{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.258610954263128,
  "eval_steps": 500,
  "global_step": 10000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00564652738565782,
      "grad_norm": 0.0635828971862793,
      "learning_rate": 1.2531328320802006e-05,
      "loss": 1.3836,
      "step": 25
    },
    {
      "epoch": 0.01129305477131564,
      "grad_norm": 0.2005225569009781,
      "learning_rate": 2.506265664160401e-05,
      "loss": 1.6497,
      "step": 50
    },
    {
      "epoch": 0.01693958215697346,
      "grad_norm": 0.07235410064458847,
      "learning_rate": 3.759398496240601e-05,
      "loss": 1.1768,
      "step": 75
    },
    {
      "epoch": 0.02258610954263128,
      "grad_norm": 0.17648495733737946,
      "learning_rate": 5.012531328320802e-05,
      "loss": 1.2146,
      "step": 100
    },
    {
      "epoch": 0.028232636928289104,
      "grad_norm": 0.06953724473714828,
      "learning_rate": 6.265664160401002e-05,
      "loss": 0.8725,
      "step": 125
    },
    {
      "epoch": 0.03387916431394692,
      "grad_norm": 0.12059248238801956,
      "learning_rate": 7.518796992481203e-05,
      "loss": 0.7548,
      "step": 150
    },
    {
      "epoch": 0.039525691699604744,
      "grad_norm": 0.0689113661646843,
      "learning_rate": 8.771929824561403e-05,
      "loss": 0.7917,
      "step": 175
    },
    {
      "epoch": 0.04517221908526256,
      "grad_norm": 0.16621273756027222,
      "learning_rate": 0.00010025062656641604,
      "loss": 0.6985,
      "step": 200
    },
    {
      "epoch": 0.050818746470920384,
      "grad_norm": 0.07838872820138931,
      "learning_rate": 0.00011278195488721806,
      "loss": 0.7874,
      "step": 225
    },
    {
      "epoch": 0.05646527385657821,
      "grad_norm": 0.11714337766170502,
      "learning_rate": 0.00012531328320802005,
      "loss": 0.689,
      "step": 250
    },
    {
      "epoch": 0.062111801242236024,
      "grad_norm": 0.07283364981412888,
      "learning_rate": 0.00013784461152882208,
      "loss": 0.7719,
      "step": 275
    },
    {
      "epoch": 0.06775832862789384,
      "grad_norm": 0.1499466449022293,
      "learning_rate": 0.00015037593984962405,
      "loss": 0.655,
      "step": 300
    },
    {
      "epoch": 0.07340485601355166,
      "grad_norm": 0.08195521682500839,
      "learning_rate": 0.00016290726817042608,
      "loss": 0.7508,
      "step": 325
    },
    {
      "epoch": 0.07905138339920949,
      "grad_norm": 0.1490202099084854,
      "learning_rate": 0.00017543859649122806,
      "loss": 0.6372,
      "step": 350
    },
    {
      "epoch": 0.08469791078486731,
      "grad_norm": 0.1617508977651596,
      "learning_rate": 0.00018796992481203009,
      "loss": 0.7396,
      "step": 375
    },
    {
      "epoch": 0.09034443817052512,
      "grad_norm": 0.180181622505188,
      "learning_rate": 0.00019999999702625888,
      "loss": 0.6318,
      "step": 400
    },
    {
      "epoch": 0.09599096555618294,
      "grad_norm": 0.08272858709096909,
      "learning_rate": 0.00019999798975772924,
      "loss": 0.724,
      "step": 425
    },
    {
      "epoch": 0.10163749294184077,
      "grad_norm": 0.120403952896595,
      "learning_rate": 0.00019999226539902187,
      "loss": 0.6271,
      "step": 450
    },
    {
      "epoch": 0.10728402032749859,
      "grad_norm": 0.07971920073032379,
      "learning_rate": 0.00019998282416292055,
      "loss": 0.7256,
      "step": 475
    },
    {
      "epoch": 0.11293054771315642,
      "grad_norm": 0.1328658014535904,
      "learning_rate": 0.00019996966640037166,
      "loss": 0.6231,
      "step": 500
    },
    {
      "epoch": 0.11857707509881422,
      "grad_norm": 0.07475866377353668,
      "learning_rate": 0.00019995279260047092,
      "loss": 0.7251,
      "step": 525
    },
    {
      "epoch": 0.12422360248447205,
      "grad_norm": 0.1371573656797409,
      "learning_rate": 0.00019993220339044524,
      "loss": 0.5907,
      "step": 550
    },
    {
      "epoch": 0.12987012987012986,
      "grad_norm": 0.07268711924552917,
      "learning_rate": 0.00019990789953562961,
      "loss": 0.7304,
      "step": 575
    },
    {
      "epoch": 0.13551665725578768,
      "grad_norm": 0.14224526286125183,
      "learning_rate": 0.0001998798819394383,
      "loss": 0.5931,
      "step": 600
    },
    {
      "epoch": 0.1411631846414455,
      "grad_norm": 0.0803467407822609,
      "learning_rate": 0.00019984815164333163,
      "loss": 0.7076,
      "step": 625
    },
    {
      "epoch": 0.14680971202710333,
      "grad_norm": 0.12445727735757828,
      "learning_rate": 0.00019981270982677698,
      "loss": 0.5566,
      "step": 650
    },
    {
      "epoch": 0.15245623941276115,
      "grad_norm": 0.07665792107582092,
      "learning_rate": 0.00019977355780720514,
      "loss": 0.6985,
      "step": 675
    },
    {
      "epoch": 0.15810276679841898,
      "grad_norm": 0.13053999841213226,
      "learning_rate": 0.00019973069703996125,
      "loss": 0.5901,
      "step": 700
    },
    {
      "epoch": 0.1637492941840768,
      "grad_norm": 0.07512692362070084,
      "learning_rate": 0.00019968412911825067,
      "loss": 0.7184,
      "step": 725
    },
    {
      "epoch": 0.16939582156973462,
      "grad_norm": 0.12033283710479736,
      "learning_rate": 0.00019963385577307987,
      "loss": 0.6013,
      "step": 750
    },
    {
      "epoch": 0.17504234895539245,
      "grad_norm": 0.0809774249792099,
      "learning_rate": 0.000199579878873192,
      "loss": 0.7054,
      "step": 775
    },
    {
      "epoch": 0.18068887634105024,
      "grad_norm": 0.12490582466125488,
      "learning_rate": 0.0001995222004249974,
      "loss": 0.5714,
      "step": 800
    },
    {
      "epoch": 0.18633540372670807,
      "grad_norm": 0.08289226144552231,
      "learning_rate": 0.00019946082257249912,
      "loss": 0.7304,
      "step": 825
    },
    {
      "epoch": 0.1919819311123659,
      "grad_norm": 0.14385385811328888,
      "learning_rate": 0.00019939574759721316,
      "loss": 0.5639,
      "step": 850
    },
    {
      "epoch": 0.1976284584980237,
      "grad_norm": 0.07623141258955002,
      "learning_rate": 0.00019932697791808366,
      "loss": 0.7126,
      "step": 875
    },
    {
      "epoch": 0.20327498588368154,
      "grad_norm": 0.11150185018777847,
      "learning_rate": 0.000199254516091393,
      "loss": 0.5903,
      "step": 900
    },
    {
      "epoch": 0.20892151326933936,
      "grad_norm": 0.07193930447101593,
      "learning_rate": 0.00019917836481066675,
      "loss": 0.6952,
      "step": 925
    },
    {
      "epoch": 0.21456804065499718,
      "grad_norm": 0.11242598295211792,
      "learning_rate": 0.00019909852690657359,
      "loss": 0.5853,
      "step": 950
    },
    {
      "epoch": 0.220214568040655,
      "grad_norm": 0.07508910447359085,
      "learning_rate": 0.0001990150053468201,
      "loss": 0.6969,
      "step": 975
    },
    {
      "epoch": 0.22586109542631283,
      "grad_norm": 0.11998436599969864,
      "learning_rate": 0.00019892780323604035,
      "loss": 0.5791,
      "step": 1000
    },
    {
      "epoch": 0.23150762281197063,
      "grad_norm": 0.07748444378376007,
      "learning_rate": 0.0001988369238156806,
      "loss": 0.7038,
      "step": 1025
    },
    {
      "epoch": 0.23715415019762845,
      "grad_norm": 0.19617140293121338,
      "learning_rate": 0.0001987423704638788,
      "loss": 0.5792,
      "step": 1050
    },
    {
      "epoch": 0.24280067758328627,
      "grad_norm": 0.08167731016874313,
      "learning_rate": 0.00019864414669533892,
      "loss": 0.6989,
      "step": 1075
    },
    {
      "epoch": 0.2484472049689441,
      "grad_norm": 0.11911099404096603,
      "learning_rate": 0.00019854225616120044,
      "loss": 0.5985,
      "step": 1100
    },
    {
      "epoch": 0.2540937323546019,
      "grad_norm": 0.0718325674533844,
      "learning_rate": 0.0001984367026489025,
      "loss": 0.6853,
      "step": 1125
    },
    {
      "epoch": 0.2597402597402597,
      "grad_norm": 0.12495870143175125,
      "learning_rate": 0.0001983274900820432,
      "loss": 0.5738,
      "step": 1150
    },
    {
      "epoch": 0.26538678712591757,
      "grad_norm": 0.08245467394590378,
      "learning_rate": 0.0001982146225202338,
      "loss": 0.6766,
      "step": 1175
    },
    {
      "epoch": 0.27103331451157536,
      "grad_norm": 0.11854438483715057,
      "learning_rate": 0.00019809810415894767,
      "loss": 0.5828,
      "step": 1200
    },
    {
      "epoch": 0.2766798418972332,
      "grad_norm": 0.07968372851610184,
      "learning_rate": 0.0001979779393293644,
      "loss": 0.6901,
      "step": 1225
    },
    {
      "epoch": 0.282326369282891,
      "grad_norm": 0.13610929250717163,
      "learning_rate": 0.00019785413249820893,
      "loss": 0.554,
      "step": 1250
    },
    {
      "epoch": 0.28797289666854886,
      "grad_norm": 0.08099998533725739,
      "learning_rate": 0.00019772668826758527,
      "loss": 0.6615,
      "step": 1275
    },
    {
      "epoch": 0.29361942405420666,
      "grad_norm": 0.10878365486860275,
      "learning_rate": 0.0001975956113748057,
      "loss": 0.547,
      "step": 1300
    },
    {
      "epoch": 0.2992659514398645,
      "grad_norm": 0.0784253254532814,
      "learning_rate": 0.0001974609066922144,
      "loss": 0.6589,
      "step": 1325
    },
    {
      "epoch": 0.3049124788255223,
      "grad_norm": 0.10678227245807648,
      "learning_rate": 0.00019732257922700655,
      "loss": 0.5443,
      "step": 1350
    },
    {
      "epoch": 0.3105590062111801,
      "grad_norm": 0.07781955599784851,
      "learning_rate": 0.00019718063412104222,
      "loss": 0.6798,
      "step": 1375
    },
    {
      "epoch": 0.31620553359683795,
      "grad_norm": 0.12910470366477966,
      "learning_rate": 0.00019703507665065498,
      "loss": 0.54,
      "step": 1400
    },
    {
      "epoch": 0.32185206098249575,
      "grad_norm": 0.08442296832799911,
      "learning_rate": 0.00019688591222645607,
      "loss": 0.6998,
      "step": 1425
    },
    {
      "epoch": 0.3274985883681536,
      "grad_norm": 0.10945732891559601,
      "learning_rate": 0.00019673314639313315,
      "loss": 0.5576,
      "step": 1450
    },
    {
      "epoch": 0.3331451157538114,
      "grad_norm": 0.07373673468828201,
      "learning_rate": 0.00019657678482924406,
      "loss": 0.709,
      "step": 1475
    },
    {
      "epoch": 0.33879164313946925,
      "grad_norm": 0.10061786323785782,
      "learning_rate": 0.00019641683334700608,
      "loss": 0.5329,
      "step": 1500
    },
    {
      "epoch": 0.34443817052512704,
      "grad_norm": 0.08854890614748001,
      "learning_rate": 0.00019625329789207949,
      "loss": 0.6753,
      "step": 1525
    },
    {
      "epoch": 0.3500846979107849,
      "grad_norm": 0.11869047582149506,
      "learning_rate": 0.00019608618454334685,
      "loss": 0.54,
      "step": 1550
    },
    {
      "epoch": 0.3557312252964427,
      "grad_norm": 0.0819987803697586,
      "learning_rate": 0.00019591549951268692,
      "loss": 0.683,
      "step": 1575
    },
    {
      "epoch": 0.3613777526821005,
      "grad_norm": 0.10941476374864578,
      "learning_rate": 0.00019574124914474374,
      "loss": 0.5713,
      "step": 1600
    },
    {
      "epoch": 0.36702428006775834,
      "grad_norm": 0.0801759734749794,
      "learning_rate": 0.00019556343991669083,
      "loss": 0.6494,
      "step": 1625
    },
    {
      "epoch": 0.37267080745341613,
      "grad_norm": 0.11291850358247757,
      "learning_rate": 0.0001953820784379904,
      "loss": 0.5428,
      "step": 1650
    },
    {
      "epoch": 0.378317334839074,
      "grad_norm": 0.0784342810511589,
      "learning_rate": 0.00019519717145014765,
      "loss": 0.6789,
      "step": 1675
    },
    {
      "epoch": 0.3839638622247318,
      "grad_norm": 0.11995444446802139,
      "learning_rate": 0.00019500872582646034,
      "loss": 0.5409,
      "step": 1700
    },
    {
      "epoch": 0.38961038961038963,
      "grad_norm": 0.10059577226638794,
      "learning_rate": 0.00019481674857176293,
      "loss": 0.6887,
      "step": 1725
    },
    {
      "epoch": 0.3952569169960474,
      "grad_norm": 0.11123017966747284,
      "learning_rate": 0.0001946212468221666,
      "loss": 0.5482,
      "step": 1750
    },
    {
      "epoch": 0.4009034443817053,
      "grad_norm": 0.0788169875741005,
      "learning_rate": 0.00019442222784479382,
      "loss": 0.6577,
      "step": 1775
    },
    {
      "epoch": 0.40654997176736307,
      "grad_norm": 0.09987206757068634,
      "learning_rate": 0.0001942196990375081,
      "loss": 0.5353,
      "step": 1800
    },
    {
      "epoch": 0.41219649915302087,
      "grad_norm": 0.081305131316185,
      "learning_rate": 0.00019401366792863914,
      "loss": 0.6757,
      "step": 1825
    },
    {
      "epoch": 0.4178430265386787,
      "grad_norm": 0.11595868319272995,
      "learning_rate": 0.00019380414217670309,
      "loss": 0.5306,
      "step": 1850
    },
    {
      "epoch": 0.4234895539243365,
      "grad_norm": 0.08652087301015854,
      "learning_rate": 0.00019359112957011764,
      "loss": 0.6634,
      "step": 1875
    },
    {
      "epoch": 0.42913608130999437,
      "grad_norm": 0.1181817501783371,
      "learning_rate": 0.00019337463802691264,
      "loss": 0.5465,
      "step": 1900
    },
    {
      "epoch": 0.43478260869565216,
      "grad_norm": 0.08145532757043839,
      "learning_rate": 0.00019315467559443574,
      "loss": 0.6639,
      "step": 1925
    },
    {
      "epoch": 0.44042913608131,
      "grad_norm": 0.10263626277446747,
      "learning_rate": 0.0001929312504490533,
      "loss": 0.5418,
      "step": 1950
    },
    {
      "epoch": 0.4460756634669678,
      "grad_norm": 0.07672405242919922,
      "learning_rate": 0.00019270437089584635,
      "loss": 0.6547,
      "step": 1975
    },
    {
      "epoch": 0.45172219085262566,
      "grad_norm": 0.0984482392668724,
      "learning_rate": 0.00019247404536830204,
      "loss": 0.5321,
      "step": 2000
    },
    {
      "epoch": 0.45736871823828346,
      "grad_norm": 0.08558057248592377,
      "learning_rate": 0.00019224028242800006,
      "loss": 0.6657,
      "step": 2025
    },
    {
      "epoch": 0.46301524562394125,
      "grad_norm": 0.10624836385250092,
      "learning_rate": 0.00019200309076429438,
      "loss": 0.5542,
      "step": 2050
    },
    {
      "epoch": 0.4686617730095991,
      "grad_norm": 0.08049538731575012,
      "learning_rate": 0.00019176247919399023,
      "loss": 0.6389,
      "step": 2075
    },
    {
      "epoch": 0.4743083003952569,
      "grad_norm": 0.11163744330406189,
      "learning_rate": 0.00019151845666101646,
      "loss": 0.5382,
      "step": 2100
    },
    {
      "epoch": 0.47995482778091475,
      "grad_norm": 0.07694784551858902,
      "learning_rate": 0.00019127103223609307,
      "loss": 0.6345,
      "step": 2125
    },
    {
      "epoch": 0.48560135516657255,
      "grad_norm": 0.11501342058181763,
      "learning_rate": 0.0001910202151163939,
      "loss": 0.5303,
      "step": 2150
    },
    {
      "epoch": 0.4912478825522304,
      "grad_norm": 0.08532056957483292,
      "learning_rate": 0.00019076601462520492,
      "loss": 0.666,
      "step": 2175
    },
    {
      "epoch": 0.4968944099378882,
      "grad_norm": 0.09018545597791672,
      "learning_rate": 0.00019050844021157752,
      "loss": 0.5363,
      "step": 2200
    },
    {
      "epoch": 0.502540937323546,
      "grad_norm": 0.08757849037647247,
      "learning_rate": 0.00019024750144997746,
      "loss": 0.6684,
      "step": 2225
    },
    {
      "epoch": 0.5081874647092038,
      "grad_norm": 0.09418553113937378,
      "learning_rate": 0.00018998320803992872,
      "loss": 0.565,
      "step": 2250
    },
    {
      "epoch": 0.5138339920948617,
      "grad_norm": 0.10246949642896652,
      "learning_rate": 0.00018971556980565329,
      "loss": 0.6524,
      "step": 2275
    },
    {
      "epoch": 0.5194805194805194,
      "grad_norm": 0.12446881830692291,
      "learning_rate": 0.00018944459669570555,
      "loss": 0.5447,
      "step": 2300
    },
    {
      "epoch": 0.5251270468661773,
      "grad_norm": 0.08439560234546661,
      "learning_rate": 0.00018917029878260294,
      "loss": 0.673,
      "step": 2325
    },
    {
      "epoch": 0.5307735742518351,
      "grad_norm": 0.1195409968495369,
      "learning_rate": 0.00018889268626245116,
      "loss": 0.51,
      "step": 2350
    },
    {
      "epoch": 0.536420101637493,
      "grad_norm": 0.09150119870901108,
      "learning_rate": 0.0001886117694545654,
      "loss": 0.6631,
      "step": 2375
    },
    {
      "epoch": 0.5420666290231507,
      "grad_norm": 0.10301247239112854,
      "learning_rate": 0.0001883275588010866,
      "loss": 0.5257,
      "step": 2400
    },
    {
      "epoch": 0.5477131564088086,
      "grad_norm": 0.08086048811674118,
      "learning_rate": 0.00018804006486659346,
      "loss": 0.6305,
      "step": 2425
    },
    {
      "epoch": 0.5533596837944664,
      "grad_norm": 0.10540500283241272,
      "learning_rate": 0.0001877492983377096,
      "loss": 0.5607,
      "step": 2450
    },
    {
      "epoch": 0.5590062111801242,
      "grad_norm": 0.0777876153588295,
      "learning_rate": 0.00018745527002270634,
      "loss": 0.6413,
      "step": 2475
    },
    {
      "epoch": 0.564652738565782,
      "grad_norm": 0.10727331042289734,
      "learning_rate": 0.00018715799085110112,
      "loss": 0.5206,
      "step": 2500
    },
    {
      "epoch": 0.5702992659514399,
      "grad_norm": 0.09144891798496246,
      "learning_rate": 0.0001868574718732508,
      "loss": 0.6655,
      "step": 2525
    },
    {
      "epoch": 0.5759457933370977,
      "grad_norm": 0.08949702978134155,
      "learning_rate": 0.00018655372425994152,
      "loss": 0.5358,
      "step": 2550
    },
    {
      "epoch": 0.5815923207227555,
      "grad_norm": 0.08744482696056366,
      "learning_rate": 0.0001862467593019728,
      "loss": 0.6396,
      "step": 2575
    },
    {
      "epoch": 0.5872388481084133,
      "grad_norm": 0.10931294411420822,
      "learning_rate": 0.0001859365884097384,
      "loss": 0.5132,
      "step": 2600
    },
    {
      "epoch": 0.5928853754940712,
      "grad_norm": 0.07518015056848526,
      "learning_rate": 0.00018562322311280186,
      "loss": 0.6378,
      "step": 2625
    },
    {
      "epoch": 0.598531902879729,
      "grad_norm": 0.09787683933973312,
      "learning_rate": 0.000185306675059468,
      "loss": 0.5272,
      "step": 2650
    },
    {
      "epoch": 0.6041784302653868,
      "grad_norm": 0.08679146319627762,
      "learning_rate": 0.00018498695601634993,
      "loss": 0.6486,
      "step": 2675
    },
    {
      "epoch": 0.6098249576510446,
      "grad_norm": 0.10611116141080856,
      "learning_rate": 0.00018466407786793174,
      "loss": 0.5237,
      "step": 2700
    },
    {
      "epoch": 0.6154714850367025,
      "grad_norm": 0.0772860199213028,
      "learning_rate": 0.00018433805261612663,
      "loss": 0.6526,
      "step": 2725
    },
    {
      "epoch": 0.6211180124223602,
      "grad_norm": 0.1005689725279808,
      "learning_rate": 0.00018400889237983086,
      "loss": 0.5297,
      "step": 2750
    },
    {
      "epoch": 0.626764539808018,
      "grad_norm": 0.08322805911302567,
      "learning_rate": 0.00018367660939447316,
      "loss": 0.6448,
      "step": 2775
    },
    {
      "epoch": 0.6324110671936759,
      "grad_norm": 0.10492440313100815,
      "learning_rate": 0.00018334121601156002,
      "loss": 0.493,
      "step": 2800
    },
    {
      "epoch": 0.6380575945793338,
      "grad_norm": 0.08439356088638306,
      "learning_rate": 0.00018300272469821662,
      "loss": 0.6421,
      "step": 2825
    },
    {
      "epoch": 0.6437041219649915,
      "grad_norm": 0.11175478994846344,
      "learning_rate": 0.00018266114803672318,
      "loss": 0.5293,
      "step": 2850
    },
    {
      "epoch": 0.6493506493506493,
      "grad_norm": 0.07700271904468536,
      "learning_rate": 0.00018231649872404754,
      "loss": 0.6506,
      "step": 2875
    },
    {
      "epoch": 0.6549971767363072,
      "grad_norm": 0.08704936504364014,
      "learning_rate": 0.00018196878957137295,
      "loss": 0.5091,
      "step": 2900
    },
    {
      "epoch": 0.6606437041219649,
      "grad_norm": 0.0863596498966217,
      "learning_rate": 0.00018161803350362198,
      "loss": 0.6531,
      "step": 2925
    },
    {
      "epoch": 0.6662902315076228,
      "grad_norm": 0.11374282091856003,
      "learning_rate": 0.00018126424355897612,
      "loss": 0.5389,
      "step": 2950
    },
    {
      "epoch": 0.6719367588932806,
      "grad_norm": 0.08936896920204163,
      "learning_rate": 0.000180907432888391,
      "loss": 0.6267,
      "step": 2975
    },
    {
      "epoch": 0.6775832862789385,
      "grad_norm": 0.08924362808465958,
      "learning_rate": 0.0001805476147551076,
      "loss": 0.4882,
      "step": 3000
    },
    {
      "epoch": 0.6832298136645962,
      "grad_norm": 0.08161070942878723,
      "learning_rate": 0.0001801848025341593,
      "loss": 0.6502,
      "step": 3025
    },
    {
      "epoch": 0.6888763410502541,
      "grad_norm": 0.11261451989412308,
      "learning_rate": 0.00017981900971187465,
      "loss": 0.5241,
      "step": 3050
    },
    {
      "epoch": 0.6945228684359119,
      "grad_norm": 0.0859522670507431,
      "learning_rate": 0.00017945024988537603,
      "loss": 0.6413,
      "step": 3075
    },
    {
      "epoch": 0.7001693958215698,
      "grad_norm": 0.11650535464286804,
      "learning_rate": 0.0001790785367620743,
      "loss": 0.519,
      "step": 3100
    },
    {
      "epoch": 0.7058159232072275,
      "grad_norm": 0.09202492237091064,
      "learning_rate": 0.00017870388415915922,
      "loss": 0.6212,
      "step": 3125
    },
    {
      "epoch": 0.7114624505928854,
      "grad_norm": 0.1048663780093193,
      "learning_rate": 0.00017832630600308585,
      "loss": 0.5266,
      "step": 3150
    },
    {
      "epoch": 0.7171089779785432,
      "grad_norm": 0.08784987032413483,
      "learning_rate": 0.00017794581632905683,
      "loss": 0.6214,
      "step": 3175
    },
    {
      "epoch": 0.722755505364201,
      "grad_norm": 0.10895237326622009,
      "learning_rate": 0.00017756242928050085,
      "loss": 0.5059,
      "step": 3200
    },
    {
      "epoch": 0.7284020327498588,
      "grad_norm": 0.09365130960941315,
      "learning_rate": 0.0001771761591085467,
      "loss": 0.6529,
      "step": 3225
    },
    {
      "epoch": 0.7340485601355167,
      "grad_norm": 0.1045169085264206,
      "learning_rate": 0.0001767870201714936,
      "loss": 0.5229,
      "step": 3250
    },
    {
      "epoch": 0.7396950875211745,
      "grad_norm": 0.08621610701084137,
      "learning_rate": 0.0001763950269342776,
      "loss": 0.6489,
      "step": 3275
    },
    {
      "epoch": 0.7453416149068323,
      "grad_norm": 0.1039639338850975,
      "learning_rate": 0.00017600019396793367,
      "loss": 0.5116,
      "step": 3300
    },
    {
      "epoch": 0.7509881422924901,
      "grad_norm": 0.09034851938486099,
      "learning_rate": 0.00017560253594905425,
      "loss": 0.6587,
      "step": 3325
    },
    {
      "epoch": 0.756634669678148,
      "grad_norm": 0.10783170908689499,
      "learning_rate": 0.00017520206765924372,
      "loss": 0.5019,
      "step": 3350
    },
    {
      "epoch": 0.7622811970638057,
      "grad_norm": 0.08911153674125671,
      "learning_rate": 0.00017479880398456871,
      "loss": 0.6242,
      "step": 3375
    },
    {
      "epoch": 0.7679277244494636,
      "grad_norm": 0.10053360462188721,
      "learning_rate": 0.00017439275991500507,
      "loss": 0.4996,
      "step": 3400
    },
    {
      "epoch": 0.7735742518351214,
      "grad_norm": 0.09367632865905762,
      "learning_rate": 0.0001739839505438804,
      "loss": 0.6347,
      "step": 3425
    },
    {
      "epoch": 0.7792207792207793,
      "grad_norm": 0.09187789261341095,
      "learning_rate": 0.00017357239106731317,
      "loss": 0.4857,
      "step": 3450
    },
    {
      "epoch": 0.784867306606437,
      "grad_norm": 0.08425033092498779,
      "learning_rate": 0.00017315809678364777,
      "loss": 0.6305,
      "step": 3475
    },
    {
      "epoch": 0.7905138339920948,
      "grad_norm": 0.10775783658027649,
      "learning_rate": 0.00017274108309288594,
      "loss": 0.5013,
      "step": 3500
    },
    {
      "epoch": 0.7961603613777527,
      "grad_norm": 0.082928866147995,
      "learning_rate": 0.00017232136549611416,
      "loss": 0.6217,
      "step": 3525
    },
    {
      "epoch": 0.8018068887634106,
      "grad_norm": 0.09590538591146469,
      "learning_rate": 0.00017189895959492772,
      "loss": 0.4871,
      "step": 3550
    },
    {
      "epoch": 0.8074534161490683,
      "grad_norm": 0.08597232401371002,
      "learning_rate": 0.00017147388109085048,
      "loss": 0.6298,
      "step": 3575
    },
    {
      "epoch": 0.8130999435347261,
      "grad_norm": 0.10229629278182983,
      "learning_rate": 0.00017104614578475135,
      "loss": 0.5191,
      "step": 3600
    },
    {
      "epoch": 0.818746470920384,
      "grad_norm": 0.09689701348543167,
      "learning_rate": 0.0001706157695762571,
      "loss": 0.6087,
      "step": 3625
    },
    {
      "epoch": 0.8243929983060417,
      "grad_norm": 0.1155649945139885,
      "learning_rate": 0.000170182768463161,
      "loss": 0.5237,
      "step": 3650
    },
    {
      "epoch": 0.8300395256916996,
      "grad_norm": 0.08635739237070084,
      "learning_rate": 0.00016974715854082848,
      "loss": 0.6337,
      "step": 3675
    },
    {
      "epoch": 0.8356860530773574,
      "grad_norm": 0.08977790176868439,
      "learning_rate": 0.00016930895600159867,
      "loss": 0.4942,
      "step": 3700
    },
    {
      "epoch": 0.8413325804630153,
      "grad_norm": 0.08372893184423447,
      "learning_rate": 0.00016886817713418264,
      "loss": 0.6224,
      "step": 3725
    },
    {
      "epoch": 0.846979107848673,
      "grad_norm": 0.09409157931804657,
      "learning_rate": 0.00016842483832305765,
      "loss": 0.4948,
      "step": 3750
    },
    {
      "epoch": 0.8526256352343309,
      "grad_norm": 0.0824190303683281,
      "learning_rate": 0.00016797895604785842,
      "loss": 0.6188,
      "step": 3775
    },
    {
      "epoch": 0.8582721626199887,
      "grad_norm": 0.11509312689304352,
      "learning_rate": 0.0001675305468827644,
      "loss": 0.513,
      "step": 3800
    },
    {
      "epoch": 0.8639186900056465,
      "grad_norm": 0.08958346396684647,
      "learning_rate": 0.0001670796274958837,
      "loss": 0.6206,
      "step": 3825
    },
    {
      "epoch": 0.8695652173913043,
      "grad_norm": 0.0959952101111412,
      "learning_rate": 0.00016662621464863338,
      "loss": 0.4885,
      "step": 3850
    },
    {
      "epoch": 0.8752117447769622,
      "grad_norm": 0.11436719447374344,
      "learning_rate": 0.00016617032519511686,
      "loss": 0.635,
      "step": 3875
    },
    {
      "epoch": 0.88085827216262,
      "grad_norm": 0.11012863367795944,
      "learning_rate": 0.00016571197608149674,
      "loss": 0.4982,
      "step": 3900
    },
    {
      "epoch": 0.8865047995482778,
      "grad_norm": 0.08522720634937286,
      "learning_rate": 0.00016525118434536546,
      "loss": 0.6105,
      "step": 3925
    },
    {
      "epoch": 0.8921513269339356,
      "grad_norm": 0.11074435710906982,
      "learning_rate": 0.00016478796711511171,
      "loss": 0.5007,
      "step": 3950
    },
    {
      "epoch": 0.8977978543195935,
      "grad_norm": 0.08319137990474701,
      "learning_rate": 0.00016432234160928378,
      "loss": 0.6008,
      "step": 3975
    },
    {
      "epoch": 0.9034443817052513,
      "grad_norm": 0.11340347677469254,
      "learning_rate": 0.00016385432513594953,
      "loss": 0.502,
      "step": 4000
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 0.08557943999767303,
      "learning_rate": 0.0001633839350920531,
      "loss": 0.598,
      "step": 4025
    },
    {
      "epoch": 0.9147374364765669,
      "grad_norm": 0.09813399612903595,
      "learning_rate": 0.00016291118896276803,
      "loss": 0.5139,
      "step": 4050
    },
    {
      "epoch": 0.9203839638622248,
      "grad_norm": 0.07974950224161148,
      "learning_rate": 0.00016243610432084755,
      "loss": 0.6227,
      "step": 4075
    },
    {
      "epoch": 0.9260304912478825,
      "grad_norm": 0.09820155799388885,
      "learning_rate": 0.0001619586988259712,
      "loss": 0.5044,
      "step": 4100
    },
    {
      "epoch": 0.9316770186335404,
      "grad_norm": 0.0881289690732956,
      "learning_rate": 0.00016147899022408852,
      "loss": 0.6192,
      "step": 4125
    },
    {
      "epoch": 0.9373235460191982,
      "grad_norm": 0.11012347787618637,
      "learning_rate": 0.00016099699634675925,
      "loss": 0.508,
      "step": 4150
    },
    {
      "epoch": 0.9429700734048561,
      "grad_norm": 0.0851251408457756,
      "learning_rate": 0.00016051273511049065,
      "loss": 0.5897,
      "step": 4175
    },
    {
      "epoch": 0.9486166007905138,
      "grad_norm": 0.10288140922784805,
      "learning_rate": 0.0001600262245160714,
      "loss": 0.4803,
      "step": 4200
    },
    {
      "epoch": 0.9542631281761716,
      "grad_norm": 0.08995950222015381,
      "learning_rate": 0.0001595374826479026,
      "loss": 0.6203,
      "step": 4225
    },
    {
      "epoch": 0.9599096555618295,
      "grad_norm": 0.10717281699180603,
      "learning_rate": 0.00015904652767332537,
      "loss": 0.5068,
      "step": 4250
    },
    {
      "epoch": 0.9655561829474872,
      "grad_norm": 0.08305976539850235,
      "learning_rate": 0.00015855337784194577,
      "loss": 0.5919,
      "step": 4275
    },
    {
      "epoch": 0.9712027103331451,
      "grad_norm": 0.09328145533800125,
      "learning_rate": 0.00015805805148495623,
      "loss": 0.4948,
      "step": 4300
    },
    {
      "epoch": 0.9768492377188029,
      "grad_norm": 0.09493458271026611,
      "learning_rate": 0.00015756056701445422,
      "loss": 0.6024,
      "step": 4325
    },
    {
      "epoch": 0.9824957651044608,
      "grad_norm": 0.1051265150308609,
      "learning_rate": 0.0001570609429227579,
      "loss": 0.4991,
      "step": 4350
    },
    {
      "epoch": 0.9881422924901185,
      "grad_norm": 0.09176123887300491,
      "learning_rate": 0.00015655919778171862,
      "loss": 0.6006,
      "step": 4375
    },
    {
      "epoch": 0.9937888198757764,
      "grad_norm": 0.0787501335144043,
      "learning_rate": 0.00015605535024203069,
      "loss": 0.5013,
      "step": 4400
    },
    {
      "epoch": 0.9994353472614342,
      "grad_norm": 0.10489033907651901,
      "learning_rate": 0.00015554941903253797,
      "loss": 0.5548,
      "step": 4425
    },
    {
      "epoch": 1.005081874647092,
      "grad_norm": 0.08116624504327774,
      "learning_rate": 0.00015504142295953783,
      "loss": 0.5719,
      "step": 4450
    },
    {
      "epoch": 1.0107284020327498,
      "grad_norm": 0.09656750410795212,
      "learning_rate": 0.000154531380906082,
      "loss": 0.4752,
      "step": 4475
    },
    {
      "epoch": 1.0163749294184077,
      "grad_norm": 0.09433398395776749,
      "learning_rate": 0.0001540193118312747,
      "loss": 0.5796,
      "step": 4500
    },
    {
      "epoch": 1.0220214568040655,
      "grad_norm": 0.10783005505800247,
      "learning_rate": 0.0001535052347695678,
      "loss": 0.5077,
      "step": 4525
    },
    {
      "epoch": 1.0276679841897234,
      "grad_norm": 0.08784560114145279,
      "learning_rate": 0.00015298916883005342,
      "loss": 0.5571,
      "step": 4550
    },
    {
      "epoch": 1.0333145115753812,
      "grad_norm": 0.11868823319673538,
      "learning_rate": 0.00015247113319575358,
      "loss": 0.5223,
      "step": 4575
    },
    {
      "epoch": 1.0389610389610389,
      "grad_norm": 0.08813630044460297,
      "learning_rate": 0.000151951147122907,
      "loss": 0.561,
      "step": 4600
    },
    {
      "epoch": 1.0446075663466967,
      "grad_norm": 0.09996681660413742,
      "learning_rate": 0.0001514292299402535,
      "loss": 0.5138,
      "step": 4625
    },
    {
      "epoch": 1.0502540937323546,
      "grad_norm": 0.07442251592874527,
      "learning_rate": 0.00015090540104831539,
      "loss": 0.5698,
      "step": 4650
    },
    {
      "epoch": 1.0559006211180124,
      "grad_norm": 0.09993384033441544,
      "learning_rate": 0.00015037967991867642,
      "loss": 0.5093,
      "step": 4675
    },
    {
      "epoch": 1.0615471485036703,
      "grad_norm": 0.0890798568725586,
      "learning_rate": 0.0001498520860932579,
      "loss": 0.5481,
      "step": 4700
    },
    {
      "epoch": 1.0671936758893281,
      "grad_norm": 0.1006857305765152,
      "learning_rate": 0.00014932263918359228,
      "loss": 0.5045,
      "step": 4725
    },
    {
      "epoch": 1.072840203274986,
      "grad_norm": 0.08858868479728699,
      "learning_rate": 0.00014879135887009435,
      "loss": 0.5772,
      "step": 4750
    },
    {
      "epoch": 1.0784867306606438,
      "grad_norm": 0.11783988773822784,
      "learning_rate": 0.00014825826490132938,
      "loss": 0.4937,
      "step": 4775
    },
    {
      "epoch": 1.0841332580463015,
      "grad_norm": 0.0875164121389389,
      "learning_rate": 0.00014772337709327923,
      "loss": 0.5554,
      "step": 4800
    },
    {
      "epoch": 1.0897797854319593,
      "grad_norm": 0.112996406853199,
      "learning_rate": 0.00014718671532860592,
      "loss": 0.5126,
      "step": 4825
    },
    {
      "epoch": 1.0954263128176172,
      "grad_norm": 0.08834437280893326,
      "learning_rate": 0.000146648299555912,
      "loss": 0.5885,
      "step": 4850
    },
    {
      "epoch": 1.101072840203275,
      "grad_norm": 0.11341985315084457,
      "learning_rate": 0.00014610814978899983,
      "loss": 0.4871,
      "step": 4875
    },
    {
      "epoch": 1.1067193675889329,
      "grad_norm": 0.09301973134279251,
      "learning_rate": 0.00014556628610612677,
      "loss": 0.5839,
      "step": 4900
    },
    {
      "epoch": 1.1123658949745907,
      "grad_norm": 0.11545081436634064,
      "learning_rate": 0.00014502272864925955,
      "loss": 0.4868,
      "step": 4925
    },
    {
      "epoch": 1.1180124223602483,
      "grad_norm": 0.08969846367835999,
      "learning_rate": 0.00014447749762332515,
      "loss": 0.5472,
      "step": 4950
    },
    {
      "epoch": 1.1236589497459062,
      "grad_norm": 0.12094131112098694,
      "learning_rate": 0.00014393061329545992,
      "loss": 0.5234,
      "step": 4975
    },
    {
      "epoch": 1.129305477131564,
      "grad_norm": 0.0875294953584671,
      "learning_rate": 0.0001433820959942561,
      "loss": 0.5852,
      "step": 5000
    },
    {
      "epoch": 1.134952004517222,
      "grad_norm": 0.12163736671209335,
      "learning_rate": 0.00014283196610900638,
      "loss": 0.4951,
      "step": 5025
    },
    {
      "epoch": 1.1405985319028797,
      "grad_norm": 0.08404785394668579,
      "learning_rate": 0.00014230234328167044,
      "loss": 0.5629,
      "step": 5050
    },
    {
      "epoch": 1.1462450592885376,
      "grad_norm": 0.11473394185304642,
      "learning_rate": 0.0001417491121057749,
      "loss": 0.5062,
      "step": 5075
    },
    {
      "epoch": 1.1518915866741954,
      "grad_norm": 0.09582609683275223,
      "learning_rate": 0.0001411943290465374,
      "loss": 0.5705,
      "step": 5100
    },
    {
      "epoch": 1.1575381140598533,
      "grad_norm": 0.11270004510879517,
      "learning_rate": 0.00014063801472615902,
      "loss": 0.4918,
      "step": 5125
    },
    {
      "epoch": 1.163184641445511,
      "grad_norm": 0.2396761178970337,
      "learning_rate": 0.00014008018982376044,
      "loss": 0.5965,
      "step": 5150
    },
    {
      "epoch": 1.1688311688311688,
      "grad_norm": 0.1252664476633072,
      "learning_rate": 0.00013952087507461321,
      "loss": 0.4814,
      "step": 5175
    },
    {
      "epoch": 1.1744776962168266,
      "grad_norm": 0.10077520459890366,
      "learning_rate": 0.0001389600912693688,
      "loss": 0.5723,
      "step": 5200
    },
    {
      "epoch": 1.1801242236024845,
      "grad_norm": 0.10539643466472626,
      "learning_rate": 0.00013839785925328605,
      "loss": 0.4476,
      "step": 5225
    },
    {
      "epoch": 1.1857707509881423,
      "grad_norm": 0.0933731198310852,
      "learning_rate": 0.0001378341999254561,
      "loss": 0.5683,
      "step": 5250
    },
    {
      "epoch": 1.1914172783738002,
      "grad_norm": 0.12707217037677765,
      "learning_rate": 0.00013726913423802562,
      "loss": 0.5142,
      "step": 5275
    },
    {
      "epoch": 1.1970638057594578,
      "grad_norm": 0.08691050857305527,
      "learning_rate": 0.0001367026831954181,
      "loss": 0.5628,
      "step": 5300
    },
    {
      "epoch": 1.2027103331451157,
      "grad_norm": 0.10984601825475693,
      "learning_rate": 0.0001361348678535528,
      "loss": 0.4924,
      "step": 5325
    },
    {
      "epoch": 1.2083568605307735,
      "grad_norm": 0.10582837462425232,
      "learning_rate": 0.00013556570931906232,
      "loss": 0.5564,
      "step": 5350
    },
    {
      "epoch": 1.2140033879164314,
      "grad_norm": 0.1085626408457756,
      "learning_rate": 0.000134995228748508,
      "loss": 0.492,
      "step": 5375
    },
    {
      "epoch": 1.2196499153020892,
      "grad_norm": 0.11668406426906586,
      "learning_rate": 0.00013442344734759332,
      "loss": 0.5651,
      "step": 5400
    },
    {
      "epoch": 1.225296442687747,
      "grad_norm": 0.10734312981367111,
      "learning_rate": 0.00013385038637037585,
      "loss": 0.4848,
      "step": 5425
    },
    {
      "epoch": 1.230942970073405,
      "grad_norm": 0.10206615924835205,
      "learning_rate": 0.00013327606711847713,
      "loss": 0.5739,
      "step": 5450
    },
    {
      "epoch": 1.2365894974590628,
      "grad_norm": 0.11335214227437973,
      "learning_rate": 0.00013270051094029075,
      "loss": 0.4757,
      "step": 5475
    },
    {
      "epoch": 1.2422360248447206,
      "grad_norm": 0.10749132186174393,
      "learning_rate": 0.00013212373923018905,
      "loss": 0.5769,
      "step": 5500
    },
    {
      "epoch": 1.2478825522303783,
      "grad_norm": 0.13299238681793213,
      "learning_rate": 0.0001315457734277275,
      "loss": 0.4758,
      "step": 5525
    },
    {
      "epoch": 1.253529079616036,
      "grad_norm": 0.09394491463899612,
      "learning_rate": 0.00013096663501684813,
      "loss": 0.5465,
      "step": 5550
    },
    {
      "epoch": 1.259175607001694,
      "grad_norm": 0.11942502111196518,
      "learning_rate": 0.00013038634552508063,
      "loss": 0.478,
      "step": 5575
    },
    {
      "epoch": 1.2648221343873518,
      "grad_norm": 0.09682100266218185,
      "learning_rate": 0.00012980492652274234,
      "loss": 0.5667,
      "step": 5600
    },
    {
      "epoch": 1.2704686617730097,
      "grad_norm": 0.1254328489303589,
      "learning_rate": 0.00012922239962213637,
      "loss": 0.5116,
      "step": 5625
    },
    {
      "epoch": 1.2761151891586673,
      "grad_norm": 0.09513936936855316,
      "learning_rate": 0.00012863878647674816,
      "loss": 0.5433,
      "step": 5650
    },
    {
      "epoch": 1.2817617165443251,
      "grad_norm": 0.10922129452228546,
      "learning_rate": 0.00012805410878044074,
      "loss": 0.4867,
      "step": 5675
    },
    {
      "epoch": 1.287408243929983,
      "grad_norm": 0.09911426901817322,
      "learning_rate": 0.00012746838826664826,
      "loss": 0.5785,
      "step": 5700
    },
    {
      "epoch": 1.2930547713156408,
      "grad_norm": 0.11289256066083908,
      "learning_rate": 0.00012688164670756802,
      "loss": 0.4761,
      "step": 5725
    },
    {
      "epoch": 1.2987012987012987,
      "grad_norm": 0.08862913399934769,
      "learning_rate": 0.00012629390591335134,
      "loss": 0.5743,
      "step": 5750
    },
    {
      "epoch": 1.3043478260869565,
      "grad_norm": 0.13005360960960388,
      "learning_rate": 0.00012570518773129277,
      "loss": 0.49,
      "step": 5775
    },
    {
      "epoch": 1.3099943534726144,
      "grad_norm": 0.08972469717264175,
      "learning_rate": 0.0001251155140450179,
      "loss": 0.5443,
      "step": 5800
    },
    {
      "epoch": 1.3156408808582722,
      "grad_norm": 0.11599931865930557,
      "learning_rate": 0.00012452490677367003,
      "loss": 0.5138,
      "step": 5825
    },
    {
      "epoch": 1.32128740824393,
      "grad_norm": 0.096194326877594,
      "learning_rate": 0.0001239333878710954,
      "loss": 0.5524,
      "step": 5850
    },
    {
      "epoch": 1.3269339356295877,
      "grad_norm": 0.11309222131967545,
      "learning_rate": 0.00012334097932502702,
      "loss": 0.485,
      "step": 5875
    },
    {
      "epoch": 1.3325804630152456,
      "grad_norm": 0.09447719901800156,
      "learning_rate": 0.00012274770315626743,
      "loss": 0.5748,
      "step": 5900
    },
    {
      "epoch": 1.3382269904009034,
      "grad_norm": 0.11994371563196182,
      "learning_rate": 0.00012215358141787016,
      "loss": 0.4826,
      "step": 5925
    },
    {
      "epoch": 1.3438735177865613,
      "grad_norm": 0.10158982127904892,
      "learning_rate": 0.00012155863619431993,
      "loss": 0.5593,
      "step": 5950
    },
    {
      "epoch": 1.3495200451722191,
      "grad_norm": 0.1210799366235733,
      "learning_rate": 0.00012096288960071178,
      "loss": 0.4977,
      "step": 5975
    },
    {
      "epoch": 1.355166572557877,
      "grad_norm": 0.0946585088968277,
      "learning_rate": 0.00012036636378192902,
      "loss": 0.5617,
      "step": 6000
    },
    {
      "epoch": 1.3608130999435346,
      "grad_norm": 0.10858285427093506,
      "learning_rate": 0.00011976908091181998,
      "loss": 0.468,
      "step": 6025
    },
    {
      "epoch": 1.3664596273291925,
      "grad_norm": 0.08753529191017151,
      "learning_rate": 0.00011917106319237386,
      "loss": 0.5854,
      "step": 6050
    },
    {
      "epoch": 1.3721061547148503,
      "grad_norm": 0.1213916763663292,
      "learning_rate": 0.00011857233285289546,
      "loss": 0.4915,
      "step": 6075
    },
    {
      "epoch": 1.3777526821005082,
      "grad_norm": 0.09568001329898834,
      "learning_rate": 0.00011797291214917881,
      "loss": 0.5699,
      "step": 6100
    },
    {
      "epoch": 1.383399209486166,
      "grad_norm": 0.10872907191514969,
      "learning_rate": 0.00011737282336267992,
      "loss": 0.4744,
      "step": 6125
    },
    {
      "epoch": 1.3890457368718239,
      "grad_norm": 0.1102764755487442,
      "learning_rate": 0.00011677208879968858,
      "loss": 0.5315,
      "step": 6150
    },
    {
      "epoch": 1.3946922642574817,
      "grad_norm": 0.1172797903418541,
      "learning_rate": 0.00011617073079049905,
      "loss": 0.493,
      "step": 6175
    },
    {
      "epoch": 1.4003387916431396,
      "grad_norm": 0.09269782900810242,
      "learning_rate": 0.0001155687716885802,
      "loss": 0.5383,
      "step": 6200
    },
    {
      "epoch": 1.4059853190287974,
      "grad_norm": 0.11610530316829681,
      "learning_rate": 0.00011496623386974454,
      "loss": 0.4748,
      "step": 6225
    },
    {
      "epoch": 1.411631846414455,
      "grad_norm": 0.10046471655368805,
      "learning_rate": 0.00011436313973131634,
      "loss": 0.5397,
      "step": 6250
    },
    {
      "epoch": 1.417278373800113,
      "grad_norm": 0.1363869458436966,
      "learning_rate": 0.00011375951169129926,
      "loss": 0.4944,
      "step": 6275
    },
    {
      "epoch": 1.4229249011857708,
      "grad_norm": 0.09395238012075424,
      "learning_rate": 0.00011315537218754295,
      "loss": 0.5614,
      "step": 6300
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 0.11986027657985687,
      "learning_rate": 0.00011255074367690897,
      "loss": 0.4914,
      "step": 6325
    },
    {
      "epoch": 1.4342179559570865,
      "grad_norm": 0.0886669009923935,
      "learning_rate": 0.0001119456486344361,
      "loss": 0.5615,
      "step": 6350
    },
    {
      "epoch": 1.439864483342744,
      "grad_norm": 0.12676212191581726,
      "learning_rate": 0.00011134010955250491,
      "loss": 0.4836,
      "step": 6375
    },
    {
      "epoch": 1.445511010728402,
      "grad_norm": 0.09607352316379547,
      "learning_rate": 0.00011073414894000161,
      "loss": 0.5505,
      "step": 6400
    },
    {
      "epoch": 1.4511575381140598,
      "grad_norm": 0.12430471181869507,
      "learning_rate": 0.00011012778932148142,
      "loss": 0.4949,
      "step": 6425
    },
    {
      "epoch": 1.4568040654997176,
      "grad_norm": 0.10339995473623276,
      "learning_rate": 0.00010952105323633126,
      "loss": 0.5583,
      "step": 6450
    },
    {
      "epoch": 1.4624505928853755,
      "grad_norm": 0.11074826866388321,
      "learning_rate": 0.00010891396323793189,
      "loss": 0.4973,
      "step": 6475
    },
    {
      "epoch": 1.4680971202710333,
      "grad_norm": 0.09366384148597717,
      "learning_rate": 0.00010830654189281968,
      "loss": 0.5329,
      "step": 6500
    },
    {
      "epoch": 1.4737436476566912,
      "grad_norm": 0.12167912721633911,
      "learning_rate": 0.00010769881177984771,
      "loss": 0.4901,
      "step": 6525
    },
    {
      "epoch": 1.479390175042349,
      "grad_norm": 0.09269159287214279,
      "learning_rate": 0.0001070907954893464,
      "loss": 0.5685,
      "step": 6550
    },
    {
      "epoch": 1.485036702428007,
      "grad_norm": 0.11428316682577133,
      "learning_rate": 0.00010648251562228386,
      "loss": 0.5,
      "step": 6575
    },
    {
      "epoch": 1.4906832298136645,
      "grad_norm": 0.10255276411771774,
      "learning_rate": 0.00010587399478942592,
      "loss": 0.5492,
      "step": 6600
    },
    {
      "epoch": 1.4963297571993224,
      "grad_norm": 0.10394936800003052,
      "learning_rate": 0.0001052652556104953,
      "loss": 0.4788,
      "step": 6625
    },
    {
      "epoch": 1.5019762845849802,
      "grad_norm": 0.0881161019206047,
      "learning_rate": 0.00010465632071333113,
      "loss": 0.5606,
      "step": 6650
    },
    {
      "epoch": 1.507622811970638,
      "grad_norm": 0.1109280213713646,
      "learning_rate": 0.00010404721273304769,
      "loss": 0.5183,
      "step": 6675
    },
    {
      "epoch": 1.513269339356296,
      "grad_norm": 0.10558915883302689,
      "learning_rate": 0.00010343795431119304,
      "loss": 0.5563,
      "step": 6700
    },
    {
      "epoch": 1.5189158667419536,
      "grad_norm": 0.13079263269901276,
      "learning_rate": 0.00010282856809490739,
      "loss": 0.4977,
      "step": 6725
    },
    {
      "epoch": 1.5245623941276114,
      "grad_norm": 0.09773046523332596,
      "learning_rate": 0.00010221907673608133,
      "loss": 0.5698,
      "step": 6750
    },
    {
      "epoch": 1.5302089215132693,
      "grad_norm": 0.11996293067932129,
      "learning_rate": 0.00010160950289051365,
      "loss": 0.4764,
      "step": 6775
    },
    {
      "epoch": 1.5358554488989271,
      "grad_norm": 0.12360116094350815,
      "learning_rate": 0.00010099986921706946,
      "loss": 0.5274,
      "step": 6800
    },
    {
      "epoch": 1.541501976284585,
      "grad_norm": 0.1305209845304489,
      "learning_rate": 0.00010039019837683767,
      "loss": 0.4531,
      "step": 6825
    },
    {
      "epoch": 1.5471485036702428,
      "grad_norm": 0.10246080160140991,
      "learning_rate": 9.978051303228875e-05,
      "loss": 0.5593,
      "step": 6850
    },
    {
      "epoch": 1.5527950310559007,
      "grad_norm": 0.13810117542743683,
      "learning_rate": 9.917083584643235e-05,
      "loss": 0.4838,
      "step": 6875
    },
    {
      "epoch": 1.5584415584415585,
      "grad_norm": 0.10199406743049622,
      "learning_rate": 9.856118948197488e-05,
      "loss": 0.5405,
      "step": 6900
    },
    {
      "epoch": 1.5640880858272164,
      "grad_norm": 0.11060495674610138,
      "learning_rate": 9.795159660047697e-05,
      "loss": 0.4974,
      "step": 6925
    },
    {
      "epoch": 1.5697346132128742,
      "grad_norm": 0.10460948944091797,
      "learning_rate": 9.734207986151126e-05,
      "loss": 0.5568,
      "step": 6950
    },
    {
      "epoch": 1.5753811405985318,
      "grad_norm": 0.11652641743421555,
      "learning_rate": 9.673266192182008e-05,
      "loss": 0.4683,
      "step": 6975
    },
    {
      "epoch": 1.5810276679841897,
      "grad_norm": 0.09894441068172455,
      "learning_rate": 9.612336543447314e-05,
      "loss": 0.5598,
      "step": 7000
    },
    {
      "epoch": 1.5866741953698476,
      "grad_norm": 0.10987823456525803,
      "learning_rate": 9.551421304802565e-05,
      "loss": 0.4582,
      "step": 7025
    },
    {
      "epoch": 1.5923207227555054,
      "grad_norm": 0.09542589634656906,
      "learning_rate": 9.490522740567633e-05,
      "loss": 0.5482,
      "step": 7050
    },
    {
      "epoch": 1.597967250141163,
      "grad_norm": 0.12428826838731766,
      "learning_rate": 9.42964311444257e-05,
      "loss": 0.4999,
      "step": 7075
    },
    {
      "epoch": 1.6036137775268209,
      "grad_norm": 0.10819413512945175,
      "learning_rate": 9.368784689423467e-05,
      "loss": 0.5652,
      "step": 7100
    },
    {
      "epoch": 1.6092603049124787,
      "grad_norm": 0.13115598261356354,
      "learning_rate": 9.307949727718346e-05,
      "loss": 0.4771,
      "step": 7125
    },
    {
      "epoch": 1.6149068322981366,
      "grad_norm": 0.1047302857041359,
      "learning_rate": 9.24714049066305e-05,
      "loss": 0.5376,
      "step": 7150
    },
    {
      "epoch": 1.6205533596837944,
      "grad_norm": 0.13034017384052277,
      "learning_rate": 9.186359238637197e-05,
      "loss": 0.4863,
      "step": 7175
    },
    {
      "epoch": 1.6261998870694523,
      "grad_norm": 0.10137925297021866,
      "learning_rate": 9.12560823098015e-05,
      "loss": 0.5442,
      "step": 7200
    },
    {
      "epoch": 1.6318464144551101,
      "grad_norm": 0.12191277742385864,
      "learning_rate": 9.064889725907043e-05,
      "loss": 0.4717,
      "step": 7225
    },
    {
      "epoch": 1.637492941840768,
      "grad_norm": 0.10191314667463303,
      "learning_rate": 9.004205980424842e-05,
      "loss": 0.5457,
      "step": 7250
    },
    {
      "epoch": 1.6431394692264258,
      "grad_norm": 0.11787448078393936,
      "learning_rate": 8.943559250248426e-05,
      "loss": 0.4858,
      "step": 7275
    },
    {
      "epoch": 1.6487859966120837,
      "grad_norm": 0.1057475134730339,
      "learning_rate": 8.88295178971677e-05,
      "loss": 0.5735,
      "step": 7300
    },
    {
      "epoch": 1.6544325239977415,
      "grad_norm": 0.12938141822814941,
      "learning_rate": 8.822385851709125e-05,
      "loss": 0.4829,
      "step": 7325
    },
    {
      "epoch": 1.6600790513833992,
      "grad_norm": 0.10724066197872162,
      "learning_rate": 8.761863687561275e-05,
      "loss": 0.5408,
      "step": 7350
    },
    {
      "epoch": 1.665725578769057,
      "grad_norm": 0.11694779992103577,
      "learning_rate": 8.701387546981868e-05,
      "loss": 0.4961,
      "step": 7375
    },
    {
      "epoch": 1.6713721061547149,
      "grad_norm": 0.10590225458145142,
      "learning_rate": 8.640959677968778e-05,
      "loss": 0.5466,
      "step": 7400
    },
    {
      "epoch": 1.6770186335403725,
      "grad_norm": 0.11699523031711578,
      "learning_rate": 8.580582326725535e-05,
      "loss": 0.4709,
      "step": 7425
    },
    {
      "epoch": 1.6826651609260304,
      "grad_norm": 0.1064140573143959,
      "learning_rate": 8.520257737577854e-05,
      "loss": 0.5349,
      "step": 7450
    },
    {
      "epoch": 1.6883116883116882,
      "grad_norm": 0.12415055185556412,
      "learning_rate": 8.459988152890188e-05,
      "loss": 0.4655,
      "step": 7475
    },
    {
      "epoch": 1.693958215697346,
      "grad_norm": 0.11068243533372879,
      "learning_rate": 8.39977581298239e-05,
      "loss": 0.5459,
      "step": 7500
    },
    {
      "epoch": 1.699604743083004,
      "grad_norm": 0.12139423191547394,
      "learning_rate": 8.339622956046417e-05,
      "loss": 0.4657,
      "step": 7525
    },
    {
      "epoch": 1.7052512704686618,
      "grad_norm": 0.10302968323230743,
      "learning_rate": 8.27953181806316e-05,
      "loss": 0.5571,
      "step": 7550
    },
    {
      "epoch": 1.7108977978543196,
      "grad_norm": 0.12433881312608719,
      "learning_rate": 8.21950463271931e-05,
      "loss": 0.4887,
      "step": 7575
    },
    {
      "epoch": 1.7165443252399775,
      "grad_norm": 0.09217355400323868,
      "learning_rate": 8.159543631324327e-05,
      "loss": 0.5281,
      "step": 7600
    },
    {
      "epoch": 1.7221908526256353,
      "grad_norm": 0.11367136240005493,
      "learning_rate": 8.099651042727515e-05,
      "loss": 0.4849,
      "step": 7625
    },
    {
      "epoch": 1.7278373800112932,
      "grad_norm": 0.10472942888736725,
      "learning_rate": 8.039829093235156e-05,
      "loss": 0.5748,
      "step": 7650
    },
    {
      "epoch": 1.733483907396951,
      "grad_norm": 0.13037872314453125,
      "learning_rate": 7.980080006527751e-05,
      "loss": 0.4781,
      "step": 7675
    },
    {
      "epoch": 1.7391304347826086,
      "grad_norm": 0.1035398542881012,
      "learning_rate": 7.920406003577394e-05,
      "loss": 0.5419,
      "step": 7700
    },
    {
      "epoch": 1.7447769621682665,
      "grad_norm": 0.12669602036476135,
      "learning_rate": 7.86080930256517e-05,
      "loss": 0.4865,
      "step": 7725
    },
    {
      "epoch": 1.7504234895539243,
      "grad_norm": 0.09713231027126312,
      "learning_rate": 7.801292118798732e-05,
      "loss": 0.564,
      "step": 7750
    },
    {
      "epoch": 1.7560700169395822,
      "grad_norm": 0.12088830769062042,
      "learning_rate": 7.74185666462995e-05,
      "loss": 0.4641,
      "step": 7775
    },
    {
      "epoch": 1.7617165443252398,
      "grad_norm": 0.11002447456121445,
      "learning_rate": 7.68250514937266e-05,
      "loss": 0.5407,
      "step": 7800
    },
    {
      "epoch": 1.7673630717108977,
      "grad_norm": 0.12338761240243912,
      "learning_rate": 7.623239779220557e-05,
      "loss": 0.4558,
      "step": 7825
    },
    {
      "epoch": 1.7730095990965555,
      "grad_norm": 0.10458722710609436,
      "learning_rate": 7.564062757165183e-05,
      "loss": 0.55,
      "step": 7850
    },
    {
      "epoch": 1.7786561264822134,
      "grad_norm": 0.1287498027086258,
      "learning_rate": 7.504976282914027e-05,
      "loss": 0.4745,
      "step": 7875
    },
    {
      "epoch": 1.7843026538678712,
      "grad_norm": 0.10165251046419144,
      "learning_rate": 7.445982552808774e-05,
      "loss": 0.567,
      "step": 7900
    },
    {
      "epoch": 1.789949181253529,
      "grad_norm": 0.12862220406532288,
      "learning_rate": 7.387083759743655e-05,
      "loss": 0.4777,
      "step": 7925
    },
    {
      "epoch": 1.795595708639187,
      "grad_norm": 0.1015312671661377,
      "learning_rate": 7.328282093083929e-05,
      "loss": 0.5648,
      "step": 7950
    },
    {
      "epoch": 1.8012422360248448,
      "grad_norm": 0.1260487586259842,
      "learning_rate": 7.269579738584513e-05,
      "loss": 0.4895,
      "step": 7975
    },
    {
      "epoch": 1.8068887634105026,
      "grad_norm": 0.10070119798183441,
      "learning_rate": 7.210978878308729e-05,
      "loss": 0.581,
      "step": 8000
    },
    {
      "epoch": 1.8125352907961605,
      "grad_norm": 0.12946705520153046,
      "learning_rate": 7.152481690547182e-05,
      "loss": 0.475,
      "step": 8025
    },
    {
      "epoch": 1.8181818181818183,
      "grad_norm": 0.09579546749591827,
      "learning_rate": 7.094090349736803e-05,
      "loss": 0.5516,
      "step": 8050
    },
    {
      "epoch": 1.823828345567476,
      "grad_norm": 0.12078605592250824,
      "learning_rate": 7.035807026380026e-05,
      "loss": 0.4882,
      "step": 8075
    },
    {
      "epoch": 1.8294748729531338,
      "grad_norm": 0.10442069917917252,
      "learning_rate": 6.977633886964081e-05,
      "loss": 0.5322,
      "step": 8100
    },
    {
      "epoch": 1.8351214003387917,
      "grad_norm": 0.13175725936889648,
      "learning_rate": 6.919573093880494e-05,
      "loss": 0.4578,
      "step": 8125
    },
    {
      "epoch": 1.8407679277244493,
      "grad_norm": 0.10551594197750092,
      "learning_rate": 6.861626805344689e-05,
      "loss": 0.547,
      "step": 8150
    },
    {
      "epoch": 1.8464144551101072,
      "grad_norm": 0.11718405783176422,
      "learning_rate": 6.803797175315761e-05,
      "loss": 0.4872,
      "step": 8175
    },
    {
      "epoch": 1.852060982495765,
      "grad_norm": 0.12317229807376862,
      "learning_rate": 6.74608635341642e-05,
      "loss": 0.5724,
      "step": 8200
    },
    {
      "epoch": 1.8577075098814229,
      "grad_norm": 0.13172867894172668,
      "learning_rate": 6.688496484853084e-05,
      "loss": 0.4506,
      "step": 8225
    },
    {
      "epoch": 1.8633540372670807,
      "grad_norm": 0.09928746521472931,
      "learning_rate": 6.631029710336133e-05,
      "loss": 0.5094,
      "step": 8250
    },
    {
      "epoch": 1.8690005646527386,
      "grad_norm": 0.12048971652984619,
      "learning_rate": 6.573688166000345e-05,
      "loss": 0.4608,
      "step": 8275
    },
    {
      "epoch": 1.8746470920383964,
      "grad_norm": 0.1059262603521347,
      "learning_rate": 6.516473983325473e-05,
      "loss": 0.5475,
      "step": 8300
    },
    {
      "epoch": 1.8802936194240543,
      "grad_norm": 0.1448829621076584,
      "learning_rate": 6.459389289057038e-05,
      "loss": 0.4695,
      "step": 8325
    },
    {
      "epoch": 1.8859401468097121,
      "grad_norm": 0.0976184606552124,
      "learning_rate": 6.40243620512726e-05,
      "loss": 0.549,
      "step": 8350
    },
    {
      "epoch": 1.89158667419537,
      "grad_norm": 0.14233841001987457,
      "learning_rate": 6.345616848576184e-05,
      "loss": 0.4716,
      "step": 8375
    },
    {
      "epoch": 1.8972332015810278,
      "grad_norm": 0.10030635446310043,
      "learning_rate": 6.288933331472988e-05,
      "loss": 0.561,
      "step": 8400
    },
    {
      "epoch": 1.9028797289666854,
      "grad_norm": 0.12384554743766785,
      "learning_rate": 6.232387760837474e-05,
      "loss": 0.4872,
      "step": 8425
    },
    {
      "epoch": 1.9085262563523433,
      "grad_norm": 0.10905805230140686,
      "learning_rate": 6.175982238561755e-05,
      "loss": 0.5347,
      "step": 8450
    },
    {
      "epoch": 1.9141727837380011,
      "grad_norm": 0.13816171884536743,
      "learning_rate": 6.119718861332098e-05,
      "loss": 0.477,
      "step": 8475
    },
    {
      "epoch": 1.919819311123659,
      "grad_norm": 0.11039167642593384,
      "learning_rate": 6.0635997205510175e-05,
      "loss": 0.5706,
      "step": 8500
    },
    {
      "epoch": 1.9254658385093166,
      "grad_norm": 0.1241937056183815,
      "learning_rate": 6.007626902259521e-05,
      "loss": 0.4787,
      "step": 8525
    },
    {
      "epoch": 1.9311123658949745,
      "grad_norm": 0.1162077784538269,
      "learning_rate": 5.951802487059559e-05,
      "loss": 0.5515,
      "step": 8550
    },
    {
      "epoch": 1.9367588932806323,
      "grad_norm": 0.12294231355190277,
      "learning_rate": 5.8961285500367034e-05,
      "loss": 0.4699,
      "step": 8575
    },
    {
      "epoch": 1.9424054206662902,
      "grad_norm": 0.10771624743938446,
      "learning_rate": 5.8406071606830026e-05,
      "loss": 0.5373,
      "step": 8600
    },
    {
      "epoch": 1.948051948051948,
      "grad_norm": 0.11943277716636658,
      "learning_rate": 5.7852403828200495e-05,
      "loss": 0.4648,
      "step": 8625
    },
    {
      "epoch": 1.9536984754376059,
      "grad_norm": 0.11853492259979248,
      "learning_rate": 5.730030274522282e-05,
      "loss": 0.5501,
      "step": 8650
    },
    {
      "epoch": 1.9593450028232637,
      "grad_norm": 0.12967811524868011,
      "learning_rate": 5.674978888040463e-05,
      "loss": 0.4808,
      "step": 8675
    },
    {
      "epoch": 1.9649915302089216,
      "grad_norm": 0.112027108669281,
      "learning_rate": 5.6200882697254154e-05,
      "loss": 0.5311,
      "step": 8700
    },
    {
      "epoch": 1.9706380575945794,
      "grad_norm": 0.1347729116678238,
      "learning_rate": 5.565360459951936e-05,
      "loss": 0.4679,
      "step": 8725
    },
    {
      "epoch": 1.9762845849802373,
      "grad_norm": 0.11015453189611435,
      "learning_rate": 5.510797493042954e-05,
      "loss": 0.5595,
      "step": 8750
    },
    {
      "epoch": 1.981931112365895,
      "grad_norm": 0.13183894753456116,
      "learning_rate": 5.456401397193936e-05,
      "loss": 0.4656,
      "step": 8775
    },
    {
      "epoch": 1.9875776397515528,
      "grad_norm": 0.10906127840280533,
      "learning_rate": 5.402174194397458e-05,
      "loss": 0.537,
      "step": 8800
    },
    {
      "epoch": 1.9932241671372106,
      "grad_norm": 0.1271781325340271,
      "learning_rate": 5.348117900368066e-05,
      "loss": 0.4639,
      "step": 8825
    },
    {
      "epoch": 1.9988706945228685,
      "grad_norm": 0.12033294886350632,
      "learning_rate": 5.2942345244673564e-05,
      "loss": 0.5076,
      "step": 8850
    },
    {
      "epoch": 2.004517221908526,
      "grad_norm": 0.11611133068799973,
      "learning_rate": 5.240526069629265e-05,
      "loss": 0.5368,
      "step": 8875
    },
    {
      "epoch": 2.010163749294184,
      "grad_norm": 0.16251997649669647,
      "learning_rate": 5.1869945322856196e-05,
      "loss": 0.4708,
      "step": 8900
    },
    {
      "epoch": 2.015810276679842,
      "grad_norm": 0.10342962294816971,
      "learning_rate": 5.1336419022919435e-05,
      "loss": 0.5186,
      "step": 8925
    },
    {
      "epoch": 2.0214568040654997,
      "grad_norm": 0.12473160028457642,
      "learning_rate": 5.080470162853472e-05,
      "loss": 0.4405,
      "step": 8950
    },
    {
      "epoch": 2.0271033314511575,
      "grad_norm": 0.10961506515741348,
      "learning_rate": 5.0274812904514346e-05,
      "loss": 0.4963,
      "step": 8975
    },
    {
      "epoch": 2.0327498588368154,
      "grad_norm": 0.13045734167099,
      "learning_rate": 4.974677254769608e-05,
      "loss": 0.4648,
      "step": 9000
    },
    {
      "epoch": 2.038396386222473,
      "grad_norm": 0.10588447749614716,
      "learning_rate": 4.922060018621066e-05,
      "loss": 0.514,
      "step": 9025
    },
    {
      "epoch": 2.044042913608131,
      "grad_norm": 0.1310020089149475,
      "learning_rate": 4.869631537875243e-05,
      "loss": 0.4581,
      "step": 9050
    },
    {
      "epoch": 2.049689440993789,
      "grad_norm": 0.11264315247535706,
      "learning_rate": 4.8173937613852296e-05,
      "loss": 0.514,
      "step": 9075
    },
    {
      "epoch": 2.0553359683794468,
      "grad_norm": 0.11348750442266464,
      "learning_rate": 4.765348630915315e-05,
      "loss": 0.4509,
      "step": 9100
    },
    {
      "epoch": 2.0609824957651046,
      "grad_norm": 0.1084527000784874,
      "learning_rate": 4.713498081068819e-05,
      "loss": 0.4836,
      "step": 9125
    },
    {
      "epoch": 2.0666290231507625,
      "grad_norm": 0.12957565486431122,
      "learning_rate": 4.6618440392161886e-05,
      "loss": 0.4467,
      "step": 9150
    },
    {
      "epoch": 2.0722755505364203,
      "grad_norm": 0.10537248849868774,
      "learning_rate": 4.610388425423336e-05,
      "loss": 0.52,
      "step": 9175
    },
    {
      "epoch": 2.0779220779220777,
      "grad_norm": 0.13615071773529053,
      "learning_rate": 4.559133152380272e-05,
      "loss": 0.4567,
      "step": 9200
    },
    {
      "epoch": 2.0835686053077356,
      "grad_norm": 0.11438791453838348,
      "learning_rate": 4.508080125330022e-05,
      "loss": 0.5111,
      "step": 9225
    },
    {
      "epoch": 2.0892151326933934,
      "grad_norm": 0.1288277804851532,
      "learning_rate": 4.457231241997788e-05,
      "loss": 0.4584,
      "step": 9250
    },
    {
      "epoch": 2.0948616600790513,
      "grad_norm": 0.10171278566122055,
      "learning_rate": 4.40658839252041e-05,
      "loss": 0.5065,
      "step": 9275
    },
    {
      "epoch": 2.100508187464709,
      "grad_norm": 0.1178368479013443,
      "learning_rate": 4.356153459376121e-05,
      "loss": 0.4426,
      "step": 9300
    },
    {
      "epoch": 2.106154714850367,
      "grad_norm": 0.12345319241285324,
      "learning_rate": 4.305928317314549e-05,
      "loss": 0.5134,
      "step": 9325
    },
    {
      "epoch": 2.111801242236025,
      "grad_norm": 0.1341843158006668,
      "learning_rate": 4.255914833287046e-05,
      "loss": 0.4526,
      "step": 9350
    },
    {
      "epoch": 2.1174477696216827,
      "grad_norm": 0.12422988563776016,
      "learning_rate": 4.206114866377291e-05,
      "loss": 0.5025,
      "step": 9375
    },
    {
      "epoch": 2.1230942970073405,
      "grad_norm": 0.1362808346748352,
      "learning_rate": 4.156530267732173e-05,
      "loss": 0.4701,
      "step": 9400
    },
    {
      "epoch": 2.1287408243929984,
      "grad_norm": 0.11891162395477295,
      "learning_rate": 4.107162880492984e-05,
      "loss": 0.5091,
      "step": 9425
    },
    {
      "epoch": 2.1343873517786562,
      "grad_norm": 0.13123708963394165,
      "learning_rate": 4.058014539726922e-05,
      "loss": 0.4593,
      "step": 9450
    },
    {
      "epoch": 2.140033879164314,
      "grad_norm": 0.11908406764268875,
      "learning_rate": 4.0090870723588606e-05,
      "loss": 0.5076,
      "step": 9475
    },
    {
      "epoch": 2.145680406549972,
      "grad_norm": 0.1421327143907547,
      "learning_rate": 3.960382297103442e-05,
      "loss": 0.4467,
      "step": 9500
    },
    {
      "epoch": 2.15132693393563,
      "grad_norm": 0.1141195222735405,
      "learning_rate": 3.911902024397473e-05,
      "loss": 0.5044,
      "step": 9525
    },
    {
      "epoch": 2.1569734613212876,
      "grad_norm": 0.11890695989131927,
      "learning_rate": 3.8636480563326425e-05,
      "loss": 0.4562,
      "step": 9550
    },
    {
      "epoch": 2.162619988706945,
      "grad_norm": 0.12307918071746826,
      "learning_rate": 3.8156221865885126e-05,
      "loss": 0.5154,
      "step": 9575
    },
    {
      "epoch": 2.168266516092603,
      "grad_norm": 0.12713749706745148,
      "learning_rate": 3.767826200365853e-05,
      "loss": 0.4578,
      "step": 9600
    },
    {
      "epoch": 2.1739130434782608,
      "grad_norm": 0.12245321273803711,
      "learning_rate": 3.7202618743202935e-05,
      "loss": 0.5022,
      "step": 9625
    },
    {
      "epoch": 2.1795595708639186,
      "grad_norm": 0.1354779750108719,
      "learning_rate": 3.6729309764962616e-05,
      "loss": 0.4534,
      "step": 9650
    },
    {
      "epoch": 2.1852060982495765,
      "grad_norm": 0.10722421109676361,
      "learning_rate": 3.625835266261287e-05,
      "loss": 0.5021,
      "step": 9675
    },
    {
      "epoch": 2.1908526256352343,
      "grad_norm": 0.1329207420349121,
      "learning_rate": 3.578976494240577e-05,
      "loss": 0.4639,
      "step": 9700
    },
    {
      "epoch": 2.196499153020892,
      "grad_norm": 0.11663591116666794,
      "learning_rate": 3.532356402251954e-05,
      "loss": 0.4882,
      "step": 9725
    },
    {
      "epoch": 2.20214568040655,
      "grad_norm": 0.13668714463710785,
      "learning_rate": 3.485976723241121e-05,
      "loss": 0.465,
      "step": 9750
    },
    {
      "epoch": 2.207792207792208,
      "grad_norm": 0.11890459805727005,
      "learning_rate": 3.439839181217227e-05,
      "loss": 0.505,
      "step": 9775
    },
    {
      "epoch": 2.2134387351778657,
      "grad_norm": 0.1332354098558426,
      "learning_rate": 3.3939454911887844e-05,
      "loss": 0.4497,
      "step": 9800
    },
    {
      "epoch": 2.2190852625635236,
      "grad_norm": 0.11797164380550385,
      "learning_rate": 3.34829735909994e-05,
      "loss": 0.536,
      "step": 9825
    },
    {
      "epoch": 2.2247317899491814,
      "grad_norm": 0.13317126035690308,
      "learning_rate": 3.302896481767034e-05,
      "loss": 0.4645,
      "step": 9850
    },
    {
      "epoch": 2.2303783173348393,
      "grad_norm": 0.11745914816856384,
      "learning_rate": 3.25774454681554e-05,
      "loss": 0.4991,
      "step": 9875
    },
    {
      "epoch": 2.2360248447204967,
      "grad_norm": 0.13392220437526703,
      "learning_rate": 3.212843232617343e-05,
      "loss": 0.4385,
      "step": 9900
    },
    {
      "epoch": 2.2416713721061545,
      "grad_norm": 0.11925249546766281,
      "learning_rate": 3.168194208228331e-05,
      "loss": 0.4801,
      "step": 9925
    },
    {
      "epoch": 2.2473178994918124,
      "grad_norm": 0.1346205770969391,
      "learning_rate": 3.123799133326366e-05,
      "loss": 0.4562,
      "step": 9950
    },
    {
      "epoch": 2.2529644268774702,
      "grad_norm": 0.1318216472864151,
      "learning_rate": 3.0796596581495963e-05,
      "loss": 0.499,
      "step": 9975
    },
    {
      "epoch": 2.258610954263128,
      "grad_norm": 0.1402980238199234,
      "learning_rate": 3.0357774234350945e-05,
      "loss": 0.4386,
      "step": 10000
    }
  ],
  "logging_steps": 25,
  "max_steps": 13281,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 5.882541565636669e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}