{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 37963, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00026341524449873725, "grad_norm": 1216.849609375, "learning_rate": 1.5803336259877089e-07, "loss": 2.3419540405273436, "step": 10 }, { "epoch": 0.0005268304889974745, "grad_norm": 55.98480987548828, "learning_rate": 3.336259877085163e-07, "loss": 0.6632080554962159, "step": 20 }, { "epoch": 0.0007902457334962118, "grad_norm": 32.826438903808594, "learning_rate": 5.092186128182617e-07, "loss": 0.2980868101119995, "step": 30 }, { "epoch": 0.001053660977994949, "grad_norm": 14.342499732971191, "learning_rate": 6.848112379280071e-07, "loss": 0.24977679252624513, "step": 40 }, { "epoch": 0.0013170762224936862, "grad_norm": 21.13075065612793, "learning_rate": 8.604038630377525e-07, "loss": 0.25669407844543457, "step": 50 }, { "epoch": 0.0015804914669924235, "grad_norm": 14.963647842407227, "learning_rate": 1.035996488147498e-06, "loss": 0.2368014335632324, "step": 60 }, { "epoch": 0.0018439067114911607, "grad_norm": 11.261399269104004, "learning_rate": 1.2115891132572432e-06, "loss": 0.22108244895935059, "step": 70 }, { "epoch": 0.002107321955989898, "grad_norm": 9.796932220458984, "learning_rate": 1.3871817383669885e-06, "loss": 0.23258085250854493, "step": 80 }, { "epoch": 0.002370737200488635, "grad_norm": 13.789313316345215, "learning_rate": 1.5627743634767342e-06, "loss": 0.215678071975708, "step": 90 }, { "epoch": 0.0026341524449873723, "grad_norm": 12.801309585571289, "learning_rate": 1.7383669885864795e-06, "loss": 0.2141350746154785, "step": 100 }, { "epoch": 0.00289756768948611, "grad_norm": 10.925505638122559, "learning_rate": 1.913959613696225e-06, "loss": 0.21718010902404786, "step": 110 }, { "epoch": 0.003160982933984847, "grad_norm": 18.31486701965332, "learning_rate": 2.08955223880597e-06, "loss": 0.2327737808227539, "step": 120 }, { "epoch": 0.003424398178483584, "grad_norm": 10.715639114379883, "learning_rate": 2.265144863915716e-06, "loss": 0.2016597270965576, "step": 130 }, { "epoch": 0.0036878134229823213, "grad_norm": 14.487353324890137, "learning_rate": 2.440737489025461e-06, "loss": 0.21462862491607665, "step": 140 }, { "epoch": 0.0039512286674810585, "grad_norm": 10.446657180786133, "learning_rate": 2.6163301141352064e-06, "loss": 0.21720280647277831, "step": 150 }, { "epoch": 0.004214643911979796, "grad_norm": 17.61886978149414, "learning_rate": 2.791922739244952e-06, "loss": 0.23317089080810546, "step": 160 }, { "epoch": 0.004478059156478534, "grad_norm": 11.794365882873535, "learning_rate": 2.9675153643546974e-06, "loss": 0.23239498138427733, "step": 170 }, { "epoch": 0.00474147440097727, "grad_norm": 10.497361183166504, "learning_rate": 3.1431079894644427e-06, "loss": 0.23022017478942872, "step": 180 }, { "epoch": 0.005004889645476008, "grad_norm": 7.195187568664551, "learning_rate": 3.318700614574188e-06, "loss": 0.2124276876449585, "step": 190 }, { "epoch": 0.005268304889974745, "grad_norm": 12.453435897827148, "learning_rate": 3.4942932396839337e-06, "loss": 0.22470149993896485, "step": 200 }, { "epoch": 0.005531720134473482, "grad_norm": 13.42723274230957, "learning_rate": 3.669885864793679e-06, "loss": 0.21294050216674804, "step": 210 }, { "epoch": 0.00579513537897222, "grad_norm": 19.55412483215332, "learning_rate": 3.845478489903424e-06, "loss": 0.20422983169555664, "step": 220 }, { "epoch": 0.0060585506234709565, "grad_norm": 11.286014556884766, "learning_rate": 4.0210711150131695e-06, "loss": 0.21368489265441895, "step": 230 }, { "epoch": 0.006321965867969694, "grad_norm": 13.068394660949707, "learning_rate": 4.196663740122916e-06, "loss": 0.22185089588165283, "step": 240 }, { "epoch": 0.006585381112468432, "grad_norm": 15.069710731506348, "learning_rate": 4.37225636523266e-06, "loss": 0.1773337244987488, "step": 250 }, { "epoch": 0.006848796356967168, "grad_norm": 14.858759880065918, "learning_rate": 4.547848990342406e-06, "loss": 0.21536788940429688, "step": 260 }, { "epoch": 0.007112211601465906, "grad_norm": 6.253750324249268, "learning_rate": 4.7234416154521515e-06, "loss": 0.1933940052986145, "step": 270 }, { "epoch": 0.007375626845964643, "grad_norm": 12.62516975402832, "learning_rate": 4.899034240561897e-06, "loss": 0.19879379272460937, "step": 280 }, { "epoch": 0.00763904209046338, "grad_norm": 8.808650016784668, "learning_rate": 5.074626865671642e-06, "loss": 0.19079053401947021, "step": 290 }, { "epoch": 0.007902457334962117, "grad_norm": 10.257344245910645, "learning_rate": 5.250219490781387e-06, "loss": 0.1984729766845703, "step": 300 }, { "epoch": 0.008165872579460855, "grad_norm": 9.45608139038086, "learning_rate": 5.425812115891133e-06, "loss": 0.19159960746765137, "step": 310 }, { "epoch": 0.008429287823959592, "grad_norm": 11.971735000610352, "learning_rate": 5.601404741000879e-06, "loss": 0.19909011125564574, "step": 320 }, { "epoch": 0.008692703068458329, "grad_norm": 7.18591833114624, "learning_rate": 5.776997366110624e-06, "loss": 0.2063279628753662, "step": 330 }, { "epoch": 0.008956118312957067, "grad_norm": 13.483114242553711, "learning_rate": 5.9525899912203685e-06, "loss": 0.19191381931304932, "step": 340 }, { "epoch": 0.009219533557455804, "grad_norm": 10.279247283935547, "learning_rate": 6.128182616330115e-06, "loss": 0.21518998146057128, "step": 350 }, { "epoch": 0.00948294880195454, "grad_norm": 9.922018051147461, "learning_rate": 6.30377524143986e-06, "loss": 0.1996060848236084, "step": 360 }, { "epoch": 0.00974636404645328, "grad_norm": 7.3402814865112305, "learning_rate": 6.479367866549605e-06, "loss": 0.2057633399963379, "step": 370 }, { "epoch": 0.010009779290952016, "grad_norm": 5.752679824829102, "learning_rate": 6.654960491659351e-06, "loss": 0.20380020141601562, "step": 380 }, { "epoch": 0.010273194535450753, "grad_norm": 12.97748851776123, "learning_rate": 6.830553116769097e-06, "loss": 0.19769517183303834, "step": 390 }, { "epoch": 0.01053660977994949, "grad_norm": 8.03429889678955, "learning_rate": 7.006145741878841e-06, "loss": 0.19357199668884278, "step": 400 }, { "epoch": 0.010800025024448228, "grad_norm": 5.049175262451172, "learning_rate": 7.181738366988587e-06, "loss": 0.19299917221069335, "step": 410 }, { "epoch": 0.011063440268946964, "grad_norm": 5.918609619140625, "learning_rate": 7.3573309920983325e-06, "loss": 0.1978721499443054, "step": 420 }, { "epoch": 0.011326855513445701, "grad_norm": 12.29189682006836, "learning_rate": 7.532923617208078e-06, "loss": 0.1786757230758667, "step": 430 }, { "epoch": 0.01159027075794444, "grad_norm": 4.807268142700195, "learning_rate": 7.708516242317823e-06, "loss": 0.1882766604423523, "step": 440 }, { "epoch": 0.011853686002443176, "grad_norm": 6.599029064178467, "learning_rate": 7.88410886742757e-06, "loss": 0.19589791297912598, "step": 450 }, { "epoch": 0.012117101246941913, "grad_norm": 6.497337818145752, "learning_rate": 8.059701492537314e-06, "loss": 0.19239648580551147, "step": 460 }, { "epoch": 0.012380516491440651, "grad_norm": 5.329344749450684, "learning_rate": 8.23529411764706e-06, "loss": 0.18113877773284912, "step": 470 }, { "epoch": 0.012643931735939388, "grad_norm": 5.775257110595703, "learning_rate": 8.410886742756806e-06, "loss": 0.19387485980987548, "step": 480 }, { "epoch": 0.012907346980438125, "grad_norm": 8.132734298706055, "learning_rate": 8.58647936786655e-06, "loss": 0.19293633699417115, "step": 490 }, { "epoch": 0.013170762224936863, "grad_norm": 14.033019065856934, "learning_rate": 8.762071992976295e-06, "loss": 0.2104552984237671, "step": 500 }, { "epoch": 0.0134341774694356, "grad_norm": 8.634580612182617, "learning_rate": 8.937664618086041e-06, "loss": 0.20963048934936523, "step": 510 }, { "epoch": 0.013697592713934337, "grad_norm": 9.856078147888184, "learning_rate": 9.113257243195787e-06, "loss": 0.20487892627716064, "step": 520 }, { "epoch": 0.013961007958433075, "grad_norm": 6.678592205047607, "learning_rate": 9.288849868305532e-06, "loss": 0.19745393991470336, "step": 530 }, { "epoch": 0.014224423202931812, "grad_norm": 10.495431900024414, "learning_rate": 9.464442493415278e-06, "loss": 0.19688940048217773, "step": 540 }, { "epoch": 0.014487838447430549, "grad_norm": 3.559129238128662, "learning_rate": 9.640035118525024e-06, "loss": 0.18834319114685058, "step": 550 }, { "epoch": 0.014751253691929285, "grad_norm": 6.145140171051025, "learning_rate": 9.815627743634768e-06, "loss": 0.18238451480865478, "step": 560 }, { "epoch": 0.015014668936428024, "grad_norm": 5.498777389526367, "learning_rate": 9.991220368744513e-06, "loss": 0.19448280334472656, "step": 570 }, { "epoch": 0.01527808418092676, "grad_norm": 6.271116256713867, "learning_rate": 1.0166812993854259e-05, "loss": 0.19710817337036132, "step": 580 }, { "epoch": 0.015541499425425497, "grad_norm": 7.789073944091797, "learning_rate": 1.0342405618964005e-05, "loss": 0.18335962295532227, "step": 590 }, { "epoch": 0.015804914669924234, "grad_norm": 10.154095649719238, "learning_rate": 1.0517998244073751e-05, "loss": 0.19115090370178223, "step": 600 }, { "epoch": 0.016068329914422974, "grad_norm": 5.740118026733398, "learning_rate": 1.0693590869183494e-05, "loss": 0.1950551986694336, "step": 610 }, { "epoch": 0.01633174515892171, "grad_norm": 4.782759189605713, "learning_rate": 1.086918349429324e-05, "loss": 0.18122336864471436, "step": 620 }, { "epoch": 0.016595160403420448, "grad_norm": 10.19804573059082, "learning_rate": 1.1044776119402986e-05, "loss": 0.2041015863418579, "step": 630 }, { "epoch": 0.016858575647919184, "grad_norm": 7.09054708480835, "learning_rate": 1.122036874451273e-05, "loss": 0.18297982215881348, "step": 640 }, { "epoch": 0.01712199089241792, "grad_norm": 7.853230953216553, "learning_rate": 1.1395961369622477e-05, "loss": 0.1865099310874939, "step": 650 }, { "epoch": 0.017385406136916658, "grad_norm": 5.164277076721191, "learning_rate": 1.1571553994732223e-05, "loss": 0.210978102684021, "step": 660 }, { "epoch": 0.017648821381415394, "grad_norm": 5.750659942626953, "learning_rate": 1.1747146619841967e-05, "loss": 0.20355663299560547, "step": 670 }, { "epoch": 0.017912236625914135, "grad_norm": 5.682198524475098, "learning_rate": 1.1922739244951713e-05, "loss": 0.18989579677581786, "step": 680 }, { "epoch": 0.01817565187041287, "grad_norm": 8.722765922546387, "learning_rate": 1.209833187006146e-05, "loss": 0.20457243919372559, "step": 690 }, { "epoch": 0.018439067114911608, "grad_norm": 6.917410373687744, "learning_rate": 1.2273924495171204e-05, "loss": 0.1991649627685547, "step": 700 }, { "epoch": 0.018702482359410345, "grad_norm": 8.688848495483398, "learning_rate": 1.244951712028095e-05, "loss": 0.18175120353698732, "step": 710 }, { "epoch": 0.01896589760390908, "grad_norm": 7.67869758605957, "learning_rate": 1.2625109745390696e-05, "loss": 0.19073717594146727, "step": 720 }, { "epoch": 0.019229312848407818, "grad_norm": 4.780980587005615, "learning_rate": 1.2800702370500439e-05, "loss": 0.18740310668945312, "step": 730 }, { "epoch": 0.01949272809290656, "grad_norm": 7.919905662536621, "learning_rate": 1.2976294995610185e-05, "loss": 0.19469163417816163, "step": 740 }, { "epoch": 0.019756143337405295, "grad_norm": 6.539961338043213, "learning_rate": 1.315188762071993e-05, "loss": 0.20438952445983888, "step": 750 }, { "epoch": 0.02001955858190403, "grad_norm": 6.450709819793701, "learning_rate": 1.3327480245829676e-05, "loss": 0.1939136028289795, "step": 760 }, { "epoch": 0.02028297382640277, "grad_norm": 7.711205959320068, "learning_rate": 1.3503072870939422e-05, "loss": 0.20302093029022217, "step": 770 }, { "epoch": 0.020546389070901505, "grad_norm": 4.519542217254639, "learning_rate": 1.3678665496049166e-05, "loss": 0.19372329711914063, "step": 780 }, { "epoch": 0.020809804315400242, "grad_norm": 6.8084917068481445, "learning_rate": 1.3854258121158912e-05, "loss": 0.18195502758026122, "step": 790 }, { "epoch": 0.02107321955989898, "grad_norm": 14.18945598602295, "learning_rate": 1.4029850746268658e-05, "loss": 0.2100691795349121, "step": 800 }, { "epoch": 0.02133663480439772, "grad_norm": 5.325632572174072, "learning_rate": 1.4205443371378403e-05, "loss": 0.1867142677307129, "step": 810 }, { "epoch": 0.021600050048896455, "grad_norm": 4.450637340545654, "learning_rate": 1.4381035996488149e-05, "loss": 0.1825059175491333, "step": 820 }, { "epoch": 0.021863465293395192, "grad_norm": 3.700251340866089, "learning_rate": 1.4556628621597895e-05, "loss": 0.18878250122070311, "step": 830 }, { "epoch": 0.02212688053789393, "grad_norm": 3.466278076171875, "learning_rate": 1.4732221246707638e-05, "loss": 0.19076406955718994, "step": 840 }, { "epoch": 0.022390295782392666, "grad_norm": 5.9916534423828125, "learning_rate": 1.4907813871817384e-05, "loss": 0.18699628114700317, "step": 850 }, { "epoch": 0.022653711026891402, "grad_norm": 4.045867919921875, "learning_rate": 1.508340649692713e-05, "loss": 0.18128135204315185, "step": 860 }, { "epoch": 0.022917126271390142, "grad_norm": 6.328861713409424, "learning_rate": 1.5258999122036876e-05, "loss": 0.18992817401885986, "step": 870 }, { "epoch": 0.02318054151588888, "grad_norm": 4.434060096740723, "learning_rate": 1.5434591747146622e-05, "loss": 0.19711986780166627, "step": 880 }, { "epoch": 0.023443956760387616, "grad_norm": 5.92913293838501, "learning_rate": 1.561018437225637e-05, "loss": 0.18557993173599244, "step": 890 }, { "epoch": 0.023707372004886353, "grad_norm": 4.944325923919678, "learning_rate": 1.578577699736611e-05, "loss": 0.17658642530441285, "step": 900 }, { "epoch": 0.02397078724938509, "grad_norm": 2.720505714416504, "learning_rate": 1.5961369622475857e-05, "loss": 0.1952008366584778, "step": 910 }, { "epoch": 0.024234202493883826, "grad_norm": 11.049121856689453, "learning_rate": 1.6136962247585604e-05, "loss": 0.19520888328552247, "step": 920 }, { "epoch": 0.024497617738382566, "grad_norm": 2.688471794128418, "learning_rate": 1.6312554872695346e-05, "loss": 0.2007538080215454, "step": 930 }, { "epoch": 0.024761032982881303, "grad_norm": 3.5442702770233154, "learning_rate": 1.6488147497805093e-05, "loss": 0.184484601020813, "step": 940 }, { "epoch": 0.02502444822738004, "grad_norm": 4.19333553314209, "learning_rate": 1.666374012291484e-05, "loss": 0.17519211769104004, "step": 950 }, { "epoch": 0.025287863471878776, "grad_norm": 4.998476982116699, "learning_rate": 1.6839332748024585e-05, "loss": 0.18412561416625978, "step": 960 }, { "epoch": 0.025551278716377513, "grad_norm": 2.6268222332000732, "learning_rate": 1.701492537313433e-05, "loss": 0.1787615180015564, "step": 970 }, { "epoch": 0.02581469396087625, "grad_norm": 8.570691108703613, "learning_rate": 1.7190517998244077e-05, "loss": 0.2003704309463501, "step": 980 }, { "epoch": 0.026078109205374986, "grad_norm": 4.213146686553955, "learning_rate": 1.736611062335382e-05, "loss": 0.18344511985778808, "step": 990 }, { "epoch": 0.026341524449873727, "grad_norm": 5.138370513916016, "learning_rate": 1.7541703248463566e-05, "loss": 0.19046225547790527, "step": 1000 }, { "epoch": 0.026604939694372463, "grad_norm": 3.461799383163452, "learning_rate": 1.771729587357331e-05, "loss": 0.19381375312805177, "step": 1010 }, { "epoch": 0.0268683549388712, "grad_norm": 5.666082382202148, "learning_rate": 1.7892888498683055e-05, "loss": 0.19855175018310547, "step": 1020 }, { "epoch": 0.027131770183369937, "grad_norm": 3.175442934036255, "learning_rate": 1.80684811237928e-05, "loss": 0.1816019058227539, "step": 1030 }, { "epoch": 0.027395185427868674, "grad_norm": 3.104397773742676, "learning_rate": 1.8244073748902547e-05, "loss": 0.18806583881378175, "step": 1040 }, { "epoch": 0.02765860067236741, "grad_norm": 4.816192150115967, "learning_rate": 1.8419666374012293e-05, "loss": 0.1795586585998535, "step": 1050 }, { "epoch": 0.02792201591686615, "grad_norm": 4.8531880378723145, "learning_rate": 1.859525899912204e-05, "loss": 0.1978344202041626, "step": 1060 }, { "epoch": 0.028185431161364887, "grad_norm": 5.264528274536133, "learning_rate": 1.8770851624231782e-05, "loss": 0.178529953956604, "step": 1070 }, { "epoch": 0.028448846405863624, "grad_norm": 2.228271484375, "learning_rate": 1.8946444249341528e-05, "loss": 0.2057192802429199, "step": 1080 }, { "epoch": 0.02871226165036236, "grad_norm": 6.317373275756836, "learning_rate": 1.9122036874451274e-05, "loss": 0.199721622467041, "step": 1090 }, { "epoch": 0.028975676894861097, "grad_norm": 3.1878602504730225, "learning_rate": 1.929762949956102e-05, "loss": 0.1858614206314087, "step": 1100 }, { "epoch": 0.029239092139359834, "grad_norm": 3.25897479057312, "learning_rate": 1.9473222124670767e-05, "loss": 0.1948639154434204, "step": 1110 }, { "epoch": 0.02950250738385857, "grad_norm": 5.339667320251465, "learning_rate": 1.9648814749780513e-05, "loss": 0.17678192853927613, "step": 1120 }, { "epoch": 0.02976592262835731, "grad_norm": 5.180049419403076, "learning_rate": 1.9824407374890256e-05, "loss": 0.18715614080429077, "step": 1130 }, { "epoch": 0.030029337872856048, "grad_norm": 3.1028027534484863, "learning_rate": 2e-05, "loss": 0.18706140518188477, "step": 1140 }, { "epoch": 0.030292753117354784, "grad_norm": 7.96759033203125, "learning_rate": 1.9999996360784282e-05, "loss": 0.1867927074432373, "step": 1150 }, { "epoch": 0.03055616836185352, "grad_norm": 5.474365711212158, "learning_rate": 1.9999985443139778e-05, "loss": 0.17893650531768798, "step": 1160 }, { "epoch": 0.030819583606352258, "grad_norm": 3.0950846672058105, "learning_rate": 1.999996724707443e-05, "loss": 0.1875794529914856, "step": 1170 }, { "epoch": 0.031082998850850994, "grad_norm": 2.613436698913574, "learning_rate": 1.9999941772601482e-05, "loss": 0.17534806728363037, "step": 1180 }, { "epoch": 0.03134641409534973, "grad_norm": 3.5791523456573486, "learning_rate": 1.9999909019739482e-05, "loss": 0.18119866847991944, "step": 1190 }, { "epoch": 0.03160982933984847, "grad_norm": 7.290497779846191, "learning_rate": 1.9999868988512265e-05, "loss": 0.17440826892852784, "step": 1200 }, { "epoch": 0.031873244584347205, "grad_norm": 13.073683738708496, "learning_rate": 1.9999821678948964e-05, "loss": 0.17784858942031861, "step": 1210 }, { "epoch": 0.03213665982884595, "grad_norm": 7.544722557067871, "learning_rate": 1.999976709108402e-05, "loss": 0.1909373641014099, "step": 1220 }, { "epoch": 0.032400075073344685, "grad_norm": 8.275629997253418, "learning_rate": 1.9999705224957156e-05, "loss": 0.2059110403060913, "step": 1230 }, { "epoch": 0.03266349031784342, "grad_norm": 12.977010726928711, "learning_rate": 1.9999636080613406e-05, "loss": 0.1972394108772278, "step": 1240 }, { "epoch": 0.03292690556234216, "grad_norm": 5.883469104766846, "learning_rate": 1.9999559658103097e-05, "loss": 0.19403891563415526, "step": 1250 }, { "epoch": 0.033190320806840895, "grad_norm": 5.391487121582031, "learning_rate": 1.9999475957481848e-05, "loss": 0.19793307781219482, "step": 1260 }, { "epoch": 0.03345373605133963, "grad_norm": 2.229424476623535, "learning_rate": 1.999938497881059e-05, "loss": 0.1997523069381714, "step": 1270 }, { "epoch": 0.03371715129583837, "grad_norm": 3.3801238536834717, "learning_rate": 1.999928672215553e-05, "loss": 0.17972661256790162, "step": 1280 }, { "epoch": 0.033980566540337105, "grad_norm": 1.9428367614746094, "learning_rate": 1.9999181187588187e-05, "loss": 0.19339605569839477, "step": 1290 }, { "epoch": 0.03424398178483584, "grad_norm": 3.8055007457733154, "learning_rate": 1.9999068375185372e-05, "loss": 0.1797175645828247, "step": 1300 }, { "epoch": 0.03450739702933458, "grad_norm": 7.663653373718262, "learning_rate": 1.99989482850292e-05, "loss": 0.1794891119003296, "step": 1310 }, { "epoch": 0.034770812273833315, "grad_norm": 2.003793478012085, "learning_rate": 1.9998820917207076e-05, "loss": 0.18258116245269776, "step": 1320 }, { "epoch": 0.03503422751833205, "grad_norm": 3.2850828170776367, "learning_rate": 1.99986862718117e-05, "loss": 0.18429110050201417, "step": 1330 }, { "epoch": 0.03529764276283079, "grad_norm": 5.312020778656006, "learning_rate": 1.999854434894108e-05, "loss": 0.18973379135131835, "step": 1340 }, { "epoch": 0.03556105800732953, "grad_norm": 4.556980133056641, "learning_rate": 1.9998395148698505e-05, "loss": 0.18756247758865358, "step": 1350 }, { "epoch": 0.03582447325182827, "grad_norm": 10.46460247039795, "learning_rate": 1.9998238671192575e-05, "loss": 0.1856353759765625, "step": 1360 }, { "epoch": 0.036087888496327006, "grad_norm": 5.454447269439697, "learning_rate": 1.999807491653718e-05, "loss": 0.1841325044631958, "step": 1370 }, { "epoch": 0.03635130374082574, "grad_norm": 4.511054992675781, "learning_rate": 1.9997903884851503e-05, "loss": 0.17717418670654297, "step": 1380 }, { "epoch": 0.03661471898532448, "grad_norm": 3.3337926864624023, "learning_rate": 1.999772557626004e-05, "loss": 0.17742083072662354, "step": 1390 }, { "epoch": 0.036878134229823216, "grad_norm": 2.8575267791748047, "learning_rate": 1.9997539990892555e-05, "loss": 0.16801300048828124, "step": 1400 }, { "epoch": 0.03714154947432195, "grad_norm": 4.745070934295654, "learning_rate": 1.9997347128884138e-05, "loss": 0.181700599193573, "step": 1410 }, { "epoch": 0.03740496471882069, "grad_norm": 1.8378353118896484, "learning_rate": 1.9997146990375162e-05, "loss": 0.1892989993095398, "step": 1420 }, { "epoch": 0.037668379963319426, "grad_norm": 4.480461120605469, "learning_rate": 1.999693957551129e-05, "loss": 0.1772119164466858, "step": 1430 }, { "epoch": 0.03793179520781816, "grad_norm": 5.885284900665283, "learning_rate": 1.999672488444349e-05, "loss": 0.17454792261123658, "step": 1440 }, { "epoch": 0.0381952104523169, "grad_norm": 3.948930263519287, "learning_rate": 1.999650291732802e-05, "loss": 0.19416134357452391, "step": 1450 }, { "epoch": 0.038458625696815636, "grad_norm": 2.330162763595581, "learning_rate": 1.9996273674326446e-05, "loss": 0.18732736110687256, "step": 1460 }, { "epoch": 0.03872204094131437, "grad_norm": 3.0682387351989746, "learning_rate": 1.9996037155605615e-05, "loss": 0.1755363702774048, "step": 1470 }, { "epoch": 0.03898545618581312, "grad_norm": 3.3513667583465576, "learning_rate": 1.9995793361337676e-05, "loss": 0.18232324123382568, "step": 1480 }, { "epoch": 0.03924887143031185, "grad_norm": 5.149270057678223, "learning_rate": 1.9995542291700075e-05, "loss": 0.18353967666625975, "step": 1490 }, { "epoch": 0.03951228667481059, "grad_norm": 2.416672468185425, "learning_rate": 1.9995283946875548e-05, "loss": 0.18040366172790528, "step": 1500 }, { "epoch": 0.03977570191930933, "grad_norm": 4.940719127655029, "learning_rate": 1.999501832705213e-05, "loss": 0.16476590633392335, "step": 1510 }, { "epoch": 0.04003911716380806, "grad_norm": 2.3744845390319824, "learning_rate": 1.9994745432423154e-05, "loss": 0.16952116489410402, "step": 1520 }, { "epoch": 0.0403025324083068, "grad_norm": 4.365732669830322, "learning_rate": 1.9994465263187242e-05, "loss": 0.16485872268676757, "step": 1530 }, { "epoch": 0.04056594765280554, "grad_norm": 4.121256351470947, "learning_rate": 1.9994177819548313e-05, "loss": 0.19694535732269286, "step": 1540 }, { "epoch": 0.040829362897304274, "grad_norm": 4.0409417152404785, "learning_rate": 1.999388310171558e-05, "loss": 0.18773218393325805, "step": 1550 }, { "epoch": 0.04109277814180301, "grad_norm": 4.8279852867126465, "learning_rate": 1.9993581109903557e-05, "loss": 0.16155391931533813, "step": 1560 }, { "epoch": 0.04135619338630175, "grad_norm": 7.383152008056641, "learning_rate": 1.999327184433204e-05, "loss": 0.18062840700149535, "step": 1570 }, { "epoch": 0.041619608630800484, "grad_norm": 3.1306707859039307, "learning_rate": 1.9992955305226127e-05, "loss": 0.17759580612182618, "step": 1580 }, { "epoch": 0.04188302387529922, "grad_norm": 2.68891978263855, "learning_rate": 1.9992631492816214e-05, "loss": 0.17717865705490113, "step": 1590 }, { "epoch": 0.04214643911979796, "grad_norm": 2.864929437637329, "learning_rate": 1.9992300407337977e-05, "loss": 0.19280545711517333, "step": 1600 }, { "epoch": 0.0424098543642967, "grad_norm": 7.820135593414307, "learning_rate": 1.9991962049032403e-05, "loss": 0.17027571201324462, "step": 1610 }, { "epoch": 0.04267326960879544, "grad_norm": 8.171343803405762, "learning_rate": 1.9991616418145757e-05, "loss": 0.186654531955719, "step": 1620 }, { "epoch": 0.042936684853294174, "grad_norm": 4.656008720397949, "learning_rate": 1.9991263514929606e-05, "loss": 0.1733616352081299, "step": 1630 }, { "epoch": 0.04320010009779291, "grad_norm": 7.261754512786865, "learning_rate": 1.999090333964081e-05, "loss": 0.18490874767303467, "step": 1640 }, { "epoch": 0.04346351534229165, "grad_norm": 1.9213471412658691, "learning_rate": 1.9990535892541524e-05, "loss": 0.16901130676269532, "step": 1650 }, { "epoch": 0.043726930586790384, "grad_norm": 4.646639823913574, "learning_rate": 1.999016117389918e-05, "loss": 0.19688060283660888, "step": 1660 }, { "epoch": 0.04399034583128912, "grad_norm": 5.26275110244751, "learning_rate": 1.998977918398653e-05, "loss": 0.19447994232177734, "step": 1670 }, { "epoch": 0.04425376107578786, "grad_norm": 2.3935093879699707, "learning_rate": 1.9989389923081587e-05, "loss": 0.17143173217773439, "step": 1680 }, { "epoch": 0.044517176320286594, "grad_norm": 2.9727699756622314, "learning_rate": 1.998899339146768e-05, "loss": 0.17035222053527832, "step": 1690 }, { "epoch": 0.04478059156478533, "grad_norm": 2.4312655925750732, "learning_rate": 1.9988589589433423e-05, "loss": 0.1920400619506836, "step": 1700 }, { "epoch": 0.04504400680928407, "grad_norm": 5.3249921798706055, "learning_rate": 1.9988178517272717e-05, "loss": 0.18626132011413574, "step": 1710 }, { "epoch": 0.045307422053782805, "grad_norm": 3.038191556930542, "learning_rate": 1.998776017528476e-05, "loss": 0.18655831813812257, "step": 1720 }, { "epoch": 0.04557083729828155, "grad_norm": 4.33893346786499, "learning_rate": 1.9987334563774038e-05, "loss": 0.1897730350494385, "step": 1730 }, { "epoch": 0.045834252542780285, "grad_norm": 5.51531457901001, "learning_rate": 1.998690168305033e-05, "loss": 0.17306649684906006, "step": 1740 }, { "epoch": 0.04609766778727902, "grad_norm": 2.1397757530212402, "learning_rate": 1.9986461533428708e-05, "loss": 0.19089939594268798, "step": 1750 }, { "epoch": 0.04636108303177776, "grad_norm": 3.0755720138549805, "learning_rate": 1.9986014115229525e-05, "loss": 0.18136264085769654, "step": 1760 }, { "epoch": 0.046624498276276495, "grad_norm": 1.7844812870025635, "learning_rate": 1.998555942877844e-05, "loss": 0.18532176017761232, "step": 1770 }, { "epoch": 0.04688791352077523, "grad_norm": 3.129939317703247, "learning_rate": 1.9985097474406387e-05, "loss": 0.18313549757003783, "step": 1780 }, { "epoch": 0.04715132876527397, "grad_norm": 3.079510450363159, "learning_rate": 1.99846282524496e-05, "loss": 0.17228524684906005, "step": 1790 }, { "epoch": 0.047414744009772705, "grad_norm": 4.4627366065979, "learning_rate": 1.9984151763249598e-05, "loss": 0.1857968807220459, "step": 1800 }, { "epoch": 0.04767815925427144, "grad_norm": 2.8774731159210205, "learning_rate": 1.9983668007153185e-05, "loss": 0.18854742050170897, "step": 1810 }, { "epoch": 0.04794157449877018, "grad_norm": 2.9625985622406006, "learning_rate": 1.998317698451247e-05, "loss": 0.1765884518623352, "step": 1820 }, { "epoch": 0.048204989743268915, "grad_norm": 2.4045097827911377, "learning_rate": 1.9982678695684833e-05, "loss": 0.17790374755859376, "step": 1830 }, { "epoch": 0.04846840498776765, "grad_norm": 2.212707281112671, "learning_rate": 1.998217314103295e-05, "loss": 0.17856059074401856, "step": 1840 }, { "epoch": 0.04873182023226639, "grad_norm": 2.4623351097106934, "learning_rate": 1.998166032092479e-05, "loss": 0.1830640196800232, "step": 1850 }, { "epoch": 0.04899523547676513, "grad_norm": 2.410910129547119, "learning_rate": 1.99811402357336e-05, "loss": 0.166536283493042, "step": 1860 }, { "epoch": 0.04925865072126387, "grad_norm": 3.749934434890747, "learning_rate": 1.9980612885837925e-05, "loss": 0.18578824996948243, "step": 1870 }, { "epoch": 0.049522065965762606, "grad_norm": 2.0501391887664795, "learning_rate": 1.998007827162159e-05, "loss": 0.17060635089874268, "step": 1880 }, { "epoch": 0.04978548121026134, "grad_norm": 6.056958198547363, "learning_rate": 1.9979536393473715e-05, "loss": 0.17913491725921632, "step": 1890 }, { "epoch": 0.05004889645476008, "grad_norm": 3.2867374420166016, "learning_rate": 1.9978987251788696e-05, "loss": 0.17854552268981932, "step": 1900 }, { "epoch": 0.050312311699258816, "grad_norm": 5.2379608154296875, "learning_rate": 1.997843084696623e-05, "loss": 0.18391668796539307, "step": 1910 }, { "epoch": 0.05057572694375755, "grad_norm": 6.350252628326416, "learning_rate": 1.9977867179411282e-05, "loss": 0.1608332395553589, "step": 1920 }, { "epoch": 0.05083914218825629, "grad_norm": 3.3664345741271973, "learning_rate": 1.997729624953412e-05, "loss": 0.18402554988861083, "step": 1930 }, { "epoch": 0.051102557432755026, "grad_norm": 1.9777660369873047, "learning_rate": 1.997671805775029e-05, "loss": 0.18098201751708984, "step": 1940 }, { "epoch": 0.05136597267725376, "grad_norm": 4.576654434204102, "learning_rate": 1.9976132604480625e-05, "loss": 0.17587220668792725, "step": 1950 }, { "epoch": 0.0516293879217525, "grad_norm": 3.6120221614837646, "learning_rate": 1.997553989015125e-05, "loss": 0.19045765399932862, "step": 1960 }, { "epoch": 0.051892803166251236, "grad_norm": 4.202922821044922, "learning_rate": 1.9974939915193557e-05, "loss": 0.17256920337677, "step": 1970 }, { "epoch": 0.05215621841074997, "grad_norm": 2.4498140811920166, "learning_rate": 1.9974332680044237e-05, "loss": 0.18274770975112914, "step": 1980 }, { "epoch": 0.05241963365524872, "grad_norm": 1.978064775466919, "learning_rate": 1.9973718185145262e-05, "loss": 0.16635706424713134, "step": 1990 }, { "epoch": 0.05268304889974745, "grad_norm": 6.576446533203125, "learning_rate": 1.997309643094389e-05, "loss": 0.17764017581939698, "step": 2000 }, { "epoch": 0.05294646414424619, "grad_norm": 1.4355237483978271, "learning_rate": 1.997246741789266e-05, "loss": 0.1741661787033081, "step": 2010 }, { "epoch": 0.05320987938874493, "grad_norm": 4.901126861572266, "learning_rate": 1.99718311464494e-05, "loss": 0.18326185941696166, "step": 2020 }, { "epoch": 0.05347329463324366, "grad_norm": 3.029806613922119, "learning_rate": 1.9971187617077204e-05, "loss": 0.1744728684425354, "step": 2030 }, { "epoch": 0.0537367098777424, "grad_norm": 3.389781951904297, "learning_rate": 1.9970536830244466e-05, "loss": 0.17451212406158448, "step": 2040 }, { "epoch": 0.05400012512224114, "grad_norm": 2.0862302780151367, "learning_rate": 1.9969878786424863e-05, "loss": 0.17360986471176149, "step": 2050 }, { "epoch": 0.054263540366739874, "grad_norm": 2.773759365081787, "learning_rate": 1.9969213486097335e-05, "loss": 0.1843271017074585, "step": 2060 }, { "epoch": 0.05452695561123861, "grad_norm": 6.461215972900391, "learning_rate": 1.996854092974613e-05, "loss": 0.19020806550979613, "step": 2070 }, { "epoch": 0.05479037085573735, "grad_norm": 4.376319408416748, "learning_rate": 1.9967861117860754e-05, "loss": 0.17380685806274415, "step": 2080 }, { "epoch": 0.055053786100236084, "grad_norm": 2.908147096633911, "learning_rate": 1.9967174050936008e-05, "loss": 0.17962567806243895, "step": 2090 }, { "epoch": 0.05531720134473482, "grad_norm": 1.7913932800292969, "learning_rate": 1.9966479729471966e-05, "loss": 0.1898908019065857, "step": 2100 }, { "epoch": 0.05558061658923356, "grad_norm": 4.740969657897949, "learning_rate": 1.9965778153973986e-05, "loss": 0.16749815940856932, "step": 2110 }, { "epoch": 0.0558440318337323, "grad_norm": 2.8134169578552246, "learning_rate": 1.9965069324952703e-05, "loss": 0.17976722717285157, "step": 2120 }, { "epoch": 0.05610744707823104, "grad_norm": 6.888369560241699, "learning_rate": 1.9964353242924038e-05, "loss": 0.18248343467712402, "step": 2130 }, { "epoch": 0.056370862322729774, "grad_norm": 2.4826152324676514, "learning_rate": 1.9963629908409184e-05, "loss": 0.1791386842727661, "step": 2140 }, { "epoch": 0.05663427756722851, "grad_norm": 2.5916008949279785, "learning_rate": 1.9962899321934616e-05, "loss": 0.16631815433502198, "step": 2150 }, { "epoch": 0.05689769281172725, "grad_norm": 2.804347276687622, "learning_rate": 1.9962161484032085e-05, "loss": 0.16161975860595704, "step": 2160 }, { "epoch": 0.057161108056225984, "grad_norm": 2.5615134239196777, "learning_rate": 1.9961416395238623e-05, "loss": 0.15676603317260743, "step": 2170 }, { "epoch": 0.05742452330072472, "grad_norm": 3.123124361038208, "learning_rate": 1.9960664056096534e-05, "loss": 0.19687743186950685, "step": 2180 }, { "epoch": 0.05768793854522346, "grad_norm": 7.861266136169434, "learning_rate": 1.9959904467153408e-05, "loss": 0.1938778877258301, "step": 2190 }, { "epoch": 0.057951353789722194, "grad_norm": 1.9199448823928833, "learning_rate": 1.99591376289621e-05, "loss": 0.16764333248138427, "step": 2200 }, { "epoch": 0.05821476903422093, "grad_norm": 4.153176784515381, "learning_rate": 1.9958363542080758e-05, "loss": 0.17420618534088134, "step": 2210 }, { "epoch": 0.05847818427871967, "grad_norm": 3.669997215270996, "learning_rate": 1.9957582207072784e-05, "loss": 0.16904778480529786, "step": 2220 }, { "epoch": 0.058741599523218405, "grad_norm": 5.193276882171631, "learning_rate": 1.9956793624506875e-05, "loss": 0.17843559980392457, "step": 2230 }, { "epoch": 0.05900501476771714, "grad_norm": 5.262317180633545, "learning_rate": 1.9955997794956994e-05, "loss": 0.16676797866821289, "step": 2240 }, { "epoch": 0.059268430012215885, "grad_norm": 3.4274308681488037, "learning_rate": 1.995519471900238e-05, "loss": 0.17407801151275634, "step": 2250 }, { "epoch": 0.05953184525671462, "grad_norm": 2.417570114135742, "learning_rate": 1.9954384397227545e-05, "loss": 0.17624794244766234, "step": 2260 }, { "epoch": 0.05979526050121336, "grad_norm": 4.500149250030518, "learning_rate": 1.9953566830222276e-05, "loss": 0.1669601321220398, "step": 2270 }, { "epoch": 0.060058675745712095, "grad_norm": 3.915579319000244, "learning_rate": 1.9952742018581634e-05, "loss": 0.17573187351226807, "step": 2280 }, { "epoch": 0.06032209099021083, "grad_norm": 3.446171760559082, "learning_rate": 1.9951909962905956e-05, "loss": 0.1647792100906372, "step": 2290 }, { "epoch": 0.06058550623470957, "grad_norm": 5.727437973022461, "learning_rate": 1.9951070663800843e-05, "loss": 0.16824188232421874, "step": 2300 }, { "epoch": 0.060848921479208305, "grad_norm": 2.199225902557373, "learning_rate": 1.995022412187718e-05, "loss": 0.1728174328804016, "step": 2310 }, { "epoch": 0.06111233672370704, "grad_norm": 3.6737754344940186, "learning_rate": 1.9949370337751105e-05, "loss": 0.18129185438156128, "step": 2320 }, { "epoch": 0.06137575196820578, "grad_norm": 3.5838427543640137, "learning_rate": 1.994850931204405e-05, "loss": 0.17500871419906616, "step": 2330 }, { "epoch": 0.061639167212704515, "grad_norm": 6.6596503257751465, "learning_rate": 1.99476410453827e-05, "loss": 0.18103907108306885, "step": 2340 }, { "epoch": 0.06190258245720325, "grad_norm": 4.493770599365234, "learning_rate": 1.994676553839902e-05, "loss": 0.18336416482925416, "step": 2350 }, { "epoch": 0.06216599770170199, "grad_norm": 2.929215431213379, "learning_rate": 1.9945882791730246e-05, "loss": 0.1864294648170471, "step": 2360 }, { "epoch": 0.06242941294620073, "grad_norm": 4.521968841552734, "learning_rate": 1.994499280601887e-05, "loss": 0.1677119493484497, "step": 2370 }, { "epoch": 0.06269282819069946, "grad_norm": 4.141523838043213, "learning_rate": 1.994409558191267e-05, "loss": 0.16287223100662232, "step": 2380 }, { "epoch": 0.0629562434351982, "grad_norm": 6.123220920562744, "learning_rate": 1.9943191120064672e-05, "loss": 0.18463417291641235, "step": 2390 }, { "epoch": 0.06321965867969694, "grad_norm": 3.535463571548462, "learning_rate": 1.99422794211332e-05, "loss": 0.16414353847503663, "step": 2400 }, { "epoch": 0.06348307392419568, "grad_norm": 1.7787047624588013, "learning_rate": 1.9941360485781814e-05, "loss": 0.17881357669830322, "step": 2410 }, { "epoch": 0.06374648916869441, "grad_norm": 1.5359500646591187, "learning_rate": 1.9940434314679363e-05, "loss": 0.17352229356765747, "step": 2420 }, { "epoch": 0.06400990441319315, "grad_norm": 2.6052448749542236, "learning_rate": 1.9939500908499947e-05, "loss": 0.1704418659210205, "step": 2430 }, { "epoch": 0.0642733196576919, "grad_norm": 8.108606338500977, "learning_rate": 1.9938560267922948e-05, "loss": 0.1841930627822876, "step": 2440 }, { "epoch": 0.06453673490219063, "grad_norm": 2.279193639755249, "learning_rate": 1.9937612393632995e-05, "loss": 0.1566016435623169, "step": 2450 }, { "epoch": 0.06480015014668937, "grad_norm": 2.2991490364074707, "learning_rate": 1.9936657286320003e-05, "loss": 0.15794761180877687, "step": 2460 }, { "epoch": 0.0650635653911881, "grad_norm": 2.228835105895996, "learning_rate": 1.993569494667913e-05, "loss": 0.17371480464935302, "step": 2470 }, { "epoch": 0.06532698063568684, "grad_norm": 3.9322290420532227, "learning_rate": 1.9934725375410812e-05, "loss": 0.18611712455749513, "step": 2480 }, { "epoch": 0.06559039588018557, "grad_norm": 2.7677993774414062, "learning_rate": 1.9933748573220745e-05, "loss": 0.18283743858337403, "step": 2490 }, { "epoch": 0.06585381112468432, "grad_norm": 5.351887226104736, "learning_rate": 1.993276454081989e-05, "loss": 0.16801278591156005, "step": 2500 }, { "epoch": 0.06611722636918305, "grad_norm": 4.0779523849487305, "learning_rate": 1.9931773278924467e-05, "loss": 0.18050620555877686, "step": 2510 }, { "epoch": 0.06638064161368179, "grad_norm": 4.539202690124512, "learning_rate": 1.9930774788255957e-05, "loss": 0.18246767520904542, "step": 2520 }, { "epoch": 0.06664405685818052, "grad_norm": 2.2406067848205566, "learning_rate": 1.992976906954111e-05, "loss": 0.18431549072265624, "step": 2530 }, { "epoch": 0.06690747210267926, "grad_norm": 2.1686861515045166, "learning_rate": 1.992875612351192e-05, "loss": 0.17098768949508666, "step": 2540 }, { "epoch": 0.067170887347178, "grad_norm": 1.420542597770691, "learning_rate": 1.9927735950905663e-05, "loss": 0.1574415683746338, "step": 2550 }, { "epoch": 0.06743430259167674, "grad_norm": 1.3667410612106323, "learning_rate": 1.9926708552464865e-05, "loss": 0.18900036811828613, "step": 2560 }, { "epoch": 0.06769771783617548, "grad_norm": 6.134460926055908, "learning_rate": 1.9925673928937307e-05, "loss": 0.18500947952270508, "step": 2570 }, { "epoch": 0.06796113308067421, "grad_norm": 2.7926952838897705, "learning_rate": 1.9924632081076032e-05, "loss": 0.17736659049987794, "step": 2580 }, { "epoch": 0.06822454832517295, "grad_norm": 2.6798529624938965, "learning_rate": 1.9923583009639344e-05, "loss": 0.18802956342697144, "step": 2590 }, { "epoch": 0.06848796356967168, "grad_norm": 4.323914051055908, "learning_rate": 1.9922526715390802e-05, "loss": 0.17529017925262452, "step": 2600 }, { "epoch": 0.06875137881417043, "grad_norm": 4.830811500549316, "learning_rate": 1.9921463199099222e-05, "loss": 0.1764782428741455, "step": 2610 }, { "epoch": 0.06901479405866916, "grad_norm": 2.899869918823242, "learning_rate": 1.992039246153868e-05, "loss": 0.18186357021331787, "step": 2620 }, { "epoch": 0.0692782093031679, "grad_norm": 4.909945011138916, "learning_rate": 1.99193145034885e-05, "loss": 0.1656233549118042, "step": 2630 }, { "epoch": 0.06954162454766663, "grad_norm": 2.1321394443511963, "learning_rate": 1.9918229325733267e-05, "loss": 0.1680976629257202, "step": 2640 }, { "epoch": 0.06980503979216537, "grad_norm": 1.6144006252288818, "learning_rate": 1.9917136929062824e-05, "loss": 0.16645102500915526, "step": 2650 }, { "epoch": 0.0700684550366641, "grad_norm": 4.522813320159912, "learning_rate": 1.991603731427226e-05, "loss": 0.16991182565689086, "step": 2660 }, { "epoch": 0.07033187028116285, "grad_norm": 6.199846267700195, "learning_rate": 1.9914930482161928e-05, "loss": 0.17128931283950805, "step": 2670 }, { "epoch": 0.07059528552566158, "grad_norm": 2.9292664527893066, "learning_rate": 1.9913816433537425e-05, "loss": 0.1582548975944519, "step": 2680 }, { "epoch": 0.07085870077016032, "grad_norm": 2.9178993701934814, "learning_rate": 1.99126951692096e-05, "loss": 0.1775449275970459, "step": 2690 }, { "epoch": 0.07112211601465906, "grad_norm": 2.847787618637085, "learning_rate": 1.991156668999456e-05, "loss": 0.1818000555038452, "step": 2700 }, { "epoch": 0.0713855312591578, "grad_norm": 1.8268945217132568, "learning_rate": 1.9910430996713668e-05, "loss": 0.16643352508544923, "step": 2710 }, { "epoch": 0.07164894650365654, "grad_norm": 3.8446929454803467, "learning_rate": 1.9909288090193518e-05, "loss": 0.18127477169036865, "step": 2720 }, { "epoch": 0.07191236174815527, "grad_norm": 2.3423750400543213, "learning_rate": 1.9908137971265977e-05, "loss": 0.1892937660217285, "step": 2730 }, { "epoch": 0.07217577699265401, "grad_norm": 2.959515333175659, "learning_rate": 1.9906980640768146e-05, "loss": 0.17546026706695556, "step": 2740 }, { "epoch": 0.07243919223715274, "grad_norm": 5.454360485076904, "learning_rate": 1.9905816099542382e-05, "loss": 0.17557764053344727, "step": 2750 }, { "epoch": 0.07270260748165149, "grad_norm": 3.9013335704803467, "learning_rate": 1.9904644348436282e-05, "loss": 0.17549136877059937, "step": 2760 }, { "epoch": 0.07296602272615021, "grad_norm": 2.8568930625915527, "learning_rate": 1.990346538830271e-05, "loss": 0.16762638092041016, "step": 2770 }, { "epoch": 0.07322943797064896, "grad_norm": 2.865736484527588, "learning_rate": 1.9902279219999752e-05, "loss": 0.16685889959335326, "step": 2780 }, { "epoch": 0.07349285321514769, "grad_norm": 4.987081527709961, "learning_rate": 1.9901085844390757e-05, "loss": 0.17137544155120848, "step": 2790 }, { "epoch": 0.07375626845964643, "grad_norm": 6.112964630126953, "learning_rate": 1.9899885262344316e-05, "loss": 0.17832401990890503, "step": 2800 }, { "epoch": 0.07401968370414516, "grad_norm": 2.9379589557647705, "learning_rate": 1.9898677474734266e-05, "loss": 0.17404618263244628, "step": 2810 }, { "epoch": 0.0742830989486439, "grad_norm": 3.033299684524536, "learning_rate": 1.9897462482439684e-05, "loss": 0.16928070783615112, "step": 2820 }, { "epoch": 0.07454651419314265, "grad_norm": 5.922482967376709, "learning_rate": 1.9896240286344892e-05, "loss": 0.18840209245681763, "step": 2830 }, { "epoch": 0.07480992943764138, "grad_norm": 3.625156879425049, "learning_rate": 1.989501088733946e-05, "loss": 0.1758962631225586, "step": 2840 }, { "epoch": 0.07507334468214012, "grad_norm": 2.5976529121398926, "learning_rate": 1.9893774286318198e-05, "loss": 0.17367956638336182, "step": 2850 }, { "epoch": 0.07533675992663885, "grad_norm": 9.213865280151367, "learning_rate": 1.9892530484181155e-05, "loss": 0.18133424520492553, "step": 2860 }, { "epoch": 0.0756001751711376, "grad_norm": 4.756237983703613, "learning_rate": 1.989127948183363e-05, "loss": 0.17546343803405762, "step": 2870 }, { "epoch": 0.07586359041563633, "grad_norm": 4.869169235229492, "learning_rate": 1.9890021280186148e-05, "loss": 0.19019633531570435, "step": 2880 }, { "epoch": 0.07612700566013507, "grad_norm": 3.024535894393921, "learning_rate": 1.9888755880154485e-05, "loss": 0.16373116970062257, "step": 2890 }, { "epoch": 0.0763904209046338, "grad_norm": 5.006529808044434, "learning_rate": 1.988748328265966e-05, "loss": 0.15127161741256714, "step": 2900 }, { "epoch": 0.07665383614913254, "grad_norm": 3.3394505977630615, "learning_rate": 1.9886203488627918e-05, "loss": 0.17995487451553344, "step": 2910 }, { "epoch": 0.07691725139363127, "grad_norm": 4.2614521980285645, "learning_rate": 1.988491649899075e-05, "loss": 0.17964822053909302, "step": 2920 }, { "epoch": 0.07718066663813002, "grad_norm": 5.49972677230835, "learning_rate": 1.988362231468488e-05, "loss": 0.16522576808929443, "step": 2930 }, { "epoch": 0.07744408188262875, "grad_norm": 1.8651714324951172, "learning_rate": 1.9882320936652278e-05, "loss": 0.1812723159790039, "step": 2940 }, { "epoch": 0.07770749712712749, "grad_norm": 3.4847335815429688, "learning_rate": 1.9881012365840138e-05, "loss": 0.16962571144104005, "step": 2950 }, { "epoch": 0.07797091237162623, "grad_norm": 2.5620877742767334, "learning_rate": 1.987969660320089e-05, "loss": 0.17313063144683838, "step": 2960 }, { "epoch": 0.07823432761612496, "grad_norm": 2.1615424156188965, "learning_rate": 1.9878373649692213e-05, "loss": 0.1855088710784912, "step": 2970 }, { "epoch": 0.0784977428606237, "grad_norm": 1.7967240810394287, "learning_rate": 1.9877043506277004e-05, "loss": 0.18424060344696044, "step": 2980 }, { "epoch": 0.07876115810512244, "grad_norm": 4.547329902648926, "learning_rate": 1.9875706173923396e-05, "loss": 0.1721693754196167, "step": 2990 }, { "epoch": 0.07902457334962118, "grad_norm": 5.679002285003662, "learning_rate": 1.9874361653604765e-05, "loss": 0.16726884841918946, "step": 3000 }, { "epoch": 0.07928798859411991, "grad_norm": 2.9822731018066406, "learning_rate": 1.9873009946299702e-05, "loss": 0.1833507537841797, "step": 3010 }, { "epoch": 0.07955140383861865, "grad_norm": 5.097931385040283, "learning_rate": 1.9871651052992046e-05, "loss": 0.16194227933883668, "step": 3020 }, { "epoch": 0.07981481908311738, "grad_norm": 4.194357395172119, "learning_rate": 1.987028497467085e-05, "loss": 0.1554744005203247, "step": 3030 }, { "epoch": 0.08007823432761613, "grad_norm": 3.353128671646118, "learning_rate": 1.986891171233041e-05, "loss": 0.1627996802330017, "step": 3040 }, { "epoch": 0.08034164957211486, "grad_norm": 5.690291404724121, "learning_rate": 1.9867531266970247e-05, "loss": 0.17662999629974366, "step": 3050 }, { "epoch": 0.0806050648166136, "grad_norm": 1.9672770500183105, "learning_rate": 1.9866143639595107e-05, "loss": 0.18338868618011475, "step": 3060 }, { "epoch": 0.08086848006111233, "grad_norm": 2.903369188308716, "learning_rate": 1.9864748831214964e-05, "loss": 0.15977989435195922, "step": 3070 }, { "epoch": 0.08113189530561107, "grad_norm": 3.179340362548828, "learning_rate": 1.986334684284502e-05, "loss": 0.18550510406494142, "step": 3080 }, { "epoch": 0.08139531055010982, "grad_norm": 5.171327590942383, "learning_rate": 1.98619376755057e-05, "loss": 0.17571659088134767, "step": 3090 }, { "epoch": 0.08165872579460855, "grad_norm": 1.9005030393600464, "learning_rate": 1.986052133022266e-05, "loss": 0.1921723484992981, "step": 3100 }, { "epoch": 0.08192214103910729, "grad_norm": 4.878815174102783, "learning_rate": 1.985909780802678e-05, "loss": 0.16665189266204833, "step": 3110 }, { "epoch": 0.08218555628360602, "grad_norm": 2.6174368858337402, "learning_rate": 1.985766710995416e-05, "loss": 0.16968307495117188, "step": 3120 }, { "epoch": 0.08244897152810476, "grad_norm": 7.275308609008789, "learning_rate": 1.9856229237046115e-05, "loss": 0.1752866268157959, "step": 3130 }, { "epoch": 0.0827123867726035, "grad_norm": 3.9535911083221436, "learning_rate": 1.9854784190349203e-05, "loss": 0.1702702283859253, "step": 3140 }, { "epoch": 0.08297580201710224, "grad_norm": 5.441196918487549, "learning_rate": 1.985333197091518e-05, "loss": 0.17779067754745484, "step": 3150 }, { "epoch": 0.08323921726160097, "grad_norm": 4.558765888214111, "learning_rate": 1.9851872579801044e-05, "loss": 0.1699903726577759, "step": 3160 }, { "epoch": 0.08350263250609971, "grad_norm": 1.5481666326522827, "learning_rate": 1.9850406018068997e-05, "loss": 0.16395806074142455, "step": 3170 }, { "epoch": 0.08376604775059844, "grad_norm": 5.0610527992248535, "learning_rate": 1.984893228678647e-05, "loss": 0.1590022325515747, "step": 3180 }, { "epoch": 0.08402946299509718, "grad_norm": 3.863367795944214, "learning_rate": 1.9847451387026103e-05, "loss": 0.1698500394821167, "step": 3190 }, { "epoch": 0.08429287823959591, "grad_norm": 4.3859052658081055, "learning_rate": 1.9845963319865758e-05, "loss": 0.1835271716117859, "step": 3200 }, { "epoch": 0.08455629348409466, "grad_norm": 2.7952327728271484, "learning_rate": 1.9844468086388523e-05, "loss": 0.16524484157562255, "step": 3210 }, { "epoch": 0.0848197087285934, "grad_norm": 5.930577754974365, "learning_rate": 1.9842965687682686e-05, "loss": 0.16260709762573242, "step": 3220 }, { "epoch": 0.08508312397309213, "grad_norm": 2.7276506423950195, "learning_rate": 1.9841456124841763e-05, "loss": 0.1762959837913513, "step": 3230 }, { "epoch": 0.08534653921759087, "grad_norm": 3.3596677780151367, "learning_rate": 1.983993939896447e-05, "loss": 0.16360392570495605, "step": 3240 }, { "epoch": 0.0856099544620896, "grad_norm": 3.4063642024993896, "learning_rate": 1.9838415511154755e-05, "loss": 0.18501075506210327, "step": 3250 }, { "epoch": 0.08587336970658835, "grad_norm": 3.643169403076172, "learning_rate": 1.9836884462521768e-05, "loss": 0.1628565549850464, "step": 3260 }, { "epoch": 0.08613678495108708, "grad_norm": 2.280477285385132, "learning_rate": 1.9835346254179865e-05, "loss": 0.17599490880966187, "step": 3270 }, { "epoch": 0.08640020019558582, "grad_norm": 2.0981392860412598, "learning_rate": 1.9833800887248628e-05, "loss": 0.17700258493423462, "step": 3280 }, { "epoch": 0.08666361544008455, "grad_norm": 5.265347480773926, "learning_rate": 1.983224836285284e-05, "loss": 0.17288534641265868, "step": 3290 }, { "epoch": 0.0869270306845833, "grad_norm": 6.0181427001953125, "learning_rate": 1.9830688682122494e-05, "loss": 0.1847670316696167, "step": 3300 }, { "epoch": 0.08719044592908202, "grad_norm": 4.942960262298584, "learning_rate": 1.982912184619279e-05, "loss": 0.1714083194732666, "step": 3310 }, { "epoch": 0.08745386117358077, "grad_norm": 7.8191375732421875, "learning_rate": 1.9827547856204147e-05, "loss": 0.16406528949737548, "step": 3320 }, { "epoch": 0.0877172764180795, "grad_norm": 3.3880088329315186, "learning_rate": 1.9825966713302172e-05, "loss": 0.17072827816009523, "step": 3330 }, { "epoch": 0.08798069166257824, "grad_norm": 4.340854644775391, "learning_rate": 1.98243784186377e-05, "loss": 0.17088290452957153, "step": 3340 }, { "epoch": 0.08824410690707699, "grad_norm": 3.746156692504883, "learning_rate": 1.982278297336675e-05, "loss": 0.17421211004257203, "step": 3350 }, { "epoch": 0.08850752215157572, "grad_norm": 1.7947607040405273, "learning_rate": 1.9821180378650563e-05, "loss": 0.17688140869140626, "step": 3360 }, { "epoch": 0.08877093739607446, "grad_norm": 3.6648716926574707, "learning_rate": 1.9819570635655577e-05, "loss": 0.1721170425415039, "step": 3370 }, { "epoch": 0.08903435264057319, "grad_norm": 5.80394172668457, "learning_rate": 1.981795374555343e-05, "loss": 0.16205997467041017, "step": 3380 }, { "epoch": 0.08929776788507193, "grad_norm": 2.763179302215576, "learning_rate": 1.981632970952096e-05, "loss": 0.16646959781646728, "step": 3390 }, { "epoch": 0.08956118312957066, "grad_norm": 1.4514175653457642, "learning_rate": 1.9814698528740222e-05, "loss": 0.17294921875, "step": 3400 }, { "epoch": 0.0898245983740694, "grad_norm": 4.321579933166504, "learning_rate": 1.9813060204398448e-05, "loss": 0.17119885683059693, "step": 3410 }, { "epoch": 0.09008801361856814, "grad_norm": 2.7648565769195557, "learning_rate": 1.9811414737688088e-05, "loss": 0.1738181233406067, "step": 3420 }, { "epoch": 0.09035142886306688, "grad_norm": 2.2346138954162598, "learning_rate": 1.9809762129806777e-05, "loss": 0.1727881908416748, "step": 3430 }, { "epoch": 0.09061484410756561, "grad_norm": 3.526754379272461, "learning_rate": 1.9808102381957364e-05, "loss": 0.1788222908973694, "step": 3440 }, { "epoch": 0.09087825935206435, "grad_norm": 3.5606818199157715, "learning_rate": 1.980643549534788e-05, "loss": 0.16893284320831298, "step": 3450 }, { "epoch": 0.0911416745965631, "grad_norm": 1.602480411529541, "learning_rate": 1.980476147119155e-05, "loss": 0.15513352155685425, "step": 3460 }, { "epoch": 0.09140508984106183, "grad_norm": 1.276413917541504, "learning_rate": 1.980308031070682e-05, "loss": 0.16534485816955566, "step": 3470 }, { "epoch": 0.09166850508556057, "grad_norm": 2.8959596157073975, "learning_rate": 1.9801392015117286e-05, "loss": 0.1872309446334839, "step": 3480 }, { "epoch": 0.0919319203300593, "grad_norm": 3.5731759071350098, "learning_rate": 1.9799696585651782e-05, "loss": 0.15785133838653564, "step": 3490 }, { "epoch": 0.09219533557455804, "grad_norm": 1.6595618724822998, "learning_rate": 1.9797994023544307e-05, "loss": 0.18157401084899902, "step": 3500 }, { "epoch": 0.09245875081905677, "grad_norm": 1.988336205482483, "learning_rate": 1.979628433003406e-05, "loss": 0.17100307941436768, "step": 3510 }, { "epoch": 0.09272216606355552, "grad_norm": 1.5123616456985474, "learning_rate": 1.9794567506365432e-05, "loss": 0.1713883876800537, "step": 3520 }, { "epoch": 0.09298558130805425, "grad_norm": 3.167529582977295, "learning_rate": 1.9792843553788e-05, "loss": 0.16416552066802978, "step": 3530 }, { "epoch": 0.09324899655255299, "grad_norm": 3.227069139480591, "learning_rate": 1.9791112473556528e-05, "loss": 0.17864915132522582, "step": 3540 }, { "epoch": 0.09351241179705172, "grad_norm": 1.5326520204544067, "learning_rate": 1.978937426693097e-05, "loss": 0.17092440128326417, "step": 3550 }, { "epoch": 0.09377582704155046, "grad_norm": 2.6923580169677734, "learning_rate": 1.9787628935176476e-05, "loss": 0.17692804336547852, "step": 3560 }, { "epoch": 0.0940392422860492, "grad_norm": 4.238792419433594, "learning_rate": 1.9785876479563367e-05, "loss": 0.17451252937316894, "step": 3570 }, { "epoch": 0.09430265753054794, "grad_norm": 2.0466699600219727, "learning_rate": 1.978411690136716e-05, "loss": 0.17538204193115234, "step": 3580 }, { "epoch": 0.09456607277504668, "grad_norm": 2.297478437423706, "learning_rate": 1.9782350201868547e-05, "loss": 0.16729767322540284, "step": 3590 }, { "epoch": 0.09482948801954541, "grad_norm": 2.931528329849243, "learning_rate": 1.978057638235341e-05, "loss": 0.16791937351226807, "step": 3600 }, { "epoch": 0.09509290326404415, "grad_norm": 3.0888381004333496, "learning_rate": 1.9778795444112812e-05, "loss": 0.16829466819763184, "step": 3610 }, { "epoch": 0.09535631850854288, "grad_norm": 7.0721964836120605, "learning_rate": 1.9777007388443e-05, "loss": 0.16711490154266356, "step": 3620 }, { "epoch": 0.09561973375304163, "grad_norm": 2.041065216064453, "learning_rate": 1.9775212216645393e-05, "loss": 0.1552499771118164, "step": 3630 }, { "epoch": 0.09588314899754036, "grad_norm": 0.9699086546897888, "learning_rate": 1.977340993002659e-05, "loss": 0.18069822788238527, "step": 3640 }, { "epoch": 0.0961465642420391, "grad_norm": 4.110236644744873, "learning_rate": 1.977160052989839e-05, "loss": 0.17651093006134033, "step": 3650 }, { "epoch": 0.09640997948653783, "grad_norm": 1.7598530054092407, "learning_rate": 1.9769784017577734e-05, "loss": 0.16150251626968384, "step": 3660 }, { "epoch": 0.09667339473103657, "grad_norm": 9.13883113861084, "learning_rate": 1.976796039438677e-05, "loss": 0.1856318235397339, "step": 3670 }, { "epoch": 0.0969368099755353, "grad_norm": 6.139298439025879, "learning_rate": 1.9766129661652803e-05, "loss": 0.1874711036682129, "step": 3680 }, { "epoch": 0.09720022522003405, "grad_norm": 4.398290634155273, "learning_rate": 1.9764291820708324e-05, "loss": 0.18312535285949708, "step": 3690 }, { "epoch": 0.09746364046453278, "grad_norm": 3.7100791931152344, "learning_rate": 1.9762446872890984e-05, "loss": 0.1663208246231079, "step": 3700 }, { "epoch": 0.09772705570903152, "grad_norm": 3.21345853805542, "learning_rate": 1.976059481954363e-05, "loss": 0.17641947269439698, "step": 3710 }, { "epoch": 0.09799047095353026, "grad_norm": 1.9989371299743652, "learning_rate": 1.9758735662014253e-05, "loss": 0.16351261138916015, "step": 3720 }, { "epoch": 0.098253886198029, "grad_norm": 3.3005268573760986, "learning_rate": 1.975686940165604e-05, "loss": 0.1614906072616577, "step": 3730 }, { "epoch": 0.09851730144252774, "grad_norm": 2.5145790576934814, "learning_rate": 1.9754996039827322e-05, "loss": 0.15959972143173218, "step": 3740 }, { "epoch": 0.09878071668702647, "grad_norm": 2.6582067012786865, "learning_rate": 1.9753115577891623e-05, "loss": 0.16446216106414796, "step": 3750 }, { "epoch": 0.09904413193152521, "grad_norm": 1.8008614778518677, "learning_rate": 1.9751228017217623e-05, "loss": 0.15692660808563233, "step": 3760 }, { "epoch": 0.09930754717602394, "grad_norm": 110.00682067871094, "learning_rate": 1.9749333359179166e-05, "loss": 1.9137077331542969, "step": 3770 }, { "epoch": 0.09957096242052269, "grad_norm": 2.482398271560669, "learning_rate": 1.974743160515527e-05, "loss": 0.37720460891723634, "step": 3780 }, { "epoch": 0.09983437766502141, "grad_norm": 4.837782382965088, "learning_rate": 1.974552275653011e-05, "loss": 0.23142449855804442, "step": 3790 }, { "epoch": 0.10009779290952016, "grad_norm": 2.732879400253296, "learning_rate": 1.974360681469303e-05, "loss": 0.2180330514907837, "step": 3800 }, { "epoch": 0.10036120815401889, "grad_norm": 6.985912799835205, "learning_rate": 1.9741683781038538e-05, "loss": 0.20520420074462892, "step": 3810 }, { "epoch": 0.10062462339851763, "grad_norm": 4.000983238220215, "learning_rate": 1.9739753656966294e-05, "loss": 0.19291975498199462, "step": 3820 }, { "epoch": 0.10088803864301636, "grad_norm": 7.733485221862793, "learning_rate": 1.9737816443881134e-05, "loss": 0.21396305561065673, "step": 3830 }, { "epoch": 0.1011514538875151, "grad_norm": 6.139857292175293, "learning_rate": 1.973587214319304e-05, "loss": 0.1897595167160034, "step": 3840 }, { "epoch": 0.10141486913201385, "grad_norm": 2.6146812438964844, "learning_rate": 1.9733920756317157e-05, "loss": 0.19322962760925294, "step": 3850 }, { "epoch": 0.10167828437651258, "grad_norm": 1.8416882753372192, "learning_rate": 1.973196228467379e-05, "loss": 0.189955735206604, "step": 3860 }, { "epoch": 0.10194169962101132, "grad_norm": 2.378429651260376, "learning_rate": 1.9729996729688403e-05, "loss": 0.19388871192932128, "step": 3870 }, { "epoch": 0.10220511486551005, "grad_norm": 2.026815176010132, "learning_rate": 1.9728024092791606e-05, "loss": 0.18580758571624756, "step": 3880 }, { "epoch": 0.1024685301100088, "grad_norm": 1.9626963138580322, "learning_rate": 1.9726044375419172e-05, "loss": 0.19047777652740477, "step": 3890 }, { "epoch": 0.10273194535450753, "grad_norm": 2.755002498626709, "learning_rate": 1.9724057579012026e-05, "loss": 0.19273111820220948, "step": 3900 }, { "epoch": 0.10299536059900627, "grad_norm": 3.489264488220215, "learning_rate": 1.972206370501624e-05, "loss": 0.1933247685432434, "step": 3910 }, { "epoch": 0.103258775843505, "grad_norm": 2.2937896251678467, "learning_rate": 1.9720062754883048e-05, "loss": 0.19142223596572877, "step": 3920 }, { "epoch": 0.10352219108800374, "grad_norm": 1.2544584274291992, "learning_rate": 1.971805473006882e-05, "loss": 0.18210973739624023, "step": 3930 }, { "epoch": 0.10378560633250247, "grad_norm": 4.163583278656006, "learning_rate": 1.9716039632035086e-05, "loss": 0.18169561624526978, "step": 3940 }, { "epoch": 0.10404902157700122, "grad_norm": 4.032989978790283, "learning_rate": 1.9714017462248523e-05, "loss": 0.20846574306488036, "step": 3950 }, { "epoch": 0.10431243682149995, "grad_norm": 1.4758186340332031, "learning_rate": 1.9711988222180958e-05, "loss": 0.20160129070281982, "step": 3960 }, { "epoch": 0.10457585206599869, "grad_norm": 3.896413803100586, "learning_rate": 1.9709951913309347e-05, "loss": 0.18543100357055664, "step": 3970 }, { "epoch": 0.10483926731049743, "grad_norm": 1.6091901063919067, "learning_rate": 1.9707908537115816e-05, "loss": 0.1812201738357544, "step": 3980 }, { "epoch": 0.10510268255499616, "grad_norm": 5.76597785949707, "learning_rate": 1.9705858095087613e-05, "loss": 0.19178397655487062, "step": 3990 }, { "epoch": 0.1053660977994949, "grad_norm": 2.2422633171081543, "learning_rate": 1.9703800588717144e-05, "loss": 0.17241008281707765, "step": 4000 }, { "epoch": 0.10562951304399364, "grad_norm": 2.0491421222686768, "learning_rate": 1.9701736019501948e-05, "loss": 0.20043487548828126, "step": 4010 }, { "epoch": 0.10589292828849238, "grad_norm": 5.606991767883301, "learning_rate": 1.969966438894471e-05, "loss": 0.18321049213409424, "step": 4020 }, { "epoch": 0.10615634353299111, "grad_norm": 3.915343761444092, "learning_rate": 1.969758569855325e-05, "loss": 0.1860215425491333, "step": 4030 }, { "epoch": 0.10641975877748985, "grad_norm": 2.6297240257263184, "learning_rate": 1.969549994984053e-05, "loss": 0.1866211175918579, "step": 4040 }, { "epoch": 0.10668317402198858, "grad_norm": 4.614986896514893, "learning_rate": 1.9693407144324642e-05, "loss": 0.16559375524520875, "step": 4050 }, { "epoch": 0.10694658926648733, "grad_norm": 4.838974952697754, "learning_rate": 1.9691307283528834e-05, "loss": 0.19147591590881347, "step": 4060 }, { "epoch": 0.10721000451098606, "grad_norm": 5.145614147186279, "learning_rate": 1.968920036898146e-05, "loss": 0.18482613563537598, "step": 4070 }, { "epoch": 0.1074734197554848, "grad_norm": 2.7278549671173096, "learning_rate": 1.9687086402216033e-05, "loss": 0.17181750535964965, "step": 4080 }, { "epoch": 0.10773683499998353, "grad_norm": 1.9629753828048706, "learning_rate": 1.968496538477118e-05, "loss": 0.17128127813339233, "step": 4090 }, { "epoch": 0.10800025024448227, "grad_norm": 4.499478816986084, "learning_rate": 1.9682837318190683e-05, "loss": 0.20414888858795166, "step": 4100 }, { "epoch": 0.10826366548898102, "grad_norm": 2.5342025756835938, "learning_rate": 1.968070220402343e-05, "loss": 0.15767689943313598, "step": 4110 }, { "epoch": 0.10852708073347975, "grad_norm": 1.578596591949463, "learning_rate": 1.967856004382345e-05, "loss": 0.18944025039672852, "step": 4120 }, { "epoch": 0.10879049597797849, "grad_norm": 4.953753471374512, "learning_rate": 1.96764108391499e-05, "loss": 0.1833498477935791, "step": 4130 }, { "epoch": 0.10905391122247722, "grad_norm": 8.799403190612793, "learning_rate": 1.967425459156707e-05, "loss": 0.19042227268218995, "step": 4140 }, { "epoch": 0.10931732646697596, "grad_norm": 3.286208391189575, "learning_rate": 1.9672091302644363e-05, "loss": 0.18313281536102294, "step": 4150 }, { "epoch": 0.1095807417114747, "grad_norm": 5.790173530578613, "learning_rate": 1.966992097395632e-05, "loss": 0.1765434503555298, "step": 4160 }, { "epoch": 0.10984415695597344, "grad_norm": 1.271097183227539, "learning_rate": 1.9667743607082593e-05, "loss": 0.17103487253189087, "step": 4170 }, { "epoch": 0.11010757220047217, "grad_norm": 2.7132771015167236, "learning_rate": 1.9665559203607965e-05, "loss": 0.17042334079742433, "step": 4180 }, { "epoch": 0.11037098744497091, "grad_norm": 3.366156816482544, "learning_rate": 1.9663367765122346e-05, "loss": 0.1780339479446411, "step": 4190 }, { "epoch": 0.11063440268946964, "grad_norm": 1.0733368396759033, "learning_rate": 1.966116929322075e-05, "loss": 0.18273066282272338, "step": 4200 }, { "epoch": 0.11089781793396838, "grad_norm": 2.2644588947296143, "learning_rate": 1.9658963789503328e-05, "loss": 0.16600320339202881, "step": 4210 }, { "epoch": 0.11116123317846711, "grad_norm": 1.7345768213272095, "learning_rate": 1.9656751255575336e-05, "loss": 0.15617727041244506, "step": 4220 }, { "epoch": 0.11142464842296586, "grad_norm": 3.5599730014801025, "learning_rate": 1.965453169304715e-05, "loss": 0.17674144506454467, "step": 4230 }, { "epoch": 0.1116880636674646, "grad_norm": 5.5997633934021, "learning_rate": 1.9652305103534265e-05, "loss": 0.17802841663360597, "step": 4240 }, { "epoch": 0.11195147891196333, "grad_norm": 2.018476963043213, "learning_rate": 1.9650071488657293e-05, "loss": 0.16660888195037843, "step": 4250 }, { "epoch": 0.11221489415646207, "grad_norm": 1.9487333297729492, "learning_rate": 1.964783085004195e-05, "loss": 0.17661192417144775, "step": 4260 }, { "epoch": 0.1124783094009608, "grad_norm": 1.9945648908615112, "learning_rate": 1.9645583189319074e-05, "loss": 0.1813114643096924, "step": 4270 }, { "epoch": 0.11274172464545955, "grad_norm": 4.824276924133301, "learning_rate": 1.9643328508124605e-05, "loss": 0.1831031084060669, "step": 4280 }, { "epoch": 0.11300513988995828, "grad_norm": 3.3287723064422607, "learning_rate": 1.9641066808099597e-05, "loss": 0.18200147151947021, "step": 4290 }, { "epoch": 0.11326855513445702, "grad_norm": 1.4271423816680908, "learning_rate": 1.963879809089022e-05, "loss": 0.15896421670913696, "step": 4300 }, { "epoch": 0.11353197037895575, "grad_norm": 3.3298463821411133, "learning_rate": 1.9636522358147737e-05, "loss": 0.18188525438308717, "step": 4310 }, { "epoch": 0.1137953856234545, "grad_norm": 2.732421875, "learning_rate": 1.9634239611528527e-05, "loss": 0.17588669061660767, "step": 4320 }, { "epoch": 0.11405880086795323, "grad_norm": 1.9230146408081055, "learning_rate": 1.9631949852694073e-05, "loss": 0.17792567014694213, "step": 4330 }, { "epoch": 0.11432221611245197, "grad_norm": 3.895253896713257, "learning_rate": 1.9629653083310956e-05, "loss": 0.18725898265838622, "step": 4340 }, { "epoch": 0.1145856313569507, "grad_norm": 3.107570171356201, "learning_rate": 1.9627349305050867e-05, "loss": 0.18585529327392578, "step": 4350 }, { "epoch": 0.11484904660144944, "grad_norm": 3.534287452697754, "learning_rate": 1.9625038519590596e-05, "loss": 0.1772523045539856, "step": 4360 }, { "epoch": 0.11511246184594819, "grad_norm": 7.265838623046875, "learning_rate": 1.962272072861203e-05, "loss": 0.18522530794143677, "step": 4370 }, { "epoch": 0.11537587709044692, "grad_norm": 3.1288363933563232, "learning_rate": 1.9620395933802158e-05, "loss": 0.185687255859375, "step": 4380 }, { "epoch": 0.11563929233494566, "grad_norm": 2.0562987327575684, "learning_rate": 1.9618064136853066e-05, "loss": 0.1608814239501953, "step": 4390 }, { "epoch": 0.11590270757944439, "grad_norm": 2.9268815517425537, "learning_rate": 1.9615725339461936e-05, "loss": 0.17945938110351561, "step": 4400 }, { "epoch": 0.11616612282394313, "grad_norm": 4.6346259117126465, "learning_rate": 1.961337954333105e-05, "loss": 0.17191479206085206, "step": 4410 }, { "epoch": 0.11642953806844186, "grad_norm": 1.7354161739349365, "learning_rate": 1.961102675016777e-05, "loss": 0.1746042251586914, "step": 4420 }, { "epoch": 0.1166929533129406, "grad_norm": 4.444930076599121, "learning_rate": 1.960866696168457e-05, "loss": 0.16373783349990845, "step": 4430 }, { "epoch": 0.11695636855743934, "grad_norm": 3.8222973346710205, "learning_rate": 1.9606300179599004e-05, "loss": 0.1643228769302368, "step": 4440 }, { "epoch": 0.11721978380193808, "grad_norm": 2.6729631423950195, "learning_rate": 1.9603926405633712e-05, "loss": 0.18037084341049195, "step": 4450 }, { "epoch": 0.11748319904643681, "grad_norm": 2.298049211502075, "learning_rate": 1.9601545641516434e-05, "loss": 0.18246314525604249, "step": 4460 }, { "epoch": 0.11774661429093555, "grad_norm": 3.1019577980041504, "learning_rate": 1.9599157888979997e-05, "loss": 0.17116327285766603, "step": 4470 }, { "epoch": 0.11801002953543428, "grad_norm": 2.2143025398254395, "learning_rate": 1.95967631497623e-05, "loss": 0.16612298488616944, "step": 4480 }, { "epoch": 0.11827344477993303, "grad_norm": 3.2999043464660645, "learning_rate": 1.9594361425606348e-05, "loss": 0.17579426765441894, "step": 4490 }, { "epoch": 0.11853686002443177, "grad_norm": 1.5329158306121826, "learning_rate": 1.959195271826021e-05, "loss": 0.16602118015289308, "step": 4500 }, { "epoch": 0.1188002752689305, "grad_norm": 3.4732444286346436, "learning_rate": 1.9589537029477052e-05, "loss": 0.16440970897674562, "step": 4510 }, { "epoch": 0.11906369051342924, "grad_norm": 1.3499354124069214, "learning_rate": 1.9587114361015118e-05, "loss": 0.17354422807693481, "step": 4520 }, { "epoch": 0.11932710575792797, "grad_norm": 3.7262778282165527, "learning_rate": 1.958468471463773e-05, "loss": 0.17068202495574952, "step": 4530 }, { "epoch": 0.11959052100242672, "grad_norm": 3.4190776348114014, "learning_rate": 1.9582248092113286e-05, "loss": 0.17158420085906984, "step": 4540 }, { "epoch": 0.11985393624692545, "grad_norm": 1.834745168685913, "learning_rate": 1.9579804495215268e-05, "loss": 0.16756100654602052, "step": 4550 }, { "epoch": 0.12011735149142419, "grad_norm": 1.821163535118103, "learning_rate": 1.9577353925722228e-05, "loss": 0.17100207805633544, "step": 4560 }, { "epoch": 0.12038076673592292, "grad_norm": 1.3369837999343872, "learning_rate": 1.9574896385417803e-05, "loss": 0.17782658338546753, "step": 4570 }, { "epoch": 0.12064418198042166, "grad_norm": 1.4802422523498535, "learning_rate": 1.957243187609069e-05, "loss": 0.17409915924072267, "step": 4580 }, { "epoch": 0.1209075972249204, "grad_norm": 2.362729787826538, "learning_rate": 1.9569960399534668e-05, "loss": 0.16943047046661378, "step": 4590 }, { "epoch": 0.12117101246941914, "grad_norm": 3.2294201850891113, "learning_rate": 1.9567481957548585e-05, "loss": 0.18151278495788575, "step": 4600 }, { "epoch": 0.12143442771391787, "grad_norm": 4.897517681121826, "learning_rate": 1.9564996551936354e-05, "loss": 0.17805304527282714, "step": 4610 }, { "epoch": 0.12169784295841661, "grad_norm": 3.037275552749634, "learning_rate": 1.9562504184506966e-05, "loss": 0.16857391595840454, "step": 4620 }, { "epoch": 0.12196125820291535, "grad_norm": 3.210296630859375, "learning_rate": 1.9560004857074472e-05, "loss": 0.16889669895172119, "step": 4630 }, { "epoch": 0.12222467344741408, "grad_norm": 2.165093421936035, "learning_rate": 1.955749857145799e-05, "loss": 0.16624642610549928, "step": 4640 }, { "epoch": 0.12248808869191283, "grad_norm": 4.191974639892578, "learning_rate": 1.95549853294817e-05, "loss": 0.16806554794311523, "step": 4650 }, { "epoch": 0.12275150393641156, "grad_norm": 1.1658375263214111, "learning_rate": 1.955246513297485e-05, "loss": 0.18285493850708007, "step": 4660 }, { "epoch": 0.1230149191809103, "grad_norm": 1.9118156433105469, "learning_rate": 1.9549937983771748e-05, "loss": 0.17636752128601074, "step": 4670 }, { "epoch": 0.12327833442540903, "grad_norm": 7.300033092498779, "learning_rate": 1.954740388371176e-05, "loss": 0.16621408462524415, "step": 4680 }, { "epoch": 0.12354174966990777, "grad_norm": 1.5830326080322266, "learning_rate": 1.9544862834639316e-05, "loss": 0.1702355146408081, "step": 4690 }, { "epoch": 0.1238051649144065, "grad_norm": 2.0232884883880615, "learning_rate": 1.9542314838403907e-05, "loss": 0.16489782333374023, "step": 4700 }, { "epoch": 0.12406858015890525, "grad_norm": 3.372507095336914, "learning_rate": 1.9539759896860064e-05, "loss": 0.1780679702758789, "step": 4710 }, { "epoch": 0.12433199540340398, "grad_norm": 5.1563897132873535, "learning_rate": 1.9537198011867383e-05, "loss": 0.17880326509475708, "step": 4720 }, { "epoch": 0.12459541064790272, "grad_norm": 3.952237129211426, "learning_rate": 1.9534629185290524e-05, "loss": 0.17994096279144287, "step": 4730 }, { "epoch": 0.12485882589240146, "grad_norm": 2.2262771129608154, "learning_rate": 1.9532053418999186e-05, "loss": 0.1797703266143799, "step": 4740 }, { "epoch": 0.1251222411369002, "grad_norm": 2.9093079566955566, "learning_rate": 1.952947071486812e-05, "loss": 0.1862213134765625, "step": 4750 }, { "epoch": 0.12538565638139892, "grad_norm": 3.2234630584716797, "learning_rate": 1.952688107477713e-05, "loss": 0.15891847610473633, "step": 4760 }, { "epoch": 0.12564907162589767, "grad_norm": 2.033785343170166, "learning_rate": 1.9524284500611076e-05, "loss": 0.18495440483093262, "step": 4770 }, { "epoch": 0.1259124868703964, "grad_norm": 3.045027494430542, "learning_rate": 1.9521680994259845e-05, "loss": 0.1792832612991333, "step": 4780 }, { "epoch": 0.12617590211489516, "grad_norm": 2.236025333404541, "learning_rate": 1.9519070557618386e-05, "loss": 0.1786027431488037, "step": 4790 }, { "epoch": 0.12643931735939387, "grad_norm": 1.4814043045043945, "learning_rate": 1.9516453192586686e-05, "loss": 0.16486458778381347, "step": 4800 }, { "epoch": 0.12670273260389261, "grad_norm": 2.6634812355041504, "learning_rate": 1.951382890106978e-05, "loss": 0.16886759996414186, "step": 4810 }, { "epoch": 0.12696614784839136, "grad_norm": 2.261880874633789, "learning_rate": 1.951119768497774e-05, "loss": 0.18634244203567504, "step": 4820 }, { "epoch": 0.1272295630928901, "grad_norm": 2.0223073959350586, "learning_rate": 1.9508559546225672e-05, "loss": 0.1650442123413086, "step": 4830 }, { "epoch": 0.12749297833738882, "grad_norm": 1.1857988834381104, "learning_rate": 1.9505914486733735e-05, "loss": 0.1712963819503784, "step": 4840 }, { "epoch": 0.12775639358188756, "grad_norm": 1.1879140138626099, "learning_rate": 1.9503262508427115e-05, "loss": 0.17160667181015016, "step": 4850 }, { "epoch": 0.1280198088263863, "grad_norm": 2.979257106781006, "learning_rate": 1.9500603613236034e-05, "loss": 0.16386330127716064, "step": 4860 }, { "epoch": 0.12828322407088505, "grad_norm": 1.4215465784072876, "learning_rate": 1.9497937803095756e-05, "loss": 0.17829800844192506, "step": 4870 }, { "epoch": 0.1285466393153838, "grad_norm": 5.067112445831299, "learning_rate": 1.9495265079946565e-05, "loss": 0.16939343214035035, "step": 4880 }, { "epoch": 0.1288100545598825, "grad_norm": 2.416609525680542, "learning_rate": 1.9492585445733788e-05, "loss": 0.1801602363586426, "step": 4890 }, { "epoch": 0.12907346980438125, "grad_norm": 4.406738758087158, "learning_rate": 1.9489898902407783e-05, "loss": 0.17718329429626464, "step": 4900 }, { "epoch": 0.12933688504888, "grad_norm": 4.473239421844482, "learning_rate": 1.9487205451923925e-05, "loss": 0.17375457286834717, "step": 4910 }, { "epoch": 0.12960030029337874, "grad_norm": 6.689163684844971, "learning_rate": 1.9484505096242628e-05, "loss": 0.17908579111099243, "step": 4920 }, { "epoch": 0.12986371553787746, "grad_norm": 4.987771987915039, "learning_rate": 1.9481797837329323e-05, "loss": 0.165803861618042, "step": 4930 }, { "epoch": 0.1301271307823762, "grad_norm": 1.4698143005371094, "learning_rate": 1.9479083677154474e-05, "loss": 0.1656733512878418, "step": 4940 }, { "epoch": 0.13039054602687494, "grad_norm": 1.136271595954895, "learning_rate": 1.9476362617693564e-05, "loss": 0.1703730344772339, "step": 4950 }, { "epoch": 0.1306539612713737, "grad_norm": 1.2784903049468994, "learning_rate": 1.9473634660927092e-05, "loss": 0.18047982454299927, "step": 4960 }, { "epoch": 0.1309173765158724, "grad_norm": 5.696463584899902, "learning_rate": 1.9470899808840585e-05, "loss": 0.18195323944091796, "step": 4970 }, { "epoch": 0.13118079176037115, "grad_norm": 2.5817012786865234, "learning_rate": 1.9468158063424593e-05, "loss": 0.1836047887802124, "step": 4980 }, { "epoch": 0.1314442070048699, "grad_norm": 3.95703387260437, "learning_rate": 1.946540942667467e-05, "loss": 0.16616756916046144, "step": 4990 }, { "epoch": 0.13170762224936863, "grad_norm": 1.019754409790039, "learning_rate": 1.9462653900591393e-05, "loss": 0.17142720222473146, "step": 5000 }, { "epoch": 0.13197103749386738, "grad_norm": 2.779637098312378, "learning_rate": 1.9459891487180353e-05, "loss": 0.16821086406707764, "step": 5010 }, { "epoch": 0.1322344527383661, "grad_norm": 1.667672872543335, "learning_rate": 1.9457122188452156e-05, "loss": 0.15829744338989257, "step": 5020 }, { "epoch": 0.13249786798286484, "grad_norm": 4.807371616363525, "learning_rate": 1.9454346006422414e-05, "loss": 0.16907657384872438, "step": 5030 }, { "epoch": 0.13276128322736358, "grad_norm": 1.5375933647155762, "learning_rate": 1.9451562943111753e-05, "loss": 0.1772385597229004, "step": 5040 }, { "epoch": 0.13302469847186232, "grad_norm": 2.3086705207824707, "learning_rate": 1.944877300054581e-05, "loss": 0.1801529884338379, "step": 5050 }, { "epoch": 0.13328811371636104, "grad_norm": 3.6870744228363037, "learning_rate": 1.944597618075522e-05, "loss": 0.16671197414398192, "step": 5060 }, { "epoch": 0.13355152896085978, "grad_norm": 2.3490004539489746, "learning_rate": 1.944317248577563e-05, "loss": 0.1580794095993042, "step": 5070 }, { "epoch": 0.13381494420535853, "grad_norm": 2.4796037673950195, "learning_rate": 1.9440361917647693e-05, "loss": 0.17637031078338622, "step": 5080 }, { "epoch": 0.13407835944985727, "grad_norm": 1.2939022779464722, "learning_rate": 1.943754447841706e-05, "loss": 0.15971529483795166, "step": 5090 }, { "epoch": 0.134341774694356, "grad_norm": 3.615168809890747, "learning_rate": 1.9434720170134387e-05, "loss": 0.17199976444244386, "step": 5100 }, { "epoch": 0.13460518993885473, "grad_norm": 3.075110912322998, "learning_rate": 1.9431888994855324e-05, "loss": 0.18256351947784424, "step": 5110 }, { "epoch": 0.13486860518335347, "grad_norm": 4.568537712097168, "learning_rate": 1.9429050954640528e-05, "loss": 0.1641966462135315, "step": 5120 }, { "epoch": 0.13513202042785222, "grad_norm": 3.4684765338897705, "learning_rate": 1.942620605155564e-05, "loss": 0.16581418514251708, "step": 5130 }, { "epoch": 0.13539543567235096, "grad_norm": 1.0667462348937988, "learning_rate": 1.9423354287671307e-05, "loss": 0.175481379032135, "step": 5140 }, { "epoch": 0.13565885091684968, "grad_norm": 1.741031289100647, "learning_rate": 1.9420495665063166e-05, "loss": 0.16695902347564698, "step": 5150 }, { "epoch": 0.13592226616134842, "grad_norm": 1.2827870845794678, "learning_rate": 1.9417630185811846e-05, "loss": 0.16612763404846193, "step": 5160 }, { "epoch": 0.13618568140584716, "grad_norm": 3.3000290393829346, "learning_rate": 1.9414757852002962e-05, "loss": 0.18099100589752198, "step": 5170 }, { "epoch": 0.1364490966503459, "grad_norm": 2.0772011280059814, "learning_rate": 1.941187866572713e-05, "loss": 0.1708845615386963, "step": 5180 }, { "epoch": 0.13671251189484462, "grad_norm": 1.3789806365966797, "learning_rate": 1.940899262907994e-05, "loss": 0.17356276512145996, "step": 5190 }, { "epoch": 0.13697592713934337, "grad_norm": 1.7965563535690308, "learning_rate": 1.940609974416198e-05, "loss": 0.17241963148117065, "step": 5200 }, { "epoch": 0.1372393423838421, "grad_norm": 7.521316051483154, "learning_rate": 1.940320001307881e-05, "loss": 0.17003297805786133, "step": 5210 }, { "epoch": 0.13750275762834085, "grad_norm": 2.51668381690979, "learning_rate": 1.940029343794098e-05, "loss": 0.17872540950775145, "step": 5220 }, { "epoch": 0.13776617287283957, "grad_norm": 0.9895384311676025, "learning_rate": 1.9397380020864024e-05, "loss": 0.17525423765182496, "step": 5230 }, { "epoch": 0.13802958811733831, "grad_norm": 3.681814193725586, "learning_rate": 1.939445976396845e-05, "loss": 0.1625270128250122, "step": 5240 }, { "epoch": 0.13829300336183706, "grad_norm": 1.9991874694824219, "learning_rate": 1.939153266937975e-05, "loss": 0.1738563895225525, "step": 5250 }, { "epoch": 0.1385564186063358, "grad_norm": 1.5650824308395386, "learning_rate": 1.938859873922839e-05, "loss": 0.18037558794021608, "step": 5260 }, { "epoch": 0.13881983385083455, "grad_norm": 2.3394012451171875, "learning_rate": 1.9385657975649807e-05, "loss": 0.1757339596748352, "step": 5270 }, { "epoch": 0.13908324909533326, "grad_norm": 1.574631929397583, "learning_rate": 1.9382710380784418e-05, "loss": 0.16790418624877929, "step": 5280 }, { "epoch": 0.139346664339832, "grad_norm": 1.2656431198120117, "learning_rate": 1.937975595677761e-05, "loss": 0.1646130919456482, "step": 5290 }, { "epoch": 0.13961007958433075, "grad_norm": 2.4529192447662354, "learning_rate": 1.9376794705779735e-05, "loss": 0.1642151355743408, "step": 5300 }, { "epoch": 0.1398734948288295, "grad_norm": 2.2502613067626953, "learning_rate": 1.937382662994613e-05, "loss": 0.16865842342376708, "step": 5310 }, { "epoch": 0.1401369100733282, "grad_norm": 3.080422878265381, "learning_rate": 1.9370851731437077e-05, "loss": 0.17471015453338623, "step": 5320 }, { "epoch": 0.14040032531782695, "grad_norm": 5.600344181060791, "learning_rate": 1.9367870012417844e-05, "loss": 0.1749514937400818, "step": 5330 }, { "epoch": 0.1406637405623257, "grad_norm": 1.4370090961456299, "learning_rate": 1.936488147505865e-05, "loss": 0.1734986662864685, "step": 5340 }, { "epoch": 0.14092715580682444, "grad_norm": 3.536433458328247, "learning_rate": 1.9361886121534686e-05, "loss": 0.1633779525756836, "step": 5350 }, { "epoch": 0.14119057105132315, "grad_norm": 1.5282236337661743, "learning_rate": 1.93588839540261e-05, "loss": 0.17345809936523438, "step": 5360 }, { "epoch": 0.1414539862958219, "grad_norm": 1.7045758962631226, "learning_rate": 1.9355874974717988e-05, "loss": 0.16580256223678588, "step": 5370 }, { "epoch": 0.14171740154032064, "grad_norm": 2.198791027069092, "learning_rate": 1.9352859185800424e-05, "loss": 0.156879198551178, "step": 5380 }, { "epoch": 0.14198081678481939, "grad_norm": 1.9409561157226562, "learning_rate": 1.934983658946843e-05, "loss": 0.17286429405212403, "step": 5390 }, { "epoch": 0.14224423202931813, "grad_norm": 2.0150980949401855, "learning_rate": 1.9346807187921983e-05, "loss": 0.16917107105255128, "step": 5400 }, { "epoch": 0.14250764727381685, "grad_norm": 5.8431806564331055, "learning_rate": 1.9343770983366004e-05, "loss": 0.16137199401855468, "step": 5410 }, { "epoch": 0.1427710625183156, "grad_norm": 1.9951761960983276, "learning_rate": 1.9340727978010377e-05, "loss": 0.17459444999694823, "step": 5420 }, { "epoch": 0.14303447776281433, "grad_norm": 2.5256850719451904, "learning_rate": 1.933767817406994e-05, "loss": 0.16633121967315673, "step": 5430 }, { "epoch": 0.14329789300731308, "grad_norm": 2.7632458209991455, "learning_rate": 1.9334621573764462e-05, "loss": 0.1718267798423767, "step": 5440 }, { "epoch": 0.1435613082518118, "grad_norm": 2.2428905963897705, "learning_rate": 1.9331558179318676e-05, "loss": 0.17521851062774657, "step": 5450 }, { "epoch": 0.14382472349631054, "grad_norm": 3.0693438053131104, "learning_rate": 1.9328487992962247e-05, "loss": 0.17082947492599487, "step": 5460 }, { "epoch": 0.14408813874080928, "grad_norm": 2.578197717666626, "learning_rate": 1.9325411016929793e-05, "loss": 0.1739073634147644, "step": 5470 }, { "epoch": 0.14435155398530802, "grad_norm": 1.2698296308517456, "learning_rate": 1.9322327253460867e-05, "loss": 0.17073020935058594, "step": 5480 }, { "epoch": 0.14461496922980674, "grad_norm": 2.772080898284912, "learning_rate": 1.931923670479997e-05, "loss": 0.166335391998291, "step": 5490 }, { "epoch": 0.14487838447430548, "grad_norm": 1.1978576183319092, "learning_rate": 1.9316139373196533e-05, "loss": 0.16980748176574706, "step": 5500 }, { "epoch": 0.14514179971880423, "grad_norm": 2.2324275970458984, "learning_rate": 1.9313035260904923e-05, "loss": 0.1717083692550659, "step": 5510 }, { "epoch": 0.14540521496330297, "grad_norm": 1.9762606620788574, "learning_rate": 1.9309924370184455e-05, "loss": 0.17277470827102662, "step": 5520 }, { "epoch": 0.1456686302078017, "grad_norm": 3.082820415496826, "learning_rate": 1.9306806703299367e-05, "loss": 0.17550311088562012, "step": 5530 }, { "epoch": 0.14593204545230043, "grad_norm": 2.1225407123565674, "learning_rate": 1.9303682262518828e-05, "loss": 0.18062272071838378, "step": 5540 }, { "epoch": 0.14619546069679917, "grad_norm": 1.8640729188919067, "learning_rate": 1.9300551050116946e-05, "loss": 0.16901164054870604, "step": 5550 }, { "epoch": 0.14645887594129792, "grad_norm": 2.439771890640259, "learning_rate": 1.9297413068372745e-05, "loss": 0.1602844476699829, "step": 5560 }, { "epoch": 0.14672229118579666, "grad_norm": 4.671451091766357, "learning_rate": 1.9294268319570195e-05, "loss": 0.16408965587615967, "step": 5570 }, { "epoch": 0.14698570643029538, "grad_norm": 2.9854133129119873, "learning_rate": 1.9291116805998168e-05, "loss": 0.17918325662612916, "step": 5580 }, { "epoch": 0.14724912167479412, "grad_norm": 1.1332268714904785, "learning_rate": 1.9287958529950477e-05, "loss": 0.17135252952575683, "step": 5590 }, { "epoch": 0.14751253691929286, "grad_norm": 3.1729419231414795, "learning_rate": 1.9284793493725852e-05, "loss": 0.1726720929145813, "step": 5600 }, { "epoch": 0.1477759521637916, "grad_norm": 2.483050584793091, "learning_rate": 1.9281621699627943e-05, "loss": 0.17062678337097167, "step": 5610 }, { "epoch": 0.14803936740829032, "grad_norm": 1.3899911642074585, "learning_rate": 1.927844314996532e-05, "loss": 0.15533797740936278, "step": 5620 }, { "epoch": 0.14830278265278907, "grad_norm": 3.8340771198272705, "learning_rate": 1.9275257847051464e-05, "loss": 0.16460051536560058, "step": 5630 }, { "epoch": 0.1485661978972878, "grad_norm": 4.6021623611450195, "learning_rate": 1.9272065793204778e-05, "loss": 0.1672244906425476, "step": 5640 }, { "epoch": 0.14882961314178655, "grad_norm": 5.646993637084961, "learning_rate": 1.9268866990748578e-05, "loss": 0.16547658443450927, "step": 5650 }, { "epoch": 0.1490930283862853, "grad_norm": 1.836978793144226, "learning_rate": 1.9265661442011087e-05, "loss": 0.18394027948379515, "step": 5660 }, { "epoch": 0.149356443630784, "grad_norm": 6.097099304199219, "learning_rate": 1.9262449149325445e-05, "loss": 0.18445050716400146, "step": 5670 }, { "epoch": 0.14961985887528276, "grad_norm": 1.3850363492965698, "learning_rate": 1.92592301150297e-05, "loss": 0.16360682249069214, "step": 5680 }, { "epoch": 0.1498832741197815, "grad_norm": 2.060900926589966, "learning_rate": 1.9256004341466795e-05, "loss": 0.16657625436782836, "step": 5690 }, { "epoch": 0.15014668936428024, "grad_norm": 3.503030776977539, "learning_rate": 1.9252771830984594e-05, "loss": 0.1537221908569336, "step": 5700 }, { "epoch": 0.15041010460877896, "grad_norm": 6.720454692840576, "learning_rate": 1.9249532585935855e-05, "loss": 0.1685310959815979, "step": 5710 }, { "epoch": 0.1506735198532777, "grad_norm": 6.196469783782959, "learning_rate": 1.9246286608678243e-05, "loss": 0.16894335746765138, "step": 5720 }, { "epoch": 0.15093693509777645, "grad_norm": 2.7375690937042236, "learning_rate": 1.9243033901574316e-05, "loss": 0.17621567249298095, "step": 5730 }, { "epoch": 0.1512003503422752, "grad_norm": 5.094202518463135, "learning_rate": 1.923977446699154e-05, "loss": 0.1775113582611084, "step": 5740 }, { "epoch": 0.1514637655867739, "grad_norm": 5.164031505584717, "learning_rate": 1.923650830730227e-05, "loss": 0.16846926212310792, "step": 5750 }, { "epoch": 0.15172718083127265, "grad_norm": 4.039833068847656, "learning_rate": 1.923323542488375e-05, "loss": 0.16426074504852295, "step": 5760 }, { "epoch": 0.1519905960757714, "grad_norm": 1.4167944192886353, "learning_rate": 1.9229955822118138e-05, "loss": 0.16697345972061156, "step": 5770 }, { "epoch": 0.15225401132027014, "grad_norm": 2.8155319690704346, "learning_rate": 1.9226669501392465e-05, "loss": 0.1675267696380615, "step": 5780 }, { "epoch": 0.15251742656476888, "grad_norm": 3.149172782897949, "learning_rate": 1.9223376465098657e-05, "loss": 0.16234806776046753, "step": 5790 }, { "epoch": 0.1527808418092676, "grad_norm": 8.26332950592041, "learning_rate": 1.9220076715633526e-05, "loss": 0.17473869323730468, "step": 5800 }, { "epoch": 0.15304425705376634, "grad_norm": 3.8364012241363525, "learning_rate": 1.921677025539877e-05, "loss": 0.16860609054565429, "step": 5810 }, { "epoch": 0.15330767229826509, "grad_norm": 0.9789993166923523, "learning_rate": 1.921345708680098e-05, "loss": 0.16231925487518312, "step": 5820 }, { "epoch": 0.15357108754276383, "grad_norm": 2.206066608428955, "learning_rate": 1.921013721225162e-05, "loss": 0.16474533081054688, "step": 5830 }, { "epoch": 0.15383450278726254, "grad_norm": 3.5293028354644775, "learning_rate": 1.9206810634167036e-05, "loss": 0.15388745069503784, "step": 5840 }, { "epoch": 0.1540979180317613, "grad_norm": 2.0668463706970215, "learning_rate": 1.920347735496846e-05, "loss": 0.16340789794921876, "step": 5850 }, { "epoch": 0.15436133327626003, "grad_norm": 1.166778802871704, "learning_rate": 1.9200137377081993e-05, "loss": 0.16839029788970947, "step": 5860 }, { "epoch": 0.15462474852075878, "grad_norm": 2.2165234088897705, "learning_rate": 1.919679070293861e-05, "loss": 0.15636686086654664, "step": 5870 }, { "epoch": 0.1548881637652575, "grad_norm": 4.523315906524658, "learning_rate": 1.9193437334974178e-05, "loss": 0.1589397668838501, "step": 5880 }, { "epoch": 0.15515157900975624, "grad_norm": 5.277530670166016, "learning_rate": 1.9190077275629408e-05, "loss": 0.16339442729949952, "step": 5890 }, { "epoch": 0.15541499425425498, "grad_norm": 1.70864737033844, "learning_rate": 1.9186710527349907e-05, "loss": 0.1652472972869873, "step": 5900 }, { "epoch": 0.15567840949875372, "grad_norm": 5.334270477294922, "learning_rate": 1.9183337092586137e-05, "loss": 0.16924357414245605, "step": 5910 }, { "epoch": 0.15594182474325247, "grad_norm": 1.385156512260437, "learning_rate": 1.9179956973793422e-05, "loss": 0.1645041823387146, "step": 5920 }, { "epoch": 0.15620523998775118, "grad_norm": 1.0010559558868408, "learning_rate": 1.917657017343197e-05, "loss": 0.17141462564468385, "step": 5930 }, { "epoch": 0.15646865523224993, "grad_norm": 1.862549901008606, "learning_rate": 1.9173176693966835e-05, "loss": 0.17259255647659302, "step": 5940 }, { "epoch": 0.15673207047674867, "grad_norm": 2.4449520111083984, "learning_rate": 1.9169776537867936e-05, "loss": 0.1712886095046997, "step": 5950 }, { "epoch": 0.1569954857212474, "grad_norm": 2.7257649898529053, "learning_rate": 1.9166369707610055e-05, "loss": 0.1664157271385193, "step": 5960 }, { "epoch": 0.15725890096574613, "grad_norm": 2.6416056156158447, "learning_rate": 1.9162956205672833e-05, "loss": 0.1707891345024109, "step": 5970 }, { "epoch": 0.15752231621024487, "grad_norm": 1.1656323671340942, "learning_rate": 1.9159536034540757e-05, "loss": 0.17157247066497802, "step": 5980 }, { "epoch": 0.15778573145474362, "grad_norm": 2.587618827819824, "learning_rate": 1.9156109196703177e-05, "loss": 0.17043046951293944, "step": 5990 }, { "epoch": 0.15804914669924236, "grad_norm": 1.2382454872131348, "learning_rate": 1.91526756946543e-05, "loss": 0.17391424179077147, "step": 6000 }, { "epoch": 0.15831256194374108, "grad_norm": 6.294238567352295, "learning_rate": 1.9149235530893172e-05, "loss": 0.17003856897354125, "step": 6010 }, { "epoch": 0.15857597718823982, "grad_norm": 2.2809574604034424, "learning_rate": 1.9145788707923692e-05, "loss": 0.18345091342926026, "step": 6020 }, { "epoch": 0.15883939243273856, "grad_norm": 2.3363735675811768, "learning_rate": 1.914233522825461e-05, "loss": 0.16345294713973998, "step": 6030 }, { "epoch": 0.1591028076772373, "grad_norm": 2.666842222213745, "learning_rate": 1.913887509439951e-05, "loss": 0.1666812777519226, "step": 6040 }, { "epoch": 0.15936622292173605, "grad_norm": 4.385406970977783, "learning_rate": 1.9135408308876834e-05, "loss": 0.169732666015625, "step": 6050 }, { "epoch": 0.15962963816623477, "grad_norm": 2.8031108379364014, "learning_rate": 1.913193487420986e-05, "loss": 0.1608938217163086, "step": 6060 }, { "epoch": 0.1598930534107335, "grad_norm": 1.7167831659317017, "learning_rate": 1.9128454792926694e-05, "loss": 0.16481552124023438, "step": 6070 }, { "epoch": 0.16015646865523225, "grad_norm": 3.9837934970855713, "learning_rate": 1.9124968067560296e-05, "loss": 0.16808457374572755, "step": 6080 }, { "epoch": 0.160419883899731, "grad_norm": 4.571429252624512, "learning_rate": 1.9121474700648454e-05, "loss": 0.1675979733467102, "step": 6090 }, { "epoch": 0.1606832991442297, "grad_norm": 3.5140507221221924, "learning_rate": 1.9117974694733795e-05, "loss": 0.16835908889770507, "step": 6100 }, { "epoch": 0.16094671438872846, "grad_norm": 2.395657539367676, "learning_rate": 1.9114468052363764e-05, "loss": 0.17260162830352782, "step": 6110 }, { "epoch": 0.1612101296332272, "grad_norm": 1.8424381017684937, "learning_rate": 1.9110954776090656e-05, "loss": 0.16851887702941895, "step": 6120 }, { "epoch": 0.16147354487772594, "grad_norm": 2.752537250518799, "learning_rate": 1.910743486847158e-05, "loss": 0.15734286308288575, "step": 6130 }, { "epoch": 0.16173696012222466, "grad_norm": 1.6771060228347778, "learning_rate": 1.9103908332068483e-05, "loss": 0.17108169794082642, "step": 6140 }, { "epoch": 0.1620003753667234, "grad_norm": 3.5005722045898438, "learning_rate": 1.910037516944812e-05, "loss": 0.15672807693481444, "step": 6150 }, { "epoch": 0.16226379061122215, "grad_norm": 3.0754384994506836, "learning_rate": 1.909683538318209e-05, "loss": 0.16478219032287597, "step": 6160 }, { "epoch": 0.1625272058557209, "grad_norm": 2.729433059692383, "learning_rate": 1.9093288975846797e-05, "loss": 0.15805864334106445, "step": 6170 }, { "epoch": 0.16279062110021963, "grad_norm": 2.15678071975708, "learning_rate": 1.9089735950023463e-05, "loss": 0.17794071435928344, "step": 6180 }, { "epoch": 0.16305403634471835, "grad_norm": 1.323996901512146, "learning_rate": 1.9086176308298147e-05, "loss": 0.16701146364212036, "step": 6190 }, { "epoch": 0.1633174515892171, "grad_norm": 8.514081954956055, "learning_rate": 1.90826100532617e-05, "loss": 0.18284302949905396, "step": 6200 }, { "epoch": 0.16358086683371584, "grad_norm": 3.538384199142456, "learning_rate": 1.90790371875098e-05, "loss": 0.17290310859680175, "step": 6210 }, { "epoch": 0.16384428207821458, "grad_norm": 1.733695387840271, "learning_rate": 1.9075457713642934e-05, "loss": 0.17540112733840943, "step": 6220 }, { "epoch": 0.1641076973227133, "grad_norm": 1.6030211448669434, "learning_rate": 1.9071871634266392e-05, "loss": 0.16189069747924806, "step": 6230 }, { "epoch": 0.16437111256721204, "grad_norm": 1.1916093826293945, "learning_rate": 1.9068278951990282e-05, "loss": 0.1598558783531189, "step": 6240 }, { "epoch": 0.16463452781171078, "grad_norm": 5.072381496429443, "learning_rate": 1.9064679669429515e-05, "loss": 0.1668604850769043, "step": 6250 }, { "epoch": 0.16489794305620953, "grad_norm": 3.5737311840057373, "learning_rate": 1.9061073789203798e-05, "loss": 0.16895081996917724, "step": 6260 }, { "epoch": 0.16516135830070824, "grad_norm": 0.983394205570221, "learning_rate": 1.9057461313937646e-05, "loss": 0.17953739166259766, "step": 6270 }, { "epoch": 0.165424773545207, "grad_norm": 1.8774077892303467, "learning_rate": 1.905384224626038e-05, "loss": 0.17450435161590577, "step": 6280 }, { "epoch": 0.16568818878970573, "grad_norm": 4.018238544464111, "learning_rate": 1.9050216588806113e-05, "loss": 0.1708436965942383, "step": 6290 }, { "epoch": 0.16595160403420448, "grad_norm": 4.041280269622803, "learning_rate": 1.9046584344213754e-05, "loss": 0.15845983028411864, "step": 6300 }, { "epoch": 0.16621501927870322, "grad_norm": 3.2483389377593994, "learning_rate": 1.9042945515127004e-05, "loss": 0.1646010160446167, "step": 6310 }, { "epoch": 0.16647843452320193, "grad_norm": 1.7113370895385742, "learning_rate": 1.9039300104194366e-05, "loss": 0.15454000234603882, "step": 6320 }, { "epoch": 0.16674184976770068, "grad_norm": 2.4345290660858154, "learning_rate": 1.903564811406912e-05, "loss": 0.16855895519256592, "step": 6330 }, { "epoch": 0.16700526501219942, "grad_norm": 3.509082794189453, "learning_rate": 1.9031989547409347e-05, "loss": 0.15935189723968507, "step": 6340 }, { "epoch": 0.16726868025669817, "grad_norm": 1.0140666961669922, "learning_rate": 1.9028324406877905e-05, "loss": 0.15747071504592897, "step": 6350 }, { "epoch": 0.16753209550119688, "grad_norm": 2.361509084701538, "learning_rate": 1.9024652695142447e-05, "loss": 0.15230283737182618, "step": 6360 }, { "epoch": 0.16779551074569563, "grad_norm": 2.1254873275756836, "learning_rate": 1.90209744148754e-05, "loss": 0.16623189449310302, "step": 6370 }, { "epoch": 0.16805892599019437, "grad_norm": 3.5191233158111572, "learning_rate": 1.901728956875398e-05, "loss": 0.15681886672973633, "step": 6380 }, { "epoch": 0.1683223412346931, "grad_norm": 2.039400339126587, "learning_rate": 1.9013598159460168e-05, "loss": 0.15070743560791017, "step": 6390 }, { "epoch": 0.16858575647919183, "grad_norm": 5.270078182220459, "learning_rate": 1.900990018968074e-05, "loss": 0.16359434127807618, "step": 6400 }, { "epoch": 0.16884917172369057, "grad_norm": 2.2537286281585693, "learning_rate": 1.900619566210723e-05, "loss": 0.14774045944213868, "step": 6410 }, { "epoch": 0.16911258696818932, "grad_norm": 1.422263264656067, "learning_rate": 1.900248457943596e-05, "loss": 0.15659677982330322, "step": 6420 }, { "epoch": 0.16937600221268806, "grad_norm": 4.971650123596191, "learning_rate": 1.899876694436801e-05, "loss": 0.16916341781616212, "step": 6430 }, { "epoch": 0.1696394174571868, "grad_norm": 2.080029249191284, "learning_rate": 1.899504275960924e-05, "loss": 0.16470694541931152, "step": 6440 }, { "epoch": 0.16990283270168552, "grad_norm": 4.356668949127197, "learning_rate": 1.899131202787027e-05, "loss": 0.17042373418807982, "step": 6450 }, { "epoch": 0.17016624794618426, "grad_norm": 1.409234642982483, "learning_rate": 1.8987574751866488e-05, "loss": 0.18997092247009278, "step": 6460 }, { "epoch": 0.170429663190683, "grad_norm": 5.923281669616699, "learning_rate": 1.8983830934318044e-05, "loss": 0.17875025272369385, "step": 6470 }, { "epoch": 0.17069307843518175, "grad_norm": 4.1839680671691895, "learning_rate": 1.8980080577949848e-05, "loss": 0.16771798133850097, "step": 6480 }, { "epoch": 0.17095649367968047, "grad_norm": 4.0030293464660645, "learning_rate": 1.8976323685491577e-05, "loss": 0.174286150932312, "step": 6490 }, { "epoch": 0.1712199089241792, "grad_norm": 3.176650285720825, "learning_rate": 1.8972560259677658e-05, "loss": 0.17063318490982055, "step": 6500 }, { "epoch": 0.17148332416867795, "grad_norm": 6.73101806640625, "learning_rate": 1.896879030324727e-05, "loss": 0.16379247903823851, "step": 6510 }, { "epoch": 0.1717467394131767, "grad_norm": 1.5533398389816284, "learning_rate": 1.8965013818944357e-05, "loss": 0.16702687740325928, "step": 6520 }, { "epoch": 0.1720101546576754, "grad_norm": 2.447460412979126, "learning_rate": 1.89612308095176e-05, "loss": 0.16827545166015626, "step": 6530 }, { "epoch": 0.17227356990217416, "grad_norm": 1.9511741399765015, "learning_rate": 1.8957441277720438e-05, "loss": 0.17476340532302856, "step": 6540 }, { "epoch": 0.1725369851466729, "grad_norm": 1.6443811655044556, "learning_rate": 1.8953645226311056e-05, "loss": 0.16144053936004638, "step": 6550 }, { "epoch": 0.17280040039117164, "grad_norm": 1.9909825325012207, "learning_rate": 1.894984265805239e-05, "loss": 0.17726457118988037, "step": 6560 }, { "epoch": 0.1730638156356704, "grad_norm": 2.311997652053833, "learning_rate": 1.8946033575712104e-05, "loss": 0.15998380184173583, "step": 6570 }, { "epoch": 0.1733272308801691, "grad_norm": 1.371277093887329, "learning_rate": 1.894221798206262e-05, "loss": 0.17248244285583497, "step": 6580 }, { "epoch": 0.17359064612466785, "grad_norm": 3.349735736846924, "learning_rate": 1.8938395879881086e-05, "loss": 0.17439991235733032, "step": 6590 }, { "epoch": 0.1738540613691666, "grad_norm": 1.8678386211395264, "learning_rate": 1.8934567271949397e-05, "loss": 0.17888455390930175, "step": 6600 }, { "epoch": 0.17411747661366533, "grad_norm": 2.1697986125946045, "learning_rate": 1.8930732161054176e-05, "loss": 0.15905191898345947, "step": 6610 }, { "epoch": 0.17438089185816405, "grad_norm": 4.61419677734375, "learning_rate": 1.8926890549986786e-05, "loss": 0.16772692203521727, "step": 6620 }, { "epoch": 0.1746443071026628, "grad_norm": 2.403304100036621, "learning_rate": 1.8923042441543313e-05, "loss": 0.16545236110687256, "step": 6630 }, { "epoch": 0.17490772234716154, "grad_norm": 5.239044666290283, "learning_rate": 1.891918783852458e-05, "loss": 0.1718190312385559, "step": 6640 }, { "epoch": 0.17517113759166028, "grad_norm": 0.8933268785476685, "learning_rate": 1.8915326743736134e-05, "loss": 0.14813650846481324, "step": 6650 }, { "epoch": 0.175434552836159, "grad_norm": 4.976561069488525, "learning_rate": 1.8911459159988238e-05, "loss": 0.17692904472351073, "step": 6660 }, { "epoch": 0.17569796808065774, "grad_norm": 1.829314112663269, "learning_rate": 1.8907585090095897e-05, "loss": 0.16733739376068116, "step": 6670 }, { "epoch": 0.17596138332515648, "grad_norm": 3.453420877456665, "learning_rate": 1.890370453687882e-05, "loss": 0.1675776720046997, "step": 6680 }, { "epoch": 0.17622479856965523, "grad_norm": 4.0025248527526855, "learning_rate": 1.889981750316144e-05, "loss": 0.1763147830963135, "step": 6690 }, { "epoch": 0.17648821381415397, "grad_norm": 2.937887668609619, "learning_rate": 1.8895923991772913e-05, "loss": 0.16918323040008545, "step": 6700 }, { "epoch": 0.1767516290586527, "grad_norm": 2.0321450233459473, "learning_rate": 1.8892024005547104e-05, "loss": 0.18218109607696534, "step": 6710 }, { "epoch": 0.17701504430315143, "grad_norm": 5.938467502593994, "learning_rate": 1.8888117547322586e-05, "loss": 0.16488065719604492, "step": 6720 }, { "epoch": 0.17727845954765017, "grad_norm": 2.506981611251831, "learning_rate": 1.888420461994265e-05, "loss": 0.1707826614379883, "step": 6730 }, { "epoch": 0.17754187479214892, "grad_norm": 3.9032886028289795, "learning_rate": 1.88802852262553e-05, "loss": 0.16202991008758544, "step": 6740 }, { "epoch": 0.17780529003664763, "grad_norm": 4.616270065307617, "learning_rate": 1.887635936911323e-05, "loss": 0.17648696899414062, "step": 6750 }, { "epoch": 0.17806870528114638, "grad_norm": 1.9402168989181519, "learning_rate": 1.8872427051373852e-05, "loss": 0.16804811954498292, "step": 6760 }, { "epoch": 0.17833212052564512, "grad_norm": 2.2847771644592285, "learning_rate": 1.886848827589928e-05, "loss": 0.16122440099716187, "step": 6770 }, { "epoch": 0.17859553577014387, "grad_norm": 3.1110544204711914, "learning_rate": 1.8864543045556318e-05, "loss": 0.17097421884536743, "step": 6780 }, { "epoch": 0.1788589510146426, "grad_norm": 5.511929512023926, "learning_rate": 1.8860591363216482e-05, "loss": 0.1705482244491577, "step": 6790 }, { "epoch": 0.17912236625914132, "grad_norm": 2.6521503925323486, "learning_rate": 1.8856633231755972e-05, "loss": 0.1704709529876709, "step": 6800 }, { "epoch": 0.17938578150364007, "grad_norm": 1.933807373046875, "learning_rate": 1.8852668654055686e-05, "loss": 0.16671547889709473, "step": 6810 }, { "epoch": 0.1796491967481388, "grad_norm": 2.195162534713745, "learning_rate": 1.884869763300122e-05, "loss": 0.17076725959777833, "step": 6820 }, { "epoch": 0.17991261199263756, "grad_norm": 1.834706425666809, "learning_rate": 1.8844720171482854e-05, "loss": 0.15827840566635132, "step": 6830 }, { "epoch": 0.18017602723713627, "grad_norm": 3.1246657371520996, "learning_rate": 1.884073627239555e-05, "loss": 0.16756949424743653, "step": 6840 }, { "epoch": 0.18043944248163502, "grad_norm": 4.0762810707092285, "learning_rate": 1.8836745938638963e-05, "loss": 0.17327953577041627, "step": 6850 }, { "epoch": 0.18070285772613376, "grad_norm": 1.5226972103118896, "learning_rate": 1.8832749173117435e-05, "loss": 0.17376726865768433, "step": 6860 }, { "epoch": 0.1809662729706325, "grad_norm": 2.940664768218994, "learning_rate": 1.8828745978739982e-05, "loss": 0.1664882183074951, "step": 6870 }, { "epoch": 0.18122968821513122, "grad_norm": 0.9831534028053284, "learning_rate": 1.88247363584203e-05, "loss": 0.16474609375, "step": 6880 }, { "epoch": 0.18149310345962996, "grad_norm": 3.8242170810699463, "learning_rate": 1.8820720315076767e-05, "loss": 0.165224027633667, "step": 6890 }, { "epoch": 0.1817565187041287, "grad_norm": 2.3860368728637695, "learning_rate": 1.8816697851632425e-05, "loss": 0.17467756271362306, "step": 6900 }, { "epoch": 0.18201993394862745, "grad_norm": 2.2875616550445557, "learning_rate": 1.8812668971015006e-05, "loss": 0.16640254259109497, "step": 6910 }, { "epoch": 0.1822833491931262, "grad_norm": 1.7849301099777222, "learning_rate": 1.88086336761569e-05, "loss": 0.1664609432220459, "step": 6920 }, { "epoch": 0.1825467644376249, "grad_norm": 1.2779046297073364, "learning_rate": 1.8804591969995167e-05, "loss": 0.17204318046569825, "step": 6930 }, { "epoch": 0.18281017968212365, "grad_norm": 2.1393074989318848, "learning_rate": 1.8800543855471535e-05, "loss": 0.168204927444458, "step": 6940 }, { "epoch": 0.1830735949266224, "grad_norm": 2.012359142303467, "learning_rate": 1.87964893355324e-05, "loss": 0.16373515129089355, "step": 6950 }, { "epoch": 0.18333701017112114, "grad_norm": 3.754606246948242, "learning_rate": 1.879242841312881e-05, "loss": 0.16655974388122557, "step": 6960 }, { "epoch": 0.18360042541561986, "grad_norm": 4.817680835723877, "learning_rate": 1.878836109121648e-05, "loss": 0.16246843338012695, "step": 6970 }, { "epoch": 0.1838638406601186, "grad_norm": 2.288623094558716, "learning_rate": 1.8784287372755795e-05, "loss": 0.17999591827392578, "step": 6980 }, { "epoch": 0.18412725590461734, "grad_norm": 5.466291427612305, "learning_rate": 1.8780207260711767e-05, "loss": 0.17384878396987916, "step": 6990 }, { "epoch": 0.1843906711491161, "grad_norm": 2.19549298286438, "learning_rate": 1.8776120758054087e-05, "loss": 0.17137726545333862, "step": 7000 }, { "epoch": 0.1846540863936148, "grad_norm": 3.5028862953186035, "learning_rate": 1.8772027867757082e-05, "loss": 0.1672600030899048, "step": 7010 }, { "epoch": 0.18491750163811355, "grad_norm": 4.96008825302124, "learning_rate": 1.8767928592799742e-05, "loss": 0.17165446281433105, "step": 7020 }, { "epoch": 0.1851809168826123, "grad_norm": 4.227313995361328, "learning_rate": 1.876382293616569e-05, "loss": 0.17855286598205566, "step": 7030 }, { "epoch": 0.18544433212711103, "grad_norm": 1.7813960313796997, "learning_rate": 1.8759710900843202e-05, "loss": 0.16458516120910643, "step": 7040 }, { "epoch": 0.18570774737160978, "grad_norm": 1.5940022468566895, "learning_rate": 1.8755592489825193e-05, "loss": 0.15135974884033204, "step": 7050 }, { "epoch": 0.1859711626161085, "grad_norm": 2.056298017501831, "learning_rate": 1.8751467706109224e-05, "loss": 0.1800605058670044, "step": 7060 }, { "epoch": 0.18623457786060724, "grad_norm": 5.005214691162109, "learning_rate": 1.8747336552697483e-05, "loss": 0.1683848738670349, "step": 7070 }, { "epoch": 0.18649799310510598, "grad_norm": 0.9452570676803589, "learning_rate": 1.874319903259681e-05, "loss": 0.1573354721069336, "step": 7080 }, { "epoch": 0.18676140834960472, "grad_norm": 1.3948955535888672, "learning_rate": 1.873905514881867e-05, "loss": 0.18253626823425292, "step": 7090 }, { "epoch": 0.18702482359410344, "grad_norm": 4.967896461486816, "learning_rate": 1.873490490437915e-05, "loss": 0.1632927656173706, "step": 7100 }, { "epoch": 0.18728823883860218, "grad_norm": 5.258554458618164, "learning_rate": 1.8730748302298993e-05, "loss": 0.17122315168380736, "step": 7110 }, { "epoch": 0.18755165408310093, "grad_norm": 2.968151092529297, "learning_rate": 1.8726585345603544e-05, "loss": 0.1974579095840454, "step": 7120 }, { "epoch": 0.18781506932759967, "grad_norm": 1.3686175346374512, "learning_rate": 1.8722416037322778e-05, "loss": 0.17359331846237183, "step": 7130 }, { "epoch": 0.1880784845720984, "grad_norm": 2.603327989578247, "learning_rate": 1.871824038049131e-05, "loss": 0.18198542594909667, "step": 7140 }, { "epoch": 0.18834189981659713, "grad_norm": 2.7815163135528564, "learning_rate": 1.871405837814835e-05, "loss": 0.16474463939666747, "step": 7150 }, { "epoch": 0.18860531506109587, "grad_norm": 1.5069242715835571, "learning_rate": 1.8709870033337753e-05, "loss": 0.18136310577392578, "step": 7160 }, { "epoch": 0.18886873030559462, "grad_norm": 4.561825275421143, "learning_rate": 1.8705675349107964e-05, "loss": 0.16998510360717772, "step": 7170 }, { "epoch": 0.18913214555009336, "grad_norm": 3.4238052368164062, "learning_rate": 1.870147432851206e-05, "loss": 0.16424649953842163, "step": 7180 }, { "epoch": 0.18939556079459208, "grad_norm": 3.702589750289917, "learning_rate": 1.8697266974607732e-05, "loss": 0.17192732095718383, "step": 7190 }, { "epoch": 0.18965897603909082, "grad_norm": 2.7156405448913574, "learning_rate": 1.8693053290457264e-05, "loss": 0.1575788974761963, "step": 7200 }, { "epoch": 0.18992239128358956, "grad_norm": 2.9449477195739746, "learning_rate": 1.868883327912756e-05, "loss": 0.16923377513885499, "step": 7210 }, { "epoch": 0.1901858065280883, "grad_norm": 1.1869418621063232, "learning_rate": 1.868460694369013e-05, "loss": 0.1736302614212036, "step": 7220 }, { "epoch": 0.19044922177258702, "grad_norm": 1.886535406112671, "learning_rate": 1.868037428722108e-05, "loss": 0.16881624460220337, "step": 7230 }, { "epoch": 0.19071263701708577, "grad_norm": 1.9265042543411255, "learning_rate": 1.867613531280112e-05, "loss": 0.16995539665222167, "step": 7240 }, { "epoch": 0.1909760522615845, "grad_norm": 2.091367483139038, "learning_rate": 1.867189002351556e-05, "loss": 0.16899628639221193, "step": 7250 }, { "epoch": 0.19123946750608325, "grad_norm": 4.666804313659668, "learning_rate": 1.8667638422454306e-05, "loss": 0.17334694862365724, "step": 7260 }, { "epoch": 0.19150288275058197, "grad_norm": 2.7512502670288086, "learning_rate": 1.8663380512711852e-05, "loss": 0.1671319603919983, "step": 7270 }, { "epoch": 0.19176629799508071, "grad_norm": 1.1483733654022217, "learning_rate": 1.865911629738729e-05, "loss": 0.16664502620697022, "step": 7280 }, { "epoch": 0.19202971323957946, "grad_norm": 1.8396552801132202, "learning_rate": 1.86548457795843e-05, "loss": 0.17336664199829102, "step": 7290 }, { "epoch": 0.1922931284840782, "grad_norm": 1.7907785177230835, "learning_rate": 1.8650568962411154e-05, "loss": 0.16533132791519164, "step": 7300 }, { "epoch": 0.19255654372857695, "grad_norm": 2.980379104614258, "learning_rate": 1.8646285848980693e-05, "loss": 0.15267860889434814, "step": 7310 }, { "epoch": 0.19281995897307566, "grad_norm": 6.746033668518066, "learning_rate": 1.8641996442410365e-05, "loss": 0.15144495964050292, "step": 7320 }, { "epoch": 0.1930833742175744, "grad_norm": 1.285355806350708, "learning_rate": 1.8637700745822176e-05, "loss": 0.17192561626434327, "step": 7330 }, { "epoch": 0.19334678946207315, "grad_norm": 0.7614986300468445, "learning_rate": 1.8633398762342722e-05, "loss": 0.1694122314453125, "step": 7340 }, { "epoch": 0.1936102047065719, "grad_norm": 3.6428616046905518, "learning_rate": 1.8629090495103175e-05, "loss": 0.17018821239471435, "step": 7350 }, { "epoch": 0.1938736199510706, "grad_norm": 3.221688747406006, "learning_rate": 1.8624775947239272e-05, "loss": 0.14748151302337648, "step": 7360 }, { "epoch": 0.19413703519556935, "grad_norm": 3.866011142730713, "learning_rate": 1.862045512189133e-05, "loss": 0.1685500144958496, "step": 7370 }, { "epoch": 0.1944004504400681, "grad_norm": 3.077223777770996, "learning_rate": 1.8616128022204233e-05, "loss": 0.166597318649292, "step": 7380 }, { "epoch": 0.19466386568456684, "grad_norm": 1.532568335533142, "learning_rate": 1.8611794651327432e-05, "loss": 0.17148557901382447, "step": 7390 }, { "epoch": 0.19492728092906556, "grad_norm": 1.3085945844650269, "learning_rate": 1.8607455012414936e-05, "loss": 0.15532616376876832, "step": 7400 }, { "epoch": 0.1951906961735643, "grad_norm": 1.4049899578094482, "learning_rate": 1.8603109108625327e-05, "loss": 0.1671807885169983, "step": 7410 }, { "epoch": 0.19545411141806304, "grad_norm": 2.705167531967163, "learning_rate": 1.859875694312174e-05, "loss": 0.153898024559021, "step": 7420 }, { "epoch": 0.19571752666256179, "grad_norm": 1.4596813917160034, "learning_rate": 1.8594398519071867e-05, "loss": 0.17268327474594117, "step": 7430 }, { "epoch": 0.19598094190706053, "grad_norm": 1.4229291677474976, "learning_rate": 1.8590033839647955e-05, "loss": 0.17692577838897705, "step": 7440 }, { "epoch": 0.19624435715155925, "grad_norm": 1.0243366956710815, "learning_rate": 1.8585662908026814e-05, "loss": 0.1621867299079895, "step": 7450 }, { "epoch": 0.196507772396058, "grad_norm": 1.408636450767517, "learning_rate": 1.8581285727389787e-05, "loss": 0.1668846607208252, "step": 7460 }, { "epoch": 0.19677118764055673, "grad_norm": 4.507820129394531, "learning_rate": 1.857690230092278e-05, "loss": 0.18005743026733398, "step": 7470 }, { "epoch": 0.19703460288505548, "grad_norm": 1.8441933393478394, "learning_rate": 1.8572512631816243e-05, "loss": 0.1600633978843689, "step": 7480 }, { "epoch": 0.1972980181295542, "grad_norm": 0.8664716482162476, "learning_rate": 1.856811672326516e-05, "loss": 0.15333106517791747, "step": 7490 }, { "epoch": 0.19756143337405294, "grad_norm": 2.036254644393921, "learning_rate": 1.8563714578469064e-05, "loss": 0.17555055618286133, "step": 7500 }, { "epoch": 0.19782484861855168, "grad_norm": 4.339207649230957, "learning_rate": 1.8559306200632036e-05, "loss": 0.16616731882095337, "step": 7510 }, { "epoch": 0.19808826386305042, "grad_norm": 1.8186001777648926, "learning_rate": 1.8554891592962666e-05, "loss": 0.15569632053375243, "step": 7520 }, { "epoch": 0.19835167910754914, "grad_norm": 2.266519069671631, "learning_rate": 1.8550470758674112e-05, "loss": 0.16194332838058473, "step": 7530 }, { "epoch": 0.19861509435204788, "grad_norm": 1.6893670558929443, "learning_rate": 1.8546043700984036e-05, "loss": 0.17057815790176392, "step": 7540 }, { "epoch": 0.19887850959654663, "grad_norm": 2.0106425285339355, "learning_rate": 1.8541610423114647e-05, "loss": 0.16869843006134033, "step": 7550 }, { "epoch": 0.19914192484104537, "grad_norm": 2.557711124420166, "learning_rate": 1.853717092829268e-05, "loss": 0.170288610458374, "step": 7560 }, { "epoch": 0.1994053400855441, "grad_norm": 1.4350111484527588, "learning_rate": 1.853272521974938e-05, "loss": 0.17431919574737548, "step": 7570 }, { "epoch": 0.19966875533004283, "grad_norm": 1.5304800271987915, "learning_rate": 1.8528273300720538e-05, "loss": 0.16503219604492186, "step": 7580 }, { "epoch": 0.19993217057454157, "grad_norm": 1.6549766063690186, "learning_rate": 1.8523815174446444e-05, "loss": 0.16844053268432618, "step": 7590 }, { "epoch": 0.20019558581904032, "grad_norm": 1.2179384231567383, "learning_rate": 1.851935084417192e-05, "loss": 0.15739092826843262, "step": 7600 }, { "epoch": 0.20045900106353906, "grad_norm": 2.3222014904022217, "learning_rate": 1.8514880313146295e-05, "loss": 0.18232017755508423, "step": 7610 }, { "epoch": 0.20072241630803778, "grad_norm": 1.7972311973571777, "learning_rate": 1.8510403584623414e-05, "loss": 0.17485233545303344, "step": 7620 }, { "epoch": 0.20098583155253652, "grad_norm": 5.872239112854004, "learning_rate": 1.850592066186163e-05, "loss": 0.16204276084899902, "step": 7630 }, { "epoch": 0.20124924679703526, "grad_norm": 2.9276838302612305, "learning_rate": 1.850143154812382e-05, "loss": 0.1588260769844055, "step": 7640 }, { "epoch": 0.201512662041534, "grad_norm": 1.9206794500350952, "learning_rate": 1.849693624667734e-05, "loss": 0.1606444835662842, "step": 7650 }, { "epoch": 0.20177607728603272, "grad_norm": 1.9077166318893433, "learning_rate": 1.8492434760794072e-05, "loss": 0.17199217081069945, "step": 7660 }, { "epoch": 0.20203949253053147, "grad_norm": 1.8749769926071167, "learning_rate": 1.848792709375039e-05, "loss": 0.16172337532043457, "step": 7670 }, { "epoch": 0.2023029077750302, "grad_norm": 1.0930864810943604, "learning_rate": 1.848341324882717e-05, "loss": 0.15727009773254394, "step": 7680 }, { "epoch": 0.20256632301952895, "grad_norm": 1.137491226196289, "learning_rate": 1.8478893229309784e-05, "loss": 0.17343828678131104, "step": 7690 }, { "epoch": 0.2028297382640277, "grad_norm": 1.8160401582717896, "learning_rate": 1.847436703848809e-05, "loss": 0.16638603210449218, "step": 7700 }, { "epoch": 0.20309315350852641, "grad_norm": 3.2144737243652344, "learning_rate": 1.8469834679656452e-05, "loss": 0.1740139603614807, "step": 7710 }, { "epoch": 0.20335656875302516, "grad_norm": 2.218784809112549, "learning_rate": 1.8465296156113715e-05, "loss": 0.16525346040725708, "step": 7720 }, { "epoch": 0.2036199839975239, "grad_norm": 1.865092396736145, "learning_rate": 1.8460751471163212e-05, "loss": 0.17438361644744874, "step": 7730 }, { "epoch": 0.20388339924202264, "grad_norm": 2.692843198776245, "learning_rate": 1.845620062811276e-05, "loss": 0.17645652294158937, "step": 7740 }, { "epoch": 0.20414681448652136, "grad_norm": 3.0948400497436523, "learning_rate": 1.8451643630274664e-05, "loss": 0.17031031847000122, "step": 7750 }, { "epoch": 0.2044102297310201, "grad_norm": 1.7944227457046509, "learning_rate": 1.8447080480965694e-05, "loss": 0.1712850570678711, "step": 7760 }, { "epoch": 0.20467364497551885, "grad_norm": 3.1319081783294678, "learning_rate": 1.844251118350711e-05, "loss": 0.16123071908950806, "step": 7770 }, { "epoch": 0.2049370602200176, "grad_norm": 2.255049467086792, "learning_rate": 1.8437935741224652e-05, "loss": 0.17684030532836914, "step": 7780 }, { "epoch": 0.2052004754645163, "grad_norm": 5.418229579925537, "learning_rate": 1.8433354157448513e-05, "loss": 0.17405959367752075, "step": 7790 }, { "epoch": 0.20546389070901505, "grad_norm": 4.76038932800293, "learning_rate": 1.8428766435513373e-05, "loss": 0.168355131149292, "step": 7800 }, { "epoch": 0.2057273059535138, "grad_norm": 6.133687496185303, "learning_rate": 1.8424172578758374e-05, "loss": 0.17043046951293944, "step": 7810 }, { "epoch": 0.20599072119801254, "grad_norm": 3.8629117012023926, "learning_rate": 1.841957259052712e-05, "loss": 0.16035149097442628, "step": 7820 }, { "epoch": 0.20625413644251128, "grad_norm": 1.378327488899231, "learning_rate": 1.8414966474167688e-05, "loss": 0.17523620128631592, "step": 7830 }, { "epoch": 0.20651755168701, "grad_norm": 5.217845439910889, "learning_rate": 1.8410354233032604e-05, "loss": 0.1648694396018982, "step": 7840 }, { "epoch": 0.20678096693150874, "grad_norm": 1.2069529294967651, "learning_rate": 1.840573587047885e-05, "loss": 0.16076507568359374, "step": 7850 }, { "epoch": 0.20704438217600749, "grad_norm": 1.7597277164459229, "learning_rate": 1.8401111389867882e-05, "loss": 0.17651643753051757, "step": 7860 }, { "epoch": 0.20730779742050623, "grad_norm": 3.693903923034668, "learning_rate": 1.8396480794565583e-05, "loss": 0.1655038356781006, "step": 7870 }, { "epoch": 0.20757121266500494, "grad_norm": 2.703272581100464, "learning_rate": 1.839184408794231e-05, "loss": 0.16957392692565917, "step": 7880 }, { "epoch": 0.2078346279095037, "grad_norm": 1.255118489265442, "learning_rate": 1.8387201273372855e-05, "loss": 0.1652735710144043, "step": 7890 }, { "epoch": 0.20809804315400243, "grad_norm": 5.703413486480713, "learning_rate": 1.838255235423646e-05, "loss": 0.15682147741317748, "step": 7900 }, { "epoch": 0.20836145839850118, "grad_norm": 1.9693714380264282, "learning_rate": 1.8377897333916808e-05, "loss": 0.16545431613922118, "step": 7910 }, { "epoch": 0.2086248736429999, "grad_norm": 6.526916027069092, "learning_rate": 1.837323621580202e-05, "loss": 0.15698487758636476, "step": 7920 }, { "epoch": 0.20888828888749864, "grad_norm": 4.026800632476807, "learning_rate": 1.8368569003284662e-05, "loss": 0.16098641157150267, "step": 7930 }, { "epoch": 0.20915170413199738, "grad_norm": 2.7874350547790527, "learning_rate": 1.8363895699761736e-05, "loss": 0.17002403736114502, "step": 7940 }, { "epoch": 0.20941511937649612, "grad_norm": 6.196747779846191, "learning_rate": 1.8359216308634668e-05, "loss": 0.1652800917625427, "step": 7950 }, { "epoch": 0.20967853462099487, "grad_norm": 1.1055113077163696, "learning_rate": 1.8354530833309323e-05, "loss": 0.17252085208892823, "step": 7960 }, { "epoch": 0.20994194986549358, "grad_norm": 3.517197847366333, "learning_rate": 1.8349839277195993e-05, "loss": 0.15186965465545654, "step": 7970 }, { "epoch": 0.21020536510999233, "grad_norm": 2.9106662273406982, "learning_rate": 1.8345141643709394e-05, "loss": 0.16966676712036133, "step": 7980 }, { "epoch": 0.21046878035449107, "grad_norm": 3.425417900085449, "learning_rate": 1.8340437936268667e-05, "loss": 0.1717716336250305, "step": 7990 }, { "epoch": 0.2107321955989898, "grad_norm": 3.8346147537231445, "learning_rate": 1.8335728158297374e-05, "loss": 0.17628781795501708, "step": 8000 }, { "epoch": 0.21099561084348853, "grad_norm": 0.8516485095024109, "learning_rate": 1.833101231322349e-05, "loss": 0.16926485300064087, "step": 8010 }, { "epoch": 0.21125902608798727, "grad_norm": 3.068061113357544, "learning_rate": 1.832629040447942e-05, "loss": 0.17728766202926635, "step": 8020 }, { "epoch": 0.21152244133248602, "grad_norm": 2.4842169284820557, "learning_rate": 1.8321562435501963e-05, "loss": 0.1872403383255005, "step": 8030 }, { "epoch": 0.21178585657698476, "grad_norm": 3.502918004989624, "learning_rate": 1.831682840973234e-05, "loss": 0.16465312242507935, "step": 8040 }, { "epoch": 0.21204927182148348, "grad_norm": 1.8608253002166748, "learning_rate": 1.8312088330616188e-05, "loss": 0.17403874397277833, "step": 8050 }, { "epoch": 0.21231268706598222, "grad_norm": 1.6936085224151611, "learning_rate": 1.8307342201603533e-05, "loss": 0.1619790554046631, "step": 8060 }, { "epoch": 0.21257610231048096, "grad_norm": 2.4120798110961914, "learning_rate": 1.830259002614881e-05, "loss": 0.15212080478668213, "step": 8070 }, { "epoch": 0.2128395175549797, "grad_norm": 1.1923518180847168, "learning_rate": 1.829783180771087e-05, "loss": 0.17571139335632324, "step": 8080 }, { "epoch": 0.21310293279947845, "grad_norm": 2.2596004009246826, "learning_rate": 1.8293067549752937e-05, "loss": 0.16750702857971192, "step": 8090 }, { "epoch": 0.21336634804397717, "grad_norm": 1.4074236154556274, "learning_rate": 1.8288297255742646e-05, "loss": 0.18897922039031984, "step": 8100 }, { "epoch": 0.2136297632884759, "grad_norm": 3.2156765460968018, "learning_rate": 1.828352092915203e-05, "loss": 0.17232608795166016, "step": 8110 }, { "epoch": 0.21389317853297465, "grad_norm": 1.1303377151489258, "learning_rate": 1.8278738573457496e-05, "loss": 0.15231051445007324, "step": 8120 }, { "epoch": 0.2141565937774734, "grad_norm": 0.9594984650611877, "learning_rate": 1.8273950192139854e-05, "loss": 0.16860682964324952, "step": 8130 }, { "epoch": 0.2144200090219721, "grad_norm": 4.005578994750977, "learning_rate": 1.826915578868429e-05, "loss": 0.17909016609191894, "step": 8140 }, { "epoch": 0.21468342426647086, "grad_norm": 3.8453660011291504, "learning_rate": 1.8264355366580387e-05, "loss": 0.16558070182800294, "step": 8150 }, { "epoch": 0.2149468395109696, "grad_norm": 1.2554932832717896, "learning_rate": 1.825954892932209e-05, "loss": 0.16216516494750977, "step": 8160 }, { "epoch": 0.21521025475546834, "grad_norm": 1.7446955442428589, "learning_rate": 1.8254736480407738e-05, "loss": 0.16154589653015136, "step": 8170 }, { "epoch": 0.21547366999996706, "grad_norm": 1.757982850074768, "learning_rate": 1.8249918023340036e-05, "loss": 0.15902376174926758, "step": 8180 }, { "epoch": 0.2157370852444658, "grad_norm": 3.209899663925171, "learning_rate": 1.8245093561626063e-05, "loss": 0.17327922582626343, "step": 8190 }, { "epoch": 0.21600050048896455, "grad_norm": 4.2555251121521, "learning_rate": 1.824026309877727e-05, "loss": 0.17914742231369019, "step": 8200 }, { "epoch": 0.2162639157334633, "grad_norm": 4.239006519317627, "learning_rate": 1.8235426638309483e-05, "loss": 0.17818307876586914, "step": 8210 }, { "epoch": 0.21652733097796203, "grad_norm": 1.1237552165985107, "learning_rate": 1.823058418374288e-05, "loss": 0.1771291971206665, "step": 8220 }, { "epoch": 0.21679074622246075, "grad_norm": 1.6851556301116943, "learning_rate": 1.822573573860201e-05, "loss": 0.16261463165283202, "step": 8230 }, { "epoch": 0.2170541614669595, "grad_norm": 0.5752393007278442, "learning_rate": 1.8220881306415783e-05, "loss": 0.16041592359542847, "step": 8240 }, { "epoch": 0.21731757671145824, "grad_norm": 1.5373705625534058, "learning_rate": 1.821602089071746e-05, "loss": 0.16052380800247193, "step": 8250 }, { "epoch": 0.21758099195595698, "grad_norm": 1.5128740072250366, "learning_rate": 1.8211154495044666e-05, "loss": 0.1726743459701538, "step": 8260 }, { "epoch": 0.2178444072004557, "grad_norm": 0.9535961151123047, "learning_rate": 1.8206282122939368e-05, "loss": 0.15116674900054933, "step": 8270 }, { "epoch": 0.21810782244495444, "grad_norm": 2.6714415550231934, "learning_rate": 1.8201403777947896e-05, "loss": 0.15430495738983155, "step": 8280 }, { "epoch": 0.21837123768945318, "grad_norm": 1.1364001035690308, "learning_rate": 1.8196519463620912e-05, "loss": 0.16514291763305664, "step": 8290 }, { "epoch": 0.21863465293395193, "grad_norm": 2.0994296073913574, "learning_rate": 1.8191629183513435e-05, "loss": 0.15993478298187255, "step": 8300 }, { "epoch": 0.21889806817845064, "grad_norm": 1.2616156339645386, "learning_rate": 1.8186732941184824e-05, "loss": 0.16645395755767822, "step": 8310 }, { "epoch": 0.2191614834229494, "grad_norm": 1.8125791549682617, "learning_rate": 1.818183074019877e-05, "loss": 0.16402847766876222, "step": 8320 }, { "epoch": 0.21942489866744813, "grad_norm": 2.7023918628692627, "learning_rate": 1.8176922584123308e-05, "loss": 0.15896377563476563, "step": 8330 }, { "epoch": 0.21968831391194688, "grad_norm": 3.822690010070801, "learning_rate": 1.817200847653081e-05, "loss": 0.1871480703353882, "step": 8340 }, { "epoch": 0.21995172915644562, "grad_norm": 1.3560961484909058, "learning_rate": 1.8167088420997975e-05, "loss": 0.18162355422973633, "step": 8350 }, { "epoch": 0.22021514440094433, "grad_norm": 2.5879030227661133, "learning_rate": 1.816216242110582e-05, "loss": 0.1739673376083374, "step": 8360 }, { "epoch": 0.22047855964544308, "grad_norm": 1.4366779327392578, "learning_rate": 1.815723048043972e-05, "loss": 0.15444252490997315, "step": 8370 }, { "epoch": 0.22074197488994182, "grad_norm": 1.2034937143325806, "learning_rate": 1.8152292602589336e-05, "loss": 0.16794877052307128, "step": 8380 }, { "epoch": 0.22100539013444057, "grad_norm": 1.0910500288009644, "learning_rate": 1.8147348791148678e-05, "loss": 0.17969624996185302, "step": 8390 }, { "epoch": 0.22126880537893928, "grad_norm": 1.036106824874878, "learning_rate": 1.8142399049716065e-05, "loss": 0.1585320234298706, "step": 8400 }, { "epoch": 0.22153222062343803, "grad_norm": 1.5667574405670166, "learning_rate": 1.8137443381894126e-05, "loss": 0.17537500858306884, "step": 8410 }, { "epoch": 0.22179563586793677, "grad_norm": 1.4150336980819702, "learning_rate": 1.813248179128982e-05, "loss": 0.15979326963424684, "step": 8420 }, { "epoch": 0.2220590511124355, "grad_norm": 0.8739526271820068, "learning_rate": 1.81275142815144e-05, "loss": 0.17160232067108155, "step": 8430 }, { "epoch": 0.22232246635693423, "grad_norm": 2.8867576122283936, "learning_rate": 1.812254085618343e-05, "loss": 0.17140371799468995, "step": 8440 }, { "epoch": 0.22258588160143297, "grad_norm": 3.1865670680999756, "learning_rate": 1.8117561518916792e-05, "loss": 0.1677507519721985, "step": 8450 }, { "epoch": 0.22284929684593172, "grad_norm": 1.7749861478805542, "learning_rate": 1.811257627333866e-05, "loss": 0.1640796422958374, "step": 8460 }, { "epoch": 0.22311271209043046, "grad_norm": 0.6798821091651917, "learning_rate": 1.810758512307751e-05, "loss": 0.1647491693496704, "step": 8470 }, { "epoch": 0.2233761273349292, "grad_norm": 4.414216041564941, "learning_rate": 1.8102588071766116e-05, "loss": 0.16936638355255126, "step": 8480 }, { "epoch": 0.22363954257942792, "grad_norm": 3.832895517349243, "learning_rate": 1.8097585123041547e-05, "loss": 0.15951838493347167, "step": 8490 }, { "epoch": 0.22390295782392666, "grad_norm": 3.928422689437866, "learning_rate": 1.809257628054517e-05, "loss": 0.16204739809036256, "step": 8500 }, { "epoch": 0.2241663730684254, "grad_norm": 1.7303009033203125, "learning_rate": 1.808756154792263e-05, "loss": 0.1639163613319397, "step": 8510 }, { "epoch": 0.22442978831292415, "grad_norm": 3.3537349700927734, "learning_rate": 1.8082540928823866e-05, "loss": 0.17022280693054198, "step": 8520 }, { "epoch": 0.22469320355742287, "grad_norm": 1.1381927728652954, "learning_rate": 1.8077514426903107e-05, "loss": 0.16847851276397705, "step": 8530 }, { "epoch": 0.2249566188019216, "grad_norm": 2.7599594593048096, "learning_rate": 1.8072482045818852e-05, "loss": 0.1544037342071533, "step": 8540 }, { "epoch": 0.22522003404642035, "grad_norm": 2.758650779724121, "learning_rate": 1.806744378923389e-05, "loss": 0.16810257434844972, "step": 8550 }, { "epoch": 0.2254834492909191, "grad_norm": 2.461770534515381, "learning_rate": 1.8062399660815277e-05, "loss": 0.16780993938446045, "step": 8560 }, { "epoch": 0.2257468645354178, "grad_norm": 1.6087642908096313, "learning_rate": 1.8057349664234348e-05, "loss": 0.1535573959350586, "step": 8570 }, { "epoch": 0.22601027977991656, "grad_norm": 2.9401779174804688, "learning_rate": 1.8052293803166714e-05, "loss": 0.1739874839782715, "step": 8580 }, { "epoch": 0.2262736950244153, "grad_norm": 2.4148335456848145, "learning_rate": 1.8047232081292242e-05, "loss": 0.16164658069610596, "step": 8590 }, { "epoch": 0.22653711026891404, "grad_norm": 2.1201369762420654, "learning_rate": 1.8042164502295074e-05, "loss": 0.17979495525360106, "step": 8600 }, { "epoch": 0.2268005255134128, "grad_norm": 0.546754002571106, "learning_rate": 1.8037091069863615e-05, "loss": 0.16347250938415528, "step": 8610 }, { "epoch": 0.2270639407579115, "grad_norm": 1.7927920818328857, "learning_rate": 1.8032011787690523e-05, "loss": 0.15808253288269042, "step": 8620 }, { "epoch": 0.22732735600241025, "grad_norm": 4.018331050872803, "learning_rate": 1.8026926659472727e-05, "loss": 0.157732355594635, "step": 8630 }, { "epoch": 0.227590771246909, "grad_norm": 1.8864773511886597, "learning_rate": 1.802183568891139e-05, "loss": 0.16508127450942994, "step": 8640 }, { "epoch": 0.22785418649140773, "grad_norm": 2.904212713241577, "learning_rate": 1.801673887971195e-05, "loss": 0.17300665378570557, "step": 8650 }, { "epoch": 0.22811760173590645, "grad_norm": 1.5542161464691162, "learning_rate": 1.801163623558408e-05, "loss": 0.17264738082885742, "step": 8660 }, { "epoch": 0.2283810169804052, "grad_norm": 1.8326663970947266, "learning_rate": 1.800652776024171e-05, "loss": 0.16461869478225707, "step": 8670 }, { "epoch": 0.22864443222490394, "grad_norm": 2.3098151683807373, "learning_rate": 1.8001413457403005e-05, "loss": 0.16393136978149414, "step": 8680 }, { "epoch": 0.22890784746940268, "grad_norm": 1.910910964012146, "learning_rate": 1.7996293330790374e-05, "loss": 0.16562087535858155, "step": 8690 }, { "epoch": 0.2291712627139014, "grad_norm": 1.1152573823928833, "learning_rate": 1.7991167384130466e-05, "loss": 0.16165882349014282, "step": 8700 }, { "epoch": 0.22943467795840014, "grad_norm": 1.9222813844680786, "learning_rate": 1.798603562115417e-05, "loss": 0.16311440467834473, "step": 8710 }, { "epoch": 0.22969809320289888, "grad_norm": 1.0290639400482178, "learning_rate": 1.79808980455966e-05, "loss": 0.16182172298431396, "step": 8720 }, { "epoch": 0.22996150844739763, "grad_norm": 1.1843317747116089, "learning_rate": 1.7975754661197106e-05, "loss": 0.16491856575012206, "step": 8730 }, { "epoch": 0.23022492369189637, "grad_norm": 1.941758155822754, "learning_rate": 1.797060547169927e-05, "loss": 0.1687218427658081, "step": 8740 }, { "epoch": 0.2304883389363951, "grad_norm": 2.0262539386749268, "learning_rate": 1.796545048085089e-05, "loss": 0.165781569480896, "step": 8750 }, { "epoch": 0.23075175418089383, "grad_norm": 1.0734362602233887, "learning_rate": 1.796028969240399e-05, "loss": 0.16549750566482543, "step": 8760 }, { "epoch": 0.23101516942539257, "grad_norm": 1.8760873079299927, "learning_rate": 1.7955123110114815e-05, "loss": 0.16795213222503663, "step": 8770 }, { "epoch": 0.23127858466989132, "grad_norm": 0.7167704701423645, "learning_rate": 1.794995073774383e-05, "loss": 0.1512695550918579, "step": 8780 }, { "epoch": 0.23154199991439003, "grad_norm": 0.8262736797332764, "learning_rate": 1.7944772579055705e-05, "loss": 0.1726707935333252, "step": 8790 }, { "epoch": 0.23180541515888878, "grad_norm": 3.6269257068634033, "learning_rate": 1.793958863781933e-05, "loss": 0.16516927480697632, "step": 8800 }, { "epoch": 0.23206883040338752, "grad_norm": 2.56542706489563, "learning_rate": 1.7934398917807802e-05, "loss": 0.19798727035522462, "step": 8810 }, { "epoch": 0.23233224564788627, "grad_norm": 1.5771878957748413, "learning_rate": 1.7929203422798426e-05, "loss": 0.16238535642623902, "step": 8820 }, { "epoch": 0.23259566089238498, "grad_norm": 1.4299075603485107, "learning_rate": 1.79240021565727e-05, "loss": 0.15599751472473145, "step": 8830 }, { "epoch": 0.23285907613688372, "grad_norm": 6.251887321472168, "learning_rate": 1.7918795122916337e-05, "loss": 0.16086426973342896, "step": 8840 }, { "epoch": 0.23312249138138247, "grad_norm": 0.7924928665161133, "learning_rate": 1.7913582325619236e-05, "loss": 0.1677180767059326, "step": 8850 }, { "epoch": 0.2333859066258812, "grad_norm": 4.477551460266113, "learning_rate": 1.7908363768475498e-05, "loss": 0.17074579000473022, "step": 8860 }, { "epoch": 0.23364932187037996, "grad_norm": 3.1063365936279297, "learning_rate": 1.7903139455283412e-05, "loss": 0.1663001298904419, "step": 8870 }, { "epoch": 0.23391273711487867, "grad_norm": 4.917378902435303, "learning_rate": 1.7897909389845462e-05, "loss": 0.16903306245803834, "step": 8880 }, { "epoch": 0.23417615235937742, "grad_norm": 6.27949333190918, "learning_rate": 1.789267357596831e-05, "loss": 0.1583454728126526, "step": 8890 }, { "epoch": 0.23443956760387616, "grad_norm": 1.421332836151123, "learning_rate": 1.7887432017462812e-05, "loss": 0.17438818216323854, "step": 8900 }, { "epoch": 0.2347029828483749, "grad_norm": 5.997462272644043, "learning_rate": 1.7882184718144002e-05, "loss": 0.17892694473266602, "step": 8910 }, { "epoch": 0.23496639809287362, "grad_norm": 2.9228973388671875, "learning_rate": 1.7876931681831087e-05, "loss": 0.16608589887619019, "step": 8920 }, { "epoch": 0.23522981333737236, "grad_norm": 1.8963850736618042, "learning_rate": 1.787167291234745e-05, "loss": 0.16159223318099974, "step": 8930 }, { "epoch": 0.2354932285818711, "grad_norm": 6.600646018981934, "learning_rate": 1.7866408413520662e-05, "loss": 0.16078169345855714, "step": 8940 }, { "epoch": 0.23575664382636985, "grad_norm": 1.3020014762878418, "learning_rate": 1.786113818918244e-05, "loss": 0.18176519870758057, "step": 8950 }, { "epoch": 0.23602005907086857, "grad_norm": 2.5187509059906006, "learning_rate": 1.7855862243168684e-05, "loss": 0.1592639684677124, "step": 8960 }, { "epoch": 0.2362834743153673, "grad_norm": 2.7679600715637207, "learning_rate": 1.785058057931946e-05, "loss": 0.16840684413909912, "step": 8970 }, { "epoch": 0.23654688955986605, "grad_norm": 4.033356666564941, "learning_rate": 1.7845293201478985e-05, "loss": 0.185532283782959, "step": 8980 }, { "epoch": 0.2368103048043648, "grad_norm": 1.7695573568344116, "learning_rate": 1.7840000113495647e-05, "loss": 0.16610854864120483, "step": 8990 }, { "epoch": 0.23707372004886354, "grad_norm": 6.348013401031494, "learning_rate": 1.7834701319221978e-05, "loss": 0.16632871627807616, "step": 9000 }, { "epoch": 0.23733713529336226, "grad_norm": 1.471947431564331, "learning_rate": 1.782939682251467e-05, "loss": 0.16530967950820924, "step": 9010 }, { "epoch": 0.237600550537861, "grad_norm": 3.016101121902466, "learning_rate": 1.7824086627234565e-05, "loss": 0.16818511486053467, "step": 9020 }, { "epoch": 0.23786396578235974, "grad_norm": 4.687798023223877, "learning_rate": 1.7818770737246653e-05, "loss": 0.17051640748977662, "step": 9030 }, { "epoch": 0.2381273810268585, "grad_norm": 0.8874970078468323, "learning_rate": 1.7813449156420072e-05, "loss": 0.14003100395202636, "step": 9040 }, { "epoch": 0.2383907962713572, "grad_norm": 2.2560031414031982, "learning_rate": 1.7808121888628093e-05, "loss": 0.17206714153289795, "step": 9050 }, { "epoch": 0.23865421151585595, "grad_norm": 1.1795598268508911, "learning_rate": 1.780278893774813e-05, "loss": 0.16662967205047607, "step": 9060 }, { "epoch": 0.2389176267603547, "grad_norm": 2.6536166667938232, "learning_rate": 1.7797450307661733e-05, "loss": 0.16411380767822265, "step": 9070 }, { "epoch": 0.23918104200485343, "grad_norm": 3.156895875930786, "learning_rate": 1.77921060022546e-05, "loss": 0.16791642904281617, "step": 9080 }, { "epoch": 0.23944445724935215, "grad_norm": 1.3276143074035645, "learning_rate": 1.7786756025416535e-05, "loss": 0.17198946475982665, "step": 9090 }, { "epoch": 0.2397078724938509, "grad_norm": 2.3232853412628174, "learning_rate": 1.7781400381041484e-05, "loss": 0.16960501670837402, "step": 9100 }, { "epoch": 0.23997128773834964, "grad_norm": 1.1325123310089111, "learning_rate": 1.777603907302752e-05, "loss": 0.16681809425354005, "step": 9110 }, { "epoch": 0.24023470298284838, "grad_norm": 1.8951051235198975, "learning_rate": 1.7770672105276833e-05, "loss": 0.145361328125, "step": 9120 }, { "epoch": 0.24049811822734712, "grad_norm": 3.080198049545288, "learning_rate": 1.776529948169573e-05, "loss": 0.16762816905975342, "step": 9130 }, { "epoch": 0.24076153347184584, "grad_norm": 1.8066705465316772, "learning_rate": 1.775992120619464e-05, "loss": 0.17394027709960938, "step": 9140 }, { "epoch": 0.24102494871634458, "grad_norm": 1.892926812171936, "learning_rate": 1.7754537282688104e-05, "loss": 0.16103861331939698, "step": 9150 }, { "epoch": 0.24128836396084333, "grad_norm": 3.4507274627685547, "learning_rate": 1.7749147715094777e-05, "loss": 0.15574970245361328, "step": 9160 }, { "epoch": 0.24155177920534207, "grad_norm": 1.2994704246520996, "learning_rate": 1.7743752507337414e-05, "loss": 0.17930738925933837, "step": 9170 }, { "epoch": 0.2418151944498408, "grad_norm": 2.8673343658447266, "learning_rate": 1.7738351663342887e-05, "loss": 0.17165014743804932, "step": 9180 }, { "epoch": 0.24207860969433953, "grad_norm": 1.580974817276001, "learning_rate": 1.7732945187042155e-05, "loss": 0.16201283931732177, "step": 9190 }, { "epoch": 0.24234202493883827, "grad_norm": 1.4002517461776733, "learning_rate": 1.7727533082370288e-05, "loss": 0.15073100328445435, "step": 9200 }, { "epoch": 0.24260544018333702, "grad_norm": 1.21250319480896, "learning_rate": 1.772211535326645e-05, "loss": 0.1801051139831543, "step": 9210 }, { "epoch": 0.24286885542783573, "grad_norm": 1.9876132011413574, "learning_rate": 1.7716692003673898e-05, "loss": 0.15822746753692626, "step": 9220 }, { "epoch": 0.24313227067233448, "grad_norm": 2.8591182231903076, "learning_rate": 1.7711263037539978e-05, "loss": 0.17523643970489503, "step": 9230 }, { "epoch": 0.24339568591683322, "grad_norm": 2.6306967735290527, "learning_rate": 1.770582845881613e-05, "loss": 0.1667020797729492, "step": 9240 }, { "epoch": 0.24365910116133196, "grad_norm": 1.7947511672973633, "learning_rate": 1.7700388271457872e-05, "loss": 0.1684574842453003, "step": 9250 }, { "epoch": 0.2439225164058307, "grad_norm": 1.1368975639343262, "learning_rate": 1.7694942479424803e-05, "loss": 0.1614912986755371, "step": 9260 }, { "epoch": 0.24418593165032942, "grad_norm": 1.7260822057724, "learning_rate": 1.768949108668061e-05, "loss": 0.15541499853134155, "step": 9270 }, { "epoch": 0.24444934689482817, "grad_norm": 1.6663665771484375, "learning_rate": 1.7684034097193055e-05, "loss": 0.1531246781349182, "step": 9280 }, { "epoch": 0.2447127621393269, "grad_norm": 5.894384384155273, "learning_rate": 1.7678571514933966e-05, "loss": 0.16822348833084105, "step": 9290 }, { "epoch": 0.24497617738382566, "grad_norm": 4.416995048522949, "learning_rate": 1.7673103343879242e-05, "loss": 0.16470463275909425, "step": 9300 }, { "epoch": 0.24523959262832437, "grad_norm": 2.042837142944336, "learning_rate": 1.7667629588008862e-05, "loss": 0.156737756729126, "step": 9310 }, { "epoch": 0.24550300787282311, "grad_norm": 1.718830943107605, "learning_rate": 1.766215025130686e-05, "loss": 0.1592686414718628, "step": 9320 }, { "epoch": 0.24576642311732186, "grad_norm": 1.0593997240066528, "learning_rate": 1.7656665337761327e-05, "loss": 0.16592564582824706, "step": 9330 }, { "epoch": 0.2460298383618206, "grad_norm": 1.4023406505584717, "learning_rate": 1.7651174851364427e-05, "loss": 0.15763157606124878, "step": 9340 }, { "epoch": 0.24629325360631932, "grad_norm": 7.004781723022461, "learning_rate": 1.764567879611237e-05, "loss": 0.16237661838531495, "step": 9350 }, { "epoch": 0.24655666885081806, "grad_norm": 2.9889674186706543, "learning_rate": 1.7640177176005425e-05, "loss": 0.16114722490310668, "step": 9360 }, { "epoch": 0.2468200840953168, "grad_norm": 1.2702510356903076, "learning_rate": 1.7634669995047904e-05, "loss": 0.17826415300369264, "step": 9370 }, { "epoch": 0.24708349933981555, "grad_norm": 1.880155324935913, "learning_rate": 1.7629157257248175e-05, "loss": 0.1567068338394165, "step": 9380 }, { "epoch": 0.2473469145843143, "grad_norm": 1.0574346780776978, "learning_rate": 1.762363896661864e-05, "loss": 0.15714157819747926, "step": 9390 }, { "epoch": 0.247610329828813, "grad_norm": 5.1040472984313965, "learning_rate": 1.761811512717576e-05, "loss": 0.16760213375091554, "step": 9400 }, { "epoch": 0.24787374507331175, "grad_norm": 1.270359754562378, "learning_rate": 1.761258574294001e-05, "loss": 0.14441289901733398, "step": 9410 }, { "epoch": 0.2481371603178105, "grad_norm": 2.840437173843384, "learning_rate": 1.760705081793593e-05, "loss": 0.16748071908950807, "step": 9420 }, { "epoch": 0.24840057556230924, "grad_norm": 3.8460018634796143, "learning_rate": 1.760151035619206e-05, "loss": 0.18048491477966308, "step": 9430 }, { "epoch": 0.24866399080680796, "grad_norm": 0.8584071397781372, "learning_rate": 1.7595964361741e-05, "loss": 0.16119476556777954, "step": 9440 }, { "epoch": 0.2489274060513067, "grad_norm": 3.8055295944213867, "learning_rate": 1.759041283861936e-05, "loss": 0.1729937434196472, "step": 9450 }, { "epoch": 0.24919082129580544, "grad_norm": 1.3219271898269653, "learning_rate": 1.7584855790867773e-05, "loss": 0.1590951919555664, "step": 9460 }, { "epoch": 0.2494542365403042, "grad_norm": 1.0455197095870972, "learning_rate": 1.7579293222530907e-05, "loss": 0.1660757303237915, "step": 9470 }, { "epoch": 0.24971765178480293, "grad_norm": 1.4723292589187622, "learning_rate": 1.7573725137657432e-05, "loss": 0.16077876091003418, "step": 9480 }, { "epoch": 0.24998106702930165, "grad_norm": 3.036785840988159, "learning_rate": 1.756815154030005e-05, "loss": 0.1613665223121643, "step": 9490 }, { "epoch": 0.2502444822738004, "grad_norm": 1.3807061910629272, "learning_rate": 1.756257243451545e-05, "loss": 0.15721571445465088, "step": 9500 }, { "epoch": 0.2505078975182991, "grad_norm": 1.120832920074463, "learning_rate": 1.7556987824364363e-05, "loss": 0.16646871566772461, "step": 9510 }, { "epoch": 0.25077131276279785, "grad_norm": 2.6750566959381104, "learning_rate": 1.75513977139115e-05, "loss": 0.163405179977417, "step": 9520 }, { "epoch": 0.2510347280072966, "grad_norm": 5.764256477355957, "learning_rate": 1.7545802107225585e-05, "loss": 0.17567172050476074, "step": 9530 }, { "epoch": 0.25129814325179534, "grad_norm": 1.2527354955673218, "learning_rate": 1.754020100837934e-05, "loss": 0.1686413049697876, "step": 9540 }, { "epoch": 0.2515615584962941, "grad_norm": 1.4089504480361938, "learning_rate": 1.753459442144949e-05, "loss": 0.1741097688674927, "step": 9550 }, { "epoch": 0.2518249737407928, "grad_norm": 4.918815612792969, "learning_rate": 1.7528982350516754e-05, "loss": 0.16421000957489013, "step": 9560 }, { "epoch": 0.25208838898529157, "grad_norm": 3.193937301635742, "learning_rate": 1.7523364799665832e-05, "loss": 0.158728289604187, "step": 9570 }, { "epoch": 0.2523518042297903, "grad_norm": 2.3161563873291016, "learning_rate": 1.7517741772985424e-05, "loss": 0.1723085403442383, "step": 9580 }, { "epoch": 0.252615219474289, "grad_norm": 1.4920235872268677, "learning_rate": 1.751211327456821e-05, "loss": 0.1691701054573059, "step": 9590 }, { "epoch": 0.25287863471878774, "grad_norm": 4.860597133636475, "learning_rate": 1.7506479308510855e-05, "loss": 0.1663133144378662, "step": 9600 }, { "epoch": 0.2531420499632865, "grad_norm": 4.8989667892456055, "learning_rate": 1.7500839878914004e-05, "loss": 0.1576486349105835, "step": 9610 }, { "epoch": 0.25340546520778523, "grad_norm": 2.858391523361206, "learning_rate": 1.749519498988227e-05, "loss": 0.17685399055480958, "step": 9620 }, { "epoch": 0.253668880452284, "grad_norm": 4.258740425109863, "learning_rate": 1.748954464552426e-05, "loss": 0.16658849716186525, "step": 9630 }, { "epoch": 0.2539322956967827, "grad_norm": 1.0467760562896729, "learning_rate": 1.7483888849952525e-05, "loss": 0.17339508533477782, "step": 9640 }, { "epoch": 0.25419571094128146, "grad_norm": 2.9826066493988037, "learning_rate": 1.7478227607283604e-05, "loss": 0.15285853147506714, "step": 9650 }, { "epoch": 0.2544591261857802, "grad_norm": 1.9731366634368896, "learning_rate": 1.7472560921637993e-05, "loss": 0.16232600212097167, "step": 9660 }, { "epoch": 0.25472254143027895, "grad_norm": 4.898499488830566, "learning_rate": 1.746688879714015e-05, "loss": 0.1630505681037903, "step": 9670 }, { "epoch": 0.25498595667477764, "grad_norm": 1.2731608152389526, "learning_rate": 1.746121123791849e-05, "loss": 0.16516265869140626, "step": 9680 }, { "epoch": 0.2552493719192764, "grad_norm": 4.5742106437683105, "learning_rate": 1.7455528248105385e-05, "loss": 0.1741267681121826, "step": 9690 }, { "epoch": 0.2555127871637751, "grad_norm": 1.2059874534606934, "learning_rate": 1.7449839831837166e-05, "loss": 0.17127046585083008, "step": 9700 }, { "epoch": 0.25577620240827387, "grad_norm": 2.1164426803588867, "learning_rate": 1.7444145993254105e-05, "loss": 0.14964035749435425, "step": 9710 }, { "epoch": 0.2560396176527726, "grad_norm": 1.3333783149719238, "learning_rate": 1.7438446736500418e-05, "loss": 0.1787782669067383, "step": 9720 }, { "epoch": 0.25630303289727135, "grad_norm": 2.4947381019592285, "learning_rate": 1.7432742065724276e-05, "loss": 0.1663106322288513, "step": 9730 }, { "epoch": 0.2565664481417701, "grad_norm": 2.768012523651123, "learning_rate": 1.7427031985077785e-05, "loss": 0.16493306159973145, "step": 9740 }, { "epoch": 0.25682986338626884, "grad_norm": 1.3557130098342896, "learning_rate": 1.7421316498716986e-05, "loss": 0.14735229015350343, "step": 9750 }, { "epoch": 0.2570932786307676, "grad_norm": 3.684373140335083, "learning_rate": 1.741559561080185e-05, "loss": 0.15304951667785643, "step": 9760 }, { "epoch": 0.2573566938752663, "grad_norm": 1.8048173189163208, "learning_rate": 1.74098693254963e-05, "loss": 0.14261555671691895, "step": 9770 }, { "epoch": 0.257620109119765, "grad_norm": 1.8212414979934692, "learning_rate": 1.7404137646968162e-05, "loss": 0.1660293698310852, "step": 9780 }, { "epoch": 0.25788352436426376, "grad_norm": 2.070185661315918, "learning_rate": 1.7398400579389204e-05, "loss": 0.15558416843414308, "step": 9790 }, { "epoch": 0.2581469396087625, "grad_norm": 2.8650565147399902, "learning_rate": 1.739265812693511e-05, "loss": 0.16668099164962769, "step": 9800 }, { "epoch": 0.25841035485326125, "grad_norm": 4.19657564163208, "learning_rate": 1.7386910293785485e-05, "loss": 0.15788636207580567, "step": 9810 }, { "epoch": 0.25867377009776, "grad_norm": 2.2906787395477295, "learning_rate": 1.738115708412385e-05, "loss": 0.15153143405914307, "step": 9820 }, { "epoch": 0.25893718534225874, "grad_norm": 1.7824980020523071, "learning_rate": 1.737539850213764e-05, "loss": 0.16558294296264647, "step": 9830 }, { "epoch": 0.2592006005867575, "grad_norm": 2.2492425441741943, "learning_rate": 1.7369634552018196e-05, "loss": 0.16128814220428467, "step": 9840 }, { "epoch": 0.25946401583125617, "grad_norm": 1.1325523853302002, "learning_rate": 1.7363865237960773e-05, "loss": 0.16194649934768676, "step": 9850 }, { "epoch": 0.2597274310757549, "grad_norm": 2.8635146617889404, "learning_rate": 1.7358090564164524e-05, "loss": 0.156399405002594, "step": 9860 }, { "epoch": 0.25999084632025365, "grad_norm": 1.472009539604187, "learning_rate": 1.735231053483251e-05, "loss": 0.16099319458007813, "step": 9870 }, { "epoch": 0.2602542615647524, "grad_norm": 0.9001330137252808, "learning_rate": 1.734652515417168e-05, "loss": 0.17213070392608643, "step": 9880 }, { "epoch": 0.26051767680925114, "grad_norm": 0.8792181611061096, "learning_rate": 1.7340734426392894e-05, "loss": 0.16357797384262085, "step": 9890 }, { "epoch": 0.2607810920537499, "grad_norm": 2.799056053161621, "learning_rate": 1.7334938355710876e-05, "loss": 0.1753699541091919, "step": 9900 }, { "epoch": 0.26104450729824863, "grad_norm": 1.3747551441192627, "learning_rate": 1.7329136946344272e-05, "loss": 0.1723180055618286, "step": 9910 }, { "epoch": 0.2613079225427474, "grad_norm": 0.9993804097175598, "learning_rate": 1.7323330202515592e-05, "loss": 0.161040997505188, "step": 9920 }, { "epoch": 0.2615713377872461, "grad_norm": 4.044465065002441, "learning_rate": 1.731751812845124e-05, "loss": 0.18333306312561035, "step": 9930 }, { "epoch": 0.2618347530317448, "grad_norm": 1.8562572002410889, "learning_rate": 1.7311700728381486e-05, "loss": 0.1618468165397644, "step": 9940 }, { "epoch": 0.26209816827624355, "grad_norm": 2.6322269439697266, "learning_rate": 1.7305878006540483e-05, "loss": 0.16122015714645385, "step": 9950 }, { "epoch": 0.2623615835207423, "grad_norm": 1.6385127305984497, "learning_rate": 1.7300049967166273e-05, "loss": 0.16125862598419188, "step": 9960 }, { "epoch": 0.26262499876524104, "grad_norm": 1.6955251693725586, "learning_rate": 1.729421661450074e-05, "loss": 0.15182697772979736, "step": 9970 }, { "epoch": 0.2628884140097398, "grad_norm": 2.0420210361480713, "learning_rate": 1.728837795278966e-05, "loss": 0.15763320922851562, "step": 9980 }, { "epoch": 0.2631518292542385, "grad_norm": 3.748227834701538, "learning_rate": 1.7282533986282656e-05, "loss": 0.16590504646301268, "step": 9990 }, { "epoch": 0.26341524449873727, "grad_norm": 9.725698471069336, "learning_rate": 1.7276684719233222e-05, "loss": 0.17882503271102906, "step": 10000 }, { "epoch": 0.263678659743236, "grad_norm": 1.6858304738998413, "learning_rate": 1.7270830155898704e-05, "loss": 0.1843624711036682, "step": 10010 }, { "epoch": 0.26394207498773475, "grad_norm": 1.8605207204818726, "learning_rate": 1.726497030054031e-05, "loss": 0.16960929632186889, "step": 10020 }, { "epoch": 0.26420549023223344, "grad_norm": 0.8358961343765259, "learning_rate": 1.72591051574231e-05, "loss": 0.16057127714157104, "step": 10030 }, { "epoch": 0.2644689054767322, "grad_norm": 1.5059683322906494, "learning_rate": 1.7253234730815964e-05, "loss": 0.17904698848724365, "step": 10040 }, { "epoch": 0.26473232072123093, "grad_norm": 0.7544341683387756, "learning_rate": 1.724735902499166e-05, "loss": 0.17043159008026124, "step": 10050 }, { "epoch": 0.2649957359657297, "grad_norm": 1.180409550666809, "learning_rate": 1.7241478044226782e-05, "loss": 0.1712867021560669, "step": 10060 }, { "epoch": 0.2652591512102284, "grad_norm": 1.1467630863189697, "learning_rate": 1.723559179280176e-05, "loss": 0.1675874948501587, "step": 10070 }, { "epoch": 0.26552256645472716, "grad_norm": 1.6263492107391357, "learning_rate": 1.722970027500086e-05, "loss": 0.16977763175964355, "step": 10080 }, { "epoch": 0.2657859816992259, "grad_norm": 3.242255210876465, "learning_rate": 1.7223803495112185e-05, "loss": 0.16149367094039918, "step": 10090 }, { "epoch": 0.26604939694372465, "grad_norm": 1.13413405418396, "learning_rate": 1.7217901457427667e-05, "loss": 0.1491476058959961, "step": 10100 }, { "epoch": 0.26631281218822334, "grad_norm": 3.3149642944335938, "learning_rate": 1.7211994166243058e-05, "loss": 0.1668346881866455, "step": 10110 }, { "epoch": 0.2665762274327221, "grad_norm": 1.3609217405319214, "learning_rate": 1.7206081625857943e-05, "loss": 0.17354416847229004, "step": 10120 }, { "epoch": 0.2668396426772208, "grad_norm": 3.135505437850952, "learning_rate": 1.7200163840575728e-05, "loss": 0.16609044075012208, "step": 10130 }, { "epoch": 0.26710305792171957, "grad_norm": 3.287947654724121, "learning_rate": 1.7194240814703626e-05, "loss": 0.1629781723022461, "step": 10140 }, { "epoch": 0.2673664731662183, "grad_norm": 2.429792881011963, "learning_rate": 1.7188312552552672e-05, "loss": 0.16484073400497437, "step": 10150 }, { "epoch": 0.26762988841071705, "grad_norm": 2.0951647758483887, "learning_rate": 1.718237905843771e-05, "loss": 0.16781973838806152, "step": 10160 }, { "epoch": 0.2678933036552158, "grad_norm": 3.0717930793762207, "learning_rate": 1.71764403366774e-05, "loss": 0.1663385033607483, "step": 10170 }, { "epoch": 0.26815671889971454, "grad_norm": 1.8054217100143433, "learning_rate": 1.7170496391594194e-05, "loss": 0.15503692626953125, "step": 10180 }, { "epoch": 0.2684201341442133, "grad_norm": 2.206944227218628, "learning_rate": 1.716454722751435e-05, "loss": 0.15935828685760497, "step": 10190 }, { "epoch": 0.268683549388712, "grad_norm": 1.6931134462356567, "learning_rate": 1.715859284876793e-05, "loss": 0.16410541534423828, "step": 10200 }, { "epoch": 0.2689469646332107, "grad_norm": 0.8303804993629456, "learning_rate": 1.7152633259688788e-05, "loss": 0.1588420867919922, "step": 10210 }, { "epoch": 0.26921037987770946, "grad_norm": 5.18996524810791, "learning_rate": 1.714666846461457e-05, "loss": 0.1583399534225464, "step": 10220 }, { "epoch": 0.2694737951222082, "grad_norm": 1.3493494987487793, "learning_rate": 1.7140698467886707e-05, "loss": 0.15646122694015502, "step": 10230 }, { "epoch": 0.26973721036670695, "grad_norm": 0.9807963967323303, "learning_rate": 1.713472327385042e-05, "loss": 0.15867644548416138, "step": 10240 }, { "epoch": 0.2700006256112057, "grad_norm": 2.3757119178771973, "learning_rate": 1.7128742886854723e-05, "loss": 0.15437426567077636, "step": 10250 }, { "epoch": 0.27026404085570443, "grad_norm": 2.6976184844970703, "learning_rate": 1.712275731125239e-05, "loss": 0.16878772974014283, "step": 10260 }, { "epoch": 0.2705274561002032, "grad_norm": 6.278127193450928, "learning_rate": 1.7116766551399986e-05, "loss": 0.15786556005477906, "step": 10270 }, { "epoch": 0.2707908713447019, "grad_norm": 1.1371541023254395, "learning_rate": 1.711077061165784e-05, "loss": 0.16343193054199218, "step": 10280 }, { "epoch": 0.2710542865892006, "grad_norm": 3.71608304977417, "learning_rate": 1.710476949639006e-05, "loss": 0.17173961400985718, "step": 10290 }, { "epoch": 0.27131770183369935, "grad_norm": 2.6103386878967285, "learning_rate": 1.7098763209964514e-05, "loss": 0.15582338571548462, "step": 10300 }, { "epoch": 0.2715811170781981, "grad_norm": 1.4415000677108765, "learning_rate": 1.7092751756752838e-05, "loss": 0.15247159004211425, "step": 10310 }, { "epoch": 0.27184453232269684, "grad_norm": 1.7643744945526123, "learning_rate": 1.7086735141130423e-05, "loss": 0.16914632320404052, "step": 10320 }, { "epoch": 0.2721079475671956, "grad_norm": 3.320629835128784, "learning_rate": 1.7080713367476428e-05, "loss": 0.16042124032974242, "step": 10330 }, { "epoch": 0.27237136281169433, "grad_norm": 0.7510604858398438, "learning_rate": 1.7074686440173752e-05, "loss": 0.17079837322235109, "step": 10340 }, { "epoch": 0.2726347780561931, "grad_norm": 1.510626196861267, "learning_rate": 1.7068654363609058e-05, "loss": 0.150113308429718, "step": 10350 }, { "epoch": 0.2728981933006918, "grad_norm": 2.6565606594085693, "learning_rate": 1.7062617142172754e-05, "loss": 0.17418532371520995, "step": 10360 }, { "epoch": 0.2731616085451905, "grad_norm": 2.4990153312683105, "learning_rate": 1.7056574780258986e-05, "loss": 0.16879881620407106, "step": 10370 }, { "epoch": 0.27342502378968925, "grad_norm": 2.874338388442993, "learning_rate": 1.705052728226564e-05, "loss": 0.1574535608291626, "step": 10380 }, { "epoch": 0.273688439034188, "grad_norm": 1.4296990633010864, "learning_rate": 1.704447465259436e-05, "loss": 0.16112793684005738, "step": 10390 }, { "epoch": 0.27395185427868674, "grad_norm": 2.883319616317749, "learning_rate": 1.7038416895650504e-05, "loss": 0.1602324962615967, "step": 10400 }, { "epoch": 0.2742152695231855, "grad_norm": 2.1264002323150635, "learning_rate": 1.7032354015843162e-05, "loss": 0.16664502620697022, "step": 10410 }, { "epoch": 0.2744786847676842, "grad_norm": 4.976718902587891, "learning_rate": 1.7026286017585168e-05, "loss": 0.15955450534820556, "step": 10420 }, { "epoch": 0.27474210001218297, "grad_norm": 1.8965716361999512, "learning_rate": 1.7020212905293072e-05, "loss": 0.16976935863494874, "step": 10430 }, { "epoch": 0.2750055152566817, "grad_norm": 1.5467193126678467, "learning_rate": 1.7014134683387147e-05, "loss": 0.16914024353027343, "step": 10440 }, { "epoch": 0.27526893050118045, "grad_norm": 1.2824007272720337, "learning_rate": 1.7008051356291382e-05, "loss": 0.1600520968437195, "step": 10450 }, { "epoch": 0.27553234574567914, "grad_norm": 1.071985125541687, "learning_rate": 1.7001962928433488e-05, "loss": 0.1693236470222473, "step": 10460 }, { "epoch": 0.2757957609901779, "grad_norm": 2.684215784072876, "learning_rate": 1.699586940424488e-05, "loss": 0.1750227689743042, "step": 10470 }, { "epoch": 0.27605917623467663, "grad_norm": 1.4290105104446411, "learning_rate": 1.6989770788160697e-05, "loss": 0.16643695831298827, "step": 10480 }, { "epoch": 0.2763225914791754, "grad_norm": 5.176525592803955, "learning_rate": 1.6983667084619768e-05, "loss": 0.16953725814819337, "step": 10490 }, { "epoch": 0.2765860067236741, "grad_norm": 1.9429701566696167, "learning_rate": 1.6977558298064636e-05, "loss": 0.16557977199554444, "step": 10500 }, { "epoch": 0.27684942196817286, "grad_norm": 2.402991533279419, "learning_rate": 1.6971444432941533e-05, "loss": 0.15807536840438843, "step": 10510 }, { "epoch": 0.2771128372126716, "grad_norm": 1.4283201694488525, "learning_rate": 1.69653254937004e-05, "loss": 0.16668376922607422, "step": 10520 }, { "epoch": 0.27737625245717035, "grad_norm": 1.4989160299301147, "learning_rate": 1.6959201484794867e-05, "loss": 0.16354482173919677, "step": 10530 }, { "epoch": 0.2776396677016691, "grad_norm": 1.1340045928955078, "learning_rate": 1.6953072410682243e-05, "loss": 0.1766970157623291, "step": 10540 }, { "epoch": 0.2779030829461678, "grad_norm": 4.314626693725586, "learning_rate": 1.694693827582354e-05, "loss": 0.1744058609008789, "step": 10550 }, { "epoch": 0.2781664981906665, "grad_norm": 2.554692506790161, "learning_rate": 1.694079908468344e-05, "loss": 0.15582860708236695, "step": 10560 }, { "epoch": 0.27842991343516527, "grad_norm": 2.0945818424224854, "learning_rate": 1.6934654841730318e-05, "loss": 0.16522890329360962, "step": 10570 }, { "epoch": 0.278693328679664, "grad_norm": 1.8489258289337158, "learning_rate": 1.6928505551436215e-05, "loss": 0.16829500198364258, "step": 10580 }, { "epoch": 0.27895674392416275, "grad_norm": 2.906353235244751, "learning_rate": 1.6922351218276852e-05, "loss": 0.16366132497787475, "step": 10590 }, { "epoch": 0.2792201591686615, "grad_norm": 3.1140546798706055, "learning_rate": 1.6916191846731615e-05, "loss": 0.15805951356887818, "step": 10600 }, { "epoch": 0.27948357441316024, "grad_norm": 3.328979730606079, "learning_rate": 1.6910027441283562e-05, "loss": 0.17527223825454713, "step": 10610 }, { "epoch": 0.279746989657659, "grad_norm": 2.48987078666687, "learning_rate": 1.6903858006419413e-05, "loss": 0.17960891723632813, "step": 10620 }, { "epoch": 0.2800104049021577, "grad_norm": 1.1361377239227295, "learning_rate": 1.689768354662955e-05, "loss": 0.16581238508224488, "step": 10630 }, { "epoch": 0.2802738201466564, "grad_norm": 2.216179609298706, "learning_rate": 1.689150406640801e-05, "loss": 0.1664319396018982, "step": 10640 }, { "epoch": 0.28053723539115516, "grad_norm": 1.9865158796310425, "learning_rate": 1.6885319570252485e-05, "loss": 0.16623914241790771, "step": 10650 }, { "epoch": 0.2808006506356539, "grad_norm": 3.4731733798980713, "learning_rate": 1.687913006266432e-05, "loss": 0.1564274549484253, "step": 10660 }, { "epoch": 0.28106406588015265, "grad_norm": 2.7204036712646484, "learning_rate": 1.6872935548148503e-05, "loss": 0.16801011562347412, "step": 10670 }, { "epoch": 0.2813274811246514, "grad_norm": 3.478161573410034, "learning_rate": 1.686673603121367e-05, "loss": 0.16165293455123902, "step": 10680 }, { "epoch": 0.28159089636915013, "grad_norm": 4.691917419433594, "learning_rate": 1.6860531516372102e-05, "loss": 0.1705470561981201, "step": 10690 }, { "epoch": 0.2818543116136489, "grad_norm": 1.765404462814331, "learning_rate": 1.6854322008139704e-05, "loss": 0.1616927742958069, "step": 10700 }, { "epoch": 0.2821177268581476, "grad_norm": 1.0414763689041138, "learning_rate": 1.6848107511036026e-05, "loss": 0.1650221586227417, "step": 10710 }, { "epoch": 0.2823811421026463, "grad_norm": 2.2459661960601807, "learning_rate": 1.6841888029584255e-05, "loss": 0.14839355945587157, "step": 10720 }, { "epoch": 0.28264455734714505, "grad_norm": 1.9454152584075928, "learning_rate": 1.6835663568311187e-05, "loss": 0.14949328899383546, "step": 10730 }, { "epoch": 0.2829079725916438, "grad_norm": 1.7181732654571533, "learning_rate": 1.682943413174726e-05, "loss": 0.16656481027603148, "step": 10740 }, { "epoch": 0.28317138783614254, "grad_norm": 2.1810972690582275, "learning_rate": 1.6823199724426523e-05, "loss": 0.17310655117034912, "step": 10750 }, { "epoch": 0.2834348030806413, "grad_norm": 1.630421757698059, "learning_rate": 1.6816960350886653e-05, "loss": 0.1742750287055969, "step": 10760 }, { "epoch": 0.28369821832514003, "grad_norm": 3.568652629852295, "learning_rate": 1.681071601566893e-05, "loss": 0.16677199602127074, "step": 10770 }, { "epoch": 0.28396163356963877, "grad_norm": 1.7243444919586182, "learning_rate": 1.6804466723318248e-05, "loss": 0.16863741874694824, "step": 10780 }, { "epoch": 0.2842250488141375, "grad_norm": 4.268502712249756, "learning_rate": 1.6798212478383118e-05, "loss": 0.16382439136505128, "step": 10790 }, { "epoch": 0.28448846405863626, "grad_norm": 3.0435984134674072, "learning_rate": 1.6791953285415645e-05, "loss": 0.15659186840057374, "step": 10800 }, { "epoch": 0.28475187930313495, "grad_norm": 2.7115702629089355, "learning_rate": 1.678568914897154e-05, "loss": 0.16617949008941652, "step": 10810 }, { "epoch": 0.2850152945476337, "grad_norm": 5.290095329284668, "learning_rate": 1.6779420073610116e-05, "loss": 0.16419715881347657, "step": 10820 }, { "epoch": 0.28527870979213243, "grad_norm": 6.081679821014404, "learning_rate": 1.6773146063894273e-05, "loss": 0.1814349889755249, "step": 10830 }, { "epoch": 0.2855421250366312, "grad_norm": 6.589959144592285, "learning_rate": 1.6766867124390505e-05, "loss": 0.16421384811401368, "step": 10840 }, { "epoch": 0.2858055402811299, "grad_norm": 2.7706680297851562, "learning_rate": 1.6760583259668897e-05, "loss": 0.16218190193176268, "step": 10850 }, { "epoch": 0.28606895552562867, "grad_norm": 1.1384458541870117, "learning_rate": 1.6754294474303116e-05, "loss": 0.16063772439956664, "step": 10860 }, { "epoch": 0.2863323707701274, "grad_norm": 1.2858425378799438, "learning_rate": 1.674800077287041e-05, "loss": 0.1689133882522583, "step": 10870 }, { "epoch": 0.28659578601462615, "grad_norm": 3.2248010635375977, "learning_rate": 1.6741702159951613e-05, "loss": 0.16612186431884765, "step": 10880 }, { "epoch": 0.2868592012591249, "grad_norm": 1.8185428380966187, "learning_rate": 1.6735398640131117e-05, "loss": 0.16680171489715576, "step": 10890 }, { "epoch": 0.2871226165036236, "grad_norm": 1.681066870689392, "learning_rate": 1.6729090217996904e-05, "loss": 0.16474003791809083, "step": 10900 }, { "epoch": 0.28738603174812233, "grad_norm": 1.776991844177246, "learning_rate": 1.6722776898140513e-05, "loss": 0.16309726238250732, "step": 10910 }, { "epoch": 0.28764944699262107, "grad_norm": 5.052682399749756, "learning_rate": 1.6716458685157047e-05, "loss": 0.16440720558166505, "step": 10920 }, { "epoch": 0.2879128622371198, "grad_norm": 1.792630910873413, "learning_rate": 1.671013558364518e-05, "loss": 0.1601893424987793, "step": 10930 }, { "epoch": 0.28817627748161856, "grad_norm": 6.50290060043335, "learning_rate": 1.6703807598207132e-05, "loss": 0.1641981601715088, "step": 10940 }, { "epoch": 0.2884396927261173, "grad_norm": 1.2072668075561523, "learning_rate": 1.669747473344869e-05, "loss": 0.1532044529914856, "step": 10950 }, { "epoch": 0.28870310797061605, "grad_norm": 3.560117721557617, "learning_rate": 1.6691136993979183e-05, "loss": 0.14998893737792968, "step": 10960 }, { "epoch": 0.2889665232151148, "grad_norm": 1.9968262910842896, "learning_rate": 1.6684794384411494e-05, "loss": 0.1571497440338135, "step": 10970 }, { "epoch": 0.2892299384596135, "grad_norm": 1.5447105169296265, "learning_rate": 1.667844690936204e-05, "loss": 0.16695429086685182, "step": 10980 }, { "epoch": 0.2894933537041122, "grad_norm": 2.4092793464660645, "learning_rate": 1.6672094573450798e-05, "loss": 0.18046700954437256, "step": 10990 }, { "epoch": 0.28975676894861097, "grad_norm": 1.7419984340667725, "learning_rate": 1.6665737381301263e-05, "loss": 0.16796635389328002, "step": 11000 }, { "epoch": 0.2900201841931097, "grad_norm": 0.8821855783462524, "learning_rate": 1.665937533754048e-05, "loss": 0.1555652618408203, "step": 11010 }, { "epoch": 0.29028359943760845, "grad_norm": 2.166727304458618, "learning_rate": 1.6653008446799014e-05, "loss": 0.16985946893692017, "step": 11020 }, { "epoch": 0.2905470146821072, "grad_norm": 0.8867211937904358, "learning_rate": 1.6646636713710967e-05, "loss": 0.16140105724334716, "step": 11030 }, { "epoch": 0.29081042992660594, "grad_norm": 3.142256259918213, "learning_rate": 1.664026014291396e-05, "loss": 0.17453646659851074, "step": 11040 }, { "epoch": 0.2910738451711047, "grad_norm": 4.292820930480957, "learning_rate": 1.6633878739049135e-05, "loss": 0.162191641330719, "step": 11050 }, { "epoch": 0.2913372604156034, "grad_norm": 2.8926777839660645, "learning_rate": 1.6627492506761155e-05, "loss": 0.1727575421333313, "step": 11060 }, { "epoch": 0.2916006756601021, "grad_norm": 2.0280048847198486, "learning_rate": 1.6621101450698193e-05, "loss": 0.17256181240081786, "step": 11070 }, { "epoch": 0.29186409090460086, "grad_norm": 2.2571053504943848, "learning_rate": 1.6614705575511937e-05, "loss": 0.16898051500320435, "step": 11080 }, { "epoch": 0.2921275061490996, "grad_norm": 3.2090907096862793, "learning_rate": 1.660830488585758e-05, "loss": 0.15685064792633058, "step": 11090 }, { "epoch": 0.29239092139359835, "grad_norm": 2.1729891300201416, "learning_rate": 1.660189938639382e-05, "loss": 0.1577543020248413, "step": 11100 }, { "epoch": 0.2926543366380971, "grad_norm": 1.3693526983261108, "learning_rate": 1.6595489081782858e-05, "loss": 0.1667433500289917, "step": 11110 }, { "epoch": 0.29291775188259583, "grad_norm": 1.3022279739379883, "learning_rate": 1.6589073976690387e-05, "loss": 0.168997859954834, "step": 11120 }, { "epoch": 0.2931811671270946, "grad_norm": 2.9788241386413574, "learning_rate": 1.65826540757856e-05, "loss": 0.1606459379196167, "step": 11130 }, { "epoch": 0.2934445823715933, "grad_norm": 1.6969257593154907, "learning_rate": 1.6576229383741177e-05, "loss": 0.1668178915977478, "step": 11140 }, { "epoch": 0.29370799761609206, "grad_norm": 1.890663743019104, "learning_rate": 1.6569799905233285e-05, "loss": 0.16062326431274415, "step": 11150 }, { "epoch": 0.29397141286059075, "grad_norm": 1.70845365524292, "learning_rate": 1.6563365644941578e-05, "loss": 0.16515648365020752, "step": 11160 }, { "epoch": 0.2942348281050895, "grad_norm": 2.5350229740142822, "learning_rate": 1.6556926607549186e-05, "loss": 0.16005698442459107, "step": 11170 }, { "epoch": 0.29449824334958824, "grad_norm": 3.285327196121216, "learning_rate": 1.655048279774272e-05, "loss": 0.16758904457092286, "step": 11180 }, { "epoch": 0.294761658594087, "grad_norm": 3.162418842315674, "learning_rate": 1.6544034220212262e-05, "loss": 0.1618780493736267, "step": 11190 }, { "epoch": 0.2950250738385857, "grad_norm": 1.1810342073440552, "learning_rate": 1.6537580879651364e-05, "loss": 0.15990326404571534, "step": 11200 }, { "epoch": 0.29528848908308447, "grad_norm": 1.6282745599746704, "learning_rate": 1.6531122780757046e-05, "loss": 0.17074034214019776, "step": 11210 }, { "epoch": 0.2955519043275832, "grad_norm": 3.562847852706909, "learning_rate": 1.6524659928229792e-05, "loss": 0.16992653608322145, "step": 11220 }, { "epoch": 0.29581531957208196, "grad_norm": 5.389997482299805, "learning_rate": 1.6518192326773546e-05, "loss": 0.16491588354110717, "step": 11230 }, { "epoch": 0.29607873481658065, "grad_norm": 3.3076746463775635, "learning_rate": 1.6511719981095703e-05, "loss": 0.17751611471176149, "step": 11240 }, { "epoch": 0.2963421500610794, "grad_norm": 3.321488857269287, "learning_rate": 1.6505242895907122e-05, "loss": 0.16879100799560548, "step": 11250 }, { "epoch": 0.29660556530557813, "grad_norm": 2.3391315937042236, "learning_rate": 1.64987610759221e-05, "loss": 0.16759494543075562, "step": 11260 }, { "epoch": 0.2968689805500769, "grad_norm": 2.0394439697265625, "learning_rate": 1.6492274525858386e-05, "loss": 0.16416202783584594, "step": 11270 }, { "epoch": 0.2971323957945756, "grad_norm": 1.7024699449539185, "learning_rate": 1.648578325043717e-05, "loss": 0.1604258179664612, "step": 11280 }, { "epoch": 0.29739581103907436, "grad_norm": 0.8760567903518677, "learning_rate": 1.6479287254383084e-05, "loss": 0.17243671417236328, "step": 11290 }, { "epoch": 0.2976592262835731, "grad_norm": 1.9681932926177979, "learning_rate": 1.6472786542424196e-05, "loss": 0.1491257905960083, "step": 11300 }, { "epoch": 0.29792264152807185, "grad_norm": 2.980560302734375, "learning_rate": 1.6466281119292004e-05, "loss": 0.17556809186935424, "step": 11310 }, { "epoch": 0.2981860567725706, "grad_norm": 1.7533862590789795, "learning_rate": 1.6459770989721433e-05, "loss": 0.17628107070922852, "step": 11320 }, { "epoch": 0.2984494720170693, "grad_norm": 3.9475035667419434, "learning_rate": 1.6453256158450835e-05, "loss": 0.16591638326644897, "step": 11330 }, { "epoch": 0.298712887261568, "grad_norm": 1.0172587633132935, "learning_rate": 1.644673663022199e-05, "loss": 0.1658787488937378, "step": 11340 }, { "epoch": 0.29897630250606677, "grad_norm": 3.082465648651123, "learning_rate": 1.6440212409780088e-05, "loss": 0.1727348804473877, "step": 11350 }, { "epoch": 0.2992397177505655, "grad_norm": 0.9696842432022095, "learning_rate": 1.643368350187374e-05, "loss": 0.17799081802368164, "step": 11360 }, { "epoch": 0.29950313299506426, "grad_norm": 2.8087120056152344, "learning_rate": 1.6427149911254973e-05, "loss": 0.15511380434036254, "step": 11370 }, { "epoch": 0.299766548239563, "grad_norm": 1.905851125717163, "learning_rate": 1.64206116426792e-05, "loss": 0.15149754285812378, "step": 11380 }, { "epoch": 0.30002996348406175, "grad_norm": 1.2876412868499756, "learning_rate": 1.641406870090527e-05, "loss": 0.15492615699768067, "step": 11390 }, { "epoch": 0.3002933787285605, "grad_norm": 1.561937689781189, "learning_rate": 1.640752109069541e-05, "loss": 0.16446017026901244, "step": 11400 }, { "epoch": 0.30055679397305923, "grad_norm": 0.8077186346054077, "learning_rate": 1.640096881681526e-05, "loss": 0.1642158031463623, "step": 11410 }, { "epoch": 0.3008202092175579, "grad_norm": 2.659479856491089, "learning_rate": 1.6394411884033844e-05, "loss": 0.16909767389297486, "step": 11420 }, { "epoch": 0.30108362446205666, "grad_norm": 0.979556143283844, "learning_rate": 1.638785029712358e-05, "loss": 0.16505942344665528, "step": 11430 }, { "epoch": 0.3013470397065554, "grad_norm": 3.009455442428589, "learning_rate": 1.638128406086027e-05, "loss": 0.17094357013702394, "step": 11440 }, { "epoch": 0.30161045495105415, "grad_norm": 1.7486342191696167, "learning_rate": 1.6374713180023114e-05, "loss": 0.15654873847961426, "step": 11450 }, { "epoch": 0.3018738701955529, "grad_norm": 0.988720178604126, "learning_rate": 1.6368137659394673e-05, "loss": 0.1656602144241333, "step": 11460 }, { "epoch": 0.30213728544005164, "grad_norm": 3.8658294677734375, "learning_rate": 1.63615575037609e-05, "loss": 0.18056061267852783, "step": 11470 }, { "epoch": 0.3024007006845504, "grad_norm": 5.6874823570251465, "learning_rate": 1.6354972717911115e-05, "loss": 0.14566688537597655, "step": 11480 }, { "epoch": 0.3026641159290491, "grad_norm": 4.095505237579346, "learning_rate": 1.634838330663801e-05, "loss": 0.16516975164413453, "step": 11490 }, { "epoch": 0.3029275311735478, "grad_norm": 3.3114511966705322, "learning_rate": 1.6341789274737644e-05, "loss": 0.16385245323181152, "step": 11500 }, { "epoch": 0.30319094641804656, "grad_norm": 1.5097795724868774, "learning_rate": 1.6335190627009432e-05, "loss": 0.17118849754333496, "step": 11510 }, { "epoch": 0.3034543616625453, "grad_norm": 4.74098014831543, "learning_rate": 1.632858736825616e-05, "loss": 0.15499483346939086, "step": 11520 }, { "epoch": 0.30371777690704405, "grad_norm": 1.965922236442566, "learning_rate": 1.632197950328396e-05, "loss": 0.15990114212036133, "step": 11530 }, { "epoch": 0.3039811921515428, "grad_norm": 2.208751916885376, "learning_rate": 1.631536703690233e-05, "loss": 0.16275135278701783, "step": 11540 }, { "epoch": 0.30424460739604153, "grad_norm": 2.0929505825042725, "learning_rate": 1.6308749973924096e-05, "loss": 0.16572580337524415, "step": 11550 }, { "epoch": 0.3045080226405403, "grad_norm": 1.6768109798431396, "learning_rate": 1.6302128319165454e-05, "loss": 0.1719291925430298, "step": 11560 }, { "epoch": 0.304771437885039, "grad_norm": 2.099522113800049, "learning_rate": 1.629550207744592e-05, "loss": 0.17303261756896973, "step": 11570 }, { "epoch": 0.30503485312953776, "grad_norm": 1.0574778318405151, "learning_rate": 1.6288871253588366e-05, "loss": 0.16289408206939698, "step": 11580 }, { "epoch": 0.30529826837403645, "grad_norm": 3.468829393386841, "learning_rate": 1.6282235852418983e-05, "loss": 0.17289677858352662, "step": 11590 }, { "epoch": 0.3055616836185352, "grad_norm": 1.9332518577575684, "learning_rate": 1.6275595878767315e-05, "loss": 0.16368546485900878, "step": 11600 }, { "epoch": 0.30582509886303394, "grad_norm": 1.4137574434280396, "learning_rate": 1.6268951337466212e-05, "loss": 0.16169341802597045, "step": 11610 }, { "epoch": 0.3060885141075327, "grad_norm": 1.1113361120224, "learning_rate": 1.626230223335186e-05, "loss": 0.1707548141479492, "step": 11620 }, { "epoch": 0.3063519293520314, "grad_norm": 1.0584062337875366, "learning_rate": 1.6255648571263762e-05, "loss": 0.17133946418762208, "step": 11630 }, { "epoch": 0.30661534459653017, "grad_norm": 2.9722650051116943, "learning_rate": 1.6248990356044742e-05, "loss": 0.17691736221313475, "step": 11640 }, { "epoch": 0.3068787598410289, "grad_norm": 2.72941255569458, "learning_rate": 1.6242327592540942e-05, "loss": 0.15838658809661865, "step": 11650 }, { "epoch": 0.30714217508552766, "grad_norm": 1.406390905380249, "learning_rate": 1.62356602856018e-05, "loss": 0.16426489353179932, "step": 11660 }, { "epoch": 0.3074055903300264, "grad_norm": 2.330312728881836, "learning_rate": 1.6228988440080068e-05, "loss": 0.15866429805755616, "step": 11670 }, { "epoch": 0.3076690055745251, "grad_norm": 3.29801344871521, "learning_rate": 1.6222312060831814e-05, "loss": 0.15335700511932374, "step": 11680 }, { "epoch": 0.30793242081902383, "grad_norm": 3.2738678455352783, "learning_rate": 1.6215631152716387e-05, "loss": 0.17218682765960694, "step": 11690 }, { "epoch": 0.3081958360635226, "grad_norm": 2.1365878582000732, "learning_rate": 1.620894572059644e-05, "loss": 0.16849242448806762, "step": 11700 }, { "epoch": 0.3084592513080213, "grad_norm": 1.9926338195800781, "learning_rate": 1.6202255769337924e-05, "loss": 0.1500581383705139, "step": 11710 }, { "epoch": 0.30872266655252006, "grad_norm": 1.938713550567627, "learning_rate": 1.619556130381007e-05, "loss": 0.16617100238800048, "step": 11720 }, { "epoch": 0.3089860817970188, "grad_norm": 2.385643243789673, "learning_rate": 1.61888623288854e-05, "loss": 0.178570556640625, "step": 11730 }, { "epoch": 0.30924949704151755, "grad_norm": 3.0208756923675537, "learning_rate": 1.6182158849439718e-05, "loss": 0.15452828407287597, "step": 11740 }, { "epoch": 0.3095129122860163, "grad_norm": 2.3420040607452393, "learning_rate": 1.61754508703521e-05, "loss": 0.16012009382247924, "step": 11750 }, { "epoch": 0.309776327530515, "grad_norm": 2.2588369846343994, "learning_rate": 1.6168738396504912e-05, "loss": 0.177504301071167, "step": 11760 }, { "epoch": 0.3100397427750137, "grad_norm": 3.3298962116241455, "learning_rate": 1.6162021432783773e-05, "loss": 0.16277997493743895, "step": 11770 }, { "epoch": 0.31030315801951247, "grad_norm": 3.3241584300994873, "learning_rate": 1.6155299984077584e-05, "loss": 0.16260052919387818, "step": 11780 }, { "epoch": 0.3105665732640112, "grad_norm": 0.9193066358566284, "learning_rate": 1.6148574055278507e-05, "loss": 0.1394677758216858, "step": 11790 }, { "epoch": 0.31082998850850996, "grad_norm": 2.4113245010375977, "learning_rate": 1.6141843651281955e-05, "loss": 0.15499608516693114, "step": 11800 }, { "epoch": 0.3110934037530087, "grad_norm": 2.7709827423095703, "learning_rate": 1.6135108776986615e-05, "loss": 0.17120856046676636, "step": 11810 }, { "epoch": 0.31135681899750745, "grad_norm": 0.8421139121055603, "learning_rate": 1.612836943729442e-05, "loss": 0.14654258489608765, "step": 11820 }, { "epoch": 0.3116202342420062, "grad_norm": 2.6580350399017334, "learning_rate": 1.6121625637110546e-05, "loss": 0.17255182266235353, "step": 11830 }, { "epoch": 0.31188364948650493, "grad_norm": 2.213434934616089, "learning_rate": 1.6114877381343422e-05, "loss": 0.16378207206726075, "step": 11840 }, { "epoch": 0.3121470647310036, "grad_norm": 2.3642630577087402, "learning_rate": 1.610812467490472e-05, "loss": 0.1765371084213257, "step": 11850 }, { "epoch": 0.31241047997550236, "grad_norm": 3.570404052734375, "learning_rate": 1.6101367522709356e-05, "loss": 0.15860459804534913, "step": 11860 }, { "epoch": 0.3126738952200011, "grad_norm": 2.6558146476745605, "learning_rate": 1.6094605929675476e-05, "loss": 0.1534063220024109, "step": 11870 }, { "epoch": 0.31293731046449985, "grad_norm": 1.4943879842758179, "learning_rate": 1.608783990072445e-05, "loss": 0.16433295011520385, "step": 11880 }, { "epoch": 0.3132007257089986, "grad_norm": 3.9556305408477783, "learning_rate": 1.6081069440780895e-05, "loss": 0.16492291688919067, "step": 11890 }, { "epoch": 0.31346414095349734, "grad_norm": 2.304826021194458, "learning_rate": 1.607429455477264e-05, "loss": 0.14827194213867187, "step": 11900 }, { "epoch": 0.3137275561979961, "grad_norm": 0.8324774503707886, "learning_rate": 1.606751524763075e-05, "loss": 0.14791879653930665, "step": 11910 }, { "epoch": 0.3139909714424948, "grad_norm": 1.2449836730957031, "learning_rate": 1.606073152428948e-05, "loss": 0.15616493225097655, "step": 11920 }, { "epoch": 0.31425438668699357, "grad_norm": 2.4629764556884766, "learning_rate": 1.6053943389686323e-05, "loss": 0.15749013423919678, "step": 11930 }, { "epoch": 0.31451780193149226, "grad_norm": 1.928178310394287, "learning_rate": 1.604715084876198e-05, "loss": 0.1560465931892395, "step": 11940 }, { "epoch": 0.314781217175991, "grad_norm": 1.009657859802246, "learning_rate": 1.6040353906460353e-05, "loss": 0.16817398071289064, "step": 11950 }, { "epoch": 0.31504463242048975, "grad_norm": 1.7378902435302734, "learning_rate": 1.6033552567728553e-05, "loss": 0.15097591876983643, "step": 11960 }, { "epoch": 0.3153080476649885, "grad_norm": 1.647192120552063, "learning_rate": 1.602674683751688e-05, "loss": 0.1695850133895874, "step": 11970 }, { "epoch": 0.31557146290948723, "grad_norm": 2.030133008956909, "learning_rate": 1.6019936720778845e-05, "loss": 0.1619551420211792, "step": 11980 }, { "epoch": 0.315834878153986, "grad_norm": 1.604456901550293, "learning_rate": 1.601312222247114e-05, "loss": 0.1647895097732544, "step": 11990 }, { "epoch": 0.3160982933984847, "grad_norm": 1.501383662223816, "learning_rate": 1.6006303347553655e-05, "loss": 0.17120485305786132, "step": 12000 }, { "epoch": 0.31636170864298346, "grad_norm": 1.4547632932662964, "learning_rate": 1.5999480100989458e-05, "loss": 0.168790602684021, "step": 12010 }, { "epoch": 0.31662512388748215, "grad_norm": 2.470362901687622, "learning_rate": 1.5992652487744806e-05, "loss": 0.1651597499847412, "step": 12020 }, { "epoch": 0.3168885391319809, "grad_norm": 0.8512497544288635, "learning_rate": 1.5985820512789128e-05, "loss": 0.1521388053894043, "step": 12030 }, { "epoch": 0.31715195437647964, "grad_norm": 2.4640891551971436, "learning_rate": 1.5978984181095034e-05, "loss": 0.16755187511444092, "step": 12040 }, { "epoch": 0.3174153696209784, "grad_norm": 2.0878474712371826, "learning_rate": 1.5972143497638294e-05, "loss": 0.16475595235824586, "step": 12050 }, { "epoch": 0.3176787848654771, "grad_norm": 1.9410008192062378, "learning_rate": 1.5965298467397855e-05, "loss": 0.16146080493927, "step": 12060 }, { "epoch": 0.31794220010997587, "grad_norm": 1.7157337665557861, "learning_rate": 1.595844909535583e-05, "loss": 0.16329643726348878, "step": 12070 }, { "epoch": 0.3182056153544746, "grad_norm": 1.5251624584197998, "learning_rate": 1.595159538649748e-05, "loss": 0.16436694860458373, "step": 12080 }, { "epoch": 0.31846903059897336, "grad_norm": 3.703779697418213, "learning_rate": 1.5944737345811238e-05, "loss": 0.1658177375793457, "step": 12090 }, { "epoch": 0.3187324458434721, "grad_norm": 0.6362970471382141, "learning_rate": 1.5937874978288676e-05, "loss": 0.15633871555328369, "step": 12100 }, { "epoch": 0.3189958610879708, "grad_norm": 1.130324363708496, "learning_rate": 1.5931008288924523e-05, "loss": 0.15796202421188354, "step": 12110 }, { "epoch": 0.31925927633246953, "grad_norm": 1.3212850093841553, "learning_rate": 1.5924137282716653e-05, "loss": 0.15752032995224, "step": 12120 }, { "epoch": 0.3195226915769683, "grad_norm": 1.8952856063842773, "learning_rate": 1.591726196466608e-05, "loss": 0.16822067499160767, "step": 12130 }, { "epoch": 0.319786106821467, "grad_norm": 0.8672000169754028, "learning_rate": 1.5910382339776953e-05, "loss": 0.1619802713394165, "step": 12140 }, { "epoch": 0.32004952206596576, "grad_norm": 3.268214702606201, "learning_rate": 1.5903498413056568e-05, "loss": 0.16947704553604126, "step": 12150 }, { "epoch": 0.3203129373104645, "grad_norm": 1.2459160089492798, "learning_rate": 1.5896610189515335e-05, "loss": 0.16258366107940675, "step": 12160 }, { "epoch": 0.32057635255496325, "grad_norm": 1.221131443977356, "learning_rate": 1.588971767416681e-05, "loss": 0.16707320213317872, "step": 12170 }, { "epoch": 0.320839767799462, "grad_norm": 4.604711055755615, "learning_rate": 1.5882820872027654e-05, "loss": 0.15980993509292601, "step": 12180 }, { "epoch": 0.32110318304396074, "grad_norm": 0.9479048848152161, "learning_rate": 1.5875919788117662e-05, "loss": 0.16379053592681886, "step": 12190 }, { "epoch": 0.3213665982884594, "grad_norm": 0.7813912034034729, "learning_rate": 1.586901442745974e-05, "loss": 0.1628089427947998, "step": 12200 }, { "epoch": 0.32163001353295817, "grad_norm": 2.48476505279541, "learning_rate": 1.5862104795079903e-05, "loss": 0.15012545585632325, "step": 12210 }, { "epoch": 0.3218934287774569, "grad_norm": 1.473235845565796, "learning_rate": 1.5855190896007287e-05, "loss": 0.18480687141418456, "step": 12220 }, { "epoch": 0.32215684402195566, "grad_norm": 0.9626656770706177, "learning_rate": 1.584827273527412e-05, "loss": 0.1471242666244507, "step": 12230 }, { "epoch": 0.3224202592664544, "grad_norm": 3.8089799880981445, "learning_rate": 1.5841350317915745e-05, "loss": 0.16191855669021607, "step": 12240 }, { "epoch": 0.32268367451095314, "grad_norm": 1.4339286088943481, "learning_rate": 1.5834423648970588e-05, "loss": 0.17242953777313233, "step": 12250 }, { "epoch": 0.3229470897554519, "grad_norm": 2.4746530055999756, "learning_rate": 1.5827492733480183e-05, "loss": 0.17091114521026612, "step": 12260 }, { "epoch": 0.32321050499995063, "grad_norm": 2.1003952026367188, "learning_rate": 1.5820557576489145e-05, "loss": 0.14633082151412963, "step": 12270 }, { "epoch": 0.3234739202444493, "grad_norm": 1.3465090990066528, "learning_rate": 1.581361818304518e-05, "loss": 0.1540090560913086, "step": 12280 }, { "epoch": 0.32373733548894806, "grad_norm": 2.0158443450927734, "learning_rate": 1.5806674558199088e-05, "loss": 0.17401129007339478, "step": 12290 }, { "epoch": 0.3240007507334468, "grad_norm": 0.9303451180458069, "learning_rate": 1.5799726707004728e-05, "loss": 0.17019643783569335, "step": 12300 }, { "epoch": 0.32426416597794555, "grad_norm": 1.5220251083374023, "learning_rate": 1.5792774634519046e-05, "loss": 0.17132353782653809, "step": 12310 }, { "epoch": 0.3245275812224443, "grad_norm": 2.8437769412994385, "learning_rate": 1.5785818345802068e-05, "loss": 0.16212716102600097, "step": 12320 }, { "epoch": 0.32479099646694304, "grad_norm": 1.514885663986206, "learning_rate": 1.5778857845916874e-05, "loss": 0.16456305980682373, "step": 12330 }, { "epoch": 0.3250544117114418, "grad_norm": 1.4115304946899414, "learning_rate": 1.577189313992962e-05, "loss": 0.15234118700027466, "step": 12340 }, { "epoch": 0.3253178269559405, "grad_norm": 0.8355842232704163, "learning_rate": 1.5764924232909518e-05, "loss": 0.16972019672393798, "step": 12350 }, { "epoch": 0.32558124220043927, "grad_norm": 1.0770502090454102, "learning_rate": 1.5757951129928837e-05, "loss": 0.16202982664108276, "step": 12360 }, { "epoch": 0.32584465744493796, "grad_norm": 0.6522654891014099, "learning_rate": 1.5750973836062907e-05, "loss": 0.15794668197631836, "step": 12370 }, { "epoch": 0.3261080726894367, "grad_norm": 3.067584991455078, "learning_rate": 1.57439923563901e-05, "loss": 0.16633697748184204, "step": 12380 }, { "epoch": 0.32637148793393544, "grad_norm": 1.736847996711731, "learning_rate": 1.573700669599184e-05, "loss": 0.1611424207687378, "step": 12390 }, { "epoch": 0.3266349031784342, "grad_norm": 0.9269422888755798, "learning_rate": 1.573001685995259e-05, "loss": 0.1480102300643921, "step": 12400 }, { "epoch": 0.32689831842293293, "grad_norm": 2.9883875846862793, "learning_rate": 1.572302285335985e-05, "loss": 0.1709004282951355, "step": 12410 }, { "epoch": 0.3271617336674317, "grad_norm": 1.3977121114730835, "learning_rate": 1.5716024681304177e-05, "loss": 0.1557589054107666, "step": 12420 }, { "epoch": 0.3274251489119304, "grad_norm": 3.0086331367492676, "learning_rate": 1.5709022348879124e-05, "loss": 0.16213662624359132, "step": 12430 }, { "epoch": 0.32768856415642916, "grad_norm": 1.1998307704925537, "learning_rate": 1.5702015861181295e-05, "loss": 0.16115376949310303, "step": 12440 }, { "epoch": 0.3279519794009279, "grad_norm": 4.521817684173584, "learning_rate": 1.569500522331032e-05, "loss": 0.16706936359405516, "step": 12450 }, { "epoch": 0.3282153946454266, "grad_norm": 4.9035468101501465, "learning_rate": 1.568799044036884e-05, "loss": 0.15946966409683228, "step": 12460 }, { "epoch": 0.32847880988992534, "grad_norm": 1.1549479961395264, "learning_rate": 1.5680971517462516e-05, "loss": 0.14752275943756105, "step": 12470 }, { "epoch": 0.3287422251344241, "grad_norm": 3.544814109802246, "learning_rate": 1.5673948459700024e-05, "loss": 0.1462238311767578, "step": 12480 }, { "epoch": 0.3290056403789228, "grad_norm": 2.394550323486328, "learning_rate": 1.5666921272193046e-05, "loss": 0.15609135627746581, "step": 12490 }, { "epoch": 0.32926905562342157, "grad_norm": 2.136793613433838, "learning_rate": 1.565988996005628e-05, "loss": 0.16593585014343262, "step": 12500 }, { "epoch": 0.3295324708679203, "grad_norm": 1.6164112091064453, "learning_rate": 1.5652854528407407e-05, "loss": 0.15311295986175538, "step": 12510 }, { "epoch": 0.32979588611241906, "grad_norm": 1.283537745475769, "learning_rate": 1.5645814982367125e-05, "loss": 0.1621862530708313, "step": 12520 }, { "epoch": 0.3300593013569178, "grad_norm": 0.9208908081054688, "learning_rate": 1.5638771327059117e-05, "loss": 0.16463770866394042, "step": 12530 }, { "epoch": 0.3303227166014165, "grad_norm": 3.32863712310791, "learning_rate": 1.563172356761006e-05, "loss": 0.16148539781570434, "step": 12540 }, { "epoch": 0.33058613184591523, "grad_norm": 0.7914148569107056, "learning_rate": 1.5624671709149618e-05, "loss": 0.15029433965682984, "step": 12550 }, { "epoch": 0.330849547090414, "grad_norm": 2.201467752456665, "learning_rate": 1.561761575681044e-05, "loss": 0.15574159622192382, "step": 12560 }, { "epoch": 0.3311129623349127, "grad_norm": 1.9135719537734985, "learning_rate": 1.5610555715728146e-05, "loss": 0.15326759815216065, "step": 12570 }, { "epoch": 0.33137637757941146, "grad_norm": 2.1477320194244385, "learning_rate": 1.560349159104134e-05, "loss": 0.16233391761779786, "step": 12580 }, { "epoch": 0.3316397928239102, "grad_norm": 2.3446784019470215, "learning_rate": 1.55964233878916e-05, "loss": 0.1607028603553772, "step": 12590 }, { "epoch": 0.33190320806840895, "grad_norm": 1.233993649482727, "learning_rate": 1.5589351111423474e-05, "loss": 0.15697981119155885, "step": 12600 }, { "epoch": 0.3321666233129077, "grad_norm": 2.265462636947632, "learning_rate": 1.5582274766784457e-05, "loss": 0.17667732238769532, "step": 12610 }, { "epoch": 0.33243003855740644, "grad_norm": 2.211516857147217, "learning_rate": 1.557519435912503e-05, "loss": 0.15212545394897461, "step": 12620 }, { "epoch": 0.3326934538019051, "grad_norm": 1.1499607563018799, "learning_rate": 1.556810989359861e-05, "loss": 0.16230915784835814, "step": 12630 }, { "epoch": 0.33295686904640387, "grad_norm": 3.572843074798584, "learning_rate": 1.5561021375361583e-05, "loss": 0.16817113161087036, "step": 12640 }, { "epoch": 0.3332202842909026, "grad_norm": 1.4487311840057373, "learning_rate": 1.5553928809573275e-05, "loss": 0.16591668128967285, "step": 12650 }, { "epoch": 0.33348369953540136, "grad_norm": 1.4479180574417114, "learning_rate": 1.5546832201395964e-05, "loss": 0.15896689891815186, "step": 12660 }, { "epoch": 0.3337471147799001, "grad_norm": 2.2414116859436035, "learning_rate": 1.5539731555994864e-05, "loss": 0.1591496706008911, "step": 12670 }, { "epoch": 0.33401053002439884, "grad_norm": 1.5464413166046143, "learning_rate": 1.5532626878538137e-05, "loss": 0.1574365496635437, "step": 12680 }, { "epoch": 0.3342739452688976, "grad_norm": 3.315234899520874, "learning_rate": 1.5525518174196867e-05, "loss": 0.15708038806915284, "step": 12690 }, { "epoch": 0.33453736051339633, "grad_norm": 2.1330811977386475, "learning_rate": 1.551840544814508e-05, "loss": 0.1502138614654541, "step": 12700 }, { "epoch": 0.3348007757578951, "grad_norm": 1.2603166103363037, "learning_rate": 1.551128870555972e-05, "loss": 0.16133569478988646, "step": 12710 }, { "epoch": 0.33506419100239376, "grad_norm": 2.4313244819641113, "learning_rate": 1.550416795162067e-05, "loss": 0.15958077907562257, "step": 12720 }, { "epoch": 0.3353276062468925, "grad_norm": 2.646758794784546, "learning_rate": 1.549704319151071e-05, "loss": 0.16651543378829955, "step": 12730 }, { "epoch": 0.33559102149139125, "grad_norm": 1.463847279548645, "learning_rate": 1.548991443041555e-05, "loss": 0.16047365665435792, "step": 12740 }, { "epoch": 0.33585443673589, "grad_norm": 2.3231146335601807, "learning_rate": 1.5482781673523816e-05, "loss": 0.16358425617218017, "step": 12750 }, { "epoch": 0.33611785198038874, "grad_norm": 2.011672258377075, "learning_rate": 1.5475644926027035e-05, "loss": 0.15335890054702758, "step": 12760 }, { "epoch": 0.3363812672248875, "grad_norm": 3.67651104927063, "learning_rate": 1.5468504193119636e-05, "loss": 0.1769193172454834, "step": 12770 }, { "epoch": 0.3366446824693862, "grad_norm": 3.234307289123535, "learning_rate": 1.5461359479998953e-05, "loss": 0.16260917186737062, "step": 12780 }, { "epoch": 0.33690809771388497, "grad_norm": 2.505861759185791, "learning_rate": 1.5454210791865213e-05, "loss": 0.15975887775421144, "step": 12790 }, { "epoch": 0.33717151295838366, "grad_norm": 1.9686745405197144, "learning_rate": 1.5447058133921552e-05, "loss": 0.17510035037994384, "step": 12800 }, { "epoch": 0.3374349282028824, "grad_norm": 2.8942151069641113, "learning_rate": 1.543990151137397e-05, "loss": 0.1630652666091919, "step": 12810 }, { "epoch": 0.33769834344738114, "grad_norm": 1.6124788522720337, "learning_rate": 1.5432740929431372e-05, "loss": 0.1451029896736145, "step": 12820 }, { "epoch": 0.3379617586918799, "grad_norm": 2.3773393630981445, "learning_rate": 1.5425576393305537e-05, "loss": 0.15746351480484008, "step": 12830 }, { "epoch": 0.33822517393637863, "grad_norm": 1.2473437786102295, "learning_rate": 1.5418407908211125e-05, "loss": 0.1554218053817749, "step": 12840 }, { "epoch": 0.3384885891808774, "grad_norm": 2.3444995880126953, "learning_rate": 1.5411235479365664e-05, "loss": 0.15373481512069703, "step": 12850 }, { "epoch": 0.3387520044253761, "grad_norm": 1.951979637145996, "learning_rate": 1.5404059111989565e-05, "loss": 0.16297742128372192, "step": 12860 }, { "epoch": 0.33901541966987486, "grad_norm": 2.6281206607818604, "learning_rate": 1.5396878811306088e-05, "loss": 0.15754342079162598, "step": 12870 }, { "epoch": 0.3392788349143736, "grad_norm": 1.3189201354980469, "learning_rate": 1.5389694582541376e-05, "loss": 0.1666562557220459, "step": 12880 }, { "epoch": 0.3395422501588723, "grad_norm": 1.2307637929916382, "learning_rate": 1.5382506430924414e-05, "loss": 0.1610428810119629, "step": 12890 }, { "epoch": 0.33980566540337104, "grad_norm": 2.2945539951324463, "learning_rate": 1.5375314361687053e-05, "loss": 0.15570056438446045, "step": 12900 }, { "epoch": 0.3400690806478698, "grad_norm": 3.7731287479400635, "learning_rate": 1.5368118380063984e-05, "loss": 0.16737178564071656, "step": 12910 }, { "epoch": 0.3403324958923685, "grad_norm": 2.903170108795166, "learning_rate": 1.536091849129276e-05, "loss": 0.1629784345626831, "step": 12920 }, { "epoch": 0.34059591113686727, "grad_norm": 0.9237650036811829, "learning_rate": 1.5353714700613768e-05, "loss": 0.1567295551300049, "step": 12930 }, { "epoch": 0.340859326381366, "grad_norm": 4.0871171951293945, "learning_rate": 1.5346507013270237e-05, "loss": 0.15360954999923707, "step": 12940 }, { "epoch": 0.34112274162586476, "grad_norm": 3.449326515197754, "learning_rate": 1.5339295434508238e-05, "loss": 0.15187500715255736, "step": 12950 }, { "epoch": 0.3413861568703635, "grad_norm": 2.3052735328674316, "learning_rate": 1.533207996957666e-05, "loss": 0.15390403270721437, "step": 12960 }, { "epoch": 0.34164957211486224, "grad_norm": 1.8002705574035645, "learning_rate": 1.5324860623727236e-05, "loss": 0.1640734314918518, "step": 12970 }, { "epoch": 0.34191298735936093, "grad_norm": 2.9759011268615723, "learning_rate": 1.5317637402214515e-05, "loss": 0.16106028556823732, "step": 12980 }, { "epoch": 0.3421764026038597, "grad_norm": 1.5406285524368286, "learning_rate": 1.5310410310295876e-05, "loss": 0.14971792697906494, "step": 12990 }, { "epoch": 0.3424398178483584, "grad_norm": 1.6310800313949585, "learning_rate": 1.5303179353231494e-05, "loss": 0.16038281917572023, "step": 13000 }, { "epoch": 0.34270323309285716, "grad_norm": 3.290271043777466, "learning_rate": 1.529594453628439e-05, "loss": 0.15406042337417603, "step": 13010 }, { "epoch": 0.3429666483373559, "grad_norm": 2.5534353256225586, "learning_rate": 1.5288705864720356e-05, "loss": 0.16079561710357665, "step": 13020 }, { "epoch": 0.34323006358185465, "grad_norm": 2.324144124984741, "learning_rate": 1.528146334380803e-05, "loss": 0.17859241962432862, "step": 13030 }, { "epoch": 0.3434934788263534, "grad_norm": 1.9880059957504272, "learning_rate": 1.5274216978818813e-05, "loss": 0.15680484771728515, "step": 13040 }, { "epoch": 0.34375689407085214, "grad_norm": 5.048871994018555, "learning_rate": 1.526696677502693e-05, "loss": 0.15659623146057128, "step": 13050 }, { "epoch": 0.3440203093153508, "grad_norm": 3.652484178543091, "learning_rate": 1.525971273770939e-05, "loss": 0.15550707578659057, "step": 13060 }, { "epoch": 0.34428372455984957, "grad_norm": 3.1007258892059326, "learning_rate": 1.5252454872146003e-05, "loss": 0.15780220031738282, "step": 13070 }, { "epoch": 0.3445471398043483, "grad_norm": 2.6222856044769287, "learning_rate": 1.5245193183619347e-05, "loss": 0.16268000602722169, "step": 13080 }, { "epoch": 0.34481055504884706, "grad_norm": 0.6623892784118652, "learning_rate": 1.5237927677414797e-05, "loss": 0.1644824743270874, "step": 13090 }, { "epoch": 0.3450739702933458, "grad_norm": 1.1987603902816772, "learning_rate": 1.5230658358820493e-05, "loss": 0.15476655960083008, "step": 13100 }, { "epoch": 0.34533738553784454, "grad_norm": 1.6103193759918213, "learning_rate": 1.5223385233127372e-05, "loss": 0.1553072452545166, "step": 13110 }, { "epoch": 0.3456008007823433, "grad_norm": 4.204086780548096, "learning_rate": 1.5216108305629123e-05, "loss": 0.15567433834075928, "step": 13120 }, { "epoch": 0.34586421602684203, "grad_norm": 2.755586624145508, "learning_rate": 1.5208827581622205e-05, "loss": 0.17732486724853516, "step": 13130 }, { "epoch": 0.3461276312713408, "grad_norm": 1.3655773401260376, "learning_rate": 1.5201543066405844e-05, "loss": 0.1507097601890564, "step": 13140 }, { "epoch": 0.34639104651583946, "grad_norm": 1.5888972282409668, "learning_rate": 1.5194254765282025e-05, "loss": 0.16741907596588135, "step": 13150 }, { "epoch": 0.3466544617603382, "grad_norm": 1.6030786037445068, "learning_rate": 1.5186962683555493e-05, "loss": 0.15579020977020264, "step": 13160 }, { "epoch": 0.34691787700483695, "grad_norm": 1.0740143060684204, "learning_rate": 1.517966682653373e-05, "loss": 0.14397659301757812, "step": 13170 }, { "epoch": 0.3471812922493357, "grad_norm": 3.8156445026397705, "learning_rate": 1.5172367199526985e-05, "loss": 0.16053030490875245, "step": 13180 }, { "epoch": 0.34744470749383444, "grad_norm": 3.534778118133545, "learning_rate": 1.5165063807848234e-05, "loss": 0.17064921855926513, "step": 13190 }, { "epoch": 0.3477081227383332, "grad_norm": 3.084017276763916, "learning_rate": 1.5157756656813207e-05, "loss": 0.17293970584869384, "step": 13200 }, { "epoch": 0.3479715379828319, "grad_norm": 2.456469774246216, "learning_rate": 1.5150445751740356e-05, "loss": 0.15269834995269777, "step": 13210 }, { "epoch": 0.34823495322733067, "grad_norm": 1.2390481233596802, "learning_rate": 1.5143131097950882e-05, "loss": 0.15709002017974855, "step": 13220 }, { "epoch": 0.3484983684718294, "grad_norm": 1.9457662105560303, "learning_rate": 1.5135812700768702e-05, "loss": 0.17282135486602784, "step": 13230 }, { "epoch": 0.3487617837163281, "grad_norm": 1.2269865274429321, "learning_rate": 1.512849056552046e-05, "loss": 0.15146687030792236, "step": 13240 }, { "epoch": 0.34902519896082684, "grad_norm": 2.312105655670166, "learning_rate": 1.512116469753552e-05, "loss": 0.1604868531227112, "step": 13250 }, { "epoch": 0.3492886142053256, "grad_norm": 3.5973973274230957, "learning_rate": 1.5113835102145968e-05, "loss": 0.15895841121673585, "step": 13260 }, { "epoch": 0.34955202944982433, "grad_norm": 2.1222429275512695, "learning_rate": 1.51065017846866e-05, "loss": 0.158112096786499, "step": 13270 }, { "epoch": 0.3498154446943231, "grad_norm": 1.798863410949707, "learning_rate": 1.509916475049492e-05, "loss": 0.15400998592376708, "step": 13280 }, { "epoch": 0.3500788599388218, "grad_norm": 3.7973215579986572, "learning_rate": 1.5091824004911137e-05, "loss": 0.18438507318496705, "step": 13290 }, { "epoch": 0.35034227518332056, "grad_norm": 4.121893405914307, "learning_rate": 1.5084479553278164e-05, "loss": 0.15698626041412353, "step": 13300 }, { "epoch": 0.3506056904278193, "grad_norm": 2.0334579944610596, "learning_rate": 1.5077131400941609e-05, "loss": 0.1497563362121582, "step": 13310 }, { "epoch": 0.350869105672318, "grad_norm": 1.0930871963500977, "learning_rate": 1.5069779553249775e-05, "loss": 0.15730983018875122, "step": 13320 }, { "epoch": 0.35113252091681674, "grad_norm": 2.9983816146850586, "learning_rate": 1.5062424015553653e-05, "loss": 0.17402067184448242, "step": 13330 }, { "epoch": 0.3513959361613155, "grad_norm": 2.275449514389038, "learning_rate": 1.505506479320692e-05, "loss": 0.1495842695236206, "step": 13340 }, { "epoch": 0.3516593514058142, "grad_norm": 1.8016490936279297, "learning_rate": 1.5047701891565938e-05, "loss": 0.16386879682540895, "step": 13350 }, { "epoch": 0.35192276665031297, "grad_norm": 1.0246485471725464, "learning_rate": 1.504033531598974e-05, "loss": 0.15369341373443604, "step": 13360 }, { "epoch": 0.3521861818948117, "grad_norm": 2.2637062072753906, "learning_rate": 1.5032965071840045e-05, "loss": 0.16410375833511354, "step": 13370 }, { "epoch": 0.35244959713931046, "grad_norm": 0.8118025660514832, "learning_rate": 1.5025591164481226e-05, "loss": 0.17548689842224122, "step": 13380 }, { "epoch": 0.3527130123838092, "grad_norm": 3.423222303390503, "learning_rate": 1.5018213599280339e-05, "loss": 0.1592349410057068, "step": 13390 }, { "epoch": 0.35297642762830794, "grad_norm": 1.6675349473953247, "learning_rate": 1.5010832381607085e-05, "loss": 0.1574568510055542, "step": 13400 }, { "epoch": 0.35323984287280663, "grad_norm": 4.665497303009033, "learning_rate": 1.500344751683384e-05, "loss": 0.16553572416305543, "step": 13410 }, { "epoch": 0.3535032581173054, "grad_norm": 5.163017749786377, "learning_rate": 1.4996059010335626e-05, "loss": 0.16418508291244507, "step": 13420 }, { "epoch": 0.3537666733618041, "grad_norm": 1.5810531377792358, "learning_rate": 1.4988666867490117e-05, "loss": 0.15166350603103637, "step": 13430 }, { "epoch": 0.35403008860630286, "grad_norm": 4.699390888214111, "learning_rate": 1.4981271093677632e-05, "loss": 0.15531929731369018, "step": 13440 }, { "epoch": 0.3542935038508016, "grad_norm": 3.358999490737915, "learning_rate": 1.4973871694281133e-05, "loss": 0.16696605682373047, "step": 13450 }, { "epoch": 0.35455691909530035, "grad_norm": 2.944150447845459, "learning_rate": 1.4966468674686226e-05, "loss": 0.15467182397842408, "step": 13460 }, { "epoch": 0.3548203343397991, "grad_norm": 4.379073619842529, "learning_rate": 1.4959062040281143e-05, "loss": 0.16225554943084716, "step": 13470 }, { "epoch": 0.35508374958429784, "grad_norm": 1.7008657455444336, "learning_rate": 1.4951651796456758e-05, "loss": 0.16663715839385987, "step": 13480 }, { "epoch": 0.3553471648287966, "grad_norm": 1.0994452238082886, "learning_rate": 1.4944237948606561e-05, "loss": 0.15402997732162477, "step": 13490 }, { "epoch": 0.35561058007329527, "grad_norm": 1.0593913793563843, "learning_rate": 1.4936820502126676e-05, "loss": 0.1523517847061157, "step": 13500 }, { "epoch": 0.355873995317794, "grad_norm": 2.7887258529663086, "learning_rate": 1.4929399462415832e-05, "loss": 0.15883421897888184, "step": 13510 }, { "epoch": 0.35613741056229276, "grad_norm": 1.2019437551498413, "learning_rate": 1.492197483487539e-05, "loss": 0.15234428644180298, "step": 13520 }, { "epoch": 0.3564008258067915, "grad_norm": 2.7698874473571777, "learning_rate": 1.491454662490931e-05, "loss": 0.163788104057312, "step": 13530 }, { "epoch": 0.35666424105129024, "grad_norm": 2.0771799087524414, "learning_rate": 1.4907114837924168e-05, "loss": 0.15734450817108153, "step": 13540 }, { "epoch": 0.356927656295789, "grad_norm": 1.18174147605896, "learning_rate": 1.4899679479329133e-05, "loss": 0.1484987497329712, "step": 13550 }, { "epoch": 0.35719107154028773, "grad_norm": 3.306624174118042, "learning_rate": 1.4892240554535981e-05, "loss": 0.1510411262512207, "step": 13560 }, { "epoch": 0.3574544867847865, "grad_norm": 0.61256343126297, "learning_rate": 1.4884798068959085e-05, "loss": 0.1535266876220703, "step": 13570 }, { "epoch": 0.3577179020292852, "grad_norm": 0.9152763485908508, "learning_rate": 1.4877352028015408e-05, "loss": 0.15852478742599488, "step": 13580 }, { "epoch": 0.3579813172737839, "grad_norm": 1.9129151105880737, "learning_rate": 1.4869902437124499e-05, "loss": 0.16506435871124267, "step": 13590 }, { "epoch": 0.35824473251828265, "grad_norm": 2.0460853576660156, "learning_rate": 1.486244930170849e-05, "loss": 0.16260350942611695, "step": 13600 }, { "epoch": 0.3585081477627814, "grad_norm": 1.3426368236541748, "learning_rate": 1.4854992627192096e-05, "loss": 0.16269999742507935, "step": 13610 }, { "epoch": 0.35877156300728014, "grad_norm": 2.8402631282806396, "learning_rate": 1.4847532419002602e-05, "loss": 0.15133587121963502, "step": 13620 }, { "epoch": 0.3590349782517789, "grad_norm": 0.7262940406799316, "learning_rate": 1.4840068682569877e-05, "loss": 0.1654261350631714, "step": 13630 }, { "epoch": 0.3592983934962776, "grad_norm": 2.5997095108032227, "learning_rate": 1.4832601423326346e-05, "loss": 0.1412306547164917, "step": 13640 }, { "epoch": 0.35956180874077637, "grad_norm": 1.6042578220367432, "learning_rate": 1.4825130646707002e-05, "loss": 0.13963737487792968, "step": 13650 }, { "epoch": 0.3598252239852751, "grad_norm": 2.9954335689544678, "learning_rate": 1.4817656358149398e-05, "loss": 0.16664291620254518, "step": 13660 }, { "epoch": 0.3600886392297738, "grad_norm": 1.3731143474578857, "learning_rate": 1.4810178563093648e-05, "loss": 0.16448842287063598, "step": 13670 }, { "epoch": 0.36035205447427254, "grad_norm": 1.3591465950012207, "learning_rate": 1.4802697266982411e-05, "loss": 0.16669241189956666, "step": 13680 }, { "epoch": 0.3606154697187713, "grad_norm": 2.7480309009552, "learning_rate": 1.4795212475260896e-05, "loss": 0.1571629524230957, "step": 13690 }, { "epoch": 0.36087888496327003, "grad_norm": 1.7366609573364258, "learning_rate": 1.4787724193376858e-05, "loss": 0.1533592939376831, "step": 13700 }, { "epoch": 0.3611423002077688, "grad_norm": 3.360100269317627, "learning_rate": 1.4780232426780592e-05, "loss": 0.15604684352874756, "step": 13710 }, { "epoch": 0.3614057154522675, "grad_norm": 1.3372886180877686, "learning_rate": 1.477273718092493e-05, "loss": 0.16217743158340453, "step": 13720 }, { "epoch": 0.36166913069676626, "grad_norm": 1.6043689250946045, "learning_rate": 1.4765238461265235e-05, "loss": 0.16155805587768554, "step": 13730 }, { "epoch": 0.361932545941265, "grad_norm": 2.5744712352752686, "learning_rate": 1.4757736273259396e-05, "loss": 0.16881512403488158, "step": 13740 }, { "epoch": 0.36219596118576375, "grad_norm": 1.6850076913833618, "learning_rate": 1.4750230622367831e-05, "loss": 0.1528583288192749, "step": 13750 }, { "epoch": 0.36245937643026244, "grad_norm": 4.381775379180908, "learning_rate": 1.474272151405348e-05, "loss": 0.15432889461517335, "step": 13760 }, { "epoch": 0.3627227916747612, "grad_norm": 2.0710999965667725, "learning_rate": 1.473520895378179e-05, "loss": 0.16034939289093017, "step": 13770 }, { "epoch": 0.3629862069192599, "grad_norm": 2.1385741233825684, "learning_rate": 1.472769294702073e-05, "loss": 0.17269303798675537, "step": 13780 }, { "epoch": 0.36324962216375867, "grad_norm": 4.452564239501953, "learning_rate": 1.4720173499240771e-05, "loss": 0.162493896484375, "step": 13790 }, { "epoch": 0.3635130374082574, "grad_norm": 2.890094041824341, "learning_rate": 1.4712650615914899e-05, "loss": 0.1599984884262085, "step": 13800 }, { "epoch": 0.36377645265275615, "grad_norm": 2.7490758895874023, "learning_rate": 1.4705124302518581e-05, "loss": 0.17631933689117432, "step": 13810 }, { "epoch": 0.3640398678972549, "grad_norm": 3.2865192890167236, "learning_rate": 1.4697594564529805e-05, "loss": 0.16575756072998046, "step": 13820 }, { "epoch": 0.36430328314175364, "grad_norm": 1.452187180519104, "learning_rate": 1.4690061407429028e-05, "loss": 0.15172350406646729, "step": 13830 }, { "epoch": 0.3645666983862524, "grad_norm": 1.1077499389648438, "learning_rate": 1.468252483669922e-05, "loss": 0.16281816959381104, "step": 13840 }, { "epoch": 0.3648301136307511, "grad_norm": 1.3438977003097534, "learning_rate": 1.4674984857825808e-05, "loss": 0.17320263385772705, "step": 13850 }, { "epoch": 0.3650935288752498, "grad_norm": 3.23172926902771, "learning_rate": 1.4667441476296723e-05, "loss": 0.1639722228050232, "step": 13860 }, { "epoch": 0.36535694411974856, "grad_norm": 1.0664628744125366, "learning_rate": 1.4659894697602358e-05, "loss": 0.16107330322265626, "step": 13870 }, { "epoch": 0.3656203593642473, "grad_norm": 3.7399046421051025, "learning_rate": 1.465234452723559e-05, "loss": 0.16172244548797607, "step": 13880 }, { "epoch": 0.36588377460874605, "grad_norm": 1.52302086353302, "learning_rate": 1.4644790970691755e-05, "loss": 0.15384399890899658, "step": 13890 }, { "epoch": 0.3661471898532448, "grad_norm": 2.207765579223633, "learning_rate": 1.4637234033468658e-05, "loss": 0.15653094053268432, "step": 13900 }, { "epoch": 0.36641060509774354, "grad_norm": 0.9654771685600281, "learning_rate": 1.4629673721066566e-05, "loss": 0.15635418891906738, "step": 13910 }, { "epoch": 0.3666740203422423, "grad_norm": 2.821613073348999, "learning_rate": 1.4622110038988196e-05, "loss": 0.16008994579315186, "step": 13920 }, { "epoch": 0.36693743558674097, "grad_norm": 0.8471297025680542, "learning_rate": 1.4614542992738727e-05, "loss": 0.16918478012084961, "step": 13930 }, { "epoch": 0.3672008508312397, "grad_norm": 0.8293030261993408, "learning_rate": 1.4606972587825779e-05, "loss": 0.16958497762680053, "step": 13940 }, { "epoch": 0.36746426607573845, "grad_norm": 2.756995439529419, "learning_rate": 1.4599398829759419e-05, "loss": 0.15741785764694213, "step": 13950 }, { "epoch": 0.3677276813202372, "grad_norm": 1.4861520528793335, "learning_rate": 1.4591821724052155e-05, "loss": 0.14604532718658447, "step": 13960 }, { "epoch": 0.36799109656473594, "grad_norm": 1.4677746295928955, "learning_rate": 1.4584241276218935e-05, "loss": 0.17056978940963746, "step": 13970 }, { "epoch": 0.3682545118092347, "grad_norm": 0.8036866784095764, "learning_rate": 1.4576657491777134e-05, "loss": 0.157308030128479, "step": 13980 }, { "epoch": 0.36851792705373343, "grad_norm": 2.0372748374938965, "learning_rate": 1.4569070376246552e-05, "loss": 0.164312744140625, "step": 13990 }, { "epoch": 0.3687813422982322, "grad_norm": 1.2364546060562134, "learning_rate": 1.4561479935149426e-05, "loss": 0.15694385766983032, "step": 14000 }, { "epoch": 0.3690447575427309, "grad_norm": 2.079413652420044, "learning_rate": 1.4553886174010405e-05, "loss": 0.17038297653198242, "step": 14010 }, { "epoch": 0.3693081727872296, "grad_norm": 0.8347895741462708, "learning_rate": 1.4546289098356556e-05, "loss": 0.14436312913894653, "step": 14020 }, { "epoch": 0.36957158803172835, "grad_norm": 1.5082721710205078, "learning_rate": 1.4538688713717357e-05, "loss": 0.1520588994026184, "step": 14030 }, { "epoch": 0.3698350032762271, "grad_norm": 110.55648040771484, "learning_rate": 1.4531085025624694e-05, "loss": 0.8677382469177246, "step": 14040 }, { "epoch": 0.37009841852072584, "grad_norm": 1.5125747919082642, "learning_rate": 1.4523478039612863e-05, "loss": 0.3159600257873535, "step": 14050 }, { "epoch": 0.3703618337652246, "grad_norm": 8.1876859664917, "learning_rate": 1.4515867761218558e-05, "loss": 2.35714054107666, "step": 14060 }, { "epoch": 0.3706252490097233, "grad_norm": 2.4898831844329834, "learning_rate": 1.4508254195980861e-05, "loss": 0.18054525852203368, "step": 14070 }, { "epoch": 0.37088866425422207, "grad_norm": 1.2984626293182373, "learning_rate": 1.450063734944126e-05, "loss": 0.18796344995498657, "step": 14080 }, { "epoch": 0.3711520794987208, "grad_norm": 2.631394624710083, "learning_rate": 1.449301722714362e-05, "loss": 0.16985509395599366, "step": 14090 }, { "epoch": 0.37141549474321955, "grad_norm": 1.4159762859344482, "learning_rate": 1.4485393834634202e-05, "loss": 0.17357406616210938, "step": 14100 }, { "epoch": 0.37167890998771824, "grad_norm": 1.9739222526550293, "learning_rate": 1.447776717746163e-05, "loss": 0.16190049648284913, "step": 14110 }, { "epoch": 0.371942325232217, "grad_norm": 1.3441163301467896, "learning_rate": 1.4470137261176922e-05, "loss": 0.16300761699676514, "step": 14120 }, { "epoch": 0.37220574047671573, "grad_norm": 4.897531986236572, "learning_rate": 1.4462504091333454e-05, "loss": 0.1605395793914795, "step": 14130 }, { "epoch": 0.3724691557212145, "grad_norm": 1.5856220722198486, "learning_rate": 1.4454867673486983e-05, "loss": 0.16154123544692994, "step": 14140 }, { "epoch": 0.3727325709657132, "grad_norm": 2.503584146499634, "learning_rate": 1.4447228013195616e-05, "loss": 0.16073012351989746, "step": 14150 }, { "epoch": 0.37299598621021196, "grad_norm": 1.0680302381515503, "learning_rate": 1.4439585116019835e-05, "loss": 0.1634621024131775, "step": 14160 }, { "epoch": 0.3732594014547107, "grad_norm": 1.505003571510315, "learning_rate": 1.4431938987522463e-05, "loss": 0.16518867015838623, "step": 14170 }, { "epoch": 0.37352281669920945, "grad_norm": 2.2127318382263184, "learning_rate": 1.4424289633268689e-05, "loss": 0.16680452823638917, "step": 14180 }, { "epoch": 0.37378623194370814, "grad_norm": 0.7366983890533447, "learning_rate": 1.4416637058826038e-05, "loss": 0.14971038103103637, "step": 14190 }, { "epoch": 0.3740496471882069, "grad_norm": 2.635831594467163, "learning_rate": 1.4408981269764386e-05, "loss": 0.161404812335968, "step": 14200 }, { "epoch": 0.3743130624327056, "grad_norm": 3.1111111640930176, "learning_rate": 1.4401322271655947e-05, "loss": 0.15270341634750367, "step": 14210 }, { "epoch": 0.37457647767720437, "grad_norm": 1.6260734796524048, "learning_rate": 1.4393660070075267e-05, "loss": 0.1615450620651245, "step": 14220 }, { "epoch": 0.3748398929217031, "grad_norm": 3.3429183959960938, "learning_rate": 1.4385994670599233e-05, "loss": 0.15854017734527587, "step": 14230 }, { "epoch": 0.37510330816620185, "grad_norm": 2.1218626499176025, "learning_rate": 1.4378326078807048e-05, "loss": 0.15605275630950927, "step": 14240 }, { "epoch": 0.3753667234107006, "grad_norm": 1.3226367235183716, "learning_rate": 1.4370654300280246e-05, "loss": 0.16683568954467773, "step": 14250 }, { "epoch": 0.37563013865519934, "grad_norm": 1.6314787864685059, "learning_rate": 1.4362979340602677e-05, "loss": 0.1635197162628174, "step": 14260 }, { "epoch": 0.3758935538996981, "grad_norm": 0.898331880569458, "learning_rate": 1.4355301205360513e-05, "loss": 0.161777126789093, "step": 14270 }, { "epoch": 0.3761569691441968, "grad_norm": 1.1046069860458374, "learning_rate": 1.4347619900142223e-05, "loss": 0.17729322910308837, "step": 14280 }, { "epoch": 0.3764203843886955, "grad_norm": 2.354898691177368, "learning_rate": 1.43399354305386e-05, "loss": 0.14472016096115112, "step": 14290 }, { "epoch": 0.37668379963319426, "grad_norm": 4.608315944671631, "learning_rate": 1.4332247802142729e-05, "loss": 0.17475535869598388, "step": 14300 }, { "epoch": 0.376947214877693, "grad_norm": 1.088617205619812, "learning_rate": 1.4324557020550001e-05, "loss": 0.15662403106689454, "step": 14310 }, { "epoch": 0.37721063012219175, "grad_norm": 0.8234291672706604, "learning_rate": 1.4316863091358093e-05, "loss": 0.15369280576705932, "step": 14320 }, { "epoch": 0.3774740453666905, "grad_norm": 1.58505380153656, "learning_rate": 1.4309166020166984e-05, "loss": 0.14472601413726807, "step": 14330 }, { "epoch": 0.37773746061118924, "grad_norm": 0.905323326587677, "learning_rate": 1.430146581257893e-05, "loss": 0.15612034797668456, "step": 14340 }, { "epoch": 0.378000875855688, "grad_norm": 2.629537582397461, "learning_rate": 1.4293762474198478e-05, "loss": 0.1532529354095459, "step": 14350 }, { "epoch": 0.3782642911001867, "grad_norm": 1.8158786296844482, "learning_rate": 1.428605601063245e-05, "loss": 0.15262103080749512, "step": 14360 }, { "epoch": 0.3785277063446854, "grad_norm": 2.0441555976867676, "learning_rate": 1.427834642748994e-05, "loss": 0.154779052734375, "step": 14370 }, { "epoch": 0.37879112158918415, "grad_norm": 1.5941073894500732, "learning_rate": 1.4270633730382315e-05, "loss": 0.16321158409118652, "step": 14380 }, { "epoch": 0.3790545368336829, "grad_norm": 2.4559993743896484, "learning_rate": 1.4262917924923212e-05, "loss": 0.15053443908691405, "step": 14390 }, { "epoch": 0.37931795207818164, "grad_norm": 2.478412389755249, "learning_rate": 1.4255199016728527e-05, "loss": 0.1608412742614746, "step": 14400 }, { "epoch": 0.3795813673226804, "grad_norm": 1.159986138343811, "learning_rate": 1.4247477011416409e-05, "loss": 0.1534440040588379, "step": 14410 }, { "epoch": 0.37984478256717913, "grad_norm": 1.4111642837524414, "learning_rate": 1.4239751914607274e-05, "loss": 0.16220602989196778, "step": 14420 }, { "epoch": 0.3801081978116779, "grad_norm": 2.474367618560791, "learning_rate": 1.4232023731923772e-05, "loss": 0.1425974726676941, "step": 14430 }, { "epoch": 0.3803716130561766, "grad_norm": 1.536802053451538, "learning_rate": 1.422429246899082e-05, "loss": 0.16660391092300414, "step": 14440 }, { "epoch": 0.3806350283006753, "grad_norm": 3.1972148418426514, "learning_rate": 1.4216558131435551e-05, "loss": 0.1604740023612976, "step": 14450 }, { "epoch": 0.38089844354517405, "grad_norm": 3.2911996841430664, "learning_rate": 1.420882072488736e-05, "loss": 0.15100218057632447, "step": 14460 }, { "epoch": 0.3811618587896728, "grad_norm": 2.438091993331909, "learning_rate": 1.420108025497786e-05, "loss": 0.16301407814025878, "step": 14470 }, { "epoch": 0.38142527403417154, "grad_norm": 1.3162639141082764, "learning_rate": 1.41933367273409e-05, "loss": 0.1548585057258606, "step": 14480 }, { "epoch": 0.3816886892786703, "grad_norm": 1.367508888244629, "learning_rate": 1.4185590147612555e-05, "loss": 0.15591106414794922, "step": 14490 }, { "epoch": 0.381952104523169, "grad_norm": 3.4567711353302, "learning_rate": 1.4177840521431117e-05, "loss": 0.14802837371826172, "step": 14500 }, { "epoch": 0.38221551976766777, "grad_norm": 5.098599433898926, "learning_rate": 1.4170087854437102e-05, "loss": 0.14606521129608155, "step": 14510 }, { "epoch": 0.3824789350121665, "grad_norm": 2.358691692352295, "learning_rate": 1.4162332152273233e-05, "loss": 0.16264336109161376, "step": 14520 }, { "epoch": 0.38274235025666525, "grad_norm": 2.65032696723938, "learning_rate": 1.4154573420584448e-05, "loss": 0.15282394886016845, "step": 14530 }, { "epoch": 0.38300576550116394, "grad_norm": 0.8389140367507935, "learning_rate": 1.414681166501788e-05, "loss": 0.15431656837463378, "step": 14540 }, { "epoch": 0.3832691807456627, "grad_norm": 1.0716118812561035, "learning_rate": 1.4139046891222877e-05, "loss": 0.16256898641586304, "step": 14550 }, { "epoch": 0.38353259599016143, "grad_norm": 1.4902602434158325, "learning_rate": 1.4131279104850968e-05, "loss": 0.17851227521896362, "step": 14560 }, { "epoch": 0.3837960112346602, "grad_norm": 0.6923970580101013, "learning_rate": 1.4123508311555894e-05, "loss": 0.1514551281929016, "step": 14570 }, { "epoch": 0.3840594264791589, "grad_norm": 1.4878734350204468, "learning_rate": 1.4115734516993562e-05, "loss": 0.16289887428283692, "step": 14580 }, { "epoch": 0.38432284172365766, "grad_norm": 4.819644927978516, "learning_rate": 1.4107957726822084e-05, "loss": 0.15796537399291993, "step": 14590 }, { "epoch": 0.3845862569681564, "grad_norm": 3.1440346240997314, "learning_rate": 1.4100177946701738e-05, "loss": 0.16468267440795897, "step": 14600 }, { "epoch": 0.38484967221265515, "grad_norm": 1.0325764417648315, "learning_rate": 1.4092395182294989e-05, "loss": 0.1577362298965454, "step": 14610 }, { "epoch": 0.3851130874571539, "grad_norm": 1.8198367357254028, "learning_rate": 1.4084609439266464e-05, "loss": 0.1614452362060547, "step": 14620 }, { "epoch": 0.3853765027016526, "grad_norm": 1.0961147546768188, "learning_rate": 1.4076820723282962e-05, "loss": 0.14957247972488402, "step": 14630 }, { "epoch": 0.3856399179461513, "grad_norm": 1.3297518491744995, "learning_rate": 1.4069029040013448e-05, "loss": 0.16798272132873535, "step": 14640 }, { "epoch": 0.38590333319065007, "grad_norm": 2.059314727783203, "learning_rate": 1.4061234395129049e-05, "loss": 0.15547800064086914, "step": 14650 }, { "epoch": 0.3861667484351488, "grad_norm": 1.6946725845336914, "learning_rate": 1.405343679430304e-05, "loss": 0.1593209147453308, "step": 14660 }, { "epoch": 0.38643016367964755, "grad_norm": 1.9300618171691895, "learning_rate": 1.4045636243210852e-05, "loss": 0.1602797508239746, "step": 14670 }, { "epoch": 0.3866935789241463, "grad_norm": 1.4237234592437744, "learning_rate": 1.4037832747530063e-05, "loss": 0.1451493501663208, "step": 14680 }, { "epoch": 0.38695699416864504, "grad_norm": 1.6536520719528198, "learning_rate": 1.403002631294039e-05, "loss": 0.14889888763427733, "step": 14690 }, { "epoch": 0.3872204094131438, "grad_norm": 1.760358214378357, "learning_rate": 1.40222169451237e-05, "loss": 0.14497737884521483, "step": 14700 }, { "epoch": 0.3874838246576425, "grad_norm": 1.6056239604949951, "learning_rate": 1.4014404649763983e-05, "loss": 0.1589387536048889, "step": 14710 }, { "epoch": 0.3877472399021412, "grad_norm": 2.7547967433929443, "learning_rate": 1.4006589432547367e-05, "loss": 0.15198566913604736, "step": 14720 }, { "epoch": 0.38801065514663996, "grad_norm": 2.2136285305023193, "learning_rate": 1.3998771299162099e-05, "loss": 0.155065655708313, "step": 14730 }, { "epoch": 0.3882740703911387, "grad_norm": 2.704988479614258, "learning_rate": 1.3990950255298565e-05, "loss": 0.15528141260147094, "step": 14740 }, { "epoch": 0.38853748563563745, "grad_norm": 1.4304112195968628, "learning_rate": 1.3983126306649243e-05, "loss": 0.17462058067321778, "step": 14750 }, { "epoch": 0.3888009008801362, "grad_norm": 2.874154567718506, "learning_rate": 1.3975299458908753e-05, "loss": 0.15062211751937865, "step": 14760 }, { "epoch": 0.38906431612463493, "grad_norm": 2.2800326347351074, "learning_rate": 1.3967469717773807e-05, "loss": 0.16044508218765258, "step": 14770 }, { "epoch": 0.3893277313691337, "grad_norm": 2.527495861053467, "learning_rate": 1.395963708894323e-05, "loss": 0.15009183883666993, "step": 14780 }, { "epoch": 0.3895911466136324, "grad_norm": 3.599330425262451, "learning_rate": 1.3951801578117946e-05, "loss": 0.1624462127685547, "step": 14790 }, { "epoch": 0.3898545618581311, "grad_norm": 1.012844204902649, "learning_rate": 1.3943963191000976e-05, "loss": 0.15119903087615966, "step": 14800 }, { "epoch": 0.39011797710262985, "grad_norm": 3.0748188495635986, "learning_rate": 1.393612193329744e-05, "loss": 0.1639049768447876, "step": 14810 }, { "epoch": 0.3903813923471286, "grad_norm": 4.500916481018066, "learning_rate": 1.3928277810714542e-05, "loss": 0.16067972183227539, "step": 14820 }, { "epoch": 0.39064480759162734, "grad_norm": 2.3998796939849854, "learning_rate": 1.3920430828961576e-05, "loss": 0.14732036590576172, "step": 14830 }, { "epoch": 0.3909082228361261, "grad_norm": 1.7657290697097778, "learning_rate": 1.3912580993749907e-05, "loss": 0.15312588214874268, "step": 14840 }, { "epoch": 0.39117163808062483, "grad_norm": 1.9918700456619263, "learning_rate": 1.3904728310792988e-05, "loss": 0.15555081367492676, "step": 14850 }, { "epoch": 0.39143505332512357, "grad_norm": 2.9762628078460693, "learning_rate": 1.3896872785806342e-05, "loss": 0.152563738822937, "step": 14860 }, { "epoch": 0.3916984685696223, "grad_norm": 3.6066579818725586, "learning_rate": 1.3889014424507557e-05, "loss": 0.14564290046691894, "step": 14870 }, { "epoch": 0.39196188381412106, "grad_norm": 1.7081985473632812, "learning_rate": 1.3881153232616284e-05, "loss": 0.15138479471206664, "step": 14880 }, { "epoch": 0.39222529905861975, "grad_norm": 1.5145131349563599, "learning_rate": 1.3873289215854247e-05, "loss": 0.14838545322418212, "step": 14890 }, { "epoch": 0.3924887143031185, "grad_norm": 1.738135814666748, "learning_rate": 1.3865422379945205e-05, "loss": 0.14054570198059083, "step": 14900 }, { "epoch": 0.39275212954761723, "grad_norm": 1.2947759628295898, "learning_rate": 1.3857552730614989e-05, "loss": 0.1709146499633789, "step": 14910 }, { "epoch": 0.393015544792116, "grad_norm": 1.4075288772583008, "learning_rate": 1.3849680273591468e-05, "loss": 0.15703871250152587, "step": 14920 }, { "epoch": 0.3932789600366147, "grad_norm": 1.7360895872116089, "learning_rate": 1.3841805014604553e-05, "loss": 0.1606353759765625, "step": 14930 }, { "epoch": 0.39354237528111347, "grad_norm": 3.379314422607422, "learning_rate": 1.3833926959386199e-05, "loss": 0.15726999044418336, "step": 14940 }, { "epoch": 0.3938057905256122, "grad_norm": 3.6775314807891846, "learning_rate": 1.3826046113670395e-05, "loss": 0.15679794549942017, "step": 14950 }, { "epoch": 0.39406920577011095, "grad_norm": 1.1248250007629395, "learning_rate": 1.3818162483193159e-05, "loss": 0.15443719625473024, "step": 14960 }, { "epoch": 0.39433262101460964, "grad_norm": 1.5600124597549438, "learning_rate": 1.3810276073692537e-05, "loss": 0.14091256856918336, "step": 14970 }, { "epoch": 0.3945960362591084, "grad_norm": 0.9576500058174133, "learning_rate": 1.3802386890908602e-05, "loss": 0.15133707523345946, "step": 14980 }, { "epoch": 0.39485945150360713, "grad_norm": 2.258395195007324, "learning_rate": 1.3794494940583437e-05, "loss": 0.1545964241027832, "step": 14990 }, { "epoch": 0.39512286674810587, "grad_norm": 1.9081882238388062, "learning_rate": 1.3786600228461147e-05, "loss": 0.15898699760437013, "step": 15000 }, { "epoch": 0.3953862819926046, "grad_norm": 3.8406331539154053, "learning_rate": 1.377870276028784e-05, "loss": 0.15432358980178834, "step": 15010 }, { "epoch": 0.39564969723710336, "grad_norm": 2.1398234367370605, "learning_rate": 1.377080254181164e-05, "loss": 0.1635402798652649, "step": 15020 }, { "epoch": 0.3959131124816021, "grad_norm": 1.2770426273345947, "learning_rate": 1.3762899578782662e-05, "loss": 0.1613847255706787, "step": 15030 }, { "epoch": 0.39617652772610085, "grad_norm": 4.164129257202148, "learning_rate": 1.375499387695303e-05, "loss": 0.156082820892334, "step": 15040 }, { "epoch": 0.3964399429705996, "grad_norm": 1.3586474657058716, "learning_rate": 1.3747085442076844e-05, "loss": 0.1503867745399475, "step": 15050 }, { "epoch": 0.3967033582150983, "grad_norm": 1.3941792249679565, "learning_rate": 1.3739174279910214e-05, "loss": 0.15002474784851075, "step": 15060 }, { "epoch": 0.396966773459597, "grad_norm": 1.4000190496444702, "learning_rate": 1.3731260396211219e-05, "loss": 0.14491009712219238, "step": 15070 }, { "epoch": 0.39723018870409577, "grad_norm": 4.2549285888671875, "learning_rate": 1.372334379673993e-05, "loss": 0.15230891704559327, "step": 15080 }, { "epoch": 0.3974936039485945, "grad_norm": 1.0720367431640625, "learning_rate": 1.3715424487258386e-05, "loss": 0.14802491664886475, "step": 15090 }, { "epoch": 0.39775701919309325, "grad_norm": 1.0913972854614258, "learning_rate": 1.37075024735306e-05, "loss": 0.16114606857299804, "step": 15100 }, { "epoch": 0.398020434437592, "grad_norm": 2.7633302211761475, "learning_rate": 1.3699577761322563e-05, "loss": 0.15363030433654784, "step": 15110 }, { "epoch": 0.39828384968209074, "grad_norm": 1.625823736190796, "learning_rate": 1.3691650356402212e-05, "loss": 0.15904229879379272, "step": 15120 }, { "epoch": 0.3985472649265895, "grad_norm": 1.7051068544387817, "learning_rate": 1.3683720264539469e-05, "loss": 0.13933794498443602, "step": 15130 }, { "epoch": 0.3988106801710882, "grad_norm": 1.8965749740600586, "learning_rate": 1.3675787491506184e-05, "loss": 0.1487110137939453, "step": 15140 }, { "epoch": 0.3990740954155869, "grad_norm": 1.192322015762329, "learning_rate": 1.3667852043076173e-05, "loss": 0.15799120664596558, "step": 15150 }, { "epoch": 0.39933751066008566, "grad_norm": 3.494121551513672, "learning_rate": 1.3659913925025203e-05, "loss": 0.16292086839675904, "step": 15160 }, { "epoch": 0.3996009259045844, "grad_norm": 3.174518346786499, "learning_rate": 1.3651973143130979e-05, "loss": 0.15039658546447754, "step": 15170 }, { "epoch": 0.39986434114908315, "grad_norm": 1.7108027935028076, "learning_rate": 1.3644029703173139e-05, "loss": 0.14339520931243896, "step": 15180 }, { "epoch": 0.4001277563935819, "grad_norm": 2.013195753097534, "learning_rate": 1.3636083610933262e-05, "loss": 0.15625673532485962, "step": 15190 }, { "epoch": 0.40039117163808063, "grad_norm": 2.329141139984131, "learning_rate": 1.3628134872194861e-05, "loss": 0.1540451765060425, "step": 15200 }, { "epoch": 0.4006545868825794, "grad_norm": 1.9733343124389648, "learning_rate": 1.362018349274337e-05, "loss": 0.14964830875396729, "step": 15210 }, { "epoch": 0.4009180021270781, "grad_norm": 2.4485409259796143, "learning_rate": 1.3612229478366144e-05, "loss": 0.14491132497787476, "step": 15220 }, { "epoch": 0.4011814173715768, "grad_norm": 2.8672173023223877, "learning_rate": 1.360427283485246e-05, "loss": 0.1539653420448303, "step": 15230 }, { "epoch": 0.40144483261607555, "grad_norm": 1.2605730295181274, "learning_rate": 1.3596313567993504e-05, "loss": 0.15691407918930053, "step": 15240 }, { "epoch": 0.4017082478605743, "grad_norm": 1.1213932037353516, "learning_rate": 1.3588351683582374e-05, "loss": 0.14566655158996583, "step": 15250 }, { "epoch": 0.40197166310507304, "grad_norm": 2.2097480297088623, "learning_rate": 1.3580387187414077e-05, "loss": 0.15260565280914307, "step": 15260 }, { "epoch": 0.4022350783495718, "grad_norm": 2.1519720554351807, "learning_rate": 1.3572420085285512e-05, "loss": 0.1462348222732544, "step": 15270 }, { "epoch": 0.4024984935940705, "grad_norm": 2.8429415225982666, "learning_rate": 1.3564450382995482e-05, "loss": 0.15579757690429688, "step": 15280 }, { "epoch": 0.40276190883856927, "grad_norm": 1.5860052108764648, "learning_rate": 1.3556478086344678e-05, "loss": 0.1445314645767212, "step": 15290 }, { "epoch": 0.403025324083068, "grad_norm": 3.490997314453125, "learning_rate": 1.3548503201135687e-05, "loss": 0.1337110161781311, "step": 15300 }, { "epoch": 0.40328873932756676, "grad_norm": 2.182732582092285, "learning_rate": 1.3540525733172966e-05, "loss": 0.16021265983581542, "step": 15310 }, { "epoch": 0.40355215457206545, "grad_norm": 1.8379122018814087, "learning_rate": 1.353254568826287e-05, "loss": 0.15252674818038942, "step": 15320 }, { "epoch": 0.4038155698165642, "grad_norm": 1.076662540435791, "learning_rate": 1.3524563072213612e-05, "loss": 0.15463267564773558, "step": 15330 }, { "epoch": 0.40407898506106293, "grad_norm": 1.7356864213943481, "learning_rate": 1.3516577890835292e-05, "loss": 0.15280613899230958, "step": 15340 }, { "epoch": 0.4043424003055617, "grad_norm": 1.8758816719055176, "learning_rate": 1.350859014993986e-05, "loss": 0.15738201141357422, "step": 15350 }, { "epoch": 0.4046058155500604, "grad_norm": 1.441892147064209, "learning_rate": 1.3500599855341146e-05, "loss": 0.14743616580963134, "step": 15360 }, { "epoch": 0.40486923079455917, "grad_norm": 2.485245704650879, "learning_rate": 1.3492607012854824e-05, "loss": 0.1594054102897644, "step": 15370 }, { "epoch": 0.4051326460390579, "grad_norm": 1.307332158088684, "learning_rate": 1.3484611628298438e-05, "loss": 0.13513885736465453, "step": 15380 }, { "epoch": 0.40539606128355665, "grad_norm": 4.127222061157227, "learning_rate": 1.3476613707491372e-05, "loss": 0.1472334861755371, "step": 15390 }, { "epoch": 0.4056594765280554, "grad_norm": 1.2048641443252563, "learning_rate": 1.3468613256254852e-05, "loss": 0.14216147661209105, "step": 15400 }, { "epoch": 0.4059228917725541, "grad_norm": 1.0668295621871948, "learning_rate": 1.3460610280411957e-05, "loss": 0.15732243061065673, "step": 15410 }, { "epoch": 0.40618630701705283, "grad_norm": 1.7076287269592285, "learning_rate": 1.3452604785787595e-05, "loss": 0.15003594160079955, "step": 15420 }, { "epoch": 0.40644972226155157, "grad_norm": 2.4249813556671143, "learning_rate": 1.3444596778208516e-05, "loss": 0.15608222484588624, "step": 15430 }, { "epoch": 0.4067131375060503, "grad_norm": 2.1549947261810303, "learning_rate": 1.3436586263503286e-05, "loss": 0.15068566799163818, "step": 15440 }, { "epoch": 0.40697655275054906, "grad_norm": 1.9242119789123535, "learning_rate": 1.3428573247502307e-05, "loss": 0.1667972445487976, "step": 15450 }, { "epoch": 0.4072399679950478, "grad_norm": 4.145184516906738, "learning_rate": 1.3420557736037796e-05, "loss": 0.15690364837646484, "step": 15460 }, { "epoch": 0.40750338323954655, "grad_norm": 1.244687557220459, "learning_rate": 1.3412539734943792e-05, "loss": 0.13695316314697265, "step": 15470 }, { "epoch": 0.4077667984840453, "grad_norm": 1.08542001247406, "learning_rate": 1.3404519250056139e-05, "loss": 0.13999866247177123, "step": 15480 }, { "epoch": 0.408030213728544, "grad_norm": 2.4434072971343994, "learning_rate": 1.3396496287212491e-05, "loss": 0.14371805191040038, "step": 15490 }, { "epoch": 0.4082936289730427, "grad_norm": 3.3540825843811035, "learning_rate": 1.3388470852252307e-05, "loss": 0.15672013759613038, "step": 15500 }, { "epoch": 0.40855704421754147, "grad_norm": 1.4336308240890503, "learning_rate": 1.3380442951016846e-05, "loss": 0.14043289422988892, "step": 15510 }, { "epoch": 0.4088204594620402, "grad_norm": 2.8913371562957764, "learning_rate": 1.3372412589349164e-05, "loss": 0.1554826021194458, "step": 15520 }, { "epoch": 0.40908387470653895, "grad_norm": 2.0113980770111084, "learning_rate": 1.3364379773094099e-05, "loss": 0.15251659154891967, "step": 15530 }, { "epoch": 0.4093472899510377, "grad_norm": 1.5593055486679077, "learning_rate": 1.335634450809828e-05, "loss": 0.15996122360229492, "step": 15540 }, { "epoch": 0.40961070519553644, "grad_norm": 0.9899659752845764, "learning_rate": 1.3348306800210127e-05, "loss": 0.15573834180831908, "step": 15550 }, { "epoch": 0.4098741204400352, "grad_norm": 1.975188136100769, "learning_rate": 1.3340266655279828e-05, "loss": 0.15128315687179567, "step": 15560 }, { "epoch": 0.4101375356845339, "grad_norm": 1.4179285764694214, "learning_rate": 1.3332224079159342e-05, "loss": 0.15050524473190308, "step": 15570 }, { "epoch": 0.4104009509290326, "grad_norm": 1.2061607837677002, "learning_rate": 1.3324179077702408e-05, "loss": 0.15532639026641845, "step": 15580 }, { "epoch": 0.41066436617353136, "grad_norm": 4.0466718673706055, "learning_rate": 1.3316131656764523e-05, "loss": 0.1364664316177368, "step": 15590 }, { "epoch": 0.4109277814180301, "grad_norm": 2.1592938899993896, "learning_rate": 1.3308081822202953e-05, "loss": 0.15600072145462035, "step": 15600 }, { "epoch": 0.41119119666252885, "grad_norm": 1.3104662895202637, "learning_rate": 1.3300029579876705e-05, "loss": 0.16084548234939575, "step": 15610 }, { "epoch": 0.4114546119070276, "grad_norm": 1.0310088396072388, "learning_rate": 1.3291974935646554e-05, "loss": 0.14898483753204345, "step": 15620 }, { "epoch": 0.41171802715152633, "grad_norm": 1.0145978927612305, "learning_rate": 1.3283917895375018e-05, "loss": 0.14728541374206544, "step": 15630 }, { "epoch": 0.4119814423960251, "grad_norm": 1.5618764162063599, "learning_rate": 1.327585846492636e-05, "loss": 0.14271261692047119, "step": 15640 }, { "epoch": 0.4122448576405238, "grad_norm": 3.1406073570251465, "learning_rate": 1.3267796650166576e-05, "loss": 0.1601223111152649, "step": 15650 }, { "epoch": 0.41250827288502256, "grad_norm": 0.919978678226471, "learning_rate": 1.3259732456963406e-05, "loss": 0.15550501346588136, "step": 15660 }, { "epoch": 0.41277168812952125, "grad_norm": 2.652468204498291, "learning_rate": 1.3251665891186313e-05, "loss": 0.14719717502593993, "step": 15670 }, { "epoch": 0.41303510337402, "grad_norm": 1.5037918090820312, "learning_rate": 1.32435969587065e-05, "loss": 0.1437836170196533, "step": 15680 }, { "epoch": 0.41329851861851874, "grad_norm": 0.9723587036132812, "learning_rate": 1.323552566539688e-05, "loss": 0.15867984294891357, "step": 15690 }, { "epoch": 0.4135619338630175, "grad_norm": 3.276059150695801, "learning_rate": 1.3227452017132085e-05, "loss": 0.15652737617492676, "step": 15700 }, { "epoch": 0.4138253491075162, "grad_norm": 2.938443183898926, "learning_rate": 1.321937601978847e-05, "loss": 0.1497204542160034, "step": 15710 }, { "epoch": 0.41408876435201497, "grad_norm": 1.6445281505584717, "learning_rate": 1.3211297679244093e-05, "loss": 0.1558853030204773, "step": 15720 }, { "epoch": 0.4143521795965137, "grad_norm": 2.233144760131836, "learning_rate": 1.3203217001378716e-05, "loss": 0.16282907724380494, "step": 15730 }, { "epoch": 0.41461559484101246, "grad_norm": 2.083892583847046, "learning_rate": 1.3195133992073808e-05, "loss": 0.1531532049179077, "step": 15740 }, { "epoch": 0.41487901008551115, "grad_norm": 2.6178719997406006, "learning_rate": 1.318704865721253e-05, "loss": 0.1518696665763855, "step": 15750 }, { "epoch": 0.4151424253300099, "grad_norm": 2.757289409637451, "learning_rate": 1.3178961002679738e-05, "loss": 0.15577858686447144, "step": 15760 }, { "epoch": 0.41540584057450863, "grad_norm": 1.7894455194473267, "learning_rate": 1.317087103436198e-05, "loss": 0.16326783895492553, "step": 15770 }, { "epoch": 0.4156692558190074, "grad_norm": 1.3461533784866333, "learning_rate": 1.3162778758147477e-05, "loss": 0.1505049228668213, "step": 15780 }, { "epoch": 0.4159326710635061, "grad_norm": 0.9790360927581787, "learning_rate": 1.315468417992614e-05, "loss": 0.15926440954208373, "step": 15790 }, { "epoch": 0.41619608630800486, "grad_norm": 1.2073346376419067, "learning_rate": 1.3146587305589552e-05, "loss": 0.1534292221069336, "step": 15800 }, { "epoch": 0.4164595015525036, "grad_norm": 1.2139570713043213, "learning_rate": 1.3138488141030967e-05, "loss": 0.1443413496017456, "step": 15810 }, { "epoch": 0.41672291679700235, "grad_norm": 3.675682544708252, "learning_rate": 1.313038669214531e-05, "loss": 0.15875647068023682, "step": 15820 }, { "epoch": 0.4169863320415011, "grad_norm": 1.0430089235305786, "learning_rate": 1.312228296482916e-05, "loss": 0.1551366329193115, "step": 15830 }, { "epoch": 0.4172497472859998, "grad_norm": 0.7181083559989929, "learning_rate": 1.311417696498076e-05, "loss": 0.15516698360443115, "step": 15840 }, { "epoch": 0.4175131625304985, "grad_norm": 1.1264762878417969, "learning_rate": 1.310606869850001e-05, "loss": 0.13597127199172973, "step": 15850 }, { "epoch": 0.41777657777499727, "grad_norm": 1.1065309047698975, "learning_rate": 1.3097958171288453e-05, "loss": 0.15049299001693725, "step": 15860 }, { "epoch": 0.418039993019496, "grad_norm": 4.83582067489624, "learning_rate": 1.3089845389249282e-05, "loss": 0.15344113111495972, "step": 15870 }, { "epoch": 0.41830340826399476, "grad_norm": 0.9538988471031189, "learning_rate": 1.308173035828733e-05, "loss": 0.1533092737197876, "step": 15880 }, { "epoch": 0.4185668235084935, "grad_norm": 1.6554758548736572, "learning_rate": 1.3073613084309065e-05, "loss": 0.1541075348854065, "step": 15890 }, { "epoch": 0.41883023875299225, "grad_norm": 3.2279746532440186, "learning_rate": 1.3065493573222594e-05, "loss": 0.15966955423355103, "step": 15900 }, { "epoch": 0.419093653997491, "grad_norm": 0.9207119345664978, "learning_rate": 1.305737183093764e-05, "loss": 0.14770147800445557, "step": 15910 }, { "epoch": 0.41935706924198973, "grad_norm": 1.5001112222671509, "learning_rate": 1.3049247863365562e-05, "loss": 0.14938533306121826, "step": 15920 }, { "epoch": 0.4196204844864884, "grad_norm": 0.6960366368293762, "learning_rate": 1.3041121676419333e-05, "loss": 0.1492404818534851, "step": 15930 }, { "epoch": 0.41988389973098716, "grad_norm": 1.526190996170044, "learning_rate": 1.3032993276013546e-05, "loss": 0.15196497440338136, "step": 15940 }, { "epoch": 0.4201473149754859, "grad_norm": 1.2631585597991943, "learning_rate": 1.3024862668064393e-05, "loss": 0.1419268012046814, "step": 15950 }, { "epoch": 0.42041073021998465, "grad_norm": 0.9071977138519287, "learning_rate": 1.3016729858489689e-05, "loss": 0.167085599899292, "step": 15960 }, { "epoch": 0.4206741454644834, "grad_norm": 1.4934310913085938, "learning_rate": 1.300859485320884e-05, "loss": 0.1601471185684204, "step": 15970 }, { "epoch": 0.42093756070898214, "grad_norm": 2.7942543029785156, "learning_rate": 1.3000457658142858e-05, "loss": 0.15877799987792968, "step": 15980 }, { "epoch": 0.4212009759534809, "grad_norm": 1.9847742319107056, "learning_rate": 1.299231827921434e-05, "loss": 0.15381815433502197, "step": 15990 }, { "epoch": 0.4214643911979796, "grad_norm": 1.1363263130187988, "learning_rate": 1.2984176722347478e-05, "loss": 0.14648709297180176, "step": 16000 }, { "epoch": 0.4217278064424783, "grad_norm": 1.0703479051589966, "learning_rate": 1.2976032993468049e-05, "loss": 0.14846906661987305, "step": 16010 }, { "epoch": 0.42199122168697706, "grad_norm": 3.4266786575317383, "learning_rate": 1.296788709850341e-05, "loss": 0.13925650119781494, "step": 16020 }, { "epoch": 0.4222546369314758, "grad_norm": 1.395427942276001, "learning_rate": 1.29597390433825e-05, "loss": 0.13807375431060792, "step": 16030 }, { "epoch": 0.42251805217597455, "grad_norm": 2.329733371734619, "learning_rate": 1.2951588834035813e-05, "loss": 0.14707591533660888, "step": 16040 }, { "epoch": 0.4227814674204733, "grad_norm": 1.3864177465438843, "learning_rate": 1.2943436476395435e-05, "loss": 0.1627244234085083, "step": 16050 }, { "epoch": 0.42304488266497203, "grad_norm": 0.8548792600631714, "learning_rate": 1.2935281976394998e-05, "loss": 0.16215136051177978, "step": 16060 }, { "epoch": 0.4233082979094708, "grad_norm": 2.1092963218688965, "learning_rate": 1.2927125339969701e-05, "loss": 0.1483527421951294, "step": 16070 }, { "epoch": 0.4235717131539695, "grad_norm": 1.5537760257720947, "learning_rate": 1.2918966573056291e-05, "loss": 0.1405902624130249, "step": 16080 }, { "epoch": 0.42383512839846826, "grad_norm": 2.060544013977051, "learning_rate": 1.2910805681593078e-05, "loss": 0.1580068826675415, "step": 16090 }, { "epoch": 0.42409854364296695, "grad_norm": 0.7225781083106995, "learning_rate": 1.2902642671519903e-05, "loss": 0.14787281751632692, "step": 16100 }, { "epoch": 0.4243619588874657, "grad_norm": 0.9733191132545471, "learning_rate": 1.2894477548778164e-05, "loss": 0.1589444637298584, "step": 16110 }, { "epoch": 0.42462537413196444, "grad_norm": 1.730712652206421, "learning_rate": 1.2886310319310787e-05, "loss": 0.15370393991470338, "step": 16120 }, { "epoch": 0.4248887893764632, "grad_norm": 2.5866243839263916, "learning_rate": 1.2878140989062233e-05, "loss": 0.15862612724304198, "step": 16130 }, { "epoch": 0.4251522046209619, "grad_norm": 1.183091163635254, "learning_rate": 1.2869969563978493e-05, "loss": 0.15600756406784058, "step": 16140 }, { "epoch": 0.42541561986546067, "grad_norm": 1.8540464639663696, "learning_rate": 1.2861796050007082e-05, "loss": 0.14891045093536376, "step": 16150 }, { "epoch": 0.4256790351099594, "grad_norm": 2.6379146575927734, "learning_rate": 1.2853620453097039e-05, "loss": 0.16843682527542114, "step": 16160 }, { "epoch": 0.42594245035445816, "grad_norm": 1.0267925262451172, "learning_rate": 1.2845442779198916e-05, "loss": 0.14657926559448242, "step": 16170 }, { "epoch": 0.4262058655989569, "grad_norm": 0.6473158597946167, "learning_rate": 1.2837263034264773e-05, "loss": 0.1418112635612488, "step": 16180 }, { "epoch": 0.4264692808434556, "grad_norm": 1.3660576343536377, "learning_rate": 1.2829081224248183e-05, "loss": 0.15127923488616943, "step": 16190 }, { "epoch": 0.42673269608795433, "grad_norm": 1.96121346950531, "learning_rate": 1.2820897355104224e-05, "loss": 0.1610713005065918, "step": 16200 }, { "epoch": 0.4269961113324531, "grad_norm": 1.6136713027954102, "learning_rate": 1.2812711432789464e-05, "loss": 0.15320825576782227, "step": 16210 }, { "epoch": 0.4272595265769518, "grad_norm": 1.6779085397720337, "learning_rate": 1.2804523463261975e-05, "loss": 0.13766815662384033, "step": 16220 }, { "epoch": 0.42752294182145056, "grad_norm": 1.883296251296997, "learning_rate": 1.2796333452481309e-05, "loss": 0.15491061210632323, "step": 16230 }, { "epoch": 0.4277863570659493, "grad_norm": 3.558016777038574, "learning_rate": 1.2788141406408516e-05, "loss": 0.156374192237854, "step": 16240 }, { "epoch": 0.42804977231044805, "grad_norm": 0.9227489829063416, "learning_rate": 1.2779947331006113e-05, "loss": 0.14470560550689698, "step": 16250 }, { "epoch": 0.4283131875549468, "grad_norm": 2.4328908920288086, "learning_rate": 1.2771751232238107e-05, "loss": 0.14235845804214478, "step": 16260 }, { "epoch": 0.42857660279944554, "grad_norm": 4.114322662353516, "learning_rate": 1.2763553116069969e-05, "loss": 0.150870943069458, "step": 16270 }, { "epoch": 0.4288400180439442, "grad_norm": 1.1268337965011597, "learning_rate": 1.2755352988468646e-05, "loss": 0.13923630714416504, "step": 16280 }, { "epoch": 0.42910343328844297, "grad_norm": 1.943317174911499, "learning_rate": 1.2747150855402541e-05, "loss": 0.1570387601852417, "step": 16290 }, { "epoch": 0.4293668485329417, "grad_norm": 0.7445914149284363, "learning_rate": 1.2738946722841519e-05, "loss": 0.14291629791259766, "step": 16300 }, { "epoch": 0.42963026377744046, "grad_norm": 3.2605090141296387, "learning_rate": 1.2730740596756908e-05, "loss": 0.1344521999359131, "step": 16310 }, { "epoch": 0.4298936790219392, "grad_norm": 3.782573938369751, "learning_rate": 1.2722532483121475e-05, "loss": 0.15694608688354492, "step": 16320 }, { "epoch": 0.43015709426643794, "grad_norm": 1.2580121755599976, "learning_rate": 1.271432238790944e-05, "loss": 0.14579999446868896, "step": 16330 }, { "epoch": 0.4304205095109367, "grad_norm": 2.4394145011901855, "learning_rate": 1.2706110317096463e-05, "loss": 0.1475456476211548, "step": 16340 }, { "epoch": 0.43068392475543543, "grad_norm": 4.306981563568115, "learning_rate": 1.269789627665965e-05, "loss": 0.1603819966316223, "step": 16350 }, { "epoch": 0.4309473399999341, "grad_norm": 1.471011757850647, "learning_rate": 1.2689680272577527e-05, "loss": 0.16800739765167236, "step": 16360 }, { "epoch": 0.43121075524443286, "grad_norm": 1.590478539466858, "learning_rate": 1.2681462310830061e-05, "loss": 0.15029234886169435, "step": 16370 }, { "epoch": 0.4314741704889316, "grad_norm": 2.9027395248413086, "learning_rate": 1.2673242397398635e-05, "loss": 0.15278072357177735, "step": 16380 }, { "epoch": 0.43173758573343035, "grad_norm": 0.9303871989250183, "learning_rate": 1.266502053826606e-05, "loss": 0.1581754207611084, "step": 16390 }, { "epoch": 0.4320010009779291, "grad_norm": 3.5779833793640137, "learning_rate": 1.2656796739416557e-05, "loss": 0.16263647079467775, "step": 16400 }, { "epoch": 0.43226441622242784, "grad_norm": 1.4720972776412964, "learning_rate": 1.2648571006835764e-05, "loss": 0.13998794555664062, "step": 16410 }, { "epoch": 0.4325278314669266, "grad_norm": 0.8417305946350098, "learning_rate": 1.2640343346510724e-05, "loss": 0.1475058078765869, "step": 16420 }, { "epoch": 0.4327912467114253, "grad_norm": 0.5784643292427063, "learning_rate": 1.2632113764429884e-05, "loss": 0.15715216398239135, "step": 16430 }, { "epoch": 0.43305466195592407, "grad_norm": 1.644727110862732, "learning_rate": 1.2623882266583083e-05, "loss": 0.15223314762115478, "step": 16440 }, { "epoch": 0.43331807720042276, "grad_norm": 1.300498366355896, "learning_rate": 1.2615648858961564e-05, "loss": 0.1410151243209839, "step": 16450 }, { "epoch": 0.4335814924449215, "grad_norm": 2.7547824382781982, "learning_rate": 1.260741354755796e-05, "loss": 0.15437376499176025, "step": 16460 }, { "epoch": 0.43384490768942024, "grad_norm": 0.5890790224075317, "learning_rate": 1.2599176338366282e-05, "loss": 0.16392863988876344, "step": 16470 }, { "epoch": 0.434108322933919, "grad_norm": 2.423630714416504, "learning_rate": 1.2590937237381927e-05, "loss": 0.14024412631988525, "step": 16480 }, { "epoch": 0.43437173817841773, "grad_norm": 1.4590803384780884, "learning_rate": 1.2582696250601667e-05, "loss": 0.1568592071533203, "step": 16490 }, { "epoch": 0.4346351534229165, "grad_norm": 2.032033681869507, "learning_rate": 1.257445338402365e-05, "loss": 0.15171303749084472, "step": 16500 }, { "epoch": 0.4348985686674152, "grad_norm": 0.817369282245636, "learning_rate": 1.256620864364739e-05, "loss": 0.15689210891723632, "step": 16510 }, { "epoch": 0.43516198391191396, "grad_norm": 1.2582857608795166, "learning_rate": 1.255796203547376e-05, "loss": 0.14704786539077758, "step": 16520 }, { "epoch": 0.4354253991564127, "grad_norm": 1.5950223207473755, "learning_rate": 1.2549713565505001e-05, "loss": 0.1466013550758362, "step": 16530 }, { "epoch": 0.4356888144009114, "grad_norm": 1.5131056308746338, "learning_rate": 1.254146323974471e-05, "loss": 0.14338183403015137, "step": 16540 }, { "epoch": 0.43595222964541014, "grad_norm": 2.0403242111206055, "learning_rate": 1.253321106419782e-05, "loss": 0.1403353214263916, "step": 16550 }, { "epoch": 0.4362156448899089, "grad_norm": 0.8720515966415405, "learning_rate": 1.2524957044870627e-05, "loss": 0.14663426876068114, "step": 16560 }, { "epoch": 0.4364790601344076, "grad_norm": 1.1131161451339722, "learning_rate": 1.2516701187770761e-05, "loss": 0.149292516708374, "step": 16570 }, { "epoch": 0.43674247537890637, "grad_norm": 1.7151411771774292, "learning_rate": 1.250844349890719e-05, "loss": 0.15290756225585939, "step": 16580 }, { "epoch": 0.4370058906234051, "grad_norm": 1.4570950269699097, "learning_rate": 1.250018398429022e-05, "loss": 0.16414263248443603, "step": 16590 }, { "epoch": 0.43726930586790386, "grad_norm": 3.5160112380981445, "learning_rate": 1.2491922649931475e-05, "loss": 0.16681311130523682, "step": 16600 }, { "epoch": 0.4375327211124026, "grad_norm": 3.249668598175049, "learning_rate": 1.2483659501843916e-05, "loss": 0.159699010848999, "step": 16610 }, { "epoch": 0.4377961363569013, "grad_norm": 0.7401993870735168, "learning_rate": 1.2475394546041822e-05, "loss": 0.1390521764755249, "step": 16620 }, { "epoch": 0.43805955160140003, "grad_norm": 1.4742740392684937, "learning_rate": 1.2467127788540774e-05, "loss": 0.13399468660354613, "step": 16630 }, { "epoch": 0.4383229668458988, "grad_norm": 1.7406609058380127, "learning_rate": 1.245885923535768e-05, "loss": 0.15438506603240967, "step": 16640 }, { "epoch": 0.4385863820903975, "grad_norm": 0.9391282796859741, "learning_rate": 1.2450588892510756e-05, "loss": 0.15628608465194702, "step": 16650 }, { "epoch": 0.43884979733489626, "grad_norm": 0.7546727061271667, "learning_rate": 1.2442316766019502e-05, "loss": 0.15937927961349488, "step": 16660 }, { "epoch": 0.439113212579395, "grad_norm": 1.6731433868408203, "learning_rate": 1.2434042861904743e-05, "loss": 0.1470130443572998, "step": 16670 }, { "epoch": 0.43937662782389375, "grad_norm": 2.7902889251708984, "learning_rate": 1.2425767186188569e-05, "loss": 0.14178574085235596, "step": 16680 }, { "epoch": 0.4396400430683925, "grad_norm": 3.4582934379577637, "learning_rate": 1.2417489744894381e-05, "loss": 0.15108039379119872, "step": 16690 }, { "epoch": 0.43990345831289124, "grad_norm": 3.6979737281799316, "learning_rate": 1.2409210544046859e-05, "loss": 0.14404938220977784, "step": 16700 }, { "epoch": 0.4401668735573899, "grad_norm": 1.7555265426635742, "learning_rate": 1.2400929589671962e-05, "loss": 0.15075608491897582, "step": 16710 }, { "epoch": 0.44043028880188867, "grad_norm": 3.6704673767089844, "learning_rate": 1.2392646887796924e-05, "loss": 0.1623008131980896, "step": 16720 }, { "epoch": 0.4406937040463874, "grad_norm": 1.0087358951568604, "learning_rate": 1.2384362444450253e-05, "loss": 0.15016934871673585, "step": 16730 }, { "epoch": 0.44095711929088616, "grad_norm": 0.6554370522499084, "learning_rate": 1.2376076265661722e-05, "loss": 0.1626883864402771, "step": 16740 }, { "epoch": 0.4412205345353849, "grad_norm": 1.4777463674545288, "learning_rate": 1.2367788357462375e-05, "loss": 0.15668357610702516, "step": 16750 }, { "epoch": 0.44148394977988364, "grad_norm": 1.5420607328414917, "learning_rate": 1.2359498725884509e-05, "loss": 0.15315401554107666, "step": 16760 }, { "epoch": 0.4417473650243824, "grad_norm": 1.7474833726882935, "learning_rate": 1.2351207376961666e-05, "loss": 0.1522204875946045, "step": 16770 }, { "epoch": 0.44201078026888113, "grad_norm": 2.248833656311035, "learning_rate": 1.234291431672866e-05, "loss": 0.16059730052947999, "step": 16780 }, { "epoch": 0.4422741955133799, "grad_norm": 1.4765688180923462, "learning_rate": 1.2334619551221531e-05, "loss": 0.14495811462402344, "step": 16790 }, { "epoch": 0.44253761075787856, "grad_norm": 1.7608479261398315, "learning_rate": 1.232632308647757e-05, "loss": 0.15640002489089966, "step": 16800 }, { "epoch": 0.4428010260023773, "grad_norm": 2.9225287437438965, "learning_rate": 1.2318024928535296e-05, "loss": 0.15085290670394896, "step": 16810 }, { "epoch": 0.44306444124687605, "grad_norm": 2.0595269203186035, "learning_rate": 1.2309725083434477e-05, "loss": 0.15033950805664062, "step": 16820 }, { "epoch": 0.4433278564913748, "grad_norm": 0.9308318495750427, "learning_rate": 1.2301423557216088e-05, "loss": 0.15390453338623047, "step": 16830 }, { "epoch": 0.44359127173587354, "grad_norm": 2.5788233280181885, "learning_rate": 1.2293120355922346e-05, "loss": 0.14212026596069335, "step": 16840 }, { "epoch": 0.4438546869803723, "grad_norm": 0.8827166557312012, "learning_rate": 1.2284815485596671e-05, "loss": 0.13745789527893065, "step": 16850 }, { "epoch": 0.444118102224871, "grad_norm": 2.1150527000427246, "learning_rate": 1.2276508952283712e-05, "loss": 0.14669418334960938, "step": 16860 }, { "epoch": 0.44438151746936977, "grad_norm": 3.668468952178955, "learning_rate": 1.2268200762029319e-05, "loss": 0.14273782968521118, "step": 16870 }, { "epoch": 0.44464493271386846, "grad_norm": 1.3969658613204956, "learning_rate": 1.2259890920880555e-05, "loss": 0.15137676000595093, "step": 16880 }, { "epoch": 0.4449083479583672, "grad_norm": 2.7017505168914795, "learning_rate": 1.2251579434885679e-05, "loss": 0.1505270004272461, "step": 16890 }, { "epoch": 0.44517176320286594, "grad_norm": 0.9987176060676575, "learning_rate": 1.2243266310094146e-05, "loss": 0.1468098759651184, "step": 16900 }, { "epoch": 0.4454351784473647, "grad_norm": 1.8225924968719482, "learning_rate": 1.2234951552556613e-05, "loss": 0.15160313844680787, "step": 16910 }, { "epoch": 0.44569859369186343, "grad_norm": 1.1762598752975464, "learning_rate": 1.2226635168324916e-05, "loss": 0.14555127620697023, "step": 16920 }, { "epoch": 0.4459620089363622, "grad_norm": 2.0142297744750977, "learning_rate": 1.2218317163452078e-05, "loss": 0.1419994592666626, "step": 16930 }, { "epoch": 0.4462254241808609, "grad_norm": 2.9710917472839355, "learning_rate": 1.2209997543992301e-05, "loss": 0.14493963718414307, "step": 16940 }, { "epoch": 0.44648883942535966, "grad_norm": 4.444246768951416, "learning_rate": 1.2201676316000964e-05, "loss": 0.15189681053161622, "step": 16950 }, { "epoch": 0.4467522546698584, "grad_norm": 1.5777620077133179, "learning_rate": 1.2193353485534614e-05, "loss": 0.14806873798370362, "step": 16960 }, { "epoch": 0.4470156699143571, "grad_norm": 1.243222951889038, "learning_rate": 1.2185029058650972e-05, "loss": 0.14677374362945556, "step": 16970 }, { "epoch": 0.44727908515885584, "grad_norm": 1.2457711696624756, "learning_rate": 1.2176703041408908e-05, "loss": 0.1465747594833374, "step": 16980 }, { "epoch": 0.4475425004033546, "grad_norm": 2.9702799320220947, "learning_rate": 1.2168375439868464e-05, "loss": 0.14445033073425292, "step": 16990 }, { "epoch": 0.4478059156478533, "grad_norm": 0.8408448100090027, "learning_rate": 1.2160046260090816e-05, "loss": 0.142634379863739, "step": 17000 }, { "epoch": 0.44806933089235207, "grad_norm": 2.1958985328674316, "learning_rate": 1.2151715508138313e-05, "loss": 0.14240033626556398, "step": 17010 }, { "epoch": 0.4483327461368508, "grad_norm": 2.4732284545898438, "learning_rate": 1.2143383190074429e-05, "loss": 0.14789618253707887, "step": 17020 }, { "epoch": 0.44859616138134956, "grad_norm": 2.809502601623535, "learning_rate": 1.2135049311963787e-05, "loss": 0.1579753875732422, "step": 17030 }, { "epoch": 0.4488595766258483, "grad_norm": 0.9917369484901428, "learning_rate": 1.2126713879872138e-05, "loss": 0.15039727687835694, "step": 17040 }, { "epoch": 0.44912299187034704, "grad_norm": 1.2324049472808838, "learning_rate": 1.2118376899866375e-05, "loss": 0.15757911205291747, "step": 17050 }, { "epoch": 0.44938640711484573, "grad_norm": 4.149001598358154, "learning_rate": 1.2110038378014512e-05, "loss": 0.14920512437820435, "step": 17060 }, { "epoch": 0.4496498223593445, "grad_norm": 1.1431922912597656, "learning_rate": 1.210169832038568e-05, "loss": 0.1469558596611023, "step": 17070 }, { "epoch": 0.4499132376038432, "grad_norm": 1.6911019086837769, "learning_rate": 1.2093356733050137e-05, "loss": 0.13894379138946533, "step": 17080 }, { "epoch": 0.45017665284834196, "grad_norm": 4.851065158843994, "learning_rate": 1.2085013622079248e-05, "loss": 0.15715425014495848, "step": 17090 }, { "epoch": 0.4504400680928407, "grad_norm": 2.2012133598327637, "learning_rate": 1.2076668993545491e-05, "loss": 0.1455470085144043, "step": 17100 }, { "epoch": 0.45070348333733945, "grad_norm": 3.192586898803711, "learning_rate": 1.206832285352244e-05, "loss": 0.16603305339813232, "step": 17110 }, { "epoch": 0.4509668985818382, "grad_norm": 1.459620475769043, "learning_rate": 1.2059975208084787e-05, "loss": 0.15295377969741822, "step": 17120 }, { "epoch": 0.45123031382633694, "grad_norm": 2.6092729568481445, "learning_rate": 1.2051626063308299e-05, "loss": 0.14584650993347167, "step": 17130 }, { "epoch": 0.4514937290708356, "grad_norm": 0.9535001516342163, "learning_rate": 1.2043275425269854e-05, "loss": 0.14776766300201416, "step": 17140 }, { "epoch": 0.45175714431533437, "grad_norm": 1.380192756652832, "learning_rate": 1.2034923300047391e-05, "loss": 0.14686298370361328, "step": 17150 }, { "epoch": 0.4520205595598331, "grad_norm": 2.2692880630493164, "learning_rate": 1.2026569693719963e-05, "loss": 0.14073947668075562, "step": 17160 }, { "epoch": 0.45228397480433186, "grad_norm": 1.1598618030548096, "learning_rate": 1.2018214612367673e-05, "loss": 0.14537529945373534, "step": 17170 }, { "epoch": 0.4525473900488306, "grad_norm": 1.3676021099090576, "learning_rate": 1.200985806207172e-05, "loss": 0.1495693802833557, "step": 17180 }, { "epoch": 0.45281080529332934, "grad_norm": 3.4384801387786865, "learning_rate": 1.2001500048914354e-05, "loss": 0.13843708038330077, "step": 17190 }, { "epoch": 0.4530742205378281, "grad_norm": 0.7140171527862549, "learning_rate": 1.1993140578978898e-05, "loss": 0.1516636848449707, "step": 17200 }, { "epoch": 0.45333763578232683, "grad_norm": 1.3953999280929565, "learning_rate": 1.1984779658349743e-05, "loss": 0.15099825859069824, "step": 17210 }, { "epoch": 0.4536010510268256, "grad_norm": 2.163011312484741, "learning_rate": 1.1976417293112318e-05, "loss": 0.1375213861465454, "step": 17220 }, { "epoch": 0.45386446627132426, "grad_norm": 2.040428400039673, "learning_rate": 1.1968053489353122e-05, "loss": 0.1371483564376831, "step": 17230 }, { "epoch": 0.454127881515823, "grad_norm": 2.632981538772583, "learning_rate": 1.195968825315968e-05, "loss": 0.15544165372848512, "step": 17240 }, { "epoch": 0.45439129676032175, "grad_norm": 2.096789598464966, "learning_rate": 1.1951321590620583e-05, "loss": 0.1493367314338684, "step": 17250 }, { "epoch": 0.4546547120048205, "grad_norm": 3.219350814819336, "learning_rate": 1.1942953507825446e-05, "loss": 0.15230982303619384, "step": 17260 }, { "epoch": 0.45491812724931924, "grad_norm": 1.630454182624817, "learning_rate": 1.1934584010864921e-05, "loss": 0.14926600456237793, "step": 17270 }, { "epoch": 0.455181542493818, "grad_norm": 0.9075918197631836, "learning_rate": 1.1926213105830683e-05, "loss": 0.1471931219100952, "step": 17280 }, { "epoch": 0.4554449577383167, "grad_norm": 2.3459527492523193, "learning_rate": 1.1917840798815448e-05, "loss": 0.16466504335403442, "step": 17290 }, { "epoch": 0.45570837298281547, "grad_norm": 1.9068161249160767, "learning_rate": 1.1909467095912932e-05, "loss": 0.16145302057266236, "step": 17300 }, { "epoch": 0.4559717882273142, "grad_norm": 4.028242111206055, "learning_rate": 1.1901092003217885e-05, "loss": 0.15057129859924318, "step": 17310 }, { "epoch": 0.4562352034718129, "grad_norm": 1.8293356895446777, "learning_rate": 1.1892715526826058e-05, "loss": 0.15925902128219604, "step": 17320 }, { "epoch": 0.45649861871631164, "grad_norm": 0.8016414642333984, "learning_rate": 1.1884337672834212e-05, "loss": 0.14280552864074708, "step": 17330 }, { "epoch": 0.4567620339608104, "grad_norm": 1.8308782577514648, "learning_rate": 1.1875958447340108e-05, "loss": 0.13338669538497924, "step": 17340 }, { "epoch": 0.45702544920530913, "grad_norm": 4.138119697570801, "learning_rate": 1.1867577856442514e-05, "loss": 0.14858274459838866, "step": 17350 }, { "epoch": 0.4572888644498079, "grad_norm": 2.331359386444092, "learning_rate": 1.1859195906241182e-05, "loss": 0.1509074330329895, "step": 17360 }, { "epoch": 0.4575522796943066, "grad_norm": 5.339405059814453, "learning_rate": 1.1850812602836851e-05, "loss": 0.1555350661277771, "step": 17370 }, { "epoch": 0.45781569493880536, "grad_norm": 1.3874084949493408, "learning_rate": 1.1842427952331263e-05, "loss": 0.1387178897857666, "step": 17380 }, { "epoch": 0.4580791101833041, "grad_norm": 2.686893939971924, "learning_rate": 1.183404196082712e-05, "loss": 0.1639796495437622, "step": 17390 }, { "epoch": 0.4583425254278028, "grad_norm": 1.7029417753219604, "learning_rate": 1.1825654634428111e-05, "loss": 0.1436835527420044, "step": 17400 }, { "epoch": 0.45860594067230154, "grad_norm": 2.4891371726989746, "learning_rate": 1.1817265979238892e-05, "loss": 0.15924472808837892, "step": 17410 }, { "epoch": 0.4588693559168003, "grad_norm": 1.0126601457595825, "learning_rate": 1.1808876001365093e-05, "loss": 0.15389304161071776, "step": 17420 }, { "epoch": 0.459132771161299, "grad_norm": 3.5715856552124023, "learning_rate": 1.1800484706913294e-05, "loss": 0.15005779266357422, "step": 17430 }, { "epoch": 0.45939618640579777, "grad_norm": 2.9089934825897217, "learning_rate": 1.1792092101991052e-05, "loss": 0.14764821529388428, "step": 17440 }, { "epoch": 0.4596596016502965, "grad_norm": 4.055206298828125, "learning_rate": 1.1783698192706856e-05, "loss": 0.15012013912200928, "step": 17450 }, { "epoch": 0.45992301689479526, "grad_norm": 2.1168246269226074, "learning_rate": 1.1775302985170161e-05, "loss": 0.15013136863708496, "step": 17460 }, { "epoch": 0.460186432139294, "grad_norm": 1.5470592975616455, "learning_rate": 1.1766906485491361e-05, "loss": 0.15642218589782714, "step": 17470 }, { "epoch": 0.46044984738379274, "grad_norm": 1.1695713996887207, "learning_rate": 1.175850869978179e-05, "loss": 0.16491591930389404, "step": 17480 }, { "epoch": 0.46071326262829143, "grad_norm": 1.0016850233078003, "learning_rate": 1.1750109634153719e-05, "loss": 0.13391361236572266, "step": 17490 }, { "epoch": 0.4609766778727902, "grad_norm": 2.997995138168335, "learning_rate": 1.1741709294720346e-05, "loss": 0.13473234176635743, "step": 17500 }, { "epoch": 0.4612400931172889, "grad_norm": 2.5508968830108643, "learning_rate": 1.173330768759581e-05, "loss": 0.1433854103088379, "step": 17510 }, { "epoch": 0.46150350836178766, "grad_norm": 2.1634647846221924, "learning_rate": 1.1724904818895158e-05, "loss": 0.14395198822021485, "step": 17520 }, { "epoch": 0.4617669236062864, "grad_norm": 1.7531120777130127, "learning_rate": 1.1716500694734359e-05, "loss": 0.16201512813568114, "step": 17530 }, { "epoch": 0.46203033885078515, "grad_norm": 1.165075421333313, "learning_rate": 1.1708095321230295e-05, "loss": 0.14956555366516114, "step": 17540 }, { "epoch": 0.4622937540952839, "grad_norm": 1.1248748302459717, "learning_rate": 1.1699688704500765e-05, "loss": 0.13733935356140137, "step": 17550 }, { "epoch": 0.46255716933978264, "grad_norm": 1.5886026620864868, "learning_rate": 1.1691280850664463e-05, "loss": 0.1441091537475586, "step": 17560 }, { "epoch": 0.4628205845842814, "grad_norm": 1.1742891073226929, "learning_rate": 1.1682871765840996e-05, "loss": 0.15277775526046752, "step": 17570 }, { "epoch": 0.46308399982878007, "grad_norm": 2.2667734622955322, "learning_rate": 1.1674461456150845e-05, "loss": 0.1559783935546875, "step": 17580 }, { "epoch": 0.4633474150732788, "grad_norm": 3.3448452949523926, "learning_rate": 1.1666049927715408e-05, "loss": 0.15163702964782716, "step": 17590 }, { "epoch": 0.46361083031777756, "grad_norm": 1.2671579122543335, "learning_rate": 1.1657637186656952e-05, "loss": 0.14019668102264404, "step": 17600 }, { "epoch": 0.4638742455622763, "grad_norm": 1.2096498012542725, "learning_rate": 1.1649223239098639e-05, "loss": 0.15536997318267823, "step": 17610 }, { "epoch": 0.46413766080677504, "grad_norm": 2.321362018585205, "learning_rate": 1.1640808091164495e-05, "loss": 0.1455417275428772, "step": 17620 }, { "epoch": 0.4644010760512738, "grad_norm": 3.397284746170044, "learning_rate": 1.1632391748979435e-05, "loss": 0.14103397130966186, "step": 17630 }, { "epoch": 0.46466449129577253, "grad_norm": 1.4136866331100464, "learning_rate": 1.162397421866923e-05, "loss": 0.14846479892730713, "step": 17640 }, { "epoch": 0.4649279065402713, "grad_norm": 2.6794941425323486, "learning_rate": 1.1615555506360526e-05, "loss": 0.1501457452774048, "step": 17650 }, { "epoch": 0.46519132178476996, "grad_norm": 2.3812649250030518, "learning_rate": 1.1607135618180823e-05, "loss": 0.1570077657699585, "step": 17660 }, { "epoch": 0.4654547370292687, "grad_norm": 2.027519702911377, "learning_rate": 1.1598714560258477e-05, "loss": 0.15251264572143555, "step": 17670 }, { "epoch": 0.46571815227376745, "grad_norm": 1.6992186307907104, "learning_rate": 1.1590292338722702e-05, "loss": 0.1550013780593872, "step": 17680 }, { "epoch": 0.4659815675182662, "grad_norm": 1.890346884727478, "learning_rate": 1.1581868959703553e-05, "loss": 0.1401194453239441, "step": 17690 }, { "epoch": 0.46624498276276494, "grad_norm": 3.3695149421691895, "learning_rate": 1.1573444429331926e-05, "loss": 0.14342164993286133, "step": 17700 }, { "epoch": 0.4665083980072637, "grad_norm": 1.7362253665924072, "learning_rate": 1.1565018753739559e-05, "loss": 0.150391149520874, "step": 17710 }, { "epoch": 0.4667718132517624, "grad_norm": 1.628693699836731, "learning_rate": 1.1556591939059024e-05, "loss": 0.14640495777130128, "step": 17720 }, { "epoch": 0.46703522849626117, "grad_norm": 4.019689083099365, "learning_rate": 1.1548163991423716e-05, "loss": 0.14892507791519166, "step": 17730 }, { "epoch": 0.4672986437407599, "grad_norm": 1.5136631727218628, "learning_rate": 1.1539734916967864e-05, "loss": 0.14380259513854982, "step": 17740 }, { "epoch": 0.4675620589852586, "grad_norm": 0.9641938209533691, "learning_rate": 1.1531304721826506e-05, "loss": 0.14055668115615844, "step": 17750 }, { "epoch": 0.46782547422975734, "grad_norm": 2.9444046020507812, "learning_rate": 1.1522873412135509e-05, "loss": 0.1518497347831726, "step": 17760 }, { "epoch": 0.4680888894742561, "grad_norm": 1.287255883216858, "learning_rate": 1.1514440994031536e-05, "loss": 0.14322730302810668, "step": 17770 }, { "epoch": 0.46835230471875483, "grad_norm": 0.892937958240509, "learning_rate": 1.1506007473652075e-05, "loss": 0.14799716472625732, "step": 17780 }, { "epoch": 0.4686157199632536, "grad_norm": 4.604252815246582, "learning_rate": 1.1497572857135397e-05, "loss": 0.14885451793670654, "step": 17790 }, { "epoch": 0.4688791352077523, "grad_norm": 3.320729970932007, "learning_rate": 1.1489137150620582e-05, "loss": 0.1593196153640747, "step": 17800 }, { "epoch": 0.46914255045225106, "grad_norm": 1.0238037109375, "learning_rate": 1.1480700360247504e-05, "loss": 0.1542841911315918, "step": 17810 }, { "epoch": 0.4694059656967498, "grad_norm": 1.6087327003479004, "learning_rate": 1.147226249215682e-05, "loss": 0.15603346824645997, "step": 17820 }, { "epoch": 0.46966938094124855, "grad_norm": 1.4765962362289429, "learning_rate": 1.1463823552489977e-05, "loss": 0.15442261695861817, "step": 17830 }, { "epoch": 0.46993279618574724, "grad_norm": 2.2599663734436035, "learning_rate": 1.1455383547389193e-05, "loss": 0.14842382669448853, "step": 17840 }, { "epoch": 0.470196211430246, "grad_norm": 1.5237165689468384, "learning_rate": 1.1446942482997479e-05, "loss": 0.13382614850997926, "step": 17850 }, { "epoch": 0.4704596266747447, "grad_norm": 1.1381851434707642, "learning_rate": 1.1438500365458593e-05, "loss": 0.14448225498199463, "step": 17860 }, { "epoch": 0.47072304191924347, "grad_norm": 1.694810390472412, "learning_rate": 1.1430057200917084e-05, "loss": 0.1516239881515503, "step": 17870 }, { "epoch": 0.4709864571637422, "grad_norm": 2.3748412132263184, "learning_rate": 1.1421612995518243e-05, "loss": 0.1465880751609802, "step": 17880 }, { "epoch": 0.47124987240824096, "grad_norm": 1.290725827217102, "learning_rate": 1.1413167755408132e-05, "loss": 0.1570582628250122, "step": 17890 }, { "epoch": 0.4715132876527397, "grad_norm": 2.1479501724243164, "learning_rate": 1.1404721486733558e-05, "loss": 0.15277817249298095, "step": 17900 }, { "epoch": 0.47177670289723844, "grad_norm": 2.4231090545654297, "learning_rate": 1.1396274195642085e-05, "loss": 0.14163817167282106, "step": 17910 }, { "epoch": 0.47204011814173713, "grad_norm": 1.5448391437530518, "learning_rate": 1.1387825888282006e-05, "loss": 0.14796342849731445, "step": 17920 }, { "epoch": 0.4723035333862359, "grad_norm": 1.4107849597930908, "learning_rate": 1.1379376570802373e-05, "loss": 0.13670077323913574, "step": 17930 }, { "epoch": 0.4725669486307346, "grad_norm": 0.9711838960647583, "learning_rate": 1.137092624935296e-05, "loss": 0.14719635248184204, "step": 17940 }, { "epoch": 0.47283036387523336, "grad_norm": 3.7942628860473633, "learning_rate": 1.1362474930084278e-05, "loss": 0.1589948058128357, "step": 17950 }, { "epoch": 0.4730937791197321, "grad_norm": 1.0718036890029907, "learning_rate": 1.1354022619147557e-05, "loss": 0.13948471546173097, "step": 17960 }, { "epoch": 0.47335719436423085, "grad_norm": 1.0305883884429932, "learning_rate": 1.1345569322694756e-05, "loss": 0.1492333769798279, "step": 17970 }, { "epoch": 0.4736206096087296, "grad_norm": 3.4060351848602295, "learning_rate": 1.133711504687855e-05, "loss": 0.15476263761520387, "step": 17980 }, { "epoch": 0.47388402485322834, "grad_norm": 1.2418816089630127, "learning_rate": 1.1328659797852325e-05, "loss": 0.1424791097640991, "step": 17990 }, { "epoch": 0.4741474400977271, "grad_norm": 2.7394845485687256, "learning_rate": 1.1320203581770177e-05, "loss": 0.15647073984146118, "step": 18000 }, { "epoch": 0.47441085534222577, "grad_norm": 4.3799567222595215, "learning_rate": 1.1311746404786903e-05, "loss": 0.15623729228973388, "step": 18010 }, { "epoch": 0.4746742705867245, "grad_norm": 2.9589388370513916, "learning_rate": 1.1303288273058002e-05, "loss": 0.14416415691375734, "step": 18020 }, { "epoch": 0.47493768583122326, "grad_norm": 1.7486742734909058, "learning_rate": 1.1294829192739665e-05, "loss": 0.1472411870956421, "step": 18030 }, { "epoch": 0.475201101075722, "grad_norm": 0.745330810546875, "learning_rate": 1.1286369169988782e-05, "loss": 0.15340206623077393, "step": 18040 }, { "epoch": 0.47546451632022074, "grad_norm": 1.7473028898239136, "learning_rate": 1.1277908210962916e-05, "loss": 0.15620192289352416, "step": 18050 }, { "epoch": 0.4757279315647195, "grad_norm": 2.8548972606658936, "learning_rate": 1.126944632182032e-05, "loss": 0.15101770162582398, "step": 18060 }, { "epoch": 0.47599134680921823, "grad_norm": 1.7741553783416748, "learning_rate": 1.1260983508719922e-05, "loss": 0.1337494969367981, "step": 18070 }, { "epoch": 0.476254762053717, "grad_norm": 2.306640625, "learning_rate": 1.1252519777821324e-05, "loss": 0.15183064937591553, "step": 18080 }, { "epoch": 0.4765181772982157, "grad_norm": 2.8808274269104004, "learning_rate": 1.1244055135284795e-05, "loss": 0.14573724269866944, "step": 18090 }, { "epoch": 0.4767815925427144, "grad_norm": 1.7639789581298828, "learning_rate": 1.123558958727126e-05, "loss": 0.15589489936828613, "step": 18100 }, { "epoch": 0.47704500778721315, "grad_norm": 2.130861759185791, "learning_rate": 1.1227123139942322e-05, "loss": 0.1461493968963623, "step": 18110 }, { "epoch": 0.4773084230317119, "grad_norm": 1.1442781686782837, "learning_rate": 1.1218655799460216e-05, "loss": 0.15028553009033202, "step": 18120 }, { "epoch": 0.47757183827621064, "grad_norm": 1.8836586475372314, "learning_rate": 1.1210187571987845e-05, "loss": 0.15023384094238282, "step": 18130 }, { "epoch": 0.4778352535207094, "grad_norm": 4.2058000564575195, "learning_rate": 1.1201718463688745e-05, "loss": 0.15794241428375244, "step": 18140 }, { "epoch": 0.4780986687652081, "grad_norm": 1.5213243961334229, "learning_rate": 1.1193248480727101e-05, "loss": 0.15223724842071534, "step": 18150 }, { "epoch": 0.47836208400970687, "grad_norm": 2.1842844486236572, "learning_rate": 1.118477762926773e-05, "loss": 0.13670740127563477, "step": 18160 }, { "epoch": 0.4786254992542056, "grad_norm": 2.1068289279937744, "learning_rate": 1.117630591547609e-05, "loss": 0.15777530670166015, "step": 18170 }, { "epoch": 0.4788889144987043, "grad_norm": 1.53719162940979, "learning_rate": 1.1167833345518248e-05, "loss": 0.14301106929779053, "step": 18180 }, { "epoch": 0.47915232974320304, "grad_norm": 1.8680100440979004, "learning_rate": 1.1159359925560918e-05, "loss": 0.1487445831298828, "step": 18190 }, { "epoch": 0.4794157449877018, "grad_norm": 3.615814447402954, "learning_rate": 1.115088566177141e-05, "loss": 0.15305876731872559, "step": 18200 }, { "epoch": 0.47967916023220053, "grad_norm": 1.7371448278427124, "learning_rate": 1.1142410560317669e-05, "loss": 0.14505462646484374, "step": 18210 }, { "epoch": 0.4799425754766993, "grad_norm": 1.507804274559021, "learning_rate": 1.113393462736823e-05, "loss": 0.16293246746063234, "step": 18220 }, { "epoch": 0.480205990721198, "grad_norm": 0.9823640584945679, "learning_rate": 1.1125457869092251e-05, "loss": 0.15500760078430176, "step": 18230 }, { "epoch": 0.48046940596569676, "grad_norm": 1.5416265726089478, "learning_rate": 1.1116980291659474e-05, "loss": 0.15551334619522095, "step": 18240 }, { "epoch": 0.4807328212101955, "grad_norm": 1.8171042203903198, "learning_rate": 1.1108501901240254e-05, "loss": 0.1418887734413147, "step": 18250 }, { "epoch": 0.48099623645469425, "grad_norm": 1.9299026727676392, "learning_rate": 1.1100022704005524e-05, "loss": 0.1506443977355957, "step": 18260 }, { "epoch": 0.48125965169919294, "grad_norm": 3.7597007751464844, "learning_rate": 1.109154270612681e-05, "loss": 0.13991340398788452, "step": 18270 }, { "epoch": 0.4815230669436917, "grad_norm": 1.5194348096847534, "learning_rate": 1.1083061913776224e-05, "loss": 0.1421539306640625, "step": 18280 }, { "epoch": 0.4817864821881904, "grad_norm": 2.288482904434204, "learning_rate": 1.1074580333126448e-05, "loss": 0.1486757516860962, "step": 18290 }, { "epoch": 0.48204989743268917, "grad_norm": 1.2839621305465698, "learning_rate": 1.1066097970350747e-05, "loss": 0.14811025857925414, "step": 18300 }, { "epoch": 0.4823133126771879, "grad_norm": 1.153472900390625, "learning_rate": 1.1057614831622942e-05, "loss": 0.13973722457885743, "step": 18310 }, { "epoch": 0.48257672792168665, "grad_norm": 0.8812919855117798, "learning_rate": 1.1049130923117437e-05, "loss": 0.1468004107475281, "step": 18320 }, { "epoch": 0.4828401431661854, "grad_norm": 1.1885762214660645, "learning_rate": 1.1040646251009181e-05, "loss": 0.15535615682601928, "step": 18330 }, { "epoch": 0.48310355841068414, "grad_norm": 1.0491232872009277, "learning_rate": 1.1032160821473692e-05, "loss": 0.15088434219360353, "step": 18340 }, { "epoch": 0.4833669736551829, "grad_norm": 4.210526943206787, "learning_rate": 1.1023674640687019e-05, "loss": 0.15865947008132936, "step": 18350 }, { "epoch": 0.4836303888996816, "grad_norm": 0.8274266719818115, "learning_rate": 1.1015187714825782e-05, "loss": 0.14661097526550293, "step": 18360 }, { "epoch": 0.4838938041441803, "grad_norm": 1.0464576482772827, "learning_rate": 1.1006700050067122e-05, "loss": 0.13058682680130004, "step": 18370 }, { "epoch": 0.48415721938867906, "grad_norm": 1.8388075828552246, "learning_rate": 1.0998211652588737e-05, "loss": 0.14245762825012206, "step": 18380 }, { "epoch": 0.4844206346331778, "grad_norm": 2.2456421852111816, "learning_rate": 1.0989722528568845e-05, "loss": 0.15558948516845703, "step": 18390 }, { "epoch": 0.48468404987767655, "grad_norm": 2.4703099727630615, "learning_rate": 1.0981232684186191e-05, "loss": 0.15526657104492186, "step": 18400 }, { "epoch": 0.4849474651221753, "grad_norm": 1.6357272863388062, "learning_rate": 1.0972742125620062e-05, "loss": 0.14495152235031128, "step": 18410 }, { "epoch": 0.48521088036667404, "grad_norm": 2.186291217803955, "learning_rate": 1.0964250859050245e-05, "loss": 0.14678369760513305, "step": 18420 }, { "epoch": 0.4854742956111728, "grad_norm": 1.6286733150482178, "learning_rate": 1.0955758890657051e-05, "loss": 0.13941894769668578, "step": 18430 }, { "epoch": 0.48573771085567147, "grad_norm": 3.0857958793640137, "learning_rate": 1.0947266226621298e-05, "loss": 0.13353216648101807, "step": 18440 }, { "epoch": 0.4860011261001702, "grad_norm": 0.8282641768455505, "learning_rate": 1.0938772873124321e-05, "loss": 0.15016164779663085, "step": 18450 }, { "epoch": 0.48626454134466895, "grad_norm": 1.0152932405471802, "learning_rate": 1.0930278836347943e-05, "loss": 0.13740923404693603, "step": 18460 }, { "epoch": 0.4865279565891677, "grad_norm": 0.9270750880241394, "learning_rate": 1.0921784122474495e-05, "loss": 0.14765238761901855, "step": 18470 }, { "epoch": 0.48679137183366644, "grad_norm": 2.4092719554901123, "learning_rate": 1.0913288737686789e-05, "loss": 0.15641547441482545, "step": 18480 }, { "epoch": 0.4870547870781652, "grad_norm": 2.239112377166748, "learning_rate": 1.090479268816814e-05, "loss": 0.15607192516326904, "step": 18490 }, { "epoch": 0.48731820232266393, "grad_norm": 3.714046001434326, "learning_rate": 1.0896295980102334e-05, "loss": 0.15532811880111694, "step": 18500 }, { "epoch": 0.4875816175671627, "grad_norm": 1.1461067199707031, "learning_rate": 1.0887798619673648e-05, "loss": 0.13753411769866944, "step": 18510 }, { "epoch": 0.4878450328116614, "grad_norm": 0.8773181438446045, "learning_rate": 1.087930061306682e-05, "loss": 0.15117764472961426, "step": 18520 }, { "epoch": 0.4881084480561601, "grad_norm": 1.408869743347168, "learning_rate": 1.0870801966467073e-05, "loss": 0.14712955951690673, "step": 18530 }, { "epoch": 0.48837186330065885, "grad_norm": 1.2071490287780762, "learning_rate": 1.0862302686060082e-05, "loss": 0.14097331762313842, "step": 18540 }, { "epoch": 0.4886352785451576, "grad_norm": 3.825190544128418, "learning_rate": 1.0853802778031995e-05, "loss": 0.15006766319274903, "step": 18550 }, { "epoch": 0.48889869378965634, "grad_norm": 4.320367813110352, "learning_rate": 1.0845302248569411e-05, "loss": 0.1507641315460205, "step": 18560 }, { "epoch": 0.4891621090341551, "grad_norm": 2.153562545776367, "learning_rate": 1.0836801103859377e-05, "loss": 0.14611829519271852, "step": 18570 }, { "epoch": 0.4894255242786538, "grad_norm": 0.993827760219574, "learning_rate": 1.0828299350089401e-05, "loss": 0.13815746307373047, "step": 18580 }, { "epoch": 0.48968893952315257, "grad_norm": 1.5341682434082031, "learning_rate": 1.0819796993447422e-05, "loss": 0.14339524507522583, "step": 18590 }, { "epoch": 0.4899523547676513, "grad_norm": 1.7963072061538696, "learning_rate": 1.081129404012182e-05, "loss": 0.14290082454681396, "step": 18600 }, { "epoch": 0.49021577001215005, "grad_norm": 2.653552770614624, "learning_rate": 1.0802790496301414e-05, "loss": 0.15341482162475586, "step": 18610 }, { "epoch": 0.49047918525664874, "grad_norm": 2.9902749061584473, "learning_rate": 1.079428636817545e-05, "loss": 0.1468043804168701, "step": 18620 }, { "epoch": 0.4907426005011475, "grad_norm": 0.7595354914665222, "learning_rate": 1.0785781661933596e-05, "loss": 0.13893625736236573, "step": 18630 }, { "epoch": 0.49100601574564623, "grad_norm": 0.9935056567192078, "learning_rate": 1.0777276383765953e-05, "loss": 0.14845588207244872, "step": 18640 }, { "epoch": 0.491269430990145, "grad_norm": 2.1420998573303223, "learning_rate": 1.0768770539863017e-05, "loss": 0.15609476566314698, "step": 18650 }, { "epoch": 0.4915328462346437, "grad_norm": 0.9916176199913025, "learning_rate": 1.076026413641572e-05, "loss": 0.15109313726425172, "step": 18660 }, { "epoch": 0.49179626147914246, "grad_norm": 2.1514151096343994, "learning_rate": 1.0751757179615378e-05, "loss": 0.15910646915435792, "step": 18670 }, { "epoch": 0.4920596767236412, "grad_norm": 3.1264259815216064, "learning_rate": 1.0743249675653733e-05, "loss": 0.14923373460769654, "step": 18680 }, { "epoch": 0.49232309196813995, "grad_norm": 1.3555231094360352, "learning_rate": 1.0734741630722904e-05, "loss": 0.15180232524871826, "step": 18690 }, { "epoch": 0.49258650721263864, "grad_norm": 0.8851569890975952, "learning_rate": 1.0726233051015415e-05, "loss": 0.14067842960357665, "step": 18700 }, { "epoch": 0.4928499224571374, "grad_norm": 1.7543411254882812, "learning_rate": 1.0717723942724182e-05, "loss": 0.13297219276428224, "step": 18710 }, { "epoch": 0.4931133377016361, "grad_norm": 2.0926973819732666, "learning_rate": 1.07092143120425e-05, "loss": 0.14050493240356446, "step": 18720 }, { "epoch": 0.49337675294613487, "grad_norm": 4.130524635314941, "learning_rate": 1.0700704165164042e-05, "loss": 0.14140387773513793, "step": 18730 }, { "epoch": 0.4936401681906336, "grad_norm": 2.3616907596588135, "learning_rate": 1.069219350828286e-05, "loss": 0.15662919282913207, "step": 18740 }, { "epoch": 0.49390358343513235, "grad_norm": 0.8465096950531006, "learning_rate": 1.068368234759338e-05, "loss": 0.13993241786956787, "step": 18750 }, { "epoch": 0.4941669986796311, "grad_norm": 1.3447726964950562, "learning_rate": 1.0675170689290393e-05, "loss": 0.14769163131713867, "step": 18760 }, { "epoch": 0.49443041392412984, "grad_norm": 3.552053928375244, "learning_rate": 1.0666658539569048e-05, "loss": 0.14910253286361694, "step": 18770 }, { "epoch": 0.4946938291686286, "grad_norm": 2.8416900634765625, "learning_rate": 1.0658145904624852e-05, "loss": 0.15689055919647216, "step": 18780 }, { "epoch": 0.4949572444131273, "grad_norm": 2.1981563568115234, "learning_rate": 1.0649632790653677e-05, "loss": 0.15081846714019775, "step": 18790 }, { "epoch": 0.495220659657626, "grad_norm": 0.7942739725112915, "learning_rate": 1.0641119203851724e-05, "loss": 0.1580065369606018, "step": 18800 }, { "epoch": 0.49548407490212476, "grad_norm": 1.8111763000488281, "learning_rate": 1.0632605150415561e-05, "loss": 0.1607745885848999, "step": 18810 }, { "epoch": 0.4957474901466235, "grad_norm": 1.8241603374481201, "learning_rate": 1.062409063654207e-05, "loss": 0.16682770252227783, "step": 18820 }, { "epoch": 0.49601090539112225, "grad_norm": 5.525199890136719, "learning_rate": 1.0615575668428494e-05, "loss": 0.14787087440490723, "step": 18830 }, { "epoch": 0.496274320635621, "grad_norm": 1.197196364402771, "learning_rate": 1.0607060252272383e-05, "loss": 0.14810891151428224, "step": 18840 }, { "epoch": 0.49653773588011973, "grad_norm": 2.7288379669189453, "learning_rate": 1.0598544394271633e-05, "loss": 0.13284921646118164, "step": 18850 }, { "epoch": 0.4968011511246185, "grad_norm": 1.4420405626296997, "learning_rate": 1.0590028100624448e-05, "loss": 0.14264323711395263, "step": 18860 }, { "epoch": 0.4970645663691172, "grad_norm": 3.7453436851501465, "learning_rate": 1.0581511377529354e-05, "loss": 0.15315052270889282, "step": 18870 }, { "epoch": 0.4973279816136159, "grad_norm": 1.504111647605896, "learning_rate": 1.0572994231185194e-05, "loss": 0.158036470413208, "step": 18880 }, { "epoch": 0.49759139685811465, "grad_norm": 1.1681901216506958, "learning_rate": 1.056447666779111e-05, "loss": 0.14853992462158203, "step": 18890 }, { "epoch": 0.4978548121026134, "grad_norm": 0.9252111315727234, "learning_rate": 1.0555958693546552e-05, "loss": 0.1467822551727295, "step": 18900 }, { "epoch": 0.49811822734711214, "grad_norm": 1.3560895919799805, "learning_rate": 1.0547440314651274e-05, "loss": 0.13149852752685548, "step": 18910 }, { "epoch": 0.4983816425916109, "grad_norm": 2.7110817432403564, "learning_rate": 1.0538921537305315e-05, "loss": 0.1612579345703125, "step": 18920 }, { "epoch": 0.49864505783610963, "grad_norm": 2.6908607482910156, "learning_rate": 1.0530402367709007e-05, "loss": 0.1494491457939148, "step": 18930 }, { "epoch": 0.4989084730806084, "grad_norm": 2.5408129692077637, "learning_rate": 1.0521882812062978e-05, "loss": 0.14720399379730226, "step": 18940 }, { "epoch": 0.4991718883251071, "grad_norm": 1.2001858949661255, "learning_rate": 1.0513362876568118e-05, "loss": 0.14585591554641725, "step": 18950 }, { "epoch": 0.49943530356960586, "grad_norm": 3.2976624965667725, "learning_rate": 1.0504842567425611e-05, "loss": 0.15524457693099974, "step": 18960 }, { "epoch": 0.49969871881410455, "grad_norm": 0.9318621158599854, "learning_rate": 1.0496321890836901e-05, "loss": 0.1471655011177063, "step": 18970 }, { "epoch": 0.4999621340586033, "grad_norm": 2.3298094272613525, "learning_rate": 1.0487800853003706e-05, "loss": 0.14976975917816163, "step": 18980 }, { "epoch": 0.500225549303102, "grad_norm": 2.93377947807312, "learning_rate": 1.0479279460128005e-05, "loss": 0.15026307106018066, "step": 18990 }, { "epoch": 0.5004889645476008, "grad_norm": 2.4815785884857178, "learning_rate": 1.0470757718412035e-05, "loss": 0.1423191785812378, "step": 19000 }, { "epoch": 0.5007523797920995, "grad_norm": 1.5470637083053589, "learning_rate": 1.0462235634058287e-05, "loss": 0.14289921522140503, "step": 19010 }, { "epoch": 0.5010157950365982, "grad_norm": 1.407267451286316, "learning_rate": 1.0453713213269503e-05, "loss": 0.14991720914840698, "step": 19020 }, { "epoch": 0.501279210281097, "grad_norm": 1.447847604751587, "learning_rate": 1.0445190462248667e-05, "loss": 0.14782290458679198, "step": 19030 }, { "epoch": 0.5015426255255957, "grad_norm": 2.751140832901001, "learning_rate": 1.0436667387199004e-05, "loss": 0.15335700511932374, "step": 19040 }, { "epoch": 0.5018060407700945, "grad_norm": 1.840086579322815, "learning_rate": 1.0428143994323978e-05, "loss": 0.14819610118865967, "step": 19050 }, { "epoch": 0.5020694560145932, "grad_norm": 2.196458578109741, "learning_rate": 1.041962028982728e-05, "loss": 0.15086121559143068, "step": 19060 }, { "epoch": 0.502332871259092, "grad_norm": 2.69435977935791, "learning_rate": 1.0411096279912835e-05, "loss": 0.1584662079811096, "step": 19070 }, { "epoch": 0.5025962865035907, "grad_norm": 1.3249642848968506, "learning_rate": 1.0402571970784778e-05, "loss": 0.14875478744506837, "step": 19080 }, { "epoch": 0.5028597017480895, "grad_norm": 1.3883168697357178, "learning_rate": 1.0394047368647473e-05, "loss": 0.13942580223083495, "step": 19090 }, { "epoch": 0.5031231169925882, "grad_norm": 2.5266542434692383, "learning_rate": 1.0385522479705493e-05, "loss": 0.13904592990875245, "step": 19100 }, { "epoch": 0.5033865322370868, "grad_norm": 0.8600502014160156, "learning_rate": 1.0376997310163622e-05, "loss": 0.15888376235961915, "step": 19110 }, { "epoch": 0.5036499474815856, "grad_norm": 1.872457504272461, "learning_rate": 1.0368471866226837e-05, "loss": 0.13882884979248047, "step": 19120 }, { "epoch": 0.5039133627260843, "grad_norm": 5.2002105712890625, "learning_rate": 1.0359946154100336e-05, "loss": 0.14301443099975586, "step": 19130 }, { "epoch": 0.5041767779705831, "grad_norm": 1.4263056516647339, "learning_rate": 1.035142017998949e-05, "loss": 0.15186753273010253, "step": 19140 }, { "epoch": 0.5044401932150818, "grad_norm": 1.0992447137832642, "learning_rate": 1.0342893950099879e-05, "loss": 0.14881274700164795, "step": 19150 }, { "epoch": 0.5047036084595806, "grad_norm": 0.9235718846321106, "learning_rate": 1.0334367470637257e-05, "loss": 0.14818904399871827, "step": 19160 }, { "epoch": 0.5049670237040793, "grad_norm": 0.8367726802825928, "learning_rate": 1.0325840747807562e-05, "loss": 0.14308500289916992, "step": 19170 }, { "epoch": 0.505230438948578, "grad_norm": 1.5765458345413208, "learning_rate": 1.0317313787816915e-05, "loss": 0.14935808181762694, "step": 19180 }, { "epoch": 0.5054938541930768, "grad_norm": 1.4545189142227173, "learning_rate": 1.0308786596871606e-05, "loss": 0.14882365465164185, "step": 19190 }, { "epoch": 0.5057572694375755, "grad_norm": 1.6349565982818604, "learning_rate": 1.0300259181178086e-05, "loss": 0.14654190540313722, "step": 19200 }, { "epoch": 0.5060206846820743, "grad_norm": 1.1328752040863037, "learning_rate": 1.0291731546942977e-05, "loss": 0.14329965114593507, "step": 19210 }, { "epoch": 0.506284099926573, "grad_norm": 1.7126339673995972, "learning_rate": 1.0283203700373065e-05, "loss": 0.14412920475006102, "step": 19220 }, { "epoch": 0.5065475151710718, "grad_norm": 1.5129188299179077, "learning_rate": 1.0274675647675281e-05, "loss": 0.15449779033660888, "step": 19230 }, { "epoch": 0.5068109304155705, "grad_norm": 3.3966386318206787, "learning_rate": 1.0266147395056712e-05, "loss": 0.15014539957046508, "step": 19240 }, { "epoch": 0.5070743456600693, "grad_norm": 2.8790266513824463, "learning_rate": 1.0257618948724583e-05, "loss": 0.1489260673522949, "step": 19250 }, { "epoch": 0.507337760904568, "grad_norm": 2.093777894973755, "learning_rate": 1.024909031488627e-05, "loss": 0.15964990854263306, "step": 19260 }, { "epoch": 0.5076011761490666, "grad_norm": 2.407132863998413, "learning_rate": 1.0240561499749278e-05, "loss": 0.13727097511291503, "step": 19270 }, { "epoch": 0.5078645913935654, "grad_norm": 0.7666745185852051, "learning_rate": 1.023203250952125e-05, "loss": 0.14497818946838378, "step": 19280 }, { "epoch": 0.5081280066380641, "grad_norm": 1.786975622177124, "learning_rate": 1.0223503350409949e-05, "loss": 0.14301712512969972, "step": 19290 }, { "epoch": 0.5083914218825629, "grad_norm": 1.024520993232727, "learning_rate": 1.0214974028623268e-05, "loss": 0.15659863948822023, "step": 19300 }, { "epoch": 0.5086548371270616, "grad_norm": 1.2466293573379517, "learning_rate": 1.0206444550369212e-05, "loss": 0.15827748775482178, "step": 19310 }, { "epoch": 0.5089182523715604, "grad_norm": 1.2743916511535645, "learning_rate": 1.0197914921855909e-05, "loss": 0.1486790657043457, "step": 19320 }, { "epoch": 0.5091816676160591, "grad_norm": 2.378237009048462, "learning_rate": 1.0189385149291586e-05, "loss": 0.152947998046875, "step": 19330 }, { "epoch": 0.5094450828605579, "grad_norm": 0.9919693470001221, "learning_rate": 1.018085523888458e-05, "loss": 0.14628078937530517, "step": 19340 }, { "epoch": 0.5097084981050566, "grad_norm": 3.939837694168091, "learning_rate": 1.0172325196843328e-05, "loss": 0.14831498861312867, "step": 19350 }, { "epoch": 0.5099719133495553, "grad_norm": 1.0459233522415161, "learning_rate": 1.0163795029376362e-05, "loss": 0.13486547470092775, "step": 19360 }, { "epoch": 0.5102353285940541, "grad_norm": 1.6289783716201782, "learning_rate": 1.015526474269231e-05, "loss": 0.14496183395385742, "step": 19370 }, { "epoch": 0.5104987438385528, "grad_norm": 2.6156904697418213, "learning_rate": 1.0146734342999875e-05, "loss": 0.14215093851089478, "step": 19380 }, { "epoch": 0.5107621590830516, "grad_norm": 1.900761604309082, "learning_rate": 1.0138203836507859e-05, "loss": 0.15481070280075074, "step": 19390 }, { "epoch": 0.5110255743275502, "grad_norm": 1.4245350360870361, "learning_rate": 1.0129673229425125e-05, "loss": 0.14856538772583008, "step": 19400 }, { "epoch": 0.511288989572049, "grad_norm": 2.1181039810180664, "learning_rate": 1.0121142527960624e-05, "loss": 0.1358765482902527, "step": 19410 }, { "epoch": 0.5115524048165477, "grad_norm": 2.3248021602630615, "learning_rate": 1.0112611738323362e-05, "loss": 0.13935425281524658, "step": 19420 }, { "epoch": 0.5118158200610465, "grad_norm": 2.281125068664551, "learning_rate": 1.0104080866722419e-05, "loss": 0.1351043701171875, "step": 19430 }, { "epoch": 0.5120792353055452, "grad_norm": 2.745157480239868, "learning_rate": 1.0095549919366928e-05, "loss": 0.13891422748565674, "step": 19440 }, { "epoch": 0.5123426505500439, "grad_norm": 2.458371639251709, "learning_rate": 1.0087018902466086e-05, "loss": 0.15685368776321412, "step": 19450 }, { "epoch": 0.5126060657945427, "grad_norm": 3.9919064044952393, "learning_rate": 1.0078487822229132e-05, "loss": 0.1566796064376831, "step": 19460 }, { "epoch": 0.5128694810390414, "grad_norm": 1.4364752769470215, "learning_rate": 1.0069956684865354e-05, "loss": 0.14714381694793702, "step": 19470 }, { "epoch": 0.5131328962835402, "grad_norm": 2.427948236465454, "learning_rate": 1.0061425496584084e-05, "loss": 0.1411956787109375, "step": 19480 }, { "epoch": 0.5133963115280389, "grad_norm": 4.321364402770996, "learning_rate": 1.0052894263594687e-05, "loss": 0.13279030323028565, "step": 19490 }, { "epoch": 0.5136597267725377, "grad_norm": 1.2665176391601562, "learning_rate": 1.0044362992106563e-05, "loss": 0.14335918426513672, "step": 19500 }, { "epoch": 0.5139231420170364, "grad_norm": 1.3005300760269165, "learning_rate": 1.0035831688329139e-05, "loss": 0.15436944961547852, "step": 19510 }, { "epoch": 0.5141865572615352, "grad_norm": 1.758601188659668, "learning_rate": 1.0027300358471865e-05, "loss": 0.145413601398468, "step": 19520 }, { "epoch": 0.5144499725060339, "grad_norm": 2.573235034942627, "learning_rate": 1.0018769008744214e-05, "loss": 0.15831689834594725, "step": 19530 }, { "epoch": 0.5147133877505325, "grad_norm": 1.2008917331695557, "learning_rate": 1.0010237645355669e-05, "loss": 0.14830610752105713, "step": 19540 }, { "epoch": 0.5149768029950313, "grad_norm": 1.3788396120071411, "learning_rate": 1.0001706274515724e-05, "loss": 0.1499114990234375, "step": 19550 }, { "epoch": 0.51524021823953, "grad_norm": 2.8953604698181152, "learning_rate": 9.993174902433877e-06, "loss": 0.13750765323638917, "step": 19560 }, { "epoch": 0.5155036334840288, "grad_norm": 0.9864858984947205, "learning_rate": 9.984643535319632e-06, "loss": 0.144044029712677, "step": 19570 }, { "epoch": 0.5157670487285275, "grad_norm": 2.9798007011413574, "learning_rate": 9.976112179382483e-06, "loss": 0.15198394060134887, "step": 19580 }, { "epoch": 0.5160304639730263, "grad_norm": 1.2696009874343872, "learning_rate": 9.967580840831922e-06, "loss": 0.15613181591033937, "step": 19590 }, { "epoch": 0.516293879217525, "grad_norm": 1.2626265287399292, "learning_rate": 9.959049525877425e-06, "loss": 0.1529670000076294, "step": 19600 }, { "epoch": 0.5165572944620238, "grad_norm": 2.084294557571411, "learning_rate": 9.950518240728446e-06, "loss": 0.15321764945983887, "step": 19610 }, { "epoch": 0.5168207097065225, "grad_norm": 0.9875438809394836, "learning_rate": 9.94198699159443e-06, "loss": 0.14430335760116578, "step": 19620 }, { "epoch": 0.5170841249510212, "grad_norm": 1.3571583032608032, "learning_rate": 9.93345578468478e-06, "loss": 0.15005358457565307, "step": 19630 }, { "epoch": 0.51734754019552, "grad_norm": 1.7244563102722168, "learning_rate": 9.924924626208884e-06, "loss": 0.13899292945861816, "step": 19640 }, { "epoch": 0.5176109554400187, "grad_norm": 1.0799201726913452, "learning_rate": 9.916393522376087e-06, "loss": 0.1502170205116272, "step": 19650 }, { "epoch": 0.5178743706845175, "grad_norm": 2.4042084217071533, "learning_rate": 9.907862479395684e-06, "loss": 0.15075285434722902, "step": 19660 }, { "epoch": 0.5181377859290162, "grad_norm": 2.5743892192840576, "learning_rate": 9.899331503476951e-06, "loss": 0.15020623207092285, "step": 19670 }, { "epoch": 0.518401201173515, "grad_norm": 1.4856822490692139, "learning_rate": 9.890800600829086e-06, "loss": 0.13771111965179444, "step": 19680 }, { "epoch": 0.5186646164180136, "grad_norm": 2.886068344116211, "learning_rate": 9.882269777661261e-06, "loss": 0.13484984636306763, "step": 19690 }, { "epoch": 0.5189280316625123, "grad_norm": 1.4326872825622559, "learning_rate": 9.873739040182569e-06, "loss": 0.14572675228118898, "step": 19700 }, { "epoch": 0.5191914469070111, "grad_norm": 1.7273085117340088, "learning_rate": 9.86520839460205e-06, "loss": 0.1591625690460205, "step": 19710 }, { "epoch": 0.5194548621515098, "grad_norm": 3.5859158039093018, "learning_rate": 9.856677847128678e-06, "loss": 0.1385071635246277, "step": 19720 }, { "epoch": 0.5197182773960086, "grad_norm": 0.9763888716697693, "learning_rate": 9.848147403971353e-06, "loss": 0.1425497055053711, "step": 19730 }, { "epoch": 0.5199816926405073, "grad_norm": 3.1045374870300293, "learning_rate": 9.839617071338897e-06, "loss": 0.14533352851867676, "step": 19740 }, { "epoch": 0.5202451078850061, "grad_norm": 1.8657279014587402, "learning_rate": 9.831086855440056e-06, "loss": 0.1594377040863037, "step": 19750 }, { "epoch": 0.5205085231295048, "grad_norm": 1.9625568389892578, "learning_rate": 9.82255676248349e-06, "loss": 0.13853838443756103, "step": 19760 }, { "epoch": 0.5207719383740036, "grad_norm": 3.259371280670166, "learning_rate": 9.814026798677767e-06, "loss": 0.1377662181854248, "step": 19770 }, { "epoch": 0.5210353536185023, "grad_norm": 1.0122004747390747, "learning_rate": 9.805496970231367e-06, "loss": 0.14080073833465576, "step": 19780 }, { "epoch": 0.521298768863001, "grad_norm": 1.3286235332489014, "learning_rate": 9.796967283352657e-06, "loss": 0.13713793754577636, "step": 19790 }, { "epoch": 0.5215621841074998, "grad_norm": 1.7585166692733765, "learning_rate": 9.788437744249925e-06, "loss": 0.1424018621444702, "step": 19800 }, { "epoch": 0.5218255993519985, "grad_norm": 2.373056173324585, "learning_rate": 9.779908359131327e-06, "loss": 0.13863191604614258, "step": 19810 }, { "epoch": 0.5220890145964973, "grad_norm": 1.9461766481399536, "learning_rate": 9.77137913420492e-06, "loss": 0.13851629495620726, "step": 19820 }, { "epoch": 0.522352429840996, "grad_norm": 1.747761845588684, "learning_rate": 9.762850075678644e-06, "loss": 0.1568089723587036, "step": 19830 }, { "epoch": 0.5226158450854947, "grad_norm": 2.716362714767456, "learning_rate": 9.754321189760312e-06, "loss": 0.14177896976470947, "step": 19840 }, { "epoch": 0.5228792603299934, "grad_norm": 0.9269213676452637, "learning_rate": 9.74579248265762e-06, "loss": 0.13697837591171264, "step": 19850 }, { "epoch": 0.5231426755744922, "grad_norm": 0.9754267334938049, "learning_rate": 9.737263960578127e-06, "loss": 0.1481477975845337, "step": 19860 }, { "epoch": 0.5234060908189909, "grad_norm": 1.6355106830596924, "learning_rate": 9.728735629729256e-06, "loss": 0.14911410808563233, "step": 19870 }, { "epoch": 0.5236695060634896, "grad_norm": 1.887662649154663, "learning_rate": 9.7202074963183e-06, "loss": 0.14617860317230225, "step": 19880 }, { "epoch": 0.5239329213079884, "grad_norm": 1.8116190433502197, "learning_rate": 9.711679566552396e-06, "loss": 0.14119701385498046, "step": 19890 }, { "epoch": 0.5241963365524871, "grad_norm": 1.9063464403152466, "learning_rate": 9.703151846638545e-06, "loss": 0.1384981393814087, "step": 19900 }, { "epoch": 0.5244597517969859, "grad_norm": 1.7278093099594116, "learning_rate": 9.694624342783591e-06, "loss": 0.14960803985595703, "step": 19910 }, { "epoch": 0.5247231670414846, "grad_norm": 1.4267579317092896, "learning_rate": 9.686097061194209e-06, "loss": 0.1590752363204956, "step": 19920 }, { "epoch": 0.5249865822859834, "grad_norm": 2.178631067276001, "learning_rate": 9.677570008076933e-06, "loss": 0.14331037998199464, "step": 19930 }, { "epoch": 0.5252499975304821, "grad_norm": 1.876650333404541, "learning_rate": 9.669043189638115e-06, "loss": 0.14153120517730713, "step": 19940 }, { "epoch": 0.5255134127749809, "grad_norm": 1.8227524757385254, "learning_rate": 9.66051661208394e-06, "loss": 0.1577275037765503, "step": 19950 }, { "epoch": 0.5257768280194796, "grad_norm": 1.0889731645584106, "learning_rate": 9.651990281620423e-06, "loss": 0.1399899959564209, "step": 19960 }, { "epoch": 0.5260402432639782, "grad_norm": 3.9613921642303467, "learning_rate": 9.643464204453392e-06, "loss": 0.14297497272491455, "step": 19970 }, { "epoch": 0.526303658508477, "grad_norm": 2.554314136505127, "learning_rate": 9.634938386788497e-06, "loss": 0.15805583000183104, "step": 19980 }, { "epoch": 0.5265670737529757, "grad_norm": 1.6342217922210693, "learning_rate": 9.626412834831193e-06, "loss": 0.1585165739059448, "step": 19990 }, { "epoch": 0.5268304889974745, "grad_norm": 2.2663068771362305, "learning_rate": 9.617887554786746e-06, "loss": 0.13413007259368898, "step": 20000 }, { "epoch": 0.5270939042419732, "grad_norm": 2.2398314476013184, "learning_rate": 9.609362552860225e-06, "loss": 0.13606531620025636, "step": 20010 }, { "epoch": 0.527357319486472, "grad_norm": 2.3952200412750244, "learning_rate": 9.600837835256486e-06, "loss": 0.14865610599517823, "step": 20020 }, { "epoch": 0.5276207347309707, "grad_norm": 2.4638209342956543, "learning_rate": 9.592313408180199e-06, "loss": 0.13173651695251465, "step": 20030 }, { "epoch": 0.5278841499754695, "grad_norm": 5.421745300292969, "learning_rate": 9.583789277835798e-06, "loss": 0.15996851921081542, "step": 20040 }, { "epoch": 0.5281475652199682, "grad_norm": 1.5352469682693481, "learning_rate": 9.575265450427517e-06, "loss": 0.15626535415649415, "step": 20050 }, { "epoch": 0.5284109804644669, "grad_norm": 1.5374013185501099, "learning_rate": 9.566741932159369e-06, "loss": 0.15007493495941163, "step": 20060 }, { "epoch": 0.5286743957089657, "grad_norm": 1.2262126207351685, "learning_rate": 9.558218729235132e-06, "loss": 0.15287033319473267, "step": 20070 }, { "epoch": 0.5289378109534644, "grad_norm": 3.8583967685699463, "learning_rate": 9.549695847858363e-06, "loss": 0.1474534511566162, "step": 20080 }, { "epoch": 0.5292012261979632, "grad_norm": 1.439863681793213, "learning_rate": 9.541173294232387e-06, "loss": 0.1543778657913208, "step": 20090 }, { "epoch": 0.5294646414424619, "grad_norm": 1.732956886291504, "learning_rate": 9.53265107456028e-06, "loss": 0.14619820117950438, "step": 20100 }, { "epoch": 0.5297280566869607, "grad_norm": 1.1455214023590088, "learning_rate": 9.524129195044886e-06, "loss": 0.1278425931930542, "step": 20110 }, { "epoch": 0.5299914719314593, "grad_norm": 3.349483013153076, "learning_rate": 9.515607661888799e-06, "loss": 0.15562151670455932, "step": 20120 }, { "epoch": 0.5302548871759581, "grad_norm": 2.898055076599121, "learning_rate": 9.507086481294345e-06, "loss": 0.16744141578674315, "step": 20130 }, { "epoch": 0.5305183024204568, "grad_norm": 0.9423732161521912, "learning_rate": 9.498565659463624e-06, "loss": 0.13924663066864013, "step": 20140 }, { "epoch": 0.5307817176649555, "grad_norm": 1.0853277444839478, "learning_rate": 9.490045202598442e-06, "loss": 0.13722290992736816, "step": 20150 }, { "epoch": 0.5310451329094543, "grad_norm": 1.041835904121399, "learning_rate": 9.48152511690037e-06, "loss": 0.14815928936004638, "step": 20160 }, { "epoch": 0.531308548153953, "grad_norm": 2.9852700233459473, "learning_rate": 9.473005408570685e-06, "loss": 0.1414161205291748, "step": 20170 }, { "epoch": 0.5315719633984518, "grad_norm": 1.4734115600585938, "learning_rate": 9.464486083810397e-06, "loss": 0.1325724959373474, "step": 20180 }, { "epoch": 0.5318353786429505, "grad_norm": 2.1086809635162354, "learning_rate": 9.455967148820243e-06, "loss": 0.14339475631713866, "step": 20190 }, { "epoch": 0.5320987938874493, "grad_norm": 3.167470693588257, "learning_rate": 9.44744860980067e-06, "loss": 0.1525442361831665, "step": 20200 }, { "epoch": 0.532362209131948, "grad_norm": 1.626052975654602, "learning_rate": 9.438930472951838e-06, "loss": 0.13544055223464965, "step": 20210 }, { "epoch": 0.5326256243764467, "grad_norm": 1.4637725353240967, "learning_rate": 9.430412744473615e-06, "loss": 0.14316288232803345, "step": 20220 }, { "epoch": 0.5328890396209455, "grad_norm": 2.0422468185424805, "learning_rate": 9.421895430565568e-06, "loss": 0.1458195686340332, "step": 20230 }, { "epoch": 0.5331524548654442, "grad_norm": 1.780738353729248, "learning_rate": 9.413378537426972e-06, "loss": 0.13934814929962158, "step": 20240 }, { "epoch": 0.533415870109943, "grad_norm": 1.4920772314071655, "learning_rate": 9.404862071256785e-06, "loss": 0.1412933349609375, "step": 20250 }, { "epoch": 0.5336792853544416, "grad_norm": 2.495103597640991, "learning_rate": 9.396346038253654e-06, "loss": 0.15525513887405396, "step": 20260 }, { "epoch": 0.5339427005989404, "grad_norm": 4.086065292358398, "learning_rate": 9.387830444615926e-06, "loss": 0.13215769529342652, "step": 20270 }, { "epoch": 0.5342061158434391, "grad_norm": 5.381545543670654, "learning_rate": 9.379315296541611e-06, "loss": 0.13802788257598878, "step": 20280 }, { "epoch": 0.5344695310879379, "grad_norm": 1.7758619785308838, "learning_rate": 9.370800600228399e-06, "loss": 0.1489700436592102, "step": 20290 }, { "epoch": 0.5347329463324366, "grad_norm": 0.8586306571960449, "learning_rate": 9.362286361873658e-06, "loss": 0.1460409164428711, "step": 20300 }, { "epoch": 0.5349963615769353, "grad_norm": 1.8673793077468872, "learning_rate": 9.353772587674413e-06, "loss": 0.14367693662643433, "step": 20310 }, { "epoch": 0.5352597768214341, "grad_norm": 1.9490680694580078, "learning_rate": 9.345259283827362e-06, "loss": 0.13765591382980347, "step": 20320 }, { "epoch": 0.5355231920659328, "grad_norm": 2.6504921913146973, "learning_rate": 9.336746456528853e-06, "loss": 0.1442537784576416, "step": 20330 }, { "epoch": 0.5357866073104316, "grad_norm": 0.6159127354621887, "learning_rate": 9.328234111974885e-06, "loss": 0.1378491759300232, "step": 20340 }, { "epoch": 0.5360500225549303, "grad_norm": 0.9772287607192993, "learning_rate": 9.319722256361117e-06, "loss": 0.15958596467971803, "step": 20350 }, { "epoch": 0.5363134377994291, "grad_norm": 5.0246052742004395, "learning_rate": 9.311210895882835e-06, "loss": 0.14241496324539185, "step": 20360 }, { "epoch": 0.5365768530439278, "grad_norm": 2.582528829574585, "learning_rate": 9.302700036734985e-06, "loss": 0.14271373748779298, "step": 20370 }, { "epoch": 0.5368402682884266, "grad_norm": 1.6841063499450684, "learning_rate": 9.294189685112133e-06, "loss": 0.15091180801391602, "step": 20380 }, { "epoch": 0.5371036835329253, "grad_norm": 1.7367057800292969, "learning_rate": 9.285679847208474e-06, "loss": 0.1306222915649414, "step": 20390 }, { "epoch": 0.537367098777424, "grad_norm": 1.3871266841888428, "learning_rate": 9.277170529217847e-06, "loss": 0.15136001110076905, "step": 20400 }, { "epoch": 0.5376305140219227, "grad_norm": 0.8208276033401489, "learning_rate": 9.268661737333691e-06, "loss": 0.14227131605148316, "step": 20410 }, { "epoch": 0.5378939292664214, "grad_norm": 1.8109486103057861, "learning_rate": 9.260153477749072e-06, "loss": 0.14330313205718995, "step": 20420 }, { "epoch": 0.5381573445109202, "grad_norm": 2.3747565746307373, "learning_rate": 9.251645756656676e-06, "loss": 0.15246498584747314, "step": 20430 }, { "epoch": 0.5384207597554189, "grad_norm": 2.2427854537963867, "learning_rate": 9.243138580248782e-06, "loss": 0.14186134338378906, "step": 20440 }, { "epoch": 0.5386841749999177, "grad_norm": 3.637981653213501, "learning_rate": 9.234631954717285e-06, "loss": 0.14912898540496827, "step": 20450 }, { "epoch": 0.5389475902444164, "grad_norm": 2.321655035018921, "learning_rate": 9.226125886253673e-06, "loss": 0.1508535623550415, "step": 20460 }, { "epoch": 0.5392110054889152, "grad_norm": 1.1580724716186523, "learning_rate": 9.217620381049026e-06, "loss": 0.14517029523849487, "step": 20470 }, { "epoch": 0.5394744207334139, "grad_norm": 0.9021688103675842, "learning_rate": 9.209115445294022e-06, "loss": 0.1418381690979004, "step": 20480 }, { "epoch": 0.5397378359779126, "grad_norm": 0.8355332612991333, "learning_rate": 9.200611085178917e-06, "loss": 0.13761613368988038, "step": 20490 }, { "epoch": 0.5400012512224114, "grad_norm": 1.602005124092102, "learning_rate": 9.192107306893556e-06, "loss": 0.14204310178756713, "step": 20500 }, { "epoch": 0.5402646664669101, "grad_norm": 4.430052280426025, "learning_rate": 9.183604116627354e-06, "loss": 0.15493441820144654, "step": 20510 }, { "epoch": 0.5405280817114089, "grad_norm": 1.5603951215744019, "learning_rate": 9.175101520569293e-06, "loss": 0.15164161920547486, "step": 20520 }, { "epoch": 0.5407914969559076, "grad_norm": 2.1534740924835205, "learning_rate": 9.166599524907937e-06, "loss": 0.1431918501853943, "step": 20530 }, { "epoch": 0.5410549122004064, "grad_norm": 2.5842092037200928, "learning_rate": 9.158098135831404e-06, "loss": 0.14997079372406005, "step": 20540 }, { "epoch": 0.541318327444905, "grad_norm": 1.9381574392318726, "learning_rate": 9.149597359527368e-06, "loss": 0.14114913940429688, "step": 20550 }, { "epoch": 0.5415817426894038, "grad_norm": 1.2551544904708862, "learning_rate": 9.141097202183063e-06, "loss": 0.14708762168884276, "step": 20560 }, { "epoch": 0.5418451579339025, "grad_norm": 1.9657739400863647, "learning_rate": 9.13259766998527e-06, "loss": 0.1416506290435791, "step": 20570 }, { "epoch": 0.5421085731784012, "grad_norm": 2.6803407669067383, "learning_rate": 9.124098769120318e-06, "loss": 0.14310951232910157, "step": 20580 }, { "epoch": 0.5423719884229, "grad_norm": 1.058471918106079, "learning_rate": 9.115600505774072e-06, "loss": 0.15263081789016725, "step": 20590 }, { "epoch": 0.5426354036673987, "grad_norm": 1.237715482711792, "learning_rate": 9.10710288613193e-06, "loss": 0.14779032468795777, "step": 20600 }, { "epoch": 0.5428988189118975, "grad_norm": 1.9577404260635376, "learning_rate": 9.098605916378837e-06, "loss": 0.14211034774780273, "step": 20610 }, { "epoch": 0.5431622341563962, "grad_norm": 2.280879259109497, "learning_rate": 9.090109602699239e-06, "loss": 0.13914272785186768, "step": 20620 }, { "epoch": 0.543425649400895, "grad_norm": 3.180820941925049, "learning_rate": 9.081613951277135e-06, "loss": 0.1433656930923462, "step": 20630 }, { "epoch": 0.5436890646453937, "grad_norm": 1.342261791229248, "learning_rate": 9.07311896829602e-06, "loss": 0.13887467384338378, "step": 20640 }, { "epoch": 0.5439524798898925, "grad_norm": 1.3423750400543213, "learning_rate": 9.064624659938902e-06, "loss": 0.1319188356399536, "step": 20650 }, { "epoch": 0.5442158951343912, "grad_norm": 1.624721646308899, "learning_rate": 9.056131032388318e-06, "loss": 0.15467072725296022, "step": 20660 }, { "epoch": 0.5444793103788899, "grad_norm": 3.0769150257110596, "learning_rate": 9.047638091826291e-06, "loss": 0.14455357789993287, "step": 20670 }, { "epoch": 0.5447427256233887, "grad_norm": 0.9229267835617065, "learning_rate": 9.039145844434344e-06, "loss": 0.15312985181808472, "step": 20680 }, { "epoch": 0.5450061408678873, "grad_norm": 2.2930448055267334, "learning_rate": 9.030654296393512e-06, "loss": 0.14086087942123413, "step": 20690 }, { "epoch": 0.5452695561123861, "grad_norm": 1.2460960149765015, "learning_rate": 9.022163453884299e-06, "loss": 0.13684014081954957, "step": 20700 }, { "epoch": 0.5455329713568848, "grad_norm": 0.8979966044425964, "learning_rate": 9.013673323086716e-06, "loss": 0.1445266842842102, "step": 20710 }, { "epoch": 0.5457963866013836, "grad_norm": 2.714168071746826, "learning_rate": 9.005183910180243e-06, "loss": 0.14360135793685913, "step": 20720 }, { "epoch": 0.5460598018458823, "grad_norm": 1.360198736190796, "learning_rate": 8.996695221343834e-06, "loss": 0.14389874935150146, "step": 20730 }, { "epoch": 0.546323217090381, "grad_norm": 1.6908087730407715, "learning_rate": 8.988207262755936e-06, "loss": 0.1402514696121216, "step": 20740 }, { "epoch": 0.5465866323348798, "grad_norm": 1.7164537906646729, "learning_rate": 8.979720040594438e-06, "loss": 0.15595004558563233, "step": 20750 }, { "epoch": 0.5468500475793785, "grad_norm": 3.058532238006592, "learning_rate": 8.971233561036718e-06, "loss": 0.1380335807800293, "step": 20760 }, { "epoch": 0.5471134628238773, "grad_norm": 1.5087753534317017, "learning_rate": 8.962747830259598e-06, "loss": 0.14112045764923095, "step": 20770 }, { "epoch": 0.547376878068376, "grad_norm": 1.1246548891067505, "learning_rate": 8.954262854439354e-06, "loss": 0.15368914604187012, "step": 20780 }, { "epoch": 0.5476402933128748, "grad_norm": 1.4366694688796997, "learning_rate": 8.945778639751723e-06, "loss": 0.1488026738166809, "step": 20790 }, { "epoch": 0.5479037085573735, "grad_norm": 3.1067862510681152, "learning_rate": 8.937295192371882e-06, "loss": 0.13751442432403566, "step": 20800 }, { "epoch": 0.5481671238018723, "grad_norm": 1.2620395421981812, "learning_rate": 8.928812518474445e-06, "loss": 0.12394924163818359, "step": 20810 }, { "epoch": 0.548430539046371, "grad_norm": 1.3588306903839111, "learning_rate": 8.920330624233477e-06, "loss": 0.12980788946151733, "step": 20820 }, { "epoch": 0.5486939542908696, "grad_norm": 1.439029574394226, "learning_rate": 8.911849515822457e-06, "loss": 0.1461361527442932, "step": 20830 }, { "epoch": 0.5489573695353684, "grad_norm": 1.1267110109329224, "learning_rate": 8.903369199414311e-06, "loss": 0.1460430860519409, "step": 20840 }, { "epoch": 0.5492207847798671, "grad_norm": 1.2704070806503296, "learning_rate": 8.894889681181374e-06, "loss": 0.14224691390991212, "step": 20850 }, { "epoch": 0.5494842000243659, "grad_norm": 3.4136037826538086, "learning_rate": 8.886410967295401e-06, "loss": 0.15671563148498535, "step": 20860 }, { "epoch": 0.5497476152688646, "grad_norm": 1.6519752740859985, "learning_rate": 8.877933063927576e-06, "loss": 0.12956888675689698, "step": 20870 }, { "epoch": 0.5500110305133634, "grad_norm": 1.0938944816589355, "learning_rate": 8.869455977248473e-06, "loss": 0.1424380898475647, "step": 20880 }, { "epoch": 0.5502744457578621, "grad_norm": 2.3729662895202637, "learning_rate": 8.860979713428085e-06, "loss": 0.13119087219238282, "step": 20890 }, { "epoch": 0.5505378610023609, "grad_norm": 2.383052110671997, "learning_rate": 8.852504278635806e-06, "loss": 0.14191830158233643, "step": 20900 }, { "epoch": 0.5508012762468596, "grad_norm": 2.3622348308563232, "learning_rate": 8.844029679040418e-06, "loss": 0.14018592834472657, "step": 20910 }, { "epoch": 0.5510646914913583, "grad_norm": 1.7247087955474854, "learning_rate": 8.835555920810104e-06, "loss": 0.13631442785263062, "step": 20920 }, { "epoch": 0.5513281067358571, "grad_norm": 1.377771019935608, "learning_rate": 8.827083010112429e-06, "loss": 0.12919967174530028, "step": 20930 }, { "epoch": 0.5515915219803558, "grad_norm": 1.0011316537857056, "learning_rate": 8.818610953114342e-06, "loss": 0.13001965284347533, "step": 20940 }, { "epoch": 0.5518549372248546, "grad_norm": 2.2174413204193115, "learning_rate": 8.810139755982173e-06, "loss": 0.1351548671722412, "step": 20950 }, { "epoch": 0.5521183524693533, "grad_norm": 1.7237423658370972, "learning_rate": 8.801669424881624e-06, "loss": 0.14243361949920655, "step": 20960 }, { "epoch": 0.5523817677138521, "grad_norm": 1.1471120119094849, "learning_rate": 8.79319996597777e-06, "loss": 0.13542890548706055, "step": 20970 }, { "epoch": 0.5526451829583507, "grad_norm": 2.892853021621704, "learning_rate": 8.784731385435049e-06, "loss": 0.14141008853912354, "step": 20980 }, { "epoch": 0.5529085982028495, "grad_norm": 1.0170912742614746, "learning_rate": 8.77626368941725e-06, "loss": 0.13097708225250243, "step": 20990 }, { "epoch": 0.5531720134473482, "grad_norm": 1.993123173713684, "learning_rate": 8.767796884087539e-06, "loss": 0.13784905672073364, "step": 21000 }, { "epoch": 0.5534354286918469, "grad_norm": 2.6566898822784424, "learning_rate": 8.759330975608418e-06, "loss": 0.14397815465927125, "step": 21010 }, { "epoch": 0.5536988439363457, "grad_norm": 1.1238187551498413, "learning_rate": 8.750865970141736e-06, "loss": 0.14017468690872192, "step": 21020 }, { "epoch": 0.5539622591808444, "grad_norm": 1.4254951477050781, "learning_rate": 8.742401873848696e-06, "loss": 0.14144866466522216, "step": 21030 }, { "epoch": 0.5542256744253432, "grad_norm": 1.8464434146881104, "learning_rate": 8.733938692889827e-06, "loss": 0.14245989322662353, "step": 21040 }, { "epoch": 0.5544890896698419, "grad_norm": 2.13127064704895, "learning_rate": 8.725476433425003e-06, "loss": 0.1456654667854309, "step": 21050 }, { "epoch": 0.5547525049143407, "grad_norm": 3.3321163654327393, "learning_rate": 8.717015101613417e-06, "loss": 0.15122706890106202, "step": 21060 }, { "epoch": 0.5550159201588394, "grad_norm": 1.094484806060791, "learning_rate": 8.708554703613592e-06, "loss": 0.13900630474090575, "step": 21070 }, { "epoch": 0.5552793354033382, "grad_norm": 2.1923232078552246, "learning_rate": 8.700095245583374e-06, "loss": 0.12591593265533446, "step": 21080 }, { "epoch": 0.5555427506478369, "grad_norm": 1.3190973997116089, "learning_rate": 8.691636733679915e-06, "loss": 0.13992612361907958, "step": 21090 }, { "epoch": 0.5558061658923356, "grad_norm": 1.7387052774429321, "learning_rate": 8.683179174059691e-06, "loss": 0.14403337240219116, "step": 21100 }, { "epoch": 0.5560695811368344, "grad_norm": 1.7909071445465088, "learning_rate": 8.67472257287848e-06, "loss": 0.14129881858825682, "step": 21110 }, { "epoch": 0.556332996381333, "grad_norm": 2.837090015411377, "learning_rate": 8.666266936291354e-06, "loss": 0.145210599899292, "step": 21120 }, { "epoch": 0.5565964116258318, "grad_norm": 2.1816370487213135, "learning_rate": 8.657812270452695e-06, "loss": 0.14532921314239503, "step": 21130 }, { "epoch": 0.5568598268703305, "grad_norm": 2.4059414863586426, "learning_rate": 8.649358581516171e-06, "loss": 0.14213002920150758, "step": 21140 }, { "epoch": 0.5571232421148293, "grad_norm": 0.9949086308479309, "learning_rate": 8.640905875634744e-06, "loss": 0.13966329097747804, "step": 21150 }, { "epoch": 0.557386657359328, "grad_norm": 2.011904716491699, "learning_rate": 8.632454158960656e-06, "loss": 0.13940563201904296, "step": 21160 }, { "epoch": 0.5576500726038268, "grad_norm": 1.5309300422668457, "learning_rate": 8.624003437645431e-06, "loss": 0.12388737201690674, "step": 21170 }, { "epoch": 0.5579134878483255, "grad_norm": 4.113550186157227, "learning_rate": 8.615553717839873e-06, "loss": 0.12901639938354492, "step": 21180 }, { "epoch": 0.5581769030928242, "grad_norm": 1.5472941398620605, "learning_rate": 8.607105005694048e-06, "loss": 0.15422927141189574, "step": 21190 }, { "epoch": 0.558440318337323, "grad_norm": 1.6891717910766602, "learning_rate": 8.598657307357293e-06, "loss": 0.14417650699615478, "step": 21200 }, { "epoch": 0.5587037335818217, "grad_norm": 1.1880971193313599, "learning_rate": 8.590210628978213e-06, "loss": 0.14086149930953978, "step": 21210 }, { "epoch": 0.5589671488263205, "grad_norm": 2.2772252559661865, "learning_rate": 8.581764976704653e-06, "loss": 0.14968719482421874, "step": 21220 }, { "epoch": 0.5592305640708192, "grad_norm": 1.6273295879364014, "learning_rate": 8.573320356683739e-06, "loss": 0.14099578857421874, "step": 21230 }, { "epoch": 0.559493979315318, "grad_norm": 1.1669639348983765, "learning_rate": 8.564876775061819e-06, "loss": 0.12837729454040528, "step": 21240 }, { "epoch": 0.5597573945598167, "grad_norm": 2.6393821239471436, "learning_rate": 8.556434237984496e-06, "loss": 0.12521579265594482, "step": 21250 }, { "epoch": 0.5600208098043153, "grad_norm": 3.615217447280884, "learning_rate": 8.547992751596617e-06, "loss": 0.14993617534637452, "step": 21260 }, { "epoch": 0.5602842250488141, "grad_norm": 1.5077860355377197, "learning_rate": 8.53955232204226e-06, "loss": 0.14530714750289916, "step": 21270 }, { "epoch": 0.5605476402933128, "grad_norm": 1.6386337280273438, "learning_rate": 8.531112955464728e-06, "loss": 0.14383423328399658, "step": 21280 }, { "epoch": 0.5608110555378116, "grad_norm": 0.8419686555862427, "learning_rate": 8.522674658006564e-06, "loss": 0.13411865234375, "step": 21290 }, { "epoch": 0.5610744707823103, "grad_norm": 1.4798076152801514, "learning_rate": 8.514237435809517e-06, "loss": 0.1382715106010437, "step": 21300 }, { "epoch": 0.5613378860268091, "grad_norm": 0.9280018210411072, "learning_rate": 8.505801295014569e-06, "loss": 0.1292569160461426, "step": 21310 }, { "epoch": 0.5616013012713078, "grad_norm": 2.205146074295044, "learning_rate": 8.497366241761907e-06, "loss": 0.14296185970306396, "step": 21320 }, { "epoch": 0.5618647165158066, "grad_norm": 0.8171665072441101, "learning_rate": 8.488932282190914e-06, "loss": 0.14292207956314087, "step": 21330 }, { "epoch": 0.5621281317603053, "grad_norm": 0.7899404764175415, "learning_rate": 8.480499422440207e-06, "loss": 0.13396944999694824, "step": 21340 }, { "epoch": 0.562391547004804, "grad_norm": 2.800156593322754, "learning_rate": 8.472067668647571e-06, "loss": 0.1405010223388672, "step": 21350 }, { "epoch": 0.5626549622493028, "grad_norm": 1.8301501274108887, "learning_rate": 8.463637026950013e-06, "loss": 0.14069957733154298, "step": 21360 }, { "epoch": 0.5629183774938015, "grad_norm": 1.8290470838546753, "learning_rate": 8.455207503483708e-06, "loss": 0.13610647916793822, "step": 21370 }, { "epoch": 0.5631817927383003, "grad_norm": 5.118052959442139, "learning_rate": 8.446779104384028e-06, "loss": 0.1490877389907837, "step": 21380 }, { "epoch": 0.563445207982799, "grad_norm": 2.1909055709838867, "learning_rate": 8.43835183578553e-06, "loss": 0.13349788188934325, "step": 21390 }, { "epoch": 0.5637086232272978, "grad_norm": 1.3672884702682495, "learning_rate": 8.429925703821941e-06, "loss": 0.14971342086791992, "step": 21400 }, { "epoch": 0.5639720384717964, "grad_norm": 2.2627880573272705, "learning_rate": 8.421500714626164e-06, "loss": 0.14399685859680175, "step": 21410 }, { "epoch": 0.5642354537162952, "grad_norm": 2.406735897064209, "learning_rate": 8.413076874330266e-06, "loss": 0.1443360209465027, "step": 21420 }, { "epoch": 0.5644988689607939, "grad_norm": 1.0025262832641602, "learning_rate": 8.404654189065486e-06, "loss": 0.13840789794921876, "step": 21430 }, { "epoch": 0.5647622842052926, "grad_norm": 1.0473756790161133, "learning_rate": 8.396232664962218e-06, "loss": 0.15714938640594484, "step": 21440 }, { "epoch": 0.5650256994497914, "grad_norm": 0.8870872259140015, "learning_rate": 8.387812308150011e-06, "loss": 0.1200251579284668, "step": 21450 }, { "epoch": 0.5652891146942901, "grad_norm": 1.4487545490264893, "learning_rate": 8.379393124757554e-06, "loss": 0.16189903020858765, "step": 21460 }, { "epoch": 0.5655525299387889, "grad_norm": 1.0473930835723877, "learning_rate": 8.370975120912708e-06, "loss": 0.14485563039779664, "step": 21470 }, { "epoch": 0.5658159451832876, "grad_norm": 3.835700035095215, "learning_rate": 8.362558302742447e-06, "loss": 0.1314859628677368, "step": 21480 }, { "epoch": 0.5660793604277864, "grad_norm": 3.778611183166504, "learning_rate": 8.354142676372898e-06, "loss": 0.14637525081634523, "step": 21490 }, { "epoch": 0.5663427756722851, "grad_norm": 0.958023726940155, "learning_rate": 8.345728247929319e-06, "loss": 0.1303019642829895, "step": 21500 }, { "epoch": 0.5666061909167839, "grad_norm": 1.705383062362671, "learning_rate": 8.33731502353609e-06, "loss": 0.14261702299118043, "step": 21510 }, { "epoch": 0.5668696061612826, "grad_norm": 2.760012626647949, "learning_rate": 8.328903009316724e-06, "loss": 0.16314454078674318, "step": 21520 }, { "epoch": 0.5671330214057813, "grad_norm": 2.1088380813598633, "learning_rate": 8.320492211393844e-06, "loss": 0.1393909454345703, "step": 21530 }, { "epoch": 0.5673964366502801, "grad_norm": 3.086470603942871, "learning_rate": 8.312082635889193e-06, "loss": 0.1524444580078125, "step": 21540 }, { "epoch": 0.5676598518947787, "grad_norm": 1.6324459314346313, "learning_rate": 8.303674288923621e-06, "loss": 0.14742389917373658, "step": 21550 }, { "epoch": 0.5679232671392775, "grad_norm": 1.788804531097412, "learning_rate": 8.29526717661709e-06, "loss": 0.14208016395568848, "step": 21560 }, { "epoch": 0.5681866823837762, "grad_norm": 1.835189938545227, "learning_rate": 8.286861305088655e-06, "loss": 0.1327332854270935, "step": 21570 }, { "epoch": 0.568450097628275, "grad_norm": 1.0260484218597412, "learning_rate": 8.278456680456476e-06, "loss": 0.13757569789886476, "step": 21580 }, { "epoch": 0.5687135128727737, "grad_norm": 1.3911114931106567, "learning_rate": 8.270053308837791e-06, "loss": 0.1323932886123657, "step": 21590 }, { "epoch": 0.5689769281172725, "grad_norm": 1.6180859804153442, "learning_rate": 8.261651196348953e-06, "loss": 0.13180105686187743, "step": 21600 }, { "epoch": 0.5692403433617712, "grad_norm": 2.1412601470947266, "learning_rate": 8.253250349105369e-06, "loss": 0.14413774013519287, "step": 21610 }, { "epoch": 0.5695037586062699, "grad_norm": 1.4858601093292236, "learning_rate": 8.24485077322154e-06, "loss": 0.14046573638916016, "step": 21620 }, { "epoch": 0.5697671738507687, "grad_norm": 1.1523431539535522, "learning_rate": 8.236452474811046e-06, "loss": 0.14215655326843263, "step": 21630 }, { "epoch": 0.5700305890952674, "grad_norm": 1.7300925254821777, "learning_rate": 8.228055459986521e-06, "loss": 0.14495772123336792, "step": 21640 }, { "epoch": 0.5702940043397662, "grad_norm": 2.425889492034912, "learning_rate": 8.219659734859686e-06, "loss": 0.12912871837615966, "step": 21650 }, { "epoch": 0.5705574195842649, "grad_norm": 2.2049291133880615, "learning_rate": 8.211265305541304e-06, "loss": 0.1409054160118103, "step": 21660 }, { "epoch": 0.5708208348287637, "grad_norm": 2.630811929702759, "learning_rate": 8.202872178141204e-06, "loss": 0.14219361543655396, "step": 21670 }, { "epoch": 0.5710842500732624, "grad_norm": 5.629793643951416, "learning_rate": 8.19448035876827e-06, "loss": 0.13419291973114014, "step": 21680 }, { "epoch": 0.5713476653177612, "grad_norm": 2.0087921619415283, "learning_rate": 8.186089853530426e-06, "loss": 0.14463146924972534, "step": 21690 }, { "epoch": 0.5716110805622598, "grad_norm": 4.82810640335083, "learning_rate": 8.177700668534646e-06, "loss": 0.14464871883392333, "step": 21700 }, { "epoch": 0.5718744958067585, "grad_norm": 1.0920850038528442, "learning_rate": 8.169312809886944e-06, "loss": 0.1431584358215332, "step": 21710 }, { "epoch": 0.5721379110512573, "grad_norm": 1.6596543788909912, "learning_rate": 8.160926283692355e-06, "loss": 0.1413326859474182, "step": 21720 }, { "epoch": 0.572401326295756, "grad_norm": 2.550588369369507, "learning_rate": 8.152541096054965e-06, "loss": 0.13455142974853515, "step": 21730 }, { "epoch": 0.5726647415402548, "grad_norm": 1.6948639154434204, "learning_rate": 8.144157253077871e-06, "loss": 0.14330716133117677, "step": 21740 }, { "epoch": 0.5729281567847535, "grad_norm": 1.4699515104293823, "learning_rate": 8.135774760863193e-06, "loss": 0.1466321587562561, "step": 21750 }, { "epoch": 0.5731915720292523, "grad_norm": 1.1117217540740967, "learning_rate": 8.127393625512077e-06, "loss": 0.13889552354812623, "step": 21760 }, { "epoch": 0.573454987273751, "grad_norm": 1.521295428276062, "learning_rate": 8.11901385312467e-06, "loss": 0.1419827461242676, "step": 21770 }, { "epoch": 0.5737184025182498, "grad_norm": 1.211470603942871, "learning_rate": 8.110635449800132e-06, "loss": 0.14538754224777223, "step": 21780 }, { "epoch": 0.5739818177627485, "grad_norm": 1.3555760383605957, "learning_rate": 8.10225842163663e-06, "loss": 0.14008935689926147, "step": 21790 }, { "epoch": 0.5742452330072472, "grad_norm": 1.3787256479263306, "learning_rate": 8.093882774731322e-06, "loss": 0.1563981294631958, "step": 21800 }, { "epoch": 0.574508648251746, "grad_norm": 1.9325231313705444, "learning_rate": 8.085508515180373e-06, "loss": 0.14591000080108643, "step": 21810 }, { "epoch": 0.5747720634962447, "grad_norm": 1.7621967792510986, "learning_rate": 8.077135649078917e-06, "loss": 0.1312650203704834, "step": 21820 }, { "epoch": 0.5750354787407435, "grad_norm": 2.963751792907715, "learning_rate": 8.068764182521102e-06, "loss": 0.15022177696228028, "step": 21830 }, { "epoch": 0.5752988939852421, "grad_norm": 1.6561936140060425, "learning_rate": 8.060394121600033e-06, "loss": 0.13827894926071166, "step": 21840 }, { "epoch": 0.5755623092297409, "grad_norm": 1.241660475730896, "learning_rate": 8.052025472407802e-06, "loss": 0.1345421552658081, "step": 21850 }, { "epoch": 0.5758257244742396, "grad_norm": 1.1229249238967896, "learning_rate": 8.043658241035477e-06, "loss": 0.1407262086868286, "step": 21860 }, { "epoch": 0.5760891397187383, "grad_norm": 1.4798916578292847, "learning_rate": 8.035292433573088e-06, "loss": 0.13003686666488648, "step": 21870 }, { "epoch": 0.5763525549632371, "grad_norm": 1.526739478111267, "learning_rate": 8.026928056109628e-06, "loss": 0.13589000701904297, "step": 21880 }, { "epoch": 0.5766159702077358, "grad_norm": 2.081207275390625, "learning_rate": 8.018565114733056e-06, "loss": 0.14820172786712646, "step": 21890 }, { "epoch": 0.5768793854522346, "grad_norm": 1.1792004108428955, "learning_rate": 8.010203615530279e-06, "loss": 0.1371913433074951, "step": 21900 }, { "epoch": 0.5771428006967333, "grad_norm": 2.9212396144866943, "learning_rate": 8.00184356458716e-06, "loss": 0.1425872802734375, "step": 21910 }, { "epoch": 0.5774062159412321, "grad_norm": 2.568932056427002, "learning_rate": 7.993484967988504e-06, "loss": 0.13770769834518432, "step": 21920 }, { "epoch": 0.5776696311857308, "grad_norm": 2.420943021774292, "learning_rate": 7.98512783181805e-06, "loss": 0.1275397300720215, "step": 21930 }, { "epoch": 0.5779330464302296, "grad_norm": 1.3351436853408813, "learning_rate": 7.976772162158498e-06, "loss": 0.15401036739349366, "step": 21940 }, { "epoch": 0.5781964616747283, "grad_norm": 1.5390398502349854, "learning_rate": 7.968417965091449e-06, "loss": 0.12342820167541504, "step": 21950 }, { "epoch": 0.578459876919227, "grad_norm": 3.4204390048980713, "learning_rate": 7.960065246697456e-06, "loss": 0.13869794607162475, "step": 21960 }, { "epoch": 0.5787232921637258, "grad_norm": 1.136920690536499, "learning_rate": 7.951714013055988e-06, "loss": 0.1474068880081177, "step": 21970 }, { "epoch": 0.5789867074082244, "grad_norm": 1.9033774137496948, "learning_rate": 7.94336427024543e-06, "loss": 0.14119110107421876, "step": 21980 }, { "epoch": 0.5792501226527232, "grad_norm": 1.554699420928955, "learning_rate": 7.93501602434309e-06, "loss": 0.1342616319656372, "step": 21990 }, { "epoch": 0.5795135378972219, "grad_norm": 1.5930439233779907, "learning_rate": 7.926669281425177e-06, "loss": 0.14565377235412597, "step": 22000 }, { "epoch": 0.5797769531417207, "grad_norm": 1.6731349229812622, "learning_rate": 7.91832404756681e-06, "loss": 0.1558019518852234, "step": 22010 }, { "epoch": 0.5800403683862194, "grad_norm": 2.727015733718872, "learning_rate": 7.909980328842015e-06, "loss": 0.13897349834442138, "step": 22020 }, { "epoch": 0.5803037836307182, "grad_norm": 1.5650964975357056, "learning_rate": 7.901638131323705e-06, "loss": 0.13345015048980713, "step": 22030 }, { "epoch": 0.5805671988752169, "grad_norm": 1.699539065361023, "learning_rate": 7.893297461083697e-06, "loss": 0.1441852331161499, "step": 22040 }, { "epoch": 0.5808306141197156, "grad_norm": 2.780862331390381, "learning_rate": 7.884958324192689e-06, "loss": 0.14753475189208984, "step": 22050 }, { "epoch": 0.5810940293642144, "grad_norm": 2.2061057090759277, "learning_rate": 7.876620726720257e-06, "loss": 0.15842347145080565, "step": 22060 }, { "epoch": 0.5813574446087131, "grad_norm": 1.6374690532684326, "learning_rate": 7.868284674734877e-06, "loss": 0.1463268756866455, "step": 22070 }, { "epoch": 0.5816208598532119, "grad_norm": 0.950644850730896, "learning_rate": 7.859950174303879e-06, "loss": 0.1451234221458435, "step": 22080 }, { "epoch": 0.5818842750977106, "grad_norm": 1.2962594032287598, "learning_rate": 7.851617231493472e-06, "loss": 0.14493000507354736, "step": 22090 }, { "epoch": 0.5821476903422094, "grad_norm": 2.6488351821899414, "learning_rate": 7.843285852368734e-06, "loss": 0.14036115407943725, "step": 22100 }, { "epoch": 0.5824111055867081, "grad_norm": 1.311676263809204, "learning_rate": 7.8349560429936e-06, "loss": 0.13850171566009523, "step": 22110 }, { "epoch": 0.5826745208312069, "grad_norm": 1.9006129503250122, "learning_rate": 7.826627809430867e-06, "loss": 0.14296238422393798, "step": 22120 }, { "epoch": 0.5829379360757055, "grad_norm": 1.6714361906051636, "learning_rate": 7.818301157742182e-06, "loss": 0.14118659496307373, "step": 22130 }, { "epoch": 0.5832013513202042, "grad_norm": 2.192225456237793, "learning_rate": 7.809976093988039e-06, "loss": 0.13998008966445924, "step": 22140 }, { "epoch": 0.583464766564703, "grad_norm": 2.148578405380249, "learning_rate": 7.801652624227782e-06, "loss": 0.147915518283844, "step": 22150 }, { "epoch": 0.5837281818092017, "grad_norm": 1.7261829376220703, "learning_rate": 7.79333075451959e-06, "loss": 0.12919468879699708, "step": 22160 }, { "epoch": 0.5839915970537005, "grad_norm": 1.79148530960083, "learning_rate": 7.785010490920477e-06, "loss": 0.13767354488372802, "step": 22170 }, { "epoch": 0.5842550122981992, "grad_norm": 3.87001371383667, "learning_rate": 7.776691839486297e-06, "loss": 0.12961785793304442, "step": 22180 }, { "epoch": 0.584518427542698, "grad_norm": 1.4971590042114258, "learning_rate": 7.76837480627171e-06, "loss": 0.14442846775054932, "step": 22190 }, { "epoch": 0.5847818427871967, "grad_norm": 3.404597043991089, "learning_rate": 7.760059397330225e-06, "loss": 0.13201324939727782, "step": 22200 }, { "epoch": 0.5850452580316955, "grad_norm": 2.4980311393737793, "learning_rate": 7.75174561871415e-06, "loss": 0.13886125087738038, "step": 22210 }, { "epoch": 0.5853086732761942, "grad_norm": 1.3482047319412231, "learning_rate": 7.743433476474608e-06, "loss": 0.1370757818222046, "step": 22220 }, { "epoch": 0.5855720885206929, "grad_norm": 3.3399910926818848, "learning_rate": 7.73512297666154e-06, "loss": 0.1325867533683777, "step": 22230 }, { "epoch": 0.5858355037651917, "grad_norm": 1.2977268695831299, "learning_rate": 7.726814125323683e-06, "loss": 0.1321321725845337, "step": 22240 }, { "epoch": 0.5860989190096904, "grad_norm": 3.0806515216827393, "learning_rate": 7.718506928508583e-06, "loss": 0.13827941417694092, "step": 22250 }, { "epoch": 0.5863623342541892, "grad_norm": 1.7497400045394897, "learning_rate": 7.71020139226257e-06, "loss": 0.12957170009613037, "step": 22260 }, { "epoch": 0.5866257494986878, "grad_norm": 1.3768906593322754, "learning_rate": 7.701897522630772e-06, "loss": 0.13224058151245116, "step": 22270 }, { "epoch": 0.5868891647431866, "grad_norm": 1.6821852922439575, "learning_rate": 7.693595325657109e-06, "loss": 0.13704986572265626, "step": 22280 }, { "epoch": 0.5871525799876853, "grad_norm": 1.0192627906799316, "learning_rate": 7.685294807384273e-06, "loss": 0.13459532260894774, "step": 22290 }, { "epoch": 0.5874159952321841, "grad_norm": 2.249652862548828, "learning_rate": 7.676995973853743e-06, "loss": 0.13428976535797119, "step": 22300 }, { "epoch": 0.5876794104766828, "grad_norm": 2.073554754257202, "learning_rate": 7.668698831105768e-06, "loss": 0.1284961223602295, "step": 22310 }, { "epoch": 0.5879428257211815, "grad_norm": 1.597286581993103, "learning_rate": 7.66040338517936e-06, "loss": 0.14986023902893067, "step": 22320 }, { "epoch": 0.5882062409656803, "grad_norm": 1.0614159107208252, "learning_rate": 7.652109642112308e-06, "loss": 0.13878920078277587, "step": 22330 }, { "epoch": 0.588469656210179, "grad_norm": 0.9910660982131958, "learning_rate": 7.643817607941157e-06, "loss": 0.12077209949493409, "step": 22340 }, { "epoch": 0.5887330714546778, "grad_norm": 1.3058289289474487, "learning_rate": 7.635527288701201e-06, "loss": 0.13027286529541016, "step": 22350 }, { "epoch": 0.5889964866991765, "grad_norm": 1.8829057216644287, "learning_rate": 7.6272386904265005e-06, "loss": 0.13670729398727416, "step": 22360 }, { "epoch": 0.5892599019436753, "grad_norm": 2.1761209964752197, "learning_rate": 7.618951819149848e-06, "loss": 0.13363440036773683, "step": 22370 }, { "epoch": 0.589523317188174, "grad_norm": 1.0063061714172363, "learning_rate": 7.610666680902791e-06, "loss": 0.11584563255310058, "step": 22380 }, { "epoch": 0.5897867324326727, "grad_norm": 1.534286618232727, "learning_rate": 7.602383281715608e-06, "loss": 0.12634413242340087, "step": 22390 }, { "epoch": 0.5900501476771715, "grad_norm": 2.930469512939453, "learning_rate": 7.594101627617312e-06, "loss": 0.13313002586364747, "step": 22400 }, { "epoch": 0.5903135629216701, "grad_norm": 3.6169049739837646, "learning_rate": 7.585821724635656e-06, "loss": 0.1373399615287781, "step": 22410 }, { "epoch": 0.5905769781661689, "grad_norm": 2.414440155029297, "learning_rate": 7.577543578797097e-06, "loss": 0.1367746114730835, "step": 22420 }, { "epoch": 0.5908403934106676, "grad_norm": 4.573324203491211, "learning_rate": 7.569267196126844e-06, "loss": 0.1311353325843811, "step": 22430 }, { "epoch": 0.5911038086551664, "grad_norm": 1.5402500629425049, "learning_rate": 7.560992582648793e-06, "loss": 0.15258944034576416, "step": 22440 }, { "epoch": 0.5913672238996651, "grad_norm": 0.858974277973175, "learning_rate": 7.552719744385566e-06, "loss": 0.13918917179107665, "step": 22450 }, { "epoch": 0.5916306391441639, "grad_norm": 1.2477006912231445, "learning_rate": 7.544448687358495e-06, "loss": 0.14019771814346313, "step": 22460 }, { "epoch": 0.5918940543886626, "grad_norm": 1.2389476299285889, "learning_rate": 7.536179417587612e-06, "loss": 0.14187148809432984, "step": 22470 }, { "epoch": 0.5921574696331613, "grad_norm": 1.895434021949768, "learning_rate": 7.527911941091644e-06, "loss": 0.1411879301071167, "step": 22480 }, { "epoch": 0.5924208848776601, "grad_norm": 2.2023749351501465, "learning_rate": 7.519646263888023e-06, "loss": 0.1446061611175537, "step": 22490 }, { "epoch": 0.5926843001221588, "grad_norm": 1.789746642112732, "learning_rate": 7.511382391992861e-06, "loss": 0.13483229875564576, "step": 22500 }, { "epoch": 0.5929477153666576, "grad_norm": 2.2878494262695312, "learning_rate": 7.503120331420963e-06, "loss": 0.1433103084564209, "step": 22510 }, { "epoch": 0.5932111306111563, "grad_norm": 1.1830567121505737, "learning_rate": 7.494860088185815e-06, "loss": 0.1381995677947998, "step": 22520 }, { "epoch": 0.5934745458556551, "grad_norm": 0.9790918827056885, "learning_rate": 7.48660166829957e-06, "loss": 0.1348963499069214, "step": 22530 }, { "epoch": 0.5937379611001538, "grad_norm": 2.0508525371551514, "learning_rate": 7.478345077773075e-06, "loss": 0.1363231897354126, "step": 22540 }, { "epoch": 0.5940013763446526, "grad_norm": 2.046990156173706, "learning_rate": 7.470090322615822e-06, "loss": 0.14585022926330565, "step": 22550 }, { "epoch": 0.5942647915891512, "grad_norm": 3.2263994216918945, "learning_rate": 7.4618374088359805e-06, "loss": 0.14340560436248778, "step": 22560 }, { "epoch": 0.5945282068336499, "grad_norm": 0.9822491407394409, "learning_rate": 7.45358634244038e-06, "loss": 0.131090247631073, "step": 22570 }, { "epoch": 0.5947916220781487, "grad_norm": 2.8987815380096436, "learning_rate": 7.4453371294345e-06, "loss": 0.13334078788757325, "step": 22580 }, { "epoch": 0.5950550373226474, "grad_norm": 1.2362428903579712, "learning_rate": 7.437089775822477e-06, "loss": 0.1523667097091675, "step": 22590 }, { "epoch": 0.5953184525671462, "grad_norm": 3.5906929969787598, "learning_rate": 7.4288442876070865e-06, "loss": 0.13981733322143555, "step": 22600 }, { "epoch": 0.5955818678116449, "grad_norm": 2.0427067279815674, "learning_rate": 7.420600670789753e-06, "loss": 0.14434040784835817, "step": 22610 }, { "epoch": 0.5958452830561437, "grad_norm": 1.8771203756332397, "learning_rate": 7.412358931370536e-06, "loss": 0.13730986118316652, "step": 22620 }, { "epoch": 0.5961086983006424, "grad_norm": 0.9514104723930359, "learning_rate": 7.404119075348129e-06, "loss": 0.13273355960845948, "step": 22630 }, { "epoch": 0.5963721135451412, "grad_norm": 1.1654376983642578, "learning_rate": 7.3958811087198555e-06, "loss": 0.14052276611328124, "step": 22640 }, { "epoch": 0.5966355287896399, "grad_norm": 1.6257917881011963, "learning_rate": 7.3876450374816644e-06, "loss": 0.12315497398376465, "step": 22650 }, { "epoch": 0.5968989440341386, "grad_norm": 2.2458884716033936, "learning_rate": 7.379410867628115e-06, "loss": 0.13571574687957763, "step": 22660 }, { "epoch": 0.5971623592786374, "grad_norm": 1.3376939296722412, "learning_rate": 7.371178605152405e-06, "loss": 0.13569440841674804, "step": 22670 }, { "epoch": 0.597425774523136, "grad_norm": 1.1231297254562378, "learning_rate": 7.362948256046321e-06, "loss": 0.12393174171447754, "step": 22680 }, { "epoch": 0.5976891897676349, "grad_norm": 1.6613476276397705, "learning_rate": 7.354719826300266e-06, "loss": 0.14171457290649414, "step": 22690 }, { "epoch": 0.5979526050121335, "grad_norm": 4.026061534881592, "learning_rate": 7.346493321903251e-06, "loss": 0.1322157859802246, "step": 22700 }, { "epoch": 0.5982160202566323, "grad_norm": 2.120121717453003, "learning_rate": 7.338268748842876e-06, "loss": 0.13787941932678222, "step": 22710 }, { "epoch": 0.598479435501131, "grad_norm": 4.237461090087891, "learning_rate": 7.330046113105345e-06, "loss": 0.14905152320861817, "step": 22720 }, { "epoch": 0.5987428507456298, "grad_norm": 1.5908935070037842, "learning_rate": 7.321825420675444e-06, "loss": 0.12358345985412597, "step": 22730 }, { "epoch": 0.5990062659901285, "grad_norm": 1.5797275304794312, "learning_rate": 7.3136066775365455e-06, "loss": 0.12466170787811279, "step": 22740 }, { "epoch": 0.5992696812346272, "grad_norm": 2.2988929748535156, "learning_rate": 7.305389889670611e-06, "loss": 0.12248709201812744, "step": 22750 }, { "epoch": 0.599533096479126, "grad_norm": 2.516937017440796, "learning_rate": 7.2971750630581675e-06, "loss": 0.14668291807174683, "step": 22760 }, { "epoch": 0.5997965117236247, "grad_norm": 1.7522269487380981, "learning_rate": 7.2889622036783245e-06, "loss": 0.13255295753479004, "step": 22770 }, { "epoch": 0.6000599269681235, "grad_norm": 1.4491451978683472, "learning_rate": 7.280751317508756e-06, "loss": 0.1354041576385498, "step": 22780 }, { "epoch": 0.6003233422126222, "grad_norm": 2.1678884029388428, "learning_rate": 7.272542410525691e-06, "loss": 0.13485198020935057, "step": 22790 }, { "epoch": 0.600586757457121, "grad_norm": 3.710333824157715, "learning_rate": 7.2643354887039395e-06, "loss": 0.144669246673584, "step": 22800 }, { "epoch": 0.6008501727016197, "grad_norm": 2.4648778438568115, "learning_rate": 7.256130558016843e-06, "loss": 0.12967554330825806, "step": 22810 }, { "epoch": 0.6011135879461185, "grad_norm": 1.7802573442459106, "learning_rate": 7.247927624436304e-06, "loss": 0.13927714824676513, "step": 22820 }, { "epoch": 0.6013770031906172, "grad_norm": 3.068798780441284, "learning_rate": 7.239726693932779e-06, "loss": 0.13999388217926026, "step": 22830 }, { "epoch": 0.6016404184351158, "grad_norm": 1.1658803224563599, "learning_rate": 7.2315277724752495e-06, "loss": 0.11987183094024659, "step": 22840 }, { "epoch": 0.6019038336796146, "grad_norm": 2.324791669845581, "learning_rate": 7.223330866031253e-06, "loss": 0.14086025953292847, "step": 22850 }, { "epoch": 0.6021672489241133, "grad_norm": 1.6821473836898804, "learning_rate": 7.215135980566847e-06, "loss": 0.12997573614120483, "step": 22860 }, { "epoch": 0.6024306641686121, "grad_norm": 2.603598117828369, "learning_rate": 7.2069431220466215e-06, "loss": 0.1340895414352417, "step": 22870 }, { "epoch": 0.6026940794131108, "grad_norm": 2.871666193008423, "learning_rate": 7.198752296433697e-06, "loss": 0.13459556102752684, "step": 22880 }, { "epoch": 0.6029574946576096, "grad_norm": 1.4650061130523682, "learning_rate": 7.190563509689705e-06, "loss": 0.13156460523605346, "step": 22890 }, { "epoch": 0.6032209099021083, "grad_norm": 2.3220767974853516, "learning_rate": 7.182376767774803e-06, "loss": 0.12977499961853028, "step": 22900 }, { "epoch": 0.603484325146607, "grad_norm": 2.105167865753174, "learning_rate": 7.174192076647653e-06, "loss": 0.14097063541412352, "step": 22910 }, { "epoch": 0.6037477403911058, "grad_norm": 0.8079142570495605, "learning_rate": 7.166009442265423e-06, "loss": 0.1239861249923706, "step": 22920 }, { "epoch": 0.6040111556356045, "grad_norm": 1.7965773344039917, "learning_rate": 7.15782887058379e-06, "loss": 0.12257696390151977, "step": 22930 }, { "epoch": 0.6042745708801033, "grad_norm": 2.3872604370117188, "learning_rate": 7.149650367556926e-06, "loss": 0.12773282527923585, "step": 22940 }, { "epoch": 0.604537986124602, "grad_norm": 1.5343528985977173, "learning_rate": 7.141473939137498e-06, "loss": 0.12927286624908446, "step": 22950 }, { "epoch": 0.6048014013691008, "grad_norm": 2.414018154144287, "learning_rate": 7.133299591276665e-06, "loss": 0.13427507877349854, "step": 22960 }, { "epoch": 0.6050648166135995, "grad_norm": 4.630444526672363, "learning_rate": 7.12512732992407e-06, "loss": 0.13147883415222167, "step": 22970 }, { "epoch": 0.6053282318580983, "grad_norm": 2.23887300491333, "learning_rate": 7.1169571610278375e-06, "loss": 0.14559919834136964, "step": 22980 }, { "epoch": 0.6055916471025969, "grad_norm": 1.5469024181365967, "learning_rate": 7.108789090534569e-06, "loss": 0.12787739038467408, "step": 22990 }, { "epoch": 0.6058550623470956, "grad_norm": 1.1178998947143555, "learning_rate": 7.100623124389334e-06, "loss": 0.12682387828826905, "step": 23000 }, { "epoch": 0.6061184775915944, "grad_norm": 1.7025002241134644, "learning_rate": 7.092459268535685e-06, "loss": 0.13139946460723878, "step": 23010 }, { "epoch": 0.6063818928360931, "grad_norm": 2.821355104446411, "learning_rate": 7.084297528915616e-06, "loss": 0.14725086688995362, "step": 23020 }, { "epoch": 0.6066453080805919, "grad_norm": 1.172680377960205, "learning_rate": 7.076137911469603e-06, "loss": 0.1284945487976074, "step": 23030 }, { "epoch": 0.6069087233250906, "grad_norm": 1.9856228828430176, "learning_rate": 7.067980422136565e-06, "loss": 0.13298280239105226, "step": 23040 }, { "epoch": 0.6071721385695894, "grad_norm": 3.352492094039917, "learning_rate": 7.059825066853872e-06, "loss": 0.1416351318359375, "step": 23050 }, { "epoch": 0.6074355538140881, "grad_norm": 1.3905404806137085, "learning_rate": 7.0516718515573445e-06, "loss": 0.13087364435195922, "step": 23060 }, { "epoch": 0.6076989690585869, "grad_norm": 1.2077624797821045, "learning_rate": 7.043520782181248e-06, "loss": 0.13078646659851073, "step": 23070 }, { "epoch": 0.6079623843030856, "grad_norm": 1.0831563472747803, "learning_rate": 7.035371864658276e-06, "loss": 0.14396519660949708, "step": 23080 }, { "epoch": 0.6082257995475843, "grad_norm": 2.369316577911377, "learning_rate": 7.0272251049195684e-06, "loss": 0.1289849042892456, "step": 23090 }, { "epoch": 0.6084892147920831, "grad_norm": 1.8382493257522583, "learning_rate": 7.019080508894683e-06, "loss": 0.13921401500701905, "step": 23100 }, { "epoch": 0.6087526300365818, "grad_norm": 0.9461648464202881, "learning_rate": 7.010938082511614e-06, "loss": 0.1363748788833618, "step": 23110 }, { "epoch": 0.6090160452810806, "grad_norm": 0.9670825600624084, "learning_rate": 7.0027978316967705e-06, "loss": 0.125679349899292, "step": 23120 }, { "epoch": 0.6092794605255792, "grad_norm": 1.8922902345657349, "learning_rate": 6.994659762374967e-06, "loss": 0.1324836015701294, "step": 23130 }, { "epoch": 0.609542875770078, "grad_norm": 2.000466823577881, "learning_rate": 6.986523880469458e-06, "loss": 0.14015238285064696, "step": 23140 }, { "epoch": 0.6098062910145767, "grad_norm": 1.4189661741256714, "learning_rate": 6.978390191901878e-06, "loss": 0.15495144128799437, "step": 23150 }, { "epoch": 0.6100697062590755, "grad_norm": 2.563649892807007, "learning_rate": 6.970258702592277e-06, "loss": 0.1426100492477417, "step": 23160 }, { "epoch": 0.6103331215035742, "grad_norm": 1.5808608531951904, "learning_rate": 6.962129418459109e-06, "loss": 0.13244004249572755, "step": 23170 }, { "epoch": 0.6105965367480729, "grad_norm": 1.276367425918579, "learning_rate": 6.954002345419214e-06, "loss": 0.13020572662353516, "step": 23180 }, { "epoch": 0.6108599519925717, "grad_norm": 2.2723937034606934, "learning_rate": 6.945877489387829e-06, "loss": 0.14894230365753175, "step": 23190 }, { "epoch": 0.6111233672370704, "grad_norm": 1.2301757335662842, "learning_rate": 6.9377548562785735e-06, "loss": 0.1416693925857544, "step": 23200 }, { "epoch": 0.6113867824815692, "grad_norm": 2.8689942359924316, "learning_rate": 6.929634452003447e-06, "loss": 0.13422183990478515, "step": 23210 }, { "epoch": 0.6116501977260679, "grad_norm": 2.4845428466796875, "learning_rate": 6.921516282472837e-06, "loss": 0.14816060066223144, "step": 23220 }, { "epoch": 0.6119136129705667, "grad_norm": 2.3896265029907227, "learning_rate": 6.91340035359549e-06, "loss": 0.14569108486175536, "step": 23230 }, { "epoch": 0.6121770282150654, "grad_norm": 1.5747030973434448, "learning_rate": 6.905286671278536e-06, "loss": 0.14343247413635254, "step": 23240 }, { "epoch": 0.6124404434595642, "grad_norm": 2.829904794692993, "learning_rate": 6.897175241427462e-06, "loss": 0.13220865726470948, "step": 23250 }, { "epoch": 0.6127038587040629, "grad_norm": 1.0545165538787842, "learning_rate": 6.889066069946107e-06, "loss": 0.13172160387039183, "step": 23260 }, { "epoch": 0.6129672739485615, "grad_norm": 1.290607213973999, "learning_rate": 6.880959162736691e-06, "loss": 0.14032938480377197, "step": 23270 }, { "epoch": 0.6132306891930603, "grad_norm": 2.5138304233551025, "learning_rate": 6.872854525699761e-06, "loss": 0.13513410091400146, "step": 23280 }, { "epoch": 0.613494104437559, "grad_norm": 2.0763134956359863, "learning_rate": 6.8647521647342205e-06, "loss": 0.12751662731170654, "step": 23290 }, { "epoch": 0.6137575196820578, "grad_norm": 1.8853834867477417, "learning_rate": 6.856652085737323e-06, "loss": 0.13823988437652587, "step": 23300 }, { "epoch": 0.6140209349265565, "grad_norm": 1.8864843845367432, "learning_rate": 6.848554294604652e-06, "loss": 0.12451525926589965, "step": 23310 }, { "epoch": 0.6142843501710553, "grad_norm": 1.1605418920516968, "learning_rate": 6.8404587972301316e-06, "loss": 0.12951352596282958, "step": 23320 }, { "epoch": 0.614547765415554, "grad_norm": 1.5637263059616089, "learning_rate": 6.8323655995060135e-06, "loss": 0.14809350967407225, "step": 23330 }, { "epoch": 0.6148111806600528, "grad_norm": 1.1943714618682861, "learning_rate": 6.824274707322874e-06, "loss": 0.13782985210418702, "step": 23340 }, { "epoch": 0.6150745959045515, "grad_norm": 3.1508052349090576, "learning_rate": 6.816186126569615e-06, "loss": 0.1430235505104065, "step": 23350 }, { "epoch": 0.6153380111490502, "grad_norm": 4.0190935134887695, "learning_rate": 6.808099863133453e-06, "loss": 0.14463436603546143, "step": 23360 }, { "epoch": 0.615601426393549, "grad_norm": 1.0300381183624268, "learning_rate": 6.800015922899923e-06, "loss": 0.13787814378738403, "step": 23370 }, { "epoch": 0.6158648416380477, "grad_norm": 1.5893959999084473, "learning_rate": 6.7919343117528655e-06, "loss": 0.14035075902938843, "step": 23380 }, { "epoch": 0.6161282568825465, "grad_norm": 3.168975591659546, "learning_rate": 6.783855035574418e-06, "loss": 0.1364384651184082, "step": 23390 }, { "epoch": 0.6163916721270452, "grad_norm": 1.8436557054519653, "learning_rate": 6.775778100245038e-06, "loss": 0.13986263275146485, "step": 23400 }, { "epoch": 0.616655087371544, "grad_norm": 2.8265841007232666, "learning_rate": 6.76770351164346e-06, "loss": 0.13474037647247314, "step": 23410 }, { "epoch": 0.6169185026160426, "grad_norm": 3.43096661567688, "learning_rate": 6.759631275646715e-06, "loss": 0.1312613606452942, "step": 23420 }, { "epoch": 0.6171819178605413, "grad_norm": 3.0895912647247314, "learning_rate": 6.751561398130133e-06, "loss": 0.13860127925872803, "step": 23430 }, { "epoch": 0.6174453331050401, "grad_norm": 1.9143036603927612, "learning_rate": 6.743493884967312e-06, "loss": 0.13529399633407593, "step": 23440 }, { "epoch": 0.6177087483495388, "grad_norm": 2.5672409534454346, "learning_rate": 6.735428742030141e-06, "loss": 0.12235558032989502, "step": 23450 }, { "epoch": 0.6179721635940376, "grad_norm": 1.7213928699493408, "learning_rate": 6.7273659751887775e-06, "loss": 0.13784817457199097, "step": 23460 }, { "epoch": 0.6182355788385363, "grad_norm": 1.876082420349121, "learning_rate": 6.719305590311649e-06, "loss": 0.13205044269561766, "step": 23470 }, { "epoch": 0.6184989940830351, "grad_norm": 1.7731503248214722, "learning_rate": 6.7112475932654554e-06, "loss": 0.13602204322814943, "step": 23480 }, { "epoch": 0.6187624093275338, "grad_norm": 2.405726432800293, "learning_rate": 6.70319198991515e-06, "loss": 0.13570882081985475, "step": 23490 }, { "epoch": 0.6190258245720326, "grad_norm": 2.759077787399292, "learning_rate": 6.695138786123952e-06, "loss": 0.12707853317260742, "step": 23500 }, { "epoch": 0.6192892398165313, "grad_norm": 1.6895772218704224, "learning_rate": 6.687087987753331e-06, "loss": 0.13015023469924927, "step": 23510 }, { "epoch": 0.61955265506103, "grad_norm": 2.662630558013916, "learning_rate": 6.679039600663001e-06, "loss": 0.1481245279312134, "step": 23520 }, { "epoch": 0.6198160703055288, "grad_norm": 1.2388712167739868, "learning_rate": 6.670993630710929e-06, "loss": 0.13234293460845947, "step": 23530 }, { "epoch": 0.6200794855500275, "grad_norm": 2.277662515640259, "learning_rate": 6.662950083753319e-06, "loss": 0.12094216346740723, "step": 23540 }, { "epoch": 0.6203429007945263, "grad_norm": 1.9058583974838257, "learning_rate": 6.654908965644609e-06, "loss": 0.14349801540374757, "step": 23550 }, { "epoch": 0.6206063160390249, "grad_norm": 3.001951217651367, "learning_rate": 6.646870282237474e-06, "loss": 0.12705061435699463, "step": 23560 }, { "epoch": 0.6208697312835237, "grad_norm": 1.3940820693969727, "learning_rate": 6.6388340393828125e-06, "loss": 0.14450104236602784, "step": 23570 }, { "epoch": 0.6211331465280224, "grad_norm": 2.7504141330718994, "learning_rate": 6.6308002429297516e-06, "loss": 0.12904806137084962, "step": 23580 }, { "epoch": 0.6213965617725212, "grad_norm": 1.2429258823394775, "learning_rate": 6.622768898725634e-06, "loss": 0.11702160835266114, "step": 23590 }, { "epoch": 0.6216599770170199, "grad_norm": 1.9391961097717285, "learning_rate": 6.614740012616017e-06, "loss": 0.13650977611541748, "step": 23600 }, { "epoch": 0.6219233922615186, "grad_norm": 3.2737348079681396, "learning_rate": 6.606713590444673e-06, "loss": 0.13055102825164794, "step": 23610 }, { "epoch": 0.6221868075060174, "grad_norm": 2.118440866470337, "learning_rate": 6.598689638053577e-06, "loss": 0.13303968906402588, "step": 23620 }, { "epoch": 0.6224502227505161, "grad_norm": 1.831742525100708, "learning_rate": 6.590668161282903e-06, "loss": 0.13425095081329347, "step": 23630 }, { "epoch": 0.6227136379950149, "grad_norm": 2.7118890285491943, "learning_rate": 6.582649165971037e-06, "loss": 0.13576512336730956, "step": 23640 }, { "epoch": 0.6229770532395136, "grad_norm": 1.5774015188217163, "learning_rate": 6.574632657954542e-06, "loss": 0.12376420497894287, "step": 23650 }, { "epoch": 0.6232404684840124, "grad_norm": 1.852129578590393, "learning_rate": 6.5666186430681835e-06, "loss": 0.12744662761688233, "step": 23660 }, { "epoch": 0.6235038837285111, "grad_norm": 1.8748012781143188, "learning_rate": 6.558607127144905e-06, "loss": 0.12209327220916748, "step": 23670 }, { "epoch": 0.6237672989730099, "grad_norm": 2.4895706176757812, "learning_rate": 6.550598116015832e-06, "loss": 0.12632498741149903, "step": 23680 }, { "epoch": 0.6240307142175086, "grad_norm": 4.458941459655762, "learning_rate": 6.54259161551027e-06, "loss": 0.1362409234046936, "step": 23690 }, { "epoch": 0.6242941294620072, "grad_norm": 1.473846197128296, "learning_rate": 6.534587631455696e-06, "loss": 0.13577524423599244, "step": 23700 }, { "epoch": 0.624557544706506, "grad_norm": 1.2624366283416748, "learning_rate": 6.5265861696777556e-06, "loss": 0.1461920142173767, "step": 23710 }, { "epoch": 0.6248209599510047, "grad_norm": 3.334437847137451, "learning_rate": 6.518587236000259e-06, "loss": 0.1325647473335266, "step": 23720 }, { "epoch": 0.6250843751955035, "grad_norm": 2.8460960388183594, "learning_rate": 6.510590836245166e-06, "loss": 0.12550268173217774, "step": 23730 }, { "epoch": 0.6253477904400022, "grad_norm": 1.8391358852386475, "learning_rate": 6.502596976232616e-06, "loss": 0.1323185682296753, "step": 23740 }, { "epoch": 0.625611205684501, "grad_norm": 2.047132730484009, "learning_rate": 6.494605661780874e-06, "loss": 0.1419266939163208, "step": 23750 }, { "epoch": 0.6258746209289997, "grad_norm": 1.3492168188095093, "learning_rate": 6.486616898706365e-06, "loss": 0.13181718587875366, "step": 23760 }, { "epoch": 0.6261380361734985, "grad_norm": 1.2479420900344849, "learning_rate": 6.47863069282366e-06, "loss": 0.13389257192611695, "step": 23770 }, { "epoch": 0.6264014514179972, "grad_norm": 2.0096030235290527, "learning_rate": 6.47064704994546e-06, "loss": 0.1291405200958252, "step": 23780 }, { "epoch": 0.6266648666624959, "grad_norm": 1.3413392305374146, "learning_rate": 6.462665975882607e-06, "loss": 0.13203966617584229, "step": 23790 }, { "epoch": 0.6269282819069947, "grad_norm": 2.916969060897827, "learning_rate": 6.454687476444071e-06, "loss": 0.1329214096069336, "step": 23800 }, { "epoch": 0.6271916971514934, "grad_norm": 1.8002668619155884, "learning_rate": 6.446711557436944e-06, "loss": 0.11846352815628051, "step": 23810 }, { "epoch": 0.6274551123959922, "grad_norm": 2.221982479095459, "learning_rate": 6.438738224666452e-06, "loss": 0.13323776721954345, "step": 23820 }, { "epoch": 0.6277185276404909, "grad_norm": 2.204660654067993, "learning_rate": 6.4307674839359225e-06, "loss": 0.1325763463973999, "step": 23830 }, { "epoch": 0.6279819428849897, "grad_norm": 2.7456281185150146, "learning_rate": 6.42279934104681e-06, "loss": 0.1334507942199707, "step": 23840 }, { "epoch": 0.6282453581294883, "grad_norm": 2.2178750038146973, "learning_rate": 6.4148338017986725e-06, "loss": 0.12899750471115112, "step": 23850 }, { "epoch": 0.6285087733739871, "grad_norm": 1.7169125080108643, "learning_rate": 6.406870871989166e-06, "loss": 0.11753368377685547, "step": 23860 }, { "epoch": 0.6287721886184858, "grad_norm": 2.1884055137634277, "learning_rate": 6.398910557414064e-06, "loss": 0.1362016201019287, "step": 23870 }, { "epoch": 0.6290356038629845, "grad_norm": 1.9754135608673096, "learning_rate": 6.390952863867224e-06, "loss": 0.12125439643859863, "step": 23880 }, { "epoch": 0.6292990191074833, "grad_norm": 3.046356201171875, "learning_rate": 6.382997797140591e-06, "loss": 0.14156651496887207, "step": 23890 }, { "epoch": 0.629562434351982, "grad_norm": 1.2974845170974731, "learning_rate": 6.375045363024217e-06, "loss": 0.1304154872894287, "step": 23900 }, { "epoch": 0.6298258495964808, "grad_norm": 3.2356581687927246, "learning_rate": 6.367095567306219e-06, "loss": 0.14443690776824952, "step": 23910 }, { "epoch": 0.6300892648409795, "grad_norm": 1.4256112575531006, "learning_rate": 6.359148415772805e-06, "loss": 0.11655707359313965, "step": 23920 }, { "epoch": 0.6303526800854783, "grad_norm": 2.6943106651306152, "learning_rate": 6.351203914208254e-06, "loss": 0.14458651542663575, "step": 23930 }, { "epoch": 0.630616095329977, "grad_norm": 1.2466981410980225, "learning_rate": 6.343262068394914e-06, "loss": 0.13440930843353271, "step": 23940 }, { "epoch": 0.6308795105744757, "grad_norm": 2.5053553581237793, "learning_rate": 6.335322884113208e-06, "loss": 0.1446087598800659, "step": 23950 }, { "epoch": 0.6311429258189745, "grad_norm": 2.871288299560547, "learning_rate": 6.3273863671416146e-06, "loss": 0.13300762176513672, "step": 23960 }, { "epoch": 0.6314063410634732, "grad_norm": 3.7825300693511963, "learning_rate": 6.319452523256673e-06, "loss": 0.1257880449295044, "step": 23970 }, { "epoch": 0.631669756307972, "grad_norm": 1.1576205492019653, "learning_rate": 6.31152135823298e-06, "loss": 0.13199980258941652, "step": 23980 }, { "epoch": 0.6319331715524706, "grad_norm": 1.5031914710998535, "learning_rate": 6.303592877843171e-06, "loss": 0.12867307662963867, "step": 23990 }, { "epoch": 0.6321965867969694, "grad_norm": 1.4035063982009888, "learning_rate": 6.295667087857948e-06, "loss": 0.12704999446868898, "step": 24000 }, { "epoch": 0.6324600020414681, "grad_norm": 3.36336350440979, "learning_rate": 6.287743994046035e-06, "loss": 0.12667622566223144, "step": 24010 }, { "epoch": 0.6327234172859669, "grad_norm": 4.668959140777588, "learning_rate": 6.279823602174202e-06, "loss": 0.1293959140777588, "step": 24020 }, { "epoch": 0.6329868325304656, "grad_norm": 3.534213066101074, "learning_rate": 6.271905918007255e-06, "loss": 0.14599183797836304, "step": 24030 }, { "epoch": 0.6332502477749643, "grad_norm": 2.475264549255371, "learning_rate": 6.263990947308022e-06, "loss": 0.13525724411010742, "step": 24040 }, { "epoch": 0.6335136630194631, "grad_norm": 1.0795565843582153, "learning_rate": 6.2560786958373645e-06, "loss": 0.13955485820770264, "step": 24050 }, { "epoch": 0.6337770782639618, "grad_norm": 1.3888095617294312, "learning_rate": 6.248169169354159e-06, "loss": 0.13811089992523193, "step": 24060 }, { "epoch": 0.6340404935084606, "grad_norm": 1.476430058479309, "learning_rate": 6.240262373615298e-06, "loss": 0.12814874649047853, "step": 24070 }, { "epoch": 0.6343039087529593, "grad_norm": 1.8520818948745728, "learning_rate": 6.232358314375693e-06, "loss": 0.1193742036819458, "step": 24080 }, { "epoch": 0.6345673239974581, "grad_norm": 1.7299988269805908, "learning_rate": 6.224456997388254e-06, "loss": 0.12077257633209229, "step": 24090 }, { "epoch": 0.6348307392419568, "grad_norm": 2.436185598373413, "learning_rate": 6.2165584284039046e-06, "loss": 0.12589282989501954, "step": 24100 }, { "epoch": 0.6350941544864556, "grad_norm": 2.8875949382781982, "learning_rate": 6.208662613171566e-06, "loss": 0.13161444664001465, "step": 24110 }, { "epoch": 0.6353575697309543, "grad_norm": 3.7011072635650635, "learning_rate": 6.200769557438144e-06, "loss": 0.13527169227600097, "step": 24120 }, { "epoch": 0.6356209849754529, "grad_norm": 1.1932460069656372, "learning_rate": 6.192879266948554e-06, "loss": 0.13035385608673095, "step": 24130 }, { "epoch": 0.6358844002199517, "grad_norm": 1.515629768371582, "learning_rate": 6.184991747445686e-06, "loss": 0.1363903522491455, "step": 24140 }, { "epoch": 0.6361478154644504, "grad_norm": 2.8399572372436523, "learning_rate": 6.1771070046704165e-06, "loss": 0.12374341487884521, "step": 24150 }, { "epoch": 0.6364112307089492, "grad_norm": 1.3435333967208862, "learning_rate": 6.169225044361604e-06, "loss": 0.13301016092300416, "step": 24160 }, { "epoch": 0.6366746459534479, "grad_norm": 4.186432361602783, "learning_rate": 6.161345872256075e-06, "loss": 0.13631868362426758, "step": 24170 }, { "epoch": 0.6369380611979467, "grad_norm": 1.8563870191574097, "learning_rate": 6.153469494088638e-06, "loss": 0.13152701854705812, "step": 24180 }, { "epoch": 0.6372014764424454, "grad_norm": 1.8636128902435303, "learning_rate": 6.1455959155920555e-06, "loss": 0.12127397060394288, "step": 24190 }, { "epoch": 0.6374648916869442, "grad_norm": 2.586728811264038, "learning_rate": 6.1377251424970555e-06, "loss": 0.12789585590362548, "step": 24200 }, { "epoch": 0.6377283069314429, "grad_norm": 2.675607442855835, "learning_rate": 6.1298571805323344e-06, "loss": 0.1265615463256836, "step": 24210 }, { "epoch": 0.6379917221759416, "grad_norm": 2.2133636474609375, "learning_rate": 6.121992035424529e-06, "loss": 0.12195169925689697, "step": 24220 }, { "epoch": 0.6382551374204404, "grad_norm": 1.6622318029403687, "learning_rate": 6.114129712898227e-06, "loss": 0.13259224891662597, "step": 24230 }, { "epoch": 0.6385185526649391, "grad_norm": 1.6061275005340576, "learning_rate": 6.106270218675975e-06, "loss": 0.1266319990158081, "step": 24240 }, { "epoch": 0.6387819679094379, "grad_norm": 2.2115859985351562, "learning_rate": 6.098413558478244e-06, "loss": 0.13731143474578858, "step": 24250 }, { "epoch": 0.6390453831539366, "grad_norm": 1.1311208009719849, "learning_rate": 6.090559738023456e-06, "loss": 0.1263245463371277, "step": 24260 }, { "epoch": 0.6393087983984354, "grad_norm": 1.8082777261734009, "learning_rate": 6.0827087630279605e-06, "loss": 0.13360313177108765, "step": 24270 }, { "epoch": 0.639572213642934, "grad_norm": 1.9897253513336182, "learning_rate": 6.074860639206028e-06, "loss": 0.12649919986724853, "step": 24280 }, { "epoch": 0.6398356288874328, "grad_norm": 1.1828176975250244, "learning_rate": 6.067015372269872e-06, "loss": 0.1275392770767212, "step": 24290 }, { "epoch": 0.6400990441319315, "grad_norm": 1.9297324419021606, "learning_rate": 6.059172967929609e-06, "loss": 0.12473292350769043, "step": 24300 }, { "epoch": 0.6403624593764302, "grad_norm": 1.4828157424926758, "learning_rate": 6.0513334318932825e-06, "loss": 0.12543715238571168, "step": 24310 }, { "epoch": 0.640625874620929, "grad_norm": 1.5018893480300903, "learning_rate": 6.043496769866848e-06, "loss": 0.13583608865737914, "step": 24320 }, { "epoch": 0.6408892898654277, "grad_norm": 1.295203447341919, "learning_rate": 6.035662987554154e-06, "loss": 0.1269433856010437, "step": 24330 }, { "epoch": 0.6411527051099265, "grad_norm": 2.287276268005371, "learning_rate": 6.027832090656982e-06, "loss": 0.1387803077697754, "step": 24340 }, { "epoch": 0.6414161203544252, "grad_norm": 1.398120403289795, "learning_rate": 6.020004084874984e-06, "loss": 0.13742258548736572, "step": 24350 }, { "epoch": 0.641679535598924, "grad_norm": 1.5785531997680664, "learning_rate": 6.012178975905725e-06, "loss": 0.13109240531921387, "step": 24360 }, { "epoch": 0.6419429508434227, "grad_norm": 3.0412886142730713, "learning_rate": 6.0043567694446566e-06, "loss": 0.11363770961761474, "step": 24370 }, { "epoch": 0.6422063660879215, "grad_norm": 2.6021878719329834, "learning_rate": 5.9965374711851175e-06, "loss": 0.12438855171203614, "step": 24380 }, { "epoch": 0.6424697813324202, "grad_norm": 1.1255004405975342, "learning_rate": 5.9887210868183325e-06, "loss": 0.13263280391693116, "step": 24390 }, { "epoch": 0.6427331965769189, "grad_norm": 1.2732558250427246, "learning_rate": 5.9809076220334024e-06, "loss": 0.1306656002998352, "step": 24400 }, { "epoch": 0.6429966118214177, "grad_norm": 1.9269022941589355, "learning_rate": 5.973097082517301e-06, "loss": 0.13384666442871093, "step": 24410 }, { "epoch": 0.6432600270659163, "grad_norm": 3.9663805961608887, "learning_rate": 5.965289473954879e-06, "loss": 0.12034479379653931, "step": 24420 }, { "epoch": 0.6435234423104151, "grad_norm": 2.7171058654785156, "learning_rate": 5.95748480202885e-06, "loss": 0.1357027292251587, "step": 24430 }, { "epoch": 0.6437868575549138, "grad_norm": 2.0196712017059326, "learning_rate": 5.949683072419792e-06, "loss": 0.12374637126922608, "step": 24440 }, { "epoch": 0.6440502727994126, "grad_norm": 2.6394429206848145, "learning_rate": 5.941884290806142e-06, "loss": 0.12326560020446778, "step": 24450 }, { "epoch": 0.6443136880439113, "grad_norm": 1.815643072128296, "learning_rate": 5.934088462864181e-06, "loss": 0.13400188684463502, "step": 24460 }, { "epoch": 0.6445771032884101, "grad_norm": 1.918445348739624, "learning_rate": 5.926295594268059e-06, "loss": 0.1312909960746765, "step": 24470 }, { "epoch": 0.6448405185329088, "grad_norm": 2.1660428047180176, "learning_rate": 5.918505690689759e-06, "loss": 0.13104074001312255, "step": 24480 }, { "epoch": 0.6451039337774075, "grad_norm": 2.6585075855255127, "learning_rate": 5.910718757799105e-06, "loss": 0.14611992835998536, "step": 24490 }, { "epoch": 0.6453673490219063, "grad_norm": 1.8190327882766724, "learning_rate": 5.902934801263766e-06, "loss": 0.14131195545196534, "step": 24500 }, { "epoch": 0.645630764266405, "grad_norm": 1.4273743629455566, "learning_rate": 5.89515382674924e-06, "loss": 0.12134603261947632, "step": 24510 }, { "epoch": 0.6458941795109038, "grad_norm": 2.2270443439483643, "learning_rate": 5.88737583991886e-06, "loss": 0.15272622108459472, "step": 24520 }, { "epoch": 0.6461575947554025, "grad_norm": 2.4026405811309814, "learning_rate": 5.879600846433777e-06, "loss": 0.1243971347808838, "step": 24530 }, { "epoch": 0.6464210099999013, "grad_norm": 2.229891061782837, "learning_rate": 5.871828851952965e-06, "loss": 0.13997454643249513, "step": 24540 }, { "epoch": 0.6466844252444, "grad_norm": 3.061748504638672, "learning_rate": 5.864059862133221e-06, "loss": 0.12894428968429567, "step": 24550 }, { "epoch": 0.6469478404888986, "grad_norm": 1.2691320180892944, "learning_rate": 5.856293882629147e-06, "loss": 0.12037899494171142, "step": 24560 }, { "epoch": 0.6472112557333974, "grad_norm": 1.6637197732925415, "learning_rate": 5.848530919093162e-06, "loss": 0.1346564769744873, "step": 24570 }, { "epoch": 0.6474746709778961, "grad_norm": 1.7493239641189575, "learning_rate": 5.840770977175485e-06, "loss": 0.12371289730072021, "step": 24580 }, { "epoch": 0.6477380862223949, "grad_norm": 2.1690022945404053, "learning_rate": 5.833014062524129e-06, "loss": 0.12209631204605102, "step": 24590 }, { "epoch": 0.6480015014668936, "grad_norm": 1.8094267845153809, "learning_rate": 5.825260180784923e-06, "loss": 0.12565549612045288, "step": 24600 }, { "epoch": 0.6482649167113924, "grad_norm": 2.50099515914917, "learning_rate": 5.817509337601475e-06, "loss": 0.1293623447418213, "step": 24610 }, { "epoch": 0.6485283319558911, "grad_norm": 2.0741772651672363, "learning_rate": 5.8097615386151686e-06, "loss": 0.1299901485443115, "step": 24620 }, { "epoch": 0.6487917472003899, "grad_norm": 2.5859313011169434, "learning_rate": 5.8020167894652035e-06, "loss": 0.12662713527679442, "step": 24630 }, { "epoch": 0.6490551624448886, "grad_norm": 2.252793550491333, "learning_rate": 5.79427509578853e-06, "loss": 0.12805483341217042, "step": 24640 }, { "epoch": 0.6493185776893873, "grad_norm": 2.338266611099243, "learning_rate": 5.786536463219899e-06, "loss": 0.1346510887145996, "step": 24650 }, { "epoch": 0.6495819929338861, "grad_norm": 1.7052805423736572, "learning_rate": 5.778800897391812e-06, "loss": 0.12310402393341065, "step": 24660 }, { "epoch": 0.6498454081783848, "grad_norm": 1.4115674495697021, "learning_rate": 5.771068403934545e-06, "loss": 0.1322182297706604, "step": 24670 }, { "epoch": 0.6501088234228836, "grad_norm": 2.148054599761963, "learning_rate": 5.76333898847615e-06, "loss": 0.12241861820220948, "step": 24680 }, { "epoch": 0.6503722386673823, "grad_norm": 1.7170543670654297, "learning_rate": 5.755612656642424e-06, "loss": 0.12502392530441284, "step": 24690 }, { "epoch": 0.650635653911881, "grad_norm": 3.3651328086853027, "learning_rate": 5.747889414056925e-06, "loss": 0.12756978273391723, "step": 24700 }, { "epoch": 0.6508990691563797, "grad_norm": 2.6391940116882324, "learning_rate": 5.740169266340963e-06, "loss": 0.12413372993469238, "step": 24710 }, { "epoch": 0.6511624844008785, "grad_norm": 1.3925915956497192, "learning_rate": 5.732452219113589e-06, "loss": 0.12853007316589354, "step": 24720 }, { "epoch": 0.6514258996453772, "grad_norm": 3.4046177864074707, "learning_rate": 5.724738277991611e-06, "loss": 0.12906875610351562, "step": 24730 }, { "epoch": 0.6516893148898759, "grad_norm": 2.1789259910583496, "learning_rate": 5.717027448589568e-06, "loss": 0.13873298168182374, "step": 24740 }, { "epoch": 0.6519527301343747, "grad_norm": 1.3314611911773682, "learning_rate": 5.709319736519725e-06, "loss": 0.1216543436050415, "step": 24750 }, { "epoch": 0.6522161453788734, "grad_norm": 1.4093897342681885, "learning_rate": 5.701615147392096e-06, "loss": 0.1387470006942749, "step": 24760 }, { "epoch": 0.6524795606233722, "grad_norm": 1.4078052043914795, "learning_rate": 5.6939136868144075e-06, "loss": 0.13369007110595704, "step": 24770 }, { "epoch": 0.6527429758678709, "grad_norm": 1.420745849609375, "learning_rate": 5.686215360392127e-06, "loss": 0.12231886386871338, "step": 24780 }, { "epoch": 0.6530063911123697, "grad_norm": 5.363104820251465, "learning_rate": 5.678520173728414e-06, "loss": 0.13323988914489746, "step": 24790 }, { "epoch": 0.6532698063568684, "grad_norm": 2.5593738555908203, "learning_rate": 5.67082813242416e-06, "loss": 0.12884156703948973, "step": 24800 }, { "epoch": 0.6535332216013672, "grad_norm": 1.0341764688491821, "learning_rate": 5.663139242077971e-06, "loss": 0.1309589385986328, "step": 24810 }, { "epoch": 0.6537966368458659, "grad_norm": 1.1500743627548218, "learning_rate": 5.6554535082861496e-06, "loss": 0.14330631494522095, "step": 24820 }, { "epoch": 0.6540600520903646, "grad_norm": 2.6035220623016357, "learning_rate": 5.647770936642705e-06, "loss": 0.12483539581298828, "step": 24830 }, { "epoch": 0.6543234673348634, "grad_norm": 1.7983334064483643, "learning_rate": 5.6400915327393435e-06, "loss": 0.12975044250488282, "step": 24840 }, { "epoch": 0.654586882579362, "grad_norm": 1.0971519947052002, "learning_rate": 5.632415302165464e-06, "loss": 0.12629423141479493, "step": 24850 }, { "epoch": 0.6548502978238608, "grad_norm": 1.9360947608947754, "learning_rate": 5.624742250508164e-06, "loss": 0.12459230422973633, "step": 24860 }, { "epoch": 0.6551137130683595, "grad_norm": 1.334364891052246, "learning_rate": 5.617072383352222e-06, "loss": 0.12909622192382814, "step": 24870 }, { "epoch": 0.6553771283128583, "grad_norm": 1.488755226135254, "learning_rate": 5.6094057062800886e-06, "loss": 0.1189306139945984, "step": 24880 }, { "epoch": 0.655640543557357, "grad_norm": 2.079637289047241, "learning_rate": 5.601742224871911e-06, "loss": 0.13290121555328369, "step": 24890 }, { "epoch": 0.6559039588018558, "grad_norm": 3.435131788253784, "learning_rate": 5.594081944705496e-06, "loss": 0.13324885368347167, "step": 24900 }, { "epoch": 0.6561673740463545, "grad_norm": 1.8128068447113037, "learning_rate": 5.586424871356337e-06, "loss": 0.12551865577697754, "step": 24910 }, { "epoch": 0.6564307892908532, "grad_norm": 1.9172890186309814, "learning_rate": 5.578771010397572e-06, "loss": 0.1347477674484253, "step": 24920 }, { "epoch": 0.656694204535352, "grad_norm": 2.321289300918579, "learning_rate": 5.571120367400008e-06, "loss": 0.12844407558441162, "step": 24930 }, { "epoch": 0.6569576197798507, "grad_norm": 3.3005616664886475, "learning_rate": 5.563472947932123e-06, "loss": 0.12128390073776245, "step": 24940 }, { "epoch": 0.6572210350243495, "grad_norm": 1.863400936126709, "learning_rate": 5.555828757560035e-06, "loss": 0.13360559940338135, "step": 24950 }, { "epoch": 0.6574844502688482, "grad_norm": 1.4204208850860596, "learning_rate": 5.548187801847513e-06, "loss": 0.13534178733825683, "step": 24960 }, { "epoch": 0.657747865513347, "grad_norm": 1.2100952863693237, "learning_rate": 5.5405500863559795e-06, "loss": 0.13554644584655762, "step": 24970 }, { "epoch": 0.6580112807578457, "grad_norm": 2.001147508621216, "learning_rate": 5.532915616644484e-06, "loss": 0.13112908601760864, "step": 24980 }, { "epoch": 0.6582746960023445, "grad_norm": 1.4218076467514038, "learning_rate": 5.525284398269733e-06, "loss": 0.11928914785385132, "step": 24990 }, { "epoch": 0.6585381112468431, "grad_norm": 1.7926222085952759, "learning_rate": 5.517656436786059e-06, "loss": 0.12551331520080566, "step": 25000 }, { "epoch": 0.6588015264913418, "grad_norm": 3.3717331886291504, "learning_rate": 5.510031737745405e-06, "loss": 0.12922897338867187, "step": 25010 }, { "epoch": 0.6590649417358406, "grad_norm": 2.262995958328247, "learning_rate": 5.50241030669737e-06, "loss": 0.1287023901939392, "step": 25020 }, { "epoch": 0.6593283569803393, "grad_norm": 1.3557202816009521, "learning_rate": 5.4947921491891565e-06, "loss": 0.1360940456390381, "step": 25030 }, { "epoch": 0.6595917722248381, "grad_norm": 2.12546443939209, "learning_rate": 5.487177270765588e-06, "loss": 0.12858593463897705, "step": 25040 }, { "epoch": 0.6598551874693368, "grad_norm": 1.8420772552490234, "learning_rate": 5.479565676969101e-06, "loss": 0.12280924320220947, "step": 25050 }, { "epoch": 0.6601186027138356, "grad_norm": 1.164612889289856, "learning_rate": 5.47195737333974e-06, "loss": 0.11869657039642334, "step": 25060 }, { "epoch": 0.6603820179583343, "grad_norm": 1.1183053255081177, "learning_rate": 5.464352365415162e-06, "loss": 0.1270078182220459, "step": 25070 }, { "epoch": 0.660645433202833, "grad_norm": 1.3102543354034424, "learning_rate": 5.456750658730616e-06, "loss": 0.13282610177993776, "step": 25080 }, { "epoch": 0.6609088484473318, "grad_norm": 1.3557251691818237, "learning_rate": 5.449152258818955e-06, "loss": 0.12374564409255981, "step": 25090 }, { "epoch": 0.6611722636918305, "grad_norm": 3.077103614807129, "learning_rate": 5.4415571712106184e-06, "loss": 0.12795616388320924, "step": 25100 }, { "epoch": 0.6614356789363293, "grad_norm": 2.5132172107696533, "learning_rate": 5.433965401433639e-06, "loss": 0.11833770275115967, "step": 25110 }, { "epoch": 0.661699094180828, "grad_norm": 1.2876734733581543, "learning_rate": 5.426376955013641e-06, "loss": 0.12727409601211548, "step": 25120 }, { "epoch": 0.6619625094253268, "grad_norm": 2.124220609664917, "learning_rate": 5.4187918374738154e-06, "loss": 0.13897523880004883, "step": 25130 }, { "epoch": 0.6622259246698254, "grad_norm": 1.4167486429214478, "learning_rate": 5.4112100543349365e-06, "loss": 0.12000253200531005, "step": 25140 }, { "epoch": 0.6624893399143242, "grad_norm": 2.885946035385132, "learning_rate": 5.403631611115362e-06, "loss": 0.1324261784553528, "step": 25150 }, { "epoch": 0.6627527551588229, "grad_norm": 2.179004669189453, "learning_rate": 5.396056513331005e-06, "loss": 0.12380422353744507, "step": 25160 }, { "epoch": 0.6630161704033216, "grad_norm": 1.7757296562194824, "learning_rate": 5.388484766495348e-06, "loss": 0.14127066135406494, "step": 25170 }, { "epoch": 0.6632795856478204, "grad_norm": 2.2996819019317627, "learning_rate": 5.3809163761194364e-06, "loss": 0.13404903411865235, "step": 25180 }, { "epoch": 0.6635430008923191, "grad_norm": 1.6885713338851929, "learning_rate": 5.373351347711865e-06, "loss": 0.11358157396316529, "step": 25190 }, { "epoch": 0.6638064161368179, "grad_norm": 2.3002102375030518, "learning_rate": 5.365789686778797e-06, "loss": 0.13842036724090576, "step": 25200 }, { "epoch": 0.6640698313813166, "grad_norm": 1.7647242546081543, "learning_rate": 5.358231398823937e-06, "loss": 0.12279163599014283, "step": 25210 }, { "epoch": 0.6643332466258154, "grad_norm": 2.5548830032348633, "learning_rate": 5.350676489348521e-06, "loss": 0.12439333200454712, "step": 25220 }, { "epoch": 0.6645966618703141, "grad_norm": 2.2618844509124756, "learning_rate": 5.3431249638513475e-06, "loss": 0.1397274374961853, "step": 25230 }, { "epoch": 0.6648600771148129, "grad_norm": 1.3898181915283203, "learning_rate": 5.335576827828737e-06, "loss": 0.12702977657318115, "step": 25240 }, { "epoch": 0.6651234923593116, "grad_norm": 2.4428248405456543, "learning_rate": 5.328032086774557e-06, "loss": 0.12072359323501587, "step": 25250 }, { "epoch": 0.6653869076038103, "grad_norm": 2.6777660846710205, "learning_rate": 5.320490746180189e-06, "loss": 0.118163800239563, "step": 25260 }, { "epoch": 0.665650322848309, "grad_norm": 2.113028049468994, "learning_rate": 5.312952811534542e-06, "loss": 0.12437589168548584, "step": 25270 }, { "epoch": 0.6659137380928077, "grad_norm": 1.7207542657852173, "learning_rate": 5.305418288324058e-06, "loss": 0.12449285984039307, "step": 25280 }, { "epoch": 0.6661771533373065, "grad_norm": 1.6646180152893066, "learning_rate": 5.297887182032686e-06, "loss": 0.12708749771118164, "step": 25290 }, { "epoch": 0.6664405685818052, "grad_norm": 1.8938424587249756, "learning_rate": 5.29035949814189e-06, "loss": 0.12001335620880127, "step": 25300 }, { "epoch": 0.666703983826304, "grad_norm": 2.333214521408081, "learning_rate": 5.2828352421306415e-06, "loss": 0.12783668041229249, "step": 25310 }, { "epoch": 0.6669673990708027, "grad_norm": 1.6135573387145996, "learning_rate": 5.275314419475417e-06, "loss": 0.11744865179061889, "step": 25320 }, { "epoch": 0.6672308143153015, "grad_norm": 3.0127480030059814, "learning_rate": 5.267797035650202e-06, "loss": 0.11971291303634643, "step": 25330 }, { "epoch": 0.6674942295598002, "grad_norm": 1.3597122430801392, "learning_rate": 5.2602830961264704e-06, "loss": 0.12995266914367676, "step": 25340 }, { "epoch": 0.6677576448042989, "grad_norm": 2.277433156967163, "learning_rate": 5.252772606373186e-06, "loss": 0.12917823791503907, "step": 25350 }, { "epoch": 0.6680210600487977, "grad_norm": 2.340583562850952, "learning_rate": 5.245265571856812e-06, "loss": 0.11424592733383179, "step": 25360 }, { "epoch": 0.6682844752932964, "grad_norm": 2.3738512992858887, "learning_rate": 5.237761998041291e-06, "loss": 0.13570637702941896, "step": 25370 }, { "epoch": 0.6685478905377952, "grad_norm": 1.1130973100662231, "learning_rate": 5.230261890388053e-06, "loss": 0.1361257553100586, "step": 25380 }, { "epoch": 0.6688113057822939, "grad_norm": 1.5573962926864624, "learning_rate": 5.2227652543559924e-06, "loss": 0.1215883731842041, "step": 25390 }, { "epoch": 0.6690747210267927, "grad_norm": 2.24992036819458, "learning_rate": 5.215272095401483e-06, "loss": 0.1260154962539673, "step": 25400 }, { "epoch": 0.6693381362712914, "grad_norm": 1.3708999156951904, "learning_rate": 5.207782418978378e-06, "loss": 0.12521425485610962, "step": 25410 }, { "epoch": 0.6696015515157901, "grad_norm": 2.0388057231903076, "learning_rate": 5.200296230537982e-06, "loss": 0.1195115327835083, "step": 25420 }, { "epoch": 0.6698649667602888, "grad_norm": 1.953464388847351, "learning_rate": 5.192813535529066e-06, "loss": 0.1312800407409668, "step": 25430 }, { "epoch": 0.6701283820047875, "grad_norm": 1.3678486347198486, "learning_rate": 5.1853343393978585e-06, "loss": 0.11994372606277466, "step": 25440 }, { "epoch": 0.6703917972492863, "grad_norm": 1.8966997861862183, "learning_rate": 5.177858647588037e-06, "loss": 0.13026702404022217, "step": 25450 }, { "epoch": 0.670655212493785, "grad_norm": 2.107496976852417, "learning_rate": 5.170386465540741e-06, "loss": 0.116668963432312, "step": 25460 }, { "epoch": 0.6709186277382838, "grad_norm": 1.8592610359191895, "learning_rate": 5.162917798694546e-06, "loss": 0.12550971508026124, "step": 25470 }, { "epoch": 0.6711820429827825, "grad_norm": 1.40725576877594, "learning_rate": 5.15545265248546e-06, "loss": 0.13045843839645385, "step": 25480 }, { "epoch": 0.6714454582272813, "grad_norm": 1.9739255905151367, "learning_rate": 5.147991032346948e-06, "loss": 0.11674325466156006, "step": 25490 }, { "epoch": 0.67170887347178, "grad_norm": 2.1533682346343994, "learning_rate": 5.140532943709893e-06, "loss": 0.13376028537750245, "step": 25500 }, { "epoch": 0.6719722887162788, "grad_norm": 1.396370530128479, "learning_rate": 5.133078392002623e-06, "loss": 0.12856731414794922, "step": 25510 }, { "epoch": 0.6722357039607775, "grad_norm": 2.433199882507324, "learning_rate": 5.125627382650873e-06, "loss": 0.11936683654785156, "step": 25520 }, { "epoch": 0.6724991192052762, "grad_norm": 3.365872621536255, "learning_rate": 5.118179921077808e-06, "loss": 0.13974249362945557, "step": 25530 }, { "epoch": 0.672762534449775, "grad_norm": 1.9295216798782349, "learning_rate": 5.1107360127040206e-06, "loss": 0.13060702085494996, "step": 25540 }, { "epoch": 0.6730259496942737, "grad_norm": 2.1406619548797607, "learning_rate": 5.103295662947504e-06, "loss": 0.12547557353973388, "step": 25550 }, { "epoch": 0.6732893649387724, "grad_norm": 3.103738784790039, "learning_rate": 5.095858877223665e-06, "loss": 0.1225476861000061, "step": 25560 }, { "epoch": 0.6735527801832711, "grad_norm": 1.633568525314331, "learning_rate": 5.088425660945319e-06, "loss": 0.12336063385009766, "step": 25570 }, { "epoch": 0.6738161954277699, "grad_norm": 1.7945574522018433, "learning_rate": 5.080996019522675e-06, "loss": 0.11496803760528565, "step": 25580 }, { "epoch": 0.6740796106722686, "grad_norm": 4.669789791107178, "learning_rate": 5.073569958363356e-06, "loss": 0.12803106307983397, "step": 25590 }, { "epoch": 0.6743430259167673, "grad_norm": 1.5114728212356567, "learning_rate": 5.06614748287237e-06, "loss": 0.11154940128326415, "step": 25600 }, { "epoch": 0.6746064411612661, "grad_norm": 1.8887349367141724, "learning_rate": 5.058728598452103e-06, "loss": 0.13002958297729492, "step": 25610 }, { "epoch": 0.6748698564057648, "grad_norm": 2.1505300998687744, "learning_rate": 5.05131331050235e-06, "loss": 0.11464660167694092, "step": 25620 }, { "epoch": 0.6751332716502636, "grad_norm": 1.7485392093658447, "learning_rate": 5.043901624420273e-06, "loss": 0.12827643156051635, "step": 25630 }, { "epoch": 0.6753966868947623, "grad_norm": 2.812822103500366, "learning_rate": 5.036493545600421e-06, "loss": 0.11719310283660889, "step": 25640 }, { "epoch": 0.6756601021392611, "grad_norm": 1.1916080713272095, "learning_rate": 5.029089079434708e-06, "loss": 0.12002074718475342, "step": 25650 }, { "epoch": 0.6759235173837598, "grad_norm": 1.843074083328247, "learning_rate": 5.021688231312423e-06, "loss": 0.1131251573562622, "step": 25660 }, { "epoch": 0.6761869326282586, "grad_norm": 1.5080571174621582, "learning_rate": 5.014291006620229e-06, "loss": 0.12474110126495361, "step": 25670 }, { "epoch": 0.6764503478727573, "grad_norm": 1.267004370689392, "learning_rate": 5.006897410742142e-06, "loss": 0.12168059349060059, "step": 25680 }, { "epoch": 0.676713763117256, "grad_norm": 1.3784109354019165, "learning_rate": 4.999507449059542e-06, "loss": 0.12375047206878662, "step": 25690 }, { "epoch": 0.6769771783617548, "grad_norm": 1.4666917324066162, "learning_rate": 4.992121126951159e-06, "loss": 0.13297535181045533, "step": 25700 }, { "epoch": 0.6772405936062534, "grad_norm": 1.8315850496292114, "learning_rate": 4.984738449793074e-06, "loss": 0.12736388444900512, "step": 25710 }, { "epoch": 0.6775040088507522, "grad_norm": 1.4517521858215332, "learning_rate": 4.977359422958731e-06, "loss": 0.12676565647125243, "step": 25720 }, { "epoch": 0.6777674240952509, "grad_norm": 1.4164433479309082, "learning_rate": 4.96998405181889e-06, "loss": 0.1316526412963867, "step": 25730 }, { "epoch": 0.6780308393397497, "grad_norm": 1.8903478384017944, "learning_rate": 4.962612341741666e-06, "loss": 0.11299426555633545, "step": 25740 }, { "epoch": 0.6782942545842484, "grad_norm": 1.527698278427124, "learning_rate": 4.955244298092514e-06, "loss": 0.12033612728118896, "step": 25750 }, { "epoch": 0.6785576698287472, "grad_norm": 1.6021922826766968, "learning_rate": 4.947879926234212e-06, "loss": 0.13145720958709717, "step": 25760 }, { "epoch": 0.6788210850732459, "grad_norm": 2.338712215423584, "learning_rate": 4.940519231526867e-06, "loss": 0.13066718578338624, "step": 25770 }, { "epoch": 0.6790845003177446, "grad_norm": 1.152863621711731, "learning_rate": 4.933162219327909e-06, "loss": 0.11894607543945312, "step": 25780 }, { "epoch": 0.6793479155622434, "grad_norm": 1.7359790802001953, "learning_rate": 4.925808894992087e-06, "loss": 0.12239128351211548, "step": 25790 }, { "epoch": 0.6796113308067421, "grad_norm": 1.245998501777649, "learning_rate": 4.918459263871474e-06, "loss": 0.120774245262146, "step": 25800 }, { "epoch": 0.6798747460512409, "grad_norm": 1.947583556175232, "learning_rate": 4.911113331315448e-06, "loss": 0.10977452993392944, "step": 25810 }, { "epoch": 0.6801381612957396, "grad_norm": 3.92486572265625, "learning_rate": 4.903771102670686e-06, "loss": 0.12560532093048096, "step": 25820 }, { "epoch": 0.6804015765402384, "grad_norm": 1.8999077081680298, "learning_rate": 4.896432583281189e-06, "loss": 0.12385615110397338, "step": 25830 }, { "epoch": 0.680664991784737, "grad_norm": 1.546134352684021, "learning_rate": 4.88909777848824e-06, "loss": 0.12220475673675538, "step": 25840 }, { "epoch": 0.6809284070292358, "grad_norm": 1.9499245882034302, "learning_rate": 4.881766693630439e-06, "loss": 0.13378565311431884, "step": 25850 }, { "epoch": 0.6811918222737345, "grad_norm": 1.7644106149673462, "learning_rate": 4.874439334043654e-06, "loss": 0.11505424976348877, "step": 25860 }, { "epoch": 0.6814552375182332, "grad_norm": 1.354675531387329, "learning_rate": 4.8671157050610525e-06, "loss": 0.1270553231239319, "step": 25870 }, { "epoch": 0.681718652762732, "grad_norm": 2.1083264350891113, "learning_rate": 4.859795812013099e-06, "loss": 0.13352460861206056, "step": 25880 }, { "epoch": 0.6819820680072307, "grad_norm": 1.9019241333007812, "learning_rate": 4.852479660227518e-06, "loss": 0.12289712429046631, "step": 25890 }, { "epoch": 0.6822454832517295, "grad_norm": 2.200294017791748, "learning_rate": 4.8451672550293236e-06, "loss": 0.11389013528823852, "step": 25900 }, { "epoch": 0.6825088984962282, "grad_norm": 1.6134483814239502, "learning_rate": 4.837858601740799e-06, "loss": 0.12364796400070191, "step": 25910 }, { "epoch": 0.682772313740727, "grad_norm": 3.032630443572998, "learning_rate": 4.830553705681496e-06, "loss": 0.12594592571258545, "step": 25920 }, { "epoch": 0.6830357289852257, "grad_norm": 1.6272228956222534, "learning_rate": 4.823252572168236e-06, "loss": 0.13017051219940184, "step": 25930 }, { "epoch": 0.6832991442297245, "grad_norm": 1.1104934215545654, "learning_rate": 4.815955206515104e-06, "loss": 0.11981709003448486, "step": 25940 }, { "epoch": 0.6835625594742232, "grad_norm": 2.9844746589660645, "learning_rate": 4.808661614033422e-06, "loss": 0.11952757835388184, "step": 25950 }, { "epoch": 0.6838259747187219, "grad_norm": 1.2262389659881592, "learning_rate": 4.801371800031796e-06, "loss": 0.12681870460510253, "step": 25960 }, { "epoch": 0.6840893899632207, "grad_norm": 2.1987059116363525, "learning_rate": 4.794085769816057e-06, "loss": 0.10583865642547607, "step": 25970 }, { "epoch": 0.6843528052077194, "grad_norm": 1.4853713512420654, "learning_rate": 4.786803528689304e-06, "loss": 0.13597750663757324, "step": 25980 }, { "epoch": 0.6846162204522181, "grad_norm": 3.4583563804626465, "learning_rate": 4.779525081951857e-06, "loss": 0.13274688720703126, "step": 25990 }, { "epoch": 0.6848796356967168, "grad_norm": 1.6022896766662598, "learning_rate": 4.772250434901281e-06, "loss": 0.12015447616577149, "step": 26000 }, { "epoch": 0.6851430509412156, "grad_norm": 1.6992111206054688, "learning_rate": 4.7649795928323875e-06, "loss": 0.11375772953033447, "step": 26010 }, { "epoch": 0.6854064661857143, "grad_norm": 1.5041215419769287, "learning_rate": 4.757712561037203e-06, "loss": 0.12051804065704345, "step": 26020 }, { "epoch": 0.6856698814302131, "grad_norm": 2.7262015342712402, "learning_rate": 4.750449344804989e-06, "loss": 0.11637892723083496, "step": 26030 }, { "epoch": 0.6859332966747118, "grad_norm": 4.875020980834961, "learning_rate": 4.743189949422227e-06, "loss": 0.14219164848327637, "step": 26040 }, { "epoch": 0.6861967119192105, "grad_norm": 2.334354877471924, "learning_rate": 4.735934380172613e-06, "loss": 0.12796710729598998, "step": 26050 }, { "epoch": 0.6864601271637093, "grad_norm": 1.745405673980713, "learning_rate": 4.728682642337072e-06, "loss": 0.12393312454223633, "step": 26060 }, { "epoch": 0.686723542408208, "grad_norm": 1.6625155210494995, "learning_rate": 4.7214347411937325e-06, "loss": 0.12193077802658081, "step": 26070 }, { "epoch": 0.6869869576527068, "grad_norm": 1.2415653467178345, "learning_rate": 4.714190682017918e-06, "loss": 0.12704617977142335, "step": 26080 }, { "epoch": 0.6872503728972055, "grad_norm": 2.1714770793914795, "learning_rate": 4.706950470082177e-06, "loss": 0.1187453031539917, "step": 26090 }, { "epoch": 0.6875137881417043, "grad_norm": 2.4250142574310303, "learning_rate": 4.699714110656242e-06, "loss": 0.11488217115402222, "step": 26100 }, { "epoch": 0.687777203386203, "grad_norm": 2.2588932514190674, "learning_rate": 4.6924816090070575e-06, "loss": 0.11745505332946778, "step": 26110 }, { "epoch": 0.6880406186307017, "grad_norm": 2.4011032581329346, "learning_rate": 4.685252970398742e-06, "loss": 0.1336587905883789, "step": 26120 }, { "epoch": 0.6883040338752004, "grad_norm": 1.612715721130371, "learning_rate": 4.678028200092607e-06, "loss": 0.12750519514083863, "step": 26130 }, { "epoch": 0.6885674491196991, "grad_norm": 1.7430686950683594, "learning_rate": 4.670807303347161e-06, "loss": 0.12025229930877686, "step": 26140 }, { "epoch": 0.6888308643641979, "grad_norm": 3.3882124423980713, "learning_rate": 4.663590285418082e-06, "loss": 0.1263115406036377, "step": 26150 }, { "epoch": 0.6890942796086966, "grad_norm": 2.8533949851989746, "learning_rate": 4.656377151558224e-06, "loss": 0.10856816768646241, "step": 26160 }, { "epoch": 0.6893576948531954, "grad_norm": 1.891230583190918, "learning_rate": 4.64916790701762e-06, "loss": 0.13140718936920165, "step": 26170 }, { "epoch": 0.6896211100976941, "grad_norm": 3.0747363567352295, "learning_rate": 4.641962557043462e-06, "loss": 0.12021911144256592, "step": 26180 }, { "epoch": 0.6898845253421929, "grad_norm": 2.3459925651550293, "learning_rate": 4.634761106880127e-06, "loss": 0.11888906955718995, "step": 26190 }, { "epoch": 0.6901479405866916, "grad_norm": 1.6854581832885742, "learning_rate": 4.627563561769136e-06, "loss": 0.12369735240936279, "step": 26200 }, { "epoch": 0.6904113558311903, "grad_norm": 1.455118179321289, "learning_rate": 4.620369926949164e-06, "loss": 0.11730568408966065, "step": 26210 }, { "epoch": 0.6906747710756891, "grad_norm": 1.7231545448303223, "learning_rate": 4.613180207656061e-06, "loss": 0.10521879196166992, "step": 26220 }, { "epoch": 0.6909381863201878, "grad_norm": 1.3967618942260742, "learning_rate": 4.60599440912281e-06, "loss": 0.1232597827911377, "step": 26230 }, { "epoch": 0.6912016015646866, "grad_norm": 2.6301991939544678, "learning_rate": 4.598812536579545e-06, "loss": 0.11642484664916992, "step": 26240 }, { "epoch": 0.6914650168091853, "grad_norm": 1.5613361597061157, "learning_rate": 4.591634595253541e-06, "loss": 0.11451929807662964, "step": 26250 }, { "epoch": 0.6917284320536841, "grad_norm": 1.7709027528762817, "learning_rate": 4.584460590369214e-06, "loss": 0.12890610694885254, "step": 26260 }, { "epoch": 0.6919918472981827, "grad_norm": 2.104465961456299, "learning_rate": 4.577290527148118e-06, "loss": 0.11669925451278687, "step": 26270 }, { "epoch": 0.6922552625426815, "grad_norm": 1.6759536266326904, "learning_rate": 4.5701244108089325e-06, "loss": 0.12767053842544557, "step": 26280 }, { "epoch": 0.6925186777871802, "grad_norm": 1.4319974184036255, "learning_rate": 4.562962246567466e-06, "loss": 0.1348447322845459, "step": 26290 }, { "epoch": 0.6927820930316789, "grad_norm": 2.991786241531372, "learning_rate": 4.555804039636649e-06, "loss": 0.1262993574142456, "step": 26300 }, { "epoch": 0.6930455082761777, "grad_norm": 4.912561893463135, "learning_rate": 4.548649795226532e-06, "loss": 0.12200181484222412, "step": 26310 }, { "epoch": 0.6933089235206764, "grad_norm": 1.6510672569274902, "learning_rate": 4.541499518544292e-06, "loss": 0.12217485904693604, "step": 26320 }, { "epoch": 0.6935723387651752, "grad_norm": 2.6622209548950195, "learning_rate": 4.5343532147941985e-06, "loss": 0.11861733198165894, "step": 26330 }, { "epoch": 0.6938357540096739, "grad_norm": 2.2439355850219727, "learning_rate": 4.527210889177639e-06, "loss": 0.13286852836608887, "step": 26340 }, { "epoch": 0.6940991692541727, "grad_norm": 2.7648415565490723, "learning_rate": 4.520072546893115e-06, "loss": 0.13548784255981444, "step": 26350 }, { "epoch": 0.6943625844986714, "grad_norm": 1.700618028640747, "learning_rate": 4.512938193136216e-06, "loss": 0.11270358562469482, "step": 26360 }, { "epoch": 0.6946259997431702, "grad_norm": 1.456101655960083, "learning_rate": 4.5058078330996315e-06, "loss": 0.11571038961410522, "step": 26370 }, { "epoch": 0.6948894149876689, "grad_norm": 2.2863965034484863, "learning_rate": 4.498681471973145e-06, "loss": 0.1144704818725586, "step": 26380 }, { "epoch": 0.6951528302321676, "grad_norm": 1.9190670251846313, "learning_rate": 4.491559114943628e-06, "loss": 0.11661056280136109, "step": 26390 }, { "epoch": 0.6954162454766664, "grad_norm": 1.7149395942687988, "learning_rate": 4.484440767195044e-06, "loss": 0.1311408281326294, "step": 26400 }, { "epoch": 0.695679660721165, "grad_norm": 1.3221155405044556, "learning_rate": 4.477326433908435e-06, "loss": 0.12428263425827027, "step": 26410 }, { "epoch": 0.6959430759656638, "grad_norm": 1.1262234449386597, "learning_rate": 4.4702161202619085e-06, "loss": 0.12761590480804444, "step": 26420 }, { "epoch": 0.6962064912101625, "grad_norm": 1.5094705820083618, "learning_rate": 4.463109831430668e-06, "loss": 0.11901404857635497, "step": 26430 }, { "epoch": 0.6964699064546613, "grad_norm": 1.8839359283447266, "learning_rate": 4.456007572586971e-06, "loss": 0.1253509044647217, "step": 26440 }, { "epoch": 0.69673332169916, "grad_norm": 2.435302972793579, "learning_rate": 4.448909348900159e-06, "loss": 0.12538061141967774, "step": 26450 }, { "epoch": 0.6969967369436588, "grad_norm": 2.3442909717559814, "learning_rate": 4.441815165536612e-06, "loss": 0.13341095447540283, "step": 26460 }, { "epoch": 0.6972601521881575, "grad_norm": 1.482391119003296, "learning_rate": 4.434725027659786e-06, "loss": 0.11527023315429688, "step": 26470 }, { "epoch": 0.6975235674326562, "grad_norm": 2.731410503387451, "learning_rate": 4.427638940430192e-06, "loss": 0.11354727745056152, "step": 26480 }, { "epoch": 0.697786982677155, "grad_norm": 2.3016715049743652, "learning_rate": 4.420556909005392e-06, "loss": 0.12296590805053711, "step": 26490 }, { "epoch": 0.6980503979216537, "grad_norm": 3.1985208988189697, "learning_rate": 4.41347893853999e-06, "loss": 0.12965693473815917, "step": 26500 }, { "epoch": 0.6983138131661525, "grad_norm": 4.110313892364502, "learning_rate": 4.40640503418564e-06, "loss": 0.11869328022003174, "step": 26510 }, { "epoch": 0.6985772284106512, "grad_norm": 1.2606509923934937, "learning_rate": 4.39933520109103e-06, "loss": 0.11100050210952758, "step": 26520 }, { "epoch": 0.69884064365515, "grad_norm": 2.8846733570098877, "learning_rate": 4.3922694444018985e-06, "loss": 0.11425089836120605, "step": 26530 }, { "epoch": 0.6991040588996487, "grad_norm": 2.3871450424194336, "learning_rate": 4.385207769261007e-06, "loss": 0.11136502027511597, "step": 26540 }, { "epoch": 0.6993674741441475, "grad_norm": 3.483402729034424, "learning_rate": 4.378150180808136e-06, "loss": 0.12597562074661256, "step": 26550 }, { "epoch": 0.6996308893886461, "grad_norm": 1.9727181196212769, "learning_rate": 4.371096684180114e-06, "loss": 0.12381397485733033, "step": 26560 }, { "epoch": 0.6998943046331448, "grad_norm": 2.056084632873535, "learning_rate": 4.364047284510774e-06, "loss": 0.10021485090255737, "step": 26570 }, { "epoch": 0.7001577198776436, "grad_norm": 2.3417341709136963, "learning_rate": 4.357001986930982e-06, "loss": 0.11826680898666382, "step": 26580 }, { "epoch": 0.7004211351221423, "grad_norm": 3.364931344985962, "learning_rate": 4.349960796568601e-06, "loss": 0.12663872241973878, "step": 26590 }, { "epoch": 0.7006845503666411, "grad_norm": 2.1192240715026855, "learning_rate": 4.342923718548512e-06, "loss": 0.12054129838943481, "step": 26600 }, { "epoch": 0.7009479656111398, "grad_norm": 2.0279369354248047, "learning_rate": 4.335890757992611e-06, "loss": 0.11460952758789063, "step": 26610 }, { "epoch": 0.7012113808556386, "grad_norm": 2.2293221950531006, "learning_rate": 4.328861920019787e-06, "loss": 0.11111844778060913, "step": 26620 }, { "epoch": 0.7014747961001373, "grad_norm": 3.7311031818389893, "learning_rate": 4.321837209745931e-06, "loss": 0.11832411289215088, "step": 26630 }, { "epoch": 0.701738211344636, "grad_norm": 1.481664776802063, "learning_rate": 4.3148166322839305e-06, "loss": 0.11968414783477783, "step": 26640 }, { "epoch": 0.7020016265891348, "grad_norm": 2.1725823879241943, "learning_rate": 4.3078001927436605e-06, "loss": 0.11374502182006836, "step": 26650 }, { "epoch": 0.7022650418336335, "grad_norm": 1.658681035041809, "learning_rate": 4.300787896231997e-06, "loss": 0.1113494873046875, "step": 26660 }, { "epoch": 0.7025284570781323, "grad_norm": 2.788360595703125, "learning_rate": 4.293779747852791e-06, "loss": 0.11965398788452149, "step": 26670 }, { "epoch": 0.702791872322631, "grad_norm": 3.5976130962371826, "learning_rate": 4.286775752706865e-06, "loss": 0.1201399564743042, "step": 26680 }, { "epoch": 0.7030552875671298, "grad_norm": 2.119969606399536, "learning_rate": 4.279775915892038e-06, "loss": 0.11991989612579346, "step": 26690 }, { "epoch": 0.7033187028116284, "grad_norm": 1.6385959386825562, "learning_rate": 4.2727802425030885e-06, "loss": 0.11290417909622193, "step": 26700 }, { "epoch": 0.7035821180561272, "grad_norm": 1.7687398195266724, "learning_rate": 4.265788737631779e-06, "loss": 0.10695328712463378, "step": 26710 }, { "epoch": 0.7038455333006259, "grad_norm": 1.7563562393188477, "learning_rate": 4.258801406366817e-06, "loss": 0.1266298770904541, "step": 26720 }, { "epoch": 0.7041089485451246, "grad_norm": 3.085947036743164, "learning_rate": 4.251818253793885e-06, "loss": 0.1286473512649536, "step": 26730 }, { "epoch": 0.7043723637896234, "grad_norm": 2.0182485580444336, "learning_rate": 4.244839284995627e-06, "loss": 0.12445807456970215, "step": 26740 }, { "epoch": 0.7046357790341221, "grad_norm": 2.0682525634765625, "learning_rate": 4.237864505051638e-06, "loss": 0.12448686361312866, "step": 26750 }, { "epoch": 0.7048991942786209, "grad_norm": 1.6319079399108887, "learning_rate": 4.230893919038461e-06, "loss": 0.1277535080909729, "step": 26760 }, { "epoch": 0.7051626095231196, "grad_norm": 1.926567554473877, "learning_rate": 4.223927532029592e-06, "loss": 0.10826990604400635, "step": 26770 }, { "epoch": 0.7054260247676184, "grad_norm": 1.8328527212142944, "learning_rate": 4.21696534909546e-06, "loss": 0.13258168697357178, "step": 26780 }, { "epoch": 0.7056894400121171, "grad_norm": 1.7361630201339722, "learning_rate": 4.2100073753034535e-06, "loss": 0.12158634662628173, "step": 26790 }, { "epoch": 0.7059528552566159, "grad_norm": 2.3599905967712402, "learning_rate": 4.203053615717884e-06, "loss": 0.12190284729003906, "step": 26800 }, { "epoch": 0.7062162705011146, "grad_norm": 1.4248684644699097, "learning_rate": 4.196104075399986e-06, "loss": 0.12570595741271973, "step": 26810 }, { "epoch": 0.7064796857456133, "grad_norm": 1.8554799556732178, "learning_rate": 4.1891587594079485e-06, "loss": 0.12967991828918457, "step": 26820 }, { "epoch": 0.7067431009901121, "grad_norm": 2.24762225151062, "learning_rate": 4.182217672796867e-06, "loss": 0.12562665939331055, "step": 26830 }, { "epoch": 0.7070065162346107, "grad_norm": 1.3729443550109863, "learning_rate": 4.175280820618763e-06, "loss": 0.11891264915466308, "step": 26840 }, { "epoch": 0.7072699314791095, "grad_norm": 2.0037741661071777, "learning_rate": 4.168348207922577e-06, "loss": 0.10033659934997559, "step": 26850 }, { "epoch": 0.7075333467236082, "grad_norm": 2.267928123474121, "learning_rate": 4.16141983975416e-06, "loss": 0.12414878606796265, "step": 26860 }, { "epoch": 0.707796761968107, "grad_norm": 3.8586292266845703, "learning_rate": 4.154495721156287e-06, "loss": 0.12230956554412842, "step": 26870 }, { "epoch": 0.7080601772126057, "grad_norm": 2.4847373962402344, "learning_rate": 4.147575857168623e-06, "loss": 0.11762416362762451, "step": 26880 }, { "epoch": 0.7083235924571045, "grad_norm": 1.6179180145263672, "learning_rate": 4.140660252827746e-06, "loss": 0.10897264480590821, "step": 26890 }, { "epoch": 0.7085870077016032, "grad_norm": 3.1633129119873047, "learning_rate": 4.13374891316713e-06, "loss": 0.11292308568954468, "step": 26900 }, { "epoch": 0.7088504229461019, "grad_norm": 1.5455693006515503, "learning_rate": 4.126841843217144e-06, "loss": 0.11423385143280029, "step": 26910 }, { "epoch": 0.7091138381906007, "grad_norm": 3.077936887741089, "learning_rate": 4.1199390480050605e-06, "loss": 0.12521083354949952, "step": 26920 }, { "epoch": 0.7093772534350994, "grad_norm": 2.1157896518707275, "learning_rate": 4.113040532555021e-06, "loss": 0.11155164241790771, "step": 26930 }, { "epoch": 0.7096406686795982, "grad_norm": 2.442798614501953, "learning_rate": 4.106146301888065e-06, "loss": 0.12597293853759767, "step": 26940 }, { "epoch": 0.7099040839240969, "grad_norm": 1.301624059677124, "learning_rate": 4.0992563610221155e-06, "loss": 0.11274127960205078, "step": 26950 }, { "epoch": 0.7101674991685957, "grad_norm": 2.0597333908081055, "learning_rate": 4.092370714971966e-06, "loss": 0.12024319171905518, "step": 26960 }, { "epoch": 0.7104309144130944, "grad_norm": 2.1247289180755615, "learning_rate": 4.085489368749287e-06, "loss": 0.12249124050140381, "step": 26970 }, { "epoch": 0.7106943296575932, "grad_norm": 1.7592837810516357, "learning_rate": 4.07861232736262e-06, "loss": 0.12732572555541993, "step": 26980 }, { "epoch": 0.7109577449020918, "grad_norm": 3.865421772003174, "learning_rate": 4.0717395958173665e-06, "loss": 0.11615945100784301, "step": 26990 }, { "epoch": 0.7112211601465905, "grad_norm": 3.308382511138916, "learning_rate": 4.064871179115807e-06, "loss": 0.11719996929168701, "step": 27000 }, { "epoch": 0.7114845753910893, "grad_norm": 2.4195311069488525, "learning_rate": 4.0580070822570694e-06, "loss": 0.1188173770904541, "step": 27010 }, { "epoch": 0.711747990635588, "grad_norm": 1.7258294820785522, "learning_rate": 4.051147310237129e-06, "loss": 0.1196479320526123, "step": 27020 }, { "epoch": 0.7120114058800868, "grad_norm": 1.8645930290222168, "learning_rate": 4.044291868048837e-06, "loss": 0.11552979946136474, "step": 27030 }, { "epoch": 0.7122748211245855, "grad_norm": 2.6764259338378906, "learning_rate": 4.0374407606818684e-06, "loss": 0.12141351699829102, "step": 27040 }, { "epoch": 0.7125382363690843, "grad_norm": 2.8595006465911865, "learning_rate": 4.03059399312277e-06, "loss": 0.1404463529586792, "step": 27050 }, { "epoch": 0.712801651613583, "grad_norm": 2.312668800354004, "learning_rate": 4.0237515703549015e-06, "loss": 0.11730341911315918, "step": 27060 }, { "epoch": 0.7130650668580818, "grad_norm": 1.3786437511444092, "learning_rate": 4.0169134973584726e-06, "loss": 0.10395466089248658, "step": 27070 }, { "epoch": 0.7133284821025805, "grad_norm": 1.815611481666565, "learning_rate": 4.010079779110537e-06, "loss": 0.12207682132720947, "step": 27080 }, { "epoch": 0.7135918973470792, "grad_norm": 1.4056907892227173, "learning_rate": 4.003250420584966e-06, "loss": 0.12945631742477418, "step": 27090 }, { "epoch": 0.713855312591578, "grad_norm": 2.373556137084961, "learning_rate": 3.996425426752459e-06, "loss": 0.1268446683883667, "step": 27100 }, { "epoch": 0.7141187278360767, "grad_norm": 2.184959888458252, "learning_rate": 3.989604802580545e-06, "loss": 0.11992778778076171, "step": 27110 }, { "epoch": 0.7143821430805755, "grad_norm": 1.7186497449874878, "learning_rate": 3.982788553033561e-06, "loss": 0.11136140823364257, "step": 27120 }, { "epoch": 0.7146455583250741, "grad_norm": 2.1087915897369385, "learning_rate": 3.975976683072676e-06, "loss": 0.12384116649627686, "step": 27130 }, { "epoch": 0.714908973569573, "grad_norm": 1.779862403869629, "learning_rate": 3.9691691976558636e-06, "loss": 0.1307281494140625, "step": 27140 }, { "epoch": 0.7151723888140716, "grad_norm": 1.903941035270691, "learning_rate": 3.9623661017378946e-06, "loss": 0.12615103721618653, "step": 27150 }, { "epoch": 0.7154358040585704, "grad_norm": 1.837939739227295, "learning_rate": 3.955567400270365e-06, "loss": 0.11715512275695801, "step": 27160 }, { "epoch": 0.7156992193030691, "grad_norm": 2.01240611076355, "learning_rate": 3.948773098201657e-06, "loss": 0.11161549091339111, "step": 27170 }, { "epoch": 0.7159626345475678, "grad_norm": 2.292512893676758, "learning_rate": 3.941983200476969e-06, "loss": 0.11496659517288207, "step": 27180 }, { "epoch": 0.7162260497920666, "grad_norm": 2.3947837352752686, "learning_rate": 3.935197712038267e-06, "loss": 0.1144145131111145, "step": 27190 }, { "epoch": 0.7164894650365653, "grad_norm": 1.313780426979065, "learning_rate": 3.928416637824326e-06, "loss": 0.11485271453857422, "step": 27200 }, { "epoch": 0.7167528802810641, "grad_norm": 1.8613970279693604, "learning_rate": 3.921639982770709e-06, "loss": 0.12563755512237548, "step": 27210 }, { "epoch": 0.7170162955255628, "grad_norm": 1.3744745254516602, "learning_rate": 3.914867751809756e-06, "loss": 0.12355408668518067, "step": 27220 }, { "epoch": 0.7172797107700616, "grad_norm": 2.3125526905059814, "learning_rate": 3.908099949870588e-06, "loss": 0.1235579490661621, "step": 27230 }, { "epoch": 0.7175431260145603, "grad_norm": 2.815727472305298, "learning_rate": 3.901336581879102e-06, "loss": 0.1130378007888794, "step": 27240 }, { "epoch": 0.717806541259059, "grad_norm": 2.366792678833008, "learning_rate": 3.894577652757968e-06, "loss": 0.12487509250640869, "step": 27250 }, { "epoch": 0.7180699565035578, "grad_norm": 1.4414470195770264, "learning_rate": 3.887823167426633e-06, "loss": 0.12217295169830322, "step": 27260 }, { "epoch": 0.7183333717480564, "grad_norm": 1.14555823802948, "learning_rate": 3.8810731308013005e-06, "loss": 0.11535923480987549, "step": 27270 }, { "epoch": 0.7185967869925552, "grad_norm": 1.588911533355713, "learning_rate": 3.874327547794931e-06, "loss": 0.11798219680786133, "step": 27280 }, { "epoch": 0.7188602022370539, "grad_norm": 2.6106529235839844, "learning_rate": 3.8675864233172575e-06, "loss": 0.11652660369873047, "step": 27290 }, { "epoch": 0.7191236174815527, "grad_norm": 1.2776234149932861, "learning_rate": 3.860849762274759e-06, "loss": 0.10741761922836304, "step": 27300 }, { "epoch": 0.7193870327260514, "grad_norm": 2.1073081493377686, "learning_rate": 3.854117569570674e-06, "loss": 0.12002412080764771, "step": 27310 }, { "epoch": 0.7196504479705502, "grad_norm": 1.4713921546936035, "learning_rate": 3.8473898501049745e-06, "loss": 0.11782877445220948, "step": 27320 }, { "epoch": 0.7199138632150489, "grad_norm": 2.3012285232543945, "learning_rate": 3.840666608774385e-06, "loss": 0.1109262228012085, "step": 27330 }, { "epoch": 0.7201772784595476, "grad_norm": 2.519268274307251, "learning_rate": 3.833947850472378e-06, "loss": 0.11556572914123535, "step": 27340 }, { "epoch": 0.7204406937040464, "grad_norm": 2.5270938873291016, "learning_rate": 3.827233580089152e-06, "loss": 0.13329869508743286, "step": 27350 }, { "epoch": 0.7207041089485451, "grad_norm": 2.875605821609497, "learning_rate": 3.82052380251164e-06, "loss": 0.12360560894012451, "step": 27360 }, { "epoch": 0.7209675241930439, "grad_norm": 2.08943510055542, "learning_rate": 3.8138185226235114e-06, "loss": 0.11058629751205444, "step": 27370 }, { "epoch": 0.7212309394375426, "grad_norm": 2.7233753204345703, "learning_rate": 3.8071177453051523e-06, "loss": 0.1271904230117798, "step": 27380 }, { "epoch": 0.7214943546820414, "grad_norm": 2.9160163402557373, "learning_rate": 3.8004214754336853e-06, "loss": 0.11686382293701172, "step": 27390 }, { "epoch": 0.7217577699265401, "grad_norm": 1.6686164140701294, "learning_rate": 3.7937297178829458e-06, "loss": 0.10483887195587158, "step": 27400 }, { "epoch": 0.7220211851710389, "grad_norm": 1.691189169883728, "learning_rate": 3.7870424775234703e-06, "loss": 0.11175205707550048, "step": 27410 }, { "epoch": 0.7222846004155375, "grad_norm": 1.2285456657409668, "learning_rate": 3.780359759222534e-06, "loss": 0.1154709815979004, "step": 27420 }, { "epoch": 0.7225480156600362, "grad_norm": 2.0866854190826416, "learning_rate": 3.773681567844103e-06, "loss": 0.11678206920623779, "step": 27430 }, { "epoch": 0.722811430904535, "grad_norm": 1.988685965538025, "learning_rate": 3.767007908248852e-06, "loss": 0.11149398088455201, "step": 27440 }, { "epoch": 0.7230748461490337, "grad_norm": 2.807574510574341, "learning_rate": 3.7603387852941608e-06, "loss": 0.11002161502838134, "step": 27450 }, { "epoch": 0.7233382613935325, "grad_norm": 2.8468539714813232, "learning_rate": 3.753674203834099e-06, "loss": 0.10433048009872437, "step": 27460 }, { "epoch": 0.7236016766380312, "grad_norm": 3.6609861850738525, "learning_rate": 3.747014168719445e-06, "loss": 0.11099872589111329, "step": 27470 }, { "epoch": 0.72386509188253, "grad_norm": 2.0834274291992188, "learning_rate": 3.740358684797656e-06, "loss": 0.11933672428131104, "step": 27480 }, { "epoch": 0.7241285071270287, "grad_norm": 1.8659721612930298, "learning_rate": 3.7337077569128797e-06, "loss": 0.11117134094238282, "step": 27490 }, { "epoch": 0.7243919223715275, "grad_norm": 2.3260390758514404, "learning_rate": 3.7270613899059495e-06, "loss": 0.10870425701141358, "step": 27500 }, { "epoch": 0.7246553376160262, "grad_norm": 1.9390106201171875, "learning_rate": 3.7204195886143736e-06, "loss": 0.10659770965576172, "step": 27510 }, { "epoch": 0.7249187528605249, "grad_norm": 2.148266077041626, "learning_rate": 3.713782357872352e-06, "loss": 0.11148576736450196, "step": 27520 }, { "epoch": 0.7251821681050237, "grad_norm": 1.7494640350341797, "learning_rate": 3.7071497025107373e-06, "loss": 0.11850289106369019, "step": 27530 }, { "epoch": 0.7254455833495224, "grad_norm": 1.9640352725982666, "learning_rate": 3.7005216273570623e-06, "loss": 0.10389602184295654, "step": 27540 }, { "epoch": 0.7257089985940212, "grad_norm": 2.854928493499756, "learning_rate": 3.693898137235532e-06, "loss": 0.11318815946578979, "step": 27550 }, { "epoch": 0.7259724138385198, "grad_norm": 2.073033094406128, "learning_rate": 3.6872792369670074e-06, "loss": 0.11658592224121093, "step": 27560 }, { "epoch": 0.7262358290830186, "grad_norm": 1.7125049829483032, "learning_rate": 3.6806649313690066e-06, "loss": 0.11368513107299805, "step": 27570 }, { "epoch": 0.7264992443275173, "grad_norm": 1.832025408744812, "learning_rate": 3.6740552252557093e-06, "loss": 0.11881229877471924, "step": 27580 }, { "epoch": 0.7267626595720161, "grad_norm": 1.753412127494812, "learning_rate": 3.667450123437939e-06, "loss": 0.11736853122711181, "step": 27590 }, { "epoch": 0.7270260748165148, "grad_norm": 2.005425214767456, "learning_rate": 3.6608496307231823e-06, "loss": 0.11939446926116944, "step": 27600 }, { "epoch": 0.7272894900610135, "grad_norm": 2.7938408851623535, "learning_rate": 3.6542537519155617e-06, "loss": 0.09597814083099365, "step": 27610 }, { "epoch": 0.7275529053055123, "grad_norm": 2.395639181137085, "learning_rate": 3.647662491815833e-06, "loss": 0.11563736200332642, "step": 27620 }, { "epoch": 0.727816320550011, "grad_norm": 2.290195941925049, "learning_rate": 3.641075855221409e-06, "loss": 0.12693580389022827, "step": 27630 }, { "epoch": 0.7280797357945098, "grad_norm": 1.555456280708313, "learning_rate": 3.634493846926321e-06, "loss": 0.1259748101234436, "step": 27640 }, { "epoch": 0.7283431510390085, "grad_norm": 2.5939781665802, "learning_rate": 3.62791647172125e-06, "loss": 0.12127366065979003, "step": 27650 }, { "epoch": 0.7286065662835073, "grad_norm": 3.358577251434326, "learning_rate": 3.621343734393482e-06, "loss": 0.12191938161849976, "step": 27660 }, { "epoch": 0.728869981528006, "grad_norm": 1.7091481685638428, "learning_rate": 3.6147756397269397e-06, "loss": 0.12069642543792725, "step": 27670 }, { "epoch": 0.7291333967725048, "grad_norm": 2.5683908462524414, "learning_rate": 3.6082121925021717e-06, "loss": 0.11628499031066894, "step": 27680 }, { "epoch": 0.7293968120170035, "grad_norm": 1.9838377237319946, "learning_rate": 3.601653397496335e-06, "loss": 0.10870476961135864, "step": 27690 }, { "epoch": 0.7296602272615021, "grad_norm": 1.913285732269287, "learning_rate": 3.5950992594832044e-06, "loss": 0.12200069427490234, "step": 27700 }, { "epoch": 0.729923642506001, "grad_norm": 1.528165578842163, "learning_rate": 3.588549783233164e-06, "loss": 0.10116341114044189, "step": 27710 }, { "epoch": 0.7301870577504996, "grad_norm": 2.605987310409546, "learning_rate": 3.582004973513201e-06, "loss": 0.11914750337600707, "step": 27720 }, { "epoch": 0.7304504729949984, "grad_norm": 3.8213164806365967, "learning_rate": 3.575464835086917e-06, "loss": 0.12269492149353027, "step": 27730 }, { "epoch": 0.7307138882394971, "grad_norm": 3.9061880111694336, "learning_rate": 3.568929372714508e-06, "loss": 0.13363962173461913, "step": 27740 }, { "epoch": 0.7309773034839959, "grad_norm": 3.5101239681243896, "learning_rate": 3.562398591152755e-06, "loss": 0.11931343078613281, "step": 27750 }, { "epoch": 0.7312407187284946, "grad_norm": 1.702319622039795, "learning_rate": 3.5558724951550515e-06, "loss": 0.10830149650573731, "step": 27760 }, { "epoch": 0.7315041339729933, "grad_norm": 1.8454805612564087, "learning_rate": 3.549351089471367e-06, "loss": 0.1051034927368164, "step": 27770 }, { "epoch": 0.7317675492174921, "grad_norm": 2.7712953090667725, "learning_rate": 3.54283437884827e-06, "loss": 0.11631652116775512, "step": 27780 }, { "epoch": 0.7320309644619908, "grad_norm": 2.3002946376800537, "learning_rate": 3.536322368028895e-06, "loss": 0.11883697509765626, "step": 27790 }, { "epoch": 0.7322943797064896, "grad_norm": 2.4720661640167236, "learning_rate": 3.5298150617529635e-06, "loss": 0.11764998435974121, "step": 27800 }, { "epoch": 0.7325577949509883, "grad_norm": 1.9907485246658325, "learning_rate": 3.5233124647567817e-06, "loss": 0.12565019130706787, "step": 27810 }, { "epoch": 0.7328212101954871, "grad_norm": 1.8452383279800415, "learning_rate": 3.516814581773216e-06, "loss": 0.11414698362350464, "step": 27820 }, { "epoch": 0.7330846254399858, "grad_norm": 2.711698055267334, "learning_rate": 3.5103214175317056e-06, "loss": 0.11439673900604248, "step": 27830 }, { "epoch": 0.7333480406844846, "grad_norm": 1.6293158531188965, "learning_rate": 3.5038329767582567e-06, "loss": 0.11911318302154542, "step": 27840 }, { "epoch": 0.7336114559289832, "grad_norm": 2.150101900100708, "learning_rate": 3.4973492641754315e-06, "loss": 0.11258704662322998, "step": 27850 }, { "epoch": 0.7338748711734819, "grad_norm": 1.769584059715271, "learning_rate": 3.4908702845023643e-06, "loss": 0.11607253551483154, "step": 27860 }, { "epoch": 0.7341382864179807, "grad_norm": 1.9424145221710205, "learning_rate": 3.484396042454734e-06, "loss": 0.12028863430023193, "step": 27870 }, { "epoch": 0.7344017016624794, "grad_norm": 2.379554271697998, "learning_rate": 3.477926542744766e-06, "loss": 0.12301790714263916, "step": 27880 }, { "epoch": 0.7346651169069782, "grad_norm": 3.7351670265197754, "learning_rate": 3.4714617900812485e-06, "loss": 0.12777965068817138, "step": 27890 }, { "epoch": 0.7349285321514769, "grad_norm": 2.6949901580810547, "learning_rate": 3.465001789169502e-06, "loss": 0.12214939594268799, "step": 27900 }, { "epoch": 0.7351919473959757, "grad_norm": 2.1627464294433594, "learning_rate": 3.458546544711404e-06, "loss": 0.12240843772888184, "step": 27910 }, { "epoch": 0.7354553626404744, "grad_norm": 2.7009963989257812, "learning_rate": 3.45209606140535e-06, "loss": 0.11173354387283325, "step": 27920 }, { "epoch": 0.7357187778849732, "grad_norm": 2.1920394897460938, "learning_rate": 3.4456503439462797e-06, "loss": 0.10970799922943116, "step": 27930 }, { "epoch": 0.7359821931294719, "grad_norm": 2.6061880588531494, "learning_rate": 3.4392093970256703e-06, "loss": 0.13154207468032836, "step": 27940 }, { "epoch": 0.7362456083739706, "grad_norm": 2.269813299179077, "learning_rate": 3.4327732253315195e-06, "loss": 0.11904635429382324, "step": 27950 }, { "epoch": 0.7365090236184694, "grad_norm": 2.3488035202026367, "learning_rate": 3.4263418335483477e-06, "loss": 0.12121987342834473, "step": 27960 }, { "epoch": 0.7367724388629681, "grad_norm": 2.151921272277832, "learning_rate": 3.419915226357203e-06, "loss": 0.11011267900466919, "step": 27970 }, { "epoch": 0.7370358541074669, "grad_norm": 2.0613608360290527, "learning_rate": 3.413493408435642e-06, "loss": 0.12241623401641846, "step": 27980 }, { "epoch": 0.7372992693519655, "grad_norm": 1.9436496496200562, "learning_rate": 3.407076384457747e-06, "loss": 0.10541741847991944, "step": 27990 }, { "epoch": 0.7375626845964643, "grad_norm": 1.7892321348190308, "learning_rate": 3.400664159094106e-06, "loss": 0.10291869640350342, "step": 28000 }, { "epoch": 0.737826099840963, "grad_norm": 2.6249537467956543, "learning_rate": 3.394256737011803e-06, "loss": 0.10971648693084717, "step": 28010 }, { "epoch": 0.7380895150854618, "grad_norm": 1.7222647666931152, "learning_rate": 3.3878541228744455e-06, "loss": 0.10481021404266358, "step": 28020 }, { "epoch": 0.7383529303299605, "grad_norm": 1.4067392349243164, "learning_rate": 3.38145632134213e-06, "loss": 0.10281584262847901, "step": 28030 }, { "epoch": 0.7386163455744592, "grad_norm": 2.1524384021759033, "learning_rate": 3.3750633370714527e-06, "loss": 0.11427837610244751, "step": 28040 }, { "epoch": 0.738879760818958, "grad_norm": 1.5214465856552124, "learning_rate": 3.368675174715501e-06, "loss": 0.11900875568389893, "step": 28050 }, { "epoch": 0.7391431760634567, "grad_norm": 3.008248805999756, "learning_rate": 3.3622918389238544e-06, "loss": 0.1164210557937622, "step": 28060 }, { "epoch": 0.7394065913079555, "grad_norm": 1.6686649322509766, "learning_rate": 3.3559133343425865e-06, "loss": 0.11529295444488526, "step": 28070 }, { "epoch": 0.7396700065524542, "grad_norm": 2.33707857131958, "learning_rate": 3.3495396656142442e-06, "loss": 0.10960131883621216, "step": 28080 }, { "epoch": 0.739933421796953, "grad_norm": 1.5696600675582886, "learning_rate": 3.343170837377858e-06, "loss": 0.11570202112197876, "step": 28090 }, { "epoch": 0.7401968370414517, "grad_norm": 2.380117416381836, "learning_rate": 3.336806854268936e-06, "loss": 0.11045346260070801, "step": 28100 }, { "epoch": 0.7404602522859505, "grad_norm": 1.6833243370056152, "learning_rate": 3.3304477209194574e-06, "loss": 0.12003291845321655, "step": 28110 }, { "epoch": 0.7407236675304492, "grad_norm": 1.8200352191925049, "learning_rate": 3.3240934419578828e-06, "loss": 0.1147235631942749, "step": 28120 }, { "epoch": 0.7409870827749478, "grad_norm": 1.1442477703094482, "learning_rate": 3.3177440220091207e-06, "loss": 0.1099840521812439, "step": 28130 }, { "epoch": 0.7412504980194466, "grad_norm": 1.9148333072662354, "learning_rate": 3.3113994656945537e-06, "loss": 0.11134450435638428, "step": 28140 }, { "epoch": 0.7415139132639453, "grad_norm": 3.2819459438323975, "learning_rate": 3.3050597776320293e-06, "loss": 0.10599570274353028, "step": 28150 }, { "epoch": 0.7417773285084441, "grad_norm": 2.009871482849121, "learning_rate": 3.298724962435842e-06, "loss": 0.1037330985069275, "step": 28160 }, { "epoch": 0.7420407437529428, "grad_norm": 3.6019153594970703, "learning_rate": 3.292395024716746e-06, "loss": 0.12058223485946655, "step": 28170 }, { "epoch": 0.7423041589974416, "grad_norm": 2.4581003189086914, "learning_rate": 3.2860699690819407e-06, "loss": 0.1168547511100769, "step": 28180 }, { "epoch": 0.7425675742419403, "grad_norm": 2.3854124546051025, "learning_rate": 3.2797498001350737e-06, "loss": 0.11284445524215699, "step": 28190 }, { "epoch": 0.7428309894864391, "grad_norm": 1.8245744705200195, "learning_rate": 3.27343452247624e-06, "loss": 0.10759907960891724, "step": 28200 }, { "epoch": 0.7430944047309378, "grad_norm": 2.0488407611846924, "learning_rate": 3.267124140701974e-06, "loss": 0.121803617477417, "step": 28210 }, { "epoch": 0.7433578199754365, "grad_norm": 1.821044921875, "learning_rate": 3.2608186594052314e-06, "loss": 0.10325452089309692, "step": 28220 }, { "epoch": 0.7436212352199353, "grad_norm": 2.174988269805908, "learning_rate": 3.2545180831754264e-06, "loss": 0.11861910820007324, "step": 28230 }, { "epoch": 0.743884650464434, "grad_norm": 1.934433102607727, "learning_rate": 3.248222416598381e-06, "loss": 0.0983465075492859, "step": 28240 }, { "epoch": 0.7441480657089328, "grad_norm": 1.8446016311645508, "learning_rate": 3.2419316642563646e-06, "loss": 0.10245616436004638, "step": 28250 }, { "epoch": 0.7444114809534315, "grad_norm": 2.3913943767547607, "learning_rate": 3.235645830728047e-06, "loss": 0.11192939281463624, "step": 28260 }, { "epoch": 0.7446748961979303, "grad_norm": 1.8201042413711548, "learning_rate": 3.2293649205885293e-06, "loss": 0.10253340005874634, "step": 28270 }, { "epoch": 0.744938311442429, "grad_norm": 1.6862819194793701, "learning_rate": 3.2230889384093355e-06, "loss": 0.10039030313491822, "step": 28280 }, { "epoch": 0.7452017266869276, "grad_norm": 3.0110037326812744, "learning_rate": 3.2168178887583947e-06, "loss": 0.11244523525238037, "step": 28290 }, { "epoch": 0.7454651419314264, "grad_norm": 4.366178035736084, "learning_rate": 3.210551776200044e-06, "loss": 0.1043402075767517, "step": 28300 }, { "epoch": 0.7457285571759251, "grad_norm": 1.9458509683609009, "learning_rate": 3.204290605295033e-06, "loss": 0.11869488954544068, "step": 28310 }, { "epoch": 0.7459919724204239, "grad_norm": 2.419450521469116, "learning_rate": 3.1980343806005077e-06, "loss": 0.11234172582626342, "step": 28320 }, { "epoch": 0.7462553876649226, "grad_norm": 1.8831310272216797, "learning_rate": 3.1917831066700255e-06, "loss": 0.11907918453216552, "step": 28330 }, { "epoch": 0.7465188029094214, "grad_norm": 2.0886130332946777, "learning_rate": 3.1855367880535336e-06, "loss": 0.12036001682281494, "step": 28340 }, { "epoch": 0.7467822181539201, "grad_norm": 2.3052499294281006, "learning_rate": 3.179295429297361e-06, "loss": 0.11626696586608887, "step": 28350 }, { "epoch": 0.7470456333984189, "grad_norm": 2.2868764400482178, "learning_rate": 3.173059034944248e-06, "loss": 0.10985469818115234, "step": 28360 }, { "epoch": 0.7473090486429176, "grad_norm": 1.799796223640442, "learning_rate": 3.1668276095333063e-06, "loss": 0.11192812919616699, "step": 28370 }, { "epoch": 0.7475724638874163, "grad_norm": 1.5298250913619995, "learning_rate": 3.160601157600045e-06, "loss": 0.112221097946167, "step": 28380 }, { "epoch": 0.7478358791319151, "grad_norm": 2.3096742630004883, "learning_rate": 3.154379683676334e-06, "loss": 0.10016924142837524, "step": 28390 }, { "epoch": 0.7480992943764138, "grad_norm": 2.918179988861084, "learning_rate": 3.148163192290432e-06, "loss": 0.10966312885284424, "step": 28400 }, { "epoch": 0.7483627096209126, "grad_norm": 3.063699245452881, "learning_rate": 3.1419516879669744e-06, "loss": 0.1080512285232544, "step": 28410 }, { "epoch": 0.7486261248654112, "grad_norm": 1.7486010789871216, "learning_rate": 3.1357451752269595e-06, "loss": 0.11417565345764161, "step": 28420 }, { "epoch": 0.74888954010991, "grad_norm": 2.629196882247925, "learning_rate": 3.1295436585877558e-06, "loss": 0.11966291666030884, "step": 28430 }, { "epoch": 0.7491529553544087, "grad_norm": 5.940299987792969, "learning_rate": 3.1233471425630947e-06, "loss": 0.11218363046646118, "step": 28440 }, { "epoch": 0.7494163705989075, "grad_norm": 3.0338101387023926, "learning_rate": 3.1171556316630626e-06, "loss": 0.10392489433288574, "step": 28450 }, { "epoch": 0.7496797858434062, "grad_norm": 2.191357374191284, "learning_rate": 3.110969130394117e-06, "loss": 0.08895260095596313, "step": 28460 }, { "epoch": 0.7499432010879049, "grad_norm": 1.9295462369918823, "learning_rate": 3.1047876432590596e-06, "loss": 0.12493178844451905, "step": 28470 }, { "epoch": 0.7502066163324037, "grad_norm": 2.0774600505828857, "learning_rate": 3.0986111747570335e-06, "loss": 0.11657562255859374, "step": 28480 }, { "epoch": 0.7504700315769024, "grad_norm": 2.4527101516723633, "learning_rate": 3.092439729383548e-06, "loss": 0.12518390417098998, "step": 28490 }, { "epoch": 0.7507334468214012, "grad_norm": 2.03100848197937, "learning_rate": 3.0862733116304435e-06, "loss": 0.1062239646911621, "step": 28500 }, { "epoch": 0.7509968620658999, "grad_norm": 2.0509085655212402, "learning_rate": 3.0801119259859115e-06, "loss": 0.1240999698638916, "step": 28510 }, { "epoch": 0.7512602773103987, "grad_norm": 2.437329053878784, "learning_rate": 3.0739555769344666e-06, "loss": 0.094437575340271, "step": 28520 }, { "epoch": 0.7515236925548974, "grad_norm": 2.7150747776031494, "learning_rate": 3.067804268956964e-06, "loss": 0.11684627532958984, "step": 28530 }, { "epoch": 0.7517871077993962, "grad_norm": 2.655888080596924, "learning_rate": 3.0616580065305977e-06, "loss": 0.10669116973876953, "step": 28540 }, { "epoch": 0.7520505230438949, "grad_norm": 1.4863263368606567, "learning_rate": 3.055516794128881e-06, "loss": 0.11018584966659546, "step": 28550 }, { "epoch": 0.7523139382883935, "grad_norm": 4.040212631225586, "learning_rate": 3.0493806362216516e-06, "loss": 0.11176619529724122, "step": 28560 }, { "epoch": 0.7525773535328923, "grad_norm": 3.6729066371917725, "learning_rate": 3.0432495372750692e-06, "loss": 0.11143187284469605, "step": 28570 }, { "epoch": 0.752840768777391, "grad_norm": 2.9625368118286133, "learning_rate": 3.037123501751611e-06, "loss": 0.10632512569427491, "step": 28580 }, { "epoch": 0.7531041840218898, "grad_norm": 3.4083032608032227, "learning_rate": 3.0310025341100757e-06, "loss": 0.11496622562408447, "step": 28590 }, { "epoch": 0.7533675992663885, "grad_norm": 2.4270379543304443, "learning_rate": 3.0248866388055674e-06, "loss": 0.11495189666748047, "step": 28600 }, { "epoch": 0.7536310145108873, "grad_norm": 1.9388585090637207, "learning_rate": 3.018775820289488e-06, "loss": 0.09470059275627137, "step": 28610 }, { "epoch": 0.753894429755386, "grad_norm": 3.4608943462371826, "learning_rate": 3.0126700830095668e-06, "loss": 0.10709999799728394, "step": 28620 }, { "epoch": 0.7541578449998848, "grad_norm": 2.7801756858825684, "learning_rate": 3.006569431409817e-06, "loss": 0.1092991828918457, "step": 28630 }, { "epoch": 0.7544212602443835, "grad_norm": 2.9604616165161133, "learning_rate": 3.000473869930558e-06, "loss": 0.10041022300720215, "step": 28640 }, { "epoch": 0.7546846754888822, "grad_norm": 2.1126275062561035, "learning_rate": 2.9943834030084006e-06, "loss": 0.12602629661560058, "step": 28650 }, { "epoch": 0.754948090733381, "grad_norm": 2.1201694011688232, "learning_rate": 2.988298035076249e-06, "loss": 0.1051748514175415, "step": 28660 }, { "epoch": 0.7552115059778797, "grad_norm": 2.4495577812194824, "learning_rate": 2.9822177705633005e-06, "loss": 0.10804407596588135, "step": 28670 }, { "epoch": 0.7554749212223785, "grad_norm": 2.305839776992798, "learning_rate": 2.976142613895031e-06, "loss": 0.10474038124084473, "step": 28680 }, { "epoch": 0.7557383364668772, "grad_norm": 2.4812045097351074, "learning_rate": 2.970072569493202e-06, "loss": 0.09777460694313049, "step": 28690 }, { "epoch": 0.756001751711376, "grad_norm": 2.9392385482788086, "learning_rate": 2.9640076417758546e-06, "loss": 0.1145308256149292, "step": 28700 }, { "epoch": 0.7562651669558746, "grad_norm": 2.941255807876587, "learning_rate": 2.9579478351573e-06, "loss": 0.111748206615448, "step": 28710 }, { "epoch": 0.7565285822003734, "grad_norm": 2.208122730255127, "learning_rate": 2.9518931540481377e-06, "loss": 0.10748116970062256, "step": 28720 }, { "epoch": 0.7567919974448721, "grad_norm": 1.1691441535949707, "learning_rate": 2.945843602855215e-06, "loss": 0.11637694835662842, "step": 28730 }, { "epoch": 0.7570554126893708, "grad_norm": 2.0026702880859375, "learning_rate": 2.9397991859816588e-06, "loss": 0.12012244462966919, "step": 28740 }, { "epoch": 0.7573188279338696, "grad_norm": 2.5259876251220703, "learning_rate": 2.9337599078268597e-06, "loss": 0.10912539958953857, "step": 28750 }, { "epoch": 0.7575822431783683, "grad_norm": 1.8610137701034546, "learning_rate": 2.927725772786464e-06, "loss": 0.11586140394210816, "step": 28760 }, { "epoch": 0.7578456584228671, "grad_norm": 1.8599004745483398, "learning_rate": 2.9216967852523747e-06, "loss": 0.10304292440414428, "step": 28770 }, { "epoch": 0.7581090736673658, "grad_norm": 2.5852270126342773, "learning_rate": 2.915672949612749e-06, "loss": 0.1029130458831787, "step": 28780 }, { "epoch": 0.7583724889118646, "grad_norm": 1.9949932098388672, "learning_rate": 2.9096542702519904e-06, "loss": 0.11361037492752075, "step": 28790 }, { "epoch": 0.7586359041563633, "grad_norm": 1.9006965160369873, "learning_rate": 2.9036407515507613e-06, "loss": 0.10212111473083496, "step": 28800 }, { "epoch": 0.758899319400862, "grad_norm": 2.0842080116271973, "learning_rate": 2.89763239788596e-06, "loss": 0.10973999500274659, "step": 28810 }, { "epoch": 0.7591627346453608, "grad_norm": 2.781633138656616, "learning_rate": 2.8916292136307147e-06, "loss": 0.11176929473876954, "step": 28820 }, { "epoch": 0.7594261498898595, "grad_norm": 2.654951572418213, "learning_rate": 2.885631203154412e-06, "loss": 0.12253010272979736, "step": 28830 }, { "epoch": 0.7596895651343583, "grad_norm": 2.3956503868103027, "learning_rate": 2.879638370822656e-06, "loss": 0.11319206953048706, "step": 28840 }, { "epoch": 0.759952980378857, "grad_norm": 2.040252923965454, "learning_rate": 2.8736507209972996e-06, "loss": 0.110560941696167, "step": 28850 }, { "epoch": 0.7602163956233557, "grad_norm": 2.1058828830718994, "learning_rate": 2.867668258036401e-06, "loss": 0.11177741289138794, "step": 28860 }, { "epoch": 0.7604798108678544, "grad_norm": 1.802140235900879, "learning_rate": 2.8616909862942565e-06, "loss": 0.10962437391281128, "step": 28870 }, { "epoch": 0.7607432261123532, "grad_norm": 2.1303961277008057, "learning_rate": 2.855718910121389e-06, "loss": 0.09885604381561279, "step": 28880 }, { "epoch": 0.7610066413568519, "grad_norm": 2.1608150005340576, "learning_rate": 2.8497520338645278e-06, "loss": 0.09919415712356568, "step": 28890 }, { "epoch": 0.7612700566013506, "grad_norm": 2.107571840286255, "learning_rate": 2.843790361866625e-06, "loss": 0.1033104419708252, "step": 28900 }, { "epoch": 0.7615334718458494, "grad_norm": 1.6616021394729614, "learning_rate": 2.837833898466842e-06, "loss": 0.09789289236068725, "step": 28910 }, { "epoch": 0.7617968870903481, "grad_norm": 2.0636703968048096, "learning_rate": 2.831882648000547e-06, "loss": 0.10367958545684815, "step": 28920 }, { "epoch": 0.7620603023348469, "grad_norm": 2.708503246307373, "learning_rate": 2.8259366147993217e-06, "loss": 0.10623703002929688, "step": 28930 }, { "epoch": 0.7623237175793456, "grad_norm": 2.25864315032959, "learning_rate": 2.819995803190948e-06, "loss": 0.11252421140670776, "step": 28940 }, { "epoch": 0.7625871328238444, "grad_norm": 3.7977662086486816, "learning_rate": 2.8140602174993927e-06, "loss": 0.10609709024429322, "step": 28950 }, { "epoch": 0.7628505480683431, "grad_norm": 2.8023130893707275, "learning_rate": 2.8081298620448427e-06, "loss": 0.10747005939483642, "step": 28960 }, { "epoch": 0.7631139633128419, "grad_norm": 2.3266921043395996, "learning_rate": 2.8022047411436604e-06, "loss": 0.09551592469215393, "step": 28970 }, { "epoch": 0.7633773785573406, "grad_norm": 2.68698787689209, "learning_rate": 2.7962848591084122e-06, "loss": 0.1218674898147583, "step": 28980 }, { "epoch": 0.7636407938018392, "grad_norm": 1.9472417831420898, "learning_rate": 2.7903702202478355e-06, "loss": 0.09952818751335143, "step": 28990 }, { "epoch": 0.763904209046338, "grad_norm": 3.2644894123077393, "learning_rate": 2.7844608288668586e-06, "loss": 0.10078929662704468, "step": 29000 }, { "epoch": 0.7641676242908367, "grad_norm": 2.2763655185699463, "learning_rate": 2.7785566892665993e-06, "loss": 0.11516474485397339, "step": 29010 }, { "epoch": 0.7644310395353355, "grad_norm": 2.7491471767425537, "learning_rate": 2.77265780574434e-06, "loss": 0.1122936487197876, "step": 29020 }, { "epoch": 0.7646944547798342, "grad_norm": 1.966784954071045, "learning_rate": 2.766764182593544e-06, "loss": 0.11118618249893189, "step": 29030 }, { "epoch": 0.764957870024333, "grad_norm": 1.8083670139312744, "learning_rate": 2.760875824103846e-06, "loss": 0.10420331954956055, "step": 29040 }, { "epoch": 0.7652212852688317, "grad_norm": 2.9980568885803223, "learning_rate": 2.7549927345610407e-06, "loss": 0.10171328783035279, "step": 29050 }, { "epoch": 0.7654847005133305, "grad_norm": 2.1054415702819824, "learning_rate": 2.749114918247102e-06, "loss": 0.10313243865966797, "step": 29060 }, { "epoch": 0.7657481157578292, "grad_norm": 2.0661847591400146, "learning_rate": 2.7432423794401598e-06, "loss": 0.10174545049667358, "step": 29070 }, { "epoch": 0.7660115310023279, "grad_norm": 2.587684154510498, "learning_rate": 2.7373751224144905e-06, "loss": 0.10846958160400391, "step": 29080 }, { "epoch": 0.7662749462468267, "grad_norm": 2.1495285034179688, "learning_rate": 2.731513151440547e-06, "loss": 0.10967214107513427, "step": 29090 }, { "epoch": 0.7665383614913254, "grad_norm": 1.5435657501220703, "learning_rate": 2.7256564707849165e-06, "loss": 0.10946977138519287, "step": 29100 }, { "epoch": 0.7668017767358242, "grad_norm": 2.6381399631500244, "learning_rate": 2.719805084710356e-06, "loss": 0.11988117694854736, "step": 29110 }, { "epoch": 0.7670651919803229, "grad_norm": 1.927667260169983, "learning_rate": 2.7139589974757454e-06, "loss": 0.09670146703720092, "step": 29120 }, { "epoch": 0.7673286072248217, "grad_norm": 2.642045736312866, "learning_rate": 2.7081182133361207e-06, "loss": 0.10289264917373657, "step": 29130 }, { "epoch": 0.7675920224693203, "grad_norm": 2.664515972137451, "learning_rate": 2.7022827365426606e-06, "loss": 0.11731024980545043, "step": 29140 }, { "epoch": 0.7678554377138191, "grad_norm": 2.2677745819091797, "learning_rate": 2.696452571342676e-06, "loss": 0.11391515731811523, "step": 29150 }, { "epoch": 0.7681188529583178, "grad_norm": 2.0739612579345703, "learning_rate": 2.6906277219796116e-06, "loss": 0.09384946823120117, "step": 29160 }, { "epoch": 0.7683822682028165, "grad_norm": 1.890362024307251, "learning_rate": 2.6848081926930447e-06, "loss": 0.08897603750228882, "step": 29170 }, { "epoch": 0.7686456834473153, "grad_norm": 1.8974030017852783, "learning_rate": 2.6789939877186753e-06, "loss": 0.11445156335830689, "step": 29180 }, { "epoch": 0.768909098691814, "grad_norm": 1.9472665786743164, "learning_rate": 2.67318511128834e-06, "loss": 0.10308734178543091, "step": 29190 }, { "epoch": 0.7691725139363128, "grad_norm": 3.3625130653381348, "learning_rate": 2.66738156762999e-06, "loss": 0.09326214790344238, "step": 29200 }, { "epoch": 0.7694359291808115, "grad_norm": 3.25256609916687, "learning_rate": 2.6615833609676846e-06, "loss": 0.10976014137268067, "step": 29210 }, { "epoch": 0.7696993444253103, "grad_norm": 1.887880563735962, "learning_rate": 2.6557904955216185e-06, "loss": 0.11623885631561279, "step": 29220 }, { "epoch": 0.769962759669809, "grad_norm": 2.076997995376587, "learning_rate": 2.650002975508086e-06, "loss": 0.10048134326934814, "step": 29230 }, { "epoch": 0.7702261749143078, "grad_norm": 2.6705241203308105, "learning_rate": 2.6442208051394936e-06, "loss": 0.1024774432182312, "step": 29240 }, { "epoch": 0.7704895901588065, "grad_norm": 2.510833740234375, "learning_rate": 2.638443988624355e-06, "loss": 0.10102903842926025, "step": 29250 }, { "epoch": 0.7707530054033052, "grad_norm": 2.0221476554870605, "learning_rate": 2.632672530167283e-06, "loss": 0.09636545181274414, "step": 29260 }, { "epoch": 0.771016420647804, "grad_norm": 1.906801700592041, "learning_rate": 2.6269064339689986e-06, "loss": 0.10308660268783569, "step": 29270 }, { "epoch": 0.7712798358923026, "grad_norm": 1.456743597984314, "learning_rate": 2.6211457042263157e-06, "loss": 0.10821094512939453, "step": 29280 }, { "epoch": 0.7715432511368014, "grad_norm": 1.9789726734161377, "learning_rate": 2.6153903451321384e-06, "loss": 0.10610942840576172, "step": 29290 }, { "epoch": 0.7718066663813001, "grad_norm": 2.14784574508667, "learning_rate": 2.6096403608754685e-06, "loss": 0.09925751686096192, "step": 29300 }, { "epoch": 0.7720700816257989, "grad_norm": 2.426790714263916, "learning_rate": 2.603895755641387e-06, "loss": 0.11734824180603028, "step": 29310 }, { "epoch": 0.7723334968702976, "grad_norm": 1.6620110273361206, "learning_rate": 2.598156533611077e-06, "loss": 0.10316600799560546, "step": 29320 }, { "epoch": 0.7725969121147963, "grad_norm": 3.411316394805908, "learning_rate": 2.592422698961781e-06, "loss": 0.1020744800567627, "step": 29330 }, { "epoch": 0.7728603273592951, "grad_norm": 2.1453964710235596, "learning_rate": 2.586694255866831e-06, "loss": 0.11895132064819336, "step": 29340 }, { "epoch": 0.7731237426037938, "grad_norm": 2.2510263919830322, "learning_rate": 2.580971208495642e-06, "loss": 0.09370179772377014, "step": 29350 }, { "epoch": 0.7733871578482926, "grad_norm": 3.6808154582977295, "learning_rate": 2.575253561013692e-06, "loss": 0.10618710517883301, "step": 29360 }, { "epoch": 0.7736505730927913, "grad_norm": 2.545285224914551, "learning_rate": 2.56954131758253e-06, "loss": 0.09490477442741393, "step": 29370 }, { "epoch": 0.7739139883372901, "grad_norm": 2.696857452392578, "learning_rate": 2.5638344823597762e-06, "loss": 0.10694308280944824, "step": 29380 }, { "epoch": 0.7741774035817888, "grad_norm": 2.344113349914551, "learning_rate": 2.558133059499105e-06, "loss": 0.10728566646575928, "step": 29390 }, { "epoch": 0.7744408188262876, "grad_norm": 3.4485204219818115, "learning_rate": 2.552437053150266e-06, "loss": 0.1030170202255249, "step": 29400 }, { "epoch": 0.7747042340707863, "grad_norm": 1.9138386249542236, "learning_rate": 2.546746467459058e-06, "loss": 0.1031941294670105, "step": 29410 }, { "epoch": 0.774967649315285, "grad_norm": 3.5698044300079346, "learning_rate": 2.541061306567325e-06, "loss": 0.11814848184585572, "step": 29420 }, { "epoch": 0.7752310645597837, "grad_norm": 2.2780137062072754, "learning_rate": 2.5353815746129827e-06, "loss": 0.0996745467185974, "step": 29430 }, { "epoch": 0.7754944798042824, "grad_norm": 2.4986331462860107, "learning_rate": 2.529707275729979e-06, "loss": 0.10389593839645386, "step": 29440 }, { "epoch": 0.7757578950487812, "grad_norm": 3.412259340286255, "learning_rate": 2.524038414048321e-06, "loss": 0.11185711622238159, "step": 29450 }, { "epoch": 0.7760213102932799, "grad_norm": 2.4932377338409424, "learning_rate": 2.518374993694043e-06, "loss": 0.10769407749176026, "step": 29460 }, { "epoch": 0.7762847255377787, "grad_norm": 2.876797914505005, "learning_rate": 2.5127170187892257e-06, "loss": 0.10841538906097412, "step": 29470 }, { "epoch": 0.7765481407822774, "grad_norm": 3.107790231704712, "learning_rate": 2.507064493451995e-06, "loss": 0.10020568370819091, "step": 29480 }, { "epoch": 0.7768115560267762, "grad_norm": 1.7113394737243652, "learning_rate": 2.501417421796498e-06, "loss": 0.1151168704032898, "step": 29490 }, { "epoch": 0.7770749712712749, "grad_norm": 2.5878920555114746, "learning_rate": 2.4957758079329173e-06, "loss": 0.10790036916732788, "step": 29500 }, { "epoch": 0.7773383865157736, "grad_norm": 1.6686619520187378, "learning_rate": 2.490139655967465e-06, "loss": 0.10620238780975341, "step": 29510 }, { "epoch": 0.7776018017602724, "grad_norm": 1.5474406480789185, "learning_rate": 2.4845089700023694e-06, "loss": 0.10233670473098755, "step": 29520 }, { "epoch": 0.7778652170047711, "grad_norm": 1.7807796001434326, "learning_rate": 2.4788837541358957e-06, "loss": 0.09680000543594361, "step": 29530 }, { "epoch": 0.7781286322492699, "grad_norm": 2.439209222793579, "learning_rate": 2.4732640124623152e-06, "loss": 0.11033328771591186, "step": 29540 }, { "epoch": 0.7783920474937686, "grad_norm": 1.9894077777862549, "learning_rate": 2.467649749071913e-06, "loss": 0.10527840852737427, "step": 29550 }, { "epoch": 0.7786554627382674, "grad_norm": 1.626600980758667, "learning_rate": 2.4620409680509993e-06, "loss": 0.10600479841232299, "step": 29560 }, { "epoch": 0.778918877982766, "grad_norm": 2.2352404594421387, "learning_rate": 2.4564376734818818e-06, "loss": 0.11401625871658325, "step": 29570 }, { "epoch": 0.7791822932272648, "grad_norm": 1.8672168254852295, "learning_rate": 2.450839869442887e-06, "loss": 0.09712263941764832, "step": 29580 }, { "epoch": 0.7794457084717635, "grad_norm": 2.3911592960357666, "learning_rate": 2.4452475600083323e-06, "loss": 0.09694463014602661, "step": 29590 }, { "epoch": 0.7797091237162622, "grad_norm": 2.405144214630127, "learning_rate": 2.4396607492485393e-06, "loss": 0.09990288019180298, "step": 29600 }, { "epoch": 0.779972538960761, "grad_norm": 2.1722824573516846, "learning_rate": 2.4340794412298364e-06, "loss": 0.11551826000213623, "step": 29610 }, { "epoch": 0.7802359542052597, "grad_norm": 2.025461435317993, "learning_rate": 2.428503640014539e-06, "loss": 0.10341572761535645, "step": 29620 }, { "epoch": 0.7804993694497585, "grad_norm": 1.729061245918274, "learning_rate": 2.4229333496609532e-06, "loss": 0.10416191816329956, "step": 29630 }, { "epoch": 0.7807627846942572, "grad_norm": 3.175477981567383, "learning_rate": 2.4173685742233787e-06, "loss": 0.1067805528640747, "step": 29640 }, { "epoch": 0.781026199938756, "grad_norm": 2.339712142944336, "learning_rate": 2.411809317752095e-06, "loss": 0.09814120531082153, "step": 29650 }, { "epoch": 0.7812896151832547, "grad_norm": 1.8640772104263306, "learning_rate": 2.4062555842933753e-06, "loss": 0.09742879867553711, "step": 29660 }, { "epoch": 0.7815530304277535, "grad_norm": 2.2506027221679688, "learning_rate": 2.4007073778894665e-06, "loss": 0.09568822979927064, "step": 29670 }, { "epoch": 0.7818164456722522, "grad_norm": 1.8795503377914429, "learning_rate": 2.395164702578583e-06, "loss": 0.10520708560943604, "step": 29680 }, { "epoch": 0.7820798609167509, "grad_norm": 3.554860830307007, "learning_rate": 2.3896275623949327e-06, "loss": 0.10897302627563477, "step": 29690 }, { "epoch": 0.7823432761612497, "grad_norm": 2.409065008163452, "learning_rate": 2.38409596136868e-06, "loss": 0.10731174945831298, "step": 29700 }, { "epoch": 0.7826066914057483, "grad_norm": 2.676300287246704, "learning_rate": 2.378569903525969e-06, "loss": 0.10726311206817626, "step": 29710 }, { "epoch": 0.7828701066502471, "grad_norm": 2.7520251274108887, "learning_rate": 2.3730493928888963e-06, "loss": 0.11374821662902831, "step": 29720 }, { "epoch": 0.7831335218947458, "grad_norm": 2.0839004516601562, "learning_rate": 2.367534433475527e-06, "loss": 0.10298928022384643, "step": 29730 }, { "epoch": 0.7833969371392446, "grad_norm": 2.532116651535034, "learning_rate": 2.3620250292998925e-06, "loss": 0.10493221282958984, "step": 29740 }, { "epoch": 0.7836603523837433, "grad_norm": 1.8307257890701294, "learning_rate": 2.356521184371973e-06, "loss": 0.11044970750808716, "step": 29750 }, { "epoch": 0.7839237676282421, "grad_norm": 1.9994794130325317, "learning_rate": 2.3510229026977018e-06, "loss": 0.10254783630371093, "step": 29760 }, { "epoch": 0.7841871828727408, "grad_norm": 3.294605016708374, "learning_rate": 2.345530188278967e-06, "loss": 0.10689717531204224, "step": 29770 }, { "epoch": 0.7844505981172395, "grad_norm": 1.650324821472168, "learning_rate": 2.3400430451136e-06, "loss": 0.10052685737609864, "step": 29780 }, { "epoch": 0.7847140133617383, "grad_norm": 3.1220364570617676, "learning_rate": 2.334561477195387e-06, "loss": 0.08753528594970703, "step": 29790 }, { "epoch": 0.784977428606237, "grad_norm": 2.0195157527923584, "learning_rate": 2.329085488514049e-06, "loss": 0.0983591377735138, "step": 29800 }, { "epoch": 0.7852408438507358, "grad_norm": 2.687804698944092, "learning_rate": 2.3236150830552374e-06, "loss": 0.09594975113868713, "step": 29810 }, { "epoch": 0.7855042590952345, "grad_norm": 2.6733813285827637, "learning_rate": 2.3181502648005593e-06, "loss": 0.10559115409851075, "step": 29820 }, { "epoch": 0.7857676743397333, "grad_norm": 2.8952486515045166, "learning_rate": 2.3126910377275415e-06, "loss": 0.09562314748764038, "step": 29830 }, { "epoch": 0.786031089584232, "grad_norm": 3.5799081325531006, "learning_rate": 2.3072374058096447e-06, "loss": 0.12148751020431518, "step": 29840 }, { "epoch": 0.7862945048287308, "grad_norm": 2.018617630004883, "learning_rate": 2.301789373016259e-06, "loss": 0.10653045177459716, "step": 29850 }, { "epoch": 0.7865579200732294, "grad_norm": 2.7650885581970215, "learning_rate": 2.296346943312693e-06, "loss": 0.1057275414466858, "step": 29860 }, { "epoch": 0.7868213353177281, "grad_norm": 2.871429443359375, "learning_rate": 2.290910120660188e-06, "loss": 0.1032137393951416, "step": 29870 }, { "epoch": 0.7870847505622269, "grad_norm": 2.1804351806640625, "learning_rate": 2.285478909015896e-06, "loss": 0.09595445990562439, "step": 29880 }, { "epoch": 0.7873481658067256, "grad_norm": 3.158144950866699, "learning_rate": 2.2800533123328874e-06, "loss": 0.11742987632751464, "step": 29890 }, { "epoch": 0.7876115810512244, "grad_norm": 3.3463327884674072, "learning_rate": 2.2746333345601435e-06, "loss": 0.10317659378051758, "step": 29900 }, { "epoch": 0.7878749962957231, "grad_norm": 2.003300189971924, "learning_rate": 2.269218979642558e-06, "loss": 0.09583626389503479, "step": 29910 }, { "epoch": 0.7881384115402219, "grad_norm": 2.6672580242156982, "learning_rate": 2.2638102515209382e-06, "loss": 0.10476980209350586, "step": 29920 }, { "epoch": 0.7884018267847206, "grad_norm": 3.622947931289673, "learning_rate": 2.258407154131982e-06, "loss": 0.09686467647552491, "step": 29930 }, { "epoch": 0.7886652420292193, "grad_norm": 2.957420587539673, "learning_rate": 2.2530096914082965e-06, "loss": 0.11391704082489014, "step": 29940 }, { "epoch": 0.7889286572737181, "grad_norm": 2.2576959133148193, "learning_rate": 2.2476178672783943e-06, "loss": 0.10042439699172974, "step": 29950 }, { "epoch": 0.7891920725182168, "grad_norm": 2.8681156635284424, "learning_rate": 2.242231685666675e-06, "loss": 0.0952836036682129, "step": 29960 }, { "epoch": 0.7894554877627156, "grad_norm": 2.990746259689331, "learning_rate": 2.2368511504934333e-06, "loss": 0.10713133811950684, "step": 29970 }, { "epoch": 0.7897189030072143, "grad_norm": 2.3730616569519043, "learning_rate": 2.2314762656748544e-06, "loss": 0.09671033024787903, "step": 29980 }, { "epoch": 0.7899823182517131, "grad_norm": 1.754572868347168, "learning_rate": 2.2261070351230107e-06, "loss": 0.10394408702850341, "step": 29990 }, { "epoch": 0.7902457334962117, "grad_norm": 2.598315954208374, "learning_rate": 2.2207434627458626e-06, "loss": 0.11047551631927491, "step": 30000 }, { "epoch": 0.7905091487407105, "grad_norm": 2.4766337871551514, "learning_rate": 2.2153855524472513e-06, "loss": 0.11273036003112794, "step": 30010 }, { "epoch": 0.7907725639852092, "grad_norm": 1.3902301788330078, "learning_rate": 2.2100333081268853e-06, "loss": 0.09790174961090088, "step": 30020 }, { "epoch": 0.7910359792297079, "grad_norm": 2.983232259750366, "learning_rate": 2.204686733680369e-06, "loss": 0.10025057792663575, "step": 30030 }, { "epoch": 0.7912993944742067, "grad_norm": 2.5917179584503174, "learning_rate": 2.1993458329991635e-06, "loss": 0.09981014132499695, "step": 30040 }, { "epoch": 0.7915628097187054, "grad_norm": 1.799491286277771, "learning_rate": 2.1940106099706137e-06, "loss": 0.10768027305603027, "step": 30050 }, { "epoch": 0.7918262249632042, "grad_norm": 1.9192618131637573, "learning_rate": 2.1886810684779204e-06, "loss": 0.09966784119606018, "step": 30060 }, { "epoch": 0.7920896402077029, "grad_norm": 2.83858323097229, "learning_rate": 2.18335721240015e-06, "loss": 0.10254143476486206, "step": 30070 }, { "epoch": 0.7923530554522017, "grad_norm": 2.5396580696105957, "learning_rate": 2.1780390456122424e-06, "loss": 0.0939185619354248, "step": 30080 }, { "epoch": 0.7926164706967004, "grad_norm": 2.4026529788970947, "learning_rate": 2.1727265719849853e-06, "loss": 0.11150778532028198, "step": 30090 }, { "epoch": 0.7928798859411992, "grad_norm": 1.9991713762283325, "learning_rate": 2.1674197953850274e-06, "loss": 0.10021584033966065, "step": 30100 }, { "epoch": 0.7931433011856979, "grad_norm": 3.1983938217163086, "learning_rate": 2.162118719674867e-06, "loss": 0.11521103382110595, "step": 30110 }, { "epoch": 0.7934067164301966, "grad_norm": 2.7197110652923584, "learning_rate": 2.1568233487128556e-06, "loss": 0.09796382188796997, "step": 30120 }, { "epoch": 0.7936701316746954, "grad_norm": 3.315239906311035, "learning_rate": 2.1515336863531976e-06, "loss": 0.10798370838165283, "step": 30130 }, { "epoch": 0.793933546919194, "grad_norm": 2.3806374073028564, "learning_rate": 2.1462497364459356e-06, "loss": 0.10321238040924072, "step": 30140 }, { "epoch": 0.7941969621636928, "grad_norm": 2.744764566421509, "learning_rate": 2.1409715028369504e-06, "loss": 0.09857550859451295, "step": 30150 }, { "epoch": 0.7944603774081915, "grad_norm": 2.2701454162597656, "learning_rate": 2.135698989367976e-06, "loss": 0.10739431381225586, "step": 30160 }, { "epoch": 0.7947237926526903, "grad_norm": 2.2036097049713135, "learning_rate": 2.1304321998765685e-06, "loss": 0.09470166563987732, "step": 30170 }, { "epoch": 0.794987207897189, "grad_norm": 1.6872278451919556, "learning_rate": 2.1251711381961336e-06, "loss": 0.1019989013671875, "step": 30180 }, { "epoch": 0.7952506231416878, "grad_norm": 3.0468897819519043, "learning_rate": 2.11991580815589e-06, "loss": 0.10629191398620605, "step": 30190 }, { "epoch": 0.7955140383861865, "grad_norm": 2.2825138568878174, "learning_rate": 2.1146662135808948e-06, "loss": 0.10543032884597778, "step": 30200 }, { "epoch": 0.7957774536306852, "grad_norm": 2.5325489044189453, "learning_rate": 2.1094223582920346e-06, "loss": 0.10275403261184693, "step": 30210 }, { "epoch": 0.796040868875184, "grad_norm": 2.544844388961792, "learning_rate": 2.1041842461060103e-06, "loss": 0.10417187213897705, "step": 30220 }, { "epoch": 0.7963042841196827, "grad_norm": 1.7205675840377808, "learning_rate": 2.098951880835346e-06, "loss": 0.10224171876907348, "step": 30230 }, { "epoch": 0.7965676993641815, "grad_norm": 2.217451333999634, "learning_rate": 2.093725266288382e-06, "loss": 0.10100599527359008, "step": 30240 }, { "epoch": 0.7968311146086802, "grad_norm": 2.4816396236419678, "learning_rate": 2.088504406269274e-06, "loss": 0.10541706085205078, "step": 30250 }, { "epoch": 0.797094529853179, "grad_norm": 2.990586280822754, "learning_rate": 2.083289304577992e-06, "loss": 0.10027196407318115, "step": 30260 }, { "epoch": 0.7973579450976777, "grad_norm": 2.730515241622925, "learning_rate": 2.0780799650103123e-06, "loss": 0.08981869220733643, "step": 30270 }, { "epoch": 0.7976213603421765, "grad_norm": 3.53499698638916, "learning_rate": 2.0728763913578097e-06, "loss": 0.10352301597595215, "step": 30280 }, { "epoch": 0.7978847755866751, "grad_norm": 2.1412360668182373, "learning_rate": 2.0676785874078773e-06, "loss": 0.09428067207336426, "step": 30290 }, { "epoch": 0.7981481908311738, "grad_norm": 1.9140503406524658, "learning_rate": 2.062486556943699e-06, "loss": 0.10483880043029785, "step": 30300 }, { "epoch": 0.7984116060756726, "grad_norm": 2.1938445568084717, "learning_rate": 2.0573003037442575e-06, "loss": 0.10821337699890136, "step": 30310 }, { "epoch": 0.7986750213201713, "grad_norm": 2.1825971603393555, "learning_rate": 2.0521198315843315e-06, "loss": 0.10174412727355957, "step": 30320 }, { "epoch": 0.7989384365646701, "grad_norm": 2.7745490074157715, "learning_rate": 2.046945144234491e-06, "loss": 0.10233132839202881, "step": 30330 }, { "epoch": 0.7992018518091688, "grad_norm": 2.4307539463043213, "learning_rate": 2.0417762454611012e-06, "loss": 0.12246164083480834, "step": 30340 }, { "epoch": 0.7994652670536676, "grad_norm": 3.152679681777954, "learning_rate": 2.0366131390263055e-06, "loss": 0.10057950019836426, "step": 30350 }, { "epoch": 0.7997286822981663, "grad_norm": 2.119067907333374, "learning_rate": 2.0314558286880383e-06, "loss": 0.10017504692077636, "step": 30360 }, { "epoch": 0.7999920975426651, "grad_norm": 3.342468023300171, "learning_rate": 2.026304318200011e-06, "loss": 0.11534726619720459, "step": 30370 }, { "epoch": 0.8002555127871638, "grad_norm": 1.961859941482544, "learning_rate": 2.021158611311712e-06, "loss": 0.09874584674835205, "step": 30380 }, { "epoch": 0.8005189280316625, "grad_norm": 2.1499791145324707, "learning_rate": 2.016018711768416e-06, "loss": 0.10974717140197754, "step": 30390 }, { "epoch": 0.8007823432761613, "grad_norm": 2.65923810005188, "learning_rate": 2.010884623311162e-06, "loss": 0.1065971851348877, "step": 30400 }, { "epoch": 0.80104575852066, "grad_norm": 2.8642849922180176, "learning_rate": 2.005756349676753e-06, "loss": 0.10228749513626098, "step": 30410 }, { "epoch": 0.8013091737651588, "grad_norm": 2.3196537494659424, "learning_rate": 2.0006338945977777e-06, "loss": 0.09260786771774292, "step": 30420 }, { "epoch": 0.8015725890096574, "grad_norm": 2.2935287952423096, "learning_rate": 1.9955172618025743e-06, "loss": 0.10210418701171875, "step": 30430 }, { "epoch": 0.8018360042541562, "grad_norm": 2.4085946083068848, "learning_rate": 1.9904064550152515e-06, "loss": 0.08897571563720703, "step": 30440 }, { "epoch": 0.8020994194986549, "grad_norm": 2.455538511276245, "learning_rate": 1.9853014779556734e-06, "loss": 0.09223738312721252, "step": 30450 }, { "epoch": 0.8023628347431536, "grad_norm": 2.2580509185791016, "learning_rate": 1.9802023343394606e-06, "loss": 0.09957746267318726, "step": 30460 }, { "epoch": 0.8026262499876524, "grad_norm": 2.6555252075195312, "learning_rate": 1.975109027877995e-06, "loss": 0.1018826723098755, "step": 30470 }, { "epoch": 0.8028896652321511, "grad_norm": 2.7593448162078857, "learning_rate": 1.9700215622784026e-06, "loss": 0.0814438819885254, "step": 30480 }, { "epoch": 0.8031530804766499, "grad_norm": 4.487726211547852, "learning_rate": 1.96493994124356e-06, "loss": 0.100040864944458, "step": 30490 }, { "epoch": 0.8034164957211486, "grad_norm": 2.2913739681243896, "learning_rate": 1.9598641684720897e-06, "loss": 0.09053912162780761, "step": 30500 }, { "epoch": 0.8036799109656474, "grad_norm": 2.4998650550842285, "learning_rate": 1.954794247658356e-06, "loss": 0.10853469371795654, "step": 30510 }, { "epoch": 0.8039433262101461, "grad_norm": 2.8899788856506348, "learning_rate": 1.949730182492473e-06, "loss": 0.09232977628707886, "step": 30520 }, { "epoch": 0.8042067414546449, "grad_norm": 3.375314712524414, "learning_rate": 1.9446719766602793e-06, "loss": 0.10147377252578735, "step": 30530 }, { "epoch": 0.8044701566991436, "grad_norm": 3.7679643630981445, "learning_rate": 1.9396196338433547e-06, "loss": 0.09945218563079834, "step": 30540 }, { "epoch": 0.8047335719436423, "grad_norm": 2.2743566036224365, "learning_rate": 1.9345731577190187e-06, "loss": 0.10181639194488526, "step": 30550 }, { "epoch": 0.804996987188141, "grad_norm": 2.8049795627593994, "learning_rate": 1.929532551960309e-06, "loss": 0.10129845142364502, "step": 30560 }, { "epoch": 0.8052604024326397, "grad_norm": 1.981246829032898, "learning_rate": 1.9244978202359974e-06, "loss": 0.08503206372261048, "step": 30570 }, { "epoch": 0.8055238176771385, "grad_norm": 2.173642635345459, "learning_rate": 1.9194689662105803e-06, "loss": 0.09259320497512817, "step": 30580 }, { "epoch": 0.8057872329216372, "grad_norm": 2.8909363746643066, "learning_rate": 1.9144459935442705e-06, "loss": 0.10425819158554077, "step": 30590 }, { "epoch": 0.806050648166136, "grad_norm": 2.0159754753112793, "learning_rate": 1.9094289058930095e-06, "loss": 0.09913983345031738, "step": 30600 }, { "epoch": 0.8063140634106347, "grad_norm": 2.077667236328125, "learning_rate": 1.9044177069084502e-06, "loss": 0.0928848147392273, "step": 30610 }, { "epoch": 0.8065774786551335, "grad_norm": 2.3188271522521973, "learning_rate": 1.899412400237951e-06, "loss": 0.09917932748794556, "step": 30620 }, { "epoch": 0.8068408938996322, "grad_norm": 2.4118943214416504, "learning_rate": 1.8944129895246e-06, "loss": 0.10388917922973633, "step": 30630 }, { "epoch": 0.8071043091441309, "grad_norm": 2.0874223709106445, "learning_rate": 1.8894194784071761e-06, "loss": 0.10086203813552856, "step": 30640 }, { "epoch": 0.8073677243886297, "grad_norm": 3.487668037414551, "learning_rate": 1.8844318705201815e-06, "loss": 0.09922369122505188, "step": 30650 }, { "epoch": 0.8076311396331284, "grad_norm": 2.9915502071380615, "learning_rate": 1.879450169493805e-06, "loss": 0.10534589290618897, "step": 30660 }, { "epoch": 0.8078945548776272, "grad_norm": 2.4788713455200195, "learning_rate": 1.874474378953942e-06, "loss": 0.10683077573776245, "step": 30670 }, { "epoch": 0.8081579701221259, "grad_norm": 3.5770530700683594, "learning_rate": 1.8695045025221936e-06, "loss": 0.09094051122665406, "step": 30680 }, { "epoch": 0.8084213853666247, "grad_norm": 2.626495361328125, "learning_rate": 1.864540543815848e-06, "loss": 0.09337626695632935, "step": 30690 }, { "epoch": 0.8086848006111234, "grad_norm": 3.5213406085968018, "learning_rate": 1.859582506447889e-06, "loss": 0.10673511028289795, "step": 30700 }, { "epoch": 0.8089482158556222, "grad_norm": 2.133284568786621, "learning_rate": 1.8546303940269883e-06, "loss": 0.09177452325820923, "step": 30710 }, { "epoch": 0.8092116311001208, "grad_norm": 2.905601739883423, "learning_rate": 1.8496842101575053e-06, "loss": 0.10023965835571289, "step": 30720 }, { "epoch": 0.8094750463446195, "grad_norm": 3.1556615829467773, "learning_rate": 1.844743958439491e-06, "loss": 0.10534753799438476, "step": 30730 }, { "epoch": 0.8097384615891183, "grad_norm": 3.252026081085205, "learning_rate": 1.839809642468674e-06, "loss": 0.10105767250061035, "step": 30740 }, { "epoch": 0.810001876833617, "grad_norm": 1.716899037361145, "learning_rate": 1.8348812658364557e-06, "loss": 0.10913071632385254, "step": 30750 }, { "epoch": 0.8102652920781158, "grad_norm": 1.929703712463379, "learning_rate": 1.8299588321299266e-06, "loss": 0.09578953981399536, "step": 30760 }, { "epoch": 0.8105287073226145, "grad_norm": 2.655484914779663, "learning_rate": 1.8250423449318434e-06, "loss": 0.09763771891593934, "step": 30770 }, { "epoch": 0.8107921225671133, "grad_norm": 2.5049266815185547, "learning_rate": 1.8201318078206443e-06, "loss": 0.09764060974121094, "step": 30780 }, { "epoch": 0.811055537811612, "grad_norm": 2.184913158416748, "learning_rate": 1.8152272243704216e-06, "loss": 0.11237150430679321, "step": 30790 }, { "epoch": 0.8113189530561108, "grad_norm": 2.5277106761932373, "learning_rate": 1.8103285981509432e-06, "loss": 0.09923526048660278, "step": 30800 }, { "epoch": 0.8115823683006095, "grad_norm": 2.203376054763794, "learning_rate": 1.8054359327276473e-06, "loss": 0.09720472097396851, "step": 30810 }, { "epoch": 0.8118457835451082, "grad_norm": 2.2958104610443115, "learning_rate": 1.8005492316616224e-06, "loss": 0.09997859597206116, "step": 30820 }, { "epoch": 0.812109198789607, "grad_norm": 1.8857488632202148, "learning_rate": 1.7956684985096217e-06, "loss": 0.08231939673423767, "step": 30830 }, { "epoch": 0.8123726140341057, "grad_norm": 2.3872406482696533, "learning_rate": 1.7907937368240525e-06, "loss": 0.09803215861320495, "step": 30840 }, { "epoch": 0.8126360292786045, "grad_norm": 3.496072292327881, "learning_rate": 1.7859249501529742e-06, "loss": 0.1041451096534729, "step": 30850 }, { "epoch": 0.8128994445231031, "grad_norm": 3.124115467071533, "learning_rate": 1.7810621420401054e-06, "loss": 0.0949807584285736, "step": 30860 }, { "epoch": 0.8131628597676019, "grad_norm": 1.708006501197815, "learning_rate": 1.776205316024806e-06, "loss": 0.09153648018836975, "step": 30870 }, { "epoch": 0.8134262750121006, "grad_norm": 1.978276014328003, "learning_rate": 1.7713544756420785e-06, "loss": 0.09005266427993774, "step": 30880 }, { "epoch": 0.8136896902565994, "grad_norm": 2.6228015422821045, "learning_rate": 1.76650962442258e-06, "loss": 0.09742398858070374, "step": 30890 }, { "epoch": 0.8139531055010981, "grad_norm": 3.08132266998291, "learning_rate": 1.761670765892599e-06, "loss": 0.09496533870697021, "step": 30900 }, { "epoch": 0.8142165207455968, "grad_norm": 2.2043445110321045, "learning_rate": 1.7568379035740669e-06, "loss": 0.09916487932205201, "step": 30910 }, { "epoch": 0.8144799359900956, "grad_norm": 2.7649190425872803, "learning_rate": 1.7520110409845493e-06, "loss": 0.09918660521507264, "step": 30920 }, { "epoch": 0.8147433512345943, "grad_norm": 4.495147228240967, "learning_rate": 1.7471901816372416e-06, "loss": 0.09559204578399658, "step": 30930 }, { "epoch": 0.8150067664790931, "grad_norm": 2.445307731628418, "learning_rate": 1.7423753290409784e-06, "loss": 0.09825032949447632, "step": 30940 }, { "epoch": 0.8152701817235918, "grad_norm": 3.459074020385742, "learning_rate": 1.7375664867002151e-06, "loss": 0.10367021560668946, "step": 30950 }, { "epoch": 0.8155335969680906, "grad_norm": 1.8238872289657593, "learning_rate": 1.7327636581150343e-06, "loss": 0.10033223628997803, "step": 30960 }, { "epoch": 0.8157970122125893, "grad_norm": 3.3433656692504883, "learning_rate": 1.7279668467811427e-06, "loss": 0.101012921333313, "step": 30970 }, { "epoch": 0.816060427457088, "grad_norm": 2.7926526069641113, "learning_rate": 1.7231760561898625e-06, "loss": 0.09803855419158936, "step": 30980 }, { "epoch": 0.8163238427015868, "grad_norm": 2.373781681060791, "learning_rate": 1.718391289828144e-06, "loss": 0.08717655539512634, "step": 30990 }, { "epoch": 0.8165872579460854, "grad_norm": 1.7345314025878906, "learning_rate": 1.7136125511785462e-06, "loss": 0.09072790145874024, "step": 31000 }, { "epoch": 0.8168506731905842, "grad_norm": 2.217297077178955, "learning_rate": 1.7088398437192344e-06, "loss": 0.0981581449508667, "step": 31010 }, { "epoch": 0.8171140884350829, "grad_norm": 1.9742287397384644, "learning_rate": 1.7040731709239967e-06, "loss": 0.10147264003753662, "step": 31020 }, { "epoch": 0.8173775036795817, "grad_norm": 2.849788188934326, "learning_rate": 1.6993125362622231e-06, "loss": 0.10579862594604492, "step": 31030 }, { "epoch": 0.8176409189240804, "grad_norm": 2.090022563934326, "learning_rate": 1.6945579431989067e-06, "loss": 0.09839152097702027, "step": 31040 }, { "epoch": 0.8179043341685792, "grad_norm": 2.480990171432495, "learning_rate": 1.689809395194647e-06, "loss": 0.09805359244346619, "step": 31050 }, { "epoch": 0.8181677494130779, "grad_norm": 3.069821357727051, "learning_rate": 1.6850668957056393e-06, "loss": 0.1061213493347168, "step": 31060 }, { "epoch": 0.8184311646575766, "grad_norm": 2.8353090286254883, "learning_rate": 1.6803304481836835e-06, "loss": 0.08926213979721069, "step": 31070 }, { "epoch": 0.8186945799020754, "grad_norm": 3.790628433227539, "learning_rate": 1.6756000560761687e-06, "loss": 0.08962167501449585, "step": 31080 }, { "epoch": 0.8189579951465741, "grad_norm": 2.32479190826416, "learning_rate": 1.6708757228260786e-06, "loss": 0.09553762674331664, "step": 31090 }, { "epoch": 0.8192214103910729, "grad_norm": 2.197507858276367, "learning_rate": 1.6661574518719859e-06, "loss": 0.08167469501495361, "step": 31100 }, { "epoch": 0.8194848256355716, "grad_norm": 2.7031753063201904, "learning_rate": 1.6614452466480502e-06, "loss": 0.09293916225433349, "step": 31110 }, { "epoch": 0.8197482408800704, "grad_norm": 2.386213541030884, "learning_rate": 1.6567391105840247e-06, "loss": 0.09291895627975463, "step": 31120 }, { "epoch": 0.820011656124569, "grad_norm": 2.840965509414673, "learning_rate": 1.6520390471052306e-06, "loss": 0.09442095160484314, "step": 31130 }, { "epoch": 0.8202750713690679, "grad_norm": 1.8939725160598755, "learning_rate": 1.647345059632578e-06, "loss": 0.08836071491241455, "step": 31140 }, { "epoch": 0.8205384866135665, "grad_norm": 2.15107798576355, "learning_rate": 1.6426571515825563e-06, "loss": 0.09849565029144287, "step": 31150 }, { "epoch": 0.8208019018580652, "grad_norm": 2.389814853668213, "learning_rate": 1.6379753263672272e-06, "loss": 0.09843569993972778, "step": 31160 }, { "epoch": 0.821065317102564, "grad_norm": 2.4673664569854736, "learning_rate": 1.6332995873942237e-06, "loss": 0.09439014196395874, "step": 31170 }, { "epoch": 0.8213287323470627, "grad_norm": 2.3336329460144043, "learning_rate": 1.6286299380667504e-06, "loss": 0.09440286755561829, "step": 31180 }, { "epoch": 0.8215921475915615, "grad_norm": 3.1666481494903564, "learning_rate": 1.6239663817835783e-06, "loss": 0.09890685081481934, "step": 31190 }, { "epoch": 0.8218555628360602, "grad_norm": 2.1362645626068115, "learning_rate": 1.6193089219390478e-06, "loss": 0.09546375870704651, "step": 31200 }, { "epoch": 0.822118978080559, "grad_norm": 2.1598165035247803, "learning_rate": 1.6146575619230597e-06, "loss": 0.10098814964294434, "step": 31210 }, { "epoch": 0.8223823933250577, "grad_norm": 2.196150779724121, "learning_rate": 1.610012305121068e-06, "loss": 0.09322503805160523, "step": 31220 }, { "epoch": 0.8226458085695565, "grad_norm": 2.6035799980163574, "learning_rate": 1.6053731549140982e-06, "loss": 0.11002190113067627, "step": 31230 }, { "epoch": 0.8229092238140552, "grad_norm": 2.880038022994995, "learning_rate": 1.6007401146787195e-06, "loss": 0.10190057754516602, "step": 31240 }, { "epoch": 0.8231726390585539, "grad_norm": 2.1250228881835938, "learning_rate": 1.596113187787064e-06, "loss": 0.09335166811943055, "step": 31250 }, { "epoch": 0.8234360543030527, "grad_norm": 2.688743829727173, "learning_rate": 1.5914923776068025e-06, "loss": 0.10010758638381959, "step": 31260 }, { "epoch": 0.8236994695475514, "grad_norm": 2.230319023132324, "learning_rate": 1.5868776875011604e-06, "loss": 0.08984702825546265, "step": 31270 }, { "epoch": 0.8239628847920502, "grad_norm": 3.7899651527404785, "learning_rate": 1.582269120828912e-06, "loss": 0.11022096872329712, "step": 31280 }, { "epoch": 0.8242263000365488, "grad_norm": 2.3253586292266846, "learning_rate": 1.5776666809443697e-06, "loss": 0.080631422996521, "step": 31290 }, { "epoch": 0.8244897152810476, "grad_norm": 2.275787830352783, "learning_rate": 1.5730703711973883e-06, "loss": 0.08979446291923524, "step": 31300 }, { "epoch": 0.8247531305255463, "grad_norm": 2.0531129837036133, "learning_rate": 1.5684801949333584e-06, "loss": 0.09065364003181457, "step": 31310 }, { "epoch": 0.8250165457700451, "grad_norm": 2.178060531616211, "learning_rate": 1.5638961554932064e-06, "loss": 0.10511047840118408, "step": 31320 }, { "epoch": 0.8252799610145438, "grad_norm": 2.7280828952789307, "learning_rate": 1.5593182562134002e-06, "loss": 0.10244340896606445, "step": 31330 }, { "epoch": 0.8255433762590425, "grad_norm": 2.8170957565307617, "learning_rate": 1.554746500425931e-06, "loss": 0.09520951509475709, "step": 31340 }, { "epoch": 0.8258067915035413, "grad_norm": 1.8742411136627197, "learning_rate": 1.5501808914583139e-06, "loss": 0.0925305426120758, "step": 31350 }, { "epoch": 0.82607020674804, "grad_norm": 2.45717453956604, "learning_rate": 1.5456214326336017e-06, "loss": 0.09665898084640503, "step": 31360 }, { "epoch": 0.8263336219925388, "grad_norm": 1.9401803016662598, "learning_rate": 1.5410681272703643e-06, "loss": 0.09275521636009217, "step": 31370 }, { "epoch": 0.8265970372370375, "grad_norm": 2.8505475521087646, "learning_rate": 1.536520978682693e-06, "loss": 0.0896225094795227, "step": 31380 }, { "epoch": 0.8268604524815363, "grad_norm": 2.1955044269561768, "learning_rate": 1.531979990180199e-06, "loss": 0.08627361059188843, "step": 31390 }, { "epoch": 0.827123867726035, "grad_norm": 2.707451105117798, "learning_rate": 1.5274451650680066e-06, "loss": 0.10340821743011475, "step": 31400 }, { "epoch": 0.8273872829705338, "grad_norm": 2.3324174880981445, "learning_rate": 1.5229165066467632e-06, "loss": 0.09459377527236938, "step": 31410 }, { "epoch": 0.8276506982150325, "grad_norm": 1.8295801877975464, "learning_rate": 1.5183940182126189e-06, "loss": 0.09243956208229065, "step": 31420 }, { "epoch": 0.8279141134595311, "grad_norm": 2.563737630844116, "learning_rate": 1.513877703057235e-06, "loss": 0.09588671922683716, "step": 31430 }, { "epoch": 0.8281775287040299, "grad_norm": 2.174198627471924, "learning_rate": 1.509367564467782e-06, "loss": 0.09247288107872009, "step": 31440 }, { "epoch": 0.8284409439485286, "grad_norm": 2.8408374786376953, "learning_rate": 1.5048636057269294e-06, "loss": 0.10550904273986816, "step": 31450 }, { "epoch": 0.8287043591930274, "grad_norm": 2.4163150787353516, "learning_rate": 1.50036583011286e-06, "loss": 0.09473810195922852, "step": 31460 }, { "epoch": 0.8289677744375261, "grad_norm": 2.8959693908691406, "learning_rate": 1.4958742408992454e-06, "loss": 0.09343429207801819, "step": 31470 }, { "epoch": 0.8292311896820249, "grad_norm": 2.704195261001587, "learning_rate": 1.4913888413552536e-06, "loss": 0.09462268948554993, "step": 31480 }, { "epoch": 0.8294946049265236, "grad_norm": 3.5941240787506104, "learning_rate": 1.486909634745557e-06, "loss": 0.0879048228263855, "step": 31490 }, { "epoch": 0.8297580201710223, "grad_norm": 2.159776449203491, "learning_rate": 1.4824366243303156e-06, "loss": 0.0897950291633606, "step": 31500 }, { "epoch": 0.8300214354155211, "grad_norm": 1.9581103324890137, "learning_rate": 1.4779698133651766e-06, "loss": 0.0954628586769104, "step": 31510 }, { "epoch": 0.8302848506600198, "grad_norm": 3.19303822517395, "learning_rate": 1.47350920510128e-06, "loss": 0.09058750867843628, "step": 31520 }, { "epoch": 0.8305482659045186, "grad_norm": 2.9495937824249268, "learning_rate": 1.469054802785246e-06, "loss": 0.09902729988098144, "step": 31530 }, { "epoch": 0.8308116811490173, "grad_norm": 2.8796355724334717, "learning_rate": 1.4646066096591838e-06, "loss": 0.10087076425552369, "step": 31540 }, { "epoch": 0.8310750963935161, "grad_norm": 2.8139986991882324, "learning_rate": 1.4601646289606809e-06, "loss": 0.10248425006866455, "step": 31550 }, { "epoch": 0.8313385116380148, "grad_norm": 2.655945301055908, "learning_rate": 1.4557288639227995e-06, "loss": 0.08357237577438355, "step": 31560 }, { "epoch": 0.8316019268825136, "grad_norm": 2.2850356101989746, "learning_rate": 1.451299317774083e-06, "loss": 0.08574151992797852, "step": 31570 }, { "epoch": 0.8318653421270122, "grad_norm": 2.6722850799560547, "learning_rate": 1.4468759937385423e-06, "loss": 0.08634769916534424, "step": 31580 }, { "epoch": 0.8321287573715109, "grad_norm": 2.5746500492095947, "learning_rate": 1.4424588950356688e-06, "loss": 0.08730106353759766, "step": 31590 }, { "epoch": 0.8323921726160097, "grad_norm": 2.2083206176757812, "learning_rate": 1.4380480248804185e-06, "loss": 0.10893418788909912, "step": 31600 }, { "epoch": 0.8326555878605084, "grad_norm": 3.3825671672821045, "learning_rate": 1.4336433864832033e-06, "loss": 0.09494304656982422, "step": 31610 }, { "epoch": 0.8329190031050072, "grad_norm": 2.8877360820770264, "learning_rate": 1.4292449830499189e-06, "loss": 0.09124175906181335, "step": 31620 }, { "epoch": 0.8331824183495059, "grad_norm": 2.376105785369873, "learning_rate": 1.4248528177819098e-06, "loss": 0.09181804656982422, "step": 31630 }, { "epoch": 0.8334458335940047, "grad_norm": 3.8344614505767822, "learning_rate": 1.4204668938759835e-06, "loss": 0.08635830879211426, "step": 31640 }, { "epoch": 0.8337092488385034, "grad_norm": 2.7187886238098145, "learning_rate": 1.416087214524403e-06, "loss": 0.09250322580337525, "step": 31650 }, { "epoch": 0.8339726640830022, "grad_norm": 3.0349602699279785, "learning_rate": 1.4117137829148885e-06, "loss": 0.0936124563217163, "step": 31660 }, { "epoch": 0.8342360793275009, "grad_norm": 1.9629329442977905, "learning_rate": 1.4073466022306148e-06, "loss": 0.09212284088134766, "step": 31670 }, { "epoch": 0.8344994945719996, "grad_norm": 2.7767128944396973, "learning_rate": 1.4029856756502026e-06, "loss": 0.10110304355621338, "step": 31680 }, { "epoch": 0.8347629098164984, "grad_norm": 3.0751240253448486, "learning_rate": 1.3986310063477215e-06, "loss": 0.0951368749141693, "step": 31690 }, { "epoch": 0.835026325060997, "grad_norm": 2.7137973308563232, "learning_rate": 1.3942825974926888e-06, "loss": 0.09454774856567383, "step": 31700 }, { "epoch": 0.8352897403054959, "grad_norm": 3.6033194065093994, "learning_rate": 1.3899404522500626e-06, "loss": 0.09119894504547119, "step": 31710 }, { "epoch": 0.8355531555499945, "grad_norm": 2.612417221069336, "learning_rate": 1.385604573780248e-06, "loss": 0.08657161593437195, "step": 31720 }, { "epoch": 0.8358165707944933, "grad_norm": 2.353431224822998, "learning_rate": 1.38127496523908e-06, "loss": 0.1050149917602539, "step": 31730 }, { "epoch": 0.836079986038992, "grad_norm": 2.344355344772339, "learning_rate": 1.376951629777833e-06, "loss": 0.09698173403739929, "step": 31740 }, { "epoch": 0.8363434012834908, "grad_norm": 3.215313673019409, "learning_rate": 1.372634570543222e-06, "loss": 0.100224769115448, "step": 31750 }, { "epoch": 0.8366068165279895, "grad_norm": 2.1175074577331543, "learning_rate": 1.3683237906773873e-06, "loss": 0.09045387506484985, "step": 31760 }, { "epoch": 0.8368702317724882, "grad_norm": 2.1124730110168457, "learning_rate": 1.3640192933178997e-06, "loss": 0.08607937693595887, "step": 31770 }, { "epoch": 0.837133647016987, "grad_norm": 2.7939929962158203, "learning_rate": 1.359721081597758e-06, "loss": 0.09623829126358033, "step": 31780 }, { "epoch": 0.8373970622614857, "grad_norm": 2.0078372955322266, "learning_rate": 1.355429158645384e-06, "loss": 0.09443071484565735, "step": 31790 }, { "epoch": 0.8376604775059845, "grad_norm": 2.5406570434570312, "learning_rate": 1.3511435275846297e-06, "loss": 0.11336057186126709, "step": 31800 }, { "epoch": 0.8379238927504832, "grad_norm": 3.4993021488189697, "learning_rate": 1.3468641915347613e-06, "loss": 0.09293740391731262, "step": 31810 }, { "epoch": 0.838187307994982, "grad_norm": 3.440072774887085, "learning_rate": 1.3425911536104573e-06, "loss": 0.0871013045310974, "step": 31820 }, { "epoch": 0.8384507232394807, "grad_norm": 2.7893383502960205, "learning_rate": 1.3383244169218268e-06, "loss": 0.08447377681732178, "step": 31830 }, { "epoch": 0.8387141384839795, "grad_norm": 3.238522529602051, "learning_rate": 1.3340639845743797e-06, "loss": 0.1014520764350891, "step": 31840 }, { "epoch": 0.8389775537284782, "grad_norm": 4.142376899719238, "learning_rate": 1.32980985966905e-06, "loss": 0.10884908437728882, "step": 31850 }, { "epoch": 0.8392409689729768, "grad_norm": 2.564800977706909, "learning_rate": 1.3255620453021656e-06, "loss": 0.0934877336025238, "step": 31860 }, { "epoch": 0.8395043842174756, "grad_norm": 3.103771686553955, "learning_rate": 1.3213205445654688e-06, "loss": 0.09230377078056336, "step": 31870 }, { "epoch": 0.8397677994619743, "grad_norm": 2.444817066192627, "learning_rate": 1.3170853605461132e-06, "loss": 0.09301216602325439, "step": 31880 }, { "epoch": 0.8400312147064731, "grad_norm": 3.4591217041015625, "learning_rate": 1.3128564963266444e-06, "loss": 0.10012927055358886, "step": 31890 }, { "epoch": 0.8402946299509718, "grad_norm": 4.490346908569336, "learning_rate": 1.3086339549850125e-06, "loss": 0.09096555113792419, "step": 31900 }, { "epoch": 0.8405580451954706, "grad_norm": 2.342413902282715, "learning_rate": 1.3044177395945657e-06, "loss": 0.08534129858016967, "step": 31910 }, { "epoch": 0.8408214604399693, "grad_norm": 2.3099617958068848, "learning_rate": 1.3002078532240447e-06, "loss": 0.09926910400390625, "step": 31920 }, { "epoch": 0.8410848756844681, "grad_norm": 2.5733957290649414, "learning_rate": 1.296004298937592e-06, "loss": 0.09479378461837769, "step": 31930 }, { "epoch": 0.8413482909289668, "grad_norm": 3.220305919647217, "learning_rate": 1.2918070797947346e-06, "loss": 0.08150842189788818, "step": 31940 }, { "epoch": 0.8416117061734655, "grad_norm": 2.510610818862915, "learning_rate": 1.2876161988503843e-06, "loss": 0.09876208305358887, "step": 31950 }, { "epoch": 0.8418751214179643, "grad_norm": 2.4870128631591797, "learning_rate": 1.2834316591548502e-06, "loss": 0.09452131986618043, "step": 31960 }, { "epoch": 0.842138536662463, "grad_norm": 3.0697240829467773, "learning_rate": 1.279253463753819e-06, "loss": 0.09898748397827148, "step": 31970 }, { "epoch": 0.8424019519069618, "grad_norm": 2.582815170288086, "learning_rate": 1.2750816156883627e-06, "loss": 0.09422307014465332, "step": 31980 }, { "epoch": 0.8426653671514605, "grad_norm": 2.6874492168426514, "learning_rate": 1.2709161179949303e-06, "loss": 0.08603411912918091, "step": 31990 }, { "epoch": 0.8429287823959593, "grad_norm": 2.8140642642974854, "learning_rate": 1.266756973705351e-06, "loss": 0.09819988012313843, "step": 32000 }, { "epoch": 0.8431921976404579, "grad_norm": 2.656205892562866, "learning_rate": 1.2626041858468318e-06, "loss": 0.09255973100662232, "step": 32010 }, { "epoch": 0.8434556128849566, "grad_norm": 3.278170585632324, "learning_rate": 1.2584577574419499e-06, "loss": 0.08686026334762573, "step": 32020 }, { "epoch": 0.8437190281294554, "grad_norm": 3.080465078353882, "learning_rate": 1.2543176915086542e-06, "loss": 0.09543790817260742, "step": 32030 }, { "epoch": 0.8439824433739541, "grad_norm": 2.7155280113220215, "learning_rate": 1.2501839910602632e-06, "loss": 0.08158477544784545, "step": 32040 }, { "epoch": 0.8442458586184529, "grad_norm": 2.5617835521698, "learning_rate": 1.24605665910546e-06, "loss": 0.10263321399688721, "step": 32050 }, { "epoch": 0.8445092738629516, "grad_norm": 2.8031229972839355, "learning_rate": 1.2419356986483e-06, "loss": 0.08959351778030396, "step": 32060 }, { "epoch": 0.8447726891074504, "grad_norm": 2.609649896621704, "learning_rate": 1.2378211126881956e-06, "loss": 0.0949513554573059, "step": 32070 }, { "epoch": 0.8450361043519491, "grad_norm": 3.0239343643188477, "learning_rate": 1.2337129042199147e-06, "loss": 0.08377125263214111, "step": 32080 }, { "epoch": 0.8452995195964479, "grad_norm": 3.2040648460388184, "learning_rate": 1.2296110762335933e-06, "loss": 0.0812368631362915, "step": 32090 }, { "epoch": 0.8455629348409466, "grad_norm": 3.1910672187805176, "learning_rate": 1.2255156317147165e-06, "loss": 0.08880370855331421, "step": 32100 }, { "epoch": 0.8458263500854453, "grad_norm": 2.659348487854004, "learning_rate": 1.2214265736441277e-06, "loss": 0.08308431506156921, "step": 32110 }, { "epoch": 0.8460897653299441, "grad_norm": 3.258995771408081, "learning_rate": 1.2173439049980174e-06, "loss": 0.0925864815711975, "step": 32120 }, { "epoch": 0.8463531805744428, "grad_norm": 3.196812152862549, "learning_rate": 1.2132676287479273e-06, "loss": 0.08168690800666809, "step": 32130 }, { "epoch": 0.8466165958189416, "grad_norm": 3.292717695236206, "learning_rate": 1.209197747860752e-06, "loss": 0.09121686220169067, "step": 32140 }, { "epoch": 0.8468800110634402, "grad_norm": 3.0903663635253906, "learning_rate": 1.2051342652987218e-06, "loss": 0.09173312783241272, "step": 32150 }, { "epoch": 0.847143426307939, "grad_norm": 3.547985792160034, "learning_rate": 1.2010771840194168e-06, "loss": 0.09189888834953308, "step": 32160 }, { "epoch": 0.8474068415524377, "grad_norm": 2.931429862976074, "learning_rate": 1.1970265069757548e-06, "loss": 0.09845501184463501, "step": 32170 }, { "epoch": 0.8476702567969365, "grad_norm": 3.15432071685791, "learning_rate": 1.1929822371159915e-06, "loss": 0.09039353728294372, "step": 32180 }, { "epoch": 0.8479336720414352, "grad_norm": 1.8092447519302368, "learning_rate": 1.188944377383725e-06, "loss": 0.09682497382164001, "step": 32190 }, { "epoch": 0.8481970872859339, "grad_norm": 1.8187437057495117, "learning_rate": 1.1849129307178841e-06, "loss": 0.07977965474128723, "step": 32200 }, { "epoch": 0.8484605025304327, "grad_norm": 2.15590500831604, "learning_rate": 1.180887900052724e-06, "loss": 0.09012556076049805, "step": 32210 }, { "epoch": 0.8487239177749314, "grad_norm": 3.262451171875, "learning_rate": 1.17686928831784e-06, "loss": 0.09555094242095948, "step": 32220 }, { "epoch": 0.8489873330194302, "grad_norm": 2.6907286643981934, "learning_rate": 1.1728570984381505e-06, "loss": 0.10112637281417847, "step": 32230 }, { "epoch": 0.8492507482639289, "grad_norm": 3.008387565612793, "learning_rate": 1.1688513333339003e-06, "loss": 0.09789667129516602, "step": 32240 }, { "epoch": 0.8495141635084277, "grad_norm": 2.9115445613861084, "learning_rate": 1.1648519959206584e-06, "loss": 0.08286113739013672, "step": 32250 }, { "epoch": 0.8497775787529264, "grad_norm": 4.350349426269531, "learning_rate": 1.1608590891093118e-06, "loss": 0.08200201988220215, "step": 32260 }, { "epoch": 0.8500409939974252, "grad_norm": 2.895373582839966, "learning_rate": 1.156872615806075e-06, "loss": 0.09111675024032592, "step": 32270 }, { "epoch": 0.8503044092419239, "grad_norm": 2.6493518352508545, "learning_rate": 1.1528925789124734e-06, "loss": 0.09141121506690979, "step": 32280 }, { "epoch": 0.8505678244864225, "grad_norm": 2.3373231887817383, "learning_rate": 1.1489189813253499e-06, "loss": 0.08749632835388184, "step": 32290 }, { "epoch": 0.8508312397309213, "grad_norm": 3.471585273742676, "learning_rate": 1.1449518259368596e-06, "loss": 0.07943681478500367, "step": 32300 }, { "epoch": 0.85109465497542, "grad_norm": 2.9846630096435547, "learning_rate": 1.1409911156344677e-06, "loss": 0.09360946416854858, "step": 32310 }, { "epoch": 0.8513580702199188, "grad_norm": 4.026863098144531, "learning_rate": 1.1370368533009557e-06, "loss": 0.08720743656158447, "step": 32320 }, { "epoch": 0.8516214854644175, "grad_norm": 2.5302255153656006, "learning_rate": 1.1330890418144014e-06, "loss": 0.08124634027481079, "step": 32330 }, { "epoch": 0.8518849007089163, "grad_norm": 4.727496147155762, "learning_rate": 1.1291476840481908e-06, "loss": 0.09234426021575928, "step": 32340 }, { "epoch": 0.852148315953415, "grad_norm": 3.6577868461608887, "learning_rate": 1.125212782871019e-06, "loss": 0.09532197713851928, "step": 32350 }, { "epoch": 0.8524117311979138, "grad_norm": 2.5577032566070557, "learning_rate": 1.1212843411468755e-06, "loss": 0.09071091413497925, "step": 32360 }, { "epoch": 0.8526751464424125, "grad_norm": 3.0990421772003174, "learning_rate": 1.1173623617350472e-06, "loss": 0.09066571593284607, "step": 32370 }, { "epoch": 0.8529385616869112, "grad_norm": 2.740687847137451, "learning_rate": 1.1134468474901227e-06, "loss": 0.09211288690567017, "step": 32380 }, { "epoch": 0.85320197693141, "grad_norm": 2.9042043685913086, "learning_rate": 1.1095378012619796e-06, "loss": 0.08573687076568604, "step": 32390 }, { "epoch": 0.8534653921759087, "grad_norm": 2.7024149894714355, "learning_rate": 1.1056352258957924e-06, "loss": 0.08381372690200806, "step": 32400 }, { "epoch": 0.8537288074204075, "grad_norm": 2.8413636684417725, "learning_rate": 1.1017391242320263e-06, "loss": 0.08438253402709961, "step": 32410 }, { "epoch": 0.8539922226649062, "grad_norm": 2.910437822341919, "learning_rate": 1.0978494991064248e-06, "loss": 0.0920901358127594, "step": 32420 }, { "epoch": 0.854255637909405, "grad_norm": 1.9963726997375488, "learning_rate": 1.0939663533500311e-06, "loss": 0.08502305150032044, "step": 32430 }, { "epoch": 0.8545190531539036, "grad_norm": 2.7349014282226562, "learning_rate": 1.0900896897891634e-06, "loss": 0.09585847854614257, "step": 32440 }, { "epoch": 0.8547824683984024, "grad_norm": 4.296167373657227, "learning_rate": 1.0862195112454289e-06, "loss": 0.09277509450912476, "step": 32450 }, { "epoch": 0.8550458836429011, "grad_norm": 2.1637766361236572, "learning_rate": 1.0823558205357054e-06, "loss": 0.08726089000701905, "step": 32460 }, { "epoch": 0.8553092988873998, "grad_norm": 4.075812339782715, "learning_rate": 1.0784986204721537e-06, "loss": 0.10289008617401123, "step": 32470 }, { "epoch": 0.8555727141318986, "grad_norm": 3.859035015106201, "learning_rate": 1.0746479138622134e-06, "loss": 0.09606506824493408, "step": 32480 }, { "epoch": 0.8558361293763973, "grad_norm": 3.339150905609131, "learning_rate": 1.070803703508595e-06, "loss": 0.09929592609405517, "step": 32490 }, { "epoch": 0.8560995446208961, "grad_norm": 2.2278714179992676, "learning_rate": 1.066965992209279e-06, "loss": 0.08483380079269409, "step": 32500 }, { "epoch": 0.8563629598653948, "grad_norm": 2.468860387802124, "learning_rate": 1.0631347827575179e-06, "loss": 0.08611012697219848, "step": 32510 }, { "epoch": 0.8566263751098936, "grad_norm": 3.9262471199035645, "learning_rate": 1.0593100779418286e-06, "loss": 0.08846106529235839, "step": 32520 }, { "epoch": 0.8568897903543923, "grad_norm": 3.019666910171509, "learning_rate": 1.0554918805460013e-06, "loss": 0.09940208792686463, "step": 32530 }, { "epoch": 0.8571532055988911, "grad_norm": 2.9627552032470703, "learning_rate": 1.051680193349084e-06, "loss": 0.08616753816604614, "step": 32540 }, { "epoch": 0.8574166208433898, "grad_norm": 2.6743030548095703, "learning_rate": 1.0478750191253817e-06, "loss": 0.08803774118423462, "step": 32550 }, { "epoch": 0.8576800360878885, "grad_norm": 2.4110939502716064, "learning_rate": 1.0440763606444694e-06, "loss": 0.08580672740936279, "step": 32560 }, { "epoch": 0.8579434513323873, "grad_norm": 3.0888314247131348, "learning_rate": 1.0402842206711728e-06, "loss": 0.10861968994140625, "step": 32570 }, { "epoch": 0.8582068665768859, "grad_norm": 4.529305934906006, "learning_rate": 1.0364986019655765e-06, "loss": 0.09467005729675293, "step": 32580 }, { "epoch": 0.8584702818213847, "grad_norm": 3.8705146312713623, "learning_rate": 1.0327195072830153e-06, "loss": 0.09445434808731079, "step": 32590 }, { "epoch": 0.8587336970658834, "grad_norm": 2.7513556480407715, "learning_rate": 1.028946939374076e-06, "loss": 0.10197792053222657, "step": 32600 }, { "epoch": 0.8589971123103822, "grad_norm": 3.6215553283691406, "learning_rate": 1.0251809009846003e-06, "loss": 0.08807246685028076, "step": 32610 }, { "epoch": 0.8592605275548809, "grad_norm": 2.152308702468872, "learning_rate": 1.0214213948556718e-06, "loss": 0.07425551414489746, "step": 32620 }, { "epoch": 0.8595239427993796, "grad_norm": 2.624995231628418, "learning_rate": 1.0176684237236212e-06, "loss": 0.09617795944213867, "step": 32630 }, { "epoch": 0.8597873580438784, "grad_norm": 2.5175278186798096, "learning_rate": 1.0139219903200227e-06, "loss": 0.09196686148643493, "step": 32640 }, { "epoch": 0.8600507732883771, "grad_norm": 3.148681879043579, "learning_rate": 1.0101820973716891e-06, "loss": 0.09571431279182434, "step": 32650 }, { "epoch": 0.8603141885328759, "grad_norm": 3.10723614692688, "learning_rate": 1.0064487476006812e-06, "loss": 0.093073570728302, "step": 32660 }, { "epoch": 0.8605776037773746, "grad_norm": 2.3128750324249268, "learning_rate": 1.0027219437242907e-06, "loss": 0.08462905883789062, "step": 32670 }, { "epoch": 0.8608410190218734, "grad_norm": 3.012324333190918, "learning_rate": 9.990016884550425e-07, "loss": 0.08213291168212891, "step": 32680 }, { "epoch": 0.8611044342663721, "grad_norm": 2.6671533584594727, "learning_rate": 9.952879845007024e-07, "loss": 0.08425029516220092, "step": 32690 }, { "epoch": 0.8613678495108709, "grad_norm": 3.6449317932128906, "learning_rate": 9.915808345642642e-07, "loss": 0.08348126411437988, "step": 32700 }, { "epoch": 0.8616312647553696, "grad_norm": 3.7885963916778564, "learning_rate": 9.878802413439514e-07, "loss": 0.08834108114242553, "step": 32710 }, { "epoch": 0.8618946799998682, "grad_norm": 3.2745168209075928, "learning_rate": 9.841862075332143e-07, "loss": 0.09154578447341918, "step": 32720 }, { "epoch": 0.862158095244367, "grad_norm": 2.7761480808258057, "learning_rate": 9.804987358207274e-07, "loss": 0.08069422245025634, "step": 32730 }, { "epoch": 0.8624215104888657, "grad_norm": 2.6606805324554443, "learning_rate": 9.768178288903985e-07, "loss": 0.0951903998851776, "step": 32740 }, { "epoch": 0.8626849257333645, "grad_norm": 2.5139992237091064, "learning_rate": 9.731434894213465e-07, "loss": 0.09112598896026611, "step": 32750 }, { "epoch": 0.8629483409778632, "grad_norm": 2.599729061126709, "learning_rate": 9.694757200879146e-07, "loss": 0.09384306669235229, "step": 32760 }, { "epoch": 0.863211756222362, "grad_norm": 2.3694984912872314, "learning_rate": 9.658145235596628e-07, "loss": 0.09160438776016236, "step": 32770 }, { "epoch": 0.8634751714668607, "grad_norm": 4.030844211578369, "learning_rate": 9.621599025013672e-07, "loss": 0.09865644574165344, "step": 32780 }, { "epoch": 0.8637385867113595, "grad_norm": 3.0722975730895996, "learning_rate": 9.585118595730226e-07, "loss": 0.09908303022384643, "step": 32790 }, { "epoch": 0.8640020019558582, "grad_norm": 3.140033721923828, "learning_rate": 9.548703974298311e-07, "loss": 0.09221501350402832, "step": 32800 }, { "epoch": 0.8642654172003569, "grad_norm": 4.03399133682251, "learning_rate": 9.512355187222e-07, "loss": 0.09138857126235962, "step": 32810 }, { "epoch": 0.8645288324448557, "grad_norm": 2.8573901653289795, "learning_rate": 9.47607226095758e-07, "loss": 0.10079200267791748, "step": 32820 }, { "epoch": 0.8647922476893544, "grad_norm": 4.9827423095703125, "learning_rate": 9.439855221913297e-07, "loss": 0.08975324630737305, "step": 32830 }, { "epoch": 0.8650556629338532, "grad_norm": 3.121626615524292, "learning_rate": 9.403704096449484e-07, "loss": 0.09126791954040528, "step": 32840 }, { "epoch": 0.8653190781783519, "grad_norm": 3.6499810218811035, "learning_rate": 9.367618910878484e-07, "loss": 0.10034589767456055, "step": 32850 }, { "epoch": 0.8655824934228507, "grad_norm": 2.691061019897461, "learning_rate": 9.331599691464644e-07, "loss": 0.09931186437606812, "step": 32860 }, { "epoch": 0.8658459086673493, "grad_norm": 2.7514877319335938, "learning_rate": 9.295646464424323e-07, "loss": 0.08623950481414795, "step": 32870 }, { "epoch": 0.8661093239118481, "grad_norm": 3.2109360694885254, "learning_rate": 9.259759255925827e-07, "loss": 0.0835805892944336, "step": 32880 }, { "epoch": 0.8663727391563468, "grad_norm": 2.9502625465393066, "learning_rate": 9.223938092089413e-07, "loss": 0.08718862533569335, "step": 32890 }, { "epoch": 0.8666361544008455, "grad_norm": 2.4284729957580566, "learning_rate": 9.188182998987261e-07, "loss": 0.08135682344436646, "step": 32900 }, { "epoch": 0.8668995696453443, "grad_norm": 2.583721399307251, "learning_rate": 9.152494002643464e-07, "loss": 0.08042798042297364, "step": 32910 }, { "epoch": 0.867162984889843, "grad_norm": 3.0301895141601562, "learning_rate": 9.116871129034066e-07, "loss": 0.09379838109016418, "step": 32920 }, { "epoch": 0.8674264001343418, "grad_norm": 2.4695072174072266, "learning_rate": 9.081314404086872e-07, "loss": 0.08825876712799072, "step": 32930 }, { "epoch": 0.8676898153788405, "grad_norm": 2.609156608581543, "learning_rate": 9.045823853681602e-07, "loss": 0.09443806409835816, "step": 32940 }, { "epoch": 0.8679532306233393, "grad_norm": 2.575552225112915, "learning_rate": 9.010399503649836e-07, "loss": 0.0981717824935913, "step": 32950 }, { "epoch": 0.868216645867838, "grad_norm": 2.942566394805908, "learning_rate": 8.975041379774951e-07, "loss": 0.10560213327407837, "step": 32960 }, { "epoch": 0.8684800611123368, "grad_norm": 3.631166696548462, "learning_rate": 8.939749507792094e-07, "loss": 0.10037409067153931, "step": 32970 }, { "epoch": 0.8687434763568355, "grad_norm": 2.545596122741699, "learning_rate": 8.904523913388218e-07, "loss": 0.08031829595565795, "step": 32980 }, { "epoch": 0.8690068916013342, "grad_norm": 2.304084300994873, "learning_rate": 8.869364622202015e-07, "loss": 0.08482091426849366, "step": 32990 }, { "epoch": 0.869270306845833, "grad_norm": 2.4793643951416016, "learning_rate": 8.834271659823967e-07, "loss": 0.09781956076622009, "step": 33000 }, { "epoch": 0.8695337220903316, "grad_norm": 2.364407777786255, "learning_rate": 8.799245051796246e-07, "loss": 0.09236001968383789, "step": 33010 }, { "epoch": 0.8697971373348304, "grad_norm": 2.2580957412719727, "learning_rate": 8.764284823612667e-07, "loss": 0.0926478624343872, "step": 33020 }, { "epoch": 0.8700605525793291, "grad_norm": 3.298521041870117, "learning_rate": 8.729391000718867e-07, "loss": 0.09437699913978577, "step": 33030 }, { "epoch": 0.8703239678238279, "grad_norm": 2.6217219829559326, "learning_rate": 8.694563608512041e-07, "loss": 0.09034361839294433, "step": 33040 }, { "epoch": 0.8705873830683266, "grad_norm": 3.364741325378418, "learning_rate": 8.659802672341078e-07, "loss": 0.09435088634490967, "step": 33050 }, { "epoch": 0.8708507983128254, "grad_norm": 2.3028182983398438, "learning_rate": 8.625108217506472e-07, "loss": 0.08283640146255493, "step": 33060 }, { "epoch": 0.8711142135573241, "grad_norm": 3.2618582248687744, "learning_rate": 8.590480269260337e-07, "loss": 0.0927950918674469, "step": 33070 }, { "epoch": 0.8713776288018228, "grad_norm": 2.759251117706299, "learning_rate": 8.555918852806421e-07, "loss": 0.09193910360336303, "step": 33080 }, { "epoch": 0.8716410440463216, "grad_norm": 2.32448673248291, "learning_rate": 8.521423993300005e-07, "loss": 0.09571081995964051, "step": 33090 }, { "epoch": 0.8719044592908203, "grad_norm": 3.145430564880371, "learning_rate": 8.486995715847923e-07, "loss": 0.08550399541854858, "step": 33100 }, { "epoch": 0.8721678745353191, "grad_norm": 2.906242847442627, "learning_rate": 8.452634045508557e-07, "loss": 0.07901805639266968, "step": 33110 }, { "epoch": 0.8724312897798178, "grad_norm": 3.763230085372925, "learning_rate": 8.41833900729182e-07, "loss": 0.09457706212997437, "step": 33120 }, { "epoch": 0.8726947050243166, "grad_norm": 2.7676591873168945, "learning_rate": 8.384110626159126e-07, "loss": 0.09270282983779907, "step": 33130 }, { "epoch": 0.8729581202688153, "grad_norm": 2.638004779815674, "learning_rate": 8.349948927023388e-07, "loss": 0.08282482624053955, "step": 33140 }, { "epoch": 0.8732215355133139, "grad_norm": 2.6019108295440674, "learning_rate": 8.315853934748908e-07, "loss": 0.10141022205352783, "step": 33150 }, { "epoch": 0.8734849507578127, "grad_norm": 2.5719172954559326, "learning_rate": 8.281825674151533e-07, "loss": 0.09693488478660583, "step": 33160 }, { "epoch": 0.8737483660023114, "grad_norm": 2.586815118789673, "learning_rate": 8.247864169998499e-07, "loss": 0.08599652051925659, "step": 33170 }, { "epoch": 0.8740117812468102, "grad_norm": 2.6763596534729004, "learning_rate": 8.213969447008441e-07, "loss": 0.07919655442237854, "step": 33180 }, { "epoch": 0.8742751964913089, "grad_norm": 2.7802670001983643, "learning_rate": 8.180141529851404e-07, "loss": 0.09354665279388427, "step": 33190 }, { "epoch": 0.8745386117358077, "grad_norm": 3.7772412300109863, "learning_rate": 8.146380443148794e-07, "loss": 0.0920637607574463, "step": 33200 }, { "epoch": 0.8748020269803064, "grad_norm": 3.73685359954834, "learning_rate": 8.11268621147342e-07, "loss": 0.08488869667053223, "step": 33210 }, { "epoch": 0.8750654422248052, "grad_norm": 2.4781301021575928, "learning_rate": 8.079058859349364e-07, "loss": 0.08893254399299622, "step": 33220 }, { "epoch": 0.8753288574693039, "grad_norm": 2.9393656253814697, "learning_rate": 8.045498411252095e-07, "loss": 0.0877815306186676, "step": 33230 }, { "epoch": 0.8755922727138026, "grad_norm": 4.216365814208984, "learning_rate": 8.012004891608327e-07, "loss": 0.09183496236801147, "step": 33240 }, { "epoch": 0.8758556879583014, "grad_norm": 3.214771270751953, "learning_rate": 7.978578324796072e-07, "loss": 0.09113993048667908, "step": 33250 }, { "epoch": 0.8761191032028001, "grad_norm": 2.4173266887664795, "learning_rate": 7.945218735144689e-07, "loss": 0.08467539548873901, "step": 33260 }, { "epoch": 0.8763825184472989, "grad_norm": 2.281757116317749, "learning_rate": 7.91192614693469e-07, "loss": 0.07962658405303955, "step": 33270 }, { "epoch": 0.8766459336917976, "grad_norm": 2.8057310581207275, "learning_rate": 7.878700584397836e-07, "loss": 0.0890978753566742, "step": 33280 }, { "epoch": 0.8769093489362964, "grad_norm": 3.5883755683898926, "learning_rate": 7.84554207171716e-07, "loss": 0.08090507984161377, "step": 33290 }, { "epoch": 0.877172764180795, "grad_norm": 3.0417821407318115, "learning_rate": 7.812450633026847e-07, "loss": 0.0789763867855072, "step": 33300 }, { "epoch": 0.8774361794252938, "grad_norm": 3.7224814891815186, "learning_rate": 7.779426292412273e-07, "loss": 0.07945485115051269, "step": 33310 }, { "epoch": 0.8776995946697925, "grad_norm": 2.6210711002349854, "learning_rate": 7.746469073909979e-07, "loss": 0.08433708548545837, "step": 33320 }, { "epoch": 0.8779630099142912, "grad_norm": 2.5212347507476807, "learning_rate": 7.713579001507632e-07, "loss": 0.0952727198600769, "step": 33330 }, { "epoch": 0.87822642515879, "grad_norm": 2.925748825073242, "learning_rate": 7.680756099144072e-07, "loss": 0.08381397724151611, "step": 33340 }, { "epoch": 0.8784898404032887, "grad_norm": 3.4362411499023438, "learning_rate": 7.648000390709221e-07, "loss": 0.09530751705169678, "step": 33350 }, { "epoch": 0.8787532556477875, "grad_norm": 3.2152698040008545, "learning_rate": 7.615311900044087e-07, "loss": 0.07658843398094177, "step": 33360 }, { "epoch": 0.8790166708922862, "grad_norm": 4.520779609680176, "learning_rate": 7.582690650940771e-07, "loss": 0.0850820004940033, "step": 33370 }, { "epoch": 0.879280086136785, "grad_norm": 3.084282159805298, "learning_rate": 7.550136667142393e-07, "loss": 0.08681073784828186, "step": 33380 }, { "epoch": 0.8795435013812837, "grad_norm": 3.0398213863372803, "learning_rate": 7.517649972343199e-07, "loss": 0.08053352832794189, "step": 33390 }, { "epoch": 0.8798069166257825, "grad_norm": 3.096859931945801, "learning_rate": 7.485230590188397e-07, "loss": 0.07598026990890502, "step": 33400 }, { "epoch": 0.8800703318702812, "grad_norm": 3.7029881477355957, "learning_rate": 7.452878544274167e-07, "loss": 0.08580458164215088, "step": 33410 }, { "epoch": 0.8803337471147799, "grad_norm": 3.056345224380493, "learning_rate": 7.420593858147773e-07, "loss": 0.088480544090271, "step": 33420 }, { "epoch": 0.8805971623592787, "grad_norm": 3.1950876712799072, "learning_rate": 7.388376555307386e-07, "loss": 0.09383445978164673, "step": 33430 }, { "epoch": 0.8808605776037773, "grad_norm": 2.186882734298706, "learning_rate": 7.356226659202137e-07, "loss": 0.08586467504501342, "step": 33440 }, { "epoch": 0.8811239928482761, "grad_norm": 2.9347822666168213, "learning_rate": 7.324144193232119e-07, "loss": 0.07982243895530701, "step": 33450 }, { "epoch": 0.8813874080927748, "grad_norm": 2.88185977935791, "learning_rate": 7.29212918074833e-07, "loss": 0.07878372669219971, "step": 33460 }, { "epoch": 0.8816508233372736, "grad_norm": 2.7084848880767822, "learning_rate": 7.260181645052689e-07, "loss": 0.08899482488632202, "step": 33470 }, { "epoch": 0.8819142385817723, "grad_norm": 3.3272249698638916, "learning_rate": 7.228301609397981e-07, "loss": 0.09588735103607178, "step": 33480 }, { "epoch": 0.8821776538262711, "grad_norm": 3.192052125930786, "learning_rate": 7.196489096987869e-07, "loss": 0.09543607831001281, "step": 33490 }, { "epoch": 0.8824410690707698, "grad_norm": 4.234053611755371, "learning_rate": 7.164744130976886e-07, "loss": 0.07533845901489258, "step": 33500 }, { "epoch": 0.8827044843152685, "grad_norm": 2.6359446048736572, "learning_rate": 7.133066734470351e-07, "loss": 0.08881040215492249, "step": 33510 }, { "epoch": 0.8829678995597673, "grad_norm": 2.958955764770508, "learning_rate": 7.101456930524509e-07, "loss": 0.09275367259979247, "step": 33520 }, { "epoch": 0.883231314804266, "grad_norm": 2.7223575115203857, "learning_rate": 7.069914742146267e-07, "loss": 0.08992067575454712, "step": 33530 }, { "epoch": 0.8834947300487648, "grad_norm": 2.6364288330078125, "learning_rate": 7.038440192293416e-07, "loss": 0.08601759672164917, "step": 33540 }, { "epoch": 0.8837581452932635, "grad_norm": 2.9146194458007812, "learning_rate": 7.007033303874511e-07, "loss": 0.0919150948524475, "step": 33550 }, { "epoch": 0.8840215605377623, "grad_norm": 2.6077463626861572, "learning_rate": 6.97569409974882e-07, "loss": 0.08837722539901734, "step": 33560 }, { "epoch": 0.884284975782261, "grad_norm": 3.447746992111206, "learning_rate": 6.944422602726375e-07, "loss": 0.0837514042854309, "step": 33570 }, { "epoch": 0.8845483910267598, "grad_norm": 2.4730594158172607, "learning_rate": 6.913218835567925e-07, "loss": 0.09460371732711792, "step": 33580 }, { "epoch": 0.8848118062712584, "grad_norm": 2.959883689880371, "learning_rate": 6.882082820984892e-07, "loss": 0.08915255665779113, "step": 33590 }, { "epoch": 0.8850752215157571, "grad_norm": 2.185070753097534, "learning_rate": 6.851014581639437e-07, "loss": 0.0795866847038269, "step": 33600 }, { "epoch": 0.8853386367602559, "grad_norm": 3.3282954692840576, "learning_rate": 6.820014140144382e-07, "loss": 0.0869919240474701, "step": 33610 }, { "epoch": 0.8856020520047546, "grad_norm": 3.403904676437378, "learning_rate": 6.789081519063134e-07, "loss": 0.08777115345001221, "step": 33620 }, { "epoch": 0.8858654672492534, "grad_norm": 3.135868549346924, "learning_rate": 6.758216740909829e-07, "loss": 0.0947476327419281, "step": 33630 }, { "epoch": 0.8861288824937521, "grad_norm": 3.3368427753448486, "learning_rate": 6.727419828149173e-07, "loss": 0.09658998250961304, "step": 33640 }, { "epoch": 0.8863922977382509, "grad_norm": 2.9154863357543945, "learning_rate": 6.696690803196482e-07, "loss": 0.09005093574523926, "step": 33650 }, { "epoch": 0.8866557129827496, "grad_norm": 2.2819297313690186, "learning_rate": 6.666029688417674e-07, "loss": 0.0949366807937622, "step": 33660 }, { "epoch": 0.8869191282272483, "grad_norm": 2.0206573009490967, "learning_rate": 6.635436506129223e-07, "loss": 0.08898308277130126, "step": 33670 }, { "epoch": 0.8871825434717471, "grad_norm": 3.246802568435669, "learning_rate": 6.60491127859818e-07, "loss": 0.08332995176315308, "step": 33680 }, { "epoch": 0.8874459587162458, "grad_norm": 2.990107774734497, "learning_rate": 6.57445402804211e-07, "loss": 0.08590669631958008, "step": 33690 }, { "epoch": 0.8877093739607446, "grad_norm": 3.470938205718994, "learning_rate": 6.544064776629123e-07, "loss": 0.0897827386856079, "step": 33700 }, { "epoch": 0.8879727892052433, "grad_norm": 3.192746162414551, "learning_rate": 6.513743546477824e-07, "loss": 0.07895702123641968, "step": 33710 }, { "epoch": 0.888236204449742, "grad_norm": 3.5786077976226807, "learning_rate": 6.483490359657296e-07, "loss": 0.09403063058853149, "step": 33720 }, { "epoch": 0.8884996196942407, "grad_norm": 2.688558578491211, "learning_rate": 6.453305238187147e-07, "loss": 0.07546652555465698, "step": 33730 }, { "epoch": 0.8887630349387395, "grad_norm": 2.4580013751983643, "learning_rate": 6.423188204037412e-07, "loss": 0.0816095530986786, "step": 33740 }, { "epoch": 0.8890264501832382, "grad_norm": 2.684009552001953, "learning_rate": 6.393139279128513e-07, "loss": 0.08621006011962891, "step": 33750 }, { "epoch": 0.8892898654277369, "grad_norm": 2.622925043106079, "learning_rate": 6.3631584853314e-07, "loss": 0.09627124071121215, "step": 33760 }, { "epoch": 0.8895532806722357, "grad_norm": 4.059910774230957, "learning_rate": 6.333245844467384e-07, "loss": 0.08654860854148864, "step": 33770 }, { "epoch": 0.8898166959167344, "grad_norm": 2.0936827659606934, "learning_rate": 6.303401378308172e-07, "loss": 0.08326364159584046, "step": 33780 }, { "epoch": 0.8900801111612332, "grad_norm": 2.774616003036499, "learning_rate": 6.273625108575843e-07, "loss": 0.08215218782424927, "step": 33790 }, { "epoch": 0.8903435264057319, "grad_norm": 2.6843671798706055, "learning_rate": 6.243917056942849e-07, "loss": 0.0828206181526184, "step": 33800 }, { "epoch": 0.8906069416502307, "grad_norm": 2.498668670654297, "learning_rate": 6.214277245032019e-07, "loss": 0.07611084580421448, "step": 33810 }, { "epoch": 0.8908703568947294, "grad_norm": 3.0947165489196777, "learning_rate": 6.184705694416471e-07, "loss": 0.07553640604019166, "step": 33820 }, { "epoch": 0.8911337721392282, "grad_norm": 3.6278204917907715, "learning_rate": 6.155202426619655e-07, "loss": 0.09477493762969971, "step": 33830 }, { "epoch": 0.8913971873837269, "grad_norm": 2.6867377758026123, "learning_rate": 6.125767463115317e-07, "loss": 0.0821012794971466, "step": 33840 }, { "epoch": 0.8916606026282256, "grad_norm": 3.2063186168670654, "learning_rate": 6.096400825327487e-07, "loss": 0.0925445318222046, "step": 33850 }, { "epoch": 0.8919240178727244, "grad_norm": 3.43690824508667, "learning_rate": 6.067102534630498e-07, "loss": 0.08596715927124024, "step": 33860 }, { "epoch": 0.892187433117223, "grad_norm": 3.4504785537719727, "learning_rate": 6.037872612348905e-07, "loss": 0.0805177390575409, "step": 33870 }, { "epoch": 0.8924508483617218, "grad_norm": 3.485471725463867, "learning_rate": 6.008711079757467e-07, "loss": 0.08290721178054809, "step": 33880 }, { "epoch": 0.8927142636062205, "grad_norm": 3.4219040870666504, "learning_rate": 5.979617958081241e-07, "loss": 0.07942273616790771, "step": 33890 }, { "epoch": 0.8929776788507193, "grad_norm": 3.138793468475342, "learning_rate": 5.950593268495464e-07, "loss": 0.08565640449523926, "step": 33900 }, { "epoch": 0.893241094095218, "grad_norm": 2.783393383026123, "learning_rate": 5.921637032125538e-07, "loss": 0.08048315048217773, "step": 33910 }, { "epoch": 0.8935045093397168, "grad_norm": 2.939290761947632, "learning_rate": 5.89274927004706e-07, "loss": 0.08324223160743713, "step": 33920 }, { "epoch": 0.8937679245842155, "grad_norm": 2.5709803104400635, "learning_rate": 5.863930003285779e-07, "loss": 0.07793058156967163, "step": 33930 }, { "epoch": 0.8940313398287142, "grad_norm": 3.9264914989471436, "learning_rate": 5.835179252817635e-07, "loss": 0.08614345788955688, "step": 33940 }, { "epoch": 0.894294755073213, "grad_norm": 2.721067190170288, "learning_rate": 5.806497039568648e-07, "loss": 0.07095701694488525, "step": 33950 }, { "epoch": 0.8945581703177117, "grad_norm": 2.716885566711426, "learning_rate": 5.777883384414973e-07, "loss": 0.0801471471786499, "step": 33960 }, { "epoch": 0.8948215855622105, "grad_norm": 2.63810658454895, "learning_rate": 5.749338308182851e-07, "loss": 0.07082397341728211, "step": 33970 }, { "epoch": 0.8950850008067092, "grad_norm": 3.663970947265625, "learning_rate": 5.720861831648617e-07, "loss": 0.09453045129776001, "step": 33980 }, { "epoch": 0.895348416051208, "grad_norm": 3.1810386180877686, "learning_rate": 5.692453975538703e-07, "loss": 0.07849202156066895, "step": 33990 }, { "epoch": 0.8956118312957067, "grad_norm": 2.416968584060669, "learning_rate": 5.664114760529571e-07, "loss": 0.07546502351760864, "step": 34000 }, { "epoch": 0.8958752465402054, "grad_norm": 3.065967321395874, "learning_rate": 5.635844207247687e-07, "loss": 0.08856717348098755, "step": 34010 }, { "epoch": 0.8961386617847041, "grad_norm": 3.133533239364624, "learning_rate": 5.607642336269615e-07, "loss": 0.07696158289909363, "step": 34020 }, { "epoch": 0.8964020770292028, "grad_norm": 3.2229976654052734, "learning_rate": 5.579509168121888e-07, "loss": 0.08148243427276611, "step": 34030 }, { "epoch": 0.8966654922737016, "grad_norm": 2.5448482036590576, "learning_rate": 5.551444723281029e-07, "loss": 0.0803115725517273, "step": 34040 }, { "epoch": 0.8969289075182003, "grad_norm": 4.3199968338012695, "learning_rate": 5.523449022173566e-07, "loss": 0.10799534320831299, "step": 34050 }, { "epoch": 0.8971923227626991, "grad_norm": 2.529757022857666, "learning_rate": 5.495522085175942e-07, "loss": 0.08125383853912353, "step": 34060 }, { "epoch": 0.8974557380071978, "grad_norm": 2.966719150543213, "learning_rate": 5.467663932614642e-07, "loss": 0.07938695549964905, "step": 34070 }, { "epoch": 0.8977191532516966, "grad_norm": 2.87141489982605, "learning_rate": 5.439874584766014e-07, "loss": 0.0888981819152832, "step": 34080 }, { "epoch": 0.8979825684961953, "grad_norm": 2.95975399017334, "learning_rate": 5.412154061856301e-07, "loss": 0.06995826959609985, "step": 34090 }, { "epoch": 0.8982459837406941, "grad_norm": 2.9924659729003906, "learning_rate": 5.38450238406174e-07, "loss": 0.08956010937690735, "step": 34100 }, { "epoch": 0.8985093989851928, "grad_norm": 2.893056869506836, "learning_rate": 5.356919571508401e-07, "loss": 0.09560940265655518, "step": 34110 }, { "epoch": 0.8987728142296915, "grad_norm": 3.7698094844818115, "learning_rate": 5.329405644272268e-07, "loss": 0.09434202909469605, "step": 34120 }, { "epoch": 0.8990362294741903, "grad_norm": 3.600966215133667, "learning_rate": 5.301960622379132e-07, "loss": 0.07977936267852784, "step": 34130 }, { "epoch": 0.899299644718689, "grad_norm": 3.0004003047943115, "learning_rate": 5.27458452580466e-07, "loss": 0.08909812569618225, "step": 34140 }, { "epoch": 0.8995630599631877, "grad_norm": 3.62264084815979, "learning_rate": 5.24727737447438e-07, "loss": 0.07693030834197997, "step": 34150 }, { "epoch": 0.8998264752076864, "grad_norm": 3.8938021659851074, "learning_rate": 5.220039188263615e-07, "loss": 0.086391681432724, "step": 34160 }, { "epoch": 0.9000898904521852, "grad_norm": 2.6259286403656006, "learning_rate": 5.192869986997484e-07, "loss": 0.07947313785552979, "step": 34170 }, { "epoch": 0.9003533056966839, "grad_norm": 2.3528265953063965, "learning_rate": 5.165769790450903e-07, "loss": 0.08525752425193786, "step": 34180 }, { "epoch": 0.9006167209411826, "grad_norm": 2.845437526702881, "learning_rate": 5.138738618348548e-07, "loss": 0.0801089882850647, "step": 34190 }, { "epoch": 0.9008801361856814, "grad_norm": 3.0594887733459473, "learning_rate": 5.111776490364906e-07, "loss": 0.09717391729354859, "step": 34200 }, { "epoch": 0.9011435514301801, "grad_norm": 2.959667682647705, "learning_rate": 5.084883426124165e-07, "loss": 0.06964563131332398, "step": 34210 }, { "epoch": 0.9014069666746789, "grad_norm": 2.9908032417297363, "learning_rate": 5.058059445200237e-07, "loss": 0.08040726184844971, "step": 34220 }, { "epoch": 0.9016703819191776, "grad_norm": 2.8591763973236084, "learning_rate": 5.031304567116802e-07, "loss": 0.08987677097320557, "step": 34230 }, { "epoch": 0.9019337971636764, "grad_norm": 4.030356407165527, "learning_rate": 5.004618811347195e-07, "loss": 0.10365896224975586, "step": 34240 }, { "epoch": 0.9021972124081751, "grad_norm": 2.806819200515747, "learning_rate": 4.978002197314469e-07, "loss": 0.08299238681793213, "step": 34250 }, { "epoch": 0.9024606276526739, "grad_norm": 2.9352335929870605, "learning_rate": 4.951454744391338e-07, "loss": 0.08494160771369934, "step": 34260 }, { "epoch": 0.9027240428971726, "grad_norm": 3.025662422180176, "learning_rate": 4.924976471900178e-07, "loss": 0.0802310824394226, "step": 34270 }, { "epoch": 0.9029874581416713, "grad_norm": 4.5202178955078125, "learning_rate": 4.89856739911303e-07, "loss": 0.0755358099937439, "step": 34280 }, { "epoch": 0.90325087338617, "grad_norm": 3.538450241088867, "learning_rate": 4.872227545251562e-07, "loss": 0.07535643577575683, "step": 34290 }, { "epoch": 0.9035142886306687, "grad_norm": 3.260566234588623, "learning_rate": 4.84595692948705e-07, "loss": 0.08181400299072265, "step": 34300 }, { "epoch": 0.9037777038751675, "grad_norm": 4.154131889343262, "learning_rate": 4.819755570940377e-07, "loss": 0.07973582148551941, "step": 34310 }, { "epoch": 0.9040411191196662, "grad_norm": 2.9599456787109375, "learning_rate": 4.793623488682008e-07, "loss": 0.08374457359313965, "step": 34320 }, { "epoch": 0.904304534364165, "grad_norm": 4.554582595825195, "learning_rate": 4.7675607017320323e-07, "loss": 0.08346216678619385, "step": 34330 }, { "epoch": 0.9045679496086637, "grad_norm": 5.843318939208984, "learning_rate": 4.741567229060062e-07, "loss": 0.07991507053375244, "step": 34340 }, { "epoch": 0.9048313648531625, "grad_norm": 3.1303529739379883, "learning_rate": 4.715643089585242e-07, "loss": 0.084356689453125, "step": 34350 }, { "epoch": 0.9050947800976612, "grad_norm": 3.037910223007202, "learning_rate": 4.6897883021763013e-07, "loss": 0.07513250708580017, "step": 34360 }, { "epoch": 0.9053581953421599, "grad_norm": 3.3070619106292725, "learning_rate": 4.664002885651475e-07, "loss": 0.08469635844230652, "step": 34370 }, { "epoch": 0.9056216105866587, "grad_norm": 3.5560755729675293, "learning_rate": 4.638286858778496e-07, "loss": 0.08587071895599366, "step": 34380 }, { "epoch": 0.9058850258311574, "grad_norm": 3.2758469581604004, "learning_rate": 4.612640240274591e-07, "loss": 0.08496148586273193, "step": 34390 }, { "epoch": 0.9061484410756562, "grad_norm": 3.125624895095825, "learning_rate": 4.587063048806462e-07, "loss": 0.08583136200904846, "step": 34400 }, { "epoch": 0.9064118563201549, "grad_norm": 2.5626845359802246, "learning_rate": 4.561555302990328e-07, "loss": 0.08126582503318787, "step": 34410 }, { "epoch": 0.9066752715646537, "grad_norm": 3.1325175762176514, "learning_rate": 4.536117021391806e-07, "loss": 0.0855245590209961, "step": 34420 }, { "epoch": 0.9069386868091524, "grad_norm": 2.971648693084717, "learning_rate": 4.5107482225259735e-07, "loss": 0.0788845419883728, "step": 34430 }, { "epoch": 0.9072021020536511, "grad_norm": 2.8236637115478516, "learning_rate": 4.4854489248573494e-07, "loss": 0.08284119963645935, "step": 34440 }, { "epoch": 0.9074655172981498, "grad_norm": 3.5005531311035156, "learning_rate": 4.460219146799838e-07, "loss": 0.07457455396652221, "step": 34450 }, { "epoch": 0.9077289325426485, "grad_norm": 3.0091183185577393, "learning_rate": 4.4350589067167717e-07, "loss": 0.0789710283279419, "step": 34460 }, { "epoch": 0.9079923477871473, "grad_norm": 4.347733020782471, "learning_rate": 4.409968222920868e-07, "loss": 0.08807016611099243, "step": 34470 }, { "epoch": 0.908255763031646, "grad_norm": 3.0344150066375732, "learning_rate": 4.384947113674165e-07, "loss": 0.08393391966819763, "step": 34480 }, { "epoch": 0.9085191782761448, "grad_norm": 3.4083476066589355, "learning_rate": 4.35999559718816e-07, "loss": 0.07647489309310913, "step": 34490 }, { "epoch": 0.9087825935206435, "grad_norm": 3.4958810806274414, "learning_rate": 4.3351136916236156e-07, "loss": 0.08955590724945069, "step": 34500 }, { "epoch": 0.9090460087651423, "grad_norm": 3.980943202972412, "learning_rate": 4.3103014150906675e-07, "loss": 0.0746675193309784, "step": 34510 }, { "epoch": 0.909309424009641, "grad_norm": 3.825028657913208, "learning_rate": 4.285558785648747e-07, "loss": 0.08634073734283447, "step": 34520 }, { "epoch": 0.9095728392541398, "grad_norm": 2.6625008583068848, "learning_rate": 4.260885821306604e-07, "loss": 0.07420116066932678, "step": 34530 }, { "epoch": 0.9098362544986385, "grad_norm": 2.899355411529541, "learning_rate": 4.236282540022307e-07, "loss": 0.08205745220184327, "step": 34540 }, { "epoch": 0.9100996697431372, "grad_norm": 2.5102171897888184, "learning_rate": 4.2117489597031767e-07, "loss": 0.08913041949272156, "step": 34550 }, { "epoch": 0.910363084987636, "grad_norm": 4.465728759765625, "learning_rate": 4.1872850982058067e-07, "loss": 0.08080312609672546, "step": 34560 }, { "epoch": 0.9106265002321347, "grad_norm": 3.521416664123535, "learning_rate": 4.1628909733360534e-07, "loss": 0.07859263420104981, "step": 34570 }, { "epoch": 0.9108899154766334, "grad_norm": 3.2381508350372314, "learning_rate": 4.1385666028490036e-07, "loss": 0.09662553071975707, "step": 34580 }, { "epoch": 0.9111533307211321, "grad_norm": 3.475811004638672, "learning_rate": 4.114312004449006e-07, "loss": 0.07686443328857422, "step": 34590 }, { "epoch": 0.9114167459656309, "grad_norm": 3.8636598587036133, "learning_rate": 4.090127195789606e-07, "loss": 0.0870194137096405, "step": 34600 }, { "epoch": 0.9116801612101296, "grad_norm": 3.0415198802948, "learning_rate": 4.066012194473512e-07, "loss": 0.09444280862808227, "step": 34610 }, { "epoch": 0.9119435764546284, "grad_norm": 3.093485116958618, "learning_rate": 4.041967018052706e-07, "loss": 0.09004673957824708, "step": 34620 }, { "epoch": 0.9122069916991271, "grad_norm": 3.1581804752349854, "learning_rate": 4.0179916840282776e-07, "loss": 0.07921929955482483, "step": 34630 }, { "epoch": 0.9124704069436258, "grad_norm": 4.280839443206787, "learning_rate": 3.9940862098505117e-07, "loss": 0.09034912586212158, "step": 34640 }, { "epoch": 0.9127338221881246, "grad_norm": 4.008432865142822, "learning_rate": 3.970250612918858e-07, "loss": 0.09784238338470459, "step": 34650 }, { "epoch": 0.9129972374326233, "grad_norm": 3.1673340797424316, "learning_rate": 3.946484910581871e-07, "loss": 0.08409302234649658, "step": 34660 }, { "epoch": 0.9132606526771221, "grad_norm": 2.7463443279266357, "learning_rate": 3.9227891201372827e-07, "loss": 0.08482457399368286, "step": 34670 }, { "epoch": 0.9135240679216208, "grad_norm": 3.488628387451172, "learning_rate": 3.899163258831895e-07, "loss": 0.08785682916641235, "step": 34680 }, { "epoch": 0.9137874831661196, "grad_norm": 2.4668867588043213, "learning_rate": 3.8756073438616203e-07, "loss": 0.08536245822906494, "step": 34690 }, { "epoch": 0.9140508984106183, "grad_norm": 2.6861660480499268, "learning_rate": 3.8521213923714775e-07, "loss": 0.09075694084167481, "step": 34700 }, { "epoch": 0.914314313655117, "grad_norm": 2.8017754554748535, "learning_rate": 3.8287054214555694e-07, "loss": 0.0789981722831726, "step": 34710 }, { "epoch": 0.9145777288996157, "grad_norm": 2.5826070308685303, "learning_rate": 3.805359448157031e-07, "loss": 0.0815956711769104, "step": 34720 }, { "epoch": 0.9148411441441144, "grad_norm": 3.7146060466766357, "learning_rate": 3.78208348946808e-07, "loss": 0.08198602199554443, "step": 34730 }, { "epoch": 0.9151045593886132, "grad_norm": 2.966437339782715, "learning_rate": 3.7588775623299433e-07, "loss": 0.0809529721736908, "step": 34740 }, { "epoch": 0.9153679746331119, "grad_norm": 3.335696220397949, "learning_rate": 3.735741683632932e-07, "loss": 0.08364248275756836, "step": 34750 }, { "epoch": 0.9156313898776107, "grad_norm": 2.0734493732452393, "learning_rate": 3.712675870216309e-07, "loss": 0.07778718471527099, "step": 34760 }, { "epoch": 0.9158948051221094, "grad_norm": 3.4670753479003906, "learning_rate": 3.689680138868379e-07, "loss": 0.08317564725875855, "step": 34770 }, { "epoch": 0.9161582203666082, "grad_norm": 2.328885793685913, "learning_rate": 3.66675450632642e-07, "loss": 0.07206978797912597, "step": 34780 }, { "epoch": 0.9164216356111069, "grad_norm": 3.1351802349090576, "learning_rate": 3.643898989276684e-07, "loss": 0.09104031324386597, "step": 34790 }, { "epoch": 0.9166850508556056, "grad_norm": 4.0925750732421875, "learning_rate": 3.6211136043544425e-07, "loss": 0.08314403891563416, "step": 34800 }, { "epoch": 0.9169484661001044, "grad_norm": 3.837477922439575, "learning_rate": 3.598398368143863e-07, "loss": 0.0863699197769165, "step": 34810 }, { "epoch": 0.9172118813446031, "grad_norm": 2.843658685684204, "learning_rate": 3.575753297178053e-07, "loss": 0.07688092589378356, "step": 34820 }, { "epoch": 0.9174752965891019, "grad_norm": 3.146148443222046, "learning_rate": 3.553178407939095e-07, "loss": 0.08333368301391601, "step": 34830 }, { "epoch": 0.9177387118336006, "grad_norm": 2.4883267879486084, "learning_rate": 3.5306737168579685e-07, "loss": 0.07140800952911378, "step": 34840 }, { "epoch": 0.9180021270780994, "grad_norm": 2.9845476150512695, "learning_rate": 3.5082392403145373e-07, "loss": 0.08734196424484253, "step": 34850 }, { "epoch": 0.918265542322598, "grad_norm": 2.6126554012298584, "learning_rate": 3.485874994637606e-07, "loss": 0.08316202163696289, "step": 34860 }, { "epoch": 0.9185289575670968, "grad_norm": 3.546268939971924, "learning_rate": 3.463580996104821e-07, "loss": 0.08321827054023742, "step": 34870 }, { "epoch": 0.9187923728115955, "grad_norm": 3.1484179496765137, "learning_rate": 3.4413572609427235e-07, "loss": 0.08204931616783143, "step": 34880 }, { "epoch": 0.9190557880560942, "grad_norm": 3.510173797607422, "learning_rate": 3.4192038053267095e-07, "loss": 0.0877204716205597, "step": 34890 }, { "epoch": 0.919319203300593, "grad_norm": 3.2526156902313232, "learning_rate": 3.3971206453810133e-07, "loss": 0.07850162386894226, "step": 34900 }, { "epoch": 0.9195826185450917, "grad_norm": 3.112830877304077, "learning_rate": 3.3751077971787115e-07, "loss": 0.07955057621002197, "step": 34910 }, { "epoch": 0.9198460337895905, "grad_norm": 3.558072328567505, "learning_rate": 3.353165276741699e-07, "loss": 0.08785622119903565, "step": 34920 }, { "epoch": 0.9201094490340892, "grad_norm": 4.1258087158203125, "learning_rate": 3.331293100040711e-07, "loss": 0.08112850189208984, "step": 34930 }, { "epoch": 0.920372864278588, "grad_norm": 4.082957744598389, "learning_rate": 3.3094912829952477e-07, "loss": 0.08605327606201171, "step": 34940 }, { "epoch": 0.9206362795230867, "grad_norm": 3.715531587600708, "learning_rate": 3.2877598414736033e-07, "loss": 0.07865510582923889, "step": 34950 }, { "epoch": 0.9208996947675855, "grad_norm": 3.0951220989227295, "learning_rate": 3.266098791292871e-07, "loss": 0.0840553879737854, "step": 34960 }, { "epoch": 0.9211631100120842, "grad_norm": 2.3330910205841064, "learning_rate": 3.244508148218895e-07, "loss": 0.07518261671066284, "step": 34970 }, { "epoch": 0.9214265252565829, "grad_norm": 3.4036641120910645, "learning_rate": 3.22298792796627e-07, "loss": 0.08438724279403687, "step": 34980 }, { "epoch": 0.9216899405010817, "grad_norm": 3.208005428314209, "learning_rate": 3.2015381461983577e-07, "loss": 0.09100117683410644, "step": 34990 }, { "epoch": 0.9219533557455803, "grad_norm": 2.7699954509735107, "learning_rate": 3.1801588185272016e-07, "loss": 0.07686798572540283, "step": 35000 }, { "epoch": 0.9222167709900791, "grad_norm": 3.0403428077697754, "learning_rate": 3.1588499605136435e-07, "loss": 0.08368722200393677, "step": 35010 }, { "epoch": 0.9224801862345778, "grad_norm": 3.0167076587677, "learning_rate": 3.1376115876671553e-07, "loss": 0.09613497257232666, "step": 35020 }, { "epoch": 0.9227436014790766, "grad_norm": 2.749744415283203, "learning_rate": 3.1164437154459494e-07, "loss": 0.07597517967224121, "step": 35030 }, { "epoch": 0.9230070167235753, "grad_norm": 4.708847522735596, "learning_rate": 3.0953463592569343e-07, "loss": 0.07972502708435059, "step": 35040 }, { "epoch": 0.9232704319680741, "grad_norm": 3.0804433822631836, "learning_rate": 3.074319534455639e-07, "loss": 0.0941243588924408, "step": 35050 }, { "epoch": 0.9235338472125728, "grad_norm": 3.511942148208618, "learning_rate": 3.053363256346331e-07, "loss": 0.09230897426605225, "step": 35060 }, { "epoch": 0.9237972624570715, "grad_norm": 4.46218204498291, "learning_rate": 3.0324775401818884e-07, "loss": 0.07381195425987244, "step": 35070 }, { "epoch": 0.9240606777015703, "grad_norm": 3.5248541831970215, "learning_rate": 3.0116624011637953e-07, "loss": 0.09282021522521973, "step": 35080 }, { "epoch": 0.924324092946069, "grad_norm": 2.7835729122161865, "learning_rate": 2.9909178544422656e-07, "loss": 0.08314822316169738, "step": 35090 }, { "epoch": 0.9245875081905678, "grad_norm": 3.8281846046447754, "learning_rate": 2.9702439151160334e-07, "loss": 0.08882142305374145, "step": 35100 }, { "epoch": 0.9248509234350665, "grad_norm": 3.2960848808288574, "learning_rate": 2.949640598232517e-07, "loss": 0.07324846386909485, "step": 35110 }, { "epoch": 0.9251143386795653, "grad_norm": 3.532371759414673, "learning_rate": 2.9291079187876656e-07, "loss": 0.08569525480270386, "step": 35120 }, { "epoch": 0.925377753924064, "grad_norm": 3.789074659347534, "learning_rate": 2.9086458917260585e-07, "loss": 0.08080468177795411, "step": 35130 }, { "epoch": 0.9256411691685628, "grad_norm": 3.35091233253479, "learning_rate": 2.8882545319408706e-07, "loss": 0.07443431615829468, "step": 35140 }, { "epoch": 0.9259045844130614, "grad_norm": 2.828800916671753, "learning_rate": 2.8679338542737745e-07, "loss": 0.0903703510761261, "step": 35150 }, { "epoch": 0.9261679996575601, "grad_norm": 4.091304302215576, "learning_rate": 2.8476838735150613e-07, "loss": 0.08213675022125244, "step": 35160 }, { "epoch": 0.9264314149020589, "grad_norm": 3.24812388420105, "learning_rate": 2.8275046044035304e-07, "loss": 0.07638444900512695, "step": 35170 }, { "epoch": 0.9266948301465576, "grad_norm": 3.0207014083862305, "learning_rate": 2.8073960616265215e-07, "loss": 0.09528135061264038, "step": 35180 }, { "epoch": 0.9269582453910564, "grad_norm": 3.742913007736206, "learning_rate": 2.7873582598199056e-07, "loss": 0.08795477747917176, "step": 35190 }, { "epoch": 0.9272216606355551, "grad_norm": 2.179283618927002, "learning_rate": 2.7673912135680714e-07, "loss": 0.06944389343261718, "step": 35200 }, { "epoch": 0.9274850758800539, "grad_norm": 3.6792173385620117, "learning_rate": 2.7474949374038717e-07, "loss": 0.07838234901428223, "step": 35210 }, { "epoch": 0.9277484911245526, "grad_norm": 2.9140920639038086, "learning_rate": 2.7276694458086894e-07, "loss": 0.08624045848846436, "step": 35220 }, { "epoch": 0.9280119063690514, "grad_norm": 3.388768434524536, "learning_rate": 2.7079147532123704e-07, "loss": 0.08725693225860595, "step": 35230 }, { "epoch": 0.9282753216135501, "grad_norm": 2.6342103481292725, "learning_rate": 2.688230873993225e-07, "loss": 0.0663377046585083, "step": 35240 }, { "epoch": 0.9285387368580488, "grad_norm": 2.7572195529937744, "learning_rate": 2.6686178224780367e-07, "loss": 0.07700353860855103, "step": 35250 }, { "epoch": 0.9288021521025476, "grad_norm": 3.206094980239868, "learning_rate": 2.6490756129420313e-07, "loss": 0.07826132774353027, "step": 35260 }, { "epoch": 0.9290655673470463, "grad_norm": 3.7026798725128174, "learning_rate": 2.6296042596088643e-07, "loss": 0.08184983730316162, "step": 35270 }, { "epoch": 0.9293289825915451, "grad_norm": 3.161498546600342, "learning_rate": 2.610203776650655e-07, "loss": 0.08443442583084107, "step": 35280 }, { "epoch": 0.9295923978360437, "grad_norm": 3.340327739715576, "learning_rate": 2.5908741781878744e-07, "loss": 0.07587623596191406, "step": 35290 }, { "epoch": 0.9298558130805425, "grad_norm": 2.5090951919555664, "learning_rate": 2.5716154782894573e-07, "loss": 0.08126513361930847, "step": 35300 }, { "epoch": 0.9301192283250412, "grad_norm": 4.8577656745910645, "learning_rate": 2.552427690972703e-07, "loss": 0.08035948276519775, "step": 35310 }, { "epoch": 0.9303826435695399, "grad_norm": 4.148387908935547, "learning_rate": 2.5333108302033394e-07, "loss": 0.08481555581092834, "step": 35320 }, { "epoch": 0.9306460588140387, "grad_norm": 3.433245897293091, "learning_rate": 2.514264909895414e-07, "loss": 0.08121865391731262, "step": 35330 }, { "epoch": 0.9309094740585374, "grad_norm": 3.76210618019104, "learning_rate": 2.495289943911361e-07, "loss": 0.091678786277771, "step": 35340 }, { "epoch": 0.9311728893030362, "grad_norm": 3.0985474586486816, "learning_rate": 2.476385946062021e-07, "loss": 0.08055325746536254, "step": 35350 }, { "epoch": 0.9314363045475349, "grad_norm": 3.9079315662384033, "learning_rate": 2.4575529301065104e-07, "loss": 0.07811841368675232, "step": 35360 }, { "epoch": 0.9316997197920337, "grad_norm": 4.106131553649902, "learning_rate": 2.43879090975232e-07, "loss": 0.08745908737182617, "step": 35370 }, { "epoch": 0.9319631350365324, "grad_norm": 5.085618495941162, "learning_rate": 2.4200998986552484e-07, "loss": 0.08088310956954955, "step": 35380 }, { "epoch": 0.9322265502810312, "grad_norm": 4.521607875823975, "learning_rate": 2.401479910419413e-07, "loss": 0.08816174864768982, "step": 35390 }, { "epoch": 0.9324899655255299, "grad_norm": 3.07519268989563, "learning_rate": 2.3829309585972738e-07, "loss": 0.08533105850219727, "step": 35400 }, { "epoch": 0.9327533807700286, "grad_norm": 3.9141364097595215, "learning_rate": 2.364453056689553e-07, "loss": 0.08687054514884948, "step": 35410 }, { "epoch": 0.9330167960145274, "grad_norm": 3.3308281898498535, "learning_rate": 2.3460462181452258e-07, "loss": 0.08242011070251465, "step": 35420 }, { "epoch": 0.933280211259026, "grad_norm": 2.7324342727661133, "learning_rate": 2.3277104563616205e-07, "loss": 0.08785771131515503, "step": 35430 }, { "epoch": 0.9335436265035248, "grad_norm": 2.8924973011016846, "learning_rate": 2.3094457846842943e-07, "loss": 0.07430617809295655, "step": 35440 }, { "epoch": 0.9338070417480235, "grad_norm": 4.432921886444092, "learning_rate": 2.291252216407047e-07, "loss": 0.07027154564857482, "step": 35450 }, { "epoch": 0.9340704569925223, "grad_norm": 2.6412835121154785, "learning_rate": 2.273129764771953e-07, "loss": 0.07310807704925537, "step": 35460 }, { "epoch": 0.934333872237021, "grad_norm": 3.493670701980591, "learning_rate": 2.2550784429693052e-07, "loss": 0.08149528503417969, "step": 35470 }, { "epoch": 0.9345972874815198, "grad_norm": 2.9807181358337402, "learning_rate": 2.2370982641376494e-07, "loss": 0.06930849552154542, "step": 35480 }, { "epoch": 0.9348607027260185, "grad_norm": 4.573816299438477, "learning_rate": 2.2191892413637173e-07, "loss": 0.07242638468742371, "step": 35490 }, { "epoch": 0.9351241179705172, "grad_norm": 3.4060733318328857, "learning_rate": 2.2013513876824933e-07, "loss": 0.0842191994190216, "step": 35500 }, { "epoch": 0.935387533215016, "grad_norm": 3.20341420173645, "learning_rate": 2.183584716077114e-07, "loss": 0.07626940608024597, "step": 35510 }, { "epoch": 0.9356509484595147, "grad_norm": 2.706181287765503, "learning_rate": 2.1658892394789244e-07, "loss": 0.07938292026519775, "step": 35520 }, { "epoch": 0.9359143637040135, "grad_norm": 3.762568950653076, "learning_rate": 2.1482649707674775e-07, "loss": 0.07192053794860839, "step": 35530 }, { "epoch": 0.9361777789485122, "grad_norm": 2.75396728515625, "learning_rate": 2.13071192277049e-07, "loss": 0.07903485298156739, "step": 35540 }, { "epoch": 0.936441194193011, "grad_norm": 2.933680772781372, "learning_rate": 2.1132301082637752e-07, "loss": 0.08154487609863281, "step": 35550 }, { "epoch": 0.9367046094375097, "grad_norm": 3.494316816329956, "learning_rate": 2.0958195399713888e-07, "loss": 0.08886071443557739, "step": 35560 }, { "epoch": 0.9369680246820085, "grad_norm": 3.6108803749084473, "learning_rate": 2.0784802305654938e-07, "loss": 0.07722725868225097, "step": 35570 }, { "epoch": 0.9372314399265071, "grad_norm": 3.81976580619812, "learning_rate": 2.061212192666373e-07, "loss": 0.08356092572212219, "step": 35580 }, { "epoch": 0.9374948551710058, "grad_norm": 3.541877031326294, "learning_rate": 2.0440154388424616e-07, "loss": 0.0801086187362671, "step": 35590 }, { "epoch": 0.9377582704155046, "grad_norm": 3.2635700702667236, "learning_rate": 2.0268899816102916e-07, "loss": 0.07644298076629638, "step": 35600 }, { "epoch": 0.9380216856600033, "grad_norm": 3.3834638595581055, "learning_rate": 2.0098358334345147e-07, "loss": 0.07850615978240967, "step": 35610 }, { "epoch": 0.9382851009045021, "grad_norm": 3.4920787811279297, "learning_rate": 1.9928530067278684e-07, "loss": 0.076918226480484, "step": 35620 }, { "epoch": 0.9385485161490008, "grad_norm": 3.014052152633667, "learning_rate": 1.9759415138511984e-07, "loss": 0.08234553337097168, "step": 35630 }, { "epoch": 0.9388119313934996, "grad_norm": 3.414965867996216, "learning_rate": 1.959101367113414e-07, "loss": 0.08025280237197877, "step": 35640 }, { "epoch": 0.9390753466379983, "grad_norm": 2.590999126434326, "learning_rate": 1.9423325787714774e-07, "loss": 0.0781197190284729, "step": 35650 }, { "epoch": 0.9393387618824971, "grad_norm": 2.8618674278259277, "learning_rate": 1.9256351610304814e-07, "loss": 0.0838323712348938, "step": 35660 }, { "epoch": 0.9396021771269958, "grad_norm": 2.49741268157959, "learning_rate": 1.909009126043504e-07, "loss": 0.07189027070999146, "step": 35670 }, { "epoch": 0.9398655923714945, "grad_norm": 2.3049356937408447, "learning_rate": 1.8924544859116656e-07, "loss": 0.08320173025131225, "step": 35680 }, { "epoch": 0.9401290076159933, "grad_norm": 3.3760218620300293, "learning_rate": 1.8759712526841943e-07, "loss": 0.09425455331802368, "step": 35690 }, { "epoch": 0.940392422860492, "grad_norm": 3.3633458614349365, "learning_rate": 1.8595594383582605e-07, "loss": 0.0887882649898529, "step": 35700 }, { "epoch": 0.9406558381049908, "grad_norm": 3.3471739292144775, "learning_rate": 1.8432190548791197e-07, "loss": 0.08985176086425781, "step": 35710 }, { "epoch": 0.9409192533494894, "grad_norm": 2.5643928050994873, "learning_rate": 1.8269501141399915e-07, "loss": 0.07684924006462097, "step": 35720 }, { "epoch": 0.9411826685939882, "grad_norm": 3.209599256515503, "learning_rate": 1.8107526279821041e-07, "loss": 0.0870524525642395, "step": 35730 }, { "epoch": 0.9414460838384869, "grad_norm": 3.153867483139038, "learning_rate": 1.794626608194705e-07, "loss": 0.08497607707977295, "step": 35740 }, { "epoch": 0.9417094990829857, "grad_norm": 2.7054896354675293, "learning_rate": 1.7785720665150054e-07, "loss": 0.08860353231430054, "step": 35750 }, { "epoch": 0.9419729143274844, "grad_norm": 3.616401195526123, "learning_rate": 1.7625890146281798e-07, "loss": 0.07892615795135498, "step": 35760 }, { "epoch": 0.9422363295719831, "grad_norm": 2.896014928817749, "learning_rate": 1.7466774641674012e-07, "loss": 0.062407302856445315, "step": 35770 }, { "epoch": 0.9424997448164819, "grad_norm": 4.017721652984619, "learning_rate": 1.7308374267137606e-07, "loss": 0.08021291494369506, "step": 35780 }, { "epoch": 0.9427631600609806, "grad_norm": 2.5488009452819824, "learning_rate": 1.715068913796336e-07, "loss": 0.08599677085876464, "step": 35790 }, { "epoch": 0.9430265753054794, "grad_norm": 3.607666254043579, "learning_rate": 1.6993719368921468e-07, "loss": 0.08262885212898255, "step": 35800 }, { "epoch": 0.9432899905499781, "grad_norm": 3.4929823875427246, "learning_rate": 1.683746507426087e-07, "loss": 0.0753445565700531, "step": 35810 }, { "epoch": 0.9435534057944769, "grad_norm": 2.8466107845306396, "learning_rate": 1.66819263677106e-07, "loss": 0.0856070101261139, "step": 35820 }, { "epoch": 0.9438168210389756, "grad_norm": 3.960296154022217, "learning_rate": 1.6527103362478202e-07, "loss": 0.08565835356712341, "step": 35830 }, { "epoch": 0.9440802362834743, "grad_norm": 3.304680347442627, "learning_rate": 1.6372996171250544e-07, "loss": 0.09105749726295471, "step": 35840 }, { "epoch": 0.9443436515279731, "grad_norm": 3.0125269889831543, "learning_rate": 1.6219604906193677e-07, "loss": 0.07870637178421021, "step": 35850 }, { "epoch": 0.9446070667724717, "grad_norm": 2.825869560241699, "learning_rate": 1.606692967895207e-07, "loss": 0.08730052709579468, "step": 35860 }, { "epoch": 0.9448704820169705, "grad_norm": 3.976679801940918, "learning_rate": 1.5914970600649615e-07, "loss": 0.09384922981262207, "step": 35870 }, { "epoch": 0.9451338972614692, "grad_norm": 2.7253527641296387, "learning_rate": 1.5763727781888726e-07, "loss": 0.06890380382537842, "step": 35880 }, { "epoch": 0.945397312505968, "grad_norm": 2.633924961090088, "learning_rate": 1.5613201332750127e-07, "loss": 0.07402856349945068, "step": 35890 }, { "epoch": 0.9456607277504667, "grad_norm": 3.7678098678588867, "learning_rate": 1.5463391362793735e-07, "loss": 0.07832558155059814, "step": 35900 }, { "epoch": 0.9459241429949655, "grad_norm": 3.598539113998413, "learning_rate": 1.5314297981057658e-07, "loss": 0.07941666841506959, "step": 35910 }, { "epoch": 0.9461875582394642, "grad_norm": 2.1917884349823, "learning_rate": 1.5165921296058427e-07, "loss": 0.08138633966445923, "step": 35920 }, { "epoch": 0.9464509734839629, "grad_norm": 2.247236967086792, "learning_rate": 1.501826141579099e-07, "loss": 0.07521520256996155, "step": 35930 }, { "epoch": 0.9467143887284617, "grad_norm": 3.5096094608306885, "learning_rate": 1.48713184477286e-07, "loss": 0.0809207797050476, "step": 35940 }, { "epoch": 0.9469778039729604, "grad_norm": 4.801087379455566, "learning_rate": 1.4725092498822814e-07, "loss": 0.0875846266746521, "step": 35950 }, { "epoch": 0.9472412192174592, "grad_norm": 2.500501871109009, "learning_rate": 1.457958367550305e-07, "loss": 0.08625367283821106, "step": 35960 }, { "epoch": 0.9475046344619579, "grad_norm": 3.2405457496643066, "learning_rate": 1.443479208367693e-07, "loss": 0.07045297622680664, "step": 35970 }, { "epoch": 0.9477680497064567, "grad_norm": 3.973538637161255, "learning_rate": 1.4290717828730037e-07, "loss": 0.07893630266189575, "step": 35980 }, { "epoch": 0.9480314649509554, "grad_norm": 3.845377206802368, "learning_rate": 1.4147361015525717e-07, "loss": 0.08700265288352967, "step": 35990 }, { "epoch": 0.9482948801954542, "grad_norm": 3.2575817108154297, "learning_rate": 1.4004721748405392e-07, "loss": 0.09324226975440979, "step": 36000 }, { "epoch": 0.9485582954399528, "grad_norm": 3.5269522666931152, "learning_rate": 1.3862800131188237e-07, "loss": 0.0839369237422943, "step": 36010 }, { "epoch": 0.9488217106844515, "grad_norm": 3.6663482189178467, "learning_rate": 1.37215962671704e-07, "loss": 0.0810430109500885, "step": 36020 }, { "epoch": 0.9490851259289503, "grad_norm": 3.444056510925293, "learning_rate": 1.3581110259126674e-07, "loss": 0.07733712792396545, "step": 36030 }, { "epoch": 0.949348541173449, "grad_norm": 3.498277187347412, "learning_rate": 1.3441342209308595e-07, "loss": 0.08756636977195739, "step": 36040 }, { "epoch": 0.9496119564179478, "grad_norm": 4.080771446228027, "learning_rate": 1.3302292219445457e-07, "loss": 0.08821588158607482, "step": 36050 }, { "epoch": 0.9498753716624465, "grad_norm": 2.860117197036743, "learning_rate": 1.3163960390743746e-07, "loss": 0.07264686226844788, "step": 36060 }, { "epoch": 0.9501387869069453, "grad_norm": 4.1588921546936035, "learning_rate": 1.302634682388726e-07, "loss": 0.0822717308998108, "step": 36070 }, { "epoch": 0.950402202151444, "grad_norm": 2.4850683212280273, "learning_rate": 1.2889451619037318e-07, "loss": 0.08082115054130554, "step": 36080 }, { "epoch": 0.9506656173959428, "grad_norm": 4.039710998535156, "learning_rate": 1.2753274875832112e-07, "loss": 0.08314835429191589, "step": 36090 }, { "epoch": 0.9509290326404415, "grad_norm": 2.050668239593506, "learning_rate": 1.2617816693386907e-07, "loss": 0.07790923118591309, "step": 36100 }, { "epoch": 0.9511924478849402, "grad_norm": 2.8227381706237793, "learning_rate": 1.2483077170293954e-07, "loss": 0.09083651304244995, "step": 36110 }, { "epoch": 0.951455863129439, "grad_norm": 3.6943459510803223, "learning_rate": 1.2349056404622472e-07, "loss": 0.0650102972984314, "step": 36120 }, { "epoch": 0.9517192783739377, "grad_norm": 2.6941018104553223, "learning_rate": 1.2215754493918763e-07, "loss": 0.08649995923042297, "step": 36130 }, { "epoch": 0.9519826936184365, "grad_norm": 2.888972043991089, "learning_rate": 1.208317153520555e-07, "loss": 0.08072331547737122, "step": 36140 }, { "epoch": 0.9522461088629351, "grad_norm": 2.566572904586792, "learning_rate": 1.195130762498231e-07, "loss": 0.07762680053710938, "step": 36150 }, { "epoch": 0.952509524107434, "grad_norm": 3.34098482131958, "learning_rate": 1.1820162859225493e-07, "loss": 0.0794985294342041, "step": 36160 }, { "epoch": 0.9527729393519326, "grad_norm": 2.9330785274505615, "learning_rate": 1.1689737333387852e-07, "loss": 0.07338705658912659, "step": 36170 }, { "epoch": 0.9530363545964314, "grad_norm": 3.8981611728668213, "learning_rate": 1.1560031142398787e-07, "loss": 0.08719243407249451, "step": 36180 }, { "epoch": 0.9532997698409301, "grad_norm": 4.021990776062012, "learning_rate": 1.1431044380663891e-07, "loss": 0.09117951393127441, "step": 36190 }, { "epoch": 0.9535631850854288, "grad_norm": 2.662592887878418, "learning_rate": 1.1302777142065402e-07, "loss": 0.07695996761322021, "step": 36200 }, { "epoch": 0.9538266003299276, "grad_norm": 3.1694343090057373, "learning_rate": 1.117522951996175e-07, "loss": 0.07791326045989991, "step": 36210 }, { "epoch": 0.9540900155744263, "grad_norm": 3.7758257389068604, "learning_rate": 1.1048401607187565e-07, "loss": 0.0826797604560852, "step": 36220 }, { "epoch": 0.9543534308189251, "grad_norm": 3.400937557220459, "learning_rate": 1.0922293496053671e-07, "loss": 0.09511750936508179, "step": 36230 }, { "epoch": 0.9546168460634238, "grad_norm": 3.4521243572235107, "learning_rate": 1.0796905278347092e-07, "loss": 0.08306851983070374, "step": 36240 }, { "epoch": 0.9548802613079226, "grad_norm": 3.7417373657226562, "learning_rate": 1.0672237045330491e-07, "loss": 0.07480708360671998, "step": 36250 }, { "epoch": 0.9551436765524213, "grad_norm": 3.276909112930298, "learning_rate": 1.0548288887743174e-07, "loss": 0.07702910900115967, "step": 36260 }, { "epoch": 0.9554070917969201, "grad_norm": 3.7112839221954346, "learning_rate": 1.0425060895799865e-07, "loss": 0.0754775047302246, "step": 36270 }, { "epoch": 0.9556705070414188, "grad_norm": 4.921500205993652, "learning_rate": 1.0302553159190932e-07, "loss": 0.08896617293357849, "step": 36280 }, { "epoch": 0.9559339222859174, "grad_norm": 3.1982383728027344, "learning_rate": 1.0180765767083045e-07, "loss": 0.08302226066589355, "step": 36290 }, { "epoch": 0.9561973375304162, "grad_norm": 3.058547258377075, "learning_rate": 1.00596988081183e-07, "loss": 0.0656394600868225, "step": 36300 }, { "epoch": 0.9564607527749149, "grad_norm": 3.056581497192383, "learning_rate": 9.939352370414435e-08, "loss": 0.07114831805229187, "step": 36310 }, { "epoch": 0.9567241680194137, "grad_norm": 2.5360565185546875, "learning_rate": 9.819726541564711e-08, "loss": 0.08367639780044556, "step": 36320 }, { "epoch": 0.9569875832639124, "grad_norm": 3.785982370376587, "learning_rate": 9.700821408638039e-08, "loss": 0.08381316065788269, "step": 36330 }, { "epoch": 0.9572509985084112, "grad_norm": 2.3921871185302734, "learning_rate": 9.582637058178635e-08, "loss": 0.07727662920951843, "step": 36340 }, { "epoch": 0.9575144137529099, "grad_norm": 2.379070997238159, "learning_rate": 9.465173576206355e-08, "loss": 0.07249605655670166, "step": 36350 }, { "epoch": 0.9577778289974086, "grad_norm": 2.8880910873413086, "learning_rate": 9.348431048215923e-08, "loss": 0.08233056068420411, "step": 36360 }, { "epoch": 0.9580412442419074, "grad_norm": 3.5517241954803467, "learning_rate": 9.232409559177701e-08, "loss": 0.08356518745422363, "step": 36370 }, { "epoch": 0.9583046594864061, "grad_norm": 4.10744571685791, "learning_rate": 9.117109193537254e-08, "loss": 0.08717795014381409, "step": 36380 }, { "epoch": 0.9585680747309049, "grad_norm": 3.2733957767486572, "learning_rate": 9.002530035215007e-08, "loss": 0.08247268199920654, "step": 36390 }, { "epoch": 0.9588314899754036, "grad_norm": 3.6574718952178955, "learning_rate": 8.888672167606693e-08, "loss": 0.0861321210861206, "step": 36400 }, { "epoch": 0.9590949052199024, "grad_norm": 4.028165817260742, "learning_rate": 8.775535673582913e-08, "loss": 0.07479958534240723, "step": 36410 }, { "epoch": 0.9593583204644011, "grad_norm": 3.3088057041168213, "learning_rate": 8.663120635489241e-08, "loss": 0.0693530261516571, "step": 36420 }, { "epoch": 0.9596217357088999, "grad_norm": 2.77164626121521, "learning_rate": 8.55142713514645e-08, "loss": 0.08226193189620971, "step": 36430 }, { "epoch": 0.9598851509533985, "grad_norm": 2.6353063583374023, "learning_rate": 8.440455253849622e-08, "loss": 0.07654183506965637, "step": 36440 }, { "epoch": 0.9601485661978972, "grad_norm": 3.1267471313476562, "learning_rate": 8.330205072369035e-08, "loss": 0.08441108465194702, "step": 36450 }, { "epoch": 0.960411981442396, "grad_norm": 3.812303066253662, "learning_rate": 8.22067667094939e-08, "loss": 0.08434191942214966, "step": 36460 }, { "epoch": 0.9606753966868947, "grad_norm": 2.3375062942504883, "learning_rate": 8.11187012931025e-08, "loss": 0.07270373702049256, "step": 36470 }, { "epoch": 0.9609388119313935, "grad_norm": 3.7170393466949463, "learning_rate": 8.003785526645824e-08, "loss": 0.08831048011779785, "step": 36480 }, { "epoch": 0.9612022271758922, "grad_norm": 4.060833930969238, "learning_rate": 7.896422941624626e-08, "loss": 0.07795121669769287, "step": 36490 }, { "epoch": 0.961465642420391, "grad_norm": 3.481768846511841, "learning_rate": 7.789782452389705e-08, "loss": 0.08123452067375184, "step": 36500 }, { "epoch": 0.9617290576648897, "grad_norm": 3.3818836212158203, "learning_rate": 7.683864136558749e-08, "loss": 0.07628883719444275, "step": 36510 }, { "epoch": 0.9619924729093885, "grad_norm": 3.869878053665161, "learning_rate": 7.578668071223649e-08, "loss": 0.07670087814331054, "step": 36520 }, { "epoch": 0.9622558881538872, "grad_norm": 3.3468594551086426, "learning_rate": 7.474194332950602e-08, "loss": 0.07334237694740295, "step": 36530 }, { "epoch": 0.9625193033983859, "grad_norm": 3.7417540550231934, "learning_rate": 7.370442997780114e-08, "loss": 0.07172559499740601, "step": 36540 }, { "epoch": 0.9627827186428847, "grad_norm": 2.1424107551574707, "learning_rate": 7.267414141227002e-08, "loss": 0.0670099139213562, "step": 36550 }, { "epoch": 0.9630461338873834, "grad_norm": 2.71066951751709, "learning_rate": 7.165107838279838e-08, "loss": 0.08618881702423095, "step": 36560 }, { "epoch": 0.9633095491318822, "grad_norm": 2.771928310394287, "learning_rate": 7.063524163401836e-08, "loss": 0.08566093444824219, "step": 36570 }, { "epoch": 0.9635729643763808, "grad_norm": 4.244553089141846, "learning_rate": 6.962663190529851e-08, "loss": 0.08089950084686279, "step": 36580 }, { "epoch": 0.9638363796208796, "grad_norm": 2.447354555130005, "learning_rate": 6.862524993074825e-08, "loss": 0.06845871210098267, "step": 36590 }, { "epoch": 0.9640997948653783, "grad_norm": 3.5460898876190186, "learning_rate": 6.763109643921683e-08, "loss": 0.07385713458061219, "step": 36600 }, { "epoch": 0.9643632101098771, "grad_norm": 3.885822296142578, "learning_rate": 6.664417215429319e-08, "loss": 0.08909210562705994, "step": 36610 }, { "epoch": 0.9646266253543758, "grad_norm": 3.103344678878784, "learning_rate": 6.566447779430163e-08, "loss": 0.07813948392868042, "step": 36620 }, { "epoch": 0.9648900405988745, "grad_norm": 3.140707015991211, "learning_rate": 6.469201407230729e-08, "loss": 0.08741220831871033, "step": 36630 }, { "epoch": 0.9651534558433733, "grad_norm": 3.3857505321502686, "learning_rate": 6.372678169611068e-08, "loss": 0.07841212153434754, "step": 36640 }, { "epoch": 0.965416871087872, "grad_norm": 3.3837666511535645, "learning_rate": 6.276878136824982e-08, "loss": 0.07031522989273072, "step": 36650 }, { "epoch": 0.9656802863323708, "grad_norm": 3.148695707321167, "learning_rate": 6.181801378599805e-08, "loss": 0.07727954387664795, "step": 36660 }, { "epoch": 0.9659437015768695, "grad_norm": 2.3277628421783447, "learning_rate": 6.087447964136517e-08, "loss": 0.07457727789878846, "step": 36670 }, { "epoch": 0.9662071168213683, "grad_norm": 4.115120887756348, "learning_rate": 5.993817962109627e-08, "loss": 0.0962709367275238, "step": 36680 }, { "epoch": 0.966470532065867, "grad_norm": 3.8336403369903564, "learning_rate": 5.9009114406671787e-08, "loss": 0.07982535362243652, "step": 36690 }, { "epoch": 0.9667339473103658, "grad_norm": 4.119904518127441, "learning_rate": 5.8087284674304136e-08, "loss": 0.0843365728855133, "step": 36700 }, { "epoch": 0.9669973625548645, "grad_norm": 4.0521745681762695, "learning_rate": 5.7172691094942165e-08, "loss": 0.07822818756103515, "step": 36710 }, { "epoch": 0.9672607777993631, "grad_norm": 3.3460183143615723, "learning_rate": 5.626533433426451e-08, "loss": 0.08418148756027222, "step": 36720 }, { "epoch": 0.967524193043862, "grad_norm": 3.250821828842163, "learning_rate": 5.5365215052686216e-08, "loss": 0.07607810497283936, "step": 36730 }, { "epoch": 0.9677876082883606, "grad_norm": 3.6597790718078613, "learning_rate": 5.4472333905352115e-08, "loss": 0.07855787873268127, "step": 36740 }, { "epoch": 0.9680510235328594, "grad_norm": 3.314641237258911, "learning_rate": 5.3586691542140135e-08, "loss": 0.06622600555419922, "step": 36750 }, { "epoch": 0.9683144387773581, "grad_norm": 2.7767255306243896, "learning_rate": 5.2708288607659084e-08, "loss": 0.06820368766784668, "step": 36760 }, { "epoch": 0.9685778540218569, "grad_norm": 3.1539251804351807, "learning_rate": 5.183712574124755e-08, "loss": 0.07855565547943115, "step": 36770 }, { "epoch": 0.9688412692663556, "grad_norm": 3.6668307781219482, "learning_rate": 5.0973203576976104e-08, "loss": 0.08385986685752869, "step": 36780 }, { "epoch": 0.9691046845108544, "grad_norm": 3.966346502304077, "learning_rate": 5.011652274364509e-08, "loss": 0.07822505831718445, "step": 36790 }, { "epoch": 0.9693680997553531, "grad_norm": 3.7337613105773926, "learning_rate": 4.926708386478352e-08, "loss": 0.07241837978363037, "step": 36800 }, { "epoch": 0.9696315149998518, "grad_norm": 3.71059513092041, "learning_rate": 4.8424887558649045e-08, "loss": 0.07815968990325928, "step": 36810 }, { "epoch": 0.9698949302443506, "grad_norm": 2.41961407661438, "learning_rate": 4.7589934438228017e-08, "loss": 0.08366822600364685, "step": 36820 }, { "epoch": 0.9701583454888493, "grad_norm": 3.185199022293091, "learning_rate": 4.676222511123762e-08, "loss": 0.08805142641067505, "step": 36830 }, { "epoch": 0.9704217607333481, "grad_norm": 4.671657085418701, "learning_rate": 4.59417601801182e-08, "loss": 0.08921749591827392, "step": 36840 }, { "epoch": 0.9706851759778468, "grad_norm": 3.3864784240722656, "learning_rate": 4.512854024203983e-08, "loss": 0.1026678204536438, "step": 36850 }, { "epoch": 0.9709485912223456, "grad_norm": 2.5230953693389893, "learning_rate": 4.432256588890016e-08, "loss": 0.07387417554855347, "step": 36860 }, { "epoch": 0.9712120064668442, "grad_norm": 3.435598850250244, "learning_rate": 4.352383770732216e-08, "loss": 0.07965551018714905, "step": 36870 }, { "epoch": 0.9714754217113429, "grad_norm": 3.8272790908813477, "learning_rate": 4.273235627865302e-08, "loss": 0.06937637329101562, "step": 36880 }, { "epoch": 0.9717388369558417, "grad_norm": 3.3400490283966064, "learning_rate": 4.19481221789686e-08, "loss": 0.08133833408355713, "step": 36890 }, { "epoch": 0.9720022522003404, "grad_norm": 3.3695645332336426, "learning_rate": 4.117113597906675e-08, "loss": 0.07292927503585815, "step": 36900 }, { "epoch": 0.9722656674448392, "grad_norm": 3.1752939224243164, "learning_rate": 4.040139824447398e-08, "loss": 0.07658216953277588, "step": 36910 }, { "epoch": 0.9725290826893379, "grad_norm": 3.9419806003570557, "learning_rate": 3.963890953543659e-08, "loss": 0.07403979301452637, "step": 36920 }, { "epoch": 0.9727924979338367, "grad_norm": 3.8072245121002197, "learning_rate": 3.88836704069262e-08, "loss": 0.07881747484207154, "step": 36930 }, { "epoch": 0.9730559131783354, "grad_norm": 3.1940648555755615, "learning_rate": 3.813568140864088e-08, "loss": 0.08951616287231445, "step": 36940 }, { "epoch": 0.9733193284228342, "grad_norm": 3.5551204681396484, "learning_rate": 3.739494308499736e-08, "loss": 0.0924375057220459, "step": 36950 }, { "epoch": 0.9735827436673329, "grad_norm": 3.30731201171875, "learning_rate": 3.66614559751377e-08, "loss": 0.07997127175331116, "step": 36960 }, { "epoch": 0.9738461589118316, "grad_norm": 3.719454288482666, "learning_rate": 3.593522061292598e-08, "loss": 0.06359333992004394, "step": 36970 }, { "epoch": 0.9741095741563304, "grad_norm": 3.329185962677002, "learning_rate": 3.521623752694603e-08, "loss": 0.08577426671981811, "step": 36980 }, { "epoch": 0.9743729894008291, "grad_norm": 3.3039937019348145, "learning_rate": 3.4504507240507027e-08, "loss": 0.07685214877128602, "step": 36990 }, { "epoch": 0.9746364046453279, "grad_norm": 2.175581455230713, "learning_rate": 3.380003027163459e-08, "loss": 0.0729967474937439, "step": 37000 }, { "epoch": 0.9748998198898265, "grad_norm": 3.8133106231689453, "learning_rate": 3.310280713307856e-08, "loss": 0.0671647310256958, "step": 37010 }, { "epoch": 0.9751632351343253, "grad_norm": 3.332765579223633, "learning_rate": 3.241283833230746e-08, "loss": 0.07470282912254333, "step": 37020 }, { "epoch": 0.975426650378824, "grad_norm": 4.164594650268555, "learning_rate": 3.1730124371511794e-08, "loss": 0.07708516120910644, "step": 37030 }, { "epoch": 0.9756900656233228, "grad_norm": 3.303919792175293, "learning_rate": 3.1054665747599635e-08, "loss": 0.06656551361083984, "step": 37040 }, { "epoch": 0.9759534808678215, "grad_norm": 4.0509467124938965, "learning_rate": 3.038646295219771e-08, "loss": 0.09405139088630676, "step": 37050 }, { "epoch": 0.9762168961123202, "grad_norm": 3.5469682216644287, "learning_rate": 2.972551647165367e-08, "loss": 0.08668878674507141, "step": 37060 }, { "epoch": 0.976480311356819, "grad_norm": 3.6157965660095215, "learning_rate": 2.9071826787033797e-08, "loss": 0.10152641534805298, "step": 37070 }, { "epoch": 0.9767437266013177, "grad_norm": 2.9603753089904785, "learning_rate": 2.8425394374120842e-08, "loss": 0.07601255774497986, "step": 37080 }, { "epoch": 0.9770071418458165, "grad_norm": 3.6463069915771484, "learning_rate": 2.7786219703415108e-08, "loss": 0.08202457427978516, "step": 37090 }, { "epoch": 0.9772705570903152, "grad_norm": 2.8076741695404053, "learning_rate": 2.715430324013779e-08, "loss": 0.06449891328811645, "step": 37100 }, { "epoch": 0.977533972334814, "grad_norm": 3.027930974960327, "learning_rate": 2.6529645444223206e-08, "loss": 0.08578472733497619, "step": 37110 }, { "epoch": 0.9777973875793127, "grad_norm": 3.4960741996765137, "learning_rate": 2.5912246770324333e-08, "loss": 0.07528491020202636, "step": 37120 }, { "epoch": 0.9780608028238115, "grad_norm": 3.592390537261963, "learning_rate": 2.5302107667811716e-08, "loss": 0.08501970767974854, "step": 37130 }, { "epoch": 0.9783242180683102, "grad_norm": 3.848507881164551, "learning_rate": 2.469922858076901e-08, "loss": 0.08198221921920776, "step": 37140 }, { "epoch": 0.9785876333128088, "grad_norm": 6.3134765625, "learning_rate": 2.410360994799965e-08, "loss": 0.08646944761276246, "step": 37150 }, { "epoch": 0.9788510485573076, "grad_norm": 3.1764931678771973, "learning_rate": 2.351525220301909e-08, "loss": 0.08379952907562256, "step": 37160 }, { "epoch": 0.9791144638018063, "grad_norm": 3.7452383041381836, "learning_rate": 2.293415577406144e-08, "loss": 0.08169332146644592, "step": 37170 }, { "epoch": 0.9793778790463051, "grad_norm": 3.819570779800415, "learning_rate": 2.2360321084071712e-08, "loss": 0.0652350902557373, "step": 37180 }, { "epoch": 0.9796412942908038, "grad_norm": 3.1669366359710693, "learning_rate": 2.179374855071137e-08, "loss": 0.08161295056343079, "step": 37190 }, { "epoch": 0.9799047095353026, "grad_norm": 3.5456583499908447, "learning_rate": 2.123443858635832e-08, "loss": 0.07710825204849243, "step": 37200 }, { "epoch": 0.9801681247798013, "grad_norm": 2.8758134841918945, "learning_rate": 2.068239159810137e-08, "loss": 0.07527859210968017, "step": 37210 }, { "epoch": 0.9804315400243001, "grad_norm": 5.146163463592529, "learning_rate": 2.0137607987743558e-08, "loss": 0.07720626592636108, "step": 37220 }, { "epoch": 0.9806949552687988, "grad_norm": 2.9801981449127197, "learning_rate": 1.9600088151803255e-08, "loss": 0.0762102723121643, "step": 37230 }, { "epoch": 0.9809583705132975, "grad_norm": 3.8560426235198975, "learning_rate": 1.906983248150973e-08, "loss": 0.07584298253059388, "step": 37240 }, { "epoch": 0.9812217857577963, "grad_norm": 2.392115831375122, "learning_rate": 1.854684136280538e-08, "loss": 0.07918848991394042, "step": 37250 }, { "epoch": 0.981485201002295, "grad_norm": 4.403531074523926, "learning_rate": 1.8031115176346815e-08, "loss": 0.08814020156860351, "step": 37260 }, { "epoch": 0.9817486162467938, "grad_norm": 2.632052183151245, "learning_rate": 1.752265429750044e-08, "loss": 0.07985200881958007, "step": 37270 }, { "epoch": 0.9820120314912925, "grad_norm": 4.001180648803711, "learning_rate": 1.7021459096348004e-08, "loss": 0.09048662781715393, "step": 37280 }, { "epoch": 0.9822754467357913, "grad_norm": 3.674745559692383, "learning_rate": 1.652752993767992e-08, "loss": 0.07973977327346801, "step": 37290 }, { "epoch": 0.98253886198029, "grad_norm": 2.4252870082855225, "learning_rate": 1.604086718099973e-08, "loss": 0.07941004037857055, "step": 37300 }, { "epoch": 0.9828022772247887, "grad_norm": 3.4525043964385986, "learning_rate": 1.556147118051965e-08, "loss": 0.07474204897880554, "step": 37310 }, { "epoch": 0.9830656924692874, "grad_norm": 3.349771022796631, "learning_rate": 1.508934228516723e-08, "loss": 0.08623042106628417, "step": 37320 }, { "epoch": 0.9833291077137861, "grad_norm": 3.767266035079956, "learning_rate": 1.4624480838576483e-08, "loss": 0.07761359214782715, "step": 37330 }, { "epoch": 0.9835925229582849, "grad_norm": 2.578791856765747, "learning_rate": 1.4166887179094535e-08, "loss": 0.0671455979347229, "step": 37340 }, { "epoch": 0.9838559382027836, "grad_norm": 2.8975610733032227, "learning_rate": 1.3716561639777192e-08, "loss": 0.0810556173324585, "step": 37350 }, { "epoch": 0.9841193534472824, "grad_norm": 3.2894015312194824, "learning_rate": 1.3273504548392269e-08, "loss": 0.0753623902797699, "step": 37360 }, { "epoch": 0.9843827686917811, "grad_norm": 3.153244972229004, "learning_rate": 1.2837716227414032e-08, "loss": 0.07019039392471313, "step": 37370 }, { "epoch": 0.9846461839362799, "grad_norm": 3.0656957626342773, "learning_rate": 1.240919699402765e-08, "loss": 0.07575136423110962, "step": 37380 }, { "epoch": 0.9849095991807786, "grad_norm": 3.6688833236694336, "learning_rate": 1.1987947160130297e-08, "loss": 0.08591063022613525, "step": 37390 }, { "epoch": 0.9851730144252773, "grad_norm": 3.7495832443237305, "learning_rate": 1.1573967032324495e-08, "loss": 0.08820271492004395, "step": 37400 }, { "epoch": 0.9854364296697761, "grad_norm": 3.1494569778442383, "learning_rate": 1.1167256911922552e-08, "loss": 0.07559375166893005, "step": 37410 }, { "epoch": 0.9856998449142748, "grad_norm": 3.208392858505249, "learning_rate": 1.0767817094945454e-08, "loss": 0.07819053530693054, "step": 37420 }, { "epoch": 0.9859632601587736, "grad_norm": 3.694384813308716, "learning_rate": 1.0375647872123974e-08, "loss": 0.07139667272567748, "step": 37430 }, { "epoch": 0.9862266754032722, "grad_norm": 5.059241771697998, "learning_rate": 9.990749528894227e-09, "loss": 0.07837204933166504, "step": 37440 }, { "epoch": 0.986490090647771, "grad_norm": 3.422377824783325, "learning_rate": 9.613122345403236e-09, "loss": 0.08149483799934387, "step": 37450 }, { "epoch": 0.9867535058922697, "grad_norm": 3.915684223175049, "learning_rate": 9.24276659650336e-09, "loss": 0.0791463851928711, "step": 37460 }, { "epoch": 0.9870169211367685, "grad_norm": 3.5605499744415283, "learning_rate": 8.879682551755643e-09, "loss": 0.08922768831253051, "step": 37470 }, { "epoch": 0.9872803363812672, "grad_norm": 2.707592725753784, "learning_rate": 8.523870475429797e-09, "loss": 0.07197659611701965, "step": 37480 }, { "epoch": 0.9875437516257659, "grad_norm": 2.7899205684661865, "learning_rate": 8.175330626498668e-09, "loss": 0.07062914371490478, "step": 37490 }, { "epoch": 0.9878071668702647, "grad_norm": 3.4334192276000977, "learning_rate": 7.834063258645997e-09, "loss": 0.08505938649177551, "step": 37500 }, { "epoch": 0.9880705821147634, "grad_norm": 3.3356189727783203, "learning_rate": 7.500068620260869e-09, "loss": 0.0849015474319458, "step": 37510 }, { "epoch": 0.9883339973592622, "grad_norm": 3.6320760250091553, "learning_rate": 7.173346954438831e-09, "loss": 0.07898875474929809, "step": 37520 }, { "epoch": 0.9885974126037609, "grad_norm": 3.8548777103424072, "learning_rate": 6.853898498982992e-09, "loss": 0.08514646291732789, "step": 37530 }, { "epoch": 0.9888608278482597, "grad_norm": 2.6057631969451904, "learning_rate": 6.541723486399587e-09, "loss": 0.07811856269836426, "step": 37540 }, { "epoch": 0.9891242430927584, "grad_norm": 3.389355182647705, "learning_rate": 6.236822143904642e-09, "loss": 0.07582002282142639, "step": 37550 }, { "epoch": 0.9893876583372572, "grad_norm": 3.735788106918335, "learning_rate": 5.939194693418415e-09, "loss": 0.07647809386253357, "step": 37560 }, { "epoch": 0.9896510735817559, "grad_norm": 4.251745223999023, "learning_rate": 5.648841351566514e-09, "loss": 0.0766704261302948, "step": 37570 }, { "epoch": 0.9899144888262545, "grad_norm": 3.7358436584472656, "learning_rate": 5.365762329681002e-09, "loss": 0.07190104126930237, "step": 37580 }, { "epoch": 0.9901779040707533, "grad_norm": 2.451616048812866, "learning_rate": 5.089957833798176e-09, "loss": 0.08347662687301635, "step": 37590 }, { "epoch": 0.990441319315252, "grad_norm": 3.589111804962158, "learning_rate": 4.8214280646619035e-09, "loss": 0.08442447185516358, "step": 37600 }, { "epoch": 0.9907047345597508, "grad_norm": 2.989448070526123, "learning_rate": 4.560173217718067e-09, "loss": 0.07338271141052247, "step": 37610 }, { "epoch": 0.9909681498042495, "grad_norm": 3.631045341491699, "learning_rate": 4.3061934831212235e-09, "loss": 0.0704586923122406, "step": 37620 }, { "epoch": 0.9912315650487483, "grad_norm": 3.2550485134124756, "learning_rate": 4.0594890457268384e-09, "loss": 0.07008362412452698, "step": 37630 }, { "epoch": 0.991494980293247, "grad_norm": 4.049292087554932, "learning_rate": 3.820060085097943e-09, "loss": 0.08079715967178344, "step": 37640 }, { "epoch": 0.9917583955377458, "grad_norm": 4.926188945770264, "learning_rate": 3.587906775501804e-09, "loss": 0.07504080533981324, "step": 37650 }, { "epoch": 0.9920218107822445, "grad_norm": 3.227933883666992, "learning_rate": 3.363029285907704e-09, "loss": 0.06813185811042785, "step": 37660 }, { "epoch": 0.9922852260267432, "grad_norm": 3.1709728240966797, "learning_rate": 3.145427779993604e-09, "loss": 0.09147645235061645, "step": 37670 }, { "epoch": 0.992548641271242, "grad_norm": 2.749650239944458, "learning_rate": 2.935102416138369e-09, "loss": 0.08093016147613526, "step": 37680 }, { "epoch": 0.9928120565157407, "grad_norm": 2.5123696327209473, "learning_rate": 2.732053347425101e-09, "loss": 0.06879023909568786, "step": 37690 }, { "epoch": 0.9930754717602395, "grad_norm": 4.169955253601074, "learning_rate": 2.536280721642248e-09, "loss": 0.0890285313129425, "step": 37700 }, { "epoch": 0.9933388870047382, "grad_norm": 4.537631034851074, "learning_rate": 2.3477846812813843e-09, "loss": 0.07276389598846436, "step": 37710 }, { "epoch": 0.993602302249237, "grad_norm": 2.988370418548584, "learning_rate": 2.16656536353832e-09, "loss": 0.0751300573348999, "step": 37720 }, { "epoch": 0.9938657174937356, "grad_norm": 4.080356597900391, "learning_rate": 1.9926229003119912e-09, "loss": 0.07531857490539551, "step": 37730 }, { "epoch": 0.9941291327382344, "grad_norm": 2.6696572303771973, "learning_rate": 1.8259574182055707e-09, "loss": 0.07473999261856079, "step": 37740 }, { "epoch": 0.9943925479827331, "grad_norm": 3.6017825603485107, "learning_rate": 1.6665690385253564e-09, "loss": 0.07620586752891541, "step": 37750 }, { "epoch": 0.9946559632272318, "grad_norm": 4.389824390411377, "learning_rate": 1.5144578772807728e-09, "loss": 0.08367294073104858, "step": 37760 }, { "epoch": 0.9949193784717306, "grad_norm": 2.6322903633117676, "learning_rate": 1.36962404518437e-09, "loss": 0.09296576976776123, "step": 37770 }, { "epoch": 0.9951827937162293, "grad_norm": 3.660442352294922, "learning_rate": 1.232067647654045e-09, "loss": 0.07984203100204468, "step": 37780 }, { "epoch": 0.9954462089607281, "grad_norm": 2.516247510910034, "learning_rate": 1.1017887848085995e-09, "loss": 0.07852885127067566, "step": 37790 }, { "epoch": 0.9957096242052268, "grad_norm": 2.360837936401367, "learning_rate": 9.78787551469962e-10, "loss": 0.0718604028224945, "step": 37800 }, { "epoch": 0.9959730394497256, "grad_norm": 3.2304515838623047, "learning_rate": 8.630640371642963e-10, "loss": 0.08577928543090821, "step": 37810 }, { "epoch": 0.9962364546942243, "grad_norm": 3.56337571144104, "learning_rate": 7.546183261197826e-10, "loss": 0.08046060800552368, "step": 37820 }, { "epoch": 0.9964998699387231, "grad_norm": 3.2088756561279297, "learning_rate": 6.534504972688372e-10, "loss": 0.07467269897460938, "step": 37830 }, { "epoch": 0.9967632851832218, "grad_norm": 3.044252872467041, "learning_rate": 5.595606242447815e-10, "loss": 0.09295543432235717, "step": 37840 }, { "epoch": 0.9970267004277205, "grad_norm": 3.945995330810547, "learning_rate": 4.729487753851736e-10, "loss": 0.06839954853057861, "step": 37850 }, { "epoch": 0.9972901156722193, "grad_norm": 3.7632291316986084, "learning_rate": 3.9361501372958686e-10, "loss": 0.08171111941337586, "step": 37860 }, { "epoch": 0.997553530916718, "grad_norm": 3.5628371238708496, "learning_rate": 3.2155939702072093e-10, "loss": 0.0817295789718628, "step": 37870 }, { "epoch": 0.9978169461612167, "grad_norm": 2.934351682662964, "learning_rate": 2.5678197770329094e-10, "loss": 0.0911809742450714, "step": 37880 }, { "epoch": 0.9980803614057154, "grad_norm": 3.403958559036255, "learning_rate": 1.9928280292624835e-10, "loss": 0.08589597940444946, "step": 37890 }, { "epoch": 0.9983437766502142, "grad_norm": 4.057007789611816, "learning_rate": 1.4906191453833985e-10, "loss": 0.07659727334976196, "step": 37900 }, { "epoch": 0.9986071918947129, "grad_norm": 3.784315586090088, "learning_rate": 1.061193490936585e-10, "loss": 0.0767064094543457, "step": 37910 }, { "epoch": 0.9988706071392117, "grad_norm": 2.6403231620788574, "learning_rate": 7.045513784831315e-11, "loss": 0.07230588793754578, "step": 37920 }, { "epoch": 0.9991340223837104, "grad_norm": 2.534409284591675, "learning_rate": 4.206930675820786e-11, "loss": 0.07276164293289185, "step": 37930 }, { "epoch": 0.9993974376282091, "grad_norm": 3.4588510990142822, "learning_rate": 2.0961876485703358e-11, "loss": 0.07399822473526001, "step": 37940 }, { "epoch": 0.9996608528727079, "grad_norm": 2.705890655517578, "learning_rate": 7.132862392955631e-12, "loss": 0.07712229490280151, "step": 37950 }, { "epoch": 0.9999242681172066, "grad_norm": 3.6945912837982178, "learning_rate": 5.822745441363964e-13, "loss": 0.07713338136672973, "step": 37960 }, { "epoch": 1.0, "step": 37963, "total_flos": 2.0205788732727283e+20, "train_loss": 0.140463082595309, "train_runtime": 327843.1423, "train_samples_per_second": 7.411, "train_steps_per_second": 0.116 } ], "logging_steps": 10, "max_steps": 37963, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 1000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 2.0205788732727283e+20, "train_batch_size": 1, "trial_name": null, "trial_params": null }