{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.082139216144577, "eval_steps": 500, "global_step": 272000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00019892283286006277, "grad_norm": 0.5478392243385315, "learning_rate": 1.9998050558176915e-05, "loss": 2.8825897216796874, "step": 50 }, { "epoch": 0.00039784566572012553, "grad_norm": 0.49451741576194763, "learning_rate": 1.9996061331826827e-05, "loss": 2.818773193359375, "step": 100 }, { "epoch": 0.0005967684985801883, "grad_norm": 0.47904306650161743, "learning_rate": 1.999407210547674e-05, "loss": 2.7205413818359374, "step": 150 }, { "epoch": 0.0007956913314402511, "grad_norm": 0.5007067918777466, "learning_rate": 1.999208287912665e-05, "loss": 2.67067626953125, "step": 200 }, { "epoch": 0.0009946141643003138, "grad_norm": 0.49141111969947815, "learning_rate": 1.9990093652776564e-05, "loss": 2.671988525390625, "step": 250 }, { "epoch": 0.0011935369971603767, "grad_norm": 0.5031938552856445, "learning_rate": 1.9988104426426476e-05, "loss": 2.6896957397460937, "step": 300 }, { "epoch": 0.0013924598300204393, "grad_norm": 0.5243322253227234, "learning_rate": 1.9986115200076388e-05, "loss": 2.678563537597656, "step": 350 }, { "epoch": 0.0015913826628805021, "grad_norm": 0.5135937929153442, "learning_rate": 1.99841259737263e-05, "loss": 2.675412902832031, "step": 400 }, { "epoch": 0.0017903054957405648, "grad_norm": 0.4568680226802826, "learning_rate": 1.9982136747376212e-05, "loss": 2.653265380859375, "step": 450 }, { "epoch": 0.0019892283286006276, "grad_norm": 0.5060641765594482, "learning_rate": 1.9980147521026124e-05, "loss": 2.64932861328125, "step": 500 }, { "epoch": 0.0021881511614606905, "grad_norm": 0.55921471118927, "learning_rate": 1.9978158294676037e-05, "loss": 2.6767303466796877, "step": 550 }, { "epoch": 0.0023870739943207533, "grad_norm": 0.4758363366127014, "learning_rate": 1.997616906832595e-05, "loss": 2.6173846435546877, "step": 600 }, { "epoch": 0.0025859968271808157, "grad_norm": 0.46849697828292847, "learning_rate": 1.997417984197586e-05, "loss": 2.65154052734375, "step": 650 }, { "epoch": 0.0027849196600408786, "grad_norm": 0.4179868698120117, "learning_rate": 1.9972190615625773e-05, "loss": 2.6094656372070313, "step": 700 }, { "epoch": 0.0029838424929009414, "grad_norm": 0.5291228890419006, "learning_rate": 1.9970201389275685e-05, "loss": 2.633155517578125, "step": 750 }, { "epoch": 0.0031827653257610043, "grad_norm": 0.4421975910663605, "learning_rate": 1.9968212162925597e-05, "loss": 2.5792889404296875, "step": 800 }, { "epoch": 0.003381688158621067, "grad_norm": 0.48195788264274597, "learning_rate": 1.996622293657551e-05, "loss": 2.5837698364257813, "step": 850 }, { "epoch": 0.0035806109914811295, "grad_norm": 0.524206280708313, "learning_rate": 1.996423371022542e-05, "loss": 2.540571594238281, "step": 900 }, { "epoch": 0.0037795338243411924, "grad_norm": 0.46174174547195435, "learning_rate": 1.9962244483875334e-05, "loss": 2.5253761291503904, "step": 950 }, { "epoch": 0.003978456657201255, "grad_norm": 0.42715102434158325, "learning_rate": 1.9960255257525246e-05, "loss": 2.569144592285156, "step": 1000 }, { "epoch": 0.004177379490061318, "grad_norm": 0.5518872141838074, "learning_rate": 1.9958266031175158e-05, "loss": 2.59046142578125, "step": 1050 }, { "epoch": 0.004376302322921381, "grad_norm": 0.4533003866672516, "learning_rate": 1.995627680482507e-05, "loss": 2.5992257690429685, "step": 1100 }, { "epoch": 0.004575225155781444, "grad_norm": 0.5420551300048828, "learning_rate": 1.995428757847498e-05, "loss": 2.6003927612304687, "step": 1150 }, { "epoch": 0.004774147988641507, "grad_norm": 0.44064828753471375, "learning_rate": 1.995229835212489e-05, "loss": 2.5384942626953126, "step": 1200 }, { "epoch": 0.004973070821501569, "grad_norm": 0.5187183618545532, "learning_rate": 1.9950309125774803e-05, "loss": 2.5839990234375, "step": 1250 }, { "epoch": 0.0051719936543616314, "grad_norm": 0.46409159898757935, "learning_rate": 1.994831989942472e-05, "loss": 2.546786041259766, "step": 1300 }, { "epoch": 0.005370916487221694, "grad_norm": 0.4641093611717224, "learning_rate": 1.994633067307463e-05, "loss": 2.546092834472656, "step": 1350 }, { "epoch": 0.005569839320081757, "grad_norm": 0.4371015131473541, "learning_rate": 1.9944341446724543e-05, "loss": 2.654139404296875, "step": 1400 }, { "epoch": 0.00576876215294182, "grad_norm": 0.4346758723258972, "learning_rate": 1.9942352220374455e-05, "loss": 2.5233253479003905, "step": 1450 }, { "epoch": 0.005967684985801883, "grad_norm": 0.47130659222602844, "learning_rate": 1.9940362994024364e-05, "loss": 2.516582183837891, "step": 1500 }, { "epoch": 0.006166607818661946, "grad_norm": 0.4723242521286011, "learning_rate": 1.9938373767674276e-05, "loss": 2.5500637817382814, "step": 1550 }, { "epoch": 0.0063655306515220085, "grad_norm": 0.48338782787323, "learning_rate": 1.9936384541324188e-05, "loss": 2.54703369140625, "step": 1600 }, { "epoch": 0.006564453484382071, "grad_norm": 0.5357130169868469, "learning_rate": 1.99343953149741e-05, "loss": 2.529511260986328, "step": 1650 }, { "epoch": 0.006763376317242134, "grad_norm": 0.504641592502594, "learning_rate": 1.9932406088624016e-05, "loss": 2.5232208251953123, "step": 1700 }, { "epoch": 0.006962299150102196, "grad_norm": 0.5541943311691284, "learning_rate": 1.9930416862273928e-05, "loss": 2.4893238830566404, "step": 1750 }, { "epoch": 0.007161221982962259, "grad_norm": 0.4146322011947632, "learning_rate": 1.9928427635923837e-05, "loss": 2.541248321533203, "step": 1800 }, { "epoch": 0.007360144815822322, "grad_norm": 0.4430202543735504, "learning_rate": 1.992643840957375e-05, "loss": 2.567982482910156, "step": 1850 }, { "epoch": 0.007559067648682385, "grad_norm": 0.45610979199409485, "learning_rate": 1.992444918322366e-05, "loss": 2.4886514282226564, "step": 1900 }, { "epoch": 0.007757990481542448, "grad_norm": 0.5374492406845093, "learning_rate": 1.9922459956873573e-05, "loss": 2.4875096130371093, "step": 1950 }, { "epoch": 0.00795691331440251, "grad_norm": 0.45553284883499146, "learning_rate": 1.9920470730523485e-05, "loss": 2.4158392333984375, "step": 2000 }, { "epoch": 0.008155836147262573, "grad_norm": 0.4749634563922882, "learning_rate": 1.9918481504173398e-05, "loss": 2.586251220703125, "step": 2050 }, { "epoch": 0.008354758980122636, "grad_norm": 0.5085834264755249, "learning_rate": 1.9916492277823313e-05, "loss": 2.5001405334472655, "step": 2100 }, { "epoch": 0.008553681812982699, "grad_norm": 0.44805675745010376, "learning_rate": 1.9914503051473222e-05, "loss": 2.5267015075683594, "step": 2150 }, { "epoch": 0.008752604645842762, "grad_norm": 0.4947705864906311, "learning_rate": 1.9912513825123134e-05, "loss": 2.5229624938964843, "step": 2200 }, { "epoch": 0.008951527478702825, "grad_norm": 0.5817417502403259, "learning_rate": 1.9910524598773046e-05, "loss": 2.513974914550781, "step": 2250 }, { "epoch": 0.009150450311562888, "grad_norm": 0.4778134822845459, "learning_rate": 1.990853537242296e-05, "loss": 2.5188621520996093, "step": 2300 }, { "epoch": 0.00934937314442295, "grad_norm": 0.5550323724746704, "learning_rate": 1.990654614607287e-05, "loss": 2.5357632446289062, "step": 2350 }, { "epoch": 0.009548295977283013, "grad_norm": 0.4921201765537262, "learning_rate": 1.9904556919722783e-05, "loss": 2.512112579345703, "step": 2400 }, { "epoch": 0.009747218810143076, "grad_norm": 0.45680075883865356, "learning_rate": 1.9902567693372695e-05, "loss": 2.4851231384277344, "step": 2450 }, { "epoch": 0.009946141643003137, "grad_norm": 0.4718796908855438, "learning_rate": 1.9900578467022607e-05, "loss": 2.5513980102539064, "step": 2500 }, { "epoch": 0.0101450644758632, "grad_norm": 0.4222698211669922, "learning_rate": 1.989858924067252e-05, "loss": 2.5116835021972657, "step": 2550 }, { "epoch": 0.010343987308723263, "grad_norm": 0.46504679322242737, "learning_rate": 1.989660001432243e-05, "loss": 2.488034210205078, "step": 2600 }, { "epoch": 0.010542910141583326, "grad_norm": 0.4974987804889679, "learning_rate": 1.9894610787972343e-05, "loss": 2.5413035583496093, "step": 2650 }, { "epoch": 0.010741832974443389, "grad_norm": 0.4955303966999054, "learning_rate": 1.9892621561622255e-05, "loss": 2.4977728271484376, "step": 2700 }, { "epoch": 0.010940755807303451, "grad_norm": 0.4842558801174164, "learning_rate": 1.9890632335272168e-05, "loss": 2.4372438049316405, "step": 2750 }, { "epoch": 0.011139678640163514, "grad_norm": 0.5429825186729431, "learning_rate": 1.988864310892208e-05, "loss": 2.4618882751464843, "step": 2800 }, { "epoch": 0.011338601473023577, "grad_norm": 0.46429315209388733, "learning_rate": 1.9886653882571992e-05, "loss": 2.4308543395996094, "step": 2850 }, { "epoch": 0.01153752430588364, "grad_norm": 0.5067711472511292, "learning_rate": 1.9884664656221904e-05, "loss": 2.4706951904296877, "step": 2900 }, { "epoch": 0.011736447138743703, "grad_norm": 0.4977436661720276, "learning_rate": 1.9882675429871816e-05, "loss": 2.4439541625976564, "step": 2950 }, { "epoch": 0.011935369971603766, "grad_norm": 0.5013620853424072, "learning_rate": 1.988068620352173e-05, "loss": 2.4183050537109376, "step": 3000 }, { "epoch": 0.012134292804463829, "grad_norm": 0.4407724142074585, "learning_rate": 1.987869697717164e-05, "loss": 2.423381042480469, "step": 3050 }, { "epoch": 0.012333215637323891, "grad_norm": 0.5213010311126709, "learning_rate": 1.9876707750821553e-05, "loss": 2.4958120727539064, "step": 3100 }, { "epoch": 0.012532138470183954, "grad_norm": 0.5605552196502686, "learning_rate": 1.9874718524471465e-05, "loss": 2.4624327087402342, "step": 3150 }, { "epoch": 0.012731061303044017, "grad_norm": 0.5809507369995117, "learning_rate": 1.9872729298121377e-05, "loss": 2.4256831359863282, "step": 3200 }, { "epoch": 0.01292998413590408, "grad_norm": 0.5351467728614807, "learning_rate": 1.987074007177129e-05, "loss": 2.436898651123047, "step": 3250 }, { "epoch": 0.013128906968764143, "grad_norm": 0.4939177334308624, "learning_rate": 1.98687508454212e-05, "loss": 2.454507141113281, "step": 3300 }, { "epoch": 0.013327829801624206, "grad_norm": 0.4666970372200012, "learning_rate": 1.9866761619071113e-05, "loss": 2.393096466064453, "step": 3350 }, { "epoch": 0.013526752634484268, "grad_norm": 0.437502920627594, "learning_rate": 1.9864772392721026e-05, "loss": 2.421041717529297, "step": 3400 }, { "epoch": 0.01372567546734433, "grad_norm": 0.4937194287776947, "learning_rate": 1.9862783166370934e-05, "loss": 2.4278868103027342, "step": 3450 }, { "epoch": 0.013924598300204392, "grad_norm": Infinity, "learning_rate": 1.986079394002085e-05, "loss": 2.4413571166992187, "step": 3500 }, { "epoch": 0.014123521133064455, "grad_norm": 0.5779012441635132, "learning_rate": 1.9858844498197763e-05, "loss": 2.5103520202636718, "step": 3550 }, { "epoch": 0.014322443965924518, "grad_norm": 0.548040509223938, "learning_rate": 1.9856855271847675e-05, "loss": 2.467832336425781, "step": 3600 }, { "epoch": 0.014521366798784581, "grad_norm": 0.44418853521347046, "learning_rate": 1.9854866045497588e-05, "loss": 2.4846151733398436, "step": 3650 }, { "epoch": 0.014720289631644644, "grad_norm": 0.4724622964859009, "learning_rate": 1.98528768191475e-05, "loss": 2.4728684997558594, "step": 3700 }, { "epoch": 0.014919212464504707, "grad_norm": 0.5267306566238403, "learning_rate": 1.985088759279741e-05, "loss": 2.465199737548828, "step": 3750 }, { "epoch": 0.01511813529736477, "grad_norm": 0.5172199010848999, "learning_rate": 1.984889836644732e-05, "loss": 2.502405700683594, "step": 3800 }, { "epoch": 0.015317058130224832, "grad_norm": 0.5150426030158997, "learning_rate": 1.9846909140097236e-05, "loss": 2.441567230224609, "step": 3850 }, { "epoch": 0.015515980963084895, "grad_norm": 0.5188329219818115, "learning_rate": 1.984491991374715e-05, "loss": 2.414941864013672, "step": 3900 }, { "epoch": 0.015714903795944958, "grad_norm": 0.49610841274261475, "learning_rate": 1.984293068739706e-05, "loss": 2.4392112731933593, "step": 3950 }, { "epoch": 0.01591382662880502, "grad_norm": 0.4555966556072235, "learning_rate": 1.9840941461046973e-05, "loss": 2.4693463134765623, "step": 4000 }, { "epoch": 0.016112749461665084, "grad_norm": 0.46359139680862427, "learning_rate": 1.9838952234696885e-05, "loss": 2.426801910400391, "step": 4050 }, { "epoch": 0.016311672294525147, "grad_norm": 0.48716065287590027, "learning_rate": 1.9836963008346794e-05, "loss": 2.412553405761719, "step": 4100 }, { "epoch": 0.01651059512738521, "grad_norm": 0.534561812877655, "learning_rate": 1.9834973781996706e-05, "loss": 2.444194488525391, "step": 4150 }, { "epoch": 0.016709517960245272, "grad_norm": 0.387003093957901, "learning_rate": 1.9832984555646618e-05, "loss": 2.4243374633789063, "step": 4200 }, { "epoch": 0.016908440793105335, "grad_norm": 0.536815881729126, "learning_rate": 1.9830995329296533e-05, "loss": 2.4047177124023436, "step": 4250 }, { "epoch": 0.017107363625965398, "grad_norm": 0.4709075391292572, "learning_rate": 1.9829006102946446e-05, "loss": 2.4110227966308595, "step": 4300 }, { "epoch": 0.01730628645882546, "grad_norm": 0.5307052135467529, "learning_rate": 1.9827016876596358e-05, "loss": 2.415077667236328, "step": 4350 }, { "epoch": 0.017505209291685524, "grad_norm": 0.4903867840766907, "learning_rate": 1.9825027650246266e-05, "loss": 2.3533612060546876, "step": 4400 }, { "epoch": 0.017704132124545587, "grad_norm": 0.5303165316581726, "learning_rate": 1.982303842389618e-05, "loss": 2.380521545410156, "step": 4450 }, { "epoch": 0.01790305495740565, "grad_norm": 0.5815038681030273, "learning_rate": 1.982104919754609e-05, "loss": 2.4268548583984373, "step": 4500 }, { "epoch": 0.018101977790265712, "grad_norm": 0.52152019739151, "learning_rate": 1.9819059971196003e-05, "loss": 2.496058044433594, "step": 4550 }, { "epoch": 0.018300900623125775, "grad_norm": 0.5518723726272583, "learning_rate": 1.9817070744845915e-05, "loss": 2.418312225341797, "step": 4600 }, { "epoch": 0.018499823455985838, "grad_norm": 0.5223607420921326, "learning_rate": 1.981508151849583e-05, "loss": 2.3793487548828125, "step": 4650 }, { "epoch": 0.0186987462888459, "grad_norm": 0.5096916556358337, "learning_rate": 1.9813092292145743e-05, "loss": 2.476354522705078, "step": 4700 }, { "epoch": 0.018897669121705964, "grad_norm": 0.5020546913146973, "learning_rate": 1.981110306579565e-05, "loss": 2.3748490905761717, "step": 4750 }, { "epoch": 0.019096591954566026, "grad_norm": 0.5447866916656494, "learning_rate": 1.9809113839445564e-05, "loss": 2.41257568359375, "step": 4800 }, { "epoch": 0.01929551478742609, "grad_norm": 0.49975496530532837, "learning_rate": 1.9807124613095476e-05, "loss": 2.391145935058594, "step": 4850 }, { "epoch": 0.019494437620286152, "grad_norm": 0.5290626287460327, "learning_rate": 1.9805135386745388e-05, "loss": 2.397941131591797, "step": 4900 }, { "epoch": 0.01969336045314621, "grad_norm": 0.43505167961120605, "learning_rate": 1.98031461603953e-05, "loss": 2.4172259521484376, "step": 4950 }, { "epoch": 0.019892283286006274, "grad_norm": 0.5370852947235107, "learning_rate": 1.9801156934045212e-05, "loss": 2.464697265625, "step": 5000 }, { "epoch": 0.020091206118866337, "grad_norm": 0.48506954312324524, "learning_rate": 1.9799167707695124e-05, "loss": 2.405416259765625, "step": 5050 }, { "epoch": 0.0202901289517264, "grad_norm": 0.478952020406723, "learning_rate": 1.9797178481345036e-05, "loss": 2.403958740234375, "step": 5100 }, { "epoch": 0.020489051784586463, "grad_norm": 0.5438531041145325, "learning_rate": 1.979518925499495e-05, "loss": 2.357369842529297, "step": 5150 }, { "epoch": 0.020687974617446526, "grad_norm": 0.4977757930755615, "learning_rate": 1.979320002864486e-05, "loss": 2.3566912841796874, "step": 5200 }, { "epoch": 0.02088689745030659, "grad_norm": 0.45990070700645447, "learning_rate": 1.9791210802294773e-05, "loss": 2.4292533874511717, "step": 5250 }, { "epoch": 0.02108582028316665, "grad_norm": 0.5140153765678406, "learning_rate": 1.9789221575944685e-05, "loss": 2.3664129638671874, "step": 5300 }, { "epoch": 0.021284743116026714, "grad_norm": 0.6015112400054932, "learning_rate": 1.9787232349594597e-05, "loss": 2.407796630859375, "step": 5350 }, { "epoch": 0.021483665948886777, "grad_norm": 0.4536919593811035, "learning_rate": 1.978524312324451e-05, "loss": 2.3508540344238282, "step": 5400 }, { "epoch": 0.02168258878174684, "grad_norm": 0.4898732304573059, "learning_rate": 1.978325389689442e-05, "loss": 2.398020935058594, "step": 5450 }, { "epoch": 0.021881511614606903, "grad_norm": 0.49619200825691223, "learning_rate": 1.9781264670544334e-05, "loss": 2.4135781860351564, "step": 5500 }, { "epoch": 0.022080434447466966, "grad_norm": 0.5108316540718079, "learning_rate": 1.9779275444194246e-05, "loss": 2.3874090576171874, "step": 5550 }, { "epoch": 0.02227935728032703, "grad_norm": 0.3860325813293457, "learning_rate": 1.9777286217844158e-05, "loss": 2.43198486328125, "step": 5600 }, { "epoch": 0.02247828011318709, "grad_norm": 0.5498777031898499, "learning_rate": 1.977529699149407e-05, "loss": 2.3947015380859376, "step": 5650 }, { "epoch": 0.022677202946047154, "grad_norm": 0.513837456703186, "learning_rate": 1.9773307765143982e-05, "loss": 2.4300851440429687, "step": 5700 }, { "epoch": 0.022876125778907217, "grad_norm": 0.540860116481781, "learning_rate": 1.9771318538793894e-05, "loss": 2.346922607421875, "step": 5750 }, { "epoch": 0.02307504861176728, "grad_norm": 0.5587403178215027, "learning_rate": 1.9769329312443807e-05, "loss": 2.3774240112304685, "step": 5800 }, { "epoch": 0.023273971444627343, "grad_norm": 0.5778207182884216, "learning_rate": 1.976734008609372e-05, "loss": 2.363583679199219, "step": 5850 }, { "epoch": 0.023472894277487406, "grad_norm": 0.5030023455619812, "learning_rate": 1.976535085974363e-05, "loss": 2.380902862548828, "step": 5900 }, { "epoch": 0.02367181711034747, "grad_norm": 0.5647470951080322, "learning_rate": 1.9763361633393543e-05, "loss": 2.396997375488281, "step": 5950 }, { "epoch": 0.02387073994320753, "grad_norm": 0.5116869211196899, "learning_rate": 1.9761372407043455e-05, "loss": 2.377525177001953, "step": 6000 }, { "epoch": 0.024069662776067594, "grad_norm": 0.4980827271938324, "learning_rate": 1.9759383180693367e-05, "loss": 2.4115278625488283, "step": 6050 }, { "epoch": 0.024268585608927657, "grad_norm": 0.2921101748943329, "learning_rate": 1.975739395434328e-05, "loss": 2.410311279296875, "step": 6100 }, { "epoch": 0.02446750844178772, "grad_norm": 0.5903090238571167, "learning_rate": 1.975540472799319e-05, "loss": 2.424467010498047, "step": 6150 }, { "epoch": 0.024666431274647783, "grad_norm": 0.508520245552063, "learning_rate": 1.9753415501643104e-05, "loss": 2.4288893127441407, "step": 6200 }, { "epoch": 0.024865354107507846, "grad_norm": 0.5103323459625244, "learning_rate": 1.9751426275293016e-05, "loss": 2.356508941650391, "step": 6250 }, { "epoch": 0.02506427694036791, "grad_norm": 0.4771401882171631, "learning_rate": 1.9749437048942928e-05, "loss": 2.439430847167969, "step": 6300 }, { "epoch": 0.02526319977322797, "grad_norm": 0.5084514617919922, "learning_rate": 1.9747447822592837e-05, "loss": 2.3550816345214844, "step": 6350 }, { "epoch": 0.025462122606088034, "grad_norm": 0.5002087354660034, "learning_rate": 1.974545859624275e-05, "loss": 2.369927978515625, "step": 6400 }, { "epoch": 0.025661045438948097, "grad_norm": 0.5110379457473755, "learning_rate": 1.974346936989266e-05, "loss": 2.4110655212402343, "step": 6450 }, { "epoch": 0.02585996827180816, "grad_norm": 0.5042800307273865, "learning_rate": 1.9741480143542577e-05, "loss": 2.41763916015625, "step": 6500 }, { "epoch": 0.026058891104668223, "grad_norm": 0.5502017736434937, "learning_rate": 1.973949091719249e-05, "loss": 2.3731988525390624, "step": 6550 }, { "epoch": 0.026257813937528286, "grad_norm": 0.5786991715431213, "learning_rate": 1.97375016908424e-05, "loss": 2.3677528381347654, "step": 6600 }, { "epoch": 0.02645673677038835, "grad_norm": 0.5242754220962524, "learning_rate": 1.973551246449231e-05, "loss": 2.383412322998047, "step": 6650 }, { "epoch": 0.02665565960324841, "grad_norm": 0.527603030204773, "learning_rate": 1.9733523238142222e-05, "loss": 2.358564910888672, "step": 6700 }, { "epoch": 0.026854582436108474, "grad_norm": 0.5067466497421265, "learning_rate": 1.9731534011792134e-05, "loss": 2.3769796752929686, "step": 6750 }, { "epoch": 0.027053505268968537, "grad_norm": 0.47563591599464417, "learning_rate": 1.9729544785442046e-05, "loss": 2.4137895202636717, "step": 6800 }, { "epoch": 0.0272524281018286, "grad_norm": 0.589998185634613, "learning_rate": 1.9727555559091958e-05, "loss": 2.3478825378417967, "step": 6850 }, { "epoch": 0.02745135093468866, "grad_norm": 0.4740649461746216, "learning_rate": 1.9725566332741874e-05, "loss": 2.3479988098144533, "step": 6900 }, { "epoch": 0.027650273767548722, "grad_norm": 0.5006279945373535, "learning_rate": 1.9723577106391786e-05, "loss": 2.4142831420898436, "step": 6950 }, { "epoch": 0.027849196600408785, "grad_norm": 0.5197291374206543, "learning_rate": 1.9721587880041695e-05, "loss": 2.394392395019531, "step": 7000 }, { "epoch": 0.028048119433268848, "grad_norm": 0.5931088328361511, "learning_rate": 1.9719598653691607e-05, "loss": 2.3524211120605467, "step": 7050 }, { "epoch": 0.02824704226612891, "grad_norm": 0.49170392751693726, "learning_rate": 1.971760942734152e-05, "loss": 2.3925900268554687, "step": 7100 }, { "epoch": 0.028445965098988973, "grad_norm": 0.47105756402015686, "learning_rate": 1.971562020099143e-05, "loss": 2.3561181640625, "step": 7150 }, { "epoch": 0.028644887931849036, "grad_norm": 0.48790988326072693, "learning_rate": 1.9713630974641343e-05, "loss": 2.3567570495605468, "step": 7200 }, { "epoch": 0.0288438107647091, "grad_norm": 0.5016824007034302, "learning_rate": 1.9711641748291255e-05, "loss": 2.3650442504882814, "step": 7250 }, { "epoch": 0.029042733597569162, "grad_norm": 0.48277729749679565, "learning_rate": 1.9709652521941168e-05, "loss": 2.315171966552734, "step": 7300 }, { "epoch": 0.029241656430429225, "grad_norm": 0.43990591168403625, "learning_rate": 1.970766329559108e-05, "loss": 2.417427825927734, "step": 7350 }, { "epoch": 0.029440579263289288, "grad_norm": 0.4770338535308838, "learning_rate": 1.9705674069240992e-05, "loss": 2.3345152282714845, "step": 7400 }, { "epoch": 0.02963950209614935, "grad_norm": 0.4737016260623932, "learning_rate": 1.9703684842890904e-05, "loss": 2.318902893066406, "step": 7450 }, { "epoch": 0.029838424929009413, "grad_norm": 0.5070662498474121, "learning_rate": 1.9701695616540816e-05, "loss": 2.468476104736328, "step": 7500 }, { "epoch": 0.030037347761869476, "grad_norm": 0.5411328077316284, "learning_rate": 1.9699706390190728e-05, "loss": 2.3685066223144533, "step": 7550 }, { "epoch": 0.03023627059472954, "grad_norm": 0.48224562406539917, "learning_rate": 1.969771716384064e-05, "loss": 2.354645538330078, "step": 7600 }, { "epoch": 0.030435193427589602, "grad_norm": 0.46831250190734863, "learning_rate": 1.9695727937490553e-05, "loss": 2.3771917724609377, "step": 7650 }, { "epoch": 0.030634116260449665, "grad_norm": 0.5086206793785095, "learning_rate": 1.9693738711140465e-05, "loss": 2.351865081787109, "step": 7700 }, { "epoch": 0.030833039093309728, "grad_norm": 0.579850971698761, "learning_rate": 1.9691749484790377e-05, "loss": 2.4043829345703127, "step": 7750 }, { "epoch": 0.03103196192616979, "grad_norm": 0.4431004524230957, "learning_rate": 1.968976025844029e-05, "loss": 2.3181077575683595, "step": 7800 }, { "epoch": 0.031230884759029853, "grad_norm": 0.5328527688980103, "learning_rate": 1.96877710320902e-05, "loss": 2.311842346191406, "step": 7850 }, { "epoch": 0.031429807591889916, "grad_norm": 0.472569078207016, "learning_rate": 1.9685781805740113e-05, "loss": 2.4528424072265627, "step": 7900 }, { "epoch": 0.03162873042474998, "grad_norm": 0.48408880829811096, "learning_rate": 1.9683792579390025e-05, "loss": 2.316387481689453, "step": 7950 }, { "epoch": 0.03182765325761004, "grad_norm": 0.5018310546875, "learning_rate": 1.9681803353039938e-05, "loss": 2.3426324462890626, "step": 8000 }, { "epoch": 0.032026576090470105, "grad_norm": 0.4374028146266937, "learning_rate": 1.967981412668985e-05, "loss": 2.358287353515625, "step": 8050 }, { "epoch": 0.03222549892333017, "grad_norm": 0.46791157126426697, "learning_rate": 1.9677824900339762e-05, "loss": 2.307162628173828, "step": 8100 }, { "epoch": 0.03242442175619023, "grad_norm": 0.47853729128837585, "learning_rate": 1.9675835673989674e-05, "loss": 2.366924591064453, "step": 8150 }, { "epoch": 0.03262334458905029, "grad_norm": 0.48778337240219116, "learning_rate": 1.9673846447639586e-05, "loss": 2.2960238647460938, "step": 8200 }, { "epoch": 0.032822267421910356, "grad_norm": 0.47376468777656555, "learning_rate": 1.9671857221289498e-05, "loss": 2.3414100646972655, "step": 8250 }, { "epoch": 0.03302119025477042, "grad_norm": 0.5410640239715576, "learning_rate": 1.966986799493941e-05, "loss": 2.3411402893066406, "step": 8300 }, { "epoch": 0.03322011308763048, "grad_norm": 0.5330605506896973, "learning_rate": 1.9667878768589323e-05, "loss": 2.373150634765625, "step": 8350 }, { "epoch": 0.033419035920490545, "grad_norm": 0.49018922448158264, "learning_rate": 1.9665889542239235e-05, "loss": 2.3983915710449217, "step": 8400 }, { "epoch": 0.03361795875335061, "grad_norm": 0.423949271440506, "learning_rate": 1.9663900315889147e-05, "loss": 2.3363468933105467, "step": 8450 }, { "epoch": 0.03381688158621067, "grad_norm": 0.4857105314731598, "learning_rate": 1.966191108953906e-05, "loss": 2.3601783752441405, "step": 8500 }, { "epoch": 0.03401580441907073, "grad_norm": 0.41596662998199463, "learning_rate": 1.965992186318897e-05, "loss": 2.3816679382324217, "step": 8550 }, { "epoch": 0.034214727251930796, "grad_norm": 0.5268593430519104, "learning_rate": 1.965793263683888e-05, "loss": 2.3548573303222655, "step": 8600 }, { "epoch": 0.03441365008479086, "grad_norm": 0.48241081833839417, "learning_rate": 1.9655943410488792e-05, "loss": 2.3939895629882812, "step": 8650 }, { "epoch": 0.03461257291765092, "grad_norm": 0.47482624650001526, "learning_rate": 1.9653954184138708e-05, "loss": 2.3296827697753906, "step": 8700 }, { "epoch": 0.034811495750510985, "grad_norm": 0.5017552971839905, "learning_rate": 1.965196495778862e-05, "loss": 2.3560113525390625, "step": 8750 }, { "epoch": 0.03501041858337105, "grad_norm": 0.4968159794807434, "learning_rate": 1.9649975731438532e-05, "loss": 2.3225521850585937, "step": 8800 }, { "epoch": 0.03520934141623111, "grad_norm": 0.4342438280582428, "learning_rate": 1.9647986505088444e-05, "loss": 2.3919108581542967, "step": 8850 }, { "epoch": 0.03540826424909117, "grad_norm": 0.48853662610054016, "learning_rate": 1.9645997278738356e-05, "loss": 2.2774807739257814, "step": 8900 }, { "epoch": 0.035607187081951236, "grad_norm": 0.45906540751457214, "learning_rate": 1.9644008052388265e-05, "loss": 2.3346908569335936, "step": 8950 }, { "epoch": 0.0358061099148113, "grad_norm": 0.45042285323143005, "learning_rate": 1.9642018826038177e-05, "loss": 2.3472596740722658, "step": 9000 }, { "epoch": 0.03600503274767136, "grad_norm": 0.4837452173233032, "learning_rate": 1.964002959968809e-05, "loss": 2.3613658142089844, "step": 9050 }, { "epoch": 0.036203955580531424, "grad_norm": 0.582648515701294, "learning_rate": 1.9638040373338005e-05, "loss": 2.327206726074219, "step": 9100 }, { "epoch": 0.03640287841339149, "grad_norm": 0.560383677482605, "learning_rate": 1.9636051146987917e-05, "loss": 2.3279966735839843, "step": 9150 }, { "epoch": 0.03660180124625155, "grad_norm": 0.5158295631408691, "learning_rate": 1.963406192063783e-05, "loss": 2.3390972900390623, "step": 9200 }, { "epoch": 0.03680072407911161, "grad_norm": 0.6468159556388855, "learning_rate": 1.9632072694287738e-05, "loss": 2.3681561279296877, "step": 9250 }, { "epoch": 0.036999646911971676, "grad_norm": 0.4599217176437378, "learning_rate": 1.963008346793765e-05, "loss": 2.3407025146484375, "step": 9300 }, { "epoch": 0.03719856974483174, "grad_norm": 0.49240174889564514, "learning_rate": 1.9628094241587562e-05, "loss": 2.299652099609375, "step": 9350 }, { "epoch": 0.0373974925776918, "grad_norm": 0.5415018796920776, "learning_rate": 1.9626105015237474e-05, "loss": 2.323467102050781, "step": 9400 }, { "epoch": 0.037596415410551864, "grad_norm": 0.4469684958457947, "learning_rate": 1.9624115788887386e-05, "loss": 2.3925804138183593, "step": 9450 }, { "epoch": 0.03779533824341193, "grad_norm": 0.5326672196388245, "learning_rate": 1.9622126562537302e-05, "loss": 2.2917111206054686, "step": 9500 }, { "epoch": 0.03799426107627199, "grad_norm": 0.43654298782348633, "learning_rate": 1.9620137336187214e-05, "loss": 2.3698780822753904, "step": 9550 }, { "epoch": 0.03819318390913205, "grad_norm": 0.5225947499275208, "learning_rate": 1.9618148109837123e-05, "loss": 2.3575570678710935, "step": 9600 }, { "epoch": 0.038392106741992116, "grad_norm": 0.5259618759155273, "learning_rate": 1.9616158883487035e-05, "loss": 2.3093251037597655, "step": 9650 }, { "epoch": 0.03859102957485218, "grad_norm": 0.5407246351242065, "learning_rate": 1.9614169657136947e-05, "loss": 2.3317826843261718, "step": 9700 }, { "epoch": 0.03878995240771224, "grad_norm": 0.5087655782699585, "learning_rate": 1.961218043078686e-05, "loss": 2.356128845214844, "step": 9750 }, { "epoch": 0.038988875240572304, "grad_norm": 0.508955717086792, "learning_rate": 1.961019120443677e-05, "loss": 2.3583940124511718, "step": 9800 }, { "epoch": 0.03918779807343237, "grad_norm": 0.3918013870716095, "learning_rate": 1.9608201978086684e-05, "loss": 2.331027374267578, "step": 9850 }, { "epoch": 0.03938672090629242, "grad_norm": 0.5776320099830627, "learning_rate": 1.9606212751736596e-05, "loss": 2.3291180419921873, "step": 9900 }, { "epoch": 0.039585643739152486, "grad_norm": 0.558190643787384, "learning_rate": 1.9604223525386508e-05, "loss": 2.312794494628906, "step": 9950 }, { "epoch": 0.03978456657201255, "grad_norm": 0.5831214785575867, "learning_rate": 1.960223429903642e-05, "loss": 2.334913330078125, "step": 10000 }, { "epoch": 0.03998348940487261, "grad_norm": 0.49272438883781433, "learning_rate": 1.9600245072686332e-05, "loss": 2.3250746154785156, "step": 10050 }, { "epoch": 0.040182412237732675, "grad_norm": 0.5371116995811462, "learning_rate": 1.9598255846336244e-05, "loss": 2.340718536376953, "step": 10100 }, { "epoch": 0.04038133507059274, "grad_norm": 0.38808244466781616, "learning_rate": 1.9596266619986156e-05, "loss": 2.339864959716797, "step": 10150 }, { "epoch": 0.0405802579034528, "grad_norm": 0.4962560832500458, "learning_rate": 1.959427739363607e-05, "loss": 2.3182144165039062, "step": 10200 }, { "epoch": 0.04077918073631286, "grad_norm": 0.5038146376609802, "learning_rate": 1.959228816728598e-05, "loss": 2.313939208984375, "step": 10250 }, { "epoch": 0.040978103569172926, "grad_norm": 0.5089634656906128, "learning_rate": 1.9590298940935893e-05, "loss": 2.301549377441406, "step": 10300 }, { "epoch": 0.04117702640203299, "grad_norm": 0.4360482096672058, "learning_rate": 1.9588309714585805e-05, "loss": 2.3436187744140624, "step": 10350 }, { "epoch": 0.04137594923489305, "grad_norm": 0.5808095932006836, "learning_rate": 1.9586320488235717e-05, "loss": 2.2774468994140626, "step": 10400 }, { "epoch": 0.041574872067753114, "grad_norm": 0.4723690152168274, "learning_rate": 1.958433126188563e-05, "loss": 2.2778999328613283, "step": 10450 }, { "epoch": 0.04177379490061318, "grad_norm": 0.5660196542739868, "learning_rate": 1.958234203553554e-05, "loss": 2.318050537109375, "step": 10500 }, { "epoch": 0.04197271773347324, "grad_norm": 0.5613991022109985, "learning_rate": 1.9580352809185454e-05, "loss": 2.3471421813964843, "step": 10550 }, { "epoch": 0.0421716405663333, "grad_norm": 0.5560259222984314, "learning_rate": 1.9578363582835366e-05, "loss": 2.33948486328125, "step": 10600 }, { "epoch": 0.042370563399193366, "grad_norm": 0.5944580435752869, "learning_rate": 1.9576374356485278e-05, "loss": 2.3539222717285155, "step": 10650 }, { "epoch": 0.04256948623205343, "grad_norm": 0.4864901006221771, "learning_rate": 1.957438513013519e-05, "loss": 2.323912811279297, "step": 10700 }, { "epoch": 0.04276840906491349, "grad_norm": 0.5965571403503418, "learning_rate": 1.9572395903785102e-05, "loss": 2.342884216308594, "step": 10750 }, { "epoch": 0.042967331897773554, "grad_norm": 0.5202301740646362, "learning_rate": 1.9570406677435014e-05, "loss": 2.3452874755859376, "step": 10800 }, { "epoch": 0.04316625473063362, "grad_norm": 0.5094255208969116, "learning_rate": 1.9568417451084926e-05, "loss": 2.309946746826172, "step": 10850 }, { "epoch": 0.04336517756349368, "grad_norm": 0.47360074520111084, "learning_rate": 1.9566428224734835e-05, "loss": 2.327744140625, "step": 10900 }, { "epoch": 0.04356410039635374, "grad_norm": 0.5249730348587036, "learning_rate": 1.956443899838475e-05, "loss": 2.298944854736328, "step": 10950 }, { "epoch": 0.043763023229213806, "grad_norm": 0.5034305453300476, "learning_rate": 1.9562489556561664e-05, "loss": 2.293399200439453, "step": 11000 }, { "epoch": 0.04396194606207387, "grad_norm": 0.4938550591468811, "learning_rate": 1.9560500330211576e-05, "loss": 2.2921197509765623, "step": 11050 }, { "epoch": 0.04416086889493393, "grad_norm": 0.5328453779220581, "learning_rate": 1.955851110386149e-05, "loss": 2.251877746582031, "step": 11100 }, { "epoch": 0.044359791727793994, "grad_norm": 0.5028183460235596, "learning_rate": 1.95565218775114e-05, "loss": 2.3029872131347657, "step": 11150 }, { "epoch": 0.04455871456065406, "grad_norm": 0.5333051085472107, "learning_rate": 1.955453265116131e-05, "loss": 2.350544738769531, "step": 11200 }, { "epoch": 0.04475763739351412, "grad_norm": 0.5336958169937134, "learning_rate": 1.9552543424811225e-05, "loss": 2.297502899169922, "step": 11250 }, { "epoch": 0.04495656022637418, "grad_norm": 0.47798606753349304, "learning_rate": 1.9550554198461137e-05, "loss": 2.3716781616210936, "step": 11300 }, { "epoch": 0.045155483059234246, "grad_norm": 0.4582976996898651, "learning_rate": 1.954856497211105e-05, "loss": 2.3393397521972656, "step": 11350 }, { "epoch": 0.04535440589209431, "grad_norm": 0.512231707572937, "learning_rate": 1.954657574576096e-05, "loss": 2.277655029296875, "step": 11400 }, { "epoch": 0.04555332872495437, "grad_norm": 0.5011807084083557, "learning_rate": 1.9544586519410874e-05, "loss": 2.3257359313964843, "step": 11450 }, { "epoch": 0.045752251557814434, "grad_norm": 0.5237681865692139, "learning_rate": 1.9542597293060786e-05, "loss": 2.3508877563476562, "step": 11500 }, { "epoch": 0.0459511743906745, "grad_norm": 0.5609585642814636, "learning_rate": 1.9540608066710695e-05, "loss": 2.352852020263672, "step": 11550 }, { "epoch": 0.04615009722353456, "grad_norm": 0.4589311480522156, "learning_rate": 1.9538618840360607e-05, "loss": 2.3124302673339843, "step": 11600 }, { "epoch": 0.04634902005639462, "grad_norm": 0.5263824462890625, "learning_rate": 1.9536629614010522e-05, "loss": 2.2939715576171875, "step": 11650 }, { "epoch": 0.046547942889254686, "grad_norm": 0.4893070459365845, "learning_rate": 1.9534640387660434e-05, "loss": 2.346824035644531, "step": 11700 }, { "epoch": 0.04674686572211475, "grad_norm": 0.4856417179107666, "learning_rate": 1.9532651161310346e-05, "loss": 2.3714889526367187, "step": 11750 }, { "epoch": 0.04694578855497481, "grad_norm": 0.6172106862068176, "learning_rate": 1.953066193496026e-05, "loss": 2.3307008361816406, "step": 11800 }, { "epoch": 0.047144711387834874, "grad_norm": 0.516497790813446, "learning_rate": 1.9528672708610167e-05, "loss": 2.285205993652344, "step": 11850 }, { "epoch": 0.04734363422069494, "grad_norm": 0.4227709472179413, "learning_rate": 1.952668348226008e-05, "loss": 2.2692947387695312, "step": 11900 }, { "epoch": 0.047542557053555, "grad_norm": 0.48502033948898315, "learning_rate": 1.952469425590999e-05, "loss": 2.3162091064453123, "step": 11950 }, { "epoch": 0.04774147988641506, "grad_norm": 0.527104377746582, "learning_rate": 1.9522705029559904e-05, "loss": 2.3269512939453123, "step": 12000 }, { "epoch": 0.047940402719275126, "grad_norm": 0.5222641229629517, "learning_rate": 1.9520715803209816e-05, "loss": 2.3050389099121094, "step": 12050 }, { "epoch": 0.04813932555213519, "grad_norm": 0.5718929767608643, "learning_rate": 1.951872657685973e-05, "loss": 2.2601531982421874, "step": 12100 }, { "epoch": 0.04833824838499525, "grad_norm": 0.5632660984992981, "learning_rate": 1.9516737350509644e-05, "loss": 2.3684675598144533, "step": 12150 }, { "epoch": 0.048537171217855314, "grad_norm": 0.5941556692123413, "learning_rate": 1.9514748124159552e-05, "loss": 2.2955555725097656, "step": 12200 }, { "epoch": 0.04873609405071538, "grad_norm": 0.5404202342033386, "learning_rate": 1.9512758897809465e-05, "loss": 2.2957498168945314, "step": 12250 }, { "epoch": 0.04893501688357544, "grad_norm": 0.47728902101516724, "learning_rate": 1.9510769671459377e-05, "loss": 2.3152812194824217, "step": 12300 }, { "epoch": 0.0491339397164355, "grad_norm": 0.5876482129096985, "learning_rate": 1.950878044510929e-05, "loss": 2.3083648681640625, "step": 12350 }, { "epoch": 0.049332862549295566, "grad_norm": 0.4640975594520569, "learning_rate": 1.95067912187592e-05, "loss": 2.298441619873047, "step": 12400 }, { "epoch": 0.04953178538215563, "grad_norm": 0.4961966276168823, "learning_rate": 1.9504801992409113e-05, "loss": 2.277954406738281, "step": 12450 }, { "epoch": 0.04973070821501569, "grad_norm": 0.4657418131828308, "learning_rate": 1.9502812766059025e-05, "loss": 2.298264465332031, "step": 12500 }, { "epoch": 0.049929631047875754, "grad_norm": 0.5002082586288452, "learning_rate": 1.9500823539708937e-05, "loss": 2.2952839660644533, "step": 12550 }, { "epoch": 0.05012855388073582, "grad_norm": 0.5634850263595581, "learning_rate": 1.949883431335885e-05, "loss": 2.3539105224609376, "step": 12600 }, { "epoch": 0.05032747671359588, "grad_norm": 0.4764060974121094, "learning_rate": 1.9496845087008762e-05, "loss": 2.3008193969726562, "step": 12650 }, { "epoch": 0.05052639954645594, "grad_norm": 0.4094434380531311, "learning_rate": 1.9494855860658674e-05, "loss": 2.3506492614746093, "step": 12700 }, { "epoch": 0.050725322379316005, "grad_norm": 0.5008903741836548, "learning_rate": 1.9492866634308586e-05, "loss": 2.248862762451172, "step": 12750 }, { "epoch": 0.05092424521217607, "grad_norm": 0.5186200141906738, "learning_rate": 1.9490877407958498e-05, "loss": 2.3135299682617188, "step": 12800 }, { "epoch": 0.05112316804503613, "grad_norm": 0.5264031887054443, "learning_rate": 1.948888818160841e-05, "loss": 2.3413407897949217, "step": 12850 }, { "epoch": 0.051322090877896194, "grad_norm": 0.47033506631851196, "learning_rate": 1.9486898955258322e-05, "loss": 2.243574371337891, "step": 12900 }, { "epoch": 0.05152101371075626, "grad_norm": 0.5113924145698547, "learning_rate": 1.9484909728908235e-05, "loss": 2.2803724670410155, "step": 12950 }, { "epoch": 0.05171993654361632, "grad_norm": 0.519797682762146, "learning_rate": 1.9482920502558147e-05, "loss": 2.316182403564453, "step": 13000 }, { "epoch": 0.05191885937647638, "grad_norm": 0.5598306059837341, "learning_rate": 1.948093127620806e-05, "loss": 2.3222360229492187, "step": 13050 }, { "epoch": 0.052117782209336445, "grad_norm": 0.5373337864875793, "learning_rate": 1.947894204985797e-05, "loss": 2.291268005371094, "step": 13100 }, { "epoch": 0.05231670504219651, "grad_norm": 0.517918050289154, "learning_rate": 1.9476952823507883e-05, "loss": 2.3252610778808593, "step": 13150 }, { "epoch": 0.05251562787505657, "grad_norm": 0.5501250624656677, "learning_rate": 1.9474963597157795e-05, "loss": 2.3216508483886718, "step": 13200 }, { "epoch": 0.052714550707916634, "grad_norm": 0.5196780562400818, "learning_rate": 1.9472974370807707e-05, "loss": 2.2962966918945313, "step": 13250 }, { "epoch": 0.0529134735407767, "grad_norm": 0.42645496129989624, "learning_rate": 1.947098514445762e-05, "loss": 2.298736114501953, "step": 13300 }, { "epoch": 0.05311239637363676, "grad_norm": 0.5262851119041443, "learning_rate": 1.9468995918107532e-05, "loss": 2.2902371215820314, "step": 13350 }, { "epoch": 0.05331131920649682, "grad_norm": 0.42818087339401245, "learning_rate": 1.9467006691757444e-05, "loss": 2.2837010192871094, "step": 13400 }, { "epoch": 0.053510242039356885, "grad_norm": 0.5155251026153564, "learning_rate": 1.9465017465407353e-05, "loss": 2.3207275390625, "step": 13450 }, { "epoch": 0.05370916487221695, "grad_norm": 0.5333742499351501, "learning_rate": 1.9463028239057268e-05, "loss": 2.280226135253906, "step": 13500 }, { "epoch": 0.05390808770507701, "grad_norm": 0.4708807170391083, "learning_rate": 1.946103901270718e-05, "loss": 2.2816253662109376, "step": 13550 }, { "epoch": 0.054107010537937074, "grad_norm": 0.5292482376098633, "learning_rate": 1.9459049786357092e-05, "loss": 2.284524688720703, "step": 13600 }, { "epoch": 0.05430593337079714, "grad_norm": 0.4885295629501343, "learning_rate": 1.9457060560007005e-05, "loss": 2.2974005126953125, "step": 13650 }, { "epoch": 0.0545048562036572, "grad_norm": 0.5740625262260437, "learning_rate": 1.9455071333656917e-05, "loss": 2.3547288513183595, "step": 13700 }, { "epoch": 0.05470377903651726, "grad_norm": 0.49187207221984863, "learning_rate": 1.945308210730683e-05, "loss": 2.2945762634277345, "step": 13750 }, { "epoch": 0.05490270186937732, "grad_norm": 0.5620005130767822, "learning_rate": 1.9451092880956738e-05, "loss": 2.246928863525391, "step": 13800 }, { "epoch": 0.05510162470223738, "grad_norm": 0.5952235460281372, "learning_rate": 1.944910365460665e-05, "loss": 2.2627517700195314, "step": 13850 }, { "epoch": 0.055300547535097444, "grad_norm": 0.5201510787010193, "learning_rate": 1.9447114428256565e-05, "loss": 2.378265380859375, "step": 13900 }, { "epoch": 0.05549947036795751, "grad_norm": 0.45743629336357117, "learning_rate": 1.9445125201906478e-05, "loss": 2.243382568359375, "step": 13950 }, { "epoch": 0.05569839320081757, "grad_norm": 0.5189955830574036, "learning_rate": 1.944313597555639e-05, "loss": 2.3361222839355467, "step": 14000 }, { "epoch": 0.05589731603367763, "grad_norm": 0.528221070766449, "learning_rate": 1.9441146749206302e-05, "loss": 2.3786846923828127, "step": 14050 }, { "epoch": 0.056096238866537695, "grad_norm": 0.504446268081665, "learning_rate": 1.943915752285621e-05, "loss": 2.348214874267578, "step": 14100 }, { "epoch": 0.05629516169939776, "grad_norm": 0.5590423941612244, "learning_rate": 1.9437168296506123e-05, "loss": 2.2670046997070314, "step": 14150 }, { "epoch": 0.05649408453225782, "grad_norm": 0.5155242681503296, "learning_rate": 1.9435179070156035e-05, "loss": 2.3145977783203127, "step": 14200 }, { "epoch": 0.056693007365117884, "grad_norm": 0.48422813415527344, "learning_rate": 1.9433189843805947e-05, "loss": 2.2731761169433593, "step": 14250 }, { "epoch": 0.05689193019797795, "grad_norm": 0.4635816514492035, "learning_rate": 1.9431200617455863e-05, "loss": 2.283515625, "step": 14300 }, { "epoch": 0.05709085303083801, "grad_norm": 0.524685263633728, "learning_rate": 1.9429211391105775e-05, "loss": 2.263406982421875, "step": 14350 }, { "epoch": 0.05728977586369807, "grad_norm": 0.49985840916633606, "learning_rate": 1.9427222164755687e-05, "loss": 2.284171295166016, "step": 14400 }, { "epoch": 0.057488698696558135, "grad_norm": 0.547370970249176, "learning_rate": 1.9425232938405596e-05, "loss": 2.277954864501953, "step": 14450 }, { "epoch": 0.0576876215294182, "grad_norm": 0.5773013830184937, "learning_rate": 1.9423243712055508e-05, "loss": 2.299970703125, "step": 14500 }, { "epoch": 0.05788654436227826, "grad_norm": 0.5027283430099487, "learning_rate": 1.942125448570542e-05, "loss": 2.3062429809570313, "step": 14550 }, { "epoch": 0.058085467195138324, "grad_norm": 0.49694421887397766, "learning_rate": 1.9419265259355332e-05, "loss": 2.3104264831542967, "step": 14600 }, { "epoch": 0.05828439002799839, "grad_norm": 0.5415092706680298, "learning_rate": 1.941731581753225e-05, "loss": 2.2892745971679687, "step": 14650 }, { "epoch": 0.05848331286085845, "grad_norm": 0.506134569644928, "learning_rate": 1.941532659118216e-05, "loss": 2.267159271240234, "step": 14700 }, { "epoch": 0.05868223569371851, "grad_norm": 0.5004725456237793, "learning_rate": 1.9413337364832073e-05, "loss": 2.2388436889648435, "step": 14750 }, { "epoch": 0.058881158526578575, "grad_norm": 0.5330361127853394, "learning_rate": 1.9411348138481982e-05, "loss": 2.2184083557128904, "step": 14800 }, { "epoch": 0.05908008135943864, "grad_norm": 0.5564002394676208, "learning_rate": 1.9409358912131894e-05, "loss": 2.263230438232422, "step": 14850 }, { "epoch": 0.0592790041922987, "grad_norm": 0.46604278683662415, "learning_rate": 1.9407369685781806e-05, "loss": 2.272127380371094, "step": 14900 }, { "epoch": 0.059477927025158764, "grad_norm": 0.5634845495223999, "learning_rate": 1.940538045943172e-05, "loss": 2.3157357788085937, "step": 14950 }, { "epoch": 0.05967684985801883, "grad_norm": 0.5179756283760071, "learning_rate": 1.940339123308163e-05, "loss": 2.2395057678222656, "step": 15000 }, { "epoch": 0.05987577269087889, "grad_norm": 0.47275763750076294, "learning_rate": 1.9401402006731546e-05, "loss": 2.2625660705566406, "step": 15050 }, { "epoch": 0.06007469552373895, "grad_norm": 0.4224807322025299, "learning_rate": 1.9399412780381455e-05, "loss": 2.321799774169922, "step": 15100 }, { "epoch": 0.060273618356599015, "grad_norm": 0.5521396994590759, "learning_rate": 1.9397423554031367e-05, "loss": 2.2350067138671874, "step": 15150 }, { "epoch": 0.06047254118945908, "grad_norm": 0.4845815896987915, "learning_rate": 1.939543432768128e-05, "loss": 2.337565460205078, "step": 15200 }, { "epoch": 0.06067146402231914, "grad_norm": 0.9439213871955872, "learning_rate": 1.939344510133119e-05, "loss": 2.240023193359375, "step": 15250 }, { "epoch": 0.060870386855179204, "grad_norm": 0.5578423738479614, "learning_rate": 1.9391455874981103e-05, "loss": 2.2590676879882814, "step": 15300 }, { "epoch": 0.06106930968803927, "grad_norm": 0.5932989120483398, "learning_rate": 1.9389466648631016e-05, "loss": 2.2708169555664064, "step": 15350 }, { "epoch": 0.06126823252089933, "grad_norm": 0.6691983342170715, "learning_rate": 1.9387477422280928e-05, "loss": 2.251698760986328, "step": 15400 }, { "epoch": 0.06146715535375939, "grad_norm": 0.5190653800964355, "learning_rate": 1.938548819593084e-05, "loss": 2.292062530517578, "step": 15450 }, { "epoch": 0.061666078186619455, "grad_norm": 0.499209463596344, "learning_rate": 1.9383498969580752e-05, "loss": 2.278829803466797, "step": 15500 }, { "epoch": 0.06186500101947952, "grad_norm": 0.47052615880966187, "learning_rate": 1.9381509743230664e-05, "loss": 2.243563232421875, "step": 15550 }, { "epoch": 0.06206392385233958, "grad_norm": 0.5466270446777344, "learning_rate": 1.9379520516880576e-05, "loss": 2.2603825378417968, "step": 15600 }, { "epoch": 0.062262846685199644, "grad_norm": 0.4138563573360443, "learning_rate": 1.937753129053049e-05, "loss": 2.246696014404297, "step": 15650 }, { "epoch": 0.06246176951805971, "grad_norm": 0.5726961493492126, "learning_rate": 1.93755420641804e-05, "loss": 2.2826632690429687, "step": 15700 }, { "epoch": 0.06266069235091977, "grad_norm": 0.6065142750740051, "learning_rate": 1.9373552837830313e-05, "loss": 2.2477983093261718, "step": 15750 }, { "epoch": 0.06285961518377983, "grad_norm": 0.5724515318870544, "learning_rate": 1.9371563611480225e-05, "loss": 2.2435235595703125, "step": 15800 }, { "epoch": 0.0630585380166399, "grad_norm": 0.5459858775138855, "learning_rate": 1.9369574385130137e-05, "loss": 2.251372833251953, "step": 15850 }, { "epoch": 0.06325746084949996, "grad_norm": 0.5499418377876282, "learning_rate": 1.936758515878005e-05, "loss": 2.243422393798828, "step": 15900 }, { "epoch": 0.06345638368236002, "grad_norm": 0.5175423622131348, "learning_rate": 1.936559593242996e-05, "loss": 2.3148013305664064, "step": 15950 }, { "epoch": 0.06365530651522008, "grad_norm": 0.5250300765037537, "learning_rate": 1.9363606706079873e-05, "loss": 2.3222923278808594, "step": 16000 }, { "epoch": 0.06385422934808015, "grad_norm": 0.528753936290741, "learning_rate": 1.9361617479729786e-05, "loss": 2.2579660034179687, "step": 16050 }, { "epoch": 0.06405315218094021, "grad_norm": 0.4630770981311798, "learning_rate": 1.9359628253379698e-05, "loss": 2.261855010986328, "step": 16100 }, { "epoch": 0.06425207501380027, "grad_norm": 0.4770432114601135, "learning_rate": 1.935763902702961e-05, "loss": 2.2707588195800783, "step": 16150 }, { "epoch": 0.06445099784666034, "grad_norm": 0.483958899974823, "learning_rate": 1.9355649800679522e-05, "loss": 2.243647155761719, "step": 16200 }, { "epoch": 0.0646499206795204, "grad_norm": 0.4701959788799286, "learning_rate": 1.9353660574329434e-05, "loss": 2.2362126159667968, "step": 16250 }, { "epoch": 0.06484884351238046, "grad_norm": 0.47445306181907654, "learning_rate": 1.9351671347979346e-05, "loss": 2.323039855957031, "step": 16300 }, { "epoch": 0.06504776634524052, "grad_norm": 0.5592694282531738, "learning_rate": 1.934968212162926e-05, "loss": 2.2386038208007815, "step": 16350 }, { "epoch": 0.06524668917810059, "grad_norm": 0.5260711908340454, "learning_rate": 1.9347692895279167e-05, "loss": 2.267103271484375, "step": 16400 }, { "epoch": 0.06544561201096065, "grad_norm": 0.49941763281822205, "learning_rate": 1.9345703668929083e-05, "loss": 2.298883056640625, "step": 16450 }, { "epoch": 0.06564453484382071, "grad_norm": 0.43885648250579834, "learning_rate": 1.9343714442578995e-05, "loss": 2.252605133056641, "step": 16500 }, { "epoch": 0.06584345767668077, "grad_norm": 0.548008918762207, "learning_rate": 1.9341725216228907e-05, "loss": 2.293037872314453, "step": 16550 }, { "epoch": 0.06604238050954084, "grad_norm": 0.47238582372665405, "learning_rate": 1.933973598987882e-05, "loss": 2.2085845947265623, "step": 16600 }, { "epoch": 0.0662413033424009, "grad_norm": 0.5320387482643127, "learning_rate": 1.933774676352873e-05, "loss": 2.234955139160156, "step": 16650 }, { "epoch": 0.06644022617526096, "grad_norm": 0.5135543942451477, "learning_rate": 1.933575753717864e-05, "loss": 2.273407287597656, "step": 16700 }, { "epoch": 0.06663914900812103, "grad_norm": 0.5423591732978821, "learning_rate": 1.9333768310828552e-05, "loss": 2.289434356689453, "step": 16750 }, { "epoch": 0.06683807184098109, "grad_norm": 0.5710050463676453, "learning_rate": 1.9331779084478464e-05, "loss": 2.3617236328125, "step": 16800 }, { "epoch": 0.06703699467384115, "grad_norm": 0.46634551882743835, "learning_rate": 1.932978985812838e-05, "loss": 2.2675123596191407, "step": 16850 }, { "epoch": 0.06723591750670121, "grad_norm": 0.49296534061431885, "learning_rate": 1.9327800631778292e-05, "loss": 2.242511901855469, "step": 16900 }, { "epoch": 0.06743484033956128, "grad_norm": 0.5517917275428772, "learning_rate": 1.9325811405428204e-05, "loss": 2.307420196533203, "step": 16950 }, { "epoch": 0.06763376317242134, "grad_norm": 0.5116357207298279, "learning_rate": 1.9323822179078116e-05, "loss": 2.2879432678222655, "step": 17000 }, { "epoch": 0.0678326860052814, "grad_norm": 0.5889848470687866, "learning_rate": 1.9321832952728025e-05, "loss": 2.260302734375, "step": 17050 }, { "epoch": 0.06803160883814147, "grad_norm": 0.5375827550888062, "learning_rate": 1.9319843726377937e-05, "loss": 2.204697723388672, "step": 17100 }, { "epoch": 0.06823053167100153, "grad_norm": 0.602928638458252, "learning_rate": 1.931785450002785e-05, "loss": 2.268499298095703, "step": 17150 }, { "epoch": 0.06842945450386159, "grad_norm": 0.4201854467391968, "learning_rate": 1.931586527367776e-05, "loss": 2.240293426513672, "step": 17200 }, { "epoch": 0.06862837733672165, "grad_norm": 0.5538254976272583, "learning_rate": 1.931391583185468e-05, "loss": 2.2632789611816406, "step": 17250 }, { "epoch": 0.06882730016958172, "grad_norm": 0.5226390957832336, "learning_rate": 1.931192660550459e-05, "loss": 2.196150360107422, "step": 17300 }, { "epoch": 0.06902622300244178, "grad_norm": 0.6434885859489441, "learning_rate": 1.93099373791545e-05, "loss": 2.2584542846679687, "step": 17350 }, { "epoch": 0.06922514583530184, "grad_norm": 0.5963517427444458, "learning_rate": 1.930794815280441e-05, "loss": 2.2612571716308594, "step": 17400 }, { "epoch": 0.0694240686681619, "grad_norm": 0.5165600180625916, "learning_rate": 1.9305958926454324e-05, "loss": 2.2604876708984376, "step": 17450 }, { "epoch": 0.06962299150102197, "grad_norm": 0.4988686442375183, "learning_rate": 1.9303969700104236e-05, "loss": 2.2345852661132812, "step": 17500 }, { "epoch": 0.06982191433388203, "grad_norm": 0.528110146522522, "learning_rate": 1.9301980473754148e-05, "loss": 2.1977088928222654, "step": 17550 }, { "epoch": 0.0700208371667421, "grad_norm": 0.5710955262184143, "learning_rate": 1.9299991247404064e-05, "loss": 2.2550419616699218, "step": 17600 }, { "epoch": 0.07021975999960216, "grad_norm": 0.5284214615821838, "learning_rate": 1.9298002021053976e-05, "loss": 2.2567848205566405, "step": 17650 }, { "epoch": 0.07041868283246222, "grad_norm": 0.6087015271186829, "learning_rate": 1.9296012794703884e-05, "loss": 2.235007781982422, "step": 17700 }, { "epoch": 0.07061760566532228, "grad_norm": 0.5690137147903442, "learning_rate": 1.9294023568353797e-05, "loss": 2.273645324707031, "step": 17750 }, { "epoch": 0.07081652849818235, "grad_norm": 0.5589634776115417, "learning_rate": 1.929203434200371e-05, "loss": 2.295856781005859, "step": 17800 }, { "epoch": 0.07101545133104241, "grad_norm": 0.48619866371154785, "learning_rate": 1.929004511565362e-05, "loss": 2.216392822265625, "step": 17850 }, { "epoch": 0.07121437416390247, "grad_norm": 0.5639730095863342, "learning_rate": 1.9288055889303533e-05, "loss": 2.254555358886719, "step": 17900 }, { "epoch": 0.07141329699676253, "grad_norm": 0.5078126192092896, "learning_rate": 1.9286066662953445e-05, "loss": 2.235054626464844, "step": 17950 }, { "epoch": 0.0716122198296226, "grad_norm": 0.49570342898368835, "learning_rate": 1.9284077436603357e-05, "loss": 2.2499732971191406, "step": 18000 }, { "epoch": 0.07181114266248266, "grad_norm": 0.5313770771026611, "learning_rate": 1.928208821025327e-05, "loss": 2.2653758239746096, "step": 18050 }, { "epoch": 0.07201006549534272, "grad_norm": 0.48001348972320557, "learning_rate": 1.928009898390318e-05, "loss": 2.257662506103516, "step": 18100 }, { "epoch": 0.07220898832820279, "grad_norm": 0.3942019045352936, "learning_rate": 1.9278109757553094e-05, "loss": 2.337117156982422, "step": 18150 }, { "epoch": 0.07240791116106285, "grad_norm": 0.5007137060165405, "learning_rate": 1.9276120531203006e-05, "loss": 2.260611114501953, "step": 18200 }, { "epoch": 0.07260683399392291, "grad_norm": 0.6752716302871704, "learning_rate": 1.9274131304852918e-05, "loss": 2.2367718505859373, "step": 18250 }, { "epoch": 0.07280575682678297, "grad_norm": 0.50809645652771, "learning_rate": 1.927214207850283e-05, "loss": 2.2623423767089843, "step": 18300 }, { "epoch": 0.07300467965964304, "grad_norm": 0.4264284372329712, "learning_rate": 1.9270152852152742e-05, "loss": 2.1989878845214843, "step": 18350 }, { "epoch": 0.0732036024925031, "grad_norm": 0.5571845769882202, "learning_rate": 1.9268163625802654e-05, "loss": 2.249786682128906, "step": 18400 }, { "epoch": 0.07340252532536316, "grad_norm": 0.5684404373168945, "learning_rate": 1.9266174399452567e-05, "loss": 2.1938783264160158, "step": 18450 }, { "epoch": 0.07360144815822323, "grad_norm": 0.49422428011894226, "learning_rate": 1.926418517310248e-05, "loss": 2.2386700439453127, "step": 18500 }, { "epoch": 0.07380037099108329, "grad_norm": 0.5562616586685181, "learning_rate": 1.926219594675239e-05, "loss": 2.3040483093261717, "step": 18550 }, { "epoch": 0.07399929382394335, "grad_norm": 0.5516597032546997, "learning_rate": 1.9260206720402303e-05, "loss": 2.2434646606445314, "step": 18600 }, { "epoch": 0.07419821665680341, "grad_norm": 0.5219873785972595, "learning_rate": 1.9258217494052215e-05, "loss": 2.2612664794921873, "step": 18650 }, { "epoch": 0.07439713948966348, "grad_norm": 0.5599794983863831, "learning_rate": 1.9256228267702127e-05, "loss": 2.3215176391601564, "step": 18700 }, { "epoch": 0.07459606232252354, "grad_norm": 0.44705837965011597, "learning_rate": 1.925423904135204e-05, "loss": 2.227407684326172, "step": 18750 }, { "epoch": 0.0747949851553836, "grad_norm": 0.49408650398254395, "learning_rate": 1.925224981500195e-05, "loss": 2.261202850341797, "step": 18800 }, { "epoch": 0.07499390798824367, "grad_norm": 0.44640490412712097, "learning_rate": 1.9250260588651864e-05, "loss": 2.207927398681641, "step": 18850 }, { "epoch": 0.07519283082110373, "grad_norm": 0.5079675912857056, "learning_rate": 1.9248271362301776e-05, "loss": 2.2557188415527345, "step": 18900 }, { "epoch": 0.07539175365396379, "grad_norm": 0.5404150485992432, "learning_rate": 1.9246282135951688e-05, "loss": 2.2587408447265624, "step": 18950 }, { "epoch": 0.07559067648682385, "grad_norm": 0.4647636413574219, "learning_rate": 1.9244292909601597e-05, "loss": 2.2662451171875, "step": 19000 }, { "epoch": 0.07578959931968392, "grad_norm": 0.5353376269340515, "learning_rate": 1.9242303683251512e-05, "loss": 2.2282760620117186, "step": 19050 }, { "epoch": 0.07598852215254398, "grad_norm": 0.4676717221736908, "learning_rate": 1.9240314456901424e-05, "loss": 2.2177734375, "step": 19100 }, { "epoch": 0.07618744498540404, "grad_norm": 0.5785231590270996, "learning_rate": 1.9238325230551337e-05, "loss": 2.293676452636719, "step": 19150 }, { "epoch": 0.0763863678182641, "grad_norm": 0.5690168142318726, "learning_rate": 1.923633600420125e-05, "loss": 2.2182044982910156, "step": 19200 }, { "epoch": 0.07658529065112417, "grad_norm": 0.44276684522628784, "learning_rate": 1.923434677785116e-05, "loss": 2.2315022277832033, "step": 19250 }, { "epoch": 0.07678421348398423, "grad_norm": 0.5890305638313293, "learning_rate": 1.923235755150107e-05, "loss": 2.2373358154296876, "step": 19300 }, { "epoch": 0.0769831363168443, "grad_norm": 0.4856083393096924, "learning_rate": 1.9230368325150982e-05, "loss": 2.273783721923828, "step": 19350 }, { "epoch": 0.07718205914970436, "grad_norm": 0.5061861872673035, "learning_rate": 1.9228379098800894e-05, "loss": 2.2115774536132813, "step": 19400 }, { "epoch": 0.07738098198256442, "grad_norm": 0.5369890332221985, "learning_rate": 1.922638987245081e-05, "loss": 2.2092352294921875, "step": 19450 }, { "epoch": 0.07757990481542448, "grad_norm": 0.5227654576301575, "learning_rate": 1.922440064610072e-05, "loss": 2.2668409729003907, "step": 19500 }, { "epoch": 0.07777882764828455, "grad_norm": 0.5890436172485352, "learning_rate": 1.9222411419750634e-05, "loss": 2.205455780029297, "step": 19550 }, { "epoch": 0.07797775048114461, "grad_norm": 0.3902466893196106, "learning_rate": 1.9220422193400546e-05, "loss": 2.2453822326660156, "step": 19600 }, { "epoch": 0.07817667331400467, "grad_norm": 0.5694035291671753, "learning_rate": 1.9218432967050455e-05, "loss": 2.2124876403808593, "step": 19650 }, { "epoch": 0.07837559614686473, "grad_norm": 0.504370927810669, "learning_rate": 1.9216443740700367e-05, "loss": 2.2534359741210936, "step": 19700 }, { "epoch": 0.07857451897972478, "grad_norm": 0.514361560344696, "learning_rate": 1.921445451435028e-05, "loss": 2.1974472045898437, "step": 19750 }, { "epoch": 0.07877344181258485, "grad_norm": 0.5535170435905457, "learning_rate": 1.921246528800019e-05, "loss": 2.2017835998535156, "step": 19800 }, { "epoch": 0.07897236464544491, "grad_norm": 0.5240291357040405, "learning_rate": 1.9210476061650107e-05, "loss": 2.2662974548339845, "step": 19850 }, { "epoch": 0.07917128747830497, "grad_norm": 0.5630162954330444, "learning_rate": 1.920848683530002e-05, "loss": 2.2635284423828126, "step": 19900 }, { "epoch": 0.07937021031116503, "grad_norm": 0.41869664192199707, "learning_rate": 1.920653739347693e-05, "loss": 2.2299433898925782, "step": 19950 }, { "epoch": 0.0795691331440251, "grad_norm": 0.44380590319633484, "learning_rate": 1.920454816712684e-05, "loss": 2.282265167236328, "step": 20000 }, { "epoch": 0.07976805597688516, "grad_norm": 0.5242075324058533, "learning_rate": 1.9202558940776753e-05, "loss": 2.1920591735839845, "step": 20050 }, { "epoch": 0.07996697880974522, "grad_norm": 0.6112689971923828, "learning_rate": 1.9200569714426665e-05, "loss": 2.1657148742675782, "step": 20100 }, { "epoch": 0.08016590164260529, "grad_norm": 0.4734936058521271, "learning_rate": 1.9198580488076578e-05, "loss": 2.239213714599609, "step": 20150 }, { "epoch": 0.08036482447546535, "grad_norm": 0.5113575458526611, "learning_rate": 1.9196591261726493e-05, "loss": 2.2156605529785156, "step": 20200 }, { "epoch": 0.08056374730832541, "grad_norm": 0.5292969942092896, "learning_rate": 1.9194602035376405e-05, "loss": 2.202944030761719, "step": 20250 }, { "epoch": 0.08076267014118547, "grad_norm": 0.4934408962726593, "learning_rate": 1.9192612809026314e-05, "loss": 2.2269247436523436, "step": 20300 }, { "epoch": 0.08096159297404554, "grad_norm": 0.5857859253883362, "learning_rate": 1.9190623582676226e-05, "loss": 2.2357867431640623, "step": 20350 }, { "epoch": 0.0811605158069056, "grad_norm": 0.533180296421051, "learning_rate": 1.9188634356326138e-05, "loss": 2.230920867919922, "step": 20400 }, { "epoch": 0.08135943863976566, "grad_norm": 0.49916261434555054, "learning_rate": 1.918664512997605e-05, "loss": 2.251059722900391, "step": 20450 }, { "epoch": 0.08155836147262573, "grad_norm": 0.5111820697784424, "learning_rate": 1.9184655903625963e-05, "loss": 2.2105255126953125, "step": 20500 }, { "epoch": 0.08175728430548579, "grad_norm": 0.5431243777275085, "learning_rate": 1.9182666677275875e-05, "loss": 2.257746124267578, "step": 20550 }, { "epoch": 0.08195620713834585, "grad_norm": 0.5016031861305237, "learning_rate": 1.9180677450925787e-05, "loss": 2.2267640686035155, "step": 20600 }, { "epoch": 0.08215512997120591, "grad_norm": 0.6616201400756836, "learning_rate": 1.91787280091027e-05, "loss": 2.2452490234375, "step": 20650 }, { "epoch": 0.08235405280406598, "grad_norm": 0.481441468000412, "learning_rate": 1.9176738782752612e-05, "loss": 2.2257070922851563, "step": 20700 }, { "epoch": 0.08255297563692604, "grad_norm": 0.49232688546180725, "learning_rate": 1.9174749556402525e-05, "loss": 2.2019859313964845, "step": 20750 }, { "epoch": 0.0827518984697861, "grad_norm": 0.40650299191474915, "learning_rate": 1.9172760330052437e-05, "loss": 2.205707550048828, "step": 20800 }, { "epoch": 0.08295082130264617, "grad_norm": 0.5505855679512024, "learning_rate": 1.917077110370235e-05, "loss": 2.249988708496094, "step": 20850 }, { "epoch": 0.08314974413550623, "grad_norm": 0.5149282217025757, "learning_rate": 1.916878187735226e-05, "loss": 2.2129244995117188, "step": 20900 }, { "epoch": 0.08334866696836629, "grad_norm": 0.4310462772846222, "learning_rate": 1.9166792651002173e-05, "loss": 2.227361755371094, "step": 20950 }, { "epoch": 0.08354758980122635, "grad_norm": 0.5928002595901489, "learning_rate": 1.9164803424652085e-05, "loss": 2.2282760620117186, "step": 21000 }, { "epoch": 0.08374651263408642, "grad_norm": 0.5114561319351196, "learning_rate": 1.9162814198301998e-05, "loss": 2.2504832458496096, "step": 21050 }, { "epoch": 0.08394543546694648, "grad_norm": 0.5711889266967773, "learning_rate": 1.916082497195191e-05, "loss": 2.227830047607422, "step": 21100 }, { "epoch": 0.08414435829980654, "grad_norm": 0.5236198306083679, "learning_rate": 1.9158835745601822e-05, "loss": 2.2332635498046876, "step": 21150 }, { "epoch": 0.0843432811326666, "grad_norm": 0.4644909203052521, "learning_rate": 1.9156846519251734e-05, "loss": 2.275635070800781, "step": 21200 }, { "epoch": 0.08454220396552667, "grad_norm": 0.5297006368637085, "learning_rate": 1.9154857292901646e-05, "loss": 2.260339813232422, "step": 21250 }, { "epoch": 0.08474112679838673, "grad_norm": 0.46736204624176025, "learning_rate": 1.9152868066551558e-05, "loss": 2.2583378601074218, "step": 21300 }, { "epoch": 0.0849400496312468, "grad_norm": 0.4722309708595276, "learning_rate": 1.915087884020147e-05, "loss": 2.224939727783203, "step": 21350 }, { "epoch": 0.08513897246410686, "grad_norm": 0.48527276515960693, "learning_rate": 1.9148889613851383e-05, "loss": 2.2234088134765626, "step": 21400 }, { "epoch": 0.08533789529696692, "grad_norm": 0.4921424984931946, "learning_rate": 1.9146900387501295e-05, "loss": 2.2215570068359374, "step": 21450 }, { "epoch": 0.08553681812982698, "grad_norm": 0.5774291157722473, "learning_rate": 1.9144911161151207e-05, "loss": 2.216626739501953, "step": 21500 }, { "epoch": 0.08573574096268705, "grad_norm": 0.5275934338569641, "learning_rate": 1.914292193480112e-05, "loss": 2.185272979736328, "step": 21550 }, { "epoch": 0.08593466379554711, "grad_norm": 0.45150667428970337, "learning_rate": 1.914093270845103e-05, "loss": 2.2577420043945313, "step": 21600 }, { "epoch": 0.08613358662840717, "grad_norm": 0.5706455111503601, "learning_rate": 1.9138943482100943e-05, "loss": 2.2663214111328127, "step": 21650 }, { "epoch": 0.08633250946126723, "grad_norm": 0.57062166929245, "learning_rate": 1.9136954255750855e-05, "loss": 2.169580078125, "step": 21700 }, { "epoch": 0.0865314322941273, "grad_norm": 0.5280207991600037, "learning_rate": 1.9134965029400768e-05, "loss": 2.237732696533203, "step": 21750 }, { "epoch": 0.08673035512698736, "grad_norm": 0.32185986638069153, "learning_rate": 1.913297580305068e-05, "loss": 2.219240875244141, "step": 21800 }, { "epoch": 0.08692927795984742, "grad_norm": 0.603204071521759, "learning_rate": 1.9130986576700592e-05, "loss": 2.297452697753906, "step": 21850 }, { "epoch": 0.08712820079270749, "grad_norm": 0.5550698637962341, "learning_rate": 1.91289973503505e-05, "loss": 2.217601470947266, "step": 21900 }, { "epoch": 0.08732712362556755, "grad_norm": 0.518869161605835, "learning_rate": 1.9127008124000416e-05, "loss": 2.2601890563964844, "step": 21950 }, { "epoch": 0.08752604645842761, "grad_norm": 0.5013507008552551, "learning_rate": 1.9125018897650328e-05, "loss": 2.2063389587402344, "step": 22000 }, { "epoch": 0.08772496929128767, "grad_norm": 0.4572283625602722, "learning_rate": 1.912302967130024e-05, "loss": 2.2068136596679686, "step": 22050 }, { "epoch": 0.08792389212414774, "grad_norm": 0.4398670494556427, "learning_rate": 1.9121040444950153e-05, "loss": 2.2205125427246095, "step": 22100 }, { "epoch": 0.0881228149570078, "grad_norm": 0.4564041197299957, "learning_rate": 1.9119051218600065e-05, "loss": 2.2912664794921875, "step": 22150 }, { "epoch": 0.08832173778986786, "grad_norm": 0.5852370858192444, "learning_rate": 1.9117061992249977e-05, "loss": 2.2339857482910155, "step": 22200 }, { "epoch": 0.08852066062272793, "grad_norm": 0.5010802149772644, "learning_rate": 1.9115072765899886e-05, "loss": 2.1703260803222655, "step": 22250 }, { "epoch": 0.08871958345558799, "grad_norm": 0.5369467735290527, "learning_rate": 1.9113083539549798e-05, "loss": 2.26935791015625, "step": 22300 }, { "epoch": 0.08891850628844805, "grad_norm": 0.537528395652771, "learning_rate": 1.9111094313199713e-05, "loss": 2.259157257080078, "step": 22350 }, { "epoch": 0.08911742912130811, "grad_norm": 0.47058555483818054, "learning_rate": 1.9109105086849625e-05, "loss": 2.22094970703125, "step": 22400 }, { "epoch": 0.08931635195416818, "grad_norm": 0.5411145687103271, "learning_rate": 1.9107115860499538e-05, "loss": 2.1865924072265623, "step": 22450 }, { "epoch": 0.08951527478702824, "grad_norm": 0.5630295872688293, "learning_rate": 1.910512663414945e-05, "loss": 2.1591645812988283, "step": 22500 }, { "epoch": 0.0897141976198883, "grad_norm": 0.5385822653770447, "learning_rate": 1.910313740779936e-05, "loss": 2.2275790405273437, "step": 22550 }, { "epoch": 0.08991312045274837, "grad_norm": 0.5663081407546997, "learning_rate": 1.910114818144927e-05, "loss": 2.1666244506835937, "step": 22600 }, { "epoch": 0.09011204328560843, "grad_norm": 0.5236029028892517, "learning_rate": 1.9099158955099183e-05, "loss": 2.2232469177246093, "step": 22650 }, { "epoch": 0.09031096611846849, "grad_norm": 0.5315884947776794, "learning_rate": 1.9097169728749095e-05, "loss": 2.2381637573242186, "step": 22700 }, { "epoch": 0.09050988895132855, "grad_norm": 0.5694779753684998, "learning_rate": 1.909518050239901e-05, "loss": 2.2820585632324217, "step": 22750 }, { "epoch": 0.09070881178418862, "grad_norm": 0.49796080589294434, "learning_rate": 1.9093191276048923e-05, "loss": 2.2527337646484376, "step": 22800 }, { "epoch": 0.09090773461704868, "grad_norm": 0.4976939260959625, "learning_rate": 1.9091202049698835e-05, "loss": 2.2228549194335936, "step": 22850 }, { "epoch": 0.09110665744990874, "grad_norm": 0.4776739776134491, "learning_rate": 1.9089212823348744e-05, "loss": 2.2447206115722658, "step": 22900 }, { "epoch": 0.0913055802827688, "grad_norm": 0.552729070186615, "learning_rate": 1.9087223596998656e-05, "loss": 2.2800949096679686, "step": 22950 }, { "epoch": 0.09150450311562887, "grad_norm": 0.5749696493148804, "learning_rate": 1.9085234370648568e-05, "loss": 2.250563201904297, "step": 23000 }, { "epoch": 0.09170342594848893, "grad_norm": 0.5456891655921936, "learning_rate": 1.908324514429848e-05, "loss": 2.2699468994140624, "step": 23050 }, { "epoch": 0.091902348781349, "grad_norm": 0.5951603651046753, "learning_rate": 1.9081255917948392e-05, "loss": 2.2452099609375, "step": 23100 }, { "epoch": 0.09210127161420906, "grad_norm": 0.5396323204040527, "learning_rate": 1.9079266691598308e-05, "loss": 2.26619873046875, "step": 23150 }, { "epoch": 0.09230019444706912, "grad_norm": 0.5890723466873169, "learning_rate": 1.9077277465248216e-05, "loss": 2.249764404296875, "step": 23200 }, { "epoch": 0.09249911727992918, "grad_norm": 0.5174845457077026, "learning_rate": 1.907528823889813e-05, "loss": 2.2315234375, "step": 23250 }, { "epoch": 0.09269804011278925, "grad_norm": 0.48812437057495117, "learning_rate": 1.907329901254804e-05, "loss": 2.226591033935547, "step": 23300 }, { "epoch": 0.09289696294564931, "grad_norm": 0.4839382469654083, "learning_rate": 1.9071309786197953e-05, "loss": 2.231209564208984, "step": 23350 }, { "epoch": 0.09309588577850937, "grad_norm": 0.5233811140060425, "learning_rate": 1.9069320559847865e-05, "loss": 2.253740692138672, "step": 23400 }, { "epoch": 0.09329480861136943, "grad_norm": 0.541310727596283, "learning_rate": 1.9067331333497777e-05, "loss": 2.209989013671875, "step": 23450 }, { "epoch": 0.0934937314442295, "grad_norm": 0.5438382625579834, "learning_rate": 1.906534210714769e-05, "loss": 2.218547821044922, "step": 23500 }, { "epoch": 0.09369265427708956, "grad_norm": 0.4714638590812683, "learning_rate": 1.90633528807976e-05, "loss": 2.1902545166015623, "step": 23550 }, { "epoch": 0.09389157710994962, "grad_norm": 0.45889031887054443, "learning_rate": 1.9061363654447514e-05, "loss": 2.1727229309082032, "step": 23600 }, { "epoch": 0.09409049994280969, "grad_norm": 0.5265594124794006, "learning_rate": 1.9059374428097426e-05, "loss": 2.165437774658203, "step": 23650 }, { "epoch": 0.09428942277566975, "grad_norm": 0.537134051322937, "learning_rate": 1.9057385201747338e-05, "loss": 2.1991175842285156, "step": 23700 }, { "epoch": 0.09448834560852981, "grad_norm": 0.531226634979248, "learning_rate": 1.905539597539725e-05, "loss": 2.258228607177734, "step": 23750 }, { "epoch": 0.09468726844138987, "grad_norm": 0.5448224544525146, "learning_rate": 1.9053406749047162e-05, "loss": 2.2098089599609376, "step": 23800 }, { "epoch": 0.09488619127424994, "grad_norm": 0.4585404098033905, "learning_rate": 1.9051417522697074e-05, "loss": 2.172798919677734, "step": 23850 }, { "epoch": 0.09508511410711, "grad_norm": 0.6402602791786194, "learning_rate": 1.9049428296346986e-05, "loss": 2.198953399658203, "step": 23900 }, { "epoch": 0.09528403693997006, "grad_norm": 0.4964609146118164, "learning_rate": 1.90474390699969e-05, "loss": 2.235067596435547, "step": 23950 }, { "epoch": 0.09548295977283013, "grad_norm": 0.5043805241584778, "learning_rate": 1.904544984364681e-05, "loss": 2.2139385986328124, "step": 24000 }, { "epoch": 0.09568188260569019, "grad_norm": 0.5152483582496643, "learning_rate": 1.9043460617296723e-05, "loss": 2.167518310546875, "step": 24050 }, { "epoch": 0.09588080543855025, "grad_norm": 0.49679529666900635, "learning_rate": 1.9041471390946635e-05, "loss": 2.2163775634765623, "step": 24100 }, { "epoch": 0.09607972827141031, "grad_norm": 0.5344534516334534, "learning_rate": 1.9039482164596547e-05, "loss": 2.230632019042969, "step": 24150 }, { "epoch": 0.09627865110427038, "grad_norm": 0.586783230304718, "learning_rate": 1.903749293824646e-05, "loss": 2.179014129638672, "step": 24200 }, { "epoch": 0.09647757393713044, "grad_norm": 0.44779062271118164, "learning_rate": 1.903550371189637e-05, "loss": 2.208468780517578, "step": 24250 }, { "epoch": 0.0966764967699905, "grad_norm": 0.5417418479919434, "learning_rate": 1.9033514485546284e-05, "loss": 2.1852215576171874, "step": 24300 }, { "epoch": 0.09687541960285057, "grad_norm": 0.45786598324775696, "learning_rate": 1.9031525259196196e-05, "loss": 2.2354312133789063, "step": 24350 }, { "epoch": 0.09707434243571063, "grad_norm": 0.5086018443107605, "learning_rate": 1.9029536032846108e-05, "loss": 2.236842041015625, "step": 24400 }, { "epoch": 0.09727326526857069, "grad_norm": 0.5502527356147766, "learning_rate": 1.902754680649602e-05, "loss": 2.2236407470703123, "step": 24450 }, { "epoch": 0.09747218810143075, "grad_norm": 0.48024752736091614, "learning_rate": 1.902555758014593e-05, "loss": 2.205392303466797, "step": 24500 }, { "epoch": 0.09767111093429082, "grad_norm": 0.5257444977760315, "learning_rate": 1.9023568353795844e-05, "loss": 2.2299461364746094, "step": 24550 }, { "epoch": 0.09787003376715088, "grad_norm": 0.5897514820098877, "learning_rate": 1.9021579127445756e-05, "loss": 2.1892210388183595, "step": 24600 }, { "epoch": 0.09806895660001094, "grad_norm": 0.45728054642677307, "learning_rate": 1.901958990109567e-05, "loss": 2.2177497863769533, "step": 24650 }, { "epoch": 0.098267879432871, "grad_norm": 0.5624350309371948, "learning_rate": 1.901760067474558e-05, "loss": 2.265419769287109, "step": 24700 }, { "epoch": 0.09846680226573107, "grad_norm": 0.5596985816955566, "learning_rate": 1.9015611448395493e-05, "loss": 2.2380415344238282, "step": 24750 }, { "epoch": 0.09866572509859113, "grad_norm": 0.4572666585445404, "learning_rate": 1.9013622222045405e-05, "loss": 2.13435791015625, "step": 24800 }, { "epoch": 0.0988646479314512, "grad_norm": 0.5419999361038208, "learning_rate": 1.9011632995695314e-05, "loss": 2.1959410095214844, "step": 24850 }, { "epoch": 0.09906357076431126, "grad_norm": 0.39732205867767334, "learning_rate": 1.9009643769345226e-05, "loss": 2.2044882202148437, "step": 24900 }, { "epoch": 0.09926249359717132, "grad_norm": 0.4838886857032776, "learning_rate": 1.9007654542995138e-05, "loss": 2.1623988342285156, "step": 24950 }, { "epoch": 0.09946141643003138, "grad_norm": 0.5340353846549988, "learning_rate": 1.9005665316645054e-05, "loss": 2.201682586669922, "step": 25000 }, { "epoch": 0.09966033926289145, "grad_norm": 0.4874856472015381, "learning_rate": 1.9003676090294966e-05, "loss": 2.179762115478516, "step": 25050 }, { "epoch": 0.09985926209575151, "grad_norm": 0.597293496131897, "learning_rate": 1.9001686863944878e-05, "loss": 2.1897151184082033, "step": 25100 }, { "epoch": 0.10005818492861157, "grad_norm": 0.616527259349823, "learning_rate": 1.8999697637594787e-05, "loss": 2.2089044189453126, "step": 25150 }, { "epoch": 0.10025710776147163, "grad_norm": 0.5018583536148071, "learning_rate": 1.89977084112447e-05, "loss": 2.1837612915039064, "step": 25200 }, { "epoch": 0.1004560305943317, "grad_norm": 0.5576973557472229, "learning_rate": 1.899571918489461e-05, "loss": 2.1766094970703125, "step": 25250 }, { "epoch": 0.10065495342719176, "grad_norm": 0.47765475511550903, "learning_rate": 1.8993729958544523e-05, "loss": 2.188108215332031, "step": 25300 }, { "epoch": 0.10085387626005182, "grad_norm": 0.5481031537055969, "learning_rate": 1.8991740732194435e-05, "loss": 2.2076034545898438, "step": 25350 }, { "epoch": 0.10105279909291189, "grad_norm": 0.5525266528129578, "learning_rate": 1.898975150584435e-05, "loss": 2.2367764282226563, "step": 25400 }, { "epoch": 0.10125172192577195, "grad_norm": 0.6358855962753296, "learning_rate": 1.898776227949426e-05, "loss": 2.277913360595703, "step": 25450 }, { "epoch": 0.10145064475863201, "grad_norm": 0.586553692817688, "learning_rate": 1.8985773053144172e-05, "loss": 2.205990905761719, "step": 25500 }, { "epoch": 0.10164956759149207, "grad_norm": 0.6233808398246765, "learning_rate": 1.8983783826794084e-05, "loss": 2.199638671875, "step": 25550 }, { "epoch": 0.10184849042435214, "grad_norm": 0.5381131768226624, "learning_rate": 1.8981794600443996e-05, "loss": 2.1774388122558594, "step": 25600 }, { "epoch": 0.1020474132572122, "grad_norm": 0.5108106136322021, "learning_rate": 1.8979805374093908e-05, "loss": 2.1605458068847656, "step": 25650 }, { "epoch": 0.10224633609007226, "grad_norm": 0.601452648639679, "learning_rate": 1.897781614774382e-05, "loss": 2.2473460388183595, "step": 25700 }, { "epoch": 0.10244525892293233, "grad_norm": 0.5824030041694641, "learning_rate": 1.8975826921393732e-05, "loss": 2.1808038330078126, "step": 25750 }, { "epoch": 0.10264418175579239, "grad_norm": 0.49952933192253113, "learning_rate": 1.8973837695043645e-05, "loss": 2.1982958984375, "step": 25800 }, { "epoch": 0.10284310458865245, "grad_norm": 0.535213828086853, "learning_rate": 1.8971848468693557e-05, "loss": 2.2291824340820314, "step": 25850 }, { "epoch": 0.10304202742151251, "grad_norm": 0.5249989628791809, "learning_rate": 1.896985924234347e-05, "loss": 2.190851287841797, "step": 25900 }, { "epoch": 0.10324095025437258, "grad_norm": 0.4699265956878662, "learning_rate": 1.896787001599338e-05, "loss": 2.204471435546875, "step": 25950 }, { "epoch": 0.10343987308723264, "grad_norm": 0.536679208278656, "learning_rate": 1.8965880789643293e-05, "loss": 2.212308044433594, "step": 26000 }, { "epoch": 0.1036387959200927, "grad_norm": 0.6690974831581116, "learning_rate": 1.8963891563293205e-05, "loss": 2.1846487426757815, "step": 26050 }, { "epoch": 0.10383771875295277, "grad_norm": 0.6133774518966675, "learning_rate": 1.8961902336943117e-05, "loss": 2.245725402832031, "step": 26100 }, { "epoch": 0.10403664158581283, "grad_norm": 0.5619173645973206, "learning_rate": 1.895991311059303e-05, "loss": 2.2335055541992186, "step": 26150 }, { "epoch": 0.10423556441867289, "grad_norm": 0.5932241678237915, "learning_rate": 1.8957923884242942e-05, "loss": 2.238319854736328, "step": 26200 }, { "epoch": 0.10443448725153295, "grad_norm": 0.46078115701675415, "learning_rate": 1.8955934657892854e-05, "loss": 2.1728929138183593, "step": 26250 }, { "epoch": 0.10463341008439302, "grad_norm": 0.5721940398216248, "learning_rate": 1.8953945431542766e-05, "loss": 2.18090087890625, "step": 26300 }, { "epoch": 0.10483233291725308, "grad_norm": 0.46019718050956726, "learning_rate": 1.8951956205192678e-05, "loss": 2.2068414306640625, "step": 26350 }, { "epoch": 0.10503125575011314, "grad_norm": 0.5252344608306885, "learning_rate": 1.894996697884259e-05, "loss": 2.183749542236328, "step": 26400 }, { "epoch": 0.1052301785829732, "grad_norm": 0.5209318399429321, "learning_rate": 1.8947977752492503e-05, "loss": 2.2603472900390624, "step": 26450 }, { "epoch": 0.10542910141583327, "grad_norm": 0.553101658821106, "learning_rate": 1.8945988526142415e-05, "loss": 2.227206115722656, "step": 26500 }, { "epoch": 0.10562802424869333, "grad_norm": 0.4108973443508148, "learning_rate": 1.8943999299792327e-05, "loss": 2.1462176513671873, "step": 26550 }, { "epoch": 0.1058269470815534, "grad_norm": 0.4923765957355499, "learning_rate": 1.894204985796924e-05, "loss": 2.2065269470214846, "step": 26600 }, { "epoch": 0.10602586991441346, "grad_norm": 0.5080229043960571, "learning_rate": 1.8940060631619152e-05, "loss": 2.155797576904297, "step": 26650 }, { "epoch": 0.10622479274727352, "grad_norm": 0.5774619579315186, "learning_rate": 1.8938071405269065e-05, "loss": 2.239965362548828, "step": 26700 }, { "epoch": 0.10642371558013358, "grad_norm": 0.5317085981369019, "learning_rate": 1.8936121963445978e-05, "loss": 2.2504208374023436, "step": 26750 }, { "epoch": 0.10662263841299364, "grad_norm": 0.5436949729919434, "learning_rate": 1.893413273709589e-05, "loss": 2.2073155212402344, "step": 26800 }, { "epoch": 0.10682156124585371, "grad_norm": 0.6102069020271301, "learning_rate": 1.8932143510745802e-05, "loss": 2.2152667236328125, "step": 26850 }, { "epoch": 0.10702048407871377, "grad_norm": 0.5976851582527161, "learning_rate": 1.8930154284395715e-05, "loss": 2.224620819091797, "step": 26900 }, { "epoch": 0.10721940691157383, "grad_norm": 0.4353698790073395, "learning_rate": 1.8928165058045627e-05, "loss": 2.174615478515625, "step": 26950 }, { "epoch": 0.1074183297444339, "grad_norm": 0.5135305523872375, "learning_rate": 1.892617583169554e-05, "loss": 2.144137420654297, "step": 27000 }, { "epoch": 0.10761725257729396, "grad_norm": 0.518898069858551, "learning_rate": 1.892418660534545e-05, "loss": 2.1823663330078125, "step": 27050 }, { "epoch": 0.10781617541015402, "grad_norm": 0.5821455121040344, "learning_rate": 1.8922197378995363e-05, "loss": 2.145536193847656, "step": 27100 }, { "epoch": 0.10801509824301408, "grad_norm": 0.442483514547348, "learning_rate": 1.8920208152645275e-05, "loss": 2.2388938903808593, "step": 27150 }, { "epoch": 0.10821402107587415, "grad_norm": 0.5532496571540833, "learning_rate": 1.8918218926295187e-05, "loss": 2.1827685546875, "step": 27200 }, { "epoch": 0.10841294390873421, "grad_norm": 0.4949336349964142, "learning_rate": 1.89162296999451e-05, "loss": 2.180545196533203, "step": 27250 }, { "epoch": 0.10861186674159427, "grad_norm": 0.5375663042068481, "learning_rate": 1.8914240473595012e-05, "loss": 2.2183297729492186, "step": 27300 }, { "epoch": 0.10881078957445434, "grad_norm": 0.4804486036300659, "learning_rate": 1.8912251247244924e-05, "loss": 2.218954315185547, "step": 27350 }, { "epoch": 0.1090097124073144, "grad_norm": 0.4665115475654602, "learning_rate": 1.8910262020894836e-05, "loss": 2.1853366088867188, "step": 27400 }, { "epoch": 0.10920863524017446, "grad_norm": 0.47396838665008545, "learning_rate": 1.8908272794544745e-05, "loss": 2.1916033935546877, "step": 27450 }, { "epoch": 0.10940755807303452, "grad_norm": 0.5480823516845703, "learning_rate": 1.890628356819466e-05, "loss": 2.23872314453125, "step": 27500 }, { "epoch": 0.10960648090589459, "grad_norm": 0.5504873394966125, "learning_rate": 1.8904294341844572e-05, "loss": 2.214105529785156, "step": 27550 }, { "epoch": 0.10980540373875464, "grad_norm": 0.49983084201812744, "learning_rate": 1.8902305115494485e-05, "loss": 2.196951446533203, "step": 27600 }, { "epoch": 0.1100043265716147, "grad_norm": 0.519454300403595, "learning_rate": 1.8900315889144397e-05, "loss": 2.2155496215820314, "step": 27650 }, { "epoch": 0.11020324940447476, "grad_norm": 0.4831204116344452, "learning_rate": 1.889832666279431e-05, "loss": 2.1964601135253905, "step": 27700 }, { "epoch": 0.11040217223733483, "grad_norm": 0.4859577715396881, "learning_rate": 1.8896337436444218e-05, "loss": 2.1954396057128904, "step": 27750 }, { "epoch": 0.11060109507019489, "grad_norm": 0.5543569326400757, "learning_rate": 1.889434821009413e-05, "loss": 2.2147335815429687, "step": 27800 }, { "epoch": 0.11080001790305495, "grad_norm": 0.586184024810791, "learning_rate": 1.8892358983744042e-05, "loss": 2.18451904296875, "step": 27850 }, { "epoch": 0.11099894073591501, "grad_norm": 0.5566709637641907, "learning_rate": 1.8890369757393957e-05, "loss": 2.250419464111328, "step": 27900 }, { "epoch": 0.11119786356877508, "grad_norm": 0.5738060474395752, "learning_rate": 1.888838053104387e-05, "loss": 2.1825433349609376, "step": 27950 }, { "epoch": 0.11139678640163514, "grad_norm": 0.44597962498664856, "learning_rate": 1.8886391304693782e-05, "loss": 2.1700035095214845, "step": 28000 }, { "epoch": 0.1115957092344952, "grad_norm": 0.4276014268398285, "learning_rate": 1.8884402078343694e-05, "loss": 2.1617501831054686, "step": 28050 }, { "epoch": 0.11179463206735527, "grad_norm": 0.5192756652832031, "learning_rate": 1.8882412851993603e-05, "loss": 2.1343067932128905, "step": 28100 }, { "epoch": 0.11199355490021533, "grad_norm": 0.5830237865447998, "learning_rate": 1.8880423625643515e-05, "loss": 2.1678652954101563, "step": 28150 }, { "epoch": 0.11219247773307539, "grad_norm": 0.5306761860847473, "learning_rate": 1.8878434399293427e-05, "loss": 2.1846363830566404, "step": 28200 }, { "epoch": 0.11239140056593545, "grad_norm": 0.5165621042251587, "learning_rate": 1.887644517294334e-05, "loss": 2.1501683044433593, "step": 28250 }, { "epoch": 0.11259032339879552, "grad_norm": 0.404417484998703, "learning_rate": 1.8874455946593255e-05, "loss": 2.184874267578125, "step": 28300 }, { "epoch": 0.11278924623165558, "grad_norm": 0.5903500318527222, "learning_rate": 1.8872466720243167e-05, "loss": 2.1885256958007813, "step": 28350 }, { "epoch": 0.11298816906451564, "grad_norm": 0.5522173047065735, "learning_rate": 1.8870477493893076e-05, "loss": 2.2119174194335938, "step": 28400 }, { "epoch": 0.1131870918973757, "grad_norm": 0.4989508092403412, "learning_rate": 1.8868488267542988e-05, "loss": 2.187999267578125, "step": 28450 }, { "epoch": 0.11338601473023577, "grad_norm": 0.49987801909446716, "learning_rate": 1.88664990411929e-05, "loss": 2.236266326904297, "step": 28500 }, { "epoch": 0.11358493756309583, "grad_norm": 0.47671663761138916, "learning_rate": 1.8864509814842812e-05, "loss": 2.219978942871094, "step": 28550 }, { "epoch": 0.1137838603959559, "grad_norm": 0.46415287256240845, "learning_rate": 1.8862520588492724e-05, "loss": 2.140933837890625, "step": 28600 }, { "epoch": 0.11398278322881596, "grad_norm": 0.5063691139221191, "learning_rate": 1.8860531362142636e-05, "loss": 2.168863067626953, "step": 28650 }, { "epoch": 0.11418170606167602, "grad_norm": 0.5380640029907227, "learning_rate": 1.885854213579255e-05, "loss": 2.193540344238281, "step": 28700 }, { "epoch": 0.11438062889453608, "grad_norm": 0.49407675862312317, "learning_rate": 1.885655290944246e-05, "loss": 2.176432342529297, "step": 28750 }, { "epoch": 0.11457955172739615, "grad_norm": 0.5572803020477295, "learning_rate": 1.8854563683092373e-05, "loss": 2.2139537048339846, "step": 28800 }, { "epoch": 0.11477847456025621, "grad_norm": 0.5513818860054016, "learning_rate": 1.8852574456742285e-05, "loss": 2.148565368652344, "step": 28850 }, { "epoch": 0.11497739739311627, "grad_norm": 0.5379766821861267, "learning_rate": 1.8850585230392197e-05, "loss": 2.208501892089844, "step": 28900 }, { "epoch": 0.11517632022597633, "grad_norm": 0.46882352232933044, "learning_rate": 1.884859600404211e-05, "loss": 2.210621643066406, "step": 28950 }, { "epoch": 0.1153752430588364, "grad_norm": 0.5301218032836914, "learning_rate": 1.884660677769202e-05, "loss": 2.1886981201171873, "step": 29000 }, { "epoch": 0.11557416589169646, "grad_norm": 0.5087281465530396, "learning_rate": 1.8844617551341933e-05, "loss": 2.1777383422851564, "step": 29050 }, { "epoch": 0.11577308872455652, "grad_norm": 0.5582488775253296, "learning_rate": 1.8842628324991846e-05, "loss": 2.2661729431152344, "step": 29100 }, { "epoch": 0.11597201155741658, "grad_norm": 0.5703056454658508, "learning_rate": 1.8840639098641758e-05, "loss": 2.1878868103027345, "step": 29150 }, { "epoch": 0.11617093439027665, "grad_norm": 0.5708887577056885, "learning_rate": 1.883864987229167e-05, "loss": 2.210337371826172, "step": 29200 }, { "epoch": 0.11636985722313671, "grad_norm": 0.48165586590766907, "learning_rate": 1.8836660645941582e-05, "loss": 2.1643089294433593, "step": 29250 }, { "epoch": 0.11656878005599677, "grad_norm": 0.5108463764190674, "learning_rate": 1.8834671419591494e-05, "loss": 2.1968260192871094, "step": 29300 }, { "epoch": 0.11676770288885684, "grad_norm": 0.5567777752876282, "learning_rate": 1.8832682193241406e-05, "loss": 2.242270965576172, "step": 29350 }, { "epoch": 0.1169666257217169, "grad_norm": 0.4966385066509247, "learning_rate": 1.883069296689132e-05, "loss": 2.1549700927734374, "step": 29400 }, { "epoch": 0.11716554855457696, "grad_norm": 0.49657946825027466, "learning_rate": 1.882870374054123e-05, "loss": 2.22048828125, "step": 29450 }, { "epoch": 0.11736447138743702, "grad_norm": 0.5250135064125061, "learning_rate": 1.8826714514191143e-05, "loss": 2.174252471923828, "step": 29500 }, { "epoch": 0.11756339422029709, "grad_norm": 0.5657070279121399, "learning_rate": 1.8824725287841055e-05, "loss": 2.13904541015625, "step": 29550 }, { "epoch": 0.11776231705315715, "grad_norm": 0.5295047163963318, "learning_rate": 1.8822736061490967e-05, "loss": 2.111061859130859, "step": 29600 }, { "epoch": 0.11796123988601721, "grad_norm": 0.6235729455947876, "learning_rate": 1.882074683514088e-05, "loss": 2.220480499267578, "step": 29650 }, { "epoch": 0.11816016271887728, "grad_norm": 0.5482920408248901, "learning_rate": 1.881875760879079e-05, "loss": 2.1231219482421877, "step": 29700 }, { "epoch": 0.11835908555173734, "grad_norm": 0.4552220106124878, "learning_rate": 1.8816768382440703e-05, "loss": 2.20981201171875, "step": 29750 }, { "epoch": 0.1185580083845974, "grad_norm": 0.4956287741661072, "learning_rate": 1.8814779156090616e-05, "loss": 2.252241973876953, "step": 29800 }, { "epoch": 0.11875693121745746, "grad_norm": 0.5354408025741577, "learning_rate": 1.8812789929740528e-05, "loss": 2.2146624755859374, "step": 29850 }, { "epoch": 0.11895585405031753, "grad_norm": 0.5344682335853577, "learning_rate": 1.881080070339044e-05, "loss": 2.1856622314453125, "step": 29900 }, { "epoch": 0.11915477688317759, "grad_norm": 0.5377901792526245, "learning_rate": 1.8808811477040352e-05, "loss": 2.1869265747070314, "step": 29950 }, { "epoch": 0.11935369971603765, "grad_norm": 0.491577684879303, "learning_rate": 1.880682225069026e-05, "loss": 2.212216033935547, "step": 30000 }, { "epoch": 0.11955262254889772, "grad_norm": 0.511406421661377, "learning_rate": 1.8804833024340173e-05, "loss": 2.1752688598632814, "step": 30050 }, { "epoch": 0.11975154538175778, "grad_norm": 0.5693467259407043, "learning_rate": 1.880284379799009e-05, "loss": 2.1894029235839843, "step": 30100 }, { "epoch": 0.11995046821461784, "grad_norm": 0.4839550852775574, "learning_rate": 1.880085457164e-05, "loss": 2.2015145874023436, "step": 30150 }, { "epoch": 0.1201493910474779, "grad_norm": 0.5705762505531311, "learning_rate": 1.8798865345289913e-05, "loss": 2.237238006591797, "step": 30200 }, { "epoch": 0.12034831388033797, "grad_norm": 0.5506167411804199, "learning_rate": 1.8796876118939825e-05, "loss": 2.176952209472656, "step": 30250 }, { "epoch": 0.12054723671319803, "grad_norm": 0.5704229474067688, "learning_rate": 1.8794886892589737e-05, "loss": 2.1433824157714843, "step": 30300 }, { "epoch": 0.1207461595460581, "grad_norm": 0.5659142136573792, "learning_rate": 1.8792897666239646e-05, "loss": 2.1825413513183594, "step": 30350 }, { "epoch": 0.12094508237891816, "grad_norm": 0.5055663585662842, "learning_rate": 1.8790908439889558e-05, "loss": 2.1980409240722656, "step": 30400 }, { "epoch": 0.12114400521177822, "grad_norm": 0.6047959327697754, "learning_rate": 1.878891921353947e-05, "loss": 2.1349932861328127, "step": 30450 }, { "epoch": 0.12134292804463828, "grad_norm": 0.6000716090202332, "learning_rate": 1.8786929987189386e-05, "loss": 2.2185733032226564, "step": 30500 }, { "epoch": 0.12154185087749834, "grad_norm": 0.6077626943588257, "learning_rate": 1.8784940760839298e-05, "loss": 2.2056739807128904, "step": 30550 }, { "epoch": 0.12174077371035841, "grad_norm": 0.5744892954826355, "learning_rate": 1.878295153448921e-05, "loss": 2.1481951904296874, "step": 30600 }, { "epoch": 0.12193969654321847, "grad_norm": 0.5395755767822266, "learning_rate": 1.878096230813912e-05, "loss": 2.1451673889160157, "step": 30650 }, { "epoch": 0.12213861937607853, "grad_norm": 0.5584396719932556, "learning_rate": 1.877897308178903e-05, "loss": 2.1642588806152343, "step": 30700 }, { "epoch": 0.1223375422089386, "grad_norm": 0.5187245011329651, "learning_rate": 1.8776983855438943e-05, "loss": 2.188314666748047, "step": 30750 }, { "epoch": 0.12253646504179866, "grad_norm": 0.5554862022399902, "learning_rate": 1.8774994629088855e-05, "loss": 2.168018035888672, "step": 30800 }, { "epoch": 0.12273538787465872, "grad_norm": 0.47904837131500244, "learning_rate": 1.8773005402738767e-05, "loss": 2.2101637268066407, "step": 30850 }, { "epoch": 0.12293431070751878, "grad_norm": 0.5011553764343262, "learning_rate": 1.877101617638868e-05, "loss": 2.1824513244628907, "step": 30900 }, { "epoch": 0.12313323354037885, "grad_norm": 0.477346807718277, "learning_rate": 1.8769026950038595e-05, "loss": 2.1872528076171873, "step": 30950 }, { "epoch": 0.12333215637323891, "grad_norm": 0.5362797975540161, "learning_rate": 1.8767037723688504e-05, "loss": 2.1490249633789062, "step": 31000 }, { "epoch": 0.12353107920609897, "grad_norm": 0.5279614329338074, "learning_rate": 1.8765048497338416e-05, "loss": 2.1785911560058593, "step": 31050 }, { "epoch": 0.12373000203895904, "grad_norm": 0.5376858711242676, "learning_rate": 1.8763059270988328e-05, "loss": 2.176360626220703, "step": 31100 }, { "epoch": 0.1239289248718191, "grad_norm": 0.5158880949020386, "learning_rate": 1.876107004463824e-05, "loss": 2.1287435913085937, "step": 31150 }, { "epoch": 0.12412784770467916, "grad_norm": 0.6041430830955505, "learning_rate": 1.8759080818288152e-05, "loss": 2.155021209716797, "step": 31200 }, { "epoch": 0.12432677053753922, "grad_norm": 0.5980244874954224, "learning_rate": 1.8757091591938064e-05, "loss": 2.157806396484375, "step": 31250 }, { "epoch": 0.12452569337039929, "grad_norm": 0.5445902943611145, "learning_rate": 1.8755102365587977e-05, "loss": 2.168020935058594, "step": 31300 }, { "epoch": 0.12472461620325935, "grad_norm": 0.485151082277298, "learning_rate": 1.875311313923789e-05, "loss": 2.1646234130859376, "step": 31350 }, { "epoch": 0.12492353903611941, "grad_norm": 0.553519606590271, "learning_rate": 1.87511239128878e-05, "loss": 2.2177861022949217, "step": 31400 }, { "epoch": 0.1251224618689795, "grad_norm": 0.5581177473068237, "learning_rate": 1.8749134686537713e-05, "loss": 2.1518345642089844, "step": 31450 }, { "epoch": 0.12532138470183954, "grad_norm": 0.6392439603805542, "learning_rate": 1.8747145460187625e-05, "loss": 2.2298741149902344, "step": 31500 }, { "epoch": 0.1255203075346996, "grad_norm": 0.5985459089279175, "learning_rate": 1.8745156233837537e-05, "loss": 2.1975074768066407, "step": 31550 }, { "epoch": 0.12571923036755966, "grad_norm": 0.5324075222015381, "learning_rate": 1.874316700748745e-05, "loss": 2.22779541015625, "step": 31600 }, { "epoch": 0.1259181532004197, "grad_norm": 0.5756731629371643, "learning_rate": 1.874117778113736e-05, "loss": 2.1729159545898438, "step": 31650 }, { "epoch": 0.1261170760332798, "grad_norm": 0.363854318857193, "learning_rate": 1.8739188554787274e-05, "loss": 2.198079376220703, "step": 31700 }, { "epoch": 0.12631599886613984, "grad_norm": 0.4574756622314453, "learning_rate": 1.8737199328437186e-05, "loss": 2.1741954040527345, "step": 31750 }, { "epoch": 0.12651492169899992, "grad_norm": 0.5937820672988892, "learning_rate": 1.8735210102087098e-05, "loss": 2.1574247741699217, "step": 31800 }, { "epoch": 0.12671384453185996, "grad_norm": 0.49845123291015625, "learning_rate": 1.873322087573701e-05, "loss": 2.1815570068359373, "step": 31850 }, { "epoch": 0.12691276736472004, "grad_norm": 0.5347491502761841, "learning_rate": 1.8731271433913924e-05, "loss": 2.192160797119141, "step": 31900 }, { "epoch": 0.1271116901975801, "grad_norm": 0.6149244904518127, "learning_rate": 1.8729282207563836e-05, "loss": 2.1725440979003907, "step": 31950 }, { "epoch": 0.12731061303044017, "grad_norm": 0.5274011492729187, "learning_rate": 1.8727292981213748e-05, "loss": 2.151186981201172, "step": 32000 }, { "epoch": 0.12750953586330022, "grad_norm": 0.5296633839607239, "learning_rate": 1.872530375486366e-05, "loss": 2.191092681884766, "step": 32050 }, { "epoch": 0.1277084586961603, "grad_norm": 0.5036041140556335, "learning_rate": 1.8723314528513572e-05, "loss": 2.2033567810058594, "step": 32100 }, { "epoch": 0.12790738152902034, "grad_norm": 0.5113654136657715, "learning_rate": 1.8721325302163484e-05, "loss": 2.1567733764648436, "step": 32150 }, { "epoch": 0.12810630436188042, "grad_norm": 0.48790356516838074, "learning_rate": 1.8719336075813397e-05, "loss": 2.19705810546875, "step": 32200 }, { "epoch": 0.12830522719474047, "grad_norm": 0.49298134446144104, "learning_rate": 1.871734684946331e-05, "loss": 2.1605485534667968, "step": 32250 }, { "epoch": 0.12850415002760054, "grad_norm": 0.5475209355354309, "learning_rate": 1.871535762311322e-05, "loss": 2.2209259033203126, "step": 32300 }, { "epoch": 0.1287030728604606, "grad_norm": 0.497631311416626, "learning_rate": 1.8713368396763133e-05, "loss": 2.2171685791015623, "step": 32350 }, { "epoch": 0.12890199569332067, "grad_norm": 0.4855594038963318, "learning_rate": 1.8711379170413045e-05, "loss": 2.133531494140625, "step": 32400 }, { "epoch": 0.12910091852618072, "grad_norm": 0.5599591732025146, "learning_rate": 1.8709389944062957e-05, "loss": 2.204909362792969, "step": 32450 }, { "epoch": 0.1292998413590408, "grad_norm": 0.5820069313049316, "learning_rate": 1.870740071771287e-05, "loss": 2.1678224182128907, "step": 32500 }, { "epoch": 0.12949876419190084, "grad_norm": 0.4789052903652191, "learning_rate": 1.870541149136278e-05, "loss": 2.218444061279297, "step": 32550 }, { "epoch": 0.12969768702476092, "grad_norm": 0.6038575172424316, "learning_rate": 1.870342226501269e-05, "loss": 2.153567199707031, "step": 32600 }, { "epoch": 0.12989660985762097, "grad_norm": 0.6868671774864197, "learning_rate": 1.8701433038662603e-05, "loss": 2.1670233154296876, "step": 32650 }, { "epoch": 0.13009553269048105, "grad_norm": 0.4067879915237427, "learning_rate": 1.8699443812312518e-05, "loss": 2.2217982482910155, "step": 32700 }, { "epoch": 0.1302944555233411, "grad_norm": 0.5354380011558533, "learning_rate": 1.869745458596243e-05, "loss": 2.149874420166016, "step": 32750 }, { "epoch": 0.13049337835620117, "grad_norm": 0.5259619951248169, "learning_rate": 1.8695465359612342e-05, "loss": 2.233143310546875, "step": 32800 }, { "epoch": 0.13069230118906122, "grad_norm": 0.4410480856895447, "learning_rate": 1.8693476133262254e-05, "loss": 2.168966979980469, "step": 32850 }, { "epoch": 0.1308912240219213, "grad_norm": 0.4852573871612549, "learning_rate": 1.8691486906912167e-05, "loss": 2.184669647216797, "step": 32900 }, { "epoch": 0.13109014685478135, "grad_norm": 0.5847207307815552, "learning_rate": 1.8689497680562075e-05, "loss": 2.2479347229003905, "step": 32950 }, { "epoch": 0.13128906968764142, "grad_norm": 0.501075267791748, "learning_rate": 1.8687508454211988e-05, "loss": 2.2366156005859374, "step": 33000 }, { "epoch": 0.13148799252050147, "grad_norm": 0.47724413871765137, "learning_rate": 1.86855192278619e-05, "loss": 2.157849884033203, "step": 33050 }, { "epoch": 0.13168691535336155, "grad_norm": 0.5641265511512756, "learning_rate": 1.8683530001511815e-05, "loss": 2.1311639404296874, "step": 33100 }, { "epoch": 0.1318858381862216, "grad_norm": 0.524878978729248, "learning_rate": 1.8681540775161727e-05, "loss": 2.142886505126953, "step": 33150 }, { "epoch": 0.13208476101908168, "grad_norm": 0.5281782150268555, "learning_rate": 1.867955154881164e-05, "loss": 2.1307553100585936, "step": 33200 }, { "epoch": 0.13228368385194172, "grad_norm": 0.5253816843032837, "learning_rate": 1.8677562322461548e-05, "loss": 2.189570007324219, "step": 33250 }, { "epoch": 0.1324826066848018, "grad_norm": 0.4937022924423218, "learning_rate": 1.867557309611146e-05, "loss": 2.1736814880371096, "step": 33300 }, { "epoch": 0.13268152951766185, "grad_norm": 0.5371102094650269, "learning_rate": 1.8673583869761373e-05, "loss": 2.151394958496094, "step": 33350 }, { "epoch": 0.13288045235052193, "grad_norm": 0.4956592619419098, "learning_rate": 1.8671594643411285e-05, "loss": 2.1409538269042967, "step": 33400 }, { "epoch": 0.13307937518338198, "grad_norm": 0.5489466786384583, "learning_rate": 1.8669605417061197e-05, "loss": 2.154979553222656, "step": 33450 }, { "epoch": 0.13327829801624205, "grad_norm": 0.476563423871994, "learning_rate": 1.8667616190711112e-05, "loss": 2.1292985534667968, "step": 33500 }, { "epoch": 0.1334772208491021, "grad_norm": 0.6655639410018921, "learning_rate": 1.8665626964361025e-05, "loss": 2.231531982421875, "step": 33550 }, { "epoch": 0.13367614368196218, "grad_norm": 0.5134598016738892, "learning_rate": 1.8663637738010933e-05, "loss": 2.1393133544921876, "step": 33600 }, { "epoch": 0.13387506651482223, "grad_norm": 0.47659748792648315, "learning_rate": 1.8661648511660845e-05, "loss": 2.177981109619141, "step": 33650 }, { "epoch": 0.1340739893476823, "grad_norm": 0.5587292313575745, "learning_rate": 1.8659659285310758e-05, "loss": 2.2105055236816407, "step": 33700 }, { "epoch": 0.13427291218054235, "grad_norm": 0.4891665577888489, "learning_rate": 1.865767005896067e-05, "loss": 2.212679595947266, "step": 33750 }, { "epoch": 0.13447183501340243, "grad_norm": 0.481309175491333, "learning_rate": 1.8655680832610582e-05, "loss": 2.2074984741210937, "step": 33800 }, { "epoch": 0.13467075784626248, "grad_norm": 0.5388895273208618, "learning_rate": 1.86537313907875e-05, "loss": 2.203882598876953, "step": 33850 }, { "epoch": 0.13486968067912256, "grad_norm": 0.5195946097373962, "learning_rate": 1.8651742164437408e-05, "loss": 2.164841766357422, "step": 33900 }, { "epoch": 0.1350686035119826, "grad_norm": 0.48848336935043335, "learning_rate": 1.864975293808732e-05, "loss": 2.2345675659179687, "step": 33950 }, { "epoch": 0.13526752634484268, "grad_norm": 0.5453406572341919, "learning_rate": 1.8647763711737232e-05, "loss": 2.1737550354003905, "step": 34000 }, { "epoch": 0.13546644917770273, "grad_norm": 0.5714558959007263, "learning_rate": 1.8645774485387144e-05, "loss": 2.137253723144531, "step": 34050 }, { "epoch": 0.1356653720105628, "grad_norm": 0.5850618481636047, "learning_rate": 1.8643785259037056e-05, "loss": 2.1590542602539062, "step": 34100 }, { "epoch": 0.13586429484342286, "grad_norm": 0.4666570723056793, "learning_rate": 1.8641796032686968e-05, "loss": 2.127194366455078, "step": 34150 }, { "epoch": 0.13606321767628293, "grad_norm": 0.6052256226539612, "learning_rate": 1.863980680633688e-05, "loss": 2.207763366699219, "step": 34200 }, { "epoch": 0.13626214050914298, "grad_norm": 0.4508417844772339, "learning_rate": 1.8637817579986793e-05, "loss": 2.176491241455078, "step": 34250 }, { "epoch": 0.13646106334200306, "grad_norm": 0.5099793672561646, "learning_rate": 1.8635828353636705e-05, "loss": 2.1613505554199217, "step": 34300 }, { "epoch": 0.1366599861748631, "grad_norm": 0.5332155823707581, "learning_rate": 1.8633839127286617e-05, "loss": 2.243791961669922, "step": 34350 }, { "epoch": 0.13685890900772318, "grad_norm": 0.49282997846603394, "learning_rate": 1.863184990093653e-05, "loss": 2.16620361328125, "step": 34400 }, { "epoch": 0.13705783184058323, "grad_norm": 0.5712761282920837, "learning_rate": 1.862986067458644e-05, "loss": 2.2024130249023437, "step": 34450 }, { "epoch": 0.1372567546734433, "grad_norm": 0.5402825474739075, "learning_rate": 1.8627871448236353e-05, "loss": 2.1810952758789064, "step": 34500 }, { "epoch": 0.13745567750630336, "grad_norm": 0.5774102210998535, "learning_rate": 1.8625882221886265e-05, "loss": 2.1258447265625, "step": 34550 }, { "epoch": 0.13765460033916344, "grad_norm": 0.5759469866752625, "learning_rate": 1.8623892995536178e-05, "loss": 2.199498596191406, "step": 34600 }, { "epoch": 0.13785352317202348, "grad_norm": 0.5580266118049622, "learning_rate": 1.862190376918609e-05, "loss": 2.1611781311035156, "step": 34650 }, { "epoch": 0.13805244600488356, "grad_norm": 0.5348168611526489, "learning_rate": 1.8619914542836002e-05, "loss": 2.186202087402344, "step": 34700 }, { "epoch": 0.1382513688377436, "grad_norm": 0.5045925378799438, "learning_rate": 1.8617925316485914e-05, "loss": 2.1726724243164064, "step": 34750 }, { "epoch": 0.1384502916706037, "grad_norm": 0.6941531300544739, "learning_rate": 1.8615936090135826e-05, "loss": 2.139268341064453, "step": 34800 }, { "epoch": 0.13864921450346374, "grad_norm": 0.4477648437023163, "learning_rate": 1.8613946863785738e-05, "loss": 2.1567646789550783, "step": 34850 }, { "epoch": 0.1388481373363238, "grad_norm": 0.6380840539932251, "learning_rate": 1.861195763743565e-05, "loss": 2.1817689514160157, "step": 34900 }, { "epoch": 0.13904706016918386, "grad_norm": 0.5002685189247131, "learning_rate": 1.8609968411085563e-05, "loss": 2.125812683105469, "step": 34950 }, { "epoch": 0.13924598300204394, "grad_norm": 0.5294092297554016, "learning_rate": 1.8607979184735475e-05, "loss": 2.2321926879882814, "step": 35000 }, { "epoch": 0.139444905834904, "grad_norm": 0.5492232441902161, "learning_rate": 1.8605989958385387e-05, "loss": 2.1485714721679687, "step": 35050 }, { "epoch": 0.13964382866776406, "grad_norm": 0.5781382322311401, "learning_rate": 1.86040007320353e-05, "loss": 2.189228057861328, "step": 35100 }, { "epoch": 0.1398427515006241, "grad_norm": 0.5632701516151428, "learning_rate": 1.860201150568521e-05, "loss": 2.1154583740234374, "step": 35150 }, { "epoch": 0.1400416743334842, "grad_norm": 0.451508104801178, "learning_rate": 1.860002227933512e-05, "loss": 2.1469572448730467, "step": 35200 }, { "epoch": 0.14024059716634424, "grad_norm": 0.45875653624534607, "learning_rate": 1.8598033052985035e-05, "loss": 2.1618345642089842, "step": 35250 }, { "epoch": 0.14043951999920432, "grad_norm": 0.48653337359428406, "learning_rate": 1.8596043826634948e-05, "loss": 2.264252166748047, "step": 35300 }, { "epoch": 0.14063844283206436, "grad_norm": 0.47829413414001465, "learning_rate": 1.859405460028486e-05, "loss": 2.1166526794433596, "step": 35350 }, { "epoch": 0.14083736566492444, "grad_norm": 0.6112663745880127, "learning_rate": 1.8592065373934772e-05, "loss": 2.194012451171875, "step": 35400 }, { "epoch": 0.1410362884977845, "grad_norm": 0.4994930624961853, "learning_rate": 1.8590076147584684e-05, "loss": 2.1654962158203124, "step": 35450 }, { "epoch": 0.14123521133064457, "grad_norm": 0.518126368522644, "learning_rate": 1.8588086921234596e-05, "loss": 2.1487379455566407, "step": 35500 }, { "epoch": 0.14143413416350462, "grad_norm": 0.6314736604690552, "learning_rate": 1.8586097694884505e-05, "loss": 2.1038522338867187, "step": 35550 }, { "epoch": 0.1416330569963647, "grad_norm": 0.4930416941642761, "learning_rate": 1.8584108468534417e-05, "loss": 2.1425904846191406, "step": 35600 }, { "epoch": 0.14183197982922474, "grad_norm": 0.5411590933799744, "learning_rate": 1.8582119242184333e-05, "loss": 2.1395449829101563, "step": 35650 }, { "epoch": 0.14203090266208482, "grad_norm": 0.5822312235832214, "learning_rate": 1.8580130015834245e-05, "loss": 2.161257781982422, "step": 35700 }, { "epoch": 0.14222982549494487, "grad_norm": 0.573778510093689, "learning_rate": 1.8578140789484157e-05, "loss": 2.1470941162109374, "step": 35750 }, { "epoch": 0.14242874832780494, "grad_norm": 0.5228402614593506, "learning_rate": 1.857615156313407e-05, "loss": 2.1398817443847657, "step": 35800 }, { "epoch": 0.142627671160665, "grad_norm": 0.5131897330284119, "learning_rate": 1.8574162336783978e-05, "loss": 2.144942169189453, "step": 35850 }, { "epoch": 0.14282659399352507, "grad_norm": 0.5600727200508118, "learning_rate": 1.857217311043389e-05, "loss": 2.200516662597656, "step": 35900 }, { "epoch": 0.14302551682638512, "grad_norm": 0.5624778866767883, "learning_rate": 1.8570183884083802e-05, "loss": 2.1801702880859377, "step": 35950 }, { "epoch": 0.1432244396592452, "grad_norm": 0.4576130211353302, "learning_rate": 1.8568194657733714e-05, "loss": 2.1693046569824217, "step": 36000 }, { "epoch": 0.14342336249210524, "grad_norm": 0.46616441011428833, "learning_rate": 1.856620543138363e-05, "loss": 2.206753997802734, "step": 36050 }, { "epoch": 0.14362228532496532, "grad_norm": 0.6365834474563599, "learning_rate": 1.8564216205033542e-05, "loss": 2.219575958251953, "step": 36100 }, { "epoch": 0.14382120815782537, "grad_norm": 0.5633766055107117, "learning_rate": 1.8562226978683454e-05, "loss": 2.1682542419433593, "step": 36150 }, { "epoch": 0.14402013099068545, "grad_norm": 0.5944415926933289, "learning_rate": 1.8560237752333363e-05, "loss": 2.1478091430664064, "step": 36200 }, { "epoch": 0.1442190538235455, "grad_norm": 0.7217615842819214, "learning_rate": 1.8558248525983275e-05, "loss": 2.1634756469726564, "step": 36250 }, { "epoch": 0.14441797665640557, "grad_norm": 0.5382615923881531, "learning_rate": 1.8556259299633187e-05, "loss": 2.198925323486328, "step": 36300 }, { "epoch": 0.14461689948926562, "grad_norm": 0.4663105309009552, "learning_rate": 1.85542700732831e-05, "loss": 2.2041712951660157, "step": 36350 }, { "epoch": 0.1448158223221257, "grad_norm": 0.5243980288505554, "learning_rate": 1.855228084693301e-05, "loss": 2.193206481933594, "step": 36400 }, { "epoch": 0.14501474515498575, "grad_norm": 0.5232993960380554, "learning_rate": 1.8550291620582927e-05, "loss": 2.1525440979003907, "step": 36450 }, { "epoch": 0.14521366798784582, "grad_norm": 0.5032647848129272, "learning_rate": 1.8548302394232836e-05, "loss": 2.165350341796875, "step": 36500 }, { "epoch": 0.14541259082070587, "grad_norm": 0.5291615724563599, "learning_rate": 1.8546313167882748e-05, "loss": 2.2130397033691405, "step": 36550 }, { "epoch": 0.14561151365356595, "grad_norm": 0.48761990666389465, "learning_rate": 1.854432394153266e-05, "loss": 2.156594543457031, "step": 36600 }, { "epoch": 0.145810436486426, "grad_norm": 0.46036043763160706, "learning_rate": 1.8542334715182572e-05, "loss": 2.1347232055664063, "step": 36650 }, { "epoch": 0.14600935931928608, "grad_norm": 0.5536908507347107, "learning_rate": 1.8540345488832484e-05, "loss": 2.15889892578125, "step": 36700 }, { "epoch": 0.14620828215214612, "grad_norm": 0.4602327346801758, "learning_rate": 1.8538356262482396e-05, "loss": 2.2102847290039063, "step": 36750 }, { "epoch": 0.1464072049850062, "grad_norm": 0.5986318588256836, "learning_rate": 1.853636703613231e-05, "loss": 2.1078199768066406, "step": 36800 }, { "epoch": 0.14660612781786625, "grad_norm": 0.5661736726760864, "learning_rate": 1.853437780978222e-05, "loss": 2.1715008544921877, "step": 36850 }, { "epoch": 0.14680505065072633, "grad_norm": 0.5083137154579163, "learning_rate": 1.8532388583432133e-05, "loss": 2.196439666748047, "step": 36900 }, { "epoch": 0.14700397348358638, "grad_norm": 0.5651976466178894, "learning_rate": 1.8530399357082045e-05, "loss": 2.2202955627441407, "step": 36950 }, { "epoch": 0.14720289631644645, "grad_norm": 0.47674089670181274, "learning_rate": 1.8528410130731957e-05, "loss": 2.147147369384766, "step": 37000 }, { "epoch": 0.1474018191493065, "grad_norm": 0.41821473836898804, "learning_rate": 1.852642090438187e-05, "loss": 2.154086151123047, "step": 37050 }, { "epoch": 0.14760074198216658, "grad_norm": 0.5619843602180481, "learning_rate": 1.852443167803178e-05, "loss": 2.201946563720703, "step": 37100 }, { "epoch": 0.14779966481502663, "grad_norm": 0.5806388854980469, "learning_rate": 1.8522442451681694e-05, "loss": 2.1645680236816407, "step": 37150 }, { "epoch": 0.1479985876478867, "grad_norm": 0.46140459179878235, "learning_rate": 1.8520453225331606e-05, "loss": 2.181314239501953, "step": 37200 }, { "epoch": 0.14819751048074675, "grad_norm": 0.5899180769920349, "learning_rate": 1.8518463998981518e-05, "loss": 2.1947332763671876, "step": 37250 }, { "epoch": 0.14839643331360683, "grad_norm": 0.5271891355514526, "learning_rate": 1.851647477263143e-05, "loss": 2.1226316833496095, "step": 37300 }, { "epoch": 0.14859535614646688, "grad_norm": 0.5666892528533936, "learning_rate": 1.8514485546281342e-05, "loss": 2.1530767822265626, "step": 37350 }, { "epoch": 0.14879427897932695, "grad_norm": 0.5814537405967712, "learning_rate": 1.8512496319931254e-05, "loss": 2.1481103515625, "step": 37400 }, { "epoch": 0.148993201812187, "grad_norm": 0.556914746761322, "learning_rate": 1.8510507093581163e-05, "loss": 2.1296116638183595, "step": 37450 }, { "epoch": 0.14919212464504708, "grad_norm": 0.535162091255188, "learning_rate": 1.850851786723108e-05, "loss": 2.151704864501953, "step": 37500 }, { "epoch": 0.14939104747790713, "grad_norm": 0.5514208078384399, "learning_rate": 1.850652864088099e-05, "loss": 2.1244245910644532, "step": 37550 }, { "epoch": 0.1495899703107672, "grad_norm": 0.47839969396591187, "learning_rate": 1.8504539414530903e-05, "loss": 2.1544000244140626, "step": 37600 }, { "epoch": 0.14978889314362726, "grad_norm": 0.49943825602531433, "learning_rate": 1.8502550188180815e-05, "loss": 2.1320985412597655, "step": 37650 }, { "epoch": 0.14998781597648733, "grad_norm": 0.5179593563079834, "learning_rate": 1.8500560961830727e-05, "loss": 2.1551641845703124, "step": 37700 }, { "epoch": 0.15018673880934738, "grad_norm": 0.46034160256385803, "learning_rate": 1.849857173548064e-05, "loss": 2.153864440917969, "step": 37750 }, { "epoch": 0.15038566164220746, "grad_norm": 0.5366671085357666, "learning_rate": 1.8496582509130548e-05, "loss": 2.13669921875, "step": 37800 }, { "epoch": 0.1505845844750675, "grad_norm": 0.5522403717041016, "learning_rate": 1.849459328278046e-05, "loss": 2.156935577392578, "step": 37850 }, { "epoch": 0.15078350730792758, "grad_norm": 0.6047018766403198, "learning_rate": 1.8492604056430376e-05, "loss": 2.1931582641601564, "step": 37900 }, { "epoch": 0.15098243014078763, "grad_norm": 0.5845444798469543, "learning_rate": 1.8490614830080288e-05, "loss": 2.1645005798339843, "step": 37950 }, { "epoch": 0.1511813529736477, "grad_norm": 0.46167877316474915, "learning_rate": 1.84886256037302e-05, "loss": 2.183463592529297, "step": 38000 }, { "epoch": 0.15138027580650776, "grad_norm": 0.5102078914642334, "learning_rate": 1.8486636377380112e-05, "loss": 2.139849853515625, "step": 38050 }, { "epoch": 0.15157919863936783, "grad_norm": 0.5184875130653381, "learning_rate": 1.848464715103002e-05, "loss": 2.157329254150391, "step": 38100 }, { "epoch": 0.15177812147222788, "grad_norm": 0.46233487129211426, "learning_rate": 1.8482657924679933e-05, "loss": 2.182453308105469, "step": 38150 }, { "epoch": 0.15197704430508796, "grad_norm": 0.5487840175628662, "learning_rate": 1.8480668698329845e-05, "loss": 2.1270643615722657, "step": 38200 }, { "epoch": 0.152175967137948, "grad_norm": 0.4635883867740631, "learning_rate": 1.8478679471979757e-05, "loss": 2.1316015625, "step": 38250 }, { "epoch": 0.1523748899708081, "grad_norm": 0.5410680174827576, "learning_rate": 1.8476690245629673e-05, "loss": 2.128941650390625, "step": 38300 }, { "epoch": 0.15257381280366814, "grad_norm": 0.6083848476409912, "learning_rate": 1.8474701019279585e-05, "loss": 2.187699890136719, "step": 38350 }, { "epoch": 0.1527727356365282, "grad_norm": 0.546951413154602, "learning_rate": 1.8472711792929497e-05, "loss": 2.1795204162597654, "step": 38400 }, { "epoch": 0.15297165846938826, "grad_norm": 0.4971100389957428, "learning_rate": 1.8470722566579406e-05, "loss": 2.0930192565917967, "step": 38450 }, { "epoch": 0.15317058130224834, "grad_norm": 0.4805755317211151, "learning_rate": 1.8468733340229318e-05, "loss": 2.187390899658203, "step": 38500 }, { "epoch": 0.1533695041351084, "grad_norm": 0.573282778263092, "learning_rate": 1.846674411387923e-05, "loss": 2.1438340759277343, "step": 38550 }, { "epoch": 0.15356842696796846, "grad_norm": 0.5589561462402344, "learning_rate": 1.8464754887529142e-05, "loss": 2.1168588256835936, "step": 38600 }, { "epoch": 0.1537673498008285, "grad_norm": 0.5510828495025635, "learning_rate": 1.8462765661179055e-05, "loss": 2.1511102294921876, "step": 38650 }, { "epoch": 0.1539662726336886, "grad_norm": 0.5737041234970093, "learning_rate": 1.846077643482897e-05, "loss": 2.1396270751953126, "step": 38700 }, { "epoch": 0.15416519546654864, "grad_norm": 0.5363348722457886, "learning_rate": 1.845878720847888e-05, "loss": 2.1079803466796876, "step": 38750 }, { "epoch": 0.15436411829940871, "grad_norm": 0.5207923650741577, "learning_rate": 1.845679798212879e-05, "loss": 2.1603125, "step": 38800 }, { "epoch": 0.15456304113226876, "grad_norm": 0.5074342489242554, "learning_rate": 1.8454848540305705e-05, "loss": 2.1293569946289064, "step": 38850 }, { "epoch": 0.15476196396512884, "grad_norm": 0.5830539464950562, "learning_rate": 1.8452859313955617e-05, "loss": 2.2031491088867186, "step": 38900 }, { "epoch": 0.1549608867979889, "grad_norm": 0.6363376975059509, "learning_rate": 1.845087008760553e-05, "loss": 2.2802389526367186, "step": 38950 }, { "epoch": 0.15515980963084897, "grad_norm": 0.5601738095283508, "learning_rate": 1.844888086125544e-05, "loss": 2.157103271484375, "step": 39000 }, { "epoch": 0.15535873246370902, "grad_norm": 0.5461059808731079, "learning_rate": 1.8446891634905357e-05, "loss": 2.120672149658203, "step": 39050 }, { "epoch": 0.1555576552965691, "grad_norm": 0.48031914234161377, "learning_rate": 1.8444902408555265e-05, "loss": 2.199144287109375, "step": 39100 }, { "epoch": 0.15575657812942914, "grad_norm": 0.647547721862793, "learning_rate": 1.8442913182205177e-05, "loss": 2.142056579589844, "step": 39150 }, { "epoch": 0.15595550096228922, "grad_norm": 0.5172239542007446, "learning_rate": 1.844092395585509e-05, "loss": 2.1915650939941407, "step": 39200 }, { "epoch": 0.15615442379514927, "grad_norm": 0.5577855110168457, "learning_rate": 1.8438934729505002e-05, "loss": 2.1744589233398437, "step": 39250 }, { "epoch": 0.15635334662800934, "grad_norm": 0.49216583371162415, "learning_rate": 1.8436945503154914e-05, "loss": 2.1649322509765625, "step": 39300 }, { "epoch": 0.1565522694608694, "grad_norm": 0.5098722577095032, "learning_rate": 1.8434956276804826e-05, "loss": 2.193984375, "step": 39350 }, { "epoch": 0.15675119229372947, "grad_norm": 0.610787034034729, "learning_rate": 1.8432967050454738e-05, "loss": 2.142711486816406, "step": 39400 }, { "epoch": 0.15695011512658952, "grad_norm": 0.4909246265888214, "learning_rate": 1.843097782410465e-05, "loss": 2.170638122558594, "step": 39450 }, { "epoch": 0.15714903795944957, "grad_norm": 0.529387354850769, "learning_rate": 1.8428988597754562e-05, "loss": 2.1489085388183593, "step": 39500 }, { "epoch": 0.15734796079230964, "grad_norm": 0.5634523630142212, "learning_rate": 1.8426999371404475e-05, "loss": 2.1517466735839843, "step": 39550 }, { "epoch": 0.1575468836251697, "grad_norm": 0.5602388978004456, "learning_rate": 1.8425010145054387e-05, "loss": 2.1450566101074218, "step": 39600 }, { "epoch": 0.15774580645802977, "grad_norm": 0.5599037408828735, "learning_rate": 1.84230209187043e-05, "loss": 2.1344581604003907, "step": 39650 }, { "epoch": 0.15794472929088982, "grad_norm": 0.5161935091018677, "learning_rate": 1.842103169235421e-05, "loss": 2.1877450561523437, "step": 39700 }, { "epoch": 0.1581436521237499, "grad_norm": 0.5297123193740845, "learning_rate": 1.8419042466004123e-05, "loss": 2.1518553161621092, "step": 39750 }, { "epoch": 0.15834257495660994, "grad_norm": 0.6005194783210754, "learning_rate": 1.8417053239654035e-05, "loss": 2.148242950439453, "step": 39800 }, { "epoch": 0.15854149778947002, "grad_norm": 0.5010674595832825, "learning_rate": 1.8415064013303947e-05, "loss": 2.218810729980469, "step": 39850 }, { "epoch": 0.15874042062233007, "grad_norm": 0.5238016247749329, "learning_rate": 1.841307478695386e-05, "loss": 2.1655929565429686, "step": 39900 }, { "epoch": 0.15893934345519015, "grad_norm": 0.6278696656227112, "learning_rate": 1.8411085560603772e-05, "loss": 2.2074554443359373, "step": 39950 }, { "epoch": 0.1591382662880502, "grad_norm": 0.5024388432502747, "learning_rate": 1.8409096334253684e-05, "loss": 2.190721130371094, "step": 40000 }, { "epoch": 0.15933718912091027, "grad_norm": 0.48540210723876953, "learning_rate": 1.8407107107903596e-05, "loss": 2.2003207397460938, "step": 40050 }, { "epoch": 0.15953611195377032, "grad_norm": 0.6029617786407471, "learning_rate": 1.8405117881553508e-05, "loss": 2.175911102294922, "step": 40100 }, { "epoch": 0.1597350347866304, "grad_norm": 0.49571964144706726, "learning_rate": 1.840312865520342e-05, "loss": 2.1573268127441407, "step": 40150 }, { "epoch": 0.15993395761949045, "grad_norm": 0.5085028409957886, "learning_rate": 1.8401139428853333e-05, "loss": 2.108054504394531, "step": 40200 }, { "epoch": 0.16013288045235052, "grad_norm": 0.5086105465888977, "learning_rate": 1.8399150202503245e-05, "loss": 2.1244049072265625, "step": 40250 }, { "epoch": 0.16033180328521057, "grad_norm": 0.5295969843864441, "learning_rate": 1.8397160976153157e-05, "loss": 2.1245350646972656, "step": 40300 }, { "epoch": 0.16053072611807065, "grad_norm": 0.42337608337402344, "learning_rate": 1.839517174980307e-05, "loss": 2.1985539245605468, "step": 40350 }, { "epoch": 0.1607296489509307, "grad_norm": 0.5186324715614319, "learning_rate": 1.8393182523452978e-05, "loss": 2.0919126892089843, "step": 40400 }, { "epoch": 0.16092857178379077, "grad_norm": 0.4460389018058777, "learning_rate": 1.8391193297102893e-05, "loss": 2.157334899902344, "step": 40450 }, { "epoch": 0.16112749461665082, "grad_norm": 0.44541773200035095, "learning_rate": 1.8389204070752805e-05, "loss": 2.130186767578125, "step": 40500 }, { "epoch": 0.1613264174495109, "grad_norm": 0.5319133400917053, "learning_rate": 1.8387214844402718e-05, "loss": 2.1504888916015625, "step": 40550 }, { "epoch": 0.16152534028237095, "grad_norm": 0.5810596942901611, "learning_rate": 1.838526540257963e-05, "loss": 2.1694981384277345, "step": 40600 }, { "epoch": 0.16172426311523103, "grad_norm": 0.5683962106704712, "learning_rate": 1.8383276176229543e-05, "loss": 2.1500694274902346, "step": 40650 }, { "epoch": 0.16192318594809108, "grad_norm": 0.5179654955863953, "learning_rate": 1.8381286949879452e-05, "loss": 2.1366615295410156, "step": 40700 }, { "epoch": 0.16212210878095115, "grad_norm": 0.563145101070404, "learning_rate": 1.8379297723529364e-05, "loss": 2.133911437988281, "step": 40750 }, { "epoch": 0.1623210316138112, "grad_norm": 0.48139455914497375, "learning_rate": 1.837730849717928e-05, "loss": 2.125212860107422, "step": 40800 }, { "epoch": 0.16251995444667128, "grad_norm": 0.5547031164169312, "learning_rate": 1.8375319270829192e-05, "loss": 2.210226745605469, "step": 40850 }, { "epoch": 0.16271887727953133, "grad_norm": 0.5316616296768188, "learning_rate": 1.8373330044479104e-05, "loss": 2.143583068847656, "step": 40900 }, { "epoch": 0.1629178001123914, "grad_norm": 0.49886417388916016, "learning_rate": 1.8371340818129016e-05, "loss": 2.1710246276855467, "step": 40950 }, { "epoch": 0.16311672294525145, "grad_norm": 0.5464217662811279, "learning_rate": 1.8369351591778928e-05, "loss": 2.093758544921875, "step": 41000 }, { "epoch": 0.16331564577811153, "grad_norm": 0.5953105688095093, "learning_rate": 1.8367362365428837e-05, "loss": 2.181264190673828, "step": 41050 }, { "epoch": 0.16351456861097158, "grad_norm": 0.5432385206222534, "learning_rate": 1.836537313907875e-05, "loss": 2.111508026123047, "step": 41100 }, { "epoch": 0.16371349144383165, "grad_norm": 0.5447415709495544, "learning_rate": 1.836338391272866e-05, "loss": 2.121979217529297, "step": 41150 }, { "epoch": 0.1639124142766917, "grad_norm": 0.5696507692337036, "learning_rate": 1.8361394686378577e-05, "loss": 2.12022705078125, "step": 41200 }, { "epoch": 0.16411133710955178, "grad_norm": 0.5090187788009644, "learning_rate": 1.835940546002849e-05, "loss": 2.1368730163574217, "step": 41250 }, { "epoch": 0.16431025994241183, "grad_norm": 0.4640263319015503, "learning_rate": 1.83574162336784e-05, "loss": 2.1371522521972657, "step": 41300 }, { "epoch": 0.1645091827752719, "grad_norm": 0.6140899062156677, "learning_rate": 1.835542700732831e-05, "loss": 2.1698487854003905, "step": 41350 }, { "epoch": 0.16470810560813195, "grad_norm": 0.5114386677742004, "learning_rate": 1.8353437780978222e-05, "loss": 2.1660557556152344, "step": 41400 }, { "epoch": 0.16490702844099203, "grad_norm": 0.5861048102378845, "learning_rate": 1.8351448554628134e-05, "loss": 2.1995115661621094, "step": 41450 }, { "epoch": 0.16510595127385208, "grad_norm": 0.602535605430603, "learning_rate": 1.8349459328278046e-05, "loss": 2.1819390869140625, "step": 41500 }, { "epoch": 0.16530487410671216, "grad_norm": 0.40237659215927124, "learning_rate": 1.834747010192796e-05, "loss": 2.1593255615234375, "step": 41550 }, { "epoch": 0.1655037969395722, "grad_norm": 0.6958070993423462, "learning_rate": 1.8345480875577874e-05, "loss": 2.169309539794922, "step": 41600 }, { "epoch": 0.16570271977243228, "grad_norm": 0.4832911193370819, "learning_rate": 1.8343491649227786e-05, "loss": 2.181063690185547, "step": 41650 }, { "epoch": 0.16590164260529233, "grad_norm": 0.5791292786598206, "learning_rate": 1.8341502422877695e-05, "loss": 2.1686924743652343, "step": 41700 }, { "epoch": 0.1661005654381524, "grad_norm": 0.43916961550712585, "learning_rate": 1.8339513196527607e-05, "loss": 2.180246124267578, "step": 41750 }, { "epoch": 0.16629948827101246, "grad_norm": 0.6350082159042358, "learning_rate": 1.833752397017752e-05, "loss": 2.1357051086425782, "step": 41800 }, { "epoch": 0.16649841110387253, "grad_norm": 0.6312410831451416, "learning_rate": 1.833553474382743e-05, "loss": 2.1771278381347656, "step": 41850 }, { "epoch": 0.16669733393673258, "grad_norm": 0.5649234652519226, "learning_rate": 1.8333545517477343e-05, "loss": 2.131345520019531, "step": 41900 }, { "epoch": 0.16689625676959266, "grad_norm": 0.5716596245765686, "learning_rate": 1.8331556291127256e-05, "loss": 2.1263584899902344, "step": 41950 }, { "epoch": 0.1670951796024527, "grad_norm": 0.5702884197235107, "learning_rate": 1.8329567064777168e-05, "loss": 2.1315142822265627, "step": 42000 }, { "epoch": 0.16729410243531279, "grad_norm": 0.5509811639785767, "learning_rate": 1.832757783842708e-05, "loss": 2.139906463623047, "step": 42050 }, { "epoch": 0.16749302526817283, "grad_norm": 0.5050264000892639, "learning_rate": 1.8325588612076992e-05, "loss": 2.170631561279297, "step": 42100 }, { "epoch": 0.1676919481010329, "grad_norm": 0.5224362015724182, "learning_rate": 1.8323599385726904e-05, "loss": 2.1422573852539064, "step": 42150 }, { "epoch": 0.16789087093389296, "grad_norm": 0.5426728129386902, "learning_rate": 1.8321610159376816e-05, "loss": 2.0940748596191407, "step": 42200 }, { "epoch": 0.16808979376675304, "grad_norm": 0.5210455060005188, "learning_rate": 1.831962093302673e-05, "loss": 2.079983367919922, "step": 42250 }, { "epoch": 0.1682887165996131, "grad_norm": 0.5269166827201843, "learning_rate": 1.831763170667664e-05, "loss": 2.1693234252929687, "step": 42300 }, { "epoch": 0.16848763943247316, "grad_norm": 0.5693178772926331, "learning_rate": 1.8315642480326553e-05, "loss": 2.1018536376953123, "step": 42350 }, { "epoch": 0.1686865622653332, "grad_norm": 0.48441752791404724, "learning_rate": 1.8313653253976465e-05, "loss": 2.1271881103515624, "step": 42400 }, { "epoch": 0.1688854850981933, "grad_norm": 0.4529274106025696, "learning_rate": 1.8311664027626377e-05, "loss": 2.1247819519042968, "step": 42450 }, { "epoch": 0.16908440793105334, "grad_norm": 0.4734072983264923, "learning_rate": 1.830967480127629e-05, "loss": 2.0998959350585937, "step": 42500 }, { "epoch": 0.16928333076391341, "grad_norm": 0.4637087881565094, "learning_rate": 1.83076855749262e-05, "loss": 2.1676026916503908, "step": 42550 }, { "epoch": 0.16948225359677346, "grad_norm": 0.5334165096282959, "learning_rate": 1.8305696348576113e-05, "loss": 2.1548570251464843, "step": 42600 }, { "epoch": 0.16968117642963354, "grad_norm": 0.5169240236282349, "learning_rate": 1.8303707122226026e-05, "loss": 2.157135314941406, "step": 42650 }, { "epoch": 0.1698800992624936, "grad_norm": 0.5185680389404297, "learning_rate": 1.8301717895875938e-05, "loss": 2.1490431213378907, "step": 42700 }, { "epoch": 0.17007902209535367, "grad_norm": 0.5165036916732788, "learning_rate": 1.829972866952585e-05, "loss": 2.135874786376953, "step": 42750 }, { "epoch": 0.17027794492821371, "grad_norm": 0.5811851620674133, "learning_rate": 1.8297739443175762e-05, "loss": 2.123148651123047, "step": 42800 }, { "epoch": 0.1704768677610738, "grad_norm": 0.4972330629825592, "learning_rate": 1.8295750216825674e-05, "loss": 2.133658599853516, "step": 42850 }, { "epoch": 0.17067579059393384, "grad_norm": 0.47997361421585083, "learning_rate": 1.8293760990475586e-05, "loss": 2.146107635498047, "step": 42900 }, { "epoch": 0.17087471342679392, "grad_norm": 0.549617350101471, "learning_rate": 1.82917717641255e-05, "loss": 2.1071017456054686, "step": 42950 }, { "epoch": 0.17107363625965397, "grad_norm": 0.5361450910568237, "learning_rate": 1.828978253777541e-05, "loss": 2.0584765625, "step": 43000 }, { "epoch": 0.17127255909251404, "grad_norm": 0.6502788662910461, "learning_rate": 1.8287793311425323e-05, "loss": 2.113952331542969, "step": 43050 }, { "epoch": 0.1714714819253741, "grad_norm": 0.5531705617904663, "learning_rate": 1.8285804085075235e-05, "loss": 2.135988311767578, "step": 43100 }, { "epoch": 0.17167040475823417, "grad_norm": 0.48471227288246155, "learning_rate": 1.8283814858725147e-05, "loss": 2.162136993408203, "step": 43150 }, { "epoch": 0.17186932759109422, "grad_norm": 0.5105908513069153, "learning_rate": 1.828182563237506e-05, "loss": 2.1342578125, "step": 43200 }, { "epoch": 0.1720682504239543, "grad_norm": 0.5353871583938599, "learning_rate": 1.827983640602497e-05, "loss": 2.129888610839844, "step": 43250 }, { "epoch": 0.17226717325681434, "grad_norm": 0.5058733820915222, "learning_rate": 1.827784717967488e-05, "loss": 2.1615864562988283, "step": 43300 }, { "epoch": 0.17246609608967442, "grad_norm": 0.5576196312904358, "learning_rate": 1.8275857953324792e-05, "loss": 2.136527252197266, "step": 43350 }, { "epoch": 0.17266501892253447, "grad_norm": 0.5185539126396179, "learning_rate": 1.8273868726974708e-05, "loss": 2.152915496826172, "step": 43400 }, { "epoch": 0.17286394175539455, "grad_norm": 0.4882115125656128, "learning_rate": 1.827187950062462e-05, "loss": 2.1535911560058594, "step": 43450 }, { "epoch": 0.1730628645882546, "grad_norm": 0.3631893992424011, "learning_rate": 1.8269890274274532e-05, "loss": 2.192937774658203, "step": 43500 }, { "epoch": 0.17326178742111467, "grad_norm": 0.5408120155334473, "learning_rate": 1.8267901047924444e-05, "loss": 2.1280343627929685, "step": 43550 }, { "epoch": 0.17346071025397472, "grad_norm": 0.37062278389930725, "learning_rate": 1.8265911821574356e-05, "loss": 2.1601911926269532, "step": 43600 }, { "epoch": 0.1736596330868348, "grad_norm": 0.48052793741226196, "learning_rate": 1.8263922595224265e-05, "loss": 2.1056723022460937, "step": 43650 }, { "epoch": 0.17385855591969485, "grad_norm": 0.5237756371498108, "learning_rate": 1.8261933368874177e-05, "loss": 2.1413424682617186, "step": 43700 }, { "epoch": 0.17405747875255492, "grad_norm": 0.6610327959060669, "learning_rate": 1.825994414252409e-05, "loss": 2.094590148925781, "step": 43750 }, { "epoch": 0.17425640158541497, "grad_norm": 0.5823939442634583, "learning_rate": 1.8257954916174e-05, "loss": 2.141654052734375, "step": 43800 }, { "epoch": 0.17445532441827505, "grad_norm": 0.5246759653091431, "learning_rate": 1.8255965689823917e-05, "loss": 2.2020492553710938, "step": 43850 }, { "epoch": 0.1746542472511351, "grad_norm": 0.5144526958465576, "learning_rate": 1.825397646347383e-05, "loss": 2.0662525939941405, "step": 43900 }, { "epoch": 0.17485317008399517, "grad_norm": 0.555125892162323, "learning_rate": 1.8251987237123738e-05, "loss": 2.155855865478516, "step": 43950 }, { "epoch": 0.17505209291685522, "grad_norm": 0.5240859389305115, "learning_rate": 1.824999801077365e-05, "loss": 2.1615524291992188, "step": 44000 }, { "epoch": 0.1752510157497153, "grad_norm": 0.5718843936920166, "learning_rate": 1.8248008784423562e-05, "loss": 2.088797760009766, "step": 44050 }, { "epoch": 0.17544993858257535, "grad_norm": 0.6328282952308655, "learning_rate": 1.8246019558073474e-05, "loss": 2.175008544921875, "step": 44100 }, { "epoch": 0.17564886141543543, "grad_norm": 0.5865340232849121, "learning_rate": 1.8244030331723387e-05, "loss": 2.1666024780273436, "step": 44150 }, { "epoch": 0.17584778424829547, "grad_norm": 0.5946438908576965, "learning_rate": 1.82420411053733e-05, "loss": 2.1361228942871096, "step": 44200 }, { "epoch": 0.17604670708115555, "grad_norm": 0.6068114638328552, "learning_rate": 1.8240051879023214e-05, "loss": 2.135209045410156, "step": 44250 }, { "epoch": 0.1762456299140156, "grad_norm": 0.658735990524292, "learning_rate": 1.8238102437200124e-05, "loss": 2.2005828857421874, "step": 44300 }, { "epoch": 0.17644455274687568, "grad_norm": 0.4727732539176941, "learning_rate": 1.8236113210850037e-05, "loss": 2.1027459716796875, "step": 44350 }, { "epoch": 0.17664347557973573, "grad_norm": 0.5632817149162292, "learning_rate": 1.823412398449995e-05, "loss": 2.125308837890625, "step": 44400 }, { "epoch": 0.1768423984125958, "grad_norm": 0.4354909658432007, "learning_rate": 1.823213475814986e-05, "loss": 2.1299986267089843, "step": 44450 }, { "epoch": 0.17704132124545585, "grad_norm": 0.5441125631332397, "learning_rate": 1.8230145531799773e-05, "loss": 2.1242144775390623, "step": 44500 }, { "epoch": 0.17724024407831593, "grad_norm": 0.49118489027023315, "learning_rate": 1.8228156305449685e-05, "loss": 2.0692225646972657, "step": 44550 }, { "epoch": 0.17743916691117598, "grad_norm": 0.5618523955345154, "learning_rate": 1.8226167079099597e-05, "loss": 2.1203350830078125, "step": 44600 }, { "epoch": 0.17763808974403605, "grad_norm": 0.5901533961296082, "learning_rate": 1.822417785274951e-05, "loss": 2.0966180419921874, "step": 44650 }, { "epoch": 0.1778370125768961, "grad_norm": 0.5654410719871521, "learning_rate": 1.822218862639942e-05, "loss": 2.0825579833984373, "step": 44700 }, { "epoch": 0.17803593540975618, "grad_norm": 0.5903883576393127, "learning_rate": 1.8220199400049334e-05, "loss": 2.0947178649902343, "step": 44750 }, { "epoch": 0.17823485824261623, "grad_norm": 0.5312032103538513, "learning_rate": 1.8218210173699246e-05, "loss": 2.1670530700683592, "step": 44800 }, { "epoch": 0.1784337810754763, "grad_norm": 0.5367507338523865, "learning_rate": 1.8216220947349158e-05, "loss": 2.088251647949219, "step": 44850 }, { "epoch": 0.17863270390833635, "grad_norm": 0.5666272044181824, "learning_rate": 1.821423172099907e-05, "loss": 2.1401718139648436, "step": 44900 }, { "epoch": 0.17883162674119643, "grad_norm": 0.509791910648346, "learning_rate": 1.8212242494648982e-05, "loss": 2.1497210693359374, "step": 44950 }, { "epoch": 0.17903054957405648, "grad_norm": 0.6391710638999939, "learning_rate": 1.8210253268298894e-05, "loss": 2.091194763183594, "step": 45000 }, { "epoch": 0.17922947240691656, "grad_norm": 0.5443548560142517, "learning_rate": 1.8208264041948807e-05, "loss": 2.1186122131347656, "step": 45050 }, { "epoch": 0.1794283952397766, "grad_norm": 0.48681846261024475, "learning_rate": 1.820627481559872e-05, "loss": 2.1649517822265625, "step": 45100 }, { "epoch": 0.17962731807263668, "grad_norm": 0.46023041009902954, "learning_rate": 1.820428558924863e-05, "loss": 2.1407095336914064, "step": 45150 }, { "epoch": 0.17982624090549673, "grad_norm": 0.5210662484169006, "learning_rate": 1.8202296362898543e-05, "loss": 2.1652874755859375, "step": 45200 }, { "epoch": 0.1800251637383568, "grad_norm": 0.565578281879425, "learning_rate": 1.8200307136548455e-05, "loss": 2.1003260803222656, "step": 45250 }, { "epoch": 0.18022408657121686, "grad_norm": 0.6309168338775635, "learning_rate": 1.8198317910198367e-05, "loss": 2.1870220947265624, "step": 45300 }, { "epoch": 0.18042300940407693, "grad_norm": 0.506210446357727, "learning_rate": 1.819632868384828e-05, "loss": 2.1262138366699217, "step": 45350 }, { "epoch": 0.18062193223693698, "grad_norm": 0.505994975566864, "learning_rate": 1.819433945749819e-05, "loss": 2.1062001037597655, "step": 45400 }, { "epoch": 0.18082085506979706, "grad_norm": 0.6078945398330688, "learning_rate": 1.8192350231148104e-05, "loss": 2.1248933410644533, "step": 45450 }, { "epoch": 0.1810197779026571, "grad_norm": 0.7542694211006165, "learning_rate": 1.8190361004798016e-05, "loss": 2.199825897216797, "step": 45500 }, { "epoch": 0.18121870073551719, "grad_norm": 0.5716150403022766, "learning_rate": 1.8188371778447928e-05, "loss": 2.0874012756347655, "step": 45550 }, { "epoch": 0.18141762356837723, "grad_norm": 0.45096081495285034, "learning_rate": 1.818638255209784e-05, "loss": 2.1136070251464845, "step": 45600 }, { "epoch": 0.1816165464012373, "grad_norm": 0.6214340329170227, "learning_rate": 1.8184393325747752e-05, "loss": 2.0947892761230467, "step": 45650 }, { "epoch": 0.18181546923409736, "grad_norm": 0.5629560351371765, "learning_rate": 1.8182404099397665e-05, "loss": 2.1450531005859377, "step": 45700 }, { "epoch": 0.18201439206695744, "grad_norm": 0.5796762108802795, "learning_rate": 1.8180414873047577e-05, "loss": 2.109466094970703, "step": 45750 }, { "epoch": 0.18221331489981749, "grad_norm": 0.5990759134292603, "learning_rate": 1.817842564669749e-05, "loss": 2.063998260498047, "step": 45800 }, { "epoch": 0.18241223773267756, "grad_norm": 0.6005164384841919, "learning_rate": 1.81764364203474e-05, "loss": 2.1139984130859375, "step": 45850 }, { "epoch": 0.1826111605655376, "grad_norm": 0.4709616005420685, "learning_rate": 1.817444719399731e-05, "loss": 2.1092459106445314, "step": 45900 }, { "epoch": 0.1828100833983977, "grad_norm": 0.5932245254516602, "learning_rate": 1.8172457967647222e-05, "loss": 2.0900770568847657, "step": 45950 }, { "epoch": 0.18300900623125774, "grad_norm": 0.5357955694198608, "learning_rate": 1.8170468741297137e-05, "loss": 2.1276821899414062, "step": 46000 }, { "epoch": 0.1832079290641178, "grad_norm": 0.6159970164299011, "learning_rate": 1.816847951494705e-05, "loss": 2.0861056518554686, "step": 46050 }, { "epoch": 0.18340685189697786, "grad_norm": 0.4905487895011902, "learning_rate": 1.816649028859696e-05, "loss": 2.151156311035156, "step": 46100 }, { "epoch": 0.18360577472983794, "grad_norm": 0.5944917798042297, "learning_rate": 1.8164501062246874e-05, "loss": 2.138446044921875, "step": 46150 }, { "epoch": 0.183804697562698, "grad_norm": 0.48137325048446655, "learning_rate": 1.8162511835896786e-05, "loss": 2.1340815734863283, "step": 46200 }, { "epoch": 0.18400362039555807, "grad_norm": 0.5680419206619263, "learning_rate": 1.8160522609546695e-05, "loss": 2.1610826110839843, "step": 46250 }, { "epoch": 0.18420254322841811, "grad_norm": 0.5551254749298096, "learning_rate": 1.8158533383196607e-05, "loss": 2.053030242919922, "step": 46300 }, { "epoch": 0.1844014660612782, "grad_norm": 0.49510595202445984, "learning_rate": 1.815654415684652e-05, "loss": 2.1285446166992186, "step": 46350 }, { "epoch": 0.18460038889413824, "grad_norm": 0.5888778567314148, "learning_rate": 1.8154554930496435e-05, "loss": 2.1571278381347656, "step": 46400 }, { "epoch": 0.18479931172699832, "grad_norm": 0.5793852806091309, "learning_rate": 1.8152565704146347e-05, "loss": 2.092467498779297, "step": 46450 }, { "epoch": 0.18499823455985837, "grad_norm": 0.5074703097343445, "learning_rate": 1.815057647779626e-05, "loss": 2.13672119140625, "step": 46500 }, { "epoch": 0.18519715739271844, "grad_norm": 0.5526040196418762, "learning_rate": 1.8148587251446168e-05, "loss": 2.1662677001953123, "step": 46550 }, { "epoch": 0.1853960802255785, "grad_norm": 0.5366107821464539, "learning_rate": 1.814659802509608e-05, "loss": 2.1034442138671876, "step": 46600 }, { "epoch": 0.18559500305843857, "grad_norm": 0.4998043477535248, "learning_rate": 1.8144608798745992e-05, "loss": 2.1350468444824218, "step": 46650 }, { "epoch": 0.18579392589129862, "grad_norm": 0.5560060739517212, "learning_rate": 1.8142619572395904e-05, "loss": 2.151147766113281, "step": 46700 }, { "epoch": 0.1859928487241587, "grad_norm": 0.5958081483840942, "learning_rate": 1.8140630346045816e-05, "loss": 2.1065985107421876, "step": 46750 }, { "epoch": 0.18619177155701874, "grad_norm": 0.5899885296821594, "learning_rate": 1.8138641119695732e-05, "loss": 2.08820556640625, "step": 46800 }, { "epoch": 0.18639069438987882, "grad_norm": 0.5921320915222168, "learning_rate": 1.8136651893345644e-05, "loss": 2.107676544189453, "step": 46850 }, { "epoch": 0.18658961722273887, "grad_norm": 0.5052732229232788, "learning_rate": 1.8134662666995553e-05, "loss": 2.100069580078125, "step": 46900 }, { "epoch": 0.18678854005559895, "grad_norm": 0.6577388644218445, "learning_rate": 1.8132673440645465e-05, "loss": 2.1215249633789064, "step": 46950 }, { "epoch": 0.186987462888459, "grad_norm": 0.3810938894748688, "learning_rate": 1.8130684214295377e-05, "loss": 2.116583251953125, "step": 47000 }, { "epoch": 0.18718638572131907, "grad_norm": 0.577003538608551, "learning_rate": 1.812869498794529e-05, "loss": 2.15046875, "step": 47050 }, { "epoch": 0.18738530855417912, "grad_norm": 0.593712329864502, "learning_rate": 1.81267057615952e-05, "loss": 2.1369598388671873, "step": 47100 }, { "epoch": 0.1875842313870392, "grad_norm": 0.5360112190246582, "learning_rate": 1.8124716535245113e-05, "loss": 2.1096051025390623, "step": 47150 }, { "epoch": 0.18778315421989925, "grad_norm": 0.5318012237548828, "learning_rate": 1.8122727308895026e-05, "loss": 2.115592498779297, "step": 47200 }, { "epoch": 0.18798207705275932, "grad_norm": 0.6232028603553772, "learning_rate": 1.8120738082544938e-05, "loss": 2.069017791748047, "step": 47250 }, { "epoch": 0.18818099988561937, "grad_norm": 0.42492058873176575, "learning_rate": 1.811874885619485e-05, "loss": 2.1733741760253906, "step": 47300 }, { "epoch": 0.18837992271847942, "grad_norm": 0.5037474036216736, "learning_rate": 1.8116759629844762e-05, "loss": 2.0966319274902343, "step": 47350 }, { "epoch": 0.1885788455513395, "grad_norm": 0.6040537357330322, "learning_rate": 1.8114810188021675e-05, "loss": 2.1440333557128906, "step": 47400 }, { "epoch": 0.18877776838419955, "grad_norm": 0.5590276122093201, "learning_rate": 1.8112820961671588e-05, "loss": 2.161787567138672, "step": 47450 }, { "epoch": 0.18897669121705962, "grad_norm": 0.4710489809513092, "learning_rate": 1.81108317353215e-05, "loss": 2.0667408752441405, "step": 47500 }, { "epoch": 0.18917561404991967, "grad_norm": 0.5742000937461853, "learning_rate": 1.8108842508971412e-05, "loss": 2.0951885986328125, "step": 47550 }, { "epoch": 0.18937453688277975, "grad_norm": 0.5507506728172302, "learning_rate": 1.8106853282621324e-05, "loss": 2.1089846801757814, "step": 47600 }, { "epoch": 0.1895734597156398, "grad_norm": 0.5853949785232544, "learning_rate": 1.8104864056271236e-05, "loss": 2.1244300842285155, "step": 47650 }, { "epoch": 0.18977238254849987, "grad_norm": 0.4255027770996094, "learning_rate": 1.810287482992115e-05, "loss": 2.0879441833496095, "step": 47700 }, { "epoch": 0.18997130538135992, "grad_norm": 0.5263792276382446, "learning_rate": 1.810088560357106e-05, "loss": 2.1161924743652345, "step": 47750 }, { "epoch": 0.19017022821422, "grad_norm": 0.542484700679779, "learning_rate": 1.8098896377220973e-05, "loss": 2.1082781982421874, "step": 47800 }, { "epoch": 0.19036915104708005, "grad_norm": 0.5251725912094116, "learning_rate": 1.8096907150870885e-05, "loss": 2.080404968261719, "step": 47850 }, { "epoch": 0.19056807387994013, "grad_norm": 0.47229382395744324, "learning_rate": 1.8094917924520797e-05, "loss": 2.1185418701171876, "step": 47900 }, { "epoch": 0.19076699671280017, "grad_norm": 0.5803781151771545, "learning_rate": 1.809292869817071e-05, "loss": 2.1176406860351564, "step": 47950 }, { "epoch": 0.19096591954566025, "grad_norm": 0.5491621494293213, "learning_rate": 1.809093947182062e-05, "loss": 2.1532012939453127, "step": 48000 }, { "epoch": 0.1911648423785203, "grad_norm": 0.5076926350593567, "learning_rate": 1.8088950245470533e-05, "loss": 2.113334503173828, "step": 48050 }, { "epoch": 0.19136376521138038, "grad_norm": 0.6164645552635193, "learning_rate": 1.8086961019120445e-05, "loss": 2.1290992736816405, "step": 48100 }, { "epoch": 0.19156268804424043, "grad_norm": 0.6230513453483582, "learning_rate": 1.8084971792770358e-05, "loss": 2.1156349182128906, "step": 48150 }, { "epoch": 0.1917616108771005, "grad_norm": 0.6171297430992126, "learning_rate": 1.808298256642027e-05, "loss": 2.124163818359375, "step": 48200 }, { "epoch": 0.19196053370996055, "grad_norm": 0.5445767045021057, "learning_rate": 1.8080993340070182e-05, "loss": 2.169175567626953, "step": 48250 }, { "epoch": 0.19215945654282063, "grad_norm": 0.4746412932872772, "learning_rate": 1.8079004113720094e-05, "loss": 2.119879150390625, "step": 48300 }, { "epoch": 0.19235837937568068, "grad_norm": 0.5314530730247498, "learning_rate": 1.8077014887370006e-05, "loss": 2.0574722290039062, "step": 48350 }, { "epoch": 0.19255730220854075, "grad_norm": 0.44336026906967163, "learning_rate": 1.807502566101992e-05, "loss": 2.133303680419922, "step": 48400 }, { "epoch": 0.1927562250414008, "grad_norm": 0.6059415936470032, "learning_rate": 1.807303643466983e-05, "loss": 2.0822216796875, "step": 48450 }, { "epoch": 0.19295514787426088, "grad_norm": 0.5473490953445435, "learning_rate": 1.807104720831974e-05, "loss": 2.1507122802734373, "step": 48500 }, { "epoch": 0.19315407070712093, "grad_norm": 0.6038652658462524, "learning_rate": 1.8069057981969655e-05, "loss": 2.125140380859375, "step": 48550 }, { "epoch": 0.193352993539981, "grad_norm": 0.6346015334129333, "learning_rate": 1.8067068755619567e-05, "loss": 2.076620788574219, "step": 48600 }, { "epoch": 0.19355191637284105, "grad_norm": 0.6015193462371826, "learning_rate": 1.806507952926948e-05, "loss": 2.112718963623047, "step": 48650 }, { "epoch": 0.19375083920570113, "grad_norm": 0.6458257436752319, "learning_rate": 1.806309030291939e-05, "loss": 2.0945472717285156, "step": 48700 }, { "epoch": 0.19394976203856118, "grad_norm": 0.5860766768455505, "learning_rate": 1.8061101076569303e-05, "loss": 2.0629450988769533, "step": 48750 }, { "epoch": 0.19414868487142126, "grad_norm": 0.4829556345939636, "learning_rate": 1.8059111850219216e-05, "loss": 2.075708923339844, "step": 48800 }, { "epoch": 0.1943476077042813, "grad_norm": 0.5787078142166138, "learning_rate": 1.8057122623869124e-05, "loss": 2.123612823486328, "step": 48850 }, { "epoch": 0.19454653053714138, "grad_norm": 0.4392569661140442, "learning_rate": 1.8055133397519036e-05, "loss": 2.127122039794922, "step": 48900 }, { "epoch": 0.19474545337000143, "grad_norm": 0.48986944556236267, "learning_rate": 1.8053144171168952e-05, "loss": 2.127798767089844, "step": 48950 }, { "epoch": 0.1949443762028615, "grad_norm": 0.6213397979736328, "learning_rate": 1.8051154944818864e-05, "loss": 2.1286424255371093, "step": 49000 }, { "epoch": 0.19514329903572156, "grad_norm": 0.527030348777771, "learning_rate": 1.8049165718468776e-05, "loss": 2.189007110595703, "step": 49050 }, { "epoch": 0.19534222186858163, "grad_norm": 0.4984337091445923, "learning_rate": 1.804717649211869e-05, "loss": 2.176228790283203, "step": 49100 }, { "epoch": 0.19554114470144168, "grad_norm": 0.5777124166488647, "learning_rate": 1.8045187265768597e-05, "loss": 2.110526275634766, "step": 49150 }, { "epoch": 0.19574006753430176, "grad_norm": 0.6358063817024231, "learning_rate": 1.804319803941851e-05, "loss": 2.1419476318359374, "step": 49200 }, { "epoch": 0.1959389903671618, "grad_norm": 0.540871262550354, "learning_rate": 1.804120881306842e-05, "loss": 2.146909027099609, "step": 49250 }, { "epoch": 0.19613791320002189, "grad_norm": 0.5553017854690552, "learning_rate": 1.8039219586718334e-05, "loss": 2.1293240356445313, "step": 49300 }, { "epoch": 0.19633683603288193, "grad_norm": 0.5584371089935303, "learning_rate": 1.803723036036825e-05, "loss": 2.138132781982422, "step": 49350 }, { "epoch": 0.196535758865742, "grad_norm": 0.4934276342391968, "learning_rate": 1.803524113401816e-05, "loss": 2.0778236389160156, "step": 49400 }, { "epoch": 0.19673468169860206, "grad_norm": 0.5102418661117554, "learning_rate": 1.803325190766807e-05, "loss": 2.1119284057617187, "step": 49450 }, { "epoch": 0.19693360453146214, "grad_norm": 0.5745898485183716, "learning_rate": 1.8031262681317982e-05, "loss": 2.094723815917969, "step": 49500 }, { "epoch": 0.19713252736432219, "grad_norm": 0.41309061646461487, "learning_rate": 1.8029273454967894e-05, "loss": 2.117995300292969, "step": 49550 }, { "epoch": 0.19733145019718226, "grad_norm": 0.6493209600448608, "learning_rate": 1.8027284228617806e-05, "loss": 2.108984069824219, "step": 49600 }, { "epoch": 0.1975303730300423, "grad_norm": 0.5947410464286804, "learning_rate": 1.802529500226772e-05, "loss": 2.107251739501953, "step": 49650 }, { "epoch": 0.1977292958629024, "grad_norm": 0.5988466143608093, "learning_rate": 1.802330577591763e-05, "loss": 2.0658416748046875, "step": 49700 }, { "epoch": 0.19792821869576244, "grad_norm": 0.4449918568134308, "learning_rate": 1.8021316549567543e-05, "loss": 2.1823863220214843, "step": 49750 }, { "epoch": 0.1981271415286225, "grad_norm": 0.5019288063049316, "learning_rate": 1.8019327323217455e-05, "loss": 2.1051866149902345, "step": 49800 }, { "epoch": 0.19832606436148256, "grad_norm": 0.4764851927757263, "learning_rate": 1.8017338096867367e-05, "loss": 2.1689080810546875, "step": 49850 }, { "epoch": 0.19852498719434264, "grad_norm": 0.5666704177856445, "learning_rate": 1.801534887051728e-05, "loss": 2.1071275329589843, "step": 49900 }, { "epoch": 0.1987239100272027, "grad_norm": 0.5777774453163147, "learning_rate": 1.801335964416719e-05, "loss": 2.1825616455078123, "step": 49950 }, { "epoch": 0.19892283286006276, "grad_norm": 0.520012378692627, "learning_rate": 1.8011370417817104e-05, "loss": 2.1209066772460936, "step": 50000 }, { "epoch": 0.1991217556929228, "grad_norm": 0.4869330823421478, "learning_rate": 1.8009381191467016e-05, "loss": 2.1445111083984374, "step": 50050 }, { "epoch": 0.1993206785257829, "grad_norm": 0.495148241519928, "learning_rate": 1.8007391965116928e-05, "loss": 2.1014970397949218, "step": 50100 }, { "epoch": 0.19951960135864294, "grad_norm": 0.42597708106040955, "learning_rate": 1.800540273876684e-05, "loss": 2.117384948730469, "step": 50150 }, { "epoch": 0.19971852419150302, "grad_norm": 0.5519025325775146, "learning_rate": 1.8003413512416752e-05, "loss": 2.171905975341797, "step": 50200 }, { "epoch": 0.19991744702436307, "grad_norm": 0.6103881597518921, "learning_rate": 1.8001424286066664e-05, "loss": 2.1125613403320314, "step": 50250 }, { "epoch": 0.20011636985722314, "grad_norm": 0.52480149269104, "learning_rate": 1.7999435059716577e-05, "loss": 2.1094009399414064, "step": 50300 }, { "epoch": 0.2003152926900832, "grad_norm": 0.48829299211502075, "learning_rate": 1.799744583336649e-05, "loss": 2.1138009643554687, "step": 50350 }, { "epoch": 0.20051421552294327, "grad_norm": 0.5021555423736572, "learning_rate": 1.79954566070164e-05, "loss": 2.130376892089844, "step": 50400 }, { "epoch": 0.20071313835580332, "grad_norm": 0.5521305799484253, "learning_rate": 1.7993467380666313e-05, "loss": 2.115046081542969, "step": 50450 }, { "epoch": 0.2009120611886634, "grad_norm": 0.5477733612060547, "learning_rate": 1.7991478154316225e-05, "loss": 2.038019104003906, "step": 50500 }, { "epoch": 0.20111098402152344, "grad_norm": 0.5957963466644287, "learning_rate": 1.7989488927966137e-05, "loss": 2.1235198974609375, "step": 50550 }, { "epoch": 0.20130990685438352, "grad_norm": 0.5176833271980286, "learning_rate": 1.798749970161605e-05, "loss": 2.1398191833496094, "step": 50600 }, { "epoch": 0.20150882968724357, "grad_norm": 0.5853161811828613, "learning_rate": 1.798551047526596e-05, "loss": 2.1682688903808596, "step": 50650 }, { "epoch": 0.20170775252010364, "grad_norm": 0.531839907169342, "learning_rate": 1.7983521248915874e-05, "loss": 2.1369612121582033, "step": 50700 }, { "epoch": 0.2019066753529637, "grad_norm": 0.5332740545272827, "learning_rate": 1.7981532022565782e-05, "loss": 2.1418275451660156, "step": 50750 }, { "epoch": 0.20210559818582377, "grad_norm": 0.517183780670166, "learning_rate": 1.7979542796215698e-05, "loss": 2.140799713134766, "step": 50800 }, { "epoch": 0.20230452101868382, "grad_norm": 0.5920785665512085, "learning_rate": 1.797755356986561e-05, "loss": 2.0862020874023437, "step": 50850 }, { "epoch": 0.2025034438515439, "grad_norm": 0.5474011898040771, "learning_rate": 1.7975564343515522e-05, "loss": 2.097864227294922, "step": 50900 }, { "epoch": 0.20270236668440395, "grad_norm": 0.5494111776351929, "learning_rate": 1.7973575117165434e-05, "loss": 2.1291688537597655, "step": 50950 }, { "epoch": 0.20290128951726402, "grad_norm": 0.5518476366996765, "learning_rate": 1.7971585890815347e-05, "loss": 2.1345248413085938, "step": 51000 }, { "epoch": 0.20310021235012407, "grad_norm": 0.5467748045921326, "learning_rate": 1.796959666446526e-05, "loss": 2.060655975341797, "step": 51050 }, { "epoch": 0.20329913518298415, "grad_norm": 0.4913685619831085, "learning_rate": 1.7967607438115167e-05, "loss": 2.1074200439453126, "step": 51100 }, { "epoch": 0.2034980580158442, "grad_norm": 0.5392553210258484, "learning_rate": 1.796561821176508e-05, "loss": 2.146559753417969, "step": 51150 }, { "epoch": 0.20369698084870427, "grad_norm": 0.5380077958106995, "learning_rate": 1.7963628985414995e-05, "loss": 2.0946806335449217, "step": 51200 }, { "epoch": 0.20389590368156432, "grad_norm": 0.47544556856155396, "learning_rate": 1.7961639759064907e-05, "loss": 2.1082591247558593, "step": 51250 }, { "epoch": 0.2040948265144244, "grad_norm": 0.6177787184715271, "learning_rate": 1.795965053271482e-05, "loss": 2.1757432556152345, "step": 51300 }, { "epoch": 0.20429374934728445, "grad_norm": 0.5078738927841187, "learning_rate": 1.795766130636473e-05, "loss": 2.095434265136719, "step": 51350 }, { "epoch": 0.20449267218014452, "grad_norm": 0.6321300268173218, "learning_rate": 1.795567208001464e-05, "loss": 2.1540676879882814, "step": 51400 }, { "epoch": 0.20469159501300457, "grad_norm": 0.49753203988075256, "learning_rate": 1.7953682853664553e-05, "loss": 2.13357177734375, "step": 51450 }, { "epoch": 0.20489051784586465, "grad_norm": 0.5321568250656128, "learning_rate": 1.7951693627314465e-05, "loss": 2.097370910644531, "step": 51500 }, { "epoch": 0.2050894406787247, "grad_norm": 0.5400456190109253, "learning_rate": 1.7949704400964377e-05, "loss": 2.065793304443359, "step": 51550 }, { "epoch": 0.20528836351158478, "grad_norm": 0.5396239757537842, "learning_rate": 1.7947715174614292e-05, "loss": 2.1179344177246096, "step": 51600 }, { "epoch": 0.20548728634444482, "grad_norm": 0.6196457147598267, "learning_rate": 1.7945725948264204e-05, "loss": 2.0958912658691404, "step": 51650 }, { "epoch": 0.2056862091773049, "grad_norm": 0.5350463390350342, "learning_rate": 1.7943736721914117e-05, "loss": 2.1416941833496095, "step": 51700 }, { "epoch": 0.20588513201016495, "grad_norm": 0.5392886996269226, "learning_rate": 1.7941787280091027e-05, "loss": 2.055662841796875, "step": 51750 }, { "epoch": 0.20608405484302503, "grad_norm": 0.4535066485404968, "learning_rate": 1.793983783826794e-05, "loss": 2.1505754089355467, "step": 51800 }, { "epoch": 0.20628297767588508, "grad_norm": 0.5707430243492126, "learning_rate": 1.7937848611917856e-05, "loss": 2.1005734252929686, "step": 51850 }, { "epoch": 0.20648190050874515, "grad_norm": 0.572016716003418, "learning_rate": 1.7935859385567768e-05, "loss": 2.0374969482421874, "step": 51900 }, { "epoch": 0.2066808233416052, "grad_norm": 0.5146169662475586, "learning_rate": 1.793387015921768e-05, "loss": 2.1657666015625, "step": 51950 }, { "epoch": 0.20687974617446528, "grad_norm": 0.5138805508613586, "learning_rate": 1.7931880932867592e-05, "loss": 2.097891387939453, "step": 52000 }, { "epoch": 0.20707866900732533, "grad_norm": 0.5451368093490601, "learning_rate": 1.7929891706517504e-05, "loss": 2.0963839721679687, "step": 52050 }, { "epoch": 0.2072775918401854, "grad_norm": 0.541108250617981, "learning_rate": 1.7927902480167413e-05, "loss": 2.118468475341797, "step": 52100 }, { "epoch": 0.20747651467304545, "grad_norm": 0.5236655473709106, "learning_rate": 1.7925913253817325e-05, "loss": 2.117195739746094, "step": 52150 }, { "epoch": 0.20767543750590553, "grad_norm": 0.5376532077789307, "learning_rate": 1.7923924027467237e-05, "loss": 2.063456726074219, "step": 52200 }, { "epoch": 0.20787436033876558, "grad_norm": 0.44165563583374023, "learning_rate": 1.792193480111715e-05, "loss": 2.1570751953125, "step": 52250 }, { "epoch": 0.20807328317162566, "grad_norm": 0.5368886590003967, "learning_rate": 1.7919945574767065e-05, "loss": 2.1198799133300783, "step": 52300 }, { "epoch": 0.2082722060044857, "grad_norm": 0.5603135824203491, "learning_rate": 1.7917956348416977e-05, "loss": 2.0819224548339843, "step": 52350 }, { "epoch": 0.20847112883734578, "grad_norm": 0.5232288837432861, "learning_rate": 1.7915967122066886e-05, "loss": 2.086439971923828, "step": 52400 }, { "epoch": 0.20867005167020583, "grad_norm": 0.6800499558448792, "learning_rate": 1.7913977895716798e-05, "loss": 2.1521617126464845, "step": 52450 }, { "epoch": 0.2088689745030659, "grad_norm": 0.6270444393157959, "learning_rate": 1.791198866936671e-05, "loss": 2.081105194091797, "step": 52500 }, { "epoch": 0.20906789733592596, "grad_norm": 0.6324753761291504, "learning_rate": 1.7909999443016622e-05, "loss": 2.1575938415527345, "step": 52550 }, { "epoch": 0.20926682016878603, "grad_norm": 0.5793786644935608, "learning_rate": 1.7908010216666535e-05, "loss": 2.1447654724121095, "step": 52600 }, { "epoch": 0.20946574300164608, "grad_norm": 0.6418116092681885, "learning_rate": 1.7906020990316447e-05, "loss": 2.094773712158203, "step": 52650 }, { "epoch": 0.20966466583450616, "grad_norm": 0.5323590040206909, "learning_rate": 1.7904031763966362e-05, "loss": 2.074276123046875, "step": 52700 }, { "epoch": 0.2098635886673662, "grad_norm": 0.5938608646392822, "learning_rate": 1.790204253761627e-05, "loss": 2.069058532714844, "step": 52750 }, { "epoch": 0.21006251150022628, "grad_norm": 0.5321609973907471, "learning_rate": 1.7900053311266183e-05, "loss": 2.1352975463867185, "step": 52800 }, { "epoch": 0.21026143433308633, "grad_norm": 0.4844023585319519, "learning_rate": 1.7898064084916095e-05, "loss": 2.0845210266113283, "step": 52850 }, { "epoch": 0.2104603571659464, "grad_norm": 0.5753064155578613, "learning_rate": 1.7896074858566007e-05, "loss": 2.095992126464844, "step": 52900 }, { "epoch": 0.21065927999880646, "grad_norm": 0.45385828614234924, "learning_rate": 1.789408563221592e-05, "loss": 2.0909527587890624, "step": 52950 }, { "epoch": 0.21085820283166654, "grad_norm": 0.6035183668136597, "learning_rate": 1.7892096405865832e-05, "loss": 2.1028466796875, "step": 53000 }, { "epoch": 0.21105712566452658, "grad_norm": 0.5806202292442322, "learning_rate": 1.7890107179515744e-05, "loss": 2.0890953063964846, "step": 53050 }, { "epoch": 0.21125604849738666, "grad_norm": 0.5960564613342285, "learning_rate": 1.7888117953165656e-05, "loss": 2.1079777526855468, "step": 53100 }, { "epoch": 0.2114549713302467, "grad_norm": 0.5162715911865234, "learning_rate": 1.7886128726815568e-05, "loss": 2.109918212890625, "step": 53150 }, { "epoch": 0.2116538941631068, "grad_norm": 0.4453166127204895, "learning_rate": 1.788413950046548e-05, "loss": 2.1260496520996095, "step": 53200 }, { "epoch": 0.21185281699596684, "grad_norm": 0.5773827433586121, "learning_rate": 1.7882150274115392e-05, "loss": 2.16668701171875, "step": 53250 }, { "epoch": 0.2120517398288269, "grad_norm": 0.4779597520828247, "learning_rate": 1.7880161047765305e-05, "loss": 2.1523265075683593, "step": 53300 }, { "epoch": 0.21225066266168696, "grad_norm": 0.510266125202179, "learning_rate": 1.7878171821415217e-05, "loss": 2.08338134765625, "step": 53350 }, { "epoch": 0.21244958549454704, "grad_norm": 0.4819304645061493, "learning_rate": 1.787618259506513e-05, "loss": 2.1621498107910155, "step": 53400 }, { "epoch": 0.2126485083274071, "grad_norm": 0.5463733077049255, "learning_rate": 1.787419336871504e-05, "loss": 2.0843191528320313, "step": 53450 }, { "epoch": 0.21284743116026716, "grad_norm": 0.6128923296928406, "learning_rate": 1.7872204142364953e-05, "loss": 2.080638122558594, "step": 53500 }, { "epoch": 0.2130463539931272, "grad_norm": 0.6175855398178101, "learning_rate": 1.7870214916014865e-05, "loss": 2.146451263427734, "step": 53550 }, { "epoch": 0.2132452768259873, "grad_norm": 0.6106910109519958, "learning_rate": 1.7868225689664777e-05, "loss": 2.125449371337891, "step": 53600 }, { "epoch": 0.21344419965884734, "grad_norm": 0.5660700798034668, "learning_rate": 1.786623646331469e-05, "loss": 2.1306915283203125, "step": 53650 }, { "epoch": 0.21364312249170742, "grad_norm": 0.5538065433502197, "learning_rate": 1.7864247236964602e-05, "loss": 2.1188090515136717, "step": 53700 }, { "epoch": 0.21384204532456746, "grad_norm": 0.6653887629508972, "learning_rate": 1.7862258010614514e-05, "loss": 2.11252685546875, "step": 53750 }, { "epoch": 0.21404096815742754, "grad_norm": 0.5701465606689453, "learning_rate": 1.7860268784264426e-05, "loss": 2.129075164794922, "step": 53800 }, { "epoch": 0.2142398909902876, "grad_norm": 0.49792560935020447, "learning_rate": 1.7858279557914338e-05, "loss": 2.1417221069335937, "step": 53850 }, { "epoch": 0.21443881382314767, "grad_norm": 0.574318528175354, "learning_rate": 1.785629033156425e-05, "loss": 2.0962696838378907, "step": 53900 }, { "epoch": 0.21463773665600772, "grad_norm": 0.6359025239944458, "learning_rate": 1.7854301105214163e-05, "loss": 2.1192987060546873, "step": 53950 }, { "epoch": 0.2148366594888678, "grad_norm": 0.5516477823257446, "learning_rate": 1.785231187886407e-05, "loss": 2.1381500244140623, "step": 54000 }, { "epoch": 0.21503558232172784, "grad_norm": 0.5511181950569153, "learning_rate": 1.7850322652513983e-05, "loss": 2.0565919494628906, "step": 54050 }, { "epoch": 0.21523450515458792, "grad_norm": 0.6368272304534912, "learning_rate": 1.78483334261639e-05, "loss": 2.115557098388672, "step": 54100 }, { "epoch": 0.21543342798744797, "grad_norm": 0.553490936756134, "learning_rate": 1.784634419981381e-05, "loss": 2.077825012207031, "step": 54150 }, { "epoch": 0.21563235082030804, "grad_norm": 0.49449899792671204, "learning_rate": 1.7844354973463723e-05, "loss": 2.103396301269531, "step": 54200 }, { "epoch": 0.2158312736531681, "grad_norm": 0.5109418630599976, "learning_rate": 1.7842365747113635e-05, "loss": 2.1335044860839845, "step": 54250 }, { "epoch": 0.21603019648602817, "grad_norm": 0.5310605764389038, "learning_rate": 1.7840376520763548e-05, "loss": 2.118711242675781, "step": 54300 }, { "epoch": 0.21622911931888822, "grad_norm": 0.5749795436859131, "learning_rate": 1.7838387294413456e-05, "loss": 2.152138977050781, "step": 54350 }, { "epoch": 0.2164280421517483, "grad_norm": 0.46946343779563904, "learning_rate": 1.783639806806337e-05, "loss": 2.0641635131835936, "step": 54400 }, { "epoch": 0.21662696498460834, "grad_norm": 0.5253238081932068, "learning_rate": 1.783440884171328e-05, "loss": 2.163521728515625, "step": 54450 }, { "epoch": 0.21682588781746842, "grad_norm": 0.5776919722557068, "learning_rate": 1.7832419615363196e-05, "loss": 2.165123291015625, "step": 54500 }, { "epoch": 0.21702481065032847, "grad_norm": 0.5192419290542603, "learning_rate": 1.7830430389013108e-05, "loss": 2.117196807861328, "step": 54550 }, { "epoch": 0.21722373348318855, "grad_norm": 0.6073774695396423, "learning_rate": 1.782844116266302e-05, "loss": 2.136197052001953, "step": 54600 }, { "epoch": 0.2174226563160486, "grad_norm": 0.5345956087112427, "learning_rate": 1.782645193631293e-05, "loss": 2.0630001831054687, "step": 54650 }, { "epoch": 0.21762157914890867, "grad_norm": 0.5464485287666321, "learning_rate": 1.782446270996284e-05, "loss": 2.1135784912109377, "step": 54700 }, { "epoch": 0.21782050198176872, "grad_norm": 0.5411350131034851, "learning_rate": 1.7822473483612753e-05, "loss": 2.061462097167969, "step": 54750 }, { "epoch": 0.2180194248146288, "grad_norm": 0.9018494486808777, "learning_rate": 1.7820484257262666e-05, "loss": 2.0863186645507814, "step": 54800 }, { "epoch": 0.21821834764748885, "grad_norm": 0.546563446521759, "learning_rate": 1.7818495030912578e-05, "loss": 2.1019192504882813, "step": 54850 }, { "epoch": 0.21841727048034892, "grad_norm": 0.6629248857498169, "learning_rate": 1.7816505804562493e-05, "loss": 2.0888929748535157, "step": 54900 }, { "epoch": 0.21861619331320897, "grad_norm": 0.5397602915763855, "learning_rate": 1.7814516578212405e-05, "loss": 2.0924151611328123, "step": 54950 }, { "epoch": 0.21881511614606905, "grad_norm": 0.6082922220230103, "learning_rate": 1.7812527351862314e-05, "loss": 2.1220172119140623, "step": 55000 }, { "epoch": 0.2190140389789291, "grad_norm": 0.5449220538139343, "learning_rate": 1.7810538125512226e-05, "loss": 2.123433074951172, "step": 55050 }, { "epoch": 0.21921296181178918, "grad_norm": 0.590976893901825, "learning_rate": 1.780854889916214e-05, "loss": 2.079989013671875, "step": 55100 }, { "epoch": 0.21941188464464922, "grad_norm": 0.5491288900375366, "learning_rate": 1.780655967281205e-05, "loss": 2.1150975036621094, "step": 55150 }, { "epoch": 0.21961080747750927, "grad_norm": 0.5454486012458801, "learning_rate": 1.7804570446461963e-05, "loss": 2.117442169189453, "step": 55200 }, { "epoch": 0.21980973031036935, "grad_norm": 0.5910441875457764, "learning_rate": 1.7802581220111875e-05, "loss": 2.135828857421875, "step": 55250 }, { "epoch": 0.2200086531432294, "grad_norm": 0.6145221590995789, "learning_rate": 1.7800591993761787e-05, "loss": 2.104983367919922, "step": 55300 }, { "epoch": 0.22020757597608948, "grad_norm": 0.646501898765564, "learning_rate": 1.77986027674117e-05, "loss": 2.1102951049804686, "step": 55350 }, { "epoch": 0.22040649880894952, "grad_norm": 0.5925225019454956, "learning_rate": 1.779661354106161e-05, "loss": 2.0993312072753905, "step": 55400 }, { "epoch": 0.2206054216418096, "grad_norm": 0.6061117649078369, "learning_rate": 1.7794624314711524e-05, "loss": 2.1124333190917968, "step": 55450 }, { "epoch": 0.22080434447466965, "grad_norm": 0.5028465986251831, "learning_rate": 1.7792635088361436e-05, "loss": 2.0848316955566406, "step": 55500 }, { "epoch": 0.22100326730752973, "grad_norm": 0.5013981461524963, "learning_rate": 1.7790645862011348e-05, "loss": 2.093883361816406, "step": 55550 }, { "epoch": 0.22120219014038978, "grad_norm": 0.5122907757759094, "learning_rate": 1.778865663566126e-05, "loss": 2.054202117919922, "step": 55600 }, { "epoch": 0.22140111297324985, "grad_norm": 0.3723560869693756, "learning_rate": 1.7786667409311172e-05, "loss": 2.1097215270996093, "step": 55650 }, { "epoch": 0.2216000358061099, "grad_norm": 0.5172881484031677, "learning_rate": 1.7784678182961084e-05, "loss": 2.1101397705078124, "step": 55700 }, { "epoch": 0.22179895863896998, "grad_norm": 0.5526776909828186, "learning_rate": 1.7782688956610996e-05, "loss": 2.1165171813964845, "step": 55750 }, { "epoch": 0.22199788147183003, "grad_norm": 0.6553623080253601, "learning_rate": 1.778069973026091e-05, "loss": 2.1071710205078125, "step": 55800 }, { "epoch": 0.2221968043046901, "grad_norm": 0.5362281799316406, "learning_rate": 1.777871050391082e-05, "loss": 2.0410360717773437, "step": 55850 }, { "epoch": 0.22239572713755015, "grad_norm": 0.5825661420822144, "learning_rate": 1.7776761062087734e-05, "loss": 2.1258535766601563, "step": 55900 }, { "epoch": 0.22259464997041023, "grad_norm": 0.4606325626373291, "learning_rate": 1.7774771835737646e-05, "loss": 2.1126214599609376, "step": 55950 }, { "epoch": 0.22279357280327028, "grad_norm": 0.4885670840740204, "learning_rate": 1.777278260938756e-05, "loss": 2.066197509765625, "step": 56000 }, { "epoch": 0.22299249563613036, "grad_norm": 0.5266088843345642, "learning_rate": 1.777079338303747e-05, "loss": 2.0883860778808594, "step": 56050 }, { "epoch": 0.2231914184689904, "grad_norm": 0.5622708201408386, "learning_rate": 1.7768804156687383e-05, "loss": 2.145188903808594, "step": 56100 }, { "epoch": 0.22339034130185048, "grad_norm": 0.5485810041427612, "learning_rate": 1.7766814930337295e-05, "loss": 2.0517657470703123, "step": 56150 }, { "epoch": 0.22358926413471053, "grad_norm": 0.602179229259491, "learning_rate": 1.7764825703987207e-05, "loss": 2.107152099609375, "step": 56200 }, { "epoch": 0.2237881869675706, "grad_norm": 0.6416996717453003, "learning_rate": 1.776283647763712e-05, "loss": 2.154295654296875, "step": 56250 }, { "epoch": 0.22398710980043066, "grad_norm": 0.5257443189620972, "learning_rate": 1.776084725128703e-05, "loss": 2.140755157470703, "step": 56300 }, { "epoch": 0.22418603263329073, "grad_norm": 0.47181421518325806, "learning_rate": 1.7758858024936943e-05, "loss": 2.116202392578125, "step": 56350 }, { "epoch": 0.22438495546615078, "grad_norm": 0.5605736970901489, "learning_rate": 1.7756868798586856e-05, "loss": 2.066280517578125, "step": 56400 }, { "epoch": 0.22458387829901086, "grad_norm": 0.6495715975761414, "learning_rate": 1.7754879572236768e-05, "loss": 2.144746856689453, "step": 56450 }, { "epoch": 0.2247828011318709, "grad_norm": 0.582656979560852, "learning_rate": 1.775289034588668e-05, "loss": 2.0983953857421875, "step": 56500 }, { "epoch": 0.22498172396473098, "grad_norm": 0.4133286774158478, "learning_rate": 1.7750901119536592e-05, "loss": 2.083669891357422, "step": 56550 }, { "epoch": 0.22518064679759103, "grad_norm": 0.5505864024162292, "learning_rate": 1.77489118931865e-05, "loss": 2.0684298706054687, "step": 56600 }, { "epoch": 0.2253795696304511, "grad_norm": 0.6337907314300537, "learning_rate": 1.7746922666836416e-05, "loss": 2.1100927734375, "step": 56650 }, { "epoch": 0.22557849246331116, "grad_norm": 0.5513405203819275, "learning_rate": 1.774493344048633e-05, "loss": 2.078378143310547, "step": 56700 }, { "epoch": 0.22577741529617124, "grad_norm": 0.8091294169425964, "learning_rate": 1.774294421413624e-05, "loss": 2.08693115234375, "step": 56750 }, { "epoch": 0.22597633812903128, "grad_norm": 0.5563457608222961, "learning_rate": 1.7740954987786153e-05, "loss": 2.178367004394531, "step": 56800 }, { "epoch": 0.22617526096189136, "grad_norm": 0.5353357195854187, "learning_rate": 1.7738965761436065e-05, "loss": 2.0984957885742186, "step": 56850 }, { "epoch": 0.2263741837947514, "grad_norm": 0.49463361501693726, "learning_rate": 1.7736976535085977e-05, "loss": 2.0294837951660156, "step": 56900 }, { "epoch": 0.2265731066276115, "grad_norm": 0.5459253787994385, "learning_rate": 1.7734987308735886e-05, "loss": 2.086062469482422, "step": 56950 }, { "epoch": 0.22677202946047154, "grad_norm": 0.5588546991348267, "learning_rate": 1.7732998082385798e-05, "loss": 2.1272007751464845, "step": 57000 }, { "epoch": 0.2269709522933316, "grad_norm": 0.5099554061889648, "learning_rate": 1.7731008856035714e-05, "loss": 2.1381842041015626, "step": 57050 }, { "epoch": 0.22716987512619166, "grad_norm": 0.5540441274642944, "learning_rate": 1.7729019629685626e-05, "loss": 2.1163394165039064, "step": 57100 }, { "epoch": 0.22736879795905174, "grad_norm": 0.5118212699890137, "learning_rate": 1.7727030403335538e-05, "loss": 2.0944720458984376, "step": 57150 }, { "epoch": 0.2275677207919118, "grad_norm": 0.5195177793502808, "learning_rate": 1.772504117698545e-05, "loss": 2.1166258239746094, "step": 57200 }, { "epoch": 0.22776664362477186, "grad_norm": 0.47923848032951355, "learning_rate": 1.772305195063536e-05, "loss": 2.109049530029297, "step": 57250 }, { "epoch": 0.2279655664576319, "grad_norm": 0.5723922252655029, "learning_rate": 1.772106272428527e-05, "loss": 2.0771588134765624, "step": 57300 }, { "epoch": 0.228164489290492, "grad_norm": 0.5889751315116882, "learning_rate": 1.7719073497935183e-05, "loss": 2.117481231689453, "step": 57350 }, { "epoch": 0.22836341212335204, "grad_norm": 0.5094491243362427, "learning_rate": 1.7717084271585095e-05, "loss": 2.0466798400878905, "step": 57400 }, { "epoch": 0.22856233495621212, "grad_norm": 0.5828971266746521, "learning_rate": 1.7715095045235007e-05, "loss": 2.0672154235839844, "step": 57450 }, { "epoch": 0.22876125778907216, "grad_norm": 0.6393038630485535, "learning_rate": 1.7713105818884923e-05, "loss": 2.0880743408203126, "step": 57500 }, { "epoch": 0.22896018062193224, "grad_norm": 0.49855342507362366, "learning_rate": 1.7711116592534835e-05, "loss": 2.0481451416015624, "step": 57550 }, { "epoch": 0.2291591034547923, "grad_norm": 0.5644420385360718, "learning_rate": 1.7709127366184744e-05, "loss": 2.1231144714355468, "step": 57600 }, { "epoch": 0.22935802628765237, "grad_norm": 0.5632004737854004, "learning_rate": 1.7707138139834656e-05, "loss": 2.112692413330078, "step": 57650 }, { "epoch": 0.22955694912051242, "grad_norm": 0.5447084307670593, "learning_rate": 1.7705148913484568e-05, "loss": 2.0711065673828126, "step": 57700 }, { "epoch": 0.2297558719533725, "grad_norm": 0.5421858429908752, "learning_rate": 1.770315968713448e-05, "loss": 2.112601318359375, "step": 57750 }, { "epoch": 0.22995479478623254, "grad_norm": 0.531826376914978, "learning_rate": 1.7701170460784392e-05, "loss": 2.043171081542969, "step": 57800 }, { "epoch": 0.23015371761909262, "grad_norm": 0.504396378993988, "learning_rate": 1.7699181234434304e-05, "loss": 2.0932551574707032, "step": 57850 }, { "epoch": 0.23035264045195267, "grad_norm": 0.6037415266036987, "learning_rate": 1.7697192008084217e-05, "loss": 2.1210896301269533, "step": 57900 }, { "epoch": 0.23055156328481274, "grad_norm": 0.6016315817832947, "learning_rate": 1.769520278173413e-05, "loss": 2.1146641540527344, "step": 57950 }, { "epoch": 0.2307504861176728, "grad_norm": 0.58851158618927, "learning_rate": 1.769321355538404e-05, "loss": 2.075622253417969, "step": 58000 }, { "epoch": 0.23094940895053287, "grad_norm": 0.5035909414291382, "learning_rate": 1.7691224329033953e-05, "loss": 2.137571105957031, "step": 58050 }, { "epoch": 0.23114833178339292, "grad_norm": 0.5603659152984619, "learning_rate": 1.7689235102683865e-05, "loss": 2.109413757324219, "step": 58100 }, { "epoch": 0.231347254616253, "grad_norm": 0.6137785911560059, "learning_rate": 1.7687245876333777e-05, "loss": 2.1055194091796876, "step": 58150 }, { "epoch": 0.23154617744911304, "grad_norm": 0.5829649567604065, "learning_rate": 1.768525664998369e-05, "loss": 2.046627655029297, "step": 58200 }, { "epoch": 0.23174510028197312, "grad_norm": 0.5780123472213745, "learning_rate": 1.76832674236336e-05, "loss": 2.0741448974609376, "step": 58250 }, { "epoch": 0.23194402311483317, "grad_norm": 0.5902813673019409, "learning_rate": 1.7681278197283514e-05, "loss": 2.06986083984375, "step": 58300 }, { "epoch": 0.23214294594769325, "grad_norm": 0.4685487747192383, "learning_rate": 1.7679288970933426e-05, "loss": 2.1098483276367186, "step": 58350 }, { "epoch": 0.2323418687805533, "grad_norm": 0.5289891362190247, "learning_rate": 1.7677299744583338e-05, "loss": 2.075914001464844, "step": 58400 }, { "epoch": 0.23254079161341337, "grad_norm": 0.5776365399360657, "learning_rate": 1.767531051823325e-05, "loss": 2.122936248779297, "step": 58450 }, { "epoch": 0.23273971444627342, "grad_norm": 0.5460101366043091, "learning_rate": 1.7673321291883162e-05, "loss": 2.1250260925292968, "step": 58500 }, { "epoch": 0.2329386372791335, "grad_norm": 0.5920984148979187, "learning_rate": 1.7671332065533075e-05, "loss": 2.111446533203125, "step": 58550 }, { "epoch": 0.23313756011199355, "grad_norm": 0.5889592170715332, "learning_rate": 1.7669342839182987e-05, "loss": 2.1062583923339844, "step": 58600 }, { "epoch": 0.23333648294485362, "grad_norm": 0.5259612798690796, "learning_rate": 1.76673536128329e-05, "loss": 2.1119305419921877, "step": 58650 }, { "epoch": 0.23353540577771367, "grad_norm": 0.5957757830619812, "learning_rate": 1.766536438648281e-05, "loss": 2.1425633239746094, "step": 58700 }, { "epoch": 0.23373432861057375, "grad_norm": 0.6141670346260071, "learning_rate": 1.7663375160132723e-05, "loss": 2.0544593811035154, "step": 58750 }, { "epoch": 0.2339332514434338, "grad_norm": 0.5471116304397583, "learning_rate": 1.7661385933782635e-05, "loss": 2.0660636901855467, "step": 58800 }, { "epoch": 0.23413217427629388, "grad_norm": 0.6129280924797058, "learning_rate": 1.7659396707432547e-05, "loss": 2.1340782165527346, "step": 58850 }, { "epoch": 0.23433109710915392, "grad_norm": 0.47556301951408386, "learning_rate": 1.765740748108246e-05, "loss": 2.0700311279296875, "step": 58900 }, { "epoch": 0.234530019942014, "grad_norm": 0.5978822708129883, "learning_rate": 1.7655418254732372e-05, "loss": 2.0590110778808595, "step": 58950 }, { "epoch": 0.23472894277487405, "grad_norm": 0.5870944261550903, "learning_rate": 1.7653429028382284e-05, "loss": 2.142244873046875, "step": 59000 }, { "epoch": 0.23492786560773413, "grad_norm": 0.5849129557609558, "learning_rate": 1.7651439802032196e-05, "loss": 2.09810791015625, "step": 59050 }, { "epoch": 0.23512678844059418, "grad_norm": 0.582249104976654, "learning_rate": 1.7649450575682108e-05, "loss": 2.146787414550781, "step": 59100 }, { "epoch": 0.23532571127345425, "grad_norm": 0.5659385323524475, "learning_rate": 1.764746134933202e-05, "loss": 2.117950439453125, "step": 59150 }, { "epoch": 0.2355246341063143, "grad_norm": 0.5322573184967041, "learning_rate": 1.764551190750893e-05, "loss": 2.1162594604492186, "step": 59200 }, { "epoch": 0.23572355693917438, "grad_norm": 0.5322219133377075, "learning_rate": 1.7643522681158846e-05, "loss": 2.105101318359375, "step": 59250 }, { "epoch": 0.23592247977203443, "grad_norm": 0.555664598941803, "learning_rate": 1.7641533454808758e-05, "loss": 2.073489990234375, "step": 59300 }, { "epoch": 0.2361214026048945, "grad_norm": 0.5301822423934937, "learning_rate": 1.763954422845867e-05, "loss": 2.0910110473632812, "step": 59350 }, { "epoch": 0.23632032543775455, "grad_norm": 0.5489861965179443, "learning_rate": 1.7637555002108582e-05, "loss": 2.1065863037109374, "step": 59400 }, { "epoch": 0.23651924827061463, "grad_norm": 0.5464277863502502, "learning_rate": 1.7635565775758495e-05, "loss": 2.105673980712891, "step": 59450 }, { "epoch": 0.23671817110347468, "grad_norm": 0.575156569480896, "learning_rate": 1.7633576549408407e-05, "loss": 2.088802490234375, "step": 59500 }, { "epoch": 0.23691709393633475, "grad_norm": 0.6302264332771301, "learning_rate": 1.7631587323058315e-05, "loss": 2.057101135253906, "step": 59550 }, { "epoch": 0.2371160167691948, "grad_norm": 0.44821757078170776, "learning_rate": 1.7629598096708228e-05, "loss": 2.095486907958984, "step": 59600 }, { "epoch": 0.23731493960205488, "grad_norm": 0.4976816773414612, "learning_rate": 1.7627608870358143e-05, "loss": 2.1363653564453124, "step": 59650 }, { "epoch": 0.23751386243491493, "grad_norm": 0.528415858745575, "learning_rate": 1.7625619644008055e-05, "loss": 2.1065969848632813, "step": 59700 }, { "epoch": 0.237712785267775, "grad_norm": 0.5656834244728088, "learning_rate": 1.7623630417657967e-05, "loss": 2.0764260864257813, "step": 59750 }, { "epoch": 0.23791170810063506, "grad_norm": 0.5804672241210938, "learning_rate": 1.762164119130788e-05, "loss": 2.1101673889160155, "step": 59800 }, { "epoch": 0.23811063093349513, "grad_norm": 0.5047582983970642, "learning_rate": 1.7619651964957788e-05, "loss": 2.1085000610351563, "step": 59850 }, { "epoch": 0.23830955376635518, "grad_norm": 0.5888086557388306, "learning_rate": 1.76176627386077e-05, "loss": 2.063953094482422, "step": 59900 }, { "epoch": 0.23850847659921526, "grad_norm": 0.5991609692573547, "learning_rate": 1.7615673512257613e-05, "loss": 2.0896150207519533, "step": 59950 }, { "epoch": 0.2387073994320753, "grad_norm": 0.5807873010635376, "learning_rate": 1.7613684285907525e-05, "loss": 2.0847837829589846, "step": 60000 }, { "epoch": 0.23890632226493538, "grad_norm": 0.529021680355072, "learning_rate": 1.761169505955744e-05, "loss": 2.1169560241699217, "step": 60050 }, { "epoch": 0.23910524509779543, "grad_norm": 0.5300664901733398, "learning_rate": 1.7609705833207352e-05, "loss": 2.0577020263671875, "step": 60100 }, { "epoch": 0.2393041679306555, "grad_norm": 0.5816794037818909, "learning_rate": 1.7607716606857265e-05, "loss": 2.1198220825195313, "step": 60150 }, { "epoch": 0.23950309076351556, "grad_norm": 0.5242066979408264, "learning_rate": 1.7605727380507173e-05, "loss": 2.1040028381347655, "step": 60200 }, { "epoch": 0.23970201359637563, "grad_norm": 0.6110667586326599, "learning_rate": 1.7603738154157085e-05, "loss": 2.12913330078125, "step": 60250 }, { "epoch": 0.23990093642923568, "grad_norm": 0.558870792388916, "learning_rate": 1.7601748927806998e-05, "loss": 2.0978648376464846, "step": 60300 }, { "epoch": 0.24009985926209576, "grad_norm": 0.46644437313079834, "learning_rate": 1.759975970145691e-05, "loss": 2.132432403564453, "step": 60350 }, { "epoch": 0.2402987820949558, "grad_norm": 0.5511854290962219, "learning_rate": 1.7597770475106822e-05, "loss": 2.082366485595703, "step": 60400 }, { "epoch": 0.2404977049278159, "grad_norm": 0.5960392951965332, "learning_rate": 1.7595781248756737e-05, "loss": 2.079195556640625, "step": 60450 }, { "epoch": 0.24069662776067594, "grad_norm": 0.55750972032547, "learning_rate": 1.7593792022406646e-05, "loss": 2.0561868286132814, "step": 60500 }, { "epoch": 0.240895550593536, "grad_norm": 0.787125825881958, "learning_rate": 1.759180279605656e-05, "loss": 2.0849993896484373, "step": 60550 }, { "epoch": 0.24109447342639606, "grad_norm": 0.5054084062576294, "learning_rate": 1.758981356970647e-05, "loss": 2.138714599609375, "step": 60600 }, { "epoch": 0.24129339625925614, "grad_norm": 0.5234236717224121, "learning_rate": 1.7587824343356383e-05, "loss": 2.0589996337890626, "step": 60650 }, { "epoch": 0.2414923190921162, "grad_norm": 0.5772512555122375, "learning_rate": 1.7585835117006295e-05, "loss": 2.1160018920898436, "step": 60700 }, { "epoch": 0.24169124192497626, "grad_norm": 0.5870538353919983, "learning_rate": 1.7583845890656207e-05, "loss": 2.0705348205566407, "step": 60750 }, { "epoch": 0.2418901647578363, "grad_norm": 0.6032446622848511, "learning_rate": 1.758185666430612e-05, "loss": 2.0962547302246093, "step": 60800 }, { "epoch": 0.2420890875906964, "grad_norm": 0.47302836179733276, "learning_rate": 1.757986743795603e-05, "loss": 2.088584136962891, "step": 60850 }, { "epoch": 0.24228801042355644, "grad_norm": 0.6513030529022217, "learning_rate": 1.7577878211605943e-05, "loss": 2.0919102478027343, "step": 60900 }, { "epoch": 0.24248693325641651, "grad_norm": 0.6404626965522766, "learning_rate": 1.7575888985255856e-05, "loss": 2.1261712646484376, "step": 60950 }, { "epoch": 0.24268585608927656, "grad_norm": 0.5548906922340393, "learning_rate": 1.7573899758905768e-05, "loss": 2.0984175109863283, "step": 61000 }, { "epoch": 0.24288477892213664, "grad_norm": 0.6324447393417358, "learning_rate": 1.757191053255568e-05, "loss": 2.0585787963867186, "step": 61050 }, { "epoch": 0.2430837017549967, "grad_norm": 0.5985217690467834, "learning_rate": 1.7569921306205592e-05, "loss": 2.1185406494140624, "step": 61100 }, { "epoch": 0.24328262458785677, "grad_norm": 0.5228272676467896, "learning_rate": 1.7567932079855504e-05, "loss": 2.0329835510253904, "step": 61150 }, { "epoch": 0.24348154742071682, "grad_norm": 0.6249879002571106, "learning_rate": 1.7565942853505416e-05, "loss": 2.18404541015625, "step": 61200 }, { "epoch": 0.2436804702535769, "grad_norm": 0.5904106497764587, "learning_rate": 1.756395362715533e-05, "loss": 2.1532582092285155, "step": 61250 }, { "epoch": 0.24387939308643694, "grad_norm": 0.5519466400146484, "learning_rate": 1.756196440080524e-05, "loss": 2.0758677673339845, "step": 61300 }, { "epoch": 0.24407831591929702, "grad_norm": 0.6222125291824341, "learning_rate": 1.7559975174455153e-05, "loss": 2.139138031005859, "step": 61350 }, { "epoch": 0.24427723875215707, "grad_norm": 0.5452578663825989, "learning_rate": 1.7557985948105065e-05, "loss": 2.1009776306152346, "step": 61400 }, { "epoch": 0.24447616158501714, "grad_norm": 0.5346028208732605, "learning_rate": 1.7555996721754977e-05, "loss": 2.077801513671875, "step": 61450 }, { "epoch": 0.2446750844178772, "grad_norm": 0.5832170248031616, "learning_rate": 1.755400749540489e-05, "loss": 2.1182281494140627, "step": 61500 }, { "epoch": 0.24487400725073727, "grad_norm": 0.549515962600708, "learning_rate": 1.75520182690548e-05, "loss": 2.093008728027344, "step": 61550 }, { "epoch": 0.24507293008359732, "grad_norm": 0.5906563997268677, "learning_rate": 1.7550029042704713e-05, "loss": 2.1056748962402345, "step": 61600 }, { "epoch": 0.2452718529164574, "grad_norm": 0.6547118425369263, "learning_rate": 1.7548039816354626e-05, "loss": 2.0915687561035154, "step": 61650 }, { "epoch": 0.24547077574931744, "grad_norm": 0.5670909881591797, "learning_rate": 1.7546050590004538e-05, "loss": 2.0890037536621096, "step": 61700 }, { "epoch": 0.24566969858217752, "grad_norm": 0.5425621271133423, "learning_rate": 1.754406136365445e-05, "loss": 2.042249908447266, "step": 61750 }, { "epoch": 0.24586862141503757, "grad_norm": 0.5368083715438843, "learning_rate": 1.754207213730436e-05, "loss": 2.117281799316406, "step": 61800 }, { "epoch": 0.24606754424789765, "grad_norm": 0.5499523282051086, "learning_rate": 1.7540082910954274e-05, "loss": 2.1508282470703124, "step": 61850 }, { "epoch": 0.2462664670807577, "grad_norm": 0.5501684546470642, "learning_rate": 1.7538093684604186e-05, "loss": 2.0762742614746093, "step": 61900 }, { "epoch": 0.24646538991361777, "grad_norm": 0.6779779195785522, "learning_rate": 1.75361044582541e-05, "loss": 2.026292419433594, "step": 61950 }, { "epoch": 0.24666431274647782, "grad_norm": 0.562542200088501, "learning_rate": 1.753411523190401e-05, "loss": 2.0635061645507813, "step": 62000 }, { "epoch": 0.2468632355793379, "grad_norm": 0.5570915341377258, "learning_rate": 1.7532126005553923e-05, "loss": 2.054787902832031, "step": 62050 }, { "epoch": 0.24706215841219795, "grad_norm": 0.5932397246360779, "learning_rate": 1.753013677920383e-05, "loss": 2.1059645080566405, "step": 62100 }, { "epoch": 0.24726108124505802, "grad_norm": 0.5168606638908386, "learning_rate": 1.7528147552853744e-05, "loss": 2.133685302734375, "step": 62150 }, { "epoch": 0.24746000407791807, "grad_norm": 0.5110442638397217, "learning_rate": 1.7526158326503656e-05, "loss": 2.0889817810058595, "step": 62200 }, { "epoch": 0.24765892691077815, "grad_norm": 0.5622414350509644, "learning_rate": 1.752416910015357e-05, "loss": 2.081254730224609, "step": 62250 }, { "epoch": 0.2478578497436382, "grad_norm": 0.657329261302948, "learning_rate": 1.7522179873803483e-05, "loss": 2.0470550537109373, "step": 62300 }, { "epoch": 0.24805677257649827, "grad_norm": 0.6496082544326782, "learning_rate": 1.7520190647453396e-05, "loss": 2.0698587036132814, "step": 62350 }, { "epoch": 0.24825569540935832, "grad_norm": 0.5955235958099365, "learning_rate": 1.7518201421103308e-05, "loss": 2.043252410888672, "step": 62400 }, { "epoch": 0.2484546182422184, "grad_norm": 0.5274964570999146, "learning_rate": 1.7516212194753217e-05, "loss": 2.0964385986328127, "step": 62450 }, { "epoch": 0.24865354107507845, "grad_norm": 0.5127412676811218, "learning_rate": 1.751422296840313e-05, "loss": 2.0654112243652345, "step": 62500 }, { "epoch": 0.24885246390793853, "grad_norm": 0.5956032872200012, "learning_rate": 1.751223374205304e-05, "loss": 2.068228454589844, "step": 62550 }, { "epoch": 0.24905138674079857, "grad_norm": 0.528687596321106, "learning_rate": 1.7510244515702953e-05, "loss": 2.094086151123047, "step": 62600 }, { "epoch": 0.24925030957365865, "grad_norm": 0.5760073065757751, "learning_rate": 1.7508255289352865e-05, "loss": 2.0547724914550782, "step": 62650 }, { "epoch": 0.2494492324065187, "grad_norm": 0.6359015107154846, "learning_rate": 1.750626606300278e-05, "loss": 1.9852583312988281, "step": 62700 }, { "epoch": 0.24964815523937878, "grad_norm": 0.5897312164306641, "learning_rate": 1.750427683665269e-05, "loss": 2.0774307250976562, "step": 62750 }, { "epoch": 0.24984707807223883, "grad_norm": 0.5883996486663818, "learning_rate": 1.75022876103026e-05, "loss": 2.023285675048828, "step": 62800 }, { "epoch": 0.2500460009050989, "grad_norm": 0.5828474164009094, "learning_rate": 1.7500298383952514e-05, "loss": 2.0563978576660156, "step": 62850 }, { "epoch": 0.250244923737959, "grad_norm": 0.5861142873764038, "learning_rate": 1.7498309157602426e-05, "loss": 2.0755076599121094, "step": 62900 }, { "epoch": 0.25044384657081903, "grad_norm": 0.5560398697853088, "learning_rate": 1.7496319931252338e-05, "loss": 2.095537567138672, "step": 62950 }, { "epoch": 0.2506427694036791, "grad_norm": 0.6064199805259705, "learning_rate": 1.749433070490225e-05, "loss": 2.0564434814453123, "step": 63000 }, { "epoch": 0.2508416922365391, "grad_norm": 0.6635186076164246, "learning_rate": 1.7492381263079167e-05, "loss": 2.088545379638672, "step": 63050 }, { "epoch": 0.2510406150693992, "grad_norm": 0.5490158200263977, "learning_rate": 1.7490392036729076e-05, "loss": 2.102775115966797, "step": 63100 }, { "epoch": 0.2512395379022593, "grad_norm": 0.7076172232627869, "learning_rate": 1.7488402810378988e-05, "loss": 2.1674977111816407, "step": 63150 }, { "epoch": 0.25143846073511933, "grad_norm": 0.5440514087677002, "learning_rate": 1.74864135840289e-05, "loss": 2.0687069702148437, "step": 63200 }, { "epoch": 0.2516373835679794, "grad_norm": 0.529190182685852, "learning_rate": 1.7484424357678812e-05, "loss": 2.136824188232422, "step": 63250 }, { "epoch": 0.2518363064008394, "grad_norm": 0.5839245915412903, "learning_rate": 1.7482435131328724e-05, "loss": 2.165859832763672, "step": 63300 }, { "epoch": 0.25203522923369953, "grad_norm": 0.5959023237228394, "learning_rate": 1.7480445904978636e-05, "loss": 2.0737550354003904, "step": 63350 }, { "epoch": 0.2522341520665596, "grad_norm": 0.5601164698600769, "learning_rate": 1.747845667862855e-05, "loss": 2.0916679382324217, "step": 63400 }, { "epoch": 0.25243307489941963, "grad_norm": 0.6586089134216309, "learning_rate": 1.747646745227846e-05, "loss": 2.087838592529297, "step": 63450 }, { "epoch": 0.2526319977322797, "grad_norm": 0.5816461443901062, "learning_rate": 1.7474478225928373e-05, "loss": 2.1060028076171875, "step": 63500 }, { "epoch": 0.2528309205651398, "grad_norm": 0.5477854609489441, "learning_rate": 1.7472488999578285e-05, "loss": 2.1108047485351564, "step": 63550 }, { "epoch": 0.25302984339799983, "grad_norm": 0.5341030955314636, "learning_rate": 1.7470499773228197e-05, "loss": 2.0964649963378905, "step": 63600 }, { "epoch": 0.2532287662308599, "grad_norm": 0.56644207239151, "learning_rate": 1.746851054687811e-05, "loss": 2.0508563232421877, "step": 63650 }, { "epoch": 0.25342768906371993, "grad_norm": 0.5584724545478821, "learning_rate": 1.746652132052802e-05, "loss": 2.08742431640625, "step": 63700 }, { "epoch": 0.25362661189658003, "grad_norm": 0.4776672422885895, "learning_rate": 1.7464532094177934e-05, "loss": 2.0941653442382813, "step": 63750 }, { "epoch": 0.2538255347294401, "grad_norm": 0.3690292239189148, "learning_rate": 1.7462542867827846e-05, "loss": 2.069451904296875, "step": 63800 }, { "epoch": 0.25402445756230013, "grad_norm": 0.5898683071136475, "learning_rate": 1.7460553641477758e-05, "loss": 2.123486328125, "step": 63850 }, { "epoch": 0.2542233803951602, "grad_norm": 0.5512010455131531, "learning_rate": 1.745856441512767e-05, "loss": 2.0334120178222657, "step": 63900 }, { "epoch": 0.2544223032280203, "grad_norm": 0.5630418062210083, "learning_rate": 1.7456575188777582e-05, "loss": 2.0982017517089844, "step": 63950 }, { "epoch": 0.25462122606088033, "grad_norm": 0.5375609397888184, "learning_rate": 1.7454585962427494e-05, "loss": 2.118358917236328, "step": 64000 }, { "epoch": 0.2548201488937404, "grad_norm": 0.5462496280670166, "learning_rate": 1.7452596736077407e-05, "loss": 2.068787078857422, "step": 64050 }, { "epoch": 0.25501907172660043, "grad_norm": 0.5676685571670532, "learning_rate": 1.745060750972732e-05, "loss": 2.109566345214844, "step": 64100 }, { "epoch": 0.25521799455946054, "grad_norm": 0.574386477470398, "learning_rate": 1.744861828337723e-05, "loss": 2.0922311401367186, "step": 64150 }, { "epoch": 0.2554169173923206, "grad_norm": 0.5409818291664124, "learning_rate": 1.7446629057027143e-05, "loss": 2.0483747863769532, "step": 64200 }, { "epoch": 0.25561584022518063, "grad_norm": 0.5326209664344788, "learning_rate": 1.7444639830677055e-05, "loss": 2.1210406494140623, "step": 64250 }, { "epoch": 0.2558147630580407, "grad_norm": 0.6145411133766174, "learning_rate": 1.7442650604326967e-05, "loss": 2.0997465515136717, "step": 64300 }, { "epoch": 0.2560136858909008, "grad_norm": 0.5148624181747437, "learning_rate": 1.744066137797688e-05, "loss": 2.11966064453125, "step": 64350 }, { "epoch": 0.25621260872376084, "grad_norm": 0.5510386228561401, "learning_rate": 1.7438672151626788e-05, "loss": 2.1011141967773437, "step": 64400 }, { "epoch": 0.2564115315566209, "grad_norm": 0.5744731426239014, "learning_rate": 1.7436682925276704e-05, "loss": 2.1205488586425782, "step": 64450 }, { "epoch": 0.25661045438948094, "grad_norm": 0.5431989431381226, "learning_rate": 1.7434693698926616e-05, "loss": 2.0545294189453127, "step": 64500 }, { "epoch": 0.25680937722234104, "grad_norm": 0.5277000069618225, "learning_rate": 1.7432704472576528e-05, "loss": 2.117791748046875, "step": 64550 }, { "epoch": 0.2570083000552011, "grad_norm": 0.5555258989334106, "learning_rate": 1.743071524622644e-05, "loss": 2.0494869995117186, "step": 64600 }, { "epoch": 0.25720722288806114, "grad_norm": 0.5575228929519653, "learning_rate": 1.7428726019876352e-05, "loss": 2.074479827880859, "step": 64650 }, { "epoch": 0.2574061457209212, "grad_norm": 0.6130039691925049, "learning_rate": 1.742673679352626e-05, "loss": 2.1040704345703123, "step": 64700 }, { "epoch": 0.2576050685537813, "grad_norm": 0.5626559257507324, "learning_rate": 1.7424747567176173e-05, "loss": 2.1282247924804687, "step": 64750 }, { "epoch": 0.25780399138664134, "grad_norm": 0.5254698991775513, "learning_rate": 1.7422758340826085e-05, "loss": 2.040620880126953, "step": 64800 }, { "epoch": 0.2580029142195014, "grad_norm": 0.5384527444839478, "learning_rate": 1.7420769114476e-05, "loss": 2.102158508300781, "step": 64850 }, { "epoch": 0.25820183705236144, "grad_norm": 0.5259802341461182, "learning_rate": 1.7418779888125913e-05, "loss": 2.0802763366699217, "step": 64900 }, { "epoch": 0.25840075988522154, "grad_norm": 0.556671142578125, "learning_rate": 1.7416790661775825e-05, "loss": 2.109660186767578, "step": 64950 }, { "epoch": 0.2585996827180816, "grad_norm": 0.626571774482727, "learning_rate": 1.7414801435425737e-05, "loss": 2.1023770141601563, "step": 65000 }, { "epoch": 0.25879860555094164, "grad_norm": 0.5993019342422485, "learning_rate": 1.7412812209075646e-05, "loss": 2.144237060546875, "step": 65050 }, { "epoch": 0.2589975283838017, "grad_norm": 0.549428403377533, "learning_rate": 1.7410822982725558e-05, "loss": 2.0908094787597657, "step": 65100 }, { "epoch": 0.2591964512166618, "grad_norm": 0.48475968837738037, "learning_rate": 1.740883375637547e-05, "loss": 2.0663572692871095, "step": 65150 }, { "epoch": 0.25939537404952184, "grad_norm": 0.5898828506469727, "learning_rate": 1.7406884314552387e-05, "loss": 2.132313995361328, "step": 65200 }, { "epoch": 0.2595942968823819, "grad_norm": 0.5708932280540466, "learning_rate": 1.74048950882023e-05, "loss": 2.0775494384765625, "step": 65250 }, { "epoch": 0.25979321971524194, "grad_norm": 0.4867013394832611, "learning_rate": 1.740290586185221e-05, "loss": 2.0983767700195313, "step": 65300 }, { "epoch": 0.25999214254810205, "grad_norm": 0.5215783715248108, "learning_rate": 1.740091663550212e-05, "loss": 2.0691648864746095, "step": 65350 }, { "epoch": 0.2601910653809621, "grad_norm": 0.5563364028930664, "learning_rate": 1.7398927409152032e-05, "loss": 2.1035835266113283, "step": 65400 }, { "epoch": 0.26038998821382214, "grad_norm": 0.5668522715568542, "learning_rate": 1.7396938182801945e-05, "loss": 2.0711408996582032, "step": 65450 }, { "epoch": 0.2605889110466822, "grad_norm": 0.6013136506080627, "learning_rate": 1.7394948956451857e-05, "loss": 2.06764404296875, "step": 65500 }, { "epoch": 0.2607878338795423, "grad_norm": 0.48542478680610657, "learning_rate": 1.739295973010177e-05, "loss": 2.102000885009766, "step": 65550 }, { "epoch": 0.26098675671240235, "grad_norm": 0.4978347718715668, "learning_rate": 1.7390970503751684e-05, "loss": 2.0702143859863282, "step": 65600 }, { "epoch": 0.2611856795452624, "grad_norm": 0.5174520015716553, "learning_rate": 1.7388981277401597e-05, "loss": 2.084870147705078, "step": 65650 }, { "epoch": 0.26138460237812244, "grad_norm": 0.5792783498764038, "learning_rate": 1.7386992051051505e-05, "loss": 2.057059783935547, "step": 65700 }, { "epoch": 0.26158352521098255, "grad_norm": 0.5429502129554749, "learning_rate": 1.7385002824701417e-05, "loss": 2.08778076171875, "step": 65750 }, { "epoch": 0.2617824480438426, "grad_norm": 0.4906705319881439, "learning_rate": 1.738301359835133e-05, "loss": 2.0726123046875, "step": 65800 }, { "epoch": 0.26198137087670265, "grad_norm": 0.579079270362854, "learning_rate": 1.7381024372001242e-05, "loss": 2.1047299194335936, "step": 65850 }, { "epoch": 0.2621802937095627, "grad_norm": 0.5314404964447021, "learning_rate": 1.7379035145651154e-05, "loss": 2.0739398193359375, "step": 65900 }, { "epoch": 0.2623792165424228, "grad_norm": 0.5481660962104797, "learning_rate": 1.7377045919301066e-05, "loss": 2.068204803466797, "step": 65950 }, { "epoch": 0.26257813937528285, "grad_norm": 0.5808189511299133, "learning_rate": 1.7375056692950978e-05, "loss": 2.085831604003906, "step": 66000 }, { "epoch": 0.2627770622081429, "grad_norm": 0.59426349401474, "learning_rate": 1.737306746660089e-05, "loss": 2.0846343994140626, "step": 66050 }, { "epoch": 0.26297598504100295, "grad_norm": 0.5478283762931824, "learning_rate": 1.7371078240250802e-05, "loss": 2.0790943908691406, "step": 66100 }, { "epoch": 0.26317490787386305, "grad_norm": 0.6115398406982422, "learning_rate": 1.7369089013900715e-05, "loss": 2.057633972167969, "step": 66150 }, { "epoch": 0.2633738307067231, "grad_norm": 0.5769546627998352, "learning_rate": 1.7367099787550627e-05, "loss": 2.1191134643554688, "step": 66200 }, { "epoch": 0.26357275353958315, "grad_norm": 0.6357139945030212, "learning_rate": 1.736511056120054e-05, "loss": 2.123850860595703, "step": 66250 }, { "epoch": 0.2637716763724432, "grad_norm": 0.49979114532470703, "learning_rate": 1.736312133485045e-05, "loss": 2.0892234802246095, "step": 66300 }, { "epoch": 0.2639705992053033, "grad_norm": 0.6080743074417114, "learning_rate": 1.7361132108500363e-05, "loss": 2.166942901611328, "step": 66350 }, { "epoch": 0.26416952203816335, "grad_norm": 0.603980541229248, "learning_rate": 1.7359142882150275e-05, "loss": 2.0950616455078124, "step": 66400 }, { "epoch": 0.2643684448710234, "grad_norm": 0.5499782562255859, "learning_rate": 1.7357153655800188e-05, "loss": 2.024669494628906, "step": 66450 }, { "epoch": 0.26456736770388345, "grad_norm": 0.61979079246521, "learning_rate": 1.73551644294501e-05, "loss": 2.057142639160156, "step": 66500 }, { "epoch": 0.26476629053674355, "grad_norm": 0.6151476502418518, "learning_rate": 1.7353175203100012e-05, "loss": 2.0661552429199217, "step": 66550 }, { "epoch": 0.2649652133696036, "grad_norm": 0.6692804098129272, "learning_rate": 1.7351185976749924e-05, "loss": 2.0729939270019533, "step": 66600 }, { "epoch": 0.26516413620246365, "grad_norm": 0.5941460132598877, "learning_rate": 1.7349196750399836e-05, "loss": 2.0674397277832033, "step": 66650 }, { "epoch": 0.2653630590353237, "grad_norm": 0.5169015526771545, "learning_rate": 1.7347207524049748e-05, "loss": 2.080846710205078, "step": 66700 }, { "epoch": 0.2655619818681838, "grad_norm": 0.468179851770401, "learning_rate": 1.734521829769966e-05, "loss": 2.105485687255859, "step": 66750 }, { "epoch": 0.26576090470104385, "grad_norm": 0.4080422818660736, "learning_rate": 1.7343229071349573e-05, "loss": 2.1088858032226563, "step": 66800 }, { "epoch": 0.2659598275339039, "grad_norm": 0.5516209006309509, "learning_rate": 1.7341239844999485e-05, "loss": 2.0725042724609377, "step": 66850 }, { "epoch": 0.26615875036676395, "grad_norm": 0.542499303817749, "learning_rate": 1.7339250618649397e-05, "loss": 2.122627410888672, "step": 66900 }, { "epoch": 0.26635767319962406, "grad_norm": 0.3886195719242096, "learning_rate": 1.733726139229931e-05, "loss": 2.129516906738281, "step": 66950 }, { "epoch": 0.2665565960324841, "grad_norm": 0.472261905670166, "learning_rate": 1.733527216594922e-05, "loss": 2.1005116271972657, "step": 67000 }, { "epoch": 0.26675551886534415, "grad_norm": 0.520654022693634, "learning_rate": 1.7333282939599133e-05, "loss": 2.0305715942382814, "step": 67050 }, { "epoch": 0.2669544416982042, "grad_norm": 0.5595295429229736, "learning_rate": 1.7331293713249045e-05, "loss": 2.0706996154785156, "step": 67100 }, { "epoch": 0.2671533645310643, "grad_norm": 0.4995846450328827, "learning_rate": 1.7329304486898958e-05, "loss": 2.066569366455078, "step": 67150 }, { "epoch": 0.26735228736392436, "grad_norm": 0.49182096123695374, "learning_rate": 1.732731526054887e-05, "loss": 2.0552632141113283, "step": 67200 }, { "epoch": 0.2675512101967844, "grad_norm": 0.6015332937240601, "learning_rate": 1.7325326034198782e-05, "loss": 2.0591932678222657, "step": 67250 }, { "epoch": 0.26775013302964445, "grad_norm": 0.5231198668479919, "learning_rate": 1.732333680784869e-05, "loss": 2.100852813720703, "step": 67300 }, { "epoch": 0.26794905586250456, "grad_norm": 0.5911427140235901, "learning_rate": 1.7321347581498603e-05, "loss": 2.100482635498047, "step": 67350 }, { "epoch": 0.2681479786953646, "grad_norm": 0.47365498542785645, "learning_rate": 1.7319358355148518e-05, "loss": 2.0361289978027344, "step": 67400 }, { "epoch": 0.26834690152822466, "grad_norm": 0.6663581132888794, "learning_rate": 1.731736912879843e-05, "loss": 2.1187127685546874, "step": 67450 }, { "epoch": 0.2685458243610847, "grad_norm": 0.5694706439971924, "learning_rate": 1.7315379902448343e-05, "loss": 2.0778599548339844, "step": 67500 }, { "epoch": 0.2687447471939448, "grad_norm": 0.5426257848739624, "learning_rate": 1.7313390676098255e-05, "loss": 2.0475588989257814, "step": 67550 }, { "epoch": 0.26894367002680486, "grad_norm": 0.5841568112373352, "learning_rate": 1.7311401449748167e-05, "loss": 2.072598571777344, "step": 67600 }, { "epoch": 0.2691425928596649, "grad_norm": 0.5567362904548645, "learning_rate": 1.7309412223398076e-05, "loss": 2.0605117797851564, "step": 67650 }, { "epoch": 0.26934151569252496, "grad_norm": 0.6150649785995483, "learning_rate": 1.7307422997047988e-05, "loss": 2.091070098876953, "step": 67700 }, { "epoch": 0.26954043852538506, "grad_norm": 0.5658698081970215, "learning_rate": 1.73054337706979e-05, "loss": 2.1065357971191405, "step": 67750 }, { "epoch": 0.2697393613582451, "grad_norm": 0.6224231123924255, "learning_rate": 1.7303444544347815e-05, "loss": 2.043379211425781, "step": 67800 }, { "epoch": 0.26993828419110516, "grad_norm": 0.5211924910545349, "learning_rate": 1.7301455317997728e-05, "loss": 2.114945068359375, "step": 67850 }, { "epoch": 0.2701372070239652, "grad_norm": 0.5017222166061401, "learning_rate": 1.729946609164764e-05, "loss": 2.049123077392578, "step": 67900 }, { "epoch": 0.2703361298568253, "grad_norm": 0.674353301525116, "learning_rate": 1.729747686529755e-05, "loss": 2.0677622985839843, "step": 67950 }, { "epoch": 0.27053505268968536, "grad_norm": 0.5669772624969482, "learning_rate": 1.729548763894746e-05, "loss": 2.112833709716797, "step": 68000 }, { "epoch": 0.2707339755225454, "grad_norm": 0.5161846876144409, "learning_rate": 1.7293498412597373e-05, "loss": 2.069213104248047, "step": 68050 }, { "epoch": 0.27093289835540546, "grad_norm": 0.45089051127433777, "learning_rate": 1.7291509186247285e-05, "loss": 2.115355224609375, "step": 68100 }, { "epoch": 0.27113182118826556, "grad_norm": 0.6123613119125366, "learning_rate": 1.7289519959897197e-05, "loss": 2.069251708984375, "step": 68150 }, { "epoch": 0.2713307440211256, "grad_norm": 0.5897037982940674, "learning_rate": 1.7287530733547113e-05, "loss": 2.1418283081054685, "step": 68200 }, { "epoch": 0.27152966685398566, "grad_norm": 0.6908036470413208, "learning_rate": 1.7285541507197025e-05, "loss": 2.085340118408203, "step": 68250 }, { "epoch": 0.2717285896868457, "grad_norm": 0.551016628742218, "learning_rate": 1.7283552280846934e-05, "loss": 2.065923614501953, "step": 68300 }, { "epoch": 0.2719275125197058, "grad_norm": 0.5544448494911194, "learning_rate": 1.7281563054496846e-05, "loss": 2.039539642333984, "step": 68350 }, { "epoch": 0.27212643535256587, "grad_norm": 0.6259133219718933, "learning_rate": 1.7279573828146758e-05, "loss": 2.057326202392578, "step": 68400 }, { "epoch": 0.2723253581854259, "grad_norm": 0.4706394672393799, "learning_rate": 1.727758460179667e-05, "loss": 2.0708619689941408, "step": 68450 }, { "epoch": 0.27252428101828596, "grad_norm": 0.5321121215820312, "learning_rate": 1.7275595375446582e-05, "loss": 2.0996995544433594, "step": 68500 }, { "epoch": 0.27272320385114607, "grad_norm": 0.5984963178634644, "learning_rate": 1.7273606149096494e-05, "loss": 2.025102996826172, "step": 68550 }, { "epoch": 0.2729221266840061, "grad_norm": 0.5567344427108765, "learning_rate": 1.7271616922746406e-05, "loss": 2.088634033203125, "step": 68600 }, { "epoch": 0.27312104951686617, "grad_norm": 0.5728836059570312, "learning_rate": 1.726962769639632e-05, "loss": 2.1306028747558594, "step": 68650 }, { "epoch": 0.2733199723497262, "grad_norm": 0.5157707333564758, "learning_rate": 1.726763847004623e-05, "loss": 2.0755426025390626, "step": 68700 }, { "epoch": 0.2735188951825863, "grad_norm": 0.47181424498558044, "learning_rate": 1.7265649243696143e-05, "loss": 2.0510324096679686, "step": 68750 }, { "epoch": 0.27371781801544637, "grad_norm": 0.41806983947753906, "learning_rate": 1.7263660017346055e-05, "loss": 2.07963623046875, "step": 68800 }, { "epoch": 0.2739167408483064, "grad_norm": 0.5520937442779541, "learning_rate": 1.7261670790995967e-05, "loss": 2.0886305236816405, "step": 68850 }, { "epoch": 0.27411566368116647, "grad_norm": 0.5429663062095642, "learning_rate": 1.725968156464588e-05, "loss": 2.1095205688476564, "step": 68900 }, { "epoch": 0.27431458651402657, "grad_norm": 0.5873296856880188, "learning_rate": 1.725769233829579e-05, "loss": 2.0497142028808595, "step": 68950 }, { "epoch": 0.2745135093468866, "grad_norm": 0.5130357146263123, "learning_rate": 1.7255703111945704e-05, "loss": 2.056355285644531, "step": 69000 }, { "epoch": 0.27471243217974667, "grad_norm": 0.5762559175491333, "learning_rate": 1.7253753670122617e-05, "loss": 2.0865032958984373, "step": 69050 }, { "epoch": 0.2749113550126067, "grad_norm": 0.5803307890892029, "learning_rate": 1.725176444377253e-05, "loss": 2.0278077697753907, "step": 69100 }, { "epoch": 0.2751102778454668, "grad_norm": 0.49866926670074463, "learning_rate": 1.724977521742244e-05, "loss": 2.0854164123535157, "step": 69150 }, { "epoch": 0.27530920067832687, "grad_norm": 0.4374507665634155, "learning_rate": 1.7247785991072354e-05, "loss": 2.0363540649414062, "step": 69200 }, { "epoch": 0.2755081235111869, "grad_norm": 0.5829781889915466, "learning_rate": 1.7245796764722266e-05, "loss": 2.08948974609375, "step": 69250 }, { "epoch": 0.27570704634404697, "grad_norm": 0.5564028024673462, "learning_rate": 1.7243807538372178e-05, "loss": 2.0673094177246094, "step": 69300 }, { "epoch": 0.2759059691769071, "grad_norm": 0.6584615707397461, "learning_rate": 1.724181831202209e-05, "loss": 2.0815892028808594, "step": 69350 }, { "epoch": 0.2761048920097671, "grad_norm": 0.677443265914917, "learning_rate": 1.7239829085672002e-05, "loss": 2.071532897949219, "step": 69400 }, { "epoch": 0.27630381484262717, "grad_norm": 0.5485062599182129, "learning_rate": 1.7237839859321914e-05, "loss": 2.0768809509277344, "step": 69450 }, { "epoch": 0.2765027376754872, "grad_norm": 0.5906112790107727, "learning_rate": 1.7235850632971826e-05, "loss": 2.0113645935058595, "step": 69500 }, { "epoch": 0.2767016605083473, "grad_norm": 0.5485422611236572, "learning_rate": 1.723386140662174e-05, "loss": 2.090776824951172, "step": 69550 }, { "epoch": 0.2769005833412074, "grad_norm": 0.5918004512786865, "learning_rate": 1.723187218027165e-05, "loss": 1.9984451293945313, "step": 69600 }, { "epoch": 0.2770995061740674, "grad_norm": 0.6977009773254395, "learning_rate": 1.7229882953921563e-05, "loss": 2.1202349853515625, "step": 69650 }, { "epoch": 0.27729842900692747, "grad_norm": 0.5893908143043518, "learning_rate": 1.7227893727571475e-05, "loss": 2.068106994628906, "step": 69700 }, { "epoch": 0.2774973518397876, "grad_norm": 0.5610613822937012, "learning_rate": 1.7225904501221387e-05, "loss": 2.0706553649902344, "step": 69750 }, { "epoch": 0.2776962746726476, "grad_norm": 0.6241777539253235, "learning_rate": 1.72239152748713e-05, "loss": 2.0479167175292967, "step": 69800 }, { "epoch": 0.2778951975055077, "grad_norm": 0.6886526346206665, "learning_rate": 1.722192604852121e-05, "loss": 2.0829905700683593, "step": 69850 }, { "epoch": 0.2780941203383677, "grad_norm": 0.6415451765060425, "learning_rate": 1.721993682217112e-05, "loss": 2.0989190673828126, "step": 69900 }, { "epoch": 0.2782930431712278, "grad_norm": 0.6080844402313232, "learning_rate": 1.7217947595821036e-05, "loss": 2.0787088012695314, "step": 69950 }, { "epoch": 0.2784919660040879, "grad_norm": 0.5175934433937073, "learning_rate": 1.7215958369470948e-05, "loss": 2.02710693359375, "step": 70000 }, { "epoch": 0.2786908888369479, "grad_norm": 0.666074275970459, "learning_rate": 1.721396914312086e-05, "loss": 2.095478820800781, "step": 70050 }, { "epoch": 0.278889811669808, "grad_norm": 0.5225263237953186, "learning_rate": 1.7211979916770772e-05, "loss": 2.061832275390625, "step": 70100 }, { "epoch": 0.2790887345026681, "grad_norm": 0.7065442204475403, "learning_rate": 1.7209990690420684e-05, "loss": 2.0809706115722655, "step": 70150 }, { "epoch": 0.27928765733552813, "grad_norm": 0.48049941658973694, "learning_rate": 1.7208001464070596e-05, "loss": 2.1171218872070314, "step": 70200 }, { "epoch": 0.2794865801683882, "grad_norm": 0.5921735167503357, "learning_rate": 1.7206012237720505e-05, "loss": 2.104828338623047, "step": 70250 }, { "epoch": 0.2796855030012482, "grad_norm": 0.5453161597251892, "learning_rate": 1.7204023011370417e-05, "loss": 2.077190704345703, "step": 70300 }, { "epoch": 0.27988442583410833, "grad_norm": 0.521581768989563, "learning_rate": 1.720203378502033e-05, "loss": 2.0359121704101564, "step": 70350 }, { "epoch": 0.2800833486669684, "grad_norm": 0.5136622786521912, "learning_rate": 1.7200044558670245e-05, "loss": 2.149873046875, "step": 70400 }, { "epoch": 0.28028227149982843, "grad_norm": 0.5623121857643127, "learning_rate": 1.7198055332320157e-05, "loss": 2.0796382141113283, "step": 70450 }, { "epoch": 0.2804811943326885, "grad_norm": 0.6272450685501099, "learning_rate": 1.719606610597007e-05, "loss": 2.0654296875, "step": 70500 }, { "epoch": 0.2806801171655486, "grad_norm": 0.6001415848731995, "learning_rate": 1.7194076879619978e-05, "loss": 2.0734193420410154, "step": 70550 }, { "epoch": 0.28087903999840863, "grad_norm": 0.44984883069992065, "learning_rate": 1.719208765326989e-05, "loss": 2.0261468505859375, "step": 70600 }, { "epoch": 0.2810779628312687, "grad_norm": 0.5410088896751404, "learning_rate": 1.7190098426919802e-05, "loss": 2.099534149169922, "step": 70650 }, { "epoch": 0.28127688566412873, "grad_norm": 0.5566971302032471, "learning_rate": 1.7188109200569715e-05, "loss": 2.1489717102050783, "step": 70700 }, { "epoch": 0.28147580849698883, "grad_norm": 0.5947183966636658, "learning_rate": 1.7186119974219627e-05, "loss": 2.0741363525390626, "step": 70750 }, { "epoch": 0.2816747313298489, "grad_norm": 0.48056983947753906, "learning_rate": 1.7184130747869542e-05, "loss": 2.0935960388183594, "step": 70800 }, { "epoch": 0.28187365416270893, "grad_norm": 0.6072959899902344, "learning_rate": 1.7182141521519454e-05, "loss": 2.091233367919922, "step": 70850 }, { "epoch": 0.282072576995569, "grad_norm": 0.5498788952827454, "learning_rate": 1.7180152295169363e-05, "loss": 2.112096405029297, "step": 70900 }, { "epoch": 0.2822714998284291, "grad_norm": 0.5714774131774902, "learning_rate": 1.7178163068819275e-05, "loss": 2.032332000732422, "step": 70950 }, { "epoch": 0.28247042266128913, "grad_norm": 0.6167571544647217, "learning_rate": 1.7176173842469187e-05, "loss": 2.1223490905761717, "step": 71000 }, { "epoch": 0.2826693454941492, "grad_norm": 0.5277132987976074, "learning_rate": 1.71741846161191e-05, "loss": 2.058310699462891, "step": 71050 }, { "epoch": 0.28286826832700923, "grad_norm": 0.5641026496887207, "learning_rate": 1.717219538976901e-05, "loss": 2.0913902282714845, "step": 71100 }, { "epoch": 0.2830671911598693, "grad_norm": 0.5271800756454468, "learning_rate": 1.7170206163418924e-05, "loss": 2.095482177734375, "step": 71150 }, { "epoch": 0.2832661139927294, "grad_norm": 0.5498457551002502, "learning_rate": 1.7168216937068836e-05, "loss": 2.100777893066406, "step": 71200 }, { "epoch": 0.28346503682558943, "grad_norm": 0.5101550221443176, "learning_rate": 1.7166227710718748e-05, "loss": 2.0549560546875, "step": 71250 }, { "epoch": 0.2836639596584495, "grad_norm": 0.5910105109214783, "learning_rate": 1.716423848436866e-05, "loss": 2.0500909423828126, "step": 71300 }, { "epoch": 0.28386288249130953, "grad_norm": 0.5660295486450195, "learning_rate": 1.7162249258018572e-05, "loss": 2.102925109863281, "step": 71350 }, { "epoch": 0.28406180532416964, "grad_norm": 0.523663341999054, "learning_rate": 1.7160260031668485e-05, "loss": 2.0430999755859376, "step": 71400 }, { "epoch": 0.2842607281570297, "grad_norm": 0.5723249316215515, "learning_rate": 1.7158270805318397e-05, "loss": 2.0233721923828125, "step": 71450 }, { "epoch": 0.28445965098988973, "grad_norm": 0.5606951117515564, "learning_rate": 1.715628157896831e-05, "loss": 2.030916442871094, "step": 71500 }, { "epoch": 0.2846585738227498, "grad_norm": 0.5494405031204224, "learning_rate": 1.715429235261822e-05, "loss": 2.1034136962890626, "step": 71550 }, { "epoch": 0.2848574966556099, "grad_norm": 0.5197064280509949, "learning_rate": 1.7152303126268133e-05, "loss": 2.0605955505371094, "step": 71600 }, { "epoch": 0.28505641948846994, "grad_norm": 0.5493270754814148, "learning_rate": 1.7150313899918045e-05, "loss": 2.0740516662597654, "step": 71650 }, { "epoch": 0.28525534232133, "grad_norm": 0.5868731141090393, "learning_rate": 1.7148324673567957e-05, "loss": 2.0358375549316405, "step": 71700 }, { "epoch": 0.28545426515419003, "grad_norm": 0.6032951474189758, "learning_rate": 1.714633544721787e-05, "loss": 2.0790025329589845, "step": 71750 }, { "epoch": 0.28565318798705014, "grad_norm": 0.6787834763526917, "learning_rate": 1.7144346220867782e-05, "loss": 2.0640936279296875, "step": 71800 }, { "epoch": 0.2858521108199102, "grad_norm": 0.5657801032066345, "learning_rate": 1.7142356994517694e-05, "loss": 2.095294189453125, "step": 71850 }, { "epoch": 0.28605103365277024, "grad_norm": 0.5791577100753784, "learning_rate": 1.7140367768167606e-05, "loss": 2.054870300292969, "step": 71900 }, { "epoch": 0.2862499564856303, "grad_norm": 0.4967980682849884, "learning_rate": 1.7138378541817518e-05, "loss": 2.069802093505859, "step": 71950 }, { "epoch": 0.2864488793184904, "grad_norm": 0.5210077166557312, "learning_rate": 1.713638931546743e-05, "loss": 2.1137060546875, "step": 72000 }, { "epoch": 0.28664780215135044, "grad_norm": 0.6740646362304688, "learning_rate": 1.7134400089117342e-05, "loss": 2.053013916015625, "step": 72050 }, { "epoch": 0.2868467249842105, "grad_norm": 0.5627349019050598, "learning_rate": 1.7132410862767255e-05, "loss": 2.0917543029785155, "step": 72100 }, { "epoch": 0.28704564781707054, "grad_norm": 0.49933120608329773, "learning_rate": 1.7130421636417163e-05, "loss": 2.1878984069824217, "step": 72150 }, { "epoch": 0.28724457064993064, "grad_norm": 0.6411178112030029, "learning_rate": 1.712843241006708e-05, "loss": 2.1032919311523437, "step": 72200 }, { "epoch": 0.2874434934827907, "grad_norm": 0.6261463761329651, "learning_rate": 1.7126482968243992e-05, "loss": 2.104584045410156, "step": 72250 }, { "epoch": 0.28764241631565074, "grad_norm": 0.4934305250644684, "learning_rate": 1.7124493741893905e-05, "loss": 2.1536656188964844, "step": 72300 }, { "epoch": 0.2878413391485108, "grad_norm": 0.5141853094100952, "learning_rate": 1.7122504515543817e-05, "loss": 2.1237521362304688, "step": 72350 }, { "epoch": 0.2880402619813709, "grad_norm": 0.5500419735908508, "learning_rate": 1.712051528919373e-05, "loss": 2.0650018310546874, "step": 72400 }, { "epoch": 0.28823918481423094, "grad_norm": 0.6214693784713745, "learning_rate": 1.711852606284364e-05, "loss": 2.0301495361328126, "step": 72450 }, { "epoch": 0.288438107647091, "grad_norm": 0.5897942185401917, "learning_rate": 1.711653683649355e-05, "loss": 2.0507293701171876, "step": 72500 }, { "epoch": 0.28863703047995104, "grad_norm": 0.5933854579925537, "learning_rate": 1.7114547610143465e-05, "loss": 2.071910705566406, "step": 72550 }, { "epoch": 0.28883595331281114, "grad_norm": 0.5978907346725464, "learning_rate": 1.7112558383793377e-05, "loss": 2.0735833740234373, "step": 72600 }, { "epoch": 0.2890348761456712, "grad_norm": 0.4992750287055969, "learning_rate": 1.711056915744329e-05, "loss": 2.0467514038085937, "step": 72650 }, { "epoch": 0.28923379897853124, "grad_norm": 0.5001775026321411, "learning_rate": 1.7108579931093202e-05, "loss": 2.07505126953125, "step": 72700 }, { "epoch": 0.2894327218113913, "grad_norm": 0.5756205916404724, "learning_rate": 1.7106590704743114e-05, "loss": 1.9970584106445313, "step": 72750 }, { "epoch": 0.2896316446442514, "grad_norm": 0.5938371419906616, "learning_rate": 1.7104601478393026e-05, "loss": 2.072227020263672, "step": 72800 }, { "epoch": 0.28983056747711144, "grad_norm": 0.564937949180603, "learning_rate": 1.7102612252042935e-05, "loss": 2.065592803955078, "step": 72850 }, { "epoch": 0.2900294903099715, "grad_norm": 0.590303897857666, "learning_rate": 1.7100623025692847e-05, "loss": 2.0223159790039062, "step": 72900 }, { "epoch": 0.29022841314283154, "grad_norm": 0.5885422825813293, "learning_rate": 1.7098633799342762e-05, "loss": 2.0550587463378904, "step": 72950 }, { "epoch": 0.29042733597569165, "grad_norm": 0.5654144287109375, "learning_rate": 1.7096644572992675e-05, "loss": 2.1253483581542967, "step": 73000 }, { "epoch": 0.2906262588085517, "grad_norm": 0.574239194393158, "learning_rate": 1.7094655346642587e-05, "loss": 2.0562672424316406, "step": 73050 }, { "epoch": 0.29082518164141175, "grad_norm": 0.5677664279937744, "learning_rate": 1.70926661202925e-05, "loss": 2.050592346191406, "step": 73100 }, { "epoch": 0.2910241044742718, "grad_norm": 0.5833063721656799, "learning_rate": 1.7090676893942408e-05, "loss": 2.0371278381347655, "step": 73150 }, { "epoch": 0.2912230273071319, "grad_norm": 0.603339672088623, "learning_rate": 1.708868766759232e-05, "loss": 2.1055633544921877, "step": 73200 }, { "epoch": 0.29142195013999195, "grad_norm": 0.514190137386322, "learning_rate": 1.7086698441242232e-05, "loss": 2.041348876953125, "step": 73250 }, { "epoch": 0.291620872972852, "grad_norm": 0.550898551940918, "learning_rate": 1.7084709214892144e-05, "loss": 2.0812977600097655, "step": 73300 }, { "epoch": 0.29181979580571205, "grad_norm": 0.6104974150657654, "learning_rate": 1.708271998854206e-05, "loss": 2.0986589050292968, "step": 73350 }, { "epoch": 0.29201871863857215, "grad_norm": 0.5753252506256104, "learning_rate": 1.7080730762191972e-05, "loss": 2.1055331420898438, "step": 73400 }, { "epoch": 0.2922176414714322, "grad_norm": 0.5665411949157715, "learning_rate": 1.7078741535841884e-05, "loss": 2.131777038574219, "step": 73450 }, { "epoch": 0.29241656430429225, "grad_norm": 0.5489396452903748, "learning_rate": 1.7076752309491793e-05, "loss": 2.0875810241699218, "step": 73500 }, { "epoch": 0.2926154871371523, "grad_norm": 0.44506680965423584, "learning_rate": 1.7074763083141705e-05, "loss": 2.078926544189453, "step": 73550 }, { "epoch": 0.2928144099700124, "grad_norm": 0.5464196801185608, "learning_rate": 1.7072773856791617e-05, "loss": 2.0548805236816405, "step": 73600 }, { "epoch": 0.29301333280287245, "grad_norm": 0.4848789870738983, "learning_rate": 1.707078463044153e-05, "loss": 2.038734893798828, "step": 73650 }, { "epoch": 0.2932122556357325, "grad_norm": 0.6000980734825134, "learning_rate": 1.706879540409144e-05, "loss": 2.03200927734375, "step": 73700 }, { "epoch": 0.29341117846859255, "grad_norm": 0.4893133044242859, "learning_rate": 1.7066806177741357e-05, "loss": 2.071172790527344, "step": 73750 }, { "epoch": 0.29361010130145265, "grad_norm": 0.6377317309379578, "learning_rate": 1.7064816951391266e-05, "loss": 2.081318359375, "step": 73800 }, { "epoch": 0.2938090241343127, "grad_norm": 0.5487603545188904, "learning_rate": 1.7062827725041178e-05, "loss": 2.1047732543945314, "step": 73850 }, { "epoch": 0.29400794696717275, "grad_norm": 0.5300293564796448, "learning_rate": 1.706083849869109e-05, "loss": 2.086111297607422, "step": 73900 }, { "epoch": 0.2942068698000328, "grad_norm": 0.5021640658378601, "learning_rate": 1.7058849272341002e-05, "loss": 2.1484686279296876, "step": 73950 }, { "epoch": 0.2944057926328929, "grad_norm": 0.5869803428649902, "learning_rate": 1.7056860045990914e-05, "loss": 2.0946588134765625, "step": 74000 }, { "epoch": 0.29460471546575295, "grad_norm": 0.5726786255836487, "learning_rate": 1.7054870819640826e-05, "loss": 2.1015565490722654, "step": 74050 }, { "epoch": 0.294803638298613, "grad_norm": 0.5223619341850281, "learning_rate": 1.705288159329074e-05, "loss": 2.079811859130859, "step": 74100 }, { "epoch": 0.29500256113147305, "grad_norm": 0.4221661686897278, "learning_rate": 1.705089236694065e-05, "loss": 2.105879211425781, "step": 74150 }, { "epoch": 0.29520148396433316, "grad_norm": 0.6226012110710144, "learning_rate": 1.7048903140590563e-05, "loss": 2.017183990478516, "step": 74200 }, { "epoch": 0.2954004067971932, "grad_norm": 0.5304207801818848, "learning_rate": 1.7046913914240475e-05, "loss": 2.0553936767578125, "step": 74250 }, { "epoch": 0.29559932963005325, "grad_norm": 0.5171613097190857, "learning_rate": 1.7044924687890387e-05, "loss": 2.0760568237304686, "step": 74300 }, { "epoch": 0.2957982524629133, "grad_norm": 0.46522465348243713, "learning_rate": 1.70429354615403e-05, "loss": 2.0787725830078125, "step": 74350 }, { "epoch": 0.2959971752957734, "grad_norm": 0.5647649765014648, "learning_rate": 1.704094623519021e-05, "loss": 2.0597259521484377, "step": 74400 }, { "epoch": 0.29619609812863346, "grad_norm": 0.5099721550941467, "learning_rate": 1.7038957008840123e-05, "loss": 2.039805450439453, "step": 74450 }, { "epoch": 0.2963950209614935, "grad_norm": 0.5710187554359436, "learning_rate": 1.7036967782490036e-05, "loss": 2.06597412109375, "step": 74500 }, { "epoch": 0.29659394379435355, "grad_norm": 0.44745492935180664, "learning_rate": 1.7034978556139948e-05, "loss": 2.062616729736328, "step": 74550 }, { "epoch": 0.29679286662721366, "grad_norm": 0.5513758063316345, "learning_rate": 1.703298932978986e-05, "loss": 2.1189541625976562, "step": 74600 }, { "epoch": 0.2969917894600737, "grad_norm": 0.5675778388977051, "learning_rate": 1.7031000103439772e-05, "loss": 2.114825439453125, "step": 74650 }, { "epoch": 0.29719071229293376, "grad_norm": 0.6296405792236328, "learning_rate": 1.7029010877089684e-05, "loss": 2.0337571716308593, "step": 74700 }, { "epoch": 0.2973896351257938, "grad_norm": 0.4917977750301361, "learning_rate": 1.7027021650739596e-05, "loss": 2.017528839111328, "step": 74750 }, { "epoch": 0.2975885579586539, "grad_norm": 0.6008421182632446, "learning_rate": 1.702503242438951e-05, "loss": 2.112841033935547, "step": 74800 }, { "epoch": 0.29778748079151396, "grad_norm": 0.5452347993850708, "learning_rate": 1.702304319803942e-05, "loss": 2.0825315856933595, "step": 74850 }, { "epoch": 0.297986403624374, "grad_norm": 0.48933812975883484, "learning_rate": 1.7021053971689333e-05, "loss": 2.0837599182128907, "step": 74900 }, { "epoch": 0.29818532645723406, "grad_norm": 0.44992920756340027, "learning_rate": 1.7019064745339245e-05, "loss": 2.066339874267578, "step": 74950 }, { "epoch": 0.29838424929009416, "grad_norm": 0.5523433089256287, "learning_rate": 1.7017075518989157e-05, "loss": 2.091971893310547, "step": 75000 }, { "epoch": 0.2985831721229542, "grad_norm": 0.5257256627082825, "learning_rate": 1.701508629263907e-05, "loss": 2.1081565856933593, "step": 75050 }, { "epoch": 0.29878209495581426, "grad_norm": 0.5658153891563416, "learning_rate": 1.7013097066288978e-05, "loss": 2.0669993591308593, "step": 75100 }, { "epoch": 0.2989810177886743, "grad_norm": 0.4661053419113159, "learning_rate": 1.7011107839938893e-05, "loss": 2.0756097412109376, "step": 75150 }, { "epoch": 0.2991799406215344, "grad_norm": 0.5531963109970093, "learning_rate": 1.7009118613588806e-05, "loss": 2.0449310302734376, "step": 75200 }, { "epoch": 0.29937886345439446, "grad_norm": 0.4590567946434021, "learning_rate": 1.7007129387238718e-05, "loss": 2.0703358459472656, "step": 75250 }, { "epoch": 0.2995777862872545, "grad_norm": 0.5862204432487488, "learning_rate": 1.700514016088863e-05, "loss": 2.028150787353516, "step": 75300 }, { "epoch": 0.29977670912011456, "grad_norm": 0.5758904814720154, "learning_rate": 1.7003150934538542e-05, "loss": 2.041060485839844, "step": 75350 }, { "epoch": 0.29997563195297466, "grad_norm": 0.6212184429168701, "learning_rate": 1.700116170818845e-05, "loss": 2.0304051208496094, "step": 75400 }, { "epoch": 0.3001745547858347, "grad_norm": 0.6714374423027039, "learning_rate": 1.6999172481838363e-05, "loss": 2.058799285888672, "step": 75450 }, { "epoch": 0.30037347761869476, "grad_norm": 0.5104044079780579, "learning_rate": 1.6997183255488275e-05, "loss": 2.0535177612304687, "step": 75500 }, { "epoch": 0.3005724004515548, "grad_norm": 0.6064487099647522, "learning_rate": 1.6995194029138187e-05, "loss": 2.0614250183105467, "step": 75550 }, { "epoch": 0.3007713232844149, "grad_norm": 0.5214465260505676, "learning_rate": 1.6993204802788103e-05, "loss": 2.0574330139160155, "step": 75600 }, { "epoch": 0.30097024611727496, "grad_norm": 0.5690756440162659, "learning_rate": 1.6991215576438015e-05, "loss": 2.095877838134766, "step": 75650 }, { "epoch": 0.301169168950135, "grad_norm": 0.6063743233680725, "learning_rate": 1.6989226350087927e-05, "loss": 2.040966796875, "step": 75700 }, { "epoch": 0.30136809178299506, "grad_norm": 0.5534660816192627, "learning_rate": 1.6987276908264837e-05, "loss": 2.0576890563964843, "step": 75750 }, { "epoch": 0.30156701461585517, "grad_norm": 0.567627489566803, "learning_rate": 1.698528768191475e-05, "loss": 2.033245391845703, "step": 75800 }, { "epoch": 0.3017659374487152, "grad_norm": 0.6075568795204163, "learning_rate": 1.698329845556466e-05, "loss": 2.057825927734375, "step": 75850 }, { "epoch": 0.30196486028157526, "grad_norm": 0.6015084981918335, "learning_rate": 1.6981309229214577e-05, "loss": 2.1111134338378905, "step": 75900 }, { "epoch": 0.3021637831144353, "grad_norm": 0.53606778383255, "learning_rate": 1.697932000286449e-05, "loss": 2.072620849609375, "step": 75950 }, { "epoch": 0.3023627059472954, "grad_norm": 0.5401234030723572, "learning_rate": 1.69773307765144e-05, "loss": 2.0654421997070314, "step": 76000 }, { "epoch": 0.30256162878015547, "grad_norm": 0.5633761882781982, "learning_rate": 1.697534155016431e-05, "loss": 2.094582977294922, "step": 76050 }, { "epoch": 0.3027605516130155, "grad_norm": 0.5267698764801025, "learning_rate": 1.6973352323814222e-05, "loss": 2.0974696350097655, "step": 76100 }, { "epoch": 0.30295947444587557, "grad_norm": 0.5797390937805176, "learning_rate": 1.6971363097464134e-05, "loss": 2.039922332763672, "step": 76150 }, { "epoch": 0.30315839727873567, "grad_norm": 0.5711543560028076, "learning_rate": 1.6969373871114047e-05, "loss": 2.0907293701171876, "step": 76200 }, { "epoch": 0.3033573201115957, "grad_norm": 0.5571615099906921, "learning_rate": 1.696738464476396e-05, "loss": 2.042142181396484, "step": 76250 }, { "epoch": 0.30355624294445577, "grad_norm": 0.5933647155761719, "learning_rate": 1.6965395418413874e-05, "loss": 2.100109100341797, "step": 76300 }, { "epoch": 0.3037551657773158, "grad_norm": 0.4766092300415039, "learning_rate": 1.6963406192063786e-05, "loss": 2.0762037658691406, "step": 76350 }, { "epoch": 0.3039540886101759, "grad_norm": 0.5763885378837585, "learning_rate": 1.6961416965713695e-05, "loss": 2.0815614318847655, "step": 76400 }, { "epoch": 0.30415301144303597, "grad_norm": 0.7345082759857178, "learning_rate": 1.6959427739363607e-05, "loss": 2.049954681396484, "step": 76450 }, { "epoch": 0.304351934275896, "grad_norm": 0.5943505167961121, "learning_rate": 1.695743851301352e-05, "loss": 2.0630975341796876, "step": 76500 }, { "epoch": 0.30455085710875607, "grad_norm": 0.5783537030220032, "learning_rate": 1.695544928666343e-05, "loss": 2.0806649780273436, "step": 76550 }, { "epoch": 0.3047497799416162, "grad_norm": 0.5654114484786987, "learning_rate": 1.6953460060313344e-05, "loss": 2.0571028137207032, "step": 76600 }, { "epoch": 0.3049487027744762, "grad_norm": 0.5419055223464966, "learning_rate": 1.6951470833963256e-05, "loss": 2.096832275390625, "step": 76650 }, { "epoch": 0.30514762560733627, "grad_norm": 0.6143912672996521, "learning_rate": 1.6949481607613168e-05, "loss": 2.083215789794922, "step": 76700 }, { "epoch": 0.3053465484401963, "grad_norm": 0.5260093212127686, "learning_rate": 1.694749238126308e-05, "loss": 2.085826416015625, "step": 76750 }, { "epoch": 0.3055454712730564, "grad_norm": 0.5624952912330627, "learning_rate": 1.6945503154912992e-05, "loss": 2.070975494384766, "step": 76800 }, { "epoch": 0.3057443941059165, "grad_norm": 0.5673276782035828, "learning_rate": 1.6943513928562904e-05, "loss": 2.037998046875, "step": 76850 }, { "epoch": 0.3059433169387765, "grad_norm": 0.5674716234207153, "learning_rate": 1.6941524702212817e-05, "loss": 2.055018005371094, "step": 76900 }, { "epoch": 0.30614223977163657, "grad_norm": 0.5710904002189636, "learning_rate": 1.693953547586273e-05, "loss": 2.0356428527832033, "step": 76950 }, { "epoch": 0.3063411626044967, "grad_norm": 0.5834421515464783, "learning_rate": 1.693754624951264e-05, "loss": 2.0607505798339845, "step": 77000 }, { "epoch": 0.3065400854373567, "grad_norm": 0.6381275653839111, "learning_rate": 1.6935557023162553e-05, "loss": 2.08331787109375, "step": 77050 }, { "epoch": 0.3067390082702168, "grad_norm": 0.4706171751022339, "learning_rate": 1.6933567796812465e-05, "loss": 2.0732046508789064, "step": 77100 }, { "epoch": 0.3069379311030768, "grad_norm": 0.5404936075210571, "learning_rate": 1.6931578570462377e-05, "loss": 2.065638122558594, "step": 77150 }, { "epoch": 0.3071368539359369, "grad_norm": 0.5091927647590637, "learning_rate": 1.692958934411229e-05, "loss": 2.046165771484375, "step": 77200 }, { "epoch": 0.307335776768797, "grad_norm": 0.3504844009876251, "learning_rate": 1.69276001177622e-05, "loss": 2.1422979736328127, "step": 77250 }, { "epoch": 0.307534699601657, "grad_norm": 0.5938659906387329, "learning_rate": 1.6925610891412114e-05, "loss": 2.0330992126464844, "step": 77300 }, { "epoch": 0.3077336224345171, "grad_norm": 0.4972893297672272, "learning_rate": 1.6923621665062026e-05, "loss": 2.0269337463378907, "step": 77350 }, { "epoch": 0.3079325452673772, "grad_norm": 0.652092695236206, "learning_rate": 1.6921632438711938e-05, "loss": 2.0775128173828126, "step": 77400 }, { "epoch": 0.3081314681002372, "grad_norm": 0.6661313772201538, "learning_rate": 1.691964321236185e-05, "loss": 2.0448216247558593, "step": 77450 }, { "epoch": 0.3083303909330973, "grad_norm": 0.7061251401901245, "learning_rate": 1.6917653986011762e-05, "loss": 2.1082579040527345, "step": 77500 }, { "epoch": 0.3085293137659573, "grad_norm": 0.5523152351379395, "learning_rate": 1.6915664759661674e-05, "loss": 2.0415931701660157, "step": 77550 }, { "epoch": 0.30872823659881743, "grad_norm": 0.5683842301368713, "learning_rate": 1.6913675533311587e-05, "loss": 2.0529496765136717, "step": 77600 }, { "epoch": 0.3089271594316775, "grad_norm": 0.5120072960853577, "learning_rate": 1.69117260914885e-05, "loss": 2.0919395446777345, "step": 77650 }, { "epoch": 0.3091260822645375, "grad_norm": 0.5415822267532349, "learning_rate": 1.6909736865138412e-05, "loss": 2.0734677124023437, "step": 77700 }, { "epoch": 0.3093250050973976, "grad_norm": 0.5844326019287109, "learning_rate": 1.6907747638788324e-05, "loss": 2.086569519042969, "step": 77750 }, { "epoch": 0.3095239279302577, "grad_norm": 0.49805131554603577, "learning_rate": 1.6905758412438237e-05, "loss": 2.0681779479980467, "step": 77800 }, { "epoch": 0.30972285076311773, "grad_norm": 0.47317761182785034, "learning_rate": 1.690376918608815e-05, "loss": 2.0852198791503906, "step": 77850 }, { "epoch": 0.3099217735959778, "grad_norm": 0.5209531188011169, "learning_rate": 1.690177995973806e-05, "loss": 2.113123474121094, "step": 77900 }, { "epoch": 0.3101206964288378, "grad_norm": 0.667961061000824, "learning_rate": 1.6899790733387973e-05, "loss": 2.022644805908203, "step": 77950 }, { "epoch": 0.31031961926169793, "grad_norm": 0.4894404411315918, "learning_rate": 1.6897801507037882e-05, "loss": 2.0402508544921876, "step": 78000 }, { "epoch": 0.310518542094558, "grad_norm": 0.6332370042800903, "learning_rate": 1.6895812280687794e-05, "loss": 2.0267755126953126, "step": 78050 }, { "epoch": 0.31071746492741803, "grad_norm": 0.5427396893501282, "learning_rate": 1.689382305433771e-05, "loss": 2.115980987548828, "step": 78100 }, { "epoch": 0.3109163877602781, "grad_norm": 0.5055705904960632, "learning_rate": 1.689183382798762e-05, "loss": 2.060768737792969, "step": 78150 }, { "epoch": 0.3111153105931382, "grad_norm": 0.5311993360519409, "learning_rate": 1.6889844601637534e-05, "loss": 2.080915832519531, "step": 78200 }, { "epoch": 0.31131423342599823, "grad_norm": 0.6259278655052185, "learning_rate": 1.6887855375287446e-05, "loss": 2.0648439025878904, "step": 78250 }, { "epoch": 0.3115131562588583, "grad_norm": 0.6003364324569702, "learning_rate": 1.6885866148937358e-05, "loss": 2.0519947814941406, "step": 78300 }, { "epoch": 0.31171207909171833, "grad_norm": 0.517795979976654, "learning_rate": 1.6883876922587267e-05, "loss": 2.0737123107910156, "step": 78350 }, { "epoch": 0.31191100192457843, "grad_norm": 0.5278976559638977, "learning_rate": 1.688188769623718e-05, "loss": 2.0397744750976563, "step": 78400 }, { "epoch": 0.3121099247574385, "grad_norm": 0.6427380442619324, "learning_rate": 1.687989846988709e-05, "loss": 2.053964080810547, "step": 78450 }, { "epoch": 0.31230884759029853, "grad_norm": 0.5273250937461853, "learning_rate": 1.6877909243537007e-05, "loss": 2.0426339721679687, "step": 78500 }, { "epoch": 0.3125077704231586, "grad_norm": 0.5919299721717834, "learning_rate": 1.687592001718692e-05, "loss": 2.0734791564941406, "step": 78550 }, { "epoch": 0.3127066932560187, "grad_norm": 0.6015663146972656, "learning_rate": 1.687393079083683e-05, "loss": 2.0205094909667967, "step": 78600 }, { "epoch": 0.31290561608887874, "grad_norm": 0.5043601989746094, "learning_rate": 1.687194156448674e-05, "loss": 2.024877166748047, "step": 78650 }, { "epoch": 0.3131045389217388, "grad_norm": 0.5176621675491333, "learning_rate": 1.6869952338136652e-05, "loss": 2.063232421875, "step": 78700 }, { "epoch": 0.31330346175459883, "grad_norm": 0.7035320401191711, "learning_rate": 1.6867963111786564e-05, "loss": 2.077167205810547, "step": 78750 }, { "epoch": 0.31350238458745894, "grad_norm": 0.5440734028816223, "learning_rate": 1.6865973885436476e-05, "loss": 2.062893981933594, "step": 78800 }, { "epoch": 0.313701307420319, "grad_norm": 0.7007911801338196, "learning_rate": 1.6863984659086388e-05, "loss": 2.0289260864257814, "step": 78850 }, { "epoch": 0.31390023025317904, "grad_norm": 0.558647096157074, "learning_rate": 1.6862035217263305e-05, "loss": 2.087496643066406, "step": 78900 }, { "epoch": 0.3140991530860391, "grad_norm": 0.5881429314613342, "learning_rate": 1.6860045990913217e-05, "loss": 2.08771484375, "step": 78950 }, { "epoch": 0.31429807591889913, "grad_norm": 0.5669193863868713, "learning_rate": 1.6858056764563126e-05, "loss": 2.059434661865234, "step": 79000 }, { "epoch": 0.31449699875175924, "grad_norm": 0.6032732725143433, "learning_rate": 1.6856067538213038e-05, "loss": 2.0682859802246094, "step": 79050 }, { "epoch": 0.3146959215846193, "grad_norm": 0.5526472330093384, "learning_rate": 1.685407831186295e-05, "loss": 2.1127810668945313, "step": 79100 }, { "epoch": 0.31489484441747934, "grad_norm": 0.4976632297039032, "learning_rate": 1.6852089085512862e-05, "loss": 2.127614288330078, "step": 79150 }, { "epoch": 0.3150937672503394, "grad_norm": 0.4939351975917816, "learning_rate": 1.6850099859162775e-05, "loss": 2.11957763671875, "step": 79200 }, { "epoch": 0.3152926900831995, "grad_norm": 0.5492998957633972, "learning_rate": 1.684811063281269e-05, "loss": 2.068908996582031, "step": 79250 }, { "epoch": 0.31549161291605954, "grad_norm": 0.48011404275894165, "learning_rate": 1.68461214064626e-05, "loss": 2.044095306396484, "step": 79300 }, { "epoch": 0.3156905357489196, "grad_norm": 0.53934246301651, "learning_rate": 1.684413218011251e-05, "loss": 2.0362548828125, "step": 79350 }, { "epoch": 0.31588945858177964, "grad_norm": 0.5596928000450134, "learning_rate": 1.6842142953762423e-05, "loss": 2.028395233154297, "step": 79400 }, { "epoch": 0.31608838141463974, "grad_norm": 0.536526620388031, "learning_rate": 1.6840153727412335e-05, "loss": 2.024553070068359, "step": 79450 }, { "epoch": 0.3162873042474998, "grad_norm": 0.5985233187675476, "learning_rate": 1.6838164501062247e-05, "loss": 2.017264099121094, "step": 79500 }, { "epoch": 0.31648622708035984, "grad_norm": 0.4387560188770294, "learning_rate": 1.683617527471216e-05, "loss": 2.019980621337891, "step": 79550 }, { "epoch": 0.3166851499132199, "grad_norm": 0.49528929591178894, "learning_rate": 1.6834186048362072e-05, "loss": 2.0479737854003908, "step": 79600 }, { "epoch": 0.31688407274608, "grad_norm": 0.5985044240951538, "learning_rate": 1.6832196822011984e-05, "loss": 2.0376939392089843, "step": 79650 }, { "epoch": 0.31708299557894004, "grad_norm": 0.5676377415657043, "learning_rate": 1.6830207595661896e-05, "loss": 2.001690673828125, "step": 79700 }, { "epoch": 0.3172819184118001, "grad_norm": 0.5911245942115784, "learning_rate": 1.6828218369311808e-05, "loss": 2.0946044921875, "step": 79750 }, { "epoch": 0.31748084124466014, "grad_norm": 0.644193708896637, "learning_rate": 1.682622914296172e-05, "loss": 2.0901885986328126, "step": 79800 }, { "epoch": 0.31767976407752024, "grad_norm": 0.4350212514400482, "learning_rate": 1.6824239916611632e-05, "loss": 2.0569970703125, "step": 79850 }, { "epoch": 0.3178786869103803, "grad_norm": 0.5059679746627808, "learning_rate": 1.6822250690261545e-05, "loss": 2.0225987243652344, "step": 79900 }, { "epoch": 0.31807760974324034, "grad_norm": 0.5285889506340027, "learning_rate": 1.6820261463911457e-05, "loss": 2.041403961181641, "step": 79950 }, { "epoch": 0.3182765325761004, "grad_norm": 0.5118769407272339, "learning_rate": 1.681827223756137e-05, "loss": 2.0663990783691406, "step": 80000 }, { "epoch": 0.3184754554089605, "grad_norm": 0.6020757555961609, "learning_rate": 1.681628301121128e-05, "loss": 2.04179931640625, "step": 80050 }, { "epoch": 0.31867437824182054, "grad_norm": 0.5389184951782227, "learning_rate": 1.6814293784861193e-05, "loss": 2.0407731628417967, "step": 80100 }, { "epoch": 0.3188733010746806, "grad_norm": 0.5985107421875, "learning_rate": 1.6812304558511105e-05, "loss": 2.108462066650391, "step": 80150 }, { "epoch": 0.31907222390754064, "grad_norm": 0.5269830226898193, "learning_rate": 1.6810315332161018e-05, "loss": 2.056180877685547, "step": 80200 }, { "epoch": 0.31927114674040075, "grad_norm": 0.6651145815849304, "learning_rate": 1.680832610581093e-05, "loss": 2.0530943298339843, "step": 80250 }, { "epoch": 0.3194700695732608, "grad_norm": 0.5350845456123352, "learning_rate": 1.6806336879460842e-05, "loss": 2.0420684814453125, "step": 80300 }, { "epoch": 0.31966899240612084, "grad_norm": 0.50331711769104, "learning_rate": 1.6804347653110754e-05, "loss": 2.0480426025390623, "step": 80350 }, { "epoch": 0.3198679152389809, "grad_norm": 0.6237742900848389, "learning_rate": 1.6802358426760666e-05, "loss": 2.03310302734375, "step": 80400 }, { "epoch": 0.320066838071841, "grad_norm": 0.4960459768772125, "learning_rate": 1.6800369200410578e-05, "loss": 2.111714630126953, "step": 80450 }, { "epoch": 0.32026576090470105, "grad_norm": 0.583190381526947, "learning_rate": 1.679837997406049e-05, "loss": 1.9881723022460938, "step": 80500 }, { "epoch": 0.3204646837375611, "grad_norm": 0.5428025126457214, "learning_rate": 1.6796390747710403e-05, "loss": 2.084092712402344, "step": 80550 }, { "epoch": 0.32066360657042114, "grad_norm": 0.6019826531410217, "learning_rate": 1.679440152136031e-05, "loss": 2.1062553405761717, "step": 80600 }, { "epoch": 0.32086252940328125, "grad_norm": 0.5786023736000061, "learning_rate": 1.6792412295010227e-05, "loss": 2.0257534790039062, "step": 80650 }, { "epoch": 0.3210614522361413, "grad_norm": 0.5944143533706665, "learning_rate": 1.679042306866014e-05, "loss": 2.063681335449219, "step": 80700 }, { "epoch": 0.32126037506900135, "grad_norm": 0.5818614959716797, "learning_rate": 1.678843384231005e-05, "loss": 2.0447235107421875, "step": 80750 }, { "epoch": 0.3214592979018614, "grad_norm": 0.5061950087547302, "learning_rate": 1.6786444615959963e-05, "loss": 2.1285026550292967, "step": 80800 }, { "epoch": 0.3216582207347215, "grad_norm": 0.6544302701950073, "learning_rate": 1.6784455389609875e-05, "loss": 2.0117842102050782, "step": 80850 }, { "epoch": 0.32185714356758155, "grad_norm": 0.6422328948974609, "learning_rate": 1.6782466163259788e-05, "loss": 2.0978543090820314, "step": 80900 }, { "epoch": 0.3220560664004416, "grad_norm": 0.5732906460762024, "learning_rate": 1.6780476936909696e-05, "loss": 2.0823300170898436, "step": 80950 }, { "epoch": 0.32225498923330165, "grad_norm": 0.5987274050712585, "learning_rate": 1.677848771055961e-05, "loss": 2.001516571044922, "step": 81000 }, { "epoch": 0.32245391206616175, "grad_norm": 0.4935825765132904, "learning_rate": 1.6776498484209524e-05, "loss": 2.092604064941406, "step": 81050 }, { "epoch": 0.3226528348990218, "grad_norm": 0.5762448310852051, "learning_rate": 1.6774509257859436e-05, "loss": 2.0461109924316405, "step": 81100 }, { "epoch": 0.32285175773188185, "grad_norm": 0.5916804075241089, "learning_rate": 1.6772520031509348e-05, "loss": 1.978888702392578, "step": 81150 }, { "epoch": 0.3230506805647419, "grad_norm": 0.5335211157798767, "learning_rate": 1.677053080515926e-05, "loss": 2.0465768432617186, "step": 81200 }, { "epoch": 0.323249603397602, "grad_norm": 0.651292085647583, "learning_rate": 1.676854157880917e-05, "loss": 2.0203501892089846, "step": 81250 }, { "epoch": 0.32344852623046205, "grad_norm": 0.5976051092147827, "learning_rate": 1.676655235245908e-05, "loss": 2.077975006103516, "step": 81300 }, { "epoch": 0.3236474490633221, "grad_norm": 0.4983349144458771, "learning_rate": 1.6764563126108993e-05, "loss": 2.0015650939941407, "step": 81350 }, { "epoch": 0.32384637189618215, "grad_norm": 0.4124213457107544, "learning_rate": 1.6762573899758906e-05, "loss": 2.0793296813964846, "step": 81400 }, { "epoch": 0.32404529472904225, "grad_norm": 0.5865325927734375, "learning_rate": 1.676058467340882e-05, "loss": 2.036834411621094, "step": 81450 }, { "epoch": 0.3242442175619023, "grad_norm": 0.6158815622329712, "learning_rate": 1.6758595447058733e-05, "loss": 2.057825469970703, "step": 81500 }, { "epoch": 0.32444314039476235, "grad_norm": 0.5751641988754272, "learning_rate": 1.6756606220708645e-05, "loss": 2.083798522949219, "step": 81550 }, { "epoch": 0.3246420632276224, "grad_norm": 0.6019526124000549, "learning_rate": 1.6754616994358554e-05, "loss": 2.029754333496094, "step": 81600 }, { "epoch": 0.3248409860604825, "grad_norm": 0.6188103556632996, "learning_rate": 1.6752627768008466e-05, "loss": 2.043090362548828, "step": 81650 }, { "epoch": 0.32503990889334256, "grad_norm": 0.6240272521972656, "learning_rate": 1.675063854165838e-05, "loss": 2.025235137939453, "step": 81700 }, { "epoch": 0.3252388317262026, "grad_norm": 0.6686885356903076, "learning_rate": 1.674864931530829e-05, "loss": 2.085228271484375, "step": 81750 }, { "epoch": 0.32543775455906265, "grad_norm": 0.7139090895652771, "learning_rate": 1.6746660088958203e-05, "loss": 2.0954241943359375, "step": 81800 }, { "epoch": 0.32563667739192276, "grad_norm": 0.5823469161987305, "learning_rate": 1.674467086260812e-05, "loss": 2.0654507446289063, "step": 81850 }, { "epoch": 0.3258356002247828, "grad_norm": 0.5375186204910278, "learning_rate": 1.6742681636258027e-05, "loss": 2.057413787841797, "step": 81900 }, { "epoch": 0.32603452305764286, "grad_norm": 0.6461812257766724, "learning_rate": 1.674069240990794e-05, "loss": 2.095292816162109, "step": 81950 }, { "epoch": 0.3262334458905029, "grad_norm": 0.48154187202453613, "learning_rate": 1.673870318355785e-05, "loss": 2.0119346618652343, "step": 82000 }, { "epoch": 0.326432368723363, "grad_norm": 0.5356967449188232, "learning_rate": 1.6736713957207764e-05, "loss": 2.014232940673828, "step": 82050 }, { "epoch": 0.32663129155622306, "grad_norm": 0.5016456842422485, "learning_rate": 1.6734724730857676e-05, "loss": 2.0315672302246095, "step": 82100 }, { "epoch": 0.3268302143890831, "grad_norm": 0.49846139550209045, "learning_rate": 1.6732735504507588e-05, "loss": 2.010486907958984, "step": 82150 }, { "epoch": 0.32702913722194316, "grad_norm": 0.6087013483047485, "learning_rate": 1.67307462781575e-05, "loss": 2.053579864501953, "step": 82200 }, { "epoch": 0.32722806005480326, "grad_norm": 0.5401783585548401, "learning_rate": 1.6728757051807412e-05, "loss": 1.996375274658203, "step": 82250 }, { "epoch": 0.3274269828876633, "grad_norm": 0.6424230933189392, "learning_rate": 1.6726767825457324e-05, "loss": 2.0224960327148436, "step": 82300 }, { "epoch": 0.32762590572052336, "grad_norm": 0.5623034238815308, "learning_rate": 1.6724778599107236e-05, "loss": 2.0220597839355468, "step": 82350 }, { "epoch": 0.3278248285533834, "grad_norm": 0.5754292011260986, "learning_rate": 1.672278937275715e-05, "loss": 2.07582763671875, "step": 82400 }, { "epoch": 0.3280237513862435, "grad_norm": 0.5997067093849182, "learning_rate": 1.672080014640706e-05, "loss": 2.022105712890625, "step": 82450 }, { "epoch": 0.32822267421910356, "grad_norm": 0.5992316603660583, "learning_rate": 1.6718810920056973e-05, "loss": 1.9628582763671876, "step": 82500 }, { "epoch": 0.3284215970519636, "grad_norm": 0.5961392521858215, "learning_rate": 1.6716821693706885e-05, "loss": 2.071912841796875, "step": 82550 }, { "epoch": 0.32862051988482366, "grad_norm": 0.4970846474170685, "learning_rate": 1.6714832467356797e-05, "loss": 2.0705442810058594, "step": 82600 }, { "epoch": 0.32881944271768376, "grad_norm": 0.5425998568534851, "learning_rate": 1.671284324100671e-05, "loss": 2.0125791931152346, "step": 82650 }, { "epoch": 0.3290183655505438, "grad_norm": 0.4557241201400757, "learning_rate": 1.671085401465662e-05, "loss": 2.056190185546875, "step": 82700 }, { "epoch": 0.32921728838340386, "grad_norm": 0.6350930333137512, "learning_rate": 1.6708864788306534e-05, "loss": 2.056061553955078, "step": 82750 }, { "epoch": 0.3294162112162639, "grad_norm": 0.5886679291725159, "learning_rate": 1.6706875561956446e-05, "loss": 2.014700164794922, "step": 82800 }, { "epoch": 0.329615134049124, "grad_norm": 0.585071861743927, "learning_rate": 1.6704886335606358e-05, "loss": 2.1038368225097654, "step": 82850 }, { "epoch": 0.32981405688198406, "grad_norm": 0.5832191705703735, "learning_rate": 1.670289710925627e-05, "loss": 2.041897888183594, "step": 82900 }, { "epoch": 0.3300129797148441, "grad_norm": 0.6465365886688232, "learning_rate": 1.6700907882906182e-05, "loss": 2.1438137817382814, "step": 82950 }, { "epoch": 0.33021190254770416, "grad_norm": 0.5311741232872009, "learning_rate": 1.6698918656556094e-05, "loss": 2.0482774353027344, "step": 83000 }, { "epoch": 0.33041082538056427, "grad_norm": 0.4981730580329895, "learning_rate": 1.6696929430206006e-05, "loss": 1.9775538635253906, "step": 83050 }, { "epoch": 0.3306097482134243, "grad_norm": 0.558242678642273, "learning_rate": 1.669494020385592e-05, "loss": 2.1075843811035155, "step": 83100 }, { "epoch": 0.33080867104628436, "grad_norm": 0.5627507567405701, "learning_rate": 1.669295097750583e-05, "loss": 2.032883605957031, "step": 83150 }, { "epoch": 0.3310075938791444, "grad_norm": 0.5430936217308044, "learning_rate": 1.669096175115574e-05, "loss": 2.053883819580078, "step": 83200 }, { "epoch": 0.3312065167120045, "grad_norm": 0.661983072757721, "learning_rate": 1.668897252480565e-05, "loss": 2.005581817626953, "step": 83250 }, { "epoch": 0.33140543954486457, "grad_norm": 0.5765153765678406, "learning_rate": 1.6686983298455567e-05, "loss": 2.1016246032714845, "step": 83300 }, { "epoch": 0.3316043623777246, "grad_norm": 0.643967866897583, "learning_rate": 1.668499407210548e-05, "loss": 2.0064129638671875, "step": 83350 }, { "epoch": 0.33180328521058466, "grad_norm": 0.6185842156410217, "learning_rate": 1.668300484575539e-05, "loss": 2.0744699096679686, "step": 83400 }, { "epoch": 0.33200220804344477, "grad_norm": 0.5793312788009644, "learning_rate": 1.6681015619405304e-05, "loss": 1.9923016357421874, "step": 83450 }, { "epoch": 0.3322011308763048, "grad_norm": 0.6905924081802368, "learning_rate": 1.6679026393055216e-05, "loss": 2.1084669494628905, "step": 83500 }, { "epoch": 0.33240005370916487, "grad_norm": 0.573638916015625, "learning_rate": 1.6677037166705125e-05, "loss": 2.0252525329589846, "step": 83550 }, { "epoch": 0.3325989765420249, "grad_norm": 0.471744179725647, "learning_rate": 1.6675047940355037e-05, "loss": 2.0147628784179688, "step": 83600 }, { "epoch": 0.332797899374885, "grad_norm": 0.5371356010437012, "learning_rate": 1.667305871400495e-05, "loss": 2.0131352233886717, "step": 83650 }, { "epoch": 0.33299682220774507, "grad_norm": 0.5315324664115906, "learning_rate": 1.6671069487654864e-05, "loss": 1.988987579345703, "step": 83700 }, { "epoch": 0.3331957450406051, "grad_norm": 0.6239974498748779, "learning_rate": 1.6669080261304777e-05, "loss": 2.0453524780273438, "step": 83750 }, { "epoch": 0.33339466787346517, "grad_norm": 0.50672447681427, "learning_rate": 1.666709103495469e-05, "loss": 2.0767033386230467, "step": 83800 }, { "epoch": 0.33359359070632527, "grad_norm": 0.4957294464111328, "learning_rate": 1.6665101808604597e-05, "loss": 2.044126281738281, "step": 83850 }, { "epoch": 0.3337925135391853, "grad_norm": 0.5274726152420044, "learning_rate": 1.666311258225451e-05, "loss": 2.082652282714844, "step": 83900 }, { "epoch": 0.33399143637204537, "grad_norm": 0.603851854801178, "learning_rate": 1.6661123355904422e-05, "loss": 2.109138946533203, "step": 83950 }, { "epoch": 0.3341903592049054, "grad_norm": 0.5502905249595642, "learning_rate": 1.6659134129554334e-05, "loss": 2.0445295715332032, "step": 84000 }, { "epoch": 0.3343892820377655, "grad_norm": 0.5660884380340576, "learning_rate": 1.6657144903204246e-05, "loss": 2.045934753417969, "step": 84050 }, { "epoch": 0.33458820487062557, "grad_norm": 0.596332848072052, "learning_rate": 1.665515567685416e-05, "loss": 2.053942108154297, "step": 84100 }, { "epoch": 0.3347871277034856, "grad_norm": 0.576592743396759, "learning_rate": 1.665316645050407e-05, "loss": 2.044197998046875, "step": 84150 }, { "epoch": 0.33498605053634567, "grad_norm": 0.6195060610771179, "learning_rate": 1.6651177224153982e-05, "loss": 1.9982008361816406, "step": 84200 }, { "epoch": 0.3351849733692058, "grad_norm": 0.5750425457954407, "learning_rate": 1.6649187997803895e-05, "loss": 2.066806182861328, "step": 84250 }, { "epoch": 0.3353838962020658, "grad_norm": 0.5850114822387695, "learning_rate": 1.6647198771453807e-05, "loss": 2.014828796386719, "step": 84300 }, { "epoch": 0.33558281903492587, "grad_norm": 0.5137722492218018, "learning_rate": 1.664520954510372e-05, "loss": 2.039979248046875, "step": 84350 }, { "epoch": 0.3357817418677859, "grad_norm": 0.6489835381507874, "learning_rate": 1.664322031875363e-05, "loss": 2.0468510437011718, "step": 84400 }, { "epoch": 0.335980664700646, "grad_norm": 0.6591596603393555, "learning_rate": 1.6641231092403543e-05, "loss": 2.0497898864746094, "step": 84450 }, { "epoch": 0.3361795875335061, "grad_norm": 0.550435483455658, "learning_rate": 1.6639241866053455e-05, "loss": 2.0151628112792968, "step": 84500 }, { "epoch": 0.3363785103663661, "grad_norm": 0.5802210569381714, "learning_rate": 1.6637252639703367e-05, "loss": 2.04058837890625, "step": 84550 }, { "epoch": 0.3365774331992262, "grad_norm": 0.46405893564224243, "learning_rate": 1.663526341335328e-05, "loss": 2.0645220947265623, "step": 84600 }, { "epoch": 0.3367763560320863, "grad_norm": 0.5472344160079956, "learning_rate": 1.6633274187003192e-05, "loss": 2.04767822265625, "step": 84650 }, { "epoch": 0.3369752788649463, "grad_norm": 0.6338501572608948, "learning_rate": 1.6631284960653104e-05, "loss": 2.0599635314941405, "step": 84700 }, { "epoch": 0.3371742016978064, "grad_norm": 0.6520425081253052, "learning_rate": 1.6629295734303016e-05, "loss": 2.0244676208496095, "step": 84750 }, { "epoch": 0.3373731245306664, "grad_norm": 0.5524817109107971, "learning_rate": 1.6627306507952928e-05, "loss": 2.022299041748047, "step": 84800 }, { "epoch": 0.33757204736352653, "grad_norm": 0.6028332114219666, "learning_rate": 1.662531728160284e-05, "loss": 2.0613754272460936, "step": 84850 }, { "epoch": 0.3377709701963866, "grad_norm": 0.5687604546546936, "learning_rate": 1.6623328055252752e-05, "loss": 2.0713172912597657, "step": 84900 }, { "epoch": 0.3379698930292466, "grad_norm": 0.6231784224510193, "learning_rate": 1.6621338828902665e-05, "loss": 2.0282832336425782, "step": 84950 }, { "epoch": 0.3381688158621067, "grad_norm": 0.5514401197433472, "learning_rate": 1.6619349602552577e-05, "loss": 2.0481527709960936, "step": 85000 }, { "epoch": 0.3383677386949668, "grad_norm": 0.6031825542449951, "learning_rate": 1.661736037620249e-05, "loss": 2.0203399658203125, "step": 85050 }, { "epoch": 0.33856666152782683, "grad_norm": 0.5769655108451843, "learning_rate": 1.66153711498524e-05, "loss": 2.057270050048828, "step": 85100 }, { "epoch": 0.3387655843606869, "grad_norm": 0.6940752863883972, "learning_rate": 1.6613381923502313e-05, "loss": 2.0287535095214846, "step": 85150 }, { "epoch": 0.3389645071935469, "grad_norm": 0.6894735097885132, "learning_rate": 1.6611392697152225e-05, "loss": 2.0492453002929687, "step": 85200 }, { "epoch": 0.33916343002640703, "grad_norm": 0.5278728604316711, "learning_rate": 1.6609403470802137e-05, "loss": 2.076227111816406, "step": 85250 }, { "epoch": 0.3393623528592671, "grad_norm": 0.5895153880119324, "learning_rate": 1.660741424445205e-05, "loss": 2.0393190002441406, "step": 85300 }, { "epoch": 0.33956127569212713, "grad_norm": 0.7955146431922913, "learning_rate": 1.6605425018101962e-05, "loss": 2.050367736816406, "step": 85350 }, { "epoch": 0.3397601985249872, "grad_norm": 0.5462652444839478, "learning_rate": 1.6603435791751874e-05, "loss": 2.015247344970703, "step": 85400 }, { "epoch": 0.3399591213578473, "grad_norm": 0.485538512468338, "learning_rate": 1.6601446565401783e-05, "loss": 1.9910733032226562, "step": 85450 }, { "epoch": 0.34015804419070733, "grad_norm": 0.5216002464294434, "learning_rate": 1.65994971235787e-05, "loss": 2.0834786987304685, "step": 85500 }, { "epoch": 0.3403569670235674, "grad_norm": 0.6241446137428284, "learning_rate": 1.6597507897228612e-05, "loss": 2.046925354003906, "step": 85550 }, { "epoch": 0.34055588985642743, "grad_norm": 0.5795531272888184, "learning_rate": 1.6595518670878524e-05, "loss": 2.078582763671875, "step": 85600 }, { "epoch": 0.34075481268928753, "grad_norm": 0.5774307250976562, "learning_rate": 1.6593529444528436e-05, "loss": 2.070340881347656, "step": 85650 }, { "epoch": 0.3409537355221476, "grad_norm": 0.6048152446746826, "learning_rate": 1.6591540218178348e-05, "loss": 2.037833251953125, "step": 85700 }, { "epoch": 0.34115265835500763, "grad_norm": 0.5610008835792542, "learning_rate": 1.658955099182826e-05, "loss": 2.0351580810546874, "step": 85750 }, { "epoch": 0.3413515811878677, "grad_norm": 0.7111213803291321, "learning_rate": 1.658756176547817e-05, "loss": 2.081864776611328, "step": 85800 }, { "epoch": 0.3415505040207278, "grad_norm": 0.5972548723220825, "learning_rate": 1.6585572539128085e-05, "loss": 2.0653733825683593, "step": 85850 }, { "epoch": 0.34174942685358783, "grad_norm": 0.5077217817306519, "learning_rate": 1.6583583312777997e-05, "loss": 1.9879757690429687, "step": 85900 }, { "epoch": 0.3419483496864479, "grad_norm": 0.5604113936424255, "learning_rate": 1.658159408642791e-05, "loss": 2.018878631591797, "step": 85950 }, { "epoch": 0.34214727251930793, "grad_norm": 0.657521665096283, "learning_rate": 1.657960486007782e-05, "loss": 2.027740478515625, "step": 86000 }, { "epoch": 0.34234619535216804, "grad_norm": 0.6551072001457214, "learning_rate": 1.6577615633727733e-05, "loss": 2.0962744140625, "step": 86050 }, { "epoch": 0.3425451181850281, "grad_norm": 0.4901462495326996, "learning_rate": 1.6575626407377642e-05, "loss": 2.023440399169922, "step": 86100 }, { "epoch": 0.34274404101788813, "grad_norm": 0.6464279294013977, "learning_rate": 1.6573637181027554e-05, "loss": 1.9882720947265624, "step": 86150 }, { "epoch": 0.3429429638507482, "grad_norm": 0.5751024484634399, "learning_rate": 1.6571647954677466e-05, "loss": 2.0661993408203125, "step": 86200 }, { "epoch": 0.3431418866836083, "grad_norm": 0.595035195350647, "learning_rate": 1.6569658728327382e-05, "loss": 2.0637969970703125, "step": 86250 }, { "epoch": 0.34334080951646834, "grad_norm": 0.6666437387466431, "learning_rate": 1.6567669501977294e-05, "loss": 2.014035339355469, "step": 86300 }, { "epoch": 0.3435397323493284, "grad_norm": 0.5311003923416138, "learning_rate": 1.6565680275627206e-05, "loss": 2.0579071044921875, "step": 86350 }, { "epoch": 0.34373865518218844, "grad_norm": 0.6159586906433105, "learning_rate": 1.6563691049277118e-05, "loss": 2.017495422363281, "step": 86400 }, { "epoch": 0.34393757801504854, "grad_norm": 0.5927170515060425, "learning_rate": 1.6561701822927027e-05, "loss": 2.026571960449219, "step": 86450 }, { "epoch": 0.3441365008479086, "grad_norm": 0.5318716168403625, "learning_rate": 1.655971259657694e-05, "loss": 2.018191070556641, "step": 86500 }, { "epoch": 0.34433542368076864, "grad_norm": 0.5866738557815552, "learning_rate": 1.655772337022685e-05, "loss": 2.049113922119141, "step": 86550 }, { "epoch": 0.3445343465136287, "grad_norm": 0.6903706192970276, "learning_rate": 1.6555734143876763e-05, "loss": 2.048307189941406, "step": 86600 }, { "epoch": 0.3447332693464888, "grad_norm": 0.6134349703788757, "learning_rate": 1.655374491752668e-05, "loss": 2.0306317138671877, "step": 86650 }, { "epoch": 0.34493219217934884, "grad_norm": 0.5205749273300171, "learning_rate": 1.655175569117659e-05, "loss": 2.0542601013183592, "step": 86700 }, { "epoch": 0.3451311150122089, "grad_norm": 0.49868643283843994, "learning_rate": 1.65497664648265e-05, "loss": 2.018701934814453, "step": 86750 }, { "epoch": 0.34533003784506894, "grad_norm": 0.6595944762229919, "learning_rate": 1.6547777238476412e-05, "loss": 2.0775567626953126, "step": 86800 }, { "epoch": 0.345528960677929, "grad_norm": 0.651270866394043, "learning_rate": 1.6545788012126324e-05, "loss": 2.0719171142578126, "step": 86850 }, { "epoch": 0.3457278835107891, "grad_norm": 0.6606582403182983, "learning_rate": 1.6543798785776236e-05, "loss": 2.0806884765625, "step": 86900 }, { "epoch": 0.34592680634364914, "grad_norm": 0.5387622714042664, "learning_rate": 1.654180955942615e-05, "loss": 2.0214588928222654, "step": 86950 }, { "epoch": 0.3461257291765092, "grad_norm": 0.6115306615829468, "learning_rate": 1.653982033307606e-05, "loss": 2.052373504638672, "step": 87000 }, { "epoch": 0.34632465200936924, "grad_norm": 0.5841149687767029, "learning_rate": 1.6537831106725976e-05, "loss": 1.9972529602050781, "step": 87050 }, { "epoch": 0.34652357484222934, "grad_norm": 0.5533328652381897, "learning_rate": 1.6535841880375885e-05, "loss": 2.0501580810546876, "step": 87100 }, { "epoch": 0.3467224976750894, "grad_norm": 0.5597710609436035, "learning_rate": 1.6533852654025797e-05, "loss": 1.9869889831542968, "step": 87150 }, { "epoch": 0.34692142050794944, "grad_norm": 0.49967432022094727, "learning_rate": 1.653186342767571e-05, "loss": 2.068500671386719, "step": 87200 }, { "epoch": 0.3471203433408095, "grad_norm": 0.4952866733074188, "learning_rate": 1.652987420132562e-05, "loss": 2.0263385009765624, "step": 87250 }, { "epoch": 0.3473192661736696, "grad_norm": 0.592666745185852, "learning_rate": 1.6527884974975533e-05, "loss": 1.9824005126953126, "step": 87300 }, { "epoch": 0.34751818900652964, "grad_norm": 0.5972239971160889, "learning_rate": 1.6525895748625446e-05, "loss": 2.079687194824219, "step": 87350 }, { "epoch": 0.3477171118393897, "grad_norm": 0.5505673289299011, "learning_rate": 1.6523906522275358e-05, "loss": 2.0676556396484376, "step": 87400 }, { "epoch": 0.34791603467224974, "grad_norm": 0.6170937418937683, "learning_rate": 1.652191729592527e-05, "loss": 2.064981231689453, "step": 87450 }, { "epoch": 0.34811495750510985, "grad_norm": 0.5567100644111633, "learning_rate": 1.6519928069575182e-05, "loss": 2.029221649169922, "step": 87500 }, { "epoch": 0.3483138803379699, "grad_norm": 0.523249089717865, "learning_rate": 1.6517938843225094e-05, "loss": 2.03578857421875, "step": 87550 }, { "epoch": 0.34851280317082994, "grad_norm": 0.5031667947769165, "learning_rate": 1.6515949616875006e-05, "loss": 2.043878936767578, "step": 87600 }, { "epoch": 0.34871172600369, "grad_norm": 0.5671429634094238, "learning_rate": 1.651396039052492e-05, "loss": 2.0083831787109374, "step": 87650 }, { "epoch": 0.3489106488365501, "grad_norm": 0.5218262672424316, "learning_rate": 1.651197116417483e-05, "loss": 2.032064971923828, "step": 87700 }, { "epoch": 0.34910957166941015, "grad_norm": 0.5464679002761841, "learning_rate": 1.6509981937824743e-05, "loss": 2.002696990966797, "step": 87750 }, { "epoch": 0.3493084945022702, "grad_norm": 0.6228607296943665, "learning_rate": 1.6507992711474655e-05, "loss": 2.021872863769531, "step": 87800 }, { "epoch": 0.34950741733513024, "grad_norm": 0.49907511472702026, "learning_rate": 1.6506003485124567e-05, "loss": 2.0254977416992186, "step": 87850 }, { "epoch": 0.34970634016799035, "grad_norm": 0.6326223015785217, "learning_rate": 1.650401425877448e-05, "loss": 2.0176123046875, "step": 87900 }, { "epoch": 0.3499052630008504, "grad_norm": 0.5580804944038391, "learning_rate": 1.650202503242439e-05, "loss": 2.0930802917480467, "step": 87950 }, { "epoch": 0.35010418583371045, "grad_norm": 0.5839262008666992, "learning_rate": 1.6500035806074303e-05, "loss": 2.038557434082031, "step": 88000 }, { "epoch": 0.3503031086665705, "grad_norm": 0.6017661094665527, "learning_rate": 1.6498046579724216e-05, "loss": 2.048898468017578, "step": 88050 }, { "epoch": 0.3505020314994306, "grad_norm": 0.6448938250541687, "learning_rate": 1.6496057353374128e-05, "loss": 2.06339599609375, "step": 88100 }, { "epoch": 0.35070095433229065, "grad_norm": 0.41249579191207886, "learning_rate": 1.649406812702404e-05, "loss": 2.027160186767578, "step": 88150 }, { "epoch": 0.3508998771651507, "grad_norm": 0.5861328840255737, "learning_rate": 1.6492078900673952e-05, "loss": 2.054144287109375, "step": 88200 }, { "epoch": 0.35109879999801075, "grad_norm": 0.5574255585670471, "learning_rate": 1.6490089674323864e-05, "loss": 2.0502474975585936, "step": 88250 }, { "epoch": 0.35129772283087085, "grad_norm": 0.4589688777923584, "learning_rate": 1.6488100447973776e-05, "loss": 2.0331103515625, "step": 88300 }, { "epoch": 0.3514966456637309, "grad_norm": 0.5335825681686401, "learning_rate": 1.648611122162369e-05, "loss": 2.0574778747558593, "step": 88350 }, { "epoch": 0.35169556849659095, "grad_norm": 0.5679098963737488, "learning_rate": 1.6484121995273597e-05, "loss": 2.0145184326171877, "step": 88400 }, { "epoch": 0.351894491329451, "grad_norm": 0.49786993861198425, "learning_rate": 1.648213276892351e-05, "loss": 2.1023289489746095, "step": 88450 }, { "epoch": 0.3520934141623111, "grad_norm": 0.4839291572570801, "learning_rate": 1.6480143542573425e-05, "loss": 2.040740203857422, "step": 88500 }, { "epoch": 0.35229233699517115, "grad_norm": 0.704627513885498, "learning_rate": 1.6478154316223337e-05, "loss": 2.0808474731445314, "step": 88550 }, { "epoch": 0.3524912598280312, "grad_norm": 0.5084697008132935, "learning_rate": 1.647616508987325e-05, "loss": 2.0698481750488282, "step": 88600 }, { "epoch": 0.35269018266089125, "grad_norm": 0.49602386355400085, "learning_rate": 1.6474215648050163e-05, "loss": 2.0240345764160157, "step": 88650 }, { "epoch": 0.35288910549375135, "grad_norm": 0.5604652762413025, "learning_rate": 1.647222642170007e-05, "loss": 2.1087382507324217, "step": 88700 }, { "epoch": 0.3530880283266114, "grad_norm": 0.514789879322052, "learning_rate": 1.6470237195349984e-05, "loss": 2.043009490966797, "step": 88750 }, { "epoch": 0.35328695115947145, "grad_norm": 0.5129432082176208, "learning_rate": 1.64682479689999e-05, "loss": 2.0527072143554688, "step": 88800 }, { "epoch": 0.3534858739923315, "grad_norm": 0.6731042265892029, "learning_rate": 1.646625874264981e-05, "loss": 2.010889587402344, "step": 88850 }, { "epoch": 0.3536847968251916, "grad_norm": 0.408336341381073, "learning_rate": 1.6464269516299723e-05, "loss": 2.0765280151367187, "step": 88900 }, { "epoch": 0.35388371965805165, "grad_norm": 0.4186793267726898, "learning_rate": 1.6462280289949636e-05, "loss": 2.041236572265625, "step": 88950 }, { "epoch": 0.3540826424909117, "grad_norm": 0.7243832349777222, "learning_rate": 1.6460291063599548e-05, "loss": 1.9787730407714843, "step": 89000 }, { "epoch": 0.35428156532377175, "grad_norm": 0.6375615000724792, "learning_rate": 1.6458301837249457e-05, "loss": 2.0136587524414065, "step": 89050 }, { "epoch": 0.35448048815663186, "grad_norm": 0.392537921667099, "learning_rate": 1.645631261089937e-05, "loss": 2.0547280883789063, "step": 89100 }, { "epoch": 0.3546794109894919, "grad_norm": 0.5666753649711609, "learning_rate": 1.645432338454928e-05, "loss": 2.0176979064941407, "step": 89150 }, { "epoch": 0.35487833382235195, "grad_norm": 0.5196659564971924, "learning_rate": 1.6452334158199193e-05, "loss": 2.057511291503906, "step": 89200 }, { "epoch": 0.355077256655212, "grad_norm": 0.622198760509491, "learning_rate": 1.645034493184911e-05, "loss": 2.1092987060546875, "step": 89250 }, { "epoch": 0.3552761794880721, "grad_norm": 0.6178663372993469, "learning_rate": 1.644835570549902e-05, "loss": 2.01769775390625, "step": 89300 }, { "epoch": 0.35547510232093216, "grad_norm": 0.5660715103149414, "learning_rate": 1.644636647914893e-05, "loss": 2.046478271484375, "step": 89350 }, { "epoch": 0.3556740251537922, "grad_norm": 0.6112529039382935, "learning_rate": 1.644437725279884e-05, "loss": 1.9806584167480468, "step": 89400 }, { "epoch": 0.35587294798665225, "grad_norm": 0.6199285984039307, "learning_rate": 1.6442388026448754e-05, "loss": 2.054867706298828, "step": 89450 }, { "epoch": 0.35607187081951236, "grad_norm": 0.5512824654579163, "learning_rate": 1.6440398800098666e-05, "loss": 2.0666786193847657, "step": 89500 }, { "epoch": 0.3562707936523724, "grad_norm": 0.5674744844436646, "learning_rate": 1.6438409573748578e-05, "loss": 2.0677272033691407, "step": 89550 }, { "epoch": 0.35646971648523246, "grad_norm": 0.543236255645752, "learning_rate": 1.643642034739849e-05, "loss": 2.060319671630859, "step": 89600 }, { "epoch": 0.3566686393180925, "grad_norm": 0.5590550303459167, "learning_rate": 1.6434431121048406e-05, "loss": 2.0467318725585937, "step": 89650 }, { "epoch": 0.3568675621509526, "grad_norm": 0.5616602301597595, "learning_rate": 1.6432441894698314e-05, "loss": 2.0335301208496093, "step": 89700 }, { "epoch": 0.35706648498381266, "grad_norm": 0.6428877711296082, "learning_rate": 1.6430452668348227e-05, "loss": 2.0398866271972658, "step": 89750 }, { "epoch": 0.3572654078166727, "grad_norm": 0.574178159236908, "learning_rate": 1.642846344199814e-05, "loss": 2.0455221557617187, "step": 89800 }, { "epoch": 0.35746433064953276, "grad_norm": 0.5465304851531982, "learning_rate": 1.642647421564805e-05, "loss": 1.996597900390625, "step": 89850 }, { "epoch": 0.35766325348239286, "grad_norm": 0.6116454601287842, "learning_rate": 1.6424484989297963e-05, "loss": 2.078422393798828, "step": 89900 }, { "epoch": 0.3578621763152529, "grad_norm": 0.5789434909820557, "learning_rate": 1.6422495762947875e-05, "loss": 2.032861633300781, "step": 89950 }, { "epoch": 0.35806109914811296, "grad_norm": 0.5201782584190369, "learning_rate": 1.6420506536597787e-05, "loss": 2.076071929931641, "step": 90000 }, { "epoch": 0.358260021980973, "grad_norm": 0.46702471375465393, "learning_rate": 1.64185173102477e-05, "loss": 2.000425109863281, "step": 90050 }, { "epoch": 0.3584589448138331, "grad_norm": 0.4956376552581787, "learning_rate": 1.641652808389761e-05, "loss": 2.0726304626464844, "step": 90100 }, { "epoch": 0.35865786764669316, "grad_norm": 0.6072651147842407, "learning_rate": 1.6414538857547524e-05, "loss": 2.027364044189453, "step": 90150 }, { "epoch": 0.3588567904795532, "grad_norm": 0.5871526002883911, "learning_rate": 1.6412549631197436e-05, "loss": 2.0584201049804687, "step": 90200 }, { "epoch": 0.35905571331241326, "grad_norm": 0.5077921152114868, "learning_rate": 1.6410560404847348e-05, "loss": 1.9916705322265624, "step": 90250 }, { "epoch": 0.35925463614527336, "grad_norm": 0.5786094665527344, "learning_rate": 1.640857117849726e-05, "loss": 2.0770745849609376, "step": 90300 }, { "epoch": 0.3594535589781334, "grad_norm": 0.6439311504364014, "learning_rate": 1.6406581952147172e-05, "loss": 2.0552996826171874, "step": 90350 }, { "epoch": 0.35965248181099346, "grad_norm": 0.5602532029151917, "learning_rate": 1.6404592725797084e-05, "loss": 2.053685302734375, "step": 90400 }, { "epoch": 0.3598514046438535, "grad_norm": 0.5744288563728333, "learning_rate": 1.6402603499446997e-05, "loss": 2.075880279541016, "step": 90450 }, { "epoch": 0.3600503274767136, "grad_norm": 0.5746824741363525, "learning_rate": 1.640061427309691e-05, "loss": 2.0444187927246094, "step": 90500 }, { "epoch": 0.36024925030957367, "grad_norm": 0.5959745645523071, "learning_rate": 1.639862504674682e-05, "loss": 1.9816581726074218, "step": 90550 }, { "epoch": 0.3604481731424337, "grad_norm": 0.5465052723884583, "learning_rate": 1.6396635820396733e-05, "loss": 2.014980926513672, "step": 90600 }, { "epoch": 0.36064709597529376, "grad_norm": 0.5597826838493347, "learning_rate": 1.6394646594046645e-05, "loss": 2.0453611755371095, "step": 90650 }, { "epoch": 0.36084601880815387, "grad_norm": 0.519737720489502, "learning_rate": 1.6392657367696557e-05, "loss": 2.0505979919433592, "step": 90700 }, { "epoch": 0.3610449416410139, "grad_norm": 0.49840137362480164, "learning_rate": 1.639066814134647e-05, "loss": 2.0391972351074217, "step": 90750 }, { "epoch": 0.36124386447387397, "grad_norm": 0.5241156220436096, "learning_rate": 1.6388718699523383e-05, "loss": 2.0663153076171876, "step": 90800 }, { "epoch": 0.361442787306734, "grad_norm": 0.4717576503753662, "learning_rate": 1.6386729473173295e-05, "loss": 2.0299105834960938, "step": 90850 }, { "epoch": 0.3616417101395941, "grad_norm": 0.5638239979743958, "learning_rate": 1.6384740246823207e-05, "loss": 2.0402313232421876, "step": 90900 }, { "epoch": 0.36184063297245417, "grad_norm": 0.5891692042350769, "learning_rate": 1.638275102047312e-05, "loss": 2.102374725341797, "step": 90950 }, { "epoch": 0.3620395558053142, "grad_norm": 0.5086619257926941, "learning_rate": 1.638076179412303e-05, "loss": 2.013863067626953, "step": 91000 }, { "epoch": 0.36223847863817427, "grad_norm": 0.5221182107925415, "learning_rate": 1.6378772567772944e-05, "loss": 2.025726318359375, "step": 91050 }, { "epoch": 0.36243740147103437, "grad_norm": 0.5985209941864014, "learning_rate": 1.6376783341422856e-05, "loss": 2.0395333862304685, "step": 91100 }, { "epoch": 0.3626363243038944, "grad_norm": 0.5011125206947327, "learning_rate": 1.6374794115072768e-05, "loss": 2.0665283203125, "step": 91150 }, { "epoch": 0.36283524713675447, "grad_norm": 0.5060053467750549, "learning_rate": 1.637280488872268e-05, "loss": 1.977097625732422, "step": 91200 }, { "epoch": 0.3630341699696145, "grad_norm": 0.5449846982955933, "learning_rate": 1.6370815662372592e-05, "loss": 2.0370004272460935, "step": 91250 }, { "epoch": 0.3632330928024746, "grad_norm": 0.5757023096084595, "learning_rate": 1.63688264360225e-05, "loss": 2.020285797119141, "step": 91300 }, { "epoch": 0.36343201563533467, "grad_norm": 0.5680784583091736, "learning_rate": 1.6366837209672413e-05, "loss": 1.990818328857422, "step": 91350 }, { "epoch": 0.3636309384681947, "grad_norm": 0.5938807725906372, "learning_rate": 1.636484798332233e-05, "loss": 2.0483995056152344, "step": 91400 }, { "epoch": 0.36382986130105477, "grad_norm": 0.5378807783126831, "learning_rate": 1.636285875697224e-05, "loss": 2.0532595825195314, "step": 91450 }, { "epoch": 0.3640287841339149, "grad_norm": 0.5866096615791321, "learning_rate": 1.6360869530622153e-05, "loss": 2.124101257324219, "step": 91500 }, { "epoch": 0.3642277069667749, "grad_norm": 0.5866751074790955, "learning_rate": 1.6358880304272065e-05, "loss": 1.99613525390625, "step": 91550 }, { "epoch": 0.36442662979963497, "grad_norm": 0.545853316783905, "learning_rate": 1.6356891077921977e-05, "loss": 2.0406849670410154, "step": 91600 }, { "epoch": 0.364625552632495, "grad_norm": 0.5492649078369141, "learning_rate": 1.6354901851571886e-05, "loss": 2.053814697265625, "step": 91650 }, { "epoch": 0.3648244754653551, "grad_norm": 0.8224371075630188, "learning_rate": 1.6352912625221798e-05, "loss": 2.01776611328125, "step": 91700 }, { "epoch": 0.3650233982982152, "grad_norm": 0.5750924348831177, "learning_rate": 1.635092339887171e-05, "loss": 2.0016851806640625, "step": 91750 }, { "epoch": 0.3652223211310752, "grad_norm": 0.6016367673873901, "learning_rate": 1.6348934172521626e-05, "loss": 2.0257583618164063, "step": 91800 }, { "epoch": 0.36542124396393527, "grad_norm": 0.600546658039093, "learning_rate": 1.6346944946171538e-05, "loss": 2.031649932861328, "step": 91850 }, { "epoch": 0.3656201667967954, "grad_norm": 0.4927789270877838, "learning_rate": 1.634495571982145e-05, "loss": 2.0217027282714843, "step": 91900 }, { "epoch": 0.3658190896296554, "grad_norm": 0.6131353974342346, "learning_rate": 1.634296649347136e-05, "loss": 1.9860787963867188, "step": 91950 }, { "epoch": 0.3660180124625155, "grad_norm": 0.5858716368675232, "learning_rate": 1.634097726712127e-05, "loss": 2.0384199523925783, "step": 92000 }, { "epoch": 0.3662169352953755, "grad_norm": 0.5541172623634338, "learning_rate": 1.6338988040771183e-05, "loss": 2.039354248046875, "step": 92050 }, { "epoch": 0.3664158581282356, "grad_norm": 0.46829909086227417, "learning_rate": 1.6336998814421095e-05, "loss": 1.9822996520996095, "step": 92100 }, { "epoch": 0.3666147809610957, "grad_norm": 0.5355618000030518, "learning_rate": 1.6335009588071008e-05, "loss": 2.019418029785156, "step": 92150 }, { "epoch": 0.3668137037939557, "grad_norm": 0.58768230676651, "learning_rate": 1.6333020361720923e-05, "loss": 2.007150421142578, "step": 92200 }, { "epoch": 0.3670126266268158, "grad_norm": 0.5514873266220093, "learning_rate": 1.6331031135370835e-05, "loss": 2.0405545043945312, "step": 92250 }, { "epoch": 0.3672115494596759, "grad_norm": 0.4621821641921997, "learning_rate": 1.6329041909020744e-05, "loss": 2.0112208557128906, "step": 92300 }, { "epoch": 0.36741047229253593, "grad_norm": 0.6601710915565491, "learning_rate": 1.6327052682670656e-05, "loss": 2.0124725341796874, "step": 92350 }, { "epoch": 0.367609395125396, "grad_norm": 0.5435765385627747, "learning_rate": 1.6325063456320568e-05, "loss": 2.0915988159179686, "step": 92400 }, { "epoch": 0.367808317958256, "grad_norm": 0.6830514669418335, "learning_rate": 1.632307422997048e-05, "loss": 2.058980712890625, "step": 92450 }, { "epoch": 0.36800724079111613, "grad_norm": 0.5820109248161316, "learning_rate": 1.6321085003620393e-05, "loss": 2.129729919433594, "step": 92500 }, { "epoch": 0.3682061636239762, "grad_norm": 0.564902126789093, "learning_rate": 1.6319095777270305e-05, "loss": 2.0737431335449217, "step": 92550 }, { "epoch": 0.36840508645683623, "grad_norm": 0.5398443341255188, "learning_rate": 1.6317106550920217e-05, "loss": 1.96593017578125, "step": 92600 }, { "epoch": 0.3686040092896963, "grad_norm": 0.6089395880699158, "learning_rate": 1.631511732457013e-05, "loss": 2.0466458129882814, "step": 92650 }, { "epoch": 0.3688029321225564, "grad_norm": 0.503431499004364, "learning_rate": 1.631312809822004e-05, "loss": 2.020011749267578, "step": 92700 }, { "epoch": 0.36900185495541643, "grad_norm": 0.6021276712417603, "learning_rate": 1.6311138871869953e-05, "loss": 2.027735595703125, "step": 92750 }, { "epoch": 0.3692007777882765, "grad_norm": 0.5959991812705994, "learning_rate": 1.6309149645519865e-05, "loss": 2.0274835205078126, "step": 92800 }, { "epoch": 0.36939970062113653, "grad_norm": 0.5487960577011108, "learning_rate": 1.6307160419169778e-05, "loss": 2.021279296875, "step": 92850 }, { "epoch": 0.36959862345399663, "grad_norm": 0.6369873881340027, "learning_rate": 1.630517119281969e-05, "loss": 2.010298614501953, "step": 92900 }, { "epoch": 0.3697975462868567, "grad_norm": 0.5231918096542358, "learning_rate": 1.6303181966469602e-05, "loss": 1.9835612487792968, "step": 92950 }, { "epoch": 0.36999646911971673, "grad_norm": 0.4540453851222992, "learning_rate": 1.6301192740119514e-05, "loss": 2.0257489013671877, "step": 93000 }, { "epoch": 0.3701953919525768, "grad_norm": 0.6153197884559631, "learning_rate": 1.6299203513769426e-05, "loss": 2.012801513671875, "step": 93050 }, { "epoch": 0.3703943147854369, "grad_norm": 0.47422701120376587, "learning_rate": 1.629721428741934e-05, "loss": 2.0352964782714844, "step": 93100 }, { "epoch": 0.37059323761829693, "grad_norm": 0.7047008872032166, "learning_rate": 1.629522506106925e-05, "loss": 2.0013504028320312, "step": 93150 }, { "epoch": 0.370792160451157, "grad_norm": 0.4803241491317749, "learning_rate": 1.6293235834719163e-05, "loss": 1.9942233276367187, "step": 93200 }, { "epoch": 0.37099108328401703, "grad_norm": 0.498456209897995, "learning_rate": 1.6291246608369075e-05, "loss": 2.05454833984375, "step": 93250 }, { "epoch": 0.37119000611687714, "grad_norm": 0.6074188947677612, "learning_rate": 1.6289257382018987e-05, "loss": 2.046153106689453, "step": 93300 }, { "epoch": 0.3713889289497372, "grad_norm": 0.6431484818458557, "learning_rate": 1.62872681556689e-05, "loss": 2.055696563720703, "step": 93350 }, { "epoch": 0.37158785178259723, "grad_norm": 0.7211117148399353, "learning_rate": 1.628527892931881e-05, "loss": 2.016543426513672, "step": 93400 }, { "epoch": 0.3717867746154573, "grad_norm": 0.5919255614280701, "learning_rate": 1.6283289702968723e-05, "loss": 1.9924659729003906, "step": 93450 }, { "epoch": 0.3719856974483174, "grad_norm": 0.5718649625778198, "learning_rate": 1.6281300476618635e-05, "loss": 2.0118788146972655, "step": 93500 }, { "epoch": 0.37218462028117744, "grad_norm": 0.5320569276809692, "learning_rate": 1.6279311250268548e-05, "loss": 2.0714735412597656, "step": 93550 }, { "epoch": 0.3723835431140375, "grad_norm": 0.467348575592041, "learning_rate": 1.627732202391846e-05, "loss": 2.014864501953125, "step": 93600 }, { "epoch": 0.37258246594689753, "grad_norm": 0.6129562258720398, "learning_rate": 1.6275332797568372e-05, "loss": 2.0185063171386717, "step": 93650 }, { "epoch": 0.37278138877975764, "grad_norm": 0.5579917430877686, "learning_rate": 1.6273343571218284e-05, "loss": 2.012803497314453, "step": 93700 }, { "epoch": 0.3729803116126177, "grad_norm": 0.5597490072250366, "learning_rate": 1.6271354344868196e-05, "loss": 2.0493235778808594, "step": 93750 }, { "epoch": 0.37317923444547774, "grad_norm": 0.5177932381629944, "learning_rate": 1.626936511851811e-05, "loss": 2.041300506591797, "step": 93800 }, { "epoch": 0.3733781572783378, "grad_norm": 0.6152750849723816, "learning_rate": 1.626737589216802e-05, "loss": 1.9773495483398438, "step": 93850 }, { "epoch": 0.3735770801111979, "grad_norm": 0.6811844706535339, "learning_rate": 1.626538666581793e-05, "loss": 2.0207893371582033, "step": 93900 }, { "epoch": 0.37377600294405794, "grad_norm": 0.6193950176239014, "learning_rate": 1.626339743946784e-05, "loss": 2.0670594787597656, "step": 93950 }, { "epoch": 0.373974925776918, "grad_norm": 0.6753005385398865, "learning_rate": 1.6261408213117757e-05, "loss": 1.9861509704589844, "step": 94000 }, { "epoch": 0.37417384860977804, "grad_norm": 0.5153489708900452, "learning_rate": 1.625941898676767e-05, "loss": 1.9928720092773438, "step": 94050 }, { "epoch": 0.37437277144263814, "grad_norm": 0.5952100157737732, "learning_rate": 1.625742976041758e-05, "loss": 2.038176727294922, "step": 94100 }, { "epoch": 0.3745716942754982, "grad_norm": 0.7013928294181824, "learning_rate": 1.6255480318594495e-05, "loss": 2.045706329345703, "step": 94150 }, { "epoch": 0.37477061710835824, "grad_norm": 0.4725378453731537, "learning_rate": 1.6253491092244407e-05, "loss": 2.063975982666016, "step": 94200 }, { "epoch": 0.3749695399412183, "grad_norm": 0.4905008375644684, "learning_rate": 1.6251501865894316e-05, "loss": 2.0387307739257814, "step": 94250 }, { "epoch": 0.3751684627740784, "grad_norm": 0.5226490497589111, "learning_rate": 1.6249512639544228e-05, "loss": 2.0154833984375, "step": 94300 }, { "epoch": 0.37536738560693844, "grad_norm": 0.49211615324020386, "learning_rate": 1.6247523413194143e-05, "loss": 2.085347900390625, "step": 94350 }, { "epoch": 0.3755663084397985, "grad_norm": 0.5117087364196777, "learning_rate": 1.6245534186844055e-05, "loss": 2.0617750549316405, "step": 94400 }, { "epoch": 0.37576523127265854, "grad_norm": 0.6000460982322693, "learning_rate": 1.6243544960493968e-05, "loss": 2.087511444091797, "step": 94450 }, { "epoch": 0.37596415410551864, "grad_norm": 0.5683834552764893, "learning_rate": 1.624155573414388e-05, "loss": 1.9879083251953125, "step": 94500 }, { "epoch": 0.3761630769383787, "grad_norm": 0.657644510269165, "learning_rate": 1.623956650779379e-05, "loss": 2.021589508056641, "step": 94550 }, { "epoch": 0.37636199977123874, "grad_norm": 0.5276411771774292, "learning_rate": 1.62375772814437e-05, "loss": 2.008267059326172, "step": 94600 }, { "epoch": 0.3765609226040988, "grad_norm": 0.6284490823745728, "learning_rate": 1.6235588055093613e-05, "loss": 2.039010009765625, "step": 94650 }, { "epoch": 0.37675984543695884, "grad_norm": 0.6193966269493103, "learning_rate": 1.6233598828743525e-05, "loss": 2.0253289794921874, "step": 94700 }, { "epoch": 0.37695876826981894, "grad_norm": 0.6467611193656921, "learning_rate": 1.623160960239344e-05, "loss": 2.04052734375, "step": 94750 }, { "epoch": 0.377157691102679, "grad_norm": 0.6562389731407166, "learning_rate": 1.6229620376043353e-05, "loss": 1.99888427734375, "step": 94800 }, { "epoch": 0.37735661393553904, "grad_norm": 0.5323068499565125, "learning_rate": 1.6227631149693265e-05, "loss": 2.0726695251464844, "step": 94850 }, { "epoch": 0.3775555367683991, "grad_norm": 0.5451204180717468, "learning_rate": 1.6225641923343174e-05, "loss": 2.023376007080078, "step": 94900 }, { "epoch": 0.3777544596012592, "grad_norm": 0.5707686543464661, "learning_rate": 1.6223652696993086e-05, "loss": 2.0456912231445314, "step": 94950 }, { "epoch": 0.37795338243411924, "grad_norm": 0.49786269664764404, "learning_rate": 1.6221663470642998e-05, "loss": 1.9971328735351563, "step": 95000 }, { "epoch": 0.3781523052669793, "grad_norm": 0.4906308352947235, "learning_rate": 1.621967424429291e-05, "loss": 2.0503999328613283, "step": 95050 }, { "epoch": 0.37835122809983934, "grad_norm": 0.5777581334114075, "learning_rate": 1.6217685017942822e-05, "loss": 2.022094421386719, "step": 95100 }, { "epoch": 0.37855015093269945, "grad_norm": 0.5770463943481445, "learning_rate": 1.6215695791592738e-05, "loss": 2.0445071411132814, "step": 95150 }, { "epoch": 0.3787490737655595, "grad_norm": 0.5627402663230896, "learning_rate": 1.6213706565242646e-05, "loss": 1.9741511535644531, "step": 95200 }, { "epoch": 0.37894799659841955, "grad_norm": 0.5896719098091125, "learning_rate": 1.621171733889256e-05, "loss": 2.0611555480957033, "step": 95250 }, { "epoch": 0.3791469194312796, "grad_norm": 0.5207870006561279, "learning_rate": 1.620972811254247e-05, "loss": 2.0356065368652345, "step": 95300 }, { "epoch": 0.3793458422641397, "grad_norm": 0.6251391768455505, "learning_rate": 1.6207738886192383e-05, "loss": 2.013406982421875, "step": 95350 }, { "epoch": 0.37954476509699975, "grad_norm": 0.5338059663772583, "learning_rate": 1.6205749659842295e-05, "loss": 2.0329022216796875, "step": 95400 }, { "epoch": 0.3797436879298598, "grad_norm": 0.5845978856086731, "learning_rate": 1.6203760433492207e-05, "loss": 2.057140350341797, "step": 95450 }, { "epoch": 0.37994261076271985, "grad_norm": 0.620983362197876, "learning_rate": 1.620177120714212e-05, "loss": 2.0844874572753906, "step": 95500 }, { "epoch": 0.38014153359557995, "grad_norm": 0.5437675714492798, "learning_rate": 1.619978198079203e-05, "loss": 2.0370474243164063, "step": 95550 }, { "epoch": 0.38034045642844, "grad_norm": 0.5652778148651123, "learning_rate": 1.6197792754441944e-05, "loss": 2.054998016357422, "step": 95600 }, { "epoch": 0.38053937926130005, "grad_norm": 0.580124020576477, "learning_rate": 1.6195803528091856e-05, "loss": 2.0621502685546873, "step": 95650 }, { "epoch": 0.3807383020941601, "grad_norm": 0.504412829875946, "learning_rate": 1.6193814301741768e-05, "loss": 2.020530090332031, "step": 95700 }, { "epoch": 0.3809372249270202, "grad_norm": 0.7666497826576233, "learning_rate": 1.619182507539168e-05, "loss": 2.0895071411132813, "step": 95750 }, { "epoch": 0.38113614775988025, "grad_norm": 0.6155702471733093, "learning_rate": 1.6189835849041592e-05, "loss": 1.9968283081054687, "step": 95800 }, { "epoch": 0.3813350705927403, "grad_norm": 0.4856646955013275, "learning_rate": 1.6187846622691504e-05, "loss": 2.07984619140625, "step": 95850 }, { "epoch": 0.38153399342560035, "grad_norm": 0.5492447018623352, "learning_rate": 1.6185857396341416e-05, "loss": 2.0115911865234377, "step": 95900 }, { "epoch": 0.38173291625846045, "grad_norm": 0.6060464978218079, "learning_rate": 1.618386816999133e-05, "loss": 2.008347625732422, "step": 95950 }, { "epoch": 0.3819318390913205, "grad_norm": 0.5498272776603699, "learning_rate": 1.618187894364124e-05, "loss": 1.9925457763671874, "step": 96000 }, { "epoch": 0.38213076192418055, "grad_norm": 0.523328959941864, "learning_rate": 1.6179889717291153e-05, "loss": 2.0234220886230467, "step": 96050 }, { "epoch": 0.3823296847570406, "grad_norm": 0.576717734336853, "learning_rate": 1.6177900490941065e-05, "loss": 2.048206481933594, "step": 96100 }, { "epoch": 0.3825286075899007, "grad_norm": 0.6017113327980042, "learning_rate": 1.6175911264590974e-05, "loss": 1.9982655334472657, "step": 96150 }, { "epoch": 0.38272753042276075, "grad_norm": 0.674514889717102, "learning_rate": 1.617392203824089e-05, "loss": 2.0146022033691406, "step": 96200 }, { "epoch": 0.3829264532556208, "grad_norm": 0.6007245779037476, "learning_rate": 1.61719328118908e-05, "loss": 2.03072509765625, "step": 96250 }, { "epoch": 0.38312537608848085, "grad_norm": 0.5751179456710815, "learning_rate": 1.6169943585540714e-05, "loss": 2.0217796325683595, "step": 96300 }, { "epoch": 0.38332429892134096, "grad_norm": 0.6757460236549377, "learning_rate": 1.6167954359190626e-05, "loss": 1.9958641052246093, "step": 96350 }, { "epoch": 0.383523221754201, "grad_norm": 0.6148249506950378, "learning_rate": 1.6165965132840538e-05, "loss": 2.036348876953125, "step": 96400 }, { "epoch": 0.38372214458706105, "grad_norm": 0.5774668455123901, "learning_rate": 1.616397590649045e-05, "loss": 2.024725036621094, "step": 96450 }, { "epoch": 0.3839210674199211, "grad_norm": 0.5276380777359009, "learning_rate": 1.616198668014036e-05, "loss": 2.032492828369141, "step": 96500 }, { "epoch": 0.3841199902527812, "grad_norm": 0.6191763877868652, "learning_rate": 1.615999745379027e-05, "loss": 2.006408386230469, "step": 96550 }, { "epoch": 0.38431891308564126, "grad_norm": 0.5241677761077881, "learning_rate": 1.6158008227440187e-05, "loss": 2.063521423339844, "step": 96600 }, { "epoch": 0.3845178359185013, "grad_norm": 0.54474276304245, "learning_rate": 1.61560190010901e-05, "loss": 2.0550297546386718, "step": 96650 }, { "epoch": 0.38471675875136135, "grad_norm": 0.6114922165870667, "learning_rate": 1.615402977474001e-05, "loss": 2.007638702392578, "step": 96700 }, { "epoch": 0.38491568158422146, "grad_norm": 0.5680885910987854, "learning_rate": 1.6152040548389923e-05, "loss": 2.0214944458007813, "step": 96750 }, { "epoch": 0.3851146044170815, "grad_norm": 0.6024768352508545, "learning_rate": 1.6150051322039832e-05, "loss": 2.021098937988281, "step": 96800 }, { "epoch": 0.38531352724994156, "grad_norm": 0.5434513092041016, "learning_rate": 1.6148062095689744e-05, "loss": 2.0551486206054688, "step": 96850 }, { "epoch": 0.3855124500828016, "grad_norm": 0.4621385633945465, "learning_rate": 1.6146072869339656e-05, "loss": 2.0075062561035155, "step": 96900 }, { "epoch": 0.3857113729156617, "grad_norm": 0.5675125122070312, "learning_rate": 1.6144083642989568e-05, "loss": 2.018296813964844, "step": 96950 }, { "epoch": 0.38591029574852176, "grad_norm": 0.6109488010406494, "learning_rate": 1.6142094416639484e-05, "loss": 2.0291567993164064, "step": 97000 }, { "epoch": 0.3861092185813818, "grad_norm": 0.6616516709327698, "learning_rate": 1.6140105190289396e-05, "loss": 2.016317138671875, "step": 97050 }, { "epoch": 0.38630814141424186, "grad_norm": 0.4911077916622162, "learning_rate": 1.6138115963939308e-05, "loss": 1.9870948791503906, "step": 97100 }, { "epoch": 0.38650706424710196, "grad_norm": 0.6342312693595886, "learning_rate": 1.6136126737589217e-05, "loss": 2.0553329467773436, "step": 97150 }, { "epoch": 0.386705987079962, "grad_norm": 0.5512543320655823, "learning_rate": 1.613413751123913e-05, "loss": 2.029798583984375, "step": 97200 }, { "epoch": 0.38690490991282206, "grad_norm": 0.5551594495773315, "learning_rate": 1.613214828488904e-05, "loss": 2.0130210876464845, "step": 97250 }, { "epoch": 0.3871038327456821, "grad_norm": 0.5981594324111938, "learning_rate": 1.6130159058538953e-05, "loss": 2.0348724365234374, "step": 97300 }, { "epoch": 0.3873027555785422, "grad_norm": 0.5589044690132141, "learning_rate": 1.6128169832188865e-05, "loss": 2.070974578857422, "step": 97350 }, { "epoch": 0.38750167841140226, "grad_norm": 0.6076593399047852, "learning_rate": 1.612618060583878e-05, "loss": 1.983700714111328, "step": 97400 }, { "epoch": 0.3877006012442623, "grad_norm": 0.5445239543914795, "learning_rate": 1.612419137948869e-05, "loss": 2.0282418823242185, "step": 97450 }, { "epoch": 0.38789952407712236, "grad_norm": 0.6489928364753723, "learning_rate": 1.6122202153138602e-05, "loss": 2.0201652526855467, "step": 97500 }, { "epoch": 0.38809844690998246, "grad_norm": 0.5381724834442139, "learning_rate": 1.6120212926788514e-05, "loss": 2.0634828186035157, "step": 97550 }, { "epoch": 0.3882973697428425, "grad_norm": 0.6122270226478577, "learning_rate": 1.6118223700438426e-05, "loss": 2.017811584472656, "step": 97600 }, { "epoch": 0.38849629257570256, "grad_norm": 0.5649616122245789, "learning_rate": 1.6116234474088338e-05, "loss": 2.0393031311035155, "step": 97650 }, { "epoch": 0.3886952154085626, "grad_norm": 0.64494788646698, "learning_rate": 1.611424524773825e-05, "loss": 1.9894749450683593, "step": 97700 }, { "epoch": 0.3888941382414227, "grad_norm": 0.6027268171310425, "learning_rate": 1.6112256021388162e-05, "loss": 2.0118592834472655, "step": 97750 }, { "epoch": 0.38909306107428276, "grad_norm": 0.6055883765220642, "learning_rate": 1.6110266795038075e-05, "loss": 2.013331298828125, "step": 97800 }, { "epoch": 0.3892919839071428, "grad_norm": 0.569087564945221, "learning_rate": 1.6108277568687987e-05, "loss": 1.9774151611328126, "step": 97850 }, { "epoch": 0.38949090674000286, "grad_norm": 0.5287646055221558, "learning_rate": 1.61062883423379e-05, "loss": 2.0453280639648437, "step": 97900 }, { "epoch": 0.38968982957286297, "grad_norm": 0.6100111603736877, "learning_rate": 1.610429911598781e-05, "loss": 2.0135382080078124, "step": 97950 }, { "epoch": 0.389888752405723, "grad_norm": 0.3390906751155853, "learning_rate": 1.6102309889637723e-05, "loss": 2.062701873779297, "step": 98000 }, { "epoch": 0.39008767523858306, "grad_norm": 0.4811865985393524, "learning_rate": 1.6100320663287635e-05, "loss": 2.0274855041503907, "step": 98050 }, { "epoch": 0.3902865980714431, "grad_norm": 0.5394128561019897, "learning_rate": 1.6098331436937548e-05, "loss": 2.0500416564941406, "step": 98100 }, { "epoch": 0.3904855209043032, "grad_norm": 0.6020705699920654, "learning_rate": 1.609638199511446e-05, "loss": 2.047488098144531, "step": 98150 }, { "epoch": 0.39068444373716327, "grad_norm": 0.5997905135154724, "learning_rate": 1.6094392768764373e-05, "loss": 2.0467294311523436, "step": 98200 }, { "epoch": 0.3908833665700233, "grad_norm": 0.6065864562988281, "learning_rate": 1.6092403542414285e-05, "loss": 2.039320831298828, "step": 98250 }, { "epoch": 0.39108228940288337, "grad_norm": 0.44010478258132935, "learning_rate": 1.6090414316064197e-05, "loss": 2.037114715576172, "step": 98300 }, { "epoch": 0.39128121223574347, "grad_norm": 0.5219911932945251, "learning_rate": 1.608842508971411e-05, "loss": 1.9747789001464844, "step": 98350 }, { "epoch": 0.3914801350686035, "grad_norm": 0.6047014594078064, "learning_rate": 1.6086435863364022e-05, "loss": 2.07247314453125, "step": 98400 }, { "epoch": 0.39167905790146357, "grad_norm": 0.5546205639839172, "learning_rate": 1.6084446637013934e-05, "loss": 2.039309234619141, "step": 98450 }, { "epoch": 0.3918779807343236, "grad_norm": 0.7586532831192017, "learning_rate": 1.6082457410663846e-05, "loss": 2.0128773498535155, "step": 98500 }, { "epoch": 0.3920769035671837, "grad_norm": 0.5513031482696533, "learning_rate": 1.608050796884076e-05, "loss": 2.033181457519531, "step": 98550 }, { "epoch": 0.39227582640004377, "grad_norm": 0.6600023508071899, "learning_rate": 1.607851874249067e-05, "loss": 2.0537559509277346, "step": 98600 }, { "epoch": 0.3924747492329038, "grad_norm": 0.45225271582603455, "learning_rate": 1.6076529516140584e-05, "loss": 1.9962574768066406, "step": 98650 }, { "epoch": 0.39267367206576387, "grad_norm": 0.5257669687271118, "learning_rate": 1.6074540289790496e-05, "loss": 2.0121795654296877, "step": 98700 }, { "epoch": 0.392872594898624, "grad_norm": 0.5748347043991089, "learning_rate": 1.6072551063440408e-05, "loss": 2.0126141357421874, "step": 98750 }, { "epoch": 0.393071517731484, "grad_norm": 0.5765624642372131, "learning_rate": 1.607056183709032e-05, "loss": 2.052080535888672, "step": 98800 }, { "epoch": 0.39327044056434407, "grad_norm": 0.5947469472885132, "learning_rate": 1.6068572610740232e-05, "loss": 2.0392755126953124, "step": 98850 }, { "epoch": 0.3934693633972041, "grad_norm": 0.46327608823776245, "learning_rate": 1.6066583384390145e-05, "loss": 2.1295545959472655, "step": 98900 }, { "epoch": 0.3936682862300642, "grad_norm": 0.5549982190132141, "learning_rate": 1.6064594158040057e-05, "loss": 2.0495957946777343, "step": 98950 }, { "epoch": 0.3938672090629243, "grad_norm": 0.4614090621471405, "learning_rate": 1.606260493168997e-05, "loss": 2.0434202575683593, "step": 99000 }, { "epoch": 0.3940661318957843, "grad_norm": 0.5898913145065308, "learning_rate": 1.606061570533988e-05, "loss": 2.007857666015625, "step": 99050 }, { "epoch": 0.39426505472864437, "grad_norm": 0.6859596371650696, "learning_rate": 1.6058626478989793e-05, "loss": 2.0428851318359373, "step": 99100 }, { "epoch": 0.3944639775615045, "grad_norm": 0.5929448008537292, "learning_rate": 1.6056637252639705e-05, "loss": 2.034345703125, "step": 99150 }, { "epoch": 0.3946629003943645, "grad_norm": 0.5206171870231628, "learning_rate": 1.6054648026289617e-05, "loss": 2.0144174194335935, "step": 99200 }, { "epoch": 0.3948618232272246, "grad_norm": 0.5479264259338379, "learning_rate": 1.605265879993953e-05, "loss": 2.0254780578613283, "step": 99250 }, { "epoch": 0.3950607460600846, "grad_norm": 0.49256765842437744, "learning_rate": 1.6050669573589442e-05, "loss": 2.048308563232422, "step": 99300 }, { "epoch": 0.3952596688929447, "grad_norm": 0.5902886390686035, "learning_rate": 1.6048680347239354e-05, "loss": 2.0256607055664064, "step": 99350 }, { "epoch": 0.3954585917258048, "grad_norm": 0.5941340923309326, "learning_rate": 1.6046691120889263e-05, "loss": 2.060225067138672, "step": 99400 }, { "epoch": 0.3956575145586648, "grad_norm": 0.5769845247268677, "learning_rate": 1.6044701894539175e-05, "loss": 2.03699951171875, "step": 99450 }, { "epoch": 0.3958564373915249, "grad_norm": 0.6020959615707397, "learning_rate": 1.604271266818909e-05, "loss": 2.046849822998047, "step": 99500 }, { "epoch": 0.396055360224385, "grad_norm": 0.5068184733390808, "learning_rate": 1.6040723441839002e-05, "loss": 2.02755615234375, "step": 99550 }, { "epoch": 0.396254283057245, "grad_norm": 0.6057761907577515, "learning_rate": 1.6038734215488915e-05, "loss": 2.051125946044922, "step": 99600 }, { "epoch": 0.3964532058901051, "grad_norm": 0.5733392834663391, "learning_rate": 1.6036744989138827e-05, "loss": 2.04673828125, "step": 99650 }, { "epoch": 0.3966521287229651, "grad_norm": 0.6285498142242432, "learning_rate": 1.603475576278874e-05, "loss": 2.016423797607422, "step": 99700 }, { "epoch": 0.39685105155582523, "grad_norm": 0.5761775374412537, "learning_rate": 1.6032766536438648e-05, "loss": 1.9997164916992187, "step": 99750 }, { "epoch": 0.3970499743886853, "grad_norm": 0.7095495462417603, "learning_rate": 1.603077731008856e-05, "loss": 1.9718325805664063, "step": 99800 }, { "epoch": 0.3972488972215453, "grad_norm": 0.6016948223114014, "learning_rate": 1.6028788083738472e-05, "loss": 2.0423333740234373, "step": 99850 }, { "epoch": 0.3974478200544054, "grad_norm": 0.4606049358844757, "learning_rate": 1.6026798857388387e-05, "loss": 2.085349578857422, "step": 99900 }, { "epoch": 0.3976467428872655, "grad_norm": 0.6217737793922424, "learning_rate": 1.60248096310383e-05, "loss": 2.009854278564453, "step": 99950 }, { "epoch": 0.39784566572012553, "grad_norm": 0.6302829384803772, "learning_rate": 1.6022820404688212e-05, "loss": 1.9848240661621093, "step": 100000 }, { "epoch": 0.3980445885529856, "grad_norm": 0.5880136489868164, "learning_rate": 1.602083117833812e-05, "loss": 2.0382843017578125, "step": 100050 }, { "epoch": 0.3982435113858456, "grad_norm": 0.44991084933280945, "learning_rate": 1.6018841951988033e-05, "loss": 2.056185455322266, "step": 100100 }, { "epoch": 0.39844243421870573, "grad_norm": 0.5599552989006042, "learning_rate": 1.6016852725637945e-05, "loss": 2.046284942626953, "step": 100150 }, { "epoch": 0.3986413570515658, "grad_norm": 0.5603101849555969, "learning_rate": 1.6014863499287857e-05, "loss": 2.0582737731933594, "step": 100200 }, { "epoch": 0.39884027988442583, "grad_norm": 0.5710169672966003, "learning_rate": 1.601287427293777e-05, "loss": 2.076534881591797, "step": 100250 }, { "epoch": 0.3990392027172859, "grad_norm": 0.4584129750728607, "learning_rate": 1.6010885046587685e-05, "loss": 2.016046600341797, "step": 100300 }, { "epoch": 0.399238125550146, "grad_norm": 0.5805577635765076, "learning_rate": 1.6008895820237597e-05, "loss": 2.0220970153808593, "step": 100350 }, { "epoch": 0.39943704838300603, "grad_norm": 0.6203105449676514, "learning_rate": 1.6006906593887506e-05, "loss": 2.0069972229003907, "step": 100400 }, { "epoch": 0.3996359712158661, "grad_norm": 0.6074467301368713, "learning_rate": 1.6004917367537418e-05, "loss": 2.018644256591797, "step": 100450 }, { "epoch": 0.39983489404872613, "grad_norm": 0.6914571523666382, "learning_rate": 1.600292814118733e-05, "loss": 2.0558351135253905, "step": 100500 }, { "epoch": 0.40003381688158623, "grad_norm": 0.6200302243232727, "learning_rate": 1.6000938914837242e-05, "loss": 2.0001289367675783, "step": 100550 }, { "epoch": 0.4002327397144463, "grad_norm": 0.5944579839706421, "learning_rate": 1.5998949688487154e-05, "loss": 2.03564208984375, "step": 100600 }, { "epoch": 0.40043166254730633, "grad_norm": 0.5863574743270874, "learning_rate": 1.5996960462137066e-05, "loss": 2.041003875732422, "step": 100650 }, { "epoch": 0.4006305853801664, "grad_norm": 0.667042076587677, "learning_rate": 1.599497123578698e-05, "loss": 2.0360150146484375, "step": 100700 }, { "epoch": 0.4008295082130265, "grad_norm": 0.5328896045684814, "learning_rate": 1.599298200943689e-05, "loss": 2.024744873046875, "step": 100750 }, { "epoch": 0.40102843104588654, "grad_norm": 0.5813261270523071, "learning_rate": 1.5991032567613804e-05, "loss": 2.0770477294921874, "step": 100800 }, { "epoch": 0.4012273538787466, "grad_norm": 0.6324332356452942, "learning_rate": 1.5989043341263716e-05, "loss": 2.0534568786621095, "step": 100850 }, { "epoch": 0.40142627671160663, "grad_norm": 0.5318690538406372, "learning_rate": 1.598705411491363e-05, "loss": 2.053411102294922, "step": 100900 }, { "epoch": 0.40162519954446674, "grad_norm": 0.6649624705314636, "learning_rate": 1.598506488856354e-05, "loss": 2.022134094238281, "step": 100950 }, { "epoch": 0.4018241223773268, "grad_norm": 0.6840423345565796, "learning_rate": 1.5983075662213453e-05, "loss": 2.0200076293945313, "step": 101000 }, { "epoch": 0.40202304521018684, "grad_norm": 0.5455690622329712, "learning_rate": 1.5981086435863365e-05, "loss": 2.0220762634277345, "step": 101050 }, { "epoch": 0.4022219680430469, "grad_norm": 0.5812079310417175, "learning_rate": 1.5979097209513277e-05, "loss": 1.945897979736328, "step": 101100 }, { "epoch": 0.402420890875907, "grad_norm": 0.5673421025276184, "learning_rate": 1.597710798316319e-05, "loss": 2.026407470703125, "step": 101150 }, { "epoch": 0.40261981370876704, "grad_norm": 0.7264680862426758, "learning_rate": 1.59751187568131e-05, "loss": 1.9915069580078124, "step": 101200 }, { "epoch": 0.4028187365416271, "grad_norm": 0.5428031086921692, "learning_rate": 1.5973129530463013e-05, "loss": 1.9972003173828126, "step": 101250 }, { "epoch": 0.40301765937448714, "grad_norm": 0.5144270658493042, "learning_rate": 1.5971140304112926e-05, "loss": 2.0037420654296874, "step": 101300 }, { "epoch": 0.40321658220734724, "grad_norm": 0.5078089237213135, "learning_rate": 1.5969151077762838e-05, "loss": 1.9882896423339844, "step": 101350 }, { "epoch": 0.4034155050402073, "grad_norm": 0.5933693647384644, "learning_rate": 1.596716185141275e-05, "loss": 2.0644549560546874, "step": 101400 }, { "epoch": 0.40361442787306734, "grad_norm": 0.6145095229148865, "learning_rate": 1.5965172625062662e-05, "loss": 2.0577508544921876, "step": 101450 }, { "epoch": 0.4038133507059274, "grad_norm": 0.5975869297981262, "learning_rate": 1.5963183398712574e-05, "loss": 1.9942539978027343, "step": 101500 }, { "epoch": 0.4040122735387875, "grad_norm": 0.4563092887401581, "learning_rate": 1.5961194172362486e-05, "loss": 2.0521421813964844, "step": 101550 }, { "epoch": 0.40421119637164754, "grad_norm": 0.5749307870864868, "learning_rate": 1.59592049460124e-05, "loss": 2.0178482055664064, "step": 101600 }, { "epoch": 0.4044101192045076, "grad_norm": 0.5967308878898621, "learning_rate": 1.595721571966231e-05, "loss": 2.079739227294922, "step": 101650 }, { "epoch": 0.40460904203736764, "grad_norm": 0.531603217124939, "learning_rate": 1.5955226493312223e-05, "loss": 2.0194381713867187, "step": 101700 }, { "epoch": 0.40480796487022774, "grad_norm": 0.5158302187919617, "learning_rate": 1.5953237266962135e-05, "loss": 2.006327362060547, "step": 101750 }, { "epoch": 0.4050068877030878, "grad_norm": 0.5346975326538086, "learning_rate": 1.5951248040612047e-05, "loss": 2.051902618408203, "step": 101800 }, { "epoch": 0.40520581053594784, "grad_norm": 0.5547789931297302, "learning_rate": 1.594925881426196e-05, "loss": 1.98944091796875, "step": 101850 }, { "epoch": 0.4054047333688079, "grad_norm": 0.5399845242500305, "learning_rate": 1.594726958791187e-05, "loss": 1.995567626953125, "step": 101900 }, { "epoch": 0.405603656201668, "grad_norm": 0.5356823801994324, "learning_rate": 1.5945280361561783e-05, "loss": 2.0268324279785155, "step": 101950 }, { "epoch": 0.40580257903452804, "grad_norm": 0.5903404355049133, "learning_rate": 1.5943291135211692e-05, "loss": 2.028431701660156, "step": 102000 }, { "epoch": 0.4060015018673881, "grad_norm": 0.6568157076835632, "learning_rate": 1.5941301908861608e-05, "loss": 2.0587448120117187, "step": 102050 }, { "epoch": 0.40620042470024814, "grad_norm": 0.6172178983688354, "learning_rate": 1.593931268251152e-05, "loss": 1.977887420654297, "step": 102100 }, { "epoch": 0.40639934753310825, "grad_norm": 0.5947266221046448, "learning_rate": 1.5937323456161432e-05, "loss": 2.029330596923828, "step": 102150 }, { "epoch": 0.4065982703659683, "grad_norm": 0.4803697466850281, "learning_rate": 1.5935334229811344e-05, "loss": 1.9967420959472657, "step": 102200 }, { "epoch": 0.40679719319882834, "grad_norm": 0.6124541163444519, "learning_rate": 1.5933345003461256e-05, "loss": 2.0563003540039064, "step": 102250 }, { "epoch": 0.4069961160316884, "grad_norm": 0.605090320110321, "learning_rate": 1.593135577711117e-05, "loss": 2.0054440307617187, "step": 102300 }, { "epoch": 0.4071950388645485, "grad_norm": 0.5930483937263489, "learning_rate": 1.5929366550761077e-05, "loss": 2.0723443603515626, "step": 102350 }, { "epoch": 0.40739396169740855, "grad_norm": 0.5767683982849121, "learning_rate": 1.592737732441099e-05, "loss": 2.0635267639160157, "step": 102400 }, { "epoch": 0.4075928845302686, "grad_norm": 0.510838508605957, "learning_rate": 1.5925427882587906e-05, "loss": 1.9927958679199218, "step": 102450 }, { "epoch": 0.40779180736312864, "grad_norm": 0.6352221965789795, "learning_rate": 1.592343865623782e-05, "loss": 2.0500636291503906, "step": 102500 }, { "epoch": 0.4079907301959887, "grad_norm": 0.6125971674919128, "learning_rate": 1.592144942988773e-05, "loss": 2.001520538330078, "step": 102550 }, { "epoch": 0.4081896530288488, "grad_norm": 0.544002115726471, "learning_rate": 1.5919460203537643e-05, "loss": 2.007682037353516, "step": 102600 }, { "epoch": 0.40838857586170885, "grad_norm": 0.44057950377464294, "learning_rate": 1.5917470977187555e-05, "loss": 2.0328909301757814, "step": 102650 }, { "epoch": 0.4085874986945689, "grad_norm": 0.6196190714836121, "learning_rate": 1.5915481750837464e-05, "loss": 2.0069094848632814, "step": 102700 }, { "epoch": 0.40878642152742894, "grad_norm": 0.5738869309425354, "learning_rate": 1.5913492524487376e-05, "loss": 1.994451904296875, "step": 102750 }, { "epoch": 0.40898534436028905, "grad_norm": 0.5392487049102783, "learning_rate": 1.591150329813729e-05, "loss": 1.973696746826172, "step": 102800 }, { "epoch": 0.4091842671931491, "grad_norm": 0.6632168292999268, "learning_rate": 1.5909514071787203e-05, "loss": 2.0037916564941405, "step": 102850 }, { "epoch": 0.40938319002600915, "grad_norm": 0.6393146514892578, "learning_rate": 1.5907524845437116e-05, "loss": 1.9819448852539063, "step": 102900 }, { "epoch": 0.4095821128588692, "grad_norm": 0.577155351638794, "learning_rate": 1.5905535619087028e-05, "loss": 2.0344342041015624, "step": 102950 }, { "epoch": 0.4097810356917293, "grad_norm": 0.6163491606712341, "learning_rate": 1.5903546392736936e-05, "loss": 2.0480390930175782, "step": 103000 }, { "epoch": 0.40997995852458935, "grad_norm": 0.6747498512268066, "learning_rate": 1.590155716638685e-05, "loss": 2.029280242919922, "step": 103050 }, { "epoch": 0.4101788813574494, "grad_norm": 0.5582685470581055, "learning_rate": 1.589956794003676e-05, "loss": 2.010467224121094, "step": 103100 }, { "epoch": 0.41037780419030945, "grad_norm": 0.6403326988220215, "learning_rate": 1.5897578713686673e-05, "loss": 2.0132489013671875, "step": 103150 }, { "epoch": 0.41057672702316955, "grad_norm": 0.5428339838981628, "learning_rate": 1.589558948733659e-05, "loss": 2.069123077392578, "step": 103200 }, { "epoch": 0.4107756498560296, "grad_norm": 0.638884961605072, "learning_rate": 1.58936002609865e-05, "loss": 1.9992475891113282, "step": 103250 }, { "epoch": 0.41097457268888965, "grad_norm": 0.6499197483062744, "learning_rate": 1.589161103463641e-05, "loss": 2.062555236816406, "step": 103300 }, { "epoch": 0.4111734955217497, "grad_norm": 0.5775219202041626, "learning_rate": 1.588962180828632e-05, "loss": 2.0888999938964843, "step": 103350 }, { "epoch": 0.4113724183546098, "grad_norm": 0.48516595363616943, "learning_rate": 1.5887632581936234e-05, "loss": 1.980084228515625, "step": 103400 }, { "epoch": 0.41157134118746985, "grad_norm": 0.6440320014953613, "learning_rate": 1.5885643355586146e-05, "loss": 2.026065673828125, "step": 103450 }, { "epoch": 0.4117702640203299, "grad_norm": 0.5253552198410034, "learning_rate": 1.5883654129236058e-05, "loss": 1.9931268310546875, "step": 103500 }, { "epoch": 0.41196918685318995, "grad_norm": 0.5661066174507141, "learning_rate": 1.588166490288597e-05, "loss": 1.9701498413085938, "step": 103550 }, { "epoch": 0.41216810968605005, "grad_norm": 0.4911993145942688, "learning_rate": 1.5879675676535886e-05, "loss": 1.9913458251953124, "step": 103600 }, { "epoch": 0.4123670325189101, "grad_norm": 0.6432226300239563, "learning_rate": 1.5877686450185794e-05, "loss": 2.018122253417969, "step": 103650 }, { "epoch": 0.41256595535177015, "grad_norm": 0.537059485912323, "learning_rate": 1.5875697223835707e-05, "loss": 1.97571044921875, "step": 103700 }, { "epoch": 0.4127648781846302, "grad_norm": 0.6042221784591675, "learning_rate": 1.587370799748562e-05, "loss": 2.04801025390625, "step": 103750 }, { "epoch": 0.4129638010174903, "grad_norm": 0.6263375282287598, "learning_rate": 1.587171877113553e-05, "loss": 2.0200340270996096, "step": 103800 }, { "epoch": 0.41316272385035036, "grad_norm": 0.5997186303138733, "learning_rate": 1.5869729544785443e-05, "loss": 2.125528564453125, "step": 103850 }, { "epoch": 0.4133616466832104, "grad_norm": 0.6131992340087891, "learning_rate": 1.5867740318435355e-05, "loss": 1.9792034912109375, "step": 103900 }, { "epoch": 0.41356056951607045, "grad_norm": 0.6910045146942139, "learning_rate": 1.5865751092085267e-05, "loss": 2.060076904296875, "step": 103950 }, { "epoch": 0.41375949234893056, "grad_norm": 0.5906137228012085, "learning_rate": 1.586376186573518e-05, "loss": 1.9856033325195312, "step": 104000 }, { "epoch": 0.4139584151817906, "grad_norm": 0.5713112950325012, "learning_rate": 1.586177263938509e-05, "loss": 1.983689422607422, "step": 104050 }, { "epoch": 0.41415733801465066, "grad_norm": 0.5596773028373718, "learning_rate": 1.5859783413035004e-05, "loss": 2.0459230041503904, "step": 104100 }, { "epoch": 0.4143562608475107, "grad_norm": 0.6240203976631165, "learning_rate": 1.5857794186684916e-05, "loss": 2.046031188964844, "step": 104150 }, { "epoch": 0.4145551836803708, "grad_norm": 0.6231343746185303, "learning_rate": 1.5855804960334828e-05, "loss": 1.9886212158203125, "step": 104200 }, { "epoch": 0.41475410651323086, "grad_norm": 0.6098334789276123, "learning_rate": 1.585381573398474e-05, "loss": 2.0150770568847656, "step": 104250 }, { "epoch": 0.4149530293460909, "grad_norm": 0.6104575395584106, "learning_rate": 1.5851826507634652e-05, "loss": 2.0012770080566407, "step": 104300 }, { "epoch": 0.41515195217895096, "grad_norm": 0.4864068329334259, "learning_rate": 1.5849837281284564e-05, "loss": 2.0416722106933594, "step": 104350 }, { "epoch": 0.41535087501181106, "grad_norm": 0.6306304931640625, "learning_rate": 1.5847848054934477e-05, "loss": 1.99920166015625, "step": 104400 }, { "epoch": 0.4155497978446711, "grad_norm": 0.5246489644050598, "learning_rate": 1.584585882858439e-05, "loss": 1.9962995910644532, "step": 104450 }, { "epoch": 0.41574872067753116, "grad_norm": 0.5989525318145752, "learning_rate": 1.58438696022343e-05, "loss": 2.014199981689453, "step": 104500 }, { "epoch": 0.4159476435103912, "grad_norm": 0.6030387878417969, "learning_rate": 1.5841880375884213e-05, "loss": 2.0264312744140627, "step": 104550 }, { "epoch": 0.4161465663432513, "grad_norm": 0.5437155961990356, "learning_rate": 1.5839891149534122e-05, "loss": 2.0293899536132813, "step": 104600 }, { "epoch": 0.41634548917611136, "grad_norm": 0.5850275158882141, "learning_rate": 1.5837901923184037e-05, "loss": 2.009674530029297, "step": 104650 }, { "epoch": 0.4165444120089714, "grad_norm": 0.6172714233398438, "learning_rate": 1.583591269683395e-05, "loss": 2.01094970703125, "step": 104700 }, { "epoch": 0.41674333484183146, "grad_norm": 0.6591300964355469, "learning_rate": 1.583392347048386e-05, "loss": 2.057313995361328, "step": 104750 }, { "epoch": 0.41694225767469156, "grad_norm": 0.590181291103363, "learning_rate": 1.5831934244133774e-05, "loss": 2.0005479431152344, "step": 104800 }, { "epoch": 0.4171411805075516, "grad_norm": 0.5789369940757751, "learning_rate": 1.5829945017783686e-05, "loss": 2.0391880798339845, "step": 104850 }, { "epoch": 0.41734010334041166, "grad_norm": 0.534511923789978, "learning_rate": 1.58279955759606e-05, "loss": 1.9802256774902345, "step": 104900 }, { "epoch": 0.4175390261732717, "grad_norm": 0.5188771486282349, "learning_rate": 1.582600634961051e-05, "loss": 2.0033021545410157, "step": 104950 }, { "epoch": 0.4177379490061318, "grad_norm": 0.5677863359451294, "learning_rate": 1.5824017123260424e-05, "loss": 2.0110342407226565, "step": 105000 }, { "epoch": 0.41793687183899186, "grad_norm": 0.5763332843780518, "learning_rate": 1.5822027896910336e-05, "loss": 1.968284149169922, "step": 105050 }, { "epoch": 0.4181357946718519, "grad_norm": 0.6452218890190125, "learning_rate": 1.5820038670560248e-05, "loss": 1.9870907592773437, "step": 105100 }, { "epoch": 0.41833471750471196, "grad_norm": 0.48686617612838745, "learning_rate": 1.581804944421016e-05, "loss": 2.0306671142578123, "step": 105150 }, { "epoch": 0.41853364033757207, "grad_norm": 0.601513683795929, "learning_rate": 1.5816060217860072e-05, "loss": 2.010018157958984, "step": 105200 }, { "epoch": 0.4187325631704321, "grad_norm": 0.6213662028312683, "learning_rate": 1.581407099150998e-05, "loss": 2.0043663024902343, "step": 105250 }, { "epoch": 0.41893148600329216, "grad_norm": 0.6800434589385986, "learning_rate": 1.5812081765159893e-05, "loss": 2.0451895141601564, "step": 105300 }, { "epoch": 0.4191304088361522, "grad_norm": 0.5756286978721619, "learning_rate": 1.581009253880981e-05, "loss": 2.037020416259766, "step": 105350 }, { "epoch": 0.4193293316690123, "grad_norm": 0.7522449493408203, "learning_rate": 1.580810331245972e-05, "loss": 2.063944549560547, "step": 105400 }, { "epoch": 0.41952825450187237, "grad_norm": 0.5189664959907532, "learning_rate": 1.5806114086109633e-05, "loss": 1.9873712158203125, "step": 105450 }, { "epoch": 0.4197271773347324, "grad_norm": 0.6365987062454224, "learning_rate": 1.5804124859759545e-05, "loss": 2.0384906005859373, "step": 105500 }, { "epoch": 0.41992610016759246, "grad_norm": 0.43872639536857605, "learning_rate": 1.5802135633409457e-05, "loss": 2.0293238830566405, "step": 105550 }, { "epoch": 0.42012502300045257, "grad_norm": 0.5242419838905334, "learning_rate": 1.5800146407059366e-05, "loss": 2.039922637939453, "step": 105600 }, { "epoch": 0.4203239458333126, "grad_norm": 0.5106091499328613, "learning_rate": 1.5798157180709278e-05, "loss": 2.0302935791015626, "step": 105650 }, { "epoch": 0.42052286866617267, "grad_norm": 0.6344772577285767, "learning_rate": 1.579616795435919e-05, "loss": 2.037156829833984, "step": 105700 }, { "epoch": 0.4207217914990327, "grad_norm": 0.6160595417022705, "learning_rate": 1.5794178728009102e-05, "loss": 2.006025390625, "step": 105750 }, { "epoch": 0.4209207143318928, "grad_norm": 0.5730109214782715, "learning_rate": 1.5792189501659018e-05, "loss": 1.9671702575683594, "step": 105800 }, { "epoch": 0.42111963716475287, "grad_norm": 0.5598406791687012, "learning_rate": 1.579020027530893e-05, "loss": 2.0537139892578127, "step": 105850 }, { "epoch": 0.4213185599976129, "grad_norm": 0.5225928425788879, "learning_rate": 1.578821104895884e-05, "loss": 2.0057420349121093, "step": 105900 }, { "epoch": 0.42151748283047297, "grad_norm": 0.33408278226852417, "learning_rate": 1.578622182260875e-05, "loss": 1.9952406311035156, "step": 105950 }, { "epoch": 0.42171640566333307, "grad_norm": 0.6214371919631958, "learning_rate": 1.5784232596258663e-05, "loss": 2.043406524658203, "step": 106000 }, { "epoch": 0.4219153284961931, "grad_norm": 0.5954177379608154, "learning_rate": 1.5782243369908575e-05, "loss": 1.9692987060546876, "step": 106050 }, { "epoch": 0.42211425132905317, "grad_norm": 0.5538131594657898, "learning_rate": 1.5780254143558487e-05, "loss": 2.0246658325195312, "step": 106100 }, { "epoch": 0.4223131741619132, "grad_norm": 0.55618816614151, "learning_rate": 1.57782649172084e-05, "loss": 2.0135539245605467, "step": 106150 }, { "epoch": 0.4225120969947733, "grad_norm": 0.5811052918434143, "learning_rate": 1.5776275690858315e-05, "loss": 2.063972625732422, "step": 106200 }, { "epoch": 0.42271101982763337, "grad_norm": 0.6621485948562622, "learning_rate": 1.5774286464508224e-05, "loss": 1.977401885986328, "step": 106250 }, { "epoch": 0.4229099426604934, "grad_norm": 0.5402007699012756, "learning_rate": 1.5772297238158136e-05, "loss": 2.083381805419922, "step": 106300 }, { "epoch": 0.42310886549335347, "grad_norm": 0.5739873051643372, "learning_rate": 1.5770308011808048e-05, "loss": 2.049042510986328, "step": 106350 }, { "epoch": 0.4233077883262136, "grad_norm": 0.5945102572441101, "learning_rate": 1.576831878545796e-05, "loss": 1.9563458251953125, "step": 106400 }, { "epoch": 0.4235067111590736, "grad_norm": 0.6311101317405701, "learning_rate": 1.5766329559107873e-05, "loss": 2.0748699951171874, "step": 106450 }, { "epoch": 0.4237056339919337, "grad_norm": 0.44188782572746277, "learning_rate": 1.5764340332757785e-05, "loss": 2.0623912048339843, "step": 106500 }, { "epoch": 0.4239045568247937, "grad_norm": 0.5757001638412476, "learning_rate": 1.5762351106407697e-05, "loss": 1.9966523742675781, "step": 106550 }, { "epoch": 0.4241034796576538, "grad_norm": 0.5873962044715881, "learning_rate": 1.576036188005761e-05, "loss": 2.0747866821289063, "step": 106600 }, { "epoch": 0.4243024024905139, "grad_norm": 0.4918270707130432, "learning_rate": 1.575837265370752e-05, "loss": 2.0040594482421876, "step": 106650 }, { "epoch": 0.4245013253233739, "grad_norm": 0.5991538166999817, "learning_rate": 1.5756383427357433e-05, "loss": 2.022547302246094, "step": 106700 }, { "epoch": 0.424700248156234, "grad_norm": 0.5533703565597534, "learning_rate": 1.5754394201007345e-05, "loss": 2.077634582519531, "step": 106750 }, { "epoch": 0.4248991709890941, "grad_norm": 0.5758677124977112, "learning_rate": 1.5752404974657258e-05, "loss": 2.0182205200195313, "step": 106800 }, { "epoch": 0.4250980938219541, "grad_norm": 0.508793830871582, "learning_rate": 1.575041574830717e-05, "loss": 1.9863348388671875, "step": 106850 }, { "epoch": 0.4252970166548142, "grad_norm": 0.5705050826072693, "learning_rate": 1.5748426521957082e-05, "loss": 2.0701097106933593, "step": 106900 }, { "epoch": 0.4254959394876742, "grad_norm": 0.47584524750709534, "learning_rate": 1.5746437295606994e-05, "loss": 1.9861077880859375, "step": 106950 }, { "epoch": 0.42569486232053433, "grad_norm": 0.467475026845932, "learning_rate": 1.5744448069256906e-05, "loss": 2.023051910400391, "step": 107000 }, { "epoch": 0.4258937851533944, "grad_norm": 0.6391743421554565, "learning_rate": 1.5742458842906818e-05, "loss": 2.045084991455078, "step": 107050 }, { "epoch": 0.4260927079862544, "grad_norm": 0.5707900524139404, "learning_rate": 1.574046961655673e-05, "loss": 2.0399209594726564, "step": 107100 }, { "epoch": 0.4262916308191145, "grad_norm": 0.6412329077720642, "learning_rate": 1.5738480390206643e-05, "loss": 1.9943521118164063, "step": 107150 }, { "epoch": 0.4264905536519746, "grad_norm": 0.44099560379981995, "learning_rate": 1.5736491163856555e-05, "loss": 2.007812805175781, "step": 107200 }, { "epoch": 0.42668947648483463, "grad_norm": 0.5545750260353088, "learning_rate": 1.5734501937506467e-05, "loss": 1.9521875, "step": 107250 }, { "epoch": 0.4268883993176947, "grad_norm": 0.6346858143806458, "learning_rate": 1.573251271115638e-05, "loss": 2.046122283935547, "step": 107300 }, { "epoch": 0.4270873221505547, "grad_norm": 0.5755708813667297, "learning_rate": 1.573052348480629e-05, "loss": 2.088772888183594, "step": 107350 }, { "epoch": 0.42728624498341483, "grad_norm": 0.5291570425033569, "learning_rate": 1.5728534258456203e-05, "loss": 2.0219297790527344, "step": 107400 }, { "epoch": 0.4274851678162749, "grad_norm": 0.6378032565116882, "learning_rate": 1.5726584816633117e-05, "loss": 2.009870758056641, "step": 107450 }, { "epoch": 0.42768409064913493, "grad_norm": 0.558759331703186, "learning_rate": 1.572459559028303e-05, "loss": 2.00464111328125, "step": 107500 }, { "epoch": 0.427883013481995, "grad_norm": 0.9345026016235352, "learning_rate": 1.572260636393294e-05, "loss": 1.9784445190429687, "step": 107550 }, { "epoch": 0.4280819363148551, "grad_norm": 0.5620555877685547, "learning_rate": 1.5720617137582853e-05, "loss": 2.02440185546875, "step": 107600 }, { "epoch": 0.42828085914771513, "grad_norm": 0.6408937573432922, "learning_rate": 1.5718627911232765e-05, "loss": 2.0326249694824217, "step": 107650 }, { "epoch": 0.4284797819805752, "grad_norm": 0.5496846437454224, "learning_rate": 1.5716638684882678e-05, "loss": 2.0465635681152343, "step": 107700 }, { "epoch": 0.42867870481343523, "grad_norm": 0.5649222135543823, "learning_rate": 1.571464945853259e-05, "loss": 2.0327662658691406, "step": 107750 }, { "epoch": 0.42887762764629533, "grad_norm": 0.5060648918151855, "learning_rate": 1.5712660232182502e-05, "loss": 2.016465301513672, "step": 107800 }, { "epoch": 0.4290765504791554, "grad_norm": 0.5293127298355103, "learning_rate": 1.571067100583241e-05, "loss": 2.014732666015625, "step": 107850 }, { "epoch": 0.42927547331201543, "grad_norm": 0.5771971344947815, "learning_rate": 1.5708681779482323e-05, "loss": 1.9961195373535157, "step": 107900 }, { "epoch": 0.4294743961448755, "grad_norm": 0.6530756950378418, "learning_rate": 1.5706692553132238e-05, "loss": 2.0242214965820313, "step": 107950 }, { "epoch": 0.4296733189777356, "grad_norm": 0.5970114469528198, "learning_rate": 1.570470332678215e-05, "loss": 1.9920317077636718, "step": 108000 }, { "epoch": 0.42987224181059563, "grad_norm": 0.6131753921508789, "learning_rate": 1.5702714100432063e-05, "loss": 2.0349490356445314, "step": 108050 }, { "epoch": 0.4300711646434557, "grad_norm": 0.5894033312797546, "learning_rate": 1.5700724874081975e-05, "loss": 2.0218357849121094, "step": 108100 }, { "epoch": 0.43027008747631573, "grad_norm": 0.5610945820808411, "learning_rate": 1.5698735647731887e-05, "loss": 2.0199755859375, "step": 108150 }, { "epoch": 0.43046901030917584, "grad_norm": 0.6034848093986511, "learning_rate": 1.5696746421381796e-05, "loss": 2.0381756591796876, "step": 108200 }, { "epoch": 0.4306679331420359, "grad_norm": 0.5624331831932068, "learning_rate": 1.5694757195031708e-05, "loss": 2.0000299072265624, "step": 108250 }, { "epoch": 0.43086685597489593, "grad_norm": 0.6468844413757324, "learning_rate": 1.569276796868162e-05, "loss": 2.037781219482422, "step": 108300 }, { "epoch": 0.431065778807756, "grad_norm": 0.5776414275169373, "learning_rate": 1.5690778742331535e-05, "loss": 1.9978172302246093, "step": 108350 }, { "epoch": 0.4312647016406161, "grad_norm": 0.5945411920547485, "learning_rate": 1.5688789515981448e-05, "loss": 2.0464556884765623, "step": 108400 }, { "epoch": 0.43146362447347614, "grad_norm": 0.5564348101615906, "learning_rate": 1.568680028963136e-05, "loss": 1.974677276611328, "step": 108450 }, { "epoch": 0.4316625473063362, "grad_norm": 0.5202946066856384, "learning_rate": 1.568481106328127e-05, "loss": 2.0140997314453126, "step": 108500 }, { "epoch": 0.43186147013919624, "grad_norm": 0.5874884128570557, "learning_rate": 1.568282183693118e-05, "loss": 1.9933041381835936, "step": 108550 }, { "epoch": 0.43206039297205634, "grad_norm": 0.5866652727127075, "learning_rate": 1.5680832610581093e-05, "loss": 2.0342747497558595, "step": 108600 }, { "epoch": 0.4322593158049164, "grad_norm": 0.5695591568946838, "learning_rate": 1.5678843384231005e-05, "loss": 2.031394348144531, "step": 108650 }, { "epoch": 0.43245823863777644, "grad_norm": 0.5568060874938965, "learning_rate": 1.5676854157880917e-05, "loss": 2.0162477111816406, "step": 108700 }, { "epoch": 0.4326571614706365, "grad_norm": 0.6552156209945679, "learning_rate": 1.5674864931530833e-05, "loss": 2.03097412109375, "step": 108750 }, { "epoch": 0.4328560843034966, "grad_norm": 0.5564373731613159, "learning_rate": 1.5672875705180745e-05, "loss": 2.053034210205078, "step": 108800 }, { "epoch": 0.43305500713635664, "grad_norm": 0.6743385195732117, "learning_rate": 1.5670886478830653e-05, "loss": 2.080913391113281, "step": 108850 }, { "epoch": 0.4332539299692167, "grad_norm": 0.5462058186531067, "learning_rate": 1.5668897252480566e-05, "loss": 2.0114535522460937, "step": 108900 }, { "epoch": 0.43345285280207674, "grad_norm": 0.638744592666626, "learning_rate": 1.5666908026130478e-05, "loss": 1.9776434326171874, "step": 108950 }, { "epoch": 0.43365177563493684, "grad_norm": 0.6746826767921448, "learning_rate": 1.566491879978039e-05, "loss": 2.0451420593261718, "step": 109000 }, { "epoch": 0.4338506984677969, "grad_norm": 0.715133011341095, "learning_rate": 1.5662929573430302e-05, "loss": 2.0730650329589846, "step": 109050 }, { "epoch": 0.43404962130065694, "grad_norm": 0.5619137287139893, "learning_rate": 1.5660940347080214e-05, "loss": 2.0520280456542968, "step": 109100 }, { "epoch": 0.434248544133517, "grad_norm": 0.6010045409202576, "learning_rate": 1.5658951120730126e-05, "loss": 2.0572149658203127, "step": 109150 }, { "epoch": 0.4344474669663771, "grad_norm": 0.4657900035381317, "learning_rate": 1.565696189438004e-05, "loss": 1.9999459838867188, "step": 109200 }, { "epoch": 0.43464638979923714, "grad_norm": 0.5727123022079468, "learning_rate": 1.565497266802995e-05, "loss": 2.003640899658203, "step": 109250 }, { "epoch": 0.4348453126320972, "grad_norm": 0.5811353325843811, "learning_rate": 1.5652983441679863e-05, "loss": 2.0513287353515626, "step": 109300 }, { "epoch": 0.43504423546495724, "grad_norm": 0.5730752944946289, "learning_rate": 1.5650994215329775e-05, "loss": 1.9886909484863282, "step": 109350 }, { "epoch": 0.43524315829781735, "grad_norm": 0.5260764360427856, "learning_rate": 1.5649004988979687e-05, "loss": 2.0524166870117186, "step": 109400 }, { "epoch": 0.4354420811306774, "grad_norm": 0.5535375475883484, "learning_rate": 1.56470157626296e-05, "loss": 2.0448207092285156, "step": 109450 }, { "epoch": 0.43564100396353744, "grad_norm": 0.6250180602073669, "learning_rate": 1.564502653627951e-05, "loss": 1.9890849304199218, "step": 109500 }, { "epoch": 0.4358399267963975, "grad_norm": 0.634581983089447, "learning_rate": 1.5643037309929424e-05, "loss": 1.9935536193847656, "step": 109550 }, { "epoch": 0.4360388496292576, "grad_norm": 0.6139950752258301, "learning_rate": 1.5641048083579336e-05, "loss": 2.023534851074219, "step": 109600 }, { "epoch": 0.43623777246211765, "grad_norm": 0.6372048258781433, "learning_rate": 1.5639058857229248e-05, "loss": 1.985661163330078, "step": 109650 }, { "epoch": 0.4364366952949777, "grad_norm": 0.5155043601989746, "learning_rate": 1.563706963087916e-05, "loss": 2.0759092712402345, "step": 109700 }, { "epoch": 0.43663561812783774, "grad_norm": 0.5532680153846741, "learning_rate": 1.5635080404529072e-05, "loss": 1.9488262939453125, "step": 109750 }, { "epoch": 0.43683454096069785, "grad_norm": 0.7025464773178101, "learning_rate": 1.5633091178178984e-05, "loss": 1.9884855651855469, "step": 109800 }, { "epoch": 0.4370334637935579, "grad_norm": 0.4990638494491577, "learning_rate": 1.5631101951828896e-05, "loss": 2.0309034729003907, "step": 109850 }, { "epoch": 0.43723238662641795, "grad_norm": 0.6727657914161682, "learning_rate": 1.562911272547881e-05, "loss": 2.032054748535156, "step": 109900 }, { "epoch": 0.437431309459278, "grad_norm": 0.6318202018737793, "learning_rate": 1.562712349912872e-05, "loss": 2.0248648071289064, "step": 109950 }, { "epoch": 0.4376302322921381, "grad_norm": 0.6151597499847412, "learning_rate": 1.5625134272778633e-05, "loss": 2.033201141357422, "step": 110000 }, { "epoch": 0.43782915512499815, "grad_norm": 0.6622610092163086, "learning_rate": 1.5623145046428545e-05, "loss": 1.987083740234375, "step": 110050 }, { "epoch": 0.4380280779578582, "grad_norm": 0.5208989977836609, "learning_rate": 1.5621155820078457e-05, "loss": 2.019400787353516, "step": 110100 }, { "epoch": 0.43822700079071825, "grad_norm": 0.5526440143585205, "learning_rate": 1.561916659372837e-05, "loss": 2.0138400268554686, "step": 110150 }, { "epoch": 0.43842592362357835, "grad_norm": 0.6110556721687317, "learning_rate": 1.561717736737828e-05, "loss": 2.025318908691406, "step": 110200 }, { "epoch": 0.4386248464564384, "grad_norm": 0.6816936135292053, "learning_rate": 1.5615188141028194e-05, "loss": 1.9968943786621094, "step": 110250 }, { "epoch": 0.43882376928929845, "grad_norm": 0.704780638217926, "learning_rate": 1.5613198914678106e-05, "loss": 2.0037492370605468, "step": 110300 }, { "epoch": 0.4390226921221585, "grad_norm": 0.6684925556182861, "learning_rate": 1.5611209688328018e-05, "loss": 2.014828338623047, "step": 110350 }, { "epoch": 0.43922161495501855, "grad_norm": 0.5475967526435852, "learning_rate": 1.560922046197793e-05, "loss": 2.0127085876464843, "step": 110400 }, { "epoch": 0.43942053778787865, "grad_norm": 0.6398653984069824, "learning_rate": 1.560723123562784e-05, "loss": 2.0104336547851562, "step": 110450 }, { "epoch": 0.4396194606207387, "grad_norm": 0.5636106133460999, "learning_rate": 1.560524200927775e-05, "loss": 1.976803741455078, "step": 110500 }, { "epoch": 0.43981838345359875, "grad_norm": 0.49990981817245483, "learning_rate": 1.5603252782927666e-05, "loss": 2.012068328857422, "step": 110550 }, { "epoch": 0.4400173062864588, "grad_norm": 0.6181021332740784, "learning_rate": 1.560126355657758e-05, "loss": 2.01276123046875, "step": 110600 }, { "epoch": 0.4402162291193189, "grad_norm": 0.5278449058532715, "learning_rate": 1.559927433022749e-05, "loss": 2.007413787841797, "step": 110650 }, { "epoch": 0.44041515195217895, "grad_norm": 0.6162600517272949, "learning_rate": 1.5597285103877403e-05, "loss": 1.992371826171875, "step": 110700 }, { "epoch": 0.440614074785039, "grad_norm": 0.5570794343948364, "learning_rate": 1.5595295877527315e-05, "loss": 2.018826904296875, "step": 110750 }, { "epoch": 0.44081299761789905, "grad_norm": 0.6182676553726196, "learning_rate": 1.5593306651177224e-05, "loss": 2.0006118774414063, "step": 110800 }, { "epoch": 0.44101192045075915, "grad_norm": 0.6144781112670898, "learning_rate": 1.5591317424827136e-05, "loss": 2.027738037109375, "step": 110850 }, { "epoch": 0.4412108432836192, "grad_norm": 0.6235699653625488, "learning_rate": 1.5589328198477048e-05, "loss": 2.0063800048828124, "step": 110900 }, { "epoch": 0.44140976611647925, "grad_norm": 0.542779266834259, "learning_rate": 1.558733897212696e-05, "loss": 2.0013998413085936, "step": 110950 }, { "epoch": 0.4416086889493393, "grad_norm": 0.5570080280303955, "learning_rate": 1.5585349745776876e-05, "loss": 2.024033660888672, "step": 111000 }, { "epoch": 0.4418076117821994, "grad_norm": 0.6727036833763123, "learning_rate": 1.5583360519426788e-05, "loss": 1.970677490234375, "step": 111050 }, { "epoch": 0.44200653461505945, "grad_norm": 0.563866376876831, "learning_rate": 1.5581371293076697e-05, "loss": 2.004641418457031, "step": 111100 }, { "epoch": 0.4422054574479195, "grad_norm": 0.5490649938583374, "learning_rate": 1.557938206672661e-05, "loss": 2.0257559204101563, "step": 111150 }, { "epoch": 0.44240438028077955, "grad_norm": 0.5704553127288818, "learning_rate": 1.557739284037652e-05, "loss": 2.0454998779296876, "step": 111200 }, { "epoch": 0.44260330311363966, "grad_norm": 0.506682813167572, "learning_rate": 1.5575443398553434e-05, "loss": 1.99658447265625, "step": 111250 }, { "epoch": 0.4428022259464997, "grad_norm": 0.491109162569046, "learning_rate": 1.557345417220335e-05, "loss": 1.9868574523925782, "step": 111300 }, { "epoch": 0.44300114877935975, "grad_norm": 0.620080292224884, "learning_rate": 1.5571464945853262e-05, "loss": 2.0224761962890625, "step": 111350 }, { "epoch": 0.4432000716122198, "grad_norm": 0.6464053392410278, "learning_rate": 1.5569475719503174e-05, "loss": 2.0256619262695312, "step": 111400 }, { "epoch": 0.4433989944450799, "grad_norm": 0.6299174427986145, "learning_rate": 1.5567486493153083e-05, "loss": 2.005110778808594, "step": 111450 }, { "epoch": 0.44359791727793996, "grad_norm": 0.5939970016479492, "learning_rate": 1.5565497266802995e-05, "loss": 2.029161834716797, "step": 111500 }, { "epoch": 0.4437968401108, "grad_norm": 0.6070258021354675, "learning_rate": 1.5563508040452907e-05, "loss": 2.044627838134766, "step": 111550 }, { "epoch": 0.44399576294366006, "grad_norm": 0.5803679823875427, "learning_rate": 1.556151881410282e-05, "loss": 2.0818785095214842, "step": 111600 }, { "epoch": 0.44419468577652016, "grad_norm": 0.6061193346977234, "learning_rate": 1.555952958775273e-05, "loss": 2.003052978515625, "step": 111650 }, { "epoch": 0.4443936086093802, "grad_norm": 0.6207433938980103, "learning_rate": 1.5557540361402644e-05, "loss": 1.9960226440429687, "step": 111700 }, { "epoch": 0.44459253144224026, "grad_norm": 0.6020652651786804, "learning_rate": 1.5555551135052556e-05, "loss": 2.0223756408691407, "step": 111750 }, { "epoch": 0.4447914542751003, "grad_norm": 0.5577155351638794, "learning_rate": 1.5553561908702468e-05, "loss": 1.9963619995117188, "step": 111800 }, { "epoch": 0.4449903771079604, "grad_norm": 0.7404701709747314, "learning_rate": 1.555157268235238e-05, "loss": 2.034799041748047, "step": 111850 }, { "epoch": 0.44518929994082046, "grad_norm": 0.5160264372825623, "learning_rate": 1.5549583456002292e-05, "loss": 2.0115602111816404, "step": 111900 }, { "epoch": 0.4453882227736805, "grad_norm": 0.5868932604789734, "learning_rate": 1.5547594229652205e-05, "loss": 1.9484437561035157, "step": 111950 }, { "epoch": 0.44558714560654056, "grad_norm": 0.3832285404205322, "learning_rate": 1.5545605003302117e-05, "loss": 1.981368865966797, "step": 112000 }, { "epoch": 0.44578606843940066, "grad_norm": 0.5974204540252686, "learning_rate": 1.554361577695203e-05, "loss": 2.03849853515625, "step": 112050 }, { "epoch": 0.4459849912722607, "grad_norm": 0.5175960063934326, "learning_rate": 1.554162655060194e-05, "loss": 2.023445739746094, "step": 112100 }, { "epoch": 0.44618391410512076, "grad_norm": 0.43435049057006836, "learning_rate": 1.5539637324251853e-05, "loss": 2.004272155761719, "step": 112150 }, { "epoch": 0.4463828369379808, "grad_norm": 0.5803554058074951, "learning_rate": 1.5537648097901765e-05, "loss": 2.023629608154297, "step": 112200 }, { "epoch": 0.4465817597708409, "grad_norm": 0.5266780257225037, "learning_rate": 1.5535658871551677e-05, "loss": 1.9808819580078125, "step": 112250 }, { "epoch": 0.44678068260370096, "grad_norm": 0.4696264863014221, "learning_rate": 1.553366964520159e-05, "loss": 2.0315789794921875, "step": 112300 }, { "epoch": 0.446979605436561, "grad_norm": 0.5486732721328735, "learning_rate": 1.55316804188515e-05, "loss": 2.0079464721679687, "step": 112350 }, { "epoch": 0.44717852826942106, "grad_norm": 0.5706218481063843, "learning_rate": 1.5529691192501414e-05, "loss": 2.009522247314453, "step": 112400 }, { "epoch": 0.44737745110228117, "grad_norm": 0.6112673282623291, "learning_rate": 1.5527701966151326e-05, "loss": 2.0396868896484377, "step": 112450 }, { "epoch": 0.4475763739351412, "grad_norm": 0.5647121667861938, "learning_rate": 1.5525712739801238e-05, "loss": 2.0467678833007814, "step": 112500 }, { "epoch": 0.44777529676800126, "grad_norm": 0.5723282694816589, "learning_rate": 1.552376329797815e-05, "loss": 2.073511505126953, "step": 112550 }, { "epoch": 0.4479742196008613, "grad_norm": 0.5481998920440674, "learning_rate": 1.5521774071628064e-05, "loss": 2.005071258544922, "step": 112600 }, { "epoch": 0.4481731424337214, "grad_norm": 0.5475088357925415, "learning_rate": 1.5519784845277976e-05, "loss": 2.0140423583984375, "step": 112650 }, { "epoch": 0.44837206526658147, "grad_norm": 0.5576059222221375, "learning_rate": 1.5517795618927888e-05, "loss": 2.0225926208496094, "step": 112700 }, { "epoch": 0.4485709880994415, "grad_norm": 0.5908367037773132, "learning_rate": 1.55158063925778e-05, "loss": 2.022996520996094, "step": 112750 }, { "epoch": 0.44876991093230156, "grad_norm": 0.5828393697738647, "learning_rate": 1.5513817166227712e-05, "loss": 2.0443072509765625, "step": 112800 }, { "epoch": 0.44896883376516167, "grad_norm": 0.5908666849136353, "learning_rate": 1.5511827939877624e-05, "loss": 2.0245626831054686, "step": 112850 }, { "epoch": 0.4491677565980217, "grad_norm": 0.5329924821853638, "learning_rate": 1.5509838713527537e-05, "loss": 2.0209780883789064, "step": 112900 }, { "epoch": 0.44936667943088177, "grad_norm": 0.5077747106552124, "learning_rate": 1.550784948717745e-05, "loss": 2.0276637268066406, "step": 112950 }, { "epoch": 0.4495656022637418, "grad_norm": 0.6954885721206665, "learning_rate": 1.550586026082736e-05, "loss": 2.0161782836914064, "step": 113000 }, { "epoch": 0.4497645250966019, "grad_norm": 0.5542087554931641, "learning_rate": 1.550387103447727e-05, "loss": 2.0161282348632814, "step": 113050 }, { "epoch": 0.44996344792946197, "grad_norm": 0.5808271169662476, "learning_rate": 1.5501881808127185e-05, "loss": 2.0148828125, "step": 113100 }, { "epoch": 0.450162370762322, "grad_norm": 0.5562579035758972, "learning_rate": 1.5499892581777097e-05, "loss": 1.9324615478515625, "step": 113150 }, { "epoch": 0.45036129359518207, "grad_norm": 0.8085353374481201, "learning_rate": 1.549790335542701e-05, "loss": 2.014761962890625, "step": 113200 }, { "epoch": 0.45056021642804217, "grad_norm": 0.6270598769187927, "learning_rate": 1.549591412907692e-05, "loss": 1.997276611328125, "step": 113250 }, { "epoch": 0.4507591392609022, "grad_norm": 0.5781266689300537, "learning_rate": 1.5493924902726834e-05, "loss": 2.0288656616210936, "step": 113300 }, { "epoch": 0.45095806209376227, "grad_norm": 0.5581923127174377, "learning_rate": 1.5491935676376746e-05, "loss": 2.0019883728027343, "step": 113350 }, { "epoch": 0.4511569849266223, "grad_norm": 0.607671856880188, "learning_rate": 1.5489946450026655e-05, "loss": 2.0448097229003905, "step": 113400 }, { "epoch": 0.4513559077594824, "grad_norm": 0.6135187149047852, "learning_rate": 1.5487957223676567e-05, "loss": 2.0292564392089845, "step": 113450 }, { "epoch": 0.45155483059234247, "grad_norm": 0.5683545470237732, "learning_rate": 1.5485967997326482e-05, "loss": 2.0007945251464845, "step": 113500 }, { "epoch": 0.4517537534252025, "grad_norm": 0.5596336126327515, "learning_rate": 1.5483978770976395e-05, "loss": 2.0085720825195312, "step": 113550 }, { "epoch": 0.45195267625806257, "grad_norm": 0.6634269952774048, "learning_rate": 1.5481989544626307e-05, "loss": 2.028948211669922, "step": 113600 }, { "epoch": 0.4521515990909227, "grad_norm": 0.6676499843597412, "learning_rate": 1.548000031827622e-05, "loss": 1.974613494873047, "step": 113650 }, { "epoch": 0.4523505219237827, "grad_norm": 0.5247558355331421, "learning_rate": 1.5478011091926128e-05, "loss": 1.9967716979980468, "step": 113700 }, { "epoch": 0.45254944475664277, "grad_norm": 0.6229835748672485, "learning_rate": 1.547602186557604e-05, "loss": 2.0071812438964844, "step": 113750 }, { "epoch": 0.4527483675895028, "grad_norm": 0.5431439876556396, "learning_rate": 1.5474032639225952e-05, "loss": 2.0084828186035155, "step": 113800 }, { "epoch": 0.4529472904223629, "grad_norm": 0.6090958714485168, "learning_rate": 1.5472043412875864e-05, "loss": 2.05111328125, "step": 113850 }, { "epoch": 0.453146213255223, "grad_norm": 0.6227231025695801, "learning_rate": 1.547005418652578e-05, "loss": 1.994132080078125, "step": 113900 }, { "epoch": 0.453345136088083, "grad_norm": 0.5136319398880005, "learning_rate": 1.5468064960175692e-05, "loss": 1.9968701171875, "step": 113950 }, { "epoch": 0.45354405892094307, "grad_norm": 0.6165713667869568, "learning_rate": 1.5466075733825604e-05, "loss": 2.0241932678222656, "step": 114000 }, { "epoch": 0.4537429817538032, "grad_norm": 0.4827836751937866, "learning_rate": 1.5464086507475513e-05, "loss": 1.9948190307617188, "step": 114050 }, { "epoch": 0.4539419045866632, "grad_norm": 0.5787628889083862, "learning_rate": 1.5462097281125425e-05, "loss": 1.9781779479980468, "step": 114100 }, { "epoch": 0.4541408274195233, "grad_norm": 0.624032735824585, "learning_rate": 1.5460108054775337e-05, "loss": 2.026113739013672, "step": 114150 }, { "epoch": 0.4543397502523833, "grad_norm": 0.6396898031234741, "learning_rate": 1.545811882842525e-05, "loss": 2.0453060913085936, "step": 114200 }, { "epoch": 0.4545386730852434, "grad_norm": 0.5855856537818909, "learning_rate": 1.545612960207516e-05, "loss": 2.033893585205078, "step": 114250 }, { "epoch": 0.4547375959181035, "grad_norm": 0.5689274072647095, "learning_rate": 1.5454140375725077e-05, "loss": 1.9813778686523438, "step": 114300 }, { "epoch": 0.4549365187509635, "grad_norm": 0.46510693430900574, "learning_rate": 1.5452151149374985e-05, "loss": 2.00052490234375, "step": 114350 }, { "epoch": 0.4551354415838236, "grad_norm": 0.5393950939178467, "learning_rate": 1.5450161923024898e-05, "loss": 2.0075914001464845, "step": 114400 }, { "epoch": 0.4553343644166837, "grad_norm": 0.4858758747577667, "learning_rate": 1.544817269667481e-05, "loss": 2.0446852111816405, "step": 114450 }, { "epoch": 0.45553328724954373, "grad_norm": 0.514686644077301, "learning_rate": 1.5446183470324722e-05, "loss": 1.966446075439453, "step": 114500 }, { "epoch": 0.4557322100824038, "grad_norm": 0.6452842950820923, "learning_rate": 1.5444194243974634e-05, "loss": 2.03128173828125, "step": 114550 }, { "epoch": 0.4559311329152638, "grad_norm": 0.6265659332275391, "learning_rate": 1.5442205017624546e-05, "loss": 1.988644561767578, "step": 114600 }, { "epoch": 0.45613005574812393, "grad_norm": 0.5857723355293274, "learning_rate": 1.544021579127446e-05, "loss": 1.9370611572265626, "step": 114650 }, { "epoch": 0.456328978580984, "grad_norm": 0.5866117477416992, "learning_rate": 1.543822656492437e-05, "loss": 2.0051222229003907, "step": 114700 }, { "epoch": 0.45652790141384403, "grad_norm": 0.5059932470321655, "learning_rate": 1.5436237338574283e-05, "loss": 2.03925048828125, "step": 114750 }, { "epoch": 0.4567268242467041, "grad_norm": 0.707645833492279, "learning_rate": 1.5434248112224195e-05, "loss": 2.0041249084472654, "step": 114800 }, { "epoch": 0.4569257470795642, "grad_norm": 0.5247070789337158, "learning_rate": 1.5432258885874107e-05, "loss": 1.977176971435547, "step": 114850 }, { "epoch": 0.45712466991242423, "grad_norm": 0.5751223564147949, "learning_rate": 1.543026965952402e-05, "loss": 1.9644808959960938, "step": 114900 }, { "epoch": 0.4573235927452843, "grad_norm": 0.631942868232727, "learning_rate": 1.542828043317393e-05, "loss": 2.0666766357421875, "step": 114950 }, { "epoch": 0.45752251557814433, "grad_norm": 0.6040699481964111, "learning_rate": 1.5426291206823843e-05, "loss": 2.021493225097656, "step": 115000 }, { "epoch": 0.45772143841100443, "grad_norm": 0.5711373090744019, "learning_rate": 1.5424301980473756e-05, "loss": 1.9686737060546875, "step": 115050 }, { "epoch": 0.4579203612438645, "grad_norm": 0.5187176465988159, "learning_rate": 1.5422312754123668e-05, "loss": 1.9778033447265626, "step": 115100 }, { "epoch": 0.45811928407672453, "grad_norm": 0.543050229549408, "learning_rate": 1.542032352777358e-05, "loss": 1.9688432312011719, "step": 115150 }, { "epoch": 0.4583182069095846, "grad_norm": 0.5528223514556885, "learning_rate": 1.5418334301423492e-05, "loss": 1.9986297607421875, "step": 115200 }, { "epoch": 0.4585171297424447, "grad_norm": 0.6575373411178589, "learning_rate": 1.5416345075073404e-05, "loss": 2.0111669921875, "step": 115250 }, { "epoch": 0.45871605257530473, "grad_norm": 0.5372204184532166, "learning_rate": 1.5414355848723316e-05, "loss": 1.9863081359863282, "step": 115300 }, { "epoch": 0.4589149754081648, "grad_norm": 0.7277659177780151, "learning_rate": 1.541236662237323e-05, "loss": 2.0258517456054688, "step": 115350 }, { "epoch": 0.45911389824102483, "grad_norm": 0.6485069394111633, "learning_rate": 1.541037739602314e-05, "loss": 1.9664781188964844, "step": 115400 }, { "epoch": 0.45931282107388494, "grad_norm": 0.6892268657684326, "learning_rate": 1.5408388169673053e-05, "loss": 2.024394683837891, "step": 115450 }, { "epoch": 0.459511743906745, "grad_norm": 0.6627073287963867, "learning_rate": 1.5406398943322965e-05, "loss": 2.0699644470214844, "step": 115500 }, { "epoch": 0.45971066673960503, "grad_norm": 0.5595639944076538, "learning_rate": 1.5404409716972877e-05, "loss": 2.0557621765136718, "step": 115550 }, { "epoch": 0.4599095895724651, "grad_norm": 0.561562180519104, "learning_rate": 1.540242049062279e-05, "loss": 1.9931936645507813, "step": 115600 }, { "epoch": 0.4601085124053252, "grad_norm": 0.537118673324585, "learning_rate": 1.5400431264272698e-05, "loss": 1.9429457092285156, "step": 115650 }, { "epoch": 0.46030743523818524, "grad_norm": 0.6607091426849365, "learning_rate": 1.5398442037922613e-05, "loss": 2.064627380371094, "step": 115700 }, { "epoch": 0.4605063580710453, "grad_norm": 0.546396017074585, "learning_rate": 1.5396452811572526e-05, "loss": 2.0554058837890623, "step": 115750 }, { "epoch": 0.46070528090390533, "grad_norm": 0.5927671194076538, "learning_rate": 1.5394463585222438e-05, "loss": 1.952802734375, "step": 115800 }, { "epoch": 0.46090420373676544, "grad_norm": 0.6105004549026489, "learning_rate": 1.539247435887235e-05, "loss": 2.008922271728516, "step": 115850 }, { "epoch": 0.4611031265696255, "grad_norm": 0.5273284316062927, "learning_rate": 1.5390485132522262e-05, "loss": 1.9505427551269532, "step": 115900 }, { "epoch": 0.46130204940248554, "grad_norm": 0.6077753305435181, "learning_rate": 1.538849590617217e-05, "loss": 1.94720458984375, "step": 115950 }, { "epoch": 0.4615009722353456, "grad_norm": 0.5908426642417908, "learning_rate": 1.5386506679822083e-05, "loss": 1.9579161071777345, "step": 116000 }, { "epoch": 0.4616998950682057, "grad_norm": 0.37486979365348816, "learning_rate": 1.5384517453471995e-05, "loss": 2.0259330749511717, "step": 116050 }, { "epoch": 0.46189881790106574, "grad_norm": 0.5658749938011169, "learning_rate": 1.538252822712191e-05, "loss": 1.9625686645507812, "step": 116100 }, { "epoch": 0.4620977407339258, "grad_norm": 0.6101911664009094, "learning_rate": 1.5380539000771823e-05, "loss": 2.035344696044922, "step": 116150 }, { "epoch": 0.46229666356678584, "grad_norm": 0.5694454312324524, "learning_rate": 1.5378549774421735e-05, "loss": 2.02791748046875, "step": 116200 }, { "epoch": 0.46249558639964594, "grad_norm": 0.5627511143684387, "learning_rate": 1.5376560548071647e-05, "loss": 2.0077809143066405, "step": 116250 }, { "epoch": 0.462694509232506, "grad_norm": 0.5999167561531067, "learning_rate": 1.5374571321721556e-05, "loss": 2.017319641113281, "step": 116300 }, { "epoch": 0.46289343206536604, "grad_norm": 0.5911429524421692, "learning_rate": 1.5372582095371468e-05, "loss": 2.039628143310547, "step": 116350 }, { "epoch": 0.4630923548982261, "grad_norm": 0.5233595371246338, "learning_rate": 1.537059286902138e-05, "loss": 1.95648193359375, "step": 116400 }, { "epoch": 0.4632912777310862, "grad_norm": 0.5229326486587524, "learning_rate": 1.5368603642671292e-05, "loss": 2.000879669189453, "step": 116450 }, { "epoch": 0.46349020056394624, "grad_norm": 0.5039988160133362, "learning_rate": 1.5366614416321208e-05, "loss": 1.9973828125, "step": 116500 }, { "epoch": 0.4636891233968063, "grad_norm": 0.5729622840881348, "learning_rate": 1.536462518997112e-05, "loss": 2.013366394042969, "step": 116550 }, { "epoch": 0.46388804622966634, "grad_norm": 0.6142416596412659, "learning_rate": 1.536263596362103e-05, "loss": 2.0277914428710937, "step": 116600 }, { "epoch": 0.46408696906252644, "grad_norm": 0.5725548267364502, "learning_rate": 1.536064673727094e-05, "loss": 2.0184695434570314, "step": 116650 }, { "epoch": 0.4642858918953865, "grad_norm": 0.5302380323410034, "learning_rate": 1.5358657510920853e-05, "loss": 1.990430145263672, "step": 116700 }, { "epoch": 0.46448481472824654, "grad_norm": 0.6113941073417664, "learning_rate": 1.5356668284570765e-05, "loss": 2.013508148193359, "step": 116750 }, { "epoch": 0.4646837375611066, "grad_norm": 0.555452823638916, "learning_rate": 1.5354679058220677e-05, "loss": 1.995169677734375, "step": 116800 }, { "epoch": 0.4648826603939667, "grad_norm": 0.7587475180625916, "learning_rate": 1.535268983187059e-05, "loss": 2.03577392578125, "step": 116850 }, { "epoch": 0.46508158322682674, "grad_norm": 0.6018828749656677, "learning_rate": 1.53507006055205e-05, "loss": 2.0491123962402344, "step": 116900 }, { "epoch": 0.4652805060596868, "grad_norm": 0.6112669110298157, "learning_rate": 1.5348711379170414e-05, "loss": 2.0488636779785154, "step": 116950 }, { "epoch": 0.46547942889254684, "grad_norm": 0.5767244696617126, "learning_rate": 1.5346722152820326e-05, "loss": 2.03196044921875, "step": 117000 }, { "epoch": 0.46567835172540695, "grad_norm": 0.5196667313575745, "learning_rate": 1.5344732926470238e-05, "loss": 1.9776998901367187, "step": 117050 }, { "epoch": 0.465877274558267, "grad_norm": 0.5763862133026123, "learning_rate": 1.534274370012015e-05, "loss": 2.003811950683594, "step": 117100 }, { "epoch": 0.46607619739112705, "grad_norm": 0.6376035213470459, "learning_rate": 1.5340754473770062e-05, "loss": 2.001146240234375, "step": 117150 }, { "epoch": 0.4662751202239871, "grad_norm": 0.6196942329406738, "learning_rate": 1.5338765247419974e-05, "loss": 1.9895262145996093, "step": 117200 }, { "epoch": 0.4664740430568472, "grad_norm": 0.6088504791259766, "learning_rate": 1.5336776021069887e-05, "loss": 2.0240104675292967, "step": 117250 }, { "epoch": 0.46667296588970725, "grad_norm": 0.6030307412147522, "learning_rate": 1.53347867947198e-05, "loss": 2.047058410644531, "step": 117300 }, { "epoch": 0.4668718887225673, "grad_norm": 0.5700313448905945, "learning_rate": 1.533279756836971e-05, "loss": 1.9933218383789062, "step": 117350 }, { "epoch": 0.46707081155542735, "grad_norm": 0.5926457643508911, "learning_rate": 1.5330808342019623e-05, "loss": 2.0435955810546873, "step": 117400 }, { "epoch": 0.46726973438828745, "grad_norm": 0.5202028155326843, "learning_rate": 1.5328819115669535e-05, "loss": 2.0034748840332033, "step": 117450 }, { "epoch": 0.4674686572211475, "grad_norm": 0.5803148150444031, "learning_rate": 1.5326829889319447e-05, "loss": 2.001894226074219, "step": 117500 }, { "epoch": 0.46766758005400755, "grad_norm": 0.4871901571750641, "learning_rate": 1.532484066296936e-05, "loss": 2.0505055236816405, "step": 117550 }, { "epoch": 0.4678665028868676, "grad_norm": 0.5709880590438843, "learning_rate": 1.532285143661927e-05, "loss": 2.070085906982422, "step": 117600 }, { "epoch": 0.4680654257197277, "grad_norm": 0.5320698022842407, "learning_rate": 1.5320862210269184e-05, "loss": 2.0384371948242186, "step": 117650 }, { "epoch": 0.46826434855258775, "grad_norm": 0.6344123482704163, "learning_rate": 1.5318872983919096e-05, "loss": 2.069898223876953, "step": 117700 }, { "epoch": 0.4684632713854478, "grad_norm": 0.4955126941204071, "learning_rate": 1.5316883757569008e-05, "loss": 2.0554856872558593, "step": 117750 }, { "epoch": 0.46866219421830785, "grad_norm": 0.6440271139144897, "learning_rate": 1.531489453121892e-05, "loss": 2.0062713623046875, "step": 117800 }, { "epoch": 0.46886111705116795, "grad_norm": 0.7260912656784058, "learning_rate": 1.5312905304868832e-05, "loss": 2.0478050231933596, "step": 117850 }, { "epoch": 0.469060039884028, "grad_norm": 0.5656759142875671, "learning_rate": 1.531091607851874e-05, "loss": 2.0424876403808594, "step": 117900 }, { "epoch": 0.46925896271688805, "grad_norm": 0.5776940584182739, "learning_rate": 1.5308926852168657e-05, "loss": 1.9510107421875, "step": 117950 }, { "epoch": 0.4694578855497481, "grad_norm": 0.6011480689048767, "learning_rate": 1.530693762581857e-05, "loss": 2.07463134765625, "step": 118000 }, { "epoch": 0.4696568083826082, "grad_norm": 0.5359660983085632, "learning_rate": 1.530494839946848e-05, "loss": 2.000857391357422, "step": 118050 }, { "epoch": 0.46985573121546825, "grad_norm": 0.580488920211792, "learning_rate": 1.5302959173118393e-05, "loss": 1.9796746826171876, "step": 118100 }, { "epoch": 0.4700546540483283, "grad_norm": 0.6538369059562683, "learning_rate": 1.5300969946768305e-05, "loss": 1.9994252014160157, "step": 118150 }, { "epoch": 0.47025357688118835, "grad_norm": 0.5884948372840881, "learning_rate": 1.5298980720418217e-05, "loss": 2.0283245849609375, "step": 118200 }, { "epoch": 0.47045249971404846, "grad_norm": 0.3713313043117523, "learning_rate": 1.5296991494068126e-05, "loss": 2.058824157714844, "step": 118250 }, { "epoch": 0.4706514225469085, "grad_norm": 0.605590283870697, "learning_rate": 1.5295002267718038e-05, "loss": 1.98261474609375, "step": 118300 }, { "epoch": 0.47085034537976855, "grad_norm": 0.6727908849716187, "learning_rate": 1.5293013041367954e-05, "loss": 2.0534275817871093, "step": 118350 }, { "epoch": 0.4710492682126286, "grad_norm": 0.6045511960983276, "learning_rate": 1.5291023815017866e-05, "loss": 2.0060107421875, "step": 118400 }, { "epoch": 0.47124819104548865, "grad_norm": 0.5037333369255066, "learning_rate": 1.5289034588667778e-05, "loss": 1.980158233642578, "step": 118450 }, { "epoch": 0.47144711387834876, "grad_norm": 0.48281586170196533, "learning_rate": 1.528704536231769e-05, "loss": 1.9870930480957032, "step": 118500 }, { "epoch": 0.4716460367112088, "grad_norm": 0.5590367317199707, "learning_rate": 1.52850561359676e-05, "loss": 1.993220672607422, "step": 118550 }, { "epoch": 0.47184495954406885, "grad_norm": 0.4943028390407562, "learning_rate": 1.528306690961751e-05, "loss": 2.0389675903320312, "step": 118600 }, { "epoch": 0.4720438823769289, "grad_norm": 0.7419543862342834, "learning_rate": 1.5281077683267423e-05, "loss": 2.0478807067871094, "step": 118650 }, { "epoch": 0.472242805209789, "grad_norm": 0.5199508666992188, "learning_rate": 1.5279088456917335e-05, "loss": 2.012337341308594, "step": 118700 }, { "epoch": 0.47244172804264906, "grad_norm": 0.6479163765907288, "learning_rate": 1.527709923056725e-05, "loss": 1.9897271728515624, "step": 118750 }, { "epoch": 0.4726406508755091, "grad_norm": 0.4132716655731201, "learning_rate": 1.5275110004217163e-05, "loss": 2.1013824462890627, "step": 118800 }, { "epoch": 0.47283957370836915, "grad_norm": 0.5090117454528809, "learning_rate": 1.5273120777867075e-05, "loss": 2.006774139404297, "step": 118850 }, { "epoch": 0.47303849654122926, "grad_norm": 0.46223363280296326, "learning_rate": 1.5271131551516984e-05, "loss": 2.03974365234375, "step": 118900 }, { "epoch": 0.4732374193740893, "grad_norm": 0.49077877402305603, "learning_rate": 1.5269142325166896e-05, "loss": 1.9707333374023437, "step": 118950 }, { "epoch": 0.47343634220694936, "grad_norm": 0.6314084529876709, "learning_rate": 1.5267153098816808e-05, "loss": 2.007742156982422, "step": 119000 }, { "epoch": 0.4736352650398094, "grad_norm": 0.6636348366737366, "learning_rate": 1.526516387246672e-05, "loss": 1.9664743041992188, "step": 119050 }, { "epoch": 0.4738341878726695, "grad_norm": 0.5514074563980103, "learning_rate": 1.5263174646116633e-05, "loss": 2.0342510986328124, "step": 119100 }, { "epoch": 0.47403311070552956, "grad_norm": 0.636132538318634, "learning_rate": 1.5261185419766548e-05, "loss": 2.0551307678222654, "step": 119150 }, { "epoch": 0.4742320335383896, "grad_norm": 0.46318429708480835, "learning_rate": 1.5259196193416457e-05, "loss": 1.988714141845703, "step": 119200 }, { "epoch": 0.47443095637124966, "grad_norm": 0.5830183029174805, "learning_rate": 1.5257206967066369e-05, "loss": 1.9664311218261719, "step": 119250 }, { "epoch": 0.47462987920410976, "grad_norm": 0.5900910496711731, "learning_rate": 1.5255217740716281e-05, "loss": 2.0022230529785157, "step": 119300 }, { "epoch": 0.4748288020369698, "grad_norm": 0.5182344317436218, "learning_rate": 1.5253228514366193e-05, "loss": 2.0140443420410157, "step": 119350 }, { "epoch": 0.47502772486982986, "grad_norm": 0.6062299013137817, "learning_rate": 1.5251239288016107e-05, "loss": 2.066478729248047, "step": 119400 }, { "epoch": 0.4752266477026899, "grad_norm": 0.6884866952896118, "learning_rate": 1.524928984619302e-05, "loss": 2.0410108947753907, "step": 119450 }, { "epoch": 0.47542557053555, "grad_norm": 0.668714702129364, "learning_rate": 1.5247340404369934e-05, "loss": 1.9530819702148436, "step": 119500 }, { "epoch": 0.47562449336841006, "grad_norm": 0.6146720051765442, "learning_rate": 1.5245390962546846e-05, "loss": 1.9697383117675782, "step": 119550 }, { "epoch": 0.4758234162012701, "grad_norm": 0.6310422420501709, "learning_rate": 1.524340173619676e-05, "loss": 1.9515463256835937, "step": 119600 }, { "epoch": 0.47602233903413016, "grad_norm": 0.558236837387085, "learning_rate": 1.5241412509846672e-05, "loss": 1.96174072265625, "step": 119650 }, { "epoch": 0.47622126186699026, "grad_norm": 0.5744043588638306, "learning_rate": 1.5239423283496584e-05, "loss": 2.0178321838378905, "step": 119700 }, { "epoch": 0.4764201846998503, "grad_norm": 0.7195176482200623, "learning_rate": 1.5237434057146496e-05, "loss": 1.997615203857422, "step": 119750 }, { "epoch": 0.47661910753271036, "grad_norm": 0.6143149137496948, "learning_rate": 1.5235444830796408e-05, "loss": 1.9793466186523438, "step": 119800 }, { "epoch": 0.4768180303655704, "grad_norm": 0.5038840174674988, "learning_rate": 1.5233455604446319e-05, "loss": 2.015872802734375, "step": 119850 }, { "epoch": 0.4770169531984305, "grad_norm": 0.565538227558136, "learning_rate": 1.5231466378096231e-05, "loss": 1.9619978332519532, "step": 119900 }, { "epoch": 0.47721587603129056, "grad_norm": 0.44865578413009644, "learning_rate": 1.5229477151746143e-05, "loss": 1.9830290222167968, "step": 119950 }, { "epoch": 0.4774147988641506, "grad_norm": 0.6705788373947144, "learning_rate": 1.5227487925396057e-05, "loss": 2.0270143127441407, "step": 120000 }, { "epoch": 0.47761372169701066, "grad_norm": 0.5602022409439087, "learning_rate": 1.5225498699045969e-05, "loss": 1.9906643676757811, "step": 120050 }, { "epoch": 0.47781264452987077, "grad_norm": 0.6182267069816589, "learning_rate": 1.5223509472695881e-05, "loss": 1.9939065551757813, "step": 120100 }, { "epoch": 0.4780115673627308, "grad_norm": 0.6775335669517517, "learning_rate": 1.5221520246345793e-05, "loss": 2.0424703979492187, "step": 120150 }, { "epoch": 0.47821049019559086, "grad_norm": 0.6986251473426819, "learning_rate": 1.5219531019995704e-05, "loss": 2.0100192260742187, "step": 120200 }, { "epoch": 0.4784094130284509, "grad_norm": 0.6762583255767822, "learning_rate": 1.5217541793645616e-05, "loss": 2.002889404296875, "step": 120250 }, { "epoch": 0.478608335861311, "grad_norm": 0.5059555768966675, "learning_rate": 1.5215552567295528e-05, "loss": 1.9817611694335937, "step": 120300 }, { "epoch": 0.47880725869417107, "grad_norm": 0.7047236561775208, "learning_rate": 1.521356334094544e-05, "loss": 2.0505047607421876, "step": 120350 }, { "epoch": 0.4790061815270311, "grad_norm": 0.6243566870689392, "learning_rate": 1.5211574114595352e-05, "loss": 1.9847987365722657, "step": 120400 }, { "epoch": 0.47920510435989117, "grad_norm": 0.4750196039676666, "learning_rate": 1.5209584888245266e-05, "loss": 1.9495933532714844, "step": 120450 }, { "epoch": 0.47940402719275127, "grad_norm": 0.5757802724838257, "learning_rate": 1.5207595661895175e-05, "loss": 1.9601972961425782, "step": 120500 }, { "epoch": 0.4796029500256113, "grad_norm": 0.6709091067314148, "learning_rate": 1.5205606435545089e-05, "loss": 2.013679962158203, "step": 120550 }, { "epoch": 0.47980187285847137, "grad_norm": 0.5989776849746704, "learning_rate": 1.5203617209195001e-05, "loss": 1.96046630859375, "step": 120600 }, { "epoch": 0.4800007956913314, "grad_norm": 0.564067542552948, "learning_rate": 1.5201627982844913e-05, "loss": 2.044736328125, "step": 120650 }, { "epoch": 0.4801997185241915, "grad_norm": 0.5669559240341187, "learning_rate": 1.5199638756494825e-05, "loss": 1.9583908081054688, "step": 120700 }, { "epoch": 0.48039864135705157, "grad_norm": 0.5996476411819458, "learning_rate": 1.5197649530144737e-05, "loss": 2.0073696899414064, "step": 120750 }, { "epoch": 0.4805975641899116, "grad_norm": 0.6494384407997131, "learning_rate": 1.519566030379465e-05, "loss": 2.0111433410644532, "step": 120800 }, { "epoch": 0.48079648702277167, "grad_norm": 0.5995691418647766, "learning_rate": 1.519367107744456e-05, "loss": 1.9725445556640624, "step": 120850 }, { "epoch": 0.4809954098556318, "grad_norm": 0.5745535492897034, "learning_rate": 1.5191681851094472e-05, "loss": 2.038809814453125, "step": 120900 }, { "epoch": 0.4811943326884918, "grad_norm": 0.529074490070343, "learning_rate": 1.5189692624744386e-05, "loss": 1.995853729248047, "step": 120950 }, { "epoch": 0.48139325552135187, "grad_norm": 0.5489996671676636, "learning_rate": 1.5187703398394298e-05, "loss": 1.9870339965820312, "step": 121000 }, { "epoch": 0.4815921783542119, "grad_norm": 0.574969470500946, "learning_rate": 1.518571417204421e-05, "loss": 2.0200018310546874, "step": 121050 }, { "epoch": 0.481791101187072, "grad_norm": 0.7172936201095581, "learning_rate": 1.5183724945694122e-05, "loss": 1.9999441528320312, "step": 121100 }, { "epoch": 0.4819900240199321, "grad_norm": 0.6878817677497864, "learning_rate": 1.5181735719344033e-05, "loss": 2.0579522705078124, "step": 121150 }, { "epoch": 0.4821889468527921, "grad_norm": 0.5850234627723694, "learning_rate": 1.5179746492993945e-05, "loss": 1.9664533996582032, "step": 121200 }, { "epoch": 0.48238786968565217, "grad_norm": 0.6747224926948547, "learning_rate": 1.5177757266643857e-05, "loss": 2.0184002685546876, "step": 121250 }, { "epoch": 0.4825867925185123, "grad_norm": 0.5963588953018188, "learning_rate": 1.517576804029377e-05, "loss": 1.9692254638671876, "step": 121300 }, { "epoch": 0.4827857153513723, "grad_norm": 0.6176520586013794, "learning_rate": 1.5173778813943683e-05, "loss": 2.0289459228515625, "step": 121350 }, { "epoch": 0.4829846381842324, "grad_norm": 0.6718717813491821, "learning_rate": 1.5171789587593595e-05, "loss": 1.994828643798828, "step": 121400 }, { "epoch": 0.4831835610170924, "grad_norm": 0.5374060869216919, "learning_rate": 1.5169800361243508e-05, "loss": 2.013385772705078, "step": 121450 }, { "epoch": 0.4833824838499525, "grad_norm": 0.5552113056182861, "learning_rate": 1.5167811134893418e-05, "loss": 1.9984938049316405, "step": 121500 }, { "epoch": 0.4835814066828126, "grad_norm": 0.5204220414161682, "learning_rate": 1.516582190854333e-05, "loss": 2.006961669921875, "step": 121550 }, { "epoch": 0.4837803295156726, "grad_norm": 0.6060960292816162, "learning_rate": 1.5163832682193242e-05, "loss": 2.026741638183594, "step": 121600 }, { "epoch": 0.4839792523485327, "grad_norm": 0.5550637245178223, "learning_rate": 1.5161843455843154e-05, "loss": 2.0085462951660156, "step": 121650 }, { "epoch": 0.4841781751813928, "grad_norm": 0.5062645673751831, "learning_rate": 1.5159854229493067e-05, "loss": 2.0195248413085936, "step": 121700 }, { "epoch": 0.4843770980142528, "grad_norm": 0.6046434640884399, "learning_rate": 1.515786500314298e-05, "loss": 2.0273896789550783, "step": 121750 }, { "epoch": 0.4845760208471129, "grad_norm": 0.5392206907272339, "learning_rate": 1.5155875776792889e-05, "loss": 1.9476104736328126, "step": 121800 }, { "epoch": 0.4847749436799729, "grad_norm": 0.5917559862136841, "learning_rate": 1.5153886550442803e-05, "loss": 1.9907534790039063, "step": 121850 }, { "epoch": 0.48497386651283303, "grad_norm": 0.6204864382743835, "learning_rate": 1.5151897324092715e-05, "loss": 1.9513897705078125, "step": 121900 }, { "epoch": 0.4851727893456931, "grad_norm": 0.5994875431060791, "learning_rate": 1.5149908097742627e-05, "loss": 2.0254534912109374, "step": 121950 }, { "epoch": 0.4853717121785531, "grad_norm": 0.48806101083755493, "learning_rate": 1.514791887139254e-05, "loss": 1.970347900390625, "step": 122000 }, { "epoch": 0.4855706350114132, "grad_norm": 0.6405678987503052, "learning_rate": 1.5145929645042452e-05, "loss": 2.036766357421875, "step": 122050 }, { "epoch": 0.4857695578442733, "grad_norm": 0.5723554491996765, "learning_rate": 1.5143940418692364e-05, "loss": 1.9996214294433594, "step": 122100 }, { "epoch": 0.48596848067713333, "grad_norm": 0.570720911026001, "learning_rate": 1.5141951192342274e-05, "loss": 2.0017042541503907, "step": 122150 }, { "epoch": 0.4861674035099934, "grad_norm": 0.5530833005905151, "learning_rate": 1.5139961965992186e-05, "loss": 2.0260569763183596, "step": 122200 }, { "epoch": 0.48636632634285343, "grad_norm": 0.545456051826477, "learning_rate": 1.51379727396421e-05, "loss": 2.0498829650878907, "step": 122250 }, { "epoch": 0.48656524917571353, "grad_norm": 0.5242537260055542, "learning_rate": 1.5135983513292012e-05, "loss": 1.9424998474121093, "step": 122300 }, { "epoch": 0.4867641720085736, "grad_norm": 0.6003217697143555, "learning_rate": 1.5133994286941924e-05, "loss": 1.993332977294922, "step": 122350 }, { "epoch": 0.48696309484143363, "grad_norm": 0.605130136013031, "learning_rate": 1.5132005060591837e-05, "loss": 2.001386871337891, "step": 122400 }, { "epoch": 0.4871620176742937, "grad_norm": 0.5328429937362671, "learning_rate": 1.5130015834241747e-05, "loss": 2.0344435119628907, "step": 122450 }, { "epoch": 0.4873609405071538, "grad_norm": 0.5181903839111328, "learning_rate": 1.512802660789166e-05, "loss": 2.0720533752441406, "step": 122500 }, { "epoch": 0.48755986334001383, "grad_norm": 0.6673620939254761, "learning_rate": 1.5126037381541571e-05, "loss": 2.0175125122070314, "step": 122550 }, { "epoch": 0.4877587861728739, "grad_norm": 0.7206003665924072, "learning_rate": 1.5124048155191483e-05, "loss": 1.9842411804199218, "step": 122600 }, { "epoch": 0.48795770900573393, "grad_norm": 0.6803132891654968, "learning_rate": 1.5122058928841397e-05, "loss": 2.0189695739746094, "step": 122650 }, { "epoch": 0.48815663183859404, "grad_norm": 0.5490893125534058, "learning_rate": 1.512006970249131e-05, "loss": 2.0097671508789063, "step": 122700 }, { "epoch": 0.4883555546714541, "grad_norm": 0.49237656593322754, "learning_rate": 1.5118080476141222e-05, "loss": 2.0136338806152345, "step": 122750 }, { "epoch": 0.48855447750431413, "grad_norm": 0.5630783438682556, "learning_rate": 1.5116091249791132e-05, "loss": 2.013887939453125, "step": 122800 }, { "epoch": 0.4887534003371742, "grad_norm": 0.5865453481674194, "learning_rate": 1.5114102023441044e-05, "loss": 1.9847630310058593, "step": 122850 }, { "epoch": 0.4889523231700343, "grad_norm": 0.5304304361343384, "learning_rate": 1.5112112797090956e-05, "loss": 2.0080471801757813, "step": 122900 }, { "epoch": 0.48915124600289434, "grad_norm": 0.6462113857269287, "learning_rate": 1.5110123570740869e-05, "loss": 1.9779347229003905, "step": 122950 }, { "epoch": 0.4893501688357544, "grad_norm": 0.6900922656059265, "learning_rate": 1.510813434439078e-05, "loss": 1.9976190185546876, "step": 123000 }, { "epoch": 0.48954909166861443, "grad_norm": 0.6204034090042114, "learning_rate": 1.5106145118040694e-05, "loss": 1.9807014465332031, "step": 123050 }, { "epoch": 0.48974801450147454, "grad_norm": 0.6066051125526428, "learning_rate": 1.5104155891690603e-05, "loss": 1.999208984375, "step": 123100 }, { "epoch": 0.4899469373343346, "grad_norm": 0.5012942552566528, "learning_rate": 1.5102166665340517e-05, "loss": 1.9743547058105468, "step": 123150 }, { "epoch": 0.49014586016719464, "grad_norm": 0.529010534286499, "learning_rate": 1.510017743899043e-05, "loss": 2.0136813354492187, "step": 123200 }, { "epoch": 0.4903447830000547, "grad_norm": 0.6548493504524231, "learning_rate": 1.5098188212640341e-05, "loss": 2.025182800292969, "step": 123250 }, { "epoch": 0.4905437058329148, "grad_norm": 0.4886799156665802, "learning_rate": 1.5096198986290254e-05, "loss": 1.9360169982910156, "step": 123300 }, { "epoch": 0.49074262866577484, "grad_norm": 0.6066966652870178, "learning_rate": 1.5094209759940166e-05, "loss": 1.9927218627929688, "step": 123350 }, { "epoch": 0.4909415514986349, "grad_norm": 0.6144793033599854, "learning_rate": 1.5092220533590078e-05, "loss": 2.007617950439453, "step": 123400 }, { "epoch": 0.49114047433149494, "grad_norm": 0.5209155678749084, "learning_rate": 1.5090231307239988e-05, "loss": 1.952456512451172, "step": 123450 }, { "epoch": 0.49133939716435504, "grad_norm": 0.6380859613418579, "learning_rate": 1.50882420808899e-05, "loss": 1.9978668212890625, "step": 123500 }, { "epoch": 0.4915383199972151, "grad_norm": 0.6001052856445312, "learning_rate": 1.5086252854539814e-05, "loss": 2.030561981201172, "step": 123550 }, { "epoch": 0.49173724283007514, "grad_norm": 0.4886865019798279, "learning_rate": 1.5084263628189726e-05, "loss": 2.023209686279297, "step": 123600 }, { "epoch": 0.4919361656629352, "grad_norm": 0.6343898773193359, "learning_rate": 1.5082274401839639e-05, "loss": 2.0072648620605467, "step": 123650 }, { "epoch": 0.4921350884957953, "grad_norm": 0.6556450724601746, "learning_rate": 1.508028517548955e-05, "loss": 2.0070091247558595, "step": 123700 }, { "epoch": 0.49233401132865534, "grad_norm": 0.57780921459198, "learning_rate": 1.5078295949139461e-05, "loss": 1.9956446838378907, "step": 123750 }, { "epoch": 0.4925329341615154, "grad_norm": 0.49793538451194763, "learning_rate": 1.5076306722789373e-05, "loss": 2.0190090942382812, "step": 123800 }, { "epoch": 0.49273185699437544, "grad_norm": 0.5418640971183777, "learning_rate": 1.5074317496439285e-05, "loss": 2.0391812133789062, "step": 123850 }, { "epoch": 0.49293077982723554, "grad_norm": 0.594974160194397, "learning_rate": 1.5072328270089198e-05, "loss": 1.9980746459960939, "step": 123900 }, { "epoch": 0.4931297026600956, "grad_norm": 0.5745349526405334, "learning_rate": 1.5070339043739111e-05, "loss": 1.9589830017089844, "step": 123950 }, { "epoch": 0.49332862549295564, "grad_norm": 0.5786569118499756, "learning_rate": 1.5068349817389024e-05, "loss": 2.011634521484375, "step": 124000 }, { "epoch": 0.4935275483258157, "grad_norm": 0.43753132224082947, "learning_rate": 1.5066360591038936e-05, "loss": 1.9961003112792968, "step": 124050 }, { "epoch": 0.4937264711586758, "grad_norm": 0.5427939891815186, "learning_rate": 1.5064371364688846e-05, "loss": 1.9843499755859375, "step": 124100 }, { "epoch": 0.49392539399153584, "grad_norm": 0.6055757403373718, "learning_rate": 1.5062382138338758e-05, "loss": 2.032040863037109, "step": 124150 }, { "epoch": 0.4941243168243959, "grad_norm": 0.5476160645484924, "learning_rate": 1.506039291198867e-05, "loss": 2.01540771484375, "step": 124200 }, { "epoch": 0.49432323965725594, "grad_norm": 0.4812210202217102, "learning_rate": 1.5058403685638583e-05, "loss": 1.990279998779297, "step": 124250 }, { "epoch": 0.49452216249011605, "grad_norm": 0.6026652455329895, "learning_rate": 1.5056414459288495e-05, "loss": 2.012262268066406, "step": 124300 }, { "epoch": 0.4947210853229761, "grad_norm": 0.5194963812828064, "learning_rate": 1.5054425232938407e-05, "loss": 2.024120330810547, "step": 124350 }, { "epoch": 0.49492000815583614, "grad_norm": 0.5816293358802795, "learning_rate": 1.5052436006588317e-05, "loss": 1.9972348022460937, "step": 124400 }, { "epoch": 0.4951189309886962, "grad_norm": 0.5216043591499329, "learning_rate": 1.5050446780238231e-05, "loss": 1.9515586853027345, "step": 124450 }, { "epoch": 0.4953178538215563, "grad_norm": 0.6754438877105713, "learning_rate": 1.5048457553888143e-05, "loss": 1.9966836547851563, "step": 124500 }, { "epoch": 0.49551677665441635, "grad_norm": 0.5765421986579895, "learning_rate": 1.5046468327538055e-05, "loss": 2.0378021240234374, "step": 124550 }, { "epoch": 0.4957156994872764, "grad_norm": 0.4878929853439331, "learning_rate": 1.5044479101187968e-05, "loss": 1.9770425415039063, "step": 124600 }, { "epoch": 0.49591462232013644, "grad_norm": 0.626633882522583, "learning_rate": 1.504248987483788e-05, "loss": 1.975751190185547, "step": 124650 }, { "epoch": 0.49611354515299655, "grad_norm": 0.5916438698768616, "learning_rate": 1.5040500648487792e-05, "loss": 1.993057861328125, "step": 124700 }, { "epoch": 0.4963124679858566, "grad_norm": 0.5193774700164795, "learning_rate": 1.5038511422137702e-05, "loss": 1.9684600830078125, "step": 124750 }, { "epoch": 0.49651139081871665, "grad_norm": 0.5676165819168091, "learning_rate": 1.5036522195787615e-05, "loss": 1.9965475463867188, "step": 124800 }, { "epoch": 0.4967103136515767, "grad_norm": 0.4924023151397705, "learning_rate": 1.5034532969437527e-05, "loss": 2.0356820678710936, "step": 124850 }, { "epoch": 0.4969092364844368, "grad_norm": 0.6230697631835938, "learning_rate": 1.503254374308744e-05, "loss": 1.990164794921875, "step": 124900 }, { "epoch": 0.49710815931729685, "grad_norm": 0.6673582196235657, "learning_rate": 1.5030554516737353e-05, "loss": 1.9731561279296874, "step": 124950 }, { "epoch": 0.4973070821501569, "grad_norm": 0.6718992590904236, "learning_rate": 1.5028565290387265e-05, "loss": 2.0053717041015626, "step": 125000 }, { "epoch": 0.49750600498301695, "grad_norm": 0.5679284334182739, "learning_rate": 1.5026576064037175e-05, "loss": 2.0349700927734373, "step": 125050 }, { "epoch": 0.49770492781587705, "grad_norm": 0.5974433422088623, "learning_rate": 1.5024586837687087e-05, "loss": 1.9259333801269531, "step": 125100 }, { "epoch": 0.4979038506487371, "grad_norm": 0.5383091568946838, "learning_rate": 1.5022597611337e-05, "loss": 1.955035400390625, "step": 125150 }, { "epoch": 0.49810277348159715, "grad_norm": 0.4752386808395386, "learning_rate": 1.5020608384986912e-05, "loss": 2.001916809082031, "step": 125200 }, { "epoch": 0.4983016963144572, "grad_norm": 0.6498311758041382, "learning_rate": 1.5018619158636824e-05, "loss": 1.9966950988769532, "step": 125250 }, { "epoch": 0.4985006191473173, "grad_norm": 0.5524863600730896, "learning_rate": 1.5016629932286738e-05, "loss": 1.9466862487792969, "step": 125300 }, { "epoch": 0.49869954198017735, "grad_norm": 0.5083037614822388, "learning_rate": 1.5014640705936646e-05, "loss": 1.9771136474609374, "step": 125350 }, { "epoch": 0.4988984648130374, "grad_norm": 0.5422106981277466, "learning_rate": 1.501265147958656e-05, "loss": 2.0619606018066405, "step": 125400 }, { "epoch": 0.49909738764589745, "grad_norm": 0.5339271426200867, "learning_rate": 1.5010662253236472e-05, "loss": 1.9646644592285156, "step": 125450 }, { "epoch": 0.49929631047875755, "grad_norm": 0.621643602848053, "learning_rate": 1.5008673026886385e-05, "loss": 1.9979812622070312, "step": 125500 }, { "epoch": 0.4994952333116176, "grad_norm": 0.6140711307525635, "learning_rate": 1.5006683800536297e-05, "loss": 2.0344552612304687, "step": 125550 }, { "epoch": 0.49969415614447765, "grad_norm": 0.5683419704437256, "learning_rate": 1.5004694574186209e-05, "loss": 2.017011871337891, "step": 125600 }, { "epoch": 0.4998930789773377, "grad_norm": 0.6096767783164978, "learning_rate": 1.5002745132363124e-05, "loss": 1.9777096557617186, "step": 125650 }, { "epoch": 0.5000920018101978, "grad_norm": 0.6325658559799194, "learning_rate": 1.5000755906013035e-05, "loss": 1.9500326538085937, "step": 125700 }, { "epoch": 0.5002909246430578, "grad_norm": 0.579384982585907, "learning_rate": 1.4998766679662947e-05, "loss": 1.9243038940429686, "step": 125750 }, { "epoch": 0.500489847475918, "grad_norm": 0.5836588740348816, "learning_rate": 1.4996777453312859e-05, "loss": 1.9972352600097656, "step": 125800 }, { "epoch": 0.500688770308778, "grad_norm": 0.6821098327636719, "learning_rate": 1.4994788226962771e-05, "loss": 1.995061492919922, "step": 125850 }, { "epoch": 0.5008876931416381, "grad_norm": 0.6649511456489563, "learning_rate": 1.4992799000612683e-05, "loss": 1.9868539428710938, "step": 125900 }, { "epoch": 0.5010866159744981, "grad_norm": 0.5975461602210999, "learning_rate": 1.4990809774262595e-05, "loss": 1.99765869140625, "step": 125950 }, { "epoch": 0.5012855388073582, "grad_norm": 0.5714362859725952, "learning_rate": 1.4988820547912507e-05, "loss": 1.9852468872070312, "step": 126000 }, { "epoch": 0.5014844616402182, "grad_norm": 0.5635392665863037, "learning_rate": 1.4986831321562418e-05, "loss": 1.9655429077148439, "step": 126050 }, { "epoch": 0.5016833844730783, "grad_norm": 0.5981415510177612, "learning_rate": 1.498484209521233e-05, "loss": 2.007593231201172, "step": 126100 }, { "epoch": 0.5018823073059383, "grad_norm": 0.5869321823120117, "learning_rate": 1.4982852868862244e-05, "loss": 1.9299591064453125, "step": 126150 }, { "epoch": 0.5020812301387984, "grad_norm": 0.5060520172119141, "learning_rate": 1.4980863642512156e-05, "loss": 2.0271360778808596, "step": 126200 }, { "epoch": 0.5022801529716585, "grad_norm": 0.5561806559562683, "learning_rate": 1.4978874416162068e-05, "loss": 1.9851162719726563, "step": 126250 }, { "epoch": 0.5024790758045186, "grad_norm": 0.519168496131897, "learning_rate": 1.497688518981198e-05, "loss": 2.0185470581054688, "step": 126300 }, { "epoch": 0.5026779986373786, "grad_norm": 0.5834574103355408, "learning_rate": 1.497489596346189e-05, "loss": 2.042341613769531, "step": 126350 }, { "epoch": 0.5028769214702387, "grad_norm": 0.6435027122497559, "learning_rate": 1.4972906737111803e-05, "loss": 2.0000570678710936, "step": 126400 }, { "epoch": 0.5030758443030987, "grad_norm": 0.6259173154830933, "learning_rate": 1.4970917510761715e-05, "loss": 1.9955043029785156, "step": 126450 }, { "epoch": 0.5032747671359588, "grad_norm": 0.5850337743759155, "learning_rate": 1.4968928284411627e-05, "loss": 1.9843165588378906, "step": 126500 }, { "epoch": 0.5034736899688188, "grad_norm": 0.513281524181366, "learning_rate": 1.4966939058061541e-05, "loss": 1.9979618835449218, "step": 126550 }, { "epoch": 0.5036726128016789, "grad_norm": 0.5664961934089661, "learning_rate": 1.4964989616238454e-05, "loss": 1.9913479614257812, "step": 126600 }, { "epoch": 0.503871535634539, "grad_norm": 0.5761082768440247, "learning_rate": 1.4963000389888367e-05, "loss": 1.9748837280273437, "step": 126650 }, { "epoch": 0.5040704584673991, "grad_norm": 0.596825897693634, "learning_rate": 1.4961011163538277e-05, "loss": 1.9515478515625, "step": 126700 }, { "epoch": 0.5042693813002591, "grad_norm": 0.4814811646938324, "learning_rate": 1.495902193718819e-05, "loss": 1.9637080383300782, "step": 126750 }, { "epoch": 0.5044683041331192, "grad_norm": 0.42759400606155396, "learning_rate": 1.4957032710838101e-05, "loss": 1.9427635192871093, "step": 126800 }, { "epoch": 0.5046672269659792, "grad_norm": 0.5888792872428894, "learning_rate": 1.4955043484488014e-05, "loss": 2.0124858093261717, "step": 126850 }, { "epoch": 0.5048661497988393, "grad_norm": 0.516567051410675, "learning_rate": 1.4953054258137927e-05, "loss": 2.001048126220703, "step": 126900 }, { "epoch": 0.5050650726316993, "grad_norm": 0.5356466174125671, "learning_rate": 1.495106503178784e-05, "loss": 1.9778439331054687, "step": 126950 }, { "epoch": 0.5052639954645594, "grad_norm": 0.6518732309341431, "learning_rate": 1.494907580543775e-05, "loss": 1.9814804077148438, "step": 127000 }, { "epoch": 0.5054629182974195, "grad_norm": 0.5219168066978455, "learning_rate": 1.4947086579087662e-05, "loss": 1.9612405395507813, "step": 127050 }, { "epoch": 0.5056618411302796, "grad_norm": 0.5582583546638489, "learning_rate": 1.4945097352737574e-05, "loss": 2.019362030029297, "step": 127100 }, { "epoch": 0.5058607639631396, "grad_norm": 0.4864736497402191, "learning_rate": 1.4943108126387486e-05, "loss": 2.030221710205078, "step": 127150 }, { "epoch": 0.5060596867959997, "grad_norm": 0.5451263785362244, "learning_rate": 1.4941118900037399e-05, "loss": 1.9710427856445312, "step": 127200 }, { "epoch": 0.5062586096288597, "grad_norm": 0.5567352771759033, "learning_rate": 1.493912967368731e-05, "loss": 2.010750427246094, "step": 127250 }, { "epoch": 0.5064575324617198, "grad_norm": 0.5572960376739502, "learning_rate": 1.4937140447337225e-05, "loss": 2.022824554443359, "step": 127300 }, { "epoch": 0.5066564552945798, "grad_norm": 0.6221963167190552, "learning_rate": 1.4935151220987133e-05, "loss": 2.0157977294921876, "step": 127350 }, { "epoch": 0.5068553781274399, "grad_norm": 0.5189698934555054, "learning_rate": 1.4933161994637047e-05, "loss": 1.9754673767089843, "step": 127400 }, { "epoch": 0.5070543009603, "grad_norm": 0.5474683046340942, "learning_rate": 1.493117276828696e-05, "loss": 1.9454388427734375, "step": 127450 }, { "epoch": 0.5072532237931601, "grad_norm": 0.5744040608406067, "learning_rate": 1.4929183541936871e-05, "loss": 2.063819580078125, "step": 127500 }, { "epoch": 0.5074521466260201, "grad_norm": 0.6755750179290771, "learning_rate": 1.4927194315586784e-05, "loss": 2.0373298645019533, "step": 127550 }, { "epoch": 0.5076510694588802, "grad_norm": 0.5589731931686401, "learning_rate": 1.4925205089236696e-05, "loss": 1.9316207885742187, "step": 127600 }, { "epoch": 0.5078499922917402, "grad_norm": 0.5950993895530701, "learning_rate": 1.4923215862886606e-05, "loss": 2.0027006530761717, "step": 127650 }, { "epoch": 0.5080489151246003, "grad_norm": 0.5288292765617371, "learning_rate": 1.4921226636536518e-05, "loss": 1.9856666564941405, "step": 127700 }, { "epoch": 0.5082478379574603, "grad_norm": 0.6159432530403137, "learning_rate": 1.491923741018643e-05, "loss": 1.987227325439453, "step": 127750 }, { "epoch": 0.5084467607903204, "grad_norm": 0.5709269046783447, "learning_rate": 1.4917248183836344e-05, "loss": 1.9776307678222655, "step": 127800 }, { "epoch": 0.5086456836231805, "grad_norm": 0.5882406234741211, "learning_rate": 1.4915258957486256e-05, "loss": 1.9394586181640625, "step": 127850 }, { "epoch": 0.5088446064560406, "grad_norm": 0.5328643321990967, "learning_rate": 1.4913269731136169e-05, "loss": 1.966687469482422, "step": 127900 }, { "epoch": 0.5090435292889006, "grad_norm": 0.5327330231666565, "learning_rate": 1.491128050478608e-05, "loss": 1.97440673828125, "step": 127950 }, { "epoch": 0.5092424521217607, "grad_norm": 0.5361045002937317, "learning_rate": 1.4909291278435991e-05, "loss": 1.9715805053710938, "step": 128000 }, { "epoch": 0.5094413749546207, "grad_norm": 0.5783485770225525, "learning_rate": 1.4907302052085903e-05, "loss": 2.040179443359375, "step": 128050 }, { "epoch": 0.5096402977874808, "grad_norm": 0.574751615524292, "learning_rate": 1.4905312825735815e-05, "loss": 1.947931365966797, "step": 128100 }, { "epoch": 0.5098392206203408, "grad_norm": 0.5803755521774292, "learning_rate": 1.4903323599385728e-05, "loss": 1.9439540100097656, "step": 128150 }, { "epoch": 0.5100381434532009, "grad_norm": 0.5408809781074524, "learning_rate": 1.4901334373035641e-05, "loss": 2.0050480651855467, "step": 128200 }, { "epoch": 0.510237066286061, "grad_norm": 0.4827142357826233, "learning_rate": 1.4899345146685554e-05, "loss": 2.024405059814453, "step": 128250 }, { "epoch": 0.5104359891189211, "grad_norm": 0.7092207670211792, "learning_rate": 1.4897355920335464e-05, "loss": 1.9768658447265626, "step": 128300 }, { "epoch": 0.5106349119517811, "grad_norm": 0.6582323908805847, "learning_rate": 1.4895366693985376e-05, "loss": 2.004188232421875, "step": 128350 }, { "epoch": 0.5108338347846412, "grad_norm": 0.4728025197982788, "learning_rate": 1.4893377467635288e-05, "loss": 2.0233364868164063, "step": 128400 }, { "epoch": 0.5110327576175012, "grad_norm": 0.48956820368766785, "learning_rate": 1.48913882412852e-05, "loss": 1.9504214477539064, "step": 128450 }, { "epoch": 0.5112316804503613, "grad_norm": 0.41225332021713257, "learning_rate": 1.4889399014935113e-05, "loss": 1.9953717041015624, "step": 128500 }, { "epoch": 0.5114306032832213, "grad_norm": 0.6332764625549316, "learning_rate": 1.4887409788585025e-05, "loss": 1.9539927673339843, "step": 128550 }, { "epoch": 0.5116295261160814, "grad_norm": 0.623853325843811, "learning_rate": 1.4885420562234935e-05, "loss": 1.9747636413574219, "step": 128600 }, { "epoch": 0.5118284489489415, "grad_norm": 0.5979064702987671, "learning_rate": 1.4883431335884847e-05, "loss": 1.9866645812988282, "step": 128650 }, { "epoch": 0.5120273717818016, "grad_norm": 0.6652819514274597, "learning_rate": 1.4881442109534761e-05, "loss": 1.992015380859375, "step": 128700 }, { "epoch": 0.5122262946146616, "grad_norm": 0.5822145342826843, "learning_rate": 1.4879452883184673e-05, "loss": 1.9674400329589843, "step": 128750 }, { "epoch": 0.5124252174475217, "grad_norm": 0.5416576266288757, "learning_rate": 1.4877463656834586e-05, "loss": 1.9901864624023438, "step": 128800 }, { "epoch": 0.5126241402803817, "grad_norm": 0.7178792953491211, "learning_rate": 1.4875474430484498e-05, "loss": 1.9738601684570312, "step": 128850 }, { "epoch": 0.5128230631132418, "grad_norm": 0.5738387107849121, "learning_rate": 1.487348520413441e-05, "loss": 2.0006890869140626, "step": 128900 }, { "epoch": 0.5130219859461018, "grad_norm": 0.6235736608505249, "learning_rate": 1.487149597778432e-05, "loss": 1.9726350402832031, "step": 128950 }, { "epoch": 0.5132209087789619, "grad_norm": 0.5685728788375854, "learning_rate": 1.4869506751434232e-05, "loss": 2.0300115966796874, "step": 129000 }, { "epoch": 0.513419831611822, "grad_norm": 0.5758592486381531, "learning_rate": 1.4867517525084145e-05, "loss": 1.9777947998046874, "step": 129050 }, { "epoch": 0.5136187544446821, "grad_norm": 0.6598905324935913, "learning_rate": 1.4865528298734058e-05, "loss": 1.9246278381347657, "step": 129100 }, { "epoch": 0.5138176772775421, "grad_norm": 0.6160648465156555, "learning_rate": 1.486353907238397e-05, "loss": 2.0089151000976564, "step": 129150 }, { "epoch": 0.5140166001104022, "grad_norm": 0.5578486323356628, "learning_rate": 1.4861549846033883e-05, "loss": 1.964869384765625, "step": 129200 }, { "epoch": 0.5142155229432622, "grad_norm": 0.5222259163856506, "learning_rate": 1.4859560619683793e-05, "loss": 1.9635917663574218, "step": 129250 }, { "epoch": 0.5144144457761223, "grad_norm": 0.5413611531257629, "learning_rate": 1.4857571393333705e-05, "loss": 1.933607177734375, "step": 129300 }, { "epoch": 0.5146133686089823, "grad_norm": 0.5819974541664124, "learning_rate": 1.4855582166983617e-05, "loss": 2.021070709228516, "step": 129350 }, { "epoch": 0.5148122914418424, "grad_norm": 0.5361678004264832, "learning_rate": 1.485359294063353e-05, "loss": 2.013627777099609, "step": 129400 }, { "epoch": 0.5150112142747025, "grad_norm": 0.6237080693244934, "learning_rate": 1.4851603714283442e-05, "loss": 2.0620787048339846, "step": 129450 }, { "epoch": 0.5152101371075626, "grad_norm": 0.6864469647407532, "learning_rate": 1.4849614487933356e-05, "loss": 2.0118507385253905, "step": 129500 }, { "epoch": 0.5154090599404226, "grad_norm": 0.595526397228241, "learning_rate": 1.4847625261583268e-05, "loss": 1.9943182373046875, "step": 129550 }, { "epoch": 0.5156079827732827, "grad_norm": 0.6702523827552795, "learning_rate": 1.4845636035233178e-05, "loss": 1.9975996398925782, "step": 129600 }, { "epoch": 0.5158069056061427, "grad_norm": 0.6493624448776245, "learning_rate": 1.484364680888309e-05, "loss": 2.029694366455078, "step": 129650 }, { "epoch": 0.5160058284390028, "grad_norm": 0.5698150396347046, "learning_rate": 1.4841657582533002e-05, "loss": 1.9632968139648437, "step": 129700 }, { "epoch": 0.5162047512718628, "grad_norm": 0.5684245228767395, "learning_rate": 1.4839668356182915e-05, "loss": 2.0261940002441405, "step": 129750 }, { "epoch": 0.5164036741047229, "grad_norm": 0.7180051803588867, "learning_rate": 1.4837679129832827e-05, "loss": 1.9502864074707031, "step": 129800 }, { "epoch": 0.516602596937583, "grad_norm": 0.569007158279419, "learning_rate": 1.4835689903482739e-05, "loss": 1.9924517822265626, "step": 129850 }, { "epoch": 0.5168015197704431, "grad_norm": 0.5201748609542847, "learning_rate": 1.483370067713265e-05, "loss": 1.9686106872558593, "step": 129900 }, { "epoch": 0.5170004426033031, "grad_norm": 0.6179969906806946, "learning_rate": 1.4831711450782562e-05, "loss": 1.926834259033203, "step": 129950 }, { "epoch": 0.5171993654361632, "grad_norm": 0.6190645694732666, "learning_rate": 1.4829722224432475e-05, "loss": 1.9599459838867188, "step": 130000 }, { "epoch": 0.5173982882690232, "grad_norm": 0.5157051682472229, "learning_rate": 1.4827732998082387e-05, "loss": 1.9544602966308593, "step": 130050 }, { "epoch": 0.5175972111018833, "grad_norm": 0.5751699209213257, "learning_rate": 1.48257437717323e-05, "loss": 2.0140234375, "step": 130100 }, { "epoch": 0.5177961339347433, "grad_norm": 0.6000940799713135, "learning_rate": 1.4823754545382212e-05, "loss": 1.9568466186523437, "step": 130150 }, { "epoch": 0.5179950567676034, "grad_norm": 0.59727942943573, "learning_rate": 1.4821765319032124e-05, "loss": 1.9519349670410155, "step": 130200 }, { "epoch": 0.5181939796004635, "grad_norm": 0.5943633913993835, "learning_rate": 1.4819776092682034e-05, "loss": 1.963807373046875, "step": 130250 }, { "epoch": 0.5183929024333236, "grad_norm": 0.6574764847755432, "learning_rate": 1.4817786866331947e-05, "loss": 1.9898019409179688, "step": 130300 }, { "epoch": 0.5185918252661836, "grad_norm": 0.5902690887451172, "learning_rate": 1.4815797639981859e-05, "loss": 1.9925393676757812, "step": 130350 }, { "epoch": 0.5187907480990437, "grad_norm": 0.5806618332862854, "learning_rate": 1.4813808413631773e-05, "loss": 1.9539178466796876, "step": 130400 }, { "epoch": 0.5189896709319037, "grad_norm": 0.5255290269851685, "learning_rate": 1.4811819187281685e-05, "loss": 1.98798095703125, "step": 130450 }, { "epoch": 0.5191885937647638, "grad_norm": 0.600732684135437, "learning_rate": 1.4809829960931597e-05, "loss": 2.0971102905273438, "step": 130500 }, { "epoch": 0.5193875165976238, "grad_norm": 0.5609550476074219, "learning_rate": 1.4807840734581507e-05, "loss": 2.004536895751953, "step": 130550 }, { "epoch": 0.5195864394304839, "grad_norm": 0.6039653420448303, "learning_rate": 1.480585150823142e-05, "loss": 1.9703262329101563, "step": 130600 }, { "epoch": 0.519785362263344, "grad_norm": 0.6970183849334717, "learning_rate": 1.4803862281881332e-05, "loss": 2.0178509521484376, "step": 130650 }, { "epoch": 0.5199842850962041, "grad_norm": 0.6386555433273315, "learning_rate": 1.4801873055531244e-05, "loss": 1.9913540649414063, "step": 130700 }, { "epoch": 0.5201832079290641, "grad_norm": 0.642580509185791, "learning_rate": 1.4799883829181156e-05, "loss": 1.9392813110351563, "step": 130750 }, { "epoch": 0.5203821307619242, "grad_norm": 0.5395886301994324, "learning_rate": 1.4797894602831068e-05, "loss": 1.9406341552734374, "step": 130800 }, { "epoch": 0.5205810535947842, "grad_norm": 0.5723294615745544, "learning_rate": 1.4795945161007983e-05, "loss": 1.98349853515625, "step": 130850 }, { "epoch": 0.5207799764276443, "grad_norm": 0.48793742060661316, "learning_rate": 1.4793955934657894e-05, "loss": 1.9705335998535156, "step": 130900 }, { "epoch": 0.5209788992605043, "grad_norm": 0.7099360227584839, "learning_rate": 1.4791966708307806e-05, "loss": 2.0020651245117187, "step": 130950 }, { "epoch": 0.5211778220933644, "grad_norm": 0.546596348285675, "learning_rate": 1.4789977481957718e-05, "loss": 2.0194183349609376, "step": 131000 }, { "epoch": 0.5213767449262245, "grad_norm": 0.5808358788490295, "learning_rate": 1.478798825560763e-05, "loss": 1.9563215637207032, "step": 131050 }, { "epoch": 0.5215756677590846, "grad_norm": 0.6210852861404419, "learning_rate": 1.4785999029257542e-05, "loss": 1.9849336242675781, "step": 131100 }, { "epoch": 0.5217745905919446, "grad_norm": 0.589349091053009, "learning_rate": 1.4784009802907456e-05, "loss": 1.9976992797851563, "step": 131150 }, { "epoch": 0.5219735134248047, "grad_norm": 0.4721739888191223, "learning_rate": 1.4782020576557365e-05, "loss": 1.9785841369628907, "step": 131200 }, { "epoch": 0.5221724362576647, "grad_norm": 0.6477406024932861, "learning_rate": 1.4780031350207279e-05, "loss": 1.9972348022460937, "step": 131250 }, { "epoch": 0.5223713590905248, "grad_norm": 0.5659725069999695, "learning_rate": 1.477804212385719e-05, "loss": 2.0311038208007814, "step": 131300 }, { "epoch": 0.5225702819233848, "grad_norm": 0.7181436419487, "learning_rate": 1.4776052897507103e-05, "loss": 1.9963134765625, "step": 131350 }, { "epoch": 0.5227692047562449, "grad_norm": 0.5828695893287659, "learning_rate": 1.4774063671157015e-05, "loss": 1.9559744262695313, "step": 131400 }, { "epoch": 0.522968127589105, "grad_norm": 0.5842617154121399, "learning_rate": 1.4772074444806927e-05, "loss": 2.0223507690429687, "step": 131450 }, { "epoch": 0.5231670504219651, "grad_norm": 0.5923817753791809, "learning_rate": 1.477008521845684e-05, "loss": 1.99238037109375, "step": 131500 }, { "epoch": 0.5233659732548251, "grad_norm": 0.5582053661346436, "learning_rate": 1.476809599210675e-05, "loss": 2.001818084716797, "step": 131550 }, { "epoch": 0.5235648960876852, "grad_norm": 0.6175490021705627, "learning_rate": 1.4766106765756662e-05, "loss": 2.0132901000976564, "step": 131600 }, { "epoch": 0.5237638189205452, "grad_norm": 0.540733814239502, "learning_rate": 1.4764117539406576e-05, "loss": 1.9618717956542968, "step": 131650 }, { "epoch": 0.5239627417534053, "grad_norm": 0.6595100164413452, "learning_rate": 1.4762128313056488e-05, "loss": 2.029294891357422, "step": 131700 }, { "epoch": 0.5241616645862653, "grad_norm": 0.563015341758728, "learning_rate": 1.47601390867064e-05, "loss": 1.966921844482422, "step": 131750 }, { "epoch": 0.5243605874191254, "grad_norm": 0.49210280179977417, "learning_rate": 1.4758149860356312e-05, "loss": 1.967769775390625, "step": 131800 }, { "epoch": 0.5245595102519856, "grad_norm": 0.5969565510749817, "learning_rate": 1.4756160634006223e-05, "loss": 1.9524075317382812, "step": 131850 }, { "epoch": 0.5247584330848456, "grad_norm": 0.5526189208030701, "learning_rate": 1.4754171407656135e-05, "loss": 1.9779212951660157, "step": 131900 }, { "epoch": 0.5249573559177056, "grad_norm": 0.4748499095439911, "learning_rate": 1.4752182181306047e-05, "loss": 1.981031494140625, "step": 131950 }, { "epoch": 0.5251562787505657, "grad_norm": 0.5662875175476074, "learning_rate": 1.475019295495596e-05, "loss": 1.9967018127441407, "step": 132000 }, { "epoch": 0.5253552015834257, "grad_norm": 0.5441047549247742, "learning_rate": 1.4748203728605871e-05, "loss": 1.9921913146972656, "step": 132050 }, { "epoch": 0.5255541244162858, "grad_norm": 0.7287087440490723, "learning_rate": 1.4746214502255785e-05, "loss": 1.9669448852539062, "step": 132100 }, { "epoch": 0.5257530472491458, "grad_norm": 0.4369541108608246, "learning_rate": 1.4744225275905697e-05, "loss": 1.9709866333007813, "step": 132150 }, { "epoch": 0.5259519700820059, "grad_norm": 0.5137024521827698, "learning_rate": 1.4742236049555608e-05, "loss": 1.9900558471679688, "step": 132200 }, { "epoch": 0.526150892914866, "grad_norm": 0.6100850701332092, "learning_rate": 1.474024682320552e-05, "loss": 1.9965133666992188, "step": 132250 }, { "epoch": 0.5263498157477261, "grad_norm": 0.6621410250663757, "learning_rate": 1.4738257596855432e-05, "loss": 1.954756622314453, "step": 132300 }, { "epoch": 0.5265487385805862, "grad_norm": 0.5852276682853699, "learning_rate": 1.4736268370505344e-05, "loss": 1.9945814514160156, "step": 132350 }, { "epoch": 0.5267476614134462, "grad_norm": 0.7179412841796875, "learning_rate": 1.473431892868226e-05, "loss": 2.0250506591796875, "step": 132400 }, { "epoch": 0.5269465842463062, "grad_norm": 0.5737753510475159, "learning_rate": 1.4732329702332172e-05, "loss": 2.012256927490234, "step": 132450 }, { "epoch": 0.5271455070791663, "grad_norm": 0.604060173034668, "learning_rate": 1.4730340475982082e-05, "loss": 1.977412567138672, "step": 132500 }, { "epoch": 0.5273444299120263, "grad_norm": 0.5189964175224304, "learning_rate": 1.4728351249631994e-05, "loss": 1.9738845825195312, "step": 132550 }, { "epoch": 0.5275433527448864, "grad_norm": 0.6751134991645813, "learning_rate": 1.4726362023281906e-05, "loss": 1.9887925720214843, "step": 132600 }, { "epoch": 0.5277422755777466, "grad_norm": 0.753276526927948, "learning_rate": 1.4724372796931818e-05, "loss": 1.9890251159667969, "step": 132650 }, { "epoch": 0.5279411984106066, "grad_norm": 0.5941011905670166, "learning_rate": 1.472238357058173e-05, "loss": 2.0009494018554688, "step": 132700 }, { "epoch": 0.5281401212434667, "grad_norm": 0.6533281207084656, "learning_rate": 1.4720394344231643e-05, "loss": 1.9822232055664062, "step": 132750 }, { "epoch": 0.5283390440763267, "grad_norm": 0.5642436742782593, "learning_rate": 1.4718405117881555e-05, "loss": 1.9206057739257814, "step": 132800 }, { "epoch": 0.5285379669091868, "grad_norm": 0.5609648823738098, "learning_rate": 1.4716415891531465e-05, "loss": 1.95505126953125, "step": 132850 }, { "epoch": 0.5287368897420468, "grad_norm": 0.6359190344810486, "learning_rate": 1.4714426665181379e-05, "loss": 1.9976475524902344, "step": 132900 }, { "epoch": 0.5289358125749068, "grad_norm": 0.5324872136116028, "learning_rate": 1.4712437438831291e-05, "loss": 2.02072021484375, "step": 132950 }, { "epoch": 0.5291347354077669, "grad_norm": 0.4865109622478485, "learning_rate": 1.4710448212481203e-05, "loss": 1.9210769653320312, "step": 133000 }, { "epoch": 0.5293336582406271, "grad_norm": 0.427327424287796, "learning_rate": 1.4708458986131116e-05, "loss": 1.9867327880859376, "step": 133050 }, { "epoch": 0.5295325810734871, "grad_norm": 0.5497506260871887, "learning_rate": 1.4706469759781028e-05, "loss": 2.014210968017578, "step": 133100 }, { "epoch": 0.5297315039063472, "grad_norm": 0.551234781742096, "learning_rate": 1.4704480533430938e-05, "loss": 1.9829441833496093, "step": 133150 }, { "epoch": 0.5299304267392072, "grad_norm": 0.5917801260948181, "learning_rate": 1.470249130708085e-05, "loss": 1.9680262756347657, "step": 133200 }, { "epoch": 0.5301293495720673, "grad_norm": 0.607344925403595, "learning_rate": 1.4700502080730762e-05, "loss": 2.0243101501464844, "step": 133250 }, { "epoch": 0.5303282724049273, "grad_norm": 0.575570821762085, "learning_rate": 1.4698512854380675e-05, "loss": 1.984763641357422, "step": 133300 }, { "epoch": 0.5305271952377874, "grad_norm": 0.6283494234085083, "learning_rate": 1.4696523628030588e-05, "loss": 1.9788751220703125, "step": 133350 }, { "epoch": 0.5307261180706474, "grad_norm": 0.6219653487205505, "learning_rate": 1.46945344016805e-05, "loss": 2.0047982788085936, "step": 133400 }, { "epoch": 0.5309250409035076, "grad_norm": 0.34114065766334534, "learning_rate": 1.4692545175330413e-05, "loss": 1.9631491088867188, "step": 133450 }, { "epoch": 0.5311239637363676, "grad_norm": 0.5912941694259644, "learning_rate": 1.4690555948980323e-05, "loss": 1.9723040771484375, "step": 133500 }, { "epoch": 0.5313228865692277, "grad_norm": 0.6024045348167419, "learning_rate": 1.4688566722630235e-05, "loss": 1.9445436096191406, "step": 133550 }, { "epoch": 0.5315218094020877, "grad_norm": 0.6185288429260254, "learning_rate": 1.4686577496280147e-05, "loss": 2.0217599487304687, "step": 133600 }, { "epoch": 0.5317207322349478, "grad_norm": 0.5638116002082825, "learning_rate": 1.468458826993006e-05, "loss": 2.0274617004394533, "step": 133650 }, { "epoch": 0.5319196550678078, "grad_norm": 0.7013239860534668, "learning_rate": 1.4682599043579972e-05, "loss": 1.9954556274414061, "step": 133700 }, { "epoch": 0.5321185779006679, "grad_norm": 0.4496331810951233, "learning_rate": 1.4680609817229886e-05, "loss": 2.0036384582519533, "step": 133750 }, { "epoch": 0.5323175007335279, "grad_norm": 0.6370352506637573, "learning_rate": 1.4678620590879794e-05, "loss": 1.963730926513672, "step": 133800 }, { "epoch": 0.5325164235663881, "grad_norm": 0.5747882127761841, "learning_rate": 1.4676631364529708e-05, "loss": 2.0301820373535158, "step": 133850 }, { "epoch": 0.5327153463992481, "grad_norm": 0.5618709921836853, "learning_rate": 1.467464213817962e-05, "loss": 1.99250244140625, "step": 133900 }, { "epoch": 0.5329142692321082, "grad_norm": 0.6167848110198975, "learning_rate": 1.4672652911829533e-05, "loss": 1.9892292785644532, "step": 133950 }, { "epoch": 0.5331131920649682, "grad_norm": 0.6024337410926819, "learning_rate": 1.4670663685479445e-05, "loss": 1.9554312133789062, "step": 134000 }, { "epoch": 0.5333121148978283, "grad_norm": 0.6326452493667603, "learning_rate": 1.4668674459129357e-05, "loss": 1.9491639709472657, "step": 134050 }, { "epoch": 0.5335110377306883, "grad_norm": 0.5984023809432983, "learning_rate": 1.4666685232779269e-05, "loss": 1.942622528076172, "step": 134100 }, { "epoch": 0.5337099605635484, "grad_norm": 0.6101610660552979, "learning_rate": 1.466469600642918e-05, "loss": 1.9977430725097656, "step": 134150 }, { "epoch": 0.5339088833964084, "grad_norm": 0.6026674509048462, "learning_rate": 1.4662706780079092e-05, "loss": 1.9834896850585937, "step": 134200 }, { "epoch": 0.5341078062292685, "grad_norm": 0.6533790230751038, "learning_rate": 1.4660717553729005e-05, "loss": 2.024469451904297, "step": 134250 }, { "epoch": 0.5343067290621286, "grad_norm": 0.6190670132637024, "learning_rate": 1.4658728327378918e-05, "loss": 1.9874452209472657, "step": 134300 }, { "epoch": 0.5345056518949887, "grad_norm": 0.6430385708808899, "learning_rate": 1.465673910102883e-05, "loss": 2.0222247314453123, "step": 134350 }, { "epoch": 0.5347045747278487, "grad_norm": 0.5764914155006409, "learning_rate": 1.4654749874678742e-05, "loss": 2.0016848754882814, "step": 134400 }, { "epoch": 0.5349034975607088, "grad_norm": 0.6664733290672302, "learning_rate": 1.4652760648328652e-05, "loss": 2.019296875, "step": 134450 }, { "epoch": 0.5351024203935688, "grad_norm": 0.49503904581069946, "learning_rate": 1.4650771421978564e-05, "loss": 1.991785430908203, "step": 134500 }, { "epoch": 0.5353013432264289, "grad_norm": 0.6201068758964539, "learning_rate": 1.4648782195628477e-05, "loss": 2.0067445373535158, "step": 134550 }, { "epoch": 0.5355002660592889, "grad_norm": 0.5670273900032043, "learning_rate": 1.4646792969278389e-05, "loss": 1.972235107421875, "step": 134600 }, { "epoch": 0.535699188892149, "grad_norm": 0.6120335459709167, "learning_rate": 1.4644803742928303e-05, "loss": 1.9687515258789063, "step": 134650 }, { "epoch": 0.5358981117250091, "grad_norm": 0.556748628616333, "learning_rate": 1.4642814516578215e-05, "loss": 2.000363616943359, "step": 134700 }, { "epoch": 0.5360970345578692, "grad_norm": 0.6693964600563049, "learning_rate": 1.4640825290228127e-05, "loss": 2.0334588623046876, "step": 134750 }, { "epoch": 0.5362959573907292, "grad_norm": 0.5835266709327698, "learning_rate": 1.4638836063878037e-05, "loss": 1.997908172607422, "step": 134800 }, { "epoch": 0.5364948802235893, "grad_norm": 0.7948305606842041, "learning_rate": 1.463684683752795e-05, "loss": 1.9677151489257811, "step": 134850 }, { "epoch": 0.5366938030564493, "grad_norm": 0.5344545245170593, "learning_rate": 1.4634857611177862e-05, "loss": 2.0431138610839845, "step": 134900 }, { "epoch": 0.5368927258893094, "grad_norm": 0.6824910640716553, "learning_rate": 1.4632868384827774e-05, "loss": 2.0400340270996096, "step": 134950 }, { "epoch": 0.5370916487221694, "grad_norm": 0.5788838863372803, "learning_rate": 1.4630879158477686e-05, "loss": 1.96456298828125, "step": 135000 }, { "epoch": 0.5372905715550295, "grad_norm": 0.5422737002372742, "learning_rate": 1.46288899321276e-05, "loss": 2.0135731506347656, "step": 135050 }, { "epoch": 0.5374894943878896, "grad_norm": 0.3555559515953064, "learning_rate": 1.4626900705777508e-05, "loss": 2.0618849182128907, "step": 135100 }, { "epoch": 0.5376884172207497, "grad_norm": 0.5616377592086792, "learning_rate": 1.4624911479427422e-05, "loss": 1.9670208740234374, "step": 135150 }, { "epoch": 0.5378873400536097, "grad_norm": 0.563460648059845, "learning_rate": 1.4622922253077334e-05, "loss": 1.9643885803222656, "step": 135200 }, { "epoch": 0.5380862628864698, "grad_norm": 0.6297844052314758, "learning_rate": 1.4620933026727247e-05, "loss": 1.9435096740722657, "step": 135250 }, { "epoch": 0.5382851857193298, "grad_norm": 0.6132398247718811, "learning_rate": 1.4618943800377159e-05, "loss": 2.005664825439453, "step": 135300 }, { "epoch": 0.5384841085521899, "grad_norm": 0.5407642126083374, "learning_rate": 1.4616954574027071e-05, "loss": 1.9859210205078126, "step": 135350 }, { "epoch": 0.5386830313850499, "grad_norm": 0.6315252780914307, "learning_rate": 1.4614965347676983e-05, "loss": 2.0066571044921875, "step": 135400 }, { "epoch": 0.53888195421791, "grad_norm": 0.5404320359230042, "learning_rate": 1.4612976121326894e-05, "loss": 1.9502531433105468, "step": 135450 }, { "epoch": 0.5390808770507701, "grad_norm": 0.6665002107620239, "learning_rate": 1.4610986894976806e-05, "loss": 2.0094073486328123, "step": 135500 }, { "epoch": 0.5392797998836302, "grad_norm": 0.46084800362586975, "learning_rate": 1.460899766862672e-05, "loss": 2.003509826660156, "step": 135550 }, { "epoch": 0.5394787227164902, "grad_norm": 0.5623598098754883, "learning_rate": 1.4607008442276632e-05, "loss": 1.9912986755371094, "step": 135600 }, { "epoch": 0.5396776455493503, "grad_norm": 0.6555529236793518, "learning_rate": 1.4605019215926544e-05, "loss": 1.981546630859375, "step": 135650 }, { "epoch": 0.5398765683822103, "grad_norm": 0.5818672776222229, "learning_rate": 1.4603029989576456e-05, "loss": 2.0128118896484377, "step": 135700 }, { "epoch": 0.5400754912150704, "grad_norm": 0.7582321166992188, "learning_rate": 1.4601040763226366e-05, "loss": 1.9742747497558595, "step": 135750 }, { "epoch": 0.5402744140479304, "grad_norm": 0.5400150418281555, "learning_rate": 1.4599051536876279e-05, "loss": 2.0153358459472654, "step": 135800 }, { "epoch": 0.5404733368807905, "grad_norm": 0.6301628947257996, "learning_rate": 1.459706231052619e-05, "loss": 1.9462680053710937, "step": 135850 }, { "epoch": 0.5406722597136506, "grad_norm": 0.5248615741729736, "learning_rate": 1.4595073084176103e-05, "loss": 2.0052011108398435, "step": 135900 }, { "epoch": 0.5408711825465107, "grad_norm": 0.5035131573677063, "learning_rate": 1.4593083857826017e-05, "loss": 2.005443267822266, "step": 135950 }, { "epoch": 0.5410701053793707, "grad_norm": 0.5533472299575806, "learning_rate": 1.4591094631475929e-05, "loss": 1.9148539733886718, "step": 136000 }, { "epoch": 0.5412690282122308, "grad_norm": 0.573483943939209, "learning_rate": 1.4589105405125841e-05, "loss": 2.005330352783203, "step": 136050 }, { "epoch": 0.5414679510450908, "grad_norm": 0.5428409576416016, "learning_rate": 1.4587116178775751e-05, "loss": 1.99164794921875, "step": 136100 }, { "epoch": 0.5416668738779509, "grad_norm": 0.5949061512947083, "learning_rate": 1.4585126952425664e-05, "loss": 2.0351382446289064, "step": 136150 }, { "epoch": 0.5418657967108109, "grad_norm": 0.5457675457000732, "learning_rate": 1.4583137726075576e-05, "loss": 2.030759735107422, "step": 136200 }, { "epoch": 0.542064719543671, "grad_norm": 0.5681372284889221, "learning_rate": 1.4581148499725488e-05, "loss": 2.0200459289550783, "step": 136250 }, { "epoch": 0.5422636423765311, "grad_norm": 0.5369942784309387, "learning_rate": 1.45791592733754e-05, "loss": 1.9609197998046874, "step": 136300 }, { "epoch": 0.5424625652093912, "grad_norm": 0.5911480784416199, "learning_rate": 1.4577170047025314e-05, "loss": 1.999911346435547, "step": 136350 }, { "epoch": 0.5426614880422512, "grad_norm": 0.5662989616394043, "learning_rate": 1.4575180820675223e-05, "loss": 1.9610858154296875, "step": 136400 }, { "epoch": 0.5428604108751113, "grad_norm": 0.5717621445655823, "learning_rate": 1.4573191594325136e-05, "loss": 2.0279249572753906, "step": 136450 }, { "epoch": 0.5430593337079713, "grad_norm": 0.5674471259117126, "learning_rate": 1.4571202367975049e-05, "loss": 2.048404541015625, "step": 136500 }, { "epoch": 0.5432582565408314, "grad_norm": 0.5653378367424011, "learning_rate": 1.456921314162496e-05, "loss": 1.9343711853027343, "step": 136550 }, { "epoch": 0.5434571793736914, "grad_norm": 0.5057408809661865, "learning_rate": 1.4567223915274873e-05, "loss": 1.993405303955078, "step": 136600 }, { "epoch": 0.5436561022065515, "grad_norm": 0.5692144632339478, "learning_rate": 1.4565234688924785e-05, "loss": 1.9839974975585937, "step": 136650 }, { "epoch": 0.5438550250394116, "grad_norm": 0.646868884563446, "learning_rate": 1.4563245462574697e-05, "loss": 1.9621478271484376, "step": 136700 }, { "epoch": 0.5440539478722717, "grad_norm": 0.609025239944458, "learning_rate": 1.4561256236224608e-05, "loss": 1.9567347717285157, "step": 136750 }, { "epoch": 0.5442528707051317, "grad_norm": 0.5184829235076904, "learning_rate": 1.455926700987452e-05, "loss": 2.000554962158203, "step": 136800 }, { "epoch": 0.5444517935379918, "grad_norm": 0.5907589793205261, "learning_rate": 1.4557277783524434e-05, "loss": 1.9507606506347657, "step": 136850 }, { "epoch": 0.5446507163708518, "grad_norm": 0.5583916902542114, "learning_rate": 1.4555288557174346e-05, "loss": 1.9932830810546875, "step": 136900 }, { "epoch": 0.5448496392037119, "grad_norm": 0.6649031639099121, "learning_rate": 1.4553299330824258e-05, "loss": 2.0394248962402344, "step": 136950 }, { "epoch": 0.5450485620365719, "grad_norm": 0.6818561553955078, "learning_rate": 1.455131010447417e-05, "loss": 1.9402870178222655, "step": 137000 }, { "epoch": 0.545247484869432, "grad_norm": 0.59517502784729, "learning_rate": 1.454932087812408e-05, "loss": 1.985692901611328, "step": 137050 }, { "epoch": 0.5454464077022921, "grad_norm": 0.5143899321556091, "learning_rate": 1.4547331651773993e-05, "loss": 1.9383245849609374, "step": 137100 }, { "epoch": 0.5456453305351522, "grad_norm": 0.6013317704200745, "learning_rate": 1.4545342425423905e-05, "loss": 1.9235859680175782, "step": 137150 }, { "epoch": 0.5458442533680122, "grad_norm": 0.5658414363861084, "learning_rate": 1.4543353199073817e-05, "loss": 1.986146240234375, "step": 137200 }, { "epoch": 0.5460431762008723, "grad_norm": 0.3182540535926819, "learning_rate": 1.4541363972723729e-05, "loss": 2.001659240722656, "step": 137250 }, { "epoch": 0.5462420990337323, "grad_norm": 0.5981470942497253, "learning_rate": 1.4539374746373643e-05, "loss": 1.9964425659179688, "step": 137300 }, { "epoch": 0.5464410218665924, "grad_norm": 0.5644211173057556, "learning_rate": 1.4537385520023553e-05, "loss": 1.9998719787597656, "step": 137350 }, { "epoch": 0.5466399446994524, "grad_norm": 0.40475115180015564, "learning_rate": 1.4535396293673466e-05, "loss": 1.9609072875976563, "step": 137400 }, { "epoch": 0.5468388675323125, "grad_norm": 0.6047788262367249, "learning_rate": 1.4533407067323378e-05, "loss": 1.9591667175292968, "step": 137450 }, { "epoch": 0.5470377903651726, "grad_norm": 0.5791077017784119, "learning_rate": 1.453141784097329e-05, "loss": 1.991928253173828, "step": 137500 }, { "epoch": 0.5472367131980327, "grad_norm": 0.4628171920776367, "learning_rate": 1.4529468399150203e-05, "loss": 2.0593721008300783, "step": 137550 }, { "epoch": 0.5474356360308927, "grad_norm": 0.5670837759971619, "learning_rate": 1.4527479172800117e-05, "loss": 2.0237471008300782, "step": 137600 }, { "epoch": 0.5476345588637528, "grad_norm": 0.5295332670211792, "learning_rate": 1.452548994645003e-05, "loss": 1.936722412109375, "step": 137650 }, { "epoch": 0.5478334816966128, "grad_norm": 0.5604135394096375, "learning_rate": 1.452350072009994e-05, "loss": 1.972434539794922, "step": 137700 }, { "epoch": 0.5480324045294729, "grad_norm": 0.4923156201839447, "learning_rate": 1.4521511493749852e-05, "loss": 2.0224632263183593, "step": 137750 }, { "epoch": 0.5482313273623329, "grad_norm": 0.6003984808921814, "learning_rate": 1.4519522267399764e-05, "loss": 2.008384094238281, "step": 137800 }, { "epoch": 0.548430250195193, "grad_norm": 0.45967915654182434, "learning_rate": 1.4517533041049676e-05, "loss": 1.9915333557128907, "step": 137850 }, { "epoch": 0.5486291730280531, "grad_norm": 0.6062825918197632, "learning_rate": 1.4515543814699588e-05, "loss": 1.934249267578125, "step": 137900 }, { "epoch": 0.5488280958609132, "grad_norm": 0.537838876247406, "learning_rate": 1.45135545883495e-05, "loss": 1.9742752075195313, "step": 137950 }, { "epoch": 0.5490270186937732, "grad_norm": 0.5354487895965576, "learning_rate": 1.4511605146526416e-05, "loss": 1.968583984375, "step": 138000 }, { "epoch": 0.5492259415266333, "grad_norm": 0.5804877877235413, "learning_rate": 1.4509615920176326e-05, "loss": 2.0030511474609374, "step": 138050 }, { "epoch": 0.5494248643594933, "grad_norm": 0.4603697657585144, "learning_rate": 1.4507626693826238e-05, "loss": 1.9818960571289062, "step": 138100 }, { "epoch": 0.5496237871923534, "grad_norm": 0.6379734873771667, "learning_rate": 1.450563746747615e-05, "loss": 1.9492803955078124, "step": 138150 }, { "epoch": 0.5498227100252134, "grad_norm": 0.6644529104232788, "learning_rate": 1.4503648241126063e-05, "loss": 2.0360142517089845, "step": 138200 }, { "epoch": 0.5500216328580735, "grad_norm": 0.612097978591919, "learning_rate": 1.4501659014775975e-05, "loss": 1.948520965576172, "step": 138250 }, { "epoch": 0.5502205556909336, "grad_norm": 0.6787688136100769, "learning_rate": 1.4499669788425887e-05, "loss": 1.9636026000976563, "step": 138300 }, { "epoch": 0.5504194785237937, "grad_norm": 0.5619426369667053, "learning_rate": 1.4497680562075797e-05, "loss": 1.9722587585449218, "step": 138350 }, { "epoch": 0.5506184013566537, "grad_norm": 0.6316298246383667, "learning_rate": 1.449569133572571e-05, "loss": 2.0018879699707033, "step": 138400 }, { "epoch": 0.5508173241895138, "grad_norm": 0.579454243183136, "learning_rate": 1.4493702109375623e-05, "loss": 1.9932966613769532, "step": 138450 }, { "epoch": 0.5510162470223738, "grad_norm": 0.8012740612030029, "learning_rate": 1.4491712883025535e-05, "loss": 1.9665487670898438, "step": 138500 }, { "epoch": 0.5512151698552339, "grad_norm": 0.6527002453804016, "learning_rate": 1.4489723656675448e-05, "loss": 2.0111540222167967, "step": 138550 }, { "epoch": 0.5514140926880939, "grad_norm": 0.6367419362068176, "learning_rate": 1.448773443032536e-05, "loss": 1.979652862548828, "step": 138600 }, { "epoch": 0.551613015520954, "grad_norm": 0.5701773166656494, "learning_rate": 1.4485745203975272e-05, "loss": 2.004332275390625, "step": 138650 }, { "epoch": 0.5518119383538141, "grad_norm": 0.5171747803688049, "learning_rate": 1.4483755977625182e-05, "loss": 1.9655303955078125, "step": 138700 }, { "epoch": 0.5520108611866742, "grad_norm": 0.5803807377815247, "learning_rate": 1.4481766751275094e-05, "loss": 2.033139190673828, "step": 138750 }, { "epoch": 0.5522097840195342, "grad_norm": 0.4500597417354584, "learning_rate": 1.4479777524925007e-05, "loss": 2.014795227050781, "step": 138800 }, { "epoch": 0.5524087068523943, "grad_norm": 0.43543240427970886, "learning_rate": 1.447778829857492e-05, "loss": 2.0638444519042967, "step": 138850 }, { "epoch": 0.5526076296852543, "grad_norm": 0.6536105275154114, "learning_rate": 1.4475799072224833e-05, "loss": 1.9339842224121093, "step": 138900 }, { "epoch": 0.5528065525181144, "grad_norm": 0.5914337038993835, "learning_rate": 1.4473809845874745e-05, "loss": 1.9958506774902345, "step": 138950 }, { "epoch": 0.5530054753509744, "grad_norm": 0.5744814276695251, "learning_rate": 1.4471820619524655e-05, "loss": 1.9591065979003905, "step": 139000 }, { "epoch": 0.5532043981838345, "grad_norm": 0.5880095958709717, "learning_rate": 1.4469831393174567e-05, "loss": 1.9845138549804688, "step": 139050 }, { "epoch": 0.5534033210166946, "grad_norm": 0.6594445109367371, "learning_rate": 1.446784216682448e-05, "loss": 1.9858013916015624, "step": 139100 }, { "epoch": 0.5536022438495547, "grad_norm": 0.5938425064086914, "learning_rate": 1.4465892725001393e-05, "loss": 1.954027862548828, "step": 139150 }, { "epoch": 0.5538011666824147, "grad_norm": 0.6355128884315491, "learning_rate": 1.4463903498651307e-05, "loss": 2.006777648925781, "step": 139200 }, { "epoch": 0.5540000895152748, "grad_norm": 0.6034892797470093, "learning_rate": 1.4461914272301219e-05, "loss": 2.0090682983398436, "step": 139250 }, { "epoch": 0.5541990123481348, "grad_norm": 0.6976091861724854, "learning_rate": 1.4459925045951131e-05, "loss": 1.9503543090820312, "step": 139300 }, { "epoch": 0.5543979351809949, "grad_norm": 0.6557671427726746, "learning_rate": 1.4457935819601042e-05, "loss": 2.012257537841797, "step": 139350 }, { "epoch": 0.5545968580138549, "grad_norm": 0.5390744209289551, "learning_rate": 1.4455946593250954e-05, "loss": 2.0580543518066405, "step": 139400 }, { "epoch": 0.554795780846715, "grad_norm": 0.5782756209373474, "learning_rate": 1.4453957366900866e-05, "loss": 1.9781130981445312, "step": 139450 }, { "epoch": 0.5549947036795752, "grad_norm": 0.5468140244483948, "learning_rate": 1.4451968140550778e-05, "loss": 1.947934112548828, "step": 139500 }, { "epoch": 0.5551936265124352, "grad_norm": 0.6176870465278625, "learning_rate": 1.444997891420069e-05, "loss": 2.0041159057617186, "step": 139550 }, { "epoch": 0.5553925493452952, "grad_norm": 0.5385349988937378, "learning_rate": 1.4447989687850604e-05, "loss": 1.9610736083984375, "step": 139600 }, { "epoch": 0.5555914721781553, "grad_norm": 0.4104554355144501, "learning_rate": 1.4446000461500513e-05, "loss": 1.9001390075683593, "step": 139650 }, { "epoch": 0.5557903950110153, "grad_norm": 0.63675856590271, "learning_rate": 1.4444011235150427e-05, "loss": 1.98862060546875, "step": 139700 }, { "epoch": 0.5559893178438754, "grad_norm": 0.7137323021888733, "learning_rate": 1.4442022008800339e-05, "loss": 1.920377197265625, "step": 139750 }, { "epoch": 0.5561882406767354, "grad_norm": 0.5542101263999939, "learning_rate": 1.4440032782450251e-05, "loss": 1.9728407287597656, "step": 139800 }, { "epoch": 0.5563871635095955, "grad_norm": 0.5742987990379333, "learning_rate": 1.4438043556100163e-05, "loss": 1.975929718017578, "step": 139850 }, { "epoch": 0.5565860863424557, "grad_norm": 0.6179028749465942, "learning_rate": 1.4436054329750075e-05, "loss": 1.922774658203125, "step": 139900 }, { "epoch": 0.5567850091753157, "grad_norm": 0.8334759473800659, "learning_rate": 1.4434065103399987e-05, "loss": 2.019426727294922, "step": 139950 }, { "epoch": 0.5569839320081758, "grad_norm": 0.6589177250862122, "learning_rate": 1.4432075877049898e-05, "loss": 1.9641763305664062, "step": 140000 }, { "epoch": 0.5571828548410358, "grad_norm": 0.6340176463127136, "learning_rate": 1.443008665069981e-05, "loss": 1.9886383056640624, "step": 140050 }, { "epoch": 0.5573817776738959, "grad_norm": 0.6810864806175232, "learning_rate": 1.4428097424349724e-05, "loss": 1.9907334899902345, "step": 140100 }, { "epoch": 0.5575807005067559, "grad_norm": 0.6632858514785767, "learning_rate": 1.4426108197999636e-05, "loss": 2.008140869140625, "step": 140150 }, { "epoch": 0.557779623339616, "grad_norm": 0.6633093953132629, "learning_rate": 1.4424118971649548e-05, "loss": 1.9751251220703125, "step": 140200 }, { "epoch": 0.557978546172476, "grad_norm": 0.4136069715023041, "learning_rate": 1.442212974529946e-05, "loss": 1.9738786315917969, "step": 140250 }, { "epoch": 0.5581774690053362, "grad_norm": 0.6734405755996704, "learning_rate": 1.442014051894937e-05, "loss": 1.9785296630859375, "step": 140300 }, { "epoch": 0.5583763918381962, "grad_norm": 0.49203380942344666, "learning_rate": 1.4418151292599283e-05, "loss": 1.997069091796875, "step": 140350 }, { "epoch": 0.5585753146710563, "grad_norm": 0.5514113306999207, "learning_rate": 1.4416162066249195e-05, "loss": 2.0156434631347655, "step": 140400 }, { "epoch": 0.5587742375039163, "grad_norm": 0.5945544838905334, "learning_rate": 1.4414172839899107e-05, "loss": 1.969998321533203, "step": 140450 }, { "epoch": 0.5589731603367764, "grad_norm": 0.6392792463302612, "learning_rate": 1.441218361354902e-05, "loss": 1.9738522338867188, "step": 140500 }, { "epoch": 0.5591720831696364, "grad_norm": 0.6721217632293701, "learning_rate": 1.4410194387198933e-05, "loss": 1.9780084228515624, "step": 140550 }, { "epoch": 0.5593710060024965, "grad_norm": 0.5780634880065918, "learning_rate": 1.4408205160848845e-05, "loss": 1.9337135314941407, "step": 140600 }, { "epoch": 0.5595699288353565, "grad_norm": 0.5803233981132507, "learning_rate": 1.4406215934498756e-05, "loss": 2.021817321777344, "step": 140650 }, { "epoch": 0.5597688516682167, "grad_norm": 0.7375008463859558, "learning_rate": 1.4404226708148668e-05, "loss": 1.9685145568847657, "step": 140700 }, { "epoch": 0.5599677745010767, "grad_norm": 0.6591814756393433, "learning_rate": 1.440223748179858e-05, "loss": 1.932411651611328, "step": 140750 }, { "epoch": 0.5601666973339368, "grad_norm": 0.5557107329368591, "learning_rate": 1.4400248255448492e-05, "loss": 1.9887742614746093, "step": 140800 }, { "epoch": 0.5603656201667968, "grad_norm": 0.6418778896331787, "learning_rate": 1.4398259029098404e-05, "loss": 1.9608639526367186, "step": 140850 }, { "epoch": 0.5605645429996569, "grad_norm": 0.5457889437675476, "learning_rate": 1.4396269802748316e-05, "loss": 2.014262237548828, "step": 140900 }, { "epoch": 0.5607634658325169, "grad_norm": 0.6079552173614502, "learning_rate": 1.4394280576398227e-05, "loss": 2.021204528808594, "step": 140950 }, { "epoch": 0.560962388665377, "grad_norm": 0.5690563321113586, "learning_rate": 1.4392291350048139e-05, "loss": 2.043801727294922, "step": 141000 }, { "epoch": 0.561161311498237, "grad_norm": 0.6132711172103882, "learning_rate": 1.4390302123698053e-05, "loss": 1.9516384887695313, "step": 141050 }, { "epoch": 0.5613602343310972, "grad_norm": 0.5761020183563232, "learning_rate": 1.4388312897347965e-05, "loss": 2.0326718139648436, "step": 141100 }, { "epoch": 0.5615591571639572, "grad_norm": 0.5108965039253235, "learning_rate": 1.4386323670997877e-05, "loss": 1.948804931640625, "step": 141150 }, { "epoch": 0.5617580799968173, "grad_norm": 0.5437946915626526, "learning_rate": 1.438433444464779e-05, "loss": 1.9520852661132813, "step": 141200 }, { "epoch": 0.5619570028296773, "grad_norm": 0.7659685015678406, "learning_rate": 1.4382345218297701e-05, "loss": 2.006627960205078, "step": 141250 }, { "epoch": 0.5621559256625374, "grad_norm": 0.5723358988761902, "learning_rate": 1.4380355991947612e-05, "loss": 2.0402294921875, "step": 141300 }, { "epoch": 0.5623548484953974, "grad_norm": 0.4778207540512085, "learning_rate": 1.4378366765597524e-05, "loss": 1.9436611938476562, "step": 141350 }, { "epoch": 0.5625537713282575, "grad_norm": 0.5694226622581482, "learning_rate": 1.4376377539247436e-05, "loss": 1.9253312683105468, "step": 141400 }, { "epoch": 0.5627526941611175, "grad_norm": 0.5481706261634827, "learning_rate": 1.437438831289735e-05, "loss": 2.0005587768554687, "step": 141450 }, { "epoch": 0.5629516169939777, "grad_norm": 0.44024255871772766, "learning_rate": 1.4372399086547262e-05, "loss": 1.9563368225097657, "step": 141500 }, { "epoch": 0.5631505398268377, "grad_norm": 0.5845614075660706, "learning_rate": 1.4370409860197174e-05, "loss": 1.9436569213867188, "step": 141550 }, { "epoch": 0.5633494626596978, "grad_norm": 0.6205153465270996, "learning_rate": 1.4368420633847085e-05, "loss": 2.025343017578125, "step": 141600 }, { "epoch": 0.5635483854925578, "grad_norm": 0.5641186237335205, "learning_rate": 1.4366431407496997e-05, "loss": 1.9684063720703124, "step": 141650 }, { "epoch": 0.5637473083254179, "grad_norm": 0.5219302773475647, "learning_rate": 1.4364442181146909e-05, "loss": 1.9830099487304687, "step": 141700 }, { "epoch": 0.5639462311582779, "grad_norm": 0.5434600114822388, "learning_rate": 1.4362452954796821e-05, "loss": 2.006369934082031, "step": 141750 }, { "epoch": 0.564145153991138, "grad_norm": 0.6727455258369446, "learning_rate": 1.4360463728446733e-05, "loss": 2.014482421875, "step": 141800 }, { "epoch": 0.564344076823998, "grad_norm": 0.492174357175827, "learning_rate": 1.4358474502096647e-05, "loss": 1.9445584106445313, "step": 141850 }, { "epoch": 0.5645429996568582, "grad_norm": 0.5662975907325745, "learning_rate": 1.435648527574656e-05, "loss": 2.0429000854492188, "step": 141900 }, { "epoch": 0.5647419224897182, "grad_norm": 0.6790714263916016, "learning_rate": 1.435449604939647e-05, "loss": 1.970211944580078, "step": 141950 }, { "epoch": 0.5649408453225783, "grad_norm": 0.5780842900276184, "learning_rate": 1.4352506823046382e-05, "loss": 1.9650445556640626, "step": 142000 }, { "epoch": 0.5651397681554383, "grad_norm": 0.6790897846221924, "learning_rate": 1.4350517596696294e-05, "loss": 2.006329345703125, "step": 142050 }, { "epoch": 0.5653386909882984, "grad_norm": 0.5293068885803223, "learning_rate": 1.4348528370346206e-05, "loss": 2.0094801330566407, "step": 142100 }, { "epoch": 0.5655376138211584, "grad_norm": 0.5844816565513611, "learning_rate": 1.4346539143996118e-05, "loss": 2.0319784545898436, "step": 142150 }, { "epoch": 0.5657365366540185, "grad_norm": 0.5282744765281677, "learning_rate": 1.434454991764603e-05, "loss": 2.0354045104980467, "step": 142200 }, { "epoch": 0.5659354594868785, "grad_norm": 0.6394404172897339, "learning_rate": 1.4342560691295941e-05, "loss": 1.9620216369628907, "step": 142250 }, { "epoch": 0.5661343823197386, "grad_norm": 0.6864558458328247, "learning_rate": 1.4340571464945853e-05, "loss": 2.0643963623046875, "step": 142300 }, { "epoch": 0.5663333051525987, "grad_norm": 0.5574891567230225, "learning_rate": 1.4338582238595767e-05, "loss": 1.9690667724609374, "step": 142350 }, { "epoch": 0.5665322279854588, "grad_norm": 0.5631962418556213, "learning_rate": 1.4336593012245679e-05, "loss": 1.9572831726074218, "step": 142400 }, { "epoch": 0.5667311508183188, "grad_norm": 0.6217043399810791, "learning_rate": 1.4334603785895591e-05, "loss": 2.0355790710449218, "step": 142450 }, { "epoch": 0.5669300736511789, "grad_norm": 0.6095539927482605, "learning_rate": 1.4332614559545503e-05, "loss": 1.9713931274414063, "step": 142500 }, { "epoch": 0.5671289964840389, "grad_norm": 0.6017058491706848, "learning_rate": 1.4330625333195414e-05, "loss": 1.9414781188964845, "step": 142550 }, { "epoch": 0.567327919316899, "grad_norm": 0.6039707064628601, "learning_rate": 1.4328636106845326e-05, "loss": 1.9969737243652343, "step": 142600 }, { "epoch": 0.567526842149759, "grad_norm": 0.6128959059715271, "learning_rate": 1.4326646880495238e-05, "loss": 2.069419708251953, "step": 142650 }, { "epoch": 0.5677257649826191, "grad_norm": 0.5798870921134949, "learning_rate": 1.432465765414515e-05, "loss": 1.9719960021972656, "step": 142700 }, { "epoch": 0.5679246878154792, "grad_norm": 0.5992769002914429, "learning_rate": 1.4322668427795064e-05, "loss": 2.0022740173339844, "step": 142750 }, { "epoch": 0.5681236106483393, "grad_norm": 0.5014349818229675, "learning_rate": 1.4320679201444976e-05, "loss": 1.9365086364746094, "step": 142800 }, { "epoch": 0.5683225334811993, "grad_norm": 0.625528872013092, "learning_rate": 1.4318689975094888e-05, "loss": 1.9391366577148437, "step": 142850 }, { "epoch": 0.5685214563140594, "grad_norm": 0.6163278818130493, "learning_rate": 1.4316700748744799e-05, "loss": 2.0245927429199218, "step": 142900 }, { "epoch": 0.5687203791469194, "grad_norm": 0.693657636642456, "learning_rate": 1.4314711522394711e-05, "loss": 1.9145317077636719, "step": 142950 }, { "epoch": 0.5689193019797795, "grad_norm": 0.6166695952415466, "learning_rate": 1.4312722296044623e-05, "loss": 2.0040457153320315, "step": 143000 }, { "epoch": 0.5691182248126395, "grad_norm": 0.49845486879348755, "learning_rate": 1.4310733069694535e-05, "loss": 1.9828764343261718, "step": 143050 }, { "epoch": 0.5693171476454996, "grad_norm": 0.554673433303833, "learning_rate": 1.4308743843344447e-05, "loss": 2.0056260681152343, "step": 143100 }, { "epoch": 0.5695160704783597, "grad_norm": 0.5082075595855713, "learning_rate": 1.4306754616994361e-05, "loss": 1.9885269165039063, "step": 143150 }, { "epoch": 0.5697149933112198, "grad_norm": 0.5870106816291809, "learning_rate": 1.430476539064427e-05, "loss": 1.9917543029785156, "step": 143200 }, { "epoch": 0.5699139161440798, "grad_norm": 0.6082667708396912, "learning_rate": 1.4302776164294184e-05, "loss": 2.020365905761719, "step": 143250 }, { "epoch": 0.5701128389769399, "grad_norm": 0.588739812374115, "learning_rate": 1.4300786937944096e-05, "loss": 1.9879873657226563, "step": 143300 }, { "epoch": 0.5703117618097999, "grad_norm": 0.5372385382652283, "learning_rate": 1.4298797711594008e-05, "loss": 2.010905914306641, "step": 143350 }, { "epoch": 0.57051068464266, "grad_norm": 0.5121227502822876, "learning_rate": 1.429680848524392e-05, "loss": 1.9909695434570311, "step": 143400 }, { "epoch": 0.57070960747552, "grad_norm": 0.57268226146698, "learning_rate": 1.4294819258893832e-05, "loss": 1.979648895263672, "step": 143450 }, { "epoch": 0.5709085303083801, "grad_norm": 0.5728833079338074, "learning_rate": 1.4292830032543745e-05, "loss": 2.0039111328125, "step": 143500 }, { "epoch": 0.5711074531412402, "grad_norm": 0.60872882604599, "learning_rate": 1.4290840806193655e-05, "loss": 2.0075390625, "step": 143550 }, { "epoch": 0.5713063759741003, "grad_norm": 0.6441833972930908, "learning_rate": 1.4288851579843567e-05, "loss": 2.007279815673828, "step": 143600 }, { "epoch": 0.5715052988069603, "grad_norm": 0.6640268564224243, "learning_rate": 1.4286862353493481e-05, "loss": 1.9268988037109376, "step": 143650 }, { "epoch": 0.5717042216398204, "grad_norm": 0.6554080843925476, "learning_rate": 1.4284873127143393e-05, "loss": 1.9831289672851562, "step": 143700 }, { "epoch": 0.5719031444726804, "grad_norm": 0.5377809405326843, "learning_rate": 1.4282883900793305e-05, "loss": 2.0208567810058593, "step": 143750 }, { "epoch": 0.5721020673055405, "grad_norm": 0.5839488506317139, "learning_rate": 1.4280894674443217e-05, "loss": 1.9903134155273436, "step": 143800 }, { "epoch": 0.5723009901384005, "grad_norm": 0.5911650657653809, "learning_rate": 1.4278905448093128e-05, "loss": 2.0269398498535156, "step": 143850 }, { "epoch": 0.5724999129712606, "grad_norm": 0.5451627373695374, "learning_rate": 1.427691622174304e-05, "loss": 1.9681072998046876, "step": 143900 }, { "epoch": 0.5726988358041207, "grad_norm": 0.5097623467445374, "learning_rate": 1.4274926995392952e-05, "loss": 1.9735379028320312, "step": 143950 }, { "epoch": 0.5728977586369808, "grad_norm": 0.5164663195610046, "learning_rate": 1.4272937769042864e-05, "loss": 1.921546630859375, "step": 144000 }, { "epoch": 0.5730966814698408, "grad_norm": 0.5730831623077393, "learning_rate": 1.427098832721978e-05, "loss": 2.013325653076172, "step": 144050 }, { "epoch": 0.5732956043027009, "grad_norm": 0.6048088669776917, "learning_rate": 1.4268999100869692e-05, "loss": 1.9902583312988282, "step": 144100 }, { "epoch": 0.5734945271355609, "grad_norm": 0.5833395719528198, "learning_rate": 1.4267009874519604e-05, "loss": 1.9467921447753906, "step": 144150 }, { "epoch": 0.573693449968421, "grad_norm": 0.654052734375, "learning_rate": 1.4265020648169514e-05, "loss": 2.0546710205078127, "step": 144200 }, { "epoch": 0.573892372801281, "grad_norm": 0.5771719813346863, "learning_rate": 1.4263031421819426e-05, "loss": 1.9548284912109375, "step": 144250 }, { "epoch": 0.5740912956341411, "grad_norm": 0.5778651833534241, "learning_rate": 1.4261042195469339e-05, "loss": 1.9957354736328126, "step": 144300 }, { "epoch": 0.5742902184670012, "grad_norm": 0.5794798731803894, "learning_rate": 1.425905296911925e-05, "loss": 1.9949826049804686, "step": 144350 }, { "epoch": 0.5744891412998613, "grad_norm": 0.4613473117351532, "learning_rate": 1.4257063742769165e-05, "loss": 2.0554876708984375, "step": 144400 }, { "epoch": 0.5746880641327213, "grad_norm": 0.5529010891914368, "learning_rate": 1.4255074516419077e-05, "loss": 1.9726644897460937, "step": 144450 }, { "epoch": 0.5748869869655814, "grad_norm": 0.5774592757225037, "learning_rate": 1.4253085290068987e-05, "loss": 1.9690704345703125, "step": 144500 }, { "epoch": 0.5750859097984414, "grad_norm": 0.49598440527915955, "learning_rate": 1.42510960637189e-05, "loss": 1.9665963745117188, "step": 144550 }, { "epoch": 0.5752848326313015, "grad_norm": 0.6494578123092651, "learning_rate": 1.4249106837368811e-05, "loss": 1.8923704528808594, "step": 144600 }, { "epoch": 0.5754837554641615, "grad_norm": 0.6097249984741211, "learning_rate": 1.4247117611018724e-05, "loss": 2.0012391662597655, "step": 144650 }, { "epoch": 0.5756826782970216, "grad_norm": 0.6304829120635986, "learning_rate": 1.4245128384668636e-05, "loss": 2.020894012451172, "step": 144700 }, { "epoch": 0.5758816011298817, "grad_norm": 0.5362088680267334, "learning_rate": 1.4243139158318548e-05, "loss": 1.9787835693359375, "step": 144750 }, { "epoch": 0.5760805239627418, "grad_norm": 0.5006948709487915, "learning_rate": 1.4241149931968462e-05, "loss": 1.97440185546875, "step": 144800 }, { "epoch": 0.5762794467956018, "grad_norm": 0.4804908335208893, "learning_rate": 1.423916070561837e-05, "loss": 1.9953359985351562, "step": 144850 }, { "epoch": 0.5764783696284619, "grad_norm": 0.5267537832260132, "learning_rate": 1.4237171479268284e-05, "loss": 2.041855926513672, "step": 144900 }, { "epoch": 0.5766772924613219, "grad_norm": 0.5544477105140686, "learning_rate": 1.4235182252918197e-05, "loss": 1.9623500061035157, "step": 144950 }, { "epoch": 0.576876215294182, "grad_norm": 0.6390044689178467, "learning_rate": 1.4233193026568109e-05, "loss": 2.0149432373046876, "step": 145000 }, { "epoch": 0.577075138127042, "grad_norm": 0.5624077320098877, "learning_rate": 1.423120380021802e-05, "loss": 1.968795166015625, "step": 145050 }, { "epoch": 0.5772740609599021, "grad_norm": 0.6305074095726013, "learning_rate": 1.4229214573867933e-05, "loss": 1.9608464050292969, "step": 145100 }, { "epoch": 0.5774729837927622, "grad_norm": 0.5752955675125122, "learning_rate": 1.4227225347517843e-05, "loss": 2.0027737426757812, "step": 145150 }, { "epoch": 0.5776719066256223, "grad_norm": 0.5830689072608948, "learning_rate": 1.4225236121167756e-05, "loss": 1.9507281494140625, "step": 145200 }, { "epoch": 0.5778708294584823, "grad_norm": 0.6456285119056702, "learning_rate": 1.4223246894817668e-05, "loss": 2.041114959716797, "step": 145250 }, { "epoch": 0.5780697522913424, "grad_norm": 0.5449653267860413, "learning_rate": 1.4221257668467582e-05, "loss": 1.984727325439453, "step": 145300 }, { "epoch": 0.5782686751242024, "grad_norm": 0.5917249917984009, "learning_rate": 1.4219268442117494e-05, "loss": 1.9873452758789063, "step": 145350 }, { "epoch": 0.5784675979570625, "grad_norm": 0.5015178322792053, "learning_rate": 1.4217279215767406e-05, "loss": 1.9860858154296874, "step": 145400 }, { "epoch": 0.5786665207899225, "grad_norm": 0.6735976338386536, "learning_rate": 1.4215289989417318e-05, "loss": 1.9475883483886718, "step": 145450 }, { "epoch": 0.5788654436227826, "grad_norm": 0.6742454171180725, "learning_rate": 1.4213300763067228e-05, "loss": 1.9570182800292968, "step": 145500 }, { "epoch": 0.5790643664556427, "grad_norm": 0.5054150819778442, "learning_rate": 1.421131153671714e-05, "loss": 1.95245849609375, "step": 145550 }, { "epoch": 0.5792632892885028, "grad_norm": 0.48625725507736206, "learning_rate": 1.4209322310367053e-05, "loss": 2.0065933227539063, "step": 145600 }, { "epoch": 0.5794622121213628, "grad_norm": 0.6264781355857849, "learning_rate": 1.4207333084016965e-05, "loss": 1.9519760131835937, "step": 145650 }, { "epoch": 0.5796611349542229, "grad_norm": 0.5935661792755127, "learning_rate": 1.4205343857666877e-05, "loss": 1.9746319580078124, "step": 145700 }, { "epoch": 0.5798600577870829, "grad_norm": 0.5865821838378906, "learning_rate": 1.4203354631316791e-05, "loss": 1.947020263671875, "step": 145750 }, { "epoch": 0.580058980619943, "grad_norm": 0.5497528910636902, "learning_rate": 1.4201365404966701e-05, "loss": 1.9284950256347657, "step": 145800 }, { "epoch": 0.580257903452803, "grad_norm": 0.5742099285125732, "learning_rate": 1.4199376178616613e-05, "loss": 1.9417924499511718, "step": 145850 }, { "epoch": 0.5804568262856631, "grad_norm": 0.5472828149795532, "learning_rate": 1.4197386952266526e-05, "loss": 1.9910791015625, "step": 145900 }, { "epoch": 0.5806557491185232, "grad_norm": 0.5790324211120605, "learning_rate": 1.4195397725916438e-05, "loss": 1.9403025817871093, "step": 145950 }, { "epoch": 0.5808546719513833, "grad_norm": 0.6105930805206299, "learning_rate": 1.419340849956635e-05, "loss": 1.9959463500976562, "step": 146000 }, { "epoch": 0.5810535947842433, "grad_norm": 0.4908815324306488, "learning_rate": 1.4191419273216262e-05, "loss": 2.04427978515625, "step": 146050 }, { "epoch": 0.5812525176171034, "grad_norm": 0.6589938402175903, "learning_rate": 1.4189430046866174e-05, "loss": 1.9590835571289062, "step": 146100 }, { "epoch": 0.5814514404499634, "grad_norm": 0.5611351132392883, "learning_rate": 1.4187440820516085e-05, "loss": 2.0119442749023437, "step": 146150 }, { "epoch": 0.5816503632828235, "grad_norm": 0.5693400502204895, "learning_rate": 1.4185451594165997e-05, "loss": 1.9472975158691406, "step": 146200 }, { "epoch": 0.5818492861156835, "grad_norm": 0.5513907074928284, "learning_rate": 1.418346236781591e-05, "loss": 2.0213916015625, "step": 146250 }, { "epoch": 0.5820482089485436, "grad_norm": 0.6124303936958313, "learning_rate": 1.4181473141465823e-05, "loss": 1.9521702575683593, "step": 146300 }, { "epoch": 0.5822471317814037, "grad_norm": 0.6999961733818054, "learning_rate": 1.4179483915115735e-05, "loss": 1.9269813537597655, "step": 146350 }, { "epoch": 0.5824460546142638, "grad_norm": 0.6211793422698975, "learning_rate": 1.4177494688765647e-05, "loss": 1.9469973754882812, "step": 146400 }, { "epoch": 0.5826449774471238, "grad_norm": 0.6881364583969116, "learning_rate": 1.4175505462415558e-05, "loss": 1.9052732849121095, "step": 146450 }, { "epoch": 0.5828439002799839, "grad_norm": 0.60647052526474, "learning_rate": 1.417351623606547e-05, "loss": 1.9879653930664063, "step": 146500 }, { "epoch": 0.5830428231128439, "grad_norm": 0.6440387964248657, "learning_rate": 1.4171527009715382e-05, "loss": 2.014398651123047, "step": 146550 }, { "epoch": 0.583241745945704, "grad_norm": 0.537166953086853, "learning_rate": 1.4169537783365294e-05, "loss": 1.9137666320800781, "step": 146600 }, { "epoch": 0.583440668778564, "grad_norm": 0.6289156079292297, "learning_rate": 1.4167548557015208e-05, "loss": 1.9554574584960938, "step": 146650 }, { "epoch": 0.5836395916114241, "grad_norm": 0.6454509496688843, "learning_rate": 1.416555933066512e-05, "loss": 1.9652674865722657, "step": 146700 }, { "epoch": 0.5838385144442843, "grad_norm": 0.6522575616836548, "learning_rate": 1.4163570104315032e-05, "loss": 2.0026651000976563, "step": 146750 }, { "epoch": 0.5840374372771443, "grad_norm": 0.701380729675293, "learning_rate": 1.4161580877964943e-05, "loss": 1.9757929992675782, "step": 146800 }, { "epoch": 0.5842363601100043, "grad_norm": 0.6463649868965149, "learning_rate": 1.4159591651614855e-05, "loss": 1.9527249145507812, "step": 146850 }, { "epoch": 0.5844352829428644, "grad_norm": 0.602682888507843, "learning_rate": 1.4157602425264767e-05, "loss": 1.9980174255371095, "step": 146900 }, { "epoch": 0.5846342057757244, "grad_norm": 0.5969951152801514, "learning_rate": 1.4155613198914679e-05, "loss": 1.987461395263672, "step": 146950 }, { "epoch": 0.5848331286085845, "grad_norm": 0.5767215490341187, "learning_rate": 1.4153623972564591e-05, "loss": 2.006571807861328, "step": 147000 }, { "epoch": 0.5850320514414445, "grad_norm": 0.5673882961273193, "learning_rate": 1.4151634746214505e-05, "loss": 1.9960049438476561, "step": 147050 }, { "epoch": 0.5852309742743046, "grad_norm": 0.615764856338501, "learning_rate": 1.4149645519864414e-05, "loss": 1.9653660583496093, "step": 147100 }, { "epoch": 0.5854298971071648, "grad_norm": 0.47032856941223145, "learning_rate": 1.4147656293514328e-05, "loss": 2.020790557861328, "step": 147150 }, { "epoch": 0.5856288199400248, "grad_norm": 0.6197739839553833, "learning_rate": 1.414566706716424e-05, "loss": 1.9627052307128907, "step": 147200 }, { "epoch": 0.5858277427728849, "grad_norm": 0.5816812515258789, "learning_rate": 1.4143677840814152e-05, "loss": 1.9829629516601563, "step": 147250 }, { "epoch": 0.5860266656057449, "grad_norm": 0.38828179240226746, "learning_rate": 1.4141688614464064e-05, "loss": 1.985186309814453, "step": 147300 }, { "epoch": 0.586225588438605, "grad_norm": 0.6422900557518005, "learning_rate": 1.4139699388113976e-05, "loss": 1.9690115356445312, "step": 147350 }, { "epoch": 0.586424511271465, "grad_norm": 0.6291143894195557, "learning_rate": 1.4137710161763888e-05, "loss": 1.9990585327148438, "step": 147400 }, { "epoch": 0.586623434104325, "grad_norm": 0.5354421138763428, "learning_rate": 1.4135720935413799e-05, "loss": 1.9765324401855469, "step": 147450 }, { "epoch": 0.5868223569371851, "grad_norm": 0.562746524810791, "learning_rate": 1.4133771493590714e-05, "loss": 2.0291256713867187, "step": 147500 }, { "epoch": 0.5870212797700453, "grad_norm": 0.5156688094139099, "learning_rate": 1.4131782267240626e-05, "loss": 1.9900469970703125, "step": 147550 }, { "epoch": 0.5872202026029053, "grad_norm": 0.56290602684021, "learning_rate": 1.4129793040890538e-05, "loss": 1.9485667419433594, "step": 147600 }, { "epoch": 0.5874191254357654, "grad_norm": 0.5890428423881531, "learning_rate": 1.412780381454045e-05, "loss": 1.9934700012207032, "step": 147650 }, { "epoch": 0.5876180482686254, "grad_norm": 0.6340422034263611, "learning_rate": 1.4125814588190363e-05, "loss": 1.9949833679199218, "step": 147700 }, { "epoch": 0.5878169711014855, "grad_norm": 0.6224383115768433, "learning_rate": 1.4123825361840273e-05, "loss": 1.9850450134277344, "step": 147750 }, { "epoch": 0.5880158939343455, "grad_norm": 0.5887055397033691, "learning_rate": 1.4121836135490185e-05, "loss": 1.9687953186035156, "step": 147800 }, { "epoch": 0.5882148167672056, "grad_norm": 0.6404150724411011, "learning_rate": 1.4119846909140097e-05, "loss": 2.0082183837890626, "step": 147850 }, { "epoch": 0.5884137396000656, "grad_norm": 0.6004981994628906, "learning_rate": 1.4117857682790011e-05, "loss": 2.0302392578125, "step": 147900 }, { "epoch": 0.5886126624329258, "grad_norm": 0.7022414207458496, "learning_rate": 1.4115868456439923e-05, "loss": 1.956790771484375, "step": 147950 }, { "epoch": 0.5888115852657858, "grad_norm": 0.6091811060905457, "learning_rate": 1.4113879230089835e-05, "loss": 1.94852783203125, "step": 148000 }, { "epoch": 0.5890105080986459, "grad_norm": 0.5603083968162537, "learning_rate": 1.4111890003739748e-05, "loss": 1.9768502807617188, "step": 148050 }, { "epoch": 0.5892094309315059, "grad_norm": 0.5908665657043457, "learning_rate": 1.4109900777389658e-05, "loss": 1.9430448913574219, "step": 148100 }, { "epoch": 0.589408353764366, "grad_norm": 0.6336021423339844, "learning_rate": 1.410791155103957e-05, "loss": 1.9636019897460937, "step": 148150 }, { "epoch": 0.589607276597226, "grad_norm": 0.4396447241306305, "learning_rate": 1.4105922324689482e-05, "loss": 1.9788621520996095, "step": 148200 }, { "epoch": 0.589806199430086, "grad_norm": 0.5524910092353821, "learning_rate": 1.4103933098339394e-05, "loss": 1.9454034423828126, "step": 148250 }, { "epoch": 0.5900051222629461, "grad_norm": 0.585757315158844, "learning_rate": 1.4101943871989308e-05, "loss": 1.9684664916992187, "step": 148300 }, { "epoch": 0.5902040450958063, "grad_norm": 0.5455991625785828, "learning_rate": 1.409995464563922e-05, "loss": 2.010518341064453, "step": 148350 }, { "epoch": 0.5904029679286663, "grad_norm": 0.536930501461029, "learning_rate": 1.4097965419289131e-05, "loss": 2.0225703430175783, "step": 148400 }, { "epoch": 0.5906018907615264, "grad_norm": 0.5385761260986328, "learning_rate": 1.4095976192939043e-05, "loss": 1.9541815185546876, "step": 148450 }, { "epoch": 0.5908008135943864, "grad_norm": 0.559441089630127, "learning_rate": 1.4093986966588955e-05, "loss": 1.9045001220703126, "step": 148500 }, { "epoch": 0.5909997364272465, "grad_norm": 0.5443758964538574, "learning_rate": 1.4091997740238867e-05, "loss": 1.953840789794922, "step": 148550 }, { "epoch": 0.5911986592601065, "grad_norm": 0.5813440680503845, "learning_rate": 1.409000851388878e-05, "loss": 1.9963191223144532, "step": 148600 }, { "epoch": 0.5913975820929666, "grad_norm": 0.5790966749191284, "learning_rate": 1.4088019287538692e-05, "loss": 1.9858781433105468, "step": 148650 }, { "epoch": 0.5915965049258266, "grad_norm": 0.6409999132156372, "learning_rate": 1.4086030061188605e-05, "loss": 1.9737269592285156, "step": 148700 }, { "epoch": 0.5917954277586868, "grad_norm": 0.5501960515975952, "learning_rate": 1.4084040834838514e-05, "loss": 1.9496388244628906, "step": 148750 }, { "epoch": 0.5919943505915468, "grad_norm": 0.6250784993171692, "learning_rate": 1.4082051608488428e-05, "loss": 1.9870468139648438, "step": 148800 }, { "epoch": 0.5921932734244069, "grad_norm": 0.510622501373291, "learning_rate": 1.408006238213834e-05, "loss": 1.9921371459960937, "step": 148850 }, { "epoch": 0.5923921962572669, "grad_norm": 0.6150009632110596, "learning_rate": 1.4078073155788252e-05, "loss": 1.9881866455078125, "step": 148900 }, { "epoch": 0.592591119090127, "grad_norm": 0.602886974811554, "learning_rate": 1.4076083929438164e-05, "loss": 1.9918135070800782, "step": 148950 }, { "epoch": 0.592790041922987, "grad_norm": 0.6694660782814026, "learning_rate": 1.4074094703088077e-05, "loss": 1.9564869689941407, "step": 149000 }, { "epoch": 0.5929889647558471, "grad_norm": 0.4486539363861084, "learning_rate": 1.4072105476737987e-05, "loss": 1.956284942626953, "step": 149050 }, { "epoch": 0.5931878875887071, "grad_norm": 0.5947849750518799, "learning_rate": 1.40701162503879e-05, "loss": 1.9775117492675782, "step": 149100 }, { "epoch": 0.5933868104215673, "grad_norm": 0.655955970287323, "learning_rate": 1.4068127024037811e-05, "loss": 1.9853074645996094, "step": 149150 }, { "epoch": 0.5935857332544273, "grad_norm": 0.6166208386421204, "learning_rate": 1.4066137797687725e-05, "loss": 1.9633848571777344, "step": 149200 }, { "epoch": 0.5937846560872874, "grad_norm": 0.5587543249130249, "learning_rate": 1.4064148571337637e-05, "loss": 1.95594970703125, "step": 149250 }, { "epoch": 0.5939835789201474, "grad_norm": 0.556018590927124, "learning_rate": 1.406215934498755e-05, "loss": 2.02090576171875, "step": 149300 }, { "epoch": 0.5941825017530075, "grad_norm": 0.5404276251792908, "learning_rate": 1.4060170118637462e-05, "loss": 2.0082075500488283, "step": 149350 }, { "epoch": 0.5943814245858675, "grad_norm": 0.5478287935256958, "learning_rate": 1.4058180892287372e-05, "loss": 1.9641944885253906, "step": 149400 }, { "epoch": 0.5945803474187276, "grad_norm": 0.6481881141662598, "learning_rate": 1.4056191665937284e-05, "loss": 1.9456768798828126, "step": 149450 }, { "epoch": 0.5947792702515876, "grad_norm": 0.6112906336784363, "learning_rate": 1.4054202439587196e-05, "loss": 2.022093505859375, "step": 149500 }, { "epoch": 0.5949781930844478, "grad_norm": 0.5400034785270691, "learning_rate": 1.4052213213237109e-05, "loss": 2.006224822998047, "step": 149550 }, { "epoch": 0.5951771159173078, "grad_norm": 0.5897115468978882, "learning_rate": 1.4050223986887022e-05, "loss": 1.9656861877441407, "step": 149600 }, { "epoch": 0.5953760387501679, "grad_norm": 0.5846022367477417, "learning_rate": 1.4048234760536935e-05, "loss": 1.9980613708496093, "step": 149650 }, { "epoch": 0.5955749615830279, "grad_norm": 0.5417883992195129, "learning_rate": 1.4046245534186845e-05, "loss": 1.9915724182128907, "step": 149700 }, { "epoch": 0.595773884415888, "grad_norm": 0.5882043242454529, "learning_rate": 1.4044256307836757e-05, "loss": 1.9116661071777343, "step": 149750 }, { "epoch": 0.595972807248748, "grad_norm": 0.6878496408462524, "learning_rate": 1.404226708148667e-05, "loss": 1.9517095947265626, "step": 149800 }, { "epoch": 0.5961717300816081, "grad_norm": 0.3665289580821991, "learning_rate": 1.4040277855136581e-05, "loss": 1.9777117919921876, "step": 149850 }, { "epoch": 0.5963706529144681, "grad_norm": 0.5359643697738647, "learning_rate": 1.4038288628786494e-05, "loss": 2.0068218994140623, "step": 149900 }, { "epoch": 0.5965695757473282, "grad_norm": 0.6627965569496155, "learning_rate": 1.4036299402436406e-05, "loss": 1.9750442504882812, "step": 149950 }, { "epoch": 0.5967684985801883, "grad_norm": 0.6517647504806519, "learning_rate": 1.403431017608632e-05, "loss": 1.95282470703125, "step": 150000 }, { "epoch": 0.5969674214130484, "grad_norm": 0.5846887230873108, "learning_rate": 1.4032320949736228e-05, "loss": 1.948846893310547, "step": 150050 }, { "epoch": 0.5971663442459084, "grad_norm": 0.553419291973114, "learning_rate": 1.4030331723386142e-05, "loss": 1.9832928466796875, "step": 150100 }, { "epoch": 0.5973652670787685, "grad_norm": 0.5942819118499756, "learning_rate": 1.4028342497036054e-05, "loss": 1.977229461669922, "step": 150150 }, { "epoch": 0.5975641899116285, "grad_norm": 0.626459538936615, "learning_rate": 1.4026353270685966e-05, "loss": 1.9773066711425782, "step": 150200 }, { "epoch": 0.5977631127444886, "grad_norm": 0.5352042317390442, "learning_rate": 1.4024364044335879e-05, "loss": 1.9491584777832032, "step": 150250 }, { "epoch": 0.5979620355773486, "grad_norm": 0.6016097068786621, "learning_rate": 1.402237481798579e-05, "loss": 1.9985699462890625, "step": 150300 }, { "epoch": 0.5981609584102087, "grad_norm": 0.6014531254768372, "learning_rate": 1.4020385591635701e-05, "loss": 1.9575650024414062, "step": 150350 }, { "epoch": 0.5983598812430688, "grad_norm": 0.5515604615211487, "learning_rate": 1.4018396365285613e-05, "loss": 1.9852590942382813, "step": 150400 }, { "epoch": 0.5985588040759289, "grad_norm": 0.5401650071144104, "learning_rate": 1.4016407138935525e-05, "loss": 1.9415841674804688, "step": 150450 }, { "epoch": 0.5987577269087889, "grad_norm": 0.5898739695549011, "learning_rate": 1.401441791258544e-05, "loss": 1.94236328125, "step": 150500 }, { "epoch": 0.598956649741649, "grad_norm": 0.6101876497268677, "learning_rate": 1.4012428686235351e-05, "loss": 1.9868145751953126, "step": 150550 }, { "epoch": 0.599155572574509, "grad_norm": 0.5203258991241455, "learning_rate": 1.4010439459885264e-05, "loss": 1.9806593322753907, "step": 150600 }, { "epoch": 0.5993544954073691, "grad_norm": 0.7103843688964844, "learning_rate": 1.4008450233535176e-05, "loss": 1.9858570861816407, "step": 150650 }, { "epoch": 0.5995534182402291, "grad_norm": 0.6403511166572571, "learning_rate": 1.4006461007185086e-05, "loss": 1.9933828735351562, "step": 150700 }, { "epoch": 0.5997523410730892, "grad_norm": 0.5638087391853333, "learning_rate": 1.4004471780834998e-05, "loss": 2.0194940185546875, "step": 150750 }, { "epoch": 0.5999512639059493, "grad_norm": 0.5866647958755493, "learning_rate": 1.400248255448491e-05, "loss": 1.9953623962402345, "step": 150800 }, { "epoch": 0.6001501867388094, "grad_norm": 0.5838631987571716, "learning_rate": 1.4000493328134823e-05, "loss": 1.9555686950683593, "step": 150850 }, { "epoch": 0.6003491095716694, "grad_norm": 0.6169543266296387, "learning_rate": 1.3998504101784735e-05, "loss": 1.965205535888672, "step": 150900 }, { "epoch": 0.6005480324045295, "grad_norm": 0.6388037800788879, "learning_rate": 1.3996514875434649e-05, "loss": 1.9168960571289062, "step": 150950 }, { "epoch": 0.6007469552373895, "grad_norm": 0.6775066256523132, "learning_rate": 1.3994525649084559e-05, "loss": 1.9992137145996094, "step": 151000 }, { "epoch": 0.6009458780702496, "grad_norm": 0.6561816334724426, "learning_rate": 1.3992536422734471e-05, "loss": 2.0386923217773436, "step": 151050 }, { "epoch": 0.6011448009031096, "grad_norm": 0.569974958896637, "learning_rate": 1.3990547196384383e-05, "loss": 1.97479248046875, "step": 151100 }, { "epoch": 0.6013437237359697, "grad_norm": 0.5715729594230652, "learning_rate": 1.3988557970034296e-05, "loss": 1.9509896850585937, "step": 151150 }, { "epoch": 0.6015426465688298, "grad_norm": 0.5517758727073669, "learning_rate": 1.3986568743684208e-05, "loss": 1.964460906982422, "step": 151200 }, { "epoch": 0.6017415694016899, "grad_norm": 0.6147448420524597, "learning_rate": 1.398457951733412e-05, "loss": 1.958734893798828, "step": 151250 }, { "epoch": 0.6019404922345499, "grad_norm": 0.5651826858520508, "learning_rate": 1.398259029098403e-05, "loss": 1.983199920654297, "step": 151300 }, { "epoch": 0.60213941506741, "grad_norm": 0.4462471008300781, "learning_rate": 1.3980601064633942e-05, "loss": 1.9541645812988282, "step": 151350 }, { "epoch": 0.60233833790027, "grad_norm": 0.5554624199867249, "learning_rate": 1.3978611838283855e-05, "loss": 2.017111511230469, "step": 151400 }, { "epoch": 0.6025372607331301, "grad_norm": 0.4818060100078583, "learning_rate": 1.3976622611933768e-05, "loss": 1.9847021484375, "step": 151450 }, { "epoch": 0.6027361835659901, "grad_norm": 0.5964573621749878, "learning_rate": 1.397463338558368e-05, "loss": 1.9740855407714843, "step": 151500 }, { "epoch": 0.6029351063988502, "grad_norm": 0.6530109643936157, "learning_rate": 1.3972644159233593e-05, "loss": 1.9666963195800782, "step": 151550 }, { "epoch": 0.6031340292317103, "grad_norm": 0.5334864258766174, "learning_rate": 1.3970654932883505e-05, "loss": 2.016317596435547, "step": 151600 }, { "epoch": 0.6033329520645704, "grad_norm": 0.6201861500740051, "learning_rate": 1.3968665706533415e-05, "loss": 1.9862095642089843, "step": 151650 }, { "epoch": 0.6035318748974304, "grad_norm": 0.5875126719474792, "learning_rate": 1.3966716264710329e-05, "loss": 1.9562677001953126, "step": 151700 }, { "epoch": 0.6037307977302905, "grad_norm": 0.5664580464363098, "learning_rate": 1.3964727038360243e-05, "loss": 2.026388854980469, "step": 151750 }, { "epoch": 0.6039297205631505, "grad_norm": 0.6049782633781433, "learning_rate": 1.3962737812010155e-05, "loss": 1.991367645263672, "step": 151800 }, { "epoch": 0.6041286433960106, "grad_norm": 0.5953044295310974, "learning_rate": 1.3960748585660067e-05, "loss": 1.9439295959472656, "step": 151850 }, { "epoch": 0.6043275662288706, "grad_norm": 0.6343777775764465, "learning_rate": 1.3958759359309979e-05, "loss": 1.9560060119628906, "step": 151900 }, { "epoch": 0.6045264890617307, "grad_norm": 0.6258072257041931, "learning_rate": 1.3956770132959891e-05, "loss": 1.996687469482422, "step": 151950 }, { "epoch": 0.6047254118945908, "grad_norm": 0.5767166614532471, "learning_rate": 1.3954780906609802e-05, "loss": 1.9831953430175782, "step": 152000 }, { "epoch": 0.6049243347274509, "grad_norm": 0.6994052529335022, "learning_rate": 1.3952791680259714e-05, "loss": 1.9268402099609374, "step": 152050 }, { "epoch": 0.6051232575603109, "grad_norm": 0.6213752627372742, "learning_rate": 1.3950802453909626e-05, "loss": 2.0003230285644533, "step": 152100 }, { "epoch": 0.605322180393171, "grad_norm": 0.5211489200592041, "learning_rate": 1.3948813227559538e-05, "loss": 1.9119802856445312, "step": 152150 }, { "epoch": 0.605521103226031, "grad_norm": 0.5017456412315369, "learning_rate": 1.3946824001209452e-05, "loss": 1.956068878173828, "step": 152200 }, { "epoch": 0.6057200260588911, "grad_norm": 0.5840393900871277, "learning_rate": 1.3944834774859364e-05, "loss": 1.9573220825195312, "step": 152250 }, { "epoch": 0.6059189488917511, "grad_norm": 0.5859197974205017, "learning_rate": 1.3942845548509275e-05, "loss": 2.0054998779296875, "step": 152300 }, { "epoch": 0.6061178717246112, "grad_norm": 0.6430534720420837, "learning_rate": 1.3940856322159187e-05, "loss": 2.0206700134277344, "step": 152350 }, { "epoch": 0.6063167945574713, "grad_norm": 0.5704708099365234, "learning_rate": 1.3938867095809099e-05, "loss": 1.960882110595703, "step": 152400 }, { "epoch": 0.6065157173903314, "grad_norm": 0.5880545973777771, "learning_rate": 1.3936877869459011e-05, "loss": 2.001459503173828, "step": 152450 }, { "epoch": 0.6067146402231914, "grad_norm": 0.5577272176742554, "learning_rate": 1.3934888643108923e-05, "loss": 2.009729919433594, "step": 152500 }, { "epoch": 0.6069135630560515, "grad_norm": 0.5502371191978455, "learning_rate": 1.3932899416758835e-05, "loss": 1.920638427734375, "step": 152550 }, { "epoch": 0.6071124858889115, "grad_norm": 0.544274091720581, "learning_rate": 1.3930910190408749e-05, "loss": 1.9505894470214844, "step": 152600 }, { "epoch": 0.6073114087217716, "grad_norm": 0.5818033814430237, "learning_rate": 1.3928920964058658e-05, "loss": 1.9972996520996094, "step": 152650 }, { "epoch": 0.6075103315546316, "grad_norm": 0.5669009685516357, "learning_rate": 1.3926931737708572e-05, "loss": 1.9448707580566407, "step": 152700 }, { "epoch": 0.6077092543874917, "grad_norm": 0.598420262336731, "learning_rate": 1.3924942511358484e-05, "loss": 1.950810546875, "step": 152750 }, { "epoch": 0.6079081772203518, "grad_norm": 0.5515417456626892, "learning_rate": 1.3922953285008396e-05, "loss": 1.9660202026367188, "step": 152800 }, { "epoch": 0.6081071000532119, "grad_norm": 0.4734908938407898, "learning_rate": 1.3920964058658308e-05, "loss": 1.9625006103515625, "step": 152850 }, { "epoch": 0.6083060228860719, "grad_norm": 0.5772661566734314, "learning_rate": 1.391897483230822e-05, "loss": 1.984127960205078, "step": 152900 }, { "epoch": 0.608504945718932, "grad_norm": 0.577225923538208, "learning_rate": 1.391698560595813e-05, "loss": 1.9701325988769531, "step": 152950 }, { "epoch": 0.608703868551792, "grad_norm": 0.7138063311576843, "learning_rate": 1.3914996379608043e-05, "loss": 2.024362335205078, "step": 153000 }, { "epoch": 0.6089027913846521, "grad_norm": 0.5378592610359192, "learning_rate": 1.3913007153257955e-05, "loss": 2.0059603881835937, "step": 153050 }, { "epoch": 0.6091017142175121, "grad_norm": 0.5859053730964661, "learning_rate": 1.3911017926907869e-05, "loss": 1.9460446166992187, "step": 153100 }, { "epoch": 0.6093006370503722, "grad_norm": 0.6691925525665283, "learning_rate": 1.3909028700557781e-05, "loss": 1.9875027465820312, "step": 153150 }, { "epoch": 0.6094995598832323, "grad_norm": 0.572543740272522, "learning_rate": 1.3907039474207693e-05, "loss": 1.9387037658691406, "step": 153200 }, { "epoch": 0.6096984827160924, "grad_norm": 0.6765785217285156, "learning_rate": 1.3905050247857604e-05, "loss": 1.9452546691894532, "step": 153250 }, { "epoch": 0.6098974055489524, "grad_norm": 0.6035070419311523, "learning_rate": 1.3903061021507516e-05, "loss": 1.9753367614746093, "step": 153300 }, { "epoch": 0.6100963283818125, "grad_norm": 0.6081525683403015, "learning_rate": 1.3901071795157428e-05, "loss": 1.9861872863769532, "step": 153350 }, { "epoch": 0.6102952512146725, "grad_norm": 0.6007336378097534, "learning_rate": 1.389908256880734e-05, "loss": 1.9634709167480469, "step": 153400 }, { "epoch": 0.6104941740475326, "grad_norm": 0.6621211171150208, "learning_rate": 1.3897093342457252e-05, "loss": 1.9429893493652344, "step": 153450 }, { "epoch": 0.6106930968803926, "grad_norm": 0.5763882994651794, "learning_rate": 1.3895104116107166e-05, "loss": 1.9602247619628905, "step": 153500 }, { "epoch": 0.6108920197132527, "grad_norm": 0.59560626745224, "learning_rate": 1.3893114889757078e-05, "loss": 1.961391143798828, "step": 153550 }, { "epoch": 0.6110909425461128, "grad_norm": 0.47044461965560913, "learning_rate": 1.3891125663406989e-05, "loss": 2.0008502197265625, "step": 153600 }, { "epoch": 0.6112898653789729, "grad_norm": 0.601344108581543, "learning_rate": 1.38891364370569e-05, "loss": 2.0047511291503906, "step": 153650 }, { "epoch": 0.611488788211833, "grad_norm": 0.6279007792472839, "learning_rate": 1.3887147210706813e-05, "loss": 1.9629794311523439, "step": 153700 }, { "epoch": 0.611687711044693, "grad_norm": 0.5837147831916809, "learning_rate": 1.3885157984356725e-05, "loss": 1.929511260986328, "step": 153750 }, { "epoch": 0.611886633877553, "grad_norm": 0.5068081617355347, "learning_rate": 1.3883168758006637e-05, "loss": 1.9458229064941406, "step": 153800 }, { "epoch": 0.6120855567104131, "grad_norm": 0.6303746104240417, "learning_rate": 1.388117953165655e-05, "loss": 1.9160655212402344, "step": 153850 }, { "epoch": 0.6122844795432731, "grad_norm": 0.5338422656059265, "learning_rate": 1.387919030530646e-05, "loss": 1.9931301879882812, "step": 153900 }, { "epoch": 0.6124834023761332, "grad_norm": 0.6389930844306946, "learning_rate": 1.3877201078956372e-05, "loss": 2.0371533203125, "step": 153950 }, { "epoch": 0.6126823252089934, "grad_norm": 0.6043959856033325, "learning_rate": 1.3875251637133287e-05, "loss": 1.9849395751953125, "step": 154000 }, { "epoch": 0.6128812480418534, "grad_norm": 0.5653139352798462, "learning_rate": 1.38732624107832e-05, "loss": 2.0354669189453123, "step": 154050 }, { "epoch": 0.6130801708747134, "grad_norm": 0.5805498361587524, "learning_rate": 1.3871273184433111e-05, "loss": 1.9823686218261718, "step": 154100 }, { "epoch": 0.6132790937075735, "grad_norm": 0.5365418791770935, "learning_rate": 1.3869283958083024e-05, "loss": 1.9407528686523436, "step": 154150 }, { "epoch": 0.6134780165404335, "grad_norm": 0.6680804491043091, "learning_rate": 1.3867294731732936e-05, "loss": 1.95293212890625, "step": 154200 }, { "epoch": 0.6136769393732936, "grad_norm": 0.6489595174789429, "learning_rate": 1.3865305505382846e-05, "loss": 1.9717840576171874, "step": 154250 }, { "epoch": 0.6138758622061536, "grad_norm": 0.599837601184845, "learning_rate": 1.3863316279032758e-05, "loss": 1.9688223266601563, "step": 154300 }, { "epoch": 0.6140747850390137, "grad_norm": 0.5932193398475647, "learning_rate": 1.3861327052682672e-05, "loss": 1.9774252319335937, "step": 154350 }, { "epoch": 0.6142737078718739, "grad_norm": 0.6093432903289795, "learning_rate": 1.3859337826332584e-05, "loss": 1.9454515075683594, "step": 154400 }, { "epoch": 0.6144726307047339, "grad_norm": 0.47541913390159607, "learning_rate": 1.3857348599982496e-05, "loss": 1.9690974426269532, "step": 154450 }, { "epoch": 0.614671553537594, "grad_norm": 0.6271063089370728, "learning_rate": 1.3855359373632409e-05, "loss": 1.9798220825195312, "step": 154500 }, { "epoch": 0.614870476370454, "grad_norm": 0.5732753276824951, "learning_rate": 1.3853370147282319e-05, "loss": 1.9431132507324218, "step": 154550 }, { "epoch": 0.615069399203314, "grad_norm": 0.5640801787376404, "learning_rate": 1.3851380920932231e-05, "loss": 1.9975662231445312, "step": 154600 }, { "epoch": 0.6152683220361741, "grad_norm": 0.5717364549636841, "learning_rate": 1.3849391694582143e-05, "loss": 1.9901031494140624, "step": 154650 }, { "epoch": 0.6154672448690341, "grad_norm": 0.6237818002700806, "learning_rate": 1.3847402468232056e-05, "loss": 1.9289057922363282, "step": 154700 }, { "epoch": 0.6156661677018942, "grad_norm": 0.5138419270515442, "learning_rate": 1.384541324188197e-05, "loss": 2.0288275146484374, "step": 154750 }, { "epoch": 0.6158650905347544, "grad_norm": 0.5367547869682312, "learning_rate": 1.3843424015531881e-05, "loss": 1.9554293823242188, "step": 154800 }, { "epoch": 0.6160640133676144, "grad_norm": 0.5017540454864502, "learning_rate": 1.3841434789181794e-05, "loss": 1.88893798828125, "step": 154850 }, { "epoch": 0.6162629362004745, "grad_norm": 0.6695616841316223, "learning_rate": 1.3839445562831704e-05, "loss": 1.9883111572265626, "step": 154900 }, { "epoch": 0.6164618590333345, "grad_norm": 0.6751031875610352, "learning_rate": 1.3837456336481616e-05, "loss": 1.9704598999023437, "step": 154950 }, { "epoch": 0.6166607818661946, "grad_norm": 0.6202064752578735, "learning_rate": 1.3835467110131528e-05, "loss": 1.9968853759765626, "step": 155000 }, { "epoch": 0.6168597046990546, "grad_norm": 0.6449732184410095, "learning_rate": 1.383347788378144e-05, "loss": 1.954940185546875, "step": 155050 }, { "epoch": 0.6170586275319146, "grad_norm": 0.6273284554481506, "learning_rate": 1.3831488657431353e-05, "loss": 1.953150177001953, "step": 155100 }, { "epoch": 0.6172575503647747, "grad_norm": 0.7410906553268433, "learning_rate": 1.3829499431081267e-05, "loss": 1.9890760803222656, "step": 155150 }, { "epoch": 0.6174564731976349, "grad_norm": 0.6304149627685547, "learning_rate": 1.3827510204731175e-05, "loss": 1.9812481689453125, "step": 155200 }, { "epoch": 0.6176553960304949, "grad_norm": 0.5794317126274109, "learning_rate": 1.3825520978381089e-05, "loss": 1.994043731689453, "step": 155250 }, { "epoch": 0.617854318863355, "grad_norm": 0.5603783130645752, "learning_rate": 1.3823531752031001e-05, "loss": 1.934547882080078, "step": 155300 }, { "epoch": 0.618053241696215, "grad_norm": 0.6409780383110046, "learning_rate": 1.3821542525680913e-05, "loss": 1.9602851867675781, "step": 155350 }, { "epoch": 0.618252164529075, "grad_norm": 0.5002322196960449, "learning_rate": 1.3819553299330826e-05, "loss": 1.9861238098144531, "step": 155400 }, { "epoch": 0.6184510873619351, "grad_norm": 0.6622902154922485, "learning_rate": 1.3817564072980738e-05, "loss": 1.9686058044433594, "step": 155450 }, { "epoch": 0.6186500101947952, "grad_norm": 0.571983814239502, "learning_rate": 1.381557484663065e-05, "loss": 1.972593994140625, "step": 155500 }, { "epoch": 0.6188489330276552, "grad_norm": 0.6237232089042664, "learning_rate": 1.381358562028056e-05, "loss": 1.9281968688964843, "step": 155550 }, { "epoch": 0.6190478558605154, "grad_norm": 0.7116290926933289, "learning_rate": 1.3811596393930472e-05, "loss": 1.9601031494140626, "step": 155600 }, { "epoch": 0.6192467786933754, "grad_norm": 0.562343180179596, "learning_rate": 1.3809607167580386e-05, "loss": 2.0049423217773437, "step": 155650 }, { "epoch": 0.6194457015262355, "grad_norm": 0.5841497778892517, "learning_rate": 1.3807617941230298e-05, "loss": 1.991405029296875, "step": 155700 }, { "epoch": 0.6196446243590955, "grad_norm": 0.6492112278938293, "learning_rate": 1.380562871488021e-05, "loss": 1.9755014038085938, "step": 155750 }, { "epoch": 0.6198435471919556, "grad_norm": 0.5975133180618286, "learning_rate": 1.3803639488530123e-05, "loss": 1.9753509521484376, "step": 155800 }, { "epoch": 0.6200424700248156, "grad_norm": 0.5992507934570312, "learning_rate": 1.3801650262180033e-05, "loss": 1.9894883728027344, "step": 155850 }, { "epoch": 0.6202413928576757, "grad_norm": 0.3753006160259247, "learning_rate": 1.3799661035829945e-05, "loss": 1.999464111328125, "step": 155900 }, { "epoch": 0.6204403156905357, "grad_norm": 0.6329013109207153, "learning_rate": 1.3797671809479857e-05, "loss": 1.9590789794921875, "step": 155950 }, { "epoch": 0.6206392385233959, "grad_norm": 0.7049049735069275, "learning_rate": 1.3795722367656773e-05, "loss": 1.9884858703613282, "step": 156000 }, { "epoch": 0.6208381613562559, "grad_norm": 0.6807284355163574, "learning_rate": 1.3793733141306685e-05, "loss": 1.982486572265625, "step": 156050 }, { "epoch": 0.621037084189116, "grad_norm": 0.5222828984260559, "learning_rate": 1.3791743914956597e-05, "loss": 1.983516845703125, "step": 156100 }, { "epoch": 0.621236007021976, "grad_norm": 0.6312127709388733, "learning_rate": 1.3789754688606509e-05, "loss": 1.9816561889648439, "step": 156150 }, { "epoch": 0.6214349298548361, "grad_norm": 0.6147653460502625, "learning_rate": 1.378776546225642e-05, "loss": 1.992464599609375, "step": 156200 }, { "epoch": 0.6216338526876961, "grad_norm": 0.7008287310600281, "learning_rate": 1.3785776235906332e-05, "loss": 1.9541145324707032, "step": 156250 }, { "epoch": 0.6218327755205562, "grad_norm": 0.6576167345046997, "learning_rate": 1.3783787009556244e-05, "loss": 1.966927032470703, "step": 156300 }, { "epoch": 0.6220316983534162, "grad_norm": 0.5392459630966187, "learning_rate": 1.3781797783206156e-05, "loss": 1.98378662109375, "step": 156350 }, { "epoch": 0.6222306211862764, "grad_norm": 0.6290916204452515, "learning_rate": 1.377980855685607e-05, "loss": 2.010179290771484, "step": 156400 }, { "epoch": 0.6224295440191364, "grad_norm": 0.6019361019134521, "learning_rate": 1.3777819330505982e-05, "loss": 1.9549087524414062, "step": 156450 }, { "epoch": 0.6226284668519965, "grad_norm": 0.531577467918396, "learning_rate": 1.3775830104155892e-05, "loss": 1.9895857238769532, "step": 156500 }, { "epoch": 0.6228273896848565, "grad_norm": 0.6078243851661682, "learning_rate": 1.3773840877805805e-05, "loss": 1.9445956420898438, "step": 156550 }, { "epoch": 0.6230263125177166, "grad_norm": 0.6565086841583252, "learning_rate": 1.3771851651455717e-05, "loss": 1.987383575439453, "step": 156600 }, { "epoch": 0.6232252353505766, "grad_norm": 0.7622785568237305, "learning_rate": 1.3769862425105629e-05, "loss": 1.9731912231445312, "step": 156650 }, { "epoch": 0.6234241581834367, "grad_norm": 0.6362915635108948, "learning_rate": 1.3767873198755541e-05, "loss": 2.0166250610351564, "step": 156700 }, { "epoch": 0.6236230810162967, "grad_norm": 0.5978673696517944, "learning_rate": 1.3765883972405453e-05, "loss": 1.9439886474609376, "step": 156750 }, { "epoch": 0.6238220038491569, "grad_norm": 0.6072251200675964, "learning_rate": 1.3763894746055367e-05, "loss": 2.0400074768066405, "step": 156800 }, { "epoch": 0.6240209266820169, "grad_norm": 0.5833436846733093, "learning_rate": 1.3761905519705276e-05, "loss": 1.9843629455566407, "step": 156850 }, { "epoch": 0.624219849514877, "grad_norm": 0.5668893456459045, "learning_rate": 1.375991629335519e-05, "loss": 1.9931692504882812, "step": 156900 }, { "epoch": 0.624418772347737, "grad_norm": 0.6226190328598022, "learning_rate": 1.3757927067005102e-05, "loss": 1.986484375, "step": 156950 }, { "epoch": 0.6246176951805971, "grad_norm": 0.5825393795967102, "learning_rate": 1.3755937840655014e-05, "loss": 1.9792022705078125, "step": 157000 }, { "epoch": 0.6248166180134571, "grad_norm": 0.5347006320953369, "learning_rate": 1.3753948614304926e-05, "loss": 1.922603759765625, "step": 157050 }, { "epoch": 0.6250155408463172, "grad_norm": 0.549511730670929, "learning_rate": 1.3751959387954838e-05, "loss": 1.9753103637695313, "step": 157100 }, { "epoch": 0.6252144636791772, "grad_norm": 0.6394692063331604, "learning_rate": 1.3749970161604749e-05, "loss": 1.93205810546875, "step": 157150 }, { "epoch": 0.6254133865120374, "grad_norm": 0.6414468884468079, "learning_rate": 1.374798093525466e-05, "loss": 1.9401458740234374, "step": 157200 }, { "epoch": 0.6256123093448974, "grad_norm": 0.6285583972930908, "learning_rate": 1.3745991708904573e-05, "loss": 1.9493617248535156, "step": 157250 }, { "epoch": 0.6258112321777575, "grad_norm": 0.6149645447731018, "learning_rate": 1.3744002482554487e-05, "loss": 1.9887678527832031, "step": 157300 }, { "epoch": 0.6260101550106175, "grad_norm": 0.6733729839324951, "learning_rate": 1.3742013256204399e-05, "loss": 1.96771484375, "step": 157350 }, { "epoch": 0.6262090778434776, "grad_norm": 0.552245557308197, "learning_rate": 1.3740024029854311e-05, "loss": 1.9087355041503906, "step": 157400 }, { "epoch": 0.6264080006763376, "grad_norm": 0.6274532675743103, "learning_rate": 1.3738034803504223e-05, "loss": 1.9755638122558594, "step": 157450 }, { "epoch": 0.6266069235091977, "grad_norm": 0.5683292746543884, "learning_rate": 1.3736045577154134e-05, "loss": 1.9658139038085938, "step": 157500 }, { "epoch": 0.6268058463420577, "grad_norm": 0.6048158407211304, "learning_rate": 1.3734056350804046e-05, "loss": 1.963988037109375, "step": 157550 }, { "epoch": 0.6270047691749179, "grad_norm": 0.5505148768424988, "learning_rate": 1.3732067124453958e-05, "loss": 1.979142303466797, "step": 157600 }, { "epoch": 0.6272036920077779, "grad_norm": 0.5433773994445801, "learning_rate": 1.373007789810387e-05, "loss": 1.9134445190429688, "step": 157650 }, { "epoch": 0.627402614840638, "grad_norm": 0.40421515703201294, "learning_rate": 1.3728088671753784e-05, "loss": 1.9981483459472655, "step": 157700 }, { "epoch": 0.627601537673498, "grad_norm": 0.7292563319206238, "learning_rate": 1.3726099445403696e-05, "loss": 2.0216238403320315, "step": 157750 }, { "epoch": 0.6278004605063581, "grad_norm": 0.5560089349746704, "learning_rate": 1.3724110219053607e-05, "loss": 1.9130973815917969, "step": 157800 }, { "epoch": 0.6279993833392181, "grad_norm": 0.5396754741668701, "learning_rate": 1.3722120992703519e-05, "loss": 1.971195831298828, "step": 157850 }, { "epoch": 0.6281983061720782, "grad_norm": 0.5565981268882751, "learning_rate": 1.372013176635343e-05, "loss": 1.9775306701660156, "step": 157900 }, { "epoch": 0.6283972290049382, "grad_norm": 0.6864540576934814, "learning_rate": 1.3718142540003343e-05, "loss": 1.9809454345703126, "step": 157950 }, { "epoch": 0.6285961518377983, "grad_norm": 0.7168575525283813, "learning_rate": 1.3716153313653255e-05, "loss": 1.9472203063964844, "step": 158000 }, { "epoch": 0.6287950746706584, "grad_norm": 0.692366361618042, "learning_rate": 1.3714164087303167e-05, "loss": 1.9985476684570314, "step": 158050 }, { "epoch": 0.6289939975035185, "grad_norm": 0.5716967582702637, "learning_rate": 1.3712174860953081e-05, "loss": 1.9684347534179687, "step": 158100 }, { "epoch": 0.6291929203363785, "grad_norm": 0.5094587802886963, "learning_rate": 1.371018563460299e-05, "loss": 1.96978515625, "step": 158150 }, { "epoch": 0.6293918431692386, "grad_norm": 0.5822829008102417, "learning_rate": 1.3708196408252904e-05, "loss": 2.0047148132324217, "step": 158200 }, { "epoch": 0.6295907660020986, "grad_norm": 0.6467392444610596, "learning_rate": 1.3706207181902816e-05, "loss": 1.9657138061523438, "step": 158250 }, { "epoch": 0.6297896888349587, "grad_norm": 0.7410619258880615, "learning_rate": 1.3704217955552728e-05, "loss": 1.9019351196289063, "step": 158300 }, { "epoch": 0.6299886116678187, "grad_norm": 0.5051903128623962, "learning_rate": 1.370222872920264e-05, "loss": 1.9634291076660155, "step": 158350 }, { "epoch": 0.6301875345006788, "grad_norm": 0.5895523428916931, "learning_rate": 1.3700239502852552e-05, "loss": 1.9402262878417968, "step": 158400 }, { "epoch": 0.6303864573335389, "grad_norm": 0.6393479108810425, "learning_rate": 1.3698250276502463e-05, "loss": 1.9650418090820312, "step": 158450 }, { "epoch": 0.630585380166399, "grad_norm": 0.558975875377655, "learning_rate": 1.3696261050152375e-05, "loss": 1.9953575134277344, "step": 158500 }, { "epoch": 0.630784302999259, "grad_norm": 0.6497620344161987, "learning_rate": 1.3694271823802287e-05, "loss": 1.9453446960449219, "step": 158550 }, { "epoch": 0.6309832258321191, "grad_norm": 0.7111606597900391, "learning_rate": 1.36922825974522e-05, "loss": 2.0140948486328125, "step": 158600 }, { "epoch": 0.6311821486649791, "grad_norm": 0.5650834441184998, "learning_rate": 1.3690293371102113e-05, "loss": 1.9637544250488281, "step": 158650 }, { "epoch": 0.6313810714978392, "grad_norm": 0.5591875910758972, "learning_rate": 1.3688304144752025e-05, "loss": 1.9615951538085938, "step": 158700 }, { "epoch": 0.6315799943306992, "grad_norm": 0.6212294101715088, "learning_rate": 1.3686314918401937e-05, "loss": 2.013516082763672, "step": 158750 }, { "epoch": 0.6317789171635593, "grad_norm": 0.549564778804779, "learning_rate": 1.3684325692051848e-05, "loss": 1.9685975646972655, "step": 158800 }, { "epoch": 0.6319778399964194, "grad_norm": 0.5179488658905029, "learning_rate": 1.368233646570176e-05, "loss": 1.9797938537597657, "step": 158850 }, { "epoch": 0.6321767628292795, "grad_norm": 0.6709555387496948, "learning_rate": 1.3680347239351672e-05, "loss": 1.887007598876953, "step": 158900 }, { "epoch": 0.6323756856621395, "grad_norm": 0.5698062777519226, "learning_rate": 1.3678397797528587e-05, "loss": 1.958289794921875, "step": 158950 }, { "epoch": 0.6325746084949996, "grad_norm": 0.558707058429718, "learning_rate": 1.36764085711785e-05, "loss": 1.9535995483398438, "step": 159000 }, { "epoch": 0.6327735313278596, "grad_norm": 0.6379057765007019, "learning_rate": 1.3674419344828412e-05, "loss": 1.9620584106445313, "step": 159050 }, { "epoch": 0.6329724541607197, "grad_norm": 0.542914867401123, "learning_rate": 1.3672430118478322e-05, "loss": 1.9624325561523437, "step": 159100 }, { "epoch": 0.6331713769935797, "grad_norm": 0.5452579855918884, "learning_rate": 1.3670440892128234e-05, "loss": 1.9534458923339844, "step": 159150 }, { "epoch": 0.6333702998264398, "grad_norm": 0.5965121388435364, "learning_rate": 1.3668451665778146e-05, "loss": 1.966311492919922, "step": 159200 }, { "epoch": 0.6335692226592999, "grad_norm": 0.5390868186950684, "learning_rate": 1.3666462439428058e-05, "loss": 1.896241455078125, "step": 159250 }, { "epoch": 0.63376814549216, "grad_norm": 0.502065122127533, "learning_rate": 1.366447321307797e-05, "loss": 1.940760040283203, "step": 159300 }, { "epoch": 0.63396706832502, "grad_norm": 0.49417397379875183, "learning_rate": 1.3662483986727884e-05, "loss": 2.0115252685546876, "step": 159350 }, { "epoch": 0.6341659911578801, "grad_norm": 0.5887784957885742, "learning_rate": 1.3660494760377797e-05, "loss": 1.9602981567382813, "step": 159400 }, { "epoch": 0.6343649139907401, "grad_norm": 0.6162486672401428, "learning_rate": 1.3658505534027707e-05, "loss": 1.9821292114257814, "step": 159450 }, { "epoch": 0.6345638368236002, "grad_norm": 0.40275248885154724, "learning_rate": 1.365651630767762e-05, "loss": 2.0109417724609373, "step": 159500 }, { "epoch": 0.6347627596564602, "grad_norm": 0.5783175826072693, "learning_rate": 1.3654527081327531e-05, "loss": 1.932457275390625, "step": 159550 }, { "epoch": 0.6349616824893203, "grad_norm": 0.6268549561500549, "learning_rate": 1.3652537854977443e-05, "loss": 1.9634390258789063, "step": 159600 }, { "epoch": 0.6351606053221804, "grad_norm": 0.590587854385376, "learning_rate": 1.3650548628627356e-05, "loss": 2.0282058715820312, "step": 159650 }, { "epoch": 0.6353595281550405, "grad_norm": 0.674794614315033, "learning_rate": 1.3648559402277268e-05, "loss": 1.9944783020019532, "step": 159700 }, { "epoch": 0.6355584509879005, "grad_norm": 0.6209328174591064, "learning_rate": 1.3646570175927178e-05, "loss": 2.030038146972656, "step": 159750 }, { "epoch": 0.6357573738207606, "grad_norm": 0.5266450643539429, "learning_rate": 1.364458094957709e-05, "loss": 1.96729736328125, "step": 159800 }, { "epoch": 0.6359562966536206, "grad_norm": 0.6151196956634521, "learning_rate": 1.3642591723227002e-05, "loss": 1.9318782043457032, "step": 159850 }, { "epoch": 0.6361552194864807, "grad_norm": 0.56081622838974, "learning_rate": 1.3640602496876916e-05, "loss": 1.9736495971679688, "step": 159900 }, { "epoch": 0.6363541423193407, "grad_norm": 0.5674142837524414, "learning_rate": 1.3638613270526828e-05, "loss": 1.9854104614257813, "step": 159950 }, { "epoch": 0.6365530651522008, "grad_norm": 0.6442809104919434, "learning_rate": 1.363662404417674e-05, "loss": 1.9086358642578125, "step": 160000 }, { "epoch": 0.6367519879850609, "grad_norm": 0.6740526556968689, "learning_rate": 1.3634634817826653e-05, "loss": 2.010825958251953, "step": 160050 }, { "epoch": 0.636950910817921, "grad_norm": 0.8030680418014526, "learning_rate": 1.3632645591476563e-05, "loss": 1.9780972290039063, "step": 160100 }, { "epoch": 0.637149833650781, "grad_norm": 0.6276201009750366, "learning_rate": 1.3630656365126475e-05, "loss": 1.9937643432617187, "step": 160150 }, { "epoch": 0.6373487564836411, "grad_norm": 0.5377590656280518, "learning_rate": 1.3628667138776388e-05, "loss": 1.9631468200683593, "step": 160200 }, { "epoch": 0.6375476793165011, "grad_norm": 0.5827679634094238, "learning_rate": 1.36266779124263e-05, "loss": 1.960668182373047, "step": 160250 }, { "epoch": 0.6377466021493612, "grad_norm": 0.5357514023780823, "learning_rate": 1.3624688686076213e-05, "loss": 1.9914674377441406, "step": 160300 }, { "epoch": 0.6379455249822212, "grad_norm": 0.5550944209098816, "learning_rate": 1.3622699459726126e-05, "loss": 2.008896789550781, "step": 160350 }, { "epoch": 0.6381444478150813, "grad_norm": 0.5841155648231506, "learning_rate": 1.3620710233376036e-05, "loss": 1.9855868530273437, "step": 160400 }, { "epoch": 0.6383433706479414, "grad_norm": 0.6229991912841797, "learning_rate": 1.3618721007025948e-05, "loss": 1.9886219787597657, "step": 160450 }, { "epoch": 0.6385422934808015, "grad_norm": 0.6183667182922363, "learning_rate": 1.361673178067586e-05, "loss": 1.9799977111816407, "step": 160500 }, { "epoch": 0.6387412163136615, "grad_norm": 0.6137898564338684, "learning_rate": 1.3614742554325773e-05, "loss": 1.9930937194824219, "step": 160550 }, { "epoch": 0.6389401391465216, "grad_norm": 0.6494871377944946, "learning_rate": 1.3612753327975685e-05, "loss": 2.0099603271484376, "step": 160600 }, { "epoch": 0.6391390619793816, "grad_norm": 0.5555556416511536, "learning_rate": 1.3610764101625597e-05, "loss": 1.955733642578125, "step": 160650 }, { "epoch": 0.6393379848122417, "grad_norm": 0.6346853375434875, "learning_rate": 1.360877487527551e-05, "loss": 1.977442626953125, "step": 160700 }, { "epoch": 0.6395369076451017, "grad_norm": 0.6672390699386597, "learning_rate": 1.360678564892542e-05, "loss": 1.9280821228027343, "step": 160750 }, { "epoch": 0.6397358304779618, "grad_norm": 0.5407641530036926, "learning_rate": 1.3604796422575333e-05, "loss": 1.9569412231445313, "step": 160800 }, { "epoch": 0.639934753310822, "grad_norm": 0.5845869779586792, "learning_rate": 1.3602807196225245e-05, "loss": 1.9714765930175782, "step": 160850 }, { "epoch": 0.640133676143682, "grad_norm": 0.6333935260772705, "learning_rate": 1.3600817969875158e-05, "loss": 1.9671722412109376, "step": 160900 }, { "epoch": 0.640332598976542, "grad_norm": 0.5269621014595032, "learning_rate": 1.359882874352507e-05, "loss": 1.9592607116699219, "step": 160950 }, { "epoch": 0.6405315218094021, "grad_norm": 0.4764185845851898, "learning_rate": 1.3596839517174982e-05, "loss": 1.945287628173828, "step": 161000 }, { "epoch": 0.6407304446422621, "grad_norm": 0.540081799030304, "learning_rate": 1.3594850290824892e-05, "loss": 2.0270420837402345, "step": 161050 }, { "epoch": 0.6409293674751222, "grad_norm": 0.6749238967895508, "learning_rate": 1.3592861064474804e-05, "loss": 1.9052273559570312, "step": 161100 }, { "epoch": 0.6411282903079822, "grad_norm": 0.5474725365638733, "learning_rate": 1.3590871838124717e-05, "loss": 1.913453369140625, "step": 161150 }, { "epoch": 0.6413272131408423, "grad_norm": 0.48960864543914795, "learning_rate": 1.358888261177463e-05, "loss": 1.9057579040527344, "step": 161200 }, { "epoch": 0.6415261359737024, "grad_norm": 0.6103407144546509, "learning_rate": 1.3586893385424543e-05, "loss": 1.9393424987792969, "step": 161250 }, { "epoch": 0.6417250588065625, "grad_norm": 0.556272566318512, "learning_rate": 1.3584904159074455e-05, "loss": 1.9571670532226562, "step": 161300 }, { "epoch": 0.6419239816394225, "grad_norm": 0.5669354796409607, "learning_rate": 1.3582914932724367e-05, "loss": 1.8965263366699219, "step": 161350 }, { "epoch": 0.6421229044722826, "grad_norm": 0.589733898639679, "learning_rate": 1.3580925706374277e-05, "loss": 1.9431239318847657, "step": 161400 }, { "epoch": 0.6423218273051426, "grad_norm": 0.5129130482673645, "learning_rate": 1.357897626455119e-05, "loss": 1.9239097595214845, "step": 161450 }, { "epoch": 0.6425207501380027, "grad_norm": 0.5446606874465942, "learning_rate": 1.3576987038201103e-05, "loss": 1.9092300415039063, "step": 161500 }, { "epoch": 0.6427196729708627, "grad_norm": 0.4678652584552765, "learning_rate": 1.3574997811851017e-05, "loss": 1.9085264587402344, "step": 161550 }, { "epoch": 0.6429185958037228, "grad_norm": 0.7246032953262329, "learning_rate": 1.3573008585500929e-05, "loss": 1.9802449035644532, "step": 161600 }, { "epoch": 0.643117518636583, "grad_norm": 0.5935914516448975, "learning_rate": 1.3571019359150841e-05, "loss": 1.9962054443359376, "step": 161650 }, { "epoch": 0.643316441469443, "grad_norm": 0.5697070956230164, "learning_rate": 1.3569030132800752e-05, "loss": 1.9499203491210937, "step": 161700 }, { "epoch": 0.643515364302303, "grad_norm": 0.6315224170684814, "learning_rate": 1.3567040906450664e-05, "loss": 1.9687602233886718, "step": 161750 }, { "epoch": 0.6437142871351631, "grad_norm": 0.627393364906311, "learning_rate": 1.3565051680100576e-05, "loss": 1.963781280517578, "step": 161800 }, { "epoch": 0.6439132099680231, "grad_norm": 0.5293607711791992, "learning_rate": 1.3563062453750488e-05, "loss": 1.9837770080566406, "step": 161850 }, { "epoch": 0.6441121328008832, "grad_norm": 0.616712212562561, "learning_rate": 1.35610732274004e-05, "loss": 2.001148986816406, "step": 161900 }, { "epoch": 0.6443110556337432, "grad_norm": 0.6321412920951843, "learning_rate": 1.3559084001050314e-05, "loss": 2.0034397888183593, "step": 161950 }, { "epoch": 0.6445099784666033, "grad_norm": 0.5694063901901245, "learning_rate": 1.3557094774700226e-05, "loss": 1.8969606018066407, "step": 162000 }, { "epoch": 0.6447089012994635, "grad_norm": 0.5974847078323364, "learning_rate": 1.3555105548350137e-05, "loss": 1.9922608947753906, "step": 162050 }, { "epoch": 0.6449078241323235, "grad_norm": 0.5921640992164612, "learning_rate": 1.3553116322000049e-05, "loss": 1.9358987426757812, "step": 162100 }, { "epoch": 0.6451067469651836, "grad_norm": 0.5600255727767944, "learning_rate": 1.3551127095649961e-05, "loss": 1.9700363159179688, "step": 162150 }, { "epoch": 0.6453056697980436, "grad_norm": 0.5767713189125061, "learning_rate": 1.3549137869299873e-05, "loss": 1.9175732421875, "step": 162200 }, { "epoch": 0.6455045926309037, "grad_norm": 0.5787684917449951, "learning_rate": 1.3547148642949785e-05, "loss": 2.0032427978515623, "step": 162250 }, { "epoch": 0.6457035154637637, "grad_norm": 0.6167246103286743, "learning_rate": 1.3545159416599697e-05, "loss": 2.0235955810546873, "step": 162300 }, { "epoch": 0.6459024382966237, "grad_norm": 0.6824870705604553, "learning_rate": 1.3543170190249608e-05, "loss": 1.9914581298828125, "step": 162350 }, { "epoch": 0.6461013611294838, "grad_norm": 0.626855194568634, "learning_rate": 1.354118096389952e-05, "loss": 1.9880133056640625, "step": 162400 }, { "epoch": 0.646300283962344, "grad_norm": 0.5489233732223511, "learning_rate": 1.3539191737549434e-05, "loss": 2.0238383483886717, "step": 162450 }, { "epoch": 0.646499206795204, "grad_norm": 0.5917928814888, "learning_rate": 1.3537202511199346e-05, "loss": 1.9801116943359376, "step": 162500 }, { "epoch": 0.6466981296280641, "grad_norm": 0.571596086025238, "learning_rate": 1.3535213284849258e-05, "loss": 1.9460870361328124, "step": 162550 }, { "epoch": 0.6468970524609241, "grad_norm": 0.5819786190986633, "learning_rate": 1.353322405849917e-05, "loss": 2.025959167480469, "step": 162600 }, { "epoch": 0.6470959752937842, "grad_norm": 0.693871259689331, "learning_rate": 1.3531234832149082e-05, "loss": 1.9886447143554689, "step": 162650 }, { "epoch": 0.6472948981266442, "grad_norm": 0.6800723075866699, "learning_rate": 1.3529245605798993e-05, "loss": 1.9484031677246094, "step": 162700 }, { "epoch": 0.6474938209595043, "grad_norm": 0.6218690276145935, "learning_rate": 1.3527256379448905e-05, "loss": 1.9636993408203125, "step": 162750 }, { "epoch": 0.6476927437923643, "grad_norm": 0.5829600691795349, "learning_rate": 1.3525267153098817e-05, "loss": 2.030223693847656, "step": 162800 }, { "epoch": 0.6478916666252245, "grad_norm": 0.569266140460968, "learning_rate": 1.3523277926748731e-05, "loss": 2.017757568359375, "step": 162850 }, { "epoch": 0.6480905894580845, "grad_norm": 0.6147276759147644, "learning_rate": 1.3521288700398643e-05, "loss": 1.9567051696777344, "step": 162900 }, { "epoch": 0.6482895122909446, "grad_norm": 0.6733900308609009, "learning_rate": 1.3519299474048555e-05, "loss": 1.9895594787597657, "step": 162950 }, { "epoch": 0.6484884351238046, "grad_norm": 0.5948969125747681, "learning_rate": 1.3517310247698466e-05, "loss": 1.9653315734863281, "step": 163000 }, { "epoch": 0.6486873579566647, "grad_norm": 0.6040372848510742, "learning_rate": 1.3515321021348378e-05, "loss": 1.9854098510742189, "step": 163050 }, { "epoch": 0.6488862807895247, "grad_norm": 0.5647464394569397, "learning_rate": 1.351333179499829e-05, "loss": 1.9886361694335937, "step": 163100 }, { "epoch": 0.6490852036223848, "grad_norm": 0.5443736910820007, "learning_rate": 1.3511342568648202e-05, "loss": 1.9872012329101563, "step": 163150 }, { "epoch": 0.6492841264552448, "grad_norm": 0.492927223443985, "learning_rate": 1.3509353342298114e-05, "loss": 1.9662643432617188, "step": 163200 }, { "epoch": 0.649483049288105, "grad_norm": 0.47702544927597046, "learning_rate": 1.3507364115948028e-05, "loss": 1.941048126220703, "step": 163250 }, { "epoch": 0.649681972120965, "grad_norm": 0.6579359173774719, "learning_rate": 1.350537488959794e-05, "loss": 1.94361083984375, "step": 163300 }, { "epoch": 0.6498808949538251, "grad_norm": 0.5335484743118286, "learning_rate": 1.350338566324785e-05, "loss": 1.9997669982910156, "step": 163350 }, { "epoch": 0.6500798177866851, "grad_norm": 0.6637337803840637, "learning_rate": 1.3501396436897763e-05, "loss": 1.910639190673828, "step": 163400 }, { "epoch": 0.6502787406195452, "grad_norm": 0.6091169118881226, "learning_rate": 1.3499407210547675e-05, "loss": 1.9579788208007813, "step": 163450 }, { "epoch": 0.6504776634524052, "grad_norm": 0.6824168562889099, "learning_rate": 1.3497417984197587e-05, "loss": 1.923697509765625, "step": 163500 }, { "epoch": 0.6506765862852653, "grad_norm": 0.6431562304496765, "learning_rate": 1.34954287578475e-05, "loss": 1.9202775573730468, "step": 163550 }, { "epoch": 0.6508755091181253, "grad_norm": 0.5020760297775269, "learning_rate": 1.3493439531497411e-05, "loss": 1.9865829467773437, "step": 163600 }, { "epoch": 0.6510744319509855, "grad_norm": 0.6798808574676514, "learning_rate": 1.3491450305147322e-05, "loss": 2.0009547424316407, "step": 163650 }, { "epoch": 0.6512733547838455, "grad_norm": 0.5825102925300598, "learning_rate": 1.3489461078797234e-05, "loss": 1.9075723266601563, "step": 163700 }, { "epoch": 0.6514722776167056, "grad_norm": 0.5377139449119568, "learning_rate": 1.3487471852447148e-05, "loss": 1.9672470092773438, "step": 163750 }, { "epoch": 0.6516712004495656, "grad_norm": 0.46384552121162415, "learning_rate": 1.3485522410624061e-05, "loss": 1.9615045166015626, "step": 163800 }, { "epoch": 0.6518701232824257, "grad_norm": 0.37120211124420166, "learning_rate": 1.3483533184273973e-05, "loss": 1.9669807434082032, "step": 163850 }, { "epoch": 0.6520690461152857, "grad_norm": 0.5860596299171448, "learning_rate": 1.3481543957923886e-05, "loss": 1.9449493408203125, "step": 163900 }, { "epoch": 0.6522679689481458, "grad_norm": 0.4850291311740875, "learning_rate": 1.3479554731573798e-05, "loss": 1.9367735290527344, "step": 163950 }, { "epoch": 0.6524668917810058, "grad_norm": 0.5289084315299988, "learning_rate": 1.3477565505223708e-05, "loss": 1.9376087951660157, "step": 164000 }, { "epoch": 0.652665814613866, "grad_norm": 0.5815974473953247, "learning_rate": 1.347557627887362e-05, "loss": 1.9806550598144532, "step": 164050 }, { "epoch": 0.652864737446726, "grad_norm": 0.68943852186203, "learning_rate": 1.3473587052523534e-05, "loss": 1.9574911499023437, "step": 164100 }, { "epoch": 0.6530636602795861, "grad_norm": 0.5552136898040771, "learning_rate": 1.3471597826173446e-05, "loss": 1.9546665954589844, "step": 164150 }, { "epoch": 0.6532625831124461, "grad_norm": 0.5323941707611084, "learning_rate": 1.3469608599823359e-05, "loss": 1.930424041748047, "step": 164200 }, { "epoch": 0.6534615059453062, "grad_norm": 0.6281698942184448, "learning_rate": 1.346761937347327e-05, "loss": 1.9715541076660157, "step": 164250 }, { "epoch": 0.6536604287781662, "grad_norm": 0.39841458201408386, "learning_rate": 1.3465630147123181e-05, "loss": 1.9847450256347656, "step": 164300 }, { "epoch": 0.6538593516110263, "grad_norm": 0.6627444624900818, "learning_rate": 1.3463640920773093e-05, "loss": 1.9538729858398438, "step": 164350 }, { "epoch": 0.6540582744438863, "grad_norm": 0.5379952788352966, "learning_rate": 1.3461651694423005e-05, "loss": 1.9198910522460937, "step": 164400 }, { "epoch": 0.6542571972767465, "grad_norm": 0.6468152403831482, "learning_rate": 1.3459662468072918e-05, "loss": 1.9744783020019532, "step": 164450 }, { "epoch": 0.6544561201096065, "grad_norm": 0.7479387521743774, "learning_rate": 1.3457673241722831e-05, "loss": 2.019791717529297, "step": 164500 }, { "epoch": 0.6546550429424666, "grad_norm": 0.6852051019668579, "learning_rate": 1.3455684015372744e-05, "loss": 2.0242691040039062, "step": 164550 }, { "epoch": 0.6548539657753266, "grad_norm": 0.6140834093093872, "learning_rate": 1.3453694789022656e-05, "loss": 1.9794888305664062, "step": 164600 }, { "epoch": 0.6550528886081867, "grad_norm": 0.5342567563056946, "learning_rate": 1.3451705562672566e-05, "loss": 1.9895552062988282, "step": 164650 }, { "epoch": 0.6552518114410467, "grad_norm": 0.6286404728889465, "learning_rate": 1.3449716336322478e-05, "loss": 1.9508856201171876, "step": 164700 }, { "epoch": 0.6554507342739068, "grad_norm": 0.5968277454376221, "learning_rate": 1.344772710997239e-05, "loss": 1.9308802795410156, "step": 164750 }, { "epoch": 0.6556496571067668, "grad_norm": 0.6471959948539734, "learning_rate": 1.3445737883622303e-05, "loss": 1.9352532958984374, "step": 164800 }, { "epoch": 0.655848579939627, "grad_norm": 0.6353601813316345, "learning_rate": 1.3443748657272215e-05, "loss": 1.9883406066894531, "step": 164850 }, { "epoch": 0.656047502772487, "grad_norm": 0.56463223695755, "learning_rate": 1.3441759430922129e-05, "loss": 2.0132080078125, "step": 164900 }, { "epoch": 0.6562464256053471, "grad_norm": 0.587805449962616, "learning_rate": 1.3439770204572037e-05, "loss": 1.9289572143554687, "step": 164950 }, { "epoch": 0.6564453484382071, "grad_norm": 0.5129795670509338, "learning_rate": 1.3437780978221951e-05, "loss": 1.979296417236328, "step": 165000 }, { "epoch": 0.6566442712710672, "grad_norm": 0.6435869932174683, "learning_rate": 1.3435791751871863e-05, "loss": 1.9966494750976562, "step": 165050 }, { "epoch": 0.6568431941039272, "grad_norm": 0.5822070837020874, "learning_rate": 1.3433802525521775e-05, "loss": 1.9834066772460937, "step": 165100 }, { "epoch": 0.6570421169367873, "grad_norm": 0.6550924181938171, "learning_rate": 1.3431813299171688e-05, "loss": 1.966595458984375, "step": 165150 }, { "epoch": 0.6572410397696473, "grad_norm": 0.6675074696540833, "learning_rate": 1.34298240728216e-05, "loss": 1.9788638305664064, "step": 165200 }, { "epoch": 0.6574399626025075, "grad_norm": 0.6465460658073425, "learning_rate": 1.3427834846471512e-05, "loss": 1.9515736389160157, "step": 165250 }, { "epoch": 0.6576388854353675, "grad_norm": 0.6127587556838989, "learning_rate": 1.3425845620121422e-05, "loss": 1.9750566101074218, "step": 165300 }, { "epoch": 0.6578378082682276, "grad_norm": 0.5290191173553467, "learning_rate": 1.3423856393771334e-05, "loss": 1.9512403869628907, "step": 165350 }, { "epoch": 0.6580367311010876, "grad_norm": 0.5971218347549438, "learning_rate": 1.3421867167421248e-05, "loss": 1.9460444641113281, "step": 165400 }, { "epoch": 0.6582356539339477, "grad_norm": 0.6609295606613159, "learning_rate": 1.341987794107116e-05, "loss": 1.9869764709472657, "step": 165450 }, { "epoch": 0.6584345767668077, "grad_norm": 0.5382452011108398, "learning_rate": 1.3417888714721073e-05, "loss": 1.9780580139160155, "step": 165500 }, { "epoch": 0.6586334995996678, "grad_norm": 0.6423865556716919, "learning_rate": 1.3415899488370985e-05, "loss": 1.9149395751953124, "step": 165550 }, { "epoch": 0.6588324224325278, "grad_norm": 0.539301872253418, "learning_rate": 1.3413910262020895e-05, "loss": 1.9187530517578124, "step": 165600 }, { "epoch": 0.6590313452653879, "grad_norm": 0.6119909286499023, "learning_rate": 1.3411921035670807e-05, "loss": 1.9090274047851563, "step": 165650 }, { "epoch": 0.659230268098248, "grad_norm": 0.7167896032333374, "learning_rate": 1.340993180932072e-05, "loss": 2.0266961669921875, "step": 165700 }, { "epoch": 0.6594291909311081, "grad_norm": 0.5827038884162903, "learning_rate": 1.3407942582970632e-05, "loss": 2.0176069641113283, "step": 165750 }, { "epoch": 0.6596281137639681, "grad_norm": 0.6521984338760376, "learning_rate": 1.3405953356620544e-05, "loss": 1.8924798583984375, "step": 165800 }, { "epoch": 0.6598270365968282, "grad_norm": 0.6141455173492432, "learning_rate": 1.3403964130270458e-05, "loss": 1.970614013671875, "step": 165850 }, { "epoch": 0.6600259594296882, "grad_norm": 0.5747098326683044, "learning_rate": 1.340197490392037e-05, "loss": 1.9276556396484374, "step": 165900 }, { "epoch": 0.6602248822625483, "grad_norm": 0.6931791305541992, "learning_rate": 1.339998567757028e-05, "loss": 1.9490968322753905, "step": 165950 }, { "epoch": 0.6604238050954083, "grad_norm": 0.5954591631889343, "learning_rate": 1.3398036235747194e-05, "loss": 1.9902989196777343, "step": 166000 }, { "epoch": 0.6606227279282684, "grad_norm": 0.6141789555549622, "learning_rate": 1.3396047009397106e-05, "loss": 2.002383575439453, "step": 166050 }, { "epoch": 0.6608216507611285, "grad_norm": 0.7226396799087524, "learning_rate": 1.3394057783047018e-05, "loss": 1.9316647338867188, "step": 166100 }, { "epoch": 0.6610205735939886, "grad_norm": 0.7044349312782288, "learning_rate": 1.3392068556696932e-05, "loss": 1.930000762939453, "step": 166150 }, { "epoch": 0.6612194964268486, "grad_norm": 0.5411977171897888, "learning_rate": 1.3390079330346844e-05, "loss": 1.9754911804199218, "step": 166200 }, { "epoch": 0.6614184192597087, "grad_norm": 0.5542563796043396, "learning_rate": 1.3388090103996754e-05, "loss": 1.9830609130859376, "step": 166250 }, { "epoch": 0.6616173420925687, "grad_norm": 0.5820098519325256, "learning_rate": 1.3386100877646667e-05, "loss": 1.9421791076660155, "step": 166300 }, { "epoch": 0.6618162649254288, "grad_norm": 0.6303835511207581, "learning_rate": 1.3384111651296579e-05, "loss": 1.948075714111328, "step": 166350 }, { "epoch": 0.6620151877582888, "grad_norm": 0.6172199845314026, "learning_rate": 1.3382122424946491e-05, "loss": 2.021645355224609, "step": 166400 }, { "epoch": 0.6622141105911489, "grad_norm": 0.7408104538917542, "learning_rate": 1.3380133198596403e-05, "loss": 1.9718833923339845, "step": 166450 }, { "epoch": 0.662413033424009, "grad_norm": 0.7401261329650879, "learning_rate": 1.3378143972246315e-05, "loss": 1.9859996032714844, "step": 166500 }, { "epoch": 0.6626119562568691, "grad_norm": 0.6164616942405701, "learning_rate": 1.3376154745896229e-05, "loss": 1.9751451110839844, "step": 166550 }, { "epoch": 0.6628108790897291, "grad_norm": 0.6468473672866821, "learning_rate": 1.3374165519546138e-05, "loss": 1.9941368103027344, "step": 166600 }, { "epoch": 0.6630098019225892, "grad_norm": 0.5252023935317993, "learning_rate": 1.3372176293196052e-05, "loss": 1.9608282470703124, "step": 166650 }, { "epoch": 0.6632087247554492, "grad_norm": 0.6459085941314697, "learning_rate": 1.3370187066845964e-05, "loss": 1.9590257263183595, "step": 166700 }, { "epoch": 0.6634076475883093, "grad_norm": 0.6578996777534485, "learning_rate": 1.3368197840495876e-05, "loss": 1.902496337890625, "step": 166750 }, { "epoch": 0.6636065704211693, "grad_norm": 0.5016270875930786, "learning_rate": 1.3366208614145788e-05, "loss": 1.9519546508789063, "step": 166800 }, { "epoch": 0.6638054932540294, "grad_norm": 0.6405324339866638, "learning_rate": 1.33642193877957e-05, "loss": 2.006944274902344, "step": 166850 }, { "epoch": 0.6640044160868895, "grad_norm": 0.4999639391899109, "learning_rate": 1.336223016144561e-05, "loss": 1.9144241333007812, "step": 166900 }, { "epoch": 0.6642033389197496, "grad_norm": 0.6348304152488708, "learning_rate": 1.3360240935095523e-05, "loss": 1.9842227172851563, "step": 166950 }, { "epoch": 0.6644022617526096, "grad_norm": 0.5686671733856201, "learning_rate": 1.3358251708745435e-05, "loss": 1.954173583984375, "step": 167000 }, { "epoch": 0.6646011845854697, "grad_norm": 0.5999440550804138, "learning_rate": 1.3356262482395347e-05, "loss": 2.0126605224609375, "step": 167050 }, { "epoch": 0.6648001074183297, "grad_norm": 0.6741089820861816, "learning_rate": 1.3354273256045261e-05, "loss": 1.9465940856933595, "step": 167100 }, { "epoch": 0.6649990302511898, "grad_norm": 0.6308878660202026, "learning_rate": 1.3352284029695173e-05, "loss": 1.9647979736328125, "step": 167150 }, { "epoch": 0.6651979530840498, "grad_norm": 0.6706940531730652, "learning_rate": 1.3350294803345085e-05, "loss": 1.9800492858886718, "step": 167200 }, { "epoch": 0.6653968759169099, "grad_norm": 0.6324967741966248, "learning_rate": 1.3348305576994996e-05, "loss": 1.970841827392578, "step": 167250 }, { "epoch": 0.66559579874977, "grad_norm": 0.582139253616333, "learning_rate": 1.3346316350644908e-05, "loss": 1.9556150817871094, "step": 167300 }, { "epoch": 0.6657947215826301, "grad_norm": 0.6220501661300659, "learning_rate": 1.334432712429482e-05, "loss": 1.9947537231445311, "step": 167350 }, { "epoch": 0.6659936444154901, "grad_norm": 0.5996485948562622, "learning_rate": 1.3342337897944732e-05, "loss": 1.937394256591797, "step": 167400 }, { "epoch": 0.6661925672483502, "grad_norm": 0.5948781967163086, "learning_rate": 1.3340348671594644e-05, "loss": 1.9691998291015624, "step": 167450 }, { "epoch": 0.6663914900812102, "grad_norm": 0.6756111979484558, "learning_rate": 1.3338359445244558e-05, "loss": 1.9638414001464843, "step": 167500 }, { "epoch": 0.6665904129140703, "grad_norm": 0.5485852956771851, "learning_rate": 1.3336370218894467e-05, "loss": 1.9534461975097657, "step": 167550 }, { "epoch": 0.6667893357469303, "grad_norm": 0.6303901076316833, "learning_rate": 1.333438099254438e-05, "loss": 1.9838272094726563, "step": 167600 }, { "epoch": 0.6669882585797904, "grad_norm": 0.5138285756111145, "learning_rate": 1.3332391766194293e-05, "loss": 1.934322052001953, "step": 167650 }, { "epoch": 0.6671871814126505, "grad_norm": 0.6528556942939758, "learning_rate": 1.3330402539844205e-05, "loss": 1.9625680541992188, "step": 167700 }, { "epoch": 0.6673861042455106, "grad_norm": 0.635908842086792, "learning_rate": 1.3328413313494117e-05, "loss": 2.0025213623046874, "step": 167750 }, { "epoch": 0.6675850270783706, "grad_norm": 0.6646952629089355, "learning_rate": 1.332642408714403e-05, "loss": 1.973856201171875, "step": 167800 }, { "epoch": 0.6677839499112307, "grad_norm": 0.5603941679000854, "learning_rate": 1.3324434860793941e-05, "loss": 1.9105946350097656, "step": 167850 }, { "epoch": 0.6679828727440907, "grad_norm": 0.6433072090148926, "learning_rate": 1.3322445634443852e-05, "loss": 1.9571037292480469, "step": 167900 }, { "epoch": 0.6681817955769508, "grad_norm": 0.7317153215408325, "learning_rate": 1.3320456408093764e-05, "loss": 1.9901412963867187, "step": 167950 }, { "epoch": 0.6683807184098108, "grad_norm": 0.647335946559906, "learning_rate": 1.3318467181743678e-05, "loss": 1.9683238220214845, "step": 168000 }, { "epoch": 0.6685796412426709, "grad_norm": 0.667595386505127, "learning_rate": 1.331647795539359e-05, "loss": 2.026106872558594, "step": 168050 }, { "epoch": 0.668778564075531, "grad_norm": 0.5840060710906982, "learning_rate": 1.3314488729043502e-05, "loss": 1.9954782104492188, "step": 168100 }, { "epoch": 0.6689774869083911, "grad_norm": 0.48702287673950195, "learning_rate": 1.3312499502693414e-05, "loss": 1.913466796875, "step": 168150 }, { "epoch": 0.6691764097412511, "grad_norm": 0.5878929495811462, "learning_rate": 1.3310510276343325e-05, "loss": 1.9474658203125, "step": 168200 }, { "epoch": 0.6693753325741112, "grad_norm": 0.7881160378456116, "learning_rate": 1.3308521049993237e-05, "loss": 1.934010009765625, "step": 168250 }, { "epoch": 0.6695742554069712, "grad_norm": 0.5897546410560608, "learning_rate": 1.3306531823643149e-05, "loss": 1.9460250854492187, "step": 168300 }, { "epoch": 0.6697731782398313, "grad_norm": 0.6475924253463745, "learning_rate": 1.3304542597293061e-05, "loss": 2.0014718627929686, "step": 168350 }, { "epoch": 0.6699721010726913, "grad_norm": 0.5906071662902832, "learning_rate": 1.3302553370942975e-05, "loss": 1.934016876220703, "step": 168400 }, { "epoch": 0.6701710239055514, "grad_norm": 0.6521267294883728, "learning_rate": 1.3300564144592887e-05, "loss": 1.9754995727539062, "step": 168450 }, { "epoch": 0.6703699467384115, "grad_norm": 0.6429029703140259, "learning_rate": 1.3298574918242798e-05, "loss": 1.9861624145507812, "step": 168500 }, { "epoch": 0.6705688695712716, "grad_norm": 0.5879282355308533, "learning_rate": 1.329658569189271e-05, "loss": 2.0254281616210936, "step": 168550 }, { "epoch": 0.6707677924041316, "grad_norm": 0.6514618396759033, "learning_rate": 1.3294596465542622e-05, "loss": 1.9853916931152344, "step": 168600 }, { "epoch": 0.6709667152369917, "grad_norm": 0.5834446549415588, "learning_rate": 1.3292607239192534e-05, "loss": 1.9620854187011718, "step": 168650 }, { "epoch": 0.6711656380698517, "grad_norm": 0.6326104402542114, "learning_rate": 1.3290618012842446e-05, "loss": 1.9624629211425781, "step": 168700 }, { "epoch": 0.6713645609027118, "grad_norm": 0.6200913190841675, "learning_rate": 1.3288628786492358e-05, "loss": 1.9830325317382813, "step": 168750 }, { "epoch": 0.6715634837355718, "grad_norm": 0.6333926320075989, "learning_rate": 1.3286679344669274e-05, "loss": 2.0141587829589844, "step": 168800 }, { "epoch": 0.6717624065684319, "grad_norm": 0.5599626302719116, "learning_rate": 1.3284690118319184e-05, "loss": 1.9256401062011719, "step": 168850 }, { "epoch": 0.671961329401292, "grad_norm": 0.45951104164123535, "learning_rate": 1.3282700891969096e-05, "loss": 1.9019442749023439, "step": 168900 }, { "epoch": 0.6721602522341521, "grad_norm": 0.5880116820335388, "learning_rate": 1.3280711665619008e-05, "loss": 1.9315281677246094, "step": 168950 }, { "epoch": 0.6723591750670121, "grad_norm": 0.6976844668388367, "learning_rate": 1.327872243926892e-05, "loss": 1.935208282470703, "step": 169000 }, { "epoch": 0.6725580978998722, "grad_norm": 0.5379548072814941, "learning_rate": 1.3276733212918833e-05, "loss": 1.9800228881835937, "step": 169050 }, { "epoch": 0.6727570207327322, "grad_norm": 0.6948954463005066, "learning_rate": 1.3274743986568745e-05, "loss": 1.9555194091796875, "step": 169100 }, { "epoch": 0.6729559435655923, "grad_norm": 0.5839225649833679, "learning_rate": 1.3272754760218659e-05, "loss": 1.9521482849121095, "step": 169150 }, { "epoch": 0.6731548663984523, "grad_norm": 0.6278376579284668, "learning_rate": 1.3270765533868567e-05, "loss": 1.9889303588867187, "step": 169200 }, { "epoch": 0.6733537892313124, "grad_norm": 0.68642258644104, "learning_rate": 1.3268776307518481e-05, "loss": 1.9656201171875, "step": 169250 }, { "epoch": 0.6735527120641726, "grad_norm": 0.6878868937492371, "learning_rate": 1.3266787081168393e-05, "loss": 1.9420561218261718, "step": 169300 }, { "epoch": 0.6737516348970326, "grad_norm": 0.6154994368553162, "learning_rate": 1.3264797854818305e-05, "loss": 1.9658859252929688, "step": 169350 }, { "epoch": 0.6739505577298927, "grad_norm": 0.6002416610717773, "learning_rate": 1.3262808628468218e-05, "loss": 1.9368766784667968, "step": 169400 }, { "epoch": 0.6741494805627527, "grad_norm": 0.703344464302063, "learning_rate": 1.326081940211813e-05, "loss": 1.94978271484375, "step": 169450 }, { "epoch": 0.6743484033956127, "grad_norm": 0.532525360584259, "learning_rate": 1.325883017576804e-05, "loss": 1.9842599487304688, "step": 169500 }, { "epoch": 0.6745473262284728, "grad_norm": 0.5728873014450073, "learning_rate": 1.3256840949417952e-05, "loss": 1.9331298828125, "step": 169550 }, { "epoch": 0.6747462490613328, "grad_norm": 0.5997127890586853, "learning_rate": 1.3254851723067865e-05, "loss": 1.9394111633300781, "step": 169600 }, { "epoch": 0.6749451718941929, "grad_norm": 0.5513131022453308, "learning_rate": 1.3252862496717778e-05, "loss": 2.0014878845214845, "step": 169650 }, { "epoch": 0.6751440947270531, "grad_norm": 0.6577293276786804, "learning_rate": 1.325087327036769e-05, "loss": 1.9719326782226563, "step": 169700 }, { "epoch": 0.6753430175599131, "grad_norm": 0.5756435990333557, "learning_rate": 1.3248884044017603e-05, "loss": 1.9586293029785156, "step": 169750 }, { "epoch": 0.6755419403927732, "grad_norm": 0.6380916833877563, "learning_rate": 1.3246894817667513e-05, "loss": 1.8623551940917968, "step": 169800 }, { "epoch": 0.6757408632256332, "grad_norm": 0.5521462559700012, "learning_rate": 1.3244905591317425e-05, "loss": 1.9947598266601563, "step": 169850 }, { "epoch": 0.6759397860584933, "grad_norm": 0.597274661064148, "learning_rate": 1.3242916364967337e-05, "loss": 1.9580844116210938, "step": 169900 }, { "epoch": 0.6761387088913533, "grad_norm": 0.6426103115081787, "learning_rate": 1.324092713861725e-05, "loss": 1.9529635620117187, "step": 169950 }, { "epoch": 0.6763376317242134, "grad_norm": 0.5529561042785645, "learning_rate": 1.3238937912267162e-05, "loss": 1.9452642822265624, "step": 170000 }, { "epoch": 0.6765365545570734, "grad_norm": 0.5679133534431458, "learning_rate": 1.3236948685917076e-05, "loss": 1.982586669921875, "step": 170050 }, { "epoch": 0.6767354773899336, "grad_norm": 0.47328463196754456, "learning_rate": 1.3234959459566988e-05, "loss": 1.9778123474121094, "step": 170100 }, { "epoch": 0.6769344002227936, "grad_norm": 0.6227615475654602, "learning_rate": 1.3232970233216898e-05, "loss": 1.93908203125, "step": 170150 }, { "epoch": 0.6771333230556537, "grad_norm": 0.5777117013931274, "learning_rate": 1.323098100686681e-05, "loss": 1.9951800537109374, "step": 170200 }, { "epoch": 0.6773322458885137, "grad_norm": 0.5344200730323792, "learning_rate": 1.3228991780516722e-05, "loss": 1.9361016845703125, "step": 170250 }, { "epoch": 0.6775311687213738, "grad_norm": 0.5521973967552185, "learning_rate": 1.3227002554166635e-05, "loss": 1.9414654541015626, "step": 170300 }, { "epoch": 0.6777300915542338, "grad_norm": 0.671845018863678, "learning_rate": 1.3225013327816547e-05, "loss": 1.9476910400390626, "step": 170350 }, { "epoch": 0.6779290143870939, "grad_norm": 0.5683905482292175, "learning_rate": 1.3223024101466459e-05, "loss": 1.9645591735839845, "step": 170400 }, { "epoch": 0.6781279372199539, "grad_norm": 0.6634063720703125, "learning_rate": 1.322103487511637e-05, "loss": 1.9352450561523438, "step": 170450 }, { "epoch": 0.6783268600528141, "grad_norm": 0.6189723014831543, "learning_rate": 1.3219045648766281e-05, "loss": 1.952721405029297, "step": 170500 }, { "epoch": 0.6785257828856741, "grad_norm": 0.6586669087409973, "learning_rate": 1.3217056422416195e-05, "loss": 2.0140541076660154, "step": 170550 }, { "epoch": 0.6787247057185342, "grad_norm": 0.6772307753562927, "learning_rate": 1.3215067196066107e-05, "loss": 1.9785269165039063, "step": 170600 }, { "epoch": 0.6789236285513942, "grad_norm": 0.7002596855163574, "learning_rate": 1.321307796971602e-05, "loss": 1.984658203125, "step": 170650 }, { "epoch": 0.6791225513842543, "grad_norm": 0.5686843991279602, "learning_rate": 1.3211088743365932e-05, "loss": 1.9551742553710938, "step": 170700 }, { "epoch": 0.6793214742171143, "grad_norm": 0.7079886794090271, "learning_rate": 1.3209099517015844e-05, "loss": 2.007868194580078, "step": 170750 }, { "epoch": 0.6795203970499744, "grad_norm": 0.6474146842956543, "learning_rate": 1.3207110290665754e-05, "loss": 1.911507110595703, "step": 170800 }, { "epoch": 0.6797193198828344, "grad_norm": 0.6474172472953796, "learning_rate": 1.3205121064315666e-05, "loss": 1.9172962951660155, "step": 170850 }, { "epoch": 0.6799182427156946, "grad_norm": 0.5055372714996338, "learning_rate": 1.3203131837965579e-05, "loss": 2.012734832763672, "step": 170900 }, { "epoch": 0.6801171655485546, "grad_norm": 0.7158119082450867, "learning_rate": 1.3201142611615492e-05, "loss": 1.985111083984375, "step": 170950 }, { "epoch": 0.6803160883814147, "grad_norm": 0.5406821370124817, "learning_rate": 1.3199153385265405e-05, "loss": 2.0253790283203124, "step": 171000 }, { "epoch": 0.6805150112142747, "grad_norm": 0.6356104016304016, "learning_rate": 1.3197164158915317e-05, "loss": 1.9518008422851563, "step": 171050 }, { "epoch": 0.6807139340471348, "grad_norm": 0.5063109993934631, "learning_rate": 1.3195174932565227e-05, "loss": 1.9643093872070312, "step": 171100 }, { "epoch": 0.6809128568799948, "grad_norm": 0.5727166533470154, "learning_rate": 1.319318570621514e-05, "loss": 1.9837641906738281, "step": 171150 }, { "epoch": 0.6811117797128549, "grad_norm": 0.6718882322311401, "learning_rate": 1.3191196479865052e-05, "loss": 1.9581068420410157, "step": 171200 }, { "epoch": 0.6813107025457149, "grad_norm": 0.610322892665863, "learning_rate": 1.3189207253514964e-05, "loss": 1.9165805053710938, "step": 171250 }, { "epoch": 0.6815096253785751, "grad_norm": 0.615227997303009, "learning_rate": 1.3187218027164876e-05, "loss": 2.0226478576660156, "step": 171300 }, { "epoch": 0.6817085482114351, "grad_norm": 0.6446302533149719, "learning_rate": 1.318522880081479e-05, "loss": 1.941097412109375, "step": 171350 }, { "epoch": 0.6819074710442952, "grad_norm": 0.7172838449478149, "learning_rate": 1.3183239574464702e-05, "loss": 2.0067097473144533, "step": 171400 }, { "epoch": 0.6821063938771552, "grad_norm": 0.6314775943756104, "learning_rate": 1.3181250348114612e-05, "loss": 2.002721710205078, "step": 171450 }, { "epoch": 0.6823053167100153, "grad_norm": 0.6212918162345886, "learning_rate": 1.3179261121764524e-05, "loss": 1.9435244750976564, "step": 171500 }, { "epoch": 0.6825042395428753, "grad_norm": 0.6317631006240845, "learning_rate": 1.3177271895414437e-05, "loss": 1.9263900756835937, "step": 171550 }, { "epoch": 0.6827031623757354, "grad_norm": 0.5076785087585449, "learning_rate": 1.3175282669064349e-05, "loss": 1.9616110229492187, "step": 171600 }, { "epoch": 0.6829020852085954, "grad_norm": 0.6071076393127441, "learning_rate": 1.317329344271426e-05, "loss": 1.9098115539550782, "step": 171650 }, { "epoch": 0.6831010080414556, "grad_norm": 0.7290217876434326, "learning_rate": 1.3171304216364173e-05, "loss": 1.9448211669921875, "step": 171700 }, { "epoch": 0.6832999308743156, "grad_norm": 0.6093226075172424, "learning_rate": 1.3169314990014083e-05, "loss": 1.9752468872070312, "step": 171750 }, { "epoch": 0.6834988537071757, "grad_norm": 0.7070204019546509, "learning_rate": 1.3167325763663996e-05, "loss": 1.975845184326172, "step": 171800 }, { "epoch": 0.6836977765400357, "grad_norm": 0.5597606301307678, "learning_rate": 1.316533653731391e-05, "loss": 1.94154541015625, "step": 171850 }, { "epoch": 0.6838966993728958, "grad_norm": 0.5869372487068176, "learning_rate": 1.3163347310963822e-05, "loss": 1.9508439636230468, "step": 171900 }, { "epoch": 0.6840956222057558, "grad_norm": 0.6118630170822144, "learning_rate": 1.3161358084613734e-05, "loss": 1.9858511352539063, "step": 171950 }, { "epoch": 0.6842945450386159, "grad_norm": 0.6398428678512573, "learning_rate": 1.3159368858263646e-05, "loss": 1.9955036926269532, "step": 172000 }, { "epoch": 0.6844934678714759, "grad_norm": 0.6506847143173218, "learning_rate": 1.3157379631913558e-05, "loss": 1.9019931030273438, "step": 172050 }, { "epoch": 0.6846923907043361, "grad_norm": 0.6495166420936584, "learning_rate": 1.3155390405563468e-05, "loss": 2.008004608154297, "step": 172100 }, { "epoch": 0.6848913135371961, "grad_norm": 0.6281445622444153, "learning_rate": 1.315340117921338e-05, "loss": 1.9365943908691405, "step": 172150 }, { "epoch": 0.6850902363700562, "grad_norm": 0.699629008769989, "learning_rate": 1.3151411952863293e-05, "loss": 2.0258665466308594, "step": 172200 }, { "epoch": 0.6852891592029162, "grad_norm": 0.7436072826385498, "learning_rate": 1.3149422726513205e-05, "loss": 1.9399179077148438, "step": 172250 }, { "epoch": 0.6854880820357763, "grad_norm": 0.6369670033454895, "learning_rate": 1.3147433500163119e-05, "loss": 1.9710055541992189, "step": 172300 }, { "epoch": 0.6856870048686363, "grad_norm": 0.47902360558509827, "learning_rate": 1.3145444273813031e-05, "loss": 1.9406222534179687, "step": 172350 }, { "epoch": 0.6858859277014964, "grad_norm": 0.6252622604370117, "learning_rate": 1.3143455047462941e-05, "loss": 1.9465228271484376, "step": 172400 }, { "epoch": 0.6860848505343564, "grad_norm": 0.5636504888534546, "learning_rate": 1.3141465821112853e-05, "loss": 1.9259956359863282, "step": 172450 }, { "epoch": 0.6862837733672166, "grad_norm": 0.5990538001060486, "learning_rate": 1.3139476594762766e-05, "loss": 1.926588134765625, "step": 172500 }, { "epoch": 0.6864826962000766, "grad_norm": 0.5589560866355896, "learning_rate": 1.3137487368412678e-05, "loss": 1.9942996215820312, "step": 172550 }, { "epoch": 0.6866816190329367, "grad_norm": 0.6102709174156189, "learning_rate": 1.313549814206259e-05, "loss": 2.0058912658691406, "step": 172600 }, { "epoch": 0.6868805418657967, "grad_norm": 0.6793992519378662, "learning_rate": 1.3133508915712502e-05, "loss": 1.9509471130371094, "step": 172650 }, { "epoch": 0.6870794646986568, "grad_norm": 0.5556241273880005, "learning_rate": 1.3131519689362416e-05, "loss": 1.9616058349609375, "step": 172700 }, { "epoch": 0.6872783875315168, "grad_norm": 0.6452838778495789, "learning_rate": 1.3129570247539328e-05, "loss": 1.9265034484863282, "step": 172750 }, { "epoch": 0.6874773103643769, "grad_norm": 0.6809362173080444, "learning_rate": 1.312758102118924e-05, "loss": 1.9550482177734374, "step": 172800 }, { "epoch": 0.6876762331972369, "grad_norm": 0.5218937993049622, "learning_rate": 1.3125591794839152e-05, "loss": 1.951771240234375, "step": 172850 }, { "epoch": 0.6878751560300971, "grad_norm": 0.5924151539802551, "learning_rate": 1.3123602568489064e-05, "loss": 1.9030979919433593, "step": 172900 }, { "epoch": 0.6880740788629571, "grad_norm": 0.5890881419181824, "learning_rate": 1.3121613342138976e-05, "loss": 1.9772962951660156, "step": 172950 }, { "epoch": 0.6882730016958172, "grad_norm": 0.5402178764343262, "learning_rate": 1.311962411578889e-05, "loss": 1.9527980041503907, "step": 173000 }, { "epoch": 0.6884719245286772, "grad_norm": 0.5939467549324036, "learning_rate": 1.3117634889438799e-05, "loss": 1.9721844482421875, "step": 173050 }, { "epoch": 0.6886708473615373, "grad_norm": 0.5823971629142761, "learning_rate": 1.3115645663088713e-05, "loss": 1.980036163330078, "step": 173100 }, { "epoch": 0.6888697701943973, "grad_norm": 0.6753098964691162, "learning_rate": 1.3113656436738625e-05, "loss": 1.9021963500976562, "step": 173150 }, { "epoch": 0.6890686930272574, "grad_norm": 0.631411075592041, "learning_rate": 1.3111667210388537e-05, "loss": 1.9873895263671875, "step": 173200 }, { "epoch": 0.6892676158601174, "grad_norm": 0.601502001285553, "learning_rate": 1.310967798403845e-05, "loss": 1.9600732421875, "step": 173250 }, { "epoch": 0.6894665386929776, "grad_norm": 0.6592881083488464, "learning_rate": 1.3107688757688361e-05, "loss": 2.0105348205566407, "step": 173300 }, { "epoch": 0.6896654615258376, "grad_norm": 0.7185329794883728, "learning_rate": 1.3105699531338273e-05, "loss": 1.9180657958984375, "step": 173350 }, { "epoch": 0.6898643843586977, "grad_norm": 0.48404160141944885, "learning_rate": 1.3103710304988184e-05, "loss": 2.0207473754882814, "step": 173400 }, { "epoch": 0.6900633071915577, "grad_norm": 0.6207731366157532, "learning_rate": 1.3101721078638096e-05, "loss": 2.007202911376953, "step": 173450 }, { "epoch": 0.6902622300244178, "grad_norm": 0.5695228576660156, "learning_rate": 1.3099731852288008e-05, "loss": 1.9591265869140626, "step": 173500 }, { "epoch": 0.6904611528572778, "grad_norm": 0.6631335020065308, "learning_rate": 1.3097742625937922e-05, "loss": 1.968793182373047, "step": 173550 }, { "epoch": 0.6906600756901379, "grad_norm": 0.5705615282058716, "learning_rate": 1.3095753399587834e-05, "loss": 2.009027099609375, "step": 173600 }, { "epoch": 0.6908589985229979, "grad_norm": 0.6442859172821045, "learning_rate": 1.3093764173237746e-05, "loss": 1.9389608764648438, "step": 173650 }, { "epoch": 0.691057921355858, "grad_norm": 0.5820040702819824, "learning_rate": 1.3091774946887657e-05, "loss": 1.9226080322265624, "step": 173700 }, { "epoch": 0.6912568441887181, "grad_norm": 0.6070265769958496, "learning_rate": 1.3089785720537569e-05, "loss": 1.9620118713378907, "step": 173750 }, { "epoch": 0.6914557670215782, "grad_norm": 0.7483996748924255, "learning_rate": 1.3087796494187481e-05, "loss": 1.9556085205078124, "step": 173800 }, { "epoch": 0.6916546898544382, "grad_norm": 0.5367985963821411, "learning_rate": 1.3085807267837393e-05, "loss": 2.00890869140625, "step": 173850 }, { "epoch": 0.6918536126872983, "grad_norm": 0.5299115777015686, "learning_rate": 1.3083818041487305e-05, "loss": 2.0068478393554687, "step": 173900 }, { "epoch": 0.6920525355201583, "grad_norm": 0.6125742793083191, "learning_rate": 1.308182881513722e-05, "loss": 1.9942910766601563, "step": 173950 }, { "epoch": 0.6922514583530184, "grad_norm": 0.6219415664672852, "learning_rate": 1.3079839588787131e-05, "loss": 1.9794418334960937, "step": 174000 }, { "epoch": 0.6924503811858784, "grad_norm": 0.5413085222244263, "learning_rate": 1.3077850362437042e-05, "loss": 1.911297607421875, "step": 174050 }, { "epoch": 0.6926493040187385, "grad_norm": 0.615614116191864, "learning_rate": 1.3075861136086954e-05, "loss": 1.9384190368652343, "step": 174100 }, { "epoch": 0.6928482268515986, "grad_norm": 0.6012570261955261, "learning_rate": 1.3073871909736866e-05, "loss": 1.9839698791503906, "step": 174150 }, { "epoch": 0.6930471496844587, "grad_norm": 0.485343873500824, "learning_rate": 1.3071882683386778e-05, "loss": 1.9112821960449218, "step": 174200 }, { "epoch": 0.6932460725173187, "grad_norm": 0.5278796553611755, "learning_rate": 1.306989345703669e-05, "loss": 1.9508197021484375, "step": 174250 }, { "epoch": 0.6934449953501788, "grad_norm": 0.5711222887039185, "learning_rate": 1.3067904230686603e-05, "loss": 1.9458718872070313, "step": 174300 }, { "epoch": 0.6936439181830388, "grad_norm": 0.6920262575149536, "learning_rate": 1.3065915004336513e-05, "loss": 1.9613363647460937, "step": 174350 }, { "epoch": 0.6938428410158989, "grad_norm": 0.5061562061309814, "learning_rate": 1.3063925777986425e-05, "loss": 1.9687796020507813, "step": 174400 }, { "epoch": 0.6940417638487589, "grad_norm": 0.5780789852142334, "learning_rate": 1.3061936551636339e-05, "loss": 1.9666909790039062, "step": 174450 }, { "epoch": 0.694240686681619, "grad_norm": 0.7168166041374207, "learning_rate": 1.3059947325286251e-05, "loss": 1.9527130126953125, "step": 174500 }, { "epoch": 0.6944396095144791, "grad_norm": 0.5422454476356506, "learning_rate": 1.3057958098936163e-05, "loss": 1.928916473388672, "step": 174550 }, { "epoch": 0.6946385323473392, "grad_norm": 0.6110217571258545, "learning_rate": 1.3055968872586075e-05, "loss": 2.0226515197753905, "step": 174600 }, { "epoch": 0.6948374551801992, "grad_norm": 0.5653586983680725, "learning_rate": 1.3053979646235988e-05, "loss": 1.960574951171875, "step": 174650 }, { "epoch": 0.6950363780130593, "grad_norm": 0.6065129041671753, "learning_rate": 1.3051990419885898e-05, "loss": 1.9584286499023438, "step": 174700 }, { "epoch": 0.6952353008459193, "grad_norm": 0.4950732886791229, "learning_rate": 1.305000119353581e-05, "loss": 1.9528445434570312, "step": 174750 }, { "epoch": 0.6954342236787794, "grad_norm": 0.6666948795318604, "learning_rate": 1.3048011967185722e-05, "loss": 1.9360560607910156, "step": 174800 }, { "epoch": 0.6956331465116394, "grad_norm": 0.609679102897644, "learning_rate": 1.3046022740835636e-05, "loss": 1.975375213623047, "step": 174850 }, { "epoch": 0.6958320693444995, "grad_norm": 0.6347205638885498, "learning_rate": 1.3044033514485548e-05, "loss": 1.9307177734375, "step": 174900 }, { "epoch": 0.6960309921773596, "grad_norm": 0.5589809417724609, "learning_rate": 1.304204428813546e-05, "loss": 2.001629638671875, "step": 174950 }, { "epoch": 0.6962299150102197, "grad_norm": 0.6916815042495728, "learning_rate": 1.3040055061785371e-05, "loss": 1.983373565673828, "step": 175000 }, { "epoch": 0.6964288378430797, "grad_norm": 0.6332627534866333, "learning_rate": 1.3038065835435283e-05, "loss": 1.9510244750976562, "step": 175050 }, { "epoch": 0.6966277606759398, "grad_norm": 0.6312013268470764, "learning_rate": 1.3036076609085195e-05, "loss": 1.957006072998047, "step": 175100 }, { "epoch": 0.6968266835087998, "grad_norm": 0.5890754461288452, "learning_rate": 1.3034087382735107e-05, "loss": 1.955267333984375, "step": 175150 }, { "epoch": 0.6970256063416599, "grad_norm": 0.6416407823562622, "learning_rate": 1.303209815638502e-05, "loss": 1.9467471313476563, "step": 175200 }, { "epoch": 0.6972245291745199, "grad_norm": 0.6036764979362488, "learning_rate": 1.3030108930034933e-05, "loss": 1.9495086669921875, "step": 175250 }, { "epoch": 0.69742345200738, "grad_norm": 0.545613706111908, "learning_rate": 1.3028159488211847e-05, "loss": 1.9642425537109376, "step": 175300 }, { "epoch": 0.6976223748402401, "grad_norm": 0.4939366579055786, "learning_rate": 1.3026170261861757e-05, "loss": 1.9371044921875, "step": 175350 }, { "epoch": 0.6978212976731002, "grad_norm": 0.5245595574378967, "learning_rate": 1.302418103551167e-05, "loss": 1.9649615478515625, "step": 175400 }, { "epoch": 0.6980202205059602, "grad_norm": 0.6181942820549011, "learning_rate": 1.3022191809161582e-05, "loss": 1.9650163269042968, "step": 175450 }, { "epoch": 0.6982191433388203, "grad_norm": 0.49875643849372864, "learning_rate": 1.3020202582811494e-05, "loss": 1.9094888305664062, "step": 175500 }, { "epoch": 0.6984180661716803, "grad_norm": 0.6102614402770996, "learning_rate": 1.3018213356461406e-05, "loss": 1.9939082336425782, "step": 175550 }, { "epoch": 0.6986169890045404, "grad_norm": 0.541060745716095, "learning_rate": 1.301622413011132e-05, "loss": 2.0051866149902344, "step": 175600 }, { "epoch": 0.6988159118374004, "grad_norm": 0.5278360247612, "learning_rate": 1.3014234903761228e-05, "loss": 1.9639212036132812, "step": 175650 }, { "epoch": 0.6990148346702605, "grad_norm": 0.6158648133277893, "learning_rate": 1.3012245677411142e-05, "loss": 1.965721893310547, "step": 175700 }, { "epoch": 0.6992137575031206, "grad_norm": 0.6036497354507446, "learning_rate": 1.3010256451061054e-05, "loss": 1.9173899841308595, "step": 175750 }, { "epoch": 0.6994126803359807, "grad_norm": 0.593086302280426, "learning_rate": 1.3008267224710967e-05, "loss": 1.995494384765625, "step": 175800 }, { "epoch": 0.6996116031688407, "grad_norm": 0.6318752765655518, "learning_rate": 1.3006277998360879e-05, "loss": 1.8934617614746094, "step": 175850 }, { "epoch": 0.6998105260017008, "grad_norm": 0.41159316897392273, "learning_rate": 1.3004288772010791e-05, "loss": 1.9622323608398438, "step": 175900 }, { "epoch": 0.7000094488345608, "grad_norm": 0.5506535768508911, "learning_rate": 1.3002299545660703e-05, "loss": 1.9925013732910157, "step": 175950 }, { "epoch": 0.7002083716674209, "grad_norm": 0.5990058183670044, "learning_rate": 1.3000310319310613e-05, "loss": 2.0235835266113282, "step": 176000 }, { "epoch": 0.7004072945002809, "grad_norm": 0.6206620931625366, "learning_rate": 1.2998321092960526e-05, "loss": 1.976660919189453, "step": 176050 }, { "epoch": 0.700606217333141, "grad_norm": 0.5789020657539368, "learning_rate": 1.299633186661044e-05, "loss": 2.0062278747558593, "step": 176100 }, { "epoch": 0.7008051401660012, "grad_norm": 0.5616990327835083, "learning_rate": 1.2994342640260352e-05, "loss": 1.9031228637695312, "step": 176150 }, { "epoch": 0.7010040629988612, "grad_norm": 0.5263570547103882, "learning_rate": 1.2992353413910264e-05, "loss": 1.9886265563964844, "step": 176200 }, { "epoch": 0.7012029858317212, "grad_norm": 0.5944104194641113, "learning_rate": 1.2990364187560176e-05, "loss": 1.9516056823730468, "step": 176250 }, { "epoch": 0.7014019086645813, "grad_norm": 0.581814169883728, "learning_rate": 1.2988374961210086e-05, "loss": 1.8842434692382812, "step": 176300 }, { "epoch": 0.7016008314974413, "grad_norm": 0.6059974431991577, "learning_rate": 1.2986385734859998e-05, "loss": 1.937283935546875, "step": 176350 }, { "epoch": 0.7017997543303014, "grad_norm": 0.5669457316398621, "learning_rate": 1.298439650850991e-05, "loss": 1.977520751953125, "step": 176400 }, { "epoch": 0.7019986771631614, "grad_norm": 0.6150415539741516, "learning_rate": 1.2982407282159823e-05, "loss": 2.04492431640625, "step": 176450 }, { "epoch": 0.7021975999960215, "grad_norm": 0.6309443116188049, "learning_rate": 1.2980418055809737e-05, "loss": 1.93208740234375, "step": 176500 }, { "epoch": 0.7023965228288817, "grad_norm": 0.4912402629852295, "learning_rate": 1.2978428829459649e-05, "loss": 1.9421348571777344, "step": 176550 }, { "epoch": 0.7025954456617417, "grad_norm": 0.5995904207229614, "learning_rate": 1.2976439603109561e-05, "loss": 1.9639002990722656, "step": 176600 }, { "epoch": 0.7027943684946018, "grad_norm": 0.5556608438491821, "learning_rate": 1.2974450376759471e-05, "loss": 1.9453939819335937, "step": 176650 }, { "epoch": 0.7029932913274618, "grad_norm": 0.6219083070755005, "learning_rate": 1.2972461150409384e-05, "loss": 1.926935577392578, "step": 176700 }, { "epoch": 0.7031922141603218, "grad_norm": 0.6007654070854187, "learning_rate": 1.2970471924059296e-05, "loss": 1.9508381652832032, "step": 176750 }, { "epoch": 0.7033911369931819, "grad_norm": 0.6512378454208374, "learning_rate": 1.2968482697709208e-05, "loss": 1.982752685546875, "step": 176800 }, { "epoch": 0.703590059826042, "grad_norm": 0.5483191013336182, "learning_rate": 1.296649347135912e-05, "loss": 1.9484310913085938, "step": 176850 }, { "epoch": 0.703788982658902, "grad_norm": 0.5786319375038147, "learning_rate": 1.2964504245009034e-05, "loss": 1.9401776123046874, "step": 176900 }, { "epoch": 0.7039879054917622, "grad_norm": 0.5390604734420776, "learning_rate": 1.2962515018658943e-05, "loss": 1.9178501892089843, "step": 176950 }, { "epoch": 0.7041868283246222, "grad_norm": 0.5585570335388184, "learning_rate": 1.2960525792308856e-05, "loss": 1.9226055908203126, "step": 177000 }, { "epoch": 0.7043857511574823, "grad_norm": 0.6650248169898987, "learning_rate": 1.2958536565958769e-05, "loss": 1.9060452270507813, "step": 177050 }, { "epoch": 0.7045846739903423, "grad_norm": 0.579165518283844, "learning_rate": 1.295654733960868e-05, "loss": 1.9132913208007813, "step": 177100 }, { "epoch": 0.7047835968232024, "grad_norm": 0.5664594769477844, "learning_rate": 1.2954558113258593e-05, "loss": 1.950393524169922, "step": 177150 }, { "epoch": 0.7049825196560624, "grad_norm": 0.7112635374069214, "learning_rate": 1.2952568886908505e-05, "loss": 1.973607635498047, "step": 177200 }, { "epoch": 0.7051814424889224, "grad_norm": 0.6210468411445618, "learning_rate": 1.2950579660558417e-05, "loss": 1.932845916748047, "step": 177250 }, { "epoch": 0.7053803653217825, "grad_norm": 0.5967161059379578, "learning_rate": 1.2948590434208328e-05, "loss": 1.9253025817871094, "step": 177300 }, { "epoch": 0.7055792881546427, "grad_norm": 0.6139297485351562, "learning_rate": 1.294660120785824e-05, "loss": 1.9639512634277343, "step": 177350 }, { "epoch": 0.7057782109875027, "grad_norm": 0.662327229976654, "learning_rate": 1.2944651766035155e-05, "loss": 1.9856129455566407, "step": 177400 }, { "epoch": 0.7059771338203628, "grad_norm": 0.6932691335678101, "learning_rate": 1.2942702324212068e-05, "loss": 1.9544270324707032, "step": 177450 }, { "epoch": 0.7061760566532228, "grad_norm": 0.5916690230369568, "learning_rate": 1.294071309786198e-05, "loss": 1.9669876098632812, "step": 177500 }, { "epoch": 0.7063749794860829, "grad_norm": 0.6373116970062256, "learning_rate": 1.2938723871511893e-05, "loss": 1.9922747802734375, "step": 177550 }, { "epoch": 0.7065739023189429, "grad_norm": 0.6146555542945862, "learning_rate": 1.2936734645161803e-05, "loss": 1.860633087158203, "step": 177600 }, { "epoch": 0.706772825151803, "grad_norm": 0.6952052712440491, "learning_rate": 1.2934745418811715e-05, "loss": 1.979315948486328, "step": 177650 }, { "epoch": 0.706971747984663, "grad_norm": 0.7104290127754211, "learning_rate": 1.2932756192461629e-05, "loss": 1.9074905395507813, "step": 177700 }, { "epoch": 0.7071706708175232, "grad_norm": 0.6554495692253113, "learning_rate": 1.2930766966111541e-05, "loss": 1.9678062438964843, "step": 177750 }, { "epoch": 0.7073695936503832, "grad_norm": 0.52723628282547, "learning_rate": 1.2928777739761453e-05, "loss": 1.9632447814941407, "step": 177800 }, { "epoch": 0.7075685164832433, "grad_norm": 0.675139844417572, "learning_rate": 1.2926788513411366e-05, "loss": 1.9647117614746095, "step": 177850 }, { "epoch": 0.7077674393161033, "grad_norm": 0.2942258417606354, "learning_rate": 1.2924799287061278e-05, "loss": 1.9615541076660157, "step": 177900 }, { "epoch": 0.7079663621489634, "grad_norm": 0.6063248515129089, "learning_rate": 1.2922810060711188e-05, "loss": 1.9777281188964844, "step": 177950 }, { "epoch": 0.7081652849818234, "grad_norm": 0.6273890137672424, "learning_rate": 1.29208208343611e-05, "loss": 1.9432994079589845, "step": 178000 }, { "epoch": 0.7083642078146835, "grad_norm": 0.5816575884819031, "learning_rate": 1.2918831608011012e-05, "loss": 1.8953849792480468, "step": 178050 }, { "epoch": 0.7085631306475435, "grad_norm": 0.5347800254821777, "learning_rate": 1.2916842381660926e-05, "loss": 1.9704765319824218, "step": 178100 }, { "epoch": 0.7087620534804037, "grad_norm": 0.5995627045631409, "learning_rate": 1.2914853155310838e-05, "loss": 1.976047821044922, "step": 178150 }, { "epoch": 0.7089609763132637, "grad_norm": 0.6148393750190735, "learning_rate": 1.291286392896075e-05, "loss": 1.9436367797851561, "step": 178200 }, { "epoch": 0.7091598991461238, "grad_norm": 0.5615351796150208, "learning_rate": 1.2910874702610661e-05, "loss": 1.9234689331054688, "step": 178250 }, { "epoch": 0.7093588219789838, "grad_norm": 0.5995496511459351, "learning_rate": 1.2908885476260573e-05, "loss": 1.9901979064941406, "step": 178300 }, { "epoch": 0.7095577448118439, "grad_norm": 0.5542240738868713, "learning_rate": 1.2906896249910485e-05, "loss": 1.936244659423828, "step": 178350 }, { "epoch": 0.7097566676447039, "grad_norm": 0.5518093705177307, "learning_rate": 1.2904907023560397e-05, "loss": 1.9265280151367188, "step": 178400 }, { "epoch": 0.709955590477564, "grad_norm": 0.6045891046524048, "learning_rate": 1.290291779721031e-05, "loss": 1.9317984008789062, "step": 178450 }, { "epoch": 0.710154513310424, "grad_norm": 0.64598548412323, "learning_rate": 1.2900928570860223e-05, "loss": 1.9719921875, "step": 178500 }, { "epoch": 0.7103534361432842, "grad_norm": 0.6398409008979797, "learning_rate": 1.2898939344510136e-05, "loss": 1.9590998840332032, "step": 178550 }, { "epoch": 0.7105523589761442, "grad_norm": 0.5914210081100464, "learning_rate": 1.2896950118160046e-05, "loss": 1.915098876953125, "step": 178600 }, { "epoch": 0.7107512818090043, "grad_norm": 0.6996634602546692, "learning_rate": 1.2894960891809958e-05, "loss": 1.9534796142578126, "step": 178650 }, { "epoch": 0.7109502046418643, "grad_norm": 0.6105952858924866, "learning_rate": 1.289297166545987e-05, "loss": 1.947200164794922, "step": 178700 }, { "epoch": 0.7111491274747244, "grad_norm": 0.5975544452667236, "learning_rate": 1.2890982439109783e-05, "loss": 1.9365597534179688, "step": 178750 }, { "epoch": 0.7113480503075844, "grad_norm": 0.5102234482765198, "learning_rate": 1.2888993212759695e-05, "loss": 1.9606388854980468, "step": 178800 }, { "epoch": 0.7115469731404445, "grad_norm": 0.5492573976516724, "learning_rate": 1.2887003986409607e-05, "loss": 1.9768655395507813, "step": 178850 }, { "epoch": 0.7117458959733045, "grad_norm": 0.6018583178520203, "learning_rate": 1.2885014760059517e-05, "loss": 1.9289340209960937, "step": 178900 }, { "epoch": 0.7119448188061647, "grad_norm": 0.7304274439811707, "learning_rate": 1.288302553370943e-05, "loss": 1.9652590942382813, "step": 178950 }, { "epoch": 0.7121437416390247, "grad_norm": 0.4647403359413147, "learning_rate": 1.2881036307359343e-05, "loss": 1.9668601989746093, "step": 179000 }, { "epoch": 0.7123426644718848, "grad_norm": 0.6534489393234253, "learning_rate": 1.2879047081009255e-05, "loss": 1.9272235107421876, "step": 179050 }, { "epoch": 0.7125415873047448, "grad_norm": 0.5593897700309753, "learning_rate": 1.2877057854659168e-05, "loss": 2.007495880126953, "step": 179100 }, { "epoch": 0.7127405101376049, "grad_norm": 0.5869185328483582, "learning_rate": 1.287506862830908e-05, "loss": 1.9515286254882813, "step": 179150 }, { "epoch": 0.7129394329704649, "grad_norm": 0.6087112426757812, "learning_rate": 1.2873079401958992e-05, "loss": 1.9795355224609374, "step": 179200 }, { "epoch": 0.713138355803325, "grad_norm": 0.6006350517272949, "learning_rate": 1.2871090175608902e-05, "loss": 1.9654049682617187, "step": 179250 }, { "epoch": 0.713337278636185, "grad_norm": 0.5007810592651367, "learning_rate": 1.2869100949258814e-05, "loss": 1.978731689453125, "step": 179300 }, { "epoch": 0.7135362014690452, "grad_norm": 0.5892319083213806, "learning_rate": 1.2867111722908727e-05, "loss": 2.016007080078125, "step": 179350 }, { "epoch": 0.7137351243019052, "grad_norm": 0.5789260864257812, "learning_rate": 1.286512249655864e-05, "loss": 1.980207977294922, "step": 179400 }, { "epoch": 0.7139340471347653, "grad_norm": 0.540278434753418, "learning_rate": 1.2863133270208553e-05, "loss": 1.9135430908203126, "step": 179450 }, { "epoch": 0.7141329699676253, "grad_norm": 0.6849889755249023, "learning_rate": 1.2861144043858465e-05, "loss": 1.987848663330078, "step": 179500 }, { "epoch": 0.7143318928004854, "grad_norm": 0.4246155023574829, "learning_rate": 1.2859154817508375e-05, "loss": 2.000691680908203, "step": 179550 }, { "epoch": 0.7145308156333454, "grad_norm": 0.5955738425254822, "learning_rate": 1.2857165591158287e-05, "loss": 1.9794935607910156, "step": 179600 }, { "epoch": 0.7147297384662055, "grad_norm": 0.44512802362442017, "learning_rate": 1.28551763648082e-05, "loss": 1.9504466247558594, "step": 179650 }, { "epoch": 0.7149286612990655, "grad_norm": 0.5868266224861145, "learning_rate": 1.2853187138458112e-05, "loss": 1.9041847229003905, "step": 179700 }, { "epoch": 0.7151275841319257, "grad_norm": 0.603209376335144, "learning_rate": 1.2851197912108024e-05, "loss": 1.9411337280273437, "step": 179750 }, { "epoch": 0.7153265069647857, "grad_norm": 0.4524536728858948, "learning_rate": 1.2849208685757938e-05, "loss": 2.011308135986328, "step": 179800 }, { "epoch": 0.7155254297976458, "grad_norm": 0.7392184734344482, "learning_rate": 1.284721945940785e-05, "loss": 1.9651771545410157, "step": 179850 }, { "epoch": 0.7157243526305058, "grad_norm": 0.5863561630249023, "learning_rate": 1.284523023305776e-05, "loss": 2.0075714111328127, "step": 179900 }, { "epoch": 0.7159232754633659, "grad_norm": 0.48099708557128906, "learning_rate": 1.2843280791234674e-05, "loss": 1.9867489624023438, "step": 179950 }, { "epoch": 0.7161221982962259, "grad_norm": 0.6536116003990173, "learning_rate": 1.2841291564884586e-05, "loss": 1.9620230102539062, "step": 180000 }, { "epoch": 0.716321121129086, "grad_norm": 0.5142342448234558, "learning_rate": 1.2839302338534498e-05, "loss": 1.9437501525878906, "step": 180050 }, { "epoch": 0.716520043961946, "grad_norm": 0.5891138315200806, "learning_rate": 1.283731311218441e-05, "loss": 1.9682029724121093, "step": 180100 }, { "epoch": 0.7167189667948062, "grad_norm": 0.6934857368469238, "learning_rate": 1.2835323885834324e-05, "loss": 1.9233088684082031, "step": 180150 }, { "epoch": 0.7169178896276662, "grad_norm": 0.5172995924949646, "learning_rate": 1.2833334659484233e-05, "loss": 1.9559222412109376, "step": 180200 }, { "epoch": 0.7171168124605263, "grad_norm": 0.604745090007782, "learning_rate": 1.2831345433134147e-05, "loss": 1.954109649658203, "step": 180250 }, { "epoch": 0.7173157352933863, "grad_norm": 0.5708123445510864, "learning_rate": 1.2829356206784059e-05, "loss": 1.9664471435546875, "step": 180300 }, { "epoch": 0.7175146581262464, "grad_norm": 0.5049630403518677, "learning_rate": 1.282736698043397e-05, "loss": 1.9549971008300782, "step": 180350 }, { "epoch": 0.7177135809591064, "grad_norm": 0.6287798881530762, "learning_rate": 1.2825377754083883e-05, "loss": 1.9684266662597656, "step": 180400 }, { "epoch": 0.7179125037919665, "grad_norm": 0.6759544610977173, "learning_rate": 1.2823388527733795e-05, "loss": 1.920924072265625, "step": 180450 }, { "epoch": 0.7181114266248265, "grad_norm": 0.7317234873771667, "learning_rate": 1.2821399301383707e-05, "loss": 1.9782460021972657, "step": 180500 }, { "epoch": 0.7183103494576867, "grad_norm": 0.5312860012054443, "learning_rate": 1.2819410075033618e-05, "loss": 1.9363829040527343, "step": 180550 }, { "epoch": 0.7185092722905467, "grad_norm": 0.5979598760604858, "learning_rate": 1.281742084868353e-05, "loss": 1.9906999206542968, "step": 180600 }, { "epoch": 0.7187081951234068, "grad_norm": 0.4774273633956909, "learning_rate": 1.2815431622333444e-05, "loss": 1.9509365844726563, "step": 180650 }, { "epoch": 0.7189071179562668, "grad_norm": 0.5506439208984375, "learning_rate": 1.2813442395983356e-05, "loss": 1.961842041015625, "step": 180700 }, { "epoch": 0.7191060407891269, "grad_norm": 0.5739799737930298, "learning_rate": 1.2811453169633268e-05, "loss": 1.9887408447265624, "step": 180750 }, { "epoch": 0.7193049636219869, "grad_norm": 0.5216155052185059, "learning_rate": 1.280946394328318e-05, "loss": 1.9508009338378907, "step": 180800 }, { "epoch": 0.719503886454847, "grad_norm": 0.4738161861896515, "learning_rate": 1.280747471693309e-05, "loss": 1.9671847534179687, "step": 180850 }, { "epoch": 0.719702809287707, "grad_norm": 0.6890722513198853, "learning_rate": 1.2805485490583003e-05, "loss": 1.936365966796875, "step": 180900 }, { "epoch": 0.7199017321205672, "grad_norm": 0.6558961868286133, "learning_rate": 1.2803496264232915e-05, "loss": 1.9122274780273438, "step": 180950 }, { "epoch": 0.7201006549534272, "grad_norm": 0.4952366054058075, "learning_rate": 1.2801507037882827e-05, "loss": 1.9483741760253905, "step": 181000 }, { "epoch": 0.7202995777862873, "grad_norm": 0.6481090188026428, "learning_rate": 1.2799517811532741e-05, "loss": 1.952149658203125, "step": 181050 }, { "epoch": 0.7204985006191473, "grad_norm": 0.640622079372406, "learning_rate": 1.2797528585182653e-05, "loss": 1.9538951110839844, "step": 181100 }, { "epoch": 0.7206974234520074, "grad_norm": 0.533951997756958, "learning_rate": 1.2795539358832565e-05, "loss": 1.8860455322265626, "step": 181150 }, { "epoch": 0.7208963462848674, "grad_norm": 0.5948426127433777, "learning_rate": 1.2793550132482476e-05, "loss": 1.9417552185058593, "step": 181200 }, { "epoch": 0.7210952691177275, "grad_norm": 0.4349178373813629, "learning_rate": 1.2791560906132388e-05, "loss": 1.99625, "step": 181250 }, { "epoch": 0.7212941919505875, "grad_norm": 0.5737036466598511, "learning_rate": 1.27895716797823e-05, "loss": 1.9532485961914063, "step": 181300 }, { "epoch": 0.7214931147834477, "grad_norm": 0.5889508128166199, "learning_rate": 1.2787582453432212e-05, "loss": 1.9547419738769531, "step": 181350 }, { "epoch": 0.7216920376163077, "grad_norm": 0.6090871095657349, "learning_rate": 1.2785593227082124e-05, "loss": 1.9321261596679689, "step": 181400 }, { "epoch": 0.7218909604491678, "grad_norm": 0.5091133117675781, "learning_rate": 1.2783604000732038e-05, "loss": 1.9728407287597656, "step": 181450 }, { "epoch": 0.7220898832820278, "grad_norm": 0.6135525703430176, "learning_rate": 1.2781614774381947e-05, "loss": 1.947920379638672, "step": 181500 }, { "epoch": 0.7222888061148879, "grad_norm": 0.5361465811729431, "learning_rate": 1.277962554803186e-05, "loss": 1.9661178588867188, "step": 181550 }, { "epoch": 0.7224877289477479, "grad_norm": 0.6853674054145813, "learning_rate": 1.2777636321681773e-05, "loss": 1.9473782348632813, "step": 181600 }, { "epoch": 0.722686651780608, "grad_norm": 0.5673430562019348, "learning_rate": 1.2775647095331685e-05, "loss": 1.9510353088378907, "step": 181650 }, { "epoch": 0.722885574613468, "grad_norm": 0.6609207987785339, "learning_rate": 1.2773657868981597e-05, "loss": 1.943274688720703, "step": 181700 }, { "epoch": 0.7230844974463281, "grad_norm": 0.6395733952522278, "learning_rate": 1.277166864263151e-05, "loss": 2.039620208740234, "step": 181750 }, { "epoch": 0.7232834202791882, "grad_norm": 0.7679088115692139, "learning_rate": 1.2769679416281421e-05, "loss": 2.000470733642578, "step": 181800 }, { "epoch": 0.7234823431120483, "grad_norm": 0.5262123942375183, "learning_rate": 1.2767690189931332e-05, "loss": 1.9216314697265624, "step": 181850 }, { "epoch": 0.7236812659449083, "grad_norm": 0.5631518959999084, "learning_rate": 1.2765700963581244e-05, "loss": 1.8880859375, "step": 181900 }, { "epoch": 0.7238801887777684, "grad_norm": 0.6441415548324585, "learning_rate": 1.2763711737231156e-05, "loss": 1.9261572265625, "step": 181950 }, { "epoch": 0.7240791116106284, "grad_norm": 0.46193066239356995, "learning_rate": 1.276172251088107e-05, "loss": 1.9431266784667969, "step": 182000 }, { "epoch": 0.7242780344434885, "grad_norm": 0.62607342004776, "learning_rate": 1.2759733284530982e-05, "loss": 1.946291961669922, "step": 182050 }, { "epoch": 0.7244769572763485, "grad_norm": 0.6097332239151001, "learning_rate": 1.2757744058180894e-05, "loss": 1.9319865417480468, "step": 182100 }, { "epoch": 0.7246758801092086, "grad_norm": 0.6353846192359924, "learning_rate": 1.2755754831830805e-05, "loss": 1.955673370361328, "step": 182150 }, { "epoch": 0.7248748029420687, "grad_norm": 0.6413089632987976, "learning_rate": 1.2753765605480717e-05, "loss": 1.9285890197753905, "step": 182200 }, { "epoch": 0.7250737257749288, "grad_norm": 0.6873670220375061, "learning_rate": 1.2751776379130629e-05, "loss": 1.9463275146484376, "step": 182250 }, { "epoch": 0.7252726486077888, "grad_norm": 0.5511295199394226, "learning_rate": 1.2749787152780541e-05, "loss": 1.943597869873047, "step": 182300 }, { "epoch": 0.7254715714406489, "grad_norm": 0.605109691619873, "learning_rate": 1.2747797926430453e-05, "loss": 1.9158323669433595, "step": 182350 }, { "epoch": 0.7256704942735089, "grad_norm": 0.4265860617160797, "learning_rate": 1.2745808700080367e-05, "loss": 1.9499404907226563, "step": 182400 }, { "epoch": 0.725869417106369, "grad_norm": 0.6422107815742493, "learning_rate": 1.274381947373028e-05, "loss": 1.946690673828125, "step": 182450 }, { "epoch": 0.726068339939229, "grad_norm": 0.6255564093589783, "learning_rate": 1.274183024738019e-05, "loss": 1.9459724426269531, "step": 182500 }, { "epoch": 0.7262672627720891, "grad_norm": 0.4674580693244934, "learning_rate": 1.2739841021030102e-05, "loss": 1.9547193908691407, "step": 182550 }, { "epoch": 0.7264661856049492, "grad_norm": 0.5686088800430298, "learning_rate": 1.2737851794680014e-05, "loss": 1.9462442016601562, "step": 182600 }, { "epoch": 0.7266651084378093, "grad_norm": 0.5645065307617188, "learning_rate": 1.2735862568329926e-05, "loss": 1.891513671875, "step": 182650 }, { "epoch": 0.7268640312706693, "grad_norm": 0.5535271167755127, "learning_rate": 1.2733873341979838e-05, "loss": 1.9974488830566406, "step": 182700 }, { "epoch": 0.7270629541035294, "grad_norm": 0.6107583045959473, "learning_rate": 1.273188411562975e-05, "loss": 1.993372039794922, "step": 182750 }, { "epoch": 0.7272618769363894, "grad_norm": 0.6780672669410706, "learning_rate": 1.2729894889279661e-05, "loss": 2.0053253173828125, "step": 182800 }, { "epoch": 0.7274607997692495, "grad_norm": 0.4288371503353119, "learning_rate": 1.2727905662929573e-05, "loss": 1.8616935729980468, "step": 182850 }, { "epoch": 0.7276597226021095, "grad_norm": 0.5747227072715759, "learning_rate": 1.2725916436579487e-05, "loss": 1.9778955078125, "step": 182900 }, { "epoch": 0.7278586454349696, "grad_norm": 0.5655699968338013, "learning_rate": 1.2723927210229399e-05, "loss": 1.8855050659179688, "step": 182950 }, { "epoch": 0.7280575682678297, "grad_norm": 0.5896579623222351, "learning_rate": 1.2721937983879311e-05, "loss": 1.971091766357422, "step": 183000 }, { "epoch": 0.7282564911006898, "grad_norm": 0.4959811270236969, "learning_rate": 1.2719948757529223e-05, "loss": 1.9846554565429688, "step": 183050 }, { "epoch": 0.7284554139335498, "grad_norm": 0.543452799320221, "learning_rate": 1.2717959531179135e-05, "loss": 1.9789924621582031, "step": 183100 }, { "epoch": 0.7286543367664099, "grad_norm": 0.6370832920074463, "learning_rate": 1.2715970304829046e-05, "loss": 1.936519775390625, "step": 183150 }, { "epoch": 0.7288532595992699, "grad_norm": 0.6420296430587769, "learning_rate": 1.2713981078478958e-05, "loss": 1.9288790893554688, "step": 183200 }, { "epoch": 0.72905218243213, "grad_norm": 0.6251072287559509, "learning_rate": 1.271199185212887e-05, "loss": 1.9537254333496095, "step": 183250 }, { "epoch": 0.72925110526499, "grad_norm": 0.558193564414978, "learning_rate": 1.2710002625778784e-05, "loss": 1.9518157958984375, "step": 183300 }, { "epoch": 0.7294500280978501, "grad_norm": 0.645496666431427, "learning_rate": 1.2708013399428696e-05, "loss": 1.9645883178710937, "step": 183350 }, { "epoch": 0.7296489509307102, "grad_norm": 0.6008288860321045, "learning_rate": 1.2706024173078608e-05, "loss": 1.9765762329101562, "step": 183400 }, { "epoch": 0.7298478737635703, "grad_norm": 0.6038889288902283, "learning_rate": 1.2704034946728519e-05, "loss": 1.8967062377929687, "step": 183450 }, { "epoch": 0.7300467965964303, "grad_norm": 0.7247796058654785, "learning_rate": 1.2702045720378431e-05, "loss": 2.0157620239257814, "step": 183500 }, { "epoch": 0.7302457194292904, "grad_norm": 0.6290791630744934, "learning_rate": 1.2700056494028343e-05, "loss": 1.9524681091308593, "step": 183550 }, { "epoch": 0.7304446422621504, "grad_norm": 0.5264587998390198, "learning_rate": 1.2698067267678255e-05, "loss": 1.9737319946289062, "step": 183600 }, { "epoch": 0.7306435650950105, "grad_norm": 0.5988287925720215, "learning_rate": 1.2696078041328167e-05, "loss": 1.9444287109375, "step": 183650 }, { "epoch": 0.7308424879278705, "grad_norm": 0.6984456777572632, "learning_rate": 1.2694088814978081e-05, "loss": 1.9128512573242187, "step": 183700 }, { "epoch": 0.7310414107607306, "grad_norm": 0.7278265357017517, "learning_rate": 1.269209958862799e-05, "loss": 1.915452423095703, "step": 183750 }, { "epoch": 0.7312403335935908, "grad_norm": 0.6737762689590454, "learning_rate": 1.2690110362277904e-05, "loss": 1.9387724304199219, "step": 183800 }, { "epoch": 0.7314392564264508, "grad_norm": 0.6666950583457947, "learning_rate": 1.2688121135927816e-05, "loss": 1.933101043701172, "step": 183850 }, { "epoch": 0.7316381792593109, "grad_norm": 0.6227391362190247, "learning_rate": 1.2686131909577728e-05, "loss": 1.9839114379882812, "step": 183900 }, { "epoch": 0.7318371020921709, "grad_norm": 0.5833902955055237, "learning_rate": 1.268414268322764e-05, "loss": 1.9706520080566405, "step": 183950 }, { "epoch": 0.732036024925031, "grad_norm": 0.6022323369979858, "learning_rate": 1.2682153456877552e-05, "loss": 1.9514497375488282, "step": 184000 }, { "epoch": 0.732234947757891, "grad_norm": 0.6850904822349548, "learning_rate": 1.2680164230527465e-05, "loss": 1.8952264404296875, "step": 184050 }, { "epoch": 0.732433870590751, "grad_norm": 0.5749801397323608, "learning_rate": 1.2678175004177375e-05, "loss": 1.9862222290039062, "step": 184100 }, { "epoch": 0.7326327934236111, "grad_norm": 0.6528343558311462, "learning_rate": 1.2676185777827287e-05, "loss": 1.9373661804199218, "step": 184150 }, { "epoch": 0.7328317162564713, "grad_norm": 0.5560740232467651, "learning_rate": 1.2674196551477201e-05, "loss": 1.9354573059082032, "step": 184200 }, { "epoch": 0.7330306390893313, "grad_norm": 0.5950138568878174, "learning_rate": 1.2672207325127113e-05, "loss": 1.9098251342773438, "step": 184250 }, { "epoch": 0.7332295619221914, "grad_norm": 1.3617678880691528, "learning_rate": 1.2670218098777025e-05, "loss": 1.9625839233398437, "step": 184300 }, { "epoch": 0.7334284847550514, "grad_norm": 0.6808899641036987, "learning_rate": 1.2668228872426937e-05, "loss": 1.9681033325195312, "step": 184350 }, { "epoch": 0.7336274075879115, "grad_norm": 0.585646390914917, "learning_rate": 1.2666279430603851e-05, "loss": 1.95216796875, "step": 184400 }, { "epoch": 0.7338263304207715, "grad_norm": 0.6211307048797607, "learning_rate": 1.2664290204253761e-05, "loss": 1.9591534423828125, "step": 184450 }, { "epoch": 0.7340252532536315, "grad_norm": 0.6043044924736023, "learning_rate": 1.2662300977903674e-05, "loss": 1.938054962158203, "step": 184500 }, { "epoch": 0.7342241760864916, "grad_norm": 0.6238291263580322, "learning_rate": 1.2660311751553587e-05, "loss": 1.946004638671875, "step": 184550 }, { "epoch": 0.7344230989193518, "grad_norm": 0.5355074405670166, "learning_rate": 1.26583225252035e-05, "loss": 1.9369435119628906, "step": 184600 }, { "epoch": 0.7346220217522118, "grad_norm": 0.6950331330299377, "learning_rate": 1.2656333298853412e-05, "loss": 1.9534225463867188, "step": 184650 }, { "epoch": 0.7348209445850719, "grad_norm": 0.6374898552894592, "learning_rate": 1.2654344072503324e-05, "loss": 1.9706024169921874, "step": 184700 }, { "epoch": 0.7350198674179319, "grad_norm": 0.5451593399047852, "learning_rate": 1.2652354846153234e-05, "loss": 1.876002655029297, "step": 184750 }, { "epoch": 0.735218790250792, "grad_norm": 0.6303057670593262, "learning_rate": 1.2650365619803146e-05, "loss": 1.9772503662109375, "step": 184800 }, { "epoch": 0.735417713083652, "grad_norm": 0.6128081679344177, "learning_rate": 1.2648376393453059e-05, "loss": 1.9035104370117188, "step": 184850 }, { "epoch": 0.735616635916512, "grad_norm": 0.6516140103340149, "learning_rate": 1.264638716710297e-05, "loss": 1.9537721252441407, "step": 184900 }, { "epoch": 0.7358155587493721, "grad_norm": 0.5399661064147949, "learning_rate": 1.2644397940752885e-05, "loss": 1.9212564086914063, "step": 184950 }, { "epoch": 0.7360144815822323, "grad_norm": 0.6399881839752197, "learning_rate": 1.2642408714402797e-05, "loss": 1.9863090515136719, "step": 185000 }, { "epoch": 0.7362134044150923, "grad_norm": 0.8219234347343445, "learning_rate": 1.2640419488052707e-05, "loss": 1.9156715393066406, "step": 185050 }, { "epoch": 0.7364123272479524, "grad_norm": 0.5949836373329163, "learning_rate": 1.263843026170262e-05, "loss": 1.9261154174804687, "step": 185100 }, { "epoch": 0.7366112500808124, "grad_norm": 0.6494293212890625, "learning_rate": 1.2636441035352531e-05, "loss": 1.965203094482422, "step": 185150 }, { "epoch": 0.7368101729136725, "grad_norm": 0.6637411117553711, "learning_rate": 1.2634451809002444e-05, "loss": 1.9389823913574218, "step": 185200 }, { "epoch": 0.7370090957465325, "grad_norm": 0.6984081864356995, "learning_rate": 1.2632462582652356e-05, "loss": 1.9994241333007812, "step": 185250 }, { "epoch": 0.7372080185793926, "grad_norm": 0.5835612416267395, "learning_rate": 1.2630473356302268e-05, "loss": 1.9245042419433593, "step": 185300 }, { "epoch": 0.7374069414122526, "grad_norm": 0.5946435928344727, "learning_rate": 1.2628484129952182e-05, "loss": 1.9509463500976563, "step": 185350 }, { "epoch": 0.7376058642451128, "grad_norm": 0.5122556686401367, "learning_rate": 1.262649490360209e-05, "loss": 1.9786224365234375, "step": 185400 }, { "epoch": 0.7378047870779728, "grad_norm": 0.6977052092552185, "learning_rate": 1.2624505677252004e-05, "loss": 2.002161560058594, "step": 185450 }, { "epoch": 0.7380037099108329, "grad_norm": 0.6616052985191345, "learning_rate": 1.2622516450901916e-05, "loss": 1.9530709838867188, "step": 185500 }, { "epoch": 0.7382026327436929, "grad_norm": 0.6438819169998169, "learning_rate": 1.2620527224551829e-05, "loss": 1.980137939453125, "step": 185550 }, { "epoch": 0.738401555576553, "grad_norm": 0.6279382109642029, "learning_rate": 1.261853799820174e-05, "loss": 1.9340579223632812, "step": 185600 }, { "epoch": 0.738600478409413, "grad_norm": 0.6998558044433594, "learning_rate": 1.2616548771851653e-05, "loss": 1.9436683654785156, "step": 185650 }, { "epoch": 0.7387994012422731, "grad_norm": 0.6279191970825195, "learning_rate": 1.2614559545501563e-05, "loss": 2.002871398925781, "step": 185700 }, { "epoch": 0.7389983240751331, "grad_norm": 0.4973312020301819, "learning_rate": 1.2612570319151476e-05, "loss": 1.919136505126953, "step": 185750 }, { "epoch": 0.7391972469079933, "grad_norm": 0.6297978162765503, "learning_rate": 1.2610581092801388e-05, "loss": 1.939774932861328, "step": 185800 }, { "epoch": 0.7393961697408533, "grad_norm": 0.6084054112434387, "learning_rate": 1.2608591866451301e-05, "loss": 1.9253514099121094, "step": 185850 }, { "epoch": 0.7395950925737134, "grad_norm": 0.7604281902313232, "learning_rate": 1.2606602640101214e-05, "loss": 2.0238528442382813, "step": 185900 }, { "epoch": 0.7397940154065734, "grad_norm": 0.539832353591919, "learning_rate": 1.2604613413751126e-05, "loss": 1.899652557373047, "step": 185950 }, { "epoch": 0.7399929382394335, "grad_norm": 0.5506983399391174, "learning_rate": 1.2602624187401038e-05, "loss": 1.9354859924316405, "step": 186000 }, { "epoch": 0.7401918610722935, "grad_norm": 0.5941636562347412, "learning_rate": 1.2600634961050948e-05, "loss": 1.9734521484375, "step": 186050 }, { "epoch": 0.7403907839051536, "grad_norm": 0.5850731134414673, "learning_rate": 1.259864573470086e-05, "loss": 1.9765145874023438, "step": 186100 }, { "epoch": 0.7405897067380136, "grad_norm": 0.6378508806228638, "learning_rate": 1.2596656508350773e-05, "loss": 1.9596424865722657, "step": 186150 }, { "epoch": 0.7407886295708738, "grad_norm": 0.6598408818244934, "learning_rate": 1.2594667282000685e-05, "loss": 1.9438648986816407, "step": 186200 }, { "epoch": 0.7409875524037338, "grad_norm": 0.6141682863235474, "learning_rate": 1.2592678055650599e-05, "loss": 1.9933685302734374, "step": 186250 }, { "epoch": 0.7411864752365939, "grad_norm": 0.5604131817817688, "learning_rate": 1.259068882930051e-05, "loss": 1.9549989318847656, "step": 186300 }, { "epoch": 0.7413853980694539, "grad_norm": 0.6214625835418701, "learning_rate": 1.2588699602950421e-05, "loss": 1.965437774658203, "step": 186350 }, { "epoch": 0.741584320902314, "grad_norm": 0.49954068660736084, "learning_rate": 1.2586710376600333e-05, "loss": 1.8918092346191406, "step": 186400 }, { "epoch": 0.741783243735174, "grad_norm": 0.4765719473361969, "learning_rate": 1.2584721150250246e-05, "loss": 1.9370394897460939, "step": 186450 }, { "epoch": 0.7419821665680341, "grad_norm": 0.6244306564331055, "learning_rate": 1.2582731923900158e-05, "loss": 1.9358932495117187, "step": 186500 }, { "epoch": 0.7421810894008941, "grad_norm": 0.543522298336029, "learning_rate": 1.258074269755007e-05, "loss": 1.9696284484863282, "step": 186550 }, { "epoch": 0.7423800122337543, "grad_norm": 0.6024730801582336, "learning_rate": 1.2578753471199982e-05, "loss": 2.0046128845214843, "step": 186600 }, { "epoch": 0.7425789350666143, "grad_norm": 0.6308577060699463, "learning_rate": 1.2576764244849896e-05, "loss": 1.9306222534179687, "step": 186650 }, { "epoch": 0.7427778578994744, "grad_norm": 0.6083489656448364, "learning_rate": 1.2574775018499805e-05, "loss": 1.8982382202148438, "step": 186700 }, { "epoch": 0.7429767807323344, "grad_norm": 0.6462936997413635, "learning_rate": 1.2572785792149718e-05, "loss": 1.9622337341308593, "step": 186750 }, { "epoch": 0.7431757035651945, "grad_norm": 0.6500908136367798, "learning_rate": 1.2570836350326632e-05, "loss": 1.933872833251953, "step": 186800 }, { "epoch": 0.7433746263980545, "grad_norm": 0.6475131511688232, "learning_rate": 1.2568847123976544e-05, "loss": 1.9592987060546876, "step": 186850 }, { "epoch": 0.7435735492309146, "grad_norm": 0.649922251701355, "learning_rate": 1.2566857897626456e-05, "loss": 1.975123291015625, "step": 186900 }, { "epoch": 0.7437724720637746, "grad_norm": 0.5681410431861877, "learning_rate": 1.2564868671276368e-05, "loss": 1.9152030944824219, "step": 186950 }, { "epoch": 0.7439713948966348, "grad_norm": 0.5884458422660828, "learning_rate": 1.2562879444926279e-05, "loss": 1.92569091796875, "step": 187000 }, { "epoch": 0.7441703177294948, "grad_norm": 0.5301565527915955, "learning_rate": 1.2560890218576191e-05, "loss": 1.9057086181640626, "step": 187050 }, { "epoch": 0.7443692405623549, "grad_norm": 0.6413630247116089, "learning_rate": 1.2558900992226105e-05, "loss": 1.9286408996582032, "step": 187100 }, { "epoch": 0.7445681633952149, "grad_norm": 0.6169342398643494, "learning_rate": 1.2556911765876017e-05, "loss": 1.9485758972167968, "step": 187150 }, { "epoch": 0.744767086228075, "grad_norm": 0.6084550619125366, "learning_rate": 1.2554922539525929e-05, "loss": 1.9712046813964843, "step": 187200 }, { "epoch": 0.744966009060935, "grad_norm": 0.52023845911026, "learning_rate": 1.2552933313175841e-05, "loss": 1.9536831665039063, "step": 187250 }, { "epoch": 0.7451649318937951, "grad_norm": 0.6068075299263, "learning_rate": 1.2550944086825753e-05, "loss": 1.973129425048828, "step": 187300 }, { "epoch": 0.7453638547266551, "grad_norm": 0.5798595547676086, "learning_rate": 1.2548954860475664e-05, "loss": 1.922500457763672, "step": 187350 }, { "epoch": 0.7455627775595153, "grad_norm": 0.552108108997345, "learning_rate": 1.2546965634125576e-05, "loss": 1.9074371337890625, "step": 187400 }, { "epoch": 0.7457617003923753, "grad_norm": 0.6848759651184082, "learning_rate": 1.2544976407775488e-05, "loss": 1.9724659729003906, "step": 187450 }, { "epoch": 0.7459606232252354, "grad_norm": 0.6164507269859314, "learning_rate": 1.2542987181425402e-05, "loss": 1.9170594787597657, "step": 187500 }, { "epoch": 0.7461595460580954, "grad_norm": 0.5924158096313477, "learning_rate": 1.2540997955075314e-05, "loss": 1.982908935546875, "step": 187550 }, { "epoch": 0.7463584688909555, "grad_norm": 0.5648995637893677, "learning_rate": 1.2539008728725226e-05, "loss": 1.9635476684570312, "step": 187600 }, { "epoch": 0.7465573917238155, "grad_norm": 0.6433842778205872, "learning_rate": 1.2537019502375137e-05, "loss": 1.9482302856445313, "step": 187650 }, { "epoch": 0.7467563145566756, "grad_norm": 0.5571646690368652, "learning_rate": 1.2535030276025049e-05, "loss": 1.9556559753417968, "step": 187700 }, { "epoch": 0.7469552373895356, "grad_norm": 0.6581423282623291, "learning_rate": 1.2533041049674961e-05, "loss": 1.9214071655273437, "step": 187750 }, { "epoch": 0.7471541602223958, "grad_norm": 0.6923399567604065, "learning_rate": 1.2531051823324873e-05, "loss": 1.9619145202636719, "step": 187800 }, { "epoch": 0.7473530830552558, "grad_norm": 0.58689284324646, "learning_rate": 1.2529062596974785e-05, "loss": 1.9438175964355469, "step": 187850 }, { "epoch": 0.7475520058881159, "grad_norm": 0.6582669019699097, "learning_rate": 1.2527073370624699e-05, "loss": 1.9305604553222657, "step": 187900 }, { "epoch": 0.7477509287209759, "grad_norm": 0.6136912107467651, "learning_rate": 1.2525084144274611e-05, "loss": 1.9348707580566407, "step": 187950 }, { "epoch": 0.747949851553836, "grad_norm": 0.6239882707595825, "learning_rate": 1.2523094917924522e-05, "loss": 1.9571202087402344, "step": 188000 }, { "epoch": 0.748148774386696, "grad_norm": 0.6022031307220459, "learning_rate": 1.2521105691574434e-05, "loss": 1.9274003601074219, "step": 188050 }, { "epoch": 0.7483476972195561, "grad_norm": 0.5684980750083923, "learning_rate": 1.2519116465224346e-05, "loss": 1.969081268310547, "step": 188100 }, { "epoch": 0.7485466200524161, "grad_norm": 0.5863581299781799, "learning_rate": 1.2517127238874258e-05, "loss": 1.9343229675292968, "step": 188150 }, { "epoch": 0.7487455428852763, "grad_norm": 0.5588148832321167, "learning_rate": 1.251513801252417e-05, "loss": 1.94597412109375, "step": 188200 }, { "epoch": 0.7489444657181363, "grad_norm": 0.46374964714050293, "learning_rate": 1.2513148786174082e-05, "loss": 1.9305307006835937, "step": 188250 }, { "epoch": 0.7491433885509964, "grad_norm": 0.6070159077644348, "learning_rate": 1.2511159559823993e-05, "loss": 1.9073593139648437, "step": 188300 }, { "epoch": 0.7493423113838564, "grad_norm": 0.3667885959148407, "learning_rate": 1.2509170333473905e-05, "loss": 1.986781463623047, "step": 188350 }, { "epoch": 0.7495412342167165, "grad_norm": 0.5583677291870117, "learning_rate": 1.2507181107123817e-05, "loss": 1.9378349304199218, "step": 188400 }, { "epoch": 0.7497401570495765, "grad_norm": 0.5900669097900391, "learning_rate": 1.2505191880773731e-05, "loss": 1.9344282531738282, "step": 188450 }, { "epoch": 0.7499390798824366, "grad_norm": 0.6816280484199524, "learning_rate": 1.2503202654423643e-05, "loss": 1.9241676330566406, "step": 188500 }, { "epoch": 0.7501380027152966, "grad_norm": 0.5383158326148987, "learning_rate": 1.2501213428073555e-05, "loss": 1.9379208374023438, "step": 188550 }, { "epoch": 0.7503369255481568, "grad_norm": 0.66990065574646, "learning_rate": 1.2499224201723467e-05, "loss": 1.9069918823242187, "step": 188600 }, { "epoch": 0.7505358483810168, "grad_norm": 0.5929400324821472, "learning_rate": 1.2497234975373378e-05, "loss": 1.9343853759765626, "step": 188650 }, { "epoch": 0.7507347712138769, "grad_norm": 0.5504324436187744, "learning_rate": 1.249524574902329e-05, "loss": 1.9397770690917968, "step": 188700 }, { "epoch": 0.7509336940467369, "grad_norm": 0.5781251788139343, "learning_rate": 1.2493256522673202e-05, "loss": 1.9464851379394532, "step": 188750 }, { "epoch": 0.751132616879597, "grad_norm": 0.6704772710800171, "learning_rate": 1.2491267296323114e-05, "loss": 1.9371870422363282, "step": 188800 }, { "epoch": 0.751331539712457, "grad_norm": 0.6421793103218079, "learning_rate": 1.2489278069973028e-05, "loss": 1.9319717407226562, "step": 188850 }, { "epoch": 0.7515304625453171, "grad_norm": 0.7113531231880188, "learning_rate": 1.248728884362294e-05, "loss": 1.9842829895019531, "step": 188900 }, { "epoch": 0.7517293853781771, "grad_norm": 0.5611582398414612, "learning_rate": 1.248529961727285e-05, "loss": 1.89019287109375, "step": 188950 }, { "epoch": 0.7519283082110373, "grad_norm": 0.5983324646949768, "learning_rate": 1.2483310390922763e-05, "loss": 1.9941795349121094, "step": 189000 }, { "epoch": 0.7521272310438973, "grad_norm": 0.5942187309265137, "learning_rate": 1.2481321164572675e-05, "loss": 1.96390869140625, "step": 189050 }, { "epoch": 0.7523261538767574, "grad_norm": 0.5692708492279053, "learning_rate": 1.2479331938222587e-05, "loss": 1.9008628845214843, "step": 189100 }, { "epoch": 0.7525250767096174, "grad_norm": 0.596514105796814, "learning_rate": 1.24773427118725e-05, "loss": 1.896670379638672, "step": 189150 }, { "epoch": 0.7527239995424775, "grad_norm": 0.636613667011261, "learning_rate": 1.2475353485522412e-05, "loss": 1.9845660400390626, "step": 189200 }, { "epoch": 0.7529229223753375, "grad_norm": 0.6595257520675659, "learning_rate": 1.2473364259172325e-05, "loss": 1.8871965026855468, "step": 189250 }, { "epoch": 0.7531218452081976, "grad_norm": 0.6547526121139526, "learning_rate": 1.2471375032822234e-05, "loss": 1.9297923278808593, "step": 189300 }, { "epoch": 0.7533207680410576, "grad_norm": 0.8229783773422241, "learning_rate": 1.2469385806472148e-05, "loss": 1.9678652954101563, "step": 189350 }, { "epoch": 0.7535196908739177, "grad_norm": 0.582162618637085, "learning_rate": 1.246739658012206e-05, "loss": 1.9449252319335937, "step": 189400 }, { "epoch": 0.7537186137067778, "grad_norm": 0.5986804366111755, "learning_rate": 1.2465407353771972e-05, "loss": 1.93623779296875, "step": 189450 }, { "epoch": 0.7539175365396379, "grad_norm": 0.6663188338279724, "learning_rate": 1.2463418127421884e-05, "loss": 1.9434333801269532, "step": 189500 }, { "epoch": 0.7541164593724979, "grad_norm": 0.6344797611236572, "learning_rate": 1.2461428901071797e-05, "loss": 1.9147743225097655, "step": 189550 }, { "epoch": 0.754315382205358, "grad_norm": 0.5701450705528259, "learning_rate": 1.2459439674721707e-05, "loss": 1.9489073181152343, "step": 189600 }, { "epoch": 0.754514305038218, "grad_norm": 0.7060979008674622, "learning_rate": 1.245745044837162e-05, "loss": 1.9806898498535157, "step": 189650 }, { "epoch": 0.7547132278710781, "grad_norm": 0.564128577709198, "learning_rate": 1.2455461222021531e-05, "loss": 1.9692385864257813, "step": 189700 }, { "epoch": 0.7549121507039381, "grad_norm": 0.5984684228897095, "learning_rate": 1.2453471995671445e-05, "loss": 1.9561831665039062, "step": 189750 }, { "epoch": 0.7551110735367982, "grad_norm": 0.6484797596931458, "learning_rate": 1.2451482769321357e-05, "loss": 1.9391996765136719, "step": 189800 }, { "epoch": 0.7553099963696583, "grad_norm": 0.6374527812004089, "learning_rate": 1.244949354297127e-05, "loss": 1.967880401611328, "step": 189850 }, { "epoch": 0.7555089192025184, "grad_norm": 0.7126078605651855, "learning_rate": 1.2447504316621182e-05, "loss": 1.987576904296875, "step": 189900 }, { "epoch": 0.7557078420353784, "grad_norm": 0.5762766003608704, "learning_rate": 1.2445515090271092e-05, "loss": 1.9278114318847657, "step": 189950 }, { "epoch": 0.7559067648682385, "grad_norm": 0.6562265753746033, "learning_rate": 1.2443525863921004e-05, "loss": 1.9669883728027344, "step": 190000 }, { "epoch": 0.7561056877010985, "grad_norm": 0.6078214645385742, "learning_rate": 1.2441536637570916e-05, "loss": 1.966114501953125, "step": 190050 }, { "epoch": 0.7563046105339586, "grad_norm": 0.6078852415084839, "learning_rate": 1.2439547411220828e-05, "loss": 1.9910740661621094, "step": 190100 }, { "epoch": 0.7565035333668186, "grad_norm": 0.6120779514312744, "learning_rate": 1.2437558184870742e-05, "loss": 1.9379685974121095, "step": 190150 }, { "epoch": 0.7567024561996787, "grad_norm": 0.5964395403862, "learning_rate": 1.2435568958520654e-05, "loss": 1.9074641418457032, "step": 190200 }, { "epoch": 0.7569013790325388, "grad_norm": 0.6070502400398254, "learning_rate": 1.2433619516697566e-05, "loss": 1.9935725402832032, "step": 190250 }, { "epoch": 0.7571003018653989, "grad_norm": 0.5682585835456848, "learning_rate": 1.2431630290347478e-05, "loss": 1.877220916748047, "step": 190300 }, { "epoch": 0.7572992246982589, "grad_norm": 0.6042733788490295, "learning_rate": 1.242964106399739e-05, "loss": 1.970352783203125, "step": 190350 }, { "epoch": 0.757498147531119, "grad_norm": 0.6396771669387817, "learning_rate": 1.2427651837647303e-05, "loss": 1.9135726928710937, "step": 190400 }, { "epoch": 0.757697070363979, "grad_norm": 0.5530056953430176, "learning_rate": 1.2425662611297215e-05, "loss": 1.950423126220703, "step": 190450 }, { "epoch": 0.7578959931968391, "grad_norm": 0.5322350859642029, "learning_rate": 1.2423673384947129e-05, "loss": 1.9104582214355468, "step": 190500 }, { "epoch": 0.7580949160296991, "grad_norm": 0.5284775495529175, "learning_rate": 1.242168415859704e-05, "loss": 1.9346527099609374, "step": 190550 }, { "epoch": 0.7582938388625592, "grad_norm": 0.6901416778564453, "learning_rate": 1.2419694932246951e-05, "loss": 1.951673583984375, "step": 190600 }, { "epoch": 0.7584927616954193, "grad_norm": 0.531219482421875, "learning_rate": 1.2417705705896863e-05, "loss": 1.947228240966797, "step": 190650 }, { "epoch": 0.7586916845282794, "grad_norm": 0.5725061893463135, "learning_rate": 1.2415716479546776e-05, "loss": 1.9337498474121093, "step": 190700 }, { "epoch": 0.7588906073611394, "grad_norm": 0.5766752362251282, "learning_rate": 1.2413727253196688e-05, "loss": 1.9773208618164062, "step": 190750 }, { "epoch": 0.7590895301939995, "grad_norm": 0.6281076073646545, "learning_rate": 1.24117380268466e-05, "loss": 1.949670867919922, "step": 190800 }, { "epoch": 0.7592884530268595, "grad_norm": 0.6839421987533569, "learning_rate": 1.2409748800496512e-05, "loss": 1.9526284790039063, "step": 190850 }, { "epoch": 0.7594873758597196, "grad_norm": 0.5899216532707214, "learning_rate": 1.2407759574146422e-05, "loss": 1.9689015197753905, "step": 190900 }, { "epoch": 0.7596862986925796, "grad_norm": 0.5896294116973877, "learning_rate": 1.2405770347796335e-05, "loss": 1.986099853515625, "step": 190950 }, { "epoch": 0.7598852215254397, "grad_norm": 0.5474247336387634, "learning_rate": 1.2403781121446248e-05, "loss": 1.9022877502441407, "step": 191000 }, { "epoch": 0.7600841443582999, "grad_norm": 0.4916377067565918, "learning_rate": 1.240179189509616e-05, "loss": 1.967469482421875, "step": 191050 }, { "epoch": 0.7602830671911599, "grad_norm": 0.7252951264381409, "learning_rate": 1.2399802668746073e-05, "loss": 1.9582240295410156, "step": 191100 }, { "epoch": 0.76048199002402, "grad_norm": 0.5798225998878479, "learning_rate": 1.2397813442395985e-05, "loss": 1.9321127319335938, "step": 191150 }, { "epoch": 0.76068091285688, "grad_norm": 0.6798336505889893, "learning_rate": 1.2395824216045897e-05, "loss": 1.918319091796875, "step": 191200 }, { "epoch": 0.76087983568974, "grad_norm": 0.5587425827980042, "learning_rate": 1.2393834989695808e-05, "loss": 1.9021572875976562, "step": 191250 }, { "epoch": 0.7610787585226001, "grad_norm": 0.6187124848365784, "learning_rate": 1.239184576334572e-05, "loss": 1.944899444580078, "step": 191300 }, { "epoch": 0.7612776813554601, "grad_norm": 0.538750410079956, "learning_rate": 1.2389856536995632e-05, "loss": 1.8865238952636718, "step": 191350 }, { "epoch": 0.7614766041883202, "grad_norm": 0.6531844735145569, "learning_rate": 1.2387867310645546e-05, "loss": 1.8986045837402343, "step": 191400 }, { "epoch": 0.7616755270211804, "grad_norm": 0.641205906867981, "learning_rate": 1.2385878084295458e-05, "loss": 1.9058287048339844, "step": 191450 }, { "epoch": 0.7618744498540404, "grad_norm": 0.6244486570358276, "learning_rate": 1.238388885794537e-05, "loss": 1.9791444396972657, "step": 191500 }, { "epoch": 0.7620733726869005, "grad_norm": 0.5876111388206482, "learning_rate": 1.238189963159528e-05, "loss": 1.9955342102050782, "step": 191550 }, { "epoch": 0.7622722955197605, "grad_norm": 0.5489661693572998, "learning_rate": 1.2379910405245193e-05, "loss": 1.9546223449707032, "step": 191600 }, { "epoch": 0.7624712183526206, "grad_norm": 0.644626796245575, "learning_rate": 1.2377921178895105e-05, "loss": 1.9395623779296876, "step": 191650 }, { "epoch": 0.7626701411854806, "grad_norm": 0.6100766658782959, "learning_rate": 1.2375931952545017e-05, "loss": 1.898372039794922, "step": 191700 }, { "epoch": 0.7628690640183406, "grad_norm": 0.5809467434883118, "learning_rate": 1.2373942726194929e-05, "loss": 1.9865158081054688, "step": 191750 }, { "epoch": 0.7630679868512007, "grad_norm": 0.5849765539169312, "learning_rate": 1.2371953499844843e-05, "loss": 1.9793511962890624, "step": 191800 }, { "epoch": 0.7632669096840609, "grad_norm": 0.643059253692627, "learning_rate": 1.2369964273494755e-05, "loss": 2.0013217163085937, "step": 191850 }, { "epoch": 0.7634658325169209, "grad_norm": 0.528741180896759, "learning_rate": 1.2367975047144665e-05, "loss": 1.9056381225585937, "step": 191900 }, { "epoch": 0.763664755349781, "grad_norm": 0.7012842893600464, "learning_rate": 1.2365985820794578e-05, "loss": 2.0410394287109375, "step": 191950 }, { "epoch": 0.763863678182641, "grad_norm": 0.6642756462097168, "learning_rate": 1.236399659444449e-05, "loss": 1.9486874389648436, "step": 192000 }, { "epoch": 0.764062601015501, "grad_norm": 0.570479691028595, "learning_rate": 1.2362007368094402e-05, "loss": 1.9307818603515625, "step": 192050 }, { "epoch": 0.7642615238483611, "grad_norm": 0.4906456470489502, "learning_rate": 1.2360018141744314e-05, "loss": 1.9305450439453125, "step": 192100 }, { "epoch": 0.7644604466812212, "grad_norm": 0.6213350892066956, "learning_rate": 1.2358028915394226e-05, "loss": 1.9421353149414062, "step": 192150 }, { "epoch": 0.7646593695140812, "grad_norm": 0.6608092188835144, "learning_rate": 1.2356039689044137e-05, "loss": 1.924646759033203, "step": 192200 }, { "epoch": 0.7648582923469414, "grad_norm": 0.6818744540214539, "learning_rate": 1.2354050462694049e-05, "loss": 2.0233750915527344, "step": 192250 }, { "epoch": 0.7650572151798014, "grad_norm": 0.5953768491744995, "learning_rate": 1.2352061236343963e-05, "loss": 1.9553729248046876, "step": 192300 }, { "epoch": 0.7652561380126615, "grad_norm": 0.4825112521648407, "learning_rate": 1.2350111794520876e-05, "loss": 1.9407960510253905, "step": 192350 }, { "epoch": 0.7654550608455215, "grad_norm": 0.6220096945762634, "learning_rate": 1.2348122568170788e-05, "loss": 1.9328691101074218, "step": 192400 }, { "epoch": 0.7656539836783816, "grad_norm": 0.5609855651855469, "learning_rate": 1.23461333418207e-05, "loss": 1.936971435546875, "step": 192450 }, { "epoch": 0.7658529065112416, "grad_norm": 0.5906381607055664, "learning_rate": 1.2344144115470613e-05, "loss": 1.9463912963867187, "step": 192500 }, { "epoch": 0.7660518293441017, "grad_norm": 0.4485863447189331, "learning_rate": 1.2342154889120523e-05, "loss": 1.95061767578125, "step": 192550 }, { "epoch": 0.7662507521769617, "grad_norm": 0.5629604458808899, "learning_rate": 1.2340165662770435e-05, "loss": 1.9699005126953124, "step": 192600 }, { "epoch": 0.7664496750098219, "grad_norm": 0.6696743369102478, "learning_rate": 1.2338176436420349e-05, "loss": 1.9793817138671874, "step": 192650 }, { "epoch": 0.7666485978426819, "grad_norm": 0.46740737557411194, "learning_rate": 1.2336187210070261e-05, "loss": 1.9452001953125, "step": 192700 }, { "epoch": 0.766847520675542, "grad_norm": 0.6513397097587585, "learning_rate": 1.2334197983720173e-05, "loss": 1.96916015625, "step": 192750 }, { "epoch": 0.767046443508402, "grad_norm": 0.5701234936714172, "learning_rate": 1.2332208757370085e-05, "loss": 1.8866084289550782, "step": 192800 }, { "epoch": 0.7672453663412621, "grad_norm": 0.5984917283058167, "learning_rate": 1.2330219531019996e-05, "loss": 1.9275721740722656, "step": 192850 }, { "epoch": 0.7674442891741221, "grad_norm": 0.6453930735588074, "learning_rate": 1.2328230304669908e-05, "loss": 1.9300213623046876, "step": 192900 }, { "epoch": 0.7676432120069822, "grad_norm": 0.6289764046669006, "learning_rate": 1.232624107831982e-05, "loss": 1.8671247863769531, "step": 192950 }, { "epoch": 0.7678421348398422, "grad_norm": 0.5564558506011963, "learning_rate": 1.2324251851969732e-05, "loss": 1.927646942138672, "step": 193000 }, { "epoch": 0.7680410576727024, "grad_norm": 0.4425605237483978, "learning_rate": 1.2322262625619646e-05, "loss": 1.9531613159179688, "step": 193050 }, { "epoch": 0.7682399805055624, "grad_norm": 0.5899165272712708, "learning_rate": 1.2320273399269558e-05, "loss": 1.9753453063964843, "step": 193100 }, { "epoch": 0.7684389033384225, "grad_norm": 0.5280678868293762, "learning_rate": 1.231828417291947e-05, "loss": 1.92173828125, "step": 193150 }, { "epoch": 0.7686378261712825, "grad_norm": 0.5987132787704468, "learning_rate": 1.2316294946569381e-05, "loss": 2.011602783203125, "step": 193200 }, { "epoch": 0.7688367490041426, "grad_norm": 0.6530902981758118, "learning_rate": 1.2314305720219293e-05, "loss": 1.9224876403808593, "step": 193250 }, { "epoch": 0.7690356718370026, "grad_norm": 0.5467698574066162, "learning_rate": 1.2312316493869205e-05, "loss": 1.9602473449707032, "step": 193300 }, { "epoch": 0.7692345946698627, "grad_norm": 0.5793338418006897, "learning_rate": 1.2310327267519117e-05, "loss": 1.922403564453125, "step": 193350 }, { "epoch": 0.7694335175027227, "grad_norm": 0.5687411427497864, "learning_rate": 1.230833804116903e-05, "loss": 1.9238687133789063, "step": 193400 }, { "epoch": 0.7696324403355829, "grad_norm": 0.5632157325744629, "learning_rate": 1.2306348814818943e-05, "loss": 1.9202723693847656, "step": 193450 }, { "epoch": 0.7698313631684429, "grad_norm": 0.5927729606628418, "learning_rate": 1.2304359588468852e-05, "loss": 1.9107838439941407, "step": 193500 }, { "epoch": 0.770030286001303, "grad_norm": 0.6932967305183411, "learning_rate": 1.2302370362118766e-05, "loss": 1.9369337463378906, "step": 193550 }, { "epoch": 0.770229208834163, "grad_norm": 0.6056913137435913, "learning_rate": 1.2300381135768678e-05, "loss": 1.9227716064453124, "step": 193600 }, { "epoch": 0.7704281316670231, "grad_norm": 0.5793431401252747, "learning_rate": 1.229839190941859e-05, "loss": 1.9356332397460938, "step": 193650 }, { "epoch": 0.7706270544998831, "grad_norm": 0.5921046137809753, "learning_rate": 1.2296402683068502e-05, "loss": 1.9070965576171874, "step": 193700 }, { "epoch": 0.7708259773327432, "grad_norm": 0.6398272514343262, "learning_rate": 1.2294413456718414e-05, "loss": 1.937576446533203, "step": 193750 }, { "epoch": 0.7710249001656032, "grad_norm": 0.7446639537811279, "learning_rate": 1.2292424230368327e-05, "loss": 1.9079478454589844, "step": 193800 }, { "epoch": 0.7712238229984634, "grad_norm": 0.5813820958137512, "learning_rate": 1.2290435004018237e-05, "loss": 1.9669277954101563, "step": 193850 }, { "epoch": 0.7714227458313234, "grad_norm": 0.6104487180709839, "learning_rate": 1.228844577766815e-05, "loss": 1.9840699768066405, "step": 193900 }, { "epoch": 0.7716216686641835, "grad_norm": 0.6108117699623108, "learning_rate": 1.2286456551318063e-05, "loss": 1.934291534423828, "step": 193950 }, { "epoch": 0.7718205914970435, "grad_norm": 0.5421832203865051, "learning_rate": 1.2284467324967975e-05, "loss": 1.9050169372558594, "step": 194000 }, { "epoch": 0.7720195143299036, "grad_norm": 0.6255433559417725, "learning_rate": 1.2282478098617887e-05, "loss": 1.9667514038085938, "step": 194050 }, { "epoch": 0.7722184371627636, "grad_norm": 0.5741948485374451, "learning_rate": 1.22804888722678e-05, "loss": 1.9044998168945313, "step": 194100 }, { "epoch": 0.7724173599956237, "grad_norm": 0.6257070302963257, "learning_rate": 1.227849964591771e-05, "loss": 1.94771240234375, "step": 194150 }, { "epoch": 0.7726162828284837, "grad_norm": 0.6391482353210449, "learning_rate": 1.2276510419567622e-05, "loss": 1.9714077758789061, "step": 194200 }, { "epoch": 0.7728152056613439, "grad_norm": 0.5842266082763672, "learning_rate": 1.2274521193217534e-05, "loss": 1.8917036437988282, "step": 194250 }, { "epoch": 0.7730141284942039, "grad_norm": 0.5203920006752014, "learning_rate": 1.2272531966867446e-05, "loss": 1.9169776916503907, "step": 194300 }, { "epoch": 0.773213051327064, "grad_norm": 0.6213693618774414, "learning_rate": 1.227054274051736e-05, "loss": 1.9766156005859374, "step": 194350 }, { "epoch": 0.773411974159924, "grad_norm": 0.6610056161880493, "learning_rate": 1.2268553514167272e-05, "loss": 1.9180789184570313, "step": 194400 }, { "epoch": 0.7736108969927841, "grad_norm": 0.5732936859130859, "learning_rate": 1.2266564287817185e-05, "loss": 1.8738339233398438, "step": 194450 }, { "epoch": 0.7738098198256441, "grad_norm": 0.7044534683227539, "learning_rate": 1.2264575061467095e-05, "loss": 1.9664344787597656, "step": 194500 }, { "epoch": 0.7740087426585042, "grad_norm": 0.6761036515235901, "learning_rate": 1.2262625619644008e-05, "loss": 1.9838343811035157, "step": 194550 }, { "epoch": 0.7742076654913642, "grad_norm": 0.57669997215271, "learning_rate": 1.226063639329392e-05, "loss": 1.9163592529296876, "step": 194600 }, { "epoch": 0.7744065883242244, "grad_norm": 0.6185814142227173, "learning_rate": 1.2258647166943833e-05, "loss": 2.0560595703125, "step": 194650 }, { "epoch": 0.7746055111570844, "grad_norm": 0.7514646053314209, "learning_rate": 1.2256657940593747e-05, "loss": 1.925876922607422, "step": 194700 }, { "epoch": 0.7748044339899445, "grad_norm": 0.6043378114700317, "learning_rate": 1.2254668714243659e-05, "loss": 1.934179229736328, "step": 194750 }, { "epoch": 0.7750033568228045, "grad_norm": 0.5704622864723206, "learning_rate": 1.225267948789357e-05, "loss": 1.947688751220703, "step": 194800 }, { "epoch": 0.7752022796556646, "grad_norm": 0.5609965920448303, "learning_rate": 1.2250690261543481e-05, "loss": 1.9850445556640626, "step": 194850 }, { "epoch": 0.7754012024885246, "grad_norm": 0.5162490010261536, "learning_rate": 1.2248701035193393e-05, "loss": 1.8894046020507813, "step": 194900 }, { "epoch": 0.7756001253213847, "grad_norm": 0.4499123692512512, "learning_rate": 1.2246711808843306e-05, "loss": 1.8772975158691407, "step": 194950 }, { "epoch": 0.7757990481542447, "grad_norm": 0.5987416505813599, "learning_rate": 1.2244722582493218e-05, "loss": 1.9595329284667968, "step": 195000 }, { "epoch": 0.7759979709871049, "grad_norm": 0.5701102614402771, "learning_rate": 1.224273335614313e-05, "loss": 1.9221231079101562, "step": 195050 }, { "epoch": 0.7761968938199649, "grad_norm": 0.5834150910377502, "learning_rate": 1.2240744129793044e-05, "loss": 1.9550486755371095, "step": 195100 }, { "epoch": 0.776395816652825, "grad_norm": 0.5884585976600647, "learning_rate": 1.2238754903442953e-05, "loss": 1.9495884704589843, "step": 195150 }, { "epoch": 0.776594739485685, "grad_norm": 0.6072821021080017, "learning_rate": 1.2236765677092866e-05, "loss": 1.9135453796386719, "step": 195200 }, { "epoch": 0.7767936623185451, "grad_norm": 0.6037423610687256, "learning_rate": 1.2234776450742779e-05, "loss": 1.9634854125976562, "step": 195250 }, { "epoch": 0.7769925851514051, "grad_norm": 0.606969952583313, "learning_rate": 1.223278722439269e-05, "loss": 1.95718505859375, "step": 195300 }, { "epoch": 0.7771915079842652, "grad_norm": 0.6905271410942078, "learning_rate": 1.2230797998042603e-05, "loss": 1.9563934326171875, "step": 195350 }, { "epoch": 0.7773904308171252, "grad_norm": 0.7048882842063904, "learning_rate": 1.2228808771692515e-05, "loss": 1.9068302917480469, "step": 195400 }, { "epoch": 0.7775893536499854, "grad_norm": 0.5893850922584534, "learning_rate": 1.2226819545342425e-05, "loss": 1.9361680603027345, "step": 195450 }, { "epoch": 0.7777882764828454, "grad_norm": 0.6321687698364258, "learning_rate": 1.2224830318992338e-05, "loss": 1.9809423828125, "step": 195500 }, { "epoch": 0.7779871993157055, "grad_norm": 0.5877612829208374, "learning_rate": 1.222284109264225e-05, "loss": 1.9931109619140626, "step": 195550 }, { "epoch": 0.7781861221485655, "grad_norm": 0.5920290946960449, "learning_rate": 1.2220851866292164e-05, "loss": 1.9525387573242188, "step": 195600 }, { "epoch": 0.7783850449814256, "grad_norm": 0.6168246269226074, "learning_rate": 1.2218862639942076e-05, "loss": 1.9184910583496093, "step": 195650 }, { "epoch": 0.7785839678142856, "grad_norm": 0.6897215843200684, "learning_rate": 1.2216873413591988e-05, "loss": 1.9200343322753906, "step": 195700 }, { "epoch": 0.7787828906471457, "grad_norm": 0.8047582507133484, "learning_rate": 1.22148841872419e-05, "loss": 1.924209442138672, "step": 195750 }, { "epoch": 0.7789818134800057, "grad_norm": 0.6524720191955566, "learning_rate": 1.221289496089181e-05, "loss": 1.9908775329589843, "step": 195800 }, { "epoch": 0.7791807363128659, "grad_norm": 0.6318880915641785, "learning_rate": 1.2210905734541723e-05, "loss": 1.95977294921875, "step": 195850 }, { "epoch": 0.7793796591457259, "grad_norm": 0.5379771590232849, "learning_rate": 1.2208916508191635e-05, "loss": 1.9307066345214843, "step": 195900 }, { "epoch": 0.779578581978586, "grad_norm": 0.680939257144928, "learning_rate": 1.2206927281841547e-05, "loss": 1.9878561401367187, "step": 195950 }, { "epoch": 0.779777504811446, "grad_norm": 0.610500156879425, "learning_rate": 1.2204938055491459e-05, "loss": 1.9211727905273437, "step": 196000 }, { "epoch": 0.7799764276443061, "grad_norm": 0.6987023949623108, "learning_rate": 1.2202948829141373e-05, "loss": 1.9507966613769532, "step": 196050 }, { "epoch": 0.7801753504771661, "grad_norm": 0.6102102398872375, "learning_rate": 1.2200959602791282e-05, "loss": 1.94753662109375, "step": 196100 }, { "epoch": 0.7803742733100262, "grad_norm": 0.6640422940254211, "learning_rate": 1.2198970376441195e-05, "loss": 1.96665771484375, "step": 196150 }, { "epoch": 0.7805731961428862, "grad_norm": 0.5211401581764221, "learning_rate": 1.2196981150091108e-05, "loss": 1.9460520935058594, "step": 196200 }, { "epoch": 0.7807721189757464, "grad_norm": 0.5555579662322998, "learning_rate": 1.219499192374102e-05, "loss": 1.9923466491699218, "step": 196250 }, { "epoch": 0.7809710418086064, "grad_norm": 0.5766854286193848, "learning_rate": 1.2193002697390932e-05, "loss": 1.9461627197265625, "step": 196300 }, { "epoch": 0.7811699646414665, "grad_norm": 0.6325852870941162, "learning_rate": 1.2191053255567847e-05, "loss": 2.002441864013672, "step": 196350 }, { "epoch": 0.7813688874743265, "grad_norm": 0.5661579370498657, "learning_rate": 1.218906402921776e-05, "loss": 1.9567849731445313, "step": 196400 }, { "epoch": 0.7815678103071866, "grad_norm": 0.7273218631744385, "learning_rate": 1.218707480286767e-05, "loss": 1.9122030639648437, "step": 196450 }, { "epoch": 0.7817667331400466, "grad_norm": 0.5474095940589905, "learning_rate": 1.2185085576517582e-05, "loss": 1.9703475952148437, "step": 196500 }, { "epoch": 0.7819656559729067, "grad_norm": 0.5441590547561646, "learning_rate": 1.2183096350167494e-05, "loss": 1.9188926696777344, "step": 196550 }, { "epoch": 0.7821645788057667, "grad_norm": 0.4744572043418884, "learning_rate": 1.2181107123817406e-05, "loss": 2.014176940917969, "step": 196600 }, { "epoch": 0.7823635016386269, "grad_norm": 0.6183485388755798, "learning_rate": 1.2179117897467318e-05, "loss": 1.9324378967285156, "step": 196650 }, { "epoch": 0.7825624244714869, "grad_norm": 0.5143067836761475, "learning_rate": 1.217712867111723e-05, "loss": 1.922552032470703, "step": 196700 }, { "epoch": 0.782761347304347, "grad_norm": 0.5756495594978333, "learning_rate": 1.2175139444767141e-05, "loss": 1.8942459106445313, "step": 196750 }, { "epoch": 0.782960270137207, "grad_norm": 0.5630496740341187, "learning_rate": 1.2173150218417053e-05, "loss": 1.9738027954101562, "step": 196800 }, { "epoch": 0.7831591929700671, "grad_norm": 0.6708203554153442, "learning_rate": 1.2171160992066967e-05, "loss": 1.9352777099609375, "step": 196850 }, { "epoch": 0.7833581158029271, "grad_norm": 0.6442953944206238, "learning_rate": 1.2169171765716879e-05, "loss": 1.9629678344726562, "step": 196900 }, { "epoch": 0.7835570386357872, "grad_norm": 0.5841683745384216, "learning_rate": 1.2167182539366791e-05, "loss": 1.923375701904297, "step": 196950 }, { "epoch": 0.7837559614686472, "grad_norm": 0.4706372022628784, "learning_rate": 1.2165193313016703e-05, "loss": 1.8878518676757812, "step": 197000 }, { "epoch": 0.7839548843015074, "grad_norm": 0.6271761655807495, "learning_rate": 1.2163204086666615e-05, "loss": 2.005648651123047, "step": 197050 }, { "epoch": 0.7841538071343674, "grad_norm": 0.5848802328109741, "learning_rate": 1.2161214860316526e-05, "loss": 1.9434217834472656, "step": 197100 }, { "epoch": 0.7843527299672275, "grad_norm": 0.5892587900161743, "learning_rate": 1.2159225633966438e-05, "loss": 1.9610653686523438, "step": 197150 }, { "epoch": 0.7845516528000875, "grad_norm": 0.5999188423156738, "learning_rate": 1.215723640761635e-05, "loss": 1.9848764038085938, "step": 197200 }, { "epoch": 0.7847505756329476, "grad_norm": 0.5441652536392212, "learning_rate": 1.2155247181266262e-05, "loss": 1.9442658996582032, "step": 197250 }, { "epoch": 0.7849494984658076, "grad_norm": 0.7878192663192749, "learning_rate": 1.2153257954916176e-05, "loss": 1.9744242858886718, "step": 197300 }, { "epoch": 0.7851484212986677, "grad_norm": 0.601412296295166, "learning_rate": 1.2151268728566088e-05, "loss": 1.9682815551757813, "step": 197350 }, { "epoch": 0.7853473441315277, "grad_norm": 0.5961347818374634, "learning_rate": 1.2149279502215999e-05, "loss": 1.892522430419922, "step": 197400 }, { "epoch": 0.7855462669643878, "grad_norm": 0.5602604150772095, "learning_rate": 1.2147290275865911e-05, "loss": 1.9047080993652343, "step": 197450 }, { "epoch": 0.785745189797248, "grad_norm": 0.6065346598625183, "learning_rate": 1.2145301049515823e-05, "loss": 1.9490904235839843, "step": 197500 }, { "epoch": 0.785944112630108, "grad_norm": 0.6242186427116394, "learning_rate": 1.2143311823165735e-05, "loss": 1.9478302001953125, "step": 197550 }, { "epoch": 0.786143035462968, "grad_norm": 0.7083110809326172, "learning_rate": 1.2141322596815647e-05, "loss": 1.9656254577636718, "step": 197600 }, { "epoch": 0.7863419582958281, "grad_norm": 0.5806460380554199, "learning_rate": 1.213933337046556e-05, "loss": 2.0170321655273438, "step": 197650 }, { "epoch": 0.7865408811286881, "grad_norm": 0.4950401186943054, "learning_rate": 1.2137344144115473e-05, "loss": 1.9703486633300782, "step": 197700 }, { "epoch": 0.7867398039615482, "grad_norm": 0.6264001727104187, "learning_rate": 1.2135354917765382e-05, "loss": 1.9491929626464843, "step": 197750 }, { "epoch": 0.7869387267944082, "grad_norm": 0.4888485372066498, "learning_rate": 1.2133365691415296e-05, "loss": 1.8944094848632813, "step": 197800 }, { "epoch": 0.7871376496272683, "grad_norm": 0.6208372116088867, "learning_rate": 1.2131376465065208e-05, "loss": 1.9098463439941407, "step": 197850 }, { "epoch": 0.7873365724601284, "grad_norm": 0.6481223702430725, "learning_rate": 1.212938723871512e-05, "loss": 1.9044229125976562, "step": 197900 }, { "epoch": 0.7875354952929885, "grad_norm": 0.48427775502204895, "learning_rate": 1.2127398012365032e-05, "loss": 1.9217359924316406, "step": 197950 }, { "epoch": 0.7877344181258485, "grad_norm": 0.5838560461997986, "learning_rate": 1.2125408786014945e-05, "loss": 1.9643988037109374, "step": 198000 }, { "epoch": 0.7879333409587086, "grad_norm": 0.6462004780769348, "learning_rate": 1.2123419559664855e-05, "loss": 1.9903878784179687, "step": 198050 }, { "epoch": 0.7881322637915686, "grad_norm": 0.6336300373077393, "learning_rate": 1.2121430333314767e-05, "loss": 1.9558709716796876, "step": 198100 }, { "epoch": 0.7883311866244287, "grad_norm": 0.5971291661262512, "learning_rate": 1.211944110696468e-05, "loss": 1.891473388671875, "step": 198150 }, { "epoch": 0.7885301094572887, "grad_norm": 0.5848082304000854, "learning_rate": 1.2117451880614593e-05, "loss": 1.9608631896972657, "step": 198200 }, { "epoch": 0.7887290322901488, "grad_norm": 0.5624761581420898, "learning_rate": 1.2115462654264505e-05, "loss": 1.9628468322753907, "step": 198250 }, { "epoch": 0.788927955123009, "grad_norm": 0.6603236198425293, "learning_rate": 1.2113473427914417e-05, "loss": 1.9702253723144532, "step": 198300 }, { "epoch": 0.789126877955869, "grad_norm": 0.6366989016532898, "learning_rate": 1.211148420156433e-05, "loss": 1.9170071411132812, "step": 198350 }, { "epoch": 0.789325800788729, "grad_norm": 0.609248161315918, "learning_rate": 1.210949497521424e-05, "loss": 1.9572970581054687, "step": 198400 }, { "epoch": 0.7895247236215891, "grad_norm": 0.64845210313797, "learning_rate": 1.2107505748864152e-05, "loss": 1.972939453125, "step": 198450 }, { "epoch": 0.7897236464544491, "grad_norm": 0.5906059145927429, "learning_rate": 1.2105516522514064e-05, "loss": 1.964358367919922, "step": 198500 }, { "epoch": 0.7899225692873092, "grad_norm": 0.6100972890853882, "learning_rate": 1.2103527296163976e-05, "loss": 1.8788162231445313, "step": 198550 }, { "epoch": 0.7901214921201692, "grad_norm": 0.6311392784118652, "learning_rate": 1.210153806981389e-05, "loss": 1.9709199523925782, "step": 198600 }, { "epoch": 0.7903204149530293, "grad_norm": 0.5994837880134583, "learning_rate": 1.2099548843463802e-05, "loss": 1.8965144348144531, "step": 198650 }, { "epoch": 0.7905193377858895, "grad_norm": 0.6310065984725952, "learning_rate": 1.2097559617113713e-05, "loss": 1.9507510375976562, "step": 198700 }, { "epoch": 0.7907182606187495, "grad_norm": 0.5415756106376648, "learning_rate": 1.2095570390763625e-05, "loss": 1.926601104736328, "step": 198750 }, { "epoch": 0.7909171834516096, "grad_norm": 0.6316993832588196, "learning_rate": 1.2093581164413537e-05, "loss": 1.9042921447753907, "step": 198800 }, { "epoch": 0.7911161062844696, "grad_norm": 0.6468856930732727, "learning_rate": 1.209159193806345e-05, "loss": 1.9027493286132813, "step": 198850 }, { "epoch": 0.7913150291173296, "grad_norm": 0.5844302177429199, "learning_rate": 1.2089602711713361e-05, "loss": 1.940150909423828, "step": 198900 }, { "epoch": 0.7915139519501897, "grad_norm": 0.652897298336029, "learning_rate": 1.2087613485363274e-05, "loss": 1.925043182373047, "step": 198950 }, { "epoch": 0.7917128747830497, "grad_norm": 0.572955310344696, "learning_rate": 1.2085624259013184e-05, "loss": 1.9161903381347656, "step": 199000 }, { "epoch": 0.7919117976159098, "grad_norm": 0.5787946581840515, "learning_rate": 1.2083635032663096e-05, "loss": 1.9500846862792969, "step": 199050 }, { "epoch": 0.79211072044877, "grad_norm": 0.6597450375556946, "learning_rate": 1.208164580631301e-05, "loss": 1.9429718017578126, "step": 199100 }, { "epoch": 0.79230964328163, "grad_norm": 0.6177992820739746, "learning_rate": 1.2079656579962922e-05, "loss": 1.9413931274414062, "step": 199150 }, { "epoch": 0.79250856611449, "grad_norm": 0.55156010389328, "learning_rate": 1.2077667353612834e-05, "loss": 1.9575128173828125, "step": 199200 }, { "epoch": 0.7927074889473501, "grad_norm": 0.6155908703804016, "learning_rate": 1.2075678127262746e-05, "loss": 1.9898081970214845, "step": 199250 }, { "epoch": 0.7929064117802102, "grad_norm": 0.6823394298553467, "learning_rate": 1.2073688900912659e-05, "loss": 1.9461488342285156, "step": 199300 }, { "epoch": 0.7931053346130702, "grad_norm": 0.6617546677589417, "learning_rate": 1.2071699674562569e-05, "loss": 1.927924346923828, "step": 199350 }, { "epoch": 0.7933042574459302, "grad_norm": 0.676479160785675, "learning_rate": 1.2069710448212481e-05, "loss": 1.9881094360351563, "step": 199400 }, { "epoch": 0.7935031802787903, "grad_norm": 0.6422195434570312, "learning_rate": 1.2067721221862393e-05, "loss": 1.9372315979003907, "step": 199450 }, { "epoch": 0.7937021031116505, "grad_norm": 0.5472992062568665, "learning_rate": 1.2065731995512307e-05, "loss": 1.90921630859375, "step": 199500 }, { "epoch": 0.7939010259445105, "grad_norm": 0.6814062595367432, "learning_rate": 1.206374276916222e-05, "loss": 1.9069778442382812, "step": 199550 }, { "epoch": 0.7940999487773706, "grad_norm": 0.6769008636474609, "learning_rate": 1.2061753542812131e-05, "loss": 1.9603173828125, "step": 199600 }, { "epoch": 0.7942988716102306, "grad_norm": 0.5881928205490112, "learning_rate": 1.2059764316462042e-05, "loss": 1.9212384033203125, "step": 199650 }, { "epoch": 0.7944977944430907, "grad_norm": 0.5083538889884949, "learning_rate": 1.2057775090111954e-05, "loss": 1.8845079040527344, "step": 199700 }, { "epoch": 0.7946967172759507, "grad_norm": 0.6237191557884216, "learning_rate": 1.2055785863761866e-05, "loss": 1.912570037841797, "step": 199750 }, { "epoch": 0.7948956401088108, "grad_norm": 0.5974951982498169, "learning_rate": 1.2053796637411778e-05, "loss": 1.9630242919921874, "step": 199800 }, { "epoch": 0.7950945629416708, "grad_norm": 0.5113774538040161, "learning_rate": 1.205180741106169e-05, "loss": 1.96699462890625, "step": 199850 }, { "epoch": 0.795293485774531, "grad_norm": 0.5616562962532043, "learning_rate": 1.2049818184711604e-05, "loss": 1.976557159423828, "step": 199900 }, { "epoch": 0.795492408607391, "grad_norm": 0.6350754499435425, "learning_rate": 1.2047828958361517e-05, "loss": 1.935892333984375, "step": 199950 }, { "epoch": 0.7956913314402511, "grad_norm": 0.5444859862327576, "learning_rate": 1.2045839732011427e-05, "loss": 1.9392668151855468, "step": 200000 }, { "epoch": 0.7958902542731111, "grad_norm": 0.5903211236000061, "learning_rate": 1.2043850505661339e-05, "loss": 1.9527198791503906, "step": 200050 }, { "epoch": 0.7960891771059712, "grad_norm": 0.6352442502975464, "learning_rate": 1.2041861279311251e-05, "loss": 1.89960693359375, "step": 200100 }, { "epoch": 0.7962880999388312, "grad_norm": 0.5924007296562195, "learning_rate": 1.2039872052961163e-05, "loss": 1.9969813537597656, "step": 200150 }, { "epoch": 0.7964870227716913, "grad_norm": 0.5782442688941956, "learning_rate": 1.2037882826611076e-05, "loss": 1.9992733764648438, "step": 200200 }, { "epoch": 0.7966859456045513, "grad_norm": 0.5626468658447266, "learning_rate": 1.2035893600260988e-05, "loss": 1.9006230163574218, "step": 200250 }, { "epoch": 0.7968848684374115, "grad_norm": 0.652040421962738, "learning_rate": 1.2033904373910898e-05, "loss": 1.9626589965820314, "step": 200300 }, { "epoch": 0.7970837912702715, "grad_norm": 0.76136714220047, "learning_rate": 1.203191514756081e-05, "loss": 1.9535882568359375, "step": 200350 }, { "epoch": 0.7972827141031316, "grad_norm": 0.5419869422912598, "learning_rate": 1.2029925921210724e-05, "loss": 1.9308979797363282, "step": 200400 }, { "epoch": 0.7974816369359916, "grad_norm": 0.6049144268035889, "learning_rate": 1.2027936694860636e-05, "loss": 1.907091827392578, "step": 200450 }, { "epoch": 0.7976805597688517, "grad_norm": 0.5432886481285095, "learning_rate": 1.2025947468510548e-05, "loss": 1.960343780517578, "step": 200500 }, { "epoch": 0.7978794826017117, "grad_norm": 0.6914539933204651, "learning_rate": 1.202395824216046e-05, "loss": 1.9184635925292968, "step": 200550 }, { "epoch": 0.7980784054345718, "grad_norm": 0.6287057399749756, "learning_rate": 1.2021969015810373e-05, "loss": 1.9392576599121094, "step": 200600 }, { "epoch": 0.7982773282674318, "grad_norm": 0.657248854637146, "learning_rate": 1.2019979789460283e-05, "loss": 1.933671875, "step": 200650 }, { "epoch": 0.798476251100292, "grad_norm": 0.6020541191101074, "learning_rate": 1.2017990563110195e-05, "loss": 1.8991355895996094, "step": 200700 }, { "epoch": 0.798675173933152, "grad_norm": 0.6398270726203918, "learning_rate": 1.2016001336760107e-05, "loss": 1.9695057678222656, "step": 200750 }, { "epoch": 0.7988740967660121, "grad_norm": 0.6187320351600647, "learning_rate": 1.2014012110410021e-05, "loss": 1.9813388061523438, "step": 200800 }, { "epoch": 0.7990730195988721, "grad_norm": 0.5614232420921326, "learning_rate": 1.2012022884059933e-05, "loss": 1.8848609924316406, "step": 200850 }, { "epoch": 0.7992719424317322, "grad_norm": 0.6393321752548218, "learning_rate": 1.2010033657709846e-05, "loss": 1.9658773803710938, "step": 200900 }, { "epoch": 0.7994708652645922, "grad_norm": 0.5726595520973206, "learning_rate": 1.2008044431359756e-05, "loss": 1.937619171142578, "step": 200950 }, { "epoch": 0.7996697880974523, "grad_norm": 0.6448296904563904, "learning_rate": 1.2006055205009668e-05, "loss": 1.9036326599121094, "step": 201000 }, { "epoch": 0.7998687109303123, "grad_norm": 0.644599199295044, "learning_rate": 1.200406597865958e-05, "loss": 1.9403640747070312, "step": 201050 }, { "epoch": 0.8000676337631725, "grad_norm": 0.45277804136276245, "learning_rate": 1.2002076752309492e-05, "loss": 1.9638076782226563, "step": 201100 }, { "epoch": 0.8002665565960325, "grad_norm": 0.5955109000205994, "learning_rate": 1.2000087525959405e-05, "loss": 1.938992462158203, "step": 201150 }, { "epoch": 0.8004654794288926, "grad_norm": 0.5810053944587708, "learning_rate": 1.1998098299609317e-05, "loss": 1.9200843811035155, "step": 201200 }, { "epoch": 0.8006644022617526, "grad_norm": 0.6798883676528931, "learning_rate": 1.199610907325923e-05, "loss": 1.9638986206054687, "step": 201250 }, { "epoch": 0.8008633250946127, "grad_norm": 0.7060837745666504, "learning_rate": 1.199411984690914e-05, "loss": 1.9932861328125, "step": 201300 }, { "epoch": 0.8010622479274727, "grad_norm": 0.5739895105361938, "learning_rate": 1.1992130620559053e-05, "loss": 1.9059382629394532, "step": 201350 }, { "epoch": 0.8012611707603328, "grad_norm": 0.6451419591903687, "learning_rate": 1.1990141394208965e-05, "loss": 1.9711541748046875, "step": 201400 }, { "epoch": 0.8014600935931928, "grad_norm": 0.5987159609794617, "learning_rate": 1.1988152167858878e-05, "loss": 1.9513308715820312, "step": 201450 }, { "epoch": 0.801659016426053, "grad_norm": 0.5701179504394531, "learning_rate": 1.198616294150879e-05, "loss": 1.95262939453125, "step": 201500 }, { "epoch": 0.801857939258913, "grad_norm": 0.6056580543518066, "learning_rate": 1.1984173715158702e-05, "loss": 1.8865708923339843, "step": 201550 }, { "epoch": 0.8020568620917731, "grad_norm": 0.6073800921440125, "learning_rate": 1.1982184488808612e-05, "loss": 1.9762660217285157, "step": 201600 }, { "epoch": 0.8022557849246331, "grad_norm": 0.6688972115516663, "learning_rate": 1.1980195262458524e-05, "loss": 1.931205291748047, "step": 201650 }, { "epoch": 0.8024547077574932, "grad_norm": 0.6274455189704895, "learning_rate": 1.1978206036108437e-05, "loss": 1.9094000244140625, "step": 201700 }, { "epoch": 0.8026536305903532, "grad_norm": 0.6355403661727905, "learning_rate": 1.197621680975835e-05, "loss": 1.9136001586914062, "step": 201750 }, { "epoch": 0.8028525534232133, "grad_norm": 0.5865337252616882, "learning_rate": 1.1974227583408263e-05, "loss": 1.916754150390625, "step": 201800 }, { "epoch": 0.8030514762560733, "grad_norm": 0.47263166308403015, "learning_rate": 1.1972238357058175e-05, "loss": 1.8938270568847657, "step": 201850 }, { "epoch": 0.8032503990889335, "grad_norm": 0.5869203805923462, "learning_rate": 1.1970249130708087e-05, "loss": 1.8704347229003906, "step": 201900 }, { "epoch": 0.8034493219217935, "grad_norm": 0.5820178985595703, "learning_rate": 1.1968259904357997e-05, "loss": 1.88730712890625, "step": 201950 }, { "epoch": 0.8036482447546536, "grad_norm": 0.5576879382133484, "learning_rate": 1.196627067800791e-05, "loss": 1.910569305419922, "step": 202000 }, { "epoch": 0.8038471675875136, "grad_norm": 0.6310747861862183, "learning_rate": 1.1964281451657822e-05, "loss": 1.9333975219726562, "step": 202050 }, { "epoch": 0.8040460904203737, "grad_norm": 0.5375069975852966, "learning_rate": 1.1962292225307734e-05, "loss": 1.928017578125, "step": 202100 }, { "epoch": 0.8042450132532337, "grad_norm": 0.6202073693275452, "learning_rate": 1.1960302998957648e-05, "loss": 1.900174560546875, "step": 202150 }, { "epoch": 0.8044439360860938, "grad_norm": 0.6028382182121277, "learning_rate": 1.195831377260756e-05, "loss": 1.909874267578125, "step": 202200 }, { "epoch": 0.8046428589189538, "grad_norm": 0.6398399472236633, "learning_rate": 1.195632454625747e-05, "loss": 1.9001702880859375, "step": 202250 }, { "epoch": 0.804841781751814, "grad_norm": 0.6194895505905151, "learning_rate": 1.1954335319907382e-05, "loss": 1.913551025390625, "step": 202300 }, { "epoch": 0.805040704584674, "grad_norm": 0.6476233601570129, "learning_rate": 1.1952385878084296e-05, "loss": 1.949206085205078, "step": 202350 }, { "epoch": 0.8052396274175341, "grad_norm": 0.6381654739379883, "learning_rate": 1.1950396651734208e-05, "loss": 1.9509616088867188, "step": 202400 }, { "epoch": 0.8054385502503941, "grad_norm": 0.6362107396125793, "learning_rate": 1.194840742538412e-05, "loss": 1.9199319458007813, "step": 202450 }, { "epoch": 0.8056374730832542, "grad_norm": 0.66179358959198, "learning_rate": 1.1946418199034034e-05, "loss": 1.9135757446289063, "step": 202500 }, { "epoch": 0.8058363959161142, "grad_norm": 0.6748512983322144, "learning_rate": 1.1944428972683946e-05, "loss": 1.8805540466308595, "step": 202550 }, { "epoch": 0.8060353187489743, "grad_norm": 0.5686279535293579, "learning_rate": 1.1942439746333857e-05, "loss": 1.9257058715820312, "step": 202600 }, { "epoch": 0.8062342415818343, "grad_norm": 0.5702764391899109, "learning_rate": 1.194049030451077e-05, "loss": 1.916185302734375, "step": 202650 }, { "epoch": 0.8064331644146945, "grad_norm": 0.6464744210243225, "learning_rate": 1.1938501078160682e-05, "loss": 1.9723316955566406, "step": 202700 }, { "epoch": 0.8066320872475545, "grad_norm": 0.6428667306900024, "learning_rate": 1.1936511851810594e-05, "loss": 1.9488914489746094, "step": 202750 }, { "epoch": 0.8068310100804146, "grad_norm": 0.8081175088882446, "learning_rate": 1.1934522625460508e-05, "loss": 1.9925521850585937, "step": 202800 }, { "epoch": 0.8070299329132746, "grad_norm": 0.6115913391113281, "learning_rate": 1.193253339911042e-05, "loss": 1.9600645446777343, "step": 202850 }, { "epoch": 0.8072288557461347, "grad_norm": 0.6853653192520142, "learning_rate": 1.193054417276033e-05, "loss": 1.9253111267089844, "step": 202900 }, { "epoch": 0.8074277785789947, "grad_norm": 0.5980024933815002, "learning_rate": 1.1928554946410243e-05, "loss": 1.8945156860351562, "step": 202950 }, { "epoch": 0.8076267014118548, "grad_norm": 0.6708159446716309, "learning_rate": 1.1926565720060155e-05, "loss": 1.9133377075195312, "step": 203000 }, { "epoch": 0.8078256242447148, "grad_norm": 0.6411265134811401, "learning_rate": 1.1924576493710067e-05, "loss": 1.9249470520019532, "step": 203050 }, { "epoch": 0.808024547077575, "grad_norm": 0.6482679843902588, "learning_rate": 1.192258726735998e-05, "loss": 1.928023223876953, "step": 203100 }, { "epoch": 0.808223469910435, "grad_norm": 0.6377198100090027, "learning_rate": 1.1920637825536895e-05, "loss": 1.9313941955566407, "step": 203150 }, { "epoch": 0.8084223927432951, "grad_norm": 0.48177415132522583, "learning_rate": 1.1918648599186807e-05, "loss": 1.9706822204589844, "step": 203200 }, { "epoch": 0.8086213155761551, "grad_norm": 0.5132790803909302, "learning_rate": 1.1916659372836717e-05, "loss": 1.9178875732421874, "step": 203250 }, { "epoch": 0.8088202384090152, "grad_norm": 0.606738269329071, "learning_rate": 1.191467014648663e-05, "loss": 1.9451702880859374, "step": 203300 }, { "epoch": 0.8090191612418752, "grad_norm": 0.8044711947441101, "learning_rate": 1.1912680920136541e-05, "loss": 1.936609344482422, "step": 203350 }, { "epoch": 0.8092180840747353, "grad_norm": 0.5441716909408569, "learning_rate": 1.1910691693786454e-05, "loss": 1.9119570922851563, "step": 203400 }, { "epoch": 0.8094170069075953, "grad_norm": 0.63392573595047, "learning_rate": 1.1908702467436366e-05, "loss": 1.8827073669433594, "step": 203450 }, { "epoch": 0.8096159297404555, "grad_norm": 0.6414943337440491, "learning_rate": 1.1906713241086278e-05, "loss": 1.944656524658203, "step": 203500 }, { "epoch": 0.8098148525733155, "grad_norm": 0.6564260721206665, "learning_rate": 1.1904724014736192e-05, "loss": 1.9861293029785156, "step": 203550 }, { "epoch": 0.8100137754061756, "grad_norm": 0.689434826374054, "learning_rate": 1.19027347883861e-05, "loss": 1.9473045349121094, "step": 203600 }, { "epoch": 0.8102126982390356, "grad_norm": 0.5930472016334534, "learning_rate": 1.1900745562036014e-05, "loss": 1.8895745849609376, "step": 203650 }, { "epoch": 0.8104116210718957, "grad_norm": 0.6171531677246094, "learning_rate": 1.1898756335685926e-05, "loss": 1.934911651611328, "step": 203700 }, { "epoch": 0.8106105439047557, "grad_norm": 0.6606844663619995, "learning_rate": 1.1896767109335839e-05, "loss": 1.937337646484375, "step": 203750 }, { "epoch": 0.8108094667376158, "grad_norm": 0.5514217615127563, "learning_rate": 1.189477788298575e-05, "loss": 1.9289376831054688, "step": 203800 }, { "epoch": 0.8110083895704758, "grad_norm": 0.592788815498352, "learning_rate": 1.1892788656635663e-05, "loss": 1.9698759460449218, "step": 203850 }, { "epoch": 0.811207312403336, "grad_norm": 0.37070879340171814, "learning_rate": 1.1890799430285573e-05, "loss": 1.9177113342285157, "step": 203900 }, { "epoch": 0.811406235236196, "grad_norm": 0.6045830845832825, "learning_rate": 1.1888810203935485e-05, "loss": 1.9430006408691407, "step": 203950 }, { "epoch": 0.8116051580690561, "grad_norm": 0.5837911367416382, "learning_rate": 1.1886820977585398e-05, "loss": 1.9771121215820313, "step": 204000 }, { "epoch": 0.8118040809019161, "grad_norm": 0.6636357307434082, "learning_rate": 1.1884831751235311e-05, "loss": 1.9600627136230468, "step": 204050 }, { "epoch": 0.8120030037347762, "grad_norm": 0.6878066062927246, "learning_rate": 1.1882842524885224e-05, "loss": 1.9370614624023437, "step": 204100 }, { "epoch": 0.8122019265676362, "grad_norm": 0.5821247696876526, "learning_rate": 1.1880853298535136e-05, "loss": 1.923699188232422, "step": 204150 }, { "epoch": 0.8124008494004963, "grad_norm": 0.9563537240028381, "learning_rate": 1.1878864072185046e-05, "loss": 1.9732893371582032, "step": 204200 }, { "epoch": 0.8125997722333563, "grad_norm": 0.925731897354126, "learning_rate": 1.1876874845834958e-05, "loss": 1.9074984741210939, "step": 204250 }, { "epoch": 0.8127986950662165, "grad_norm": 0.5961201190948486, "learning_rate": 1.187488561948487e-05, "loss": 1.9425650024414063, "step": 204300 }, { "epoch": 0.8129976178990765, "grad_norm": 0.5629404187202454, "learning_rate": 1.1872896393134783e-05, "loss": 1.9246463012695312, "step": 204350 }, { "epoch": 0.8131965407319366, "grad_norm": 0.5594759583473206, "learning_rate": 1.1870907166784695e-05, "loss": 1.9253865051269532, "step": 204400 }, { "epoch": 0.8133954635647966, "grad_norm": 0.5558532476425171, "learning_rate": 1.1868917940434607e-05, "loss": 1.9061091613769532, "step": 204450 }, { "epoch": 0.8135943863976567, "grad_norm": 0.7064305543899536, "learning_rate": 1.186692871408452e-05, "loss": 1.944756317138672, "step": 204500 }, { "epoch": 0.8137933092305167, "grad_norm": 0.5678713917732239, "learning_rate": 1.186493948773443e-05, "loss": 1.9592678833007813, "step": 204550 }, { "epoch": 0.8139922320633768, "grad_norm": 0.6021910309791565, "learning_rate": 1.1862950261384343e-05, "loss": 1.9096923828125, "step": 204600 }, { "epoch": 0.8141911548962368, "grad_norm": 0.516226589679718, "learning_rate": 1.1860961035034256e-05, "loss": 1.9142036437988281, "step": 204650 }, { "epoch": 0.814390077729097, "grad_norm": 0.5937727093696594, "learning_rate": 1.1858971808684168e-05, "loss": 1.9488975524902343, "step": 204700 }, { "epoch": 0.814589000561957, "grad_norm": 0.6735427975654602, "learning_rate": 1.185698258233408e-05, "loss": 1.9009121704101561, "step": 204750 }, { "epoch": 0.8147879233948171, "grad_norm": 0.644920825958252, "learning_rate": 1.1854993355983992e-05, "loss": 1.8844627380371093, "step": 204800 }, { "epoch": 0.8149868462276771, "grad_norm": 0.5184299349784851, "learning_rate": 1.1853004129633902e-05, "loss": 2.000657958984375, "step": 204850 }, { "epoch": 0.8151857690605372, "grad_norm": 0.6461679935455322, "learning_rate": 1.1851014903283815e-05, "loss": 1.9222718811035155, "step": 204900 }, { "epoch": 0.8153846918933972, "grad_norm": 0.5876919031143188, "learning_rate": 1.1849025676933727e-05, "loss": 1.929976043701172, "step": 204950 }, { "epoch": 0.8155836147262573, "grad_norm": 0.6697443127632141, "learning_rate": 1.184703645058364e-05, "loss": 1.9602760314941405, "step": 205000 }, { "epoch": 0.8157825375591173, "grad_norm": 0.5060864686965942, "learning_rate": 1.1845047224233553e-05, "loss": 1.934090118408203, "step": 205050 }, { "epoch": 0.8159814603919774, "grad_norm": 0.4711911082267761, "learning_rate": 1.1843057997883465e-05, "loss": 1.9327578735351563, "step": 205100 }, { "epoch": 0.8161803832248375, "grad_norm": 0.514394223690033, "learning_rate": 1.1841068771533377e-05, "loss": 1.9621234130859375, "step": 205150 }, { "epoch": 0.8163793060576976, "grad_norm": 0.4528726637363434, "learning_rate": 1.1839079545183287e-05, "loss": 1.9762413024902343, "step": 205200 }, { "epoch": 0.8165782288905576, "grad_norm": 0.5297009944915771, "learning_rate": 1.18370903188332e-05, "loss": 1.9536723327636718, "step": 205250 }, { "epoch": 0.8167771517234177, "grad_norm": 0.5327929258346558, "learning_rate": 1.1835101092483112e-05, "loss": 1.9043463134765626, "step": 205300 }, { "epoch": 0.8169760745562777, "grad_norm": 0.6070038676261902, "learning_rate": 1.1833111866133024e-05, "loss": 1.9294908142089844, "step": 205350 }, { "epoch": 0.8171749973891378, "grad_norm": 0.6602621674537659, "learning_rate": 1.1831122639782938e-05, "loss": 1.929132080078125, "step": 205400 }, { "epoch": 0.8173739202219978, "grad_norm": 0.46977105736732483, "learning_rate": 1.182913341343285e-05, "loss": 1.9414102172851562, "step": 205450 }, { "epoch": 0.8175728430548579, "grad_norm": 0.5864723920822144, "learning_rate": 1.182714418708276e-05, "loss": 1.9217626953125, "step": 205500 }, { "epoch": 0.817771765887718, "grad_norm": 0.6229313015937805, "learning_rate": 1.1825154960732672e-05, "loss": 1.9186408996582032, "step": 205550 }, { "epoch": 0.8179706887205781, "grad_norm": 0.5908211469650269, "learning_rate": 1.1823165734382585e-05, "loss": 1.9281588745117189, "step": 205600 }, { "epoch": 0.8181696115534381, "grad_norm": 0.591331958770752, "learning_rate": 1.1821176508032497e-05, "loss": 1.8984895324707032, "step": 205650 }, { "epoch": 0.8183685343862982, "grad_norm": 0.5680346488952637, "learning_rate": 1.1819187281682409e-05, "loss": 1.9394061279296875, "step": 205700 }, { "epoch": 0.8185674572191582, "grad_norm": 0.6184287071228027, "learning_rate": 1.1817198055332321e-05, "loss": 1.971583251953125, "step": 205750 }, { "epoch": 0.8187663800520183, "grad_norm": 0.7268353700637817, "learning_rate": 1.1815208828982235e-05, "loss": 1.9909523010253907, "step": 205800 }, { "epoch": 0.8189653028848783, "grad_norm": 0.6831215023994446, "learning_rate": 1.1813219602632144e-05, "loss": 1.9908415222167968, "step": 205850 }, { "epoch": 0.8191642257177384, "grad_norm": 0.6043176651000977, "learning_rate": 1.1811230376282057e-05, "loss": 1.9414056396484376, "step": 205900 }, { "epoch": 0.8193631485505986, "grad_norm": 0.570694625377655, "learning_rate": 1.180924114993197e-05, "loss": 1.9644305419921875, "step": 205950 }, { "epoch": 0.8195620713834586, "grad_norm": 0.6286222338676453, "learning_rate": 1.1807251923581882e-05, "loss": 1.937244873046875, "step": 206000 }, { "epoch": 0.8197609942163187, "grad_norm": 0.5913300514221191, "learning_rate": 1.1805262697231794e-05, "loss": 1.9197712707519532, "step": 206050 }, { "epoch": 0.8199599170491787, "grad_norm": 0.6675813794136047, "learning_rate": 1.1803273470881706e-05, "loss": 1.9451231384277343, "step": 206100 }, { "epoch": 0.8201588398820387, "grad_norm": 0.557918906211853, "learning_rate": 1.1801284244531617e-05, "loss": 1.9686985778808594, "step": 206150 }, { "epoch": 0.8203577627148988, "grad_norm": 0.6666540503501892, "learning_rate": 1.1799295018181529e-05, "loss": 1.9689979553222656, "step": 206200 }, { "epoch": 0.8205566855477588, "grad_norm": 0.6398258805274963, "learning_rate": 1.179730579183144e-05, "loss": 1.9119720458984375, "step": 206250 }, { "epoch": 0.8207556083806189, "grad_norm": 0.5879294872283936, "learning_rate": 1.1795316565481355e-05, "loss": 1.9669918823242187, "step": 206300 }, { "epoch": 0.820954531213479, "grad_norm": 0.671687126159668, "learning_rate": 1.1793327339131267e-05, "loss": 1.9144615173339843, "step": 206350 }, { "epoch": 0.8211534540463391, "grad_norm": 0.6216904520988464, "learning_rate": 1.1791338112781179e-05, "loss": 1.8894331359863281, "step": 206400 }, { "epoch": 0.8213523768791992, "grad_norm": 0.6221476793289185, "learning_rate": 1.1789348886431091e-05, "loss": 1.957665252685547, "step": 206450 }, { "epoch": 0.8215512997120592, "grad_norm": 0.5956272482872009, "learning_rate": 1.1787359660081002e-05, "loss": 1.9262820434570314, "step": 206500 }, { "epoch": 0.8217502225449193, "grad_norm": 0.6108434200286865, "learning_rate": 1.1785370433730914e-05, "loss": 1.9149017333984375, "step": 206550 }, { "epoch": 0.8219491453777793, "grad_norm": 0.5517480969429016, "learning_rate": 1.1783381207380826e-05, "loss": 1.935614013671875, "step": 206600 }, { "epoch": 0.8221480682106393, "grad_norm": 0.6292403936386108, "learning_rate": 1.1781391981030738e-05, "loss": 1.9096968078613281, "step": 206650 }, { "epoch": 0.8223469910434994, "grad_norm": 0.6802545785903931, "learning_rate": 1.1779402754680652e-05, "loss": 1.9490821838378907, "step": 206700 }, { "epoch": 0.8225459138763596, "grad_norm": 0.575722336769104, "learning_rate": 1.1777413528330564e-05, "loss": 1.94589599609375, "step": 206750 }, { "epoch": 0.8227448367092196, "grad_norm": 0.5980589985847473, "learning_rate": 1.1775424301980474e-05, "loss": 1.9637127685546876, "step": 206800 }, { "epoch": 0.8229437595420797, "grad_norm": 0.6817991137504578, "learning_rate": 1.1773435075630387e-05, "loss": 1.9150698852539063, "step": 206850 }, { "epoch": 0.8231426823749397, "grad_norm": 0.544969916343689, "learning_rate": 1.1771445849280299e-05, "loss": 1.9145596313476563, "step": 206900 }, { "epoch": 0.8233416052077998, "grad_norm": 0.567372739315033, "learning_rate": 1.1769456622930211e-05, "loss": 1.904954376220703, "step": 206950 }, { "epoch": 0.8235405280406598, "grad_norm": 0.5836435556411743, "learning_rate": 1.1767467396580123e-05, "loss": 1.9204917907714845, "step": 207000 }, { "epoch": 0.8237394508735199, "grad_norm": 0.42430543899536133, "learning_rate": 1.1765478170230035e-05, "loss": 1.9240060424804688, "step": 207050 }, { "epoch": 0.8239383737063799, "grad_norm": 0.6444391012191772, "learning_rate": 1.1763488943879949e-05, "loss": 1.8923988342285156, "step": 207100 }, { "epoch": 0.8241372965392401, "grad_norm": 0.6949103474617004, "learning_rate": 1.1761499717529858e-05, "loss": 1.9219076538085937, "step": 207150 }, { "epoch": 0.8243362193721001, "grad_norm": 0.6886915564537048, "learning_rate": 1.1759510491179772e-05, "loss": 1.923365936279297, "step": 207200 }, { "epoch": 0.8245351422049602, "grad_norm": 0.5968340039253235, "learning_rate": 1.1757521264829684e-05, "loss": 1.9662918090820312, "step": 207250 }, { "epoch": 0.8247340650378202, "grad_norm": 0.5925848484039307, "learning_rate": 1.1755532038479596e-05, "loss": 1.980953369140625, "step": 207300 }, { "epoch": 0.8249329878706803, "grad_norm": 0.6016888618469238, "learning_rate": 1.1753542812129508e-05, "loss": 1.9100657653808595, "step": 207350 }, { "epoch": 0.8251319107035403, "grad_norm": 0.5548856258392334, "learning_rate": 1.175155358577942e-05, "loss": 1.9257012939453124, "step": 207400 }, { "epoch": 0.8253308335364004, "grad_norm": 0.7126498222351074, "learning_rate": 1.174956435942933e-05, "loss": 1.9097593688964845, "step": 207450 }, { "epoch": 0.8255297563692604, "grad_norm": 0.5808058381080627, "learning_rate": 1.1747575133079243e-05, "loss": 1.934638671875, "step": 207500 }, { "epoch": 0.8257286792021206, "grad_norm": 0.4960010349750519, "learning_rate": 1.1745585906729155e-05, "loss": 1.9234754943847656, "step": 207550 }, { "epoch": 0.8259276020349806, "grad_norm": 0.6449662446975708, "learning_rate": 1.1743596680379069e-05, "loss": 1.8940724182128905, "step": 207600 }, { "epoch": 0.8261265248678407, "grad_norm": 0.48623237013816833, "learning_rate": 1.1741607454028981e-05, "loss": 1.9639317321777343, "step": 207650 }, { "epoch": 0.8263254477007007, "grad_norm": 0.6512259244918823, "learning_rate": 1.1739618227678893e-05, "loss": 1.9387054443359375, "step": 207700 }, { "epoch": 0.8265243705335608, "grad_norm": 0.5819956064224243, "learning_rate": 1.1737629001328805e-05, "loss": 1.928591766357422, "step": 207750 }, { "epoch": 0.8267232933664208, "grad_norm": 0.4529256522655487, "learning_rate": 1.1735639774978716e-05, "loss": 1.972596435546875, "step": 207800 }, { "epoch": 0.8269222161992809, "grad_norm": 0.7142247557640076, "learning_rate": 1.1733650548628628e-05, "loss": 1.9314170837402345, "step": 207850 }, { "epoch": 0.8271211390321409, "grad_norm": 0.6209463477134705, "learning_rate": 1.173166132227854e-05, "loss": 1.9917280578613281, "step": 207900 }, { "epoch": 0.8273200618650011, "grad_norm": 0.6392563581466675, "learning_rate": 1.1729672095928452e-05, "loss": 1.9234043884277343, "step": 207950 }, { "epoch": 0.8275189846978611, "grad_norm": 0.6705324649810791, "learning_rate": 1.1727682869578366e-05, "loss": 1.9054949951171876, "step": 208000 }, { "epoch": 0.8277179075307212, "grad_norm": 0.6455082893371582, "learning_rate": 1.1725693643228278e-05, "loss": 1.924134521484375, "step": 208050 }, { "epoch": 0.8279168303635812, "grad_norm": 0.5574994683265686, "learning_rate": 1.1723704416878189e-05, "loss": 1.9594349670410156, "step": 208100 }, { "epoch": 0.8281157531964413, "grad_norm": 0.7356171011924744, "learning_rate": 1.17217151905281e-05, "loss": 1.902176971435547, "step": 208150 }, { "epoch": 0.8283146760293013, "grad_norm": 0.6891601085662842, "learning_rate": 1.1719725964178013e-05, "loss": 1.895074462890625, "step": 208200 }, { "epoch": 0.8285135988621614, "grad_norm": 0.6259900331497192, "learning_rate": 1.1717736737827925e-05, "loss": 1.922183074951172, "step": 208250 }, { "epoch": 0.8287125216950214, "grad_norm": 0.5519786477088928, "learning_rate": 1.1715747511477837e-05, "loss": 1.957453155517578, "step": 208300 }, { "epoch": 0.8289114445278816, "grad_norm": 0.49793893098831177, "learning_rate": 1.171375828512775e-05, "loss": 1.9783203125, "step": 208350 }, { "epoch": 0.8291103673607416, "grad_norm": 0.656904935836792, "learning_rate": 1.1711769058777661e-05, "loss": 1.965161590576172, "step": 208400 }, { "epoch": 0.8293092901936017, "grad_norm": 0.5914871692657471, "learning_rate": 1.1709779832427572e-05, "loss": 1.9249737548828125, "step": 208450 }, { "epoch": 0.8295082130264617, "grad_norm": 0.5772649049758911, "learning_rate": 1.1707790606077486e-05, "loss": 1.9677003479003907, "step": 208500 }, { "epoch": 0.8297071358593218, "grad_norm": 0.557574450969696, "learning_rate": 1.1705801379727398e-05, "loss": 1.952630615234375, "step": 208550 }, { "epoch": 0.8299060586921818, "grad_norm": 0.5043084621429443, "learning_rate": 1.170381215337731e-05, "loss": 1.9177597045898438, "step": 208600 }, { "epoch": 0.8301049815250419, "grad_norm": 0.6076444983482361, "learning_rate": 1.1701822927027222e-05, "loss": 1.9241766357421874, "step": 208650 }, { "epoch": 0.8303039043579019, "grad_norm": 0.6763280034065247, "learning_rate": 1.1699833700677134e-05, "loss": 1.883985595703125, "step": 208700 }, { "epoch": 0.8305028271907621, "grad_norm": 0.621986448764801, "learning_rate": 1.1697844474327045e-05, "loss": 1.9486863708496094, "step": 208750 }, { "epoch": 0.8307017500236221, "grad_norm": 0.5700458884239197, "learning_rate": 1.1695855247976957e-05, "loss": 1.9304861450195312, "step": 208800 }, { "epoch": 0.8309006728564822, "grad_norm": 0.5738588571548462, "learning_rate": 1.1693866021626869e-05, "loss": 1.9280113220214843, "step": 208850 }, { "epoch": 0.8310995956893422, "grad_norm": 0.6514074206352234, "learning_rate": 1.1691916579803784e-05, "loss": 1.9379647827148438, "step": 208900 }, { "epoch": 0.8312985185222023, "grad_norm": 0.7108434438705444, "learning_rate": 1.1689927353453696e-05, "loss": 1.9241740417480468, "step": 208950 }, { "epoch": 0.8314974413550623, "grad_norm": 0.6936744451522827, "learning_rate": 1.1687938127103609e-05, "loss": 1.94306396484375, "step": 209000 }, { "epoch": 0.8316963641879224, "grad_norm": 0.6919686794281006, "learning_rate": 1.168594890075352e-05, "loss": 1.964879913330078, "step": 209050 }, { "epoch": 0.8318952870207824, "grad_norm": 0.5050113797187805, "learning_rate": 1.1683959674403431e-05, "loss": 1.962517547607422, "step": 209100 }, { "epoch": 0.8320942098536426, "grad_norm": 0.8232921361923218, "learning_rate": 1.1681970448053343e-05, "loss": 1.9237339782714844, "step": 209150 }, { "epoch": 0.8322931326865026, "grad_norm": 0.5891081690788269, "learning_rate": 1.1679981221703255e-05, "loss": 1.9412220764160155, "step": 209200 }, { "epoch": 0.8324920555193627, "grad_norm": 0.5505880117416382, "learning_rate": 1.167799199535317e-05, "loss": 1.959088592529297, "step": 209250 }, { "epoch": 0.8326909783522227, "grad_norm": 0.7531107664108276, "learning_rate": 1.1676002769003081e-05, "loss": 1.9609640502929688, "step": 209300 }, { "epoch": 0.8328899011850828, "grad_norm": 0.5565246343612671, "learning_rate": 1.1674013542652994e-05, "loss": 1.9298582458496094, "step": 209350 }, { "epoch": 0.8330888240179428, "grad_norm": 0.2807236611843109, "learning_rate": 1.1672024316302904e-05, "loss": 1.9083894348144532, "step": 209400 }, { "epoch": 0.8332877468508029, "grad_norm": 0.5097759366035461, "learning_rate": 1.1670035089952816e-05, "loss": 1.9377508544921875, "step": 209450 }, { "epoch": 0.8334866696836629, "grad_norm": 0.650393009185791, "learning_rate": 1.1668045863602728e-05, "loss": 1.9558175659179688, "step": 209500 }, { "epoch": 0.8336855925165231, "grad_norm": 0.6020397543907166, "learning_rate": 1.166605663725264e-05, "loss": 1.891405792236328, "step": 209550 }, { "epoch": 0.8338845153493831, "grad_norm": 0.7075285315513611, "learning_rate": 1.1664067410902553e-05, "loss": 1.9389588928222656, "step": 209600 }, { "epoch": 0.8340834381822432, "grad_norm": 0.5990845561027527, "learning_rate": 1.1662078184552465e-05, "loss": 1.9328645324707032, "step": 209650 }, { "epoch": 0.8342823610151032, "grad_norm": 0.5895177721977234, "learning_rate": 1.1660088958202379e-05, "loss": 1.926114044189453, "step": 209700 }, { "epoch": 0.8344812838479633, "grad_norm": 0.6469086408615112, "learning_rate": 1.1658099731852289e-05, "loss": 1.9559515380859376, "step": 209750 }, { "epoch": 0.8346802066808233, "grad_norm": 0.5920442938804626, "learning_rate": 1.1656110505502201e-05, "loss": 1.9122323608398437, "step": 209800 }, { "epoch": 0.8348791295136834, "grad_norm": 0.6410161852836609, "learning_rate": 1.1654121279152113e-05, "loss": 1.9342259216308593, "step": 209850 }, { "epoch": 0.8350780523465434, "grad_norm": 0.5085039138793945, "learning_rate": 1.1652132052802025e-05, "loss": 1.970059814453125, "step": 209900 }, { "epoch": 0.8352769751794036, "grad_norm": 0.5838212370872498, "learning_rate": 1.1650142826451938e-05, "loss": 1.9399833679199219, "step": 209950 }, { "epoch": 0.8354758980122636, "grad_norm": 0.5611534714698792, "learning_rate": 1.164815360010185e-05, "loss": 1.8976557922363282, "step": 210000 }, { "epoch": 0.8356748208451237, "grad_norm": 0.6090266108512878, "learning_rate": 1.164616437375176e-05, "loss": 1.9164849853515624, "step": 210050 }, { "epoch": 0.8358737436779837, "grad_norm": 0.6296951770782471, "learning_rate": 1.1644175147401672e-05, "loss": 1.9262872314453126, "step": 210100 }, { "epoch": 0.8360726665108438, "grad_norm": 0.568274974822998, "learning_rate": 1.1642185921051584e-05, "loss": 1.9822140502929688, "step": 210150 }, { "epoch": 0.8362715893437038, "grad_norm": 0.6262255907058716, "learning_rate": 1.1640196694701498e-05, "loss": 1.9597926330566406, "step": 210200 }, { "epoch": 0.8364705121765639, "grad_norm": 0.5515707731246948, "learning_rate": 1.163820746835141e-05, "loss": 1.9165638732910155, "step": 210250 }, { "epoch": 0.8366694350094239, "grad_norm": 0.6906166672706604, "learning_rate": 1.1636218242001323e-05, "loss": 2.024930877685547, "step": 210300 }, { "epoch": 0.8368683578422841, "grad_norm": 0.6283827424049377, "learning_rate": 1.1634229015651235e-05, "loss": 1.8858364868164061, "step": 210350 }, { "epoch": 0.8370672806751441, "grad_norm": 0.7495071887969971, "learning_rate": 1.1632239789301145e-05, "loss": 1.9189154052734374, "step": 210400 }, { "epoch": 0.8372662035080042, "grad_norm": 0.6132140755653381, "learning_rate": 1.1630250562951057e-05, "loss": 1.8919903564453124, "step": 210450 }, { "epoch": 0.8374651263408642, "grad_norm": 0.6176371574401855, "learning_rate": 1.162826133660097e-05, "loss": 1.9134951782226564, "step": 210500 }, { "epoch": 0.8376640491737243, "grad_norm": 0.683551549911499, "learning_rate": 1.1626272110250882e-05, "loss": 1.9371099853515625, "step": 210550 }, { "epoch": 0.8378629720065843, "grad_norm": 0.7370846271514893, "learning_rate": 1.1624282883900795e-05, "loss": 1.9859391784667968, "step": 210600 }, { "epoch": 0.8380618948394444, "grad_norm": 0.7459537386894226, "learning_rate": 1.1622293657550708e-05, "loss": 1.9699516296386719, "step": 210650 }, { "epoch": 0.8382608176723044, "grad_norm": 0.5770606994628906, "learning_rate": 1.1620304431200618e-05, "loss": 1.9818150329589843, "step": 210700 }, { "epoch": 0.8384597405051646, "grad_norm": 0.5001510977745056, "learning_rate": 1.161831520485053e-05, "loss": 1.964301300048828, "step": 210750 }, { "epoch": 0.8386586633380246, "grad_norm": 0.669557511806488, "learning_rate": 1.1616325978500442e-05, "loss": 1.961199951171875, "step": 210800 }, { "epoch": 0.8388575861708847, "grad_norm": 0.6228396892547607, "learning_rate": 1.1614336752150355e-05, "loss": 1.8926579284667968, "step": 210850 }, { "epoch": 0.8390565090037447, "grad_norm": 0.5950161814689636, "learning_rate": 1.1612347525800267e-05, "loss": 1.8931910705566406, "step": 210900 }, { "epoch": 0.8392554318366048, "grad_norm": 0.6197401881217957, "learning_rate": 1.1610358299450179e-05, "loss": 1.8973829650878906, "step": 210950 }, { "epoch": 0.8394543546694648, "grad_norm": 0.6542137861251831, "learning_rate": 1.160836907310009e-05, "loss": 1.924210662841797, "step": 211000 }, { "epoch": 0.8396532775023249, "grad_norm": 0.6311703324317932, "learning_rate": 1.1606379846750001e-05, "loss": 1.907010040283203, "step": 211050 }, { "epoch": 0.8398522003351849, "grad_norm": 0.558673620223999, "learning_rate": 1.1604390620399915e-05, "loss": 1.9347369384765625, "step": 211100 }, { "epoch": 0.8400511231680451, "grad_norm": 0.5698249936103821, "learning_rate": 1.1602401394049827e-05, "loss": 1.8982814025878907, "step": 211150 }, { "epoch": 0.8402500460009051, "grad_norm": 0.6414156556129456, "learning_rate": 1.160041216769974e-05, "loss": 1.8896379089355468, "step": 211200 }, { "epoch": 0.8404489688337652, "grad_norm": 0.6301663517951965, "learning_rate": 1.1598422941349652e-05, "loss": 1.9243630981445312, "step": 211250 }, { "epoch": 0.8406478916666252, "grad_norm": 0.6065858602523804, "learning_rate": 1.1596433714999564e-05, "loss": 1.9530815124511718, "step": 211300 }, { "epoch": 0.8408468144994853, "grad_norm": 0.5268362164497375, "learning_rate": 1.1594444488649474e-05, "loss": 1.9198362731933594, "step": 211350 }, { "epoch": 0.8410457373323453, "grad_norm": 0.7179563641548157, "learning_rate": 1.1592455262299386e-05, "loss": 1.90609375, "step": 211400 }, { "epoch": 0.8412446601652054, "grad_norm": 0.6546129584312439, "learning_rate": 1.1590466035949299e-05, "loss": 1.979363250732422, "step": 211450 }, { "epoch": 0.8414435829980654, "grad_norm": 0.6355382204055786, "learning_rate": 1.1588476809599212e-05, "loss": 1.9518145751953124, "step": 211500 }, { "epoch": 0.8416425058309256, "grad_norm": 0.6397769451141357, "learning_rate": 1.1586487583249125e-05, "loss": 1.9806494140625, "step": 211550 }, { "epoch": 0.8418414286637856, "grad_norm": 0.5840262174606323, "learning_rate": 1.1584498356899037e-05, "loss": 1.898048095703125, "step": 211600 }, { "epoch": 0.8420403514966457, "grad_norm": 0.5778433084487915, "learning_rate": 1.1582509130548947e-05, "loss": 1.9438729858398438, "step": 211650 }, { "epoch": 0.8422392743295057, "grad_norm": 0.5482605695724487, "learning_rate": 1.158051990419886e-05, "loss": 1.9661053466796874, "step": 211700 }, { "epoch": 0.8424381971623658, "grad_norm": 0.5958054661750793, "learning_rate": 1.1578530677848771e-05, "loss": 1.9139743041992188, "step": 211750 }, { "epoch": 0.8426371199952258, "grad_norm": 0.5691726803779602, "learning_rate": 1.1576541451498684e-05, "loss": 1.9380068969726563, "step": 211800 }, { "epoch": 0.8428360428280859, "grad_norm": 0.5956195592880249, "learning_rate": 1.1574552225148596e-05, "loss": 1.9679975891113282, "step": 211850 }, { "epoch": 0.8430349656609459, "grad_norm": 0.6445286870002747, "learning_rate": 1.157256299879851e-05, "loss": 1.9717092895507813, "step": 211900 }, { "epoch": 0.8432338884938061, "grad_norm": 0.6742223501205444, "learning_rate": 1.1570573772448422e-05, "loss": 2.0083876037597657, "step": 211950 }, { "epoch": 0.8434328113266661, "grad_norm": 0.5866472125053406, "learning_rate": 1.1568584546098332e-05, "loss": 1.9353785705566406, "step": 212000 }, { "epoch": 0.8436317341595262, "grad_norm": 0.5757023096084595, "learning_rate": 1.1566595319748244e-05, "loss": 1.9332723999023438, "step": 212050 }, { "epoch": 0.8438306569923862, "grad_norm": 0.6257780194282532, "learning_rate": 1.1564606093398156e-05, "loss": 1.9447152709960938, "step": 212100 }, { "epoch": 0.8440295798252463, "grad_norm": 0.6069192290306091, "learning_rate": 1.1562616867048069e-05, "loss": 2.0005229187011717, "step": 212150 }, { "epoch": 0.8442285026581063, "grad_norm": 0.505377471446991, "learning_rate": 1.156062764069798e-05, "loss": 1.902984619140625, "step": 212200 }, { "epoch": 0.8444274254909664, "grad_norm": 0.6585236191749573, "learning_rate": 1.1558638414347893e-05, "loss": 1.9078497314453124, "step": 212250 }, { "epoch": 0.8446263483238264, "grad_norm": 0.5781847834587097, "learning_rate": 1.1556649187997803e-05, "loss": 1.9331019592285157, "step": 212300 }, { "epoch": 0.8448252711566866, "grad_norm": 0.6455986499786377, "learning_rate": 1.1554659961647716e-05, "loss": 1.9312408447265625, "step": 212350 }, { "epoch": 0.8450241939895466, "grad_norm": 0.5845618844032288, "learning_rate": 1.155267073529763e-05, "loss": 1.9313485717773438, "step": 212400 }, { "epoch": 0.8452231168224067, "grad_norm": 0.5063814520835876, "learning_rate": 1.1550681508947542e-05, "loss": 1.9384896850585938, "step": 212450 }, { "epoch": 0.8454220396552667, "grad_norm": 0.6063165068626404, "learning_rate": 1.1548692282597454e-05, "loss": 1.923958740234375, "step": 212500 }, { "epoch": 0.8456209624881268, "grad_norm": 0.6225978136062622, "learning_rate": 1.1546742840774367e-05, "loss": 1.9620504760742188, "step": 212550 }, { "epoch": 0.8458198853209868, "grad_norm": 0.7327156662940979, "learning_rate": 1.154475361442428e-05, "loss": 1.9390629577636718, "step": 212600 }, { "epoch": 0.8460188081538469, "grad_norm": 0.6606374979019165, "learning_rate": 1.154276438807419e-05, "loss": 1.868778076171875, "step": 212650 }, { "epoch": 0.8462177309867069, "grad_norm": 0.7270863056182861, "learning_rate": 1.1540775161724102e-05, "loss": 1.9223854064941406, "step": 212700 }, { "epoch": 0.8464166538195671, "grad_norm": 0.5616810321807861, "learning_rate": 1.1538785935374016e-05, "loss": 1.9317889404296875, "step": 212750 }, { "epoch": 0.8466155766524271, "grad_norm": 0.6163923740386963, "learning_rate": 1.1536796709023928e-05, "loss": 1.9714573669433593, "step": 212800 }, { "epoch": 0.8468144994852872, "grad_norm": 0.5913294553756714, "learning_rate": 1.153480748267384e-05, "loss": 1.9104539489746093, "step": 212850 }, { "epoch": 0.8470134223181472, "grad_norm": 0.7181819081306458, "learning_rate": 1.1532818256323752e-05, "loss": 1.93475830078125, "step": 212900 }, { "epoch": 0.8472123451510073, "grad_norm": 0.5646179914474487, "learning_rate": 1.1530829029973663e-05, "loss": 1.9246844482421874, "step": 212950 }, { "epoch": 0.8474112679838673, "grad_norm": 0.7234300971031189, "learning_rate": 1.1528839803623575e-05, "loss": 1.9343130493164062, "step": 213000 }, { "epoch": 0.8476101908167274, "grad_norm": 0.5829530954360962, "learning_rate": 1.1526850577273487e-05, "loss": 1.9393263244628907, "step": 213050 }, { "epoch": 0.8478091136495874, "grad_norm": 0.7502500414848328, "learning_rate": 1.1524861350923399e-05, "loss": 1.9735955810546875, "step": 213100 }, { "epoch": 0.8480080364824475, "grad_norm": 0.5712578892707825, "learning_rate": 1.1522872124573313e-05, "loss": 1.9799261474609375, "step": 213150 }, { "epoch": 0.8482069593153077, "grad_norm": 0.5155539512634277, "learning_rate": 1.1520882898223225e-05, "loss": 1.9213578796386719, "step": 213200 }, { "epoch": 0.8484058821481677, "grad_norm": 0.6431692838668823, "learning_rate": 1.1518893671873137e-05, "loss": 1.9703546142578126, "step": 213250 }, { "epoch": 0.8486048049810277, "grad_norm": 0.5749709010124207, "learning_rate": 1.1516904445523048e-05, "loss": 1.945083465576172, "step": 213300 }, { "epoch": 0.8488037278138878, "grad_norm": 0.5568137764930725, "learning_rate": 1.151491521917296e-05, "loss": 1.8711387634277343, "step": 213350 }, { "epoch": 0.8490026506467478, "grad_norm": 0.5460834503173828, "learning_rate": 1.1512925992822872e-05, "loss": 1.9648956298828124, "step": 213400 }, { "epoch": 0.8492015734796079, "grad_norm": 0.6014933586120605, "learning_rate": 1.1510936766472784e-05, "loss": 1.930263671875, "step": 213450 }, { "epoch": 0.849400496312468, "grad_norm": 0.6106795072555542, "learning_rate": 1.1508947540122696e-05, "loss": 1.9750901794433593, "step": 213500 }, { "epoch": 0.849599419145328, "grad_norm": 0.6001163125038147, "learning_rate": 1.150695831377261e-05, "loss": 1.9469068908691407, "step": 213550 }, { "epoch": 0.8497983419781882, "grad_norm": 0.5736371874809265, "learning_rate": 1.1504969087422519e-05, "loss": 1.967566680908203, "step": 213600 }, { "epoch": 0.8499972648110482, "grad_norm": 0.7644039392471313, "learning_rate": 1.1502979861072433e-05, "loss": 1.9208114624023438, "step": 213650 }, { "epoch": 0.8501961876439083, "grad_norm": 0.619987428188324, "learning_rate": 1.1500990634722345e-05, "loss": 1.9274147033691407, "step": 213700 }, { "epoch": 0.8503951104767683, "grad_norm": 0.5981545448303223, "learning_rate": 1.1499001408372257e-05, "loss": 1.9569071960449218, "step": 213750 }, { "epoch": 0.8505940333096284, "grad_norm": 0.6500292420387268, "learning_rate": 1.1497012182022169e-05, "loss": 1.9070004272460936, "step": 213800 }, { "epoch": 0.8507929561424884, "grad_norm": 0.5305215716362, "learning_rate": 1.1495022955672081e-05, "loss": 1.9861996459960938, "step": 213850 }, { "epoch": 0.8509918789753484, "grad_norm": 0.5630095601081848, "learning_rate": 1.1493033729321993e-05, "loss": 1.9780197143554688, "step": 213900 }, { "epoch": 0.8511908018082085, "grad_norm": 0.7283774614334106, "learning_rate": 1.1491044502971904e-05, "loss": 1.8871011352539062, "step": 213950 }, { "epoch": 0.8513897246410687, "grad_norm": 0.685675859451294, "learning_rate": 1.1489055276621816e-05, "loss": 1.9667735290527344, "step": 214000 }, { "epoch": 0.8515886474739287, "grad_norm": 0.5748606324195862, "learning_rate": 1.148706605027173e-05, "loss": 1.9238798522949219, "step": 214050 }, { "epoch": 0.8517875703067888, "grad_norm": 0.6377625465393066, "learning_rate": 1.1485076823921642e-05, "loss": 1.9134541320800782, "step": 214100 }, { "epoch": 0.8519864931396488, "grad_norm": 0.5557973384857178, "learning_rate": 1.1483087597571554e-05, "loss": 1.951644744873047, "step": 214150 }, { "epoch": 0.8521854159725089, "grad_norm": 0.527421772480011, "learning_rate": 1.1481098371221466e-05, "loss": 1.9227384948730468, "step": 214200 }, { "epoch": 0.8523843388053689, "grad_norm": 0.7462273240089417, "learning_rate": 1.1479109144871377e-05, "loss": 1.9467478942871095, "step": 214250 }, { "epoch": 0.852583261638229, "grad_norm": 0.6306065917015076, "learning_rate": 1.1477119918521289e-05, "loss": 1.8771975708007813, "step": 214300 }, { "epoch": 0.852782184471089, "grad_norm": 0.7268533110618591, "learning_rate": 1.1475130692171201e-05, "loss": 1.9232687377929687, "step": 214350 }, { "epoch": 0.8529811073039492, "grad_norm": 0.6400559544563293, "learning_rate": 1.1473141465821113e-05, "loss": 1.96115234375, "step": 214400 }, { "epoch": 0.8531800301368092, "grad_norm": 0.613132894039154, "learning_rate": 1.1471152239471027e-05, "loss": 1.8926913452148437, "step": 214450 }, { "epoch": 0.8533789529696693, "grad_norm": 0.6917703747749329, "learning_rate": 1.146916301312094e-05, "loss": 1.9429953002929687, "step": 214500 }, { "epoch": 0.8535778758025293, "grad_norm": 0.5257333517074585, "learning_rate": 1.1467173786770851e-05, "loss": 1.9823600769042968, "step": 214550 }, { "epoch": 0.8537767986353894, "grad_norm": 0.46394631266593933, "learning_rate": 1.1465184560420762e-05, "loss": 1.9437443542480468, "step": 214600 }, { "epoch": 0.8539757214682494, "grad_norm": 0.6373779773712158, "learning_rate": 1.1463195334070674e-05, "loss": 1.9858999633789063, "step": 214650 }, { "epoch": 0.8541746443011095, "grad_norm": 0.6195667386054993, "learning_rate": 1.1461206107720586e-05, "loss": 1.8856198120117187, "step": 214700 }, { "epoch": 0.8543735671339695, "grad_norm": 0.5735679864883423, "learning_rate": 1.1459216881370498e-05, "loss": 1.9428950500488282, "step": 214750 }, { "epoch": 0.8545724899668297, "grad_norm": 0.5869314074516296, "learning_rate": 1.145722765502041e-05, "loss": 1.9116716003417968, "step": 214800 }, { "epoch": 0.8547714127996897, "grad_norm": 0.7007542848587036, "learning_rate": 1.1455238428670322e-05, "loss": 1.9172283935546874, "step": 214850 }, { "epoch": 0.8549703356325498, "grad_norm": 0.7021847367286682, "learning_rate": 1.1453249202320233e-05, "loss": 1.9460203552246094, "step": 214900 }, { "epoch": 0.8551692584654098, "grad_norm": 0.6074678301811218, "learning_rate": 1.1451259975970147e-05, "loss": 1.9760806274414062, "step": 214950 }, { "epoch": 0.8553681812982699, "grad_norm": 0.5945807695388794, "learning_rate": 1.1449270749620059e-05, "loss": 1.8825433349609375, "step": 215000 }, { "epoch": 0.8555671041311299, "grad_norm": 0.7189461588859558, "learning_rate": 1.1447281523269971e-05, "loss": 1.9425294494628906, "step": 215050 }, { "epoch": 0.85576602696399, "grad_norm": 0.5463271141052246, "learning_rate": 1.1445292296919883e-05, "loss": 1.9475885009765626, "step": 215100 }, { "epoch": 0.85596494979685, "grad_norm": 0.5798172354698181, "learning_rate": 1.1443303070569795e-05, "loss": 1.9652301025390626, "step": 215150 }, { "epoch": 0.8561638726297102, "grad_norm": 0.6104555726051331, "learning_rate": 1.144135362874671e-05, "loss": 1.9578791809082032, "step": 215200 }, { "epoch": 0.8563627954625702, "grad_norm": 0.5822669863700867, "learning_rate": 1.143936440239662e-05, "loss": 1.9273841857910157, "step": 215250 }, { "epoch": 0.8565617182954303, "grad_norm": 0.599808394908905, "learning_rate": 1.1437375176046533e-05, "loss": 1.96667236328125, "step": 215300 }, { "epoch": 0.8567606411282903, "grad_norm": 0.7003628015518188, "learning_rate": 1.1435385949696445e-05, "loss": 1.8956475830078126, "step": 215350 }, { "epoch": 0.8569595639611504, "grad_norm": 0.5344306826591492, "learning_rate": 1.1433396723346357e-05, "loss": 1.9236648559570313, "step": 215400 }, { "epoch": 0.8571584867940104, "grad_norm": 0.567785382270813, "learning_rate": 1.143140749699627e-05, "loss": 1.9893817138671874, "step": 215450 }, { "epoch": 0.8573574096268705, "grad_norm": 0.5731047987937927, "learning_rate": 1.1429418270646182e-05, "loss": 1.9708712768554688, "step": 215500 }, { "epoch": 0.8575563324597305, "grad_norm": 0.6477253437042236, "learning_rate": 1.1427429044296092e-05, "loss": 1.9472151184082032, "step": 215550 }, { "epoch": 0.8577552552925907, "grad_norm": 0.624497652053833, "learning_rate": 1.1425439817946004e-05, "loss": 1.8871937561035157, "step": 215600 }, { "epoch": 0.8579541781254507, "grad_norm": 0.6622951030731201, "learning_rate": 1.1423450591595916e-05, "loss": 1.9310238647460938, "step": 215650 }, { "epoch": 0.8581531009583108, "grad_norm": 0.6454523801803589, "learning_rate": 1.142146136524583e-05, "loss": 1.9012623596191407, "step": 215700 }, { "epoch": 0.8583520237911708, "grad_norm": 0.6692964434623718, "learning_rate": 1.1419472138895742e-05, "loss": 1.9180978393554688, "step": 215750 }, { "epoch": 0.8585509466240309, "grad_norm": 0.6301824450492859, "learning_rate": 1.1417482912545655e-05, "loss": 1.943038787841797, "step": 215800 }, { "epoch": 0.8587498694568909, "grad_norm": 0.6905278563499451, "learning_rate": 1.1415493686195567e-05, "loss": 1.9181369018554688, "step": 215850 }, { "epoch": 0.858948792289751, "grad_norm": 0.5362820625305176, "learning_rate": 1.1413504459845477e-05, "loss": 1.8985392761230468, "step": 215900 }, { "epoch": 0.859147715122611, "grad_norm": 0.626020610332489, "learning_rate": 1.141151523349539e-05, "loss": 1.9681942749023438, "step": 215950 }, { "epoch": 0.8593466379554712, "grad_norm": 0.6250451803207397, "learning_rate": 1.1409526007145302e-05, "loss": 1.8958180236816407, "step": 216000 }, { "epoch": 0.8595455607883312, "grad_norm": 0.5527949333190918, "learning_rate": 1.1407536780795214e-05, "loss": 1.9573249816894531, "step": 216050 }, { "epoch": 0.8597444836211913, "grad_norm": 0.7547492980957031, "learning_rate": 1.1405547554445126e-05, "loss": 1.9547657775878906, "step": 216100 }, { "epoch": 0.8599434064540513, "grad_norm": 0.5741900205612183, "learning_rate": 1.140355832809504e-05, "loss": 1.885413818359375, "step": 216150 }, { "epoch": 0.8601423292869114, "grad_norm": 0.6019918918609619, "learning_rate": 1.140156910174495e-05, "loss": 1.9766813659667968, "step": 216200 }, { "epoch": 0.8603412521197714, "grad_norm": 0.7054733633995056, "learning_rate": 1.1399579875394862e-05, "loss": 1.9521377563476563, "step": 216250 }, { "epoch": 0.8605401749526315, "grad_norm": 0.7433952689170837, "learning_rate": 1.1397590649044774e-05, "loss": 1.9354963684082032, "step": 216300 }, { "epoch": 0.8607390977854915, "grad_norm": 0.6620587706565857, "learning_rate": 1.1395601422694687e-05, "loss": 1.9145053100585938, "step": 216350 }, { "epoch": 0.8609380206183517, "grad_norm": 0.5488362908363342, "learning_rate": 1.1393612196344599e-05, "loss": 1.9275808715820313, "step": 216400 }, { "epoch": 0.8611369434512117, "grad_norm": 0.5598395466804504, "learning_rate": 1.139162296999451e-05, "loss": 1.9295442199707031, "step": 216450 }, { "epoch": 0.8613358662840718, "grad_norm": 0.6341652274131775, "learning_rate": 1.1389633743644423e-05, "loss": 1.9661074829101564, "step": 216500 }, { "epoch": 0.8615347891169318, "grad_norm": 0.6037054657936096, "learning_rate": 1.1387644517294333e-05, "loss": 1.9084565734863281, "step": 216550 }, { "epoch": 0.8617337119497919, "grad_norm": 0.6471906900405884, "learning_rate": 1.1385655290944246e-05, "loss": 1.9280072021484376, "step": 216600 }, { "epoch": 0.8619326347826519, "grad_norm": 0.5411344766616821, "learning_rate": 1.138366606459416e-05, "loss": 1.9688650512695312, "step": 216650 }, { "epoch": 0.862131557615512, "grad_norm": 0.6156086921691895, "learning_rate": 1.1381676838244072e-05, "loss": 1.8703094482421876, "step": 216700 }, { "epoch": 0.862330480448372, "grad_norm": 0.6087055802345276, "learning_rate": 1.1379687611893984e-05, "loss": 1.9522201538085937, "step": 216750 }, { "epoch": 0.8625294032812322, "grad_norm": 0.6411495804786682, "learning_rate": 1.1377698385543896e-05, "loss": 1.8557887268066406, "step": 216800 }, { "epoch": 0.8627283261140922, "grad_norm": 0.5591506361961365, "learning_rate": 1.1375709159193806e-05, "loss": 1.9526930236816407, "step": 216850 }, { "epoch": 0.8629272489469523, "grad_norm": 0.6146740913391113, "learning_rate": 1.1373719932843718e-05, "loss": 1.9055413818359375, "step": 216900 }, { "epoch": 0.8631261717798123, "grad_norm": 0.577820360660553, "learning_rate": 1.137173070649363e-05, "loss": 1.9216885375976562, "step": 216950 }, { "epoch": 0.8633250946126724, "grad_norm": 0.5950506925582886, "learning_rate": 1.1369741480143543e-05, "loss": 1.95046630859375, "step": 217000 }, { "epoch": 0.8635240174455324, "grad_norm": 0.5630297064781189, "learning_rate": 1.1367752253793457e-05, "loss": 1.9040882873535157, "step": 217050 }, { "epoch": 0.8637229402783925, "grad_norm": 0.6169308423995972, "learning_rate": 1.1365763027443369e-05, "loss": 1.9761778259277343, "step": 217100 }, { "epoch": 0.8639218631112525, "grad_norm": 0.6300691366195679, "learning_rate": 1.1363773801093281e-05, "loss": 1.962735137939453, "step": 217150 }, { "epoch": 0.8641207859441127, "grad_norm": 0.5249264240264893, "learning_rate": 1.1361784574743191e-05, "loss": 1.9199826049804687, "step": 217200 }, { "epoch": 0.8643197087769727, "grad_norm": 0.5879119634628296, "learning_rate": 1.1359795348393103e-05, "loss": 1.940924072265625, "step": 217250 }, { "epoch": 0.8645186316098328, "grad_norm": 0.5811590552330017, "learning_rate": 1.1357806122043016e-05, "loss": 1.9732730102539062, "step": 217300 }, { "epoch": 0.8647175544426928, "grad_norm": 0.6407005786895752, "learning_rate": 1.1355816895692928e-05, "loss": 1.928155517578125, "step": 217350 }, { "epoch": 0.8649164772755529, "grad_norm": 0.6850847601890564, "learning_rate": 1.135382766934284e-05, "loss": 1.9366145324707031, "step": 217400 }, { "epoch": 0.8651154001084129, "grad_norm": 0.643543004989624, "learning_rate": 1.1351838442992754e-05, "loss": 1.9303726196289062, "step": 217450 }, { "epoch": 0.865314322941273, "grad_norm": 0.5941541194915771, "learning_rate": 1.1349849216642662e-05, "loss": 1.8959526062011718, "step": 217500 }, { "epoch": 0.865513245774133, "grad_norm": 0.7034996747970581, "learning_rate": 1.1347859990292576e-05, "loss": 1.9225515747070312, "step": 217550 }, { "epoch": 0.8657121686069932, "grad_norm": 0.5801730751991272, "learning_rate": 1.1345870763942488e-05, "loss": 1.905291290283203, "step": 217600 }, { "epoch": 0.8659110914398532, "grad_norm": 0.6274155378341675, "learning_rate": 1.13438815375924e-05, "loss": 1.9270387268066407, "step": 217650 }, { "epoch": 0.8661100142727133, "grad_norm": 0.6516984701156616, "learning_rate": 1.1341892311242313e-05, "loss": 1.9215530395507812, "step": 217700 }, { "epoch": 0.8663089371055733, "grad_norm": 0.5740735530853271, "learning_rate": 1.1339903084892225e-05, "loss": 1.8933514404296874, "step": 217750 }, { "epoch": 0.8665078599384334, "grad_norm": 0.6877278089523315, "learning_rate": 1.1337913858542137e-05, "loss": 1.9406141662597656, "step": 217800 }, { "epoch": 0.8667067827712934, "grad_norm": 0.6477134227752686, "learning_rate": 1.1335924632192048e-05, "loss": 1.8857249450683593, "step": 217850 }, { "epoch": 0.8669057056041535, "grad_norm": 0.5977353453636169, "learning_rate": 1.133393540584196e-05, "loss": 1.9888002014160155, "step": 217900 }, { "epoch": 0.8671046284370135, "grad_norm": 0.5931159257888794, "learning_rate": 1.1331946179491874e-05, "loss": 1.9402664184570313, "step": 217950 }, { "epoch": 0.8673035512698737, "grad_norm": 0.6582433581352234, "learning_rate": 1.1329956953141786e-05, "loss": 1.8994943237304687, "step": 218000 }, { "epoch": 0.8675024741027337, "grad_norm": 0.6327993273735046, "learning_rate": 1.1327967726791698e-05, "loss": 1.9207461547851563, "step": 218050 }, { "epoch": 0.8677013969355938, "grad_norm": 0.7025743722915649, "learning_rate": 1.132597850044161e-05, "loss": 1.9109939575195312, "step": 218100 }, { "epoch": 0.8679003197684538, "grad_norm": 0.6352485418319702, "learning_rate": 1.132398927409152e-05, "loss": 1.8941560363769532, "step": 218150 }, { "epoch": 0.8680992426013139, "grad_norm": 0.5531399846076965, "learning_rate": 1.1322039832268434e-05, "loss": 2.0233819580078123, "step": 218200 }, { "epoch": 0.8682981654341739, "grad_norm": 0.5171266794204712, "learning_rate": 1.1320050605918346e-05, "loss": 1.92316162109375, "step": 218250 }, { "epoch": 0.868497088267034, "grad_norm": 0.5491836071014404, "learning_rate": 1.131806137956826e-05, "loss": 1.9108509826660156, "step": 218300 }, { "epoch": 0.868696011099894, "grad_norm": 0.6068410277366638, "learning_rate": 1.1316072153218172e-05, "loss": 1.936142578125, "step": 218350 }, { "epoch": 0.8688949339327542, "grad_norm": 0.6879481077194214, "learning_rate": 1.1314082926868084e-05, "loss": 1.962052764892578, "step": 218400 }, { "epoch": 0.8690938567656142, "grad_norm": 0.6055728197097778, "learning_rate": 1.1312093700517996e-05, "loss": 1.9360531616210936, "step": 218450 }, { "epoch": 0.8692927795984743, "grad_norm": 0.6953355669975281, "learning_rate": 1.1310104474167907e-05, "loss": 1.8746357727050782, "step": 218500 }, { "epoch": 0.8694917024313343, "grad_norm": 0.6900410056114197, "learning_rate": 1.1308115247817819e-05, "loss": 1.9048068237304687, "step": 218550 }, { "epoch": 0.8696906252641944, "grad_norm": 0.6547292470932007, "learning_rate": 1.1306126021467731e-05, "loss": 1.9592405700683593, "step": 218600 }, { "epoch": 0.8698895480970544, "grad_norm": 0.5156456828117371, "learning_rate": 1.1304136795117643e-05, "loss": 1.955734405517578, "step": 218650 }, { "epoch": 0.8700884709299145, "grad_norm": 0.49590760469436646, "learning_rate": 1.1302147568767557e-05, "loss": 1.9200552368164063, "step": 218700 }, { "epoch": 0.8702873937627745, "grad_norm": 0.7173185348510742, "learning_rate": 1.130015834241747e-05, "loss": 1.9767991638183593, "step": 218750 }, { "epoch": 0.8704863165956347, "grad_norm": 0.5751887559890747, "learning_rate": 1.129816911606738e-05, "loss": 1.9464707946777344, "step": 218800 }, { "epoch": 0.8706852394284947, "grad_norm": 0.6083251237869263, "learning_rate": 1.1296179889717292e-05, "loss": 1.9450634765625, "step": 218850 }, { "epoch": 0.8708841622613548, "grad_norm": 0.6123972535133362, "learning_rate": 1.1294190663367204e-05, "loss": 1.9199693298339844, "step": 218900 }, { "epoch": 0.8710830850942148, "grad_norm": 0.6090331673622131, "learning_rate": 1.1292201437017116e-05, "loss": 1.923795166015625, "step": 218950 }, { "epoch": 0.8712820079270749, "grad_norm": 0.5434225797653198, "learning_rate": 1.1290212210667028e-05, "loss": 1.8900889587402343, "step": 219000 }, { "epoch": 0.8714809307599349, "grad_norm": 0.6677340269088745, "learning_rate": 1.128822298431694e-05, "loss": 1.959548797607422, "step": 219050 }, { "epoch": 0.871679853592795, "grad_norm": 0.6385114789009094, "learning_rate": 1.1286233757966854e-05, "loss": 1.9449916076660156, "step": 219100 }, { "epoch": 0.871878776425655, "grad_norm": 0.6640167832374573, "learning_rate": 1.1284244531616763e-05, "loss": 1.89880615234375, "step": 219150 }, { "epoch": 0.8720776992585152, "grad_norm": 0.6221321821212769, "learning_rate": 1.1282255305266677e-05, "loss": 1.954414825439453, "step": 219200 }, { "epoch": 0.8722766220913752, "grad_norm": 0.6528215408325195, "learning_rate": 1.1280266078916589e-05, "loss": 1.8970356750488282, "step": 219250 }, { "epoch": 0.8724755449242353, "grad_norm": 0.6248766779899597, "learning_rate": 1.1278276852566501e-05, "loss": 1.9516249084472657, "step": 219300 }, { "epoch": 0.8726744677570953, "grad_norm": 0.6270657181739807, "learning_rate": 1.1276287626216413e-05, "loss": 1.9413319396972657, "step": 219350 }, { "epoch": 0.8728733905899554, "grad_norm": 0.5819460153579712, "learning_rate": 1.1274298399866325e-05, "loss": 1.9063313293457032, "step": 219400 }, { "epoch": 0.8730723134228154, "grad_norm": 0.5915355682373047, "learning_rate": 1.1272309173516236e-05, "loss": 1.921127471923828, "step": 219450 }, { "epoch": 0.8732712362556755, "grad_norm": 0.6398605108261108, "learning_rate": 1.1270319947166148e-05, "loss": 1.9020269775390626, "step": 219500 }, { "epoch": 0.8734701590885355, "grad_norm": 0.5635426640510559, "learning_rate": 1.126833072081606e-05, "loss": 1.908949432373047, "step": 219550 }, { "epoch": 0.8736690819213957, "grad_norm": 0.6318166255950928, "learning_rate": 1.1266341494465974e-05, "loss": 1.917608184814453, "step": 219600 }, { "epoch": 0.8738680047542557, "grad_norm": 0.5126222968101501, "learning_rate": 1.1264352268115886e-05, "loss": 1.9806678771972657, "step": 219650 }, { "epoch": 0.8740669275871158, "grad_norm": 0.5380896329879761, "learning_rate": 1.1262363041765798e-05, "loss": 1.8911984252929688, "step": 219700 }, { "epoch": 0.8742658504199758, "grad_norm": 0.5762720108032227, "learning_rate": 1.126037381541571e-05, "loss": 1.9079759216308594, "step": 219750 }, { "epoch": 0.8744647732528359, "grad_norm": 0.6123332381248474, "learning_rate": 1.1258384589065621e-05, "loss": 1.929702911376953, "step": 219800 }, { "epoch": 0.8746636960856959, "grad_norm": 0.636237621307373, "learning_rate": 1.1256395362715533e-05, "loss": 1.942957305908203, "step": 219850 }, { "epoch": 0.874862618918556, "grad_norm": 0.7017678618431091, "learning_rate": 1.1254406136365445e-05, "loss": 1.9581777954101562, "step": 219900 }, { "epoch": 0.875061541751416, "grad_norm": 0.6752484440803528, "learning_rate": 1.1252416910015357e-05, "loss": 1.9308639526367188, "step": 219950 }, { "epoch": 0.8752604645842762, "grad_norm": 0.5783019661903381, "learning_rate": 1.1250427683665271e-05, "loss": 1.9820010375976562, "step": 220000 }, { "epoch": 0.8754593874171362, "grad_norm": 0.6243264675140381, "learning_rate": 1.1248438457315183e-05, "loss": 1.9499790954589844, "step": 220050 }, { "epoch": 0.8756583102499963, "grad_norm": 0.6657593846321106, "learning_rate": 1.1246449230965094e-05, "loss": 1.9345014953613282, "step": 220100 }, { "epoch": 0.8758572330828563, "grad_norm": 0.6391299366950989, "learning_rate": 1.1244460004615006e-05, "loss": 1.8937440490722657, "step": 220150 }, { "epoch": 0.8760561559157164, "grad_norm": 0.5431379079818726, "learning_rate": 1.1242470778264918e-05, "loss": 1.9378057861328124, "step": 220200 }, { "epoch": 0.8762550787485764, "grad_norm": 0.610934317111969, "learning_rate": 1.124048155191483e-05, "loss": 1.909503173828125, "step": 220250 }, { "epoch": 0.8764540015814365, "grad_norm": 0.5545766949653625, "learning_rate": 1.1238492325564742e-05, "loss": 1.9193946838378906, "step": 220300 }, { "epoch": 0.8766529244142965, "grad_norm": 0.6160212159156799, "learning_rate": 1.1236503099214654e-05, "loss": 1.9768515014648438, "step": 220350 }, { "epoch": 0.8768518472471567, "grad_norm": 0.5757206678390503, "learning_rate": 1.1234513872864568e-05, "loss": 1.9549777221679687, "step": 220400 }, { "epoch": 0.8770507700800168, "grad_norm": 0.5173248052597046, "learning_rate": 1.1232524646514477e-05, "loss": 1.9042588806152343, "step": 220450 }, { "epoch": 0.8772496929128768, "grad_norm": 0.6685081124305725, "learning_rate": 1.1230535420164391e-05, "loss": 1.9411927795410155, "step": 220500 }, { "epoch": 0.8774486157457368, "grad_norm": 0.5660359263420105, "learning_rate": 1.1228546193814303e-05, "loss": 1.9326475524902345, "step": 220550 }, { "epoch": 0.8776475385785969, "grad_norm": 0.6231837272644043, "learning_rate": 1.1226556967464215e-05, "loss": 1.901799774169922, "step": 220600 }, { "epoch": 0.877846461411457, "grad_norm": 0.6873891949653625, "learning_rate": 1.1224567741114127e-05, "loss": 1.8944703674316405, "step": 220650 }, { "epoch": 0.878045384244317, "grad_norm": 0.5787459015846252, "learning_rate": 1.122257851476404e-05, "loss": 1.9330412292480468, "step": 220700 }, { "epoch": 0.878244307077177, "grad_norm": 0.7884812355041504, "learning_rate": 1.1220629072940951e-05, "loss": 1.9617262268066407, "step": 220750 }, { "epoch": 0.8784432299100371, "grad_norm": 0.6117864847183228, "learning_rate": 1.1218639846590863e-05, "loss": 1.895228271484375, "step": 220800 }, { "epoch": 0.8786421527428973, "grad_norm": 0.6124635934829712, "learning_rate": 1.1216650620240777e-05, "loss": 1.9378880310058593, "step": 220850 }, { "epoch": 0.8788410755757573, "grad_norm": 0.7376399636268616, "learning_rate": 1.121466139389069e-05, "loss": 1.8450863647460938, "step": 220900 }, { "epoch": 0.8790399984086174, "grad_norm": 0.5877828001976013, "learning_rate": 1.1212672167540602e-05, "loss": 1.8632431030273438, "step": 220950 }, { "epoch": 0.8792389212414774, "grad_norm": 0.5736125707626343, "learning_rate": 1.1210682941190514e-05, "loss": 1.9166824340820312, "step": 221000 }, { "epoch": 0.8794378440743374, "grad_norm": 0.5664158463478088, "learning_rate": 1.1208693714840426e-05, "loss": 1.981499481201172, "step": 221050 }, { "epoch": 0.8796367669071975, "grad_norm": 0.6189256310462952, "learning_rate": 1.1206704488490336e-05, "loss": 1.9516265869140625, "step": 221100 }, { "epoch": 0.8798356897400575, "grad_norm": 0.5831999778747559, "learning_rate": 1.1204715262140248e-05, "loss": 1.9551504516601563, "step": 221150 }, { "epoch": 0.8800346125729176, "grad_norm": 0.6613122224807739, "learning_rate": 1.120272603579016e-05, "loss": 1.9393507385253905, "step": 221200 }, { "epoch": 0.8802335354057778, "grad_norm": 0.665408194065094, "learning_rate": 1.1200736809440074e-05, "loss": 1.877306671142578, "step": 221250 }, { "epoch": 0.8804324582386378, "grad_norm": 0.5903525352478027, "learning_rate": 1.1198747583089987e-05, "loss": 1.9361476135253906, "step": 221300 }, { "epoch": 0.8806313810714979, "grad_norm": 0.5276656746864319, "learning_rate": 1.1196758356739899e-05, "loss": 1.9540072631835939, "step": 221350 }, { "epoch": 0.8808303039043579, "grad_norm": 0.6070125699043274, "learning_rate": 1.119476913038981e-05, "loss": 1.919258575439453, "step": 221400 }, { "epoch": 0.881029226737218, "grad_norm": 0.6498346328735352, "learning_rate": 1.1192779904039721e-05, "loss": 1.960454559326172, "step": 221450 }, { "epoch": 0.881228149570078, "grad_norm": 0.5533048510551453, "learning_rate": 1.1190790677689634e-05, "loss": 1.934718475341797, "step": 221500 }, { "epoch": 0.881427072402938, "grad_norm": 0.5712659955024719, "learning_rate": 1.1188801451339546e-05, "loss": 1.9302668762207031, "step": 221550 }, { "epoch": 0.8816259952357981, "grad_norm": 0.7120301723480225, "learning_rate": 1.1186812224989458e-05, "loss": 1.9298854064941406, "step": 221600 }, { "epoch": 0.8818249180686583, "grad_norm": 0.6324278116226196, "learning_rate": 1.1184822998639372e-05, "loss": 1.9531689453125, "step": 221650 }, { "epoch": 0.8820238409015183, "grad_norm": 0.5759391784667969, "learning_rate": 1.1182833772289284e-05, "loss": 1.925886993408203, "step": 221700 }, { "epoch": 0.8822227637343784, "grad_norm": 0.5305859446525574, "learning_rate": 1.1180844545939194e-05, "loss": 1.9558599853515626, "step": 221750 }, { "epoch": 0.8824216865672384, "grad_norm": 0.6190885305404663, "learning_rate": 1.1178855319589106e-05, "loss": 1.927156982421875, "step": 221800 }, { "epoch": 0.8826206094000985, "grad_norm": 0.5777256488800049, "learning_rate": 1.1176866093239019e-05, "loss": 1.9000357055664063, "step": 221850 }, { "epoch": 0.8828195322329585, "grad_norm": 0.6907039880752563, "learning_rate": 1.117487686688893e-05, "loss": 1.9191108703613282, "step": 221900 }, { "epoch": 0.8830184550658186, "grad_norm": 0.7281268239021301, "learning_rate": 1.1172887640538843e-05, "loss": 1.8955012512207032, "step": 221950 }, { "epoch": 0.8832173778986786, "grad_norm": 0.6446425914764404, "learning_rate": 1.1170898414188755e-05, "loss": 1.9221229553222656, "step": 222000 }, { "epoch": 0.8834163007315388, "grad_norm": 0.6447655558586121, "learning_rate": 1.1168909187838665e-05, "loss": 1.9434263610839844, "step": 222050 }, { "epoch": 0.8836152235643988, "grad_norm": 0.6043435335159302, "learning_rate": 1.1166919961488578e-05, "loss": 1.8976824951171876, "step": 222100 }, { "epoch": 0.8838141463972589, "grad_norm": 0.604036808013916, "learning_rate": 1.1164930735138491e-05, "loss": 1.9196354675292968, "step": 222150 }, { "epoch": 0.8840130692301189, "grad_norm": 0.5723575353622437, "learning_rate": 1.1162941508788404e-05, "loss": 1.9659048461914062, "step": 222200 }, { "epoch": 0.884211992062979, "grad_norm": 0.7525479793548584, "learning_rate": 1.1160952282438316e-05, "loss": 1.9164225769042968, "step": 222250 }, { "epoch": 0.884410914895839, "grad_norm": 0.6905035376548767, "learning_rate": 1.1158963056088228e-05, "loss": 1.9787330627441406, "step": 222300 }, { "epoch": 0.884609837728699, "grad_norm": 0.7173888087272644, "learning_rate": 1.115697382973814e-05, "loss": 1.9021894836425781, "step": 222350 }, { "epoch": 0.8848087605615591, "grad_norm": 0.6712430119514465, "learning_rate": 1.115498460338805e-05, "loss": 1.9273959350585939, "step": 222400 }, { "epoch": 0.8850076833944193, "grad_norm": 0.7278544902801514, "learning_rate": 1.1152995377037963e-05, "loss": 1.961613311767578, "step": 222450 }, { "epoch": 0.8852066062272793, "grad_norm": 0.5615749955177307, "learning_rate": 1.1151006150687875e-05, "loss": 1.9044671630859376, "step": 222500 }, { "epoch": 0.8854055290601394, "grad_norm": 0.5849507451057434, "learning_rate": 1.1149016924337787e-05, "loss": 1.942889404296875, "step": 222550 }, { "epoch": 0.8856044518929994, "grad_norm": 0.6644327640533447, "learning_rate": 1.11470276979877e-05, "loss": 1.8979757690429688, "step": 222600 }, { "epoch": 0.8858033747258595, "grad_norm": 0.6475374102592468, "learning_rate": 1.1145038471637613e-05, "loss": 1.9421456909179688, "step": 222650 }, { "epoch": 0.8860022975587195, "grad_norm": 0.6409767866134644, "learning_rate": 1.1143049245287523e-05, "loss": 1.9160177612304687, "step": 222700 }, { "epoch": 0.8862012203915796, "grad_norm": 0.6156197190284729, "learning_rate": 1.1141060018937435e-05, "loss": 1.937078094482422, "step": 222750 }, { "epoch": 0.8864001432244396, "grad_norm": 0.399371862411499, "learning_rate": 1.1139070792587348e-05, "loss": 1.8742782592773437, "step": 222800 }, { "epoch": 0.8865990660572998, "grad_norm": 0.6382929086685181, "learning_rate": 1.113708156623726e-05, "loss": 1.9397962951660157, "step": 222850 }, { "epoch": 0.8867979888901598, "grad_norm": 0.5985254049301147, "learning_rate": 1.1135092339887172e-05, "loss": 1.9600015258789063, "step": 222900 }, { "epoch": 0.8869969117230199, "grad_norm": 0.5872058868408203, "learning_rate": 1.1133103113537084e-05, "loss": 1.89747314453125, "step": 222950 }, { "epoch": 0.8871958345558799, "grad_norm": 0.5111750960350037, "learning_rate": 1.1131113887186994e-05, "loss": 1.9104873657226562, "step": 223000 }, { "epoch": 0.88739475738874, "grad_norm": 0.6436880230903625, "learning_rate": 1.1129124660836907e-05, "loss": 1.9391154479980468, "step": 223050 }, { "epoch": 0.8875936802216, "grad_norm": 0.6035071015357971, "learning_rate": 1.112713543448682e-05, "loss": 1.9271095275878907, "step": 223100 }, { "epoch": 0.8877926030544601, "grad_norm": 0.5014072060585022, "learning_rate": 1.1125146208136733e-05, "loss": 1.92951171875, "step": 223150 }, { "epoch": 0.8879915258873201, "grad_norm": 0.6577699184417725, "learning_rate": 1.1123156981786645e-05, "loss": 1.9211735534667969, "step": 223200 }, { "epoch": 0.8881904487201803, "grad_norm": 0.6396734714508057, "learning_rate": 1.1121167755436557e-05, "loss": 1.9544715881347656, "step": 223250 }, { "epoch": 0.8883893715530403, "grad_norm": 0.5685601234436035, "learning_rate": 1.1119178529086469e-05, "loss": 1.9410409545898437, "step": 223300 }, { "epoch": 0.8885882943859004, "grad_norm": 0.6804109215736389, "learning_rate": 1.111718930273638e-05, "loss": 1.9523860168457032, "step": 223350 }, { "epoch": 0.8887872172187604, "grad_norm": 0.4458373486995697, "learning_rate": 1.1115200076386292e-05, "loss": 1.9433836364746093, "step": 223400 }, { "epoch": 0.8889861400516205, "grad_norm": 0.7652501463890076, "learning_rate": 1.1113210850036204e-05, "loss": 1.9778579711914062, "step": 223450 }, { "epoch": 0.8891850628844805, "grad_norm": 0.6295596957206726, "learning_rate": 1.1111221623686118e-05, "loss": 1.9462210083007812, "step": 223500 }, { "epoch": 0.8893839857173406, "grad_norm": 0.5542837977409363, "learning_rate": 1.110923239733603e-05, "loss": 1.9082203674316407, "step": 223550 }, { "epoch": 0.8895829085502006, "grad_norm": 0.6703467965126038, "learning_rate": 1.1107243170985942e-05, "loss": 1.9502638244628907, "step": 223600 }, { "epoch": 0.8897818313830608, "grad_norm": 0.6765260100364685, "learning_rate": 1.1105253944635852e-05, "loss": 1.93293212890625, "step": 223650 }, { "epoch": 0.8899807542159208, "grad_norm": 0.5537295341491699, "learning_rate": 1.1103264718285765e-05, "loss": 1.9178536987304688, "step": 223700 }, { "epoch": 0.8901796770487809, "grad_norm": 0.5549466013908386, "learning_rate": 1.1101275491935677e-05, "loss": 1.9439083862304687, "step": 223750 }, { "epoch": 0.8903785998816409, "grad_norm": 0.5732191205024719, "learning_rate": 1.1099286265585589e-05, "loss": 1.942816925048828, "step": 223800 }, { "epoch": 0.890577522714501, "grad_norm": 0.686173141002655, "learning_rate": 1.1097297039235501e-05, "loss": 1.8691085815429687, "step": 223850 }, { "epoch": 0.890776445547361, "grad_norm": 0.5485436916351318, "learning_rate": 1.1095307812885415e-05, "loss": 1.8950807189941405, "step": 223900 }, { "epoch": 0.8909753683802211, "grad_norm": 0.5186026096343994, "learning_rate": 1.1093318586535327e-05, "loss": 1.9179188537597656, "step": 223950 }, { "epoch": 0.8911742912130811, "grad_norm": 0.6279614567756653, "learning_rate": 1.1091329360185237e-05, "loss": 1.9364456176757812, "step": 224000 }, { "epoch": 0.8913732140459413, "grad_norm": 0.5964539051055908, "learning_rate": 1.108934013383515e-05, "loss": 1.938394775390625, "step": 224050 }, { "epoch": 0.8915721368788013, "grad_norm": 0.6837378740310669, "learning_rate": 1.1087350907485062e-05, "loss": 1.9413041687011718, "step": 224100 }, { "epoch": 0.8917710597116614, "grad_norm": 0.5875136256217957, "learning_rate": 1.1085361681134974e-05, "loss": 1.855977783203125, "step": 224150 }, { "epoch": 0.8919699825445214, "grad_norm": 0.5796180963516235, "learning_rate": 1.1083372454784886e-05, "loss": 1.9104042053222656, "step": 224200 }, { "epoch": 0.8921689053773815, "grad_norm": 0.6919858455657959, "learning_rate": 1.1081383228434798e-05, "loss": 1.9297837829589843, "step": 224250 }, { "epoch": 0.8923678282102415, "grad_norm": 0.6361483931541443, "learning_rate": 1.1079394002084709e-05, "loss": 1.9344776916503905, "step": 224300 }, { "epoch": 0.8925667510431016, "grad_norm": 0.6649602055549622, "learning_rate": 1.107740477573462e-05, "loss": 1.941664276123047, "step": 224350 }, { "epoch": 0.8927656738759616, "grad_norm": 0.5682276487350464, "learning_rate": 1.1075415549384535e-05, "loss": 1.9155776977539063, "step": 224400 }, { "epoch": 0.8929645967088218, "grad_norm": 0.6340628266334534, "learning_rate": 1.1073426323034447e-05, "loss": 1.94013916015625, "step": 224450 }, { "epoch": 0.8931635195416818, "grad_norm": 0.634718656539917, "learning_rate": 1.1071437096684359e-05, "loss": 1.8926068115234376, "step": 224500 }, { "epoch": 0.8933624423745419, "grad_norm": 0.5490519404411316, "learning_rate": 1.1069447870334271e-05, "loss": 1.9180816650390624, "step": 224550 }, { "epoch": 0.8935613652074019, "grad_norm": 0.6155017614364624, "learning_rate": 1.1067458643984183e-05, "loss": 1.8532546997070312, "step": 224600 }, { "epoch": 0.893760288040262, "grad_norm": 0.6475931406021118, "learning_rate": 1.1065469417634094e-05, "loss": 1.8862167358398438, "step": 224650 }, { "epoch": 0.893959210873122, "grad_norm": 0.7006517052650452, "learning_rate": 1.1063480191284006e-05, "loss": 1.9552494812011718, "step": 224700 }, { "epoch": 0.8941581337059821, "grad_norm": 0.5434738993644714, "learning_rate": 1.1061490964933918e-05, "loss": 1.8860700988769532, "step": 224750 }, { "epoch": 0.8943570565388421, "grad_norm": 0.5947401523590088, "learning_rate": 1.1059541523110833e-05, "loss": 1.8781314086914063, "step": 224800 }, { "epoch": 0.8945559793717023, "grad_norm": 0.6063098311424255, "learning_rate": 1.1057552296760745e-05, "loss": 1.9251873779296875, "step": 224850 }, { "epoch": 0.8947549022045623, "grad_norm": 0.5919012427330017, "learning_rate": 1.1055563070410657e-05, "loss": 1.9746186828613281, "step": 224900 }, { "epoch": 0.8949538250374224, "grad_norm": 0.6723925471305847, "learning_rate": 1.1053573844060568e-05, "loss": 1.8820173645019531, "step": 224950 }, { "epoch": 0.8951527478702824, "grad_norm": 0.6475527882575989, "learning_rate": 1.105158461771048e-05, "loss": 1.8939320373535156, "step": 225000 }, { "epoch": 0.8953516707031425, "grad_norm": 0.5807192921638489, "learning_rate": 1.1049595391360392e-05, "loss": 1.8909532165527343, "step": 225050 }, { "epoch": 0.8955505935360025, "grad_norm": 0.6642412543296814, "learning_rate": 1.1047606165010304e-05, "loss": 1.9394908142089844, "step": 225100 }, { "epoch": 0.8957495163688626, "grad_norm": 0.5121974945068359, "learning_rate": 1.1045616938660218e-05, "loss": 1.9029864501953124, "step": 225150 }, { "epoch": 0.8959484392017226, "grad_norm": 0.6928761005401611, "learning_rate": 1.104362771231013e-05, "loss": 1.9488180541992188, "step": 225200 }, { "epoch": 0.8961473620345828, "grad_norm": 0.4898155629634857, "learning_rate": 1.1041638485960042e-05, "loss": 1.9117129516601563, "step": 225250 }, { "epoch": 0.8963462848674428, "grad_norm": 0.5766781568527222, "learning_rate": 1.1039649259609953e-05, "loss": 1.9244635009765625, "step": 225300 }, { "epoch": 0.8965452077003029, "grad_norm": 0.6879872679710388, "learning_rate": 1.1037660033259865e-05, "loss": 1.9666525268554687, "step": 225350 }, { "epoch": 0.8967441305331629, "grad_norm": 0.66246098279953, "learning_rate": 1.1035670806909777e-05, "loss": 1.8841244506835937, "step": 225400 }, { "epoch": 0.896943053366023, "grad_norm": 0.48639997839927673, "learning_rate": 1.103368158055969e-05, "loss": 1.9807769775390625, "step": 225450 }, { "epoch": 0.897141976198883, "grad_norm": 0.579512357711792, "learning_rate": 1.1031692354209601e-05, "loss": 1.9568048095703126, "step": 225500 }, { "epoch": 0.8973408990317431, "grad_norm": 0.553118109703064, "learning_rate": 1.1029703127859515e-05, "loss": 1.9684303283691407, "step": 225550 }, { "epoch": 0.8975398218646031, "grad_norm": 0.6881111860275269, "learning_rate": 1.1027713901509424e-05, "loss": 1.8461769104003907, "step": 225600 }, { "epoch": 0.8977387446974633, "grad_norm": 0.6656067371368408, "learning_rate": 1.1025724675159338e-05, "loss": 1.9171856689453124, "step": 225650 }, { "epoch": 0.8979376675303233, "grad_norm": 0.6071397066116333, "learning_rate": 1.102373544880925e-05, "loss": 1.9435455322265625, "step": 225700 }, { "epoch": 0.8981365903631834, "grad_norm": 0.5213660597801208, "learning_rate": 1.1021746222459162e-05, "loss": 1.9143936157226562, "step": 225750 }, { "epoch": 0.8983355131960434, "grad_norm": 0.6415132284164429, "learning_rate": 1.1019756996109074e-05, "loss": 1.9058457946777343, "step": 225800 }, { "epoch": 0.8985344360289035, "grad_norm": 0.59126216173172, "learning_rate": 1.1017767769758986e-05, "loss": 1.9096009826660156, "step": 225850 }, { "epoch": 0.8987333588617635, "grad_norm": 0.5900043845176697, "learning_rate": 1.1015778543408899e-05, "loss": 1.9340980529785157, "step": 225900 }, { "epoch": 0.8989322816946236, "grad_norm": 0.5825771689414978, "learning_rate": 1.1013789317058809e-05, "loss": 1.9622889709472657, "step": 225950 }, { "epoch": 0.8991312045274836, "grad_norm": 0.6767287254333496, "learning_rate": 1.1011839875235724e-05, "loss": 1.8952239990234374, "step": 226000 }, { "epoch": 0.8993301273603438, "grad_norm": 0.5524104237556458, "learning_rate": 1.1009850648885636e-05, "loss": 1.8928086853027344, "step": 226050 }, { "epoch": 0.8995290501932038, "grad_norm": 0.6196005940437317, "learning_rate": 1.1007861422535549e-05, "loss": 1.911769256591797, "step": 226100 }, { "epoch": 0.8997279730260639, "grad_norm": 0.6324009299278259, "learning_rate": 1.100587219618546e-05, "loss": 1.9780096435546874, "step": 226150 }, { "epoch": 0.8999268958589239, "grad_norm": 0.5380609035491943, "learning_rate": 1.1003882969835373e-05, "loss": 1.8817185974121093, "step": 226200 }, { "epoch": 0.900125818691784, "grad_norm": 0.6311644911766052, "learning_rate": 1.1001893743485283e-05, "loss": 1.9468260192871094, "step": 226250 }, { "epoch": 0.900324741524644, "grad_norm": 0.5975486040115356, "learning_rate": 1.0999904517135195e-05, "loss": 1.9333229064941406, "step": 226300 }, { "epoch": 0.9005236643575041, "grad_norm": 0.6405425071716309, "learning_rate": 1.0997915290785108e-05, "loss": 1.8939724731445313, "step": 226350 }, { "epoch": 0.9007225871903641, "grad_norm": 0.5561459064483643, "learning_rate": 1.0995926064435021e-05, "loss": 1.9713471984863282, "step": 226400 }, { "epoch": 0.9009215100232243, "grad_norm": 0.6261287927627563, "learning_rate": 1.0993936838084934e-05, "loss": 1.9032380676269531, "step": 226450 }, { "epoch": 0.9011204328560843, "grad_norm": 0.5726646780967712, "learning_rate": 1.0991947611734846e-05, "loss": 1.9366361999511719, "step": 226500 }, { "epoch": 0.9013193556889444, "grad_norm": 0.6788557171821594, "learning_rate": 1.0989958385384758e-05, "loss": 1.9048040771484376, "step": 226550 }, { "epoch": 0.9015182785218044, "grad_norm": 0.6291944980621338, "learning_rate": 1.0987969159034668e-05, "loss": 1.8998806762695313, "step": 226600 }, { "epoch": 0.9017172013546645, "grad_norm": 0.5373190641403198, "learning_rate": 1.098597993268458e-05, "loss": 1.93375732421875, "step": 226650 }, { "epoch": 0.9019161241875245, "grad_norm": 0.550346851348877, "learning_rate": 1.0983990706334493e-05, "loss": 1.9146177673339844, "step": 226700 }, { "epoch": 0.9021150470203846, "grad_norm": 0.5904232859611511, "learning_rate": 1.0982001479984405e-05, "loss": 1.9162184143066405, "step": 226750 }, { "epoch": 0.9023139698532446, "grad_norm": 0.7205482125282288, "learning_rate": 1.0980012253634319e-05, "loss": 1.9009622192382813, "step": 226800 }, { "epoch": 0.9025128926861048, "grad_norm": 0.4989933669567108, "learning_rate": 1.097802302728423e-05, "loss": 1.8946682739257812, "step": 226850 }, { "epoch": 0.9027118155189648, "grad_norm": 0.5925897359848022, "learning_rate": 1.0976033800934141e-05, "loss": 1.9221925354003906, "step": 226900 }, { "epoch": 0.9029107383518249, "grad_norm": 0.6164125204086304, "learning_rate": 1.0974044574584053e-05, "loss": 1.939385223388672, "step": 226950 }, { "epoch": 0.9031096611846849, "grad_norm": 0.6408931016921997, "learning_rate": 1.0972055348233966e-05, "loss": 1.8958078002929688, "step": 227000 }, { "epoch": 0.903308584017545, "grad_norm": 0.6035169363021851, "learning_rate": 1.0970066121883878e-05, "loss": 1.9613101196289062, "step": 227050 }, { "epoch": 0.903507506850405, "grad_norm": 0.6093360185623169, "learning_rate": 1.096807689553379e-05, "loss": 1.8826109313964843, "step": 227100 }, { "epoch": 0.9037064296832651, "grad_norm": 0.5826935768127441, "learning_rate": 1.0966087669183702e-05, "loss": 1.9242561340332032, "step": 227150 }, { "epoch": 0.9039053525161251, "grad_norm": 0.5688390135765076, "learning_rate": 1.0964098442833616e-05, "loss": 1.8951885986328125, "step": 227200 }, { "epoch": 0.9041042753489853, "grad_norm": 0.7196418046951294, "learning_rate": 1.0962109216483525e-05, "loss": 1.94840576171875, "step": 227250 }, { "epoch": 0.9043031981818453, "grad_norm": 0.6550236344337463, "learning_rate": 1.0960119990133438e-05, "loss": 1.9039561462402343, "step": 227300 }, { "epoch": 0.9045021210147054, "grad_norm": 0.6127256155014038, "learning_rate": 1.095813076378335e-05, "loss": 1.905359344482422, "step": 227350 }, { "epoch": 0.9047010438475654, "grad_norm": 0.5871273875236511, "learning_rate": 1.0956141537433263e-05, "loss": 1.9415583801269531, "step": 227400 }, { "epoch": 0.9048999666804255, "grad_norm": 0.5856467485427856, "learning_rate": 1.0954152311083175e-05, "loss": 1.948673095703125, "step": 227450 }, { "epoch": 0.9050988895132855, "grad_norm": 0.6496508121490479, "learning_rate": 1.0952163084733087e-05, "loss": 1.9487629699707032, "step": 227500 }, { "epoch": 0.9052978123461456, "grad_norm": 0.5598734021186829, "learning_rate": 1.0950173858382997e-05, "loss": 1.91888671875, "step": 227550 }, { "epoch": 0.9054967351790056, "grad_norm": 0.6565378308296204, "learning_rate": 1.094818463203291e-05, "loss": 1.9227108764648437, "step": 227600 }, { "epoch": 0.9056956580118658, "grad_norm": 0.5218026041984558, "learning_rate": 1.0946195405682822e-05, "loss": 1.9259234619140626, "step": 227650 }, { "epoch": 0.9058945808447258, "grad_norm": 0.5599445104598999, "learning_rate": 1.0944206179332736e-05, "loss": 1.8878080749511719, "step": 227700 }, { "epoch": 0.9060935036775859, "grad_norm": 0.6236605048179626, "learning_rate": 1.0942216952982648e-05, "loss": 1.9341790771484375, "step": 227750 }, { "epoch": 0.906292426510446, "grad_norm": 0.6432846784591675, "learning_rate": 1.094022772663256e-05, "loss": 1.9218905639648438, "step": 227800 }, { "epoch": 0.906491349343306, "grad_norm": 0.6531063914299011, "learning_rate": 1.0938238500282472e-05, "loss": 1.9736004638671876, "step": 227850 }, { "epoch": 0.906690272176166, "grad_norm": 0.5483478307723999, "learning_rate": 1.0936249273932382e-05, "loss": 1.9402565002441405, "step": 227900 }, { "epoch": 0.9068891950090261, "grad_norm": 0.587835431098938, "learning_rate": 1.0934260047582295e-05, "loss": 1.9090899658203124, "step": 227950 }, { "epoch": 0.9070881178418861, "grad_norm": 0.6633743643760681, "learning_rate": 1.0932270821232207e-05, "loss": 1.9180288696289063, "step": 228000 }, { "epoch": 0.9072870406747463, "grad_norm": 0.7409719824790955, "learning_rate": 1.0930281594882119e-05, "loss": 1.890068817138672, "step": 228050 }, { "epoch": 0.9074859635076064, "grad_norm": 0.6247624754905701, "learning_rate": 1.0928292368532033e-05, "loss": 1.947421112060547, "step": 228100 }, { "epoch": 0.9076848863404664, "grad_norm": 0.5514896512031555, "learning_rate": 1.0926303142181945e-05, "loss": 1.9361227416992188, "step": 228150 }, { "epoch": 0.9078838091733265, "grad_norm": 0.48431795835494995, "learning_rate": 1.0924313915831855e-05, "loss": 1.9182011413574218, "step": 228200 }, { "epoch": 0.9080827320061865, "grad_norm": 0.7198428511619568, "learning_rate": 1.0922324689481767e-05, "loss": 1.9488133239746093, "step": 228250 }, { "epoch": 0.9082816548390465, "grad_norm": 0.6173345446586609, "learning_rate": 1.092033546313168e-05, "loss": 1.9069586181640625, "step": 228300 }, { "epoch": 0.9084805776719066, "grad_norm": 0.632891058921814, "learning_rate": 1.0918346236781592e-05, "loss": 1.9055538940429688, "step": 228350 }, { "epoch": 0.9086795005047666, "grad_norm": 0.6128008365631104, "learning_rate": 1.0916357010431504e-05, "loss": 1.9503704833984374, "step": 228400 }, { "epoch": 0.9088784233376268, "grad_norm": 0.5766360759735107, "learning_rate": 1.0914367784081416e-05, "loss": 1.969549560546875, "step": 228450 }, { "epoch": 0.9090773461704869, "grad_norm": 0.6820964217185974, "learning_rate": 1.0912378557731328e-05, "loss": 1.900177001953125, "step": 228500 }, { "epoch": 0.9092762690033469, "grad_norm": 0.6105976700782776, "learning_rate": 1.0910389331381239e-05, "loss": 1.9098358154296875, "step": 228550 }, { "epoch": 0.909475191836207, "grad_norm": 0.6829099655151367, "learning_rate": 1.0908400105031152e-05, "loss": 1.9380203247070313, "step": 228600 }, { "epoch": 0.909674114669067, "grad_norm": 0.6268090605735779, "learning_rate": 1.0906410878681065e-05, "loss": 1.934151153564453, "step": 228650 }, { "epoch": 0.909873037501927, "grad_norm": 0.6450990438461304, "learning_rate": 1.0904421652330977e-05, "loss": 1.9630435180664063, "step": 228700 }, { "epoch": 0.9100719603347871, "grad_norm": 0.6547089219093323, "learning_rate": 1.0902432425980889e-05, "loss": 1.9259080505371093, "step": 228750 }, { "epoch": 0.9102708831676471, "grad_norm": 0.6584132313728333, "learning_rate": 1.0900443199630801e-05, "loss": 1.9463133239746093, "step": 228800 }, { "epoch": 0.9104698060005072, "grad_norm": 0.451412171125412, "learning_rate": 1.0898453973280712e-05, "loss": 1.9051239013671875, "step": 228850 }, { "epoch": 0.9106687288333674, "grad_norm": 0.5851405262947083, "learning_rate": 1.0896464746930624e-05, "loss": 1.9403804016113282, "step": 228900 }, { "epoch": 0.9108676516662274, "grad_norm": 0.6276551485061646, "learning_rate": 1.0894475520580536e-05, "loss": 1.894445343017578, "step": 228950 }, { "epoch": 0.9110665744990875, "grad_norm": 0.659429132938385, "learning_rate": 1.0892486294230448e-05, "loss": 1.9344497680664063, "step": 229000 }, { "epoch": 0.9112654973319475, "grad_norm": 0.6274235844612122, "learning_rate": 1.0890497067880362e-05, "loss": 1.9407608032226562, "step": 229050 }, { "epoch": 0.9114644201648076, "grad_norm": 0.5747768878936768, "learning_rate": 1.0888507841530274e-05, "loss": 1.9118910217285157, "step": 229100 }, { "epoch": 0.9116633429976676, "grad_norm": 0.6739235520362854, "learning_rate": 1.0886558399707187e-05, "loss": 1.93840087890625, "step": 229150 }, { "epoch": 0.9118622658305277, "grad_norm": 0.5252765417098999, "learning_rate": 1.0884569173357098e-05, "loss": 1.914055633544922, "step": 229200 }, { "epoch": 0.9120611886633877, "grad_norm": 0.6535395383834839, "learning_rate": 1.088257994700701e-05, "loss": 1.9242620849609375, "step": 229250 }, { "epoch": 0.9122601114962479, "grad_norm": 0.7595146894454956, "learning_rate": 1.0880590720656922e-05, "loss": 1.8955218505859375, "step": 229300 }, { "epoch": 0.9124590343291079, "grad_norm": 0.6792664527893066, "learning_rate": 1.0878601494306836e-05, "loss": 1.9429049682617188, "step": 229350 }, { "epoch": 0.912657957161968, "grad_norm": 0.6074337363243103, "learning_rate": 1.0876612267956748e-05, "loss": 1.9324122619628907, "step": 229400 }, { "epoch": 0.912856879994828, "grad_norm": 0.3786737024784088, "learning_rate": 1.087462304160666e-05, "loss": 1.9461293029785156, "step": 229450 }, { "epoch": 0.9130558028276881, "grad_norm": 0.6527878046035767, "learning_rate": 1.087263381525657e-05, "loss": 1.955788116455078, "step": 229500 }, { "epoch": 0.9132547256605481, "grad_norm": 0.545608639717102, "learning_rate": 1.0870644588906483e-05, "loss": 1.9024819946289062, "step": 229550 }, { "epoch": 0.9134536484934082, "grad_norm": 0.5421317219734192, "learning_rate": 1.0868655362556395e-05, "loss": 1.9499827575683595, "step": 229600 }, { "epoch": 0.9136525713262682, "grad_norm": 0.6329551339149475, "learning_rate": 1.0866666136206307e-05, "loss": 1.913050079345703, "step": 229650 }, { "epoch": 0.9138514941591284, "grad_norm": 0.6586899757385254, "learning_rate": 1.086467690985622e-05, "loss": 1.9235958862304687, "step": 229700 }, { "epoch": 0.9140504169919884, "grad_norm": 0.5341224670410156, "learning_rate": 1.0862687683506132e-05, "loss": 1.8844110107421874, "step": 229750 }, { "epoch": 0.9142493398248485, "grad_norm": 0.597934901714325, "learning_rate": 1.0860698457156045e-05, "loss": 1.9887937927246093, "step": 229800 }, { "epoch": 0.9144482626577085, "grad_norm": 0.61006760597229, "learning_rate": 1.0858709230805956e-05, "loss": 1.9280287170410155, "step": 229850 }, { "epoch": 0.9146471854905686, "grad_norm": 0.5090768337249756, "learning_rate": 1.0856720004455868e-05, "loss": 1.8717605590820312, "step": 229900 }, { "epoch": 0.9148461083234286, "grad_norm": 0.6123177409172058, "learning_rate": 1.085473077810578e-05, "loss": 1.962834930419922, "step": 229950 }, { "epoch": 0.9150450311562887, "grad_norm": 0.6022371649742126, "learning_rate": 1.0852741551755692e-05, "loss": 1.8916763305664062, "step": 230000 }, { "epoch": 0.9152439539891487, "grad_norm": 0.6218931674957275, "learning_rate": 1.0850752325405604e-05, "loss": 1.9066773986816405, "step": 230050 }, { "epoch": 0.9154428768220089, "grad_norm": 0.734586238861084, "learning_rate": 1.0848763099055517e-05, "loss": 1.9156329345703125, "step": 230100 }, { "epoch": 0.9156417996548689, "grad_norm": 0.6416453719139099, "learning_rate": 1.0846773872705427e-05, "loss": 1.8960810852050782, "step": 230150 }, { "epoch": 0.915840722487729, "grad_norm": 0.5915756225585938, "learning_rate": 1.0844784646355339e-05, "loss": 1.953059539794922, "step": 230200 }, { "epoch": 0.916039645320589, "grad_norm": 0.6072124242782593, "learning_rate": 1.0842795420005251e-05, "loss": 1.8907638549804688, "step": 230250 }, { "epoch": 0.9162385681534491, "grad_norm": 0.6341316103935242, "learning_rate": 1.0840806193655165e-05, "loss": 1.8978326416015625, "step": 230300 }, { "epoch": 0.9164374909863091, "grad_norm": 0.46004995703697205, "learning_rate": 1.0838816967305077e-05, "loss": 1.9269921875, "step": 230350 }, { "epoch": 0.9166364138191692, "grad_norm": 0.5639070272445679, "learning_rate": 1.083682774095499e-05, "loss": 1.9351493835449218, "step": 230400 }, { "epoch": 0.9168353366520292, "grad_norm": 0.5075497627258301, "learning_rate": 1.0834838514604902e-05, "loss": 1.9216276550292968, "step": 230450 }, { "epoch": 0.9170342594848894, "grad_norm": 0.6178284287452698, "learning_rate": 1.0832849288254812e-05, "loss": 1.9456755065917968, "step": 230500 }, { "epoch": 0.9172331823177494, "grad_norm": 0.47985634207725525, "learning_rate": 1.0830860061904724e-05, "loss": 1.9322805786132813, "step": 230550 }, { "epoch": 0.9174321051506095, "grad_norm": 0.5993374586105347, "learning_rate": 1.0828870835554636e-05, "loss": 1.9527481079101563, "step": 230600 }, { "epoch": 0.9176310279834695, "grad_norm": 0.716899573802948, "learning_rate": 1.0826881609204548e-05, "loss": 1.8706086730957032, "step": 230650 }, { "epoch": 0.9178299508163296, "grad_norm": 0.6385080814361572, "learning_rate": 1.0824892382854462e-05, "loss": 1.91847900390625, "step": 230700 }, { "epoch": 0.9180288736491896, "grad_norm": 0.6812397837638855, "learning_rate": 1.0822903156504374e-05, "loss": 1.9360130310058594, "step": 230750 }, { "epoch": 0.9182277964820497, "grad_norm": 0.7055907249450684, "learning_rate": 1.0820913930154285e-05, "loss": 1.9307174682617188, "step": 230800 }, { "epoch": 0.9184267193149097, "grad_norm": 0.5744057893753052, "learning_rate": 1.0818924703804197e-05, "loss": 1.9099070739746093, "step": 230850 }, { "epoch": 0.9186256421477699, "grad_norm": 0.6451530456542969, "learning_rate": 1.081693547745411e-05, "loss": 1.9584945678710937, "step": 230900 }, { "epoch": 0.9188245649806299, "grad_norm": 0.6653836965560913, "learning_rate": 1.0814946251104021e-05, "loss": 1.9221078491210937, "step": 230950 }, { "epoch": 0.91902348781349, "grad_norm": 0.6452102661132812, "learning_rate": 1.0812957024753933e-05, "loss": 1.9130601501464843, "step": 231000 }, { "epoch": 0.91922241064635, "grad_norm": 0.6436131000518799, "learning_rate": 1.0810967798403846e-05, "loss": 1.946580352783203, "step": 231050 }, { "epoch": 0.9194213334792101, "grad_norm": 0.6149972677230835, "learning_rate": 1.080897857205376e-05, "loss": 1.9376229858398437, "step": 231100 }, { "epoch": 0.9196202563120701, "grad_norm": 0.8901051878929138, "learning_rate": 1.0807029130230671e-05, "loss": 1.8660087585449219, "step": 231150 }, { "epoch": 0.9198191791449302, "grad_norm": 0.5368551015853882, "learning_rate": 1.0805039903880583e-05, "loss": 1.9210671997070312, "step": 231200 }, { "epoch": 0.9200181019777902, "grad_norm": 0.6838755011558533, "learning_rate": 1.0803050677530496e-05, "loss": 1.9231654357910157, "step": 231250 }, { "epoch": 0.9202170248106504, "grad_norm": 0.4597887396812439, "learning_rate": 1.0801061451180408e-05, "loss": 1.917322998046875, "step": 231300 }, { "epoch": 0.9204159476435104, "grad_norm": 0.7366844415664673, "learning_rate": 1.079907222483032e-05, "loss": 1.921740264892578, "step": 231350 }, { "epoch": 0.9206148704763705, "grad_norm": 0.5812135934829712, "learning_rate": 1.0797082998480232e-05, "loss": 1.9487799072265626, "step": 231400 }, { "epoch": 0.9208137933092305, "grad_norm": 0.6536857485771179, "learning_rate": 1.0795093772130142e-05, "loss": 1.9212730407714844, "step": 231450 }, { "epoch": 0.9210127161420906, "grad_norm": 0.6540970802307129, "learning_rate": 1.0793104545780055e-05, "loss": 1.884788818359375, "step": 231500 }, { "epoch": 0.9212116389749506, "grad_norm": 0.6136449575424194, "learning_rate": 1.0791115319429968e-05, "loss": 1.9047573852539061, "step": 231550 }, { "epoch": 0.9214105618078107, "grad_norm": 0.6203199625015259, "learning_rate": 1.078912609307988e-05, "loss": 1.9157823181152345, "step": 231600 }, { "epoch": 0.9216094846406707, "grad_norm": 0.6299610733985901, "learning_rate": 1.0787136866729793e-05, "loss": 2.0171653747558596, "step": 231650 }, { "epoch": 0.9218084074735309, "grad_norm": 0.6327623724937439, "learning_rate": 1.0785147640379705e-05, "loss": 1.9314285278320313, "step": 231700 }, { "epoch": 0.9220073303063909, "grad_norm": 0.6564725041389465, "learning_rate": 1.0783158414029617e-05, "loss": 1.9513960266113282, "step": 231750 }, { "epoch": 0.922206253139251, "grad_norm": 0.6163555979728699, "learning_rate": 1.0781169187679527e-05, "loss": 1.9541761779785156, "step": 231800 }, { "epoch": 0.922405175972111, "grad_norm": 0.6565739512443542, "learning_rate": 1.077917996132944e-05, "loss": 1.9360581970214843, "step": 231850 }, { "epoch": 0.9226040988049711, "grad_norm": 0.6457798480987549, "learning_rate": 1.0777190734979352e-05, "loss": 2.0084933471679687, "step": 231900 }, { "epoch": 0.9228030216378311, "grad_norm": 0.6878944635391235, "learning_rate": 1.0775201508629266e-05, "loss": 1.904193115234375, "step": 231950 }, { "epoch": 0.9230019444706912, "grad_norm": 0.5953393578529358, "learning_rate": 1.0773212282279178e-05, "loss": 1.9521371459960937, "step": 232000 }, { "epoch": 0.9232008673035512, "grad_norm": 0.6405996680259705, "learning_rate": 1.077122305592909e-05, "loss": 1.918250274658203, "step": 232050 }, { "epoch": 0.9233997901364114, "grad_norm": 0.5376003384590149, "learning_rate": 1.0769233829579e-05, "loss": 1.9497967529296876, "step": 232100 }, { "epoch": 0.9235987129692714, "grad_norm": 0.6833957433700562, "learning_rate": 1.0767244603228912e-05, "loss": 1.9153547668457032, "step": 232150 }, { "epoch": 0.9237976358021315, "grad_norm": 0.6420559287071228, "learning_rate": 1.0765255376878825e-05, "loss": 1.910691680908203, "step": 232200 }, { "epoch": 0.9239965586349915, "grad_norm": 0.6343680024147034, "learning_rate": 1.0763266150528737e-05, "loss": 1.9627192687988282, "step": 232250 }, { "epoch": 0.9241954814678516, "grad_norm": 0.6200809478759766, "learning_rate": 1.0761276924178649e-05, "loss": 1.8902838134765625, "step": 232300 }, { "epoch": 0.9243944043007116, "grad_norm": 0.5639551877975464, "learning_rate": 1.0759287697828563e-05, "loss": 1.94596923828125, "step": 232350 }, { "epoch": 0.9245933271335717, "grad_norm": 0.5816938877105713, "learning_rate": 1.0757298471478475e-05, "loss": 1.9239553833007812, "step": 232400 }, { "epoch": 0.9247922499664317, "grad_norm": 0.7097365260124207, "learning_rate": 1.0755309245128385e-05, "loss": 1.9317727661132813, "step": 232450 }, { "epoch": 0.9249911727992919, "grad_norm": 0.7004187107086182, "learning_rate": 1.0753320018778298e-05, "loss": 1.9710708618164063, "step": 232500 }, { "epoch": 0.9251900956321519, "grad_norm": 0.5438631772994995, "learning_rate": 1.075133079242821e-05, "loss": 1.9234356689453125, "step": 232550 }, { "epoch": 0.925389018465012, "grad_norm": 0.6525129675865173, "learning_rate": 1.0749341566078122e-05, "loss": 1.9083155822753906, "step": 232600 }, { "epoch": 0.925587941297872, "grad_norm": 0.6160567402839661, "learning_rate": 1.0747352339728034e-05, "loss": 1.9040162658691406, "step": 232650 }, { "epoch": 0.9257868641307321, "grad_norm": 0.5733749866485596, "learning_rate": 1.0745363113377946e-05, "loss": 1.88248291015625, "step": 232700 }, { "epoch": 0.9259857869635921, "grad_norm": 0.6245303153991699, "learning_rate": 1.0743373887027857e-05, "loss": 1.902928466796875, "step": 232750 }, { "epoch": 0.9261847097964522, "grad_norm": 0.5646101832389832, "learning_rate": 1.0741384660677769e-05, "loss": 1.9134494018554689, "step": 232800 }, { "epoch": 0.9263836326293122, "grad_norm": 0.49396026134490967, "learning_rate": 1.0739395434327683e-05, "loss": 1.9624591064453125, "step": 232850 }, { "epoch": 0.9265825554621724, "grad_norm": 0.5172989368438721, "learning_rate": 1.0737406207977595e-05, "loss": 1.9501734924316407, "step": 232900 }, { "epoch": 0.9267814782950324, "grad_norm": 0.5803762674331665, "learning_rate": 1.0735416981627507e-05, "loss": 1.9444729614257812, "step": 232950 }, { "epoch": 0.9269804011278925, "grad_norm": 0.4600427746772766, "learning_rate": 1.0733427755277419e-05, "loss": 1.949195556640625, "step": 233000 }, { "epoch": 0.9271793239607525, "grad_norm": 0.5835740566253662, "learning_rate": 1.0731438528927331e-05, "loss": 1.9725299072265625, "step": 233050 }, { "epoch": 0.9273782467936126, "grad_norm": 0.5540322661399841, "learning_rate": 1.0729449302577242e-05, "loss": 1.8881875610351562, "step": 233100 }, { "epoch": 0.9275771696264726, "grad_norm": 0.5471137762069702, "learning_rate": 1.0727460076227154e-05, "loss": 1.9684683227539062, "step": 233150 }, { "epoch": 0.9277760924593327, "grad_norm": 0.6631507277488708, "learning_rate": 1.0725470849877066e-05, "loss": 1.9132102966308593, "step": 233200 }, { "epoch": 0.9279750152921927, "grad_norm": 0.7062480449676514, "learning_rate": 1.072348162352698e-05, "loss": 1.907645721435547, "step": 233250 }, { "epoch": 0.9281739381250529, "grad_norm": 0.4926985502243042, "learning_rate": 1.0721492397176892e-05, "loss": 1.9168260192871094, "step": 233300 }, { "epoch": 0.9283728609579129, "grad_norm": 0.4926934540271759, "learning_rate": 1.0719503170826804e-05, "loss": 1.94060791015625, "step": 233350 }, { "epoch": 0.928571783790773, "grad_norm": 0.6758716106414795, "learning_rate": 1.0717513944476714e-05, "loss": 1.8773670959472657, "step": 233400 }, { "epoch": 0.928770706623633, "grad_norm": 0.7020239233970642, "learning_rate": 1.0715524718126627e-05, "loss": 1.9004042053222656, "step": 233450 }, { "epoch": 0.9289696294564931, "grad_norm": 0.6718794703483582, "learning_rate": 1.0713535491776539e-05, "loss": 1.9737875366210937, "step": 233500 }, { "epoch": 0.9291685522893531, "grad_norm": 0.565423846244812, "learning_rate": 1.0711546265426451e-05, "loss": 1.903275909423828, "step": 233550 }, { "epoch": 0.9293674751222132, "grad_norm": 0.6294510364532471, "learning_rate": 1.0709557039076363e-05, "loss": 1.898836669921875, "step": 233600 }, { "epoch": 0.9295663979550732, "grad_norm": 0.6055424809455872, "learning_rate": 1.0707567812726277e-05, "loss": 1.9025059509277344, "step": 233650 }, { "epoch": 0.9297653207879334, "grad_norm": 0.5995309352874756, "learning_rate": 1.0705578586376189e-05, "loss": 1.9286323547363282, "step": 233700 }, { "epoch": 0.9299642436207934, "grad_norm": 0.5415207147598267, "learning_rate": 1.07035893600261e-05, "loss": 1.9303929138183593, "step": 233750 }, { "epoch": 0.9301631664536535, "grad_norm": 0.6369578838348389, "learning_rate": 1.0701600133676012e-05, "loss": 1.8901957702636718, "step": 233800 }, { "epoch": 0.9303620892865135, "grad_norm": 0.6120049357414246, "learning_rate": 1.0699610907325924e-05, "loss": 1.9239190673828126, "step": 233850 }, { "epoch": 0.9305610121193736, "grad_norm": 0.5447914004325867, "learning_rate": 1.0697621680975836e-05, "loss": 1.906934814453125, "step": 233900 }, { "epoch": 0.9307599349522336, "grad_norm": 0.5786536335945129, "learning_rate": 1.0695632454625748e-05, "loss": 1.9319070434570313, "step": 233950 }, { "epoch": 0.9309588577850937, "grad_norm": 0.7326710820198059, "learning_rate": 1.069364322827566e-05, "loss": 1.9060972595214845, "step": 234000 }, { "epoch": 0.9311577806179537, "grad_norm": 0.6062663793563843, "learning_rate": 1.069165400192557e-05, "loss": 1.913736572265625, "step": 234050 }, { "epoch": 0.9313567034508139, "grad_norm": 0.6364566683769226, "learning_rate": 1.0689664775575483e-05, "loss": 1.9104634094238282, "step": 234100 }, { "epoch": 0.9315556262836739, "grad_norm": 0.42524006962776184, "learning_rate": 1.0687675549225397e-05, "loss": 1.8628076171875, "step": 234150 }, { "epoch": 0.931754549116534, "grad_norm": 0.6180304288864136, "learning_rate": 1.0685686322875309e-05, "loss": 1.8189385986328126, "step": 234200 }, { "epoch": 0.931953471949394, "grad_norm": 0.5954170823097229, "learning_rate": 1.0683697096525221e-05, "loss": 1.8743667602539062, "step": 234250 }, { "epoch": 0.9321523947822541, "grad_norm": 0.5553351640701294, "learning_rate": 1.0681747654702134e-05, "loss": 1.9317161560058593, "step": 234300 }, { "epoch": 0.9323513176151141, "grad_norm": 0.48569655418395996, "learning_rate": 1.067979821287905e-05, "loss": 1.93604248046875, "step": 234350 }, { "epoch": 0.9325502404479742, "grad_norm": 0.5544672608375549, "learning_rate": 1.0677808986528958e-05, "loss": 1.933385009765625, "step": 234400 }, { "epoch": 0.9327491632808342, "grad_norm": 0.658482015132904, "learning_rate": 1.0675819760178872e-05, "loss": 1.9624000549316407, "step": 234450 }, { "epoch": 0.9329480861136944, "grad_norm": 0.6072056889533997, "learning_rate": 1.0673830533828784e-05, "loss": 1.9437002563476562, "step": 234500 }, { "epoch": 0.9331470089465544, "grad_norm": 0.6247831583023071, "learning_rate": 1.0671841307478697e-05, "loss": 1.928236541748047, "step": 234550 }, { "epoch": 0.9333459317794145, "grad_norm": 0.6199629306793213, "learning_rate": 1.0669852081128609e-05, "loss": 1.9089601135253906, "step": 234600 }, { "epoch": 0.9335448546122745, "grad_norm": 0.6775707006454468, "learning_rate": 1.066786285477852e-05, "loss": 1.9006468200683593, "step": 234650 }, { "epoch": 0.9337437774451346, "grad_norm": 0.5861600637435913, "learning_rate": 1.0665873628428431e-05, "loss": 1.9306159973144532, "step": 234700 }, { "epoch": 0.9339427002779946, "grad_norm": 0.6402588486671448, "learning_rate": 1.0663884402078343e-05, "loss": 1.913106689453125, "step": 234750 }, { "epoch": 0.9341416231108547, "grad_norm": 0.6317773461341858, "learning_rate": 1.0661895175728256e-05, "loss": 1.9392640686035156, "step": 234800 }, { "epoch": 0.9343405459437147, "grad_norm": 0.668425440788269, "learning_rate": 1.065990594937817e-05, "loss": 1.9422705078125, "step": 234850 }, { "epoch": 0.9345394687765749, "grad_norm": 0.5611419677734375, "learning_rate": 1.0657916723028082e-05, "loss": 1.9394789123535157, "step": 234900 }, { "epoch": 0.934738391609435, "grad_norm": 0.623313844203949, "learning_rate": 1.0655927496677994e-05, "loss": 1.8604270935058593, "step": 234950 }, { "epoch": 0.934937314442295, "grad_norm": 0.6091877818107605, "learning_rate": 1.0653938270327906e-05, "loss": 1.901603240966797, "step": 235000 }, { "epoch": 0.935136237275155, "grad_norm": 0.6381043195724487, "learning_rate": 1.0651949043977816e-05, "loss": 1.9068524169921874, "step": 235050 }, { "epoch": 0.9353351601080151, "grad_norm": 0.5666685104370117, "learning_rate": 1.0649959817627728e-05, "loss": 1.8906936645507812, "step": 235100 }, { "epoch": 0.9355340829408751, "grad_norm": 0.7163417935371399, "learning_rate": 1.064797059127764e-05, "loss": 1.9279371643066405, "step": 235150 }, { "epoch": 0.9357330057737352, "grad_norm": 0.6118545532226562, "learning_rate": 1.0645981364927553e-05, "loss": 1.9456436157226562, "step": 235200 }, { "epoch": 0.9359319286065952, "grad_norm": 0.5566832423210144, "learning_rate": 1.0643992138577467e-05, "loss": 1.9217083740234375, "step": 235250 }, { "epoch": 0.9361308514394554, "grad_norm": 0.6105031371116638, "learning_rate": 1.0642002912227379e-05, "loss": 1.9049606323242188, "step": 235300 }, { "epoch": 0.9363297742723155, "grad_norm": 0.650223970413208, "learning_rate": 1.0640013685877289e-05, "loss": 1.9318284606933593, "step": 235350 }, { "epoch": 0.9365286971051755, "grad_norm": 0.6047725677490234, "learning_rate": 1.0638024459527201e-05, "loss": 1.9252621459960937, "step": 235400 }, { "epoch": 0.9367276199380355, "grad_norm": 0.6085692048072815, "learning_rate": 1.0636035233177113e-05, "loss": 1.8926002502441406, "step": 235450 }, { "epoch": 0.9369265427708956, "grad_norm": 0.46376505494117737, "learning_rate": 1.0634046006827026e-05, "loss": 1.9593043518066406, "step": 235500 }, { "epoch": 0.9371254656037556, "grad_norm": 0.5944023728370667, "learning_rate": 1.0632056780476938e-05, "loss": 1.9745895385742187, "step": 235550 }, { "epoch": 0.9373243884366157, "grad_norm": 0.7104045748710632, "learning_rate": 1.063006755412685e-05, "loss": 1.895697784423828, "step": 235600 }, { "epoch": 0.9375233112694757, "grad_norm": 0.6473132371902466, "learning_rate": 1.0628078327776764e-05, "loss": 1.9017767333984374, "step": 235650 }, { "epoch": 0.9377222341023359, "grad_norm": 0.6545693874359131, "learning_rate": 1.0626089101426672e-05, "loss": 1.9112281799316406, "step": 235700 }, { "epoch": 0.937921156935196, "grad_norm": 0.5772590041160583, "learning_rate": 1.0624099875076586e-05, "loss": 1.8945428466796874, "step": 235750 }, { "epoch": 0.938120079768056, "grad_norm": 0.7202886343002319, "learning_rate": 1.0622110648726498e-05, "loss": 1.9661146545410155, "step": 235800 }, { "epoch": 0.938319002600916, "grad_norm": 0.6725397109985352, "learning_rate": 1.062012142237641e-05, "loss": 1.8665631103515625, "step": 235850 }, { "epoch": 0.9385179254337761, "grad_norm": 0.6136829853057861, "learning_rate": 1.0618132196026323e-05, "loss": 1.9085435485839843, "step": 235900 }, { "epoch": 0.9387168482666362, "grad_norm": 0.6532204747200012, "learning_rate": 1.0616142969676235e-05, "loss": 1.9545469665527344, "step": 235950 }, { "epoch": 0.9389157710994962, "grad_norm": 0.504672110080719, "learning_rate": 1.0614153743326145e-05, "loss": 1.9377317810058594, "step": 236000 }, { "epoch": 0.9391146939323562, "grad_norm": 0.5518671274185181, "learning_rate": 1.0612164516976057e-05, "loss": 1.8858714294433594, "step": 236050 }, { "epoch": 0.9393136167652164, "grad_norm": 0.5786515474319458, "learning_rate": 1.061017529062597e-05, "loss": 1.9306340026855469, "step": 236100 }, { "epoch": 0.9395125395980765, "grad_norm": 0.612690806388855, "learning_rate": 1.0608186064275883e-05, "loss": 1.9588072204589844, "step": 236150 }, { "epoch": 0.9397114624309365, "grad_norm": 0.5294341444969177, "learning_rate": 1.0606196837925796e-05, "loss": 1.9090284729003906, "step": 236200 }, { "epoch": 0.9399103852637966, "grad_norm": 0.5872505903244019, "learning_rate": 1.0604207611575708e-05, "loss": 1.9007992553710937, "step": 236250 }, { "epoch": 0.9401093080966566, "grad_norm": 0.4759756624698639, "learning_rate": 1.060221838522562e-05, "loss": 1.8953553771972655, "step": 236300 }, { "epoch": 0.9403082309295167, "grad_norm": 0.6716158986091614, "learning_rate": 1.060022915887553e-05, "loss": 1.9396353149414063, "step": 236350 }, { "epoch": 0.9405071537623767, "grad_norm": 0.5608217120170593, "learning_rate": 1.0598239932525443e-05, "loss": 1.9411224365234374, "step": 236400 }, { "epoch": 0.9407060765952368, "grad_norm": 0.654404878616333, "learning_rate": 1.0596250706175355e-05, "loss": 1.9239242553710938, "step": 236450 }, { "epoch": 0.9409049994280969, "grad_norm": 0.6292003393173218, "learning_rate": 1.0594261479825267e-05, "loss": 1.8650701904296876, "step": 236500 }, { "epoch": 0.941103922260957, "grad_norm": 0.48972779512405396, "learning_rate": 1.059227225347518e-05, "loss": 1.94176513671875, "step": 236550 }, { "epoch": 0.941302845093817, "grad_norm": 0.7840906977653503, "learning_rate": 1.0590283027125093e-05, "loss": 1.90143798828125, "step": 236600 }, { "epoch": 0.9415017679266771, "grad_norm": 0.6391528844833374, "learning_rate": 1.0588293800775003e-05, "loss": 1.88985595703125, "step": 236650 }, { "epoch": 0.9417006907595371, "grad_norm": 0.6776024699211121, "learning_rate": 1.0586304574424915e-05, "loss": 1.9154135131835937, "step": 236700 }, { "epoch": 0.9418996135923972, "grad_norm": 0.6365889310836792, "learning_rate": 1.0584315348074828e-05, "loss": 1.9166128540039062, "step": 236750 }, { "epoch": 0.9420985364252572, "grad_norm": 0.7456234693527222, "learning_rate": 1.058232612172474e-05, "loss": 1.9620626831054688, "step": 236800 }, { "epoch": 0.9422974592581173, "grad_norm": 0.6520381569862366, "learning_rate": 1.0580336895374652e-05, "loss": 1.905435791015625, "step": 236850 }, { "epoch": 0.9424963820909773, "grad_norm": 0.5680698752403259, "learning_rate": 1.0578347669024564e-05, "loss": 1.967888641357422, "step": 236900 }, { "epoch": 0.9426953049238375, "grad_norm": 0.5461956262588501, "learning_rate": 1.0576358442674478e-05, "loss": 1.9008837890625, "step": 236950 }, { "epoch": 0.9428942277566975, "grad_norm": 0.40603774785995483, "learning_rate": 1.0574369216324387e-05, "loss": 1.896591796875, "step": 237000 }, { "epoch": 0.9430931505895576, "grad_norm": 0.5911737084388733, "learning_rate": 1.05723799899743e-05, "loss": 1.978851318359375, "step": 237050 }, { "epoch": 0.9432920734224176, "grad_norm": 0.5834019184112549, "learning_rate": 1.0570390763624213e-05, "loss": 1.8850782775878907, "step": 237100 }, { "epoch": 0.9434909962552777, "grad_norm": 0.6166366338729858, "learning_rate": 1.0568401537274125e-05, "loss": 1.9402523803710938, "step": 237150 }, { "epoch": 0.9436899190881377, "grad_norm": 0.5727928876876831, "learning_rate": 1.0566412310924037e-05, "loss": 1.9248457336425782, "step": 237200 }, { "epoch": 0.9438888419209978, "grad_norm": 0.5632866024971008, "learning_rate": 1.0564423084573949e-05, "loss": 1.8818719482421875, "step": 237250 }, { "epoch": 0.9440877647538578, "grad_norm": 0.688393771648407, "learning_rate": 1.056243385822386e-05, "loss": 1.921998291015625, "step": 237300 }, { "epoch": 0.944286687586718, "grad_norm": 0.5477756857872009, "learning_rate": 1.0560444631873772e-05, "loss": 1.9491267395019531, "step": 237350 }, { "epoch": 0.944485610419578, "grad_norm": 0.5028636455535889, "learning_rate": 1.0558455405523684e-05, "loss": 1.9200875854492188, "step": 237400 }, { "epoch": 0.9446845332524381, "grad_norm": 0.7763418555259705, "learning_rate": 1.0556466179173596e-05, "loss": 1.927711944580078, "step": 237450 }, { "epoch": 0.9448834560852981, "grad_norm": 0.5595151782035828, "learning_rate": 1.055447695282351e-05, "loss": 1.9113302612304688, "step": 237500 }, { "epoch": 0.9450823789181582, "grad_norm": 0.667367696762085, "learning_rate": 1.0552487726473422e-05, "loss": 1.937039337158203, "step": 237550 }, { "epoch": 0.9452813017510182, "grad_norm": 0.5791504383087158, "learning_rate": 1.0550498500123334e-05, "loss": 1.9153955078125, "step": 237600 }, { "epoch": 0.9454802245838783, "grad_norm": 0.6607599854469299, "learning_rate": 1.0548509273773244e-05, "loss": 1.9171717834472657, "step": 237650 }, { "epoch": 0.9456791474167383, "grad_norm": 0.6162293553352356, "learning_rate": 1.0546520047423157e-05, "loss": 1.8677418518066407, "step": 237700 }, { "epoch": 0.9458780702495985, "grad_norm": 0.5549947023391724, "learning_rate": 1.0544530821073069e-05, "loss": 1.9560806274414062, "step": 237750 }, { "epoch": 0.9460769930824585, "grad_norm": 0.6237648725509644, "learning_rate": 1.0542541594722981e-05, "loss": 1.9184910583496093, "step": 237800 }, { "epoch": 0.9462759159153186, "grad_norm": 0.5505233407020569, "learning_rate": 1.0540552368372893e-05, "loss": 1.9193536376953124, "step": 237850 }, { "epoch": 0.9464748387481786, "grad_norm": 0.568585991859436, "learning_rate": 1.0538563142022807e-05, "loss": 1.93814208984375, "step": 237900 }, { "epoch": 0.9466737615810387, "grad_norm": 0.7294853925704956, "learning_rate": 1.0536573915672716e-05, "loss": 1.903561553955078, "step": 237950 }, { "epoch": 0.9468726844138987, "grad_norm": 0.6234079003334045, "learning_rate": 1.053458468932263e-05, "loss": 1.9329733276367187, "step": 238000 }, { "epoch": 0.9470716072467588, "grad_norm": 0.7294489741325378, "learning_rate": 1.0532595462972542e-05, "loss": 1.9060862731933594, "step": 238050 }, { "epoch": 0.9472705300796188, "grad_norm": 0.5623129606246948, "learning_rate": 1.0530606236622454e-05, "loss": 1.9424150085449219, "step": 238100 }, { "epoch": 0.947469452912479, "grad_norm": 0.6124558448791504, "learning_rate": 1.0528617010272366e-05, "loss": 1.9586563110351562, "step": 238150 }, { "epoch": 0.947668375745339, "grad_norm": 0.622071385383606, "learning_rate": 1.0526627783922278e-05, "loss": 1.8979301452636719, "step": 238200 }, { "epoch": 0.9478672985781991, "grad_norm": 0.5140340328216553, "learning_rate": 1.052463855757219e-05, "loss": 1.9061123657226562, "step": 238250 }, { "epoch": 0.9480662214110591, "grad_norm": 0.542145311832428, "learning_rate": 1.05226493312221e-05, "loss": 1.897386474609375, "step": 238300 }, { "epoch": 0.9482651442439192, "grad_norm": 0.6214171051979065, "learning_rate": 1.0520699889399016e-05, "loss": 1.9263421630859374, "step": 238350 }, { "epoch": 0.9484640670767792, "grad_norm": 0.6262476444244385, "learning_rate": 1.0518710663048928e-05, "loss": 1.913343963623047, "step": 238400 }, { "epoch": 0.9486629899096393, "grad_norm": 0.6929861903190613, "learning_rate": 1.051672143669884e-05, "loss": 1.8417431640625, "step": 238450 }, { "epoch": 0.9488619127424993, "grad_norm": 0.5049673914909363, "learning_rate": 1.0514732210348752e-05, "loss": 1.9310472106933594, "step": 238500 }, { "epoch": 0.9490608355753595, "grad_norm": 0.6077099442481995, "learning_rate": 1.0512742983998664e-05, "loss": 1.9445468139648439, "step": 238550 }, { "epoch": 0.9492597584082195, "grad_norm": 0.6249160766601562, "learning_rate": 1.0510753757648575e-05, "loss": 1.9081985473632812, "step": 238600 }, { "epoch": 0.9494586812410796, "grad_norm": 0.6770691871643066, "learning_rate": 1.0508764531298487e-05, "loss": 1.972194061279297, "step": 238650 }, { "epoch": 0.9496576040739396, "grad_norm": 0.6396705508232117, "learning_rate": 1.05067753049484e-05, "loss": 1.9495191955566407, "step": 238700 }, { "epoch": 0.9498565269067997, "grad_norm": 0.5762730240821838, "learning_rate": 1.0504786078598313e-05, "loss": 1.9279307556152343, "step": 238750 }, { "epoch": 0.9500554497396597, "grad_norm": 0.5921325087547302, "learning_rate": 1.0502796852248225e-05, "loss": 1.9059658813476563, "step": 238800 }, { "epoch": 0.9502543725725198, "grad_norm": 0.6154397130012512, "learning_rate": 1.0500807625898137e-05, "loss": 1.9211721801757813, "step": 238850 }, { "epoch": 0.9504532954053798, "grad_norm": 0.6338339447975159, "learning_rate": 1.049881839954805e-05, "loss": 1.9125506591796875, "step": 238900 }, { "epoch": 0.95065221823824, "grad_norm": 0.6900097727775574, "learning_rate": 1.049682917319796e-05, "loss": 1.9035533142089844, "step": 238950 }, { "epoch": 0.9508511410711, "grad_norm": 0.7764650583267212, "learning_rate": 1.0494839946847872e-05, "loss": 1.9132319641113282, "step": 239000 }, { "epoch": 0.9510500639039601, "grad_norm": 0.5468494296073914, "learning_rate": 1.0492850720497784e-05, "loss": 1.8866262817382813, "step": 239050 }, { "epoch": 0.9512489867368201, "grad_norm": 0.7926319241523743, "learning_rate": 1.0490861494147696e-05, "loss": 1.9076968383789064, "step": 239100 }, { "epoch": 0.9514479095696802, "grad_norm": 0.6392654776573181, "learning_rate": 1.048887226779761e-05, "loss": 1.9165953063964845, "step": 239150 }, { "epoch": 0.9516468324025402, "grad_norm": 0.5417115688323975, "learning_rate": 1.0486883041447522e-05, "loss": 1.9231671142578124, "step": 239200 }, { "epoch": 0.9518457552354003, "grad_norm": 0.6417663097381592, "learning_rate": 1.0484893815097433e-05, "loss": 1.9399490356445312, "step": 239250 }, { "epoch": 0.9520446780682603, "grad_norm": 0.49757152795791626, "learning_rate": 1.0482904588747345e-05, "loss": 1.9082144165039063, "step": 239300 }, { "epoch": 0.9522436009011205, "grad_norm": 0.518656313419342, "learning_rate": 1.0480915362397257e-05, "loss": 1.9078726196289062, "step": 239350 }, { "epoch": 0.9524425237339805, "grad_norm": 0.5400851368904114, "learning_rate": 1.047892613604717e-05, "loss": 1.8698248291015624, "step": 239400 }, { "epoch": 0.9526414465668406, "grad_norm": 0.5749469995498657, "learning_rate": 1.0476936909697081e-05, "loss": 1.944681396484375, "step": 239450 }, { "epoch": 0.9528403693997006, "grad_norm": 0.6410975456237793, "learning_rate": 1.0474947683346994e-05, "loss": 1.9139337158203125, "step": 239500 }, { "epoch": 0.9530392922325607, "grad_norm": 0.6058395504951477, "learning_rate": 1.0472958456996907e-05, "loss": 1.8771553039550781, "step": 239550 }, { "epoch": 0.9532382150654207, "grad_norm": 0.6373329162597656, "learning_rate": 1.0470969230646816e-05, "loss": 1.9495498657226562, "step": 239600 }, { "epoch": 0.9534371378982808, "grad_norm": 0.6202234029769897, "learning_rate": 1.046898000429673e-05, "loss": 1.9091264343261718, "step": 239650 }, { "epoch": 0.9536360607311408, "grad_norm": 0.5355921983718872, "learning_rate": 1.0466990777946642e-05, "loss": 1.9815560913085937, "step": 239700 }, { "epoch": 0.953834983564001, "grad_norm": 0.5651111602783203, "learning_rate": 1.0465001551596554e-05, "loss": 1.9198472595214844, "step": 239750 }, { "epoch": 0.954033906396861, "grad_norm": 0.692969560623169, "learning_rate": 1.0463012325246466e-05, "loss": 1.9538079833984374, "step": 239800 }, { "epoch": 0.9542328292297211, "grad_norm": 0.7618812918663025, "learning_rate": 1.0461023098896379e-05, "loss": 1.91029052734375, "step": 239850 }, { "epoch": 0.9544317520625811, "grad_norm": 0.4916873574256897, "learning_rate": 1.0459033872546289e-05, "loss": 1.8860359191894531, "step": 239900 }, { "epoch": 0.9546306748954412, "grad_norm": 0.7254324555397034, "learning_rate": 1.0457044646196201e-05, "loss": 1.8792997741699218, "step": 239950 }, { "epoch": 0.9548295977283012, "grad_norm": 0.5744848847389221, "learning_rate": 1.0455055419846113e-05, "loss": 1.8714381408691407, "step": 240000 }, { "epoch": 0.9550285205611613, "grad_norm": 0.6008189916610718, "learning_rate": 1.0453066193496027e-05, "loss": 1.9046153259277343, "step": 240050 }, { "epoch": 0.9552274433940213, "grad_norm": 0.6582966446876526, "learning_rate": 1.045107696714594e-05, "loss": 1.9344465637207031, "step": 240100 }, { "epoch": 0.9554263662268815, "grad_norm": 0.6837573051452637, "learning_rate": 1.0449087740795851e-05, "loss": 1.9137882995605469, "step": 240150 }, { "epoch": 0.9556252890597415, "grad_norm": 0.6165426969528198, "learning_rate": 1.0447098514445762e-05, "loss": 1.9468858337402344, "step": 240200 }, { "epoch": 0.9558242118926016, "grad_norm": 0.6755176782608032, "learning_rate": 1.0445149072622675e-05, "loss": 1.900701446533203, "step": 240250 }, { "epoch": 0.9560231347254616, "grad_norm": 0.6787141561508179, "learning_rate": 1.0443159846272588e-05, "loss": 1.9321601867675782, "step": 240300 }, { "epoch": 0.9562220575583217, "grad_norm": 0.5573254227638245, "learning_rate": 1.04411706199225e-05, "loss": 1.8858865356445313, "step": 240350 }, { "epoch": 0.9564209803911817, "grad_norm": 0.7159140706062317, "learning_rate": 1.0439181393572414e-05, "loss": 1.966884765625, "step": 240400 }, { "epoch": 0.9566199032240418, "grad_norm": 0.6339734196662903, "learning_rate": 1.0437192167222326e-05, "loss": 1.9757920837402343, "step": 240450 }, { "epoch": 0.9568188260569018, "grad_norm": 0.6325072050094604, "learning_rate": 1.0435202940872238e-05, "loss": 1.9313299560546875, "step": 240500 }, { "epoch": 0.957017748889762, "grad_norm": 0.5915086269378662, "learning_rate": 1.0433213714522148e-05, "loss": 1.8848136901855468, "step": 240550 }, { "epoch": 0.957216671722622, "grad_norm": 0.6902451515197754, "learning_rate": 1.043122448817206e-05, "loss": 1.9018406677246094, "step": 240600 }, { "epoch": 0.9574155945554821, "grad_norm": 0.48660939931869507, "learning_rate": 1.0429235261821973e-05, "loss": 1.9347105407714844, "step": 240650 }, { "epoch": 0.9576145173883421, "grad_norm": 0.799964964389801, "learning_rate": 1.0427246035471885e-05, "loss": 1.9430644226074218, "step": 240700 }, { "epoch": 0.9578134402212022, "grad_norm": 0.6434339284896851, "learning_rate": 1.0425256809121797e-05, "loss": 1.9322361755371094, "step": 240750 }, { "epoch": 0.9580123630540622, "grad_norm": 0.5443002581596375, "learning_rate": 1.042326758277171e-05, "loss": 1.867418212890625, "step": 240800 }, { "epoch": 0.9582112858869223, "grad_norm": 0.5730472207069397, "learning_rate": 1.0421278356421623e-05, "loss": 1.9581816101074219, "step": 240850 }, { "epoch": 0.9584102087197823, "grad_norm": 0.7170524597167969, "learning_rate": 1.0419289130071533e-05, "loss": 1.94066650390625, "step": 240900 }, { "epoch": 0.9586091315526425, "grad_norm": 0.6205130815505981, "learning_rate": 1.0417299903721445e-05, "loss": 1.8923208618164062, "step": 240950 }, { "epoch": 0.9588080543855025, "grad_norm": 0.548316240310669, "learning_rate": 1.0415310677371358e-05, "loss": 1.9426502990722656, "step": 241000 }, { "epoch": 0.9590069772183626, "grad_norm": 0.6514095067977905, "learning_rate": 1.041332145102127e-05, "loss": 1.8886692810058594, "step": 241050 }, { "epoch": 0.9592059000512226, "grad_norm": 0.8296677470207214, "learning_rate": 1.0411372009198183e-05, "loss": 2.007939910888672, "step": 241100 }, { "epoch": 0.9594048228840827, "grad_norm": 0.5712206959724426, "learning_rate": 1.0409382782848097e-05, "loss": 1.8747763061523437, "step": 241150 }, { "epoch": 0.9596037457169427, "grad_norm": 0.6001817584037781, "learning_rate": 1.0407393556498006e-05, "loss": 1.8805859375, "step": 241200 }, { "epoch": 0.9598026685498028, "grad_norm": 0.7541288733482361, "learning_rate": 1.040540433014792e-05, "loss": 1.8952577209472656, "step": 241250 }, { "epoch": 0.9600015913826628, "grad_norm": 0.6176155805587769, "learning_rate": 1.0403415103797832e-05, "loss": 1.913729248046875, "step": 241300 }, { "epoch": 0.960200514215523, "grad_norm": 0.5914310812950134, "learning_rate": 1.0401425877447744e-05, "loss": 1.8579347229003906, "step": 241350 }, { "epoch": 0.960399437048383, "grad_norm": 0.56103515625, "learning_rate": 1.0399436651097656e-05, "loss": 1.8964533996582031, "step": 241400 }, { "epoch": 0.9605983598812431, "grad_norm": 0.6055774688720703, "learning_rate": 1.0397447424747568e-05, "loss": 1.9036825561523438, "step": 241450 }, { "epoch": 0.9607972827141031, "grad_norm": 0.5948882699012756, "learning_rate": 1.039545819839748e-05, "loss": 1.9251222229003906, "step": 241500 }, { "epoch": 0.9609962055469632, "grad_norm": 0.5388286709785461, "learning_rate": 1.0393468972047391e-05, "loss": 1.9505113220214845, "step": 241550 }, { "epoch": 0.9611951283798232, "grad_norm": 0.6626489758491516, "learning_rate": 1.0391479745697303e-05, "loss": 1.916793670654297, "step": 241600 }, { "epoch": 0.9613940512126833, "grad_norm": 0.5026517510414124, "learning_rate": 1.0389490519347217e-05, "loss": 1.9443006896972657, "step": 241650 }, { "epoch": 0.9615929740455433, "grad_norm": 0.6298881769180298, "learning_rate": 1.0387501292997129e-05, "loss": 1.9732008361816407, "step": 241700 }, { "epoch": 0.9617918968784035, "grad_norm": 0.610779881477356, "learning_rate": 1.0385512066647041e-05, "loss": 1.9042294311523438, "step": 241750 }, { "epoch": 0.9619908197112635, "grad_norm": 0.7128349542617798, "learning_rate": 1.0383522840296953e-05, "loss": 1.8753564453125, "step": 241800 }, { "epoch": 0.9621897425441236, "grad_norm": 0.481121301651001, "learning_rate": 1.0381533613946864e-05, "loss": 1.988456268310547, "step": 241850 }, { "epoch": 0.9623886653769836, "grad_norm": 0.6348769068717957, "learning_rate": 1.0379544387596776e-05, "loss": 1.9568064880371094, "step": 241900 }, { "epoch": 0.9625875882098437, "grad_norm": 0.5815595984458923, "learning_rate": 1.0377555161246688e-05, "loss": 1.8818145751953126, "step": 241950 }, { "epoch": 0.9627865110427037, "grad_norm": 0.6902779340744019, "learning_rate": 1.03755659348966e-05, "loss": 1.9359037780761719, "step": 242000 }, { "epoch": 0.9629854338755638, "grad_norm": 0.5587375164031982, "learning_rate": 1.0373576708546514e-05, "loss": 1.8915548706054688, "step": 242050 }, { "epoch": 0.9631843567084238, "grad_norm": 0.582061231136322, "learning_rate": 1.0371587482196426e-05, "loss": 1.903466796875, "step": 242100 }, { "epoch": 0.963383279541284, "grad_norm": 0.6447423696517944, "learning_rate": 1.0369598255846338e-05, "loss": 1.8987191772460938, "step": 242150 }, { "epoch": 0.963582202374144, "grad_norm": 0.6254127621650696, "learning_rate": 1.0367609029496249e-05, "loss": 1.9189710998535157, "step": 242200 }, { "epoch": 0.9637811252070041, "grad_norm": 0.6983504295349121, "learning_rate": 1.0365619803146161e-05, "loss": 1.91719482421875, "step": 242250 }, { "epoch": 0.9639800480398641, "grad_norm": 0.6372857093811035, "learning_rate": 1.0363630576796073e-05, "loss": 1.9491447448730468, "step": 242300 }, { "epoch": 0.9641789708727242, "grad_norm": 0.588111937046051, "learning_rate": 1.0361641350445985e-05, "loss": 1.8811054992675782, "step": 242350 }, { "epoch": 0.9643778937055842, "grad_norm": 0.5462092757225037, "learning_rate": 1.0359652124095897e-05, "loss": 1.9127647399902343, "step": 242400 }, { "epoch": 0.9645768165384443, "grad_norm": 0.5950251817703247, "learning_rate": 1.0357662897745811e-05, "loss": 1.9391397094726563, "step": 242450 }, { "epoch": 0.9647757393713043, "grad_norm": 0.704819917678833, "learning_rate": 1.035567367139572e-05, "loss": 1.9035536193847655, "step": 242500 }, { "epoch": 0.9649746622041645, "grad_norm": 0.546552300453186, "learning_rate": 1.0353684445045634e-05, "loss": 1.9600177001953125, "step": 242550 }, { "epoch": 0.9651735850370246, "grad_norm": 0.6316819190979004, "learning_rate": 1.0351695218695546e-05, "loss": 1.8941580200195312, "step": 242600 }, { "epoch": 0.9653725078698846, "grad_norm": 0.6525806188583374, "learning_rate": 1.0349705992345458e-05, "loss": 1.9324412536621094, "step": 242650 }, { "epoch": 0.9655714307027446, "grad_norm": 0.6369016170501709, "learning_rate": 1.034771676599537e-05, "loss": 1.8854600524902343, "step": 242700 }, { "epoch": 0.9657703535356047, "grad_norm": 0.6855711340904236, "learning_rate": 1.0345727539645282e-05, "loss": 1.9086158752441407, "step": 242750 }, { "epoch": 0.9659692763684647, "grad_norm": 0.6204760074615479, "learning_rate": 1.0343738313295195e-05, "loss": 1.887755126953125, "step": 242800 }, { "epoch": 0.9661681992013248, "grad_norm": 0.6385406255722046, "learning_rate": 1.0341749086945105e-05, "loss": 1.9601602172851562, "step": 242850 }, { "epoch": 0.9663671220341848, "grad_norm": 0.6517096757888794, "learning_rate": 1.0339759860595017e-05, "loss": 1.8767666625976562, "step": 242900 }, { "epoch": 0.966566044867045, "grad_norm": 0.6122742295265198, "learning_rate": 1.0337770634244931e-05, "loss": 1.915108184814453, "step": 242950 }, { "epoch": 0.966764967699905, "grad_norm": 0.6660585999488831, "learning_rate": 1.0335781407894843e-05, "loss": 1.9423338317871093, "step": 243000 }, { "epoch": 0.9669638905327651, "grad_norm": 0.557931125164032, "learning_rate": 1.0333792181544755e-05, "loss": 1.9065643310546876, "step": 243050 }, { "epoch": 0.9671628133656252, "grad_norm": 0.6134954690933228, "learning_rate": 1.0331802955194667e-05, "loss": 1.9477294921875, "step": 243100 }, { "epoch": 0.9673617361984852, "grad_norm": 0.6871724724769592, "learning_rate": 1.0329813728844578e-05, "loss": 1.9070616149902344, "step": 243150 }, { "epoch": 0.9675606590313452, "grad_norm": 0.5424934029579163, "learning_rate": 1.032782450249449e-05, "loss": 1.9059043884277345, "step": 243200 }, { "epoch": 0.9677595818642053, "grad_norm": 0.61090487241745, "learning_rate": 1.0325835276144402e-05, "loss": 1.9268267822265626, "step": 243250 }, { "epoch": 0.9679585046970653, "grad_norm": 0.6551761031150818, "learning_rate": 1.0323846049794314e-05, "loss": 1.9135491943359375, "step": 243300 }, { "epoch": 0.9681574275299255, "grad_norm": 0.5975556373596191, "learning_rate": 1.0321856823444228e-05, "loss": 1.8885710144042969, "step": 243350 }, { "epoch": 0.9683563503627856, "grad_norm": 0.5696113705635071, "learning_rate": 1.031986759709414e-05, "loss": 1.9209294128417969, "step": 243400 }, { "epoch": 0.9685552731956456, "grad_norm": 0.6540392637252808, "learning_rate": 1.031787837074405e-05, "loss": 1.948734893798828, "step": 243450 }, { "epoch": 0.9687541960285057, "grad_norm": 0.6703161001205444, "learning_rate": 1.0315889144393963e-05, "loss": 1.9176615905761718, "step": 243500 }, { "epoch": 0.9689531188613657, "grad_norm": 0.5989747047424316, "learning_rate": 1.0313899918043875e-05, "loss": 1.9083944702148437, "step": 243550 }, { "epoch": 0.9691520416942258, "grad_norm": 0.6624101400375366, "learning_rate": 1.0311910691693787e-05, "loss": 1.9422030639648438, "step": 243600 }, { "epoch": 0.9693509645270858, "grad_norm": 0.600473940372467, "learning_rate": 1.03099214653437e-05, "loss": 1.90163818359375, "step": 243650 }, { "epoch": 0.9695498873599459, "grad_norm": 0.7076425552368164, "learning_rate": 1.0307932238993611e-05, "loss": 1.9128273010253907, "step": 243700 }, { "epoch": 0.969748810192806, "grad_norm": 0.6856187582015991, "learning_rate": 1.0305943012643525e-05, "loss": 1.9516293334960937, "step": 243750 }, { "epoch": 0.9699477330256661, "grad_norm": 0.6449884176254272, "learning_rate": 1.0303953786293434e-05, "loss": 1.9034547424316406, "step": 243800 }, { "epoch": 0.9701466558585261, "grad_norm": 0.617494523525238, "learning_rate": 1.0301964559943348e-05, "loss": 1.947199249267578, "step": 243850 }, { "epoch": 0.9703455786913862, "grad_norm": 0.5593754053115845, "learning_rate": 1.029997533359326e-05, "loss": 1.9253515625, "step": 243900 }, { "epoch": 0.9705445015242462, "grad_norm": 0.6139973998069763, "learning_rate": 1.0297986107243172e-05, "loss": 1.9256224060058593, "step": 243950 }, { "epoch": 0.9707434243571063, "grad_norm": 0.6591303944587708, "learning_rate": 1.0295996880893084e-05, "loss": 1.942705841064453, "step": 244000 }, { "epoch": 0.9709423471899663, "grad_norm": 0.6316538453102112, "learning_rate": 1.0294007654542996e-05, "loss": 1.9358213806152345, "step": 244050 }, { "epoch": 0.9711412700228264, "grad_norm": 0.5371450781822205, "learning_rate": 1.0292018428192907e-05, "loss": 1.9636196899414062, "step": 244100 }, { "epoch": 0.9713401928556865, "grad_norm": 0.6029168963432312, "learning_rate": 1.0290029201842819e-05, "loss": 1.9244255065917968, "step": 244150 }, { "epoch": 0.9715391156885466, "grad_norm": 0.6171601414680481, "learning_rate": 1.0288039975492731e-05, "loss": 1.9059135437011718, "step": 244200 }, { "epoch": 0.9717380385214066, "grad_norm": 0.5804551839828491, "learning_rate": 1.0286050749142645e-05, "loss": 1.8887741088867187, "step": 244250 }, { "epoch": 0.9719369613542667, "grad_norm": 0.5731338858604431, "learning_rate": 1.0284061522792557e-05, "loss": 1.97151123046875, "step": 244300 }, { "epoch": 0.9721358841871267, "grad_norm": 0.6440850496292114, "learning_rate": 1.028207229644247e-05, "loss": 1.8808251953125, "step": 244350 }, { "epoch": 0.9723348070199868, "grad_norm": 0.6473183631896973, "learning_rate": 1.0280083070092381e-05, "loss": 1.91527587890625, "step": 244400 }, { "epoch": 0.9725337298528468, "grad_norm": 0.6243454217910767, "learning_rate": 1.0278093843742292e-05, "loss": 1.95357177734375, "step": 244450 }, { "epoch": 0.9727326526857069, "grad_norm": 0.6541977524757385, "learning_rate": 1.0276104617392204e-05, "loss": 1.9430314636230468, "step": 244500 }, { "epoch": 0.9729315755185669, "grad_norm": 0.5540120601654053, "learning_rate": 1.0274115391042116e-05, "loss": 1.911558837890625, "step": 244550 }, { "epoch": 0.9731304983514271, "grad_norm": 0.6561856269836426, "learning_rate": 1.0272126164692028e-05, "loss": 1.869613800048828, "step": 244600 }, { "epoch": 0.9733294211842871, "grad_norm": 0.5905885100364685, "learning_rate": 1.027013693834194e-05, "loss": 1.9556697082519532, "step": 244650 }, { "epoch": 0.9735283440171472, "grad_norm": 0.6944381594657898, "learning_rate": 1.0268147711991854e-05, "loss": 1.8728489685058594, "step": 244700 }, { "epoch": 0.9737272668500072, "grad_norm": 0.7015166878700256, "learning_rate": 1.0266158485641765e-05, "loss": 1.898538818359375, "step": 244750 }, { "epoch": 0.9739261896828673, "grad_norm": 0.5817675590515137, "learning_rate": 1.0264169259291677e-05, "loss": 2.000113067626953, "step": 244800 }, { "epoch": 0.9741251125157273, "grad_norm": 0.5359107255935669, "learning_rate": 1.0262180032941589e-05, "loss": 1.8844624328613282, "step": 244850 }, { "epoch": 0.9743240353485874, "grad_norm": 0.6687321066856384, "learning_rate": 1.0260190806591501e-05, "loss": 1.9637448120117187, "step": 244900 }, { "epoch": 0.9745229581814474, "grad_norm": 0.6553518176078796, "learning_rate": 1.0258201580241413e-05, "loss": 1.896499481201172, "step": 244950 }, { "epoch": 0.9747218810143076, "grad_norm": 0.6376892328262329, "learning_rate": 1.0256212353891326e-05, "loss": 1.87160888671875, "step": 245000 }, { "epoch": 0.9749208038471676, "grad_norm": 0.5796130299568176, "learning_rate": 1.0254223127541238e-05, "loss": 1.9286050415039062, "step": 245050 }, { "epoch": 0.9751197266800277, "grad_norm": 0.6113117933273315, "learning_rate": 1.0252233901191148e-05, "loss": 1.9638938903808594, "step": 245100 }, { "epoch": 0.9753186495128877, "grad_norm": 0.5944579243659973, "learning_rate": 1.025024467484106e-05, "loss": 1.8882757568359374, "step": 245150 }, { "epoch": 0.9755175723457478, "grad_norm": 0.6721470355987549, "learning_rate": 1.0248255448490974e-05, "loss": 1.8970320129394531, "step": 245200 }, { "epoch": 0.9757164951786078, "grad_norm": 0.5955960750579834, "learning_rate": 1.0246266222140886e-05, "loss": 1.9114176940917968, "step": 245250 }, { "epoch": 0.9759154180114679, "grad_norm": 0.6729441285133362, "learning_rate": 1.0244276995790798e-05, "loss": 1.94185546875, "step": 245300 }, { "epoch": 0.9761143408443279, "grad_norm": 0.5482197999954224, "learning_rate": 1.024228776944071e-05, "loss": 1.9019509887695312, "step": 245350 }, { "epoch": 0.9763132636771881, "grad_norm": 0.6458950042724609, "learning_rate": 1.0240298543090621e-05, "loss": 1.8994615173339844, "step": 245400 }, { "epoch": 0.9765121865100481, "grad_norm": 0.6757642030715942, "learning_rate": 1.0238309316740533e-05, "loss": 1.9479469299316405, "step": 245450 }, { "epoch": 0.9767111093429082, "grad_norm": 0.6010845303535461, "learning_rate": 1.0236320090390445e-05, "loss": 1.9664825439453124, "step": 245500 }, { "epoch": 0.9769100321757682, "grad_norm": 0.6418342590332031, "learning_rate": 1.0234330864040357e-05, "loss": 1.9488198852539063, "step": 245550 }, { "epoch": 0.9771089550086283, "grad_norm": 0.6469101905822754, "learning_rate": 1.0232341637690271e-05, "loss": 1.9012454223632813, "step": 245600 }, { "epoch": 0.9773078778414883, "grad_norm": 0.6192911863327026, "learning_rate": 1.0230352411340183e-05, "loss": 1.9222616577148437, "step": 245650 }, { "epoch": 0.9775068006743484, "grad_norm": 0.5643234848976135, "learning_rate": 1.0228363184990096e-05, "loss": 1.9170468139648438, "step": 245700 }, { "epoch": 0.9777057235072084, "grad_norm": 0.6067061424255371, "learning_rate": 1.0226373958640006e-05, "loss": 1.9656002807617188, "step": 245750 }, { "epoch": 0.9779046463400686, "grad_norm": 0.6690434813499451, "learning_rate": 1.0224384732289918e-05, "loss": 1.9387535095214843, "step": 245800 }, { "epoch": 0.9781035691729286, "grad_norm": 0.6456690430641174, "learning_rate": 1.022239550593983e-05, "loss": 1.9622518920898437, "step": 245850 }, { "epoch": 0.9783024920057887, "grad_norm": 0.6498227715492249, "learning_rate": 1.0220406279589742e-05, "loss": 1.8959686279296875, "step": 245900 }, { "epoch": 0.9785014148386487, "grad_norm": 0.6293913125991821, "learning_rate": 1.0218417053239655e-05, "loss": 1.8999160766601562, "step": 245950 }, { "epoch": 0.9787003376715088, "grad_norm": 0.6864774227142334, "learning_rate": 1.0216427826889568e-05, "loss": 1.9292723083496093, "step": 246000 }, { "epoch": 0.9788992605043688, "grad_norm": 0.67593914270401, "learning_rate": 1.0214438600539477e-05, "loss": 1.942269287109375, "step": 246050 }, { "epoch": 0.9790981833372289, "grad_norm": 0.6162497997283936, "learning_rate": 1.0212449374189391e-05, "loss": 1.9072084045410156, "step": 246100 }, { "epoch": 0.9792971061700889, "grad_norm": 0.6035870313644409, "learning_rate": 1.0210460147839303e-05, "loss": 1.9712637329101563, "step": 246150 }, { "epoch": 0.9794960290029491, "grad_norm": 0.6097192764282227, "learning_rate": 1.0208470921489215e-05, "loss": 1.9648431396484376, "step": 246200 }, { "epoch": 0.9796949518358091, "grad_norm": 0.5927661061286926, "learning_rate": 1.0206481695139128e-05, "loss": 1.9037298583984374, "step": 246250 }, { "epoch": 0.9798938746686692, "grad_norm": 0.544912576675415, "learning_rate": 1.020449246878904e-05, "loss": 1.8590675354003907, "step": 246300 }, { "epoch": 0.9800927975015292, "grad_norm": 0.5603645443916321, "learning_rate": 1.0202503242438952e-05, "loss": 2.0039741516113283, "step": 246350 }, { "epoch": 0.9802917203343893, "grad_norm": 0.5427824854850769, "learning_rate": 1.0200514016088862e-05, "loss": 1.9672409057617188, "step": 246400 }, { "epoch": 0.9804906431672493, "grad_norm": 0.6192269921302795, "learning_rate": 1.0198524789738774e-05, "loss": 1.8912234497070313, "step": 246450 }, { "epoch": 0.9806895660001094, "grad_norm": 0.6181466579437256, "learning_rate": 1.0196535563388688e-05, "loss": 1.931571044921875, "step": 246500 }, { "epoch": 0.9808884888329694, "grad_norm": 0.5753871202468872, "learning_rate": 1.01945463370386e-05, "loss": 1.90349853515625, "step": 246550 }, { "epoch": 0.9810874116658296, "grad_norm": 0.6013830304145813, "learning_rate": 1.0192557110688513e-05, "loss": 1.9503553771972657, "step": 246600 }, { "epoch": 0.9812863344986896, "grad_norm": 0.653654158115387, "learning_rate": 1.0190567884338425e-05, "loss": 1.9457223510742188, "step": 246650 }, { "epoch": 0.9814852573315497, "grad_norm": 0.6672387719154358, "learning_rate": 1.0188578657988335e-05, "loss": 1.9379335021972657, "step": 246700 }, { "epoch": 0.9816841801644097, "grad_norm": 0.6665735840797424, "learning_rate": 1.0186589431638247e-05, "loss": 1.887244415283203, "step": 246750 }, { "epoch": 0.9818831029972698, "grad_norm": 0.6708016991615295, "learning_rate": 1.018460020528816e-05, "loss": 1.9189866638183595, "step": 246800 }, { "epoch": 0.9820820258301298, "grad_norm": 0.6706681847572327, "learning_rate": 1.0182610978938072e-05, "loss": 1.911614990234375, "step": 246850 }, { "epoch": 0.9822809486629899, "grad_norm": 0.6705705523490906, "learning_rate": 1.0180621752587985e-05, "loss": 1.9175505065917968, "step": 246900 }, { "epoch": 0.9824798714958499, "grad_norm": 0.7080526351928711, "learning_rate": 1.0178632526237898e-05, "loss": 1.9201045227050781, "step": 246950 }, { "epoch": 0.9826787943287101, "grad_norm": 0.6153372526168823, "learning_rate": 1.017664329988781e-05, "loss": 1.8795199584960938, "step": 247000 }, { "epoch": 0.9828777171615701, "grad_norm": 0.6897954940795898, "learning_rate": 1.017465407353772e-05, "loss": 1.9038558959960938, "step": 247050 }, { "epoch": 0.9830766399944302, "grad_norm": 0.5068761706352234, "learning_rate": 1.0172664847187632e-05, "loss": 1.9292738342285156, "step": 247100 }, { "epoch": 0.9832755628272902, "grad_norm": 0.7253710627555847, "learning_rate": 1.0170675620837544e-05, "loss": 1.9574131774902344, "step": 247150 }, { "epoch": 0.9834744856601503, "grad_norm": 0.7372097969055176, "learning_rate": 1.0168686394487457e-05, "loss": 1.8956150817871094, "step": 247200 }, { "epoch": 0.9836734084930103, "grad_norm": 0.6370980739593506, "learning_rate": 1.0166697168137369e-05, "loss": 1.9654953002929687, "step": 247250 }, { "epoch": 0.9838723313258704, "grad_norm": 0.5540226697921753, "learning_rate": 1.0164707941787283e-05, "loss": 1.9107920837402343, "step": 247300 }, { "epoch": 0.9840712541587304, "grad_norm": 0.7139046788215637, "learning_rate": 1.0162718715437191e-05, "loss": 1.860429229736328, "step": 247350 }, { "epoch": 0.9842701769915906, "grad_norm": 0.554252564907074, "learning_rate": 1.0160729489087105e-05, "loss": 1.8949037170410157, "step": 247400 }, { "epoch": 0.9844690998244506, "grad_norm": 0.6724543571472168, "learning_rate": 1.0158740262737017e-05, "loss": 1.9050335693359375, "step": 247450 }, { "epoch": 0.9846680226573107, "grad_norm": 0.6138593554496765, "learning_rate": 1.015675103638693e-05, "loss": 1.923331298828125, "step": 247500 }, { "epoch": 0.9848669454901707, "grad_norm": 0.7383430004119873, "learning_rate": 1.0154761810036842e-05, "loss": 1.9388945007324219, "step": 247550 }, { "epoch": 0.9850658683230308, "grad_norm": 0.5760554671287537, "learning_rate": 1.0152772583686754e-05, "loss": 1.9268403625488282, "step": 247600 }, { "epoch": 0.9852647911558908, "grad_norm": 0.6433870196342468, "learning_rate": 1.0150823141863669e-05, "loss": 1.8965213012695312, "step": 247650 }, { "epoch": 0.9854637139887509, "grad_norm": 0.48212650418281555, "learning_rate": 1.0148833915513578e-05, "loss": 1.923721466064453, "step": 247700 }, { "epoch": 0.9856626368216109, "grad_norm": 0.5958032608032227, "learning_rate": 1.0146844689163492e-05, "loss": 1.9452217102050782, "step": 247750 }, { "epoch": 0.9858615596544711, "grad_norm": 0.6539788246154785, "learning_rate": 1.0144855462813404e-05, "loss": 1.9275408935546876, "step": 247800 }, { "epoch": 0.9860604824873311, "grad_norm": 0.571832001209259, "learning_rate": 1.0142866236463316e-05, "loss": 1.8937176513671874, "step": 247850 }, { "epoch": 0.9862594053201912, "grad_norm": 0.7066701054573059, "learning_rate": 1.0140877010113228e-05, "loss": 1.9274713134765624, "step": 247900 }, { "epoch": 0.9864583281530512, "grad_norm": 0.5407889485359192, "learning_rate": 1.013888778376314e-05, "loss": 1.9175184631347657, "step": 247950 }, { "epoch": 0.9866572509859113, "grad_norm": 0.6122527718544006, "learning_rate": 1.013689855741305e-05, "loss": 1.89196044921875, "step": 248000 }, { "epoch": 0.9868561738187713, "grad_norm": 0.7285898327827454, "learning_rate": 1.0134909331062963e-05, "loss": 1.9164396667480468, "step": 248050 }, { "epoch": 0.9870550966516314, "grad_norm": 0.5840665102005005, "learning_rate": 1.0132920104712875e-05, "loss": 1.8787826538085937, "step": 248100 }, { "epoch": 0.9872540194844914, "grad_norm": 0.5071786642074585, "learning_rate": 1.0130930878362789e-05, "loss": 1.896571807861328, "step": 248150 }, { "epoch": 0.9874529423173516, "grad_norm": 0.6585614681243896, "learning_rate": 1.0128941652012701e-05, "loss": 1.9474591064453124, "step": 248200 }, { "epoch": 0.9876518651502116, "grad_norm": 0.663258969783783, "learning_rate": 1.0126952425662613e-05, "loss": 1.8664691162109375, "step": 248250 }, { "epoch": 0.9878507879830717, "grad_norm": 0.6068998575210571, "learning_rate": 1.0124963199312525e-05, "loss": 1.9154470825195313, "step": 248300 }, { "epoch": 0.9880497108159317, "grad_norm": 0.5780694484710693, "learning_rate": 1.0122973972962436e-05, "loss": 1.9177000427246094, "step": 248350 }, { "epoch": 0.9882486336487918, "grad_norm": 0.5170224905014038, "learning_rate": 1.0120984746612348e-05, "loss": 1.8947718811035157, "step": 248400 }, { "epoch": 0.9884475564816518, "grad_norm": 0.6350798010826111, "learning_rate": 1.011899552026226e-05, "loss": 1.9158985900878907, "step": 248450 }, { "epoch": 0.9886464793145119, "grad_norm": 0.5939338803291321, "learning_rate": 1.0117006293912172e-05, "loss": 1.935623016357422, "step": 248500 }, { "epoch": 0.9888454021473719, "grad_norm": 0.6410792469978333, "learning_rate": 1.0115017067562086e-05, "loss": 1.8768634033203124, "step": 248550 }, { "epoch": 0.9890443249802321, "grad_norm": 0.6192904114723206, "learning_rate": 1.0113027841211998e-05, "loss": 1.9477171325683593, "step": 248600 }, { "epoch": 0.9892432478130921, "grad_norm": 0.606308102607727, "learning_rate": 1.0111038614861908e-05, "loss": 1.856270294189453, "step": 248650 }, { "epoch": 0.9894421706459522, "grad_norm": 0.5590584874153137, "learning_rate": 1.010904938851182e-05, "loss": 1.9156098937988282, "step": 248700 }, { "epoch": 0.9896410934788122, "grad_norm": 0.6633460521697998, "learning_rate": 1.0107060162161733e-05, "loss": 1.9637896728515625, "step": 248750 }, { "epoch": 0.9898400163116723, "grad_norm": 0.601179301738739, "learning_rate": 1.0105070935811645e-05, "loss": 1.893760986328125, "step": 248800 }, { "epoch": 0.9900389391445323, "grad_norm": 0.618304431438446, "learning_rate": 1.0103081709461557e-05, "loss": 1.9302642822265625, "step": 248850 }, { "epoch": 0.9902378619773924, "grad_norm": 0.6128675937652588, "learning_rate": 1.010109248311147e-05, "loss": 1.9527375793457031, "step": 248900 }, { "epoch": 0.9904367848102524, "grad_norm": 0.648382842540741, "learning_rate": 1.0099103256761383e-05, "loss": 1.9048361206054687, "step": 248950 }, { "epoch": 0.9906357076431126, "grad_norm": 0.6071004271507263, "learning_rate": 1.0097114030411292e-05, "loss": 1.9355154418945313, "step": 249000 }, { "epoch": 0.9908346304759726, "grad_norm": 0.6122827529907227, "learning_rate": 1.0095124804061206e-05, "loss": 1.91973388671875, "step": 249050 }, { "epoch": 0.9910335533088327, "grad_norm": 0.5537679195404053, "learning_rate": 1.0093135577711118e-05, "loss": 1.9540095520019531, "step": 249100 }, { "epoch": 0.9912324761416927, "grad_norm": 0.6574816703796387, "learning_rate": 1.009114635136103e-05, "loss": 1.8759490966796875, "step": 249150 }, { "epoch": 0.9914313989745528, "grad_norm": 0.5065442323684692, "learning_rate": 1.0089157125010942e-05, "loss": 1.9441424560546876, "step": 249200 }, { "epoch": 0.9916303218074128, "grad_norm": 0.7768864035606384, "learning_rate": 1.0087167898660854e-05, "loss": 1.9598579406738281, "step": 249250 }, { "epoch": 0.9918292446402729, "grad_norm": 0.556033730506897, "learning_rate": 1.0085178672310765e-05, "loss": 1.9121978759765625, "step": 249300 }, { "epoch": 0.9920281674731329, "grad_norm": 0.6813965439796448, "learning_rate": 1.0083189445960677e-05, "loss": 1.9479873657226563, "step": 249350 }, { "epoch": 0.9922270903059931, "grad_norm": 0.5741145014762878, "learning_rate": 1.0081200219610589e-05, "loss": 1.9159388732910156, "step": 249400 }, { "epoch": 0.9924260131388531, "grad_norm": 0.595835268497467, "learning_rate": 1.0079210993260503e-05, "loss": 1.9542781066894532, "step": 249450 }, { "epoch": 0.9926249359717132, "grad_norm": 0.6424641013145447, "learning_rate": 1.0077221766910415e-05, "loss": 1.8710554504394532, "step": 249500 }, { "epoch": 0.9928238588045732, "grad_norm": 0.5937554836273193, "learning_rate": 1.0075232540560327e-05, "loss": 1.904057159423828, "step": 249550 }, { "epoch": 0.9930227816374333, "grad_norm": 0.647436261177063, "learning_rate": 1.007324331421024e-05, "loss": 1.912535400390625, "step": 249600 }, { "epoch": 0.9932217044702933, "grad_norm": 0.6413513422012329, "learning_rate": 1.007125408786015e-05, "loss": 1.9223583984375, "step": 249650 }, { "epoch": 0.9934206273031534, "grad_norm": 0.6518971920013428, "learning_rate": 1.0069304646037063e-05, "loss": 1.902694091796875, "step": 249700 }, { "epoch": 0.9936195501360134, "grad_norm": 0.5258127450942993, "learning_rate": 1.0067315419686975e-05, "loss": 1.9044206237792969, "step": 249750 }, { "epoch": 0.9938184729688736, "grad_norm": 0.668860673904419, "learning_rate": 1.006532619333689e-05, "loss": 1.9230158996582032, "step": 249800 }, { "epoch": 0.9940173958017336, "grad_norm": 0.5906127691268921, "learning_rate": 1.0063336966986801e-05, "loss": 1.9156489562988281, "step": 249850 }, { "epoch": 0.9942163186345937, "grad_norm": 0.6483474969863892, "learning_rate": 1.0061347740636713e-05, "loss": 1.9107369995117187, "step": 249900 }, { "epoch": 0.9944152414674537, "grad_norm": 0.677234947681427, "learning_rate": 1.0059358514286624e-05, "loss": 1.9101661682128905, "step": 249950 }, { "epoch": 0.9946141643003138, "grad_norm": 0.7855494022369385, "learning_rate": 1.0057369287936536e-05, "loss": 1.9671038818359374, "step": 250000 }, { "epoch": 0.9948130871331738, "grad_norm": 0.7035454511642456, "learning_rate": 1.0055380061586448e-05, "loss": 1.902901611328125, "step": 250050 }, { "epoch": 0.9950120099660339, "grad_norm": 0.5657080411911011, "learning_rate": 1.005339083523636e-05, "loss": 1.852921142578125, "step": 250100 }, { "epoch": 0.9952109327988939, "grad_norm": 0.5378425717353821, "learning_rate": 1.0051401608886273e-05, "loss": 1.888086700439453, "step": 250150 }, { "epoch": 0.9954098556317541, "grad_norm": 0.5977141261100769, "learning_rate": 1.0049412382536186e-05, "loss": 1.9552357482910157, "step": 250200 }, { "epoch": 0.9956087784646142, "grad_norm": 0.5800620913505554, "learning_rate": 1.0047423156186099e-05, "loss": 1.9754966735839843, "step": 250250 }, { "epoch": 0.9958077012974742, "grad_norm": 0.6341877579689026, "learning_rate": 1.0045433929836009e-05, "loss": 1.9672845458984376, "step": 250300 }, { "epoch": 0.9960066241303343, "grad_norm": 0.6625377535820007, "learning_rate": 1.0043444703485921e-05, "loss": 1.938709716796875, "step": 250350 }, { "epoch": 0.9962055469631943, "grad_norm": 0.6227965354919434, "learning_rate": 1.0041455477135833e-05, "loss": 1.8945455932617188, "step": 250400 }, { "epoch": 0.9964044697960543, "grad_norm": 0.5626790523529053, "learning_rate": 1.0039466250785745e-05, "loss": 1.9008050537109376, "step": 250450 }, { "epoch": 0.9966033926289144, "grad_norm": 0.5866935849189758, "learning_rate": 1.0037477024435658e-05, "loss": 1.9275590515136718, "step": 250500 }, { "epoch": 0.9968023154617744, "grad_norm": 0.5917348861694336, "learning_rate": 1.003548779808557e-05, "loss": 1.9039111328125, "step": 250550 }, { "epoch": 0.9970012382946346, "grad_norm": 0.6096296906471252, "learning_rate": 1.003349857173548e-05, "loss": 1.9409449768066407, "step": 250600 }, { "epoch": 0.9972001611274947, "grad_norm": 0.6178233623504639, "learning_rate": 1.0031509345385392e-05, "loss": 1.9176309204101563, "step": 250650 }, { "epoch": 0.9973990839603547, "grad_norm": 0.6470136642456055, "learning_rate": 1.0029520119035306e-05, "loss": 1.9467898559570314, "step": 250700 }, { "epoch": 0.9975980067932148, "grad_norm": 0.5740290284156799, "learning_rate": 1.0027530892685218e-05, "loss": 1.9207559204101563, "step": 250750 }, { "epoch": 0.9977969296260748, "grad_norm": 0.7818469405174255, "learning_rate": 1.002554166633513e-05, "loss": 1.8845933532714845, "step": 250800 }, { "epoch": 0.9979958524589349, "grad_norm": 0.5903899669647217, "learning_rate": 1.0023552439985043e-05, "loss": 1.9146995544433594, "step": 250850 }, { "epoch": 0.9981947752917949, "grad_norm": 0.6697097420692444, "learning_rate": 1.0021563213634955e-05, "loss": 1.9708030700683594, "step": 250900 }, { "epoch": 0.998393698124655, "grad_norm": 0.6758453845977783, "learning_rate": 1.0019573987284865e-05, "loss": 1.8832516479492187, "step": 250950 }, { "epoch": 0.9985926209575151, "grad_norm": 0.7555890083312988, "learning_rate": 1.0017584760934777e-05, "loss": 1.9079380798339844, "step": 251000 }, { "epoch": 0.9987915437903752, "grad_norm": 0.6184512376785278, "learning_rate": 1.001559553458469e-05, "loss": 1.9131947326660157, "step": 251050 }, { "epoch": 0.9989904666232352, "grad_norm": 0.5666922330856323, "learning_rate": 1.0013606308234602e-05, "loss": 1.9065623474121094, "step": 251100 }, { "epoch": 0.9991893894560953, "grad_norm": 0.7005065083503723, "learning_rate": 1.0011617081884515e-05, "loss": 1.9003958129882812, "step": 251150 }, { "epoch": 0.9993883122889553, "grad_norm": 0.5615023970603943, "learning_rate": 1.0009627855534428e-05, "loss": 1.981317901611328, "step": 251200 }, { "epoch": 0.9995872351218154, "grad_norm": 0.636727511882782, "learning_rate": 1.0007638629184338e-05, "loss": 1.9224723815917968, "step": 251250 }, { "epoch": 0.9997861579546754, "grad_norm": 0.6550567150115967, "learning_rate": 1.000564940283425e-05, "loss": 1.9448435974121094, "step": 251300 }, { "epoch": 0.9999850807875355, "grad_norm": 0.6913137435913086, "learning_rate": 1.0003660176484162e-05, "loss": 1.9256886291503905, "step": 251350 }, { "epoch": 1.0001830090062314, "grad_norm": 0.663007915019989, "learning_rate": 1.0001670950134074e-05, "loss": 1.8653013610839844, "step": 251400 }, { "epoch": 1.0003819318390914, "grad_norm": 0.6957257390022278, "learning_rate": 9.999681723783987e-06, "loss": 1.8942214965820312, "step": 251450 }, { "epoch": 1.0005808546719515, "grad_norm": 0.5917683243751526, "learning_rate": 9.997692497433899e-06, "loss": 1.8750599670410155, "step": 251500 }, { "epoch": 1.0007797775048115, "grad_norm": 0.6294801831245422, "learning_rate": 9.995703271083811e-06, "loss": 1.9150604248046874, "step": 251550 }, { "epoch": 1.0009787003376716, "grad_norm": 0.7615512609481812, "learning_rate": 9.993714044733723e-06, "loss": 1.8886976623535157, "step": 251600 }, { "epoch": 1.0011776231705316, "grad_norm": 0.6076008081436157, "learning_rate": 9.991724818383635e-06, "loss": 1.908482208251953, "step": 251650 }, { "epoch": 1.0013765460033917, "grad_norm": 0.601330578327179, "learning_rate": 9.989735592033547e-06, "loss": 1.8911129760742187, "step": 251700 }, { "epoch": 1.0015754688362517, "grad_norm": 0.6619437336921692, "learning_rate": 9.98774636568346e-06, "loss": 1.8445364379882812, "step": 251750 }, { "epoch": 1.0017743916691118, "grad_norm": 0.6732245683670044, "learning_rate": 9.98575713933337e-06, "loss": 1.8375041198730468, "step": 251800 }, { "epoch": 1.0019733145019718, "grad_norm": 0.9901807308197021, "learning_rate": 9.983807697510285e-06, "loss": 1.893560333251953, "step": 251850 }, { "epoch": 1.0021722373348319, "grad_norm": 0.5372016429901123, "learning_rate": 9.981818471160197e-06, "loss": 1.8582180786132811, "step": 251900 }, { "epoch": 1.002371160167692, "grad_norm": 0.5836195349693298, "learning_rate": 9.97982924481011e-06, "loss": 1.9303892517089845, "step": 251950 }, { "epoch": 1.002570083000552, "grad_norm": 0.6318982839584351, "learning_rate": 9.977840018460022e-06, "loss": 1.838524627685547, "step": 252000 }, { "epoch": 1.002769005833412, "grad_norm": 0.6991223692893982, "learning_rate": 9.975850792109934e-06, "loss": 1.9129774475097656, "step": 252050 }, { "epoch": 1.002967928666272, "grad_norm": 0.628959596157074, "learning_rate": 9.973861565759846e-06, "loss": 1.8642665100097657, "step": 252100 }, { "epoch": 1.003166851499132, "grad_norm": 0.6877052783966064, "learning_rate": 9.971872339409758e-06, "loss": 1.9059783935546875, "step": 252150 }, { "epoch": 1.0033657743319924, "grad_norm": 0.5670610666275024, "learning_rate": 9.96988311305967e-06, "loss": 1.9057485961914062, "step": 252200 }, { "epoch": 1.0035646971648524, "grad_norm": 0.6073005795478821, "learning_rate": 9.967893886709582e-06, "loss": 1.8685903930664063, "step": 252250 }, { "epoch": 1.0037636199977125, "grad_norm": 0.5577684640884399, "learning_rate": 9.965904660359493e-06, "loss": 1.8508987426757812, "step": 252300 }, { "epoch": 1.0039625428305725, "grad_norm": 0.7252808809280396, "learning_rate": 9.963915434009405e-06, "loss": 1.9165936279296876, "step": 252350 }, { "epoch": 1.0041614656634326, "grad_norm": 0.6336053609848022, "learning_rate": 9.961926207659319e-06, "loss": 1.879228057861328, "step": 252400 }, { "epoch": 1.0043603884962926, "grad_norm": 0.6951159834861755, "learning_rate": 9.95993698130923e-06, "loss": 1.8905538940429687, "step": 252450 }, { "epoch": 1.0045593113291527, "grad_norm": 0.5236043334007263, "learning_rate": 9.957947754959141e-06, "loss": 1.8597976684570312, "step": 252500 }, { "epoch": 1.0047582341620127, "grad_norm": 0.6304293274879456, "learning_rate": 9.955958528609053e-06, "loss": 1.8633041381835938, "step": 252550 }, { "epoch": 1.0049571569948728, "grad_norm": 0.5307576060295105, "learning_rate": 9.953969302258967e-06, "loss": 1.8697064208984375, "step": 252600 }, { "epoch": 1.0051560798277328, "grad_norm": 0.6343225240707397, "learning_rate": 9.951980075908878e-06, "loss": 1.8936494445800782, "step": 252650 }, { "epoch": 1.0053550026605929, "grad_norm": 0.7177470922470093, "learning_rate": 9.94999084955879e-06, "loss": 1.9246275329589844, "step": 252700 }, { "epoch": 1.005553925493453, "grad_norm": 0.7445796132087708, "learning_rate": 9.948001623208702e-06, "loss": 1.9156585693359376, "step": 252750 }, { "epoch": 1.005752848326313, "grad_norm": 0.5856938362121582, "learning_rate": 9.946012396858614e-06, "loss": 1.8793405151367188, "step": 252800 }, { "epoch": 1.005951771159173, "grad_norm": 0.6489778161048889, "learning_rate": 9.944023170508526e-06, "loss": 1.8628688049316406, "step": 252850 }, { "epoch": 1.006150693992033, "grad_norm": 0.6374128460884094, "learning_rate": 9.942033944158439e-06, "loss": 1.8918190002441406, "step": 252900 }, { "epoch": 1.006349616824893, "grad_norm": 0.5927213430404663, "learning_rate": 9.94004471780835e-06, "loss": 1.8840217590332031, "step": 252950 }, { "epoch": 1.0065485396577534, "grad_norm": 0.6330884099006653, "learning_rate": 9.938055491458263e-06, "loss": 1.873417510986328, "step": 253000 }, { "epoch": 1.0067474624906134, "grad_norm": 0.7273655533790588, "learning_rate": 9.936066265108175e-06, "loss": 1.8402052307128907, "step": 253050 }, { "epoch": 1.0069463853234735, "grad_norm": 0.6250107288360596, "learning_rate": 9.934077038758087e-06, "loss": 1.8345233154296876, "step": 253100 }, { "epoch": 1.0071453081563335, "grad_norm": 0.7557152509689331, "learning_rate": 9.932087812408e-06, "loss": 1.8605644226074218, "step": 253150 }, { "epoch": 1.0073442309891936, "grad_norm": 0.6777316331863403, "learning_rate": 9.930098586057911e-06, "loss": 1.8478384399414063, "step": 253200 }, { "epoch": 1.0075431538220536, "grad_norm": 0.8045459389686584, "learning_rate": 9.928109359707824e-06, "loss": 1.868666534423828, "step": 253250 }, { "epoch": 1.0077420766549137, "grad_norm": 0.5515750646591187, "learning_rate": 9.926120133357736e-06, "loss": 1.883946990966797, "step": 253300 }, { "epoch": 1.0079409994877737, "grad_norm": 0.697730541229248, "learning_rate": 9.924130907007648e-06, "loss": 1.8983616638183594, "step": 253350 }, { "epoch": 1.0081399223206338, "grad_norm": 0.5475822687149048, "learning_rate": 9.92214168065756e-06, "loss": 1.8870222473144531, "step": 253400 }, { "epoch": 1.0083388451534938, "grad_norm": 0.6573666930198669, "learning_rate": 9.92015245430747e-06, "loss": 1.79459228515625, "step": 253450 }, { "epoch": 1.0085377679863539, "grad_norm": 0.5969873070716858, "learning_rate": 9.918163227957384e-06, "loss": 1.8431642150878906, "step": 253500 }, { "epoch": 1.008736690819214, "grad_norm": 0.6238161325454712, "learning_rate": 9.916174001607296e-06, "loss": 1.9010209655761718, "step": 253550 }, { "epoch": 1.008935613652074, "grad_norm": 0.6457561254501343, "learning_rate": 9.914184775257207e-06, "loss": 1.8689320373535157, "step": 253600 }, { "epoch": 1.009134536484934, "grad_norm": 0.6076177954673767, "learning_rate": 9.912195548907119e-06, "loss": 1.8937008666992188, "step": 253650 }, { "epoch": 1.009333459317794, "grad_norm": 0.8053234219551086, "learning_rate": 9.910206322557033e-06, "loss": 1.9047125244140626, "step": 253700 }, { "epoch": 1.009532382150654, "grad_norm": 0.6767064332962036, "learning_rate": 9.908217096206943e-06, "loss": 1.9065286254882812, "step": 253750 }, { "epoch": 1.0097313049835144, "grad_norm": 0.606854259967804, "learning_rate": 9.906227869856855e-06, "loss": 1.915023193359375, "step": 253800 }, { "epoch": 1.0099302278163744, "grad_norm": 0.7092632055282593, "learning_rate": 9.904238643506768e-06, "loss": 1.8906723022460938, "step": 253850 }, { "epoch": 1.0101291506492345, "grad_norm": 0.6145822405815125, "learning_rate": 9.902249417156681e-06, "loss": 1.8999525451660155, "step": 253900 }, { "epoch": 1.0103280734820945, "grad_norm": 0.632665753364563, "learning_rate": 9.900260190806592e-06, "loss": 1.9327192687988282, "step": 253950 }, { "epoch": 1.0105269963149546, "grad_norm": 0.5513363480567932, "learning_rate": 9.898270964456504e-06, "loss": 1.8757563781738282, "step": 254000 }, { "epoch": 1.0107259191478146, "grad_norm": 0.6066961884498596, "learning_rate": 9.896281738106416e-06, "loss": 1.8868580627441407, "step": 254050 }, { "epoch": 1.0109248419806747, "grad_norm": 0.7070772051811218, "learning_rate": 9.894292511756328e-06, "loss": 1.8898472595214844, "step": 254100 }, { "epoch": 1.0111237648135347, "grad_norm": 0.6875351667404175, "learning_rate": 9.89230328540624e-06, "loss": 1.8580447387695314, "step": 254150 }, { "epoch": 1.0113226876463948, "grad_norm": 0.6496527194976807, "learning_rate": 9.890314059056153e-06, "loss": 1.9315260314941407, "step": 254200 }, { "epoch": 1.0115216104792548, "grad_norm": 0.5718750357627869, "learning_rate": 9.888324832706065e-06, "loss": 1.8741143798828126, "step": 254250 }, { "epoch": 1.0117205333121149, "grad_norm": 0.530620276927948, "learning_rate": 9.886335606355977e-06, "loss": 1.9117524719238281, "step": 254300 }, { "epoch": 1.011919456144975, "grad_norm": 0.6061214208602905, "learning_rate": 9.884346380005889e-06, "loss": 1.8808187866210937, "step": 254350 }, { "epoch": 1.012118378977835, "grad_norm": 0.5472016930580139, "learning_rate": 9.882357153655801e-06, "loss": 1.8529573059082032, "step": 254400 }, { "epoch": 1.012317301810695, "grad_norm": 0.5850628614425659, "learning_rate": 9.880367927305713e-06, "loss": 1.8698190307617188, "step": 254450 }, { "epoch": 1.012516224643555, "grad_norm": 0.636215090751648, "learning_rate": 9.878378700955626e-06, "loss": 1.8872486877441406, "step": 254500 }, { "epoch": 1.0127151474764151, "grad_norm": 0.6252774596214294, "learning_rate": 9.876389474605538e-06, "loss": 1.851514129638672, "step": 254550 }, { "epoch": 1.0129140703092754, "grad_norm": 0.6450246572494507, "learning_rate": 9.874440032782451e-06, "loss": 1.9252299499511718, "step": 254600 }, { "epoch": 1.0131129931421354, "grad_norm": 0.619499146938324, "learning_rate": 9.872450806432363e-06, "loss": 1.8710369873046875, "step": 254650 }, { "epoch": 1.0133119159749955, "grad_norm": 0.5939483642578125, "learning_rate": 9.870461580082275e-06, "loss": 1.8879385375976563, "step": 254700 }, { "epoch": 1.0135108388078555, "grad_norm": 0.6723388433456421, "learning_rate": 9.868472353732188e-06, "loss": 1.9153790283203125, "step": 254750 }, { "epoch": 1.0137097616407156, "grad_norm": 0.48007604479789734, "learning_rate": 9.8664831273821e-06, "loss": 1.8481800842285157, "step": 254800 }, { "epoch": 1.0139086844735756, "grad_norm": 0.6145443916320801, "learning_rate": 9.864493901032012e-06, "loss": 1.8525567626953126, "step": 254850 }, { "epoch": 1.0141076073064357, "grad_norm": 0.6599355936050415, "learning_rate": 9.862504674681922e-06, "loss": 1.8700782775878906, "step": 254900 }, { "epoch": 1.0143065301392957, "grad_norm": 0.6186160445213318, "learning_rate": 9.860515448331836e-06, "loss": 1.8901029968261718, "step": 254950 }, { "epoch": 1.0145054529721558, "grad_norm": 0.4841049611568451, "learning_rate": 9.858526221981748e-06, "loss": 1.8496394348144531, "step": 255000 }, { "epoch": 1.0147043758050158, "grad_norm": 0.6821517944335938, "learning_rate": 9.856536995631659e-06, "loss": 1.8802005004882814, "step": 255050 }, { "epoch": 1.0149032986378759, "grad_norm": 0.6353356242179871, "learning_rate": 9.854547769281571e-06, "loss": 1.853887481689453, "step": 255100 }, { "epoch": 1.015102221470736, "grad_norm": 0.7253705859184265, "learning_rate": 9.852558542931485e-06, "loss": 1.8952073669433593, "step": 255150 }, { "epoch": 1.015301144303596, "grad_norm": 0.6677622199058533, "learning_rate": 9.850569316581397e-06, "loss": 1.8928912353515626, "step": 255200 }, { "epoch": 1.015500067136456, "grad_norm": 0.6906731724739075, "learning_rate": 9.848580090231307e-06, "loss": 1.9389906311035157, "step": 255250 }, { "epoch": 1.015698989969316, "grad_norm": 0.6206244230270386, "learning_rate": 9.84659086388122e-06, "loss": 1.92907470703125, "step": 255300 }, { "epoch": 1.0158979128021761, "grad_norm": 0.5110836625099182, "learning_rate": 9.844601637531133e-06, "loss": 1.8785592651367187, "step": 255350 }, { "epoch": 1.0160968356350364, "grad_norm": 0.565740704536438, "learning_rate": 9.842612411181044e-06, "loss": 1.8633815002441407, "step": 255400 }, { "epoch": 1.0162957584678964, "grad_norm": 0.5344323515892029, "learning_rate": 9.840623184830956e-06, "loss": 1.8762019348144532, "step": 255450 }, { "epoch": 1.0164946813007565, "grad_norm": 0.5598470568656921, "learning_rate": 9.838633958480868e-06, "loss": 1.878065643310547, "step": 255500 }, { "epoch": 1.0166936041336165, "grad_norm": 0.5874972343444824, "learning_rate": 9.83664473213078e-06, "loss": 1.9259896850585938, "step": 255550 }, { "epoch": 1.0168925269664766, "grad_norm": 0.6752815246582031, "learning_rate": 9.834655505780692e-06, "loss": 1.9122482299804688, "step": 255600 }, { "epoch": 1.0170914497993366, "grad_norm": 0.492195725440979, "learning_rate": 9.832666279430605e-06, "loss": 1.8571037292480468, "step": 255650 }, { "epoch": 1.0172903726321967, "grad_norm": 0.5761342644691467, "learning_rate": 9.830677053080517e-06, "loss": 1.8381065368652343, "step": 255700 }, { "epoch": 1.0174892954650567, "grad_norm": 0.5867770910263062, "learning_rate": 9.828687826730429e-06, "loss": 1.9101959228515626, "step": 255750 }, { "epoch": 1.0176882182979168, "grad_norm": 0.573582112789154, "learning_rate": 9.826698600380341e-06, "loss": 1.857204132080078, "step": 255800 }, { "epoch": 1.0178871411307768, "grad_norm": 0.5236883759498596, "learning_rate": 9.824709374030253e-06, "loss": 1.902135772705078, "step": 255850 }, { "epoch": 1.0180860639636369, "grad_norm": 0.5826842188835144, "learning_rate": 9.822720147680165e-06, "loss": 1.861331787109375, "step": 255900 }, { "epoch": 1.018284986796497, "grad_norm": 0.5348026156425476, "learning_rate": 9.820730921330077e-06, "loss": 1.89557861328125, "step": 255950 }, { "epoch": 1.018483909629357, "grad_norm": 0.5774080753326416, "learning_rate": 9.81874169497999e-06, "loss": 1.8613897705078124, "step": 256000 }, { "epoch": 1.018682832462217, "grad_norm": 0.6352986693382263, "learning_rate": 9.816752468629902e-06, "loss": 1.8727606201171876, "step": 256050 }, { "epoch": 1.018881755295077, "grad_norm": 0.7375523447990417, "learning_rate": 9.814763242279814e-06, "loss": 1.9005732727050781, "step": 256100 }, { "epoch": 1.0190806781279371, "grad_norm": 0.6387300491333008, "learning_rate": 9.812774015929726e-06, "loss": 1.8979774475097657, "step": 256150 }, { "epoch": 1.0192796009607972, "grad_norm": 0.5934498310089111, "learning_rate": 9.810784789579636e-06, "loss": 1.8740339660644532, "step": 256200 }, { "epoch": 1.0194785237936574, "grad_norm": 0.6700533628463745, "learning_rate": 9.80879556322955e-06, "loss": 1.8970132446289063, "step": 256250 }, { "epoch": 1.0196774466265175, "grad_norm": 0.6630362272262573, "learning_rate": 9.806806336879462e-06, "loss": 1.912611083984375, "step": 256300 }, { "epoch": 1.0198763694593775, "grad_norm": 0.5860635042190552, "learning_rate": 9.804817110529373e-06, "loss": 1.8568833923339845, "step": 256350 }, { "epoch": 1.0200752922922376, "grad_norm": 0.7379211783409119, "learning_rate": 9.802827884179285e-06, "loss": 1.852220001220703, "step": 256400 }, { "epoch": 1.0202742151250976, "grad_norm": 0.7468568086624146, "learning_rate": 9.800838657829199e-06, "loss": 1.889986572265625, "step": 256450 }, { "epoch": 1.0204731379579577, "grad_norm": 0.6122701168060303, "learning_rate": 9.798849431479111e-06, "loss": 1.891746826171875, "step": 256500 }, { "epoch": 1.0206720607908177, "grad_norm": 0.6177230477333069, "learning_rate": 9.796860205129021e-06, "loss": 1.9106272888183593, "step": 256550 }, { "epoch": 1.0208709836236778, "grad_norm": 0.6706838011741638, "learning_rate": 9.794870978778934e-06, "loss": 1.8684068298339844, "step": 256600 }, { "epoch": 1.0210699064565378, "grad_norm": 0.6582521200180054, "learning_rate": 9.792881752428847e-06, "loss": 1.89190185546875, "step": 256650 }, { "epoch": 1.0212688292893979, "grad_norm": 0.5429525971412659, "learning_rate": 9.790892526078758e-06, "loss": 1.8877377319335937, "step": 256700 }, { "epoch": 1.021467752122258, "grad_norm": 0.6448703408241272, "learning_rate": 9.78890329972867e-06, "loss": 1.8898738098144532, "step": 256750 }, { "epoch": 1.021666674955118, "grad_norm": 0.7637540698051453, "learning_rate": 9.786914073378582e-06, "loss": 1.9276960754394532, "step": 256800 }, { "epoch": 1.021865597787978, "grad_norm": 0.7247146368026733, "learning_rate": 9.784924847028494e-06, "loss": 1.89789794921875, "step": 256850 }, { "epoch": 1.022064520620838, "grad_norm": 0.7395443320274353, "learning_rate": 9.782935620678406e-06, "loss": 1.9052491760253907, "step": 256900 }, { "epoch": 1.0222634434536981, "grad_norm": 0.63163161277771, "learning_rate": 9.780946394328319e-06, "loss": 1.9510372924804686, "step": 256950 }, { "epoch": 1.0224623662865584, "grad_norm": 0.5869089365005493, "learning_rate": 9.77895716797823e-06, "loss": 1.85730712890625, "step": 257000 }, { "epoch": 1.0226612891194184, "grad_norm": 0.6314305663108826, "learning_rate": 9.776967941628143e-06, "loss": 1.893710174560547, "step": 257050 }, { "epoch": 1.0228602119522785, "grad_norm": 0.6359471082687378, "learning_rate": 9.774978715278055e-06, "loss": 1.913923797607422, "step": 257100 }, { "epoch": 1.0230591347851385, "grad_norm": 0.6235742568969727, "learning_rate": 9.772989488927967e-06, "loss": 1.9163273620605468, "step": 257150 }, { "epoch": 1.0232580576179986, "grad_norm": 0.6084278225898743, "learning_rate": 9.77100026257788e-06, "loss": 1.8457591247558593, "step": 257200 }, { "epoch": 1.0234569804508586, "grad_norm": 0.7007471323013306, "learning_rate": 9.769011036227792e-06, "loss": 1.8965428161621094, "step": 257250 }, { "epoch": 1.0236559032837187, "grad_norm": 0.681214451789856, "learning_rate": 9.767021809877704e-06, "loss": 1.8593476867675782, "step": 257300 }, { "epoch": 1.0238548261165787, "grad_norm": 0.5158607959747314, "learning_rate": 9.765032583527616e-06, "loss": 1.8619869995117186, "step": 257350 }, { "epoch": 1.0240537489494388, "grad_norm": 0.5832627415657043, "learning_rate": 9.763043357177528e-06, "loss": 1.8979989624023437, "step": 257400 }, { "epoch": 1.0242526717822988, "grad_norm": 0.6694321036338806, "learning_rate": 9.76105413082744e-06, "loss": 1.9225486755371093, "step": 257450 }, { "epoch": 1.024451594615159, "grad_norm": 0.6645919680595398, "learning_rate": 9.75906490447735e-06, "loss": 1.8990182495117187, "step": 257500 }, { "epoch": 1.024650517448019, "grad_norm": 0.7053396701812744, "learning_rate": 9.757075678127263e-06, "loss": 1.8210897827148438, "step": 257550 }, { "epoch": 1.024849440280879, "grad_norm": 0.6404819488525391, "learning_rate": 9.755086451777177e-06, "loss": 1.8640167236328125, "step": 257600 }, { "epoch": 1.025048363113739, "grad_norm": 0.6578997373580933, "learning_rate": 9.753097225427087e-06, "loss": 1.9383010864257812, "step": 257650 }, { "epoch": 1.025247285946599, "grad_norm": 0.6821269392967224, "learning_rate": 9.751107999076999e-06, "loss": 1.8791221618652343, "step": 257700 }, { "epoch": 1.0254462087794591, "grad_norm": 0.5032035708427429, "learning_rate": 9.749118772726911e-06, "loss": 1.9569149780273438, "step": 257750 }, { "epoch": 1.0256451316123192, "grad_norm": 0.61460942029953, "learning_rate": 9.747129546376823e-06, "loss": 1.927760467529297, "step": 257800 }, { "epoch": 1.0258440544451795, "grad_norm": 0.7026063203811646, "learning_rate": 9.745140320026736e-06, "loss": 1.8748060607910155, "step": 257850 }, { "epoch": 1.0260429772780395, "grad_norm": 0.6341108679771423, "learning_rate": 9.743151093676648e-06, "loss": 1.8886819458007813, "step": 257900 }, { "epoch": 1.0262419001108996, "grad_norm": 0.597046971321106, "learning_rate": 9.74116186732656e-06, "loss": 1.8676763916015624, "step": 257950 }, { "epoch": 1.0264408229437596, "grad_norm": 0.6147946715354919, "learning_rate": 9.739172640976472e-06, "loss": 1.8572691345214845, "step": 258000 }, { "epoch": 1.0266397457766196, "grad_norm": 0.6145636439323425, "learning_rate": 9.737183414626384e-06, "loss": 1.8491786193847657, "step": 258050 }, { "epoch": 1.0268386686094797, "grad_norm": 0.6276530027389526, "learning_rate": 9.735194188276296e-06, "loss": 1.8865261840820313, "step": 258100 }, { "epoch": 1.0270375914423397, "grad_norm": 0.7244125604629517, "learning_rate": 9.733204961926208e-06, "loss": 1.9062977600097657, "step": 258150 }, { "epoch": 1.0272365142751998, "grad_norm": 0.7201836109161377, "learning_rate": 9.73121573557612e-06, "loss": 1.9384808349609375, "step": 258200 }, { "epoch": 1.0274354371080598, "grad_norm": 0.507075846195221, "learning_rate": 9.729226509226033e-06, "loss": 1.843745574951172, "step": 258250 }, { "epoch": 1.02763435994092, "grad_norm": 0.641710102558136, "learning_rate": 9.727237282875945e-06, "loss": 1.8405276489257814, "step": 258300 }, { "epoch": 1.02783328277378, "grad_norm": 0.6593363285064697, "learning_rate": 9.725248056525857e-06, "loss": 1.9053070068359375, "step": 258350 }, { "epoch": 1.02803220560664, "grad_norm": 0.7351531386375427, "learning_rate": 9.72325883017577e-06, "loss": 1.8836354064941405, "step": 258400 }, { "epoch": 1.0282311284395, "grad_norm": 0.6183541417121887, "learning_rate": 9.72126960382568e-06, "loss": 1.867131805419922, "step": 258450 }, { "epoch": 1.02843005127236, "grad_norm": 0.5708067417144775, "learning_rate": 9.719280377475593e-06, "loss": 1.8596905517578124, "step": 258500 }, { "epoch": 1.0286289741052201, "grad_norm": 0.5859741568565369, "learning_rate": 9.717291151125506e-06, "loss": 1.8887228393554687, "step": 258550 }, { "epoch": 1.0288278969380802, "grad_norm": 0.5991242527961731, "learning_rate": 9.715301924775418e-06, "loss": 1.8634402465820312, "step": 258600 }, { "epoch": 1.0290268197709405, "grad_norm": 0.5543881058692932, "learning_rate": 9.713352482952331e-06, "loss": 1.883154296875, "step": 258650 }, { "epoch": 1.0292257426038005, "grad_norm": 0.6636466383934021, "learning_rate": 9.711363256602243e-06, "loss": 1.8941946411132813, "step": 258700 }, { "epoch": 1.0294246654366606, "grad_norm": 0.6281542778015137, "learning_rate": 9.709374030252156e-06, "loss": 1.8326919555664063, "step": 258750 }, { "epoch": 1.0296235882695206, "grad_norm": 0.5325491428375244, "learning_rate": 9.707384803902066e-06, "loss": 1.8767198181152345, "step": 258800 }, { "epoch": 1.0298225111023807, "grad_norm": 0.5909613966941833, "learning_rate": 9.70539557755198e-06, "loss": 1.8727146911621093, "step": 258850 }, { "epoch": 1.0300214339352407, "grad_norm": 0.7021273374557495, "learning_rate": 9.703406351201892e-06, "loss": 1.8527052307128906, "step": 258900 }, { "epoch": 1.0302203567681008, "grad_norm": 0.6574035882949829, "learning_rate": 9.701417124851802e-06, "loss": 1.860978546142578, "step": 258950 }, { "epoch": 1.0304192796009608, "grad_norm": 0.5330532193183899, "learning_rate": 9.699427898501715e-06, "loss": 1.8963494873046876, "step": 259000 }, { "epoch": 1.0306182024338209, "grad_norm": 0.7181360125541687, "learning_rate": 9.697438672151628e-06, "loss": 1.8374513244628907, "step": 259050 }, { "epoch": 1.030817125266681, "grad_norm": 0.5183095335960388, "learning_rate": 9.695449445801539e-06, "loss": 1.860401611328125, "step": 259100 }, { "epoch": 1.031016048099541, "grad_norm": 0.5539997816085815, "learning_rate": 9.693460219451451e-06, "loss": 1.8491575622558594, "step": 259150 }, { "epoch": 1.031214970932401, "grad_norm": 0.5846754312515259, "learning_rate": 9.691470993101363e-06, "loss": 1.8662948608398438, "step": 259200 }, { "epoch": 1.031413893765261, "grad_norm": 0.6919736266136169, "learning_rate": 9.689481766751277e-06, "loss": 1.9011042785644532, "step": 259250 }, { "epoch": 1.031612816598121, "grad_norm": 0.6082684397697449, "learning_rate": 9.687492540401187e-06, "loss": 1.87983154296875, "step": 259300 }, { "epoch": 1.0318117394309811, "grad_norm": 0.623401939868927, "learning_rate": 9.6855033140511e-06, "loss": 1.906424560546875, "step": 259350 }, { "epoch": 1.0320106622638412, "grad_norm": 0.6008098721504211, "learning_rate": 9.683514087701012e-06, "loss": 1.8408601379394531, "step": 259400 }, { "epoch": 1.0322095850967015, "grad_norm": 0.6078923940658569, "learning_rate": 9.681524861350924e-06, "loss": 1.8746121215820313, "step": 259450 }, { "epoch": 1.0324085079295615, "grad_norm": 0.6927788853645325, "learning_rate": 9.679535635000836e-06, "loss": 1.8471406555175782, "step": 259500 }, { "epoch": 1.0326074307624216, "grad_norm": 0.6284947395324707, "learning_rate": 9.677546408650748e-06, "loss": 1.8958302307128907, "step": 259550 }, { "epoch": 1.0328063535952816, "grad_norm": 0.7512503266334534, "learning_rate": 9.67555718230066e-06, "loss": 1.8641091918945312, "step": 259600 }, { "epoch": 1.0330052764281417, "grad_norm": 0.7014064192771912, "learning_rate": 9.673567955950572e-06, "loss": 1.9101864624023437, "step": 259650 }, { "epoch": 1.0332041992610017, "grad_norm": 0.5683279633522034, "learning_rate": 9.671578729600485e-06, "loss": 1.8983297729492188, "step": 259700 }, { "epoch": 1.0334031220938618, "grad_norm": 0.5974324941635132, "learning_rate": 9.669589503250397e-06, "loss": 1.9009919738769532, "step": 259750 }, { "epoch": 1.0336020449267218, "grad_norm": 0.6620243787765503, "learning_rate": 9.66764006142731e-06, "loss": 1.8480413818359376, "step": 259800 }, { "epoch": 1.0338009677595819, "grad_norm": 0.6835086941719055, "learning_rate": 9.665650835077222e-06, "loss": 1.874827880859375, "step": 259850 }, { "epoch": 1.033999890592442, "grad_norm": 0.7015897035598755, "learning_rate": 9.663661608727135e-06, "loss": 1.9663484191894531, "step": 259900 }, { "epoch": 1.034198813425302, "grad_norm": 0.6043884754180908, "learning_rate": 9.661672382377047e-06, "loss": 1.8845654296875, "step": 259950 }, { "epoch": 1.034397736258162, "grad_norm": 0.6236907839775085, "learning_rate": 9.659683156026959e-06, "loss": 1.862676239013672, "step": 260000 }, { "epoch": 1.034596659091022, "grad_norm": 0.524391770362854, "learning_rate": 9.657693929676871e-06, "loss": 1.8847653198242187, "step": 260050 }, { "epoch": 1.034795581923882, "grad_norm": 0.6829896569252014, "learning_rate": 9.655704703326783e-06, "loss": 1.8137904357910157, "step": 260100 }, { "epoch": 1.0349945047567422, "grad_norm": 0.6015082001686096, "learning_rate": 9.653715476976695e-06, "loss": 1.8369068908691406, "step": 260150 }, { "epoch": 1.0351934275896022, "grad_norm": 0.5970872044563293, "learning_rate": 9.651726250626607e-06, "loss": 1.8617063903808593, "step": 260200 }, { "epoch": 1.0353923504224625, "grad_norm": 0.6283843517303467, "learning_rate": 9.649737024276518e-06, "loss": 1.8536799621582032, "step": 260250 }, { "epoch": 1.0355912732553225, "grad_norm": 0.5480306148529053, "learning_rate": 9.647747797926432e-06, "loss": 1.8551432800292968, "step": 260300 }, { "epoch": 1.0357901960881826, "grad_norm": 0.4847591519355774, "learning_rate": 9.645758571576344e-06, "loss": 1.8906094360351562, "step": 260350 }, { "epoch": 1.0359891189210426, "grad_norm": 0.6431349515914917, "learning_rate": 9.643769345226254e-06, "loss": 1.8738662719726562, "step": 260400 }, { "epoch": 1.0361880417539027, "grad_norm": 0.5708069205284119, "learning_rate": 9.641780118876166e-06, "loss": 1.9238229370117188, "step": 260450 }, { "epoch": 1.0363869645867627, "grad_norm": 0.6640785932540894, "learning_rate": 9.63979089252608e-06, "loss": 1.833105010986328, "step": 260500 }, { "epoch": 1.0365858874196228, "grad_norm": 0.7465188503265381, "learning_rate": 9.637801666175992e-06, "loss": 1.8642556762695313, "step": 260550 }, { "epoch": 1.0367848102524828, "grad_norm": 0.6559619307518005, "learning_rate": 9.635812439825903e-06, "loss": 1.934838104248047, "step": 260600 }, { "epoch": 1.0369837330853429, "grad_norm": 0.5223390460014343, "learning_rate": 9.633823213475815e-06, "loss": 1.8778968811035157, "step": 260650 }, { "epoch": 1.037182655918203, "grad_norm": 0.5808517932891846, "learning_rate": 9.631833987125729e-06, "loss": 1.8996585083007813, "step": 260700 }, { "epoch": 1.037381578751063, "grad_norm": 0.6422610282897949, "learning_rate": 9.62984476077564e-06, "loss": 1.8517823791503907, "step": 260750 }, { "epoch": 1.037580501583923, "grad_norm": 0.6262533664703369, "learning_rate": 9.627855534425551e-06, "loss": 1.782093505859375, "step": 260800 }, { "epoch": 1.037779424416783, "grad_norm": 0.6626811027526855, "learning_rate": 9.625866308075464e-06, "loss": 1.9283137512207031, "step": 260850 }, { "epoch": 1.037978347249643, "grad_norm": 0.5263513326644897, "learning_rate": 9.623877081725376e-06, "loss": 1.829597930908203, "step": 260900 }, { "epoch": 1.0381772700825032, "grad_norm": 0.42281270027160645, "learning_rate": 9.621887855375288e-06, "loss": 1.8536158752441407, "step": 260950 }, { "epoch": 1.0383761929153632, "grad_norm": 0.7298934459686279, "learning_rate": 9.6198986290252e-06, "loss": 1.9061985778808594, "step": 261000 }, { "epoch": 1.0385751157482235, "grad_norm": 0.6010802984237671, "learning_rate": 9.617909402675112e-06, "loss": 1.8600108337402343, "step": 261050 }, { "epoch": 1.0387740385810835, "grad_norm": 0.5786093473434448, "learning_rate": 9.615920176325024e-06, "loss": 1.8821290588378907, "step": 261100 }, { "epoch": 1.0389729614139436, "grad_norm": 0.6123414039611816, "learning_rate": 9.613930949974937e-06, "loss": 1.8455352783203125, "step": 261150 }, { "epoch": 1.0391718842468036, "grad_norm": 0.6175932884216309, "learning_rate": 9.611941723624849e-06, "loss": 1.8241085815429687, "step": 261200 }, { "epoch": 1.0393708070796637, "grad_norm": 0.628718376159668, "learning_rate": 9.609992281801762e-06, "loss": 1.8182077026367187, "step": 261250 }, { "epoch": 1.0395697299125237, "grad_norm": 0.6529210209846497, "learning_rate": 9.608003055451674e-06, "loss": 1.8353369140625, "step": 261300 }, { "epoch": 1.0397686527453838, "grad_norm": 0.6513029932975769, "learning_rate": 9.606013829101586e-06, "loss": 1.9248509216308594, "step": 261350 }, { "epoch": 1.0399675755782438, "grad_norm": 0.6878504157066345, "learning_rate": 9.604024602751499e-06, "loss": 1.907515411376953, "step": 261400 }, { "epoch": 1.0401664984111039, "grad_norm": 0.521683931350708, "learning_rate": 9.60203537640141e-06, "loss": 1.8893211364746094, "step": 261450 }, { "epoch": 1.040365421243964, "grad_norm": 0.605110228061676, "learning_rate": 9.600046150051323e-06, "loss": 1.8579322814941406, "step": 261500 }, { "epoch": 1.040564344076824, "grad_norm": 0.7203890085220337, "learning_rate": 9.598056923701235e-06, "loss": 1.863970947265625, "step": 261550 }, { "epoch": 1.040763266909684, "grad_norm": 0.6502427458763123, "learning_rate": 9.596067697351147e-06, "loss": 1.9101947021484376, "step": 261600 }, { "epoch": 1.040962189742544, "grad_norm": 0.6471958756446838, "learning_rate": 9.59407847100106e-06, "loss": 1.8900592041015625, "step": 261650 }, { "epoch": 1.0411611125754041, "grad_norm": 0.545592188835144, "learning_rate": 9.59208924465097e-06, "loss": 1.8224896240234374, "step": 261700 }, { "epoch": 1.0413600354082642, "grad_norm": 0.7195369601249695, "learning_rate": 9.590100018300884e-06, "loss": 1.8624359130859376, "step": 261750 }, { "epoch": 1.0415589582411242, "grad_norm": 0.6180253028869629, "learning_rate": 9.588110791950796e-06, "loss": 1.8375921630859375, "step": 261800 }, { "epoch": 1.0417578810739845, "grad_norm": 0.6671930551528931, "learning_rate": 9.586121565600708e-06, "loss": 1.8808790588378905, "step": 261850 }, { "epoch": 1.0419568039068445, "grad_norm": 0.6489983797073364, "learning_rate": 9.584132339250618e-06, "loss": 1.8799461364746093, "step": 261900 }, { "epoch": 1.0421557267397046, "grad_norm": 0.6603778004646301, "learning_rate": 9.582143112900532e-06, "loss": 1.954881591796875, "step": 261950 }, { "epoch": 1.0423546495725646, "grad_norm": 0.7068319320678711, "learning_rate": 9.580153886550444e-06, "loss": 1.8450994873046875, "step": 262000 }, { "epoch": 1.0425535724054247, "grad_norm": 0.5921096801757812, "learning_rate": 9.578164660200355e-06, "loss": 1.9493905639648437, "step": 262050 }, { "epoch": 1.0427524952382847, "grad_norm": 0.5533425211906433, "learning_rate": 9.576175433850267e-06, "loss": 1.8593856811523437, "step": 262100 }, { "epoch": 1.0429514180711448, "grad_norm": 0.35249966382980347, "learning_rate": 9.57418620750018e-06, "loss": 1.8588339233398437, "step": 262150 }, { "epoch": 1.0431503409040048, "grad_norm": 0.7198031544685364, "learning_rate": 9.572196981150091e-06, "loss": 1.9350636291503907, "step": 262200 }, { "epoch": 1.0433492637368649, "grad_norm": 0.5793904066085815, "learning_rate": 9.570207754800003e-06, "loss": 1.8702484130859376, "step": 262250 }, { "epoch": 1.043548186569725, "grad_norm": 0.5372281670570374, "learning_rate": 9.568218528449916e-06, "loss": 1.8869432067871095, "step": 262300 }, { "epoch": 1.043747109402585, "grad_norm": 0.6973446011543274, "learning_rate": 9.566229302099828e-06, "loss": 1.853865966796875, "step": 262350 }, { "epoch": 1.043946032235445, "grad_norm": 0.6235744953155518, "learning_rate": 9.56424007574974e-06, "loss": 1.8887113952636718, "step": 262400 }, { "epoch": 1.044144955068305, "grad_norm": 0.5945005416870117, "learning_rate": 9.562250849399652e-06, "loss": 1.8474716186523437, "step": 262450 }, { "epoch": 1.0443438779011651, "grad_norm": 0.5485174655914307, "learning_rate": 9.560261623049564e-06, "loss": 1.94450439453125, "step": 262500 }, { "epoch": 1.0445428007340252, "grad_norm": 0.6896798610687256, "learning_rate": 9.558272396699476e-06, "loss": 1.815343475341797, "step": 262550 }, { "epoch": 1.0447417235668852, "grad_norm": 0.6991839408874512, "learning_rate": 9.556283170349388e-06, "loss": 1.9014129638671875, "step": 262600 }, { "epoch": 1.0449406463997453, "grad_norm": 0.4804098606109619, "learning_rate": 9.5542939439993e-06, "loss": 1.8364962768554687, "step": 262650 }, { "epoch": 1.0451395692326055, "grad_norm": 0.6059601306915283, "learning_rate": 9.552304717649213e-06, "loss": 1.8820654296875, "step": 262700 }, { "epoch": 1.0453384920654656, "grad_norm": 0.5038729310035706, "learning_rate": 9.550315491299125e-06, "loss": 1.89551025390625, "step": 262750 }, { "epoch": 1.0455374148983256, "grad_norm": 0.5393328666687012, "learning_rate": 9.548326264949037e-06, "loss": 1.8857843017578124, "step": 262800 }, { "epoch": 1.0457363377311857, "grad_norm": 0.5619882345199585, "learning_rate": 9.546337038598949e-06, "loss": 1.9056781005859376, "step": 262850 }, { "epoch": 1.0459352605640457, "grad_norm": 0.5239191651344299, "learning_rate": 9.544347812248861e-06, "loss": 1.88131103515625, "step": 262900 }, { "epoch": 1.0461341833969058, "grad_norm": 0.5753674507141113, "learning_rate": 9.542358585898773e-06, "loss": 1.8695494079589843, "step": 262950 }, { "epoch": 1.0463331062297658, "grad_norm": 0.5799781084060669, "learning_rate": 9.540369359548684e-06, "loss": 1.8604440307617187, "step": 263000 }, { "epoch": 1.0465320290626259, "grad_norm": 0.7709992527961731, "learning_rate": 9.538380133198598e-06, "loss": 1.8377479553222655, "step": 263050 }, { "epoch": 1.046730951895486, "grad_norm": 0.6588295698165894, "learning_rate": 9.53639090684851e-06, "loss": 1.914293975830078, "step": 263100 }, { "epoch": 1.046929874728346, "grad_norm": 0.631282389163971, "learning_rate": 9.534401680498422e-06, "loss": 1.8826754760742188, "step": 263150 }, { "epoch": 1.047128797561206, "grad_norm": 0.6383616924285889, "learning_rate": 9.532412454148332e-06, "loss": 1.8883631896972657, "step": 263200 }, { "epoch": 1.047327720394066, "grad_norm": 0.650747537612915, "learning_rate": 9.530423227798246e-06, "loss": 1.857861328125, "step": 263250 }, { "epoch": 1.0475266432269261, "grad_norm": 0.689197301864624, "learning_rate": 9.528434001448158e-06, "loss": 1.8350485229492188, "step": 263300 }, { "epoch": 1.0477255660597862, "grad_norm": 0.6331049203872681, "learning_rate": 9.526444775098069e-06, "loss": 1.8799798583984375, "step": 263350 }, { "epoch": 1.0479244888926462, "grad_norm": 0.48363184928894043, "learning_rate": 9.524455548747981e-06, "loss": 1.8581239318847655, "step": 263400 }, { "epoch": 1.0481234117255065, "grad_norm": 0.6738073229789734, "learning_rate": 9.522466322397895e-06, "loss": 1.8692526245117187, "step": 263450 }, { "epoch": 1.0483223345583665, "grad_norm": 0.568525493144989, "learning_rate": 9.520477096047805e-06, "loss": 1.8424604797363282, "step": 263500 }, { "epoch": 1.0485212573912266, "grad_norm": 0.6035695672035217, "learning_rate": 9.518487869697717e-06, "loss": 1.8726716613769532, "step": 263550 }, { "epoch": 1.0487201802240866, "grad_norm": 0.6807388663291931, "learning_rate": 9.51649864334763e-06, "loss": 1.8672163391113281, "step": 263600 }, { "epoch": 1.0489191030569467, "grad_norm": 0.585543155670166, "learning_rate": 9.514509416997542e-06, "loss": 1.8938174438476563, "step": 263650 }, { "epoch": 1.0491180258898067, "grad_norm": 0.4952694773674011, "learning_rate": 9.512520190647454e-06, "loss": 1.9022499084472657, "step": 263700 }, { "epoch": 1.0493169487226668, "grad_norm": 0.7038636803627014, "learning_rate": 9.510530964297366e-06, "loss": 1.8468365478515625, "step": 263750 }, { "epoch": 1.0495158715555268, "grad_norm": 0.6454856395721436, "learning_rate": 9.508581522474281e-06, "loss": 1.8886021423339843, "step": 263800 }, { "epoch": 1.0497147943883869, "grad_norm": 0.6268760561943054, "learning_rate": 9.506592296124192e-06, "loss": 1.8891110229492187, "step": 263850 }, { "epoch": 1.049913717221247, "grad_norm": 0.5634494423866272, "learning_rate": 9.504603069774104e-06, "loss": 1.8823629760742187, "step": 263900 }, { "epoch": 1.050112640054107, "grad_norm": 0.7149844169616699, "learning_rate": 9.502613843424016e-06, "loss": 1.8664120483398436, "step": 263950 }, { "epoch": 1.050311562886967, "grad_norm": 0.651964545249939, "learning_rate": 9.500624617073928e-06, "loss": 1.867458038330078, "step": 264000 }, { "epoch": 1.050510485719827, "grad_norm": 0.6484234929084778, "learning_rate": 9.49863539072384e-06, "loss": 1.8521803283691407, "step": 264050 }, { "epoch": 1.0507094085526871, "grad_norm": 0.6410173773765564, "learning_rate": 9.496646164373752e-06, "loss": 1.8703036499023438, "step": 264100 }, { "epoch": 1.0509083313855472, "grad_norm": 0.662363588809967, "learning_rate": 9.494656938023665e-06, "loss": 1.8561386108398437, "step": 264150 }, { "epoch": 1.0511072542184072, "grad_norm": 0.7751960158348083, "learning_rate": 9.492667711673577e-06, "loss": 1.920080108642578, "step": 264200 }, { "epoch": 1.0513061770512673, "grad_norm": 0.6252279281616211, "learning_rate": 9.490678485323489e-06, "loss": 1.8815374755859375, "step": 264250 }, { "epoch": 1.0515050998841275, "grad_norm": 0.5433325171470642, "learning_rate": 9.488689258973401e-06, "loss": 1.8473121643066406, "step": 264300 }, { "epoch": 1.0517040227169876, "grad_norm": 0.5888289213180542, "learning_rate": 9.486700032623313e-06, "loss": 1.8852064514160156, "step": 264350 }, { "epoch": 1.0519029455498476, "grad_norm": 0.6340243220329285, "learning_rate": 9.484710806273225e-06, "loss": 1.8693482971191406, "step": 264400 }, { "epoch": 1.0521018683827077, "grad_norm": 0.6706527471542358, "learning_rate": 9.482721579923137e-06, "loss": 1.9300204467773439, "step": 264450 }, { "epoch": 1.0523007912155677, "grad_norm": 0.5415666103363037, "learning_rate": 9.48073235357305e-06, "loss": 1.9043295288085937, "step": 264500 }, { "epoch": 1.0524997140484278, "grad_norm": 0.6219034194946289, "learning_rate": 9.478743127222962e-06, "loss": 1.8548638916015625, "step": 264550 }, { "epoch": 1.0526986368812878, "grad_norm": 0.5688084959983826, "learning_rate": 9.476753900872874e-06, "loss": 1.8459353637695313, "step": 264600 }, { "epoch": 1.052897559714148, "grad_norm": 0.5908061861991882, "learning_rate": 9.474764674522784e-06, "loss": 1.8222459411621095, "step": 264650 }, { "epoch": 1.053096482547008, "grad_norm": 0.6849122643470764, "learning_rate": 9.472775448172698e-06, "loss": 1.8346090698242188, "step": 264700 }, { "epoch": 1.053295405379868, "grad_norm": 0.4396716356277466, "learning_rate": 9.47078622182261e-06, "loss": 1.8426675415039062, "step": 264750 }, { "epoch": 1.053494328212728, "grad_norm": 0.5771105289459229, "learning_rate": 9.46879699547252e-06, "loss": 1.927616729736328, "step": 264800 }, { "epoch": 1.053693251045588, "grad_norm": 0.7041809558868408, "learning_rate": 9.466807769122433e-06, "loss": 1.8961770629882813, "step": 264850 }, { "epoch": 1.0538921738784481, "grad_norm": 0.6075960993766785, "learning_rate": 9.464818542772347e-06, "loss": 1.8667286682128905, "step": 264900 }, { "epoch": 1.0540910967113082, "grad_norm": 0.4222847819328308, "learning_rate": 9.462829316422257e-06, "loss": 1.8898463439941406, "step": 264950 }, { "epoch": 1.0542900195441682, "grad_norm": 0.6952036619186401, "learning_rate": 9.46084009007217e-06, "loss": 1.8725468444824218, "step": 265000 }, { "epoch": 1.0544889423770285, "grad_norm": 0.7494192123413086, "learning_rate": 9.458850863722082e-06, "loss": 1.8563240051269532, "step": 265050 }, { "epoch": 1.0546878652098886, "grad_norm": 0.6566715240478516, "learning_rate": 9.456861637371995e-06, "loss": 1.8615177917480468, "step": 265100 }, { "epoch": 1.0548867880427486, "grad_norm": 0.5887587666511536, "learning_rate": 9.454872411021906e-06, "loss": 1.9117372131347656, "step": 265150 }, { "epoch": 1.0550857108756087, "grad_norm": 0.5685933828353882, "learning_rate": 9.452883184671818e-06, "loss": 1.9123858642578124, "step": 265200 }, { "epoch": 1.0552846337084687, "grad_norm": 0.6770960688591003, "learning_rate": 9.45089395832173e-06, "loss": 1.8404962158203124, "step": 265250 }, { "epoch": 1.0554835565413287, "grad_norm": 0.6490153670310974, "learning_rate": 9.448904731971642e-06, "loss": 1.8600569152832032, "step": 265300 }, { "epoch": 1.0556824793741888, "grad_norm": 0.6299229264259338, "learning_rate": 9.446915505621554e-06, "loss": 1.8888406372070312, "step": 265350 }, { "epoch": 1.0558814022070488, "grad_norm": 0.6196044683456421, "learning_rate": 9.444926279271467e-06, "loss": 1.8889834594726562, "step": 265400 }, { "epoch": 1.056080325039909, "grad_norm": 0.7024624347686768, "learning_rate": 9.442937052921379e-06, "loss": 1.9108087158203124, "step": 265450 }, { "epoch": 1.056279247872769, "grad_norm": 0.5954587459564209, "learning_rate": 9.440947826571291e-06, "loss": 1.86869384765625, "step": 265500 }, { "epoch": 1.056478170705629, "grad_norm": 0.5181295871734619, "learning_rate": 9.438958600221203e-06, "loss": 1.8925869750976563, "step": 265550 }, { "epoch": 1.056677093538489, "grad_norm": 0.6212870478630066, "learning_rate": 9.436969373871115e-06, "loss": 1.8807272338867187, "step": 265600 }, { "epoch": 1.056876016371349, "grad_norm": 0.6538793444633484, "learning_rate": 9.434980147521027e-06, "loss": 1.8573257446289062, "step": 265650 }, { "epoch": 1.0570749392042091, "grad_norm": 0.5590440630912781, "learning_rate": 9.43299092117094e-06, "loss": 1.9407464599609374, "step": 265700 }, { "epoch": 1.0572738620370692, "grad_norm": 0.5655008554458618, "learning_rate": 9.431001694820852e-06, "loss": 1.8643595886230468, "step": 265750 }, { "epoch": 1.0574727848699292, "grad_norm": 0.7182729244232178, "learning_rate": 9.429012468470764e-06, "loss": 1.8731797790527345, "step": 265800 }, { "epoch": 1.0576717077027893, "grad_norm": 0.7361655831336975, "learning_rate": 9.427023242120676e-06, "loss": 1.865340576171875, "step": 265850 }, { "epoch": 1.0578706305356496, "grad_norm": 0.7162418365478516, "learning_rate": 9.425034015770588e-06, "loss": 1.860111083984375, "step": 265900 }, { "epoch": 1.0580695533685096, "grad_norm": 0.7416098713874817, "learning_rate": 9.423044789420498e-06, "loss": 1.8711544799804687, "step": 265950 }, { "epoch": 1.0582684762013697, "grad_norm": 0.7443413734436035, "learning_rate": 9.421055563070412e-06, "loss": 1.853267364501953, "step": 266000 }, { "epoch": 1.0584673990342297, "grad_norm": 0.6337122917175293, "learning_rate": 9.419066336720324e-06, "loss": 1.8652471923828124, "step": 266050 }, { "epoch": 1.0586663218670898, "grad_norm": 0.5833685994148254, "learning_rate": 9.417077110370235e-06, "loss": 1.8913813781738282, "step": 266100 }, { "epoch": 1.0588652446999498, "grad_norm": 0.7319800853729248, "learning_rate": 9.415087884020147e-06, "loss": 1.9135328674316405, "step": 266150 }, { "epoch": 1.0590641675328099, "grad_norm": 0.4789610803127289, "learning_rate": 9.41309865767006e-06, "loss": 1.84915771484375, "step": 266200 }, { "epoch": 1.05926309036567, "grad_norm": 0.5732915997505188, "learning_rate": 9.411109431319971e-06, "loss": 1.837672576904297, "step": 266250 }, { "epoch": 1.05946201319853, "grad_norm": 0.7027587890625, "learning_rate": 9.409120204969883e-06, "loss": 1.8206956481933594, "step": 266300 }, { "epoch": 1.05966093603139, "grad_norm": 0.6365215182304382, "learning_rate": 9.407130978619796e-06, "loss": 1.8746185302734375, "step": 266350 }, { "epoch": 1.05985985886425, "grad_norm": 0.5829334855079651, "learning_rate": 9.405141752269708e-06, "loss": 1.9599501037597655, "step": 266400 }, { "epoch": 1.06005878169711, "grad_norm": 0.6613931059837341, "learning_rate": 9.40315252591962e-06, "loss": 1.9066015625, "step": 266450 }, { "epoch": 1.0602577045299701, "grad_norm": 0.6348958611488342, "learning_rate": 9.401163299569532e-06, "loss": 1.88394775390625, "step": 266500 }, { "epoch": 1.0604566273628302, "grad_norm": 0.6007611751556396, "learning_rate": 9.399174073219444e-06, "loss": 1.8814988708496094, "step": 266550 }, { "epoch": 1.0606555501956902, "grad_norm": 0.580308198928833, "learning_rate": 9.397184846869356e-06, "loss": 1.8716061401367188, "step": 266600 }, { "epoch": 1.0608544730285503, "grad_norm": 0.5624076724052429, "learning_rate": 9.395195620519269e-06, "loss": 1.8424232482910157, "step": 266650 }, { "epoch": 1.0610533958614106, "grad_norm": 0.5428366661071777, "learning_rate": 9.39320639416918e-06, "loss": 1.867242889404297, "step": 266700 }, { "epoch": 1.0612523186942706, "grad_norm": 0.5674212574958801, "learning_rate": 9.391217167819093e-06, "loss": 1.8529629516601562, "step": 266750 }, { "epoch": 1.0614512415271307, "grad_norm": 0.6194356083869934, "learning_rate": 9.389227941469005e-06, "loss": 1.8351373291015625, "step": 266800 }, { "epoch": 1.0616501643599907, "grad_norm": 0.6059789657592773, "learning_rate": 9.387238715118917e-06, "loss": 1.8477204895019532, "step": 266850 }, { "epoch": 1.0618490871928508, "grad_norm": 0.5394636988639832, "learning_rate": 9.385249488768828e-06, "loss": 1.9031150817871094, "step": 266900 }, { "epoch": 1.0620480100257108, "grad_norm": 0.675679624080658, "learning_rate": 9.383260262418741e-06, "loss": 1.8910929870605468, "step": 266950 }, { "epoch": 1.0622469328585709, "grad_norm": 0.6541621088981628, "learning_rate": 9.381271036068654e-06, "loss": 1.8796199035644532, "step": 267000 }, { "epoch": 1.062445855691431, "grad_norm": 0.6065433621406555, "learning_rate": 9.379281809718564e-06, "loss": 1.89404296875, "step": 267050 }, { "epoch": 1.062644778524291, "grad_norm": 0.6268543601036072, "learning_rate": 9.377292583368476e-06, "loss": 1.8920953369140625, "step": 267100 }, { "epoch": 1.062843701357151, "grad_norm": 0.5514017343521118, "learning_rate": 9.37530335701839e-06, "loss": 1.835260009765625, "step": 267150 }, { "epoch": 1.063042624190011, "grad_norm": 0.7439023852348328, "learning_rate": 9.373314130668302e-06, "loss": 1.817509765625, "step": 267200 }, { "epoch": 1.063241547022871, "grad_norm": 0.6218692660331726, "learning_rate": 9.371324904318213e-06, "loss": 1.8783116149902344, "step": 267250 }, { "epoch": 1.0634404698557312, "grad_norm": 0.5950149297714233, "learning_rate": 9.369335677968125e-06, "loss": 1.8305401611328125, "step": 267300 }, { "epoch": 1.0636393926885912, "grad_norm": 0.5860691070556641, "learning_rate": 9.367346451618039e-06, "loss": 1.8901319885253907, "step": 267350 }, { "epoch": 1.0638383155214512, "grad_norm": 0.6263824701309204, "learning_rate": 9.365357225267949e-06, "loss": 1.870574951171875, "step": 267400 }, { "epoch": 1.0640372383543113, "grad_norm": 0.6274541616439819, "learning_rate": 9.363367998917861e-06, "loss": 1.890783233642578, "step": 267450 }, { "epoch": 1.0642361611871716, "grad_norm": 0.5152260065078735, "learning_rate": 9.361378772567773e-06, "loss": 1.8745797729492188, "step": 267500 }, { "epoch": 1.0644350840200316, "grad_norm": 0.5632141828536987, "learning_rate": 9.359389546217685e-06, "loss": 1.8313015747070311, "step": 267550 }, { "epoch": 1.0646340068528917, "grad_norm": 0.6926330327987671, "learning_rate": 9.357400319867598e-06, "loss": 1.851780242919922, "step": 267600 }, { "epoch": 1.0648329296857517, "grad_norm": 0.6184064149856567, "learning_rate": 9.35541109351751e-06, "loss": 1.909456787109375, "step": 267650 }, { "epoch": 1.0650318525186118, "grad_norm": 0.641886830329895, "learning_rate": 9.353421867167422e-06, "loss": 1.8809385681152344, "step": 267700 }, { "epoch": 1.0652307753514718, "grad_norm": 0.5994720458984375, "learning_rate": 9.351432640817334e-06, "loss": 1.8841299438476562, "step": 267750 }, { "epoch": 1.0654296981843319, "grad_norm": 0.6329610347747803, "learning_rate": 9.349443414467246e-06, "loss": 1.86563232421875, "step": 267800 }, { "epoch": 1.065628621017192, "grad_norm": 0.6877838969230652, "learning_rate": 9.347454188117158e-06, "loss": 1.8430026245117188, "step": 267850 }, { "epoch": 1.065827543850052, "grad_norm": 0.5922356843948364, "learning_rate": 9.34546496176707e-06, "loss": 1.8349441528320312, "step": 267900 }, { "epoch": 1.066026466682912, "grad_norm": 0.6055331230163574, "learning_rate": 9.343475735416983e-06, "loss": 1.9200869750976564, "step": 267950 }, { "epoch": 1.066225389515772, "grad_norm": 0.6393352746963501, "learning_rate": 9.341486509066895e-06, "loss": 1.8639056396484375, "step": 268000 }, { "epoch": 1.066424312348632, "grad_norm": 0.6428718566894531, "learning_rate": 9.339537067243808e-06, "loss": 1.8377525329589843, "step": 268050 }, { "epoch": 1.0666232351814922, "grad_norm": 0.6535479426383972, "learning_rate": 9.33754784089372e-06, "loss": 1.9113967895507813, "step": 268100 }, { "epoch": 1.0668221580143522, "grad_norm": 0.6193712949752808, "learning_rate": 9.335558614543633e-06, "loss": 1.8878929138183593, "step": 268150 }, { "epoch": 1.0670210808472123, "grad_norm": 0.587677538394928, "learning_rate": 9.333569388193545e-06, "loss": 1.8403662109375, "step": 268200 }, { "epoch": 1.0672200036800723, "grad_norm": 0.6768253445625305, "learning_rate": 9.331580161843457e-06, "loss": 1.875331573486328, "step": 268250 }, { "epoch": 1.0674189265129326, "grad_norm": 0.6257195472717285, "learning_rate": 9.329590935493369e-06, "loss": 1.8849887084960937, "step": 268300 }, { "epoch": 1.0676178493457926, "grad_norm": 0.6509926319122314, "learning_rate": 9.327641493670282e-06, "loss": 1.8565574645996095, "step": 268350 }, { "epoch": 1.0678167721786527, "grad_norm": 0.558862030506134, "learning_rate": 9.325652267320195e-06, "loss": 1.884517822265625, "step": 268400 }, { "epoch": 1.0680156950115127, "grad_norm": 0.6795443892478943, "learning_rate": 9.323663040970107e-06, "loss": 1.850327911376953, "step": 268450 }, { "epoch": 1.0682146178443728, "grad_norm": 0.6557990908622742, "learning_rate": 9.321673814620019e-06, "loss": 1.9090376281738282, "step": 268500 }, { "epoch": 1.0684135406772328, "grad_norm": 0.5950473546981812, "learning_rate": 9.319684588269931e-06, "loss": 1.8551162719726562, "step": 268550 }, { "epoch": 1.0686124635100929, "grad_norm": 0.6147762537002563, "learning_rate": 9.317695361919843e-06, "loss": 1.8956448364257812, "step": 268600 }, { "epoch": 1.068811386342953, "grad_norm": 0.7363340258598328, "learning_rate": 9.315706135569755e-06, "loss": 1.8560765075683594, "step": 268650 }, { "epoch": 1.069010309175813, "grad_norm": 0.6923497319221497, "learning_rate": 9.313716909219666e-06, "loss": 1.9283355712890624, "step": 268700 }, { "epoch": 1.069209232008673, "grad_norm": 0.689206063747406, "learning_rate": 9.31172768286958e-06, "loss": 1.8750993347167968, "step": 268750 }, { "epoch": 1.069408154841533, "grad_norm": 0.7170327305793762, "learning_rate": 9.309738456519492e-06, "loss": 1.8442167663574218, "step": 268800 }, { "epoch": 1.0696070776743931, "grad_norm": 0.6501282453536987, "learning_rate": 9.307749230169402e-06, "loss": 1.8912986755371093, "step": 268850 }, { "epoch": 1.0698060005072532, "grad_norm": 0.6949161887168884, "learning_rate": 9.305760003819314e-06, "loss": 1.8390809631347655, "step": 268900 }, { "epoch": 1.0700049233401132, "grad_norm": 0.6791054606437683, "learning_rate": 9.303770777469228e-06, "loss": 1.877064208984375, "step": 268950 }, { "epoch": 1.0702038461729733, "grad_norm": 0.6764287352561951, "learning_rate": 9.30178155111914e-06, "loss": 1.8855059814453126, "step": 269000 }, { "epoch": 1.0704027690058333, "grad_norm": 0.6893775463104248, "learning_rate": 9.299792324769051e-06, "loss": 1.9432833862304688, "step": 269050 }, { "epoch": 1.0706016918386934, "grad_norm": 0.6718994379043579, "learning_rate": 9.297803098418963e-06, "loss": 1.8464599609375, "step": 269100 }, { "epoch": 1.0708006146715536, "grad_norm": 0.7483124732971191, "learning_rate": 9.295813872068877e-06, "loss": 1.91895751953125, "step": 269150 }, { "epoch": 1.0709995375044137, "grad_norm": 0.5344007611274719, "learning_rate": 9.293824645718787e-06, "loss": 1.89832763671875, "step": 269200 }, { "epoch": 1.0711984603372737, "grad_norm": 0.7723394632339478, "learning_rate": 9.2918354193687e-06, "loss": 1.8924528503417968, "step": 269250 }, { "epoch": 1.0713973831701338, "grad_norm": 0.7245182394981384, "learning_rate": 9.289846193018612e-06, "loss": 1.902819366455078, "step": 269300 }, { "epoch": 1.0715963060029938, "grad_norm": 0.6273779273033142, "learning_rate": 9.287856966668524e-06, "loss": 1.928563995361328, "step": 269350 }, { "epoch": 1.0717952288358539, "grad_norm": 0.6647968292236328, "learning_rate": 9.285867740318436e-06, "loss": 1.8980628967285156, "step": 269400 }, { "epoch": 1.071994151668714, "grad_norm": 0.6018015742301941, "learning_rate": 9.283878513968348e-06, "loss": 1.8665359497070313, "step": 269450 }, { "epoch": 1.072193074501574, "grad_norm": 0.6247149109840393, "learning_rate": 9.28188928761826e-06, "loss": 1.87740966796875, "step": 269500 }, { "epoch": 1.072391997334434, "grad_norm": 0.6755090355873108, "learning_rate": 9.279900061268172e-06, "loss": 1.8871389770507812, "step": 269550 }, { "epoch": 1.072590920167294, "grad_norm": 0.6195138692855835, "learning_rate": 9.277910834918084e-06, "loss": 1.8815512084960937, "step": 269600 }, { "epoch": 1.0727898430001541, "grad_norm": 0.574266791343689, "learning_rate": 9.275921608567997e-06, "loss": 1.89640625, "step": 269650 }, { "epoch": 1.0729887658330142, "grad_norm": 0.604550302028656, "learning_rate": 9.273932382217909e-06, "loss": 1.8563735961914063, "step": 269700 }, { "epoch": 1.0731876886658742, "grad_norm": 0.67275470495224, "learning_rate": 9.271982940394822e-06, "loss": 1.92895263671875, "step": 269750 }, { "epoch": 1.0733866114987343, "grad_norm": 0.6321632266044617, "learning_rate": 9.269993714044734e-06, "loss": 1.889093017578125, "step": 269800 }, { "epoch": 1.0735855343315943, "grad_norm": 0.6153204441070557, "learning_rate": 9.268004487694647e-06, "loss": 1.844456024169922, "step": 269850 }, { "epoch": 1.0737844571644546, "grad_norm": 0.6166037321090698, "learning_rate": 9.266015261344559e-06, "loss": 1.9122280883789062, "step": 269900 }, { "epoch": 1.0739833799973146, "grad_norm": 0.613170862197876, "learning_rate": 9.26402603499447e-06, "loss": 1.8950270080566407, "step": 269950 }, { "epoch": 1.0741823028301747, "grad_norm": 0.5967901349067688, "learning_rate": 9.262036808644383e-06, "loss": 1.8827897644042968, "step": 270000 }, { "epoch": 1.0743812256630347, "grad_norm": 0.6425639986991882, "learning_rate": 9.260047582294295e-06, "loss": 1.8577304077148438, "step": 270050 }, { "epoch": 1.0745801484958948, "grad_norm": 0.632366418838501, "learning_rate": 9.258058355944207e-06, "loss": 1.832289581298828, "step": 270100 }, { "epoch": 1.0747790713287548, "grad_norm": 0.6946215033531189, "learning_rate": 9.256069129594118e-06, "loss": 1.8083085632324218, "step": 270150 }, { "epoch": 1.0749779941616149, "grad_norm": 0.558143675327301, "learning_rate": 9.254079903244032e-06, "loss": 1.8153231811523438, "step": 270200 }, { "epoch": 1.075176916994475, "grad_norm": 0.623989462852478, "learning_rate": 9.252090676893944e-06, "loss": 1.9144880676269531, "step": 270250 }, { "epoch": 1.075375839827335, "grad_norm": 0.7715136408805847, "learning_rate": 9.250101450543856e-06, "loss": 1.9031687927246095, "step": 270300 }, { "epoch": 1.075574762660195, "grad_norm": 0.6820271611213684, "learning_rate": 9.248112224193766e-06, "loss": 1.853416748046875, "step": 270350 }, { "epoch": 1.075773685493055, "grad_norm": 0.686138391494751, "learning_rate": 9.24612299784368e-06, "loss": 1.91078369140625, "step": 270400 }, { "epoch": 1.0759726083259151, "grad_norm": 0.6232523322105408, "learning_rate": 9.244133771493592e-06, "loss": 1.8895962524414063, "step": 270450 }, { "epoch": 1.0761715311587752, "grad_norm": 0.6018140912055969, "learning_rate": 9.242144545143503e-06, "loss": 1.8940597534179688, "step": 270500 }, { "epoch": 1.0763704539916352, "grad_norm": 0.5801100134849548, "learning_rate": 9.240155318793415e-06, "loss": 1.8633877563476562, "step": 270550 }, { "epoch": 1.0765693768244953, "grad_norm": 0.6395415663719177, "learning_rate": 9.238166092443329e-06, "loss": 1.8970233154296876, "step": 270600 }, { "epoch": 1.0767682996573553, "grad_norm": 0.7211031317710876, "learning_rate": 9.23617686609324e-06, "loss": 1.8687437438964845, "step": 270650 }, { "epoch": 1.0769672224902154, "grad_norm": 0.6277488470077515, "learning_rate": 9.234187639743151e-06, "loss": 1.9001950073242186, "step": 270700 }, { "epoch": 1.0771661453230756, "grad_norm": 0.6054127812385559, "learning_rate": 9.232198413393063e-06, "loss": 1.8886616516113282, "step": 270750 }, { "epoch": 1.0773650681559357, "grad_norm": 0.5405490398406982, "learning_rate": 9.230209187042976e-06, "loss": 1.8630986022949219, "step": 270800 }, { "epoch": 1.0775639909887957, "grad_norm": 0.6916034817695618, "learning_rate": 9.228219960692888e-06, "loss": 1.849470977783203, "step": 270850 }, { "epoch": 1.0777629138216558, "grad_norm": 0.6500036716461182, "learning_rate": 9.2262307343428e-06, "loss": 1.8527452087402343, "step": 270900 }, { "epoch": 1.0779618366545158, "grad_norm": 0.6793097853660583, "learning_rate": 9.224241507992712e-06, "loss": 1.8954487609863282, "step": 270950 }, { "epoch": 1.0781607594873759, "grad_norm": 0.7582953572273254, "learning_rate": 9.222252281642624e-06, "loss": 1.8728752136230469, "step": 271000 }, { "epoch": 1.078359682320236, "grad_norm": 0.7790319323539734, "learning_rate": 9.220263055292536e-06, "loss": 1.8644960021972656, "step": 271050 }, { "epoch": 1.078558605153096, "grad_norm": 0.5368496179580688, "learning_rate": 9.218273828942448e-06, "loss": 1.8466316223144532, "step": 271100 }, { "epoch": 1.078757527985956, "grad_norm": 0.7309476733207703, "learning_rate": 9.21628460259236e-06, "loss": 1.9013536071777344, "step": 271150 }, { "epoch": 1.078956450818816, "grad_norm": 0.6499818563461304, "learning_rate": 9.214295376242273e-06, "loss": 1.8223004150390625, "step": 271200 }, { "epoch": 1.0791553736516761, "grad_norm": 0.6889240741729736, "learning_rate": 9.212306149892185e-06, "loss": 1.8725773620605468, "step": 271250 }, { "epoch": 1.0793542964845362, "grad_norm": 0.5480566620826721, "learning_rate": 9.210316923542097e-06, "loss": 1.8849288940429687, "step": 271300 }, { "epoch": 1.0795532193173962, "grad_norm": 0.5143357515335083, "learning_rate": 9.20832769719201e-06, "loss": 1.85267822265625, "step": 271350 }, { "epoch": 1.0797521421502563, "grad_norm": 0.631222665309906, "learning_rate": 9.206338470841921e-06, "loss": 1.950607147216797, "step": 271400 }, { "epoch": 1.0799510649831163, "grad_norm": 0.6545188426971436, "learning_rate": 9.204349244491832e-06, "loss": 1.8644122314453124, "step": 271450 }, { "epoch": 1.0801499878159766, "grad_norm": 0.60652756690979, "learning_rate": 9.202360018141746e-06, "loss": 1.8797467041015625, "step": 271500 }, { "epoch": 1.0803489106488366, "grad_norm": 0.6616390943527222, "learning_rate": 9.200370791791658e-06, "loss": 1.8445945739746095, "step": 271550 }, { "epoch": 1.0805478334816967, "grad_norm": 0.5264456272125244, "learning_rate": 9.198381565441568e-06, "loss": 1.8516595458984375, "step": 271600 }, { "epoch": 1.0807467563145567, "grad_norm": 0.758707582950592, "learning_rate": 9.19639233909148e-06, "loss": 1.8583859252929686, "step": 271650 }, { "epoch": 1.0809456791474168, "grad_norm": 0.6942134499549866, "learning_rate": 9.194403112741394e-06, "loss": 1.884098663330078, "step": 271700 }, { "epoch": 1.0811446019802768, "grad_norm": 0.48993220925331116, "learning_rate": 9.192413886391306e-06, "loss": 1.9031867980957031, "step": 271750 }, { "epoch": 1.081343524813137, "grad_norm": 0.758561372756958, "learning_rate": 9.190424660041217e-06, "loss": 1.8381544494628905, "step": 271800 }, { "epoch": 1.081542447645997, "grad_norm": 0.6642873287200928, "learning_rate": 9.188435433691129e-06, "loss": 1.90783935546875, "step": 271850 }, { "epoch": 1.081741370478857, "grad_norm": 0.723678469657898, "learning_rate": 9.186446207341043e-06, "loss": 1.8865420532226562, "step": 271900 }, { "epoch": 1.081940293311717, "grad_norm": 0.5821460485458374, "learning_rate": 9.184456980990953e-06, "loss": 1.8818360900878905, "step": 271950 }, { "epoch": 1.082139216144577, "grad_norm": 0.5194848775863647, "learning_rate": 9.182467754640865e-06, "loss": 1.8400106811523438, "step": 272000 } ], "logging_steps": 50, "max_steps": 502708, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 2000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 3.302196075463219e+18, "train_batch_size": 6, "trial_name": null, "trial_params": null }