623154640197753, "step": 16800 }, { "epoch": 0.4428010260023773, "grad_norm": 2.646394968032837, "learning_rate": 1.2318024928535296e-05, "loss": 0.15131595134735107, "step": 16810 }, { "epoch": 0.44306444124687605, "grad_norm": 1.683043122291565, "learning_rate": 1.2309725083434477e-05, "loss": 0.15034570693969726, "step": 16820 }, { "epoch": 0.4433278564913748, "grad_norm": 0.9766079783439636, "learning_rate": 1.2301423557216088e-05, "loss": 0.15631994009017944, "step": 16830 }, { "epoch": 0.44359127173587354, "grad_norm": 2.5650908946990967, "learning_rate": 1.2293120355922346e-05, "loss": 0.14178526401519775, "step": 16840 }, { "epoch": 0.4438546869803723, "grad_norm": 1.0305907726287842, "learning_rate": 1.2284815485596671e-05, "loss": 0.133766770362854, "step": 16850 }, { "epoch": 0.444118102224871, "grad_norm": 1.9650368690490723, "learning_rate": 1.2276508952283712e-05, "loss": 0.1498384118080139, "step": 16860 }, { "epoch": 0.44438151746936977, "grad_norm": 2.8900256156921387, "learning_rate": 1.2268200762029319e-05, "loss": 0.14820647239685059, "step": 16870 }, { "epoch": 0.44464493271386846, "grad_norm": 1.1540567874908447, "learning_rate": 1.2259890920880555e-05, "loss": 0.153932523727417, "step": 16880 }, { "epoch": 0.4449083479583672, "grad_norm": 2.974741220474243, "learning_rate": 1.2251579434885679e-05, "loss": 0.14975526332855224, "step": 16890 }, { "epoch": 0.44517176320286594, "grad_norm": 0.9767012596130371, "learning_rate": 1.2243266310094146e-05, "loss": 0.14379959106445311, "step": 16900 }, { "epoch": 0.4454351784473647, "grad_norm": 1.875224232673645, "learning_rate": 1.2234951552556613e-05, "loss": 0.1537942409515381, "step": 16910 }, { "epoch": 0.44569859369186343, "grad_norm": 1.284682035446167, "learning_rate": 1.2226635168324916e-05, "loss": 0.14391570091247557, "step": 16920 }, { "epoch": 0.4459620089363622, "grad_norm": 1.838684320449829, "learning_rate": 1.2218317163452078e-05, "loss": 0.14195505380630494, "step": 16930 }, { "epoch": 0.4462254241808609, "grad_norm": 2.1460957527160645, "learning_rate": 1.2209997543992301e-05, "loss": 0.1497251272201538, "step": 16940 }, { "epoch": 0.44648883942535966, "grad_norm": 4.345428466796875, "learning_rate": 1.2201676316000964e-05, "loss": 0.1531028628349304, "step": 16950 }, { "epoch": 0.4467522546698584, "grad_norm": 1.0765693187713623, "learning_rate": 1.2193353485534614e-05, "loss": 0.1478060007095337, "step": 16960 }, { "epoch": 0.4470156699143571, "grad_norm": 1.1346330642700195, "learning_rate": 1.2185029058650972e-05, "loss": 0.1484243392944336, "step": 16970 }, { "epoch": 0.44727908515885584, "grad_norm": 0.9893016815185547, "learning_rate": 1.2176703041408908e-05, "loss": 0.14196276664733887, "step": 16980 }, { "epoch": 0.4475425004033546, "grad_norm": 2.9343435764312744, "learning_rate": 1.2168375439868464e-05, "loss": 0.15076253414154053, "step": 16990 }, { "epoch": 0.4478059156478533, "grad_norm": 1.2750167846679688, "learning_rate": 1.2160046260090816e-05, "loss": 0.13910696506500245, "step": 17000 }, { "epoch": 0.44806933089235207, "grad_norm": 1.789567232131958, "learning_rate": 1.2151715508138313e-05, "loss": 0.14134323596954346, "step": 17010 }, { "epoch": 0.4483327461368508, "grad_norm": 2.2539844512939453, "learning_rate": 1.2143383190074429e-05, "loss": 0.14925127029418944, "step": 17020 }, { "epoch": 0.44859616138134956, "grad_norm": 2.45245361328125, "learning_rate": 1.2135049311963787e-05, "loss": 0.15928752422332765, "step": 17030 }, { "epoch": 0.4488595766258483, "grad_norm": 0.8531352281570435, "learning_rate": 1.2126713879872138e-05, "loss": 0.15417776107788086, "step": 17040 }, { "epoch": 0.44912299187034704, "grad_norm": 1.3483890295028687, "learning_rate": 1.2118376899866375e-05, "loss": 0.16025030612945557, "step": 17050 }, { "epoch": 0.44938640711484573, "grad_norm": 3.445101499557495, "learning_rate": 1.2110038378014512e-05, "loss": 0.14956881999969482, "step": 17060 }, { "epoch": 0.4496498223593445, "grad_norm": 1.2502912282943726, "learning_rate": 1.210169832038568e-05, "loss": 0.14961498975753784, "step": 17070 }, { "epoch": 0.4499132376038432, "grad_norm": 1.5227994918823242, "learning_rate": 1.2093356733050137e-05, "loss": 0.14225741624832153, "step": 17080 }, { "epoch": 0.45017665284834196, "grad_norm": 4.727108478546143, "learning_rate": 1.2085013622079248e-05, "loss": 0.16117268800735474, "step": 17090 }, { "epoch": 0.4504400680928407, "grad_norm": 2.449047088623047, "learning_rate": 1.2076668993545491e-05, "loss": 0.1472479820251465, "step": 17100 }, { "epoch": 0.45070348333733945, "grad_norm": 3.556539535522461, "learning_rate": 1.206832285352244e-05, "loss": 0.16525906324386597, "step": 17110 }, { "epoch": 0.4509668985818382, "grad_norm": 1.639600157737732, "learning_rate": 1.2059975208084787e-05, "loss": 0.15425395965576172, "step": 17120 }, { "epoch": 0.45123031382633694, "grad_norm": 2.9474778175354004, "learning_rate": 1.2051626063308299e-05, "loss": 0.14884862899780274, "step": 17130 }, { "epoch": 0.4514937290708356, "grad_norm": 0.8923338651657104, "learning_rate": 1.2043275425269854e-05, "loss": 0.14607470035552977, "step": 17140 }, { "epoch": 0.45175714431533437, "grad_norm": 1.2357548475265503, "learning_rate": 1.2034923300047391e-05, "loss": 0.1496651887893677, "step": 17150 }, { "epoch": 0.4520205595598331, "grad_norm": 2.968639612197876, "learning_rate": 1.2026569693719963e-05, "loss": 0.1443983793258667, "step": 17160 }, { "epoch": 0.45228397480433186, "grad_norm": 1.2279843091964722, "learning_rate": 1.2018214612367673e-05, "loss": 0.14604599475860597, "step": 17170 }, { "epoch": 0.4525473900488306, "grad_norm": 1.3404622077941895, "learning_rate": 1.200985806207172e-05, "loss": 0.15193593502044678, "step": 17180 }, { "epoch": 0.45281080529332934, "grad_norm": 2.9767227172851562, "learning_rate": 1.2001500048914354e-05, "loss": 0.13879504203796386, "step": 17190 }, { "epoch": 0.4530742205378281, "grad_norm": 0.7766321897506714, "learning_rate": 1.1993140578978898e-05, "loss": 0.1566009998321533, "step": 17200 }, { "epoch": 0.45333763578232683, "grad_norm": 0.9427021741867065, "learning_rate": 1.1984779658349743e-05, "loss": 0.14897780418395995, "step": 17210 }, { "epoch": 0.4536010510268256, "grad_norm": 1.9569770097732544, "learning_rate": 1.1976417293112318e-05, "loss": 0.14055848121643066, "step": 17220 }, { "epoch": 0.45386446627132426, "grad_norm": 1.8961104154586792, "learning_rate": 1.1968053489353122e-05, "loss": 0.13999367952346803, "step": 17230 }, { "epoch": 0.454127881515823, "grad_norm": 2.8398401737213135, "learning_rate": 1.195968825315968e-05, "loss": 0.15664639472961425, "step": 17240 }, { "epoch": 0.45439129676032175, "grad_norm": 1.438865303993225, "learning_rate": 1.1951321590620583e-05, "loss": 0.1489412307739258, "step": 17250 }, { "epoch": 0.4546547120048205, "grad_norm": 3.137953042984009, "learning_rate": 1.1942953507825446e-05, "loss": 0.1518186092376709, "step": 17260 }, { "epoch": 0.45491812724931924, "grad_norm": 1.7225931882858276, "learning_rate": 1.1934584010864921e-05, "loss": 0.15331606864929198, "step": 17270 }, { "epoch": 0.455181542493818, "grad_norm": 0.9200688004493713, "learning_rate": 1.1926213105830683e-05, "loss": 0.1494528293609619, "step": 17280 }, { "epoch": 0.4554449577383167, "grad_norm": 1.6750571727752686, "learning_rate": 1.1917840798815448e-05, "loss": 0.16344165802001953, "step": 17290 }, { "epoch": 0.45570837298281547, "grad_norm": 1.3462178707122803, "learning_rate": 1.1909467095912932e-05, "loss": 0.15850260257720947, "step": 17300 }, { "epoch": 0.4559717882273142, "grad_norm": 3.8179421424865723, "learning_rate": 1.1901092003217885e-05, "loss": 0.15335993766784667, "step": 17310 }, { "epoch": 0.4562352034718129, "grad_norm": 1.628098726272583, "learning_rate": 1.1892715526826058e-05, "loss": 0.15867993831634522, "step": 17320 }, { "epoch": 0.45649861871631164, "grad_norm": 0.9509814977645874, "learning_rate": 1.1884337672834212e-05, "loss": 0.14082229137420654, "step": 17330 }, { "epoch": 0.4567620339608104, "grad_norm": 1.446303129196167, "learning_rate": 1.1875958447340108e-05, "loss": 0.13985121250152588, "step": 17340 }, { "epoch": 0.45702544920530913, "grad_norm": 3.4943900108337402, "learning_rate": 1.1867577856442514e-05, "loss": 0.14966773986816406, "step": 17350 }, { "epoch": 0.4572888644498079, "grad_norm": 1.8536168336868286, "learning_rate": 1.1859195906241182e-05, "loss": 0.15243176221847535, "step": 17360 }, { "epoch": 0.4575522796943066, "grad_norm": 5.024237632751465, "learning_rate": 1.1850812602836851e-05, "loss": 0.15628604888916015, "step": 17370 }, { "epoch": 0.45781569493880536, "grad_norm": 1.4698835611343384, "learning_rate": 1.1842427952331263e-05, "loss": 0.14327077865600585, "step": 17380 }, { "epoch": 0.4580791101833041, "grad_norm": 2.7152578830718994, "learning_rate": 1.183404196082712e-05, "loss": 0.1625860333442688, "step": 17390 }, { "epoch": 0.4583425254278028, "grad_norm": 1.329917073249817, "learning_rate": 1.1825654634428111e-05, "loss": 0.14229516983032225, "step": 17400 }, { "epoch": 0.45860594067230154, "grad_norm": 2.1527976989746094, "learning_rate": 1.1817265979238892e-05, "loss": 0.15544474124908447, "step": 17410 }, { "epoch": 0.4588693559168003, "grad_norm": 0.6782963275909424, "learning_rate": 1.1808876001365093e-05, "loss": 0.15680547952651977, "step": 17420 }, { "epoch": 0.459132771161299, "grad_norm": 2.761730670928955, "learning_rate": 1.1800484706913294e-05, "loss": 0.14986045360565187, "step": 17430 }, { "epoch": 0.45939618640579777, "grad_norm": 1.6894460916519165, "learning_rate": 1.1792092101991052e-05, "loss": 0.1497400403022766, "step": 17440 }, { "epoch": 0.4596596016502965, "grad_norm": 4.036172389984131, "learning_rate": 1.1783698192706856e-05, "loss": 0.15447641611099244, "step": 17450 }, { "epoch": 0.45992301689479526, "grad_norm": 2.0545763969421387, "learning_rate": 1.1775302985170161e-05, "loss": 0.150688099861145, "step": 17460 }, { "epoch": 0.460186432139294, "grad_norm": 1.3328825235366821, "learning_rate": 1.1766906485491361e-05, "loss": 0.1573621392250061, "step": 17470 }, { "epoch": 0.46044984738379274, "grad_norm": 1.4910085201263428, "learning_rate": 1.175850869978179e-05, "loss": 0.16070592403411865, "step": 17480 }, { "epoch": 0.46071326262829143, "grad_norm": 1.1707394123077393, "learning_rate": 1.1750109634153719e-05, "loss": 0.13951433897018434, "step": 17490 }, { "epoch": 0.4609766778727902, "grad_norm": 3.4571006298065186, "learning_rate": 1.1741709294720346e-05, "loss": 0.13785455226898194, "step": 17500 }, { "epoch": 0.4612400931172889, "grad_norm": 2.8531880378723145, "learning_rate": 1.173330768759581e-05, "loss": 0.1511009693145752, "step": 17510 }, { "epoch": 0.46150350836178766, "grad_norm": 1.9163192510604858, "learning_rate": 1.1724904818895158e-05, "loss": 0.14194061756134033, "step": 17520 }, { "epoch": 0.4617669236062864, "grad_norm": 1.6884974241256714, "learning_rate": 1.1716500694734359e-05, "loss": 0.15971513986587524, "step": 17530 }, { "epoch": 0.46203033885078515, "grad_norm": 1.244564414024353, "learning_rate": 1.1708095321230295e-05, "loss": 0.14340317249298096, "step": 17540 }, { "epoch": 0.4622937540952839, "grad_norm": 1.8520945310592651, "learning_rate": 1.1699688704500765e-05, "loss": 0.13762768507003784, "step": 17550 }, { "epoch": 0.46255716933978264, "grad_norm": 1.8706395626068115, "learning_rate": 1.1691280850664463e-05, "loss": 0.14655916690826415, "step": 17560 }, { "epoch": 0.4628205845842814, "grad_norm": 1.313513994216919, "learning_rate": 1.1682871765840996e-05, "loss": 0.15292127132415773, "step": 17570 }, { "epoch": 0.46308399982878007, "grad_norm": 2.3409857749938965, "learning_rate": 1.1674461456150845e-05, "loss": 0.15527639389038086, "step": 17580 }, { "epoch": 0.4633474150732788, "grad_norm": 3.4967844486236572, "learning_rate": 1.1666049927715408e-05, "loss": 0.1533466100692749, "step": 17590 }, { "epoch": 0.46361083031777756, "grad_norm": 1.4909512996673584, "learning_rate": 1.1657637186656952e-05, "loss": 0.13895961046218872, "step": 17600 }, { "epoch": 0.4638742455622763, "grad_norm": 0.92853844165802, "learning_rate": 1.1649223239098639e-05, "loss": 0.15322024822235109, "step": 17610 }, { "epoch": 0.46413766080677504, "grad_norm": 2.0991015434265137, "learning_rate": 1.1640808091164495e-05, "loss": 0.14418009519577027, "step": 17620 }, { "epoch": 0.4644010760512738, "grad_norm": 3.061347484588623, "learning_rate": 1.1632391748979435e-05, "loss": 0.145028018951416, "step": 17630 }, { "epoch": 0.46466449129577253, "grad_norm": 1.130261778831482, "learning_rate": 1.162397421866923e-05, "loss": 0.15289132595062255, "step": 17640 }, { "epoch": 0.4649279065402713, "grad_norm": 2.471585512161255, "learning_rate": 1.1615555506360526e-05, "loss": 0.15105606317520143, "step": 17650 }, { "epoch": 0.46519132178476996, "grad_norm": 2.459324598312378, "learning_rate": 1.1607135618180823e-05, "loss": 0.158202064037323, "step": 17660 }, { "epoch": 0.4654547370292687, "grad_norm": 2.1110057830810547, "learning_rate": 1.1598714560258477e-05, "loss": 0.15097283124923705, "step": 17670 }, { "epoch": 0.46571815227376745, "grad_norm": 1.738840103149414, "learning_rate": 1.1590292338722702e-05, "loss": 0.15506334304809571, "step": 17680 }, { "epoch": 0.4659815675182662, "grad_norm": 1.9595896005630493, "learning_rate": 1.1581868959703553e-05, "loss": 0.13926827907562256, "step": 17690 }, { "epoch": 0.46624498276276494, "grad_norm": 4.133377552032471, "learning_rate": 1.1573444429331926e-05, "loss": 0.14339203834533693, "step": 17700 }, { "epoch": 0.4665083980072637, "grad_norm": 1.5465182065963745, "learning_rate": 1.1565018753739559e-05, "loss": 0.14850131273269654, "step": 17710 }, { "epoch": 0.4667718132517624, "grad_norm": 1.1263813972473145, "learning_rate": 1.1556591939059024e-05, "loss": 0.14304622411727905, "step": 17720 }, { "epoch": 0.46703522849626117, "grad_norm": 3.4909861087799072, "learning_rate": 1.1548163991423716e-05, "loss": 0.14903321266174316, "step": 17730 }, { "epoch": 0.4672986437407599, "grad_norm": 1.6992733478546143, "learning_rate": 1.1539734916967864e-05, "loss": 0.14757784605026245, "step": 17740 }, { "epoch": 0.4675620589852586, "grad_norm": 0.9667653441429138, "learning_rate": 1.1531304721826506e-05, "loss": 0.13715872764587403, "step": 17750 }, { "epoch": 0.46782547422975734, "grad_norm": 2.650465250015259, "learning_rate": 1.1522873412135509e-05, "loss": 0.14797415733337402, "step": 17760 }, { "epoch": 0.4680888894742561, "grad_norm": 1.4438735246658325, "learning_rate": 1.1514440994031536e-05, "loss": 0.1434715747833252, "step": 17770 }, { "epoch": 0.46835230471875483, "grad_norm": 1.0159199237823486, "learning_rate": 1.1506007473652075e-05, "loss": 0.15052027702331544, "step": 17780 }, { "epoch": 0.4686157199632536, "grad_norm": 4.78308629989624, "learning_rate": 1.1497572857135397e-05, "loss": 0.1508150577545166, "step": 17790 }, { "epoch": 0.4688791352077523, "grad_norm": 3.5128257274627686, "learning_rate": 1.1489137150620582e-05, "loss": 0.1562565565109253, "step": 17800 }, { "epoch": 0.46914255045225106, "grad_norm": 1.1340678930282593, "learning_rate": 1.1480700360247504e-05, "loss": 0.15312700271606444, "step": 17810 }, { "epoch": 0.4694059656967498, "grad_norm": 1.8238115310668945, "learning_rate": 1.147226249215682e-05, "loss": 0.15652620792388916, "step": 17820 }, { "epoch": 0.46966938094124855, "grad_norm": 1.0575859546661377, "learning_rate": 1.1463823552489977e-05, "loss": 0.15420888662338256, "step": 17830 }, { "epoch": 0.46993279618574724, "grad_norm": 1.8895812034606934, "learning_rate": 1.1455383547389193e-05, "loss": 0.14628111124038695, "step": 17840 }, { "epoch": 0.470196211430246, "grad_norm": 1.5251905918121338, "learning_rate": 1.1446942482997479e-05, "loss": 0.13715728521347045, "step": 17850 }, { "epoch": 0.4704596266747447, "grad_norm": 0.9832710027694702, "learning_rate": 1.1438500365458593e-05, "loss": 0.14384706020355226, "step": 17860 }, { "epoch": 0.47072304191924347, "grad_norm": 1.536123514175415, "learning_rate": 1.1430057200917084e-05, "loss": 0.15606427192687988, "step": 17870 }, { "epoch": 0.4709864571637422, "grad_norm": 1.9826369285583496, "learning_rate": 1.1421612995518243e-05, "loss": 0.14592002630233764, "step": 17880 }, { "epoch": 0.47124987240824096, "grad_norm": 1.3019416332244873, "learning_rate": 1.1413167755408132e-05, "loss": 0.15666815042495727, "step": 17890 }, { "epoch": 0.4715132876527397, "grad_norm": 2.042876958847046, "learning_rate": 1.1404721486733558e-05, "loss": 0.15538184642791747, "step": 17900 }, { "epoch": 0.47177670289723844, "grad_norm": 1.7893743515014648, "learning_rate": 1.1396274195642085e-05, "loss": 0.14032083749771118, "step": 17910 }, { "epoch": 0.47204011814173713, "grad_norm": 1.3982996940612793, "learning_rate": 1.1387825888282006e-05, "loss": 0.15186548233032227, "step": 17920 }, { "epoch": 0.4723035333862359, "grad_norm": 1.5872682332992554, "learning_rate": 1.1379376570802373e-05, "loss": 0.13808106184005736, "step": 17930 }, { "epoch": 0.4725669486307346, "grad_norm": 1.0906713008880615, "learning_rate": 1.137092624935296e-05, "loss": 0.1482973575592041, "step": 17940 }, { "epoch": 0.47283036387523336, "grad_norm": 3.6784591674804688, "learning_rate": 1.1362474930084278e-05, "loss": 0.15731369256973265, "step": 17950 }, { "epoch": 0.4730937791197321, "grad_norm": 0.792427122592926, "learning_rate": 1.1354022619147557e-05, "loss": 0.13906947374343873, "step": 17960 }, { "epoch": 0.47335719436423085, "grad_norm": 1.2121319770812988, "learning_rate": 1.1345569322694756e-05, "loss": 0.15105795860290527, "step": 17970 }, { "epoch": 0.4736206096087296, "grad_norm": 2.784905433654785, "learning_rate": 1.133711504687855e-05, "loss": 0.15419318675994872, "step": 17980 }, { "epoch": 0.47388402485322834, "grad_norm": 1.142905354499817, "learning_rate": 1.1328659797852325e-05, "loss": 0.14162498712539673, "step": 17990 }, { "epoch": 0.4741474400977271, "grad_norm": 1.980897068977356, "learning_rate": 1.1320203581770177e-05, "loss": 0.15828737020492553, "step": 18000 }, { "epoch": 0.47441085534222577, "grad_norm": 4.201733589172363, "learning_rate": 1.1311746404786903e-05, "loss": 0.15529043674468995, "step": 18010 }, { "epoch": 0.4746742705867245, "grad_norm": 3.0601203441619873, "learning_rate": 1.1303288273058002e-05, "loss": 0.14564769268035888, "step": 18020 }, { "epoch": 0.47493768583122326, "grad_norm": 1.4950872659683228, "learning_rate": 1.1294829192739665e-05, "loss": 0.1466737747192383, "step": 18030 }, { "epoch": 0.475201101075722, "grad_norm": 0.9880825877189636, "learning_rate": 1.1286369169988782e-05, "loss": 0.1569310188293457, "step": 18040 }, { "epoch": 0.47546451632022074, "grad_norm": 1.8625187873840332, "learning_rate": 1.1277908210962916e-05, "loss": 0.15615841150283813, "step": 18050 }, { "epoch": 0.4757279315647195, "grad_norm": 2.874906063079834, "learning_rate": 1.126944632182032e-05, "loss": 0.15533580780029296, "step": 18060 }, { "epoch": 0.47599134680921823, "grad_norm": 1.2136099338531494, "learning_rate": 1.1260983508719922e-05, "loss": 0.1375659465789795, "step": 18070 }, { "epoch": 0.476254762053717, "grad_norm": 2.4155545234680176, "learning_rate": 1.1252519777821324e-05, "loss": 0.1480556011199951, "step": 18080 }, { "epoch": 0.4765181772982157, "grad_norm": 2.919327974319458, "learning_rate": 1.1244055135284795e-05, "loss": 0.14825891256332396, "step": 18090 }, { "epoch": 0.4767815925427144, "grad_norm": 1.5521396398544312, "learning_rate": 1.123558958727126e-05, "loss": 0.15398693084716797, "step": 18100 }, { "epoch": 0.47704500778721315, "grad_norm": 2.3654139041900635, "learning_rate": 1.1227123139942322e-05, "loss": 0.15224874019622803, "step": 18110 }, { "epoch": 0.4773084230317119, "grad_norm": 1.1736451387405396, "learning_rate": 1.1218655799460216e-05, "loss": 0.15073370933532715, "step": 18120 }, { "epoch": 0.47757183827621064, "grad_norm": 1.7723840475082397, "learning_rate": 1.1210187571987845e-05, "loss": 0.14814176559448242, "step": 18130 }, { "epoch": 0.4778352535207094, "grad_norm": 3.782442331314087, "learning_rate": 1.1201718463688745e-05, "loss": 0.16173114776611328, "step": 18140 }, { "epoch": 0.4780986687652081, "grad_norm": 1.6605783700942993, "learning_rate": 1.1193248480727101e-05, "loss": 0.14851503372192382, "step": 18150 }, { "epoch": 0.47836208400970687, "grad_norm": 1.452580213546753, "learning_rate": 1.118477762926773e-05, "loss": 0.13570823669433593, "step": 18160 }, { "epoch": 0.4786254992542056, "grad_norm": 1.7634947299957275, "learning_rate": 1.117630591547609e-05, "loss": 0.15321476459503175, "step": 18170 }, { "epoch": 0.4788889144987043, "grad_norm": 1.5684349536895752, "learning_rate": 1.1167833345518248e-05, "loss": 0.14010260105133057, "step": 18180 }, { "epoch": 0.47915232974320304, "grad_norm": 2.0537235736846924, "learning_rate": 1.1159359925560918e-05, "loss": 0.1504213809967041, "step": 18190 }, { "epoch": 0.4794157449877018, "grad_norm": 3.8816847801208496, "learning_rate": 1.115088566177141e-05, "loss": 0.14935474395751952, "step": 18200 }, { "epoch": 0.47967916023220053, "grad_norm": 1.8404109477996826, "learning_rate": 1.1142410560317669e-05, "loss": 0.15153253078460693, "step": 18210 }, { "epoch": 0.4799425754766993, "grad_norm": 1.6634432077407837, "learning_rate": 1.113393462736823e-05, "loss": 0.15887999534606934, "step": 18220 }, { "epoch": 0.480205990721198, "grad_norm": 1.2520396709442139, "learning_rate": 1.1125457869092251e-05, "loss": 0.1553717851638794, "step": 18230 }, { "epoch": 0.48046940596569676, "grad_norm": 1.0677928924560547, "learning_rate": 1.1116980291659474e-05, "loss": 0.15312449932098388, "step": 18240 }, { "epoch": 0.4807328212101955, "grad_norm": 2.086299419403076, "learning_rate": 1.1108501901240254e-05, "loss": 0.14098457098007203, "step": 18250 }, { "epoch": 0.48099623645469425, "grad_norm": 2.055757522583008, "learning_rate": 1.1100022704005524e-05, "loss": 0.14811468124389648, "step": 18260 }, { "epoch": 0.48125965169919294, "grad_norm": 3.8819217681884766, "learning_rate": 1.109154270612681e-05, "loss": 0.14931104183197022, "step": 18270 }, { "epoch": 0.4815230669436917, "grad_norm": 1.5263441801071167, "learning_rate": 1.1083061913776224e-05, "loss": 0.14118808507919312, "step": 18280 }, { "epoch": 0.4817864821881904, "grad_norm": 1.9673794507980347, "learning_rate": 1.1074580333126448e-05, "loss": 0.14240214824676514, "step": 18290 }, { "epoch": 0.48204989743268917, "grad_norm": 1.4254584312438965, "learning_rate": 1.1066097970350747e-05, "loss": 0.14740424156188964, "step": 18300 }, { "epoch": 0.4823133126771879, "grad_norm": 1.2202329635620117, "learning_rate": 1.1057614831622942e-05, "loss": 0.13788931369781493, "step": 18310 }, { "epoch": 0.48257672792168665, "grad_norm": 0.7324032783508301, "learning_rate": 1.1049130923117437e-05, "loss": 0.1461780309677124, "step": 18320 }, { "epoch": 0.4828401431661854, "grad_norm": 1.1829161643981934, "learning_rate": 1.1040646251009181e-05, "loss": 0.15613738298416138, "step": 18330 }, { "epoch": 0.48310355841068414, "grad_norm": 1.296500563621521, "learning_rate": 1.1032160821473692e-05, "loss": 0.15774736404418946, "step": 18340 }, { "epoch": 0.4833669736551829, "grad_norm": 3.5447306632995605, "learning_rate": 1.1023674640687019e-05, "loss": 0.1535026788711548, "step": 18350 }, { "epoch": 0.4836303888996816, "grad_norm": 0.8083765506744385, "learning_rate": 1.1015187714825782e-05, "loss": 0.14617033004760743, "step": 18360 }, { "epoch": 0.4838938041441803, "grad_norm": 0.8711820244789124, "learning_rate": 1.1006700050067122e-05, "loss": 0.13277046680450438, "step": 18370 }, { "epoch": 0.48415721938867906, "grad_norm": 1.7798388004302979, "learning_rate": 1.0998211652588737e-05, "loss": 0.14242184162139893, "step": 18380 }, { "epoch": 0.4844206346331778, "grad_norm": 2.284360885620117, "learning_rate": 1.0989722528568845e-05, "loss": 0.15653977394104004, "step": 18390 }, { "epoch": 0.48468404987767655, "grad_norm": 2.6032681465148926, "learning_rate": 1.0981232684186191e-05, "loss": 0.1583958625793457, "step": 18400 }, { "epoch": 0.4849474651221753, "grad_norm": 1.7215116024017334, "learning_rate": 1.0972742125620062e-05, "loss": 0.14430637359619142, "step": 18410 }, { "epoch": 0.48521088036667404, "grad_norm": 1.4732708930969238, "learning_rate": 1.0964250859050245e-05, "loss": 0.14465309381484986, "step": 18420 }, { "epoch": 0.4854742956111728, "grad_norm": 1.2183047533035278, "learning_rate": 1.0955758890657051e-05, "loss": 0.1398926019668579, "step": 18430 }, { "epoch": 0.48573771085567147, "grad_norm": 3.0126452445983887, "learning_rate": 1.0947266226621298e-05, "loss": 0.12756593227386476, "step": 18440 }, { "epoch": 0.4860011261001702, "grad_norm": 0.8885126113891602, "learning_rate": 1.0938772873124321e-05, "loss": 0.15579553842544555, "step": 18450 }, { "epoch": 0.48626454134466895, "grad_norm": 1.3476073741912842, "learning_rate": 1.0930278836347943e-05, "loss": 0.13780038356781005, "step": 18460 }, { "epoch": 0.4865279565891677, "grad_norm": 0.9061648845672607, "learning_rate": 1.0921784122474495e-05, "loss": 0.14520190954208373, "step": 18470 }, { "epoch": 0.48679137183366644, "grad_norm": 2.5007712841033936, "learning_rate": 1.0913288737686789e-05, "loss": 0.15873112678527831, "step": 18480 }, { "epoch": 0.4870547870781652, "grad_norm": 1.347628116607666, "learning_rate": 1.090479268816814e-05, "loss": 0.15681650638580322, "step": 18490 }, { "epoch": 0.48731820232266393, "grad_norm": 3.4196879863739014, "learning_rate": 1.0896295980102334e-05, "loss": 0.15255430936813355, "step": 18500 }, { "epoch": 0.4875816175671627, "grad_norm": 0.8679484724998474, "learning_rate": 1.0887798619673648e-05, "loss": 0.14275577068328857, "step": 18510 }, { "epoch": 0.4878450328116614, "grad_norm": 0.872025191783905, "learning_rate": 1.087930061306682e-05, "loss": 0.14769902229309081, "step": 18520 }, { "epoch": 0.4881084480561601, "grad_norm": 1.001449465751648, "learning_rate": 1.0870801966467073e-05, "loss": 0.14916601181030273, "step": 18530 }, { "epoch": 0.48837186330065885, "grad_norm": 1.2605620622634888, "learning_rate": 1.0862302686060082e-05, "loss": 0.14083054065704345, "step": 18540 }, { "epoch": 0.4886352785451576, "grad_norm": 2.972100019454956, "learning_rate": 1.0853802778031995e-05, "loss": 0.15204312801361083, "step": 18550 }, { "epoch": 0.48889869378965634, "grad_norm": 4.103074073791504, "learning_rate": 1.0845302248569411e-05, "loss": 0.1487704873085022, "step": 18560 }, { "epoch": 0.4891621090341551, "grad_norm": 2.0001602172851562, "learning_rate": 1.0836801103859377e-05, "loss": 0.15012904405593872, "step": 18570 }, { "epoch": 0.4894255242786538, "grad_norm": 1.2687581777572632, "learning_rate": 1.0828299350089401e-05, "loss": 0.13954813480377198, "step": 18580 }, { "epoch": 0.48968893952315257, "grad_norm": 1.5466644763946533, "learning_rate": 1.0819796993447422e-05, "loss": 0.15000382661819459, "step": 18590 }, { "epoch": 0.4899523547676513, "grad_norm": 1.4072458744049072, "learning_rate": 1.081129404012182e-05, "loss": 0.14852621555328369, "step": 18600 }, { "epoch": 0.49021577001215005, "grad_norm": 1.7601369619369507, "learning_rate": 1.0802790496301414e-05, "loss": 0.1500128149986267, "step": 18610 }, { "epoch": 0.49047918525664874, "grad_norm": 2.9029011726379395, "learning_rate": 1.079428636817545e-05, "loss": 0.15106464624404908, "step": 18620 }, { "epoch": 0.4907426005011475, "grad_norm": 0.6162680387496948, "learning_rate": 1.0785781661933596e-05, "loss": 0.13770226240158082, "step": 18630 }, { "epoch": 0.49100601574564623, "grad_norm": 0.8835515379905701, "learning_rate": 1.0777276383765953e-05, "loss": 0.1472805619239807, "step": 18640 }, { "epoch": 0.491269430990145, "grad_norm": 1.7909375429153442, "learning_rate": 1.0768770539863017e-05, "loss": 0.15940303802490235, "step": 18650 }, { "epoch": 0.4915328462346437, "grad_norm": 1.1798428297042847, "learning_rate": 1.076026413641572e-05, "loss": 0.1513391613960266, "step": 18660 }, { "epoch": 0.49179626147914246, "grad_norm": 2.4369900226593018, "learning_rate": 1.0751757179615378e-05, "loss": 0.16143847703933717, "step": 18670 }, { "epoch": 0.4920596767236412, "grad_norm": 3.5619142055511475, "learning_rate": 1.0743249675653733e-05, "loss": 0.150250244140625, "step": 18680 }, { "epoch": 0.49232309196813995, "grad_norm": 1.2104181051254272, "learning_rate": 1.0734741630722904e-05, "loss": 0.15016967058181763, "step": 18690 }, { "epoch": 0.49258650721263864, "grad_norm": 0.8815870881080627, "learning_rate": 1.0726233051015415e-05, "loss": 0.14395836591720582, "step": 18700 }, { "epoch": 0.4928499224571374, "grad_norm": 1.6377395391464233, "learning_rate": 1.0717723942724182e-05, "loss": 0.1373901844024658, "step": 18710 }, { "epoch": 0.4931133377016361, "grad_norm": 2.61862850189209, "learning_rate": 1.07092143120425e-05, "loss": 0.1429473876953125, "step": 18720 }, { "epoch": 0.49337675294613487, "grad_norm": 3.8500008583068848, "learning_rate": 1.0700704165164042e-05, "loss": 0.1438666820526123, "step": 18730 }, { "epoch": 0.4936401681906336, "grad_norm": 2.0458407402038574, "learning_rate": 1.069219350828286e-05, "loss": 0.15368828773498536, "step": 18740 }, { "epoch": 0.49390358343513235, "grad_norm": 1.0606168508529663, "learning_rate": 1.068368234759338e-05, "loss": 0.1415130615234375, "step": 18750 }, { "epoch": 0.4941669986796311, "grad_norm": 1.6496132612228394, "learning_rate": 1.0675170689290393e-05, "loss": 0.15449204444885253, "step": 18760 }, { "epoch": 0.49443041392412984, "grad_norm": 2.774550437927246, "learning_rate": 1.0666658539569048e-05, "loss": 0.14569132328033446, "step": 18770 }, { "epoch": 0.4946938291686286, "grad_norm": 2.5911879539489746, "learning_rate": 1.0658145904624852e-05, "loss": 0.1584687352180481, "step": 18780 }, { "epoch": 0.4949572444131273, "grad_norm": 1.9410018920898438, "learning_rate": 1.0649632790653677e-05, "loss": 0.1514638304710388, "step": 18790 }, { "epoch": 0.495220659657626, "grad_norm": 0.957203209400177, "learning_rate": 1.0641119203851724e-05, "loss": 0.16125142574310303, "step": 18800 }, { "epoch": 0.49548407490212476, "grad_norm": 1.8354356288909912, "learning_rate": 1.0632605150415561e-05, "loss": 0.15763838291168214, "step": 18810 }, { "epoch": 0.4957474901466235, "grad_norm": 1.7231494188308716, "learning_rate": 1.062409063654207e-05, "loss": 0.15820086002349854, "step": 18820 }, { "epoch": 0.49601090539112225, "grad_norm": 5.516703128814697, "learning_rate": 1.0615575668428494e-05, "loss": 0.14348316192626953, "step": 18830 }, { "epoch": 0.496274320635621, "grad_norm": 1.2780945301055908, "learning_rate": 1.0607060252272383e-05, "loss": 0.15029621124267578, "step": 18840 }, { "epoch": 0.49653773588011973, "grad_norm": 2.3022310733795166, "learning_rate": 1.0598544394271633e-05, "loss": 0.13349947929382325, "step": 18850 }, { "epoch": 0.4968011511246185, "grad_norm": 1.3398736715316772, "learning_rate": 1.0590028100624448e-05, "loss": 0.14757699966430665, "step": 18860 }, { "epoch": 0.4970645663691172, "grad_norm": 3.2989799976348877, "learning_rate": 1.0581511377529354e-05, "loss": 0.15610888004302978, "step": 18870 }, { "epoch": 0.4973279816136159, "grad_norm": 1.1222224235534668, "learning_rate": 1.0572994231185194e-05, "loss": 0.15859498977661132, "step": 18880 }, { "epoch": 0.49759139685811465, "grad_norm": 0.8748815059661865, "learning_rate": 1.056447666779111e-05, "loss": 0.14748927354812622, "step": 18890 }, { "epoch": 0.4978548121026134, "grad_norm": 1.0314973592758179, "learning_rate": 1.0555958693546552e-05, "loss": 0.14599355459213256, "step": 18900 }, { "epoch": 0.49811822734711214, "grad_norm": 1.7011762857437134, "learning_rate": 1.0547440314651274e-05, "loss": 0.13399827480316162, "step": 18910 }, { "epoch": 0.4983816425916109, "grad_norm": 1.8207489252090454, "learning_rate": 1.0538921537305315e-05, "loss": 0.15630850791931153, "step": 18920 }, { "epoch": 0.49864505783610963, "grad_norm": 2.344480514526367, "learning_rate": 1.0530402367709007e-05, "loss": 0.1465930700302124, "step": 18930 }, { "epoch": 0.4989084730806084, "grad_norm": 3.1001288890838623, "learning_rate": 1.0521882812062978e-05, "loss": 0.14682961702346803, "step": 18940 }, { "epoch": 0.4991718883251071, "grad_norm": 1.2954356670379639, "learning_rate": 1.0513362876568118e-05, "loss": 0.14684603214263917, "step": 18950 }, { "epoch": 0.49943530356960586, "grad_norm": 2.8522756099700928, "learning_rate": 1.0504842567425611e-05, "loss": 0.1498933792114258, "step": 18960 }, { "epoch": 0.49969871881410455, "grad_norm": 1.1009010076522827, "learning_rate": 1.0496321890836901e-05, "loss": 0.1463375687599182, "step": 18970 }, { "epoch": 0.4999621340586033, "grad_norm": 1.7055238485336304, "learning_rate": 1.0487800853003706e-05, "loss": 0.15278724431991578, "step": 18980 }, { "epoch": 0.500225549303102, "grad_norm": 2.5821444988250732, "learning_rate": 1.0479279460128005e-05, "loss": 0.15081819295883178, "step": 18990 }, { "epoch": 0.5004889645476008, "grad_norm": 2.296334981918335, "learning_rate": 1.0470757718412035e-05, "loss": 0.14137314558029174, "step": 19000 }, { "epoch": 0.5007523797920995, "grad_norm": 1.3831411600112915, "learning_rate": 1.0462235634058287e-05, "loss": 0.14380195140838622, "step": 19010 }, { "epoch": 0.5010157950365982, "grad_norm": 1.1968296766281128, "learning_rate": 1.0453713213269503e-05, "loss": 0.15122706890106202, "step": 19020 }, { "epoch": 0.501279210281097, "grad_norm": 1.5785256624221802, "learning_rate": 1.0445190462248667e-05, "loss": 0.14160070419311524, "step": 19030 }, { "epoch": 0.5015426255255957, "grad_norm": 2.7329204082489014, "learning_rate": 1.0436667387199004e-05, "loss": 0.15293660163879394, "step": 19040 }, { "epoch": 0.5018060407700945, "grad_norm": 2.1165730953216553, "learning_rate": 1.0428143994323978e-05, "loss": 0.15318228006362916, "step": 19050 }, { "epoch": 0.5020694560145932, "grad_norm": 1.6908777952194214, "learning_rate": 1.041962028982728e-05, "loss": 0.15033068656921386, "step": 19060 }, { "epoch": 0.502332871259092, "grad_norm": 2.910203218460083, "learning_rate": 1.0411096279912835e-05, "loss": 0.16284884214401246, "step": 19070 }, { "epoch": 0.5025962865035907, "grad_norm": 1.4251660108566284, "learning_rate": 1.0402571970784778e-05, "loss": 0.14858992099761964, "step": 19080 }, { "epoch": 0.5028597017480895, "grad_norm": 1.441221833229065, "learning_rate": 1.0394047368647473e-05, "loss": 0.13931167125701904, "step": 19090 }, { "epoch": 0.5031231169925882, "grad_norm": 2.526003837585449, "learning_rate": 1.0385522479705493e-05, "loss": 0.14409732818603516, "step": 19100 }, { "epoch": 0.5033865322370868, "grad_norm": 0.8512628078460693, "learning_rate": 1.0376997310163622e-05, "loss": 0.1578314185142517, "step": 19110 }, { "epoch": 0.5036499474815856, "grad_norm": 1.6929161548614502, "learning_rate": 1.0368471866226837e-05, "loss": 0.13998717069625854, "step": 19120 }, { "epoch": 0.5039133627260843, "grad_norm": 4.7928690910339355, "learning_rate": 1.0359946154100336e-05, "loss": 0.14840054512023926, "step": 19130 }, { "epoch": 0.5041767779705831, "grad_norm": 1.3837658166885376, "learning_rate": 1.035142017998949e-05, "loss": 0.14901335239410402, "step": 19140 }, { "epoch": 0.5044401932150818, "grad_norm": 1.2238867282867432, "learning_rate": 1.0342893950099879e-05, "loss": 0.14921576976776124, "step": 19150 }, { "epoch": 0.5047036084595806, "grad_norm": 1.0269334316253662, "learning_rate": 1.0334367470637257e-05, "loss": 0.14461692571640014, "step": 19160 }, { "epoch": 0.5049670237040793, "grad_norm": 0.8911460041999817, "learning_rate": 1.0325840747807562e-05, "loss": 0.14486613273620605, "step": 19170 }, { "epoch": 0.505230438948578, "grad_norm": 1.613764762878418, "learning_rate": 1.0317313787816915e-05, "loss": 0.1511072874069214, "step": 19180 }, { "epoch": 0.5054938541930768, "grad_norm": 2.213819980621338, "learning_rate": 1.0308786596871606e-05, "loss": 0.1553111791610718, "step": 19190 }, { "epoch": 0.5057572694375755, "grad_norm": 1.493815302848816, "learning_rate": 1.0300259181178086e-05, "loss": 0.14763203859329224, "step": 19200 }, { "epoch": 0.5060206846820743, "grad_norm": 1.3102303743362427, "learning_rate": 1.0291731546942977e-05, "loss": 0.13903203010559081, "step": 19210 }, { "epoch": 0.506284099926573, "grad_norm": 1.6567535400390625, "learning_rate": 1.0283203700373065e-05, "loss": 0.14065907001495362, "step": 19220 }, { "epoch": 0.5065475151710718, "grad_norm": 0.8574995398521423, "learning_rate": 1.0274675647675281e-05, "loss": 0.15614211559295654, "step": 19230 }, { "epoch": 0.5068109304155705, "grad_norm": 2.707500457763672, "learning_rate": 1.0266147395056712e-05, "loss": 0.14740206003189088, "step": 19240 }, { "epoch": 0.5070743456600693, "grad_norm": 2.0802040100097656, "learning_rate": 1.0257618948724583e-05, "loss": 0.14689288139343262, "step": 19250 }, { "epoch": 0.507337760904568, "grad_norm": 1.9342554807662964, "learning_rate": 1.024909031488627e-05, "loss": 0.15543537139892577, "step": 19260 }, { "epoch": 0.5076011761490666, "grad_norm": 2.083660125732422, "learning_rate": 1.0240561499749278e-05, "loss": 0.13417415618896483, "step": 19270 }, { "epoch": 0.5078645913935654, "grad_norm": 0.7762096524238586, "learning_rate": 1.023203250952125e-05, "loss": 0.1427147388458252, "step": 19280 }, { "epoch": 0.5081280066380641, "grad_norm": 1.7857309579849243, "learning_rate": 1.0223503350409949e-05, "loss": 0.14666883945465087, "step": 19290 }, { "epoch": 0.5083914218825629, "grad_norm": 1.404138207435608, "learning_rate": 1.0214974028623268e-05, "loss": 0.15481972694396973, "step": 19300 }, { "epoch": 0.5086548371270616, "grad_norm": 1.2745575904846191, "learning_rate": 1.0206444550369212e-05, "loss": 0.15805697441101074, "step": 19310 }, { "epoch": 0.5089182523715604, "grad_norm": 1.3005937337875366, "learning_rate": 1.0197914921855909e-05, "loss": 0.14839000701904298, "step": 19320 }, { "epoch": 0.5091816676160591, "grad_norm": 2.3447771072387695, "learning_rate": 1.0189385149291586e-05, "loss": 0.15499320030212402, "step": 19330 }, { "epoch": 0.5094450828605579, "grad_norm": 1.3091232776641846, "learning_rate": 1.018085523888458e-05, "loss": 0.1433521628379822, "step": 19340 }, { "epoch": 0.5097084981050566, "grad_norm": 3.2911200523376465, "learning_rate": 1.0172325196843328e-05, "loss": 0.14603545665740966, "step": 19350 }, { "epoch": 0.5099719133495553, "grad_norm": 1.5031732320785522, "learning_rate": 1.0163795029376362e-05, "loss": 0.14106171131134032, "step": 19360 }, { "epoch": 0.5102353285940541, "grad_norm": 1.295548915863037, "learning_rate": 1.015526474269231e-05, "loss": 0.14711834192276002, "step": 19370 }, { "epoch": 0.5104987438385528, "grad_norm": 2.5744686126708984, "learning_rate": 1.0146734342999875e-05, "loss": 0.14206490516662598, "step": 19380 }, { "epoch": 0.5107621590830516, "grad_norm": 2.0234646797180176, "learning_rate": 1.0138203836507859e-05, "loss": 0.1584465503692627, "step": 19390 }, { "epoch": 0.5110255743275502, "grad_norm": 1.376537799835205, "learning_rate": 1.0129673229425125e-05, "loss": 0.1537250757217407, "step": 19400 }, { "epoch": 0.511288989572049, "grad_norm": 2.566356897354126, "learning_rate": 1.0121142527960624e-05, "loss": 0.1371258020401001, "step": 19410 }, { "epoch": 0.5115524048165477, "grad_norm": 2.764714002609253, "learning_rate": 1.0112611738323362e-05, "loss": 0.13787879943847656, "step": 19420 }, { "epoch": 0.5118158200610465, "grad_norm": 2.147087335586548, "learning_rate": 1.0104080866722419e-05, "loss": 0.14009652137756348, "step": 19430 }, { "epoch": 0.5120792353055452, "grad_norm": 2.6052086353302, "learning_rate": 1.0095549919366928e-05, "loss": 0.136492121219635, "step": 19440 }, { "epoch": 0.5123426505500439, "grad_norm": 1.5396333932876587, "learning_rate": 1.0087018902466086e-05, "loss": 0.160950231552124, "step": 19450 }, { "epoch": 0.5126060657945427, "grad_norm": 3.300063371658325, "learning_rate": 1.0078487822229132e-05, "loss": 0.15649837255477905, "step": 19460 }, { "epoch": 0.5128694810390414, "grad_norm": 1.2869011163711548, "learning_rate": 1.0069956684865354e-05, "loss": 0.15008795261383057, "step": 19470 }, { "epoch": 0.5131328962835402, "grad_norm": 2.1325693130493164, "learning_rate": 1.0061425496584084e-05, "loss": 0.13719067573547364, "step": 19480 }, { "epoch": 0.5133963115280389, "grad_norm": 2.946072578430176, "learning_rate": 1.0052894263594687e-05, "loss": 0.1261027693748474, "step": 19490 }, { "epoch": 0.5136597267725377, "grad_norm": 1.1933695077896118, "learning_rate": 1.0044362992106563e-05, "loss": 0.13904714584350586, "step": 19500 }, { "epoch": 0.5139231420170364, "grad_norm": 1.5769648551940918, "learning_rate": 1.0035831688329139e-05, "loss": 0.15225353240966796, "step": 19510 }, { "epoch": 0.5141865572615352, "grad_norm": 1.5696725845336914, "learning_rate": 1.0027300358471865e-05, "loss": 0.15327017307281493, "step": 19520 }, { "epoch": 0.5144499725060339, "grad_norm": 2.2455971240997314, "learning_rate": 1.0018769008744214e-05, "loss": 0.15598268508911134, "step": 19530 }, { "epoch": 0.5147133877505325, "grad_norm": 1.2916336059570312, "learning_rate": 1.0010237645355669e-05, "loss": 0.14737370014190673, "step": 19540 }, { "epoch": 0.5149768029950313, "grad_norm": 1.6060940027236938, "learning_rate": 1.0001706274515724e-05, "loss": 0.15535176992416383, "step": 19550 }, { "epoch": 0.51524021823953, "grad_norm": 2.3548643589019775, "learning_rate": 9.993174902433877e-06, "loss": 0.1397170066833496, "step": 19560 }, { "epoch": 0.5155036334840288, "grad_norm": 1.138669729232788, "learning_rate": 9.984643535319632e-06, "loss": 0.1475850224494934, "step": 19570 }, { "epoch": 0.5157670487285275, "grad_norm": 2.8895747661590576, "learning_rate": 9.976112179382483e-06, "loss": 0.15294718742370605, "step": 19580 }, { "epoch": 0.5160304639730263, "grad_norm": 1.2594094276428223, "learning_rate": 9.967580840831922e-06, "loss": 0.15314366817474365, "step": 19590 }, { "epoch": 0.516293879217525, "grad_norm": 1.462028980255127, "learning_rate": 9.959049525877425e-06, "loss": 0.15053083896636962, "step": 19600 }, { "epoch": 0.5165572944620238, "grad_norm": 2.9741766452789307, "learning_rate": 9.950518240728446e-06, "loss": 0.15493392944335938, "step": 19610 }, { "epoch": 0.5168207097065225, "grad_norm": 0.9846689105033875, "learning_rate": 9.94198699159443e-06, "loss": 0.1459495782852173, "step": 19620 }, { "epoch": 0.5170841249510212, "grad_norm": 1.4399325847625732, "learning_rate": 9.93345578468478e-06, "loss": 0.1534660816192627, "step": 19630 }, { "epoch": 0.51734754019552, "grad_norm": 1.6225000619888306, "learning_rate": 9.924924626208884e-06, "loss": 0.14182379245758056, "step": 19640 }, { "epoch": 0.5176109554400187, "grad_norm": 1.2644728422164917, "learning_rate": 9.916393522376087e-06, "loss": 0.14891959428787233, "step": 19650 }, { "epoch": 0.5178743706845175, "grad_norm": 2.0562632083892822, "learning_rate": 9.907862479395684e-06, "loss": 0.1478731393814087, "step": 19660 }, { "epoch": 0.5181377859290162, "grad_norm": 2.132044792175293, "learning_rate": 9.899331503476951e-06, "loss": 0.15217673778533936, "step": 19670 }, { "epoch": 0.518401201173515, "grad_norm": 1.3713669776916504, "learning_rate": 9.890800600829086e-06, "loss": 0.1363386631011963, "step": 19680 }, { "epoch": 0.5186646164180136, "grad_norm": 2.216437816619873, "learning_rate": 9.882269777661261e-06, "loss": 0.13815715312957763, "step": 19690 }, { "epoch": 0.5189280316625123, "grad_norm": 1.3409593105316162, "learning_rate": 9.873739040182569e-06, "loss": 0.14492301940917968, "step": 19700 }, { "epoch": 0.5191914469070111, "grad_norm": 1.8516956567764282, "learning_rate": 9.86520839460205e-06, "loss": 0.15929148197174073, "step": 19710 }, { "epoch": 0.5194548621515098, "grad_norm": 3.830980062484741, "learning_rate": 9.856677847128678e-06, "loss": 0.14092130661010743, "step": 19720 }, { "epoch": 0.5197182773960086, "grad_norm": 1.0137544870376587, "learning_rate": 9.848147403971353e-06, "loss": 0.1399618148803711, "step": 19730 }, { "epoch": 0.5199816926405073, "grad_norm": 2.5912046432495117, "learning_rate": 9.839617071338897e-06, "loss": 0.14828503131866455, "step": 19740 }, { "epoch": 0.5202451078850061, "grad_norm": 1.8286877870559692, "learning_rate": 9.831086855440056e-06, "loss": 0.15824906826019286, "step": 19750 }, { "epoch": 0.5205085231295048, "grad_norm": 1.6712734699249268, "learning_rate": 9.82255676248349e-06, "loss": 0.1399045467376709, "step": 19760 }, { "epoch": 0.5207719383740036, "grad_norm": 2.844741106033325, "learning_rate": 9.814026798677767e-06, "loss": 0.13921115398406983, "step": 19770 }, { "epoch": 0.5210353536185023, "grad_norm": 1.1526365280151367, "learning_rate": 9.805496970231367e-06, "loss": 0.14069212675094606, "step": 19780 }, { "epoch": 0.521298768863001, "grad_norm": 0.934292733669281, "learning_rate": 9.796967283352657e-06, "loss": 0.13650329113006593, "step": 19790 }, { "epoch": 0.5215621841074998, "grad_norm": 1.5175528526306152, "learning_rate": 9.788437744249925e-06, "loss": 0.14219241142272948, "step": 19800 }, { "epoch": 0.5218255993519985, "grad_norm": 2.402474880218506, "learning_rate": 9.779908359131327e-06, "loss": 0.14422991275787353, "step": 19810 }, { "epoch": 0.5220890145964973, "grad_norm": 1.1302165985107422, "learning_rate": 9.77137913420492e-06, "loss": 0.13654613494873047, "step": 19820 }, { "epoch": 0.522352429840996, "grad_norm": 1.3850123882293701, "learning_rate": 9.762850075678644e-06, "loss": 0.15341765880584718, "step": 19830 }, { "epoch": 0.5226158450854947, "grad_norm": 2.8786439895629883, "learning_rate": 9.754321189760312e-06, "loss": 0.14083011150360109, "step": 19840 }, { "epoch": 0.5228792603299934, "grad_norm": 1.1039745807647705, "learning_rate": 9.74579248265762e-06, "loss": 0.13715198040008544, "step": 19850 }, { "epoch": 0.5231426755744922, "grad_norm": 0.8777554631233215, "learning_rate": 9.737263960578127e-06, "loss": 0.14758198261260985, "step": 19860 }, { "epoch": 0.5234060908189909, "grad_norm": 1.5226448774337769, "learning_rate": 9.728735629729256e-06, "loss": 0.14914212226867676, "step": 19870 }, { "epoch": 0.5236695060634896, "grad_norm": 1.929233431816101, "learning_rate": 9.7202074963183e-06, "loss": 0.14758516550064088, "step": 19880 }, { "epoch": 0.5239329213079884, "grad_norm": 1.764902114868164, "learning_rate": 9.711679566552396e-06, "loss": 0.14065263271331788, "step": 19890 }, { "epoch": 0.5241963365524871, "grad_norm": 1.9090557098388672, "learning_rate": 9.703151846638545e-06, "loss": 0.14486370086669922, "step": 19900 }, { "epoch": 0.5244597517969859, "grad_norm": 1.574471116065979, "learning_rate": 9.694624342783591e-06, "loss": 0.15089013576507568, "step": 19910 }, { "epoch": 0.5247231670414846, "grad_norm": 1.3782514333724976, "learning_rate": 9.686097061194209e-06, "loss": 0.1588992714881897, "step": 19920 }, { "epoch": 0.5249865822859834, "grad_norm": 2.3547823429107666, "learning_rate": 9.677570008076933e-06, "loss": 0.1435283064842224, "step": 19930 }, { "epoch": 0.5252499975304821, "grad_norm": 1.50046968460083, "learning_rate": 9.669043189638115e-06, "loss": 0.14338748455047606, "step": 19940 }, { "epoch": 0.5255134127749809, "grad_norm": 1.9057642221450806, "learning_rate": 9.66051661208394e-06, "loss": 0.16099207401275634, "step": 19950 }, { "epoch": 0.5257768280194796, "grad_norm": 1.1039315462112427, "learning_rate": 9.651990281620423e-06, "loss": 0.13748084306716918, "step": 19960 }, { "epoch": 0.5260402432639782, "grad_norm": 3.9484622478485107, "learning_rate": 9.643464204453392e-06, "loss": 0.14291170835494996, "step": 19970 }, { "epoch": 0.526303658508477, "grad_norm": 2.9198038578033447, "learning_rate": 9.634938386788497e-06, "loss": 0.15875372886657715, "step": 19980 }, { "epoch": 0.5265670737529757, "grad_norm": 2.063274621963501, "learning_rate": 9.626412834831193e-06, "loss": 0.15551395416259767, "step": 19990 }, { "epoch": 0.5268304889974745, "grad_norm": 2.846961498260498, "learning_rate": 9.617887554786746e-06, "loss": 0.13903977870941162, "step": 20000 }, { "epoch": 0.5270939042419732, "grad_norm": 2.1181421279907227, "learning_rate": 9.609362552860225e-06, "loss": 0.13668001890182496, "step": 20010 }, { "epoch": 0.527357319486472, "grad_norm": 2.570470094680786, "learning_rate": 9.600837835256486e-06, "loss": 0.14672960042953492, "step": 20020 }, { "epoch": 0.5276207347309707, "grad_norm": 2.425945281982422, "learning_rate": 9.592313408180199e-06, "loss": 0.13038918972015381, "step": 20030 }, { "epoch": 0.5278841499754695, "grad_norm": 5.163753986358643, "learning_rate": 9.583789277835798e-06, "loss": 0.15892621278762817, "step": 20040 }, { "epoch": 0.5281475652199682, "grad_norm": 1.1371424198150635, "learning_rate": 9.575265450427517e-06, "loss": 0.1532713532447815, "step": 20050 }, { "epoch": 0.5284109804644669, "grad_norm": 1.027183175086975, "learning_rate": 9.566741932159369e-06, "loss": 0.15664281845092773, "step": 20060 }, { "epoch": 0.5286743957089657, "grad_norm": 1.2192305326461792, "learning_rate": 9.558218729235132e-06, "loss": 0.1559069871902466, "step": 20070 }, { "epoch": 0.5289378109534644, "grad_norm": 4.024788856506348, "learning_rate": 9.549695847858363e-06, "loss": 0.15106983184814454, "step": 20080 }, { "epoch": 0.5292012261979632, "grad_norm": 1.3952815532684326, "learning_rate": 9.541173294232387e-06, "loss": 0.15476707220077515, "step": 20090 }, { "epoch": 0.5294646414424619, "grad_norm": 1.2210808992385864, "learning_rate": 9.53265107456028e-06, "loss": 0.14874216318130493, "step": 20100 }, { "epoch": 0.5297280566869607, "grad_norm": 1.079765796661377, "learning_rate": 9.524129195044886e-06, "loss": 0.13327379226684571, "step": 20110 }, { "epoch": 0.5299914719314593, "grad_norm": 3.8039815425872803, "learning_rate": 9.515607661888799e-06, "loss": 0.1516256809234619, "step": 20120 }, { "epoch": 0.5302548871759581, "grad_norm": 2.3745832443237305, "learning_rate": 9.507086481294345e-06, "loss": 0.16660836935043336, "step": 20130 }, { "epoch": 0.5305183024204568, "grad_norm": 0.8721898794174194, "learning_rate": 9.498565659463624e-06, "loss": 0.13887007236480714, "step": 20140 }, { "epoch": 0.5307817176649555, "grad_norm": 0.9482285380363464, "learning_rate": 9.490045202598442e-06, "loss": 0.13912563323974608, "step": 20150 }, { "epoch": 0.5310451329094543, "grad_norm": 1.2928142547607422, "learning_rate": 9.48152511690037e-06, "loss": 0.15345723628997804, "step": 20160 }, { "epoch": 0.531308548153953, "grad_norm": 2.7951371669769287, "learning_rate": 9.473005408570685e-06, "loss": 0.1434560775756836, "step": 20170 }, { "epoch": 0.5315719633984518, "grad_norm": 1.729980230331421, "learning_rate": 9.464486083810397e-06, "loss": 0.13917311429977416, "step": 20180 }, { "epoch": 0.5318353786429505, "grad_norm": 1.6099958419799805, "learning_rate": 9.455967148820243e-06, "loss": 0.14360905885696412, "step": 20190 }, { "epoch": 0.5320987938874493, "grad_norm": 3.5707077980041504, "learning_rate": 9.44744860980067e-06, "loss": 0.15448061227798462, "step": 20200 }, { "epoch": 0.532362209131948, "grad_norm": 1.3288524150848389, "learning_rate": 9.438930472951838e-06, "loss": 0.13376864194869995, "step": 20210 }, { "epoch": 0.5326256243764467, "grad_norm": 1.4450596570968628, "learning_rate": 9.430412744473615e-06, "loss": 0.13862965106964112, "step": 20220 }, { "epoch": 0.5328890396209455, "grad_norm": 1.5008383989334106, "learning_rate": 9.421895430565568e-06, "loss": 0.1452367424964905, "step": 20230 }, { "epoch": 0.5331524548654442, "grad_norm": 1.8925951719284058, "learning_rate": 9.413378537426972e-06, "loss": 0.14189867973327636, "step": 20240 }, { "epoch": 0.533415870109943, "grad_norm": 2.021700143814087, "learning_rate": 9.404862071256785e-06, "loss": 0.14434316158294677, "step": 20250 }, { "epoch": 0.5336792853544416, "grad_norm": 2.1280877590179443, "learning_rate": 9.396346038253654e-06, "loss": 0.1527738094329834, "step": 20260 }, { "epoch": 0.5339427005989404, "grad_norm": 4.1475019454956055, "learning_rate": 9.387830444615926e-06, "loss": 0.13773940801620482, "step": 20270 }, { "epoch": 0.5342061158434391, "grad_norm": 4.258584022521973, "learning_rate": 9.379315296541611e-06, "loss": 0.13826172351837157, "step": 20280 }, { "epoch": 0.5344695310879379, "grad_norm": 1.647410273551941, "learning_rate": 9.370800600228399e-06, "loss": 0.15303510427474976, "step": 20290 }, { "epoch": 0.5347329463324366, "grad_norm": 0.8665757179260254, "learning_rate": 9.362286361873658e-06, "loss": 0.14823758602142334, "step": 20300 }, { "epoch": 0.5349963615769353, "grad_norm": 1.6981199979782104, "learning_rate": 9.353772587674413e-06, "loss": 0.14973682165145874, "step": 20310 }, { "epoch": 0.5352597768214341, "grad_norm": 1.425737977027893, "learning_rate": 9.345259283827362e-06, "loss": 0.1404339075088501, "step": 20320 }, { "epoch": 0.5355231920659328, "grad_norm": 2.1598215103149414, "learning_rate": 9.336746456528853e-06, "loss": 0.1474231958389282, "step": 20330 }, { "epoch": 0.5357866073104316, "grad_norm": 1.2057077884674072, "learning_rate": 9.328234111974885e-06, "loss": 0.1387540102005005, "step": 20340 }, { "epoch": 0.5360500225549303, "grad_norm": 1.6868776082992554, "learning_rate": 9.319722256361117e-06, "loss": 0.15921127796173096, "step": 20350 }, { "epoch": 0.5363134377994291, "grad_norm": 4.724923133850098, "learning_rate": 9.311210895882835e-06, "loss": 0.14425122737884521, "step": 20360 }, { "epoch": 0.5365768530439278, "grad_norm": 2.165400505065918, "learning_rate": 9.302700036734985e-06, "loss": 0.14418394565582277, "step": 20370 }, { "epoch": 0.5368402682884266, "grad_norm": 1.1999578475952148, "learning_rate": 9.294189685112133e-06, "loss": 0.15139025449752808, "step": 20380 }, { "epoch": 0.5371036835329253, "grad_norm": 1.3079744577407837, "learning_rate": 9.285679847208474e-06, "loss": 0.1304416060447693, "step": 20390 }, { "epoch": 0.537367098777424, "grad_norm": 1.5684911012649536, "learning_rate": 9.277170529217847e-06, "loss": 0.15356409549713135, "step": 20400 }, { "epoch": 0.5376305140219227, "grad_norm": 1.0099949836730957, "learning_rate": 9.268661737333691e-06, "loss": 0.14221857786178588, "step": 20410 }, { "epoch": 0.5378939292664214, "grad_norm": 1.498012900352478, "learning_rate": 9.260153477749072e-06, "loss": 0.14058010578155516, "step": 20420 }, { "epoch": 0.5381573445109202, "grad_norm": 2.3070154190063477, "learning_rate": 9.251645756656676e-06, "loss": 0.15172650814056396, "step": 20430 }, { "epoch": 0.5384207597554189, "grad_norm": 1.8467692136764526, "learning_rate": 9.243138580248782e-06, "loss": 0.13658182621002196, "step": 20440 }, { "epoch": 0.5386841749999177, "grad_norm": 3.9503414630889893, "learning_rate": 9.234631954717285e-06, "loss": 0.14964704513549804, "step": 20450 }, { "epoch": 0.5389475902444164, "grad_norm": 2.938934803009033, "learning_rate": 9.226125886253673e-06, "loss": 0.1488774061203003, "step": 20460 }, { "epoch": 0.5392110054889152, "grad_norm": 0.9103949069976807, "learning_rate": 9.217620381049026e-06, "loss": 0.14805376529693604, "step": 20470 }, { "epoch": 0.5394744207334139, "grad_norm": 1.4039576053619385, "learning_rate": 9.209115445294022e-06, "loss": 0.14412531852722169, "step": 20480 }, { "epoch": 0.5397378359779126, "grad_norm": 0.7360811233520508, "learning_rate": 9.200611085178917e-06, "loss": 0.13909471035003662, "step": 20490 }, { "epoch": 0.5400012512224114, "grad_norm": 1.4395889043807983, "learning_rate": 9.192107306893556e-06, "loss": 0.1426942229270935, "step": 20500 }, { "epoch": 0.5402646664669101, "grad_norm": 4.433235168457031, "learning_rate": 9.183604116627354e-06, "loss": 0.1551394581794739, "step": 20510 }, { "epoch": 0.5405280817114089, "grad_norm": 1.4808717966079712, "learning_rate": 9.175101520569293e-06, "loss": 0.15379025936126708, "step": 20520 }, { "epoch": 0.5407914969559076, "grad_norm": 2.053586959838867, "learning_rate": 9.166599524907937e-06, "loss": 0.14754023551940917, "step": 20530 }, { "epoch": 0.5410549122004064, "grad_norm": 2.477853536605835, "learning_rate": 9.158098135831404e-06, "loss": 0.1490910530090332, "step": 20540 }, { "epoch": 0.541318327444905, "grad_norm": 1.4300862550735474, "learning_rate": 9.149597359527368e-06, "loss": 0.1458722710609436, "step": 20550 }, { "epoch": 0.5415817426894038, "grad_norm": 0.8753971457481384, "learning_rate": 9.141097202183063e-06, "loss": 0.1451590895652771, "step": 20560 }, { "epoch": 0.5418451579339025, "grad_norm": 1.9063854217529297, "learning_rate": 9.13259766998527e-06, "loss": 0.14207327365875244, "step": 20570 }, { "epoch": 0.5421085731784012, "grad_norm": 2.634308099746704, "learning_rate": 9.124098769120318e-06, "loss": 0.14396668672561647, "step": 20580 }, { "epoch": 0.5423719884229, "grad_norm": 1.1063477993011475, "learning_rate": 9.115600505774072e-06, "loss": 0.1519090414047241, "step": 20590 }, { "epoch": 0.5426354036673987, "grad_norm": 1.9760692119598389, "learning_rate": 9.10710288613193e-06, "loss": 0.14466817378997804, "step": 20600 }, { "epoch": 0.5428988189118975, "grad_norm": 1.6995593309402466, "learning_rate": 9.098605916378837e-06, "loss": 0.14294489622116088, "step": 20610 }, { "epoch": 0.5431622341563962, "grad_norm": 2.6123838424682617, "learning_rate": 9.090109602699239e-06, "loss": 0.14380329847335815, "step": 20620 }, { "epoch": 0.543425649400895, "grad_norm": 2.859376907348633, "learning_rate": 9.081613951277135e-06, "loss": 0.1457177519798279, "step": 20630 }, { "epoch": 0.5436890646453937, "grad_norm": 0.9787763953208923, "learning_rate": 9.07311896829602e-06, "loss": 0.14217019081115723, "step": 20640 }, { "epoch": 0.5439524798898925, "grad_norm": 1.3614226579666138, "learning_rate": 9.064624659938902e-06, "loss": 0.13428668975830077, "step": 20650 }, { "epoch": 0.5442158951343912, "grad_norm": 1.6199923753738403, "learning_rate": 9.056131032388318e-06, "loss": 0.15025068521499635, "step": 20660 }, { "epoch": 0.5444793103788899, "grad_norm": 3.290330410003662, "learning_rate": 9.047638091826291e-06, "loss": 0.1459529161453247, "step": 20670 }, { "epoch": 0.5447427256233887, "grad_norm": 1.0707005262374878, "learning_rate": 9.039145844434344e-06, "loss": 0.14878835678100585, "step": 20680 }, { "epoch": 0.5450061408678873, "grad_norm": 2.3444111347198486, "learning_rate": 9.030654296393512e-06, "loss": 0.13953973054885865, "step": 20690 }, { "epoch": 0.5452695561123861, "grad_norm": 1.2372063398361206, "learning_rate": 9.022163453884299e-06, "loss": 0.1419472336769104, "step": 20700 }, { "epoch": 0.5455329713568848, "grad_norm": 1.206382393836975, "learning_rate": 9.013673323086716e-06, "loss": 0.1472236156463623, "step": 20710 }, { "epoch": 0.5457963866013836, "grad_norm": 2.5940349102020264, "learning_rate": 9.005183910180243e-06, "loss": 0.14005050659179688, "step": 20720 }, { "epoch": 0.5460598018458823, "grad_norm": 1.6014878749847412, "learning_rate": 8.996695221343834e-06, "loss": 0.14512170553207399, "step": 20730 }, { "epoch": 0.546323217090381, "grad_norm": 1.9984054565429688, "learning_rate": 8.988207262755936e-06, "loss": 0.13643984794616698, "step": 20740 }, { "epoch": 0.5465866323348798, "grad_norm": 1.7795199155807495, "learning_rate": 8.979720040594438e-06, "loss": 0.1517580270767212, "step": 20750 }, { "epoch": 0.5468500475793785, "grad_norm": 3.037431001663208, "learning_rate": 8.971233561036718e-06, "loss": 0.1344667673110962, "step": 20760 }, { "epoch": 0.5471134628238773, "grad_norm": 1.3158403635025024, "learning_rate": 8.962747830259598e-06, "loss": 0.14314764738082886, "step": 20770 }, { "epoch": 0.547376878068376, "grad_norm": 1.398511528968811, "learning_rate": 8.954262854439354e-06, "loss": 0.15594903230667115, "step": 20780 }, { "epoch": 0.5476402933128748, "grad_norm": 1.4542665481567383, "learning_rate": 8.945778639751723e-06, "loss": 0.14854226112365723, "step": 20790 }, { "epoch": 0.5479037085573735, "grad_norm": 2.640573263168335, "learning_rate": 8.937295192371882e-06, "loss": 0.13924386501312255, "step": 20800 }, { "epoch": 0.5481671238018723, "grad_norm": 0.8788541555404663, "learning_rate": 8.928812518474445e-06, "loss": 0.12493373155593872, "step": 20810 }, { "epoch": 0.548430539046371, "grad_norm": 1.473767876625061, "learning_rate": 8.920330624233477e-06, "loss": 0.13367819786071777, "step": 20820 }, { "epoch": 0.5486939542908696, "grad_norm": 1.5962122678756714, "learning_rate": 8.911849515822457e-06, "loss": 0.146856951713562, "step": 20830 }, { "epoch": 0.5489573695353684, "grad_norm": 1.0507304668426514, "learning_rate": 8.903369199414311e-06, "loss": 0.14669082164764405, "step": 20840 }, { "epoch": 0.5492207847798671, "grad_norm": 1.2556434869766235, "learning_rate": 8.894889681181374e-06, "loss": 0.14285554885864257, "step": 20850 }, { "epoch": 0.5494842000243659, "grad_norm": 2.8843748569488525, "learning_rate": 8.886410967295401e-06, "loss": 0.15043903589248658, "step": 20860 }, { "epoch": 0.5497476152688646, "grad_norm": 1.455613374710083, "learning_rate": 8.877933063927576e-06, "loss": 0.12866997718811035, "step": 20870 }, { "epoch": 0.5500110305133634, "grad_norm": 1.2143800258636475, "learning_rate": 8.869455977248473e-06, "loss": 0.14446790218353273, "step": 20880 }, { "epoch": 0.5502744457578621, "grad_norm": 2.373098611831665, "learning_rate": 8.860979713428085e-06, "loss": 0.1284008026123047, "step": 20890 }, { "epoch": 0.5505378610023609, "grad_norm": 2.4396185874938965, "learning_rate": 8.852504278635806e-06, "loss": 0.14261178970336913, "step": 20900 }, { "epoch": 0.5508012762468596, "grad_norm": 2.798598527908325, "learning_rate": 8.844029679040418e-06, "loss": 0.13772372007369996, "step": 20910 }, { "epoch": 0.5510646914913583, "grad_norm": 1.8013896942138672, "learning_rate": 8.835555920810104e-06, "loss": 0.13402650356292725, "step": 20920 }, { "epoch": 0.5513281067358571, "grad_norm": 1.5705934762954712, "learning_rate": 8.827083010112429e-06, "loss": 0.13004162311553955, "step": 20930 }, { "epoch": 0.5515915219803558, "grad_norm": 1.007167935371399, "learning_rate": 8.818610953114342e-06, "loss": 0.13803149461746217, "step": 20940 }, { "epoch": 0.5518549372248546, "grad_norm": 2.239243745803833, "learning_rate": 8.810139755982173e-06, "loss": 0.1362785816192627, "step": 20950 }, { "epoch": 0.5521183524693533, "grad_norm": 1.7880754470825195, "learning_rate": 8.801669424881624e-06, "loss": 0.1455116868019104, "step": 20960 }, { "epoch": 0.5523817677138521, "grad_norm": 1.0791460275650024, "learning_rate": 8.79319996597777e-06, "loss": 0.1356194257736206, "step": 20970 }, { "epoch": 0.5526451829583507, "grad_norm": 2.673792839050293, "learning_rate": 8.784731385435049e-06, "loss": 0.13961164951324462, "step": 20980 }, { "epoch": 0.5529085982028495, "grad_norm": 1.234612226486206, "learning_rate": 8.77626368941725e-06, "loss": 0.12904365062713624, "step": 20990 }, { "epoch": 0.5531720134473482, "grad_norm": 1.921331524848938, "learning_rate": 8.767796884087539e-06, "loss": 0.14220647811889647, "step": 21000 }, { "epoch": 0.5534354286918469, "grad_norm": 1.59220290184021, "learning_rate": 8.759330975608418e-06, "loss": 0.14628324508666993, "step": 21010 }, { "epoch": 0.5536988439363457, "grad_norm": 1.0279210805892944, "learning_rate": 8.750865970141736e-06, "loss": 0.13822426795959472, "step": 21020 }, { "epoch": 0.5539622591808444, "grad_norm": 1.364558219909668, "learning_rate": 8.742401873848696e-06, "loss": 0.1441588282585144, "step": 21030 }, { "epoch": 0.5542256744253432, "grad_norm": 1.7580287456512451, "learning_rate": 8.733938692889827e-06, "loss": 0.1441972851753235, "step": 21040 }, { "epoch": 0.5544890896698419, "grad_norm": 1.9328315258026123, "learning_rate": 8.725476433425003e-06, "loss": 0.14438742399215698, "step": 21050 }, { "epoch": 0.5547525049143407, "grad_norm": 2.6036219596862793, "learning_rate": 8.717015101613417e-06, "loss": 0.15445878505706787, "step": 21060 }, { "epoch": 0.5550159201588394, "grad_norm": 1.0863735675811768, "learning_rate": 8.708554703613592e-06, "loss": 0.13840298652648925, "step": 21070 }, { "epoch": 0.5552793354033382, "grad_norm": 1.5532797574996948, "learning_rate": 8.700095245583374e-06, "loss": 0.12546783685684204, "step": 21080 }, { "epoch": 0.5555427506478369, "grad_norm": 1.215814471244812, "learning_rate": 8.691636733679915e-06, "loss": 0.1474788784980774, "step": 21090 }, { "epoch": 0.5558061658923356, "grad_norm": 1.035428524017334, "learning_rate": 8.683179174059691e-06, "loss": 0.14591716527938842, "step": 21100 }, { "epoch": 0.5560695811368344, "grad_norm": 1.405126929283142, "learning_rate": 8.67472257287848e-06, "loss": 0.14031503200531006, "step": 21110 }, { "epoch": 0.556332996381333, "grad_norm": 2.9720284938812256, "learning_rate": 8.666266936291354e-06, "loss": 0.1470266819000244, "step": 21120 }, { "epoch": 0.5565964116258318, "grad_norm": 2.036269426345825, "learning_rate": 8.657812270452695e-06, "loss": 0.14854304790496825, "step": 21130 }, { "epoch": 0.5568598268703305, "grad_norm": 3.0547292232513428, "learning_rate": 8.649358581516171e-06, "loss": 0.14850430488586425, "step": 21140 }, { "epoch": 0.5571232421148293, "grad_norm": 1.1473997831344604, "learning_rate": 8.640905875634744e-06, "loss": 0.14032835960388185, "step": 21150 }, { "epoch": 0.557386657359328, "grad_norm": 1.9324498176574707, "learning_rate": 8.632454158960656e-06, "loss": 0.1416162967681885, "step": 21160 }, { "epoch": 0.5576500726038268, "grad_norm": 1.510184645652771, "learning_rate": 8.624003437645431e-06, "loss": 0.12798300981521607, "step": 21170 }, { "epoch": 0.5579134878483255, "grad_norm": 3.5571646690368652, "learning_rate": 8.615553717839873e-06, "loss": 0.12958848476409912, "step": 21180 }, { "epoch": 0.5581769030928242, "grad_norm": 1.9652025699615479, "learning_rate": 8.607105005694048e-06, "loss": 0.15796165466308593, "step": 21190 }, { "epoch": 0.558440318337323, "grad_norm": 1.2466548681259155, "learning_rate": 8.598657307357293e-06, "loss": 0.14633381366729736, "step": 21200 }, { "epoch": 0.5587037335818217, "grad_norm": 1.2790203094482422, "learning_rate": 8.590210628978213e-06, "loss": 0.13542633056640624, "step": 21210 }, { "epoch": 0.5589671488263205, "grad_norm": 2.716552972793579, "learning_rate": 8.581764976704653e-06, "loss": 0.15139721632003783, "step": 21220 }, { "epoch": 0.5592305640708192, "grad_norm": 1.605116367340088, "learning_rate": 8.573320356683739e-06, "loss": 0.1433122158050537, "step": 21230 }, { "epoch": 0.559493979315318, "grad_norm": 1.4077167510986328, "learning_rate": 8.564876775061819e-06, "loss": 0.13082010746002198, "step": 21240 }, { "epoch": 0.5597573945598167, "grad_norm": 2.1443498134613037, "learning_rate": 8.556434237984496e-06, "loss": 0.12629921436309816, "step": 21250 }, { "epoch": 0.5600208098043153, "grad_norm": 2.76877760887146, "learning_rate": 8.547992751596617e-06, "loss": 0.15175271034240723, "step": 21260 }, { "epoch": 0.5602842250488141, "grad_norm": 1.3197495937347412, "learning_rate": 8.53955232204226e-06, "loss": 0.14688124656677246, "step": 21270 }, { "epoch": 0.5605476402933128, "grad_norm": 1.8022853136062622, "learning_rate": 8.531112955464728e-06, "loss": 0.14115902185440063, "step": 21280 }, { "epoch": 0.5608110555378116, "grad_norm": 0.820013701915741, "learning_rate": 8.522674658006564e-06, "loss": 0.13342647552490233, "step": 21290 }, { "epoch": 0.5610744707823103, "grad_norm": 1.9270676374435425, "learning_rate": 8.514237435809517e-06, "loss": 0.13708875179290772, "step": 21300 }, { "epoch": 0.5613378860268091, "grad_norm": 0.9707344174385071, "learning_rate": 8.505801295014569e-06, "loss": 0.12839465141296386, "step": 21310 }, { "epoch": 0.5616013012713078, "grad_norm": 2.7273123264312744, "learning_rate": 8.497366241761907e-06, "loss": 0.14307765960693358, "step": 21320 }, { "epoch": 0.5618647165158066, "grad_norm": 0.9454347491264343, "learning_rate": 8.488932282190914e-06, "loss": 0.1418527841567993, "step": 21330 }, { "epoch": 0.5621281317603053, "grad_norm": 1.1497737169265747, "learning_rate": 8.480499422440207e-06, "loss": 0.13174790143966675, "step": 21340 }, { "epoch": 0.562391547004804, "grad_norm": 2.2739737033843994, "learning_rate": 8.472067668647571e-06, "loss": 0.1370586633682251, "step": 21350 }, { "epoch": 0.5626549622493028, "grad_norm": 1.5624831914901733, "learning_rate": 8.463637026950013e-06, "loss": 0.14252681732177735, "step": 21360 }, { "epoch": 0.5629183774938015, "grad_norm": 1.4014149904251099, "learning_rate": 8.455207503483708e-06, "loss": 0.13889676332473755, "step": 21370 }, { "epoch": 0.5631817927383003, "grad_norm": 4.310088157653809, "learning_rate": 8.446779104384028e-06, "loss": 0.14517813920974731, "step": 21380 }, { "epoch": 0.563445207982799, "grad_norm": 2.3990464210510254, "learning_rate": 8.43835183578553e-06, "loss": 0.13661344051361085, "step": 21390 }, { "epoch": 0.5637086232272978, "grad_norm": 1.2284396886825562, "learning_rate": 8.429925703821941e-06, "loss": 0.1489706516265869, "step": 21400 }, { "epoch": 0.5639720384717964, "grad_norm": 2.3163232803344727, "learning_rate": 8.421500714626164e-06, "loss": 0.14896812438964843, "step": 21410 }, { "epoch": 0.5642354537162952, "grad_norm": 2.077481269836426, "learning_rate": 8.413076874330266e-06, "loss": 0.14536721706390382, "step": 21420 }, { "epoch": 0.5644988689607939, "grad_norm": 1.0778180360794067, "learning_rate": 8.404654189065486e-06, "loss": 0.14029300212860107, "step": 21430 }, { "epoch": 0.5647622842052926, "grad_norm": 1.182979702949524, "learning_rate": 8.396232664962218e-06, "loss": 0.15225814580917357, "step": 21440 }, { "epoch": 0.5650256994497914, "grad_norm": 1.1856297254562378, "learning_rate": 8.387812308150011e-06, "loss": 0.1212612509727478, "step": 21450 }, { "epoch": 0.5652891146942901, "grad_norm": 1.4856414794921875, "learning_rate": 8.379393124757554e-06, "loss": 0.15635645389556885, "step": 21460 }, { "epoch": 0.5655525299387889, "grad_norm": 1.2284791469573975, "learning_rate": 8.370975120912708e-06, "loss": 0.1459421157836914, "step": 21470 }, { "epoch": 0.5658159451832876, "grad_norm": 3.7784557342529297, "learning_rate": 8.362558302742447e-06, "loss": 0.13513576984405518, "step": 21480 }, { "epoch": 0.5660793604277864, "grad_norm": 3.185525894165039, "learning_rate": 8.354142676372898e-06, "loss": 0.1465972661972046, "step": 21490 }, { "epoch": 0.5663427756722851, "grad_norm": 1.0058479309082031, "learning_rate": 8.345728247929319e-06, "loss": 0.13593921661376954, "step": 21500 }, { "epoch": 0.5666061909167839, "grad_norm": 2.013692855834961, "learning_rate": 8.33731502353609e-06, "loss": 0.13696660995483398, "step": 21510 }, { "epoch": 0.5668696061612826, "grad_norm": 2.339216947555542, "learning_rate": 8.328903009316724e-06, "loss": 0.16057662963867186, "step": 21520 }, { "epoch": 0.5671330214057813, "grad_norm": 2.0091419219970703, "learning_rate": 8.320492211393844e-06, "loss": 0.1400207281112671, "step": 21530 }, { "epoch": 0.5673964366502801, "grad_norm": 3.307774066925049, "learning_rate": 8.312082635889193e-06, "loss": 0.14703035354614258, "step": 21540 }, { "epoch": 0.5676598518947787, "grad_norm": 1.4040311574935913, "learning_rate": 8.303674288923621e-06, "loss": 0.1478423595428467, "step": 21550 }, { "epoch": 0.5679232671392775, "grad_norm": 2.3392491340637207, "learning_rate": 8.29526717661709e-06, "loss": 0.14175894260406494, "step": 21560 }, { "epoch": 0.5681866823837762, "grad_norm": 2.1681931018829346, "learning_rate": 8.286861305088655e-06, "loss": 0.1342236280441284, "step": 21570 }, { "epoch": 0.568450097628275, "grad_norm": 1.132524013519287, "learning_rate": 8.278456680456476e-06, "loss": 0.136657190322876, "step": 21580 }, { "epoch": 0.5687135128727737, "grad_norm": 1.633362054824829, "learning_rate": 8.270053308837791e-06, "loss": 0.13014538288116456, "step": 21590 }, { "epoch": 0.5689769281172725, "grad_norm": 1.7421880960464478, "learning_rate": 8.261651196348953e-06, "loss": 0.14268186092376708, "step": 21600 }, { "epoch": 0.5692403433617712, "grad_norm": 3.0321390628814697, "learning_rate": 8.253250349105369e-06, "loss": 0.14048962593078612, "step": 21610 }, { "epoch": 0.5695037586062699, "grad_norm": 1.277105450630188, "learning_rate": 8.24485077322154e-06, "loss": 0.13367104530334473, "step": 21620 }, { "epoch": 0.5697671738507687, "grad_norm": 1.4577267169952393, "learning_rate": 8.236452474811046e-06, "loss": 0.14081515073776246, "step": 21630 }, { "epoch": 0.5700305890952674, "grad_norm": 1.5705159902572632, "learning_rate": 8.228055459986521e-06, "loss": 0.14443295001983641, "step": 21640 }, { "epoch": 0.5702940043397662, "grad_norm": 2.6392416954040527, "learning_rate": 8.219659734859686e-06, "loss": 0.13086326122283937, "step": 21650 }, { "epoch": 0.5705574195842649, "grad_norm": 2.2307488918304443, "learning_rate": 8.211265305541304e-06, "loss": 0.13940789699554443, "step": 21660 }, { "epoch": 0.5708208348287637, "grad_norm": 2.447232246398926, "learning_rate": 8.202872178141204e-06, "loss": 0.1437532424926758, "step": 21670 }, { "epoch": 0.5710842500732624, "grad_norm": 4.781264305114746, "learning_rate": 8.19448035876827e-06, "loss": 0.13025875091552735, "step": 21680 }, { "epoch": 0.5713476653177612, "grad_norm": 1.7582088708877563, "learning_rate": 8.186089853530426e-06, "loss": 0.13976566791534423, "step": 21690 }, { "epoch": 0.5716110805622598, "grad_norm": 4.437719345092773, "learning_rate": 8.177700668534646e-06, "loss": 0.1477077603340149, "step": 21700 }, { "epoch": 0.5718744958067585, "grad_norm": 1.3231457471847534, "learning_rate": 8.169312809886944e-06, "loss": 0.14540590047836305, "step": 21710 }, { "epoch": 0.5721379110512573, "grad_norm": 1.768678069114685, "learning_rate": 8.160926283692355e-06, "loss": 0.1366886854171753, "step": 21720 }, { "epoch": 0.572401326295756, "grad_norm": 2.415264844894409, "learning_rate": 8.152541096054965e-06, "loss": 0.13476073741912842, "step": 21730 }, { "epoch": 0.5726647415402548, "grad_norm": 1.3911515474319458, "learning_rate": 8.144157253077871e-06, "loss": 0.14522867202758788, "step": 21740 }, { "epoch": 0.5729281567847535, "grad_norm": 1.776887059211731, "learning_rate": 8.135774760863193e-06, "loss": 0.14630258083343506, "step": 21750 }, { "epoch": 0.5731915720292523, "grad_norm": 0.9247317910194397, "learning_rate": 8.127393625512077e-06, "loss": 0.13907299041748047, "step": 21760 }, { "epoch": 0.573454987273751, "grad_norm": 1.287216067314148, "learning_rate": 8.11901385312467e-06, "loss": 0.1454732060432434, "step": 21770 }, { "epoch": 0.5737184025182498, "grad_norm": 1.1424788236618042, "learning_rate": 8.110635449800132e-06, "loss": 0.1432805061340332, "step": 21780 }, { "epoch": 0.5739818177627485, "grad_norm": 1.9062596559524536, "learning_rate": 8.10225842163663e-06, "loss": 0.13991246223449708, "step": 21790 }, { "epoch": 0.5742452330072472, "grad_norm": 1.8231836557388306, "learning_rate": 8.093882774731322e-06, "loss": 0.15258758068084716, "step": 21800 }, { "epoch": 0.574508648251746, "grad_norm": 2.6037895679473877, "learning_rate": 8.085508515180373e-06, "loss": 0.14622726440429687, "step": 21810 }, { "epoch": 0.5747720634962447, "grad_norm": 1.62034273147583, "learning_rate": 8.077135649078917e-06, "loss": 0.13441712856292726, "step": 21820 }, { "epoch": 0.5750354787407435, "grad_norm": 2.6069250106811523, "learning_rate": 8.068764182521102e-06, "loss": 0.14745705127716063, "step": 21830 }, { "epoch": 0.5752988939852421, "grad_norm": 2.262434244155884, "learning_rate": 8.060394121600033e-06, "loss": 0.1385944128036499, "step": 21840 }, { "epoch": 0.5755623092297409, "grad_norm": 0.9521980285644531, "learning_rate": 8.052025472407802e-06, "loss": 0.14219447374343872, "step": 21850 }, { "epoch": 0.5758257244742396, "grad_norm": 1.3309924602508545, "learning_rate": 8.043658241035477e-06, "loss": 0.15094131231307983, "step": 21860 }, { "epoch": 0.5760891397187383, "grad_norm": 1.0624761581420898, "learning_rate": 8.035292433573088e-06, "loss": 0.13712711334228517, "step": 21870 }, { "epoch": 0.5763525549632371, "grad_norm": 2.049549102783203, "learning_rate": 8.026928056109628e-06, "loss": 0.13426334857940675, "step": 21880 }, { "epoch": 0.5766159702077358, "grad_norm": 2.4750239849090576, "learning_rate": 8.018565114733056e-06, "loss": 0.152967631816864, "step": 21890 }, { "epoch": 0.5768793854522346, "grad_norm": 1.1282082796096802, "learning_rate": 8.010203615530279e-06, "loss": 0.13822417259216307, "step": 21900 }, { "epoch": 0.5771428006967333, "grad_norm": 2.00154185295105, "learning_rate": 8.00184356458716e-06, "loss": 0.13928136825561524, "step": 21910 }, { "epoch": 0.5774062159412321, "grad_norm": 2.4279444217681885, "learning_rate": 7.993484967988504e-06, "loss": 0.1367687463760376, "step": 21920 }, { "epoch": 0.5776696311857308, "grad_norm": 1.760727882385254, "learning_rate": 7.98512783181805e-06, "loss": 0.13040430545806886, "step": 21930 }, { "epoch": 0.5779330464302296, "grad_norm": 0.9972143769264221, "learning_rate": 7.976772162158498e-06, "loss": 0.14616258144378663, "step": 21940 }, { "epoch": 0.5781964616747283, "grad_norm": 1.2001657485961914, "learning_rate": 7.968417965091449e-06, "loss": 0.13007659912109376, "step": 21950 }, { "epoch": 0.578459876919227, "grad_norm": 3.403923511505127, "learning_rate": 7.960065246697456e-06, "loss": 0.13981012105941773, "step": 21960 }, { "epoch": 0.5787232921637258, "grad_norm": 1.2256461381912231, "learning_rate": 7.951714013055988e-06, "loss": 0.14254635572433472, "step": 21970 }, { "epoch": 0.5789867074082244, "grad_norm": 2.4260880947113037, "learning_rate": 7.94336427024543e-06, "loss": 0.141426944732666, "step": 21980 }, { "epoch": 0.5792501226527232, "grad_norm": 2.1684815883636475, "learning_rate": 7.93501602434309e-06, "loss": 0.13691939115524293, "step": 21990 }, { "epoch": 0.5795135378972219, "grad_norm": 1.7566275596618652, "learning_rate": 7.926669281425177e-06, "loss": 0.14665026664733888, "step": 22000 }, { "epoch": 0.5797769531417207, "grad_norm": 2.041823625564575, "learning_rate": 7.91832404756681e-06, "loss": 0.15235153436660767, "step": 22010 }, { "epoch": 0.5800403683862194, "grad_norm": 3.349639892578125, "learning_rate": 7.909980328842015e-06, "loss": 0.14169347286224365, "step": 22020 }, { "epoch": 0.5803037836307182, "grad_norm": 1.3941168785095215, "learning_rate": 7.901638131323705e-06, "loss": 0.13447632789611816, "step": 22030 }, { "epoch": 0.5805671988752169, "grad_norm": 1.5212807655334473, "learning_rate": 7.893297461083697e-06, "loss": 0.14364492893218994, "step": 22040 }, { "epoch": 0.5808306141197156, "grad_norm": 2.154919147491455, "learning_rate": 7.884958324192689e-06, "loss": 0.15233582258224487, "step": 22050 }, { "epoch": 0.5810940293642144, "grad_norm": 2.325606346130371, "learning_rate": 7.876620726720257e-06, "loss": 0.15916038751602174, "step": 22060 }, { "epoch": 0.5813574446087131, "grad_norm": 1.3664339780807495, "learning_rate": 7.868284674734877e-06, "loss": 0.14454200267791747, "step": 22070 }, { "epoch": 0.5816208598532119, "grad_norm": 1.2808334827423096, "learning_rate": 7.859950174303879e-06, "loss": 0.13964483737945557, "step": 22080 }, { "epoch": 0.5818842750977106, "grad_norm": 1.326959252357483, "learning_rate": 7.851617231493472e-06, "loss": 0.14288663864135742, "step": 22090 }, { "epoch": 0.5821476903422094, "grad_norm": 3.025688648223877, "learning_rate": 7.843285852368734e-06, "loss": 0.1438010334968567, "step": 22100 }, { "epoch": 0.5824111055867081, "grad_norm": 0.9006655216217041, "learning_rate": 7.8349560429936e-06, "loss": 0.13613922595977784, "step": 22110 }, { "epoch": 0.5826745208312069, "grad_norm": 2.00726056098938, "learning_rate": 7.826627809430867e-06, "loss": 0.14535844326019287, "step": 22120 }, { "epoch": 0.5829379360757055, "grad_norm": 1.2734129428863525, "learning_rate": 7.818301157742182e-06, "loss": 0.14210078716278077, "step": 22130 }, { "epoch": 0.5832013513202042, "grad_norm": 1.9776004552841187, "learning_rate": 7.809976093988039e-06, "loss": 0.1412169098854065, "step": 22140 }, { "epoch": 0.583464766564703, "grad_norm": 2.3372480869293213, "learning_rate": 7.801652624227782e-06, "loss": 0.14606118202209473, "step": 22150 }, { "epoch": 0.5837281818092017, "grad_norm": 1.957200527191162, "learning_rate": 7.79333075451959e-06, "loss": 0.13367938995361328, "step": 22160 }, { "epoch": 0.5839915970537005, "grad_norm": 2.0938234329223633, "learning_rate": 7.785010490920477e-06, "loss": 0.1360210657119751, "step": 22170 }, { "epoch": 0.5842550122981992, "grad_norm": 3.259354591369629, "learning_rate": 7.776691839486297e-06, "loss": 0.13232021331787108, "step": 22180 }, { "epoch": 0.584518427542698, "grad_norm": 1.5120712518692017, "learning_rate": 7.76837480627171e-06, "loss": 0.14411338567733764, "step": 22190 }, { "epoch": 0.5847818427871967, "grad_norm": 2.7178122997283936, "learning_rate": 7.760059397330225e-06, "loss": 0.12883739471435546, "step": 22200 }, { "epoch": 0.5850452580316955, "grad_norm": 2.1749377250671387, "learning_rate": 7.75174561871415e-06, "loss": 0.13742561340332032, "step": 22210 }, { "epoch": 0.5853086732761942, "grad_norm": 1.376511573791504, "learning_rate": 7.743433476474608e-06, "loss": 0.14099645614624023, "step": 22220 }, { "epoch": 0.5855720885206929, "grad_norm": 2.7851803302764893, "learning_rate": 7.73512297666154e-06, "loss": 0.137251615524292, "step": 22230 }, { "epoch": 0.5858355037651917, "grad_norm": 1.7320144176483154, "learning_rate": 7.726814125323683e-06, "loss": 0.13355311155319213, "step": 22240 }, { "epoch": 0.5860989190096904, "grad_norm": 3.3466720581054688, "learning_rate": 7.718506928508583e-06, "loss": 0.13157451152801514, "step": 22250 }, { "epoch": 0.5863623342541892, "grad_norm": 1.454717755317688, "learning_rate": 7.71020139226257e-06, "loss": 0.128670597076416, "step": 22260 }, { "epoch": 0.5866257494986878, "grad_norm": 1.260093092918396, "learning_rate": 7.701897522630772e-06, "loss": 0.13451685905456542, "step": 22270 }, { "epoch": 0.5868891647431866, "grad_norm": 1.9417003393173218, "learning_rate": 7.693595325657109e-06, "loss": 0.14411141872406005, "step": 22280 }, { "epoch": 0.5871525799876853, "grad_norm": 1.3007649183273315, "learning_rate": 7.685294807384273e-06, "loss": 0.13818674087524413, "step": 22290 }, { "epoch": 0.5874159952321841, "grad_norm": 2.8054728507995605, "learning_rate": 7.676995973853743e-06, "loss": 0.13809475898742676, "step": 22300 }, { "epoch": 0.5876794104766828, "grad_norm": 2.359215497970581, "learning_rate": 7.668698831105768e-06, "loss": 0.12153959274291992, "step": 22310 }, { "epoch": 0.5879428257211815, "grad_norm": 1.6965869665145874, "learning_rate": 7.66040338517936e-06, "loss": 0.1519408941268921, "step": 22320 }, { "epoch": 0.5882062409656803, "grad_norm": 1.1389813423156738, "learning_rate": 7.652109642112308e-06, "loss": 0.14197875261306764, "step": 22330 }, { "epoch": 0.588469656210179, "grad_norm": 0.879774808883667, "learning_rate": 7.643817607941157e-06, "loss": 0.1247597336769104, "step": 22340 }, { "epoch": 0.5887330714546778, "grad_norm": 1.5250407457351685, "learning_rate": 7.635527288701201e-06, "loss": 0.12969813346862794, "step": 22350 }, { "epoch": 0.5889964866991765, "grad_norm": 1.9490437507629395, "learning_rate": 7.6272386904265005e-06, "loss": 0.14052047729492187, "step": 22360 }, { "epoch": 0.5892599019436753, "grad_norm": 1.3290855884552002, "learning_rate": 7.618951819149848e-06, "loss": 0.13489744663238526, "step": 22370 }, { "epoch": 0.589523317188174, "grad_norm": 1.0907402038574219, "learning_rate": 7.610666680902791e-06, "loss": 0.11795734167098999, "step": 22380 }, { "epoch": 0.5897867324326727, "grad_norm": 1.4133387804031372, "learning_rate": 7.602383281715608e-06, "loss": 0.12640891075134278, "step": 22390 }, { "epoch": 0.5900501476771715, "grad_norm": 3.5032525062561035, "learning_rate": 7.594101627617312e-06, "loss": 0.13037655353546143, "step": 22400 }, { "epoch": 0.5903135629216701, "grad_norm": 3.667081594467163, "learning_rate": 7.585821724635656e-06, "loss": 0.13933902978897095, "step": 22410 }, { "epoch": 0.5905769781661689, "grad_norm": 2.4519429206848145, "learning_rate": 7.577543578797097e-06, "loss": 0.1340112566947937, "step": 22420 }, { "epoch": 0.5908403934106676, "grad_norm": 4.345276355743408, "learning_rate": 7.569267196126844e-06, "loss": 0.13415942192077637, "step": 22430 }, { "epoch": 0.5911038086551664, "grad_norm": 1.3273944854736328, "learning_rate": 7.560992582648793e-06, "loss": 0.15048201084136964, "step": 22440 }, { "epoch": 0.5913672238996651, "grad_norm": 0.8380498290061951, "learning_rate": 7.552719744385566e-06, "loss": 0.14122240543365477, "step": 22450 }, { "epoch": 0.5916306391441639, "grad_norm": 1.030920147895813, "learning_rate": 7.544448687358495e-06, "loss": 0.14346637725830078, "step": 22460 }, { "epoch": 0.5918940543886626, "grad_norm": 1.533831000328064, "learning_rate": 7.536179417587612e-06, "loss": 0.1467394709587097, "step": 22470 }, { "epoch": 0.5921574696331613, "grad_norm": 1.68788743019104, "learning_rate": 7.527911941091644e-06, "loss": 0.13858261108398437, "step": 22480 }, { "epoch": 0.5924208848776601, "grad_norm": 2.478898048400879, "learning_rate": 7.519646263888023e-06, "loss": 0.1466692328453064, "step": 22490 }, { "epoch": 0.5926843001221588, "grad_norm": 1.737900733947754, "learning_rate": 7.511382391992861e-06, "loss": 0.132220995426178, "step": 22500 }, { "epoch": 0.5929477153666576, "grad_norm": 2.6237199306488037, "learning_rate": 7.503120331420963e-06, "loss": 0.1401451349258423, "step": 22510 }, { "epoch": 0.5932111306111563, "grad_norm": 1.329270362854004, "learning_rate": 7.494860088185815e-06, "loss": 0.13867149353027344, "step": 22520 }, { "epoch": 0.5934745458556551, "grad_norm": 1.0304137468338013, "learning_rate": 7.48660166829957e-06, "loss": 0.1359700918197632, "step": 22530 }, { "epoch": 0.5937379611001538, "grad_norm": 1.5632143020629883, "learning_rate": 7.478345077773075e-06, "loss": 0.13265614509582518, "step": 22540 }, { "epoch": 0.5940013763446526, "grad_norm": 1.7573111057281494, "learning_rate": 7.470090322615822e-06, "loss": 0.1461291193962097, "step": 22550 }, { "epoch": 0.5942647915891512, "grad_norm": 2.549546241760254, "learning_rate": 7.4618374088359805e-06, "loss": 0.1453535795211792, "step": 22560 }, { "epoch": 0.5945282068336499, "grad_norm": 0.8963285088539124, "learning_rate": 7.45358634244038e-06, "loss": 0.12899158000946045, "step": 22570 }, { "epoch": 0.5947916220781487, "grad_norm": 1.8742681741714478, "learning_rate": 7.4453371294345e-06, "loss": 0.13969855308532714, "step": 22580 }, { "epoch": 0.5950550373226474, "grad_norm": 1.1914149522781372, "learning_rate": 7.437089775822477e-06, "loss": 0.15606969594955444, "step": 22590 }, { "epoch": 0.5953184525671462, "grad_norm": 3.112884759902954, "learning_rate": 7.4288442876070865e-06, "loss": 0.1387503981590271, "step": 22600 }, { "epoch": 0.5955818678116449, "grad_norm": 1.4798784255981445, "learning_rate": 7.420600670789753e-06, "loss": 0.1450800895690918, "step": 22610 }, { "epoch": 0.5958452830561437, "grad_norm": 1.9173600673675537, "learning_rate": 7.412358931370536e-06, "loss": 0.13388354778289796, "step": 22620 }, { "epoch": 0.5961086983006424, "grad_norm": 1.1699302196502686, "learning_rate": 7.404119075348129e-06, "loss": 0.1332531213760376, "step": 22630 }, { "epoch": 0.5963721135451412, "grad_norm": 1.226805567741394, "learning_rate": 7.3958811087198555e-06, "loss": 0.1351555347442627, "step": 22640 }, { "epoch": 0.5966355287896399, "grad_norm": 2.2241768836975098, "learning_rate": 7.3876450374816644e-06, "loss": 0.1242680311203003, "step": 22650 }, { "epoch": 0.5968989440341386, "grad_norm": 2.1293294429779053, "learning_rate": 7.379410867628115e-06, "loss": 0.13863098621368408, "step": 22660 }, { "epoch": 0.5971623592786374, "grad_norm": 1.2961041927337646, "learning_rate": 7.371178605152405e-06, "loss": 0.13768538236618041, "step": 22670 }, { "epoch": 0.597425774523136, "grad_norm": 1.511897325515747, "learning_rate": 7.362948256046321e-06, "loss": 0.12705087661743164, "step": 22680 }, { "epoch": 0.5976891897676349, "grad_norm": 1.9035743474960327, "learning_rate": 7.354719826300266e-06, "loss": 0.14341967105865477, "step": 22690 }, { "epoch": 0.5979526050121335, "grad_norm": 3.5648131370544434, "learning_rate": 7.346493321903251e-06, "loss": 0.13168864250183104, "step": 22700 }, { "epoch": 0.5982160202566323, "grad_norm": 2.061309576034546, "learning_rate": 7.338268748842876e-06, "loss": 0.1346265435218811, "step": 22710 }, { "epoch": 0.598479435501131, "grad_norm": 4.308382034301758, "learning_rate": 7.330046113105345e-06, "loss": 0.15230922698974608, "step": 22720 }, { "epoch": 0.5987428507456298, "grad_norm": 2.058213233947754, "learning_rate": 7.321825420675444e-06, "loss": 0.1256236433982849, "step": 22730 }, { "epoch": 0.5990062659901285, "grad_norm": 1.4238194227218628, "learning_rate": 7.3136066775365455e-06, "loss": 0.12268483638763428, "step": 22740 }, { "epoch": 0.5992696812346272, "grad_norm": 1.464016079902649, "learning_rate": 7.305389889670611e-06, "loss": 0.12579131126403809, "step": 22750 }, { "epoch": 0.599533096479126, "grad_norm": 2.8080224990844727, "learning_rate": 7.2971750630581675e-06, "loss": 0.14506924152374268, "step": 22760 }, { "epoch": 0.5997965117236247, "grad_norm": 2.32289981842041, "learning_rate": 7.2889622036783245e-06, "loss": 0.13142220973968505, "step": 22770 }, { "epoch": 0.6000599269681235, "grad_norm": 1.3357194662094116, "learning_rate": 7.280751317508756e-06, "loss": 0.13835612535476685, "step": 22780 }, { "epoch": 0.6003233422126222, "grad_norm": 2.9217703342437744, "learning_rate": 7.272542410525691e-06, "loss": 0.13725976943969725, "step": 22790 }, { "epoch": 0.600586757457121, "grad_norm": 3.767174005508423, "learning_rate": 7.2643354887039395e-06, "loss": 0.14237921237945556, "step": 22800 }, { "epoch": 0.6008501727016197, "grad_norm": 1.952309250831604, "learning_rate": 7.256130558016843e-06, "loss": 0.13486299514770508, "step": 22810 }, { "epoch": 0.6011135879461185, "grad_norm": 1.2877047061920166, "learning_rate": 7.247927624436304e-06, "loss": 0.13497045040130615, "step": 22820 }, { "epoch": 0.6013770031906172, "grad_norm": 3.3456454277038574, "learning_rate": 7.239726693932779e-06, "loss": 0.1384181261062622, "step": 22830 }, { "epoch": 0.6016404184351158, "grad_norm": 1.2117992639541626, "learning_rate": 7.2315277724752495e-06, "loss": 0.11705526113510131, "step": 22840 }, { "epoch": 0.6019038336796146, "grad_norm": 1.851601243019104, "learning_rate": 7.223330866031253e-06, "loss": 0.1367639422416687, "step": 22850 }, { "epoch": 0.6021672489241133, "grad_norm": 2.0845065116882324, "learning_rate": 7.215135980566847e-06, "loss": 0.12931843996047973, "step": 22860 }, { "epoch": 0.6024306641686121, "grad_norm": 2.2736008167266846, "learning_rate": 7.2069431220466215e-06, "loss": 0.12982580661773682, "step": 22870 }, { "epoch": 0.6026940794131108, "grad_norm": 2.674480676651001, "learning_rate": 7.198752296433697e-06, "loss": 0.13703490495681764, "step": 22880 }, { "epoch": 0.6029574946576096, "grad_norm": 1.565246820449829, "learning_rate": 7.190563509689705e-06, "loss": 0.13530919551849366, "step": 22890 }, { "epoch": 0.6032209099021083, "grad_norm": 2.219738721847534, "learning_rate": 7.182376767774803e-06, "loss": 0.13727716207504273, "step": 22900 }, { "epoch": 0.603484325146607, "grad_norm": 2.170437812805176, "learning_rate": 7.174192076647653e-06, "loss": 0.13908464908599855, "step": 22910 }, { "epoch": 0.6037477403911058, "grad_norm": 0.9131832122802734, "learning_rate": 7.166009442265423e-06, "loss": 0.12554715871810912, "step": 22920 }, { "epoch": 0.6040111556356045, "grad_norm": 1.6522296667099, "learning_rate": 7.15782887058379e-06, "loss": 0.12362624406814575, "step": 22930 }, { "epoch": 0.6042745708801033, "grad_norm": 2.423560857772827, "learning_rate": 7.149650367556926e-06, "loss": 0.12956205606460572, "step": 22940 }, { "epoch": 0.604537986124602, "grad_norm": 1.5958635807037354, "learning_rate": 7.141473939137498e-06, "loss": 0.1315408706665039, "step": 22950 }, { "epoch": 0.6048014013691008, "grad_norm": 2.3409807682037354, "learning_rate": 7.133299591276665e-06, "loss": 0.13859401941299437, "step": 22960 }, { "epoch": 0.6050648166135995, "grad_norm": 4.046905994415283, "learning_rate": 7.12512732992407e-06, "loss": 0.12936657667160034, "step": 22970 }, { "epoch": 0.6053282318580983, "grad_norm": 1.829798698425293, "learning_rate": 7.1169571610278375e-06, "loss": 0.14649610519409179, "step": 22980 }, { "epoch": 0.6055916471025969, "grad_norm": 1.6535916328430176, "learning_rate": 7.108789090534569e-06, "loss": 0.1285887598991394, "step": 22990 }, { "epoch": 0.6058550623470956, "grad_norm": 1.2781596183776855, "learning_rate": 7.100623124389334e-06, "loss": 0.12564703226089477, "step": 23000 }, { "epoch": 0.6061184775915944, "grad_norm": 1.779761552810669, "learning_rate": 7.092459268535685e-06, "loss": 0.1300681471824646, "step": 23010 }, { "epoch": 0.6063818928360931, "grad_norm": 2.2924916744232178, "learning_rate": 7.084297528915616e-06, "loss": 0.14320025444030762, "step": 23020 }, { "epoch": 0.6066453080805919, "grad_norm": 1.7958859205245972, "learning_rate": 7.076137911469603e-06, "loss": 0.13335188627243041, "step": 23030 }, { "epoch": 0.6069087233250906, "grad_norm": 2.0676112174987793, "learning_rate": 7.067980422136565e-06, "loss": 0.13191760778427125, "step": 23040 }, { "epoch": 0.6071721385695894, "grad_norm": 3.3237154483795166, "learning_rate": 7.059825066853872e-06, "loss": 0.14130390882492067, "step": 23050 }, { "epoch": 0.6074355538140881, "grad_norm": 1.3252068758010864, "learning_rate": 7.0516718515573445e-06, "loss": 0.1346278667449951, "step": 23060 }, { "epoch": 0.6076989690585869, "grad_norm": 1.21701180934906, "learning_rate": 7.043520782181248e-06, "loss": 0.12566326856613158, "step": 23070 }, { "epoch": 0.6079623843030856, "grad_norm": 0.9701268672943115, "learning_rate": 7.035371864658276e-06, "loss": 0.14326462745666504, "step": 23080 }, { "epoch": 0.6082257995475843, "grad_norm": 3.011596918106079, "learning_rate": 7.0272251049195684e-06, "loss": 0.12657630443572998, "step": 23090 }, { "epoch": 0.6084892147920831, "grad_norm": 1.5859776735305786, "learning_rate": 7.019080508894683e-06, "loss": 0.1393064022064209, "step": 23100 }, { "epoch": 0.6087526300365818, "grad_norm": 1.0201854705810547, "learning_rate": 7.010938082511614e-06, "loss": 0.13520464897155762, "step": 23110 }, { "epoch": 0.6090160452810806, "grad_norm": 0.92243492603302, "learning_rate": 7.0027978316967705e-06, "loss": 0.11962107419967652, "step": 23120 }, { "epoch": 0.6092794605255792, "grad_norm": 1.6745864152908325, "learning_rate": 6.994659762374967e-06, "loss": 0.13828139305114745, "step": 23130 }, { "epoch": 0.609542875770078, "grad_norm": 1.9463509321212769, "learning_rate": 6.986523880469458e-06, "loss": 0.1379551887512207, "step": 23140 }, { "epoch": 0.6098062910145767, "grad_norm": 1.6240496635437012, "learning_rate": 6.978390191901878e-06, "loss": 0.15803852081298828, "step": 23150 }, { "epoch": 0.6100697062590755, "grad_norm": 1.9957606792449951, "learning_rate": 6.970258702592277e-06, "loss": 0.14842696189880372, "step": 23160 }, { "epoch": 0.6103331215035742, "grad_norm": 1.5118907690048218, "learning_rate": 6.962129418459109e-06, "loss": 0.13108162879943847, "step": 23170 }, { "epoch": 0.6105965367480729, "grad_norm": 1.3585834503173828, "learning_rate": 6.954002345419214e-06, "loss": 0.13095296621322633, "step": 23180 }, { "epoch": 0.6108599519925717, "grad_norm": 1.7361661195755005, "learning_rate": 6.945877489387829e-06, "loss": 0.14684677124023438, "step": 23190 }, { "epoch": 0.6111233672370704, "grad_norm": 1.1690677404403687, "learning_rate": 6.9377548562785735e-06, "loss": 0.1438913106918335, "step": 23200 }, { "epoch": 0.6113867824815692, "grad_norm": 3.321448564529419, "learning_rate": 6.929634452003447e-06, "loss": 0.13421289920806884, "step": 23210 }, { "epoch": 0.6116501977260679, "grad_norm": 2.7688100337982178, "learning_rate": 6.921516282472837e-06, "loss": 0.14494483470916747, "step": 23220 }, { "epoch": 0.6119136129705667, "grad_norm": 1.16612708568573, "learning_rate": 6.91340035359549e-06, "loss": 0.14535285234451295, "step": 23230 }, { "epoch": 0.6121770282150654, "grad_norm": 1.7695574760437012, "learning_rate": 6.905286671278536e-06, "loss": 0.14812817573547363, "step": 23240 }, { "epoch": 0.6124404434595642, "grad_norm": 2.757967472076416, "learning_rate": 6.897175241427462e-06, "loss": 0.13392614126205443, "step": 23250 }, { "epoch": 0.6127038587040629, "grad_norm": 1.1154521703720093, "learning_rate": 6.889066069946107e-06, "loss": 0.13807668685913085, "step": 23260 }, { "epoch": 0.6129672739485615, "grad_norm": 1.4567043781280518, "learning_rate": 6.880959162736691e-06, "loss": 0.13970824480056762, "step": 23270 }, { "epoch": 0.6132306891930603, "grad_norm": 2.15162992477417, "learning_rate": 6.872854525699761e-06, "loss": 0.14394410848617553, "step": 23280 }, { "epoch": 0.613494104437559, "grad_norm": 1.6757107973098755, "learning_rate": 6.8647521647342205e-06, "loss": 0.13444560766220093, "step": 23290 }, { "epoch": 0.6137575196820578, "grad_norm": 1.544282078742981, "learning_rate": 6.856652085737323e-06, "loss": 0.13872048854827881, "step": 23300 }, { "epoch": 0.6140209349265565, "grad_norm": 1.5487407445907593, "learning_rate": 6.848554294604652e-06, "loss": 0.12985005378723144, "step": 23310 }, { "epoch": 0.6142843501710553, "grad_norm": 1.1773853302001953, "learning_rate": 6.8404587972301316e-06, "loss": 0.12821224927902222, "step": 23320 }, { "epoch": 0.614547765415554, "grad_norm": 1.5287106037139893, "learning_rate": 6.8323655995060135e-06, "loss": 0.14662374258041383, "step": 23330 }, { "epoch": 0.6148111806600528, "grad_norm": 1.9292664527893066, "learning_rate": 6.824274707322874e-06, "loss": 0.14386749267578125, "step": 23340 }, { "epoch": 0.6150745959045515, "grad_norm": 2.674490213394165, "learning_rate": 6.816186126569615e-06, "loss": 0.14352866411209106, "step": 23350 }, { "epoch": 0.6153380111490502, "grad_norm": 3.913530111312866, "learning_rate": 6.808099863133453e-06, "loss": 0.14514703750610353, "step": 23360 }, { "epoch": 0.615601426393549, "grad_norm": 1.3314069509506226, "learning_rate": 6.800015922899923e-06, "loss": 0.14441139698028566, "step": 23370 }, { "epoch": 0.6158648416380477, "grad_norm": 1.3594778776168823, "learning_rate": 6.7919343117528655e-06, "loss": 0.13278831243515016, "step": 23380 }, { "epoch": 0.6161282568825465, "grad_norm": 3.1915454864501953, "learning_rate": 6.783855035574418e-06, "loss": 0.13886797428131104, "step": 23390 }, { "epoch": 0.6163916721270452, "grad_norm": 2.2043368816375732, "learning_rate": 6.775778100245038e-06, "loss": 0.13809691667556762, "step": 23400 }, { "epoch": 0.616655087371544, "grad_norm": 2.495242118835449, "learning_rate": 6.76770351164346e-06, "loss": 0.13231287002563477, "step": 23410 }, { "epoch": 0.6169185026160426, "grad_norm": 3.5685060024261475, "learning_rate": 6.759631275646715e-06, "loss": 0.1338672161102295, "step": 23420 }, { "epoch": 0.6171819178605413, "grad_norm": 2.747638702392578, "learning_rate": 6.751561398130133e-06, "loss": 0.13973145484924315, "step": 23430 }, { "epoch": 0.6174453331050401, "grad_norm": 1.6452780961990356, "learning_rate": 6.743493884967312e-06, "loss": 0.1344381093978882, "step": 23440 }, { "epoch": 0.6177087483495388, "grad_norm": 2.1730122566223145, "learning_rate": 6.735428742030141e-06, "loss": 0.12619454860687257, "step": 23450 }, { "epoch": 0.6179721635940376, "grad_norm": 1.504738450050354, "learning_rate": 6.7273659751887775e-06, "loss": 0.13614765405654908, "step": 23460 }, { "epoch": 0.6182355788385363, "grad_norm": 1.910217523574829, "learning_rate": 6.719305590311649e-06, "loss": 0.14206268787384033, "step": 23470 }, { "epoch": 0.6184989940830351, "grad_norm": 1.5908321142196655, "learning_rate": 6.7112475932654554e-06, "loss": 0.1348401427268982, "step": 23480 }, { "epoch": 0.6187624093275338, "grad_norm": 1.550385594367981, "learning_rate": 6.70319198991515e-06, "loss": 0.13613076210021974, "step": 23490 }, { "epoch": 0.6190258245720326, "grad_norm": 1.5345630645751953, "learning_rate": 6.695138786123952e-06, "loss": 0.1250666618347168, "step": 23500 }, { "epoch": 0.6192892398165313, "grad_norm": 1.3813756704330444, "learning_rate": 6.687087987753331e-06, "loss": 0.1287366509437561, "step": 23510 }, { "epoch": 0.61955265506103, "grad_norm": 2.284297466278076, "learning_rate": 6.679039600663001e-06, "loss": 0.1473633885383606, "step": 23520 }, { "epoch": 0.6198160703055288, "grad_norm": 1.0218144655227661, "learning_rate": 6.670993630710929e-06, "loss": 0.13433544635772704, "step": 23530 }, { "epoch": 0.6200794855500275, "grad_norm": 3.259692430496216, "learning_rate": 6.662950083753319e-06, "loss": 0.12265582084655761, "step": 23540 }, { "epoch": 0.6203429007945263, "grad_norm": 1.9981532096862793, "learning_rate": 6.654908965644609e-06, "loss": 0.14202003479003905, "step": 23550 }, { "epoch": 0.6206063160390249, "grad_norm": 2.7271335124969482, "learning_rate": 6.646870282237474e-06, "loss": 0.12764648199081421, "step": 23560 }, { "epoch": 0.6208697312835237, "grad_norm": 1.1461126804351807, "learning_rate": 6.6388340393828125e-06, "loss": 0.139501690864563, "step": 23570 }, { "epoch": 0.6211331465280224, "grad_norm": 1.8472166061401367, "learning_rate": 6.6308002429297516e-06, "loss": 0.1305922508239746, "step": 23580 }, { "epoch": 0.6213965617725212, "grad_norm": 1.5326427221298218, "learning_rate": 6.622768898725634e-06, "loss": 0.12125575542449951, "step": 23590 }, { "epoch": 0.6216599770170199, "grad_norm": 1.99599027633667, "learning_rate": 6.614740012616017e-06, "loss": 0.13971397876739503, "step": 23600 }, { "epoch": 0.6219233922615186, "grad_norm": 3.304590940475464, "learning_rate": 6.606713590444673e-06, "loss": 0.1330944776535034, "step": 23610 }, { "epoch": 0.6221868075060174, "grad_norm": 2.0400049686431885, "learning_rate": 6.598689638053577e-06, "loss": 0.1339116334915161, "step": 23620 }, { "epoch": 0.6224502227505161, "grad_norm": 1.4354807138442993, "learning_rate": 6.590668161282903e-06, "loss": 0.13692710399627686, "step": 23630 }, { "epoch": 0.6227136379950149, "grad_norm": 2.2645506858825684, "learning_rate": 6.582649165971037e-06, "loss": 0.13564627170562743, "step": 23640 }, { "epoch": 0.6229770532395136, "grad_norm": 1.0720829963684082, "learning_rate": 6.574632657954542e-06, "loss": 0.12081665992736816, "step": 23650 }, { "epoch": 0.6232404684840124, "grad_norm": 1.6837486028671265, "learning_rate": 6.5666186430681835e-06, "loss": 0.13120708465576172, "step": 23660 }, { "epoch": 0.6235038837285111, "grad_norm": 2.6598060131073, "learning_rate": 6.558607127144905e-06, "loss": 0.12443850040435792, "step": 23670 }, { "epoch": 0.6237672989730099, "grad_norm": 2.0662739276885986, "learning_rate": 6.550598116015832e-06, "loss": 0.12586724758148193, "step": 23680 }, { "epoch": 0.6240307142175086, "grad_norm": 4.700531005859375, "learning_rate": 6.54259161551027e-06, "loss": 0.1430070996284485, "step": 23690 }, { "epoch": 0.6242941294620072, "grad_norm": 1.4063715934753418, "learning_rate": 6.534587631455696e-06, "loss": 0.13047313690185547, "step": 23700 }, { "epoch": 0.624557544706506, "grad_norm": 1.150915265083313, "learning_rate": 6.5265861696777556e-06, "loss": 0.1458498239517212, "step": 23710 }, { "epoch": 0.6248209599510047, "grad_norm": 3.0944011211395264, "learning_rate": 6.518587236000259e-06, "loss": 0.13266512155532836, "step": 23720 }, { "epoch": 0.6250843751955035, "grad_norm": 2.7176270484924316, "learning_rate": 6.510590836245166e-06, "loss": 0.12522993087768555, "step": 23730 }, { "epoch": 0.6253477904400022, "grad_norm": 1.875853419303894, "learning_rate": 6.502596976232616e-06, "loss": 0.12853543758392333, "step": 23740 }, { "epoch": 0.625611205684501, "grad_norm": 2.2285783290863037, "learning_rate": 6.494605661780874e-06, "loss": 0.1365283489227295, "step": 23750 }, { "epoch": 0.6258746209289997, "grad_norm": 1.5051079988479614, "learning_rate": 6.486616898706365e-06, "loss": 0.13102661371231078, "step": 23760 }, { "epoch": 0.6261380361734985, "grad_norm": 1.3255470991134644, "learning_rate": 6.47863069282366e-06, "loss": 0.12624235153198243, "step": 23770 }, { "epoch": 0.6264014514179972, "grad_norm": 1.4270707368850708, "learning_rate": 6.47064704994546e-06, "loss": 0.1267557978630066, "step": 23780 }, { "epoch": 0.6266648666624959, "grad_norm": 1.4308260679244995, "learning_rate": 6.462665975882607e-06, "loss": 0.1366610646247864, "step": 23790 }, { "epoch": 0.6269282819069947, "grad_norm": 2.4969780445098877, "learning_rate": 6.454687476444071e-06, "loss": 0.13154151439666747, "step": 23800 }, { "epoch": 0.6271916971514934, "grad_norm": 1.627360224723816, "learning_rate": 6.446711557436944e-06, "loss": 0.11908701658248902, "step": 23810 }, { "epoch": 0.6274551123959922, "grad_norm": 1.7008200883865356, "learning_rate": 6.438738224666452e-06, "loss": 0.13665759563446045, "step": 23820 }, { "epoch": 0.6277185276404909, "grad_norm": 1.6215964555740356, "learning_rate": 6.4307674839359225e-06, "loss": 0.13394253253936766, "step": 23830 }, { "epoch": 0.6279819428849897, "grad_norm": 2.3973870277404785, "learning_rate": 6.42279934104681e-06, "loss": 0.1313725471496582, "step": 23840 }, { "epoch": 0.6282453581294883, "grad_norm": 2.5614023208618164, "learning_rate": 6.4148338017986725e-06, "loss": 0.12715213298797606, "step": 23850 }, { "epoch": 0.6285087733739871, "grad_norm": 1.6852234601974487, "learning_rate": 6.406870871989166e-06, "loss": 0.12283146381378174, "step": 23860 }, { "epoch": 0.6287721886184858, "grad_norm": 1.9976269006729126, "learning_rate": 6.398910557414064e-06, "loss": 0.13673146963119506, "step": 23870 }, { "epoch": 0.6290356038629845, "grad_norm": 1.8903156518936157, "learning_rate": 6.390952863867224e-06, "loss": 0.12278593778610229, "step": 23880 }, { "epoch": 0.6292990191074833, "grad_norm": 2.64963698387146, "learning_rate": 6.382997797140591e-06, "loss": 0.13851743936538696, "step": 23890 }, { "epoch": 0.629562434351982, "grad_norm": 1.6137570142745972, "learning_rate": 6.375045363024217e-06, "loss": 0.13014848232269288, "step": 23900 }, { "epoch": 0.6298258495964808, "grad_norm": 2.422612428665161, "learning_rate": 6.367095567306219e-06, "loss": 0.1384115219116211, "step": 23910 }, { "epoch": 0.6300892648409795, "grad_norm": 1.1889007091522217, "learning_rate": 6.359148415772805e-06, "loss": 0.11252074241638184, "step": 23920 }, { "epoch": 0.6303526800854783, "grad_norm": 2.2318878173828125, "learning_rate": 6.351203914208254e-06, "loss": 0.1394122362136841, "step": 23930 }, { "epoch": 0.630616095329977, "grad_norm": 1.1691190004348755, "learning_rate": 6.343262068394914e-06, "loss": 0.13707913160324098, "step": 23940 }, { "epoch": 0.6308795105744757, "grad_norm": 2.1827099323272705, "learning_rate": 6.335322884113208e-06, "loss": 0.14432525634765625, "step": 23950 }, { "epoch": 0.6311429258189745, "grad_norm": 3.422816276550293, "learning_rate": 6.3273863671416146e-06, "loss": 0.13551578521728516, "step": 23960 }, { "epoch": 0.6314063410634732, "grad_norm": 3.1348302364349365, "learning_rate": 6.319452523256673e-06, "loss": 0.13028421401977539, "step": 23970 }, { "epoch": 0.631669756307972, "grad_norm": 1.5547503232955933, "learning_rate": 6.31152135823298e-06, "loss": 0.1391129493713379, "step": 23980 }, { "epoch": 0.6319331715524706, "grad_norm": 1.0959995985031128, "learning_rate": 6.303592877843171e-06, "loss": 0.1317953109741211, "step": 23990 }, { "epoch": 0.6321965867969694, "grad_norm": 1.724988341331482, "learning_rate": 6.295667087857948e-06, "loss": 0.1295749068260193, "step": 24000 }, { "epoch": 0.6324600020414681, "grad_norm": 3.3054990768432617, "learning_rate": 6.287743994046035e-06, "loss": 0.12933266162872314, "step": 24010 }, { "epoch": 0.6327234172859669, "grad_norm": 4.245477199554443, "learning_rate": 6.279823602174202e-06, "loss": 0.12974278926849364, "step": 24020 }, { "epoch": 0.6329868325304656, "grad_norm": 2.8604722023010254, "learning_rate": 6.271905918007255e-06, "loss": 0.13997501134872437, "step": 24030 }, { "epoch": 0.6332502477749643, "grad_norm": 1.8221216201782227, "learning_rate": 6.263990947308022e-06, "loss": 0.1371808171272278, "step": 24040 }, { "epoch": 0.6335136630194631, "grad_norm": 1.0448698997497559, "learning_rate": 6.2560786958373645e-06, "loss": 0.13806909322738647, "step": 24050 }, { "epoch": 0.6337770782639618, "grad_norm": 1.36404287815094, "learning_rate": 6.248169169354159e-06, "loss": 0.1330324649810791, "step": 24060 }, { "epoch": 0.6340404935084606, "grad_norm": 1.6602745056152344, "learning_rate": 6.240262373615298e-06, "loss": 0.1280613899230957, "step": 24070 }, { "epoch": 0.6343039087529593, "grad_norm": 1.4272998571395874, "learning_rate": 6.232358314375693e-06, "loss": 0.11995165348052979, "step": 24080 }, { "epoch": 0.6345673239974581, "grad_norm": 2.1884028911590576, "learning_rate": 6.224456997388254e-06, "loss": 0.12345999479293823, "step": 24090 }, { "epoch": 0.6348307392419568, "grad_norm": 2.227618455886841, "learning_rate": 6.2165584284039046e-06, "loss": 0.12737948894500734, "step": 24100 }, { "epoch": 0.6350941544864556, "grad_norm": 3.1600940227508545, "learning_rate": 6.208662613171566e-06, "loss": 0.13346943855285645, "step": 24110 }, { "epoch": 0.6353575697309543, "grad_norm": 3.7033908367156982, "learning_rate": 6.200769557438144e-06, "loss": 0.13762266635894777, "step": 24120 }, { "epoch": 0.6356209849754529, "grad_norm": 1.253118634223938, "learning_rate": 6.192879266948554e-06, "loss": 0.1360788583755493, "step": 24130 }, { "epoch": 0.6358844002199517, "grad_norm": 1.3203182220458984, "learning_rate": 6.184991747445686e-06, "loss": 0.13832876682281495, "step": 24140 }, { "epoch": 0.6361478154644504, "grad_norm": 2.560349464416504, "learning_rate": 6.1771070046704165e-06, "loss": 0.12174944877624512, "step": 24150 }, { "epoch": 0.6364112307089492, "grad_norm": 1.6534390449523926, "learning_rate": 6.169225044361604e-06, "loss": 0.1372016429901123, "step": 24160 }, { "epoch": 0.6366746459534479, "grad_norm": 4.137404441833496, "learning_rate": 6.161345872256075e-06, "loss": 0.13803132772445678, "step": 24170 }, { "epoch": 0.6369380611979467, "grad_norm": 2.0106258392333984, "learning_rate": 6.153469494088638e-06, "loss": 0.13373289108276368, "step": 24180 }, { "epoch": 0.6372014764424454, "grad_norm": 1.6663179397583008, "learning_rate": 6.1455959155920555e-06, "loss": 0.12680363655090332, "step": 24190 }, { "epoch": 0.6374648916869442, "grad_norm": 1.831756830215454, "learning_rate": 6.1377251424970555e-06, "loss": 0.12424938678741455, "step": 24200 }, { "epoch": 0.6377283069314429, "grad_norm": 2.8127782344818115, "learning_rate": 6.1298571805323344e-06, "loss": 0.13134782314300536, "step": 24210 }, { "epoch": 0.6379917221759416, "grad_norm": 2.4022109508514404, "learning_rate": 6.121992035424529e-06, "loss": 0.1261478066444397, "step": 24220 }, { "epoch": 0.6382551374204404, "grad_norm": 1.6404449939727783, "learning_rate": 6.114129712898227e-06, "loss": 0.13752682209014894, "step": 24230 }, { "epoch": 0.6385185526649391, "grad_norm": 1.7709808349609375, "learning_rate": 6.106270218675975e-06, "loss": 0.1271979331970215, "step": 24240 }, { "epoch": 0.6387819679094379, "grad_norm": 2.088421106338501, "learning_rate": 6.098413558478244e-06, "loss": 0.13680522441864013, "step": 24250 }, { "epoch": 0.6390453831539366, "grad_norm": 1.3346798419952393, "learning_rate": 6.090559738023456e-06, "loss": 0.1322550058364868, "step": 24260 }, { "epoch": 0.6393087983984354, "grad_norm": 1.6157045364379883, "learning_rate": 6.0827087630279605e-06, "loss": 0.13618019819259644, "step": 24270 }, { "epoch": 0.639572213642934, "grad_norm": 1.6210790872573853, "learning_rate": 6.074860639206028e-06, "loss": 0.1288818597793579, "step": 24280 }, { "epoch": 0.6398356288874328, "grad_norm": 1.4276721477508545, "learning_rate": 6.067015372269872e-06, "loss": 0.1256876230239868, "step": 24290 }, { "epoch": 0.6400990441319315, "grad_norm": 2.0554749965667725, "learning_rate": 6.059172967929609e-06, "loss": 0.1284841537475586, "step": 24300 }, { "epoch": 0.6403624593764302, "grad_norm": 1.4668656587600708, "learning_rate": 6.0513334318932825e-06, "loss": 0.12317128181457519, "step": 24310 }, { "epoch": 0.640625874620929, "grad_norm": 1.8496456146240234, "learning_rate": 6.043496769866848e-06, "loss": 0.13468897342681885, "step": 24320 }, { "epoch": 0.6408892898654277, "grad_norm": 1.5227057933807373, "learning_rate": 6.035662987554154e-06, "loss": 0.12461931705474853, "step": 24330 }, { "epoch": 0.6411527051099265, "grad_norm": 1.8189256191253662, "learning_rate": 6.027832090656982e-06, "loss": 0.13242685794830322, "step": 24340 }, { "epoch": 0.6414161203544252, "grad_norm": 1.5173455476760864, "learning_rate": 6.020004084874984e-06, "loss": 0.13580559492111205, "step": 24350 }, { "epoch": 0.641679535598924, "grad_norm": 1.7864166498184204, "learning_rate": 6.012178975905725e-06, "loss": 0.12710734605789184, "step": 24360 }, { "epoch": 0.6419429508434227, "grad_norm": 2.0899741649627686, "learning_rate": 6.0043567694446566e-06, "loss": 0.11553562879562378, "step": 24370 }, { "epoch": 0.6422063660879215, "grad_norm": 1.798837423324585, "learning_rate": 5.9965374711851175e-06, "loss": 0.1283977746963501, "step": 24380 }, { "epoch": 0.6424697813324202, "grad_norm": 1.1978095769882202, "learning_rate": 5.9887210868183325e-06, "loss": 0.13371679782867432, "step": 24390 }, { "epoch": 0.6427331965769189, "grad_norm": 1.374138355255127, "learning_rate": 5.9809076220334024e-06, "loss": 0.13051991462707518, "step": 24400 }, { "epoch": 0.6429966118214177, "grad_norm": 2.272029161453247, "learning_rate": 5.973097082517301e-06, "loss": 0.1348263740539551, "step": 24410 }, { "epoch": 0.6432600270659163, "grad_norm": 3.602541208267212, "learning_rate": 5.965289473954879e-06, "loss": 0.11954959630966186, "step": 24420 }, { "epoch": 0.6435234423104151, "grad_norm": 2.0131843090057373, "learning_rate": 5.95748480202885e-06, "loss": 0.13434914350509644, "step": 24430 }, { "epoch": 0.6437868575549138, "grad_norm": 2.0701003074645996, "learning_rate": 5.949683072419792e-06, "loss": 0.12710556983947754, "step": 24440 }, { "epoch": 0.6440502727994126, "grad_norm": 2.0348830223083496, "learning_rate": 5.941884290806142e-06, "loss": 0.12920215129852294, "step": 24450 }, { "epoch": 0.6443136880439113, "grad_norm": 1.4433223009109497, "learning_rate": 5.934088462864181e-06, "loss": 0.13371219635009765, "step": 24460 }, { "epoch": 0.6445771032884101, "grad_norm": 1.8408464193344116, "learning_rate": 5.926295594268059e-06, "loss": 0.13458621501922607, "step": 24470 }, { "epoch": 0.6448405185329088, "grad_norm": 2.092369556427002, "learning_rate": 5.918505690689759e-06, "loss": 0.12987751960754396, "step": 24480 }, { "epoch": 0.6451039337774075, "grad_norm": 2.7278904914855957, "learning_rate": 5.910718757799105e-06, "loss": 0.1444784164428711, "step": 24490 }, { "epoch": 0.6453673490219063, "grad_norm": 1.8620538711547852, "learning_rate": 5.902934801263766e-06, "loss": 0.1400262475013733, "step": 24500 }, { "epoch": 0.645630764266405, "grad_norm": 1.0312343835830688, "learning_rate": 5.89515382674924e-06, "loss": 0.12466871738433838, "step": 24510 }, { "epoch": 0.6458941795109038, "grad_norm": 1.8427543640136719, "learning_rate": 5.88737583991886e-06, "loss": 0.15371588468551636, "step": 24520 }, { "epoch": 0.6461575947554025, "grad_norm": 1.9758057594299316, "learning_rate": 5.879600846433777e-06, "loss": 0.12534481287002563, "step": 24530 }, { "epoch": 0.6464210099999013, "grad_norm": 2.5196967124938965, "learning_rate": 5.871828851952965e-06, "loss": 0.14227464199066162, "step": 24540 }, { "epoch": 0.6466844252444, "grad_norm": 1.8154116868972778, "learning_rate": 5.864059862133221e-06, "loss": 0.12929396629333495, "step": 24550 }, { "epoch": 0.6469478404888986, "grad_norm": 1.346178650856018, "learning_rate": 5.856293882629147e-06, "loss": 0.11740961074829101, "step": 24560 }, { "epoch": 0.6472112557333974, "grad_norm": 1.839348554611206, "learning_rate": 5.848530919093162e-06, "loss": 0.1310088038444519, "step": 24570 }, { "epoch": 0.6474746709778961, "grad_norm": 1.875753402709961, "learning_rate": 5.840770977175485e-06, "loss": 0.1289583206176758, "step": 24580 }, { "epoch": 0.6477380862223949, "grad_norm": 2.4698779582977295, "learning_rate": 5.833014062524129e-06, "loss": 0.12414966821670533, "step": 24590 }, { "epoch": 0.6480015014668936, "grad_norm": 1.6359995603561401, "learning_rate": 5.825260180784923e-06, "loss": 0.12633421421051025, "step": 24600 }, { "epoch": 0.6482649167113924, "grad_norm": 2.162179470062256, "learning_rate": 5.817509337601475e-06, "loss": 0.12699227333068847, "step": 24610 }, { "epoch": 0.6485283319558911, "grad_norm": 1.784746527671814, "learning_rate": 5.8097615386151686e-06, "loss": 0.13216261863708495, "step": 24620 }, { "epoch": 0.6487917472003899, "grad_norm": 2.4121928215026855, "learning_rate": 5.8020167894652035e-06, "loss": 0.13587512969970703, "step": 24630 }, { "epoch": 0.6490551624448886, "grad_norm": 2.542466640472412, "learning_rate": 5.79427509578853e-06, "loss": 0.13314205408096313, "step": 24640 }, { "epoch": 0.6493185776893873, "grad_norm": 2.0856382846832275, "learning_rate": 5.786536463219899e-06, "loss": 0.1310916304588318, "step": 24650 }, { "epoch": 0.6495819929338861, "grad_norm": 1.694027066230774, "learning_rate": 5.778800897391812e-06, "loss": 0.12049665451049804, "step": 24660 }, { "epoch": 0.6498454081783848, "grad_norm": 1.1813323497772217, "learning_rate": 5.771068403934545e-06, "loss": 0.14300485849380493, "step": 24670 }, { "epoch": 0.6501088234228836, "grad_norm": 1.3493103981018066, "learning_rate": 5.76333898847615e-06, "loss": 0.12597864866256714, "step": 24680 }, { "epoch": 0.6503722386673823, "grad_norm": 1.4694185256958008, "learning_rate": 5.755612656642424e-06, "loss": 0.12433321475982666, "step": 24690 }, { "epoch": 0.650635653911881, "grad_norm": 3.042088031768799, "learning_rate": 5.747889414056925e-06, "loss": 0.12182512283325195, "step": 24700 }, { "epoch": 0.6508990691563797, "grad_norm": 2.1728391647338867, "learning_rate": 5.740169266340963e-06, "loss": 0.12224781513214111, "step": 24710 }, { "epoch": 0.6511624844008785, "grad_norm": 1.528135895729065, "learning_rate": 5.732452219113589e-06, "loss": 0.13370733261108397, "step": 24720 }, { "epoch": 0.6514258996453772, "grad_norm": 3.1132843494415283, "learning_rate": 5.724738277991611e-06, "loss": 0.1307976722717285, "step": 24730 }, { "epoch": 0.6516893148898759, "grad_norm": 2.563917875289917, "learning_rate": 5.717027448589568e-06, "loss": 0.14210879802703857, "step": 24740 }, { "epoch": 0.6519527301343747, "grad_norm": 1.6086431741714478, "learning_rate": 5.709319736519725e-06, "loss": 0.11860090494155884, "step": 24750 }, { "epoch": 0.6522161453788734, "grad_norm": 1.4324969053268433, "learning_rate": 5.701615147392096e-06, "loss": 0.14161217212677002, "step": 24760 }, { "epoch": 0.6524795606233722, "grad_norm": 1.7354272603988647, "learning_rate": 5.6939136868144075e-06, "loss": 0.13406260013580323, "step": 24770 }, { "epoch": 0.6527429758678709, "grad_norm": 1.3666917085647583, "learning_rate": 5.686215360392127e-06, "loss": 0.12092156410217285, "step": 24780 }, { "epoch": 0.6530063911123697, "grad_norm": 5.712231636047363, "learning_rate": 5.678520173728414e-06, "loss": 0.12562429904937744, "step": 24790 }, { "epoch": 0.6532698063568684, "grad_norm": 2.0803067684173584, "learning_rate": 5.67082813242416e-06, "loss": 0.13077110052108765, "step": 24800 }, { "epoch": 0.6535332216013672, "grad_norm": 1.2049423456192017, "learning_rate": 5.663139242077971e-06, "loss": 0.13381545543670653, "step": 24810 }, { "epoch": 0.6537966368458659, "grad_norm": 1.229687213897705, "learning_rate": 5.6554535082861496e-06, "loss": 0.1418265700340271, "step": 24820 }, { "epoch": 0.6540600520903646, "grad_norm": 2.3280253410339355, "learning_rate": 5.647770936642705e-06, "loss": 0.12455275058746337, "step": 24830 }, { "epoch": 0.6543234673348634, "grad_norm": 2.0843141078948975, "learning_rate": 5.6400915327393435e-06, "loss": 0.13337942361831664, "step": 24840 }, { "epoch": 0.654586882579362, "grad_norm": 1.3061219453811646, "learning_rate": 5.632415302165464e-06, "loss": 0.12887468338012695, "step": 24850 }, { "epoch": 0.6548502978238608, "grad_norm": 1.9063363075256348, "learning_rate": 5.624742250508164e-06, "loss": 0.12121727466583251, "step": 24860 }, { "epoch": 0.6551137130683595, "grad_norm": 1.555111050605774, "learning_rate": 5.617072383352222e-06, "loss": 0.12721095085144044, "step": 24870 }, { "epoch": 0.6553771283128583, "grad_norm": 1.5318697690963745, "learning_rate": 5.6094057062800886e-06, "loss": 0.11696898937225342, "step": 24880 }, { "epoch": 0.655640543557357, "grad_norm": 2.3066818714141846, "learning_rate": 5.601742224871911e-06, "loss": 0.13737345933914186, "step": 24890 }, { "epoch": 0.6559039588018558, "grad_norm": 2.5696446895599365, "learning_rate": 5.594081944705496e-06, "loss": 0.1331905245780945, "step": 24900 }, { "epoch": 0.6561673740463545, "grad_norm": 1.8279051780700684, "learning_rate": 5.586424871356337e-06, "loss": 0.12376551628112793, "step": 24910 }, { "epoch": 0.6564307892908532, "grad_norm": 1.7451695203781128, "learning_rate": 5.578771010397572e-06, "loss": 0.13452720642089844, "step": 24920 }, { "epoch": 0.656694204535352, "grad_norm": 1.4065666198730469, "learning_rate": 5.571120367400008e-06, "loss": 0.13088314533233641, "step": 24930 }, { "epoch": 0.6569576197798507, "grad_norm": 3.7273714542388916, "learning_rate": 5.563472947932123e-06, "loss": 0.1261131763458252, "step": 24940 }, { "epoch": 0.6572210350243495, "grad_norm": 2.202598810195923, "learning_rate": 5.555828757560035e-06, "loss": 0.1294947385787964, "step": 24950 }, { "epoch": 0.6574844502688482, "grad_norm": 1.2772908210754395, "learning_rate": 5.548187801847513e-06, "loss": 0.13418323993682862, "step": 24960 }, { "epoch": 0.657747865513347, "grad_norm": 1.2455147504806519, "learning_rate": 5.5405500863559795e-06, "loss": 0.12910327911376954, "step": 24970 }, { "epoch": 0.6580112807578457, "grad_norm": 1.9194133281707764, "learning_rate": 5.532915616644484e-06, "loss": 0.1291234612464905, "step": 24980 }, { "epoch": 0.6582746960023445, "grad_norm": 1.6292808055877686, "learning_rate": 5.525284398269733e-06, "loss": 0.12155508995056152, "step": 24990 }, { "epoch": 0.6585381112468431, "grad_norm": 1.9782670736312866, "learning_rate": 5.517656436786059e-06, "loss": 0.12570990324020387, "step": 25000 }, { "epoch": 0.6588015264913418, "grad_norm": 3.1713387966156006, "learning_rate": 5.510031737745405e-06, "loss": 0.13109610080718995, "step": 25010 }, { "epoch": 0.6590649417358406, "grad_norm": 2.06780743598938, "learning_rate": 5.50241030669737e-06, "loss": 0.13122116327285765, "step": 25020 }, { "epoch": 0.6593283569803393, "grad_norm": 1.8963149785995483, "learning_rate": 5.4947921491891565e-06, "loss": 0.13014676570892333, "step": 25030 }, { "epoch": 0.6595917722248381, "grad_norm": 1.4531915187835693, "learning_rate": 5.487177270765588e-06, "loss": 0.1266641616821289, "step": 25040 }, { "epoch": 0.6598551874693368, "grad_norm": 1.696946620941162, "learning_rate": 5.479565676969101e-06, "loss": 0.12354198694229127, "step": 25050 }, { "epoch": 0.6601186027138356, "grad_norm": 1.2855199575424194, "learning_rate": 5.47195737333974e-06, "loss": 0.11868727207183838, "step": 25060 }, { "epoch": 0.6603820179583343, "grad_norm": 1.4470393657684326, "learning_rate": 5.464352365415162e-06, "loss": 0.1257412075996399, "step": 25070 }, { "epoch": 0.660645433202833, "grad_norm": 1.6042226552963257, "learning_rate": 5.456750658730616e-06, "loss": 0.13276149034500123, "step": 25080 }, { "epoch": 0.6609088484473318, "grad_norm": 1.606757640838623, "learning_rate": 5.449152258818955e-06, "loss": 0.12110370397567749, "step": 25090 }, { "epoch": 0.6611722636918305, "grad_norm": 3.2371175289154053, "learning_rate": 5.4415571712106184e-06, "loss": 0.13425033092498778, "step": 25100 }, { "epoch": 0.6614356789363293, "grad_norm": 2.1244049072265625, "learning_rate": 5.433965401433639e-06, "loss": 0.12086924314498901, "step": 25110 }, { "epoch": 0.661699094180828, "grad_norm": 1.784631609916687, "learning_rate": 5.426376955013641e-06, "loss": 0.12270606756210327, "step": 25120 }, { "epoch": 0.6619625094253268, "grad_norm": 1.9316493272781372, "learning_rate": 5.4187918374738154e-06, "loss": 0.1349888801574707, "step": 25130 }, { "epoch": 0.6622259246698254, "grad_norm": 1.0944693088531494, "learning_rate": 5.4112100543349365e-06, "loss": 0.12086101770401, "step": 25140 }, { "epoch": 0.6624893399143242, "grad_norm": 1.9021093845367432, "learning_rate": 5.403631611115362e-06, "loss": 0.12636126279830934, "step": 25150 }, { "epoch": 0.6627527551588229, "grad_norm": 1.8203885555267334, "learning_rate": 5.396056513331005e-06, "loss": 0.1231919527053833, "step": 25160 }, { "epoch": 0.6630161704033216, "grad_norm": 2.1627469062805176, "learning_rate": 5.388484766495348e-06, "loss": 0.14149143695831298, "step": 25170 }, { "epoch": 0.6632795856478204, "grad_norm": 3.0342249870300293, "learning_rate": 5.3809163761194364e-06, "loss": 0.13779067993164062, "step": 25180 }, { "epoch": 0.6635430008923191, "grad_norm": 1.171353816986084, "learning_rate": 5.373351347711865e-06, "loss": 0.10944166183471679, "step": 25190 }, { "epoch": 0.6638064161368179, "grad_norm": 2.5788824558258057, "learning_rate": 5.365789686778797e-06, "loss": 0.13720269203186036, "step": 25200 }, { "epoch": 0.6640698313813166, "grad_norm": 1.9044944047927856, "learning_rate": 5.358231398823937e-06, "loss": 0.12338168621063232, "step": 25210 }, { "epoch": 0.6643332466258154, "grad_norm": 1.9700770378112793, "learning_rate": 5.350676489348521e-06, "loss": 0.12927050590515138, "step": 25220 }, { "epoch": 0.6645966618703141, "grad_norm": 2.744641065597534, "learning_rate": 5.3431249638513475e-06, "loss": 0.1352560043334961, "step": 25230 }, { "epoch": 0.6648600771148129, "grad_norm": 1.3122572898864746, "learning_rate": 5.335576827828737e-06, "loss": 0.12737841606140138, "step": 25240 }, { "epoch": 0.6651234923593116, "grad_norm": 2.4719221591949463, "learning_rate": 5.328032086774557e-06, "loss": 0.12473784685134888, "step": 25250 }, { "epoch": 0.6653869076038103, "grad_norm": 2.3306639194488525, "learning_rate": 5.320490746180189e-06, "loss": 0.1166311264038086, "step": 25260 }, { "epoch": 0.665650322848309, "grad_norm": 1.9911569356918335, "learning_rate": 5.312952811534542e-06, "loss": 0.12518186569213868, "step": 25270 }, { "epoch": 0.6659137380928077, "grad_norm": 1.8676825761795044, "learning_rate": 5.305418288324058e-06, "loss": 0.12478146553039551, "step": 25280 }, { "epoch": 0.6661771533373065, "grad_norm": 2.3110830783843994, "learning_rate": 5.297887182032686e-06, "loss": 0.13054546117782592, "step": 25290 }, { "epoch": 0.6664405685818052, "grad_norm": 2.1304142475128174, "learning_rate": 5.29035949814189e-06, "loss": 0.1171417474746704, "step": 25300 }, { "epoch": 0.666703983826304, "grad_norm": 2.4559216499328613, "learning_rate": 5.2828352421306415e-06, "loss": 0.13195037841796875, "step": 25310 }, { "epoch": 0.6669673990708027, "grad_norm": 1.4834100008010864, "learning_rate": 5.275314419475417e-06, "loss": 0.11782493591308593, "step": 25320 }, { "epoch": 0.6672308143153015, "grad_norm": 2.5889475345611572, "learning_rate": 5.267797035650202e-06, "loss": 0.12402417659759521, "step": 25330 }, { "epoch": 0.6674942295598002, "grad_norm": 1.3680787086486816, "learning_rate": 5.2602830961264704e-06, "loss": 0.13225946426391602, "step": 25340 }, { "epoch": 0.6677576448042989, "grad_norm": 1.9655052423477173, "learning_rate": 5.252772606373186e-06, "loss": 0.12484322786331177, "step": 25350 }, { "epoch": 0.6680210600487977, "grad_norm": 1.7021794319152832, "learning_rate": 5.245265571856812e-06, "loss": 0.11781144142150879, "step": 25360 }, { "epoch": 0.6682844752932964, "grad_norm": 2.7563509941101074, "learning_rate": 5.237761998041291e-06, "loss": 0.12861459255218505, "step": 25370 }, { "epoch": 0.6685478905377952, "grad_norm": 1.2990914583206177, "learning_rate": 5.230261890388053e-06, "loss": 0.13863215446472169, "step": 25380 }, { "epoch": 0.6688113057822939, "grad_norm": 1.654746413230896, "learning_rate": 5.2227652543559924e-06, "loss": 0.12575931549072267, "step": 25390 }, { "epoch": 0.6690747210267927, "grad_norm": 2.800497055053711, "learning_rate": 5.215272095401483e-06, "loss": 0.12697529792785645, "step": 25400 }, { "epoch": 0.6693381362712914, "grad_norm": 1.5713207721710205, "learning_rate": 5.207782418978378e-06, "loss": 0.122649085521698, "step": 25410 }, { "epoch": 0.6696015515157901, "grad_norm": 1.619488000869751, "learning_rate": 5.200296230537982e-06, "loss": 0.12497291564941407, "step": 25420 }, { "epoch": 0.6698649667602888, "grad_norm": 1.6509993076324463, "learning_rate": 5.192813535529066e-06, "loss": 0.1380634903907776, "step": 25430 }, { "epoch": 0.6701283820047875, "grad_norm": 1.3368176221847534, "learning_rate": 5.1853343393978585e-06, "loss": 0.11691343784332275, "step": 25440 }, { "epoch": 0.6703917972492863, "grad_norm": 1.8093023300170898, "learning_rate": 5.177858647588037e-06, "loss": 0.13117260932922364, "step": 25450 }, { "epoch": 0.670655212493785, "grad_norm": 1.6703089475631714, "learning_rate": 5.170386465540741e-06, "loss": 0.1219447135925293, "step": 25460 }, { "epoch": 0.6709186277382838, "grad_norm": 2.0706889629364014, "learning_rate": 5.162917798694546e-06, "loss": 0.12395532131195068, "step": 25470 }, { "epoch": 0.6711820429827825, "grad_norm": 1.3583873510360718, "learning_rate": 5.15545265248546e-06, "loss": 0.13142640590667726, "step": 25480 }, { "epoch": 0.6714454582272813, "grad_norm": 1.4103294610977173, "learning_rate": 5.147991032346948e-06, "loss": 0.12016454935073853, "step": 25490 }, { "epoch": 0.67170887347178, "grad_norm": 1.8333032131195068, "learning_rate": 5.140532943709893e-06, "loss": 0.1298065185546875, "step": 25500 }, { "epoch": 0.6719722887162788, "grad_norm": 1.6899319887161255, "learning_rate": 5.133078392002623e-06, "loss": 0.13012888431549072, "step": 25510 }, { "epoch": 0.6722357039607775, "grad_norm": 1.82323157787323, "learning_rate": 5.125627382650873e-06, "loss": 0.12062652111053467, "step": 25520 }, { "epoch": 0.6724991192052762, "grad_norm": 2.7338321208953857, "learning_rate": 5.118179921077808e-06, "loss": 0.1372469663619995, "step": 25530 }, { "epoch": 0.672762534449775, "grad_norm": 2.0487847328186035, "learning_rate": 5.1107360127040206e-06, "loss": 0.13580604791641235, "step": 25540 }, { "epoch": 0.6730259496942737, "grad_norm": 1.8529961109161377, "learning_rate": 5.103295662947504e-06, "loss": 0.12588456869125367, "step": 25550 }, { "epoch": 0.6732893649387724, "grad_norm": 3.0763297080993652, "learning_rate": 5.095858877223665e-06, "loss": 0.11587395668029785, "step": 25560 }, { "epoch": 0.6735527801832711, "grad_norm": 1.2763031721115112, "learning_rate": 5.088425660945319e-06, "loss": 0.12391849756240844, "step": 25570 }, { "epoch": 0.6738161954277699, "grad_norm": 1.5106593370437622, "learning_rate": 5.080996019522675e-06, "loss": 0.11506240367889405, "step": 25580 }, { "epoch": 0.6740796106722686, "grad_norm": 4.416794300079346, "learning_rate": 5.073569958363356e-06, "loss": 0.13482291698455812, "step": 25590 }, { "epoch": 0.6743430259167673, "grad_norm": 1.5156432390213013, "learning_rate": 5.06614748287237e-06, "loss": 0.11483114957809448, "step": 25600 }, { "epoch": 0.6746064411612661, "grad_norm": 1.918158769607544, "learning_rate": 5.058728598452103e-06, "loss": 0.13551087379455568, "step": 25610 }, { "epoch": 0.6748698564057648, "grad_norm": 1.7390077114105225, "learning_rate": 5.05131331050235e-06, "loss": 0.11405411958694459, "step": 25620 }, { "epoch": 0.6751332716502636, "grad_norm": 1.428131341934204, "learning_rate": 5.043901624420273e-06, "loss": 0.1392799973487854, "step": 25630 }, { "epoch": 0.6753966868947623, "grad_norm": 2.8325181007385254, "learning_rate": 5.036493545600421e-06, "loss": 0.1230050802230835, "step": 25640 }, { "epoch": 0.6756601021392611, "grad_norm": 1.53987717628479, "learning_rate": 5.029089079434708e-06, "loss": 0.11534477472305298, "step": 25650 }, { "epoch": 0.6759235173837598, "grad_norm": 2.5627479553222656, "learning_rate": 5.021688231312423e-06, "loss": 0.11439530849456787, "step": 25660 }, { "epoch": 0.6761869326282586, "grad_norm": 1.742884874343872, "learning_rate": 5.014291006620229e-06, "loss": 0.12806396484375, "step": 25670 }, { "epoch": 0.6764503478727573, "grad_norm": 1.2483717203140259, "learning_rate": 5.006897410742142e-06, "loss": 0.12371904850006103, "step": 25680 }, { "epoch": 0.676713763117256, "grad_norm": 1.4000067710876465, "learning_rate": 4.999507449059542e-06, "loss": 0.12417756319046021, "step": 25690 }, { "epoch": 0.6769771783617548, "grad_norm": 1.272212028503418, "learning_rate": 4.992121126951159e-06, "loss": 0.1285032272338867, "step": 25700 }, { "epoch": 0.6772405936062534, "grad_norm": 1.9524126052856445, "learning_rate": 4.984738449793074e-06, "loss": 0.1245455265045166, "step": 25710 }, { "epoch": 0.6775040088507522, "grad_norm": 1.5634527206420898, "learning_rate": 4.977359422958731e-06, "loss": 0.12452538013458252, "step": 25720 }, { "epoch": 0.6777674240952509, "grad_norm": 1.2464637756347656, "learning_rate": 4.96998405181889e-06, "loss": 0.12708033323287965, "step": 25730 }, { "epoch": 0.6780308393397497, "grad_norm": 2.0223917961120605, "learning_rate": 4.962612341741666e-06, "loss": 0.11389310359954834, "step": 25740 }, { "epoch": 0.6782942545842484, "grad_norm": 1.86430025100708, "learning_rate": 4.955244298092514e-06, "loss": 0.11981027126312256, "step": 25750 }, { "epoch": 0.6785576698287472, "grad_norm": 1.4811643362045288, "learning_rate": 4.947879926234212e-06, "loss": 0.1293141722679138, "step": 25760 }, { "epoch": 0.6788210850732459, "grad_norm": 2.7949416637420654, "learning_rate": 4.940519231526867e-06, "loss": 0.12884972095489503, "step": 25770 }, { "epoch": 0.6790845003177446, "grad_norm": 1.2332656383514404, "learning_rate": 4.933162219327909e-06, "loss": 0.11381412744522094, "step": 25780 }, { "epoch": 0.6793479155622434, "grad_norm": 2.008965253829956, "learning_rate": 4.925808894992087e-06, "loss": 0.12360856533050538, "step": 25790 }, { "epoch": 0.6796113308067421, "grad_norm": 1.8749215602874756, "learning_rate": 4.918459263871474e-06, "loss": 0.1269216299057007, "step": 25800 }, { "epoch": 0.6798747460512409, "grad_norm": 2.1946496963500977, "learning_rate": 4.911113331315448e-06, "loss": 0.1101183533668518, "step": 25810 }, { "epoch": 0.6801381612957396, "grad_norm": 4.422580242156982, "learning_rate": 4.903771102670686e-06, "loss": 0.12049601078033448, "step": 25820 }, { "epoch": 0.6804015765402384, "grad_norm": 1.8093156814575195, "learning_rate": 4.896432583281189e-06, "loss": 0.12078680992126464, "step": 25830 }, { "epoch": 0.680664991784737, "grad_norm": 1.5898609161376953, "learning_rate": 4.88909777848824e-06, "loss": 0.13151074647903443, "step": 25840 }, { "epoch": 0.6809284070292358, "grad_norm": 2.056182384490967, "learning_rate": 4.881766693630439e-06, "loss": 0.13233299255371095, "step": 25850 }, { "epoch": 0.6811918222737345, "grad_norm": 1.807580828666687, "learning_rate": 4.874439334043654e-06, "loss": 0.12474935054779053, "step": 25860 }, { "epoch": 0.6814552375182332, "grad_norm": 1.2002320289611816, "learning_rate": 4.8671157050610525e-06, "loss": 0.12797946929931642, "step": 25870 }, { "epoch": 0.681718652762732, "grad_norm": 2.351151704788208, "learning_rate": 4.859795812013099e-06, "loss": 0.14106898307800292, "step": 25880 }, { "epoch": 0.6819820680072307, "grad_norm": 2.382876396179199, "learning_rate": 4.852479660227518e-06, "loss": 0.1298104405403137, "step": 25890 }, { "epoch": 0.6822454832517295, "grad_norm": 2.5652520656585693, "learning_rate": 4.8451672550293236e-06, "loss": 0.1105840802192688, "step": 25900 }, { "epoch": 0.6825088984962282, "grad_norm": 2.1069867610931396, "learning_rate": 4.837858601740799e-06, "loss": 0.12068567276000977, "step": 25910 }, { "epoch": 0.682772313740727, "grad_norm": 1.9716886281967163, "learning_rate": 4.830553705681496e-06, "loss": 0.12574610710144044, "step": 25920 }, { "epoch": 0.6830357289852257, "grad_norm": 1.7726651430130005, "learning_rate": 4.823252572168236e-06, "loss": 0.1271679639816284, "step": 25930 }, { "epoch": 0.6832991442297245, "grad_norm": 1.2279834747314453, "learning_rate": 4.815955206515104e-06, "loss": 0.11919870376586914, "step": 25940 }, { "epoch": 0.6835625594742232, "grad_norm": 3.405165672302246, "learning_rate": 4.808661614033422e-06, "loss": 0.12636829614639283, "step": 25950 }, { "epoch": 0.6838259747187219, "grad_norm": 1.215681791305542, "learning_rate": 4.801371800031796e-06, "loss": 0.12469648122787476, "step": 25960 }, { "epoch": 0.6840893899632207, "grad_norm": 2.108569860458374, "learning_rate": 4.794085769816057e-06, "loss": 0.10692461729049682, "step": 25970 }, { "epoch": 0.6843528052077194, "grad_norm": 1.714445948600769, "learning_rate": 4.786803528689304e-06, "loss": 0.13543431758880614, "step": 25980 }, { "epoch": 0.6846162204522181, "grad_norm": 3.1820855140686035, "learning_rate": 4.779525081951857e-06, "loss": 0.1373348593711853, "step": 25990 }, { "epoch": 0.6848796356967168, "grad_norm": 2.116178512573242, "learning_rate": 4.772250434901281e-06, "loss": 0.12168015241622925, "step": 26000 }, { "epoch": 0.6851430509412156, "grad_norm": 1.7399358749389648, "learning_rate": 4.7649795928323875e-06, "loss": 0.11371382474899291, "step": 26010 }, { "epoch": 0.6854064661857143, "grad_norm": 1.4321684837341309, "learning_rate": 4.757712561037203e-06, "loss": 0.12620742321014405, "step": 26020 }, { "epoch": 0.6856698814302131, "grad_norm": 2.6821742057800293, "learning_rate": 4.750449344804989e-06, "loss": 0.12006151676177979, "step": 26030 }, { "epoch": 0.6859332966747118, "grad_norm": 4.946223258972168, "learning_rate": 4.743189949422227e-06, "loss": 0.14459127187728882, "step": 26040 }, { "epoch": 0.6861967119192105, "grad_norm": 2.0664851665496826, "learning_rate": 4.735934380172613e-06, "loss": 0.12284313440322876, "step": 26050 }, { "epoch": 0.6864601271637093, "grad_norm": 2.0438828468322754, "learning_rate": 4.728682642337072e-06, "loss": 0.11939235925674438, "step": 26060 }, { "epoch": 0.686723542408208, "grad_norm": 1.8481909036636353, "learning_rate": 4.7214347411937325e-06, "loss": 0.1260448455810547, "step": 26070 }, { "epoch": 0.6869869576527068, "grad_norm": 1.3179068565368652, "learning_rate": 4.714190682017918e-06, "loss": 0.12346014976501465, "step": 26080 }, { "epoch": 0.6872503728972055, "grad_norm": 2.2629871368408203, "learning_rate": 4.706950470082177e-06, "loss": 0.12489397525787353, "step": 26090 }, { "epoch": 0.6875137881417043, "grad_norm": 2.063953399658203, "learning_rate": 4.699714110656242e-06, "loss": 0.1154388427734375, "step": 26100 }, { "epoch": 0.687777203386203, "grad_norm": 2.150942087173462, "learning_rate": 4.6924816090070575e-06, "loss": 0.12156517505645752, "step": 26110 }, { "epoch": 0.6880406186307017, "grad_norm": 2.355780601501465, "learning_rate": 4.685252970398742e-06, "loss": 0.13822814226150512, "step": 26120 }, { "epoch": 0.6883040338752004, "grad_norm": 1.7914671897888184, "learning_rate": 4.678028200092607e-06, "loss": 0.12704215049743653, "step": 26130 }, { "epoch": 0.6885674491196991, "grad_norm": 1.7357968091964722, "learning_rate": 4.670807303347161e-06, "loss": 0.12705230712890625, "step": 26140 }, { "epoch": 0.6888308643641979, "grad_norm": 3.4553778171539307, "learning_rate": 4.663590285418082e-06, "loss": 0.12357593774795532, "step": 26150 }, { "epoch": 0.6890942796086966, "grad_norm": 2.3434877395629883, "learning_rate": 4.656377151558224e-06, "loss": 0.1068561315536499, "step": 26160 }, { "epoch": 0.6893576948531954, "grad_norm": 2.8103110790252686, "learning_rate": 4.64916790701762e-06, "loss": 0.13464488983154296, "step": 26170 }, { "epoch": 0.6896211100976941, "grad_norm": 2.6613752841949463, "learning_rate": 4.641962557043462e-06, "loss": 0.11191914081573487, "step": 26180 }, { "epoch": 0.6898845253421929, "grad_norm": 2.0675411224365234, "learning_rate": 4.634761106880127e-06, "loss": 0.1174379825592041, "step": 26190 }, { "epoch": 0.6901479405866916, "grad_norm": 1.9495398998260498, "learning_rate": 4.627563561769136e-06, "loss": 0.12114928960800171, "step": 26200 }, { "epoch": 0.6904113558311903, "grad_norm": 1.6473041772842407, "learning_rate": 4.620369926949164e-06, "loss": 0.11989161968231202, "step": 26210 }, { "epoch": 0.6906747710756891, "grad_norm": 1.6470181941986084, "learning_rate": 4.613180207656061e-06, "loss": 0.1065227746963501, "step": 26220 }, { "epoch": 0.6909381863201878, "grad_norm": 1.3795465230941772, "learning_rate": 4.60599440912281e-06, "loss": 0.12378441095352173, "step": 26230 }, { "epoch": 0.6912016015646866, "grad_norm": 2.3694615364074707, "learning_rate": 4.598812536579545e-06, "loss": 0.11974658966064453, "step": 26240 }, { "epoch": 0.6914650168091853, "grad_norm": 1.6323717832565308, "learning_rate": 4.591634595253541e-06, "loss": 0.12075891494750976, "step": 26250 }, { "epoch": 0.6917284320536841, "grad_norm": 1.9604911804199219, "learning_rate": 4.584460590369214e-06, "loss": 0.1291203022003174, "step": 26260 }, { "epoch": 0.6919918472981827, "grad_norm": 2.3474843502044678, "learning_rate": 4.577290527148118e-06, "loss": 0.11758077144622803, "step": 26270 }, { "epoch": 0.6922552625426815, "grad_norm": 2.0430150032043457, "learning_rate": 4.5701244108089325e-06, "loss": 0.1278796076774597, "step": 26280 }, { "epoch": 0.6925186777871802, "grad_norm": 1.4421242475509644, "learning_rate": 4.562962246567466e-06, "loss": 0.12938508987426758, "step": 26290 }, { "epoch": 0.6927820930316789, "grad_norm": 2.9324119091033936, "learning_rate": 4.555804039636649e-06, "loss": 0.12735079526901244, "step": 26300 }, { "epoch": 0.6930455082761777, "grad_norm": 5.1951212882995605, "learning_rate": 4.548649795226532e-06, "loss": 0.12643522024154663, "step": 26310 }, { "epoch": 0.6933089235206764, "grad_norm": 1.3423683643341064, "learning_rate": 4.541499518544292e-06, "loss": 0.12079728841781616, "step": 26320 }, { "epoch": 0.6935723387651752, "grad_norm": 2.5702064037323, "learning_rate": 4.5343532147941985e-06, "loss": 0.1177213191986084, "step": 26330 }, { "epoch": 0.6938357540096739, "grad_norm": 2.129239797592163, "learning_rate": 4.527210889177639e-06, "loss": 0.1335367202758789, "step": 26340 }, { "epoch": 0.6940991692541727, "grad_norm": 3.814293146133423, "learning_rate": 4.520072546893115e-06, "loss": 0.12976784706115724, "step": 26350 }, { "epoch": 0.6943625844986714, "grad_norm": 1.5025544166564941, "learning_rate": 4.512938193136216e-06, "loss": 0.11609649658203125, "step": 26360 }, { "epoch": 0.6946259997431702, "grad_norm": 1.269335150718689, "learning_rate": 4.5058078330996315e-06, "loss": 0.12822502851486206, "step": 26370 }, { "epoch": 0.6948894149876689, "grad_norm": 2.130838394165039, "learning_rate": 4.498681471973145e-06, "loss": 0.11332626342773437, "step": 26380 }, { "epoch": 0.6951528302321676, "grad_norm": 2.109058380126953, "learning_rate": 4.491559114943628e-06, "loss": 0.1186053991317749, "step": 26390 }, { "epoch": 0.6954162454766664, "grad_norm": 2.6686954498291016, "learning_rate": 4.484440767195044e-06, "loss": 0.13588383197784423, "step": 26400 }, { "epoch": 0.695679660721165, "grad_norm": 1.8459012508392334, "learning_rate": 4.477326433908435e-06, "loss": 0.1243659496307373, "step": 26410 }, { "epoch": 0.6959430759656638, "grad_norm": 1.231094241142273, "learning_rate": 4.4702161202619085e-06, "loss": 0.12031877040863037, "step": 26420 }, { "epoch": 0.6962064912101625, "grad_norm": 1.2308181524276733, "learning_rate": 4.463109831430668e-06, "loss": 0.12340245246887208, "step": 26430 }, { "epoch": 0.6964699064546613, "grad_norm": 2.132349729537964, "learning_rate": 4.456007572586971e-06, "loss": 0.1182745099067688, "step": 26440 }, { "epoch": 0.69673332169916, "grad_norm": 2.991576671600342, "learning_rate": 4.448909348900159e-06, "loss": 0.11971229314804077, "step": 26450 }, { "epoch": 0.6969967369436588, "grad_norm": 2.0846188068389893, "learning_rate": 4.441815165536612e-06, "loss": 0.13808531761169435, "step": 26460 }, { "epoch": 0.6972601521881575, "grad_norm": 1.958300232887268, "learning_rate": 4.434725027659786e-06, "loss": 0.11723496913909912, "step": 26470 }, { "epoch": 0.6975235674326562, "grad_norm": 2.168423652648926, "learning_rate": 4.427638940430192e-06, "loss": 0.12141797542572022, "step": 26480 }, { "epoch": 0.697786982677155, "grad_norm": 2.052990674972534, "learning_rate": 4.420556909005392e-06, "loss": 0.12376704216003417, "step": 26490 }, { "epoch": 0.6980503979216537, "grad_norm": 3.2533085346221924, "learning_rate": 4.41347893853999e-06, "loss": 0.12941720485687255, "step": 26500 }, { "epoch": 0.6983138131661525, "grad_norm": 3.2121074199676514, "learning_rate": 4.40640503418564e-06, "loss": 0.11420257091522217, "step": 26510 }, { "epoch": 0.6985772284106512, "grad_norm": 1.1480803489685059, "learning_rate": 4.39933520109103e-06, "loss": 0.11270418167114257, "step": 26520 }, { "epoch": 0.69884064365515, "grad_norm": 3.4381442070007324, "learning_rate": 4.3922694444018985e-06, "loss": 0.12457791566848755, "step": 26530 }, { "epoch": 0.6991040588996487, "grad_norm": 2.2118258476257324, "learning_rate": 4.385207769261007e-06, "loss": 0.1121901273727417, "step": 26540 }, { "epoch": 0.6993674741441475, "grad_norm": 2.79447603225708, "learning_rate": 4.378150180808136e-06, "loss": 0.12083771228790283, "step": 26550 }, { "epoch": 0.6996308893886461, "grad_norm": 1.7967365980148315, "learning_rate": 4.371096684180114e-06, "loss": 0.12458817958831787, "step": 26560 }, { "epoch": 0.6998943046331448, "grad_norm": 2.0166683197021484, "learning_rate": 4.364047284510774e-06, "loss": 0.10277330875396729, "step": 26570 }, { "epoch": 0.7001577198776436, "grad_norm": 2.0267763137817383, "learning_rate": 4.357001986930982e-06, "loss": 0.11925275325775146, "step": 26580 }, { "epoch": 0.7004211351221423, "grad_norm": 3.7672572135925293, "learning_rate": 4.349960796568601e-06, "loss": 0.13294790983200072, "step": 26590 }, { "epoch": 0.7006845503666411, "grad_norm": 2.1720027923583984, "learning_rate": 4.342923718548512e-06, "loss": 0.11752657890319824, "step": 26600 }, { "epoch": 0.7009479656111398, "grad_norm": 2.058204412460327, "learning_rate": 4.335890757992611e-06, "loss": 0.11439242362976074, "step": 26610 }, { "epoch": 0.7012113808556386, "grad_norm": 1.4568963050842285, "learning_rate": 4.328861920019787e-06, "loss": 0.11525685787200927, "step": 26620 }, { "epoch": 0.7014747961001373, "grad_norm": 3.662971019744873, "learning_rate": 4.321837209745931e-06, "loss": 0.12142734527587891, "step": 26630 }, { "epoch": 0.701738211344636, "grad_norm": 1.5251866579055786, "learning_rate": 4.3148166322839305e-06, "loss": 0.11993449926376343, "step": 26640 }, { "epoch": 0.7020016265891348, "grad_norm": 2.085624933242798, "learning_rate": 4.3078001927436605e-06, "loss": 0.10990924835205078, "step": 26650 }, { "epoch": 0.7022650418336335, "grad_norm": 1.9079967737197876, "learning_rate": 4.300787896231997e-06, "loss": 0.11731863021850586, "step": 26660 }, { "epoch": 0.7025284570781323, "grad_norm": 3.1984214782714844, "learning_rate": 4.293779747852791e-06, "loss": 0.12214882373809814, "step": 26670 }, { "epoch": 0.702791872322631, "grad_norm": 3.5481972694396973, "learning_rate": 4.286775752706865e-06, "loss": 0.11891759634017944, "step": 26680 }, { "epoch": 0.7030552875671298, "grad_norm": 2.511335849761963, "learning_rate": 4.279775915892038e-06, "loss": 0.12528226375579835, "step": 26690 }, { "epoch": 0.7033187028116284, "grad_norm": 1.7737181186676025, "learning_rate": 4.2727802425030885e-06, "loss": 0.10950570106506348, "step": 26700 }, { "epoch": 0.7035821180561272, "grad_norm": 1.5914318561553955, "learning_rate": 4.265788737631779e-06, "loss": 0.1068609118461609, "step": 26710 }, { "epoch": 0.7038455333006259, "grad_norm": 1.3537935018539429, "learning_rate": 4.258801406366817e-06, "loss": 0.12202236652374268, "step": 26720 }, { "epoch": 0.7041089485451246, "grad_norm": 2.957810640335083, "learning_rate": 4.251818253793885e-06, "loss": 0.1261681079864502, "step": 26730 }, { "epoch": 0.7043723637896234, "grad_norm": 2.1929779052734375, "learning_rate": 4.244839284995627e-06, "loss": 0.12943620681762696, "step": 26740 }, { "epoch": 0.7046357790341221, "grad_norm": 2.0407323837280273, "learning_rate": 4.237864505051638e-06, "loss": 0.11938461065292358, "step": 26750 }, { "epoch": 0.7048991942786209, "grad_norm": 2.077540874481201, "learning_rate": 4.230893919038461e-06, "loss": 0.1302335262298584, "step": 26760 }, { "epoch": 0.7051626095231196, "grad_norm": 2.145817518234253, "learning_rate": 4.223927532029592e-06, "loss": 0.11161129474639893, "step": 26770 }, { "epoch": 0.7054260247676184, "grad_norm": 1.7028160095214844, "learning_rate": 4.21696534909546e-06, "loss": 0.13086264133453368, "step": 26780 }, { "epoch": 0.7056894400121171, "grad_norm": 1.6992765665054321, "learning_rate": 4.2100073753034535e-06, "loss": 0.12247508764266968, "step": 26790 }, { "epoch": 0.7059528552566159, "grad_norm": 1.8617280721664429, "learning_rate": 4.203053615717884e-06, "loss": 0.12427725791931152, "step": 26800 }, { "epoch": 0.7062162705011146, "grad_norm": 1.51960027217865, "learning_rate": 4.196104075399986e-06, "loss": 0.12828404903411866, "step": 26810 }, { "epoch": 0.7064796857456133, "grad_norm": 1.2900197505950928, "learning_rate": 4.1891587594079485e-06, "loss": 0.1273495078086853, "step": 26820 }, { "epoch": 0.7067431009901121, "grad_norm": 1.701558232307434, "learning_rate": 4.182217672796867e-06, "loss": 0.1243804931640625, "step": 26830 }, { "epoch": 0.7070065162346107, "grad_norm": 1.723868489265442, "learning_rate": 4.175280820618763e-06, "loss": 0.11460634469985961, "step": 26840 }, { "epoch": 0.7072699314791095, "grad_norm": 1.3977923393249512, "learning_rate": 4.168348207922577e-06, "loss": 0.10018012523651124, "step": 26850 }, { "epoch": 0.7075333467236082, "grad_norm": 2.301748514175415, "learning_rate": 4.16141983975416e-06, "loss": 0.12059073448181153, "step": 26860 }, { "epoch": 0.707796761968107, "grad_norm": 3.1372742652893066, "learning_rate": 4.154495721156287e-06, "loss": 0.12499253749847412, "step": 26870 }, { "epoch": 0.7080601772126057, "grad_norm": 2.6181466579437256, "learning_rate": 4.147575857168623e-06, "loss": 0.12180033922195435, "step": 26880 }, { "epoch": 0.7083235924571045, "grad_norm": 1.5796067714691162, "learning_rate": 4.140660252827746e-06, "loss": 0.10701879262924194, "step": 26890 }, { "epoch": 0.7085870077016032, "grad_norm": 2.0819873809814453, "learning_rate": 4.13374891316713e-06, "loss": 0.11343830823898315, "step": 26900 }, { "epoch": 0.7088504229461019, "grad_norm": 1.350573182106018, "learning_rate": 4.126841843217144e-06, "loss": 0.11633208990097046, "step": 26910 }, { "epoch": 0.7091138381906007, "grad_norm": 3.0124897956848145, "learning_rate": 4.1199390480050605e-06, "loss": 0.12166750431060791, "step": 26920 }, { "epoch": 0.7093772534350994, "grad_norm": 1.8941220045089722, "learning_rate": 4.113040532555021e-06, "loss": 0.1140750527381897, "step": 26930 }, { "epoch": 0.7096406686795982, "grad_norm": 2.1463167667388916, "learning_rate": 4.106146301888065e-06, "loss": 0.13006289005279542, "step": 26940 }, { "epoch": 0.7099040839240969, "grad_norm": 1.4107600450515747, "learning_rate": 4.0992563610221155e-06, "loss": 0.11203041076660156, "step": 26950 }, { "epoch": 0.7101674991685957, "grad_norm": 2.2753612995147705, "learning_rate": 4.092370714971966e-06, "loss": 0.1166377067565918, "step": 26960 }, { "epoch": 0.7104309144130944, "grad_norm": 2.359844446182251, "learning_rate": 4.085489368749287e-06, "loss": 0.1280975103378296, "step": 26970 }, { "epoch": 0.7106943296575932, "grad_norm": 1.6562803983688354, "learning_rate": 4.07861232736262e-06, "loss": 0.12276105880737305, "step": 26980 }, { "epoch": 0.7109577449020918, "grad_norm": 3.2865426540374756, "learning_rate": 4.0717395958173665e-06, "loss": 0.11929903030395508, "step": 26990 }, { "epoch": 0.7112211601465905, "grad_norm": 3.377168893814087, "learning_rate": 4.064871179115807e-06, "loss": 0.12178035974502563, "step": 27000 }, { "epoch": 0.7114845753910893, "grad_norm": 1.999619960784912, "learning_rate": 4.0580070822570694e-06, "loss": 0.11535389423370361, "step": 27010 }, { "epoch": 0.711747990635588, "grad_norm": 1.3976091146469116, "learning_rate": 4.051147310237129e-06, "loss": 0.11862726211547851, "step": 27020 }, { "epoch": 0.7120114058800868, "grad_norm": 1.5103439092636108, "learning_rate": 4.044291868048837e-06, "loss": 0.11388988494873047, "step": 27030 }, { "epoch": 0.7122748211245855, "grad_norm": 2.203603506088257, "learning_rate": 4.0374407606818684e-06, "loss": 0.12180958986282349, "step": 27040 }, { "epoch": 0.7125382363690843, "grad_norm": 3.1265618801116943, "learning_rate": 4.03059399312277e-06, "loss": 0.13434205055236817, "step": 27050 }, { "epoch": 0.712801651613583, "grad_norm": 2.3688511848449707, "learning_rate": 4.0237515703549015e-06, "loss": 0.11743377447128296, "step": 27060 }, { "epoch": 0.7130650668580818, "grad_norm": 1.710732102394104, "learning_rate": 4.0169134973584726e-06, "loss": 0.10317066907882691, "step": 27070 }, { "epoch": 0.7133284821025805, "grad_norm": 1.4397358894348145, "learning_rate": 4.010079779110537e-06, "loss": 0.12313956022262573, "step": 27080 }, { "epoch": 0.7135918973470792, "grad_norm": 1.5474827289581299, "learning_rate": 4.003250420584966e-06, "loss": 0.13281220197677612, "step": 27090 }, { "epoch": 0.713855312591578, "grad_norm": 2.6661055088043213, "learning_rate": 3.996425426752459e-06, "loss": 0.13001519441604614, "step": 27100 }, { "epoch": 0.7141187278360767, "grad_norm": 2.1769914627075195, "learning_rate": 3.989604802580545e-06, "loss": 0.122047758102417, "step": 27110 }, { "epoch": 0.7143821430805755, "grad_norm": 1.4996957778930664, "learning_rate": 3.982788553033561e-06, "loss": 0.12069787979125976, "step": 27120 }, { "epoch": 0.7146455583250741, "grad_norm": 2.0206100940704346, "learning_rate": 3.975976683072676e-06, "loss": 0.12115275859832764, "step": 27130 }, { "epoch": 0.714908973569573, "grad_norm": 2.8667988777160645, "learning_rate": 3.9691691976558636e-06, "loss": 0.1272792100906372, "step": 27140 }, { "epoch": 0.7151723888140716, "grad_norm": 1.77174711227417, "learning_rate": 3.9623661017378946e-06, "loss": 0.12218490839004517, "step": 27150 }, { "epoch": 0.7154358040585704, "grad_norm": 2.206533193588257, "learning_rate": 3.955567400270365e-06, "loss": 0.12706682682037354, "step": 27160 }, { "epoch": 0.7156992193030691, "grad_norm": 2.275637149810791, "learning_rate": 3.948773098201657e-06, "loss": 0.11216809749603271, "step": 27170 }, { "epoch": 0.7159626345475678, "grad_norm": 2.8700687885284424, "learning_rate": 3.941983200476969e-06, "loss": 0.1241675615310669, "step": 27180 }, { "epoch": 0.7162260497920666, "grad_norm": 2.7569961547851562, "learning_rate": 3.935197712038267e-06, "loss": 0.12088764905929565, "step": 27190 }, { "epoch": 0.7164894650365653, "grad_norm": 1.1326334476470947, "learning_rate": 3.928416637824326e-06, "loss": 0.10992059707641602, "step": 27200 }, { "epoch": 0.7167528802810641, "grad_norm": 2.319352149963379, "learning_rate": 3.921639982770709e-06, "loss": 0.12357844114303589, "step": 27210 }, { "epoch": 0.7170162955255628, "grad_norm": 1.4442996978759766, "learning_rate": 3.914867751809756e-06, "loss": 0.12068275213241578, "step": 27220 }, { "epoch": 0.7172797107700616, "grad_norm": 2.5674023628234863, "learning_rate": 3.908099949870588e-06, "loss": 0.1201859474182129, "step": 27230 }, { "epoch": 0.7175431260145603, "grad_norm": 2.256633996963501, "learning_rate": 3.901336581879102e-06, "loss": 0.11195951700210571, "step": 27240 }, { "epoch": 0.717806541259059, "grad_norm": 2.381932020187378, "learning_rate": 3.894577652757968e-06, "loss": 0.13151636123657226, "step": 27250 }, { "epoch": 0.7180699565035578, "grad_norm": 1.3511911630630493, "learning_rate": 3.887823167426633e-06, "loss": 0.12355430126190185, "step": 27260 }, { "epoch": 0.7183333717480564, "grad_norm": 1.2164198160171509, "learning_rate": 3.8810731308013005e-06, "loss": 0.11391173601150513, "step": 27270 }, { "epoch": 0.7185967869925552, "grad_norm": 1.362744688987732, "learning_rate": 3.874327547794931e-06, "loss": 0.11488176584243774, "step": 27280 }, { "epoch": 0.7188602022370539, "grad_norm": 2.4343509674072266, "learning_rate": 3.8675864233172575e-06, "loss": 0.12159625291824341, "step": 27290 }, { "epoch": 0.7191236174815527, "grad_norm": 1.6597450971603394, "learning_rate": 3.860849762274759e-06, "loss": 0.10943484306335449, "step": 27300 }, { "epoch": 0.7193870327260514, "grad_norm": 2.217503547668457, "learning_rate": 3.854117569570674e-06, "loss": 0.11840810775756835, "step": 27310 }, { "epoch": 0.7196504479705502, "grad_norm": 1.7280086278915405, "learning_rate": 3.8473898501049745e-06, "loss": 0.1195081353187561, "step": 27320 }, { "epoch": 0.7199138632150489, "grad_norm": 1.4849473237991333, "learning_rate": 3.840666608774385e-06, "loss": 0.11385328769683838, "step": 27330 }, { "epoch": 0.7201772784595476, "grad_norm": 2.552304267883301, "learning_rate": 3.833947850472378e-06, "loss": 0.12050127983093262, "step": 27340 }, { "epoch": 0.7204406937040464, "grad_norm": 2.3716795444488525, "learning_rate": 3.827233580089152e-06, "loss": 0.13216915130615234, "step": 27350 }, { "epoch": 0.7207041089485451, "grad_norm": 2.6861393451690674, "learning_rate": 3.82052380251164e-06, "loss": 0.12387621402740479, "step": 27360 }, { "epoch": 0.7209675241930439, "grad_norm": 2.5226187705993652, "learning_rate": 3.8138185226235114e-06, "loss": 0.1122541069984436, "step": 27370 }, { "epoch": 0.7212309394375426, "grad_norm": 2.8787310123443604, "learning_rate": 3.8071177453051523e-06, "loss": 0.12965736389160157, "step": 27380 }, { "epoch": 0.7214943546820414, "grad_norm": 3.056363821029663, "learning_rate": 3.8004214754336853e-06, "loss": 0.11874854564666748, "step": 27390 }, { "epoch": 0.7217577699265401, "grad_norm": 1.568199634552002, "learning_rate": 3.7937297178829458e-06, "loss": 0.10609240531921386, "step": 27400 }, { "epoch": 0.7220211851710389, "grad_norm": 1.9935908317565918, "learning_rate": 3.7870424775234703e-06, "loss": 0.11081697940826415, "step": 27410 }, { "epoch": 0.7222846004155375, "grad_norm": 1.504525065422058, "learning_rate": 3.780359759222534e-06, "loss": 0.12490205764770508, "step": 27420 }, { "epoch": 0.7225480156600362, "grad_norm": 1.8643258810043335, "learning_rate": 3.773681567844103e-06, "loss": 0.11936498880386352, "step": 27430 }, { "epoch": 0.722811430904535, "grad_norm": 1.506584644317627, "learning_rate": 3.767007908248852e-06, "loss": 0.1135859489440918, "step": 27440 }, { "epoch": 0.7230748461490337, "grad_norm": 2.903513193130493, "learning_rate": 3.7603387852941608e-06, "loss": 0.10729714632034301, "step": 27450 }, { "epoch": 0.7233382613935325, "grad_norm": 1.9728156328201294, "learning_rate": 3.753674203834099e-06, "loss": 0.10310795307159423, "step": 27460 }, { "epoch": 0.7236016766380312, "grad_norm": 3.1666200160980225, "learning_rate": 3.747014168719445e-06, "loss": 0.11041045188903809, "step": 27470 }, { "epoch": 0.72386509188253, "grad_norm": 2.669316291809082, "learning_rate": 3.740358684797656e-06, "loss": 0.11760118007659912, "step": 27480 }, { "epoch": 0.7241285071270287, "grad_norm": 1.8571949005126953, "learning_rate": 3.7337077569128797e-06, "loss": 0.12018194198608398, "step": 27490 }, { "epoch": 0.7243919223715275, "grad_norm": 2.363863945007324, "learning_rate": 3.7270613899059495e-06, "loss": 0.10539257526397705, "step": 27500 }, { "epoch": 0.7246553376160262, "grad_norm": 1.878325343132019, "learning_rate": 3.7204195886143736e-06, "loss": 0.10915906429290771, "step": 27510 }, { "epoch": 0.7249187528605249, "grad_norm": 1.7607008218765259, "learning_rate": 3.713782357872352e-06, "loss": 0.1091078281402588, "step": 27520 }, { "epoch": 0.7251821681050237, "grad_norm": 2.3129920959472656, "learning_rate": 3.7071497025107373e-06, "loss": 0.11928732395172119, "step": 27530 }, { "epoch": 0.7254455833495224, "grad_norm": 2.2486212253570557, "learning_rate": 3.7005216273570623e-06, "loss": 0.11010732650756835, "step": 27540 }, { "epoch": 0.7257089985940212, "grad_norm": 3.0828161239624023, "learning_rate": 3.693898137235532e-06, "loss": 0.11611711978912354, "step": 27550 }, { "epoch": 0.7259724138385198, "grad_norm": 1.6542943716049194, "learning_rate": 3.6872792369670074e-06, "loss": 0.1131793737411499, "step": 27560 }, { "epoch": 0.7262358290830186, "grad_norm": 1.8356387615203857, "learning_rate": 3.6806649313690066e-06, "loss": 0.11925468444824219, "step": 27570 }, { "epoch": 0.7264992443275173, "grad_norm": 1.3766295909881592, "learning_rate": 3.6740552252557093e-06, "loss": 0.11659814119338989, "step": 27580 }, { "epoch": 0.7267626595720161, "grad_norm": 1.4008095264434814, "learning_rate": 3.667450123437939e-06, "loss": 0.11889472007751464, "step": 27590 }, { "epoch": 0.7270260748165148, "grad_norm": 1.8063406944274902, "learning_rate": 3.6608496307231823e-06, "loss": 0.11573033332824707, "step": 27600 }, { "epoch": 0.7272894900610135, "grad_norm": 2.152568817138672, "learning_rate": 3.6542537519155617e-06, "loss": 0.10332356691360474, "step": 27610 }, { "epoch": 0.7275529053055123, "grad_norm": 2.4253616333007812, "learning_rate": 3.647662491815833e-06, "loss": 0.11919510364532471, "step": 27620 }, { "epoch": 0.727816320550011, "grad_norm": 1.8263999223709106, "learning_rate": 3.641075855221409e-06, "loss": 0.12627885341644288, "step": 27630 }, { "epoch": 0.7280797357945098, "grad_norm": 1.3348997831344604, "learning_rate": 3.634493846926321e-06, "loss": 0.1245943546295166, "step": 27640 }, { "epoch": 0.7283431510390085, "grad_norm": 2.0186729431152344, "learning_rate": 3.62791647172125e-06, "loss": 0.12326951026916504, "step": 27650 }, { "epoch": 0.7286065662835073, "grad_norm": 3.228224515914917, "learning_rate": 3.621343734393482e-06, "loss": 0.12548767328262328, "step": 27660 }, { "epoch": 0.728869981528006, "grad_norm": 1.8546370267868042, "learning_rate": 3.6147756397269397e-06, "loss": 0.11851780414581299, "step": 27670 }, { "epoch": 0.7291333967725048, "grad_norm": 2.0766730308532715, "learning_rate": 3.6082121925021717e-06, "loss": 0.11335618495941162, "step": 27680 }, { "epoch": 0.7293968120170035, "grad_norm": 1.9104712009429932, "learning_rate": 3.601653397496335e-06, "loss": 0.10981366634368897, "step": 27690 }, { "epoch": 0.7296602272615021, "grad_norm": 2.113523006439209, "learning_rate": 3.5950992594832044e-06, "loss": 0.11782114505767823, "step": 27700 }, { "epoch": 0.729923642506001, "grad_norm": 1.8722351789474487, "learning_rate": 3.588549783233164e-06, "loss": 0.10086860656738281, "step": 27710 }, { "epoch": 0.7301870577504996, "grad_norm": 2.3482377529144287, "learning_rate": 3.582004973513201e-06, "loss": 0.11865497827529907, "step": 27720 }, { "epoch": 0.7304504729949984, "grad_norm": 3.016274929046631, "learning_rate": 3.575464835086917e-06, "loss": 0.12270463705062866, "step": 27730 }, { "epoch": 0.7307138882394971, "grad_norm": 2.8257830142974854, "learning_rate": 3.568929372714508e-06, "loss": 0.1328423023223877, "step": 27740 }, { "epoch": 0.7309773034839959, "grad_norm": 3.176060676574707, "learning_rate": 3.562398591152755e-06, "loss": 0.11597945690155029, "step": 27750 }, { "epoch": 0.7312407187284946, "grad_norm": 2.2613794803619385, "learning_rate": 3.5558724951550515e-06, "loss": 0.11276696920394898, "step": 27760 }, { "epoch": 0.7315041339729933, "grad_norm": 1.869027018547058, "learning_rate": 3.549351089471367e-06, "loss": 0.10939429998397827, "step": 27770 }, { "epoch": 0.7317675492174921, "grad_norm": 2.859475612640381, "learning_rate": 3.54283437884827e-06, "loss": 0.11807434558868408, "step": 27780 }, { "epoch": 0.7320309644619908, "grad_norm": 2.2375760078430176, "learning_rate": 3.536322368028895e-06, "loss": 0.11523926258087158, "step": 27790 }, { "epoch": 0.7322943797064896, "grad_norm": 2.206886053085327, "learning_rate": 3.5298150617529635e-06, "loss": 0.11804792881011963, "step": 27800 }, { "epoch": 0.7325577949509883, "grad_norm": 2.251879930496216, "learning_rate": 3.5233124647567817e-06, "loss": 0.13296449184417725, "step": 27810 }, { "epoch": 0.7328212101954871, "grad_norm": 2.2789032459259033, "learning_rate": 3.516814581773216e-06, "loss": 0.11435232162475586, "step": 27820 }, { "epoch": 0.7330846254399858, "grad_norm": 2.386967182159424, "learning_rate": 3.5103214175317056e-06, "loss": 0.11401557922363281, "step": 27830 }, { "epoch": 0.7333480406844846, "grad_norm": 2.238673210144043, "learning_rate": 3.5038329767582567e-06, "loss": 0.11456202268600464, "step": 27840 }, { "epoch": 0.7336114559289832, "grad_norm": 2.0560734272003174, "learning_rate": 3.4973492641754315e-06, "loss": 0.11025478839874267, "step": 27850 }, { "epoch": 0.7338748711734819, "grad_norm": 1.7557488679885864, "learning_rate": 3.4908702845023643e-06, "loss": 0.11059154272079467, "step": 27860 }, { "epoch": 0.7341382864179807, "grad_norm": 2.1201231479644775, "learning_rate": 3.484396042454734e-06, "loss": 0.12221239805221558, "step": 27870 }, { "epoch": 0.7344017016624794, "grad_norm": 2.076723575592041, "learning_rate": 3.477926542744766e-06, "loss": 0.11762017011642456, "step": 27880 }, { "epoch": 0.7346651169069782, "grad_norm": 3.1764354705810547, "learning_rate": 3.4714617900812485e-06, "loss": 0.12177612781524658, "step": 27890 }, { "epoch": 0.7349285321514769, "grad_norm": 3.1152279376983643, "learning_rate": 3.465001789169502e-06, "loss": 0.11665505170822144, "step": 27900 }, { "epoch": 0.7351919473959757, "grad_norm": 2.2053651809692383, "learning_rate": 3.458546544711404e-06, "loss": 0.11907396316528321, "step": 27910 }, { "epoch": 0.7354553626404744, "grad_norm": 2.6329429149627686, "learning_rate": 3.45209606140535e-06, "loss": 0.11509928703308106, "step": 27920 }, { "epoch": 0.7357187778849732, "grad_norm": 2.066983222961426, "learning_rate": 3.4456503439462797e-06, "loss": 0.11111185550689698, "step": 27930 }, { "epoch": 0.7359821931294719, "grad_norm": 1.7517598867416382, "learning_rate": 3.4392093970256703e-06, "loss": 0.13237440586090088, "step": 27940 }, { "epoch": 0.7362456083739706, "grad_norm": 1.7166496515274048, "learning_rate": 3.4327732253315195e-06, "loss": 0.12192974090576172, "step": 27950 }, { "epoch": 0.7365090236184694, "grad_norm": 2.022427558898926, "learning_rate": 3.4263418335483477e-06, "loss": 0.1187664270401001, "step": 27960 }, { "epoch": 0.7367724388629681, "grad_norm": 1.9366487264633179, "learning_rate": 3.419915226357203e-06, "loss": 0.10542230606079102, "step": 27970 }, { "epoch": 0.7370358541074669, "grad_norm": 2.17488431930542, "learning_rate": 3.413493408435642e-06, "loss": 0.11749818325042724, "step": 27980 }, { "epoch": 0.7372992693519655, "grad_norm": 2.5876777172088623, "learning_rate": 3.407076384457747e-06, "loss": 0.11597220897674561, "step": 27990 }, { "epoch": 0.7375626845964643, "grad_norm": 1.7562578916549683, "learning_rate": 3.400664159094106e-06, "loss": 0.09718084335327148, "step": 28000 }, { "epoch": 0.737826099840963, "grad_norm": 1.3945207595825195, "learning_rate": 3.394256737011803e-06, "loss": 0.10905537605285645, "step": 28010 }, { "epoch": 0.7380895150854618, "grad_norm": 1.9697245359420776, "learning_rate": 3.3878541228744455e-06, "loss": 0.10522754192352295, "step": 28020 }, { "epoch": 0.7383529303299605, "grad_norm": 1.5413663387298584, "learning_rate": 3.38145632134213e-06, "loss": 0.10748559236526489, "step": 28030 }, { "epoch": 0.7386163455744592, "grad_norm": 1.8743482828140259, "learning_rate": 3.3750633370714527e-06, "loss": 0.11116278171539307, "step": 28040 }, { "epoch": 0.738879760818958, "grad_norm": 1.7392059564590454, "learning_rate": 3.368675174715501e-06, "loss": 0.11926674842834473, "step": 28050 }, { "epoch": 0.7391431760634567, "grad_norm": 3.287235975265503, "learning_rate": 3.3622918389238544e-06, "loss": 0.11945244073867797, "step": 28060 }, { "epoch": 0.7394065913079555, "grad_norm": 1.898306131362915, "learning_rate": 3.3559133343425865e-06, "loss": 0.11678509712219239, "step": 28070 }, { "epoch": 0.7396700065524542, "grad_norm": 2.4186251163482666, "learning_rate": 3.3495396656142442e-06, "loss": 0.10424538850784301, "step": 28080 }, { "epoch": 0.739933421796953, "grad_norm": 2.227473020553589, "learning_rate": 3.343170837377858e-06, "loss": 0.11742508411407471, "step": 28090 }, { "epoch": 0.7401968370414517, "grad_norm": 2.6863045692443848, "learning_rate": 3.336806854268936e-06, "loss": 0.10530037879943847, "step": 28100 }, { "epoch": 0.7404602522859505, "grad_norm": 2.0438859462738037, "learning_rate": 3.3304477209194574e-06, "loss": 0.12672765254974366, "step": 28110 }, { "epoch": 0.7407236675304492, "grad_norm": 1.8428547382354736, "learning_rate": 3.3240934419578828e-06, "loss": 0.11815779209136963, "step": 28120 }, { "epoch": 0.7409870827749478, "grad_norm": 1.3075140714645386, "learning_rate": 3.3177440220091207e-06, "loss": 0.11608549356460571, "step": 28130 }, { "epoch": 0.7412504980194466, "grad_norm": 1.558368444442749, "learning_rate": 3.3113994656945537e-06, "loss": 0.1170583963394165, "step": 28140 }, { "epoch": 0.7415139132639453, "grad_norm": 3.034384250640869, "learning_rate": 3.3050597776320293e-06, "loss": 0.10185598134994507, "step": 28150 }, { "epoch": 0.7417773285084441, "grad_norm": 3.2401156425476074, "learning_rate": 3.298724962435842e-06, "loss": 0.10779002904891968, "step": 28160 }, { "epoch": 0.7420407437529428, "grad_norm": 3.5300724506378174, "learning_rate": 3.292395024716746e-06, "loss": 0.1255251169204712, "step": 28170 }, { "epoch": 0.7423041589974416, "grad_norm": 1.9361032247543335, "learning_rate": 3.2860699690819407e-06, "loss": 0.11120243072509765, "step": 28180 }, { "epoch": 0.7425675742419403, "grad_norm": 3.200296401977539, "learning_rate": 3.2797498001350737e-06, "loss": 0.10667557716369629, "step": 28190 }, { "epoch": 0.7428309894864391, "grad_norm": 2.4836487770080566, "learning_rate": 3.27343452247624e-06, "loss": 0.10796170234680176, "step": 28200 }, { "epoch": 0.7430944047309378, "grad_norm": 2.188218355178833, "learning_rate": 3.267124140701974e-06, "loss": 0.1239927053451538, "step": 28210 }, { "epoch": 0.7433578199754365, "grad_norm": 1.8974895477294922, "learning_rate": 3.2608186594052314e-06, "loss": 0.10830600261688232, "step": 28220 }, { "epoch": 0.7436212352199353, "grad_norm": 2.121480703353882, "learning_rate": 3.2545180831754264e-06, "loss": 0.12367551326751709, "step": 28230 }, { "epoch": 0.743884650464434, "grad_norm": 2.108844041824341, "learning_rate": 3.248222416598381e-06, "loss": 0.10254689455032348, "step": 28240 }, { "epoch": 0.7441480657089328, "grad_norm": 1.7494664192199707, "learning_rate": 3.2419316642563646e-06, "loss": 0.105080246925354, "step": 28250 }, { "epoch": 0.7444114809534315, "grad_norm": 1.9637352228164673, "learning_rate": 3.235645830728047e-06, "loss": 0.11864218711853028, "step": 28260 }, { "epoch": 0.7446748961979303, "grad_norm": 1.4382879734039307, "learning_rate": 3.2293649205885293e-06, "loss": 0.10492277145385742, "step": 28270 }, { "epoch": 0.744938311442429, "grad_norm": 1.470000982284546, "learning_rate": 3.2230889384093355e-06, "loss": 0.09536184668540955, "step": 28280 }, { "epoch": 0.7452017266869276, "grad_norm": 2.7277021408081055, "learning_rate": 3.2168178887583947e-06, "loss": 0.11595797538757324, "step": 28290 }, { "epoch": 0.7454651419314264, "grad_norm": 4.285806655883789, "learning_rate": 3.210551776200044e-06, "loss": 0.10496206283569336, "step": 28300 }, { "epoch": 0.7457285571759251, "grad_norm": 2.16125226020813, "learning_rate": 3.204290605295033e-06, "loss": 0.11472570896148682, "step": 28310 }, { "epoch": 0.7459919724204239, "grad_norm": 2.6700003147125244, "learning_rate": 3.1980343806005077e-06, "loss": 0.11249222755432128, "step": 28320 }, { "epoch": 0.7462553876649226, "grad_norm": 2.10435152053833, "learning_rate": 3.1917831066700255e-06, "loss": 0.11721837520599365, "step": 28330 }, { "epoch": 0.7465188029094214, "grad_norm": 2.033230781555176, "learning_rate": 3.1855367880535336e-06, "loss": 0.12278184890747071, "step": 28340 }, { "epoch": 0.7467822181539201, "grad_norm": 2.2207746505737305, "learning_rate": 3.179295429297361e-06, "loss": 0.1235379695892334, "step": 28350 }, { "epoch": 0.7470456333984189, "grad_norm": 2.037698984146118, "learning_rate": 3.173059034944248e-06, "loss": 0.11027657985687256, "step": 28360 }, { "epoch": 0.7473090486429176, "grad_norm": 1.7984880208969116, "learning_rate": 3.1668276095333063e-06, "loss": 0.1110999584197998, "step": 28370 }, { "epoch": 0.7475724638874163, "grad_norm": 1.5147708654403687, "learning_rate": 3.160601157600045e-06, "loss": 0.11355626583099365, "step": 28380 }, { "epoch": 0.7478358791319151, "grad_norm": 2.35507869720459, "learning_rate": 3.154379683676334e-06, "loss": 0.10166778564453124, "step": 28390 }, { "epoch": 0.7480992943764138, "grad_norm": 2.322338104248047, "learning_rate": 3.148163192290432e-06, "loss": 0.11124299764633179, "step": 28400 }, { "epoch": 0.7483627096209126, "grad_norm": 2.8296945095062256, "learning_rate": 3.1419516879669744e-06, "loss": 0.10554044246673584, "step": 28410 }, { "epoch": 0.7486261248654112, "grad_norm": 1.8247795104980469, "learning_rate": 3.1357451752269595e-06, "loss": 0.11891016960144044, "step": 28420 }, { "epoch": 0.74888954010991, "grad_norm": 1.998046636581421, "learning_rate": 3.1295436585877558e-06, "loss": 0.127939772605896, "step": 28430 }, { "epoch": 0.7491529553544087, "grad_norm": 5.937948226928711, "learning_rate": 3.1233471425630947e-06, "loss": 0.11036579608917237, "step": 28440 }, { "epoch": 0.7494163705989075, "grad_norm": 2.7310123443603516, "learning_rate": 3.1171556316630626e-06, "loss": 0.11051354408264161, "step": 28450 }, { "epoch": 0.7496797858434062, "grad_norm": 2.4129650592803955, "learning_rate": 3.110969130394117e-06, "loss": 0.09427891969680786, "step": 28460 }, { "epoch": 0.7499432010879049, "grad_norm": 1.76536226272583, "learning_rate": 3.1047876432590596e-06, "loss": 0.12534472942352295, "step": 28470 }, { "epoch": 0.7502066163324037, "grad_norm": 2.0553677082061768, "learning_rate": 3.0986111747570335e-06, "loss": 0.11535053253173828, "step": 28480 }, { "epoch": 0.7504700315769024, "grad_norm": 1.9919973611831665, "learning_rate": 3.092439729383548e-06, "loss": 0.12119061946868896, "step": 28490 }, { "epoch": 0.7507334468214012, "grad_norm": 2.0095741748809814, "learning_rate": 3.0862733116304435e-06, "loss": 0.10667043924331665, "step": 28500 }, { "epoch": 0.7509968620658999, "grad_norm": 1.905532717704773, "learning_rate": 3.0801119259859115e-06, "loss": 0.12211775779724121, "step": 28510 }, { "epoch": 0.7512602773103987, "grad_norm": 2.9823901653289795, "learning_rate": 3.0739555769344666e-06, "loss": 0.09633402228355407, "step": 28520 }, { "epoch": 0.7515236925548974, "grad_norm": 2.2431678771972656, "learning_rate": 3.067804268956964e-06, "loss": 0.1209635615348816, "step": 28530 }, { "epoch": 0.7517871077993962, "grad_norm": 2.3395068645477295, "learning_rate": 3.0616580065305977e-06, "loss": 0.10804212093353271, "step": 28540 }, { "epoch": 0.7520505230438949, "grad_norm": 1.5747613906860352, "learning_rate": 3.055516794128881e-06, "loss": 0.11929118633270264, "step": 28550 }, { "epoch": 0.7523139382883935, "grad_norm": 3.19789457321167, "learning_rate": 3.0493806362216516e-06, "loss": 0.11153368949890137, "step": 28560 }, { "epoch": 0.7525773535328923, "grad_norm": 2.4614808559417725, "learning_rate": 3.0432495372750692e-06, "loss": 0.10772961378097534, "step": 28570 }, { "epoch": 0.752840768777391, "grad_norm": 2.4080495834350586, "learning_rate": 3.037123501751611e-06, "loss": 0.11543608903884887, "step": 28580 }, { "epoch": 0.7531041840218898, "grad_norm": 3.637078285217285, "learning_rate": 3.0310025341100757e-06, "loss": 0.11580777168273926, "step": 28590 }, { "epoch": 0.7533675992663885, "grad_norm": 2.2048704624176025, "learning_rate": 3.0248866388055674e-06, "loss": 0.1210524320602417, "step": 28600 }, { "epoch": 0.7536310145108873, "grad_norm": 1.8417234420776367, "learning_rate": 3.018775820289488e-06, "loss": 0.0960583209991455, "step": 28610 }, { "epoch": 0.753894429755386, "grad_norm": 2.4272708892822266, "learning_rate": 3.0126700830095668e-06, "loss": 0.10932953357696533, "step": 28620 }, { "epoch": 0.7541578449998848, "grad_norm": 2.939303159713745, "learning_rate": 3.006569431409817e-06, "loss": 0.10756092071533203, "step": 28630 }, { "epoch": 0.7544212602443835, "grad_norm": 2.9699339866638184, "learning_rate": 3.000473869930558e-06, "loss": 0.10357744693756103, "step": 28640 }, { "epoch": 0.7546846754888822, "grad_norm": 2.119982957839966, "learning_rate": 2.9943834030084006e-06, "loss": 0.12711561918258668, "step": 28650 }, { "epoch": 0.754948090733381, "grad_norm": 1.8950426578521729, "learning_rate": 2.988298035076249e-06, "loss": 0.10415318012237548, "step": 28660 }, { "epoch": 0.7552115059778797, "grad_norm": 2.410963296890259, "learning_rate": 2.9822177705633005e-06, "loss": 0.10509796142578125, "step": 28670 }, { "epoch": 0.7554749212223785, "grad_norm": 2.272277593612671, "learning_rate": 2.976142613895031e-06, "loss": 0.10469913482666016, "step": 28680 }, { "epoch": 0.7557383364668772, "grad_norm": 2.1021344661712646, "learning_rate": 2.970072569493202e-06, "loss": 0.09530800580978394, "step": 28690 }, { "epoch": 0.756001751711376, "grad_norm": 2.9050676822662354, "learning_rate": 2.9640076417758546e-06, "loss": 0.12170753479003907, "step": 28700 }, { "epoch": 0.7562651669558746, "grad_norm": 2.5585105419158936, "learning_rate": 2.9579478351573e-06, "loss": 0.1069373607635498, "step": 28710 }, { "epoch": 0.7565285822003734, "grad_norm": 2.863450527191162, "learning_rate": 2.9518931540481377e-06, "loss": 0.11424611806869507, "step": 28720 }, { "epoch": 0.7567919974448721, "grad_norm": 1.8509392738342285, "learning_rate": 2.945843602855215e-06, "loss": 0.11711499691009522, "step": 28730 }, { "epoch": 0.7570554126893708, "grad_norm": 1.9220157861709595, "learning_rate": 2.9397991859816588e-06, "loss": 0.11851334571838379, "step": 28740 }, { "epoch": 0.7573188279338696, "grad_norm": 3.2745108604431152, "learning_rate": 2.9337599078268597e-06, "loss": 0.11104247570037842, "step": 28750 }, { "epoch": 0.7575822431783683, "grad_norm": 1.8345122337341309, "learning_rate": 2.927725772786464e-06, "loss": 0.11821900606155396, "step": 28760 }, { "epoch": 0.7578456584228671, "grad_norm": 1.4489829540252686, "learning_rate": 2.9216967852523747e-06, "loss": 0.10336267948150635, "step": 28770 }, { "epoch": 0.7581090736673658, "grad_norm": 2.9268996715545654, "learning_rate": 2.915672949612749e-06, "loss": 0.10543432235717773, "step": 28780 }, { "epoch": 0.7583724889118646, "grad_norm": 1.7115247249603271, "learning_rate": 2.9096542702519904e-06, "loss": 0.10714178085327149, "step": 28790 }, { "epoch": 0.7586359041563633, "grad_norm": 2.1690173149108887, "learning_rate": 2.9036407515507613e-06, "loss": 0.10738196372985839, "step": 28800 }, { "epoch": 0.758899319400862, "grad_norm": 1.6772655248641968, "learning_rate": 2.89763239788596e-06, "loss": 0.10994408130645753, "step": 28810 }, { "epoch": 0.7591627346453608, "grad_norm": 3.0155115127563477, "learning_rate": 2.8916292136307147e-06, "loss": 0.11954339742660522, "step": 28820 }, { "epoch": 0.7594261498898595, "grad_norm": 1.6000062227249146, "learning_rate": 2.885631203154412e-06, "loss": 0.12171907424926758, "step": 28830 }, { "epoch": 0.7596895651343583, "grad_norm": 2.359022855758667, "learning_rate": 2.879638370822656e-06, "loss": 0.11847488880157471, "step": 28840 }, { "epoch": 0.759952980378857, "grad_norm": 1.8246443271636963, "learning_rate": 2.8736507209972996e-06, "loss": 0.10971739292144775, "step": 28850 }, { "epoch": 0.7602163956233557, "grad_norm": 2.2961809635162354, "learning_rate": 2.867668258036401e-06, "loss": 0.11407297849655151, "step": 28860 }, { "epoch": 0.7604798108678544, "grad_norm": 2.1467247009277344, "learning_rate": 2.8616909862942565e-06, "loss": 0.11372796297073365, "step": 28870 }, { "epoch": 0.7607432261123532, "grad_norm": 1.641113042831421, "learning_rate": 2.855718910121389e-06, "loss": 0.10242385864257812, "step": 28880 }, { "epoch": 0.7610066413568519, "grad_norm": 2.1368539333343506, "learning_rate": 2.8497520338645278e-06, "loss": 0.09708724021911622, "step": 28890 }, { "epoch": 0.7612700566013506, "grad_norm": 2.021902561187744, "learning_rate": 2.843790361866625e-06, "loss": 0.10462512969970703, "step": 28900 }, { "epoch": 0.7615334718458494, "grad_norm": 1.572171926498413, "learning_rate": 2.837833898466842e-06, "loss": 0.10361733436584472, "step": 28910 }, { "epoch": 0.7617968870903481, "grad_norm": 1.8740203380584717, "learning_rate": 2.831882648000547e-06, "loss": 0.10623611211776733, "step": 28920 }, { "epoch": 0.7620603023348469, "grad_norm": 2.014639139175415, "learning_rate": 2.8259366147993217e-06, "loss": 0.10677577257156372, "step": 28930 }, { "epoch": 0.7623237175793456, "grad_norm": 2.261223793029785, "learning_rate": 2.819995803190948e-06, "loss": 0.11526358127593994, "step": 28940 }, { "epoch": 0.7625871328238444, "grad_norm": 3.1811957359313965, "learning_rate": 2.8140602174993927e-06, "loss": 0.1092390775680542, "step": 28950 }, { "epoch": 0.7628505480683431, "grad_norm": 2.5311501026153564, "learning_rate": 2.8081298620448427e-06, "loss": 0.11192741394042968, "step": 28960 }, { "epoch": 0.7631139633128419, "grad_norm": 1.6075079441070557, "learning_rate": 2.8022047411436604e-06, "loss": 0.09511278867721558, "step": 28970 }, { "epoch": 0.7633773785573406, "grad_norm": 2.9141297340393066, "learning_rate": 2.7962848591084122e-06, "loss": 0.1126785397529602, "step": 28980 }, { "epoch": 0.7636407938018392, "grad_norm": 2.7581024169921875, "learning_rate": 2.7903702202478355e-06, "loss": 0.10833187103271484, "step": 28990 }, { "epoch": 0.763904209046338, "grad_norm": 4.034667491912842, "learning_rate": 2.7844608288668586e-06, "loss": 0.09964309930801392, "step": 29000 }, { "epoch": 0.7641676242908367, "grad_norm": 2.3027641773223877, "learning_rate": 2.7785566892665993e-06, "loss": 0.12119672298431397, "step": 29010 }, { "epoch": 0.7644310395353355, "grad_norm": 2.277366876602173, "learning_rate": 2.77265780574434e-06, "loss": 0.10786187648773193, "step": 29020 }, { "epoch": 0.7646944547798342, "grad_norm": 1.9306695461273193, "learning_rate": 2.766764182593544e-06, "loss": 0.10970476865768433, "step": 29030 }, { "epoch": 0.764957870024333, "grad_norm": 2.0467212200164795, "learning_rate": 2.760875824103846e-06, "loss": 0.11122227907180786, "step": 29040 }, { "epoch": 0.7652212852688317, "grad_norm": 2.2333085536956787, "learning_rate": 2.7549927345610407e-06, "loss": 0.10652447938919067, "step": 29050 }, { "epoch": 0.7654847005133305, "grad_norm": 2.3755807876586914, "learning_rate": 2.749114918247102e-06, "loss": 0.10802233219146729, "step": 29060 }, { "epoch": 0.7657481157578292, "grad_norm": 2.1852335929870605, "learning_rate": 2.7432423794401598e-06, "loss": 0.1013180136680603, "step": 29070 }, { "epoch": 0.7660115310023279, "grad_norm": 2.5609419345855713, "learning_rate": 2.7373751224144905e-06, "loss": 0.10801432132720948, "step": 29080 }, { "epoch": 0.7662749462468267, "grad_norm": 2.10229229927063, "learning_rate": 2.731513151440547e-06, "loss": 0.1104661226272583, "step": 29090 }, { "epoch": 0.7665383614913254, "grad_norm": 1.24983549118042, "learning_rate": 2.7256564707849165e-06, "loss": 0.10590221881866455, "step": 29100 }, { "epoch": 0.7668017767358242, "grad_norm": 2.7401950359344482, "learning_rate": 2.719805084710356e-06, "loss": 0.12367777824401856, "step": 29110 }, { "epoch": 0.7670651919803229, "grad_norm": 2.3819541931152344, "learning_rate": 2.7139589974757454e-06, "loss": 0.1005718469619751, "step": 29120 }, { "epoch": 0.7673286072248217, "grad_norm": 2.9638209342956543, "learning_rate": 2.7081182133361207e-06, "loss": 0.11054753065109253, "step": 29130 }, { "epoch": 0.7675920224693203, "grad_norm": 1.6928718090057373, "learning_rate": 2.7022827365426606e-06, "loss": 0.11563515663146973, "step": 29140 }, { "epoch": 0.7678554377138191, "grad_norm": 1.8320391178131104, "learning_rate": 2.696452571342676e-06, "loss": 0.11093462705612182, "step": 29150 }, { "epoch": 0.7681188529583178, "grad_norm": 2.044863700866699, "learning_rate": 2.6906277219796116e-06, "loss": 0.09268667101860047, "step": 29160 }, { "epoch": 0.7683822682028165, "grad_norm": 2.657179355621338, "learning_rate": 2.6848081926930447e-06, "loss": 0.09113399982452393, "step": 29170 }, { "epoch": 0.7686456834473153, "grad_norm": 2.247852325439453, "learning_rate": 2.6789939877186753e-06, "loss": 0.11599489450454711, "step": 29180 }, { "epoch": 0.768909098691814, "grad_norm": 1.8965418338775635, "learning_rate": 2.67318511128834e-06, "loss": 0.10232670307159424, "step": 29190 }, { "epoch": 0.7691725139363128, "grad_norm": 3.024156332015991, "learning_rate": 2.66738156762999e-06, "loss": 0.09571147561073304, "step": 29200 }, { "epoch": 0.7694359291808115, "grad_norm": 3.0941543579101562, "learning_rate": 2.6615833609676846e-06, "loss": 0.106933331489563, "step": 29210 }, { "epoch": 0.7696993444253103, "grad_norm": 2.020129919052124, "learning_rate": 2.6557904955216185e-06, "loss": 0.12130868434906006, "step": 29220 }, { "epoch": 0.769962759669809, "grad_norm": 1.9398326873779297, "learning_rate": 2.650002975508086e-06, "loss": 0.09663882255554199, "step": 29230 }, { "epoch": 0.7702261749143078, "grad_norm": 2.1656861305236816, "learning_rate": 2.6442208051394936e-06, "loss": 0.09977526664733886, "step": 29240 }, { "epoch": 0.7704895901588065, "grad_norm": 2.1519765853881836, "learning_rate": 2.638443988624355e-06, "loss": 0.10152437686920165, "step": 29250 }, { "epoch": 0.7707530054033052, "grad_norm": 2.5024948120117188, "learning_rate": 2.632672530167283e-06, "loss": 0.09868078231811524, "step": 29260 }, { "epoch": 0.771016420647804, "grad_norm": 2.0807976722717285, "learning_rate": 2.6269064339689986e-06, "loss": 0.1019057273864746, "step": 29270 }, { "epoch": 0.7712798358923026, "grad_norm": 2.0656261444091797, "learning_rate": 2.6211457042263157e-06, "loss": 0.10401287078857421, "step": 29280 }, { "epoch": 0.7715432511368014, "grad_norm": 1.8568929433822632, "learning_rate": 2.6153903451321384e-06, "loss": 0.09852112531661987, "step": 29290 }, { "epoch": 0.7718066663813001, "grad_norm": 2.068373203277588, "learning_rate": 2.6096403608754685e-06, "loss": 0.09420151710510254, "step": 29300 }, { "epoch": 0.7720700816257989, "grad_norm": 2.4069042205810547, "learning_rate": 2.603895755641387e-06, "loss": 0.1254136323928833, "step": 29310 }, { "epoch": 0.7723334968702976, "grad_norm": 1.4999173879623413, "learning_rate": 2.598156533611077e-06, "loss": 0.10195057392120362, "step": 29320 }, { "epoch": 0.7725969121147963, "grad_norm": 3.5391945838928223, "learning_rate": 2.592422698961781e-06, "loss": 0.10377495288848877, "step": 29330 }, { "epoch": 0.7728603273592951, "grad_norm": 2.3333640098571777, "learning_rate": 2.586694255866831e-06, "loss": 0.11542732715606689, "step": 29340 }, { "epoch": 0.7731237426037938, "grad_norm": 2.814791679382324, "learning_rate": 2.580971208495642e-06, "loss": 0.09955225586891174, "step": 29350 }, { "epoch": 0.7733871578482926, "grad_norm": 2.9963035583496094, "learning_rate": 2.575253561013692e-06, "loss": 0.11123573780059814, "step": 29360 }, { "epoch": 0.7736505730927913, "grad_norm": 2.2336044311523438, "learning_rate": 2.56954131758253e-06, "loss": 0.09252492189407349, "step": 29370 }, { "epoch": 0.7739139883372901, "grad_norm": 2.796008586883545, "learning_rate": 2.5638344823597762e-06, "loss": 0.11316423416137696, "step": 29380 }, { "epoch": 0.7741774035817888, "grad_norm": 2.3672842979431152, "learning_rate": 2.558133059499105e-06, "loss": 0.10953431129455567, "step": 29390 }, { "epoch": 0.7744408188262876, "grad_norm": 3.602856159210205, "learning_rate": 2.552437053150266e-06, "loss": 0.10522642135620117, "step": 29400 }, { "epoch": 0.7747042340707863, "grad_norm": 1.977079153060913, "learning_rate": 2.546746467459058e-06, "loss": 0.10458725690841675, "step": 29410 }, { "epoch": 0.774967649315285, "grad_norm": 3.4106850624084473, "learning_rate": 2.541061306567325e-06, "loss": 0.11442704200744629, "step": 29420 }, { "epoch": 0.7752310645597837, "grad_norm": 2.635674476623535, "learning_rate": 2.5353815746129827e-06, "loss": 0.10444284677505493, "step": 29430 }, { "epoch": 0.7754944798042824, "grad_norm": 2.5873570442199707, "learning_rate": 2.529707275729979e-06, "loss": 0.10766317844390869, "step": 29440 }, { "epoch": 0.7757578950487812, "grad_norm": 2.766479730606079, "learning_rate": 2.524038414048321e-06, "loss": 0.1129724621772766, "step": 29450 }, { "epoch": 0.7760213102932799, "grad_norm": 2.692570447921753, "learning_rate": 2.518374993694043e-06, "loss": 0.10506348609924317, "step": 29460 }, { "epoch": 0.7762847255377787, "grad_norm": 2.855803966522217, "learning_rate": 2.5127170187892257e-06, "loss": 0.1038158655166626, "step": 29470 }, { "epoch": 0.7765481407822774, "grad_norm": 3.050288200378418, "learning_rate": 2.507064493451995e-06, "loss": 0.09646067023277283, "step": 29480 }, { "epoch": 0.7768115560267762, "grad_norm": 1.7398812770843506, "learning_rate": 2.501417421796498e-06, "loss": 0.10976638793945312, "step": 29490 }, { "epoch": 0.7770749712712749, "grad_norm": 2.5950212478637695, "learning_rate": 2.4957758079329173e-06, "loss": 0.10820250511169434, "step": 29500 }, { "epoch": 0.7773383865157736, "grad_norm": 1.663110613822937, "learning_rate": 2.490139655967465e-06, "loss": 0.11371781826019287, "step": 29510 }, { "epoch": 0.7776018017602724, "grad_norm": 1.6788477897644043, "learning_rate": 2.4845089700023694e-06, "loss": 0.11050618886947632, "step": 29520 }, { "epoch": 0.7778652170047711, "grad_norm": 1.6401376724243164, "learning_rate": 2.4788837541358957e-06, "loss": 0.0990557074546814, "step": 29530 }, { "epoch": 0.7781286322492699, "grad_norm": 2.7304439544677734, "learning_rate": 2.4732640124623152e-06, "loss": 0.10990705490112304, "step": 29540 }, { "epoch": 0.7783920474937686, "grad_norm": 2.117602586746216, "learning_rate": 2.467649749071913e-06, "loss": 0.10667607784271241, "step": 29550 }, { "epoch": 0.7786554627382674, "grad_norm": 2.053899049758911, "learning_rate": 2.4620409680509993e-06, "loss": 0.10857541561126709, "step": 29560 }, { "epoch": 0.778918877982766, "grad_norm": 1.9639462232589722, "learning_rate": 2.4564376734818818e-06, "loss": 0.11312531232833863, "step": 29570 }, { "epoch": 0.7791822932272648, "grad_norm": 1.9451096057891846, "learning_rate": 2.450839869442887e-06, "loss": 0.10286128520965576, "step": 29580 }, { "epoch": 0.7794457084717635, "grad_norm": 2.405618667602539, "learning_rate": 2.4452475600083323e-06, "loss": 0.09600199460983276, "step": 29590 }, { "epoch": 0.7797091237162622, "grad_norm": 1.9897747039794922, "learning_rate": 2.4396607492485393e-06, "loss": 0.10202583074569702, "step": 29600 }, { "epoch": 0.779972538960761, "grad_norm": 2.358945608139038, "learning_rate": 2.4340794412298364e-06, "loss": 0.11529362201690674, "step": 29610 }, { "epoch": 0.7802359542052597, "grad_norm": 1.8943089246749878, "learning_rate": 2.428503640014539e-06, "loss": 0.09958317279815673, "step": 29620 }, { "epoch": 0.7804993694497585, "grad_norm": 1.6416140794754028, "learning_rate": 2.4229333496609532e-06, "loss": 0.10747349262237549, "step": 29630 }, { "epoch": 0.7807627846942572, "grad_norm": 2.480982542037964, "learning_rate": 2.4173685742233787e-06, "loss": 0.10971136093139648, "step": 29640 }, { "epoch": 0.781026199938756, "grad_norm": 2.303393840789795, "learning_rate": 2.411809317752095e-06, "loss": 0.09402797222137452, "step": 29650 }, { "epoch": 0.7812896151832547, "grad_norm": 2.1966054439544678, "learning_rate": 2.4062555842933753e-06, "loss": 0.09764624834060669, "step": 29660 }, { "epoch": 0.7815530304277535, "grad_norm": 1.9560797214508057, "learning_rate": 2.4007073778894665e-06, "loss": 0.09743269681930541, "step": 29670 }, { "epoch": 0.7818164456722522, "grad_norm": 2.4328038692474365, "learning_rate": 2.395164702578583e-06, "loss": 0.10395692586898804, "step": 29680 }, { "epoch": 0.7820798609167509, "grad_norm": 3.7305431365966797, "learning_rate": 2.3896275623949327e-06, "loss": 0.11257811784744262, "step": 29690 }, { "epoch": 0.7823432761612497, "grad_norm": 2.311058282852173, "learning_rate": 2.38409596136868e-06, "loss": 0.10153965950012207, "step": 29700 }, { "epoch": 0.7826066914057483, "grad_norm": 2.9370200634002686, "learning_rate": 2.378569903525969e-06, "loss": 0.10973742008209228, "step": 29710 }, { "epoch": 0.7828701066502471, "grad_norm": 2.531160354614258, "learning_rate": 2.3730493928888963e-06, "loss": 0.11163433790206909, "step": 29720 }, { "epoch": 0.7831335218947458, "grad_norm": 2.928375244140625, "learning_rate": 2.367534433475527e-06, "loss": 0.10655643939971923, "step": 29730 }, { "epoch": 0.7833969371392446, "grad_norm": 2.3430402278900146, "learning_rate": 2.3620250292998925e-06, "loss": 0.11001017093658447, "step": 29740 }, { "epoch": 0.7836603523837433, "grad_norm": 1.7908682823181152, "learning_rate": 2.356521184371973e-06, "loss": 0.10372365713119507, "step": 29750 }, { "epoch": 0.7839237676282421, "grad_norm": 2.112163543701172, "learning_rate": 2.3510229026977018e-06, "loss": 0.1100579857826233, "step": 29760 }, { "epoch": 0.7841871828727408, "grad_norm": 3.1351816654205322, "learning_rate": 2.345530188278967e-06, "loss": 0.09762536883354186, "step": 29770 }, { "epoch": 0.7844505981172395, "grad_norm": 1.4174375534057617, "learning_rate": 2.3400430451136e-06, "loss": 0.10326206684112549, "step": 29780 }, { "epoch": 0.7847140133617383, "grad_norm": 3.5724594593048096, "learning_rate": 2.334561477195387e-06, "loss": 0.08905689120292663, "step": 29790 }, { "epoch": 0.784977428606237, "grad_norm": 2.188936948776245, "learning_rate": 2.329085488514049e-06, "loss": 0.10084326267242431, "step": 29800 }, { "epoch": 0.7852408438507358, "grad_norm": 2.450495481491089, "learning_rate": 2.3236150830552374e-06, "loss": 0.09803738594055175, "step": 29810 }, { "epoch": 0.7855042590952345, "grad_norm": 2.919785261154175, "learning_rate": 2.3181502648005593e-06, "loss": 0.10097002983093262, "step": 29820 }, { "epoch": 0.7857676743397333, "grad_norm": 2.511720657348633, "learning_rate": 2.3126910377275415e-06, "loss": 0.10023025274276734, "step": 29830 }, { "epoch": 0.786031089584232, "grad_norm": 3.3814475536346436, "learning_rate": 2.3072374058096447e-06, "loss": 0.12299985885620117, "step": 29840 }, { "epoch": 0.7862945048287308, "grad_norm": 2.124335527420044, "learning_rate": 2.301789373016259e-06, "loss": 0.09933788776397705, "step": 29850 }, { "epoch": 0.7865579200732294, "grad_norm": 3.488847494125366, "learning_rate": 2.296346943312693e-06, "loss": 0.10544464588165284, "step": 29860 }, { "epoch": 0.7868213353177281, "grad_norm": 2.4754655361175537, "learning_rate": 2.290910120660188e-06, "loss": 0.10901446342468261, "step": 29870 }, { "epoch": 0.7870847505622269, "grad_norm": 2.1373841762542725, "learning_rate": 2.285478909015896e-06, "loss": 0.10075972080230713, "step": 29880 }, { "epoch": 0.7873481658067256, "grad_norm": 2.5028953552246094, "learning_rate": 2.2800533123328874e-06, "loss": 0.11232705116271972, "step": 29890 }, { "epoch": 0.7876115810512244, "grad_norm": 2.5802552700042725, "learning_rate": 2.2746333345601435e-06, "loss": 0.09860701560974121, "step": 29900 }, { "epoch": 0.7878749962957231, "grad_norm": 2.0076375007629395, "learning_rate": 2.269218979642558e-06, "loss": 0.09961537122726441, "step": 29910 }, { "epoch": 0.7881384115402219, "grad_norm": 2.5779826641082764, "learning_rate": 2.2638102515209382e-06, "loss": 0.10269268751144409, "step": 29920 }, { "epoch": 0.7884018267847206, "grad_norm": 3.2313971519470215, "learning_rate": 2.258407154131982e-06, "loss": 0.09377456307411194, "step": 29930 }, { "epoch": 0.7886652420292193, "grad_norm": 2.355905771255493, "learning_rate": 2.2530096914082965e-06, "loss": 0.11937890052795411, "step": 29940 }, { "epoch": 0.7889286572737181, "grad_norm": 1.9105643033981323, "learning_rate": 2.2476178672783943e-06, "loss": 0.09724319577217103, "step": 29950 }, { "epoch": 0.7891920725182168, "grad_norm": 2.523822784423828, "learning_rate": 2.242231685666675e-06, "loss": 0.10384583473205566, "step": 29960 }, { "epoch": 0.7894554877627156, "grad_norm": 3.3274412155151367, "learning_rate": 2.2368511504934333e-06, "loss": 0.10601608753204346, "step": 29970 }, { "epoch": 0.7897189030072143, "grad_norm": 1.8603743314743042, "learning_rate": 2.2314762656748544e-06, "loss": 0.0987348735332489, "step": 29980 }, { "epoch": 0.7899823182517131, "grad_norm": 1.734856128692627, "learning_rate": 2.2261070351230107e-06, "loss": 0.10241851806640626, "step": 29990 }, { "epoch": 0.7902457334962117, "grad_norm": 2.490375518798828, "learning_rate": 2.2207434627458626e-06, "loss": 0.11146395206451416, "step": 30000 }, { "epoch": 0.7905091487407105, "grad_norm": 2.4172778129577637, "learning_rate": 2.2153855524472513e-06, "loss": 0.11379624605178833, "step": 30010 }, { "epoch": 0.7907725639852092, "grad_norm": 1.6662647724151611, "learning_rate": 2.2100333081268853e-06, "loss": 0.1030120611190796, "step": 30020 }, { "epoch": 0.7910359792297079, "grad_norm": 3.6563820838928223, "learning_rate": 2.204686733680369e-06, "loss": 0.10208138227462768, "step": 30030 }, { "epoch": 0.7912993944742067, "grad_norm": 2.518568992614746, "learning_rate": 2.1993458329991635e-06, "loss": 0.09794659614562988, "step": 30040 }, { "epoch": 0.7915628097187054, "grad_norm": 1.8513543605804443, "learning_rate": 2.1940106099706137e-06, "loss": 0.10632678270339965, "step": 30050 }, { "epoch": 0.7918262249632042, "grad_norm": 1.5457658767700195, "learning_rate": 2.1886810684779204e-06, "loss": 0.10127389430999756, "step": 30060 }, { "epoch": 0.7920896402077029, "grad_norm": 2.2213306427001953, "learning_rate": 2.18335721240015e-06, "loss": 0.10120739936828613, "step": 30070 }, { "epoch": 0.7923530554522017, "grad_norm": 2.544038772583008, "learning_rate": 2.1780390456122424e-06, "loss": 0.09335299730300903, "step": 30080 }, { "epoch": 0.7926164706967004, "grad_norm": 2.375037431716919, "learning_rate": 2.1727265719849853e-06, "loss": 0.10432586669921876, "step": 30090 }, { "epoch": 0.7928798859411992, "grad_norm": 1.921794056892395, "learning_rate": 2.1674197953850274e-06, "loss": 0.0962272047996521, "step": 30100 }, { "epoch": 0.7931433011856979, "grad_norm": 2.5376110076904297, "learning_rate": 2.162118719674867e-06, "loss": 0.11189782619476318, "step": 30110 }, { "epoch": 0.7934067164301966, "grad_norm": 2.6436848640441895, "learning_rate": 2.1568233487128556e-06, "loss": 0.09308662414550781, "step": 30120 }, { "epoch": 0.7936701316746954, "grad_norm": 3.1172335147857666, "learning_rate": 2.1515336863531976e-06, "loss": 0.11103767156600952, "step": 30130 }, { "epoch": 0.793933546919194, "grad_norm": 3.1956169605255127, "learning_rate": 2.1462497364459356e-06, "loss": 0.10227203369140625, "step": 30140 }, { "epoch": 0.7941969621636928, "grad_norm": 2.4752631187438965, "learning_rate": 2.1409715028369504e-06, "loss": 0.09799787998199463, "step": 30150 }, { "epoch": 0.7944603774081915, "grad_norm": 1.9479243755340576, "learning_rate": 2.135698989367976e-06, "loss": 0.10510900020599365, "step": 30160 }, { "epoch": 0.7947237926526903, "grad_norm": 2.585387945175171, "learning_rate": 2.1304321998765685e-06, "loss": 0.09306852221488952, "step": 30170 }, { "epoch": 0.794987207897189, "grad_norm": 1.7084771394729614, "learning_rate": 2.1251711381961336e-06, "loss": 0.10570292472839356, "step": 30180 }, { "epoch": 0.7952506231416878, "grad_norm": 3.8540170192718506, "learning_rate": 2.11991580815589e-06, "loss": 0.09919472932815551, "step": 30190 }, { "epoch": 0.7955140383861865, "grad_norm": 2.1741890907287598, "learning_rate": 2.1146662135808948e-06, "loss": 0.10253365039825439, "step": 30200 }, { "epoch": 0.7957774536306852, "grad_norm": 2.675208330154419, "learning_rate": 2.1094223582920346e-06, "loss": 0.10344504117965699, "step": 30210 }, { "epoch": 0.796040868875184, "grad_norm": 3.122464418411255, "learning_rate": 2.1041842461060103e-06, "loss": 0.1039923906326294, "step": 30220 }, { "epoch": 0.7963042841196827, "grad_norm": 2.193250894546509, "learning_rate": 2.098951880835346e-06, "loss": 0.10336925983428955, "step": 30230 }, { "epoch": 0.7965676993641815, "grad_norm": 2.154554605484009, "learning_rate": 2.093725266288382e-06, "loss": 0.09681297540664673, "step": 30240 }, { "epoch": 0.7968311146086802, "grad_norm": 2.792383909225464, "learning_rate": 2.088504406269274e-06, "loss": 0.10914685726165771, "step": 30250 }, { "epoch": 0.797094529853179, "grad_norm": 3.1740972995758057, "learning_rate": 2.083289304577992e-06, "loss": 0.10009956359863281, "step": 30260 }, { "epoch": 0.7973579450976777, "grad_norm": 2.7721705436706543, "learning_rate": 2.0780799650103123e-06, "loss": 0.09519697427749634, "step": 30270 }, { "epoch": 0.7976213603421765, "grad_norm": 3.988821268081665, "learning_rate": 2.0728763913578097e-06, "loss": 0.106632399559021, "step": 30280 }, { "epoch": 0.7978847755866751, "grad_norm": 2.6161763668060303, "learning_rate": 2.0676785874078773e-06, "loss": 0.09430739879608155, "step": 30290 }, { "epoch": 0.7981481908311738, "grad_norm": 2.162923574447632, "learning_rate": 2.062486556943699e-06, "loss": 0.10770424604415893, "step": 30300 }, { "epoch": 0.7984116060756726, "grad_norm": 2.1281509399414062, "learning_rate": 2.0573003037442575e-06, "loss": 0.10835130214691162, "step": 30310 }, { "epoch": 0.7986750213201713, "grad_norm": 2.7398102283477783, "learning_rate": 2.0521198315843315e-06, "loss": 0.09537909626960754, "step": 30320 }, { "epoch": 0.7989384365646701, "grad_norm": 2.307100772857666, "learning_rate": 2.046945144234491e-06, "loss": 0.1071700930595398, "step": 30330 }, { "epoch": 0.7992018518091688, "grad_norm": 2.337705612182617, "learning_rate": 2.0417762454611012e-06, "loss": 0.126465368270874, "step": 30340 }, { "epoch": 0.7994652670536676, "grad_norm": 2.534097671508789, "learning_rate": 2.0366131390263055e-06, "loss": 0.10068594217300415, "step": 30350 }, { "epoch": 0.7997286822981663, "grad_norm": 2.212820529937744, "learning_rate": 2.0314558286880383e-06, "loss": 0.10018962621688843, "step": 30360 }, { "epoch": 0.7999920975426651, "grad_norm": 2.9768569469451904, "learning_rate": 2.026304318200011e-06, "loss": 0.11340570449829102, "step": 30370 }, { "epoch": 0.8002555127871638, "grad_norm": 2.7264480590820312, "learning_rate": 2.021158611311712e-06, "loss": 0.1038978099822998, "step": 30380 }, { "epoch": 0.8005189280316625, "grad_norm": 2.1975619792938232, "learning_rate": 2.016018711768416e-06, "loss": 0.1095590591430664, "step": 30390 }, { "epoch": 0.8007823432761613, "grad_norm": 2.89817214012146, "learning_rate": 2.010884623311162e-06, "loss": 0.10926902294158936, "step": 30400 }, { "epoch": 0.80104575852066, "grad_norm": 2.945384979248047, "learning_rate": 2.005756349676753e-06, "loss": 0.10542826652526856, "step": 30410 }, { "epoch": 0.8013091737651588, "grad_norm": 1.9850443601608276, "learning_rate": 2.0006338945977777e-06, "loss": 0.0963804304599762, "step": 30420 }, { "epoch": 0.8015725890096574, "grad_norm": 2.956050157546997, "learning_rate": 1.9955172618025743e-06, "loss": 0.10212506055831909, "step": 30430 }, { "epoch": 0.8018360042541562, "grad_norm": 2.5569851398468018, "learning_rate": 1.9904064550152515e-06, "loss": 0.09042570590972901, "step": 30440 }, { "epoch": 0.8020994194986549, "grad_norm": 2.1711983680725098, "learning_rate": 1.9853014779556734e-06, "loss": 0.0970804214477539, "step": 30450 }, { "epoch": 0.8023628347431536, "grad_norm": 2.3872787952423096, "learning_rate": 1.9802023343394606e-06, "loss": 0.09862137436866761, "step": 30460 }, { "epoch": 0.8026262499876524, "grad_norm": 2.6491286754608154, "learning_rate": 1.975109027877995e-06, "loss": 0.10594279766082763, "step": 30470 }, { "epoch": 0.8028896652321511, "grad_norm": 2.3902764320373535, "learning_rate": 1.9700215622784026e-06, "loss": 0.0824286937713623, "step": 30480 }, { "epoch": 0.8031530804766499, "grad_norm": 3.9095709323883057, "learning_rate": 1.96493994124356e-06, "loss": 0.10033643245697021, "step": 30490 }, { "epoch": 0.8034164957211486, "grad_norm": 2.0353457927703857, "learning_rate": 1.9598641684720897e-06, "loss": 0.09039275050163269, "step": 30500 }, { "epoch": 0.8036799109656474, "grad_norm": 2.0480873584747314, "learning_rate": 1.954794247658356e-06, "loss": 0.11212444305419922, "step": 30510 }, { "epoch": 0.8039433262101461, "grad_norm": 2.6248531341552734, "learning_rate": 1.949730182492473e-06, "loss": 0.0965346097946167, "step": 30520 }, { "epoch": 0.8042067414546449, "grad_norm": 2.586522340774536, "learning_rate": 1.9446719766602793e-06, "loss": 0.10060727596282959, "step": 30530 }, { "epoch": 0.8044701566991436, "grad_norm": 3.5560975074768066, "learning_rate": 1.9396196338433547e-06, "loss": 0.10060412883758545, "step": 30540 }, { "epoch": 0.8047335719436423, "grad_norm": 2.4846441745758057, "learning_rate": 1.9345731577190187e-06, "loss": 0.10485575199127198, "step": 30550 }, { "epoch": 0.804996987188141, "grad_norm": 2.2964694499969482, "learning_rate": 1.929532551960309e-06, "loss": 0.10503683090209961, "step": 30560 }, { "epoch": 0.8052604024326397, "grad_norm": 2.239311695098877, "learning_rate": 1.9244978202359974e-06, "loss": 0.08414986729621887, "step": 30570 }, { "epoch": 0.8055238176771385, "grad_norm": 1.9318010807037354, "learning_rate": 1.9194689662105803e-06, "loss": 0.09182456731796265, "step": 30580 }, { "epoch": 0.8057872329216372, "grad_norm": 2.6315736770629883, "learning_rate": 1.9144459935442705e-06, "loss": 0.10163826942443847, "step": 30590 }, { "epoch": 0.806050648166136, "grad_norm": 2.1264753341674805, "learning_rate": 1.9094289058930095e-06, "loss": 0.09478009939193725, "step": 30600 }, { "epoch": 0.8063140634106347, "grad_norm": 2.385493755340576, "learning_rate": 1.9044177069084502e-06, "loss": 0.09102095365524292, "step": 30610 }, { "epoch": 0.8065774786551335, "grad_norm": 2.5342350006103516, "learning_rate": 1.899412400237951e-06, "loss": 0.10170502662658691, "step": 30620 }, { "epoch": 0.8068408938996322, "grad_norm": 2.298704147338867, "learning_rate": 1.8944129895246e-06, "loss": 0.10850121974945068, "step": 30630 }, { "epoch": 0.8071043091441309, "grad_norm": 2.5242741107940674, "learning_rate": 1.8894194784071761e-06, "loss": 0.09807728528976441, "step": 30640 }, { "epoch": 0.8073677243886297, "grad_norm": 3.150552272796631, "learning_rate": 1.8844318705201815e-06, "loss": 0.10528196096420288, "step": 30650 }, { "epoch": 0.8076311396331284, "grad_norm": 2.3674068450927734, "learning_rate": 1.879450169493805e-06, "loss": 0.09862388372421264, "step": 30660 }, { "epoch": 0.8078945548776272, "grad_norm": 2.309098243713379, "learning_rate": 1.874474378953942e-06, "loss": 0.10587178468704224, "step": 30670 }, { "epoch": 0.8081579701221259, "grad_norm": 4.493068218231201, "learning_rate": 1.8695045025221936e-06, "loss": 0.0997033417224884, "step": 30680 }, { "epoch": 0.8084213853666247, "grad_norm": 2.1331241130828857, "learning_rate": 1.864540543815848e-06, "loss": 0.08914066553115844, "step": 30690 }, { "epoch": 0.8086848006111234, "grad_norm": 3.4952340126037598, "learning_rate": 1.859582506447889e-06, "loss": 0.09995833039283752, "step": 30700 }, { "epoch": 0.8089482158556222, "grad_norm": 2.2555301189422607, "learning_rate": 1.8546303940269883e-06, "loss": 0.09096700549125672, "step": 30710 }, { "epoch": 0.8092116311001208, "grad_norm": 3.0101637840270996, "learning_rate": 1.8496842101575053e-06, "loss": 0.10132811069488526, "step": 30720 }, { "epoch": 0.8094750463446195, "grad_norm": 3.173882007598877, "learning_rate": 1.844743958439491e-06, "loss": 0.10961766242980957, "step": 30730 }, { "epoch": 0.8097384615891183, "grad_norm": 3.2490084171295166, "learning_rate": 1.839809642468674e-06, "loss": 0.09935485124588013, "step": 30740 }, { "epoch": 0.810001876833617, "grad_norm": 2.0775156021118164, "learning_rate": 1.8348812658364557e-06, "loss": 0.10972459316253662, "step": 30750 }, { "epoch": 0.8102652920781158, "grad_norm": 2.3018429279327393, "learning_rate": 1.8299588321299266e-06, "loss": 0.09412057399749756, "step": 30760 }, { "epoch": 0.8105287073226145, "grad_norm": 2.4381978511810303, "learning_rate": 1.8250423449318434e-06, "loss": 0.10183088779449463, "step": 30770 }, { "epoch": 0.8107921225671133, "grad_norm": 2.639451503753662, "learning_rate": 1.8201318078206443e-06, "loss": 0.11048798561096192, "step": 30780 }, { "epoch": 0.811055537811612, "grad_norm": 1.9445831775665283, "learning_rate": 1.8152272243704216e-06, "loss": 0.11534185409545898, "step": 30790 }, { "epoch": 0.8113189530561108, "grad_norm": 2.2157483100891113, "learning_rate": 1.8103285981509432e-06, "loss": 0.10321683883666992, "step": 30800 }, { "epoch": 0.8115823683006095, "grad_norm": 1.9329686164855957, "learning_rate": 1.8054359327276473e-06, "loss": 0.09956198930740356, "step": 30810 }, { "epoch": 0.8118457835451082, "grad_norm": 2.024501085281372, "learning_rate": 1.8005492316616224e-06, "loss": 0.10059921741485596, "step": 30820 }, { "epoch": 0.812109198789607, "grad_norm": 1.9053144454956055, "learning_rate": 1.7956684985096217e-06, "loss": 0.088255375623703, "step": 30830 }, { "epoch": 0.8123726140341057, "grad_norm": 2.2927353382110596, "learning_rate": 1.7907937368240525e-06, "loss": 0.09460504651069641, "step": 30840 }, { "epoch": 0.8126360292786045, "grad_norm": 3.6126105785369873, "learning_rate": 1.7859249501529742e-06, "loss": 0.10557039976119995, "step": 30850 }, { "epoch": 0.8128994445231031, "grad_norm": 2.972585439682007, "learning_rate": 1.7810621420401054e-06, "loss": 0.10062664747238159, "step": 30860 }, { "epoch": 0.8131628597676019, "grad_norm": 2.228623151779175, "learning_rate": 1.776205316024806e-06, "loss": 0.09596153497695922, "step": 30870 }, { "epoch": 0.8134262750121006, "grad_norm": 1.8532334566116333, "learning_rate": 1.7713544756420785e-06, "loss": 0.09649009704589843, "step": 30880 }, { "epoch": 0.8136896902565994, "grad_norm": 2.177025079727173, "learning_rate": 1.76650962442258e-06, "loss": 0.09855520129203796, "step": 30890 }, { "epoch": 0.8139531055010981, "grad_norm": 4.6057305335998535, "learning_rate": 1.761670765892599e-06, "loss": 0.10131417512893677, "step": 30900 }, { "epoch": 0.8142165207455968, "grad_norm": 2.3926656246185303, "learning_rate": 1.7568379035740669e-06, "loss": 0.10379652976989746, "step": 30910 }, { "epoch": 0.8144799359900956, "grad_norm": 2.4922659397125244, "learning_rate": 1.7520110409845493e-06, "loss": 0.09709064960479737, "step": 30920 }, { "epoch": 0.8147433512345943, "grad_norm": 3.4670639038085938, "learning_rate": 1.7471901816372416e-06, "loss": 0.10063086748123169, "step": 30930 }, { "epoch": 0.8150067664790931, "grad_norm": 2.3955297470092773, "learning_rate": 1.7423753290409784e-06, "loss": 0.10178632736206054, "step": 30940 }, { "epoch": 0.8152701817235918, "grad_norm": 3.31606125831604, "learning_rate": 1.7375664867002151e-06, "loss": 0.09885970354080201, "step": 30950 }, { "epoch": 0.8155335969680906, "grad_norm": 1.6375770568847656, "learning_rate": 1.7327636581150343e-06, "loss": 0.09941215515136718, "step": 30960 }, { "epoch": 0.8157970122125893, "grad_norm": 2.7898635864257812, "learning_rate": 1.7279668467811427e-06, "loss": 0.09457095265388489, "step": 30970 }, { "epoch": 0.816060427457088, "grad_norm": 2.7095017433166504, "learning_rate": 1.7231760561898625e-06, "loss": 0.09405511617660522, "step": 30980 }, { "epoch": 0.8163238427015868, "grad_norm": 2.308763027191162, "learning_rate": 1.718391289828144e-06, "loss": 0.08920605182647705, "step": 30990 }, { "epoch": 0.8165872579460854, "grad_norm": 2.4709558486938477, "learning_rate": 1.7136125511785462e-06, "loss": 0.09693531394004821, "step": 31000 }, { "epoch": 0.8168506731905842, "grad_norm": 2.153144359588623, "learning_rate": 1.7088398437192344e-06, "loss": 0.10188546180725097, "step": 31010 }, { "epoch": 0.8171140884350829, "grad_norm": 2.1626625061035156, "learning_rate": 1.7040731709239967e-06, "loss": 0.097905433177948, "step": 31020 }, { "epoch": 0.8173775036795817, "grad_norm": 3.041135311126709, "learning_rate": 1.6993125362622231e-06, "loss": 0.095755934715271, "step": 31030 }, { "epoch": 0.8176409189240804, "grad_norm": 2.6440601348876953, "learning_rate": 1.6945579431989067e-06, "loss": 0.09246640205383301, "step": 31040 }, { "epoch": 0.8179043341685792, "grad_norm": 2.25240159034729, "learning_rate": 1.689809395194647e-06, "loss": 0.08927754163742066, "step": 31050 }, { "epoch": 0.8181677494130779, "grad_norm": 3.601128101348877, "learning_rate": 1.6850668957056393e-06, "loss": 0.11250033378601074, "step": 31060 }, { "epoch": 0.8184311646575766, "grad_norm": 2.391826868057251, "learning_rate": 1.6803304481836835e-06, "loss": 0.09215079545974732, "step": 31070 }, { "epoch": 0.8186945799020754, "grad_norm": 3.7391746044158936, "learning_rate": 1.6756000560761687e-06, "loss": 0.0970126748085022, "step": 31080 }, { "epoch": 0.8189579951465741, "grad_norm": 2.5933451652526855, "learning_rate": 1.6708757228260786e-06, "loss": 0.10198168754577637, "step": 31090 }, { "epoch": 0.8192214103910729, "grad_norm": 1.9861392974853516, "learning_rate": 1.6661574518719859e-06, "loss": 0.0907946228981018, "step": 31100 }, { "epoch": 0.8194848256355716, "grad_norm": 2.311957836151123, "learning_rate": 1.6614452466480502e-06, "loss": 0.09679244756698609, "step": 31110 }, { "epoch": 0.8197482408800704, "grad_norm": 2.3531041145324707, "learning_rate": 1.6567391105840247e-06, "loss": 0.09049997925758362, "step": 31120 }, { "epoch": 0.820011656124569, "grad_norm": 2.875424861907959, "learning_rate": 1.6520390471052306e-06, "loss": 0.09348170757293701, "step": 31130 }, { "epoch": 0.8202750713690679, "grad_norm": 2.0123066902160645, "learning_rate": 1.647345059632578e-06, "loss": 0.08808712959289551, "step": 31140 }, { "epoch": 0.8205384866135665, "grad_norm": 2.6033809185028076, "learning_rate": 1.6426571515825563e-06, "loss": 0.09451888799667359, "step": 31150 }, { "epoch": 0.8208019018580652, "grad_norm": 2.636636257171631, "learning_rate": 1.6379753263672272e-06, "loss": 0.09870589971542358, "step": 31160 }, { "epoch": 0.821065317102564, "grad_norm": 2.5029401779174805, "learning_rate": 1.6332995873942237e-06, "loss": 0.0860779881477356, "step": 31170 }, { "epoch": 0.8213287323470627, "grad_norm": 2.949045181274414, "learning_rate": 1.6286299380667504e-06, "loss": 0.09192924499511719, "step": 31180 }, { "epoch": 0.8215921475915615, "grad_norm": 2.542492628097534, "learning_rate": 1.6239663817835783e-06, "loss": 0.10354038476943969, "step": 31190 }, { "epoch": 0.8218555628360602, "grad_norm": 2.444443464279175, "learning_rate": 1.6193089219390478e-06, "loss": 0.10250027179718017, "step": 31200 }, { "epoch": 0.822118978080559, "grad_norm": 2.5327839851379395, "learning_rate": 1.6146575619230597e-06, "loss": 0.10577535629272461, "step": 31210 }, { "epoch": 0.8223823933250577, "grad_norm": 2.210672616958618, "learning_rate": 1.610012305121068e-06, "loss": 0.09472029209136963, "step": 31220 }, { "epoch": 0.8226458085695565, "grad_norm": 2.613236427307129, "learning_rate": 1.6053731549140982e-06, "loss": 0.10975174903869629, "step": 31230 }, { "epoch": 0.8229092238140552, "grad_norm": 2.583828926086426, "learning_rate": 1.6007401146787195e-06, "loss": 0.10098938941955567, "step": 31240 }, { "epoch": 0.8231726390585539, "grad_norm": 2.3501384258270264, "learning_rate": 1.596113187787064e-06, "loss": 0.0891477882862091, "step": 31250 }, { "epoch": 0.8234360543030527, "grad_norm": 2.785311698913574, "learning_rate": 1.5914923776068025e-06, "loss": 0.09783970713615417, "step": 31260 }, { "epoch": 0.8236994695475514, "grad_norm": 2.357874631881714, "learning_rate": 1.5868776875011604e-06, "loss": 0.09212687015533447, "step": 31270 }, { "epoch": 0.8239628847920502, "grad_norm": 3.5716354846954346, "learning_rate": 1.582269120828912e-06, "loss": 0.10760661363601684, "step": 31280 }, { "epoch": 0.8242263000365488, "grad_norm": 1.9684737920761108, "learning_rate": 1.5776666809443697e-06, "loss": 0.08421471118927001, "step": 31290 }, { "epoch": 0.8244897152810476, "grad_norm": 2.6339125633239746, "learning_rate": 1.5730703711973883e-06, "loss": 0.08660909533500671, "step": 31300 }, { "epoch": 0.8247531305255463, "grad_norm": 2.033888101577759, "learning_rate": 1.5684801949333584e-06, "loss": 0.09185928702354432, "step": 31310 }, { "epoch": 0.8250165457700451, "grad_norm": 1.9448575973510742, "learning_rate": 1.5638961554932064e-06, "loss": 0.10358326435089112, "step": 31320 }, { "epoch": 0.8252799610145438, "grad_norm": 2.5716307163238525, "learning_rate": 1.5593182562134002e-06, "loss": 0.10626088380813599, "step": 31330 }, { "epoch": 0.8255433762590425, "grad_norm": 2.5840651988983154, "learning_rate": 1.554746500425931e-06, "loss": 0.09409339427947998, "step": 31340 }, { "epoch": 0.8258067915035413, "grad_norm": 1.9957154989242554, "learning_rate": 1.5501808914583139e-06, "loss": 0.09343056678771973, "step": 31350 }, { "epoch": 0.82607020674804, "grad_norm": 2.361879825592041, "learning_rate": 1.5456214326336017e-06, "loss": 0.09168766140937805, "step": 31360 }, { "epoch": 0.8263336219925388, "grad_norm": 2.0754916667938232, "learning_rate": 1.5410681272703643e-06, "loss": 0.09610361456871033, "step": 31370 }, { "epoch": 0.8265970372370375, "grad_norm": 2.4348602294921875, "learning_rate": 1.536520978682693e-06, "loss": 0.08634010553359986, "step": 31380 }, { "epoch": 0.8268604524815363, "grad_norm": 2.3042643070220947, "learning_rate": 1.531979990180199e-06, "loss": 0.08448027968406677, "step": 31390 }, { "epoch": 0.827123867726035, "grad_norm": 2.5539491176605225, "learning_rate": 1.5274451650680066e-06, "loss": 0.10882412195205689, "step": 31400 }, { "epoch": 0.8273872829705338, "grad_norm": 2.44153094291687, "learning_rate": 1.5229165066467632e-06, "loss": 0.09531021714210511, "step": 31410 }, { "epoch": 0.8276506982150325, "grad_norm": 1.9842381477355957, "learning_rate": 1.5183940182126189e-06, "loss": 0.08564068078994751, "step": 31420 }, { "epoch": 0.8279141134595311, "grad_norm": 2.349813222885132, "learning_rate": 1.513877703057235e-06, "loss": 0.09401295185089112, "step": 31430 }, { "epoch": 0.8281775287040299, "grad_norm": 2.174201488494873, "learning_rate": 1.509367564467782e-06, "loss": 0.09458405375480652, "step": 31440 }, { "epoch": 0.8284409439485286, "grad_norm": 2.7290008068084717, "learning_rate": 1.5048636057269294e-06, "loss": 0.10662031173706055, "step": 31450 }, { "epoch": 0.8287043591930274, "grad_norm": 2.265287399291992, "learning_rate": 1.50036583011286e-06, "loss": 0.09395207166671753, "step": 31460 }, { "epoch": 0.8289677744375261, "grad_norm": 2.2634289264678955, "learning_rate": 1.4958742408992454e-06, "loss": 0.09494591951370239, "step": 31470 }, { "epoch": 0.8292311896820249, "grad_norm": 1.7269690036773682, "learning_rate": 1.4913888413552536e-06, "loss": 0.08747838139533996, "step": 31480 }, { "epoch": 0.8294946049265236, "grad_norm": 4.4071455001831055, "learning_rate": 1.486909634745557e-06, "loss": 0.09001590013504028, "step": 31490 }, { "epoch": 0.8297580201710223, "grad_norm": 2.466275930404663, "learning_rate": 1.4824366243303156e-06, "loss": 0.0982825756072998, "step": 31500 }, { "epoch": 0.8300214354155211, "grad_norm": 2.1502861976623535, "learning_rate": 1.4779698133651766e-06, "loss": 0.09334533214569092, "step": 31510 }, { "epoch": 0.8302848506600198, "grad_norm": 2.7752034664154053, "learning_rate": 1.47350920510128e-06, "loss": 0.09430872201919556, "step": 31520 }, { "epoch": 0.8305482659045186, "grad_norm": 2.7200710773468018, "learning_rate": 1.469054802785246e-06, "loss": 0.1011959195137024, "step": 31530 }, { "epoch": 0.8308116811490173, "grad_norm": 3.9094719886779785, "learning_rate": 1.4646066096591838e-06, "loss": 0.10393025875091552, "step": 31540 }, { "epoch": 0.8310750963935161, "grad_norm": 3.068401575088501, "learning_rate": 1.4601646289606809e-06, "loss": 0.10906866788864136, "step": 31550 }, { "epoch": 0.8313385116380148, "grad_norm": 2.3668620586395264, "learning_rate": 1.4557288639227995e-06, "loss": 0.08477731347084046, "step": 31560 }, { "epoch": 0.8316019268825136, "grad_norm": 2.03501558303833, "learning_rate": 1.451299317774083e-06, "loss": 0.08680410385131836, "step": 31570 }, { "epoch": 0.8318653421270122, "grad_norm": 1.814716100692749, "learning_rate": 1.4468759937385423e-06, "loss": 0.08237886428833008, "step": 31580 }, { "epoch": 0.8321287573715109, "grad_norm": 2.90429425239563, "learning_rate": 1.4424588950356688e-06, "loss": 0.09658218622207641, "step": 31590 }, { "epoch": 0.8323921726160097, "grad_norm": 2.066568374633789, "learning_rate": 1.4380480248804185e-06, "loss": 0.11154341697692871, "step": 31600 }, { "epoch": 0.8326555878605084, "grad_norm": 3.2611019611358643, "learning_rate": 1.4336433864832033e-06, "loss": 0.09978151917457581, "step": 31610 }, { "epoch": 0.8329190031050072, "grad_norm": 2.549973487854004, "learning_rate": 1.4292449830499189e-06, "loss": 0.09879053831100464, "step": 31620 }, { "epoch": 0.8331824183495059, "grad_norm": 2.1689846515655518, "learning_rate": 1.4248528177819098e-06, "loss": 0.09394351243972779, "step": 31630 }, { "epoch": 0.8334458335940047, "grad_norm": 3.4968786239624023, "learning_rate": 1.4204668938759835e-06, "loss": 0.08469510674476624, "step": 31640 }, { "epoch": 0.8337092488385034, "grad_norm": 3.0571653842926025, "learning_rate": 1.416087214524403e-06, "loss": 0.0948081910610199, "step": 31650 }, { "epoch": 0.8339726640830022, "grad_norm": 3.0026471614837646, "learning_rate": 1.4117137829148885e-06, "loss": 0.0948763370513916, "step": 31660 }, { "epoch": 0.8342360793275009, "grad_norm": 2.0308921337127686, "learning_rate": 1.4073466022306148e-06, "loss": 0.0991324543952942, "step": 31670 }, { "epoch": 0.8344994945719996, "grad_norm": 2.7394890785217285, "learning_rate": 1.4029856756502026e-06, "loss": 0.09867826700210572, "step": 31680 }, { "epoch": 0.8347629098164984, "grad_norm": 2.494990587234497, "learning_rate": 1.3986310063477215e-06, "loss": 0.09718393087387085, "step": 31690 }, { "epoch": 0.835026325060997, "grad_norm": 2.3750226497650146, "learning_rate": 1.3942825974926888e-06, "loss": 0.09932678937911987, "step": 31700 }, { "epoch": 0.8352897403054959, "grad_norm": 2.5075037479400635, "learning_rate": 1.3899404522500626e-06, "loss": 0.09052485227584839, "step": 31710 }, { "epoch": 0.8355531555499945, "grad_norm": 2.7136781215667725, "learning_rate": 1.385604573780248e-06, "loss": 0.0859736979007721, "step": 31720 }, { "epoch": 0.8358165707944933, "grad_norm": 2.250074625015259, "learning_rate": 1.38127496523908e-06, "loss": 0.10791199207305908, "step": 31730 }, { "epoch": 0.836079986038992, "grad_norm": 2.141667366027832, "learning_rate": 1.376951629777833e-06, "loss": 0.09276441931724548, "step": 31740 }, { "epoch": 0.8363434012834908, "grad_norm": 2.7324423789978027, "learning_rate": 1.372634570543222e-06, "loss": 0.09273275136947631, "step": 31750 }, { "epoch": 0.8366068165279895, "grad_norm": 2.030651092529297, "learning_rate": 1.3683237906773873e-06, "loss": 0.09983213543891907, "step": 31760 }, { "epoch": 0.8368702317724882, "grad_norm": 2.142014741897583, "learning_rate": 1.3640192933178997e-06, "loss": 0.09424407482147217, "step": 31770 }, { "epoch": 0.837133647016987, "grad_norm": 2.7139089107513428, "learning_rate": 1.359721081597758e-06, "loss": 0.10267586708068847, "step": 31780 }, { "epoch": 0.8373970622614857, "grad_norm": 2.303256034851074, "learning_rate": 1.355429158645384e-06, "loss": 0.0911514163017273, "step": 31790 }, { "epoch": 0.8376604775059845, "grad_norm": 2.4281301498413086, "learning_rate": 1.3511435275846297e-06, "loss": 0.10740969181060792, "step": 31800 }, { "epoch": 0.8379238927504832, "grad_norm": 4.822164535522461, "learning_rate": 1.3468641915347613e-06, "loss": 0.09987375736236573, "step": 31810 }, { "epoch": 0.838187307994982, "grad_norm": 3.1201322078704834, "learning_rate": 1.3425911536104573e-06, "loss": 0.09119333028793335, "step": 31820 }, { "epoch": 0.8384507232394807, "grad_norm": 2.123441219329834, "learning_rate": 1.3383244169218268e-06, "loss": 0.085197114944458, "step": 31830 }, { "epoch": 0.8387141384839795, "grad_norm": 3.6737749576568604, "learning_rate": 1.3340639845743797e-06, "loss": 0.10608772039413453, "step": 31840 }, { "epoch": 0.8389775537284782, "grad_norm": 3.315472364425659, "learning_rate": 1.32980985966905e-06, "loss": 0.10263235569000244, "step": 31850 }, { "epoch": 0.8392409689729768, "grad_norm": 2.622882127761841, "learning_rate": 1.3255620453021656e-06, "loss": 0.09394654035568237, "step": 31860 }, { "epoch": 0.8395043842174756, "grad_norm": 3.495643377304077, "learning_rate": 1.3213205445654688e-06, "loss": 0.08981291055679322, "step": 31870 }, { "epoch": 0.8397677994619743, "grad_norm": 2.7612721920013428, "learning_rate": 1.3170853605461132e-06, "loss": 0.09914720058441162, "step": 31880 }, { "epoch": 0.8400312147064731, "grad_norm": 2.9124300479888916, "learning_rate": 1.3128564963266444e-06, "loss": 0.10380191802978515, "step": 31890 }, { "epoch": 0.8402946299509718, "grad_norm": 3.9218204021453857, "learning_rate": 1.3086339549850125e-06, "loss": 0.09471712708473205, "step": 31900 }, { "epoch": 0.8405580451954706, "grad_norm": 2.595320701599121, "learning_rate": 1.3044177395945657e-06, "loss": 0.09133238792419433, "step": 31910 }, { "epoch": 0.8408214604399693, "grad_norm": 2.4830710887908936, "learning_rate": 1.3002078532240447e-06, "loss": 0.10204141139984131, "step": 31920 }, { "epoch": 0.8410848756844681, "grad_norm": 2.281064033508301, "learning_rate": 1.296004298937592e-06, "loss": 0.09520718455314636, "step": 31930 }, { "epoch": 0.8413482909289668, "grad_norm": 3.3670694828033447, "learning_rate": 1.2918070797947346e-06, "loss": 0.07970075607299805, "step": 31940 }, { "epoch": 0.8416117061734655, "grad_norm": 2.2784476280212402, "learning_rate": 1.2876161988503843e-06, "loss": 0.10221903324127198, "step": 31950 }, { "epoch": 0.8418751214179643, "grad_norm": 2.845970630645752, "learning_rate": 1.2834316591548502e-06, "loss": 0.09553806185722351, "step": 31960 }, { "epoch": 0.842138536662463, "grad_norm": 3.359551429748535, "learning_rate": 1.279253463753819e-06, "loss": 0.10189886093139648, "step": 31970 }, { "epoch": 0.8424019519069618, "grad_norm": 3.594364881515503, "learning_rate": 1.2750816156883627e-06, "loss": 0.08864935040473938, "step": 31980 }, { "epoch": 0.8426653671514605, "grad_norm": 2.6195340156555176, "learning_rate": 1.2709161179949303e-06, "loss": 0.08854737281799316, "step": 31990 }, { "epoch": 0.8429287823959593, "grad_norm": 2.6784579753875732, "learning_rate": 1.266756973705351e-06, "loss": 0.09849635362625123, "step": 32000 }, { "epoch": 0.8431921976404579, "grad_norm": 2.8888518810272217, "learning_rate": 1.2626041858468318e-06, "loss": 0.09851099252700805, "step": 32010 }, { "epoch": 0.8434556128849566, "grad_norm": 2.464057445526123, "learning_rate": 1.2584577574419499e-06, "loss": 0.0868833303451538, "step": 32020 }, { "epoch": 0.8437190281294554, "grad_norm": 2.368112802505493, "learning_rate": 1.2543176915086542e-06, "loss": 0.0962238848209381, "step": 32030 }, { "epoch": 0.8439824433739541, "grad_norm": 3.5888867378234863, "learning_rate": 1.2501839910602632e-06, "loss": 0.09100362062454223, "step": 32040 }, { "epoch": 0.8442458586184529, "grad_norm": 2.8645694255828857, "learning_rate": 1.24605665910546e-06, "loss": 0.09829869270324706, "step": 32050 }, { "epoch": 0.8445092738629516, "grad_norm": 2.4353415966033936, "learning_rate": 1.2419356986483e-06, "loss": 0.09606446027755737, "step": 32060 }, { "epoch": 0.8447726891074504, "grad_norm": 2.771317958831787, "learning_rate": 1.2378211126881956e-06, "loss": 0.09497038722038269, "step": 32070 }, { "epoch": 0.8450361043519491, "grad_norm": 2.8086097240448, "learning_rate": 1.2337129042199147e-06, "loss": 0.08896275758743286, "step": 32080 }, { "epoch": 0.8452995195964479, "grad_norm": 3.4449453353881836, "learning_rate": 1.2296110762335933e-06, "loss": 0.08229601979255677, "step": 32090 }, { "epoch": 0.8455629348409466, "grad_norm": 2.6683766841888428, "learning_rate": 1.2255156317147165e-06, "loss": 0.08750817775726319, "step": 32100 }, { "epoch": 0.8458263500854453, "grad_norm": 3.4030373096466064, "learning_rate": 1.2214265736441277e-06, "loss": 0.08436395525932312, "step": 32110 }, { "epoch": 0.8460897653299441, "grad_norm": 3.3822243213653564, "learning_rate": 1.2173439049980174e-06, "loss": 0.08922678828239441, "step": 32120 }, { "epoch": 0.8463531805744428, "grad_norm": 3.51133394241333, "learning_rate": 1.2132676287479273e-06, "loss": 0.09211612939834594, "step": 32130 }, { "epoch": 0.8466165958189416, "grad_norm": 2.9267029762268066, "learning_rate": 1.209197747860752e-06, "loss": 0.08575397133827209, "step": 32140 }, { "epoch": 0.8468800110634402, "grad_norm": 3.039691209793091, "learning_rate": 1.2051342652987218e-06, "loss": 0.09874678254127503, "step": 32150 }, { "epoch": 0.847143426307939, "grad_norm": 3.790069818496704, "learning_rate": 1.2010771840194168e-06, "loss": 0.08581607341766358, "step": 32160 }, { "epoch": 0.8474068415524377, "grad_norm": 2.3769869804382324, "learning_rate": 1.1970265069757548e-06, "loss": 0.09554259181022644, "step": 32170 }, { "epoch": 0.8476702567969365, "grad_norm": 2.9863059520721436, "learning_rate": 1.1929822371159915e-06, "loss": 0.09346143007278443, "step": 32180 }, { "epoch": 0.8479336720414352, "grad_norm": 1.8375312089920044, "learning_rate": 1.188944377383725e-06, "loss": 0.09583905935287476, "step": 32190 }, { "epoch": 0.8481970872859339, "grad_norm": 2.3805463314056396, "learning_rate": 1.1849129307178841e-06, "loss": 0.09002001881599427, "step": 32200 }, { "epoch": 0.8484605025304327, "grad_norm": 2.7119216918945312, "learning_rate": 1.180887900052724e-06, "loss": 0.09272799491882325, "step": 32210 }, { "epoch": 0.8487239177749314, "grad_norm": 3.577364444732666, "learning_rate": 1.17686928831784e-06, "loss": 0.09631770849227905, "step": 32220 }, { "epoch": 0.8489873330194302, "grad_norm": 2.3039727210998535, "learning_rate": 1.1728570984381505e-06, "loss": 0.10124878883361817, "step": 32230 }, { "epoch": 0.8492507482639289, "grad_norm": 3.4403727054595947, "learning_rate": 1.1688513333339003e-06, "loss": 0.10467803478240967, "step": 32240 }, { "epoch": 0.8495141635084277, "grad_norm": 2.6790881156921387, "learning_rate": 1.1648519959206584e-06, "loss": 0.08570701479911805, "step": 32250 }, { "epoch": 0.8497775787529264, "grad_norm": 3.7121057510375977, "learning_rate": 1.1608590891093118e-06, "loss": 0.08273664712905884, "step": 32260 }, { "epoch": 0.8500409939974252, "grad_norm": 3.236447811126709, "learning_rate": 1.156872615806075e-06, "loss": 0.09565279483795167, "step": 32270 }, { "epoch": 0.8503044092419239, "grad_norm": 2.421407461166382, "learning_rate": 1.1528925789124734e-06, "loss": 0.0890588104724884, "step": 32280 }, { "epoch": 0.8505678244864225, "grad_norm": 2.0658349990844727, "learning_rate": 1.1489189813253499e-06, "loss": 0.0941986620426178, "step": 32290 }, { "epoch": 0.8508312397309213, "grad_norm": 2.43864107131958, "learning_rate": 1.1449518259368596e-06, "loss": 0.08663682341575622, "step": 32300 }, { "epoch": 0.85109465497542, "grad_norm": 2.62886118888855, "learning_rate": 1.1409911156344677e-06, "loss": 0.09357197284698486, "step": 32310 }, { "epoch": 0.8513580702199188, "grad_norm": 2.768280029296875, "learning_rate": 1.1370368533009557e-06, "loss": 0.08691961765289306, "step": 32320 }, { "epoch": 0.8516214854644175, "grad_norm": 1.9661054611206055, "learning_rate": 1.1330890418144014e-06, "loss": 0.08777657747268677, "step": 32330 }, { "epoch": 0.8518849007089163, "grad_norm": 3.143247604370117, "learning_rate": 1.1291476840481908e-06, "loss": 0.0976134181022644, "step": 32340 }, { "epoch": 0.852148315953415, "grad_norm": 3.8330531120300293, "learning_rate": 1.125212782871019e-06, "loss": 0.09618738889694214, "step": 32350 }, { "epoch": 0.8524117311979138, "grad_norm": 2.6147446632385254, "learning_rate": 1.1212843411468755e-06, "loss": 0.08619116544723511, "step": 32360 }, { "epoch": 0.8526751464424125, "grad_norm": 2.8868918418884277, "learning_rate": 1.1173623617350472e-06, "loss": 0.08923477530479432, "step": 32370 }, { "epoch": 0.8529385616869112, "grad_norm": 2.231767416000366, "learning_rate": 1.1134468474901227e-06, "loss": 0.09023739099502563, "step": 32380 }, { "epoch": 0.85320197693141, "grad_norm": 3.167440176010132, "learning_rate": 1.1095378012619796e-06, "loss": 0.08718887567520142, "step": 32390 }, { "epoch": 0.8534653921759087, "grad_norm": 2.905679702758789, "learning_rate": 1.1056352258957924e-06, "loss": 0.08304146528244019, "step": 32400 }, { "epoch": 0.8537288074204075, "grad_norm": 3.0718421936035156, "learning_rate": 1.1017391242320263e-06, "loss": 0.07914116382598876, "step": 32410 }, { "epoch": 0.8539922226649062, "grad_norm": 3.4653873443603516, "learning_rate": 1.0978494991064248e-06, "loss": 0.09271668791770935, "step": 32420 }, { "epoch": 0.854255637909405, "grad_norm": 2.3801381587982178, "learning_rate": 1.0939663533500311e-06, "loss": 0.08476808071136474, "step": 32430 }, { "epoch": 0.8545190531539036, "grad_norm": 1.7799657583236694, "learning_rate": 1.0900896897891634e-06, "loss": 0.09792077541351318, "step": 32440 }, { "epoch": 0.8547824683984024, "grad_norm": 4.517430305480957, "learning_rate": 1.0862195112454289e-06, "loss": 0.09406340718269349, "step": 32450 }, { "epoch": 0.8550458836429011, "grad_norm": 2.1165103912353516, "learning_rate": 1.0823558205357054e-06, "loss": 0.08327579498291016, "step": 32460 }, { "epoch": 0.8553092988873998, "grad_norm": 4.096268177032471, "learning_rate": 1.0784986204721537e-06, "loss": 0.09475295543670655, "step": 32470 }, { "epoch": 0.8555727141318986, "grad_norm": 4.412981033325195, "learning_rate": 1.0746479138622134e-06, "loss": 0.09297220706939698, "step": 32480 }, { "epoch": 0.8558361293763973, "grad_norm": 3.818556070327759, "learning_rate": 1.070803703508595e-06, "loss": 0.09954985380172729, "step": 32490 }, { "epoch": 0.8560995446208961, "grad_norm": 2.9838595390319824, "learning_rate": 1.066965992209279e-06, "loss": 0.08974815607070923, "step": 32500 }, { "epoch": 0.8563629598653948, "grad_norm": 2.6627280712127686, "learning_rate": 1.0631347827575179e-06, "loss": 0.08754905462265014, "step": 32510 }, { "epoch": 0.8566263751098936, "grad_norm": 4.114582061767578, "learning_rate": 1.0593100779418286e-06, "loss": 0.09870373010635376, "step": 32520 }, { "epoch": 0.8568897903543923, "grad_norm": 3.3311448097229004, "learning_rate": 1.0554918805460013e-06, "loss": 0.10722858905792236, "step": 32530 }, { "epoch": 0.8571532055988911, "grad_norm": 2.727813959121704, "learning_rate": 1.051680193349084e-06, "loss": 0.08591344356536865, "step": 32540 }, { "epoch": 0.8574166208433898, "grad_norm": 2.7076852321624756, "learning_rate": 1.0478750191253817e-06, "loss": 0.09174219369888306, "step": 32550 }, { "epoch": 0.8576800360878885, "grad_norm": 2.5887250900268555, "learning_rate": 1.0440763606444694e-06, "loss": 0.08186925649642944, "step": 32560 }, { "epoch": 0.8579434513323873, "grad_norm": 3.1188864707946777, "learning_rate": 1.0402842206711728e-06, "loss": 0.10322426557540894, "step": 32570 }, { "epoch": 0.8582068665768859, "grad_norm": 3.9307589530944824, "learning_rate": 1.0364986019655765e-06, "loss": 0.09305580854415893, "step": 32580 }, { "epoch": 0.8584702818213847, "grad_norm": 3.1091504096984863, "learning_rate": 1.0327195072830153e-06, "loss": 0.09280110597610473, "step": 32590 }, { "epoch": 0.8587336970658834, "grad_norm": 2.7193806171417236, "learning_rate": 1.028946939374076e-06, "loss": 0.10416502952575683, "step": 32600 }, { "epoch": 0.8589971123103822, "grad_norm": 3.207411050796509, "learning_rate": 1.0251809009846003e-06, "loss": 0.09636226892471314, "step": 32610 }, { "epoch": 0.8592605275548809, "grad_norm": 1.9670557975769043, "learning_rate": 1.0214213948556718e-06, "loss": 0.08387893438339233, "step": 32620 }, { "epoch": 0.8595239427993796, "grad_norm": 2.4855244159698486, "learning_rate": 1.0176684237236212e-06, "loss": 0.09378713965415955, "step": 32630 }, { "epoch": 0.8597873580438784, "grad_norm": 2.5902743339538574, "learning_rate": 1.0139219903200227e-06, "loss": 0.09726693630218505, "step": 32640 }, { "epoch": 0.8600507732883771, "grad_norm": 2.7057278156280518, "learning_rate": 1.0101820973716891e-06, "loss": 0.10099732875823975, "step": 32650 }, { "epoch": 0.8603141885328759, "grad_norm": 2.618551015853882, "learning_rate": 1.0064487476006812e-06, "loss": 0.08715583682060242, "step": 32660 }, { "epoch": 0.8605776037773746, "grad_norm": 2.4142792224884033, "learning_rate": 1.0027219437242907e-06, "loss": 0.09058166742324829, "step": 32670 }, { "epoch": 0.8608410190218734, "grad_norm": 2.912184238433838, "learning_rate": 9.990016884550425e-07, "loss": 0.09535446166992187, "step": 32680 }, { "epoch": 0.8611044342663721, "grad_norm": 2.679105043411255, "learning_rate": 9.952879845007024e-07, "loss": 0.08376578688621521, "step": 32690 }, { "epoch": 0.8613678495108709, "grad_norm": 4.099848747253418, "learning_rate": 9.915808345642642e-07, "loss": 0.0827410340309143, "step": 32700 }, { "epoch": 0.8616312647553696, "grad_norm": 2.932405710220337, "learning_rate": 9.878802413439514e-07, "loss": 0.08036942481994629, "step": 32710 }, { "epoch": 0.8618946799998682, "grad_norm": 3.552690029144287, "learning_rate": 9.841862075332143e-07, "loss": 0.09070841073989869, "step": 32720 }, { "epoch": 0.862158095244367, "grad_norm": 2.6638283729553223, "learning_rate": 9.804987358207274e-07, "loss": 0.08785814046859741, "step": 32730 }, { "epoch": 0.8624215104888657, "grad_norm": 2.399714469909668, "learning_rate": 9.768178288903985e-07, "loss": 0.09459115266799926, "step": 32740 }, { "epoch": 0.8626849257333645, "grad_norm": 2.579263925552368, "learning_rate": 9.731434894213465e-07, "loss": 0.09448435306549072, "step": 32750 }, { "epoch": 0.8629483409778632, "grad_norm": 2.9273815155029297, "learning_rate": 9.694757200879146e-07, "loss": 0.09674578905105591, "step": 32760 }, { "epoch": 0.863211756222362, "grad_norm": 2.494175672531128, "learning_rate": 9.658145235596628e-07, "loss": 0.09741510152816772, "step": 32770 }, { "epoch": 0.8634751714668607, "grad_norm": 3.871100664138794, "learning_rate": 9.621599025013672e-07, "loss": 0.09589773416519165, "step": 32780 }, { "epoch": 0.8637385867113595, "grad_norm": 2.6983232498168945, "learning_rate": 9.585118595730226e-07, "loss": 0.09504358768463135, "step": 32790 }, { "epoch": 0.8640020019558582, "grad_norm": 3.165138006210327, "learning_rate": 9.548703974298311e-07, "loss": 0.08987783193588257, "step": 32800 }, { "epoch": 0.8642654172003569, "grad_norm": 3.382502317428589, "learning_rate": 9.512355187222e-07, "loss": 0.09026905298233032, "step": 32810 }, { "epoch": 0.8645288324448557, "grad_norm": 1.7286157608032227, "learning_rate": 9.47607226095758e-07, "loss": 0.09850457310676575, "step": 32820 }, { "epoch": 0.8647922476893544, "grad_norm": 3.8584365844726562, "learning_rate": 9.439855221913297e-07, "loss": 0.09939780235290527, "step": 32830 }, { "epoch": 0.8650556629338532, "grad_norm": 2.8806002140045166, "learning_rate": 9.403704096449484e-07, "loss": 0.09173498153686524, "step": 32840 }, { "epoch": 0.8653190781783519, "grad_norm": 3.7886080741882324, "learning_rate": 9.367618910878484e-07, "loss": 0.09859368205070496, "step": 32850 }, { "epoch": 0.8655824934228507, "grad_norm": 2.7386887073516846, "learning_rate": 9.331599691464644e-07, "loss": 0.09828336238861084, "step": 32860 }, { "epoch": 0.8658459086673493, "grad_norm": 2.9955248832702637, "learning_rate": 9.295646464424323e-07, "loss": 0.08756810426712036, "step": 32870 }, { "epoch": 0.8661093239118481, "grad_norm": 3.6513478755950928, "learning_rate": 9.259759255925827e-07, "loss": 0.08096688389778137, "step": 32880 }, { "epoch": 0.8663727391563468, "grad_norm": 2.7545456886291504, "learning_rate": 9.223938092089413e-07, "loss": 0.09404809474945068, "step": 32890 }, { "epoch": 0.8666361544008455, "grad_norm": 1.9293814897537231, "learning_rate": 9.188182998987261e-07, "loss": 0.08108824491500854, "step": 32900 }, { "epoch": 0.8668995696453443, "grad_norm": 2.989119529724121, "learning_rate": 9.152494002643464e-07, "loss": 0.09082106947898864, "step": 32910 }, { "epoch": 0.867162984889843, "grad_norm": 3.224207639694214, "learning_rate": 9.116871129034066e-07, "loss": 0.09115540981292725, "step": 32920 }, { "epoch": 0.8674264001343418, "grad_norm": 2.923759698867798, "learning_rate": 9.081314404086872e-07, "loss": 0.08806981444358826, "step": 32930 }, { "epoch": 0.8676898153788405, "grad_norm": 3.381455183029175, "learning_rate": 9.045823853681602e-07, "loss": 0.09026565551757812, "step": 32940 }, { "epoch": 0.8679532306233393, "grad_norm": 2.9612879753112793, "learning_rate": 9.010399503649836e-07, "loss": 0.09641886353492737, "step": 32950 }, { "epoch": 0.868216645867838, "grad_norm": 3.5638515949249268, "learning_rate": 8.975041379774951e-07, "loss": 0.10316740274429322, "step": 32960 }, { "epoch": 0.8684800611123368, "grad_norm": 3.954968214035034, "learning_rate": 8.939749507792094e-07, "loss": 0.09808475971221924, "step": 32970 }, { "epoch": 0.8687434763568355, "grad_norm": 2.3529140949249268, "learning_rate": 8.904523913388218e-07, "loss": 0.08385375738143921, "step": 32980 }, { "epoch": 0.8690068916013342, "grad_norm": 2.9294066429138184, "learning_rate": 8.869364622202015e-07, "loss": 0.08308814167976379, "step": 32990 }, { "epoch": 0.869270306845833, "grad_norm": 2.6465377807617188, "learning_rate": 8.834271659823967e-07, "loss": 0.09238020181655884, "step": 33000 }, { "epoch": 0.8695337220903316, "grad_norm": 2.7824628353118896, "learning_rate": 8.799245051796246e-07, "loss": 0.08480805158615112, "step": 33010 }, { "epoch": 0.8697971373348304, "grad_norm": 2.520437240600586, "learning_rate": 8.764284823612667e-07, "loss": 0.0885523021221161, "step": 33020 }, { "epoch": 0.8700605525793291, "grad_norm": 3.3460330963134766, "learning_rate": 8.729391000718867e-07, "loss": 0.0946550965309143, "step": 33030 }, { "epoch": 0.8703239678238279, "grad_norm": 3.3788416385650635, "learning_rate": 8.694563608512041e-07, "loss": 0.09560973644256592, "step": 33040 }, { "epoch": 0.8705873830683266, "grad_norm": 3.1502280235290527, "learning_rate": 8.659802672341078e-07, "loss": 0.08820499181747436, "step": 33050 }, { "epoch": 0.8708507983128254, "grad_norm": 2.467752456665039, "learning_rate": 8.625108217506472e-07, "loss": 0.08558632135391235, "step": 33060 }, { "epoch": 0.8711142135573241, "grad_norm": 3.286099672317505, "learning_rate": 8.590480269260337e-07, "loss": 0.08870554566383362, "step": 33070 }, { "epoch": 0.8713776288018228, "grad_norm": 3.040219306945801, "learning_rate": 8.555918852806421e-07, "loss": 0.09499970674514771, "step": 33080 }, { "epoch": 0.8716410440463216, "grad_norm": 2.5188846588134766, "learning_rate": 8.521423993300005e-07, "loss": 0.09503021240234374, "step": 33090 }, { "epoch": 0.8719044592908203, "grad_norm": 2.995647668838501, "learning_rate": 8.486995715847923e-07, "loss": 0.08325605988502502, "step": 33100 }, { "epoch": 0.8721678745353191, "grad_norm": 3.2356150150299072, "learning_rate": 8.452634045508557e-07, "loss": 0.0868962287902832, "step": 33110 }, { "epoch": 0.8724312897798178, "grad_norm": 3.5717859268188477, "learning_rate": 8.41833900729182e-07, "loss": 0.0911331295967102, "step": 33120 }, { "epoch": 0.8726947050243166, "grad_norm": 2.471527576446533, "learning_rate": 8.384110626159126e-07, "loss": 0.09377825260162354, "step": 33130 }, { "epoch": 0.8729581202688153, "grad_norm": 2.479393243789673, "learning_rate": 8.349948927023388e-07, "loss": 0.0905720829963684, "step": 33140 }, { "epoch": 0.8732215355133139, "grad_norm": 3.3268473148345947, "learning_rate": 8.315853934748908e-07, "loss": 0.09636127948760986, "step": 33150 }, { "epoch": 0.8734849507578127, "grad_norm": 3.1197669506073, "learning_rate": 8.281825674151533e-07, "loss": 0.09548139572143555, "step": 33160 }, { "epoch": 0.8737483660023114, "grad_norm": 2.4029557704925537, "learning_rate": 8.247864169998499e-07, "loss": 0.0899128794670105, "step": 33170 }, { "epoch": 0.8740117812468102, "grad_norm": 2.3152081966400146, "learning_rate": 8.213969447008441e-07, "loss": 0.08304970264434815, "step": 33180 }, { "epoch": 0.8742751964913089, "grad_norm": 2.941631555557251, "learning_rate": 8.180141529851404e-07, "loss": 0.09519110918045044, "step": 33190 }, { "epoch": 0.8745386117358077, "grad_norm": 3.679533004760742, "learning_rate": 8.146380443148794e-07, "loss": 0.09257259964942932, "step": 33200 }, { "epoch": 0.8748020269803064, "grad_norm": 3.0056254863739014, "learning_rate": 8.11268621147342e-07, "loss": 0.09300330877304078, "step": 33210 }, { "epoch": 0.8750654422248052, "grad_norm": 3.2489771842956543, "learning_rate": 8.079058859349364e-07, "loss": 0.09475352764129638, "step": 33220 }, { "epoch": 0.8753288574693039, "grad_norm": 2.248199224472046, "learning_rate": 8.045498411252095e-07, "loss": 0.09297013282775879, "step": 33230 }, { "epoch": 0.8755922727138026, "grad_norm": 3.879396915435791, "learning_rate": 8.012004891608327e-07, "loss": 0.09872850775718689, "step": 33240 }, { "epoch": 0.8758556879583014, "grad_norm": 3.748809814453125, "learning_rate": 7.978578324796072e-07, "loss": 0.08885804414749146, "step": 33250 }, { "epoch": 0.8761191032028001, "grad_norm": 2.2327675819396973, "learning_rate": 7.945218735144689e-07, "loss": 0.08309193849563598, "step": 33260 }, { "epoch": 0.8763825184472989, "grad_norm": 3.0661985874176025, "learning_rate": 7.91192614693469e-07, "loss": 0.08637797236442565, "step": 33270 }, { "epoch": 0.8766459336917976, "grad_norm": 3.4623770713806152, "learning_rate": 7.878700584397836e-07, "loss": 0.08220633268356323, "step": 33280 }, { "epoch": 0.8769093489362964, "grad_norm": 3.894061803817749, "learning_rate": 7.84554207171716e-07, "loss": 0.08610810041427612, "step": 33290 }, { "epoch": 0.877172764180795, "grad_norm": 3.335930824279785, "learning_rate": 7.812450633026847e-07, "loss": 0.07992979288101196, "step": 33300 }, { "epoch": 0.8774361794252938, "grad_norm": 3.5593507289886475, "learning_rate": 7.779426292412273e-07, "loss": 0.07768430709838867, "step": 33310 }, { "epoch": 0.8776995946697925, "grad_norm": 2.567718267440796, "learning_rate": 7.746469073909979e-07, "loss": 0.08015058636665344, "step": 33320 }, { "epoch": 0.8779630099142912, "grad_norm": 2.6669721603393555, "learning_rate": 7.713579001507632e-07, "loss": 0.0942508339881897, "step": 33330 }, { "epoch": 0.87822642515879, "grad_norm": 1.973449468612671, "learning_rate": 7.680756099144072e-07, "loss": 0.08820632696151734, "step": 33340 }, { "epoch": 0.8784898404032887, "grad_norm": 3.5294077396392822, "learning_rate": 7.648000390709221e-07, "loss": 0.09451942443847657, "step": 33350 }, { "epoch": 0.8787532556477875, "grad_norm": 2.6963579654693604, "learning_rate": 7.615311900044087e-07, "loss": 0.07109616994857788, "step": 33360 }, { "epoch": 0.8790166708922862, "grad_norm": 4.417686939239502, "learning_rate": 7.582690650940771e-07, "loss": 0.0931973397731781, "step": 33370 }, { "epoch": 0.879280086136785, "grad_norm": 2.7317097187042236, "learning_rate": 7.550136667142393e-07, "loss": 0.08440893888473511, "step": 33380 }, { "epoch": 0.8795435013812837, "grad_norm": 3.529381275177002, "learning_rate": 7.517649972343199e-07, "loss": 0.07867162823677062, "step": 33390 }, { "epoch": 0.8798069166257825, "grad_norm": 3.21932053565979, "learning_rate": 7.485230590188397e-07, "loss": 0.08154574632644654, "step": 33400 }, { "epoch": 0.8800703318702812, "grad_norm": 3.5386691093444824, "learning_rate": 7.452878544274167e-07, "loss": 0.09059983491897583, "step": 33410 }, { "epoch": 0.8803337471147799, "grad_norm": 3.2739651203155518, "learning_rate": 7.420593858147773e-07, "loss": 0.09664689898490905, "step": 33420 }, { "epoch": 0.8805971623592787, "grad_norm": 3.156334400177002, "learning_rate": 7.388376555307386e-07, "loss": 0.09249724149703979, "step": 33430 }, { "epoch": 0.8808605776037773, "grad_norm": 2.9653429985046387, "learning_rate": 7.356226659202137e-07, "loss": 0.0808334469795227, "step": 33440 }, { "epoch": 0.8811239928482761, "grad_norm": 2.656538724899292, "learning_rate": 7.324144193232119e-07, "loss": 0.08069849014282227, "step": 33450 }, { "epoch": 0.8813874080927748, "grad_norm": 2.6552228927612305, "learning_rate": 7.29212918074833e-07, "loss": 0.07665507793426514, "step": 33460 }, { "epoch": 0.8816508233372736, "grad_norm": 3.1570897102355957, "learning_rate": 7.260181645052689e-07, "loss": 0.09205914735794067, "step": 33470 }, { "epoch": 0.8819142385817723, "grad_norm": 2.447861671447754, "learning_rate": 7.228301609397981e-07, "loss": 0.09368026256561279, "step": 33480 }, { "epoch": 0.8821776538262711, "grad_norm": 2.9855613708496094, "learning_rate": 7.196489096987869e-07, "loss": 0.09669078588485717, "step": 33490 }, { "epoch": 0.8824410690707698, "grad_norm": 3.2997336387634277, "learning_rate": 7.164744130976886e-07, "loss": 0.07191516160964966, "step": 33500 }, { "epoch": 0.8827044843152685, "grad_norm": 2.5468127727508545, "learning_rate": 7.133066734470351e-07, "loss": 0.09474445581436157, "step": 33510 }, { "epoch": 0.8829678995597673, "grad_norm": 2.9569320678710938, "learning_rate": 7.101456930524509e-07, "loss": 0.09007320404052735, "step": 33520 }, { "epoch": 0.883231314804266, "grad_norm": 3.0181479454040527, "learning_rate": 7.069914742146267e-07, "loss": 0.09460103511810303, "step": 33530 }, { "epoch": 0.8834947300487648, "grad_norm": 3.328418493270874, "learning_rate": 7.038440192293416e-07, "loss": 0.08958646655082703, "step": 33540 }, { "epoch": 0.8837581452932635, "grad_norm": 2.1959099769592285, "learning_rate": 7.007033303874511e-07, "loss": 0.08472674489021301, "step": 33550 }, { "epoch": 0.8840215605377623, "grad_norm": 2.634690523147583, "learning_rate": 6.97569409974882e-07, "loss": 0.09307442903518677, "step": 33560 }, { "epoch": 0.884284975782261, "grad_norm": 3.1193883419036865, "learning_rate": 6.944422602726375e-07, "loss": 0.07444275617599487, "step": 33570 }, { "epoch": 0.8845483910267598, "grad_norm": 3.240511655807495, "learning_rate": 6.913218835567925e-07, "loss": 0.09137159585952759, "step": 33580 }, { "epoch": 0.8848118062712584, "grad_norm": 3.961134433746338, "learning_rate": 6.882082820984892e-07, "loss": 0.09516191482543945, "step": 33590 }, { "epoch": 0.8850752215157571, "grad_norm": 2.364638328552246, "learning_rate": 6.851014581639437e-07, "loss": 0.07433972358703614, "step": 33600 }, { "epoch": 0.8853386367602559, "grad_norm": 2.030583143234253, "learning_rate": 6.820014140144382e-07, "loss": 0.08454200029373168, "step": 33610 }, { "epoch": 0.8856020520047546, "grad_norm": 3.0560145378112793, "learning_rate": 6.789081519063134e-07, "loss": 0.0935598373413086, "step": 33620 }, { "epoch": 0.8858654672492534, "grad_norm": 2.9175541400909424, "learning_rate": 6.758216740909829e-07, "loss": 0.09533923864364624, "step": 33630 }, { "epoch": 0.8861288824937521, "grad_norm": 2.701786756515503, "learning_rate": 6.727419828149173e-07, "loss": 0.09324216842651367, "step": 33640 }, { "epoch": 0.8863922977382509, "grad_norm": 2.5443944931030273, "learning_rate": 6.696690803196482e-07, "loss": 0.08605523109436035, "step": 33650 }, { "epoch": 0.8866557129827496, "grad_norm": 2.8409669399261475, "learning_rate": 6.666029688417674e-07, "loss": 0.0943009912967682, "step": 33660 }, { "epoch": 0.8869191282272483, "grad_norm": 2.658543825149536, "learning_rate": 6.635436506129223e-07, "loss": 0.0861518144607544, "step": 33670 }, { "epoch": 0.8871825434717471, "grad_norm": 3.5103096961975098, "learning_rate": 6.60491127859818e-07, "loss": 0.08964366912841797, "step": 33680 }, { "epoch": 0.8874459587162458, "grad_norm": 2.467257261276245, "learning_rate": 6.57445402804211e-07, "loss": 0.08627638816833497, "step": 33690 }, { "epoch": 0.8877093739607446, "grad_norm": 3.5641093254089355, "learning_rate": 6.544064776629123e-07, "loss": 0.0882696270942688, "step": 33700 }, { "epoch": 0.8879727892052433, "grad_norm": 2.548527956008911, "learning_rate": 6.513743546477824e-07, "loss": 0.07462223172187805, "step": 33710 }, { "epoch": 0.888236204449742, "grad_norm": 3.1176202297210693, "learning_rate": 6.483490359657296e-07, "loss": 0.08858217597007752, "step": 33720 }, { "epoch": 0.8884996196942407, "grad_norm": 2.148355484008789, "learning_rate": 6.453305238187147e-07, "loss": 0.07962720394134522, "step": 33730 }, { "epoch": 0.8887630349387395, "grad_norm": 2.5675721168518066, "learning_rate": 6.423188204037412e-07, "loss": 0.08759270310401916, "step": 33740 }, { "epoch": 0.8890264501832382, "grad_norm": 2.5894200801849365, "learning_rate": 6.393139279128513e-07, "loss": 0.08137593269348145, "step": 33750 }, { "epoch": 0.8892898654277369, "grad_norm": 3.0000603199005127, "learning_rate": 6.3631584853314e-07, "loss": 0.0890834629535675, "step": 33760 }, { "epoch": 0.8895532806722357, "grad_norm": 3.6637256145477295, "learning_rate": 6.333245844467384e-07, "loss": 0.09022608995437623, "step": 33770 }, { "epoch": 0.8898166959167344, "grad_norm": 2.444878339767456, "learning_rate": 6.303401378308172e-07, "loss": 0.08021628856658936, "step": 33780 }, { "epoch": 0.8900801111612332, "grad_norm": 2.591137409210205, "learning_rate": 6.273625108575843e-07, "loss": 0.0842495620250702, "step": 33790 }, { "epoch": 0.8903435264057319, "grad_norm": 2.72381854057312, "learning_rate": 6.243917056942849e-07, "loss": 0.0849428653717041, "step": 33800 }, { "epoch": 0.8906069416502307, "grad_norm": 3.718505859375, "learning_rate": 6.214277245032019e-07, "loss": 0.08183403611183167, "step": 33810 }, { "epoch": 0.8908703568947294, "grad_norm": 3.9367048740386963, "learning_rate": 6.184705694416471e-07, "loss": 0.08650360107421876, "step": 33820 }, { "epoch": 0.8911337721392282, "grad_norm": 3.2969703674316406, "learning_rate": 6.155202426619655e-07, "loss": 0.0888897180557251, "step": 33830 }, { "epoch": 0.8913971873837269, "grad_norm": 3.014298915863037, "learning_rate": 6.125767463115317e-07, "loss": 0.08852174878120422, "step": 33840 }, { "epoch": 0.8916606026282256, "grad_norm": 3.2000608444213867, "learning_rate": 6.096400825327487e-07, "loss": 0.09367623925209045, "step": 33850 }, { "epoch": 0.8919240178727244, "grad_norm": 3.106494665145874, "learning_rate": 6.067102534630498e-07, "loss": 0.08540844917297363, "step": 33860 }, { "epoch": 0.892187433117223, "grad_norm": 3.2809414863586426, "learning_rate": 6.037872612348905e-07, "loss": 0.07725362777709961, "step": 33870 }, { "epoch": 0.8924508483617218, "grad_norm": 3.141187906265259, "learning_rate": 6.008711079757467e-07, "loss": 0.08399592638015747, "step": 33880 }, { "epoch": 0.8927142636062205, "grad_norm": 3.0430586338043213, "learning_rate": 5.979617958081241e-07, "loss": 0.07982717752456665, "step": 33890 }, { "epoch": 0.8929776788507193, "grad_norm": 2.485133647918701, "learning_rate": 5.950593268495464e-07, "loss": 0.08526681661605835, "step": 33900 }, { "epoch": 0.893241094095218, "grad_norm": 2.3360507488250732, "learning_rate": 5.921637032125538e-07, "loss": 0.07849393486976623, "step": 33910 }, { "epoch": 0.8935045093397168, "grad_norm": 3.0273056030273438, "learning_rate": 5.89274927004706e-07, "loss": 0.08327957391738891, "step": 33920 }, { "epoch": 0.8937679245842155, "grad_norm": 2.6604342460632324, "learning_rate": 5.863930003285779e-07, "loss": 0.07669328451156616, "step": 33930 }, { "epoch": 0.8940313398287142, "grad_norm": 3.432288408279419, "learning_rate": 5.835179252817635e-07, "loss": 0.08898953199386597, "step": 33940 }, { "epoch": 0.894294755073213, "grad_norm": 2.2986581325531006, "learning_rate": 5.806497039568648e-07, "loss": 0.07497620582580566, "step": 33950 }, { "epoch": 0.8945581703177117, "grad_norm": 2.62571120262146, "learning_rate": 5.777883384414973e-07, "loss": 0.07749274969100953, "step": 33960 }, { "epoch": 0.8948215855622105, "grad_norm": 2.605116128921509, "learning_rate": 5.749338308182851e-07, "loss": 0.0791547179222107, "step": 33970 }, { "epoch": 0.8950850008067092, "grad_norm": 3.373314619064331, "learning_rate": 5.720861831648617e-07, "loss": 0.09009881019592285, "step": 33980 }, { "epoch": 0.895348416051208, "grad_norm": 3.18033504486084, "learning_rate": 5.692453975538703e-07, "loss": 0.0780484914779663, "step": 33990 }, { "epoch": 0.8956118312957067, "grad_norm": 1.9454399347305298, "learning_rate": 5.664114760529571e-07, "loss": 0.08226714134216309, "step": 34000 }, { "epoch": 0.8958752465402054, "grad_norm": 2.5894973278045654, "learning_rate": 5.635844207247687e-07, "loss": 0.08866448402404785, "step": 34010 }, { "epoch": 0.8961386617847041, "grad_norm": 2.8551077842712402, "learning_rate": 5.607642336269615e-07, "loss": 0.07710912227630615, "step": 34020 }, { "epoch": 0.8964020770292028, "grad_norm": 3.1512179374694824, "learning_rate": 5.579509168121888e-07, "loss": 0.08187508583068848, "step": 34030 }, { "epoch": 0.8966654922737016, "grad_norm": 3.1806459426879883, "learning_rate": 5.551444723281029e-07, "loss": 0.08382105827331543, "step": 34040 }, { "epoch": 0.8969289075182003, "grad_norm": 3.6934244632720947, "learning_rate": 5.523449022173566e-07, "loss": 0.10374695062637329, "step": 34050 }, { "epoch": 0.8971923227626991, "grad_norm": 2.292755365371704, "learning_rate": 5.495522085175942e-07, "loss": 0.07769957184791565, "step": 34060 }, { "epoch": 0.8974557380071978, "grad_norm": 3.475640058517456, "learning_rate": 5.467663932614642e-07, "loss": 0.08345956802368164, "step": 34070 }, { "epoch": 0.8977191532516966, "grad_norm": 2.878262996673584, "learning_rate": 5.439874584766014e-07, "loss": 0.09233187437057495, "step": 34080 }, { "epoch": 0.8979825684961953, "grad_norm": 2.855509042739868, "learning_rate": 5.412154061856301e-07, "loss": 0.07575578093528748, "step": 34090 }, { "epoch": 0.8982459837406941, "grad_norm": 3.013843297958374, "learning_rate": 5.38450238406174e-07, "loss": 0.08478127717971802, "step": 34100 }, { "epoch": 0.8985093989851928, "grad_norm": 2.82547664642334, "learning_rate": 5.356919571508401e-07, "loss": 0.09485266208648682, "step": 34110 }, { "epoch": 0.8987728142296915, "grad_norm": 4.216487884521484, "learning_rate": 5.329405644272268e-07, "loss": 0.09612655043601989, "step": 34120 }, { "epoch": 0.8990362294741903, "grad_norm": 3.271833658218384, "learning_rate": 5.301960622379132e-07, "loss": 0.08470184803009033, "step": 34130 }, { "epoch": 0.899299644718689, "grad_norm": 4.0348801612854, "learning_rate": 5.27458452580466e-07, "loss": 0.09290415048599243, "step": 34140 }, { "epoch": 0.8995630599631877, "grad_norm": 3.52473783493042, "learning_rate": 5.24727737447438e-07, "loss": 0.08443785309791565, "step": 34150 }, { "epoch": 0.8998264752076864, "grad_norm": 3.1763179302215576, "learning_rate": 5.220039188263615e-07, "loss": 0.07959506511688233, "step": 34160 }, { "epoch": 0.9000898904521852, "grad_norm": 2.7289421558380127, "learning_rate": 5.192869986997484e-07, "loss": 0.08822963833808899, "step": 34170 }, { "epoch": 0.9003533056966839, "grad_norm": 3.439203977584839, "learning_rate": 5.165769790450903e-07, "loss": 0.08368765711784362, "step": 34180 }, { "epoch": 0.9006167209411826, "grad_norm": 3.032095193862915, "learning_rate": 5.138738618348548e-07, "loss": 0.08699430227279663, "step": 34190 }, { "epoch": 0.9008801361856814, "grad_norm": 3.117347478866577, "learning_rate": 5.111776490364906e-07, "loss": 0.09787341952323914, "step": 34200 }, { "epoch": 0.9011435514301801, "grad_norm": 2.6589720249176025, "learning_rate": 5.084883426124165e-07, "loss": 0.06956454515457153, "step": 34210 }, { "epoch": 0.9014069666746789, "grad_norm": 2.780609369277954, "learning_rate": 5.058059445200237e-07, "loss": 0.07746808528900147, "step": 34220 }, { "epoch": 0.9016703819191776, "grad_norm": 3.2057740688323975, "learning_rate": 5.031304567116802e-07, "loss": 0.08737049102783204, "step": 34230 }, { "epoch": 0.9019337971636764, "grad_norm": 3.680811882019043, "learning_rate": 5.004618811347195e-07, "loss": 0.08970011472702026, "step": 34240 }, { "epoch": 0.9021972124081751, "grad_norm": 3.1675562858581543, "learning_rate": 4.978002197314469e-07, "loss": 0.0942071259021759, "step": 34250 }, { "epoch": 0.9024606276526739, "grad_norm": 2.48836612701416, "learning_rate": 4.951454744391338e-07, "loss": 0.0863286793231964, "step": 34260 }, { "epoch": 0.9027240428971726, "grad_norm": 2.7650160789489746, "learning_rate": 4.924976471900178e-07, "loss": 0.0759728193283081, "step": 34270 }, { "epoch": 0.9029874581416713, "grad_norm": 4.61401891708374, "learning_rate": 4.89856739911303e-07, "loss": 0.07565252780914307, "step": 34280 }, { "epoch": 0.90325087338617, "grad_norm": 3.5134854316711426, "learning_rate": 4.872227545251562e-07, "loss": 0.0793230414390564, "step": 34290 }, { "epoch": 0.9035142886306687, "grad_norm": 2.899937391281128, "learning_rate": 4.84595692948705e-07, "loss": 0.07650419473648071, "step": 34300 }, { "epoch": 0.9037777038751675, "grad_norm": 3.8000197410583496, "learning_rate": 4.819755570940377e-07, "loss": 0.09128966331481933, "step": 34310 }, { "epoch": 0.9040411191196662, "grad_norm": 2.8499674797058105, "learning_rate": 4.793623488682008e-07, "loss": 0.08331849575042724, "step": 34320 }, { "epoch": 0.904304534364165, "grad_norm": 3.1989285945892334, "learning_rate": 4.7675607017320323e-07, "loss": 0.08080207705497741, "step": 34330 }, { "epoch": 0.9045679496086637, "grad_norm": 4.765405654907227, "learning_rate": 4.741567229060062e-07, "loss": 0.08041363954544067, "step": 34340 }, { "epoch": 0.9048313648531625, "grad_norm": 3.254171848297119, "learning_rate": 4.715643089585242e-07, "loss": 0.08600177764892578, "step": 34350 }, { "epoch": 0.9050947800976612, "grad_norm": 2.8716530799865723, "learning_rate": 4.6897883021763013e-07, "loss": 0.07945472598075867, "step": 34360 }, { "epoch": 0.9053581953421599, "grad_norm": 3.073305130004883, "learning_rate": 4.664002885651475e-07, "loss": 0.08417291641235351, "step": 34370 }, { "epoch": 0.9056216105866587, "grad_norm": 4.2801384925842285, "learning_rate": 4.638286858778496e-07, "loss": 0.08378046751022339, "step": 34380 }, { "epoch": 0.9058850258311574, "grad_norm": 4.2126994132995605, "learning_rate": 4.612640240274591e-07, "loss": 0.09171774983406067, "step": 34390 }, { "epoch": 0.9061484410756562, "grad_norm": 3.448819637298584, "learning_rate": 4.587063048806462e-07, "loss": 0.08948748111724854, "step": 34400 }, { "epoch": 0.9064118563201549, "grad_norm": 2.5603842735290527, "learning_rate": 4.561555302990328e-07, "loss": 0.09010646343231202, "step": 34410 }, { "epoch": 0.9066752715646537, "grad_norm": 3.6640329360961914, "learning_rate": 4.536117021391806e-07, "loss": 0.08889051675796508, "step": 34420 }, { "epoch": 0.9069386868091524, "grad_norm": 3.0882344245910645, "learning_rate": 4.5107482225259735e-07, "loss": 0.08195672035217286, "step": 34430 }, { "epoch": 0.9072021020536511, "grad_norm": 2.8446543216705322, "learning_rate": 4.4854489248573494e-07, "loss": 0.07883623242378235, "step": 34440 }, { "epoch": 0.9074655172981498, "grad_norm": 3.054704189300537, "learning_rate": 4.460219146799838e-07, "loss": 0.07816315293312073, "step": 34450 }, { "epoch": 0.9077289325426485, "grad_norm": 2.85972261428833, "learning_rate": 4.4350589067167717e-07, "loss": 0.07812363505363465, "step": 34460 }, { "epoch": 0.9079923477871473, "grad_norm": 3.679262161254883, "learning_rate": 4.409968222920868e-07, "loss": 0.09302793145179748, "step": 34470 }, { "epoch": 0.908255763031646, "grad_norm": 3.5264246463775635, "learning_rate": 4.384947113674165e-07, "loss": 0.08623006343841552, "step": 34480 }, { "epoch": 0.9085191782761448, "grad_norm": 2.653489351272583, "learning_rate": 4.35999559718816e-07, "loss": 0.0822284460067749, "step": 34490 }, { "epoch": 0.9087825935206435, "grad_norm": 3.262420892715454, "learning_rate": 4.3351136916236156e-07, "loss": 0.0927403450012207, "step": 34500 }, { "epoch": 0.9090460087651423, "grad_norm": 3.646331310272217, "learning_rate": 4.3103014150906675e-07, "loss": 0.07886837720870972, "step": 34510 }, { "epoch": 0.909309424009641, "grad_norm": 3.677265167236328, "learning_rate": 4.285558785648747e-07, "loss": 0.08859394788742066, "step": 34520 }, { "epoch": 0.9095728392541398, "grad_norm": 2.810969352722168, "learning_rate": 4.260885821306604e-07, "loss": 0.07968290448188782, "step": 34530 }, { "epoch": 0.9098362544986385, "grad_norm": 2.7076566219329834, "learning_rate": 4.236282540022307e-07, "loss": 0.07704629302024842, "step": 34540 }, { "epoch": 0.9100996697431372, "grad_norm": 3.138796091079712, "learning_rate": 4.2117489597031767e-07, "loss": 0.09392867088317872, "step": 34550 }, { "epoch": 0.910363084987636, "grad_norm": 4.243546009063721, "learning_rate": 4.1872850982058067e-07, "loss": 0.08957300186157227, "step": 34560 }, { "epoch": 0.9106265002321347, "grad_norm": 2.8912806510925293, "learning_rate": 4.1628909733360534e-07, "loss": 0.0848383903503418, "step": 34570 }, { "epoch": 0.9108899154766334, "grad_norm": 3.4595119953155518, "learning_rate": 4.1385666028490036e-07, "loss": 0.08813338875770568, "step": 34580 }, { "epoch": 0.9111533307211321, "grad_norm": 2.9882876873016357, "learning_rate": 4.114312004449006e-07, "loss": 0.0734938621520996, "step": 34590 }, { "epoch": 0.9114167459656309, "grad_norm": 3.7188119888305664, "learning_rate": 4.090127195789606e-07, "loss": 0.08238164186477662, "step": 34600 }, { "epoch": 0.9116801612101296, "grad_norm": 3.217554807662964, "learning_rate": 4.066012194473512e-07, "loss": 0.09467202425003052, "step": 34610 }, { "epoch": 0.9119435764546284, "grad_norm": 2.7586052417755127, "learning_rate": 4.041967018052706e-07, "loss": 0.09081991910934448, "step": 34620 }, { "epoch": 0.9122069916991271, "grad_norm": 2.95207142829895, "learning_rate": 4.0179916840282776e-07, "loss": 0.0805223047733307, "step": 34630 }, { "epoch": 0.9124704069436258, "grad_norm": 2.584061861038208, "learning_rate": 3.9940862098505117e-07, "loss": 0.07862425446510315, "step": 34640 }, { "epoch": 0.9127338221881246, "grad_norm": 3.72965669631958, "learning_rate": 3.970250612918858e-07, "loss": 0.1009333610534668, "step": 34650 }, { "epoch": 0.9129972374326233, "grad_norm": 2.698894739151001, "learning_rate": 3.946484910581871e-07, "loss": 0.0781123161315918, "step": 34660 }, { "epoch": 0.9132606526771221, "grad_norm": 2.8210604190826416, "learning_rate": 3.9227891201372827e-07, "loss": 0.07511930465698242, "step": 34670 }, { "epoch": 0.9135240679216208, "grad_norm": 3.763813018798828, "learning_rate": 3.899163258831895e-07, "loss": 0.08558266162872315, "step": 34680 }, { "epoch": 0.9137874831661196, "grad_norm": 2.7194886207580566, "learning_rate": 3.8756073438616203e-07, "loss": 0.08160321712493897, "step": 34690 }, { "epoch": 0.9140508984106183, "grad_norm": 2.591460704803467, "learning_rate": 3.8521213923714775e-07, "loss": 0.08520693778991699, "step": 34700 }, { "epoch": 0.914314313655117, "grad_norm": 2.4975626468658447, "learning_rate": 3.8287054214555694e-07, "loss": 0.07685672044754029, "step": 34710 }, { "epoch": 0.9145777288996157, "grad_norm": 2.337996006011963, "learning_rate": 3.805359448157031e-07, "loss": 0.08111155033111572, "step": 34720 }, { "epoch": 0.9148411441441144, "grad_norm": 3.118986129760742, "learning_rate": 3.78208348946808e-07, "loss": 0.0774409830570221, "step": 34730 }, { "epoch": 0.9151045593886132, "grad_norm": 2.1018879413604736, "learning_rate": 3.7588775623299433e-07, "loss": 0.07624192237854004, "step": 34740 }, { "epoch": 0.9153679746331119, "grad_norm": 3.5074310302734375, "learning_rate": 3.735741683632932e-07, "loss": 0.08506640791893005, "step": 34750 }, { "epoch": 0.9156313898776107, "grad_norm": 2.5031561851501465, "learning_rate": 3.712675870216309e-07, "loss": 0.07678319811820984, "step": 34760 }, { "epoch": 0.9158948051221094, "grad_norm": 3.650597333908081, "learning_rate": 3.689680138868379e-07, "loss": 0.08356554508209228, "step": 34770 }, { "epoch": 0.9161582203666082, "grad_norm": 2.6690211296081543, "learning_rate": 3.66675450632642e-07, "loss": 0.07278918623924255, "step": 34780 }, { "epoch": 0.9164216356111069, "grad_norm": 2.8692140579223633, "learning_rate": 3.643898989276684e-07, "loss": 0.08918020725250245, "step": 34790 }, { "epoch": 0.9166850508556056, "grad_norm": 4.430020809173584, "learning_rate": 3.6211136043544425e-07, "loss": 0.08986327648162842, "step": 34800 }, { "epoch": 0.9169484661001044, "grad_norm": 2.8953988552093506, "learning_rate": 3.598398368143863e-07, "loss": 0.0898177981376648, "step": 34810 }, { "epoch": 0.9172118813446031, "grad_norm": 3.044879674911499, "learning_rate": 3.575753297178053e-07, "loss": 0.07525169849395752, "step": 34820 }, { "epoch": 0.9174752965891019, "grad_norm": 3.1370959281921387, "learning_rate": 3.553178407939095e-07, "loss": 0.08240444660186767, "step": 34830 }, { "epoch": 0.9177387118336006, "grad_norm": 2.649618625640869, "learning_rate": 3.5306737168579685e-07, "loss": 0.08106522560119629, "step": 34840 }, { "epoch": 0.9180021270780994, "grad_norm": 2.754831314086914, "learning_rate": 3.5082392403145373e-07, "loss": 0.08210763931274415, "step": 34850 }, { "epoch": 0.918265542322598, "grad_norm": 2.366929531097412, "learning_rate": 3.485874994637606e-07, "loss": 0.08231836557388306, "step": 34860 }, { "epoch": 0.9185289575670968, "grad_norm": 3.033207893371582, "learning_rate": 3.463580996104821e-07, "loss": 0.08935953378677368, "step": 34870 }, { "epoch": 0.9187923728115955, "grad_norm": 3.5123379230499268, "learning_rate": 3.4413572609427235e-07, "loss": 0.08430693149566651, "step": 34880 }, { "epoch": 0.9190557880560942, "grad_norm": 4.412215709686279, "learning_rate": 3.4192038053267095e-07, "loss": 0.08261313438415527, "step": 34890 }, { "epoch": 0.919319203300593, "grad_norm": 3.0322530269622803, "learning_rate": 3.3971206453810133e-07, "loss": 0.07788345813751221, "step": 34900 }, { "epoch": 0.9195826185450917, "grad_norm": 3.153395414352417, "learning_rate": 3.3751077971787115e-07, "loss": 0.07962493896484375, "step": 34910 }, { "epoch": 0.9198460337895905, "grad_norm": 4.094183444976807, "learning_rate": 3.353165276741699e-07, "loss": 0.08258055448532105, "step": 34920 }, { "epoch": 0.9201094490340892, "grad_norm": 4.154822826385498, "learning_rate": 3.331293100040711e-07, "loss": 0.0922630250453949, "step": 34930 }, { "epoch": 0.920372864278588, "grad_norm": 3.7039809226989746, "learning_rate": 3.3094912829952477e-07, "loss": 0.09151618480682373, "step": 34940 }, { "epoch": 0.9206362795230867, "grad_norm": 3.7935538291931152, "learning_rate": 3.2877598414736033e-07, "loss": 0.07825679183006287, "step": 34950 }, { "epoch": 0.9208996947675855, "grad_norm": 3.2053699493408203, "learning_rate": 3.266098791292871e-07, "loss": 0.09213950634002685, "step": 34960 }, { "epoch": 0.9211631100120842, "grad_norm": 2.5485000610351562, "learning_rate": 3.244508148218895e-07, "loss": 0.07155948877334595, "step": 34970 }, { "epoch": 0.9214265252565829, "grad_norm": 2.944997787475586, "learning_rate": 3.22298792796627e-07, "loss": 0.07939380407333374, "step": 34980 }, { "epoch": 0.9216899405010817, "grad_norm": 3.0600476264953613, "learning_rate": 3.2015381461983577e-07, "loss": 0.0914269208908081, "step": 34990 }, { "epoch": 0.9219533557455803, "grad_norm": 3.0276448726654053, "learning_rate": 3.1801588185272016e-07, "loss": 0.0782006859779358, "step": 35000 }, { "epoch": 0.9222167709900791, "grad_norm": 2.520739793777466, "learning_rate": 3.1588499605136435e-07, "loss": 0.08572744131088257, "step": 35010 }, { "epoch": 0.9224801862345778, "grad_norm": 3.3987812995910645, "learning_rate": 3.1376115876671553e-07, "loss": 0.09401571750640869, "step": 35020 }, { "epoch": 0.9227436014790766, "grad_norm": 2.574411392211914, "learning_rate": 3.1164437154459494e-07, "loss": 0.08154506683349609, "step": 35030 }, { "epoch": 0.9230070167235753, "grad_norm": 4.683283805847168, "learning_rate": 3.0953463592569343e-07, "loss": 0.08338724374771118, "step": 35040 }, { "epoch": 0.9232704319680741, "grad_norm": 3.203033924102783, "learning_rate": 3.074319534455639e-07, "loss": 0.09532843828201294, "step": 35050 }, { "epoch": 0.9235338472125728, "grad_norm": 2.8619184494018555, "learning_rate": 3.053363256346331e-07, "loss": 0.08692805767059326, "step": 35060 }, { "epoch": 0.9237972624570715, "grad_norm": 5.181122779846191, "learning_rate": 3.0324775401818884e-07, "loss": 0.08019742369651794, "step": 35070 }, { "epoch": 0.9240606777015703, "grad_norm": 3.820136070251465, "learning_rate": 3.0116624011637953e-07, "loss": 0.09378492832183838, "step": 35080 }, { "epoch": 0.924324092946069, "grad_norm": 2.54590106010437, "learning_rate": 2.9909178544422656e-07, "loss": 0.08156596422195435, "step": 35090 }, { "epoch": 0.9245875081905678, "grad_norm": 3.163952589035034, "learning_rate": 2.9702439151160334e-07, "loss": 0.08385679125785828, "step": 35100 }, { "epoch": 0.9248509234350665, "grad_norm": 2.8266377449035645, "learning_rate": 2.949640598232517e-07, "loss": 0.08469934463500976, "step": 35110 }, { "epoch": 0.9251143386795653, "grad_norm": 4.161337375640869, "learning_rate": 2.9291079187876656e-07, "loss": 0.0908672034740448, "step": 35120 }, { "epoch": 0.925377753924064, "grad_norm": 3.0389933586120605, "learning_rate": 2.9086458917260585e-07, "loss": 0.08378652930259704, "step": 35130 }, { "epoch": 0.9256411691685628, "grad_norm": 2.8954708576202393, "learning_rate": 2.8882545319408706e-07, "loss": 0.07764111757278443, "step": 35140 }, { "epoch": 0.9259045844130614, "grad_norm": 3.0216848850250244, "learning_rate": 2.8679338542737745e-07, "loss": 0.08855594992637635, "step": 35150 }, { "epoch": 0.9261679996575601, "grad_norm": 4.876061916351318, "learning_rate": 2.8476838735150613e-07, "loss": 0.0804531455039978, "step": 35160 }, { "epoch": 0.9264314149020589, "grad_norm": 2.770002841949463, "learning_rate": 2.8275046044035304e-07, "loss": 0.07646973133087158, "step": 35170 }, { "epoch": 0.9266948301465576, "grad_norm": 2.7345683574676514, "learning_rate": 2.8073960616265215e-07, "loss": 0.09046350717544556, "step": 35180 }, { "epoch": 0.9269582453910564, "grad_norm": 3.8922088146209717, "learning_rate": 2.7873582598199056e-07, "loss": 0.08700125217437744, "step": 35190 }, { "epoch": 0.9272216606355551, "grad_norm": 2.4086813926696777, "learning_rate": 2.7673912135680714e-07, "loss": 0.06409684419631959, "step": 35200 }, { "epoch": 0.9274850758800539, "grad_norm": 3.649402379989624, "learning_rate": 2.7474949374038717e-07, "loss": 0.07815578579902649, "step": 35210 }, { "epoch": 0.9277484911245526, "grad_norm": 2.1178648471832275, "learning_rate": 2.7276694458086894e-07, "loss": 0.08324924111366272, "step": 35220 }, { "epoch": 0.9280119063690514, "grad_norm": 3.6975040435791016, "learning_rate": 2.7079147532123704e-07, "loss": 0.09063262939453125, "step": 35230 }, { "epoch": 0.9282753216135501, "grad_norm": 2.734313488006592, "learning_rate": 2.688230873993225e-07, "loss": 0.06810388565063477, "step": 35240 }, { "epoch": 0.9285387368580488, "grad_norm": 3.7272145748138428, "learning_rate": 2.6686178224780367e-07, "loss": 0.0840737521648407, "step": 35250 }, { "epoch": 0.9288021521025476, "grad_norm": 2.961775541305542, "learning_rate": 2.6490756129420313e-07, "loss": 0.07827574014663696, "step": 35260 }, { "epoch": 0.9290655673470463, "grad_norm": 3.273270845413208, "learning_rate": 2.6296042596088643e-07, "loss": 0.08296308517456055, "step": 35270 }, { "epoch": 0.9293289825915451, "grad_norm": 3.542968988418579, "learning_rate": 2.610203776650655e-07, "loss": 0.08110016584396362, "step": 35280 }, { "epoch": 0.9295923978360437, "grad_norm": 3.35563063621521, "learning_rate": 2.5908741781878744e-07, "loss": 0.07821820974349976, "step": 35290 }, { "epoch": 0.9298558130805425, "grad_norm": 2.335418224334717, "learning_rate": 2.5716154782894573e-07, "loss": 0.07944345474243164, "step": 35300 }, { "epoch": 0.9301192283250412, "grad_norm": 4.577977180480957, "learning_rate": 2.552427690972703e-07, "loss": 0.08066850304603576, "step": 35310 }, { "epoch": 0.9303826435695399, "grad_norm": 3.0748043060302734, "learning_rate": 2.5333108302033394e-07, "loss": 0.07338759303092957, "step": 35320 }, { "epoch": 0.9306460588140387, "grad_norm": 3.3539915084838867, "learning_rate": 2.514264909895414e-07, "loss": 0.08345267772674561, "step": 35330 }, { "epoch": 0.9309094740585374, "grad_norm": 2.5673065185546875, "learning_rate": 2.495289943911361e-07, "loss": 0.08423304557800293, "step": 35340 }, { "epoch": 0.9311728893030362, "grad_norm": 3.4291775226593018, "learning_rate": 2.476385946062021e-07, "loss": 0.0850480079650879, "step": 35350 }, { "epoch": 0.9314363045475349, "grad_norm": 3.263091802597046, "learning_rate": 2.4575529301065104e-07, "loss": 0.08066965341567993, "step": 35360 }, { "epoch": 0.9316997197920337, "grad_norm": 3.420093536376953, "learning_rate": 2.43879090975232e-07, "loss": 0.08985942602157593, "step": 35370 }, { "epoch": 0.9319631350365324, "grad_norm": 4.725570201873779, "learning_rate": 2.4200998986552484e-07, "loss": 0.08463276624679565, "step": 35380 }, { "epoch": 0.9322265502810312, "grad_norm": 4.0341339111328125, "learning_rate": 2.401479910419413e-07, "loss": 0.08458285331726074, "step": 35390 }, { "epoch": 0.9324899655255299, "grad_norm": 3.1198737621307373, "learning_rate": 2.3829309585972738e-07, "loss": 0.09211444854736328, "step": 35400 }, { "epoch": 0.9327533807700286, "grad_norm": 3.4856975078582764, "learning_rate": 2.364453056689553e-07, "loss": 0.0816168487071991, "step": 35410 }, { "epoch": 0.9330167960145274, "grad_norm": 3.6900625228881836, "learning_rate": 2.3460462181452258e-07, "loss": 0.08490262031555176, "step": 35420 }, { "epoch": 0.933280211259026, "grad_norm": 2.5466692447662354, "learning_rate": 2.3277104563616205e-07, "loss": 0.08269588351249695, "step": 35430 }, { "epoch": 0.9335436265035248, "grad_norm": 3.3097083568573, "learning_rate": 2.3094457846842943e-07, "loss": 0.08120981454849244, "step": 35440 }, { "epoch": 0.9338070417480235, "grad_norm": 3.898371458053589, "learning_rate": 2.291252216407047e-07, "loss": 0.07624321579933166, "step": 35450 }, { "epoch": 0.9340704569925223, "grad_norm": 2.8873770236968994, "learning_rate": 2.273129764771953e-07, "loss": 0.07686638832092285, "step": 35460 }, { "epoch": 0.934333872237021, "grad_norm": 3.8425076007843018, "learning_rate": 2.2550784429693052e-07, "loss": 0.08361493349075318, "step": 35470 }, { "epoch": 0.9345972874815198, "grad_norm": 2.98122239112854, "learning_rate": 2.2370982641376494e-07, "loss": 0.06990942358970642, "step": 35480 }, { "epoch": 0.9348607027260185, "grad_norm": 4.177785396575928, "learning_rate": 2.2191892413637173e-07, "loss": 0.07994742393493652, "step": 35490 }, { "epoch": 0.9351241179705172, "grad_norm": 3.5331342220306396, "learning_rate": 2.2013513876824933e-07, "loss": 0.07860888242721557, "step": 35500 }, { "epoch": 0.935387533215016, "grad_norm": 2.0477001667022705, "learning_rate": 2.183584716077114e-07, "loss": 0.07816175222396851, "step": 35510 }, { "epoch": 0.9356509484595147, "grad_norm": 2.6872293949127197, "learning_rate": 2.1658892394789244e-07, "loss": 0.07564923763275147, "step": 35520 }, { "epoch": 0.9359143637040135, "grad_norm": 4.008882522583008, "learning_rate": 2.1482649707674775e-07, "loss": 0.07334855794906617, "step": 35530 }, { "epoch": 0.9361777789485122, "grad_norm": 2.709103584289551, "learning_rate": 2.13071192277049e-07, "loss": 0.07849140167236328, "step": 35540 }, { "epoch": 0.936441194193011, "grad_norm": 2.8963963985443115, "learning_rate": 2.1132301082637752e-07, "loss": 0.08238277435302735, "step": 35550 }, { "epoch": 0.9367046094375097, "grad_norm": 3.7175703048706055, "learning_rate": 2.0958195399713888e-07, "loss": 0.08582147359848022, "step": 35560 }, { "epoch": 0.9369680246820085, "grad_norm": 3.530756711959839, "learning_rate": 2.0784802305654938e-07, "loss": 0.07623060941696166, "step": 35570 }, { "epoch": 0.9372314399265071, "grad_norm": 4.509089469909668, "learning_rate": 2.061212192666373e-07, "loss": 0.0841529369354248, "step": 35580 }, { "epoch": 0.9374948551710058, "grad_norm": 2.459733724594116, "learning_rate": 2.0440154388424616e-07, "loss": 0.08146090507507324, "step": 35590 }, { "epoch": 0.9377582704155046, "grad_norm": 2.207768201828003, "learning_rate": 2.0268899816102916e-07, "loss": 0.07200020551681519, "step": 35600 }, { "epoch": 0.9380216856600033, "grad_norm": 3.4613680839538574, "learning_rate": 2.0098358334345147e-07, "loss": 0.08028180599212646, "step": 35610 }, { "epoch": 0.9382851009045021, "grad_norm": 3.5348610877990723, "learning_rate": 1.9928530067278684e-07, "loss": 0.07898968458175659, "step": 35620 }, { "epoch": 0.9385485161490008, "grad_norm": 2.946798086166382, "learning_rate": 1.9759415138511984e-07, "loss": 0.07843701243400573, "step": 35630 }, { "epoch": 0.9388119313934996, "grad_norm": 3.2260661125183105, "learning_rate": 1.959101367113414e-07, "loss": 0.08270301818847656, "step": 35640 }, { "epoch": 0.9390753466379983, "grad_norm": 2.3831088542938232, "learning_rate": 1.9423325787714774e-07, "loss": 0.07937952876091003, "step": 35650 }, { "epoch": 0.9393387618824971, "grad_norm": 2.5834858417510986, "learning_rate": 1.9256351610304814e-07, "loss": 0.08144803047180176, "step": 35660 }, { "epoch": 0.9396021771269958, "grad_norm": 3.5276248455047607, "learning_rate": 1.909009126043504e-07, "loss": 0.07851221561431884, "step": 35670 }, { "epoch": 0.9398655923714945, "grad_norm": 1.9899604320526123, "learning_rate": 1.8924544859116656e-07, "loss": 0.08498393297195435, "step": 35680 }, { "epoch": 0.9401290076159933, "grad_norm": 3.3724558353424072, "learning_rate": 1.8759712526841943e-07, "loss": 0.08519054651260376, "step": 35690 }, { "epoch": 0.940392422860492, "grad_norm": 3.1534764766693115, "learning_rate": 1.8595594383582605e-07, "loss": 0.08579696416854858, "step": 35700 }, { "epoch": 0.9406558381049908, "grad_norm": 2.841399908065796, "learning_rate": 1.8432190548791197e-07, "loss": 0.08654026985168457, "step": 35710 }, { "epoch": 0.9409192533494894, "grad_norm": 2.5337512493133545, "learning_rate": 1.8269501141399915e-07, "loss": 0.0816759467124939, "step": 35720 }, { "epoch": 0.9411826685939882, "grad_norm": 3.031423330307007, "learning_rate": 1.8107526279821041e-07, "loss": 0.08929321765899659, "step": 35730 }, { "epoch": 0.9414460838384869, "grad_norm": 3.3655343055725098, "learning_rate": 1.794626608194705e-07, "loss": 0.0924578845500946, "step": 35740 }, { "epoch": 0.9417094990829857, "grad_norm": 3.031622886657715, "learning_rate": 1.7785720665150054e-07, "loss": 0.08676246404647828, "step": 35750 }, { "epoch": 0.9419729143274844, "grad_norm": 4.014092922210693, "learning_rate": 1.7625890146281798e-07, "loss": 0.07827741503715516, "step": 35760 }, { "epoch": 0.9422363295719831, "grad_norm": 1.876304268836975, "learning_rate": 1.7466774641674012e-07, "loss": 0.06667996644973755, "step": 35770 }, { "epoch": 0.9424997448164819, "grad_norm": 3.044618606567383, "learning_rate": 1.7308374267137606e-07, "loss": 0.0807801365852356, "step": 35780 }, { "epoch": 0.9427631600609806, "grad_norm": 2.9418880939483643, "learning_rate": 1.715068913796336e-07, "loss": 0.08017163276672364, "step": 35790 }, { "epoch": 0.9430265753054794, "grad_norm": 3.3804163932800293, "learning_rate": 1.6993719368921468e-07, "loss": 0.0866655170917511, "step": 35800 }, { "epoch": 0.9432899905499781, "grad_norm": 4.000646114349365, "learning_rate": 1.683746507426087e-07, "loss": 0.0695287823677063, "step": 35810 }, { "epoch": 0.9435534057944769, "grad_norm": 3.232266426086426, "learning_rate": 1.66819263677106e-07, "loss": 0.08777822256088257, "step": 35820 }, { "epoch": 0.9438168210389756, "grad_norm": 3.764951229095459, "learning_rate": 1.6527103362478202e-07, "loss": 0.08376954793930054, "step": 35830 }, { "epoch": 0.9440802362834743, "grad_norm": 3.149622678756714, "learning_rate": 1.6372996171250544e-07, "loss": 0.0951123595237732, "step": 35840 }, { "epoch": 0.9443436515279731, "grad_norm": 3.5430500507354736, "learning_rate": 1.6219604906193677e-07, "loss": 0.07717002630233764, "step": 35850 }, { "epoch": 0.9446070667724717, "grad_norm": 3.3928873538970947, "learning_rate": 1.606692967895207e-07, "loss": 0.08329054117202758, "step": 35860 }, { "epoch": 0.9448704820169705, "grad_norm": 3.9568843841552734, "learning_rate": 1.5914970600649615e-07, "loss": 0.09206262826919556, "step": 35870 }, { "epoch": 0.9451338972614692, "grad_norm": 2.5194854736328125, "learning_rate": 1.5763727781888726e-07, "loss": 0.06686234474182129, "step": 35880 }, { "epoch": 0.945397312505968, "grad_norm": 2.8747916221618652, "learning_rate": 1.5613201332750127e-07, "loss": 0.08410470485687256, "step": 35890 }, { "epoch": 0.9456607277504667, "grad_norm": 3.2890589237213135, "learning_rate": 1.5463391362793735e-07, "loss": 0.07727016806602478, "step": 35900 }, { "epoch": 0.9459241429949655, "grad_norm": 3.111259698867798, "learning_rate": 1.5314297981057658e-07, "loss": 0.08279317021369934, "step": 35910 }, { "epoch": 0.9461875582394642, "grad_norm": 1.746019959449768, "learning_rate": 1.5165921296058427e-07, "loss": 0.08291352391242982, "step": 35920 }, { "epoch": 0.9464509734839629, "grad_norm": 2.475541353225708, "learning_rate": 1.501826141579099e-07, "loss": 0.0736750066280365, "step": 35930 }, { "epoch": 0.9467143887284617, "grad_norm": 2.893404960632324, "learning_rate": 1.48713184477286e-07, "loss": 0.08124839067459107, "step": 35940 }, { "epoch": 0.9469778039729604, "grad_norm": 4.350130081176758, "learning_rate": 1.4725092498822814e-07, "loss": 0.08544806241989136, "step": 35950 }, { "epoch": 0.9472412192174592, "grad_norm": 2.836643695831299, "learning_rate": 1.457958367550305e-07, "loss": 0.08648765087127686, "step": 35960 }, { "epoch": 0.9475046344619579, "grad_norm": 2.2592244148254395, "learning_rate": 1.443479208367693e-07, "loss": 0.07370012998580933, "step": 35970 }, { "epoch": 0.9477680497064567, "grad_norm": 3.009108781814575, "learning_rate": 1.4290717828730037e-07, "loss": 0.07679461240768433, "step": 35980 }, { "epoch": 0.9480314649509554, "grad_norm": 3.7179698944091797, "learning_rate": 1.4147361015525717e-07, "loss": 0.09428122639656067, "step": 35990 }, { "epoch": 0.9482948801954542, "grad_norm": 3.3837287425994873, "learning_rate": 1.4004721748405392e-07, "loss": 0.09692404866218567, "step": 36000 }, { "epoch": 0.9485582954399528, "grad_norm": 3.829810619354248, "learning_rate": 1.3862800131188237e-07, "loss": 0.08357189893722534, "step": 36010 }, { "epoch": 0.9488217106844515, "grad_norm": 3.384469509124756, "learning_rate": 1.37215962671704e-07, "loss": 0.08102420568466187, "step": 36020 }, { "epoch": 0.9490851259289503, "grad_norm": 4.51561164855957, "learning_rate": 1.3581110259126674e-07, "loss": 0.08111618161201477, "step": 36030 }, { "epoch": 0.949348541173449, "grad_norm": 3.5002734661102295, "learning_rate": 1.3441342209308595e-07, "loss": 0.08235270977020263, "step": 36040 }, { "epoch": 0.9496119564179478, "grad_norm": 3.6034958362579346, "learning_rate": 1.3302292219445457e-07, "loss": 0.08336960077285767, "step": 36050 }, { "epoch": 0.9498753716624465, "grad_norm": 3.004848003387451, "learning_rate": 1.3163960390743746e-07, "loss": 0.07823350429534912, "step": 36060 }, { "epoch": 0.9501387869069453, "grad_norm": 4.166321754455566, "learning_rate": 1.302634682388726e-07, "loss": 0.07965397834777832, "step": 36070 }, { "epoch": 0.950402202151444, "grad_norm": 2.8222098350524902, "learning_rate": 1.2889451619037318e-07, "loss": 0.07635408639907837, "step": 36080 }, { "epoch": 0.9506656173959428, "grad_norm": 3.0926573276519775, "learning_rate": 1.2753274875832112e-07, "loss": 0.08114373683929443, "step": 36090 }, { "epoch": 0.9509290326404415, "grad_norm": 2.971815824508667, "learning_rate": 1.2617816693386907e-07, "loss": 0.07958526611328125, "step": 36100 }, { "epoch": 0.9511924478849402, "grad_norm": 3.242311477661133, "learning_rate": 1.2483077170293954e-07, "loss": 0.08798760771751404, "step": 36110 }, { "epoch": 0.951455863129439, "grad_norm": 3.8018639087677, "learning_rate": 1.2349056404622472e-07, "loss": 0.0701157808303833, "step": 36120 }, { "epoch": 0.9517192783739377, "grad_norm": 2.4764046669006348, "learning_rate": 1.2215754493918763e-07, "loss": 0.0841257095336914, "step": 36130 }, { "epoch": 0.9519826936184365, "grad_norm": 3.724006414413452, "learning_rate": 1.208317153520555e-07, "loss": 0.09398032426834106, "step": 36140 }, { "epoch": 0.9522461088629351, "grad_norm": 3.2525408267974854, "learning_rate": 1.195130762498231e-07, "loss": 0.07752150893211365, "step": 36150 }, { "epoch": 0.952509524107434, "grad_norm": 3.4881813526153564, "learning_rate": 1.1820162859225493e-07, "loss": 0.07277869582176208, "step": 36160 }, { "epoch": 0.9527729393519326, "grad_norm": 2.5891242027282715, "learning_rate": 1.1689737333387852e-07, "loss": 0.06967814564704895, "step": 36170 }, { "epoch": 0.9530363545964314, "grad_norm": 3.1579718589782715, "learning_rate": 1.1560031142398787e-07, "loss": 0.07979346513748169, "step": 36180 }, { "epoch": 0.9532997698409301, "grad_norm": 3.3242859840393066, "learning_rate": 1.1431044380663891e-07, "loss": 0.0965050220489502, "step": 36190 }, { "epoch": 0.9535631850854288, "grad_norm": 2.7520744800567627, "learning_rate": 1.1302777142065402e-07, "loss": 0.08319125175476075, "step": 36200 }, { "epoch": 0.9538266003299276, "grad_norm": 2.5709004402160645, "learning_rate": 1.117522951996175e-07, "loss": 0.07848449945449829, "step": 36210 }, { "epoch": 0.9540900155744263, "grad_norm": 3.2640438079833984, "learning_rate": 1.1048401607187565e-07, "loss": 0.08073311448097228, "step": 36220 }, { "epoch": 0.9543534308189251, "grad_norm": 2.7794549465179443, "learning_rate": 1.0922293496053671e-07, "loss": 0.09120905995368958, "step": 36230 }, { "epoch": 0.9546168460634238, "grad_norm": 3.02392840385437, "learning_rate": 1.0796905278347092e-07, "loss": 0.08796538710594178, "step": 36240 }, { "epoch": 0.9548802613079226, "grad_norm": 2.9759809970855713, "learning_rate": 1.0672237045330491e-07, "loss": 0.07931755781173706, "step": 36250 }, { "epoch": 0.9551436765524213, "grad_norm": 3.700474262237549, "learning_rate": 1.0548288887743174e-07, "loss": 0.07721196413040161, "step": 36260 }, { "epoch": 0.9554070917969201, "grad_norm": 4.357350826263428, "learning_rate": 1.0425060895799865e-07, "loss": 0.0697629451751709, "step": 36270 }, { "epoch": 0.9556705070414188, "grad_norm": 3.284801721572876, "learning_rate": 1.0302553159190932e-07, "loss": 0.08495320677757263, "step": 36280 }, { "epoch": 0.9559339222859174, "grad_norm": 2.5708534717559814, "learning_rate": 1.0180765767083045e-07, "loss": 0.07388817071914673, "step": 36290 }, { "epoch": 0.9561973375304162, "grad_norm": 2.5194027423858643, "learning_rate": 1.00596988081183e-07, "loss": 0.07813369035720825, "step": 36300 }, { "epoch": 0.9564607527749149, "grad_norm": 2.698702812194824, "learning_rate": 9.939352370414435e-08, "loss": 0.07642086744308471, "step": 36310 }, { "epoch": 0.9567241680194137, "grad_norm": 3.3082234859466553, "learning_rate": 9.819726541564711e-08, "loss": 0.08367544412612915, "step": 36320 }, { "epoch": 0.9569875832639124, "grad_norm": 2.9854254722595215, "learning_rate": 9.700821408638039e-08, "loss": 0.0874977469444275, "step": 36330 }, { "epoch": 0.9572509985084112, "grad_norm": 2.0901169776916504, "learning_rate": 9.582637058178635e-08, "loss": 0.07700308561325073, "step": 36340 }, { "epoch": 0.9575144137529099, "grad_norm": 3.599608898162842, "learning_rate": 9.465173576206355e-08, "loss": 0.07082630395889282, "step": 36350 }, { "epoch": 0.9577778289974086, "grad_norm": 2.889281988143921, "learning_rate": 9.348431048215923e-08, "loss": 0.09003482460975647, "step": 36360 }, { "epoch": 0.9580412442419074, "grad_norm": 3.6109964847564697, "learning_rate": 9.232409559177701e-08, "loss": 0.07910590171813965, "step": 36370 }, { "epoch": 0.9583046594864061, "grad_norm": 3.9687063694000244, "learning_rate": 9.117109193537254e-08, "loss": 0.09916083812713623, "step": 36380 }, { "epoch": 0.9585680747309049, "grad_norm": 3.17942476272583, "learning_rate": 9.002530035215007e-08, "loss": 0.0898890495300293, "step": 36390 }, { "epoch": 0.9588314899754036, "grad_norm": 3.3118698596954346, "learning_rate": 8.888672167606693e-08, "loss": 0.0823095202445984, "step": 36400 }, { "epoch": 0.9590949052199024, "grad_norm": 3.6837875843048096, "learning_rate": 8.775535673582913e-08, "loss": 0.07875177264213562, "step": 36410 }, { "epoch": 0.9593583204644011, "grad_norm": 3.2860729694366455, "learning_rate": 8.663120635489241e-08, "loss": 0.07415508031845093, "step": 36420 }, { "epoch": 0.9596217357088999, "grad_norm": 2.917374610900879, "learning_rate": 8.55142713514645e-08, "loss": 0.08280390501022339, "step": 36430 }, { "epoch": 0.9598851509533985, "grad_norm": 3.889368772506714, "learning_rate": 8.440455253849622e-08, "loss": 0.08281577825546264, "step": 36440 }, { "epoch": 0.9601485661978972, "grad_norm": 2.2072746753692627, "learning_rate": 8.330205072369035e-08, "loss": 0.08309106230735779, "step": 36450 }, { "epoch": 0.960411981442396, "grad_norm": 3.4981441497802734, "learning_rate": 8.22067667094939e-08, "loss": 0.0871547818183899, "step": 36460 }, { "epoch": 0.9606753966868947, "grad_norm": 2.981234550476074, "learning_rate": 8.11187012931025e-08, "loss": 0.0827444076538086, "step": 36470 }, { "epoch": 0.9609388119313935, "grad_norm": 3.209834098815918, "learning_rate": 8.003785526645824e-08, "loss": 0.0873442530632019, "step": 36480 }, { "epoch": 0.9612022271758922, "grad_norm": 4.154417514801025, "learning_rate": 7.896422941624626e-08, "loss": 0.08467501401901245, "step": 36490 }, { "epoch": 0.961465642420391, "grad_norm": 3.7958478927612305, "learning_rate": 7.789782452389705e-08, "loss": 0.08534530401229859, "step": 36500 }, { "epoch": 0.9617290576648897, "grad_norm": 3.4120121002197266, "learning_rate": 7.683864136558749e-08, "loss": 0.07912418246269226, "step": 36510 }, { "epoch": 0.9619924729093885, "grad_norm": 4.081188201904297, "learning_rate": 7.578668071223649e-08, "loss": 0.07732291221618652, "step": 36520 }, { "epoch": 0.9622558881538872, "grad_norm": 3.326331615447998, "learning_rate": 7.474194332950602e-08, "loss": 0.08105028867721557, "step": 36530 }, { "epoch": 0.9625193033983859, "grad_norm": 3.964747667312622, "learning_rate": 7.370442997780114e-08, "loss": 0.07582420110702515, "step": 36540 }, { "epoch": 0.9627827186428847, "grad_norm": 2.787489414215088, "learning_rate": 7.267414141227002e-08, "loss": 0.07220847606658935, "step": 36550 }, { "epoch": 0.9630461338873834, "grad_norm": 3.1846039295196533, "learning_rate": 7.165107838279838e-08, "loss": 0.08357859849929809, "step": 36560 }, { "epoch": 0.9633095491318822, "grad_norm": 2.2518930435180664, "learning_rate": 7.063524163401836e-08, "loss": 0.08369896411895753, "step": 36570 }, { "epoch": 0.9635729643763808, "grad_norm": 3.401231050491333, "learning_rate": 6.962663190529851e-08, "loss": 0.07641544342041015, "step": 36580 }, { "epoch": 0.9638363796208796, "grad_norm": 3.2759571075439453, "learning_rate": 6.862524993074825e-08, "loss": 0.07068753242492676, "step": 36590 }, { "epoch": 0.9640997948653783, "grad_norm": 3.354495048522949, "learning_rate": 6.763109643921683e-08, "loss": 0.07256864309310913, "step": 36600 }, { "epoch": 0.9643632101098771, "grad_norm": 3.5453596115112305, "learning_rate": 6.664417215429319e-08, "loss": 0.09654340744018555, "step": 36610 }, { "epoch": 0.9646266253543758, "grad_norm": 2.716707229614258, "learning_rate": 6.566447779430163e-08, "loss": 0.08553112149238587, "step": 36620 }, { "epoch": 0.9648900405988745, "grad_norm": 2.9721720218658447, "learning_rate": 6.469201407230729e-08, "loss": 0.07585623264312744, "step": 36630 }, { "epoch": 0.9651534558433733, "grad_norm": 3.808044910430908, "learning_rate": 6.372678169611068e-08, "loss": 0.08250983953475952, "step": 36640 }, { "epoch": 0.965416871087872, "grad_norm": 3.4781010150909424, "learning_rate": 6.276878136824982e-08, "loss": 0.07726793885231018, "step": 36650 }, { "epoch": 0.9656802863323708, "grad_norm": 3.040804862976074, "learning_rate": 6.181801378599805e-08, "loss": 0.07886203527450561, "step": 36660 }, { "epoch": 0.9659437015768695, "grad_norm": 3.492788553237915, "learning_rate": 6.087447964136517e-08, "loss": 0.07375862002372742, "step": 36670 }, { "epoch": 0.9662071168213683, "grad_norm": 4.2508721351623535, "learning_rate": 5.993817962109627e-08, "loss": 0.09451094865798951, "step": 36680 }, { "epoch": 0.966470532065867, "grad_norm": 3.2241013050079346, "learning_rate": 5.9009114406671787e-08, "loss": 0.07990506887435914, "step": 36690 }, { "epoch": 0.9667339473103658, "grad_norm": 3.807445526123047, "learning_rate": 5.8087284674304136e-08, "loss": 0.08487989902496337, "step": 36700 }, { "epoch": 0.9669973625548645, "grad_norm": 3.739630937576294, "learning_rate": 5.7172691094942165e-08, "loss": 0.07583245635032654, "step": 36710 }, { "epoch": 0.9672607777993631, "grad_norm": 3.5349555015563965, "learning_rate": 5.626533433426451e-08, "loss": 0.08879632949829101, "step": 36720 }, { "epoch": 0.967524193043862, "grad_norm": 3.3093833923339844, "learning_rate": 5.5365215052686216e-08, "loss": 0.08618174195289612, "step": 36730 }, { "epoch": 0.9677876082883606, "grad_norm": 3.3158507347106934, "learning_rate": 5.4472333905352115e-08, "loss": 0.07573038339614868, "step": 36740 }, { "epoch": 0.9680510235328594, "grad_norm": 2.6586568355560303, "learning_rate": 5.3586691542140135e-08, "loss": 0.06304721832275391, "step": 36750 }, { "epoch": 0.9683144387773581, "grad_norm": 2.893648862838745, "learning_rate": 5.2708288607659084e-08, "loss": 0.07695536613464356, "step": 36760 }, { "epoch": 0.9685778540218569, "grad_norm": 3.2751572132110596, "learning_rate": 5.183712574124755e-08, "loss": 0.07732014656066895, "step": 36770 }, { "epoch": 0.9688412692663556, "grad_norm": 2.9944405555725098, "learning_rate": 5.0973203576976104e-08, "loss": 0.08034420013427734, "step": 36780 }, { "epoch": 0.9691046845108544, "grad_norm": 3.288200616836548, "learning_rate": 5.011652274364509e-08, "loss": 0.07620214819908142, "step": 36790 }, { "epoch": 0.9693680997553531, "grad_norm": 3.50715970993042, "learning_rate": 4.926708386478352e-08, "loss": 0.0757490336894989, "step": 36800 }, { "epoch": 0.9696315149998518, "grad_norm": 3.6662344932556152, "learning_rate": 4.8424887558649045e-08, "loss": 0.08044755458831787, "step": 36810 }, { "epoch": 0.9698949302443506, "grad_norm": 2.475908041000366, "learning_rate": 4.7589934438228017e-08, "loss": 0.0815010905265808, "step": 36820 }, { "epoch": 0.9701583454888493, "grad_norm": 2.9961423873901367, "learning_rate": 4.676222511123762e-08, "loss": 0.08875769376754761, "step": 36830 }, { "epoch": 0.9704217607333481, "grad_norm": 5.053282260894775, "learning_rate": 4.59417601801182e-08, "loss": 0.08446967601776123, "step": 36840 }, { "epoch": 0.9706851759778468, "grad_norm": 3.151216983795166, "learning_rate": 4.512854024203983e-08, "loss": 0.10245009660720825, "step": 36850 }, { "epoch": 0.9709485912223456, "grad_norm": 2.5282604694366455, "learning_rate": 4.432256588890016e-08, "loss": 0.07618271112442017, "step": 36860 }, { "epoch": 0.9712120064668442, "grad_norm": 2.8965675830841064, "learning_rate": 4.352383770732216e-08, "loss": 0.07567130327224732, "step": 36870 }, { "epoch": 0.9714754217113429, "grad_norm": 3.8412792682647705, "learning_rate": 4.273235627865302e-08, "loss": 0.07046204805374146, "step": 36880 }, { "epoch": 0.9717388369558417, "grad_norm": 2.9773266315460205, "learning_rate": 4.19481221789686e-08, "loss": 0.08639901876449585, "step": 36890 }, { "epoch": 0.9720022522003404, "grad_norm": 3.703511953353882, "learning_rate": 4.117113597906675e-08, "loss": 0.0780713438987732, "step": 36900 }, { "epoch": 0.9722656674448392, "grad_norm": 3.1514089107513428, "learning_rate": 4.040139824447398e-08, "loss": 0.07442628741264343, "step": 36910 }, { "epoch": 0.9725290826893379, "grad_norm": 3.5664398670196533, "learning_rate": 3.963890953543659e-08, "loss": 0.07638385891914368, "step": 36920 }, { "epoch": 0.9727924979338367, "grad_norm": 3.3384945392608643, "learning_rate": 3.88836704069262e-08, "loss": 0.08037492632865906, "step": 36930 }, { "epoch": 0.9730559131783354, "grad_norm": 3.0511796474456787, "learning_rate": 3.813568140864088e-08, "loss": 0.08619625568389892, "step": 36940 }, { "epoch": 0.9733193284228342, "grad_norm": 2.968306064605713, "learning_rate": 3.739494308499736e-08, "loss": 0.08999336361885071, "step": 36950 }, { "epoch": 0.9735827436673329, "grad_norm": 2.6208736896514893, "learning_rate": 3.66614559751377e-08, "loss": 0.08314566612243653, "step": 36960 }, { "epoch": 0.9738461589118316, "grad_norm": 3.1985766887664795, "learning_rate": 3.593522061292598e-08, "loss": 0.05562950968742371, "step": 36970 }, { "epoch": 0.9741095741563304, "grad_norm": 4.636105060577393, "learning_rate": 3.521623752694603e-08, "loss": 0.0839102566242218, "step": 36980 }, { "epoch": 0.9743729894008291, "grad_norm": 2.8375236988067627, "learning_rate": 3.4504507240507027e-08, "loss": 0.08293615579605103, "step": 36990 }, { "epoch": 0.9746364046453279, "grad_norm": 2.756716251373291, "learning_rate": 3.380003027163459e-08, "loss": 0.07357977628707886, "step": 37000 }, { "epoch": 0.9748998198898265, "grad_norm": 3.684147834777832, "learning_rate": 3.310280713307856e-08, "loss": 0.07824200391769409, "step": 37010 }, { "epoch": 0.9751632351343253, "grad_norm": 3.5350635051727295, "learning_rate": 3.241283833230746e-08, "loss": 0.07372047901153564, "step": 37020 }, { "epoch": 0.975426650378824, "grad_norm": 4.0182695388793945, "learning_rate": 3.1730124371511794e-08, "loss": 0.07222009897232055, "step": 37030 }, { "epoch": 0.9756900656233228, "grad_norm": 3.3388671875, "learning_rate": 3.1054665747599635e-08, "loss": 0.07237493991851807, "step": 37040 }, { "epoch": 0.9759534808678215, "grad_norm": 3.4152801036834717, "learning_rate": 3.038646295219771e-08, "loss": 0.09342195391654969, "step": 37050 }, { "epoch": 0.9762168961123202, "grad_norm": 3.073044538497925, "learning_rate": 2.972551647165367e-08, "loss": 0.09007441997528076, "step": 37060 }, { "epoch": 0.976480311356819, "grad_norm": 3.8755555152893066, "learning_rate": 2.9071826787033797e-08, "loss": 0.09282814264297486, "step": 37070 }, { "epoch": 0.9767437266013177, "grad_norm": 2.6534504890441895, "learning_rate": 2.8425394374120842e-08, "loss": 0.0744934320449829, "step": 37080 }, { "epoch": 0.9770071418458165, "grad_norm": 3.374199390411377, "learning_rate": 2.7786219703415108e-08, "loss": 0.08524580001831054, "step": 37090 }, { "epoch": 0.9772705570903152, "grad_norm": 3.120309829711914, "learning_rate": 2.715430324013779e-08, "loss": 0.0623554527759552, "step": 37100 }, { "epoch": 0.977533972334814, "grad_norm": 3.503580331802368, "learning_rate": 2.6529645444223206e-08, "loss": 0.07870006561279297, "step": 37110 }, { "epoch": 0.9777973875793127, "grad_norm": 3.004288673400879, "learning_rate": 2.5912246770324333e-08, "loss": 0.08116482496261597, "step": 37120 }, { "epoch": 0.9780608028238115, "grad_norm": 3.613814115524292, "learning_rate": 2.5302107667811716e-08, "loss": 0.08808611631393433, "step": 37130 }, { "epoch": 0.9783242180683102, "grad_norm": 3.535007953643799, "learning_rate": 2.469922858076901e-08, "loss": 0.08463242650032043, "step": 37140 }, { "epoch": 0.9785876333128088, "grad_norm": 5.741289138793945, "learning_rate": 2.410360994799965e-08, "loss": 0.092824387550354, "step": 37150 }, { "epoch": 0.9788510485573076, "grad_norm": 3.226278781890869, "learning_rate": 2.351525220301909e-08, "loss": 0.0844203531742096, "step": 37160 }, { "epoch": 0.9791144638018063, "grad_norm": 3.079904317855835, "learning_rate": 2.293415577406144e-08, "loss": 0.0878986120223999, "step": 37170 }, { "epoch": 0.9793778790463051, "grad_norm": 3.4174656867980957, "learning_rate": 2.2360321084071712e-08, "loss": 0.0683456838130951, "step": 37180 }, { "epoch": 0.9796412942908038, "grad_norm": 2.8130745887756348, "learning_rate": 2.179374855071137e-08, "loss": 0.08688818216323853, "step": 37190 }, { "epoch": 0.9799047095353026, "grad_norm": 3.864722490310669, "learning_rate": 2.123443858635832e-08, "loss": 0.08264228105545043, "step": 37200 }, { "epoch": 0.9801681247798013, "grad_norm": 2.9820022583007812, "learning_rate": 2.068239159810137e-08, "loss": 0.07725671529769898, "step": 37210 }, { "epoch": 0.9804315400243001, "grad_norm": 4.6537766456604, "learning_rate": 2.0137607987743558e-08, "loss": 0.08055187463760376, "step": 37220 }, { "epoch": 0.9806949552687988, "grad_norm": 3.5157113075256348, "learning_rate": 1.9600088151803255e-08, "loss": 0.07928937673568726, "step": 37230 }, { "epoch": 0.9809583705132975, "grad_norm": 3.2602479457855225, "learning_rate": 1.906983248150973e-08, "loss": 0.0725189208984375, "step": 37240 }, { "epoch": 0.9812217857577963, "grad_norm": 3.1686742305755615, "learning_rate": 1.854684136280538e-08, "loss": 0.08302997350692749, "step": 37250 }, { "epoch": 0.981485201002295, "grad_norm": 3.7543227672576904, "learning_rate": 1.8031115176346815e-08, "loss": 0.08461867570877075, "step": 37260 }, { "epoch": 0.9817486162467938, "grad_norm": 3.171018600463867, "learning_rate": 1.752265429750044e-08, "loss": 0.08550771474838256, "step": 37270 }, { "epoch": 0.9820120314912925, "grad_norm": 3.970806837081909, "learning_rate": 1.7021459096348004e-08, "loss": 0.08170843124389648, "step": 37280 }, { "epoch": 0.9822754467357913, "grad_norm": 3.3310792446136475, "learning_rate": 1.652752993767992e-08, "loss": 0.08274611234664916, "step": 37290 }, { "epoch": 0.98253886198029, "grad_norm": 3.3696699142456055, "learning_rate": 1.604086718099973e-08, "loss": 0.07477965354919433, "step": 37300 }, { "epoch": 0.9828022772247887, "grad_norm": 3.3569984436035156, "learning_rate": 1.556147118051965e-08, "loss": 0.07115514278411865, "step": 37310 }, { "epoch": 0.9830656924692874, "grad_norm": 3.0897152423858643, "learning_rate": 1.508934228516723e-08, "loss": 0.07724997997283936, "step": 37320 }, { "epoch": 0.9833291077137861, "grad_norm": 2.9402048587799072, "learning_rate": 1.4624480838576483e-08, "loss": 0.08261755704879761, "step": 37330 }, { "epoch": 0.9835925229582849, "grad_norm": 2.8943748474121094, "learning_rate": 1.4166887179094535e-08, "loss": 0.07149138450622558, "step": 37340 }, { "epoch": 0.9838559382027836, "grad_norm": 3.159092426300049, "learning_rate": 1.3716561639777192e-08, "loss": 0.08272376060485839, "step": 37350 }, { "epoch": 0.9841193534472824, "grad_norm": 3.173973321914673, "learning_rate": 1.3273504548392269e-08, "loss": 0.07817894220352173, "step": 37360 }, { "epoch": 0.9843827686917811, "grad_norm": 3.200773000717163, "learning_rate": 1.2837716227414032e-08, "loss": 0.07244307398796082, "step": 37370 }, { "epoch": 0.9846461839362799, "grad_norm": 2.3096089363098145, "learning_rate": 1.240919699402765e-08, "loss": 0.07797722816467285, "step": 37380 }, { "epoch": 0.9849095991807786, "grad_norm": 2.5614027976989746, "learning_rate": 1.1987947160130297e-08, "loss": 0.08569172620773316, "step": 37390 }, { "epoch": 0.9851730144252773, "grad_norm": 4.086820602416992, "learning_rate": 1.1573967032324495e-08, "loss": 0.08904114365577698, "step": 37400 }, { "epoch": 0.9854364296697761, "grad_norm": 3.472987651824951, "learning_rate": 1.1167256911922552e-08, "loss": 0.07231451272964477, "step": 37410 }, { "epoch": 0.9856998449142748, "grad_norm": 3.103458881378174, "learning_rate": 1.0767817094945454e-08, "loss": 0.0797403633594513, "step": 37420 }, { "epoch": 0.9859632601587736, "grad_norm": 4.382533550262451, "learning_rate": 1.0375647872123974e-08, "loss": 0.07205737829208374, "step": 37430 }, { "epoch": 0.9862266754032722, "grad_norm": 4.394754409790039, "learning_rate": 9.990749528894227e-09, "loss": 0.08119440078735352, "step": 37440 }, { "epoch": 0.986490090647771, "grad_norm": 2.7803127765655518, "learning_rate": 9.613122345403236e-09, "loss": 0.0741765022277832, "step": 37450 }, { "epoch": 0.9867535058922697, "grad_norm": 2.9060475826263428, "learning_rate": 9.24276659650336e-09, "loss": 0.07708529233932496, "step": 37460 }, { "epoch": 0.9870169211367685, "grad_norm": 3.1955819129943848, "learning_rate": 8.879682551755643e-09, "loss": 0.08262293338775635, "step": 37470 }, { "epoch": 0.9872803363812672, "grad_norm": 2.92024564743042, "learning_rate": 8.523870475429797e-09, "loss": 0.07627305984497071, "step": 37480 }, { "epoch": 0.9875437516257659, "grad_norm": 2.9942944049835205, "learning_rate": 8.175330626498668e-09, "loss": 0.06842379570007324, "step": 37490 }, { "epoch": 0.9878071668702647, "grad_norm": 3.4299986362457275, "learning_rate": 7.834063258645997e-09, "loss": 0.08280959129333496, "step": 37500 }, { "epoch": 0.9880705821147634, "grad_norm": 2.8208703994750977, "learning_rate": 7.500068620260869e-09, "loss": 0.09040405154228211, "step": 37510 }, { "epoch": 0.9883339973592622, "grad_norm": 3.520425796508789, "learning_rate": 7.173346954438831e-09, "loss": 0.08398497104644775, "step": 37520 }, { "epoch": 0.9885974126037609, "grad_norm": 3.8190815448760986, "learning_rate": 6.853898498982992e-09, "loss": 0.08018972277641297, "step": 37530 }, { "epoch": 0.9888608278482597, "grad_norm": 2.6620306968688965, "learning_rate": 6.541723486399587e-09, "loss": 0.06985781788825988, "step": 37540 }, { "epoch": 0.9891242430927584, "grad_norm": 3.6336278915405273, "learning_rate": 6.236822143904642e-09, "loss": 0.07285656929016113, "step": 37550 }, { "epoch": 0.9893876583372572, "grad_norm": 3.1808278560638428, "learning_rate": 5.939194693418415e-09, "loss": 0.07602039575576783, "step": 37560 }, { "epoch": 0.9896510735817559, "grad_norm": 4.176185607910156, "learning_rate": 5.648841351566514e-09, "loss": 0.07851302623748779, "step": 37570 }, { "epoch": 0.9899144888262545, "grad_norm": 4.184758186340332, "learning_rate": 5.365762329681002e-09, "loss": 0.07683154344558715, "step": 37580 }, { "epoch": 0.9901779040707533, "grad_norm": 2.7175676822662354, "learning_rate": 5.089957833798176e-09, "loss": 0.08509884476661682, "step": 37590 }, { "epoch": 0.990441319315252, "grad_norm": 2.9967455863952637, "learning_rate": 4.8214280646619035e-09, "loss": 0.08276500701904296, "step": 37600 }, { "epoch": 0.9907047345597508, "grad_norm": 2.7768867015838623, "learning_rate": 4.560173217718067e-09, "loss": 0.07753145694732666, "step": 37610 }, { "epoch": 0.9909681498042495, "grad_norm": 3.8592708110809326, "learning_rate": 4.3061934831212235e-09, "loss": 0.06718257069587708, "step": 37620 }, { "epoch": 0.9912315650487483, "grad_norm": 2.3110647201538086, "learning_rate": 4.0594890457268384e-09, "loss": 0.0689778447151184, "step": 37630 }, { "epoch": 0.991494980293247, "grad_norm": 5.250881195068359, "learning_rate": 3.820060085097943e-09, "loss": 0.08211426734924317, "step": 37640 }, { "epoch": 0.9917583955377458, "grad_norm": 5.522868633270264, "learning_rate": 3.587906775501804e-09, "loss": 0.08195667266845703, "step": 37650 }, { "epoch": 0.9920218107822445, "grad_norm": 2.664371967315674, "learning_rate": 3.363029285907704e-09, "loss": 0.06776826977729797, "step": 37660 }, { "epoch": 0.9922852260267432, "grad_norm": 3.6841681003570557, "learning_rate": 3.145427779993604e-09, "loss": 0.08931547403335571, "step": 37670 }, { "epoch": 0.992548641271242, "grad_norm": 2.964348554611206, "learning_rate": 2.935102416138369e-09, "loss": 0.08838666081428528, "step": 37680 }, { "epoch": 0.9928120565157407, "grad_norm": 2.725663423538208, "learning_rate": 2.732053347425101e-09, "loss": 0.07676072120666504, "step": 37690 }, { "epoch": 0.9930754717602395, "grad_norm": 3.475229024887085, "learning_rate": 2.536280721642248e-09, "loss": 0.0785722017288208, "step": 37700 }, { "epoch": 0.9933388870047382, "grad_norm": 3.126929998397827, "learning_rate": 2.3477846812813843e-09, "loss": 0.07754260301589966, "step": 37710 }, { "epoch": 0.993602302249237, "grad_norm": 3.6615025997161865, "learning_rate": 2.16656536353832e-09, "loss": 0.07481368780136108, "step": 37720 }, { "epoch": 0.9938657174937356, "grad_norm": 3.7577998638153076, "learning_rate": 1.9926229003119912e-09, "loss": 0.07667208909988403, "step": 37730 }, { "epoch": 0.9941291327382344, "grad_norm": 3.9907915592193604, "learning_rate": 1.8259574182055707e-09, "loss": 0.0789380669593811, "step": 37740 }, { "epoch": 0.9943925479827331, "grad_norm": 3.4211857318878174, "learning_rate": 1.6665690385253564e-09, "loss": 0.07980055809020996, "step": 37750 }, { "epoch": 0.9946559632272318, "grad_norm": 3.0441811084747314, "learning_rate": 1.5144578772807728e-09, "loss": 0.08317612409591675, "step": 37760 }, { "epoch": 0.9949193784717306, "grad_norm": 2.4671711921691895, "learning_rate": 1.36962404518437e-09, "loss": 0.08756675720214843, "step": 37770 }, { "epoch": 0.9951827937162293, "grad_norm": 4.265773296356201, "learning_rate": 1.232067647654045e-09, "loss": 0.08295713067054748, "step": 37780 }, { "epoch": 0.9954462089607281, "grad_norm": 3.3054327964782715, "learning_rate": 1.1017887848085995e-09, "loss": 0.07086915969848633, "step": 37790 }, { "epoch": 0.9957096242052268, "grad_norm": 2.593196153640747, "learning_rate": 9.78787551469962e-10, "loss": 0.07671125531196595, "step": 37800 }, { "epoch": 0.9959730394497256, "grad_norm": 3.84032940864563, "learning_rate": 8.630640371642963e-10, "loss": 0.08153756856918334, "step": 37810 }, { "epoch": 0.9962364546942243, "grad_norm": 3.4387857913970947, "learning_rate": 7.546183261197826e-10, "loss": 0.08345973491668701, "step": 37820 }, { "epoch": 0.9964998699387231, "grad_norm": 3.0871620178222656, "learning_rate": 6.534504972688372e-10, "loss": 0.08171728253364563, "step": 37830 }, { "epoch": 0.9967632851832218, "grad_norm": 2.3930070400238037, "learning_rate": 5.595606242447815e-10, "loss": 0.09065428376197815, "step": 37840 }, { "epoch": 0.9970267004277205, "grad_norm": 3.6464085578918457, "learning_rate": 4.729487753851736e-10, "loss": 0.07586486339569092, "step": 37850 }, { "epoch": 0.9972901156722193, "grad_norm": 4.6180572509765625, "learning_rate": 3.9361501372958686e-10, "loss": 0.07660998106002807, "step": 37860 }, { "epoch": 0.997553530916718, "grad_norm": 3.130533218383789, "learning_rate": 3.2155939702072093e-10, "loss": 0.0875010371208191, "step": 37870 }, { "epoch": 0.9978169461612167, "grad_norm": 2.6391897201538086, "learning_rate": 2.5678197770329094e-10, "loss": 0.095641028881073, "step": 37880 }, { "epoch": 0.9980803614057154, "grad_norm": 3.2637107372283936, "learning_rate": 1.9928280292624835e-10, "loss": 0.08152107000350953, "step": 37890 }, { "epoch": 0.9983437766502142, "grad_norm": 3.1285104751586914, "learning_rate": 1.4906191453833985e-10, "loss": 0.08140809535980224, "step": 37900 }, { "epoch": 0.9986071918947129, "grad_norm": 3.316500186920166, "learning_rate": 1.061193490936585e-10, "loss": 0.07635051608085633, "step": 37910 }, { "epoch": 0.9988706071392117, "grad_norm": 2.4398677349090576, "learning_rate": 7.045513784831315e-11, "loss": 0.07451021075248718, "step": 37920 }, { "epoch": 0.9991340223837104, "grad_norm": 2.348752498626709, "learning_rate": 4.206930675820786e-11, "loss": 0.06354578733444213, "step": 37930 }, { "epoch": 0.9993974376282091, "grad_norm": 3.893136739730835, "learning_rate": 2.0961876485703358e-11, "loss": 0.0716029167175293, "step": 37940 }, { "epoch": 0.9996608528727079, "grad_norm": 2.803866147994995, "learning_rate": 7.132862392955631e-12, "loss": 0.0761520266532898, "step": 37950 }, { "epoch": 0.9999242681172066, "grad_norm": 3.2351529598236084, "learning_rate": 5.822745441363964e-13, "loss": 0.07414671182632446, "step": 37960 }, { "epoch": 1.0, "step": 37963, "total_flos": 2.04116045707887e+20, "train_loss": 0.1424806285298725, "train_runtime": 424613.2059, "train_samples_per_second": 5.722, "train_steps_per_second": 0.089 } ], "logging_steps": 10, "max_steps": 37963, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 1000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 2.04116045707887e+20, "train_batch_size": 1, "trial_name": null, "trial_params": null }