{ "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": 0.0002634156781718952, "grad_norm": 1344.1268310546875, "learning_rate": 7.901668129938544e-08, "loss": 2.2199325561523438, "step": 10 }, { "epoch": 0.0005268313563437904, "grad_norm": 305.907958984375, "learning_rate": 1.6681299385425814e-07, "loss": 1.2107165336608887, "step": 20 }, { "epoch": 0.0007902470345156856, "grad_norm": 53.14158630371094, "learning_rate": 2.5460930640913084e-07, "loss": 0.3427035093307495, "step": 30 }, { "epoch": 0.0010536627126875807, "grad_norm": 26.266605377197266, "learning_rate": 3.4240561896400353e-07, "loss": 0.27893781661987305, "step": 40 }, { "epoch": 0.001317078390859476, "grad_norm": 31.805789947509766, "learning_rate": 4.3020193151887623e-07, "loss": 0.25335125923156737, "step": 50 }, { "epoch": 0.0015804940690313711, "grad_norm": 26.22760009765625, "learning_rate": 5.17998244073749e-07, "loss": 0.2507523536682129, "step": 60 }, { "epoch": 0.0018439097472032664, "grad_norm": 18.792566299438477, "learning_rate": 6.057945566286216e-07, "loss": 0.23843445777893066, "step": 70 }, { "epoch": 0.0021073254253751615, "grad_norm": 14.454401969909668, "learning_rate": 6.935908691834943e-07, "loss": 0.23929898738861083, "step": 80 }, { "epoch": 0.0023707411035470568, "grad_norm": 12.31025218963623, "learning_rate": 7.813871817383671e-07, "loss": 0.22582824230194093, "step": 90 }, { "epoch": 0.002634156781718952, "grad_norm": 15.075723648071289, "learning_rate": 8.691834942932398e-07, "loss": 0.21818239688873292, "step": 100 }, { "epoch": 0.002897572459890847, "grad_norm": 27.868282318115234, "learning_rate": 9.569798068481124e-07, "loss": 0.22904341220855712, "step": 110 }, { "epoch": 0.0031609881380627422, "grad_norm": 13.215436935424805, "learning_rate": 1.044776119402985e-06, "loss": 0.22476511001586913, "step": 120 }, { "epoch": 0.0034244038162346375, "grad_norm": 13.760505676269531, "learning_rate": 1.132572431957858e-06, "loss": 0.22371923923492432, "step": 130 }, { "epoch": 0.003687819494406533, "grad_norm": 20.607589721679688, "learning_rate": 1.2203687445127305e-06, "loss": 0.2427896499633789, "step": 140 }, { "epoch": 0.003951235172578428, "grad_norm": 12.707477569580078, "learning_rate": 1.3081650570676032e-06, "loss": 0.23230397701263428, "step": 150 }, { "epoch": 0.004214650850750323, "grad_norm": 11.90628433227539, "learning_rate": 1.395961369622476e-06, "loss": 0.22111644744873046, "step": 160 }, { "epoch": 0.004478066528922219, "grad_norm": 15.41542911529541, "learning_rate": 1.4837576821773487e-06, "loss": 0.21396496295928955, "step": 170 }, { "epoch": 0.0047414822070941135, "grad_norm": 15.13331127166748, "learning_rate": 1.5715539947322213e-06, "loss": 0.20910873413085937, "step": 180 }, { "epoch": 0.005004897885266008, "grad_norm": 23.847087860107422, "learning_rate": 1.659350307287094e-06, "loss": 0.19665462970733644, "step": 190 }, { "epoch": 0.005268313563437904, "grad_norm": 12.69528865814209, "learning_rate": 1.7471466198419668e-06, "loss": 0.20307505130767822, "step": 200 }, { "epoch": 0.005531729241609799, "grad_norm": 13.585241317749023, "learning_rate": 1.8349429323968395e-06, "loss": 0.22064239978790284, "step": 210 }, { "epoch": 0.005795144919781694, "grad_norm": 13.295137405395508, "learning_rate": 1.922739244951712e-06, "loss": 0.20313763618469238, "step": 220 }, { "epoch": 0.00605856059795359, "grad_norm": 17.71592140197754, "learning_rate": 2.0105355575065848e-06, "loss": 0.2114720344543457, "step": 230 }, { "epoch": 0.0063219762761254844, "grad_norm": 21.647708892822266, "learning_rate": 2.098331870061458e-06, "loss": 0.20645666122436523, "step": 240 }, { "epoch": 0.00658539195429738, "grad_norm": 15.717618942260742, "learning_rate": 2.18612818261633e-06, "loss": 0.20930118560791017, "step": 250 }, { "epoch": 0.006848807632469275, "grad_norm": 16.260637283325195, "learning_rate": 2.273924495171203e-06, "loss": 0.21160573959350587, "step": 260 }, { "epoch": 0.00711222331064117, "grad_norm": 15.06156063079834, "learning_rate": 2.3617208077260758e-06, "loss": 0.21642675399780273, "step": 270 }, { "epoch": 0.007375638988813066, "grad_norm": 11.326884269714355, "learning_rate": 2.4495171202809484e-06, "loss": 0.21193346977233887, "step": 280 }, { "epoch": 0.0076390546669849605, "grad_norm": 11.127808570861816, "learning_rate": 2.537313432835821e-06, "loss": 0.2029552698135376, "step": 290 }, { "epoch": 0.007902470345156856, "grad_norm": 16.06747817993164, "learning_rate": 2.6251097453906937e-06, "loss": 0.20278167724609375, "step": 300 }, { "epoch": 0.008165886023328751, "grad_norm": 7.38133430480957, "learning_rate": 2.7129060579455663e-06, "loss": 0.2083155632019043, "step": 310 }, { "epoch": 0.008429301701500646, "grad_norm": 11.810487747192383, "learning_rate": 2.8007023705004394e-06, "loss": 0.21380987167358398, "step": 320 }, { "epoch": 0.00869271737967254, "grad_norm": 24.387250900268555, "learning_rate": 2.888498683055312e-06, "loss": 0.22559595108032227, "step": 330 }, { "epoch": 0.008956133057844437, "grad_norm": 11.376081466674805, "learning_rate": 2.9762949956101843e-06, "loss": 0.18619390726089477, "step": 340 }, { "epoch": 0.009219548736016332, "grad_norm": 19.877111434936523, "learning_rate": 3.0640913081650573e-06, "loss": 0.1898792028427124, "step": 350 }, { "epoch": 0.009482964414188227, "grad_norm": 14.953822135925293, "learning_rate": 3.15188762071993e-06, "loss": 0.20459203720092772, "step": 360 }, { "epoch": 0.009746380092360122, "grad_norm": 7.993106842041016, "learning_rate": 3.2396839332748026e-06, "loss": 0.20634255409240723, "step": 370 }, { "epoch": 0.010009795770532017, "grad_norm": 17.376502990722656, "learning_rate": 3.3274802458296757e-06, "loss": 0.20443801879882811, "step": 380 }, { "epoch": 0.010273211448703913, "grad_norm": 13.012223243713379, "learning_rate": 3.4152765583845483e-06, "loss": 0.20967729091644288, "step": 390 }, { "epoch": 0.010536627126875808, "grad_norm": 8.289230346679688, "learning_rate": 3.5030728709394205e-06, "loss": 0.2098015785217285, "step": 400 }, { "epoch": 0.010800042805047703, "grad_norm": 13.055809020996094, "learning_rate": 3.5908691834942936e-06, "loss": 0.19794878959655762, "step": 410 }, { "epoch": 0.011063458483219598, "grad_norm": 15.049177169799805, "learning_rate": 3.6786654960491663e-06, "loss": 0.21246964931488038, "step": 420 }, { "epoch": 0.011326874161391493, "grad_norm": 13.16029167175293, "learning_rate": 3.766461808604039e-06, "loss": 0.20530548095703124, "step": 430 }, { "epoch": 0.011590289839563388, "grad_norm": 7.8615875244140625, "learning_rate": 3.8542581211589115e-06, "loss": 0.19117324352264403, "step": 440 }, { "epoch": 0.011853705517735284, "grad_norm": 11.592045783996582, "learning_rate": 3.942054433713785e-06, "loss": 0.18817155361175536, "step": 450 }, { "epoch": 0.01211712119590718, "grad_norm": 12.132209777832031, "learning_rate": 4.029850746268657e-06, "loss": 0.19603559970855713, "step": 460 }, { "epoch": 0.012380536874079074, "grad_norm": 10.021072387695312, "learning_rate": 4.11764705882353e-06, "loss": 0.1973685145378113, "step": 470 }, { "epoch": 0.012643952552250969, "grad_norm": 10.676888465881348, "learning_rate": 4.205443371378403e-06, "loss": 0.200744104385376, "step": 480 }, { "epoch": 0.012907368230422864, "grad_norm": 12.021708488464355, "learning_rate": 4.293239683933275e-06, "loss": 0.19789866209030152, "step": 490 }, { "epoch": 0.01317078390859476, "grad_norm": 9.941787719726562, "learning_rate": 4.381035996488147e-06, "loss": 0.1754964828491211, "step": 500 }, { "epoch": 0.013434199586766655, "grad_norm": 11.566021919250488, "learning_rate": 4.4688323090430205e-06, "loss": 0.18835082054138183, "step": 510 }, { "epoch": 0.01369761526493855, "grad_norm": 11.374008178710938, "learning_rate": 4.5566286215978935e-06, "loss": 0.18918077945709227, "step": 520 }, { "epoch": 0.013961030943110445, "grad_norm": 11.883458137512207, "learning_rate": 4.644424934152766e-06, "loss": 0.19024355411529542, "step": 530 }, { "epoch": 0.01422444662128234, "grad_norm": 12.175081253051758, "learning_rate": 4.732221246707639e-06, "loss": 0.1985282778739929, "step": 540 }, { "epoch": 0.014487862299454236, "grad_norm": 8.81667423248291, "learning_rate": 4.820017559262512e-06, "loss": 0.19168518781661986, "step": 550 }, { "epoch": 0.014751277977626131, "grad_norm": 12.070093154907227, "learning_rate": 4.907813871817384e-06, "loss": 0.2064610481262207, "step": 560 }, { "epoch": 0.015014693655798026, "grad_norm": 13.000725746154785, "learning_rate": 4.995610184372256e-06, "loss": 0.1862930178642273, "step": 570 }, { "epoch": 0.015278109333969921, "grad_norm": 7.2626142501831055, "learning_rate": 5.083406496927129e-06, "loss": 0.1941167116165161, "step": 580 }, { "epoch": 0.015541525012141816, "grad_norm": 8.632307052612305, "learning_rate": 5.1712028094820025e-06, "loss": 0.18595916032791138, "step": 590 }, { "epoch": 0.015804940690313712, "grad_norm": 9.535715103149414, "learning_rate": 5.2589991220368755e-06, "loss": 0.18071119785308837, "step": 600 }, { "epoch": 0.016068356368485606, "grad_norm": 8.21552848815918, "learning_rate": 5.346795434591747e-06, "loss": 0.19939901828765869, "step": 610 }, { "epoch": 0.016331772046657502, "grad_norm": 5.622674942016602, "learning_rate": 5.43459174714662e-06, "loss": 0.19190480709075927, "step": 620 }, { "epoch": 0.0165951877248294, "grad_norm": 8.804400444030762, "learning_rate": 5.522388059701493e-06, "loss": 0.19782911539077758, "step": 630 }, { "epoch": 0.016858603403001292, "grad_norm": 8.93156623840332, "learning_rate": 5.610184372256365e-06, "loss": 0.1780896782875061, "step": 640 }, { "epoch": 0.01712201908117319, "grad_norm": 7.705193996429443, "learning_rate": 5.697980684811238e-06, "loss": 0.21331636905670165, "step": 650 }, { "epoch": 0.01738543475934508, "grad_norm": 12.41068172454834, "learning_rate": 5.785776997366111e-06, "loss": 0.1851435899734497, "step": 660 }, { "epoch": 0.017648850437516978, "grad_norm": 12.323648452758789, "learning_rate": 5.873573309920984e-06, "loss": 0.18900887966156005, "step": 670 }, { "epoch": 0.017912266115688875, "grad_norm": 16.626461029052734, "learning_rate": 5.961369622475857e-06, "loss": 0.18545730113983155, "step": 680 }, { "epoch": 0.018175681793860768, "grad_norm": 9.513431549072266, "learning_rate": 6.04916593503073e-06, "loss": 0.18819092512130736, "step": 690 }, { "epoch": 0.018439097472032664, "grad_norm": 11.296788215637207, "learning_rate": 6.136962247585602e-06, "loss": 0.18345155715942382, "step": 700 }, { "epoch": 0.018702513150204558, "grad_norm": 6.111549377441406, "learning_rate": 6.224758560140475e-06, "loss": 0.16233972311019898, "step": 710 }, { "epoch": 0.018965928828376454, "grad_norm": 5.305216312408447, "learning_rate": 6.312554872695348e-06, "loss": 0.19602582454681397, "step": 720 }, { "epoch": 0.01922934450654835, "grad_norm": 10.138668060302734, "learning_rate": 6.4003511852502195e-06, "loss": 0.19271395206451417, "step": 730 }, { "epoch": 0.019492760184720244, "grad_norm": 8.349844932556152, "learning_rate": 6.4881474978050925e-06, "loss": 0.18569488525390626, "step": 740 }, { "epoch": 0.01975617586289214, "grad_norm": 11.33891773223877, "learning_rate": 6.575943810359965e-06, "loss": 0.20322816371917723, "step": 750 }, { "epoch": 0.020019591541064034, "grad_norm": 8.491156578063965, "learning_rate": 6.663740122914838e-06, "loss": 0.18714966773986816, "step": 760 }, { "epoch": 0.02028300721923593, "grad_norm": 8.559356689453125, "learning_rate": 6.751536435469711e-06, "loss": 0.20190515518188476, "step": 770 }, { "epoch": 0.020546422897407827, "grad_norm": 4.2997283935546875, "learning_rate": 6.839332748024583e-06, "loss": 0.17488479614257812, "step": 780 }, { "epoch": 0.02080983857557972, "grad_norm": 6.1610846519470215, "learning_rate": 6.927129060579456e-06, "loss": 0.18732242584228515, "step": 790 }, { "epoch": 0.021073254253751617, "grad_norm": 6.851006984710693, "learning_rate": 7.014925373134329e-06, "loss": 0.19470269680023194, "step": 800 }, { "epoch": 0.02133666993192351, "grad_norm": 5.979601860046387, "learning_rate": 7.1027216856892015e-06, "loss": 0.17967044115066527, "step": 810 }, { "epoch": 0.021600085610095406, "grad_norm": 7.648305416107178, "learning_rate": 7.1905179982440745e-06, "loss": 0.202604079246521, "step": 820 }, { "epoch": 0.0218635012882673, "grad_norm": 3.9838502407073975, "learning_rate": 7.278314310798948e-06, "loss": 0.18650438785552978, "step": 830 }, { "epoch": 0.022126916966439196, "grad_norm": 6.095693588256836, "learning_rate": 7.366110623353819e-06, "loss": 0.18570631742477417, "step": 840 }, { "epoch": 0.022390332644611093, "grad_norm": 5.933147430419922, "learning_rate": 7.453906935908692e-06, "loss": 0.21298682689666748, "step": 850 }, { "epoch": 0.022653748322782986, "grad_norm": 11.681578636169434, "learning_rate": 7.541703248463565e-06, "loss": 0.19844311475753784, "step": 860 }, { "epoch": 0.022917164000954882, "grad_norm": 6.235942840576172, "learning_rate": 7.629499561018438e-06, "loss": 0.19478822946548463, "step": 870 }, { "epoch": 0.023180579679126775, "grad_norm": 15.257841110229492, "learning_rate": 7.717295873573311e-06, "loss": 0.19437431097030639, "step": 880 }, { "epoch": 0.023443995357298672, "grad_norm": 3.784950017929077, "learning_rate": 7.805092186128184e-06, "loss": 0.19444185495376587, "step": 890 }, { "epoch": 0.02370741103547057, "grad_norm": 5.221986770629883, "learning_rate": 7.892888498683056e-06, "loss": 0.19580020904541015, "step": 900 }, { "epoch": 0.02397082671364246, "grad_norm": 9.73990535736084, "learning_rate": 7.980684811237929e-06, "loss": 0.20281095504760743, "step": 910 }, { "epoch": 0.02423424239181436, "grad_norm": 12.60679817199707, "learning_rate": 8.068481123792802e-06, "loss": 0.18488106727600098, "step": 920 }, { "epoch": 0.02449765806998625, "grad_norm": 11.014083862304688, "learning_rate": 8.156277436347673e-06, "loss": 0.19988620281219482, "step": 930 }, { "epoch": 0.024761073748158148, "grad_norm": 6.512516975402832, "learning_rate": 8.244073748902546e-06, "loss": 0.19108359813690184, "step": 940 }, { "epoch": 0.025024489426330045, "grad_norm": 9.021238327026367, "learning_rate": 8.33187006145742e-06, "loss": 0.1851930022239685, "step": 950 }, { "epoch": 0.025287905104501938, "grad_norm": 3.2743828296661377, "learning_rate": 8.419666374012292e-06, "loss": 0.1942622423171997, "step": 960 }, { "epoch": 0.025551320782673834, "grad_norm": 9.790884017944336, "learning_rate": 8.507462686567165e-06, "loss": 0.19740605354309082, "step": 970 }, { "epoch": 0.025814736460845727, "grad_norm": 8.795883178710938, "learning_rate": 8.595258999122039e-06, "loss": 0.1972440719604492, "step": 980 }, { "epoch": 0.026078152139017624, "grad_norm": 7.4297027587890625, "learning_rate": 8.68305531167691e-06, "loss": 0.18811614513397218, "step": 990 }, { "epoch": 0.02634156781718952, "grad_norm": 9.566286087036133, "learning_rate": 8.770851624231783e-06, "loss": 0.17567049264907836, "step": 1000 }, { "epoch": 0.026604983495361414, "grad_norm": 11.592842102050781, "learning_rate": 8.858647936786654e-06, "loss": 0.19291529655456544, "step": 1010 }, { "epoch": 0.02686839917353331, "grad_norm": 3.248302459716797, "learning_rate": 8.946444249341527e-06, "loss": 0.1775096297264099, "step": 1020 }, { "epoch": 0.027131814851705204, "grad_norm": 3.624112844467163, "learning_rate": 9.0342405618964e-06, "loss": 0.18569350242614746, "step": 1030 }, { "epoch": 0.0273952305298771, "grad_norm": 6.304218769073486, "learning_rate": 9.122036874451274e-06, "loss": 0.2043032169342041, "step": 1040 }, { "epoch": 0.027658646208048997, "grad_norm": 2.564544439315796, "learning_rate": 9.209833187006147e-06, "loss": 0.17472259998321532, "step": 1050 }, { "epoch": 0.02792206188622089, "grad_norm": 7.185754776000977, "learning_rate": 9.29762949956102e-06, "loss": 0.1770520329475403, "step": 1060 }, { "epoch": 0.028185477564392786, "grad_norm": 7.146871566772461, "learning_rate": 9.385425812115891e-06, "loss": 0.18197908401489257, "step": 1070 }, { "epoch": 0.02844889324256468, "grad_norm": 3.530480146408081, "learning_rate": 9.473222124670764e-06, "loss": 0.19095704555511475, "step": 1080 }, { "epoch": 0.028712308920736576, "grad_norm": 3.8762218952178955, "learning_rate": 9.561018437225637e-06, "loss": 0.20388154983520507, "step": 1090 }, { "epoch": 0.028975724598908473, "grad_norm": 5.993069648742676, "learning_rate": 9.64881474978051e-06, "loss": 0.19987068176269532, "step": 1100 }, { "epoch": 0.029239140277080366, "grad_norm": 4.165266036987305, "learning_rate": 9.736611062335383e-06, "loss": 0.20893466472625732, "step": 1110 }, { "epoch": 0.029502555955252262, "grad_norm": 6.6156816482543945, "learning_rate": 9.824407374890256e-06, "loss": 0.19175448417663574, "step": 1120 }, { "epoch": 0.029765971633424156, "grad_norm": 5.939418315887451, "learning_rate": 9.912203687445128e-06, "loss": 0.17565816640853882, "step": 1130 }, { "epoch": 0.030029387311596052, "grad_norm": 5.934043884277344, "learning_rate": 1e-05, "loss": 0.18105318546295165, "step": 1140 }, { "epoch": 0.03029280298976795, "grad_norm": 9.065888404846191, "learning_rate": 9.999998180392141e-06, "loss": 0.175040340423584, "step": 1150 }, { "epoch": 0.030556218667939842, "grad_norm": 4.711532115936279, "learning_rate": 9.999992721569889e-06, "loss": 0.1790454864501953, "step": 1160 }, { "epoch": 0.03081963434611174, "grad_norm": 8.532089233398438, "learning_rate": 9.999983623537215e-06, "loss": 0.17937886714935303, "step": 1170 }, { "epoch": 0.03108305002428363, "grad_norm": 6.500422954559326, "learning_rate": 9.999970886300741e-06, "loss": 0.18563116788864137, "step": 1180 }, { "epoch": 0.03134646570245553, "grad_norm": 4.196733474731445, "learning_rate": 9.999954509869741e-06, "loss": 0.18297022581100464, "step": 1190 }, { "epoch": 0.031609881380627425, "grad_norm": 4.825767993927002, "learning_rate": 9.999934494256132e-06, "loss": 0.18033648729324342, "step": 1200 }, { "epoch": 0.03187329705879932, "grad_norm": 8.849143028259277, "learning_rate": 9.999910839474482e-06, "loss": 0.1829996943473816, "step": 1210 }, { "epoch": 0.03213671273697121, "grad_norm": 6.684520244598389, "learning_rate": 9.99988354554201e-06, "loss": 0.18787226676940919, "step": 1220 }, { "epoch": 0.03240012841514311, "grad_norm": 3.6262423992156982, "learning_rate": 9.999852612478578e-06, "loss": 0.1871333599090576, "step": 1230 }, { "epoch": 0.032663544093315004, "grad_norm": 8.569589614868164, "learning_rate": 9.999818040306703e-06, "loss": 0.17978298664093018, "step": 1240 }, { "epoch": 0.0329269597714869, "grad_norm": 8.344002723693848, "learning_rate": 9.999779829051549e-06, "loss": 0.17525920867919922, "step": 1250 }, { "epoch": 0.0331903754496588, "grad_norm": 4.077969551086426, "learning_rate": 9.999737978740924e-06, "loss": 0.1870884418487549, "step": 1260 }, { "epoch": 0.03345379112783069, "grad_norm": 6.319187164306641, "learning_rate": 9.999692489405295e-06, "loss": 0.17011215686798095, "step": 1270 }, { "epoch": 0.033717206806002584, "grad_norm": 5.510404109954834, "learning_rate": 9.999643361077764e-06, "loss": 0.190757417678833, "step": 1280 }, { "epoch": 0.03398062248417448, "grad_norm": 7.065680503845215, "learning_rate": 9.999590593794093e-06, "loss": 0.18821250200271605, "step": 1290 }, { "epoch": 0.03424403816234638, "grad_norm": 6.519702911376953, "learning_rate": 9.999534187592686e-06, "loss": 0.18498709201812744, "step": 1300 }, { "epoch": 0.03450745384051827, "grad_norm": 8.461880683898926, "learning_rate": 9.9994741425146e-06, "loss": 0.16216530799865722, "step": 1310 }, { "epoch": 0.03477086951869016, "grad_norm": 3.5656545162200928, "learning_rate": 9.999410458603538e-06, "loss": 0.1864970326423645, "step": 1320 }, { "epoch": 0.03503428519686206, "grad_norm": 4.843153953552246, "learning_rate": 9.99934313590585e-06, "loss": 0.1943099856376648, "step": 1330 }, { "epoch": 0.035297700875033956, "grad_norm": 3.7693324089050293, "learning_rate": 9.99927217447054e-06, "loss": 0.1819278836250305, "step": 1340 }, { "epoch": 0.03556111655320585, "grad_norm": 2.6344802379608154, "learning_rate": 9.999197574349253e-06, "loss": 0.18364635705947877, "step": 1350 }, { "epoch": 0.03582453223137775, "grad_norm": 6.5771050453186035, "learning_rate": 9.999119335596288e-06, "loss": 0.1793527603149414, "step": 1360 }, { "epoch": 0.03608794790954964, "grad_norm": 4.72206449508667, "learning_rate": 9.99903745826859e-06, "loss": 0.1787046194076538, "step": 1370 }, { "epoch": 0.036351363587721536, "grad_norm": 9.33901309967041, "learning_rate": 9.998951942425752e-06, "loss": 0.17359887361526488, "step": 1380 }, { "epoch": 0.03661477926589343, "grad_norm": 8.251105308532715, "learning_rate": 9.99886278813002e-06, "loss": 0.1822443723678589, "step": 1390 }, { "epoch": 0.03687819494406533, "grad_norm": 7.739146709442139, "learning_rate": 9.998769995446278e-06, "loss": 0.17990760803222655, "step": 1400 }, { "epoch": 0.037141610622237226, "grad_norm": 13.497612953186035, "learning_rate": 9.998673564442069e-06, "loss": 0.18518345355987548, "step": 1410 }, { "epoch": 0.037405026300409115, "grad_norm": 5.407529830932617, "learning_rate": 9.998573495187581e-06, "loss": 0.18771715164184571, "step": 1420 }, { "epoch": 0.03766844197858101, "grad_norm": 4.613569736480713, "learning_rate": 9.998469787755645e-06, "loss": 0.18477816581726075, "step": 1430 }, { "epoch": 0.03793185765675291, "grad_norm": 5.355501174926758, "learning_rate": 9.998362442221745e-06, "loss": 0.17269386053085328, "step": 1440 }, { "epoch": 0.038195273334924805, "grad_norm": 7.305898189544678, "learning_rate": 9.99825145866401e-06, "loss": 0.18713274002075195, "step": 1450 }, { "epoch": 0.0384586890130967, "grad_norm": 4.597102642059326, "learning_rate": 9.998136837163223e-06, "loss": 0.17728980779647827, "step": 1460 }, { "epoch": 0.03872210469126859, "grad_norm": 5.449532985687256, "learning_rate": 9.998018577802807e-06, "loss": 0.18461439609527588, "step": 1470 }, { "epoch": 0.03898552036944049, "grad_norm": 4.28111457824707, "learning_rate": 9.997896680668838e-06, "loss": 0.1856493592262268, "step": 1480 }, { "epoch": 0.039248936047612384, "grad_norm": 3.8021585941314697, "learning_rate": 9.997771145850037e-06, "loss": 0.1731380581855774, "step": 1490 }, { "epoch": 0.03951235172578428, "grad_norm": 2.9554989337921143, "learning_rate": 9.997641973437774e-06, "loss": 0.17533363103866578, "step": 1500 }, { "epoch": 0.03977576740395618, "grad_norm": 4.171002388000488, "learning_rate": 9.997509163526065e-06, "loss": 0.17744704484939575, "step": 1510 }, { "epoch": 0.04003918308212807, "grad_norm": 5.216943264007568, "learning_rate": 9.997372716211577e-06, "loss": 0.17577649354934693, "step": 1520 }, { "epoch": 0.040302598760299964, "grad_norm": 11.554439544677734, "learning_rate": 9.997232631593621e-06, "loss": 0.18165099620819092, "step": 1530 }, { "epoch": 0.04056601443847186, "grad_norm": 5.1125006675720215, "learning_rate": 9.997088909774157e-06, "loss": 0.17027554512023926, "step": 1540 }, { "epoch": 0.04082943011664376, "grad_norm": 2.3249332904815674, "learning_rate": 9.99694155085779e-06, "loss": 0.18529691696166992, "step": 1550 }, { "epoch": 0.041092845794815654, "grad_norm": 3.974057674407959, "learning_rate": 9.996790554951779e-06, "loss": 0.1808495283126831, "step": 1560 }, { "epoch": 0.04135626147298754, "grad_norm": 5.702089309692383, "learning_rate": 9.99663592216602e-06, "loss": 0.18029215335845947, "step": 1570 }, { "epoch": 0.04161967715115944, "grad_norm": 6.553534030914307, "learning_rate": 9.996477652613064e-06, "loss": 0.17806292772293092, "step": 1580 }, { "epoch": 0.041883092829331336, "grad_norm": 3.196510076522827, "learning_rate": 9.996315746408107e-06, "loss": 0.16703250408172607, "step": 1590 }, { "epoch": 0.04214650850750323, "grad_norm": 8.843392372131348, "learning_rate": 9.996150203668988e-06, "loss": 0.18210901021957399, "step": 1600 }, { "epoch": 0.04240992418567512, "grad_norm": 4.603781700134277, "learning_rate": 9.995981024516202e-06, "loss": 0.16934611797332763, "step": 1610 }, { "epoch": 0.04267333986384702, "grad_norm": 4.432097434997559, "learning_rate": 9.995808209072878e-06, "loss": 0.18311220407485962, "step": 1620 }, { "epoch": 0.042936755542018916, "grad_norm": 9.392411231994629, "learning_rate": 9.995631757464803e-06, "loss": 0.1783134937286377, "step": 1630 }, { "epoch": 0.04320017122019081, "grad_norm": 5.44857120513916, "learning_rate": 9.995451669820406e-06, "loss": 0.1757124423980713, "step": 1640 }, { "epoch": 0.04346358689836271, "grad_norm": 3.0323898792266846, "learning_rate": 9.995267946270762e-06, "loss": 0.18201433420181273, "step": 1650 }, { "epoch": 0.0437270025765346, "grad_norm": 5.219616413116455, "learning_rate": 9.99508058694959e-06, "loss": 0.1845093250274658, "step": 1660 }, { "epoch": 0.043990418254706495, "grad_norm": 3.4398066997528076, "learning_rate": 9.994889591993265e-06, "loss": 0.16824851036071778, "step": 1670 }, { "epoch": 0.04425383393287839, "grad_norm": 2.923576593399048, "learning_rate": 9.994694961540793e-06, "loss": 0.17890536785125732, "step": 1680 }, { "epoch": 0.04451724961105029, "grad_norm": 3.6247940063476562, "learning_rate": 9.99449669573384e-06, "loss": 0.1725233316421509, "step": 1690 }, { "epoch": 0.044780665289222185, "grad_norm": 6.585686206817627, "learning_rate": 9.994294794716712e-06, "loss": 0.17246040105819702, "step": 1700 }, { "epoch": 0.045044080967394075, "grad_norm": 4.336049556732178, "learning_rate": 9.994089258636358e-06, "loss": 0.18124196529388428, "step": 1710 }, { "epoch": 0.04530749664556597, "grad_norm": 3.5992558002471924, "learning_rate": 9.99388008764238e-06, "loss": 0.18031561374664307, "step": 1720 }, { "epoch": 0.04557091232373787, "grad_norm": 5.828647136688232, "learning_rate": 9.993667281887019e-06, "loss": 0.16835880279541016, "step": 1730 }, { "epoch": 0.045834328001909765, "grad_norm": 4.531713962554932, "learning_rate": 9.993450841525165e-06, "loss": 0.17848966121673585, "step": 1740 }, { "epoch": 0.04609774368008166, "grad_norm": 10.467463493347168, "learning_rate": 9.993230766714354e-06, "loss": 0.17850607633590698, "step": 1750 }, { "epoch": 0.04636115935825355, "grad_norm": 6.638392448425293, "learning_rate": 9.993007057614762e-06, "loss": 0.17940282821655273, "step": 1760 }, { "epoch": 0.04662457503642545, "grad_norm": 6.660160541534424, "learning_rate": 9.99277971438922e-06, "loss": 0.18261781930923462, "step": 1770 }, { "epoch": 0.046887990714597344, "grad_norm": 3.0174694061279297, "learning_rate": 9.992548737203194e-06, "loss": 0.17863271236419678, "step": 1780 }, { "epoch": 0.04715140639276924, "grad_norm": 4.4181060791015625, "learning_rate": 9.9923141262248e-06, "loss": 0.1681815505027771, "step": 1790 }, { "epoch": 0.04741482207094114, "grad_norm": 5.652539253234863, "learning_rate": 9.992075881624799e-06, "loss": 0.19338574409484863, "step": 1800 }, { "epoch": 0.04767823774911303, "grad_norm": 3.861172676086426, "learning_rate": 9.991834003576593e-06, "loss": 0.15959618091583253, "step": 1810 }, { "epoch": 0.04794165342728492, "grad_norm": 7.027517318725586, "learning_rate": 9.991588492256235e-06, "loss": 0.18614208698272705, "step": 1820 }, { "epoch": 0.04820506910545682, "grad_norm": 2.7895519733428955, "learning_rate": 9.991339347842417e-06, "loss": 0.1840449094772339, "step": 1830 }, { "epoch": 0.04846848478362872, "grad_norm": 5.513394355773926, "learning_rate": 9.991086570516475e-06, "loss": 0.17185078859329223, "step": 1840 }, { "epoch": 0.04873190046180061, "grad_norm": 4.806138515472412, "learning_rate": 9.990830160462396e-06, "loss": 0.18247382640838622, "step": 1850 }, { "epoch": 0.0489953161399725, "grad_norm": 7.067373752593994, "learning_rate": 9.9905701178668e-06, "loss": 0.17807323932647706, "step": 1860 }, { "epoch": 0.0492587318181444, "grad_norm": 8.617301940917969, "learning_rate": 9.990306442918962e-06, "loss": 0.1764454126358032, "step": 1870 }, { "epoch": 0.049522147496316296, "grad_norm": 3.5616393089294434, "learning_rate": 9.990039135810795e-06, "loss": 0.17069439888000487, "step": 1880 }, { "epoch": 0.04978556317448819, "grad_norm": 5.709587574005127, "learning_rate": 9.989768196736858e-06, "loss": 0.1781458377838135, "step": 1890 }, { "epoch": 0.05004897885266009, "grad_norm": 4.567939758300781, "learning_rate": 9.989493625894348e-06, "loss": 0.16590038537979127, "step": 1900 }, { "epoch": 0.05031239453083198, "grad_norm": 3.0389530658721924, "learning_rate": 9.989215423483114e-06, "loss": 0.16719043254852295, "step": 1910 }, { "epoch": 0.050575810209003876, "grad_norm": 2.309375286102295, "learning_rate": 9.988933589705641e-06, "loss": 0.17616744041442872, "step": 1920 }, { "epoch": 0.05083922588717577, "grad_norm": 9.633587837219238, "learning_rate": 9.98864812476706e-06, "loss": 0.17945950031280516, "step": 1930 }, { "epoch": 0.05110264156534767, "grad_norm": 3.640535593032837, "learning_rate": 9.988359028875145e-06, "loss": 0.16640561819076538, "step": 1940 }, { "epoch": 0.051366057243519565, "grad_norm": 4.192374229431152, "learning_rate": 9.988066302240313e-06, "loss": 0.15957386493682862, "step": 1950 }, { "epoch": 0.051629472921691455, "grad_norm": 3.2084476947784424, "learning_rate": 9.987769945075625e-06, "loss": 0.16970627307891845, "step": 1960 }, { "epoch": 0.05189288859986335, "grad_norm": 5.630716800689697, "learning_rate": 9.987469957596779e-06, "loss": 0.18318512439727783, "step": 1970 }, { "epoch": 0.05215630427803525, "grad_norm": 4.690696716308594, "learning_rate": 9.987166340022119e-06, "loss": 0.17159146070480347, "step": 1980 }, { "epoch": 0.052419719956207145, "grad_norm": 5.638988971710205, "learning_rate": 9.986859092572631e-06, "loss": 0.1768037796020508, "step": 1990 }, { "epoch": 0.05268313563437904, "grad_norm": 3.2345919609069824, "learning_rate": 9.986548215471946e-06, "loss": 0.1783995270729065, "step": 2000 }, { "epoch": 0.05294655131255093, "grad_norm": 4.385927677154541, "learning_rate": 9.98623370894633e-06, "loss": 0.1681928515434265, "step": 2010 }, { "epoch": 0.05320996699072283, "grad_norm": 5.731701374053955, "learning_rate": 9.9859155732247e-06, "loss": 0.17034542560577393, "step": 2020 }, { "epoch": 0.053473382668894724, "grad_norm": 6.412173271179199, "learning_rate": 9.985593808538602e-06, "loss": 0.1800058364868164, "step": 2030 }, { "epoch": 0.05373679834706662, "grad_norm": 6.551823616027832, "learning_rate": 9.985268415122233e-06, "loss": 0.18046289682388306, "step": 2040 }, { "epoch": 0.05400021402523852, "grad_norm": 13.142145156860352, "learning_rate": 9.984939393212432e-06, "loss": 0.18263757228851318, "step": 2050 }, { "epoch": 0.05426362970341041, "grad_norm": 9.710610389709473, "learning_rate": 9.984606743048668e-06, "loss": 0.18364241123199462, "step": 2060 }, { "epoch": 0.054527045381582304, "grad_norm": 5.7839884757995605, "learning_rate": 9.984270464873066e-06, "loss": 0.1821704626083374, "step": 2070 }, { "epoch": 0.0547904610597542, "grad_norm": 5.236260414123535, "learning_rate": 9.983930558930377e-06, "loss": 0.192392098903656, "step": 2080 }, { "epoch": 0.0550538767379261, "grad_norm": 4.689997673034668, "learning_rate": 9.983587025468004e-06, "loss": 0.17405197620391846, "step": 2090 }, { "epoch": 0.05531729241609799, "grad_norm": 6.825615406036377, "learning_rate": 9.983239864735983e-06, "loss": 0.1792796492576599, "step": 2100 }, { "epoch": 0.05558070809426988, "grad_norm": 4.607245922088623, "learning_rate": 9.982889076986993e-06, "loss": 0.17602474689483644, "step": 2110 }, { "epoch": 0.05584412377244178, "grad_norm": 2.4154186248779297, "learning_rate": 9.982534662476351e-06, "loss": 0.17389429807662965, "step": 2120 }, { "epoch": 0.056107539450613676, "grad_norm": 2.812485694885254, "learning_rate": 9.982176621462019e-06, "loss": 0.17141962051391602, "step": 2130 }, { "epoch": 0.05637095512878557, "grad_norm": 6.351594924926758, "learning_rate": 9.981814954204592e-06, "loss": 0.16910955905914307, "step": 2140 }, { "epoch": 0.05663437080695747, "grad_norm": 6.654417037963867, "learning_rate": 9.981449660967308e-06, "loss": 0.17163118124008178, "step": 2150 }, { "epoch": 0.05689778648512936, "grad_norm": 3.920734405517578, "learning_rate": 9.981080742016043e-06, "loss": 0.16758649349212645, "step": 2160 }, { "epoch": 0.057161202163301256, "grad_norm": 3.127882719039917, "learning_rate": 9.980708197619312e-06, "loss": 0.1923213481903076, "step": 2170 }, { "epoch": 0.05742461784147315, "grad_norm": 9.761358261108398, "learning_rate": 9.980332028048267e-06, "loss": 0.1816110134124756, "step": 2180 }, { "epoch": 0.05768803351964505, "grad_norm": 4.610804557800293, "learning_rate": 9.979952233576704e-06, "loss": 0.18122586011886596, "step": 2190 }, { "epoch": 0.057951449197816945, "grad_norm": 4.378683567047119, "learning_rate": 9.97956881448105e-06, "loss": 0.18641605377197265, "step": 2200 }, { "epoch": 0.058214864875988835, "grad_norm": 3.810063600540161, "learning_rate": 9.979181771040379e-06, "loss": 0.17549887895584107, "step": 2210 }, { "epoch": 0.05847828055416073, "grad_norm": 8.251422882080078, "learning_rate": 9.978791103536392e-06, "loss": 0.17437469959259033, "step": 2220 }, { "epoch": 0.05874169623233263, "grad_norm": 3.3551743030548096, "learning_rate": 9.978396812253437e-06, "loss": 0.1545167088508606, "step": 2230 }, { "epoch": 0.059005111910504525, "grad_norm": 3.294245958328247, "learning_rate": 9.977998897478497e-06, "loss": 0.16629478931427003, "step": 2240 }, { "epoch": 0.05926852758867642, "grad_norm": 6.414180755615234, "learning_rate": 9.97759735950119e-06, "loss": 0.1888200044631958, "step": 2250 }, { "epoch": 0.05953194326684831, "grad_norm": 5.040812969207764, "learning_rate": 9.977192198613773e-06, "loss": 0.18157453536987306, "step": 2260 }, { "epoch": 0.05979535894502021, "grad_norm": 3.5713319778442383, "learning_rate": 9.976783415111138e-06, "loss": 0.1783649206161499, "step": 2270 }, { "epoch": 0.060058774623192104, "grad_norm": 3.637068510055542, "learning_rate": 9.976371009290817e-06, "loss": 0.18128604888916017, "step": 2280 }, { "epoch": 0.060322190301364, "grad_norm": 3.397573947906494, "learning_rate": 9.975954981452978e-06, "loss": 0.17890770435333253, "step": 2290 }, { "epoch": 0.0605856059795359, "grad_norm": 4.930329322814941, "learning_rate": 9.975535331900421e-06, "loss": 0.17515015602111816, "step": 2300 }, { "epoch": 0.06084902165770779, "grad_norm": 4.8215250968933105, "learning_rate": 9.97511206093859e-06, "loss": 0.1726822853088379, "step": 2310 }, { "epoch": 0.061112437335879684, "grad_norm": 7.905542373657227, "learning_rate": 9.974685168875553e-06, "loss": 0.1737690806388855, "step": 2320 }, { "epoch": 0.06137585301405158, "grad_norm": 4.693175315856934, "learning_rate": 9.974254656022025e-06, "loss": 0.15787088871002197, "step": 2330 }, { "epoch": 0.06163926869222348, "grad_norm": 5.123935699462891, "learning_rate": 9.97382052269135e-06, "loss": 0.18975412845611572, "step": 2340 }, { "epoch": 0.061902684370395374, "grad_norm": 6.798859596252441, "learning_rate": 9.97338276919951e-06, "loss": 0.1742071509361267, "step": 2350 }, { "epoch": 0.06216610004856726, "grad_norm": 2.482302188873291, "learning_rate": 9.972941395865123e-06, "loss": 0.17897799015045165, "step": 2360 }, { "epoch": 0.06242951572673916, "grad_norm": 2.3944039344787598, "learning_rate": 9.972496403009436e-06, "loss": 0.17237546443939208, "step": 2370 }, { "epoch": 0.06269293140491106, "grad_norm": 5.554666996002197, "learning_rate": 9.972047790956334e-06, "loss": 0.18098526000976561, "step": 2380 }, { "epoch": 0.06295634708308295, "grad_norm": 3.5538368225097656, "learning_rate": 9.971595560032336e-06, "loss": 0.16777126789093016, "step": 2390 }, { "epoch": 0.06321976276125485, "grad_norm": 4.554999828338623, "learning_rate": 9.9711397105666e-06, "loss": 0.16913223266601562, "step": 2400 }, { "epoch": 0.06348317843942675, "grad_norm": 3.3898963928222656, "learning_rate": 9.970680242890907e-06, "loss": 0.1759082317352295, "step": 2410 }, { "epoch": 0.06374659411759864, "grad_norm": 2.025590658187866, "learning_rate": 9.970217157339682e-06, "loss": 0.16734870672225952, "step": 2420 }, { "epoch": 0.06401000979577053, "grad_norm": 2.9668233394622803, "learning_rate": 9.969750454249974e-06, "loss": 0.16997015476226807, "step": 2430 }, { "epoch": 0.06427342547394242, "grad_norm": 2.3612496852874756, "learning_rate": 9.969280133961474e-06, "loss": 0.178621244430542, "step": 2440 }, { "epoch": 0.06453684115211432, "grad_norm": 2.664839744567871, "learning_rate": 9.968806196816498e-06, "loss": 0.17586450576782225, "step": 2450 }, { "epoch": 0.06480025683028622, "grad_norm": 3.5967133045196533, "learning_rate": 9.968328643160002e-06, "loss": 0.17572236061096191, "step": 2460 }, { "epoch": 0.06506367250845811, "grad_norm": 2.954834222793579, "learning_rate": 9.967847473339565e-06, "loss": 0.15950028896331786, "step": 2470 }, { "epoch": 0.06532708818663001, "grad_norm": 3.569617748260498, "learning_rate": 9.967362687705406e-06, "loss": 0.18101515769958496, "step": 2480 }, { "epoch": 0.0655905038648019, "grad_norm": 4.696358680725098, "learning_rate": 9.966874286610372e-06, "loss": 0.1677725076675415, "step": 2490 }, { "epoch": 0.0658539195429738, "grad_norm": 3.9838178157806396, "learning_rate": 9.966382270409945e-06, "loss": 0.1774571418762207, "step": 2500 }, { "epoch": 0.0661173352211457, "grad_norm": 5.277383327484131, "learning_rate": 9.965886639462234e-06, "loss": 0.17272512912750243, "step": 2510 }, { "epoch": 0.0663807508993176, "grad_norm": 2.0417838096618652, "learning_rate": 9.965387394127978e-06, "loss": 0.18972607851028442, "step": 2520 }, { "epoch": 0.06664416657748948, "grad_norm": 4.180023193359375, "learning_rate": 9.964884534770554e-06, "loss": 0.17800111770629884, "step": 2530 }, { "epoch": 0.06690758225566137, "grad_norm": 4.402682304382324, "learning_rate": 9.96437806175596e-06, "loss": 0.1782094955444336, "step": 2540 }, { "epoch": 0.06717099793383327, "grad_norm": 2.1040942668914795, "learning_rate": 9.963867975452831e-06, "loss": 0.17999727725982667, "step": 2550 }, { "epoch": 0.06743441361200517, "grad_norm": 6.498521327972412, "learning_rate": 9.963354276232433e-06, "loss": 0.18282980918884278, "step": 2560 }, { "epoch": 0.06769782929017706, "grad_norm": 2.4788119792938232, "learning_rate": 9.962836964468654e-06, "loss": 0.17045732736587524, "step": 2570 }, { "epoch": 0.06796124496834896, "grad_norm": 2.879821300506592, "learning_rate": 9.962316040538016e-06, "loss": 0.17459412813186645, "step": 2580 }, { "epoch": 0.06822466064652086, "grad_norm": 8.045357704162598, "learning_rate": 9.961791504819672e-06, "loss": 0.17094774246215821, "step": 2590 }, { "epoch": 0.06848807632469275, "grad_norm": 2.533884286880493, "learning_rate": 9.961263357695401e-06, "loss": 0.17857229709625244, "step": 2600 }, { "epoch": 0.06875149200286465, "grad_norm": 5.01926851272583, "learning_rate": 9.960731599549611e-06, "loss": 0.19099299907684325, "step": 2610 }, { "epoch": 0.06901490768103655, "grad_norm": 4.660578727722168, "learning_rate": 9.96019623076934e-06, "loss": 0.16811307668685913, "step": 2620 }, { "epoch": 0.06927832335920843, "grad_norm": 8.007450103759766, "learning_rate": 9.95965725174425e-06, "loss": 0.1757439374923706, "step": 2630 }, { "epoch": 0.06954173903738033, "grad_norm": 6.926934719085693, "learning_rate": 9.959114662866633e-06, "loss": 0.17755310535430907, "step": 2640 }, { "epoch": 0.06980515471555222, "grad_norm": 7.66856575012207, "learning_rate": 9.958568464531412e-06, "loss": 0.186178457736969, "step": 2650 }, { "epoch": 0.07006857039372412, "grad_norm": 7.592636585235596, "learning_rate": 9.95801865713613e-06, "loss": 0.17526824474334718, "step": 2660 }, { "epoch": 0.07033198607189602, "grad_norm": 1.8758985996246338, "learning_rate": 9.957465241080964e-06, "loss": 0.1783423900604248, "step": 2670 }, { "epoch": 0.07059540175006791, "grad_norm": 3.5383291244506836, "learning_rate": 9.956908216768713e-06, "loss": 0.1716325283050537, "step": 2680 }, { "epoch": 0.07085881742823981, "grad_norm": 6.753291606903076, "learning_rate": 9.9563475846048e-06, "loss": 0.1612412929534912, "step": 2690 }, { "epoch": 0.0711222331064117, "grad_norm": 2.82960844039917, "learning_rate": 9.95578334499728e-06, "loss": 0.17423734664916993, "step": 2700 }, { "epoch": 0.0713856487845836, "grad_norm": 2.954469680786133, "learning_rate": 9.955215498356834e-06, "loss": 0.17226710319519042, "step": 2710 }, { "epoch": 0.0716490644627555, "grad_norm": 5.028881549835205, "learning_rate": 9.954644045096759e-06, "loss": 0.1769021511077881, "step": 2720 }, { "epoch": 0.07191248014092738, "grad_norm": 4.052404403686523, "learning_rate": 9.954068985632988e-06, "loss": 0.19077894687652588, "step": 2730 }, { "epoch": 0.07217589581909928, "grad_norm": 4.628846168518066, "learning_rate": 9.953490320384073e-06, "loss": 0.15937019586563111, "step": 2740 }, { "epoch": 0.07243931149727117, "grad_norm": 5.416739463806152, "learning_rate": 9.952908049771191e-06, "loss": 0.16966618299484254, "step": 2750 }, { "epoch": 0.07270272717544307, "grad_norm": 3.4305198192596436, "learning_rate": 9.952322174218141e-06, "loss": 0.1744523525238037, "step": 2760 }, { "epoch": 0.07296614285361497, "grad_norm": 6.216912746429443, "learning_rate": 9.951732694151355e-06, "loss": 0.171920108795166, "step": 2770 }, { "epoch": 0.07322955853178686, "grad_norm": 3.5175063610076904, "learning_rate": 9.951139609999876e-06, "loss": 0.15650513172149658, "step": 2780 }, { "epoch": 0.07349297420995876, "grad_norm": 4.00062370300293, "learning_rate": 9.950542922195379e-06, "loss": 0.18186039924621583, "step": 2790 }, { "epoch": 0.07375638988813066, "grad_norm": 2.4293506145477295, "learning_rate": 9.949942631172158e-06, "loss": 0.16667344570159912, "step": 2800 }, { "epoch": 0.07401980556630255, "grad_norm": 3.579925775527954, "learning_rate": 9.949338737367133e-06, "loss": 0.1642069101333618, "step": 2810 }, { "epoch": 0.07428322124447445, "grad_norm": 7.0450639724731445, "learning_rate": 9.948731241219842e-06, "loss": 0.1777646064758301, "step": 2820 }, { "epoch": 0.07454663692264633, "grad_norm": 1.835729956626892, "learning_rate": 9.948120143172446e-06, "loss": 0.17344406843185425, "step": 2830 }, { "epoch": 0.07481005260081823, "grad_norm": 5.397601127624512, "learning_rate": 9.94750544366973e-06, "loss": 0.17330164909362794, "step": 2840 }, { "epoch": 0.07507346827899013, "grad_norm": 3.4178895950317383, "learning_rate": 9.946887143159099e-06, "loss": 0.17044928073883056, "step": 2850 }, { "epoch": 0.07533688395716202, "grad_norm": 3.047327995300293, "learning_rate": 9.946265242090577e-06, "loss": 0.17530510425567628, "step": 2860 }, { "epoch": 0.07560029963533392, "grad_norm": 2.7362687587738037, "learning_rate": 9.945639740916816e-06, "loss": 0.1693619966506958, "step": 2870 }, { "epoch": 0.07586371531350582, "grad_norm": 2.561385154724121, "learning_rate": 9.945010640093074e-06, "loss": 0.17678675651550294, "step": 2880 }, { "epoch": 0.07612713099167771, "grad_norm": 3.140603542327881, "learning_rate": 9.944377940077243e-06, "loss": 0.16699635982513428, "step": 2890 }, { "epoch": 0.07639054666984961, "grad_norm": 4.115111827850342, "learning_rate": 9.94374164132983e-06, "loss": 0.18036831617355348, "step": 2900 }, { "epoch": 0.0766539623480215, "grad_norm": 2.8503637313842773, "learning_rate": 9.943101744313959e-06, "loss": 0.17674968242645264, "step": 2910 }, { "epoch": 0.0769173780261934, "grad_norm": 3.1077005863189697, "learning_rate": 9.942458249495374e-06, "loss": 0.16769458055496217, "step": 2920 }, { "epoch": 0.07718079370436529, "grad_norm": 9.071978569030762, "learning_rate": 9.94181115734244e-06, "loss": 0.1688298225402832, "step": 2930 }, { "epoch": 0.07744420938253718, "grad_norm": 2.3998019695281982, "learning_rate": 9.941160468326139e-06, "loss": 0.17626709938049318, "step": 2940 }, { "epoch": 0.07770762506070908, "grad_norm": 6.1342034339904785, "learning_rate": 9.940506182920069e-06, "loss": 0.175899875164032, "step": 2950 }, { "epoch": 0.07797104073888098, "grad_norm": 2.5122575759887695, "learning_rate": 9.939848301600445e-06, "loss": 0.17090299129486083, "step": 2960 }, { "epoch": 0.07823445641705287, "grad_norm": 4.607466697692871, "learning_rate": 9.939186824846106e-06, "loss": 0.166940176486969, "step": 2970 }, { "epoch": 0.07849787209522477, "grad_norm": 3.417919397354126, "learning_rate": 9.938521753138502e-06, "loss": 0.16935186386108397, "step": 2980 }, { "epoch": 0.07876128777339667, "grad_norm": 2.15596866607666, "learning_rate": 9.937853086961698e-06, "loss": 0.16227587461471557, "step": 2990 }, { "epoch": 0.07902470345156856, "grad_norm": 4.537834644317627, "learning_rate": 9.937180826802382e-06, "loss": 0.16452592611312866, "step": 3000 }, { "epoch": 0.07928811912974046, "grad_norm": 3.403853416442871, "learning_rate": 9.936504973149851e-06, "loss": 0.18245265483856202, "step": 3010 }, { "epoch": 0.07955153480791236, "grad_norm": 3.6958892345428467, "learning_rate": 9.935825526496023e-06, "loss": 0.15633554458618165, "step": 3020 }, { "epoch": 0.07981495048608424, "grad_norm": 4.926299571990967, "learning_rate": 9.935142487335425e-06, "loss": 0.17498354911804198, "step": 3030 }, { "epoch": 0.08007836616425613, "grad_norm": 2.2159178256988525, "learning_rate": 9.934455856165206e-06, "loss": 0.17082390785217286, "step": 3040 }, { "epoch": 0.08034178184242803, "grad_norm": 3.392390012741089, "learning_rate": 9.933765633485124e-06, "loss": 0.17010934352874757, "step": 3050 }, { "epoch": 0.08060519752059993, "grad_norm": 5.884143829345703, "learning_rate": 9.933071819797554e-06, "loss": 0.17608139514923096, "step": 3060 }, { "epoch": 0.08086861319877182, "grad_norm": 1.9652087688446045, "learning_rate": 9.932374415607482e-06, "loss": 0.16665477752685548, "step": 3070 }, { "epoch": 0.08113202887694372, "grad_norm": 3.39043927192688, "learning_rate": 9.93167342142251e-06, "loss": 0.17293229103088378, "step": 3080 }, { "epoch": 0.08139544455511562, "grad_norm": 2.772920846939087, "learning_rate": 9.93096883775285e-06, "loss": 0.17765508890151976, "step": 3090 }, { "epoch": 0.08165886023328751, "grad_norm": 4.838275909423828, "learning_rate": 9.93026066511133e-06, "loss": 0.18570774793624878, "step": 3100 }, { "epoch": 0.08192227591145941, "grad_norm": 2.683899164199829, "learning_rate": 9.92954890401339e-06, "loss": 0.1823618769645691, "step": 3110 }, { "epoch": 0.08218569158963131, "grad_norm": 4.381089687347412, "learning_rate": 9.92883355497708e-06, "loss": 0.17548794746398927, "step": 3120 }, { "epoch": 0.08244910726780319, "grad_norm": 3.9094552993774414, "learning_rate": 9.928114618523057e-06, "loss": 0.17378846406936646, "step": 3130 }, { "epoch": 0.08271252294597509, "grad_norm": 3.8950488567352295, "learning_rate": 9.927392095174601e-06, "loss": 0.17923357486724853, "step": 3140 }, { "epoch": 0.08297593862414698, "grad_norm": 2.716707229614258, "learning_rate": 9.92666598545759e-06, "loss": 0.17810927629470824, "step": 3150 }, { "epoch": 0.08323935430231888, "grad_norm": 2.6691155433654785, "learning_rate": 9.925936289900522e-06, "loss": 0.18233566284179686, "step": 3160 }, { "epoch": 0.08350276998049078, "grad_norm": 3.9785141944885254, "learning_rate": 9.925203009034498e-06, "loss": 0.1608000636100769, "step": 3170 }, { "epoch": 0.08376618565866267, "grad_norm": 4.4101338386535645, "learning_rate": 9.924466143393235e-06, "loss": 0.17613718509674073, "step": 3180 }, { "epoch": 0.08402960133683457, "grad_norm": 3.0034584999084473, "learning_rate": 9.923725693513052e-06, "loss": 0.17042592763900757, "step": 3190 }, { "epoch": 0.08429301701500647, "grad_norm": 2.6542105674743652, "learning_rate": 9.922981659932879e-06, "loss": 0.1745414614677429, "step": 3200 }, { "epoch": 0.08455643269317836, "grad_norm": 3.0383827686309814, "learning_rate": 9.922234043194261e-06, "loss": 0.17171072959899902, "step": 3210 }, { "epoch": 0.08481984837135025, "grad_norm": 4.414712905883789, "learning_rate": 9.921482843841343e-06, "loss": 0.16019524335861207, "step": 3220 }, { "epoch": 0.08508326404952214, "grad_norm": 5.328068733215332, "learning_rate": 9.920728062420881e-06, "loss": 0.15469095706939698, "step": 3230 }, { "epoch": 0.08534667972769404, "grad_norm": 3.429518461227417, "learning_rate": 9.919969699482236e-06, "loss": 0.1688215970993042, "step": 3240 }, { "epoch": 0.08561009540586594, "grad_norm": 4.567251682281494, "learning_rate": 9.919207755577378e-06, "loss": 0.176293683052063, "step": 3250 }, { "epoch": 0.08587351108403783, "grad_norm": 5.307624340057373, "learning_rate": 9.918442231260884e-06, "loss": 0.1755120038986206, "step": 3260 }, { "epoch": 0.08613692676220973, "grad_norm": 2.8510897159576416, "learning_rate": 9.917673127089932e-06, "loss": 0.1694265842437744, "step": 3270 }, { "epoch": 0.08640034244038163, "grad_norm": 4.382963180541992, "learning_rate": 9.916900443624314e-06, "loss": 0.17118558883666993, "step": 3280 }, { "epoch": 0.08666375811855352, "grad_norm": 1.934726357460022, "learning_rate": 9.91612418142642e-06, "loss": 0.17842687368392945, "step": 3290 }, { "epoch": 0.08692717379672542, "grad_norm": 4.895179271697998, "learning_rate": 9.915344341061247e-06, "loss": 0.17419590950012206, "step": 3300 }, { "epoch": 0.08719058947489731, "grad_norm": 2.7725980281829834, "learning_rate": 9.914560923096395e-06, "loss": 0.16590306758880616, "step": 3310 }, { "epoch": 0.0874540051530692, "grad_norm": 6.229907989501953, "learning_rate": 9.913773928102074e-06, "loss": 0.16355650424957274, "step": 3320 }, { "epoch": 0.0877174208312411, "grad_norm": 3.8654050827026367, "learning_rate": 9.912983356651086e-06, "loss": 0.16611862182617188, "step": 3330 }, { "epoch": 0.08798083650941299, "grad_norm": 2.6104843616485596, "learning_rate": 9.91218920931885e-06, "loss": 0.1624467968940735, "step": 3340 }, { "epoch": 0.08824425218758489, "grad_norm": 3.7739033699035645, "learning_rate": 9.911391486683375e-06, "loss": 0.18190379142761232, "step": 3350 }, { "epoch": 0.08850766786575678, "grad_norm": 2.7176663875579834, "learning_rate": 9.910590189325281e-06, "loss": 0.18036720752716065, "step": 3360 }, { "epoch": 0.08877108354392868, "grad_norm": 4.0332112312316895, "learning_rate": 9.909785317827789e-06, "loss": 0.1651561975479126, "step": 3370 }, { "epoch": 0.08903449922210058, "grad_norm": 2.397739887237549, "learning_rate": 9.908976872776715e-06, "loss": 0.16609110832214355, "step": 3380 }, { "epoch": 0.08929791490027247, "grad_norm": 3.128810405731201, "learning_rate": 9.90816485476048e-06, "loss": 0.1626802444458008, "step": 3390 }, { "epoch": 0.08956133057844437, "grad_norm": 2.5435752868652344, "learning_rate": 9.907349264370111e-06, "loss": 0.15757231712341307, "step": 3400 }, { "epoch": 0.08982474625661627, "grad_norm": 3.1860246658325195, "learning_rate": 9.906530102199224e-06, "loss": 0.16501498222351074, "step": 3410 }, { "epoch": 0.09008816193478815, "grad_norm": 2.746338129043579, "learning_rate": 9.905707368844044e-06, "loss": 0.1728563666343689, "step": 3420 }, { "epoch": 0.09035157761296005, "grad_norm": 2.4093639850616455, "learning_rate": 9.904881064903388e-06, "loss": 0.17124122381210327, "step": 3430 }, { "epoch": 0.09061499329113194, "grad_norm": 4.1353349685668945, "learning_rate": 9.904051190978682e-06, "loss": 0.1711814284324646, "step": 3440 }, { "epoch": 0.09087840896930384, "grad_norm": 5.948703289031982, "learning_rate": 9.90321774767394e-06, "loss": 0.17230840921401977, "step": 3450 }, { "epoch": 0.09114182464747574, "grad_norm": 2.8386919498443604, "learning_rate": 9.902380735595775e-06, "loss": 0.16732176542282104, "step": 3460 }, { "epoch": 0.09140524032564763, "grad_norm": 2.7843146324157715, "learning_rate": 9.90154015535341e-06, "loss": 0.17970805168151854, "step": 3470 }, { "epoch": 0.09166865600381953, "grad_norm": 6.109985828399658, "learning_rate": 9.900696007558643e-06, "loss": 0.1796471118927002, "step": 3480 }, { "epoch": 0.09193207168199143, "grad_norm": 3.6352314949035645, "learning_rate": 9.899848292825891e-06, "loss": 0.170535409450531, "step": 3490 }, { "epoch": 0.09219548736016332, "grad_norm": 4.128434658050537, "learning_rate": 9.898997011772153e-06, "loss": 0.15852088928222657, "step": 3500 }, { "epoch": 0.09245890303833522, "grad_norm": 6.701513290405273, "learning_rate": 9.89814216501703e-06, "loss": 0.16977696418762206, "step": 3510 }, { "epoch": 0.0927223187165071, "grad_norm": 5.3691816329956055, "learning_rate": 9.897283753182716e-06, "loss": 0.16492484807968139, "step": 3520 }, { "epoch": 0.092985734394679, "grad_norm": 3.401210069656372, "learning_rate": 9.896421776894e-06, "loss": 0.15144444704055787, "step": 3530 }, { "epoch": 0.0932491500728509, "grad_norm": 2.752540111541748, "learning_rate": 9.895556236778264e-06, "loss": 0.18635543584823608, "step": 3540 }, { "epoch": 0.09351256575102279, "grad_norm": 3.0596628189086914, "learning_rate": 9.894687133465485e-06, "loss": 0.17428913116455078, "step": 3550 }, { "epoch": 0.09377598142919469, "grad_norm": 5.025325775146484, "learning_rate": 9.893814467588238e-06, "loss": 0.18881447315216066, "step": 3560 }, { "epoch": 0.09403939710736658, "grad_norm": 2.672713279724121, "learning_rate": 9.892938239781684e-06, "loss": 0.1793952226638794, "step": 3570 }, { "epoch": 0.09430281278553848, "grad_norm": 3.1337313652038574, "learning_rate": 9.89205845068358e-06, "loss": 0.16922385692596437, "step": 3580 }, { "epoch": 0.09456622846371038, "grad_norm": 3.6950156688690186, "learning_rate": 9.891175100934273e-06, "loss": 0.16640559434890748, "step": 3590 }, { "epoch": 0.09482964414188227, "grad_norm": 3.4377262592315674, "learning_rate": 9.890288191176705e-06, "loss": 0.157254159450531, "step": 3600 }, { "epoch": 0.09509305982005417, "grad_norm": 3.561224937438965, "learning_rate": 9.889397722056406e-06, "loss": 0.16631264686584474, "step": 3610 }, { "epoch": 0.09535647549822605, "grad_norm": 2.398782730102539, "learning_rate": 9.8885036942215e-06, "loss": 0.18043169975280762, "step": 3620 }, { "epoch": 0.09561989117639795, "grad_norm": 4.002199172973633, "learning_rate": 9.887606108322696e-06, "loss": 0.1701803207397461, "step": 3630 }, { "epoch": 0.09588330685456985, "grad_norm": 3.212526321411133, "learning_rate": 9.886704965013296e-06, "loss": 0.15747467279434205, "step": 3640 }, { "epoch": 0.09614672253274174, "grad_norm": 4.185349464416504, "learning_rate": 9.885800264949195e-06, "loss": 0.17131781578063965, "step": 3650 }, { "epoch": 0.09641013821091364, "grad_norm": 5.12830924987793, "learning_rate": 9.884892008788867e-06, "loss": 0.1774187922477722, "step": 3660 }, { "epoch": 0.09667355388908554, "grad_norm": 6.301506996154785, "learning_rate": 9.883980197193384e-06, "loss": 0.17564842700958253, "step": 3670 }, { "epoch": 0.09693696956725743, "grad_norm": 4.441055774688721, "learning_rate": 9.883064830826402e-06, "loss": 0.17102891206741333, "step": 3680 }, { "epoch": 0.09720038524542933, "grad_norm": 6.233587741851807, "learning_rate": 9.882145910354162e-06, "loss": 0.17215858697891234, "step": 3690 }, { "epoch": 0.09746380092360123, "grad_norm": 2.8411970138549805, "learning_rate": 9.881223436445492e-06, "loss": 0.16051944494247436, "step": 3700 }, { "epoch": 0.09772721660177312, "grad_norm": 4.1526384353637695, "learning_rate": 9.880297409771815e-06, "loss": 0.17355886697769166, "step": 3710 }, { "epoch": 0.097990632279945, "grad_norm": 4.603363037109375, "learning_rate": 9.879367831007127e-06, "loss": 0.19117426872253418, "step": 3720 }, { "epoch": 0.0982540479581169, "grad_norm": 3.160926103591919, "learning_rate": 9.87843470082802e-06, "loss": 0.17697830200195314, "step": 3730 }, { "epoch": 0.0985174636362888, "grad_norm": 2.9471399784088135, "learning_rate": 9.877498019913661e-06, "loss": 0.15404584407806396, "step": 3740 }, { "epoch": 0.0987808793144607, "grad_norm": 2.283642053604126, "learning_rate": 9.876557788945812e-06, "loss": 0.17097052335739135, "step": 3750 }, { "epoch": 0.09904429499263259, "grad_norm": 4.767331123352051, "learning_rate": 9.875614008608812e-06, "loss": 0.15659927129745482, "step": 3760 }, { "epoch": 0.09930771067080449, "grad_norm": 3.572770357131958, "learning_rate": 9.874666679589583e-06, "loss": 0.1654239535331726, "step": 3770 }, { "epoch": 0.09957112634897639, "grad_norm": 2.073718547821045, "learning_rate": 9.873715802577635e-06, "loss": 0.1678226113319397, "step": 3780 }, { "epoch": 0.09983454202714828, "grad_norm": 3.2708232402801514, "learning_rate": 9.872761378265054e-06, "loss": 0.1641343355178833, "step": 3790 }, { "epoch": 0.10009795770532018, "grad_norm": 1.5196765661239624, "learning_rate": 9.871803407346516e-06, "loss": 0.15393414497375488, "step": 3800 }, { "epoch": 0.10036137338349208, "grad_norm": 2.5738019943237305, "learning_rate": 9.870841890519269e-06, "loss": 0.1603121876716614, "step": 3810 }, { "epoch": 0.10062478906166396, "grad_norm": 5.439716815948486, "learning_rate": 9.869876828483147e-06, "loss": 0.17558579444885253, "step": 3820 }, { "epoch": 0.10088820473983585, "grad_norm": 3.0882883071899414, "learning_rate": 9.868908221940567e-06, "loss": 0.1755610227584839, "step": 3830 }, { "epoch": 0.10115162041800775, "grad_norm": 4.240593433380127, "learning_rate": 9.86793607159652e-06, "loss": 0.16080873012542723, "step": 3840 }, { "epoch": 0.10141503609617965, "grad_norm": 2.2652337551116943, "learning_rate": 9.866960378158579e-06, "loss": 0.1757303237915039, "step": 3850 }, { "epoch": 0.10167845177435154, "grad_norm": 3.19911527633667, "learning_rate": 9.865981142336895e-06, "loss": 0.17372434139251708, "step": 3860 }, { "epoch": 0.10194186745252344, "grad_norm": 7.881778717041016, "learning_rate": 9.864998364844201e-06, "loss": 0.1807868719100952, "step": 3870 }, { "epoch": 0.10220528313069534, "grad_norm": 2.1664397716522217, "learning_rate": 9.864012046395803e-06, "loss": 0.16683511734008788, "step": 3880 }, { "epoch": 0.10246869880886723, "grad_norm": 8.275367736816406, "learning_rate": 9.863022187709586e-06, "loss": 0.15736680030822753, "step": 3890 }, { "epoch": 0.10273211448703913, "grad_norm": 4.21075439453125, "learning_rate": 9.862028789506013e-06, "loss": 0.1720592498779297, "step": 3900 }, { "epoch": 0.10299553016521103, "grad_norm": 4.109136581420898, "learning_rate": 9.86103185250812e-06, "loss": 0.16696598529815673, "step": 3910 }, { "epoch": 0.10325894584338291, "grad_norm": 2.2028980255126953, "learning_rate": 9.860031377441524e-06, "loss": 0.1768753409385681, "step": 3920 }, { "epoch": 0.1035223615215548, "grad_norm": 2.864097833633423, "learning_rate": 9.85902736503441e-06, "loss": 0.1822575807571411, "step": 3930 }, { "epoch": 0.1037857771997267, "grad_norm": 3.7098782062530518, "learning_rate": 9.858019816017543e-06, "loss": 0.16604347229003907, "step": 3940 }, { "epoch": 0.1040491928778986, "grad_norm": 4.514284133911133, "learning_rate": 9.857008731124262e-06, "loss": 0.17571552991867065, "step": 3950 }, { "epoch": 0.1043126085560705, "grad_norm": 1.9210429191589355, "learning_rate": 9.855994111090479e-06, "loss": 0.1641901731491089, "step": 3960 }, { "epoch": 0.10457602423424239, "grad_norm": 5.029963493347168, "learning_rate": 9.854975956654674e-06, "loss": 0.1734936475753784, "step": 3970 }, { "epoch": 0.10483943991241429, "grad_norm": 4.544711112976074, "learning_rate": 9.853954268557908e-06, "loss": 0.16894351243972777, "step": 3980 }, { "epoch": 0.10510285559058619, "grad_norm": 2.8703453540802, "learning_rate": 9.852929047543806e-06, "loss": 0.17181472778320311, "step": 3990 }, { "epoch": 0.10536627126875808, "grad_norm": 2.036325693130493, "learning_rate": 9.851900294358572e-06, "loss": 0.16685943603515624, "step": 4000 }, { "epoch": 0.10562968694692998, "grad_norm": 2.4548354148864746, "learning_rate": 9.850868009750974e-06, "loss": 0.15847234725952147, "step": 4010 }, { "epoch": 0.10589310262510186, "grad_norm": 2.214195728302002, "learning_rate": 9.849832194472355e-06, "loss": 0.16268811225891114, "step": 4020 }, { "epoch": 0.10615651830327376, "grad_norm": 4.237188339233398, "learning_rate": 9.848792849276625e-06, "loss": 0.16978241205215455, "step": 4030 }, { "epoch": 0.10641993398144566, "grad_norm": 4.119980335235596, "learning_rate": 9.847749974920265e-06, "loss": 0.16910700798034667, "step": 4040 }, { "epoch": 0.10668334965961755, "grad_norm": 3.436638593673706, "learning_rate": 9.846703572162321e-06, "loss": 0.15880491733551025, "step": 4050 }, { "epoch": 0.10694676533778945, "grad_norm": 2.3937947750091553, "learning_rate": 9.845653641764417e-06, "loss": 0.159345281124115, "step": 4060 }, { "epoch": 0.10721018101596134, "grad_norm": 2.7862117290496826, "learning_rate": 9.84460018449073e-06, "loss": 0.17379924058914184, "step": 4070 }, { "epoch": 0.10747359669413324, "grad_norm": 2.503821611404419, "learning_rate": 9.843543201108016e-06, "loss": 0.16549936532974244, "step": 4080 }, { "epoch": 0.10773701237230514, "grad_norm": 3.844198226928711, "learning_rate": 9.84248269238559e-06, "loss": 0.174782133102417, "step": 4090 }, { "epoch": 0.10800042805047703, "grad_norm": 7.247418403625488, "learning_rate": 9.841418659095342e-06, "loss": 0.17283053398132325, "step": 4100 }, { "epoch": 0.10826384372864893, "grad_norm": 3.4268503189086914, "learning_rate": 9.840351102011714e-06, "loss": 0.17175915241241455, "step": 4110 }, { "epoch": 0.10852725940682081, "grad_norm": 3.157460927963257, "learning_rate": 9.839280021911725e-06, "loss": 0.17053359746932983, "step": 4120 }, { "epoch": 0.10879067508499271, "grad_norm": 1.2920844554901123, "learning_rate": 9.83820541957495e-06, "loss": 0.16356279850006103, "step": 4130 }, { "epoch": 0.10905409076316461, "grad_norm": 3.6077077388763428, "learning_rate": 9.837127295783536e-06, "loss": 0.16810119152069092, "step": 4140 }, { "epoch": 0.1093175064413365, "grad_norm": 2.4065489768981934, "learning_rate": 9.836045651322182e-06, "loss": 0.17232627868652345, "step": 4150 }, { "epoch": 0.1095809221195084, "grad_norm": 1.9678699970245361, "learning_rate": 9.83496048697816e-06, "loss": 0.16805639266967773, "step": 4160 }, { "epoch": 0.1098443377976803, "grad_norm": 2.5184433460235596, "learning_rate": 9.833871803541297e-06, "loss": 0.15849825143814086, "step": 4170 }, { "epoch": 0.1101077534758522, "grad_norm": 3.3827006816864014, "learning_rate": 9.832779601803983e-06, "loss": 0.16413098573684692, "step": 4180 }, { "epoch": 0.11037116915402409, "grad_norm": 3.1858675479888916, "learning_rate": 9.831683882561173e-06, "loss": 0.16094229221343995, "step": 4190 }, { "epoch": 0.11063458483219599, "grad_norm": 3.274988889694214, "learning_rate": 9.830584646610374e-06, "loss": 0.18580846786499022, "step": 4200 }, { "epoch": 0.11089800051036787, "grad_norm": 3.2807934284210205, "learning_rate": 9.829481894751664e-06, "loss": 0.1616978645324707, "step": 4210 }, { "epoch": 0.11116141618853977, "grad_norm": 2.3957319259643555, "learning_rate": 9.828375627787668e-06, "loss": 0.16793627738952638, "step": 4220 }, { "epoch": 0.11142483186671166, "grad_norm": 2.511997938156128, "learning_rate": 9.827265846523575e-06, "loss": 0.18320152759552003, "step": 4230 }, { "epoch": 0.11168824754488356, "grad_norm": 6.297463893890381, "learning_rate": 9.826152551767133e-06, "loss": 0.1684071898460388, "step": 4240 }, { "epoch": 0.11195166322305546, "grad_norm": 5.183200836181641, "learning_rate": 9.825035744328646e-06, "loss": 0.1817143201828003, "step": 4250 }, { "epoch": 0.11221507890122735, "grad_norm": 3.79255747795105, "learning_rate": 9.823915425020976e-06, "loss": 0.17427668571472169, "step": 4260 }, { "epoch": 0.11247849457939925, "grad_norm": 2.95603346824646, "learning_rate": 9.822791594659537e-06, "loss": 0.17544338703155518, "step": 4270 }, { "epoch": 0.11274191025757115, "grad_norm": 2.32023024559021, "learning_rate": 9.821664254062302e-06, "loss": 0.1702479362487793, "step": 4280 }, { "epoch": 0.11300532593574304, "grad_norm": 4.39347505569458, "learning_rate": 9.820533404049799e-06, "loss": 0.156931734085083, "step": 4290 }, { "epoch": 0.11326874161391494, "grad_norm": 3.8751087188720703, "learning_rate": 9.81939904544511e-06, "loss": 0.169882333278656, "step": 4300 }, { "epoch": 0.11353215729208682, "grad_norm": 3.173220157623291, "learning_rate": 9.818261179073868e-06, "loss": 0.16245390176773072, "step": 4310 }, { "epoch": 0.11379557297025872, "grad_norm": 2.113257884979248, "learning_rate": 9.817119805764264e-06, "loss": 0.16759731769561767, "step": 4320 }, { "epoch": 0.11405898864843061, "grad_norm": 3.2754828929901123, "learning_rate": 9.815974926347037e-06, "loss": 0.17754795551300048, "step": 4330 }, { "epoch": 0.11432240432660251, "grad_norm": 2.027573823928833, "learning_rate": 9.814826541655478e-06, "loss": 0.15920422077178956, "step": 4340 }, { "epoch": 0.11458582000477441, "grad_norm": 2.4571707248687744, "learning_rate": 9.813674652525434e-06, "loss": 0.16064209938049318, "step": 4350 }, { "epoch": 0.1148492356829463, "grad_norm": 6.10903263092041, "learning_rate": 9.812519259795298e-06, "loss": 0.17641305923461914, "step": 4360 }, { "epoch": 0.1151126513611182, "grad_norm": 1.6925591230392456, "learning_rate": 9.811360364306016e-06, "loss": 0.1691579580307007, "step": 4370 }, { "epoch": 0.1153760670392901, "grad_norm": 2.5731632709503174, "learning_rate": 9.810197966901079e-06, "loss": 0.16652841567993165, "step": 4380 }, { "epoch": 0.115639482717462, "grad_norm": 5.663952827453613, "learning_rate": 9.809032068426533e-06, "loss": 0.16285874843597412, "step": 4390 }, { "epoch": 0.11590289839563389, "grad_norm": 4.1746978759765625, "learning_rate": 9.807862669730968e-06, "loss": 0.17178828716278077, "step": 4400 }, { "epoch": 0.11616631407380577, "grad_norm": 2.9327147006988525, "learning_rate": 9.806689771665525e-06, "loss": 0.16871594190597533, "step": 4410 }, { "epoch": 0.11642972975197767, "grad_norm": 3.93292236328125, "learning_rate": 9.805513375083885e-06, "loss": 0.15886952877044677, "step": 4420 }, { "epoch": 0.11669314543014957, "grad_norm": 3.637643575668335, "learning_rate": 9.804333480842285e-06, "loss": 0.1664292335510254, "step": 4430 }, { "epoch": 0.11695656110832146, "grad_norm": 1.8327094316482544, "learning_rate": 9.803150089799502e-06, "loss": 0.16783179044723512, "step": 4440 }, { "epoch": 0.11721997678649336, "grad_norm": 1.9156196117401123, "learning_rate": 9.801963202816856e-06, "loss": 0.16294782161712645, "step": 4450 }, { "epoch": 0.11748339246466526, "grad_norm": 2.908419132232666, "learning_rate": 9.800772820758217e-06, "loss": 0.16427005529403688, "step": 4460 }, { "epoch": 0.11774680814283715, "grad_norm": 3.6535825729370117, "learning_rate": 9.799578944489998e-06, "loss": 0.1684000849723816, "step": 4470 }, { "epoch": 0.11801022382100905, "grad_norm": 1.546571969985962, "learning_rate": 9.79838157488115e-06, "loss": 0.16413952112197877, "step": 4480 }, { "epoch": 0.11827363949918095, "grad_norm": 5.033501148223877, "learning_rate": 9.797180712803174e-06, "loss": 0.16789562702178956, "step": 4490 }, { "epoch": 0.11853705517735284, "grad_norm": 6.079884052276611, "learning_rate": 9.795976359130104e-06, "loss": 0.16320583820343018, "step": 4500 }, { "epoch": 0.11880047085552473, "grad_norm": 2.4318063259124756, "learning_rate": 9.794768514738526e-06, "loss": 0.190287184715271, "step": 4510 }, { "epoch": 0.11906388653369662, "grad_norm": 4.5098557472229, "learning_rate": 9.793557180507559e-06, "loss": 0.16669838428497313, "step": 4520 }, { "epoch": 0.11932730221186852, "grad_norm": 5.393552303314209, "learning_rate": 9.792342357318864e-06, "loss": 0.16669390201568604, "step": 4530 }, { "epoch": 0.11959071789004042, "grad_norm": 3.4502272605895996, "learning_rate": 9.791124046056643e-06, "loss": 0.16174768209457396, "step": 4540 }, { "epoch": 0.11985413356821231, "grad_norm": 1.8575506210327148, "learning_rate": 9.789902247607634e-06, "loss": 0.1610277533531189, "step": 4550 }, { "epoch": 0.12011754924638421, "grad_norm": 2.6013693809509277, "learning_rate": 9.788676962861114e-06, "loss": 0.15605707168579103, "step": 4560 }, { "epoch": 0.1203809649245561, "grad_norm": 5.321904182434082, "learning_rate": 9.787448192708901e-06, "loss": 0.15303454399108887, "step": 4570 }, { "epoch": 0.120644380602728, "grad_norm": 4.009253978729248, "learning_rate": 9.786215938045345e-06, "loss": 0.17740910053253173, "step": 4580 }, { "epoch": 0.1209077962808999, "grad_norm": 3.2433927059173584, "learning_rate": 9.784980199767334e-06, "loss": 0.16623291969299317, "step": 4590 }, { "epoch": 0.1211712119590718, "grad_norm": 2.9820473194122314, "learning_rate": 9.783740978774293e-06, "loss": 0.16785770654678345, "step": 4600 }, { "epoch": 0.12143462763724368, "grad_norm": 2.8499245643615723, "learning_rate": 9.782498275968177e-06, "loss": 0.15386636257171632, "step": 4610 }, { "epoch": 0.12169804331541557, "grad_norm": 6.944137096405029, "learning_rate": 9.781252092253483e-06, "loss": 0.16960813999176025, "step": 4620 }, { "epoch": 0.12196145899358747, "grad_norm": 2.020089626312256, "learning_rate": 9.780002428537236e-06, "loss": 0.15778486728668212, "step": 4630 }, { "epoch": 0.12222487467175937, "grad_norm": 1.9659085273742676, "learning_rate": 9.778749285728995e-06, "loss": 0.18655229806900026, "step": 4640 }, { "epoch": 0.12248829034993126, "grad_norm": 4.004394054412842, "learning_rate": 9.77749266474085e-06, "loss": 0.16540874242782594, "step": 4650 }, { "epoch": 0.12275170602810316, "grad_norm": 2.441570997238159, "learning_rate": 9.776232566487425e-06, "loss": 0.16210159063339233, "step": 4660 }, { "epoch": 0.12301512170627506, "grad_norm": 2.0107946395874023, "learning_rate": 9.774968991885874e-06, "loss": 0.1746441125869751, "step": 4670 }, { "epoch": 0.12327853738444695, "grad_norm": 1.7371212244033813, "learning_rate": 9.77370194185588e-06, "loss": 0.15772935152053832, "step": 4680 }, { "epoch": 0.12354195306261885, "grad_norm": 7.0804219245910645, "learning_rate": 9.772431417319658e-06, "loss": 0.1558934211730957, "step": 4690 }, { "epoch": 0.12380536874079075, "grad_norm": 2.772345542907715, "learning_rate": 9.771157419201953e-06, "loss": 0.16216224431991577, "step": 4700 }, { "epoch": 0.12406878441896263, "grad_norm": 4.577585697174072, "learning_rate": 9.769879948430032e-06, "loss": 0.17400180101394652, "step": 4710 }, { "epoch": 0.12433220009713453, "grad_norm": 1.6157857179641724, "learning_rate": 9.768599005933692e-06, "loss": 0.1730104446411133, "step": 4720 }, { "epoch": 0.12459561577530642, "grad_norm": 1.9515554904937744, "learning_rate": 9.767314592645262e-06, "loss": 0.17084411382675171, "step": 4730 }, { "epoch": 0.12485903145347832, "grad_norm": 4.80374002456665, "learning_rate": 9.766026709499593e-06, "loss": 0.16273683309555054, "step": 4740 }, { "epoch": 0.12512244713165022, "grad_norm": 4.336164474487305, "learning_rate": 9.76473535743406e-06, "loss": 0.16135666370391846, "step": 4750 }, { "epoch": 0.1253858628098221, "grad_norm": 4.102790832519531, "learning_rate": 9.763440537388566e-06, "loss": 0.1918026924133301, "step": 4760 }, { "epoch": 0.125649278487994, "grad_norm": 5.267177104949951, "learning_rate": 9.762142250305538e-06, "loss": 0.16765716075897216, "step": 4770 }, { "epoch": 0.1259126941661659, "grad_norm": 2.3330159187316895, "learning_rate": 9.760840497129923e-06, "loss": 0.16704720258712769, "step": 4780 }, { "epoch": 0.1261761098443378, "grad_norm": 3.0629043579101562, "learning_rate": 9.759535278809193e-06, "loss": 0.17000982761383057, "step": 4790 }, { "epoch": 0.1264395255225097, "grad_norm": 2.149223566055298, "learning_rate": 9.758226596293343e-06, "loss": 0.17449510097503662, "step": 4800 }, { "epoch": 0.1267029412006816, "grad_norm": 5.307435035705566, "learning_rate": 9.75691445053489e-06, "loss": 0.16411055326461793, "step": 4810 }, { "epoch": 0.1269663568788535, "grad_norm": 1.795188307762146, "learning_rate": 9.75559884248887e-06, "loss": 0.16369513273239136, "step": 4820 }, { "epoch": 0.1272297725570254, "grad_norm": 3.170353651046753, "learning_rate": 9.754279773112836e-06, "loss": 0.17900640964508058, "step": 4830 }, { "epoch": 0.12749318823519729, "grad_norm": 2.2426586151123047, "learning_rate": 9.752957243366867e-06, "loss": 0.178638231754303, "step": 4840 }, { "epoch": 0.12775660391336918, "grad_norm": 4.800877094268799, "learning_rate": 9.751631254213557e-06, "loss": 0.163227117061615, "step": 4850 }, { "epoch": 0.12802001959154105, "grad_norm": 3.080871343612671, "learning_rate": 9.750301806618017e-06, "loss": 0.17038533687591553, "step": 4860 }, { "epoch": 0.12828343526971295, "grad_norm": 3.2225890159606934, "learning_rate": 9.748968901547878e-06, "loss": 0.1698058843612671, "step": 4870 }, { "epoch": 0.12854685094788484, "grad_norm": 4.657205104827881, "learning_rate": 9.747632539973282e-06, "loss": 0.18746333122253417, "step": 4880 }, { "epoch": 0.12881026662605674, "grad_norm": 3.5219497680664062, "learning_rate": 9.746292722866894e-06, "loss": 0.1643688917160034, "step": 4890 }, { "epoch": 0.12907368230422864, "grad_norm": 2.7926576137542725, "learning_rate": 9.744949451203891e-06, "loss": 0.16628879308700562, "step": 4900 }, { "epoch": 0.12933709798240053, "grad_norm": 1.7833871841430664, "learning_rate": 9.743602725961962e-06, "loss": 0.15988944768905639, "step": 4910 }, { "epoch": 0.12960051366057243, "grad_norm": 2.4460504055023193, "learning_rate": 9.742252548121314e-06, "loss": 0.17260122299194336, "step": 4920 }, { "epoch": 0.12986392933874433, "grad_norm": 3.5619354248046875, "learning_rate": 9.740898918664662e-06, "loss": 0.17193541526794434, "step": 4930 }, { "epoch": 0.13012734501691622, "grad_norm": 2.4360830783843994, "learning_rate": 9.739541838577237e-06, "loss": 0.1646174669265747, "step": 4940 }, { "epoch": 0.13039076069508812, "grad_norm": 1.1423580646514893, "learning_rate": 9.738181308846782e-06, "loss": 0.16466157436370848, "step": 4950 }, { "epoch": 0.13065417637326002, "grad_norm": 1.434401035308838, "learning_rate": 9.736817330463546e-06, "loss": 0.1676031231880188, "step": 4960 }, { "epoch": 0.1309175920514319, "grad_norm": 3.0087714195251465, "learning_rate": 9.735449904420292e-06, "loss": 0.15848714113235474, "step": 4970 }, { "epoch": 0.1311810077296038, "grad_norm": 1.7581169605255127, "learning_rate": 9.734079031712297e-06, "loss": 0.16906239986419677, "step": 4980 }, { "epoch": 0.1314444234077757, "grad_norm": 2.261765480041504, "learning_rate": 9.732704713337335e-06, "loss": 0.16801410913467407, "step": 4990 }, { "epoch": 0.1317078390859476, "grad_norm": 5.943589687347412, "learning_rate": 9.731326950295697e-06, "loss": 0.1527409315109253, "step": 5000 }, { "epoch": 0.1319712547641195, "grad_norm": 1.7594643831253052, "learning_rate": 9.729945743590176e-06, "loss": 0.17155909538269043, "step": 5010 }, { "epoch": 0.1322346704422914, "grad_norm": 4.3215508460998535, "learning_rate": 9.728561094226078e-06, "loss": 0.17556730508804322, "step": 5020 }, { "epoch": 0.1324980861204633, "grad_norm": 1.807218313217163, "learning_rate": 9.727173003211207e-06, "loss": 0.15856626033782958, "step": 5030 }, { "epoch": 0.1327615017986352, "grad_norm": 2.232227087020874, "learning_rate": 9.725781471555877e-06, "loss": 0.17623924016952514, "step": 5040 }, { "epoch": 0.1330249174768071, "grad_norm": 1494.5362548828125, "learning_rate": 9.724386500272905e-06, "loss": 1.5791622161865235, "step": 5050 }, { "epoch": 0.13328833315497896, "grad_norm": 7.055463790893555, "learning_rate": 9.72298809037761e-06, "loss": 0.6168725967407227, "step": 5060 }, { "epoch": 0.13355174883315085, "grad_norm": 2.580298900604248, "learning_rate": 9.721586242887814e-06, "loss": 0.20951552391052247, "step": 5070 }, { "epoch": 0.13381516451132275, "grad_norm": 4.19932746887207, "learning_rate": 9.720180958823847e-06, "loss": 0.17689931392669678, "step": 5080 }, { "epoch": 0.13407858018949464, "grad_norm": 5.4914021492004395, "learning_rate": 9.71877223920853e-06, "loss": 0.17071865797042846, "step": 5090 }, { "epoch": 0.13434199586766654, "grad_norm": 2.998354196548462, "learning_rate": 9.717360085067194e-06, "loss": 0.17885173559188844, "step": 5100 }, { "epoch": 0.13460541154583844, "grad_norm": 3.481097936630249, "learning_rate": 9.715944497427662e-06, "loss": 0.19116644859313964, "step": 5110 }, { "epoch": 0.13486882722401033, "grad_norm": 2.436389684677124, "learning_rate": 9.714525477320264e-06, "loss": 0.1796579122543335, "step": 5120 }, { "epoch": 0.13513224290218223, "grad_norm": 4.5535383224487305, "learning_rate": 9.71310302577782e-06, "loss": 0.17560794353485107, "step": 5130 }, { "epoch": 0.13539565858035413, "grad_norm": 4.112987995147705, "learning_rate": 9.711677143835653e-06, "loss": 0.17897002696990966, "step": 5140 }, { "epoch": 0.13565907425852602, "grad_norm": 3.697187900543213, "learning_rate": 9.710247832531583e-06, "loss": 0.1753474473953247, "step": 5150 }, { "epoch": 0.13592248993669792, "grad_norm": 2.4299402236938477, "learning_rate": 9.708815092905923e-06, "loss": 0.17777698040008544, "step": 5160 }, { "epoch": 0.13618590561486982, "grad_norm": 4.875292778015137, "learning_rate": 9.707378926001481e-06, "loss": 0.1812560439109802, "step": 5170 }, { "epoch": 0.13644932129304171, "grad_norm": 3.584601879119873, "learning_rate": 9.705939332863565e-06, "loss": 0.15839152336120604, "step": 5180 }, { "epoch": 0.1367127369712136, "grad_norm": 2.0921945571899414, "learning_rate": 9.70449631453997e-06, "loss": 0.17104249000549315, "step": 5190 }, { "epoch": 0.1369761526493855, "grad_norm": 3.8008408546447754, "learning_rate": 9.70304987208099e-06, "loss": 0.1702852725982666, "step": 5200 }, { "epoch": 0.1372395683275574, "grad_norm": 1.7643136978149414, "learning_rate": 9.701600006539405e-06, "loss": 0.16888428926467897, "step": 5210 }, { "epoch": 0.1375029840057293, "grad_norm": 3.285553455352783, "learning_rate": 9.70014671897049e-06, "loss": 0.1698533773422241, "step": 5220 }, { "epoch": 0.1377663996839012, "grad_norm": 3.0126004219055176, "learning_rate": 9.698690010432012e-06, "loss": 0.16269547939300538, "step": 5230 }, { "epoch": 0.1380298153620731, "grad_norm": 2.9137465953826904, "learning_rate": 9.697229881984226e-06, "loss": 0.15606882572174072, "step": 5240 }, { "epoch": 0.13829323104024496, "grad_norm": 3.2809126377105713, "learning_rate": 9.695766334689875e-06, "loss": 0.18339684009552001, "step": 5250 }, { "epoch": 0.13855664671841686, "grad_norm": 5.341607570648193, "learning_rate": 9.694299369614196e-06, "loss": 0.17618310451507568, "step": 5260 }, { "epoch": 0.13882006239658876, "grad_norm": 3.399418354034424, "learning_rate": 9.692828987824903e-06, "loss": 0.1724669575691223, "step": 5270 }, { "epoch": 0.13908347807476065, "grad_norm": 4.866036415100098, "learning_rate": 9.691355190392209e-06, "loss": 0.18255048990249634, "step": 5280 }, { "epoch": 0.13934689375293255, "grad_norm": 4.047554969787598, "learning_rate": 9.689877978388805e-06, "loss": 0.16879990100860595, "step": 5290 }, { "epoch": 0.13961030943110445, "grad_norm": 4.400942325592041, "learning_rate": 9.688397352889867e-06, "loss": 0.1660370111465454, "step": 5300 }, { "epoch": 0.13987372510927634, "grad_norm": 1.975563406944275, "learning_rate": 9.686913314973065e-06, "loss": 0.1535104990005493, "step": 5310 }, { "epoch": 0.14013714078744824, "grad_norm": 3.734795570373535, "learning_rate": 9.685425865718539e-06, "loss": 0.17044694423675538, "step": 5320 }, { "epoch": 0.14040055646562014, "grad_norm": 5.591304779052734, "learning_rate": 9.683935006208922e-06, "loss": 0.15766427516937256, "step": 5330 }, { "epoch": 0.14066397214379203, "grad_norm": 1.4861453771591187, "learning_rate": 9.682440737529326e-06, "loss": 0.17813096046447754, "step": 5340 }, { "epoch": 0.14092738782196393, "grad_norm": 2.697916269302368, "learning_rate": 9.680943060767343e-06, "loss": 0.15383286476135255, "step": 5350 }, { "epoch": 0.14119080350013583, "grad_norm": 4.568086624145508, "learning_rate": 9.67944197701305e-06, "loss": 0.1698022723197937, "step": 5360 }, { "epoch": 0.14145421917830772, "grad_norm": 2.967345714569092, "learning_rate": 9.677937487358994e-06, "loss": 0.17545194625854493, "step": 5370 }, { "epoch": 0.14171763485647962, "grad_norm": 2.166377067565918, "learning_rate": 9.676429592900212e-06, "loss": 0.15628767013549805, "step": 5380 }, { "epoch": 0.14198105053465152, "grad_norm": 4.788552284240723, "learning_rate": 9.674918294734215e-06, "loss": 0.16951594352722169, "step": 5390 }, { "epoch": 0.1422444662128234, "grad_norm": 4.124420166015625, "learning_rate": 9.673403593960991e-06, "loss": 0.16637558937072755, "step": 5400 }, { "epoch": 0.1425078818909953, "grad_norm": 2.8478031158447266, "learning_rate": 9.671885491683002e-06, "loss": 0.17900489568710326, "step": 5410 }, { "epoch": 0.1427712975691672, "grad_norm": 2.5203487873077393, "learning_rate": 9.670363989005188e-06, "loss": 0.1616595506668091, "step": 5420 }, { "epoch": 0.1430347132473391, "grad_norm": 3.1967074871063232, "learning_rate": 9.66883908703497e-06, "loss": 0.17033400535583496, "step": 5430 }, { "epoch": 0.143298128925511, "grad_norm": 2.5320680141448975, "learning_rate": 9.667310786882231e-06, "loss": 0.1684849739074707, "step": 5440 }, { "epoch": 0.14356154460368287, "grad_norm": 2.7041988372802734, "learning_rate": 9.665779089659338e-06, "loss": 0.1627207040786743, "step": 5450 }, { "epoch": 0.14382496028185476, "grad_norm": 3.792678117752075, "learning_rate": 9.664243996481124e-06, "loss": 0.17064294815063477, "step": 5460 }, { "epoch": 0.14408837596002666, "grad_norm": 2.244291305541992, "learning_rate": 9.662705508464896e-06, "loss": 0.16057488918304444, "step": 5470 }, { "epoch": 0.14435179163819856, "grad_norm": 3.239868402481079, "learning_rate": 9.661163626730434e-06, "loss": 0.1615046262741089, "step": 5480 }, { "epoch": 0.14461520731637045, "grad_norm": 1.7064470052719116, "learning_rate": 9.659618352399985e-06, "loss": 0.16662144660949707, "step": 5490 }, { "epoch": 0.14487862299454235, "grad_norm": 5.720874309539795, "learning_rate": 9.658069686598266e-06, "loss": 0.15769007205963134, "step": 5500 }, { "epoch": 0.14514203867271425, "grad_norm": 4.025033473968506, "learning_rate": 9.656517630452462e-06, "loss": 0.1666627883911133, "step": 5510 }, { "epoch": 0.14540545435088614, "grad_norm": 3.1421165466308594, "learning_rate": 9.654962185092228e-06, "loss": 0.16918705701828002, "step": 5520 }, { "epoch": 0.14566887002905804, "grad_norm": 4.918827056884766, "learning_rate": 9.653403351649683e-06, "loss": 0.164570415019989, "step": 5530 }, { "epoch": 0.14593228570722994, "grad_norm": 1.8642512559890747, "learning_rate": 9.651841131259414e-06, "loss": 0.158362078666687, "step": 5540 }, { "epoch": 0.14619570138540183, "grad_norm": 2.4675352573394775, "learning_rate": 9.650275525058473e-06, "loss": 0.1667211174964905, "step": 5550 }, { "epoch": 0.14645911706357373, "grad_norm": 2.4067506790161133, "learning_rate": 9.648706534186372e-06, "loss": 0.16074509620666505, "step": 5560 }, { "epoch": 0.14672253274174563, "grad_norm": 4.815004348754883, "learning_rate": 9.647134159785097e-06, "loss": 0.15410547256469725, "step": 5570 }, { "epoch": 0.14698594841991752, "grad_norm": 2.4200658798217773, "learning_rate": 9.645558402999084e-06, "loss": 0.17619353532791138, "step": 5580 }, { "epoch": 0.14724936409808942, "grad_norm": 3.922086000442505, "learning_rate": 9.643979264975239e-06, "loss": 0.1809668779373169, "step": 5590 }, { "epoch": 0.14751277977626132, "grad_norm": 1.5727133750915527, "learning_rate": 9.642396746862926e-06, "loss": 0.15674135684967042, "step": 5600 }, { "epoch": 0.1477761954544332, "grad_norm": 2.6375794410705566, "learning_rate": 9.640810849813972e-06, "loss": 0.14658334255218505, "step": 5610 }, { "epoch": 0.1480396111326051, "grad_norm": 2.185835599899292, "learning_rate": 9.63922157498266e-06, "loss": 0.16380313634872437, "step": 5620 }, { "epoch": 0.148303026810777, "grad_norm": 3.29579496383667, "learning_rate": 9.637628923525732e-06, "loss": 0.15296344757080077, "step": 5630 }, { "epoch": 0.1485664424889489, "grad_norm": 3.688950777053833, "learning_rate": 9.636032896602389e-06, "loss": 0.16867597103118898, "step": 5640 }, { "epoch": 0.14882985816712077, "grad_norm": 2.8733432292938232, "learning_rate": 9.634433495374289e-06, "loss": 0.16811735630035402, "step": 5650 }, { "epoch": 0.14909327384529267, "grad_norm": 2.2633261680603027, "learning_rate": 9.632830721005543e-06, "loss": 0.16475944519042968, "step": 5660 }, { "epoch": 0.14935668952346456, "grad_norm": 2.293163776397705, "learning_rate": 9.631224574662723e-06, "loss": 0.168978750705719, "step": 5670 }, { "epoch": 0.14962010520163646, "grad_norm": 3.640787363052368, "learning_rate": 9.62961505751485e-06, "loss": 0.17635656595230104, "step": 5680 }, { "epoch": 0.14988352087980836, "grad_norm": 2.2616353034973145, "learning_rate": 9.628002170733398e-06, "loss": 0.16491618156433105, "step": 5690 }, { "epoch": 0.15014693655798025, "grad_norm": 4.613709926605225, "learning_rate": 9.626385915492297e-06, "loss": 0.16729882955551148, "step": 5700 }, { "epoch": 0.15041035223615215, "grad_norm": 1.4747636318206787, "learning_rate": 9.624766292967927e-06, "loss": 0.16437418460845948, "step": 5710 }, { "epoch": 0.15067376791432405, "grad_norm": 3.3972105979919434, "learning_rate": 9.623143304339121e-06, "loss": 0.1618602991104126, "step": 5720 }, { "epoch": 0.15093718359249594, "grad_norm": 2.7554948329925537, "learning_rate": 9.621516950787158e-06, "loss": 0.15285061597824096, "step": 5730 }, { "epoch": 0.15120059927066784, "grad_norm": 2.3432235717773438, "learning_rate": 9.61988723349577e-06, "loss": 0.16568363904953004, "step": 5740 }, { "epoch": 0.15146401494883974, "grad_norm": 1.8628116846084595, "learning_rate": 9.618254153651134e-06, "loss": 0.15628483295440673, "step": 5750 }, { "epoch": 0.15172743062701163, "grad_norm": 1.670379877090454, "learning_rate": 9.616617712441875e-06, "loss": 0.1756540298461914, "step": 5760 }, { "epoch": 0.15199084630518353, "grad_norm": 2.4805352687835693, "learning_rate": 9.614977911059069e-06, "loss": 0.1488049030303955, "step": 5770 }, { "epoch": 0.15225426198335543, "grad_norm": 1.837319254875183, "learning_rate": 9.613334750696232e-06, "loss": 0.16532495021820068, "step": 5780 }, { "epoch": 0.15251767766152732, "grad_norm": 2.659468173980713, "learning_rate": 9.611688232549328e-06, "loss": 0.1759103536605835, "step": 5790 }, { "epoch": 0.15278109333969922, "grad_norm": 3.742863416671753, "learning_rate": 9.610038357816763e-06, "loss": 0.16152455806732177, "step": 5800 }, { "epoch": 0.15304450901787112, "grad_norm": 6.973571300506592, "learning_rate": 9.608385127699385e-06, "loss": 0.17643280029296876, "step": 5810 }, { "epoch": 0.153307924696043, "grad_norm": 5.333430290222168, "learning_rate": 9.60672854340049e-06, "loss": 0.176023268699646, "step": 5820 }, { "epoch": 0.1535713403742149, "grad_norm": 3.9568634033203125, "learning_rate": 9.60506860612581e-06, "loss": 0.16783303022384644, "step": 5830 }, { "epoch": 0.1538347560523868, "grad_norm": 3.4925036430358887, "learning_rate": 9.603405317083518e-06, "loss": 0.1653972864151001, "step": 5840 }, { "epoch": 0.15409817173055868, "grad_norm": 1.7794376611709595, "learning_rate": 9.60173867748423e-06, "loss": 0.16521236896514893, "step": 5850 }, { "epoch": 0.15436158740873057, "grad_norm": 2.183302402496338, "learning_rate": 9.600068688540996e-06, "loss": 0.1560445547103882, "step": 5860 }, { "epoch": 0.15462500308690247, "grad_norm": 5.903050899505615, "learning_rate": 9.598395351469305e-06, "loss": 0.16573083400726318, "step": 5870 }, { "epoch": 0.15488841876507436, "grad_norm": 4.326621055603027, "learning_rate": 9.596718667487089e-06, "loss": 0.15907764434814453, "step": 5880 }, { "epoch": 0.15515183444324626, "grad_norm": 2.58107852935791, "learning_rate": 9.595038637814704e-06, "loss": 0.15729308128356934, "step": 5890 }, { "epoch": 0.15541525012141816, "grad_norm": 2.2300422191619873, "learning_rate": 9.593355263674953e-06, "loss": 0.17150934934616088, "step": 5900 }, { "epoch": 0.15567866579959005, "grad_norm": 1.7954256534576416, "learning_rate": 9.591668546293068e-06, "loss": 0.14950352907180786, "step": 5910 }, { "epoch": 0.15594208147776195, "grad_norm": 3.346726417541504, "learning_rate": 9.589978486896711e-06, "loss": 0.16373677253723146, "step": 5920 }, { "epoch": 0.15620549715593385, "grad_norm": 2.1400082111358643, "learning_rate": 9.588285086715986e-06, "loss": 0.16124367713928223, "step": 5930 }, { "epoch": 0.15646891283410574, "grad_norm": 2.5346076488494873, "learning_rate": 9.586588346983418e-06, "loss": 0.17515432834625244, "step": 5940 }, { "epoch": 0.15673232851227764, "grad_norm": 1.7779539823532104, "learning_rate": 9.584888268933968e-06, "loss": 0.15915879011154174, "step": 5950 }, { "epoch": 0.15699574419044954, "grad_norm": 2.8913257122039795, "learning_rate": 9.583184853805028e-06, "loss": 0.161309814453125, "step": 5960 }, { "epoch": 0.15725915986862143, "grad_norm": 1.4160218238830566, "learning_rate": 9.581478102836416e-06, "loss": 0.16039915084838868, "step": 5970 }, { "epoch": 0.15752257554679333, "grad_norm": 2.0453052520751953, "learning_rate": 9.579768017270379e-06, "loss": 0.1688627004623413, "step": 5980 }, { "epoch": 0.15778599122496523, "grad_norm": 1.890442132949829, "learning_rate": 9.578054598351589e-06, "loss": 0.1454911708831787, "step": 5990 }, { "epoch": 0.15804940690313712, "grad_norm": 4.56081485748291, "learning_rate": 9.57633784732715e-06, "loss": 0.15731607675552367, "step": 6000 }, { "epoch": 0.15831282258130902, "grad_norm": 3.615267038345337, "learning_rate": 9.574617765446586e-06, "loss": 0.155615496635437, "step": 6010 }, { "epoch": 0.15857623825948092, "grad_norm": 3.139923095703125, "learning_rate": 9.572894353961846e-06, "loss": 0.1650020956993103, "step": 6020 }, { "epoch": 0.1588396539376528, "grad_norm": 3.618896722793579, "learning_rate": 9.571167614127305e-06, "loss": 0.15662777423858643, "step": 6030 }, { "epoch": 0.1591030696158247, "grad_norm": 2.4890599250793457, "learning_rate": 9.569437547199755e-06, "loss": 0.17709574699401856, "step": 6040 }, { "epoch": 0.15936648529399658, "grad_norm": 4.5045247077941895, "learning_rate": 9.567704154438417e-06, "loss": 0.17104616165161132, "step": 6050 }, { "epoch": 0.15962990097216848, "grad_norm": 2.1992714405059814, "learning_rate": 9.56596743710493e-06, "loss": 0.16846375465393065, "step": 6060 }, { "epoch": 0.15989331665034037, "grad_norm": 3.5368101596832275, "learning_rate": 9.564227396463347e-06, "loss": 0.1571272611618042, "step": 6070 }, { "epoch": 0.16015673232851227, "grad_norm": 1.251259446144104, "learning_rate": 9.562484033780148e-06, "loss": 0.1504136800765991, "step": 6080 }, { "epoch": 0.16042014800668417, "grad_norm": 2.1487412452697754, "learning_rate": 9.560737350324227e-06, "loss": 0.15664137601852418, "step": 6090 }, { "epoch": 0.16068356368485606, "grad_norm": 5.20465612411499, "learning_rate": 9.558987347366898e-06, "loss": 0.1702374577522278, "step": 6100 }, { "epoch": 0.16094697936302796, "grad_norm": 2.334980010986328, "learning_rate": 9.557234026181882e-06, "loss": 0.17294702529907227, "step": 6110 }, { "epoch": 0.16121039504119986, "grad_norm": 2.281681776046753, "learning_rate": 9.555477388045328e-06, "loss": 0.1519400358200073, "step": 6120 }, { "epoch": 0.16147381071937175, "grad_norm": 2.891550064086914, "learning_rate": 9.55371743423579e-06, "loss": 0.15976316928863527, "step": 6130 }, { "epoch": 0.16173722639754365, "grad_norm": 2.126455307006836, "learning_rate": 9.551954166034241e-06, "loss": 0.16687612533569335, "step": 6140 }, { "epoch": 0.16200064207571555, "grad_norm": 6.906123161315918, "learning_rate": 9.55018758472406e-06, "loss": 0.171065092086792, "step": 6150 }, { "epoch": 0.16226405775388744, "grad_norm": 3.193197250366211, "learning_rate": 9.548417691591044e-06, "loss": 0.14835066795349122, "step": 6160 }, { "epoch": 0.16252747343205934, "grad_norm": 1.8526387214660645, "learning_rate": 9.546644487923398e-06, "loss": 0.16218342781066894, "step": 6170 }, { "epoch": 0.16279088911023123, "grad_norm": 1.9017181396484375, "learning_rate": 9.544867975011732e-06, "loss": 0.16469185352325438, "step": 6180 }, { "epoch": 0.16305430478840313, "grad_norm": 3.5160176753997803, "learning_rate": 9.543088154149074e-06, "loss": 0.15527141094207764, "step": 6190 }, { "epoch": 0.16331772046657503, "grad_norm": 2.31353759765625, "learning_rate": 9.54130502663085e-06, "loss": 0.1647395372390747, "step": 6200 }, { "epoch": 0.16358113614474692, "grad_norm": 1.657252550125122, "learning_rate": 9.5395185937549e-06, "loss": 0.16268293857574462, "step": 6210 }, { "epoch": 0.16384455182291882, "grad_norm": 5.580503463745117, "learning_rate": 9.537728856821467e-06, "loss": 0.14743629693984986, "step": 6220 }, { "epoch": 0.16410796750109072, "grad_norm": 2.8178088665008545, "learning_rate": 9.535935817133196e-06, "loss": 0.17540860176086426, "step": 6230 }, { "epoch": 0.16437138317926261, "grad_norm": 2.801985740661621, "learning_rate": 9.534139475995141e-06, "loss": 0.1771228313446045, "step": 6240 }, { "epoch": 0.16463479885743448, "grad_norm": 3.9080746173858643, "learning_rate": 9.532339834714757e-06, "loss": 0.1706920623779297, "step": 6250 }, { "epoch": 0.16489821453560638, "grad_norm": 2.926011562347412, "learning_rate": 9.530536894601899e-06, "loss": 0.1508044958114624, "step": 6260 }, { "epoch": 0.16516163021377828, "grad_norm": 2.081136465072632, "learning_rate": 9.528730656968823e-06, "loss": 0.16593313217163086, "step": 6270 }, { "epoch": 0.16542504589195017, "grad_norm": 2.4576966762542725, "learning_rate": 9.52692112313019e-06, "loss": 0.1643066883087158, "step": 6280 }, { "epoch": 0.16568846157012207, "grad_norm": 1.7805933952331543, "learning_rate": 9.525108294403057e-06, "loss": 0.1618055820465088, "step": 6290 }, { "epoch": 0.16595187724829397, "grad_norm": 3.5364837646484375, "learning_rate": 9.523292172106877e-06, "loss": 0.1609845757484436, "step": 6300 }, { "epoch": 0.16621529292646586, "grad_norm": 3.148632287979126, "learning_rate": 9.521472757563502e-06, "loss": 0.1639179468154907, "step": 6310 }, { "epoch": 0.16647870860463776, "grad_norm": 2.5965583324432373, "learning_rate": 9.519650052097183e-06, "loss": 0.17998669147491456, "step": 6320 }, { "epoch": 0.16674212428280966, "grad_norm": 4.440237998962402, "learning_rate": 9.51782405703456e-06, "loss": 0.16292529106140136, "step": 6330 }, { "epoch": 0.16700553996098155, "grad_norm": 4.285465240478516, "learning_rate": 9.515994773704673e-06, "loss": 0.14668445587158202, "step": 6340 }, { "epoch": 0.16726895563915345, "grad_norm": 2.621535301208496, "learning_rate": 9.514162203438952e-06, "loss": 0.16532602310180664, "step": 6350 }, { "epoch": 0.16753237131732535, "grad_norm": 3.2964911460876465, "learning_rate": 9.512326347571223e-06, "loss": 0.16341776847839357, "step": 6360 }, { "epoch": 0.16779578699549724, "grad_norm": 1.6526225805282593, "learning_rate": 9.5104872074377e-06, "loss": 0.16083877086639403, "step": 6370 }, { "epoch": 0.16805920267366914, "grad_norm": 4.152899742126465, "learning_rate": 9.50864478437699e-06, "loss": 0.16251096725463868, "step": 6380 }, { "epoch": 0.16832261835184104, "grad_norm": 3.8006722927093506, "learning_rate": 9.506799079730084e-06, "loss": 0.1554914355278015, "step": 6390 }, { "epoch": 0.16858603403001293, "grad_norm": 1.7266108989715576, "learning_rate": 9.50495009484037e-06, "loss": 0.16420042514801025, "step": 6400 }, { "epoch": 0.16884944970818483, "grad_norm": 2.033888101577759, "learning_rate": 9.503097831053615e-06, "loss": 0.17049846649169922, "step": 6410 }, { "epoch": 0.16911286538635673, "grad_norm": 3.31581449508667, "learning_rate": 9.50124228971798e-06, "loss": 0.16058666706085206, "step": 6420 }, { "epoch": 0.16937628106452862, "grad_norm": 2.709162712097168, "learning_rate": 9.499383472184005e-06, "loss": 0.1507721424102783, "step": 6430 }, { "epoch": 0.1696396967427005, "grad_norm": 3.2523279190063477, "learning_rate": 9.49752137980462e-06, "loss": 0.17518296241760253, "step": 6440 }, { "epoch": 0.1699031124208724, "grad_norm": 5.409176349639893, "learning_rate": 9.495656013935135e-06, "loss": 0.15878530740737914, "step": 6450 }, { "epoch": 0.17016652809904428, "grad_norm": 3.0873818397521973, "learning_rate": 9.493787375933244e-06, "loss": 0.16938438415527343, "step": 6460 }, { "epoch": 0.17042994377721618, "grad_norm": 2.362013816833496, "learning_rate": 9.491915467159022e-06, "loss": 0.16343544721603392, "step": 6470 }, { "epoch": 0.17069335945538808, "grad_norm": 1.5519912242889404, "learning_rate": 9.490040288974924e-06, "loss": 0.1624719738960266, "step": 6480 }, { "epoch": 0.17095677513355997, "grad_norm": 3.7493700981140137, "learning_rate": 9.488161842745788e-06, "loss": 0.16382393836975098, "step": 6490 }, { "epoch": 0.17122019081173187, "grad_norm": 5.438973903656006, "learning_rate": 9.486280129838829e-06, "loss": 0.14994332790374756, "step": 6500 }, { "epoch": 0.17148360648990377, "grad_norm": 2.965027332305908, "learning_rate": 9.484395151623636e-06, "loss": 0.14929412603378295, "step": 6510 }, { "epoch": 0.17174702216807566, "grad_norm": 2.8033692836761475, "learning_rate": 9.482506909472178e-06, "loss": 0.17265799045562744, "step": 6520 }, { "epoch": 0.17201043784624756, "grad_norm": 4.258802890777588, "learning_rate": 9.4806154047588e-06, "loss": 0.15901799201965333, "step": 6530 }, { "epoch": 0.17227385352441946, "grad_norm": 1.8897523880004883, "learning_rate": 9.478720638860219e-06, "loss": 0.16402206420898438, "step": 6540 }, { "epoch": 0.17253726920259135, "grad_norm": 3.2578485012054443, "learning_rate": 9.476822613155528e-06, "loss": 0.16416295766830444, "step": 6550 }, { "epoch": 0.17280068488076325, "grad_norm": 2.636996269226074, "learning_rate": 9.474921329026195e-06, "loss": 0.17562596797943114, "step": 6560 }, { "epoch": 0.17306410055893515, "grad_norm": 2.6528289318084717, "learning_rate": 9.473016787856052e-06, "loss": 0.17106266021728517, "step": 6570 }, { "epoch": 0.17332751623710704, "grad_norm": 2.7548158168792725, "learning_rate": 9.47110899103131e-06, "loss": 0.16846374273300171, "step": 6580 }, { "epoch": 0.17359093191527894, "grad_norm": 2.8913047313690186, "learning_rate": 9.469197939940543e-06, "loss": 0.14618172645568847, "step": 6590 }, { "epoch": 0.17385434759345084, "grad_norm": 1.5763473510742188, "learning_rate": 9.467283635974699e-06, "loss": 0.1552791714668274, "step": 6600 }, { "epoch": 0.17411776327162273, "grad_norm": 2.175835132598877, "learning_rate": 9.465366080527088e-06, "loss": 0.15705740451812744, "step": 6610 }, { "epoch": 0.17438117894979463, "grad_norm": 2.005786895751953, "learning_rate": 9.463445274993393e-06, "loss": 0.1634977102279663, "step": 6620 }, { "epoch": 0.17464459462796653, "grad_norm": 1.3269760608673096, "learning_rate": 9.461521220771657e-06, "loss": 0.1491564154624939, "step": 6630 }, { "epoch": 0.1749080103061384, "grad_norm": 1.8701040744781494, "learning_rate": 9.45959391926229e-06, "loss": 0.1597490429878235, "step": 6640 }, { "epoch": 0.1751714259843103, "grad_norm": 3.5837721824645996, "learning_rate": 9.457663371868067e-06, "loss": 0.1690082311630249, "step": 6650 }, { "epoch": 0.1754348416624822, "grad_norm": 2.2362959384918213, "learning_rate": 9.455729579994119e-06, "loss": 0.16199724674224852, "step": 6660 }, { "epoch": 0.17569825734065408, "grad_norm": 2.4087870121002197, "learning_rate": 9.453792545047948e-06, "loss": 0.15500166416168212, "step": 6670 }, { "epoch": 0.17596167301882598, "grad_norm": 3.3926615715026855, "learning_rate": 9.45185226843941e-06, "loss": 0.15974433422088624, "step": 6680 }, { "epoch": 0.17622508869699788, "grad_norm": 3.38918137550354, "learning_rate": 9.44990875158072e-06, "loss": 0.13677480220794677, "step": 6690 }, { "epoch": 0.17648850437516977, "grad_norm": 1.9878153800964355, "learning_rate": 9.447961995886457e-06, "loss": 0.1544861078262329, "step": 6700 }, { "epoch": 0.17675192005334167, "grad_norm": 2.9755771160125732, "learning_rate": 9.446012002773552e-06, "loss": 0.16434826850891113, "step": 6710 }, { "epoch": 0.17701533573151357, "grad_norm": 6.011094570159912, "learning_rate": 9.444058773661293e-06, "loss": 0.16279008388519287, "step": 6720 }, { "epoch": 0.17727875140968546, "grad_norm": 2.010594129562378, "learning_rate": 9.442102309971325e-06, "loss": 0.15422968864440917, "step": 6730 }, { "epoch": 0.17754216708785736, "grad_norm": 2.117446184158325, "learning_rate": 9.44014261312765e-06, "loss": 0.16246540546417237, "step": 6740 }, { "epoch": 0.17780558276602926, "grad_norm": 2.1554601192474365, "learning_rate": 9.438179684556615e-06, "loss": 0.15567526817321778, "step": 6750 }, { "epoch": 0.17806899844420115, "grad_norm": 3.4268338680267334, "learning_rate": 9.436213525686926e-06, "loss": 0.15393097400665284, "step": 6760 }, { "epoch": 0.17833241412237305, "grad_norm": 3.309215784072876, "learning_rate": 9.43424413794964e-06, "loss": 0.1504669427871704, "step": 6770 }, { "epoch": 0.17859582980054495, "grad_norm": 3.159430742263794, "learning_rate": 9.432271522778159e-06, "loss": 0.16126105785369874, "step": 6780 }, { "epoch": 0.17885924547871684, "grad_norm": 2.957378625869751, "learning_rate": 9.430295681608241e-06, "loss": 0.14963533878326415, "step": 6790 }, { "epoch": 0.17912266115688874, "grad_norm": 6.341246604919434, "learning_rate": 9.428316615877986e-06, "loss": 0.1646751880645752, "step": 6800 }, { "epoch": 0.17938607683506064, "grad_norm": 3.3716623783111572, "learning_rate": 9.426334327027843e-06, "loss": 0.16151554584503175, "step": 6810 }, { "epoch": 0.17964949251323253, "grad_norm": 2.3764779567718506, "learning_rate": 9.42434881650061e-06, "loss": 0.1663457155227661, "step": 6820 }, { "epoch": 0.17991290819140443, "grad_norm": 4.476960182189941, "learning_rate": 9.422360085741427e-06, "loss": 0.15629475116729735, "step": 6830 }, { "epoch": 0.1801763238695763, "grad_norm": 5.695446014404297, "learning_rate": 9.420368136197775e-06, "loss": 0.17141170501708985, "step": 6840 }, { "epoch": 0.1804397395477482, "grad_norm": 5.565206050872803, "learning_rate": 9.418372969319481e-06, "loss": 0.17278221845626832, "step": 6850 }, { "epoch": 0.1807031552259201, "grad_norm": 3.380385637283325, "learning_rate": 9.416374586558717e-06, "loss": 0.15293947458267212, "step": 6860 }, { "epoch": 0.180966570904092, "grad_norm": 3.930631399154663, "learning_rate": 9.414372989369991e-06, "loss": 0.16947923898696898, "step": 6870 }, { "epoch": 0.18122998658226389, "grad_norm": 3.4796152114868164, "learning_rate": 9.41236817921015e-06, "loss": 0.15807442665100097, "step": 6880 }, { "epoch": 0.18149340226043578, "grad_norm": 2.523235321044922, "learning_rate": 9.410360157538383e-06, "loss": 0.17156808376312255, "step": 6890 }, { "epoch": 0.18175681793860768, "grad_norm": 5.6764349937438965, "learning_rate": 9.408348925816212e-06, "loss": 0.1663515329360962, "step": 6900 }, { "epoch": 0.18202023361677958, "grad_norm": 3.3185782432556152, "learning_rate": 9.406334485507503e-06, "loss": 0.15219409465789796, "step": 6910 }, { "epoch": 0.18228364929495147, "grad_norm": 2.630089044570923, "learning_rate": 9.40431683807845e-06, "loss": 0.1759208083152771, "step": 6920 }, { "epoch": 0.18254706497312337, "grad_norm": 1.5062687397003174, "learning_rate": 9.402295984997583e-06, "loss": 0.1649142861366272, "step": 6930 }, { "epoch": 0.18281048065129527, "grad_norm": 3.346684455871582, "learning_rate": 9.400271927735767e-06, "loss": 0.16449811458587646, "step": 6940 }, { "epoch": 0.18307389632946716, "grad_norm": 3.5317800045013428, "learning_rate": 9.3982446677662e-06, "loss": 0.16395266056060792, "step": 6950 }, { "epoch": 0.18333731200763906, "grad_norm": 4.187323570251465, "learning_rate": 9.396214206564405e-06, "loss": 0.1532060384750366, "step": 6960 }, { "epoch": 0.18360072768581095, "grad_norm": 4.266016006469727, "learning_rate": 9.39418054560824e-06, "loss": 0.1544421672821045, "step": 6970 }, { "epoch": 0.18386414336398285, "grad_norm": 2.4494972229003906, "learning_rate": 9.392143686377898e-06, "loss": 0.16287569999694823, "step": 6980 }, { "epoch": 0.18412755904215475, "grad_norm": 2.2286198139190674, "learning_rate": 9.390103630355883e-06, "loss": 0.1686761736869812, "step": 6990 }, { "epoch": 0.18439097472032664, "grad_norm": 4.347607135772705, "learning_rate": 9.388060379027044e-06, "loss": 0.15254151821136475, "step": 7000 }, { "epoch": 0.18465439039849854, "grad_norm": 1.4812270402908325, "learning_rate": 9.386013933878541e-06, "loss": 0.1569042205810547, "step": 7010 }, { "epoch": 0.18491780607667044, "grad_norm": 3.1481423377990723, "learning_rate": 9.383964296399871e-06, "loss": 0.15730068683624268, "step": 7020 }, { "epoch": 0.18518122175484233, "grad_norm": 2.2468903064727783, "learning_rate": 9.381911468082845e-06, "loss": 0.1556534767150879, "step": 7030 }, { "epoch": 0.1854446374330142, "grad_norm": 2.0355112552642822, "learning_rate": 9.379855450421601e-06, "loss": 0.15449790954589843, "step": 7040 }, { "epoch": 0.1857080531111861, "grad_norm": 1.4318108558654785, "learning_rate": 9.377796244912596e-06, "loss": 0.16282023191452027, "step": 7050 }, { "epoch": 0.185971468789358, "grad_norm": 2.3883957862854004, "learning_rate": 9.375733853054612e-06, "loss": 0.1626462459564209, "step": 7060 }, { "epoch": 0.1862348844675299, "grad_norm": 1.6662840843200684, "learning_rate": 9.373668276348742e-06, "loss": 0.16591007709503175, "step": 7070 }, { "epoch": 0.1864983001457018, "grad_norm": 1.8027677536010742, "learning_rate": 9.371599516298405e-06, "loss": 0.14958369731903076, "step": 7080 }, { "epoch": 0.1867617158238737, "grad_norm": 1.9331743717193604, "learning_rate": 9.369527574409335e-06, "loss": 0.15669631958007812, "step": 7090 }, { "epoch": 0.18702513150204558, "grad_norm": 3.626664638519287, "learning_rate": 9.367452452189576e-06, "loss": 0.16248295307159424, "step": 7100 }, { "epoch": 0.18728854718021748, "grad_norm": 1.8000528812408447, "learning_rate": 9.365374151149496e-06, "loss": 0.15469173192977906, "step": 7110 }, { "epoch": 0.18755196285838938, "grad_norm": 2.308020830154419, "learning_rate": 9.363292672801772e-06, "loss": 0.16277776956558226, "step": 7120 }, { "epoch": 0.18781537853656127, "grad_norm": 2.6496388912200928, "learning_rate": 9.361208018661389e-06, "loss": 0.1601394534111023, "step": 7130 }, { "epoch": 0.18807879421473317, "grad_norm": 3.177948236465454, "learning_rate": 9.359120190245654e-06, "loss": 0.16084367036819458, "step": 7140 }, { "epoch": 0.18834220989290507, "grad_norm": 4.236626148223877, "learning_rate": 9.357029189074176e-06, "loss": 0.17409416437149047, "step": 7150 }, { "epoch": 0.18860562557107696, "grad_norm": 1.6272380352020264, "learning_rate": 9.354935016668876e-06, "loss": 0.16155076026916504, "step": 7160 }, { "epoch": 0.18886904124924886, "grad_norm": 3.2393546104431152, "learning_rate": 9.352837674553982e-06, "loss": 0.14930622577667235, "step": 7170 }, { "epoch": 0.18913245692742076, "grad_norm": 3.2108850479125977, "learning_rate": 9.35073716425603e-06, "loss": 0.1575726628303528, "step": 7180 }, { "epoch": 0.18939587260559265, "grad_norm": 5.226456642150879, "learning_rate": 9.348633487303866e-06, "loss": 0.1593010187149048, "step": 7190 }, { "epoch": 0.18965928828376455, "grad_norm": 1.9226504564285278, "learning_rate": 9.346526645228632e-06, "loss": 0.15329430103302003, "step": 7200 }, { "epoch": 0.18992270396193645, "grad_norm": 2.3458497524261475, "learning_rate": 9.34441663956378e-06, "loss": 0.15764977931976318, "step": 7210 }, { "epoch": 0.19018611964010834, "grad_norm": 1.614496111869812, "learning_rate": 9.342303471845065e-06, "loss": 0.15011906623840332, "step": 7220 }, { "epoch": 0.19044953531828024, "grad_norm": 5.360962390899658, "learning_rate": 9.34018714361054e-06, "loss": 0.17166576385498047, "step": 7230 }, { "epoch": 0.1907129509964521, "grad_norm": 2.0665454864501953, "learning_rate": 9.33806765640056e-06, "loss": 0.17163195610046386, "step": 7240 }, { "epoch": 0.190976366674624, "grad_norm": 3.19826340675354, "learning_rate": 9.33594501175778e-06, "loss": 0.1578364610671997, "step": 7250 }, { "epoch": 0.1912397823527959, "grad_norm": 3.2034387588500977, "learning_rate": 9.333819211227153e-06, "loss": 0.17079854011535645, "step": 7260 }, { "epoch": 0.1915031980309678, "grad_norm": 1.6135073900222778, "learning_rate": 9.331690256355926e-06, "loss": 0.16498885154724122, "step": 7270 }, { "epoch": 0.1917666137091397, "grad_norm": 2.3838768005371094, "learning_rate": 9.329558148693645e-06, "loss": 0.15808719396591187, "step": 7280 }, { "epoch": 0.1920300293873116, "grad_norm": 2.9321160316467285, "learning_rate": 9.32742288979215e-06, "loss": 0.1571478486061096, "step": 7290 }, { "epoch": 0.1922934450654835, "grad_norm": 5.342441082000732, "learning_rate": 9.325284481205577e-06, "loss": 0.15866379737854003, "step": 7300 }, { "epoch": 0.19255686074365538, "grad_norm": 2.8926515579223633, "learning_rate": 9.323142924490347e-06, "loss": 0.16318860054016113, "step": 7310 }, { "epoch": 0.19282027642182728, "grad_norm": 4.6519341468811035, "learning_rate": 9.320998221205182e-06, "loss": 0.16788477897644044, "step": 7320 }, { "epoch": 0.19308369209999918, "grad_norm": 2.4587111473083496, "learning_rate": 9.318850372911088e-06, "loss": 0.16177833080291748, "step": 7330 }, { "epoch": 0.19334710777817107, "grad_norm": 1.884494662284851, "learning_rate": 9.316699381171361e-06, "loss": 0.16653990745544434, "step": 7340 }, { "epoch": 0.19361052345634297, "grad_norm": 3.794616460800171, "learning_rate": 9.314545247551588e-06, "loss": 0.14977695941925048, "step": 7350 }, { "epoch": 0.19387393913451487, "grad_norm": 6.14423942565918, "learning_rate": 9.312387973619636e-06, "loss": 0.15543452501296998, "step": 7360 }, { "epoch": 0.19413735481268676, "grad_norm": 1.608450174331665, "learning_rate": 9.310227560945665e-06, "loss": 0.1784190058708191, "step": 7370 }, { "epoch": 0.19440077049085866, "grad_norm": 1.6143823862075806, "learning_rate": 9.308064011102116e-06, "loss": 0.1601337432861328, "step": 7380 }, { "epoch": 0.19466418616903056, "grad_norm": 1.6970078945159912, "learning_rate": 9.305897325663716e-06, "loss": 0.1678074359893799, "step": 7390 }, { "epoch": 0.19492760184720245, "grad_norm": 3.0269763469696045, "learning_rate": 9.303727506207468e-06, "loss": 0.16174378395080566, "step": 7400 }, { "epoch": 0.19519101752537435, "grad_norm": 4.8217315673828125, "learning_rate": 9.301554554312664e-06, "loss": 0.1445288062095642, "step": 7410 }, { "epoch": 0.19545443320354625, "grad_norm": 2.771101474761963, "learning_rate": 9.29937847156087e-06, "loss": 0.17192678451538085, "step": 7420 }, { "epoch": 0.19571784888171812, "grad_norm": 3.7780098915100098, "learning_rate": 9.297199259535933e-06, "loss": 0.1777357578277588, "step": 7430 }, { "epoch": 0.19598126455989, "grad_norm": 2.3590927124023438, "learning_rate": 9.295016919823978e-06, "loss": 0.1643328309059143, "step": 7440 }, { "epoch": 0.1962446802380619, "grad_norm": 3.465576410293579, "learning_rate": 9.292831454013407e-06, "loss": 0.1632140874862671, "step": 7450 }, { "epoch": 0.1965080959162338, "grad_norm": 2.904449462890625, "learning_rate": 9.290642863694894e-06, "loss": 0.14923250675201416, "step": 7460 }, { "epoch": 0.1967715115944057, "grad_norm": 1.9922462701797485, "learning_rate": 9.28845115046139e-06, "loss": 0.17221757173538207, "step": 7470 }, { "epoch": 0.1970349272725776, "grad_norm": 2.3931546211242676, "learning_rate": 9.286256315908121e-06, "loss": 0.17456512451171874, "step": 7480 }, { "epoch": 0.1972983429507495, "grad_norm": 2.6548211574554443, "learning_rate": 9.28405836163258e-06, "loss": 0.1714434027671814, "step": 7490 }, { "epoch": 0.1975617586289214, "grad_norm": 2.06655216217041, "learning_rate": 9.281857289234532e-06, "loss": 0.16431554555892944, "step": 7500 }, { "epoch": 0.1978251743070933, "grad_norm": 5.934584140777588, "learning_rate": 9.279653100316018e-06, "loss": 0.16635960340499878, "step": 7510 }, { "epoch": 0.19808858998526518, "grad_norm": 2.567450761795044, "learning_rate": 9.277445796481333e-06, "loss": 0.16446259021759033, "step": 7520 }, { "epoch": 0.19835200566343708, "grad_norm": 1.5596301555633545, "learning_rate": 9.275235379337056e-06, "loss": 0.15794095993041993, "step": 7530 }, { "epoch": 0.19861542134160898, "grad_norm": 4.1717209815979, "learning_rate": 9.273021850492018e-06, "loss": 0.161451256275177, "step": 7540 }, { "epoch": 0.19887883701978087, "grad_norm": 3.723606824874878, "learning_rate": 9.270805211557324e-06, "loss": 0.15438941717147828, "step": 7550 }, { "epoch": 0.19914225269795277, "grad_norm": 1.9820066690444946, "learning_rate": 9.26858546414634e-06, "loss": 0.14983400106430053, "step": 7560 }, { "epoch": 0.19940566837612467, "grad_norm": 2.1370341777801514, "learning_rate": 9.26636260987469e-06, "loss": 0.16017599105834962, "step": 7570 }, { "epoch": 0.19966908405429656, "grad_norm": 1.4647772312164307, "learning_rate": 9.264136650360269e-06, "loss": 0.1649295210838318, "step": 7580 }, { "epoch": 0.19993249973246846, "grad_norm": 2.8997933864593506, "learning_rate": 9.261907587223222e-06, "loss": 0.15629860162734985, "step": 7590 }, { "epoch": 0.20019591541064036, "grad_norm": 4.374965190887451, "learning_rate": 9.25967542208596e-06, "loss": 0.1663503646850586, "step": 7600 }, { "epoch": 0.20045933108881225, "grad_norm": 1.485316276550293, "learning_rate": 9.257440156573147e-06, "loss": 0.15532341003417968, "step": 7610 }, { "epoch": 0.20072274676698415, "grad_norm": 2.207362174987793, "learning_rate": 9.255201792311707e-06, "loss": 0.17302726507186889, "step": 7620 }, { "epoch": 0.20098616244515602, "grad_norm": 1.4272087812423706, "learning_rate": 9.252960330930816e-06, "loss": 0.1532304048538208, "step": 7630 }, { "epoch": 0.20124957812332792, "grad_norm": 2.5025594234466553, "learning_rate": 9.25071577406191e-06, "loss": 0.16170103549957277, "step": 7640 }, { "epoch": 0.2015129938014998, "grad_norm": 3.305387496948242, "learning_rate": 9.24846812333867e-06, "loss": 0.15532870292663575, "step": 7650 }, { "epoch": 0.2017764094796717, "grad_norm": 2.7693276405334473, "learning_rate": 9.246217380397036e-06, "loss": 0.1749914765357971, "step": 7660 }, { "epoch": 0.2020398251578436, "grad_norm": 3.7836434841156006, "learning_rate": 9.243963546875195e-06, "loss": 0.15469319820404054, "step": 7670 }, { "epoch": 0.2023032408360155, "grad_norm": 2.111919641494751, "learning_rate": 9.241706624413584e-06, "loss": 0.1416559934616089, "step": 7680 }, { "epoch": 0.2025666565141874, "grad_norm": 1.3297168016433716, "learning_rate": 9.239446614654892e-06, "loss": 0.16851226091384888, "step": 7690 }, { "epoch": 0.2028300721923593, "grad_norm": 1.1494040489196777, "learning_rate": 9.237183519244045e-06, "loss": 0.159784197807312, "step": 7700 }, { "epoch": 0.2030934878705312, "grad_norm": 1.4149024486541748, "learning_rate": 9.234917339828226e-06, "loss": 0.15279344320297242, "step": 7710 }, { "epoch": 0.2033569035487031, "grad_norm": 2.9850072860717773, "learning_rate": 9.232648078056858e-06, "loss": 0.1638326406478882, "step": 7720 }, { "epoch": 0.20362031922687499, "grad_norm": 2.3541128635406494, "learning_rate": 9.230375735581606e-06, "loss": 0.15825501680374146, "step": 7730 }, { "epoch": 0.20388373490504688, "grad_norm": 5.68491792678833, "learning_rate": 9.22810031405638e-06, "loss": 0.15651507377624513, "step": 7740 }, { "epoch": 0.20414715058321878, "grad_norm": 4.614163875579834, "learning_rate": 9.225821815137332e-06, "loss": 0.16692949533462526, "step": 7750 }, { "epoch": 0.20441056626139067, "grad_norm": 1.2842075824737549, "learning_rate": 9.223540240482847e-06, "loss": 0.15286213159561157, "step": 7760 }, { "epoch": 0.20467398193956257, "grad_norm": 1.711776614189148, "learning_rate": 9.221255591753556e-06, "loss": 0.15505523681640626, "step": 7770 }, { "epoch": 0.20493739761773447, "grad_norm": 4.292652606964111, "learning_rate": 9.218967870612326e-06, "loss": 0.1681240439414978, "step": 7780 }, { "epoch": 0.20520081329590636, "grad_norm": 2.58949613571167, "learning_rate": 9.216677078724257e-06, "loss": 0.14896212816238402, "step": 7790 }, { "epoch": 0.20546422897407826, "grad_norm": 2.353163003921509, "learning_rate": 9.214383217756687e-06, "loss": 0.1536668062210083, "step": 7800 }, { "epoch": 0.20572764465225016, "grad_norm": 4.285794258117676, "learning_rate": 9.212086289379187e-06, "loss": 0.1548104166984558, "step": 7810 }, { "epoch": 0.20599106033042205, "grad_norm": 4.126558303833008, "learning_rate": 9.20978629526356e-06, "loss": 0.14919464588165282, "step": 7820 }, { "epoch": 0.20625447600859392, "grad_norm": 3.2525529861450195, "learning_rate": 9.207483237083844e-06, "loss": 0.16921472549438477, "step": 7830 }, { "epoch": 0.20651789168676582, "grad_norm": 2.727025032043457, "learning_rate": 9.205177116516302e-06, "loss": 0.15440857410430908, "step": 7840 }, { "epoch": 0.20678130736493772, "grad_norm": 4.873241424560547, "learning_rate": 9.202867935239426e-06, "loss": 0.16930251121520995, "step": 7850 }, { "epoch": 0.2070447230431096, "grad_norm": 2.4318389892578125, "learning_rate": 9.200555694933941e-06, "loss": 0.16031253337860107, "step": 7860 }, { "epoch": 0.2073081387212815, "grad_norm": 1.6009972095489502, "learning_rate": 9.198240397282791e-06, "loss": 0.1361618995666504, "step": 7870 }, { "epoch": 0.2075715543994534, "grad_norm": 2.304414987564087, "learning_rate": 9.195922043971155e-06, "loss": 0.16158114671707152, "step": 7880 }, { "epoch": 0.2078349700776253, "grad_norm": 2.6898601055145264, "learning_rate": 9.193600636686428e-06, "loss": 0.1503298282623291, "step": 7890 }, { "epoch": 0.2080983857557972, "grad_norm": 5.916249752044678, "learning_rate": 9.19127617711823e-06, "loss": 0.17227243185043334, "step": 7900 }, { "epoch": 0.2083618014339691, "grad_norm": 3.402132987976074, "learning_rate": 9.188948666958404e-06, "loss": 0.1634302854537964, "step": 7910 }, { "epoch": 0.208625217112141, "grad_norm": 1.6800005435943604, "learning_rate": 9.18661810790101e-06, "loss": 0.16907598972320556, "step": 7920 }, { "epoch": 0.2088886327903129, "grad_norm": 4.796390056610107, "learning_rate": 9.184284501642331e-06, "loss": 0.14973186254501342, "step": 7930 }, { "epoch": 0.20915204846848479, "grad_norm": 1.9380837678909302, "learning_rate": 9.181947849880868e-06, "loss": 0.1741116762161255, "step": 7940 }, { "epoch": 0.20941546414665668, "grad_norm": 2.2395403385162354, "learning_rate": 9.179608154317334e-06, "loss": 0.1589387059211731, "step": 7950 }, { "epoch": 0.20967887982482858, "grad_norm": 4.1374125480651855, "learning_rate": 9.177265416654662e-06, "loss": 0.1515341281890869, "step": 7960 }, { "epoch": 0.20994229550300048, "grad_norm": 3.5229086875915527, "learning_rate": 9.174919638597996e-06, "loss": 0.16062772274017334, "step": 7970 }, { "epoch": 0.21020571118117237, "grad_norm": 1.9373637437820435, "learning_rate": 9.172570821854697e-06, "loss": 0.14937136173248292, "step": 7980 }, { "epoch": 0.21046912685934427, "grad_norm": 3.934753656387329, "learning_rate": 9.170218968134334e-06, "loss": 0.15157160758972169, "step": 7990 }, { "epoch": 0.21073254253751617, "grad_norm": 2.6855368614196777, "learning_rate": 9.167864079148687e-06, "loss": 0.14570785760879518, "step": 8000 }, { "epoch": 0.21099595821568806, "grad_norm": 9.849858283996582, "learning_rate": 9.165506156611745e-06, "loss": 0.15157243013381957, "step": 8010 }, { "epoch": 0.21125937389385996, "grad_norm": 2.5532679557800293, "learning_rate": 9.16314520223971e-06, "loss": 0.17082276344299316, "step": 8020 }, { "epoch": 0.21152278957203183, "grad_norm": 2.3719701766967773, "learning_rate": 9.160781217750982e-06, "loss": 0.17286163568496704, "step": 8030 }, { "epoch": 0.21178620525020372, "grad_norm": 6.552882194519043, "learning_rate": 9.15841420486617e-06, "loss": 0.15885871648788452, "step": 8040 }, { "epoch": 0.21204962092837562, "grad_norm": 3.4485201835632324, "learning_rate": 9.156044165308094e-06, "loss": 0.1670440435409546, "step": 8050 }, { "epoch": 0.21231303660654752, "grad_norm": 3.9054269790649414, "learning_rate": 9.153671100801767e-06, "loss": 0.17230713367462158, "step": 8060 }, { "epoch": 0.2125764522847194, "grad_norm": 2.2353808879852295, "learning_rate": 9.151295013074405e-06, "loss": 0.155734384059906, "step": 8070 }, { "epoch": 0.2128398679628913, "grad_norm": 4.009214878082275, "learning_rate": 9.148915903855435e-06, "loss": 0.15333133935928345, "step": 8080 }, { "epoch": 0.2131032836410632, "grad_norm": 2.0219457149505615, "learning_rate": 9.146533774876469e-06, "loss": 0.16475909948349, "step": 8090 }, { "epoch": 0.2133666993192351, "grad_norm": 4.046748161315918, "learning_rate": 9.144148627871323e-06, "loss": 0.16950169801712037, "step": 8100 }, { "epoch": 0.213630114997407, "grad_norm": 1.632441759109497, "learning_rate": 9.141760464576015e-06, "loss": 0.16771329641342164, "step": 8110 }, { "epoch": 0.2138935306755789, "grad_norm": 2.628774404525757, "learning_rate": 9.139369286728748e-06, "loss": 0.16754781007766723, "step": 8120 }, { "epoch": 0.2141569463537508, "grad_norm": 2.0999362468719482, "learning_rate": 9.136975096069927e-06, "loss": 0.15748915672302247, "step": 8130 }, { "epoch": 0.2144203620319227, "grad_norm": 5.282872200012207, "learning_rate": 9.134577894342146e-06, "loss": 0.15990610122680665, "step": 8140 }, { "epoch": 0.2146837777100946, "grad_norm": 3.857347249984741, "learning_rate": 9.132177683290194e-06, "loss": 0.1556852340698242, "step": 8150 }, { "epoch": 0.21494719338826648, "grad_norm": 1.5862919092178345, "learning_rate": 9.129774464661046e-06, "loss": 0.1623629331588745, "step": 8160 }, { "epoch": 0.21521060906643838, "grad_norm": 3.2748546600341797, "learning_rate": 9.127368240203869e-06, "loss": 0.16577328443527223, "step": 8170 }, { "epoch": 0.21547402474461028, "grad_norm": 2.1174869537353516, "learning_rate": 9.124959011670018e-06, "loss": 0.17539384365081787, "step": 8180 }, { "epoch": 0.21573744042278217, "grad_norm": 7.460575580596924, "learning_rate": 9.122546780813032e-06, "loss": 0.17344415187835693, "step": 8190 }, { "epoch": 0.21600085610095407, "grad_norm": 1.7456245422363281, "learning_rate": 9.120131549388635e-06, "loss": 0.15350668430328368, "step": 8200 }, { "epoch": 0.21626427177912597, "grad_norm": 2.546351194381714, "learning_rate": 9.117713319154742e-06, "loss": 0.1549446702003479, "step": 8210 }, { "epoch": 0.21652768745729786, "grad_norm": 7.08259916305542, "learning_rate": 9.11529209187144e-06, "loss": 0.16445393562316896, "step": 8220 }, { "epoch": 0.21679110313546973, "grad_norm": 1.5833829641342163, "learning_rate": 9.112867869301005e-06, "loss": 0.16179561614990234, "step": 8230 }, { "epoch": 0.21705451881364163, "grad_norm": 4.281214237213135, "learning_rate": 9.110440653207892e-06, "loss": 0.16846580505371095, "step": 8240 }, { "epoch": 0.21731793449181352, "grad_norm": 1.4090991020202637, "learning_rate": 9.10801044535873e-06, "loss": 0.17512372732162476, "step": 8250 }, { "epoch": 0.21758135016998542, "grad_norm": 3.1827614307403564, "learning_rate": 9.105577247522333e-06, "loss": 0.17014739513397217, "step": 8260 }, { "epoch": 0.21784476584815732, "grad_norm": 3.16504168510437, "learning_rate": 9.103141061469684e-06, "loss": 0.16010141372680664, "step": 8270 }, { "epoch": 0.21810818152632921, "grad_norm": 2.7402150630950928, "learning_rate": 9.100701888973948e-06, "loss": 0.17409296035766603, "step": 8280 }, { "epoch": 0.2183715972045011, "grad_norm": 2.7215652465820312, "learning_rate": 9.098259731810456e-06, "loss": 0.15278446674346924, "step": 8290 }, { "epoch": 0.218635012882673, "grad_norm": 2.1530613899230957, "learning_rate": 9.095814591756718e-06, "loss": 0.1633763313293457, "step": 8300 }, { "epoch": 0.2188984285608449, "grad_norm": 1.5115028619766235, "learning_rate": 9.093366470592412e-06, "loss": 0.1602092981338501, "step": 8310 }, { "epoch": 0.2191618442390168, "grad_norm": 4.178118705749512, "learning_rate": 9.090915370099385e-06, "loss": 0.1496757984161377, "step": 8320 }, { "epoch": 0.2194252599171887, "grad_norm": 1.8900744915008545, "learning_rate": 9.088461292061654e-06, "loss": 0.1597871422767639, "step": 8330 }, { "epoch": 0.2196886755953606, "grad_norm": 5.230164051055908, "learning_rate": 9.086004238265404e-06, "loss": 0.1590323805809021, "step": 8340 }, { "epoch": 0.2199520912735325, "grad_norm": 2.8199658393859863, "learning_rate": 9.083544210498988e-06, "loss": 0.1472614049911499, "step": 8350 }, { "epoch": 0.2202155069517044, "grad_norm": 3.172503709793091, "learning_rate": 9.08108121055291e-06, "loss": 0.1604283094406128, "step": 8360 }, { "epoch": 0.22047892262987628, "grad_norm": 6.766580581665039, "learning_rate": 9.07861524021986e-06, "loss": 0.14410111904144288, "step": 8370 }, { "epoch": 0.22074233830804818, "grad_norm": 2.3865718841552734, "learning_rate": 9.076146301294668e-06, "loss": 0.16513171195983886, "step": 8380 }, { "epoch": 0.22100575398622008, "grad_norm": 2.058744192123413, "learning_rate": 9.073674395574339e-06, "loss": 0.15956262350082398, "step": 8390 }, { "epoch": 0.22126916966439197, "grad_norm": 1.3015062808990479, "learning_rate": 9.071199524858033e-06, "loss": 0.1641825556755066, "step": 8400 }, { "epoch": 0.22153258534256387, "grad_norm": 2.801400661468506, "learning_rate": 9.068721690947063e-06, "loss": 0.15922194719314575, "step": 8410 }, { "epoch": 0.22179600102073574, "grad_norm": 2.3748791217803955, "learning_rate": 9.06624089564491e-06, "loss": 0.16635613441467284, "step": 8420 }, { "epoch": 0.22205941669890764, "grad_norm": 2.7498114109039307, "learning_rate": 9.0637571407572e-06, "loss": 0.15818424224853517, "step": 8430 }, { "epoch": 0.22232283237707953, "grad_norm": 1.6713098287582397, "learning_rate": 9.061270428091716e-06, "loss": 0.15592970848083496, "step": 8440 }, { "epoch": 0.22258624805525143, "grad_norm": 1.622771978378296, "learning_rate": 9.058780759458396e-06, "loss": 0.16155658960342406, "step": 8450 }, { "epoch": 0.22284966373342333, "grad_norm": 2.687009334564209, "learning_rate": 9.05628813666933e-06, "loss": 0.15963484048843385, "step": 8460 }, { "epoch": 0.22311307941159522, "grad_norm": 1.166480302810669, "learning_rate": 9.053792561538755e-06, "loss": 0.1643437385559082, "step": 8470 }, { "epoch": 0.22337649508976712, "grad_norm": 2.691371440887451, "learning_rate": 9.051294035883058e-06, "loss": 0.1634756803512573, "step": 8480 }, { "epoch": 0.22363991076793902, "grad_norm": 1.6486828327178955, "learning_rate": 9.048792561520774e-06, "loss": 0.1452239990234375, "step": 8490 }, { "epoch": 0.2239033264461109, "grad_norm": 2.8967273235321045, "learning_rate": 9.046288140272585e-06, "loss": 0.15406816005706786, "step": 8500 }, { "epoch": 0.2241667421242828, "grad_norm": 1.8676751852035522, "learning_rate": 9.043780773961316e-06, "loss": 0.1612716555595398, "step": 8510 }, { "epoch": 0.2244301578024547, "grad_norm": 1.1001532077789307, "learning_rate": 9.041270464411933e-06, "loss": 0.164845871925354, "step": 8520 }, { "epoch": 0.2246935734806266, "grad_norm": 2.024291515350342, "learning_rate": 9.038757213451554e-06, "loss": 0.15984742641448973, "step": 8530 }, { "epoch": 0.2249569891587985, "grad_norm": 2.6794867515563965, "learning_rate": 9.036241022909426e-06, "loss": 0.17022271156311036, "step": 8540 }, { "epoch": 0.2252204048369704, "grad_norm": 1.641011118888855, "learning_rate": 9.033721894616945e-06, "loss": 0.16878123283386232, "step": 8550 }, { "epoch": 0.2254838205151423, "grad_norm": 3.306769847869873, "learning_rate": 9.031199830407638e-06, "loss": 0.15801280736923218, "step": 8560 }, { "epoch": 0.2257472361933142, "grad_norm": 2.3847122192382812, "learning_rate": 9.028674832117174e-06, "loss": 0.16523709297180175, "step": 8570 }, { "epoch": 0.22601065187148608, "grad_norm": 5.088786602020264, "learning_rate": 9.026146901583357e-06, "loss": 0.15568678379058837, "step": 8580 }, { "epoch": 0.22627406754965798, "grad_norm": 2.5084633827209473, "learning_rate": 9.023616040646121e-06, "loss": 0.15518828630447387, "step": 8590 }, { "epoch": 0.22653748322782988, "grad_norm": 5.859106540679932, "learning_rate": 9.021082251147537e-06, "loss": 0.16596362590789795, "step": 8600 }, { "epoch": 0.22680089890600177, "grad_norm": 3.1459579467773438, "learning_rate": 9.018545534931808e-06, "loss": 0.16295735836029052, "step": 8610 }, { "epoch": 0.22706431458417364, "grad_norm": 3.5846993923187256, "learning_rate": 9.016005893845262e-06, "loss": 0.16564003229141236, "step": 8620 }, { "epoch": 0.22732773026234554, "grad_norm": 6.221827030181885, "learning_rate": 9.013463329736363e-06, "loss": 0.1628058910369873, "step": 8630 }, { "epoch": 0.22759114594051744, "grad_norm": 3.833805561065674, "learning_rate": 9.010917844455696e-06, "loss": 0.1664734125137329, "step": 8640 }, { "epoch": 0.22785456161868933, "grad_norm": 3.4085628986358643, "learning_rate": 9.008369439855975e-06, "loss": 0.15829131603240967, "step": 8650 }, { "epoch": 0.22811797729686123, "grad_norm": 2.173166513442993, "learning_rate": 9.00581811779204e-06, "loss": 0.15189130306243898, "step": 8660 }, { "epoch": 0.22838139297503313, "grad_norm": 2.0736770629882812, "learning_rate": 9.003263880120856e-06, "loss": 0.15421407222747802, "step": 8670 }, { "epoch": 0.22864480865320502, "grad_norm": 3.4258475303649902, "learning_rate": 9.000706728701502e-06, "loss": 0.171176278591156, "step": 8680 }, { "epoch": 0.22890822433137692, "grad_norm": 2.0914885997772217, "learning_rate": 8.998146665395187e-06, "loss": 0.1694069266319275, "step": 8690 }, { "epoch": 0.22917164000954882, "grad_norm": 2.060819149017334, "learning_rate": 8.995583692065233e-06, "loss": 0.16404141187667848, "step": 8700 }, { "epoch": 0.2294350556877207, "grad_norm": 3.7161221504211426, "learning_rate": 8.993017810577085e-06, "loss": 0.15688248872756957, "step": 8710 }, { "epoch": 0.2296984713658926, "grad_norm": 2.0565145015716553, "learning_rate": 8.9904490227983e-06, "loss": 0.15675766468048097, "step": 8720 }, { "epoch": 0.2299618870440645, "grad_norm": 2.256774425506592, "learning_rate": 8.987877330598553e-06, "loss": 0.1600069522857666, "step": 8730 }, { "epoch": 0.2302253027222364, "grad_norm": 4.741102695465088, "learning_rate": 8.985302735849635e-06, "loss": 0.161454439163208, "step": 8740 }, { "epoch": 0.2304887184004083, "grad_norm": 1.90498685836792, "learning_rate": 8.982725240425445e-06, "loss": 0.1613085985183716, "step": 8750 }, { "epoch": 0.2307521340785802, "grad_norm": 1.9667786359786987, "learning_rate": 8.980144846201995e-06, "loss": 0.17005567550659179, "step": 8760 }, { "epoch": 0.2310155497567521, "grad_norm": 3.007249593734741, "learning_rate": 8.977561555057407e-06, "loss": 0.16683908700942993, "step": 8770 }, { "epoch": 0.231278965434924, "grad_norm": 2.1584832668304443, "learning_rate": 8.974975368871915e-06, "loss": 0.1523761749267578, "step": 8780 }, { "epoch": 0.23154238111309589, "grad_norm": 3.0813510417938232, "learning_rate": 8.972386289527853e-06, "loss": 0.16379189491271973, "step": 8790 }, { "epoch": 0.23180579679126778, "grad_norm": 2.1121973991394043, "learning_rate": 8.969794318909665e-06, "loss": 0.15196012258529662, "step": 8800 }, { "epoch": 0.23206921246943968, "grad_norm": 3.699657440185547, "learning_rate": 8.967199458903901e-06, "loss": 0.16299126148223878, "step": 8810 }, { "epoch": 0.23233262814761155, "grad_norm": 3.588277816772461, "learning_rate": 8.964601711399213e-06, "loss": 0.15680264234542846, "step": 8820 }, { "epoch": 0.23259604382578344, "grad_norm": 2.2897233963012695, "learning_rate": 8.96200107828635e-06, "loss": 0.17046637535095216, "step": 8830 }, { "epoch": 0.23285945950395534, "grad_norm": 5.105335235595703, "learning_rate": 8.959397561458169e-06, "loss": 0.16920385360717774, "step": 8840 }, { "epoch": 0.23312287518212724, "grad_norm": 1.77255117893219, "learning_rate": 8.956791162809618e-06, "loss": 0.15233737230300903, "step": 8850 }, { "epoch": 0.23338629086029913, "grad_norm": 2.0160791873931885, "learning_rate": 8.954181884237749e-06, "loss": 0.1719114899635315, "step": 8860 }, { "epoch": 0.23364970653847103, "grad_norm": 4.157822132110596, "learning_rate": 8.951569727641706e-06, "loss": 0.16411283016204833, "step": 8870 }, { "epoch": 0.23391312221664293, "grad_norm": 3.8906242847442627, "learning_rate": 8.948954694922731e-06, "loss": 0.15914535522460938, "step": 8880 }, { "epoch": 0.23417653789481482, "grad_norm": 3.2138893604278564, "learning_rate": 8.946336787984156e-06, "loss": 0.14889336824417115, "step": 8890 }, { "epoch": 0.23443995357298672, "grad_norm": 2.799572467803955, "learning_rate": 8.943716008731406e-06, "loss": 0.16782524585723876, "step": 8900 }, { "epoch": 0.23470336925115862, "grad_norm": 2.4984912872314453, "learning_rate": 8.941092359072001e-06, "loss": 0.1567391872406006, "step": 8910 }, { "epoch": 0.2349667849293305, "grad_norm": 2.4963748455047607, "learning_rate": 8.938465840915543e-06, "loss": 0.16283987760543822, "step": 8920 }, { "epoch": 0.2352302006075024, "grad_norm": 1.7579084634780884, "learning_rate": 8.935836456173726e-06, "loss": 0.16331655979156495, "step": 8930 }, { "epoch": 0.2354936162856743, "grad_norm": 4.824265956878662, "learning_rate": 8.933204206760331e-06, "loss": 0.1560542106628418, "step": 8940 }, { "epoch": 0.2357570319638462, "grad_norm": 2.797057867050171, "learning_rate": 8.93056909459122e-06, "loss": 0.15669792890548706, "step": 8950 }, { "epoch": 0.2360204476420181, "grad_norm": 2.3994078636169434, "learning_rate": 8.927931121584342e-06, "loss": 0.15852956771850585, "step": 8960 }, { "epoch": 0.23628386332019, "grad_norm": 1.3274592161178589, "learning_rate": 8.92529028965973e-06, "loss": 0.1484583616256714, "step": 8970 }, { "epoch": 0.2365472789983619, "grad_norm": 2.30122971534729, "learning_rate": 8.922646600739492e-06, "loss": 0.164860999584198, "step": 8980 }, { "epoch": 0.2368106946765338, "grad_norm": 2.800806999206543, "learning_rate": 8.920000056747824e-06, "loss": 0.17204182147979735, "step": 8990 }, { "epoch": 0.23707411035470569, "grad_norm": 3.122523546218872, "learning_rate": 8.917350659610989e-06, "loss": 0.1563807487487793, "step": 9000 }, { "epoch": 0.23733752603287758, "grad_norm": 1.840315818786621, "learning_rate": 8.914698411257334e-06, "loss": 0.16457796096801758, "step": 9010 }, { "epoch": 0.23760094171104945, "grad_norm": 2.3199074268341064, "learning_rate": 8.912043313617282e-06, "loss": 0.16841673851013184, "step": 9020 }, { "epoch": 0.23786435738922135, "grad_norm": 5.059749126434326, "learning_rate": 8.909385368623327e-06, "loss": 0.15710383653640747, "step": 9030 }, { "epoch": 0.23812777306739324, "grad_norm": 1.5548529624938965, "learning_rate": 8.906724578210036e-06, "loss": 0.15609900951385497, "step": 9040 }, { "epoch": 0.23839118874556514, "grad_norm": 1.47372305393219, "learning_rate": 8.904060944314047e-06, "loss": 0.15774188041687012, "step": 9050 }, { "epoch": 0.23865460442373704, "grad_norm": 2.2975046634674072, "learning_rate": 8.901394468874064e-06, "loss": 0.16142666339874268, "step": 9060 }, { "epoch": 0.23891802010190893, "grad_norm": 2.5057778358459473, "learning_rate": 8.898725153830867e-06, "loss": 0.16984405517578124, "step": 9070 }, { "epoch": 0.23918143578008083, "grad_norm": 1.7219542264938354, "learning_rate": 8.8960530011273e-06, "loss": 0.1623287558555603, "step": 9080 }, { "epoch": 0.23944485145825273, "grad_norm": 2.7141594886779785, "learning_rate": 8.893378012708267e-06, "loss": 0.1621023893356323, "step": 9090 }, { "epoch": 0.23970826713642462, "grad_norm": 3.7328109741210938, "learning_rate": 8.890700190520742e-06, "loss": 0.15648460388183594, "step": 9100 }, { "epoch": 0.23997168281459652, "grad_norm": 2.0318422317504883, "learning_rate": 8.88801953651376e-06, "loss": 0.14856163263320923, "step": 9110 }, { "epoch": 0.24023509849276842, "grad_norm": 2.599996328353882, "learning_rate": 8.885336052638417e-06, "loss": 0.16607420444488524, "step": 9120 }, { "epoch": 0.24049851417094031, "grad_norm": 1.3064515590667725, "learning_rate": 8.882649740847865e-06, "loss": 0.16008396148681642, "step": 9130 }, { "epoch": 0.2407619298491122, "grad_norm": 1.4866786003112793, "learning_rate": 8.87996060309732e-06, "loss": 0.16296811103820802, "step": 9140 }, { "epoch": 0.2410253455272841, "grad_norm": 1.822614312171936, "learning_rate": 8.877268641344052e-06, "loss": 0.15362731218338013, "step": 9150 }, { "epoch": 0.241288761205456, "grad_norm": 1.7250608205795288, "learning_rate": 8.874573857547388e-06, "loss": 0.14807240962982177, "step": 9160 }, { "epoch": 0.2415521768836279, "grad_norm": 1.412099003791809, "learning_rate": 8.871876253668707e-06, "loss": 0.17366211414337157, "step": 9170 }, { "epoch": 0.2418155925617998, "grad_norm": 1.9296458959579468, "learning_rate": 8.869175831671444e-06, "loss": 0.14754871129989625, "step": 9180 }, { "epoch": 0.2420790082399717, "grad_norm": 3.8179678916931152, "learning_rate": 8.866472593521077e-06, "loss": 0.16143412590026857, "step": 9190 }, { "epoch": 0.2423424239181436, "grad_norm": 1.5294030904769897, "learning_rate": 8.863766541185144e-06, "loss": 0.15424509048461915, "step": 9200 }, { "epoch": 0.2426058395963155, "grad_norm": 2.330871105194092, "learning_rate": 8.861057676633225e-06, "loss": 0.17234952449798585, "step": 9210 }, { "epoch": 0.24286925527448736, "grad_norm": 2.5336177349090576, "learning_rate": 8.858346001836949e-06, "loss": 0.16644033193588256, "step": 9220 }, { "epoch": 0.24313267095265925, "grad_norm": 1.7618353366851807, "learning_rate": 8.855631518769989e-06, "loss": 0.13928444385528566, "step": 9230 }, { "epoch": 0.24339608663083115, "grad_norm": 2.9945931434631348, "learning_rate": 8.852914229408065e-06, "loss": 0.16863577365875243, "step": 9240 }, { "epoch": 0.24365950230900305, "grad_norm": 2.2148611545562744, "learning_rate": 8.850194135728936e-06, "loss": 0.158562171459198, "step": 9250 }, { "epoch": 0.24392291798717494, "grad_norm": 2.760073184967041, "learning_rate": 8.847471239712401e-06, "loss": 0.16951042413711548, "step": 9260 }, { "epoch": 0.24418633366534684, "grad_norm": 4.50670051574707, "learning_rate": 8.844745543340306e-06, "loss": 0.17311689853668213, "step": 9270 }, { "epoch": 0.24444974934351874, "grad_norm": 2.495347023010254, "learning_rate": 8.842017048596527e-06, "loss": 0.15540337562561035, "step": 9280 }, { "epoch": 0.24471316502169063, "grad_norm": 3.4242136478424072, "learning_rate": 8.839285757466983e-06, "loss": 0.14489219188690186, "step": 9290 }, { "epoch": 0.24497658069986253, "grad_norm": 2.3201262950897217, "learning_rate": 8.836551671939621e-06, "loss": 0.16978344917297364, "step": 9300 }, { "epoch": 0.24523999637803442, "grad_norm": 2.3927440643310547, "learning_rate": 8.833814794004431e-06, "loss": 0.16652183532714843, "step": 9310 }, { "epoch": 0.24550341205620632, "grad_norm": 1.3264234066009521, "learning_rate": 8.83107512565343e-06, "loss": 0.15099196434020995, "step": 9320 }, { "epoch": 0.24576682773437822, "grad_norm": 5.862160682678223, "learning_rate": 8.828332668880664e-06, "loss": 0.145677649974823, "step": 9330 }, { "epoch": 0.24603024341255011, "grad_norm": 3.245896816253662, "learning_rate": 8.825587425682213e-06, "loss": 0.15179002285003662, "step": 9340 }, { "epoch": 0.246293659090722, "grad_norm": 2.908393383026123, "learning_rate": 8.822839398056186e-06, "loss": 0.15241063833236695, "step": 9350 }, { "epoch": 0.2465570747688939, "grad_norm": 2.4011471271514893, "learning_rate": 8.820088588002712e-06, "loss": 0.1604534149169922, "step": 9360 }, { "epoch": 0.2468204904470658, "grad_norm": 1.9211633205413818, "learning_rate": 8.817334997523952e-06, "loss": 0.15280871391296386, "step": 9370 }, { "epoch": 0.2470839061252377, "grad_norm": 2.162684440612793, "learning_rate": 8.814578628624087e-06, "loss": 0.16439309120178222, "step": 9380 }, { "epoch": 0.2473473218034096, "grad_norm": 1.531745195388794, "learning_rate": 8.81181948330932e-06, "loss": 0.16309999227523803, "step": 9390 }, { "epoch": 0.2476107374815815, "grad_norm": 5.067048072814941, "learning_rate": 8.80905756358788e-06, "loss": 0.1620950698852539, "step": 9400 }, { "epoch": 0.2478741531597534, "grad_norm": 1.6467033624649048, "learning_rate": 8.806292871470006e-06, "loss": 0.15698472261428834, "step": 9410 }, { "epoch": 0.24813756883792526, "grad_norm": 2.1796987056732178, "learning_rate": 8.803525408967964e-06, "loss": 0.16618902683258058, "step": 9420 }, { "epoch": 0.24840098451609716, "grad_norm": 2.877361536026001, "learning_rate": 8.80075517809603e-06, "loss": 0.17449507713317872, "step": 9430 }, { "epoch": 0.24866440019426905, "grad_norm": 1.9902689456939697, "learning_rate": 8.7979821808705e-06, "loss": 0.16091582775115967, "step": 9440 }, { "epoch": 0.24892781587244095, "grad_norm": 4.519589900970459, "learning_rate": 8.79520641930968e-06, "loss": 0.1622037410736084, "step": 9450 }, { "epoch": 0.24919123155061285, "grad_norm": 2.1092770099639893, "learning_rate": 8.792427895433887e-06, "loss": 0.15606818199157715, "step": 9460 }, { "epoch": 0.24945464722878474, "grad_norm": 3.5475494861602783, "learning_rate": 8.789646611265454e-06, "loss": 0.15774829387664796, "step": 9470 }, { "epoch": 0.24971806290695664, "grad_norm": 1.8247815370559692, "learning_rate": 8.786862568828716e-06, "loss": 0.16346067190170288, "step": 9480 }, { "epoch": 0.24998147858512854, "grad_norm": 1.3374102115631104, "learning_rate": 8.784075770150024e-06, "loss": 0.17037945985794067, "step": 9490 }, { "epoch": 0.25024489426330043, "grad_norm": 1.996683955192566, "learning_rate": 8.781286217257725e-06, "loss": 0.17302203178405762, "step": 9500 }, { "epoch": 0.25050830994147233, "grad_norm": 3.139533042907715, "learning_rate": 8.778493912182181e-06, "loss": 0.16581701040267943, "step": 9510 }, { "epoch": 0.2507717256196442, "grad_norm": 1.8831709623336792, "learning_rate": 8.77569885695575e-06, "loss": 0.16950664520263672, "step": 9520 }, { "epoch": 0.2510351412978161, "grad_norm": 3.345874309539795, "learning_rate": 8.772901053612793e-06, "loss": 0.17033586502075196, "step": 9530 }, { "epoch": 0.251298556975988, "grad_norm": 2.02817440032959, "learning_rate": 8.77010050418967e-06, "loss": 0.15563113689422609, "step": 9540 }, { "epoch": 0.2515619726541599, "grad_norm": 4.048304080963135, "learning_rate": 8.767297210724745e-06, "loss": 0.18140754699707032, "step": 9550 }, { "epoch": 0.2518253883323318, "grad_norm": 1.861695408821106, "learning_rate": 8.764491175258377e-06, "loss": 0.15654791593551637, "step": 9560 }, { "epoch": 0.2520888040105037, "grad_norm": 2.869658946990967, "learning_rate": 8.761682399832916e-06, "loss": 0.14573278427124023, "step": 9570 }, { "epoch": 0.2523522196886756, "grad_norm": 5.089749336242676, "learning_rate": 8.758870886492712e-06, "loss": 0.17571573257446288, "step": 9580 }, { "epoch": 0.2526156353668475, "grad_norm": 1.067087173461914, "learning_rate": 8.756056637284105e-06, "loss": 0.15994391441345215, "step": 9590 }, { "epoch": 0.2528790510450194, "grad_norm": 2.5959644317626953, "learning_rate": 8.753239654255427e-06, "loss": 0.1483623743057251, "step": 9600 }, { "epoch": 0.2531424667231913, "grad_norm": 1.9962823390960693, "learning_rate": 8.750419939457002e-06, "loss": 0.1617504119873047, "step": 9610 }, { "epoch": 0.2534058824013632, "grad_norm": 1.91349458694458, "learning_rate": 8.747597494941135e-06, "loss": 0.1619559645652771, "step": 9620 }, { "epoch": 0.2536692980795351, "grad_norm": 1.1921491622924805, "learning_rate": 8.74477232276213e-06, "loss": 0.16088442802429198, "step": 9630 }, { "epoch": 0.253932713757707, "grad_norm": 6.383599281311035, "learning_rate": 8.741944424976263e-06, "loss": 0.15694539546966552, "step": 9640 }, { "epoch": 0.2541961294358789, "grad_norm": 5.765538215637207, "learning_rate": 8.739113803641802e-06, "loss": 0.16297593116760253, "step": 9650 }, { "epoch": 0.2544595451140508, "grad_norm": 3.400621175765991, "learning_rate": 8.736280460818997e-06, "loss": 0.16458886861801147, "step": 9660 }, { "epoch": 0.2547229607922227, "grad_norm": 2.669766664505005, "learning_rate": 8.733444398570075e-06, "loss": 0.16935944557189941, "step": 9670 }, { "epoch": 0.25498637647039457, "grad_norm": 2.749565839767456, "learning_rate": 8.730605618959244e-06, "loss": 0.16778312921524047, "step": 9680 }, { "epoch": 0.25524979214856647, "grad_norm": 4.596763610839844, "learning_rate": 8.727764124052693e-06, "loss": 0.16659178733825683, "step": 9690 }, { "epoch": 0.25551320782673836, "grad_norm": 3.5215673446655273, "learning_rate": 8.724919915918583e-06, "loss": 0.1622615098953247, "step": 9700 }, { "epoch": 0.2557766235049102, "grad_norm": 1.5542415380477905, "learning_rate": 8.722072996627052e-06, "loss": 0.15729684829711915, "step": 9710 }, { "epoch": 0.2560400391830821, "grad_norm": 3.2951509952545166, "learning_rate": 8.719223368250209e-06, "loss": 0.1505728244781494, "step": 9720 }, { "epoch": 0.256303454861254, "grad_norm": 1.5923088788986206, "learning_rate": 8.716371032862138e-06, "loss": 0.15378050804138182, "step": 9730 }, { "epoch": 0.2565668705394259, "grad_norm": 3.6390750408172607, "learning_rate": 8.713515992538893e-06, "loss": 0.14985698461532593, "step": 9740 }, { "epoch": 0.2568302862175978, "grad_norm": 6.172433853149414, "learning_rate": 8.710658249358493e-06, "loss": 0.1634383201599121, "step": 9750 }, { "epoch": 0.2570937018957697, "grad_norm": 2.7735016345977783, "learning_rate": 8.707797805400926e-06, "loss": 0.14753029346466065, "step": 9760 }, { "epoch": 0.2573571175739416, "grad_norm": 5.477529525756836, "learning_rate": 8.70493466274815e-06, "loss": 0.15848854780197144, "step": 9770 }, { "epoch": 0.2576205332521135, "grad_norm": 3.0180394649505615, "learning_rate": 8.702068823484081e-06, "loss": 0.16901767253875732, "step": 9780 }, { "epoch": 0.2578839489302854, "grad_norm": 3.423121452331543, "learning_rate": 8.699200289694602e-06, "loss": 0.1445619583129883, "step": 9790 }, { "epoch": 0.2581473646084573, "grad_norm": 1.4966647624969482, "learning_rate": 8.696329063467555e-06, "loss": 0.155652916431427, "step": 9800 }, { "epoch": 0.25841078028662917, "grad_norm": 1.749238133430481, "learning_rate": 8.693455146892743e-06, "loss": 0.15694258213043213, "step": 9810 }, { "epoch": 0.25867419596480107, "grad_norm": 1.222490668296814, "learning_rate": 8.690578542061925e-06, "loss": 0.16606473922729492, "step": 9820 }, { "epoch": 0.25893761164297296, "grad_norm": 1.7225552797317505, "learning_rate": 8.68769925106882e-06, "loss": 0.15755114555358887, "step": 9830 }, { "epoch": 0.25920102732114486, "grad_norm": 1.6634786128997803, "learning_rate": 8.684817276009098e-06, "loss": 0.15253400802612305, "step": 9840 }, { "epoch": 0.25946444299931676, "grad_norm": 2.851823091506958, "learning_rate": 8.681932618980387e-06, "loss": 0.16305571794509888, "step": 9850 }, { "epoch": 0.25972785867748865, "grad_norm": 6.671760082244873, "learning_rate": 8.679045282082262e-06, "loss": 0.1627673864364624, "step": 9860 }, { "epoch": 0.25999127435566055, "grad_norm": 1.2561050653457642, "learning_rate": 8.676155267416255e-06, "loss": 0.15623934268951417, "step": 9870 }, { "epoch": 0.26025469003383245, "grad_norm": 3.166581630706787, "learning_rate": 8.67326257708584e-06, "loss": 0.1652980089187622, "step": 9880 }, { "epoch": 0.26051810571200434, "grad_norm": 2.681283712387085, "learning_rate": 8.670367213196447e-06, "loss": 0.15842158794403077, "step": 9890 }, { "epoch": 0.26078152139017624, "grad_norm": 1.6627826690673828, "learning_rate": 8.667469177855438e-06, "loss": 0.16247366666793822, "step": 9900 }, { "epoch": 0.26104493706834814, "grad_norm": 1.5521444082260132, "learning_rate": 8.664568473172136e-06, "loss": 0.16585423946380615, "step": 9910 }, { "epoch": 0.26130835274652003, "grad_norm": 1.6658296585083008, "learning_rate": 8.661665101257796e-06, "loss": 0.16085288524627686, "step": 9920 }, { "epoch": 0.26157176842469193, "grad_norm": 2.5381081104278564, "learning_rate": 8.65875906422562e-06, "loss": 0.16058698892593384, "step": 9930 }, { "epoch": 0.2618351841028638, "grad_norm": 3.809955596923828, "learning_rate": 8.655850364190743e-06, "loss": 0.15225706100463868, "step": 9940 }, { "epoch": 0.2620985997810357, "grad_norm": 3.475472927093506, "learning_rate": 8.652939003270242e-06, "loss": 0.15038652420043946, "step": 9950 }, { "epoch": 0.2623620154592076, "grad_norm": 2.832509994506836, "learning_rate": 8.650024983583136e-06, "loss": 0.16575868129730226, "step": 9960 }, { "epoch": 0.2626254311373795, "grad_norm": 2.201223373413086, "learning_rate": 8.64710830725037e-06, "loss": 0.15513894557952881, "step": 9970 }, { "epoch": 0.2628888468155514, "grad_norm": 2.475552797317505, "learning_rate": 8.64418897639483e-06, "loss": 0.14898781776428222, "step": 9980 }, { "epoch": 0.2631522624937233, "grad_norm": 4.816128253936768, "learning_rate": 8.641266993141328e-06, "loss": 0.16609668731689453, "step": 9990 }, { "epoch": 0.2634156781718952, "grad_norm": 1.793197751045227, "learning_rate": 8.638342359616611e-06, "loss": 0.15695908069610595, "step": 10000 }, { "epoch": 0.2636790938500671, "grad_norm": 2.404160499572754, "learning_rate": 8.635415077949352e-06, "loss": 0.14947168827056884, "step": 10010 }, { "epoch": 0.263942509528239, "grad_norm": 4.972931861877441, "learning_rate": 8.632485150270155e-06, "loss": 0.15364493131637574, "step": 10020 }, { "epoch": 0.2642059252064109, "grad_norm": 3.8173422813415527, "learning_rate": 8.62955257871155e-06, "loss": 0.15477635860443115, "step": 10030 }, { "epoch": 0.2644693408845828, "grad_norm": 3.551318407058716, "learning_rate": 8.626617365407982e-06, "loss": 0.15510861873626708, "step": 10040 }, { "epoch": 0.2647327565627547, "grad_norm": 1.0379395484924316, "learning_rate": 8.62367951249583e-06, "loss": 0.15710582733154296, "step": 10050 }, { "epoch": 0.2649961722409266, "grad_norm": 1.2080576419830322, "learning_rate": 8.620739022113391e-06, "loss": 0.16301966905593873, "step": 10060 }, { "epoch": 0.2652595879190985, "grad_norm": 2.6542580127716064, "learning_rate": 8.61779589640088e-06, "loss": 0.15781991481781005, "step": 10070 }, { "epoch": 0.2655230035972704, "grad_norm": 2.314462661743164, "learning_rate": 8.61485013750043e-06, "loss": 0.15660469532012938, "step": 10080 }, { "epoch": 0.2657864192754423, "grad_norm": 2.050550937652588, "learning_rate": 8.611901747556093e-06, "loss": 0.161323082447052, "step": 10090 }, { "epoch": 0.2660498349536142, "grad_norm": 1.9342694282531738, "learning_rate": 8.608950728713834e-06, "loss": 0.15460388660430907, "step": 10100 }, { "epoch": 0.266313250631786, "grad_norm": 3.7896642684936523, "learning_rate": 8.605997083121529e-06, "loss": 0.15351288318634032, "step": 10110 }, { "epoch": 0.2665766663099579, "grad_norm": 1.800229549407959, "learning_rate": 8.603040812928971e-06, "loss": 0.16823055744171142, "step": 10120 }, { "epoch": 0.2668400819881298, "grad_norm": 1.0924490690231323, "learning_rate": 8.600081920287864e-06, "loss": 0.1687934398651123, "step": 10130 }, { "epoch": 0.2671034976663017, "grad_norm": 2.7982077598571777, "learning_rate": 8.597120407351813e-06, "loss": 0.13443381786346437, "step": 10140 }, { "epoch": 0.2673669133444736, "grad_norm": 2.768848180770874, "learning_rate": 8.594156276276336e-06, "loss": 0.1574810266494751, "step": 10150 }, { "epoch": 0.2676303290226455, "grad_norm": 3.160487174987793, "learning_rate": 8.591189529218855e-06, "loss": 0.14141353368759155, "step": 10160 }, { "epoch": 0.2678937447008174, "grad_norm": 1.2375527620315552, "learning_rate": 8.5882201683387e-06, "loss": 0.16304287910461426, "step": 10170 }, { "epoch": 0.2681571603789893, "grad_norm": 1.7061452865600586, "learning_rate": 8.585248195797097e-06, "loss": 0.1571220874786377, "step": 10180 }, { "epoch": 0.2684205760571612, "grad_norm": 3.09134840965271, "learning_rate": 8.582273613757175e-06, "loss": 0.15358266830444336, "step": 10190 }, { "epoch": 0.2686839917353331, "grad_norm": 2.0934906005859375, "learning_rate": 8.579296424383964e-06, "loss": 0.1637447237968445, "step": 10200 }, { "epoch": 0.268947407413505, "grad_norm": 2.253718614578247, "learning_rate": 8.576316629844394e-06, "loss": 0.1568474292755127, "step": 10210 }, { "epoch": 0.2692108230916769, "grad_norm": 4.135497093200684, "learning_rate": 8.573334232307284e-06, "loss": 0.1591839909553528, "step": 10220 }, { "epoch": 0.2694742387698488, "grad_norm": 3.7061378955841064, "learning_rate": 8.570349233943353e-06, "loss": 0.16083271503448487, "step": 10230 }, { "epoch": 0.26973765444802067, "grad_norm": 1.9966962337493896, "learning_rate": 8.56736163692521e-06, "loss": 0.15321484804153443, "step": 10240 }, { "epoch": 0.27000107012619257, "grad_norm": 4.6938886642456055, "learning_rate": 8.564371443427362e-06, "loss": 0.14912147521972657, "step": 10250 }, { "epoch": 0.27026448580436446, "grad_norm": 3.44415283203125, "learning_rate": 8.561378655626195e-06, "loss": 0.16667466163635253, "step": 10260 }, { "epoch": 0.27052790148253636, "grad_norm": 2.5307276248931885, "learning_rate": 8.558383275699993e-06, "loss": 0.15552324056625366, "step": 10270 }, { "epoch": 0.27079131716070826, "grad_norm": 3.1042275428771973, "learning_rate": 8.55538530582892e-06, "loss": 0.1551973819732666, "step": 10280 }, { "epoch": 0.27105473283888015, "grad_norm": 1.5978809595108032, "learning_rate": 8.55238474819503e-06, "loss": 0.17426931858062744, "step": 10290 }, { "epoch": 0.27131814851705205, "grad_norm": 2.9458115100860596, "learning_rate": 8.549381604982257e-06, "loss": 0.15902330875396728, "step": 10300 }, { "epoch": 0.27158156419522395, "grad_norm": 1.490597128868103, "learning_rate": 8.546375878376419e-06, "loss": 0.16411356925964354, "step": 10310 }, { "epoch": 0.27184497987339584, "grad_norm": 2.991586446762085, "learning_rate": 8.543367570565212e-06, "loss": 0.14475102424621583, "step": 10320 }, { "epoch": 0.27210839555156774, "grad_norm": 5.638370990753174, "learning_rate": 8.540356683738214e-06, "loss": 0.16617579460144044, "step": 10330 }, { "epoch": 0.27237181122973964, "grad_norm": 2.2730600833892822, "learning_rate": 8.537343220086876e-06, "loss": 0.15647420883178711, "step": 10340 }, { "epoch": 0.27263522690791153, "grad_norm": 3.4475557804107666, "learning_rate": 8.534327181804529e-06, "loss": 0.15798119306564332, "step": 10350 }, { "epoch": 0.27289864258608343, "grad_norm": 2.4169955253601074, "learning_rate": 8.531308571086377e-06, "loss": 0.16074966192245482, "step": 10360 }, { "epoch": 0.2731620582642553, "grad_norm": 3.541322708129883, "learning_rate": 8.528287390129493e-06, "loss": 0.16794984340667723, "step": 10370 }, { "epoch": 0.2734254739424272, "grad_norm": 1.2107927799224854, "learning_rate": 8.52526364113282e-06, "loss": 0.15782207250595093, "step": 10380 }, { "epoch": 0.2736888896205991, "grad_norm": 2.3889052867889404, "learning_rate": 8.52223732629718e-06, "loss": 0.14845848083496094, "step": 10390 }, { "epoch": 0.273952305298771, "grad_norm": 1.9943636655807495, "learning_rate": 8.519208447825252e-06, "loss": 0.15369392633438111, "step": 10400 }, { "epoch": 0.2742157209769429, "grad_norm": 3.5213592052459717, "learning_rate": 8.516177007921581e-06, "loss": 0.15171506404876708, "step": 10410 }, { "epoch": 0.2744791366551148, "grad_norm": 5.486818313598633, "learning_rate": 8.513143008792584e-06, "loss": 0.15276501178741456, "step": 10420 }, { "epoch": 0.2747425523332867, "grad_norm": 2.0169663429260254, "learning_rate": 8.510106452646536e-06, "loss": 0.15512624979019166, "step": 10430 }, { "epoch": 0.2750059680114586, "grad_norm": 4.291604995727539, "learning_rate": 8.507067341693574e-06, "loss": 0.16040148735046386, "step": 10440 }, { "epoch": 0.2752693836896305, "grad_norm": 3.779468536376953, "learning_rate": 8.504025678145691e-06, "loss": 0.16578627824783326, "step": 10450 }, { "epoch": 0.2755327993678024, "grad_norm": 2.681323766708374, "learning_rate": 8.500981464216744e-06, "loss": 0.16184996366500853, "step": 10460 }, { "epoch": 0.2757962150459743, "grad_norm": 6.232359409332275, "learning_rate": 8.49793470212244e-06, "loss": 0.15841634273529054, "step": 10470 }, { "epoch": 0.2760596307241462, "grad_norm": 1.5947937965393066, "learning_rate": 8.494885394080349e-06, "loss": 0.16080541610717775, "step": 10480 }, { "epoch": 0.2763230464023181, "grad_norm": 3.2494988441467285, "learning_rate": 8.491833542309884e-06, "loss": 0.1673792600631714, "step": 10490 }, { "epoch": 0.2765864620804899, "grad_norm": 2.233052968978882, "learning_rate": 8.488779149032318e-06, "loss": 0.16296247243881226, "step": 10500 }, { "epoch": 0.2768498777586618, "grad_norm": 1.295525074005127, "learning_rate": 8.485722216470767e-06, "loss": 0.15611603260040283, "step": 10510 }, { "epoch": 0.2771132934368337, "grad_norm": 2.5284488201141357, "learning_rate": 8.4826627468502e-06, "loss": 0.1516016960144043, "step": 10520 }, { "epoch": 0.2773767091150056, "grad_norm": 1.8574440479278564, "learning_rate": 8.479600742397433e-06, "loss": 0.1580876350402832, "step": 10530 }, { "epoch": 0.2776401247931775, "grad_norm": 2.617496967315674, "learning_rate": 8.476536205341121e-06, "loss": 0.15167546272277832, "step": 10540 }, { "epoch": 0.2779035404713494, "grad_norm": 5.058963775634766, "learning_rate": 8.47346913791177e-06, "loss": 0.16220694780349731, "step": 10550 }, { "epoch": 0.2781669561495213, "grad_norm": 1.6570475101470947, "learning_rate": 8.47039954234172e-06, "loss": 0.15841941833496093, "step": 10560 }, { "epoch": 0.2784303718276932, "grad_norm": 1.73186457157135, "learning_rate": 8.467327420865159e-06, "loss": 0.15970678329467775, "step": 10570 }, { "epoch": 0.2786937875058651, "grad_norm": 4.058094501495361, "learning_rate": 8.464252775718108e-06, "loss": 0.15356396436691283, "step": 10580 }, { "epoch": 0.278957203184037, "grad_norm": 1.596170425415039, "learning_rate": 8.461175609138426e-06, "loss": 0.15697054862976073, "step": 10590 }, { "epoch": 0.2792206188622089, "grad_norm": 3.5947647094726562, "learning_rate": 8.458095923365807e-06, "loss": 0.15875551700592042, "step": 10600 }, { "epoch": 0.2794840345403808, "grad_norm": 1.034542202949524, "learning_rate": 8.455013720641781e-06, "loss": 0.15736348628997804, "step": 10610 }, { "epoch": 0.2797474502185527, "grad_norm": 2.3508353233337402, "learning_rate": 8.451929003209707e-06, "loss": 0.17189180850982666, "step": 10620 }, { "epoch": 0.2800108658967246, "grad_norm": 2.2756283283233643, "learning_rate": 8.448841773314775e-06, "loss": 0.15174827575683594, "step": 10630 }, { "epoch": 0.2802742815748965, "grad_norm": 3.848141670227051, "learning_rate": 8.445752033204005e-06, "loss": 0.152091908454895, "step": 10640 }, { "epoch": 0.2805376972530684, "grad_norm": 2.23557448387146, "learning_rate": 8.442659785126242e-06, "loss": 0.15513279438018798, "step": 10650 }, { "epoch": 0.28080111293124027, "grad_norm": 2.5446348190307617, "learning_rate": 8.43956503133216e-06, "loss": 0.16328028440475464, "step": 10660 }, { "epoch": 0.28106452860941217, "grad_norm": 1.522862434387207, "learning_rate": 8.436467774074252e-06, "loss": 0.1660879969596863, "step": 10670 }, { "epoch": 0.28132794428758406, "grad_norm": 1.8834360837936401, "learning_rate": 8.433368015606835e-06, "loss": 0.16038596630096436, "step": 10680 }, { "epoch": 0.28159135996575596, "grad_norm": 2.0980231761932373, "learning_rate": 8.430265758186051e-06, "loss": 0.15557916164398194, "step": 10690 }, { "epoch": 0.28185477564392786, "grad_norm": 4.8933868408203125, "learning_rate": 8.427161004069852e-06, "loss": 0.1565809965133667, "step": 10700 }, { "epoch": 0.28211819132209975, "grad_norm": 1.8026556968688965, "learning_rate": 8.424053755518013e-06, "loss": 0.16953803300857545, "step": 10710 }, { "epoch": 0.28238160700027165, "grad_norm": 2.7713541984558105, "learning_rate": 8.420944014792127e-06, "loss": 0.1578225612640381, "step": 10720 }, { "epoch": 0.28264502267844355, "grad_norm": 2.624530553817749, "learning_rate": 8.417831784155593e-06, "loss": 0.1524683952331543, "step": 10730 }, { "epoch": 0.28290843835661544, "grad_norm": 5.091099739074707, "learning_rate": 8.41471706587363e-06, "loss": 0.16272624731063842, "step": 10740 }, { "epoch": 0.28317185403478734, "grad_norm": 1.479628562927246, "learning_rate": 8.411599862213261e-06, "loss": 0.14781986474990844, "step": 10750 }, { "epoch": 0.28343526971295924, "grad_norm": 1.3618533611297607, "learning_rate": 8.408480175443327e-06, "loss": 0.1606688141822815, "step": 10760 }, { "epoch": 0.28369868539113113, "grad_norm": 1.770098328590393, "learning_rate": 8.405358007834465e-06, "loss": 0.16384464502334595, "step": 10770 }, { "epoch": 0.28396210106930303, "grad_norm": 6.1391921043396, "learning_rate": 8.402233361659124e-06, "loss": 0.15669751167297363, "step": 10780 }, { "epoch": 0.2842255167474749, "grad_norm": 3.181853771209717, "learning_rate": 8.399106239191559e-06, "loss": 0.16431407928466796, "step": 10790 }, { "epoch": 0.2844889324256468, "grad_norm": 8.367878913879395, "learning_rate": 8.395976642707823e-06, "loss": 0.15155136585235596, "step": 10800 }, { "epoch": 0.2847523481038187, "grad_norm": 2.398562431335449, "learning_rate": 8.39284457448577e-06, "loss": 0.16833133697509767, "step": 10810 }, { "epoch": 0.2850157637819906, "grad_norm": 2.895080089569092, "learning_rate": 8.389710036805058e-06, "loss": 0.14511170387268066, "step": 10820 }, { "epoch": 0.2852791794601625, "grad_norm": 2.2605254650115967, "learning_rate": 8.386573031947137e-06, "loss": 0.15258705615997314, "step": 10830 }, { "epoch": 0.2855425951383344, "grad_norm": 1.437419056892395, "learning_rate": 8.383433562195253e-06, "loss": 0.16526925563812256, "step": 10840 }, { "epoch": 0.2858060108165063, "grad_norm": 5.544875621795654, "learning_rate": 8.380291629834449e-06, "loss": 0.16078643798828124, "step": 10850 }, { "epoch": 0.2860694264946782, "grad_norm": 1.347224235534668, "learning_rate": 8.377147237151558e-06, "loss": 0.16359195709228516, "step": 10860 }, { "epoch": 0.2863328421728501, "grad_norm": 4.9796600341796875, "learning_rate": 8.374000386435206e-06, "loss": 0.16216386556625367, "step": 10870 }, { "epoch": 0.286596257851022, "grad_norm": 2.594545841217041, "learning_rate": 8.370851079975806e-06, "loss": 0.15977106094360352, "step": 10880 }, { "epoch": 0.2868596735291939, "grad_norm": 4.301241874694824, "learning_rate": 8.367699320065558e-06, "loss": 0.1588426113128662, "step": 10890 }, { "epoch": 0.28712308920736573, "grad_norm": 5.651462078094482, "learning_rate": 8.364545108998452e-06, "loss": 0.1654369831085205, "step": 10900 }, { "epoch": 0.28738650488553763, "grad_norm": 1.683243751525879, "learning_rate": 8.361388449070256e-06, "loss": 0.16154738664627075, "step": 10910 }, { "epoch": 0.2876499205637095, "grad_norm": 2.1023733615875244, "learning_rate": 8.358229342578523e-06, "loss": 0.16250889301300048, "step": 10920 }, { "epoch": 0.2879133362418814, "grad_norm": 1.4962489604949951, "learning_rate": 8.35506779182259e-06, "loss": 0.156103515625, "step": 10930 }, { "epoch": 0.2881767519200533, "grad_norm": 2.486417293548584, "learning_rate": 8.351903799103566e-06, "loss": 0.16014814376831055, "step": 10940 }, { "epoch": 0.2884401675982252, "grad_norm": 2.6601850986480713, "learning_rate": 8.348737366724346e-06, "loss": 0.1588624358177185, "step": 10950 }, { "epoch": 0.2887035832763971, "grad_norm": 1.4099342823028564, "learning_rate": 8.345568496989591e-06, "loss": 0.14137305021286012, "step": 10960 }, { "epoch": 0.288966998954569, "grad_norm": 2.5081686973571777, "learning_rate": 8.342397192205747e-06, "loss": 0.1708172082901001, "step": 10970 }, { "epoch": 0.2892304146327409, "grad_norm": 1.9384422302246094, "learning_rate": 8.33922345468102e-06, "loss": 0.13939714431762695, "step": 10980 }, { "epoch": 0.2894938303109128, "grad_norm": 2.3935341835021973, "learning_rate": 8.336047286725399e-06, "loss": 0.15466525554656982, "step": 10990 }, { "epoch": 0.2897572459890847, "grad_norm": 3.0324933528900146, "learning_rate": 8.332868690650631e-06, "loss": 0.14788538217544556, "step": 11000 }, { "epoch": 0.2900206616672566, "grad_norm": 3.7712459564208984, "learning_rate": 8.32968766877024e-06, "loss": 0.16136906147003174, "step": 11010 }, { "epoch": 0.2902840773454285, "grad_norm": 2.0734639167785645, "learning_rate": 8.326504223399507e-06, "loss": 0.16661523580551146, "step": 11020 }, { "epoch": 0.2905474930236004, "grad_norm": 2.1997039318084717, "learning_rate": 8.323318356855483e-06, "loss": 0.16197621822357178, "step": 11030 }, { "epoch": 0.2908109087017723, "grad_norm": 1.8514530658721924, "learning_rate": 8.32013007145698e-06, "loss": 0.15541130304336548, "step": 11040 }, { "epoch": 0.2910743243799442, "grad_norm": 1.846004843711853, "learning_rate": 8.316939369524567e-06, "loss": 0.16802034378051758, "step": 11050 }, { "epoch": 0.2913377400581161, "grad_norm": 1.3969190120697021, "learning_rate": 8.313746253380577e-06, "loss": 0.15534493923187256, "step": 11060 }, { "epoch": 0.291601155736288, "grad_norm": 1.364698052406311, "learning_rate": 8.310550725349097e-06, "loss": 0.14771625995635987, "step": 11070 }, { "epoch": 0.2918645714144599, "grad_norm": 3.224146604537964, "learning_rate": 8.307352787755969e-06, "loss": 0.16814959049224854, "step": 11080 }, { "epoch": 0.29212798709263177, "grad_norm": 2.971562147140503, "learning_rate": 8.30415244292879e-06, "loss": 0.16459670066833496, "step": 11090 }, { "epoch": 0.29239140277080367, "grad_norm": 6.80068302154541, "learning_rate": 8.30094969319691e-06, "loss": 0.16517716646194458, "step": 11100 }, { "epoch": 0.29265481844897556, "grad_norm": 4.385847568511963, "learning_rate": 8.297744540891429e-06, "loss": 0.14858862161636352, "step": 11110 }, { "epoch": 0.29291823412714746, "grad_norm": 1.5277810096740723, "learning_rate": 8.294536988345194e-06, "loss": 0.16369664669036865, "step": 11120 }, { "epoch": 0.29318164980531936, "grad_norm": 4.157609939575195, "learning_rate": 8.2913270378928e-06, "loss": 0.15604760646820068, "step": 11130 }, { "epoch": 0.29344506548349125, "grad_norm": 1.65449059009552, "learning_rate": 8.288114691870588e-06, "loss": 0.16389386653900145, "step": 11140 }, { "epoch": 0.29370848116166315, "grad_norm": 2.2482802867889404, "learning_rate": 8.284899952616642e-06, "loss": 0.1491938829421997, "step": 11150 }, { "epoch": 0.29397189683983505, "grad_norm": 2.0224082469940186, "learning_rate": 8.281682822470789e-06, "loss": 0.1697561264038086, "step": 11160 }, { "epoch": 0.29423531251800694, "grad_norm": 2.9462873935699463, "learning_rate": 8.278463303774593e-06, "loss": 0.15995868444442748, "step": 11170 }, { "epoch": 0.29449872819617884, "grad_norm": 1.6222071647644043, "learning_rate": 8.27524139887136e-06, "loss": 0.15895507335662842, "step": 11180 }, { "epoch": 0.29476214387435073, "grad_norm": 1.9216192960739136, "learning_rate": 8.272017110106131e-06, "loss": 0.15432652235031127, "step": 11190 }, { "epoch": 0.29502555955252263, "grad_norm": 2.1766152381896973, "learning_rate": 8.268790439825682e-06, "loss": 0.15857467651367188, "step": 11200 }, { "epoch": 0.29528897523069453, "grad_norm": 2.954742193222046, "learning_rate": 8.265561390378523e-06, "loss": 0.14197831153869628, "step": 11210 }, { "epoch": 0.2955523909088664, "grad_norm": 3.028432846069336, "learning_rate": 8.262329964114896e-06, "loss": 0.166753888130188, "step": 11220 }, { "epoch": 0.2958158065870383, "grad_norm": 6.183155536651611, "learning_rate": 8.259096163386773e-06, "loss": 0.16783249378204346, "step": 11230 }, { "epoch": 0.2960792222652102, "grad_norm": 2.2830865383148193, "learning_rate": 8.255859990547852e-06, "loss": 0.1465087652206421, "step": 11240 }, { "epoch": 0.2963426379433821, "grad_norm": 4.53877067565918, "learning_rate": 8.252621447953561e-06, "loss": 0.15948338508605958, "step": 11250 }, { "epoch": 0.296606053621554, "grad_norm": 2.0927956104278564, "learning_rate": 8.24938053796105e-06, "loss": 0.15400232076644899, "step": 11260 }, { "epoch": 0.2968694692997259, "grad_norm": 1.2028024196624756, "learning_rate": 8.246137262929193e-06, "loss": 0.15991986989974977, "step": 11270 }, { "epoch": 0.2971328849778978, "grad_norm": 2.9866247177124023, "learning_rate": 8.242891625218585e-06, "loss": 0.15311973094940184, "step": 11280 }, { "epoch": 0.2973963006560697, "grad_norm": 2.2202672958374023, "learning_rate": 8.239643627191542e-06, "loss": 0.16671154499053956, "step": 11290 }, { "epoch": 0.29765971633424154, "grad_norm": 4.878757953643799, "learning_rate": 8.236393271212098e-06, "loss": 0.1633157253265381, "step": 11300 }, { "epoch": 0.29792313201241344, "grad_norm": 2.5543951988220215, "learning_rate": 8.233140559646002e-06, "loss": 0.1655956506729126, "step": 11310 }, { "epoch": 0.29818654769058534, "grad_norm": 1.8410897254943848, "learning_rate": 8.229885494860717e-06, "loss": 0.1573964238166809, "step": 11320 }, { "epoch": 0.29844996336875723, "grad_norm": 2.892646074295044, "learning_rate": 8.226628079225418e-06, "loss": 0.16095168590545655, "step": 11330 }, { "epoch": 0.29871337904692913, "grad_norm": 4.569720268249512, "learning_rate": 8.223368315110995e-06, "loss": 0.17167081832885742, "step": 11340 }, { "epoch": 0.298976794725101, "grad_norm": 2.182654619216919, "learning_rate": 8.220106204890044e-06, "loss": 0.15885473489761354, "step": 11350 }, { "epoch": 0.2992402104032729, "grad_norm": 3.0104711055755615, "learning_rate": 8.21684175093687e-06, "loss": 0.16826395988464354, "step": 11360 }, { "epoch": 0.2995036260814448, "grad_norm": 2.302745819091797, "learning_rate": 8.213574955627487e-06, "loss": 0.15673271417617798, "step": 11370 }, { "epoch": 0.2997670417596167, "grad_norm": 3.080853223800659, "learning_rate": 8.2103058213396e-06, "loss": 0.15800166130065918, "step": 11380 }, { "epoch": 0.3000304574377886, "grad_norm": 1.2394447326660156, "learning_rate": 8.207034350452636e-06, "loss": 0.1628638982772827, "step": 11390 }, { "epoch": 0.3002938731159605, "grad_norm": 1.5785843133926392, "learning_rate": 8.203760545347706e-06, "loss": 0.1555301070213318, "step": 11400 }, { "epoch": 0.3005572887941324, "grad_norm": 2.3620972633361816, "learning_rate": 8.20048440840763e-06, "loss": 0.16689765453338623, "step": 11410 }, { "epoch": 0.3008207044723043, "grad_norm": 3.9141249656677246, "learning_rate": 8.197205942016922e-06, "loss": 0.16500532627105713, "step": 11420 }, { "epoch": 0.3010841201504762, "grad_norm": 3.3929176330566406, "learning_rate": 8.19392514856179e-06, "loss": 0.15392765998840333, "step": 11430 }, { "epoch": 0.3013475358286481, "grad_norm": 4.396742343902588, "learning_rate": 8.190642030430135e-06, "loss": 0.15740277767181396, "step": 11440 }, { "epoch": 0.30161095150682, "grad_norm": 2.724858522415161, "learning_rate": 8.187356590011557e-06, "loss": 0.1648500680923462, "step": 11450 }, { "epoch": 0.3018743671849919, "grad_norm": 4.367386817932129, "learning_rate": 8.184068829697336e-06, "loss": 0.15725140571594237, "step": 11460 }, { "epoch": 0.3021377828631638, "grad_norm": 1.956723928451538, "learning_rate": 8.18077875188045e-06, "loss": 0.15037686824798585, "step": 11470 }, { "epoch": 0.3024011985413357, "grad_norm": 2.8972530364990234, "learning_rate": 8.177486358955557e-06, "loss": 0.15435919761657715, "step": 11480 }, { "epoch": 0.3026646142195076, "grad_norm": 4.581657409667969, "learning_rate": 8.174191653319005e-06, "loss": 0.1675273895263672, "step": 11490 }, { "epoch": 0.3029280298976795, "grad_norm": 5.621101379394531, "learning_rate": 8.170894637368822e-06, "loss": 0.1489580273628235, "step": 11500 }, { "epoch": 0.30319144557585137, "grad_norm": 1.3820918798446655, "learning_rate": 8.167595313504716e-06, "loss": 0.1677138924598694, "step": 11510 }, { "epoch": 0.30345486125402327, "grad_norm": 1.4645776748657227, "learning_rate": 8.16429368412808e-06, "loss": 0.14780491590499878, "step": 11520 }, { "epoch": 0.30371827693219516, "grad_norm": 3.0391030311584473, "learning_rate": 8.16098975164198e-06, "loss": 0.1643594741821289, "step": 11530 }, { "epoch": 0.30398169261036706, "grad_norm": 3.2991085052490234, "learning_rate": 8.157683518451166e-06, "loss": 0.13920021057128906, "step": 11540 }, { "epoch": 0.30424510828853896, "grad_norm": 1.5487558841705322, "learning_rate": 8.154374986962048e-06, "loss": 0.15550589561462402, "step": 11550 }, { "epoch": 0.30450852396671085, "grad_norm": 1.4376201629638672, "learning_rate": 8.151064159582727e-06, "loss": 0.1510746121406555, "step": 11560 }, { "epoch": 0.30477193964488275, "grad_norm": 1.096184253692627, "learning_rate": 8.14775103872296e-06, "loss": 0.1587050437927246, "step": 11570 }, { "epoch": 0.30503535532305465, "grad_norm": 2.430419921875, "learning_rate": 8.144435626794183e-06, "loss": 0.15684106349945068, "step": 11580 }, { "epoch": 0.30529877100122654, "grad_norm": 3.18609619140625, "learning_rate": 8.141117926209491e-06, "loss": 0.15325263738632203, "step": 11590 }, { "epoch": 0.30556218667939844, "grad_norm": 1.5335938930511475, "learning_rate": 8.137797939383657e-06, "loss": 0.15951294898986818, "step": 11600 }, { "epoch": 0.30582560235757034, "grad_norm": 1.8142049312591553, "learning_rate": 8.134475668733106e-06, "loss": 0.15773750543594361, "step": 11610 }, { "epoch": 0.30608901803574223, "grad_norm": 4.7463579177856445, "learning_rate": 8.13115111667593e-06, "loss": 0.1595916748046875, "step": 11620 }, { "epoch": 0.30635243371391413, "grad_norm": 2.446218490600586, "learning_rate": 8.127824285631881e-06, "loss": 0.16883062124252318, "step": 11630 }, { "epoch": 0.306615849392086, "grad_norm": 2.217352867126465, "learning_rate": 8.124495178022371e-06, "loss": 0.14536213874816895, "step": 11640 }, { "epoch": 0.3068792650702579, "grad_norm": 3.1856772899627686, "learning_rate": 8.121163796270471e-06, "loss": 0.14009135961532593, "step": 11650 }, { "epoch": 0.3071426807484298, "grad_norm": 1.4182173013687134, "learning_rate": 8.1178301428009e-06, "loss": 0.17646697759628296, "step": 11660 }, { "epoch": 0.3074060964266017, "grad_norm": 3.2831907272338867, "learning_rate": 8.114494220040034e-06, "loss": 0.16723157167434693, "step": 11670 }, { "epoch": 0.3076695121047736, "grad_norm": 3.996643304824829, "learning_rate": 8.111156030415907e-06, "loss": 0.15154814720153809, "step": 11680 }, { "epoch": 0.30793292778294545, "grad_norm": 1.3829375505447388, "learning_rate": 8.107815576358194e-06, "loss": 0.15266649723052977, "step": 11690 }, { "epoch": 0.30819634346111735, "grad_norm": 2.5565595626831055, "learning_rate": 8.10447286029822e-06, "loss": 0.15872280597686766, "step": 11700 }, { "epoch": 0.30845975913928925, "grad_norm": 6.428664684295654, "learning_rate": 8.101127884668962e-06, "loss": 0.15473661422729493, "step": 11710 }, { "epoch": 0.30872317481746114, "grad_norm": 3.808699131011963, "learning_rate": 8.097780651905034e-06, "loss": 0.15926940441131593, "step": 11720 }, { "epoch": 0.30898659049563304, "grad_norm": 0.9584486484527588, "learning_rate": 8.0944311644427e-06, "loss": 0.1671532392501831, "step": 11730 }, { "epoch": 0.30925000617380494, "grad_norm": 3.093376636505127, "learning_rate": 8.091079424719859e-06, "loss": 0.1575162172317505, "step": 11740 }, { "epoch": 0.30951342185197683, "grad_norm": 2.4473655223846436, "learning_rate": 8.08772543517605e-06, "loss": 0.17319004535675048, "step": 11750 }, { "epoch": 0.30977683753014873, "grad_norm": 6.977925777435303, "learning_rate": 8.084369198252456e-06, "loss": 0.15737323760986327, "step": 11760 }, { "epoch": 0.3100402532083206, "grad_norm": 2.4821410179138184, "learning_rate": 8.081010716391886e-06, "loss": 0.15206152200698853, "step": 11770 }, { "epoch": 0.3103036688864925, "grad_norm": 1.6690306663513184, "learning_rate": 8.077649992038792e-06, "loss": 0.15770766735076905, "step": 11780 }, { "epoch": 0.3105670845646644, "grad_norm": 2.5174965858459473, "learning_rate": 8.074287027639254e-06, "loss": 0.17453882694244385, "step": 11790 }, { "epoch": 0.3108305002428363, "grad_norm": 5.018013954162598, "learning_rate": 8.070921825640978e-06, "loss": 0.1574064612388611, "step": 11800 }, { "epoch": 0.3110939159210082, "grad_norm": 1.5902824401855469, "learning_rate": 8.067554388493308e-06, "loss": 0.15057902336120604, "step": 11810 }, { "epoch": 0.3113573315991801, "grad_norm": 2.147907257080078, "learning_rate": 8.06418471864721e-06, "loss": 0.1579190492630005, "step": 11820 }, { "epoch": 0.311620747277352, "grad_norm": 1.9824093580245972, "learning_rate": 8.060812818555273e-06, "loss": 0.16128698587417603, "step": 11830 }, { "epoch": 0.3118841629555239, "grad_norm": 1.7757132053375244, "learning_rate": 8.057438690671711e-06, "loss": 0.1546817183494568, "step": 11840 }, { "epoch": 0.3121475786336958, "grad_norm": 2.6712281703948975, "learning_rate": 8.05406233745236e-06, "loss": 0.16255300045013427, "step": 11850 }, { "epoch": 0.3124109943118677, "grad_norm": 2.1078941822052, "learning_rate": 8.050683761354678e-06, "loss": 0.15356498956680298, "step": 11860 }, { "epoch": 0.3126744099900396, "grad_norm": 2.760641574859619, "learning_rate": 8.047302964837738e-06, "loss": 0.15280603170394896, "step": 11870 }, { "epoch": 0.3129378256682115, "grad_norm": 3.209705114364624, "learning_rate": 8.043919950362225e-06, "loss": 0.16047120094299316, "step": 11880 }, { "epoch": 0.3132012413463834, "grad_norm": 4.930066108703613, "learning_rate": 8.040534720390448e-06, "loss": 0.15874006748199462, "step": 11890 }, { "epoch": 0.3134646570245553, "grad_norm": 2.9616668224334717, "learning_rate": 8.03714727738632e-06, "loss": 0.1752923846244812, "step": 11900 }, { "epoch": 0.3137280727027272, "grad_norm": 1.6341551542282104, "learning_rate": 8.033757623815374e-06, "loss": 0.1534088969230652, "step": 11910 }, { "epoch": 0.3139914883808991, "grad_norm": 1.7116886377334595, "learning_rate": 8.03036576214474e-06, "loss": 0.14870784282684327, "step": 11920 }, { "epoch": 0.31425490405907097, "grad_norm": 2.1760261058807373, "learning_rate": 8.026971694843162e-06, "loss": 0.15859845876693726, "step": 11930 }, { "epoch": 0.31451831973724287, "grad_norm": 1.2668797969818115, "learning_rate": 8.02357542438099e-06, "loss": 0.15510209798812866, "step": 11940 }, { "epoch": 0.31478173541541477, "grad_norm": 2.770878553390503, "learning_rate": 8.020176953230177e-06, "loss": 0.14972324371337892, "step": 11950 }, { "epoch": 0.31504515109358666, "grad_norm": 1.3920047283172607, "learning_rate": 8.016776283864277e-06, "loss": 0.16389498710632325, "step": 11960 }, { "epoch": 0.31530856677175856, "grad_norm": 1.0892244577407837, "learning_rate": 8.01337341875844e-06, "loss": 0.1577028751373291, "step": 11970 }, { "epoch": 0.31557198244993045, "grad_norm": 1.123152256011963, "learning_rate": 8.009968360389423e-06, "loss": 0.1740635871887207, "step": 11980 }, { "epoch": 0.31583539812810235, "grad_norm": 2.7222096920013428, "learning_rate": 8.00656111123557e-06, "loss": 0.1605099081993103, "step": 11990 }, { "epoch": 0.31609881380627425, "grad_norm": 2.4316375255584717, "learning_rate": 8.003151673776828e-06, "loss": 0.14770286083221434, "step": 12000 }, { "epoch": 0.31636222948444614, "grad_norm": 2.0689496994018555, "learning_rate": 7.999740050494729e-06, "loss": 0.1613134980201721, "step": 12010 }, { "epoch": 0.31662564516261804, "grad_norm": 1.5754585266113281, "learning_rate": 7.996326243872403e-06, "loss": 0.1555086135864258, "step": 12020 }, { "epoch": 0.31688906084078994, "grad_norm": 2.4428482055664062, "learning_rate": 7.992910256394564e-06, "loss": 0.15659799575805664, "step": 12030 }, { "epoch": 0.31715247651896183, "grad_norm": 4.312586307525635, "learning_rate": 7.989492090547517e-06, "loss": 0.16939811706542968, "step": 12040 }, { "epoch": 0.31741589219713373, "grad_norm": 3.395068407058716, "learning_rate": 7.986071748819147e-06, "loss": 0.16682946681976318, "step": 12050 }, { "epoch": 0.3176793078753056, "grad_norm": 1.507655143737793, "learning_rate": 7.982649233698927e-06, "loss": 0.1578940987586975, "step": 12060 }, { "epoch": 0.3179427235534775, "grad_norm": 2.6470837593078613, "learning_rate": 7.979224547677914e-06, "loss": 0.14111907482147218, "step": 12070 }, { "epoch": 0.3182061392316494, "grad_norm": 3.7012877464294434, "learning_rate": 7.97579769324874e-06, "loss": 0.17112149000167848, "step": 12080 }, { "epoch": 0.31846955490982126, "grad_norm": 2.1660044193267822, "learning_rate": 7.972368672905619e-06, "loss": 0.16644034385681153, "step": 12090 }, { "epoch": 0.31873297058799316, "grad_norm": 1.1130017042160034, "learning_rate": 7.968937489144338e-06, "loss": 0.15489256381988525, "step": 12100 }, { "epoch": 0.31899638626616506, "grad_norm": 2.0093142986297607, "learning_rate": 7.965504144462261e-06, "loss": 0.1572236180305481, "step": 12110 }, { "epoch": 0.31925980194433695, "grad_norm": 2.203456163406372, "learning_rate": 7.962068641358327e-06, "loss": 0.16010260581970215, "step": 12120 }, { "epoch": 0.31952321762250885, "grad_norm": 3.9908926486968994, "learning_rate": 7.95863098233304e-06, "loss": 0.1603947877883911, "step": 12130 }, { "epoch": 0.31978663330068074, "grad_norm": 3.0812323093414307, "learning_rate": 7.955191169888477e-06, "loss": 0.15577644109725952, "step": 12140 }, { "epoch": 0.32005004897885264, "grad_norm": 3.0179929733276367, "learning_rate": 7.951749206528284e-06, "loss": 0.15829795598983765, "step": 12150 }, { "epoch": 0.32031346465702454, "grad_norm": 1.9581305980682373, "learning_rate": 7.948305094757667e-06, "loss": 0.15538308620452881, "step": 12160 }, { "epoch": 0.32057688033519643, "grad_norm": 2.732630968093872, "learning_rate": 7.944858837083404e-06, "loss": 0.14984959363937378, "step": 12170 }, { "epoch": 0.32084029601336833, "grad_norm": 3.709571123123169, "learning_rate": 7.941410436013827e-06, "loss": 0.1546069860458374, "step": 12180 }, { "epoch": 0.32110371169154023, "grad_norm": 1.7401002645492554, "learning_rate": 7.937959894058831e-06, "loss": 0.14299635887145995, "step": 12190 }, { "epoch": 0.3213671273697121, "grad_norm": 3.9423162937164307, "learning_rate": 7.93450721372987e-06, "loss": 0.15708280801773072, "step": 12200 }, { "epoch": 0.321630543047884, "grad_norm": 3.6224658489227295, "learning_rate": 7.931052397539952e-06, "loss": 0.16248010396957396, "step": 12210 }, { "epoch": 0.3218939587260559, "grad_norm": 5.869894027709961, "learning_rate": 7.927595448003644e-06, "loss": 0.17110226154327393, "step": 12220 }, { "epoch": 0.3221573744042278, "grad_norm": 2.694669008255005, "learning_rate": 7.92413636763706e-06, "loss": 0.15015414953231812, "step": 12230 }, { "epoch": 0.3224207900823997, "grad_norm": 2.416583776473999, "learning_rate": 7.920675158957872e-06, "loss": 0.16210170984268188, "step": 12240 }, { "epoch": 0.3226842057605716, "grad_norm": 2.155585765838623, "learning_rate": 7.917211824485294e-06, "loss": 0.15400009155273436, "step": 12250 }, { "epoch": 0.3229476214387435, "grad_norm": 4.035641193389893, "learning_rate": 7.913746366740092e-06, "loss": 0.1533929228782654, "step": 12260 }, { "epoch": 0.3232110371169154, "grad_norm": 1.1619298458099365, "learning_rate": 7.910278788244572e-06, "loss": 0.14385954141616822, "step": 12270 }, { "epoch": 0.3234744527950873, "grad_norm": 3.5638487339019775, "learning_rate": 7.90680909152259e-06, "loss": 0.15970590114593505, "step": 12280 }, { "epoch": 0.3237378684732592, "grad_norm": 1.5640407800674438, "learning_rate": 7.903337279099544e-06, "loss": 0.15372685194015503, "step": 12290 }, { "epoch": 0.3240012841514311, "grad_norm": 2.6447665691375732, "learning_rate": 7.899863353502364e-06, "loss": 0.17491139173507692, "step": 12300 }, { "epoch": 0.324264699829603, "grad_norm": 2.531977653503418, "learning_rate": 7.896387317259523e-06, "loss": 0.15996792316436767, "step": 12310 }, { "epoch": 0.3245281155077749, "grad_norm": 3.3120946884155273, "learning_rate": 7.892909172901034e-06, "loss": 0.16039881706237794, "step": 12320 }, { "epoch": 0.3247915311859468, "grad_norm": 1.8746389150619507, "learning_rate": 7.889428922958437e-06, "loss": 0.15448683500289917, "step": 12330 }, { "epoch": 0.3250549468641187, "grad_norm": 4.157230854034424, "learning_rate": 7.88594656996481e-06, "loss": 0.14700837135314943, "step": 12340 }, { "epoch": 0.3253183625422906, "grad_norm": 1.1061670780181885, "learning_rate": 7.882462116454759e-06, "loss": 0.1730874538421631, "step": 12350 }, { "epoch": 0.32558177822046247, "grad_norm": 2.434802770614624, "learning_rate": 7.878975564964418e-06, "loss": 0.16496973037719725, "step": 12360 }, { "epoch": 0.32584519389863437, "grad_norm": 2.409303665161133, "learning_rate": 7.875486918031453e-06, "loss": 0.16228758096694945, "step": 12370 }, { "epoch": 0.32610860957680626, "grad_norm": 2.879586935043335, "learning_rate": 7.87199617819505e-06, "loss": 0.16254823207855223, "step": 12380 }, { "epoch": 0.32637202525497816, "grad_norm": 1.8433382511138916, "learning_rate": 7.86850334799592e-06, "loss": 0.15596156120300292, "step": 12390 }, { "epoch": 0.32663544093315006, "grad_norm": 2.0082647800445557, "learning_rate": 7.865008429976294e-06, "loss": 0.16093608140945434, "step": 12400 }, { "epoch": 0.32689885661132195, "grad_norm": 1.6446747779846191, "learning_rate": 7.861511426679926e-06, "loss": 0.1478700280189514, "step": 12410 }, { "epoch": 0.32716227228949385, "grad_norm": 1.5778939723968506, "learning_rate": 7.858012340652089e-06, "loss": 0.160161030292511, "step": 12420 }, { "epoch": 0.32742568796766575, "grad_norm": 2.284885883331299, "learning_rate": 7.854511174439562e-06, "loss": 0.168194580078125, "step": 12430 }, { "epoch": 0.32768910364583764, "grad_norm": 1.7785662412643433, "learning_rate": 7.851007930590648e-06, "loss": 0.16532100439071656, "step": 12440 }, { "epoch": 0.32795251932400954, "grad_norm": 2.1990461349487305, "learning_rate": 7.84750261165516e-06, "loss": 0.14106061458587646, "step": 12450 }, { "epoch": 0.32821593500218144, "grad_norm": 2.6171913146972656, "learning_rate": 7.84399522018442e-06, "loss": 0.14835790395736695, "step": 12460 }, { "epoch": 0.32847935068035333, "grad_norm": 1.621957778930664, "learning_rate": 7.840485758731258e-06, "loss": 0.16930917501449586, "step": 12470 }, { "epoch": 0.32874276635852523, "grad_norm": 2.238229751586914, "learning_rate": 7.836974229850012e-06, "loss": 0.151144278049469, "step": 12480 }, { "epoch": 0.32900618203669707, "grad_norm": 3.2286601066589355, "learning_rate": 7.833460636096523e-06, "loss": 0.14756326675415038, "step": 12490 }, { "epoch": 0.32926959771486897, "grad_norm": 3.4476726055145264, "learning_rate": 7.82994498002814e-06, "loss": 0.15749547481536866, "step": 12500 }, { "epoch": 0.32953301339304086, "grad_norm": 1.895652413368225, "learning_rate": 7.826427264203704e-06, "loss": 0.15500099658966066, "step": 12510 }, { "epoch": 0.32979642907121276, "grad_norm": 3.679414749145508, "learning_rate": 7.822907491183562e-06, "loss": 0.15255534648895264, "step": 12520 }, { "epoch": 0.33005984474938466, "grad_norm": 2.293696403503418, "learning_rate": 7.819385663529558e-06, "loss": 0.15509982109069825, "step": 12530 }, { "epoch": 0.33032326042755655, "grad_norm": 1.8949875831604004, "learning_rate": 7.81586178380503e-06, "loss": 0.14099475145339965, "step": 12540 }, { "epoch": 0.33058667610572845, "grad_norm": 5.1779985427856445, "learning_rate": 7.812335854574809e-06, "loss": 0.14552035331726074, "step": 12550 }, { "epoch": 0.33085009178390035, "grad_norm": 3.278249502182007, "learning_rate": 7.80880787840522e-06, "loss": 0.1602412223815918, "step": 12560 }, { "epoch": 0.33111350746207224, "grad_norm": 2.088294506072998, "learning_rate": 7.805277857864073e-06, "loss": 0.15387682914733886, "step": 12570 }, { "epoch": 0.33137692314024414, "grad_norm": 2.807631254196167, "learning_rate": 7.80174579552067e-06, "loss": 0.15725715160369874, "step": 12580 }, { "epoch": 0.33164033881841604, "grad_norm": 1.2922018766403198, "learning_rate": 7.7982116939458e-06, "loss": 0.15916819572448732, "step": 12590 }, { "epoch": 0.33190375449658793, "grad_norm": 2.43411922454834, "learning_rate": 7.794675555711737e-06, "loss": 0.16070555448532103, "step": 12600 }, { "epoch": 0.33216717017475983, "grad_norm": 1.9366732835769653, "learning_rate": 7.791137383392228e-06, "loss": 0.17239704132080078, "step": 12610 }, { "epoch": 0.3324305858529317, "grad_norm": 1.6403884887695312, "learning_rate": 7.787597179562514e-06, "loss": 0.1545855760574341, "step": 12620 }, { "epoch": 0.3326940015311036, "grad_norm": 2.3334953784942627, "learning_rate": 7.784054946799305e-06, "loss": 0.14304754734039307, "step": 12630 }, { "epoch": 0.3329574172092755, "grad_norm": 1.4525283575057983, "learning_rate": 7.780510687680792e-06, "loss": 0.15622247457504274, "step": 12640 }, { "epoch": 0.3332208328874474, "grad_norm": 1.4847906827926636, "learning_rate": 7.776964404786637e-06, "loss": 0.14625895023345947, "step": 12650 }, { "epoch": 0.3334842485656193, "grad_norm": 4.325026512145996, "learning_rate": 7.773416100697982e-06, "loss": 0.15918948650360107, "step": 12660 }, { "epoch": 0.3337476642437912, "grad_norm": 1.3270394802093506, "learning_rate": 7.769865777997432e-06, "loss": 0.16142865419387817, "step": 12670 }, { "epoch": 0.3340110799219631, "grad_norm": 2.3473784923553467, "learning_rate": 7.766313439269068e-06, "loss": 0.1565181016921997, "step": 12680 }, { "epoch": 0.334274495600135, "grad_norm": 2.4844396114349365, "learning_rate": 7.762759087098434e-06, "loss": 0.16409614086151122, "step": 12690 }, { "epoch": 0.3345379112783069, "grad_norm": 1.219543218612671, "learning_rate": 7.75920272407254e-06, "loss": 0.1477281332015991, "step": 12700 }, { "epoch": 0.3348013269564788, "grad_norm": 2.2829813957214355, "learning_rate": 7.75564435277986e-06, "loss": 0.16347991228103637, "step": 12710 }, { "epoch": 0.3350647426346507, "grad_norm": 1.1703470945358276, "learning_rate": 7.752083975810334e-06, "loss": 0.15259404182434083, "step": 12720 }, { "epoch": 0.3353281583128226, "grad_norm": 2.4653024673461914, "learning_rate": 7.748521595755354e-06, "loss": 0.14611705541610717, "step": 12730 }, { "epoch": 0.3355915739909945, "grad_norm": 2.6397640705108643, "learning_rate": 7.744957215207775e-06, "loss": 0.16313130855560304, "step": 12740 }, { "epoch": 0.3358549896691664, "grad_norm": 2.447828531265259, "learning_rate": 7.741390836761908e-06, "loss": 0.15507395267486573, "step": 12750 }, { "epoch": 0.3361184053473383, "grad_norm": 2.87931227684021, "learning_rate": 7.737822463013517e-06, "loss": 0.15698788166046143, "step": 12760 }, { "epoch": 0.3363818210255102, "grad_norm": 1.1227527856826782, "learning_rate": 7.734252096559818e-06, "loss": 0.1619136095046997, "step": 12770 }, { "epoch": 0.33664523670368207, "grad_norm": 2.4995994567871094, "learning_rate": 7.730679739999477e-06, "loss": 0.14922993183135985, "step": 12780 }, { "epoch": 0.33690865238185397, "grad_norm": 2.2936222553253174, "learning_rate": 7.727105395932606e-06, "loss": 0.13739773035049438, "step": 12790 }, { "epoch": 0.33717206806002586, "grad_norm": 2.4305310249328613, "learning_rate": 7.723529066960776e-06, "loss": 0.15217764377593995, "step": 12800 }, { "epoch": 0.33743548373819776, "grad_norm": 2.0313262939453125, "learning_rate": 7.719950755686985e-06, "loss": 0.15356451272964478, "step": 12810 }, { "epoch": 0.33769889941636966, "grad_norm": 1.766194462776184, "learning_rate": 7.716370464715686e-06, "loss": 0.1456224203109741, "step": 12820 }, { "epoch": 0.33796231509454155, "grad_norm": 2.048628568649292, "learning_rate": 7.712788196652769e-06, "loss": 0.1502050280570984, "step": 12830 }, { "epoch": 0.33822573077271345, "grad_norm": 1.4468728303909302, "learning_rate": 7.709203954105563e-06, "loss": 0.1635649561882019, "step": 12840 }, { "epoch": 0.33848914645088535, "grad_norm": 1.4037877321243286, "learning_rate": 7.705617739682832e-06, "loss": 0.1517292380332947, "step": 12850 }, { "epoch": 0.33875256212905724, "grad_norm": 2.1157853603363037, "learning_rate": 7.702029555994782e-06, "loss": 0.161482834815979, "step": 12860 }, { "epoch": 0.33901597780722914, "grad_norm": 2.7231223583221436, "learning_rate": 7.698439405653044e-06, "loss": 0.15729483366012573, "step": 12870 }, { "epoch": 0.339279393485401, "grad_norm": 2.802156448364258, "learning_rate": 7.694847291270688e-06, "loss": 0.146986722946167, "step": 12880 }, { "epoch": 0.3395428091635729, "grad_norm": 3.5794875621795654, "learning_rate": 7.691253215462207e-06, "loss": 0.17475041151046752, "step": 12890 }, { "epoch": 0.3398062248417448, "grad_norm": 1.0813332796096802, "learning_rate": 7.687657180843526e-06, "loss": 0.1625834345817566, "step": 12900 }, { "epoch": 0.34006964051991667, "grad_norm": 3.97351336479187, "learning_rate": 7.684059190031992e-06, "loss": 0.16296870708465577, "step": 12910 }, { "epoch": 0.34033305619808857, "grad_norm": 1.660177230834961, "learning_rate": 7.68045924564638e-06, "loss": 0.15692007541656494, "step": 12920 }, { "epoch": 0.34059647187626046, "grad_norm": 2.854034423828125, "learning_rate": 7.676857350306884e-06, "loss": 0.16269850730895996, "step": 12930 }, { "epoch": 0.34085988755443236, "grad_norm": 2.6028525829315186, "learning_rate": 7.673253506635118e-06, "loss": 0.16490187644958496, "step": 12940 }, { "epoch": 0.34112330323260426, "grad_norm": 3.207892656326294, "learning_rate": 7.669647717254119e-06, "loss": 0.1586683511734009, "step": 12950 }, { "epoch": 0.34138671891077615, "grad_norm": 2.239905595779419, "learning_rate": 7.66603998478833e-06, "loss": 0.15224087238311768, "step": 12960 }, { "epoch": 0.34165013458894805, "grad_norm": 2.796421527862549, "learning_rate": 7.662430311863618e-06, "loss": 0.1352986216545105, "step": 12970 }, { "epoch": 0.34191355026711995, "grad_norm": 1.3237196207046509, "learning_rate": 7.658818701107257e-06, "loss": 0.15695552825927733, "step": 12980 }, { "epoch": 0.34217696594529184, "grad_norm": 1.8287980556488037, "learning_rate": 7.655205155147938e-06, "loss": 0.16111421585083008, "step": 12990 }, { "epoch": 0.34244038162346374, "grad_norm": 2.462913990020752, "learning_rate": 7.651589676615747e-06, "loss": 0.1717458486557007, "step": 13000 }, { "epoch": 0.34270379730163564, "grad_norm": 2.8652329444885254, "learning_rate": 7.647972268142195e-06, "loss": 0.15346906185150147, "step": 13010 }, { "epoch": 0.34296721297980753, "grad_norm": 3.492868185043335, "learning_rate": 7.644352932360178e-06, "loss": 0.15493910312652587, "step": 13020 }, { "epoch": 0.34323062865797943, "grad_norm": 1.8509849309921265, "learning_rate": 7.640731671904014e-06, "loss": 0.13133736848831176, "step": 13030 }, { "epoch": 0.3434940443361513, "grad_norm": 1.9040125608444214, "learning_rate": 7.637108489409406e-06, "loss": 0.14387544393539428, "step": 13040 }, { "epoch": 0.3437574600143232, "grad_norm": 3.6936957836151123, "learning_rate": 7.633483387513464e-06, "loss": 0.15886847972869872, "step": 13050 }, { "epoch": 0.3440208756924951, "grad_norm": 4.411664962768555, "learning_rate": 7.629856368854696e-06, "loss": 0.15596847534179686, "step": 13060 }, { "epoch": 0.344284291370667, "grad_norm": 2.053666114807129, "learning_rate": 7.6262274360730015e-06, "loss": 0.1505640745162964, "step": 13070 }, { "epoch": 0.3445477070488389, "grad_norm": 2.903235912322998, "learning_rate": 7.622596591809673e-06, "loss": 0.16509186029434203, "step": 13080 }, { "epoch": 0.3448111227270108, "grad_norm": 4.877506256103516, "learning_rate": 7.6189638387073984e-06, "loss": 0.16761014461517335, "step": 13090 }, { "epoch": 0.3450745384051827, "grad_norm": 2.5473551750183105, "learning_rate": 7.615329179410247e-06, "loss": 0.14867552518844604, "step": 13100 }, { "epoch": 0.3453379540833546, "grad_norm": 5.547333717346191, "learning_rate": 7.611692616563686e-06, "loss": 0.16853631734848024, "step": 13110 }, { "epoch": 0.3456013697615265, "grad_norm": 2.529336452484131, "learning_rate": 7.608054152814562e-06, "loss": 0.14345754384994508, "step": 13120 }, { "epoch": 0.3458647854396984, "grad_norm": 1.3136566877365112, "learning_rate": 7.6044137908111025e-06, "loss": 0.15178314447402955, "step": 13130 }, { "epoch": 0.3461282011178703, "grad_norm": 2.72537899017334, "learning_rate": 7.600771533202922e-06, "loss": 0.15415207147598267, "step": 13140 }, { "epoch": 0.3463916167960422, "grad_norm": 2.0062742233276367, "learning_rate": 7.597127382641013e-06, "loss": 0.1582885980606079, "step": 13150 }, { "epoch": 0.3466550324742141, "grad_norm": 1.7060130834579468, "learning_rate": 7.593481341777747e-06, "loss": 0.17073653936386107, "step": 13160 }, { "epoch": 0.346918448152386, "grad_norm": 1.1254653930664062, "learning_rate": 7.589833413266865e-06, "loss": 0.1476193070411682, "step": 13170 }, { "epoch": 0.3471818638305579, "grad_norm": 4.645252227783203, "learning_rate": 7.586183599763492e-06, "loss": 0.16144931316375732, "step": 13180 }, { "epoch": 0.3474452795087298, "grad_norm": 2.0150206089019775, "learning_rate": 7.582531903924117e-06, "loss": 0.17595876455307008, "step": 13190 }, { "epoch": 0.3477086951869017, "grad_norm": 1.835935115814209, "learning_rate": 7.5788783284066035e-06, "loss": 0.1533725380897522, "step": 13200 }, { "epoch": 0.34797211086507357, "grad_norm": 3.904733896255493, "learning_rate": 7.575222875870178e-06, "loss": 0.14364899396896363, "step": 13210 }, { "epoch": 0.34823552654324547, "grad_norm": 3.2019550800323486, "learning_rate": 7.571565548975441e-06, "loss": 0.16467952728271484, "step": 13220 }, { "epoch": 0.34849894222141736, "grad_norm": 1.7494443655014038, "learning_rate": 7.567906350384351e-06, "loss": 0.14979363679885865, "step": 13230 }, { "epoch": 0.34876235789958926, "grad_norm": 2.349619150161743, "learning_rate": 7.56424528276023e-06, "loss": 0.14600720405578613, "step": 13240 }, { "epoch": 0.34902577357776116, "grad_norm": 3.052905797958374, "learning_rate": 7.56058234876776e-06, "loss": 0.16964478492736818, "step": 13250 }, { "epoch": 0.34928918925593305, "grad_norm": 1.2032142877578735, "learning_rate": 7.556917551072984e-06, "loss": 0.161082661151886, "step": 13260 }, { "epoch": 0.34955260493410495, "grad_norm": 4.358856201171875, "learning_rate": 7.5532508923433e-06, "loss": 0.15455496311187744, "step": 13270 }, { "epoch": 0.3498160206122768, "grad_norm": 1.1356419324874878, "learning_rate": 7.54958237524746e-06, "loss": 0.15137808322906493, "step": 13280 }, { "epoch": 0.3500794362904487, "grad_norm": 2.17096209526062, "learning_rate": 7.545912002455568e-06, "loss": 0.14099076986312867, "step": 13290 }, { "epoch": 0.3503428519686206, "grad_norm": 2.3319942951202393, "learning_rate": 7.542239776639082e-06, "loss": 0.15532206296920775, "step": 13300 }, { "epoch": 0.3506062676467925, "grad_norm": 5.108238697052002, "learning_rate": 7.538565700470804e-06, "loss": 0.15812911987304687, "step": 13310 }, { "epoch": 0.3508696833249644, "grad_norm": 3.4226088523864746, "learning_rate": 7.534889776624887e-06, "loss": 0.1554775595664978, "step": 13320 }, { "epoch": 0.3511330990031363, "grad_norm": 1.5496662855148315, "learning_rate": 7.531212007776827e-06, "loss": 0.1613817811012268, "step": 13330 }, { "epoch": 0.35139651468130817, "grad_norm": 4.262597560882568, "learning_rate": 7.52753239660346e-06, "loss": 0.15572547912597656, "step": 13340 }, { "epoch": 0.35165993035948007, "grad_norm": 2.212737560272217, "learning_rate": 7.523850945782969e-06, "loss": 0.16245946884155274, "step": 13350 }, { "epoch": 0.35192334603765196, "grad_norm": 1.4702465534210205, "learning_rate": 7.52016765799487e-06, "loss": 0.1602800726890564, "step": 13360 }, { "epoch": 0.35218676171582386, "grad_norm": 2.4720523357391357, "learning_rate": 7.5164825359200224e-06, "loss": 0.16089627742767335, "step": 13370 }, { "epoch": 0.35245017739399576, "grad_norm": 1.8819516897201538, "learning_rate": 7.512795582240613e-06, "loss": 0.1582469940185547, "step": 13380 }, { "epoch": 0.35271359307216765, "grad_norm": 4.330505847930908, "learning_rate": 7.509106799640169e-06, "loss": 0.15643465518951416, "step": 13390 }, { "epoch": 0.35297700875033955, "grad_norm": 1.7429240942001343, "learning_rate": 7.505416190803543e-06, "loss": 0.15575532913208007, "step": 13400 }, { "epoch": 0.35324042442851145, "grad_norm": 2.0692179203033447, "learning_rate": 7.50172375841692e-06, "loss": 0.14814934730529786, "step": 13410 }, { "epoch": 0.35350384010668334, "grad_norm": 2.884669780731201, "learning_rate": 7.498029505167813e-06, "loss": 0.12885565757751466, "step": 13420 }, { "epoch": 0.35376725578485524, "grad_norm": 3.9249584674835205, "learning_rate": 7.494333433745059e-06, "loss": 0.15530478954315186, "step": 13430 }, { "epoch": 0.35403067146302714, "grad_norm": 5.93233585357666, "learning_rate": 7.490635546838816e-06, "loss": 0.14896724224090577, "step": 13440 }, { "epoch": 0.35429408714119903, "grad_norm": 1.2789051532745361, "learning_rate": 7.486935847140566e-06, "loss": 0.16096010208129882, "step": 13450 }, { "epoch": 0.35455750281937093, "grad_norm": 1.7106578350067139, "learning_rate": 7.483234337343113e-06, "loss": 0.14813342094421386, "step": 13460 }, { "epoch": 0.3548209184975428, "grad_norm": 3.5352299213409424, "learning_rate": 7.4795310201405715e-06, "loss": 0.15001928806304932, "step": 13470 }, { "epoch": 0.3550843341757147, "grad_norm": 2.7511401176452637, "learning_rate": 7.475825898228379e-06, "loss": 0.154707670211792, "step": 13480 }, { "epoch": 0.3553477498538866, "grad_norm": 2.7203011512756348, "learning_rate": 7.472118974303281e-06, "loss": 0.15527286529541015, "step": 13490 }, { "epoch": 0.3556111655320585, "grad_norm": 1.9486416578292847, "learning_rate": 7.468410251063338e-06, "loss": 0.14719027280807495, "step": 13500 }, { "epoch": 0.3558745812102304, "grad_norm": 3.595423698425293, "learning_rate": 7.464699731207916e-06, "loss": 0.16172373294830322, "step": 13510 }, { "epoch": 0.3561379968884023, "grad_norm": 4.484795093536377, "learning_rate": 7.460987417437695e-06, "loss": 0.1497706651687622, "step": 13520 }, { "epoch": 0.3564014125665742, "grad_norm": 1.6296826601028442, "learning_rate": 7.457273312454655e-06, "loss": 0.14718077182769776, "step": 13530 }, { "epoch": 0.3566648282447461, "grad_norm": 2.2242093086242676, "learning_rate": 7.453557418962084e-06, "loss": 0.1552443504333496, "step": 13540 }, { "epoch": 0.356928243922918, "grad_norm": 3.5229523181915283, "learning_rate": 7.4498397396645665e-06, "loss": 0.16148858070373534, "step": 13550 }, { "epoch": 0.3571916596010899, "grad_norm": 1.8101776838302612, "learning_rate": 7.4461202772679906e-06, "loss": 0.1641092300415039, "step": 13560 }, { "epoch": 0.3574550752792618, "grad_norm": 1.4399155378341675, "learning_rate": 7.442399034479542e-06, "loss": 0.15762633085250854, "step": 13570 }, { "epoch": 0.3577184909574337, "grad_norm": 1.7754364013671875, "learning_rate": 7.438676014007704e-06, "loss": 0.1498098611831665, "step": 13580 }, { "epoch": 0.3579819066356056, "grad_norm": 2.9469943046569824, "learning_rate": 7.434951218562249e-06, "loss": 0.16071555614471436, "step": 13590 }, { "epoch": 0.3582453223137775, "grad_norm": 2.4698426723480225, "learning_rate": 7.431224650854245e-06, "loss": 0.15564444065093994, "step": 13600 }, { "epoch": 0.3585087379919494, "grad_norm": 1.724709153175354, "learning_rate": 7.427496313596048e-06, "loss": 0.14849807024002076, "step": 13610 }, { "epoch": 0.3587721536701213, "grad_norm": 1.449575662612915, "learning_rate": 7.423766209501301e-06, "loss": 0.16036772727966309, "step": 13620 }, { "epoch": 0.35903556934829317, "grad_norm": 2.244264602661133, "learning_rate": 7.4200343412849386e-06, "loss": 0.15116536617279053, "step": 13630 }, { "epoch": 0.35929898502646507, "grad_norm": 2.139819622039795, "learning_rate": 7.416300711663173e-06, "loss": 0.1580702543258667, "step": 13640 }, { "epoch": 0.35956240070463696, "grad_norm": 2.042898178100586, "learning_rate": 7.412565323353501e-06, "loss": 0.14647188186645507, "step": 13650 }, { "epoch": 0.35982581638280886, "grad_norm": 1.2595385313034058, "learning_rate": 7.408828179074699e-06, "loss": 0.14937809705734253, "step": 13660 }, { "epoch": 0.3600892320609807, "grad_norm": 6.403219699859619, "learning_rate": 7.405089281546824e-06, "loss": 0.15882670879364014, "step": 13670 }, { "epoch": 0.3603526477391526, "grad_norm": 1.090378999710083, "learning_rate": 7.401348633491206e-06, "loss": 0.15299998521804808, "step": 13680 }, { "epoch": 0.3606160634173245, "grad_norm": 3.7082161903381348, "learning_rate": 7.397606237630448e-06, "loss": 0.14566655158996583, "step": 13690 }, { "epoch": 0.3608794790954964, "grad_norm": 3.5322868824005127, "learning_rate": 7.393862096688429e-06, "loss": 0.15789875984191895, "step": 13700 }, { "epoch": 0.3611428947736683, "grad_norm": 2.7086243629455566, "learning_rate": 7.390116213390296e-06, "loss": 0.1295977473258972, "step": 13710 }, { "epoch": 0.3614063104518402, "grad_norm": 2.0715794563293457, "learning_rate": 7.386368590462465e-06, "loss": 0.15986883640289307, "step": 13720 }, { "epoch": 0.3616697261300121, "grad_norm": 3.02510404586792, "learning_rate": 7.3826192306326175e-06, "loss": 0.16753077507019043, "step": 13730 }, { "epoch": 0.361933141808184, "grad_norm": 1.509665608406067, "learning_rate": 7.378868136629698e-06, "loss": 0.16070101261138917, "step": 13740 }, { "epoch": 0.3621965574863559, "grad_norm": 1.4314463138580322, "learning_rate": 7.3751153111839155e-06, "loss": 0.1493173599243164, "step": 13750 }, { "epoch": 0.36245997316452777, "grad_norm": 1.0433897972106934, "learning_rate": 7.37136075702674e-06, "loss": 0.1652389645576477, "step": 13760 }, { "epoch": 0.36272338884269967, "grad_norm": 2.1839208602905273, "learning_rate": 7.367604476890895e-06, "loss": 0.15115659236907958, "step": 13770 }, { "epoch": 0.36298680452087156, "grad_norm": 1.843639612197876, "learning_rate": 7.363846473510365e-06, "loss": 0.14953770637512206, "step": 13780 }, { "epoch": 0.36325022019904346, "grad_norm": 1.7627949714660645, "learning_rate": 7.360086749620386e-06, "loss": 0.16616148948669435, "step": 13790 }, { "epoch": 0.36351363587721536, "grad_norm": 2.8135437965393066, "learning_rate": 7.356325307957449e-06, "loss": 0.16407381296157836, "step": 13800 }, { "epoch": 0.36377705155538725, "grad_norm": 1.2057462930679321, "learning_rate": 7.3525621512592905e-06, "loss": 0.15414373874664306, "step": 13810 }, { "epoch": 0.36404046723355915, "grad_norm": 5.482474327087402, "learning_rate": 7.348797282264903e-06, "loss": 0.15304806232452392, "step": 13820 }, { "epoch": 0.36430388291173105, "grad_norm": 1.3889625072479248, "learning_rate": 7.345030703714514e-06, "loss": 0.1490722417831421, "step": 13830 }, { "epoch": 0.36456729858990294, "grad_norm": 1.311397910118103, "learning_rate": 7.34126241834961e-06, "loss": 0.16018280982971192, "step": 13840 }, { "epoch": 0.36483071426807484, "grad_norm": 2.434105396270752, "learning_rate": 7.337492428912904e-06, "loss": 0.165458083152771, "step": 13850 }, { "epoch": 0.36509412994624674, "grad_norm": 2.041186809539795, "learning_rate": 7.3337207381483615e-06, "loss": 0.14607341289520265, "step": 13860 }, { "epoch": 0.36535754562441863, "grad_norm": 1.4404289722442627, "learning_rate": 7.329947348801179e-06, "loss": 0.15802072286605834, "step": 13870 }, { "epoch": 0.36562096130259053, "grad_norm": 3.510484457015991, "learning_rate": 7.326172263617795e-06, "loss": 0.15259923934936523, "step": 13880 }, { "epoch": 0.3658843769807624, "grad_norm": 3.5216615200042725, "learning_rate": 7.3223954853458774e-06, "loss": 0.15482797622680664, "step": 13890 }, { "epoch": 0.3661477926589343, "grad_norm": 1.3858215808868408, "learning_rate": 7.318617016734329e-06, "loss": 0.16019145250320435, "step": 13900 }, { "epoch": 0.3664112083371062, "grad_norm": 1.3431435823440552, "learning_rate": 7.314836860533283e-06, "loss": 0.14411551952362062, "step": 13910 }, { "epoch": 0.3666746240152781, "grad_norm": 1.4774547815322876, "learning_rate": 7.311055019494098e-06, "loss": 0.16434578895568847, "step": 13920 }, { "epoch": 0.36693803969345, "grad_norm": 1.2507346868515015, "learning_rate": 7.307271496369363e-06, "loss": 0.15729610919952391, "step": 13930 }, { "epoch": 0.3672014553716219, "grad_norm": 1.4514005184173584, "learning_rate": 7.3034862939128895e-06, "loss": 0.13955093622207643, "step": 13940 }, { "epoch": 0.3674648710497938, "grad_norm": 1.4886364936828613, "learning_rate": 7.299699414879709e-06, "loss": 0.16227163076400758, "step": 13950 }, { "epoch": 0.3677282867279657, "grad_norm": 0.9489860534667969, "learning_rate": 7.295910862026077e-06, "loss": 0.16131337881088256, "step": 13960 }, { "epoch": 0.3679917024061376, "grad_norm": 2.0338523387908936, "learning_rate": 7.292120638109468e-06, "loss": 0.15471408367156983, "step": 13970 }, { "epoch": 0.3682551180843095, "grad_norm": 2.7773895263671875, "learning_rate": 7.288328745888567e-06, "loss": 0.15599849224090576, "step": 13980 }, { "epoch": 0.3685185337624814, "grad_norm": 3.1201765537261963, "learning_rate": 7.284535188123276e-06, "loss": 0.15350518226623536, "step": 13990 }, { "epoch": 0.3687819494406533, "grad_norm": 1.6980465650558472, "learning_rate": 7.280739967574713e-06, "loss": 0.14180777072906495, "step": 14000 }, { "epoch": 0.3690453651188252, "grad_norm": 2.5995304584503174, "learning_rate": 7.276943087005202e-06, "loss": 0.1532144784927368, "step": 14010 }, { "epoch": 0.3693087807969971, "grad_norm": 1.5375802516937256, "learning_rate": 7.273144549178278e-06, "loss": 0.15289616584777832, "step": 14020 }, { "epoch": 0.369572196475169, "grad_norm": 1.6715967655181885, "learning_rate": 7.269344356858678e-06, "loss": 0.1592520594596863, "step": 14030 }, { "epoch": 0.3698356121533409, "grad_norm": 2.7061359882354736, "learning_rate": 7.265542512812347e-06, "loss": 0.16271758079528809, "step": 14040 }, { "epoch": 0.3700990278315128, "grad_norm": 2.491403579711914, "learning_rate": 7.261739019806432e-06, "loss": 0.15580294132232667, "step": 14050 }, { "epoch": 0.37036244350968467, "grad_norm": 1.018552541732788, "learning_rate": 7.257933880609279e-06, "loss": 0.15194109678268433, "step": 14060 }, { "epoch": 0.3706258591878565, "grad_norm": 3.7935116291046143, "learning_rate": 7.254127097990431e-06, "loss": 0.1718468427658081, "step": 14070 }, { "epoch": 0.3708892748660284, "grad_norm": 4.2864990234375, "learning_rate": 7.25031867472063e-06, "loss": 0.1558953881263733, "step": 14080 }, { "epoch": 0.3711526905442003, "grad_norm": 1.169075846672058, "learning_rate": 7.24650861357181e-06, "loss": 0.14671988487243653, "step": 14090 }, { "epoch": 0.3714161062223722, "grad_norm": 1.511738657951355, "learning_rate": 7.242696917317101e-06, "loss": 0.15932952165603637, "step": 14100 }, { "epoch": 0.3716795219005441, "grad_norm": 3.583207607269287, "learning_rate": 7.238883588730815e-06, "loss": 0.162739896774292, "step": 14110 }, { "epoch": 0.371942937578716, "grad_norm": 2.726109504699707, "learning_rate": 7.235068630588461e-06, "loss": 0.1554970622062683, "step": 14120 }, { "epoch": 0.3722063532568879, "grad_norm": 3.2294845581054688, "learning_rate": 7.231252045666727e-06, "loss": 0.15125024318695068, "step": 14130 }, { "epoch": 0.3724697689350598, "grad_norm": 1.906017780303955, "learning_rate": 7.227433836743492e-06, "loss": 0.1548757314682007, "step": 14140 }, { "epoch": 0.3727331846132317, "grad_norm": 1.5844364166259766, "learning_rate": 7.223614006597808e-06, "loss": 0.15931277275085448, "step": 14150 }, { "epoch": 0.3729966002914036, "grad_norm": 1.6444039344787598, "learning_rate": 7.219792558009917e-06, "loss": 0.15171133279800414, "step": 14160 }, { "epoch": 0.3732600159695755, "grad_norm": 2.37109112739563, "learning_rate": 7.215969493761231e-06, "loss": 0.14182509183883668, "step": 14170 }, { "epoch": 0.3735234316477474, "grad_norm": 2.519780397415161, "learning_rate": 7.2121448166343446e-06, "loss": 0.14984216690063476, "step": 14180 }, { "epoch": 0.37378684732591927, "grad_norm": 1.4622950553894043, "learning_rate": 7.208318529413019e-06, "loss": 0.14300186634063722, "step": 14190 }, { "epoch": 0.37405026300409117, "grad_norm": 2.3116230964660645, "learning_rate": 7.204490634882193e-06, "loss": 0.1438391089439392, "step": 14200 }, { "epoch": 0.37431367868226306, "grad_norm": 2.457341194152832, "learning_rate": 7.2006611358279735e-06, "loss": 0.15219467878341675, "step": 14210 }, { "epoch": 0.37457709436043496, "grad_norm": 1.323244571685791, "learning_rate": 7.196830035037634e-06, "loss": 0.13737831115722657, "step": 14220 }, { "epoch": 0.37484051003860686, "grad_norm": 1.6529995203018188, "learning_rate": 7.1929973352996165e-06, "loss": 0.15404808521270752, "step": 14230 }, { "epoch": 0.37510392571677875, "grad_norm": 1.4308706521987915, "learning_rate": 7.189163039403524e-06, "loss": 0.16234565973281861, "step": 14240 }, { "epoch": 0.37536734139495065, "grad_norm": 1.7528282403945923, "learning_rate": 7.185327150140123e-06, "loss": 0.1438475728034973, "step": 14250 }, { "epoch": 0.37563075707312255, "grad_norm": 2.2275502681732178, "learning_rate": 7.1814896703013385e-06, "loss": 0.14082204103469848, "step": 14260 }, { "epoch": 0.37589417275129444, "grad_norm": 4.195061683654785, "learning_rate": 7.1776506026802565e-06, "loss": 0.16757893562316895, "step": 14270 }, { "epoch": 0.37615758842946634, "grad_norm": 2.2586750984191895, "learning_rate": 7.1738099500711115e-06, "loss": 0.14958466291427613, "step": 14280 }, { "epoch": 0.37642100410763824, "grad_norm": 3.3764665126800537, "learning_rate": 7.1699677152693e-06, "loss": 0.14840222597122193, "step": 14290 }, { "epoch": 0.37668441978581013, "grad_norm": 1.3436477184295654, "learning_rate": 7.166123901071364e-06, "loss": 0.15582242012023925, "step": 14300 }, { "epoch": 0.37694783546398203, "grad_norm": 1.7756497859954834, "learning_rate": 7.1622785102750005e-06, "loss": 0.15267500877380372, "step": 14310 }, { "epoch": 0.3772112511421539, "grad_norm": 1.6340587139129639, "learning_rate": 7.158431545679046e-06, "loss": 0.15218218564987182, "step": 14320 }, { "epoch": 0.3774746668203258, "grad_norm": 2.9876198768615723, "learning_rate": 7.154583010083492e-06, "loss": 0.15662260055541993, "step": 14330 }, { "epoch": 0.3777380824984977, "grad_norm": 1.7915198802947998, "learning_rate": 7.150732906289465e-06, "loss": 0.15686209201812745, "step": 14340 }, { "epoch": 0.3780014981766696, "grad_norm": 2.0315020084381104, "learning_rate": 7.146881237099239e-06, "loss": 0.1397937297821045, "step": 14350 }, { "epoch": 0.3782649138548415, "grad_norm": 1.8205928802490234, "learning_rate": 7.143028005316225e-06, "loss": 0.15171622037887572, "step": 14360 }, { "epoch": 0.3785283295330134, "grad_norm": 1.8804856538772583, "learning_rate": 7.13917321374497e-06, "loss": 0.16057965755462647, "step": 14370 }, { "epoch": 0.3787917452111853, "grad_norm": 1.7356617450714111, "learning_rate": 7.135316865191157e-06, "loss": 0.16657500267028807, "step": 14380 }, { "epoch": 0.3790551608893572, "grad_norm": 1.6962946653366089, "learning_rate": 7.131458962461606e-06, "loss": 0.1593935489654541, "step": 14390 }, { "epoch": 0.3793185765675291, "grad_norm": 3.122250556945801, "learning_rate": 7.127599508364264e-06, "loss": 0.14525513648986815, "step": 14400 }, { "epoch": 0.379581992245701, "grad_norm": 4.390376091003418, "learning_rate": 7.1237385057082046e-06, "loss": 0.16974735260009766, "step": 14410 }, { "epoch": 0.3798454079238729, "grad_norm": 1.8558112382888794, "learning_rate": 7.119875957303637e-06, "loss": 0.14992204904556275, "step": 14420 }, { "epoch": 0.3801088236020448, "grad_norm": 4.664941787719727, "learning_rate": 7.116011865961886e-06, "loss": 0.16463587284088135, "step": 14430 }, { "epoch": 0.3803722392802167, "grad_norm": 4.31261682510376, "learning_rate": 7.11214623449541e-06, "loss": 0.15657806396484375, "step": 14440 }, { "epoch": 0.3806356549583886, "grad_norm": 4.336062431335449, "learning_rate": 7.108279065717776e-06, "loss": 0.1523685336112976, "step": 14450 }, { "epoch": 0.3808990706365605, "grad_norm": 2.5805258750915527, "learning_rate": 7.10441036244368e-06, "loss": 0.1471353530883789, "step": 14460 }, { "epoch": 0.3811624863147323, "grad_norm": 1.8932186365127563, "learning_rate": 7.10054012748893e-06, "loss": 0.14370994567871093, "step": 14470 }, { "epoch": 0.3814259019929042, "grad_norm": 2.2590808868408203, "learning_rate": 7.09666836367045e-06, "loss": 0.15718519687652588, "step": 14480 }, { "epoch": 0.3816893176710761, "grad_norm": 3.7195332050323486, "learning_rate": 7.0927950738062774e-06, "loss": 0.15751456022262572, "step": 14490 }, { "epoch": 0.381952733349248, "grad_norm": 2.7838809490203857, "learning_rate": 7.088920260715558e-06, "loss": 0.15015147924423217, "step": 14500 }, { "epoch": 0.3822161490274199, "grad_norm": 1.6218609809875488, "learning_rate": 7.085043927218551e-06, "loss": 0.14421465396881103, "step": 14510 }, { "epoch": 0.3824795647055918, "grad_norm": 2.228858232498169, "learning_rate": 7.0811660761366165e-06, "loss": 0.15153076648712158, "step": 14520 }, { "epoch": 0.3827429803837637, "grad_norm": 2.5488924980163574, "learning_rate": 7.077286710292224e-06, "loss": 0.15887593030929564, "step": 14530 }, { "epoch": 0.3830063960619356, "grad_norm": 3.3246054649353027, "learning_rate": 7.07340583250894e-06, "loss": 0.1398059368133545, "step": 14540 }, { "epoch": 0.3832698117401075, "grad_norm": 1.616461992263794, "learning_rate": 7.069523445611438e-06, "loss": 0.14656665325164794, "step": 14550 }, { "epoch": 0.3835332274182794, "grad_norm": 3.4832921028137207, "learning_rate": 7.065639552425484e-06, "loss": 0.17558248043060304, "step": 14560 }, { "epoch": 0.3837966430964513, "grad_norm": 2.767500638961792, "learning_rate": 7.061754155777947e-06, "loss": 0.14878178834915162, "step": 14570 }, { "epoch": 0.3840600587746232, "grad_norm": 5.436642646789551, "learning_rate": 7.057867258496781e-06, "loss": 0.1569177031517029, "step": 14580 }, { "epoch": 0.3843234744527951, "grad_norm": 2.3578312397003174, "learning_rate": 7.053978863411042e-06, "loss": 0.15260705947875977, "step": 14590 }, { "epoch": 0.384586890130967, "grad_norm": 1.8397252559661865, "learning_rate": 7.050088973350869e-06, "loss": 0.14984655380249023, "step": 14600 }, { "epoch": 0.38485030580913887, "grad_norm": 2.044610023498535, "learning_rate": 7.0461975911474945e-06, "loss": 0.1639110565185547, "step": 14610 }, { "epoch": 0.38511372148731077, "grad_norm": 4.145593643188477, "learning_rate": 7.042304719633232e-06, "loss": 0.15118982791900634, "step": 14620 }, { "epoch": 0.38537713716548266, "grad_norm": 3.40494966506958, "learning_rate": 7.038410361641481e-06, "loss": 0.14631330966949463, "step": 14630 }, { "epoch": 0.38564055284365456, "grad_norm": 3.150256633758545, "learning_rate": 7.034514520006724e-06, "loss": 0.15599969625473023, "step": 14640 }, { "epoch": 0.38590396852182646, "grad_norm": 1.089646816253662, "learning_rate": 7.030617197564524e-06, "loss": 0.15873456001281738, "step": 14650 }, { "epoch": 0.38616738419999835, "grad_norm": 1.2397304773330688, "learning_rate": 7.02671839715152e-06, "loss": 0.15663055181503296, "step": 14660 }, { "epoch": 0.38643079987817025, "grad_norm": 2.9588797092437744, "learning_rate": 7.022818121605426e-06, "loss": 0.14629750251770018, "step": 14670 }, { "epoch": 0.38669421555634215, "grad_norm": 3.1346864700317383, "learning_rate": 7.0189163737650315e-06, "loss": 0.12537091970443726, "step": 14680 }, { "epoch": 0.38695763123451404, "grad_norm": 2.5749590396881104, "learning_rate": 7.015013156470195e-06, "loss": 0.15571563243865966, "step": 14690 }, { "epoch": 0.38722104691268594, "grad_norm": 2.496086835861206, "learning_rate": 7.01110847256185e-06, "loss": 0.14613039493560792, "step": 14700 }, { "epoch": 0.38748446259085784, "grad_norm": 1.1619689464569092, "learning_rate": 7.007202324881992e-06, "loss": 0.1517017126083374, "step": 14710 }, { "epoch": 0.38774787826902973, "grad_norm": 3.4924020767211914, "learning_rate": 7.003294716273683e-06, "loss": 0.1491444945335388, "step": 14720 }, { "epoch": 0.38801129394720163, "grad_norm": 2.9700355529785156, "learning_rate": 6.9993856495810496e-06, "loss": 0.15014162063598632, "step": 14730 }, { "epoch": 0.3882747096253735, "grad_norm": 2.1629016399383545, "learning_rate": 6.995475127649282e-06, "loss": 0.14860868453979492, "step": 14740 }, { "epoch": 0.3885381253035454, "grad_norm": 3.9265480041503906, "learning_rate": 6.991563153324622e-06, "loss": 0.15113554000854493, "step": 14750 }, { "epoch": 0.3888015409817173, "grad_norm": 1.765947937965393, "learning_rate": 6.9876497294543765e-06, "loss": 0.1447599411010742, "step": 14760 }, { "epoch": 0.3890649566598892, "grad_norm": 4.240228652954102, "learning_rate": 6.983734858886903e-06, "loss": 0.15431568622589112, "step": 14770 }, { "epoch": 0.3893283723380611, "grad_norm": 2.385678768157959, "learning_rate": 6.979818544471615e-06, "loss": 0.1475411295890808, "step": 14780 }, { "epoch": 0.389591788016233, "grad_norm": 2.317544937133789, "learning_rate": 6.975900789058973e-06, "loss": 0.15385537147521972, "step": 14790 }, { "epoch": 0.3898552036944049, "grad_norm": 2.182826042175293, "learning_rate": 6.971981595500488e-06, "loss": 0.165794038772583, "step": 14800 }, { "epoch": 0.3901186193725768, "grad_norm": 3.473280191421509, "learning_rate": 6.96806096664872e-06, "loss": 0.15069515705108644, "step": 14810 }, { "epoch": 0.3903820350507487, "grad_norm": 2.1884214878082275, "learning_rate": 6.964138905357271e-06, "loss": 0.15762898921966553, "step": 14820 }, { "epoch": 0.3906454507289206, "grad_norm": 3.230773448944092, "learning_rate": 6.960215414480788e-06, "loss": 0.139265775680542, "step": 14830 }, { "epoch": 0.3909088664070925, "grad_norm": 1.6145650148391724, "learning_rate": 6.956290496874954e-06, "loss": 0.1616771101951599, "step": 14840 }, { "epoch": 0.3911722820852644, "grad_norm": 1.0805432796478271, "learning_rate": 6.952364155396494e-06, "loss": 0.15177109241485595, "step": 14850 }, { "epoch": 0.39143569776343623, "grad_norm": 2.573309898376465, "learning_rate": 6.948436392903171e-06, "loss": 0.15149340629577637, "step": 14860 }, { "epoch": 0.3916991134416081, "grad_norm": 1.8439639806747437, "learning_rate": 6.944507212253778e-06, "loss": 0.15371508598327638, "step": 14870 }, { "epoch": 0.39196252911978, "grad_norm": 4.540953636169434, "learning_rate": 6.940576616308142e-06, "loss": 0.16377884149551392, "step": 14880 }, { "epoch": 0.3922259447979519, "grad_norm": 2.845085382461548, "learning_rate": 6.936644607927124e-06, "loss": 0.1557713747024536, "step": 14890 }, { "epoch": 0.3924893604761238, "grad_norm": 4.319413661956787, "learning_rate": 6.9327111899726026e-06, "loss": 0.15743638277053834, "step": 14900 }, { "epoch": 0.3927527761542957, "grad_norm": 2.343224287033081, "learning_rate": 6.928776365307494e-06, "loss": 0.15994821786880492, "step": 14910 }, { "epoch": 0.3930161918324676, "grad_norm": 2.019127130508423, "learning_rate": 6.924840136795734e-06, "loss": 0.15934523344039916, "step": 14920 }, { "epoch": 0.3932796075106395, "grad_norm": 1.058731198310852, "learning_rate": 6.920902507302276e-06, "loss": 0.15077754259109497, "step": 14930 }, { "epoch": 0.3935430231888114, "grad_norm": 1.739488959312439, "learning_rate": 6.916963479693099e-06, "loss": 0.1470818281173706, "step": 14940 }, { "epoch": 0.3938064388669833, "grad_norm": 3.6511330604553223, "learning_rate": 6.913023056835198e-06, "loss": 0.1557040572166443, "step": 14950 }, { "epoch": 0.3940698545451552, "grad_norm": 2.0465829372406006, "learning_rate": 6.909081241596579e-06, "loss": 0.16470963954925538, "step": 14960 }, { "epoch": 0.3943332702233271, "grad_norm": 4.5905327796936035, "learning_rate": 6.905138036846269e-06, "loss": 0.14224824905395508, "step": 14970 }, { "epoch": 0.394596685901499, "grad_norm": 3.484959363937378, "learning_rate": 6.901193445454301e-06, "loss": 0.15862753391265869, "step": 14980 }, { "epoch": 0.3948601015796709, "grad_norm": 3.459350824356079, "learning_rate": 6.897247470291718e-06, "loss": 0.15845341682434083, "step": 14990 }, { "epoch": 0.3951235172578428, "grad_norm": 1.682303786277771, "learning_rate": 6.893300114230573e-06, "loss": 0.16438491344451905, "step": 15000 }, { "epoch": 0.3953869329360147, "grad_norm": 1.6633336544036865, "learning_rate": 6.88935138014392e-06, "loss": 0.13771259784698486, "step": 15010 }, { "epoch": 0.3956503486141866, "grad_norm": 1.7457149028778076, "learning_rate": 6.88540127090582e-06, "loss": 0.16378339529037475, "step": 15020 }, { "epoch": 0.39591376429235847, "grad_norm": 1.6133472919464111, "learning_rate": 6.881449789391331e-06, "loss": 0.15882420539855957, "step": 15030 }, { "epoch": 0.39617717997053037, "grad_norm": 1.4689128398895264, "learning_rate": 6.877496938476515e-06, "loss": 0.14161667823791504, "step": 15040 }, { "epoch": 0.39644059564870227, "grad_norm": 2.240866184234619, "learning_rate": 6.873542721038422e-06, "loss": 0.15115902423858643, "step": 15050 }, { "epoch": 0.39670401132687416, "grad_norm": 1.8431570529937744, "learning_rate": 6.869587139955107e-06, "loss": 0.14806027412414552, "step": 15060 }, { "epoch": 0.39696742700504606, "grad_norm": 2.484621286392212, "learning_rate": 6.865630198105609e-06, "loss": 0.155122971534729, "step": 15070 }, { "epoch": 0.39723084268321796, "grad_norm": 2.153735399246216, "learning_rate": 6.861671898369965e-06, "loss": 0.15208015441894532, "step": 15080 }, { "epoch": 0.39749425836138985, "grad_norm": 1.8708301782608032, "learning_rate": 6.857712243629193e-06, "loss": 0.14204760789871215, "step": 15090 }, { "epoch": 0.39775767403956175, "grad_norm": 1.9889510869979858, "learning_rate": 6.8537512367653e-06, "loss": 0.15054476261138916, "step": 15100 }, { "epoch": 0.39802108971773364, "grad_norm": 1.9190653562545776, "learning_rate": 6.8497888806612815e-06, "loss": 0.14857161045074463, "step": 15110 }, { "epoch": 0.39828450539590554, "grad_norm": 1.0409042835235596, "learning_rate": 6.845825178201106e-06, "loss": 0.16333379745483398, "step": 15120 }, { "epoch": 0.39854792107407744, "grad_norm": 3.800628185272217, "learning_rate": 6.8418601322697346e-06, "loss": 0.15872656106948851, "step": 15130 }, { "epoch": 0.39881133675224933, "grad_norm": 1.6401008367538452, "learning_rate": 6.837893745753092e-06, "loss": 0.13947489261627197, "step": 15140 }, { "epoch": 0.39907475243042123, "grad_norm": 3.2302064895629883, "learning_rate": 6.833926021538086e-06, "loss": 0.1568101167678833, "step": 15150 }, { "epoch": 0.39933816810859313, "grad_norm": 3.6806745529174805, "learning_rate": 6.829956962512602e-06, "loss": 0.15356864929199218, "step": 15160 }, { "epoch": 0.399601583786765, "grad_norm": 1.1895506381988525, "learning_rate": 6.825986571565489e-06, "loss": 0.14030866622924804, "step": 15170 }, { "epoch": 0.3998649994649369, "grad_norm": 1.9752578735351562, "learning_rate": 6.822014851586569e-06, "loss": 0.15619266033172607, "step": 15180 }, { "epoch": 0.4001284151431088, "grad_norm": 1.5992157459259033, "learning_rate": 6.818041805466631e-06, "loss": 0.16117502450942994, "step": 15190 }, { "epoch": 0.4003918308212807, "grad_norm": 1.7376328706741333, "learning_rate": 6.8140674360974305e-06, "loss": 0.15224387645721435, "step": 15200 }, { "epoch": 0.4006552464994526, "grad_norm": 1.9963774681091309, "learning_rate": 6.810091746371685e-06, "loss": 0.15300889015197755, "step": 15210 }, { "epoch": 0.4009186621776245, "grad_norm": 2.2298736572265625, "learning_rate": 6.806114739183072e-06, "loss": 0.15824166536331177, "step": 15220 }, { "epoch": 0.4011820778557964, "grad_norm": 2.2138445377349854, "learning_rate": 6.80213641742623e-06, "loss": 0.15028576850891112, "step": 15230 }, { "epoch": 0.4014454935339683, "grad_norm": 2.07983136177063, "learning_rate": 6.798156783996752e-06, "loss": 0.14576435089111328, "step": 15240 }, { "epoch": 0.4017089092121402, "grad_norm": 2.968247175216675, "learning_rate": 6.794175841791187e-06, "loss": 0.14736198186874389, "step": 15250 }, { "epoch": 0.40197232489031204, "grad_norm": 2.0666275024414062, "learning_rate": 6.7901935937070385e-06, "loss": 0.15271320343017578, "step": 15260 }, { "epoch": 0.40223574056848393, "grad_norm": 1.3685168027877808, "learning_rate": 6.786210042642756e-06, "loss": 0.1579584002494812, "step": 15270 }, { "epoch": 0.40249915624665583, "grad_norm": 2.2259645462036133, "learning_rate": 6.782225191497741e-06, "loss": 0.1611625909805298, "step": 15280 }, { "epoch": 0.40276257192482773, "grad_norm": 1.2034912109375, "learning_rate": 6.778239043172339e-06, "loss": 0.1482660174369812, "step": 15290 }, { "epoch": 0.4030259876029996, "grad_norm": 3.932584047317505, "learning_rate": 6.7742516005678435e-06, "loss": 0.15537188053131104, "step": 15300 }, { "epoch": 0.4032894032811715, "grad_norm": 2.6423778533935547, "learning_rate": 6.770262866586483e-06, "loss": 0.14538400173187255, "step": 15310 }, { "epoch": 0.4035528189593434, "grad_norm": 2.2796359062194824, "learning_rate": 6.766272844131435e-06, "loss": 0.15186498165130616, "step": 15320 }, { "epoch": 0.4038162346375153, "grad_norm": 3.334988832473755, "learning_rate": 6.762281536106806e-06, "loss": 0.1521677255630493, "step": 15330 }, { "epoch": 0.4040796503156872, "grad_norm": 1.7045364379882812, "learning_rate": 6.758288945417646e-06, "loss": 0.15566176176071167, "step": 15340 }, { "epoch": 0.4043430659938591, "grad_norm": 2.523469924926758, "learning_rate": 6.75429507496993e-06, "loss": 0.15350182056427003, "step": 15350 }, { "epoch": 0.404606481672031, "grad_norm": 1.1940371990203857, "learning_rate": 6.750299927670573e-06, "loss": 0.1522940993309021, "step": 15360 }, { "epoch": 0.4048698973502029, "grad_norm": 2.0505709648132324, "learning_rate": 6.746303506427412e-06, "loss": 0.151654052734375, "step": 15370 }, { "epoch": 0.4051333130283748, "grad_norm": 4.722397327423096, "learning_rate": 6.742305814149219e-06, "loss": 0.15147918462753296, "step": 15380 }, { "epoch": 0.4053967287065467, "grad_norm": 1.5813013315200806, "learning_rate": 6.738306853745686e-06, "loss": 0.15582174062728882, "step": 15390 }, { "epoch": 0.4056601443847186, "grad_norm": 1.598442792892456, "learning_rate": 6.734306628127426e-06, "loss": 0.1548468589782715, "step": 15400 }, { "epoch": 0.4059235600628905, "grad_norm": 2.6900899410247803, "learning_rate": 6.730305140205978e-06, "loss": 0.16457654237747193, "step": 15410 }, { "epoch": 0.4061869757410624, "grad_norm": 1.7876075506210327, "learning_rate": 6.726302392893797e-06, "loss": 0.1397912859916687, "step": 15420 }, { "epoch": 0.4064503914192343, "grad_norm": 1.8887474536895752, "learning_rate": 6.722298389104258e-06, "loss": 0.16292905807495117, "step": 15430 }, { "epoch": 0.4067138070974062, "grad_norm": 1.7844163179397583, "learning_rate": 6.718293131751643e-06, "loss": 0.1592571973800659, "step": 15440 }, { "epoch": 0.4069772227755781, "grad_norm": 6.603170871734619, "learning_rate": 6.714286623751154e-06, "loss": 0.15450613498687743, "step": 15450 }, { "epoch": 0.40724063845374997, "grad_norm": 3.2693963050842285, "learning_rate": 6.710278868018898e-06, "loss": 0.14406148195266724, "step": 15460 }, { "epoch": 0.40750405413192187, "grad_norm": 3.117006301879883, "learning_rate": 6.706269867471896e-06, "loss": 0.146087646484375, "step": 15470 }, { "epoch": 0.40776746981009376, "grad_norm": 1.5090850591659546, "learning_rate": 6.7022596250280696e-06, "loss": 0.15344529151916503, "step": 15480 }, { "epoch": 0.40803088548826566, "grad_norm": 2.740276336669922, "learning_rate": 6.6982481436062455e-06, "loss": 0.1472179412841797, "step": 15490 }, { "epoch": 0.40829430116643756, "grad_norm": 1.5627132654190063, "learning_rate": 6.6942354261261535e-06, "loss": 0.1574077606201172, "step": 15500 }, { "epoch": 0.40855771684460945, "grad_norm": 1.4387496709823608, "learning_rate": 6.690221475508423e-06, "loss": 0.16032590866088867, "step": 15510 }, { "epoch": 0.40882113252278135, "grad_norm": 5.710041046142578, "learning_rate": 6.686206294674582e-06, "loss": 0.15103448629379274, "step": 15520 }, { "epoch": 0.40908454820095325, "grad_norm": 1.4006835222244263, "learning_rate": 6.6821898865470496e-06, "loss": 0.14309828281402587, "step": 15530 }, { "epoch": 0.40934796387912514, "grad_norm": 2.2762560844421387, "learning_rate": 6.67817225404914e-06, "loss": 0.15536186695098878, "step": 15540 }, { "epoch": 0.40961137955729704, "grad_norm": 1.5115293264389038, "learning_rate": 6.674153400105063e-06, "loss": 0.13918720483779906, "step": 15550 }, { "epoch": 0.40987479523546894, "grad_norm": 1.5558501482009888, "learning_rate": 6.670133327639914e-06, "loss": 0.162703537940979, "step": 15560 }, { "epoch": 0.41013821091364083, "grad_norm": 2.2454631328582764, "learning_rate": 6.666112039579671e-06, "loss": 0.15129215717315675, "step": 15570 }, { "epoch": 0.41040162659181273, "grad_norm": 2.060605764389038, "learning_rate": 6.662089538851204e-06, "loss": 0.14612984657287598, "step": 15580 }, { "epoch": 0.4106650422699846, "grad_norm": 1.5460985898971558, "learning_rate": 6.658065828382262e-06, "loss": 0.1356353521347046, "step": 15590 }, { "epoch": 0.4109284579481565, "grad_norm": 4.363698482513428, "learning_rate": 6.654040911101476e-06, "loss": 0.15506410598754883, "step": 15600 }, { "epoch": 0.4111918736263284, "grad_norm": 3.449538469314575, "learning_rate": 6.6500147899383525e-06, "loss": 0.14638261795043944, "step": 15610 }, { "epoch": 0.4114552893045003, "grad_norm": 1.586491346359253, "learning_rate": 6.645987467823277e-06, "loss": 0.13291072845458984, "step": 15620 }, { "epoch": 0.4117187049826722, "grad_norm": 2.4653236865997314, "learning_rate": 6.641958947687509e-06, "loss": 0.13612362146377563, "step": 15630 }, { "epoch": 0.4119821206608441, "grad_norm": 3.289590358734131, "learning_rate": 6.63792923246318e-06, "loss": 0.1525922656059265, "step": 15640 }, { "epoch": 0.412245536339016, "grad_norm": 1.8911653757095337, "learning_rate": 6.633898325083288e-06, "loss": 0.15989203453063966, "step": 15650 }, { "epoch": 0.41250895201718785, "grad_norm": 3.153531074523926, "learning_rate": 6.629866228481703e-06, "loss": 0.15104719400405883, "step": 15660 }, { "epoch": 0.41277236769535974, "grad_norm": 4.890185356140137, "learning_rate": 6.6258329455931565e-06, "loss": 0.13993368148803711, "step": 15670 }, { "epoch": 0.41303578337353164, "grad_norm": 1.1893421411514282, "learning_rate": 6.62179847935325e-06, "loss": 0.1468294620513916, "step": 15680 }, { "epoch": 0.41329919905170354, "grad_norm": 1.2914068698883057, "learning_rate": 6.61776283269844e-06, "loss": 0.16962041854858398, "step": 15690 }, { "epoch": 0.41356261472987543, "grad_norm": 1.310512900352478, "learning_rate": 6.613726008566043e-06, "loss": 0.1479944944381714, "step": 15700 }, { "epoch": 0.41382603040804733, "grad_norm": 2.691415548324585, "learning_rate": 6.609688009894235e-06, "loss": 0.1445556402206421, "step": 15710 }, { "epoch": 0.4140894460862192, "grad_norm": 3.46160888671875, "learning_rate": 6.605648839622046e-06, "loss": 0.13898346424102784, "step": 15720 }, { "epoch": 0.4143528617643911, "grad_norm": 1.9372191429138184, "learning_rate": 6.601608500689358e-06, "loss": 0.14924004077911376, "step": 15730 }, { "epoch": 0.414616277442563, "grad_norm": 2.426513195037842, "learning_rate": 6.597566996036904e-06, "loss": 0.1600063681602478, "step": 15740 }, { "epoch": 0.4148796931207349, "grad_norm": 2.3759570121765137, "learning_rate": 6.593524328606265e-06, "loss": 0.16115932464599608, "step": 15750 }, { "epoch": 0.4151431087989068, "grad_norm": 1.6453131437301636, "learning_rate": 6.589480501339869e-06, "loss": 0.15410730838775635, "step": 15760 }, { "epoch": 0.4154065244770787, "grad_norm": 1.518022060394287, "learning_rate": 6.58543551718099e-06, "loss": 0.15027801990509032, "step": 15770 }, { "epoch": 0.4156699401552506, "grad_norm": 1.784364104270935, "learning_rate": 6.581389379073738e-06, "loss": 0.14731462001800538, "step": 15780 }, { "epoch": 0.4159333558334225, "grad_norm": 1.4193766117095947, "learning_rate": 6.57734208996307e-06, "loss": 0.1526337146759033, "step": 15790 }, { "epoch": 0.4161967715115944, "grad_norm": 2.5405631065368652, "learning_rate": 6.573293652794776e-06, "loss": 0.14769449234008789, "step": 15800 }, { "epoch": 0.4164601871897663, "grad_norm": 3.135498523712158, "learning_rate": 6.569244070515484e-06, "loss": 0.144717800617218, "step": 15810 }, { "epoch": 0.4167236028679382, "grad_norm": 1.7804893255233765, "learning_rate": 6.565193346072655e-06, "loss": 0.14811220169067382, "step": 15820 }, { "epoch": 0.4169870185461101, "grad_norm": 1.5557409524917603, "learning_rate": 6.56114148241458e-06, "loss": 0.1483596920967102, "step": 15830 }, { "epoch": 0.417250434224282, "grad_norm": 2.270578384399414, "learning_rate": 6.55708848249038e-06, "loss": 0.1552456021308899, "step": 15840 }, { "epoch": 0.4175138499024539, "grad_norm": 1.4422086477279663, "learning_rate": 6.553034349250005e-06, "loss": 0.15225268602371217, "step": 15850 }, { "epoch": 0.4177772655806258, "grad_norm": 2.1741247177124023, "learning_rate": 6.548979085644227e-06, "loss": 0.15079454183578492, "step": 15860 }, { "epoch": 0.4180406812587977, "grad_norm": 3.0607070922851562, "learning_rate": 6.544922694624641e-06, "loss": 0.1374043345451355, "step": 15870 }, { "epoch": 0.41830409693696957, "grad_norm": 5.034666061401367, "learning_rate": 6.540865179143665e-06, "loss": 0.15435483455657958, "step": 15880 }, { "epoch": 0.41856751261514147, "grad_norm": 3.2251176834106445, "learning_rate": 6.536806542154533e-06, "loss": 0.14171075820922852, "step": 15890 }, { "epoch": 0.41883092829331336, "grad_norm": 4.608428001403809, "learning_rate": 6.532746786611297e-06, "loss": 0.15162315368652343, "step": 15900 }, { "epoch": 0.41909434397148526, "grad_norm": 2.872896909713745, "learning_rate": 6.52868591546882e-06, "loss": 0.1595179557800293, "step": 15910 }, { "epoch": 0.41935775964965716, "grad_norm": 1.2327673435211182, "learning_rate": 6.524623931682781e-06, "loss": 0.15009040832519532, "step": 15920 }, { "epoch": 0.41962117532782905, "grad_norm": 1.538149356842041, "learning_rate": 6.520560838209666e-06, "loss": 0.1444324254989624, "step": 15930 }, { "epoch": 0.41988459100600095, "grad_norm": 2.2667038440704346, "learning_rate": 6.516496638006773e-06, "loss": 0.14313054084777832, "step": 15940 }, { "epoch": 0.42014800668417285, "grad_norm": 2.1941897869110107, "learning_rate": 6.512431334032197e-06, "loss": 0.14708259105682372, "step": 15950 }, { "epoch": 0.42041142236234474, "grad_norm": 2.549858331680298, "learning_rate": 6.5083649292448446e-06, "loss": 0.15672290325164795, "step": 15960 }, { "epoch": 0.42067483804051664, "grad_norm": 1.0367659330368042, "learning_rate": 6.50429742660442e-06, "loss": 0.14951598644256592, "step": 15970 }, { "epoch": 0.42093825371868854, "grad_norm": 1.150990605354309, "learning_rate": 6.500228829071429e-06, "loss": 0.15651794672012329, "step": 15980 }, { "epoch": 0.42120166939686043, "grad_norm": 3.141935348510742, "learning_rate": 6.49615913960717e-06, "loss": 0.15344038009643554, "step": 15990 }, { "epoch": 0.42146508507503233, "grad_norm": 1.5346391201019287, "learning_rate": 6.492088361173739e-06, "loss": 0.14766507148742675, "step": 16000 }, { "epoch": 0.4217285007532042, "grad_norm": 2.454958200454712, "learning_rate": 6.4880164967340244e-06, "loss": 0.14876046180725097, "step": 16010 }, { "epoch": 0.4219919164313761, "grad_norm": 3.0405571460723877, "learning_rate": 6.483943549251705e-06, "loss": 0.15905874967575073, "step": 16020 }, { "epoch": 0.422255332109548, "grad_norm": 1.7550724744796753, "learning_rate": 6.47986952169125e-06, "loss": 0.16277883052825928, "step": 16030 }, { "epoch": 0.4225187477877199, "grad_norm": 2.9173028469085693, "learning_rate": 6.4757944170179066e-06, "loss": 0.13737844228744506, "step": 16040 }, { "epoch": 0.42278216346589176, "grad_norm": 1.3207155466079712, "learning_rate": 6.4717182381977175e-06, "loss": 0.13488609790802003, "step": 16050 }, { "epoch": 0.42304557914406365, "grad_norm": 1.4663002490997314, "learning_rate": 6.467640988197499e-06, "loss": 0.1699681043624878, "step": 16060 }, { "epoch": 0.42330899482223555, "grad_norm": 1.3112162351608276, "learning_rate": 6.4635626699848505e-06, "loss": 0.15303276777267455, "step": 16070 }, { "epoch": 0.42357241050040745, "grad_norm": 2.7668979167938232, "learning_rate": 6.459483286528146e-06, "loss": 0.15468331575393676, "step": 16080 }, { "epoch": 0.42383582617857934, "grad_norm": 2.2401654720306396, "learning_rate": 6.455402840796539e-06, "loss": 0.14539921283721924, "step": 16090 }, { "epoch": 0.42409924185675124, "grad_norm": 1.785015344619751, "learning_rate": 6.4513213357599516e-06, "loss": 0.15210950374603271, "step": 16100 }, { "epoch": 0.42436265753492314, "grad_norm": 2.06204891204834, "learning_rate": 6.447238774389082e-06, "loss": 0.14167630672454834, "step": 16110 }, { "epoch": 0.42462607321309503, "grad_norm": 3.9393198490142822, "learning_rate": 6.4431551596553935e-06, "loss": 0.15188243389129638, "step": 16120 }, { "epoch": 0.42488948889126693, "grad_norm": 2.906160354614258, "learning_rate": 6.4390704945311166e-06, "loss": 0.14970197677612304, "step": 16130 }, { "epoch": 0.4251529045694388, "grad_norm": 1.437076449394226, "learning_rate": 6.434984781989246e-06, "loss": 0.16051533222198486, "step": 16140 }, { "epoch": 0.4254163202476107, "grad_norm": 1.5065956115722656, "learning_rate": 6.430898025003541e-06, "loss": 0.15327062606811523, "step": 16150 }, { "epoch": 0.4256797359257826, "grad_norm": 2.755612850189209, "learning_rate": 6.4268102265485195e-06, "loss": 0.14970991611480713, "step": 16160 }, { "epoch": 0.4259431516039545, "grad_norm": 1.3324555158615112, "learning_rate": 6.422721389599458e-06, "loss": 0.15704214572906494, "step": 16170 }, { "epoch": 0.4262065672821264, "grad_norm": 6.0861406326293945, "learning_rate": 6.418631517132387e-06, "loss": 0.15375300645828247, "step": 16180 }, { "epoch": 0.4264699829602983, "grad_norm": 1.3318520784378052, "learning_rate": 6.414540612124092e-06, "loss": 0.15960067510604858, "step": 16190 }, { "epoch": 0.4267333986384702, "grad_norm": 2.634021043777466, "learning_rate": 6.410448677552112e-06, "loss": 0.15466182231903075, "step": 16200 }, { "epoch": 0.4269968143166421, "grad_norm": 1.2228977680206299, "learning_rate": 6.406355716394732e-06, "loss": 0.14957298040390016, "step": 16210 }, { "epoch": 0.427260229994814, "grad_norm": 4.7478461265563965, "learning_rate": 6.402261731630988e-06, "loss": 0.15005232095718385, "step": 16220 }, { "epoch": 0.4275236456729859, "grad_norm": 2.6820881366729736, "learning_rate": 6.3981667262406546e-06, "loss": 0.1456017255783081, "step": 16230 }, { "epoch": 0.4277870613511578, "grad_norm": 1.4689512252807617, "learning_rate": 6.394070703204258e-06, "loss": 0.14401158094406127, "step": 16240 }, { "epoch": 0.4280504770293297, "grad_norm": 1.9369829893112183, "learning_rate": 6.3899736655030566e-06, "loss": 0.1536228895187378, "step": 16250 }, { "epoch": 0.4283138927075016, "grad_norm": 2.7333920001983643, "learning_rate": 6.385875616119054e-06, "loss": 0.15573830604553224, "step": 16260 }, { "epoch": 0.4285773083856735, "grad_norm": 1.8335044384002686, "learning_rate": 6.3817765580349845e-06, "loss": 0.13915979862213135, "step": 16270 }, { "epoch": 0.4288407240638454, "grad_norm": 2.106356382369995, "learning_rate": 6.377676494234323e-06, "loss": 0.1554669737815857, "step": 16280 }, { "epoch": 0.4291041397420173, "grad_norm": 2.2071127891540527, "learning_rate": 6.373575427701271e-06, "loss": 0.1454970121383667, "step": 16290 }, { "epoch": 0.4293675554201892, "grad_norm": 3.7389230728149414, "learning_rate": 6.369473361420759e-06, "loss": 0.13949190378189086, "step": 16300 }, { "epoch": 0.42963097109836107, "grad_norm": 4.067681312561035, "learning_rate": 6.365370298378454e-06, "loss": 0.15759599208831787, "step": 16310 }, { "epoch": 0.42989438677653297, "grad_norm": 1.6478079557418823, "learning_rate": 6.361266241560738e-06, "loss": 0.13914363384246825, "step": 16320 }, { "epoch": 0.43015780245470486, "grad_norm": 1.0137507915496826, "learning_rate": 6.35716119395472e-06, "loss": 0.14374741315841674, "step": 16330 }, { "epoch": 0.43042121813287676, "grad_norm": 1.6468734741210938, "learning_rate": 6.3530551585482315e-06, "loss": 0.15082697868347167, "step": 16340 }, { "epoch": 0.43068463381104866, "grad_norm": 2.0203092098236084, "learning_rate": 6.348948138329825e-06, "loss": 0.14464151859283447, "step": 16350 }, { "epoch": 0.43094804948922055, "grad_norm": 3.4143576622009277, "learning_rate": 6.344840136288763e-06, "loss": 0.15339642763137817, "step": 16360 }, { "epoch": 0.43121146516739245, "grad_norm": 3.501739025115967, "learning_rate": 6.340731155415031e-06, "loss": 0.14761340618133545, "step": 16370 }, { "epoch": 0.43147488084556435, "grad_norm": 1.4899625778198242, "learning_rate": 6.336621198699318e-06, "loss": 0.13236310482025146, "step": 16380 }, { "epoch": 0.43173829652373624, "grad_norm": 1.607964038848877, "learning_rate": 6.33251026913303e-06, "loss": 0.15171414613723755, "step": 16390 }, { "epoch": 0.43200171220190814, "grad_norm": 2.294689178466797, "learning_rate": 6.3283983697082785e-06, "loss": 0.14912980794906616, "step": 16400 }, { "epoch": 0.43226512788008004, "grad_norm": 2.390573263168335, "learning_rate": 6.324285503417882e-06, "loss": 0.1515910267829895, "step": 16410 }, { "epoch": 0.43252854355825193, "grad_norm": 1.6783506870269775, "learning_rate": 6.320171673255362e-06, "loss": 0.146044921875, "step": 16420 }, { "epoch": 0.43279195923642383, "grad_norm": 1.7116674184799194, "learning_rate": 6.316056882214942e-06, "loss": 0.14504778385162354, "step": 16430 }, { "epoch": 0.4330553749145957, "grad_norm": 4.2102580070495605, "learning_rate": 6.311941133291542e-06, "loss": 0.15100377798080444, "step": 16440 }, { "epoch": 0.43331879059276757, "grad_norm": 1.7697337865829468, "learning_rate": 6.307824429480782e-06, "loss": 0.1479719400405884, "step": 16450 }, { "epoch": 0.43358220627093946, "grad_norm": 2.188962697982788, "learning_rate": 6.30370677377898e-06, "loss": 0.1466852068901062, "step": 16460 }, { "epoch": 0.43384562194911136, "grad_norm": 1.3373472690582275, "learning_rate": 6.299588169183141e-06, "loss": 0.14600744247436523, "step": 16470 }, { "epoch": 0.43410903762728326, "grad_norm": 2.0224876403808594, "learning_rate": 6.295468618690963e-06, "loss": 0.14458153247833253, "step": 16480 }, { "epoch": 0.43437245330545515, "grad_norm": 1.5441287755966187, "learning_rate": 6.291348125300833e-06, "loss": 0.13869547843933105, "step": 16490 }, { "epoch": 0.43463586898362705, "grad_norm": 1.632568359375, "learning_rate": 6.287226692011825e-06, "loss": 0.1475994825363159, "step": 16500 }, { "epoch": 0.43489928466179895, "grad_norm": 2.012096405029297, "learning_rate": 6.283104321823695e-06, "loss": 0.15259721279144287, "step": 16510 }, { "epoch": 0.43516270033997084, "grad_norm": 2.142911911010742, "learning_rate": 6.27898101773688e-06, "loss": 0.14030208587646484, "step": 16520 }, { "epoch": 0.43542611601814274, "grad_norm": 4.793797969818115, "learning_rate": 6.274856782752501e-06, "loss": 0.13321223258972167, "step": 16530 }, { "epoch": 0.43568953169631464, "grad_norm": 2.147339344024658, "learning_rate": 6.270731619872355e-06, "loss": 0.14893636703491211, "step": 16540 }, { "epoch": 0.43595294737448653, "grad_norm": 2.2924280166625977, "learning_rate": 6.26660553209891e-06, "loss": 0.15431872606277466, "step": 16550 }, { "epoch": 0.43621636305265843, "grad_norm": 1.8759636878967285, "learning_rate": 6.2624785224353135e-06, "loss": 0.1450061559677124, "step": 16560 }, { "epoch": 0.4364797787308303, "grad_norm": 1.512276530265808, "learning_rate": 6.2583505938853806e-06, "loss": 0.13692736625671387, "step": 16570 }, { "epoch": 0.4367431944090022, "grad_norm": 4.236037254333496, "learning_rate": 6.254221749453595e-06, "loss": 0.1396812081336975, "step": 16580 }, { "epoch": 0.4370066100871741, "grad_norm": 4.1130266189575195, "learning_rate": 6.25009199214511e-06, "loss": 0.14769372940063477, "step": 16590 }, { "epoch": 0.437270025765346, "grad_norm": 1.3525081872940063, "learning_rate": 6.245961324965738e-06, "loss": 0.15028784275054932, "step": 16600 }, { "epoch": 0.4375334414435179, "grad_norm": 1.465043306350708, "learning_rate": 6.241829750921958e-06, "loss": 0.1548154830932617, "step": 16610 }, { "epoch": 0.4377968571216898, "grad_norm": 4.600656032562256, "learning_rate": 6.237697273020911e-06, "loss": 0.16282808780670166, "step": 16620 }, { "epoch": 0.4380602727998617, "grad_norm": 5.0683746337890625, "learning_rate": 6.233563894270387e-06, "loss": 0.15450737476348878, "step": 16630 }, { "epoch": 0.4383236884780336, "grad_norm": 1.5249533653259277, "learning_rate": 6.22942961767884e-06, "loss": 0.16193701028823854, "step": 16640 }, { "epoch": 0.4385871041562055, "grad_norm": 3.7332944869995117, "learning_rate": 6.225294446255378e-06, "loss": 0.1438573956489563, "step": 16650 }, { "epoch": 0.4388505198343774, "grad_norm": 2.3294336795806885, "learning_rate": 6.221158383009751e-06, "loss": 0.16028069257736205, "step": 16660 }, { "epoch": 0.4391139355125493, "grad_norm": 1.5141488313674927, "learning_rate": 6.217021430952372e-06, "loss": 0.15237128734588623, "step": 16670 }, { "epoch": 0.4393773511907212, "grad_norm": 2.379164218902588, "learning_rate": 6.2128835930942844e-06, "loss": 0.1542523145675659, "step": 16680 }, { "epoch": 0.4396407668688931, "grad_norm": 1.4815995693206787, "learning_rate": 6.208744872447191e-06, "loss": 0.1455615758895874, "step": 16690 }, { "epoch": 0.439904182547065, "grad_norm": 1.19709312915802, "learning_rate": 6.204605272023429e-06, "loss": 0.13376195430755616, "step": 16700 }, { "epoch": 0.4401675982252369, "grad_norm": 2.640594959259033, "learning_rate": 6.200464794835981e-06, "loss": 0.1509217143058777, "step": 16710 }, { "epoch": 0.4404310139034088, "grad_norm": 2.8308043479919434, "learning_rate": 6.196323443898462e-06, "loss": 0.15017290115356446, "step": 16720 }, { "epoch": 0.44069442958158067, "grad_norm": 2.3400871753692627, "learning_rate": 6.192181222225127e-06, "loss": 0.14604611396789552, "step": 16730 }, { "epoch": 0.44095784525975257, "grad_norm": 4.403182029724121, "learning_rate": 6.188038132830861e-06, "loss": 0.1482859492301941, "step": 16740 }, { "epoch": 0.44122126093792446, "grad_norm": 2.2023098468780518, "learning_rate": 6.183894178731188e-06, "loss": 0.14751477241516114, "step": 16750 }, { "epoch": 0.44148467661609636, "grad_norm": 3.595388650894165, "learning_rate": 6.179749362942254e-06, "loss": 0.14795805215835572, "step": 16760 }, { "epoch": 0.44174809229426826, "grad_norm": 3.4070022106170654, "learning_rate": 6.175603688480833e-06, "loss": 0.14023003578186036, "step": 16770 }, { "epoch": 0.44201150797244015, "grad_norm": 1.56428062915802, "learning_rate": 6.17145715836433e-06, "loss": 0.14628701210021972, "step": 16780 }, { "epoch": 0.44227492365061205, "grad_norm": 3.3190855979919434, "learning_rate": 6.167309775610766e-06, "loss": 0.14024125337600707, "step": 16790 }, { "epoch": 0.44253833932878395, "grad_norm": 3.161921501159668, "learning_rate": 6.163161543238785e-06, "loss": 0.15881772041320802, "step": 16800 }, { "epoch": 0.44280175500695584, "grad_norm": 3.6263015270233154, "learning_rate": 6.159012464267648e-06, "loss": 0.15821958780288697, "step": 16810 }, { "epoch": 0.44306517068512774, "grad_norm": 1.7532603740692139, "learning_rate": 6.154862541717238e-06, "loss": 0.13711202144622803, "step": 16820 }, { "epoch": 0.44332858636329964, "grad_norm": 2.201416254043579, "learning_rate": 6.150711778608044e-06, "loss": 0.14972487688064576, "step": 16830 }, { "epoch": 0.4435920020414715, "grad_norm": 1.5862149000167847, "learning_rate": 6.146560177961173e-06, "loss": 0.14887919425964355, "step": 16840 }, { "epoch": 0.4438554177196434, "grad_norm": 1.1372967958450317, "learning_rate": 6.1424077427983355e-06, "loss": 0.14093688726425171, "step": 16850 }, { "epoch": 0.44411883339781527, "grad_norm": 1.2772263288497925, "learning_rate": 6.138254476141856e-06, "loss": 0.14239020347595216, "step": 16860 }, { "epoch": 0.44438224907598717, "grad_norm": 1.4477242231369019, "learning_rate": 6.1341003810146595e-06, "loss": 0.13184576034545897, "step": 16870 }, { "epoch": 0.44464566475415906, "grad_norm": 1.2322765588760376, "learning_rate": 6.129945460440277e-06, "loss": 0.1486804485321045, "step": 16880 }, { "epoch": 0.44490908043233096, "grad_norm": 2.6681172847747803, "learning_rate": 6.1257897174428396e-06, "loss": 0.15069050788879396, "step": 16890 }, { "epoch": 0.44517249611050286, "grad_norm": 2.8424599170684814, "learning_rate": 6.121633155047073e-06, "loss": 0.1373998522758484, "step": 16900 }, { "epoch": 0.44543591178867475, "grad_norm": 2.953927755355835, "learning_rate": 6.117475776278307e-06, "loss": 0.13699796199798583, "step": 16910 }, { "epoch": 0.44569932746684665, "grad_norm": 2.1966259479522705, "learning_rate": 6.113317584162458e-06, "loss": 0.1405511498451233, "step": 16920 }, { "epoch": 0.44596274314501855, "grad_norm": 1.8140463829040527, "learning_rate": 6.109158581726039e-06, "loss": 0.15166935920715333, "step": 16930 }, { "epoch": 0.44622615882319044, "grad_norm": 3.0729331970214844, "learning_rate": 6.1049987719961504e-06, "loss": 0.14744925498962402, "step": 16940 }, { "epoch": 0.44648957450136234, "grad_norm": 1.700394868850708, "learning_rate": 6.100838158000482e-06, "loss": 0.1399103283882141, "step": 16950 }, { "epoch": 0.44675299017953424, "grad_norm": 1.7499452829360962, "learning_rate": 6.096676742767307e-06, "loss": 0.13812379837036132, "step": 16960 }, { "epoch": 0.44701640585770613, "grad_norm": 2.354326009750366, "learning_rate": 6.092514529325486e-06, "loss": 0.1482361674308777, "step": 16970 }, { "epoch": 0.44727982153587803, "grad_norm": 1.9238204956054688, "learning_rate": 6.088351520704454e-06, "loss": 0.1543351650238037, "step": 16980 }, { "epoch": 0.4475432372140499, "grad_norm": 2.1748263835906982, "learning_rate": 6.084187719934232e-06, "loss": 0.16594841480255126, "step": 16990 }, { "epoch": 0.4478066528922218, "grad_norm": 1.4593000411987305, "learning_rate": 6.080023130045408e-06, "loss": 0.15026239156723023, "step": 17000 }, { "epoch": 0.4480700685703937, "grad_norm": 3.1289052963256836, "learning_rate": 6.075857754069156e-06, "loss": 0.13961890935897828, "step": 17010 }, { "epoch": 0.4483334842485656, "grad_norm": 3.1836490631103516, "learning_rate": 6.071691595037214e-06, "loss": 0.1485459804534912, "step": 17020 }, { "epoch": 0.4485968999267375, "grad_norm": 1.7854381799697876, "learning_rate": 6.067524655981893e-06, "loss": 0.14502793550491333, "step": 17030 }, { "epoch": 0.4488603156049094, "grad_norm": 2.0681543350219727, "learning_rate": 6.063356939936069e-06, "loss": 0.1414552688598633, "step": 17040 }, { "epoch": 0.4491237312830813, "grad_norm": 2.6014842987060547, "learning_rate": 6.0591884499331875e-06, "loss": 0.1388018846511841, "step": 17050 }, { "epoch": 0.4493871469612532, "grad_norm": 3.123791217803955, "learning_rate": 6.055019189007256e-06, "loss": 0.15301136970520018, "step": 17060 }, { "epoch": 0.4496505626394251, "grad_norm": 1.8580714464187622, "learning_rate": 6.05084916019284e-06, "loss": 0.13559948205947875, "step": 17070 }, { "epoch": 0.449913978317597, "grad_norm": 2.891113519668579, "learning_rate": 6.046678366525069e-06, "loss": 0.1582431674003601, "step": 17080 }, { "epoch": 0.4501773939957689, "grad_norm": 1.8693711757659912, "learning_rate": 6.042506811039624e-06, "loss": 0.14409924745559693, "step": 17090 }, { "epoch": 0.4504408096739408, "grad_norm": 4.499271392822266, "learning_rate": 6.038334496772746e-06, "loss": 0.1309260368347168, "step": 17100 }, { "epoch": 0.4507042253521127, "grad_norm": 1.5088309049606323, "learning_rate": 6.03416142676122e-06, "loss": 0.13293330669403075, "step": 17110 }, { "epoch": 0.4509676410302846, "grad_norm": 1.744957685470581, "learning_rate": 6.029987604042393e-06, "loss": 0.14559526443481446, "step": 17120 }, { "epoch": 0.4512310567084565, "grad_norm": 4.35652494430542, "learning_rate": 6.0258130316541495e-06, "loss": 0.1503550887107849, "step": 17130 }, { "epoch": 0.4514944723866284, "grad_norm": 3.759652853012085, "learning_rate": 6.021637712634927e-06, "loss": 0.1369057297706604, "step": 17140 }, { "epoch": 0.4517578880648003, "grad_norm": 3.461033821105957, "learning_rate": 6.017461650023696e-06, "loss": 0.14631725549697877, "step": 17150 }, { "epoch": 0.45202130374297217, "grad_norm": 1.0745638608932495, "learning_rate": 6.0132848468599815e-06, "loss": 0.1537693738937378, "step": 17160 }, { "epoch": 0.45228471942114407, "grad_norm": 1.8844915628433228, "learning_rate": 6.0091073061838365e-06, "loss": 0.14988033771514891, "step": 17170 }, { "epoch": 0.45254813509931596, "grad_norm": 2.2800543308258057, "learning_rate": 6.00492903103586e-06, "loss": 0.16232092380523683, "step": 17180 }, { "epoch": 0.45281155077748786, "grad_norm": 2.3637845516204834, "learning_rate": 6.000750024457177e-06, "loss": 0.16100505590438843, "step": 17190 }, { "epoch": 0.45307496645565976, "grad_norm": 1.3831192255020142, "learning_rate": 5.996570289489449e-06, "loss": 0.15988516807556152, "step": 17200 }, { "epoch": 0.45333838213383165, "grad_norm": 1.3034965991973877, "learning_rate": 5.9923898291748715e-06, "loss": 0.15907716751098633, "step": 17210 }, { "epoch": 0.45360179781200355, "grad_norm": 2.1796045303344727, "learning_rate": 5.988208646556159e-06, "loss": 0.13732213973999025, "step": 17220 }, { "epoch": 0.45386521349017545, "grad_norm": 2.1247997283935547, "learning_rate": 5.984026744676561e-06, "loss": 0.14329245090484619, "step": 17230 }, { "epoch": 0.4541286291683473, "grad_norm": 1.5014820098876953, "learning_rate": 5.97984412657984e-06, "loss": 0.14172859191894532, "step": 17240 }, { "epoch": 0.4543920448465192, "grad_norm": 4.303319931030273, "learning_rate": 5.9756607953102915e-06, "loss": 0.13495737314224243, "step": 17250 }, { "epoch": 0.4546554605246911, "grad_norm": 1.563039779663086, "learning_rate": 5.971476753912723e-06, "loss": 0.1518675923347473, "step": 17260 }, { "epoch": 0.454918876202863, "grad_norm": 1.5586975812911987, "learning_rate": 5.9672920054324604e-06, "loss": 0.14079636335372925, "step": 17270 }, { "epoch": 0.4551822918810349, "grad_norm": 2.8090996742248535, "learning_rate": 5.963106552915342e-06, "loss": 0.14094760417938232, "step": 17280 }, { "epoch": 0.45544570755920677, "grad_norm": 2.1591086387634277, "learning_rate": 5.958920399407724e-06, "loss": 0.14761432409286498, "step": 17290 }, { "epoch": 0.45570912323737867, "grad_norm": 1.748212456703186, "learning_rate": 5.954733547956466e-06, "loss": 0.14390639066696168, "step": 17300 }, { "epoch": 0.45597253891555056, "grad_norm": 2.6911537647247314, "learning_rate": 5.9505460016089425e-06, "loss": 0.15332934856414795, "step": 17310 }, { "epoch": 0.45623595459372246, "grad_norm": 1.2271353006362915, "learning_rate": 5.946357763413029e-06, "loss": 0.14174140691757203, "step": 17320 }, { "epoch": 0.45649937027189436, "grad_norm": 3.126795530319214, "learning_rate": 5.942168836417106e-06, "loss": 0.1595513939857483, "step": 17330 }, { "epoch": 0.45676278595006625, "grad_norm": 2.2719435691833496, "learning_rate": 5.937979223670054e-06, "loss": 0.14815218448638917, "step": 17340 }, { "epoch": 0.45702620162823815, "grad_norm": 2.3176655769348145, "learning_rate": 5.933788928221257e-06, "loss": 0.1456812620162964, "step": 17350 }, { "epoch": 0.45728961730641005, "grad_norm": 2.743713855743408, "learning_rate": 5.929597953120591e-06, "loss": 0.15763320922851562, "step": 17360 }, { "epoch": 0.45755303298458194, "grad_norm": 1.7211270332336426, "learning_rate": 5.925406301418426e-06, "loss": 0.14592043161392212, "step": 17370 }, { "epoch": 0.45781644866275384, "grad_norm": 1.301701545715332, "learning_rate": 5.921213976165631e-06, "loss": 0.1297528028488159, "step": 17380 }, { "epoch": 0.45807986434092574, "grad_norm": 1.6062417030334473, "learning_rate": 5.91702098041356e-06, "loss": 0.15597076416015626, "step": 17390 }, { "epoch": 0.45834328001909763, "grad_norm": 2.206861972808838, "learning_rate": 5.912827317214056e-06, "loss": 0.1528489351272583, "step": 17400 }, { "epoch": 0.45860669569726953, "grad_norm": 3.691978693008423, "learning_rate": 5.908632989619446e-06, "loss": 0.15051462650299072, "step": 17410 }, { "epoch": 0.4588701113754414, "grad_norm": 2.1743180751800537, "learning_rate": 5.904438000682546e-06, "loss": 0.15018620491027831, "step": 17420 }, { "epoch": 0.4591335270536133, "grad_norm": 2.5640792846679688, "learning_rate": 5.900242353456647e-06, "loss": 0.15078413486480713, "step": 17430 }, { "epoch": 0.4593969427317852, "grad_norm": 2.214796543121338, "learning_rate": 5.896046050995526e-06, "loss": 0.15288491249084474, "step": 17440 }, { "epoch": 0.4596603584099571, "grad_norm": 2.418914318084717, "learning_rate": 5.891849096353428e-06, "loss": 0.13953371047973634, "step": 17450 }, { "epoch": 0.459923774088129, "grad_norm": 3.8722994327545166, "learning_rate": 5.887651492585081e-06, "loss": 0.14007656574249266, "step": 17460 }, { "epoch": 0.4601871897663009, "grad_norm": 1.1646711826324463, "learning_rate": 5.8834532427456805e-06, "loss": 0.13613628149032592, "step": 17470 }, { "epoch": 0.4604506054444728, "grad_norm": 1.5274829864501953, "learning_rate": 5.879254349890895e-06, "loss": 0.14948222637176514, "step": 17480 }, { "epoch": 0.4607140211226447, "grad_norm": 1.9735231399536133, "learning_rate": 5.8750548170768595e-06, "loss": 0.14253268241882325, "step": 17490 }, { "epoch": 0.4609774368008166, "grad_norm": 1.9856270551681519, "learning_rate": 5.870854647360173e-06, "loss": 0.1552174210548401, "step": 17500 }, { "epoch": 0.4612408524789885, "grad_norm": 1.5222209692001343, "learning_rate": 5.866653843797905e-06, "loss": 0.16193435192108155, "step": 17510 }, { "epoch": 0.4615042681571604, "grad_norm": 2.2559616565704346, "learning_rate": 5.862452409447579e-06, "loss": 0.15969789028167725, "step": 17520 }, { "epoch": 0.4617676838353323, "grad_norm": 2.411724328994751, "learning_rate": 5.858250347367179e-06, "loss": 0.14121036529541015, "step": 17530 }, { "epoch": 0.4620310995135042, "grad_norm": 1.4517048597335815, "learning_rate": 5.854047660615147e-06, "loss": 0.14823817014694213, "step": 17540 }, { "epoch": 0.4622945151916761, "grad_norm": 2.12085223197937, "learning_rate": 5.849844352250382e-06, "loss": 0.14436326026916504, "step": 17550 }, { "epoch": 0.462557930869848, "grad_norm": 2.079655885696411, "learning_rate": 5.845640425332232e-06, "loss": 0.15419869422912597, "step": 17560 }, { "epoch": 0.4628213465480199, "grad_norm": 3.3567464351654053, "learning_rate": 5.841435882920498e-06, "loss": 0.15868698358535765, "step": 17570 }, { "epoch": 0.46308476222619177, "grad_norm": 3.8736555576324463, "learning_rate": 5.837230728075423e-06, "loss": 0.14700710773468018, "step": 17580 }, { "epoch": 0.46334817790436367, "grad_norm": 2.1360180377960205, "learning_rate": 5.833024963857704e-06, "loss": 0.14285759925842284, "step": 17590 }, { "epoch": 0.46361159358253556, "grad_norm": 3.92887806892395, "learning_rate": 5.828818593328476e-06, "loss": 0.13460931777954102, "step": 17600 }, { "epoch": 0.46387500926070746, "grad_norm": 2.399545431137085, "learning_rate": 5.8246116195493195e-06, "loss": 0.16416647434234619, "step": 17610 }, { "epoch": 0.46413842493887936, "grad_norm": 2.7334015369415283, "learning_rate": 5.8204040455822474e-06, "loss": 0.15301427841186524, "step": 17620 }, { "epoch": 0.46440184061705125, "grad_norm": 4.20542049407959, "learning_rate": 5.8161958744897176e-06, "loss": 0.1365823984146118, "step": 17630 }, { "epoch": 0.4646652562952231, "grad_norm": 3.279937744140625, "learning_rate": 5.811987109334615e-06, "loss": 0.15683703422546386, "step": 17640 }, { "epoch": 0.464928671973395, "grad_norm": 1.187078595161438, "learning_rate": 5.807777753180263e-06, "loss": 0.14532148838043213, "step": 17650 }, { "epoch": 0.4651920876515669, "grad_norm": 1.747628927230835, "learning_rate": 5.803567809090412e-06, "loss": 0.14478319883346558, "step": 17660 }, { "epoch": 0.4654555033297388, "grad_norm": 1.464833378791809, "learning_rate": 5.7993572801292385e-06, "loss": 0.13789682388305663, "step": 17670 }, { "epoch": 0.4657189190079107, "grad_norm": 2.057285785675049, "learning_rate": 5.795146169361351e-06, "loss": 0.13705368041992189, "step": 17680 }, { "epoch": 0.4659823346860826, "grad_norm": 5.425158500671387, "learning_rate": 5.7909344798517765e-06, "loss": 0.14337252378463744, "step": 17690 }, { "epoch": 0.4662457503642545, "grad_norm": 3.1683456897735596, "learning_rate": 5.786722214665963e-06, "loss": 0.145883309841156, "step": 17700 }, { "epoch": 0.46650916604242637, "grad_norm": 1.4125372171401978, "learning_rate": 5.782509376869779e-06, "loss": 0.1423474907875061, "step": 17710 }, { "epoch": 0.46677258172059827, "grad_norm": 2.0376851558685303, "learning_rate": 5.778295969529512e-06, "loss": 0.138421893119812, "step": 17720 }, { "epoch": 0.46703599739877016, "grad_norm": 1.239922285079956, "learning_rate": 5.774081995711858e-06, "loss": 0.1376358985900879, "step": 17730 }, { "epoch": 0.46729941307694206, "grad_norm": 2.731863498687744, "learning_rate": 5.769867458483932e-06, "loss": 0.14605828523635864, "step": 17740 }, { "epoch": 0.46756282875511396, "grad_norm": 2.3121206760406494, "learning_rate": 5.765652360913253e-06, "loss": 0.14343745708465577, "step": 17750 }, { "epoch": 0.46782624443328585, "grad_norm": 1.6062726974487305, "learning_rate": 5.7614367060677546e-06, "loss": 0.14877159595489503, "step": 17760 }, { "epoch": 0.46808966011145775, "grad_norm": 1.365917682647705, "learning_rate": 5.757220497015768e-06, "loss": 0.14833760261535645, "step": 17770 }, { "epoch": 0.46835307578962965, "grad_norm": 1.1974480152130127, "learning_rate": 5.753003736826037e-06, "loss": 0.15707905292510987, "step": 17780 }, { "epoch": 0.46861649146780154, "grad_norm": 2.3657050132751465, "learning_rate": 5.748786428567699e-06, "loss": 0.1441969871520996, "step": 17790 }, { "epoch": 0.46887990714597344, "grad_norm": 2.471867084503174, "learning_rate": 5.744568575310291e-06, "loss": 0.15078890323638916, "step": 17800 }, { "epoch": 0.46914332282414534, "grad_norm": 1.7599260807037354, "learning_rate": 5.740350180123752e-06, "loss": 0.15212721824645997, "step": 17810 }, { "epoch": 0.46940673850231723, "grad_norm": 1.6215797662734985, "learning_rate": 5.73613124607841e-06, "loss": 0.12542831897735596, "step": 17820 }, { "epoch": 0.46967015418048913, "grad_norm": 2.3267533779144287, "learning_rate": 5.731911776244989e-06, "loss": 0.15488011837005616, "step": 17830 }, { "epoch": 0.469933569858661, "grad_norm": 2.2566370964050293, "learning_rate": 5.727691773694597e-06, "loss": 0.1460500478744507, "step": 17840 }, { "epoch": 0.4701969855368329, "grad_norm": 2.5930662155151367, "learning_rate": 5.7234712414987395e-06, "loss": 0.1439279317855835, "step": 17850 }, { "epoch": 0.4704604012150048, "grad_norm": 3.1820244789123535, "learning_rate": 5.719250182729297e-06, "loss": 0.15782448053359985, "step": 17860 }, { "epoch": 0.4707238168931767, "grad_norm": 2.3682971000671387, "learning_rate": 5.715028600458542e-06, "loss": 0.1472579836845398, "step": 17870 }, { "epoch": 0.4709872325713486, "grad_norm": 2.659548044204712, "learning_rate": 5.710806497759122e-06, "loss": 0.1312854766845703, "step": 17880 }, { "epoch": 0.4712506482495205, "grad_norm": 1.897870421409607, "learning_rate": 5.706583877704066e-06, "loss": 0.13122586011886597, "step": 17890 }, { "epoch": 0.4715140639276924, "grad_norm": 1.1603546142578125, "learning_rate": 5.702360743366779e-06, "loss": 0.15601255893707275, "step": 17900 }, { "epoch": 0.4717774796058643, "grad_norm": 3.8000757694244385, "learning_rate": 5.6981370978210425e-06, "loss": 0.14122170209884644, "step": 17910 }, { "epoch": 0.4720408952840362, "grad_norm": 1.5430359840393066, "learning_rate": 5.693912944141003e-06, "loss": 0.1375321626663208, "step": 17920 }, { "epoch": 0.4723043109622081, "grad_norm": 1.234584927558899, "learning_rate": 5.689688285401187e-06, "loss": 0.14089139699935913, "step": 17930 }, { "epoch": 0.47256772664038, "grad_norm": 2.102989435195923, "learning_rate": 5.68546312467648e-06, "loss": 0.14093201160430907, "step": 17940 }, { "epoch": 0.4728311423185519, "grad_norm": 2.1104815006256104, "learning_rate": 5.681237465042139e-06, "loss": 0.1480846881866455, "step": 17950 }, { "epoch": 0.4730945579967238, "grad_norm": 1.9918612241744995, "learning_rate": 5.677011309573779e-06, "loss": 0.14128050804138184, "step": 17960 }, { "epoch": 0.4733579736748957, "grad_norm": 1.885278344154358, "learning_rate": 5.672784661347378e-06, "loss": 0.14521405696868897, "step": 17970 }, { "epoch": 0.4736213893530676, "grad_norm": 8.04674243927002, "learning_rate": 5.668557523439275e-06, "loss": 0.13526253700256347, "step": 17980 }, { "epoch": 0.4738848050312395, "grad_norm": 2.037785530090332, "learning_rate": 5.664329898926163e-06, "loss": 0.16028138399124145, "step": 17990 }, { "epoch": 0.47414822070941137, "grad_norm": 4.678908824920654, "learning_rate": 5.660101790885089e-06, "loss": 0.14716428518295288, "step": 18000 }, { "epoch": 0.47441163638758327, "grad_norm": 1.907875657081604, "learning_rate": 5.655873202393451e-06, "loss": 0.14808976650238037, "step": 18010 }, { "epoch": 0.47467505206575517, "grad_norm": 2.9079415798187256, "learning_rate": 5.651644136529001e-06, "loss": 0.14537472724914552, "step": 18020 }, { "epoch": 0.474938467743927, "grad_norm": 2.905191659927368, "learning_rate": 5.647414596369833e-06, "loss": 0.14657440185546874, "step": 18030 }, { "epoch": 0.4752018834220989, "grad_norm": 1.4559390544891357, "learning_rate": 5.643184584994391e-06, "loss": 0.14194376468658448, "step": 18040 }, { "epoch": 0.4754652991002708, "grad_norm": 3.440054416656494, "learning_rate": 5.638954105481458e-06, "loss": 0.1426222324371338, "step": 18050 }, { "epoch": 0.4757287147784427, "grad_norm": 2.8800995349884033, "learning_rate": 5.63472316091016e-06, "loss": 0.1491086959838867, "step": 18060 }, { "epoch": 0.4759921304566146, "grad_norm": 3.287726640701294, "learning_rate": 5.630491754359961e-06, "loss": 0.13020648956298828, "step": 18070 }, { "epoch": 0.4762555461347865, "grad_norm": 2.0814006328582764, "learning_rate": 5.626259888910662e-06, "loss": 0.14881324768066406, "step": 18080 }, { "epoch": 0.4765189618129584, "grad_norm": 1.9755629301071167, "learning_rate": 5.622027567642398e-06, "loss": 0.14008885622024536, "step": 18090 }, { "epoch": 0.4767823774911303, "grad_norm": 1.716796875, "learning_rate": 5.61779479363563e-06, "loss": 0.14128826856613158, "step": 18100 }, { "epoch": 0.4770457931693022, "grad_norm": 4.101381301879883, "learning_rate": 5.613561569971161e-06, "loss": 0.1409127950668335, "step": 18110 }, { "epoch": 0.4773092088474741, "grad_norm": 1.850378155708313, "learning_rate": 5.609327899730108e-06, "loss": 0.13625950813293458, "step": 18120 }, { "epoch": 0.477572624525646, "grad_norm": 3.5068037509918213, "learning_rate": 5.6050937859939226e-06, "loss": 0.13477089405059814, "step": 18130 }, { "epoch": 0.47783604020381787, "grad_norm": 2.874803066253662, "learning_rate": 5.6008592318443725e-06, "loss": 0.13625428676605225, "step": 18140 }, { "epoch": 0.47809945588198977, "grad_norm": 2.504413366317749, "learning_rate": 5.5966242403635505e-06, "loss": 0.14537959098815917, "step": 18150 }, { "epoch": 0.47836287156016166, "grad_norm": 1.3291864395141602, "learning_rate": 5.592388814633865e-06, "loss": 0.14074037075042725, "step": 18160 }, { "epoch": 0.47862628723833356, "grad_norm": 1.454416036605835, "learning_rate": 5.588152957738045e-06, "loss": 0.1391150712966919, "step": 18170 }, { "epoch": 0.47888970291650546, "grad_norm": 2.860384702682495, "learning_rate": 5.583916672759124e-06, "loss": 0.12910842895507812, "step": 18180 }, { "epoch": 0.47915311859467735, "grad_norm": 1.5538500547409058, "learning_rate": 5.579679962780459e-06, "loss": 0.14271175861358643, "step": 18190 }, { "epoch": 0.47941653427284925, "grad_norm": 1.5497580766677856, "learning_rate": 5.575442830885705e-06, "loss": 0.13282084465026855, "step": 18200 }, { "epoch": 0.47967994995102115, "grad_norm": 1.2355475425720215, "learning_rate": 5.5712052801588346e-06, "loss": 0.14341599941253663, "step": 18210 }, { "epoch": 0.47994336562919304, "grad_norm": 2.8838236331939697, "learning_rate": 5.566967313684115e-06, "loss": 0.1549671173095703, "step": 18220 }, { "epoch": 0.48020678130736494, "grad_norm": 2.899287700653076, "learning_rate": 5.562728934546126e-06, "loss": 0.1343597412109375, "step": 18230 }, { "epoch": 0.48047019698553683, "grad_norm": 2.998612642288208, "learning_rate": 5.558490145829737e-06, "loss": 0.1454639434814453, "step": 18240 }, { "epoch": 0.48073361266370873, "grad_norm": 2.643777370452881, "learning_rate": 5.554250950620127e-06, "loss": 0.14366319179534912, "step": 18250 }, { "epoch": 0.48099702834188063, "grad_norm": 1.9688695669174194, "learning_rate": 5.550011352002762e-06, "loss": 0.14371049404144287, "step": 18260 }, { "epoch": 0.4812604440200525, "grad_norm": 2.898845911026001, "learning_rate": 5.545771353063405e-06, "loss": 0.1392603874206543, "step": 18270 }, { "epoch": 0.4815238596982244, "grad_norm": 1.4264365434646606, "learning_rate": 5.541530956888112e-06, "loss": 0.14267749786376954, "step": 18280 }, { "epoch": 0.4817872753763963, "grad_norm": 3.086918592453003, "learning_rate": 5.537290166563224e-06, "loss": 0.1374131202697754, "step": 18290 }, { "epoch": 0.4820506910545682, "grad_norm": 3.1176393032073975, "learning_rate": 5.5330489851753734e-06, "loss": 0.14096349477767944, "step": 18300 }, { "epoch": 0.4823141067327401, "grad_norm": 1.7948700189590454, "learning_rate": 5.528807415811471e-06, "loss": 0.1274722099304199, "step": 18310 }, { "epoch": 0.482577522410912, "grad_norm": 2.540515184402466, "learning_rate": 5.524565461558719e-06, "loss": 0.14160608053207396, "step": 18320 }, { "epoch": 0.4828409380890839, "grad_norm": 2.529797077178955, "learning_rate": 5.5203231255045905e-06, "loss": 0.14976648092269898, "step": 18330 }, { "epoch": 0.4831043537672558, "grad_norm": 2.805732011795044, "learning_rate": 5.516080410736846e-06, "loss": 0.14173274040222167, "step": 18340 }, { "epoch": 0.4833677694454277, "grad_norm": 3.5098202228546143, "learning_rate": 5.511837320343509e-06, "loss": 0.15078202486038209, "step": 18350 }, { "epoch": 0.4836311851235996, "grad_norm": 3.2757208347320557, "learning_rate": 5.507593857412891e-06, "loss": 0.1657733678817749, "step": 18360 }, { "epoch": 0.4838946008017715, "grad_norm": 2.2345352172851562, "learning_rate": 5.503350025033561e-06, "loss": 0.14100033044815063, "step": 18370 }, { "epoch": 0.4841580164799434, "grad_norm": 2.3748531341552734, "learning_rate": 5.499105826294369e-06, "loss": 0.13849987983703613, "step": 18380 }, { "epoch": 0.4844214321581153, "grad_norm": 3.5841994285583496, "learning_rate": 5.4948612642844224e-06, "loss": 0.14649826288223267, "step": 18390 }, { "epoch": 0.4846848478362872, "grad_norm": 3.814359188079834, "learning_rate": 5.4906163420930955e-06, "loss": 0.143965482711792, "step": 18400 }, { "epoch": 0.4849482635144591, "grad_norm": 1.3691574335098267, "learning_rate": 5.486371062810031e-06, "loss": 0.12757693529129027, "step": 18410 }, { "epoch": 0.485211679192631, "grad_norm": 3.84836483001709, "learning_rate": 5.4821254295251225e-06, "loss": 0.14332720041275024, "step": 18420 }, { "epoch": 0.4854750948708028, "grad_norm": 1.5705739259719849, "learning_rate": 5.4778794453285255e-06, "loss": 0.1420516014099121, "step": 18430 }, { "epoch": 0.4857385105489747, "grad_norm": 1.5531708002090454, "learning_rate": 5.473633113310649e-06, "loss": 0.13043406009674072, "step": 18440 }, { "epoch": 0.4860019262271466, "grad_norm": 3.9959728717803955, "learning_rate": 5.4693864365621605e-06, "loss": 0.14591087102890016, "step": 18450 }, { "epoch": 0.4862653419053185, "grad_norm": 1.655192255973816, "learning_rate": 5.465139418173971e-06, "loss": 0.14264202117919922, "step": 18460 }, { "epoch": 0.4865287575834904, "grad_norm": 1.8850191831588745, "learning_rate": 5.4608920612372476e-06, "loss": 0.14135334491729737, "step": 18470 }, { "epoch": 0.4867921732616623, "grad_norm": 1.114485502243042, "learning_rate": 5.4566443688433945e-06, "loss": 0.13029305934906005, "step": 18480 }, { "epoch": 0.4870555889398342, "grad_norm": 3.5821821689605713, "learning_rate": 5.45239634408407e-06, "loss": 0.1357923150062561, "step": 18490 }, { "epoch": 0.4873190046180061, "grad_norm": 3.192997455596924, "learning_rate": 5.448147990051167e-06, "loss": 0.1471573829650879, "step": 18500 }, { "epoch": 0.487582420296178, "grad_norm": 2.5217995643615723, "learning_rate": 5.443899309836824e-06, "loss": 0.12886770963668823, "step": 18510 }, { "epoch": 0.4878458359743499, "grad_norm": 1.7177528142929077, "learning_rate": 5.43965030653341e-06, "loss": 0.1419088363647461, "step": 18520 }, { "epoch": 0.4881092516525218, "grad_norm": 1.7948020696640015, "learning_rate": 5.435400983233537e-06, "loss": 0.14795236587524413, "step": 18530 }, { "epoch": 0.4883726673306937, "grad_norm": 1.1748770475387573, "learning_rate": 5.431151343030041e-06, "loss": 0.1480305314064026, "step": 18540 }, { "epoch": 0.4886360830088656, "grad_norm": 2.444002628326416, "learning_rate": 5.426901389015997e-06, "loss": 0.15152437686920167, "step": 18550 }, { "epoch": 0.48889949868703747, "grad_norm": 1.5251662731170654, "learning_rate": 5.4226511242847055e-06, "loss": 0.13584296703338622, "step": 18560 }, { "epoch": 0.48916291436520937, "grad_norm": 2.1006367206573486, "learning_rate": 5.418400551929689e-06, "loss": 0.1370074510574341, "step": 18570 }, { "epoch": 0.48942633004338126, "grad_norm": 4.213191986083984, "learning_rate": 5.414149675044701e-06, "loss": 0.131225061416626, "step": 18580 }, { "epoch": 0.48968974572155316, "grad_norm": 2.970073699951172, "learning_rate": 5.409898496723711e-06, "loss": 0.14706742763519287, "step": 18590 }, { "epoch": 0.48995316139972506, "grad_norm": 0.9901520609855652, "learning_rate": 5.40564702006091e-06, "loss": 0.14622113704681397, "step": 18600 }, { "epoch": 0.49021657707789695, "grad_norm": 2.7229409217834473, "learning_rate": 5.401395248150707e-06, "loss": 0.14392144680023194, "step": 18610 }, { "epoch": 0.49047999275606885, "grad_norm": 1.3640849590301514, "learning_rate": 5.397143184087725e-06, "loss": 0.1460471749305725, "step": 18620 }, { "epoch": 0.49074340843424075, "grad_norm": 3.0282156467437744, "learning_rate": 5.392890830966798e-06, "loss": 0.1353626847267151, "step": 18630 }, { "epoch": 0.49100682411241264, "grad_norm": 1.5364916324615479, "learning_rate": 5.3886381918829765e-06, "loss": 0.13824858665466308, "step": 18640 }, { "epoch": 0.49127023979058454, "grad_norm": 1.4559924602508545, "learning_rate": 5.384385269931509e-06, "loss": 0.13934930562973022, "step": 18650 }, { "epoch": 0.49153365546875644, "grad_norm": 1.9785271883010864, "learning_rate": 5.38013206820786e-06, "loss": 0.1455094575881958, "step": 18660 }, { "epoch": 0.49179707114692833, "grad_norm": 1.7244083881378174, "learning_rate": 5.375878589807689e-06, "loss": 0.15018974542617797, "step": 18670 }, { "epoch": 0.49206048682510023, "grad_norm": 2.561774730682373, "learning_rate": 5.3716248378268665e-06, "loss": 0.12747342586517335, "step": 18680 }, { "epoch": 0.4923239025032721, "grad_norm": 1.7682346105575562, "learning_rate": 5.367370815361452e-06, "loss": 0.15260062217712403, "step": 18690 }, { "epoch": 0.492587318181444, "grad_norm": 2.114384174346924, "learning_rate": 5.363116525507707e-06, "loss": 0.15419647693634034, "step": 18700 }, { "epoch": 0.4928507338596159, "grad_norm": 2.634662389755249, "learning_rate": 5.358861971362091e-06, "loss": 0.14291844367980958, "step": 18710 }, { "epoch": 0.4931141495377878, "grad_norm": 2.3363022804260254, "learning_rate": 5.35460715602125e-06, "loss": 0.13601624965667725, "step": 18720 }, { "epoch": 0.4933775652159597, "grad_norm": 1.6400359869003296, "learning_rate": 5.350352082582021e-06, "loss": 0.14293333292007446, "step": 18730 }, { "epoch": 0.4936409808941316, "grad_norm": 1.2720519304275513, "learning_rate": 5.34609675414143e-06, "loss": 0.13254144191741943, "step": 18740 }, { "epoch": 0.4939043965723035, "grad_norm": 1.5769433975219727, "learning_rate": 5.34184117379669e-06, "loss": 0.13499858379364013, "step": 18750 }, { "epoch": 0.4941678122504754, "grad_norm": 1.394892930984497, "learning_rate": 5.3375853446451966e-06, "loss": 0.12642416954040528, "step": 18760 }, { "epoch": 0.4944312279286473, "grad_norm": 2.3348660469055176, "learning_rate": 5.333329269784524e-06, "loss": 0.14674112796783448, "step": 18770 }, { "epoch": 0.4946946436068192, "grad_norm": 1.9982770681381226, "learning_rate": 5.329072952312426e-06, "loss": 0.13160502910614014, "step": 18780 }, { "epoch": 0.4949580592849911, "grad_norm": 2.455533266067505, "learning_rate": 5.324816395326838e-06, "loss": 0.12283289432525635, "step": 18790 }, { "epoch": 0.495221474963163, "grad_norm": 2.6831514835357666, "learning_rate": 5.320559601925862e-06, "loss": 0.1350919008255005, "step": 18800 }, { "epoch": 0.4954848906413349, "grad_norm": 1.3435512781143188, "learning_rate": 5.3163025752077805e-06, "loss": 0.14073636531829833, "step": 18810 }, { "epoch": 0.4957483063195068, "grad_norm": 1.3339688777923584, "learning_rate": 5.312045318271035e-06, "loss": 0.13640906810760497, "step": 18820 }, { "epoch": 0.4960117219976786, "grad_norm": 1.2051626443862915, "learning_rate": 5.307787834214247e-06, "loss": 0.14622626304626465, "step": 18830 }, { "epoch": 0.4962751376758505, "grad_norm": 2.569983959197998, "learning_rate": 5.303530126136191e-06, "loss": 0.14718589782714844, "step": 18840 }, { "epoch": 0.4965385533540224, "grad_norm": 2.6273882389068604, "learning_rate": 5.299272197135817e-06, "loss": 0.1467883586883545, "step": 18850 }, { "epoch": 0.4968019690321943, "grad_norm": 2.9425160884857178, "learning_rate": 5.295014050312224e-06, "loss": 0.128196382522583, "step": 18860 }, { "epoch": 0.4970653847103662, "grad_norm": 3.3877885341644287, "learning_rate": 5.290755688764677e-06, "loss": 0.13856801986694336, "step": 18870 }, { "epoch": 0.4973288003885381, "grad_norm": 2.01535701751709, "learning_rate": 5.286497115592597e-06, "loss": 0.1317427158355713, "step": 18880 }, { "epoch": 0.49759221606671, "grad_norm": 2.8944292068481445, "learning_rate": 5.282238333895555e-06, "loss": 0.14184650182723998, "step": 18890 }, { "epoch": 0.4978556317448819, "grad_norm": 1.8495569229125977, "learning_rate": 5.277979346773276e-06, "loss": 0.13471646308898927, "step": 18900 }, { "epoch": 0.4981190474230538, "grad_norm": 1.960176706314087, "learning_rate": 5.273720157325637e-06, "loss": 0.14029712677001954, "step": 18910 }, { "epoch": 0.4983824631012257, "grad_norm": 2.3795981407165527, "learning_rate": 5.269460768652658e-06, "loss": 0.1358123779296875, "step": 18920 }, { "epoch": 0.4986458787793976, "grad_norm": 1.7630113363265991, "learning_rate": 5.2652011838545035e-06, "loss": 0.1402115821838379, "step": 18930 }, { "epoch": 0.4989092944575695, "grad_norm": 2.8990559577941895, "learning_rate": 5.260941406031489e-06, "loss": 0.13851475715637207, "step": 18940 }, { "epoch": 0.4991727101357414, "grad_norm": 3.7432711124420166, "learning_rate": 5.256681438284059e-06, "loss": 0.14531610012054444, "step": 18950 }, { "epoch": 0.4994361258139133, "grad_norm": 1.6401911973953247, "learning_rate": 5.2524212837128054e-06, "loss": 0.1396859645843506, "step": 18960 }, { "epoch": 0.4996995414920852, "grad_norm": 2.5443549156188965, "learning_rate": 5.2481609454184505e-06, "loss": 0.1338839888572693, "step": 18970 }, { "epoch": 0.49996295717025707, "grad_norm": 1.7639614343643188, "learning_rate": 5.243900426501853e-06, "loss": 0.14507176876068115, "step": 18980 }, { "epoch": 0.500226372848429, "grad_norm": 2.229494333267212, "learning_rate": 5.239639730064002e-06, "loss": 0.12940523624420167, "step": 18990 }, { "epoch": 0.5004897885266009, "grad_norm": 2.17891001701355, "learning_rate": 5.2353788592060174e-06, "loss": 0.12903740406036376, "step": 19000 }, { "epoch": 0.5007532042047728, "grad_norm": 2.0580294132232666, "learning_rate": 5.231117817029143e-06, "loss": 0.12872315645217897, "step": 19010 }, { "epoch": 0.5010166198829447, "grad_norm": 1.6362437009811401, "learning_rate": 5.2268566066347515e-06, "loss": 0.12707269191741943, "step": 19020 }, { "epoch": 0.5012800355611166, "grad_norm": 2.0130724906921387, "learning_rate": 5.222595231124333e-06, "loss": 0.1520641565322876, "step": 19030 }, { "epoch": 0.5015434512392885, "grad_norm": 2.2461965084075928, "learning_rate": 5.218333693599502e-06, "loss": 0.1389616012573242, "step": 19040 }, { "epoch": 0.5018068669174603, "grad_norm": 4.3327860832214355, "learning_rate": 5.214071997161989e-06, "loss": 0.14821286201477052, "step": 19050 }, { "epoch": 0.5020702825956322, "grad_norm": 1.897737741470337, "learning_rate": 5.20981014491364e-06, "loss": 0.14282385110855103, "step": 19060 }, { "epoch": 0.5023336982738041, "grad_norm": 1.5631507635116577, "learning_rate": 5.205548139956418e-06, "loss": 0.13766177892684936, "step": 19070 }, { "epoch": 0.502597113951976, "grad_norm": 1.58620285987854, "learning_rate": 5.201285985392389e-06, "loss": 0.1517622709274292, "step": 19080 }, { "epoch": 0.5028605296301479, "grad_norm": 4.766759395599365, "learning_rate": 5.197023684323736e-06, "loss": 0.14052560329437255, "step": 19090 }, { "epoch": 0.5031239453083198, "grad_norm": 1.9730724096298218, "learning_rate": 5.192761239852746e-06, "loss": 0.12912242412567138, "step": 19100 }, { "epoch": 0.5033873609864917, "grad_norm": 2.1934258937835693, "learning_rate": 5.188498655081811e-06, "loss": 0.1334105372428894, "step": 19110 }, { "epoch": 0.5036507766646636, "grad_norm": 2.202141284942627, "learning_rate": 5.184235933113419e-06, "loss": 0.13639789819717407, "step": 19120 }, { "epoch": 0.5039141923428355, "grad_norm": 1.9857056140899658, "learning_rate": 5.179973077050168e-06, "loss": 0.13905935287475585, "step": 19130 }, { "epoch": 0.5041776080210074, "grad_norm": 2.3106470108032227, "learning_rate": 5.175710089994745e-06, "loss": 0.13236756324768068, "step": 19140 }, { "epoch": 0.5044410236991793, "grad_norm": 1.7005298137664795, "learning_rate": 5.171446975049939e-06, "loss": 0.14108203649520873, "step": 19150 }, { "epoch": 0.5047044393773512, "grad_norm": 2.3785173892974854, "learning_rate": 5.167183735318629e-06, "loss": 0.14222462177276612, "step": 19160 }, { "epoch": 0.5049678550555231, "grad_norm": 2.459904909133911, "learning_rate": 5.162920373903781e-06, "loss": 0.1432353973388672, "step": 19170 }, { "epoch": 0.505231270733695, "grad_norm": 1.590559482574463, "learning_rate": 5.158656893908458e-06, "loss": 0.14701273441314697, "step": 19180 }, { "epoch": 0.5054946864118669, "grad_norm": 2.9150383472442627, "learning_rate": 5.154393298435803e-06, "loss": 0.15156294107437135, "step": 19190 }, { "epoch": 0.5057581020900388, "grad_norm": 2.010150194168091, "learning_rate": 5.150129590589043e-06, "loss": 0.1356189489364624, "step": 19200 }, { "epoch": 0.5060215177682107, "grad_norm": 2.9375698566436768, "learning_rate": 5.145865773471489e-06, "loss": 0.1354214668273926, "step": 19210 }, { "epoch": 0.5062849334463826, "grad_norm": 2.3435966968536377, "learning_rate": 5.1416018501865325e-06, "loss": 0.1456549048423767, "step": 19220 }, { "epoch": 0.5065483491245545, "grad_norm": 2.57275390625, "learning_rate": 5.137337823837641e-06, "loss": 0.15092352628707886, "step": 19230 }, { "epoch": 0.5068117648027264, "grad_norm": 1.6775504350662231, "learning_rate": 5.133073697528356e-06, "loss": 0.1521836519241333, "step": 19240 }, { "epoch": 0.5070751804808983, "grad_norm": 1.7718104124069214, "learning_rate": 5.1288094743622914e-06, "loss": 0.15692909955978393, "step": 19250 }, { "epoch": 0.5073385961590702, "grad_norm": 1.9878945350646973, "learning_rate": 5.124545157443135e-06, "loss": 0.14417952299118042, "step": 19260 }, { "epoch": 0.5076020118372421, "grad_norm": 3.1535160541534424, "learning_rate": 5.120280749874639e-06, "loss": 0.14736754894256593, "step": 19270 }, { "epoch": 0.507865427515414, "grad_norm": 3.398618698120117, "learning_rate": 5.116016254760625e-06, "loss": 0.13532621860504152, "step": 19280 }, { "epoch": 0.5081288431935859, "grad_norm": 1.8125567436218262, "learning_rate": 5.111751675204974e-06, "loss": 0.1547936201095581, "step": 19290 }, { "epoch": 0.5083922588717578, "grad_norm": 2.58874773979187, "learning_rate": 5.107487014311634e-06, "loss": 0.15335203409194947, "step": 19300 }, { "epoch": 0.5086556745499297, "grad_norm": 2.05243182182312, "learning_rate": 5.103222275184606e-06, "loss": 0.13420605659484863, "step": 19310 }, { "epoch": 0.5089190902281016, "grad_norm": 1.9483964443206787, "learning_rate": 5.098957460927954e-06, "loss": 0.14007506370544434, "step": 19320 }, { "epoch": 0.5091825059062735, "grad_norm": 2.357628583908081, "learning_rate": 5.094692574645793e-06, "loss": 0.12944576740264893, "step": 19330 }, { "epoch": 0.5094459215844453, "grad_norm": 1.4965506792068481, "learning_rate": 5.09042761944229e-06, "loss": 0.13293569087982177, "step": 19340 }, { "epoch": 0.5097093372626172, "grad_norm": 1.6737098693847656, "learning_rate": 5.086162598421664e-06, "loss": 0.1359407424926758, "step": 19350 }, { "epoch": 0.5099727529407891, "grad_norm": 1.30066978931427, "learning_rate": 5.081897514688181e-06, "loss": 0.12502914667129517, "step": 19360 }, { "epoch": 0.510236168618961, "grad_norm": 1.6391781568527222, "learning_rate": 5.077632371346155e-06, "loss": 0.12214529514312744, "step": 19370 }, { "epoch": 0.5104995842971329, "grad_norm": 1.7952793836593628, "learning_rate": 5.0733671714999375e-06, "loss": 0.1324016809463501, "step": 19380 }, { "epoch": 0.5107629999753048, "grad_norm": 3.7966790199279785, "learning_rate": 5.069101918253929e-06, "loss": 0.13778285980224608, "step": 19390 }, { "epoch": 0.5110264156534767, "grad_norm": 3.3981823921203613, "learning_rate": 5.0648366147125625e-06, "loss": 0.1465572237968445, "step": 19400 }, { "epoch": 0.5112898313316486, "grad_norm": 1.573104739189148, "learning_rate": 5.060571263980312e-06, "loss": 0.13335013389587402, "step": 19410 }, { "epoch": 0.5115532470098204, "grad_norm": 1.6081868410110474, "learning_rate": 5.056305869161681e-06, "loss": 0.12916651964187623, "step": 19420 }, { "epoch": 0.5118166626879923, "grad_norm": 2.0184319019317627, "learning_rate": 5.052040433361209e-06, "loss": 0.1379850387573242, "step": 19430 }, { "epoch": 0.5120800783661642, "grad_norm": 1.8569562435150146, "learning_rate": 5.047774959683464e-06, "loss": 0.13471927642822265, "step": 19440 }, { "epoch": 0.5123434940443361, "grad_norm": 1.430996060371399, "learning_rate": 5.043509451233043e-06, "loss": 0.1258331298828125, "step": 19450 }, { "epoch": 0.512606909722508, "grad_norm": 1.5845624208450317, "learning_rate": 5.039243911114566e-06, "loss": 0.13706626892089843, "step": 19460 }, { "epoch": 0.5128703254006799, "grad_norm": 3.233156204223633, "learning_rate": 5.034978342432677e-06, "loss": 0.13646594285964966, "step": 19470 }, { "epoch": 0.5131337410788518, "grad_norm": 2.035828113555908, "learning_rate": 5.030712748292042e-06, "loss": 0.13977733850479127, "step": 19480 }, { "epoch": 0.5133971567570237, "grad_norm": 2.31390380859375, "learning_rate": 5.026447131797343e-06, "loss": 0.14869040250778198, "step": 19490 }, { "epoch": 0.5136605724351956, "grad_norm": 3.237780809402466, "learning_rate": 5.0221814960532815e-06, "loss": 0.14864155054092407, "step": 19500 }, { "epoch": 0.5139239881133675, "grad_norm": 3.159925937652588, "learning_rate": 5.017915844164569e-06, "loss": 0.14235682487487794, "step": 19510 }, { "epoch": 0.5141874037915394, "grad_norm": 2.4178614616394043, "learning_rate": 5.013650179235932e-06, "loss": 0.14195343255996704, "step": 19520 }, { "epoch": 0.5144508194697113, "grad_norm": 1.8643718957901, "learning_rate": 5.009384504372107e-06, "loss": 0.1341008186340332, "step": 19530 }, { "epoch": 0.5147142351478832, "grad_norm": 1.8526992797851562, "learning_rate": 5.005118822677834e-06, "loss": 0.13740912675857545, "step": 19540 }, { "epoch": 0.5149776508260551, "grad_norm": 1.701194405555725, "learning_rate": 5.000853137257862e-06, "loss": 0.13222482204437255, "step": 19550 }, { "epoch": 0.515241066504227, "grad_norm": 1.7992476224899292, "learning_rate": 4.996587451216939e-06, "loss": 0.14158836603164673, "step": 19560 }, { "epoch": 0.5155044821823989, "grad_norm": 2.796937942504883, "learning_rate": 4.992321767659816e-06, "loss": 0.14530186653137206, "step": 19570 }, { "epoch": 0.5157678978605708, "grad_norm": 2.950636386871338, "learning_rate": 4.9880560896912416e-06, "loss": 0.13715574741363526, "step": 19580 }, { "epoch": 0.5160313135387427, "grad_norm": 1.785079002380371, "learning_rate": 4.983790420415961e-06, "loss": 0.14244403839111328, "step": 19590 }, { "epoch": 0.5162947292169145, "grad_norm": 1.9133107662200928, "learning_rate": 4.979524762938713e-06, "loss": 0.1259798526763916, "step": 19600 }, { "epoch": 0.5165581448950864, "grad_norm": 2.4808757305145264, "learning_rate": 4.975259120364223e-06, "loss": 0.15936048030853273, "step": 19610 }, { "epoch": 0.5168215605732583, "grad_norm": 2.152336597442627, "learning_rate": 4.970993495797215e-06, "loss": 0.13833547830581666, "step": 19620 }, { "epoch": 0.5170849762514302, "grad_norm": 5.709275722503662, "learning_rate": 4.96672789234239e-06, "loss": 0.12779189348220826, "step": 19630 }, { "epoch": 0.5173483919296021, "grad_norm": 2.8789491653442383, "learning_rate": 4.962462313104442e-06, "loss": 0.14552606344223024, "step": 19640 }, { "epoch": 0.517611807607774, "grad_norm": 1.6305001974105835, "learning_rate": 4.958196761188044e-06, "loss": 0.13692636489868165, "step": 19650 }, { "epoch": 0.5178752232859459, "grad_norm": 2.300482988357544, "learning_rate": 4.953931239697842e-06, "loss": 0.13995141983032228, "step": 19660 }, { "epoch": 0.5181386389641178, "grad_norm": 2.1744332313537598, "learning_rate": 4.949665751738476e-06, "loss": 0.12951843738555907, "step": 19670 }, { "epoch": 0.5184020546422897, "grad_norm": 3.4736945629119873, "learning_rate": 4.945400300414543e-06, "loss": 0.14185818433761596, "step": 19680 }, { "epoch": 0.5186654703204616, "grad_norm": 1.810952067375183, "learning_rate": 4.9411348888306305e-06, "loss": 0.14338741302490235, "step": 19690 }, { "epoch": 0.5189288859986335, "grad_norm": 2.280569314956665, "learning_rate": 4.936869520091284e-06, "loss": 0.13578567504882813, "step": 19700 }, { "epoch": 0.5191923016768054, "grad_norm": 2.1247174739837646, "learning_rate": 4.932604197301025e-06, "loss": 0.14559452533721923, "step": 19710 }, { "epoch": 0.5194557173549773, "grad_norm": 1.6835228204727173, "learning_rate": 4.928338923564339e-06, "loss": 0.12928686141967774, "step": 19720 }, { "epoch": 0.5197191330331492, "grad_norm": 6.23717737197876, "learning_rate": 4.924073701985676e-06, "loss": 0.14367005825042725, "step": 19730 }, { "epoch": 0.5199825487113211, "grad_norm": 3.7512450218200684, "learning_rate": 4.919808535669448e-06, "loss": 0.1467425584793091, "step": 19740 }, { "epoch": 0.520245964389493, "grad_norm": 3.0444841384887695, "learning_rate": 4.915543427720028e-06, "loss": 0.14180649518966676, "step": 19750 }, { "epoch": 0.5205093800676649, "grad_norm": 1.6019421815872192, "learning_rate": 4.911278381241745e-06, "loss": 0.13086180686950682, "step": 19760 }, { "epoch": 0.5207727957458368, "grad_norm": 3.609929323196411, "learning_rate": 4.907013399338883e-06, "loss": 0.12537416219711303, "step": 19770 }, { "epoch": 0.5210362114240087, "grad_norm": 3.549104690551758, "learning_rate": 4.902748485115684e-06, "loss": 0.15198224782943726, "step": 19780 }, { "epoch": 0.5212996271021806, "grad_norm": 2.8509409427642822, "learning_rate": 4.898483641676329e-06, "loss": 0.130735445022583, "step": 19790 }, { "epoch": 0.5215630427803525, "grad_norm": 1.3470938205718994, "learning_rate": 4.894218872124962e-06, "loss": 0.1261749744415283, "step": 19800 }, { "epoch": 0.5218264584585244, "grad_norm": 5.123560905456543, "learning_rate": 4.8899541795656635e-06, "loss": 0.13914868831634522, "step": 19810 }, { "epoch": 0.5220898741366963, "grad_norm": 1.8163535594940186, "learning_rate": 4.88568956710246e-06, "loss": 0.13756562471389772, "step": 19820 }, { "epoch": 0.5223532898148682, "grad_norm": 2.323626756668091, "learning_rate": 4.881425037839322e-06, "loss": 0.1409894585609436, "step": 19830 }, { "epoch": 0.5226167054930401, "grad_norm": 2.3458051681518555, "learning_rate": 4.877160594880156e-06, "loss": 0.14443496465682984, "step": 19840 }, { "epoch": 0.522880121171212, "grad_norm": 1.4859132766723633, "learning_rate": 4.87289624132881e-06, "loss": 0.12979531288146973, "step": 19850 }, { "epoch": 0.5231435368493839, "grad_norm": 2.1264123916625977, "learning_rate": 4.868631980289063e-06, "loss": 0.1508224368095398, "step": 19860 }, { "epoch": 0.5234069525275558, "grad_norm": 1.333258867263794, "learning_rate": 4.864367814864628e-06, "loss": 0.13955552577972413, "step": 19870 }, { "epoch": 0.5236703682057277, "grad_norm": 2.864813804626465, "learning_rate": 4.86010374815915e-06, "loss": 0.12067651748657227, "step": 19880 }, { "epoch": 0.5239337838838996, "grad_norm": 1.7403103113174438, "learning_rate": 4.855839783276198e-06, "loss": 0.12629566192626954, "step": 19890 }, { "epoch": 0.5241971995620714, "grad_norm": 2.4935171604156494, "learning_rate": 4.851575923319272e-06, "loss": 0.13486642837524415, "step": 19900 }, { "epoch": 0.5244606152402433, "grad_norm": 2.2344112396240234, "learning_rate": 4.8473121713917955e-06, "loss": 0.12539105415344237, "step": 19910 }, { "epoch": 0.5247240309184152, "grad_norm": 2.696828603744507, "learning_rate": 4.8430485305971046e-06, "loss": 0.12799862623214722, "step": 19920 }, { "epoch": 0.5249874465965871, "grad_norm": 1.3933926820755005, "learning_rate": 4.838785004038467e-06, "loss": 0.13515297174453736, "step": 19930 }, { "epoch": 0.525250862274759, "grad_norm": 2.018460273742676, "learning_rate": 4.834521594819057e-06, "loss": 0.13791160583496093, "step": 19940 }, { "epoch": 0.5255142779529309, "grad_norm": 1.8480734825134277, "learning_rate": 4.83025830604197e-06, "loss": 0.1396145224571228, "step": 19950 }, { "epoch": 0.5257776936311028, "grad_norm": 1.6446614265441895, "learning_rate": 4.825995140810211e-06, "loss": 0.1380678176879883, "step": 19960 }, { "epoch": 0.5260411093092747, "grad_norm": 3.4837074279785156, "learning_rate": 4.821732102226696e-06, "loss": 0.12724263668060304, "step": 19970 }, { "epoch": 0.5263045249874466, "grad_norm": 1.9457160234451294, "learning_rate": 4.8174691933942485e-06, "loss": 0.13550647497177123, "step": 19980 }, { "epoch": 0.5265679406656185, "grad_norm": 2.035127878189087, "learning_rate": 4.8132064174155965e-06, "loss": 0.14220039844512938, "step": 19990 }, { "epoch": 0.5268313563437904, "grad_norm": 1.669464349746704, "learning_rate": 4.808943777393373e-06, "loss": 0.14167497158050538, "step": 20000 }, { "epoch": 0.5270947720219623, "grad_norm": 3.514099359512329, "learning_rate": 4.8046812764301125e-06, "loss": 0.14162155389785766, "step": 20010 }, { "epoch": 0.5273581877001342, "grad_norm": 1.764420509338379, "learning_rate": 4.800418917628243e-06, "loss": 0.14210946559906007, "step": 20020 }, { "epoch": 0.5276216033783061, "grad_norm": 1.9246875047683716, "learning_rate": 4.796156704090099e-06, "loss": 0.13040463924407958, "step": 20030 }, { "epoch": 0.527885019056478, "grad_norm": 3.4937515258789062, "learning_rate": 4.791894638917899e-06, "loss": 0.13520776033401488, "step": 20040 }, { "epoch": 0.5281484347346499, "grad_norm": 2.23240065574646, "learning_rate": 4.787632725213759e-06, "loss": 0.13096802234649657, "step": 20050 }, { "epoch": 0.5284118504128218, "grad_norm": 1.6924092769622803, "learning_rate": 4.783370966079684e-06, "loss": 0.11875994205474853, "step": 20060 }, { "epoch": 0.5286752660909937, "grad_norm": 2.4654476642608643, "learning_rate": 4.779109364617566e-06, "loss": 0.1423788070678711, "step": 20070 }, { "epoch": 0.5289386817691656, "grad_norm": 2.1279349327087402, "learning_rate": 4.7748479239291815e-06, "loss": 0.14903066158294678, "step": 20080 }, { "epoch": 0.5292020974473375, "grad_norm": 3.550809621810913, "learning_rate": 4.7705866471161935e-06, "loss": 0.12554992437362672, "step": 20090 }, { "epoch": 0.5294655131255094, "grad_norm": 1.9478639364242554, "learning_rate": 4.76632553728014e-06, "loss": 0.13538841009140015, "step": 20100 }, { "epoch": 0.5297289288036813, "grad_norm": 1.7173869609832764, "learning_rate": 4.762064597522443e-06, "loss": 0.14310216903686523, "step": 20110 }, { "epoch": 0.5299923444818532, "grad_norm": 4.513588905334473, "learning_rate": 4.7578038309443995e-06, "loss": 0.1369636058807373, "step": 20120 }, { "epoch": 0.5302557601600251, "grad_norm": 2.3692543506622314, "learning_rate": 4.753543240647173e-06, "loss": 0.13891317844390869, "step": 20130 }, { "epoch": 0.530519175838197, "grad_norm": 2.147625207901001, "learning_rate": 4.749282829731812e-06, "loss": 0.1365249991416931, "step": 20140 }, { "epoch": 0.5307825915163689, "grad_norm": 1.702759027481079, "learning_rate": 4.745022601299221e-06, "loss": 0.11913318634033203, "step": 20150 }, { "epoch": 0.5310460071945408, "grad_norm": 2.27864933013916, "learning_rate": 4.740762558450185e-06, "loss": 0.12597562074661256, "step": 20160 }, { "epoch": 0.5313094228727127, "grad_norm": 2.8484110832214355, "learning_rate": 4.7365027042853425e-06, "loss": 0.14115272760391234, "step": 20170 }, { "epoch": 0.5315728385508846, "grad_norm": 1.634049892425537, "learning_rate": 4.732243041905199e-06, "loss": 0.13286406993865968, "step": 20180 }, { "epoch": 0.5318362542290564, "grad_norm": 1.8478319644927979, "learning_rate": 4.727983574410122e-06, "loss": 0.14365191459655763, "step": 20190 }, { "epoch": 0.5320996699072283, "grad_norm": 1.5212639570236206, "learning_rate": 4.723724304900335e-06, "loss": 0.13201889991760254, "step": 20200 }, { "epoch": 0.5323630855854001, "grad_norm": 1.3791258335113525, "learning_rate": 4.719465236475919e-06, "loss": 0.13907430171966553, "step": 20210 }, { "epoch": 0.532626501263572, "grad_norm": 3.4023168087005615, "learning_rate": 4.7152063722368076e-06, "loss": 0.13470664024353027, "step": 20220 }, { "epoch": 0.5328899169417439, "grad_norm": 4.2973833084106445, "learning_rate": 4.710947715282784e-06, "loss": 0.1371508479118347, "step": 20230 }, { "epoch": 0.5331533326199158, "grad_norm": 1.5328816175460815, "learning_rate": 4.706689268713486e-06, "loss": 0.13795154094696044, "step": 20240 }, { "epoch": 0.5334167482980877, "grad_norm": 2.4011592864990234, "learning_rate": 4.702431035628393e-06, "loss": 0.13528114557266235, "step": 20250 }, { "epoch": 0.5336801639762596, "grad_norm": 1.918984293937683, "learning_rate": 4.698173019126827e-06, "loss": 0.13733992576599122, "step": 20260 }, { "epoch": 0.5339435796544315, "grad_norm": 2.1617414951324463, "learning_rate": 4.693915222307963e-06, "loss": 0.12777307033538818, "step": 20270 }, { "epoch": 0.5342069953326034, "grad_norm": 1.6112240552902222, "learning_rate": 4.689657648270806e-06, "loss": 0.1370531678199768, "step": 20280 }, { "epoch": 0.5344704110107753, "grad_norm": 2.0768651962280273, "learning_rate": 4.685400300114199e-06, "loss": 0.12874703407287597, "step": 20290 }, { "epoch": 0.5347338266889472, "grad_norm": 1.5599384307861328, "learning_rate": 4.681143180936829e-06, "loss": 0.1305445432662964, "step": 20300 }, { "epoch": 0.5349972423671191, "grad_norm": 3.523266553878784, "learning_rate": 4.676886293837207e-06, "loss": 0.1463677167892456, "step": 20310 }, { "epoch": 0.535260658045291, "grad_norm": 5.62265157699585, "learning_rate": 4.672629641913681e-06, "loss": 0.12938518524169923, "step": 20320 }, { "epoch": 0.5355240737234629, "grad_norm": 3.0576584339141846, "learning_rate": 4.668373228264427e-06, "loss": 0.13960654735565187, "step": 20330 }, { "epoch": 0.5357874894016348, "grad_norm": 1.9255996942520142, "learning_rate": 4.664117055987443e-06, "loss": 0.14287514686584474, "step": 20340 }, { "epoch": 0.5360509050798067, "grad_norm": 1.8373359441757202, "learning_rate": 4.659861128180558e-06, "loss": 0.13238670825958251, "step": 20350 }, { "epoch": 0.5363143207579786, "grad_norm": 2.3095862865448, "learning_rate": 4.655605447941418e-06, "loss": 0.1376890778541565, "step": 20360 }, { "epoch": 0.5365777364361505, "grad_norm": 2.8820533752441406, "learning_rate": 4.651350018367493e-06, "loss": 0.1349032163619995, "step": 20370 }, { "epoch": 0.5368411521143224, "grad_norm": 5.61858606338501, "learning_rate": 4.647094842556067e-06, "loss": 0.1464466333389282, "step": 20380 }, { "epoch": 0.5371045677924943, "grad_norm": 3.2173357009887695, "learning_rate": 4.642839923604237e-06, "loss": 0.14335615634918214, "step": 20390 }, { "epoch": 0.5373679834706662, "grad_norm": 1.8683401346206665, "learning_rate": 4.638585264608923e-06, "loss": 0.14333133697509765, "step": 20400 }, { "epoch": 0.5376313991488381, "grad_norm": 1.4656524658203125, "learning_rate": 4.634330868666846e-06, "loss": 0.14101370573043823, "step": 20410 }, { "epoch": 0.53789481482701, "grad_norm": 2.3900468349456787, "learning_rate": 4.630076738874536e-06, "loss": 0.13990843296051025, "step": 20420 }, { "epoch": 0.5381582305051819, "grad_norm": 4.7448296546936035, "learning_rate": 4.625822878328338e-06, "loss": 0.12755422592163085, "step": 20430 }, { "epoch": 0.5384216461833538, "grad_norm": 3.1186652183532715, "learning_rate": 4.621569290124391e-06, "loss": 0.14285359382629395, "step": 20440 }, { "epoch": 0.5386850618615256, "grad_norm": 1.6415798664093018, "learning_rate": 4.617315977358643e-06, "loss": 0.12846605777740477, "step": 20450 }, { "epoch": 0.5389484775396975, "grad_norm": 2.0264251232147217, "learning_rate": 4.613062943126836e-06, "loss": 0.1478576183319092, "step": 20460 }, { "epoch": 0.5392118932178694, "grad_norm": 2.1445131301879883, "learning_rate": 4.608810190524513e-06, "loss": 0.14479511976242065, "step": 20470 }, { "epoch": 0.5394753088960413, "grad_norm": 1.9773221015930176, "learning_rate": 4.604557722647011e-06, "loss": 0.13257691860198975, "step": 20480 }, { "epoch": 0.5397387245742132, "grad_norm": 1.6608942747116089, "learning_rate": 4.6003055425894585e-06, "loss": 0.1281324028968811, "step": 20490 }, { "epoch": 0.5400021402523851, "grad_norm": 1.4467105865478516, "learning_rate": 4.596053653446778e-06, "loss": 0.14014493227005004, "step": 20500 }, { "epoch": 0.540265555930557, "grad_norm": 2.8413960933685303, "learning_rate": 4.591802058313677e-06, "loss": 0.13678011894226075, "step": 20510 }, { "epoch": 0.5405289716087289, "grad_norm": 2.189793109893799, "learning_rate": 4.587550760284646e-06, "loss": 0.14064927101135255, "step": 20520 }, { "epoch": 0.5407923872869008, "grad_norm": 3.035292625427246, "learning_rate": 4.583299762453969e-06, "loss": 0.1428179383277893, "step": 20530 }, { "epoch": 0.5410558029650727, "grad_norm": 1.6624163389205933, "learning_rate": 4.579049067915702e-06, "loss": 0.1355337142944336, "step": 20540 }, { "epoch": 0.5413192186432446, "grad_norm": 3.6518476009368896, "learning_rate": 4.574798679763684e-06, "loss": 0.13243021965026855, "step": 20550 }, { "epoch": 0.5415826343214165, "grad_norm": 4.035533905029297, "learning_rate": 4.570548601091531e-06, "loss": 0.1527944803237915, "step": 20560 }, { "epoch": 0.5418460499995884, "grad_norm": 2.617851734161377, "learning_rate": 4.566298834992635e-06, "loss": 0.1471501350402832, "step": 20570 }, { "epoch": 0.5421094656777603, "grad_norm": 1.6976699829101562, "learning_rate": 4.562049384560159e-06, "loss": 0.13507213592529296, "step": 20580 }, { "epoch": 0.5423728813559322, "grad_norm": 2.964749813079834, "learning_rate": 4.557800252887036e-06, "loss": 0.1340899109840393, "step": 20590 }, { "epoch": 0.5426362970341041, "grad_norm": 1.6238031387329102, "learning_rate": 4.553551443065965e-06, "loss": 0.13760581016540527, "step": 20600 }, { "epoch": 0.542899712712276, "grad_norm": 1.9850322008132935, "learning_rate": 4.549302958189418e-06, "loss": 0.13309946060180664, "step": 20610 }, { "epoch": 0.5431631283904479, "grad_norm": 1.8128193616867065, "learning_rate": 4.545054801349619e-06, "loss": 0.13975069522857667, "step": 20620 }, { "epoch": 0.5434265440686198, "grad_norm": 2.9887242317199707, "learning_rate": 4.540806975638567e-06, "loss": 0.13117597103118897, "step": 20630 }, { "epoch": 0.5436899597467917, "grad_norm": 1.898065447807312, "learning_rate": 4.53655948414801e-06, "loss": 0.12239429950714112, "step": 20640 }, { "epoch": 0.5439533754249636, "grad_norm": 2.525339126586914, "learning_rate": 4.532312329969451e-06, "loss": 0.14216487407684325, "step": 20650 }, { "epoch": 0.5442167911031355, "grad_norm": 2.2039639949798584, "learning_rate": 4.528065516194159e-06, "loss": 0.14969344139099122, "step": 20660 }, { "epoch": 0.5444802067813074, "grad_norm": 2.3414247035980225, "learning_rate": 4.523819045913146e-06, "loss": 0.13960039615631104, "step": 20670 }, { "epoch": 0.5447436224594793, "grad_norm": 1.5319684743881226, "learning_rate": 4.519572922217172e-06, "loss": 0.12809514999389648, "step": 20680 }, { "epoch": 0.5450070381376512, "grad_norm": 1.453079342842102, "learning_rate": 4.515327148196756e-06, "loss": 0.12802581787109374, "step": 20690 }, { "epoch": 0.5452704538158231, "grad_norm": 1.8855524063110352, "learning_rate": 4.5110817269421495e-06, "loss": 0.13680677413940429, "step": 20700 }, { "epoch": 0.545533869493995, "grad_norm": 3.526644229888916, "learning_rate": 4.506836661543358e-06, "loss": 0.13691110610961915, "step": 20710 }, { "epoch": 0.5457972851721669, "grad_norm": 3.682056188583374, "learning_rate": 4.502591955090121e-06, "loss": 0.1409568667411804, "step": 20720 }, { "epoch": 0.5460607008503388, "grad_norm": 2.4587209224700928, "learning_rate": 4.498347610671917e-06, "loss": 0.12809102535247802, "step": 20730 }, { "epoch": 0.5463241165285107, "grad_norm": 1.9089041948318481, "learning_rate": 4.494103631377968e-06, "loss": 0.12506097555160522, "step": 20740 }, { "epoch": 0.5465875322066825, "grad_norm": 2.704148054122925, "learning_rate": 4.489860020297219e-06, "loss": 0.14206401109695435, "step": 20750 }, { "epoch": 0.5468509478848544, "grad_norm": 2.033205986022949, "learning_rate": 4.485616780518359e-06, "loss": 0.12905409336090087, "step": 20760 }, { "epoch": 0.5471143635630263, "grad_norm": 2.099118232727051, "learning_rate": 4.481373915129799e-06, "loss": 0.14498754739761352, "step": 20770 }, { "epoch": 0.5473777792411982, "grad_norm": 3.6679062843322754, "learning_rate": 4.477131427219677e-06, "loss": 0.13570890426635743, "step": 20780 }, { "epoch": 0.5476411949193701, "grad_norm": 3.119908094406128, "learning_rate": 4.472889319875862e-06, "loss": 0.1354449510574341, "step": 20790 }, { "epoch": 0.547904610597542, "grad_norm": 2.221277952194214, "learning_rate": 4.468647596185941e-06, "loss": 0.1369931936264038, "step": 20800 }, { "epoch": 0.5481680262757139, "grad_norm": 2.0804619789123535, "learning_rate": 4.4644062592372225e-06, "loss": 0.13368194103240966, "step": 20810 }, { "epoch": 0.5484314419538858, "grad_norm": 2.323885679244995, "learning_rate": 4.4601653121167385e-06, "loss": 0.1322077512741089, "step": 20820 }, { "epoch": 0.5486948576320577, "grad_norm": 1.7422395944595337, "learning_rate": 4.455924757911229e-06, "loss": 0.13402752876281737, "step": 20830 }, { "epoch": 0.5489582733102296, "grad_norm": 1.548571228981018, "learning_rate": 4.4516845997071554e-06, "loss": 0.13891966342926027, "step": 20840 }, { "epoch": 0.5492216889884015, "grad_norm": 2.4584827423095703, "learning_rate": 4.447444840590687e-06, "loss": 0.12417029142379761, "step": 20850 }, { "epoch": 0.5494851046665734, "grad_norm": 2.889810562133789, "learning_rate": 4.4432054836477004e-06, "loss": 0.14359058141708375, "step": 20860 }, { "epoch": 0.5497485203447453, "grad_norm": 2.7967286109924316, "learning_rate": 4.438966531963788e-06, "loss": 0.133516263961792, "step": 20870 }, { "epoch": 0.5500119360229172, "grad_norm": 4.724292755126953, "learning_rate": 4.4347279886242366e-06, "loss": 0.11559607982635497, "step": 20880 }, { "epoch": 0.5502753517010891, "grad_norm": 1.8218227624893188, "learning_rate": 4.430489856714043e-06, "loss": 0.12121713161468506, "step": 20890 }, { "epoch": 0.550538767379261, "grad_norm": 1.5844134092330933, "learning_rate": 4.426252139317903e-06, "loss": 0.1249232292175293, "step": 20900 }, { "epoch": 0.5508021830574329, "grad_norm": 1.8020943403244019, "learning_rate": 4.422014839520209e-06, "loss": 0.11954154968261718, "step": 20910 }, { "epoch": 0.5510655987356048, "grad_norm": 1.9867700338363647, "learning_rate": 4.417777960405052e-06, "loss": 0.13439559936523438, "step": 20920 }, { "epoch": 0.5513290144137767, "grad_norm": 1.8392605781555176, "learning_rate": 4.4135415050562145e-06, "loss": 0.13347477912902833, "step": 20930 }, { "epoch": 0.5515924300919486, "grad_norm": 1.661994457244873, "learning_rate": 4.409305476557171e-06, "loss": 0.1291187286376953, "step": 20940 }, { "epoch": 0.5518558457701205, "grad_norm": 1.598652958869934, "learning_rate": 4.4050698779910864e-06, "loss": 0.14096572399139404, "step": 20950 }, { "epoch": 0.5521192614482924, "grad_norm": 2.2051193714141846, "learning_rate": 4.400834712440812e-06, "loss": 0.15182640552520751, "step": 20960 }, { "epoch": 0.5523826771264643, "grad_norm": 2.6309103965759277, "learning_rate": 4.396599982988885e-06, "loss": 0.1325032591819763, "step": 20970 }, { "epoch": 0.5526460928046362, "grad_norm": 3.5130796432495117, "learning_rate": 4.392365692717524e-06, "loss": 0.12666491270065308, "step": 20980 }, { "epoch": 0.5529095084828081, "grad_norm": 1.9079008102416992, "learning_rate": 4.388131844708625e-06, "loss": 0.13316987752914428, "step": 20990 }, { "epoch": 0.5531729241609799, "grad_norm": 2.458517551422119, "learning_rate": 4.383898442043769e-06, "loss": 0.1329815626144409, "step": 21000 }, { "epoch": 0.5534363398391517, "grad_norm": 2.569074869155884, "learning_rate": 4.379665487804209e-06, "loss": 0.14320549964904786, "step": 21010 }, { "epoch": 0.5536997555173236, "grad_norm": 1.8870995044708252, "learning_rate": 4.375432985070868e-06, "loss": 0.14309269189834595, "step": 21020 }, { "epoch": 0.5539631711954955, "grad_norm": 3.118781328201294, "learning_rate": 4.371200936924348e-06, "loss": 0.13711209297180177, "step": 21030 }, { "epoch": 0.5542265868736674, "grad_norm": 2.2684459686279297, "learning_rate": 4.366969346444914e-06, "loss": 0.1318830132484436, "step": 21040 }, { "epoch": 0.5544900025518393, "grad_norm": 1.6228333711624146, "learning_rate": 4.3627382167125016e-06, "loss": 0.12508151531219483, "step": 21050 }, { "epoch": 0.5547534182300112, "grad_norm": 2.098179340362549, "learning_rate": 4.358507550806709e-06, "loss": 0.12967910766601562, "step": 21060 }, { "epoch": 0.5550168339081831, "grad_norm": 1.8968486785888672, "learning_rate": 4.354277351806796e-06, "loss": 0.1444178342819214, "step": 21070 }, { "epoch": 0.555280249586355, "grad_norm": 1.891221284866333, "learning_rate": 4.350047622791687e-06, "loss": 0.13645308017730712, "step": 21080 }, { "epoch": 0.5555436652645269, "grad_norm": 2.333322048187256, "learning_rate": 4.3458183668399576e-06, "loss": 0.12890172004699707, "step": 21090 }, { "epoch": 0.5558070809426988, "grad_norm": 1.6218081712722778, "learning_rate": 4.341589587029846e-06, "loss": 0.1356906294822693, "step": 21100 }, { "epoch": 0.5560704966208707, "grad_norm": 3.511535167694092, "learning_rate": 4.33736128643924e-06, "loss": 0.13850884437561034, "step": 21110 }, { "epoch": 0.5563339122990426, "grad_norm": 2.1705968379974365, "learning_rate": 4.333133468145677e-06, "loss": 0.14896667003631592, "step": 21120 }, { "epoch": 0.5565973279772145, "grad_norm": 1.405661702156067, "learning_rate": 4.328906135226348e-06, "loss": 0.12855122089385987, "step": 21130 }, { "epoch": 0.5568607436553864, "grad_norm": 2.0944745540618896, "learning_rate": 4.3246792907580855e-06, "loss": 0.13303428888320923, "step": 21140 }, { "epoch": 0.5571241593335583, "grad_norm": 3.7147648334503174, "learning_rate": 4.320452937817372e-06, "loss": 0.14500449895858764, "step": 21150 }, { "epoch": 0.5573875750117302, "grad_norm": 1.7977783679962158, "learning_rate": 4.316227079480328e-06, "loss": 0.13188475370407104, "step": 21160 }, { "epoch": 0.5576509906899021, "grad_norm": 2.6217598915100098, "learning_rate": 4.3120017188227155e-06, "loss": 0.12950352430343628, "step": 21170 }, { "epoch": 0.557914406368074, "grad_norm": 2.5517799854278564, "learning_rate": 4.307776858919936e-06, "loss": 0.1332114338874817, "step": 21180 }, { "epoch": 0.5581778220462459, "grad_norm": 1.3777574300765991, "learning_rate": 4.303552502847024e-06, "loss": 0.1349979519844055, "step": 21190 }, { "epoch": 0.5584412377244178, "grad_norm": 2.429018259048462, "learning_rate": 4.299328653678646e-06, "loss": 0.13213963508605958, "step": 21200 }, { "epoch": 0.5587046534025897, "grad_norm": 1.5968317985534668, "learning_rate": 4.295105314489107e-06, "loss": 0.1469145894050598, "step": 21210 }, { "epoch": 0.5589680690807616, "grad_norm": 2.51765513420105, "learning_rate": 4.290882488352327e-06, "loss": 0.13702330589294434, "step": 21220 }, { "epoch": 0.5592314847589335, "grad_norm": 2.567270278930664, "learning_rate": 4.286660178341869e-06, "loss": 0.1443018913269043, "step": 21230 }, { "epoch": 0.5594949004371054, "grad_norm": 3.725196123123169, "learning_rate": 4.2824383875309095e-06, "loss": 0.1332364559173584, "step": 21240 }, { "epoch": 0.5597583161152773, "grad_norm": 3.556394338607788, "learning_rate": 4.278217118992248e-06, "loss": 0.14332108497619628, "step": 21250 }, { "epoch": 0.5600217317934492, "grad_norm": 1.5069345235824585, "learning_rate": 4.273996375798308e-06, "loss": 0.14163665771484374, "step": 21260 }, { "epoch": 0.5602851474716211, "grad_norm": 2.256439685821533, "learning_rate": 4.26977616102113e-06, "loss": 0.13447606563568115, "step": 21270 }, { "epoch": 0.560548563149793, "grad_norm": 1.5984922647476196, "learning_rate": 4.265556477732364e-06, "loss": 0.13992457389831542, "step": 21280 }, { "epoch": 0.5608119788279649, "grad_norm": 3.155763626098633, "learning_rate": 4.261337329003282e-06, "loss": 0.13564667701721192, "step": 21290 }, { "epoch": 0.5610753945061367, "grad_norm": 2.3860087394714355, "learning_rate": 4.2571187179047586e-06, "loss": 0.12827737331390382, "step": 21300 }, { "epoch": 0.5613388101843086, "grad_norm": 2.074540138244629, "learning_rate": 4.2529006475072845e-06, "loss": 0.12415748834609985, "step": 21310 }, { "epoch": 0.5616022258624805, "grad_norm": 3.591670036315918, "learning_rate": 4.2486831208809535e-06, "loss": 0.13174235820770264, "step": 21320 }, { "epoch": 0.5618656415406524, "grad_norm": 3.4881653785705566, "learning_rate": 4.244466141095457e-06, "loss": 0.1249419093132019, "step": 21330 }, { "epoch": 0.5621290572188243, "grad_norm": 4.7174882888793945, "learning_rate": 4.240249711220103e-06, "loss": 0.13798723220825196, "step": 21340 }, { "epoch": 0.5623924728969962, "grad_norm": 4.171737194061279, "learning_rate": 4.2360338343237854e-06, "loss": 0.1363377571105957, "step": 21350 }, { "epoch": 0.5626558885751681, "grad_norm": 2.280640125274658, "learning_rate": 4.231818513475006e-06, "loss": 0.11869561672210693, "step": 21360 }, { "epoch": 0.56291930425334, "grad_norm": 2.167562961578369, "learning_rate": 4.227603751741854e-06, "loss": 0.14160356521606446, "step": 21370 }, { "epoch": 0.5631827199315119, "grad_norm": 2.442617893218994, "learning_rate": 4.223389552192014e-06, "loss": 0.11797411441802978, "step": 21380 }, { "epoch": 0.5634461356096838, "grad_norm": 2.1853559017181396, "learning_rate": 4.219175917892765e-06, "loss": 0.12056951522827149, "step": 21390 }, { "epoch": 0.5637095512878557, "grad_norm": 3.352212429046631, "learning_rate": 4.214962851910971e-06, "loss": 0.1403264284133911, "step": 21400 }, { "epoch": 0.5639729669660276, "grad_norm": 3.174177885055542, "learning_rate": 4.210750357313082e-06, "loss": 0.14410895109176636, "step": 21410 }, { "epoch": 0.5642363826441995, "grad_norm": 2.1124093532562256, "learning_rate": 4.206538437165133e-06, "loss": 0.13427309989929198, "step": 21420 }, { "epoch": 0.5644997983223714, "grad_norm": 3.9634311199188232, "learning_rate": 4.202327094532743e-06, "loss": 0.13495813608169555, "step": 21430 }, { "epoch": 0.5647632140005433, "grad_norm": 5.545526504516602, "learning_rate": 4.198116332481109e-06, "loss": 0.12295725345611572, "step": 21440 }, { "epoch": 0.5650266296787152, "grad_norm": 1.7385995388031006, "learning_rate": 4.1939061540750055e-06, "loss": 0.1404163122177124, "step": 21450 }, { "epoch": 0.5652900453568871, "grad_norm": 3.712170362472534, "learning_rate": 4.189696562378777e-06, "loss": 0.12386704683303833, "step": 21460 }, { "epoch": 0.565553461035059, "grad_norm": 2.927086591720581, "learning_rate": 4.185487560456354e-06, "loss": 0.14123773574829102, "step": 21470 }, { "epoch": 0.5658168767132309, "grad_norm": 2.030923366546631, "learning_rate": 4.181279151371224e-06, "loss": 0.1279660940170288, "step": 21480 }, { "epoch": 0.5660802923914028, "grad_norm": 1.8974653482437134, "learning_rate": 4.177071338186449e-06, "loss": 0.1354508399963379, "step": 21490 }, { "epoch": 0.5663437080695747, "grad_norm": 2.1938464641571045, "learning_rate": 4.1728641239646594e-06, "loss": 0.1401708960533142, "step": 21500 }, { "epoch": 0.5666071237477466, "grad_norm": 2.3896515369415283, "learning_rate": 4.168657511768045e-06, "loss": 0.1370328187942505, "step": 21510 }, { "epoch": 0.5668705394259185, "grad_norm": 2.687933921813965, "learning_rate": 4.164451504658362e-06, "loss": 0.12622809410095215, "step": 21520 }, { "epoch": 0.5671339551040904, "grad_norm": 2.6971435546875, "learning_rate": 4.160246105696922e-06, "loss": 0.13385852575302123, "step": 21530 }, { "epoch": 0.5673973707822623, "grad_norm": 1.8668965101242065, "learning_rate": 4.156041317944596e-06, "loss": 0.13614945411682128, "step": 21540 }, { "epoch": 0.5676607864604342, "grad_norm": 1.3284648656845093, "learning_rate": 4.151837144461811e-06, "loss": 0.13309569358825685, "step": 21550 }, { "epoch": 0.5679242021386061, "grad_norm": 1.6155864000320435, "learning_rate": 4.147633588308545e-06, "loss": 0.12980506420135499, "step": 21560 }, { "epoch": 0.568187617816778, "grad_norm": 2.7010326385498047, "learning_rate": 4.143430652544328e-06, "loss": 0.1344562888145447, "step": 21570 }, { "epoch": 0.5684510334949499, "grad_norm": 1.8443022966384888, "learning_rate": 4.139228340228238e-06, "loss": 0.12095389366149903, "step": 21580 }, { "epoch": 0.5687144491731217, "grad_norm": 1.5575666427612305, "learning_rate": 4.135026654418896e-06, "loss": 0.13819508552551268, "step": 21590 }, { "epoch": 0.5689778648512936, "grad_norm": 2.4828615188598633, "learning_rate": 4.130825598174476e-06, "loss": 0.12328238487243652, "step": 21600 }, { "epoch": 0.5692412805294655, "grad_norm": 2.2836036682128906, "learning_rate": 4.126625174552684e-06, "loss": 0.13579146862030028, "step": 21610 }, { "epoch": 0.5695046962076374, "grad_norm": 2.3828530311584473, "learning_rate": 4.12242538661077e-06, "loss": 0.12357559204101562, "step": 21620 }, { "epoch": 0.5697681118858093, "grad_norm": 2.370561122894287, "learning_rate": 4.118226237405523e-06, "loss": 0.13542389869689941, "step": 21630 }, { "epoch": 0.5700315275639812, "grad_norm": 3.160602331161499, "learning_rate": 4.1140277299932605e-06, "loss": 0.12639126777648926, "step": 21640 }, { "epoch": 0.5702949432421531, "grad_norm": 2.658536195755005, "learning_rate": 4.109829867429843e-06, "loss": 0.1396339178085327, "step": 21650 }, { "epoch": 0.570558358920325, "grad_norm": 1.893929362297058, "learning_rate": 4.105632652770652e-06, "loss": 0.13183940649032594, "step": 21660 }, { "epoch": 0.5708217745984969, "grad_norm": 2.809687614440918, "learning_rate": 4.101436089070602e-06, "loss": 0.12059086561203003, "step": 21670 }, { "epoch": 0.5710851902766688, "grad_norm": 2.3205738067626953, "learning_rate": 4.097240179384135e-06, "loss": 0.13291616439819337, "step": 21680 }, { "epoch": 0.5713486059548407, "grad_norm": 3.0841519832611084, "learning_rate": 4.093044926765213e-06, "loss": 0.12721662521362304, "step": 21690 }, { "epoch": 0.5716120216330126, "grad_norm": 3.1890251636505127, "learning_rate": 4.088850334267323e-06, "loss": 0.13504871129989623, "step": 21700 }, { "epoch": 0.5718754373111845, "grad_norm": 3.1179065704345703, "learning_rate": 4.084656404943472e-06, "loss": 0.1283905625343323, "step": 21710 }, { "epoch": 0.5721388529893564, "grad_norm": 3.25480055809021, "learning_rate": 4.080463141846178e-06, "loss": 0.127738094329834, "step": 21720 }, { "epoch": 0.5724022686675283, "grad_norm": 2.7209136486053467, "learning_rate": 4.0762705480274826e-06, "loss": 0.1234935164451599, "step": 21730 }, { "epoch": 0.5726656843457002, "grad_norm": 2.3642139434814453, "learning_rate": 4.0720786265389354e-06, "loss": 0.1255425214767456, "step": 21740 }, { "epoch": 0.5729291000238721, "grad_norm": 2.0770580768585205, "learning_rate": 4.067887380431596e-06, "loss": 0.14760177135467528, "step": 21750 }, { "epoch": 0.573192515702044, "grad_norm": 2.102221727371216, "learning_rate": 4.063696812756038e-06, "loss": 0.12956554889678956, "step": 21760 }, { "epoch": 0.5734559313802159, "grad_norm": 2.615806818008423, "learning_rate": 4.059506926562335e-06, "loss": 0.12201014757156373, "step": 21770 }, { "epoch": 0.5737193470583878, "grad_norm": 3.3552322387695312, "learning_rate": 4.055317724900066e-06, "loss": 0.130226993560791, "step": 21780 }, { "epoch": 0.5739827627365597, "grad_norm": 5.179234504699707, "learning_rate": 4.051129210818315e-06, "loss": 0.1353604555130005, "step": 21790 }, { "epoch": 0.5742461784147315, "grad_norm": 3.977402925491333, "learning_rate": 4.046941387365661e-06, "loss": 0.12543569803237914, "step": 21800 }, { "epoch": 0.5745095940929034, "grad_norm": 2.3667426109313965, "learning_rate": 4.042754257590186e-06, "loss": 0.12604576349258423, "step": 21810 }, { "epoch": 0.5747730097710753, "grad_norm": 2.1768980026245117, "learning_rate": 4.038567824539458e-06, "loss": 0.12007803916931152, "step": 21820 }, { "epoch": 0.5750364254492472, "grad_norm": 2.131080389022827, "learning_rate": 4.034382091260551e-06, "loss": 0.1378009557723999, "step": 21830 }, { "epoch": 0.575299841127419, "grad_norm": 2.989100456237793, "learning_rate": 4.030197060800017e-06, "loss": 0.13803030252456666, "step": 21840 }, { "epoch": 0.575563256805591, "grad_norm": 1.8281649351119995, "learning_rate": 4.026012736203901e-06, "loss": 0.13617171049118043, "step": 21850 }, { "epoch": 0.5758266724837628, "grad_norm": 2.687771797180176, "learning_rate": 4.021829120517739e-06, "loss": 0.12987686395645143, "step": 21860 }, { "epoch": 0.5760900881619347, "grad_norm": 2.031223773956299, "learning_rate": 4.017646216786544e-06, "loss": 0.1199730634689331, "step": 21870 }, { "epoch": 0.5763535038401066, "grad_norm": 1.834753155708313, "learning_rate": 4.013464028054814e-06, "loss": 0.13006072044372557, "step": 21880 }, { "epoch": 0.5766169195182785, "grad_norm": 2.5884554386138916, "learning_rate": 4.009282557366528e-06, "loss": 0.13327773809432983, "step": 21890 }, { "epoch": 0.5768803351964504, "grad_norm": 1.7873220443725586, "learning_rate": 4.0051018077651396e-06, "loss": 0.1262887239456177, "step": 21900 }, { "epoch": 0.5771437508746223, "grad_norm": 3.6306731700897217, "learning_rate": 4.00092178229358e-06, "loss": 0.13910949230194092, "step": 21910 }, { "epoch": 0.5774071665527942, "grad_norm": 2.809868812561035, "learning_rate": 3.996742483994252e-06, "loss": 0.12671215534210206, "step": 21920 }, { "epoch": 0.5776705822309661, "grad_norm": 2.4695005416870117, "learning_rate": 3.992563915909025e-06, "loss": 0.1363283157348633, "step": 21930 }, { "epoch": 0.577933997909138, "grad_norm": 1.3777921199798584, "learning_rate": 3.988386081079249e-06, "loss": 0.1337173819541931, "step": 21940 }, { "epoch": 0.5781974135873099, "grad_norm": 2.696974754333496, "learning_rate": 3.984208982545724e-06, "loss": 0.13579747676849366, "step": 21950 }, { "epoch": 0.5784608292654818, "grad_norm": 1.7722229957580566, "learning_rate": 3.980032623348728e-06, "loss": 0.11987957954406739, "step": 21960 }, { "epoch": 0.5787242449436537, "grad_norm": 4.758094787597656, "learning_rate": 3.975857006527994e-06, "loss": 0.13810080289840698, "step": 21970 }, { "epoch": 0.5789876606218256, "grad_norm": 2.497462749481201, "learning_rate": 3.971682135122715e-06, "loss": 0.12408257722854614, "step": 21980 }, { "epoch": 0.5792510762999975, "grad_norm": 3.5587899684906006, "learning_rate": 3.967508012171545e-06, "loss": 0.12880512475967407, "step": 21990 }, { "epoch": 0.5795144919781694, "grad_norm": 2.5186595916748047, "learning_rate": 3.963334640712589e-06, "loss": 0.1357123851776123, "step": 22000 }, { "epoch": 0.5797779076563413, "grad_norm": 2.0612847805023193, "learning_rate": 3.959162023783405e-06, "loss": 0.13223860263824463, "step": 22010 }, { "epoch": 0.5800413233345132, "grad_norm": 1.7861900329589844, "learning_rate": 3.954990164421007e-06, "loss": 0.12702382802963258, "step": 22020 }, { "epoch": 0.5803047390126851, "grad_norm": 2.203338861465454, "learning_rate": 3.9508190656618524e-06, "loss": 0.1283339262008667, "step": 22030 }, { "epoch": 0.580568154690857, "grad_norm": 2.217123031616211, "learning_rate": 3.946648730541849e-06, "loss": 0.12408993244171143, "step": 22040 }, { "epoch": 0.5808315703690289, "grad_norm": 2.5536532402038574, "learning_rate": 3.9424791620963444e-06, "loss": 0.1274345636367798, "step": 22050 }, { "epoch": 0.5810949860472008, "grad_norm": 2.0478968620300293, "learning_rate": 3.938310363360128e-06, "loss": 0.13318288326263428, "step": 22060 }, { "epoch": 0.5813584017253727, "grad_norm": 2.7748405933380127, "learning_rate": 3.9341423373674386e-06, "loss": 0.12087569236755372, "step": 22070 }, { "epoch": 0.5816218174035446, "grad_norm": 3.0599420070648193, "learning_rate": 3.9299750871519395e-06, "loss": 0.12809181213378906, "step": 22080 }, { "epoch": 0.5818852330817165, "grad_norm": 2.078580856323242, "learning_rate": 3.925808615746736e-06, "loss": 0.12570446729660034, "step": 22090 }, { "epoch": 0.5821486487598884, "grad_norm": 2.053379535675049, "learning_rate": 3.921642926184367e-06, "loss": 0.12978770732879638, "step": 22100 }, { "epoch": 0.5824120644380603, "grad_norm": 2.189293146133423, "learning_rate": 3.9174780214968e-06, "loss": 0.1184650182723999, "step": 22110 }, { "epoch": 0.5826754801162322, "grad_norm": 2.4717302322387695, "learning_rate": 3.913313904715434e-06, "loss": 0.12526307106018067, "step": 22120 }, { "epoch": 0.582938895794404, "grad_norm": 4.227387428283691, "learning_rate": 3.909150578871091e-06, "loss": 0.13350226879119872, "step": 22130 }, { "epoch": 0.583202311472576, "grad_norm": 2.003053665161133, "learning_rate": 3.904988046994019e-06, "loss": 0.12967095375061036, "step": 22140 }, { "epoch": 0.5834657271507478, "grad_norm": 1.5415750741958618, "learning_rate": 3.900826312113891e-06, "loss": 0.1280951976776123, "step": 22150 }, { "epoch": 0.5837291428289197, "grad_norm": 1.7290335893630981, "learning_rate": 3.896665377259795e-06, "loss": 0.13179235458374022, "step": 22160 }, { "epoch": 0.5839925585070916, "grad_norm": 2.596628427505493, "learning_rate": 3.892505245460239e-06, "loss": 0.11725041866302491, "step": 22170 }, { "epoch": 0.5842559741852635, "grad_norm": 2.2777435779571533, "learning_rate": 3.8883459197431485e-06, "loss": 0.12681028842926026, "step": 22180 }, { "epoch": 0.5845193898634354, "grad_norm": 3.076306104660034, "learning_rate": 3.884187403135855e-06, "loss": 0.122258460521698, "step": 22190 }, { "epoch": 0.5847828055416073, "grad_norm": 4.542661666870117, "learning_rate": 3.880029698665112e-06, "loss": 0.1292683005332947, "step": 22200 }, { "epoch": 0.5850462212197792, "grad_norm": 2.8762049674987793, "learning_rate": 3.875872809357075e-06, "loss": 0.13193390369415284, "step": 22210 }, { "epoch": 0.5853096368979511, "grad_norm": 2.4910225868225098, "learning_rate": 3.871716738237304e-06, "loss": 0.1347934603691101, "step": 22220 }, { "epoch": 0.585573052576123, "grad_norm": 1.7353469133377075, "learning_rate": 3.86756148833077e-06, "loss": 0.12562637329101561, "step": 22230 }, { "epoch": 0.5858364682542949, "grad_norm": 2.9017837047576904, "learning_rate": 3.863407062661842e-06, "loss": 0.13216594457626343, "step": 22240 }, { "epoch": 0.5860998839324668, "grad_norm": 2.431450605392456, "learning_rate": 3.859253464254291e-06, "loss": 0.12333316802978515, "step": 22250 }, { "epoch": 0.5863632996106387, "grad_norm": 2.0725481510162354, "learning_rate": 3.855100696131285e-06, "loss": 0.12868824005126953, "step": 22260 }, { "epoch": 0.5866267152888106, "grad_norm": 2.2778072357177734, "learning_rate": 3.850948761315386e-06, "loss": 0.1235345482826233, "step": 22270 }, { "epoch": 0.5868901309669825, "grad_norm": 1.9497935771942139, "learning_rate": 3.8467976628285544e-06, "loss": 0.12448123693466187, "step": 22280 }, { "epoch": 0.5871535466451544, "grad_norm": 2.3346807956695557, "learning_rate": 3.842647403692136e-06, "loss": 0.125064480304718, "step": 22290 }, { "epoch": 0.5874169623233263, "grad_norm": 1.7987090349197388, "learning_rate": 3.838497986926872e-06, "loss": 0.11004828214645386, "step": 22300 }, { "epoch": 0.5876803780014982, "grad_norm": 2.7511703968048096, "learning_rate": 3.834349415552884e-06, "loss": 0.13261985778808594, "step": 22310 }, { "epoch": 0.5879437936796701, "grad_norm": 2.4968044757843018, "learning_rate": 3.83020169258968e-06, "loss": 0.13703794479370118, "step": 22320 }, { "epoch": 0.588207209357842, "grad_norm": 2.2358906269073486, "learning_rate": 3.826054821056154e-06, "loss": 0.12774877548217772, "step": 22330 }, { "epoch": 0.5884706250360139, "grad_norm": 1.7466636896133423, "learning_rate": 3.821908803970578e-06, "loss": 0.11678352355957031, "step": 22340 }, { "epoch": 0.5887340407141858, "grad_norm": 1.8086521625518799, "learning_rate": 3.8177636443506004e-06, "loss": 0.12211852073669434, "step": 22350 }, { "epoch": 0.5889974563923577, "grad_norm": 4.526615619659424, "learning_rate": 3.8136193452132503e-06, "loss": 0.13346760272979735, "step": 22360 }, { "epoch": 0.5892608720705296, "grad_norm": 1.9881575107574463, "learning_rate": 3.809475909574924e-06, "loss": 0.1258033037185669, "step": 22370 }, { "epoch": 0.5895242877487015, "grad_norm": 2.7161970138549805, "learning_rate": 3.8053333404513953e-06, "loss": 0.12922383546829225, "step": 22380 }, { "epoch": 0.5897877034268734, "grad_norm": 1.8220045566558838, "learning_rate": 3.801191640857804e-06, "loss": 0.12802419662475586, "step": 22390 }, { "epoch": 0.5900511191050453, "grad_norm": 3.5105249881744385, "learning_rate": 3.797050813808656e-06, "loss": 0.1322782516479492, "step": 22400 }, { "epoch": 0.5903145347832172, "grad_norm": 1.9031541347503662, "learning_rate": 3.792910862317828e-06, "loss": 0.12465867996215821, "step": 22410 }, { "epoch": 0.5905779504613891, "grad_norm": 3.6875150203704834, "learning_rate": 3.7887717893985486e-06, "loss": 0.12958377599716187, "step": 22420 }, { "epoch": 0.590841366139561, "grad_norm": 2.437816619873047, "learning_rate": 3.784633598063422e-06, "loss": 0.1241357684135437, "step": 22430 }, { "epoch": 0.5911047818177328, "grad_norm": 2.3900794982910156, "learning_rate": 3.7804962913243965e-06, "loss": 0.12959365844726561, "step": 22440 }, { "epoch": 0.5913681974959047, "grad_norm": 2.394575595855713, "learning_rate": 3.776359872192783e-06, "loss": 0.12284499406814575, "step": 22450 }, { "epoch": 0.5916316131740766, "grad_norm": 4.903549671173096, "learning_rate": 3.7722243436792475e-06, "loss": 0.13827073574066162, "step": 22460 }, { "epoch": 0.5918950288522485, "grad_norm": 2.3692641258239746, "learning_rate": 3.768089708793806e-06, "loss": 0.12379606962203979, "step": 22470 }, { "epoch": 0.5921584445304204, "grad_norm": 2.224165678024292, "learning_rate": 3.763955970545822e-06, "loss": 0.12273013591766357, "step": 22480 }, { "epoch": 0.5924218602085923, "grad_norm": 2.2318661212921143, "learning_rate": 3.7598231319440116e-06, "loss": 0.13135216236114503, "step": 22490 }, { "epoch": 0.5926852758867642, "grad_norm": 2.928675651550293, "learning_rate": 3.7556911959964303e-06, "loss": 0.14510462284088135, "step": 22500 }, { "epoch": 0.5929486915649361, "grad_norm": 2.794027090072632, "learning_rate": 3.7515601657104815e-06, "loss": 0.12177119255065919, "step": 22510 }, { "epoch": 0.593212107243108, "grad_norm": 1.7826192378997803, "learning_rate": 3.7474300440929077e-06, "loss": 0.12508493661880493, "step": 22520 }, { "epoch": 0.5934755229212799, "grad_norm": 2.997476577758789, "learning_rate": 3.743300834149785e-06, "loss": 0.11897438764572144, "step": 22530 }, { "epoch": 0.5937389385994518, "grad_norm": 3.8211686611175537, "learning_rate": 3.7391725388865375e-06, "loss": 0.13730974197387696, "step": 22540 }, { "epoch": 0.5940023542776237, "grad_norm": 1.8420450687408447, "learning_rate": 3.735045161307911e-06, "loss": 0.11844782829284668, "step": 22550 }, { "epoch": 0.5942657699557956, "grad_norm": 4.944300174713135, "learning_rate": 3.7309187044179902e-06, "loss": 0.12582311630249024, "step": 22560 }, { "epoch": 0.5945291856339675, "grad_norm": 2.8229892253875732, "learning_rate": 3.72679317122019e-06, "loss": 0.1287372589111328, "step": 22570 }, { "epoch": 0.5947926013121394, "grad_norm": 2.165531873703003, "learning_rate": 3.72266856471725e-06, "loss": 0.13208975791931152, "step": 22580 }, { "epoch": 0.5950560169903112, "grad_norm": 2.864448308944702, "learning_rate": 3.7185448879112384e-06, "loss": 0.11269798278808593, "step": 22590 }, { "epoch": 0.5953194326684831, "grad_norm": 2.884035348892212, "learning_rate": 3.7144221438035432e-06, "loss": 0.1285560369491577, "step": 22600 }, { "epoch": 0.595582848346655, "grad_norm": 2.584427833557129, "learning_rate": 3.7103003353948765e-06, "loss": 0.12418673038482667, "step": 22610 }, { "epoch": 0.5958462640248269, "grad_norm": 1.7165073156356812, "learning_rate": 3.706179465685268e-06, "loss": 0.1320462703704834, "step": 22620 }, { "epoch": 0.5961096797029988, "grad_norm": 3.2471578121185303, "learning_rate": 3.7020595376740643e-06, "loss": 0.1390128254890442, "step": 22630 }, { "epoch": 0.5963730953811707, "grad_norm": 2.4147567749023438, "learning_rate": 3.6979405543599277e-06, "loss": 0.13784596920013428, "step": 22640 }, { "epoch": 0.5966365110593426, "grad_norm": 3.2297849655151367, "learning_rate": 3.6938225187408322e-06, "loss": 0.12606568336486818, "step": 22650 }, { "epoch": 0.5968999267375145, "grad_norm": 3.4991626739501953, "learning_rate": 3.6897054338140577e-06, "loss": 0.13155841827392578, "step": 22660 }, { "epoch": 0.5971633424156864, "grad_norm": 1.7457139492034912, "learning_rate": 3.6855893025762025e-06, "loss": 0.13113303184509278, "step": 22670 }, { "epoch": 0.5974267580938583, "grad_norm": 1.8928301334381104, "learning_rate": 3.6814741280231604e-06, "loss": 0.135604190826416, "step": 22680 }, { "epoch": 0.5976901737720302, "grad_norm": 2.2369179725646973, "learning_rate": 3.677359913150133e-06, "loss": 0.11485333442687988, "step": 22690 }, { "epoch": 0.597953589450202, "grad_norm": 2.6418075561523438, "learning_rate": 3.6732466609516255e-06, "loss": 0.13562785387039183, "step": 22700 }, { "epoch": 0.598217005128374, "grad_norm": 2.2143771648406982, "learning_rate": 3.669134374421438e-06, "loss": 0.11963753700256348, "step": 22710 }, { "epoch": 0.5984804208065458, "grad_norm": 2.676806926727295, "learning_rate": 3.6650230565526723e-06, "loss": 0.13327527046203613, "step": 22720 }, { "epoch": 0.5987438364847177, "grad_norm": 2.564476728439331, "learning_rate": 3.660912710337722e-06, "loss": 0.12471976280212402, "step": 22730 }, { "epoch": 0.5990072521628896, "grad_norm": 2.431180477142334, "learning_rate": 3.6568033387682728e-06, "loss": 0.12194406986236572, "step": 22740 }, { "epoch": 0.5992706678410615, "grad_norm": 2.2639153003692627, "learning_rate": 3.6526949448353054e-06, "loss": 0.12971887588500977, "step": 22750 }, { "epoch": 0.5995340835192334, "grad_norm": 2.4884438514709473, "learning_rate": 3.6485875315290838e-06, "loss": 0.12208094596862792, "step": 22760 }, { "epoch": 0.5997974991974053, "grad_norm": 2.3454301357269287, "learning_rate": 3.6444811018391623e-06, "loss": 0.12881826162338256, "step": 22770 }, { "epoch": 0.6000609148755772, "grad_norm": 4.1044158935546875, "learning_rate": 3.640375658754378e-06, "loss": 0.12772848606109619, "step": 22780 }, { "epoch": 0.6003243305537491, "grad_norm": 3.0759477615356445, "learning_rate": 3.6362712052628456e-06, "loss": 0.12375588417053222, "step": 22790 }, { "epoch": 0.600587746231921, "grad_norm": 2.6843836307525635, "learning_rate": 3.6321677443519697e-06, "loss": 0.14049363136291504, "step": 22800 }, { "epoch": 0.6008511619100929, "grad_norm": 1.8918125629425049, "learning_rate": 3.6280652790084216e-06, "loss": 0.1123171329498291, "step": 22810 }, { "epoch": 0.6011145775882648, "grad_norm": 1.8264336585998535, "learning_rate": 3.623963812218152e-06, "loss": 0.122064208984375, "step": 22820 }, { "epoch": 0.6013779932664367, "grad_norm": 4.605893611907959, "learning_rate": 3.6198633469663893e-06, "loss": 0.12271931171417236, "step": 22830 }, { "epoch": 0.6016414089446086, "grad_norm": 2.1430976390838623, "learning_rate": 3.6157638862376248e-06, "loss": 0.12433744668960571, "step": 22840 }, { "epoch": 0.6019048246227805, "grad_norm": 3.9196298122406006, "learning_rate": 3.6116654330156264e-06, "loss": 0.1341320514678955, "step": 22850 }, { "epoch": 0.6021682403009524, "grad_norm": 2.3716626167297363, "learning_rate": 3.6075679902834237e-06, "loss": 0.1217086911201477, "step": 22860 }, { "epoch": 0.6024316559791243, "grad_norm": 2.4169795513153076, "learning_rate": 3.6034715610233107e-06, "loss": 0.1294262170791626, "step": 22870 }, { "epoch": 0.6026950716572962, "grad_norm": 2.600573778152466, "learning_rate": 3.5993761482168486e-06, "loss": 0.12885284423828125, "step": 22880 }, { "epoch": 0.6029584873354681, "grad_norm": 3.164271593093872, "learning_rate": 3.5952817548448525e-06, "loss": 0.1331647515296936, "step": 22890 }, { "epoch": 0.60322190301364, "grad_norm": 2.2757606506347656, "learning_rate": 3.5911883838874016e-06, "loss": 0.11285326480865479, "step": 22900 }, { "epoch": 0.6034853186918119, "grad_norm": 2.470964193344116, "learning_rate": 3.5870960383238264e-06, "loss": 0.13413074016571044, "step": 22910 }, { "epoch": 0.6037487343699838, "grad_norm": 3.2275807857513428, "learning_rate": 3.5830047211327113e-06, "loss": 0.13061604499816895, "step": 22920 }, { "epoch": 0.6040121500481557, "grad_norm": 2.0661678314208984, "learning_rate": 3.578914435291895e-06, "loss": 0.1354766607284546, "step": 22930 }, { "epoch": 0.6042755657263276, "grad_norm": 4.668251037597656, "learning_rate": 3.574825183778463e-06, "loss": 0.12548720836639404, "step": 22940 }, { "epoch": 0.6045389814044995, "grad_norm": 3.47021746635437, "learning_rate": 3.570736969568749e-06, "loss": 0.12810604572296141, "step": 22950 }, { "epoch": 0.6048023970826714, "grad_norm": 2.2198047637939453, "learning_rate": 3.5666497956383327e-06, "loss": 0.13037872314453125, "step": 22960 }, { "epoch": 0.6050658127608433, "grad_norm": 3.0127902030944824, "learning_rate": 3.562563664962035e-06, "loss": 0.12136558294296265, "step": 22970 }, { "epoch": 0.6053292284390152, "grad_norm": 1.6122785806655884, "learning_rate": 3.5584785805139187e-06, "loss": 0.1250173807144165, "step": 22980 }, { "epoch": 0.605592644117187, "grad_norm": 4.222856044769287, "learning_rate": 3.5543945452672844e-06, "loss": 0.12161707878112793, "step": 22990 }, { "epoch": 0.605856059795359, "grad_norm": 2.0557656288146973, "learning_rate": 3.550311562194667e-06, "loss": 0.12732152938842772, "step": 23000 }, { "epoch": 0.6061194754735308, "grad_norm": 2.583043336868286, "learning_rate": 3.5462296342678426e-06, "loss": 0.1210411787033081, "step": 23010 }, { "epoch": 0.6063828911517027, "grad_norm": 2.5632660388946533, "learning_rate": 3.542148764457808e-06, "loss": 0.13166234493255616, "step": 23020 }, { "epoch": 0.6066463068298746, "grad_norm": 2.5352141857147217, "learning_rate": 3.5380689557348017e-06, "loss": 0.12997801303863527, "step": 23030 }, { "epoch": 0.6069097225080465, "grad_norm": 2.8083512783050537, "learning_rate": 3.5339902110682824e-06, "loss": 0.13225014209747316, "step": 23040 }, { "epoch": 0.6071731381862184, "grad_norm": 3.4644949436187744, "learning_rate": 3.529912533426936e-06, "loss": 0.13520503044128418, "step": 23050 }, { "epoch": 0.6074365538643903, "grad_norm": 2.9388885498046875, "learning_rate": 3.5258359257786723e-06, "loss": 0.1263191819190979, "step": 23060 }, { "epoch": 0.6076999695425622, "grad_norm": 2.367889642715454, "learning_rate": 3.521760391090624e-06, "loss": 0.1196635127067566, "step": 23070 }, { "epoch": 0.6079633852207341, "grad_norm": 2.1362061500549316, "learning_rate": 3.517685932329138e-06, "loss": 0.11747736930847168, "step": 23080 }, { "epoch": 0.608226800898906, "grad_norm": 2.7506356239318848, "learning_rate": 3.5136125524597842e-06, "loss": 0.12283203601837159, "step": 23090 }, { "epoch": 0.6084902165770779, "grad_norm": 2.12363338470459, "learning_rate": 3.5095402544473415e-06, "loss": 0.13028972148895263, "step": 23100 }, { "epoch": 0.6087536322552498, "grad_norm": 2.781496524810791, "learning_rate": 3.505469041255807e-06, "loss": 0.122990882396698, "step": 23110 }, { "epoch": 0.6090170479334217, "grad_norm": 2.558166027069092, "learning_rate": 3.5013989158483853e-06, "loss": 0.12411248683929443, "step": 23120 }, { "epoch": 0.6092804636115936, "grad_norm": 2.374274492263794, "learning_rate": 3.4973298811874834e-06, "loss": 0.12141618728637696, "step": 23130 }, { "epoch": 0.6095438792897655, "grad_norm": 2.3881947994232178, "learning_rate": 3.493261940234729e-06, "loss": 0.10265347957611085, "step": 23140 }, { "epoch": 0.6098072949679374, "grad_norm": 2.448676586151123, "learning_rate": 3.489195095950939e-06, "loss": 0.11386871337890625, "step": 23150 }, { "epoch": 0.6100707106461093, "grad_norm": 1.797037124633789, "learning_rate": 3.4851293512961385e-06, "loss": 0.1182174563407898, "step": 23160 }, { "epoch": 0.6103341263242812, "grad_norm": 1.8615537881851196, "learning_rate": 3.4810647092295546e-06, "loss": 0.11780400276184082, "step": 23170 }, { "epoch": 0.6105975420024531, "grad_norm": 1.5518016815185547, "learning_rate": 3.477001172709607e-06, "loss": 0.11924316883087158, "step": 23180 }, { "epoch": 0.610860957680625, "grad_norm": 2.1772754192352295, "learning_rate": 3.4729387446939146e-06, "loss": 0.11596050262451171, "step": 23190 }, { "epoch": 0.6111243733587969, "grad_norm": 4.406439304351807, "learning_rate": 3.4688774281392867e-06, "loss": 0.13890573978424073, "step": 23200 }, { "epoch": 0.6113877890369688, "grad_norm": 3.7423558235168457, "learning_rate": 3.4648172260017236e-06, "loss": 0.12223515510559083, "step": 23210 }, { "epoch": 0.6116512047151407, "grad_norm": 2.4099833965301514, "learning_rate": 3.4607581412364184e-06, "loss": 0.11184835433959961, "step": 23220 }, { "epoch": 0.6119146203933126, "grad_norm": 2.7842750549316406, "learning_rate": 3.456700176797745e-06, "loss": 0.13279592990875244, "step": 23230 }, { "epoch": 0.6121780360714845, "grad_norm": 3.3651652336120605, "learning_rate": 3.452643335639268e-06, "loss": 0.12687897682189941, "step": 23240 }, { "epoch": 0.6124414517496564, "grad_norm": 2.0658223628997803, "learning_rate": 3.448587620713731e-06, "loss": 0.11800944805145264, "step": 23250 }, { "epoch": 0.6127048674278283, "grad_norm": 2.5865256786346436, "learning_rate": 3.4445330349730537e-06, "loss": 0.13187053203582763, "step": 23260 }, { "epoch": 0.6129682831060002, "grad_norm": 2.0853512287139893, "learning_rate": 3.4404795813683457e-06, "loss": 0.12467128038406372, "step": 23270 }, { "epoch": 0.613231698784172, "grad_norm": 2.428072690963745, "learning_rate": 3.4364272628498805e-06, "loss": 0.12122999429702759, "step": 23280 }, { "epoch": 0.613495114462344, "grad_norm": 3.1884171962738037, "learning_rate": 3.4323760823671102e-06, "loss": 0.13016290664672853, "step": 23290 }, { "epoch": 0.6137585301405158, "grad_norm": 2.4736034870147705, "learning_rate": 3.4283260428686616e-06, "loss": 0.12387146949768066, "step": 23300 }, { "epoch": 0.6140219458186877, "grad_norm": 2.456927537918091, "learning_rate": 3.424277147302326e-06, "loss": 0.12008165121078491, "step": 23310 }, { "epoch": 0.6142853614968596, "grad_norm": 4.318734169006348, "learning_rate": 3.4202293986150658e-06, "loss": 0.13773094415664672, "step": 23320 }, { "epoch": 0.6145487771750315, "grad_norm": 4.734594345092773, "learning_rate": 3.4161827997530068e-06, "loss": 0.11647615432739258, "step": 23330 }, { "epoch": 0.6148121928532034, "grad_norm": 2.4321818351745605, "learning_rate": 3.412137353661437e-06, "loss": 0.1284739851951599, "step": 23340 }, { "epoch": 0.6150756085313753, "grad_norm": 2.55171799659729, "learning_rate": 3.4080930632848076e-06, "loss": 0.1120875358581543, "step": 23350 }, { "epoch": 0.6153390242095472, "grad_norm": 2.94914174079895, "learning_rate": 3.4040499315667264e-06, "loss": 0.11644241809844971, "step": 23360 }, { "epoch": 0.6156024398877191, "grad_norm": 2.3274481296539307, "learning_rate": 3.4000079614499613e-06, "loss": 0.12558523416519166, "step": 23370 }, { "epoch": 0.6158658555658909, "grad_norm": 2.3487744331359863, "learning_rate": 3.3959671558764328e-06, "loss": 0.11588023900985718, "step": 23380 }, { "epoch": 0.6161292712440628, "grad_norm": 3.472057819366455, "learning_rate": 3.391927517787209e-06, "loss": 0.1303808331489563, "step": 23390 }, { "epoch": 0.6163926869222347, "grad_norm": 2.496522903442383, "learning_rate": 3.387889050122519e-06, "loss": 0.12238261699676514, "step": 23400 }, { "epoch": 0.6166561026004066, "grad_norm": 3.0535645484924316, "learning_rate": 3.38385175582173e-06, "loss": 0.13920724391937256, "step": 23410 }, { "epoch": 0.6169195182785785, "grad_norm": 1.7302124500274658, "learning_rate": 3.3798156378233576e-06, "loss": 0.1185429573059082, "step": 23420 }, { "epoch": 0.6171829339567504, "grad_norm": 3.3547096252441406, "learning_rate": 3.3757806990650664e-06, "loss": 0.11007449626922608, "step": 23430 }, { "epoch": 0.6174463496349223, "grad_norm": 2.8807621002197266, "learning_rate": 3.371746942483656e-06, "loss": 0.13201302289962769, "step": 23440 }, { "epoch": 0.6177097653130942, "grad_norm": 2.4488303661346436, "learning_rate": 3.3677143710150705e-06, "loss": 0.12725274562835692, "step": 23450 }, { "epoch": 0.6179731809912661, "grad_norm": 2.878904342651367, "learning_rate": 3.3636829875943887e-06, "loss": 0.1425493836402893, "step": 23460 }, { "epoch": 0.618236596669438, "grad_norm": 2.837351083755493, "learning_rate": 3.3596527951558246e-06, "loss": 0.13250598907470704, "step": 23470 }, { "epoch": 0.6185000123476099, "grad_norm": 2.0364131927490234, "learning_rate": 3.3556237966327277e-06, "loss": 0.12187659740447998, "step": 23480 }, { "epoch": 0.6187634280257818, "grad_norm": 4.47922945022583, "learning_rate": 3.351595994957575e-06, "loss": 0.12007933855056763, "step": 23490 }, { "epoch": 0.6190268437039537, "grad_norm": 2.6108498573303223, "learning_rate": 3.347569393061976e-06, "loss": 0.1379100799560547, "step": 23500 }, { "epoch": 0.6192902593821256, "grad_norm": 2.513328790664673, "learning_rate": 3.3435439938766656e-06, "loss": 0.1305760622024536, "step": 23510 }, { "epoch": 0.6195536750602975, "grad_norm": 2.495913028717041, "learning_rate": 3.3395198003315005e-06, "loss": 0.12748534679412843, "step": 23520 }, { "epoch": 0.6198170907384694, "grad_norm": 2.54156494140625, "learning_rate": 3.3354968153554646e-06, "loss": 0.13134814500808717, "step": 23530 }, { "epoch": 0.6200805064166413, "grad_norm": 2.965956211090088, "learning_rate": 3.3314750418766594e-06, "loss": 0.10730558633804321, "step": 23540 }, { "epoch": 0.6203439220948131, "grad_norm": 3.4443209171295166, "learning_rate": 3.3274544828223044e-06, "loss": 0.11522771120071411, "step": 23550 }, { "epoch": 0.620607337772985, "grad_norm": 3.1467788219451904, "learning_rate": 3.323435141118737e-06, "loss": 0.12207612991333008, "step": 23560 }, { "epoch": 0.6208707534511569, "grad_norm": 2.4730241298675537, "learning_rate": 3.3194170196914063e-06, "loss": 0.12694976329803467, "step": 23570 }, { "epoch": 0.6211341691293288, "grad_norm": 2.0990047454833984, "learning_rate": 3.3154001214648758e-06, "loss": 0.13738300800323486, "step": 23580 }, { "epoch": 0.6213975848075007, "grad_norm": 2.582045078277588, "learning_rate": 3.311384449362817e-06, "loss": 0.11470125913619995, "step": 23590 }, { "epoch": 0.6216610004856726, "grad_norm": 2.4275944232940674, "learning_rate": 3.3073700063080084e-06, "loss": 0.12161095142364502, "step": 23600 }, { "epoch": 0.6219244161638445, "grad_norm": 5.335278511047363, "learning_rate": 3.3033567952223365e-06, "loss": 0.12367000579833984, "step": 23610 }, { "epoch": 0.6221878318420164, "grad_norm": 2.4545369148254395, "learning_rate": 3.2993448190267886e-06, "loss": 0.13421206474304198, "step": 23620 }, { "epoch": 0.6224512475201883, "grad_norm": 2.973435640335083, "learning_rate": 3.2953340806414517e-06, "loss": 0.12166861295700074, "step": 23630 }, { "epoch": 0.6227146631983602, "grad_norm": 2.756399393081665, "learning_rate": 3.2913245829855183e-06, "loss": 0.12215623855590821, "step": 23640 }, { "epoch": 0.6229780788765321, "grad_norm": 3.282759428024292, "learning_rate": 3.287316328977271e-06, "loss": 0.11498003005981446, "step": 23650 }, { "epoch": 0.623241494554704, "grad_norm": 2.8214809894561768, "learning_rate": 3.2833093215340918e-06, "loss": 0.11918017864227295, "step": 23660 }, { "epoch": 0.6235049102328759, "grad_norm": 2.4085898399353027, "learning_rate": 3.2793035635724524e-06, "loss": 0.11552636623382569, "step": 23670 }, { "epoch": 0.6237683259110478, "grad_norm": 2.9338810443878174, "learning_rate": 3.275299058007916e-06, "loss": 0.12741081714630126, "step": 23680 }, { "epoch": 0.6240317415892197, "grad_norm": 2.459594488143921, "learning_rate": 3.271295807755135e-06, "loss": 0.12024431228637696, "step": 23690 }, { "epoch": 0.6242951572673916, "grad_norm": 4.610580921173096, "learning_rate": 3.267293815727848e-06, "loss": 0.13162194490432738, "step": 23700 }, { "epoch": 0.6245585729455635, "grad_norm": 2.3511743545532227, "learning_rate": 3.2632930848388778e-06, "loss": 0.12141695022583007, "step": 23710 }, { "epoch": 0.6248219886237354, "grad_norm": 4.917429447174072, "learning_rate": 3.2592936180001294e-06, "loss": 0.12422250509262085, "step": 23720 }, { "epoch": 0.6250854043019073, "grad_norm": 2.5514767169952393, "learning_rate": 3.255295418122583e-06, "loss": 0.1301575183868408, "step": 23730 }, { "epoch": 0.6253488199800792, "grad_norm": 2.2970054149627686, "learning_rate": 3.251298488116308e-06, "loss": 0.12139105796813965, "step": 23740 }, { "epoch": 0.6256122356582511, "grad_norm": 3.113424777984619, "learning_rate": 3.247302830890437e-06, "loss": 0.12495527267456055, "step": 23750 }, { "epoch": 0.625875651336423, "grad_norm": 3.6112256050109863, "learning_rate": 3.2433084493531827e-06, "loss": 0.11417839527130128, "step": 23760 }, { "epoch": 0.6261390670145949, "grad_norm": 3.34676194190979, "learning_rate": 3.23931534641183e-06, "loss": 0.11126763820648193, "step": 23770 }, { "epoch": 0.6264024826927668, "grad_norm": 2.6432557106018066, "learning_rate": 3.23532352497273e-06, "loss": 0.11676840782165528, "step": 23780 }, { "epoch": 0.6266658983709387, "grad_norm": 2.357750654220581, "learning_rate": 3.2313329879413037e-06, "loss": 0.12608895301818848, "step": 23790 }, { "epoch": 0.6269293140491106, "grad_norm": 3.3793115615844727, "learning_rate": 3.2273437382220353e-06, "loss": 0.12386773824691773, "step": 23800 }, { "epoch": 0.6271927297272825, "grad_norm": 2.586679220199585, "learning_rate": 3.223355778718472e-06, "loss": 0.12701816558837892, "step": 23810 }, { "epoch": 0.6274561454054544, "grad_norm": 2.380385398864746, "learning_rate": 3.219369112333226e-06, "loss": 0.12230054140090943, "step": 23820 }, { "epoch": 0.6277195610836263, "grad_norm": 1.6929152011871338, "learning_rate": 3.2153837419679612e-06, "loss": 0.11891394853591919, "step": 23830 }, { "epoch": 0.6279829767617982, "grad_norm": 1.7673449516296387, "learning_rate": 3.211399670523405e-06, "loss": 0.12221062183380127, "step": 23840 }, { "epoch": 0.62824639243997, "grad_norm": 3.114377975463867, "learning_rate": 3.2074169008993363e-06, "loss": 0.131032395362854, "step": 23850 }, { "epoch": 0.6285098081181419, "grad_norm": 2.8875365257263184, "learning_rate": 3.203435435994583e-06, "loss": 0.13271820545196533, "step": 23860 }, { "epoch": 0.6287732237963138, "grad_norm": 4.523441314697266, "learning_rate": 3.199455278707032e-06, "loss": 0.11486630439758301, "step": 23870 }, { "epoch": 0.6290366394744857, "grad_norm": 3.2643020153045654, "learning_rate": 3.195476431933612e-06, "loss": 0.12325931787490844, "step": 23880 }, { "epoch": 0.6293000551526576, "grad_norm": 4.585756778717041, "learning_rate": 3.1914988985702953e-06, "loss": 0.1230825662612915, "step": 23890 }, { "epoch": 0.6295634708308295, "grad_norm": 2.2555787563323975, "learning_rate": 3.1875226815121085e-06, "loss": 0.12313153743743896, "step": 23900 }, { "epoch": 0.6298268865090014, "grad_norm": 2.6201295852661133, "learning_rate": 3.1835477836531093e-06, "loss": 0.12838162183761598, "step": 23910 }, { "epoch": 0.6300903021871733, "grad_norm": 2.145540714263916, "learning_rate": 3.1795742078864023e-06, "loss": 0.106130051612854, "step": 23920 }, { "epoch": 0.6303537178653452, "grad_norm": 2.923250436782837, "learning_rate": 3.175601957104127e-06, "loss": 0.11295863389968872, "step": 23930 }, { "epoch": 0.6306171335435171, "grad_norm": 4.316046714782715, "learning_rate": 3.171631034197457e-06, "loss": 0.11873629093170165, "step": 23940 }, { "epoch": 0.630880549221689, "grad_norm": 2.8914637565612793, "learning_rate": 3.167661442056604e-06, "loss": 0.1252763032913208, "step": 23950 }, { "epoch": 0.6311439648998609, "grad_norm": 2.658574104309082, "learning_rate": 3.1636931835708073e-06, "loss": 0.1295474052429199, "step": 23960 }, { "epoch": 0.6314073805780328, "grad_norm": 2.154350519180298, "learning_rate": 3.1597262616283365e-06, "loss": 0.12613314390182495, "step": 23970 }, { "epoch": 0.6316707962562047, "grad_norm": 2.705881118774414, "learning_rate": 3.15576067911649e-06, "loss": 0.13527780771255493, "step": 23980 }, { "epoch": 0.6319342119343766, "grad_norm": 2.0605289936065674, "learning_rate": 3.1517964389215854e-06, "loss": 0.11365605592727661, "step": 23990 }, { "epoch": 0.6321976276125485, "grad_norm": 2.609292984008789, "learning_rate": 3.147833543928974e-06, "loss": 0.11431009769439697, "step": 24000 }, { "epoch": 0.6324610432907204, "grad_norm": 2.476713180541992, "learning_rate": 3.1438719970230173e-06, "loss": 0.12263433933258057, "step": 24010 }, { "epoch": 0.6327244589688923, "grad_norm": 5.001091480255127, "learning_rate": 3.139911801087101e-06, "loss": 0.1214903473854065, "step": 24020 }, { "epoch": 0.6329878746470642, "grad_norm": 2.623734712600708, "learning_rate": 3.1359529590036277e-06, "loss": 0.1110072374343872, "step": 24030 }, { "epoch": 0.6332512903252361, "grad_norm": 3.158118486404419, "learning_rate": 3.131995473654011e-06, "loss": 0.1048850655555725, "step": 24040 }, { "epoch": 0.633514706003408, "grad_norm": 2.0271780490875244, "learning_rate": 3.1280393479186823e-06, "loss": 0.11439371109008789, "step": 24050 }, { "epoch": 0.6337781216815799, "grad_norm": 3.3587658405303955, "learning_rate": 3.1240845846770795e-06, "loss": 0.12994734048843384, "step": 24060 }, { "epoch": 0.6340415373597518, "grad_norm": 3.2419753074645996, "learning_rate": 3.120131186807649e-06, "loss": 0.11930358409881592, "step": 24070 }, { "epoch": 0.6343049530379237, "grad_norm": 2.3114473819732666, "learning_rate": 3.1161791571878465e-06, "loss": 0.12131333351135254, "step": 24080 }, { "epoch": 0.6345683687160956, "grad_norm": 3.206367254257202, "learning_rate": 3.112228498694127e-06, "loss": 0.12943692207336427, "step": 24090 }, { "epoch": 0.6348317843942675, "grad_norm": 2.388355255126953, "learning_rate": 3.1082792142019523e-06, "loss": 0.09998196959495545, "step": 24100 }, { "epoch": 0.6350952000724394, "grad_norm": 2.462841749191284, "learning_rate": 3.104331306585783e-06, "loss": 0.12988637685775756, "step": 24110 }, { "epoch": 0.6353586157506113, "grad_norm": 2.7204349040985107, "learning_rate": 3.100384778719072e-06, "loss": 0.12684280872344972, "step": 24120 }, { "epoch": 0.6356220314287832, "grad_norm": 3.688076972961426, "learning_rate": 3.096439633474277e-06, "loss": 0.1166692852973938, "step": 24130 }, { "epoch": 0.635885447106955, "grad_norm": 3.0794143676757812, "learning_rate": 3.092495873722843e-06, "loss": 0.13378913402557374, "step": 24140 }, { "epoch": 0.636148862785127, "grad_norm": 2.775099992752075, "learning_rate": 3.0885535023352083e-06, "loss": 0.11741632223129272, "step": 24150 }, { "epoch": 0.6364122784632988, "grad_norm": 2.5943403244018555, "learning_rate": 3.084612522180802e-06, "loss": 0.12390294075012206, "step": 24160 }, { "epoch": 0.6366756941414706, "grad_norm": 3.0607879161834717, "learning_rate": 3.0806729361280376e-06, "loss": 0.12245540618896485, "step": 24170 }, { "epoch": 0.6369391098196425, "grad_norm": 2.628296375274658, "learning_rate": 3.076734747044319e-06, "loss": 0.12055552005767822, "step": 24180 }, { "epoch": 0.6372025254978144, "grad_norm": 2.496307849884033, "learning_rate": 3.0727979577960277e-06, "loss": 0.11185879707336426, "step": 24190 }, { "epoch": 0.6374659411759863, "grad_norm": 3.649921178817749, "learning_rate": 3.0688625712485278e-06, "loss": 0.11775965690612793, "step": 24200 }, { "epoch": 0.6377293568541582, "grad_norm": 3.215426206588745, "learning_rate": 3.0649285902661672e-06, "loss": 0.1230310320854187, "step": 24210 }, { "epoch": 0.6379927725323301, "grad_norm": 2.3270530700683594, "learning_rate": 3.0609960177122643e-06, "loss": 0.1154504656791687, "step": 24220 }, { "epoch": 0.638256188210502, "grad_norm": 2.613698959350586, "learning_rate": 3.0570648564491133e-06, "loss": 0.1270003318786621, "step": 24230 }, { "epoch": 0.6385196038886739, "grad_norm": 2.9012203216552734, "learning_rate": 3.0531351093379876e-06, "loss": 0.11346316337585449, "step": 24240 }, { "epoch": 0.6387830195668458, "grad_norm": 2.0433974266052246, "learning_rate": 3.049206779239122e-06, "loss": 0.1113702893257141, "step": 24250 }, { "epoch": 0.6390464352450177, "grad_norm": 3.531996250152588, "learning_rate": 3.045279869011728e-06, "loss": 0.0959729552268982, "step": 24260 }, { "epoch": 0.6393098509231896, "grad_norm": 3.151883125305176, "learning_rate": 3.0413543815139802e-06, "loss": 0.1256310224533081, "step": 24270 }, { "epoch": 0.6395732666013615, "grad_norm": 2.6928093433380127, "learning_rate": 3.037430319603014e-06, "loss": 0.12238354682922363, "step": 24280 }, { "epoch": 0.6398366822795334, "grad_norm": 2.834062337875366, "learning_rate": 3.033507686134936e-06, "loss": 0.11792067289352418, "step": 24290 }, { "epoch": 0.6401000979577053, "grad_norm": 3.1054563522338867, "learning_rate": 3.0295864839648044e-06, "loss": 0.11254940032958985, "step": 24300 }, { "epoch": 0.6403635136358772, "grad_norm": 2.719982385635376, "learning_rate": 3.0256667159466412e-06, "loss": 0.10870676040649414, "step": 24310 }, { "epoch": 0.6406269293140491, "grad_norm": 1.9632734060287476, "learning_rate": 3.021748384933424e-06, "loss": 0.12608575820922852, "step": 24320 }, { "epoch": 0.640890344992221, "grad_norm": 3.848644256591797, "learning_rate": 3.017831493777077e-06, "loss": 0.10948371887207031, "step": 24330 }, { "epoch": 0.6411537606703929, "grad_norm": 2.8230934143066406, "learning_rate": 3.013916045328491e-06, "loss": 0.11746448278427124, "step": 24340 }, { "epoch": 0.6414171763485648, "grad_norm": 3.008772850036621, "learning_rate": 3.010002042437492e-06, "loss": 0.13041865825653076, "step": 24350 }, { "epoch": 0.6416805920267367, "grad_norm": 1.6233799457550049, "learning_rate": 3.0060894879528625e-06, "loss": 0.11208891868591309, "step": 24360 }, { "epoch": 0.6419440077049086, "grad_norm": 3.562762975692749, "learning_rate": 3.0021783847223283e-06, "loss": 0.11429373025894166, "step": 24370 }, { "epoch": 0.6422074233830805, "grad_norm": 3.4432666301727295, "learning_rate": 2.9982687355925588e-06, "loss": 0.12796835899353026, "step": 24380 }, { "epoch": 0.6424708390612524, "grad_norm": 3.103698492050171, "learning_rate": 2.9943605434091662e-06, "loss": 0.11936488151550292, "step": 24390 }, { "epoch": 0.6427342547394242, "grad_norm": 3.6637706756591797, "learning_rate": 2.9904538110167012e-06, "loss": 0.127247953414917, "step": 24400 }, { "epoch": 0.6429976704175961, "grad_norm": 3.5430774688720703, "learning_rate": 2.9865485412586503e-06, "loss": 0.12560734748840333, "step": 24410 }, { "epoch": 0.643261086095768, "grad_norm": 2.384099006652832, "learning_rate": 2.9826447369774397e-06, "loss": 0.11653772592544556, "step": 24420 }, { "epoch": 0.6435245017739399, "grad_norm": 2.568544387817383, "learning_rate": 2.978742401014425e-06, "loss": 0.12924363613128662, "step": 24430 }, { "epoch": 0.6437879174521118, "grad_norm": 2.9251511096954346, "learning_rate": 2.974841536209896e-06, "loss": 0.10754568576812744, "step": 24440 }, { "epoch": 0.6440513331302837, "grad_norm": 1.861548662185669, "learning_rate": 2.970942145403071e-06, "loss": 0.10745203495025635, "step": 24450 }, { "epoch": 0.6443147488084556, "grad_norm": 3.764988660812378, "learning_rate": 2.9670442314320903e-06, "loss": 0.10650243759155273, "step": 24460 }, { "epoch": 0.6445781644866275, "grad_norm": 2.6557018756866455, "learning_rate": 2.9631477971340295e-06, "loss": 0.1120842695236206, "step": 24470 }, { "epoch": 0.6448415801647994, "grad_norm": 2.5983035564422607, "learning_rate": 2.9592528453448793e-06, "loss": 0.11742967367172241, "step": 24480 }, { "epoch": 0.6451049958429713, "grad_norm": 2.6013574600219727, "learning_rate": 2.9553593788995523e-06, "loss": 0.1082618236541748, "step": 24490 }, { "epoch": 0.6453684115211432, "grad_norm": 3.814281940460205, "learning_rate": 2.951467400631883e-06, "loss": 0.11635637283325195, "step": 24500 }, { "epoch": 0.6456318271993151, "grad_norm": 2.469648599624634, "learning_rate": 2.94757691337462e-06, "loss": 0.11566181182861328, "step": 24510 }, { "epoch": 0.645895242877487, "grad_norm": 3.7861592769622803, "learning_rate": 2.94368791995943e-06, "loss": 0.1111712098121643, "step": 24520 }, { "epoch": 0.6461586585556589, "grad_norm": 2.881183624267578, "learning_rate": 2.9398004232168885e-06, "loss": 0.1272052526473999, "step": 24530 }, { "epoch": 0.6464220742338308, "grad_norm": 3.5110080242156982, "learning_rate": 2.9359144259764827e-06, "loss": 0.12110795974731445, "step": 24540 }, { "epoch": 0.6466854899120027, "grad_norm": 3.2635984420776367, "learning_rate": 2.9320299310666106e-06, "loss": 0.11259396076202392, "step": 24550 }, { "epoch": 0.6469489055901746, "grad_norm": 2.4563848972320557, "learning_rate": 2.9281469413145735e-06, "loss": 0.10924066305160522, "step": 24560 }, { "epoch": 0.6472123212683465, "grad_norm": 3.34437894821167, "learning_rate": 2.924265459546581e-06, "loss": 0.12706785202026366, "step": 24570 }, { "epoch": 0.6474757369465184, "grad_norm": 2.522022008895874, "learning_rate": 2.9203854885877425e-06, "loss": 0.11784616708755494, "step": 24580 }, { "epoch": 0.6477391526246903, "grad_norm": 1.9957727193832397, "learning_rate": 2.9165070312620645e-06, "loss": 0.11378804445266724, "step": 24590 }, { "epoch": 0.6480025683028622, "grad_norm": 3.077784538269043, "learning_rate": 2.9126300903924617e-06, "loss": 0.12464244365692138, "step": 24600 }, { "epoch": 0.6482659839810341, "grad_norm": 3.970508337020874, "learning_rate": 2.9087546688007374e-06, "loss": 0.10963358879089355, "step": 24610 }, { "epoch": 0.648529399659206, "grad_norm": 2.9565441608428955, "learning_rate": 2.9048807693075843e-06, "loss": 0.12117307186126709, "step": 24620 }, { "epoch": 0.6487928153373779, "grad_norm": 2.3716745376586914, "learning_rate": 2.9010083947326018e-06, "loss": 0.12062971591949463, "step": 24630 }, { "epoch": 0.6490562310155498, "grad_norm": 2.2692832946777344, "learning_rate": 2.897137547894265e-06, "loss": 0.10543913841247558, "step": 24640 }, { "epoch": 0.6493196466937217, "grad_norm": 2.9198529720306396, "learning_rate": 2.8932682316099497e-06, "loss": 0.1192082405090332, "step": 24650 }, { "epoch": 0.6495830623718936, "grad_norm": 3.073503255844116, "learning_rate": 2.889400448695906e-06, "loss": 0.11182321310043335, "step": 24660 }, { "epoch": 0.6498464780500655, "grad_norm": 4.688616752624512, "learning_rate": 2.8855342019672727e-06, "loss": 0.11771847009658813, "step": 24670 }, { "epoch": 0.6501098937282374, "grad_norm": 2.4727277755737305, "learning_rate": 2.881669494238075e-06, "loss": 0.11775968074798585, "step": 24680 }, { "epoch": 0.6503733094064093, "grad_norm": 3.192176580429077, "learning_rate": 2.877806328321212e-06, "loss": 0.12153617143630982, "step": 24690 }, { "epoch": 0.6506367250845811, "grad_norm": 2.7756457328796387, "learning_rate": 2.8739447070284626e-06, "loss": 0.12455902099609376, "step": 24700 }, { "epoch": 0.650900140762753, "grad_norm": 3.695838451385498, "learning_rate": 2.8700846331704814e-06, "loss": 0.12373428344726563, "step": 24710 }, { "epoch": 0.6511635564409249, "grad_norm": 2.554427146911621, "learning_rate": 2.8662261095567945e-06, "loss": 0.1104500412940979, "step": 24720 }, { "epoch": 0.6514269721190968, "grad_norm": 3.28295636177063, "learning_rate": 2.8623691389958054e-06, "loss": 0.108315110206604, "step": 24730 }, { "epoch": 0.6516903877972687, "grad_norm": 2.7121622562408447, "learning_rate": 2.858513724294784e-06, "loss": 0.11914795637130737, "step": 24740 }, { "epoch": 0.6519538034754406, "grad_norm": 4.29885721206665, "learning_rate": 2.8546598682598626e-06, "loss": 0.10332653522491456, "step": 24750 }, { "epoch": 0.6522172191536125, "grad_norm": 2.8944320678710938, "learning_rate": 2.850807573696048e-06, "loss": 0.10852460861206055, "step": 24760 }, { "epoch": 0.6524806348317844, "grad_norm": 3.372185230255127, "learning_rate": 2.8469568434072037e-06, "loss": 0.10881602764129639, "step": 24770 }, { "epoch": 0.6527440505099563, "grad_norm": 2.374237060546875, "learning_rate": 2.8431076801960634e-06, "loss": 0.10632716417312622, "step": 24780 }, { "epoch": 0.6530074661881282, "grad_norm": 2.966813564300537, "learning_rate": 2.839260086864207e-06, "loss": 0.11166094541549683, "step": 24790 }, { "epoch": 0.6532708818663001, "grad_norm": 3.482280969619751, "learning_rate": 2.83541406621208e-06, "loss": 0.11436113119125366, "step": 24800 }, { "epoch": 0.653534297544472, "grad_norm": 4.343987941741943, "learning_rate": 2.8315696210389855e-06, "loss": 0.10960443019866943, "step": 24810 }, { "epoch": 0.6537977132226439, "grad_norm": 2.854264259338379, "learning_rate": 2.8277267541430748e-06, "loss": 0.12261903285980225, "step": 24820 }, { "epoch": 0.6540611289008158, "grad_norm": 5.584127426147461, "learning_rate": 2.8238854683213524e-06, "loss": 0.1306680917739868, "step": 24830 }, { "epoch": 0.6543245445789877, "grad_norm": 4.114769458770752, "learning_rate": 2.8200457663696718e-06, "loss": 0.11208564043045044, "step": 24840 }, { "epoch": 0.6545879602571596, "grad_norm": 2.292876958847046, "learning_rate": 2.816207651082732e-06, "loss": 0.1203843593597412, "step": 24850 }, { "epoch": 0.6548513759353315, "grad_norm": 2.8770806789398193, "learning_rate": 2.812371125254082e-06, "loss": 0.1257917881011963, "step": 24860 }, { "epoch": 0.6551147916135034, "grad_norm": 3.153813600540161, "learning_rate": 2.808536191676111e-06, "loss": 0.10686259269714356, "step": 24870 }, { "epoch": 0.6553782072916753, "grad_norm": 2.686276912689209, "learning_rate": 2.8047028531400443e-06, "loss": 0.12878574132919313, "step": 24880 }, { "epoch": 0.6556416229698472, "grad_norm": 3.388819694519043, "learning_rate": 2.8008711124359555e-06, "loss": 0.13049275875091554, "step": 24890 }, { "epoch": 0.6559050386480191, "grad_norm": 2.7866709232330322, "learning_rate": 2.797040972352748e-06, "loss": 0.10994243621826172, "step": 24900 }, { "epoch": 0.656168454326191, "grad_norm": 3.2714884281158447, "learning_rate": 2.7932124356781686e-06, "loss": 0.1202237844467163, "step": 24910 }, { "epoch": 0.6564318700043629, "grad_norm": 2.6273765563964844, "learning_rate": 2.789385505198786e-06, "loss": 0.11073067188262939, "step": 24920 }, { "epoch": 0.6566952856825348, "grad_norm": 3.1556880474090576, "learning_rate": 2.785560183700004e-06, "loss": 0.1294865846633911, "step": 24930 }, { "epoch": 0.6569587013607067, "grad_norm": 6.235063552856445, "learning_rate": 2.7817364739660614e-06, "loss": 0.11986312866210938, "step": 24940 }, { "epoch": 0.6572221170388786, "grad_norm": 2.1650497913360596, "learning_rate": 2.7779143787800174e-06, "loss": 0.11409270763397217, "step": 24950 }, { "epoch": 0.6574855327170505, "grad_norm": 2.4846959114074707, "learning_rate": 2.7740939009237565e-06, "loss": 0.0945557713508606, "step": 24960 }, { "epoch": 0.6577489483952222, "grad_norm": 2.9631478786468506, "learning_rate": 2.7702750431779898e-06, "loss": 0.11263142824172974, "step": 24970 }, { "epoch": 0.6580123640733941, "grad_norm": 2.8969860076904297, "learning_rate": 2.766457808322242e-06, "loss": 0.11843494176864625, "step": 24980 }, { "epoch": 0.658275779751566, "grad_norm": 3.1029863357543945, "learning_rate": 2.7626421991348666e-06, "loss": 0.11414914131164551, "step": 24990 }, { "epoch": 0.6585391954297379, "grad_norm": 2.5379154682159424, "learning_rate": 2.7588282183930295e-06, "loss": 0.11008422374725342, "step": 25000 }, { "epoch": 0.6588026111079098, "grad_norm": 2.9321110248565674, "learning_rate": 2.7550158688727025e-06, "loss": 0.11609867811203003, "step": 25010 }, { "epoch": 0.6590660267860817, "grad_norm": 5.529200077056885, "learning_rate": 2.751205153348685e-06, "loss": 0.12009205818176269, "step": 25020 }, { "epoch": 0.6593294424642536, "grad_norm": 4.210806846618652, "learning_rate": 2.7473960745945783e-06, "loss": 0.10854455232620239, "step": 25030 }, { "epoch": 0.6595928581424255, "grad_norm": 4.553875923156738, "learning_rate": 2.743588635382794e-06, "loss": 0.11549054384231568, "step": 25040 }, { "epoch": 0.6598562738205974, "grad_norm": 2.283313274383545, "learning_rate": 2.7397828384845505e-06, "loss": 0.10696853399276733, "step": 25050 }, { "epoch": 0.6601196894987693, "grad_norm": 4.018795967102051, "learning_rate": 2.73597868666987e-06, "loss": 0.10278706550598145, "step": 25060 }, { "epoch": 0.6603831051769412, "grad_norm": 3.3152213096618652, "learning_rate": 2.732176182707581e-06, "loss": 0.12107992172241211, "step": 25070 }, { "epoch": 0.6606465208551131, "grad_norm": 3.408907890319824, "learning_rate": 2.728375329365308e-06, "loss": 0.11210966110229492, "step": 25080 }, { "epoch": 0.660909936533285, "grad_norm": 3.783745765686035, "learning_rate": 2.7245761294094776e-06, "loss": 0.1146162986755371, "step": 25090 }, { "epoch": 0.6611733522114569, "grad_norm": 3.2815563678741455, "learning_rate": 2.7207785856053092e-06, "loss": 0.12035796642303467, "step": 25100 }, { "epoch": 0.6614367678896288, "grad_norm": 2.198704957962036, "learning_rate": 2.7169827007168193e-06, "loss": 0.10170509815216064, "step": 25110 }, { "epoch": 0.6617001835678007, "grad_norm": 4.210973739624023, "learning_rate": 2.7131884775068206e-06, "loss": 0.10617606639862061, "step": 25120 }, { "epoch": 0.6619635992459726, "grad_norm": 3.17079496383667, "learning_rate": 2.7093959187369077e-06, "loss": 0.11053540706634521, "step": 25130 }, { "epoch": 0.6622270149241445, "grad_norm": 3.718024969100952, "learning_rate": 2.7056050271674683e-06, "loss": 0.10515704154968261, "step": 25140 }, { "epoch": 0.6624904306023164, "grad_norm": 3.3350579738616943, "learning_rate": 2.701815805557681e-06, "loss": 0.11673305034637452, "step": 25150 }, { "epoch": 0.6627538462804883, "grad_norm": 3.9233150482177734, "learning_rate": 2.6980282566655026e-06, "loss": 0.11677680015563965, "step": 25160 }, { "epoch": 0.6630172619586602, "grad_norm": 4.322581768035889, "learning_rate": 2.694242383247674e-06, "loss": 0.10874423980712891, "step": 25170 }, { "epoch": 0.6632806776368321, "grad_norm": 3.404844284057617, "learning_rate": 2.6904581880597182e-06, "loss": 0.10725884437561035, "step": 25180 }, { "epoch": 0.663544093315004, "grad_norm": 2.72664475440979, "learning_rate": 2.6866756738559325e-06, "loss": 0.10713019371032714, "step": 25190 }, { "epoch": 0.6638075089931759, "grad_norm": 2.85174822807312, "learning_rate": 2.6828948433893986e-06, "loss": 0.11715389490127563, "step": 25200 }, { "epoch": 0.6640709246713478, "grad_norm": 2.905125141143799, "learning_rate": 2.6791156994119686e-06, "loss": 0.09377734661102295, "step": 25210 }, { "epoch": 0.6643343403495197, "grad_norm": 3.086224317550659, "learning_rate": 2.6753382446742605e-06, "loss": 0.10656131505966186, "step": 25220 }, { "epoch": 0.6645977560276916, "grad_norm": 3.2144110202789307, "learning_rate": 2.6715624819256737e-06, "loss": 0.11856937408447266, "step": 25230 }, { "epoch": 0.6648611717058635, "grad_norm": 4.772554874420166, "learning_rate": 2.6677884139143686e-06, "loss": 0.12402088642120361, "step": 25240 }, { "epoch": 0.6651245873840353, "grad_norm": 4.569657802581787, "learning_rate": 2.6640160433872784e-06, "loss": 0.12574902772903443, "step": 25250 }, { "epoch": 0.6653880030622072, "grad_norm": 3.388349771499634, "learning_rate": 2.6602453730900944e-06, "loss": 0.11564584970474243, "step": 25260 }, { "epoch": 0.6656514187403791, "grad_norm": 2.9507381916046143, "learning_rate": 2.656476405767271e-06, "loss": 0.10620934963226318, "step": 25270 }, { "epoch": 0.665914834418551, "grad_norm": 3.056903123855591, "learning_rate": 2.652709144162029e-06, "loss": 0.11968944072723389, "step": 25280 }, { "epoch": 0.6661782500967229, "grad_norm": 2.12337327003479, "learning_rate": 2.648943591016343e-06, "loss": 0.11837570667266846, "step": 25290 }, { "epoch": 0.6664416657748948, "grad_norm": 2.883511543273926, "learning_rate": 2.645179749070945e-06, "loss": 0.10655338764190674, "step": 25300 }, { "epoch": 0.6667050814530667, "grad_norm": 3.510563611984253, "learning_rate": 2.6414176210653207e-06, "loss": 0.1111327052116394, "step": 25310 }, { "epoch": 0.6669684971312386, "grad_norm": 5.694674491882324, "learning_rate": 2.6376572097377084e-06, "loss": 0.10859237909317017, "step": 25320 }, { "epoch": 0.6672319128094105, "grad_norm": 3.2381303310394287, "learning_rate": 2.633898517825101e-06, "loss": 0.10707827806472778, "step": 25330 }, { "epoch": 0.6674953284875824, "grad_norm": 3.0387277603149414, "learning_rate": 2.6301415480632352e-06, "loss": 0.12421590089797974, "step": 25340 }, { "epoch": 0.6677587441657543, "grad_norm": 3.1801066398620605, "learning_rate": 2.626386303186593e-06, "loss": 0.11025110483169556, "step": 25350 }, { "epoch": 0.6680221598439262, "grad_norm": 2.8792693614959717, "learning_rate": 2.622632785928406e-06, "loss": 0.08887631893157959, "step": 25360 }, { "epoch": 0.6682855755220981, "grad_norm": 2.274061918258667, "learning_rate": 2.6188809990206454e-06, "loss": 0.11229277849197387, "step": 25370 }, { "epoch": 0.66854899120027, "grad_norm": 2.9713480472564697, "learning_rate": 2.6151309451940265e-06, "loss": 0.10803337097167968, "step": 25380 }, { "epoch": 0.6688124068784419, "grad_norm": 3.6643171310424805, "learning_rate": 2.6113826271779962e-06, "loss": 0.10781102180480957, "step": 25390 }, { "epoch": 0.6690758225566138, "grad_norm": 1.7788892984390259, "learning_rate": 2.6076360477007417e-06, "loss": 0.11150619983673096, "step": 25400 }, { "epoch": 0.6693392382347857, "grad_norm": 2.3318653106689453, "learning_rate": 2.603891209489189e-06, "loss": 0.12244215011596679, "step": 25410 }, { "epoch": 0.6696026539129576, "grad_norm": 2.9219472408294678, "learning_rate": 2.600148115268991e-06, "loss": 0.11026906967163086, "step": 25420 }, { "epoch": 0.6698660695911295, "grad_norm": 2.756891965866089, "learning_rate": 2.596406767764533e-06, "loss": 0.11399608850479126, "step": 25430 }, { "epoch": 0.6701294852693014, "grad_norm": 3.4315826892852783, "learning_rate": 2.5926671696989292e-06, "loss": 0.10312392711639404, "step": 25440 }, { "epoch": 0.6703929009474733, "grad_norm": 2.4664463996887207, "learning_rate": 2.5889293237940185e-06, "loss": 0.09743618965148926, "step": 25450 }, { "epoch": 0.6706563166256452, "grad_norm": 3.4518957138061523, "learning_rate": 2.5851932327703704e-06, "loss": 0.12021793127059936, "step": 25460 }, { "epoch": 0.6709197323038171, "grad_norm": 2.731499671936035, "learning_rate": 2.581458899347273e-06, "loss": 0.09939135909080506, "step": 25470 }, { "epoch": 0.671183147981989, "grad_norm": 3.0915725231170654, "learning_rate": 2.57772632624273e-06, "loss": 0.12075417041778565, "step": 25480 }, { "epoch": 0.6714465636601609, "grad_norm": 2.6086478233337402, "learning_rate": 2.573995516173474e-06, "loss": 0.11095212697982788, "step": 25490 }, { "epoch": 0.6717099793383328, "grad_norm": 3.110666513442993, "learning_rate": 2.5702664718549466e-06, "loss": 0.12555222511291503, "step": 25500 }, { "epoch": 0.6719733950165047, "grad_norm": 3.401984691619873, "learning_rate": 2.5665391960013114e-06, "loss": 0.10172966718673707, "step": 25510 }, { "epoch": 0.6722368106946766, "grad_norm": 3.7102396488189697, "learning_rate": 2.5628136913254365e-06, "loss": 0.10344456434249878, "step": 25520 }, { "epoch": 0.6725002263728485, "grad_norm": 3.7058167457580566, "learning_rate": 2.559089960538904e-06, "loss": 0.10672394037246705, "step": 25530 }, { "epoch": 0.6727636420510203, "grad_norm": 3.753652334213257, "learning_rate": 2.5553680063520103e-06, "loss": 0.12159366607666015, "step": 25540 }, { "epoch": 0.6730270577291922, "grad_norm": 3.723423719406128, "learning_rate": 2.551647831473752e-06, "loss": 0.10932610034942628, "step": 25550 }, { "epoch": 0.6732904734073641, "grad_norm": 4.314154148101807, "learning_rate": 2.5479294386118327e-06, "loss": 0.11240166425704956, "step": 25560 }, { "epoch": 0.673553889085536, "grad_norm": 2.912149667739868, "learning_rate": 2.5442128304726595e-06, "loss": 0.11091694831848145, "step": 25570 }, { "epoch": 0.6738173047637079, "grad_norm": 3.5483193397521973, "learning_rate": 2.5404980097613373e-06, "loss": 0.11592013835906982, "step": 25580 }, { "epoch": 0.6740807204418798, "grad_norm": 4.553089618682861, "learning_rate": 2.536784979181678e-06, "loss": 0.10939216613769531, "step": 25590 }, { "epoch": 0.6743441361200517, "grad_norm": 2.969892740249634, "learning_rate": 2.533073741436185e-06, "loss": 0.11284785270690918, "step": 25600 }, { "epoch": 0.6746075517982236, "grad_norm": 4.154445171356201, "learning_rate": 2.5293642992260516e-06, "loss": 0.1173860788345337, "step": 25610 }, { "epoch": 0.6748709674763955, "grad_norm": 3.508108615875244, "learning_rate": 2.525656655251175e-06, "loss": 0.09992167949676514, "step": 25620 }, { "epoch": 0.6751343831545674, "grad_norm": 3.325054407119751, "learning_rate": 2.5219508122101366e-06, "loss": 0.12005486488342285, "step": 25630 }, { "epoch": 0.6753977988327393, "grad_norm": 3.35009765625, "learning_rate": 2.5182467728002104e-06, "loss": 0.10688414573669433, "step": 25640 }, { "epoch": 0.6756612145109112, "grad_norm": 3.838104248046875, "learning_rate": 2.514544539717354e-06, "loss": 0.11796457767486572, "step": 25650 }, { "epoch": 0.6759246301890831, "grad_norm": 2.783097982406616, "learning_rate": 2.5108441156562115e-06, "loss": 0.11617124080657959, "step": 25660 }, { "epoch": 0.676188045867255, "grad_norm": 2.2816219329833984, "learning_rate": 2.5071455033101145e-06, "loss": 0.09466310739517211, "step": 25670 }, { "epoch": 0.6764514615454269, "grad_norm": 2.8917324542999268, "learning_rate": 2.503448705371071e-06, "loss": 0.10369296073913574, "step": 25680 }, { "epoch": 0.6767148772235988, "grad_norm": 2.8682541847229004, "learning_rate": 2.499753724529771e-06, "loss": 0.1149591088294983, "step": 25690 }, { "epoch": 0.6769782929017707, "grad_norm": 3.7936112880706787, "learning_rate": 2.4960605634755797e-06, "loss": 0.10577161312103271, "step": 25700 }, { "epoch": 0.6772417085799426, "grad_norm": 3.5723989009857178, "learning_rate": 2.492369224896537e-06, "loss": 0.0947296142578125, "step": 25710 }, { "epoch": 0.6775051242581145, "grad_norm": 2.5375709533691406, "learning_rate": 2.4886797114793654e-06, "loss": 0.10541235208511353, "step": 25720 }, { "epoch": 0.6777685399362864, "grad_norm": 2.8873140811920166, "learning_rate": 2.484992025909445e-06, "loss": 0.11364556550979614, "step": 25730 }, { "epoch": 0.6780319556144583, "grad_norm": 3.258396625518799, "learning_rate": 2.481306170870833e-06, "loss": 0.11248053312301635, "step": 25740 }, { "epoch": 0.6782953712926302, "grad_norm": 3.7899906635284424, "learning_rate": 2.477622149046257e-06, "loss": 0.11169273853302002, "step": 25750 }, { "epoch": 0.678558786970802, "grad_norm": 3.372112989425659, "learning_rate": 2.473939963117106e-06, "loss": 0.12105734348297119, "step": 25760 }, { "epoch": 0.6788222026489739, "grad_norm": 3.8011581897735596, "learning_rate": 2.4702596157634334e-06, "loss": 0.09603264927864075, "step": 25770 }, { "epoch": 0.6790856183271458, "grad_norm": 3.7443599700927734, "learning_rate": 2.4665811096639546e-06, "loss": 0.11645911931991577, "step": 25780 }, { "epoch": 0.6793490340053177, "grad_norm": 3.7125415802001953, "learning_rate": 2.4629044474960436e-06, "loss": 0.11594899892807006, "step": 25790 }, { "epoch": 0.6796124496834896, "grad_norm": 3.9553699493408203, "learning_rate": 2.459229631935737e-06, "loss": 0.1184651494026184, "step": 25800 }, { "epoch": 0.6798758653616614, "grad_norm": 2.7081973552703857, "learning_rate": 2.455556665657724e-06, "loss": 0.11189932823181152, "step": 25810 }, { "epoch": 0.6801392810398333, "grad_norm": 4.245368003845215, "learning_rate": 2.451885551335343e-06, "loss": 0.10550389289855958, "step": 25820 }, { "epoch": 0.6804026967180052, "grad_norm": 2.790367603302002, "learning_rate": 2.4482162916405945e-06, "loss": 0.10580356121063232, "step": 25830 }, { "epoch": 0.6806661123961771, "grad_norm": 2.5726587772369385, "learning_rate": 2.44454888924412e-06, "loss": 0.08987393975257874, "step": 25840 }, { "epoch": 0.680929528074349, "grad_norm": 3.7659261226654053, "learning_rate": 2.4408833468152194e-06, "loss": 0.11120569705963135, "step": 25850 }, { "epoch": 0.6811929437525209, "grad_norm": 3.036900281906128, "learning_rate": 2.437219667021827e-06, "loss": 0.10787836313247681, "step": 25860 }, { "epoch": 0.6814563594306928, "grad_norm": 2.2533514499664307, "learning_rate": 2.4335578525305263e-06, "loss": 0.10217570066452027, "step": 25870 }, { "epoch": 0.6817197751088647, "grad_norm": 2.6588165760040283, "learning_rate": 2.4298979060065493e-06, "loss": 0.10756950378417969, "step": 25880 }, { "epoch": 0.6819831907870366, "grad_norm": 2.8403542041778564, "learning_rate": 2.426239830113759e-06, "loss": 0.11090338230133057, "step": 25890 }, { "epoch": 0.6822466064652085, "grad_norm": 3.3657097816467285, "learning_rate": 2.4225836275146618e-06, "loss": 0.10513982772827149, "step": 25900 }, { "epoch": 0.6825100221433804, "grad_norm": 2.408155918121338, "learning_rate": 2.4189293008703994e-06, "loss": 0.11620292663574219, "step": 25910 }, { "epoch": 0.6827734378215523, "grad_norm": 3.3969976902008057, "learning_rate": 2.415276852840748e-06, "loss": 0.11619067192077637, "step": 25920 }, { "epoch": 0.6830368534997242, "grad_norm": 3.0009021759033203, "learning_rate": 2.411626286084118e-06, "loss": 0.11038908958435059, "step": 25930 }, { "epoch": 0.6833002691778961, "grad_norm": 3.0929226875305176, "learning_rate": 2.407977603257552e-06, "loss": 0.11489462852478027, "step": 25940 }, { "epoch": 0.683563684856068, "grad_norm": 2.819519281387329, "learning_rate": 2.404330807016711e-06, "loss": 0.10260770320892335, "step": 25950 }, { "epoch": 0.6838271005342399, "grad_norm": 2.680586338043213, "learning_rate": 2.400685900015898e-06, "loss": 0.1162951946258545, "step": 25960 }, { "epoch": 0.6840905162124118, "grad_norm": 5.404940128326416, "learning_rate": 2.3970428849080284e-06, "loss": 0.10647079944610596, "step": 25970 }, { "epoch": 0.6843539318905837, "grad_norm": 3.220351219177246, "learning_rate": 2.393401764344652e-06, "loss": 0.10399073362350464, "step": 25980 }, { "epoch": 0.6846173475687556, "grad_norm": 2.4492669105529785, "learning_rate": 2.3897625409759285e-06, "loss": 0.11296727657318115, "step": 25990 }, { "epoch": 0.6848807632469275, "grad_norm": 3.459095001220703, "learning_rate": 2.3861252174506404e-06, "loss": 0.10852348804473877, "step": 26000 }, { "epoch": 0.6851441789250994, "grad_norm": 4.367877006530762, "learning_rate": 2.3824897964161938e-06, "loss": 0.11037248373031616, "step": 26010 }, { "epoch": 0.6854075946032713, "grad_norm": 3.1582038402557373, "learning_rate": 2.3788562805186014e-06, "loss": 0.10988651514053345, "step": 26020 }, { "epoch": 0.6856710102814432, "grad_norm": 3.6721103191375732, "learning_rate": 2.3752246724024947e-06, "loss": 0.11904603242874146, "step": 26030 }, { "epoch": 0.6859344259596151, "grad_norm": 3.3870737552642822, "learning_rate": 2.3715949747111134e-06, "loss": 0.11150531768798828, "step": 26040 }, { "epoch": 0.686197841637787, "grad_norm": 2.846611261367798, "learning_rate": 2.3679671900863065e-06, "loss": 0.11679497957229615, "step": 26050 }, { "epoch": 0.6864612573159589, "grad_norm": 3.494178533554077, "learning_rate": 2.364341321168536e-06, "loss": 0.1080620527267456, "step": 26060 }, { "epoch": 0.6867246729941308, "grad_norm": 2.7333743572235107, "learning_rate": 2.3607173705968662e-06, "loss": 0.1187061071395874, "step": 26070 }, { "epoch": 0.6869880886723027, "grad_norm": 3.1057722568511963, "learning_rate": 2.357095341008959e-06, "loss": 0.09848875999450683, "step": 26080 }, { "epoch": 0.6872515043504746, "grad_norm": 5.390413761138916, "learning_rate": 2.3534752350410884e-06, "loss": 0.10273572206497192, "step": 26090 }, { "epoch": 0.6875149200286464, "grad_norm": 3.1308844089508057, "learning_rate": 2.349857055328121e-06, "loss": 0.11408486366271972, "step": 26100 }, { "epoch": 0.6877783357068183, "grad_norm": 4.486187934875488, "learning_rate": 2.3462408045035288e-06, "loss": 0.11217341423034669, "step": 26110 }, { "epoch": 0.6880417513849902, "grad_norm": 3.2233119010925293, "learning_rate": 2.342626485199371e-06, "loss": 0.08936789631843567, "step": 26120 }, { "epoch": 0.6883051670631621, "grad_norm": 4.5385050773620605, "learning_rate": 2.3390141000463033e-06, "loss": 0.1134761095046997, "step": 26130 }, { "epoch": 0.688568582741334, "grad_norm": 3.425934076309204, "learning_rate": 2.3354036516735804e-06, "loss": 0.10954992771148682, "step": 26140 }, { "epoch": 0.6888319984195059, "grad_norm": 2.622252941131592, "learning_rate": 2.331795142709041e-06, "loss": 0.1061814308166504, "step": 26150 }, { "epoch": 0.6890954140976778, "grad_norm": 4.402109622955322, "learning_rate": 2.328188575779112e-06, "loss": 0.10218690633773804, "step": 26160 }, { "epoch": 0.6893588297758497, "grad_norm": 3.9372735023498535, "learning_rate": 2.32458395350881e-06, "loss": 0.09847736358642578, "step": 26170 }, { "epoch": 0.6896222454540216, "grad_norm": 3.4456095695495605, "learning_rate": 2.320981278521731e-06, "loss": 0.09698905944824218, "step": 26180 }, { "epoch": 0.6898856611321935, "grad_norm": 3.5701100826263428, "learning_rate": 2.3173805534400634e-06, "loss": 0.11491179466247559, "step": 26190 }, { "epoch": 0.6901490768103654, "grad_norm": 3.145247220993042, "learning_rate": 2.313781780884568e-06, "loss": 0.10933127403259277, "step": 26200 }, { "epoch": 0.6904124924885373, "grad_norm": 3.0600826740264893, "learning_rate": 2.310184963474582e-06, "loss": 0.10643621683120727, "step": 26210 }, { "epoch": 0.6906759081667092, "grad_norm": 3.30098557472229, "learning_rate": 2.3065901038280305e-06, "loss": 0.10094537734985351, "step": 26220 }, { "epoch": 0.6909393238448811, "grad_norm": 3.178356885910034, "learning_rate": 2.302997204561405e-06, "loss": 0.10326957702636719, "step": 26230 }, { "epoch": 0.691202739523053, "grad_norm": 2.9094038009643555, "learning_rate": 2.2994062682897723e-06, "loss": 0.1010839581489563, "step": 26240 }, { "epoch": 0.6914661552012249, "grad_norm": 3.692124605178833, "learning_rate": 2.2958172976267706e-06, "loss": 0.10452010631561279, "step": 26250 }, { "epoch": 0.6917295708793968, "grad_norm": 3.6975386142730713, "learning_rate": 2.292230295184607e-06, "loss": 0.11328403949737549, "step": 26260 }, { "epoch": 0.6919929865575687, "grad_norm": 3.5366854667663574, "learning_rate": 2.288645263574059e-06, "loss": 0.09137587547302246, "step": 26270 }, { "epoch": 0.6922564022357406, "grad_norm": 3.069153070449829, "learning_rate": 2.2850622054044662e-06, "loss": 0.11131906509399414, "step": 26280 }, { "epoch": 0.6925198179139125, "grad_norm": 2.7383220195770264, "learning_rate": 2.281481123283733e-06, "loss": 0.09686543345451355, "step": 26290 }, { "epoch": 0.6927832335920844, "grad_norm": 3.3225014209747314, "learning_rate": 2.2779020198183245e-06, "loss": 0.10982685089111328, "step": 26300 }, { "epoch": 0.6930466492702563, "grad_norm": 2.6670613288879395, "learning_rate": 2.274324897613266e-06, "loss": 0.11284756660461426, "step": 26310 }, { "epoch": 0.6933100649484282, "grad_norm": 3.5113346576690674, "learning_rate": 2.270749759272146e-06, "loss": 0.09191510081291199, "step": 26320 }, { "epoch": 0.6935734806266001, "grad_norm": 4.331082820892334, "learning_rate": 2.2671766073970992e-06, "loss": 0.10011788606643676, "step": 26330 }, { "epoch": 0.693836896304772, "grad_norm": 3.5940802097320557, "learning_rate": 2.2636054445888197e-06, "loss": 0.10613981485366822, "step": 26340 }, { "epoch": 0.6941003119829439, "grad_norm": 3.946420669555664, "learning_rate": 2.2600362734465576e-06, "loss": 0.108770751953125, "step": 26350 }, { "epoch": 0.6943637276611158, "grad_norm": 3.4027411937713623, "learning_rate": 2.256469096568108e-06, "loss": 0.1224778413772583, "step": 26360 }, { "epoch": 0.6946271433392877, "grad_norm": 3.0258421897888184, "learning_rate": 2.2529039165498157e-06, "loss": 0.1074376106262207, "step": 26370 }, { "epoch": 0.6948905590174596, "grad_norm": 2.5843398571014404, "learning_rate": 2.2493407359865726e-06, "loss": 0.10143035650253296, "step": 26380 }, { "epoch": 0.6951539746956314, "grad_norm": 2.926558256149292, "learning_rate": 2.245779557471814e-06, "loss": 0.11692759990692139, "step": 26390 }, { "epoch": 0.6954173903738033, "grad_norm": 2.8935706615448, "learning_rate": 2.242220383597522e-06, "loss": 0.11124600172042846, "step": 26400 }, { "epoch": 0.6956808060519752, "grad_norm": 3.1259958744049072, "learning_rate": 2.2386632169542176e-06, "loss": 0.11778128147125244, "step": 26410 }, { "epoch": 0.6959442217301471, "grad_norm": 5.9719014167785645, "learning_rate": 2.2351080601309543e-06, "loss": 0.11975780725479127, "step": 26420 }, { "epoch": 0.696207637408319, "grad_norm": 2.5672099590301514, "learning_rate": 2.231554915715334e-06, "loss": 0.10818579196929931, "step": 26430 }, { "epoch": 0.6964710530864909, "grad_norm": 4.219062805175781, "learning_rate": 2.2280037862934855e-06, "loss": 0.11223628520965576, "step": 26440 }, { "epoch": 0.6967344687646628, "grad_norm": 2.618046283721924, "learning_rate": 2.2244546744500796e-06, "loss": 0.10779823064804077, "step": 26450 }, { "epoch": 0.6969978844428347, "grad_norm": 3.2996327877044678, "learning_rate": 2.220907582768306e-06, "loss": 0.10554187297821045, "step": 26460 }, { "epoch": 0.6972613001210066, "grad_norm": 4.052340507507324, "learning_rate": 2.217362513829893e-06, "loss": 0.10270823240280151, "step": 26470 }, { "epoch": 0.6975247157991785, "grad_norm": 3.712590456008911, "learning_rate": 2.213819470215096e-06, "loss": 0.09805716276168823, "step": 26480 }, { "epoch": 0.6977881314773504, "grad_norm": 5.503606796264648, "learning_rate": 2.210278454502696e-06, "loss": 0.10094224214553833, "step": 26490 }, { "epoch": 0.6980515471555223, "grad_norm": 4.629671573638916, "learning_rate": 2.206739469269995e-06, "loss": 0.1048054814338684, "step": 26500 }, { "epoch": 0.6983149628336942, "grad_norm": 2.6796250343322754, "learning_rate": 2.20320251709282e-06, "loss": 0.1112579345703125, "step": 26510 }, { "epoch": 0.6985783785118661, "grad_norm": 4.96635103225708, "learning_rate": 2.199667600545515e-06, "loss": 0.1049856424331665, "step": 26520 }, { "epoch": 0.698841794190038, "grad_norm": 3.5373027324676514, "learning_rate": 2.1961347222009492e-06, "loss": 0.10315179824829102, "step": 26530 }, { "epoch": 0.6991052098682099, "grad_norm": 4.9180426597595215, "learning_rate": 2.1926038846305035e-06, "loss": 0.10826668739318848, "step": 26540 }, { "epoch": 0.6993686255463817, "grad_norm": 2.576366662979126, "learning_rate": 2.189075090404068e-06, "loss": 0.10357961654663086, "step": 26550 }, { "epoch": 0.6996320412245536, "grad_norm": 2.871032476425171, "learning_rate": 2.185548342090057e-06, "loss": 0.10527329444885254, "step": 26560 }, { "epoch": 0.6998954569027255, "grad_norm": 3.6798346042633057, "learning_rate": 2.182023642255387e-06, "loss": 0.11158610582351684, "step": 26570 }, { "epoch": 0.7001588725808974, "grad_norm": 3.7934491634368896, "learning_rate": 2.178500993465491e-06, "loss": 0.09459185600280762, "step": 26580 }, { "epoch": 0.7004222882590693, "grad_norm": 3.961074113845825, "learning_rate": 2.1749803982843004e-06, "loss": 0.10522691011428834, "step": 26590 }, { "epoch": 0.7006857039372412, "grad_norm": 3.129549503326416, "learning_rate": 2.171461859274256e-06, "loss": 0.10156681537628173, "step": 26600 }, { "epoch": 0.7009491196154131, "grad_norm": 3.07570481300354, "learning_rate": 2.1679453789963057e-06, "loss": 0.09856622219085694, "step": 26610 }, { "epoch": 0.701212535293585, "grad_norm": 3.132774829864502, "learning_rate": 2.1644309600098934e-06, "loss": 0.12019525766372681, "step": 26620 }, { "epoch": 0.7014759509717569, "grad_norm": 4.521572589874268, "learning_rate": 2.1609186048729654e-06, "loss": 0.10420346260070801, "step": 26630 }, { "epoch": 0.7017393666499288, "grad_norm": 4.140403747558594, "learning_rate": 2.1574083161419652e-06, "loss": 0.09794155359268189, "step": 26640 }, { "epoch": 0.7020027823281006, "grad_norm": 3.9795494079589844, "learning_rate": 2.1539000963718303e-06, "loss": 0.1040927529335022, "step": 26650 }, { "epoch": 0.7022661980062725, "grad_norm": 4.5158820152282715, "learning_rate": 2.1503939481159985e-06, "loss": 0.1005709171295166, "step": 26660 }, { "epoch": 0.7025296136844444, "grad_norm": 3.2296559810638428, "learning_rate": 2.1468898739263954e-06, "loss": 0.10108985900878906, "step": 26670 }, { "epoch": 0.7027930293626163, "grad_norm": 3.108938217163086, "learning_rate": 2.1433878763534327e-06, "loss": 0.11714327335357666, "step": 26680 }, { "epoch": 0.7030564450407882, "grad_norm": 4.431373119354248, "learning_rate": 2.139887957946019e-06, "loss": 0.10528436899185181, "step": 26690 }, { "epoch": 0.7033198607189601, "grad_norm": 3.879404067993164, "learning_rate": 2.1363901212515443e-06, "loss": 0.11277740001678467, "step": 26700 }, { "epoch": 0.703583276397132, "grad_norm": 3.8899760246276855, "learning_rate": 2.1328943688158895e-06, "loss": 0.1002467393875122, "step": 26710 }, { "epoch": 0.7038466920753039, "grad_norm": 3.060574531555176, "learning_rate": 2.1294007031834085e-06, "loss": 0.08990076184272766, "step": 26720 }, { "epoch": 0.7041101077534758, "grad_norm": 2.769913911819458, "learning_rate": 2.1259091268969424e-06, "loss": 0.09785466194152832, "step": 26730 }, { "epoch": 0.7043735234316477, "grad_norm": 3.5596940517425537, "learning_rate": 2.1224196424978136e-06, "loss": 0.1089470386505127, "step": 26740 }, { "epoch": 0.7046369391098196, "grad_norm": 3.5626533031463623, "learning_rate": 2.118932252525819e-06, "loss": 0.11625206470489502, "step": 26750 }, { "epoch": 0.7049003547879915, "grad_norm": 3.6297738552093506, "learning_rate": 2.1154469595192307e-06, "loss": 0.10715310573577881, "step": 26760 }, { "epoch": 0.7051637704661634, "grad_norm": 3.0822882652282715, "learning_rate": 2.111963766014796e-06, "loss": 0.11203312873840332, "step": 26770 }, { "epoch": 0.7054271861443353, "grad_norm": 2.8101561069488525, "learning_rate": 2.10848267454773e-06, "loss": 0.10927287340164185, "step": 26780 }, { "epoch": 0.7056906018225072, "grad_norm": 3.6541147232055664, "learning_rate": 2.1050036876517267e-06, "loss": 0.10252600908279419, "step": 26790 }, { "epoch": 0.7059540175006791, "grad_norm": 4.048943996429443, "learning_rate": 2.101526807858942e-06, "loss": 0.10661617517471314, "step": 26800 }, { "epoch": 0.706217433178851, "grad_norm": 3.1938133239746094, "learning_rate": 2.098052037699993e-06, "loss": 0.1033708095550537, "step": 26810 }, { "epoch": 0.7064808488570229, "grad_norm": 3.4096226692199707, "learning_rate": 2.0945793797039742e-06, "loss": 0.10008957386016845, "step": 26820 }, { "epoch": 0.7067442645351948, "grad_norm": 3.505347728729248, "learning_rate": 2.0911088363984333e-06, "loss": 0.113127601146698, "step": 26830 }, { "epoch": 0.7070076802133667, "grad_norm": 3.3899874687194824, "learning_rate": 2.0876404103093813e-06, "loss": 0.10977120399475097, "step": 26840 }, { "epoch": 0.7072710958915386, "grad_norm": 3.4720592498779297, "learning_rate": 2.0841741039612884e-06, "loss": 0.10904867649078369, "step": 26850 }, { "epoch": 0.7075345115697105, "grad_norm": 4.982717990875244, "learning_rate": 2.08070991987708e-06, "loss": 0.10998989343643188, "step": 26860 }, { "epoch": 0.7077979272478824, "grad_norm": 3.9245307445526123, "learning_rate": 2.0772478605781434e-06, "loss": 0.11458059549331664, "step": 26870 }, { "epoch": 0.7080613429260543, "grad_norm": 3.3238778114318848, "learning_rate": 2.0737879285843115e-06, "loss": 0.09092840552330017, "step": 26880 }, { "epoch": 0.7083247586042262, "grad_norm": 3.5682084560394287, "learning_rate": 2.070330126413873e-06, "loss": 0.09213030934333802, "step": 26890 }, { "epoch": 0.7085881742823981, "grad_norm": 2.9670841693878174, "learning_rate": 2.066874456583565e-06, "loss": 0.10544644594192505, "step": 26900 }, { "epoch": 0.70885158996057, "grad_norm": 3.5426790714263916, "learning_rate": 2.063420921608572e-06, "loss": 0.09596118927001954, "step": 26910 }, { "epoch": 0.7091150056387419, "grad_norm": 5.049426078796387, "learning_rate": 2.0599695240025303e-06, "loss": 0.10810914039611816, "step": 26920 }, { "epoch": 0.7093784213169138, "grad_norm": 4.391149997711182, "learning_rate": 2.0565202662775106e-06, "loss": 0.09412248134613037, "step": 26930 }, { "epoch": 0.7096418369950857, "grad_norm": 4.060124397277832, "learning_rate": 2.0530731509440323e-06, "loss": 0.09252305030822754, "step": 26940 }, { "epoch": 0.7099052526732575, "grad_norm": 3.706254482269287, "learning_rate": 2.0496281805110577e-06, "loss": 0.09808987379074097, "step": 26950 }, { "epoch": 0.7101686683514294, "grad_norm": 4.573536396026611, "learning_rate": 2.046185357485983e-06, "loss": 0.1145779848098755, "step": 26960 }, { "epoch": 0.7104320840296013, "grad_norm": 3.3356969356536865, "learning_rate": 2.0427446843746433e-06, "loss": 0.12005563974380493, "step": 26970 }, { "epoch": 0.7106954997077732, "grad_norm": 4.4115071296691895, "learning_rate": 2.03930616368131e-06, "loss": 0.11148649454116821, "step": 26980 }, { "epoch": 0.7109589153859451, "grad_norm": 3.052751302719116, "learning_rate": 2.0358697979086833e-06, "loss": 0.09642496705055237, "step": 26990 }, { "epoch": 0.711222331064117, "grad_norm": 3.6347060203552246, "learning_rate": 2.0324355895579034e-06, "loss": 0.1047884464263916, "step": 27000 }, { "epoch": 0.7114857467422889, "grad_norm": 3.7381770610809326, "learning_rate": 2.0290035411285347e-06, "loss": 0.10981459617614746, "step": 27010 }, { "epoch": 0.7117491624204608, "grad_norm": 3.799668073654175, "learning_rate": 2.0255736551185647e-06, "loss": 0.10210485458374023, "step": 27020 }, { "epoch": 0.7120125780986327, "grad_norm": 3.7293734550476074, "learning_rate": 2.0221459340244183e-06, "loss": 0.10518734455108643, "step": 27030 }, { "epoch": 0.7122759937768046, "grad_norm": 2.8717336654663086, "learning_rate": 2.0187203803409342e-06, "loss": 0.10179827213287354, "step": 27040 }, { "epoch": 0.7125394094549765, "grad_norm": 3.644158363342285, "learning_rate": 2.015296996561385e-06, "loss": 0.11675543785095215, "step": 27050 }, { "epoch": 0.7128028251331484, "grad_norm": 4.260745525360107, "learning_rate": 2.0118757851774508e-06, "loss": 0.08853446841239929, "step": 27060 }, { "epoch": 0.7130662408113203, "grad_norm": 3.9677414894104004, "learning_rate": 2.0084567486792363e-06, "loss": 0.10409824848175049, "step": 27070 }, { "epoch": 0.7133296564894922, "grad_norm": 4.244760990142822, "learning_rate": 2.0050398895552687e-06, "loss": 0.10256661176681518, "step": 27080 }, { "epoch": 0.7135930721676641, "grad_norm": 3.270134449005127, "learning_rate": 2.001625210292483e-06, "loss": 0.10811138153076172, "step": 27090 }, { "epoch": 0.713856487845836, "grad_norm": 3.757765054702759, "learning_rate": 1.9982127133762296e-06, "loss": 0.1046480417251587, "step": 27100 }, { "epoch": 0.7141199035240079, "grad_norm": 2.9812161922454834, "learning_rate": 1.9948024012902723e-06, "loss": 0.09231893420219421, "step": 27110 }, { "epoch": 0.7143833192021798, "grad_norm": 3.9692490100860596, "learning_rate": 1.9913942765167804e-06, "loss": 0.10164837837219239, "step": 27120 }, { "epoch": 0.7146467348803517, "grad_norm": 4.222838878631592, "learning_rate": 1.987988341536338e-06, "loss": 0.10043127536773681, "step": 27130 }, { "epoch": 0.7149101505585236, "grad_norm": 5.2626543045043945, "learning_rate": 1.9845845988279318e-06, "loss": 0.10538914203643798, "step": 27140 }, { "epoch": 0.7151735662366955, "grad_norm": 3.8027243614196777, "learning_rate": 1.9811830508689473e-06, "loss": 0.11004562377929687, "step": 27150 }, { "epoch": 0.7154369819148674, "grad_norm": 4.222452163696289, "learning_rate": 1.9777837001351823e-06, "loss": 0.0983609139919281, "step": 27160 }, { "epoch": 0.7157003975930393, "grad_norm": 5.1378960609436035, "learning_rate": 1.9743865491008286e-06, "loss": 0.11343395709991455, "step": 27170 }, { "epoch": 0.7159638132712112, "grad_norm": 2.823137044906616, "learning_rate": 1.9709916002384845e-06, "loss": 0.1002117395401001, "step": 27180 }, { "epoch": 0.7162272289493831, "grad_norm": 3.6207215785980225, "learning_rate": 1.9675988560191333e-06, "loss": 0.09887394905090333, "step": 27190 }, { "epoch": 0.716490644627555, "grad_norm": 4.152927875518799, "learning_rate": 1.964208318912163e-06, "loss": 0.0973321795463562, "step": 27200 }, { "epoch": 0.7167540603057269, "grad_norm": 3.5721747875213623, "learning_rate": 1.9608199913853544e-06, "loss": 0.10357490777969361, "step": 27210 }, { "epoch": 0.7170174759838988, "grad_norm": 4.524234294891357, "learning_rate": 1.957433875904878e-06, "loss": 0.099628746509552, "step": 27220 }, { "epoch": 0.7172808916620707, "grad_norm": 5.271790981292725, "learning_rate": 1.954049974935294e-06, "loss": 0.09474201202392578, "step": 27230 }, { "epoch": 0.7175443073402425, "grad_norm": 3.3967108726501465, "learning_rate": 1.950668290939551e-06, "loss": 0.10071954727172852, "step": 27240 }, { "epoch": 0.7178077230184144, "grad_norm": 4.2232561111450195, "learning_rate": 1.947288826378984e-06, "loss": 0.09365304112434387, "step": 27250 }, { "epoch": 0.7180711386965863, "grad_norm": 3.5593152046203613, "learning_rate": 1.9439115837133164e-06, "loss": 0.10006537437438964, "step": 27260 }, { "epoch": 0.7183345543747582, "grad_norm": 4.238886833190918, "learning_rate": 1.9405365654006503e-06, "loss": 0.10677295923233032, "step": 27270 }, { "epoch": 0.7185979700529301, "grad_norm": 3.817258358001709, "learning_rate": 1.9371637738974653e-06, "loss": 0.11084342002868652, "step": 27280 }, { "epoch": 0.718861385731102, "grad_norm": 4.984395503997803, "learning_rate": 1.9337932116586287e-06, "loss": 0.09949308633804321, "step": 27290 }, { "epoch": 0.7191248014092739, "grad_norm": 3.2974796295166016, "learning_rate": 1.9304248811373796e-06, "loss": 0.09279572367668151, "step": 27300 }, { "epoch": 0.7193882170874458, "grad_norm": 3.3029417991638184, "learning_rate": 1.927058784785337e-06, "loss": 0.10532760620117188, "step": 27310 }, { "epoch": 0.7196516327656177, "grad_norm": 4.557961463928223, "learning_rate": 1.9236949250524873e-06, "loss": 0.09847838878631592, "step": 27320 }, { "epoch": 0.7199150484437896, "grad_norm": 3.149838924407959, "learning_rate": 1.9203333043871925e-06, "loss": 0.10369429588317872, "step": 27330 }, { "epoch": 0.7201784641219614, "grad_norm": 3.0593607425689697, "learning_rate": 1.916973925236189e-06, "loss": 0.09979007840156555, "step": 27340 }, { "epoch": 0.7204418798001333, "grad_norm": 3.094723701477051, "learning_rate": 1.913616790044576e-06, "loss": 0.09742650985717774, "step": 27350 }, { "epoch": 0.7207052954783052, "grad_norm": 4.384091854095459, "learning_rate": 1.91026190125582e-06, "loss": 0.09860106706619262, "step": 27360 }, { "epoch": 0.7209687111564771, "grad_norm": 4.103701591491699, "learning_rate": 1.9069092613117557e-06, "loss": 0.11339250802993775, "step": 27370 }, { "epoch": 0.721232126834649, "grad_norm": 4.611839771270752, "learning_rate": 1.9035588726525762e-06, "loss": 0.10661396980285645, "step": 27380 }, { "epoch": 0.7214955425128209, "grad_norm": 3.151756763458252, "learning_rate": 1.9002107377168426e-06, "loss": 0.11403312683105468, "step": 27390 }, { "epoch": 0.7217589581909928, "grad_norm": 3.779676914215088, "learning_rate": 1.8968648589414729e-06, "loss": 0.10394763946533203, "step": 27400 }, { "epoch": 0.7220223738691647, "grad_norm": 3.19480037689209, "learning_rate": 1.8935212387617352e-06, "loss": 0.10015017986297607, "step": 27410 }, { "epoch": 0.7222857895473366, "grad_norm": 3.2719905376434326, "learning_rate": 1.890179879611267e-06, "loss": 0.09143954515457153, "step": 27420 }, { "epoch": 0.7225492052255085, "grad_norm": 4.138509273529053, "learning_rate": 1.8868407839220516e-06, "loss": 0.096924090385437, "step": 27430 }, { "epoch": 0.7228126209036804, "grad_norm": 4.220264434814453, "learning_rate": 1.883503954124426e-06, "loss": 0.106324303150177, "step": 27440 }, { "epoch": 0.7230760365818523, "grad_norm": 3.9944608211517334, "learning_rate": 1.8801693926470804e-06, "loss": 0.09377041459083557, "step": 27450 }, { "epoch": 0.7233394522600242, "grad_norm": 2.7044548988342285, "learning_rate": 1.8768371019170495e-06, "loss": 0.08792471885681152, "step": 27460 }, { "epoch": 0.7236028679381961, "grad_norm": 4.736961364746094, "learning_rate": 1.8735070843597224e-06, "loss": 0.103824782371521, "step": 27470 }, { "epoch": 0.723866283616368, "grad_norm": 3.4301302433013916, "learning_rate": 1.870179342398828e-06, "loss": 0.09065918326377868, "step": 27480 }, { "epoch": 0.7241296992945399, "grad_norm": 3.8860247135162354, "learning_rate": 1.8668538784564399e-06, "loss": 0.09676328897476197, "step": 27490 }, { "epoch": 0.7243931149727117, "grad_norm": 4.267942428588867, "learning_rate": 1.8635306949529747e-06, "loss": 0.10136878490447998, "step": 27500 }, { "epoch": 0.7246565306508836, "grad_norm": 4.560184955596924, "learning_rate": 1.8602097943071868e-06, "loss": 0.10132628679275513, "step": 27510 }, { "epoch": 0.7249199463290555, "grad_norm": 3.988456964492798, "learning_rate": 1.856891178936176e-06, "loss": 0.09813179969787597, "step": 27520 }, { "epoch": 0.7251833620072274, "grad_norm": 4.137581825256348, "learning_rate": 1.8535748512553686e-06, "loss": 0.09265244603157044, "step": 27530 }, { "epoch": 0.7254467776853993, "grad_norm": 4.807163238525391, "learning_rate": 1.8502608136785311e-06, "loss": 0.094341242313385, "step": 27540 }, { "epoch": 0.7257101933635712, "grad_norm": 3.4143476486206055, "learning_rate": 1.846949068617766e-06, "loss": 0.10235728025436401, "step": 27550 }, { "epoch": 0.7259736090417431, "grad_norm": 3.221083164215088, "learning_rate": 1.8436396184835037e-06, "loss": 0.10847761631011962, "step": 27560 }, { "epoch": 0.726237024719915, "grad_norm": 3.230811834335327, "learning_rate": 1.8403324656845033e-06, "loss": 0.10579416751861573, "step": 27570 }, { "epoch": 0.7265004403980869, "grad_norm": 3.797189474105835, "learning_rate": 1.8370276126278546e-06, "loss": 0.09530079364776611, "step": 27580 }, { "epoch": 0.7267638560762588, "grad_norm": 3.269115924835205, "learning_rate": 1.8337250617189695e-06, "loss": 0.0962449073791504, "step": 27590 }, { "epoch": 0.7270272717544307, "grad_norm": 3.028817653656006, "learning_rate": 1.8304248153615912e-06, "loss": 0.10141491889953613, "step": 27600 }, { "epoch": 0.7272906874326026, "grad_norm": 3.860799789428711, "learning_rate": 1.8271268759577808e-06, "loss": 0.1110733151435852, "step": 27610 }, { "epoch": 0.7275541031107745, "grad_norm": 3.7293474674224854, "learning_rate": 1.8238312459079166e-06, "loss": 0.09322381019592285, "step": 27620 }, { "epoch": 0.7278175187889464, "grad_norm": 3.810595750808716, "learning_rate": 1.8205379276107044e-06, "loss": 0.09654682874679565, "step": 27630 }, { "epoch": 0.7280809344671183, "grad_norm": 4.502783298492432, "learning_rate": 1.8172469234631606e-06, "loss": 0.1014490008354187, "step": 27640 }, { "epoch": 0.7283443501452902, "grad_norm": 4.663501262664795, "learning_rate": 1.813958235860625e-06, "loss": 0.095614755153656, "step": 27650 }, { "epoch": 0.7286077658234621, "grad_norm": 3.619304895401001, "learning_rate": 1.810671867196741e-06, "loss": 0.09377279877662659, "step": 27660 }, { "epoch": 0.728871181501634, "grad_norm": 3.840089797973633, "learning_rate": 1.8073878198634698e-06, "loss": 0.11089950799942017, "step": 27670 }, { "epoch": 0.7291345971798059, "grad_norm": 3.829578161239624, "learning_rate": 1.8041060962510859e-06, "loss": 0.10113773345947266, "step": 27680 }, { "epoch": 0.7293980128579778, "grad_norm": 3.8716702461242676, "learning_rate": 1.8008266987481675e-06, "loss": 0.10914849042892456, "step": 27690 }, { "epoch": 0.7296614285361497, "grad_norm": 3.166590929031372, "learning_rate": 1.7975496297416022e-06, "loss": 0.098094242811203, "step": 27700 }, { "epoch": 0.7299248442143216, "grad_norm": 3.849968433380127, "learning_rate": 1.794274891616582e-06, "loss": 0.09216553568840027, "step": 27710 }, { "epoch": 0.7301882598924935, "grad_norm": 3.8580455780029297, "learning_rate": 1.7910024867566005e-06, "loss": 0.09851531982421875, "step": 27720 }, { "epoch": 0.7304516755706654, "grad_norm": 4.132777214050293, "learning_rate": 1.7877324175434585e-06, "loss": 0.09960825443267822, "step": 27730 }, { "epoch": 0.7307150912488373, "grad_norm": 4.080651760101318, "learning_rate": 1.784464686357254e-06, "loss": 0.10683759450912475, "step": 27740 }, { "epoch": 0.7309785069270092, "grad_norm": 3.8246729373931885, "learning_rate": 1.7811992955763774e-06, "loss": 0.09525618553161622, "step": 27750 }, { "epoch": 0.7312419226051811, "grad_norm": 4.837862491607666, "learning_rate": 1.7779362475775257e-06, "loss": 0.10363579988479614, "step": 27760 }, { "epoch": 0.731505338283353, "grad_norm": 3.3535146713256836, "learning_rate": 1.7746755447356834e-06, "loss": 0.09333987832069397, "step": 27770 }, { "epoch": 0.7317687539615249, "grad_norm": 4.776111125946045, "learning_rate": 1.771417189424135e-06, "loss": 0.09735201597213745, "step": 27780 }, { "epoch": 0.7320321696396968, "grad_norm": 3.6245920658111572, "learning_rate": 1.7681611840144475e-06, "loss": 0.08648754954338074, "step": 27790 }, { "epoch": 0.7322955853178686, "grad_norm": 4.086280822753906, "learning_rate": 1.7649075308764818e-06, "loss": 0.10370461940765381, "step": 27800 }, { "epoch": 0.7325590009960405, "grad_norm": 3.5990829467773438, "learning_rate": 1.7616562323783908e-06, "loss": 0.09604442119598389, "step": 27810 }, { "epoch": 0.7328224166742124, "grad_norm": 4.1878204345703125, "learning_rate": 1.758407290886608e-06, "loss": 0.09502832889556885, "step": 27820 }, { "epoch": 0.7330858323523843, "grad_norm": 4.989692211151123, "learning_rate": 1.7551607087658528e-06, "loss": 0.10411978960037231, "step": 27830 }, { "epoch": 0.7333492480305562, "grad_norm": 5.54674768447876, "learning_rate": 1.7519164883791283e-06, "loss": 0.0951109230518341, "step": 27840 }, { "epoch": 0.7336126637087281, "grad_norm": 5.086316108703613, "learning_rate": 1.7486746320877157e-06, "loss": 0.11072013378143311, "step": 27850 }, { "epoch": 0.7338760793869, "grad_norm": 2.9218616485595703, "learning_rate": 1.7454351422511822e-06, "loss": 0.09701960682868957, "step": 27860 }, { "epoch": 0.7341394950650719, "grad_norm": 3.9180593490600586, "learning_rate": 1.742198021227367e-06, "loss": 0.09903166294097901, "step": 27870 }, { "epoch": 0.7344029107432438, "grad_norm": 3.5816235542297363, "learning_rate": 1.738963271372383e-06, "loss": 0.09210343360900879, "step": 27880 }, { "epoch": 0.7346663264214157, "grad_norm": 4.425755977630615, "learning_rate": 1.7357308950406242e-06, "loss": 0.09560789465904236, "step": 27890 }, { "epoch": 0.7349297420995876, "grad_norm": 4.021759510040283, "learning_rate": 1.732500894584751e-06, "loss": 0.1000502347946167, "step": 27900 }, { "epoch": 0.7351931577777595, "grad_norm": 4.083045959472656, "learning_rate": 1.729273272355702e-06, "loss": 0.08691344857215881, "step": 27910 }, { "epoch": 0.7354565734559314, "grad_norm": 3.1217234134674072, "learning_rate": 1.726048030702675e-06, "loss": 0.09523463249206543, "step": 27920 }, { "epoch": 0.7357199891341033, "grad_norm": 4.330427646636963, "learning_rate": 1.7228251719731398e-06, "loss": 0.08994665145874023, "step": 27930 }, { "epoch": 0.7359834048122752, "grad_norm": 3.3579814434051514, "learning_rate": 1.7196046985128352e-06, "loss": 0.09446653127670288, "step": 27940 }, { "epoch": 0.7362468204904471, "grad_norm": 4.140453815460205, "learning_rate": 1.7163866126657598e-06, "loss": 0.10321574211120606, "step": 27950 }, { "epoch": 0.736510236168619, "grad_norm": 4.01765251159668, "learning_rate": 1.7131709167741738e-06, "loss": 0.10606765747070312, "step": 27960 }, { "epoch": 0.7367736518467909, "grad_norm": 4.285407543182373, "learning_rate": 1.7099576131786016e-06, "loss": 0.0988570511341095, "step": 27970 }, { "epoch": 0.7370370675249628, "grad_norm": 3.5935721397399902, "learning_rate": 1.706746704217821e-06, "loss": 0.09467272162437439, "step": 27980 }, { "epoch": 0.7373004832031347, "grad_norm": 3.0668373107910156, "learning_rate": 1.7035381922288735e-06, "loss": 0.09746504426002503, "step": 27990 }, { "epoch": 0.7375638988813066, "grad_norm": 2.3875937461853027, "learning_rate": 1.700332079547053e-06, "loss": 0.09225062131881714, "step": 28000 }, { "epoch": 0.7378273145594785, "grad_norm": 4.447257995605469, "learning_rate": 1.6971283685059015e-06, "loss": 0.08933742046356201, "step": 28010 }, { "epoch": 0.7380907302376504, "grad_norm": 3.930229902267456, "learning_rate": 1.6939270614372227e-06, "loss": 0.08913971185684204, "step": 28020 }, { "epoch": 0.7383541459158223, "grad_norm": 3.1525702476501465, "learning_rate": 1.690728160671065e-06, "loss": 0.0942482054233551, "step": 28030 }, { "epoch": 0.7386175615939942, "grad_norm": 5.699667930603027, "learning_rate": 1.6875316685357264e-06, "loss": 0.10229434967041015, "step": 28040 }, { "epoch": 0.7388809772721661, "grad_norm": 2.9590229988098145, "learning_rate": 1.6843375873577505e-06, "loss": 0.09440884590148926, "step": 28050 }, { "epoch": 0.739144392950338, "grad_norm": 2.8406620025634766, "learning_rate": 1.6811459194619272e-06, "loss": 0.08861490488052368, "step": 28060 }, { "epoch": 0.7394078086285099, "grad_norm": 5.595778942108154, "learning_rate": 1.6779566671712933e-06, "loss": 0.09488941431045532, "step": 28070 }, { "epoch": 0.7396712243066818, "grad_norm": 6.7842936515808105, "learning_rate": 1.6747698328071221e-06, "loss": 0.0933147668838501, "step": 28080 }, { "epoch": 0.7399346399848536, "grad_norm": 3.4611501693725586, "learning_rate": 1.671585418688929e-06, "loss": 0.08023813962936402, "step": 28090 }, { "epoch": 0.7401980556630255, "grad_norm": 3.1986215114593506, "learning_rate": 1.668403427134468e-06, "loss": 0.09805140495300294, "step": 28100 }, { "epoch": 0.7404614713411974, "grad_norm": 3.7319674491882324, "learning_rate": 1.6652238604597287e-06, "loss": 0.09511905312538146, "step": 28110 }, { "epoch": 0.7407248870193693, "grad_norm": 5.542313575744629, "learning_rate": 1.6620467209789414e-06, "loss": 0.09238584041595459, "step": 28120 }, { "epoch": 0.7409883026975412, "grad_norm": 5.3329620361328125, "learning_rate": 1.6588720110045603e-06, "loss": 0.10941808223724366, "step": 28130 }, { "epoch": 0.741251718375713, "grad_norm": 3.9589107036590576, "learning_rate": 1.6556997328472769e-06, "loss": 0.09103896617889404, "step": 28140 }, { "epoch": 0.7415151340538849, "grad_norm": 3.7262258529663086, "learning_rate": 1.6525298888160147e-06, "loss": 0.09488558769226074, "step": 28150 }, { "epoch": 0.7417785497320568, "grad_norm": 3.9207606315612793, "learning_rate": 1.649362481217921e-06, "loss": 0.09121515750885009, "step": 28160 }, { "epoch": 0.7420419654102287, "grad_norm": 2.9725303649902344, "learning_rate": 1.646197512358373e-06, "loss": 0.10181220769882202, "step": 28170 }, { "epoch": 0.7423053810884006, "grad_norm": 3.1318085193634033, "learning_rate": 1.6430349845409703e-06, "loss": 0.09501914978027344, "step": 28180 }, { "epoch": 0.7425687967665725, "grad_norm": 4.456239700317383, "learning_rate": 1.6398749000675369e-06, "loss": 0.09772093892097473, "step": 28190 }, { "epoch": 0.7428322124447444, "grad_norm": 3.5815346240997314, "learning_rate": 1.63671726123812e-06, "loss": 0.10425121784210205, "step": 28200 }, { "epoch": 0.7430956281229163, "grad_norm": 4.284858703613281, "learning_rate": 1.633562070350987e-06, "loss": 0.09456642866134643, "step": 28210 }, { "epoch": 0.7433590438010882, "grad_norm": 3.680537462234497, "learning_rate": 1.6304093297026157e-06, "loss": 0.09752981662750244, "step": 28220 }, { "epoch": 0.7436224594792601, "grad_norm": 5.10155725479126, "learning_rate": 1.6272590415877132e-06, "loss": 0.10648648738861084, "step": 28230 }, { "epoch": 0.743885875157432, "grad_norm": 3.0203301906585693, "learning_rate": 1.6241112082991905e-06, "loss": 0.09672675132751465, "step": 28240 }, { "epoch": 0.7441492908356039, "grad_norm": 4.567969799041748, "learning_rate": 1.6209658321281823e-06, "loss": 0.09464845657348633, "step": 28250 }, { "epoch": 0.7444127065137758, "grad_norm": 3.3290855884552, "learning_rate": 1.6178229153640234e-06, "loss": 0.08340182900428772, "step": 28260 }, { "epoch": 0.7446761221919477, "grad_norm": 5.620272636413574, "learning_rate": 1.6146824602942646e-06, "loss": 0.09807928204536438, "step": 28270 }, { "epoch": 0.7449395378701196, "grad_norm": 4.067939758300781, "learning_rate": 1.6115444692046677e-06, "loss": 0.08792108297348022, "step": 28280 }, { "epoch": 0.7452029535482915, "grad_norm": 3.3927059173583984, "learning_rate": 1.6084089443791973e-06, "loss": 0.08680953979492187, "step": 28290 }, { "epoch": 0.7454663692264634, "grad_norm": 4.149141311645508, "learning_rate": 1.605275888100022e-06, "loss": 0.10212616920471192, "step": 28300 }, { "epoch": 0.7457297849046353, "grad_norm": 3.2074875831604004, "learning_rate": 1.6021453026475165e-06, "loss": 0.09560261964797974, "step": 28310 }, { "epoch": 0.7459932005828072, "grad_norm": 4.903495788574219, "learning_rate": 1.5990171903002539e-06, "loss": 0.10674247741699219, "step": 28320 }, { "epoch": 0.746256616260979, "grad_norm": 5.226134777069092, "learning_rate": 1.5958915533350128e-06, "loss": 0.08812123537063599, "step": 28330 }, { "epoch": 0.746520031939151, "grad_norm": 4.357222557067871, "learning_rate": 1.5927683940267668e-06, "loss": 0.10379999876022339, "step": 28340 }, { "epoch": 0.7467834476173228, "grad_norm": 4.056844234466553, "learning_rate": 1.5896477146486806e-06, "loss": 0.10656731128692627, "step": 28350 }, { "epoch": 0.7470468632954947, "grad_norm": 4.41269063949585, "learning_rate": 1.586529517472124e-06, "loss": 0.09822221994400024, "step": 28360 }, { "epoch": 0.7473102789736666, "grad_norm": 4.1405253410339355, "learning_rate": 1.5834138047666532e-06, "loss": 0.09426534175872803, "step": 28370 }, { "epoch": 0.7475736946518385, "grad_norm": 4.8449273109436035, "learning_rate": 1.5803005788000225e-06, "loss": 0.08603735566139221, "step": 28380 }, { "epoch": 0.7478371103300104, "grad_norm": 4.493166923522949, "learning_rate": 1.577189841838167e-06, "loss": 0.08977635502815247, "step": 28390 }, { "epoch": 0.7481005260081823, "grad_norm": 4.6330156326293945, "learning_rate": 1.574081596145216e-06, "loss": 0.09504960775375366, "step": 28400 }, { "epoch": 0.7483639416863542, "grad_norm": 5.4016594886779785, "learning_rate": 1.5709758439834872e-06, "loss": 0.10126268863677979, "step": 28410 }, { "epoch": 0.7486273573645261, "grad_norm": 4.682175159454346, "learning_rate": 1.5678725876134797e-06, "loss": 0.10614111423492431, "step": 28420 }, { "epoch": 0.748890773042698, "grad_norm": 3.1608355045318604, "learning_rate": 1.5647718292938779e-06, "loss": 0.09646879434585572, "step": 28430 }, { "epoch": 0.7491541887208699, "grad_norm": 3.665823221206665, "learning_rate": 1.5616735712815474e-06, "loss": 0.10114550590515137, "step": 28440 }, { "epoch": 0.7494176043990418, "grad_norm": 3.5184922218322754, "learning_rate": 1.5585778158315313e-06, "loss": 0.09694585800170899, "step": 28450 }, { "epoch": 0.7496810200772137, "grad_norm": 4.552343368530273, "learning_rate": 1.5554845651970584e-06, "loss": 0.10010421276092529, "step": 28460 }, { "epoch": 0.7499444357553856, "grad_norm": 3.5863609313964844, "learning_rate": 1.5523938216295298e-06, "loss": 0.08383595943450928, "step": 28470 }, { "epoch": 0.7502078514335575, "grad_norm": 3.6256537437438965, "learning_rate": 1.5493055873785168e-06, "loss": 0.09524009823799133, "step": 28480 }, { "epoch": 0.7504712671117294, "grad_norm": 4.787498474121094, "learning_rate": 1.546219864691774e-06, "loss": 0.1049943208694458, "step": 28490 }, { "epoch": 0.7507346827899013, "grad_norm": 4.109622001647949, "learning_rate": 1.5431366558152218e-06, "loss": 0.09796488881111146, "step": 28500 }, { "epoch": 0.7509980984680732, "grad_norm": 3.45849609375, "learning_rate": 1.5400559629929558e-06, "loss": 0.09075171351432801, "step": 28510 }, { "epoch": 0.7512615141462451, "grad_norm": 5.011972904205322, "learning_rate": 1.5369777884672333e-06, "loss": 0.10000107288360596, "step": 28520 }, { "epoch": 0.751524929824417, "grad_norm": 4.635426998138428, "learning_rate": 1.533902134478482e-06, "loss": 0.10514397621154785, "step": 28530 }, { "epoch": 0.7517883455025889, "grad_norm": 3.0275964736938477, "learning_rate": 1.5308290032652989e-06, "loss": 0.08492684364318848, "step": 28540 }, { "epoch": 0.7520517611807608, "grad_norm": 4.034580707550049, "learning_rate": 1.5277583970644405e-06, "loss": 0.09531847834587097, "step": 28550 }, { "epoch": 0.7523151768589327, "grad_norm": 3.8110949993133545, "learning_rate": 1.5246903181108258e-06, "loss": 0.08597640991210938, "step": 28560 }, { "epoch": 0.7525785925371046, "grad_norm": 4.641963958740234, "learning_rate": 1.5216247686375346e-06, "loss": 0.09052932262420654, "step": 28570 }, { "epoch": 0.7528420082152765, "grad_norm": 4.300163745880127, "learning_rate": 1.5185617508758055e-06, "loss": 0.08998308181762696, "step": 28580 }, { "epoch": 0.7531054238934484, "grad_norm": 5.218149662017822, "learning_rate": 1.5155012670550378e-06, "loss": 0.09843692779541016, "step": 28590 }, { "epoch": 0.7533688395716203, "grad_norm": 6.330803871154785, "learning_rate": 1.5124433194027837e-06, "loss": 0.09054800271987914, "step": 28600 }, { "epoch": 0.7536322552497922, "grad_norm": 4.017702102661133, "learning_rate": 1.509387910144744e-06, "loss": 0.09582479000091552, "step": 28610 }, { "epoch": 0.7538956709279641, "grad_norm": 3.742475986480713, "learning_rate": 1.5063350415047834e-06, "loss": 0.09053283929824829, "step": 28620 }, { "epoch": 0.754159086606136, "grad_norm": 3.5800342559814453, "learning_rate": 1.5032847157049084e-06, "loss": 0.09431655406951904, "step": 28630 }, { "epoch": 0.7544225022843078, "grad_norm": 2.96553373336792, "learning_rate": 1.500236934965279e-06, "loss": 0.08940424919128417, "step": 28640 }, { "epoch": 0.7546859179624797, "grad_norm": 3.689202308654785, "learning_rate": 1.4971917015042003e-06, "loss": 0.09759345054626464, "step": 28650 }, { "epoch": 0.7549493336406516, "grad_norm": 4.096159934997559, "learning_rate": 1.4941490175381245e-06, "loss": 0.08148479461669922, "step": 28660 }, { "epoch": 0.7552127493188235, "grad_norm": 4.784111976623535, "learning_rate": 1.4911088852816503e-06, "loss": 0.09910074472427369, "step": 28670 }, { "epoch": 0.7554761649969954, "grad_norm": 4.6106791496276855, "learning_rate": 1.4880713069475155e-06, "loss": 0.09291614890098572, "step": 28680 }, { "epoch": 0.7557395806751673, "grad_norm": 4.031212329864502, "learning_rate": 1.485036284746601e-06, "loss": 0.08179665207862855, "step": 28690 }, { "epoch": 0.7560029963533392, "grad_norm": 4.287348747253418, "learning_rate": 1.4820038208879273e-06, "loss": 0.10108697414398193, "step": 28700 }, { "epoch": 0.7562664120315111, "grad_norm": 3.585833787918091, "learning_rate": 1.47897391757865e-06, "loss": 0.09386075139045716, "step": 28710 }, { "epoch": 0.756529827709683, "grad_norm": 3.7468173503875732, "learning_rate": 1.4759465770240689e-06, "loss": 0.08809939622879029, "step": 28720 }, { "epoch": 0.7567932433878549, "grad_norm": 3.6541922092437744, "learning_rate": 1.4729218014276075e-06, "loss": 0.09451398849487305, "step": 28730 }, { "epoch": 0.7570566590660268, "grad_norm": 5.071716785430908, "learning_rate": 1.4698995929908294e-06, "loss": 0.09398081302642822, "step": 28740 }, { "epoch": 0.7573200747441987, "grad_norm": 2.480135202407837, "learning_rate": 1.4668799539134299e-06, "loss": 0.087790447473526, "step": 28750 }, { "epoch": 0.7575834904223706, "grad_norm": 3.3764028549194336, "learning_rate": 1.463862886393232e-06, "loss": 0.10083763599395752, "step": 28760 }, { "epoch": 0.7578469061005425, "grad_norm": 2.9048850536346436, "learning_rate": 1.4608483926261874e-06, "loss": 0.09739016890525817, "step": 28770 }, { "epoch": 0.7581103217787144, "grad_norm": 3.7227513790130615, "learning_rate": 1.4578364748063745e-06, "loss": 0.11022648811340333, "step": 28780 }, { "epoch": 0.7583737374568863, "grad_norm": 5.020188331604004, "learning_rate": 1.4548271351259952e-06, "loss": 0.09915226697921753, "step": 28790 }, { "epoch": 0.7586371531350582, "grad_norm": 4.847402095794678, "learning_rate": 1.4518203757753807e-06, "loss": 0.08277674913406372, "step": 28800 }, { "epoch": 0.7589005688132301, "grad_norm": 3.9860823154449463, "learning_rate": 1.44881619894298e-06, "loss": 0.07933559417724609, "step": 28810 }, { "epoch": 0.759163984491402, "grad_norm": 4.745926380157471, "learning_rate": 1.4458146068153573e-06, "loss": 0.09759080410003662, "step": 28820 }, { "epoch": 0.7594274001695739, "grad_norm": 3.6357150077819824, "learning_rate": 1.442815601577206e-06, "loss": 0.0855612337589264, "step": 28830 }, { "epoch": 0.7596908158477458, "grad_norm": 3.5143730640411377, "learning_rate": 1.439819185411328e-06, "loss": 0.0794224739074707, "step": 28840 }, { "epoch": 0.7599542315259177, "grad_norm": 4.035648345947266, "learning_rate": 1.4368253604986498e-06, "loss": 0.08827892541885377, "step": 28850 }, { "epoch": 0.7602176472040896, "grad_norm": 4.0527119636535645, "learning_rate": 1.4338341290182006e-06, "loss": 0.10405241250991822, "step": 28860 }, { "epoch": 0.7604810628822615, "grad_norm": 3.8532660007476807, "learning_rate": 1.4308454931471282e-06, "loss": 0.09408077001571655, "step": 28870 }, { "epoch": 0.7607444785604334, "grad_norm": 5.114193439483643, "learning_rate": 1.4278594550606945e-06, "loss": 0.08863416314125061, "step": 28880 }, { "epoch": 0.7610078942386053, "grad_norm": 7.687660217285156, "learning_rate": 1.4248760169322639e-06, "loss": 0.08587610721588135, "step": 28890 }, { "epoch": 0.7612713099167772, "grad_norm": 5.2714643478393555, "learning_rate": 1.4218951809333125e-06, "loss": 0.09245144724845886, "step": 28900 }, { "epoch": 0.7615347255949491, "grad_norm": 4.777899742126465, "learning_rate": 1.418916949233421e-06, "loss": 0.10860984325408936, "step": 28910 }, { "epoch": 0.761798141273121, "grad_norm": 4.460324287414551, "learning_rate": 1.4159413240002734e-06, "loss": 0.10670716762542724, "step": 28920 }, { "epoch": 0.7620615569512927, "grad_norm": 4.426624774932861, "learning_rate": 1.4129683073996608e-06, "loss": 0.10253567695617676, "step": 28930 }, { "epoch": 0.7623249726294646, "grad_norm": 3.684602737426758, "learning_rate": 1.409997901595474e-06, "loss": 0.08418475389480591, "step": 28940 }, { "epoch": 0.7625883883076365, "grad_norm": 5.514393329620361, "learning_rate": 1.4070301087496964e-06, "loss": 0.09164879322052003, "step": 28950 }, { "epoch": 0.7628518039858084, "grad_norm": 4.15283203125, "learning_rate": 1.4040649310224214e-06, "loss": 0.09345952272415162, "step": 28960 }, { "epoch": 0.7631152196639803, "grad_norm": 4.153979778289795, "learning_rate": 1.4011023705718302e-06, "loss": 0.09801514148712158, "step": 28970 }, { "epoch": 0.7633786353421522, "grad_norm": 4.64967679977417, "learning_rate": 1.3981424295542061e-06, "loss": 0.08549781441688538, "step": 28980 }, { "epoch": 0.7636420510203241, "grad_norm": 3.8691554069519043, "learning_rate": 1.3951851101239177e-06, "loss": 0.08145334720611572, "step": 28990 }, { "epoch": 0.763905466698496, "grad_norm": 5.216892719268799, "learning_rate": 1.3922304144334293e-06, "loss": 0.10301430225372314, "step": 29000 }, { "epoch": 0.7641688823766679, "grad_norm": 4.320065021514893, "learning_rate": 1.3892783446332996e-06, "loss": 0.09908955097198487, "step": 29010 }, { "epoch": 0.7644322980548398, "grad_norm": 4.102685451507568, "learning_rate": 1.38632890287217e-06, "loss": 0.07821484804153442, "step": 29020 }, { "epoch": 0.7646957137330117, "grad_norm": 3.7294647693634033, "learning_rate": 1.383382091296772e-06, "loss": 0.09788757562637329, "step": 29030 }, { "epoch": 0.7649591294111836, "grad_norm": 4.736691474914551, "learning_rate": 1.380437912051923e-06, "loss": 0.08540483117103577, "step": 29040 }, { "epoch": 0.7652225450893555, "grad_norm": 4.178765296936035, "learning_rate": 1.3774963672805204e-06, "loss": 0.09730057716369629, "step": 29050 }, { "epoch": 0.7654859607675274, "grad_norm": 4.191710472106934, "learning_rate": 1.374557459123551e-06, "loss": 0.08888072967529297, "step": 29060 }, { "epoch": 0.7657493764456993, "grad_norm": 4.339596748352051, "learning_rate": 1.3716211897200799e-06, "loss": 0.09318064451217652, "step": 29070 }, { "epoch": 0.7660127921238712, "grad_norm": 2.97560715675354, "learning_rate": 1.3686875612072452e-06, "loss": 0.09461700320243835, "step": 29080 }, { "epoch": 0.7662762078020431, "grad_norm": 4.140294551849365, "learning_rate": 1.3657565757202734e-06, "loss": 0.08196247816085815, "step": 29090 }, { "epoch": 0.766539623480215, "grad_norm": 6.919163703918457, "learning_rate": 1.3628282353924582e-06, "loss": 0.08395624160766602, "step": 29100 }, { "epoch": 0.7668030391583869, "grad_norm": 4.32041072845459, "learning_rate": 1.359902542355178e-06, "loss": 0.09375180006027221, "step": 29110 }, { "epoch": 0.7670664548365588, "grad_norm": 4.383762836456299, "learning_rate": 1.3569794987378727e-06, "loss": 0.08688862919807434, "step": 29120 }, { "epoch": 0.7673298705147307, "grad_norm": 4.742859840393066, "learning_rate": 1.3540591066680604e-06, "loss": 0.09771167039871216, "step": 29130 }, { "epoch": 0.7675932861929026, "grad_norm": 4.333304405212402, "learning_rate": 1.3511413682713303e-06, "loss": 0.0822193443775177, "step": 29140 }, { "epoch": 0.7678567018710745, "grad_norm": 4.244485855102539, "learning_rate": 1.348226285671338e-06, "loss": 0.1099165439605713, "step": 29150 }, { "epoch": 0.7681201175492464, "grad_norm": 4.959094047546387, "learning_rate": 1.3453138609898058e-06, "loss": 0.08962397575378418, "step": 29160 }, { "epoch": 0.7683835332274183, "grad_norm": 2.643094062805176, "learning_rate": 1.3424040963465223e-06, "loss": 0.09073961973190307, "step": 29170 }, { "epoch": 0.7686469489055902, "grad_norm": 4.864450454711914, "learning_rate": 1.3394969938593376e-06, "loss": 0.08678988218307496, "step": 29180 }, { "epoch": 0.768910364583762, "grad_norm": 4.395050525665283, "learning_rate": 1.33659255564417e-06, "loss": 0.09002103209495545, "step": 29190 }, { "epoch": 0.769173780261934, "grad_norm": 4.335933208465576, "learning_rate": 1.333690783814995e-06, "loss": 0.08247475624084473, "step": 29200 }, { "epoch": 0.7694371959401058, "grad_norm": 3.398036241531372, "learning_rate": 1.3307916804838423e-06, "loss": 0.0953490138053894, "step": 29210 }, { "epoch": 0.7697006116182777, "grad_norm": 4.23280143737793, "learning_rate": 1.3278952477608093e-06, "loss": 0.08757768273353576, "step": 29220 }, { "epoch": 0.7699640272964496, "grad_norm": 5.802147388458252, "learning_rate": 1.325001487754043e-06, "loss": 0.09002179503440857, "step": 29230 }, { "epoch": 0.7702274429746215, "grad_norm": 4.100104331970215, "learning_rate": 1.3221104025697468e-06, "loss": 0.08818495273590088, "step": 29240 }, { "epoch": 0.7704908586527934, "grad_norm": 3.537170648574829, "learning_rate": 1.3192219943121776e-06, "loss": 0.08237948417663574, "step": 29250 }, { "epoch": 0.7707542743309653, "grad_norm": 4.285287380218506, "learning_rate": 1.3163362650836415e-06, "loss": 0.09606741666793824, "step": 29260 }, { "epoch": 0.7710176900091372, "grad_norm": 4.164183139801025, "learning_rate": 1.3134532169844993e-06, "loss": 0.09096962809562684, "step": 29270 }, { "epoch": 0.7712811056873091, "grad_norm": 4.199525833129883, "learning_rate": 1.3105728521131579e-06, "loss": 0.0985458493232727, "step": 29280 }, { "epoch": 0.771544521365481, "grad_norm": 5.331406116485596, "learning_rate": 1.3076951725660692e-06, "loss": 0.11085216999053955, "step": 29290 }, { "epoch": 0.7718079370436529, "grad_norm": 3.6802852153778076, "learning_rate": 1.3048201804377343e-06, "loss": 0.09145951271057129, "step": 29300 }, { "epoch": 0.7720713527218248, "grad_norm": 4.8262810707092285, "learning_rate": 1.3019478778206935e-06, "loss": 0.10871182680130005, "step": 29310 }, { "epoch": 0.7723347683999967, "grad_norm": 4.34891414642334, "learning_rate": 1.2990782668055384e-06, "loss": 0.09857029914855957, "step": 29320 }, { "epoch": 0.7725981840781686, "grad_norm": 5.490789890289307, "learning_rate": 1.2962113494808904e-06, "loss": 0.09608142375946045, "step": 29330 }, { "epoch": 0.7728615997563405, "grad_norm": 3.6673948764801025, "learning_rate": 1.2933471279334154e-06, "loss": 0.09089683294296265, "step": 29340 }, { "epoch": 0.7731250154345124, "grad_norm": 4.397944450378418, "learning_rate": 1.290485604247821e-06, "loss": 0.09247716069221497, "step": 29350 }, { "epoch": 0.7733884311126843, "grad_norm": 3.786855936050415, "learning_rate": 1.287626780506846e-06, "loss": 0.08947888612747193, "step": 29360 }, { "epoch": 0.7736518467908562, "grad_norm": 4.429439067840576, "learning_rate": 1.284770658791265e-06, "loss": 0.10070818662643433, "step": 29370 }, { "epoch": 0.7739152624690281, "grad_norm": 4.0725603103637695, "learning_rate": 1.2819172411798881e-06, "loss": 0.09188002347946167, "step": 29380 }, { "epoch": 0.7741786781472, "grad_norm": 5.955046653747559, "learning_rate": 1.2790665297495526e-06, "loss": 0.09744277596473694, "step": 29390 }, { "epoch": 0.7744420938253719, "grad_norm": 4.019720554351807, "learning_rate": 1.276218526575133e-06, "loss": 0.09405975341796875, "step": 29400 }, { "epoch": 0.7747055095035438, "grad_norm": 4.124478816986084, "learning_rate": 1.273373233729529e-06, "loss": 0.08949400782585144, "step": 29410 }, { "epoch": 0.7749689251817157, "grad_norm": 5.387200355529785, "learning_rate": 1.2705306532836625e-06, "loss": 0.08301681280136108, "step": 29420 }, { "epoch": 0.7752323408598876, "grad_norm": 5.048255920410156, "learning_rate": 1.2676907873064913e-06, "loss": 0.09192358255386353, "step": 29430 }, { "epoch": 0.7754957565380595, "grad_norm": 3.667151689529419, "learning_rate": 1.2648536378649894e-06, "loss": 0.08869022130966187, "step": 29440 }, { "epoch": 0.7757591722162314, "grad_norm": 3.7798736095428467, "learning_rate": 1.2620192070241604e-06, "loss": 0.07594031095504761, "step": 29450 }, { "epoch": 0.7760225878944033, "grad_norm": 4.381338596343994, "learning_rate": 1.2591874968470214e-06, "loss": 0.08218237161636352, "step": 29460 }, { "epoch": 0.7762860035725752, "grad_norm": 3.8121540546417236, "learning_rate": 1.2563585093946129e-06, "loss": 0.08616187572479247, "step": 29470 }, { "epoch": 0.776549419250747, "grad_norm": 4.319460391998291, "learning_rate": 1.2535322467259975e-06, "loss": 0.09111983776092529, "step": 29480 }, { "epoch": 0.776812834928919, "grad_norm": 5.053682327270508, "learning_rate": 1.250708710898249e-06, "loss": 0.08978927731513978, "step": 29490 }, { "epoch": 0.7770762506070908, "grad_norm": 3.8018605709075928, "learning_rate": 1.2478879039664587e-06, "loss": 0.08550801277160644, "step": 29500 }, { "epoch": 0.7773396662852627, "grad_norm": 4.873858451843262, "learning_rate": 1.2450698279837325e-06, "loss": 0.07883515954017639, "step": 29510 }, { "epoch": 0.7776030819634346, "grad_norm": 3.9109573364257812, "learning_rate": 1.2422544850011847e-06, "loss": 0.08781424760818482, "step": 29520 }, { "epoch": 0.7778664976416065, "grad_norm": 3.771965980529785, "learning_rate": 1.2394418770679478e-06, "loss": 0.09056259393692016, "step": 29530 }, { "epoch": 0.7781299133197784, "grad_norm": 4.17999267578125, "learning_rate": 1.2366320062311576e-06, "loss": 0.09345223307609558, "step": 29540 }, { "epoch": 0.7783933289979503, "grad_norm": 3.426865339279175, "learning_rate": 1.2338248745359565e-06, "loss": 0.08404537439346313, "step": 29550 }, { "epoch": 0.7786567446761222, "grad_norm": 4.551056385040283, "learning_rate": 1.2310204840254996e-06, "loss": 0.09431884884834289, "step": 29560 }, { "epoch": 0.7789201603542941, "grad_norm": 3.159792423248291, "learning_rate": 1.2282188367409409e-06, "loss": 0.07816271781921387, "step": 29570 }, { "epoch": 0.779183576032466, "grad_norm": 4.775075435638428, "learning_rate": 1.2254199347214434e-06, "loss": 0.09022020101547241, "step": 29580 }, { "epoch": 0.7794469917106379, "grad_norm": 3.9869937896728516, "learning_rate": 1.2226237800041662e-06, "loss": 0.08488813042640686, "step": 29590 }, { "epoch": 0.7797104073888098, "grad_norm": 3.9529080390930176, "learning_rate": 1.2198303746242697e-06, "loss": 0.08689780831336975, "step": 29600 }, { "epoch": 0.7799738230669817, "grad_norm": 3.8580551147460938, "learning_rate": 1.2170397206149182e-06, "loss": 0.0850167989730835, "step": 29610 }, { "epoch": 0.7802372387451536, "grad_norm": 4.321518898010254, "learning_rate": 1.2142518200072695e-06, "loss": 0.10055444240570069, "step": 29620 }, { "epoch": 0.7805006544233255, "grad_norm": 4.583689212799072, "learning_rate": 1.2114666748304766e-06, "loss": 0.09132797122001649, "step": 29630 }, { "epoch": 0.7807640701014974, "grad_norm": 5.361180305480957, "learning_rate": 1.2086842871116894e-06, "loss": 0.08865251541137695, "step": 29640 }, { "epoch": 0.7810274857796693, "grad_norm": 3.887558698654175, "learning_rate": 1.2059046588760476e-06, "loss": 0.0925408124923706, "step": 29650 }, { "epoch": 0.7812909014578412, "grad_norm": 5.362173557281494, "learning_rate": 1.2031277921466877e-06, "loss": 0.08974781036376953, "step": 29660 }, { "epoch": 0.7815543171360131, "grad_norm": 4.938616752624512, "learning_rate": 1.2003536889447332e-06, "loss": 0.0916597604751587, "step": 29670 }, { "epoch": 0.781817732814185, "grad_norm": 4.059353351593018, "learning_rate": 1.1975823512892914e-06, "loss": 0.09048691987991334, "step": 29680 }, { "epoch": 0.7820811484923569, "grad_norm": 10.388577461242676, "learning_rate": 1.1948137811974663e-06, "loss": 0.10072776079177856, "step": 29690 }, { "epoch": 0.7823445641705288, "grad_norm": 3.9009954929351807, "learning_rate": 1.19204798068434e-06, "loss": 0.07447519302368164, "step": 29700 }, { "epoch": 0.7826079798487007, "grad_norm": 4.508605480194092, "learning_rate": 1.1892849517629845e-06, "loss": 0.08846876621246338, "step": 29710 }, { "epoch": 0.7828713955268725, "grad_norm": 3.9755797386169434, "learning_rate": 1.1865246964444481e-06, "loss": 0.0836969256401062, "step": 29720 }, { "epoch": 0.7831348112050444, "grad_norm": 4.098480224609375, "learning_rate": 1.1837672167377635e-06, "loss": 0.09000782370567321, "step": 29730 }, { "epoch": 0.7833982268832163, "grad_norm": 5.726623058319092, "learning_rate": 1.1810125146499462e-06, "loss": 0.08525713682174682, "step": 29740 }, { "epoch": 0.7836616425613881, "grad_norm": 5.477299690246582, "learning_rate": 1.1782605921859864e-06, "loss": 0.08130136132240295, "step": 29750 }, { "epoch": 0.78392505823956, "grad_norm": 4.39647912979126, "learning_rate": 1.1755114513488509e-06, "loss": 0.08269641995429992, "step": 29760 }, { "epoch": 0.7841884739177319, "grad_norm": 5.739989280700684, "learning_rate": 1.1727650941394835e-06, "loss": 0.0927831470966339, "step": 29770 }, { "epoch": 0.7844518895959038, "grad_norm": 4.444987773895264, "learning_rate": 1.1700215225568e-06, "loss": 0.09732595086097717, "step": 29780 }, { "epoch": 0.7847153052740757, "grad_norm": 4.332214832305908, "learning_rate": 1.1672807385976936e-06, "loss": 0.0901439905166626, "step": 29790 }, { "epoch": 0.7849787209522476, "grad_norm": 3.292328357696533, "learning_rate": 1.1645427442570246e-06, "loss": 0.08471733331680298, "step": 29800 }, { "epoch": 0.7852421366304195, "grad_norm": 4.016160011291504, "learning_rate": 1.1618075415276187e-06, "loss": 0.09965349435806274, "step": 29810 }, { "epoch": 0.7855055523085914, "grad_norm": 5.627131938934326, "learning_rate": 1.1590751324002797e-06, "loss": 0.08750731945037842, "step": 29820 }, { "epoch": 0.7857689679867633, "grad_norm": 4.098034858703613, "learning_rate": 1.1563455188637707e-06, "loss": 0.09003034830093384, "step": 29830 }, { "epoch": 0.7860323836649352, "grad_norm": 5.401447296142578, "learning_rate": 1.1536187029048223e-06, "loss": 0.09022240042686462, "step": 29840 }, { "epoch": 0.7862957993431071, "grad_norm": 4.102752208709717, "learning_rate": 1.1508946865081294e-06, "loss": 0.092575204372406, "step": 29850 }, { "epoch": 0.786559215021279, "grad_norm": 4.489665985107422, "learning_rate": 1.1481734716563464e-06, "loss": 0.08393112421035767, "step": 29860 }, { "epoch": 0.7868226306994509, "grad_norm": 3.6956892013549805, "learning_rate": 1.145455060330094e-06, "loss": 0.09027407169342042, "step": 29870 }, { "epoch": 0.7870860463776228, "grad_norm": 5.5702667236328125, "learning_rate": 1.142739454507948e-06, "loss": 0.09350897073745727, "step": 29880 }, { "epoch": 0.7873494620557947, "grad_norm": 4.843353748321533, "learning_rate": 1.1400266561664437e-06, "loss": 0.10020835399627685, "step": 29890 }, { "epoch": 0.7876128777339666, "grad_norm": 4.1632490158081055, "learning_rate": 1.1373166672800718e-06, "loss": 0.08929897546768188, "step": 29900 }, { "epoch": 0.7878762934121385, "grad_norm": 4.584317207336426, "learning_rate": 1.134609489821279e-06, "loss": 0.08722149133682251, "step": 29910 }, { "epoch": 0.7881397090903104, "grad_norm": 5.937442302703857, "learning_rate": 1.1319051257604691e-06, "loss": 0.08825990557670593, "step": 29920 }, { "epoch": 0.7884031247684823, "grad_norm": 7.031275749206543, "learning_rate": 1.129203577065991e-06, "loss": 0.08152559995651246, "step": 29930 }, { "epoch": 0.7886665404466542, "grad_norm": 4.679234027862549, "learning_rate": 1.1265048457041483e-06, "loss": 0.09043717980384827, "step": 29940 }, { "epoch": 0.7889299561248261, "grad_norm": 5.020290851593018, "learning_rate": 1.1238089336391972e-06, "loss": 0.08677439093589782, "step": 29950 }, { "epoch": 0.789193371802998, "grad_norm": 5.997182369232178, "learning_rate": 1.1211158428333375e-06, "loss": 0.0885589599609375, "step": 29960 }, { "epoch": 0.7894567874811699, "grad_norm": 3.7963099479675293, "learning_rate": 1.1184255752467166e-06, "loss": 0.0796135425567627, "step": 29970 }, { "epoch": 0.7897202031593418, "grad_norm": 4.187689781188965, "learning_rate": 1.1157381328374272e-06, "loss": 0.07424429655075074, "step": 29980 }, { "epoch": 0.7899836188375137, "grad_norm": 3.970623731613159, "learning_rate": 1.1130535175615054e-06, "loss": 0.08754203319549561, "step": 29990 }, { "epoch": 0.7902470345156856, "grad_norm": 5.2499566078186035, "learning_rate": 1.1103717313729313e-06, "loss": 0.0802493691444397, "step": 30000 }, { "epoch": 0.7905104501938575, "grad_norm": 5.37473726272583, "learning_rate": 1.1076927762236256e-06, "loss": 0.09062831401824951, "step": 30010 }, { "epoch": 0.7907738658720294, "grad_norm": 7.074263572692871, "learning_rate": 1.1050166540634426e-06, "loss": 0.07366795539855957, "step": 30020 }, { "epoch": 0.7910372815502013, "grad_norm": 6.1802568435668945, "learning_rate": 1.1023433668401845e-06, "loss": 0.08597080707550049, "step": 30030 }, { "epoch": 0.7913006972283732, "grad_norm": 4.720902919769287, "learning_rate": 1.0996729164995818e-06, "loss": 0.09473084211349488, "step": 30040 }, { "epoch": 0.791564112906545, "grad_norm": 6.398552894592285, "learning_rate": 1.0970053049853069e-06, "loss": 0.09768956899642944, "step": 30050 }, { "epoch": 0.7918275285847169, "grad_norm": 5.2351274490356445, "learning_rate": 1.0943405342389602e-06, "loss": 0.08334207534790039, "step": 30060 }, { "epoch": 0.7920909442628888, "grad_norm": 4.035182952880859, "learning_rate": 1.091678606200075e-06, "loss": 0.08881734609603882, "step": 30070 }, { "epoch": 0.7923543599410607, "grad_norm": 6.274366855621338, "learning_rate": 1.0890195228061212e-06, "loss": 0.09039437770843506, "step": 30080 }, { "epoch": 0.7926177756192326, "grad_norm": 4.610469341278076, "learning_rate": 1.0863632859924926e-06, "loss": 0.08122246861457824, "step": 30090 }, { "epoch": 0.7928811912974045, "grad_norm": 3.907362461090088, "learning_rate": 1.0837098976925137e-06, "loss": 0.08431149125099183, "step": 30100 }, { "epoch": 0.7931446069755764, "grad_norm": 4.306833744049072, "learning_rate": 1.0810593598374335e-06, "loss": 0.09061402082443237, "step": 30110 }, { "epoch": 0.7934080226537483, "grad_norm": 3.610238790512085, "learning_rate": 1.0784116743564278e-06, "loss": 0.09327855110168456, "step": 30120 }, { "epoch": 0.7936714383319202, "grad_norm": 5.6648173332214355, "learning_rate": 1.0757668431765988e-06, "loss": 0.07722029685974122, "step": 30130 }, { "epoch": 0.7939348540100921, "grad_norm": 4.836004257202148, "learning_rate": 1.0731248682229678e-06, "loss": 0.08202507495880126, "step": 30140 }, { "epoch": 0.794198269688264, "grad_norm": 4.606919288635254, "learning_rate": 1.0704857514184752e-06, "loss": 0.07476601004600525, "step": 30150 }, { "epoch": 0.7944616853664359, "grad_norm": 4.3553266525268555, "learning_rate": 1.067849494683988e-06, "loss": 0.08585894703865052, "step": 30160 }, { "epoch": 0.7947251010446078, "grad_norm": 3.9905378818511963, "learning_rate": 1.0652160999382843e-06, "loss": 0.08364387154579163, "step": 30170 }, { "epoch": 0.7949885167227797, "grad_norm": 4.312774658203125, "learning_rate": 1.0625855690980668e-06, "loss": 0.09737979769706726, "step": 30180 }, { "epoch": 0.7952519324009516, "grad_norm": 5.024652481079102, "learning_rate": 1.059957904077945e-06, "loss": 0.09156490564346313, "step": 30190 }, { "epoch": 0.7955153480791235, "grad_norm": 5.033726692199707, "learning_rate": 1.0573331067904474e-06, "loss": 0.07555769085884094, "step": 30200 }, { "epoch": 0.7957787637572954, "grad_norm": 4.705256462097168, "learning_rate": 1.0547111791460173e-06, "loss": 0.07970902919769288, "step": 30210 }, { "epoch": 0.7960421794354673, "grad_norm": 5.146127700805664, "learning_rate": 1.0520921230530052e-06, "loss": 0.08489696979522705, "step": 30220 }, { "epoch": 0.7963055951136392, "grad_norm": 4.886209487915039, "learning_rate": 1.049475940417673e-06, "loss": 0.08838655352592469, "step": 30230 }, { "epoch": 0.7965690107918111, "grad_norm": 5.122250080108643, "learning_rate": 1.046862633144191e-06, "loss": 0.09198371171951295, "step": 30240 }, { "epoch": 0.796832426469983, "grad_norm": 5.28523588180542, "learning_rate": 1.044252203134637e-06, "loss": 0.08487998247146607, "step": 30250 }, { "epoch": 0.7970958421481549, "grad_norm": 5.321743488311768, "learning_rate": 1.041644652288996e-06, "loss": 0.09144835472106934, "step": 30260 }, { "epoch": 0.7973592578263268, "grad_norm": 5.345059871673584, "learning_rate": 1.0390399825051562e-06, "loss": 0.08643338680267335, "step": 30270 }, { "epoch": 0.7976226735044987, "grad_norm": 5.13613224029541, "learning_rate": 1.0364381956789048e-06, "loss": 0.09150579571723938, "step": 30280 }, { "epoch": 0.7978860891826706, "grad_norm": 2.424846649169922, "learning_rate": 1.0338392937039387e-06, "loss": 0.07724002003669739, "step": 30290 }, { "epoch": 0.7981495048608425, "grad_norm": 3.970581531524658, "learning_rate": 1.0312432784718495e-06, "loss": 0.0800978660583496, "step": 30300 }, { "epoch": 0.7984129205390144, "grad_norm": 4.0262131690979, "learning_rate": 1.0286501518721287e-06, "loss": 0.07910290956497193, "step": 30310 }, { "epoch": 0.7986763362171863, "grad_norm": 4.769623756408691, "learning_rate": 1.0260599157921657e-06, "loss": 0.09457944631576538, "step": 30320 }, { "epoch": 0.7989397518953582, "grad_norm": 5.876136779785156, "learning_rate": 1.0234725721172455e-06, "loss": 0.10108047723770142, "step": 30330 }, { "epoch": 0.79920316757353, "grad_norm": 4.881933689117432, "learning_rate": 1.0208881227305506e-06, "loss": 0.08550227284431458, "step": 30340 }, { "epoch": 0.799466583251702, "grad_norm": 6.347262382507324, "learning_rate": 1.0183065695131527e-06, "loss": 0.08984037041664124, "step": 30350 }, { "epoch": 0.7997299989298738, "grad_norm": 3.6137712001800537, "learning_rate": 1.0157279143440192e-06, "loss": 0.0834839940071106, "step": 30360 }, { "epoch": 0.7999934146080457, "grad_norm": 4.907365798950195, "learning_rate": 1.0131521591000055e-06, "loss": 0.08130674362182617, "step": 30370 }, { "epoch": 0.8002568302862176, "grad_norm": 5.030773162841797, "learning_rate": 1.010579305655856e-06, "loss": 0.08528135418891906, "step": 30380 }, { "epoch": 0.8005202459643895, "grad_norm": 4.955322742462158, "learning_rate": 1.008009355884208e-06, "loss": 0.08186321258544922, "step": 30390 }, { "epoch": 0.8007836616425614, "grad_norm": 5.852077960968018, "learning_rate": 1.005442311655581e-06, "loss": 0.08747317790985107, "step": 30400 }, { "epoch": 0.8010470773207333, "grad_norm": 4.709561347961426, "learning_rate": 1.0028781748383765e-06, "loss": 0.08652628064155579, "step": 30410 }, { "epoch": 0.8013104929989052, "grad_norm": 5.787837505340576, "learning_rate": 1.0003169472988889e-06, "loss": 0.08700047731399536, "step": 30420 }, { "epoch": 0.8015739086770771, "grad_norm": 5.023800373077393, "learning_rate": 9.977586309012872e-07, "loss": 0.09224957227706909, "step": 30430 }, { "epoch": 0.801837324355249, "grad_norm": 4.7506937980651855, "learning_rate": 9.952032275076258e-07, "loss": 0.08639053106307984, "step": 30440 }, { "epoch": 0.8021007400334209, "grad_norm": 5.297988414764404, "learning_rate": 9.926507389778367e-07, "loss": 0.09065714478492737, "step": 30450 }, { "epoch": 0.8023641557115928, "grad_norm": 4.267396450042725, "learning_rate": 9.901011671697303e-07, "loss": 0.09971692562103271, "step": 30460 }, { "epoch": 0.8026275713897647, "grad_norm": 3.0927155017852783, "learning_rate": 9.875545139389976e-07, "loss": 0.08269031047821045, "step": 30470 }, { "epoch": 0.8028909870679366, "grad_norm": 5.627140522003174, "learning_rate": 9.850107811392013e-07, "loss": 0.0879435420036316, "step": 30480 }, { "epoch": 0.8031544027461085, "grad_norm": 4.323553085327148, "learning_rate": 9.8246997062178e-07, "loss": 0.07244246006011963, "step": 30490 }, { "epoch": 0.8034178184242804, "grad_norm": 5.2452545166015625, "learning_rate": 9.799320842360448e-07, "loss": 0.08242865204811096, "step": 30500 }, { "epoch": 0.8036812341024522, "grad_norm": 3.7158753871917725, "learning_rate": 9.77397123829178e-07, "loss": 0.07617055177688599, "step": 30510 }, { "epoch": 0.8039446497806241, "grad_norm": 4.747710227966309, "learning_rate": 9.748650912462366e-07, "loss": 0.08153734207153321, "step": 30520 }, { "epoch": 0.804208065458796, "grad_norm": 5.212290287017822, "learning_rate": 9.723359883301397e-07, "loss": 0.08793048858642578, "step": 30530 }, { "epoch": 0.8044714811369679, "grad_norm": 4.2453813552856445, "learning_rate": 9.698098169216773e-07, "loss": 0.09090328216552734, "step": 30540 }, { "epoch": 0.8047348968151398, "grad_norm": 4.317033767700195, "learning_rate": 9.672865788595094e-07, "loss": 0.0830617070198059, "step": 30550 }, { "epoch": 0.8049983124933117, "grad_norm": 3.625471830368042, "learning_rate": 9.647662759801546e-07, "loss": 0.07807804942131043, "step": 30560 }, { "epoch": 0.8052617281714836, "grad_norm": 5.423619270324707, "learning_rate": 9.622489101179987e-07, "loss": 0.08043608069419861, "step": 30570 }, { "epoch": 0.8055251438496555, "grad_norm": 5.315919399261475, "learning_rate": 9.597344831052902e-07, "loss": 0.08476563096046448, "step": 30580 }, { "epoch": 0.8057885595278274, "grad_norm": 4.2643513679504395, "learning_rate": 9.572229967721352e-07, "loss": 0.07542253136634827, "step": 30590 }, { "epoch": 0.8060519752059992, "grad_norm": 4.487305641174316, "learning_rate": 9.547144529465047e-07, "loss": 0.07610756754875184, "step": 30600 }, { "epoch": 0.8063153908841711, "grad_norm": 4.880791187286377, "learning_rate": 9.522088534542251e-07, "loss": 0.09135344624519348, "step": 30610 }, { "epoch": 0.806578806562343, "grad_norm": 4.1522297859191895, "learning_rate": 9.497062001189755e-07, "loss": 0.08374677896499634, "step": 30620 }, { "epoch": 0.8068422222405149, "grad_norm": 4.992159366607666, "learning_rate": 9.472064947623e-07, "loss": 0.08126925230026245, "step": 30630 }, { "epoch": 0.8071056379186868, "grad_norm": 4.3784661293029785, "learning_rate": 9.447097392035881e-07, "loss": 0.07306690812110901, "step": 30640 }, { "epoch": 0.8073690535968587, "grad_norm": 5.361541748046875, "learning_rate": 9.422159352600907e-07, "loss": 0.074763023853302, "step": 30650 }, { "epoch": 0.8076324692750306, "grad_norm": 4.95919942855835, "learning_rate": 9.397250847469024e-07, "loss": 0.06576765179634095, "step": 30660 }, { "epoch": 0.8078958849532025, "grad_norm": 6.412187099456787, "learning_rate": 9.37237189476971e-07, "loss": 0.08671959042549134, "step": 30670 }, { "epoch": 0.8081593006313744, "grad_norm": 3.948826313018799, "learning_rate": 9.347522512610968e-07, "loss": 0.08621563315391541, "step": 30680 }, { "epoch": 0.8084227163095463, "grad_norm": 4.8524489402771, "learning_rate": 9.32270271907924e-07, "loss": 0.07663238048553467, "step": 30690 }, { "epoch": 0.8086861319877182, "grad_norm": 4.436318874359131, "learning_rate": 9.297912532239445e-07, "loss": 0.07220046520233155, "step": 30700 }, { "epoch": 0.8089495476658901, "grad_norm": 5.427560806274414, "learning_rate": 9.273151970134942e-07, "loss": 0.08664069175720215, "step": 30710 }, { "epoch": 0.809212963344062, "grad_norm": 5.563728332519531, "learning_rate": 9.248421050787527e-07, "loss": 0.09650423526763915, "step": 30720 }, { "epoch": 0.8094763790222339, "grad_norm": 5.015710353851318, "learning_rate": 9.223719792197455e-07, "loss": 0.0824009895324707, "step": 30730 }, { "epoch": 0.8097397947004058, "grad_norm": 4.194884777069092, "learning_rate": 9.19904821234337e-07, "loss": 0.08657122254371644, "step": 30740 }, { "epoch": 0.8100032103785777, "grad_norm": 4.767070770263672, "learning_rate": 9.174406329182279e-07, "loss": 0.08874998688697815, "step": 30750 }, { "epoch": 0.8102666260567496, "grad_norm": 4.444025993347168, "learning_rate": 9.149794160649633e-07, "loss": 0.08570545315742492, "step": 30760 }, { "epoch": 0.8105300417349215, "grad_norm": 4.731954574584961, "learning_rate": 9.125211724659217e-07, "loss": 0.0887310802936554, "step": 30770 }, { "epoch": 0.8107934574130934, "grad_norm": 3.6990468502044678, "learning_rate": 9.100659039103221e-07, "loss": 0.07682586908340454, "step": 30780 }, { "epoch": 0.8110568730912653, "grad_norm": 2.8541676998138428, "learning_rate": 9.076136121852108e-07, "loss": 0.08219149708747864, "step": 30790 }, { "epoch": 0.8113202887694372, "grad_norm": 5.0188164710998535, "learning_rate": 9.051642990754716e-07, "loss": 0.07174860239028931, "step": 30800 }, { "epoch": 0.8115837044476091, "grad_norm": 5.122946262359619, "learning_rate": 9.027179663638236e-07, "loss": 0.08063205480575561, "step": 30810 }, { "epoch": 0.811847120125781, "grad_norm": 4.911550998687744, "learning_rate": 9.002746158308112e-07, "loss": 0.0708670437335968, "step": 30820 }, { "epoch": 0.8121105358039529, "grad_norm": 4.663455963134766, "learning_rate": 8.978342492548109e-07, "loss": 0.07228104472160339, "step": 30830 }, { "epoch": 0.8123739514821248, "grad_norm": 6.168410778045654, "learning_rate": 8.953968684120263e-07, "loss": 0.08558881878852845, "step": 30840 }, { "epoch": 0.8126373671602967, "grad_norm": 5.551052570343018, "learning_rate": 8.929624750764871e-07, "loss": 0.08175166845321655, "step": 30850 }, { "epoch": 0.8129007828384686, "grad_norm": 4.263894557952881, "learning_rate": 8.905310710200527e-07, "loss": 0.0779283344745636, "step": 30860 }, { "epoch": 0.8131641985166405, "grad_norm": 4.801138877868652, "learning_rate": 8.88102658012403e-07, "loss": 0.09366016387939453, "step": 30870 }, { "epoch": 0.8134276141948124, "grad_norm": 3.005610942840576, "learning_rate": 8.856772378210393e-07, "loss": 0.06275113821029663, "step": 30880 }, { "epoch": 0.8136910298729843, "grad_norm": 5.094855308532715, "learning_rate": 8.8325481221129e-07, "loss": 0.09273748397827149, "step": 30890 }, { "epoch": 0.8139544455511561, "grad_norm": 5.889703750610352, "learning_rate": 8.808353829462995e-07, "loss": 0.08036493062973023, "step": 30900 }, { "epoch": 0.814217861229328, "grad_norm": 3.8354692459106445, "learning_rate": 8.784189517870334e-07, "loss": 0.072219979763031, "step": 30910 }, { "epoch": 0.8144812769074999, "grad_norm": 4.926088333129883, "learning_rate": 8.760055204922747e-07, "loss": 0.08127403259277344, "step": 30920 }, { "epoch": 0.8147446925856718, "grad_norm": 4.2780442237854, "learning_rate": 8.735950908186208e-07, "loss": 0.08653423190116882, "step": 30930 }, { "epoch": 0.8150081082638437, "grad_norm": 5.050280570983887, "learning_rate": 8.711876645204892e-07, "loss": 0.09068005681037902, "step": 30940 }, { "epoch": 0.8152715239420156, "grad_norm": 3.599581718444824, "learning_rate": 8.687832433501076e-07, "loss": 0.07536685466766357, "step": 30950 }, { "epoch": 0.8155349396201875, "grad_norm": 5.851410388946533, "learning_rate": 8.663818290575172e-07, "loss": 0.08020581007003784, "step": 30960 }, { "epoch": 0.8157983552983594, "grad_norm": 4.317624568939209, "learning_rate": 8.639834233905714e-07, "loss": 0.07618138790130616, "step": 30970 }, { "epoch": 0.8160617709765313, "grad_norm": 4.512801647186279, "learning_rate": 8.615880280949312e-07, "loss": 0.07984042167663574, "step": 30980 }, { "epoch": 0.8163251866547032, "grad_norm": 5.043650150299072, "learning_rate": 8.59195644914072e-07, "loss": 0.08380064964294434, "step": 30990 }, { "epoch": 0.8165886023328751, "grad_norm": 4.99387788772583, "learning_rate": 8.568062755892731e-07, "loss": 0.07117041945457458, "step": 31000 }, { "epoch": 0.816852018011047, "grad_norm": 4.312022686004639, "learning_rate": 8.544199218596172e-07, "loss": 0.08153777122497559, "step": 31010 }, { "epoch": 0.8171154336892189, "grad_norm": 5.2878217697143555, "learning_rate": 8.520365854619983e-07, "loss": 0.08546740412712098, "step": 31020 }, { "epoch": 0.8173788493673908, "grad_norm": 4.164046287536621, "learning_rate": 8.496562681311116e-07, "loss": 0.08181557655334473, "step": 31030 }, { "epoch": 0.8176422650455627, "grad_norm": 5.6565260887146, "learning_rate": 8.472789715994533e-07, "loss": 0.10434579849243164, "step": 31040 }, { "epoch": 0.8179056807237346, "grad_norm": 5.646178722381592, "learning_rate": 8.449046975973236e-07, "loss": 0.0742089331150055, "step": 31050 }, { "epoch": 0.8181690964019065, "grad_norm": 6.043826580047607, "learning_rate": 8.425334478528196e-07, "loss": 0.08550845980644226, "step": 31060 }, { "epoch": 0.8184325120800784, "grad_norm": 5.487966537475586, "learning_rate": 8.401652240918417e-07, "loss": 0.086655592918396, "step": 31070 }, { "epoch": 0.8186959277582503, "grad_norm": 4.147116661071777, "learning_rate": 8.378000280380843e-07, "loss": 0.07881866693496704, "step": 31080 }, { "epoch": 0.8189593434364222, "grad_norm": 4.939593315124512, "learning_rate": 8.354378614130393e-07, "loss": 0.07723884582519532, "step": 31090 }, { "epoch": 0.8192227591145941, "grad_norm": 4.716099262237549, "learning_rate": 8.330787259359929e-07, "loss": 0.06925337314605713, "step": 31100 }, { "epoch": 0.819486174792766, "grad_norm": 7.121088027954102, "learning_rate": 8.307226233240251e-07, "loss": 0.08551349639892578, "step": 31110 }, { "epoch": 0.8197495904709379, "grad_norm": 3.502572774887085, "learning_rate": 8.283695552920124e-07, "loss": 0.07812363505363465, "step": 31120 }, { "epoch": 0.8200130061491098, "grad_norm": 3.555320978164673, "learning_rate": 8.260195235526153e-07, "loss": 0.07277803421020508, "step": 31130 }, { "epoch": 0.8202764218272817, "grad_norm": 4.661558628082275, "learning_rate": 8.23672529816289e-07, "loss": 0.08041008710861205, "step": 31140 }, { "epoch": 0.8205398375054536, "grad_norm": 4.052892684936523, "learning_rate": 8.213285757912781e-07, "loss": 0.09020032882690429, "step": 31150 }, { "epoch": 0.8208032531836255, "grad_norm": 5.344539165496826, "learning_rate": 8.189876631836136e-07, "loss": 0.07760941982269287, "step": 31160 }, { "epoch": 0.8210666688617974, "grad_norm": 5.70853328704834, "learning_rate": 8.166497936971119e-07, "loss": 0.07286595106124878, "step": 31170 }, { "epoch": 0.8213300845399693, "grad_norm": 5.490346431732178, "learning_rate": 8.143149690333752e-07, "loss": 0.07146261930465699, "step": 31180 }, { "epoch": 0.8215935002181411, "grad_norm": 9.388407707214355, "learning_rate": 8.119831908917891e-07, "loss": 0.08253732919692994, "step": 31190 }, { "epoch": 0.821856915896313, "grad_norm": 5.372498989105225, "learning_rate": 8.096544609695239e-07, "loss": 0.07520736455917358, "step": 31200 }, { "epoch": 0.8221203315744849, "grad_norm": 5.8255934715271, "learning_rate": 8.073287809615299e-07, "loss": 0.07812113165855408, "step": 31210 }, { "epoch": 0.8223837472526568, "grad_norm": 5.768524646759033, "learning_rate": 8.05006152560534e-07, "loss": 0.08083974719047546, "step": 31220 }, { "epoch": 0.8226471629308287, "grad_norm": 3.827751636505127, "learning_rate": 8.026865774570491e-07, "loss": 0.08309556245803833, "step": 31230 }, { "epoch": 0.8229105786090006, "grad_norm": 4.858229160308838, "learning_rate": 8.003700573393597e-07, "loss": 0.07134859561920166, "step": 31240 }, { "epoch": 0.8231739942871725, "grad_norm": 4.963382720947266, "learning_rate": 7.98056593893532e-07, "loss": 0.09155150651931762, "step": 31250 }, { "epoch": 0.8234374099653444, "grad_norm": 3.9895472526550293, "learning_rate": 7.957461888034012e-07, "loss": 0.0750070035457611, "step": 31260 }, { "epoch": 0.8237008256435163, "grad_norm": 5.723723888397217, "learning_rate": 7.934388437505802e-07, "loss": 0.0869828224182129, "step": 31270 }, { "epoch": 0.8239642413216882, "grad_norm": 5.207704544067383, "learning_rate": 7.91134560414456e-07, "loss": 0.08393595218658448, "step": 31280 }, { "epoch": 0.8242276569998601, "grad_norm": 5.1081743240356445, "learning_rate": 7.888333404721849e-07, "loss": 0.07347937822341918, "step": 31290 }, { "epoch": 0.824491072678032, "grad_norm": 5.345296382904053, "learning_rate": 7.865351855986942e-07, "loss": 0.08198739290237426, "step": 31300 }, { "epoch": 0.8247544883562038, "grad_norm": 4.887561321258545, "learning_rate": 7.842400974666792e-07, "loss": 0.07999675273895264, "step": 31310 }, { "epoch": 0.8250179040343757, "grad_norm": 3.8669984340667725, "learning_rate": 7.819480777466032e-07, "loss": 0.07727011442184448, "step": 31320 }, { "epoch": 0.8252813197125476, "grad_norm": 4.7489495277404785, "learning_rate": 7.796591281067001e-07, "loss": 0.07358676791191102, "step": 31330 }, { "epoch": 0.8255447353907195, "grad_norm": 5.9463887214660645, "learning_rate": 7.773732502129655e-07, "loss": 0.07620112895965576, "step": 31340 }, { "epoch": 0.8258081510688914, "grad_norm": 6.446361541748047, "learning_rate": 7.750904457291569e-07, "loss": 0.08016021251678467, "step": 31350 }, { "epoch": 0.8260715667470633, "grad_norm": 5.0554327964782715, "learning_rate": 7.728107163168009e-07, "loss": 0.08025645017623902, "step": 31360 }, { "epoch": 0.8263349824252352, "grad_norm": 5.538835525512695, "learning_rate": 7.705340636351821e-07, "loss": 0.07411657571792603, "step": 31370 }, { "epoch": 0.8265983981034071, "grad_norm": 3.4906351566314697, "learning_rate": 7.682604893413465e-07, "loss": 0.06869803667068482, "step": 31380 }, { "epoch": 0.826861813781579, "grad_norm": 6.193586826324463, "learning_rate": 7.659899950900995e-07, "loss": 0.07098222374916077, "step": 31390 }, { "epoch": 0.8271252294597509, "grad_norm": 5.026836395263672, "learning_rate": 7.637225825340033e-07, "loss": 0.07484133243560791, "step": 31400 }, { "epoch": 0.8273886451379228, "grad_norm": 6.694081783294678, "learning_rate": 7.614582533233816e-07, "loss": 0.08465704917907715, "step": 31410 }, { "epoch": 0.8276520608160947, "grad_norm": 3.8790853023529053, "learning_rate": 7.591970091063094e-07, "loss": 0.07398212552070618, "step": 31420 }, { "epoch": 0.8279154764942666, "grad_norm": 6.61421012878418, "learning_rate": 7.569388515286175e-07, "loss": 0.10040605068206787, "step": 31430 }, { "epoch": 0.8281788921724385, "grad_norm": 5.0623064041137695, "learning_rate": 7.54683782233891e-07, "loss": 0.07903608083724975, "step": 31440 }, { "epoch": 0.8284423078506103, "grad_norm": 5.354896545410156, "learning_rate": 7.524318028634647e-07, "loss": 0.0805105447769165, "step": 31450 }, { "epoch": 0.8287057235287822, "grad_norm": 4.423668384552002, "learning_rate": 7.5018291505643e-07, "loss": 0.0774347424507141, "step": 31460 }, { "epoch": 0.8289691392069541, "grad_norm": 4.0350494384765625, "learning_rate": 7.479371204496227e-07, "loss": 0.06736937761306763, "step": 31470 }, { "epoch": 0.829232554885126, "grad_norm": 4.6804094314575195, "learning_rate": 7.456944206776268e-07, "loss": 0.07858949899673462, "step": 31480 }, { "epoch": 0.8294959705632979, "grad_norm": 4.820195198059082, "learning_rate": 7.434548173727785e-07, "loss": 0.07446542978286744, "step": 31490 }, { "epoch": 0.8297593862414698, "grad_norm": 5.780622959136963, "learning_rate": 7.412183121651578e-07, "loss": 0.07875934839248658, "step": 31500 }, { "epoch": 0.8300228019196417, "grad_norm": 5.139027118682861, "learning_rate": 7.389849066825883e-07, "loss": 0.06500355005264283, "step": 31510 }, { "epoch": 0.8302862175978136, "grad_norm": 4.520771503448486, "learning_rate": 7.3675460255064e-07, "loss": 0.06863974928855895, "step": 31520 }, { "epoch": 0.8305496332759855, "grad_norm": 4.703767776489258, "learning_rate": 7.34527401392623e-07, "loss": 0.07418888807296753, "step": 31530 }, { "epoch": 0.8308130489541574, "grad_norm": 5.052894592285156, "learning_rate": 7.323033048295919e-07, "loss": 0.08645644187927246, "step": 31540 }, { "epoch": 0.8310764646323293, "grad_norm": 5.090928077697754, "learning_rate": 7.300823144803404e-07, "loss": 0.07714306116104126, "step": 31550 }, { "epoch": 0.8313398803105012, "grad_norm": 3.9545950889587402, "learning_rate": 7.278644319613997e-07, "loss": 0.07133831977844238, "step": 31560 }, { "epoch": 0.8316032959886731, "grad_norm": 4.703083515167236, "learning_rate": 7.256496588870415e-07, "loss": 0.09612720012664795, "step": 31570 }, { "epoch": 0.831866711666845, "grad_norm": 5.311788558959961, "learning_rate": 7.234379968692712e-07, "loss": 0.08146046400070191, "step": 31580 }, { "epoch": 0.8321301273450169, "grad_norm": 6.411531448364258, "learning_rate": 7.212294475178344e-07, "loss": 0.08017048835754395, "step": 31590 }, { "epoch": 0.8323935430231888, "grad_norm": 6.088335990905762, "learning_rate": 7.190240124402093e-07, "loss": 0.08957271575927735, "step": 31600 }, { "epoch": 0.8326569587013607, "grad_norm": 4.299225330352783, "learning_rate": 7.168216932416016e-07, "loss": 0.08574577569961547, "step": 31610 }, { "epoch": 0.8329203743795326, "grad_norm": 5.839494705200195, "learning_rate": 7.146224915249594e-07, "loss": 0.07960354089736939, "step": 31620 }, { "epoch": 0.8331837900577045, "grad_norm": 5.283490180969238, "learning_rate": 7.124264088909549e-07, "loss": 0.0858063519001007, "step": 31630 }, { "epoch": 0.8334472057358764, "grad_norm": 4.314868450164795, "learning_rate": 7.102334469379918e-07, "loss": 0.07227194309234619, "step": 31640 }, { "epoch": 0.8337106214140483, "grad_norm": 4.6340718269348145, "learning_rate": 7.080436072622015e-07, "loss": 0.08073803186416625, "step": 31650 }, { "epoch": 0.8339740370922202, "grad_norm": 5.5768513679504395, "learning_rate": 7.058568914574443e-07, "loss": 0.07416656613349915, "step": 31660 }, { "epoch": 0.8342374527703921, "grad_norm": 4.102316379547119, "learning_rate": 7.036733011153074e-07, "loss": 0.06594237089157104, "step": 31670 }, { "epoch": 0.834500868448564, "grad_norm": 4.601358890533447, "learning_rate": 7.014928378251013e-07, "loss": 0.08588827848434448, "step": 31680 }, { "epoch": 0.8347642841267359, "grad_norm": 4.68295955657959, "learning_rate": 6.993155031738607e-07, "loss": 0.06972827315330506, "step": 31690 }, { "epoch": 0.8350276998049078, "grad_norm": 5.631277084350586, "learning_rate": 6.971412987463444e-07, "loss": 0.0882671058177948, "step": 31700 }, { "epoch": 0.8352911154830797, "grad_norm": 6.173096179962158, "learning_rate": 6.949702261250313e-07, "loss": 0.09268280863761902, "step": 31710 }, { "epoch": 0.8355545311612516, "grad_norm": 5.556522846221924, "learning_rate": 6.92802286890124e-07, "loss": 0.07242724895477295, "step": 31720 }, { "epoch": 0.8358179468394235, "grad_norm": 5.238094806671143, "learning_rate": 6.9063748261954e-07, "loss": 0.08323575854301453, "step": 31730 }, { "epoch": 0.8360813625175954, "grad_norm": 4.622897624969482, "learning_rate": 6.884758148889165e-07, "loss": 0.07706230878829956, "step": 31740 }, { "epoch": 0.8363447781957672, "grad_norm": 4.781893253326416, "learning_rate": 6.86317285271611e-07, "loss": 0.07353424429893493, "step": 31750 }, { "epoch": 0.8366081938739391, "grad_norm": 4.41598653793335, "learning_rate": 6.841618953386936e-07, "loss": 0.08167965412139892, "step": 31760 }, { "epoch": 0.836871609552111, "grad_norm": 4.791264533996582, "learning_rate": 6.820096466589499e-07, "loss": 0.0767871081829071, "step": 31770 }, { "epoch": 0.8371350252302829, "grad_norm": 3.6721038818359375, "learning_rate": 6.79860540798879e-07, "loss": 0.06541986465454101, "step": 31780 }, { "epoch": 0.8373984409084548, "grad_norm": 6.003962993621826, "learning_rate": 6.77714579322692e-07, "loss": 0.08473332524299622, "step": 31790 }, { "epoch": 0.8376618565866267, "grad_norm": 4.888148784637451, "learning_rate": 6.755717637923148e-07, "loss": 0.08621639013290405, "step": 31800 }, { "epoch": 0.8379252722647986, "grad_norm": 5.129510879516602, "learning_rate": 6.734320957673807e-07, "loss": 0.09260346293449402, "step": 31810 }, { "epoch": 0.8381886879429705, "grad_norm": 5.025836944580078, "learning_rate": 6.712955768052287e-07, "loss": 0.08261330127716064, "step": 31820 }, { "epoch": 0.8384521036211424, "grad_norm": 5.926937103271484, "learning_rate": 6.691622084609134e-07, "loss": 0.0968386948108673, "step": 31830 }, { "epoch": 0.8387155192993143, "grad_norm": 5.763984680175781, "learning_rate": 6.670319922871898e-07, "loss": 0.07448248267173767, "step": 31840 }, { "epoch": 0.8389789349774862, "grad_norm": 4.5573201179504395, "learning_rate": 6.64904929834525e-07, "loss": 0.07614850997924805, "step": 31850 }, { "epoch": 0.8392423506556581, "grad_norm": 3.1595680713653564, "learning_rate": 6.627810226510828e-07, "loss": 0.07219552993774414, "step": 31860 }, { "epoch": 0.83950576633383, "grad_norm": 2.7542195320129395, "learning_rate": 6.606602722827344e-07, "loss": 0.06348174810409546, "step": 31870 }, { "epoch": 0.8397691820120019, "grad_norm": 3.300661087036133, "learning_rate": 6.585426802730566e-07, "loss": 0.07885074615478516, "step": 31880 }, { "epoch": 0.8400325976901738, "grad_norm": 4.681591510772705, "learning_rate": 6.564282481633222e-07, "loss": 0.06641957759857178, "step": 31890 }, { "epoch": 0.8402960133683457, "grad_norm": 5.787496566772461, "learning_rate": 6.543169774925062e-07, "loss": 0.07473848462104797, "step": 31900 }, { "epoch": 0.8405594290465176, "grad_norm": 5.650522232055664, "learning_rate": 6.522088697972829e-07, "loss": 0.08234779834747315, "step": 31910 }, { "epoch": 0.8408228447246895, "grad_norm": 6.966228008270264, "learning_rate": 6.501039266120224e-07, "loss": 0.0826346755027771, "step": 31920 }, { "epoch": 0.8410862604028614, "grad_norm": 3.354759931564331, "learning_rate": 6.48002149468796e-07, "loss": 0.0682068943977356, "step": 31930 }, { "epoch": 0.8413496760810333, "grad_norm": 5.411659240722656, "learning_rate": 6.459035398973673e-07, "loss": 0.07787307500839233, "step": 31940 }, { "epoch": 0.8416130917592052, "grad_norm": 5.961213111877441, "learning_rate": 6.438080994251921e-07, "loss": 0.07653105854988099, "step": 31950 }, { "epoch": 0.8418765074373771, "grad_norm": 6.3599371910095215, "learning_rate": 6.417158295774251e-07, "loss": 0.0751143217086792, "step": 31960 }, { "epoch": 0.842139923115549, "grad_norm": 5.059743404388428, "learning_rate": 6.396267318769095e-07, "loss": 0.0751610517501831, "step": 31970 }, { "epoch": 0.8424033387937209, "grad_norm": 4.777114391326904, "learning_rate": 6.375408078441814e-07, "loss": 0.08043112754821777, "step": 31980 }, { "epoch": 0.8426667544718928, "grad_norm": 4.792617321014404, "learning_rate": 6.354580589974651e-07, "loss": 0.08715733289718627, "step": 31990 }, { "epoch": 0.8429301701500647, "grad_norm": 5.798806190490723, "learning_rate": 6.333784868526755e-07, "loss": 0.07921286225318909, "step": 32000 }, { "epoch": 0.8431935858282366, "grad_norm": 4.565712928771973, "learning_rate": 6.313020929234159e-07, "loss": 0.07192705273628235, "step": 32010 }, { "epoch": 0.8434570015064085, "grad_norm": 3.8303520679473877, "learning_rate": 6.292288787209749e-07, "loss": 0.07702075242996216, "step": 32020 }, { "epoch": 0.8437204171845804, "grad_norm": 6.3231282234191895, "learning_rate": 6.271588457543271e-07, "loss": 0.07882853746414184, "step": 32030 }, { "epoch": 0.8439838328627522, "grad_norm": 4.333111763000488, "learning_rate": 6.250919955301316e-07, "loss": 0.07223476767539978, "step": 32040 }, { "epoch": 0.8442472485409241, "grad_norm": 4.918465614318848, "learning_rate": 6.2302832955273e-07, "loss": 0.0726864755153656, "step": 32050 }, { "epoch": 0.844510664219096, "grad_norm": 5.892886161804199, "learning_rate": 6.2096784932415e-07, "loss": 0.07572718262672425, "step": 32060 }, { "epoch": 0.8447740798972679, "grad_norm": 3.751098871231079, "learning_rate": 6.189105563440978e-07, "loss": 0.07154562473297119, "step": 32070 }, { "epoch": 0.8450374955754398, "grad_norm": 8.482404708862305, "learning_rate": 6.168564521099573e-07, "loss": 0.07150105237960816, "step": 32080 }, { "epoch": 0.8453009112536117, "grad_norm": 6.0936174392700195, "learning_rate": 6.148055381167966e-07, "loss": 0.08131300210952759, "step": 32090 }, { "epoch": 0.8455643269317835, "grad_norm": 5.31117057800293, "learning_rate": 6.127578158573582e-07, "loss": 0.09315271377563476, "step": 32100 }, { "epoch": 0.8458277426099554, "grad_norm": 3.622591972351074, "learning_rate": 6.107132868220638e-07, "loss": 0.07032872438430786, "step": 32110 }, { "epoch": 0.8460911582881273, "grad_norm": 5.1842780113220215, "learning_rate": 6.086719524990087e-07, "loss": 0.08144761919975281, "step": 32120 }, { "epoch": 0.8463545739662992, "grad_norm": 5.201797962188721, "learning_rate": 6.066338143739637e-07, "loss": 0.07280389666557312, "step": 32130 }, { "epoch": 0.8466179896444711, "grad_norm": 4.965435981750488, "learning_rate": 6.04598873930376e-07, "loss": 0.07931302785873413, "step": 32140 }, { "epoch": 0.846881405322643, "grad_norm": 4.545825958251953, "learning_rate": 6.025671326493609e-07, "loss": 0.06940202116966247, "step": 32150 }, { "epoch": 0.8471448210008149, "grad_norm": 4.408278942108154, "learning_rate": 6.005385920097084e-07, "loss": 0.06976890563964844, "step": 32160 }, { "epoch": 0.8474082366789868, "grad_norm": 4.233046531677246, "learning_rate": 5.985132534878774e-07, "loss": 0.06413868665695191, "step": 32170 }, { "epoch": 0.8476716523571587, "grad_norm": 4.691256046295166, "learning_rate": 5.964911185579958e-07, "loss": 0.084266197681427, "step": 32180 }, { "epoch": 0.8479350680353306, "grad_norm": 5.393982887268066, "learning_rate": 5.944721886918625e-07, "loss": 0.07752063274383544, "step": 32190 }, { "epoch": 0.8481984837135025, "grad_norm": 4.516205310821533, "learning_rate": 5.924564653589421e-07, "loss": 0.07107915878295898, "step": 32200 }, { "epoch": 0.8484618993916744, "grad_norm": 4.561002731323242, "learning_rate": 5.90443950026362e-07, "loss": 0.08083338737487793, "step": 32210 }, { "epoch": 0.8487253150698463, "grad_norm": 5.940286636352539, "learning_rate": 5.8843464415892e-07, "loss": 0.07326593399047851, "step": 32220 }, { "epoch": 0.8489887307480182, "grad_norm": 4.810128211975098, "learning_rate": 5.864285492190752e-07, "loss": 0.06976037621498107, "step": 32230 }, { "epoch": 0.8492521464261901, "grad_norm": 4.330216884613037, "learning_rate": 5.844256666669501e-07, "loss": 0.06620849967002869, "step": 32240 }, { "epoch": 0.849515562104362, "grad_norm": 7.214298725128174, "learning_rate": 5.824259979603292e-07, "loss": 0.0847528338432312, "step": 32250 }, { "epoch": 0.8497789777825339, "grad_norm": 4.062180995941162, "learning_rate": 5.804295445546559e-07, "loss": 0.07764975428581238, "step": 32260 }, { "epoch": 0.8500423934607058, "grad_norm": 3.622363805770874, "learning_rate": 5.784363079030375e-07, "loss": 0.07621608972549439, "step": 32270 }, { "epoch": 0.8503058091388777, "grad_norm": 5.349930286407471, "learning_rate": 5.764462894562367e-07, "loss": 0.07867676019668579, "step": 32280 }, { "epoch": 0.8505692248170496, "grad_norm": 5.04694128036499, "learning_rate": 5.744594906626749e-07, "loss": 0.07382283210754395, "step": 32290 }, { "epoch": 0.8508326404952214, "grad_norm": 3.7472333908081055, "learning_rate": 5.724759129684298e-07, "loss": 0.06251742839813232, "step": 32300 }, { "epoch": 0.8510960561733933, "grad_norm": 5.048600196838379, "learning_rate": 5.704955578172339e-07, "loss": 0.08042585253715515, "step": 32310 }, { "epoch": 0.8513594718515652, "grad_norm": 5.233602046966553, "learning_rate": 5.685184266504779e-07, "loss": 0.06364687085151673, "step": 32320 }, { "epoch": 0.8516228875297371, "grad_norm": 4.903939247131348, "learning_rate": 5.665445209072007e-07, "loss": 0.07347503304481506, "step": 32330 }, { "epoch": 0.851886303207909, "grad_norm": 5.645471096038818, "learning_rate": 5.645738420240954e-07, "loss": 0.07106645107269287, "step": 32340 }, { "epoch": 0.8521497188860809, "grad_norm": 4.469570636749268, "learning_rate": 5.626063914355095e-07, "loss": 0.08392513990402221, "step": 32350 }, { "epoch": 0.8524131345642528, "grad_norm": 5.436281681060791, "learning_rate": 5.606421705734378e-07, "loss": 0.0792550265789032, "step": 32360 }, { "epoch": 0.8526765502424247, "grad_norm": 4.923274993896484, "learning_rate": 5.586811808675236e-07, "loss": 0.08124710321426391, "step": 32370 }, { "epoch": 0.8529399659205966, "grad_norm": 5.786137580871582, "learning_rate": 5.567234237450614e-07, "loss": 0.07915770411491393, "step": 32380 }, { "epoch": 0.8532033815987685, "grad_norm": 5.088027477264404, "learning_rate": 5.547689006309898e-07, "loss": 0.06427665948867797, "step": 32390 }, { "epoch": 0.8534667972769404, "grad_norm": 5.910856246948242, "learning_rate": 5.528176129478962e-07, "loss": 0.06772054433822632, "step": 32400 }, { "epoch": 0.8537302129551123, "grad_norm": 3.809624671936035, "learning_rate": 5.508695621160131e-07, "loss": 0.07939841747283935, "step": 32410 }, { "epoch": 0.8539936286332842, "grad_norm": 4.931519031524658, "learning_rate": 5.489247495532124e-07, "loss": 0.08447555303573609, "step": 32420 }, { "epoch": 0.8542570443114561, "grad_norm": 4.341862201690674, "learning_rate": 5.469831766750156e-07, "loss": 0.07273249626159668, "step": 32430 }, { "epoch": 0.854520459989628, "grad_norm": 5.4490838050842285, "learning_rate": 5.450448448945817e-07, "loss": 0.07474428415298462, "step": 32440 }, { "epoch": 0.8547838756677999, "grad_norm": 4.387855529785156, "learning_rate": 5.431097556227144e-07, "loss": 0.08519123792648316, "step": 32450 }, { "epoch": 0.8550472913459718, "grad_norm": 5.7125563621521, "learning_rate": 5.411779102678527e-07, "loss": 0.0674207329750061, "step": 32460 }, { "epoch": 0.8553107070241437, "grad_norm": 3.584217071533203, "learning_rate": 5.392493102360769e-07, "loss": 0.0738269567489624, "step": 32470 }, { "epoch": 0.8555741227023156, "grad_norm": 5.506666660308838, "learning_rate": 5.373239569311067e-07, "loss": 0.08226313591003417, "step": 32480 }, { "epoch": 0.8558375383804875, "grad_norm": 5.444710731506348, "learning_rate": 5.354018517542975e-07, "loss": 0.07743103504180908, "step": 32490 }, { "epoch": 0.8561009540586594, "grad_norm": 5.985889434814453, "learning_rate": 5.334829961046395e-07, "loss": 0.08753328323364258, "step": 32500 }, { "epoch": 0.8563643697368313, "grad_norm": 4.980691909790039, "learning_rate": 5.315673913787589e-07, "loss": 0.07154960036277772, "step": 32510 }, { "epoch": 0.8566277854150032, "grad_norm": 5.396378517150879, "learning_rate": 5.296550389709143e-07, "loss": 0.07108542919158936, "step": 32520 }, { "epoch": 0.8568912010931751, "grad_norm": 4.71735954284668, "learning_rate": 5.277459402730007e-07, "loss": 0.0775387167930603, "step": 32530 }, { "epoch": 0.857154616771347, "grad_norm": 5.669427394866943, "learning_rate": 5.25840096674542e-07, "loss": 0.08209952116012573, "step": 32540 }, { "epoch": 0.8574180324495189, "grad_norm": 4.821783065795898, "learning_rate": 5.239375095626909e-07, "loss": 0.0718672513961792, "step": 32550 }, { "epoch": 0.8576814481276908, "grad_norm": 6.007558345794678, "learning_rate": 5.220381803222347e-07, "loss": 0.07812989354133607, "step": 32560 }, { "epoch": 0.8579448638058627, "grad_norm": 6.419043064117432, "learning_rate": 5.201421103355864e-07, "loss": 0.06519050598144531, "step": 32570 }, { "epoch": 0.8582082794840346, "grad_norm": 3.6989283561706543, "learning_rate": 5.182493009827883e-07, "loss": 0.06834309101104737, "step": 32580 }, { "epoch": 0.8584716951622064, "grad_norm": 6.59525728225708, "learning_rate": 5.163597536415077e-07, "loss": 0.07613659501075745, "step": 32590 }, { "epoch": 0.8587351108403783, "grad_norm": 5.783602714538574, "learning_rate": 5.14473469687038e-07, "loss": 0.08812054991722107, "step": 32600 }, { "epoch": 0.8589985265185502, "grad_norm": 4.488767623901367, "learning_rate": 5.125904504923002e-07, "loss": 0.06454411149024963, "step": 32610 }, { "epoch": 0.8592619421967221, "grad_norm": 5.941965579986572, "learning_rate": 5.107106974278359e-07, "loss": 0.07363466024398804, "step": 32620 }, { "epoch": 0.859525357874894, "grad_norm": 4.697628021240234, "learning_rate": 5.088342118618106e-07, "loss": 0.07838757038116455, "step": 32630 }, { "epoch": 0.8597887735530659, "grad_norm": 5.249103546142578, "learning_rate": 5.069609951600113e-07, "loss": 0.08180283308029175, "step": 32640 }, { "epoch": 0.8600521892312378, "grad_norm": 4.489104747772217, "learning_rate": 5.050910486858446e-07, "loss": 0.07609991431236267, "step": 32650 }, { "epoch": 0.8603156049094097, "grad_norm": 6.146313190460205, "learning_rate": 5.032243738003406e-07, "loss": 0.07747346758842469, "step": 32660 }, { "epoch": 0.8605790205875816, "grad_norm": 6.6306047439575195, "learning_rate": 5.013609718621454e-07, "loss": 0.07741038799285889, "step": 32670 }, { "epoch": 0.8608424362657535, "grad_norm": 6.247570514678955, "learning_rate": 4.995008442275212e-07, "loss": 0.07518025636672973, "step": 32680 }, { "epoch": 0.8611058519439254, "grad_norm": 5.691739082336426, "learning_rate": 4.976439922503512e-07, "loss": 0.0679767906665802, "step": 32690 }, { "epoch": 0.8613692676220973, "grad_norm": 5.449418067932129, "learning_rate": 4.957904172821321e-07, "loss": 0.07160573601722717, "step": 32700 }, { "epoch": 0.8616326833002692, "grad_norm": 3.6766839027404785, "learning_rate": 4.939401206719757e-07, "loss": 0.0770403802394867, "step": 32710 }, { "epoch": 0.8618960989784411, "grad_norm": 5.240781784057617, "learning_rate": 4.920931037666072e-07, "loss": 0.07660043835639954, "step": 32720 }, { "epoch": 0.862159514656613, "grad_norm": 5.345116138458252, "learning_rate": 4.902493679103637e-07, "loss": 0.08093698620796204, "step": 32730 }, { "epoch": 0.8624229303347849, "grad_norm": 6.029928684234619, "learning_rate": 4.884089144451992e-07, "loss": 0.07553762197494507, "step": 32740 }, { "epoch": 0.8626863460129568, "grad_norm": 5.067325592041016, "learning_rate": 4.865717447106733e-07, "loss": 0.07431252598762512, "step": 32750 }, { "epoch": 0.8629497616911287, "grad_norm": 6.143150806427002, "learning_rate": 4.847378600439573e-07, "loss": 0.07147436141967774, "step": 32760 }, { "epoch": 0.8632131773693006, "grad_norm": 6.1793928146362305, "learning_rate": 4.829072617798314e-07, "loss": 0.07443949580192566, "step": 32770 }, { "epoch": 0.8634765930474725, "grad_norm": 6.234432697296143, "learning_rate": 4.810799512506836e-07, "loss": 0.06829504370689392, "step": 32780 }, { "epoch": 0.8637400087256444, "grad_norm": 4.725831031799316, "learning_rate": 4.792559297865113e-07, "loss": 0.07854129076004028, "step": 32790 }, { "epoch": 0.8640034244038163, "grad_norm": 5.855370044708252, "learning_rate": 4.774351987149156e-07, "loss": 0.07967990040779113, "step": 32800 }, { "epoch": 0.8642668400819882, "grad_norm": 6.085787773132324, "learning_rate": 4.756177593611e-07, "loss": 0.08028197884559632, "step": 32810 }, { "epoch": 0.8645302557601601, "grad_norm": 5.2228312492370605, "learning_rate": 4.73803613047879e-07, "loss": 0.07407301068305969, "step": 32820 }, { "epoch": 0.864793671438332, "grad_norm": 4.949946880340576, "learning_rate": 4.7199276109566484e-07, "loss": 0.0746949553489685, "step": 32830 }, { "epoch": 0.8650570871165039, "grad_norm": 6.884580612182617, "learning_rate": 4.701852048224742e-07, "loss": 0.06822417974472046, "step": 32840 }, { "epoch": 0.8653205027946758, "grad_norm": 4.359294414520264, "learning_rate": 4.683809455439242e-07, "loss": 0.07607890367507934, "step": 32850 }, { "epoch": 0.8655839184728477, "grad_norm": 6.414586544036865, "learning_rate": 4.665799845732322e-07, "loss": 0.07206892967224121, "step": 32860 }, { "epoch": 0.8658473341510196, "grad_norm": 4.810818672180176, "learning_rate": 4.6478232322121617e-07, "loss": 0.08124241828918458, "step": 32870 }, { "epoch": 0.8661107498291915, "grad_norm": 5.1546220779418945, "learning_rate": 4.6298796279629134e-07, "loss": 0.08888628482818603, "step": 32880 }, { "epoch": 0.8663741655073632, "grad_norm": 7.489560604095459, "learning_rate": 4.6119690460447063e-07, "loss": 0.08017309308052063, "step": 32890 }, { "epoch": 0.8666375811855351, "grad_norm": 3.558959722518921, "learning_rate": 4.5940914994936304e-07, "loss": 0.06943416595458984, "step": 32900 }, { "epoch": 0.866900996863707, "grad_norm": 6.117400646209717, "learning_rate": 4.576247001321732e-07, "loss": 0.07118109464645386, "step": 32910 }, { "epoch": 0.8671644125418789, "grad_norm": 6.2267656326293945, "learning_rate": 4.558435564517033e-07, "loss": 0.0648075819015503, "step": 32920 }, { "epoch": 0.8674278282200508, "grad_norm": 4.013854026794434, "learning_rate": 4.540657202043436e-07, "loss": 0.08029189705848694, "step": 32930 }, { "epoch": 0.8676912438982227, "grad_norm": 4.632141590118408, "learning_rate": 4.522911926840801e-07, "loss": 0.07026289701461792, "step": 32940 }, { "epoch": 0.8679546595763946, "grad_norm": 6.364835739135742, "learning_rate": 4.505199751824918e-07, "loss": 0.07643074989318847, "step": 32950 }, { "epoch": 0.8682180752545665, "grad_norm": 4.330821990966797, "learning_rate": 4.4875206898874757e-07, "loss": 0.0668033480644226, "step": 32960 }, { "epoch": 0.8684814909327384, "grad_norm": 5.24167013168335, "learning_rate": 4.469874753896047e-07, "loss": 0.06809518337249756, "step": 32970 }, { "epoch": 0.8687449066109103, "grad_norm": 4.414320468902588, "learning_rate": 4.452261956694109e-07, "loss": 0.08158721923828124, "step": 32980 }, { "epoch": 0.8690083222890822, "grad_norm": 5.507406234741211, "learning_rate": 4.4346823111010076e-07, "loss": 0.08897801041603089, "step": 32990 }, { "epoch": 0.8692717379672541, "grad_norm": 5.314439296722412, "learning_rate": 4.4171358299119835e-07, "loss": 0.07276229858398438, "step": 33000 }, { "epoch": 0.869535153645426, "grad_norm": 5.7653608322143555, "learning_rate": 4.399622525898123e-07, "loss": 0.06964116096496582, "step": 33010 }, { "epoch": 0.8697985693235979, "grad_norm": 5.359834671020508, "learning_rate": 4.3821424118063336e-07, "loss": 0.0793626070022583, "step": 33020 }, { "epoch": 0.8700619850017698, "grad_norm": 6.041033744812012, "learning_rate": 4.3646955003594337e-07, "loss": 0.07728523015975952, "step": 33030 }, { "epoch": 0.8703254006799417, "grad_norm": 4.801740646362305, "learning_rate": 4.3472818042560207e-07, "loss": 0.08416228890419006, "step": 33040 }, { "epoch": 0.8705888163581136, "grad_norm": 5.779715061187744, "learning_rate": 4.329901336170539e-07, "loss": 0.06567152738571166, "step": 33050 }, { "epoch": 0.8708522320362855, "grad_norm": 4.190917015075684, "learning_rate": 4.312554108753236e-07, "loss": 0.07318891286849975, "step": 33060 }, { "epoch": 0.8711156477144574, "grad_norm": 6.292623996734619, "learning_rate": 4.2952401346301685e-07, "loss": 0.07660646438598633, "step": 33070 }, { "epoch": 0.8713790633926293, "grad_norm": 5.764009475708008, "learning_rate": 4.2779594264032107e-07, "loss": 0.07482507228851318, "step": 33080 }, { "epoch": 0.8716424790708012, "grad_norm": 6.3215651512146, "learning_rate": 4.2607119966500023e-07, "loss": 0.08039939999580384, "step": 33090 }, { "epoch": 0.8719058947489731, "grad_norm": 4.784054756164551, "learning_rate": 4.2434978579239617e-07, "loss": 0.08142877221107483, "step": 33100 }, { "epoch": 0.872169310427145, "grad_norm": 3.983722448348999, "learning_rate": 4.2263170227542785e-07, "loss": 0.06873674392700195, "step": 33110 }, { "epoch": 0.8724327261053169, "grad_norm": 5.635846138000488, "learning_rate": 4.20916950364591e-07, "loss": 0.0804761826992035, "step": 33120 }, { "epoch": 0.8726961417834888, "grad_norm": 4.160591125488281, "learning_rate": 4.192055313079563e-07, "loss": 0.07128607034683228, "step": 33130 }, { "epoch": 0.8729595574616607, "grad_norm": 5.097555160522461, "learning_rate": 4.174974463511694e-07, "loss": 0.0673882782459259, "step": 33140 }, { "epoch": 0.8732229731398325, "grad_norm": 4.346644401550293, "learning_rate": 4.157926967374454e-07, "loss": 0.06604174971580505, "step": 33150 }, { "epoch": 0.8734863888180044, "grad_norm": 5.3926262855529785, "learning_rate": 4.1409128370757667e-07, "loss": 0.06766842603683472, "step": 33160 }, { "epoch": 0.8737498044961763, "grad_norm": 7.953225612640381, "learning_rate": 4.123932084999249e-07, "loss": 0.07144508957862854, "step": 33170 }, { "epoch": 0.8740132201743482, "grad_norm": 6.1560187339782715, "learning_rate": 4.1069847235042204e-07, "loss": 0.07946867942810058, "step": 33180 }, { "epoch": 0.8742766358525201, "grad_norm": 4.007959842681885, "learning_rate": 4.090070764925702e-07, "loss": 0.07492286562919617, "step": 33190 }, { "epoch": 0.874540051530692, "grad_norm": 6.366214275360107, "learning_rate": 4.073190221574397e-07, "loss": 0.07014946341514587, "step": 33200 }, { "epoch": 0.8748034672088639, "grad_norm": 5.940914630889893, "learning_rate": 4.05634310573671e-07, "loss": 0.07425289154052735, "step": 33210 }, { "epoch": 0.8750668828870358, "grad_norm": 5.989242076873779, "learning_rate": 4.039529429674682e-07, "loss": 0.07787153124809265, "step": 33220 }, { "epoch": 0.8753302985652077, "grad_norm": 6.508825302124023, "learning_rate": 4.0227492056260476e-07, "loss": 0.07007117867469788, "step": 33230 }, { "epoch": 0.8755937142433796, "grad_norm": 5.163427829742432, "learning_rate": 4.0060024458041637e-07, "loss": 0.07344157695770263, "step": 33240 }, { "epoch": 0.8758571299215515, "grad_norm": 5.79435920715332, "learning_rate": 3.989289162398036e-07, "loss": 0.08380992412567138, "step": 33250 }, { "epoch": 0.8761205455997234, "grad_norm": 5.249152183532715, "learning_rate": 3.9726093675723444e-07, "loss": 0.08146946430206299, "step": 33260 }, { "epoch": 0.8763839612778953, "grad_norm": 5.93942403793335, "learning_rate": 3.955963073467345e-07, "loss": 0.07841423749923707, "step": 33270 }, { "epoch": 0.8766473769560672, "grad_norm": 5.575658321380615, "learning_rate": 3.939350292198918e-07, "loss": 0.08999207615852356, "step": 33280 }, { "epoch": 0.8769107926342391, "grad_norm": 6.2806172370910645, "learning_rate": 3.92277103585858e-07, "loss": 0.08085976839065552, "step": 33290 }, { "epoch": 0.877174208312411, "grad_norm": 6.41100549697876, "learning_rate": 3.9062253165134235e-07, "loss": 0.06761197447776794, "step": 33300 }, { "epoch": 0.8774376239905829, "grad_norm": 5.136833667755127, "learning_rate": 3.8897131462061365e-07, "loss": 0.0751099169254303, "step": 33310 }, { "epoch": 0.8777010396687548, "grad_norm": 4.235189914703369, "learning_rate": 3.8732345369549895e-07, "loss": 0.07305091619491577, "step": 33320 }, { "epoch": 0.8779644553469267, "grad_norm": 5.163221836090088, "learning_rate": 3.856789500753816e-07, "loss": 0.07306385040283203, "step": 33330 }, { "epoch": 0.8782278710250986, "grad_norm": 6.109523296356201, "learning_rate": 3.840378049572036e-07, "loss": 0.07884476184844971, "step": 33340 }, { "epoch": 0.8784912867032705, "grad_norm": 6.853399276733398, "learning_rate": 3.8240001953546104e-07, "loss": 0.06557489037513733, "step": 33350 }, { "epoch": 0.8787547023814424, "grad_norm": 4.235916614532471, "learning_rate": 3.8076559500220435e-07, "loss": 0.06513867378234864, "step": 33360 }, { "epoch": 0.8790181180596143, "grad_norm": 5.908555507659912, "learning_rate": 3.7913453254703856e-07, "loss": 0.0653915524482727, "step": 33370 }, { "epoch": 0.8792815337377862, "grad_norm": 4.324704647064209, "learning_rate": 3.775068333571197e-07, "loss": 0.05895015597343445, "step": 33380 }, { "epoch": 0.8795449494159581, "grad_norm": 5.08020544052124, "learning_rate": 3.7588249861715997e-07, "loss": 0.06343244314193726, "step": 33390 }, { "epoch": 0.87980836509413, "grad_norm": 5.276219844818115, "learning_rate": 3.7426152950941986e-07, "loss": 0.06657284498214722, "step": 33400 }, { "epoch": 0.8800717807723019, "grad_norm": 6.993581295013428, "learning_rate": 3.7264392721370836e-07, "loss": 0.07225316762924194, "step": 33410 }, { "epoch": 0.8803351964504738, "grad_norm": 5.370820999145508, "learning_rate": 3.7102969290738865e-07, "loss": 0.06710069179534912, "step": 33420 }, { "epoch": 0.8805986121286457, "grad_norm": 3.287144660949707, "learning_rate": 3.694188277653693e-07, "loss": 0.09049853086471557, "step": 33430 }, { "epoch": 0.8808620278068175, "grad_norm": 6.243648052215576, "learning_rate": 3.6781133296010684e-07, "loss": 0.06985522508621216, "step": 33440 }, { "epoch": 0.8811254434849894, "grad_norm": 5.142467021942139, "learning_rate": 3.6620720966160593e-07, "loss": 0.07378503084182739, "step": 33450 }, { "epoch": 0.8813888591631613, "grad_norm": 6.002206802368164, "learning_rate": 3.646064590374165e-07, "loss": 0.06531034708023072, "step": 33460 }, { "epoch": 0.8816522748413332, "grad_norm": 4.167457580566406, "learning_rate": 3.6300908225263443e-07, "loss": 0.0808516263961792, "step": 33470 }, { "epoch": 0.8819156905195051, "grad_norm": 6.482204914093018, "learning_rate": 3.6141508046989905e-07, "loss": 0.07479463815689087, "step": 33480 }, { "epoch": 0.882179106197677, "grad_norm": 4.611236572265625, "learning_rate": 3.5982445484939343e-07, "loss": 0.07022485733032227, "step": 33490 }, { "epoch": 0.8824425218758489, "grad_norm": 5.3725152015686035, "learning_rate": 3.582372065488443e-07, "loss": 0.06428645849227906, "step": 33500 }, { "epoch": 0.8827059375540208, "grad_norm": 5.806538105010986, "learning_rate": 3.5665333672351753e-07, "loss": 0.06775869131088257, "step": 33510 }, { "epoch": 0.8829693532321927, "grad_norm": 5.74099063873291, "learning_rate": 3.5507284652622543e-07, "loss": 0.0717349886894226, "step": 33520 }, { "epoch": 0.8832327689103646, "grad_norm": 3.8851239681243896, "learning_rate": 3.5349573710731333e-07, "loss": 0.0678119719028473, "step": 33530 }, { "epoch": 0.8834961845885365, "grad_norm": 5.288051128387451, "learning_rate": 3.519220096146708e-07, "loss": 0.07408688068389893, "step": 33540 }, { "epoch": 0.8837596002667084, "grad_norm": 5.049154758453369, "learning_rate": 3.5035166519372556e-07, "loss": 0.07078350782394409, "step": 33550 }, { "epoch": 0.8840230159448803, "grad_norm": 4.97130012512207, "learning_rate": 3.48784704987441e-07, "loss": 0.06357211470603943, "step": 33560 }, { "epoch": 0.8842864316230522, "grad_norm": 4.390218257904053, "learning_rate": 3.4722113013631876e-07, "loss": 0.06986448764801026, "step": 33570 }, { "epoch": 0.8845498473012241, "grad_norm": 4.954543113708496, "learning_rate": 3.4566094177839624e-07, "loss": 0.06345463395118714, "step": 33580 }, { "epoch": 0.884813262979396, "grad_norm": 5.34235954284668, "learning_rate": 3.441041410492446e-07, "loss": 0.06047357320785522, "step": 33590 }, { "epoch": 0.8850766786575679, "grad_norm": 8.823299407958984, "learning_rate": 3.4255072908197183e-07, "loss": 0.07525157928466797, "step": 33600 }, { "epoch": 0.8853400943357398, "grad_norm": 5.55422830581665, "learning_rate": 3.410007070072191e-07, "loss": 0.0750880241394043, "step": 33610 }, { "epoch": 0.8856035100139117, "grad_norm": 5.7291259765625, "learning_rate": 3.394540759531567e-07, "loss": 0.07480108737945557, "step": 33620 }, { "epoch": 0.8858669256920836, "grad_norm": 6.970531940460205, "learning_rate": 3.3791083704549144e-07, "loss": 0.06806767582893372, "step": 33630 }, { "epoch": 0.8861303413702555, "grad_norm": 5.832605361938477, "learning_rate": 3.3637099140745867e-07, "loss": 0.06699259281158447, "step": 33640 }, { "epoch": 0.8863937570484274, "grad_norm": 6.512474536895752, "learning_rate": 3.348345401598241e-07, "loss": 0.06975846290588379, "step": 33650 }, { "epoch": 0.8866571727265993, "grad_norm": 7.0161848068237305, "learning_rate": 3.333014844208837e-07, "loss": 0.07895577549934388, "step": 33660 }, { "epoch": 0.8869205884047712, "grad_norm": 6.1517767906188965, "learning_rate": 3.3177182530646114e-07, "loss": 0.0773487389087677, "step": 33670 }, { "epoch": 0.887184004082943, "grad_norm": 2.5547077655792236, "learning_rate": 3.30245563929909e-07, "loss": 0.061495077610015866, "step": 33680 }, { "epoch": 0.8874474197611149, "grad_norm": 4.1165900230407715, "learning_rate": 3.287227014021055e-07, "loss": 0.07754195332527161, "step": 33690 }, { "epoch": 0.8877108354392867, "grad_norm": 5.181070327758789, "learning_rate": 3.2720323883145613e-07, "loss": 0.0712195634841919, "step": 33700 }, { "epoch": 0.8879742511174586, "grad_norm": 2.8546557426452637, "learning_rate": 3.256871773238912e-07, "loss": 0.07102495431900024, "step": 33710 }, { "epoch": 0.8882376667956305, "grad_norm": 5.203920841217041, "learning_rate": 3.241745179828648e-07, "loss": 0.06250168085098266, "step": 33720 }, { "epoch": 0.8885010824738024, "grad_norm": 4.97288703918457, "learning_rate": 3.2266526190935735e-07, "loss": 0.06593352556228638, "step": 33730 }, { "epoch": 0.8887644981519743, "grad_norm": 5.939675807952881, "learning_rate": 3.211594102018706e-07, "loss": 0.06460586786270142, "step": 33740 }, { "epoch": 0.8890279138301462, "grad_norm": 5.941983699798584, "learning_rate": 3.1965696395642565e-07, "loss": 0.07210645079612732, "step": 33750 }, { "epoch": 0.8892913295083181, "grad_norm": 5.226228713989258, "learning_rate": 3.1815792426657e-07, "loss": 0.06461670398712158, "step": 33760 }, { "epoch": 0.88955474518649, "grad_norm": 7.6564507484436035, "learning_rate": 3.166622922233692e-07, "loss": 0.07966884970664978, "step": 33770 }, { "epoch": 0.8898181608646619, "grad_norm": 6.470453262329102, "learning_rate": 3.151700689154086e-07, "loss": 0.06388614177703858, "step": 33780 }, { "epoch": 0.8900815765428338, "grad_norm": 4.304139614105225, "learning_rate": 3.1368125542879215e-07, "loss": 0.06468616724014283, "step": 33790 }, { "epoch": 0.8903449922210057, "grad_norm": 6.088304042816162, "learning_rate": 3.1219585284714247e-07, "loss": 0.07135077118873596, "step": 33800 }, { "epoch": 0.8906084078991776, "grad_norm": 5.465991020202637, "learning_rate": 3.1071386225160096e-07, "loss": 0.07805630564689636, "step": 33810 }, { "epoch": 0.8908718235773495, "grad_norm": 5.405788898468018, "learning_rate": 3.0923528472082353e-07, "loss": 0.07511802315711975, "step": 33820 }, { "epoch": 0.8911352392555214, "grad_norm": 5.793666839599609, "learning_rate": 3.0776012133098274e-07, "loss": 0.07194539308547973, "step": 33830 }, { "epoch": 0.8913986549336933, "grad_norm": 5.879497051239014, "learning_rate": 3.0628837315576586e-07, "loss": 0.0762730062007904, "step": 33840 }, { "epoch": 0.8916620706118652, "grad_norm": 5.061279773712158, "learning_rate": 3.0482004126637433e-07, "loss": 0.06992514133453369, "step": 33850 }, { "epoch": 0.8919254862900371, "grad_norm": 7.448275089263916, "learning_rate": 3.033551267315249e-07, "loss": 0.05825051069259644, "step": 33860 }, { "epoch": 0.892188901968209, "grad_norm": 5.963418006896973, "learning_rate": 3.0189363061744525e-07, "loss": 0.06934090852737426, "step": 33870 }, { "epoch": 0.8924523176463809, "grad_norm": 5.356656551361084, "learning_rate": 3.004355539878734e-07, "loss": 0.07314136028289794, "step": 33880 }, { "epoch": 0.8927157333245528, "grad_norm": 5.841501712799072, "learning_rate": 2.9898089790406205e-07, "loss": 0.07589088678359986, "step": 33890 }, { "epoch": 0.8929791490027247, "grad_norm": 5.510014057159424, "learning_rate": 2.975296634247732e-07, "loss": 0.06873184442520142, "step": 33900 }, { "epoch": 0.8932425646808966, "grad_norm": 7.339508533477783, "learning_rate": 2.960818516062769e-07, "loss": 0.07846101522445678, "step": 33910 }, { "epoch": 0.8935059803590685, "grad_norm": 5.46766471862793, "learning_rate": 2.94637463502353e-07, "loss": 0.06304433345794677, "step": 33920 }, { "epoch": 0.8937693960372404, "grad_norm": 6.585385322570801, "learning_rate": 2.9319650016428894e-07, "loss": 0.07005903720855713, "step": 33930 }, { "epoch": 0.8940328117154123, "grad_norm": 5.355676174163818, "learning_rate": 2.9175896264088176e-07, "loss": 0.07499082088470459, "step": 33940 }, { "epoch": 0.8942962273935842, "grad_norm": 6.6426873207092285, "learning_rate": 2.903248519784324e-07, "loss": 0.07398062944412231, "step": 33950 }, { "epoch": 0.8945596430717561, "grad_norm": 5.773818016052246, "learning_rate": 2.8889416922074867e-07, "loss": 0.06760952472686768, "step": 33960 }, { "epoch": 0.894823058749928, "grad_norm": 6.808023929595947, "learning_rate": 2.8746691540914253e-07, "loss": 0.06460528969764709, "step": 33970 }, { "epoch": 0.8950864744280999, "grad_norm": 5.505505084991455, "learning_rate": 2.8604309158243084e-07, "loss": 0.06676326990127564, "step": 33980 }, { "epoch": 0.8953498901062718, "grad_norm": 7.084875583648682, "learning_rate": 2.8462269877693516e-07, "loss": 0.08285461068153381, "step": 33990 }, { "epoch": 0.8956133057844436, "grad_norm": 5.282711982727051, "learning_rate": 2.8320573802647857e-07, "loss": 0.06643307209014893, "step": 34000 }, { "epoch": 0.8958767214626155, "grad_norm": 6.492991924285889, "learning_rate": 2.8179221036238437e-07, "loss": 0.0716627836227417, "step": 34010 }, { "epoch": 0.8961401371407874, "grad_norm": 6.592701435089111, "learning_rate": 2.8038211681348073e-07, "loss": 0.06684476733207703, "step": 34020 }, { "epoch": 0.8964035528189593, "grad_norm": 3.8093695640563965, "learning_rate": 2.789754584060944e-07, "loss": 0.07512718439102173, "step": 34030 }, { "epoch": 0.8966669684971312, "grad_norm": 4.75612735748291, "learning_rate": 2.7757223616405146e-07, "loss": 0.07386713027954102, "step": 34040 }, { "epoch": 0.8969303841753031, "grad_norm": 6.369325637817383, "learning_rate": 2.761724511086783e-07, "loss": 0.07944008111953735, "step": 34050 }, { "epoch": 0.897193799853475, "grad_norm": 5.48779296875, "learning_rate": 2.747761042587971e-07, "loss": 0.07172590494155884, "step": 34060 }, { "epoch": 0.8974572155316469, "grad_norm": 6.928581237792969, "learning_rate": 2.733831966307321e-07, "loss": 0.0740256130695343, "step": 34070 }, { "epoch": 0.8977206312098188, "grad_norm": 5.4985785484313965, "learning_rate": 2.719937292383007e-07, "loss": 0.06957334280014038, "step": 34080 }, { "epoch": 0.8979840468879907, "grad_norm": 3.242140293121338, "learning_rate": 2.7060770309281506e-07, "loss": 0.08318769931793213, "step": 34090 }, { "epoch": 0.8982474625661626, "grad_norm": 5.357398509979248, "learning_rate": 2.69225119203087e-07, "loss": 0.0752285361289978, "step": 34100 }, { "epoch": 0.8985108782443345, "grad_norm": 5.715819835662842, "learning_rate": 2.6784597857542005e-07, "loss": 0.06561976671218872, "step": 34110 }, { "epoch": 0.8987742939225064, "grad_norm": 5.664666652679443, "learning_rate": 2.664702822136134e-07, "loss": 0.06634891033172607, "step": 34120 }, { "epoch": 0.8990377096006783, "grad_norm": 5.493467807769775, "learning_rate": 2.650980311189566e-07, "loss": 0.06386145949363708, "step": 34130 }, { "epoch": 0.8993011252788502, "grad_norm": 6.171666622161865, "learning_rate": 2.63729226290233e-07, "loss": 0.0704851508140564, "step": 34140 }, { "epoch": 0.8995645409570221, "grad_norm": 5.7214508056640625, "learning_rate": 2.62363868723719e-07, "loss": 0.06735482215881347, "step": 34150 }, { "epoch": 0.899827956635194, "grad_norm": 4.811983585357666, "learning_rate": 2.6100195941318075e-07, "loss": 0.06733543872833252, "step": 34160 }, { "epoch": 0.9000913723133659, "grad_norm": 6.464471340179443, "learning_rate": 2.596434993498742e-07, "loss": 0.06695042848587036, "step": 34170 }, { "epoch": 0.9003547879915378, "grad_norm": 6.876003742218018, "learning_rate": 2.5828848952254515e-07, "loss": 0.07683340311050416, "step": 34180 }, { "epoch": 0.9006182036697097, "grad_norm": 3.98159122467041, "learning_rate": 2.569369309174274e-07, "loss": 0.07473430633544922, "step": 34190 }, { "epoch": 0.9008816193478816, "grad_norm": 4.95936393737793, "learning_rate": 2.555888245182453e-07, "loss": 0.05649802088737488, "step": 34200 }, { "epoch": 0.9011450350260535, "grad_norm": 3.8291494846343994, "learning_rate": 2.5424417130620826e-07, "loss": 0.0774224877357483, "step": 34210 }, { "epoch": 0.9014084507042254, "grad_norm": 6.445468425750732, "learning_rate": 2.5290297226001183e-07, "loss": 0.06939530968666077, "step": 34220 }, { "epoch": 0.9016718663823973, "grad_norm": 6.131041526794434, "learning_rate": 2.515652283558401e-07, "loss": 0.0768373191356659, "step": 34230 }, { "epoch": 0.9019352820605692, "grad_norm": 6.116026401519775, "learning_rate": 2.502309405673598e-07, "loss": 0.06526549458503723, "step": 34240 }, { "epoch": 0.9021986977387411, "grad_norm": 6.335423946380615, "learning_rate": 2.4890010986572345e-07, "loss": 0.07528613209724426, "step": 34250 }, { "epoch": 0.902462113416913, "grad_norm": 4.588692665100098, "learning_rate": 2.475727372195669e-07, "loss": 0.0660396695137024, "step": 34260 }, { "epoch": 0.9027255290950849, "grad_norm": 3.7281010150909424, "learning_rate": 2.462488235950089e-07, "loss": 0.0681867241859436, "step": 34270 }, { "epoch": 0.9029889447732568, "grad_norm": 5.156481742858887, "learning_rate": 2.449283699556515e-07, "loss": 0.05615466833114624, "step": 34280 }, { "epoch": 0.9032523604514286, "grad_norm": 5.904376983642578, "learning_rate": 2.436113772625781e-07, "loss": 0.0725115716457367, "step": 34290 }, { "epoch": 0.9035157761296005, "grad_norm": 4.358409404754639, "learning_rate": 2.422978464743525e-07, "loss": 0.056048065423965454, "step": 34300 }, { "epoch": 0.9037791918077724, "grad_norm": 5.097048282623291, "learning_rate": 2.4098777854701883e-07, "loss": 0.06853222846984863, "step": 34310 }, { "epoch": 0.9040426074859443, "grad_norm": 5.105210304260254, "learning_rate": 2.396811744341004e-07, "loss": 0.06185494661331177, "step": 34320 }, { "epoch": 0.9043060231641162, "grad_norm": 3.179060935974121, "learning_rate": 2.3837803508660161e-07, "loss": 0.06964973211288453, "step": 34330 }, { "epoch": 0.9045694388422881, "grad_norm": 5.962168216705322, "learning_rate": 2.370783614530031e-07, "loss": 0.06864007711410522, "step": 34340 }, { "epoch": 0.90483285452046, "grad_norm": 4.989119052886963, "learning_rate": 2.357821544792621e-07, "loss": 0.08202240467071534, "step": 34350 }, { "epoch": 0.9050962701986319, "grad_norm": 6.560608386993408, "learning_rate": 2.3448941510881507e-07, "loss": 0.07403069138526916, "step": 34360 }, { "epoch": 0.9053596858768038, "grad_norm": 7.725258827209473, "learning_rate": 2.3320014428257375e-07, "loss": 0.06982209682464599, "step": 34370 }, { "epoch": 0.9056231015549757, "grad_norm": 3.7714040279388428, "learning_rate": 2.319143429389248e-07, "loss": 0.06368421912193298, "step": 34380 }, { "epoch": 0.9058865172331476, "grad_norm": 6.790114879608154, "learning_rate": 2.3063201201372954e-07, "loss": 0.08135839700698852, "step": 34390 }, { "epoch": 0.9061499329113195, "grad_norm": 4.836159706115723, "learning_rate": 2.293531524403231e-07, "loss": 0.06039778590202331, "step": 34400 }, { "epoch": 0.9064133485894914, "grad_norm": 5.0525031089782715, "learning_rate": 2.280777651495164e-07, "loss": 0.06444722414016724, "step": 34410 }, { "epoch": 0.9066767642676633, "grad_norm": 5.082833766937256, "learning_rate": 2.268058510695903e-07, "loss": 0.06761437058448791, "step": 34420 }, { "epoch": 0.9069401799458352, "grad_norm": 5.067622661590576, "learning_rate": 2.2553741112629867e-07, "loss": 0.06560583114624023, "step": 34430 }, { "epoch": 0.9072035956240071, "grad_norm": 6.336693286895752, "learning_rate": 2.2427244624286747e-07, "loss": 0.07906868457794189, "step": 34440 }, { "epoch": 0.907467011302179, "grad_norm": 5.716230869293213, "learning_rate": 2.230109573399919e-07, "loss": 0.08145074844360352, "step": 34450 }, { "epoch": 0.9077304269803509, "grad_norm": 5.459799289703369, "learning_rate": 2.2175294533583858e-07, "loss": 0.05981930494308472, "step": 34460 }, { "epoch": 0.9079938426585228, "grad_norm": 5.762171745300293, "learning_rate": 2.204984111460434e-07, "loss": 0.05878556966781616, "step": 34470 }, { "epoch": 0.9082572583366946, "grad_norm": 6.114683151245117, "learning_rate": 2.1924735568370826e-07, "loss": 0.06748496294021607, "step": 34480 }, { "epoch": 0.9085206740148665, "grad_norm": 5.93878698348999, "learning_rate": 2.17999779859408e-07, "loss": 0.07502113580703736, "step": 34490 }, { "epoch": 0.9087840896930384, "grad_norm": 5.789543628692627, "learning_rate": 2.1675568458118078e-07, "loss": 0.08155208230018615, "step": 34500 }, { "epoch": 0.9090475053712103, "grad_norm": 6.097339153289795, "learning_rate": 2.1551507075453337e-07, "loss": 0.06569490432739258, "step": 34510 }, { "epoch": 0.9093109210493822, "grad_norm": 7.855108261108398, "learning_rate": 2.1427793928243734e-07, "loss": 0.07161854505538941, "step": 34520 }, { "epoch": 0.9095743367275541, "grad_norm": 5.675601482391357, "learning_rate": 2.130442910653302e-07, "loss": 0.06569446325302124, "step": 34530 }, { "epoch": 0.909837752405726, "grad_norm": 5.811634063720703, "learning_rate": 2.1181412700111536e-07, "loss": 0.07177256345748902, "step": 34540 }, { "epoch": 0.9101011680838978, "grad_norm": 5.590038299560547, "learning_rate": 2.1058744798515883e-07, "loss": 0.09018337726593018, "step": 34550 }, { "epoch": 0.9103645837620697, "grad_norm": 7.380630016326904, "learning_rate": 2.0936425491029034e-07, "loss": 0.08277182579040528, "step": 34560 }, { "epoch": 0.9106279994402416, "grad_norm": 5.904324531555176, "learning_rate": 2.0814454866680267e-07, "loss": 0.06906275749206543, "step": 34570 }, { "epoch": 0.9108914151184135, "grad_norm": 3.632570505142212, "learning_rate": 2.0692833014245018e-07, "loss": 0.058166444301605225, "step": 34580 }, { "epoch": 0.9111548307965854, "grad_norm": 4.553287506103516, "learning_rate": 2.057156002224503e-07, "loss": 0.06221001148223877, "step": 34590 }, { "epoch": 0.9114182464747573, "grad_norm": 6.36288595199585, "learning_rate": 2.045063597894803e-07, "loss": 0.06983338594436646, "step": 34600 }, { "epoch": 0.9116816621529292, "grad_norm": 5.407476425170898, "learning_rate": 2.033006097236756e-07, "loss": 0.07205095291137695, "step": 34610 }, { "epoch": 0.9119450778311011, "grad_norm": 5.985437870025635, "learning_rate": 2.020983509026353e-07, "loss": 0.05922784805297852, "step": 34620 }, { "epoch": 0.912208493509273, "grad_norm": 6.617381572723389, "learning_rate": 2.0089958420141388e-07, "loss": 0.08104726076126098, "step": 34630 }, { "epoch": 0.9124719091874449, "grad_norm": 4.32213830947876, "learning_rate": 1.9970431049252558e-07, "loss": 0.07266393303871155, "step": 34640 }, { "epoch": 0.9127353248656168, "grad_norm": 6.09397554397583, "learning_rate": 1.985125306459429e-07, "loss": 0.07329621315002441, "step": 34650 }, { "epoch": 0.9129987405437887, "grad_norm": 4.5721635818481445, "learning_rate": 1.9732424552909356e-07, "loss": 0.06788917779922485, "step": 34660 }, { "epoch": 0.9132621562219606, "grad_norm": 6.309351921081543, "learning_rate": 1.9613945600686414e-07, "loss": 0.05351153612136841, "step": 34670 }, { "epoch": 0.9135255719001325, "grad_norm": 7.595926284790039, "learning_rate": 1.9495816294159475e-07, "loss": 0.062231671810150144, "step": 34680 }, { "epoch": 0.9137889875783044, "grad_norm": 5.954641819000244, "learning_rate": 1.9378036719308101e-07, "loss": 0.0604053258895874, "step": 34690 }, { "epoch": 0.9140524032564763, "grad_norm": 6.228517055511475, "learning_rate": 1.9260606961857388e-07, "loss": 0.07181408405303955, "step": 34700 }, { "epoch": 0.9143158189346482, "grad_norm": 5.423251628875732, "learning_rate": 1.9143527107277847e-07, "loss": 0.06767276525497437, "step": 34710 }, { "epoch": 0.9145792346128201, "grad_norm": 4.584412574768066, "learning_rate": 1.9026797240785155e-07, "loss": 0.06844727993011475, "step": 34720 }, { "epoch": 0.914842650290992, "grad_norm": 4.220582485198975, "learning_rate": 1.89104174473404e-07, "loss": 0.06327673196792602, "step": 34730 }, { "epoch": 0.9151060659691639, "grad_norm": 6.096381664276123, "learning_rate": 1.8794387811649717e-07, "loss": 0.07161741256713867, "step": 34740 }, { "epoch": 0.9153694816473358, "grad_norm": 4.903972148895264, "learning_rate": 1.867870841816466e-07, "loss": 0.06693909168243409, "step": 34750 }, { "epoch": 0.9156328973255077, "grad_norm": 4.796250820159912, "learning_rate": 1.8563379351081544e-07, "loss": 0.06895486116409302, "step": 34760 }, { "epoch": 0.9158963130036796, "grad_norm": 4.012337684631348, "learning_rate": 1.8448400694341894e-07, "loss": 0.07556260824203491, "step": 34770 }, { "epoch": 0.9161597286818515, "grad_norm": 5.93644380569458, "learning_rate": 1.83337725316321e-07, "loss": 0.06469382047653198, "step": 34780 }, { "epoch": 0.9164231443600234, "grad_norm": 5.874996662139893, "learning_rate": 1.821949494638342e-07, "loss": 0.07384329438209533, "step": 34790 }, { "epoch": 0.9166865600381953, "grad_norm": 4.893446922302246, "learning_rate": 1.8105568021772212e-07, "loss": 0.07271872758865357, "step": 34800 }, { "epoch": 0.9169499757163672, "grad_norm": 6.830514430999756, "learning_rate": 1.7991991840719315e-07, "loss": 0.06558432579040527, "step": 34810 }, { "epoch": 0.9172133913945391, "grad_norm": 6.771401882171631, "learning_rate": 1.7878766485890266e-07, "loss": 0.0694648265838623, "step": 34820 }, { "epoch": 0.917476807072711, "grad_norm": 6.8097920417785645, "learning_rate": 1.7765892039695476e-07, "loss": 0.07804046869277954, "step": 34830 }, { "epoch": 0.9177402227508829, "grad_norm": 8.777909278869629, "learning_rate": 1.7653368584289843e-07, "loss": 0.06944088935852051, "step": 34840 }, { "epoch": 0.9180036384290547, "grad_norm": 4.970886707305908, "learning_rate": 1.7541196201572686e-07, "loss": 0.06574259996414185, "step": 34850 }, { "epoch": 0.9182670541072266, "grad_norm": 5.708157539367676, "learning_rate": 1.742937497318803e-07, "loss": 0.06960084438323974, "step": 34860 }, { "epoch": 0.9185304697853985, "grad_norm": 4.1143903732299805, "learning_rate": 1.7317904980524104e-07, "loss": 0.06447882056236268, "step": 34870 }, { "epoch": 0.9187938854635704, "grad_norm": 7.141999244689941, "learning_rate": 1.7206786304713617e-07, "loss": 0.08687325716018676, "step": 34880 }, { "epoch": 0.9190573011417423, "grad_norm": 4.928611755371094, "learning_rate": 1.7096019026633547e-07, "loss": 0.07675333619117737, "step": 34890 }, { "epoch": 0.9193207168199142, "grad_norm": 6.496432304382324, "learning_rate": 1.6985603226905066e-07, "loss": 0.06074240803718567, "step": 34900 }, { "epoch": 0.9195841324980861, "grad_norm": 5.556860446929932, "learning_rate": 1.6875538985893557e-07, "loss": 0.062767893075943, "step": 34910 }, { "epoch": 0.919847548176258, "grad_norm": 7.1152167320251465, "learning_rate": 1.6765826383708495e-07, "loss": 0.0638836681842804, "step": 34920 }, { "epoch": 0.9201109638544299, "grad_norm": 6.871767520904541, "learning_rate": 1.6656465500203555e-07, "loss": 0.06289691925048828, "step": 34930 }, { "epoch": 0.9203743795326018, "grad_norm": 4.842080593109131, "learning_rate": 1.6547456414976238e-07, "loss": 0.061293959617614746, "step": 34940 }, { "epoch": 0.9206377952107737, "grad_norm": 5.307104110717773, "learning_rate": 1.6438799207368017e-07, "loss": 0.06622807383537292, "step": 34950 }, { "epoch": 0.9209012108889456, "grad_norm": 4.311829566955566, "learning_rate": 1.6330493956464355e-07, "loss": 0.07627484202384949, "step": 34960 }, { "epoch": 0.9211646265671175, "grad_norm": 5.377438545227051, "learning_rate": 1.6222540741094474e-07, "loss": 0.076544189453125, "step": 34970 }, { "epoch": 0.9214280422452894, "grad_norm": 6.486123085021973, "learning_rate": 1.611493963983135e-07, "loss": 0.07216006517410278, "step": 34980 }, { "epoch": 0.9216914579234613, "grad_norm": 4.523367881774902, "learning_rate": 1.6007690730991789e-07, "loss": 0.06266617774963379, "step": 34990 }, { "epoch": 0.9219548736016332, "grad_norm": 6.369196891784668, "learning_rate": 1.5900794092636008e-07, "loss": 0.07034525871276856, "step": 35000 }, { "epoch": 0.9222182892798051, "grad_norm": 4.795995712280273, "learning_rate": 1.5794249802568218e-07, "loss": 0.07201036810874939, "step": 35010 }, { "epoch": 0.922481704957977, "grad_norm": 4.982271194458008, "learning_rate": 1.5688057938335776e-07, "loss": 0.060153496265411374, "step": 35020 }, { "epoch": 0.9227451206361489, "grad_norm": 7.112264633178711, "learning_rate": 1.5582218577229747e-07, "loss": 0.07285494804382324, "step": 35030 }, { "epoch": 0.9230085363143208, "grad_norm": 4.6727094650268555, "learning_rate": 1.5476731796284671e-07, "loss": 0.06597906351089478, "step": 35040 }, { "epoch": 0.9232719519924927, "grad_norm": 6.770757675170898, "learning_rate": 1.5371597672278194e-07, "loss": 0.08299517631530762, "step": 35050 }, { "epoch": 0.9235353676706646, "grad_norm": 4.746904373168945, "learning_rate": 1.5266816281731655e-07, "loss": 0.05979776382446289, "step": 35060 }, { "epoch": 0.9237987833488365, "grad_norm": 5.579225540161133, "learning_rate": 1.5162387700909442e-07, "loss": 0.06490904688835145, "step": 35070 }, { "epoch": 0.9240621990270084, "grad_norm": 5.9532599449157715, "learning_rate": 1.5058312005818976e-07, "loss": 0.07227004170417786, "step": 35080 }, { "epoch": 0.9243256147051803, "grad_norm": 4.716990947723389, "learning_rate": 1.4954589272211328e-07, "loss": 0.06300992965698242, "step": 35090 }, { "epoch": 0.9245890303833522, "grad_norm": 5.233424663543701, "learning_rate": 1.4851219575580167e-07, "loss": 0.055896437168121337, "step": 35100 }, { "epoch": 0.9248524460615241, "grad_norm": 5.663751125335693, "learning_rate": 1.4748202991162585e-07, "loss": 0.0748557448387146, "step": 35110 }, { "epoch": 0.925115861739696, "grad_norm": 5.789803981781006, "learning_rate": 1.4645539593938328e-07, "loss": 0.0740208387374878, "step": 35120 }, { "epoch": 0.9253792774178679, "grad_norm": 3.9712610244750977, "learning_rate": 1.4543229458630293e-07, "loss": 0.06616172790527344, "step": 35130 }, { "epoch": 0.9256426930960397, "grad_norm": 7.402288913726807, "learning_rate": 1.4441272659704353e-07, "loss": 0.06026633977890015, "step": 35140 }, { "epoch": 0.9259061087742116, "grad_norm": 5.337446212768555, "learning_rate": 1.4339669271368872e-07, "loss": 0.07850565910339355, "step": 35150 }, { "epoch": 0.9261695244523835, "grad_norm": 6.07331657409668, "learning_rate": 1.4238419367575306e-07, "loss": 0.07119031548500061, "step": 35160 }, { "epoch": 0.9264329401305554, "grad_norm": 7.419558525085449, "learning_rate": 1.4137523022017652e-07, "loss": 0.07891860008239746, "step": 35170 }, { "epoch": 0.9266963558087273, "grad_norm": 5.142488479614258, "learning_rate": 1.4036980308132607e-07, "loss": 0.0656275749206543, "step": 35180 }, { "epoch": 0.9269597714868992, "grad_norm": 5.187500953674316, "learning_rate": 1.3936791299099528e-07, "loss": 0.0702303946018219, "step": 35190 }, { "epoch": 0.9272231871650711, "grad_norm": 5.953988552093506, "learning_rate": 1.3836956067840357e-07, "loss": 0.0686109185218811, "step": 35200 }, { "epoch": 0.927486602843243, "grad_norm": 5.9956207275390625, "learning_rate": 1.3737474687019358e-07, "loss": 0.07424167394638062, "step": 35210 }, { "epoch": 0.9277500185214149, "grad_norm": 4.6628947257995605, "learning_rate": 1.3638347229043447e-07, "loss": 0.06963099241256714, "step": 35220 }, { "epoch": 0.9280134341995868, "grad_norm": 4.931395530700684, "learning_rate": 1.3539573766061852e-07, "loss": 0.06379730105400086, "step": 35230 }, { "epoch": 0.9282768498777587, "grad_norm": 6.33162260055542, "learning_rate": 1.3441154369966124e-07, "loss": 0.0707069218158722, "step": 35240 }, { "epoch": 0.9285402655559306, "grad_norm": 6.7892165184021, "learning_rate": 1.3343089112390184e-07, "loss": 0.07037928104400634, "step": 35250 }, { "epoch": 0.9288036812341025, "grad_norm": 5.430028438568115, "learning_rate": 1.3245378064710157e-07, "loss": 0.0654416561126709, "step": 35260 }, { "epoch": 0.9290670969122743, "grad_norm": 4.481483459472656, "learning_rate": 1.3148021298044322e-07, "loss": 0.07088671326637268, "step": 35270 }, { "epoch": 0.9293305125904462, "grad_norm": 6.671574592590332, "learning_rate": 1.3051018883253274e-07, "loss": 0.07619264125823974, "step": 35280 }, { "epoch": 0.9295939282686181, "grad_norm": 4.858748435974121, "learning_rate": 1.2954370890939372e-07, "loss": 0.06986613869667054, "step": 35290 }, { "epoch": 0.92985734394679, "grad_norm": 4.673599720001221, "learning_rate": 1.2858077391447287e-07, "loss": 0.060229575634002684, "step": 35300 }, { "epoch": 0.9301207596249619, "grad_norm": 3.892575263977051, "learning_rate": 1.2762138454863514e-07, "loss": 0.054621005058288576, "step": 35310 }, { "epoch": 0.9303841753031338, "grad_norm": 7.621439456939697, "learning_rate": 1.2666554151016697e-07, "loss": 0.07224227786064148, "step": 35320 }, { "epoch": 0.9306475909813057, "grad_norm": 6.018758773803711, "learning_rate": 1.257132454947707e-07, "loss": 0.07638453245162964, "step": 35330 }, { "epoch": 0.9309110066594776, "grad_norm": 7.288653373718262, "learning_rate": 1.2476449719556804e-07, "loss": 0.06289854645729065, "step": 35340 }, { "epoch": 0.9311744223376495, "grad_norm": 5.73895788192749, "learning_rate": 1.2381929730310105e-07, "loss": 0.06203625202178955, "step": 35350 }, { "epoch": 0.9314378380158214, "grad_norm": 6.4723358154296875, "learning_rate": 1.2287764650532552e-07, "loss": 0.06695555448532105, "step": 35360 }, { "epoch": 0.9317012536939933, "grad_norm": 7.237545967102051, "learning_rate": 1.21939545487616e-07, "loss": 0.06340780854225159, "step": 35370 }, { "epoch": 0.9319646693721652, "grad_norm": 6.053020477294922, "learning_rate": 1.2100499493276242e-07, "loss": 0.06569419503211975, "step": 35380 }, { "epoch": 0.932228085050337, "grad_norm": 4.402467250823975, "learning_rate": 1.2007399552097066e-07, "loss": 0.06845844984054565, "step": 35390 }, { "epoch": 0.932491500728509, "grad_norm": 5.714812278747559, "learning_rate": 1.1914654792986369e-07, "loss": 0.0693937361240387, "step": 35400 }, { "epoch": 0.9327549164066808, "grad_norm": 5.58941125869751, "learning_rate": 1.1822265283447765e-07, "loss": 0.07254610061645508, "step": 35410 }, { "epoch": 0.9330183320848527, "grad_norm": 5.469064235687256, "learning_rate": 1.1730231090726129e-07, "loss": 0.08381598591804504, "step": 35420 }, { "epoch": 0.9332817477630246, "grad_norm": 5.581043243408203, "learning_rate": 1.1638552281808102e-07, "loss": 0.06834365129470825, "step": 35430 }, { "epoch": 0.9335451634411965, "grad_norm": 4.413814544677734, "learning_rate": 1.1547228923421471e-07, "loss": 0.06424667239189148, "step": 35440 }, { "epoch": 0.9338085791193684, "grad_norm": 3.39452862739563, "learning_rate": 1.1456261082035236e-07, "loss": 0.06025948524475098, "step": 35450 }, { "epoch": 0.9340719947975403, "grad_norm": 7.197231292724609, "learning_rate": 1.1365648823859765e-07, "loss": 0.075601464509964, "step": 35460 }, { "epoch": 0.9343354104757122, "grad_norm": 5.3006415367126465, "learning_rate": 1.1275392214846526e-07, "loss": 0.05906842947006226, "step": 35470 }, { "epoch": 0.9345988261538841, "grad_norm": 6.0418195724487305, "learning_rate": 1.1185491320688247e-07, "loss": 0.06224328279495239, "step": 35480 }, { "epoch": 0.934862241832056, "grad_norm": 8.438061714172363, "learning_rate": 1.1095946206818586e-07, "loss": 0.06753079891204834, "step": 35490 }, { "epoch": 0.9351256575102279, "grad_norm": 6.736607074737549, "learning_rate": 1.1006756938412467e-07, "loss": 0.07127790451049805, "step": 35500 }, { "epoch": 0.9353890731883998, "grad_norm": 5.328470706939697, "learning_rate": 1.091792358038557e-07, "loss": 0.051967072486877444, "step": 35510 }, { "epoch": 0.9356524888665717, "grad_norm": 6.37784481048584, "learning_rate": 1.0829446197394622e-07, "loss": 0.06115795373916626, "step": 35520 }, { "epoch": 0.9359159045447436, "grad_norm": 4.607280731201172, "learning_rate": 1.0741324853837387e-07, "loss": 0.07296897768974304, "step": 35530 }, { "epoch": 0.9361793202229155, "grad_norm": 5.554794788360596, "learning_rate": 1.065355961385245e-07, "loss": 0.07200950980186463, "step": 35540 }, { "epoch": 0.9364427359010874, "grad_norm": 5.063072204589844, "learning_rate": 1.0566150541318876e-07, "loss": 0.0643563985824585, "step": 35550 }, { "epoch": 0.9367061515792593, "grad_norm": 6.935144901275635, "learning_rate": 1.0479097699856944e-07, "loss": 0.06724834442138672, "step": 35560 }, { "epoch": 0.9369695672574312, "grad_norm": 4.576374530792236, "learning_rate": 1.0392401152827469e-07, "loss": 0.05920727252960205, "step": 35570 }, { "epoch": 0.9372329829356031, "grad_norm": 6.494014739990234, "learning_rate": 1.0306060963331866e-07, "loss": 0.06793543100357055, "step": 35580 }, { "epoch": 0.937496398613775, "grad_norm": 4.301703453063965, "learning_rate": 1.0220077194212308e-07, "loss": 0.06283259987831116, "step": 35590 }, { "epoch": 0.9377598142919469, "grad_norm": 7.672483444213867, "learning_rate": 1.0134449908051458e-07, "loss": 0.06449269056320191, "step": 35600 }, { "epoch": 0.9380232299701188, "grad_norm": 6.831697940826416, "learning_rate": 1.0049179167172573e-07, "loss": 0.05412636995315552, "step": 35610 }, { "epoch": 0.9382866456482907, "grad_norm": 5.062432765960693, "learning_rate": 9.964265033639342e-08, "loss": 0.07070403695106506, "step": 35620 }, { "epoch": 0.9385500613264626, "grad_norm": 6.59227180480957, "learning_rate": 9.879707569255992e-08, "loss": 0.07246467471122742, "step": 35630 }, { "epoch": 0.9388134770046345, "grad_norm": 5.356546878814697, "learning_rate": 9.79550683556707e-08, "loss": 0.0738629162311554, "step": 35640 }, { "epoch": 0.9390768926828064, "grad_norm": 8.023842811584473, "learning_rate": 9.711662893857387e-08, "loss": 0.07110199928283692, "step": 35650 }, { "epoch": 0.9393403083609783, "grad_norm": 7.299337863922119, "learning_rate": 9.628175805152407e-08, "loss": 0.05945212841033935, "step": 35660 }, { "epoch": 0.9396037240391502, "grad_norm": 5.728797435760498, "learning_rate": 9.54504563021752e-08, "loss": 0.07019490599632264, "step": 35670 }, { "epoch": 0.939867139717322, "grad_norm": 6.0744242668151855, "learning_rate": 9.462272429558328e-08, "loss": 0.07509149312973022, "step": 35680 }, { "epoch": 0.940130555395494, "grad_norm": 8.237571716308594, "learning_rate": 9.379856263420972e-08, "loss": 0.06613152027130127, "step": 35690 }, { "epoch": 0.9403939710736658, "grad_norm": 5.32693338394165, "learning_rate": 9.297797191791302e-08, "loss": 0.05287469625473022, "step": 35700 }, { "epoch": 0.9406573867518377, "grad_norm": 4.858672142028809, "learning_rate": 9.216095274395598e-08, "loss": 0.06537923812866211, "step": 35710 }, { "epoch": 0.9409208024300096, "grad_norm": 4.568402290344238, "learning_rate": 9.134750570699958e-08, "loss": 0.06383408308029175, "step": 35720 }, { "epoch": 0.9411842181081815, "grad_norm": 7.543764591217041, "learning_rate": 9.053763139910521e-08, "loss": 0.06743710041046143, "step": 35730 }, { "epoch": 0.9414476337863534, "grad_norm": 4.2281622886657715, "learning_rate": 8.973133040973525e-08, "loss": 0.06681735515594482, "step": 35740 }, { "epoch": 0.9417110494645253, "grad_norm": 6.157707691192627, "learning_rate": 8.892860332575027e-08, "loss": 0.06295348405838012, "step": 35750 }, { "epoch": 0.9419744651426972, "grad_norm": 3.898946523666382, "learning_rate": 8.812945073140899e-08, "loss": 0.0693973183631897, "step": 35760 }, { "epoch": 0.9422378808208691, "grad_norm": 5.398693084716797, "learning_rate": 8.733387320837006e-08, "loss": 0.0783270239830017, "step": 35770 }, { "epoch": 0.942501296499041, "grad_norm": 6.057741641998291, "learning_rate": 8.654187133568803e-08, "loss": 0.055128300189971925, "step": 35780 }, { "epoch": 0.9427647121772129, "grad_norm": 6.9728522300720215, "learning_rate": 8.57534456898168e-08, "loss": 0.06548448801040649, "step": 35790 }, { "epoch": 0.9430281278553848, "grad_norm": 6.918217182159424, "learning_rate": 8.496859684460734e-08, "loss": 0.07425929307937622, "step": 35800 }, { "epoch": 0.9432915435335567, "grad_norm": 4.970094203948975, "learning_rate": 8.418732537130436e-08, "loss": 0.0640799880027771, "step": 35810 }, { "epoch": 0.9435549592117286, "grad_norm": 7.283233165740967, "learning_rate": 8.3409631838553e-08, "loss": 0.06104466915130615, "step": 35820 }, { "epoch": 0.9438183748899005, "grad_norm": 5.150959491729736, "learning_rate": 8.263551681239101e-08, "loss": 0.06836221218109131, "step": 35830 }, { "epoch": 0.9440817905680724, "grad_norm": 5.257936954498291, "learning_rate": 8.186498085625272e-08, "loss": 0.05514991879463196, "step": 35840 }, { "epoch": 0.9443452062462443, "grad_norm": 6.5023345947265625, "learning_rate": 8.109802453096838e-08, "loss": 0.0654185175895691, "step": 35850 }, { "epoch": 0.9446086219244162, "grad_norm": 7.183347225189209, "learning_rate": 8.033464839476034e-08, "loss": 0.0603692889213562, "step": 35860 }, { "epoch": 0.9448720376025881, "grad_norm": 5.698358535766602, "learning_rate": 7.957485300324807e-08, "loss": 0.0651313066482544, "step": 35870 }, { "epoch": 0.94513545328076, "grad_norm": 6.6603851318359375, "learning_rate": 7.881863890944363e-08, "loss": 0.06648149490356445, "step": 35880 }, { "epoch": 0.9453988689589319, "grad_norm": 6.251184940338135, "learning_rate": 7.806600666375064e-08, "loss": 0.06783372163772583, "step": 35890 }, { "epoch": 0.9456622846371038, "grad_norm": 3.9651734828948975, "learning_rate": 7.731695681396867e-08, "loss": 0.0656977653503418, "step": 35900 }, { "epoch": 0.9459257003152757, "grad_norm": 7.803674221038818, "learning_rate": 7.657148990528829e-08, "loss": 0.07707302570343018, "step": 35910 }, { "epoch": 0.9461891159934476, "grad_norm": 5.508210182189941, "learning_rate": 7.582960648029214e-08, "loss": 0.06935359835624695, "step": 35920 }, { "epoch": 0.9464525316716195, "grad_norm": 4.850083351135254, "learning_rate": 7.509130707895495e-08, "loss": 0.07225944995880126, "step": 35930 }, { "epoch": 0.9467159473497914, "grad_norm": 5.353855609893799, "learning_rate": 7.4356592238643e-08, "loss": 0.0640016496181488, "step": 35940 }, { "epoch": 0.9469793630279633, "grad_norm": 7.661026954650879, "learning_rate": 7.362546249411407e-08, "loss": 0.06703571677207946, "step": 35950 }, { "epoch": 0.9472427787061352, "grad_norm": 4.055479049682617, "learning_rate": 7.289791837751525e-08, "loss": 0.06159047484397888, "step": 35960 }, { "epoch": 0.947506194384307, "grad_norm": 6.63674783706665, "learning_rate": 7.217396041838465e-08, "loss": 0.060929661989212035, "step": 35970 }, { "epoch": 0.947769610062479, "grad_norm": 6.776385307312012, "learning_rate": 7.145358914365019e-08, "loss": 0.07157870531082153, "step": 35980 }, { "epoch": 0.9480330257406508, "grad_norm": 6.586204528808594, "learning_rate": 7.073680507762859e-08, "loss": 0.07006927728652954, "step": 35990 }, { "epoch": 0.9482964414188227, "grad_norm": 5.649869918823242, "learning_rate": 7.002360874202696e-08, "loss": 0.06366340517997741, "step": 36000 }, { "epoch": 0.9485598570969946, "grad_norm": 6.1436967849731445, "learning_rate": 6.931400065594118e-08, "loss": 0.056836044788360594, "step": 36010 }, { "epoch": 0.9488232727751665, "grad_norm": 6.760861396789551, "learning_rate": 6.8607981335852e-08, "loss": 0.06293433308601379, "step": 36020 }, { "epoch": 0.9490866884533384, "grad_norm": 5.060582637786865, "learning_rate": 6.790555129563337e-08, "loss": 0.06988099813461304, "step": 36030 }, { "epoch": 0.9493501041315103, "grad_norm": 7.822937488555908, "learning_rate": 6.720671104654298e-08, "loss": 0.06509079933166503, "step": 36040 }, { "epoch": 0.9496135198096822, "grad_norm": 6.351588726043701, "learning_rate": 6.651146109722728e-08, "loss": 0.07001382112503052, "step": 36050 }, { "epoch": 0.949876935487854, "grad_norm": 7.336408615112305, "learning_rate": 6.581980195371873e-08, "loss": 0.07185168862342835, "step": 36060 }, { "epoch": 0.9501403511660259, "grad_norm": 5.77691650390625, "learning_rate": 6.51317341194363e-08, "loss": 0.0632008969783783, "step": 36070 }, { "epoch": 0.9504037668441978, "grad_norm": 5.924773216247559, "learning_rate": 6.444725809518659e-08, "loss": 0.06829032897949219, "step": 36080 }, { "epoch": 0.9506671825223697, "grad_norm": 4.939152717590332, "learning_rate": 6.376637437916056e-08, "loss": 0.06032882928848267, "step": 36090 }, { "epoch": 0.9509305982005416, "grad_norm": 5.884328842163086, "learning_rate": 6.308908346693454e-08, "loss": 0.0655362844467163, "step": 36100 }, { "epoch": 0.9511940138787135, "grad_norm": 6.620552062988281, "learning_rate": 6.241538585146977e-08, "loss": 0.06128358840942383, "step": 36110 }, { "epoch": 0.9514574295568854, "grad_norm": 5.927768230438232, "learning_rate": 6.174528202311236e-08, "loss": 0.055311274528503415, "step": 36120 }, { "epoch": 0.9517208452350573, "grad_norm": 4.237771034240723, "learning_rate": 6.107877246959382e-08, "loss": 0.07580230236053467, "step": 36130 }, { "epoch": 0.9519842609132292, "grad_norm": 6.712155342102051, "learning_rate": 6.041585767602776e-08, "loss": 0.06970521211624145, "step": 36140 }, { "epoch": 0.9522476765914011, "grad_norm": 5.582298755645752, "learning_rate": 5.975653812491155e-08, "loss": 0.06623654365539551, "step": 36150 }, { "epoch": 0.952511092269573, "grad_norm": 4.315589904785156, "learning_rate": 5.9100814296127464e-08, "loss": 0.061502587795257566, "step": 36160 }, { "epoch": 0.9527745079477449, "grad_norm": 4.995179176330566, "learning_rate": 5.844868666693926e-08, "loss": 0.059338992834091185, "step": 36170 }, { "epoch": 0.9530379236259168, "grad_norm": 5.296846866607666, "learning_rate": 5.7800155711993933e-08, "loss": 0.059754788875579834, "step": 36180 }, { "epoch": 0.9533013393040887, "grad_norm": 5.251497268676758, "learning_rate": 5.715522190331946e-08, "loss": 0.05843801498413086, "step": 36190 }, { "epoch": 0.9535647549822606, "grad_norm": 6.376216411590576, "learning_rate": 5.651388571032701e-08, "loss": 0.05959095358848572, "step": 36200 }, { "epoch": 0.9538281706604325, "grad_norm": 4.961588382720947, "learning_rate": 5.587614759980875e-08, "loss": 0.06208651065826416, "step": 36210 }, { "epoch": 0.9540915863386044, "grad_norm": 5.687039375305176, "learning_rate": 5.524200803593782e-08, "loss": 0.06734711527824402, "step": 36220 }, { "epoch": 0.9543550020167763, "grad_norm": 6.862206935882568, "learning_rate": 5.4611467480268355e-08, "loss": 0.06925317049026489, "step": 36230 }, { "epoch": 0.9546184176949482, "grad_norm": 7.172873497009277, "learning_rate": 5.398452639173546e-08, "loss": 0.06839576959609986, "step": 36240 }, { "epoch": 0.95488183337312, "grad_norm": 6.741777420043945, "learning_rate": 5.336118522665246e-08, "loss": 0.06648638248443603, "step": 36250 }, { "epoch": 0.955145249051292, "grad_norm": 5.955129623413086, "learning_rate": 5.274144443871587e-08, "loss": 0.06339420676231385, "step": 36260 }, { "epoch": 0.9554086647294638, "grad_norm": 7.361164569854736, "learning_rate": 5.2125304478999325e-08, "loss": 0.06285771131515502, "step": 36270 }, { "epoch": 0.9556720804076357, "grad_norm": 5.189977645874023, "learning_rate": 5.151276579595466e-08, "loss": 0.05854572653770447, "step": 36280 }, { "epoch": 0.9559354960858076, "grad_norm": 5.914910316467285, "learning_rate": 5.0903828835415224e-08, "loss": 0.06665432453155518, "step": 36290 }, { "epoch": 0.9561989117639795, "grad_norm": 7.5871076583862305, "learning_rate": 5.02984940405915e-08, "loss": 0.07610466480255126, "step": 36300 }, { "epoch": 0.9564623274421514, "grad_norm": 3.9452733993530273, "learning_rate": 4.9696761852072176e-08, "loss": 0.07336453795433044, "step": 36310 }, { "epoch": 0.9567257431203233, "grad_norm": 5.721592426300049, "learning_rate": 4.9098632707823556e-08, "loss": 0.05400235056877136, "step": 36320 }, { "epoch": 0.9569891587984952, "grad_norm": 6.8557047843933105, "learning_rate": 4.8504107043190197e-08, "loss": 0.0736454427242279, "step": 36330 }, { "epoch": 0.9572525744766671, "grad_norm": 6.00373649597168, "learning_rate": 4.791318529089317e-08, "loss": 0.061535096168518065, "step": 36340 }, { "epoch": 0.957515990154839, "grad_norm": 6.460165977478027, "learning_rate": 4.7325867881031775e-08, "loss": 0.05872797966003418, "step": 36350 }, { "epoch": 0.9577794058330109, "grad_norm": 4.995926856994629, "learning_rate": 4.6742155241079615e-08, "loss": 0.07762036323547364, "step": 36360 }, { "epoch": 0.9580428215111828, "grad_norm": 5.245771884918213, "learning_rate": 4.616204779588851e-08, "loss": 0.07231931686401367, "step": 36370 }, { "epoch": 0.9583062371893547, "grad_norm": 4.425645351409912, "learning_rate": 4.558554596768627e-08, "loss": 0.0725597620010376, "step": 36380 }, { "epoch": 0.9585696528675266, "grad_norm": 4.790252685546875, "learning_rate": 4.501265017607504e-08, "loss": 0.0534113883972168, "step": 36390 }, { "epoch": 0.9588330685456985, "grad_norm": 4.243388652801514, "learning_rate": 4.4443360838033466e-08, "loss": 0.0653024137020111, "step": 36400 }, { "epoch": 0.9590964842238704, "grad_norm": 7.297702789306641, "learning_rate": 4.3877678367914565e-08, "loss": 0.07590036988258361, "step": 36410 }, { "epoch": 0.9593598999020423, "grad_norm": 4.119722843170166, "learning_rate": 4.331560317744621e-08, "loss": 0.07080004811286926, "step": 36420 }, { "epoch": 0.9596233155802142, "grad_norm": 7.442885398864746, "learning_rate": 4.275713567573225e-08, "loss": 0.06148940324783325, "step": 36430 }, { "epoch": 0.9598867312583861, "grad_norm": 4.233079433441162, "learning_rate": 4.220227626924811e-08, "loss": 0.06861129999160767, "step": 36440 }, { "epoch": 0.960150146936558, "grad_norm": 5.106351375579834, "learning_rate": 4.1651025361845174e-08, "loss": 0.06436707377433777, "step": 36450 }, { "epoch": 0.9604135626147299, "grad_norm": 5.281909465789795, "learning_rate": 4.110338335474695e-08, "loss": 0.05556044578552246, "step": 36460 }, { "epoch": 0.9606769782929018, "grad_norm": 7.1951189041137695, "learning_rate": 4.055935064655125e-08, "loss": 0.06234622597694397, "step": 36470 }, { "epoch": 0.9609403939710737, "grad_norm": 4.311394691467285, "learning_rate": 4.001892763322912e-08, "loss": 0.05872285962104797, "step": 36480 }, { "epoch": 0.9612038096492456, "grad_norm": 7.510067462921143, "learning_rate": 3.948211470812313e-08, "loss": 0.08302432894706727, "step": 36490 }, { "epoch": 0.9614672253274175, "grad_norm": 5.547579288482666, "learning_rate": 3.8948912261948524e-08, "loss": 0.06989257335662842, "step": 36500 }, { "epoch": 0.9617306410055894, "grad_norm": 5.397747039794922, "learning_rate": 3.8419320682793746e-08, "loss": 0.07505647540092468, "step": 36510 }, { "epoch": 0.9619940566837613, "grad_norm": 5.501282215118408, "learning_rate": 3.7893340356118244e-08, "loss": 0.06589398384094239, "step": 36520 }, { "epoch": 0.9622574723619332, "grad_norm": 5.8289337158203125, "learning_rate": 3.737097166475301e-08, "loss": 0.06514733433723449, "step": 36530 }, { "epoch": 0.962520888040105, "grad_norm": 6.158104419708252, "learning_rate": 3.685221498890057e-08, "loss": 0.07110238075256348, "step": 36540 }, { "epoch": 0.962784303718277, "grad_norm": 5.582512855529785, "learning_rate": 3.633707070613501e-08, "loss": 0.061816167831420896, "step": 36550 }, { "epoch": 0.9630477193964488, "grad_norm": 5.798737525939941, "learning_rate": 3.582553919139919e-08, "loss": 0.06852765679359436, "step": 36560 }, { "epoch": 0.9633111350746207, "grad_norm": 4.159910202026367, "learning_rate": 3.531762081700918e-08, "loss": 0.05833841562271118, "step": 36570 }, { "epoch": 0.9635745507527926, "grad_norm": 4.691751480102539, "learning_rate": 3.4813315952649254e-08, "loss": 0.06432532072067261, "step": 36580 }, { "epoch": 0.9638379664309645, "grad_norm": 6.653680324554443, "learning_rate": 3.431262496537413e-08, "loss": 0.07285019159317016, "step": 36590 }, { "epoch": 0.9641013821091364, "grad_norm": 5.338353157043457, "learning_rate": 3.381554821960842e-08, "loss": 0.07145187854766846, "step": 36600 }, { "epoch": 0.9643647977873083, "grad_norm": 8.066006660461426, "learning_rate": 3.3322086077146596e-08, "loss": 0.0750081181526184, "step": 36610 }, { "epoch": 0.9646282134654802, "grad_norm": 6.212028980255127, "learning_rate": 3.283223889715082e-08, "loss": 0.059063839912414554, "step": 36620 }, { "epoch": 0.9648916291436521, "grad_norm": 6.701766014099121, "learning_rate": 3.2346007036153645e-08, "loss": 0.060190731287002565, "step": 36630 }, { "epoch": 0.965155044821824, "grad_norm": 7.734141826629639, "learning_rate": 3.186339084805534e-08, "loss": 0.05597991347312927, "step": 36640 }, { "epoch": 0.9654184604999959, "grad_norm": 3.2360527515411377, "learning_rate": 3.138439068412491e-08, "loss": 0.06286150813102723, "step": 36650 }, { "epoch": 0.9656818761781678, "grad_norm": 4.843702793121338, "learning_rate": 3.090900689299903e-08, "loss": 0.06770287156105041, "step": 36660 }, { "epoch": 0.9659452918563397, "grad_norm": 7.978590488433838, "learning_rate": 3.043723982068258e-08, "loss": 0.06856831312179565, "step": 36670 }, { "epoch": 0.9662087075345116, "grad_norm": 7.180423736572266, "learning_rate": 2.9969089810548135e-08, "loss": 0.06120140552520752, "step": 36680 }, { "epoch": 0.9664721232126835, "grad_norm": 5.826457500457764, "learning_rate": 2.9504557203335893e-08, "loss": 0.07044256925582885, "step": 36690 }, { "epoch": 0.9667355388908554, "grad_norm": 7.430706977844238, "learning_rate": 2.9043642337152068e-08, "loss": 0.06648545265197754, "step": 36700 }, { "epoch": 0.9669989545690273, "grad_norm": 5.590300559997559, "learning_rate": 2.8586345547471083e-08, "loss": 0.06817799806594849, "step": 36710 }, { "epoch": 0.9672623702471992, "grad_norm": 6.8960700035095215, "learning_rate": 2.8132667167132254e-08, "loss": 0.06377623677253723, "step": 36720 }, { "epoch": 0.9675257859253711, "grad_norm": 3.867990016937256, "learning_rate": 2.7682607526343108e-08, "loss": 0.052723240852355954, "step": 36730 }, { "epoch": 0.967789201603543, "grad_norm": 6.30410099029541, "learning_rate": 2.7236166952676058e-08, "loss": 0.06676490902900696, "step": 36740 }, { "epoch": 0.9680526172817149, "grad_norm": 5.931967735290527, "learning_rate": 2.6793345771070068e-08, "loss": 0.06996378898620606, "step": 36750 }, { "epoch": 0.9683160329598868, "grad_norm": 5.073814392089844, "learning_rate": 2.6354144303829542e-08, "loss": 0.0552100419998169, "step": 36760 }, { "epoch": 0.9685794486380587, "grad_norm": 8.039228439331055, "learning_rate": 2.5918562870623776e-08, "loss": 0.07394251823425294, "step": 36770 }, { "epoch": 0.9688428643162306, "grad_norm": 5.217129707336426, "learning_rate": 2.5486601788488052e-08, "loss": 0.05521014928817749, "step": 36780 }, { "epoch": 0.9691062799944025, "grad_norm": 6.591895580291748, "learning_rate": 2.5058261371822545e-08, "loss": 0.06319929361343384, "step": 36790 }, { "epoch": 0.9693696956725744, "grad_norm": 7.434736251831055, "learning_rate": 2.463354193239176e-08, "loss": 0.07098909616470336, "step": 36800 }, { "epoch": 0.9696331113507463, "grad_norm": 5.633517742156982, "learning_rate": 2.4212443779324523e-08, "loss": 0.06044158935546875, "step": 36810 }, { "epoch": 0.9698965270289182, "grad_norm": 6.287377834320068, "learning_rate": 2.3794967219114008e-08, "loss": 0.06678688526153564, "step": 36820 }, { "epoch": 0.97015994270709, "grad_norm": 6.220147609710693, "learning_rate": 2.338111255561881e-08, "loss": 0.08154328465461731, "step": 36830 }, { "epoch": 0.970423358385262, "grad_norm": 5.24041748046875, "learning_rate": 2.29708800900591e-08, "loss": 0.06472904086112977, "step": 36840 }, { "epoch": 0.9706867740634337, "grad_norm": 6.759183883666992, "learning_rate": 2.2564270121019915e-08, "loss": 0.07327622175216675, "step": 36850 }, { "epoch": 0.9709501897416056, "grad_norm": 6.574487209320068, "learning_rate": 2.216128294445008e-08, "loss": 0.07475240230560302, "step": 36860 }, { "epoch": 0.9712136054197775, "grad_norm": 6.2432966232299805, "learning_rate": 2.176191885366108e-08, "loss": 0.062034893035888675, "step": 36870 }, { "epoch": 0.9714770210979494, "grad_norm": 5.475354194641113, "learning_rate": 2.136617813932651e-08, "loss": 0.058783209323883055, "step": 36880 }, { "epoch": 0.9717404367761213, "grad_norm": 7.834117889404297, "learning_rate": 2.09740610894843e-08, "loss": 0.07178152799606323, "step": 36890 }, { "epoch": 0.9720038524542932, "grad_norm": 3.419811964035034, "learning_rate": 2.0585567989533374e-08, "loss": 0.057636797428131104, "step": 36900 }, { "epoch": 0.9722672681324651, "grad_norm": 6.313663005828857, "learning_rate": 2.020069912223699e-08, "loss": 0.062239301204681394, "step": 36910 }, { "epoch": 0.972530683810637, "grad_norm": 3.9548115730285645, "learning_rate": 1.9819454767718294e-08, "loss": 0.07435002326965331, "step": 36920 }, { "epoch": 0.9727940994888089, "grad_norm": 6.037209987640381, "learning_rate": 1.94418352034631e-08, "loss": 0.05673007965087891, "step": 36930 }, { "epoch": 0.9730575151669808, "grad_norm": 7.650024890899658, "learning_rate": 1.906784070432044e-08, "loss": 0.07393332719802856, "step": 36940 }, { "epoch": 0.9733209308451527, "grad_norm": 6.945317268371582, "learning_rate": 1.869747154249868e-08, "loss": 0.08201557397842407, "step": 36950 }, { "epoch": 0.9735843465233246, "grad_norm": 7.263277053833008, "learning_rate": 1.833072798756885e-08, "loss": 0.07076058387756348, "step": 36960 }, { "epoch": 0.9738477622014965, "grad_norm": 6.328479766845703, "learning_rate": 1.796761030646299e-08, "loss": 0.0579103946685791, "step": 36970 }, { "epoch": 0.9741111778796684, "grad_norm": 4.988516807556152, "learning_rate": 1.7608118763473014e-08, "loss": 0.056540220975875854, "step": 36980 }, { "epoch": 0.9743745935578403, "grad_norm": 3.6467700004577637, "learning_rate": 1.7252253620253513e-08, "loss": 0.06294828653335571, "step": 36990 }, { "epoch": 0.9746380092360122, "grad_norm": 9.195568084716797, "learning_rate": 1.6900015135817293e-08, "loss": 0.0745604932308197, "step": 37000 }, { "epoch": 0.9749014249141841, "grad_norm": 7.57575798034668, "learning_rate": 1.655140356653928e-08, "loss": 0.06327281594276428, "step": 37010 }, { "epoch": 0.975164840592356, "grad_norm": 4.991515159606934, "learning_rate": 1.620641916615373e-08, "loss": 0.06538687944412232, "step": 37020 }, { "epoch": 0.9754282562705279, "grad_norm": 4.734572410583496, "learning_rate": 1.5865062185755897e-08, "loss": 0.06043705940246582, "step": 37030 }, { "epoch": 0.9756916719486998, "grad_norm": 7.013640403747559, "learning_rate": 1.5527332873799818e-08, "loss": 0.06838288307189941, "step": 37040 }, { "epoch": 0.9759550876268717, "grad_norm": 6.396208763122559, "learning_rate": 1.5193231476098856e-08, "loss": 0.07199996709823608, "step": 37050 }, { "epoch": 0.9762185033050436, "grad_norm": 6.1417341232299805, "learning_rate": 1.4862758235826835e-08, "loss": 0.07055475711822509, "step": 37060 }, { "epoch": 0.9764819189832155, "grad_norm": 6.6729960441589355, "learning_rate": 1.4535913393516898e-08, "loss": 0.05867383480072021, "step": 37070 }, { "epoch": 0.9767453346613874, "grad_norm": 7.775796413421631, "learning_rate": 1.4212697187060421e-08, "loss": 0.07369301319122315, "step": 37080 }, { "epoch": 0.9770087503395593, "grad_norm": 6.948468208312988, "learning_rate": 1.3893109851707554e-08, "loss": 0.06544946432113648, "step": 37090 }, { "epoch": 0.9772721660177311, "grad_norm": 5.273996353149414, "learning_rate": 1.3577151620068896e-08, "loss": 0.05814324021339416, "step": 37100 }, { "epoch": 0.977535581695903, "grad_norm": 5.256224632263184, "learning_rate": 1.3264822722111603e-08, "loss": 0.07065773606300355, "step": 37110 }, { "epoch": 0.9777989973740749, "grad_norm": 6.111777305603027, "learning_rate": 1.2956123385162166e-08, "loss": 0.0669223427772522, "step": 37120 }, { "epoch": 0.9780624130522468, "grad_norm": 5.362745761871338, "learning_rate": 1.2651053833905858e-08, "loss": 0.06799451112747193, "step": 37130 }, { "epoch": 0.9783258287304187, "grad_norm": 5.621744155883789, "learning_rate": 1.2349614290384504e-08, "loss": 0.07373179197311401, "step": 37140 }, { "epoch": 0.9785892444085906, "grad_norm": 5.071153163909912, "learning_rate": 1.2051804973999826e-08, "loss": 0.07245575785636901, "step": 37150 }, { "epoch": 0.9788526600867625, "grad_norm": 6.472638130187988, "learning_rate": 1.1757626101509545e-08, "loss": 0.05747482180595398, "step": 37160 }, { "epoch": 0.9791160757649344, "grad_norm": 5.421839714050293, "learning_rate": 1.146707788703072e-08, "loss": 0.06459977626800537, "step": 37170 }, { "epoch": 0.9793794914431063, "grad_norm": 4.624818325042725, "learning_rate": 1.1180160542035856e-08, "loss": 0.05836014747619629, "step": 37180 }, { "epoch": 0.9796429071212782, "grad_norm": 6.494551658630371, "learning_rate": 1.0896874275355685e-08, "loss": 0.06806309223175049, "step": 37190 }, { "epoch": 0.9799063227994501, "grad_norm": 5.821423530578613, "learning_rate": 1.061721929317916e-08, "loss": 0.06131135225296021, "step": 37200 }, { "epoch": 0.980169738477622, "grad_norm": 6.766383171081543, "learning_rate": 1.0341195799050685e-08, "loss": 0.06572745442390442, "step": 37210 }, { "epoch": 0.9804331541557939, "grad_norm": 7.693059921264648, "learning_rate": 1.0068803993871779e-08, "loss": 0.06990413665771485, "step": 37220 }, { "epoch": 0.9806965698339658, "grad_norm": 7.0911760330200195, "learning_rate": 9.800044075901627e-09, "loss": 0.07493871450424194, "step": 37230 }, { "epoch": 0.9809599855121377, "grad_norm": 6.55902099609375, "learning_rate": 9.534916240754865e-09, "loss": 0.06789945363998413, "step": 37240 }, { "epoch": 0.9812234011903096, "grad_norm": 5.324575901031494, "learning_rate": 9.27342068140269e-09, "loss": 0.04862588047981262, "step": 37250 }, { "epoch": 0.9814868168684815, "grad_norm": 8.053727149963379, "learning_rate": 9.015557588173407e-09, "loss": 0.073375403881073, "step": 37260 }, { "epoch": 0.9817502325466534, "grad_norm": 5.141667366027832, "learning_rate": 8.76132714875022e-09, "loss": 0.06945719122886658, "step": 37270 }, { "epoch": 0.9820136482248253, "grad_norm": 7.080410480499268, "learning_rate": 8.510729548174002e-09, "loss": 0.06752517223358154, "step": 37280 }, { "epoch": 0.9822770639029972, "grad_norm": 4.0047149658203125, "learning_rate": 8.26376496883996e-09, "loss": 0.07019937634468079, "step": 37290 }, { "epoch": 0.9825404795811691, "grad_norm": 4.442424297332764, "learning_rate": 8.020433590499865e-09, "loss": 0.06368646025657654, "step": 37300 }, { "epoch": 0.982803895259341, "grad_norm": 5.984375476837158, "learning_rate": 7.780735590259825e-09, "loss": 0.060888683795928954, "step": 37310 }, { "epoch": 0.9830673109375129, "grad_norm": 6.8336968421936035, "learning_rate": 7.544671142583615e-09, "loss": 0.06479942798614502, "step": 37320 }, { "epoch": 0.9833307266156848, "grad_norm": 3.9791646003723145, "learning_rate": 7.3122404192882415e-09, "loss": 0.07301832437515259, "step": 37330 }, { "epoch": 0.9835941422938567, "grad_norm": 6.356592178344727, "learning_rate": 7.083443589547267e-09, "loss": 0.058867454528808594, "step": 37340 }, { "epoch": 0.9838575579720286, "grad_norm": 6.538567066192627, "learning_rate": 6.858280819888596e-09, "loss": 0.05941101312637329, "step": 37350 }, { "epoch": 0.9841209736502005, "grad_norm": 5.761967658996582, "learning_rate": 6.6367522741961345e-09, "loss": 0.05589503049850464, "step": 37360 }, { "epoch": 0.9843843893283724, "grad_norm": 3.6406936645507812, "learning_rate": 6.418858113707016e-09, "loss": 0.06752811670303345, "step": 37370 }, { "epoch": 0.9846478050065443, "grad_norm": 5.097436428070068, "learning_rate": 6.204598497013825e-09, "loss": 0.0535605251789093, "step": 37380 }, { "epoch": 0.9849112206847161, "grad_norm": 4.746931552886963, "learning_rate": 5.993973580065149e-09, "loss": 0.06417444944381714, "step": 37390 }, { "epoch": 0.985174636362888, "grad_norm": 6.94158935546875, "learning_rate": 5.786983516162248e-09, "loss": 0.06256886720657348, "step": 37400 }, { "epoch": 0.9854380520410599, "grad_norm": 5.345177173614502, "learning_rate": 5.583628455961276e-09, "loss": 0.06276562809944153, "step": 37410 }, { "epoch": 0.9857014677192318, "grad_norm": 4.337769031524658, "learning_rate": 5.383908547472727e-09, "loss": 0.06293184757232666, "step": 37420 }, { "epoch": 0.9859648833974037, "grad_norm": 4.205045700073242, "learning_rate": 5.187823936061987e-09, "loss": 0.06038523912429809, "step": 37430 }, { "epoch": 0.9862282990755756, "grad_norm": 6.68577241897583, "learning_rate": 4.995374764447114e-09, "loss": 0.06390358209609985, "step": 37440 }, { "epoch": 0.9864917147537475, "grad_norm": 7.047734260559082, "learning_rate": 4.806561172701618e-09, "loss": 0.06922236680984498, "step": 37450 }, { "epoch": 0.9867551304319194, "grad_norm": 5.525099277496338, "learning_rate": 4.62138329825168e-09, "loss": 0.06486598253250123, "step": 37460 }, { "epoch": 0.9870185461100913, "grad_norm": 7.6596808433532715, "learning_rate": 4.439841275877821e-09, "loss": 0.07299812436103821, "step": 37470 }, { "epoch": 0.9872819617882632, "grad_norm": 5.138876438140869, "learning_rate": 4.2619352377148984e-09, "loss": 0.05927410125732422, "step": 37480 }, { "epoch": 0.9875453774664351, "grad_norm": 6.925347328186035, "learning_rate": 4.087665313249334e-09, "loss": 0.0716974139213562, "step": 37490 }, { "epoch": 0.987808793144607, "grad_norm": 3.6451642513275146, "learning_rate": 3.917031629322998e-09, "loss": 0.06295325756072997, "step": 37500 }, { "epoch": 0.9880722088227789, "grad_norm": 4.7704644203186035, "learning_rate": 3.7500343101304345e-09, "loss": 0.07546420097351074, "step": 37510 }, { "epoch": 0.9883356245009508, "grad_norm": 6.574708938598633, "learning_rate": 3.5866734772194156e-09, "loss": 0.08265715837478638, "step": 37520 }, { "epoch": 0.9885990401791227, "grad_norm": 6.622854709625244, "learning_rate": 3.426949249491496e-09, "loss": 0.08091981410980224, "step": 37530 }, { "epoch": 0.9888624558572946, "grad_norm": 8.757187843322754, "learning_rate": 3.2708617431997934e-09, "loss": 0.06746656894683838, "step": 37540 }, { "epoch": 0.9891258715354665, "grad_norm": 5.516408443450928, "learning_rate": 3.118411071952321e-09, "loss": 0.06874260902404786, "step": 37550 }, { "epoch": 0.9893892872136384, "grad_norm": 6.3855743408203125, "learning_rate": 2.9695973467092077e-09, "loss": 0.0637604057788849, "step": 37560 }, { "epoch": 0.9896527028918103, "grad_norm": 4.424450397491455, "learning_rate": 2.824420675783257e-09, "loss": 0.05262402892112732, "step": 37570 }, { "epoch": 0.9899161185699822, "grad_norm": 5.775226593017578, "learning_rate": 2.682881164840501e-09, "loss": 0.06545880436897278, "step": 37580 }, { "epoch": 0.9901795342481541, "grad_norm": 8.472465515136719, "learning_rate": 2.544978916899088e-09, "loss": 0.06493521928787231, "step": 37590 }, { "epoch": 0.990442949926326, "grad_norm": 4.754128932952881, "learning_rate": 2.4107140323309518e-09, "loss": 0.06019158959388733, "step": 37600 }, { "epoch": 0.9907063656044979, "grad_norm": 8.50888442993164, "learning_rate": 2.2800866088590335e-09, "loss": 0.06392685174942017, "step": 37610 }, { "epoch": 0.9909697812826698, "grad_norm": 4.668750762939453, "learning_rate": 2.1530967415606118e-09, "loss": 0.06621480584144593, "step": 37620 }, { "epoch": 0.9912331969608417, "grad_norm": 5.83742618560791, "learning_rate": 2.0297445228634192e-09, "loss": 0.06681252717971801, "step": 37630 }, { "epoch": 0.9914966126390136, "grad_norm": 8.721904754638672, "learning_rate": 1.9100300425489716e-09, "loss": 0.07057573795318603, "step": 37640 }, { "epoch": 0.9917600283171853, "grad_norm": 3.8988401889801025, "learning_rate": 1.793953387750902e-09, "loss": 0.062050366401672365, "step": 37650 }, { "epoch": 0.9920234439953572, "grad_norm": 4.9478373527526855, "learning_rate": 1.681514642953852e-09, "loss": 0.0731929898262024, "step": 37660 }, { "epoch": 0.9922868596735291, "grad_norm": 4.374867916107178, "learning_rate": 1.572713889996802e-09, "loss": 0.06485964059829712, "step": 37670 }, { "epoch": 0.992550275351701, "grad_norm": 4.319451332092285, "learning_rate": 1.4675512080691844e-09, "loss": 0.063744056224823, "step": 37680 }, { "epoch": 0.9928136910298729, "grad_norm": 6.559131622314453, "learning_rate": 1.3660266737125505e-09, "loss": 0.07417099475860596, "step": 37690 }, { "epoch": 0.9930771067080448, "grad_norm": 5.525753498077393, "learning_rate": 1.268140360821124e-09, "loss": 0.05865771174430847, "step": 37700 }, { "epoch": 0.9933405223862167, "grad_norm": 5.652621746063232, "learning_rate": 1.1738923406406922e-09, "loss": 0.06592180728912353, "step": 37710 }, { "epoch": 0.9936039380643886, "grad_norm": 7.458366870880127, "learning_rate": 1.08328268176916e-09, "loss": 0.06608673930168152, "step": 37720 }, { "epoch": 0.9938673537425605, "grad_norm": 8.86489486694336, "learning_rate": 9.963114501559956e-10, "loss": 0.07391177415847779, "step": 37730 }, { "epoch": 0.9941307694207324, "grad_norm": 4.826699733734131, "learning_rate": 9.129787091027853e-10, "loss": 0.06710981130599976, "step": 37740 }, { "epoch": 0.9943941850989043, "grad_norm": 5.1160736083984375, "learning_rate": 8.332845192626782e-10, "loss": 0.0677121639251709, "step": 37750 }, { "epoch": 0.9946576007770762, "grad_norm": 5.0544281005859375, "learning_rate": 7.572289386403864e-10, "loss": 0.06217951774597168, "step": 37760 }, { "epoch": 0.9949210164552481, "grad_norm": 7.326071262359619, "learning_rate": 6.84812022592185e-10, "loss": 0.06091086864471436, "step": 37770 }, { "epoch": 0.99518443213342, "grad_norm": 6.265317916870117, "learning_rate": 6.160338238270225e-10, "loss": 0.0683281421661377, "step": 37780 }, { "epoch": 0.9954478478115919, "grad_norm": 7.3858489990234375, "learning_rate": 5.508943924042998e-10, "loss": 0.06833511590957642, "step": 37790 }, { "epoch": 0.9957112634897638, "grad_norm": 6.896548748016357, "learning_rate": 4.89393775734981e-10, "loss": 0.06411347389221192, "step": 37800 }, { "epoch": 0.9959746791679357, "grad_norm": 4.397863864898682, "learning_rate": 4.3153201858214813e-10, "loss": 0.056560182571411134, "step": 37810 }, { "epoch": 0.9962380948461076, "grad_norm": 6.754597187042236, "learning_rate": 3.773091630598913e-10, "loss": 0.07671601176261902, "step": 37820 }, { "epoch": 0.9965015105242795, "grad_norm": 9.075762748718262, "learning_rate": 3.267252486344186e-10, "loss": 0.06589534878730774, "step": 37830 }, { "epoch": 0.9967649262024514, "grad_norm": 5.408308506011963, "learning_rate": 2.7978031212239074e-10, "loss": 0.062205415964126584, "step": 37840 }, { "epoch": 0.9970283418806233, "grad_norm": 4.617432594299316, "learning_rate": 2.364743876925868e-10, "loss": 0.0685584306716919, "step": 37850 }, { "epoch": 0.9972917575587952, "grad_norm": 7.057038307189941, "learning_rate": 1.9680750686479343e-10, "loss": 0.0691515326499939, "step": 37860 }, { "epoch": 0.9975551732369671, "grad_norm": 4.538537979125977, "learning_rate": 1.6077969851036047e-10, "loss": 0.06360369920730591, "step": 37870 }, { "epoch": 0.997818588915139, "grad_norm": 5.838778972625732, "learning_rate": 1.2839098885164547e-10, "loss": 0.07551488876342774, "step": 37880 }, { "epoch": 0.9980820045933109, "grad_norm": 4.130828380584717, "learning_rate": 9.964140146312417e-11, "loss": 0.05819079875946045, "step": 37890 }, { "epoch": 0.9983454202714828, "grad_norm": 5.411816120147705, "learning_rate": 7.453095726916993e-11, "loss": 0.05652863979339599, "step": 37900 }, { "epoch": 0.9986088359496547, "grad_norm": 5.820533275604248, "learning_rate": 5.305967454682925e-11, "loss": 0.06711377501487732, "step": 37910 }, { "epoch": 0.9988722516278266, "grad_norm": 6.046236991882324, "learning_rate": 3.5227568924156575e-11, "loss": 0.06704609394073487, "step": 37920 }, { "epoch": 0.9991356673059985, "grad_norm": 4.961540699005127, "learning_rate": 2.103465337910393e-11, "loss": 0.05933867692947388, "step": 37930 }, { "epoch": 0.9993990829841704, "grad_norm": 6.669266700744629, "learning_rate": 1.0480938242851679e-11, "loss": 0.06851111650466919, "step": 37940 }, { "epoch": 0.9996624986623422, "grad_norm": 6.282093048095703, "learning_rate": 3.5664311964778153e-12, "loss": 0.06631701588630676, "step": 37950 }, { "epoch": 0.9999259143405141, "grad_norm": 5.239515781402588, "learning_rate": 2.911372720681982e-13, "loss": 0.06399661898612977, "step": 37960 }, { "epoch": 1.0, "step": 37963, "total_flos": 1.3554249512862823e+20, "train_loss": 0.03804711307403464, "train_runtime": 205187.7644, "train_samples_per_second": 11.841, "train_steps_per_second": 0.185 } ], "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": 1.3554249512862823e+20, "train_batch_size": 1, "trial_name": null, "trial_params": null }