[
  {
    "loss": 1.5673105239868164,
    "grad_norm": 17.69691276550293,
    "learning_rate": 1.8e-05,
    "epoch": 0.0003215744284014535,
    "step": 10
  },
  {
    "loss": 0.3935260772705078,
    "grad_norm": 4.892251491546631,
    "learning_rate": 3.8e-05,
    "epoch": 0.000643148856802907,
    "step": 20
  },
  {
    "loss": 0.28224611282348633,
    "grad_norm": 3.7167558670043945,
    "learning_rate": 5.8e-05,
    "epoch": 0.0009647232852043606,
    "step": 30
  },
  {
    "loss": 0.21302554607391358,
    "grad_norm": 2.2682483196258545,
    "learning_rate": 7.800000000000001e-05,
    "epoch": 0.001286297713605814,
    "step": 40
  },
  {
    "loss": 0.19408349990844725,
    "grad_norm": 2.2181177139282227,
    "learning_rate": 9.8e-05,
    "epoch": 0.0016078721420072675,
    "step": 50
  },
  {
    "loss": 0.21122798919677735,
    "grad_norm": 1.552700400352478,
    "learning_rate": 0.000118,
    "epoch": 0.0019294465704087212,
    "step": 60
  },
  {
    "loss": 0.17427901029586793,
    "grad_norm": 1.7993711233139038,
    "learning_rate": 0.000138,
    "epoch": 0.0022510209988101746,
    "step": 70
  },
  {
    "loss": 0.18862361907958985,
    "grad_norm": 1.6078184843063354,
    "learning_rate": 0.00015800000000000002,
    "epoch": 0.002572595427211628,
    "step": 80
  },
  {
    "loss": 0.15153205394744873,
    "grad_norm": 3.307420492172241,
    "learning_rate": 0.00017800000000000002,
    "epoch": 0.0028941698556130815,
    "step": 90
  },
  {
    "loss": 0.19316819906234742,
    "grad_norm": 1.4174200296401978,
    "learning_rate": 0.00019800000000000002,
    "epoch": 0.003215744284014535,
    "step": 100
  },
  {
    "loss": 0.18397116661071777,
    "grad_norm": 2.051401138305664,
    "learning_rate": 0.00019999995839788484,
    "epoch": 0.003537318712415989,
    "step": 110
  },
  {
    "loss": 0.15460578203201295,
    "grad_norm": 1.2434544563293457,
    "learning_rate": 0.00019999981458814847,
    "epoch": 0.0038588931408174423,
    "step": 120
  },
  {
    "loss": 0.16346523761749268,
    "grad_norm": 2.292675256729126,
    "learning_rate": 0.00019999956805733223,
    "epoch": 0.004180467569218895,
    "step": 130
  },
  {
    "loss": 0.1633118748664856,
    "grad_norm": 1.3027211427688599,
    "learning_rate": 0.00019999921880568935,
    "epoch": 0.004502041997620349,
    "step": 140
  },
  {
    "loss": 0.14462895393371583,
    "grad_norm": 1.0612585544586182,
    "learning_rate": 0.00019999876683357855,
    "epoch": 0.004823616426021803,
    "step": 150
  },
  {
    "loss": 0.16672265529632568,
    "grad_norm": 1.4711265563964844,
    "learning_rate": 0.00019999821214146418,
    "epoch": 0.005145190854423256,
    "step": 160
  },
  {
    "loss": 0.1713968276977539,
    "grad_norm": 0.9445065259933472,
    "learning_rate": 0.00019999755472991594,
    "epoch": 0.00546676528282471,
    "step": 170
  },
  {
    "loss": 0.15581530332565308,
    "grad_norm": 1.4779735803604126,
    "learning_rate": 0.0001999967945996092,
    "epoch": 0.005788339711226163,
    "step": 180
  },
  {
    "loss": 0.15593582391738892,
    "grad_norm": 1.2822918891906738,
    "learning_rate": 0.00019999593175132476,
    "epoch": 0.006109914139627617,
    "step": 190
  },
  {
    "loss": 0.19365686178207397,
    "grad_norm": 1.0924097299575806,
    "learning_rate": 0.00019999496618594893,
    "epoch": 0.00643148856802907,
    "step": 200
  },
  {
    "loss": 0.15486612319946289,
    "grad_norm": 0.7953126430511475,
    "learning_rate": 0.00019999389790447355,
    "epoch": 0.006753062996430524,
    "step": 210
  },
  {
    "loss": 0.16763465404510497,
    "grad_norm": 1.3703619241714478,
    "learning_rate": 0.00019999272690799603,
    "epoch": 0.007074637424831978,
    "step": 220
  },
  {
    "loss": 0.15374833345413208,
    "grad_norm": 0.6760343909263611,
    "learning_rate": 0.00019999145319771913,
    "epoch": 0.007396211853233431,
    "step": 230
  },
  {
    "loss": 0.17234739065170288,
    "grad_norm": 1.6318228244781494,
    "learning_rate": 0.00019999007677495127,
    "epoch": 0.007717786281634885,
    "step": 240
  },
  {
    "loss": 0.1620724081993103,
    "grad_norm": 0.681673526763916,
    "learning_rate": 0.00019998859764110637,
    "epoch": 0.008039360710036338,
    "step": 250
  },
  {
    "loss": 0.16377369165420533,
    "grad_norm": 1.023579716682434,
    "learning_rate": 0.00019998701579770377,
    "epoch": 0.00836093513843779,
    "step": 260
  },
  {
    "loss": 0.15195246934890747,
    "grad_norm": 1.2718433141708374,
    "learning_rate": 0.00019998533124636838,
    "epoch": 0.008682509566839245,
    "step": 270
  },
  {
    "loss": 0.19328283071517943,
    "grad_norm": 1.8994773626327515,
    "learning_rate": 0.00019998354398883056,
    "epoch": 0.009004083995240698,
    "step": 280
  },
  {
    "loss": 0.13401613235473633,
    "grad_norm": 0.902221143245697,
    "learning_rate": 0.00019998165402692624,
    "epoch": 0.009325658423642151,
    "step": 290
  },
  {
    "loss": 0.17590770721435547,
    "grad_norm": 0.6986466646194458,
    "learning_rate": 0.0001999796613625968,
    "epoch": 0.009647232852043606,
    "step": 300
  },
  {
    "loss": 0.15391089916229247,
    "grad_norm": 1.071215033531189,
    "learning_rate": 0.00019997756599788913,
    "epoch": 0.00996880728044506,
    "step": 310
  },
  {
    "loss": 0.156918466091156,
    "grad_norm": 0.8470816612243652,
    "learning_rate": 0.0001999753679349556,
    "epoch": 0.010290381708846512,
    "step": 320
  },
  {
    "loss": 0.1681678533554077,
    "grad_norm": 1.6104198694229126,
    "learning_rate": 0.00019997306717605413,
    "epoch": 0.010611956137247965,
    "step": 330
  },
  {
    "loss": 0.13588624000549315,
    "grad_norm": 0.7630122900009155,
    "learning_rate": 0.00019997066372354804,
    "epoch": 0.01093353056564942,
    "step": 340
  },
  {
    "loss": 0.16677809953689576,
    "grad_norm": 1.205472707748413,
    "learning_rate": 0.0001999681575799062,
    "epoch": 0.011255104994050873,
    "step": 350
  },
  {
    "loss": 0.12492926120758056,
    "grad_norm": 0.7292776107788086,
    "learning_rate": 0.000199965548747703,
    "epoch": 0.011576679422452326,
    "step": 360
  },
  {
    "loss": 0.14817183017730712,
    "grad_norm": 0.6910330653190613,
    "learning_rate": 0.0001999628372296182,
    "epoch": 0.01189825385085378,
    "step": 370
  },
  {
    "loss": 0.15440335273742675,
    "grad_norm": 1.0198880434036255,
    "learning_rate": 0.00019996002302843712,
    "epoch": 0.012219828279255234,
    "step": 380
  },
  {
    "loss": 0.1331814169883728,
    "grad_norm": 0.7604308128356934,
    "learning_rate": 0.00019995710614705058,
    "epoch": 0.012541402707656687,
    "step": 390
  },
  {
    "loss": 0.10600050687789916,
    "grad_norm": 0.8126104474067688,
    "learning_rate": 0.00019995408658845482,
    "epoch": 0.01286297713605814,
    "step": 400
  },
  {
    "loss": 0.14562102556228637,
    "grad_norm": 0.6844388842582703,
    "learning_rate": 0.00019995096435575157,
    "epoch": 0.013184551564459595,
    "step": 410
  },
  {
    "loss": 0.12237757444381714,
    "grad_norm": 0.8213316798210144,
    "learning_rate": 0.00019994773945214797,
    "epoch": 0.013506125992861048,
    "step": 420
  },
  {
    "loss": 0.16479880809783937,
    "grad_norm": 1.0815225839614868,
    "learning_rate": 0.00019994441188095677,
    "epoch": 0.0138277004212625,
    "step": 430
  },
  {
    "loss": 0.14453794956207275,
    "grad_norm": 1.1022553443908691,
    "learning_rate": 0.00019994098164559601,
    "epoch": 0.014149274849663955,
    "step": 440
  },
  {
    "loss": 0.15858374834060668,
    "grad_norm": 0.820088267326355,
    "learning_rate": 0.00019993744874958937,
    "epoch": 0.014470849278065408,
    "step": 450
  },
  {
    "loss": 0.12721436023712157,
    "grad_norm": 1.0894126892089844,
    "learning_rate": 0.00019993381319656582,
    "epoch": 0.014792423706466861,
    "step": 460
  },
  {
    "loss": 0.15983102321624756,
    "grad_norm": 1.233209252357483,
    "learning_rate": 0.00019993007499025984,
    "epoch": 0.015113998134868314,
    "step": 470
  },
  {
    "loss": 0.12201125621795654,
    "grad_norm": 0.6897937655448914,
    "learning_rate": 0.0001999262341345114,
    "epoch": 0.01543557256326977,
    "step": 480
  },
  {
    "loss": 0.13091821670532228,
    "grad_norm": 1.1130845546722412,
    "learning_rate": 0.00019992229063326583,
    "epoch": 0.015757146991671224,
    "step": 490
  },
  {
    "loss": 0.11298969984054566,
    "grad_norm": 1.0091015100479126,
    "learning_rate": 0.000199918244490574,
    "epoch": 0.016078721420072675,
    "step": 500
  },
  {
    "loss": 0.133112633228302,
    "grad_norm": 0.7309413552284241,
    "learning_rate": 0.0001999140957105921,
    "epoch": 0.01640029584847413,
    "step": 510
  },
  {
    "loss": 0.13967225551605225,
    "grad_norm": 0.8078439235687256,
    "learning_rate": 0.00019990984429758187,
    "epoch": 0.01672187027687558,
    "step": 520
  },
  {
    "loss": 0.15573676824569702,
    "grad_norm": 0.8659370541572571,
    "learning_rate": 0.00019990549025591036,
    "epoch": 0.017043444705277036,
    "step": 530
  },
  {
    "loss": 0.1315290927886963,
    "grad_norm": 0.9505926966667175,
    "learning_rate": 0.00019990103359005014,
    "epoch": 0.01736501913367849,
    "step": 540
  },
  {
    "loss": 0.12422749996185303,
    "grad_norm": 0.8183798789978027,
    "learning_rate": 0.00019989647430457912,
    "epoch": 0.017686593562079942,
    "step": 550
  },
  {
    "loss": 0.14997694492340088,
    "grad_norm": 1.2214590311050415,
    "learning_rate": 0.00019989181240418069,
    "epoch": 0.018008167990481397,
    "step": 560
  },
  {
    "loss": 0.13226089477539063,
    "grad_norm": 0.6610310673713684,
    "learning_rate": 0.0001998870478936436,
    "epoch": 0.01832974241888285,
    "step": 570
  },
  {
    "loss": 0.13494133949279785,
    "grad_norm": 0.7559733986854553,
    "learning_rate": 0.000199882180777862,
    "epoch": 0.018651316847284303,
    "step": 580
  },
  {
    "loss": 0.13185596466064453,
    "grad_norm": 0.9044352173805237,
    "learning_rate": 0.00019987721106183546,
    "epoch": 0.018972891275685758,
    "step": 590
  },
  {
    "loss": 0.14558923244476318,
    "grad_norm": 0.9192641377449036,
    "learning_rate": 0.00019987213875066894,
    "epoch": 0.019294465704087212,
    "step": 600
  },
  {
    "loss": 0.1581730842590332,
    "grad_norm": 0.5637726187705994,
    "learning_rate": 0.0001998669638495728,
    "epoch": 0.019616040132488664,
    "step": 610
  },
  {
    "loss": 0.11631412506103515,
    "grad_norm": 0.7248178124427795,
    "learning_rate": 0.0001998616863638627,
    "epoch": 0.01993761456089012,
    "step": 620
  },
  {
    "loss": 0.13501194715499878,
    "grad_norm": 1.1470476388931274,
    "learning_rate": 0.00019985630629895985,
    "epoch": 0.020259188989291573,
    "step": 630
  },
  {
    "loss": 0.14213244915008544,
    "grad_norm": 1.4807144403457642,
    "learning_rate": 0.0001998508236603906,
    "epoch": 0.020580763417693024,
    "step": 640
  },
  {
    "loss": 0.14146710634231568,
    "grad_norm": 0.8103519678115845,
    "learning_rate": 0.0001998452384537869,
    "epoch": 0.02090233784609448,
    "step": 650
  },
  {
    "loss": 0.1671810507774353,
    "grad_norm": 1.4063045978546143,
    "learning_rate": 0.00019983955068488584,
    "epoch": 0.02122391227449593,
    "step": 660
  },
  {
    "loss": 0.1154186487197876,
    "grad_norm": 0.6153373122215271,
    "learning_rate": 0.00019983376035953004,
    "epoch": 0.021545486702897385,
    "step": 670
  },
  {
    "loss": 0.13083285093307495,
    "grad_norm": 0.7791343927383423,
    "learning_rate": 0.0001998278674836674,
    "epoch": 0.02186706113129884,
    "step": 680
  },
  {
    "loss": 0.09962020516395569,
    "grad_norm": 0.7736439108848572,
    "learning_rate": 0.00019982187206335107,
    "epoch": 0.02218863555970029,
    "step": 690
  },
  {
    "loss": 0.12725504636764526,
    "grad_norm": 0.6886352300643921,
    "learning_rate": 0.00019981577410473972,
    "epoch": 0.022510209988101746,
    "step": 700
  },
  {
    "loss": 0.13899757862091064,
    "grad_norm": 0.837552011013031,
    "learning_rate": 0.0001998095736140972,
    "epoch": 0.0228317844165032,
    "step": 710
  },
  {
    "loss": 0.15298879146575928,
    "grad_norm": 0.9249375462532043,
    "learning_rate": 0.00019980327059779276,
    "epoch": 0.023153358844904652,
    "step": 720
  },
  {
    "loss": 0.10059181451797486,
    "grad_norm": 0.5768576264381409,
    "learning_rate": 0.00019979686506230092,
    "epoch": 0.023474933273306107,
    "step": 730
  },
  {
    "loss": 0.13777350187301635,
    "grad_norm": 1.2064447402954102,
    "learning_rate": 0.0001997903570142015,
    "epoch": 0.02379650770170756,
    "step": 740
  },
  {
    "loss": 0.1392066478729248,
    "grad_norm": 0.5575463771820068,
    "learning_rate": 0.00019978374646017973,
    "epoch": 0.024118082130109013,
    "step": 750
  },
  {
    "loss": 0.14810303449630738,
    "grad_norm": 0.7672901153564453,
    "learning_rate": 0.00019977703340702595,
    "epoch": 0.024439656558510468,
    "step": 760
  },
  {
    "loss": 0.150481915473938,
    "grad_norm": 0.8861374258995056,
    "learning_rate": 0.00019977021786163598,
    "epoch": 0.024761230986911922,
    "step": 770
  },
  {
    "loss": 0.1382421851158142,
    "grad_norm": 0.7194581627845764,
    "learning_rate": 0.00019976329983101084,
    "epoch": 0.025082805415313374,
    "step": 780
  },
  {
    "loss": 0.12950103282928466,
    "grad_norm": 1.3259282112121582,
    "learning_rate": 0.00019975627932225675,
    "epoch": 0.02540437984371483,
    "step": 790
  },
  {
    "loss": 0.11794104576110839,
    "grad_norm": 0.7138015031814575,
    "learning_rate": 0.0001997491563425853,
    "epoch": 0.02572595427211628,
    "step": 800
  },
  {
    "loss": 0.11722338199615479,
    "grad_norm": 0.9002670049667358,
    "learning_rate": 0.00019974193089931328,
    "epoch": 0.026047528700517734,
    "step": 810
  },
  {
    "loss": 0.1378617525100708,
    "grad_norm": 0.7109375596046448,
    "learning_rate": 0.0001997346029998628,
    "epoch": 0.02636910312891919,
    "step": 820
  },
  {
    "loss": 0.1343904733657837,
    "grad_norm": 0.6565284729003906,
    "learning_rate": 0.00019972717265176115,
    "epoch": 0.02669067755732064,
    "step": 830
  },
  {
    "loss": 0.11249516010284424,
    "grad_norm": 0.5646419525146484,
    "learning_rate": 0.0001997196398626409,
    "epoch": 0.027012251985722095,
    "step": 840
  },
  {
    "loss": 0.11767094135284424,
    "grad_norm": 0.6782432198524475,
    "learning_rate": 0.00019971200464023977,
    "epoch": 0.02733382641412355,
    "step": 850
  },
  {
    "loss": 0.11552731990814209,
    "grad_norm": 0.748511016368866,
    "learning_rate": 0.00019970426699240083,
    "epoch": 0.027655400842525,
    "step": 860
  },
  {
    "loss": 0.12385071516036987,
    "grad_norm": 0.6928240060806274,
    "learning_rate": 0.00019969642692707225,
    "epoch": 0.027976975270926456,
    "step": 870
  },
  {
    "loss": 0.13992183208465575,
    "grad_norm": 1.11572265625,
    "learning_rate": 0.00019968848445230743,
    "epoch": 0.02829854969932791,
    "step": 880
  },
  {
    "loss": 0.15401984453201295,
    "grad_norm": 0.6393543481826782,
    "learning_rate": 0.00019968043957626502,
    "epoch": 0.028620124127729362,
    "step": 890
  },
  {
    "loss": 0.12747691869735717,
    "grad_norm": 0.7813221216201782,
    "learning_rate": 0.00019967229230720881,
    "epoch": 0.028941698556130817,
    "step": 900
  },
  {
    "loss": 0.12719355821609496,
    "grad_norm": 0.7403981685638428,
    "learning_rate": 0.00019966404265350774,
    "epoch": 0.02926327298453227,
    "step": 910
  },
  {
    "loss": 0.10643273591995239,
    "grad_norm": 0.48086461424827576,
    "learning_rate": 0.000199655690623636,
    "epoch": 0.029584847412933723,
    "step": 920
  },
  {
    "loss": 0.12535429000854492,
    "grad_norm": 1.4569687843322754,
    "learning_rate": 0.0001996472362261729,
    "epoch": 0.029906421841335178,
    "step": 930
  },
  {
    "loss": 0.11564887762069702,
    "grad_norm": 0.6724082827568054,
    "learning_rate": 0.00019963867946980285,
    "epoch": 0.03022799626973663,
    "step": 940
  },
  {
    "loss": 0.12263286113739014,
    "grad_norm": 0.8747388124465942,
    "learning_rate": 0.00019963002036331554,
    "epoch": 0.030549570698138084,
    "step": 950
  },
  {
    "loss": 0.11707538366317749,
    "grad_norm": 1.0104649066925049,
    "learning_rate": 0.00019962125891560568,
    "epoch": 0.03087114512653954,
    "step": 960
  },
  {
    "loss": 0.12445554733276368,
    "grad_norm": 0.7107091546058655,
    "learning_rate": 0.00019961239513567312,
    "epoch": 0.03119271955494099,
    "step": 970
  },
  {
    "loss": 0.10379760265350342,
    "grad_norm": 0.5619798898696899,
    "learning_rate": 0.00019960342903262284,
    "epoch": 0.03151429398334245,
    "step": 980
  },
  {
    "loss": 0.12779237031936647,
    "grad_norm": 0.5871363878250122,
    "learning_rate": 0.000199594360615665,
    "epoch": 0.0318358684117439,
    "step": 990
  },
  {
    "loss": 0.12199541330337524,
    "grad_norm": 0.6088923215866089,
    "learning_rate": 0.0001995851898941147,
    "epoch": 0.03215744284014535,
    "step": 1000
  },
  {
    "loss": 0.13736236095428467,
    "grad_norm": 0.8911682963371277,
    "learning_rate": 0.00019957591687739233,
    "epoch": 0.0324790172685468,
    "step": 1010
  },
  {
    "loss": 0.11516963243484497,
    "grad_norm": 0.6520726680755615,
    "learning_rate": 0.00019956654157502314,
    "epoch": 0.03280059169694826,
    "step": 1020
  },
  {
    "loss": 0.14110876321792604,
    "grad_norm": 0.6925550103187561,
    "learning_rate": 0.00019955706399663766,
    "epoch": 0.03312216612534971,
    "step": 1030
  },
  {
    "loss": 0.14986019134521483,
    "grad_norm": 0.6888828873634338,
    "learning_rate": 0.00019954748415197134,
    "epoch": 0.03344374055375116,
    "step": 1040
  },
  {
    "loss": 0.10206960439682007,
    "grad_norm": 0.5661484599113464,
    "learning_rate": 0.00019953780205086464,
    "epoch": 0.03376531498215262,
    "step": 1050
  },
  {
    "loss": 0.11590293645858765,
    "grad_norm": 0.6978943943977356,
    "learning_rate": 0.00019952801770326328,
    "epoch": 0.03408688941055407,
    "step": 1060
  },
  {
    "loss": 0.10875160694122314,
    "grad_norm": 0.501489520072937,
    "learning_rate": 0.00019951813111921776,
    "epoch": 0.03440846383895552,
    "step": 1070
  },
  {
    "loss": 0.07669517397880554,
    "grad_norm": 0.5798457264900208,
    "learning_rate": 0.00019950814230888374,
    "epoch": 0.03473003826735698,
    "step": 1080
  },
  {
    "loss": 0.12092061042785644,
    "grad_norm": 0.9621400237083435,
    "learning_rate": 0.00019949805128252187,
    "epoch": 0.03505161269575843,
    "step": 1090
  },
  {
    "loss": 0.14484325647354127,
    "grad_norm": 0.6315482258796692,
    "learning_rate": 0.00019948785805049774,
    "epoch": 0.035373187124159884,
    "step": 1100
  },
  {
    "loss": 0.14884073734283448,
    "grad_norm": 0.905430793762207,
    "learning_rate": 0.000199477562623282,
    "epoch": 0.03569476155256134,
    "step": 1110
  },
  {
    "loss": 0.13443020582199097,
    "grad_norm": 0.9212183356285095,
    "learning_rate": 0.00019946716501145022,
    "epoch": 0.036016335980962794,
    "step": 1120
  },
  {
    "loss": 0.12805633544921874,
    "grad_norm": 0.9336081743240356,
    "learning_rate": 0.00019945666522568298,
    "epoch": 0.036337910409364245,
    "step": 1130
  },
  {
    "loss": 0.1238585352897644,
    "grad_norm": 1.1159182786941528,
    "learning_rate": 0.00019944606327676579,
    "epoch": 0.0366594848377657,
    "step": 1140
  },
  {
    "loss": 0.11155736446380615,
    "grad_norm": 0.7781017422676086,
    "learning_rate": 0.0001994353591755891,
    "epoch": 0.036981059266167154,
    "step": 1150
  },
  {
    "loss": 0.12316350936889649,
    "grad_norm": 0.9370765089988708,
    "learning_rate": 0.0001994245529331483,
    "epoch": 0.037302633694568606,
    "step": 1160
  },
  {
    "loss": 0.12969765663146973,
    "grad_norm": 0.8996332883834839,
    "learning_rate": 0.00019941364456054367,
    "epoch": 0.037624208122970064,
    "step": 1170
  },
  {
    "loss": 0.1030587911605835,
    "grad_norm": 1.21480131149292,
    "learning_rate": 0.0001994026340689805,
    "epoch": 0.037945782551371515,
    "step": 1180
  },
  {
    "loss": 0.14748505353927613,
    "grad_norm": 0.8236110806465149,
    "learning_rate": 0.00019939152146976886,
    "epoch": 0.038267356979772967,
    "step": 1190
  },
  {
    "loss": 0.15111169815063477,
    "grad_norm": 0.8031110763549805,
    "learning_rate": 0.00019938030677432372,
    "epoch": 0.038588931408174425,
    "step": 1200
  },
  {
    "loss": 0.12816275358200074,
    "grad_norm": 0.6093372702598572,
    "learning_rate": 0.00019936898999416505,
    "epoch": 0.038910505836575876,
    "step": 1210
  },
  {
    "loss": 0.13696140050888062,
    "grad_norm": 0.9523636102676392,
    "learning_rate": 0.00019935757114091751,
    "epoch": 0.03923208026497733,
    "step": 1220
  },
  {
    "loss": 0.10912299156188965,
    "grad_norm": 0.7931325435638428,
    "learning_rate": 0.0001993460502263107,
    "epoch": 0.039553654693378786,
    "step": 1230
  },
  {
    "loss": 0.1422716498374939,
    "grad_norm": 0.7136402130126953,
    "learning_rate": 0.0001993344272621791,
    "epoch": 0.03987522912178024,
    "step": 1240
  },
  {
    "loss": 0.1305735230445862,
    "grad_norm": 0.7761098146438599,
    "learning_rate": 0.00019932270226046193,
    "epoch": 0.04019680355018169,
    "step": 1250
  },
  {
    "loss": 0.13865208625793457,
    "grad_norm": 0.8858702182769775,
    "learning_rate": 0.00019931087523320326,
    "epoch": 0.040518377978583146,
    "step": 1260
  },
  {
    "loss": 0.12371792793273925,
    "grad_norm": 0.6040947437286377,
    "learning_rate": 0.00019929894619255195,
    "epoch": 0.0408399524069846,
    "step": 1270
  },
  {
    "loss": 0.13337819576263427,
    "grad_norm": 0.9483299851417542,
    "learning_rate": 0.0001992869151507617,
    "epoch": 0.04116152683538605,
    "step": 1280
  },
  {
    "loss": 0.1270521402359009,
    "grad_norm": 0.6090492010116577,
    "learning_rate": 0.00019927478212019092,
    "epoch": 0.0414831012637875,
    "step": 1290
  },
  {
    "loss": 0.12199642658233642,
    "grad_norm": 1.03255033493042,
    "learning_rate": 0.00019926254711330281,
    "epoch": 0.04180467569218896,
    "step": 1300
  },
  {
    "loss": 0.12788064479827882,
    "grad_norm": 1.19109046459198,
    "learning_rate": 0.00019925021014266536,
    "epoch": 0.04212625012059041,
    "step": 1310
  },
  {
    "loss": 0.12751251459121704,
    "grad_norm": 0.9528569579124451,
    "learning_rate": 0.00019923777122095123,
    "epoch": 0.04244782454899186,
    "step": 1320
  },
  {
    "loss": 0.12552311420440673,
    "grad_norm": 0.6547811031341553,
    "learning_rate": 0.00019922523036093782,
    "epoch": 0.04276939897739332,
    "step": 1330
  },
  {
    "loss": 0.11283427476882935,
    "grad_norm": 0.7006787657737732,
    "learning_rate": 0.00019921258757550734,
    "epoch": 0.04309097340579477,
    "step": 1340
  },
  {
    "loss": 0.1332577347755432,
    "grad_norm": 0.7077374458312988,
    "learning_rate": 0.00019919984287764653,
    "epoch": 0.04341254783419622,
    "step": 1350
  },
  {
    "loss": 0.12081760168075562,
    "grad_norm": 0.5344422459602356,
    "learning_rate": 0.00019918699628044694,
    "epoch": 0.04373412226259768,
    "step": 1360
  },
  {
    "loss": 0.11974830627441406,
    "grad_norm": 0.8550666570663452,
    "learning_rate": 0.00019917404779710478,
    "epoch": 0.04405569669099913,
    "step": 1370
  },
  {
    "loss": 0.108591628074646,
    "grad_norm": 0.583908200263977,
    "learning_rate": 0.00019916099744092088,
    "epoch": 0.04437727111940058,
    "step": 1380
  },
  {
    "loss": 0.1248102068901062,
    "grad_norm": 1.138697862625122,
    "learning_rate": 0.00019914784522530074,
    "epoch": 0.04469884554780204,
    "step": 1390
  },
  {
    "loss": 0.11697250604629517,
    "grad_norm": 0.914976954460144,
    "learning_rate": 0.00019913459116375446,
    "epoch": 0.04502041997620349,
    "step": 1400
  },
  {
    "loss": 0.10801193714141846,
    "grad_norm": 0.618378758430481,
    "learning_rate": 0.0001991212352698968,
    "epoch": 0.04534199440460494,
    "step": 1410
  },
  {
    "loss": 0.11103023290634155,
    "grad_norm": 0.9175752401351929,
    "learning_rate": 0.00019910777755744708,
    "epoch": 0.0456635688330064,
    "step": 1420
  },
  {
    "loss": 0.11435778141021728,
    "grad_norm": 0.7195553779602051,
    "learning_rate": 0.00019909421804022924,
    "epoch": 0.04598514326140785,
    "step": 1430
  },
  {
    "loss": 0.1225479245185852,
    "grad_norm": 0.567287266254425,
    "learning_rate": 0.0001990805567321718,
    "epoch": 0.046306717689809304,
    "step": 1440
  },
  {
    "loss": 0.13343338966369628,
    "grad_norm": 1.069581389427185,
    "learning_rate": 0.00019906679364730786,
    "epoch": 0.04662829211821076,
    "step": 1450
  },
  {
    "loss": 0.13520500659942628,
    "grad_norm": 0.8290631175041199,
    "learning_rate": 0.00019905292879977498,
    "epoch": 0.046949866546612214,
    "step": 1460
  },
  {
    "loss": 0.10813941955566406,
    "grad_norm": 0.8197263479232788,
    "learning_rate": 0.00019903896220381534,
    "epoch": 0.047271440975013665,
    "step": 1470
  },
  {
    "loss": 0.10801063776016236,
    "grad_norm": 0.6460339426994324,
    "learning_rate": 0.00019902489387377557,
    "epoch": 0.04759301540341512,
    "step": 1480
  },
  {
    "loss": 0.12939759492874145,
    "grad_norm": 0.958020031452179,
    "learning_rate": 0.0001990107238241069,
    "epoch": 0.047914589831816574,
    "step": 1490
  },
  {
    "loss": 0.12598515748977662,
    "grad_norm": 0.8492549657821655,
    "learning_rate": 0.00019899645206936494,
    "epoch": 0.048236164260218026,
    "step": 1500
  },
  {
    "loss": 0.11651817560195923,
    "grad_norm": 0.5606909394264221,
    "learning_rate": 0.0001989820786242098,
    "epoch": 0.048557738688619484,
    "step": 1510
  },
  {
    "loss": 0.11292610168457032,
    "grad_norm": 0.7311549782752991,
    "learning_rate": 0.00019896760350340608,
    "epoch": 0.048879313117020935,
    "step": 1520
  },
  {
    "loss": 0.09686515927314758,
    "grad_norm": 0.5886459946632385,
    "learning_rate": 0.00019895302672182287,
    "epoch": 0.049200887545422387,
    "step": 1530
  },
  {
    "loss": 0.11024473905563355,
    "grad_norm": 0.8052958250045776,
    "learning_rate": 0.00019893834829443354,
    "epoch": 0.049522461973823845,
    "step": 1540
  },
  {
    "loss": 0.11257057189941407,
    "grad_norm": 0.8030698299407959,
    "learning_rate": 0.000198923568236316,
    "epoch": 0.049844036402225296,
    "step": 1550
  },
  {
    "eval_loss": 0.11128632724285126,
    "eval_runtime": 46.063,
    "eval_samples_per_second": 109.111,
    "eval_steps_per_second": 27.289,
    "epoch": 0.049972666173585874,
    "step": 1554
  },
  {
    "loss": 0.11879251003265381,
    "grad_norm": 1.0847232341766357,
    "learning_rate": 0.00019890868656265251,
    "epoch": 0.05016561083062675,
    "step": 1560
  },
  {
    "loss": 0.09545501470565795,
    "grad_norm": 0.6696542501449585,
    "learning_rate": 0.0001988937032887297,
    "epoch": 0.050487185259028206,
    "step": 1570
  },
  {
    "loss": 0.1338610291481018,
    "grad_norm": 0.8771032691001892,
    "learning_rate": 0.00019887861842993857,
    "epoch": 0.05080875968742966,
    "step": 1580
  },
  {
    "loss": 0.13209749460220338,
    "grad_norm": 0.6574100255966187,
    "learning_rate": 0.00019886343200177455,
    "epoch": 0.05113033411583111,
    "step": 1590
  },
  {
    "loss": 0.10262678861618042,
    "grad_norm": 0.6385945081710815,
    "learning_rate": 0.0001988481440198372,
    "epoch": 0.05145190854423256,
    "step": 1600
  },
  {
    "loss": 0.13754674196243286,
    "grad_norm": 0.8028077483177185,
    "learning_rate": 0.0001988327544998307,
    "epoch": 0.05177348297263402,
    "step": 1610
  },
  {
    "loss": 0.09865680932998658,
    "grad_norm": 0.8681406378746033,
    "learning_rate": 0.00019881726345756322,
    "epoch": 0.05209505740103547,
    "step": 1620
  },
  {
    "loss": 0.118634033203125,
    "grad_norm": 0.7602818012237549,
    "learning_rate": 0.0001988016709089474,
    "epoch": 0.05241663182943692,
    "step": 1630
  },
  {
    "loss": 0.10665820837020874,
    "grad_norm": 0.5052031874656677,
    "learning_rate": 0.00019878597687000014,
    "epoch": 0.05273820625783838,
    "step": 1640
  },
  {
    "loss": 0.13749959468841552,
    "grad_norm": 0.5683307647705078,
    "learning_rate": 0.00019877018135684246,
    "epoch": 0.05305978068623983,
    "step": 1650
  },
  {
    "loss": 0.12095248699188232,
    "grad_norm": 0.5208232402801514,
    "learning_rate": 0.00019875428438569982,
    "epoch": 0.05338135511464128,
    "step": 1660
  },
  {
    "loss": 0.1353804349899292,
    "grad_norm": 0.7039898633956909,
    "learning_rate": 0.00019873828597290175,
    "epoch": 0.05370292954304274,
    "step": 1670
  },
  {
    "loss": 0.10228586196899414,
    "grad_norm": 0.3282141089439392,
    "learning_rate": 0.00019872218613488203,
    "epoch": 0.05402450397144419,
    "step": 1680
  },
  {
    "loss": 0.1056720495223999,
    "grad_norm": 0.5899987816810608,
    "learning_rate": 0.00019870598488817858,
    "epoch": 0.05434607839984564,
    "step": 1690
  },
  {
    "loss": 0.09971145391464234,
    "grad_norm": 1.2377015352249146,
    "learning_rate": 0.00019868968224943356,
    "epoch": 0.0546676528282471,
    "step": 1700
  },
  {
    "loss": 0.10820199251174926,
    "grad_norm": 0.5962371230125427,
    "learning_rate": 0.00019867327823539323,
    "epoch": 0.05498922725664855,
    "step": 1710
  },
  {
    "loss": 0.12856364250183105,
    "grad_norm": 1.2653589248657227,
    "learning_rate": 0.00019865677286290802,
    "epoch": 0.05531080168505,
    "step": 1720
  },
  {
    "loss": 0.12928249835968017,
    "grad_norm": 1.5054935216903687,
    "learning_rate": 0.00019864016614893246,
    "epoch": 0.05563237611345146,
    "step": 1730
  },
  {
    "loss": 0.1322105646133423,
    "grad_norm": 0.7001881003379822,
    "learning_rate": 0.00019862345811052517,
    "epoch": 0.05595395054185291,
    "step": 1740
  },
  {
    "loss": 0.09415720701217652,
    "grad_norm": 0.3953130841255188,
    "learning_rate": 0.00019860664876484885,
    "epoch": 0.05627552497025436,
    "step": 1750
  },
  {
    "loss": 0.11939146518707275,
    "grad_norm": 0.9448032379150391,
    "learning_rate": 0.0001985897381291703,
    "epoch": 0.05659709939865582,
    "step": 1760
  },
  {
    "loss": 0.1330693244934082,
    "grad_norm": 0.6536581516265869,
    "learning_rate": 0.00019857272622086033,
    "epoch": 0.05691867382705727,
    "step": 1770
  },
  {
    "loss": 0.12102277278900146,
    "grad_norm": 0.746406078338623,
    "learning_rate": 0.00019855561305739375,
    "epoch": 0.057240248255458724,
    "step": 1780
  },
  {
    "loss": 0.11163841485977173,
    "grad_norm": 0.4073915481567383,
    "learning_rate": 0.00019853839865634944,
    "epoch": 0.05756182268386018,
    "step": 1790
  },
  {
    "loss": 0.12480851411819457,
    "grad_norm": 0.9843113422393799,
    "learning_rate": 0.00019852108303541029,
    "epoch": 0.057883397112261634,
    "step": 1800
  },
  {
    "loss": 0.10980266332626343,
    "grad_norm": 0.41751521825790405,
    "learning_rate": 0.00019850366621236307,
    "epoch": 0.058204971540663085,
    "step": 1810
  },
  {
    "loss": 0.10909017324447631,
    "grad_norm": 0.8801973462104797,
    "learning_rate": 0.00019848614820509858,
    "epoch": 0.05852654596906454,
    "step": 1820
  },
  {
    "loss": 0.12614597082138063,
    "grad_norm": 0.8948915004730225,
    "learning_rate": 0.0001984685290316116,
    "epoch": 0.058848120397465994,
    "step": 1830
  },
  {
    "loss": 0.12774603366851806,
    "grad_norm": 0.9197909832000732,
    "learning_rate": 0.00019845080871000068,
    "epoch": 0.059169694825867446,
    "step": 1840
  },
  {
    "loss": 0.10084892511367798,
    "grad_norm": 0.9967162013053894,
    "learning_rate": 0.0001984329872584684,
    "epoch": 0.059491269254268904,
    "step": 1850
  },
  {
    "loss": 0.11144490242004394,
    "grad_norm": 0.5707647800445557,
    "learning_rate": 0.00019841506469532117,
    "epoch": 0.059812843682670355,
    "step": 1860
  },
  {
    "loss": 0.10548099279403686,
    "grad_norm": 0.528365969657898,
    "learning_rate": 0.00019839704103896927,
    "epoch": 0.060134418111071807,
    "step": 1870
  },
  {
    "loss": 0.10982582569122315,
    "grad_norm": 0.33779260516166687,
    "learning_rate": 0.00019837891630792686,
    "epoch": 0.06045599253947326,
    "step": 1880
  },
  {
    "loss": 0.11971958875656127,
    "grad_norm": 0.7719383239746094,
    "learning_rate": 0.00019836069052081188,
    "epoch": 0.060777566967874716,
    "step": 1890
  },
  {
    "loss": 0.11518895626068115,
    "grad_norm": 0.809697151184082,
    "learning_rate": 0.00019834236369634608,
    "epoch": 0.06109914139627617,
    "step": 1900
  },
  {
    "loss": 0.10948337316513061,
    "grad_norm": 0.8163278102874756,
    "learning_rate": 0.00019832393585335502,
    "epoch": 0.06142071582467762,
    "step": 1910
  },
  {
    "loss": 0.118583083152771,
    "grad_norm": 0.46315309405326843,
    "learning_rate": 0.000198305407010768,
    "epoch": 0.06174229025307908,
    "step": 1920
  },
  {
    "loss": 0.10560156106948852,
    "grad_norm": 0.5412709712982178,
    "learning_rate": 0.00019828677718761805,
    "epoch": 0.06206386468148053,
    "step": 1930
  },
  {
    "loss": 0.12434232234954834,
    "grad_norm": 0.7563145756721497,
    "learning_rate": 0.00019826804640304206,
    "epoch": 0.06238543910988198,
    "step": 1940
  },
  {
    "loss": 0.12473642826080322,
    "grad_norm": 0.8467011451721191,
    "learning_rate": 0.00019824921467628045,
    "epoch": 0.06270701353828344,
    "step": 1950
  },
  {
    "loss": 0.10392096042633056,
    "grad_norm": 0.4988096356391907,
    "learning_rate": 0.0001982302820266774,
    "epoch": 0.0630285879666849,
    "step": 1960
  },
  {
    "loss": 0.13397929668426514,
    "grad_norm": 0.6903615593910217,
    "learning_rate": 0.00019821124847368082,
    "epoch": 0.06335016239508634,
    "step": 1970
  },
  {
    "loss": 0.11959564685821533,
    "grad_norm": 1.0386449098587036,
    "learning_rate": 0.0001981921140368422,
    "epoch": 0.0636717368234878,
    "step": 1980
  },
  {
    "loss": 0.1142409086227417,
    "grad_norm": 0.6791936159133911,
    "learning_rate": 0.00019817287873581666,
    "epoch": 0.06399331125188926,
    "step": 1990
  },
  {
    "loss": 0.15232725143432618,
    "grad_norm": 1.3021385669708252,
    "learning_rate": 0.00019815354259036295,
    "epoch": 0.0643148856802907,
    "step": 2000
  },
  {
    "loss": 0.11425412893295288,
    "grad_norm": 0.5564187169075012,
    "learning_rate": 0.0001981341056203434,
    "epoch": 0.06463646010869216,
    "step": 2010
  },
  {
    "loss": 0.1432906746864319,
    "grad_norm": 0.9113200306892395,
    "learning_rate": 0.00019811456784572394,
    "epoch": 0.0649580345370936,
    "step": 2020
  },
  {
    "loss": 0.13707234859466552,
    "grad_norm": 0.6878952980041504,
    "learning_rate": 0.000198094929286574,
    "epoch": 0.06527960896549506,
    "step": 2030
  },
  {
    "loss": 0.14629603624343873,
    "grad_norm": 0.8060909509658813,
    "learning_rate": 0.00019807518996306657,
    "epoch": 0.06560118339389652,
    "step": 2040
  },
  {
    "loss": 0.14162900447845458,
    "grad_norm": 0.8747666478157043,
    "learning_rate": 0.0001980553498954781,
    "epoch": 0.06592275782229796,
    "step": 2050
  },
  {
    "loss": 0.1435432553291321,
    "grad_norm": 0.9168952703475952,
    "learning_rate": 0.00019803540910418862,
    "epoch": 0.06624433225069942,
    "step": 2060
  },
  {
    "loss": 0.13169909715652467,
    "grad_norm": 0.8191623687744141,
    "learning_rate": 0.0001980153676096815,
    "epoch": 0.06656590667910088,
    "step": 2070
  },
  {
    "loss": 0.10620007514953614,
    "grad_norm": 0.58042311668396,
    "learning_rate": 0.00019799522543254366,
    "epoch": 0.06688748110750233,
    "step": 2080
  },
  {
    "loss": 0.11155413389205933,
    "grad_norm": 0.6616443991661072,
    "learning_rate": 0.00019797498259346537,
    "epoch": 0.06720905553590378,
    "step": 2090
  },
  {
    "loss": 0.13231509923934937,
    "grad_norm": 0.7794826626777649,
    "learning_rate": 0.00019795463911324035,
    "epoch": 0.06753062996430524,
    "step": 2100
  },
  {
    "loss": 0.13808358907699586,
    "grad_norm": 0.9658569097518921,
    "learning_rate": 0.00019793419501276568,
    "epoch": 0.06785220439270669,
    "step": 2110
  },
  {
    "loss": 0.132569420337677,
    "grad_norm": 0.7185949683189392,
    "learning_rate": 0.0001979136503130418,
    "epoch": 0.06817377882110814,
    "step": 2120
  },
  {
    "loss": 0.10459121465682983,
    "grad_norm": 0.6469611525535583,
    "learning_rate": 0.00019789300503517249,
    "epoch": 0.0684953532495096,
    "step": 2130
  },
  {
    "loss": 0.12158732414245606,
    "grad_norm": 1.1708670854568481,
    "learning_rate": 0.00019787225920036482,
    "epoch": 0.06881692767791105,
    "step": 2140
  },
  {
    "loss": 0.11037311553955079,
    "grad_norm": 0.7065096497535706,
    "learning_rate": 0.00019785141282992923,
    "epoch": 0.0691385021063125,
    "step": 2150
  },
  {
    "loss": 0.10084078311920167,
    "grad_norm": 0.7217418551445007,
    "learning_rate": 0.0001978304659452793,
    "epoch": 0.06946007653471396,
    "step": 2160
  },
  {
    "loss": 0.10967367887496948,
    "grad_norm": 0.5506851673126221,
    "learning_rate": 0.000197809418567932,
    "epoch": 0.06978165096311541,
    "step": 2170
  },
  {
    "loss": 0.11019937992095948,
    "grad_norm": 0.6507880091667175,
    "learning_rate": 0.00019778827071950742,
    "epoch": 0.07010322539151687,
    "step": 2180
  },
  {
    "loss": 0.11477915048599244,
    "grad_norm": 0.6157989501953125,
    "learning_rate": 0.00019776702242172892,
    "epoch": 0.07042479981991832,
    "step": 2190
  },
  {
    "loss": 0.134388530254364,
    "grad_norm": 0.8920512199401855,
    "learning_rate": 0.00019774567369642301,
    "epoch": 0.07074637424831977,
    "step": 2200
  },
  {
    "loss": 0.093971848487854,
    "grad_norm": 0.6216977834701538,
    "learning_rate": 0.0001977242245655194,
    "epoch": 0.07106794867672123,
    "step": 2210
  },
  {
    "loss": 0.1283738136291504,
    "grad_norm": 0.7825973033905029,
    "learning_rate": 0.00019770267505105088,
    "epoch": 0.07138952310512268,
    "step": 2220
  },
  {
    "loss": 0.1231916069984436,
    "grad_norm": 0.8116011619567871,
    "learning_rate": 0.00019768102517515336,
    "epoch": 0.07171109753352413,
    "step": 2230
  },
  {
    "loss": 0.07891560792922973,
    "grad_norm": 0.8718013763427734,
    "learning_rate": 0.00019765927496006596,
    "epoch": 0.07203267196192559,
    "step": 2240
  },
  {
    "loss": 0.09509350061416626,
    "grad_norm": 0.6453697085380554,
    "learning_rate": 0.00019763742442813068,
    "epoch": 0.07235424639032705,
    "step": 2250
  },
  {
    "loss": 0.11959600448608398,
    "grad_norm": 0.559689462184906,
    "learning_rate": 0.00019761547360179268,
    "epoch": 0.07267582081872849,
    "step": 2260
  },
  {
    "loss": 0.11340382099151611,
    "grad_norm": 0.42303594946861267,
    "learning_rate": 0.0001975934225036002,
    "epoch": 0.07299739524712995,
    "step": 2270
  },
  {
    "loss": 0.10959010124206543,
    "grad_norm": 0.5502386093139648,
    "learning_rate": 0.00019757127115620435,
    "epoch": 0.0733189696755314,
    "step": 2280
  },
  {
    "loss": 0.1179365873336792,
    "grad_norm": 1.1213687658309937,
    "learning_rate": 0.00019754901958235927,
    "epoch": 0.07364054410393285,
    "step": 2290
  },
  {
    "loss": 0.12511249780654907,
    "grad_norm": 0.611772358417511,
    "learning_rate": 0.00019752666780492207,
    "epoch": 0.07396211853233431,
    "step": 2300
  },
  {
    "loss": 0.13076653480529785,
    "grad_norm": 0.5212668180465698,
    "learning_rate": 0.0001975042158468528,
    "epoch": 0.07428369296073577,
    "step": 2310
  },
  {
    "loss": 0.10176039934158325,
    "grad_norm": 1.010329246520996,
    "learning_rate": 0.00019748166373121436,
    "epoch": 0.07460526738913721,
    "step": 2320
  },
  {
    "loss": 0.10914967060089112,
    "grad_norm": 0.7520009279251099,
    "learning_rate": 0.0001974590114811726,
    "epoch": 0.07492684181753867,
    "step": 2330
  },
  {
    "loss": 0.08320418596267701,
    "grad_norm": 0.6081514358520508,
    "learning_rate": 0.0001974362591199962,
    "epoch": 0.07524841624594013,
    "step": 2340
  },
  {
    "loss": 0.11121056079864503,
    "grad_norm": 0.779585599899292,
    "learning_rate": 0.00019741340667105665,
    "epoch": 0.07556999067434157,
    "step": 2350
  },
  {
    "loss": 0.09840821027755738,
    "grad_norm": 0.7340258359909058,
    "learning_rate": 0.0001973904541578283,
    "epoch": 0.07589156510274303,
    "step": 2360
  },
  {
    "loss": 0.13002623319625856,
    "grad_norm": 0.6933698058128357,
    "learning_rate": 0.00019736740160388824,
    "epoch": 0.07621313953114449,
    "step": 2370
  },
  {
    "loss": 0.12301526069641114,
    "grad_norm": 0.4505113363265991,
    "learning_rate": 0.00019734424903291635,
    "epoch": 0.07653471395954593,
    "step": 2380
  },
  {
    "loss": 0.11275660991668701,
    "grad_norm": 0.9241744875907898,
    "learning_rate": 0.00019732099646869528,
    "epoch": 0.07685628838794739,
    "step": 2390
  },
  {
    "loss": 0.13603054285049437,
    "grad_norm": 0.9413949847221375,
    "learning_rate": 0.00019729764393511032,
    "epoch": 0.07717786281634885,
    "step": 2400
  },
  {
    "loss": 0.10451099872589112,
    "grad_norm": 0.7786242961883545,
    "learning_rate": 0.0001972741914561495,
    "epoch": 0.0774994372447503,
    "step": 2410
  },
  {
    "loss": 0.10279628038406372,
    "grad_norm": 0.8419315218925476,
    "learning_rate": 0.00019725063905590352,
    "epoch": 0.07782101167315175,
    "step": 2420
  },
  {
    "loss": 0.12877086400985718,
    "grad_norm": 0.8151343464851379,
    "learning_rate": 0.00019722698675856564,
    "epoch": 0.07814258610155321,
    "step": 2430
  },
  {
    "loss": 0.11589571237564086,
    "grad_norm": 0.7058280110359192,
    "learning_rate": 0.0001972032345884319,
    "epoch": 0.07846416052995465,
    "step": 2440
  },
  {
    "loss": 0.11044619083404542,
    "grad_norm": 0.7708027958869934,
    "learning_rate": 0.00019717938256990077,
    "epoch": 0.07878573495835611,
    "step": 2450
  },
  {
    "loss": 0.13818830251693726,
    "grad_norm": 0.5580777525901794,
    "learning_rate": 0.00019715543072747335,
    "epoch": 0.07910730938675757,
    "step": 2460
  },
  {
    "loss": 0.12281345129013062,
    "grad_norm": 0.5901768803596497,
    "learning_rate": 0.0001971313790857533,
    "epoch": 0.07942888381515902,
    "step": 2470
  },
  {
    "loss": 0.10907325744628907,
    "grad_norm": 0.8018592596054077,
    "learning_rate": 0.00019710722766944674,
    "epoch": 0.07975045824356047,
    "step": 2480
  },
  {
    "loss": 0.12140153646469116,
    "grad_norm": 0.6740052700042725,
    "learning_rate": 0.00019708297650336234,
    "epoch": 0.08007203267196193,
    "step": 2490
  },
  {
    "loss": 0.1088398814201355,
    "grad_norm": 0.5753097534179688,
    "learning_rate": 0.00019705862561241117,
    "epoch": 0.08039360710036338,
    "step": 2500
  },
  {
    "loss": 0.11455504894256592,
    "grad_norm": 0.49554967880249023,
    "learning_rate": 0.00019703417502160684,
    "epoch": 0.08071518152876483,
    "step": 2510
  },
  {
    "loss": 0.12731305360794068,
    "grad_norm": 0.9352779388427734,
    "learning_rate": 0.00019700962475606524,
    "epoch": 0.08103675595716629,
    "step": 2520
  },
  {
    "loss": 0.10423781871795654,
    "grad_norm": 0.6742168664932251,
    "learning_rate": 0.00019698497484100472,
    "epoch": 0.08135833038556774,
    "step": 2530
  },
  {
    "loss": 0.11573268175125122,
    "grad_norm": 0.5347793102264404,
    "learning_rate": 0.00019696022530174605,
    "epoch": 0.0816799048139692,
    "step": 2540
  },
  {
    "loss": 0.11344225406646728,
    "grad_norm": 1.1676106452941895,
    "learning_rate": 0.0001969353761637122,
    "epoch": 0.08200147924237065,
    "step": 2550
  },
  {
    "loss": 0.11661807298660279,
    "grad_norm": 0.7781999707221985,
    "learning_rate": 0.00019691042745242858,
    "epoch": 0.0823230536707721,
    "step": 2560
  },
  {
    "loss": 0.11757798194885254,
    "grad_norm": 0.6178641319274902,
    "learning_rate": 0.00019688537919352274,
    "epoch": 0.08264462809917356,
    "step": 2570
  },
  {
    "loss": 0.11788698434829711,
    "grad_norm": 0.7667732238769531,
    "learning_rate": 0.00019686023141272466,
    "epoch": 0.082966202527575,
    "step": 2580
  },
  {
    "loss": 0.09521175622940063,
    "grad_norm": 1.0335626602172852,
    "learning_rate": 0.00019683498413586642,
    "epoch": 0.08328777695597646,
    "step": 2590
  },
  {
    "loss": 0.11917680501937866,
    "grad_norm": 0.582943856716156,
    "learning_rate": 0.00019680963738888233,
    "epoch": 0.08360935138437792,
    "step": 2600
  },
  {
    "loss": 0.11382901668548584,
    "grad_norm": 0.7423402070999146,
    "learning_rate": 0.00019678419119780892,
    "epoch": 0.08393092581277936,
    "step": 2610
  },
  {
    "loss": 0.12486684322357178,
    "grad_norm": 0.6837660074234009,
    "learning_rate": 0.00019675864558878483,
    "epoch": 0.08425250024118082,
    "step": 2620
  },
  {
    "loss": 0.10534079074859619,
    "grad_norm": 0.7795819044113159,
    "learning_rate": 0.0001967330005880508,
    "epoch": 0.08457407466958228,
    "step": 2630
  },
  {
    "loss": 0.11522390842437744,
    "grad_norm": 0.5674572587013245,
    "learning_rate": 0.00019670725622194976,
    "epoch": 0.08489564909798372,
    "step": 2640
  },
  {
    "loss": 0.11588020324707031,
    "grad_norm": 0.7193135023117065,
    "learning_rate": 0.0001966814125169266,
    "epoch": 0.08521722352638518,
    "step": 2650
  },
  {
    "loss": 0.11439142227172852,
    "grad_norm": 0.7520074844360352,
    "learning_rate": 0.00019665546949952835,
    "epoch": 0.08553879795478664,
    "step": 2660
  },
  {
    "loss": 0.1080745816230774,
    "grad_norm": 0.7337353825569153,
    "learning_rate": 0.00019662942719640396,
    "epoch": 0.08586037238318808,
    "step": 2670
  },
  {
    "loss": 0.09863368272781373,
    "grad_norm": 0.9156220555305481,
    "learning_rate": 0.00019660328563430443,
    "epoch": 0.08618194681158954,
    "step": 2680
  },
  {
    "loss": 0.1464648127555847,
    "grad_norm": 0.5568152666091919,
    "learning_rate": 0.00019657704484008274,
    "epoch": 0.086503521239991,
    "step": 2690
  },
  {
    "loss": 0.12296129465103149,
    "grad_norm": 1.0155256986618042,
    "learning_rate": 0.00019655070484069368,
    "epoch": 0.08682509566839244,
    "step": 2700
  },
  {
    "loss": 0.14590543508529663,
    "grad_norm": 0.6132897138595581,
    "learning_rate": 0.00019652426566319414,
    "epoch": 0.0871466700967939,
    "step": 2710
  },
  {
    "loss": 0.08513785004615784,
    "grad_norm": 0.6537352800369263,
    "learning_rate": 0.00019649772733474269,
    "epoch": 0.08746824452519536,
    "step": 2720
  },
  {
    "loss": 0.12123624086380005,
    "grad_norm": 0.8966730237007141,
    "learning_rate": 0.00019647108988259988,
    "epoch": 0.0877898189535968,
    "step": 2730
  },
  {
    "loss": 0.08953339457511902,
    "grad_norm": 0.6152775287628174,
    "learning_rate": 0.00019644435333412806,
    "epoch": 0.08811139338199826,
    "step": 2740
  },
  {
    "loss": 0.10960085391998291,
    "grad_norm": 0.48908594250679016,
    "learning_rate": 0.0001964175177167913,
    "epoch": 0.08843296781039972,
    "step": 2750
  },
  {
    "loss": 0.11797866821289063,
    "grad_norm": 0.7124961018562317,
    "learning_rate": 0.00019639058305815553,
    "epoch": 0.08875454223880117,
    "step": 2760
  },
  {
    "loss": 0.11341865062713623,
    "grad_norm": 0.579239547252655,
    "learning_rate": 0.00019636354938588836,
    "epoch": 0.08907611666720262,
    "step": 2770
  },
  {
    "loss": 0.14069968461990356,
    "grad_norm": 0.7737932205200195,
    "learning_rate": 0.0001963364167277591,
    "epoch": 0.08939769109560408,
    "step": 2780
  },
  {
    "loss": 0.10434212684631347,
    "grad_norm": 0.7406035661697388,
    "learning_rate": 0.0001963091851116388,
    "epoch": 0.08971926552400553,
    "step": 2790
  },
  {
    "loss": 0.1168674349784851,
    "grad_norm": 0.9775945544242859,
    "learning_rate": 0.00019628185456550007,
    "epoch": 0.09004083995240698,
    "step": 2800
  },
  {
    "loss": 0.12225688695907592,
    "grad_norm": 0.44528913497924805,
    "learning_rate": 0.00019625442511741727,
    "epoch": 0.09036241438080844,
    "step": 2810
  },
  {
    "loss": 0.12359638214111328,
    "grad_norm": 0.4269179105758667,
    "learning_rate": 0.00019622689679556622,
    "epoch": 0.09068398880920989,
    "step": 2820
  },
  {
    "loss": 0.1075709342956543,
    "grad_norm": 0.6485812067985535,
    "learning_rate": 0.00019619926962822435,
    "epoch": 0.09100556323761134,
    "step": 2830
  },
  {
    "loss": 0.11307716369628906,
    "grad_norm": 1.5778266191482544,
    "learning_rate": 0.00019617154364377068,
    "epoch": 0.0913271376660128,
    "step": 2840
  },
  {
    "loss": 0.11073293685913085,
    "grad_norm": 0.8932989239692688,
    "learning_rate": 0.0001961437188706857,
    "epoch": 0.09164871209441425,
    "step": 2850
  },
  {
    "loss": 0.11458207368850708,
    "grad_norm": 0.7158249616622925,
    "learning_rate": 0.0001961157953375513,
    "epoch": 0.0919702865228157,
    "step": 2860
  },
  {
    "loss": 0.10697572231292725,
    "grad_norm": 0.5875152945518494,
    "learning_rate": 0.00019608777307305095,
    "epoch": 0.09229186095121716,
    "step": 2870
  },
  {
    "loss": 0.1203433632850647,
    "grad_norm": 1.4160408973693848,
    "learning_rate": 0.00019605965210596948,
    "epoch": 0.09261343537961861,
    "step": 2880
  },
  {
    "loss": 0.13875064849853516,
    "grad_norm": 0.6960219740867615,
    "learning_rate": 0.00019603143246519307,
    "epoch": 0.09293500980802007,
    "step": 2890
  },
  {
    "loss": 0.1032223105430603,
    "grad_norm": 0.635543704032898,
    "learning_rate": 0.00019600311417970932,
    "epoch": 0.09325658423642152,
    "step": 2900
  },
  {
    "loss": 0.12232608795166015,
    "grad_norm": 0.7557558417320251,
    "learning_rate": 0.0001959746972786071,
    "epoch": 0.09357815866482297,
    "step": 2910
  },
  {
    "loss": 0.11726034879684448,
    "grad_norm": 0.6953281164169312,
    "learning_rate": 0.00019594618179107664,
    "epoch": 0.09389973309322443,
    "step": 2920
  },
  {
    "loss": 0.10258761644363404,
    "grad_norm": 0.452549546957016,
    "learning_rate": 0.00019591756774640944,
    "epoch": 0.09422130752162589,
    "step": 2930
  },
  {
    "loss": 0.12623608112335205,
    "grad_norm": 0.7142686247825623,
    "learning_rate": 0.00019588885517399812,
    "epoch": 0.09454288195002733,
    "step": 2940
  },
  {
    "loss": 0.12421194314956666,
    "grad_norm": 0.5964231491088867,
    "learning_rate": 0.0001958600441033367,
    "epoch": 0.09486445637842879,
    "step": 2950
  },
  {
    "loss": 0.11757348775863648,
    "grad_norm": 0.6275390386581421,
    "learning_rate": 0.0001958311345640202,
    "epoch": 0.09518603080683025,
    "step": 2960
  },
  {
    "loss": 0.1174355149269104,
    "grad_norm": 0.611051082611084,
    "learning_rate": 0.00019580212658574492,
    "epoch": 0.09550760523523169,
    "step": 2970
  },
  {
    "loss": 0.11205054521560669,
    "grad_norm": 0.7474158406257629,
    "learning_rate": 0.00019577302019830817,
    "epoch": 0.09582917966363315,
    "step": 2980
  },
  {
    "loss": 0.12011423110961914,
    "grad_norm": 1.0890192985534668,
    "learning_rate": 0.00019574381543160843,
    "epoch": 0.09615075409203461,
    "step": 2990
  },
  {
    "loss": 0.09580199718475342,
    "grad_norm": 0.6116976737976074,
    "learning_rate": 0.00019571451231564525,
    "epoch": 0.09647232852043605,
    "step": 3000
  },
  {
    "loss": 0.1161598801612854,
    "grad_norm": 0.6288789510726929,
    "learning_rate": 0.0001956851108805191,
    "epoch": 0.09679390294883751,
    "step": 3010
  },
  {
    "loss": 0.10462770462036133,
    "grad_norm": 0.9161404967308044,
    "learning_rate": 0.00019565561115643152,
    "epoch": 0.09711547737723897,
    "step": 3020
  },
  {
    "loss": 0.12679061889648438,
    "grad_norm": 0.5990720987319946,
    "learning_rate": 0.00019562601317368503,
    "epoch": 0.09743705180564041,
    "step": 3030
  },
  {
    "loss": 0.10021860599517822,
    "grad_norm": 1.0173943042755127,
    "learning_rate": 0.00019559631696268302,
    "epoch": 0.09775862623404187,
    "step": 3040
  },
  {
    "loss": 0.09998620748519897,
    "grad_norm": 0.5726870894432068,
    "learning_rate": 0.00019556652255392983,
    "epoch": 0.09808020066244333,
    "step": 3050
  },
  {
    "loss": 0.13044171333312987,
    "grad_norm": 0.8290376663208008,
    "learning_rate": 0.00019553662997803066,
    "epoch": 0.09840177509084477,
    "step": 3060
  },
  {
    "loss": 0.10107682943344116,
    "grad_norm": 0.6910918354988098,
    "learning_rate": 0.00019550663926569154,
    "epoch": 0.09872334951924623,
    "step": 3070
  },
  {
    "loss": 0.12903016805648804,
    "grad_norm": 1.1203150749206543,
    "learning_rate": 0.00019547655044771932,
    "epoch": 0.09904492394764769,
    "step": 3080
  },
  {
    "loss": 0.10066742897033691,
    "grad_norm": 0.37552332878112793,
    "learning_rate": 0.00019544636355502156,
    "epoch": 0.09936649837604913,
    "step": 3090
  },
  {
    "loss": 0.11238259077072144,
    "grad_norm": 0.7681753039360046,
    "learning_rate": 0.00019541607861860673,
    "epoch": 0.09968807280445059,
    "step": 3100
  },
  {
    "eval_loss": 0.10922306776046753,
    "eval_runtime": 34.3459,
    "eval_samples_per_second": 146.335,
    "eval_steps_per_second": 36.598,
    "epoch": 0.09994533234717175,
    "step": 3108
  },
  {
    "loss": 0.08798834085464477,
    "grad_norm": 0.6184873580932617,
    "learning_rate": 0.0001953856956695838,
    "epoch": 0.10000964723285205,
    "step": 3110
  },
  {
    "loss": 0.13838959932327272,
    "grad_norm": 0.7986240386962891,
    "learning_rate": 0.00019535521473916257,
    "epoch": 0.1003312216612535,
    "step": 3120
  },
  {
    "loss": 0.10403268337249756,
    "grad_norm": 0.7976729869842529,
    "learning_rate": 0.0001953246358586534,
    "epoch": 0.10065279608965495,
    "step": 3130
  },
  {
    "loss": 0.10503251552581787,
    "grad_norm": 0.5033293962478638,
    "learning_rate": 0.00019529395905946737,
    "epoch": 0.10097437051805641,
    "step": 3140
  },
  {
    "loss": 0.13334119319915771,
    "grad_norm": 0.6307993531227112,
    "learning_rate": 0.00019526318437311605,
    "epoch": 0.10129594494645786,
    "step": 3150
  },
  {
    "loss": 0.0990634799003601,
    "grad_norm": 0.9739433526992798,
    "learning_rate": 0.0001952323118312115,
    "epoch": 0.10161751937485931,
    "step": 3160
  },
  {
    "loss": 0.10344570875167847,
    "grad_norm": 0.8039160966873169,
    "learning_rate": 0.00019520134146546654,
    "epoch": 0.10193909380326076,
    "step": 3170
  },
  {
    "loss": 0.1280883550643921,
    "grad_norm": 0.8540322184562683,
    "learning_rate": 0.00019517027330769423,
    "epoch": 0.10226066823166222,
    "step": 3180
  },
  {
    "loss": 0.10271800756454467,
    "grad_norm": 0.9251324534416199,
    "learning_rate": 0.0001951391073898082,
    "epoch": 0.10258224266006367,
    "step": 3190
  },
  {
    "loss": 0.12908483743667604,
    "grad_norm": 0.8988800048828125,
    "learning_rate": 0.00019510784374382245,
    "epoch": 0.10290381708846512,
    "step": 3200
  },
  {
    "loss": 0.11154094934463502,
    "grad_norm": 0.6570804715156555,
    "learning_rate": 0.00019507648240185143,
    "epoch": 0.10322539151686658,
    "step": 3210
  },
  {
    "loss": 0.15660790205001832,
    "grad_norm": 0.40354835987091064,
    "learning_rate": 0.0001950450233961099,
    "epoch": 0.10354696594526804,
    "step": 3220
  },
  {
    "loss": 0.11898343563079834,
    "grad_norm": 0.5271028876304626,
    "learning_rate": 0.00019501346675891292,
    "epoch": 0.10386854037366948,
    "step": 3230
  },
  {
    "loss": 0.08217236399650574,
    "grad_norm": 0.6451640129089355,
    "learning_rate": 0.00019498181252267593,
    "epoch": 0.10419011480207094,
    "step": 3240
  },
  {
    "loss": 0.12326722145080567,
    "grad_norm": 0.6667381525039673,
    "learning_rate": 0.0001949500607199145,
    "epoch": 0.1045116892304724,
    "step": 3250
  },
  {
    "loss": 0.09181814193725586,
    "grad_norm": 0.7040378451347351,
    "learning_rate": 0.00019491821138324452,
    "epoch": 0.10483326365887384,
    "step": 3260
  },
  {
    "loss": 0.1170729398727417,
    "grad_norm": 0.9717530012130737,
    "learning_rate": 0.00019488626454538204,
    "epoch": 0.1051548380872753,
    "step": 3270
  },
  {
    "loss": 0.11985208988189697,
    "grad_norm": 0.8807070255279541,
    "learning_rate": 0.00019485422023914322,
    "epoch": 0.10547641251567676,
    "step": 3280
  },
  {
    "loss": 0.1229805588722229,
    "grad_norm": 0.8843317627906799,
    "learning_rate": 0.00019482207849744442,
    "epoch": 0.1057979869440782,
    "step": 3290
  },
  {
    "loss": 0.11908345222473145,
    "grad_norm": 0.6614115834236145,
    "learning_rate": 0.00019478983935330199,
    "epoch": 0.10611956137247966,
    "step": 3300
  },
  {
    "loss": 0.11808741092681885,
    "grad_norm": 0.7376132011413574,
    "learning_rate": 0.00019475750283983248,
    "epoch": 0.10644113580088112,
    "step": 3310
  },
  {
    "loss": 0.0965549647808075,
    "grad_norm": 0.6531370282173157,
    "learning_rate": 0.00019472506899025226,
    "epoch": 0.10676271022928256,
    "step": 3320
  },
  {
    "loss": 0.10322840213775634,
    "grad_norm": 0.39049506187438965,
    "learning_rate": 0.00019469253783787788,
    "epoch": 0.10708428465768402,
    "step": 3330
  },
  {
    "loss": 0.09751293063163757,
    "grad_norm": 0.7664731740951538,
    "learning_rate": 0.00019465990941612573,
    "epoch": 0.10740585908608548,
    "step": 3340
  },
  {
    "loss": 0.10536651611328125,
    "grad_norm": 0.8594152331352234,
    "learning_rate": 0.0001946271837585121,
    "epoch": 0.10772743351448692,
    "step": 3350
  },
  {
    "loss": 0.13831456899642944,
    "grad_norm": 0.48862630128860474,
    "learning_rate": 0.00019459436089865325,
    "epoch": 0.10804900794288838,
    "step": 3360
  },
  {
    "loss": 0.1286901354789734,
    "grad_norm": 0.7956599593162537,
    "learning_rate": 0.0001945614408702652,
    "epoch": 0.10837058237128984,
    "step": 3370
  },
  {
    "loss": 0.09993658065795899,
    "grad_norm": 0.6222932934761047,
    "learning_rate": 0.00019452842370716386,
    "epoch": 0.10869215679969128,
    "step": 3380
  },
  {
    "loss": 0.11540353298187256,
    "grad_norm": 0.6490775942802429,
    "learning_rate": 0.0001944953094432648,
    "epoch": 0.10901373122809274,
    "step": 3390
  },
  {
    "loss": 0.11278325319290161,
    "grad_norm": 0.7428007125854492,
    "learning_rate": 0.00019446209811258352,
    "epoch": 0.1093353056564942,
    "step": 3400
  },
  {
    "loss": 0.09938254356384277,
    "grad_norm": 0.5227949023246765,
    "learning_rate": 0.00019442878974923507,
    "epoch": 0.10965688008489564,
    "step": 3410
  },
  {
    "loss": 0.13301162719726561,
    "grad_norm": 0.8114507794380188,
    "learning_rate": 0.00019439538438743422,
    "epoch": 0.1099784545132971,
    "step": 3420
  },
  {
    "loss": 0.1044914722442627,
    "grad_norm": 0.6826546788215637,
    "learning_rate": 0.0001943618820614954,
    "epoch": 0.11030002894169856,
    "step": 3430
  },
  {
    "loss": 0.10617796182632447,
    "grad_norm": 0.5690863132476807,
    "learning_rate": 0.00019432828280583259,
    "epoch": 0.1106216033701,
    "step": 3440
  },
  {
    "loss": 0.11815530061721802,
    "grad_norm": 0.6827008128166199,
    "learning_rate": 0.00019429458665495941,
    "epoch": 0.11094317779850146,
    "step": 3450
  },
  {
    "loss": 0.08990739583969116,
    "grad_norm": 0.815086841583252,
    "learning_rate": 0.00019426079364348895,
    "epoch": 0.11126475222690292,
    "step": 3460
  },
  {
    "loss": 0.12010560035705567,
    "grad_norm": 0.5905897617340088,
    "learning_rate": 0.0001942269038061338,
    "epoch": 0.11158632665530437,
    "step": 3470
  },
  {
    "loss": 0.08246596455574036,
    "grad_norm": 0.7142632007598877,
    "learning_rate": 0.0001941929171777061,
    "epoch": 0.11190790108370582,
    "step": 3480
  },
  {
    "loss": 0.12285442352294922,
    "grad_norm": 0.48332491517066956,
    "learning_rate": 0.00019415883379311733,
    "epoch": 0.11222947551210728,
    "step": 3490
  },
  {
    "loss": 0.10786828994750977,
    "grad_norm": 0.720145046710968,
    "learning_rate": 0.00019412465368737832,
    "epoch": 0.11255104994050873,
    "step": 3500
  },
  {
    "loss": 0.118403160572052,
    "grad_norm": 0.9487232565879822,
    "learning_rate": 0.00019409037689559932,
    "epoch": 0.11287262436891018,
    "step": 3510
  },
  {
    "loss": 0.13273926973342895,
    "grad_norm": 0.5884988903999329,
    "learning_rate": 0.00019405600345298994,
    "epoch": 0.11319419879731164,
    "step": 3520
  },
  {
    "loss": 0.09212377071380615,
    "grad_norm": 0.6843329071998596,
    "learning_rate": 0.00019402153339485897,
    "epoch": 0.11351577322571309,
    "step": 3530
  },
  {
    "loss": 0.09920811653137207,
    "grad_norm": 0.6576499342918396,
    "learning_rate": 0.0001939869667566145,
    "epoch": 0.11383734765411455,
    "step": 3540
  },
  {
    "loss": 0.10872898101806641,
    "grad_norm": 0.7083240747451782,
    "learning_rate": 0.00019395230357376387,
    "epoch": 0.114158922082516,
    "step": 3550
  },
  {
    "loss": 0.1173086166381836,
    "grad_norm": 0.8997201323509216,
    "learning_rate": 0.00019391754388191346,
    "epoch": 0.11448049651091745,
    "step": 3560
  },
  {
    "loss": 0.11752113103866577,
    "grad_norm": 0.5752458572387695,
    "learning_rate": 0.0001938826877167689,
    "epoch": 0.1148020709393189,
    "step": 3570
  },
  {
    "loss": 0.09167235493659973,
    "grad_norm": 0.28855404257774353,
    "learning_rate": 0.00019384773511413493,
    "epoch": 0.11512364536772036,
    "step": 3580
  },
  {
    "loss": 0.11027594804763793,
    "grad_norm": 0.5791109204292297,
    "learning_rate": 0.00019381268610991521,
    "epoch": 0.11544521979612181,
    "step": 3590
  },
  {
    "loss": 0.0963161051273346,
    "grad_norm": 0.9037119150161743,
    "learning_rate": 0.0001937775407401126,
    "epoch": 0.11576679422452327,
    "step": 3600
  },
  {
    "loss": 0.10461825132369995,
    "grad_norm": 0.7854295969009399,
    "learning_rate": 0.00019374229904082889,
    "epoch": 0.11608836865292473,
    "step": 3610
  },
  {
    "loss": 0.11341136693954468,
    "grad_norm": 0.7756221890449524,
    "learning_rate": 0.00019370696104826474,
    "epoch": 0.11640994308132617,
    "step": 3620
  },
  {
    "loss": 0.11079100370407105,
    "grad_norm": 0.8821120858192444,
    "learning_rate": 0.0001936715267987198,
    "epoch": 0.11673151750972763,
    "step": 3630
  },
  {
    "loss": 0.10128632783889771,
    "grad_norm": 1.0691667795181274,
    "learning_rate": 0.00019363599632859257,
    "epoch": 0.11705309193812909,
    "step": 3640
  },
  {
    "loss": 0.10989173650741577,
    "grad_norm": 0.773428738117218,
    "learning_rate": 0.00019360036967438045,
    "epoch": 0.11737466636653053,
    "step": 3650
  },
  {
    "loss": 0.12041290998458862,
    "grad_norm": 0.6354496479034424,
    "learning_rate": 0.00019356464687267952,
    "epoch": 0.11769624079493199,
    "step": 3660
  },
  {
    "loss": 0.11748863458633423,
    "grad_norm": 0.4678747355937958,
    "learning_rate": 0.00019352882796018475,
    "epoch": 0.11801781522333345,
    "step": 3670
  },
  {
    "loss": 0.12304327487945557,
    "grad_norm": 0.934917688369751,
    "learning_rate": 0.0001934929129736898,
    "epoch": 0.11833938965173489,
    "step": 3680
  },
  {
    "loss": 0.119256591796875,
    "grad_norm": 0.6330606341362,
    "learning_rate": 0.00019345690195008694,
    "epoch": 0.11866096408013635,
    "step": 3690
  },
  {
    "loss": 0.11701215505599975,
    "grad_norm": 0.6313028335571289,
    "learning_rate": 0.0001934207949263672,
    "epoch": 0.11898253850853781,
    "step": 3700
  },
  {
    "loss": 0.08980629444122315,
    "grad_norm": 0.5904396772384644,
    "learning_rate": 0.00019338459193962013,
    "epoch": 0.11930411293693925,
    "step": 3710
  },
  {
    "loss": 0.10626832246780396,
    "grad_norm": 0.7841988205909729,
    "learning_rate": 0.00019334829302703393,
    "epoch": 0.11962568736534071,
    "step": 3720
  },
  {
    "loss": 0.1016309142112732,
    "grad_norm": 0.8079530000686646,
    "learning_rate": 0.0001933118982258953,
    "epoch": 0.11994726179374215,
    "step": 3730
  },
  {
    "loss": 0.10360891819000244,
    "grad_norm": 0.6402804851531982,
    "learning_rate": 0.00019327540757358943,
    "epoch": 0.12026883622214361,
    "step": 3740
  },
  {
    "loss": 0.11208920478820801,
    "grad_norm": 0.535378098487854,
    "learning_rate": 0.00019323882110759997,
    "epoch": 0.12059041065054507,
    "step": 3750
  },
  {
    "loss": 0.13467209339141845,
    "grad_norm": 0.6953940391540527,
    "learning_rate": 0.00019320213886550903,
    "epoch": 0.12091198507894652,
    "step": 3760
  },
  {
    "loss": 0.0965237021446228,
    "grad_norm": 0.6152417659759521,
    "learning_rate": 0.00019316536088499707,
    "epoch": 0.12123355950734797,
    "step": 3770
  },
  {
    "loss": 0.1212538480758667,
    "grad_norm": 1.36095130443573,
    "learning_rate": 0.0001931284872038429,
    "epoch": 0.12155513393574943,
    "step": 3780
  },
  {
    "loss": 0.09757543206214905,
    "grad_norm": 0.7375221252441406,
    "learning_rate": 0.00019309151785992364,
    "epoch": 0.12187670836415088,
    "step": 3790
  },
  {
    "loss": 0.10885839462280274,
    "grad_norm": 0.8550742864608765,
    "learning_rate": 0.00019305445289121464,
    "epoch": 0.12219828279255233,
    "step": 3800
  },
  {
    "loss": 0.0911238431930542,
    "grad_norm": 0.7616929411888123,
    "learning_rate": 0.00019301729233578953,
    "epoch": 0.12251985722095379,
    "step": 3810
  },
  {
    "loss": 0.0981452465057373,
    "grad_norm": 0.5779492855072021,
    "learning_rate": 0.0001929800362318201,
    "epoch": 0.12284143164935524,
    "step": 3820
  },
  {
    "loss": 0.1250522494316101,
    "grad_norm": 0.5627166032791138,
    "learning_rate": 0.00019294268461757626,
    "epoch": 0.1231630060777567,
    "step": 3830
  },
  {
    "loss": 0.11018778085708618,
    "grad_norm": 0.7056961059570312,
    "learning_rate": 0.00019290523753142617,
    "epoch": 0.12348458050615815,
    "step": 3840
  },
  {
    "loss": 0.10934178829193116,
    "grad_norm": 0.5087243318557739,
    "learning_rate": 0.0001928676950118358,
    "epoch": 0.1238061549345596,
    "step": 3850
  },
  {
    "loss": 0.10876210927963256,
    "grad_norm": 0.5404261350631714,
    "learning_rate": 0.00019283005709736943,
    "epoch": 0.12412772936296106,
    "step": 3860
  },
  {
    "loss": 0.11565039157867432,
    "grad_norm": 0.6383166909217834,
    "learning_rate": 0.00019279232382668916,
    "epoch": 0.12444930379136251,
    "step": 3870
  },
  {
    "loss": 0.12286177873611451,
    "grad_norm": 0.5657302737236023,
    "learning_rate": 0.00019275449523855503,
    "epoch": 0.12477087821976396,
    "step": 3880
  },
  {
    "loss": 0.09651960134506225,
    "grad_norm": 0.5566875338554382,
    "learning_rate": 0.00019271657137182513,
    "epoch": 0.12509245264816543,
    "step": 3890
  },
  {
    "loss": 0.09890485405921937,
    "grad_norm": 0.4659144878387451,
    "learning_rate": 0.0001926785522654553,
    "epoch": 0.12541402707656688,
    "step": 3900
  },
  {
    "loss": 0.0990807831287384,
    "grad_norm": 0.8052150011062622,
    "learning_rate": 0.00019264043795849927,
    "epoch": 0.12573560150496832,
    "step": 3910
  },
  {
    "loss": 0.11594893932342529,
    "grad_norm": 1.0626065731048584,
    "learning_rate": 0.00019260222849010848,
    "epoch": 0.1260571759333698,
    "step": 3920
  },
  {
    "loss": 0.12493889331817627,
    "grad_norm": 0.8749810457229614,
    "learning_rate": 0.00019256392389953227,
    "epoch": 0.12637875036177124,
    "step": 3930
  },
  {
    "loss": 0.1070855975151062,
    "grad_norm": 0.6140100359916687,
    "learning_rate": 0.00019252552422611752,
    "epoch": 0.12670032479017268,
    "step": 3940
  },
  {
    "loss": 0.08729969263076783,
    "grad_norm": 0.4629083275794983,
    "learning_rate": 0.0001924870295093089,
    "epoch": 0.12702189921857415,
    "step": 3950
  },
  {
    "loss": 0.12051993608474731,
    "grad_norm": 0.8901966214179993,
    "learning_rate": 0.0001924484397886487,
    "epoch": 0.1273434736469756,
    "step": 3960
  },
  {
    "loss": 0.10127103328704834,
    "grad_norm": 0.9668626189231873,
    "learning_rate": 0.00019240975510377673,
    "epoch": 0.12766504807537704,
    "step": 3970
  },
  {
    "loss": 0.09287093877792359,
    "grad_norm": 0.6984792947769165,
    "learning_rate": 0.00019237097549443038,
    "epoch": 0.1279866225037785,
    "step": 3980
  },
  {
    "loss": 0.12040114402770996,
    "grad_norm": 0.9510995149612427,
    "learning_rate": 0.00019233210100044458,
    "epoch": 0.12830819693217996,
    "step": 3990
  },
  {
    "loss": 0.12150945663452148,
    "grad_norm": 0.6838380694389343,
    "learning_rate": 0.00019229313166175174,
    "epoch": 0.1286297713605814,
    "step": 4000
  },
  {
    "loss": 0.1207665205001831,
    "grad_norm": 0.6368831992149353,
    "learning_rate": 0.0001922540675183816,
    "epoch": 0.12895134578898285,
    "step": 4010
  },
  {
    "loss": 0.10408351421356202,
    "grad_norm": 0.921284556388855,
    "learning_rate": 0.00019221490861046138,
    "epoch": 0.12927292021738432,
    "step": 4020
  },
  {
    "loss": 0.11057004928588868,
    "grad_norm": 0.7935436964035034,
    "learning_rate": 0.00019217565497821557,
    "epoch": 0.12959449464578576,
    "step": 4030
  },
  {
    "loss": 0.09651212692260742,
    "grad_norm": 0.5812630653381348,
    "learning_rate": 0.00019213630666196604,
    "epoch": 0.1299160690741872,
    "step": 4040
  },
  {
    "loss": 0.09005358219146728,
    "grad_norm": 0.3681035339832306,
    "learning_rate": 0.00019209686370213187,
    "epoch": 0.13023764350258868,
    "step": 4050
  },
  {
    "loss": 0.10046392679214478,
    "grad_norm": 0.6533730626106262,
    "learning_rate": 0.00019205732613922937,
    "epoch": 0.13055921793099012,
    "step": 4060
  },
  {
    "loss": 0.10442250967025757,
    "grad_norm": 0.43741610646247864,
    "learning_rate": 0.000192017694013872,
    "epoch": 0.13088079235939157,
    "step": 4070
  },
  {
    "loss": 0.08786097168922424,
    "grad_norm": 0.6568700075149536,
    "learning_rate": 0.0001919779673667704,
    "epoch": 0.13120236678779304,
    "step": 4080
  },
  {
    "loss": 0.10092254877090454,
    "grad_norm": 0.9713712334632874,
    "learning_rate": 0.0001919381462387323,
    "epoch": 0.13152394121619448,
    "step": 4090
  },
  {
    "loss": 0.1092909812927246,
    "grad_norm": 0.6388143301010132,
    "learning_rate": 0.00019189823067066247,
    "epoch": 0.13184551564459593,
    "step": 4100
  },
  {
    "loss": 0.11206631660461426,
    "grad_norm": 0.6925147175788879,
    "learning_rate": 0.00019185822070356268,
    "epoch": 0.1321670900729974,
    "step": 4110
  },
  {
    "loss": 0.10706722736358643,
    "grad_norm": 0.7303212285041809,
    "learning_rate": 0.00019181811637853166,
    "epoch": 0.13248866450139885,
    "step": 4120
  },
  {
    "loss": 0.10609568357467651,
    "grad_norm": 0.3527115285396576,
    "learning_rate": 0.00019177791773676513,
    "epoch": 0.1328102389298003,
    "step": 4130
  },
  {
    "loss": 0.09472618699073791,
    "grad_norm": 0.37147092819213867,
    "learning_rate": 0.00019173762481955556,
    "epoch": 0.13313181335820176,
    "step": 4140
  },
  {
    "loss": 0.12064085006713868,
    "grad_norm": 0.7030910849571228,
    "learning_rate": 0.0001916972376682924,
    "epoch": 0.1334533877866032,
    "step": 4150
  },
  {
    "loss": 0.11871163845062256,
    "grad_norm": 1.1606062650680542,
    "learning_rate": 0.00019165675632446186,
    "epoch": 0.13377496221500465,
    "step": 4160
  },
  {
    "loss": 0.1308833122253418,
    "grad_norm": 0.9086573719978333,
    "learning_rate": 0.00019161618082964691,
    "epoch": 0.13409653664340612,
    "step": 4170
  },
  {
    "loss": 0.11038376092910766,
    "grad_norm": 0.7408903241157532,
    "learning_rate": 0.00019157551122552712,
    "epoch": 0.13441811107180757,
    "step": 4180
  },
  {
    "loss": 0.08468644618988037,
    "grad_norm": 0.9437806606292725,
    "learning_rate": 0.00019153474755387892,
    "epoch": 0.134739685500209,
    "step": 4190
  },
  {
    "loss": 0.08930802345275879,
    "grad_norm": 0.7443150877952576,
    "learning_rate": 0.00019149388985657522,
    "epoch": 0.13506125992861048,
    "step": 4200
  },
  {
    "loss": 0.08937476873397827,
    "grad_norm": 0.9245612025260925,
    "learning_rate": 0.00019145293817558558,
    "epoch": 0.13538283435701193,
    "step": 4210
  },
  {
    "loss": 0.1095966100692749,
    "grad_norm": 0.564530074596405,
    "learning_rate": 0.00019141189255297608,
    "epoch": 0.13570440878541337,
    "step": 4220
  },
  {
    "loss": 0.09740201234817505,
    "grad_norm": 0.6564256548881531,
    "learning_rate": 0.00019137075303090934,
    "epoch": 0.13602598321381484,
    "step": 4230
  },
  {
    "loss": 0.10432863235473633,
    "grad_norm": 0.6156473159790039,
    "learning_rate": 0.0001913295196516443,
    "epoch": 0.1363475576422163,
    "step": 4240
  },
  {
    "loss": 0.09986402988433837,
    "grad_norm": 0.9064126014709473,
    "learning_rate": 0.00019128819245753652,
    "epoch": 0.13666913207061773,
    "step": 4250
  },
  {
    "loss": 0.09308716058731079,
    "grad_norm": 0.5122585296630859,
    "learning_rate": 0.0001912467714910378,
    "epoch": 0.1369907064990192,
    "step": 4260
  },
  {
    "loss": 0.10415486097335816,
    "grad_norm": 0.966684103012085,
    "learning_rate": 0.00019120525679469622,
    "epoch": 0.13731228092742065,
    "step": 4270
  },
  {
    "loss": 0.09036895036697387,
    "grad_norm": 0.8586301803588867,
    "learning_rate": 0.00019116364841115628,
    "epoch": 0.1376338553558221,
    "step": 4280
  },
  {
    "loss": 0.09274351596832275,
    "grad_norm": 0.5143318772315979,
    "learning_rate": 0.00019112194638315862,
    "epoch": 0.13795542978422357,
    "step": 4290
  },
  {
    "loss": 0.10373387336730958,
    "grad_norm": 0.765060305595398,
    "learning_rate": 0.00019108015075354002,
    "epoch": 0.138277004212625,
    "step": 4300
  },
  {
    "loss": 0.09613911509513855,
    "grad_norm": 0.6508856415748596,
    "learning_rate": 0.0001910382615652336,
    "epoch": 0.13859857864102645,
    "step": 4310
  },
  {
    "loss": 0.11943503618240356,
    "grad_norm": 0.7866203188896179,
    "learning_rate": 0.00019099627886126839,
    "epoch": 0.13892015306942793,
    "step": 4320
  },
  {
    "loss": 0.11114203929901123,
    "grad_norm": 0.4582284092903137,
    "learning_rate": 0.00019095420268476956,
    "epoch": 0.13924172749782937,
    "step": 4330
  },
  {
    "loss": 0.08318037986755371,
    "grad_norm": 0.7825440168380737,
    "learning_rate": 0.0001909120330789583,
    "epoch": 0.13956330192623081,
    "step": 4340
  },
  {
    "loss": 0.10253852605819702,
    "grad_norm": 0.7011892795562744,
    "learning_rate": 0.0001908697700871518,
    "epoch": 0.1398848763546323,
    "step": 4350
  },
  {
    "loss": 0.1049849271774292,
    "grad_norm": 0.34880417585372925,
    "learning_rate": 0.00019082741375276308,
    "epoch": 0.14020645078303373,
    "step": 4360
  },
  {
    "loss": 0.10247675180435181,
    "grad_norm": 0.5124948620796204,
    "learning_rate": 0.00019078496411930118,
    "epoch": 0.14052802521143518,
    "step": 4370
  },
  {
    "loss": 0.11412211656570434,
    "grad_norm": 0.3058924674987793,
    "learning_rate": 0.00019074242123037083,
    "epoch": 0.14084959963983665,
    "step": 4380
  },
  {
    "loss": 0.12336907386779786,
    "grad_norm": 0.8137784004211426,
    "learning_rate": 0.00019069978512967266,
    "epoch": 0.1411711740682381,
    "step": 4390
  },
  {
    "loss": 0.09460805058479309,
    "grad_norm": 1.2490235567092896,
    "learning_rate": 0.00019065705586100303,
    "epoch": 0.14149274849663954,
    "step": 4400
  },
  {
    "loss": 0.11223421096801758,
    "grad_norm": 0.8012495040893555,
    "learning_rate": 0.00019061423346825395,
    "epoch": 0.141814322925041,
    "step": 4410
  },
  {
    "loss": 0.09242308735847474,
    "grad_norm": 0.43816885352134705,
    "learning_rate": 0.00019057131799541318,
    "epoch": 0.14213589735344245,
    "step": 4420
  },
  {
    "loss": 0.10507001876831054,
    "grad_norm": 0.4903210401535034,
    "learning_rate": 0.00019052830948656398,
    "epoch": 0.1424574717818439,
    "step": 4430
  },
  {
    "loss": 0.09061969518661499,
    "grad_norm": 0.7580713629722595,
    "learning_rate": 0.00019048520798588525,
    "epoch": 0.14277904621024537,
    "step": 4440
  },
  {
    "loss": 0.09204312562942504,
    "grad_norm": 0.6889179348945618,
    "learning_rate": 0.00019044201353765142,
    "epoch": 0.1431006206386468,
    "step": 4450
  },
  {
    "loss": 0.1126853346824646,
    "grad_norm": 0.43935051560401917,
    "learning_rate": 0.00019039872618623234,
    "epoch": 0.14342219506704826,
    "step": 4460
  },
  {
    "loss": 0.11694524288177491,
    "grad_norm": 1.1189888715744019,
    "learning_rate": 0.00019035534597609339,
    "epoch": 0.14374376949544973,
    "step": 4470
  },
  {
    "loss": 0.12891401052474977,
    "grad_norm": 0.94930499792099,
    "learning_rate": 0.0001903118729517952,
    "epoch": 0.14406534392385117,
    "step": 4480
  },
  {
    "loss": 0.10222654342651367,
    "grad_norm": 0.7653531432151794,
    "learning_rate": 0.00019026830715799387,
    "epoch": 0.14438691835225262,
    "step": 4490
  },
  {
    "loss": 0.08867776989936829,
    "grad_norm": 0.6891590356826782,
    "learning_rate": 0.00019022464863944073,
    "epoch": 0.1447084927806541,
    "step": 4500
  },
  {
    "loss": 0.10994209051132202,
    "grad_norm": 0.8964745998382568,
    "learning_rate": 0.0001901808974409823,
    "epoch": 0.14503006720905554,
    "step": 4510
  },
  {
    "loss": 0.0991736888885498,
    "grad_norm": 0.5754009485244751,
    "learning_rate": 0.0001901370536075605,
    "epoch": 0.14535164163745698,
    "step": 4520
  },
  {
    "loss": 0.09913696646690369,
    "grad_norm": 0.744156539440155,
    "learning_rate": 0.00019009311718421213,
    "epoch": 0.14567321606585845,
    "step": 4530
  },
  {
    "loss": 0.1233474612236023,
    "grad_norm": 0.5341463685035706,
    "learning_rate": 0.0001900490882160693,
    "epoch": 0.1459947904942599,
    "step": 4540
  },
  {
    "loss": 0.08937983512878418,
    "grad_norm": 0.7251766324043274,
    "learning_rate": 0.00019000496674835915,
    "epoch": 0.14631636492266134,
    "step": 4550
  },
  {
    "loss": 0.09609851837158204,
    "grad_norm": 0.5500466823577881,
    "learning_rate": 0.00018996075282640379,
    "epoch": 0.1466379393510628,
    "step": 4560
  },
  {
    "loss": 0.11425167322158813,
    "grad_norm": 1.043576717376709,
    "learning_rate": 0.00018991644649562028,
    "epoch": 0.14695951377946426,
    "step": 4570
  },
  {
    "loss": 0.11258641481399537,
    "grad_norm": 0.4216415286064148,
    "learning_rate": 0.00018987204780152072,
    "epoch": 0.1472810882078657,
    "step": 4580
  },
  {
    "loss": 0.12220112085342408,
    "grad_norm": 1.0332443714141846,
    "learning_rate": 0.000189827556789712,
    "epoch": 0.14760266263626717,
    "step": 4590
  },
  {
    "loss": 0.11982320547103882,
    "grad_norm": 1.0259549617767334,
    "learning_rate": 0.0001897829735058958,
    "epoch": 0.14792423706466862,
    "step": 4600
  },
  {
    "loss": 0.10923948287963867,
    "grad_norm": 0.7364688515663147,
    "learning_rate": 0.00018973829799586868,
    "epoch": 0.14824581149307006,
    "step": 4610
  },
  {
    "loss": 0.10167646408081055,
    "grad_norm": 0.8405701518058777,
    "learning_rate": 0.00018969353030552188,
    "epoch": 0.14856738592147153,
    "step": 4620
  },
  {
    "loss": 0.09561271667480468,
    "grad_norm": 0.8111253976821899,
    "learning_rate": 0.00018964867048084134,
    "epoch": 0.14888896034987298,
    "step": 4630
  },
  {
    "loss": 0.10397641658782959,
    "grad_norm": 0.8518447279930115,
    "learning_rate": 0.00018960371856790764,
    "epoch": 0.14921053477827442,
    "step": 4640
  },
  {
    "loss": 0.12073606252670288,
    "grad_norm": 0.766708493232727,
    "learning_rate": 0.00018955867461289597,
    "epoch": 0.1495321092066759,
    "step": 4650
  },
  {
    "loss": 0.1148116946220398,
    "grad_norm": 0.5033470392227173,
    "learning_rate": 0.00018951353866207603,
    "epoch": 0.14985368363507734,
    "step": 4660
  },
  {
    "eval_loss": 0.10172691941261292,
    "eval_runtime": 34.399,
    "eval_samples_per_second": 146.109,
    "eval_steps_per_second": 36.542,
    "epoch": 0.14991799852075763,
    "step": 4662
  },
  {
    "loss": 0.1416254758834839,
    "grad_norm": 0.5513647198677063,
    "learning_rate": 0.00018946831076181204,
    "epoch": 0.15017525806347878,
    "step": 4670
  },
  {
    "loss": 0.1062854290008545,
    "grad_norm": 0.40923750400543213,
    "learning_rate": 0.00018942299095856267,
    "epoch": 0.15049683249188026,
    "step": 4680
  },
  {
    "loss": 0.0943223237991333,
    "grad_norm": 0.5009618401527405,
    "learning_rate": 0.00018937757929888104,
    "epoch": 0.1508184069202817,
    "step": 4690
  },
  {
    "loss": 0.11033749580383301,
    "grad_norm": 0.3853249251842499,
    "learning_rate": 0.0001893320758294146,
    "epoch": 0.15113998134868314,
    "step": 4700
  },
  {
    "loss": 0.0948384165763855,
    "grad_norm": 0.8394500613212585,
    "learning_rate": 0.00018928648059690498,
    "epoch": 0.15146155577708462,
    "step": 4710
  },
  {
    "loss": 0.09225496649742126,
    "grad_norm": 1.1599781513214111,
    "learning_rate": 0.00018924079364818826,
    "epoch": 0.15178313020548606,
    "step": 4720
  },
  {
    "loss": 0.13177945613861083,
    "grad_norm": 0.8892496228218079,
    "learning_rate": 0.0001891950150301947,
    "epoch": 0.1521047046338875,
    "step": 4730
  },
  {
    "loss": 0.09583035111427307,
    "grad_norm": 0.7567657828330994,
    "learning_rate": 0.0001891491447899486,
    "epoch": 0.15242627906228898,
    "step": 4740
  },
  {
    "loss": 0.08644075989723206,
    "grad_norm": 0.488677442073822,
    "learning_rate": 0.00018910318297456845,
    "epoch": 0.15274785349069042,
    "step": 4750
  },
  {
    "loss": 0.09413954019546508,
    "grad_norm": 0.5898675918579102,
    "learning_rate": 0.00018905712963126684,
    "epoch": 0.15306942791909187,
    "step": 4760
  },
  {
    "loss": 0.11096861362457275,
    "grad_norm": 1.2141278982162476,
    "learning_rate": 0.00018901098480735037,
    "epoch": 0.15339100234749334,
    "step": 4770
  },
  {
    "loss": 0.09794067144393921,
    "grad_norm": 0.6396763920783997,
    "learning_rate": 0.00018896474855021955,
    "epoch": 0.15371257677589478,
    "step": 4780
  },
  {
    "loss": 0.11338918209075928,
    "grad_norm": 0.7813658714294434,
    "learning_rate": 0.00018891842090736888,
    "epoch": 0.15403415120429623,
    "step": 4790
  },
  {
    "loss": 0.10229338407516479,
    "grad_norm": 0.5102593898773193,
    "learning_rate": 0.00018887200192638668,
    "epoch": 0.1543557256326977,
    "step": 4800
  },
  {
    "loss": 0.11056987047195435,
    "grad_norm": 0.933292031288147,
    "learning_rate": 0.0001888254916549551,
    "epoch": 0.15467730006109914,
    "step": 4810
  },
  {
    "loss": 0.11467688083648682,
    "grad_norm": 0.8504388928413391,
    "learning_rate": 0.00018877889014085015,
    "epoch": 0.1549988744895006,
    "step": 4820
  },
  {
    "loss": 0.10686609745025635,
    "grad_norm": 0.7789998054504395,
    "learning_rate": 0.00018873219743194145,
    "epoch": 0.15532044891790206,
    "step": 4830
  },
  {
    "loss": 0.08134291768074035,
    "grad_norm": 0.5356742143630981,
    "learning_rate": 0.00018868541357619235,
    "epoch": 0.1556420233463035,
    "step": 4840
  },
  {
    "loss": 0.12447811365127563,
    "grad_norm": 0.6773043870925903,
    "learning_rate": 0.0001886385386216598,
    "epoch": 0.15596359777470495,
    "step": 4850
  },
  {
    "loss": 0.1055222988128662,
    "grad_norm": 0.5099416375160217,
    "learning_rate": 0.0001885915726164944,
    "epoch": 0.15628517220310642,
    "step": 4860
  },
  {
    "loss": 0.11184096336364746,
    "grad_norm": 1.135096549987793,
    "learning_rate": 0.00018854451560894016,
    "epoch": 0.15660674663150786,
    "step": 4870
  },
  {
    "loss": 0.11401809453964233,
    "grad_norm": 0.5366411805152893,
    "learning_rate": 0.00018849736764733468,
    "epoch": 0.1569283210599093,
    "step": 4880
  },
  {
    "loss": 0.09933592081069946,
    "grad_norm": 0.5675321221351624,
    "learning_rate": 0.00018845012878010895,
    "epoch": 0.15724989548831078,
    "step": 4890
  },
  {
    "loss": 0.09386941194534301,
    "grad_norm": 0.517961323261261,
    "learning_rate": 0.00018840279905578733,
    "epoch": 0.15757146991671223,
    "step": 4900
  },
  {
    "loss": 0.11577614545822143,
    "grad_norm": 0.9143649339675903,
    "learning_rate": 0.0001883553785229875,
    "epoch": 0.15789304434511367,
    "step": 4910
  },
  {
    "loss": 0.08844751715660096,
    "grad_norm": 0.6606913805007935,
    "learning_rate": 0.00018830786723042042,
    "epoch": 0.15821461877351514,
    "step": 4920
  },
  {
    "loss": 0.09490571618080139,
    "grad_norm": 0.6583348512649536,
    "learning_rate": 0.00018826026522689033,
    "epoch": 0.1585361932019166,
    "step": 4930
  },
  {
    "loss": 0.09571805000305175,
    "grad_norm": 0.8969154953956604,
    "learning_rate": 0.00018821257256129462,
    "epoch": 0.15885776763031803,
    "step": 4940
  },
  {
    "loss": 0.10312877893447876,
    "grad_norm": 0.32090887427330017,
    "learning_rate": 0.00018816478928262377,
    "epoch": 0.1591793420587195,
    "step": 4950
  },
  {
    "loss": 0.09461336731910705,
    "grad_norm": 0.5978054404258728,
    "learning_rate": 0.00018811691543996136,
    "epoch": 0.15950091648712095,
    "step": 4960
  },
  {
    "loss": 0.09903188943862914,
    "grad_norm": 0.30837303400039673,
    "learning_rate": 0.000188068951082484,
    "epoch": 0.1598224909155224,
    "step": 4970
  },
  {
    "loss": 0.1048235297203064,
    "grad_norm": 0.667777419090271,
    "learning_rate": 0.00018802089625946135,
    "epoch": 0.16014406534392386,
    "step": 4980
  },
  {
    "loss": 0.0945176899433136,
    "grad_norm": 0.5308945178985596,
    "learning_rate": 0.0001879727510202559,
    "epoch": 0.1604656397723253,
    "step": 4990
  },
  {
    "loss": 0.11314963102340699,
    "grad_norm": 0.8820076584815979,
    "learning_rate": 0.00018792451541432302,
    "epoch": 0.16078721420072675,
    "step": 5000
  },
  {
    "loss": 0.09827512502670288,
    "grad_norm": 0.4974328875541687,
    "learning_rate": 0.000187876189491211,
    "epoch": 0.16110878862912822,
    "step": 5010
  },
  {
    "loss": 0.14028635025024414,
    "grad_norm": 0.7012617588043213,
    "learning_rate": 0.00018782777330056082,
    "epoch": 0.16143036305752967,
    "step": 5020
  },
  {
    "loss": 0.09744994044303894,
    "grad_norm": 0.5276012420654297,
    "learning_rate": 0.00018777926689210619,
    "epoch": 0.1617519374859311,
    "step": 5030
  },
  {
    "loss": 0.11486378908157349,
    "grad_norm": 0.6841879487037659,
    "learning_rate": 0.0001877306703156735,
    "epoch": 0.16207351191433259,
    "step": 5040
  },
  {
    "loss": 0.11499395370483398,
    "grad_norm": 0.5109455585479736,
    "learning_rate": 0.0001876819836211818,
    "epoch": 0.16239508634273403,
    "step": 5050
  },
  {
    "loss": 0.11083143949508667,
    "grad_norm": 0.9456536769866943,
    "learning_rate": 0.00018763320685864264,
    "epoch": 0.16271666077113547,
    "step": 5060
  },
  {
    "loss": 0.09588128924369813,
    "grad_norm": 0.7423805594444275,
    "learning_rate": 0.00018758434007816018,
    "epoch": 0.16303823519953695,
    "step": 5070
  },
  {
    "loss": 0.11503241062164307,
    "grad_norm": 0.44641563296318054,
    "learning_rate": 0.000187535383329931,
    "epoch": 0.1633598096279384,
    "step": 5080
  },
  {
    "loss": 0.12453733682632447,
    "grad_norm": 1.7256242036819458,
    "learning_rate": 0.00018748633666424404,
    "epoch": 0.16368138405633983,
    "step": 5090
  },
  {
    "loss": 0.13549509048461914,
    "grad_norm": 0.8264243006706238,
    "learning_rate": 0.00018743720013148068,
    "epoch": 0.1640029584847413,
    "step": 5100
  },
  {
    "loss": 0.13447096347808837,
    "grad_norm": 0.44616052508354187,
    "learning_rate": 0.0001873879737821146,
    "epoch": 0.16432453291314275,
    "step": 5110
  },
  {
    "loss": 0.11622662544250488,
    "grad_norm": 0.5612630248069763,
    "learning_rate": 0.00018733865766671175,
    "epoch": 0.1646461073415442,
    "step": 5120
  },
  {
    "loss": 0.10126912593841553,
    "grad_norm": 0.6290414333343506,
    "learning_rate": 0.00018728925183593024,
    "epoch": 0.16496768176994567,
    "step": 5130
  },
  {
    "loss": 0.11963098049163819,
    "grad_norm": 0.7760690450668335,
    "learning_rate": 0.00018723975634052037,
    "epoch": 0.1652892561983471,
    "step": 5140
  },
  {
    "loss": 0.13792331218719484,
    "grad_norm": 1.2590798139572144,
    "learning_rate": 0.00018719017123132456,
    "epoch": 0.16561083062674856,
    "step": 5150
  },
  {
    "loss": 0.09700539112091064,
    "grad_norm": 0.6620594263076782,
    "learning_rate": 0.0001871404965592772,
    "epoch": 0.16593240505515,
    "step": 5160
  },
  {
    "loss": 0.0915036141872406,
    "grad_norm": 0.529654324054718,
    "learning_rate": 0.00018709073237540487,
    "epoch": 0.16625397948355147,
    "step": 5170
  },
  {
    "loss": 0.11442764997482299,
    "grad_norm": 0.749620258808136,
    "learning_rate": 0.00018704087873082584,
    "epoch": 0.16657555391195292,
    "step": 5180
  },
  {
    "loss": 0.1019809365272522,
    "grad_norm": 0.9167417883872986,
    "learning_rate": 0.0001869909356767505,
    "epoch": 0.16689712834035436,
    "step": 5190
  },
  {
    "loss": 0.08411781787872315,
    "grad_norm": 0.7064757943153381,
    "learning_rate": 0.00018694090326448098,
    "epoch": 0.16721870276875583,
    "step": 5200
  },
  {
    "loss": 0.09732612371444702,
    "grad_norm": 0.8301167488098145,
    "learning_rate": 0.00018689078154541112,
    "epoch": 0.16754027719715728,
    "step": 5210
  },
  {
    "loss": 0.09780767560005188,
    "grad_norm": 0.8980233073234558,
    "learning_rate": 0.0001868405705710267,
    "epoch": 0.16786185162555872,
    "step": 5220
  },
  {
    "loss": 0.10536625385284423,
    "grad_norm": 0.7485562562942505,
    "learning_rate": 0.00018679027039290497,
    "epoch": 0.1681834260539602,
    "step": 5230
  },
  {
    "loss": 0.11484253406524658,
    "grad_norm": 0.8951029777526855,
    "learning_rate": 0.000186739881062715,
    "epoch": 0.16850500048236164,
    "step": 5240
  },
  {
    "loss": 0.12118645906448364,
    "grad_norm": 0.6397461295127869,
    "learning_rate": 0.00018668940263221734,
    "epoch": 0.16882657491076308,
    "step": 5250
  },
  {
    "loss": 0.09094181060791015,
    "grad_norm": 0.5097070932388306,
    "learning_rate": 0.000186638835153264,
    "epoch": 0.16914814933916456,
    "step": 5260
  },
  {
    "loss": 0.1319599747657776,
    "grad_norm": 0.5678567886352539,
    "learning_rate": 0.0001865881786777986,
    "epoch": 0.169469723767566,
    "step": 5270
  },
  {
    "loss": 0.10581289529800415,
    "grad_norm": 0.7563341856002808,
    "learning_rate": 0.0001865374332578561,
    "epoch": 0.16979129819596744,
    "step": 5280
  },
  {
    "loss": 0.10248388051986694,
    "grad_norm": 0.7660218477249146,
    "learning_rate": 0.00018648659894556283,
    "epoch": 0.17011287262436892,
    "step": 5290
  },
  {
    "loss": 0.09813046455383301,
    "grad_norm": 0.33348825573921204,
    "learning_rate": 0.00018643567579313647,
    "epoch": 0.17043444705277036,
    "step": 5300
  },
  {
    "loss": 0.10771477222442627,
    "grad_norm": 0.7716835737228394,
    "learning_rate": 0.00018638466385288588,
    "epoch": 0.1707560214811718,
    "step": 5310
  },
  {
    "loss": 0.08386213183403016,
    "grad_norm": 0.3895624876022339,
    "learning_rate": 0.00018633356317721125,
    "epoch": 0.17107759590957328,
    "step": 5320
  },
  {
    "loss": 0.09969307780265808,
    "grad_norm": 0.5911557078361511,
    "learning_rate": 0.00018628237381860377,
    "epoch": 0.17139917033797472,
    "step": 5330
  },
  {
    "loss": 0.10970526933670044,
    "grad_norm": 0.7556646466255188,
    "learning_rate": 0.0001862310958296458,
    "epoch": 0.17172074476637617,
    "step": 5340
  },
  {
    "loss": 0.10416722297668457,
    "grad_norm": 0.6202457547187805,
    "learning_rate": 0.0001861797292630108,
    "epoch": 0.17204231919477764,
    "step": 5350
  },
  {
    "loss": 0.09194405674934387,
    "grad_norm": 0.5211172699928284,
    "learning_rate": 0.00018612827417146315,
    "epoch": 0.17236389362317908,
    "step": 5360
  },
  {
    "loss": 0.12105774879455566,
    "grad_norm": 0.9380770921707153,
    "learning_rate": 0.00018607673060785812,
    "epoch": 0.17268546805158053,
    "step": 5370
  },
  {
    "loss": 0.08323792219161988,
    "grad_norm": 0.9509512782096863,
    "learning_rate": 0.00018602509862514194,
    "epoch": 0.173007042479982,
    "step": 5380
  },
  {
    "loss": 0.09783884286880493,
    "grad_norm": 0.532935619354248,
    "learning_rate": 0.00018597337827635165,
    "epoch": 0.17332861690838344,
    "step": 5390
  },
  {
    "loss": 0.11766867637634278,
    "grad_norm": 0.4933967590332031,
    "learning_rate": 0.00018592156961461508,
    "epoch": 0.1736501913367849,
    "step": 5400
  },
  {
    "loss": 0.10667554140090943,
    "grad_norm": 0.463371217250824,
    "learning_rate": 0.0001858696726931507,
    "epoch": 0.17397176576518636,
    "step": 5410
  },
  {
    "loss": 0.1050226092338562,
    "grad_norm": 1.000925064086914,
    "learning_rate": 0.00018581768756526762,
    "epoch": 0.1742933401935878,
    "step": 5420
  },
  {
    "loss": 0.0907900094985962,
    "grad_norm": 0.8843211531639099,
    "learning_rate": 0.00018576561428436576,
    "epoch": 0.17461491462198925,
    "step": 5430
  },
  {
    "loss": 0.09724093675613403,
    "grad_norm": 0.581278383731842,
    "learning_rate": 0.00018571345290393539,
    "epoch": 0.17493648905039072,
    "step": 5440
  },
  {
    "loss": 0.10390642881393433,
    "grad_norm": 0.5287315249443054,
    "learning_rate": 0.00018566120347755734,
    "epoch": 0.17525806347879216,
    "step": 5450
  },
  {
    "loss": 0.11524388790130616,
    "grad_norm": 0.5281156301498413,
    "learning_rate": 0.0001856088660589029,
    "epoch": 0.1755796379071936,
    "step": 5460
  },
  {
    "loss": 0.10431855916976929,
    "grad_norm": 0.7305079698562622,
    "learning_rate": 0.0001855564407017337,
    "epoch": 0.17590121233559508,
    "step": 5470
  },
  {
    "loss": 0.10620676279067993,
    "grad_norm": 0.29204392433166504,
    "learning_rate": 0.00018550392745990177,
    "epoch": 0.17622278676399653,
    "step": 5480
  },
  {
    "loss": 0.11840394735336304,
    "grad_norm": 0.6711931824684143,
    "learning_rate": 0.00018545132638734935,
    "epoch": 0.17654436119239797,
    "step": 5490
  },
  {
    "loss": 0.11159731149673462,
    "grad_norm": 0.7318800687789917,
    "learning_rate": 0.00018539863753810898,
    "epoch": 0.17686593562079944,
    "step": 5500
  },
  {
    "loss": 0.09772720336914062,
    "grad_norm": 0.5916475057601929,
    "learning_rate": 0.00018534586096630325,
    "epoch": 0.17718751004920089,
    "step": 5510
  },
  {
    "loss": 0.11480109691619873,
    "grad_norm": 0.5805314183235168,
    "learning_rate": 0.00018529299672614497,
    "epoch": 0.17750908447760233,
    "step": 5520
  },
  {
    "loss": 0.12061715126037598,
    "grad_norm": 0.8949525952339172,
    "learning_rate": 0.00018524004487193696,
    "epoch": 0.1778306589060038,
    "step": 5530
  },
  {
    "loss": 0.12319254875183105,
    "grad_norm": 0.7165164351463318,
    "learning_rate": 0.000185187005458072,
    "epoch": 0.17815223333440525,
    "step": 5540
  },
  {
    "loss": 0.09097777605056763,
    "grad_norm": 0.3125317692756653,
    "learning_rate": 0.00018513387853903287,
    "epoch": 0.1784738077628067,
    "step": 5550
  },
  {
    "loss": 0.12090181112289429,
    "grad_norm": 0.8263435363769531,
    "learning_rate": 0.0001850806641693922,
    "epoch": 0.17879538219120816,
    "step": 5560
  },
  {
    "loss": 0.09867244362831115,
    "grad_norm": 0.9423380494117737,
    "learning_rate": 0.00018502736240381251,
    "epoch": 0.1791169566196096,
    "step": 5570
  },
  {
    "loss": 0.0943190574645996,
    "grad_norm": 0.5803310871124268,
    "learning_rate": 0.000184973973297046,
    "epoch": 0.17943853104801105,
    "step": 5580
  },
  {
    "loss": 0.11495518684387207,
    "grad_norm": 0.9183040857315063,
    "learning_rate": 0.0001849204969039347,
    "epoch": 0.17976010547641252,
    "step": 5590
  },
  {
    "loss": 0.11546754837036133,
    "grad_norm": 0.6139799952507019,
    "learning_rate": 0.0001848669332794102,
    "epoch": 0.18008167990481397,
    "step": 5600
  },
  {
    "loss": 0.10671877861022949,
    "grad_norm": 0.5461157560348511,
    "learning_rate": 0.0001848132824784937,
    "epoch": 0.1804032543332154,
    "step": 5610
  },
  {
    "loss": 0.1120229721069336,
    "grad_norm": 0.9857230186462402,
    "learning_rate": 0.00018475954455629605,
    "epoch": 0.18072482876161688,
    "step": 5620
  },
  {
    "loss": 0.10375741720199586,
    "grad_norm": 0.7333449721336365,
    "learning_rate": 0.0001847057195680175,
    "epoch": 0.18104640319001833,
    "step": 5630
  },
  {
    "loss": 0.14358291625976563,
    "grad_norm": 0.8412182331085205,
    "learning_rate": 0.00018465180756894775,
    "epoch": 0.18136797761841977,
    "step": 5640
  },
  {
    "loss": 0.09533635377883912,
    "grad_norm": 1.1052337884902954,
    "learning_rate": 0.0001845978086144659,
    "epoch": 0.18168955204682125,
    "step": 5650
  },
  {
    "loss": 0.09434363842010499,
    "grad_norm": 0.8389447331428528,
    "learning_rate": 0.0001845437227600404,
    "epoch": 0.1820111264752227,
    "step": 5660
  },
  {
    "loss": 0.11739598512649536,
    "grad_norm": 0.46825969219207764,
    "learning_rate": 0.00018448955006122885,
    "epoch": 0.18233270090362413,
    "step": 5670
  },
  {
    "loss": 0.10412952899932862,
    "grad_norm": 0.725148618221283,
    "learning_rate": 0.00018443529057367817,
    "epoch": 0.1826542753320256,
    "step": 5680
  },
  {
    "loss": 0.11758382320404052,
    "grad_norm": 0.6468080878257751,
    "learning_rate": 0.00018438094435312445,
    "epoch": 0.18297584976042705,
    "step": 5690
  },
  {
    "loss": 0.09282848238945007,
    "grad_norm": 0.8106268644332886,
    "learning_rate": 0.00018432651145539272,
    "epoch": 0.1832974241888285,
    "step": 5700
  },
  {
    "loss": 0.09028403759002686,
    "grad_norm": 0.5799915790557861,
    "learning_rate": 0.0001842719919363972,
    "epoch": 0.18361899861722997,
    "step": 5710
  },
  {
    "loss": 0.09892613887786865,
    "grad_norm": 0.7924032211303711,
    "learning_rate": 0.00018421738585214104,
    "epoch": 0.1839405730456314,
    "step": 5720
  },
  {
    "loss": 0.1065057396888733,
    "grad_norm": 0.7232165932655334,
    "learning_rate": 0.00018416269325871628,
    "epoch": 0.18426214747403286,
    "step": 5730
  },
  {
    "loss": 0.10866960287094116,
    "grad_norm": 0.753584086894989,
    "learning_rate": 0.00018410791421230385,
    "epoch": 0.18458372190243433,
    "step": 5740
  },
  {
    "loss": 0.11250456571578979,
    "grad_norm": 0.6191846132278442,
    "learning_rate": 0.00018405304876917348,
    "epoch": 0.18490529633083577,
    "step": 5750
  },
  {
    "loss": 0.08145703673362732,
    "grad_norm": 0.817142903804779,
    "learning_rate": 0.0001839980969856837,
    "epoch": 0.18522687075923722,
    "step": 5760
  },
  {
    "loss": 0.10531235933303833,
    "grad_norm": 0.5584696531295776,
    "learning_rate": 0.0001839430589182816,
    "epoch": 0.1855484451876387,
    "step": 5770
  },
  {
    "loss": 0.10580016374588012,
    "grad_norm": 0.3925810754299164,
    "learning_rate": 0.000183887934623503,
    "epoch": 0.18587001961604013,
    "step": 5780
  },
  {
    "loss": 0.09895262718200684,
    "grad_norm": 0.5467655658721924,
    "learning_rate": 0.00018383272415797234,
    "epoch": 0.18619159404444158,
    "step": 5790
  },
  {
    "loss": 0.13267139196395875,
    "grad_norm": 1.0078212022781372,
    "learning_rate": 0.00018377742757840244,
    "epoch": 0.18651316847284305,
    "step": 5800
  },
  {
    "loss": 0.11166650056838989,
    "grad_norm": 0.737362802028656,
    "learning_rate": 0.00018372204494159465,
    "epoch": 0.1868347429012445,
    "step": 5810
  },
  {
    "loss": 0.10281800031661988,
    "grad_norm": 1.0692893266677856,
    "learning_rate": 0.00018366657630443882,
    "epoch": 0.18715631732964594,
    "step": 5820
  },
  {
    "loss": 0.0977216124534607,
    "grad_norm": 0.6687707304954529,
    "learning_rate": 0.00018361102172391293,
    "epoch": 0.1874778917580474,
    "step": 5830
  },
  {
    "loss": 0.10584867000579834,
    "grad_norm": 0.7733855843544006,
    "learning_rate": 0.00018355538125708334,
    "epoch": 0.18779946618644885,
    "step": 5840
  },
  {
    "loss": 0.09958687424659729,
    "grad_norm": 0.7798100113868713,
    "learning_rate": 0.00018349965496110472,
    "epoch": 0.1881210406148503,
    "step": 5850
  },
  {
    "loss": 0.10751545429229736,
    "grad_norm": 0.6445668339729309,
    "learning_rate": 0.00018344384289321978,
    "epoch": 0.18844261504325177,
    "step": 5860
  },
  {
    "loss": 0.09561465978622437,
    "grad_norm": 0.7363864183425903,
    "learning_rate": 0.00018338794511075943,
    "epoch": 0.18876418947165322,
    "step": 5870
  },
  {
    "loss": 0.11222798824310302,
    "grad_norm": 0.9353572726249695,
    "learning_rate": 0.00018333196167114255,
    "epoch": 0.18908576390005466,
    "step": 5880
  },
  {
    "loss": 0.09195201396942139,
    "grad_norm": 0.7091829180717468,
    "learning_rate": 0.00018327589263187602,
    "epoch": 0.18940733832845613,
    "step": 5890
  },
  {
    "loss": 0.07509946227073669,
    "grad_norm": 0.6560493111610413,
    "learning_rate": 0.00018321973805055474,
    "epoch": 0.18972891275685758,
    "step": 5900
  },
  {
    "loss": 0.09939819574356079,
    "grad_norm": 0.44442832469940186,
    "learning_rate": 0.00018316349798486132,
    "epoch": 0.19005048718525902,
    "step": 5910
  },
  {
    "loss": 0.09033665060997009,
    "grad_norm": 0.7245857119560242,
    "learning_rate": 0.00018310717249256633,
    "epoch": 0.1903720616136605,
    "step": 5920
  },
  {
    "loss": 0.13576501607894897,
    "grad_norm": 0.8852182030677795,
    "learning_rate": 0.00018305076163152798,
    "epoch": 0.19069363604206194,
    "step": 5930
  },
  {
    "loss": 0.0938784658908844,
    "grad_norm": 0.6078718900680542,
    "learning_rate": 0.0001829942654596923,
    "epoch": 0.19101521047046338,
    "step": 5940
  },
  {
    "loss": 0.11368472576141357,
    "grad_norm": 0.5211871862411499,
    "learning_rate": 0.00018293768403509282,
    "epoch": 0.19133678489886485,
    "step": 5950
  },
  {
    "loss": 0.08751682639122009,
    "grad_norm": 0.5084543824195862,
    "learning_rate": 0.00018288101741585072,
    "epoch": 0.1916583593272663,
    "step": 5960
  },
  {
    "loss": 0.12171189785003662,
    "grad_norm": 0.9775353074073792,
    "learning_rate": 0.00018282426566017463,
    "epoch": 0.19197993375566774,
    "step": 5970
  },
  {
    "loss": 0.09461681842803955,
    "grad_norm": 0.6826309561729431,
    "learning_rate": 0.00018276742882636073,
    "epoch": 0.19230150818406921,
    "step": 5980
  },
  {
    "loss": 0.12977845668792726,
    "grad_norm": 0.9831529855728149,
    "learning_rate": 0.0001827105069727925,
    "epoch": 0.19262308261247066,
    "step": 5990
  },
  {
    "loss": 0.11812064647674561,
    "grad_norm": 0.4797765910625458,
    "learning_rate": 0.00018265350015794082,
    "epoch": 0.1929446570408721,
    "step": 6000
  },
  {
    "loss": 0.08487967252731324,
    "grad_norm": 0.6206749081611633,
    "learning_rate": 0.00018259640844036382,
    "epoch": 0.19326623146927357,
    "step": 6010
  },
  {
    "loss": 0.10489945411682129,
    "grad_norm": 0.8495069146156311,
    "learning_rate": 0.00018253923187870678,
    "epoch": 0.19358780589767502,
    "step": 6020
  },
  {
    "loss": 0.09415996074676514,
    "grad_norm": 0.512051522731781,
    "learning_rate": 0.00018248197053170225,
    "epoch": 0.19390938032607646,
    "step": 6030
  },
  {
    "loss": 0.10890600681304932,
    "grad_norm": 0.696121096611023,
    "learning_rate": 0.00018242462445816976,
    "epoch": 0.19423095475447794,
    "step": 6040
  },
  {
    "loss": 0.12515428066253662,
    "grad_norm": 0.6889166235923767,
    "learning_rate": 0.000182367193717016,
    "epoch": 0.19455252918287938,
    "step": 6050
  },
  {
    "loss": 0.12075965404510498,
    "grad_norm": 0.6734789609909058,
    "learning_rate": 0.0001823096783672345,
    "epoch": 0.19487410361128082,
    "step": 6060
  },
  {
    "loss": 0.10391931533813477,
    "grad_norm": 1.2357184886932373,
    "learning_rate": 0.00018225207846790572,
    "epoch": 0.1951956780396823,
    "step": 6070
  },
  {
    "loss": 0.12936931848526,
    "grad_norm": 0.5635527968406677,
    "learning_rate": 0.00018219439407819713,
    "epoch": 0.19551725246808374,
    "step": 6080
  },
  {
    "loss": 0.10976414680480957,
    "grad_norm": 0.9435555338859558,
    "learning_rate": 0.00018213662525736274,
    "epoch": 0.19583882689648519,
    "step": 6090
  },
  {
    "loss": 0.09746204018592834,
    "grad_norm": 0.4754216969013214,
    "learning_rate": 0.0001820787720647435,
    "epoch": 0.19616040132488666,
    "step": 6100
  },
  {
    "loss": 0.10500679016113282,
    "grad_norm": 0.4885876476764679,
    "learning_rate": 0.00018202083455976694,
    "epoch": 0.1964819757532881,
    "step": 6110
  },
  {
    "loss": 0.09986729621887207,
    "grad_norm": 0.674632728099823,
    "learning_rate": 0.00018196281280194714,
    "epoch": 0.19680355018168955,
    "step": 6120
  },
  {
    "loss": 0.10967297554016113,
    "grad_norm": 0.794371485710144,
    "learning_rate": 0.00018190470685088483,
    "epoch": 0.19712512461009102,
    "step": 6130
  },
  {
    "loss": 0.0897097647190094,
    "grad_norm": 0.5588971376419067,
    "learning_rate": 0.0001818465167662672,
    "epoch": 0.19744669903849246,
    "step": 6140
  },
  {
    "loss": 0.10592899322509766,
    "grad_norm": 1.0722829103469849,
    "learning_rate": 0.00018178824260786782,
    "epoch": 0.1977682734668939,
    "step": 6150
  },
  {
    "loss": 0.10921781063079834,
    "grad_norm": 0.8634899854660034,
    "learning_rate": 0.0001817298844355466,
    "epoch": 0.19808984789529538,
    "step": 6160
  },
  {
    "loss": 0.08706367015838623,
    "grad_norm": 0.7577278017997742,
    "learning_rate": 0.0001816714423092499,
    "epoch": 0.19841142232369682,
    "step": 6170
  },
  {
    "loss": 0.10604641437530518,
    "grad_norm": 0.6018843054771423,
    "learning_rate": 0.00018161291628901013,
    "epoch": 0.19873299675209827,
    "step": 6180
  },
  {
    "loss": 0.1325182557106018,
    "grad_norm": 0.8310794830322266,
    "learning_rate": 0.000181554306434946,
    "epoch": 0.19905457118049974,
    "step": 6190
  },
  {
    "loss": 0.09288270473480224,
    "grad_norm": 0.479806512594223,
    "learning_rate": 0.0001814956128072623,
    "epoch": 0.19937614560890118,
    "step": 6200
  },
  {
    "loss": 0.11677355766296386,
    "grad_norm": 0.6561117768287659,
    "learning_rate": 0.0001814368354662498,
    "epoch": 0.19969772003730263,
    "step": 6210
  },
  {
    "eval_loss": 0.09828303754329681,
    "eval_runtime": 34.4157,
    "eval_samples_per_second": 146.038,
    "eval_steps_per_second": 36.524,
    "epoch": 0.1998906646943435,
    "step": 6216
  },
  {
    "loss": 0.10979170799255371,
    "grad_norm": 0.6072927713394165,
    "learning_rate": 0.00018137797447228545,
    "epoch": 0.2000192944657041,
    "step": 6220
  },
  {
    "loss": 0.07831120491027832,
    "grad_norm": 0.46518734097480774,
    "learning_rate": 0.00018131902988583192,
    "epoch": 0.20034086889410554,
    "step": 6230
  },
  {
    "loss": 0.0935763657093048,
    "grad_norm": 0.878280520439148,
    "learning_rate": 0.00018126000176743785,
    "epoch": 0.200662443322507,
    "step": 6240
  },
  {
    "loss": 0.08065707087516785,
    "grad_norm": 1.1618974208831787,
    "learning_rate": 0.0001812008901777377,
    "epoch": 0.20098401775090846,
    "step": 6250
  },
  {
    "loss": 0.10877575874328613,
    "grad_norm": 0.9426342844963074,
    "learning_rate": 0.00018114169517745163,
    "epoch": 0.2013055921793099,
    "step": 6260
  },
  {
    "loss": 0.09039924144744874,
    "grad_norm": 0.5820074677467346,
    "learning_rate": 0.0001810824168273855,
    "epoch": 0.20162716660771135,
    "step": 6270
  },
  {
    "loss": 0.10094892978668213,
    "grad_norm": 0.999288022518158,
    "learning_rate": 0.0001810230551884308,
    "epoch": 0.20194874103611282,
    "step": 6280
  },
  {
    "loss": 0.06847935914993286,
    "grad_norm": 0.4741264283657074,
    "learning_rate": 0.00018096361032156452,
    "epoch": 0.20227031546451427,
    "step": 6290
  },
  {
    "loss": 0.08855631351470947,
    "grad_norm": 0.5086546540260315,
    "learning_rate": 0.00018090408228784927,
    "epoch": 0.2025918898929157,
    "step": 6300
  },
  {
    "loss": 0.08249475955963134,
    "grad_norm": 0.7000566124916077,
    "learning_rate": 0.00018084447114843287,
    "epoch": 0.20291346432131716,
    "step": 6310
  },
  {
    "loss": 0.10006040334701538,
    "grad_norm": 0.7028230428695679,
    "learning_rate": 0.0001807847769645487,
    "epoch": 0.20323503874971863,
    "step": 6320
  },
  {
    "loss": 0.10323059558868408,
    "grad_norm": 0.9211930632591248,
    "learning_rate": 0.00018072499979751542,
    "epoch": 0.20355661317812007,
    "step": 6330
  },
  {
    "loss": 0.10563849210739136,
    "grad_norm": 0.547076940536499,
    "learning_rate": 0.0001806651397087369,
    "epoch": 0.20387818760652152,
    "step": 6340
  },
  {
    "loss": 0.10014694929122925,
    "grad_norm": 0.5817078351974487,
    "learning_rate": 0.00018060519675970214,
    "epoch": 0.204199762034923,
    "step": 6350
  },
  {
    "loss": 0.1062752366065979,
    "grad_norm": 1.991796851158142,
    "learning_rate": 0.00018054517101198528,
    "epoch": 0.20452133646332443,
    "step": 6360
  },
  {
    "loss": 0.10059846639633178,
    "grad_norm": 0.5084323883056641,
    "learning_rate": 0.00018048506252724557,
    "epoch": 0.20484291089172588,
    "step": 6370
  },
  {
    "loss": 0.06880200505256653,
    "grad_norm": 0.3965473771095276,
    "learning_rate": 0.0001804248713672272,
    "epoch": 0.20516448532012735,
    "step": 6380
  },
  {
    "loss": 0.0917474627494812,
    "grad_norm": 0.7427871823310852,
    "learning_rate": 0.0001803645975937593,
    "epoch": 0.2054860597485288,
    "step": 6390
  },
  {
    "loss": 0.10400476455688476,
    "grad_norm": 0.862004280090332,
    "learning_rate": 0.00018030424126875584,
    "epoch": 0.20580763417693024,
    "step": 6400
  },
  {
    "loss": 0.11660934686660766,
    "grad_norm": 0.8293169140815735,
    "learning_rate": 0.00018024380245421564,
    "epoch": 0.2061292086053317,
    "step": 6410
  },
  {
    "loss": 0.10720772743225097,
    "grad_norm": 0.5888752341270447,
    "learning_rate": 0.00018018328121222217,
    "epoch": 0.20645078303373315,
    "step": 6420
  },
  {
    "loss": 0.11179831027984619,
    "grad_norm": 0.6469850540161133,
    "learning_rate": 0.00018012267760494363,
    "epoch": 0.2067723574621346,
    "step": 6430
  },
  {
    "loss": 0.09034927487373352,
    "grad_norm": 0.4641270935535431,
    "learning_rate": 0.00018006199169463285,
    "epoch": 0.20709393189053607,
    "step": 6440
  },
  {
    "loss": 0.09115891456604004,
    "grad_norm": 0.683894693851471,
    "learning_rate": 0.0001800012235436271,
    "epoch": 0.20741550631893751,
    "step": 6450
  },
  {
    "loss": 0.10708479881286621,
    "grad_norm": 0.8873001337051392,
    "learning_rate": 0.0001799403732143483,
    "epoch": 0.20773708074733896,
    "step": 6460
  },
  {
    "loss": 0.08460823893547058,
    "grad_norm": 0.35685697197914124,
    "learning_rate": 0.0001798794407693026,
    "epoch": 0.20805865517574043,
    "step": 6470
  },
  {
    "loss": 0.12977834939956664,
    "grad_norm": 1.287055253982544,
    "learning_rate": 0.0001798184262710806,
    "epoch": 0.20838022960414188,
    "step": 6480
  },
  {
    "loss": 0.127907657623291,
    "grad_norm": 0.3801760673522949,
    "learning_rate": 0.00017975732978235716,
    "epoch": 0.20870180403254332,
    "step": 6490
  },
  {
    "loss": 0.11531537771224976,
    "grad_norm": 1.2351274490356445,
    "learning_rate": 0.00017969615136589141,
    "epoch": 0.2090233784609448,
    "step": 6500
  },
  {
    "loss": 0.10003373622894288,
    "grad_norm": 0.43053725361824036,
    "learning_rate": 0.00017963489108452656,
    "epoch": 0.20934495288934624,
    "step": 6510
  },
  {
    "loss": 0.11411433219909668,
    "grad_norm": 0.7325254678726196,
    "learning_rate": 0.00017957354900118992,
    "epoch": 0.20966652731774768,
    "step": 6520
  },
  {
    "loss": 0.10711169242858887,
    "grad_norm": 1.153060793876648,
    "learning_rate": 0.00017951212517889294,
    "epoch": 0.20998810174614915,
    "step": 6530
  },
  {
    "loss": 0.13314961194992064,
    "grad_norm": 0.879439115524292,
    "learning_rate": 0.00017945061968073088,
    "epoch": 0.2103096761745506,
    "step": 6540
  },
  {
    "loss": 0.10021822452545166,
    "grad_norm": 1.6467922925949097,
    "learning_rate": 0.000179389032569883,
    "epoch": 0.21063125060295204,
    "step": 6550
  },
  {
    "loss": 0.10448052883148193,
    "grad_norm": 0.8538579940795898,
    "learning_rate": 0.00017932736390961234,
    "epoch": 0.2109528250313535,
    "step": 6560
  },
  {
    "loss": 0.10208958387374878,
    "grad_norm": 0.7283279299736023,
    "learning_rate": 0.00017926561376326578,
    "epoch": 0.21127439945975496,
    "step": 6570
  },
  {
    "loss": 0.10935897827148437,
    "grad_norm": 1.3195074796676636,
    "learning_rate": 0.00017920378219427377,
    "epoch": 0.2115959738881564,
    "step": 6580
  },
  {
    "loss": 0.12804192304611206,
    "grad_norm": 1.063864827156067,
    "learning_rate": 0.00017914186926615058,
    "epoch": 0.21191754831655787,
    "step": 6590
  },
  {
    "loss": 0.10533335208892822,
    "grad_norm": 0.8524525165557861,
    "learning_rate": 0.00017907987504249388,
    "epoch": 0.21223912274495932,
    "step": 6600
  },
  {
    "loss": 0.07631253004074097,
    "grad_norm": 0.6262599229812622,
    "learning_rate": 0.00017901779958698499,
    "epoch": 0.21256069717336076,
    "step": 6610
  },
  {
    "loss": 0.08477730751037597,
    "grad_norm": 0.8015865683555603,
    "learning_rate": 0.00017895564296338857,
    "epoch": 0.21288227160176224,
    "step": 6620
  },
  {
    "loss": 0.0852261483669281,
    "grad_norm": 0.7002324461936951,
    "learning_rate": 0.00017889340523555267,
    "epoch": 0.21320384603016368,
    "step": 6630
  },
  {
    "loss": 0.12231900691986083,
    "grad_norm": 0.36369815468788147,
    "learning_rate": 0.00017883108646740868,
    "epoch": 0.21352542045856512,
    "step": 6640
  },
  {
    "loss": 0.11484540700912475,
    "grad_norm": 0.6854612827301025,
    "learning_rate": 0.0001787686867229713,
    "epoch": 0.2138469948869666,
    "step": 6650
  },
  {
    "loss": 0.11369551420211792,
    "grad_norm": 0.6591107845306396,
    "learning_rate": 0.00017870620606633825,
    "epoch": 0.21416856931536804,
    "step": 6660
  },
  {
    "loss": 0.12295722961425781,
    "grad_norm": 0.40087050199508667,
    "learning_rate": 0.00017864364456169054,
    "epoch": 0.21449014374376948,
    "step": 6670
  },
  {
    "loss": 0.10497041940689086,
    "grad_norm": 0.7661556601524353,
    "learning_rate": 0.00017858100227329207,
    "epoch": 0.21481171817217096,
    "step": 6680
  },
  {
    "loss": 0.1099507451057434,
    "grad_norm": 0.7987531423568726,
    "learning_rate": 0.00017851827926548982,
    "epoch": 0.2151332926005724,
    "step": 6690
  },
  {
    "loss": 0.10177501440048217,
    "grad_norm": 0.8020938634872437,
    "learning_rate": 0.00017845547560271365,
    "epoch": 0.21545486702897385,
    "step": 6700
  },
  {
    "loss": 0.09196376204490661,
    "grad_norm": 0.4774525761604309,
    "learning_rate": 0.0001783925913494763,
    "epoch": 0.21577644145737532,
    "step": 6710
  },
  {
    "loss": 0.11327618360519409,
    "grad_norm": 0.4038717746734619,
    "learning_rate": 0.00017832962657037327,
    "epoch": 0.21609801588577676,
    "step": 6720
  },
  {
    "loss": 0.10533188581466675,
    "grad_norm": 1.034292459487915,
    "learning_rate": 0.0001782665813300828,
    "epoch": 0.2164195903141782,
    "step": 6730
  },
  {
    "loss": 0.1124314546585083,
    "grad_norm": 0.9090122580528259,
    "learning_rate": 0.0001782034556933657,
    "epoch": 0.21674116474257968,
    "step": 6740
  },
  {
    "loss": 0.13018221855163575,
    "grad_norm": 1.0018212795257568,
    "learning_rate": 0.00017814024972506544,
    "epoch": 0.21706273917098112,
    "step": 6750
  },
  {
    "loss": 0.12027702331542969,
    "grad_norm": 0.7981282472610474,
    "learning_rate": 0.00017807696349010808,
    "epoch": 0.21738431359938257,
    "step": 6760
  },
  {
    "loss": 0.08453170061111451,
    "grad_norm": 0.5748383402824402,
    "learning_rate": 0.00017801359705350197,
    "epoch": 0.21770588802778404,
    "step": 6770
  },
  {
    "loss": 0.09876747727394104,
    "grad_norm": 0.7399095892906189,
    "learning_rate": 0.0001779501504803379,
    "epoch": 0.21802746245618548,
    "step": 6780
  },
  {
    "loss": 0.10025395154953003,
    "grad_norm": 0.6438655853271484,
    "learning_rate": 0.00017788662383578905,
    "epoch": 0.21834903688458693,
    "step": 6790
  },
  {
    "loss": 0.11686220169067382,
    "grad_norm": 0.4488239884376526,
    "learning_rate": 0.00017782301718511075,
    "epoch": 0.2186706113129884,
    "step": 6800
  },
  {
    "loss": 0.11589747667312622,
    "grad_norm": 1.082696557044983,
    "learning_rate": 0.00017775933059364061,
    "epoch": 0.21899218574138984,
    "step": 6810
  },
  {
    "loss": 0.08020769357681275,
    "grad_norm": 0.6219139099121094,
    "learning_rate": 0.0001776955641267983,
    "epoch": 0.2193137601697913,
    "step": 6820
  },
  {
    "loss": 0.1270586371421814,
    "grad_norm": 0.6280378103256226,
    "learning_rate": 0.00017763171785008545,
    "epoch": 0.21963533459819276,
    "step": 6830
  },
  {
    "loss": 0.09877665042877197,
    "grad_norm": 0.6493573784828186,
    "learning_rate": 0.0001775677918290859,
    "epoch": 0.2199569090265942,
    "step": 6840
  },
  {
    "loss": 0.08704429864883423,
    "grad_norm": 0.637754499912262,
    "learning_rate": 0.00017750378612946518,
    "epoch": 0.22027848345499565,
    "step": 6850
  },
  {
    "loss": 0.10899631977081299,
    "grad_norm": 0.6064416766166687,
    "learning_rate": 0.0001774397008169708,
    "epoch": 0.22060005788339712,
    "step": 6860
  },
  {
    "loss": 0.09086751341819763,
    "grad_norm": 0.6109260320663452,
    "learning_rate": 0.00017737553595743201,
    "epoch": 0.22092163231179857,
    "step": 6870
  },
  {
    "loss": 0.11162760257720947,
    "grad_norm": 0.8633526563644409,
    "learning_rate": 0.00017731129161675972,
    "epoch": 0.2212432067402,
    "step": 6880
  },
  {
    "loss": 0.10341193675994872,
    "grad_norm": 0.6742216944694519,
    "learning_rate": 0.00017724696786094662,
    "epoch": 0.22156478116860148,
    "step": 6890
  },
  {
    "loss": 0.08535621166229249,
    "grad_norm": 0.6198155283927917,
    "learning_rate": 0.0001771825647560668,
    "epoch": 0.22188635559700293,
    "step": 6900
  },
  {
    "loss": 0.10448358058929444,
    "grad_norm": 0.7120717763900757,
    "learning_rate": 0.00017711808236827597,
    "epoch": 0.22220793002540437,
    "step": 6910
  },
  {
    "loss": 0.08835526704788207,
    "grad_norm": 0.6049490571022034,
    "learning_rate": 0.00017705352076381127,
    "epoch": 0.22252950445380584,
    "step": 6920
  },
  {
    "loss": 0.0722814679145813,
    "grad_norm": 0.3987206518650055,
    "learning_rate": 0.00017698888000899124,
    "epoch": 0.2228510788822073,
    "step": 6930
  },
  {
    "loss": 0.10768955945968628,
    "grad_norm": 0.5643488764762878,
    "learning_rate": 0.00017692416017021562,
    "epoch": 0.22317265331060873,
    "step": 6940
  },
  {
    "loss": 0.12390257120132446,
    "grad_norm": 0.39716973900794983,
    "learning_rate": 0.00017685936131396548,
    "epoch": 0.2234942277390102,
    "step": 6950
  },
  {
    "loss": 0.08557771444320679,
    "grad_norm": 0.38591131567955017,
    "learning_rate": 0.00017679448350680304,
    "epoch": 0.22381580216741165,
    "step": 6960
  },
  {
    "loss": 0.08185508251190185,
    "grad_norm": 0.9277451634407043,
    "learning_rate": 0.00017672952681537155,
    "epoch": 0.2241373765958131,
    "step": 6970
  },
  {
    "loss": 0.11774744987487792,
    "grad_norm": 0.5688632726669312,
    "learning_rate": 0.00017666449130639543,
    "epoch": 0.22445895102421456,
    "step": 6980
  },
  {
    "loss": 0.10373042821884156,
    "grad_norm": 0.4881923496723175,
    "learning_rate": 0.00017659937704667992,
    "epoch": 0.224780525452616,
    "step": 6990
  },
  {
    "loss": 0.1025202989578247,
    "grad_norm": 0.761814296245575,
    "learning_rate": 0.00017653418410311126,
    "epoch": 0.22510209988101745,
    "step": 7000
  },
  {
    "loss": 0.10153430700302124,
    "grad_norm": 1.0617280006408691,
    "learning_rate": 0.00017646891254265645,
    "epoch": 0.22542367430941893,
    "step": 7010
  },
  {
    "loss": 0.08137912154197693,
    "grad_norm": 0.5417892932891846,
    "learning_rate": 0.00017640356243236327,
    "epoch": 0.22574524873782037,
    "step": 7020
  },
  {
    "loss": 0.11546515226364136,
    "grad_norm": 0.5282878875732422,
    "learning_rate": 0.00017633813383936015,
    "epoch": 0.22606682316622181,
    "step": 7030
  },
  {
    "loss": 0.1052852988243103,
    "grad_norm": 0.6157312989234924,
    "learning_rate": 0.00017627262683085618,
    "epoch": 0.2263883975946233,
    "step": 7040
  },
  {
    "loss": 0.10838897228240967,
    "grad_norm": 0.6011908650398254,
    "learning_rate": 0.00017620704147414107,
    "epoch": 0.22670997202302473,
    "step": 7050
  },
  {
    "loss": 0.0947287917137146,
    "grad_norm": 0.5182114243507385,
    "learning_rate": 0.00017614137783658482,
    "epoch": 0.22703154645142617,
    "step": 7060
  },
  {
    "loss": 0.09444833993911743,
    "grad_norm": 0.5272310376167297,
    "learning_rate": 0.000176075635985638,
    "epoch": 0.22735312087982765,
    "step": 7070
  },
  {
    "loss": 0.09998944401741028,
    "grad_norm": 0.702633798122406,
    "learning_rate": 0.00017600981598883152,
    "epoch": 0.2276746953082291,
    "step": 7080
  },
  {
    "loss": 0.07797504663467407,
    "grad_norm": 0.9731748700141907,
    "learning_rate": 0.00017594391791377647,
    "epoch": 0.22799626973663054,
    "step": 7090
  },
  {
    "loss": 0.09055452346801758,
    "grad_norm": 0.6219237446784973,
    "learning_rate": 0.00017587794182816416,
    "epoch": 0.228317844165032,
    "step": 7100
  },
  {
    "loss": 0.10952922105789184,
    "grad_norm": 0.6282539963722229,
    "learning_rate": 0.00017581188779976613,
    "epoch": 0.22863941859343345,
    "step": 7110
  },
  {
    "loss": 0.10705915689468384,
    "grad_norm": 1.1422780752182007,
    "learning_rate": 0.00017574575589643384,
    "epoch": 0.2289609930218349,
    "step": 7120
  },
  {
    "loss": 0.09305276870727539,
    "grad_norm": 0.6948840618133545,
    "learning_rate": 0.00017567954618609894,
    "epoch": 0.22928256745023637,
    "step": 7130
  },
  {
    "loss": 0.09609219431877136,
    "grad_norm": 0.5128272175788879,
    "learning_rate": 0.00017561325873677276,
    "epoch": 0.2296041418786378,
    "step": 7140
  },
  {
    "loss": 0.10106382369995118,
    "grad_norm": 0.9541149139404297,
    "learning_rate": 0.00017554689361654668,
    "epoch": 0.22992571630703926,
    "step": 7150
  },
  {
    "loss": 0.11649857759475708,
    "grad_norm": 0.8821794986724854,
    "learning_rate": 0.00017548045089359177,
    "epoch": 0.23024729073544073,
    "step": 7160
  },
  {
    "loss": 0.09836030006408691,
    "grad_norm": 0.839900016784668,
    "learning_rate": 0.00017541393063615884,
    "epoch": 0.23056886516384217,
    "step": 7170
  },
  {
    "loss": 0.09720966219902039,
    "grad_norm": 0.6266375780105591,
    "learning_rate": 0.00017534733291257838,
    "epoch": 0.23089043959224362,
    "step": 7180
  },
  {
    "loss": 0.10686802864074707,
    "grad_norm": 0.9709400534629822,
    "learning_rate": 0.00017528065779126033,
    "epoch": 0.2312120140206451,
    "step": 7190
  },
  {
    "loss": 0.0964643657207489,
    "grad_norm": 0.6434114575386047,
    "learning_rate": 0.0001752139053406943,
    "epoch": 0.23153358844904653,
    "step": 7200
  },
  {
    "loss": 0.09592649936676026,
    "grad_norm": 0.6001937389373779,
    "learning_rate": 0.00017514707562944927,
    "epoch": 0.23185516287744798,
    "step": 7210
  },
  {
    "loss": 0.09946632385253906,
    "grad_norm": 0.5942351222038269,
    "learning_rate": 0.0001750801687261735,
    "epoch": 0.23217673730584945,
    "step": 7220
  },
  {
    "loss": 0.11563431024551392,
    "grad_norm": 0.9182334542274475,
    "learning_rate": 0.00017501318469959463,
    "epoch": 0.2324983117342509,
    "step": 7230
  },
  {
    "loss": 0.10812370777130127,
    "grad_norm": 0.5348995327949524,
    "learning_rate": 0.00017494612361851957,
    "epoch": 0.23281988616265234,
    "step": 7240
  },
  {
    "loss": 0.10731531381607055,
    "grad_norm": 0.696471095085144,
    "learning_rate": 0.0001748789855518342,
    "epoch": 0.2331414605910538,
    "step": 7250
  },
  {
    "loss": 0.09992768168449402,
    "grad_norm": 1.2543549537658691,
    "learning_rate": 0.00017481177056850368,
    "epoch": 0.23346303501945526,
    "step": 7260
  },
  {
    "loss": 0.0912170648574829,
    "grad_norm": 1.0720438957214355,
    "learning_rate": 0.00017474447873757208,
    "epoch": 0.2337846094478567,
    "step": 7270
  },
  {
    "loss": 0.10393401384353637,
    "grad_norm": 0.7423423528671265,
    "learning_rate": 0.0001746771101281624,
    "epoch": 0.23410618387625817,
    "step": 7280
  },
  {
    "loss": 0.11489673852920532,
    "grad_norm": 0.9315942525863647,
    "learning_rate": 0.00017460966480947656,
    "epoch": 0.23442775830465962,
    "step": 7290
  },
  {
    "loss": 0.08871862292289734,
    "grad_norm": 0.6180339455604553,
    "learning_rate": 0.0001745421428507952,
    "epoch": 0.23474933273306106,
    "step": 7300
  },
  {
    "loss": 0.08185545206069947,
    "grad_norm": 0.3975384533405304,
    "learning_rate": 0.00017447454432147782,
    "epoch": 0.23507090716146253,
    "step": 7310
  },
  {
    "loss": 0.0861358404159546,
    "grad_norm": 0.49309736490249634,
    "learning_rate": 0.00017440686929096242,
    "epoch": 0.23539248158986398,
    "step": 7320
  },
  {
    "loss": 0.09481869339942932,
    "grad_norm": 0.8725064992904663,
    "learning_rate": 0.00017433911782876565,
    "epoch": 0.23571405601826542,
    "step": 7330
  },
  {
    "loss": 0.09213900566101074,
    "grad_norm": 0.7655817270278931,
    "learning_rate": 0.00017427129000448268,
    "epoch": 0.2360356304466669,
    "step": 7340
  },
  {
    "loss": 0.09312846064567566,
    "grad_norm": 0.8202439546585083,
    "learning_rate": 0.00017420338588778714,
    "epoch": 0.23635720487506834,
    "step": 7350
  },
  {
    "loss": 0.09380109310150146,
    "grad_norm": 0.8451393842697144,
    "learning_rate": 0.00017413540554843098,
    "epoch": 0.23667877930346978,
    "step": 7360
  },
  {
    "loss": 0.10455808639526368,
    "grad_norm": 0.7913851141929626,
    "learning_rate": 0.00017406734905624447,
    "epoch": 0.23700035373187125,
    "step": 7370
  },
  {
    "loss": 0.11372662782669067,
    "grad_norm": 0.5296722054481506,
    "learning_rate": 0.0001739992164811361,
    "epoch": 0.2373219281602727,
    "step": 7380
  },
  {
    "loss": 0.10835660696029663,
    "grad_norm": 0.6034450531005859,
    "learning_rate": 0.0001739310078930925,
    "epoch": 0.23764350258867414,
    "step": 7390
  },
  {
    "loss": 0.10947132110595703,
    "grad_norm": 1.676663875579834,
    "learning_rate": 0.00017386272336217845,
    "epoch": 0.23796507701707562,
    "step": 7400
  },
  {
    "loss": 0.09162078499794006,
    "grad_norm": 0.5721451044082642,
    "learning_rate": 0.00017379436295853664,
    "epoch": 0.23828665144547706,
    "step": 7410
  },
  {
    "loss": 0.1085892677307129,
    "grad_norm": 0.8040403723716736,
    "learning_rate": 0.00017372592675238776,
    "epoch": 0.2386082258738785,
    "step": 7420
  },
  {
    "loss": 0.09048799872398376,
    "grad_norm": 0.5279026031494141,
    "learning_rate": 0.00017365741481403037,
    "epoch": 0.23892980030227998,
    "step": 7430
  },
  {
    "loss": 0.1112370252609253,
    "grad_norm": 0.596389651298523,
    "learning_rate": 0.00017358882721384077,
    "epoch": 0.23925137473068142,
    "step": 7440
  },
  {
    "loss": 0.12354772090911866,
    "grad_norm": 0.4829653799533844,
    "learning_rate": 0.00017352016402227304,
    "epoch": 0.23957294915908287,
    "step": 7450
  },
  {
    "loss": 0.10008889436721802,
    "grad_norm": 0.8064790964126587,
    "learning_rate": 0.00017345142530985887,
    "epoch": 0.2398945235874843,
    "step": 7460
  },
  {
    "loss": 0.08722924590110778,
    "grad_norm": 0.5152531862258911,
    "learning_rate": 0.00017338261114720752,
    "epoch": 0.24021609801588578,
    "step": 7470
  },
  {
    "loss": 0.10029129981994629,
    "grad_norm": 0.5598156452178955,
    "learning_rate": 0.00017331372160500584,
    "epoch": 0.24053767244428723,
    "step": 7480
  },
  {
    "loss": 0.09828886985778809,
    "grad_norm": 0.30673664808273315,
    "learning_rate": 0.00017324475675401795,
    "epoch": 0.24085924687268867,
    "step": 7490
  },
  {
    "loss": 0.09425093531608582,
    "grad_norm": 2.1410884857177734,
    "learning_rate": 0.0001731757166650855,
    "epoch": 0.24118082130109014,
    "step": 7500
  },
  {
    "loss": 0.09232192039489746,
    "grad_norm": 0.9482953548431396,
    "learning_rate": 0.0001731066014091273,
    "epoch": 0.2415023957294916,
    "step": 7510
  },
  {
    "loss": 0.08345150351524352,
    "grad_norm": 0.6554622054100037,
    "learning_rate": 0.0001730374110571394,
    "epoch": 0.24182397015789303,
    "step": 7520
  },
  {
    "loss": 0.08806174993515015,
    "grad_norm": 0.5079277157783508,
    "learning_rate": 0.00017296814568019508,
    "epoch": 0.2421455445862945,
    "step": 7530
  },
  {
    "loss": 0.08180884122848511,
    "grad_norm": 0.5006304383277893,
    "learning_rate": 0.00017289880534944455,
    "epoch": 0.24246711901469595,
    "step": 7540
  },
  {
    "loss": 0.08461836576461793,
    "grad_norm": 0.6327130794525146,
    "learning_rate": 0.00017282939013611514,
    "epoch": 0.2427886934430974,
    "step": 7550
  },
  {
    "loss": 0.11851485967636108,
    "grad_norm": 0.6704835891723633,
    "learning_rate": 0.000172759900111511,
    "epoch": 0.24311026787149886,
    "step": 7560
  },
  {
    "loss": 0.09942713379859924,
    "grad_norm": 0.5339265465736389,
    "learning_rate": 0.00017269033534701313,
    "epoch": 0.2434318422999003,
    "step": 7570
  },
  {
    "loss": 0.12718113660812377,
    "grad_norm": 0.6755968928337097,
    "learning_rate": 0.00017262069591407938,
    "epoch": 0.24375341672830175,
    "step": 7580
  },
  {
    "loss": 0.08766931891441346,
    "grad_norm": 0.40743404626846313,
    "learning_rate": 0.0001725509818842443,
    "epoch": 0.24407499115670322,
    "step": 7590
  },
  {
    "loss": 0.08795983195304871,
    "grad_norm": 1.024290919303894,
    "learning_rate": 0.00017248119332911895,
    "epoch": 0.24439656558510467,
    "step": 7600
  },
  {
    "loss": 0.10301086902618409,
    "grad_norm": 0.6083598136901855,
    "learning_rate": 0.0001724113303203911,
    "epoch": 0.2447181400135061,
    "step": 7610
  },
  {
    "loss": 0.09267066121101379,
    "grad_norm": 0.47813767194747925,
    "learning_rate": 0.00017234139292982486,
    "epoch": 0.24503971444190759,
    "step": 7620
  },
  {
    "loss": 0.09036524295806884,
    "grad_norm": 1.0438446998596191,
    "learning_rate": 0.0001722713812292608,
    "epoch": 0.24536128887030903,
    "step": 7630
  },
  {
    "loss": 0.13027456998825074,
    "grad_norm": 0.6518002152442932,
    "learning_rate": 0.00017220129529061585,
    "epoch": 0.24568286329871047,
    "step": 7640
  },
  {
    "loss": 0.09606041312217713,
    "grad_norm": 0.9624584913253784,
    "learning_rate": 0.0001721311351858832,
    "epoch": 0.24600443772711195,
    "step": 7650
  },
  {
    "loss": 0.10024822950363159,
    "grad_norm": 0.7557567954063416,
    "learning_rate": 0.00017206090098713218,
    "epoch": 0.2463260121555134,
    "step": 7660
  },
  {
    "loss": 0.08903319239616395,
    "grad_norm": 1.0236475467681885,
    "learning_rate": 0.00017199059276650825,
    "epoch": 0.24664758658391484,
    "step": 7670
  },
  {
    "loss": 0.11563560962677003,
    "grad_norm": 1.1230127811431885,
    "learning_rate": 0.00017192021059623288,
    "epoch": 0.2469691610123163,
    "step": 7680
  },
  {
    "loss": 0.08626883029937744,
    "grad_norm": 0.8625466227531433,
    "learning_rate": 0.00017184975454860357,
    "epoch": 0.24729073544071775,
    "step": 7690
  },
  {
    "loss": 0.11050916910171509,
    "grad_norm": 0.9562323093414307,
    "learning_rate": 0.00017177922469599363,
    "epoch": 0.2476123098691192,
    "step": 7700
  },
  {
    "loss": 0.12541766166687013,
    "grad_norm": 0.9169332981109619,
    "learning_rate": 0.00017170862111085227,
    "epoch": 0.24793388429752067,
    "step": 7710
  },
  {
    "loss": 0.08499926924705506,
    "grad_norm": 0.7092348337173462,
    "learning_rate": 0.00017163794386570438,
    "epoch": 0.2482554587259221,
    "step": 7720
  },
  {
    "loss": 0.11418701410293579,
    "grad_norm": 0.5422489643096924,
    "learning_rate": 0.0001715671930331505,
    "epoch": 0.24857703315432356,
    "step": 7730
  },
  {
    "loss": 0.07331929206848145,
    "grad_norm": 0.717381477355957,
    "learning_rate": 0.0001714963686858668,
    "epoch": 0.24889860758272503,
    "step": 7740
  },
  {
    "loss": 0.09337686896324157,
    "grad_norm": 0.7024790644645691,
    "learning_rate": 0.0001714254708966049,
    "epoch": 0.24922018201112647,
    "step": 7750
  },
  {
    "loss": 0.12471973896026611,
    "grad_norm": 0.9360175132751465,
    "learning_rate": 0.00017135449973819196,
    "epoch": 0.24954175643952792,
    "step": 7760
  },
  {
    "loss": 0.0979580819606781,
    "grad_norm": 0.7486076951026917,
    "learning_rate": 0.00017128345528353046,
    "epoch": 0.2498633308679294,
    "step": 7770
  },
  {
    "eval_loss": 0.09749862551689148,
    "eval_runtime": 34.3673,
    "eval_samples_per_second": 146.244,
    "eval_steps_per_second": 36.575,
    "epoch": 0.2498633308679294,
    "step": 7770
  },
  {
    "loss": 0.12439451217651368,
    "grad_norm": 1.1453461647033691,
    "learning_rate": 0.00017121233760559817,
    "epoch": 0.25018490529633086,
    "step": 7780
  },
  {
    "loss": 0.09368115663528442,
    "grad_norm": 0.5301191210746765,
    "learning_rate": 0.000171141146777448,
    "epoch": 0.2505064797247323,
    "step": 7790
  },
  {
    "loss": 0.09548683166503906,
    "grad_norm": 1.0390958786010742,
    "learning_rate": 0.0001710698828722082,
    "epoch": 0.25082805415313375,
    "step": 7800
  },
  {
    "loss": 0.10714445114135743,
    "grad_norm": 0.6461021304130554,
    "learning_rate": 0.00017099854596308185,
    "epoch": 0.2511496285815352,
    "step": 7810
  },
  {
    "loss": 0.10078861713409423,
    "grad_norm": 0.9467542171478271,
    "learning_rate": 0.00017092713612334713,
    "epoch": 0.25147120300993664,
    "step": 7820
  },
  {
    "loss": 0.10617483854293823,
    "grad_norm": 0.6734544634819031,
    "learning_rate": 0.00017085565342635724,
    "epoch": 0.2517927774383381,
    "step": 7830
  },
  {
    "loss": 0.11229211091995239,
    "grad_norm": 0.7666560411453247,
    "learning_rate": 0.00017078409794554,
    "epoch": 0.2521143518667396,
    "step": 7840
  },
  {
    "loss": 0.08903430700302124,
    "grad_norm": 1.0733190774917603,
    "learning_rate": 0.00017071246975439817,
    "epoch": 0.252435926295141,
    "step": 7850
  },
  {
    "loss": 0.10291492938995361,
    "grad_norm": 0.6430104970932007,
    "learning_rate": 0.00017064076892650914,
    "epoch": 0.25275750072354247,
    "step": 7860
  },
  {
    "loss": 0.0859697163105011,
    "grad_norm": 0.7043799161911011,
    "learning_rate": 0.00017056899553552485,
    "epoch": 0.25307907515194394,
    "step": 7870
  },
  {
    "loss": 0.08448026180267335,
    "grad_norm": 1.1739349365234375,
    "learning_rate": 0.00017049714965517189,
    "epoch": 0.25340064958034536,
    "step": 7880
  },
  {
    "loss": 0.10753465890884399,
    "grad_norm": 0.6605959534645081,
    "learning_rate": 0.0001704252313592513,
    "epoch": 0.25372222400874683,
    "step": 7890
  },
  {
    "loss": 0.09561908841133118,
    "grad_norm": 0.505441427230835,
    "learning_rate": 0.00017035324072163833,
    "epoch": 0.2540437984371483,
    "step": 7900
  },
  {
    "loss": 0.12803629636764527,
    "grad_norm": 0.6273232698440552,
    "learning_rate": 0.00017028117781628282,
    "epoch": 0.2543653728655497,
    "step": 7910
  },
  {
    "loss": 0.08171590566635131,
    "grad_norm": 0.582779049873352,
    "learning_rate": 0.00017020904271720867,
    "epoch": 0.2546869472939512,
    "step": 7920
  },
  {
    "loss": 0.08273233771324158,
    "grad_norm": 0.4706325829029083,
    "learning_rate": 0.00017013683549851392,
    "epoch": 0.25500852172235267,
    "step": 7930
  },
  {
    "loss": 0.10103654861450195,
    "grad_norm": 0.6312887668609619,
    "learning_rate": 0.00017006455623437078,
    "epoch": 0.2553300961507541,
    "step": 7940
  },
  {
    "loss": 0.08203045129776002,
    "grad_norm": 0.4585205614566803,
    "learning_rate": 0.00016999220499902543,
    "epoch": 0.25565167057915555,
    "step": 7950
  },
  {
    "loss": 0.0920904517173767,
    "grad_norm": 0.48585036396980286,
    "learning_rate": 0.00016991978186679796,
    "epoch": 0.255973245007557,
    "step": 7960
  },
  {
    "loss": 0.08931515216827393,
    "grad_norm": 0.6253461837768555,
    "learning_rate": 0.00016984728691208232,
    "epoch": 0.25629481943595844,
    "step": 7970
  },
  {
    "loss": 0.09320071935653687,
    "grad_norm": 0.49154773354530334,
    "learning_rate": 0.0001697747202093463,
    "epoch": 0.2566163938643599,
    "step": 7980
  },
  {
    "loss": 0.10840188264846802,
    "grad_norm": 0.4476393163204193,
    "learning_rate": 0.00016970208183313126,
    "epoch": 0.2569379682927614,
    "step": 7990
  },
  {
    "loss": 0.09145287275314332,
    "grad_norm": 0.7409923076629639,
    "learning_rate": 0.00016962937185805235,
    "epoch": 0.2572595427211628,
    "step": 8000
  },
  {
    "loss": 0.10165148973464966,
    "grad_norm": 0.8097208738327026,
    "learning_rate": 0.00016955659035879813,
    "epoch": 0.2575811171495643,
    "step": 8010
  },
  {
    "loss": 0.11168540716171264,
    "grad_norm": 0.9544158577919006,
    "learning_rate": 0.0001694837374101307,
    "epoch": 0.2579026915779657,
    "step": 8020
  },
  {
    "loss": 0.10602585077285767,
    "grad_norm": 0.6628233194351196,
    "learning_rate": 0.00016941081308688555,
    "epoch": 0.25822426600636716,
    "step": 8030
  },
  {
    "loss": 0.08922889828681946,
    "grad_norm": 0.8359578847885132,
    "learning_rate": 0.00016933781746397136,
    "epoch": 0.25854584043476864,
    "step": 8040
  },
  {
    "loss": 0.0837334394454956,
    "grad_norm": 0.8613901138305664,
    "learning_rate": 0.0001692647506163703,
    "epoch": 0.25886741486317005,
    "step": 8050
  },
  {
    "loss": 0.0981156051158905,
    "grad_norm": 0.6106037497520447,
    "learning_rate": 0.0001691916126191375,
    "epoch": 0.2591889892915715,
    "step": 8060
  },
  {
    "loss": 0.09956081509590149,
    "grad_norm": 0.8334420323371887,
    "learning_rate": 0.00016911840354740128,
    "epoch": 0.259510563719973,
    "step": 8070
  },
  {
    "loss": 0.10414061546325684,
    "grad_norm": 0.7758110761642456,
    "learning_rate": 0.0001690451234763629,
    "epoch": 0.2598321381483744,
    "step": 8080
  },
  {
    "loss": 0.0904352068901062,
    "grad_norm": 0.4925134479999542,
    "learning_rate": 0.00016897177248129653,
    "epoch": 0.2601537125767759,
    "step": 8090
  },
  {
    "loss": 0.08173861503601074,
    "grad_norm": 0.5777755975723267,
    "learning_rate": 0.0001688983506375493,
    "epoch": 0.26047528700517736,
    "step": 8100
  },
  {
    "loss": 0.07746504545211792,
    "grad_norm": 0.4453226327896118,
    "learning_rate": 0.0001688248580205411,
    "epoch": 0.2607968614335788,
    "step": 8110
  },
  {
    "loss": 0.09968498945236207,
    "grad_norm": 0.5532230734825134,
    "learning_rate": 0.00016875129470576437,
    "epoch": 0.26111843586198025,
    "step": 8120
  },
  {
    "loss": 0.09109488129615784,
    "grad_norm": 0.44665494561195374,
    "learning_rate": 0.0001686776607687844,
    "epoch": 0.2614400102903817,
    "step": 8130
  },
  {
    "loss": 0.10301159620285034,
    "grad_norm": 0.9509112238883972,
    "learning_rate": 0.00016860395628523885,
    "epoch": 0.26176158471878314,
    "step": 8140
  },
  {
    "loss": 0.1160635232925415,
    "grad_norm": 0.46450453996658325,
    "learning_rate": 0.0001685301813308379,
    "epoch": 0.2620831591471846,
    "step": 8150
  },
  {
    "loss": 0.10146403312683105,
    "grad_norm": 0.7742125391960144,
    "learning_rate": 0.00016845633598136408,
    "epoch": 0.2624047335755861,
    "step": 8160
  },
  {
    "loss": 0.11969466209411621,
    "grad_norm": 0.43207746744155884,
    "learning_rate": 0.00016838242031267235,
    "epoch": 0.2627263080039875,
    "step": 8170
  },
  {
    "loss": 0.10110535621643066,
    "grad_norm": 0.6822372078895569,
    "learning_rate": 0.00016830843440068978,
    "epoch": 0.26304788243238897,
    "step": 8180
  },
  {
    "loss": 0.08360062837600708,
    "grad_norm": 0.8064790964126587,
    "learning_rate": 0.00016823437832141563,
    "epoch": 0.26336945686079044,
    "step": 8190
  },
  {
    "loss": 0.09945321083068848,
    "grad_norm": 0.7815895676612854,
    "learning_rate": 0.00016816025215092128,
    "epoch": 0.26369103128919186,
    "step": 8200
  },
  {
    "loss": 0.09642932415008545,
    "grad_norm": 0.7080429196357727,
    "learning_rate": 0.00016808605596535005,
    "epoch": 0.26401260571759333,
    "step": 8210
  },
  {
    "loss": 0.08843590021133423,
    "grad_norm": 0.6929242610931396,
    "learning_rate": 0.0001680117898409172,
    "epoch": 0.2643341801459948,
    "step": 8220
  },
  {
    "loss": 0.07515991926193237,
    "grad_norm": 0.5778882503509521,
    "learning_rate": 0.00016793745385390986,
    "epoch": 0.2646557545743962,
    "step": 8230
  },
  {
    "loss": 0.13268963098526002,
    "grad_norm": 1.0547934770584106,
    "learning_rate": 0.0001678630480806869,
    "epoch": 0.2649773290027977,
    "step": 8240
  },
  {
    "loss": 0.08606036305427552,
    "grad_norm": 0.35389772057533264,
    "learning_rate": 0.00016778857259767884,
    "epoch": 0.26529890343119916,
    "step": 8250
  },
  {
    "loss": 0.08920258283615112,
    "grad_norm": 1.022499680519104,
    "learning_rate": 0.00016771402748138784,
    "epoch": 0.2656204778596006,
    "step": 8260
  },
  {
    "loss": 0.09268574118614196,
    "grad_norm": 0.842609703540802,
    "learning_rate": 0.00016763941280838765,
    "epoch": 0.26594205228800205,
    "step": 8270
  },
  {
    "loss": 0.09981530904769897,
    "grad_norm": 0.9523076415061951,
    "learning_rate": 0.00016756472865532337,
    "epoch": 0.2662636267164035,
    "step": 8280
  },
  {
    "loss": 0.1299252986907959,
    "grad_norm": 0.909056544303894,
    "learning_rate": 0.00016748997509891152,
    "epoch": 0.26658520114480494,
    "step": 8290
  },
  {
    "loss": 0.09837405681610108,
    "grad_norm": 0.9995813965797424,
    "learning_rate": 0.00016741515221593988,
    "epoch": 0.2669067755732064,
    "step": 8300
  },
  {
    "loss": 0.11543529033660889,
    "grad_norm": 0.43808504939079285,
    "learning_rate": 0.0001673402600832675,
    "epoch": 0.2672283500016079,
    "step": 8310
  },
  {
    "loss": 0.10366925001144409,
    "grad_norm": 0.8612508773803711,
    "learning_rate": 0.00016726529877782452,
    "epoch": 0.2675499244300093,
    "step": 8320
  },
  {
    "loss": 0.10360457897186279,
    "grad_norm": 1.0201222896575928,
    "learning_rate": 0.00016719026837661214,
    "epoch": 0.2678714988584108,
    "step": 8330
  },
  {
    "loss": 0.0959369421005249,
    "grad_norm": 0.5185285210609436,
    "learning_rate": 0.0001671151689567025,
    "epoch": 0.26819307328681224,
    "step": 8340
  },
  {
    "loss": 0.11198742389678955,
    "grad_norm": 0.9601693153381348,
    "learning_rate": 0.00016704000059523876,
    "epoch": 0.26851464771521366,
    "step": 8350
  },
  {
    "loss": 0.10278092622756958,
    "grad_norm": 0.7881720662117004,
    "learning_rate": 0.00016696476336943477,
    "epoch": 0.26883622214361513,
    "step": 8360
  },
  {
    "loss": 0.10845927000045777,
    "grad_norm": 0.5752276182174683,
    "learning_rate": 0.00016688945735657515,
    "epoch": 0.2691577965720166,
    "step": 8370
  },
  {
    "loss": 0.1280617117881775,
    "grad_norm": 0.5794875025749207,
    "learning_rate": 0.00016681408263401524,
    "epoch": 0.269479371000418,
    "step": 8380
  },
  {
    "loss": 0.0788794457912445,
    "grad_norm": 0.33127912878990173,
    "learning_rate": 0.00016673863927918085,
    "epoch": 0.2698009454288195,
    "step": 8390
  },
  {
    "loss": 0.11556386947631836,
    "grad_norm": 0.7478880286216736,
    "learning_rate": 0.00016666312736956842,
    "epoch": 0.27012251985722097,
    "step": 8400
  },
  {
    "loss": 0.10363576412200928,
    "grad_norm": 0.6624906063079834,
    "learning_rate": 0.00016658754698274468,
    "epoch": 0.2704440942856224,
    "step": 8410
  },
  {
    "loss": 0.10713640451431275,
    "grad_norm": 0.7080273628234863,
    "learning_rate": 0.0001665118981963468,
    "epoch": 0.27076566871402385,
    "step": 8420
  },
  {
    "loss": 0.11110186576843262,
    "grad_norm": 0.5911359190940857,
    "learning_rate": 0.00016643618108808217,
    "epoch": 0.2710872431424253,
    "step": 8430
  },
  {
    "loss": 0.07568821907043458,
    "grad_norm": 0.6160790324211121,
    "learning_rate": 0.00016636039573572837,
    "epoch": 0.27140881757082674,
    "step": 8440
  },
  {
    "loss": 0.0935560405254364,
    "grad_norm": 0.47352269291877747,
    "learning_rate": 0.00016628454221713306,
    "epoch": 0.2717303919992282,
    "step": 8450
  },
  {
    "loss": 0.10653687715530395,
    "grad_norm": 0.8653551340103149,
    "learning_rate": 0.00016620862061021393,
    "epoch": 0.2720519664276297,
    "step": 8460
  },
  {
    "loss": 0.10586884021759033,
    "grad_norm": 0.7118867039680481,
    "learning_rate": 0.0001661326309929587,
    "epoch": 0.2723735408560311,
    "step": 8470
  },
  {
    "loss": 0.10795516967773437,
    "grad_norm": 0.7776938676834106,
    "learning_rate": 0.00016605657344342473,
    "epoch": 0.2726951152844326,
    "step": 8480
  },
  {
    "loss": 0.08810576796531677,
    "grad_norm": 0.6349517703056335,
    "learning_rate": 0.00016598044803973943,
    "epoch": 0.27301668971283405,
    "step": 8490
  },
  {
    "loss": 0.09347708225250244,
    "grad_norm": 0.2226419895887375,
    "learning_rate": 0.00016590425486009972,
    "epoch": 0.27333826414123547,
    "step": 8500
  },
  {
    "loss": 0.08297247290611268,
    "grad_norm": 0.24822402000427246,
    "learning_rate": 0.00016582799398277219,
    "epoch": 0.27365983856963694,
    "step": 8510
  },
  {
    "loss": 0.0796057939529419,
    "grad_norm": 0.34090086817741394,
    "learning_rate": 0.00016575166548609302,
    "epoch": 0.2739814129980384,
    "step": 8520
  },
  {
    "loss": 0.10786666870117187,
    "grad_norm": 0.7712501287460327,
    "learning_rate": 0.00016567526944846777,
    "epoch": 0.2743029874264398,
    "step": 8530
  },
  {
    "loss": 0.10834369659423829,
    "grad_norm": 0.7854555249214172,
    "learning_rate": 0.00016559880594837147,
    "epoch": 0.2746245618548413,
    "step": 8540
  },
  {
    "loss": 0.10700218677520752,
    "grad_norm": 0.5751836895942688,
    "learning_rate": 0.00016552227506434837,
    "epoch": 0.27494613628324277,
    "step": 8550
  },
  {
    "loss": 0.09334764480590821,
    "grad_norm": 0.5211126804351807,
    "learning_rate": 0.00016544567687501197,
    "epoch": 0.2752677107116442,
    "step": 8560
  },
  {
    "loss": 0.1230047106742859,
    "grad_norm": 0.35446053743362427,
    "learning_rate": 0.0001653690114590449,
    "epoch": 0.27558928514004566,
    "step": 8570
  },
  {
    "loss": 0.104975426197052,
    "grad_norm": 0.4927564561367035,
    "learning_rate": 0.00016529227889519886,
    "epoch": 0.27591085956844713,
    "step": 8580
  },
  {
    "loss": 0.11820746660232544,
    "grad_norm": 0.551878035068512,
    "learning_rate": 0.0001652154792622945,
    "epoch": 0.27623243399684855,
    "step": 8590
  },
  {
    "loss": 0.09666290879249573,
    "grad_norm": 0.8831762671470642,
    "learning_rate": 0.0001651386126392214,
    "epoch": 0.27655400842525,
    "step": 8600
  },
  {
    "loss": 0.08981671929359436,
    "grad_norm": 0.34058356285095215,
    "learning_rate": 0.00016506167910493788,
    "epoch": 0.2768755828536515,
    "step": 8610
  },
  {
    "loss": 0.13624900579452515,
    "grad_norm": 2.133625030517578,
    "learning_rate": 0.0001649846787384711,
    "epoch": 0.2771971572820529,
    "step": 8620
  },
  {
    "loss": 0.10119202136993408,
    "grad_norm": 0.6735418438911438,
    "learning_rate": 0.0001649076116189168,
    "epoch": 0.2775187317104544,
    "step": 8630
  },
  {
    "loss": 0.10100538730621338,
    "grad_norm": 0.5642951130867004,
    "learning_rate": 0.0001648304778254393,
    "epoch": 0.27784030613885585,
    "step": 8640
  },
  {
    "loss": 0.0916166365146637,
    "grad_norm": 0.40952157974243164,
    "learning_rate": 0.0001647532774372714,
    "epoch": 0.27816188056725727,
    "step": 8650
  },
  {
    "loss": 0.09529081583023072,
    "grad_norm": 0.6564289331436157,
    "learning_rate": 0.00016467601053371436,
    "epoch": 0.27848345499565874,
    "step": 8660
  },
  {
    "loss": 0.08425755500793457,
    "grad_norm": 0.46509331464767456,
    "learning_rate": 0.0001645986771941377,
    "epoch": 0.2788050294240602,
    "step": 8670
  },
  {
    "loss": 0.10253841876983642,
    "grad_norm": 0.8329104781150818,
    "learning_rate": 0.00016452127749797915,
    "epoch": 0.27912660385246163,
    "step": 8680
  },
  {
    "loss": 0.08099549412727355,
    "grad_norm": 0.502581000328064,
    "learning_rate": 0.00016444381152474477,
    "epoch": 0.2794481782808631,
    "step": 8690
  },
  {
    "loss": 0.11632890701293945,
    "grad_norm": 0.6065788865089417,
    "learning_rate": 0.0001643662793540085,
    "epoch": 0.2797697527092646,
    "step": 8700
  },
  {
    "loss": 0.0974295198917389,
    "grad_norm": 0.5808644890785217,
    "learning_rate": 0.0001642886810654124,
    "epoch": 0.280091327137666,
    "step": 8710
  },
  {
    "loss": 0.0918147623538971,
    "grad_norm": 0.35146135091781616,
    "learning_rate": 0.0001642110167386665,
    "epoch": 0.28041290156606746,
    "step": 8720
  },
  {
    "loss": 0.11327881813049316,
    "grad_norm": 0.8766664266586304,
    "learning_rate": 0.00016413328645354844,
    "epoch": 0.28073447599446893,
    "step": 8730
  },
  {
    "loss": 0.09906228184700012,
    "grad_norm": 0.639585554599762,
    "learning_rate": 0.00016405549028990386,
    "epoch": 0.28105605042287035,
    "step": 8740
  },
  {
    "loss": 0.08947083353996277,
    "grad_norm": 0.5350400805473328,
    "learning_rate": 0.00016397762832764593,
    "epoch": 0.2813776248512718,
    "step": 8750
  },
  {
    "loss": 0.10893043279647827,
    "grad_norm": 0.7001042366027832,
    "learning_rate": 0.00016389970064675544,
    "epoch": 0.2816991992796733,
    "step": 8760
  },
  {
    "loss": 0.10575506687164307,
    "grad_norm": 0.70212721824646,
    "learning_rate": 0.0001638217073272807,
    "epoch": 0.2820207737080747,
    "step": 8770
  },
  {
    "loss": 0.1042213797569275,
    "grad_norm": 0.9001713395118713,
    "learning_rate": 0.00016374364844933746,
    "epoch": 0.2823423481364762,
    "step": 8780
  },
  {
    "loss": 0.08165057897567748,
    "grad_norm": 0.8453369140625,
    "learning_rate": 0.00016366552409310874,
    "epoch": 0.28266392256487766,
    "step": 8790
  },
  {
    "loss": 0.09870893955230713,
    "grad_norm": 0.24046075344085693,
    "learning_rate": 0.00016358733433884492,
    "epoch": 0.2829854969932791,
    "step": 8800
  },
  {
    "loss": 0.09446768164634704,
    "grad_norm": 1.1681368350982666,
    "learning_rate": 0.0001635090792668635,
    "epoch": 0.28330707142168055,
    "step": 8810
  },
  {
    "loss": 0.0984427809715271,
    "grad_norm": 0.8611108064651489,
    "learning_rate": 0.0001634307589575491,
    "epoch": 0.283628645850082,
    "step": 8820
  },
  {
    "loss": 0.14263057708740234,
    "grad_norm": 0.9652822017669678,
    "learning_rate": 0.00016335237349135326,
    "epoch": 0.28395022027848343,
    "step": 8830
  },
  {
    "loss": 0.09047924280166626,
    "grad_norm": 1.2011903524398804,
    "learning_rate": 0.0001632739229487946,
    "epoch": 0.2842717947068849,
    "step": 8840
  },
  {
    "loss": 0.07878822684288025,
    "grad_norm": 0.4895530045032501,
    "learning_rate": 0.00016319540741045852,
    "epoch": 0.2845933691352864,
    "step": 8850
  },
  {
    "loss": 0.10191992521286011,
    "grad_norm": 0.9160004258155823,
    "learning_rate": 0.00016311682695699714,
    "epoch": 0.2849149435636878,
    "step": 8860
  },
  {
    "loss": 0.14895331859588623,
    "grad_norm": 0.8889757990837097,
    "learning_rate": 0.00016303818166912932,
    "epoch": 0.28523651799208927,
    "step": 8870
  },
  {
    "loss": 0.09080453515052796,
    "grad_norm": 0.9305357933044434,
    "learning_rate": 0.0001629594716276405,
    "epoch": 0.28555809242049074,
    "step": 8880
  },
  {
    "loss": 0.11647429466247558,
    "grad_norm": 0.41978469491004944,
    "learning_rate": 0.00016288069691338257,
    "epoch": 0.28587966684889216,
    "step": 8890
  },
  {
    "loss": 0.11093926429748535,
    "grad_norm": 0.5302769541740417,
    "learning_rate": 0.00016280185760727403,
    "epoch": 0.2862012412772936,
    "step": 8900
  },
  {
    "loss": 0.10094484090805053,
    "grad_norm": 1.0867546796798706,
    "learning_rate": 0.00016272295379029953,
    "epoch": 0.2865228157056951,
    "step": 8910
  },
  {
    "loss": 0.10699278116226196,
    "grad_norm": 0.5287032723426819,
    "learning_rate": 0.00016264398554351016,
    "epoch": 0.2868443901340965,
    "step": 8920
  },
  {
    "loss": 0.10200968980789185,
    "grad_norm": 0.7823114991188049,
    "learning_rate": 0.00016256495294802298,
    "epoch": 0.287165964562498,
    "step": 8930
  },
  {
    "loss": 0.07948703169822693,
    "grad_norm": 0.5769001245498657,
    "learning_rate": 0.0001624858560850214,
    "epoch": 0.28748753899089946,
    "step": 8940
  },
  {
    "loss": 0.11868566274642944,
    "grad_norm": 0.8702729940414429,
    "learning_rate": 0.00016240669503575462,
    "epoch": 0.2878091134193009,
    "step": 8950
  },
  {
    "loss": 0.12385741472244263,
    "grad_norm": 0.432796835899353,
    "learning_rate": 0.00016232746988153793,
    "epoch": 0.28813068784770235,
    "step": 8960
  },
  {
    "loss": 0.09350829720497131,
    "grad_norm": 0.5241557955741882,
    "learning_rate": 0.00016224818070375245,
    "epoch": 0.2884522622761038,
    "step": 8970
  },
  {
    "loss": 0.10613198280334472,
    "grad_norm": 0.6393463015556335,
    "learning_rate": 0.00016216882758384496,
    "epoch": 0.28877383670450524,
    "step": 8980
  },
  {
    "loss": 0.09328456521034241,
    "grad_norm": 0.9340083599090576,
    "learning_rate": 0.000162089410603328,
    "epoch": 0.2890954111329067,
    "step": 8990
  },
  {
    "loss": 0.09708858728408813,
    "grad_norm": 0.40460801124572754,
    "learning_rate": 0.00016200992984377976,
    "epoch": 0.2894169855613082,
    "step": 9000
  },
  {
    "loss": 0.11160033941268921,
    "grad_norm": 0.6721453666687012,
    "learning_rate": 0.00016193038538684385,
    "epoch": 0.2897385599897096,
    "step": 9010
  },
  {
    "loss": 0.07697651386260987,
    "grad_norm": 0.6095776557922363,
    "learning_rate": 0.00016185077731422932,
    "epoch": 0.29006013441811107,
    "step": 9020
  },
  {
    "loss": 0.08334786295890809,
    "grad_norm": 0.4550764858722687,
    "learning_rate": 0.00016177110570771063,
    "epoch": 0.29038170884651254,
    "step": 9030
  },
  {
    "loss": 0.09118367433547973,
    "grad_norm": 0.712659478187561,
    "learning_rate": 0.00016169137064912746,
    "epoch": 0.29070328327491396,
    "step": 9040
  },
  {
    "loss": 0.08763358592987061,
    "grad_norm": 0.4912260174751282,
    "learning_rate": 0.00016161157222038466,
    "epoch": 0.29102485770331543,
    "step": 9050
  },
  {
    "loss": 0.11886682510375976,
    "grad_norm": 0.635463297367096,
    "learning_rate": 0.0001615317105034522,
    "epoch": 0.2913464321317169,
    "step": 9060
  },
  {
    "loss": 0.09079389572143555,
    "grad_norm": 0.48032060265541077,
    "learning_rate": 0.0001614517855803651,
    "epoch": 0.2916680065601183,
    "step": 9070
  },
  {
    "loss": 0.09244561195373535,
    "grad_norm": 0.8139591813087463,
    "learning_rate": 0.00016137179753322316,
    "epoch": 0.2919895809885198,
    "step": 9080
  },
  {
    "loss": 0.08526058197021484,
    "grad_norm": 0.7081871628761292,
    "learning_rate": 0.00016129174644419116,
    "epoch": 0.29231115541692126,
    "step": 9090
  },
  {
    "loss": 0.09352965950965882,
    "grad_norm": 0.47736358642578125,
    "learning_rate": 0.00016121163239549862,
    "epoch": 0.2926327298453227,
    "step": 9100
  },
  {
    "loss": 0.11282715797424317,
    "grad_norm": 0.5685864686965942,
    "learning_rate": 0.00016113145546943968,
    "epoch": 0.29295430427372415,
    "step": 9110
  },
  {
    "loss": 0.12271944284439087,
    "grad_norm": 0.8204271793365479,
    "learning_rate": 0.0001610512157483731,
    "epoch": 0.2932758787021256,
    "step": 9120
  },
  {
    "loss": 0.08996185660362244,
    "grad_norm": 0.8587689995765686,
    "learning_rate": 0.00016097091331472216,
    "epoch": 0.29359745313052704,
    "step": 9130
  },
  {
    "loss": 0.09678568243980408,
    "grad_norm": 0.7826471328735352,
    "learning_rate": 0.00016089054825097453,
    "epoch": 0.2939190275589285,
    "step": 9140
  },
  {
    "loss": 0.09184351563453674,
    "grad_norm": 0.4400404691696167,
    "learning_rate": 0.0001608101206396822,
    "epoch": 0.29424060198733,
    "step": 9150
  },
  {
    "loss": 0.11043401956558227,
    "grad_norm": 0.9030609130859375,
    "learning_rate": 0.0001607296305634615,
    "epoch": 0.2945621764157314,
    "step": 9160
  },
  {
    "loss": 0.08099195957183838,
    "grad_norm": 0.7449926137924194,
    "learning_rate": 0.00016064907810499276,
    "epoch": 0.2948837508441329,
    "step": 9170
  },
  {
    "loss": 0.0982122004032135,
    "grad_norm": 0.8458951711654663,
    "learning_rate": 0.00016056846334702054,
    "epoch": 0.29520532527253435,
    "step": 9180
  },
  {
    "loss": 0.1208492398262024,
    "grad_norm": 0.6890580654144287,
    "learning_rate": 0.00016048778637235337,
    "epoch": 0.29552689970093576,
    "step": 9190
  },
  {
    "loss": 0.10449072122573852,
    "grad_norm": 0.7368547320365906,
    "learning_rate": 0.00016040704726386366,
    "epoch": 0.29584847412933724,
    "step": 9200
  },
  {
    "loss": 0.13114418983459472,
    "grad_norm": 0.5611347556114197,
    "learning_rate": 0.00016032624610448762,
    "epoch": 0.2961700485577387,
    "step": 9210
  },
  {
    "loss": 0.1002449631690979,
    "grad_norm": 0.31099745631217957,
    "learning_rate": 0.0001602453829772252,
    "epoch": 0.2964916229861401,
    "step": 9220
  },
  {
    "loss": 0.09459711909294129,
    "grad_norm": 0.5505207777023315,
    "learning_rate": 0.00016016445796514011,
    "epoch": 0.2968131974145416,
    "step": 9230
  },
  {
    "loss": 0.05700767636299133,
    "grad_norm": 0.2523616850376129,
    "learning_rate": 0.00016008347115135946,
    "epoch": 0.29713477184294307,
    "step": 9240
  },
  {
    "loss": 0.10036879777908325,
    "grad_norm": 0.5966441035270691,
    "learning_rate": 0.00016000242261907403,
    "epoch": 0.2974563462713445,
    "step": 9250
  },
  {
    "loss": 0.11783348321914673,
    "grad_norm": 0.8323610424995422,
    "learning_rate": 0.00015992131245153783,
    "epoch": 0.29777792069974596,
    "step": 9260
  },
  {
    "loss": 0.10106947422027587,
    "grad_norm": 0.7663300633430481,
    "learning_rate": 0.00015984014073206827,
    "epoch": 0.29809949512814743,
    "step": 9270
  },
  {
    "loss": 0.10415028333663941,
    "grad_norm": 0.8918988108634949,
    "learning_rate": 0.00015975890754404596,
    "epoch": 0.29842106955654885,
    "step": 9280
  },
  {
    "loss": 0.09814226031303405,
    "grad_norm": 0.30188941955566406,
    "learning_rate": 0.00015967761297091472,
    "epoch": 0.2987426439849503,
    "step": 9290
  },
  {
    "loss": 0.09462435245513916,
    "grad_norm": 0.6528173089027405,
    "learning_rate": 0.0001595962570961813,
    "epoch": 0.2990642184133518,
    "step": 9300
  },
  {
    "loss": 0.10978235006332397,
    "grad_norm": 0.45726704597473145,
    "learning_rate": 0.00015951484000341552,
    "epoch": 0.2993857928417532,
    "step": 9310
  },
  {
    "loss": 0.10773928165435791,
    "grad_norm": 0.39979204535484314,
    "learning_rate": 0.00015943336177625007,
    "epoch": 0.2997073672701547,
    "step": 9320
  },
  {
    "eval_loss": 0.0934017226099968,
    "eval_runtime": 34.8471,
    "eval_samples_per_second": 144.23,
    "eval_steps_per_second": 36.072,
    "epoch": 0.29983599704151526,
    "step": 9324
  },
  {
    "loss": 0.11000183820724488,
    "grad_norm": 0.6095293760299683,
    "learning_rate": 0.0001593518224983804,
    "epoch": 0.30002894169855615,
    "step": 9330
  },
  {
    "loss": 0.08486317992210388,
    "grad_norm": 0.4363544285297394,
    "learning_rate": 0.00015927022225356469,
    "epoch": 0.30035051612695757,
    "step": 9340
  },
  {
    "loss": 0.09942440390586853,
    "grad_norm": 0.6024127006530762,
    "learning_rate": 0.00015918856112562372,
    "epoch": 0.30067209055535904,
    "step": 9350
  },
  {
    "loss": 0.10907498598098755,
    "grad_norm": 0.5117012858390808,
    "learning_rate": 0.00015910683919844089,
    "epoch": 0.3009936649837605,
    "step": 9360
  },
  {
    "loss": 0.10354846715927124,
    "grad_norm": 0.48038336634635925,
    "learning_rate": 0.00015902505655596198,
    "epoch": 0.30131523941216193,
    "step": 9370
  },
  {
    "loss": 0.09846914410591126,
    "grad_norm": 0.7857918739318848,
    "learning_rate": 0.00015894321328219517,
    "epoch": 0.3016368138405634,
    "step": 9380
  },
  {
    "loss": 0.09118065237998962,
    "grad_norm": 0.8895276784896851,
    "learning_rate": 0.00015886130946121088,
    "epoch": 0.3019583882689649,
    "step": 9390
  },
  {
    "loss": 0.09130282402038574,
    "grad_norm": 0.5060708522796631,
    "learning_rate": 0.00015877934517714176,
    "epoch": 0.3022799626973663,
    "step": 9400
  },
  {
    "loss": 0.08922684788703919,
    "grad_norm": 0.6186339855194092,
    "learning_rate": 0.0001586973205141826,
    "epoch": 0.30260153712576776,
    "step": 9410
  },
  {
    "loss": 0.10332038402557372,
    "grad_norm": 0.7059121131896973,
    "learning_rate": 0.00015861523555659016,
    "epoch": 0.30292311155416923,
    "step": 9420
  },
  {
    "loss": 0.10528864860534667,
    "grad_norm": 0.4513157606124878,
    "learning_rate": 0.00015853309038868315,
    "epoch": 0.30324468598257065,
    "step": 9430
  },
  {
    "loss": 0.09138191342353821,
    "grad_norm": 0.7483622431755066,
    "learning_rate": 0.00015845088509484207,
    "epoch": 0.3035662604109721,
    "step": 9440
  },
  {
    "loss": 0.08996622562408448,
    "grad_norm": 1.0308533906936646,
    "learning_rate": 0.00015836861975950935,
    "epoch": 0.3038878348393736,
    "step": 9450
  },
  {
    "loss": 0.07570862770080566,
    "grad_norm": 0.5102515816688538,
    "learning_rate": 0.00015828629446718894,
    "epoch": 0.304209409267775,
    "step": 9460
  },
  {
    "loss": 0.08456374406814575,
    "grad_norm": 0.38768988847732544,
    "learning_rate": 0.00015820390930244637,
    "epoch": 0.3045309836961765,
    "step": 9470
  },
  {
    "loss": 0.1260838747024536,
    "grad_norm": 0.943644106388092,
    "learning_rate": 0.00015812146434990884,
    "epoch": 0.30485255812457795,
    "step": 9480
  },
  {
    "loss": 0.08907887935638428,
    "grad_norm": 0.4711124300956726,
    "learning_rate": 0.00015803895969426478,
    "epoch": 0.30517413255297937,
    "step": 9490
  },
  {
    "loss": 0.10089514255523682,
    "grad_norm": 0.7654880881309509,
    "learning_rate": 0.000157956395420264,
    "epoch": 0.30549570698138084,
    "step": 9500
  },
  {
    "loss": 0.07927495241165161,
    "grad_norm": 0.4468529224395752,
    "learning_rate": 0.00015787377161271762,
    "epoch": 0.3058172814097823,
    "step": 9510
  },
  {
    "loss": 0.09040161371231079,
    "grad_norm": 0.46876248717308044,
    "learning_rate": 0.00015779108835649783,
    "epoch": 0.30613885583818373,
    "step": 9520
  },
  {
    "loss": 0.09458811283111572,
    "grad_norm": 0.5352162718772888,
    "learning_rate": 0.00015770834573653795,
    "epoch": 0.3064604302665852,
    "step": 9530
  },
  {
    "loss": 0.0799996018409729,
    "grad_norm": 0.6501600742340088,
    "learning_rate": 0.00015762554383783224,
    "epoch": 0.3067820046949867,
    "step": 9540
  },
  {
    "loss": 0.11077144145965576,
    "grad_norm": 0.5619154572486877,
    "learning_rate": 0.00015754268274543586,
    "epoch": 0.3071035791233881,
    "step": 9550
  },
  {
    "loss": 0.10113047361373902,
    "grad_norm": 0.8997209072113037,
    "learning_rate": 0.00015745976254446478,
    "epoch": 0.30742515355178957,
    "step": 9560
  },
  {
    "loss": 0.09810148477554322,
    "grad_norm": 0.7278547286987305,
    "learning_rate": 0.0001573767833200957,
    "epoch": 0.30774672798019104,
    "step": 9570
  },
  {
    "loss": 0.11046998500823975,
    "grad_norm": 0.3747931122779846,
    "learning_rate": 0.0001572937451575659,
    "epoch": 0.30806830240859245,
    "step": 9580
  },
  {
    "loss": 0.12715048789978028,
    "grad_norm": 1.1526594161987305,
    "learning_rate": 0.0001572106481421732,
    "epoch": 0.3083898768369939,
    "step": 9590
  },
  {
    "loss": 0.10706145763397217,
    "grad_norm": 0.4139595031738281,
    "learning_rate": 0.00015712749235927603,
    "epoch": 0.3087114512653954,
    "step": 9600
  },
  {
    "loss": 0.11112467050552369,
    "grad_norm": 0.6514472961425781,
    "learning_rate": 0.00015704427789429298,
    "epoch": 0.3090330256937968,
    "step": 9610
  },
  {
    "loss": 0.08490209579467774,
    "grad_norm": 0.4603259563446045,
    "learning_rate": 0.000156961004832703,
    "epoch": 0.3093546001221983,
    "step": 9620
  },
  {
    "loss": 0.11466584205627442,
    "grad_norm": 0.8583003282546997,
    "learning_rate": 0.00015687767326004527,
    "epoch": 0.30967617455059976,
    "step": 9630
  },
  {
    "loss": 0.10502034425735474,
    "grad_norm": 0.6610942482948303,
    "learning_rate": 0.00015679428326191905,
    "epoch": 0.3099977489790012,
    "step": 9640
  },
  {
    "loss": 0.09413716197013855,
    "grad_norm": 0.6740992665290833,
    "learning_rate": 0.00015671083492398356,
    "epoch": 0.31031932340740265,
    "step": 9650
  },
  {
    "loss": 0.10678738355636597,
    "grad_norm": 0.4354815185070038,
    "learning_rate": 0.000156627328331958,
    "epoch": 0.3106408978358041,
    "step": 9660
  },
  {
    "loss": 0.08357634544372558,
    "grad_norm": 0.39228543639183044,
    "learning_rate": 0.00015654376357162147,
    "epoch": 0.31096247226420554,
    "step": 9670
  },
  {
    "loss": 0.09768832325935364,
    "grad_norm": 0.7303394675254822,
    "learning_rate": 0.00015646014072881267,
    "epoch": 0.311284046692607,
    "step": 9680
  },
  {
    "loss": 0.10822383165359498,
    "grad_norm": 0.5756319761276245,
    "learning_rate": 0.0001563764598894301,
    "epoch": 0.3116056211210085,
    "step": 9690
  },
  {
    "loss": 0.10973855257034301,
    "grad_norm": 0.553230345249176,
    "learning_rate": 0.00015629272113943174,
    "epoch": 0.3119271955494099,
    "step": 9700
  },
  {
    "loss": 0.09531382918357849,
    "grad_norm": 0.6640179753303528,
    "learning_rate": 0.00015620892456483514,
    "epoch": 0.31224876997781137,
    "step": 9710
  },
  {
    "loss": 0.08025224208831787,
    "grad_norm": 0.864226222038269,
    "learning_rate": 0.00015612507025171714,
    "epoch": 0.31257034440621284,
    "step": 9720
  },
  {
    "loss": 0.10060276985168456,
    "grad_norm": 0.9080818891525269,
    "learning_rate": 0.00015604115828621402,
    "epoch": 0.31289191883461426,
    "step": 9730
  },
  {
    "loss": 0.09277968406677246,
    "grad_norm": 0.4111309349536896,
    "learning_rate": 0.0001559571887545212,
    "epoch": 0.31321349326301573,
    "step": 9740
  },
  {
    "loss": 0.0899915099143982,
    "grad_norm": 0.769721508026123,
    "learning_rate": 0.00015587316174289325,
    "epoch": 0.3135350676914172,
    "step": 9750
  },
  {
    "loss": 0.10643835067749023,
    "grad_norm": 0.720486044883728,
    "learning_rate": 0.0001557890773376438,
    "epoch": 0.3138566421198186,
    "step": 9760
  },
  {
    "loss": 0.09196939468383789,
    "grad_norm": 0.5397993326187134,
    "learning_rate": 0.0001557049356251454,
    "epoch": 0.3141782165482201,
    "step": 9770
  },
  {
    "loss": 0.07454143166542053,
    "grad_norm": 0.6117302775382996,
    "learning_rate": 0.00015562073669182945,
    "epoch": 0.31449979097662156,
    "step": 9780
  },
  {
    "loss": 0.10379166603088379,
    "grad_norm": 0.8357234001159668,
    "learning_rate": 0.0001555364806241862,
    "epoch": 0.314821365405023,
    "step": 9790
  },
  {
    "loss": 0.11008110046386718,
    "grad_norm": 1.1281821727752686,
    "learning_rate": 0.00015545216750876458,
    "epoch": 0.31514293983342445,
    "step": 9800
  },
  {
    "loss": 0.09503305554389954,
    "grad_norm": 0.5694693326950073,
    "learning_rate": 0.00015536779743217206,
    "epoch": 0.3154645142618259,
    "step": 9810
  },
  {
    "loss": 0.10269997119903565,
    "grad_norm": 0.6259468197822571,
    "learning_rate": 0.00015528337048107465,
    "epoch": 0.31578608869022734,
    "step": 9820
  },
  {
    "loss": 0.07942078709602356,
    "grad_norm": 0.5972447395324707,
    "learning_rate": 0.0001551988867421968,
    "epoch": 0.3161076631186288,
    "step": 9830
  },
  {
    "loss": 0.09182611703872681,
    "grad_norm": 0.7243683338165283,
    "learning_rate": 0.00015511434630232127,
    "epoch": 0.3164292375470303,
    "step": 9840
  },
  {
    "loss": 0.08888370990753174,
    "grad_norm": 0.5140639543533325,
    "learning_rate": 0.00015502974924828906,
    "epoch": 0.3167508119754317,
    "step": 9850
  },
  {
    "loss": 0.08769227266311645,
    "grad_norm": 0.8622345924377441,
    "learning_rate": 0.0001549450956669994,
    "epoch": 0.3170723864038332,
    "step": 9860
  },
  {
    "loss": 0.09619449377059937,
    "grad_norm": 0.3897106647491455,
    "learning_rate": 0.00015486038564540948,
    "epoch": 0.31739396083223465,
    "step": 9870
  },
  {
    "loss": 0.10312601327896118,
    "grad_norm": 0.7995700240135193,
    "learning_rate": 0.00015477561927053446,
    "epoch": 0.31771553526063606,
    "step": 9880
  },
  {
    "loss": 0.11048386096954346,
    "grad_norm": 0.8452174663543701,
    "learning_rate": 0.00015469079662944757,
    "epoch": 0.31803710968903753,
    "step": 9890
  },
  {
    "loss": 0.06555310487747193,
    "grad_norm": 0.5714414715766907,
    "learning_rate": 0.00015460591780927957,
    "epoch": 0.318358684117439,
    "step": 9900
  },
  {
    "loss": 0.10164577960968017,
    "grad_norm": 0.5852811336517334,
    "learning_rate": 0.00015452098289721913,
    "epoch": 0.3186802585458404,
    "step": 9910
  },
  {
    "loss": 0.1025164246559143,
    "grad_norm": 0.8045811653137207,
    "learning_rate": 0.00015443599198051246,
    "epoch": 0.3190018329742419,
    "step": 9920
  },
  {
    "loss": 0.07474237084388732,
    "grad_norm": 0.5519168376922607,
    "learning_rate": 0.0001543509451464633,
    "epoch": 0.31932340740264337,
    "step": 9930
  },
  {
    "loss": 0.10572299957275391,
    "grad_norm": 0.666706919670105,
    "learning_rate": 0.00015426584248243286,
    "epoch": 0.3196449818310448,
    "step": 9940
  },
  {
    "loss": 0.11387902498245239,
    "grad_norm": 0.5598200559616089,
    "learning_rate": 0.00015418068407583967,
    "epoch": 0.31996655625944626,
    "step": 9950
  },
  {
    "loss": 0.1391116499900818,
    "grad_norm": 2.396148920059204,
    "learning_rate": 0.00015409547001415946,
    "epoch": 0.3202881306878477,
    "step": 9960
  },
  {
    "loss": 0.09753554463386535,
    "grad_norm": 0.7613327503204346,
    "learning_rate": 0.0001540102003849253,
    "epoch": 0.32060970511624914,
    "step": 9970
  },
  {
    "loss": 0.10358134508132935,
    "grad_norm": 0.5053406357765198,
    "learning_rate": 0.0001539248752757272,
    "epoch": 0.3209312795446506,
    "step": 9980
  },
  {
    "loss": 0.14178507328033446,
    "grad_norm": 0.7225109934806824,
    "learning_rate": 0.0001538394947742122,
    "epoch": 0.3212528539730521,
    "step": 9990
  },
  {
    "loss": 0.08253385424613953,
    "grad_norm": 0.5420551896095276,
    "learning_rate": 0.00015375405896808422,
    "epoch": 0.3215744284014535,
    "step": 10000
  },
  {
    "loss": 0.09111645221710205,
    "grad_norm": 0.6374110579490662,
    "learning_rate": 0.00015366856794510398,
    "epoch": 0.321896002829855,
    "step": 10010
  },
  {
    "loss": 0.12315458059310913,
    "grad_norm": 0.6631460189819336,
    "learning_rate": 0.00015358302179308898,
    "epoch": 0.32221757725825645,
    "step": 10020
  },
  {
    "loss": 0.11130249500274658,
    "grad_norm": 0.5429269075393677,
    "learning_rate": 0.00015349742059991332,
    "epoch": 0.32253915168665787,
    "step": 10030
  },
  {
    "loss": 0.06493273973464966,
    "grad_norm": 0.6440100073814392,
    "learning_rate": 0.00015341176445350763,
    "epoch": 0.32286072611505934,
    "step": 10040
  },
  {
    "loss": 0.10156418085098266,
    "grad_norm": 0.5670591592788696,
    "learning_rate": 0.00015332605344185897,
    "epoch": 0.3231823005434608,
    "step": 10050
  },
  {
    "loss": 0.09578874111175537,
    "grad_norm": 0.8024997711181641,
    "learning_rate": 0.0001532402876530108,
    "epoch": 0.3235038749718622,
    "step": 10060
  },
  {
    "loss": 0.0960733413696289,
    "grad_norm": 0.5642770528793335,
    "learning_rate": 0.0001531544671750628,
    "epoch": 0.3238254494002637,
    "step": 10070
  },
  {
    "loss": 0.12410423755645753,
    "grad_norm": 0.5657545924186707,
    "learning_rate": 0.0001530685920961709,
    "epoch": 0.32414702382866517,
    "step": 10080
  },
  {
    "loss": 0.0712913691997528,
    "grad_norm": 0.47271034121513367,
    "learning_rate": 0.00015298266250454704,
    "epoch": 0.3244685982570666,
    "step": 10090
  },
  {
    "loss": 0.094403076171875,
    "grad_norm": 0.5328975915908813,
    "learning_rate": 0.00015289667848845923,
    "epoch": 0.32479017268546806,
    "step": 10100
  },
  {
    "loss": 0.08306188583374023,
    "grad_norm": 0.6818568110466003,
    "learning_rate": 0.0001528106401362313,
    "epoch": 0.32511174711386953,
    "step": 10110
  },
  {
    "loss": 0.09016135931015015,
    "grad_norm": 0.6705556511878967,
    "learning_rate": 0.00015272454753624293,
    "epoch": 0.32543332154227095,
    "step": 10120
  },
  {
    "loss": 0.10912590026855469,
    "grad_norm": 0.4123816192150116,
    "learning_rate": 0.00015263840077692954,
    "epoch": 0.3257548959706724,
    "step": 10130
  },
  {
    "loss": 0.09481947422027588,
    "grad_norm": 0.4842923581600189,
    "learning_rate": 0.0001525521999467822,
    "epoch": 0.3260764703990739,
    "step": 10140
  },
  {
    "loss": 0.11109416484832764,
    "grad_norm": 0.6905588507652283,
    "learning_rate": 0.00015246594513434748,
    "epoch": 0.3263980448274753,
    "step": 10150
  },
  {
    "loss": 0.1307373523712158,
    "grad_norm": 0.7909269332885742,
    "learning_rate": 0.00015237963642822746,
    "epoch": 0.3267196192558768,
    "step": 10160
  },
  {
    "loss": 0.09989275336265564,
    "grad_norm": 0.6991463899612427,
    "learning_rate": 0.00015229327391707944,
    "epoch": 0.32704119368427825,
    "step": 10170
  },
  {
    "loss": 0.08000798225402832,
    "grad_norm": 0.6140600442886353,
    "learning_rate": 0.0001522068576896162,
    "epoch": 0.32736276811267967,
    "step": 10180
  },
  {
    "loss": 0.10280264616012573,
    "grad_norm": 0.9740838408470154,
    "learning_rate": 0.0001521203878346055,
    "epoch": 0.32768434254108114,
    "step": 10190
  },
  {
    "loss": 0.0873986840248108,
    "grad_norm": 0.2624066174030304,
    "learning_rate": 0.00015203386444087035,
    "epoch": 0.3280059169694826,
    "step": 10200
  },
  {
    "loss": 0.09654288291931153,
    "grad_norm": 0.6746272444725037,
    "learning_rate": 0.00015194728759728864,
    "epoch": 0.32832749139788403,
    "step": 10210
  },
  {
    "loss": 0.11106714010238647,
    "grad_norm": 0.7071442008018494,
    "learning_rate": 0.00015186065739279317,
    "epoch": 0.3286490658262855,
    "step": 10220
  },
  {
    "loss": 0.10884366035461426,
    "grad_norm": 0.6379486322402954,
    "learning_rate": 0.00015177397391637163,
    "epoch": 0.328970640254687,
    "step": 10230
  },
  {
    "loss": 0.11274509429931641,
    "grad_norm": 0.6626328825950623,
    "learning_rate": 0.00015168723725706638,
    "epoch": 0.3292922146830884,
    "step": 10240
  },
  {
    "loss": 0.10389626026153564,
    "grad_norm": 0.5899779796600342,
    "learning_rate": 0.00015160044750397441,
    "epoch": 0.32961378911148986,
    "step": 10250
  },
  {
    "loss": 0.11639647483825684,
    "grad_norm": 0.6206676959991455,
    "learning_rate": 0.00015151360474624729,
    "epoch": 0.32993536353989134,
    "step": 10260
  },
  {
    "loss": 0.11473977565765381,
    "grad_norm": 0.6148909330368042,
    "learning_rate": 0.00015142670907309098,
    "epoch": 0.33025693796829275,
    "step": 10270
  },
  {
    "loss": 0.09996345043182372,
    "grad_norm": 0.44436290860176086,
    "learning_rate": 0.0001513397605737658,
    "epoch": 0.3305785123966942,
    "step": 10280
  },
  {
    "loss": 0.09571771621704102,
    "grad_norm": 0.6439607739448547,
    "learning_rate": 0.00015125275933758637,
    "epoch": 0.33090008682509564,
    "step": 10290
  },
  {
    "loss": 0.08798513412475586,
    "grad_norm": 0.6966301798820496,
    "learning_rate": 0.00015116570545392147,
    "epoch": 0.3312216612534971,
    "step": 10300
  },
  {
    "loss": 0.10646851062774658,
    "grad_norm": 0.459700345993042,
    "learning_rate": 0.00015107859901219395,
    "epoch": 0.3315432356818986,
    "step": 10310
  },
  {
    "loss": 0.10774039030075074,
    "grad_norm": 0.613568902015686,
    "learning_rate": 0.00015099144010188067,
    "epoch": 0.3318648101103,
    "step": 10320
  },
  {
    "loss": 0.10567674636840821,
    "grad_norm": 0.8810312151908875,
    "learning_rate": 0.0001509042288125123,
    "epoch": 0.3321863845387015,
    "step": 10330
  },
  {
    "loss": 0.11079902648925781,
    "grad_norm": 0.727370023727417,
    "learning_rate": 0.0001508169652336735,
    "epoch": 0.33250795896710295,
    "step": 10340
  },
  {
    "loss": 0.08177146911621094,
    "grad_norm": 0.5525153875350952,
    "learning_rate": 0.00015072964945500242,
    "epoch": 0.33282953339550436,
    "step": 10350
  },
  {
    "loss": 0.09797362089157105,
    "grad_norm": 0.5044106841087341,
    "learning_rate": 0.00015064228156619103,
    "epoch": 0.33315110782390583,
    "step": 10360
  },
  {
    "loss": 0.10018237829208373,
    "grad_norm": 0.7975836992263794,
    "learning_rate": 0.00015055486165698465,
    "epoch": 0.3334726822523073,
    "step": 10370
  },
  {
    "loss": 0.0940122663974762,
    "grad_norm": 0.5432740449905396,
    "learning_rate": 0.0001504673898171822,
    "epoch": 0.3337942566807087,
    "step": 10380
  },
  {
    "loss": 0.07373791337013244,
    "grad_norm": 0.6882239580154419,
    "learning_rate": 0.00015037986613663574,
    "epoch": 0.3341158311091102,
    "step": 10390
  },
  {
    "loss": 0.09904557466506958,
    "grad_norm": 0.5263988375663757,
    "learning_rate": 0.00015029229070525086,
    "epoch": 0.33443740553751167,
    "step": 10400
  },
  {
    "loss": 0.08657509088516235,
    "grad_norm": 0.8776784539222717,
    "learning_rate": 0.0001502046636129861,
    "epoch": 0.3347589799659131,
    "step": 10410
  },
  {
    "loss": 0.10448260307312011,
    "grad_norm": 0.8565472364425659,
    "learning_rate": 0.00015011698494985307,
    "epoch": 0.33508055439431456,
    "step": 10420
  },
  {
    "loss": 0.10305610895156861,
    "grad_norm": 0.779041051864624,
    "learning_rate": 0.00015002925480591642,
    "epoch": 0.33540212882271603,
    "step": 10430
  },
  {
    "loss": 0.1126700758934021,
    "grad_norm": 0.8828395009040833,
    "learning_rate": 0.00014994147327129374,
    "epoch": 0.33572370325111744,
    "step": 10440
  },
  {
    "loss": 0.08342687487602234,
    "grad_norm": 0.624516487121582,
    "learning_rate": 0.00014985364043615528,
    "epoch": 0.3360452776795189,
    "step": 10450
  },
  {
    "loss": 0.09242086410522461,
    "grad_norm": 1.018872618675232,
    "learning_rate": 0.0001497657563907241,
    "epoch": 0.3363668521079204,
    "step": 10460
  },
  {
    "loss": 0.08708491325378417,
    "grad_norm": 0.2908959686756134,
    "learning_rate": 0.00014967782122527566,
    "epoch": 0.3366884265363218,
    "step": 10470
  },
  {
    "loss": 0.08389623165130615,
    "grad_norm": 0.43761253356933594,
    "learning_rate": 0.00014958983503013826,
    "epoch": 0.3370100009647233,
    "step": 10480
  },
  {
    "loss": 0.10015045404434204,
    "grad_norm": 0.7681123614311218,
    "learning_rate": 0.00014950179789569232,
    "epoch": 0.33733157539312475,
    "step": 10490
  },
  {
    "loss": 0.10144755840301514,
    "grad_norm": 0.5038902759552002,
    "learning_rate": 0.00014941370991237077,
    "epoch": 0.33765314982152617,
    "step": 10500
  },
  {
    "loss": 0.08995380401611328,
    "grad_norm": 0.3418147563934326,
    "learning_rate": 0.00014932557117065866,
    "epoch": 0.33797472424992764,
    "step": 10510
  },
  {
    "loss": 0.10645936727523804,
    "grad_norm": 0.872319221496582,
    "learning_rate": 0.00014923738176109323,
    "epoch": 0.3382962986783291,
    "step": 10520
  },
  {
    "loss": 0.07098982334136963,
    "grad_norm": 0.42508718371391296,
    "learning_rate": 0.0001491491417742638,
    "epoch": 0.3386178731067305,
    "step": 10530
  },
  {
    "loss": 0.09140524864196778,
    "grad_norm": 0.6166735291481018,
    "learning_rate": 0.00014906085130081157,
    "epoch": 0.338939447535132,
    "step": 10540
  },
  {
    "loss": 0.10014245510101319,
    "grad_norm": 0.1629897654056549,
    "learning_rate": 0.00014897251043142964,
    "epoch": 0.33926102196353347,
    "step": 10550
  },
  {
    "loss": 0.12346161603927612,
    "grad_norm": 0.5388593077659607,
    "learning_rate": 0.00014888411925686288,
    "epoch": 0.3395825963919349,
    "step": 10560
  },
  {
    "loss": 0.09298954606056213,
    "grad_norm": 0.9729686379432678,
    "learning_rate": 0.00014879567786790784,
    "epoch": 0.33990417082033636,
    "step": 10570
  },
  {
    "loss": 0.10407984256744385,
    "grad_norm": 0.6590120196342468,
    "learning_rate": 0.00014870718635541259,
    "epoch": 0.34022574524873783,
    "step": 10580
  },
  {
    "loss": 0.1211812973022461,
    "grad_norm": 0.7960436940193176,
    "learning_rate": 0.0001486186448102768,
    "epoch": 0.34054731967713925,
    "step": 10590
  },
  {
    "loss": 0.09517050385475159,
    "grad_norm": 0.4327518939971924,
    "learning_rate": 0.00014853005332345144,
    "epoch": 0.3408688941055407,
    "step": 10600
  },
  {
    "loss": 0.0846939206123352,
    "grad_norm": 0.7907319664955139,
    "learning_rate": 0.0001484414119859388,
    "epoch": 0.3411904685339422,
    "step": 10610
  },
  {
    "loss": 0.08831584453582764,
    "grad_norm": 0.7718275785446167,
    "learning_rate": 0.0001483527208887924,
    "epoch": 0.3415120429623436,
    "step": 10620
  },
  {
    "loss": 0.11044741868972778,
    "grad_norm": 1.1279953718185425,
    "learning_rate": 0.00014826398012311688,
    "epoch": 0.3418336173907451,
    "step": 10630
  },
  {
    "loss": 0.09619516134262085,
    "grad_norm": 0.40065017342567444,
    "learning_rate": 0.00014817518978006784,
    "epoch": 0.34215519181914655,
    "step": 10640
  },
  {
    "loss": 0.1134846568107605,
    "grad_norm": 0.48741984367370605,
    "learning_rate": 0.00014808634995085187,
    "epoch": 0.34247676624754797,
    "step": 10650
  },
  {
    "loss": 0.0754625678062439,
    "grad_norm": 0.6248498558998108,
    "learning_rate": 0.00014799746072672638,
    "epoch": 0.34279834067594944,
    "step": 10660
  },
  {
    "loss": 0.08936263918876648,
    "grad_norm": 0.5139371156692505,
    "learning_rate": 0.0001479085221989995,
    "epoch": 0.3431199151043509,
    "step": 10670
  },
  {
    "loss": 0.09549421072006226,
    "grad_norm": 0.8832308053970337,
    "learning_rate": 0.00014781953445903003,
    "epoch": 0.34344148953275233,
    "step": 10680
  },
  {
    "loss": 0.10033886432647705,
    "grad_norm": 0.478547066450119,
    "learning_rate": 0.00014773049759822727,
    "epoch": 0.3437630639611538,
    "step": 10690
  },
  {
    "loss": 0.09127762913703918,
    "grad_norm": 0.669744074344635,
    "learning_rate": 0.00014764141170805107,
    "epoch": 0.3440846383895553,
    "step": 10700
  },
  {
    "loss": 0.0837335467338562,
    "grad_norm": 0.4499804675579071,
    "learning_rate": 0.0001475522768800115,
    "epoch": 0.3444062128179567,
    "step": 10710
  },
  {
    "loss": 0.0952267050743103,
    "grad_norm": 0.5402239561080933,
    "learning_rate": 0.0001474630932056691,
    "epoch": 0.34472778724635816,
    "step": 10720
  },
  {
    "loss": 0.13682855367660524,
    "grad_norm": 0.9907369017601013,
    "learning_rate": 0.0001473738607766344,
    "epoch": 0.34504936167475964,
    "step": 10730
  },
  {
    "loss": 0.08810251951217651,
    "grad_norm": 0.4828674793243408,
    "learning_rate": 0.0001472845796845681,
    "epoch": 0.34537093610316105,
    "step": 10740
  },
  {
    "loss": 0.11136370897293091,
    "grad_norm": 0.699180006980896,
    "learning_rate": 0.00014719525002118087,
    "epoch": 0.3456925105315625,
    "step": 10750
  },
  {
    "loss": 0.08398210406303405,
    "grad_norm": 0.6285035014152527,
    "learning_rate": 0.0001471058718782333,
    "epoch": 0.346014084959964,
    "step": 10760
  },
  {
    "loss": 0.08259177803993226,
    "grad_norm": 0.8560423851013184,
    "learning_rate": 0.00014701644534753568,
    "epoch": 0.3463356593883654,
    "step": 10770
  },
  {
    "loss": 0.07981761693954467,
    "grad_norm": 0.3733856976032257,
    "learning_rate": 0.0001469269705209482,
    "epoch": 0.3466572338167669,
    "step": 10780
  },
  {
    "loss": 0.08973867297172547,
    "grad_norm": 0.7273002862930298,
    "learning_rate": 0.00014683744749038045,
    "epoch": 0.34697880824516836,
    "step": 10790
  },
  {
    "loss": 0.10675228834152221,
    "grad_norm": 0.7694258093833923,
    "learning_rate": 0.0001467478763477916,
    "epoch": 0.3473003826735698,
    "step": 10800
  },
  {
    "loss": 0.07637117505073547,
    "grad_norm": 0.9570711851119995,
    "learning_rate": 0.00014665825718519036,
    "epoch": 0.34762195710197125,
    "step": 10810
  },
  {
    "loss": 0.12145779132843018,
    "grad_norm": 0.9601441621780396,
    "learning_rate": 0.00014656859009463458,
    "epoch": 0.3479435315303727,
    "step": 10820
  },
  {
    "loss": 0.07213935852050782,
    "grad_norm": 0.8995713591575623,
    "learning_rate": 0.00014647887516823152,
    "epoch": 0.34826510595877413,
    "step": 10830
  },
  {
    "loss": 0.12633739709854125,
    "grad_norm": 1.2538115978240967,
    "learning_rate": 0.00014638911249813746,
    "epoch": 0.3485866803871756,
    "step": 10840
  },
  {
    "loss": 0.09029225111007691,
    "grad_norm": 0.6341054439544678,
    "learning_rate": 0.0001462993021765577,
    "epoch": 0.3489082548155771,
    "step": 10850
  },
  {
    "loss": 0.08201388716697693,
    "grad_norm": 0.48226431012153625,
    "learning_rate": 0.00014620944429574658,
    "epoch": 0.3492298292439785,
    "step": 10860
  },
  {
    "loss": 0.07496067881584167,
    "grad_norm": 0.7661513090133667,
    "learning_rate": 0.0001461195389480073,
    "epoch": 0.34955140367237997,
    "step": 10870
  },
  {
    "eval_loss": 0.09110351651906967,
    "eval_runtime": 34.3603,
    "eval_samples_per_second": 146.274,
    "eval_steps_per_second": 36.583,
    "epoch": 0.3498086632151011,
    "step": 10878
  },
  {
    "loss": 0.0849322497844696,
    "grad_norm": 0.9236146211624146,
    "learning_rate": 0.00014602958622569174,
    "epoch": 0.34987297810078144,
    "step": 10880
  },
  {
    "loss": 0.09441561698913574,
    "grad_norm": 0.5579268336296082,
    "learning_rate": 0.00014593958622120045,
    "epoch": 0.35019455252918286,
    "step": 10890
  },
  {
    "loss": 0.09172694087028503,
    "grad_norm": 0.45912206172943115,
    "learning_rate": 0.00014584953902698261,
    "epoch": 0.35051612695758433,
    "step": 10900
  },
  {
    "loss": 0.09633524417877197,
    "grad_norm": 0.5850641131401062,
    "learning_rate": 0.00014575944473553583,
    "epoch": 0.3508377013859858,
    "step": 10910
  },
  {
    "loss": 0.09510902166366578,
    "grad_norm": 0.9028190970420837,
    "learning_rate": 0.00014566930343940613,
    "epoch": 0.3511592758143872,
    "step": 10920
  },
  {
    "loss": 0.10485677719116211,
    "grad_norm": 0.5650879740715027,
    "learning_rate": 0.0001455791152311878,
    "epoch": 0.3514808502427887,
    "step": 10930
  },
  {
    "loss": 0.11609166860580444,
    "grad_norm": 7.561282157897949,
    "learning_rate": 0.00014548888020352328,
    "epoch": 0.35180242467119016,
    "step": 10940
  },
  {
    "loss": 0.09210947751998902,
    "grad_norm": 0.38646945357322693,
    "learning_rate": 0.00014539859844910319,
    "epoch": 0.3521239990995916,
    "step": 10950
  },
  {
    "loss": 0.08583818674087525,
    "grad_norm": 0.38824743032455444,
    "learning_rate": 0.000145308270060666,
    "epoch": 0.35244557352799305,
    "step": 10960
  },
  {
    "loss": 0.08737082481384277,
    "grad_norm": 0.40544432401657104,
    "learning_rate": 0.00014521789513099827,
    "epoch": 0.3527671479563945,
    "step": 10970
  },
  {
    "loss": 0.08975743055343628,
    "grad_norm": 0.8956839442253113,
    "learning_rate": 0.00014512747375293423,
    "epoch": 0.35308872238479594,
    "step": 10980
  },
  {
    "loss": 0.10598586797714234,
    "grad_norm": 0.9445635676383972,
    "learning_rate": 0.0001450370060193559,
    "epoch": 0.3534102968131974,
    "step": 10990
  },
  {
    "loss": 0.08878449201583863,
    "grad_norm": 0.46113863587379456,
    "learning_rate": 0.00014494649202319282,
    "epoch": 0.3537318712415989,
    "step": 11000
  },
  {
    "loss": 0.08599361181259155,
    "grad_norm": 0.3610915243625641,
    "learning_rate": 0.0001448559318574222,
    "epoch": 0.3540534456700003,
    "step": 11010
  },
  {
    "loss": 0.08179433941841126,
    "grad_norm": 0.7075566053390503,
    "learning_rate": 0.00014476532561506856,
    "epoch": 0.35437502009840177,
    "step": 11020
  },
  {
    "loss": 0.07348551750183105,
    "grad_norm": 0.7016006708145142,
    "learning_rate": 0.0001446746733892037,
    "epoch": 0.35469659452680324,
    "step": 11030
  },
  {
    "loss": 0.08005476593971253,
    "grad_norm": 0.4309087097644806,
    "learning_rate": 0.00014458397527294692,
    "epoch": 0.35501816895520466,
    "step": 11040
  },
  {
    "loss": 0.11042691469192505,
    "grad_norm": 0.7881134748458862,
    "learning_rate": 0.00014449323135946434,
    "epoch": 0.35533974338360613,
    "step": 11050
  },
  {
    "loss": 0.08432947397232056,
    "grad_norm": 0.7723735570907593,
    "learning_rate": 0.00014440244174196925,
    "epoch": 0.3556613178120076,
    "step": 11060
  },
  {
    "loss": 0.10001605749130249,
    "grad_norm": 0.8078780174255371,
    "learning_rate": 0.00014431160651372197,
    "epoch": 0.355982892240409,
    "step": 11070
  },
  {
    "loss": 0.08675259351730347,
    "grad_norm": 0.47705569863319397,
    "learning_rate": 0.00014422072576802958,
    "epoch": 0.3563044666688105,
    "step": 11080
  },
  {
    "loss": 0.084046471118927,
    "grad_norm": 0.9069617986679077,
    "learning_rate": 0.00014412979959824593,
    "epoch": 0.35662604109721197,
    "step": 11090
  },
  {
    "loss": 0.10433726310729981,
    "grad_norm": 0.8548468351364136,
    "learning_rate": 0.00014403882809777154,
    "epoch": 0.3569476155256134,
    "step": 11100
  },
  {
    "loss": 0.10171997547149658,
    "grad_norm": 0.6188079714775085,
    "learning_rate": 0.00014394781136005345,
    "epoch": 0.35726918995401485,
    "step": 11110
  },
  {
    "loss": 0.09457895755767823,
    "grad_norm": 0.5230410099029541,
    "learning_rate": 0.00014385674947858527,
    "epoch": 0.3575907643824163,
    "step": 11120
  },
  {
    "loss": 0.10628050565719604,
    "grad_norm": 1.2344039678573608,
    "learning_rate": 0.0001437656425469069,
    "epoch": 0.35791233881081774,
    "step": 11130
  },
  {
    "loss": 0.070767343044281,
    "grad_norm": 0.5525897145271301,
    "learning_rate": 0.00014367449065860453,
    "epoch": 0.3582339132392192,
    "step": 11140
  },
  {
    "loss": 0.08599192500114441,
    "grad_norm": 0.5532211661338806,
    "learning_rate": 0.00014358329390731057,
    "epoch": 0.3585554876676207,
    "step": 11150
  },
  {
    "loss": 0.09453628659248352,
    "grad_norm": 0.57228684425354,
    "learning_rate": 0.00014349205238670343,
    "epoch": 0.3588770620960221,
    "step": 11160
  },
  {
    "loss": 0.07618891596794128,
    "grad_norm": 0.5582618713378906,
    "learning_rate": 0.00014340076619050754,
    "epoch": 0.3591986365244236,
    "step": 11170
  },
  {
    "loss": 0.062136721611022946,
    "grad_norm": 0.43636590242385864,
    "learning_rate": 0.00014330943541249335,
    "epoch": 0.35952021095282505,
    "step": 11180
  },
  {
    "loss": 0.09901108741760253,
    "grad_norm": 0.5479429364204407,
    "learning_rate": 0.0001432180601464769,
    "epoch": 0.35984178538122646,
    "step": 11190
  },
  {
    "loss": 0.0857438862323761,
    "grad_norm": 0.5808840990066528,
    "learning_rate": 0.00014312664048632008,
    "epoch": 0.36016335980962794,
    "step": 11200
  },
  {
    "loss": 0.08954665064811707,
    "grad_norm": 0.6454494595527649,
    "learning_rate": 0.00014303517652593025,
    "epoch": 0.3604849342380294,
    "step": 11210
  },
  {
    "loss": 0.09943202137947083,
    "grad_norm": 0.602885365486145,
    "learning_rate": 0.0001429436683592604,
    "epoch": 0.3608065086664308,
    "step": 11220
  },
  {
    "loss": 0.0752259373664856,
    "grad_norm": 0.4851940870285034,
    "learning_rate": 0.00014285211608030892,
    "epoch": 0.3611280830948323,
    "step": 11230
  },
  {
    "loss": 0.0808374285697937,
    "grad_norm": 0.6741466522216797,
    "learning_rate": 0.00014276051978311943,
    "epoch": 0.36144965752323377,
    "step": 11240
  },
  {
    "loss": 0.09836857914924621,
    "grad_norm": 0.398953378200531,
    "learning_rate": 0.00014266887956178081,
    "epoch": 0.3617712319516352,
    "step": 11250
  },
  {
    "loss": 0.10557745695114136,
    "grad_norm": 0.6230091452598572,
    "learning_rate": 0.00014257719551042706,
    "epoch": 0.36209280638003666,
    "step": 11260
  },
  {
    "loss": 0.09444069862365723,
    "grad_norm": 0.8618478178977966,
    "learning_rate": 0.0001424854677232372,
    "epoch": 0.36241438080843813,
    "step": 11270
  },
  {
    "loss": 0.07490069270133973,
    "grad_norm": 0.6330289840698242,
    "learning_rate": 0.00014239369629443519,
    "epoch": 0.36273595523683955,
    "step": 11280
  },
  {
    "loss": 0.10039694309234619,
    "grad_norm": 0.496137410402298,
    "learning_rate": 0.0001423018813182898,
    "epoch": 0.363057529665241,
    "step": 11290
  },
  {
    "loss": 0.09477540254592895,
    "grad_norm": 0.8526451587677002,
    "learning_rate": 0.00014221002288911458,
    "epoch": 0.3633791040936425,
    "step": 11300
  },
  {
    "loss": 0.08862870931625366,
    "grad_norm": 0.8001587390899658,
    "learning_rate": 0.0001421181211012676,
    "epoch": 0.3637006785220439,
    "step": 11310
  },
  {
    "loss": 0.09040194749832153,
    "grad_norm": 0.46519744396209717,
    "learning_rate": 0.00014202617604915161,
    "epoch": 0.3640222529504454,
    "step": 11320
  },
  {
    "loss": 0.10902376174926758,
    "grad_norm": 0.8219383358955383,
    "learning_rate": 0.0001419341878272137,
    "epoch": 0.36434382737884685,
    "step": 11330
  },
  {
    "loss": 0.08118026852607726,
    "grad_norm": 0.3547578752040863,
    "learning_rate": 0.00014184215652994532,
    "epoch": 0.36466540180724827,
    "step": 11340
  },
  {
    "loss": 0.09958731532096862,
    "grad_norm": 0.46915847063064575,
    "learning_rate": 0.00014175008225188226,
    "epoch": 0.36498697623564974,
    "step": 11350
  },
  {
    "loss": 0.11956846714019775,
    "grad_norm": 1.017486810684204,
    "learning_rate": 0.0001416579650876043,
    "epoch": 0.3653085506640512,
    "step": 11360
  },
  {
    "loss": 0.07706149220466614,
    "grad_norm": 0.2844119071960449,
    "learning_rate": 0.00014156580513173544,
    "epoch": 0.36563012509245263,
    "step": 11370
  },
  {
    "loss": 0.086731219291687,
    "grad_norm": 0.4540342092514038,
    "learning_rate": 0.00014147360247894348,
    "epoch": 0.3659516995208541,
    "step": 11380
  },
  {
    "loss": 0.09100791215896606,
    "grad_norm": 0.6833384037017822,
    "learning_rate": 0.00014138135722394023,
    "epoch": 0.3662732739492556,
    "step": 11390
  },
  {
    "loss": 0.1060667872428894,
    "grad_norm": 0.550255298614502,
    "learning_rate": 0.00014128906946148114,
    "epoch": 0.366594848377657,
    "step": 11400
  },
  {
    "loss": 0.08594973087310791,
    "grad_norm": 0.6789250373840332,
    "learning_rate": 0.00014119673928636542,
    "epoch": 0.36691642280605846,
    "step": 11410
  },
  {
    "loss": 0.08912253379821777,
    "grad_norm": 0.5137705206871033,
    "learning_rate": 0.00014110436679343576,
    "epoch": 0.36723799723445993,
    "step": 11420
  },
  {
    "loss": 0.094441157579422,
    "grad_norm": 0.8425323963165283,
    "learning_rate": 0.00014101195207757837,
    "epoch": 0.36755957166286135,
    "step": 11430
  },
  {
    "loss": 0.09220321178436279,
    "grad_norm": 0.7896093726158142,
    "learning_rate": 0.00014091949523372277,
    "epoch": 0.3678811460912628,
    "step": 11440
  },
  {
    "loss": 0.09599274396896362,
    "grad_norm": 0.8824585676193237,
    "learning_rate": 0.00014082699635684188,
    "epoch": 0.3682027205196643,
    "step": 11450
  },
  {
    "loss": 0.10436227321624755,
    "grad_norm": 1.1513099670410156,
    "learning_rate": 0.00014073445554195173,
    "epoch": 0.3685242949480657,
    "step": 11460
  },
  {
    "loss": 0.08480619192123413,
    "grad_norm": 0.5512106418609619,
    "learning_rate": 0.00014064187288411132,
    "epoch": 0.3688458693764672,
    "step": 11470
  },
  {
    "loss": 0.09227036833763122,
    "grad_norm": 0.5305521488189697,
    "learning_rate": 0.0001405492484784228,
    "epoch": 0.36916744380486866,
    "step": 11480
  },
  {
    "loss": 0.07005593776702881,
    "grad_norm": 0.2772514224052429,
    "learning_rate": 0.00014045658242003116,
    "epoch": 0.3694890182332701,
    "step": 11490
  },
  {
    "loss": 0.08746758699417115,
    "grad_norm": 0.6788966059684753,
    "learning_rate": 0.00014036387480412412,
    "epoch": 0.36981059266167154,
    "step": 11500
  },
  {
    "loss": 0.09942635297775268,
    "grad_norm": 0.2725159525871277,
    "learning_rate": 0.00014027112572593213,
    "epoch": 0.370132167090073,
    "step": 11510
  },
  {
    "loss": 0.08165991306304932,
    "grad_norm": 0.7385566830635071,
    "learning_rate": 0.0001401783352807282,
    "epoch": 0.37045374151847443,
    "step": 11520
  },
  {
    "loss": 0.09183178544044494,
    "grad_norm": 0.42521658539772034,
    "learning_rate": 0.00014008550356382787,
    "epoch": 0.3707753159468759,
    "step": 11530
  },
  {
    "loss": 0.09264058470726014,
    "grad_norm": 0.7075470089912415,
    "learning_rate": 0.00013999263067058905,
    "epoch": 0.3710968903752774,
    "step": 11540
  },
  {
    "loss": 0.10476528406143189,
    "grad_norm": 0.48848652839660645,
    "learning_rate": 0.00013989971669641199,
    "epoch": 0.3714184648036788,
    "step": 11550
  },
  {
    "loss": 0.10007588863372803,
    "grad_norm": 0.7352356314659119,
    "learning_rate": 0.00013980676173673906,
    "epoch": 0.37174003923208027,
    "step": 11560
  },
  {
    "loss": 0.09925302863121033,
    "grad_norm": 0.6485668420791626,
    "learning_rate": 0.0001397137658870548,
    "epoch": 0.37206161366048174,
    "step": 11570
  },
  {
    "loss": 0.09216221570968627,
    "grad_norm": 0.4631863534450531,
    "learning_rate": 0.00013962072924288572,
    "epoch": 0.37238318808888315,
    "step": 11580
  },
  {
    "loss": 0.10099647045135499,
    "grad_norm": 0.7892594933509827,
    "learning_rate": 0.00013952765189980025,
    "epoch": 0.3727047625172846,
    "step": 11590
  },
  {
    "loss": 0.08765991926193237,
    "grad_norm": 0.5219521522521973,
    "learning_rate": 0.00013943453395340858,
    "epoch": 0.3730263369456861,
    "step": 11600
  },
  {
    "loss": 0.0861083447933197,
    "grad_norm": 0.860974907875061,
    "learning_rate": 0.00013934137549936272,
    "epoch": 0.3733479113740875,
    "step": 11610
  },
  {
    "loss": 0.0927880048751831,
    "grad_norm": 0.8678325414657593,
    "learning_rate": 0.00013924817663335618,
    "epoch": 0.373669485802489,
    "step": 11620
  },
  {
    "loss": 0.09621718525886536,
    "grad_norm": 0.794443666934967,
    "learning_rate": 0.00013915493745112397,
    "epoch": 0.37399106023089046,
    "step": 11630
  },
  {
    "loss": 0.08052088022232055,
    "grad_norm": 0.5400531888008118,
    "learning_rate": 0.00013906165804844267,
    "epoch": 0.3743126346592919,
    "step": 11640
  },
  {
    "loss": 0.0943139672279358,
    "grad_norm": 0.9415629506111145,
    "learning_rate": 0.00013896833852113,
    "epoch": 0.37463420908769335,
    "step": 11650
  },
  {
    "loss": 0.08249766230583191,
    "grad_norm": 0.7084611654281616,
    "learning_rate": 0.00013887497896504495,
    "epoch": 0.3749557835160948,
    "step": 11660
  },
  {
    "loss": 0.09889563918113708,
    "grad_norm": 0.5836837291717529,
    "learning_rate": 0.00013878157947608765,
    "epoch": 0.37527735794449624,
    "step": 11670
  },
  {
    "loss": 0.08768828511238098,
    "grad_norm": 0.5825384855270386,
    "learning_rate": 0.0001386881401501993,
    "epoch": 0.3755989323728977,
    "step": 11680
  },
  {
    "loss": 0.10164982080459595,
    "grad_norm": 0.8690288662910461,
    "learning_rate": 0.0001385946610833619,
    "epoch": 0.3759205068012992,
    "step": 11690
  },
  {
    "loss": 0.085177081823349,
    "grad_norm": 0.8311650156974792,
    "learning_rate": 0.00013850114237159834,
    "epoch": 0.3762420812297006,
    "step": 11700
  },
  {
    "loss": 0.09417153596878051,
    "grad_norm": 0.44406217336654663,
    "learning_rate": 0.00013840758411097224,
    "epoch": 0.37656365565810207,
    "step": 11710
  },
  {
    "loss": 0.09739122986793518,
    "grad_norm": 0.690326452255249,
    "learning_rate": 0.00013831398639758787,
    "epoch": 0.37688523008650354,
    "step": 11720
  },
  {
    "loss": 0.08809913396835327,
    "grad_norm": 0.80722975730896,
    "learning_rate": 0.0001382203493275899,
    "epoch": 0.37720680451490496,
    "step": 11730
  },
  {
    "loss": 0.10492403507232666,
    "grad_norm": 0.5662553310394287,
    "learning_rate": 0.00013812667299716358,
    "epoch": 0.37752837894330643,
    "step": 11740
  },
  {
    "loss": 0.08379825353622436,
    "grad_norm": 0.5302266478538513,
    "learning_rate": 0.00013803295750253444,
    "epoch": 0.3778499533717079,
    "step": 11750
  },
  {
    "loss": 0.10625137090682983,
    "grad_norm": 0.8091083765029907,
    "learning_rate": 0.0001379392029399682,
    "epoch": 0.3781715278001093,
    "step": 11760
  },
  {
    "loss": 0.06244894862174988,
    "grad_norm": 0.5015480518341064,
    "learning_rate": 0.00013784540940577078,
    "epoch": 0.3784931022285108,
    "step": 11770
  },
  {
    "loss": 0.07549222707748413,
    "grad_norm": 0.34560057520866394,
    "learning_rate": 0.00013775157699628802,
    "epoch": 0.37881467665691226,
    "step": 11780
  },
  {
    "loss": 0.06123560070991516,
    "grad_norm": 0.4780142903327942,
    "learning_rate": 0.00013765770580790581,
    "epoch": 0.3791362510853137,
    "step": 11790
  },
  {
    "loss": 0.09585493803024292,
    "grad_norm": 0.5067535042762756,
    "learning_rate": 0.0001375637959370498,
    "epoch": 0.37945782551371515,
    "step": 11800
  },
  {
    "loss": 0.10015147924423218,
    "grad_norm": 0.5134025812149048,
    "learning_rate": 0.00013746984748018543,
    "epoch": 0.3797793999421166,
    "step": 11810
  },
  {
    "loss": 0.08560925722122192,
    "grad_norm": 0.7771230936050415,
    "learning_rate": 0.00013737586053381773,
    "epoch": 0.38010097437051804,
    "step": 11820
  },
  {
    "loss": 0.08708308935165406,
    "grad_norm": 0.5460475087165833,
    "learning_rate": 0.0001372818351944913,
    "epoch": 0.3804225487989195,
    "step": 11830
  },
  {
    "loss": 0.09841415882110596,
    "grad_norm": 0.6091669797897339,
    "learning_rate": 0.00013718777155879012,
    "epoch": 0.380744123227321,
    "step": 11840
  },
  {
    "loss": 0.1109818696975708,
    "grad_norm": 0.7367703914642334,
    "learning_rate": 0.0001370936697233376,
    "epoch": 0.3810656976557224,
    "step": 11850
  },
  {
    "loss": 0.11050435304641723,
    "grad_norm": 0.7365938425064087,
    "learning_rate": 0.00013699952978479634,
    "epoch": 0.3813872720841239,
    "step": 11860
  },
  {
    "loss": 0.08165845274925232,
    "grad_norm": 0.8522195816040039,
    "learning_rate": 0.00013690535183986805,
    "epoch": 0.38170884651252535,
    "step": 11870
  },
  {
    "loss": 0.09400575757026672,
    "grad_norm": 0.5556536912918091,
    "learning_rate": 0.00013681113598529356,
    "epoch": 0.38203042094092676,
    "step": 11880
  },
  {
    "loss": 0.08722950220108032,
    "grad_norm": 0.5536159873008728,
    "learning_rate": 0.0001367168823178525,
    "epoch": 0.38235199536932823,
    "step": 11890
  },
  {
    "loss": 0.08267619609832763,
    "grad_norm": 0.701835572719574,
    "learning_rate": 0.00013662259093436357,
    "epoch": 0.3826735697977297,
    "step": 11900
  },
  {
    "loss": 0.09268981218338013,
    "grad_norm": 1.0157867670059204,
    "learning_rate": 0.00013652826193168393,
    "epoch": 0.3829951442261311,
    "step": 11910
  },
  {
    "loss": 0.09985225200653076,
    "grad_norm": 0.48218435049057007,
    "learning_rate": 0.00013643389540670962,
    "epoch": 0.3833167186545326,
    "step": 11920
  },
  {
    "loss": 0.08864061832427979,
    "grad_norm": 0.6998544335365295,
    "learning_rate": 0.00013633949145637516,
    "epoch": 0.38363829308293407,
    "step": 11930
  },
  {
    "loss": 0.10292292833328247,
    "grad_norm": 0.6836029887199402,
    "learning_rate": 0.0001362450501776534,
    "epoch": 0.3839598675113355,
    "step": 11940
  },
  {
    "loss": 0.07776576280593872,
    "grad_norm": 0.4844030439853668,
    "learning_rate": 0.00013615057166755566,
    "epoch": 0.38428144193973696,
    "step": 11950
  },
  {
    "loss": 0.08032951354980469,
    "grad_norm": 0.4513522982597351,
    "learning_rate": 0.0001360560560231315,
    "epoch": 0.38460301636813843,
    "step": 11960
  },
  {
    "loss": 0.09304511547088623,
    "grad_norm": 0.4937937259674072,
    "learning_rate": 0.00013596150334146853,
    "epoch": 0.38492459079653984,
    "step": 11970
  },
  {
    "loss": 0.09666025638580322,
    "grad_norm": 0.716979444026947,
    "learning_rate": 0.00013586691371969252,
    "epoch": 0.3852461652249413,
    "step": 11980
  },
  {
    "loss": 0.08163623809814453,
    "grad_norm": 0.41440537571907043,
    "learning_rate": 0.0001357722872549671,
    "epoch": 0.3855677396533428,
    "step": 11990
  },
  {
    "loss": 0.08722010254859924,
    "grad_norm": 0.5127982497215271,
    "learning_rate": 0.00013567762404449377,
    "epoch": 0.3858893140817442,
    "step": 12000
  },
  {
    "loss": 0.08775334358215332,
    "grad_norm": 0.5351473093032837,
    "learning_rate": 0.00013558292418551176,
    "epoch": 0.3862108885101457,
    "step": 12010
  },
  {
    "loss": 0.08145318031311036,
    "grad_norm": 0.6967948079109192,
    "learning_rate": 0.0001354881877752981,
    "epoch": 0.38653246293854715,
    "step": 12020
  },
  {
    "loss": 0.09338672161102295,
    "grad_norm": 0.4476488530635834,
    "learning_rate": 0.00013539341491116707,
    "epoch": 0.38685403736694857,
    "step": 12030
  },
  {
    "loss": 0.09646210074424744,
    "grad_norm": 0.4933212101459503,
    "learning_rate": 0.00013529860569047066,
    "epoch": 0.38717561179535004,
    "step": 12040
  },
  {
    "loss": 0.10104117393493653,
    "grad_norm": 0.7293670773506165,
    "learning_rate": 0.00013520376021059803,
    "epoch": 0.3874971862237515,
    "step": 12050
  },
  {
    "loss": 0.09380577206611633,
    "grad_norm": 0.5708062648773193,
    "learning_rate": 0.00013510887856897573,
    "epoch": 0.3878187606521529,
    "step": 12060
  },
  {
    "loss": 0.08531683683395386,
    "grad_norm": 0.51509690284729,
    "learning_rate": 0.00013501396086306738,
    "epoch": 0.3881403350805544,
    "step": 12070
  },
  {
    "loss": 0.10160613059997559,
    "grad_norm": 0.4665564000606537,
    "learning_rate": 0.0001349190071903736,
    "epoch": 0.38846190950895587,
    "step": 12080
  },
  {
    "loss": 0.09982122182846069,
    "grad_norm": 0.5586184859275818,
    "learning_rate": 0.00013482401764843205,
    "epoch": 0.3887834839373573,
    "step": 12090
  },
  {
    "loss": 0.08948178887367249,
    "grad_norm": 0.5822875499725342,
    "learning_rate": 0.0001347289923348172,
    "epoch": 0.38910505836575876,
    "step": 12100
  },
  {
    "loss": 0.08374988436698913,
    "grad_norm": 0.44458073377609253,
    "learning_rate": 0.00013463393134714028,
    "epoch": 0.38942663279416023,
    "step": 12110
  },
  {
    "loss": 0.11546404361724853,
    "grad_norm": 0.8156406879425049,
    "learning_rate": 0.00013453883478304905,
    "epoch": 0.38974820722256165,
    "step": 12120
  },
  {
    "loss": 0.08677539825439454,
    "grad_norm": 0.5954530835151672,
    "learning_rate": 0.00013444370274022803,
    "epoch": 0.3900697816509631,
    "step": 12130
  },
  {
    "loss": 0.09240028262138367,
    "grad_norm": 0.4999423623085022,
    "learning_rate": 0.00013434853531639796,
    "epoch": 0.3903913560793646,
    "step": 12140
  },
  {
    "loss": 0.1019981861114502,
    "grad_norm": 0.7203190922737122,
    "learning_rate": 0.0001342533326093161,
    "epoch": 0.390712930507766,
    "step": 12150
  },
  {
    "loss": 0.0922100841999054,
    "grad_norm": 0.5576680898666382,
    "learning_rate": 0.00013415809471677584,
    "epoch": 0.3910345049361675,
    "step": 12160
  },
  {
    "loss": 0.09082117080688476,
    "grad_norm": 0.6921060681343079,
    "learning_rate": 0.00013406282173660674,
    "epoch": 0.39135607936456895,
    "step": 12170
  },
  {
    "loss": 0.08153097033500671,
    "grad_norm": 0.42663201689720154,
    "learning_rate": 0.00013396751376667447,
    "epoch": 0.39167765379297037,
    "step": 12180
  },
  {
    "loss": 0.1282093644142151,
    "grad_norm": 0.7410852909088135,
    "learning_rate": 0.00013387217090488057,
    "epoch": 0.39199922822137184,
    "step": 12190
  },
  {
    "loss": 0.08008560538291931,
    "grad_norm": 0.8880549669265747,
    "learning_rate": 0.0001337767932491624,
    "epoch": 0.3923208026497733,
    "step": 12200
  },
  {
    "loss": 0.0876152753829956,
    "grad_norm": 0.5166379809379578,
    "learning_rate": 0.00013368138089749312,
    "epoch": 0.39264237707817473,
    "step": 12210
  },
  {
    "loss": 0.09751138091087341,
    "grad_norm": 0.7990224361419678,
    "learning_rate": 0.00013358593394788152,
    "epoch": 0.3929639515065762,
    "step": 12220
  },
  {
    "loss": 0.07425323724746705,
    "grad_norm": 0.6802085638046265,
    "learning_rate": 0.00013349045249837194,
    "epoch": 0.3932855259349777,
    "step": 12230
  },
  {
    "loss": 0.08448225259780884,
    "grad_norm": 0.41255390644073486,
    "learning_rate": 0.00013339493664704408,
    "epoch": 0.3936071003633791,
    "step": 12240
  },
  {
    "loss": 0.09719268679618835,
    "grad_norm": 0.6711472272872925,
    "learning_rate": 0.0001332993864920131,
    "epoch": 0.39392867479178056,
    "step": 12250
  },
  {
    "loss": 0.06422892212867737,
    "grad_norm": 0.4163860082626343,
    "learning_rate": 0.00013320380213142926,
    "epoch": 0.39425024922018204,
    "step": 12260
  },
  {
    "loss": 0.08449453115463257,
    "grad_norm": 0.616906464099884,
    "learning_rate": 0.0001331081836634781,
    "epoch": 0.39457182364858345,
    "step": 12270
  },
  {
    "loss": 0.09407067894935608,
    "grad_norm": 0.6051538586616516,
    "learning_rate": 0.00013301253118638007,
    "epoch": 0.3948933980769849,
    "step": 12280
  },
  {
    "loss": 0.08197566270828247,
    "grad_norm": 0.5638774633407593,
    "learning_rate": 0.00013291684479839065,
    "epoch": 0.3952149725053864,
    "step": 12290
  },
  {
    "loss": 0.0981809675693512,
    "grad_norm": 0.822028636932373,
    "learning_rate": 0.0001328211245978001,
    "epoch": 0.3955365469337878,
    "step": 12300
  },
  {
    "loss": 0.10272608995437622,
    "grad_norm": 0.7296173572540283,
    "learning_rate": 0.00013272537068293342,
    "epoch": 0.3958581213621893,
    "step": 12310
  },
  {
    "loss": 0.09320492148399354,
    "grad_norm": 0.6523847579956055,
    "learning_rate": 0.00013262958315215026,
    "epoch": 0.39617969579059076,
    "step": 12320
  },
  {
    "loss": 0.08853684663772583,
    "grad_norm": 0.8760294318199158,
    "learning_rate": 0.00013253376210384477,
    "epoch": 0.3965012702189922,
    "step": 12330
  },
  {
    "loss": 0.08399147987365722,
    "grad_norm": 0.6736103296279907,
    "learning_rate": 0.00013243790763644555,
    "epoch": 0.39682284464739365,
    "step": 12340
  },
  {
    "loss": 0.09626861810684204,
    "grad_norm": 0.668816328048706,
    "learning_rate": 0.00013234201984841558,
    "epoch": 0.3971444190757951,
    "step": 12350
  },
  {
    "loss": 0.06886869668960571,
    "grad_norm": 0.8333956003189087,
    "learning_rate": 0.0001322460988382519,
    "epoch": 0.39746599350419654,
    "step": 12360
  },
  {
    "loss": 0.07311125993728637,
    "grad_norm": 0.7170328497886658,
    "learning_rate": 0.0001321501447044859,
    "epoch": 0.397787567932598,
    "step": 12370
  },
  {
    "loss": 0.11943765878677368,
    "grad_norm": 0.6446574330329895,
    "learning_rate": 0.00013205415754568285,
    "epoch": 0.3981091423609995,
    "step": 12380
  },
  {
    "loss": 0.07889516353607177,
    "grad_norm": 0.37039467692375183,
    "learning_rate": 0.00013195813746044195,
    "epoch": 0.3984307167894009,
    "step": 12390
  },
  {
    "loss": 0.09751490354537964,
    "grad_norm": 0.48914358019828796,
    "learning_rate": 0.0001318620845473963,
    "epoch": 0.39875229121780237,
    "step": 12400
  },
  {
    "loss": 0.09465900659561158,
    "grad_norm": 0.6299532651901245,
    "learning_rate": 0.00013176599890521264,
    "epoch": 0.39907386564620384,
    "step": 12410
  },
  {
    "loss": 0.09306957125663758,
    "grad_norm": 0.7221790552139282,
    "learning_rate": 0.00013166988063259137,
    "epoch": 0.39939544007460526,
    "step": 12420
  },
  {
    "loss": 0.09147388339042664,
    "grad_norm": 0.34951910376548767,
    "learning_rate": 0.00013157372982826638,
    "epoch": 0.39971701450300673,
    "step": 12430
  },
  {
    "eval_loss": 0.08492592722177505,
    "eval_runtime": 34.3707,
    "eval_samples_per_second": 146.229,
    "eval_steps_per_second": 36.572,
    "epoch": 0.399781329388687,
    "step": 12432
  },
  {
    "loss": 0.07842938303947448,
    "grad_norm": 0.5732839703559875,
    "learning_rate": 0.00013147754659100503,
    "epoch": 0.4000385889314082,
    "step": 12440
  },
  {
    "loss": 0.08517730832099915,
    "grad_norm": 0.6061631441116333,
    "learning_rate": 0.00013138133101960796,
    "epoch": 0.4003601633598096,
    "step": 12450
  },
  {
    "loss": 0.11794278621673585,
    "grad_norm": 0.962023913860321,
    "learning_rate": 0.00013128508321290903,
    "epoch": 0.4006817377882111,
    "step": 12460
  },
  {
    "loss": 0.10742299556732178,
    "grad_norm": 0.4836251139640808,
    "learning_rate": 0.00013118880326977515,
    "epoch": 0.40100331221661256,
    "step": 12470
  },
  {
    "loss": 0.09351553916931152,
    "grad_norm": 1.062370777130127,
    "learning_rate": 0.0001310924912891064,
    "epoch": 0.401324886645014,
    "step": 12480
  },
  {
    "loss": 0.07564758062362671,
    "grad_norm": 0.5744089484214783,
    "learning_rate": 0.00013099614736983558,
    "epoch": 0.40164646107341545,
    "step": 12490
  },
  {
    "loss": 0.07484514713287353,
    "grad_norm": 0.4435352087020874,
    "learning_rate": 0.00013089977161092847,
    "epoch": 0.4019680355018169,
    "step": 12500
  },
  {
    "loss": 0.11803302764892579,
    "grad_norm": 0.6136540174484253,
    "learning_rate": 0.00013080336411138338,
    "epoch": 0.40228960993021834,
    "step": 12510
  },
  {
    "loss": 0.08221138119697571,
    "grad_norm": 0.4812261760234833,
    "learning_rate": 0.00013070692497023137,
    "epoch": 0.4026111843586198,
    "step": 12520
  },
  {
    "loss": 0.0985057532787323,
    "grad_norm": 0.6960001587867737,
    "learning_rate": 0.00013061045428653599,
    "epoch": 0.4029327587870213,
    "step": 12530
  },
  {
    "loss": 0.08646383881568909,
    "grad_norm": 0.7220068573951721,
    "learning_rate": 0.0001305139521593931,
    "epoch": 0.4032543332154227,
    "step": 12540
  },
  {
    "loss": 0.09498414993286133,
    "grad_norm": 0.40549272298812866,
    "learning_rate": 0.0001304174186879309,
    "epoch": 0.40357590764382417,
    "step": 12550
  },
  {
    "loss": 0.08594885468482971,
    "grad_norm": 0.7031739950180054,
    "learning_rate": 0.00013032085397130983,
    "epoch": 0.40389748207222564,
    "step": 12560
  },
  {
    "loss": 0.09841355085372924,
    "grad_norm": 0.8005086779594421,
    "learning_rate": 0.00013022425810872238,
    "epoch": 0.40421905650062706,
    "step": 12570
  },
  {
    "loss": 0.08631751537322999,
    "grad_norm": 0.6979205012321472,
    "learning_rate": 0.00013012763119939303,
    "epoch": 0.40454063092902853,
    "step": 12580
  },
  {
    "loss": 0.0870095133781433,
    "grad_norm": 0.34911099076271057,
    "learning_rate": 0.00013003097334257822,
    "epoch": 0.40486220535742995,
    "step": 12590
  },
  {
    "loss": 0.08911849856376648,
    "grad_norm": 0.7210829257965088,
    "learning_rate": 0.0001299342846375661,
    "epoch": 0.4051837797858314,
    "step": 12600
  },
  {
    "loss": 0.07384041547775269,
    "grad_norm": 0.3820243775844574,
    "learning_rate": 0.00012983756518367652,
    "epoch": 0.4055053542142329,
    "step": 12610
  },
  {
    "loss": 0.08409932255744934,
    "grad_norm": 0.483556866645813,
    "learning_rate": 0.00012974081508026095,
    "epoch": 0.4058269286426343,
    "step": 12620
  },
  {
    "loss": 0.0932869553565979,
    "grad_norm": 0.983331024646759,
    "learning_rate": 0.00012964403442670231,
    "epoch": 0.4061485030710358,
    "step": 12630
  },
  {
    "loss": 0.09054937958717346,
    "grad_norm": 0.399808406829834,
    "learning_rate": 0.00012954722332241496,
    "epoch": 0.40647007749943725,
    "step": 12640
  },
  {
    "loss": 0.09392053484916688,
    "grad_norm": 0.38033151626586914,
    "learning_rate": 0.0001294503818668444,
    "epoch": 0.40679165192783867,
    "step": 12650
  },
  {
    "loss": 0.07037632465362549,
    "grad_norm": 0.3877726197242737,
    "learning_rate": 0.00012935351015946756,
    "epoch": 0.40711322635624014,
    "step": 12660
  },
  {
    "loss": 0.08089154958724976,
    "grad_norm": 0.4302811026573181,
    "learning_rate": 0.00012925660829979214,
    "epoch": 0.4074348007846416,
    "step": 12670
  },
  {
    "loss": 0.08582212328910828,
    "grad_norm": 0.7823520302772522,
    "learning_rate": 0.000129159676387357,
    "epoch": 0.40775637521304303,
    "step": 12680
  },
  {
    "loss": 0.0841569185256958,
    "grad_norm": 0.5166428089141846,
    "learning_rate": 0.0001290627145217319,
    "epoch": 0.4080779496414445,
    "step": 12690
  },
  {
    "loss": 0.09074650406837463,
    "grad_norm": 0.5572818517684937,
    "learning_rate": 0.00012896572280251723,
    "epoch": 0.408399524069846,
    "step": 12700
  },
  {
    "loss": 0.09654123783111572,
    "grad_norm": 0.9365857839584351,
    "learning_rate": 0.0001288687013293441,
    "epoch": 0.4087210984982474,
    "step": 12710
  },
  {
    "loss": 0.10211057662963867,
    "grad_norm": 0.7367502450942993,
    "learning_rate": 0.00012877165020187427,
    "epoch": 0.40904267292664886,
    "step": 12720
  },
  {
    "loss": 0.09811394214630127,
    "grad_norm": 0.9345642924308777,
    "learning_rate": 0.00012867456951979985,
    "epoch": 0.40936424735505034,
    "step": 12730
  },
  {
    "loss": 0.08300668001174927,
    "grad_norm": 0.5749287009239197,
    "learning_rate": 0.00012857745938284337,
    "epoch": 0.40968582178345175,
    "step": 12740
  },
  {
    "loss": 0.08812336921691895,
    "grad_norm": 1.3258109092712402,
    "learning_rate": 0.00012848031989075754,
    "epoch": 0.4100073962118532,
    "step": 12750
  },
  {
    "loss": 0.10080107450485229,
    "grad_norm": 0.4471311867237091,
    "learning_rate": 0.00012838315114332532,
    "epoch": 0.4103289706402547,
    "step": 12760
  },
  {
    "loss": 0.09509003758430482,
    "grad_norm": 0.6037778854370117,
    "learning_rate": 0.00012828595324035976,
    "epoch": 0.4106505450686561,
    "step": 12770
  },
  {
    "loss": 0.08809060454368592,
    "grad_norm": 0.37151727080345154,
    "learning_rate": 0.00012818872628170361,
    "epoch": 0.4109721194970576,
    "step": 12780
  },
  {
    "loss": 0.09643104076385497,
    "grad_norm": 0.492166668176651,
    "learning_rate": 0.0001280914703672298,
    "epoch": 0.41129369392545906,
    "step": 12790
  },
  {
    "loss": 0.08205317258834839,
    "grad_norm": 0.621078372001648,
    "learning_rate": 0.00012799418559684072,
    "epoch": 0.4116152683538605,
    "step": 12800
  },
  {
    "loss": 0.09125524759292603,
    "grad_norm": 0.31062090396881104,
    "learning_rate": 0.0001278968720704686,
    "epoch": 0.41193684278226195,
    "step": 12810
  },
  {
    "loss": 0.11376011371612549,
    "grad_norm": 0.8067668080329895,
    "learning_rate": 0.00012779952988807512,
    "epoch": 0.4122584172106634,
    "step": 12820
  },
  {
    "loss": 0.0906079888343811,
    "grad_norm": 0.9971017241477966,
    "learning_rate": 0.00012770215914965138,
    "epoch": 0.41257999163906484,
    "step": 12830
  },
  {
    "loss": 0.097061026096344,
    "grad_norm": 0.6321334838867188,
    "learning_rate": 0.00012760475995521783,
    "epoch": 0.4129015660674663,
    "step": 12840
  },
  {
    "loss": 0.104645037651062,
    "grad_norm": 0.6004388928413391,
    "learning_rate": 0.0001275073324048242,
    "epoch": 0.4132231404958678,
    "step": 12850
  },
  {
    "loss": 0.09864118695259094,
    "grad_norm": 0.6671878695487976,
    "learning_rate": 0.0001274098765985493,
    "epoch": 0.4135447149242692,
    "step": 12860
  },
  {
    "loss": 0.10216143131256103,
    "grad_norm": 0.3585256040096283,
    "learning_rate": 0.00012731239263650095,
    "epoch": 0.41386628935267067,
    "step": 12870
  },
  {
    "loss": 0.08778411149978638,
    "grad_norm": 0.5079051852226257,
    "learning_rate": 0.0001272148806188159,
    "epoch": 0.41418786378107214,
    "step": 12880
  },
  {
    "loss": 0.07836889028549195,
    "grad_norm": 0.449163556098938,
    "learning_rate": 0.0001271173406456598,
    "epoch": 0.41450943820947356,
    "step": 12890
  },
  {
    "loss": 0.07983227968215942,
    "grad_norm": 0.5763904452323914,
    "learning_rate": 0.00012701977281722688,
    "epoch": 0.41483101263787503,
    "step": 12900
  },
  {
    "loss": 0.09611907601356506,
    "grad_norm": 0.617703914642334,
    "learning_rate": 0.00012692217723374007,
    "epoch": 0.4151525870662765,
    "step": 12910
  },
  {
    "loss": 0.09903550148010254,
    "grad_norm": 0.8212956786155701,
    "learning_rate": 0.00012682455399545082,
    "epoch": 0.4154741614946779,
    "step": 12920
  },
  {
    "loss": 0.09544740319252014,
    "grad_norm": 0.4063360095024109,
    "learning_rate": 0.00012672690320263895,
    "epoch": 0.4157957359230794,
    "step": 12930
  },
  {
    "loss": 0.101250159740448,
    "grad_norm": 0.8265182971954346,
    "learning_rate": 0.00012662922495561258,
    "epoch": 0.41611731035148086,
    "step": 12940
  },
  {
    "loss": 0.09107673168182373,
    "grad_norm": 0.5787447690963745,
    "learning_rate": 0.00012653151935470804,
    "epoch": 0.4164388847798823,
    "step": 12950
  },
  {
    "loss": 0.07801653146743774,
    "grad_norm": 0.5555682182312012,
    "learning_rate": 0.0001264337865002898,
    "epoch": 0.41676045920828375,
    "step": 12960
  },
  {
    "loss": 0.10017772912979125,
    "grad_norm": 0.634788453578949,
    "learning_rate": 0.00012633602649275027,
    "epoch": 0.4170820336366852,
    "step": 12970
  },
  {
    "loss": 0.11236907243728637,
    "grad_norm": 0.6519593000411987,
    "learning_rate": 0.00012623823943250977,
    "epoch": 0.41740360806508664,
    "step": 12980
  },
  {
    "loss": 0.10066357851028443,
    "grad_norm": 1.0568861961364746,
    "learning_rate": 0.00012614042542001642,
    "epoch": 0.4177251824934881,
    "step": 12990
  },
  {
    "loss": 0.07760431170463562,
    "grad_norm": 0.6190695762634277,
    "learning_rate": 0.00012604258455574597,
    "epoch": 0.4180467569218896,
    "step": 13000
  },
  {
    "loss": 0.09244945049285888,
    "grad_norm": 0.5714675784111023,
    "learning_rate": 0.00012594471694020185,
    "epoch": 0.418368331350291,
    "step": 13010
  },
  {
    "loss": 0.09857093691825866,
    "grad_norm": 0.498703271150589,
    "learning_rate": 0.0001258468226739149,
    "epoch": 0.4186899057786925,
    "step": 13020
  },
  {
    "loss": 0.0735785722732544,
    "grad_norm": 0.5419074892997742,
    "learning_rate": 0.00012574890185744337,
    "epoch": 0.41901148020709394,
    "step": 13030
  },
  {
    "loss": 0.10949951410293579,
    "grad_norm": 0.49285179376602173,
    "learning_rate": 0.00012565095459137274,
    "epoch": 0.41933305463549536,
    "step": 13040
  },
  {
    "loss": 0.09090704917907715,
    "grad_norm": 0.5541001558303833,
    "learning_rate": 0.00012555298097631572,
    "epoch": 0.41965462906389683,
    "step": 13050
  },
  {
    "loss": 0.0569901168346405,
    "grad_norm": 0.5824894905090332,
    "learning_rate": 0.00012545498111291202,
    "epoch": 0.4199762034922983,
    "step": 13060
  },
  {
    "loss": 0.08063774108886719,
    "grad_norm": 0.6675398945808411,
    "learning_rate": 0.00012535695510182838,
    "epoch": 0.4202977779206997,
    "step": 13070
  },
  {
    "loss": 0.09944126009941101,
    "grad_norm": 0.7908313870429993,
    "learning_rate": 0.00012525890304375831,
    "epoch": 0.4206193523491012,
    "step": 13080
  },
  {
    "loss": 0.07430846691131592,
    "grad_norm": 0.5902954936027527,
    "learning_rate": 0.0001251608250394222,
    "epoch": 0.42094092677750267,
    "step": 13090
  },
  {
    "loss": 0.09310410618782043,
    "grad_norm": 0.8124235272407532,
    "learning_rate": 0.00012506272118956692,
    "epoch": 0.4212625012059041,
    "step": 13100
  },
  {
    "loss": 0.09318177700042725,
    "grad_norm": 0.7630248069763184,
    "learning_rate": 0.00012496459159496608,
    "epoch": 0.42158407563430556,
    "step": 13110
  },
  {
    "loss": 0.07195741534233094,
    "grad_norm": 0.5920671820640564,
    "learning_rate": 0.00012486643635641965,
    "epoch": 0.421905650062707,
    "step": 13120
  },
  {
    "loss": 0.10478745698928833,
    "grad_norm": 0.9728026986122131,
    "learning_rate": 0.00012476825557475386,
    "epoch": 0.42222722449110844,
    "step": 13130
  },
  {
    "loss": 0.09816946983337402,
    "grad_norm": 0.5509280562400818,
    "learning_rate": 0.00012467004935082135,
    "epoch": 0.4225487989195099,
    "step": 13140
  },
  {
    "loss": 0.09047418236732482,
    "grad_norm": 0.5737391114234924,
    "learning_rate": 0.00012457181778550072,
    "epoch": 0.4228703733479114,
    "step": 13150
  },
  {
    "loss": 0.09822179675102234,
    "grad_norm": 0.8015565872192383,
    "learning_rate": 0.00012447356097969672,
    "epoch": 0.4231919477763128,
    "step": 13160
  },
  {
    "loss": 0.11566193103790283,
    "grad_norm": 0.6863377094268799,
    "learning_rate": 0.00012437527903433996,
    "epoch": 0.4235135222047143,
    "step": 13170
  },
  {
    "loss": 0.09870879650115967,
    "grad_norm": 0.6073296666145325,
    "learning_rate": 0.00012427697205038697,
    "epoch": 0.42383509663311575,
    "step": 13180
  },
  {
    "loss": 0.1031153678894043,
    "grad_norm": 1.089884877204895,
    "learning_rate": 0.00012417864012881987,
    "epoch": 0.42415667106151717,
    "step": 13190
  },
  {
    "loss": 0.07383702993392945,
    "grad_norm": 0.6829261779785156,
    "learning_rate": 0.0001240802833706465,
    "epoch": 0.42447824548991864,
    "step": 13200
  },
  {
    "loss": 0.08646566271781922,
    "grad_norm": 0.5114777684211731,
    "learning_rate": 0.00012398190187690012,
    "epoch": 0.4247998199183201,
    "step": 13210
  },
  {
    "loss": 0.09108679294586182,
    "grad_norm": 0.7739109396934509,
    "learning_rate": 0.00012388349574863952,
    "epoch": 0.4251213943467215,
    "step": 13220
  },
  {
    "loss": 0.09738523364067078,
    "grad_norm": 0.5807838439941406,
    "learning_rate": 0.00012378506508694864,
    "epoch": 0.425442968775123,
    "step": 13230
  },
  {
    "loss": 0.09927533268928528,
    "grad_norm": 0.6175647974014282,
    "learning_rate": 0.00012368660999293683,
    "epoch": 0.42576454320352447,
    "step": 13240
  },
  {
    "loss": 0.08608570694923401,
    "grad_norm": 1.083479881286621,
    "learning_rate": 0.00012358813056773826,
    "epoch": 0.4260861176319259,
    "step": 13250
  },
  {
    "loss": 0.07202335000038147,
    "grad_norm": 0.6211294531822205,
    "learning_rate": 0.00012348962691251234,
    "epoch": 0.42640769206032736,
    "step": 13260
  },
  {
    "loss": 0.11884490251541138,
    "grad_norm": 0.4971469044685364,
    "learning_rate": 0.00012339109912844326,
    "epoch": 0.42672926648872883,
    "step": 13270
  },
  {
    "loss": 0.06431439518928528,
    "grad_norm": 0.6088893413543701,
    "learning_rate": 0.00012329254731674,
    "epoch": 0.42705084091713025,
    "step": 13280
  },
  {
    "loss": 0.0804969072341919,
    "grad_norm": 0.5218471884727478,
    "learning_rate": 0.00012319397157863627,
    "epoch": 0.4273724153455317,
    "step": 13290
  },
  {
    "loss": 0.07834722995758056,
    "grad_norm": 0.6221503615379333,
    "learning_rate": 0.00012309537201539027,
    "epoch": 0.4276939897739332,
    "step": 13300
  },
  {
    "loss": 0.09761518239974976,
    "grad_norm": 0.7556706070899963,
    "learning_rate": 0.00012299674872828477,
    "epoch": 0.4280155642023346,
    "step": 13310
  },
  {
    "loss": 0.08869344592094422,
    "grad_norm": 0.6085411906242371,
    "learning_rate": 0.0001228981018186268,
    "epoch": 0.4283371386307361,
    "step": 13320
  },
  {
    "loss": 0.07709804773330689,
    "grad_norm": 0.9054849147796631,
    "learning_rate": 0.00012279943138774778,
    "epoch": 0.42865871305913755,
    "step": 13330
  },
  {
    "loss": 0.08806559443473816,
    "grad_norm": 0.48757463693618774,
    "learning_rate": 0.0001227007375370032,
    "epoch": 0.42898028748753897,
    "step": 13340
  },
  {
    "loss": 0.09110906720161438,
    "grad_norm": 0.6032870411872864,
    "learning_rate": 0.00012260202036777268,
    "epoch": 0.42930186191594044,
    "step": 13350
  },
  {
    "loss": 0.08105543255805969,
    "grad_norm": 0.6139655709266663,
    "learning_rate": 0.0001225032799814597,
    "epoch": 0.4296234363443419,
    "step": 13360
  },
  {
    "loss": 0.10194287300109864,
    "grad_norm": 0.4390583336353302,
    "learning_rate": 0.00012240451647949164,
    "epoch": 0.42994501077274333,
    "step": 13370
  },
  {
    "loss": 0.07526599764823913,
    "grad_norm": 0.4364306926727295,
    "learning_rate": 0.00012230572996331965,
    "epoch": 0.4302665852011448,
    "step": 13380
  },
  {
    "loss": 0.09649503231048584,
    "grad_norm": 0.5593969821929932,
    "learning_rate": 0.0001222069205344185,
    "epoch": 0.4305881596295463,
    "step": 13390
  },
  {
    "loss": 0.06970558166503907,
    "grad_norm": 0.40434157848358154,
    "learning_rate": 0.00012210808829428644,
    "epoch": 0.4309097340579477,
    "step": 13400
  },
  {
    "loss": 0.0974462866783142,
    "grad_norm": 0.7777182459831238,
    "learning_rate": 0.00012200923334444522,
    "epoch": 0.43123130848634916,
    "step": 13410
  },
  {
    "loss": 0.09920287132263184,
    "grad_norm": 0.644882082939148,
    "learning_rate": 0.00012191035578643997,
    "epoch": 0.43155288291475064,
    "step": 13420
  },
  {
    "loss": 0.09246551394462585,
    "grad_norm": 0.9769138097763062,
    "learning_rate": 0.00012181145572183889,
    "epoch": 0.43187445734315205,
    "step": 13430
  },
  {
    "loss": 0.08350251913070679,
    "grad_norm": 0.3481636047363281,
    "learning_rate": 0.00012171253325223344,
    "epoch": 0.4321960317715535,
    "step": 13440
  },
  {
    "loss": 0.06871581673622132,
    "grad_norm": 0.5685707926750183,
    "learning_rate": 0.00012161358847923801,
    "epoch": 0.432517606199955,
    "step": 13450
  },
  {
    "loss": 0.09470840096473694,
    "grad_norm": 0.5111106038093567,
    "learning_rate": 0.00012151462150448991,
    "epoch": 0.4328391806283564,
    "step": 13460
  },
  {
    "loss": 0.08720097541809083,
    "grad_norm": 0.7694170475006104,
    "learning_rate": 0.00012141563242964929,
    "epoch": 0.4331607550567579,
    "step": 13470
  },
  {
    "loss": 0.081611967086792,
    "grad_norm": 0.5865517854690552,
    "learning_rate": 0.00012131662135639898,
    "epoch": 0.43348232948515936,
    "step": 13480
  },
  {
    "loss": 0.09730910658836364,
    "grad_norm": 0.7233736515045166,
    "learning_rate": 0.00012121758838644444,
    "epoch": 0.4338039039135608,
    "step": 13490
  },
  {
    "loss": 0.08976944684982299,
    "grad_norm": 0.6866528987884521,
    "learning_rate": 0.00012111853362151357,
    "epoch": 0.43412547834196225,
    "step": 13500
  },
  {
    "loss": 0.10347449779510498,
    "grad_norm": 0.5420098900794983,
    "learning_rate": 0.00012101945716335668,
    "epoch": 0.4344470527703637,
    "step": 13510
  },
  {
    "loss": 0.08698602318763733,
    "grad_norm": 0.2645670771598816,
    "learning_rate": 0.00012092035911374639,
    "epoch": 0.43476862719876513,
    "step": 13520
  },
  {
    "loss": 0.08184699416160583,
    "grad_norm": 0.564333975315094,
    "learning_rate": 0.00012082123957447744,
    "epoch": 0.4350902016271666,
    "step": 13530
  },
  {
    "loss": 0.09781272411346435,
    "grad_norm": 0.6272760629653931,
    "learning_rate": 0.0001207220986473667,
    "epoch": 0.4354117760555681,
    "step": 13540
  },
  {
    "loss": 0.08091724514961243,
    "grad_norm": 0.5868409872055054,
    "learning_rate": 0.00012062293643425301,
    "epoch": 0.4357333504839695,
    "step": 13550
  },
  {
    "loss": 0.11358673572540283,
    "grad_norm": 0.9962054491043091,
    "learning_rate": 0.00012052375303699703,
    "epoch": 0.43605492491237097,
    "step": 13560
  },
  {
    "loss": 0.0930856466293335,
    "grad_norm": 0.641973614692688,
    "learning_rate": 0.00012042454855748118,
    "epoch": 0.43637649934077244,
    "step": 13570
  },
  {
    "loss": 0.10741612911224366,
    "grad_norm": 0.7930047512054443,
    "learning_rate": 0.00012032532309760962,
    "epoch": 0.43669807376917386,
    "step": 13580
  },
  {
    "loss": 0.08490241765975952,
    "grad_norm": 0.5818084478378296,
    "learning_rate": 0.00012022607675930795,
    "epoch": 0.43701964819757533,
    "step": 13590
  },
  {
    "loss": 0.11442885398864747,
    "grad_norm": 0.4617845416069031,
    "learning_rate": 0.00012012680964452332,
    "epoch": 0.4373412226259768,
    "step": 13600
  },
  {
    "loss": 0.09673363566398621,
    "grad_norm": 0.7523597478866577,
    "learning_rate": 0.00012002752185522412,
    "epoch": 0.4376627970543782,
    "step": 13610
  },
  {
    "loss": 0.08678258061408997,
    "grad_norm": 0.5370033383369446,
    "learning_rate": 0.00011992821349340009,
    "epoch": 0.4379843714827797,
    "step": 13620
  },
  {
    "loss": 0.11078073978424072,
    "grad_norm": 0.42315545678138733,
    "learning_rate": 0.00011982888466106197,
    "epoch": 0.43830594591118116,
    "step": 13630
  },
  {
    "loss": 0.0647172212600708,
    "grad_norm": 0.5130701661109924,
    "learning_rate": 0.00011972953546024164,
    "epoch": 0.4386275203395826,
    "step": 13640
  },
  {
    "loss": 0.0814866840839386,
    "grad_norm": 0.7799618244171143,
    "learning_rate": 0.00011963016599299184,
    "epoch": 0.43894909476798405,
    "step": 13650
  },
  {
    "loss": 0.08262754678726196,
    "grad_norm": 0.5943609476089478,
    "learning_rate": 0.0001195307763613862,
    "epoch": 0.4392706691963855,
    "step": 13660
  },
  {
    "loss": 0.0892327070236206,
    "grad_norm": 0.5623515844345093,
    "learning_rate": 0.00011943136666751895,
    "epoch": 0.43959224362478694,
    "step": 13670
  },
  {
    "loss": 0.0836462676525116,
    "grad_norm": 0.4980289936065674,
    "learning_rate": 0.00011933193701350501,
    "epoch": 0.4399138180531884,
    "step": 13680
  },
  {
    "loss": 0.07117368578910828,
    "grad_norm": 0.41098299622535706,
    "learning_rate": 0.00011923248750147979,
    "epoch": 0.4402353924815899,
    "step": 13690
  },
  {
    "loss": 0.08613250255584717,
    "grad_norm": 0.5188729166984558,
    "learning_rate": 0.0001191330182335991,
    "epoch": 0.4405569669099913,
    "step": 13700
  },
  {
    "loss": 0.10002766847610474,
    "grad_norm": 0.6116454601287842,
    "learning_rate": 0.00011903352931203903,
    "epoch": 0.44087854133839277,
    "step": 13710
  },
  {
    "loss": 0.08848408460617066,
    "grad_norm": 0.8638411164283752,
    "learning_rate": 0.00011893402083899588,
    "epoch": 0.44120011576679424,
    "step": 13720
  },
  {
    "loss": 0.09257006645202637,
    "grad_norm": 0.5978802442550659,
    "learning_rate": 0.00011883449291668594,
    "epoch": 0.44152169019519566,
    "step": 13730
  },
  {
    "loss": 0.09721587300300598,
    "grad_norm": 0.47306108474731445,
    "learning_rate": 0.00011873494564734566,
    "epoch": 0.44184326462359713,
    "step": 13740
  },
  {
    "loss": 0.07312923073768615,
    "grad_norm": 0.34300699830055237,
    "learning_rate": 0.00011863537913323119,
    "epoch": 0.4421648390519986,
    "step": 13750
  },
  {
    "loss": 0.08672202825546264,
    "grad_norm": 0.8224014639854431,
    "learning_rate": 0.00011853579347661856,
    "epoch": 0.4424864134804,
    "step": 13760
  },
  {
    "loss": 0.07279933094978333,
    "grad_norm": 0.8059155941009521,
    "learning_rate": 0.0001184361887798034,
    "epoch": 0.4428079879088015,
    "step": 13770
  },
  {
    "loss": 0.09367861747741699,
    "grad_norm": 0.7235598564147949,
    "learning_rate": 0.00011833656514510094,
    "epoch": 0.44312956233720296,
    "step": 13780
  },
  {
    "loss": 0.08866561055183411,
    "grad_norm": 0.34332987666130066,
    "learning_rate": 0.00011823692267484581,
    "epoch": 0.4434511367656044,
    "step": 13790
  },
  {
    "loss": 0.097174733877182,
    "grad_norm": 0.6409239768981934,
    "learning_rate": 0.00011813726147139201,
    "epoch": 0.44377271119400585,
    "step": 13800
  },
  {
    "loss": 0.06715927720069885,
    "grad_norm": 0.6418638825416565,
    "learning_rate": 0.00011803758163711287,
    "epoch": 0.4440942856224073,
    "step": 13810
  },
  {
    "loss": 0.11879804134368896,
    "grad_norm": 0.533685564994812,
    "learning_rate": 0.00011793788327440072,
    "epoch": 0.44441586005080874,
    "step": 13820
  },
  {
    "loss": 0.07027478814125061,
    "grad_norm": 0.4666531980037689,
    "learning_rate": 0.00011783816648566697,
    "epoch": 0.4447374344792102,
    "step": 13830
  },
  {
    "loss": 0.08207365274429321,
    "grad_norm": 0.32438936829566956,
    "learning_rate": 0.00011773843137334201,
    "epoch": 0.4450590089076117,
    "step": 13840
  },
  {
    "loss": 0.07866615056991577,
    "grad_norm": 0.566194474697113,
    "learning_rate": 0.00011763867803987505,
    "epoch": 0.4453805833360131,
    "step": 13850
  },
  {
    "loss": 0.08325292468070984,
    "grad_norm": 0.7310763597488403,
    "learning_rate": 0.0001175389065877339,
    "epoch": 0.4457021577644146,
    "step": 13860
  },
  {
    "loss": 0.10103902816772461,
    "grad_norm": 0.4220094382762909,
    "learning_rate": 0.0001174391171194051,
    "epoch": 0.44602373219281605,
    "step": 13870
  },
  {
    "loss": 0.07679594159126282,
    "grad_norm": 0.6618533134460449,
    "learning_rate": 0.00011733930973739366,
    "epoch": 0.44634530662121746,
    "step": 13880
  },
  {
    "loss": 0.10045582056045532,
    "grad_norm": 0.7397884130477905,
    "learning_rate": 0.00011723948454422298,
    "epoch": 0.44666688104961894,
    "step": 13890
  },
  {
    "loss": 0.086998850107193,
    "grad_norm": 0.628219485282898,
    "learning_rate": 0.0001171396416424348,
    "epoch": 0.4469884554780204,
    "step": 13900
  },
  {
    "loss": 0.08655040264129639,
    "grad_norm": 0.8280029296875,
    "learning_rate": 0.00011703978113458898,
    "epoch": 0.4473100299064218,
    "step": 13910
  },
  {
    "loss": 0.08575885891914367,
    "grad_norm": 0.538664698600769,
    "learning_rate": 0.00011693990312326352,
    "epoch": 0.4476316043348233,
    "step": 13920
  },
  {
    "loss": 0.07931332588195801,
    "grad_norm": 0.43020099401474,
    "learning_rate": 0.00011684000771105438,
    "epoch": 0.44795317876322477,
    "step": 13930
  },
  {
    "loss": 0.09089007377624511,
    "grad_norm": 1.0443196296691895,
    "learning_rate": 0.00011674009500057541,
    "epoch": 0.4482747531916262,
    "step": 13940
  },
  {
    "loss": 0.09716281890869141,
    "grad_norm": 0.9259188175201416,
    "learning_rate": 0.00011664016509445819,
    "epoch": 0.44859632762002766,
    "step": 13950
  },
  {
    "loss": 0.08684880137443543,
    "grad_norm": 0.44154876470565796,
    "learning_rate": 0.00011654021809535196,
    "epoch": 0.44891790204842913,
    "step": 13960
  },
  {
    "loss": 0.08301849961280823,
    "grad_norm": 0.7156301736831665,
    "learning_rate": 0.00011644025410592366,
    "epoch": 0.44923947647683055,
    "step": 13970
  },
  {
    "loss": 0.07928972244262696,
    "grad_norm": 0.6469382047653198,
    "learning_rate": 0.00011634027322885743,
    "epoch": 0.449561050905232,
    "step": 13980
  },
  {
    "eval_loss": 0.08378300815820694,
    "eval_runtime": 34.4401,
    "eval_samples_per_second": 145.935,
    "eval_steps_per_second": 36.498,
    "epoch": 0.4497539955622729,
    "step": 13986
  },
  {
    "loss": 0.07586904764175414,
    "grad_norm": 0.6927252411842346,
    "learning_rate": 0.00011624027556685498,
    "epoch": 0.4498826253336335,
    "step": 13990
  },
  {
    "loss": 0.1109076738357544,
    "grad_norm": 0.6233476996421814,
    "learning_rate": 0.00011614026122263517,
    "epoch": 0.4502041997620349,
    "step": 14000
  },
  {
    "loss": 0.08457133173942566,
    "grad_norm": 0.658160924911499,
    "learning_rate": 0.00011604023029893398,
    "epoch": 0.4505257741904364,
    "step": 14010
  },
  {
    "loss": 0.08707954287528992,
    "grad_norm": 0.7234919667243958,
    "learning_rate": 0.00011594018289850448,
    "epoch": 0.45084734861883785,
    "step": 14020
  },
  {
    "loss": 0.07688099145889282,
    "grad_norm": 0.5437519550323486,
    "learning_rate": 0.0001158401191241166,
    "epoch": 0.45116892304723927,
    "step": 14030
  },
  {
    "loss": 0.09311116337776185,
    "grad_norm": 0.8065460920333862,
    "learning_rate": 0.00011574003907855715,
    "epoch": 0.45149049747564074,
    "step": 14040
  },
  {
    "loss": 0.0929793119430542,
    "grad_norm": 0.8462100028991699,
    "learning_rate": 0.00011563994286462957,
    "epoch": 0.4518120719040422,
    "step": 14050
  },
  {
    "loss": 0.0810061514377594,
    "grad_norm": 0.8836320042610168,
    "learning_rate": 0.00011553983058515405,
    "epoch": 0.45213364633244363,
    "step": 14060
  },
  {
    "loss": 0.07508513331413269,
    "grad_norm": 1.3527172803878784,
    "learning_rate": 0.0001154397023429671,
    "epoch": 0.4524552207608451,
    "step": 14070
  },
  {
    "loss": 0.08644511103630066,
    "grad_norm": 0.47087258100509644,
    "learning_rate": 0.00011533955824092179,
    "epoch": 0.4527767951892466,
    "step": 14080
  },
  {
    "loss": 0.10185540914535522,
    "grad_norm": 0.7062323093414307,
    "learning_rate": 0.0001152393983818874,
    "epoch": 0.453098369617648,
    "step": 14090
  },
  {
    "loss": 0.07294328212738037,
    "grad_norm": 0.5222784280776978,
    "learning_rate": 0.00011513922286874941,
    "epoch": 0.45341994404604946,
    "step": 14100
  },
  {
    "loss": 0.10024436712265014,
    "grad_norm": 0.5364047884941101,
    "learning_rate": 0.00011503903180440932,
    "epoch": 0.45374151847445093,
    "step": 14110
  },
  {
    "loss": 0.09642171263694763,
    "grad_norm": 0.9905641078948975,
    "learning_rate": 0.00011493882529178477,
    "epoch": 0.45406309290285235,
    "step": 14120
  },
  {
    "loss": 0.10040140151977539,
    "grad_norm": 0.5326999425888062,
    "learning_rate": 0.00011483860343380908,
    "epoch": 0.4543846673312538,
    "step": 14130
  },
  {
    "loss": 0.09813051223754883,
    "grad_norm": 0.5978994965553284,
    "learning_rate": 0.00011473836633343144,
    "epoch": 0.4547062417596553,
    "step": 14140
  },
  {
    "loss": 0.09117928147315979,
    "grad_norm": 0.6803304553031921,
    "learning_rate": 0.00011463811409361667,
    "epoch": 0.4550278161880567,
    "step": 14150
  },
  {
    "loss": 0.07710156440734864,
    "grad_norm": 0.6730096340179443,
    "learning_rate": 0.00011453784681734516,
    "epoch": 0.4553493906164582,
    "step": 14160
  },
  {
    "loss": 0.10482146739959716,
    "grad_norm": 0.8572554588317871,
    "learning_rate": 0.00011443756460761273,
    "epoch": 0.45567096504485965,
    "step": 14170
  },
  {
    "loss": 0.09611796736717224,
    "grad_norm": 1.1262205839157104,
    "learning_rate": 0.0001143372675674305,
    "epoch": 0.45599253947326107,
    "step": 14180
  },
  {
    "loss": 0.10889110565185547,
    "grad_norm": 0.39554649591445923,
    "learning_rate": 0.00011423695579982492,
    "epoch": 0.45631411390166254,
    "step": 14190
  },
  {
    "loss": 0.09329955577850342,
    "grad_norm": 0.35157912969589233,
    "learning_rate": 0.00011413662940783747,
    "epoch": 0.456635688330064,
    "step": 14200
  },
  {
    "loss": 0.08668434023857116,
    "grad_norm": 0.5843801498413086,
    "learning_rate": 0.0001140362884945247,
    "epoch": 0.45695726275846543,
    "step": 14210
  },
  {
    "loss": 0.08654572367668152,
    "grad_norm": 0.6317597031593323,
    "learning_rate": 0.00011393593316295809,
    "epoch": 0.4572788371868669,
    "step": 14220
  },
  {
    "loss": 0.0918462872505188,
    "grad_norm": 0.428110271692276,
    "learning_rate": 0.00011383556351622393,
    "epoch": 0.4576004116152684,
    "step": 14230
  },
  {
    "loss": 0.10685827732086181,
    "grad_norm": 0.5201165080070496,
    "learning_rate": 0.00011373517965742313,
    "epoch": 0.4579219860436698,
    "step": 14240
  },
  {
    "loss": 0.08834422826766967,
    "grad_norm": 0.6804199814796448,
    "learning_rate": 0.00011363478168967132,
    "epoch": 0.45824356047207127,
    "step": 14250
  },
  {
    "loss": 0.09454160332679748,
    "grad_norm": 0.3859182894229889,
    "learning_rate": 0.00011353436971609854,
    "epoch": 0.45856513490047274,
    "step": 14260
  },
  {
    "loss": 0.08517266511917114,
    "grad_norm": 0.6662791967391968,
    "learning_rate": 0.00011343394383984925,
    "epoch": 0.45888670932887415,
    "step": 14270
  },
  {
    "loss": 0.09720152616500854,
    "grad_norm": 1.1802897453308105,
    "learning_rate": 0.00011333350416408222,
    "epoch": 0.4592082837572756,
    "step": 14280
  },
  {
    "loss": 0.10304139852523804,
    "grad_norm": 0.4195810854434967,
    "learning_rate": 0.00011323305079197033,
    "epoch": 0.4595298581856771,
    "step": 14290
  },
  {
    "loss": 0.10177587270736695,
    "grad_norm": 0.6575989723205566,
    "learning_rate": 0.00011313258382670055,
    "epoch": 0.4598514326140785,
    "step": 14300
  },
  {
    "loss": 0.07869631052017212,
    "grad_norm": 0.701774537563324,
    "learning_rate": 0.00011303210337147381,
    "epoch": 0.46017300704248,
    "step": 14310
  },
  {
    "loss": 0.1009220838546753,
    "grad_norm": 0.7061283588409424,
    "learning_rate": 0.00011293160952950496,
    "epoch": 0.46049458147088146,
    "step": 14320
  },
  {
    "loss": 0.0825681209564209,
    "grad_norm": 0.5821817517280579,
    "learning_rate": 0.00011283110240402252,
    "epoch": 0.4608161558992829,
    "step": 14330
  },
  {
    "loss": 0.07777943611145019,
    "grad_norm": 0.5349202752113342,
    "learning_rate": 0.0001127305820982687,
    "epoch": 0.46113773032768435,
    "step": 14340
  },
  {
    "loss": 0.10388762950897217,
    "grad_norm": 1.0531601905822754,
    "learning_rate": 0.00011263004871549923,
    "epoch": 0.4614593047560858,
    "step": 14350
  },
  {
    "loss": 0.0702982485294342,
    "grad_norm": 0.5619316697120667,
    "learning_rate": 0.00011252950235898328,
    "epoch": 0.46178087918448724,
    "step": 14360
  },
  {
    "loss": 0.07497292160987853,
    "grad_norm": 0.6734415292739868,
    "learning_rate": 0.00011242894313200332,
    "epoch": 0.4621024536128887,
    "step": 14370
  },
  {
    "loss": 0.09227219820022584,
    "grad_norm": 0.2237335443496704,
    "learning_rate": 0.00011232837113785508,
    "epoch": 0.4624240280412902,
    "step": 14380
  },
  {
    "loss": 0.07480285167694092,
    "grad_norm": 0.7597417235374451,
    "learning_rate": 0.00011222778647984744,
    "epoch": 0.4627456024696916,
    "step": 14390
  },
  {
    "loss": 0.09297425150871277,
    "grad_norm": 0.5696741342544556,
    "learning_rate": 0.00011212718926130215,
    "epoch": 0.46306717689809307,
    "step": 14400
  },
  {
    "loss": 0.09020763039588928,
    "grad_norm": 0.5711169838905334,
    "learning_rate": 0.000112026579585554,
    "epoch": 0.46338875132649454,
    "step": 14410
  },
  {
    "loss": 0.09613139629364013,
    "grad_norm": 0.5713536143302917,
    "learning_rate": 0.00011192595755595054,
    "epoch": 0.46371032575489596,
    "step": 14420
  },
  {
    "loss": 0.07961199283599854,
    "grad_norm": 0.7792538404464722,
    "learning_rate": 0.00011182532327585194,
    "epoch": 0.46403190018329743,
    "step": 14430
  },
  {
    "loss": 0.08362306952476502,
    "grad_norm": 0.4089360237121582,
    "learning_rate": 0.0001117246768486311,
    "epoch": 0.4643534746116989,
    "step": 14440
  },
  {
    "loss": 0.08947086334228516,
    "grad_norm": 0.8442490696907043,
    "learning_rate": 0.00011162401837767324,
    "epoch": 0.4646750490401003,
    "step": 14450
  },
  {
    "loss": 0.0991512417793274,
    "grad_norm": 0.5721593499183655,
    "learning_rate": 0.00011152334796637602,
    "epoch": 0.4649966234685018,
    "step": 14460
  },
  {
    "loss": 0.10386481285095214,
    "grad_norm": 0.9631685018539429,
    "learning_rate": 0.00011142266571814941,
    "epoch": 0.46531819789690326,
    "step": 14470
  },
  {
    "loss": 0.11372385025024415,
    "grad_norm": 0.6378226280212402,
    "learning_rate": 0.00011132197173641546,
    "epoch": 0.4656397723253047,
    "step": 14480
  },
  {
    "loss": 0.07930097579956055,
    "grad_norm": 0.39188480377197266,
    "learning_rate": 0.00011122126612460831,
    "epoch": 0.46596134675370615,
    "step": 14490
  },
  {
    "loss": 0.10634902715682984,
    "grad_norm": 0.7518141865730286,
    "learning_rate": 0.00011112054898617403,
    "epoch": 0.4662829211821076,
    "step": 14500
  },
  {
    "loss": 0.06619417667388916,
    "grad_norm": 0.3403424620628357,
    "learning_rate": 0.0001110198204245706,
    "epoch": 0.46660449561050904,
    "step": 14510
  },
  {
    "loss": 0.10738317966461182,
    "grad_norm": 0.7575116157531738,
    "learning_rate": 0.00011091908054326763,
    "epoch": 0.4669260700389105,
    "step": 14520
  },
  {
    "loss": 0.06974626183509827,
    "grad_norm": 0.22222907841205597,
    "learning_rate": 0.0001108183294457464,
    "epoch": 0.467247644467312,
    "step": 14530
  },
  {
    "loss": 0.07733258008956909,
    "grad_norm": 0.5152522921562195,
    "learning_rate": 0.00011071756723549973,
    "epoch": 0.4675692188957134,
    "step": 14540
  },
  {
    "loss": 0.10055925846099853,
    "grad_norm": 0.5919005274772644,
    "learning_rate": 0.00011061679401603189,
    "epoch": 0.4678907933241149,
    "step": 14550
  },
  {
    "loss": 0.10045595169067383,
    "grad_norm": 0.5742218494415283,
    "learning_rate": 0.00011051600989085831,
    "epoch": 0.46821236775251635,
    "step": 14560
  },
  {
    "loss": 0.09662904739379882,
    "grad_norm": 0.6789755821228027,
    "learning_rate": 0.00011041521496350582,
    "epoch": 0.46853394218091776,
    "step": 14570
  },
  {
    "loss": 0.10636229515075683,
    "grad_norm": 0.6205703020095825,
    "learning_rate": 0.0001103144093375122,
    "epoch": 0.46885551660931923,
    "step": 14580
  },
  {
    "loss": 0.09271013140678405,
    "grad_norm": 0.6065817475318909,
    "learning_rate": 0.00011021359311642629,
    "epoch": 0.4691770910377207,
    "step": 14590
  },
  {
    "loss": 0.10116171836853027,
    "grad_norm": 0.6617358326911926,
    "learning_rate": 0.00011011276640380776,
    "epoch": 0.4694986654661221,
    "step": 14600
  },
  {
    "loss": 0.07309399843215943,
    "grad_norm": 0.4628552496433258,
    "learning_rate": 0.00011001192930322715,
    "epoch": 0.4698202398945236,
    "step": 14610
  },
  {
    "loss": 0.09705093502998352,
    "grad_norm": 0.5517852902412415,
    "learning_rate": 0.00010991108191826553,
    "epoch": 0.47014181432292507,
    "step": 14620
  },
  {
    "loss": 0.079444420337677,
    "grad_norm": 0.5863767266273499,
    "learning_rate": 0.0001098102243525147,
    "epoch": 0.4704633887513265,
    "step": 14630
  },
  {
    "loss": 0.08762736916542054,
    "grad_norm": 0.8725037574768066,
    "learning_rate": 0.00010970935670957676,
    "epoch": 0.47078496317972796,
    "step": 14640
  },
  {
    "loss": 0.109092116355896,
    "grad_norm": 9.460938453674316,
    "learning_rate": 0.00010960847909306429,
    "epoch": 0.4711065376081294,
    "step": 14650
  },
  {
    "loss": 0.08770105242729187,
    "grad_norm": 0.46180495619773865,
    "learning_rate": 0.00010950759160660002,
    "epoch": 0.47142811203653084,
    "step": 14660
  },
  {
    "loss": 0.08387916088104248,
    "grad_norm": 0.7593671083450317,
    "learning_rate": 0.00010940669435381689,
    "epoch": 0.4717496864649323,
    "step": 14670
  },
  {
    "loss": 0.092861807346344,
    "grad_norm": 0.8851556777954102,
    "learning_rate": 0.00010930578743835782,
    "epoch": 0.4720712608933338,
    "step": 14680
  },
  {
    "loss": 0.0722186267375946,
    "grad_norm": 0.3864416480064392,
    "learning_rate": 0.00010920487096387564,
    "epoch": 0.4723928353217352,
    "step": 14690
  },
  {
    "loss": 0.0851593255996704,
    "grad_norm": 0.885635495185852,
    "learning_rate": 0.00010910394503403314,
    "epoch": 0.4727144097501367,
    "step": 14700
  },
  {
    "loss": 0.09186208248138428,
    "grad_norm": 0.5280024409294128,
    "learning_rate": 0.00010900300975250261,
    "epoch": 0.47303598417853815,
    "step": 14710
  },
  {
    "loss": 0.09017241597175599,
    "grad_norm": 0.3785780966281891,
    "learning_rate": 0.00010890206522296609,
    "epoch": 0.47335755860693957,
    "step": 14720
  },
  {
    "loss": 0.10533970594406128,
    "grad_norm": 0.47699448466300964,
    "learning_rate": 0.0001088011115491151,
    "epoch": 0.47367913303534104,
    "step": 14730
  },
  {
    "loss": 0.08337773680686951,
    "grad_norm": 0.5329119563102722,
    "learning_rate": 0.00010870014883465052,
    "epoch": 0.4740007074637425,
    "step": 14740
  },
  {
    "loss": 0.06845346093177795,
    "grad_norm": 0.6583190560340881,
    "learning_rate": 0.00010859917718328251,
    "epoch": 0.4743222818921439,
    "step": 14750
  },
  {
    "loss": 0.09752205610275269,
    "grad_norm": 0.6177473068237305,
    "learning_rate": 0.0001084981966987305,
    "epoch": 0.4746438563205454,
    "step": 14760
  },
  {
    "loss": 0.09779452085494995,
    "grad_norm": 0.5897741913795471,
    "learning_rate": 0.00010839720748472286,
    "epoch": 0.47496543074894687,
    "step": 14770
  },
  {
    "loss": 0.09048707485198974,
    "grad_norm": 1.006715178489685,
    "learning_rate": 0.00010829620964499702,
    "epoch": 0.4752870051773483,
    "step": 14780
  },
  {
    "loss": 0.08427447080612183,
    "grad_norm": 0.42974308133125305,
    "learning_rate": 0.00010819520328329927,
    "epoch": 0.47560857960574976,
    "step": 14790
  },
  {
    "loss": 0.08650519847869872,
    "grad_norm": 0.428631991147995,
    "learning_rate": 0.00010809418850338457,
    "epoch": 0.47593015403415123,
    "step": 14800
  },
  {
    "loss": 0.09579732418060302,
    "grad_norm": 0.6555226445198059,
    "learning_rate": 0.00010799316540901665,
    "epoch": 0.47625172846255265,
    "step": 14810
  },
  {
    "loss": 0.08336011171340943,
    "grad_norm": 0.41812366247177124,
    "learning_rate": 0.00010789213410396764,
    "epoch": 0.4765733028909541,
    "step": 14820
  },
  {
    "loss": 0.0817683219909668,
    "grad_norm": 0.9479015469551086,
    "learning_rate": 0.00010779109469201825,
    "epoch": 0.4768948773193556,
    "step": 14830
  },
  {
    "loss": 0.07006314992904664,
    "grad_norm": 0.6454442143440247,
    "learning_rate": 0.0001076900472769574,
    "epoch": 0.477216451747757,
    "step": 14840
  },
  {
    "loss": 0.10103509426116944,
    "grad_norm": 0.8812865614891052,
    "learning_rate": 0.00010758899196258228,
    "epoch": 0.4775380261761585,
    "step": 14850
  },
  {
    "loss": 0.07926849722862243,
    "grad_norm": 1.1739628314971924,
    "learning_rate": 0.00010748792885269827,
    "epoch": 0.47785960060455995,
    "step": 14860
  },
  {
    "loss": 0.10146574974060059,
    "grad_norm": 0.4796530604362488,
    "learning_rate": 0.00010738685805111859,
    "epoch": 0.47818117503296137,
    "step": 14870
  },
  {
    "loss": 0.09071105122566223,
    "grad_norm": 0.6305550336837769,
    "learning_rate": 0.00010728577966166447,
    "epoch": 0.47850274946136284,
    "step": 14880
  },
  {
    "loss": 0.10043426752090454,
    "grad_norm": 0.6661666035652161,
    "learning_rate": 0.00010718469378816491,
    "epoch": 0.47882432388976426,
    "step": 14890
  },
  {
    "loss": 0.09648953080177307,
    "grad_norm": 0.5236237049102783,
    "learning_rate": 0.00010708360053445665,
    "epoch": 0.47914589831816573,
    "step": 14900
  },
  {
    "loss": 0.08368160128593445,
    "grad_norm": 0.458296537399292,
    "learning_rate": 0.00010698250000438394,
    "epoch": 0.4794674727465672,
    "step": 14910
  },
  {
    "loss": 0.09150348901748658,
    "grad_norm": 0.5067198872566223,
    "learning_rate": 0.00010688139230179852,
    "epoch": 0.4797890471749686,
    "step": 14920
  },
  {
    "loss": 0.08629463911056519,
    "grad_norm": 0.8166804909706116,
    "learning_rate": 0.00010678027753055952,
    "epoch": 0.4801106216033701,
    "step": 14930
  },
  {
    "loss": 0.07823537588119507,
    "grad_norm": 0.5457392930984497,
    "learning_rate": 0.00010667915579453327,
    "epoch": 0.48043219603177156,
    "step": 14940
  },
  {
    "loss": 0.10616787672042846,
    "grad_norm": 0.9504504203796387,
    "learning_rate": 0.00010657802719759337,
    "epoch": 0.480753770460173,
    "step": 14950
  },
  {
    "loss": 0.09666532874107361,
    "grad_norm": 0.8659750819206238,
    "learning_rate": 0.00010647689184362037,
    "epoch": 0.48107534488857445,
    "step": 14960
  },
  {
    "loss": 0.0889670729637146,
    "grad_norm": 0.3620588481426239,
    "learning_rate": 0.00010637574983650178,
    "epoch": 0.4813969193169759,
    "step": 14970
  },
  {
    "loss": 0.08167270421981812,
    "grad_norm": 0.5511088371276855,
    "learning_rate": 0.00010627460128013199,
    "epoch": 0.48171849374537734,
    "step": 14980
  },
  {
    "loss": 0.08849920630455017,
    "grad_norm": 0.5692965388298035,
    "learning_rate": 0.00010617344627841204,
    "epoch": 0.4820400681737788,
    "step": 14990
  },
  {
    "loss": 0.0818578541278839,
    "grad_norm": 0.6101301312446594,
    "learning_rate": 0.00010607228493524963,
    "epoch": 0.4823616426021803,
    "step": 15000
  },
  {
    "loss": 0.09417021870613099,
    "grad_norm": 0.5213767886161804,
    "learning_rate": 0.00010597111735455902,
    "epoch": 0.4826832170305817,
    "step": 15010
  },
  {
    "loss": 0.07528434991836548,
    "grad_norm": 0.5721901655197144,
    "learning_rate": 0.00010586994364026082,
    "epoch": 0.4830047914589832,
    "step": 15020
  },
  {
    "loss": 0.093703293800354,
    "grad_norm": 0.4465864598751068,
    "learning_rate": 0.00010576876389628192,
    "epoch": 0.48332636588738465,
    "step": 15030
  },
  {
    "loss": 0.09364048838615417,
    "grad_norm": 0.6290184855461121,
    "learning_rate": 0.00010566757822655544,
    "epoch": 0.48364794031578606,
    "step": 15040
  },
  {
    "loss": 0.0754182755947113,
    "grad_norm": 0.4668210446834564,
    "learning_rate": 0.00010556638673502063,
    "epoch": 0.48396951474418753,
    "step": 15050
  },
  {
    "loss": 0.10114016532897949,
    "grad_norm": 0.7117626070976257,
    "learning_rate": 0.00010546518952562267,
    "epoch": 0.484291089172589,
    "step": 15060
  },
  {
    "loss": 0.0868488073348999,
    "grad_norm": 0.6074398159980774,
    "learning_rate": 0.0001053639867023126,
    "epoch": 0.4846126636009904,
    "step": 15070
  },
  {
    "loss": 0.07774009704589843,
    "grad_norm": 0.43255162239074707,
    "learning_rate": 0.00010526277836904726,
    "epoch": 0.4849342380293919,
    "step": 15080
  },
  {
    "loss": 0.07248947024345398,
    "grad_norm": 0.28224503993988037,
    "learning_rate": 0.0001051615646297891,
    "epoch": 0.48525581245779337,
    "step": 15090
  },
  {
    "loss": 0.09597593545913696,
    "grad_norm": 0.43093428015708923,
    "learning_rate": 0.00010506034558850618,
    "epoch": 0.4855773868861948,
    "step": 15100
  },
  {
    "loss": 0.09701937437057495,
    "grad_norm": 0.3181447684764862,
    "learning_rate": 0.00010495912134917199,
    "epoch": 0.48589896131459626,
    "step": 15110
  },
  {
    "loss": 0.08382670879364014,
    "grad_norm": 0.8677729964256287,
    "learning_rate": 0.00010485789201576539,
    "epoch": 0.48622053574299773,
    "step": 15120
  },
  {
    "loss": 0.08611851334571838,
    "grad_norm": 0.8189012408256531,
    "learning_rate": 0.00010475665769227034,
    "epoch": 0.48654211017139914,
    "step": 15130
  },
  {
    "loss": 0.09586833715438843,
    "grad_norm": 0.655185878276825,
    "learning_rate": 0.00010465541848267613,
    "epoch": 0.4868636845998006,
    "step": 15140
  },
  {
    "loss": 0.08906262516975402,
    "grad_norm": 0.6886798143386841,
    "learning_rate": 0.00010455417449097689,
    "epoch": 0.4871852590282021,
    "step": 15150
  },
  {
    "loss": 0.09160004258155822,
    "grad_norm": 0.44516444206237793,
    "learning_rate": 0.00010445292582117177,
    "epoch": 0.4875068334566035,
    "step": 15160
  },
  {
    "loss": 0.07538365125656128,
    "grad_norm": 0.743714451789856,
    "learning_rate": 0.00010435167257726468,
    "epoch": 0.487828407885005,
    "step": 15170
  },
  {
    "loss": 0.08665584921836852,
    "grad_norm": 0.6289533376693726,
    "learning_rate": 0.00010425041486326424,
    "epoch": 0.48814998231340645,
    "step": 15180
  },
  {
    "loss": 0.07832264304161071,
    "grad_norm": 0.6426935791969299,
    "learning_rate": 0.00010414915278318367,
    "epoch": 0.48847155674180787,
    "step": 15190
  },
  {
    "loss": 0.06755251884460449,
    "grad_norm": 0.5041143298149109,
    "learning_rate": 0.0001040478864410406,
    "epoch": 0.48879313117020934,
    "step": 15200
  },
  {
    "loss": 0.07771340012550354,
    "grad_norm": 0.42428064346313477,
    "learning_rate": 0.0001039466159408572,
    "epoch": 0.4891147055986108,
    "step": 15210
  },
  {
    "loss": 0.09802461862564087,
    "grad_norm": 0.9530168771743774,
    "learning_rate": 0.00010384534138665975,
    "epoch": 0.4894362800270122,
    "step": 15220
  },
  {
    "loss": 0.10731257200241089,
    "grad_norm": 0.7082659006118774,
    "learning_rate": 0.0001037440628824788,
    "epoch": 0.4897578544554137,
    "step": 15230
  },
  {
    "loss": 0.1051175832748413,
    "grad_norm": 0.7191018462181091,
    "learning_rate": 0.00010364278053234889,
    "epoch": 0.49007942888381517,
    "step": 15240
  },
  {
    "loss": 0.0677166998386383,
    "grad_norm": 0.49812623858451843,
    "learning_rate": 0.00010354149444030854,
    "epoch": 0.4904010033122166,
    "step": 15250
  },
  {
    "loss": 0.09486074447631836,
    "grad_norm": 1.0251941680908203,
    "learning_rate": 0.0001034402047104001,
    "epoch": 0.49072257774061806,
    "step": 15260
  },
  {
    "loss": 0.08186472058296204,
    "grad_norm": 0.5029348731040955,
    "learning_rate": 0.00010333891144666968,
    "epoch": 0.49104415216901953,
    "step": 15270
  },
  {
    "loss": 0.08887283802032471,
    "grad_norm": 0.6932932138442993,
    "learning_rate": 0.00010323761475316705,
    "epoch": 0.49136572659742095,
    "step": 15280
  },
  {
    "loss": 0.11510454416275025,
    "grad_norm": 0.8928279876708984,
    "learning_rate": 0.00010313631473394536,
    "epoch": 0.4916873010258224,
    "step": 15290
  },
  {
    "loss": 0.07902852892875671,
    "grad_norm": 0.6534966230392456,
    "learning_rate": 0.00010303501149306136,
    "epoch": 0.4920088754542239,
    "step": 15300
  },
  {
    "loss": 0.08356289267539978,
    "grad_norm": 0.7262250781059265,
    "learning_rate": 0.00010293370513457498,
    "epoch": 0.4923304498826253,
    "step": 15310
  },
  {
    "loss": 0.09315394163131714,
    "grad_norm": 1.0881065130233765,
    "learning_rate": 0.00010283239576254944,
    "epoch": 0.4926520243110268,
    "step": 15320
  },
  {
    "loss": 0.10244693756103515,
    "grad_norm": 0.7236805558204651,
    "learning_rate": 0.00010273108348105099,
    "epoch": 0.49297359873942825,
    "step": 15330
  },
  {
    "loss": 0.07832973003387451,
    "grad_norm": 0.48782211542129517,
    "learning_rate": 0.00010262976839414888,
    "epoch": 0.49329517316782967,
    "step": 15340
  },
  {
    "loss": 0.07148158550262451,
    "grad_norm": 0.5727787017822266,
    "learning_rate": 0.0001025284506059153,
    "epoch": 0.49361674759623114,
    "step": 15350
  },
  {
    "loss": 0.0850343406200409,
    "grad_norm": 0.7264379858970642,
    "learning_rate": 0.00010242713022042509,
    "epoch": 0.4939383220246326,
    "step": 15360
  },
  {
    "loss": 0.08559871315956116,
    "grad_norm": 0.626274049282074,
    "learning_rate": 0.00010232580734175591,
    "epoch": 0.49425989645303403,
    "step": 15370
  },
  {
    "loss": 0.07028576731681824,
    "grad_norm": 0.6762192845344543,
    "learning_rate": 0.0001022244820739879,
    "epoch": 0.4945814708814355,
    "step": 15380
  },
  {
    "loss": 0.09092403054237366,
    "grad_norm": 0.7768816947937012,
    "learning_rate": 0.00010212315452120362,
    "epoch": 0.494903045309837,
    "step": 15390
  },
  {
    "loss": 0.09877740144729615,
    "grad_norm": 0.8710550665855408,
    "learning_rate": 0.00010202182478748805,
    "epoch": 0.4952246197382384,
    "step": 15400
  },
  {
    "loss": 0.09296669363975525,
    "grad_norm": 0.5330905318260193,
    "learning_rate": 0.00010192049297692839,
    "epoch": 0.49554619416663986,
    "step": 15410
  },
  {
    "loss": 0.10949316024780273,
    "grad_norm": 0.6242778897285461,
    "learning_rate": 0.00010181915919361391,
    "epoch": 0.49586776859504134,
    "step": 15420
  },
  {
    "loss": 0.08684821128845215,
    "grad_norm": 0.5489434003829956,
    "learning_rate": 0.00010171782354163604,
    "epoch": 0.49618934302344275,
    "step": 15430
  },
  {
    "loss": 0.07282103300094604,
    "grad_norm": 0.3458095192909241,
    "learning_rate": 0.00010161648612508799,
    "epoch": 0.4965109174518442,
    "step": 15440
  },
  {
    "loss": 0.09404530525207519,
    "grad_norm": 0.5336048007011414,
    "learning_rate": 0.00010151514704806484,
    "epoch": 0.4968324918802457,
    "step": 15450
  },
  {
    "loss": 0.07916721701622009,
    "grad_norm": 1.0568702220916748,
    "learning_rate": 0.0001014138064146634,
    "epoch": 0.4971540663086471,
    "step": 15460
  },
  {
    "loss": 0.09599422812461852,
    "grad_norm": 0.5730618834495544,
    "learning_rate": 0.00010131246432898206,
    "epoch": 0.4974756407370486,
    "step": 15470
  },
  {
    "loss": 0.09638643264770508,
    "grad_norm": 0.6685811877250671,
    "learning_rate": 0.0001012111208951207,
    "epoch": 0.49779721516545006,
    "step": 15480
  },
  {
    "loss": 0.09276297092437744,
    "grad_norm": 0.2251046597957611,
    "learning_rate": 0.00010110977621718056,
    "epoch": 0.4981187895938515,
    "step": 15490
  },
  {
    "loss": 0.10188002586364746,
    "grad_norm": 0.6626012325286865,
    "learning_rate": 0.0001010084303992642,
    "epoch": 0.49844036402225295,
    "step": 15500
  },
  {
    "loss": 0.07401142120361329,
    "grad_norm": 0.5241210460662842,
    "learning_rate": 0.00010090708354547531,
    "epoch": 0.4987619384506544,
    "step": 15510
  },
  {
    "loss": 0.07829184532165527,
    "grad_norm": 0.4019498825073242,
    "learning_rate": 0.00010080573575991867,
    "epoch": 0.49908351287905584,
    "step": 15520
  },
  {
    "loss": 0.07195114493370056,
    "grad_norm": 0.4469619393348694,
    "learning_rate": 0.00010070438714670002,
    "epoch": 0.4994050873074573,
    "step": 15530
  },
  {
    "loss": 0.0959662914276123,
    "grad_norm": 0.4518377482891083,
    "learning_rate": 0.00010060303780992599,
    "epoch": 0.4997266617358588,
    "step": 15540
  },
  {
    "eval_loss": 0.0814763680100441,
    "eval_runtime": 34.9513,
    "eval_samples_per_second": 143.8,
    "eval_steps_per_second": 35.964,
    "epoch": 0.4997266617358588,
    "step": 15540
  },
  {
    "loss": 0.08837600946426391,
    "grad_norm": 0.2871614396572113,
    "learning_rate": 0.00010050168785370382,
    "epoch": 0.5000482361642602,
    "step": 15550
  },
  {
    "loss": 0.09808437824249268,
    "grad_norm": 0.546829104423523,
    "learning_rate": 0.00010040033738214154,
    "epoch": 0.5003698105926617,
    "step": 15560
  },
  {
    "loss": 0.06960054636001586,
    "grad_norm": 0.4438028335571289,
    "learning_rate": 0.00010029898649934759,
    "epoch": 0.5006913850210631,
    "step": 15570
  },
  {
    "loss": 0.09131770133972168,
    "grad_norm": 0.7185057401657104,
    "learning_rate": 0.00010019763530943092,
    "epoch": 0.5010129594494646,
    "step": 15580
  },
  {
    "loss": 0.0653073251247406,
    "grad_norm": 0.5891302227973938,
    "learning_rate": 0.00010009628391650077,
    "epoch": 0.5013345338778661,
    "step": 15590
  },
  {
    "loss": 0.09859957098960877,
    "grad_norm": 0.7047641277313232,
    "learning_rate": 9.999493242466654e-05,
    "epoch": 0.5016561083062675,
    "step": 15600
  },
  {
    "loss": 0.09193578362464905,
    "grad_norm": 0.5679323077201843,
    "learning_rate": 9.989358093803777e-05,
    "epoch": 0.5019776827346689,
    "step": 15610
  },
  {
    "loss": 0.09180984497070313,
    "grad_norm": 0.6876428127288818,
    "learning_rate": 9.979222956072404e-05,
    "epoch": 0.5022992571630704,
    "step": 15620
  },
  {
    "loss": 0.10084093809127807,
    "grad_norm": 0.2858849763870239,
    "learning_rate": 9.969087839683468e-05,
    "epoch": 0.5026208315914719,
    "step": 15630
  },
  {
    "loss": 0.07768757343292236,
    "grad_norm": 0.2807064354419708,
    "learning_rate": 9.958952755047905e-05,
    "epoch": 0.5029424060198733,
    "step": 15640
  },
  {
    "loss": 0.11399703025817871,
    "grad_norm": 0.7593263983726501,
    "learning_rate": 9.94881771257659e-05,
    "epoch": 0.5032639804482748,
    "step": 15650
  },
  {
    "loss": 0.07852271795272828,
    "grad_norm": 0.565068244934082,
    "learning_rate": 9.93868272268037e-05,
    "epoch": 0.5035855548766762,
    "step": 15660
  },
  {
    "loss": 0.0828331708908081,
    "grad_norm": 0.6039252877235413,
    "learning_rate": 9.928547795770036e-05,
    "epoch": 0.5039071293050776,
    "step": 15670
  },
  {
    "loss": 0.07960823178291321,
    "grad_norm": 0.39925456047058105,
    "learning_rate": 9.918412942256315e-05,
    "epoch": 0.5042287037334792,
    "step": 15680
  },
  {
    "loss": 0.07466946840286255,
    "grad_norm": 0.564476490020752,
    "learning_rate": 9.908278172549854e-05,
    "epoch": 0.5045502781618806,
    "step": 15690
  },
  {
    "loss": 0.08819963335990906,
    "grad_norm": 0.3453824520111084,
    "learning_rate": 9.89814349706122e-05,
    "epoch": 0.504871852590282,
    "step": 15700
  },
  {
    "loss": 0.1143036127090454,
    "grad_norm": 0.9882622361183167,
    "learning_rate": 9.888008926200881e-05,
    "epoch": 0.5051934270186835,
    "step": 15710
  },
  {
    "loss": 0.08639020919799804,
    "grad_norm": 0.7949790358543396,
    "learning_rate": 9.877874470379195e-05,
    "epoch": 0.5055150014470849,
    "step": 15720
  },
  {
    "loss": 0.07429859042167664,
    "grad_norm": 0.6657055616378784,
    "learning_rate": 9.867740140006406e-05,
    "epoch": 0.5058365758754864,
    "step": 15730
  },
  {
    "loss": 0.07545442581176758,
    "grad_norm": 0.48248526453971863,
    "learning_rate": 9.857605945492624e-05,
    "epoch": 0.5061581503038879,
    "step": 15740
  },
  {
    "loss": 0.07939736843109131,
    "grad_norm": 0.5649476647377014,
    "learning_rate": 9.847471897247823e-05,
    "epoch": 0.5064797247322893,
    "step": 15750
  },
  {
    "loss": 0.0747692346572876,
    "grad_norm": 0.6028984189033508,
    "learning_rate": 9.837338005681827e-05,
    "epoch": 0.5068012991606907,
    "step": 15760
  },
  {
    "loss": 0.06601083874702454,
    "grad_norm": 0.3233937919139862,
    "learning_rate": 9.827204281204298e-05,
    "epoch": 0.5071228735890922,
    "step": 15770
  },
  {
    "loss": 0.08770567774772645,
    "grad_norm": 0.48834502696990967,
    "learning_rate": 9.817070734224726e-05,
    "epoch": 0.5074444480174937,
    "step": 15780
  },
  {
    "loss": 0.10232018232345581,
    "grad_norm": 0.5461081862449646,
    "learning_rate": 9.806937375152422e-05,
    "epoch": 0.5077660224458951,
    "step": 15790
  },
  {
    "loss": 0.09111140370368957,
    "grad_norm": 0.23557019233703613,
    "learning_rate": 9.796804214396498e-05,
    "epoch": 0.5080875968742966,
    "step": 15800
  },
  {
    "loss": 0.094402277469635,
    "grad_norm": 0.8346803188323975,
    "learning_rate": 9.786671262365868e-05,
    "epoch": 0.508409171302698,
    "step": 15810
  },
  {
    "loss": 0.10833892822265626,
    "grad_norm": 0.817200779914856,
    "learning_rate": 9.77653852946922e-05,
    "epoch": 0.5087307457310994,
    "step": 15820
  },
  {
    "loss": 0.09572397470474243,
    "grad_norm": 0.7927505970001221,
    "learning_rate": 9.766406026115038e-05,
    "epoch": 0.509052320159501,
    "step": 15830
  },
  {
    "loss": 0.08352859020233154,
    "grad_norm": 0.6671968102455139,
    "learning_rate": 9.756273762711554e-05,
    "epoch": 0.5093738945879024,
    "step": 15840
  },
  {
    "loss": 0.06232442855834961,
    "grad_norm": 0.5571510791778564,
    "learning_rate": 9.746141749666759e-05,
    "epoch": 0.5096954690163038,
    "step": 15850
  },
  {
    "loss": 0.07288626432418824,
    "grad_norm": 0.49520349502563477,
    "learning_rate": 9.736009997388382e-05,
    "epoch": 0.5100170434447053,
    "step": 15860
  },
  {
    "loss": 0.10119060277938843,
    "grad_norm": 0.6404814124107361,
    "learning_rate": 9.725878516283889e-05,
    "epoch": 0.5103386178731067,
    "step": 15870
  },
  {
    "loss": 0.08675326108932495,
    "grad_norm": 0.7203865647315979,
    "learning_rate": 9.71574731676046e-05,
    "epoch": 0.5106601923015082,
    "step": 15880
  },
  {
    "loss": 0.08972042798995972,
    "grad_norm": 0.6852189302444458,
    "learning_rate": 9.705616409225004e-05,
    "epoch": 0.5109817667299097,
    "step": 15890
  },
  {
    "loss": 0.0967637062072754,
    "grad_norm": 0.5318194031715393,
    "learning_rate": 9.695485804084114e-05,
    "epoch": 0.5113033411583111,
    "step": 15900
  },
  {
    "loss": 0.08468080759048462,
    "grad_norm": 1.649613857269287,
    "learning_rate": 9.685355511744075e-05,
    "epoch": 0.5116249155867125,
    "step": 15910
  },
  {
    "loss": 0.08541063070297242,
    "grad_norm": 0.6673186421394348,
    "learning_rate": 9.675225542610847e-05,
    "epoch": 0.511946490015114,
    "step": 15920
  },
  {
    "loss": 0.07971563339233398,
    "grad_norm": 0.5428572297096252,
    "learning_rate": 9.66509590709007e-05,
    "epoch": 0.5122680644435155,
    "step": 15930
  },
  {
    "loss": 0.07468971610069275,
    "grad_norm": 0.8105533719062805,
    "learning_rate": 9.654966615587031e-05,
    "epoch": 0.5125896388719169,
    "step": 15940
  },
  {
    "loss": 0.08704789280891419,
    "grad_norm": 0.38906678557395935,
    "learning_rate": 9.644837678506666e-05,
    "epoch": 0.5129112133003184,
    "step": 15950
  },
  {
    "loss": 0.07773864269256592,
    "grad_norm": 0.6124730706214905,
    "learning_rate": 9.63470910625355e-05,
    "epoch": 0.5132327877287198,
    "step": 15960
  },
  {
    "loss": 0.09297425150871277,
    "grad_norm": 1.118937611579895,
    "learning_rate": 9.624580909231882e-05,
    "epoch": 0.5135543621571212,
    "step": 15970
  },
  {
    "loss": 0.08964041471481324,
    "grad_norm": 0.8257461190223694,
    "learning_rate": 9.614453097845473e-05,
    "epoch": 0.5138759365855228,
    "step": 15980
  },
  {
    "loss": 0.06911572813987732,
    "grad_norm": 0.3326082229614258,
    "learning_rate": 9.604325682497738e-05,
    "epoch": 0.5141975110139242,
    "step": 15990
  },
  {
    "loss": 0.08799180388450623,
    "grad_norm": 1.2826687097549438,
    "learning_rate": 9.59419867359169e-05,
    "epoch": 0.5145190854423256,
    "step": 16000
  },
  {
    "loss": 0.09639973044395447,
    "grad_norm": 0.8958343863487244,
    "learning_rate": 9.584072081529916e-05,
    "epoch": 0.514840659870727,
    "step": 16010
  },
  {
    "loss": 0.10252895355224609,
    "grad_norm": 0.7098028063774109,
    "learning_rate": 9.573945916714589e-05,
    "epoch": 0.5151622342991286,
    "step": 16020
  },
  {
    "loss": 0.09719353914260864,
    "grad_norm": 1.3733038902282715,
    "learning_rate": 9.563820189547425e-05,
    "epoch": 0.51548380872753,
    "step": 16030
  },
  {
    "loss": 0.07945320010185242,
    "grad_norm": 0.3698774576187134,
    "learning_rate": 9.553694910429704e-05,
    "epoch": 0.5158053831559314,
    "step": 16040
  },
  {
    "loss": 0.07129523158073425,
    "grad_norm": 0.4975164532661438,
    "learning_rate": 9.543570089762242e-05,
    "epoch": 0.5161269575843329,
    "step": 16050
  },
  {
    "loss": 0.0827067732810974,
    "grad_norm": 0.41556140780448914,
    "learning_rate": 9.533445737945382e-05,
    "epoch": 0.5164485320127343,
    "step": 16060
  },
  {
    "loss": 0.07715771794319153,
    "grad_norm": 0.6161404252052307,
    "learning_rate": 9.523321865378982e-05,
    "epoch": 0.5167701064411357,
    "step": 16070
  },
  {
    "loss": 0.112063467502594,
    "grad_norm": 0.3975111246109009,
    "learning_rate": 9.513198482462422e-05,
    "epoch": 0.5170916808695373,
    "step": 16080
  },
  {
    "loss": 0.08155578374862671,
    "grad_norm": 0.4807094633579254,
    "learning_rate": 9.503075599594564e-05,
    "epoch": 0.5174132552979387,
    "step": 16090
  },
  {
    "loss": 0.09136075973510742,
    "grad_norm": 0.8183568716049194,
    "learning_rate": 9.492953227173763e-05,
    "epoch": 0.5177348297263401,
    "step": 16100
  },
  {
    "loss": 0.08842723369598389,
    "grad_norm": 0.7488901615142822,
    "learning_rate": 9.482831375597848e-05,
    "epoch": 0.5180564041547416,
    "step": 16110
  },
  {
    "loss": 0.09355822801589966,
    "grad_norm": 0.6784147024154663,
    "learning_rate": 9.472710055264115e-05,
    "epoch": 0.518377978583143,
    "step": 16120
  },
  {
    "loss": 0.09595037698745727,
    "grad_norm": 0.4905896484851837,
    "learning_rate": 9.462589276569311e-05,
    "epoch": 0.5186995530115445,
    "step": 16130
  },
  {
    "loss": 0.10003188848495484,
    "grad_norm": 0.6823483109474182,
    "learning_rate": 9.452469049909625e-05,
    "epoch": 0.519021127439946,
    "step": 16140
  },
  {
    "loss": 0.09580358862876892,
    "grad_norm": 0.7502665519714355,
    "learning_rate": 9.442349385680689e-05,
    "epoch": 0.5193427018683474,
    "step": 16150
  },
  {
    "loss": 0.09321611523628234,
    "grad_norm": 0.40309739112854004,
    "learning_rate": 9.43223029427755e-05,
    "epoch": 0.5196642762967488,
    "step": 16160
  },
  {
    "loss": 0.09390120506286621,
    "grad_norm": 0.3821161091327667,
    "learning_rate": 9.422111786094666e-05,
    "epoch": 0.5199858507251504,
    "step": 16170
  },
  {
    "loss": 0.07894954681396485,
    "grad_norm": 0.4174655079841614,
    "learning_rate": 9.411993871525895e-05,
    "epoch": 0.5203074251535518,
    "step": 16180
  },
  {
    "loss": 0.07645484805107117,
    "grad_norm": 0.843494176864624,
    "learning_rate": 9.401876560964488e-05,
    "epoch": 0.5206289995819532,
    "step": 16190
  },
  {
    "loss": 0.07298891544342041,
    "grad_norm": 0.6622037291526794,
    "learning_rate": 9.391759864803073e-05,
    "epoch": 0.5209505740103547,
    "step": 16200
  },
  {
    "loss": 0.08789741396903991,
    "grad_norm": 0.7242502570152283,
    "learning_rate": 9.381643793433657e-05,
    "epoch": 0.5212721484387561,
    "step": 16210
  },
  {
    "loss": 0.08091871738433838,
    "grad_norm": 0.6538041830062866,
    "learning_rate": 9.371528357247589e-05,
    "epoch": 0.5215937228671575,
    "step": 16220
  },
  {
    "loss": 0.08055868744850159,
    "grad_norm": 0.7289121747016907,
    "learning_rate": 9.361413566635576e-05,
    "epoch": 0.5219152972955591,
    "step": 16230
  },
  {
    "loss": 0.09357218742370606,
    "grad_norm": 0.6839609146118164,
    "learning_rate": 9.35129943198766e-05,
    "epoch": 0.5222368717239605,
    "step": 16240
  },
  {
    "loss": 0.09125481843948365,
    "grad_norm": 0.654301106929779,
    "learning_rate": 9.341185963693209e-05,
    "epoch": 0.5225584461523619,
    "step": 16250
  },
  {
    "loss": 0.08754734992980957,
    "grad_norm": 0.6056458950042725,
    "learning_rate": 9.331073172140902e-05,
    "epoch": 0.5228800205807634,
    "step": 16260
  },
  {
    "loss": 0.07239229083061219,
    "grad_norm": 0.5807142853736877,
    "learning_rate": 9.320961067718733e-05,
    "epoch": 0.5232015950091649,
    "step": 16270
  },
  {
    "loss": 0.09158975481987,
    "grad_norm": 0.7827548384666443,
    "learning_rate": 9.31084966081398e-05,
    "epoch": 0.5235231694375663,
    "step": 16280
  },
  {
    "loss": 0.08536049127578735,
    "grad_norm": 0.5343862771987915,
    "learning_rate": 9.300738961813212e-05,
    "epoch": 0.5238447438659678,
    "step": 16290
  },
  {
    "loss": 0.07365837693214417,
    "grad_norm": 0.6306098103523254,
    "learning_rate": 9.290628981102265e-05,
    "epoch": 0.5241663182943692,
    "step": 16300
  },
  {
    "loss": 0.08824599981307983,
    "grad_norm": 0.4955608546733856,
    "learning_rate": 9.28051972906624e-05,
    "epoch": 0.5244878927227706,
    "step": 16310
  },
  {
    "loss": 0.10049338340759277,
    "grad_norm": 0.6357713341712952,
    "learning_rate": 9.270411216089487e-05,
    "epoch": 0.5248094671511722,
    "step": 16320
  },
  {
    "loss": 0.08429625034332275,
    "grad_norm": 0.692136824131012,
    "learning_rate": 9.2603034525556e-05,
    "epoch": 0.5251310415795736,
    "step": 16330
  },
  {
    "loss": 0.07850704789161682,
    "grad_norm": 0.9536001086235046,
    "learning_rate": 9.250196448847407e-05,
    "epoch": 0.525452616007975,
    "step": 16340
  },
  {
    "loss": 0.07307186126708984,
    "grad_norm": 0.7166012525558472,
    "learning_rate": 9.240090215346947e-05,
    "epoch": 0.5257741904363765,
    "step": 16350
  },
  {
    "loss": 0.09322544336318969,
    "grad_norm": 0.6299146413803101,
    "learning_rate": 9.229984762435469e-05,
    "epoch": 0.5260957648647779,
    "step": 16360
  },
  {
    "loss": 0.09763809442520141,
    "grad_norm": 0.4739687144756317,
    "learning_rate": 9.219880100493426e-05,
    "epoch": 0.5264173392931794,
    "step": 16370
  },
  {
    "loss": 0.0830114483833313,
    "grad_norm": 0.7146755456924438,
    "learning_rate": 9.209776239900453e-05,
    "epoch": 0.5267389137215809,
    "step": 16380
  },
  {
    "loss": 0.09164667129516602,
    "grad_norm": 0.924308717250824,
    "learning_rate": 9.199673191035363e-05,
    "epoch": 0.5270604881499823,
    "step": 16390
  },
  {
    "loss": 0.08107462525367737,
    "grad_norm": 0.6615635752677917,
    "learning_rate": 9.189570964276141e-05,
    "epoch": 0.5273820625783837,
    "step": 16400
  },
  {
    "loss": 0.07831215262413024,
    "grad_norm": 0.29422977566719055,
    "learning_rate": 9.179469569999919e-05,
    "epoch": 0.5277036370067852,
    "step": 16410
  },
  {
    "loss": 0.09040266275405884,
    "grad_norm": 0.6438028812408447,
    "learning_rate": 9.169369018582978e-05,
    "epoch": 0.5280252114351867,
    "step": 16420
  },
  {
    "loss": 0.09664230346679688,
    "grad_norm": 0.5955362915992737,
    "learning_rate": 9.159269320400735e-05,
    "epoch": 0.5283467858635881,
    "step": 16430
  },
  {
    "loss": 0.08227856755256653,
    "grad_norm": 0.7709397077560425,
    "learning_rate": 9.149170485827725e-05,
    "epoch": 0.5286683602919896,
    "step": 16440
  },
  {
    "loss": 0.06169787049293518,
    "grad_norm": 0.682957112789154,
    "learning_rate": 9.139072525237595e-05,
    "epoch": 0.528989934720391,
    "step": 16450
  },
  {
    "loss": 0.07773295640945435,
    "grad_norm": 0.5797860622406006,
    "learning_rate": 9.12897544900311e-05,
    "epoch": 0.5293115091487924,
    "step": 16460
  },
  {
    "loss": 0.0780327320098877,
    "grad_norm": 0.29846492409706116,
    "learning_rate": 9.118879267496107e-05,
    "epoch": 0.529633083577194,
    "step": 16470
  },
  {
    "loss": 0.10499351024627686,
    "grad_norm": 0.6282970309257507,
    "learning_rate": 9.108783991087515e-05,
    "epoch": 0.5299546580055954,
    "step": 16480
  },
  {
    "loss": 0.07932085394859315,
    "grad_norm": 0.6547428965568542,
    "learning_rate": 9.098689630147328e-05,
    "epoch": 0.5302762324339968,
    "step": 16490
  },
  {
    "loss": 0.07153794169425964,
    "grad_norm": 0.45561927556991577,
    "learning_rate": 9.088596195044602e-05,
    "epoch": 0.5305978068623983,
    "step": 16500
  },
  {
    "loss": 0.07359915971755981,
    "grad_norm": 0.6223723292350769,
    "learning_rate": 9.078503696147443e-05,
    "epoch": 0.5309193812907997,
    "step": 16510
  },
  {
    "loss": 0.0706337034702301,
    "grad_norm": 0.7523877024650574,
    "learning_rate": 9.06841214382299e-05,
    "epoch": 0.5312409557192012,
    "step": 16520
  },
  {
    "loss": 0.09963027238845826,
    "grad_norm": 0.557845413684845,
    "learning_rate": 9.058321548437422e-05,
    "epoch": 0.5315625301476027,
    "step": 16530
  },
  {
    "loss": 0.0828576385974884,
    "grad_norm": 0.7379382252693176,
    "learning_rate": 9.048231920355919e-05,
    "epoch": 0.5318841045760041,
    "step": 16540
  },
  {
    "loss": 0.07978498935699463,
    "grad_norm": 0.6947690844535828,
    "learning_rate": 9.03814326994268e-05,
    "epoch": 0.5322056790044055,
    "step": 16550
  },
  {
    "loss": 0.09839178323745727,
    "grad_norm": 0.790624737739563,
    "learning_rate": 9.028055607560892e-05,
    "epoch": 0.532527253432807,
    "step": 16560
  },
  {
    "loss": 0.07617455720901489,
    "grad_norm": 0.8281884789466858,
    "learning_rate": 9.017968943572732e-05,
    "epoch": 0.5328488278612085,
    "step": 16570
  },
  {
    "loss": 0.0846377968788147,
    "grad_norm": 0.36391204595565796,
    "learning_rate": 9.007883288339345e-05,
    "epoch": 0.5331704022896099,
    "step": 16580
  },
  {
    "loss": 0.0801987111568451,
    "grad_norm": 0.6115204095840454,
    "learning_rate": 8.99779865222085e-05,
    "epoch": 0.5334919767180114,
    "step": 16590
  },
  {
    "loss": 0.09012607932090759,
    "grad_norm": 0.5107051134109497,
    "learning_rate": 8.98771504557631e-05,
    "epoch": 0.5338135511464128,
    "step": 16600
  },
  {
    "loss": 0.08063666820526123,
    "grad_norm": 0.3612475097179413,
    "learning_rate": 8.977632478763735e-05,
    "epoch": 0.5341351255748142,
    "step": 16610
  },
  {
    "loss": 0.09662036299705505,
    "grad_norm": 0.7268196940422058,
    "learning_rate": 8.967550962140066e-05,
    "epoch": 0.5344567000032158,
    "step": 16620
  },
  {
    "loss": 0.07537992596626282,
    "grad_norm": 0.8529553413391113,
    "learning_rate": 8.957470506061164e-05,
    "epoch": 0.5347782744316172,
    "step": 16630
  },
  {
    "loss": 0.08451104164123535,
    "grad_norm": 0.9954885244369507,
    "learning_rate": 8.947391120881802e-05,
    "epoch": 0.5350998488600186,
    "step": 16640
  },
  {
    "loss": 0.07832143306732178,
    "grad_norm": 0.5641348958015442,
    "learning_rate": 8.93731281695565e-05,
    "epoch": 0.5354214232884201,
    "step": 16650
  },
  {
    "loss": 0.07633439302444459,
    "grad_norm": 0.712746798992157,
    "learning_rate": 8.927235604635274e-05,
    "epoch": 0.5357429977168215,
    "step": 16660
  },
  {
    "loss": 0.06378179192543029,
    "grad_norm": 0.5640179514884949,
    "learning_rate": 8.917159494272112e-05,
    "epoch": 0.536064572145223,
    "step": 16670
  },
  {
    "loss": 0.08095942735671997,
    "grad_norm": 0.6787976622581482,
    "learning_rate": 8.907084496216474e-05,
    "epoch": 0.5363861465736245,
    "step": 16680
  },
  {
    "loss": 0.07943691611289978,
    "grad_norm": 0.4951293468475342,
    "learning_rate": 8.897010620817527e-05,
    "epoch": 0.5367077210020259,
    "step": 16690
  },
  {
    "loss": 0.08359546661376953,
    "grad_norm": 0.5979715585708618,
    "learning_rate": 8.886937878423281e-05,
    "epoch": 0.5370292954304273,
    "step": 16700
  },
  {
    "loss": 0.05622500777244568,
    "grad_norm": 0.35826343297958374,
    "learning_rate": 8.876866279380582e-05,
    "epoch": 0.5373508698588289,
    "step": 16710
  },
  {
    "loss": 0.0896606206893921,
    "grad_norm": 0.6729931235313416,
    "learning_rate": 8.866795834035111e-05,
    "epoch": 0.5376724442872303,
    "step": 16720
  },
  {
    "loss": 0.08825923204421997,
    "grad_norm": 0.6600592732429504,
    "learning_rate": 8.856726552731355e-05,
    "epoch": 0.5379940187156317,
    "step": 16730
  },
  {
    "loss": 0.07278905510902405,
    "grad_norm": 0.5084134936332703,
    "learning_rate": 8.84665844581261e-05,
    "epoch": 0.5383155931440332,
    "step": 16740
  },
  {
    "loss": 0.07857139706611634,
    "grad_norm": 0.7476891875267029,
    "learning_rate": 8.836591523620959e-05,
    "epoch": 0.5386371675724346,
    "step": 16750
  },
  {
    "loss": 0.0739391565322876,
    "grad_norm": 1.2707487344741821,
    "learning_rate": 8.826525796497273e-05,
    "epoch": 0.538958742000836,
    "step": 16760
  },
  {
    "loss": 0.08058958649635314,
    "grad_norm": 0.28494247794151306,
    "learning_rate": 8.816461274781193e-05,
    "epoch": 0.5392803164292376,
    "step": 16770
  },
  {
    "loss": 0.06720038652420043,
    "grad_norm": 0.48339444398880005,
    "learning_rate": 8.80639796881113e-05,
    "epoch": 0.539601890857639,
    "step": 16780
  },
  {
    "loss": 0.06298373341560363,
    "grad_norm": 0.4123383164405823,
    "learning_rate": 8.796335888924239e-05,
    "epoch": 0.5399234652860404,
    "step": 16790
  },
  {
    "loss": 0.09222426414489746,
    "grad_norm": 0.8985582590103149,
    "learning_rate": 8.78627504545641e-05,
    "epoch": 0.5402450397144419,
    "step": 16800
  },
  {
    "loss": 0.07999668121337891,
    "grad_norm": 0.57627934217453,
    "learning_rate": 8.776215448742274e-05,
    "epoch": 0.5405666141428433,
    "step": 16810
  },
  {
    "loss": 0.09616152048110962,
    "grad_norm": 0.5940356254577637,
    "learning_rate": 8.766157109115174e-05,
    "epoch": 0.5408881885712448,
    "step": 16820
  },
  {
    "loss": 0.0929766058921814,
    "grad_norm": 1.051363468170166,
    "learning_rate": 8.756100036907166e-05,
    "epoch": 0.5412097629996463,
    "step": 16830
  },
  {
    "loss": 0.08186267614364624,
    "grad_norm": 0.942518949508667,
    "learning_rate": 8.746044242449001e-05,
    "epoch": 0.5415313374280477,
    "step": 16840
  },
  {
    "loss": 0.08210628032684326,
    "grad_norm": 0.37757983803749084,
    "learning_rate": 8.735989736070121e-05,
    "epoch": 0.5418529118564491,
    "step": 16850
  },
  {
    "loss": 0.07749378681182861,
    "grad_norm": 0.6084452867507935,
    "learning_rate": 8.72593652809864e-05,
    "epoch": 0.5421744862848507,
    "step": 16860
  },
  {
    "loss": 0.0946018636226654,
    "grad_norm": 0.681425154209137,
    "learning_rate": 8.715884628861344e-05,
    "epoch": 0.5424960607132521,
    "step": 16870
  },
  {
    "loss": 0.09242323637008668,
    "grad_norm": 0.5492236018180847,
    "learning_rate": 8.705834048683669e-05,
    "epoch": 0.5428176351416535,
    "step": 16880
  },
  {
    "loss": 0.08967363834381104,
    "grad_norm": 0.3688710331916809,
    "learning_rate": 8.6957847978897e-05,
    "epoch": 0.543139209570055,
    "step": 16890
  },
  {
    "loss": 0.0760436475276947,
    "grad_norm": 0.4894743859767914,
    "learning_rate": 8.685736886802152e-05,
    "epoch": 0.5434607839984564,
    "step": 16900
  },
  {
    "loss": 0.06940741539001465,
    "grad_norm": 0.28525763750076294,
    "learning_rate": 8.675690325742371e-05,
    "epoch": 0.5437823584268578,
    "step": 16910
  },
  {
    "loss": 0.07846690416336059,
    "grad_norm": 0.7112274169921875,
    "learning_rate": 8.665645125030311e-05,
    "epoch": 0.5441039328552594,
    "step": 16920
  },
  {
    "loss": 0.10269311666488648,
    "grad_norm": 0.4686936140060425,
    "learning_rate": 8.65560129498453e-05,
    "epoch": 0.5444255072836608,
    "step": 16930
  },
  {
    "loss": 0.08768737316131592,
    "grad_norm": 0.5227848887443542,
    "learning_rate": 8.645558845922177e-05,
    "epoch": 0.5447470817120622,
    "step": 16940
  },
  {
    "loss": 0.08848978281021118,
    "grad_norm": 0.7514484524726868,
    "learning_rate": 8.635517788158982e-05,
    "epoch": 0.5450686561404637,
    "step": 16950
  },
  {
    "loss": 0.06979380249977112,
    "grad_norm": 0.5240973234176636,
    "learning_rate": 8.625478132009248e-05,
    "epoch": 0.5453902305688652,
    "step": 16960
  },
  {
    "loss": 0.09580703973770141,
    "grad_norm": 0.374254435300827,
    "learning_rate": 8.615439887785838e-05,
    "epoch": 0.5457118049972666,
    "step": 16970
  },
  {
    "loss": 0.060118526220321655,
    "grad_norm": 1.8248990774154663,
    "learning_rate": 8.605403065800168e-05,
    "epoch": 0.5460333794256681,
    "step": 16980
  },
  {
    "loss": 0.0934792399406433,
    "grad_norm": 0.6583986282348633,
    "learning_rate": 8.595367676362181e-05,
    "epoch": 0.5463549538540695,
    "step": 16990
  },
  {
    "loss": 0.09412904381752014,
    "grad_norm": 0.46310746669769287,
    "learning_rate": 8.585333729780361e-05,
    "epoch": 0.5466765282824709,
    "step": 17000
  },
  {
    "loss": 0.08042793273925782,
    "grad_norm": 0.7318806052207947,
    "learning_rate": 8.575301236361708e-05,
    "epoch": 0.5469981027108725,
    "step": 17010
  },
  {
    "loss": 0.08339504599571228,
    "grad_norm": 0.817503035068512,
    "learning_rate": 8.565270206411722e-05,
    "epoch": 0.5473196771392739,
    "step": 17020
  },
  {
    "loss": 0.08974653482437134,
    "grad_norm": 0.582861602306366,
    "learning_rate": 8.555240650234399e-05,
    "epoch": 0.5476412515676753,
    "step": 17030
  },
  {
    "loss": 0.0690845251083374,
    "grad_norm": 0.4486590325832367,
    "learning_rate": 8.54521257813224e-05,
    "epoch": 0.5479628259960768,
    "step": 17040
  },
  {
    "loss": 0.08085647821426392,
    "grad_norm": 0.48037925362586975,
    "learning_rate": 8.535186000406204e-05,
    "epoch": 0.5482844004244782,
    "step": 17050
  },
  {
    "loss": 0.10313069820404053,
    "grad_norm": 0.935422420501709,
    "learning_rate": 8.525160927355712e-05,
    "epoch": 0.5486059748528797,
    "step": 17060
  },
  {
    "loss": 0.08438910245895385,
    "grad_norm": 0.7324216365814209,
    "learning_rate": 8.515137369278651e-05,
    "epoch": 0.5489275492812812,
    "step": 17070
  },
  {
    "loss": 0.06928063035011292,
    "grad_norm": 0.3410673439502716,
    "learning_rate": 8.50511533647135e-05,
    "epoch": 0.5492491237096826,
    "step": 17080
  },
  {
    "loss": 0.06782970428466797,
    "grad_norm": 0.6539098024368286,
    "learning_rate": 8.49509483922856e-05,
    "epoch": 0.549570698138084,
    "step": 17090
  },
  {
    "eval_loss": 0.07829523831605911,
    "eval_runtime": 35.216,
    "eval_samples_per_second": 142.719,
    "eval_steps_per_second": 35.694,
    "epoch": 0.5496993279094446,
    "step": 17094
  },
  {
    "loss": 0.08837011456489563,
    "grad_norm": 0.6394001245498657,
    "learning_rate": 8.485075887843473e-05,
    "epoch": 0.5498922725664855,
    "step": 17100
  },
  {
    "loss": 0.06662584543228149,
    "grad_norm": 0.5925083756446838,
    "learning_rate": 8.475058492607677e-05,
    "epoch": 0.550213846994887,
    "step": 17110
  },
  {
    "loss": 0.09058014750480652,
    "grad_norm": 0.6222550868988037,
    "learning_rate": 8.465042663811172e-05,
    "epoch": 0.5505354214232884,
    "step": 17120
  },
  {
    "loss": 0.07723432183265685,
    "grad_norm": 0.5994893312454224,
    "learning_rate": 8.455028411742339e-05,
    "epoch": 0.5508569958516899,
    "step": 17130
  },
  {
    "loss": 0.07140976786613465,
    "grad_norm": 0.998021125793457,
    "learning_rate": 8.445015746687948e-05,
    "epoch": 0.5511785702800913,
    "step": 17140
  },
  {
    "loss": 0.08850755095481873,
    "grad_norm": 0.33564454317092896,
    "learning_rate": 8.435004678933134e-05,
    "epoch": 0.5515001447084927,
    "step": 17150
  },
  {
    "loss": 0.09072214961051941,
    "grad_norm": 0.5762131214141846,
    "learning_rate": 8.424995218761397e-05,
    "epoch": 0.5518217191368943,
    "step": 17160
  },
  {
    "loss": 0.09939899444580078,
    "grad_norm": 0.6515547037124634,
    "learning_rate": 8.41498737645458e-05,
    "epoch": 0.5521432935652957,
    "step": 17170
  },
  {
    "loss": 0.0911516547203064,
    "grad_norm": 0.4145786166191101,
    "learning_rate": 8.404981162292862e-05,
    "epoch": 0.5524648679936971,
    "step": 17180
  },
  {
    "loss": 0.10890164375305175,
    "grad_norm": 0.7293618321418762,
    "learning_rate": 8.39497658655476e-05,
    "epoch": 0.5527864424220986,
    "step": 17190
  },
  {
    "loss": 0.08640298247337341,
    "grad_norm": 0.7666877508163452,
    "learning_rate": 8.384973659517094e-05,
    "epoch": 0.5531080168505,
    "step": 17200
  },
  {
    "loss": 0.0860232651233673,
    "grad_norm": 0.5176871418952942,
    "learning_rate": 8.374972391455002e-05,
    "epoch": 0.5534295912789015,
    "step": 17210
  },
  {
    "loss": 0.07132713198661804,
    "grad_norm": 0.23162242770195007,
    "learning_rate": 8.364972792641912e-05,
    "epoch": 0.553751165707303,
    "step": 17220
  },
  {
    "loss": 0.05892365574836731,
    "grad_norm": 0.5613875985145569,
    "learning_rate": 8.354974873349542e-05,
    "epoch": 0.5540727401357044,
    "step": 17230
  },
  {
    "loss": 0.07454466819763184,
    "grad_norm": 0.8335756063461304,
    "learning_rate": 8.344978643847879e-05,
    "epoch": 0.5543943145641058,
    "step": 17240
  },
  {
    "loss": 0.07813173532485962,
    "grad_norm": 0.833672046661377,
    "learning_rate": 8.334984114405177e-05,
    "epoch": 0.5547158889925073,
    "step": 17250
  },
  {
    "loss": 0.057703042030334474,
    "grad_norm": 0.38010871410369873,
    "learning_rate": 8.324991295287943e-05,
    "epoch": 0.5550374634209088,
    "step": 17260
  },
  {
    "loss": 0.0920906662940979,
    "grad_norm": 0.5197877287864685,
    "learning_rate": 8.315000196760933e-05,
    "epoch": 0.5553590378493102,
    "step": 17270
  },
  {
    "loss": 0.08447323441505432,
    "grad_norm": 0.7848365902900696,
    "learning_rate": 8.305010829087118e-05,
    "epoch": 0.5556806122777117,
    "step": 17280
  },
  {
    "loss": 0.08707159161567687,
    "grad_norm": 0.5702301859855652,
    "learning_rate": 8.295023202527718e-05,
    "epoch": 0.5560021867061131,
    "step": 17290
  },
  {
    "loss": 0.07622004747390747,
    "grad_norm": 0.6026646494865417,
    "learning_rate": 8.285037327342142e-05,
    "epoch": 0.5563237611345145,
    "step": 17300
  },
  {
    "loss": 0.06785175204277039,
    "grad_norm": 0.8649535179138184,
    "learning_rate": 8.275053213788007e-05,
    "epoch": 0.5566453355629161,
    "step": 17310
  },
  {
    "loss": 0.1002342700958252,
    "grad_norm": 0.7633518576622009,
    "learning_rate": 8.265070872121129e-05,
    "epoch": 0.5569669099913175,
    "step": 17320
  },
  {
    "loss": 0.0922668993473053,
    "grad_norm": 0.5194448828697205,
    "learning_rate": 8.255090312595487e-05,
    "epoch": 0.5572884844197189,
    "step": 17330
  },
  {
    "loss": 0.10162649154663086,
    "grad_norm": 0.6360050439834595,
    "learning_rate": 8.245111545463243e-05,
    "epoch": 0.5576100588481204,
    "step": 17340
  },
  {
    "loss": 0.06755626797676087,
    "grad_norm": 0.5018109083175659,
    "learning_rate": 8.235134580974706e-05,
    "epoch": 0.5579316332765218,
    "step": 17350
  },
  {
    "loss": 0.06583888530731201,
    "grad_norm": 0.6680736541748047,
    "learning_rate": 8.225159429378353e-05,
    "epoch": 0.5582532077049233,
    "step": 17360
  },
  {
    "loss": 0.07393178939819336,
    "grad_norm": 0.823982298374176,
    "learning_rate": 8.215186100920783e-05,
    "epoch": 0.5585747821333248,
    "step": 17370
  },
  {
    "loss": 0.08291969895362854,
    "grad_norm": 0.7191532850265503,
    "learning_rate": 8.205214605846721e-05,
    "epoch": 0.5588963565617262,
    "step": 17380
  },
  {
    "loss": 0.08974564075469971,
    "grad_norm": 0.5096963047981262,
    "learning_rate": 8.195244954399014e-05,
    "epoch": 0.5592179309901276,
    "step": 17390
  },
  {
    "loss": 0.09367521405220032,
    "grad_norm": 0.5104748010635376,
    "learning_rate": 8.185277156818619e-05,
    "epoch": 0.5595395054185291,
    "step": 17400
  },
  {
    "loss": 0.08481093645095825,
    "grad_norm": 0.5822654366493225,
    "learning_rate": 8.175311223344578e-05,
    "epoch": 0.5598610798469306,
    "step": 17410
  },
  {
    "loss": 0.06985856294631958,
    "grad_norm": 0.5147339701652527,
    "learning_rate": 8.16534716421403e-05,
    "epoch": 0.560182654275332,
    "step": 17420
  },
  {
    "loss": 0.09017146229743958,
    "grad_norm": 0.6477312445640564,
    "learning_rate": 8.155384989662181e-05,
    "epoch": 0.5605042287037335,
    "step": 17430
  },
  {
    "loss": 0.10362013578414916,
    "grad_norm": 0.530701756477356,
    "learning_rate": 8.145424709922305e-05,
    "epoch": 0.5608258031321349,
    "step": 17440
  },
  {
    "loss": 0.10210756063461304,
    "grad_norm": 0.5977575182914734,
    "learning_rate": 8.135466335225726e-05,
    "epoch": 0.5611473775605363,
    "step": 17450
  },
  {
    "loss": 0.08222898840904236,
    "grad_norm": 0.5708121657371521,
    "learning_rate": 8.125509875801813e-05,
    "epoch": 0.5614689519889379,
    "step": 17460
  },
  {
    "loss": 0.07522875666618348,
    "grad_norm": 0.5921570062637329,
    "learning_rate": 8.115555341877965e-05,
    "epoch": 0.5617905264173393,
    "step": 17470
  },
  {
    "loss": 0.08083952069282532,
    "grad_norm": 0.9275753498077393,
    "learning_rate": 8.10560274367961e-05,
    "epoch": 0.5621121008457407,
    "step": 17480
  },
  {
    "loss": 0.10283910036087036,
    "grad_norm": 0.4716049134731293,
    "learning_rate": 8.095652091430184e-05,
    "epoch": 0.5624336752741422,
    "step": 17490
  },
  {
    "loss": 0.08383597135543823,
    "grad_norm": 0.839942455291748,
    "learning_rate": 8.08570339535112e-05,
    "epoch": 0.5627552497025436,
    "step": 17500
  },
  {
    "loss": 0.08465657234191895,
    "grad_norm": 0.7459936141967773,
    "learning_rate": 8.075756665661848e-05,
    "epoch": 0.5630768241309451,
    "step": 17510
  },
  {
    "loss": 0.08436853289604188,
    "grad_norm": 0.8659219741821289,
    "learning_rate": 8.065811912579771e-05,
    "epoch": 0.5633983985593466,
    "step": 17520
  },
  {
    "loss": 0.07398121356964112,
    "grad_norm": 0.43199411034584045,
    "learning_rate": 8.055869146320269e-05,
    "epoch": 0.563719972987748,
    "step": 17530
  },
  {
    "loss": 0.0951408565044403,
    "grad_norm": 0.3734264373779297,
    "learning_rate": 8.045928377096676e-05,
    "epoch": 0.5640415474161494,
    "step": 17540
  },
  {
    "loss": 0.08055580258369446,
    "grad_norm": 0.4257162809371948,
    "learning_rate": 8.035989615120279e-05,
    "epoch": 0.564363121844551,
    "step": 17550
  },
  {
    "loss": 0.07695200443267822,
    "grad_norm": 0.49492180347442627,
    "learning_rate": 8.026052870600298e-05,
    "epoch": 0.5646846962729524,
    "step": 17560
  },
  {
    "loss": 0.09232667088508606,
    "grad_norm": 0.6467422246932983,
    "learning_rate": 8.016118153743886e-05,
    "epoch": 0.5650062707013538,
    "step": 17570
  },
  {
    "loss": 0.07816627621650696,
    "grad_norm": 1.0633784532546997,
    "learning_rate": 8.006185474756106e-05,
    "epoch": 0.5653278451297553,
    "step": 17580
  },
  {
    "loss": 0.09850267767906189,
    "grad_norm": 0.46513116359710693,
    "learning_rate": 7.996254843839938e-05,
    "epoch": 0.5656494195581567,
    "step": 17590
  },
  {
    "loss": 0.08328604102134704,
    "grad_norm": 0.6261907815933228,
    "learning_rate": 7.986326271196243e-05,
    "epoch": 0.5659709939865581,
    "step": 17600
  },
  {
    "loss": 0.07954009175300598,
    "grad_norm": 0.5492364168167114,
    "learning_rate": 7.976399767023786e-05,
    "epoch": 0.5662925684149597,
    "step": 17610
  },
  {
    "loss": 0.08056249022483826,
    "grad_norm": 0.6301096081733704,
    "learning_rate": 7.966475341519194e-05,
    "epoch": 0.5666141428433611,
    "step": 17620
  },
  {
    "loss": 0.08544570803642274,
    "grad_norm": 0.5183818936347961,
    "learning_rate": 7.956553004876967e-05,
    "epoch": 0.5669357172717625,
    "step": 17630
  },
  {
    "loss": 0.06803440451622009,
    "grad_norm": 0.4413444697856903,
    "learning_rate": 7.94663276728945e-05,
    "epoch": 0.567257291700164,
    "step": 17640
  },
  {
    "loss": 0.09343262910842895,
    "grad_norm": 0.8371837735176086,
    "learning_rate": 7.936714638946838e-05,
    "epoch": 0.5675788661285655,
    "step": 17650
  },
  {
    "loss": 0.08109647035598755,
    "grad_norm": 0.8725554347038269,
    "learning_rate": 7.926798630037156e-05,
    "epoch": 0.5679004405569669,
    "step": 17660
  },
  {
    "loss": 0.09562431573867798,
    "grad_norm": 0.6995370388031006,
    "learning_rate": 7.916884750746263e-05,
    "epoch": 0.5682220149853684,
    "step": 17670
  },
  {
    "loss": 0.06416831612586975,
    "grad_norm": 0.455130010843277,
    "learning_rate": 7.90697301125782e-05,
    "epoch": 0.5685435894137698,
    "step": 17680
  },
  {
    "loss": 0.07064977884292603,
    "grad_norm": 0.7157491445541382,
    "learning_rate": 7.897063421753282e-05,
    "epoch": 0.5688651638421712,
    "step": 17690
  },
  {
    "loss": 0.0642152488231659,
    "grad_norm": 0.6208144426345825,
    "learning_rate": 7.887155992411914e-05,
    "epoch": 0.5691867382705728,
    "step": 17700
  },
  {
    "loss": 0.092697274684906,
    "grad_norm": 0.5451779961585999,
    "learning_rate": 7.877250733410752e-05,
    "epoch": 0.5695083126989742,
    "step": 17710
  },
  {
    "loss": 0.06400271058082581,
    "grad_norm": 0.5939658880233765,
    "learning_rate": 7.867347654924601e-05,
    "epoch": 0.5698298871273756,
    "step": 17720
  },
  {
    "loss": 0.062121731042861936,
    "grad_norm": 0.31174877285957336,
    "learning_rate": 7.857446767126029e-05,
    "epoch": 0.5701514615557771,
    "step": 17730
  },
  {
    "loss": 0.08236455917358398,
    "grad_norm": 0.9238488674163818,
    "learning_rate": 7.847548080185357e-05,
    "epoch": 0.5704730359841785,
    "step": 17740
  },
  {
    "loss": 0.08480204343795776,
    "grad_norm": 0.625153124332428,
    "learning_rate": 7.837651604270642e-05,
    "epoch": 0.57079461041258,
    "step": 17750
  },
  {
    "loss": 0.08454369306564331,
    "grad_norm": 0.43398886919021606,
    "learning_rate": 7.827757349547666e-05,
    "epoch": 0.5711161848409815,
    "step": 17760
  },
  {
    "loss": 0.0682714343070984,
    "grad_norm": 0.5858274698257446,
    "learning_rate": 7.817865326179937e-05,
    "epoch": 0.5714377592693829,
    "step": 17770
  },
  {
    "loss": 0.07586719393730164,
    "grad_norm": 0.4799453914165497,
    "learning_rate": 7.807975544328666e-05,
    "epoch": 0.5717593336977843,
    "step": 17780
  },
  {
    "loss": 0.07879873514175414,
    "grad_norm": 0.6759800910949707,
    "learning_rate": 7.798088014152758e-05,
    "epoch": 0.5720809081261858,
    "step": 17790
  },
  {
    "loss": 0.08980271220207214,
    "grad_norm": 0.9335778951644897,
    "learning_rate": 7.788202745808817e-05,
    "epoch": 0.5724024825545873,
    "step": 17800
  },
  {
    "loss": 0.09147007465362549,
    "grad_norm": 0.6196408867835999,
    "learning_rate": 7.778319749451113e-05,
    "epoch": 0.5727240569829887,
    "step": 17810
  },
  {
    "loss": 0.09004471302032471,
    "grad_norm": 0.5582491755485535,
    "learning_rate": 7.768439035231584e-05,
    "epoch": 0.5730456314113902,
    "step": 17820
  },
  {
    "loss": 0.07299330830574036,
    "grad_norm": 0.45894795656204224,
    "learning_rate": 7.758560613299827e-05,
    "epoch": 0.5733672058397916,
    "step": 17830
  },
  {
    "loss": 0.07036964297294616,
    "grad_norm": 0.45925047993659973,
    "learning_rate": 7.748684493803082e-05,
    "epoch": 0.573688780268193,
    "step": 17840
  },
  {
    "loss": 0.06470816135406494,
    "grad_norm": 0.4903404116630554,
    "learning_rate": 7.738810686886221e-05,
    "epoch": 0.5740103546965946,
    "step": 17850
  },
  {
    "loss": 0.07539921402931213,
    "grad_norm": 0.9620845317840576,
    "learning_rate": 7.72893920269175e-05,
    "epoch": 0.574331929124996,
    "step": 17860
  },
  {
    "loss": 0.06352456212043762,
    "grad_norm": 0.4858039319515228,
    "learning_rate": 7.719070051359781e-05,
    "epoch": 0.5746535035533974,
    "step": 17870
  },
  {
    "loss": 0.08455619215965271,
    "grad_norm": 0.7511271834373474,
    "learning_rate": 7.70920324302803e-05,
    "epoch": 0.5749750779817989,
    "step": 17880
  },
  {
    "loss": 0.08338568210601807,
    "grad_norm": 0.6056551933288574,
    "learning_rate": 7.699338787831809e-05,
    "epoch": 0.5752966524102003,
    "step": 17890
  },
  {
    "loss": 0.0965720534324646,
    "grad_norm": 0.7961472868919373,
    "learning_rate": 7.689476695904013e-05,
    "epoch": 0.5756182268386018,
    "step": 17900
  },
  {
    "loss": 0.07739994525909424,
    "grad_norm": 0.44330519437789917,
    "learning_rate": 7.679616977375105e-05,
    "epoch": 0.5759398012670033,
    "step": 17910
  },
  {
    "loss": 0.06674630641937256,
    "grad_norm": 0.5733117461204529,
    "learning_rate": 7.66975964237311e-05,
    "epoch": 0.5762613756954047,
    "step": 17920
  },
  {
    "loss": 0.0786827027797699,
    "grad_norm": 0.5420511960983276,
    "learning_rate": 7.659904701023616e-05,
    "epoch": 0.5765829501238061,
    "step": 17930
  },
  {
    "loss": 0.09021250009536744,
    "grad_norm": 1.1385753154754639,
    "learning_rate": 7.650052163449743e-05,
    "epoch": 0.5769045245522076,
    "step": 17940
  },
  {
    "loss": 0.09374557733535767,
    "grad_norm": 0.6973399519920349,
    "learning_rate": 7.640202039772137e-05,
    "epoch": 0.5772260989806091,
    "step": 17950
  },
  {
    "loss": 0.10186971426010132,
    "grad_norm": 0.611345112323761,
    "learning_rate": 7.630354340108974e-05,
    "epoch": 0.5775476734090105,
    "step": 17960
  },
  {
    "loss": 0.09474144577980041,
    "grad_norm": 0.7261148691177368,
    "learning_rate": 7.620509074575934e-05,
    "epoch": 0.577869247837412,
    "step": 17970
  },
  {
    "loss": 0.09362642765045166,
    "grad_norm": 0.8504903316497803,
    "learning_rate": 7.610666253286197e-05,
    "epoch": 0.5781908222658134,
    "step": 17980
  },
  {
    "loss": 0.07732483148574829,
    "grad_norm": 0.6735202074050903,
    "learning_rate": 7.600825886350444e-05,
    "epoch": 0.5785123966942148,
    "step": 17990
  },
  {
    "loss": 0.07429722547531128,
    "grad_norm": 1.00436532497406,
    "learning_rate": 7.590987983876815e-05,
    "epoch": 0.5788339711226164,
    "step": 18000
  },
  {
    "loss": 0.09903724193572998,
    "grad_norm": 0.7012553215026855,
    "learning_rate": 7.581152555970931e-05,
    "epoch": 0.5791555455510178,
    "step": 18010
  },
  {
    "loss": 0.08933169841766357,
    "grad_norm": 0.8168038129806519,
    "learning_rate": 7.571319612735871e-05,
    "epoch": 0.5794771199794192,
    "step": 18020
  },
  {
    "loss": 0.06783590316772461,
    "grad_norm": 0.6012395024299622,
    "learning_rate": 7.561489164272157e-05,
    "epoch": 0.5797986944078207,
    "step": 18030
  },
  {
    "loss": 0.07372558116912842,
    "grad_norm": 0.5882187485694885,
    "learning_rate": 7.551661220677749e-05,
    "epoch": 0.5801202688362221,
    "step": 18040
  },
  {
    "loss": 0.0641398310661316,
    "grad_norm": 0.48085817694664,
    "learning_rate": 7.541835792048037e-05,
    "epoch": 0.5804418432646236,
    "step": 18050
  },
  {
    "loss": 0.07381256818771362,
    "grad_norm": 0.8459762930870056,
    "learning_rate": 7.532012888475827e-05,
    "epoch": 0.5807634176930251,
    "step": 18060
  },
  {
    "loss": 0.08163591623306274,
    "grad_norm": 0.803181529045105,
    "learning_rate": 7.522192520051329e-05,
    "epoch": 0.5810849921214265,
    "step": 18070
  },
  {
    "loss": 0.06541717648506165,
    "grad_norm": 0.44433683156967163,
    "learning_rate": 7.512374696862154e-05,
    "epoch": 0.5814065665498279,
    "step": 18080
  },
  {
    "loss": 0.07731590270996094,
    "grad_norm": 0.9130017161369324,
    "learning_rate": 7.50255942899329e-05,
    "epoch": 0.5817281409782294,
    "step": 18090
  },
  {
    "loss": 0.05984402298927307,
    "grad_norm": 0.3957984447479248,
    "learning_rate": 7.492746726527107e-05,
    "epoch": 0.5820497154066309,
    "step": 18100
  },
  {
    "loss": 0.09862715601921082,
    "grad_norm": 0.398759663105011,
    "learning_rate": 7.482936599543333e-05,
    "epoch": 0.5823712898350323,
    "step": 18110
  },
  {
    "loss": 0.08806217908859253,
    "grad_norm": 0.5182679891586304,
    "learning_rate": 7.473129058119063e-05,
    "epoch": 0.5826928642634338,
    "step": 18120
  },
  {
    "loss": 0.09343465566635131,
    "grad_norm": 0.7294725775718689,
    "learning_rate": 7.463324112328725e-05,
    "epoch": 0.5830144386918352,
    "step": 18130
  },
  {
    "loss": 0.07456960678100585,
    "grad_norm": 0.7115042805671692,
    "learning_rate": 7.453521772244082e-05,
    "epoch": 0.5833360131202366,
    "step": 18140
  },
  {
    "loss": 0.08490001559257507,
    "grad_norm": 0.680609941482544,
    "learning_rate": 7.443722047934223e-05,
    "epoch": 0.5836575875486382,
    "step": 18150
  },
  {
    "loss": 0.08044389486312867,
    "grad_norm": 1.0964044332504272,
    "learning_rate": 7.433924949465552e-05,
    "epoch": 0.5839791619770396,
    "step": 18160
  },
  {
    "loss": 0.0721604585647583,
    "grad_norm": 0.4555046260356903,
    "learning_rate": 7.424130486901764e-05,
    "epoch": 0.584300736405441,
    "step": 18170
  },
  {
    "loss": 0.07597453594207763,
    "grad_norm": 0.55930095911026,
    "learning_rate": 7.414338670303868e-05,
    "epoch": 0.5846223108338425,
    "step": 18180
  },
  {
    "loss": 0.08302205204963684,
    "grad_norm": 0.7470789551734924,
    "learning_rate": 7.404549509730135e-05,
    "epoch": 0.584943885262244,
    "step": 18190
  },
  {
    "loss": 0.07486485242843628,
    "grad_norm": 0.6482924818992615,
    "learning_rate": 7.394763015236117e-05,
    "epoch": 0.5852654596906454,
    "step": 18200
  },
  {
    "loss": 0.06772093176841736,
    "grad_norm": 0.8526209592819214,
    "learning_rate": 7.384979196874627e-05,
    "epoch": 0.5855870341190469,
    "step": 18210
  },
  {
    "loss": 0.10163276195526123,
    "grad_norm": 0.48233070969581604,
    "learning_rate": 7.375198064695724e-05,
    "epoch": 0.5859086085474483,
    "step": 18220
  },
  {
    "loss": 0.09970284700393676,
    "grad_norm": 0.8146294951438904,
    "learning_rate": 7.365419628746713e-05,
    "epoch": 0.5862301829758497,
    "step": 18230
  },
  {
    "loss": 0.07390713691711426,
    "grad_norm": 0.48546773195266724,
    "learning_rate": 7.355643899072123e-05,
    "epoch": 0.5865517574042513,
    "step": 18240
  },
  {
    "loss": 0.07537122964859008,
    "grad_norm": 0.6072354316711426,
    "learning_rate": 7.345870885713717e-05,
    "epoch": 0.5868733318326527,
    "step": 18250
  },
  {
    "loss": 0.07224909067153931,
    "grad_norm": 0.807327389717102,
    "learning_rate": 7.336100598710452e-05,
    "epoch": 0.5871949062610541,
    "step": 18260
  },
  {
    "loss": 0.07525655627250671,
    "grad_norm": 0.7244537472724915,
    "learning_rate": 7.326333048098494e-05,
    "epoch": 0.5875164806894556,
    "step": 18270
  },
  {
    "loss": 0.06138322353363037,
    "grad_norm": 0.3798077702522278,
    "learning_rate": 7.316568243911193e-05,
    "epoch": 0.587838055117857,
    "step": 18280
  },
  {
    "loss": 0.0677099347114563,
    "grad_norm": 0.6162866353988647,
    "learning_rate": 7.306806196179079e-05,
    "epoch": 0.5881596295462584,
    "step": 18290
  },
  {
    "loss": 0.09046462774276734,
    "grad_norm": 0.4453006684780121,
    "learning_rate": 7.297046914929853e-05,
    "epoch": 0.58848120397466,
    "step": 18300
  },
  {
    "loss": 0.0929905891418457,
    "grad_norm": 0.745503842830658,
    "learning_rate": 7.287290410188373e-05,
    "epoch": 0.5888027784030614,
    "step": 18310
  },
  {
    "loss": 0.07695024609565734,
    "grad_norm": 0.38427412509918213,
    "learning_rate": 7.277536691976646e-05,
    "epoch": 0.5891243528314628,
    "step": 18320
  },
  {
    "loss": 0.0723729133605957,
    "grad_norm": 0.40777745842933655,
    "learning_rate": 7.267785770313811e-05,
    "epoch": 0.5894459272598643,
    "step": 18330
  },
  {
    "loss": 0.08466371297836303,
    "grad_norm": 0.6734020709991455,
    "learning_rate": 7.258037655216143e-05,
    "epoch": 0.5897675016882657,
    "step": 18340
  },
  {
    "loss": 0.07116898298263549,
    "grad_norm": 0.6805622577667236,
    "learning_rate": 7.248292356697027e-05,
    "epoch": 0.5900890761166672,
    "step": 18350
  },
  {
    "loss": 0.07267593145370484,
    "grad_norm": 0.6734875440597534,
    "learning_rate": 7.238549884766956e-05,
    "epoch": 0.5904106505450687,
    "step": 18360
  },
  {
    "loss": 0.06956661939620971,
    "grad_norm": 0.8455624580383301,
    "learning_rate": 7.228810249433527e-05,
    "epoch": 0.5907322249734701,
    "step": 18370
  },
  {
    "loss": 0.0653856337070465,
    "grad_norm": 0.5372180342674255,
    "learning_rate": 7.21907346070141e-05,
    "epoch": 0.5910537994018715,
    "step": 18380
  },
  {
    "loss": 0.08260570168495178,
    "grad_norm": 0.6930351257324219,
    "learning_rate": 7.209339528572363e-05,
    "epoch": 0.591375373830273,
    "step": 18390
  },
  {
    "loss": 0.07606642842292785,
    "grad_norm": 0.4146069586277008,
    "learning_rate": 7.199608463045204e-05,
    "epoch": 0.5916969482586745,
    "step": 18400
  },
  {
    "loss": 0.07034127712249756,
    "grad_norm": 0.39609354734420776,
    "learning_rate": 7.189880274115806e-05,
    "epoch": 0.5920185226870759,
    "step": 18410
  },
  {
    "loss": 0.08222041130065919,
    "grad_norm": 0.7135039567947388,
    "learning_rate": 7.180154971777087e-05,
    "epoch": 0.5923400971154774,
    "step": 18420
  },
  {
    "loss": 0.06555980443954468,
    "grad_norm": 0.7420735955238342,
    "learning_rate": 7.170432566019001e-05,
    "epoch": 0.5926616715438788,
    "step": 18430
  },
  {
    "loss": 0.09034892320632934,
    "grad_norm": 0.7950654029846191,
    "learning_rate": 7.160713066828531e-05,
    "epoch": 0.5929832459722802,
    "step": 18440
  },
  {
    "loss": 0.07953452467918395,
    "grad_norm": 0.7548782825469971,
    "learning_rate": 7.150996484189665e-05,
    "epoch": 0.5933048204006818,
    "step": 18450
  },
  {
    "loss": 0.06989264488220215,
    "grad_norm": 0.4734945595264435,
    "learning_rate": 7.141282828083404e-05,
    "epoch": 0.5936263948290832,
    "step": 18460
  },
  {
    "loss": 0.07788462042808533,
    "grad_norm": 0.5163804292678833,
    "learning_rate": 7.13157210848774e-05,
    "epoch": 0.5939479692574846,
    "step": 18470
  },
  {
    "loss": 0.08228410482406616,
    "grad_norm": 0.49889716506004333,
    "learning_rate": 7.121864335377644e-05,
    "epoch": 0.5942695436858861,
    "step": 18480
  },
  {
    "loss": 0.09119044542312622,
    "grad_norm": 0.8352193236351013,
    "learning_rate": 7.112159518725057e-05,
    "epoch": 0.5945911181142876,
    "step": 18490
  },
  {
    "loss": 0.09865041375160218,
    "grad_norm": 0.9083040356636047,
    "learning_rate": 7.102457668498906e-05,
    "epoch": 0.594912692542689,
    "step": 18500
  },
  {
    "loss": 0.07140249013900757,
    "grad_norm": 0.5583008527755737,
    "learning_rate": 7.092758794665042e-05,
    "epoch": 0.5952342669710905,
    "step": 18510
  },
  {
    "loss": 0.05078734159469604,
    "grad_norm": 0.6857233047485352,
    "learning_rate": 7.083062907186281e-05,
    "epoch": 0.5955558413994919,
    "step": 18520
  },
  {
    "loss": 0.10113188028335571,
    "grad_norm": 0.8305119872093201,
    "learning_rate": 7.073370016022354e-05,
    "epoch": 0.5958774158278933,
    "step": 18530
  },
  {
    "loss": 0.0677301526069641,
    "grad_norm": 0.37434622645378113,
    "learning_rate": 7.063680131129924e-05,
    "epoch": 0.5961989902562949,
    "step": 18540
  },
  {
    "loss": 0.07215762138366699,
    "grad_norm": 0.38779565691947937,
    "learning_rate": 7.053993262462566e-05,
    "epoch": 0.5965205646846963,
    "step": 18550
  },
  {
    "loss": 0.06991016268730163,
    "grad_norm": 0.5665784478187561,
    "learning_rate": 7.044309419970747e-05,
    "epoch": 0.5968421391130977,
    "step": 18560
  },
  {
    "loss": 0.09297497868537903,
    "grad_norm": 0.6514776945114136,
    "learning_rate": 7.034628613601846e-05,
    "epoch": 0.5971637135414992,
    "step": 18570
  },
  {
    "loss": 0.0842542052268982,
    "grad_norm": 0.755832850933075,
    "learning_rate": 7.024950853300101e-05,
    "epoch": 0.5974852879699006,
    "step": 18580
  },
  {
    "loss": 0.08134500980377198,
    "grad_norm": 0.5388493537902832,
    "learning_rate": 7.015276149006633e-05,
    "epoch": 0.597806862398302,
    "step": 18590
  },
  {
    "loss": 0.06592969298362732,
    "grad_norm": 0.3698655664920807,
    "learning_rate": 7.005604510659421e-05,
    "epoch": 0.5981284368267036,
    "step": 18600
  },
  {
    "loss": 0.09740493297576905,
    "grad_norm": 0.7075668573379517,
    "learning_rate": 6.995935948193294e-05,
    "epoch": 0.598450011255105,
    "step": 18610
  },
  {
    "loss": 0.08867404460906983,
    "grad_norm": 0.6851850152015686,
    "learning_rate": 6.986270471539922e-05,
    "epoch": 0.5987715856835064,
    "step": 18620
  },
  {
    "loss": 0.10935039520263672,
    "grad_norm": 0.5081934928894043,
    "learning_rate": 6.976608090627807e-05,
    "epoch": 0.5990931601119079,
    "step": 18630
  },
  {
    "loss": 0.09188774824142457,
    "grad_norm": 0.5268943905830383,
    "learning_rate": 6.966948815382268e-05,
    "epoch": 0.5994147345403094,
    "step": 18640
  },
  {
    "eval_loss": 0.07528992742300034,
    "eval_runtime": 34.3822,
    "eval_samples_per_second": 146.18,
    "eval_steps_per_second": 36.56,
    "epoch": 0.5996719940830305,
    "step": 18648
  },
  {
    "loss": 0.07428829073905945,
    "grad_norm": 0.4399220049381256,
    "learning_rate": 6.957292655725437e-05,
    "epoch": 0.5997363089687108,
    "step": 18650
  },
  {
    "loss": 0.09513623714447021,
    "grad_norm": 0.7418004870414734,
    "learning_rate": 6.947639621576243e-05,
    "epoch": 0.6000578833971123,
    "step": 18660
  },
  {
    "loss": 0.09230533838272095,
    "grad_norm": 0.6606807112693787,
    "learning_rate": 6.937989722850405e-05,
    "epoch": 0.6003794578255137,
    "step": 18670
  },
  {
    "loss": 0.10270729064941406,
    "grad_norm": 0.7914863228797913,
    "learning_rate": 6.928342969460418e-05,
    "epoch": 0.6007010322539151,
    "step": 18680
  },
  {
    "loss": 0.07258764505386353,
    "grad_norm": 0.5238516330718994,
    "learning_rate": 6.918699371315556e-05,
    "epoch": 0.6010226066823167,
    "step": 18690
  },
  {
    "loss": 0.08154450654983521,
    "grad_norm": 0.6537436246871948,
    "learning_rate": 6.909058938321842e-05,
    "epoch": 0.6013441811107181,
    "step": 18700
  },
  {
    "loss": 0.065982985496521,
    "grad_norm": 0.4509539306163788,
    "learning_rate": 6.899421680382053e-05,
    "epoch": 0.6016657555391195,
    "step": 18710
  },
  {
    "loss": 0.0736770749092102,
    "grad_norm": 0.6178690195083618,
    "learning_rate": 6.889787607395702e-05,
    "epoch": 0.601987329967521,
    "step": 18720
  },
  {
    "loss": 0.09866932630538941,
    "grad_norm": 0.47050192952156067,
    "learning_rate": 6.88015672925903e-05,
    "epoch": 0.6023089043959224,
    "step": 18730
  },
  {
    "loss": 0.08328339457511902,
    "grad_norm": 0.5887206792831421,
    "learning_rate": 6.870529055865e-05,
    "epoch": 0.6026304788243239,
    "step": 18740
  },
  {
    "loss": 0.06292853951454162,
    "grad_norm": 0.6852194666862488,
    "learning_rate": 6.860904597103273e-05,
    "epoch": 0.6029520532527254,
    "step": 18750
  },
  {
    "loss": 0.0628040075302124,
    "grad_norm": 0.7905822992324829,
    "learning_rate": 6.851283362860226e-05,
    "epoch": 0.6032736276811268,
    "step": 18760
  },
  {
    "loss": 0.07740421295166015,
    "grad_norm": 0.6432583928108215,
    "learning_rate": 6.841665363018907e-05,
    "epoch": 0.6035952021095282,
    "step": 18770
  },
  {
    "loss": 0.0845758557319641,
    "grad_norm": 0.6782318353652954,
    "learning_rate": 6.832050607459049e-05,
    "epoch": 0.6039167765379297,
    "step": 18780
  },
  {
    "loss": 0.07212685942649841,
    "grad_norm": 0.7194960713386536,
    "learning_rate": 6.822439106057051e-05,
    "epoch": 0.6042383509663312,
    "step": 18790
  },
  {
    "loss": 0.08795191645622254,
    "grad_norm": 0.4270845055580139,
    "learning_rate": 6.812830868685969e-05,
    "epoch": 0.6045599253947326,
    "step": 18800
  },
  {
    "loss": 0.07008818984031677,
    "grad_norm": 0.4772111177444458,
    "learning_rate": 6.803225905215503e-05,
    "epoch": 0.6048814998231341,
    "step": 18810
  },
  {
    "loss": 0.0846167504787445,
    "grad_norm": 0.7406246662139893,
    "learning_rate": 6.793624225511998e-05,
    "epoch": 0.6052030742515355,
    "step": 18820
  },
  {
    "loss": 0.0772022008895874,
    "grad_norm": 0.9471773505210876,
    "learning_rate": 6.784025839438425e-05,
    "epoch": 0.6055246486799369,
    "step": 18830
  },
  {
    "loss": 0.08314455151557923,
    "grad_norm": 1.0049670934677124,
    "learning_rate": 6.774430756854361e-05,
    "epoch": 0.6058462231083385,
    "step": 18840
  },
  {
    "loss": 0.09380330443382263,
    "grad_norm": 0.6991022229194641,
    "learning_rate": 6.764838987615998e-05,
    "epoch": 0.6061677975367399,
    "step": 18850
  },
  {
    "loss": 0.09013092517852783,
    "grad_norm": 0.8749735951423645,
    "learning_rate": 6.755250541576126e-05,
    "epoch": 0.6064893719651413,
    "step": 18860
  },
  {
    "loss": 0.08810775876045226,
    "grad_norm": 0.5447385311126709,
    "learning_rate": 6.745665428584111e-05,
    "epoch": 0.6068109463935428,
    "step": 18870
  },
  {
    "loss": 0.10087782144546509,
    "grad_norm": 0.35421833395957947,
    "learning_rate": 6.736083658485916e-05,
    "epoch": 0.6071325208219442,
    "step": 18880
  },
  {
    "loss": 0.06867042183876038,
    "grad_norm": 0.5186894536018372,
    "learning_rate": 6.726505241124048e-05,
    "epoch": 0.6074540952503457,
    "step": 18890
  },
  {
    "loss": 0.07856658101081848,
    "grad_norm": 0.4858158528804779,
    "learning_rate": 6.71693018633758e-05,
    "epoch": 0.6077756696787472,
    "step": 18900
  },
  {
    "loss": 0.07544032335281373,
    "grad_norm": 0.5628518462181091,
    "learning_rate": 6.707358503962131e-05,
    "epoch": 0.6080972441071486,
    "step": 18910
  },
  {
    "loss": 0.0858623206615448,
    "grad_norm": 0.5891218781471252,
    "learning_rate": 6.697790203829854e-05,
    "epoch": 0.60841881853555,
    "step": 18920
  },
  {
    "loss": 0.08889785408973694,
    "grad_norm": 0.7737987041473389,
    "learning_rate": 6.68822529576943e-05,
    "epoch": 0.6087403929639515,
    "step": 18930
  },
  {
    "loss": 0.07854949235916138,
    "grad_norm": 0.630588173866272,
    "learning_rate": 6.67866378960605e-05,
    "epoch": 0.609061967392353,
    "step": 18940
  },
  {
    "loss": 0.1184352993965149,
    "grad_norm": 0.6254100203514099,
    "learning_rate": 6.66910569516142e-05,
    "epoch": 0.6093835418207544,
    "step": 18950
  },
  {
    "loss": 0.08089698553085327,
    "grad_norm": 0.3980773985385895,
    "learning_rate": 6.659551022253734e-05,
    "epoch": 0.6097051162491559,
    "step": 18960
  },
  {
    "loss": 0.08007602095603943,
    "grad_norm": 0.7637392282485962,
    "learning_rate": 6.649999780697671e-05,
    "epoch": 0.6100266906775573,
    "step": 18970
  },
  {
    "loss": 0.0555802583694458,
    "grad_norm": 0.632079005241394,
    "learning_rate": 6.640451980304392e-05,
    "epoch": 0.6103482651059587,
    "step": 18980
  },
  {
    "loss": 0.08289624452590942,
    "grad_norm": 0.23457559943199158,
    "learning_rate": 6.630907630881516e-05,
    "epoch": 0.6106698395343603,
    "step": 18990
  },
  {
    "loss": 0.06061472296714783,
    "grad_norm": 0.5475881695747375,
    "learning_rate": 6.621366742233116e-05,
    "epoch": 0.6109914139627617,
    "step": 19000
  },
  {
    "loss": 0.08997538685798645,
    "grad_norm": 0.34357795119285583,
    "learning_rate": 6.611829324159722e-05,
    "epoch": 0.6113129883911631,
    "step": 19010
  },
  {
    "loss": 0.0812469482421875,
    "grad_norm": 0.8159956932067871,
    "learning_rate": 6.602295386458289e-05,
    "epoch": 0.6116345628195646,
    "step": 19020
  },
  {
    "loss": 0.10333518981933594,
    "grad_norm": 0.8582615256309509,
    "learning_rate": 6.592764938922193e-05,
    "epoch": 0.611956137247966,
    "step": 19030
  },
  {
    "loss": 0.07760070562362671,
    "grad_norm": 0.3401944637298584,
    "learning_rate": 6.583237991341239e-05,
    "epoch": 0.6122777116763675,
    "step": 19040
  },
  {
    "loss": 0.07692185044288635,
    "grad_norm": 0.4970964193344116,
    "learning_rate": 6.573714553501626e-05,
    "epoch": 0.612599286104769,
    "step": 19050
  },
  {
    "loss": 0.07856242656707764,
    "grad_norm": 0.5750195980072021,
    "learning_rate": 6.564194635185942e-05,
    "epoch": 0.6129208605331704,
    "step": 19060
  },
  {
    "loss": 0.08100840449333191,
    "grad_norm": 0.7286192178726196,
    "learning_rate": 6.55467824617318e-05,
    "epoch": 0.6132424349615718,
    "step": 19070
  },
  {
    "loss": 0.08202658891677857,
    "grad_norm": 0.5252206325531006,
    "learning_rate": 6.545165396238691e-05,
    "epoch": 0.6135640093899734,
    "step": 19080
  },
  {
    "loss": 0.08576998710632325,
    "grad_norm": 0.4926624894142151,
    "learning_rate": 6.535656095154193e-05,
    "epoch": 0.6138855838183748,
    "step": 19090
  },
  {
    "loss": 0.08644883036613464,
    "grad_norm": 0.5414864420890808,
    "learning_rate": 6.526150352687766e-05,
    "epoch": 0.6142071582467762,
    "step": 19100
  },
  {
    "loss": 0.08598004579544068,
    "grad_norm": 0.730283796787262,
    "learning_rate": 6.516648178603826e-05,
    "epoch": 0.6145287326751777,
    "step": 19110
  },
  {
    "loss": 0.07616907358169556,
    "grad_norm": 0.5509003400802612,
    "learning_rate": 6.507149582663126e-05,
    "epoch": 0.6148503071035791,
    "step": 19120
  },
  {
    "loss": 0.07464420199394226,
    "grad_norm": 0.627088725566864,
    "learning_rate": 6.497654574622741e-05,
    "epoch": 0.6151718815319805,
    "step": 19130
  },
  {
    "loss": 0.06419848203659058,
    "grad_norm": 0.7218136191368103,
    "learning_rate": 6.488163164236075e-05,
    "epoch": 0.6154934559603821,
    "step": 19140
  },
  {
    "loss": 0.08014584183692933,
    "grad_norm": 0.8419432044029236,
    "learning_rate": 6.478675361252818e-05,
    "epoch": 0.6158150303887835,
    "step": 19150
  },
  {
    "loss": 0.055648809671401976,
    "grad_norm": 0.5282542109489441,
    "learning_rate": 6.46919117541896e-05,
    "epoch": 0.6161366048171849,
    "step": 19160
  },
  {
    "loss": 0.06651024222373962,
    "grad_norm": 0.4777286648750305,
    "learning_rate": 6.459710616476782e-05,
    "epoch": 0.6164581792455864,
    "step": 19170
  },
  {
    "loss": 0.0624203085899353,
    "grad_norm": 0.6265890598297119,
    "learning_rate": 6.450233694164832e-05,
    "epoch": 0.6167797536739879,
    "step": 19180
  },
  {
    "loss": 0.07577801942825317,
    "grad_norm": 0.5991108417510986,
    "learning_rate": 6.440760418217923e-05,
    "epoch": 0.6171013281023893,
    "step": 19190
  },
  {
    "loss": 0.06770421266555786,
    "grad_norm": 0.5578089952468872,
    "learning_rate": 6.431290798367129e-05,
    "epoch": 0.6174229025307908,
    "step": 19200
  },
  {
    "loss": 0.08427031636238098,
    "grad_norm": 0.7189592719078064,
    "learning_rate": 6.421824844339761e-05,
    "epoch": 0.6177444769591922,
    "step": 19210
  },
  {
    "loss": 0.0903325617313385,
    "grad_norm": 0.4093859791755676,
    "learning_rate": 6.41236256585937e-05,
    "epoch": 0.6180660513875936,
    "step": 19220
  },
  {
    "loss": 0.08933579921722412,
    "grad_norm": 0.7538068890571594,
    "learning_rate": 6.402903972645722e-05,
    "epoch": 0.6183876258159952,
    "step": 19230
  },
  {
    "loss": 0.08776495456695557,
    "grad_norm": 0.6686062812805176,
    "learning_rate": 6.393449074414812e-05,
    "epoch": 0.6187092002443966,
    "step": 19240
  },
  {
    "loss": 0.08356345891952514,
    "grad_norm": 0.9015601873397827,
    "learning_rate": 6.383997880878828e-05,
    "epoch": 0.619030774672798,
    "step": 19250
  },
  {
    "loss": 0.08526620864868165,
    "grad_norm": 0.9928838610649109,
    "learning_rate": 6.374550401746154e-05,
    "epoch": 0.6193523491011995,
    "step": 19260
  },
  {
    "loss": 0.07604313492774964,
    "grad_norm": 0.8028160333633423,
    "learning_rate": 6.365106646721364e-05,
    "epoch": 0.6196739235296009,
    "step": 19270
  },
  {
    "loss": 0.07956154346466064,
    "grad_norm": 0.8392983675003052,
    "learning_rate": 6.355666625505201e-05,
    "epoch": 0.6199954979580024,
    "step": 19280
  },
  {
    "loss": 0.08473367691040039,
    "grad_norm": 0.5166031122207642,
    "learning_rate": 6.346230347794577e-05,
    "epoch": 0.6203170723864039,
    "step": 19290
  },
  {
    "loss": 0.09186525344848633,
    "grad_norm": 0.8711339235305786,
    "learning_rate": 6.336797823282556e-05,
    "epoch": 0.6206386468148053,
    "step": 19300
  },
  {
    "loss": 0.07015078067779541,
    "grad_norm": 0.48920947313308716,
    "learning_rate": 6.327369061658344e-05,
    "epoch": 0.6209602212432067,
    "step": 19310
  },
  {
    "loss": 0.06253830194473267,
    "grad_norm": 1.0902957916259766,
    "learning_rate": 6.317944072607285e-05,
    "epoch": 0.6212817956716082,
    "step": 19320
  },
  {
    "loss": 0.07039433717727661,
    "grad_norm": 0.9401423335075378,
    "learning_rate": 6.308522865810852e-05,
    "epoch": 0.6216033701000097,
    "step": 19330
  },
  {
    "loss": 0.07885023355484008,
    "grad_norm": 1.2769575119018555,
    "learning_rate": 6.299105450946626e-05,
    "epoch": 0.6219249445284111,
    "step": 19340
  },
  {
    "loss": 0.07023831605911254,
    "grad_norm": 0.8095871210098267,
    "learning_rate": 6.289691837688291e-05,
    "epoch": 0.6222465189568126,
    "step": 19350
  },
  {
    "loss": 0.06982327699661255,
    "grad_norm": 0.3558923304080963,
    "learning_rate": 6.28028203570564e-05,
    "epoch": 0.622568093385214,
    "step": 19360
  },
  {
    "loss": 0.08719192147254944,
    "grad_norm": 0.6851743459701538,
    "learning_rate": 6.270876054664531e-05,
    "epoch": 0.6228896678136154,
    "step": 19370
  },
  {
    "loss": 0.08308027982711792,
    "grad_norm": 0.3055587112903595,
    "learning_rate": 6.261473904226908e-05,
    "epoch": 0.623211242242017,
    "step": 19380
  },
  {
    "loss": 0.08074118494987488,
    "grad_norm": 0.8412294983863831,
    "learning_rate": 6.252075594050785e-05,
    "epoch": 0.6235328166704184,
    "step": 19390
  },
  {
    "loss": 0.07668668031692505,
    "grad_norm": 0.8082206845283508,
    "learning_rate": 6.242681133790225e-05,
    "epoch": 0.6238543910988198,
    "step": 19400
  },
  {
    "loss": 0.07915959358215333,
    "grad_norm": 0.6886305212974548,
    "learning_rate": 6.233290533095335e-05,
    "epoch": 0.6241759655272213,
    "step": 19410
  },
  {
    "loss": 0.0796347975730896,
    "grad_norm": 0.6029472947120667,
    "learning_rate": 6.223903801612258e-05,
    "epoch": 0.6244975399556227,
    "step": 19420
  },
  {
    "loss": 0.09030648469924926,
    "grad_norm": 0.8511977791786194,
    "learning_rate": 6.214520948983166e-05,
    "epoch": 0.6248191143840242,
    "step": 19430
  },
  {
    "loss": 0.07107316851615905,
    "grad_norm": 0.7600314617156982,
    "learning_rate": 6.205141984846244e-05,
    "epoch": 0.6251406888124257,
    "step": 19440
  },
  {
    "loss": 0.08018980622291565,
    "grad_norm": 0.592937707901001,
    "learning_rate": 6.195766918835678e-05,
    "epoch": 0.6254622632408271,
    "step": 19450
  },
  {
    "loss": 0.08703637719154358,
    "grad_norm": 0.5489158630371094,
    "learning_rate": 6.186395760581664e-05,
    "epoch": 0.6257838376692285,
    "step": 19460
  },
  {
    "loss": 0.07732821106910706,
    "grad_norm": 0.2843504548072815,
    "learning_rate": 6.177028519710369e-05,
    "epoch": 0.62610541209763,
    "step": 19470
  },
  {
    "loss": 0.10473834276199341,
    "grad_norm": 0.7938198447227478,
    "learning_rate": 6.167665205843944e-05,
    "epoch": 0.6264269865260315,
    "step": 19480
  },
  {
    "loss": 0.09615716934204102,
    "grad_norm": 1.1492993831634521,
    "learning_rate": 6.158305828600502e-05,
    "epoch": 0.6267485609544329,
    "step": 19490
  },
  {
    "loss": 0.06346037983894348,
    "grad_norm": 0.5835652351379395,
    "learning_rate": 6.148950397594117e-05,
    "epoch": 0.6270701353828344,
    "step": 19500
  },
  {
    "loss": 0.072739577293396,
    "grad_norm": 1.014746069908142,
    "learning_rate": 6.139598922434802e-05,
    "epoch": 0.6273917098112358,
    "step": 19510
  },
  {
    "loss": 0.09916123151779174,
    "grad_norm": 0.715579628944397,
    "learning_rate": 6.130251412728516e-05,
    "epoch": 0.6277132842396372,
    "step": 19520
  },
  {
    "loss": 0.07230787873268127,
    "grad_norm": 0.7324387431144714,
    "learning_rate": 6.120907878077138e-05,
    "epoch": 0.6280348586680388,
    "step": 19530
  },
  {
    "loss": 0.09201850891113281,
    "grad_norm": 0.526698648929596,
    "learning_rate": 6.111568328078465e-05,
    "epoch": 0.6283564330964402,
    "step": 19540
  },
  {
    "loss": 0.08800469636917115,
    "grad_norm": 0.989494800567627,
    "learning_rate": 6.102232772326202e-05,
    "epoch": 0.6286780075248416,
    "step": 19550
  },
  {
    "loss": 0.08789398670196533,
    "grad_norm": 0.9045369625091553,
    "learning_rate": 6.092901220409949e-05,
    "epoch": 0.6289995819532431,
    "step": 19560
  },
  {
    "loss": 0.06860705614089965,
    "grad_norm": 0.7513046860694885,
    "learning_rate": 6.0835736819151946e-05,
    "epoch": 0.6293211563816445,
    "step": 19570
  },
  {
    "loss": 0.06425130963325501,
    "grad_norm": 0.6448571085929871,
    "learning_rate": 6.074250166423308e-05,
    "epoch": 0.629642730810046,
    "step": 19580
  },
  {
    "loss": 0.08776512742042542,
    "grad_norm": 0.6769627928733826,
    "learning_rate": 6.0649306835115206e-05,
    "epoch": 0.6299643052384475,
    "step": 19590
  },
  {
    "loss": 0.08552072048187256,
    "grad_norm": 0.45814409852027893,
    "learning_rate": 6.055615242752923e-05,
    "epoch": 0.6302858796668489,
    "step": 19600
  },
  {
    "loss": 0.10555309057235718,
    "grad_norm": 1.470810055732727,
    "learning_rate": 6.046303853716455e-05,
    "epoch": 0.6306074540952503,
    "step": 19610
  },
  {
    "loss": 0.07876392602920532,
    "grad_norm": 0.4727279543876648,
    "learning_rate": 6.036996525966896e-05,
    "epoch": 0.6309290285236518,
    "step": 19620
  },
  {
    "loss": 0.06964776515960694,
    "grad_norm": 0.635856568813324,
    "learning_rate": 6.027693269064849e-05,
    "epoch": 0.6312506029520533,
    "step": 19630
  },
  {
    "loss": 0.07364420294761657,
    "grad_norm": 0.34876182675361633,
    "learning_rate": 6.0183940925667326e-05,
    "epoch": 0.6315721773804547,
    "step": 19640
  },
  {
    "loss": 0.08741930723190308,
    "grad_norm": 0.3650606870651245,
    "learning_rate": 6.009099006024788e-05,
    "epoch": 0.6318937518088562,
    "step": 19650
  },
  {
    "loss": 0.07190605998039246,
    "grad_norm": 0.8150784373283386,
    "learning_rate": 5.999808018987041e-05,
    "epoch": 0.6322153262372576,
    "step": 19660
  },
  {
    "loss": 0.08492738008499146,
    "grad_norm": 0.7493435144424438,
    "learning_rate": 5.990521140997316e-05,
    "epoch": 0.632536900665659,
    "step": 19670
  },
  {
    "loss": 0.08940092325210572,
    "grad_norm": 0.8140876293182373,
    "learning_rate": 5.9812383815952066e-05,
    "epoch": 0.6328584750940606,
    "step": 19680
  },
  {
    "loss": 0.09434642791748046,
    "grad_norm": 0.6462841033935547,
    "learning_rate": 5.971959750316083e-05,
    "epoch": 0.633180049522462,
    "step": 19690
  },
  {
    "loss": 0.07376375794410706,
    "grad_norm": 0.45153164863586426,
    "learning_rate": 5.962685256691071e-05,
    "epoch": 0.6335016239508634,
    "step": 19700
  },
  {
    "loss": 0.08791798949241639,
    "grad_norm": 0.5423551201820374,
    "learning_rate": 5.953414910247054e-05,
    "epoch": 0.6338231983792649,
    "step": 19710
  },
  {
    "loss": 0.08712515830993653,
    "grad_norm": 0.47862866520881653,
    "learning_rate": 5.944148720506648e-05,
    "epoch": 0.6341447728076663,
    "step": 19720
  },
  {
    "loss": 0.07166895866394044,
    "grad_norm": 0.6706460118293762,
    "learning_rate": 5.9348866969881975e-05,
    "epoch": 0.6344663472360678,
    "step": 19730
  },
  {
    "loss": 0.07896190881729126,
    "grad_norm": 0.8685805201530457,
    "learning_rate": 5.9256288492057734e-05,
    "epoch": 0.6347879216644693,
    "step": 19740
  },
  {
    "loss": 0.078327876329422,
    "grad_norm": 0.5587207674980164,
    "learning_rate": 5.916375186669153e-05,
    "epoch": 0.6351094960928707,
    "step": 19750
  },
  {
    "loss": 0.06806607842445374,
    "grad_norm": 0.29360270500183105,
    "learning_rate": 5.907125718883812e-05,
    "epoch": 0.6354310705212721,
    "step": 19760
  },
  {
    "loss": 0.09198864698410034,
    "grad_norm": 1.0249218940734863,
    "learning_rate": 5.897880455350933e-05,
    "epoch": 0.6357526449496737,
    "step": 19770
  },
  {
    "loss": 0.07347887754440308,
    "grad_norm": 0.3725737929344177,
    "learning_rate": 5.888639405567359e-05,
    "epoch": 0.6360742193780751,
    "step": 19780
  },
  {
    "loss": 0.07164725065231323,
    "grad_norm": 0.49716490507125854,
    "learning_rate": 5.879402579025615e-05,
    "epoch": 0.6363957938064765,
    "step": 19790
  },
  {
    "loss": 0.08011369109153747,
    "grad_norm": 0.7196645736694336,
    "learning_rate": 5.8701699852138867e-05,
    "epoch": 0.636717368234878,
    "step": 19800
  },
  {
    "loss": 0.07078960537910461,
    "grad_norm": 0.4283914864063263,
    "learning_rate": 5.860941633616015e-05,
    "epoch": 0.6370389426632794,
    "step": 19810
  },
  {
    "loss": 0.09213285446166992,
    "grad_norm": 0.6686605215072632,
    "learning_rate": 5.851717533711475e-05,
    "epoch": 0.6373605170916808,
    "step": 19820
  },
  {
    "loss": 0.07392160296440124,
    "grad_norm": 0.7527466416358948,
    "learning_rate": 5.8424976949753796e-05,
    "epoch": 0.6376820915200824,
    "step": 19830
  },
  {
    "loss": 0.06171455979347229,
    "grad_norm": 0.5245692729949951,
    "learning_rate": 5.83328212687847e-05,
    "epoch": 0.6380036659484838,
    "step": 19840
  },
  {
    "loss": 0.10391281843185425,
    "grad_norm": 0.3888297379016876,
    "learning_rate": 5.824070838887091e-05,
    "epoch": 0.6383252403768852,
    "step": 19850
  },
  {
    "loss": 0.09777231216430664,
    "grad_norm": 0.5975055694580078,
    "learning_rate": 5.814863840463192e-05,
    "epoch": 0.6386468148052867,
    "step": 19860
  },
  {
    "loss": 0.0733383059501648,
    "grad_norm": 0.4285428524017334,
    "learning_rate": 5.8056611410643246e-05,
    "epoch": 0.6389683892336881,
    "step": 19870
  },
  {
    "loss": 0.08428534269332885,
    "grad_norm": 1.0025172233581543,
    "learning_rate": 5.7964627501436116e-05,
    "epoch": 0.6392899636620896,
    "step": 19880
  },
  {
    "loss": 0.08907008171081543,
    "grad_norm": 0.6572055220603943,
    "learning_rate": 5.7872686771497606e-05,
    "epoch": 0.6396115380904911,
    "step": 19890
  },
  {
    "loss": 0.07242008447647094,
    "grad_norm": 0.5621013045310974,
    "learning_rate": 5.7780789315270444e-05,
    "epoch": 0.6399331125188925,
    "step": 19900
  },
  {
    "loss": 0.0710676670074463,
    "grad_norm": 0.3812589943408966,
    "learning_rate": 5.768893522715279e-05,
    "epoch": 0.6402546869472939,
    "step": 19910
  },
  {
    "loss": 0.09399187564849854,
    "grad_norm": 0.8854357004165649,
    "learning_rate": 5.75971246014984e-05,
    "epoch": 0.6405762613756955,
    "step": 19920
  },
  {
    "loss": 0.08624742031097413,
    "grad_norm": 0.3859586715698242,
    "learning_rate": 5.750535753261623e-05,
    "epoch": 0.6408978358040969,
    "step": 19930
  },
  {
    "loss": 0.08188838958740234,
    "grad_norm": 0.7565677165985107,
    "learning_rate": 5.7413634114770676e-05,
    "epoch": 0.6412194102324983,
    "step": 19940
  },
  {
    "loss": 0.08392425179481507,
    "grad_norm": 0.4465886950492859,
    "learning_rate": 5.7321954442181114e-05,
    "epoch": 0.6415409846608998,
    "step": 19950
  },
  {
    "loss": 0.08943371772766114,
    "grad_norm": 0.901117205619812,
    "learning_rate": 5.7230318609022106e-05,
    "epoch": 0.6418625590893012,
    "step": 19960
  },
  {
    "loss": 0.09252094030380249,
    "grad_norm": 0.49248000979423523,
    "learning_rate": 5.7138726709423175e-05,
    "epoch": 0.6421841335177026,
    "step": 19970
  },
  {
    "loss": 0.07759752869606018,
    "grad_norm": 0.3645312190055847,
    "learning_rate": 5.704717883746861e-05,
    "epoch": 0.6425057079461042,
    "step": 19980
  },
  {
    "loss": 0.07568894028663635,
    "grad_norm": 0.5617930889129639,
    "learning_rate": 5.695567508719762e-05,
    "epoch": 0.6428272823745056,
    "step": 19990
  },
  {
    "loss": 0.0660637617111206,
    "grad_norm": 0.4228644073009491,
    "learning_rate": 5.6864215552603974e-05,
    "epoch": 0.643148856802907,
    "step": 20000
  },
  {
    "loss": 0.08293818831443786,
    "grad_norm": 0.5796824097633362,
    "learning_rate": 5.67728003276361e-05,
    "epoch": 0.6434704312313085,
    "step": 20010
  },
  {
    "loss": 0.08752073645591736,
    "grad_norm": 0.6435496211051941,
    "learning_rate": 5.6681429506196825e-05,
    "epoch": 0.64379200565971,
    "step": 20020
  },
  {
    "loss": 0.08041639924049378,
    "grad_norm": 0.40585389733314514,
    "learning_rate": 5.659010318214346e-05,
    "epoch": 0.6441135800881114,
    "step": 20030
  },
  {
    "loss": 0.07967058420181275,
    "grad_norm": 0.5377001762390137,
    "learning_rate": 5.64988214492876e-05,
    "epoch": 0.6444351545165129,
    "step": 20040
  },
  {
    "loss": 0.07384193539619446,
    "grad_norm": 0.4058140516281128,
    "learning_rate": 5.640758440139491e-05,
    "epoch": 0.6447567289449143,
    "step": 20050
  },
  {
    "loss": 0.07030548453330994,
    "grad_norm": 0.3950488567352295,
    "learning_rate": 5.631639213218534e-05,
    "epoch": 0.6450783033733157,
    "step": 20060
  },
  {
    "loss": 0.06710969805717468,
    "grad_norm": 0.6428706049919128,
    "learning_rate": 5.622524473533269e-05,
    "epoch": 0.6453998778017173,
    "step": 20070
  },
  {
    "loss": 0.06697219610214233,
    "grad_norm": 0.7152665257453918,
    "learning_rate": 5.613414230446467e-05,
    "epoch": 0.6457214522301187,
    "step": 20080
  },
  {
    "loss": 0.07348057627677917,
    "grad_norm": 0.7897191643714905,
    "learning_rate": 5.604308493316299e-05,
    "epoch": 0.6460430266585201,
    "step": 20090
  },
  {
    "loss": 0.07560304403305054,
    "grad_norm": 0.2724365293979645,
    "learning_rate": 5.595207271496283e-05,
    "epoch": 0.6463646010869216,
    "step": 20100
  },
  {
    "loss": 0.09230870008468628,
    "grad_norm": 0.31173989176750183,
    "learning_rate": 5.5861105743353173e-05,
    "epoch": 0.646686175515323,
    "step": 20110
  },
  {
    "loss": 0.06364266872406006,
    "grad_norm": 0.5667968988418579,
    "learning_rate": 5.577018411177641e-05,
    "epoch": 0.6470077499437245,
    "step": 20120
  },
  {
    "loss": 0.06696459054946899,
    "grad_norm": 0.5604732036590576,
    "learning_rate": 5.5679307913628365e-05,
    "epoch": 0.647329324372126,
    "step": 20130
  },
  {
    "loss": 0.09952617287635804,
    "grad_norm": 0.5189045667648315,
    "learning_rate": 5.558847724225828e-05,
    "epoch": 0.6476508988005274,
    "step": 20140
  },
  {
    "loss": 0.07203555703163148,
    "grad_norm": 0.4901675283908844,
    "learning_rate": 5.549769219096848e-05,
    "epoch": 0.6479724732289288,
    "step": 20150
  },
  {
    "loss": 0.07998304963111877,
    "grad_norm": 0.5910018086433411,
    "learning_rate": 5.540695285301465e-05,
    "epoch": 0.6482940476573303,
    "step": 20160
  },
  {
    "loss": 0.08121640682220459,
    "grad_norm": 0.40404200553894043,
    "learning_rate": 5.531625932160536e-05,
    "epoch": 0.6486156220857318,
    "step": 20170
  },
  {
    "loss": 0.06217573285102844,
    "grad_norm": 0.8459587693214417,
    "learning_rate": 5.522561168990206e-05,
    "epoch": 0.6489371965141332,
    "step": 20180
  },
  {
    "loss": 0.07579060792922973,
    "grad_norm": 0.6870545148849487,
    "learning_rate": 5.5135010051019275e-05,
    "epoch": 0.6492587709425347,
    "step": 20190
  },
  {
    "loss": 0.0787390947341919,
    "grad_norm": 0.8223719596862793,
    "learning_rate": 5.5044454498024065e-05,
    "epoch": 0.6495803453709361,
    "step": 20200
  },
  {
    "eval_loss": 0.07322059571743011,
    "eval_runtime": 34.7708,
    "eval_samples_per_second": 144.547,
    "eval_steps_per_second": 36.151,
    "epoch": 0.6496446602566164,
    "step": 20202
  },
  {
    "loss": 0.07401165962219239,
    "grad_norm": 0.4529065489768982,
    "learning_rate": 5.495394512393628e-05,
    "epoch": 0.6499019197993375,
    "step": 20210
  },
  {
    "loss": 0.07655703425407409,
    "grad_norm": 0.5374553203582764,
    "learning_rate": 5.486348202172833e-05,
    "epoch": 0.6502234942277391,
    "step": 20220
  },
  {
    "loss": 0.0764824092388153,
    "grad_norm": 0.637631356716156,
    "learning_rate": 5.4773065284324996e-05,
    "epoch": 0.6505450686561405,
    "step": 20230
  },
  {
    "loss": 0.08825883865356446,
    "grad_norm": 0.6297155022621155,
    "learning_rate": 5.468269500460359e-05,
    "epoch": 0.6508666430845419,
    "step": 20240
  },
  {
    "loss": 0.0745690643787384,
    "grad_norm": 0.7897636294364929,
    "learning_rate": 5.4592371275393494e-05,
    "epoch": 0.6511882175129434,
    "step": 20250
  },
  {
    "loss": 0.07205396890640259,
    "grad_norm": 0.4687490463256836,
    "learning_rate": 5.450209418947652e-05,
    "epoch": 0.6515097919413448,
    "step": 20260
  },
  {
    "loss": 0.07531830668449402,
    "grad_norm": 0.5674930214881897,
    "learning_rate": 5.441186383958632e-05,
    "epoch": 0.6518313663697463,
    "step": 20270
  },
  {
    "loss": 0.07985269427299499,
    "grad_norm": 0.686039388179779,
    "learning_rate": 5.4321680318408695e-05,
    "epoch": 0.6521529407981478,
    "step": 20280
  },
  {
    "loss": 0.07927957773208619,
    "grad_norm": 0.54157555103302,
    "learning_rate": 5.423154371858138e-05,
    "epoch": 0.6524745152265492,
    "step": 20290
  },
  {
    "loss": 0.07954124212265015,
    "grad_norm": 0.7790160775184631,
    "learning_rate": 5.414145413269371e-05,
    "epoch": 0.6527960896549506,
    "step": 20300
  },
  {
    "loss": 0.07763062715530396,
    "grad_norm": 0.6327701210975647,
    "learning_rate": 5.405141165328697e-05,
    "epoch": 0.6531176640833521,
    "step": 20310
  },
  {
    "loss": 0.07775360345840454,
    "grad_norm": 0.6861497163772583,
    "learning_rate": 5.396141637285383e-05,
    "epoch": 0.6534392385117536,
    "step": 20320
  },
  {
    "loss": 0.059467607736587526,
    "grad_norm": 0.7230491638183594,
    "learning_rate": 5.387146838383867e-05,
    "epoch": 0.653760812940155,
    "step": 20330
  },
  {
    "loss": 0.0787361741065979,
    "grad_norm": 0.4892818033695221,
    "learning_rate": 5.378156777863717e-05,
    "epoch": 0.6540823873685565,
    "step": 20340
  },
  {
    "loss": 0.06754761934280396,
    "grad_norm": 0.4849034547805786,
    "learning_rate": 5.369171464959635e-05,
    "epoch": 0.6544039617969579,
    "step": 20350
  },
  {
    "loss": 0.06928637623786926,
    "grad_norm": 0.8103519678115845,
    "learning_rate": 5.360190908901455e-05,
    "epoch": 0.6547255362253593,
    "step": 20360
  },
  {
    "loss": 0.0871894121170044,
    "grad_norm": 0.7014930248260498,
    "learning_rate": 5.35121511891411e-05,
    "epoch": 0.6550471106537609,
    "step": 20370
  },
  {
    "loss": 0.07648140788078309,
    "grad_norm": 0.5109164714813232,
    "learning_rate": 5.3422441042176566e-05,
    "epoch": 0.6553686850821623,
    "step": 20380
  },
  {
    "loss": 0.0827454686164856,
    "grad_norm": 0.6321818232536316,
    "learning_rate": 5.333277874027227e-05,
    "epoch": 0.6556902595105637,
    "step": 20390
  },
  {
    "loss": 0.07089014649391175,
    "grad_norm": 0.7621293067932129,
    "learning_rate": 5.324316437553041e-05,
    "epoch": 0.6560118339389652,
    "step": 20400
  },
  {
    "loss": 0.09134788513183593,
    "grad_norm": 0.3434469699859619,
    "learning_rate": 5.3153598040004146e-05,
    "epoch": 0.6563334083673666,
    "step": 20410
  },
  {
    "loss": 0.06535201668739318,
    "grad_norm": 0.6273918151855469,
    "learning_rate": 5.306407982569704e-05,
    "epoch": 0.6566549827957681,
    "step": 20420
  },
  {
    "loss": 0.08040034770965576,
    "grad_norm": 0.2526308596134186,
    "learning_rate": 5.297460982456344e-05,
    "epoch": 0.6569765572241696,
    "step": 20430
  },
  {
    "loss": 0.0793336808681488,
    "grad_norm": 0.807705819606781,
    "learning_rate": 5.288518812850799e-05,
    "epoch": 0.657298131652571,
    "step": 20440
  },
  {
    "loss": 0.0994856297969818,
    "grad_norm": 0.4748459756374359,
    "learning_rate": 5.279581482938578e-05,
    "epoch": 0.6576197060809724,
    "step": 20450
  },
  {
    "loss": 0.07574291825294495,
    "grad_norm": 0.42891186475753784,
    "learning_rate": 5.270649001900226e-05,
    "epoch": 0.657941280509374,
    "step": 20460
  },
  {
    "loss": 0.07210445404052734,
    "grad_norm": 0.7261443138122559,
    "learning_rate": 5.2617213789112906e-05,
    "epoch": 0.6582628549377754,
    "step": 20470
  },
  {
    "loss": 0.07575239539146424,
    "grad_norm": 0.4555186927318573,
    "learning_rate": 5.2527986231423554e-05,
    "epoch": 0.6585844293661768,
    "step": 20480
  },
  {
    "loss": 0.08972333669662476,
    "grad_norm": 0.8295562267303467,
    "learning_rate": 5.24388074375898e-05,
    "epoch": 0.6589060037945783,
    "step": 20490
  },
  {
    "loss": 0.07050554752349854,
    "grad_norm": 0.28110215067863464,
    "learning_rate": 5.2349677499217185e-05,
    "epoch": 0.6592275782229797,
    "step": 20500
  },
  {
    "loss": 0.0773880124092102,
    "grad_norm": 0.49791833758354187,
    "learning_rate": 5.226059650786122e-05,
    "epoch": 0.6595491526513811,
    "step": 20510
  },
  {
    "loss": 0.0658577024936676,
    "grad_norm": 0.4790436625480652,
    "learning_rate": 5.217156455502692e-05,
    "epoch": 0.6598707270797827,
    "step": 20520
  },
  {
    "loss": 0.08007944226264954,
    "grad_norm": 0.3390914797782898,
    "learning_rate": 5.208258173216909e-05,
    "epoch": 0.6601923015081841,
    "step": 20530
  },
  {
    "loss": 0.08911491632461548,
    "grad_norm": 0.6809537410736084,
    "learning_rate": 5.1993648130692055e-05,
    "epoch": 0.6605138759365855,
    "step": 20540
  },
  {
    "loss": 0.0683016836643219,
    "grad_norm": 0.3764163851737976,
    "learning_rate": 5.1904763841949466e-05,
    "epoch": 0.660835450364987,
    "step": 20550
  },
  {
    "loss": 0.07006845474243165,
    "grad_norm": 0.2706121802330017,
    "learning_rate": 5.181592895724447e-05,
    "epoch": 0.6611570247933884,
    "step": 20560
  },
  {
    "loss": 0.08022654056549072,
    "grad_norm": 0.3748409152030945,
    "learning_rate": 5.172714356782932e-05,
    "epoch": 0.6614785992217899,
    "step": 20570
  },
  {
    "loss": 0.06296700835227967,
    "grad_norm": 0.5847156643867493,
    "learning_rate": 5.163840776490555e-05,
    "epoch": 0.6618001736501913,
    "step": 20580
  },
  {
    "loss": 0.06332173347473144,
    "grad_norm": 0.44201889634132385,
    "learning_rate": 5.154972163962365e-05,
    "epoch": 0.6621217480785928,
    "step": 20590
  },
  {
    "loss": 0.08053407669067383,
    "grad_norm": 0.4436214864253998,
    "learning_rate": 5.1461085283083154e-05,
    "epoch": 0.6624433225069942,
    "step": 20600
  },
  {
    "loss": 0.0634561836719513,
    "grad_norm": 0.4250709116458893,
    "learning_rate": 5.137249878633251e-05,
    "epoch": 0.6627648969353956,
    "step": 20610
  },
  {
    "loss": 0.07566155791282654,
    "grad_norm": 0.5983794927597046,
    "learning_rate": 5.128396224036881e-05,
    "epoch": 0.6630864713637972,
    "step": 20620
  },
  {
    "loss": 0.05465579628944397,
    "grad_norm": 0.6458154320716858,
    "learning_rate": 5.119547573613799e-05,
    "epoch": 0.6634080457921986,
    "step": 20630
  },
  {
    "loss": 0.07174187898635864,
    "grad_norm": 0.5902681946754456,
    "learning_rate": 5.1107039364534436e-05,
    "epoch": 0.6637296202206,
    "step": 20640
  },
  {
    "loss": 0.07929787039756775,
    "grad_norm": 0.4368651807308197,
    "learning_rate": 5.101865321640119e-05,
    "epoch": 0.6640511946490015,
    "step": 20650
  },
  {
    "loss": 0.07327082753181458,
    "grad_norm": 0.9439780116081238,
    "learning_rate": 5.0930317382529555e-05,
    "epoch": 0.664372769077403,
    "step": 20660
  },
  {
    "loss": 0.09078122973442078,
    "grad_norm": 0.7786108255386353,
    "learning_rate": 5.084203195365924e-05,
    "epoch": 0.6646943435058044,
    "step": 20670
  },
  {
    "loss": 0.07400400042533875,
    "grad_norm": 0.46253257989883423,
    "learning_rate": 5.0753797020478235e-05,
    "epoch": 0.6650159179342059,
    "step": 20680
  },
  {
    "loss": 0.07291633486747742,
    "grad_norm": 0.9724017381668091,
    "learning_rate": 5.066561267362251e-05,
    "epoch": 0.6653374923626073,
    "step": 20690
  },
  {
    "loss": 0.10171788930892944,
    "grad_norm": 0.5624548196792603,
    "learning_rate": 5.0577479003676134e-05,
    "epoch": 0.6656590667910087,
    "step": 20700
  },
  {
    "loss": 0.07935294508934021,
    "grad_norm": 0.7655509114265442,
    "learning_rate": 5.048939610117118e-05,
    "epoch": 0.6659806412194103,
    "step": 20710
  },
  {
    "loss": 0.06642114520072936,
    "grad_norm": 0.802371621131897,
    "learning_rate": 5.040136405658743e-05,
    "epoch": 0.6663022156478117,
    "step": 20720
  },
  {
    "loss": 0.0731515884399414,
    "grad_norm": 0.5871564149856567,
    "learning_rate": 5.031338296035266e-05,
    "epoch": 0.6666237900762131,
    "step": 20730
  },
  {
    "loss": 0.09851167798042297,
    "grad_norm": 0.6241931319236755,
    "learning_rate": 5.0225452902842116e-05,
    "epoch": 0.6669453645046146,
    "step": 20740
  },
  {
    "loss": 0.09080070853233338,
    "grad_norm": 0.5461302995681763,
    "learning_rate": 5.013757397437862e-05,
    "epoch": 0.667266938933016,
    "step": 20750
  },
  {
    "loss": 0.07654562592506409,
    "grad_norm": 0.7314095497131348,
    "learning_rate": 5.004974626523258e-05,
    "epoch": 0.6675885133614174,
    "step": 20760
  },
  {
    "loss": 0.08786712288856506,
    "grad_norm": 0.6281319260597229,
    "learning_rate": 4.9961969865621684e-05,
    "epoch": 0.667910087789819,
    "step": 20770
  },
  {
    "loss": 0.07048113346099853,
    "grad_norm": 0.4600871801376343,
    "learning_rate": 4.9874244865711e-05,
    "epoch": 0.6682316622182204,
    "step": 20780
  },
  {
    "loss": 0.08273769021034241,
    "grad_norm": 0.746705174446106,
    "learning_rate": 4.9786571355612786e-05,
    "epoch": 0.6685532366466218,
    "step": 20790
  },
  {
    "loss": 0.09555152654647828,
    "grad_norm": 0.4937872886657715,
    "learning_rate": 4.969894942538632e-05,
    "epoch": 0.6688748110750233,
    "step": 20800
  },
  {
    "loss": 0.063187974691391,
    "grad_norm": 0.6522345542907715,
    "learning_rate": 4.961137916503803e-05,
    "epoch": 0.6691963855034248,
    "step": 20810
  },
  {
    "loss": 0.08229098320007325,
    "grad_norm": 0.5910975933074951,
    "learning_rate": 4.952386066452114e-05,
    "epoch": 0.6695179599318262,
    "step": 20820
  },
  {
    "loss": 0.060083693265914916,
    "grad_norm": 0.6340411901473999,
    "learning_rate": 4.943639401373582e-05,
    "epoch": 0.6698395343602277,
    "step": 20830
  },
  {
    "loss": 0.07824947834014892,
    "grad_norm": 0.5943097472190857,
    "learning_rate": 4.934897930252886e-05,
    "epoch": 0.6701611087886291,
    "step": 20840
  },
  {
    "loss": 0.08365764021873474,
    "grad_norm": 0.9509609341621399,
    "learning_rate": 4.9261616620693804e-05,
    "epoch": 0.6704826832170305,
    "step": 20850
  },
  {
    "loss": 0.08415056467056274,
    "grad_norm": 0.4306361675262451,
    "learning_rate": 4.917430605797074e-05,
    "epoch": 0.6708042576454321,
    "step": 20860
  },
  {
    "loss": 0.08667734265327454,
    "grad_norm": 0.6566128134727478,
    "learning_rate": 4.908704770404611e-05,
    "epoch": 0.6711258320738335,
    "step": 20870
  },
  {
    "loss": 0.0774792730808258,
    "grad_norm": 0.6821538209915161,
    "learning_rate": 4.899984164855289e-05,
    "epoch": 0.6714474065022349,
    "step": 20880
  },
  {
    "loss": 0.06847630739212036,
    "grad_norm": 0.4911244511604309,
    "learning_rate": 4.891268798107016e-05,
    "epoch": 0.6717689809306364,
    "step": 20890
  },
  {
    "loss": 0.08165109753608704,
    "grad_norm": 0.5228062868118286,
    "learning_rate": 4.8825586791123356e-05,
    "epoch": 0.6720905553590378,
    "step": 20900
  },
  {
    "loss": 0.058252638578414916,
    "grad_norm": 0.3973667621612549,
    "learning_rate": 4.8738538168183845e-05,
    "epoch": 0.6724121297874392,
    "step": 20910
  },
  {
    "loss": 0.08311399221420288,
    "grad_norm": 0.7493540644645691,
    "learning_rate": 4.8651542201669134e-05,
    "epoch": 0.6727337042158408,
    "step": 20920
  },
  {
    "loss": 0.07413286566734315,
    "grad_norm": 0.4782170057296753,
    "learning_rate": 4.85645989809426e-05,
    "epoch": 0.6730552786442422,
    "step": 20930
  },
  {
    "loss": 0.0762550950050354,
    "grad_norm": 0.7324798703193665,
    "learning_rate": 4.8477708595313356e-05,
    "epoch": 0.6733768530726436,
    "step": 20940
  },
  {
    "loss": 0.05731356143951416,
    "grad_norm": 0.5743696093559265,
    "learning_rate": 4.8390871134036396e-05,
    "epoch": 0.6736984275010451,
    "step": 20950
  },
  {
    "loss": 0.08263179659843445,
    "grad_norm": 0.6200225949287415,
    "learning_rate": 4.830408668631221e-05,
    "epoch": 0.6740200019294466,
    "step": 20960
  },
  {
    "loss": 0.08547630310058593,
    "grad_norm": 0.8113134503364563,
    "learning_rate": 4.821735534128682e-05,
    "epoch": 0.674341576357848,
    "step": 20970
  },
  {
    "loss": 0.07483196258544922,
    "grad_norm": 0.8760203719139099,
    "learning_rate": 4.8130677188051906e-05,
    "epoch": 0.6746631507862495,
    "step": 20980
  },
  {
    "loss": 0.0705351173877716,
    "grad_norm": 0.49309197068214417,
    "learning_rate": 4.804405231564427e-05,
    "epoch": 0.6749847252146509,
    "step": 20990
  },
  {
    "loss": 0.05743688344955444,
    "grad_norm": 0.2787437438964844,
    "learning_rate": 4.795748081304615e-05,
    "epoch": 0.6753062996430523,
    "step": 21000
  },
  {
    "loss": 0.07269101142883301,
    "grad_norm": 0.7541826963424683,
    "learning_rate": 4.787096276918486e-05,
    "epoch": 0.6756278740714539,
    "step": 21010
  },
  {
    "loss": 0.09004797935485839,
    "grad_norm": 1.4871023893356323,
    "learning_rate": 4.7784498272932777e-05,
    "epoch": 0.6759494484998553,
    "step": 21020
  },
  {
    "loss": 0.07304201126098633,
    "grad_norm": 0.3444749712944031,
    "learning_rate": 4.7698087413107416e-05,
    "epoch": 0.6762710229282567,
    "step": 21030
  },
  {
    "loss": 0.07011285424232483,
    "grad_norm": 0.7484362721443176,
    "learning_rate": 4.761173027847101e-05,
    "epoch": 0.6765925973566582,
    "step": 21040
  },
  {
    "loss": 0.08315953612327576,
    "grad_norm": 0.6611018180847168,
    "learning_rate": 4.7525426957730845e-05,
    "epoch": 0.6769141717850596,
    "step": 21050
  },
  {
    "loss": 0.0854596197605133,
    "grad_norm": 0.5237786173820496,
    "learning_rate": 4.743917753953871e-05,
    "epoch": 0.677235746213461,
    "step": 21060
  },
  {
    "loss": 0.07612773776054382,
    "grad_norm": 0.8620582818984985,
    "learning_rate": 4.7352982112491074e-05,
    "epoch": 0.6775573206418626,
    "step": 21070
  },
  {
    "loss": 0.06872981786727905,
    "grad_norm": 0.9304988384246826,
    "learning_rate": 4.726684076512904e-05,
    "epoch": 0.677878895070264,
    "step": 21080
  },
  {
    "loss": 0.09908568859100342,
    "grad_norm": 0.7491776943206787,
    "learning_rate": 4.718075358593802e-05,
    "epoch": 0.6782004694986654,
    "step": 21090
  },
  {
    "loss": 0.07195566296577453,
    "grad_norm": 0.506563127040863,
    "learning_rate": 4.70947206633479e-05,
    "epoch": 0.6785220439270669,
    "step": 21100
  },
  {
    "loss": 0.0828722059726715,
    "grad_norm": 0.5606396794319153,
    "learning_rate": 4.700874208573284e-05,
    "epoch": 0.6788436183554684,
    "step": 21110
  },
  {
    "loss": 0.08981196284294128,
    "grad_norm": 0.8274250030517578,
    "learning_rate": 4.692281794141102e-05,
    "epoch": 0.6791651927838698,
    "step": 21120
  },
  {
    "loss": 0.07490351200103759,
    "grad_norm": 0.6135382652282715,
    "learning_rate": 4.683694831864492e-05,
    "epoch": 0.6794867672122713,
    "step": 21130
  },
  {
    "loss": 0.08893336653709412,
    "grad_norm": 0.43025079369544983,
    "learning_rate": 4.675113330564082e-05,
    "epoch": 0.6798083416406727,
    "step": 21140
  },
  {
    "loss": 0.06783245205879211,
    "grad_norm": 0.8443710207939148,
    "learning_rate": 4.666537299054906e-05,
    "epoch": 0.6801299160690741,
    "step": 21150
  },
  {
    "loss": 0.08298020958900451,
    "grad_norm": 0.7228168249130249,
    "learning_rate": 4.657966746146366e-05,
    "epoch": 0.6804514904974757,
    "step": 21160
  },
  {
    "loss": 0.07848789095878601,
    "grad_norm": 0.6177375912666321,
    "learning_rate": 4.649401680642246e-05,
    "epoch": 0.6807730649258771,
    "step": 21170
  },
  {
    "loss": 0.07725452184677124,
    "grad_norm": 0.8629878163337708,
    "learning_rate": 4.640842111340694e-05,
    "epoch": 0.6810946393542785,
    "step": 21180
  },
  {
    "loss": 0.08047922849655151,
    "grad_norm": 0.865945041179657,
    "learning_rate": 4.632288047034201e-05,
    "epoch": 0.68141621378268,
    "step": 21190
  },
  {
    "loss": 0.056549960374832155,
    "grad_norm": 0.5094005465507507,
    "learning_rate": 4.623739496509618e-05,
    "epoch": 0.6817377882110814,
    "step": 21200
  },
  {
    "loss": 0.06563356518745422,
    "grad_norm": 0.5315867066383362,
    "learning_rate": 4.6151964685481175e-05,
    "epoch": 0.6820593626394829,
    "step": 21210
  },
  {
    "loss": 0.08418481349945069,
    "grad_norm": 0.7525455951690674,
    "learning_rate": 4.606658971925213e-05,
    "epoch": 0.6823809370678844,
    "step": 21220
  },
  {
    "loss": 0.07750657796859742,
    "grad_norm": 0.8312000632286072,
    "learning_rate": 4.598127015410723e-05,
    "epoch": 0.6827025114962858,
    "step": 21230
  },
  {
    "loss": 0.0824580192565918,
    "grad_norm": 0.6511006951332092,
    "learning_rate": 4.5896006077687835e-05,
    "epoch": 0.6830240859246872,
    "step": 21240
  },
  {
    "loss": 0.10098731517791748,
    "grad_norm": 0.3902733027935028,
    "learning_rate": 4.581079757757835e-05,
    "epoch": 0.6833456603530887,
    "step": 21250
  },
  {
    "loss": 0.07251612544059753,
    "grad_norm": 0.6815239787101746,
    "learning_rate": 4.572564474130593e-05,
    "epoch": 0.6836672347814902,
    "step": 21260
  },
  {
    "loss": 0.07855257987976075,
    "grad_norm": 0.7673274278640747,
    "learning_rate": 4.564054765634074e-05,
    "epoch": 0.6839888092098916,
    "step": 21270
  },
  {
    "loss": 0.07359289526939392,
    "grad_norm": 0.3823041617870331,
    "learning_rate": 4.555550641009555e-05,
    "epoch": 0.6843103836382931,
    "step": 21280
  },
  {
    "loss": 0.07962161302566528,
    "grad_norm": 0.39447155594825745,
    "learning_rate": 4.5470521089925736e-05,
    "epoch": 0.6846319580666945,
    "step": 21290
  },
  {
    "loss": 0.07411338686943054,
    "grad_norm": 0.42518749833106995,
    "learning_rate": 4.5385591783129424e-05,
    "epoch": 0.6849535324950959,
    "step": 21300
  },
  {
    "loss": 0.07336281538009644,
    "grad_norm": 0.5015262365341187,
    "learning_rate": 4.5300718576946986e-05,
    "epoch": 0.6852751069234975,
    "step": 21310
  },
  {
    "loss": 0.07329775094985962,
    "grad_norm": 0.45550772547721863,
    "learning_rate": 4.521590155856133e-05,
    "epoch": 0.6855966813518989,
    "step": 21320
  },
  {
    "loss": 0.07568760514259339,
    "grad_norm": 0.5514888763427734,
    "learning_rate": 4.513114081509755e-05,
    "epoch": 0.6859182557803003,
    "step": 21330
  },
  {
    "loss": 0.0703023374080658,
    "grad_norm": 0.5648098587989807,
    "learning_rate": 4.50464364336229e-05,
    "epoch": 0.6862398302087018,
    "step": 21340
  },
  {
    "loss": 0.07627668976783752,
    "grad_norm": 0.5147767066955566,
    "learning_rate": 4.4961788501146875e-05,
    "epoch": 0.6865614046371032,
    "step": 21350
  },
  {
    "loss": 0.07581337094306946,
    "grad_norm": 0.5529208183288574,
    "learning_rate": 4.487719710462083e-05,
    "epoch": 0.6868829790655047,
    "step": 21360
  },
  {
    "loss": 0.08803019523620606,
    "grad_norm": 0.47467994689941406,
    "learning_rate": 4.4792662330938176e-05,
    "epoch": 0.6872045534939062,
    "step": 21370
  },
  {
    "loss": 0.06819509267807007,
    "grad_norm": 0.5418660640716553,
    "learning_rate": 4.470818426693413e-05,
    "epoch": 0.6875261279223076,
    "step": 21380
  },
  {
    "loss": 0.08300604224205017,
    "grad_norm": 0.41477343440055847,
    "learning_rate": 4.462376299938555e-05,
    "epoch": 0.687847702350709,
    "step": 21390
  },
  {
    "loss": 0.10577502250671386,
    "grad_norm": 0.9017356634140015,
    "learning_rate": 4.453939861501113e-05,
    "epoch": 0.6881692767791106,
    "step": 21400
  },
  {
    "loss": 0.08897504210472107,
    "grad_norm": 0.7038023471832275,
    "learning_rate": 4.445509120047094e-05,
    "epoch": 0.688490851207512,
    "step": 21410
  },
  {
    "loss": 0.06395829319953919,
    "grad_norm": 0.7179198861122131,
    "learning_rate": 4.437084084236669e-05,
    "epoch": 0.6888124256359134,
    "step": 21420
  },
  {
    "loss": 0.08118244409561157,
    "grad_norm": 0.6487843990325928,
    "learning_rate": 4.428664762724144e-05,
    "epoch": 0.6891340000643149,
    "step": 21430
  },
  {
    "loss": 0.07925102710723878,
    "grad_norm": 0.4036339223384857,
    "learning_rate": 4.420251164157945e-05,
    "epoch": 0.6894555744927163,
    "step": 21440
  },
  {
    "loss": 0.08218384385108948,
    "grad_norm": 0.7086281776428223,
    "learning_rate": 4.411843297180636e-05,
    "epoch": 0.6897771489211177,
    "step": 21450
  },
  {
    "loss": 0.07362968921661377,
    "grad_norm": 0.7492254972457886,
    "learning_rate": 4.403441170428877e-05,
    "epoch": 0.6900987233495193,
    "step": 21460
  },
  {
    "loss": 0.07590994834899903,
    "grad_norm": 0.4599405527114868,
    "learning_rate": 4.395044792533446e-05,
    "epoch": 0.6904202977779207,
    "step": 21470
  },
  {
    "loss": 0.07032697200775147,
    "grad_norm": 0.5740514993667603,
    "learning_rate": 4.386654172119204e-05,
    "epoch": 0.6907418722063221,
    "step": 21480
  },
  {
    "loss": 0.08793300986289979,
    "grad_norm": 0.6255429983139038,
    "learning_rate": 4.3782693178051026e-05,
    "epoch": 0.6910634466347236,
    "step": 21490
  },
  {
    "loss": 0.07888756394386291,
    "grad_norm": 0.4935165345668793,
    "learning_rate": 4.369890238204177e-05,
    "epoch": 0.691385021063125,
    "step": 21500
  },
  {
    "loss": 0.07613331079483032,
    "grad_norm": 0.8250544667243958,
    "learning_rate": 4.361516941923518e-05,
    "epoch": 0.6917065954915265,
    "step": 21510
  },
  {
    "loss": 0.06617268323898315,
    "grad_norm": 0.5559676289558411,
    "learning_rate": 4.3531494375642854e-05,
    "epoch": 0.692028169919928,
    "step": 21520
  },
  {
    "loss": 0.08175877332687378,
    "grad_norm": 0.5441673994064331,
    "learning_rate": 4.3447877337216825e-05,
    "epoch": 0.6923497443483294,
    "step": 21530
  },
  {
    "loss": 0.11042051315307617,
    "grad_norm": 0.8127007484436035,
    "learning_rate": 4.3364318389849635e-05,
    "epoch": 0.6926713187767308,
    "step": 21540
  },
  {
    "loss": 0.06440520882606507,
    "grad_norm": 0.5828354954719543,
    "learning_rate": 4.328081761937401e-05,
    "epoch": 0.6929928932051324,
    "step": 21550
  },
  {
    "loss": 0.06377939581871032,
    "grad_norm": 0.41448041796684265,
    "learning_rate": 4.319737511156307e-05,
    "epoch": 0.6933144676335338,
    "step": 21560
  },
  {
    "loss": 0.08158342242240905,
    "grad_norm": 0.8715920448303223,
    "learning_rate": 4.3113990952130024e-05,
    "epoch": 0.6936360420619352,
    "step": 21570
  },
  {
    "loss": 0.07553659677505493,
    "grad_norm": 0.7690626382827759,
    "learning_rate": 4.303066522672812e-05,
    "epoch": 0.6939576164903367,
    "step": 21580
  },
  {
    "loss": 0.06266080141067505,
    "grad_norm": 1.2075152397155762,
    "learning_rate": 4.2947398020950556e-05,
    "epoch": 0.6942791909187381,
    "step": 21590
  },
  {
    "loss": 0.07179356813430786,
    "grad_norm": 0.7406636476516724,
    "learning_rate": 4.2864189420330544e-05,
    "epoch": 0.6946007653471395,
    "step": 21600
  },
  {
    "loss": 0.06792376041412354,
    "grad_norm": 0.717032790184021,
    "learning_rate": 4.278103951034092e-05,
    "epoch": 0.6949223397755411,
    "step": 21610
  },
  {
    "loss": 0.06622045636177062,
    "grad_norm": 0.5181135535240173,
    "learning_rate": 4.269794837639444e-05,
    "epoch": 0.6952439142039425,
    "step": 21620
  },
  {
    "loss": 0.08370713591575622,
    "grad_norm": 0.9519104361534119,
    "learning_rate": 4.261491610384331e-05,
    "epoch": 0.6955654886323439,
    "step": 21630
  },
  {
    "loss": 0.07882735133171082,
    "grad_norm": 0.6479547619819641,
    "learning_rate": 4.2531942777979325e-05,
    "epoch": 0.6958870630607454,
    "step": 21640
  },
  {
    "loss": 0.059241306781768796,
    "grad_norm": 0.4295456111431122,
    "learning_rate": 4.244902848403378e-05,
    "epoch": 0.6962086374891469,
    "step": 21650
  },
  {
    "loss": 0.06421804428100586,
    "grad_norm": 0.28864648938179016,
    "learning_rate": 4.236617330717725e-05,
    "epoch": 0.6965302119175483,
    "step": 21660
  },
  {
    "loss": 0.0704532504081726,
    "grad_norm": 0.639294445514679,
    "learning_rate": 4.228337733251962e-05,
    "epoch": 0.6968517863459498,
    "step": 21670
  },
  {
    "loss": 0.05593020915985107,
    "grad_norm": 0.6021806001663208,
    "learning_rate": 4.220064064511004e-05,
    "epoch": 0.6971733607743512,
    "step": 21680
  },
  {
    "loss": 0.07564791440963745,
    "grad_norm": 0.6179336309432983,
    "learning_rate": 4.2117963329936597e-05,
    "epoch": 0.6974949352027526,
    "step": 21690
  },
  {
    "loss": 0.06977922320365906,
    "grad_norm": 0.45775020122528076,
    "learning_rate": 4.2035345471926545e-05,
    "epoch": 0.6978165096311542,
    "step": 21700
  },
  {
    "loss": 0.07761858105659485,
    "grad_norm": 0.34109559655189514,
    "learning_rate": 4.195278715594595e-05,
    "epoch": 0.6981380840595556,
    "step": 21710
  },
  {
    "loss": 0.05580984354019165,
    "grad_norm": 0.3880707323551178,
    "learning_rate": 4.1870288466799815e-05,
    "epoch": 0.698459658487957,
    "step": 21720
  },
  {
    "loss": 0.0565685510635376,
    "grad_norm": 0.27682071924209595,
    "learning_rate": 4.178784948923177e-05,
    "epoch": 0.6987812329163585,
    "step": 21730
  },
  {
    "loss": 0.07089447379112243,
    "grad_norm": 0.8211534023284912,
    "learning_rate": 4.170547030792422e-05,
    "epoch": 0.6991028073447599,
    "step": 21740
  },
  {
    "loss": 0.07182880640029907,
    "grad_norm": 0.33115750551223755,
    "learning_rate": 4.162315100749815e-05,
    "epoch": 0.6994243817731614,
    "step": 21750
  },
  {
    "eval_loss": 0.07118770480155945,
    "eval_runtime": 35.4323,
    "eval_samples_per_second": 141.848,
    "eval_steps_per_second": 35.476,
    "epoch": 0.6996173264302022,
    "step": 21756
  },
  {
    "loss": 0.07729780077934265,
    "grad_norm": 0.9225740432739258,
    "learning_rate": 4.1540891672512904e-05,
    "epoch": 0.6997459562015629,
    "step": 21760
  },
  {
    "loss": 0.07148987650871277,
    "grad_norm": 0.6877495646476746,
    "learning_rate": 4.14586923874664e-05,
    "epoch": 0.7000675306299643,
    "step": 21770
  },
  {
    "loss": 0.07752482295036316,
    "grad_norm": 0.5832339525222778,
    "learning_rate": 4.137655323679471e-05,
    "epoch": 0.7003891050583657,
    "step": 21780
  },
  {
    "loss": 0.06601099967956543,
    "grad_norm": 0.5884093046188354,
    "learning_rate": 4.129447430487229e-05,
    "epoch": 0.7007106794867672,
    "step": 21790
  },
  {
    "loss": 0.08792551755905151,
    "grad_norm": 0.8232205510139465,
    "learning_rate": 4.1212455676011583e-05,
    "epoch": 0.7010322539151687,
    "step": 21800
  },
  {
    "loss": 0.056697070598602295,
    "grad_norm": 0.6312450766563416,
    "learning_rate": 4.11304974344632e-05,
    "epoch": 0.7013538283435701,
    "step": 21810
  },
  {
    "loss": 0.06890648007392883,
    "grad_norm": 0.6079735159873962,
    "learning_rate": 4.104859966441574e-05,
    "epoch": 0.7016754027719716,
    "step": 21820
  },
  {
    "loss": 0.08191646933555603,
    "grad_norm": 0.9938222169876099,
    "learning_rate": 4.096676244999554e-05,
    "epoch": 0.701996977200373,
    "step": 21830
  },
  {
    "loss": 0.08274092078208924,
    "grad_norm": 0.509135365486145,
    "learning_rate": 4.0884985875266925e-05,
    "epoch": 0.7023185516287744,
    "step": 21840
  },
  {
    "loss": 0.07548218369483947,
    "grad_norm": 0.6436309218406677,
    "learning_rate": 4.0803270024231763e-05,
    "epoch": 0.702640126057176,
    "step": 21850
  },
  {
    "loss": 0.08233748078346252,
    "grad_norm": 0.7960435748100281,
    "learning_rate": 4.0721614980829606e-05,
    "epoch": 0.7029617004855774,
    "step": 21860
  },
  {
    "loss": 0.08757339119911194,
    "grad_norm": 0.5708218216896057,
    "learning_rate": 4.064002082893758e-05,
    "epoch": 0.7032832749139788,
    "step": 21870
  },
  {
    "loss": 0.08269885182380676,
    "grad_norm": 0.7067870497703552,
    "learning_rate": 4.055848765237021e-05,
    "epoch": 0.7036048493423803,
    "step": 21880
  },
  {
    "loss": 0.08364191055297851,
    "grad_norm": 0.6864334940910339,
    "learning_rate": 4.047701553487948e-05,
    "epoch": 0.7039264237707817,
    "step": 21890
  },
  {
    "loss": 0.06692585349082947,
    "grad_norm": 0.5575340986251831,
    "learning_rate": 4.039560456015453e-05,
    "epoch": 0.7042479981991832,
    "step": 21900
  },
  {
    "loss": 0.07891395688056946,
    "grad_norm": 0.8325002193450928,
    "learning_rate": 4.031425481182173e-05,
    "epoch": 0.7045695726275847,
    "step": 21910
  },
  {
    "loss": 0.07879780530929566,
    "grad_norm": 0.7286157011985779,
    "learning_rate": 4.023296637344462e-05,
    "epoch": 0.7048911470559861,
    "step": 21920
  },
  {
    "loss": 0.07090092897415161,
    "grad_norm": 0.6933513879776001,
    "learning_rate": 4.015173932852363e-05,
    "epoch": 0.7052127214843875,
    "step": 21930
  },
  {
    "loss": 0.08005400896072387,
    "grad_norm": 0.7384825944900513,
    "learning_rate": 4.007057376049634e-05,
    "epoch": 0.705534295912789,
    "step": 21940
  },
  {
    "loss": 0.07087463140487671,
    "grad_norm": 0.531400740146637,
    "learning_rate": 3.998946975273699e-05,
    "epoch": 0.7058558703411905,
    "step": 21950
  },
  {
    "loss": 0.06927424073219299,
    "grad_norm": 0.5209558010101318,
    "learning_rate": 3.9908427388556625e-05,
    "epoch": 0.7061774447695919,
    "step": 21960
  },
  {
    "loss": 0.0795640230178833,
    "grad_norm": 0.2959861755371094,
    "learning_rate": 3.982744675120304e-05,
    "epoch": 0.7064990191979934,
    "step": 21970
  },
  {
    "loss": 0.09337790012359619,
    "grad_norm": 0.8974547982215881,
    "learning_rate": 3.97465279238605e-05,
    "epoch": 0.7068205936263948,
    "step": 21980
  },
  {
    "loss": 0.08562746047973632,
    "grad_norm": 0.8576725721359253,
    "learning_rate": 3.9665670989649904e-05,
    "epoch": 0.7071421680547962,
    "step": 21990
  },
  {
    "loss": 0.097445809841156,
    "grad_norm": 0.673926830291748,
    "learning_rate": 3.958487603162856e-05,
    "epoch": 0.7074637424831978,
    "step": 22000
  },
  {
    "loss": 0.0739307940006256,
    "grad_norm": 0.6029888987541199,
    "learning_rate": 3.9504143132789997e-05,
    "epoch": 0.7077853169115992,
    "step": 22010
  },
  {
    "loss": 0.06431577801704406,
    "grad_norm": 0.26620832085609436,
    "learning_rate": 3.942347237606413e-05,
    "epoch": 0.7081068913400006,
    "step": 22020
  },
  {
    "loss": 0.0735122561454773,
    "grad_norm": 0.5332812070846558,
    "learning_rate": 3.9342863844316944e-05,
    "epoch": 0.7084284657684021,
    "step": 22030
  },
  {
    "loss": 0.06432998776435853,
    "grad_norm": 0.34079718589782715,
    "learning_rate": 3.92623176203506e-05,
    "epoch": 0.7087500401968035,
    "step": 22040
  },
  {
    "loss": 0.05873666405677795,
    "grad_norm": 0.5827487111091614,
    "learning_rate": 3.918183378690313e-05,
    "epoch": 0.709071614625205,
    "step": 22050
  },
  {
    "loss": 0.0894324004650116,
    "grad_norm": 0.8233846426010132,
    "learning_rate": 3.9101412426648596e-05,
    "epoch": 0.7093931890536065,
    "step": 22060
  },
  {
    "loss": 0.06385500431060791,
    "grad_norm": 0.5678383708000183,
    "learning_rate": 3.902105362219687e-05,
    "epoch": 0.7097147634820079,
    "step": 22070
  },
  {
    "loss": 0.06556373834609985,
    "grad_norm": 0.5707252025604248,
    "learning_rate": 3.8940757456093456e-05,
    "epoch": 0.7100363379104093,
    "step": 22080
  },
  {
    "loss": 0.06791006922721862,
    "grad_norm": 0.5018502473831177,
    "learning_rate": 3.8860524010819684e-05,
    "epoch": 0.7103579123388108,
    "step": 22090
  },
  {
    "loss": 0.09631652235984803,
    "grad_norm": 0.49287688732147217,
    "learning_rate": 3.878035336879229e-05,
    "epoch": 0.7106794867672123,
    "step": 22100
  },
  {
    "loss": 0.06533873677253724,
    "grad_norm": 0.5646507143974304,
    "learning_rate": 3.870024561236365e-05,
    "epoch": 0.7110010611956137,
    "step": 22110
  },
  {
    "loss": 0.06178664565086365,
    "grad_norm": 0.1747838854789734,
    "learning_rate": 3.86202008238214e-05,
    "epoch": 0.7113226356240152,
    "step": 22120
  },
  {
    "loss": 0.12747459411621093,
    "grad_norm": 0.631783127784729,
    "learning_rate": 3.854021908538857e-05,
    "epoch": 0.7116442100524166,
    "step": 22130
  },
  {
    "loss": 0.08871740698814393,
    "grad_norm": 0.5668731331825256,
    "learning_rate": 3.84603004792235e-05,
    "epoch": 0.711965784480818,
    "step": 22140
  },
  {
    "loss": 0.08891339898109436,
    "grad_norm": 0.673675537109375,
    "learning_rate": 3.8380445087419505e-05,
    "epoch": 0.7122873589092196,
    "step": 22150
  },
  {
    "loss": 0.07999474406242371,
    "grad_norm": 0.5109632015228271,
    "learning_rate": 3.8300652992005116e-05,
    "epoch": 0.712608933337621,
    "step": 22160
  },
  {
    "loss": 0.07962061762809754,
    "grad_norm": 0.7328252196311951,
    "learning_rate": 3.822092427494377e-05,
    "epoch": 0.7129305077660224,
    "step": 22170
  },
  {
    "loss": 0.07219462394714356,
    "grad_norm": 0.625394880771637,
    "learning_rate": 3.814125901813374e-05,
    "epoch": 0.7132520821944239,
    "step": 22180
  },
  {
    "loss": 0.07665095329284669,
    "grad_norm": 0.7968199253082275,
    "learning_rate": 3.8061657303408324e-05,
    "epoch": 0.7135736566228253,
    "step": 22190
  },
  {
    "loss": 0.06330210566520691,
    "grad_norm": 0.5527894496917725,
    "learning_rate": 3.798211921253533e-05,
    "epoch": 0.7138952310512268,
    "step": 22200
  },
  {
    "loss": 0.08288524746894836,
    "grad_norm": 0.228219136595726,
    "learning_rate": 3.790264482721735e-05,
    "epoch": 0.7142168054796283,
    "step": 22210
  },
  {
    "loss": 0.06985173225402833,
    "grad_norm": 0.6191315054893494,
    "learning_rate": 3.7823234229091445e-05,
    "epoch": 0.7145383799080297,
    "step": 22220
  },
  {
    "loss": 0.077042555809021,
    "grad_norm": 0.5698592066764832,
    "learning_rate": 3.774388749972916e-05,
    "epoch": 0.7148599543364311,
    "step": 22230
  },
  {
    "loss": 0.07586207985877991,
    "grad_norm": 0.8893696069717407,
    "learning_rate": 3.766460472063653e-05,
    "epoch": 0.7151815287648327,
    "step": 22240
  },
  {
    "loss": 0.0670695424079895,
    "grad_norm": 1.1401852369308472,
    "learning_rate": 3.758538597325377e-05,
    "epoch": 0.7155031031932341,
    "step": 22250
  },
  {
    "loss": 0.07725749015808106,
    "grad_norm": 0.7471514940261841,
    "learning_rate": 3.7506231338955414e-05,
    "epoch": 0.7158246776216355,
    "step": 22260
  },
  {
    "loss": 0.07863717675209045,
    "grad_norm": 0.543266773223877,
    "learning_rate": 3.742714089905014e-05,
    "epoch": 0.716146252050037,
    "step": 22270
  },
  {
    "loss": 0.06630213856697083,
    "grad_norm": 0.24376331269741058,
    "learning_rate": 3.734811473478059e-05,
    "epoch": 0.7164678264784384,
    "step": 22280
  },
  {
    "loss": 0.05101625919342041,
    "grad_norm": 0.8426804542541504,
    "learning_rate": 3.7269152927323506e-05,
    "epoch": 0.7167894009068398,
    "step": 22290
  },
  {
    "loss": 0.06777915954589844,
    "grad_norm": 0.41970622539520264,
    "learning_rate": 3.71902555577894e-05,
    "epoch": 0.7171109753352414,
    "step": 22300
  },
  {
    "loss": 0.06783168911933898,
    "grad_norm": 0.6039134860038757,
    "learning_rate": 3.711142270722268e-05,
    "epoch": 0.7174325497636428,
    "step": 22310
  },
  {
    "loss": 0.05689939856529236,
    "grad_norm": 0.609046220779419,
    "learning_rate": 3.703265445660148e-05,
    "epoch": 0.7177541241920442,
    "step": 22320
  },
  {
    "loss": 0.07239497303962708,
    "grad_norm": 0.654071569442749,
    "learning_rate": 3.695395088683749e-05,
    "epoch": 0.7180756986204457,
    "step": 22330
  },
  {
    "loss": 0.07121925354003907,
    "grad_norm": 0.5291280150413513,
    "learning_rate": 3.687531207877607e-05,
    "epoch": 0.7183972730488472,
    "step": 22340
  },
  {
    "loss": 0.07519608736038208,
    "grad_norm": 0.7853771448135376,
    "learning_rate": 3.679673811319596e-05,
    "epoch": 0.7187188474772486,
    "step": 22350
  },
  {
    "loss": 0.08853870034217834,
    "grad_norm": 0.3993820250034332,
    "learning_rate": 3.671822907080938e-05,
    "epoch": 0.7190404219056501,
    "step": 22360
  },
  {
    "loss": 0.07715902328491211,
    "grad_norm": 0.8283431529998779,
    "learning_rate": 3.6639785032261764e-05,
    "epoch": 0.7193619963340515,
    "step": 22370
  },
  {
    "loss": 0.07304844856262208,
    "grad_norm": 0.7513096332550049,
    "learning_rate": 3.6561406078131845e-05,
    "epoch": 0.7196835707624529,
    "step": 22380
  },
  {
    "loss": 0.07067491412162781,
    "grad_norm": 0.9522845149040222,
    "learning_rate": 3.648309228893152e-05,
    "epoch": 0.7200051451908545,
    "step": 22390
  },
  {
    "loss": 0.07188115119934083,
    "grad_norm": 0.574332058429718,
    "learning_rate": 3.640484374510564e-05,
    "epoch": 0.7203267196192559,
    "step": 22400
  },
  {
    "loss": 0.0681536316871643,
    "grad_norm": 0.29085707664489746,
    "learning_rate": 3.632666052703218e-05,
    "epoch": 0.7206482940476573,
    "step": 22410
  },
  {
    "loss": 0.07831199765205384,
    "grad_norm": 0.7932929396629333,
    "learning_rate": 3.624854271502186e-05,
    "epoch": 0.7209698684760588,
    "step": 22420
  },
  {
    "loss": 0.07898212671279907,
    "grad_norm": 0.5751041769981384,
    "learning_rate": 3.6170490389318346e-05,
    "epoch": 0.7212914429044602,
    "step": 22430
  },
  {
    "loss": 0.09148269891738892,
    "grad_norm": 0.6542364358901978,
    "learning_rate": 3.609250363009793e-05,
    "epoch": 0.7216130173328617,
    "step": 22440
  },
  {
    "loss": 0.08846168518066407,
    "grad_norm": 0.705448567867279,
    "learning_rate": 3.601458251746962e-05,
    "epoch": 0.7219345917612632,
    "step": 22450
  },
  {
    "loss": 0.0739761769771576,
    "grad_norm": 0.47682419419288635,
    "learning_rate": 3.593672713147501e-05,
    "epoch": 0.7222561661896646,
    "step": 22460
  },
  {
    "loss": 0.06625583171844482,
    "grad_norm": 0.896597146987915,
    "learning_rate": 3.585893755208811e-05,
    "epoch": 0.722577740618066,
    "step": 22470
  },
  {
    "loss": 0.08493435978889466,
    "grad_norm": 0.48587241768836975,
    "learning_rate": 3.578121385921533e-05,
    "epoch": 0.7228993150464675,
    "step": 22480
  },
  {
    "loss": 0.06791111230850219,
    "grad_norm": 0.3798055052757263,
    "learning_rate": 3.570355613269551e-05,
    "epoch": 0.723220889474869,
    "step": 22490
  },
  {
    "loss": 0.07332050800323486,
    "grad_norm": 0.6129491329193115,
    "learning_rate": 3.562596445229954e-05,
    "epoch": 0.7235424639032704,
    "step": 22500
  },
  {
    "loss": 0.06496210098266601,
    "grad_norm": 0.512057363986969,
    "learning_rate": 3.5548438897730726e-05,
    "epoch": 0.7238640383316719,
    "step": 22510
  },
  {
    "loss": 0.07144524455070496,
    "grad_norm": 0.854631781578064,
    "learning_rate": 3.547097954862422e-05,
    "epoch": 0.7241856127600733,
    "step": 22520
  },
  {
    "loss": 0.06772378087043762,
    "grad_norm": 0.6502050161361694,
    "learning_rate": 3.5393586484547225e-05,
    "epoch": 0.7245071871884747,
    "step": 22530
  },
  {
    "loss": 0.06079759001731873,
    "grad_norm": 0.721480667591095,
    "learning_rate": 3.531625978499895e-05,
    "epoch": 0.7248287616168763,
    "step": 22540
  },
  {
    "loss": 0.11076893806457519,
    "grad_norm": 0.5731940865516663,
    "learning_rate": 3.52389995294103e-05,
    "epoch": 0.7251503360452777,
    "step": 22550
  },
  {
    "loss": 0.07782159447669983,
    "grad_norm": 0.5176689624786377,
    "learning_rate": 3.5161805797144035e-05,
    "epoch": 0.7254719104736791,
    "step": 22560
  },
  {
    "loss": 0.08271768689155579,
    "grad_norm": 0.8279057145118713,
    "learning_rate": 3.508467866749449e-05,
    "epoch": 0.7257934849020806,
    "step": 22570
  },
  {
    "loss": 0.09557575583457947,
    "grad_norm": 0.969346821308136,
    "learning_rate": 3.500761821968767e-05,
    "epoch": 0.726115059330482,
    "step": 22580
  },
  {
    "loss": 0.06104801893234253,
    "grad_norm": 0.49495741724967957,
    "learning_rate": 3.4930624532881064e-05,
    "epoch": 0.7264366337588835,
    "step": 22590
  },
  {
    "loss": 0.08584245443344116,
    "grad_norm": 0.7284782528877258,
    "learning_rate": 3.48536976861635e-05,
    "epoch": 0.726758208187285,
    "step": 22600
  },
  {
    "loss": 0.0828737735748291,
    "grad_norm": 0.6159257292747498,
    "learning_rate": 3.4776837758555284e-05,
    "epoch": 0.7270797826156864,
    "step": 22610
  },
  {
    "loss": 0.0713537335395813,
    "grad_norm": 0.4589337110519409,
    "learning_rate": 3.470004482900785e-05,
    "epoch": 0.7274013570440878,
    "step": 22620
  },
  {
    "loss": 0.06789127588272095,
    "grad_norm": 0.6434908509254456,
    "learning_rate": 3.4623318976403895e-05,
    "epoch": 0.7277229314724893,
    "step": 22630
  },
  {
    "loss": 0.06783631443977356,
    "grad_norm": 0.7492509484291077,
    "learning_rate": 3.454666027955722e-05,
    "epoch": 0.7280445059008908,
    "step": 22640
  },
  {
    "loss": 0.06551461815834045,
    "grad_norm": 0.4112277925014496,
    "learning_rate": 3.447006881721256e-05,
    "epoch": 0.7283660803292922,
    "step": 22650
  },
  {
    "loss": 0.07875913977622986,
    "grad_norm": 0.5258214473724365,
    "learning_rate": 3.43935446680457e-05,
    "epoch": 0.7286876547576937,
    "step": 22660
  },
  {
    "loss": 0.09203838109970093,
    "grad_norm": 0.7541487812995911,
    "learning_rate": 3.431708791066313e-05,
    "epoch": 0.7290092291860951,
    "step": 22670
  },
  {
    "loss": 0.06775216460227966,
    "grad_norm": 0.5672399401664734,
    "learning_rate": 3.42406986236023e-05,
    "epoch": 0.7293308036144965,
    "step": 22680
  },
  {
    "loss": 0.068998521566391,
    "grad_norm": 0.562434196472168,
    "learning_rate": 3.4164376885331165e-05,
    "epoch": 0.7296523780428981,
    "step": 22690
  },
  {
    "loss": 0.08524279594421387,
    "grad_norm": 0.7267500162124634,
    "learning_rate": 3.408812277424843e-05,
    "epoch": 0.7299739524712995,
    "step": 22700
  },
  {
    "loss": 0.07056506872177123,
    "grad_norm": 0.5633522868156433,
    "learning_rate": 3.401193636868332e-05,
    "epoch": 0.7302955268997009,
    "step": 22710
  },
  {
    "loss": 0.07487800121307372,
    "grad_norm": 0.5191984176635742,
    "learning_rate": 3.393581774689541e-05,
    "epoch": 0.7306171013281024,
    "step": 22720
  },
  {
    "loss": 0.05978419780731201,
    "grad_norm": 0.38137316703796387,
    "learning_rate": 3.385976698707478e-05,
    "epoch": 0.7309386757565038,
    "step": 22730
  },
  {
    "loss": 0.08034716248512268,
    "grad_norm": 0.7938235998153687,
    "learning_rate": 3.3783784167341716e-05,
    "epoch": 0.7312602501849053,
    "step": 22740
  },
  {
    "loss": 0.062334203720092775,
    "grad_norm": 0.49491098523139954,
    "learning_rate": 3.3707869365746683e-05,
    "epoch": 0.7315818246133068,
    "step": 22750
  },
  {
    "loss": 0.09403547048568725,
    "grad_norm": 0.5901187658309937,
    "learning_rate": 3.363202266027037e-05,
    "epoch": 0.7319033990417082,
    "step": 22760
  },
  {
    "loss": 0.06843697428703308,
    "grad_norm": 0.421527236700058,
    "learning_rate": 3.3556244128823475e-05,
    "epoch": 0.7322249734701096,
    "step": 22770
  },
  {
    "loss": 0.076862633228302,
    "grad_norm": 0.558758556842804,
    "learning_rate": 3.348053384924671e-05,
    "epoch": 0.7325465478985111,
    "step": 22780
  },
  {
    "loss": 0.060287344455719,
    "grad_norm": 0.3895277678966522,
    "learning_rate": 3.3404891899310585e-05,
    "epoch": 0.7328681223269126,
    "step": 22790
  },
  {
    "loss": 0.07556776404380798,
    "grad_norm": 0.5408258438110352,
    "learning_rate": 3.3329318356715466e-05,
    "epoch": 0.733189696755314,
    "step": 22800
  },
  {
    "loss": 0.07142072319984435,
    "grad_norm": 0.565141499042511,
    "learning_rate": 3.325381329909149e-05,
    "epoch": 0.7335112711837155,
    "step": 22810
  },
  {
    "loss": 0.084370356798172,
    "grad_norm": 0.7071492075920105,
    "learning_rate": 3.317837680399834e-05,
    "epoch": 0.7338328456121169,
    "step": 22820
  },
  {
    "loss": 0.09431569576263428,
    "grad_norm": 0.657227635383606,
    "learning_rate": 3.310300894892546e-05,
    "epoch": 0.7341544200405183,
    "step": 22830
  },
  {
    "loss": 0.08125004172325134,
    "grad_norm": 0.3469249904155731,
    "learning_rate": 3.302770981129162e-05,
    "epoch": 0.7344759944689199,
    "step": 22840
  },
  {
    "loss": 0.056107521057128906,
    "grad_norm": 0.5997012257575989,
    "learning_rate": 3.295247946844499e-05,
    "epoch": 0.7347975688973213,
    "step": 22850
  },
  {
    "loss": 0.07729174494743347,
    "grad_norm": 0.619046151638031,
    "learning_rate": 3.2877317997663206e-05,
    "epoch": 0.7351191433257227,
    "step": 22860
  },
  {
    "loss": 0.09256871342658997,
    "grad_norm": 0.5356802344322205,
    "learning_rate": 3.2802225476153036e-05,
    "epoch": 0.7354407177541242,
    "step": 22870
  },
  {
    "loss": 0.061709392070770266,
    "grad_norm": 0.34582701325416565,
    "learning_rate": 3.272720198105049e-05,
    "epoch": 0.7357622921825256,
    "step": 22880
  },
  {
    "loss": 0.08347746729850769,
    "grad_norm": 0.36255910992622375,
    "learning_rate": 3.265224758942069e-05,
    "epoch": 0.7360838666109271,
    "step": 22890
  },
  {
    "loss": 0.0869390308856964,
    "grad_norm": 0.4565773606300354,
    "learning_rate": 3.257736237825768e-05,
    "epoch": 0.7364054410393286,
    "step": 22900
  },
  {
    "loss": 0.06486717462539673,
    "grad_norm": 0.4815742075443268,
    "learning_rate": 3.250254642448457e-05,
    "epoch": 0.73672701546773,
    "step": 22910
  },
  {
    "loss": 0.05989484190940857,
    "grad_norm": 0.39747703075408936,
    "learning_rate": 3.242779980495318e-05,
    "epoch": 0.7370485898961314,
    "step": 22920
  },
  {
    "loss": 0.07348278164863586,
    "grad_norm": 0.7308650016784668,
    "learning_rate": 3.235312259644426e-05,
    "epoch": 0.737370164324533,
    "step": 22930
  },
  {
    "loss": 0.07345086336135864,
    "grad_norm": 0.8181717395782471,
    "learning_rate": 3.2278514875667124e-05,
    "epoch": 0.7376917387529344,
    "step": 22940
  },
  {
    "loss": 0.08308499455451965,
    "grad_norm": 0.6992674469947815,
    "learning_rate": 3.220397671925979e-05,
    "epoch": 0.7380133131813358,
    "step": 22950
  },
  {
    "loss": 0.08991779088973999,
    "grad_norm": 0.6428148150444031,
    "learning_rate": 3.2129508203788836e-05,
    "epoch": 0.7383348876097373,
    "step": 22960
  },
  {
    "loss": 0.06239253878593445,
    "grad_norm": 0.564574122428894,
    "learning_rate": 3.2055109405749214e-05,
    "epoch": 0.7386564620381387,
    "step": 22970
  },
  {
    "loss": 0.07732295989990234,
    "grad_norm": 0.6411125659942627,
    "learning_rate": 3.1980780401564384e-05,
    "epoch": 0.7389780364665401,
    "step": 22980
  },
  {
    "loss": 0.07028377056121826,
    "grad_norm": 0.5637052059173584,
    "learning_rate": 3.1906521267585956e-05,
    "epoch": 0.7392996108949417,
    "step": 22990
  },
  {
    "loss": 0.07760335206985473,
    "grad_norm": 0.591373860836029,
    "learning_rate": 3.183233208009393e-05,
    "epoch": 0.7396211853233431,
    "step": 23000
  },
  {
    "loss": 0.06789370179176331,
    "grad_norm": 0.7457368969917297,
    "learning_rate": 3.175821291529632e-05,
    "epoch": 0.7399427597517445,
    "step": 23010
  },
  {
    "loss": 0.08317228555679321,
    "grad_norm": 0.7319291830062866,
    "learning_rate": 3.16841638493293e-05,
    "epoch": 0.740264334180146,
    "step": 23020
  },
  {
    "loss": 0.07127461433410645,
    "grad_norm": 0.7333948612213135,
    "learning_rate": 3.161018495825705e-05,
    "epoch": 0.7405859086085474,
    "step": 23030
  },
  {
    "loss": 0.0749741554260254,
    "grad_norm": 0.6388927102088928,
    "learning_rate": 3.1536276318071564e-05,
    "epoch": 0.7409074830369489,
    "step": 23040
  },
  {
    "loss": 0.06397749185562134,
    "grad_norm": 0.2538682520389557,
    "learning_rate": 3.1462438004692796e-05,
    "epoch": 0.7412290574653504,
    "step": 23050
  },
  {
    "loss": 0.06646652817726136,
    "grad_norm": 0.5968759059906006,
    "learning_rate": 3.138867009396836e-05,
    "epoch": 0.7415506318937518,
    "step": 23060
  },
  {
    "loss": 0.05975983738899231,
    "grad_norm": 0.4150048494338989,
    "learning_rate": 3.131497266167357e-05,
    "epoch": 0.7418722063221532,
    "step": 23070
  },
  {
    "loss": 0.06749844551086426,
    "grad_norm": 0.5127459168434143,
    "learning_rate": 3.12413457835114e-05,
    "epoch": 0.7421937807505548,
    "step": 23080
  },
  {
    "loss": 0.06747809648513795,
    "grad_norm": 1.4364615678787231,
    "learning_rate": 3.116778953511233e-05,
    "epoch": 0.7425153551789562,
    "step": 23090
  },
  {
    "loss": 0.06821455359458924,
    "grad_norm": 0.8889177441596985,
    "learning_rate": 3.1094303992034234e-05,
    "epoch": 0.7428369296073576,
    "step": 23100
  },
  {
    "loss": 0.07234838008880615,
    "grad_norm": 0.5564526319503784,
    "learning_rate": 3.1020889229762427e-05,
    "epoch": 0.7431585040357591,
    "step": 23110
  },
  {
    "loss": 0.07779403924942016,
    "grad_norm": 0.5989982485771179,
    "learning_rate": 3.094754532370945e-05,
    "epoch": 0.7434800784641605,
    "step": 23120
  },
  {
    "loss": 0.07060418128967286,
    "grad_norm": 0.6478323936462402,
    "learning_rate": 3.087427234921515e-05,
    "epoch": 0.743801652892562,
    "step": 23130
  },
  {
    "loss": 0.06922822594642639,
    "grad_norm": 0.33565738797187805,
    "learning_rate": 3.080107038154638e-05,
    "epoch": 0.7441232273209635,
    "step": 23140
  },
  {
    "loss": 0.05781306624412537,
    "grad_norm": 0.6275683641433716,
    "learning_rate": 3.072793949589718e-05,
    "epoch": 0.7444448017493649,
    "step": 23150
  },
  {
    "loss": 0.0717390537261963,
    "grad_norm": 0.9177708029747009,
    "learning_rate": 3.0654879767388546e-05,
    "epoch": 0.7447663761777663,
    "step": 23160
  },
  {
    "loss": 0.08312966227531433,
    "grad_norm": 0.7901967167854309,
    "learning_rate": 3.0581891271068305e-05,
    "epoch": 0.7450879506061678,
    "step": 23170
  },
  {
    "loss": 0.07024783492088318,
    "grad_norm": 0.6758073568344116,
    "learning_rate": 3.0508974081911224e-05,
    "epoch": 0.7454095250345693,
    "step": 23180
  },
  {
    "loss": 0.06823955178260803,
    "grad_norm": 0.5088121891021729,
    "learning_rate": 3.043612827481871e-05,
    "epoch": 0.7457310994629707,
    "step": 23190
  },
  {
    "loss": 0.07273564338684083,
    "grad_norm": 0.5752536654472351,
    "learning_rate": 3.0363353924618922e-05,
    "epoch": 0.7460526738913722,
    "step": 23200
  },
  {
    "loss": 0.06960035562515259,
    "grad_norm": 0.9873530268669128,
    "learning_rate": 3.0290651106066635e-05,
    "epoch": 0.7463742483197736,
    "step": 23210
  },
  {
    "loss": 0.06923828125,
    "grad_norm": 0.23608741164207458,
    "learning_rate": 3.0218019893843054e-05,
    "epoch": 0.746695822748175,
    "step": 23220
  },
  {
    "loss": 0.06952532529830932,
    "grad_norm": 0.5447070002555847,
    "learning_rate": 3.014546036255592e-05,
    "epoch": 0.7470173971765766,
    "step": 23230
  },
  {
    "loss": 0.06679846048355102,
    "grad_norm": 0.5517072081565857,
    "learning_rate": 3.0072972586739267e-05,
    "epoch": 0.747338971604978,
    "step": 23240
  },
  {
    "loss": 0.08158867955207824,
    "grad_norm": 0.46643969416618347,
    "learning_rate": 3.0000556640853515e-05,
    "epoch": 0.7476605460333794,
    "step": 23250
  },
  {
    "loss": 0.051595062017440796,
    "grad_norm": 0.6006515622138977,
    "learning_rate": 2.9928212599285177e-05,
    "epoch": 0.7479821204617809,
    "step": 23260
  },
  {
    "loss": 0.06733589768409728,
    "grad_norm": 0.5819828510284424,
    "learning_rate": 2.9855940536347004e-05,
    "epoch": 0.7483036948901823,
    "step": 23270
  },
  {
    "loss": 0.09315139055252075,
    "grad_norm": 0.5738916993141174,
    "learning_rate": 2.9783740526277803e-05,
    "epoch": 0.7486252693185838,
    "step": 23280
  },
  {
    "loss": 0.059750539064407346,
    "grad_norm": 0.9257296919822693,
    "learning_rate": 2.9711612643242304e-05,
    "epoch": 0.7489468437469853,
    "step": 23290
  },
  {
    "loss": 0.07237733006477357,
    "grad_norm": 0.38398125767707825,
    "learning_rate": 2.9639556961331217e-05,
    "epoch": 0.7492684181753867,
    "step": 23300
  },
  {
    "loss": 0.07754396796226501,
    "grad_norm": 0.7336930632591248,
    "learning_rate": 2.9567573554561033e-05,
    "epoch": 0.7495899926037881,
    "step": 23310
  },
  {
    "eval_loss": 0.07043655216693878,
    "eval_runtime": 34.5596,
    "eval_samples_per_second": 145.43,
    "eval_steps_per_second": 36.372,
    "epoch": 0.7495899926037881,
    "step": 23310
  },
  {
    "loss": 0.06745206713676452,
    "grad_norm": 0.65149986743927,
    "learning_rate": 2.9495662496873988e-05,
    "epoch": 0.7499115670321896,
    "step": 23320
  },
  {
    "loss": 0.06239452362060547,
    "grad_norm": 0.6391117572784424,
    "learning_rate": 2.9423823862138076e-05,
    "epoch": 0.7502331414605911,
    "step": 23330
  },
  {
    "loss": 0.07942205071449279,
    "grad_norm": 0.4198969602584839,
    "learning_rate": 2.9352057724146853e-05,
    "epoch": 0.7505547158889925,
    "step": 23340
  },
  {
    "loss": 0.0804401993751526,
    "grad_norm": 0.7204815745353699,
    "learning_rate": 2.928036415661942e-05,
    "epoch": 0.750876290317394,
    "step": 23350
  },
  {
    "loss": 0.06356263160705566,
    "grad_norm": 0.49734702706336975,
    "learning_rate": 2.9208743233200318e-05,
    "epoch": 0.7511978647457954,
    "step": 23360
  },
  {
    "loss": 0.06504639983177185,
    "grad_norm": 0.4685254693031311,
    "learning_rate": 2.9137195027459418e-05,
    "epoch": 0.7515194391741968,
    "step": 23370
  },
  {
    "loss": 0.06596590876579285,
    "grad_norm": 0.5883439183235168,
    "learning_rate": 2.9065719612892018e-05,
    "epoch": 0.7518410136025984,
    "step": 23380
  },
  {
    "loss": 0.05896996855735779,
    "grad_norm": 0.4913024604320526,
    "learning_rate": 2.8994317062918462e-05,
    "epoch": 0.7521625880309998,
    "step": 23390
  },
  {
    "loss": 0.07391130924224854,
    "grad_norm": 0.690483033657074,
    "learning_rate": 2.8922987450884496e-05,
    "epoch": 0.7524841624594012,
    "step": 23400
  },
  {
    "loss": 0.0676592767238617,
    "grad_norm": 0.8345645070075989,
    "learning_rate": 2.8851730850060743e-05,
    "epoch": 0.7528057368878027,
    "step": 23410
  },
  {
    "loss": 0.0608561635017395,
    "grad_norm": 0.536762535572052,
    "learning_rate": 2.878054733364286e-05,
    "epoch": 0.7531273113162041,
    "step": 23420
  },
  {
    "loss": 0.056133496761322024,
    "grad_norm": 0.4661767780780792,
    "learning_rate": 2.87094369747515e-05,
    "epoch": 0.7534488857446056,
    "step": 23430
  },
  {
    "loss": 0.0817335844039917,
    "grad_norm": 0.5510740876197815,
    "learning_rate": 2.8638399846432083e-05,
    "epoch": 0.7537704601730071,
    "step": 23440
  },
  {
    "loss": 0.06697210073471069,
    "grad_norm": 0.7082464098930359,
    "learning_rate": 2.8567436021654914e-05,
    "epoch": 0.7540920346014085,
    "step": 23450
  },
  {
    "loss": 0.07213478684425353,
    "grad_norm": 0.3922865688800812,
    "learning_rate": 2.849654557331486e-05,
    "epoch": 0.7544136090298099,
    "step": 23460
  },
  {
    "loss": 0.07061719298362731,
    "grad_norm": 0.5049762725830078,
    "learning_rate": 2.8425728574231537e-05,
    "epoch": 0.7547351834582114,
    "step": 23470
  },
  {
    "loss": 0.08335615992546082,
    "grad_norm": 0.24221433699131012,
    "learning_rate": 2.835498509714908e-05,
    "epoch": 0.7550567578866129,
    "step": 23480
  },
  {
    "loss": 0.09131989479064942,
    "grad_norm": 0.6416606307029724,
    "learning_rate": 2.8284315214736036e-05,
    "epoch": 0.7553783323150143,
    "step": 23490
  },
  {
    "loss": 0.07481069564819336,
    "grad_norm": 0.9378443956375122,
    "learning_rate": 2.821371899958547e-05,
    "epoch": 0.7556999067434158,
    "step": 23500
  },
  {
    "loss": 0.09523290395736694,
    "grad_norm": 0.5338574051856995,
    "learning_rate": 2.814319652421463e-05,
    "epoch": 0.7560214811718172,
    "step": 23510
  },
  {
    "loss": 0.08019073605537415,
    "grad_norm": 0.6044603586196899,
    "learning_rate": 2.8072747861065164e-05,
    "epoch": 0.7563430556002186,
    "step": 23520
  },
  {
    "loss": 0.05918534398078919,
    "grad_norm": 0.5362686514854431,
    "learning_rate": 2.800237308250283e-05,
    "epoch": 0.7566646300286202,
    "step": 23530
  },
  {
    "loss": 0.06204081177711487,
    "grad_norm": 0.5132609605789185,
    "learning_rate": 2.7932072260817453e-05,
    "epoch": 0.7569862044570216,
    "step": 23540
  },
  {
    "loss": 0.07708427309989929,
    "grad_norm": 0.719713568687439,
    "learning_rate": 2.7861845468222992e-05,
    "epoch": 0.757307778885423,
    "step": 23550
  },
  {
    "loss": 0.053664219379425046,
    "grad_norm": 0.40919116139411926,
    "learning_rate": 2.7791692776857225e-05,
    "epoch": 0.7576293533138245,
    "step": 23560
  },
  {
    "loss": 0.0843777060508728,
    "grad_norm": 0.8862894773483276,
    "learning_rate": 2.772161425878196e-05,
    "epoch": 0.7579509277422259,
    "step": 23570
  },
  {
    "loss": 0.08774664998054504,
    "grad_norm": 0.7200607657432556,
    "learning_rate": 2.7651609985982674e-05,
    "epoch": 0.7582725021706274,
    "step": 23580
  },
  {
    "loss": 0.07632976174354553,
    "grad_norm": 0.7512308359146118,
    "learning_rate": 2.7581680030368685e-05,
    "epoch": 0.7585940765990289,
    "step": 23590
  },
  {
    "loss": 0.05972144603729248,
    "grad_norm": 0.4622749090194702,
    "learning_rate": 2.751182446377295e-05,
    "epoch": 0.7589156510274303,
    "step": 23600
  },
  {
    "loss": 0.0827726423740387,
    "grad_norm": 0.8650885224342346,
    "learning_rate": 2.744204335795194e-05,
    "epoch": 0.7592372254558317,
    "step": 23610
  },
  {
    "loss": 0.10206010341644287,
    "grad_norm": 0.9212839007377625,
    "learning_rate": 2.7372336784585762e-05,
    "epoch": 0.7595587998842332,
    "step": 23620
  },
  {
    "loss": 0.058333611488342284,
    "grad_norm": 0.5360256433486938,
    "learning_rate": 2.7302704815277848e-05,
    "epoch": 0.7598803743126347,
    "step": 23630
  },
  {
    "loss": 0.08178957104682923,
    "grad_norm": 0.7339324355125427,
    "learning_rate": 2.7233147521555025e-05,
    "epoch": 0.7602019487410361,
    "step": 23640
  },
  {
    "loss": 0.07520390748977661,
    "grad_norm": 0.4708062410354614,
    "learning_rate": 2.716366497486744e-05,
    "epoch": 0.7605235231694376,
    "step": 23650
  },
  {
    "loss": 0.07630211710929871,
    "grad_norm": 0.7849609851837158,
    "learning_rate": 2.7094257246588474e-05,
    "epoch": 0.760845097597839,
    "step": 23660
  },
  {
    "loss": 0.07333303689956665,
    "grad_norm": 0.6058045625686646,
    "learning_rate": 2.702492440801464e-05,
    "epoch": 0.7611666720262404,
    "step": 23670
  },
  {
    "loss": 0.08173723220825195,
    "grad_norm": 0.5990397334098816,
    "learning_rate": 2.6955666530365476e-05,
    "epoch": 0.761488246454642,
    "step": 23680
  },
  {
    "loss": 0.08724742531776428,
    "grad_norm": 0.5341375470161438,
    "learning_rate": 2.6886483684783536e-05,
    "epoch": 0.7618098208830434,
    "step": 23690
  },
  {
    "loss": 0.06513100862503052,
    "grad_norm": 0.5419607162475586,
    "learning_rate": 2.6817375942334376e-05,
    "epoch": 0.7621313953114448,
    "step": 23700
  },
  {
    "loss": 0.0673224687576294,
    "grad_norm": 0.6282056570053101,
    "learning_rate": 2.6748343374006256e-05,
    "epoch": 0.7624529697398463,
    "step": 23710
  },
  {
    "loss": 0.06728132367134095,
    "grad_norm": 0.7399762272834778,
    "learning_rate": 2.6679386050710418e-05,
    "epoch": 0.7627745441682477,
    "step": 23720
  },
  {
    "loss": 0.06391294002532959,
    "grad_norm": 0.3468599021434784,
    "learning_rate": 2.661050404328065e-05,
    "epoch": 0.7630961185966492,
    "step": 23730
  },
  {
    "loss": 0.07310450673103333,
    "grad_norm": 0.7776421904563904,
    "learning_rate": 2.65416974224734e-05,
    "epoch": 0.7634176930250507,
    "step": 23740
  },
  {
    "loss": 0.06410598754882812,
    "grad_norm": 0.4238409698009491,
    "learning_rate": 2.6472966258967746e-05,
    "epoch": 0.7637392674534521,
    "step": 23750
  },
  {
    "loss": 0.0651436448097229,
    "grad_norm": 0.6460195183753967,
    "learning_rate": 2.640431062336517e-05,
    "epoch": 0.7640608418818535,
    "step": 23760
  },
  {
    "loss": 0.08409774303436279,
    "grad_norm": 0.5648590326309204,
    "learning_rate": 2.6335730586189654e-05,
    "epoch": 0.764382416310255,
    "step": 23770
  },
  {
    "loss": 0.07830084562301635,
    "grad_norm": 1.1920636892318726,
    "learning_rate": 2.626722621788744e-05,
    "epoch": 0.7647039907386565,
    "step": 23780
  },
  {
    "loss": 0.052944546937942503,
    "grad_norm": 0.7563640475273132,
    "learning_rate": 2.619879758882712e-05,
    "epoch": 0.7650255651670579,
    "step": 23790
  },
  {
    "loss": 0.0660877764225006,
    "grad_norm": 0.3552987575531006,
    "learning_rate": 2.6130444769299456e-05,
    "epoch": 0.7653471395954594,
    "step": 23800
  },
  {
    "loss": 0.07947134375572204,
    "grad_norm": 0.5222949385643005,
    "learning_rate": 2.606216782951729e-05,
    "epoch": 0.7656687140238608,
    "step": 23810
  },
  {
    "loss": 0.07922396659851075,
    "grad_norm": 0.9539393782615662,
    "learning_rate": 2.5993966839615635e-05,
    "epoch": 0.7659902884522622,
    "step": 23820
  },
  {
    "loss": 0.0755894124507904,
    "grad_norm": 0.4629724621772766,
    "learning_rate": 2.5925841869651323e-05,
    "epoch": 0.7663118628806638,
    "step": 23830
  },
  {
    "loss": 0.05568032264709473,
    "grad_norm": 0.380663126707077,
    "learning_rate": 2.5857792989603236e-05,
    "epoch": 0.7666334373090652,
    "step": 23840
  },
  {
    "loss": 0.07649948596954345,
    "grad_norm": 0.47023138403892517,
    "learning_rate": 2.578982026937208e-05,
    "epoch": 0.7669550117374666,
    "step": 23850
  },
  {
    "loss": 0.0708249032497406,
    "grad_norm": 0.6598566174507141,
    "learning_rate": 2.5721923778780233e-05,
    "epoch": 0.7672765861658681,
    "step": 23860
  },
  {
    "loss": 0.07922458052635192,
    "grad_norm": 0.7775975465774536,
    "learning_rate": 2.565410358757189e-05,
    "epoch": 0.7675981605942696,
    "step": 23870
  },
  {
    "loss": 0.0724449872970581,
    "grad_norm": 0.3954971432685852,
    "learning_rate": 2.5586359765412758e-05,
    "epoch": 0.767919735022671,
    "step": 23880
  },
  {
    "loss": 0.06616113781929016,
    "grad_norm": 0.603389322757721,
    "learning_rate": 2.5518692381890207e-05,
    "epoch": 0.7682413094510725,
    "step": 23890
  },
  {
    "loss": 0.07546033263206482,
    "grad_norm": 0.6811168789863586,
    "learning_rate": 2.5451101506512977e-05,
    "epoch": 0.7685628838794739,
    "step": 23900
  },
  {
    "loss": 0.07617400288581848,
    "grad_norm": 1.2383785247802734,
    "learning_rate": 2.53835872087113e-05,
    "epoch": 0.7688844583078753,
    "step": 23910
  },
  {
    "loss": 0.07465756535530091,
    "grad_norm": 0.565261960029602,
    "learning_rate": 2.5316149557836753e-05,
    "epoch": 0.7692060327362769,
    "step": 23920
  },
  {
    "loss": 0.06164060235023498,
    "grad_norm": 0.33716362714767456,
    "learning_rate": 2.524878862316209e-05,
    "epoch": 0.7695276071646783,
    "step": 23930
  },
  {
    "loss": 0.08892890214920043,
    "grad_norm": 0.6663774251937866,
    "learning_rate": 2.5181504473881378e-05,
    "epoch": 0.7698491815930797,
    "step": 23940
  },
  {
    "loss": 0.08810426592826844,
    "grad_norm": 0.7194262146949768,
    "learning_rate": 2.5114297179109714e-05,
    "epoch": 0.7701707560214812,
    "step": 23950
  },
  {
    "loss": 0.07091661691665649,
    "grad_norm": 0.4901579022407532,
    "learning_rate": 2.504716680788325e-05,
    "epoch": 0.7704923304498826,
    "step": 23960
  },
  {
    "loss": 0.06549544334411621,
    "grad_norm": 0.6214858889579773,
    "learning_rate": 2.49801134291592e-05,
    "epoch": 0.770813904878284,
    "step": 23970
  },
  {
    "loss": 0.06393520832061768,
    "grad_norm": 0.49949368834495544,
    "learning_rate": 2.4913137111815676e-05,
    "epoch": 0.7711354793066856,
    "step": 23980
  },
  {
    "loss": 0.06656065583229065,
    "grad_norm": 0.7732129096984863,
    "learning_rate": 2.4846237924651537e-05,
    "epoch": 0.771457053735087,
    "step": 23990
  },
  {
    "loss": 0.07589731812477112,
    "grad_norm": 0.3842383921146393,
    "learning_rate": 2.477941593638654e-05,
    "epoch": 0.7717786281634884,
    "step": 24000
  },
  {
    "loss": 0.06695120334625244,
    "grad_norm": 0.6397907137870789,
    "learning_rate": 2.471267121566101e-05,
    "epoch": 0.7721002025918899,
    "step": 24010
  },
  {
    "loss": 0.07762157917022705,
    "grad_norm": 0.8455729484558105,
    "learning_rate": 2.4646003831036048e-05,
    "epoch": 0.7724217770202914,
    "step": 24020
  },
  {
    "loss": 0.07489639520645142,
    "grad_norm": 0.48678871989250183,
    "learning_rate": 2.457941385099317e-05,
    "epoch": 0.7727433514486928,
    "step": 24030
  },
  {
    "loss": 0.08373913168907166,
    "grad_norm": 0.6976548433303833,
    "learning_rate": 2.451290134393448e-05,
    "epoch": 0.7730649258770943,
    "step": 24040
  },
  {
    "loss": 0.04959548711776733,
    "grad_norm": 0.21124552190303802,
    "learning_rate": 2.44464663781825e-05,
    "epoch": 0.7733865003054957,
    "step": 24050
  },
  {
    "loss": 0.08356249928474427,
    "grad_norm": 0.789567232131958,
    "learning_rate": 2.4380109021980002e-05,
    "epoch": 0.7737080747338971,
    "step": 24060
  },
  {
    "loss": 0.058235388994216916,
    "grad_norm": 0.684293806552887,
    "learning_rate": 2.431382934349018e-05,
    "epoch": 0.7740296491622987,
    "step": 24070
  },
  {
    "loss": 0.0808126986026764,
    "grad_norm": 0.6744261980056763,
    "learning_rate": 2.4247627410796304e-05,
    "epoch": 0.7743512235907001,
    "step": 24080
  },
  {
    "loss": 0.07473966479301453,
    "grad_norm": 0.7529351711273193,
    "learning_rate": 2.4181503291901852e-05,
    "epoch": 0.7746727980191015,
    "step": 24090
  },
  {
    "loss": 0.07294783592224122,
    "grad_norm": 0.6820980310440063,
    "learning_rate": 2.4115457054730406e-05,
    "epoch": 0.774994372447503,
    "step": 24100
  },
  {
    "loss": 0.05817674994468689,
    "grad_norm": 0.41013917326927185,
    "learning_rate": 2.404948876712543e-05,
    "epoch": 0.7753159468759044,
    "step": 24110
  },
  {
    "loss": 0.0703526794910431,
    "grad_norm": 0.7143634557723999,
    "learning_rate": 2.3983598496850445e-05,
    "epoch": 0.7756375213043059,
    "step": 24120
  },
  {
    "loss": 0.07235896587371826,
    "grad_norm": 0.7329757213592529,
    "learning_rate": 2.391778631158873e-05,
    "epoch": 0.7759590957327074,
    "step": 24130
  },
  {
    "loss": 0.06436983942985534,
    "grad_norm": 0.4656648337841034,
    "learning_rate": 2.3852052278943437e-05,
    "epoch": 0.7762806701611088,
    "step": 24140
  },
  {
    "loss": 0.06897568106651306,
    "grad_norm": 0.8447166085243225,
    "learning_rate": 2.3786396466437356e-05,
    "epoch": 0.7766022445895102,
    "step": 24150
  },
  {
    "loss": 0.066708505153656,
    "grad_norm": 0.6804686188697815,
    "learning_rate": 2.372081894151299e-05,
    "epoch": 0.7769238190179117,
    "step": 24160
  },
  {
    "loss": 0.05716971158981323,
    "grad_norm": 0.5383962988853455,
    "learning_rate": 2.3655319771532448e-05,
    "epoch": 0.7772453934463132,
    "step": 24170
  },
  {
    "loss": 0.0641344964504242,
    "grad_norm": 0.4108293950557709,
    "learning_rate": 2.3589899023777228e-05,
    "epoch": 0.7775669678747146,
    "step": 24180
  },
  {
    "loss": 0.059736895561218264,
    "grad_norm": 0.5595316886901855,
    "learning_rate": 2.3524556765448436e-05,
    "epoch": 0.7778885423031161,
    "step": 24190
  },
  {
    "loss": 0.058888697624206544,
    "grad_norm": 0.24895723164081573,
    "learning_rate": 2.3459293063666434e-05,
    "epoch": 0.7782101167315175,
    "step": 24200
  },
  {
    "loss": 0.08416748046875,
    "grad_norm": 0.7485470771789551,
    "learning_rate": 2.339410798547088e-05,
    "epoch": 0.7785316911599189,
    "step": 24210
  },
  {
    "loss": 0.0762722373008728,
    "grad_norm": 0.5820397734642029,
    "learning_rate": 2.3329001597820766e-05,
    "epoch": 0.7788532655883205,
    "step": 24220
  },
  {
    "loss": 0.09172817468643188,
    "grad_norm": 0.7628319263458252,
    "learning_rate": 2.3263973967594188e-05,
    "epoch": 0.7791748400167219,
    "step": 24230
  },
  {
    "loss": 0.06545842289924622,
    "grad_norm": 0.53376704454422,
    "learning_rate": 2.3199025161588385e-05,
    "epoch": 0.7794964144451233,
    "step": 24240
  },
  {
    "loss": 0.08564090728759766,
    "grad_norm": 0.6967863440513611,
    "learning_rate": 2.3134155246519574e-05,
    "epoch": 0.7798179888735248,
    "step": 24250
  },
  {
    "loss": 0.08012107610702515,
    "grad_norm": 0.5019391179084778,
    "learning_rate": 2.306936428902291e-05,
    "epoch": 0.7801395633019262,
    "step": 24260
  },
  {
    "loss": 0.0676426649093628,
    "grad_norm": 0.7035797834396362,
    "learning_rate": 2.300465235565257e-05,
    "epoch": 0.7804611377303277,
    "step": 24270
  },
  {
    "loss": 0.08119889497756957,
    "grad_norm": 1.0672072172164917,
    "learning_rate": 2.2940019512881363e-05,
    "epoch": 0.7807827121587292,
    "step": 24280
  },
  {
    "loss": 0.06657284498214722,
    "grad_norm": 0.7331158518791199,
    "learning_rate": 2.2875465827101082e-05,
    "epoch": 0.7811042865871306,
    "step": 24290
  },
  {
    "loss": 0.07054007053375244,
    "grad_norm": 0.5323323607444763,
    "learning_rate": 2.2810991364622057e-05,
    "epoch": 0.781425861015532,
    "step": 24300
  },
  {
    "loss": 0.0654269278049469,
    "grad_norm": 0.8353674411773682,
    "learning_rate": 2.274659619167324e-05,
    "epoch": 0.7817474354439335,
    "step": 24310
  },
  {
    "loss": 0.06039928197860718,
    "grad_norm": 0.8380869626998901,
    "learning_rate": 2.268228037440222e-05,
    "epoch": 0.782069009872335,
    "step": 24320
  },
  {
    "loss": 0.08483896255493165,
    "grad_norm": 1.4444787502288818,
    "learning_rate": 2.2618043978874957e-05,
    "epoch": 0.7823905843007364,
    "step": 24330
  },
  {
    "loss": 0.07325981259346008,
    "grad_norm": 0.6243902444839478,
    "learning_rate": 2.2553887071075963e-05,
    "epoch": 0.7827121587291379,
    "step": 24340
  },
  {
    "loss": 0.07584742903709411,
    "grad_norm": 0.8110746741294861,
    "learning_rate": 2.2489809716907972e-05,
    "epoch": 0.7830337331575393,
    "step": 24350
  },
  {
    "loss": 0.0839657187461853,
    "grad_norm": 0.5827569365501404,
    "learning_rate": 2.242581198219208e-05,
    "epoch": 0.7833553075859407,
    "step": 24360
  },
  {
    "loss": 0.06504515409469605,
    "grad_norm": 0.7146581411361694,
    "learning_rate": 2.23618939326676e-05,
    "epoch": 0.7836768820143423,
    "step": 24370
  },
  {
    "loss": 0.08918578624725342,
    "grad_norm": 0.8743788599967957,
    "learning_rate": 2.229805563399191e-05,
    "epoch": 0.7839984564427437,
    "step": 24380
  },
  {
    "loss": 0.061603051424026486,
    "grad_norm": 0.7088946104049683,
    "learning_rate": 2.223429715174058e-05,
    "epoch": 0.7843200308711451,
    "step": 24390
  },
  {
    "loss": 0.07485917806625367,
    "grad_norm": 0.4197693467140198,
    "learning_rate": 2.2170618551407075e-05,
    "epoch": 0.7846416052995466,
    "step": 24400
  },
  {
    "loss": 0.0752731740474701,
    "grad_norm": 0.7681658267974854,
    "learning_rate": 2.2107019898402892e-05,
    "epoch": 0.784963179727948,
    "step": 24410
  },
  {
    "loss": 0.06716175079345703,
    "grad_norm": 0.5857636332511902,
    "learning_rate": 2.2043501258057396e-05,
    "epoch": 0.7852847541563495,
    "step": 24420
  },
  {
    "loss": 0.08920994400978088,
    "grad_norm": 1.1019350290298462,
    "learning_rate": 2.1980062695617686e-05,
    "epoch": 0.785606328584751,
    "step": 24430
  },
  {
    "loss": 0.07643487453460693,
    "grad_norm": 0.5771538615226746,
    "learning_rate": 2.1916704276248724e-05,
    "epoch": 0.7859279030131524,
    "step": 24440
  },
  {
    "loss": 0.06678688526153564,
    "grad_norm": 0.3816686272621155,
    "learning_rate": 2.1853426065033022e-05,
    "epoch": 0.7862494774415538,
    "step": 24450
  },
  {
    "loss": 0.07544881701469422,
    "grad_norm": 0.3709632158279419,
    "learning_rate": 2.1790228126970825e-05,
    "epoch": 0.7865710518699554,
    "step": 24460
  },
  {
    "loss": 0.07349275946617126,
    "grad_norm": 0.8929899334907532,
    "learning_rate": 2.172711052697982e-05,
    "epoch": 0.7868926262983568,
    "step": 24470
  },
  {
    "loss": 0.08367403745651245,
    "grad_norm": 0.44464945793151855,
    "learning_rate": 2.1664073329895153e-05,
    "epoch": 0.7872142007267582,
    "step": 24480
  },
  {
    "loss": 0.07415645718574523,
    "grad_norm": 0.4882439076900482,
    "learning_rate": 2.160111660046955e-05,
    "epoch": 0.7875357751551597,
    "step": 24490
  },
  {
    "loss": 0.07126249670982361,
    "grad_norm": 0.5081717371940613,
    "learning_rate": 2.1538240403372877e-05,
    "epoch": 0.7878573495835611,
    "step": 24500
  },
  {
    "loss": 0.05568731427192688,
    "grad_norm": 0.4507085084915161,
    "learning_rate": 2.1475444803192424e-05,
    "epoch": 0.7881789240119625,
    "step": 24510
  },
  {
    "loss": 0.06841294169425964,
    "grad_norm": 0.4606808125972748,
    "learning_rate": 2.141272986443259e-05,
    "epoch": 0.7885004984403641,
    "step": 24520
  },
  {
    "loss": 0.07865042686462402,
    "grad_norm": 0.5381951928138733,
    "learning_rate": 2.1350095651514935e-05,
    "epoch": 0.7888220728687655,
    "step": 24530
  },
  {
    "loss": 0.06005370020866394,
    "grad_norm": 0.5262276530265808,
    "learning_rate": 2.1287542228778145e-05,
    "epoch": 0.7891436472971669,
    "step": 24540
  },
  {
    "loss": 0.06507790684700013,
    "grad_norm": 0.741633951663971,
    "learning_rate": 2.122506966047789e-05,
    "epoch": 0.7894652217255684,
    "step": 24550
  },
  {
    "loss": 0.07987558841705322,
    "grad_norm": 0.6250978112220764,
    "learning_rate": 2.116267801078682e-05,
    "epoch": 0.7897867961539699,
    "step": 24560
  },
  {
    "loss": 0.06909897923469543,
    "grad_norm": 0.5917408466339111,
    "learning_rate": 2.1100367343794392e-05,
    "epoch": 0.7901083705823713,
    "step": 24570
  },
  {
    "loss": 0.07253650426864625,
    "grad_norm": 0.8479233384132385,
    "learning_rate": 2.1038137723506903e-05,
    "epoch": 0.7904299450107728,
    "step": 24580
  },
  {
    "loss": 0.06407266855239868,
    "grad_norm": 0.5140347480773926,
    "learning_rate": 2.0975989213847447e-05,
    "epoch": 0.7907515194391742,
    "step": 24590
  },
  {
    "loss": 0.08873333334922791,
    "grad_norm": 0.687673807144165,
    "learning_rate": 2.0913921878655683e-05,
    "epoch": 0.7910730938675756,
    "step": 24600
  },
  {
    "loss": 0.08552309274673461,
    "grad_norm": 0.9812746047973633,
    "learning_rate": 2.085193578168808e-05,
    "epoch": 0.7913946682959772,
    "step": 24610
  },
  {
    "loss": 0.082278311252594,
    "grad_norm": 0.524066686630249,
    "learning_rate": 2.0790030986617493e-05,
    "epoch": 0.7917162427243786,
    "step": 24620
  },
  {
    "loss": 0.07650412917137146,
    "grad_norm": 0.30247119069099426,
    "learning_rate": 2.0728207557033286e-05,
    "epoch": 0.79203781715278,
    "step": 24630
  },
  {
    "loss": 0.06983165144920349,
    "grad_norm": 0.737944483757019,
    "learning_rate": 2.0666465556441317e-05,
    "epoch": 0.7923593915811815,
    "step": 24640
  },
  {
    "loss": 0.07627383470535279,
    "grad_norm": 0.4333503544330597,
    "learning_rate": 2.060480504826371e-05,
    "epoch": 0.7926809660095829,
    "step": 24650
  },
  {
    "loss": 0.0832338273525238,
    "grad_norm": 0.8529727458953857,
    "learning_rate": 2.0543226095838964e-05,
    "epoch": 0.7930025404379843,
    "step": 24660
  },
  {
    "loss": 0.0783322811126709,
    "grad_norm": 0.8573735356330872,
    "learning_rate": 2.0481728762421726e-05,
    "epoch": 0.7933241148663859,
    "step": 24670
  },
  {
    "loss": 0.07682687044143677,
    "grad_norm": 0.5854844450950623,
    "learning_rate": 2.0420313111182843e-05,
    "epoch": 0.7936456892947873,
    "step": 24680
  },
  {
    "loss": 0.07580558657646179,
    "grad_norm": 0.6965992450714111,
    "learning_rate": 2.0358979205209295e-05,
    "epoch": 0.7939672637231887,
    "step": 24690
  },
  {
    "loss": 0.07111702561378479,
    "grad_norm": 0.688252329826355,
    "learning_rate": 2.0297727107503993e-05,
    "epoch": 0.7942888381515902,
    "step": 24700
  },
  {
    "loss": 0.07677547931671143,
    "grad_norm": 0.5384435653686523,
    "learning_rate": 2.0236556880985912e-05,
    "epoch": 0.7946104125799917,
    "step": 24710
  },
  {
    "loss": 0.0677608609199524,
    "grad_norm": 0.5243275165557861,
    "learning_rate": 2.0175468588489844e-05,
    "epoch": 0.7949319870083931,
    "step": 24720
  },
  {
    "loss": 0.0720300555229187,
    "grad_norm": 0.4929002523422241,
    "learning_rate": 2.011446229276649e-05,
    "epoch": 0.7952535614367946,
    "step": 24730
  },
  {
    "loss": 0.08252569437026977,
    "grad_norm": 0.35379138588905334,
    "learning_rate": 2.0053538056482303e-05,
    "epoch": 0.795575135865196,
    "step": 24740
  },
  {
    "loss": 0.07043085098266602,
    "grad_norm": 0.5493075847625732,
    "learning_rate": 1.9992695942219385e-05,
    "epoch": 0.7958967102935974,
    "step": 24750
  },
  {
    "loss": 0.09129286408424378,
    "grad_norm": 0.7228595018386841,
    "learning_rate": 1.9931936012475592e-05,
    "epoch": 0.796218284721999,
    "step": 24760
  },
  {
    "loss": 0.06351727247238159,
    "grad_norm": 1.022997498512268,
    "learning_rate": 1.9871258329664233e-05,
    "epoch": 0.7965398591504004,
    "step": 24770
  },
  {
    "loss": 0.07550753355026245,
    "grad_norm": 0.4523955285549164,
    "learning_rate": 1.9810662956114246e-05,
    "epoch": 0.7968614335788018,
    "step": 24780
  },
  {
    "loss": 0.07917873859405518,
    "grad_norm": 0.42558956146240234,
    "learning_rate": 1.975014995406993e-05,
    "epoch": 0.7971830080072033,
    "step": 24790
  },
  {
    "loss": 0.06907861828804016,
    "grad_norm": 0.5384016633033752,
    "learning_rate": 1.968971938569101e-05,
    "epoch": 0.7975045824356047,
    "step": 24800
  },
  {
    "loss": 0.0709206521511078,
    "grad_norm": 0.7265986800193787,
    "learning_rate": 1.962937131305258e-05,
    "epoch": 0.7978261568640062,
    "step": 24810
  },
  {
    "loss": 0.06628232002258301,
    "grad_norm": 0.4229384660720825,
    "learning_rate": 1.9569105798144925e-05,
    "epoch": 0.7981477312924077,
    "step": 24820
  },
  {
    "loss": 0.05692559480667114,
    "grad_norm": 0.4667295515537262,
    "learning_rate": 1.9508922902873505e-05,
    "epoch": 0.7984693057208091,
    "step": 24830
  },
  {
    "loss": 0.06895289421081544,
    "grad_norm": 0.5784633755683899,
    "learning_rate": 1.9448822689059008e-05,
    "epoch": 0.7987908801492105,
    "step": 24840
  },
  {
    "loss": 0.07624900341033936,
    "grad_norm": 0.3647468090057373,
    "learning_rate": 1.9388805218437102e-05,
    "epoch": 0.799112454577612,
    "step": 24850
  },
  {
    "loss": 0.06718692779541016,
    "grad_norm": 0.5916265845298767,
    "learning_rate": 1.9328870552658497e-05,
    "epoch": 0.7994340290060135,
    "step": 24860
  },
  {
    "eval_loss": 0.06779318302869797,
    "eval_runtime": 34.5798,
    "eval_samples_per_second": 145.345,
    "eval_steps_per_second": 36.351,
    "epoch": 0.799562658777374,
    "step": 24864
  },
  {
    "loss": 0.054065996408462526,
    "grad_norm": 0.465282142162323,
    "learning_rate": 1.926901875328889e-05,
    "epoch": 0.7997556034344149,
    "step": 24870
  },
  {
    "loss": 0.09027044773101807,
    "grad_norm": 0.6328346133232117,
    "learning_rate": 1.9209249881808745e-05,
    "epoch": 0.8000771778628164,
    "step": 24880
  },
  {
    "loss": 0.06480810046195984,
    "grad_norm": 0.5475115180015564,
    "learning_rate": 1.9149563999613464e-05,
    "epoch": 0.8003987522912178,
    "step": 24890
  },
  {
    "loss": 0.09089791178703308,
    "grad_norm": 0.676400899887085,
    "learning_rate": 1.9089961168013093e-05,
    "epoch": 0.8007203267196192,
    "step": 24900
  },
  {
    "loss": 0.08152996897697448,
    "grad_norm": 0.7551012635231018,
    "learning_rate": 1.903044144823246e-05,
    "epoch": 0.8010419011480208,
    "step": 24910
  },
  {
    "loss": 0.07394970655441284,
    "grad_norm": 0.49486297369003296,
    "learning_rate": 1.8971004901410937e-05,
    "epoch": 0.8013634755764222,
    "step": 24920
  },
  {
    "loss": 0.07575193643569947,
    "grad_norm": 0.1976037621498108,
    "learning_rate": 1.891165158860252e-05,
    "epoch": 0.8016850500048236,
    "step": 24930
  },
  {
    "loss": 0.05858004093170166,
    "grad_norm": 0.4293280243873596,
    "learning_rate": 1.8852381570775714e-05,
    "epoch": 0.8020066244332251,
    "step": 24940
  },
  {
    "loss": 0.07835475206375123,
    "grad_norm": 0.40252768993377686,
    "learning_rate": 1.879319490881337e-05,
    "epoch": 0.8023281988616265,
    "step": 24950
  },
  {
    "loss": 0.05356748700141907,
    "grad_norm": 0.821725606918335,
    "learning_rate": 1.8734091663512854e-05,
    "epoch": 0.802649773290028,
    "step": 24960
  },
  {
    "loss": 0.07878881692886353,
    "grad_norm": 0.8050437569618225,
    "learning_rate": 1.8675071895585707e-05,
    "epoch": 0.8029713477184295,
    "step": 24970
  },
  {
    "loss": 0.0742866039276123,
    "grad_norm": 0.5441704988479614,
    "learning_rate": 1.861613566565783e-05,
    "epoch": 0.8032929221468309,
    "step": 24980
  },
  {
    "loss": 0.0770497441291809,
    "grad_norm": 0.48621344566345215,
    "learning_rate": 1.8557283034269225e-05,
    "epoch": 0.8036144965752323,
    "step": 24990
  },
  {
    "loss": 0.05645939111709595,
    "grad_norm": 0.5151429176330566,
    "learning_rate": 1.8498514061874084e-05,
    "epoch": 0.8039360710036338,
    "step": 25000
  },
  {
    "loss": 0.07547401785850524,
    "grad_norm": 0.6428143978118896,
    "learning_rate": 1.8439828808840664e-05,
    "epoch": 0.8042576454320353,
    "step": 25010
  },
  {
    "loss": 0.08763245344161988,
    "grad_norm": 0.5403868556022644,
    "learning_rate": 1.8381227335451166e-05,
    "epoch": 0.8045792198604367,
    "step": 25020
  },
  {
    "loss": 0.059272265434265135,
    "grad_norm": 0.26042598485946655,
    "learning_rate": 1.8322709701901798e-05,
    "epoch": 0.8049007942888382,
    "step": 25030
  },
  {
    "loss": 0.07984400391578675,
    "grad_norm": 0.47204700112342834,
    "learning_rate": 1.8264275968302602e-05,
    "epoch": 0.8052223687172396,
    "step": 25040
  },
  {
    "loss": 0.049272620677948,
    "grad_norm": 0.38910478353500366,
    "learning_rate": 1.8205926194677403e-05,
    "epoch": 0.805543943145641,
    "step": 25050
  },
  {
    "loss": 0.07370315790176392,
    "grad_norm": 0.5536523461341858,
    "learning_rate": 1.8147660440963932e-05,
    "epoch": 0.8058655175740426,
    "step": 25060
  },
  {
    "loss": 0.07952592968940735,
    "grad_norm": 0.7388674020767212,
    "learning_rate": 1.8089478767013423e-05,
    "epoch": 0.806187092002444,
    "step": 25070
  },
  {
    "loss": 0.058814072608947755,
    "grad_norm": 0.5940684080123901,
    "learning_rate": 1.8031381232590893e-05,
    "epoch": 0.8065086664308454,
    "step": 25080
  },
  {
    "loss": 0.06479309201240539,
    "grad_norm": 0.5064087510108948,
    "learning_rate": 1.797336789737484e-05,
    "epoch": 0.8068302408592469,
    "step": 25090
  },
  {
    "loss": 0.06768692135810853,
    "grad_norm": 0.42703741788864136,
    "learning_rate": 1.7915438820957277e-05,
    "epoch": 0.8071518152876483,
    "step": 25100
  },
  {
    "loss": 0.0990111231803894,
    "grad_norm": 0.7536817789077759,
    "learning_rate": 1.7857594062843695e-05,
    "epoch": 0.8074733897160498,
    "step": 25110
  },
  {
    "loss": 0.07579346895217895,
    "grad_norm": 0.8095626831054688,
    "learning_rate": 1.7799833682452994e-05,
    "epoch": 0.8077949641444513,
    "step": 25120
  },
  {
    "loss": 0.07740336656570435,
    "grad_norm": 0.5823168158531189,
    "learning_rate": 1.7742157739117372e-05,
    "epoch": 0.8081165385728527,
    "step": 25130
  },
  {
    "loss": 0.07928024530410767,
    "grad_norm": 0.6158061623573303,
    "learning_rate": 1.7684566292082273e-05,
    "epoch": 0.8084381130012541,
    "step": 25140
  },
  {
    "loss": 0.05865381360054016,
    "grad_norm": 0.25727465748786926,
    "learning_rate": 1.762705940050633e-05,
    "epoch": 0.8087596874296556,
    "step": 25150
  },
  {
    "loss": 0.056818431615829466,
    "grad_norm": 0.87153559923172,
    "learning_rate": 1.7569637123461415e-05,
    "epoch": 0.8090812618580571,
    "step": 25160
  },
  {
    "loss": 0.06351528763771057,
    "grad_norm": 0.3815484046936035,
    "learning_rate": 1.7512299519932363e-05,
    "epoch": 0.8094028362864585,
    "step": 25170
  },
  {
    "loss": 0.07543556094169616,
    "grad_norm": 0.42525655031204224,
    "learning_rate": 1.7455046648817118e-05,
    "epoch": 0.8097244107148599,
    "step": 25180
  },
  {
    "loss": 0.07463630437850952,
    "grad_norm": 0.7410084009170532,
    "learning_rate": 1.7397878568926584e-05,
    "epoch": 0.8100459851432614,
    "step": 25190
  },
  {
    "loss": 0.056423962116241455,
    "grad_norm": 0.8661516308784485,
    "learning_rate": 1.7340795338984484e-05,
    "epoch": 0.8103675595716628,
    "step": 25200
  },
  {
    "loss": 0.06369969248771667,
    "grad_norm": 0.6466599702835083,
    "learning_rate": 1.7283797017627478e-05,
    "epoch": 0.8106891340000643,
    "step": 25210
  },
  {
    "loss": 0.050625795125961305,
    "grad_norm": 0.3340155780315399,
    "learning_rate": 1.7226883663404935e-05,
    "epoch": 0.8110107084284658,
    "step": 25220
  },
  {
    "loss": 0.06831510663032532,
    "grad_norm": 0.3890591263771057,
    "learning_rate": 1.7170055334779012e-05,
    "epoch": 0.8113322828568672,
    "step": 25230
  },
  {
    "loss": 0.07306399941444397,
    "grad_norm": 0.5340134501457214,
    "learning_rate": 1.7113312090124445e-05,
    "epoch": 0.8116538572852686,
    "step": 25240
  },
  {
    "loss": 0.06650834083557129,
    "grad_norm": 0.37586766481399536,
    "learning_rate": 1.7056653987728644e-05,
    "epoch": 0.8119754317136701,
    "step": 25250
  },
  {
    "loss": 0.06394178867340088,
    "grad_norm": 0.5065869092941284,
    "learning_rate": 1.700008108579154e-05,
    "epoch": 0.8122970061420716,
    "step": 25260
  },
  {
    "loss": 0.07049349546432496,
    "grad_norm": 0.9211828708648682,
    "learning_rate": 1.694359344242551e-05,
    "epoch": 0.812618580570473,
    "step": 25270
  },
  {
    "loss": 0.07769249081611633,
    "grad_norm": 0.5380610227584839,
    "learning_rate": 1.6887191115655422e-05,
    "epoch": 0.8129401549988745,
    "step": 25280
  },
  {
    "loss": 0.057320517301559445,
    "grad_norm": 0.641497015953064,
    "learning_rate": 1.68308741634184e-05,
    "epoch": 0.8132617294272759,
    "step": 25290
  },
  {
    "loss": 0.08097407221794128,
    "grad_norm": 0.4522627294063568,
    "learning_rate": 1.6774642643563953e-05,
    "epoch": 0.8135833038556773,
    "step": 25300
  },
  {
    "loss": 0.07374048829078675,
    "grad_norm": 0.6357822418212891,
    "learning_rate": 1.6718496613853872e-05,
    "epoch": 0.8139048782840789,
    "step": 25310
  },
  {
    "loss": 0.06144424080848694,
    "grad_norm": 0.3151220977306366,
    "learning_rate": 1.666243613196199e-05,
    "epoch": 0.8142264527124803,
    "step": 25320
  },
  {
    "loss": 0.08905712366104127,
    "grad_norm": 0.6542190909385681,
    "learning_rate": 1.6606461255474404e-05,
    "epoch": 0.8145480271408817,
    "step": 25330
  },
  {
    "loss": 0.07413828372955322,
    "grad_norm": 0.5541084408760071,
    "learning_rate": 1.6550572041889178e-05,
    "epoch": 0.8148696015692832,
    "step": 25340
  },
  {
    "loss": 0.07905344367027282,
    "grad_norm": 0.29783719778060913,
    "learning_rate": 1.6494768548616456e-05,
    "epoch": 0.8151911759976846,
    "step": 25350
  },
  {
    "loss": 0.08787615299224853,
    "grad_norm": 0.808488130569458,
    "learning_rate": 1.6439050832978285e-05,
    "epoch": 0.8155127504260861,
    "step": 25360
  },
  {
    "loss": 0.06256959438323975,
    "grad_norm": 0.5121647119522095,
    "learning_rate": 1.6383418952208553e-05,
    "epoch": 0.8158343248544876,
    "step": 25370
  },
  {
    "loss": 0.056725984811782836,
    "grad_norm": 0.7378894090652466,
    "learning_rate": 1.632787296345314e-05,
    "epoch": 0.816155899282889,
    "step": 25380
  },
  {
    "loss": 0.06233987808227539,
    "grad_norm": 0.7205318808555603,
    "learning_rate": 1.6272412923769507e-05,
    "epoch": 0.8164774737112904,
    "step": 25390
  },
  {
    "loss": 0.06884159445762635,
    "grad_norm": 1.139447808265686,
    "learning_rate": 1.6217038890126966e-05,
    "epoch": 0.816799048139692,
    "step": 25400
  },
  {
    "loss": 0.07286211252212524,
    "grad_norm": 0.6898419260978699,
    "learning_rate": 1.6161750919406405e-05,
    "epoch": 0.8171206225680934,
    "step": 25410
  },
  {
    "loss": 0.07253461480140685,
    "grad_norm": 0.6165859699249268,
    "learning_rate": 1.6106549068400312e-05,
    "epoch": 0.8174421969964948,
    "step": 25420
  },
  {
    "loss": 0.06999062299728394,
    "grad_norm": 0.685233473777771,
    "learning_rate": 1.6051433393812733e-05,
    "epoch": 0.8177637714248963,
    "step": 25430
  },
  {
    "loss": 0.06932086944580078,
    "grad_norm": 0.765102744102478,
    "learning_rate": 1.599640395225921e-05,
    "epoch": 0.8180853458532977,
    "step": 25440
  },
  {
    "loss": 0.0570664644241333,
    "grad_norm": 0.4864404797554016,
    "learning_rate": 1.5941460800266693e-05,
    "epoch": 0.8184069202816991,
    "step": 25450
  },
  {
    "loss": 0.08184292912483215,
    "grad_norm": 0.5153435468673706,
    "learning_rate": 1.5886603994273473e-05,
    "epoch": 0.8187284947101007,
    "step": 25460
  },
  {
    "loss": 0.0793727159500122,
    "grad_norm": 0.7540408372879028,
    "learning_rate": 1.5831833590629132e-05,
    "epoch": 0.8190500691385021,
    "step": 25470
  },
  {
    "loss": 0.05399051308631897,
    "grad_norm": 0.45676225423812866,
    "learning_rate": 1.577714964559457e-05,
    "epoch": 0.8193716435669035,
    "step": 25480
  },
  {
    "loss": 0.06276117563247681,
    "grad_norm": 0.5364438891410828,
    "learning_rate": 1.572255221534178e-05,
    "epoch": 0.819693217995305,
    "step": 25490
  },
  {
    "loss": 0.07922245264053344,
    "grad_norm": 1.0108367204666138,
    "learning_rate": 1.5668041355953956e-05,
    "epoch": 0.8200147924237065,
    "step": 25500
  },
  {
    "loss": 0.0870794415473938,
    "grad_norm": 0.7691150903701782,
    "learning_rate": 1.5613617123425384e-05,
    "epoch": 0.8203363668521079,
    "step": 25510
  },
  {
    "loss": 0.09856227636337281,
    "grad_norm": 0.706281840801239,
    "learning_rate": 1.5559279573661267e-05,
    "epoch": 0.8206579412805094,
    "step": 25520
  },
  {
    "loss": 0.09251796603202819,
    "grad_norm": 0.7108443975448608,
    "learning_rate": 1.550502876247788e-05,
    "epoch": 0.8209795157089108,
    "step": 25530
  },
  {
    "loss": 0.07778313755989075,
    "grad_norm": 0.8512086272239685,
    "learning_rate": 1.54508647456023e-05,
    "epoch": 0.8213010901373122,
    "step": 25540
  },
  {
    "loss": 0.08167454004287719,
    "grad_norm": 0.6187142133712769,
    "learning_rate": 1.5396787578672532e-05,
    "epoch": 0.8216226645657138,
    "step": 25550
  },
  {
    "loss": 0.06370655298233033,
    "grad_norm": 0.27371832728385925,
    "learning_rate": 1.5342797317237288e-05,
    "epoch": 0.8219442389941152,
    "step": 25560
  },
  {
    "loss": 0.07302612662315369,
    "grad_norm": 0.5439883470535278,
    "learning_rate": 1.5288894016756072e-05,
    "epoch": 0.8222658134225166,
    "step": 25570
  },
  {
    "loss": 0.06198353767395019,
    "grad_norm": 0.9168033599853516,
    "learning_rate": 1.5235077732599057e-05,
    "epoch": 0.8225873878509181,
    "step": 25580
  },
  {
    "loss": 0.08354194164276123,
    "grad_norm": 0.28686732053756714,
    "learning_rate": 1.5181348520046979e-05,
    "epoch": 0.8229089622793195,
    "step": 25590
  },
  {
    "loss": 0.0665959358215332,
    "grad_norm": 0.4895515441894531,
    "learning_rate": 1.5127706434291201e-05,
    "epoch": 0.823230536707721,
    "step": 25600
  },
  {
    "loss": 0.06269041895866394,
    "grad_norm": 0.15579339861869812,
    "learning_rate": 1.5074151530433512e-05,
    "epoch": 0.8235521111361225,
    "step": 25610
  },
  {
    "loss": 0.08237427473068237,
    "grad_norm": 0.6324445009231567,
    "learning_rate": 1.5020683863486219e-05,
    "epoch": 0.8238736855645239,
    "step": 25620
  },
  {
    "loss": 0.0931522786617279,
    "grad_norm": 0.5454768538475037,
    "learning_rate": 1.496730348837202e-05,
    "epoch": 0.8241952599929253,
    "step": 25630
  },
  {
    "loss": 0.06490244269371033,
    "grad_norm": 0.5473190546035767,
    "learning_rate": 1.4914010459923844e-05,
    "epoch": 0.8245168344213268,
    "step": 25640
  },
  {
    "loss": 0.07663293480873108,
    "grad_norm": 0.9343203902244568,
    "learning_rate": 1.4860804832885023e-05,
    "epoch": 0.8248384088497283,
    "step": 25650
  },
  {
    "loss": 0.07581743001937866,
    "grad_norm": 0.8527815937995911,
    "learning_rate": 1.4807686661909015e-05,
    "epoch": 0.8251599832781297,
    "step": 25660
  },
  {
    "loss": 0.0709666907787323,
    "grad_norm": 0.7864364981651306,
    "learning_rate": 1.475465600155953e-05,
    "epoch": 0.8254815577065312,
    "step": 25670
  },
  {
    "loss": 0.0744117796421051,
    "grad_norm": 0.6601343154907227,
    "learning_rate": 1.4701712906310295e-05,
    "epoch": 0.8258031321349326,
    "step": 25680
  },
  {
    "loss": 0.06409841179847717,
    "grad_norm": 0.33034786581993103,
    "learning_rate": 1.4648857430545092e-05,
    "epoch": 0.826124706563334,
    "step": 25690
  },
  {
    "loss": 0.0730972170829773,
    "grad_norm": 0.4441629648208618,
    "learning_rate": 1.459608962855783e-05,
    "epoch": 0.8264462809917356,
    "step": 25700
  },
  {
    "loss": 0.07730489373207092,
    "grad_norm": 0.6629310846328735,
    "learning_rate": 1.4543409554552211e-05,
    "epoch": 0.826767855420137,
    "step": 25710
  },
  {
    "loss": 0.08678494691848755,
    "grad_norm": 0.5934807062149048,
    "learning_rate": 1.4490817262641876e-05,
    "epoch": 0.8270894298485384,
    "step": 25720
  },
  {
    "loss": 0.07854796648025512,
    "grad_norm": 0.3622133731842041,
    "learning_rate": 1.4438312806850319e-05,
    "epoch": 0.8274110042769399,
    "step": 25730
  },
  {
    "loss": 0.06401524543762208,
    "grad_norm": 0.5912713408470154,
    "learning_rate": 1.4385896241110752e-05,
    "epoch": 0.8277325787053413,
    "step": 25740
  },
  {
    "loss": 0.07179030179977416,
    "grad_norm": 0.6012607216835022,
    "learning_rate": 1.4333567619266153e-05,
    "epoch": 0.8280541531337428,
    "step": 25750
  },
  {
    "loss": 0.07031106948852539,
    "grad_norm": 1.0617568492889404,
    "learning_rate": 1.4281326995069166e-05,
    "epoch": 0.8283757275621443,
    "step": 25760
  },
  {
    "loss": 0.05340087413787842,
    "grad_norm": 0.5312886834144592,
    "learning_rate": 1.422917442218198e-05,
    "epoch": 0.8286973019905457,
    "step": 25770
  },
  {
    "loss": 0.07728091478347779,
    "grad_norm": 1.6353224515914917,
    "learning_rate": 1.4177109954176437e-05,
    "epoch": 0.8290188764189471,
    "step": 25780
  },
  {
    "loss": 0.061128294467926024,
    "grad_norm": 0.2983153760433197,
    "learning_rate": 1.4125133644533761e-05,
    "epoch": 0.8293404508473486,
    "step": 25790
  },
  {
    "loss": 0.07462538480758667,
    "grad_norm": 0.6013174653053284,
    "learning_rate": 1.4073245546644731e-05,
    "epoch": 0.8296620252757501,
    "step": 25800
  },
  {
    "loss": 0.05741596817970276,
    "grad_norm": 0.28202420473098755,
    "learning_rate": 1.4021445713809422e-05,
    "epoch": 0.8299835997041515,
    "step": 25810
  },
  {
    "loss": 0.07711042165756225,
    "grad_norm": 0.7175576090812683,
    "learning_rate": 1.3969734199237283e-05,
    "epoch": 0.830305174132553,
    "step": 25820
  },
  {
    "loss": 0.06802661418914795,
    "grad_norm": 0.5284676551818848,
    "learning_rate": 1.3918111056047067e-05,
    "epoch": 0.8306267485609544,
    "step": 25830
  },
  {
    "loss": 0.07056057453155518,
    "grad_norm": 0.570586621761322,
    "learning_rate": 1.386657633726669e-05,
    "epoch": 0.8309483229893558,
    "step": 25840
  },
  {
    "loss": 0.07420239448547364,
    "grad_norm": 1.6333948373794556,
    "learning_rate": 1.3815130095833306e-05,
    "epoch": 0.8312698974177574,
    "step": 25850
  },
  {
    "loss": 0.05796622633934021,
    "grad_norm": 0.6089661717414856,
    "learning_rate": 1.376377238459311e-05,
    "epoch": 0.8315914718461588,
    "step": 25860
  },
  {
    "loss": 0.07319130897521972,
    "grad_norm": 0.568272590637207,
    "learning_rate": 1.3712503256301424e-05,
    "epoch": 0.8319130462745602,
    "step": 25870
  },
  {
    "loss": 0.06356196999549865,
    "grad_norm": 0.36745578050613403,
    "learning_rate": 1.3661322763622519e-05,
    "epoch": 0.8322346207029617,
    "step": 25880
  },
  {
    "loss": 0.07408368587493896,
    "grad_norm": 1.0645989179611206,
    "learning_rate": 1.3610230959129654e-05,
    "epoch": 0.8325561951313631,
    "step": 25890
  },
  {
    "loss": 0.06604211330413819,
    "grad_norm": 0.7959291338920593,
    "learning_rate": 1.3559227895305015e-05,
    "epoch": 0.8328777695597646,
    "step": 25900
  },
  {
    "loss": 0.07473559975624085,
    "grad_norm": 0.46967774629592896,
    "learning_rate": 1.3508313624539558e-05,
    "epoch": 0.8331993439881661,
    "step": 25910
  },
  {
    "loss": 0.0604839563369751,
    "grad_norm": 0.6299682855606079,
    "learning_rate": 1.3457488199133096e-05,
    "epoch": 0.8335209184165675,
    "step": 25920
  },
  {
    "loss": 0.07758623361587524,
    "grad_norm": 0.60549396276474,
    "learning_rate": 1.3406751671294149e-05,
    "epoch": 0.8338424928449689,
    "step": 25930
  },
  {
    "loss": 0.07174047231674194,
    "grad_norm": 0.376686155796051,
    "learning_rate": 1.3356104093139855e-05,
    "epoch": 0.8341640672733704,
    "step": 25940
  },
  {
    "loss": 0.05498989224433899,
    "grad_norm": 0.33224010467529297,
    "learning_rate": 1.330554551669615e-05,
    "epoch": 0.8344856417017719,
    "step": 25950
  },
  {
    "loss": 0.08045591115951538,
    "grad_norm": 0.4750721752643585,
    "learning_rate": 1.3255075993897392e-05,
    "epoch": 0.8348072161301733,
    "step": 25960
  },
  {
    "loss": 0.039634838700294495,
    "grad_norm": 0.40264174342155457,
    "learning_rate": 1.3204695576586534e-05,
    "epoch": 0.8351287905585748,
    "step": 25970
  },
  {
    "loss": 0.08607410788536071,
    "grad_norm": 0.9619850516319275,
    "learning_rate": 1.3154404316514979e-05,
    "epoch": 0.8354503649869762,
    "step": 25980
  },
  {
    "loss": 0.08280614018440247,
    "grad_norm": 0.46870899200439453,
    "learning_rate": 1.3104202265342525e-05,
    "epoch": 0.8357719394153776,
    "step": 25990
  },
  {
    "loss": 0.05811293721199036,
    "grad_norm": 0.901194155216217,
    "learning_rate": 1.3054089474637366e-05,
    "epoch": 0.8360935138437792,
    "step": 26000
  },
  {
    "loss": 0.0757293939590454,
    "grad_norm": 0.8040971159934998,
    "learning_rate": 1.3004065995876013e-05,
    "epoch": 0.8364150882721806,
    "step": 26010
  },
  {
    "loss": 0.08551060557365417,
    "grad_norm": 0.6271659731864929,
    "learning_rate": 1.2954131880443222e-05,
    "epoch": 0.836736662700582,
    "step": 26020
  },
  {
    "loss": 0.07875540852546692,
    "grad_norm": 0.8364705443382263,
    "learning_rate": 1.2904287179631946e-05,
    "epoch": 0.8370582371289835,
    "step": 26030
  },
  {
    "loss": 0.08773240447044373,
    "grad_norm": 0.49071305990219116,
    "learning_rate": 1.2854531944643266e-05,
    "epoch": 0.837379811557385,
    "step": 26040
  },
  {
    "loss": 0.0672372043132782,
    "grad_norm": 0.6522132754325867,
    "learning_rate": 1.2804866226586443e-05,
    "epoch": 0.8377013859857864,
    "step": 26050
  },
  {
    "loss": 0.07467586994171142,
    "grad_norm": 0.6706381440162659,
    "learning_rate": 1.2755290076478677e-05,
    "epoch": 0.8380229604141879,
    "step": 26060
  },
  {
    "loss": 0.08341933488845825,
    "grad_norm": 0.5691989660263062,
    "learning_rate": 1.2705803545245243e-05,
    "epoch": 0.8383445348425893,
    "step": 26070
  },
  {
    "loss": 0.06978496313095092,
    "grad_norm": 0.717632532119751,
    "learning_rate": 1.2656406683719357e-05,
    "epoch": 0.8386661092709907,
    "step": 26080
  },
  {
    "loss": 0.08308391571044922,
    "grad_norm": 0.7718671560287476,
    "learning_rate": 1.2607099542642075e-05,
    "epoch": 0.8389876836993923,
    "step": 26090
  },
  {
    "loss": 0.07776856422424316,
    "grad_norm": 0.8327365517616272,
    "learning_rate": 1.2557882172662338e-05,
    "epoch": 0.8393092581277937,
    "step": 26100
  },
  {
    "loss": 0.06721643805503845,
    "grad_norm": 0.6579548120498657,
    "learning_rate": 1.2508754624336827e-05,
    "epoch": 0.8396308325561951,
    "step": 26110
  },
  {
    "loss": 0.05926239490509033,
    "grad_norm": 0.40918952226638794,
    "learning_rate": 1.2459716948130018e-05,
    "epoch": 0.8399524069845966,
    "step": 26120
  },
  {
    "loss": 0.08606613874435425,
    "grad_norm": 0.7024818062782288,
    "learning_rate": 1.2410769194413995e-05,
    "epoch": 0.840273981412998,
    "step": 26130
  },
  {
    "loss": 0.07684431672096252,
    "grad_norm": 0.7775634527206421,
    "learning_rate": 1.2361911413468541e-05,
    "epoch": 0.8405955558413994,
    "step": 26140
  },
  {
    "loss": 0.06418314576148987,
    "grad_norm": 0.8045727014541626,
    "learning_rate": 1.2313143655480997e-05,
    "epoch": 0.840917130269801,
    "step": 26150
  },
  {
    "loss": 0.08686638474464417,
    "grad_norm": 0.8366597294807434,
    "learning_rate": 1.226446597054619e-05,
    "epoch": 0.8412387046982024,
    "step": 26160
  },
  {
    "loss": 0.06793270111083985,
    "grad_norm": 0.80096435546875,
    "learning_rate": 1.2215878408666493e-05,
    "epoch": 0.8415602791266038,
    "step": 26170
  },
  {
    "loss": 0.0783788800239563,
    "grad_norm": 0.8146688342094421,
    "learning_rate": 1.2167381019751634e-05,
    "epoch": 0.8418818535550053,
    "step": 26180
  },
  {
    "loss": 0.08650415539741516,
    "grad_norm": 0.7247388958930969,
    "learning_rate": 1.2118973853618765e-05,
    "epoch": 0.8422034279834067,
    "step": 26190
  },
  {
    "loss": 0.06807527542114258,
    "grad_norm": 0.5796247124671936,
    "learning_rate": 1.2070656959992354e-05,
    "epoch": 0.8425250024118082,
    "step": 26200
  },
  {
    "loss": 0.06200519800186157,
    "grad_norm": 0.36104416847229004,
    "learning_rate": 1.2022430388504091e-05,
    "epoch": 0.8428465768402097,
    "step": 26210
  },
  {
    "loss": 0.04987512826919556,
    "grad_norm": 0.5600906014442444,
    "learning_rate": 1.1974294188692959e-05,
    "epoch": 0.8431681512686111,
    "step": 26220
  },
  {
    "loss": 0.061176568269729614,
    "grad_norm": 0.41778096556663513,
    "learning_rate": 1.1926248410005047e-05,
    "epoch": 0.8434897256970125,
    "step": 26230
  },
  {
    "loss": 0.06344327926635743,
    "grad_norm": 0.2527564764022827,
    "learning_rate": 1.1878293101793625e-05,
    "epoch": 0.843811300125414,
    "step": 26240
  },
  {
    "loss": 0.06565487980842591,
    "grad_norm": 0.4772278368473053,
    "learning_rate": 1.1830428313318976e-05,
    "epoch": 0.8441328745538155,
    "step": 26250
  },
  {
    "loss": 0.0762079894542694,
    "grad_norm": 0.583329439163208,
    "learning_rate": 1.1782654093748357e-05,
    "epoch": 0.8444544489822169,
    "step": 26260
  },
  {
    "loss": 0.08452026844024658,
    "grad_norm": 0.4808464050292969,
    "learning_rate": 1.1734970492156171e-05,
    "epoch": 0.8447760234106184,
    "step": 26270
  },
  {
    "loss": 0.06882460117340088,
    "grad_norm": 0.47750189900398254,
    "learning_rate": 1.1687377557523526e-05,
    "epoch": 0.8450975978390198,
    "step": 26280
  },
  {
    "loss": 0.05743715763092041,
    "grad_norm": 0.6582987308502197,
    "learning_rate": 1.1639875338738549e-05,
    "epoch": 0.8454191722674212,
    "step": 26290
  },
  {
    "loss": 0.06682893633842468,
    "grad_norm": 0.45502129197120667,
    "learning_rate": 1.1592463884596094e-05,
    "epoch": 0.8457407466958228,
    "step": 26300
  },
  {
    "loss": 0.07056695818901063,
    "grad_norm": 0.6482895016670227,
    "learning_rate": 1.1545143243797785e-05,
    "epoch": 0.8460623211242242,
    "step": 26310
  },
  {
    "loss": 0.06616935729980469,
    "grad_norm": 0.6247773766517639,
    "learning_rate": 1.1497913464952015e-05,
    "epoch": 0.8463838955526256,
    "step": 26320
  },
  {
    "loss": 0.09493278861045837,
    "grad_norm": 0.5621542930603027,
    "learning_rate": 1.145077459657381e-05,
    "epoch": 0.8467054699810271,
    "step": 26330
  },
  {
    "loss": 0.06745593547821045,
    "grad_norm": 0.7123182415962219,
    "learning_rate": 1.1403726687084838e-05,
    "epoch": 0.8470270444094286,
    "step": 26340
  },
  {
    "loss": 0.05028054118156433,
    "grad_norm": 0.6581669449806213,
    "learning_rate": 1.1356769784813271e-05,
    "epoch": 0.84734861883783,
    "step": 26350
  },
  {
    "loss": 0.06702528595924377,
    "grad_norm": 0.8366682529449463,
    "learning_rate": 1.1309903937993838e-05,
    "epoch": 0.8476701932662315,
    "step": 26360
  },
  {
    "loss": 0.06623958349227906,
    "grad_norm": 0.8056640028953552,
    "learning_rate": 1.1263129194767764e-05,
    "epoch": 0.8479917676946329,
    "step": 26370
  },
  {
    "loss": 0.08432796597480774,
    "grad_norm": 1.285785436630249,
    "learning_rate": 1.1216445603182612e-05,
    "epoch": 0.8483133421230343,
    "step": 26380
  },
  {
    "loss": 0.058904987573623654,
    "grad_norm": 0.43198439478874207,
    "learning_rate": 1.1169853211192394e-05,
    "epoch": 0.8486349165514359,
    "step": 26390
  },
  {
    "loss": 0.07768229246139527,
    "grad_norm": 1.1017895936965942,
    "learning_rate": 1.1123352066657422e-05,
    "epoch": 0.8489564909798373,
    "step": 26400
  },
  {
    "loss": 0.04701220989227295,
    "grad_norm": 0.5291910171508789,
    "learning_rate": 1.1076942217344221e-05,
    "epoch": 0.8492780654082387,
    "step": 26410
  },
  {
    "eval_loss": 0.06639111787080765,
    "eval_runtime": 34.6001,
    "eval_samples_per_second": 145.26,
    "eval_steps_per_second": 36.329,
    "epoch": 0.8495353249509598,
    "step": 26418
  },
  {
    "loss": 0.06604416966438294,
    "grad_norm": 0.5437458157539368,
    "learning_rate": 1.1030623710925603e-05,
    "epoch": 0.8495996398366402,
    "step": 26420
  },
  {
    "loss": 0.05475022792816162,
    "grad_norm": 0.7437111139297485,
    "learning_rate": 1.0984396594980516e-05,
    "epoch": 0.8499212142650416,
    "step": 26430
  },
  {
    "loss": 0.055143731832504275,
    "grad_norm": 0.7033617496490479,
    "learning_rate": 1.093826091699406e-05,
    "epoch": 0.850242788693443,
    "step": 26440
  },
  {
    "loss": 0.05738850831985474,
    "grad_norm": 1.1026557683944702,
    "learning_rate": 1.0892216724357351e-05,
    "epoch": 0.8505643631218446,
    "step": 26450
  },
  {
    "loss": 0.08413679003715516,
    "grad_norm": 0.5151999592781067,
    "learning_rate": 1.0846264064367595e-05,
    "epoch": 0.850885937550246,
    "step": 26460
  },
  {
    "loss": 0.0639430046081543,
    "grad_norm": 0.6788651943206787,
    "learning_rate": 1.0800402984227964e-05,
    "epoch": 0.8512075119786474,
    "step": 26470
  },
  {
    "loss": 0.07391080856323243,
    "grad_norm": 0.619463324546814,
    "learning_rate": 1.0754633531047508e-05,
    "epoch": 0.8515290864070489,
    "step": 26480
  },
  {
    "loss": 0.051558637619018556,
    "grad_norm": 0.6007417440414429,
    "learning_rate": 1.070895575184122e-05,
    "epoch": 0.8518506608354504,
    "step": 26490
  },
  {
    "loss": 0.0619118869304657,
    "grad_norm": 0.610027015209198,
    "learning_rate": 1.0663369693529868e-05,
    "epoch": 0.8521722352638518,
    "step": 26500
  },
  {
    "loss": 0.05492755174636841,
    "grad_norm": 0.4237189292907715,
    "learning_rate": 1.0617875402940036e-05,
    "epoch": 0.8524938096922533,
    "step": 26510
  },
  {
    "loss": 0.07027317881584168,
    "grad_norm": 0.36540210247039795,
    "learning_rate": 1.0572472926804067e-05,
    "epoch": 0.8528153841206547,
    "step": 26520
  },
  {
    "loss": 0.07017855644226074,
    "grad_norm": 0.4847584068775177,
    "learning_rate": 1.052716231175992e-05,
    "epoch": 0.8531369585490561,
    "step": 26530
  },
  {
    "loss": 0.08128488063812256,
    "grad_norm": 0.8945302963256836,
    "learning_rate": 1.0481943604351274e-05,
    "epoch": 0.8534585329774577,
    "step": 26540
  },
  {
    "loss": 0.08098180294036865,
    "grad_norm": 0.341545045375824,
    "learning_rate": 1.043681685102732e-05,
    "epoch": 0.8537801074058591,
    "step": 26550
  },
  {
    "loss": 0.08044508099555969,
    "grad_norm": 0.3028389811515808,
    "learning_rate": 1.0391782098142867e-05,
    "epoch": 0.8541016818342605,
    "step": 26560
  },
  {
    "loss": 0.06877010464668273,
    "grad_norm": 0.481671541929245,
    "learning_rate": 1.0346839391958175e-05,
    "epoch": 0.854423256262662,
    "step": 26570
  },
  {
    "loss": 0.07255155444145203,
    "grad_norm": 0.5929657220840454,
    "learning_rate": 1.0301988778638916e-05,
    "epoch": 0.8547448306910634,
    "step": 26580
  },
  {
    "loss": 0.07767698764801026,
    "grad_norm": 0.6092710494995117,
    "learning_rate": 1.0257230304256282e-05,
    "epoch": 0.8550664051194649,
    "step": 26590
  },
  {
    "loss": 0.08522970676422119,
    "grad_norm": 0.6273548603057861,
    "learning_rate": 1.021256401478673e-05,
    "epoch": 0.8553879795478664,
    "step": 26600
  },
  {
    "loss": 0.07369956970214844,
    "grad_norm": 0.33282068371772766,
    "learning_rate": 1.016798995611199e-05,
    "epoch": 0.8557095539762678,
    "step": 26610
  },
  {
    "loss": 0.08735382556915283,
    "grad_norm": 0.6816231608390808,
    "learning_rate": 1.0123508174019169e-05,
    "epoch": 0.8560311284046692,
    "step": 26620
  },
  {
    "loss": 0.07665892243385315,
    "grad_norm": 0.4102591574192047,
    "learning_rate": 1.0079118714200464e-05,
    "epoch": 0.8563527028330707,
    "step": 26630
  },
  {
    "loss": 0.07389369606971741,
    "grad_norm": 0.4486946165561676,
    "learning_rate": 1.003482162225331e-05,
    "epoch": 0.8566742772614722,
    "step": 26640
  },
  {
    "loss": 0.0670365035533905,
    "grad_norm": 0.6316031217575073,
    "learning_rate": 9.990616943680265e-06,
    "epoch": 0.8569958516898736,
    "step": 26650
  },
  {
    "loss": 0.06709270477294922,
    "grad_norm": 0.6071740984916687,
    "learning_rate": 9.946504723888905e-06,
    "epoch": 0.8573174261182751,
    "step": 26660
  },
  {
    "loss": 0.062121140956878665,
    "grad_norm": 0.4220135509967804,
    "learning_rate": 9.902485008191897e-06,
    "epoch": 0.8576390005466765,
    "step": 26670
  },
  {
    "loss": 0.07908244729042054,
    "grad_norm": 0.6105177402496338,
    "learning_rate": 9.858557841806803e-06,
    "epoch": 0.8579605749750779,
    "step": 26680
  },
  {
    "loss": 0.06170150637626648,
    "grad_norm": 0.6949994564056396,
    "learning_rate": 9.814723269856219e-06,
    "epoch": 0.8582821494034795,
    "step": 26690
  },
  {
    "loss": 0.07273870706558228,
    "grad_norm": 0.549635648727417,
    "learning_rate": 9.770981337367524e-06,
    "epoch": 0.8586037238318809,
    "step": 26700
  },
  {
    "loss": 0.07126254439353943,
    "grad_norm": 0.8531264066696167,
    "learning_rate": 9.727332089273022e-06,
    "epoch": 0.8589252982602823,
    "step": 26710
  },
  {
    "loss": 0.0822081744670868,
    "grad_norm": 0.5738287568092346,
    "learning_rate": 9.683775570409781e-06,
    "epoch": 0.8592468726886838,
    "step": 26720
  },
  {
    "loss": 0.05138362646102905,
    "grad_norm": 0.35763096809387207,
    "learning_rate": 9.640311825519577e-06,
    "epoch": 0.8595684471170852,
    "step": 26730
  },
  {
    "loss": 0.08185197710990906,
    "grad_norm": 0.8491652011871338,
    "learning_rate": 9.596940899248951e-06,
    "epoch": 0.8598900215454867,
    "step": 26740
  },
  {
    "loss": 0.07479264736175537,
    "grad_norm": 0.4308110773563385,
    "learning_rate": 9.553662836149047e-06,
    "epoch": 0.8602115959738882,
    "step": 26750
  },
  {
    "loss": 0.043882808089256285,
    "grad_norm": 0.49413225054740906,
    "learning_rate": 9.510477680675656e-06,
    "epoch": 0.8605331704022896,
    "step": 26760
  },
  {
    "loss": 0.0769636869430542,
    "grad_norm": 0.8833135962486267,
    "learning_rate": 9.467385477189105e-06,
    "epoch": 0.860854744830691,
    "step": 26770
  },
  {
    "loss": 0.05496814846992493,
    "grad_norm": 0.462801069021225,
    "learning_rate": 9.424386269954266e-06,
    "epoch": 0.8611763192590925,
    "step": 26780
  },
  {
    "loss": 0.080049329996109,
    "grad_norm": 0.6054528951644897,
    "learning_rate": 9.381480103140483e-06,
    "epoch": 0.861497893687494,
    "step": 26790
  },
  {
    "loss": 0.08080293536186219,
    "grad_norm": 0.5494915246963501,
    "learning_rate": 9.338667020821468e-06,
    "epoch": 0.8618194681158954,
    "step": 26800
  },
  {
    "loss": 0.08168855309486389,
    "grad_norm": 0.6569522619247437,
    "learning_rate": 9.29594706697543e-06,
    "epoch": 0.8621410425442969,
    "step": 26810
  },
  {
    "loss": 0.07000950574874878,
    "grad_norm": 0.34261035919189453,
    "learning_rate": 9.2533202854848e-06,
    "epoch": 0.8624626169726983,
    "step": 26820
  },
  {
    "loss": 0.07513426542282105,
    "grad_norm": 0.47495919466018677,
    "learning_rate": 9.210786720136333e-06,
    "epoch": 0.8627841914010997,
    "step": 26830
  },
  {
    "loss": 0.08019859790802002,
    "grad_norm": 0.7717815637588501,
    "learning_rate": 9.168346414621109e-06,
    "epoch": 0.8631057658295013,
    "step": 26840
  },
  {
    "loss": 0.06261076331138611,
    "grad_norm": 0.47469109296798706,
    "learning_rate": 9.125999412534291e-06,
    "epoch": 0.8634273402579027,
    "step": 26850
  },
  {
    "loss": 0.07787419557571411,
    "grad_norm": 0.7436442971229553,
    "learning_rate": 9.083745757375307e-06,
    "epoch": 0.8637489146863041,
    "step": 26860
  },
  {
    "loss": 0.05460736751556396,
    "grad_norm": 0.42278188467025757,
    "learning_rate": 9.041585492547633e-06,
    "epoch": 0.8640704891147056,
    "step": 26870
  },
  {
    "loss": 0.08874971866607666,
    "grad_norm": 0.5448344945907593,
    "learning_rate": 8.999518661358785e-06,
    "epoch": 0.864392063543107,
    "step": 26880
  },
  {
    "loss": 0.07377313375473023,
    "grad_norm": 0.5940506458282471,
    "learning_rate": 8.957545307020399e-06,
    "epoch": 0.8647136379715085,
    "step": 26890
  },
  {
    "loss": 0.06170197129249573,
    "grad_norm": 0.3291904926300049,
    "learning_rate": 8.915665472648004e-06,
    "epoch": 0.86503521239991,
    "step": 26900
  },
  {
    "loss": 0.06614161133766175,
    "grad_norm": 0.5995969772338867,
    "learning_rate": 8.873879201261138e-06,
    "epoch": 0.8653567868283114,
    "step": 26910
  },
  {
    "loss": 0.08219808340072632,
    "grad_norm": 0.5390191078186035,
    "learning_rate": 8.83218653578317e-06,
    "epoch": 0.8656783612567128,
    "step": 26920
  },
  {
    "loss": 0.07147032022476196,
    "grad_norm": 0.7250548005104065,
    "learning_rate": 8.790587519041305e-06,
    "epoch": 0.8659999356851144,
    "step": 26930
  },
  {
    "loss": 0.07538982033729554,
    "grad_norm": 0.3211059272289276,
    "learning_rate": 8.749082193766633e-06,
    "epoch": 0.8663215101135158,
    "step": 26940
  },
  {
    "loss": 0.048902788758277894,
    "grad_norm": 0.6315880417823792,
    "learning_rate": 8.707670602593909e-06,
    "epoch": 0.8666430845419172,
    "step": 26950
  },
  {
    "loss": 0.06604506373405457,
    "grad_norm": 0.5350209474563599,
    "learning_rate": 8.66635278806166e-06,
    "epoch": 0.8669646589703187,
    "step": 26960
  },
  {
    "loss": 0.06788616180419922,
    "grad_norm": 0.7068520784378052,
    "learning_rate": 8.625128792612081e-06,
    "epoch": 0.8672862333987201,
    "step": 26970
  },
  {
    "loss": 0.05400567650794983,
    "grad_norm": 0.781935453414917,
    "learning_rate": 8.583998658590975e-06,
    "epoch": 0.8676078078271215,
    "step": 26980
  },
  {
    "loss": 0.07868753671646118,
    "grad_norm": 0.6845842003822327,
    "learning_rate": 8.542962428247748e-06,
    "epoch": 0.8679293822555231,
    "step": 26990
  },
  {
    "loss": 0.07282733917236328,
    "grad_norm": 0.39165160059928894,
    "learning_rate": 8.502020143735302e-06,
    "epoch": 0.8682509566839245,
    "step": 27000
  },
  {
    "loss": 0.06988500952720642,
    "grad_norm": 0.6668932437896729,
    "learning_rate": 8.461171847110127e-06,
    "epoch": 0.8685725311123259,
    "step": 27010
  },
  {
    "loss": 0.06271352767944335,
    "grad_norm": 0.6527853012084961,
    "learning_rate": 8.420417580332052e-06,
    "epoch": 0.8688941055407274,
    "step": 27020
  },
  {
    "loss": 0.059419095516204834,
    "grad_norm": 0.514185905456543,
    "learning_rate": 8.37975738526442e-06,
    "epoch": 0.8692156799691289,
    "step": 27030
  },
  {
    "loss": 0.06655218005180359,
    "grad_norm": 0.41537120938301086,
    "learning_rate": 8.339191303673899e-06,
    "epoch": 0.8695372543975303,
    "step": 27040
  },
  {
    "loss": 0.053801441192626955,
    "grad_norm": 0.34744489192962646,
    "learning_rate": 8.298719377230457e-06,
    "epoch": 0.8698588288259318,
    "step": 27050
  },
  {
    "loss": 0.07902992367744446,
    "grad_norm": 0.9833202362060547,
    "learning_rate": 8.258341647507407e-06,
    "epoch": 0.8701804032543332,
    "step": 27060
  },
  {
    "loss": 0.06786506175994873,
    "grad_norm": 0.5024024248123169,
    "learning_rate": 8.218058155981223e-06,
    "epoch": 0.8705019776827346,
    "step": 27070
  },
  {
    "loss": 0.0703201413154602,
    "grad_norm": 0.6945049166679382,
    "learning_rate": 8.177868944031663e-06,
    "epoch": 0.8708235521111362,
    "step": 27080
  },
  {
    "loss": 0.0715039312839508,
    "grad_norm": 0.35131388902664185,
    "learning_rate": 8.13777405294156e-06,
    "epoch": 0.8711451265395376,
    "step": 27090
  },
  {
    "loss": 0.07369782328605652,
    "grad_norm": 0.5261259078979492,
    "learning_rate": 8.097773523896902e-06,
    "epoch": 0.871466700967939,
    "step": 27100
  },
  {
    "loss": 0.07828069925308227,
    "grad_norm": 1.102333903312683,
    "learning_rate": 8.057867397986773e-06,
    "epoch": 0.8717882753963405,
    "step": 27110
  },
  {
    "loss": 0.06868954300880432,
    "grad_norm": 0.34812596440315247,
    "learning_rate": 8.018055716203198e-06,
    "epoch": 0.8721098498247419,
    "step": 27120
  },
  {
    "loss": 0.07218150496482849,
    "grad_norm": 0.43147793412208557,
    "learning_rate": 7.978338519441308e-06,
    "epoch": 0.8724314242531434,
    "step": 27130
  },
  {
    "loss": 0.05935378670692444,
    "grad_norm": 0.36542806029319763,
    "learning_rate": 7.938715848499068e-06,
    "epoch": 0.8727529986815449,
    "step": 27140
  },
  {
    "loss": 0.08468295931816101,
    "grad_norm": 0.5510056614875793,
    "learning_rate": 7.899187744077353e-06,
    "epoch": 0.8730745731099463,
    "step": 27150
  },
  {
    "loss": 0.09099004268646241,
    "grad_norm": 1.0482388734817505,
    "learning_rate": 7.859754246780027e-06,
    "epoch": 0.8733961475383477,
    "step": 27160
  },
  {
    "loss": 0.06953157782554627,
    "grad_norm": 0.5658690333366394,
    "learning_rate": 7.820415397113601e-06,
    "epoch": 0.8737177219667492,
    "step": 27170
  },
  {
    "loss": 0.08543871641159058,
    "grad_norm": 0.7472510933876038,
    "learning_rate": 7.781171235487494e-06,
    "epoch": 0.8740392963951507,
    "step": 27180
  },
  {
    "loss": 0.04514435529708862,
    "grad_norm": 0.5680370926856995,
    "learning_rate": 7.742021802213795e-06,
    "epoch": 0.8743608708235521,
    "step": 27190
  },
  {
    "loss": 0.07006924748420715,
    "grad_norm": 0.7031437158584595,
    "learning_rate": 7.702967137507256e-06,
    "epoch": 0.8746824452519536,
    "step": 27200
  },
  {
    "loss": 0.0831302523612976,
    "grad_norm": 0.8786540627479553,
    "learning_rate": 7.664007281485363e-06,
    "epoch": 0.875004019680355,
    "step": 27210
  },
  {
    "loss": 0.055636584758758545,
    "grad_norm": 0.5265671014785767,
    "learning_rate": 7.625142274168184e-06,
    "epoch": 0.8753255941087564,
    "step": 27220
  },
  {
    "loss": 0.09252220988273621,
    "grad_norm": 0.5989711880683899,
    "learning_rate": 7.586372155478316e-06,
    "epoch": 0.875647168537158,
    "step": 27230
  },
  {
    "loss": 0.05664872527122498,
    "grad_norm": 0.47775182127952576,
    "learning_rate": 7.547696965240958e-06,
    "epoch": 0.8759687429655594,
    "step": 27240
  },
  {
    "loss": 0.0712408185005188,
    "grad_norm": 0.8774183392524719,
    "learning_rate": 7.509116743183708e-06,
    "epoch": 0.8762903173939608,
    "step": 27250
  },
  {
    "loss": 0.07321785688400269,
    "grad_norm": 0.8138816356658936,
    "learning_rate": 7.470631528936711e-06,
    "epoch": 0.8766118918223623,
    "step": 27260
  },
  {
    "loss": 0.07316104769706726,
    "grad_norm": 0.537562906742096,
    "learning_rate": 7.4322413620324215e-06,
    "epoch": 0.8769334662507637,
    "step": 27270
  },
  {
    "loss": 0.06197282075881958,
    "grad_norm": 0.5482314825057983,
    "learning_rate": 7.393946281905728e-06,
    "epoch": 0.8772550406791652,
    "step": 27280
  },
  {
    "loss": 0.06532438397407532,
    "grad_norm": 0.8577420115470886,
    "learning_rate": 7.355746327893831e-06,
    "epoch": 0.8775766151075667,
    "step": 27290
  },
  {
    "loss": 0.058403903245925905,
    "grad_norm": 0.7714664340019226,
    "learning_rate": 7.317641539236198e-06,
    "epoch": 0.8778981895359681,
    "step": 27300
  },
  {
    "loss": 0.06350155472755432,
    "grad_norm": 0.6568384170532227,
    "learning_rate": 7.2796319550745615e-06,
    "epoch": 0.8782197639643695,
    "step": 27310
  },
  {
    "loss": 0.07977257370948791,
    "grad_norm": 0.7861108183860779,
    "learning_rate": 7.241717614452826e-06,
    "epoch": 0.878541338392771,
    "step": 27320
  },
  {
    "loss": 0.0729379415512085,
    "grad_norm": 0.3218686878681183,
    "learning_rate": 7.203898556317123e-06,
    "epoch": 0.8788629128211725,
    "step": 27330
  },
  {
    "loss": 0.05332390666007995,
    "grad_norm": 0.3437587022781372,
    "learning_rate": 7.166174819515626e-06,
    "epoch": 0.8791844872495739,
    "step": 27340
  },
  {
    "loss": 0.05867266654968262,
    "grad_norm": 0.6583501100540161,
    "learning_rate": 7.128546442798656e-06,
    "epoch": 0.8795060616779754,
    "step": 27350
  },
  {
    "loss": 0.06763720512390137,
    "grad_norm": 0.8608779907226562,
    "learning_rate": 7.0910134648185854e-06,
    "epoch": 0.8798276361063768,
    "step": 27360
  },
  {
    "loss": 0.055006229877471925,
    "grad_norm": 0.5322985649108887,
    "learning_rate": 7.053575924129718e-06,
    "epoch": 0.8801492105347782,
    "step": 27370
  },
  {
    "loss": 0.055677604675292966,
    "grad_norm": 0.43913590908050537,
    "learning_rate": 7.016233859188415e-06,
    "epoch": 0.8804707849631798,
    "step": 27380
  },
  {
    "loss": 0.07852984070777894,
    "grad_norm": 0.650809109210968,
    "learning_rate": 6.97898730835288e-06,
    "epoch": 0.8807923593915812,
    "step": 27390
  },
  {
    "loss": 0.06146419048309326,
    "grad_norm": 0.5928049683570862,
    "learning_rate": 6.941836309883254e-06,
    "epoch": 0.8811139338199826,
    "step": 27400
  },
  {
    "loss": 0.08269037008285522,
    "grad_norm": 0.49598103761672974,
    "learning_rate": 6.904780901941555e-06,
    "epoch": 0.8814355082483841,
    "step": 27410
  },
  {
    "loss": 0.06693826913833618,
    "grad_norm": 0.33657902479171753,
    "learning_rate": 6.867821122591511e-06,
    "epoch": 0.8817570826767855,
    "step": 27420
  },
  {
    "loss": 0.08126822710037232,
    "grad_norm": 0.8901121020317078,
    "learning_rate": 6.830957009798722e-06,
    "epoch": 0.882078657105187,
    "step": 27430
  },
  {
    "loss": 0.07340906858444214,
    "grad_norm": 0.4934292733669281,
    "learning_rate": 6.794188601430451e-06,
    "epoch": 0.8824002315335885,
    "step": 27440
  },
  {
    "loss": 0.08674216270446777,
    "grad_norm": 0.6380556225776672,
    "learning_rate": 6.757515935255665e-06,
    "epoch": 0.8827218059619899,
    "step": 27450
  },
  {
    "loss": 0.07352336049079895,
    "grad_norm": 0.5957514643669128,
    "learning_rate": 6.720939048945018e-06,
    "epoch": 0.8830433803903913,
    "step": 27460
  },
  {
    "loss": 0.06992534399032593,
    "grad_norm": 0.9841471910476685,
    "learning_rate": 6.6844579800707016e-06,
    "epoch": 0.8833649548187928,
    "step": 27470
  },
  {
    "loss": 0.08985934257507325,
    "grad_norm": 1.4467130899429321,
    "learning_rate": 6.6480727661066055e-06,
    "epoch": 0.8836865292471943,
    "step": 27480
  },
  {
    "loss": 0.06457531452178955,
    "grad_norm": 0.6235713362693787,
    "learning_rate": 6.6117834444280545e-06,
    "epoch": 0.8840081036755957,
    "step": 27490
  },
  {
    "loss": 0.06599720120429993,
    "grad_norm": 0.4847148060798645,
    "learning_rate": 6.575590052311875e-06,
    "epoch": 0.8843296781039972,
    "step": 27500
  },
  {
    "loss": 0.08554141521453858,
    "grad_norm": 0.4832707345485687,
    "learning_rate": 6.539492626936417e-06,
    "epoch": 0.8846512525323986,
    "step": 27510
  },
  {
    "loss": 0.0616130530834198,
    "grad_norm": 0.6885722279548645,
    "learning_rate": 6.503491205381374e-06,
    "epoch": 0.8849728269608,
    "step": 27520
  },
  {
    "loss": 0.06359918117523193,
    "grad_norm": 0.5734153389930725,
    "learning_rate": 6.467585824627887e-06,
    "epoch": 0.8852944013892016,
    "step": 27530
  },
  {
    "loss": 0.07236401438713073,
    "grad_norm": 0.9058264493942261,
    "learning_rate": 6.431776521558441e-06,
    "epoch": 0.885615975817603,
    "step": 27540
  },
  {
    "loss": 0.05739564299583435,
    "grad_norm": 0.626163899898529,
    "learning_rate": 6.3960633329567585e-06,
    "epoch": 0.8859375502460044,
    "step": 27550
  },
  {
    "loss": 0.10272804498672486,
    "grad_norm": 0.348671019077301,
    "learning_rate": 6.360446295507927e-06,
    "epoch": 0.8862591246744059,
    "step": 27560
  },
  {
    "loss": 0.0653518259525299,
    "grad_norm": 0.6510688066482544,
    "learning_rate": 6.324925445798191e-06,
    "epoch": 0.8865806991028073,
    "step": 27570
  },
  {
    "loss": 0.0677101731300354,
    "grad_norm": 0.595810055732727,
    "learning_rate": 6.289500820315031e-06,
    "epoch": 0.8869022735312088,
    "step": 27580
  },
  {
    "loss": 0.08281749486923218,
    "grad_norm": 2.272451400756836,
    "learning_rate": 6.2541724554470605e-06,
    "epoch": 0.8872238479596103,
    "step": 27590
  },
  {
    "loss": 0.08681055307388305,
    "grad_norm": 0.5360921621322632,
    "learning_rate": 6.218940387484018e-06,
    "epoch": 0.8875454223880117,
    "step": 27600
  },
  {
    "loss": 0.05985884070396423,
    "grad_norm": 0.651020348072052,
    "learning_rate": 6.183804652616776e-06,
    "epoch": 0.8878669968164131,
    "step": 27610
  },
  {
    "loss": 0.09359393119812012,
    "grad_norm": 0.6078817844390869,
    "learning_rate": 6.148765286937153e-06,
    "epoch": 0.8881885712448147,
    "step": 27620
  },
  {
    "loss": 0.0729813039302826,
    "grad_norm": 0.8955041766166687,
    "learning_rate": 6.113822326438079e-06,
    "epoch": 0.8885101456732161,
    "step": 27630
  },
  {
    "loss": 0.06026132702827454,
    "grad_norm": 0.3575083911418915,
    "learning_rate": 6.078975807013354e-06,
    "epoch": 0.8888317201016175,
    "step": 27640
  },
  {
    "loss": 0.07106353044509887,
    "grad_norm": 0.6659265160560608,
    "learning_rate": 6.0442257644578225e-06,
    "epoch": 0.889153294530019,
    "step": 27650
  },
  {
    "loss": 0.0630476176738739,
    "grad_norm": 0.6418014764785767,
    "learning_rate": 6.00957223446712e-06,
    "epoch": 0.8894748689584204,
    "step": 27660
  },
  {
    "loss": 0.07165031433105469,
    "grad_norm": 0.5182862281799316,
    "learning_rate": 5.975015252637806e-06,
    "epoch": 0.8897964433868218,
    "step": 27670
  },
  {
    "loss": 0.08316783308982849,
    "grad_norm": 0.7247962951660156,
    "learning_rate": 5.940554854467284e-06,
    "epoch": 0.8901180178152234,
    "step": 27680
  },
  {
    "loss": 0.07041310071945191,
    "grad_norm": 0.5974995493888855,
    "learning_rate": 5.906191075353673e-06,
    "epoch": 0.8904395922436248,
    "step": 27690
  },
  {
    "loss": 0.07408487200736999,
    "grad_norm": 0.5634307861328125,
    "learning_rate": 5.871923950595903e-06,
    "epoch": 0.8907611666720262,
    "step": 27700
  },
  {
    "loss": 0.06824617981910705,
    "grad_norm": 0.7515895962715149,
    "learning_rate": 5.837753515393607e-06,
    "epoch": 0.8910827411004277,
    "step": 27710
  },
  {
    "loss": 0.05768151879310608,
    "grad_norm": 0.399549663066864,
    "learning_rate": 5.80367980484704e-06,
    "epoch": 0.8914043155288291,
    "step": 27720
  },
  {
    "loss": 0.07902880907058715,
    "grad_norm": 0.8869091868400574,
    "learning_rate": 5.769702853957226e-06,
    "epoch": 0.8917258899572306,
    "step": 27730
  },
  {
    "loss": 0.06725460886955262,
    "grad_norm": 0.35859981179237366,
    "learning_rate": 5.73582269762567e-06,
    "epoch": 0.8920474643856321,
    "step": 27740
  },
  {
    "loss": 0.06240668296813965,
    "grad_norm": 0.5596455335617065,
    "learning_rate": 5.702039370654522e-06,
    "epoch": 0.8923690388140335,
    "step": 27750
  },
  {
    "loss": 0.06892146468162537,
    "grad_norm": 0.7789075970649719,
    "learning_rate": 5.668352907746433e-06,
    "epoch": 0.8926906132424349,
    "step": 27760
  },
  {
    "loss": 0.06604178547859192,
    "grad_norm": 0.256012499332428,
    "learning_rate": 5.634763343504546e-06,
    "epoch": 0.8930121876708365,
    "step": 27770
  },
  {
    "loss": 0.07076762318611145,
    "grad_norm": 0.9413619637489319,
    "learning_rate": 5.601270712432538e-06,
    "epoch": 0.8933337620992379,
    "step": 27780
  },
  {
    "loss": 0.0627564549446106,
    "grad_norm": 0.7740989923477173,
    "learning_rate": 5.567875048934401e-06,
    "epoch": 0.8936553365276393,
    "step": 27790
  },
  {
    "loss": 0.08443350791931152,
    "grad_norm": 0.8512759804725647,
    "learning_rate": 5.534576387314649e-06,
    "epoch": 0.8939769109560408,
    "step": 27800
  },
  {
    "loss": 0.061937087774276735,
    "grad_norm": 0.7870593667030334,
    "learning_rate": 5.5013747617780554e-06,
    "epoch": 0.8942984853844422,
    "step": 27810
  },
  {
    "loss": 0.05405592918395996,
    "grad_norm": 0.5975480079650879,
    "learning_rate": 5.4682702064297485e-06,
    "epoch": 0.8946200598128436,
    "step": 27820
  },
  {
    "loss": 0.06346514225006103,
    "grad_norm": 0.6930370330810547,
    "learning_rate": 5.435262755275172e-06,
    "epoch": 0.8949416342412452,
    "step": 27830
  },
  {
    "loss": 0.07561540603637695,
    "grad_norm": 1.0342068672180176,
    "learning_rate": 5.402352442219971e-06,
    "epoch": 0.8952632086696466,
    "step": 27840
  },
  {
    "loss": 0.08634264469146728,
    "grad_norm": 0.8079270720481873,
    "learning_rate": 5.369539301070059e-06,
    "epoch": 0.895584783098048,
    "step": 27850
  },
  {
    "loss": 0.07511700391769409,
    "grad_norm": 0.8758837580680847,
    "learning_rate": 5.336823365531507e-06,
    "epoch": 0.8959063575264495,
    "step": 27860
  },
  {
    "loss": 0.06092031598091126,
    "grad_norm": 0.5970578789710999,
    "learning_rate": 5.3042046692105214e-06,
    "epoch": 0.896227931954851,
    "step": 27870
  },
  {
    "loss": 0.08482131361961365,
    "grad_norm": 1.5843493938446045,
    "learning_rate": 5.271683245613468e-06,
    "epoch": 0.8965495063832524,
    "step": 27880
  },
  {
    "loss": 0.06481119394302368,
    "grad_norm": 0.6575544476509094,
    "learning_rate": 5.239259128146745e-06,
    "epoch": 0.8968710808116539,
    "step": 27890
  },
  {
    "loss": 0.07339617013931274,
    "grad_norm": 0.574725329875946,
    "learning_rate": 5.206932350116845e-06,
    "epoch": 0.8971926552400553,
    "step": 27900
  },
  {
    "loss": 0.06229846477508545,
    "grad_norm": 0.4941605031490326,
    "learning_rate": 5.174702944730203e-06,
    "epoch": 0.8975142296684567,
    "step": 27910
  },
  {
    "loss": 0.07401782870292664,
    "grad_norm": 0.7536153197288513,
    "learning_rate": 5.142570945093305e-06,
    "epoch": 0.8978358040968583,
    "step": 27920
  },
  {
    "loss": 0.06073766350746155,
    "grad_norm": 0.7745930552482605,
    "learning_rate": 5.11053638421255e-06,
    "epoch": 0.8981573785252597,
    "step": 27930
  },
  {
    "loss": 0.055574798583984376,
    "grad_norm": 0.6120657920837402,
    "learning_rate": 5.078599294994224e-06,
    "epoch": 0.8984789529536611,
    "step": 27940
  },
  {
    "loss": 0.07941893339157105,
    "grad_norm": 0.3994508683681488,
    "learning_rate": 5.046759710244542e-06,
    "epoch": 0.8988005273820626,
    "step": 27950
  },
  {
    "loss": 0.06143311858177185,
    "grad_norm": 0.42189642786979675,
    "learning_rate": 5.015017662669497e-06,
    "epoch": 0.899122101810464,
    "step": 27960
  },
  {
    "loss": 0.06756633520126343,
    "grad_norm": 0.48182159662246704,
    "learning_rate": 4.983373184874951e-06,
    "epoch": 0.8994436762388655,
    "step": 27970
  },
  {
    "eval_loss": 0.06576023995876312,
    "eval_runtime": 34.5615,
    "eval_samples_per_second": 145.422,
    "eval_steps_per_second": 36.37,
    "epoch": 0.8995079911245458,
    "step": 27972
  },
  {
    "loss": 0.08180526494979859,
    "grad_norm": 0.5850648283958435,
    "learning_rate": 4.951826309366481e-06,
    "epoch": 0.899765250667267,
    "step": 27980
  },
  {
    "loss": 0.07834064364433288,
    "grad_norm": 0.5346769690513611,
    "learning_rate": 4.920377068549475e-06,
    "epoch": 0.9000868250956684,
    "step": 27990
  },
  {
    "loss": 0.07173922061920165,
    "grad_norm": 0.5718779563903809,
    "learning_rate": 4.8890254947289714e-06,
    "epoch": 0.9004083995240698,
    "step": 28000
  },
  {
    "loss": 0.05986430048942566,
    "grad_norm": 0.8651319742202759,
    "learning_rate": 4.857771620109708e-06,
    "epoch": 0.9007299739524713,
    "step": 28010
  },
  {
    "loss": 0.08395033478736877,
    "grad_norm": 0.5755200982093811,
    "learning_rate": 4.826615476796071e-06,
    "epoch": 0.9010515483808728,
    "step": 28020
  },
  {
    "loss": 0.07701780796051025,
    "grad_norm": 0.5145189762115479,
    "learning_rate": 4.7955570967920475e-06,
    "epoch": 0.9013731228092742,
    "step": 28030
  },
  {
    "loss": 0.07752405405044556,
    "grad_norm": 0.40484297275543213,
    "learning_rate": 4.764596512001162e-06,
    "epoch": 0.9016946972376757,
    "step": 28040
  },
  {
    "loss": 0.07332845330238343,
    "grad_norm": 0.3679267168045044,
    "learning_rate": 4.7337337542265656e-06,
    "epoch": 0.9020162716660771,
    "step": 28050
  },
  {
    "loss": 0.06816344857215881,
    "grad_norm": 0.4973434805870056,
    "learning_rate": 4.702968855170864e-06,
    "epoch": 0.9023378460944785,
    "step": 28060
  },
  {
    "loss": 0.0697591781616211,
    "grad_norm": 0.9307938814163208,
    "learning_rate": 4.672301846436155e-06,
    "epoch": 0.9026594205228801,
    "step": 28070
  },
  {
    "loss": 0.06903193593025207,
    "grad_norm": 0.707920253276825,
    "learning_rate": 4.641732759523964e-06,
    "epoch": 0.9029809949512815,
    "step": 28080
  },
  {
    "loss": 0.08139731884002685,
    "grad_norm": 0.4234994053840637,
    "learning_rate": 4.61126162583525e-06,
    "epoch": 0.9033025693796829,
    "step": 28090
  },
  {
    "loss": 0.09809705018997192,
    "grad_norm": 0.8795967102050781,
    "learning_rate": 4.580888476670353e-06,
    "epoch": 0.9036241438080844,
    "step": 28100
  },
  {
    "loss": 0.05017074942588806,
    "grad_norm": 0.7633052468299866,
    "learning_rate": 4.55061334322896e-06,
    "epoch": 0.9039457182364858,
    "step": 28110
  },
  {
    "loss": 0.06319268345832825,
    "grad_norm": 0.5459538102149963,
    "learning_rate": 4.520436256610072e-06,
    "epoch": 0.9042672926648873,
    "step": 28120
  },
  {
    "loss": 0.06305754780769349,
    "grad_norm": 0.1714586466550827,
    "learning_rate": 4.4903572478119826e-06,
    "epoch": 0.9045888670932888,
    "step": 28130
  },
  {
    "loss": 0.07959019541740417,
    "grad_norm": 0.6232452392578125,
    "learning_rate": 4.460376347732198e-06,
    "epoch": 0.9049104415216902,
    "step": 28140
  },
  {
    "loss": 0.06855315566062928,
    "grad_norm": 0.5981919169425964,
    "learning_rate": 4.430493587167528e-06,
    "epoch": 0.9052320159500916,
    "step": 28150
  },
  {
    "loss": 0.08559853434562684,
    "grad_norm": 0.8783455491065979,
    "learning_rate": 4.400708996813884e-06,
    "epoch": 0.9055535903784931,
    "step": 28160
  },
  {
    "loss": 0.07695115208625794,
    "grad_norm": 0.3367885947227478,
    "learning_rate": 4.371022607266395e-06,
    "epoch": 0.9058751648068946,
    "step": 28170
  },
  {
    "loss": 0.05580394268035889,
    "grad_norm": 0.482756644487381,
    "learning_rate": 4.34143444901931e-06,
    "epoch": 0.906196739235296,
    "step": 28180
  },
  {
    "loss": 0.06382061243057251,
    "grad_norm": 0.2901076078414917,
    "learning_rate": 4.31194455246593e-06,
    "epoch": 0.9065183136636975,
    "step": 28190
  },
  {
    "loss": 0.06710493564605713,
    "grad_norm": 0.6091313362121582,
    "learning_rate": 4.282552947898677e-06,
    "epoch": 0.9068398880920989,
    "step": 28200
  },
  {
    "loss": 0.05544446706771851,
    "grad_norm": 0.4454199969768524,
    "learning_rate": 4.253259665508958e-06,
    "epoch": 0.9071614625205003,
    "step": 28210
  },
  {
    "loss": 0.07407185435295105,
    "grad_norm": 0.5129989385604858,
    "learning_rate": 4.224064735387234e-06,
    "epoch": 0.9074830369489019,
    "step": 28220
  },
  {
    "loss": 0.07657233476638795,
    "grad_norm": 0.5245542526245117,
    "learning_rate": 4.194968187522874e-06,
    "epoch": 0.9078046113773033,
    "step": 28230
  },
  {
    "loss": 0.0796772301197052,
    "grad_norm": 0.641661524772644,
    "learning_rate": 4.165970051804235e-06,
    "epoch": 0.9081261858057047,
    "step": 28240
  },
  {
    "loss": 0.06403705477714539,
    "grad_norm": 0.3376065790653229,
    "learning_rate": 4.137070358018602e-06,
    "epoch": 0.9084477602341062,
    "step": 28250
  },
  {
    "loss": 0.06514493227005005,
    "grad_norm": 0.3697962760925293,
    "learning_rate": 4.1082691358520496e-06,
    "epoch": 0.9087693346625076,
    "step": 28260
  },
  {
    "loss": 0.06303318738937377,
    "grad_norm": 0.6878823637962341,
    "learning_rate": 4.079566414889613e-06,
    "epoch": 0.9090909090909091,
    "step": 28270
  },
  {
    "loss": 0.06593666672706604,
    "grad_norm": 0.5542412400245667,
    "learning_rate": 4.050962224615051e-06,
    "epoch": 0.9094124835193106,
    "step": 28280
  },
  {
    "loss": 0.08710123896598816,
    "grad_norm": 0.7127916216850281,
    "learning_rate": 4.022456594410984e-06,
    "epoch": 0.909734057947712,
    "step": 28290
  },
  {
    "loss": 0.06951413154602051,
    "grad_norm": 0.7896924018859863,
    "learning_rate": 3.994049553558732e-06,
    "epoch": 0.9100556323761134,
    "step": 28300
  },
  {
    "loss": 0.08014414310455323,
    "grad_norm": 0.8470936417579651,
    "learning_rate": 3.9657411312383765e-06,
    "epoch": 0.910377206804515,
    "step": 28310
  },
  {
    "loss": 0.06745404601097107,
    "grad_norm": 0.4270573556423187,
    "learning_rate": 3.93753135652869e-06,
    "epoch": 0.9106987812329164,
    "step": 28320
  },
  {
    "loss": 0.07204972505569458,
    "grad_norm": 0.8026243448257446,
    "learning_rate": 3.909420258407126e-06,
    "epoch": 0.9110203556613178,
    "step": 28330
  },
  {
    "loss": 0.09062637090682983,
    "grad_norm": 0.9312450289726257,
    "learning_rate": 3.881407865749731e-06,
    "epoch": 0.9113419300897193,
    "step": 28340
  },
  {
    "loss": 0.06378825306892395,
    "grad_norm": 0.7994136214256287,
    "learning_rate": 3.853494207331221e-06,
    "epoch": 0.9116635045181207,
    "step": 28350
  },
  {
    "loss": 0.06912797093391418,
    "grad_norm": 0.7310376167297363,
    "learning_rate": 3.825679311824815e-06,
    "epoch": 0.9119850789465221,
    "step": 28360
  },
  {
    "loss": 0.05452196598052979,
    "grad_norm": 0.4820187985897064,
    "learning_rate": 3.797963207802391e-06,
    "epoch": 0.9123066533749237,
    "step": 28370
  },
  {
    "loss": 0.07047023177146912,
    "grad_norm": 0.451716810464859,
    "learning_rate": 3.770345923734242e-06,
    "epoch": 0.9126282278033251,
    "step": 28380
  },
  {
    "loss": 0.07474204897880554,
    "grad_norm": 0.7766392827033997,
    "learning_rate": 3.7428274879891865e-06,
    "epoch": 0.9129498022317265,
    "step": 28390
  },
  {
    "loss": 0.06405274868011475,
    "grad_norm": 0.5303685665130615,
    "learning_rate": 3.715407928834547e-06,
    "epoch": 0.913271376660128,
    "step": 28400
  },
  {
    "loss": 0.07459316849708557,
    "grad_norm": 0.33874908089637756,
    "learning_rate": 3.688087274436014e-06,
    "epoch": 0.9135929510885294,
    "step": 28410
  },
  {
    "loss": 0.09029234647750854,
    "grad_norm": 0.7368858456611633,
    "learning_rate": 3.6608655528577064e-06,
    "epoch": 0.9139145255169309,
    "step": 28420
  },
  {
    "loss": 0.0907670497894287,
    "grad_norm": 1.2996673583984375,
    "learning_rate": 3.633742792062156e-06,
    "epoch": 0.9142360999453324,
    "step": 28430
  },
  {
    "loss": 0.06846472024917602,
    "grad_norm": 0.47941938042640686,
    "learning_rate": 3.6067190199101765e-06,
    "epoch": 0.9145576743737338,
    "step": 28440
  },
  {
    "loss": 0.07348551750183105,
    "grad_norm": 0.6636233925819397,
    "learning_rate": 3.5797942641609405e-06,
    "epoch": 0.9148792488021352,
    "step": 28450
  },
  {
    "loss": 0.08235877752304077,
    "grad_norm": 0.9041732549667358,
    "learning_rate": 3.5529685524718913e-06,
    "epoch": 0.9152008232305368,
    "step": 28460
  },
  {
    "loss": 0.07450997829437256,
    "grad_norm": 0.520968496799469,
    "learning_rate": 3.5262419123987643e-06,
    "epoch": 0.9155223976589382,
    "step": 28470
  },
  {
    "loss": 0.05997384190559387,
    "grad_norm": 0.3654121458530426,
    "learning_rate": 3.4996143713954653e-06,
    "epoch": 0.9158439720873396,
    "step": 28480
  },
  {
    "loss": 0.07119588851928711,
    "grad_norm": 0.5396535396575928,
    "learning_rate": 3.4730859568141703e-06,
    "epoch": 0.9161655465157411,
    "step": 28490
  },
  {
    "loss": 0.061024290323257444,
    "grad_norm": 0.5486513376235962,
    "learning_rate": 3.446656695905215e-06,
    "epoch": 0.9164871209441425,
    "step": 28500
  },
  {
    "loss": 0.06300418376922608,
    "grad_norm": 0.5239927768707275,
    "learning_rate": 3.4203266158170376e-06,
    "epoch": 0.916808695372544,
    "step": 28510
  },
  {
    "loss": 0.07844939231872558,
    "grad_norm": 0.7887962460517883,
    "learning_rate": 3.3940957435962597e-06,
    "epoch": 0.9171302698009455,
    "step": 28520
  },
  {
    "loss": 0.06970057487487794,
    "grad_norm": 0.6157139539718628,
    "learning_rate": 3.367964106187549e-06,
    "epoch": 0.9174518442293469,
    "step": 28530
  },
  {
    "loss": 0.06430141925811768,
    "grad_norm": 0.40715357661247253,
    "learning_rate": 3.341931730433667e-06,
    "epoch": 0.9177734186577483,
    "step": 28540
  },
  {
    "loss": 0.07139861583709717,
    "grad_norm": 0.35261547565460205,
    "learning_rate": 3.31599864307538e-06,
    "epoch": 0.9180949930861498,
    "step": 28550
  },
  {
    "loss": 0.07542607784271241,
    "grad_norm": 0.9486578106880188,
    "learning_rate": 3.290164870751511e-06,
    "epoch": 0.9184165675145513,
    "step": 28560
  },
  {
    "loss": 0.0655196487903595,
    "grad_norm": 0.69345623254776,
    "learning_rate": 3.2644304399988225e-06,
    "epoch": 0.9187381419429527,
    "step": 28570
  },
  {
    "loss": 0.08214229941368104,
    "grad_norm": 0.9937447905540466,
    "learning_rate": 3.2387953772520574e-06,
    "epoch": 0.9190597163713542,
    "step": 28580
  },
  {
    "loss": 0.08389940857887268,
    "grad_norm": 0.32834765315055847,
    "learning_rate": 3.2132597088438855e-06,
    "epoch": 0.9193812907997556,
    "step": 28590
  },
  {
    "loss": 0.0574823260307312,
    "grad_norm": 0.302408903837204,
    "learning_rate": 3.1878234610048464e-06,
    "epoch": 0.919702865228157,
    "step": 28600
  },
  {
    "loss": 0.06297532320022584,
    "grad_norm": 0.7567213177680969,
    "learning_rate": 3.162486659863373e-06,
    "epoch": 0.9200244396565586,
    "step": 28610
  },
  {
    "loss": 0.06948093771934509,
    "grad_norm": 0.39594024419784546,
    "learning_rate": 3.1372493314457906e-06,
    "epoch": 0.92034601408496,
    "step": 28620
  },
  {
    "loss": 0.06656071543693542,
    "grad_norm": 0.7741539478302002,
    "learning_rate": 3.1121115016761736e-06,
    "epoch": 0.9206675885133614,
    "step": 28630
  },
  {
    "loss": 0.0799754023551941,
    "grad_norm": 0.567747175693512,
    "learning_rate": 3.087073196376422e-06,
    "epoch": 0.9209891629417629,
    "step": 28640
  },
  {
    "loss": 0.05742677450180054,
    "grad_norm": 0.4938858449459076,
    "learning_rate": 3.062134441266218e-06,
    "epoch": 0.9213107373701643,
    "step": 28650
  },
  {
    "loss": 0.09684234261512756,
    "grad_norm": 0.6577500700950623,
    "learning_rate": 3.0372952619629356e-06,
    "epoch": 0.9216323117985658,
    "step": 28660
  },
  {
    "loss": 0.08077757954597473,
    "grad_norm": 0.9018112421035767,
    "learning_rate": 3.0125556839817214e-06,
    "epoch": 0.9219538862269673,
    "step": 28670
  },
  {
    "loss": 0.07379125952720642,
    "grad_norm": 0.6212087273597717,
    "learning_rate": 2.987915732735358e-06,
    "epoch": 0.9222754606553687,
    "step": 28680
  },
  {
    "loss": 0.06398572921752929,
    "grad_norm": 0.5765277147293091,
    "learning_rate": 2.9633754335343656e-06,
    "epoch": 0.9225970350837701,
    "step": 28690
  },
  {
    "loss": 0.05859953761100769,
    "grad_norm": 0.7695963978767395,
    "learning_rate": 2.938934811586802e-06,
    "epoch": 0.9229186095121716,
    "step": 28700
  },
  {
    "loss": 0.07643457055091858,
    "grad_norm": 0.7159215807914734,
    "learning_rate": 2.9145938919984075e-06,
    "epoch": 0.9232401839405731,
    "step": 28710
  },
  {
    "loss": 0.0891688585281372,
    "grad_norm": 0.9027368426322937,
    "learning_rate": 2.8903526997724804e-06,
    "epoch": 0.9235617583689745,
    "step": 28720
  },
  {
    "loss": 0.05892614722251892,
    "grad_norm": 0.478014200925827,
    "learning_rate": 2.8662112598098813e-06,
    "epoch": 0.923883332797376,
    "step": 28730
  },
  {
    "loss": 0.0703760802745819,
    "grad_norm": 0.38507896661758423,
    "learning_rate": 2.8421695969089946e-06,
    "epoch": 0.9242049072257774,
    "step": 28740
  },
  {
    "loss": 0.06641483306884766,
    "grad_norm": 1.0187102556228638,
    "learning_rate": 2.818227735765733e-06,
    "epoch": 0.9245264816541788,
    "step": 28750
  },
  {
    "loss": 0.08073602914810181,
    "grad_norm": 0.4585058093070984,
    "learning_rate": 2.7943857009734785e-06,
    "epoch": 0.9248480560825804,
    "step": 28760
  },
  {
    "loss": 0.07679535746574402,
    "grad_norm": 0.5923634171485901,
    "learning_rate": 2.770643517023086e-06,
    "epoch": 0.9251696305109818,
    "step": 28770
  },
  {
    "loss": 0.0702307403087616,
    "grad_norm": 0.8317908644676208,
    "learning_rate": 2.747001208302791e-06,
    "epoch": 0.9254912049393832,
    "step": 28780
  },
  {
    "loss": 0.06400009393692016,
    "grad_norm": 0.3549131751060486,
    "learning_rate": 2.723458799098311e-06,
    "epoch": 0.9258127793677847,
    "step": 28790
  },
  {
    "loss": 0.058918613195419314,
    "grad_norm": 0.38816264271736145,
    "learning_rate": 2.7000163135926793e-06,
    "epoch": 0.9261343537961861,
    "step": 28800
  },
  {
    "loss": 0.07531256675720215,
    "grad_norm": 0.44148746132850647,
    "learning_rate": 2.676673775866334e-06,
    "epoch": 0.9264559282245876,
    "step": 28810
  },
  {
    "loss": 0.07543994188308716,
    "grad_norm": 0.4504343569278717,
    "learning_rate": 2.653431209897028e-06,
    "epoch": 0.9267775026529891,
    "step": 28820
  },
  {
    "loss": 0.06393010020256043,
    "grad_norm": 0.5456916689872742,
    "learning_rate": 2.630288639559808e-06,
    "epoch": 0.9270990770813905,
    "step": 28830
  },
  {
    "loss": 0.06197878122329712,
    "grad_norm": 0.26009121537208557,
    "learning_rate": 2.6072460886270243e-06,
    "epoch": 0.9274206515097919,
    "step": 28840
  },
  {
    "loss": 0.06824758648872375,
    "grad_norm": 0.45600271224975586,
    "learning_rate": 2.584303580768266e-06,
    "epoch": 0.9277422259381934,
    "step": 28850
  },
  {
    "loss": 0.06901856660842895,
    "grad_norm": 0.6517654657363892,
    "learning_rate": 2.561461139550392e-06,
    "epoch": 0.9280638003665949,
    "step": 28860
  },
  {
    "loss": 0.06693161725997925,
    "grad_norm": 0.7441944479942322,
    "learning_rate": 2.538718788437422e-06,
    "epoch": 0.9283853747949963,
    "step": 28870
  },
  {
    "loss": 0.08583144545555114,
    "grad_norm": 0.8883978724479675,
    "learning_rate": 2.5160765507905915e-06,
    "epoch": 0.9287069492233978,
    "step": 28880
  },
  {
    "loss": 0.06471338272094726,
    "grad_norm": 0.5437493920326233,
    "learning_rate": 2.493534449868318e-06,
    "epoch": 0.9290285236517992,
    "step": 28890
  },
  {
    "loss": 0.07316511273384094,
    "grad_norm": 0.5847581028938293,
    "learning_rate": 2.4710925088261115e-06,
    "epoch": 0.9293500980802006,
    "step": 28900
  },
  {
    "loss": 0.053254157304763794,
    "grad_norm": 0.49556058645248413,
    "learning_rate": 2.448750750716633e-06,
    "epoch": 0.9296716725086022,
    "step": 28910
  },
  {
    "loss": 0.05694871544837952,
    "grad_norm": 0.47599726915359497,
    "learning_rate": 2.4265091984896237e-06,
    "epoch": 0.9299932469370036,
    "step": 28920
  },
  {
    "loss": 0.06631427407264709,
    "grad_norm": 0.9774466753005981,
    "learning_rate": 2.404367874991842e-06,
    "epoch": 0.930314821365405,
    "step": 28930
  },
  {
    "loss": 0.05324622988700867,
    "grad_norm": 0.4933931827545166,
    "learning_rate": 2.3823268029672054e-06,
    "epoch": 0.9306363957938065,
    "step": 28940
  },
  {
    "loss": 0.06808727383613586,
    "grad_norm": 0.4201553165912628,
    "learning_rate": 2.3603860050565473e-06,
    "epoch": 0.9309579702222079,
    "step": 28950
  },
  {
    "loss": 0.07126197814941407,
    "grad_norm": 0.49235799908638,
    "learning_rate": 2.3385455037977176e-06,
    "epoch": 0.9312795446506094,
    "step": 28960
  },
  {
    "loss": 0.07900972366333008,
    "grad_norm": 0.6312693357467651,
    "learning_rate": 2.3168053216255924e-06,
    "epoch": 0.9316011190790109,
    "step": 28970
  },
  {
    "loss": 0.06725811958312988,
    "grad_norm": 0.5075303316116333,
    "learning_rate": 2.2951654808719414e-06,
    "epoch": 0.9319226935074123,
    "step": 28980
  },
  {
    "loss": 0.06910203099250793,
    "grad_norm": 0.4444521367549896,
    "learning_rate": 2.2736260037654943e-06,
    "epoch": 0.9322442679358137,
    "step": 28990
  },
  {
    "loss": 0.08351538777351379,
    "grad_norm": 0.839489758014679,
    "learning_rate": 2.252186912431864e-06,
    "epoch": 0.9325658423642152,
    "step": 29000
  },
  {
    "loss": 0.07000740766525268,
    "grad_norm": 0.7335992455482483,
    "learning_rate": 2.230848228893567e-06,
    "epoch": 0.9328874167926167,
    "step": 29010
  },
  {
    "loss": 0.05521320104598999,
    "grad_norm": 0.4982221722602844,
    "learning_rate": 2.2096099750699707e-06,
    "epoch": 0.9332089912210181,
    "step": 29020
  },
  {
    "loss": 0.07880055904388428,
    "grad_norm": 0.947349488735199,
    "learning_rate": 2.1884721727772672e-06,
    "epoch": 0.9335305656494196,
    "step": 29030
  },
  {
    "loss": 0.07375386357307434,
    "grad_norm": 0.7958569526672363,
    "learning_rate": 2.1674348437284997e-06,
    "epoch": 0.933852140077821,
    "step": 29040
  },
  {
    "loss": 0.06184510588645935,
    "grad_norm": 0.5542522668838501,
    "learning_rate": 2.1464980095334486e-06,
    "epoch": 0.9341737145062224,
    "step": 29050
  },
  {
    "loss": 0.07052304148674012,
    "grad_norm": 0.7163567543029785,
    "learning_rate": 2.1256616916987216e-06,
    "epoch": 0.934495288934624,
    "step": 29060
  },
  {
    "loss": 0.06895366907119752,
    "grad_norm": 0.6882383227348328,
    "learning_rate": 2.104925911627631e-06,
    "epoch": 0.9348168633630254,
    "step": 29070
  },
  {
    "loss": 0.07813504934310914,
    "grad_norm": 0.6083703637123108,
    "learning_rate": 2.0842906906202496e-06,
    "epoch": 0.9351384377914268,
    "step": 29080
  },
  {
    "loss": 0.06795021295547485,
    "grad_norm": 0.7039796710014343,
    "learning_rate": 2.063756049873333e-06,
    "epoch": 0.9354600122198283,
    "step": 29090
  },
  {
    "loss": 0.07217402458190918,
    "grad_norm": 0.7208359837532043,
    "learning_rate": 2.0433220104803088e-06,
    "epoch": 0.9357815866482297,
    "step": 29100
  },
  {
    "loss": 0.0501723051071167,
    "grad_norm": 0.7383726239204407,
    "learning_rate": 2.0229885934313074e-06,
    "epoch": 0.9361031610766312,
    "step": 29110
  },
  {
    "loss": 0.07170373201370239,
    "grad_norm": 0.7307721376419067,
    "learning_rate": 2.002755819613056e-06,
    "epoch": 0.9364247355050327,
    "step": 29120
  },
  {
    "loss": 0.07169081568717957,
    "grad_norm": 0.4697757065296173,
    "learning_rate": 1.982623709808917e-06,
    "epoch": 0.9367463099334341,
    "step": 29130
  },
  {
    "loss": 0.05431884527206421,
    "grad_norm": 0.6256595253944397,
    "learning_rate": 1.962592284698872e-06,
    "epoch": 0.9370678843618355,
    "step": 29140
  },
  {
    "loss": 0.06069269776344299,
    "grad_norm": 0.49124467372894287,
    "learning_rate": 1.942661564859427e-06,
    "epoch": 0.937389458790237,
    "step": 29150
  },
  {
    "loss": 0.05852615833282471,
    "grad_norm": 0.8317437767982483,
    "learning_rate": 1.9228315707636836e-06,
    "epoch": 0.9377110332186385,
    "step": 29160
  },
  {
    "loss": 0.0563937246799469,
    "grad_norm": 0.5100141763687134,
    "learning_rate": 1.9031023227812695e-06,
    "epoch": 0.9380326076470399,
    "step": 29170
  },
  {
    "loss": 0.07683940529823304,
    "grad_norm": 0.6290521621704102,
    "learning_rate": 1.883473841178296e-06,
    "epoch": 0.9383541820754414,
    "step": 29180
  },
  {
    "loss": 0.08868919610977173,
    "grad_norm": 0.6542617082595825,
    "learning_rate": 1.8639461461174013e-06,
    "epoch": 0.9386757565038428,
    "step": 29190
  },
  {
    "loss": 0.07720979452133178,
    "grad_norm": 0.48346221446990967,
    "learning_rate": 1.8445192576576952e-06,
    "epoch": 0.9389973309322442,
    "step": 29200
  },
  {
    "loss": 0.08283687829971313,
    "grad_norm": 0.9130428433418274,
    "learning_rate": 1.8251931957547154e-06,
    "epoch": 0.9393189053606458,
    "step": 29210
  },
  {
    "loss": 0.06735263466835022,
    "grad_norm": 0.24367094039916992,
    "learning_rate": 1.8059679802604368e-06,
    "epoch": 0.9396404797890472,
    "step": 29220
  },
  {
    "loss": 0.0744469404220581,
    "grad_norm": 0.5839953422546387,
    "learning_rate": 1.7868436309232296e-06,
    "epoch": 0.9399620542174486,
    "step": 29230
  },
  {
    "loss": 0.06536500453948975,
    "grad_norm": 0.5514238476753235,
    "learning_rate": 1.7678201673879013e-06,
    "epoch": 0.9402836286458501,
    "step": 29240
  },
  {
    "loss": 0.058148926496505736,
    "grad_norm": 0.4055790603160858,
    "learning_rate": 1.7488976091955545e-06,
    "epoch": 0.9406052030742516,
    "step": 29250
  },
  {
    "loss": 0.07389817237854004,
    "grad_norm": 0.4462706446647644,
    "learning_rate": 1.7300759757837182e-06,
    "epoch": 0.940926777502653,
    "step": 29260
  },
  {
    "loss": 0.08689826726913452,
    "grad_norm": 0.8113035559654236,
    "learning_rate": 1.7113552864862048e-06,
    "epoch": 0.9412483519310545,
    "step": 29270
  },
  {
    "loss": 0.049148404598236085,
    "grad_norm": 0.3998713493347168,
    "learning_rate": 1.6927355605331318e-06,
    "epoch": 0.9415699263594559,
    "step": 29280
  },
  {
    "loss": 0.06236416697502136,
    "grad_norm": 0.6018595099449158,
    "learning_rate": 1.674216817050933e-06,
    "epoch": 0.9418915007878573,
    "step": 29290
  },
  {
    "loss": 0.06123303771018982,
    "grad_norm": 0.4179343283176422,
    "learning_rate": 1.6557990750622699e-06,
    "epoch": 0.9422130752162589,
    "step": 29300
  },
  {
    "loss": 0.058761775493621826,
    "grad_norm": 0.5038360357284546,
    "learning_rate": 1.6374823534861084e-06,
    "epoch": 0.9425346496446603,
    "step": 29310
  },
  {
    "loss": 0.06190430521965027,
    "grad_norm": 0.3785480558872223,
    "learning_rate": 1.6192666711376204e-06,
    "epoch": 0.9428562240730617,
    "step": 29320
  },
  {
    "loss": 0.07154954075813294,
    "grad_norm": 0.6645725965499878,
    "learning_rate": 1.6011520467281605e-06,
    "epoch": 0.9431777985014632,
    "step": 29330
  },
  {
    "loss": 0.08225449323654174,
    "grad_norm": 0.5627620816230774,
    "learning_rate": 1.5831384988653331e-06,
    "epoch": 0.9434993729298646,
    "step": 29340
  },
  {
    "loss": 0.07556136846542358,
    "grad_norm": 0.3942989408969879,
    "learning_rate": 1.5652260460528478e-06,
    "epoch": 0.943820947358266,
    "step": 29350
  },
  {
    "loss": 0.06251301765441894,
    "grad_norm": 0.4809018075466156,
    "learning_rate": 1.5474147066906308e-06,
    "epoch": 0.9441425217866676,
    "step": 29360
  },
  {
    "loss": 0.06617502570152282,
    "grad_norm": 0.40315186977386475,
    "learning_rate": 1.5297044990747022e-06,
    "epoch": 0.944464096215069,
    "step": 29370
  },
  {
    "loss": 0.07782045006752014,
    "grad_norm": 0.8132786154747009,
    "learning_rate": 1.512095441397221e-06,
    "epoch": 0.9447856706434704,
    "step": 29380
  },
  {
    "loss": 0.07609405517578124,
    "grad_norm": 0.42774105072021484,
    "learning_rate": 1.4945875517464293e-06,
    "epoch": 0.9451072450718719,
    "step": 29390
  },
  {
    "loss": 0.05179577469825745,
    "grad_norm": 0.5319843292236328,
    "learning_rate": 1.4771808481066518e-06,
    "epoch": 0.9454288195002734,
    "step": 29400
  },
  {
    "loss": 0.07737534046173096,
    "grad_norm": 0.7647845149040222,
    "learning_rate": 1.4598753483582972e-06,
    "epoch": 0.9457503939286748,
    "step": 29410
  },
  {
    "loss": 0.05944972634315491,
    "grad_norm": 0.4028695523738861,
    "learning_rate": 1.4426710702777568e-06,
    "epoch": 0.9460719683570763,
    "step": 29420
  },
  {
    "loss": 0.07734155058860778,
    "grad_norm": 0.9073440432548523,
    "learning_rate": 1.4255680315375164e-06,
    "epoch": 0.9463935427854777,
    "step": 29430
  },
  {
    "loss": 0.06692083477973938,
    "grad_norm": 0.7518401741981506,
    "learning_rate": 1.4085662497060115e-06,
    "epoch": 0.9467151172138791,
    "step": 29440
  },
  {
    "loss": 0.0926957905292511,
    "grad_norm": 0.5793595910072327,
    "learning_rate": 1.3916657422476942e-06,
    "epoch": 0.9470366916422807,
    "step": 29450
  },
  {
    "loss": 0.07510214447975158,
    "grad_norm": 0.776908278465271,
    "learning_rate": 1.374866526523e-06,
    "epoch": 0.9473582660706821,
    "step": 29460
  },
  {
    "loss": 0.07716012001037598,
    "grad_norm": 0.8807123899459839,
    "learning_rate": 1.3581686197882694e-06,
    "epoch": 0.9476798404990835,
    "step": 29470
  },
  {
    "loss": 0.06747672557830811,
    "grad_norm": 0.4527372419834137,
    "learning_rate": 1.3415720391958265e-06,
    "epoch": 0.948001414927485,
    "step": 29480
  },
  {
    "loss": 0.08032204508781433,
    "grad_norm": 0.7324082851409912,
    "learning_rate": 1.3250768017938787e-06,
    "epoch": 0.9483229893558864,
    "step": 29490
  },
  {
    "loss": 0.06460022926330566,
    "grad_norm": 0.6987818479537964,
    "learning_rate": 1.3086829245265387e-06,
    "epoch": 0.9486445637842879,
    "step": 29500
  },
  {
    "loss": 0.06491804122924805,
    "grad_norm": 0.7714876532554626,
    "learning_rate": 1.2923904242338136e-06,
    "epoch": 0.9489661382126894,
    "step": 29510
  },
  {
    "loss": 0.07516576647758484,
    "grad_norm": 0.6882345080375671,
    "learning_rate": 1.2761993176515607e-06,
    "epoch": 0.9492877126410908,
    "step": 29520
  },
  {
    "eval_loss": 0.06555379927158356,
    "eval_runtime": 34.9932,
    "eval_samples_per_second": 143.628,
    "eval_steps_per_second": 35.921,
    "epoch": 0.9494806572981317,
    "step": 29526
  },
  {
    "loss": 0.08071179986000061,
    "grad_norm": 1.0186065435409546,
    "learning_rate": 1.2601096214114982e-06,
    "epoch": 0.9496092870694922,
    "step": 29530
  },
  {
    "loss": 0.07067824602127075,
    "grad_norm": 0.6346520781517029,
    "learning_rate": 1.2441213520411722e-06,
    "epoch": 0.9499308614978937,
    "step": 29540
  },
  {
    "loss": 0.08398991823196411,
    "grad_norm": 0.7575256824493408,
    "learning_rate": 1.2282345259639116e-06,
    "epoch": 0.9502524359262952,
    "step": 29550
  },
  {
    "loss": 0.06456127166748046,
    "grad_norm": 0.5739431977272034,
    "learning_rate": 1.2124491594988852e-06,
    "epoch": 0.9505740103546966,
    "step": 29560
  },
  {
    "loss": 0.07740659713745117,
    "grad_norm": 0.42999500036239624,
    "learning_rate": 1.1967652688610108e-06,
    "epoch": 0.9508955847830981,
    "step": 29570
  },
  {
    "loss": 0.07819628715515137,
    "grad_norm": 0.38845735788345337,
    "learning_rate": 1.1811828701609906e-06,
    "epoch": 0.9512171592114995,
    "step": 29580
  },
  {
    "loss": 0.0833408534526825,
    "grad_norm": 1.2384730577468872,
    "learning_rate": 1.1657019794052648e-06,
    "epoch": 0.9515387336399009,
    "step": 29590
  },
  {
    "loss": 0.07756343483924866,
    "grad_norm": 0.5256941914558411,
    "learning_rate": 1.150322612495991e-06,
    "epoch": 0.9518603080683025,
    "step": 29600
  },
  {
    "loss": 0.069635409116745,
    "grad_norm": 0.5696307420730591,
    "learning_rate": 1.1350447852310541e-06,
    "epoch": 0.9521818824967039,
    "step": 29610
  },
  {
    "loss": 0.06680102348327636,
    "grad_norm": 0.6343799829483032,
    "learning_rate": 1.1198685133040343e-06,
    "epoch": 0.9525034569251053,
    "step": 29620
  },
  {
    "loss": 0.060926932096481326,
    "grad_norm": 0.2505302131175995,
    "learning_rate": 1.1047938123041835e-06,
    "epoch": 0.9528250313535068,
    "step": 29630
  },
  {
    "loss": 0.05705304145812988,
    "grad_norm": 0.28949621319770813,
    "learning_rate": 1.0898206977164371e-06,
    "epoch": 0.9531466057819082,
    "step": 29640
  },
  {
    "loss": 0.05276017785072327,
    "grad_norm": 0.5729100704193115,
    "learning_rate": 1.0749491849213588e-06,
    "epoch": 0.9534681802103097,
    "step": 29650
  },
  {
    "loss": 0.07277541160583496,
    "grad_norm": 0.47938433289527893,
    "learning_rate": 1.0601792891951623e-06,
    "epoch": 0.9537897546387112,
    "step": 29660
  },
  {
    "loss": 0.0780954122543335,
    "grad_norm": 0.9656330943107605,
    "learning_rate": 1.0455110257096667e-06,
    "epoch": 0.9541113290671126,
    "step": 29670
  },
  {
    "loss": 0.08907517194747924,
    "grad_norm": 0.9449646472930908,
    "learning_rate": 1.0309444095323083e-06,
    "epoch": 0.954432903495514,
    "step": 29680
  },
  {
    "loss": 0.06952834129333496,
    "grad_norm": 0.4525853991508484,
    "learning_rate": 1.0164794556260849e-06,
    "epoch": 0.9547544779239155,
    "step": 29690
  },
  {
    "loss": 0.0694973349571228,
    "grad_norm": 0.48509544134140015,
    "learning_rate": 1.0021161788495882e-06,
    "epoch": 0.955076052352317,
    "step": 29700
  },
  {
    "loss": 0.08033158779144287,
    "grad_norm": 1.0671344995498657,
    "learning_rate": 9.878545939569717e-07,
    "epoch": 0.9553976267807184,
    "step": 29710
  },
  {
    "loss": 0.06861055493354798,
    "grad_norm": 0.259791761636734,
    "learning_rate": 9.736947155978948e-07,
    "epoch": 0.9557192012091199,
    "step": 29720
  },
  {
    "loss": 0.06976649165153503,
    "grad_norm": 0.44905713200569153,
    "learning_rate": 9.596365583175782e-07,
    "epoch": 0.9560407756375213,
    "step": 29730
  },
  {
    "loss": 0.058680230379104616,
    "grad_norm": 0.6654853224754333,
    "learning_rate": 9.456801365567258e-07,
    "epoch": 0.9563623500659227,
    "step": 29740
  },
  {
    "loss": 0.0554412305355072,
    "grad_norm": 3.4719386100769043,
    "learning_rate": 9.318254646515811e-07,
    "epoch": 0.9566839244943242,
    "step": 29750
  },
  {
    "loss": 0.08116706609725952,
    "grad_norm": 0.8351684212684631,
    "learning_rate": 9.180725568338044e-07,
    "epoch": 0.9570054989227257,
    "step": 29760
  },
  {
    "loss": 0.055116087198257446,
    "grad_norm": 1.078339695930481,
    "learning_rate": 9.044214272305728e-07,
    "epoch": 0.9573270733511271,
    "step": 29770
  },
  {
    "loss": 0.06081744432449341,
    "grad_norm": 0.408652663230896,
    "learning_rate": 8.908720898645029e-07,
    "epoch": 0.9576486477795285,
    "step": 29780
  },
  {
    "loss": 0.0641602635383606,
    "grad_norm": 0.4921199083328247,
    "learning_rate": 8.774245586536389e-07,
    "epoch": 0.95797022220793,
    "step": 29790
  },
  {
    "loss": 0.07612771391868592,
    "grad_norm": 0.7166224122047424,
    "learning_rate": 8.640788474114647e-07,
    "epoch": 0.9582917966363315,
    "step": 29800
  },
  {
    "loss": 0.06867678165435791,
    "grad_norm": 0.6871895790100098,
    "learning_rate": 8.508349698468588e-07,
    "epoch": 0.9586133710647329,
    "step": 29810
  },
  {
    "loss": 0.060842263698577884,
    "grad_norm": 0.6564211845397949,
    "learning_rate": 8.376929395640831e-07,
    "epoch": 0.9589349454931344,
    "step": 29820
  },
  {
    "loss": 0.06288072466850281,
    "grad_norm": 0.863722026348114,
    "learning_rate": 8.246527700628393e-07,
    "epoch": 0.9592565199215358,
    "step": 29830
  },
  {
    "loss": 0.0851693332195282,
    "grad_norm": 0.7785186767578125,
    "learning_rate": 8.117144747381233e-07,
    "epoch": 0.9595780943499372,
    "step": 29840
  },
  {
    "loss": 0.04293628633022308,
    "grad_norm": 0.3203319311141968,
    "learning_rate": 7.988780668803153e-07,
    "epoch": 0.9598996687783388,
    "step": 29850
  },
  {
    "loss": 0.07816066145896912,
    "grad_norm": 0.9088801145553589,
    "learning_rate": 7.861435596751676e-07,
    "epoch": 0.9602212432067402,
    "step": 29860
  },
  {
    "loss": 0.06818323731422424,
    "grad_norm": 0.649931788444519,
    "learning_rate": 7.735109662036832e-07,
    "epoch": 0.9605428176351416,
    "step": 29870
  },
  {
    "loss": 0.08106167912483216,
    "grad_norm": 0.210426926612854,
    "learning_rate": 7.609802994422599e-07,
    "epoch": 0.9608643920635431,
    "step": 29880
  },
  {
    "loss": 0.06616997718811035,
    "grad_norm": 0.4277474880218506,
    "learning_rate": 7.485515722625347e-07,
    "epoch": 0.9611859664919445,
    "step": 29890
  },
  {
    "loss": 0.05833558440208435,
    "grad_norm": 0.4171481430530548,
    "learning_rate": 7.362247974314617e-07,
    "epoch": 0.961507540920346,
    "step": 29900
  },
  {
    "loss": 0.06775500178337097,
    "grad_norm": 0.5039193630218506,
    "learning_rate": 7.23999987611279e-07,
    "epoch": 0.9618291153487475,
    "step": 29910
  },
  {
    "loss": 0.06911481022834778,
    "grad_norm": 0.5150966048240662,
    "learning_rate": 7.118771553594305e-07,
    "epoch": 0.9621506897771489,
    "step": 29920
  },
  {
    "loss": 0.05216036438941955,
    "grad_norm": 0.5649464726448059,
    "learning_rate": 6.998563131286662e-07,
    "epoch": 0.9624722642055503,
    "step": 29930
  },
  {
    "loss": 0.0770330011844635,
    "grad_norm": 0.4386029541492462,
    "learning_rate": 6.87937473266953e-07,
    "epoch": 0.9627938386339518,
    "step": 29940
  },
  {
    "loss": 0.06624605655670165,
    "grad_norm": 0.4605594575405121,
    "learning_rate": 6.761206480174531e-07,
    "epoch": 0.9631154130623533,
    "step": 29950
  },
  {
    "loss": 0.06516779661178589,
    "grad_norm": 0.6219736337661743,
    "learning_rate": 6.64405849518579e-07,
    "epoch": 0.9634369874907547,
    "step": 29960
  },
  {
    "loss": 0.06415200829505921,
    "grad_norm": 0.317819207906723,
    "learning_rate": 6.527930898039047e-07,
    "epoch": 0.9637585619191562,
    "step": 29970
  },
  {
    "loss": 0.06802978515625,
    "grad_norm": 0.6788613200187683,
    "learning_rate": 6.412823808022328e-07,
    "epoch": 0.9640801363475576,
    "step": 29980
  },
  {
    "loss": 0.056847357749938966,
    "grad_norm": 0.5521454215049744,
    "learning_rate": 6.298737343374717e-07,
    "epoch": 0.964401710775959,
    "step": 29990
  },
  {
    "loss": 0.06040582060813904,
    "grad_norm": 0.637925386428833,
    "learning_rate": 6.185671621287581e-07,
    "epoch": 0.9647232852043606,
    "step": 30000
  },
  {
    "loss": 0.0742426872253418,
    "grad_norm": 0.5535704493522644,
    "learning_rate": 6.073626757903239e-07,
    "epoch": 0.965044859632762,
    "step": 30010
  },
  {
    "loss": 0.06893228888511657,
    "grad_norm": 0.6192355751991272,
    "learning_rate": 5.962602868315737e-07,
    "epoch": 0.9653664340611634,
    "step": 30020
  },
  {
    "loss": 0.06794821619987487,
    "grad_norm": 0.6708097457885742,
    "learning_rate": 5.852600066570291e-07,
    "epoch": 0.9656880084895649,
    "step": 30030
  },
  {
    "loss": 0.0759381353855133,
    "grad_norm": 0.4879767596721649,
    "learning_rate": 5.743618465662959e-07,
    "epoch": 0.9660095829179663,
    "step": 30040
  },
  {
    "loss": 0.074285888671875,
    "grad_norm": 0.5703880190849304,
    "learning_rate": 5.635658177541192e-07,
    "epoch": 0.9663311573463678,
    "step": 30050
  },
  {
    "loss": 0.06602758765220643,
    "grad_norm": 0.5126018524169922,
    "learning_rate": 5.52871931310317e-07,
    "epoch": 0.9666527317747693,
    "step": 30060
  },
  {
    "loss": 0.04887774586677551,
    "grad_norm": 0.6106338500976562,
    "learning_rate": 5.4228019821978e-07,
    "epoch": 0.9669743062031707,
    "step": 30070
  },
  {
    "loss": 0.08942516446113587,
    "grad_norm": 0.6533125638961792,
    "learning_rate": 5.317906293624719e-07,
    "epoch": 0.9672958806315721,
    "step": 30080
  },
  {
    "loss": 0.059960460662841795,
    "grad_norm": 0.5349124073982239,
    "learning_rate": 5.214032355134069e-07,
    "epoch": 0.9676174550599737,
    "step": 30090
  },
  {
    "loss": 0.06590099334716797,
    "grad_norm": 0.6322242021560669,
    "learning_rate": 5.111180273426497e-07,
    "epoch": 0.9679390294883751,
    "step": 30100
  },
  {
    "loss": 0.08278042078018188,
    "grad_norm": 0.5157520174980164,
    "learning_rate": 5.009350154153158e-07,
    "epoch": 0.9682606039167765,
    "step": 30110
  },
  {
    "loss": 0.08131775259971619,
    "grad_norm": 0.7335174679756165,
    "learning_rate": 4.908542101914937e-07,
    "epoch": 0.968582178345178,
    "step": 30120
  },
  {
    "loss": 0.05691747665405274,
    "grad_norm": 0.6210181713104248,
    "learning_rate": 4.808756220263333e-07,
    "epoch": 0.9689037527735794,
    "step": 30130
  },
  {
    "loss": 0.08510000705718994,
    "grad_norm": 0.43013817071914673,
    "learning_rate": 4.709992611699465e-07,
    "epoch": 0.9692253272019808,
    "step": 30140
  },
  {
    "loss": 0.079147469997406,
    "grad_norm": 0.47200024127960205,
    "learning_rate": 4.6122513776749586e-07,
    "epoch": 0.9695469016303824,
    "step": 30150
  },
  {
    "loss": 0.0846597671508789,
    "grad_norm": 1.025412678718567,
    "learning_rate": 4.5155326185905014e-07,
    "epoch": 0.9698684760587838,
    "step": 30160
  },
  {
    "loss": 0.06067612171173096,
    "grad_norm": 0.48970702290534973,
    "learning_rate": 4.4198364337969535e-07,
    "epoch": 0.9701900504871852,
    "step": 30170
  },
  {
    "loss": 0.05486201643943787,
    "grad_norm": 0.8192650675773621,
    "learning_rate": 4.325162921594683e-07,
    "epoch": 0.9705116249155867,
    "step": 30180
  },
  {
    "loss": 0.0711601734161377,
    "grad_norm": 0.49607720971107483,
    "learning_rate": 4.2315121792335654e-07,
    "epoch": 0.9708331993439882,
    "step": 30190
  },
  {
    "loss": 0.05947260856628418,
    "grad_norm": 0.5873615741729736,
    "learning_rate": 4.138884302912649e-07,
    "epoch": 0.9711547737723896,
    "step": 30200
  },
  {
    "loss": 0.06383942365646363,
    "grad_norm": 0.4902679920196533,
    "learning_rate": 4.0472793877806004e-07,
    "epoch": 0.9714763482007911,
    "step": 30210
  },
  {
    "loss": 0.06064661741256714,
    "grad_norm": 0.3690512478351593,
    "learning_rate": 3.9566975279350385e-07,
    "epoch": 0.9717979226291925,
    "step": 30220
  },
  {
    "loss": 0.05835795998573303,
    "grad_norm": 0.7086575031280518,
    "learning_rate": 3.867138816422977e-07,
    "epoch": 0.9721194970575939,
    "step": 30230
  },
  {
    "loss": 0.07011948823928833,
    "grad_norm": 0.5723183751106262,
    "learning_rate": 3.7786033452401616e-07,
    "epoch": 0.9724410714859955,
    "step": 30240
  },
  {
    "loss": 0.08009362816810608,
    "grad_norm": 0.8066279292106628,
    "learning_rate": 3.6910912053313983e-07,
    "epoch": 0.9727626459143969,
    "step": 30250
  },
  {
    "loss": 0.0474946528673172,
    "grad_norm": 0.796147346496582,
    "learning_rate": 3.604602486590114e-07,
    "epoch": 0.9730842203427983,
    "step": 30260
  },
  {
    "loss": 0.08515186905860901,
    "grad_norm": 0.6759833097457886,
    "learning_rate": 3.519137277858797e-07,
    "epoch": 0.9734057947711998,
    "step": 30270
  },
  {
    "loss": 0.06661058664321899,
    "grad_norm": 0.7066734433174133,
    "learning_rate": 3.434695666928445e-07,
    "epoch": 0.9737273691996012,
    "step": 30280
  },
  {
    "loss": 0.06675366759300232,
    "grad_norm": 0.6176568269729614,
    "learning_rate": 3.351277740538339e-07,
    "epoch": 0.9740489436280027,
    "step": 30290
  },
  {
    "loss": 0.07360825538635254,
    "grad_norm": 0.7115502953529358,
    "learning_rate": 3.268883584376603e-07,
    "epoch": 0.9743705180564042,
    "step": 30300
  },
  {
    "loss": 0.05721208453178406,
    "grad_norm": 0.8995148539543152,
    "learning_rate": 3.187513283079424e-07,
    "epoch": 0.9746920924848056,
    "step": 30310
  },
  {
    "loss": 0.0957691490650177,
    "grad_norm": 1.0181430578231812,
    "learning_rate": 3.1071669202314966e-07,
    "epoch": 0.975013666913207,
    "step": 30320
  },
  {
    "loss": 0.06637667417526245,
    "grad_norm": 0.7077873349189758,
    "learning_rate": 3.0278445783655794e-07,
    "epoch": 0.9753352413416085,
    "step": 30330
  },
  {
    "loss": 0.08069146275520325,
    "grad_norm": 0.5280202031135559,
    "learning_rate": 2.9495463389627166e-07,
    "epoch": 0.97565681577001,
    "step": 30340
  },
  {
    "loss": 0.06518712043762206,
    "grad_norm": 0.6744681000709534,
    "learning_rate": 2.8722722824516825e-07,
    "epoch": 0.9759783901984114,
    "step": 30350
  },
  {
    "loss": 0.07043516635894775,
    "grad_norm": 0.6811769604682922,
    "learning_rate": 2.796022488209427e-07,
    "epoch": 0.9762999646268129,
    "step": 30360
  },
  {
    "loss": 0.08094441890716553,
    "grad_norm": 0.4031400978565216,
    "learning_rate": 2.720797034560851e-07,
    "epoch": 0.9766215390552143,
    "step": 30370
  },
  {
    "loss": 0.07110047340393066,
    "grad_norm": 0.7690785527229309,
    "learning_rate": 2.6465959987782563e-07,
    "epoch": 0.9769431134836157,
    "step": 30380
  },
  {
    "loss": 0.08905300498008728,
    "grad_norm": 0.5113458633422852,
    "learning_rate": 2.5734194570820045e-07,
    "epoch": 0.9772646879120173,
    "step": 30390
  },
  {
    "loss": 0.05021012425422668,
    "grad_norm": 0.4673030376434326,
    "learning_rate": 2.5012674846399685e-07,
    "epoch": 0.9775862623404187,
    "step": 30400
  },
  {
    "loss": 0.0810099720954895,
    "grad_norm": 0.7750974297523499,
    "learning_rate": 2.43014015556764e-07,
    "epoch": 0.9779078367688201,
    "step": 30410
  },
  {
    "loss": 0.06320219039916992,
    "grad_norm": 0.5255265235900879,
    "learning_rate": 2.3600375429277954e-07,
    "epoch": 0.9782294111972216,
    "step": 30420
  },
  {
    "loss": 0.08997372388839722,
    "grad_norm": 0.6714896559715271,
    "learning_rate": 2.290959718730834e-07,
    "epoch": 0.978550985625623,
    "step": 30430
  },
  {
    "loss": 0.06944092512130737,
    "grad_norm": 0.5551565885543823,
    "learning_rate": 2.2229067539343286e-07,
    "epoch": 0.9788725600540245,
    "step": 30440
  },
  {
    "loss": 0.0534528911113739,
    "grad_norm": 0.5738435983657837,
    "learning_rate": 2.1558787184430274e-07,
    "epoch": 0.979194134482426,
    "step": 30450
  },
  {
    "loss": 0.06853411197662354,
    "grad_norm": 0.3571486473083496,
    "learning_rate": 2.089875681109188e-07,
    "epoch": 0.9795157089108274,
    "step": 30460
  },
  {
    "loss": 0.07802314162254334,
    "grad_norm": 0.6826774477958679,
    "learning_rate": 2.024897709731799e-07,
    "epoch": 0.9798372833392288,
    "step": 30470
  },
  {
    "loss": 0.07004804015159607,
    "grad_norm": 0.29291650652885437,
    "learning_rate": 1.9609448710572464e-07,
    "epoch": 0.9801588577676303,
    "step": 30480
  },
  {
    "loss": 0.06426869630813599,
    "grad_norm": 0.7378115653991699,
    "learning_rate": 1.898017230778537e-07,
    "epoch": 0.9804804321960318,
    "step": 30490
  },
  {
    "loss": 0.08708415627479553,
    "grad_norm": 0.951000452041626,
    "learning_rate": 1.8361148535357419e-07,
    "epoch": 0.9808020066244332,
    "step": 30500
  },
  {
    "loss": 0.0709414005279541,
    "grad_norm": 0.2556989789009094,
    "learning_rate": 1.7752378029157745e-07,
    "epoch": 0.9811235810528347,
    "step": 30510
  },
  {
    "loss": 0.07617509365081787,
    "grad_norm": 0.4664788842201233,
    "learning_rate": 1.7153861414525018e-07,
    "epoch": 0.9814451554812361,
    "step": 30520
  },
  {
    "loss": 0.06983622908592224,
    "grad_norm": 0.6659867167472839,
    "learning_rate": 1.6565599306260782e-07,
    "epoch": 0.9817667299096375,
    "step": 30530
  },
  {
    "loss": 0.08357095718383789,
    "grad_norm": 0.6239607334136963,
    "learning_rate": 1.5987592308636113e-07,
    "epoch": 0.9820883043380391,
    "step": 30540
  },
  {
    "loss": 0.06588581800460816,
    "grad_norm": 0.43989256024360657,
    "learning_rate": 1.5419841015388293e-07,
    "epoch": 0.9824098787664405,
    "step": 30550
  },
  {
    "loss": 0.07492035031318664,
    "grad_norm": 0.6496413350105286,
    "learning_rate": 1.4862346009716365e-07,
    "epoch": 0.9827314531948419,
    "step": 30560
  },
  {
    "loss": 0.06939973831176757,
    "grad_norm": 0.5781214237213135,
    "learning_rate": 1.4315107864287802e-07,
    "epoch": 0.9830530276232434,
    "step": 30570
  },
  {
    "loss": 0.07107747197151185,
    "grad_norm": 0.7525464296340942,
    "learning_rate": 1.3778127141232943e-07,
    "epoch": 0.9833746020516448,
    "step": 30580
  },
  {
    "loss": 0.06889268159866332,
    "grad_norm": 0.5934503674507141,
    "learning_rate": 1.3251404392143896e-07,
    "epoch": 0.9836961764800463,
    "step": 30590
  },
  {
    "loss": 0.08742887377738953,
    "grad_norm": 0.8118358850479126,
    "learning_rate": 1.2734940158078967e-07,
    "epoch": 0.9840177509084478,
    "step": 30600
  },
  {
    "loss": 0.058663409948348996,
    "grad_norm": 0.6564080119132996,
    "learning_rate": 1.222873496955379e-07,
    "epoch": 0.9843393253368492,
    "step": 30610
  },
  {
    "loss": 0.06013064980506897,
    "grad_norm": 0.6733872890472412,
    "learning_rate": 1.173278934655131e-07,
    "epoch": 0.9846608997652506,
    "step": 30620
  },
  {
    "loss": 0.07635838985443115,
    "grad_norm": 0.632681667804718,
    "learning_rate": 1.1247103798511793e-07,
    "epoch": 0.9849824741936521,
    "step": 30630
  },
  {
    "loss": 0.08892632722854614,
    "grad_norm": 0.9751865267753601,
    "learning_rate": 1.0771678824339493e-07,
    "epoch": 0.9853040486220536,
    "step": 30640
  },
  {
    "loss": 0.06905696392059327,
    "grad_norm": 0.4078354835510254,
    "learning_rate": 1.0306514912393761e-07,
    "epoch": 0.985625623050455,
    "step": 30650
  },
  {
    "loss": 0.06030969619750977,
    "grad_norm": 0.5289317965507507,
    "learning_rate": 9.851612540500155e-08,
    "epoch": 0.9859471974788565,
    "step": 30660
  },
  {
    "loss": 0.06162986159324646,
    "grad_norm": 0.6373181343078613,
    "learning_rate": 9.406972175938222e-08,
    "epoch": 0.9862687719072579,
    "step": 30670
  },
  {
    "loss": 0.06717531085014343,
    "grad_norm": 0.5047315359115601,
    "learning_rate": 8.972594275448165e-08,
    "epoch": 0.9865903463356593,
    "step": 30680
  },
  {
    "loss": 0.07401962280273437,
    "grad_norm": 0.7474780082702637,
    "learning_rate": 8.548479285229727e-08,
    "epoch": 0.9869119207640609,
    "step": 30690
  },
  {
    "loss": 0.07880670428276063,
    "grad_norm": 0.7355194687843323,
    "learning_rate": 8.134627640937754e-08,
    "epoch": 0.9872334951924623,
    "step": 30700
  },
  {
    "loss": 0.07137477397918701,
    "grad_norm": 0.6599552631378174,
    "learning_rate": 7.731039767687743e-08,
    "epoch": 0.9875550696208637,
    "step": 30710
  },
  {
    "loss": 0.10589978694915772,
    "grad_norm": 0.5183787941932678,
    "learning_rate": 7.337716080048074e-08,
    "epoch": 0.9878766440492652,
    "step": 30720
  },
  {
    "loss": 0.07905386686325074,
    "grad_norm": 0.7177006602287292,
    "learning_rate": 6.954656982047781e-08,
    "epoch": 0.9881982184776666,
    "step": 30730
  },
  {
    "loss": 0.05417203903198242,
    "grad_norm": 0.43448272347450256,
    "learning_rate": 6.581862867168776e-08,
    "epoch": 0.9885197929060681,
    "step": 30740
  },
  {
    "loss": 0.0657284438610077,
    "grad_norm": 0.6650566458702087,
    "learning_rate": 6.219334118351405e-08,
    "epoch": 0.9888413673344696,
    "step": 30750
  },
  {
    "loss": 0.06718289256095886,
    "grad_norm": 0.6515393257141113,
    "learning_rate": 5.8670711079877874e-08,
    "epoch": 0.989162941762871,
    "step": 30760
  },
  {
    "loss": 0.054975324869155885,
    "grad_norm": 0.3662674129009247,
    "learning_rate": 5.52507419792736e-08,
    "epoch": 0.9894845161912724,
    "step": 30770
  },
  {
    "loss": 0.06047446727752685,
    "grad_norm": 0.5469169616699219,
    "learning_rate": 5.193343739474665e-08,
    "epoch": 0.989806090619674,
    "step": 30780
  },
  {
    "loss": 0.05281834602355957,
    "grad_norm": 0.690144419670105,
    "learning_rate": 4.871880073388235e-08,
    "epoch": 0.9901276650480754,
    "step": 30790
  },
  {
    "loss": 0.07430117130279541,
    "grad_norm": 0.5861929059028625,
    "learning_rate": 4.560683529877263e-08,
    "epoch": 0.9904492394764768,
    "step": 30800
  },
  {
    "loss": 0.061764204502105714,
    "grad_norm": 0.6248287558555603,
    "learning_rate": 4.259754428608265e-08,
    "epoch": 0.9907708139048783,
    "step": 30810
  },
  {
    "loss": 0.06891185641288758,
    "grad_norm": 0.8176858425140381,
    "learning_rate": 3.9690930786995264e-08,
    "epoch": 0.9910923883332797,
    "step": 30820
  },
  {
    "loss": 0.06561875939369202,
    "grad_norm": 0.5709847211837769,
    "learning_rate": 3.6886997787211054e-08,
    "epoch": 0.9914139627616811,
    "step": 30830
  },
  {
    "loss": 0.07824975848197938,
    "grad_norm": 1.156491994857788,
    "learning_rate": 3.418574816697051e-08,
    "epoch": 0.9917355371900827,
    "step": 30840
  },
  {
    "loss": 0.08055688142776489,
    "grad_norm": 0.5808718800544739,
    "learning_rate": 3.158718470103183e-08,
    "epoch": 0.9920571116184841,
    "step": 30850
  },
  {
    "loss": 0.07981680631637574,
    "grad_norm": 0.45614954829216003,
    "learning_rate": 2.909131005868204e-08,
    "epoch": 0.9923786860468855,
    "step": 30860
  },
  {
    "loss": 0.06255224347114563,
    "grad_norm": 0.7705869674682617,
    "learning_rate": 2.6698126803703648e-08,
    "epoch": 0.992700260475287,
    "step": 30870
  },
  {
    "loss": 0.05859441757202148,
    "grad_norm": 0.5446732640266418,
    "learning_rate": 2.4407637394396886e-08,
    "epoch": 0.9930218349036884,
    "step": 30880
  },
  {
    "loss": 0.0605139434337616,
    "grad_norm": 0.8045651912689209,
    "learning_rate": 2.2219844183601902e-08,
    "epoch": 0.9933434093320899,
    "step": 30890
  },
  {
    "loss": 0.06261725425720215,
    "grad_norm": 0.4299694895744324,
    "learning_rate": 2.013474941864324e-08,
    "epoch": 0.9936649837604914,
    "step": 30900
  },
  {
    "loss": 0.07130892276763916,
    "grad_norm": 0.4894041419029236,
    "learning_rate": 1.815235524134096e-08,
    "epoch": 0.9939865581888928,
    "step": 30910
  },
  {
    "loss": 0.07848079800605774,
    "grad_norm": 0.8134866952896118,
    "learning_rate": 1.627266368804392e-08,
    "epoch": 0.9943081326172942,
    "step": 30920
  },
  {
    "loss": 0.07628424167633056,
    "grad_norm": 0.8397912383079529,
    "learning_rate": 1.4495676689607606e-08,
    "epoch": 0.9946297070456958,
    "step": 30930
  },
  {
    "loss": 0.07152285575866699,
    "grad_norm": 0.5438010096549988,
    "learning_rate": 1.2821396071360792e-08,
    "epoch": 0.9949512814740972,
    "step": 30940
  },
  {
    "loss": 0.0670372188091278,
    "grad_norm": 0.6954264640808105,
    "learning_rate": 1.1249823553149962e-08,
    "epoch": 0.9952728559024986,
    "step": 30950
  },
  {
    "loss": 0.06887378692626953,
    "grad_norm": 0.6697443127632141,
    "learning_rate": 9.780960749317114e-09,
    "epoch": 0.9955944303309001,
    "step": 30960
  },
  {
    "loss": 0.08694977760314941,
    "grad_norm": 0.4334903657436371,
    "learning_rate": 8.414809168688643e-09,
    "epoch": 0.9959160047593015,
    "step": 30970
  },
  {
    "loss": 0.08538188934326171,
    "grad_norm": 0.8963655829429626,
    "learning_rate": 7.151370214608654e-09,
    "epoch": 0.996237579187703,
    "step": 30980
  },
  {
    "loss": 0.04286316633224487,
    "grad_norm": 0.40512776374816895,
    "learning_rate": 5.990645184883459e-09,
    "epoch": 0.9965591536161045,
    "step": 30990
  },
  {
    "loss": 0.06450687050819397,
    "grad_norm": 0.2572462260723114,
    "learning_rate": 4.932635271825969e-09,
    "epoch": 0.9968807280445059,
    "step": 31000
  },
  {
    "loss": 0.05183506608009338,
    "grad_norm": 0.4054562747478485,
    "learning_rate": 3.977341562244608e-09,
    "epoch": 0.9972023024729073,
    "step": 31010
  },
  {
    "loss": 0.06269176006317138,
    "grad_norm": 0.5525853633880615,
    "learning_rate": 3.1247650374210957e-09,
    "epoch": 0.9975238769013088,
    "step": 31020
  },
  {
    "loss": 0.05774378776550293,
    "grad_norm": 0.5034267902374268,
    "learning_rate": 2.3749065731326627e-09,
    "epoch": 0.9978454513297103,
    "step": 31030
  },
  {
    "loss": 0.08453238010406494,
    "grad_norm": 0.7438255548477173,
    "learning_rate": 1.7277669396520425e-09,
    "epoch": 0.9981670257581117,
    "step": 31040
  },
  {
    "loss": 0.06554477214813233,
    "grad_norm": 0.40568795800209045,
    "learning_rate": 1.1833468017252714e-09,
    "epoch": 0.9984886001865132,
    "step": 31050
  },
  {
    "loss": 0.0662672460079193,
    "grad_norm": 0.3644300699234009,
    "learning_rate": 7.416467185827891e-10,
    "epoch": 0.9988101746149146,
    "step": 31060
  },
  {
    "loss": 0.0686276376247406,
    "grad_norm": 0.8694908618927002,
    "learning_rate": 4.026671439616436e-10,
    "epoch": 0.999131749043316,
    "step": 31070
  },
  {
    "loss": 0.07674849033355713,
    "grad_norm": 0.7753235697746277,
    "learning_rate": 1.6640842603887762e-10,
    "epoch": 0.9994533234717176,
    "step": 31080
  },
  {
    "eval_loss": 0.06542336940765381,
    "eval_runtime": 34.5075,
    "eval_samples_per_second": 145.649,
    "eval_steps_per_second": 36.427,
    "epoch": 0.9994533234717176,
    "step": 31080
  },
  {
    "loss": 0.08255035877227783,
    "grad_norm": 0.5661125183105469,
    "learning_rate": 3.2870807520346546e-11,
    "epoch": 0.999774897900119,
    "step": 31090
  },
  {
    "eval_loss": 0.06546882539987564,
    "eval_runtime": 34.5111,
    "eval_samples_per_second": 145.634,
    "eval_steps_per_second": 36.423,
    "epoch": 1.0,
    "step": 31097
  },
  {
    "train_runtime": 4570.1661,
    "train_samples_per_second": 108.867,
    "train_steps_per_second": 6.804,
    "total_flos": 1.6262875457184e+17,
    "train_loss": 0.09000451330583219,
    "epoch": 1.0,
    "step": 31097
  }
]