{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.07300375571942479,
  "eval_steps": 500,
  "global_step": 124000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.9436998273961606e-05,
      "grad_norm": 1.0310300588607788,
      "learning_rate": 0.0002999956727618906,
      "loss": 7.841340942382812,
      "step": 50
    },
    {
      "epoch": 5.887399654792321e-05,
      "grad_norm": 1.0565732717514038,
      "learning_rate": 0.0002999912572127994,
      "loss": 7.134805908203125,
      "step": 100
    },
    {
      "epoch": 8.831099482188482e-05,
      "grad_norm": 1.094490647315979,
      "learning_rate": 0.0002999868416637082,
      "loss": 6.7894677734375,
      "step": 150
    },
    {
      "epoch": 0.00011774799309584643,
      "grad_norm": 0.9034406542778015,
      "learning_rate": 0.000299982426114617,
      "loss": 6.520293579101563,
      "step": 200
    },
    {
      "epoch": 0.00014718499136980804,
      "grad_norm": 0.8733723759651184,
      "learning_rate": 0.00029997801056552584,
      "loss": 6.357908935546875,
      "step": 250
    },
    {
      "epoch": 0.00017662198964376965,
      "grad_norm": 0.9817787408828735,
      "learning_rate": 0.00029997359501643465,
      "loss": 6.31034912109375,
      "step": 300
    },
    {
      "epoch": 0.00020605898791773125,
      "grad_norm": 0.8127152919769287,
      "learning_rate": 0.00029996917946734347,
      "loss": 6.12135986328125,
      "step": 350
    },
    {
      "epoch": 0.00023549598619169285,
      "grad_norm": 0.9790456295013428,
      "learning_rate": 0.0002999647639182523,
      "loss": 6.075621337890625,
      "step": 400
    },
    {
      "epoch": 0.00026493298446565445,
      "grad_norm": 3.255697011947632,
      "learning_rate": 0.00029996034836916104,
      "loss": 6.001095581054687,
      "step": 450
    },
    {
      "epoch": 0.0002943699827396161,
      "grad_norm": 0.5592721700668335,
      "learning_rate": 0.00029995593282006986,
      "loss": 5.892413940429687,
      "step": 500
    },
    {
      "epoch": 0.00032380698101357766,
      "grad_norm": 0.8910121917724609,
      "learning_rate": 0.0002999515172709787,
      "loss": 5.842003173828125,
      "step": 550
    },
    {
      "epoch": 0.0003532439792875393,
      "grad_norm": 0.8119663596153259,
      "learning_rate": 0.0002999471017218875,
      "loss": 5.782672729492187,
      "step": 600
    },
    {
      "epoch": 0.00038268097756150087,
      "grad_norm": 0.7697049379348755,
      "learning_rate": 0.0002999426861727963,
      "loss": 5.751597900390625,
      "step": 650
    },
    {
      "epoch": 0.0004121179758354625,
      "grad_norm": 0.9113023281097412,
      "learning_rate": 0.0002999382706237051,
      "loss": 5.7005029296875,
      "step": 700
    },
    {
      "epoch": 0.00044155497410942407,
      "grad_norm": 0.7396610975265503,
      "learning_rate": 0.00029993385507461393,
      "loss": 5.608889770507813,
      "step": 750
    },
    {
      "epoch": 0.0004709919723833857,
      "grad_norm": 0.7552371025085449,
      "learning_rate": 0.00029992943952552274,
      "loss": 5.620023193359375,
      "step": 800
    },
    {
      "epoch": 0.0005004289706573473,
      "grad_norm": 0.8388517498970032,
      "learning_rate": 0.00029992502397643156,
      "loss": 5.578570556640625,
      "step": 850
    },
    {
      "epoch": 0.0005298659689313089,
      "grad_norm": 0.7705523371696472,
      "learning_rate": 0.00029992060842734037,
      "loss": 5.543483276367187,
      "step": 900
    },
    {
      "epoch": 0.0005593029672052705,
      "grad_norm": 0.8905978798866272,
      "learning_rate": 0.0002999161928782492,
      "loss": 5.429735107421875,
      "step": 950
    },
    {
      "epoch": 0.0005887399654792322,
      "grad_norm": 0.6083276271820068,
      "learning_rate": 0.00029991177732915795,
      "loss": 5.423040771484375,
      "step": 1000
    },
    {
      "epoch": 0.0006181769637531937,
      "grad_norm": 0.8251233100891113,
      "learning_rate": 0.00029990736178006676,
      "loss": 5.36895263671875,
      "step": 1050
    },
    {
      "epoch": 0.0006476139620271553,
      "grad_norm": 0.7911943793296814,
      "learning_rate": 0.0002999029462309756,
      "loss": 5.387249145507813,
      "step": 1100
    },
    {
      "epoch": 0.000677050960301117,
      "grad_norm": 0.8078892230987549,
      "learning_rate": 0.0002998985306818844,
      "loss": 5.263767700195313,
      "step": 1150
    },
    {
      "epoch": 0.0007064879585750786,
      "grad_norm": 0.6848020553588867,
      "learning_rate": 0.0002998941151327932,
      "loss": 5.338374633789062,
      "step": 1200
    },
    {
      "epoch": 0.0007359249568490401,
      "grad_norm": 0.6651701331138611,
      "learning_rate": 0.000299889699583702,
      "loss": 5.227874755859375,
      "step": 1250
    },
    {
      "epoch": 0.0007653619551230017,
      "grad_norm": 0.7592987418174744,
      "learning_rate": 0.00029988528403461083,
      "loss": 5.264542236328125,
      "step": 1300
    },
    {
      "epoch": 0.0007947989533969634,
      "grad_norm": 0.9838700890541077,
      "learning_rate": 0.00029988086848551965,
      "loss": 5.1772802734375,
      "step": 1350
    },
    {
      "epoch": 0.000824235951670925,
      "grad_norm": 0.6935842633247375,
      "learning_rate": 0.0002998764529364284,
      "loss": 5.171509399414062,
      "step": 1400
    },
    {
      "epoch": 0.0008536729499448866,
      "grad_norm": 0.5578235983848572,
      "learning_rate": 0.0002998720373873372,
      "loss": 5.181893920898437,
      "step": 1450
    },
    {
      "epoch": 0.0008831099482188481,
      "grad_norm": 0.6788482069969177,
      "learning_rate": 0.00029986762183824604,
      "loss": 5.19407470703125,
      "step": 1500
    },
    {
      "epoch": 0.0009125469464928098,
      "grad_norm": 1.8122488260269165,
      "learning_rate": 0.00029986320628915485,
      "loss": 5.149515991210937,
      "step": 1550
    },
    {
      "epoch": 0.0009419839447667714,
      "grad_norm": 0.7572510838508606,
      "learning_rate": 0.00029985879074006367,
      "loss": 5.040107727050781,
      "step": 1600
    },
    {
      "epoch": 0.000971420943040733,
      "grad_norm": 0.8198531270027161,
      "learning_rate": 0.0002998543751909725,
      "loss": 5.046656494140625,
      "step": 1650
    },
    {
      "epoch": 0.0010008579413146946,
      "grad_norm": 0.6418925523757935,
      "learning_rate": 0.0002998499596418813,
      "loss": 5.12117431640625,
      "step": 1700
    },
    {
      "epoch": 0.0010302949395886562,
      "grad_norm": 0.6125956177711487,
      "learning_rate": 0.00029984554409279006,
      "loss": 5.012218627929688,
      "step": 1750
    },
    {
      "epoch": 0.0010597319378626178,
      "grad_norm": 0.7468776106834412,
      "learning_rate": 0.00029984112854369887,
      "loss": 5.009073486328125,
      "step": 1800
    },
    {
      "epoch": 0.0010891689361365794,
      "grad_norm": 0.7665901780128479,
      "learning_rate": 0.00029983671299460774,
      "loss": 5.024835815429688,
      "step": 1850
    },
    {
      "epoch": 0.001118605934410541,
      "grad_norm": 0.7550573348999023,
      "learning_rate": 0.0002998322974455165,
      "loss": 5.006237487792969,
      "step": 1900
    },
    {
      "epoch": 0.0011480429326845027,
      "grad_norm": 0.68825763463974,
      "learning_rate": 0.0002998278818964253,
      "loss": 4.983524780273438,
      "step": 1950
    },
    {
      "epoch": 0.0011774799309584643,
      "grad_norm": 0.5875523686408997,
      "learning_rate": 0.00029982346634733413,
      "loss": 4.9259228515625,
      "step": 2000
    },
    {
      "epoch": 0.0012069169292324257,
      "grad_norm": 0.7606977820396423,
      "learning_rate": 0.00029981905079824294,
      "loss": 4.913408203125,
      "step": 2050
    },
    {
      "epoch": 0.0012363539275063874,
      "grad_norm": 1.1867406368255615,
      "learning_rate": 0.00029981463524915176,
      "loss": 4.881487731933594,
      "step": 2100
    },
    {
      "epoch": 0.001265790925780349,
      "grad_norm": 0.7306165099143982,
      "learning_rate": 0.0002998102197000605,
      "loss": 4.899647521972656,
      "step": 2150
    },
    {
      "epoch": 0.0012952279240543106,
      "grad_norm": 0.7093265056610107,
      "learning_rate": 0.0002998058041509694,
      "loss": 4.873468017578125,
      "step": 2200
    },
    {
      "epoch": 0.0013246649223282723,
      "grad_norm": 0.972162663936615,
      "learning_rate": 0.0002998013886018782,
      "loss": 4.896275939941407,
      "step": 2250
    },
    {
      "epoch": 0.001354101920602234,
      "grad_norm": 0.5523830652236938,
      "learning_rate": 0.00029979697305278696,
      "loss": 4.8471337890625,
      "step": 2300
    },
    {
      "epoch": 0.0013835389188761955,
      "grad_norm": 0.7724469900131226,
      "learning_rate": 0.0002997925575036958,
      "loss": 4.875270690917969,
      "step": 2350
    },
    {
      "epoch": 0.0014129759171501572,
      "grad_norm": 0.9278637170791626,
      "learning_rate": 0.0002997881419546046,
      "loss": 4.822630004882813,
      "step": 2400
    },
    {
      "epoch": 0.0014424129154241188,
      "grad_norm": 0.5320096015930176,
      "learning_rate": 0.0002997837264055134,
      "loss": 4.791508483886719,
      "step": 2450
    },
    {
      "epoch": 0.0014718499136980802,
      "grad_norm": 0.6005656123161316,
      "learning_rate": 0.0002997793108564222,
      "loss": 4.861024475097656,
      "step": 2500
    },
    {
      "epoch": 0.0015012869119720418,
      "grad_norm": 0.6439042687416077,
      "learning_rate": 0.00029977489530733103,
      "loss": 4.824080505371094,
      "step": 2550
    },
    {
      "epoch": 0.0015307239102460035,
      "grad_norm": 0.7070223093032837,
      "learning_rate": 0.00029977047975823985,
      "loss": 4.752425537109375,
      "step": 2600
    },
    {
      "epoch": 0.001560160908519965,
      "grad_norm": 0.7973089814186096,
      "learning_rate": 0.00029976606420914866,
      "loss": 4.726258850097656,
      "step": 2650
    },
    {
      "epoch": 0.0015895979067939267,
      "grad_norm": 0.7084776759147644,
      "learning_rate": 0.0002997616486600574,
      "loss": 4.724707336425781,
      "step": 2700
    },
    {
      "epoch": 0.0016190349050678884,
      "grad_norm": 0.7327000498771667,
      "learning_rate": 0.00029975723311096624,
      "loss": 4.720442504882812,
      "step": 2750
    },
    {
      "epoch": 0.00164847190334185,
      "grad_norm": 0.5960451364517212,
      "learning_rate": 0.00029975281756187505,
      "loss": 4.7130670166015625,
      "step": 2800
    },
    {
      "epoch": 0.0016779089016158116,
      "grad_norm": 0.6381566524505615,
      "learning_rate": 0.00029974840201278387,
      "loss": 4.721955871582031,
      "step": 2850
    },
    {
      "epoch": 0.0017073458998897732,
      "grad_norm": 0.7925652861595154,
      "learning_rate": 0.0002997439864636927,
      "loss": 4.703958435058594,
      "step": 2900
    },
    {
      "epoch": 0.0017367828981637347,
      "grad_norm": 0.5750381946563721,
      "learning_rate": 0.0002997395709146015,
      "loss": 4.687227172851562,
      "step": 2950
    },
    {
      "epoch": 0.0017662198964376963,
      "grad_norm": 0.6599341034889221,
      "learning_rate": 0.0002997351553655103,
      "loss": 4.644759826660156,
      "step": 3000
    },
    {
      "epoch": 0.001795656894711658,
      "grad_norm": 0.8492552638053894,
      "learning_rate": 0.0002997307398164191,
      "loss": 4.689851379394531,
      "step": 3050
    },
    {
      "epoch": 0.0018250938929856195,
      "grad_norm": 1.2669733762741089,
      "learning_rate": 0.0002997263242673279,
      "loss": 4.667993469238281,
      "step": 3100
    },
    {
      "epoch": 0.0018545308912595812,
      "grad_norm": 0.7495269179344177,
      "learning_rate": 0.00029972190871823675,
      "loss": 4.645321960449219,
      "step": 3150
    },
    {
      "epoch": 0.0018839678895335428,
      "grad_norm": 0.8966331481933594,
      "learning_rate": 0.0002997174931691455,
      "loss": 4.585001525878906,
      "step": 3200
    },
    {
      "epoch": 0.0019134048878075044,
      "grad_norm": 0.7717205882072449,
      "learning_rate": 0.0002997130776200543,
      "loss": 4.702607727050781,
      "step": 3250
    },
    {
      "epoch": 0.001942841886081466,
      "grad_norm": 0.7275465726852417,
      "learning_rate": 0.00029970866207096314,
      "loss": 4.617936401367188,
      "step": 3300
    },
    {
      "epoch": 0.0019722788843554275,
      "grad_norm": 0.6321767568588257,
      "learning_rate": 0.00029970424652187196,
      "loss": 4.693574829101562,
      "step": 3350
    },
    {
      "epoch": 0.002001715882629389,
      "grad_norm": 0.763538658618927,
      "learning_rate": 0.00029969983097278077,
      "loss": 4.673726501464844,
      "step": 3400
    },
    {
      "epoch": 0.0020311528809033507,
      "grad_norm": 0.5985351800918579,
      "learning_rate": 0.00029969541542368953,
      "loss": 4.605963134765625,
      "step": 3450
    },
    {
      "epoch": 0.0020605898791773124,
      "grad_norm": 0.6944624185562134,
      "learning_rate": 0.0002996909998745984,
      "loss": 4.599256591796875,
      "step": 3500
    },
    {
      "epoch": 0.002090026877451274,
      "grad_norm": 0.6094560027122498,
      "learning_rate": 0.0002996865843255072,
      "loss": 4.583138122558593,
      "step": 3550
    },
    {
      "epoch": 0.0021194638757252356,
      "grad_norm": 0.6929382085800171,
      "learning_rate": 0.000299682168776416,
      "loss": 4.577537536621094,
      "step": 3600
    },
    {
      "epoch": 0.0021489008739991973,
      "grad_norm": 0.6469603776931763,
      "learning_rate": 0.0002996777532273248,
      "loss": 4.61540283203125,
      "step": 3650
    },
    {
      "epoch": 0.002178337872273159,
      "grad_norm": 0.5631442666053772,
      "learning_rate": 0.0002996733376782336,
      "loss": 4.576884155273437,
      "step": 3700
    },
    {
      "epoch": 0.0022077748705471205,
      "grad_norm": 1.0786155462265015,
      "learning_rate": 0.0002996689221291424,
      "loss": 4.610385131835938,
      "step": 3750
    },
    {
      "epoch": 0.002237211868821082,
      "grad_norm": 0.6526992917060852,
      "learning_rate": 0.00029966450658005123,
      "loss": 4.575941467285157,
      "step": 3800
    },
    {
      "epoch": 0.0022666488670950438,
      "grad_norm": 0.4798518121242523,
      "learning_rate": 0.00029966009103096005,
      "loss": 4.584553833007813,
      "step": 3850
    },
    {
      "epoch": 0.0022960858653690054,
      "grad_norm": 0.7173407673835754,
      "learning_rate": 0.00029965567548186886,
      "loss": 4.5248956298828125,
      "step": 3900
    },
    {
      "epoch": 0.002325522863642967,
      "grad_norm": 0.5315476059913635,
      "learning_rate": 0.0002996512599327777,
      "loss": 4.518644409179688,
      "step": 3950
    },
    {
      "epoch": 0.0023549598619169287,
      "grad_norm": 0.7507267594337463,
      "learning_rate": 0.00029964684438368644,
      "loss": 4.609258422851562,
      "step": 4000
    },
    {
      "epoch": 0.0023843968601908903,
      "grad_norm": 0.6924713850021362,
      "learning_rate": 0.00029964242883459525,
      "loss": 4.518807983398437,
      "step": 4050
    },
    {
      "epoch": 0.0024138338584648515,
      "grad_norm": 0.6268355250358582,
      "learning_rate": 0.00029963801328550406,
      "loss": 4.530610961914062,
      "step": 4100
    },
    {
      "epoch": 0.002443270856738813,
      "grad_norm": 0.5507285594940186,
      "learning_rate": 0.0002996335977364129,
      "loss": 4.535648803710938,
      "step": 4150
    },
    {
      "epoch": 0.0024727078550127748,
      "grad_norm": 0.5368896126747131,
      "learning_rate": 0.0002996291821873217,
      "loss": 4.460059509277344,
      "step": 4200
    },
    {
      "epoch": 0.0025021448532867364,
      "grad_norm": 1.015692114830017,
      "learning_rate": 0.0002996247666382305,
      "loss": 4.478104248046875,
      "step": 4250
    },
    {
      "epoch": 0.002531581851560698,
      "grad_norm": 0.7693761587142944,
      "learning_rate": 0.0002996203510891393,
      "loss": 4.490260620117187,
      "step": 4300
    },
    {
      "epoch": 0.0025610188498346597,
      "grad_norm": 0.6537253260612488,
      "learning_rate": 0.00029961593554004814,
      "loss": 4.4494970703125,
      "step": 4350
    },
    {
      "epoch": 0.0025904558481086213,
      "grad_norm": 0.7599261403083801,
      "learning_rate": 0.0002996115199909569,
      "loss": 4.518837585449218,
      "step": 4400
    },
    {
      "epoch": 0.002619892846382583,
      "grad_norm": 0.6092306971549988,
      "learning_rate": 0.00029960710444186577,
      "loss": 4.453702087402344,
      "step": 4450
    },
    {
      "epoch": 0.0026493298446565445,
      "grad_norm": 0.6066957116127014,
      "learning_rate": 0.0002996026888927746,
      "loss": 4.471926574707031,
      "step": 4500
    },
    {
      "epoch": 0.002678766842930506,
      "grad_norm": 0.7276062369346619,
      "learning_rate": 0.00029959827334368334,
      "loss": 4.469823608398437,
      "step": 4550
    },
    {
      "epoch": 0.002708203841204468,
      "grad_norm": 0.79478919506073,
      "learning_rate": 0.00029959385779459216,
      "loss": 4.428940124511719,
      "step": 4600
    },
    {
      "epoch": 0.0027376408394784294,
      "grad_norm": 0.590240478515625,
      "learning_rate": 0.00029958944224550097,
      "loss": 4.4477783203125,
      "step": 4650
    },
    {
      "epoch": 0.002767077837752391,
      "grad_norm": 0.5499537587165833,
      "learning_rate": 0.0002995850266964098,
      "loss": 4.420744018554688,
      "step": 4700
    },
    {
      "epoch": 0.0027965148360263527,
      "grad_norm": 0.6672317981719971,
      "learning_rate": 0.0002995806111473186,
      "loss": 4.388303833007813,
      "step": 4750
    },
    {
      "epoch": 0.0028259518343003143,
      "grad_norm": 0.8488474488258362,
      "learning_rate": 0.0002995761955982274,
      "loss": 4.473275756835937,
      "step": 4800
    },
    {
      "epoch": 0.002855388832574276,
      "grad_norm": 0.7180399298667908,
      "learning_rate": 0.00029957178004913623,
      "loss": 4.446622314453125,
      "step": 4850
    },
    {
      "epoch": 0.0028848258308482376,
      "grad_norm": 0.6537158489227295,
      "learning_rate": 0.000299567364500045,
      "loss": 4.422460021972657,
      "step": 4900
    },
    {
      "epoch": 0.002914262829122199,
      "grad_norm": 0.504582941532135,
      "learning_rate": 0.0002995629489509538,
      "loss": 4.375197448730469,
      "step": 4950
    },
    {
      "epoch": 0.0029436998273961604,
      "grad_norm": 0.6466673612594604,
      "learning_rate": 0.0002995585334018626,
      "loss": 4.374354248046875,
      "step": 5000
    },
    {
      "epoch": 0.002973136825670122,
      "grad_norm": 0.8836430907249451,
      "learning_rate": 0.00029955411785277143,
      "loss": 4.348298950195312,
      "step": 5050
    },
    {
      "epoch": 0.0030025738239440837,
      "grad_norm": 0.6507533192634583,
      "learning_rate": 0.00029954970230368025,
      "loss": 4.341182556152344,
      "step": 5100
    },
    {
      "epoch": 0.0030320108222180453,
      "grad_norm": 0.7422476410865784,
      "learning_rate": 0.00029954528675458906,
      "loss": 4.323683471679687,
      "step": 5150
    },
    {
      "epoch": 0.003061447820492007,
      "grad_norm": 0.47801393270492554,
      "learning_rate": 0.0002995408712054979,
      "loss": 4.36968994140625,
      "step": 5200
    },
    {
      "epoch": 0.0030908848187659686,
      "grad_norm": 0.6749917268753052,
      "learning_rate": 0.0002995364556564067,
      "loss": 4.310074768066406,
      "step": 5250
    },
    {
      "epoch": 0.00312032181703993,
      "grad_norm": 0.7628033757209778,
      "learning_rate": 0.00029953204010731545,
      "loss": 4.32994384765625,
      "step": 5300
    },
    {
      "epoch": 0.003149758815313892,
      "grad_norm": 0.7435867786407471,
      "learning_rate": 0.00029952762455822426,
      "loss": 4.277841491699219,
      "step": 5350
    },
    {
      "epoch": 0.0031791958135878534,
      "grad_norm": 1.274184226989746,
      "learning_rate": 0.0002995232090091331,
      "loss": 4.273157348632813,
      "step": 5400
    },
    {
      "epoch": 0.003208632811861815,
      "grad_norm": 0.7267596125602722,
      "learning_rate": 0.0002995187934600419,
      "loss": 4.301112365722656,
      "step": 5450
    },
    {
      "epoch": 0.0032380698101357767,
      "grad_norm": 0.5402125716209412,
      "learning_rate": 0.0002995143779109507,
      "loss": 4.287984619140625,
      "step": 5500
    },
    {
      "epoch": 0.0032675068084097383,
      "grad_norm": 0.743585467338562,
      "learning_rate": 0.0002995099623618595,
      "loss": 4.291321716308594,
      "step": 5550
    },
    {
      "epoch": 0.0032969438066837,
      "grad_norm": 0.6032958030700684,
      "learning_rate": 0.00029950554681276834,
      "loss": 4.2258169555664065,
      "step": 5600
    },
    {
      "epoch": 0.0033263808049576616,
      "grad_norm": 0.4633687436580658,
      "learning_rate": 0.00029950113126367715,
      "loss": 4.294220886230469,
      "step": 5650
    },
    {
      "epoch": 0.0033558178032316232,
      "grad_norm": 0.5703911185264587,
      "learning_rate": 0.0002994967157145859,
      "loss": 4.319907531738282,
      "step": 5700
    },
    {
      "epoch": 0.003385254801505585,
      "grad_norm": 0.6908669471740723,
      "learning_rate": 0.0002994923001654948,
      "loss": 4.211581726074218,
      "step": 5750
    },
    {
      "epoch": 0.0034146917997795465,
      "grad_norm": 0.5593363642692566,
      "learning_rate": 0.0002994878846164036,
      "loss": 4.2632989501953125,
      "step": 5800
    },
    {
      "epoch": 0.003444128798053508,
      "grad_norm": 0.9172580242156982,
      "learning_rate": 0.00029948346906731235,
      "loss": 4.300866394042969,
      "step": 5850
    },
    {
      "epoch": 0.0034735657963274693,
      "grad_norm": 0.41300103068351746,
      "learning_rate": 0.00029947905351822117,
      "loss": 4.235009765625,
      "step": 5900
    },
    {
      "epoch": 0.003503002794601431,
      "grad_norm": 0.7957214713096619,
      "learning_rate": 0.00029947463796913,
      "loss": 4.225993957519531,
      "step": 5950
    },
    {
      "epoch": 0.0035324397928753926,
      "grad_norm": 0.676835298538208,
      "learning_rate": 0.0002994702224200388,
      "loss": 4.2083743286132815,
      "step": 6000
    },
    {
      "epoch": 0.003561876791149354,
      "grad_norm": 0.5486562252044678,
      "learning_rate": 0.0002994658068709476,
      "loss": 4.197705078125,
      "step": 6050
    },
    {
      "epoch": 0.003591313789423316,
      "grad_norm": 0.6647422313690186,
      "learning_rate": 0.00029946139132185643,
      "loss": 4.248899536132813,
      "step": 6100
    },
    {
      "epoch": 0.0036207507876972775,
      "grad_norm": 0.5490094423294067,
      "learning_rate": 0.00029945697577276524,
      "loss": 4.164762878417969,
      "step": 6150
    },
    {
      "epoch": 0.003650187785971239,
      "grad_norm": 0.7341269850730896,
      "learning_rate": 0.00029945256022367406,
      "loss": 4.19623046875,
      "step": 6200
    },
    {
      "epoch": 0.0036796247842452007,
      "grad_norm": 0.6762226223945618,
      "learning_rate": 0.0002994481446745828,
      "loss": 4.222210998535156,
      "step": 6250
    },
    {
      "epoch": 0.0037090617825191624,
      "grad_norm": 0.5429332256317139,
      "learning_rate": 0.00029944372912549163,
      "loss": 4.199527893066406,
      "step": 6300
    },
    {
      "epoch": 0.003738498780793124,
      "grad_norm": 0.9981448650360107,
      "learning_rate": 0.00029943931357640045,
      "loss": 4.147794494628906,
      "step": 6350
    },
    {
      "epoch": 0.0037679357790670856,
      "grad_norm": 0.7597318291664124,
      "learning_rate": 0.00029943489802730926,
      "loss": 4.1709222412109375,
      "step": 6400
    },
    {
      "epoch": 0.0037973727773410472,
      "grad_norm": 0.5356363654136658,
      "learning_rate": 0.0002994304824782181,
      "loss": 4.1450894165039065,
      "step": 6450
    },
    {
      "epoch": 0.003826809775615009,
      "grad_norm": 0.565214216709137,
      "learning_rate": 0.0002994260669291269,
      "loss": 4.16892578125,
      "step": 6500
    },
    {
      "epoch": 0.0038562467738889705,
      "grad_norm": 0.8412073254585266,
      "learning_rate": 0.0002994216513800357,
      "loss": 4.1308377075195315,
      "step": 6550
    },
    {
      "epoch": 0.003885683772162932,
      "grad_norm": 0.49122345447540283,
      "learning_rate": 0.00029941723583094446,
      "loss": 4.194474487304688,
      "step": 6600
    },
    {
      "epoch": 0.003915120770436893,
      "grad_norm": 0.59850013256073,
      "learning_rate": 0.0002994128202818533,
      "loss": 4.144138488769531,
      "step": 6650
    },
    {
      "epoch": 0.003944557768710855,
      "grad_norm": 1.6896206140518188,
      "learning_rate": 0.0002994084047327621,
      "loss": 4.144266052246094,
      "step": 6700
    },
    {
      "epoch": 0.003973994766984817,
      "grad_norm": 0.6903894543647766,
      "learning_rate": 0.0002994039891836709,
      "loss": 4.160715942382812,
      "step": 6750
    },
    {
      "epoch": 0.004003431765258778,
      "grad_norm": 0.8733358383178711,
      "learning_rate": 0.0002993995736345797,
      "loss": 4.150364990234375,
      "step": 6800
    },
    {
      "epoch": 0.00403286876353274,
      "grad_norm": 0.7538623809814453,
      "learning_rate": 0.00029939515808548854,
      "loss": 4.118778076171875,
      "step": 6850
    },
    {
      "epoch": 0.0040623057618067015,
      "grad_norm": 0.5204249620437622,
      "learning_rate": 0.00029939074253639735,
      "loss": 4.127062377929687,
      "step": 6900
    },
    {
      "epoch": 0.004091742760080663,
      "grad_norm": 0.6425044536590576,
      "learning_rate": 0.00029938632698730617,
      "loss": 4.087417297363281,
      "step": 6950
    },
    {
      "epoch": 0.004121179758354625,
      "grad_norm": 0.697498619556427,
      "learning_rate": 0.0002993819114382149,
      "loss": 4.0751953125,
      "step": 7000
    },
    {
      "epoch": 0.004150616756628586,
      "grad_norm": 0.5834298729896545,
      "learning_rate": 0.0002993774958891238,
      "loss": 4.107063903808593,
      "step": 7050
    },
    {
      "epoch": 0.004180053754902548,
      "grad_norm": 0.6233769655227661,
      "learning_rate": 0.0002993730803400326,
      "loss": 4.052364807128907,
      "step": 7100
    },
    {
      "epoch": 0.00420949075317651,
      "grad_norm": 0.6466637253761292,
      "learning_rate": 0.00029936866479094137,
      "loss": 4.079283142089844,
      "step": 7150
    },
    {
      "epoch": 0.004238927751450471,
      "grad_norm": 0.4679613411426544,
      "learning_rate": 0.0002993642492418502,
      "loss": 4.054103088378906,
      "step": 7200
    },
    {
      "epoch": 0.004268364749724433,
      "grad_norm": 0.5627173185348511,
      "learning_rate": 0.000299359833692759,
      "loss": 4.0287039184570315,
      "step": 7250
    },
    {
      "epoch": 0.0042978017479983945,
      "grad_norm": 0.521531879901886,
      "learning_rate": 0.0002993554181436678,
      "loss": 4.085139465332031,
      "step": 7300
    },
    {
      "epoch": 0.004327238746272356,
      "grad_norm": 0.4891389310359955,
      "learning_rate": 0.0002993510025945766,
      "loss": 4.065169982910156,
      "step": 7350
    },
    {
      "epoch": 0.004356675744546318,
      "grad_norm": 0.5580434203147888,
      "learning_rate": 0.00029934658704548544,
      "loss": 4.018386535644531,
      "step": 7400
    },
    {
      "epoch": 0.004386112742820279,
      "grad_norm": 0.6094513535499573,
      "learning_rate": 0.00029934217149639426,
      "loss": 4.053096008300781,
      "step": 7450
    },
    {
      "epoch": 0.004415549741094241,
      "grad_norm": 0.50697261095047,
      "learning_rate": 0.00029933775594730307,
      "loss": 4.032203369140625,
      "step": 7500
    },
    {
      "epoch": 0.004444986739368203,
      "grad_norm": 0.5194651484489441,
      "learning_rate": 0.00029933334039821183,
      "loss": 4.0294384765625,
      "step": 7550
    },
    {
      "epoch": 0.004474423737642164,
      "grad_norm": 0.5441867113113403,
      "learning_rate": 0.00029932892484912065,
      "loss": 4.039732666015625,
      "step": 7600
    },
    {
      "epoch": 0.004503860735916126,
      "grad_norm": 0.5797628164291382,
      "learning_rate": 0.00029932450930002946,
      "loss": 3.98349853515625,
      "step": 7650
    },
    {
      "epoch": 0.0045332977341900876,
      "grad_norm": 0.5529788732528687,
      "learning_rate": 0.0002993200937509383,
      "loss": 4.002181091308594,
      "step": 7700
    },
    {
      "epoch": 0.004562734732464049,
      "grad_norm": 0.8079077005386353,
      "learning_rate": 0.0002993156782018471,
      "loss": 4.038533325195313,
      "step": 7750
    },
    {
      "epoch": 0.004592171730738011,
      "grad_norm": 0.6113392114639282,
      "learning_rate": 0.0002993112626527559,
      "loss": 4.0467922973632815,
      "step": 7800
    },
    {
      "epoch": 0.0046216087290119725,
      "grad_norm": 0.5018461346626282,
      "learning_rate": 0.0002993068471036647,
      "loss": 4.01013671875,
      "step": 7850
    },
    {
      "epoch": 0.004651045727285934,
      "grad_norm": 0.8082796931266785,
      "learning_rate": 0.00029930243155457353,
      "loss": 4.024381713867188,
      "step": 7900
    },
    {
      "epoch": 0.004680482725559896,
      "grad_norm": 0.4463624656200409,
      "learning_rate": 0.0002992980160054823,
      "loss": 3.948871765136719,
      "step": 7950
    },
    {
      "epoch": 0.004709919723833857,
      "grad_norm": 0.7486303448677063,
      "learning_rate": 0.0002992936004563911,
      "loss": 4.015348205566406,
      "step": 8000
    },
    {
      "epoch": 0.004739356722107819,
      "grad_norm": 0.5753210783004761,
      "learning_rate": 0.0002992891849073,
      "loss": 3.9402618408203125,
      "step": 8050
    },
    {
      "epoch": 0.004768793720381781,
      "grad_norm": 0.7120587229728699,
      "learning_rate": 0.00029928476935820874,
      "loss": 4.0204296875,
      "step": 8100
    },
    {
      "epoch": 0.004798230718655742,
      "grad_norm": 0.6206667423248291,
      "learning_rate": 0.00029928035380911755,
      "loss": 4.006302490234375,
      "step": 8150
    },
    {
      "epoch": 0.004827667716929703,
      "grad_norm": 0.6058242917060852,
      "learning_rate": 0.00029927593826002636,
      "loss": 4.018255004882812,
      "step": 8200
    },
    {
      "epoch": 0.004857104715203665,
      "grad_norm": 0.5593596696853638,
      "learning_rate": 0.0002992715227109352,
      "loss": 3.9617959594726564,
      "step": 8250
    },
    {
      "epoch": 0.004886541713477626,
      "grad_norm": 0.5687791705131531,
      "learning_rate": 0.00029926710716184394,
      "loss": 3.990050964355469,
      "step": 8300
    },
    {
      "epoch": 0.004915978711751588,
      "grad_norm": 0.41162046790122986,
      "learning_rate": 0.00029926269161275275,
      "loss": 3.9532760620117187,
      "step": 8350
    },
    {
      "epoch": 0.0049454157100255495,
      "grad_norm": 0.6884109377861023,
      "learning_rate": 0.0002992582760636616,
      "loss": 3.9734075927734374,
      "step": 8400
    },
    {
      "epoch": 0.004974852708299511,
      "grad_norm": 0.6932796835899353,
      "learning_rate": 0.0002992538605145704,
      "loss": 3.985809020996094,
      "step": 8450
    },
    {
      "epoch": 0.005004289706573473,
      "grad_norm": 0.5691799521446228,
      "learning_rate": 0.0002992494449654792,
      "loss": 4.008411254882812,
      "step": 8500
    },
    {
      "epoch": 0.005033726704847434,
      "grad_norm": 0.4985423982143402,
      "learning_rate": 0.000299245029416388,
      "loss": 3.9584500122070314,
      "step": 8550
    },
    {
      "epoch": 0.005063163703121396,
      "grad_norm": 0.6533710360527039,
      "learning_rate": 0.0002992406138672968,
      "loss": 3.938670349121094,
      "step": 8600
    },
    {
      "epoch": 0.005092600701395358,
      "grad_norm": 0.46181944012641907,
      "learning_rate": 0.00029923619831820564,
      "loss": 3.8755999755859376,
      "step": 8650
    },
    {
      "epoch": 0.005122037699669319,
      "grad_norm": 0.5635668635368347,
      "learning_rate": 0.00029923178276911446,
      "loss": 3.8917141723632813,
      "step": 8700
    },
    {
      "epoch": 0.005151474697943281,
      "grad_norm": 0.5262820720672607,
      "learning_rate": 0.00029922736722002327,
      "loss": 3.9362564086914062,
      "step": 8750
    },
    {
      "epoch": 0.0051809116962172426,
      "grad_norm": 0.6007362604141235,
      "learning_rate": 0.0002992229516709321,
      "loss": 3.9071429443359373,
      "step": 8800
    },
    {
      "epoch": 0.005210348694491204,
      "grad_norm": 0.6234773397445679,
      "learning_rate": 0.00029921853612184084,
      "loss": 3.9056655883789064,
      "step": 8850
    },
    {
      "epoch": 0.005239785692765166,
      "grad_norm": 0.48420482873916626,
      "learning_rate": 0.00029921412057274966,
      "loss": 3.9205499267578126,
      "step": 8900
    },
    {
      "epoch": 0.0052692226910391275,
      "grad_norm": 0.5922645926475525,
      "learning_rate": 0.0002992097050236585,
      "loss": 3.882735290527344,
      "step": 8950
    },
    {
      "epoch": 0.005298659689313089,
      "grad_norm": 0.4976702034473419,
      "learning_rate": 0.0002992052894745673,
      "loss": 3.9225946044921876,
      "step": 9000
    },
    {
      "epoch": 0.005328096687587051,
      "grad_norm": 0.5053274035453796,
      "learning_rate": 0.0002992008739254761,
      "loss": 3.8859259033203126,
      "step": 9050
    },
    {
      "epoch": 0.005357533685861012,
      "grad_norm": 0.40719425678253174,
      "learning_rate": 0.0002991964583763849,
      "loss": 3.8392767333984374,
      "step": 9100
    },
    {
      "epoch": 0.005386970684134974,
      "grad_norm": 0.5912914872169495,
      "learning_rate": 0.00029919204282729373,
      "loss": 3.891049499511719,
      "step": 9150
    },
    {
      "epoch": 0.005416407682408936,
      "grad_norm": 0.6306865811347961,
      "learning_rate": 0.00029918762727820255,
      "loss": 3.90007568359375,
      "step": 9200
    },
    {
      "epoch": 0.005445844680682897,
      "grad_norm": 0.4902883768081665,
      "learning_rate": 0.0002991832117291113,
      "loss": 3.9148297119140625,
      "step": 9250
    },
    {
      "epoch": 0.005475281678956859,
      "grad_norm": 0.5049835443496704,
      "learning_rate": 0.0002991787961800201,
      "loss": 3.824625244140625,
      "step": 9300
    },
    {
      "epoch": 0.0055047186772308205,
      "grad_norm": 0.5601292848587036,
      "learning_rate": 0.000299174380630929,
      "loss": 3.934866943359375,
      "step": 9350
    },
    {
      "epoch": 0.005534155675504782,
      "grad_norm": 0.6363621950149536,
      "learning_rate": 0.00029916996508183775,
      "loss": 3.854745178222656,
      "step": 9400
    },
    {
      "epoch": 0.005563592673778744,
      "grad_norm": 0.5438066720962524,
      "learning_rate": 0.00029916554953274656,
      "loss": 3.8880401611328126,
      "step": 9450
    },
    {
      "epoch": 0.005593029672052705,
      "grad_norm": 0.5508623719215393,
      "learning_rate": 0.0002991611339836554,
      "loss": 3.8316836547851563,
      "step": 9500
    },
    {
      "epoch": 0.005622466670326667,
      "grad_norm": 0.587388813495636,
      "learning_rate": 0.0002991567184345642,
      "loss": 3.9239862060546873,
      "step": 9550
    },
    {
      "epoch": 0.005651903668600629,
      "grad_norm": 0.5406348705291748,
      "learning_rate": 0.000299152302885473,
      "loss": 3.8461328125,
      "step": 9600
    },
    {
      "epoch": 0.00568134066687459,
      "grad_norm": 0.5968824028968811,
      "learning_rate": 0.00029914788733638177,
      "loss": 3.834413757324219,
      "step": 9650
    },
    {
      "epoch": 0.005710777665148552,
      "grad_norm": 0.582781195640564,
      "learning_rate": 0.00029914347178729064,
      "loss": 3.8392242431640624,
      "step": 9700
    },
    {
      "epoch": 0.0057402146634225135,
      "grad_norm": 0.6148483753204346,
      "learning_rate": 0.00029913905623819945,
      "loss": 3.8635601806640625,
      "step": 9750
    },
    {
      "epoch": 0.005769651661696475,
      "grad_norm": 0.633619487285614,
      "learning_rate": 0.0002991346406891082,
      "loss": 3.856841735839844,
      "step": 9800
    },
    {
      "epoch": 0.005799088659970437,
      "grad_norm": 0.607355535030365,
      "learning_rate": 0.000299130225140017,
      "loss": 3.8151614379882814,
      "step": 9850
    },
    {
      "epoch": 0.005828525658244398,
      "grad_norm": 0.5380693078041077,
      "learning_rate": 0.00029912580959092584,
      "loss": 3.860657043457031,
      "step": 9900
    },
    {
      "epoch": 0.00585796265651836,
      "grad_norm": 0.47209692001342773,
      "learning_rate": 0.00029912139404183465,
      "loss": 3.8271856689453125,
      "step": 9950
    },
    {
      "epoch": 0.005887399654792321,
      "grad_norm": 0.493587851524353,
      "learning_rate": 0.00029911697849274347,
      "loss": 3.767059020996094,
      "step": 10000
    },
    {
      "epoch": 0.0059168366530662824,
      "grad_norm": 0.6219985485076904,
      "learning_rate": 0.0002991125629436523,
      "loss": 3.8319293212890626,
      "step": 10050
    },
    {
      "epoch": 0.005946273651340244,
      "grad_norm": 0.5084136128425598,
      "learning_rate": 0.0002991081473945611,
      "loss": 3.83237548828125,
      "step": 10100
    },
    {
      "epoch": 0.005975710649614206,
      "grad_norm": 0.4244358539581299,
      "learning_rate": 0.00029910373184546986,
      "loss": 3.8493280029296875,
      "step": 10150
    },
    {
      "epoch": 0.006005147647888167,
      "grad_norm": 0.48822182416915894,
      "learning_rate": 0.0002990993162963787,
      "loss": 3.8313192749023437,
      "step": 10200
    },
    {
      "epoch": 0.006034584646162129,
      "grad_norm": 0.6106606721878052,
      "learning_rate": 0.0002990949007472875,
      "loss": 3.8355197143554687,
      "step": 10250
    },
    {
      "epoch": 0.006064021644436091,
      "grad_norm": 0.7521558403968811,
      "learning_rate": 0.0002990904851981963,
      "loss": 3.84543212890625,
      "step": 10300
    },
    {
      "epoch": 0.006093458642710052,
      "grad_norm": 0.5872248411178589,
      "learning_rate": 0.0002990860696491051,
      "loss": 3.8179525756835937,
      "step": 10350
    },
    {
      "epoch": 0.006122895640984014,
      "grad_norm": 0.6529163718223572,
      "learning_rate": 0.00029908165410001393,
      "loss": 3.785858154296875,
      "step": 10400
    },
    {
      "epoch": 0.0061523326392579755,
      "grad_norm": 0.5854930281639099,
      "learning_rate": 0.00029907723855092275,
      "loss": 3.784961853027344,
      "step": 10450
    },
    {
      "epoch": 0.006181769637531937,
      "grad_norm": 0.5023581385612488,
      "learning_rate": 0.00029907282300183156,
      "loss": 3.8292022705078126,
      "step": 10500
    },
    {
      "epoch": 0.006211206635805899,
      "grad_norm": 0.5287584066390991,
      "learning_rate": 0.0002990684074527403,
      "loss": 3.77739013671875,
      "step": 10550
    },
    {
      "epoch": 0.00624064363407986,
      "grad_norm": 0.4760802686214447,
      "learning_rate": 0.00029906399190364913,
      "loss": 3.766812744140625,
      "step": 10600
    },
    {
      "epoch": 0.006270080632353822,
      "grad_norm": 0.4092632234096527,
      "learning_rate": 0.000299059576354558,
      "loss": 3.7543515014648436,
      "step": 10650
    },
    {
      "epoch": 0.006299517630627784,
      "grad_norm": 0.4743334650993347,
      "learning_rate": 0.00029905516080546676,
      "loss": 3.7739321899414064,
      "step": 10700
    },
    {
      "epoch": 0.006328954628901745,
      "grad_norm": 0.4299439489841461,
      "learning_rate": 0.0002990507452563756,
      "loss": 3.8369293212890625,
      "step": 10750
    },
    {
      "epoch": 0.006358391627175707,
      "grad_norm": 1.4698443412780762,
      "learning_rate": 0.0002990463297072844,
      "loss": 3.7604751586914062,
      "step": 10800
    },
    {
      "epoch": 0.0063878286254496685,
      "grad_norm": 0.4397929310798645,
      "learning_rate": 0.0002990419141581932,
      "loss": 3.8025958251953127,
      "step": 10850
    },
    {
      "epoch": 0.00641726562372363,
      "grad_norm": 0.6336701512336731,
      "learning_rate": 0.000299037498609102,
      "loss": 3.737349548339844,
      "step": 10900
    },
    {
      "epoch": 0.006446702621997592,
      "grad_norm": 0.41684436798095703,
      "learning_rate": 0.0002990330830600108,
      "loss": 3.737449951171875,
      "step": 10950
    },
    {
      "epoch": 0.006476139620271553,
      "grad_norm": 0.7781590819358826,
      "learning_rate": 0.00029902866751091965,
      "loss": 3.727691650390625,
      "step": 11000
    },
    {
      "epoch": 0.006505576618545515,
      "grad_norm": 0.5045955181121826,
      "learning_rate": 0.00029902425196182846,
      "loss": 3.7893685913085937,
      "step": 11050
    },
    {
      "epoch": 0.006535013616819477,
      "grad_norm": 0.49081897735595703,
      "learning_rate": 0.0002990198364127372,
      "loss": 3.8071365356445312,
      "step": 11100
    },
    {
      "epoch": 0.006564450615093438,
      "grad_norm": 0.6123353838920593,
      "learning_rate": 0.00029901542086364604,
      "loss": 3.7899688720703124,
      "step": 11150
    },
    {
      "epoch": 0.0065938876133674,
      "grad_norm": 0.5137403011322021,
      "learning_rate": 0.00029901100531455485,
      "loss": 3.7796560668945314,
      "step": 11200
    },
    {
      "epoch": 0.006623324611641362,
      "grad_norm": 0.45117515325546265,
      "learning_rate": 0.00029900658976546367,
      "loss": 3.7371389770507815,
      "step": 11250
    },
    {
      "epoch": 0.006652761609915323,
      "grad_norm": 0.44178131222724915,
      "learning_rate": 0.0002990021742163725,
      "loss": 3.726701354980469,
      "step": 11300
    },
    {
      "epoch": 0.006682198608189285,
      "grad_norm": 0.3967849016189575,
      "learning_rate": 0.0002989977586672813,
      "loss": 3.712687683105469,
      "step": 11350
    },
    {
      "epoch": 0.0067116356064632465,
      "grad_norm": 0.8253253102302551,
      "learning_rate": 0.0002989933431181901,
      "loss": 3.785948486328125,
      "step": 11400
    },
    {
      "epoch": 0.006741072604737208,
      "grad_norm": 0.45558246970176697,
      "learning_rate": 0.0002989889275690989,
      "loss": 3.764185791015625,
      "step": 11450
    },
    {
      "epoch": 0.00677050960301117,
      "grad_norm": 0.6788157820701599,
      "learning_rate": 0.0002989845120200077,
      "loss": 3.709161376953125,
      "step": 11500
    },
    {
      "epoch": 0.006799946601285131,
      "grad_norm": 0.45410284399986267,
      "learning_rate": 0.0002989800964709165,
      "loss": 3.806800537109375,
      "step": 11550
    },
    {
      "epoch": 0.006829383599559093,
      "grad_norm": 0.5299049615859985,
      "learning_rate": 0.0002989756809218253,
      "loss": 3.7569219970703127,
      "step": 11600
    },
    {
      "epoch": 0.006858820597833055,
      "grad_norm": 0.4577332139015198,
      "learning_rate": 0.00029897126537273413,
      "loss": 3.7707864379882814,
      "step": 11650
    },
    {
      "epoch": 0.006888257596107016,
      "grad_norm": 0.4310346841812134,
      "learning_rate": 0.00029896684982364294,
      "loss": 3.7734756469726562,
      "step": 11700
    },
    {
      "epoch": 0.006917694594380978,
      "grad_norm": 0.5276893973350525,
      "learning_rate": 0.00029896243427455176,
      "loss": 3.7383084106445312,
      "step": 11750
    },
    {
      "epoch": 0.006947131592654939,
      "grad_norm": 0.805614709854126,
      "learning_rate": 0.0002989580187254606,
      "loss": 3.7143414306640623,
      "step": 11800
    },
    {
      "epoch": 0.0069765685909289,
      "grad_norm": 0.5810989737510681,
      "learning_rate": 0.00029895360317636933,
      "loss": 3.741022644042969,
      "step": 11850
    },
    {
      "epoch": 0.007006005589202862,
      "grad_norm": 0.5632728338241577,
      "learning_rate": 0.00029894918762727815,
      "loss": 3.743821716308594,
      "step": 11900
    },
    {
      "epoch": 0.0070354425874768235,
      "grad_norm": 0.4481128752231598,
      "learning_rate": 0.000298944772078187,
      "loss": 3.7933114624023436,
      "step": 11950
    },
    {
      "epoch": 0.007064879585750785,
      "grad_norm": 0.5808817148208618,
      "learning_rate": 0.0002989403565290958,
      "loss": 3.7356488037109377,
      "step": 12000
    },
    {
      "epoch": 0.007094316584024747,
      "grad_norm": 0.5947512984275818,
      "learning_rate": 0.0002989359409800046,
      "loss": 3.6974673461914063,
      "step": 12050
    },
    {
      "epoch": 0.007123753582298708,
      "grad_norm": 0.3888339400291443,
      "learning_rate": 0.0002989315254309134,
      "loss": 3.7360104370117186,
      "step": 12100
    },
    {
      "epoch": 0.00715319058057267,
      "grad_norm": 0.5698621273040771,
      "learning_rate": 0.0002989271098818222,
      "loss": 3.751977844238281,
      "step": 12150
    },
    {
      "epoch": 0.007182627578846632,
      "grad_norm": 0.8227885961532593,
      "learning_rate": 0.00029892269433273104,
      "loss": 3.7191256713867187,
      "step": 12200
    },
    {
      "epoch": 0.007212064577120593,
      "grad_norm": 0.7471373081207275,
      "learning_rate": 0.0002989182787836398,
      "loss": 3.7307159423828127,
      "step": 12250
    },
    {
      "epoch": 0.007241501575394555,
      "grad_norm": 0.5397338271141052,
      "learning_rate": 0.00029891386323454866,
      "loss": 3.733307189941406,
      "step": 12300
    },
    {
      "epoch": 0.0072709385736685166,
      "grad_norm": 0.6278129816055298,
      "learning_rate": 0.0002989094476854575,
      "loss": 3.6902597045898435,
      "step": 12350
    },
    {
      "epoch": 0.007300375571942478,
      "grad_norm": 0.6754025220870972,
      "learning_rate": 0.00029890503213636624,
      "loss": 3.677956848144531,
      "step": 12400
    },
    {
      "epoch": 0.00732981257021644,
      "grad_norm": 0.4242638051509857,
      "learning_rate": 0.00029890061658727505,
      "loss": 3.7114801025390625,
      "step": 12450
    },
    {
      "epoch": 0.0073592495684904015,
      "grad_norm": 0.629284143447876,
      "learning_rate": 0.00029889620103818387,
      "loss": 3.687984619140625,
      "step": 12500
    },
    {
      "epoch": 0.007388686566764363,
      "grad_norm": 0.4492742717266083,
      "learning_rate": 0.0002988917854890927,
      "loss": 3.7003213500976564,
      "step": 12550
    },
    {
      "epoch": 0.007418123565038325,
      "grad_norm": 0.5105596780776978,
      "learning_rate": 0.0002988873699400015,
      "loss": 3.674200134277344,
      "step": 12600
    },
    {
      "epoch": 0.007447560563312286,
      "grad_norm": 0.6817083358764648,
      "learning_rate": 0.0002988829543909103,
      "loss": 3.692904052734375,
      "step": 12650
    },
    {
      "epoch": 0.007476997561586248,
      "grad_norm": 0.5896594524383545,
      "learning_rate": 0.0002988785388418191,
      "loss": 3.675799560546875,
      "step": 12700
    },
    {
      "epoch": 0.00750643455986021,
      "grad_norm": 0.7632763385772705,
      "learning_rate": 0.00029887412329272794,
      "loss": 3.7060870361328124,
      "step": 12750
    },
    {
      "epoch": 0.007535871558134171,
      "grad_norm": 0.4764072000980377,
      "learning_rate": 0.0002988697077436367,
      "loss": 3.7198944091796875,
      "step": 12800
    },
    {
      "epoch": 0.007565308556408133,
      "grad_norm": 0.5221238732337952,
      "learning_rate": 0.0002988652921945455,
      "loss": 3.6674383544921874,
      "step": 12850
    },
    {
      "epoch": 0.0075947455546820945,
      "grad_norm": 0.45908764004707336,
      "learning_rate": 0.00029886087664545433,
      "loss": 3.694283447265625,
      "step": 12900
    },
    {
      "epoch": 0.007624182552956056,
      "grad_norm": 0.6536809206008911,
      "learning_rate": 0.00029885646109636314,
      "loss": 3.6452874755859375,
      "step": 12950
    },
    {
      "epoch": 0.007653619551230018,
      "grad_norm": 0.43446388840675354,
      "learning_rate": 0.00029885204554727196,
      "loss": 3.637245788574219,
      "step": 13000
    },
    {
      "epoch": 0.007683056549503979,
      "grad_norm": 0.48633864521980286,
      "learning_rate": 0.0002988476299981808,
      "loss": 3.6976507568359374,
      "step": 13050
    },
    {
      "epoch": 0.007712493547777941,
      "grad_norm": 0.5349531173706055,
      "learning_rate": 0.0002988432144490896,
      "loss": 3.6864990234375,
      "step": 13100
    },
    {
      "epoch": 0.007741930546051903,
      "grad_norm": 0.6183847188949585,
      "learning_rate": 0.0002988387988999984,
      "loss": 3.5945095825195312,
      "step": 13150
    },
    {
      "epoch": 0.007771367544325864,
      "grad_norm": 0.6288663744926453,
      "learning_rate": 0.00029883438335090716,
      "loss": 3.6879092407226564,
      "step": 13200
    },
    {
      "epoch": 0.007800804542599826,
      "grad_norm": 0.5025980472564697,
      "learning_rate": 0.00029882996780181603,
      "loss": 3.6523333740234376,
      "step": 13250
    },
    {
      "epoch": 0.007830241540873787,
      "grad_norm": 0.43017107248306274,
      "learning_rate": 0.00029882555225272485,
      "loss": 3.6921331787109377,
      "step": 13300
    },
    {
      "epoch": 0.00785967853914775,
      "grad_norm": 0.4185076951980591,
      "learning_rate": 0.0002988211367036336,
      "loss": 3.7400692749023436,
      "step": 13350
    },
    {
      "epoch": 0.00788911553742171,
      "grad_norm": 0.48009127378463745,
      "learning_rate": 0.0002988167211545424,
      "loss": 3.651595153808594,
      "step": 13400
    },
    {
      "epoch": 0.007918552535695672,
      "grad_norm": 0.8024508953094482,
      "learning_rate": 0.00029881230560545123,
      "loss": 3.642919006347656,
      "step": 13450
    },
    {
      "epoch": 0.007947989533969633,
      "grad_norm": 0.5198597311973572,
      "learning_rate": 0.00029880789005636005,
      "loss": 3.6964532470703126,
      "step": 13500
    },
    {
      "epoch": 0.007977426532243596,
      "grad_norm": 0.885084331035614,
      "learning_rate": 0.0002988034745072688,
      "loss": 3.7308453369140624,
      "step": 13550
    },
    {
      "epoch": 0.008006863530517556,
      "grad_norm": 0.42279329895973206,
      "learning_rate": 0.0002987990589581777,
      "loss": 3.7394009399414063,
      "step": 13600
    },
    {
      "epoch": 0.008036300528791519,
      "grad_norm": 0.5081620812416077,
      "learning_rate": 0.0002987946434090865,
      "loss": 3.6672674560546876,
      "step": 13650
    },
    {
      "epoch": 0.00806573752706548,
      "grad_norm": 0.44669193029403687,
      "learning_rate": 0.00029879022785999525,
      "loss": 3.687270202636719,
      "step": 13700
    },
    {
      "epoch": 0.008095174525339442,
      "grad_norm": 0.38171207904815674,
      "learning_rate": 0.00029878581231090407,
      "loss": 3.634102783203125,
      "step": 13750
    },
    {
      "epoch": 0.008124611523613403,
      "grad_norm": 0.3953717052936554,
      "learning_rate": 0.0002987813967618129,
      "loss": 3.5988174438476563,
      "step": 13800
    },
    {
      "epoch": 0.008154048521887365,
      "grad_norm": 0.6244080662727356,
      "learning_rate": 0.0002987769812127217,
      "loss": 3.65818603515625,
      "step": 13850
    },
    {
      "epoch": 0.008183485520161326,
      "grad_norm": 0.4855039715766907,
      "learning_rate": 0.0002987725656636305,
      "loss": 3.6699600219726562,
      "step": 13900
    },
    {
      "epoch": 0.008212922518435289,
      "grad_norm": 0.5454128980636597,
      "learning_rate": 0.0002987681501145393,
      "loss": 3.6320401000976563,
      "step": 13950
    },
    {
      "epoch": 0.00824235951670925,
      "grad_norm": 0.6604651808738708,
      "learning_rate": 0.00029876373456544814,
      "loss": 3.6718475341796877,
      "step": 14000
    },
    {
      "epoch": 0.008271796514983212,
      "grad_norm": 2.177231788635254,
      "learning_rate": 0.00029875931901635695,
      "loss": 3.5616244506835937,
      "step": 14050
    },
    {
      "epoch": 0.008301233513257173,
      "grad_norm": 0.38931483030319214,
      "learning_rate": 0.0002987549034672657,
      "loss": 3.611388854980469,
      "step": 14100
    },
    {
      "epoch": 0.008330670511531135,
      "grad_norm": 0.6183057427406311,
      "learning_rate": 0.00029875048791817453,
      "loss": 3.6276507568359375,
      "step": 14150
    },
    {
      "epoch": 0.008360107509805096,
      "grad_norm": 0.44795918464660645,
      "learning_rate": 0.00029874607236908334,
      "loss": 3.6345623779296874,
      "step": 14200
    },
    {
      "epoch": 0.008389544508079059,
      "grad_norm": 0.607596218585968,
      "learning_rate": 0.00029874165681999216,
      "loss": 3.610406494140625,
      "step": 14250
    },
    {
      "epoch": 0.00841898150635302,
      "grad_norm": 0.46461787819862366,
      "learning_rate": 0.00029873724127090097,
      "loss": 3.6695147705078126,
      "step": 14300
    },
    {
      "epoch": 0.008448418504626982,
      "grad_norm": 0.60659259557724,
      "learning_rate": 0.0002987328257218098,
      "loss": 3.627927551269531,
      "step": 14350
    },
    {
      "epoch": 0.008477855502900943,
      "grad_norm": 0.4924245774745941,
      "learning_rate": 0.0002987284101727186,
      "loss": 3.634527587890625,
      "step": 14400
    },
    {
      "epoch": 0.008507292501174905,
      "grad_norm": 0.5174017548561096,
      "learning_rate": 0.0002987239946236274,
      "loss": 3.639061279296875,
      "step": 14450
    },
    {
      "epoch": 0.008536729499448866,
      "grad_norm": 0.5955654382705688,
      "learning_rate": 0.0002987195790745362,
      "loss": 3.651937255859375,
      "step": 14500
    },
    {
      "epoch": 0.008566166497722827,
      "grad_norm": 0.4568476378917694,
      "learning_rate": 0.000298715163525445,
      "loss": 3.6458395385742186,
      "step": 14550
    },
    {
      "epoch": 0.008595603495996789,
      "grad_norm": 0.44520625472068787,
      "learning_rate": 0.00029871074797635386,
      "loss": 3.6363034057617187,
      "step": 14600
    },
    {
      "epoch": 0.00862504049427075,
      "grad_norm": 0.5278362035751343,
      "learning_rate": 0.0002987063324272626,
      "loss": 3.6151483154296873,
      "step": 14650
    },
    {
      "epoch": 0.008654477492544712,
      "grad_norm": 0.8412871360778809,
      "learning_rate": 0.00029870191687817143,
      "loss": 3.602203063964844,
      "step": 14700
    },
    {
      "epoch": 0.008683914490818673,
      "grad_norm": 0.4524139165878296,
      "learning_rate": 0.00029869750132908025,
      "loss": 3.644312744140625,
      "step": 14750
    },
    {
      "epoch": 0.008713351489092636,
      "grad_norm": 0.65975022315979,
      "learning_rate": 0.00029869308577998906,
      "loss": 3.6604104614257813,
      "step": 14800
    },
    {
      "epoch": 0.008742788487366596,
      "grad_norm": 0.5053302645683289,
      "learning_rate": 0.0002986886702308979,
      "loss": 3.6287246704101563,
      "step": 14850
    },
    {
      "epoch": 0.008772225485640559,
      "grad_norm": 0.4059031307697296,
      "learning_rate": 0.0002986842546818067,
      "loss": 3.5573049926757814,
      "step": 14900
    },
    {
      "epoch": 0.00880166248391452,
      "grad_norm": 0.3878139555454254,
      "learning_rate": 0.0002986798391327155,
      "loss": 3.5893215942382812,
      "step": 14950
    },
    {
      "epoch": 0.008831099482188482,
      "grad_norm": 0.5638760328292847,
      "learning_rate": 0.0002986754235836243,
      "loss": 3.5955633544921874,
      "step": 15000
    },
    {
      "epoch": 0.008860536480462443,
      "grad_norm": 0.3776034414768219,
      "learning_rate": 0.0002986710080345331,
      "loss": 3.630446472167969,
      "step": 15050
    },
    {
      "epoch": 0.008889973478736405,
      "grad_norm": 0.5500254034996033,
      "learning_rate": 0.0002986665924854419,
      "loss": 3.584858703613281,
      "step": 15100
    },
    {
      "epoch": 0.008919410477010366,
      "grad_norm": 0.40754443407058716,
      "learning_rate": 0.0002986621769363507,
      "loss": 3.6151776123046875,
      "step": 15150
    },
    {
      "epoch": 0.008948847475284329,
      "grad_norm": 0.48376139998435974,
      "learning_rate": 0.0002986577613872595,
      "loss": 3.608404541015625,
      "step": 15200
    },
    {
      "epoch": 0.00897828447355829,
      "grad_norm": 0.5149811506271362,
      "learning_rate": 0.00029865334583816834,
      "loss": 3.59355224609375,
      "step": 15250
    },
    {
      "epoch": 0.009007721471832252,
      "grad_norm": 0.5751956701278687,
      "learning_rate": 0.00029864893028907715,
      "loss": 3.579403076171875,
      "step": 15300
    },
    {
      "epoch": 0.009037158470106213,
      "grad_norm": 0.46526873111724854,
      "learning_rate": 0.00029864451473998597,
      "loss": 3.6673153686523436,
      "step": 15350
    },
    {
      "epoch": 0.009066595468380175,
      "grad_norm": 0.41156938672065735,
      "learning_rate": 0.00029864009919089473,
      "loss": 3.5787530517578126,
      "step": 15400
    },
    {
      "epoch": 0.009096032466654136,
      "grad_norm": 0.5940393805503845,
      "learning_rate": 0.00029863568364180354,
      "loss": 3.5977484130859376,
      "step": 15450
    },
    {
      "epoch": 0.009125469464928098,
      "grad_norm": 0.5436207056045532,
      "learning_rate": 0.00029863126809271236,
      "loss": 3.647479248046875,
      "step": 15500
    },
    {
      "epoch": 0.00915490646320206,
      "grad_norm": 0.42523619532585144,
      "learning_rate": 0.00029862685254362117,
      "loss": 3.6125033569335936,
      "step": 15550
    },
    {
      "epoch": 0.009184343461476022,
      "grad_norm": 0.4045310914516449,
      "learning_rate": 0.00029862243699453,
      "loss": 3.59714599609375,
      "step": 15600
    },
    {
      "epoch": 0.009213780459749982,
      "grad_norm": 0.42987343668937683,
      "learning_rate": 0.0002986180214454388,
      "loss": 3.679014892578125,
      "step": 15650
    },
    {
      "epoch": 0.009243217458023945,
      "grad_norm": 0.6422933340072632,
      "learning_rate": 0.0002986136058963476,
      "loss": 3.5856646728515624,
      "step": 15700
    },
    {
      "epoch": 0.009272654456297906,
      "grad_norm": 0.5232404470443726,
      "learning_rate": 0.00029860919034725643,
      "loss": 3.5996435546875,
      "step": 15750
    },
    {
      "epoch": 0.009302091454571868,
      "grad_norm": 0.3661065995693207,
      "learning_rate": 0.0002986047747981652,
      "loss": 3.5876702880859375,
      "step": 15800
    },
    {
      "epoch": 0.009331528452845829,
      "grad_norm": 0.43355804681777954,
      "learning_rate": 0.000298600359249074,
      "loss": 3.6162896728515626,
      "step": 15850
    },
    {
      "epoch": 0.009360965451119791,
      "grad_norm": 0.4002598524093628,
      "learning_rate": 0.0002985959436999829,
      "loss": 3.5765066528320313,
      "step": 15900
    },
    {
      "epoch": 0.009390402449393752,
      "grad_norm": 0.6322005987167358,
      "learning_rate": 0.00029859152815089163,
      "loss": 3.580166931152344,
      "step": 15950
    },
    {
      "epoch": 0.009419839447667715,
      "grad_norm": 0.5093302130699158,
      "learning_rate": 0.00029858711260180045,
      "loss": 3.528994445800781,
      "step": 16000
    },
    {
      "epoch": 0.009449276445941675,
      "grad_norm": 0.522774875164032,
      "learning_rate": 0.00029858269705270926,
      "loss": 3.631521911621094,
      "step": 16050
    },
    {
      "epoch": 0.009478713444215638,
      "grad_norm": 0.43726950883865356,
      "learning_rate": 0.0002985782815036181,
      "loss": 3.655953369140625,
      "step": 16100
    },
    {
      "epoch": 0.009508150442489599,
      "grad_norm": 0.468383252620697,
      "learning_rate": 0.0002985738659545269,
      "loss": 3.6045965576171874,
      "step": 16150
    },
    {
      "epoch": 0.009537587440763561,
      "grad_norm": 0.4689418375492096,
      "learning_rate": 0.0002985694504054357,
      "loss": 3.5463125610351565,
      "step": 16200
    },
    {
      "epoch": 0.009567024439037522,
      "grad_norm": 0.4200306236743927,
      "learning_rate": 0.0002985650348563445,
      "loss": 3.5866131591796875,
      "step": 16250
    },
    {
      "epoch": 0.009596461437311484,
      "grad_norm": 0.43396779894828796,
      "learning_rate": 0.00029856061930725334,
      "loss": 3.5495150756835936,
      "step": 16300
    },
    {
      "epoch": 0.009625898435585445,
      "grad_norm": 0.3856494426727295,
      "learning_rate": 0.0002985562037581621,
      "loss": 3.5594110107421875,
      "step": 16350
    },
    {
      "epoch": 0.009655335433859406,
      "grad_norm": 0.4668438732624054,
      "learning_rate": 0.0002985517882090709,
      "loss": 3.6121929931640624,
      "step": 16400
    },
    {
      "epoch": 0.009684772432133368,
      "grad_norm": 0.45074912905693054,
      "learning_rate": 0.0002985473726599797,
      "loss": 3.5758798217773435,
      "step": 16450
    },
    {
      "epoch": 0.00971420943040733,
      "grad_norm": 0.4053564965724945,
      "learning_rate": 0.00029854295711088854,
      "loss": 3.560591735839844,
      "step": 16500
    },
    {
      "epoch": 0.009743646428681292,
      "grad_norm": 0.4977295398712158,
      "learning_rate": 0.00029853854156179735,
      "loss": 3.5219677734375,
      "step": 16550
    },
    {
      "epoch": 0.009773083426955253,
      "grad_norm": 0.49057912826538086,
      "learning_rate": 0.00029853412601270617,
      "loss": 3.527646484375,
      "step": 16600
    },
    {
      "epoch": 0.009802520425229215,
      "grad_norm": 0.49237316846847534,
      "learning_rate": 0.000298529710463615,
      "loss": 3.615787353515625,
      "step": 16650
    },
    {
      "epoch": 0.009831957423503176,
      "grad_norm": 0.6688051223754883,
      "learning_rate": 0.0002985252949145238,
      "loss": 3.558399353027344,
      "step": 16700
    },
    {
      "epoch": 0.009861394421777138,
      "grad_norm": 0.4354502856731415,
      "learning_rate": 0.00029852087936543256,
      "loss": 3.4828875732421873,
      "step": 16750
    },
    {
      "epoch": 0.009890831420051099,
      "grad_norm": 0.5252079963684082,
      "learning_rate": 0.00029851646381634137,
      "loss": 3.5738836669921876,
      "step": 16800
    },
    {
      "epoch": 0.009920268418325062,
      "grad_norm": 0.5839505195617676,
      "learning_rate": 0.00029851204826725024,
      "loss": 3.5098342895507812,
      "step": 16850
    },
    {
      "epoch": 0.009949705416599022,
      "grad_norm": 0.5899139046669006,
      "learning_rate": 0.000298507632718159,
      "loss": 3.49849365234375,
      "step": 16900
    },
    {
      "epoch": 0.009979142414872985,
      "grad_norm": 0.5717675685882568,
      "learning_rate": 0.0002985032171690678,
      "loss": 3.5612716674804688,
      "step": 16950
    },
    {
      "epoch": 0.010008579413146946,
      "grad_norm": 0.5068641304969788,
      "learning_rate": 0.00029849880161997663,
      "loss": 3.564539794921875,
      "step": 17000
    },
    {
      "epoch": 0.010038016411420908,
      "grad_norm": 0.5135955214500427,
      "learning_rate": 0.00029849438607088544,
      "loss": 3.5552685546875,
      "step": 17050
    },
    {
      "epoch": 0.010067453409694869,
      "grad_norm": 0.7299718856811523,
      "learning_rate": 0.0002984899705217942,
      "loss": 3.5529974365234374,
      "step": 17100
    },
    {
      "epoch": 0.010096890407968831,
      "grad_norm": 0.4719160497188568,
      "learning_rate": 0.000298485554972703,
      "loss": 3.537672119140625,
      "step": 17150
    },
    {
      "epoch": 0.010126327406242792,
      "grad_norm": 0.40575429797172546,
      "learning_rate": 0.0002984811394236119,
      "loss": 3.541861267089844,
      "step": 17200
    },
    {
      "epoch": 0.010155764404516755,
      "grad_norm": 0.5006675124168396,
      "learning_rate": 0.00029847672387452065,
      "loss": 3.583106689453125,
      "step": 17250
    },
    {
      "epoch": 0.010185201402790715,
      "grad_norm": 0.37169545888900757,
      "learning_rate": 0.00029847230832542946,
      "loss": 3.5338613891601565,
      "step": 17300
    },
    {
      "epoch": 0.010214638401064678,
      "grad_norm": 0.47614967823028564,
      "learning_rate": 0.0002984678927763383,
      "loss": 3.4752316284179687,
      "step": 17350
    },
    {
      "epoch": 0.010244075399338639,
      "grad_norm": 0.46387287974357605,
      "learning_rate": 0.0002984634772272471,
      "loss": 3.5490206909179687,
      "step": 17400
    },
    {
      "epoch": 0.010273512397612601,
      "grad_norm": 0.4477062523365021,
      "learning_rate": 0.0002984590616781559,
      "loss": 3.54969970703125,
      "step": 17450
    },
    {
      "epoch": 0.010302949395886562,
      "grad_norm": 0.49527642130851746,
      "learning_rate": 0.0002984546461290647,
      "loss": 3.524862976074219,
      "step": 17500
    },
    {
      "epoch": 0.010332386394160524,
      "grad_norm": 0.48474085330963135,
      "learning_rate": 0.00029845023057997353,
      "loss": 3.5332382202148436,
      "step": 17550
    },
    {
      "epoch": 0.010361823392434485,
      "grad_norm": 0.41731753945350647,
      "learning_rate": 0.00029844581503088235,
      "loss": 3.5810397338867186,
      "step": 17600
    },
    {
      "epoch": 0.010391260390708448,
      "grad_norm": 0.3864580988883972,
      "learning_rate": 0.0002984413994817911,
      "loss": 3.561311950683594,
      "step": 17650
    },
    {
      "epoch": 0.010420697388982408,
      "grad_norm": 0.38801077008247375,
      "learning_rate": 0.0002984369839326999,
      "loss": 3.5356646728515626,
      "step": 17700
    },
    {
      "epoch": 0.010450134387256371,
      "grad_norm": 0.36691591143608093,
      "learning_rate": 0.00029843256838360874,
      "loss": 3.527373046875,
      "step": 17750
    },
    {
      "epoch": 0.010479571385530332,
      "grad_norm": 0.40595176815986633,
      "learning_rate": 0.00029842815283451755,
      "loss": 3.547958984375,
      "step": 17800
    },
    {
      "epoch": 0.010509008383804294,
      "grad_norm": 0.5037673115730286,
      "learning_rate": 0.00029842373728542637,
      "loss": 3.521947021484375,
      "step": 17850
    },
    {
      "epoch": 0.010538445382078255,
      "grad_norm": 0.48201414942741394,
      "learning_rate": 0.0002984193217363352,
      "loss": 3.5618252563476562,
      "step": 17900
    },
    {
      "epoch": 0.010567882380352217,
      "grad_norm": 0.49372074007987976,
      "learning_rate": 0.000298414906187244,
      "loss": 3.5487844848632815,
      "step": 17950
    },
    {
      "epoch": 0.010597319378626178,
      "grad_norm": 0.5821223855018616,
      "learning_rate": 0.0002984104906381528,
      "loss": 3.5350967407226563,
      "step": 18000
    },
    {
      "epoch": 0.01062675637690014,
      "grad_norm": 0.4274904131889343,
      "learning_rate": 0.00029840607508906157,
      "loss": 3.4852664184570314,
      "step": 18050
    },
    {
      "epoch": 0.010656193375174101,
      "grad_norm": 0.406267374753952,
      "learning_rate": 0.0002984016595399704,
      "loss": 3.573661193847656,
      "step": 18100
    },
    {
      "epoch": 0.010685630373448062,
      "grad_norm": 0.39492371678352356,
      "learning_rate": 0.00029839724399087925,
      "loss": 3.47896240234375,
      "step": 18150
    },
    {
      "epoch": 0.010715067371722025,
      "grad_norm": 0.5441620945930481,
      "learning_rate": 0.000298392828441788,
      "loss": 3.553616943359375,
      "step": 18200
    },
    {
      "epoch": 0.010744504369995985,
      "grad_norm": 0.48637712001800537,
      "learning_rate": 0.00029838841289269683,
      "loss": 3.560233459472656,
      "step": 18250
    },
    {
      "epoch": 0.010773941368269948,
      "grad_norm": 0.45168715715408325,
      "learning_rate": 0.00029838399734360564,
      "loss": 3.5423883056640624,
      "step": 18300
    },
    {
      "epoch": 0.010803378366543909,
      "grad_norm": 0.7381873726844788,
      "learning_rate": 0.00029837958179451446,
      "loss": 3.4583544921875,
      "step": 18350
    },
    {
      "epoch": 0.010832815364817871,
      "grad_norm": 0.38064125180244446,
      "learning_rate": 0.00029837516624542327,
      "loss": 3.538427734375,
      "step": 18400
    },
    {
      "epoch": 0.010862252363091832,
      "grad_norm": 0.4867687523365021,
      "learning_rate": 0.00029837075069633203,
      "loss": 3.5146575927734376,
      "step": 18450
    },
    {
      "epoch": 0.010891689361365794,
      "grad_norm": 0.44844138622283936,
      "learning_rate": 0.0002983663351472409,
      "loss": 3.44462158203125,
      "step": 18500
    },
    {
      "epoch": 0.010921126359639755,
      "grad_norm": 0.5791236162185669,
      "learning_rate": 0.0002983619195981497,
      "loss": 3.4751181030273437,
      "step": 18550
    },
    {
      "epoch": 0.010950563357913718,
      "grad_norm": 0.6232174038887024,
      "learning_rate": 0.0002983575040490585,
      "loss": 3.5045120239257814,
      "step": 18600
    },
    {
      "epoch": 0.010980000356187678,
      "grad_norm": 0.5178031921386719,
      "learning_rate": 0.0002983530884999673,
      "loss": 3.438559875488281,
      "step": 18650
    },
    {
      "epoch": 0.011009437354461641,
      "grad_norm": 0.5196633338928223,
      "learning_rate": 0.0002983486729508761,
      "loss": 3.489582214355469,
      "step": 18700
    },
    {
      "epoch": 0.011038874352735602,
      "grad_norm": 0.46026378870010376,
      "learning_rate": 0.0002983442574017849,
      "loss": 3.4725405883789064,
      "step": 18750
    },
    {
      "epoch": 0.011068311351009564,
      "grad_norm": 0.37205612659454346,
      "learning_rate": 0.00029833984185269373,
      "loss": 3.475272216796875,
      "step": 18800
    },
    {
      "epoch": 0.011097748349283525,
      "grad_norm": 0.49362805485725403,
      "learning_rate": 0.00029833542630360255,
      "loss": 3.5596636962890624,
      "step": 18850
    },
    {
      "epoch": 0.011127185347557488,
      "grad_norm": 0.5207547545433044,
      "learning_rate": 0.00029833101075451136,
      "loss": 3.5052130126953127,
      "step": 18900
    },
    {
      "epoch": 0.011156622345831448,
      "grad_norm": 0.4408878684043884,
      "learning_rate": 0.0002983265952054201,
      "loss": 3.5353207397460937,
      "step": 18950
    },
    {
      "epoch": 0.01118605934410541,
      "grad_norm": 0.4946652948856354,
      "learning_rate": 0.00029832217965632894,
      "loss": 3.5356231689453126,
      "step": 19000
    },
    {
      "epoch": 0.011215496342379372,
      "grad_norm": 0.5670718550682068,
      "learning_rate": 0.00029831776410723775,
      "loss": 3.489798583984375,
      "step": 19050
    },
    {
      "epoch": 0.011244933340653334,
      "grad_norm": 0.33738967776298523,
      "learning_rate": 0.00029831334855814657,
      "loss": 3.4970278930664063,
      "step": 19100
    },
    {
      "epoch": 0.011274370338927295,
      "grad_norm": 0.35943078994750977,
      "learning_rate": 0.0002983089330090554,
      "loss": 3.5354910278320313,
      "step": 19150
    },
    {
      "epoch": 0.011303807337201257,
      "grad_norm": 0.3899170458316803,
      "learning_rate": 0.0002983045174599642,
      "loss": 3.516681823730469,
      "step": 19200
    },
    {
      "epoch": 0.011333244335475218,
      "grad_norm": 0.4377695620059967,
      "learning_rate": 0.000298300101910873,
      "loss": 3.4598489379882813,
      "step": 19250
    },
    {
      "epoch": 0.01136268133374918,
      "grad_norm": 0.47681164741516113,
      "learning_rate": 0.0002982956863617818,
      "loss": 3.515033874511719,
      "step": 19300
    },
    {
      "epoch": 0.011392118332023141,
      "grad_norm": 0.5486381649971008,
      "learning_rate": 0.0002982912708126906,
      "loss": 3.409705810546875,
      "step": 19350
    },
    {
      "epoch": 0.011421555330297104,
      "grad_norm": 0.3963373899459839,
      "learning_rate": 0.0002982868552635994,
      "loss": 3.517978515625,
      "step": 19400
    },
    {
      "epoch": 0.011450992328571065,
      "grad_norm": 0.4109646677970886,
      "learning_rate": 0.00029828243971450827,
      "loss": 3.481412658691406,
      "step": 19450
    },
    {
      "epoch": 0.011480429326845027,
      "grad_norm": 0.3841334283351898,
      "learning_rate": 0.00029827802416541703,
      "loss": 3.553403015136719,
      "step": 19500
    },
    {
      "epoch": 0.011509866325118988,
      "grad_norm": 0.42860153317451477,
      "learning_rate": 0.00029827360861632584,
      "loss": 3.4732568359375,
      "step": 19550
    },
    {
      "epoch": 0.01153930332339295,
      "grad_norm": 0.4739302098751068,
      "learning_rate": 0.00029826919306723466,
      "loss": 3.513778076171875,
      "step": 19600
    },
    {
      "epoch": 0.011568740321666911,
      "grad_norm": 0.5933806896209717,
      "learning_rate": 0.00029826477751814347,
      "loss": 3.5428399658203125,
      "step": 19650
    },
    {
      "epoch": 0.011598177319940874,
      "grad_norm": 0.6243063807487488,
      "learning_rate": 0.0002982603619690523,
      "loss": 3.4831170654296875,
      "step": 19700
    },
    {
      "epoch": 0.011627614318214834,
      "grad_norm": 0.45110854506492615,
      "learning_rate": 0.00029825594641996105,
      "loss": 3.4830203247070313,
      "step": 19750
    },
    {
      "epoch": 0.011657051316488797,
      "grad_norm": 0.4324924945831299,
      "learning_rate": 0.0002982515308708699,
      "loss": 3.5214218139648437,
      "step": 19800
    },
    {
      "epoch": 0.011686488314762758,
      "grad_norm": 0.3741806447505951,
      "learning_rate": 0.00029824711532177873,
      "loss": 3.4795611572265623,
      "step": 19850
    },
    {
      "epoch": 0.01171592531303672,
      "grad_norm": 0.4048309028148651,
      "learning_rate": 0.0002982426997726875,
      "loss": 3.4825677490234375,
      "step": 19900
    },
    {
      "epoch": 0.011745362311310681,
      "grad_norm": 0.49649855494499207,
      "learning_rate": 0.0002982382842235963,
      "loss": 3.4073495483398437,
      "step": 19950
    },
    {
      "epoch": 0.011774799309584642,
      "grad_norm": 0.4935346245765686,
      "learning_rate": 0.0002982338686745051,
      "loss": 3.443994140625,
      "step": 20000
    },
    {
      "epoch": 0.011804236307858604,
      "grad_norm": 0.5009509325027466,
      "learning_rate": 0.00029822945312541393,
      "loss": 3.4948416137695313,
      "step": 20050
    },
    {
      "epoch": 0.011833673306132565,
      "grad_norm": 0.43454840779304504,
      "learning_rate": 0.00029822503757632275,
      "loss": 3.4720803833007814,
      "step": 20100
    },
    {
      "epoch": 0.011863110304406527,
      "grad_norm": 0.5259291529655457,
      "learning_rate": 0.00029822062202723156,
      "loss": 3.5045376586914063,
      "step": 20150
    },
    {
      "epoch": 0.011892547302680488,
      "grad_norm": 0.39222630858421326,
      "learning_rate": 0.0002982162064781404,
      "loss": 3.4370843505859376,
      "step": 20200
    },
    {
      "epoch": 0.01192198430095445,
      "grad_norm": 0.596000075340271,
      "learning_rate": 0.0002982117909290492,
      "loss": 3.514737243652344,
      "step": 20250
    },
    {
      "epoch": 0.011951421299228411,
      "grad_norm": 0.37992945313453674,
      "learning_rate": 0.00029820737537995795,
      "loss": 3.4894146728515625,
      "step": 20300
    },
    {
      "epoch": 0.011980858297502374,
      "grad_norm": 0.4079105854034424,
      "learning_rate": 0.00029820295983086677,
      "loss": 3.490022888183594,
      "step": 20350
    },
    {
      "epoch": 0.012010295295776335,
      "grad_norm": 0.4086385667324066,
      "learning_rate": 0.0002981985442817756,
      "loss": 3.459764099121094,
      "step": 20400
    },
    {
      "epoch": 0.012039732294050297,
      "grad_norm": 0.434776246547699,
      "learning_rate": 0.0002981941287326844,
      "loss": 3.479716796875,
      "step": 20450
    },
    {
      "epoch": 0.012069169292324258,
      "grad_norm": 0.42425671219825745,
      "learning_rate": 0.0002981897131835932,
      "loss": 3.45964111328125,
      "step": 20500
    },
    {
      "epoch": 0.01209860629059822,
      "grad_norm": 0.5245711207389832,
      "learning_rate": 0.000298185297634502,
      "loss": 3.4733544921875,
      "step": 20550
    },
    {
      "epoch": 0.012128043288872181,
      "grad_norm": 0.5935816168785095,
      "learning_rate": 0.00029818088208541084,
      "loss": 3.5029949951171875,
      "step": 20600
    },
    {
      "epoch": 0.012157480287146144,
      "grad_norm": 0.6593071818351746,
      "learning_rate": 0.0002981764665363196,
      "loss": 3.464559326171875,
      "step": 20650
    },
    {
      "epoch": 0.012186917285420104,
      "grad_norm": 0.35918304324150085,
      "learning_rate": 0.0002981720509872284,
      "loss": 3.4420980834960937,
      "step": 20700
    },
    {
      "epoch": 0.012216354283694067,
      "grad_norm": 0.47429054975509644,
      "learning_rate": 0.0002981676354381373,
      "loss": 3.4344183349609376,
      "step": 20750
    },
    {
      "epoch": 0.012245791281968028,
      "grad_norm": 0.44259122014045715,
      "learning_rate": 0.00029816321988904604,
      "loss": 3.4902572631835938,
      "step": 20800
    },
    {
      "epoch": 0.01227522828024199,
      "grad_norm": 0.5009517073631287,
      "learning_rate": 0.00029815880433995486,
      "loss": 3.4809439086914065,
      "step": 20850
    },
    {
      "epoch": 0.012304665278515951,
      "grad_norm": 0.4942546784877777,
      "learning_rate": 0.00029815438879086367,
      "loss": 3.43767822265625,
      "step": 20900
    },
    {
      "epoch": 0.012334102276789913,
      "grad_norm": 0.5792289972305298,
      "learning_rate": 0.0002981499732417725,
      "loss": 3.46686279296875,
      "step": 20950
    },
    {
      "epoch": 0.012363539275063874,
      "grad_norm": 0.4464184045791626,
      "learning_rate": 0.0002981455576926813,
      "loss": 3.49584716796875,
      "step": 21000
    },
    {
      "epoch": 0.012392976273337837,
      "grad_norm": 0.4120526611804962,
      "learning_rate": 0.00029814114214359006,
      "loss": 3.47660888671875,
      "step": 21050
    },
    {
      "epoch": 0.012422413271611797,
      "grad_norm": 0.4187904894351959,
      "learning_rate": 0.00029813672659449893,
      "loss": 3.4622796630859374,
      "step": 21100
    },
    {
      "epoch": 0.01245185026988576,
      "grad_norm": 0.49883967638015747,
      "learning_rate": 0.00029813231104540774,
      "loss": 3.458304443359375,
      "step": 21150
    },
    {
      "epoch": 0.01248128726815972,
      "grad_norm": 0.4888683557510376,
      "learning_rate": 0.0002981278954963165,
      "loss": 3.4426510620117186,
      "step": 21200
    },
    {
      "epoch": 0.012510724266433683,
      "grad_norm": 0.5375713109970093,
      "learning_rate": 0.0002981234799472253,
      "loss": 3.420921630859375,
      "step": 21250
    },
    {
      "epoch": 0.012540161264707644,
      "grad_norm": 0.4862951338291168,
      "learning_rate": 0.00029811906439813413,
      "loss": 3.4694705200195313,
      "step": 21300
    },
    {
      "epoch": 0.012569598262981607,
      "grad_norm": 0.5152239799499512,
      "learning_rate": 0.00029811464884904295,
      "loss": 3.4707720947265623,
      "step": 21350
    },
    {
      "epoch": 0.012599035261255567,
      "grad_norm": 0.4863365590572357,
      "learning_rate": 0.00029811023329995176,
      "loss": 3.435427551269531,
      "step": 21400
    },
    {
      "epoch": 0.01262847225952953,
      "grad_norm": 0.726618766784668,
      "learning_rate": 0.0002981058177508606,
      "loss": 3.467184753417969,
      "step": 21450
    },
    {
      "epoch": 0.01265790925780349,
      "grad_norm": 0.6539440155029297,
      "learning_rate": 0.0002981014022017694,
      "loss": 3.445046081542969,
      "step": 21500
    },
    {
      "epoch": 0.012687346256077453,
      "grad_norm": 0.3547012507915497,
      "learning_rate": 0.0002980969866526782,
      "loss": 3.4722955322265623,
      "step": 21550
    },
    {
      "epoch": 0.012716783254351414,
      "grad_norm": 0.6535599231719971,
      "learning_rate": 0.00029809257110358697,
      "loss": 3.4239157104492186,
      "step": 21600
    },
    {
      "epoch": 0.012746220252625376,
      "grad_norm": 0.38955897092819214,
      "learning_rate": 0.0002980881555544958,
      "loss": 3.4442071533203125,
      "step": 21650
    },
    {
      "epoch": 0.012775657250899337,
      "grad_norm": 0.44046950340270996,
      "learning_rate": 0.0002980837400054046,
      "loss": 3.44865966796875,
      "step": 21700
    },
    {
      "epoch": 0.0128050942491733,
      "grad_norm": 0.3660692572593689,
      "learning_rate": 0.0002980793244563134,
      "loss": 3.424599304199219,
      "step": 21750
    },
    {
      "epoch": 0.01283453124744726,
      "grad_norm": 0.48850390315055847,
      "learning_rate": 0.0002980749089072222,
      "loss": 3.4641839599609376,
      "step": 21800
    },
    {
      "epoch": 0.012863968245721221,
      "grad_norm": 0.5812660455703735,
      "learning_rate": 0.00029807049335813104,
      "loss": 3.4295440673828126,
      "step": 21850
    },
    {
      "epoch": 0.012893405243995184,
      "grad_norm": 0.3338938355445862,
      "learning_rate": 0.00029806607780903985,
      "loss": 3.4390789794921877,
      "step": 21900
    },
    {
      "epoch": 0.012922842242269144,
      "grad_norm": 0.4721289277076721,
      "learning_rate": 0.00029806166225994867,
      "loss": 3.403880920410156,
      "step": 21950
    },
    {
      "epoch": 0.012952279240543107,
      "grad_norm": 0.30594301223754883,
      "learning_rate": 0.00029805724671085743,
      "loss": 3.4181478881835936,
      "step": 22000
    },
    {
      "epoch": 0.012981716238817068,
      "grad_norm": 0.5043311715126038,
      "learning_rate": 0.00029805283116176624,
      "loss": 3.4707977294921877,
      "step": 22050
    },
    {
      "epoch": 0.01301115323709103,
      "grad_norm": 0.7442666292190552,
      "learning_rate": 0.0002980484156126751,
      "loss": 3.4397872924804687,
      "step": 22100
    },
    {
      "epoch": 0.01304059023536499,
      "grad_norm": 0.45529353618621826,
      "learning_rate": 0.00029804400006358387,
      "loss": 3.3912274169921877,
      "step": 22150
    },
    {
      "epoch": 0.013070027233638953,
      "grad_norm": 0.4400083124637604,
      "learning_rate": 0.0002980395845144927,
      "loss": 3.410342102050781,
      "step": 22200
    },
    {
      "epoch": 0.013099464231912914,
      "grad_norm": 0.8006680607795715,
      "learning_rate": 0.0002980351689654015,
      "loss": 3.476138000488281,
      "step": 22250
    },
    {
      "epoch": 0.013128901230186877,
      "grad_norm": 0.5463877320289612,
      "learning_rate": 0.0002980307534163103,
      "loss": 3.4059017944335936,
      "step": 22300
    },
    {
      "epoch": 0.013158338228460837,
      "grad_norm": 0.44419100880622864,
      "learning_rate": 0.0002980263378672191,
      "loss": 3.4625335693359376,
      "step": 22350
    },
    {
      "epoch": 0.0131877752267348,
      "grad_norm": 0.6025996208190918,
      "learning_rate": 0.00029802192231812794,
      "loss": 3.3805877685546877,
      "step": 22400
    },
    {
      "epoch": 0.01321721222500876,
      "grad_norm": 0.45474642515182495,
      "learning_rate": 0.00029801750676903676,
      "loss": 3.3783990478515626,
      "step": 22450
    },
    {
      "epoch": 0.013246649223282723,
      "grad_norm": 0.5288981199264526,
      "learning_rate": 0.0002980130912199455,
      "loss": 3.4146542358398437,
      "step": 22500
    },
    {
      "epoch": 0.013276086221556684,
      "grad_norm": 0.3935588300228119,
      "learning_rate": 0.00029800867567085433,
      "loss": 3.418365478515625,
      "step": 22550
    },
    {
      "epoch": 0.013305523219830646,
      "grad_norm": 0.5212574005126953,
      "learning_rate": 0.00029800426012176315,
      "loss": 3.380755615234375,
      "step": 22600
    },
    {
      "epoch": 0.013334960218104607,
      "grad_norm": 0.5719454884529114,
      "learning_rate": 0.00029799984457267196,
      "loss": 3.37811767578125,
      "step": 22650
    },
    {
      "epoch": 0.01336439721637857,
      "grad_norm": 0.38075482845306396,
      "learning_rate": 0.0002979954290235808,
      "loss": 3.441306457519531,
      "step": 22700
    },
    {
      "epoch": 0.01339383421465253,
      "grad_norm": 0.355879545211792,
      "learning_rate": 0.0002979910134744896,
      "loss": 3.4041323852539063,
      "step": 22750
    },
    {
      "epoch": 0.013423271212926493,
      "grad_norm": 0.4075736403465271,
      "learning_rate": 0.0002979865979253984,
      "loss": 3.448081359863281,
      "step": 22800
    },
    {
      "epoch": 0.013452708211200454,
      "grad_norm": 0.5658280849456787,
      "learning_rate": 0.0002979821823763072,
      "loss": 3.4567001342773436,
      "step": 22850
    },
    {
      "epoch": 0.013482145209474416,
      "grad_norm": 0.38906845450401306,
      "learning_rate": 0.000297977766827216,
      "loss": 3.450931396484375,
      "step": 22900
    },
    {
      "epoch": 0.013511582207748377,
      "grad_norm": 0.43356847763061523,
      "learning_rate": 0.0002979733512781248,
      "loss": 3.4321600341796876,
      "step": 22950
    },
    {
      "epoch": 0.01354101920602234,
      "grad_norm": 0.430110901594162,
      "learning_rate": 0.0002979689357290336,
      "loss": 3.439841613769531,
      "step": 23000
    },
    {
      "epoch": 0.0135704562042963,
      "grad_norm": 0.44754284620285034,
      "learning_rate": 0.0002979645201799424,
      "loss": 3.344889831542969,
      "step": 23050
    },
    {
      "epoch": 0.013599893202570263,
      "grad_norm": 0.47216561436653137,
      "learning_rate": 0.00029796010463085124,
      "loss": 3.4178656005859374,
      "step": 23100
    },
    {
      "epoch": 0.013629330200844223,
      "grad_norm": 0.46415188908576965,
      "learning_rate": 0.00029795568908176005,
      "loss": 3.4077655029296876,
      "step": 23150
    },
    {
      "epoch": 0.013658767199118186,
      "grad_norm": 0.577034592628479,
      "learning_rate": 0.00029795127353266887,
      "loss": 3.4405526733398437,
      "step": 23200
    },
    {
      "epoch": 0.013688204197392147,
      "grad_norm": 0.4615389108657837,
      "learning_rate": 0.0002979468579835777,
      "loss": 3.4445947265625,
      "step": 23250
    },
    {
      "epoch": 0.01371764119566611,
      "grad_norm": 0.43379858136177063,
      "learning_rate": 0.00029794244243448644,
      "loss": 3.4423086547851565,
      "step": 23300
    },
    {
      "epoch": 0.01374707819394007,
      "grad_norm": 0.3689925968647003,
      "learning_rate": 0.00029793802688539526,
      "loss": 3.420665283203125,
      "step": 23350
    },
    {
      "epoch": 0.013776515192214032,
      "grad_norm": 0.4253310561180115,
      "learning_rate": 0.0002979336113363041,
      "loss": 3.4298782348632812,
      "step": 23400
    },
    {
      "epoch": 0.013805952190487993,
      "grad_norm": 0.5310836434364319,
      "learning_rate": 0.0002979291957872129,
      "loss": 3.394134216308594,
      "step": 23450
    },
    {
      "epoch": 0.013835389188761956,
      "grad_norm": 0.43368637561798096,
      "learning_rate": 0.0002979247802381217,
      "loss": 3.399542236328125,
      "step": 23500
    },
    {
      "epoch": 0.013864826187035917,
      "grad_norm": 0.3927505910396576,
      "learning_rate": 0.0002979203646890305,
      "loss": 3.3879150390625,
      "step": 23550
    },
    {
      "epoch": 0.013894263185309877,
      "grad_norm": 0.4290965497493744,
      "learning_rate": 0.00029791594913993933,
      "loss": 3.394036865234375,
      "step": 23600
    },
    {
      "epoch": 0.01392370018358384,
      "grad_norm": 0.5398778915405273,
      "learning_rate": 0.00029791153359084814,
      "loss": 3.356642150878906,
      "step": 23650
    },
    {
      "epoch": 0.0139531371818578,
      "grad_norm": 0.3859393298625946,
      "learning_rate": 0.00029790711804175696,
      "loss": 3.431065673828125,
      "step": 23700
    },
    {
      "epoch": 0.013982574180131763,
      "grad_norm": 0.727910041809082,
      "learning_rate": 0.00029790270249266577,
      "loss": 3.4031881713867187,
      "step": 23750
    },
    {
      "epoch": 0.014012011178405724,
      "grad_norm": 0.515120804309845,
      "learning_rate": 0.0002978982869435746,
      "loss": 3.3853668212890624,
      "step": 23800
    },
    {
      "epoch": 0.014041448176679686,
      "grad_norm": 0.518987774848938,
      "learning_rate": 0.00029789387139448335,
      "loss": 3.400589599609375,
      "step": 23850
    },
    {
      "epoch": 0.014070885174953647,
      "grad_norm": 0.45390310883522034,
      "learning_rate": 0.00029788945584539216,
      "loss": 3.3766375732421876,
      "step": 23900
    },
    {
      "epoch": 0.01410032217322761,
      "grad_norm": 0.5676884651184082,
      "learning_rate": 0.000297885040296301,
      "loss": 3.365919189453125,
      "step": 23950
    },
    {
      "epoch": 0.01412975917150157,
      "grad_norm": 0.4782746434211731,
      "learning_rate": 0.0002978806247472098,
      "loss": 3.386842956542969,
      "step": 24000
    },
    {
      "epoch": 0.014159196169775533,
      "grad_norm": 0.3966592848300934,
      "learning_rate": 0.0002978762091981186,
      "loss": 3.4607968139648437,
      "step": 24050
    },
    {
      "epoch": 0.014188633168049494,
      "grad_norm": 0.44004762172698975,
      "learning_rate": 0.0002978717936490274,
      "loss": 3.3500274658203124,
      "step": 24100
    },
    {
      "epoch": 0.014218070166323456,
      "grad_norm": 0.39873751997947693,
      "learning_rate": 0.00029786737809993623,
      "loss": 3.334212341308594,
      "step": 24150
    },
    {
      "epoch": 0.014247507164597417,
      "grad_norm": 0.4278111159801483,
      "learning_rate": 0.000297862962550845,
      "loss": 3.385756530761719,
      "step": 24200
    },
    {
      "epoch": 0.01427694416287138,
      "grad_norm": 0.4314025342464447,
      "learning_rate": 0.0002978585470017538,
      "loss": 3.374639892578125,
      "step": 24250
    },
    {
      "epoch": 0.01430638116114534,
      "grad_norm": 0.6482324600219727,
      "learning_rate": 0.0002978541314526626,
      "loss": 3.378689880371094,
      "step": 24300
    },
    {
      "epoch": 0.014335818159419303,
      "grad_norm": 0.4158913493156433,
      "learning_rate": 0.00029784971590357144,
      "loss": 3.4083786010742188,
      "step": 24350
    },
    {
      "epoch": 0.014365255157693263,
      "grad_norm": 0.42048919200897217,
      "learning_rate": 0.00029784530035448025,
      "loss": 3.384471435546875,
      "step": 24400
    },
    {
      "epoch": 0.014394692155967226,
      "grad_norm": 0.5253003835678101,
      "learning_rate": 0.00029784088480538907,
      "loss": 3.3694854736328126,
      "step": 24450
    },
    {
      "epoch": 0.014424129154241187,
      "grad_norm": 0.4466915726661682,
      "learning_rate": 0.0002978364692562979,
      "loss": 3.3921905517578126,
      "step": 24500
    },
    {
      "epoch": 0.014453566152515149,
      "grad_norm": 0.4002355635166168,
      "learning_rate": 0.0002978320537072067,
      "loss": 3.395529479980469,
      "step": 24550
    },
    {
      "epoch": 0.01448300315078911,
      "grad_norm": 0.38272619247436523,
      "learning_rate": 0.00029782763815811546,
      "loss": 3.375052185058594,
      "step": 24600
    },
    {
      "epoch": 0.014512440149063072,
      "grad_norm": 0.3593185245990753,
      "learning_rate": 0.00029782322260902427,
      "loss": 3.40505615234375,
      "step": 24650
    },
    {
      "epoch": 0.014541877147337033,
      "grad_norm": 0.3784303069114685,
      "learning_rate": 0.00029781880705993314,
      "loss": 3.4301278686523435,
      "step": 24700
    },
    {
      "epoch": 0.014571314145610996,
      "grad_norm": 0.4828411638736725,
      "learning_rate": 0.0002978143915108419,
      "loss": 3.433629455566406,
      "step": 24750
    },
    {
      "epoch": 0.014600751143884956,
      "grad_norm": 0.6270247101783752,
      "learning_rate": 0.0002978099759617507,
      "loss": 3.403695068359375,
      "step": 24800
    },
    {
      "epoch": 0.014630188142158919,
      "grad_norm": 0.34578442573547363,
      "learning_rate": 0.00029780556041265953,
      "loss": 3.3867828369140627,
      "step": 24850
    },
    {
      "epoch": 0.01465962514043288,
      "grad_norm": 0.38490423560142517,
      "learning_rate": 0.00029780114486356834,
      "loss": 3.372447204589844,
      "step": 24900
    },
    {
      "epoch": 0.014689062138706842,
      "grad_norm": 0.486529141664505,
      "learning_rate": 0.00029779672931447716,
      "loss": 3.3599844360351563,
      "step": 24950
    },
    {
      "epoch": 0.014718499136980803,
      "grad_norm": 0.4052560031414032,
      "learning_rate": 0.00029779231376538597,
      "loss": 3.317383117675781,
      "step": 25000
    },
    {
      "epoch": 0.014747936135254765,
      "grad_norm": 0.45657026767730713,
      "learning_rate": 0.0002977878982162948,
      "loss": 3.4177618408203125,
      "step": 25050
    },
    {
      "epoch": 0.014777373133528726,
      "grad_norm": 0.4954637587070465,
      "learning_rate": 0.0002977834826672036,
      "loss": 3.383525390625,
      "step": 25100
    },
    {
      "epoch": 0.014806810131802689,
      "grad_norm": 0.3853021264076233,
      "learning_rate": 0.00029777906711811236,
      "loss": 3.3750616455078126,
      "step": 25150
    },
    {
      "epoch": 0.01483624713007665,
      "grad_norm": 0.430012971162796,
      "learning_rate": 0.0002977746515690212,
      "loss": 3.4155706787109374,
      "step": 25200
    },
    {
      "epoch": 0.014865684128350612,
      "grad_norm": 0.540747344493866,
      "learning_rate": 0.00029777023601993,
      "loss": 3.396037292480469,
      "step": 25250
    },
    {
      "epoch": 0.014895121126624573,
      "grad_norm": 0.4994257986545563,
      "learning_rate": 0.0002977658204708388,
      "loss": 3.377688293457031,
      "step": 25300
    },
    {
      "epoch": 0.014924558124898535,
      "grad_norm": 0.4284316897392273,
      "learning_rate": 0.0002977614049217476,
      "loss": 3.3426641845703124,
      "step": 25350
    },
    {
      "epoch": 0.014953995123172496,
      "grad_norm": 0.4560491740703583,
      "learning_rate": 0.00029775698937265643,
      "loss": 3.356062927246094,
      "step": 25400
    },
    {
      "epoch": 0.014983432121446457,
      "grad_norm": 0.40159013867378235,
      "learning_rate": 0.00029775257382356525,
      "loss": 3.3540130615234376,
      "step": 25450
    },
    {
      "epoch": 0.01501286911972042,
      "grad_norm": 0.4888107180595398,
      "learning_rate": 0.00029774815827447406,
      "loss": 3.361806640625,
      "step": 25500
    },
    {
      "epoch": 0.01504230611799438,
      "grad_norm": 0.512381911277771,
      "learning_rate": 0.0002977437427253828,
      "loss": 3.4839321899414064,
      "step": 25550
    },
    {
      "epoch": 0.015071743116268342,
      "grad_norm": 0.3701493442058563,
      "learning_rate": 0.00029773932717629164,
      "loss": 3.411085510253906,
      "step": 25600
    },
    {
      "epoch": 0.015101180114542303,
      "grad_norm": 0.3493577837944031,
      "learning_rate": 0.0002977349116272005,
      "loss": 3.380983581542969,
      "step": 25650
    },
    {
      "epoch": 0.015130617112816266,
      "grad_norm": 0.3666781783103943,
      "learning_rate": 0.00029773049607810927,
      "loss": 3.397652893066406,
      "step": 25700
    },
    {
      "epoch": 0.015160054111090226,
      "grad_norm": 0.554735004901886,
      "learning_rate": 0.0002977260805290181,
      "loss": 3.3539276123046875,
      "step": 25750
    },
    {
      "epoch": 0.015189491109364189,
      "grad_norm": 0.5524623990058899,
      "learning_rate": 0.0002977216649799269,
      "loss": 3.3955001831054688,
      "step": 25800
    },
    {
      "epoch": 0.01521892810763815,
      "grad_norm": 0.5510681867599487,
      "learning_rate": 0.0002977172494308357,
      "loss": 3.3120462036132814,
      "step": 25850
    },
    {
      "epoch": 0.015248365105912112,
      "grad_norm": 0.3941030502319336,
      "learning_rate": 0.00029771283388174447,
      "loss": 3.3690505981445313,
      "step": 25900
    },
    {
      "epoch": 0.015277802104186073,
      "grad_norm": 0.47000548243522644,
      "learning_rate": 0.0002977084183326533,
      "loss": 3.365960998535156,
      "step": 25950
    },
    {
      "epoch": 0.015307239102460036,
      "grad_norm": 0.40282467007637024,
      "learning_rate": 0.00029770400278356215,
      "loss": 3.371784973144531,
      "step": 26000
    },
    {
      "epoch": 0.015336676100733996,
      "grad_norm": 0.4420921504497528,
      "learning_rate": 0.0002976995872344709,
      "loss": 3.314345397949219,
      "step": 26050
    },
    {
      "epoch": 0.015366113099007959,
      "grad_norm": 0.43119996786117554,
      "learning_rate": 0.00029769517168537973,
      "loss": 3.364818115234375,
      "step": 26100
    },
    {
      "epoch": 0.01539555009728192,
      "grad_norm": 0.3786391019821167,
      "learning_rate": 0.00029769075613628854,
      "loss": 3.3849285888671874,
      "step": 26150
    },
    {
      "epoch": 0.015424987095555882,
      "grad_norm": 0.5491346120834351,
      "learning_rate": 0.00029768634058719736,
      "loss": 3.418643798828125,
      "step": 26200
    },
    {
      "epoch": 0.015454424093829843,
      "grad_norm": 0.3066082000732422,
      "learning_rate": 0.00029768192503810617,
      "loss": 3.3546438598632813,
      "step": 26250
    },
    {
      "epoch": 0.015483861092103805,
      "grad_norm": 0.49278736114501953,
      "learning_rate": 0.000297677509489015,
      "loss": 3.336306457519531,
      "step": 26300
    },
    {
      "epoch": 0.015513298090377766,
      "grad_norm": 1.1000179052352905,
      "learning_rate": 0.0002976730939399238,
      "loss": 3.3506759643554687,
      "step": 26350
    },
    {
      "epoch": 0.015542735088651729,
      "grad_norm": 0.5370403528213501,
      "learning_rate": 0.0002976686783908326,
      "loss": 3.4566995239257814,
      "step": 26400
    },
    {
      "epoch": 0.01557217208692569,
      "grad_norm": 0.5099030137062073,
      "learning_rate": 0.0002976642628417414,
      "loss": 3.366683349609375,
      "step": 26450
    },
    {
      "epoch": 0.015601609085199652,
      "grad_norm": 0.3194481432437897,
      "learning_rate": 0.0002976598472926502,
      "loss": 3.3628201293945312,
      "step": 26500
    },
    {
      "epoch": 0.015631046083473613,
      "grad_norm": 0.3804369866847992,
      "learning_rate": 0.000297655431743559,
      "loss": 3.334012145996094,
      "step": 26550
    },
    {
      "epoch": 0.015660483081747573,
      "grad_norm": 0.49965882301330566,
      "learning_rate": 0.0002976510161944678,
      "loss": 3.2990542602539064,
      "step": 26600
    },
    {
      "epoch": 0.015689920080021538,
      "grad_norm": 0.48515787720680237,
      "learning_rate": 0.00029764660064537663,
      "loss": 3.376439208984375,
      "step": 26650
    },
    {
      "epoch": 0.0157193570782955,
      "grad_norm": 0.42730218172073364,
      "learning_rate": 0.00029764218509628545,
      "loss": 3.4149545288085936,
      "step": 26700
    },
    {
      "epoch": 0.01574879407656946,
      "grad_norm": 0.613269031047821,
      "learning_rate": 0.00029763776954719426,
      "loss": 3.362144775390625,
      "step": 26750
    },
    {
      "epoch": 0.01577823107484342,
      "grad_norm": 0.3500162363052368,
      "learning_rate": 0.0002976333539981031,
      "loss": 3.331238098144531,
      "step": 26800
    },
    {
      "epoch": 0.015807668073117384,
      "grad_norm": 0.41752153635025024,
      "learning_rate": 0.00029762893844901184,
      "loss": 3.3848641967773436,
      "step": 26850
    },
    {
      "epoch": 0.015837105071391345,
      "grad_norm": 0.35849088430404663,
      "learning_rate": 0.00029762452289992065,
      "loss": 3.4043838500976564,
      "step": 26900
    },
    {
      "epoch": 0.015866542069665306,
      "grad_norm": 0.41820886731147766,
      "learning_rate": 0.0002976201073508295,
      "loss": 3.3487591552734375,
      "step": 26950
    },
    {
      "epoch": 0.015895979067939266,
      "grad_norm": 0.578575074672699,
      "learning_rate": 0.0002976156918017383,
      "loss": 3.3172650146484375,
      "step": 27000
    },
    {
      "epoch": 0.01592541606621323,
      "grad_norm": 0.33827081322669983,
      "learning_rate": 0.0002976112762526471,
      "loss": 3.348172607421875,
      "step": 27050
    },
    {
      "epoch": 0.01595485306448719,
      "grad_norm": 0.726349413394928,
      "learning_rate": 0.0002976068607035559,
      "loss": 3.3456759643554688,
      "step": 27100
    },
    {
      "epoch": 0.015984290062761152,
      "grad_norm": 0.4574359059333801,
      "learning_rate": 0.0002976024451544647,
      "loss": 3.3350204467773437,
      "step": 27150
    },
    {
      "epoch": 0.016013727061035113,
      "grad_norm": 0.3130719065666199,
      "learning_rate": 0.00029759802960537354,
      "loss": 3.3717822265625,
      "step": 27200
    },
    {
      "epoch": 0.016043164059309074,
      "grad_norm": 0.40829357504844666,
      "learning_rate": 0.0002975936140562823,
      "loss": 3.3468899536132812,
      "step": 27250
    },
    {
      "epoch": 0.016072601057583038,
      "grad_norm": 0.4914036989212036,
      "learning_rate": 0.00029758919850719117,
      "loss": 3.372924499511719,
      "step": 27300
    },
    {
      "epoch": 0.016102038055857,
      "grad_norm": 0.4152793884277344,
      "learning_rate": 0.0002975847829581,
      "loss": 3.3826412963867187,
      "step": 27350
    },
    {
      "epoch": 0.01613147505413096,
      "grad_norm": 0.39618635177612305,
      "learning_rate": 0.00029758036740900874,
      "loss": 3.354123229980469,
      "step": 27400
    },
    {
      "epoch": 0.01616091205240492,
      "grad_norm": 0.5938746333122253,
      "learning_rate": 0.00029757595185991756,
      "loss": 3.37505859375,
      "step": 27450
    },
    {
      "epoch": 0.016190349050678884,
      "grad_norm": 0.4311826825141907,
      "learning_rate": 0.00029757153631082637,
      "loss": 3.376994323730469,
      "step": 27500
    },
    {
      "epoch": 0.016219786048952845,
      "grad_norm": 0.43647196888923645,
      "learning_rate": 0.0002975671207617352,
      "loss": 3.3455404663085937,
      "step": 27550
    },
    {
      "epoch": 0.016249223047226806,
      "grad_norm": 0.388649582862854,
      "learning_rate": 0.000297562705212644,
      "loss": 3.3337249755859375,
      "step": 27600
    },
    {
      "epoch": 0.016278660045500767,
      "grad_norm": 0.33769577741622925,
      "learning_rate": 0.0002975582896635528,
      "loss": 3.35056396484375,
      "step": 27650
    },
    {
      "epoch": 0.01630809704377473,
      "grad_norm": 0.31094321608543396,
      "learning_rate": 0.00029755387411446163,
      "loss": 3.348333740234375,
      "step": 27700
    },
    {
      "epoch": 0.01633753404204869,
      "grad_norm": 0.3767419755458832,
      "learning_rate": 0.0002975494585653704,
      "loss": 3.302420654296875,
      "step": 27750
    },
    {
      "epoch": 0.016366971040322652,
      "grad_norm": 0.4392068386077881,
      "learning_rate": 0.0002975450430162792,
      "loss": 3.33163330078125,
      "step": 27800
    },
    {
      "epoch": 0.016396408038596613,
      "grad_norm": 0.3817487955093384,
      "learning_rate": 0.000297540627467188,
      "loss": 3.3207916259765624,
      "step": 27850
    },
    {
      "epoch": 0.016425845036870577,
      "grad_norm": 0.5849345922470093,
      "learning_rate": 0.00029753621191809683,
      "loss": 3.3781588745117186,
      "step": 27900
    },
    {
      "epoch": 0.016455282035144538,
      "grad_norm": 0.4472985863685608,
      "learning_rate": 0.00029753179636900565,
      "loss": 3.37745361328125,
      "step": 27950
    },
    {
      "epoch": 0.0164847190334185,
      "grad_norm": 0.37014859914779663,
      "learning_rate": 0.00029752738081991446,
      "loss": 3.331134033203125,
      "step": 28000
    },
    {
      "epoch": 0.01651415603169246,
      "grad_norm": 0.5482437014579773,
      "learning_rate": 0.0002975229652708233,
      "loss": 3.349537353515625,
      "step": 28050
    },
    {
      "epoch": 0.016543593029966424,
      "grad_norm": 0.3221342861652374,
      "learning_rate": 0.0002975185497217321,
      "loss": 3.343796691894531,
      "step": 28100
    },
    {
      "epoch": 0.016573030028240385,
      "grad_norm": 0.3486942648887634,
      "learning_rate": 0.00029751413417264085,
      "loss": 3.360440673828125,
      "step": 28150
    },
    {
      "epoch": 0.016602467026514345,
      "grad_norm": 0.35363051295280457,
      "learning_rate": 0.00029750971862354966,
      "loss": 3.2973663330078127,
      "step": 28200
    },
    {
      "epoch": 0.016631904024788306,
      "grad_norm": 0.39263415336608887,
      "learning_rate": 0.0002975053030744585,
      "loss": 3.324180908203125,
      "step": 28250
    },
    {
      "epoch": 0.01666134102306227,
      "grad_norm": 0.38221877813339233,
      "learning_rate": 0.0002975008875253673,
      "loss": 3.3487457275390624,
      "step": 28300
    },
    {
      "epoch": 0.01669077802133623,
      "grad_norm": 0.3821994364261627,
      "learning_rate": 0.0002974964719762761,
      "loss": 3.350211486816406,
      "step": 28350
    },
    {
      "epoch": 0.016720215019610192,
      "grad_norm": 0.2700080871582031,
      "learning_rate": 0.0002974920564271849,
      "loss": 3.3812020874023436,
      "step": 28400
    },
    {
      "epoch": 0.016749652017884153,
      "grad_norm": 0.3635845184326172,
      "learning_rate": 0.00029748764087809374,
      "loss": 3.3605377197265627,
      "step": 28450
    },
    {
      "epoch": 0.016779089016158117,
      "grad_norm": 0.3892115652561188,
      "learning_rate": 0.00029748322532900255,
      "loss": 3.370294494628906,
      "step": 28500
    },
    {
      "epoch": 0.016808526014432078,
      "grad_norm": 0.3583516478538513,
      "learning_rate": 0.0002974788097799113,
      "loss": 3.335756530761719,
      "step": 28550
    },
    {
      "epoch": 0.01683796301270604,
      "grad_norm": 0.4082677662372589,
      "learning_rate": 0.0002974743942308202,
      "loss": 3.384571533203125,
      "step": 28600
    },
    {
      "epoch": 0.01686740001098,
      "grad_norm": 0.3369288742542267,
      "learning_rate": 0.000297469978681729,
      "loss": 3.281658935546875,
      "step": 28650
    },
    {
      "epoch": 0.016896837009253964,
      "grad_norm": 0.38718873262405396,
      "learning_rate": 0.00029746556313263776,
      "loss": 3.3714190673828126,
      "step": 28700
    },
    {
      "epoch": 0.016926274007527924,
      "grad_norm": 0.30549246072769165,
      "learning_rate": 0.00029746114758354657,
      "loss": 3.3159780883789063,
      "step": 28750
    },
    {
      "epoch": 0.016955711005801885,
      "grad_norm": 0.45632609724998474,
      "learning_rate": 0.0002974567320344554,
      "loss": 3.3036492919921874,
      "step": 28800
    },
    {
      "epoch": 0.016985148004075846,
      "grad_norm": 0.5262589454650879,
      "learning_rate": 0.0002974523164853642,
      "loss": 3.3226739501953126,
      "step": 28850
    },
    {
      "epoch": 0.01701458500234981,
      "grad_norm": 0.40035825967788696,
      "learning_rate": 0.000297447900936273,
      "loss": 3.253521728515625,
      "step": 28900
    },
    {
      "epoch": 0.01704402200062377,
      "grad_norm": 0.3243553042411804,
      "learning_rate": 0.00029744348538718183,
      "loss": 3.32811767578125,
      "step": 28950
    },
    {
      "epoch": 0.01707345899889773,
      "grad_norm": 0.44102713465690613,
      "learning_rate": 0.00029743906983809064,
      "loss": 3.3178717041015626,
      "step": 29000
    },
    {
      "epoch": 0.017102895997171692,
      "grad_norm": 0.46506810188293457,
      "learning_rate": 0.00029743465428899946,
      "loss": 3.2894418334960935,
      "step": 29050
    },
    {
      "epoch": 0.017132332995445653,
      "grad_norm": 0.3836856782436371,
      "learning_rate": 0.0002974302387399082,
      "loss": 3.3274307250976562,
      "step": 29100
    },
    {
      "epoch": 0.017161769993719617,
      "grad_norm": 0.2887013554573059,
      "learning_rate": 0.00029742582319081703,
      "loss": 3.3618783569335937,
      "step": 29150
    },
    {
      "epoch": 0.017191206991993578,
      "grad_norm": 0.5205537676811218,
      "learning_rate": 0.00029742140764172585,
      "loss": 3.283646240234375,
      "step": 29200
    },
    {
      "epoch": 0.01722064399026754,
      "grad_norm": 0.38793429732322693,
      "learning_rate": 0.00029741699209263466,
      "loss": 3.361673583984375,
      "step": 29250
    },
    {
      "epoch": 0.0172500809885415,
      "grad_norm": 0.44497397541999817,
      "learning_rate": 0.0002974125765435435,
      "loss": 3.3310006713867186,
      "step": 29300
    },
    {
      "epoch": 0.017279517986815464,
      "grad_norm": 0.49681124091148376,
      "learning_rate": 0.0002974081609944523,
      "loss": 3.311955871582031,
      "step": 29350
    },
    {
      "epoch": 0.017308954985089425,
      "grad_norm": 0.38389280438423157,
      "learning_rate": 0.0002974037454453611,
      "loss": 3.3127511596679686,
      "step": 29400
    },
    {
      "epoch": 0.017338391983363385,
      "grad_norm": 0.5392832159996033,
      "learning_rate": 0.00029739932989626986,
      "loss": 3.3291131591796876,
      "step": 29450
    },
    {
      "epoch": 0.017367828981637346,
      "grad_norm": 0.34322652220726013,
      "learning_rate": 0.0002973949143471787,
      "loss": 3.308919677734375,
      "step": 29500
    },
    {
      "epoch": 0.01739726597991131,
      "grad_norm": 0.3400014340877533,
      "learning_rate": 0.0002973904987980875,
      "loss": 3.2641754150390625,
      "step": 29550
    },
    {
      "epoch": 0.01742670297818527,
      "grad_norm": 0.5295704007148743,
      "learning_rate": 0.0002973860832489963,
      "loss": 3.322881774902344,
      "step": 29600
    },
    {
      "epoch": 0.017456139976459232,
      "grad_norm": 0.5808720588684082,
      "learning_rate": 0.0002973816676999051,
      "loss": 3.2736151123046877,
      "step": 29650
    },
    {
      "epoch": 0.017485576974733193,
      "grad_norm": 0.3465769588947296,
      "learning_rate": 0.00029737725215081394,
      "loss": 3.37275634765625,
      "step": 29700
    },
    {
      "epoch": 0.017515013973007157,
      "grad_norm": 0.3314492702484131,
      "learning_rate": 0.00029737283660172275,
      "loss": 3.3245355224609376,
      "step": 29750
    },
    {
      "epoch": 0.017544450971281118,
      "grad_norm": 0.46414682269096375,
      "learning_rate": 0.00029736842105263157,
      "loss": 3.267093505859375,
      "step": 29800
    },
    {
      "epoch": 0.01757388796955508,
      "grad_norm": 0.3225749731063843,
      "learning_rate": 0.0002973640055035403,
      "loss": 3.316886291503906,
      "step": 29850
    },
    {
      "epoch": 0.01760332496782904,
      "grad_norm": 0.3984546363353729,
      "learning_rate": 0.0002973595899544492,
      "loss": 3.282295837402344,
      "step": 29900
    },
    {
      "epoch": 0.017632761966103003,
      "grad_norm": 0.4070683717727661,
      "learning_rate": 0.000297355174405358,
      "loss": 3.286484375,
      "step": 29950
    },
    {
      "epoch": 0.017662198964376964,
      "grad_norm": 0.2999691367149353,
      "learning_rate": 0.00029735075885626677,
      "loss": 3.3321603393554686,
      "step": 30000
    },
    {
      "epoch": 0.017691635962650925,
      "grad_norm": 0.5980725884437561,
      "learning_rate": 0.0002973463433071756,
      "loss": 3.2900433349609375,
      "step": 30050
    },
    {
      "epoch": 0.017721072960924886,
      "grad_norm": 0.42477095127105713,
      "learning_rate": 0.0002973419277580844,
      "loss": 3.3282089233398438,
      "step": 30100
    },
    {
      "epoch": 0.01775050995919885,
      "grad_norm": 0.36253875494003296,
      "learning_rate": 0.0002973375122089932,
      "loss": 3.281840515136719,
      "step": 30150
    },
    {
      "epoch": 0.01777994695747281,
      "grad_norm": 0.5856444835662842,
      "learning_rate": 0.00029733309665990203,
      "loss": 3.3522579956054686,
      "step": 30200
    },
    {
      "epoch": 0.01780938395574677,
      "grad_norm": 0.49632006883621216,
      "learning_rate": 0.00029732868111081084,
      "loss": 3.3109591674804686,
      "step": 30250
    },
    {
      "epoch": 0.017838820954020732,
      "grad_norm": 0.3469674587249756,
      "learning_rate": 0.00029732426556171966,
      "loss": 3.36434814453125,
      "step": 30300
    },
    {
      "epoch": 0.017868257952294696,
      "grad_norm": 0.45036742091178894,
      "learning_rate": 0.00029731985001262847,
      "loss": 3.3615988159179686,
      "step": 30350
    },
    {
      "epoch": 0.017897694950568657,
      "grad_norm": 0.3846302330493927,
      "learning_rate": 0.00029731543446353723,
      "loss": 3.3379400634765624,
      "step": 30400
    },
    {
      "epoch": 0.017927131948842618,
      "grad_norm": 0.4314449727535248,
      "learning_rate": 0.00029731101891444605,
      "loss": 3.2939883422851564,
      "step": 30450
    },
    {
      "epoch": 0.01795656894711658,
      "grad_norm": 0.299419641494751,
      "learning_rate": 0.00029730660336535486,
      "loss": 3.297882385253906,
      "step": 30500
    },
    {
      "epoch": 0.017986005945390543,
      "grad_norm": 0.3429957926273346,
      "learning_rate": 0.0002973021878162637,
      "loss": 3.3773703002929687,
      "step": 30550
    },
    {
      "epoch": 0.018015442943664504,
      "grad_norm": 0.43083450198173523,
      "learning_rate": 0.0002972977722671725,
      "loss": 3.281604919433594,
      "step": 30600
    },
    {
      "epoch": 0.018044879941938465,
      "grad_norm": 0.3471200168132782,
      "learning_rate": 0.0002972933567180813,
      "loss": 3.3604840087890624,
      "step": 30650
    },
    {
      "epoch": 0.018074316940212425,
      "grad_norm": 0.37335827946662903,
      "learning_rate": 0.0002972889411689901,
      "loss": 3.2873150634765627,
      "step": 30700
    },
    {
      "epoch": 0.01810375393848639,
      "grad_norm": 0.5689088702201843,
      "learning_rate": 0.00029728452561989893,
      "loss": 3.3099801635742185,
      "step": 30750
    },
    {
      "epoch": 0.01813319093676035,
      "grad_norm": 0.40712639689445496,
      "learning_rate": 0.0002972801100708077,
      "loss": 3.3137161254882814,
      "step": 30800
    },
    {
      "epoch": 0.01816262793503431,
      "grad_norm": 0.35073089599609375,
      "learning_rate": 0.0002972756945217165,
      "loss": 3.3088815307617185,
      "step": 30850
    },
    {
      "epoch": 0.018192064933308272,
      "grad_norm": 0.3126571476459503,
      "learning_rate": 0.0002972712789726253,
      "loss": 3.2765985107421876,
      "step": 30900
    },
    {
      "epoch": 0.018221501931582233,
      "grad_norm": 0.6170899271965027,
      "learning_rate": 0.00029726686342353414,
      "loss": 3.3012152099609375,
      "step": 30950
    },
    {
      "epoch": 0.018250938929856197,
      "grad_norm": 0.45087316632270813,
      "learning_rate": 0.00029726244787444295,
      "loss": 3.3077316284179688,
      "step": 31000
    },
    {
      "epoch": 0.018280375928130158,
      "grad_norm": 0.391489714384079,
      "learning_rate": 0.00029725803232535177,
      "loss": 3.297548828125,
      "step": 31050
    },
    {
      "epoch": 0.01830981292640412,
      "grad_norm": 0.4983297884464264,
      "learning_rate": 0.0002972536167762606,
      "loss": 3.2785791015625,
      "step": 31100
    },
    {
      "epoch": 0.01833924992467808,
      "grad_norm": 0.3342491388320923,
      "learning_rate": 0.00029724920122716934,
      "loss": 3.2611868286132815,
      "step": 31150
    },
    {
      "epoch": 0.018368686922952043,
      "grad_norm": 0.42416471242904663,
      "learning_rate": 0.0002972447856780782,
      "loss": 3.274851989746094,
      "step": 31200
    },
    {
      "epoch": 0.018398123921226004,
      "grad_norm": 0.4636434018611908,
      "learning_rate": 0.000297240370128987,
      "loss": 3.322803955078125,
      "step": 31250
    },
    {
      "epoch": 0.018427560919499965,
      "grad_norm": 0.47901177406311035,
      "learning_rate": 0.0002972359545798958,
      "loss": 3.2717050170898436,
      "step": 31300
    },
    {
      "epoch": 0.018456997917773926,
      "grad_norm": 0.6366715431213379,
      "learning_rate": 0.0002972315390308046,
      "loss": 3.2557296752929688,
      "step": 31350
    },
    {
      "epoch": 0.01848643491604789,
      "grad_norm": 0.3158472180366516,
      "learning_rate": 0.0002972271234817134,
      "loss": 3.2912554931640625,
      "step": 31400
    },
    {
      "epoch": 0.01851587191432185,
      "grad_norm": 0.368108332157135,
      "learning_rate": 0.0002972227079326222,
      "loss": 3.293320617675781,
      "step": 31450
    },
    {
      "epoch": 0.01854530891259581,
      "grad_norm": 0.6759923100471497,
      "learning_rate": 0.00029721829238353104,
      "loss": 3.3131890869140626,
      "step": 31500
    },
    {
      "epoch": 0.018574745910869772,
      "grad_norm": 0.43098312616348267,
      "learning_rate": 0.00029721387683443986,
      "loss": 3.298433837890625,
      "step": 31550
    },
    {
      "epoch": 0.018604182909143736,
      "grad_norm": 0.48529452085494995,
      "learning_rate": 0.00029720946128534867,
      "loss": 3.261856994628906,
      "step": 31600
    },
    {
      "epoch": 0.018633619907417697,
      "grad_norm": 0.3283274173736572,
      "learning_rate": 0.0002972050457362575,
      "loss": 3.2708428955078124,
      "step": 31650
    },
    {
      "epoch": 0.018663056905691658,
      "grad_norm": 0.3790273070335388,
      "learning_rate": 0.00029720063018716625,
      "loss": 3.2871994018554687,
      "step": 31700
    },
    {
      "epoch": 0.01869249390396562,
      "grad_norm": 0.4226006865501404,
      "learning_rate": 0.00029719621463807506,
      "loss": 3.267833557128906,
      "step": 31750
    },
    {
      "epoch": 0.018721930902239583,
      "grad_norm": 0.43577900528907776,
      "learning_rate": 0.0002971917990889839,
      "loss": 3.329666748046875,
      "step": 31800
    },
    {
      "epoch": 0.018751367900513544,
      "grad_norm": 0.25473862886428833,
      "learning_rate": 0.0002971873835398927,
      "loss": 3.2748297119140624,
      "step": 31850
    },
    {
      "epoch": 0.018780804898787504,
      "grad_norm": 0.523444414138794,
      "learning_rate": 0.0002971829679908015,
      "loss": 3.313940124511719,
      "step": 31900
    },
    {
      "epoch": 0.018810241897061465,
      "grad_norm": 0.37976422905921936,
      "learning_rate": 0.0002971785524417103,
      "loss": 3.312727355957031,
      "step": 31950
    },
    {
      "epoch": 0.01883967889533543,
      "grad_norm": 0.380183607339859,
      "learning_rate": 0.00029717413689261913,
      "loss": 3.2337860107421874,
      "step": 32000
    },
    {
      "epoch": 0.01886911589360939,
      "grad_norm": 0.47505006194114685,
      "learning_rate": 0.00029716972134352795,
      "loss": 3.2548904418945312,
      "step": 32050
    },
    {
      "epoch": 0.01889855289188335,
      "grad_norm": 0.3463679552078247,
      "learning_rate": 0.0002971653057944367,
      "loss": 3.2216262817382812,
      "step": 32100
    },
    {
      "epoch": 0.01892798989015731,
      "grad_norm": 0.39284828305244446,
      "learning_rate": 0.0002971608902453455,
      "loss": 3.2954400634765624,
      "step": 32150
    },
    {
      "epoch": 0.018957426888431276,
      "grad_norm": 0.4283110797405243,
      "learning_rate": 0.0002971564746962544,
      "loss": 3.316858215332031,
      "step": 32200
    },
    {
      "epoch": 0.018986863886705237,
      "grad_norm": 0.39009371399879456,
      "learning_rate": 0.00029715205914716315,
      "loss": 3.26821044921875,
      "step": 32250
    },
    {
      "epoch": 0.019016300884979197,
      "grad_norm": 0.46450120210647583,
      "learning_rate": 0.00029714764359807196,
      "loss": 3.2764462280273436,
      "step": 32300
    },
    {
      "epoch": 0.019045737883253158,
      "grad_norm": 0.3075593411922455,
      "learning_rate": 0.0002971432280489808,
      "loss": 3.2684234619140624,
      "step": 32350
    },
    {
      "epoch": 0.019075174881527122,
      "grad_norm": 0.30727073550224304,
      "learning_rate": 0.0002971388124998896,
      "loss": 3.3120562744140627,
      "step": 32400
    },
    {
      "epoch": 0.019104611879801083,
      "grad_norm": 0.392340749502182,
      "learning_rate": 0.0002971343969507984,
      "loss": 3.3214962768554686,
      "step": 32450
    },
    {
      "epoch": 0.019134048878075044,
      "grad_norm": 0.3629240393638611,
      "learning_rate": 0.0002971299814017072,
      "loss": 3.26174072265625,
      "step": 32500
    },
    {
      "epoch": 0.019163485876349005,
      "grad_norm": 0.3989822566509247,
      "learning_rate": 0.00029712556585261604,
      "loss": 3.21981201171875,
      "step": 32550
    },
    {
      "epoch": 0.01919292287462297,
      "grad_norm": 0.37422770261764526,
      "learning_rate": 0.0002971211503035248,
      "loss": 3.2795956420898436,
      "step": 32600
    },
    {
      "epoch": 0.01922235987289693,
      "grad_norm": 0.5168017745018005,
      "learning_rate": 0.0002971167347544336,
      "loss": 3.2654104614257813,
      "step": 32650
    },
    {
      "epoch": 0.01925179687117089,
      "grad_norm": 0.4679248034954071,
      "learning_rate": 0.0002971123192053424,
      "loss": 3.3218377685546874,
      "step": 32700
    },
    {
      "epoch": 0.01928123386944485,
      "grad_norm": 0.3650444746017456,
      "learning_rate": 0.00029710790365625124,
      "loss": 3.29232177734375,
      "step": 32750
    },
    {
      "epoch": 0.019310670867718812,
      "grad_norm": 0.356915682554245,
      "learning_rate": 0.00029710348810716006,
      "loss": 3.2414105224609373,
      "step": 32800
    },
    {
      "epoch": 0.019340107865992776,
      "grad_norm": 0.33356329798698425,
      "learning_rate": 0.00029709907255806887,
      "loss": 3.258599853515625,
      "step": 32850
    },
    {
      "epoch": 0.019369544864266737,
      "grad_norm": 0.4028301239013672,
      "learning_rate": 0.0002970946570089777,
      "loss": 3.2692025756835936,
      "step": 32900
    },
    {
      "epoch": 0.019398981862540698,
      "grad_norm": 0.39388805627822876,
      "learning_rate": 0.0002970902414598865,
      "loss": 3.237523193359375,
      "step": 32950
    },
    {
      "epoch": 0.01942841886081466,
      "grad_norm": 0.4775613248348236,
      "learning_rate": 0.00029708582591079526,
      "loss": 3.2717633056640625,
      "step": 33000
    },
    {
      "epoch": 0.019457855859088623,
      "grad_norm": 0.345913290977478,
      "learning_rate": 0.0002970814103617041,
      "loss": 3.268056640625,
      "step": 33050
    },
    {
      "epoch": 0.019487292857362584,
      "grad_norm": 0.349027544260025,
      "learning_rate": 0.0002970769948126129,
      "loss": 3.296630859375,
      "step": 33100
    },
    {
      "epoch": 0.019516729855636544,
      "grad_norm": 0.2826789319515228,
      "learning_rate": 0.0002970725792635217,
      "loss": 3.2405035400390627,
      "step": 33150
    },
    {
      "epoch": 0.019546166853910505,
      "grad_norm": 0.35219669342041016,
      "learning_rate": 0.0002970681637144305,
      "loss": 3.289042663574219,
      "step": 33200
    },
    {
      "epoch": 0.01957560385218447,
      "grad_norm": 0.37532779574394226,
      "learning_rate": 0.00029706374816533933,
      "loss": 3.2595831298828126,
      "step": 33250
    },
    {
      "epoch": 0.01960504085045843,
      "grad_norm": 0.3855534791946411,
      "learning_rate": 0.00029705933261624815,
      "loss": 3.253723449707031,
      "step": 33300
    },
    {
      "epoch": 0.01963447784873239,
      "grad_norm": 0.29480841755867004,
      "learning_rate": 0.00029705491706715696,
      "loss": 3.2731698608398436,
      "step": 33350
    },
    {
      "epoch": 0.01966391484700635,
      "grad_norm": 0.33075422048568726,
      "learning_rate": 0.0002970505015180657,
      "loss": 3.300667724609375,
      "step": 33400
    },
    {
      "epoch": 0.019693351845280316,
      "grad_norm": 0.3828047215938568,
      "learning_rate": 0.00029704608596897454,
      "loss": 3.25794189453125,
      "step": 33450
    },
    {
      "epoch": 0.019722788843554277,
      "grad_norm": 0.4717664122581482,
      "learning_rate": 0.0002970416704198834,
      "loss": 3.248912353515625,
      "step": 33500
    },
    {
      "epoch": 0.019752225841828237,
      "grad_norm": 0.35531502962112427,
      "learning_rate": 0.00029703725487079216,
      "loss": 3.209087219238281,
      "step": 33550
    },
    {
      "epoch": 0.019781662840102198,
      "grad_norm": 0.3733054995536804,
      "learning_rate": 0.000297032839321701,
      "loss": 3.2344329833984373,
      "step": 33600
    },
    {
      "epoch": 0.019811099838376162,
      "grad_norm": 0.49009817838668823,
      "learning_rate": 0.0002970284237726098,
      "loss": 3.2712063598632812,
      "step": 33650
    },
    {
      "epoch": 0.019840536836650123,
      "grad_norm": 0.3877944350242615,
      "learning_rate": 0.0002970240082235186,
      "loss": 3.2666790771484373,
      "step": 33700
    },
    {
      "epoch": 0.019869973834924084,
      "grad_norm": 0.39670881628990173,
      "learning_rate": 0.0002970195926744274,
      "loss": 3.2456088256835938,
      "step": 33750
    },
    {
      "epoch": 0.019899410833198045,
      "grad_norm": 0.38322582840919495,
      "learning_rate": 0.00029701517712533624,
      "loss": 3.3093191528320314,
      "step": 33800
    },
    {
      "epoch": 0.01992884783147201,
      "grad_norm": 0.470154732465744,
      "learning_rate": 0.00029701076157624505,
      "loss": 3.276842041015625,
      "step": 33850
    },
    {
      "epoch": 0.01995828482974597,
      "grad_norm": 0.36762791872024536,
      "learning_rate": 0.00029700634602715387,
      "loss": 3.2773919677734376,
      "step": 33900
    },
    {
      "epoch": 0.01998772182801993,
      "grad_norm": 0.4013325870037079,
      "learning_rate": 0.0002970019304780626,
      "loss": 3.2987176513671876,
      "step": 33950
    },
    {
      "epoch": 0.02001715882629389,
      "grad_norm": 0.44891357421875,
      "learning_rate": 0.00029699751492897144,
      "loss": 3.312467041015625,
      "step": 34000
    },
    {
      "epoch": 0.020046595824567855,
      "grad_norm": 0.29401934146881104,
      "learning_rate": 0.00029699309937988025,
      "loss": 3.291597900390625,
      "step": 34050
    },
    {
      "epoch": 0.020076032822841816,
      "grad_norm": 0.4446222484111786,
      "learning_rate": 0.00029698868383078907,
      "loss": 3.265635986328125,
      "step": 34100
    },
    {
      "epoch": 0.020105469821115777,
      "grad_norm": 0.3942376673221588,
      "learning_rate": 0.0002969842682816979,
      "loss": 3.2501806640625,
      "step": 34150
    },
    {
      "epoch": 0.020134906819389738,
      "grad_norm": 0.3174414336681366,
      "learning_rate": 0.0002969798527326067,
      "loss": 3.292089538574219,
      "step": 34200
    },
    {
      "epoch": 0.020164343817663702,
      "grad_norm": 0.36766913533210754,
      "learning_rate": 0.0002969754371835155,
      "loss": 3.27181884765625,
      "step": 34250
    },
    {
      "epoch": 0.020193780815937663,
      "grad_norm": 0.3638394773006439,
      "learning_rate": 0.0002969710216344243,
      "loss": 3.248637390136719,
      "step": 34300
    },
    {
      "epoch": 0.020223217814211623,
      "grad_norm": 0.29044631123542786,
      "learning_rate": 0.0002969666060853331,
      "loss": 3.2892794799804688,
      "step": 34350
    },
    {
      "epoch": 0.020252654812485584,
      "grad_norm": 0.4154435992240906,
      "learning_rate": 0.0002969621905362419,
      "loss": 3.278795166015625,
      "step": 34400
    },
    {
      "epoch": 0.02028209181075955,
      "grad_norm": 0.32950422167778015,
      "learning_rate": 0.0002969577749871507,
      "loss": 3.209134826660156,
      "step": 34450
    },
    {
      "epoch": 0.02031152880903351,
      "grad_norm": 0.4390483498573303,
      "learning_rate": 0.00029695335943805953,
      "loss": 3.270066833496094,
      "step": 34500
    },
    {
      "epoch": 0.02034096580730747,
      "grad_norm": 0.5260713696479797,
      "learning_rate": 0.00029694894388896835,
      "loss": 3.2392535400390625,
      "step": 34550
    },
    {
      "epoch": 0.02037040280558143,
      "grad_norm": 0.3461753726005554,
      "learning_rate": 0.00029694452833987716,
      "loss": 3.280260009765625,
      "step": 34600
    },
    {
      "epoch": 0.02039983980385539,
      "grad_norm": 0.4638700783252716,
      "learning_rate": 0.000296940112790786,
      "loss": 3.1883340454101563,
      "step": 34650
    },
    {
      "epoch": 0.020429276802129356,
      "grad_norm": 0.38366228342056274,
      "learning_rate": 0.00029693569724169473,
      "loss": 3.3088299560546877,
      "step": 34700
    },
    {
      "epoch": 0.020458713800403316,
      "grad_norm": 0.4099796712398529,
      "learning_rate": 0.00029693128169260355,
      "loss": 3.2568243408203124,
      "step": 34750
    },
    {
      "epoch": 0.020488150798677277,
      "grad_norm": 0.4446597397327423,
      "learning_rate": 0.0002969268661435124,
      "loss": 3.2848880004882814,
      "step": 34800
    },
    {
      "epoch": 0.020517587796951238,
      "grad_norm": 0.39452287554740906,
      "learning_rate": 0.0002969224505944212,
      "loss": 3.2659152221679686,
      "step": 34850
    },
    {
      "epoch": 0.020547024795225202,
      "grad_norm": 0.27860361337661743,
      "learning_rate": 0.00029691803504533,
      "loss": 3.2957858276367187,
      "step": 34900
    },
    {
      "epoch": 0.020576461793499163,
      "grad_norm": 0.6348734498023987,
      "learning_rate": 0.0002969136194962388,
      "loss": 3.275418701171875,
      "step": 34950
    },
    {
      "epoch": 0.020605898791773124,
      "grad_norm": 0.39326348900794983,
      "learning_rate": 0.0002969092039471476,
      "loss": 3.2997402954101562,
      "step": 35000
    },
    {
      "epoch": 0.020635335790047084,
      "grad_norm": 0.47493162751197815,
      "learning_rate": 0.00029690478839805644,
      "loss": 3.194827880859375,
      "step": 35050
    },
    {
      "epoch": 0.02066477278832105,
      "grad_norm": 0.38788190484046936,
      "learning_rate": 0.0002969003728489652,
      "loss": 3.2157391357421874,
      "step": 35100
    },
    {
      "epoch": 0.02069420978659501,
      "grad_norm": 0.5584072470664978,
      "learning_rate": 0.00029689595729987407,
      "loss": 3.25318359375,
      "step": 35150
    },
    {
      "epoch": 0.02072364678486897,
      "grad_norm": 0.4178030788898468,
      "learning_rate": 0.0002968915417507829,
      "loss": 3.191486511230469,
      "step": 35200
    },
    {
      "epoch": 0.02075308378314293,
      "grad_norm": 0.3560614287853241,
      "learning_rate": 0.00029688712620169164,
      "loss": 3.254569396972656,
      "step": 35250
    },
    {
      "epoch": 0.020782520781416895,
      "grad_norm": 0.35875198245048523,
      "learning_rate": 0.00029688271065260045,
      "loss": 3.2342538452148437,
      "step": 35300
    },
    {
      "epoch": 0.020811957779690856,
      "grad_norm": 0.3988212049007416,
      "learning_rate": 0.00029687829510350927,
      "loss": 3.2514971923828124,
      "step": 35350
    },
    {
      "epoch": 0.020841394777964817,
      "grad_norm": 0.5272156596183777,
      "learning_rate": 0.0002968738795544181,
      "loss": 3.2225341796875,
      "step": 35400
    },
    {
      "epoch": 0.020870831776238778,
      "grad_norm": 0.33339884877204895,
      "learning_rate": 0.0002968694640053269,
      "loss": 3.292405700683594,
      "step": 35450
    },
    {
      "epoch": 0.020900268774512742,
      "grad_norm": 0.32296136021614075,
      "learning_rate": 0.0002968650484562357,
      "loss": 3.2639926147460936,
      "step": 35500
    },
    {
      "epoch": 0.020929705772786703,
      "grad_norm": 0.34447261691093445,
      "learning_rate": 0.0002968606329071445,
      "loss": 3.2302169799804688,
      "step": 35550
    },
    {
      "epoch": 0.020959142771060663,
      "grad_norm": 0.5167677998542786,
      "learning_rate": 0.00029685621735805334,
      "loss": 3.227037353515625,
      "step": 35600
    },
    {
      "epoch": 0.020988579769334624,
      "grad_norm": 0.35853904485702515,
      "learning_rate": 0.0002968518018089621,
      "loss": 3.2712725830078124,
      "step": 35650
    },
    {
      "epoch": 0.02101801676760859,
      "grad_norm": 0.5661894679069519,
      "learning_rate": 0.0002968473862598709,
      "loss": 3.298123474121094,
      "step": 35700
    },
    {
      "epoch": 0.02104745376588255,
      "grad_norm": 0.3757530152797699,
      "learning_rate": 0.00029684297071077973,
      "loss": 3.243282470703125,
      "step": 35750
    },
    {
      "epoch": 0.02107689076415651,
      "grad_norm": 0.3488098680973053,
      "learning_rate": 0.00029683855516168854,
      "loss": 3.243183288574219,
      "step": 35800
    },
    {
      "epoch": 0.02110632776243047,
      "grad_norm": 0.4681428074836731,
      "learning_rate": 0.00029683413961259736,
      "loss": 3.2547308349609376,
      "step": 35850
    },
    {
      "epoch": 0.021135764760704435,
      "grad_norm": 0.3497195243835449,
      "learning_rate": 0.0002968297240635062,
      "loss": 3.260167236328125,
      "step": 35900
    },
    {
      "epoch": 0.021165201758978396,
      "grad_norm": 0.30120912194252014,
      "learning_rate": 0.000296825308514415,
      "loss": 3.270296630859375,
      "step": 35950
    },
    {
      "epoch": 0.021194638757252356,
      "grad_norm": 0.39518237113952637,
      "learning_rate": 0.00029682089296532375,
      "loss": 3.264321594238281,
      "step": 36000
    },
    {
      "epoch": 0.021224075755526317,
      "grad_norm": 0.3133420944213867,
      "learning_rate": 0.00029681647741623256,
      "loss": 3.3047528076171875,
      "step": 36050
    },
    {
      "epoch": 0.02125351275380028,
      "grad_norm": 0.42700058221817017,
      "learning_rate": 0.00029681206186714143,
      "loss": 3.2570266723632812,
      "step": 36100
    },
    {
      "epoch": 0.021282949752074242,
      "grad_norm": 0.40154770016670227,
      "learning_rate": 0.0002968076463180502,
      "loss": 3.264352111816406,
      "step": 36150
    },
    {
      "epoch": 0.021312386750348203,
      "grad_norm": 0.4905824661254883,
      "learning_rate": 0.000296803230768959,
      "loss": 3.2731582641601564,
      "step": 36200
    },
    {
      "epoch": 0.021341823748622164,
      "grad_norm": 0.3846723437309265,
      "learning_rate": 0.0002967988152198678,
      "loss": 3.2935968017578126,
      "step": 36250
    },
    {
      "epoch": 0.021371260746896124,
      "grad_norm": 0.4314635396003723,
      "learning_rate": 0.00029679439967077664,
      "loss": 3.213384704589844,
      "step": 36300
    },
    {
      "epoch": 0.02140069774517009,
      "grad_norm": 0.28772783279418945,
      "learning_rate": 0.00029678998412168545,
      "loss": 3.2769509887695314,
      "step": 36350
    },
    {
      "epoch": 0.02143013474344405,
      "grad_norm": 0.38190874457359314,
      "learning_rate": 0.0002967855685725942,
      "loss": 3.2555355834960937,
      "step": 36400
    },
    {
      "epoch": 0.02145957174171801,
      "grad_norm": 0.5213097333908081,
      "learning_rate": 0.0002967811530235031,
      "loss": 3.2585787963867188,
      "step": 36450
    },
    {
      "epoch": 0.02148900873999197,
      "grad_norm": 0.3730711340904236,
      "learning_rate": 0.0002967767374744119,
      "loss": 3.2463885498046876,
      "step": 36500
    },
    {
      "epoch": 0.021518445738265935,
      "grad_norm": 0.38471561670303345,
      "learning_rate": 0.00029677232192532065,
      "loss": 3.2244219970703125,
      "step": 36550
    },
    {
      "epoch": 0.021547882736539896,
      "grad_norm": 0.30918052792549133,
      "learning_rate": 0.00029676790637622947,
      "loss": 3.1877487182617186,
      "step": 36600
    },
    {
      "epoch": 0.021577319734813857,
      "grad_norm": 0.28681689500808716,
      "learning_rate": 0.0002967634908271383,
      "loss": 3.2241549682617188,
      "step": 36650
    },
    {
      "epoch": 0.021606756733087817,
      "grad_norm": 0.32386428117752075,
      "learning_rate": 0.0002967590752780471,
      "loss": 3.1756130981445314,
      "step": 36700
    },
    {
      "epoch": 0.02163619373136178,
      "grad_norm": 0.34625327587127686,
      "learning_rate": 0.0002967546597289559,
      "loss": 3.2177505493164062,
      "step": 36750
    },
    {
      "epoch": 0.021665630729635742,
      "grad_norm": 0.3503381013870239,
      "learning_rate": 0.0002967502441798647,
      "loss": 3.199660339355469,
      "step": 36800
    },
    {
      "epoch": 0.021695067727909703,
      "grad_norm": 0.4342331886291504,
      "learning_rate": 0.00029674582863077354,
      "loss": 3.2295327758789063,
      "step": 36850
    },
    {
      "epoch": 0.021724504726183664,
      "grad_norm": 0.37443122267723083,
      "learning_rate": 0.00029674141308168236,
      "loss": 3.2920269775390625,
      "step": 36900
    },
    {
      "epoch": 0.021753941724457628,
      "grad_norm": 0.28918153047561646,
      "learning_rate": 0.0002967369975325911,
      "loss": 3.229051208496094,
      "step": 36950
    },
    {
      "epoch": 0.02178337872273159,
      "grad_norm": 0.3045409619808197,
      "learning_rate": 0.00029673258198349993,
      "loss": 3.2039840698242186,
      "step": 37000
    },
    {
      "epoch": 0.02181281572100555,
      "grad_norm": 0.4873865842819214,
      "learning_rate": 0.00029672816643440874,
      "loss": 3.2407376098632814,
      "step": 37050
    },
    {
      "epoch": 0.02184225271927951,
      "grad_norm": 0.37862130999565125,
      "learning_rate": 0.00029672375088531756,
      "loss": 3.247666931152344,
      "step": 37100
    },
    {
      "epoch": 0.021871689717553475,
      "grad_norm": 0.4435122311115265,
      "learning_rate": 0.0002967193353362264,
      "loss": 3.2112185668945314,
      "step": 37150
    },
    {
      "epoch": 0.021901126715827435,
      "grad_norm": 0.38887667655944824,
      "learning_rate": 0.0002967149197871352,
      "loss": 3.2241061401367186,
      "step": 37200
    },
    {
      "epoch": 0.021930563714101396,
      "grad_norm": 0.350846529006958,
      "learning_rate": 0.000296710504238044,
      "loss": 3.273980712890625,
      "step": 37250
    },
    {
      "epoch": 0.021960000712375357,
      "grad_norm": 0.2736048102378845,
      "learning_rate": 0.0002967060886889528,
      "loss": 3.2106414794921876,
      "step": 37300
    },
    {
      "epoch": 0.02198943771064932,
      "grad_norm": 0.3231438398361206,
      "learning_rate": 0.0002967016731398616,
      "loss": 3.2106222534179687,
      "step": 37350
    },
    {
      "epoch": 0.022018874708923282,
      "grad_norm": 0.3412911593914032,
      "learning_rate": 0.00029669725759077045,
      "loss": 3.2196392822265625,
      "step": 37400
    },
    {
      "epoch": 0.022048311707197243,
      "grad_norm": 0.25073426961898804,
      "learning_rate": 0.00029669284204167926,
      "loss": 3.2493182373046876,
      "step": 37450
    },
    {
      "epoch": 0.022077748705471203,
      "grad_norm": 0.3087325692176819,
      "learning_rate": 0.000296688426492588,
      "loss": 3.2223541259765627,
      "step": 37500
    },
    {
      "epoch": 0.022107185703745168,
      "grad_norm": 0.3697931468486786,
      "learning_rate": 0.00029668401094349684,
      "loss": 3.19544921875,
      "step": 37550
    },
    {
      "epoch": 0.02213662270201913,
      "grad_norm": 0.340283066034317,
      "learning_rate": 0.00029667959539440565,
      "loss": 3.15642333984375,
      "step": 37600
    },
    {
      "epoch": 0.02216605970029309,
      "grad_norm": 0.4194275736808777,
      "learning_rate": 0.00029667517984531446,
      "loss": 3.23004150390625,
      "step": 37650
    },
    {
      "epoch": 0.02219549669856705,
      "grad_norm": 0.4830194413661957,
      "learning_rate": 0.0002966707642962232,
      "loss": 3.2090496826171875,
      "step": 37700
    },
    {
      "epoch": 0.022224933696841014,
      "grad_norm": 0.3489872217178345,
      "learning_rate": 0.0002966663487471321,
      "loss": 3.19129150390625,
      "step": 37750
    },
    {
      "epoch": 0.022254370695114975,
      "grad_norm": 0.3724016547203064,
      "learning_rate": 0.0002966619331980409,
      "loss": 3.2332894897460935,
      "step": 37800
    },
    {
      "epoch": 0.022283807693388936,
      "grad_norm": 0.6296040415763855,
      "learning_rate": 0.00029665751764894967,
      "loss": 3.254714050292969,
      "step": 37850
    },
    {
      "epoch": 0.022313244691662897,
      "grad_norm": 0.3687453866004944,
      "learning_rate": 0.0002966531020998585,
      "loss": 3.1854006958007814,
      "step": 37900
    },
    {
      "epoch": 0.02234268168993686,
      "grad_norm": 0.5763857960700989,
      "learning_rate": 0.0002966486865507673,
      "loss": 3.1993389892578126,
      "step": 37950
    },
    {
      "epoch": 0.02237211868821082,
      "grad_norm": 0.3069726526737213,
      "learning_rate": 0.0002966442710016761,
      "loss": 3.22733642578125,
      "step": 38000
    },
    {
      "epoch": 0.022401555686484782,
      "grad_norm": 0.41398942470550537,
      "learning_rate": 0.0002966398554525849,
      "loss": 3.2762020874023436,
      "step": 38050
    },
    {
      "epoch": 0.022430992684758743,
      "grad_norm": 0.43235674500465393,
      "learning_rate": 0.00029663543990349374,
      "loss": 3.165482482910156,
      "step": 38100
    },
    {
      "epoch": 0.022460429683032704,
      "grad_norm": 0.47886621952056885,
      "learning_rate": 0.00029663102435440255,
      "loss": 3.2218707275390623,
      "step": 38150
    },
    {
      "epoch": 0.022489866681306668,
      "grad_norm": 0.3219209909439087,
      "learning_rate": 0.00029662660880531137,
      "loss": 3.23199462890625,
      "step": 38200
    },
    {
      "epoch": 0.02251930367958063,
      "grad_norm": 0.3860849142074585,
      "learning_rate": 0.00029662219325622013,
      "loss": 3.2467312622070312,
      "step": 38250
    },
    {
      "epoch": 0.02254874067785459,
      "grad_norm": 0.46902450919151306,
      "learning_rate": 0.00029661777770712894,
      "loss": 3.13831787109375,
      "step": 38300
    },
    {
      "epoch": 0.02257817767612855,
      "grad_norm": 0.3348481059074402,
      "learning_rate": 0.00029661336215803776,
      "loss": 3.2303497314453127,
      "step": 38350
    },
    {
      "epoch": 0.022607614674402515,
      "grad_norm": 0.43958839774131775,
      "learning_rate": 0.0002966089466089466,
      "loss": 3.214053039550781,
      "step": 38400
    },
    {
      "epoch": 0.022637051672676475,
      "grad_norm": 0.44772258400917053,
      "learning_rate": 0.0002966045310598554,
      "loss": 3.181685485839844,
      "step": 38450
    },
    {
      "epoch": 0.022666488670950436,
      "grad_norm": 0.3495638370513916,
      "learning_rate": 0.0002966001155107642,
      "loss": 3.241583251953125,
      "step": 38500
    },
    {
      "epoch": 0.022695925669224397,
      "grad_norm": 0.3698650300502777,
      "learning_rate": 0.000296595699961673,
      "loss": 3.242015380859375,
      "step": 38550
    },
    {
      "epoch": 0.02272536266749836,
      "grad_norm": 0.38187599182128906,
      "learning_rate": 0.00029659128441258183,
      "loss": 3.191323547363281,
      "step": 38600
    },
    {
      "epoch": 0.022754799665772322,
      "grad_norm": 0.3345622718334198,
      "learning_rate": 0.0002965868688634906,
      "loss": 3.2035260009765625,
      "step": 38650
    },
    {
      "epoch": 0.022784236664046283,
      "grad_norm": 0.4270396828651428,
      "learning_rate": 0.00029658245331439946,
      "loss": 3.2305447387695314,
      "step": 38700
    },
    {
      "epoch": 0.022813673662320243,
      "grad_norm": 0.37730589509010315,
      "learning_rate": 0.0002965780377653083,
      "loss": 3.273265380859375,
      "step": 38750
    },
    {
      "epoch": 0.022843110660594208,
      "grad_norm": 0.42993709444999695,
      "learning_rate": 0.00029657362221621703,
      "loss": 3.221704406738281,
      "step": 38800
    },
    {
      "epoch": 0.02287254765886817,
      "grad_norm": 0.3500145971775055,
      "learning_rate": 0.00029656920666712585,
      "loss": 3.241849670410156,
      "step": 38850
    },
    {
      "epoch": 0.02290198465714213,
      "grad_norm": 0.45233696699142456,
      "learning_rate": 0.00029656479111803466,
      "loss": 3.2352999877929687,
      "step": 38900
    },
    {
      "epoch": 0.02293142165541609,
      "grad_norm": 0.5874031782150269,
      "learning_rate": 0.0002965603755689435,
      "loss": 3.2484613037109376,
      "step": 38950
    },
    {
      "epoch": 0.022960858653690054,
      "grad_norm": 0.42383909225463867,
      "learning_rate": 0.0002965559600198523,
      "loss": 3.1296536254882814,
      "step": 39000
    },
    {
      "epoch": 0.022990295651964015,
      "grad_norm": 0.34032824635505676,
      "learning_rate": 0.0002965515444707611,
      "loss": 3.1846774291992186,
      "step": 39050
    },
    {
      "epoch": 0.023019732650237976,
      "grad_norm": 0.3453627824783325,
      "learning_rate": 0.0002965471289216699,
      "loss": 3.2342343139648437,
      "step": 39100
    },
    {
      "epoch": 0.023049169648511936,
      "grad_norm": 0.34359368681907654,
      "learning_rate": 0.00029654271337257874,
      "loss": 3.1763653564453125,
      "step": 39150
    },
    {
      "epoch": 0.0230786066467859,
      "grad_norm": 0.3368850350379944,
      "learning_rate": 0.0002965382978234875,
      "loss": 3.253162841796875,
      "step": 39200
    },
    {
      "epoch": 0.02310804364505986,
      "grad_norm": 0.3329690992832184,
      "learning_rate": 0.0002965338822743963,
      "loss": 3.2332745361328126,
      "step": 39250
    },
    {
      "epoch": 0.023137480643333822,
      "grad_norm": 0.3107996881008148,
      "learning_rate": 0.0002965294667253051,
      "loss": 3.185979309082031,
      "step": 39300
    },
    {
      "epoch": 0.023166917641607783,
      "grad_norm": 0.42847561836242676,
      "learning_rate": 0.00029652505117621394,
      "loss": 3.1941619873046876,
      "step": 39350
    },
    {
      "epoch": 0.023196354639881747,
      "grad_norm": 0.36494210362434387,
      "learning_rate": 0.00029652063562712275,
      "loss": 3.1869110107421874,
      "step": 39400
    },
    {
      "epoch": 0.023225791638155708,
      "grad_norm": 0.3492168188095093,
      "learning_rate": 0.00029651622007803157,
      "loss": 3.2310455322265623,
      "step": 39450
    },
    {
      "epoch": 0.02325522863642967,
      "grad_norm": 0.44260844588279724,
      "learning_rate": 0.0002965118045289404,
      "loss": 3.1989288330078125,
      "step": 39500
    },
    {
      "epoch": 0.02328466563470363,
      "grad_norm": 0.4228355884552002,
      "learning_rate": 0.00029650738897984914,
      "loss": 3.2357833862304686,
      "step": 39550
    },
    {
      "epoch": 0.023314102632977594,
      "grad_norm": 0.4615185856819153,
      "learning_rate": 0.00029650297343075796,
      "loss": 3.143358154296875,
      "step": 39600
    },
    {
      "epoch": 0.023343539631251554,
      "grad_norm": 0.3806557357311249,
      "learning_rate": 0.00029649855788166677,
      "loss": 3.2239413452148438,
      "step": 39650
    },
    {
      "epoch": 0.023372976629525515,
      "grad_norm": 0.37798282504081726,
      "learning_rate": 0.0002964941423325756,
      "loss": 3.214391174316406,
      "step": 39700
    },
    {
      "epoch": 0.023402413627799476,
      "grad_norm": 0.4408104419708252,
      "learning_rate": 0.0002964897267834844,
      "loss": 3.1564434814453124,
      "step": 39750
    },
    {
      "epoch": 0.02343185062607344,
      "grad_norm": 0.3642081916332245,
      "learning_rate": 0.0002964853112343932,
      "loss": 3.1948565673828124,
      "step": 39800
    },
    {
      "epoch": 0.0234612876243474,
      "grad_norm": 0.35929012298583984,
      "learning_rate": 0.00029648089568530203,
      "loss": 3.198641052246094,
      "step": 39850
    },
    {
      "epoch": 0.023490724622621362,
      "grad_norm": 0.42813315987586975,
      "learning_rate": 0.00029647648013621084,
      "loss": 3.2415835571289064,
      "step": 39900
    },
    {
      "epoch": 0.023520161620895323,
      "grad_norm": 0.37044960260391235,
      "learning_rate": 0.0002964720645871196,
      "loss": 3.15937255859375,
      "step": 39950
    },
    {
      "epoch": 0.023549598619169283,
      "grad_norm": 0.32759889960289,
      "learning_rate": 0.0002964676490380285,
      "loss": 3.2516879272460937,
      "step": 40000
    },
    {
      "epoch": 0.023579035617443247,
      "grad_norm": 0.406696081161499,
      "learning_rate": 0.0002964632334889373,
      "loss": 3.196860656738281,
      "step": 40050
    },
    {
      "epoch": 0.023608472615717208,
      "grad_norm": 0.3679472506046295,
      "learning_rate": 0.00029645881793984605,
      "loss": 3.0965701293945314,
      "step": 40100
    },
    {
      "epoch": 0.02363790961399117,
      "grad_norm": 0.382213830947876,
      "learning_rate": 0.00029645440239075486,
      "loss": 3.147886047363281,
      "step": 40150
    },
    {
      "epoch": 0.02366734661226513,
      "grad_norm": 0.34691283106803894,
      "learning_rate": 0.0002964499868416637,
      "loss": 3.1714291381835937,
      "step": 40200
    },
    {
      "epoch": 0.023696783610539094,
      "grad_norm": 0.4403378665447235,
      "learning_rate": 0.0002964455712925725,
      "loss": 3.2615594482421875,
      "step": 40250
    },
    {
      "epoch": 0.023726220608813055,
      "grad_norm": 0.3168937563896179,
      "learning_rate": 0.0002964411557434813,
      "loss": 3.203696594238281,
      "step": 40300
    },
    {
      "epoch": 0.023755657607087016,
      "grad_norm": 0.3772541582584381,
      "learning_rate": 0.0002964367401943901,
      "loss": 3.213150939941406,
      "step": 40350
    },
    {
      "epoch": 0.023785094605360976,
      "grad_norm": 0.3792049288749695,
      "learning_rate": 0.00029643232464529894,
      "loss": 3.1750115966796875,
      "step": 40400
    },
    {
      "epoch": 0.02381453160363494,
      "grad_norm": 0.4307151734828949,
      "learning_rate": 0.00029642790909620775,
      "loss": 3.187103271484375,
      "step": 40450
    },
    {
      "epoch": 0.0238439686019089,
      "grad_norm": 0.4279058575630188,
      "learning_rate": 0.0002964234935471165,
      "loss": 3.1354241943359376,
      "step": 40500
    },
    {
      "epoch": 0.023873405600182862,
      "grad_norm": 0.3371472954750061,
      "learning_rate": 0.0002964190779980253,
      "loss": 3.1760455322265626,
      "step": 40550
    },
    {
      "epoch": 0.023902842598456823,
      "grad_norm": 0.26466459035873413,
      "learning_rate": 0.00029641466244893414,
      "loss": 3.1997940063476564,
      "step": 40600
    },
    {
      "epoch": 0.023932279596730787,
      "grad_norm": 0.55608069896698,
      "learning_rate": 0.00029641024689984295,
      "loss": 3.2449749755859374,
      "step": 40650
    },
    {
      "epoch": 0.023961716595004748,
      "grad_norm": 0.3074549436569214,
      "learning_rate": 0.00029640583135075177,
      "loss": 3.2330783081054686,
      "step": 40700
    },
    {
      "epoch": 0.02399115359327871,
      "grad_norm": 0.4021517038345337,
      "learning_rate": 0.0002964014158016606,
      "loss": 3.186973876953125,
      "step": 40750
    },
    {
      "epoch": 0.02402059059155267,
      "grad_norm": 0.4001486301422119,
      "learning_rate": 0.0002963970002525694,
      "loss": 3.174773864746094,
      "step": 40800
    },
    {
      "epoch": 0.024050027589826634,
      "grad_norm": 0.3407411575317383,
      "learning_rate": 0.0002963925847034782,
      "loss": 3.176812744140625,
      "step": 40850
    },
    {
      "epoch": 0.024079464588100594,
      "grad_norm": 0.30683135986328125,
      "learning_rate": 0.00029638816915438697,
      "loss": 3.1756277465820313,
      "step": 40900
    },
    {
      "epoch": 0.024108901586374555,
      "grad_norm": 0.4113587737083435,
      "learning_rate": 0.0002963837536052958,
      "loss": 3.167937927246094,
      "step": 40950
    },
    {
      "epoch": 0.024138338584648516,
      "grad_norm": 0.33438652753829956,
      "learning_rate": 0.00029637933805620466,
      "loss": 3.2123373413085936,
      "step": 41000
    },
    {
      "epoch": 0.02416777558292248,
      "grad_norm": 0.3400481939315796,
      "learning_rate": 0.0002963749225071134,
      "loss": 3.22949462890625,
      "step": 41050
    },
    {
      "epoch": 0.02419721258119644,
      "grad_norm": 0.36114609241485596,
      "learning_rate": 0.00029637050695802223,
      "loss": 3.2336968994140625,
      "step": 41100
    },
    {
      "epoch": 0.0242266495794704,
      "grad_norm": 0.3652268946170807,
      "learning_rate": 0.00029636609140893104,
      "loss": 3.1876785278320314,
      "step": 41150
    },
    {
      "epoch": 0.024256086577744362,
      "grad_norm": 0.3127206861972809,
      "learning_rate": 0.00029636167585983986,
      "loss": 3.1875930786132813,
      "step": 41200
    },
    {
      "epoch": 0.024285523576018327,
      "grad_norm": 0.3461538553237915,
      "learning_rate": 0.0002963572603107486,
      "loss": 3.1771975708007814,
      "step": 41250
    },
    {
      "epoch": 0.024314960574292287,
      "grad_norm": 0.36535224318504333,
      "learning_rate": 0.00029635284476165743,
      "loss": 3.207527770996094,
      "step": 41300
    },
    {
      "epoch": 0.024344397572566248,
      "grad_norm": 0.33108192682266235,
      "learning_rate": 0.0002963484292125663,
      "loss": 3.1871432495117187,
      "step": 41350
    },
    {
      "epoch": 0.02437383457084021,
      "grad_norm": 0.2600831687450409,
      "learning_rate": 0.00029634401366347506,
      "loss": 3.19192138671875,
      "step": 41400
    },
    {
      "epoch": 0.024403271569114173,
      "grad_norm": 0.35874274373054504,
      "learning_rate": 0.0002963395981143839,
      "loss": 3.2277426147460937,
      "step": 41450
    },
    {
      "epoch": 0.024432708567388134,
      "grad_norm": 0.2718585431575775,
      "learning_rate": 0.0002963351825652927,
      "loss": 3.137620849609375,
      "step": 41500
    },
    {
      "epoch": 0.024462145565662095,
      "grad_norm": 0.2987973988056183,
      "learning_rate": 0.0002963307670162015,
      "loss": 3.20704833984375,
      "step": 41550
    },
    {
      "epoch": 0.024491582563936055,
      "grad_norm": 0.4565908908843994,
      "learning_rate": 0.0002963263514671103,
      "loss": 3.133039855957031,
      "step": 41600
    },
    {
      "epoch": 0.02452101956221002,
      "grad_norm": 0.2958295941352844,
      "learning_rate": 0.00029632193591801913,
      "loss": 3.15419921875,
      "step": 41650
    },
    {
      "epoch": 0.02455045656048398,
      "grad_norm": 0.3589365482330322,
      "learning_rate": 0.00029631752036892795,
      "loss": 3.1318484497070314,
      "step": 41700
    },
    {
      "epoch": 0.02457989355875794,
      "grad_norm": 0.3567074239253998,
      "learning_rate": 0.00029631310481983676,
      "loss": 3.2281689453125,
      "step": 41750
    },
    {
      "epoch": 0.024609330557031902,
      "grad_norm": 0.3548557460308075,
      "learning_rate": 0.0002963086892707455,
      "loss": 3.1178799438476563,
      "step": 41800
    },
    {
      "epoch": 0.024638767555305863,
      "grad_norm": 0.34161025285720825,
      "learning_rate": 0.00029630427372165434,
      "loss": 3.2214999389648438,
      "step": 41850
    },
    {
      "epoch": 0.024668204553579827,
      "grad_norm": 0.38996338844299316,
      "learning_rate": 0.00029629985817256315,
      "loss": 3.1360794067382813,
      "step": 41900
    },
    {
      "epoch": 0.024697641551853788,
      "grad_norm": 0.28301674127578735,
      "learning_rate": 0.00029629544262347197,
      "loss": 3.196499938964844,
      "step": 41950
    },
    {
      "epoch": 0.02472707855012775,
      "grad_norm": 0.30809110403060913,
      "learning_rate": 0.0002962910270743808,
      "loss": 3.1905001831054687,
      "step": 42000
    },
    {
      "epoch": 0.02475651554840171,
      "grad_norm": 0.41510361433029175,
      "learning_rate": 0.0002962866115252896,
      "loss": 3.162966613769531,
      "step": 42050
    },
    {
      "epoch": 0.024785952546675673,
      "grad_norm": 0.33097076416015625,
      "learning_rate": 0.0002962821959761984,
      "loss": 3.192654724121094,
      "step": 42100
    },
    {
      "epoch": 0.024815389544949634,
      "grad_norm": 0.36313241720199585,
      "learning_rate": 0.0002962777804271072,
      "loss": 3.1612957763671874,
      "step": 42150
    },
    {
      "epoch": 0.024844826543223595,
      "grad_norm": 0.614332377910614,
      "learning_rate": 0.000296273364878016,
      "loss": 3.158175048828125,
      "step": 42200
    },
    {
      "epoch": 0.024874263541497556,
      "grad_norm": 0.3696703314781189,
      "learning_rate": 0.0002962689493289248,
      "loss": 3.1503616333007813,
      "step": 42250
    },
    {
      "epoch": 0.02490370053977152,
      "grad_norm": 0.3125178813934326,
      "learning_rate": 0.00029626453377983367,
      "loss": 3.197416076660156,
      "step": 42300
    },
    {
      "epoch": 0.02493313753804548,
      "grad_norm": 0.5288768410682678,
      "learning_rate": 0.00029626011823074243,
      "loss": 3.140680236816406,
      "step": 42350
    },
    {
      "epoch": 0.02496257453631944,
      "grad_norm": 0.29424306750297546,
      "learning_rate": 0.00029625570268165124,
      "loss": 3.180574951171875,
      "step": 42400
    },
    {
      "epoch": 0.024992011534593402,
      "grad_norm": 0.2737865746021271,
      "learning_rate": 0.00029625128713256006,
      "loss": 3.1252154541015624,
      "step": 42450
    },
    {
      "epoch": 0.025021448532867367,
      "grad_norm": 0.4092576801776886,
      "learning_rate": 0.00029624687158346887,
      "loss": 3.1750225830078125,
      "step": 42500
    },
    {
      "epoch": 0.025050885531141327,
      "grad_norm": 0.3022167682647705,
      "learning_rate": 0.0002962424560343777,
      "loss": 3.1669683837890625,
      "step": 42550
    },
    {
      "epoch": 0.025080322529415288,
      "grad_norm": 0.4774382710456848,
      "learning_rate": 0.00029623804048528645,
      "loss": 3.2211724853515626,
      "step": 42600
    },
    {
      "epoch": 0.02510975952768925,
      "grad_norm": 0.44779539108276367,
      "learning_rate": 0.0002962336249361953,
      "loss": 3.243386535644531,
      "step": 42650
    },
    {
      "epoch": 0.025139196525963213,
      "grad_norm": 0.523693859577179,
      "learning_rate": 0.00029622920938710413,
      "loss": 3.213955078125,
      "step": 42700
    },
    {
      "epoch": 0.025168633524237174,
      "grad_norm": 0.3470745086669922,
      "learning_rate": 0.0002962247938380129,
      "loss": 3.1458065795898436,
      "step": 42750
    },
    {
      "epoch": 0.025198070522511135,
      "grad_norm": 0.3841785192489624,
      "learning_rate": 0.0002962203782889217,
      "loss": 3.209721984863281,
      "step": 42800
    },
    {
      "epoch": 0.025227507520785095,
      "grad_norm": 0.2547699213027954,
      "learning_rate": 0.0002962159627398305,
      "loss": 3.1615567016601562,
      "step": 42850
    },
    {
      "epoch": 0.02525694451905906,
      "grad_norm": 0.4901289641857147,
      "learning_rate": 0.00029621154719073933,
      "loss": 3.105577087402344,
      "step": 42900
    },
    {
      "epoch": 0.02528638151733302,
      "grad_norm": 0.3299269676208496,
      "learning_rate": 0.00029620713164164815,
      "loss": 3.2119760131835937,
      "step": 42950
    },
    {
      "epoch": 0.02531581851560698,
      "grad_norm": 0.37671026587486267,
      "learning_rate": 0.00029620271609255696,
      "loss": 3.227357177734375,
      "step": 43000
    },
    {
      "epoch": 0.025345255513880942,
      "grad_norm": 0.2885051369667053,
      "learning_rate": 0.0002961983005434658,
      "loss": 3.1754940795898436,
      "step": 43050
    },
    {
      "epoch": 0.025374692512154906,
      "grad_norm": 0.41340088844299316,
      "learning_rate": 0.00029619388499437454,
      "loss": 3.1230868530273437,
      "step": 43100
    },
    {
      "epoch": 0.025404129510428867,
      "grad_norm": 0.4929138720035553,
      "learning_rate": 0.00029618946944528335,
      "loss": 3.1410311889648437,
      "step": 43150
    },
    {
      "epoch": 0.025433566508702828,
      "grad_norm": 0.3325177729129791,
      "learning_rate": 0.00029618505389619217,
      "loss": 3.147987976074219,
      "step": 43200
    },
    {
      "epoch": 0.02546300350697679,
      "grad_norm": 0.31573086977005005,
      "learning_rate": 0.000296180638347101,
      "loss": 3.1988986206054686,
      "step": 43250
    },
    {
      "epoch": 0.025492440505250753,
      "grad_norm": 0.2920289635658264,
      "learning_rate": 0.0002961762227980098,
      "loss": 3.1426736450195314,
      "step": 43300
    },
    {
      "epoch": 0.025521877503524713,
      "grad_norm": 0.5890534520149231,
      "learning_rate": 0.0002961718072489186,
      "loss": 3.179766845703125,
      "step": 43350
    },
    {
      "epoch": 0.025551314501798674,
      "grad_norm": 0.3430126905441284,
      "learning_rate": 0.0002961673916998274,
      "loss": 3.211826171875,
      "step": 43400
    },
    {
      "epoch": 0.025580751500072635,
      "grad_norm": 0.3847080171108246,
      "learning_rate": 0.00029616297615073624,
      "loss": 3.174338684082031,
      "step": 43450
    },
    {
      "epoch": 0.0256101884983466,
      "grad_norm": 0.3434971868991852,
      "learning_rate": 0.000296158560601645,
      "loss": 3.163721923828125,
      "step": 43500
    },
    {
      "epoch": 0.02563962549662056,
      "grad_norm": 0.3899316191673279,
      "learning_rate": 0.0002961541450525538,
      "loss": 3.1820196533203124,
      "step": 43550
    },
    {
      "epoch": 0.02566906249489452,
      "grad_norm": 0.3538844585418701,
      "learning_rate": 0.0002961497295034627,
      "loss": 3.192971496582031,
      "step": 43600
    },
    {
      "epoch": 0.02569849949316848,
      "grad_norm": 0.4491868019104004,
      "learning_rate": 0.00029614531395437144,
      "loss": 3.147752380371094,
      "step": 43650
    },
    {
      "epoch": 0.025727936491442442,
      "grad_norm": 0.35895365476608276,
      "learning_rate": 0.00029614089840528026,
      "loss": 3.215623474121094,
      "step": 43700
    },
    {
      "epoch": 0.025757373489716406,
      "grad_norm": 0.2585475444793701,
      "learning_rate": 0.00029613648285618907,
      "loss": 3.214078063964844,
      "step": 43750
    },
    {
      "epoch": 0.025786810487990367,
      "grad_norm": 0.5539636611938477,
      "learning_rate": 0.0002961320673070979,
      "loss": 3.1590765380859374,
      "step": 43800
    },
    {
      "epoch": 0.025816247486264328,
      "grad_norm": 0.3721752166748047,
      "learning_rate": 0.0002961276517580067,
      "loss": 3.178636169433594,
      "step": 43850
    },
    {
      "epoch": 0.02584568448453829,
      "grad_norm": 0.32989269495010376,
      "learning_rate": 0.00029612323620891546,
      "loss": 3.1558328247070313,
      "step": 43900
    },
    {
      "epoch": 0.025875121482812253,
      "grad_norm": 0.29314619302749634,
      "learning_rate": 0.00029611882065982433,
      "loss": 3.124910888671875,
      "step": 43950
    },
    {
      "epoch": 0.025904558481086214,
      "grad_norm": 0.31654226779937744,
      "learning_rate": 0.00029611440511073314,
      "loss": 3.129943542480469,
      "step": 44000
    },
    {
      "epoch": 0.025933995479360174,
      "grad_norm": 0.3399984836578369,
      "learning_rate": 0.0002961099895616419,
      "loss": 3.12861572265625,
      "step": 44050
    },
    {
      "epoch": 0.025963432477634135,
      "grad_norm": 0.32837533950805664,
      "learning_rate": 0.0002961055740125507,
      "loss": 3.147439270019531,
      "step": 44100
    },
    {
      "epoch": 0.0259928694759081,
      "grad_norm": 0.4116758704185486,
      "learning_rate": 0.00029610115846345953,
      "loss": 3.17194091796875,
      "step": 44150
    },
    {
      "epoch": 0.02602230647418206,
      "grad_norm": 0.4066229462623596,
      "learning_rate": 0.00029609674291436835,
      "loss": 3.195138244628906,
      "step": 44200
    },
    {
      "epoch": 0.02605174347245602,
      "grad_norm": 0.3364512324333191,
      "learning_rate": 0.00029609232736527716,
      "loss": 3.133672790527344,
      "step": 44250
    },
    {
      "epoch": 0.02608118047072998,
      "grad_norm": 0.4148530662059784,
      "learning_rate": 0.000296087911816186,
      "loss": 3.2248992919921875,
      "step": 44300
    },
    {
      "epoch": 0.026110617469003946,
      "grad_norm": 0.313284695148468,
      "learning_rate": 0.0002960834962670948,
      "loss": 3.2324069213867186,
      "step": 44350
    },
    {
      "epoch": 0.026140054467277907,
      "grad_norm": 0.5524761080741882,
      "learning_rate": 0.0002960790807180036,
      "loss": 3.1233184814453123,
      "step": 44400
    },
    {
      "epoch": 0.026169491465551867,
      "grad_norm": 0.3143770694732666,
      "learning_rate": 0.00029607466516891237,
      "loss": 3.2011932373046874,
      "step": 44450
    },
    {
      "epoch": 0.026198928463825828,
      "grad_norm": 0.3605794608592987,
      "learning_rate": 0.0002960702496198212,
      "loss": 3.2210684204101563,
      "step": 44500
    },
    {
      "epoch": 0.026228365462099792,
      "grad_norm": 0.3087241053581238,
      "learning_rate": 0.00029606583407073,
      "loss": 3.177120361328125,
      "step": 44550
    },
    {
      "epoch": 0.026257802460373753,
      "grad_norm": 0.26876217126846313,
      "learning_rate": 0.0002960614185216388,
      "loss": 3.1752853393554688,
      "step": 44600
    },
    {
      "epoch": 0.026287239458647714,
      "grad_norm": 0.3752470314502716,
      "learning_rate": 0.0002960570029725476,
      "loss": 3.1999221801757813,
      "step": 44650
    },
    {
      "epoch": 0.026316676456921675,
      "grad_norm": 0.4119313061237335,
      "learning_rate": 0.00029605258742345644,
      "loss": 3.172496337890625,
      "step": 44700
    },
    {
      "epoch": 0.02634611345519564,
      "grad_norm": 0.37847885489463806,
      "learning_rate": 0.00029604817187436525,
      "loss": 3.142645263671875,
      "step": 44750
    },
    {
      "epoch": 0.0263755504534696,
      "grad_norm": 0.31118616461753845,
      "learning_rate": 0.000296043756325274,
      "loss": 3.1398992919921875,
      "step": 44800
    },
    {
      "epoch": 0.02640498745174356,
      "grad_norm": 0.2863033413887024,
      "learning_rate": 0.00029603934077618283,
      "loss": 3.164620056152344,
      "step": 44850
    },
    {
      "epoch": 0.02643442445001752,
      "grad_norm": 0.4101048409938812,
      "learning_rate": 0.0002960349252270917,
      "loss": 3.1827798461914063,
      "step": 44900
    },
    {
      "epoch": 0.026463861448291486,
      "grad_norm": 0.33134451508522034,
      "learning_rate": 0.00029603050967800046,
      "loss": 3.146309509277344,
      "step": 44950
    },
    {
      "epoch": 0.026493298446565446,
      "grad_norm": 0.3401559293270111,
      "learning_rate": 0.00029602609412890927,
      "loss": 3.1552133178710937,
      "step": 45000
    },
    {
      "epoch": 0.026522735444839407,
      "grad_norm": 0.344407856464386,
      "learning_rate": 0.0002960216785798181,
      "loss": 3.1383566284179687,
      "step": 45050
    },
    {
      "epoch": 0.026552172443113368,
      "grad_norm": 0.29192396998405457,
      "learning_rate": 0.0002960172630307269,
      "loss": 3.210421447753906,
      "step": 45100
    },
    {
      "epoch": 0.026581609441387332,
      "grad_norm": 0.35983890295028687,
      "learning_rate": 0.0002960128474816357,
      "loss": 3.15560302734375,
      "step": 45150
    },
    {
      "epoch": 0.026611046439661293,
      "grad_norm": 0.35120198130607605,
      "learning_rate": 0.0002960084319325445,
      "loss": 3.113871765136719,
      "step": 45200
    },
    {
      "epoch": 0.026640483437935254,
      "grad_norm": 0.40215861797332764,
      "learning_rate": 0.00029600401638345334,
      "loss": 3.168427734375,
      "step": 45250
    },
    {
      "epoch": 0.026669920436209214,
      "grad_norm": 0.28755468130111694,
      "learning_rate": 0.00029599960083436216,
      "loss": 3.1991452026367186,
      "step": 45300
    },
    {
      "epoch": 0.02669935743448318,
      "grad_norm": 0.29394471645355225,
      "learning_rate": 0.0002959951852852709,
      "loss": 3.125282897949219,
      "step": 45350
    },
    {
      "epoch": 0.02672879443275714,
      "grad_norm": 0.28387323021888733,
      "learning_rate": 0.00029599076973617973,
      "loss": 3.171048583984375,
      "step": 45400
    },
    {
      "epoch": 0.0267582314310311,
      "grad_norm": 0.44179606437683105,
      "learning_rate": 0.00029598635418708855,
      "loss": 3.1246942138671874,
      "step": 45450
    },
    {
      "epoch": 0.02678766842930506,
      "grad_norm": 0.40504053235054016,
      "learning_rate": 0.00029598193863799736,
      "loss": 3.1745074462890623,
      "step": 45500
    },
    {
      "epoch": 0.02681710542757902,
      "grad_norm": 0.5301349759101868,
      "learning_rate": 0.0002959775230889062,
      "loss": 3.1266854858398436,
      "step": 45550
    },
    {
      "epoch": 0.026846542425852986,
      "grad_norm": 0.34842824935913086,
      "learning_rate": 0.000295973107539815,
      "loss": 3.1407440185546873,
      "step": 45600
    },
    {
      "epoch": 0.026875979424126947,
      "grad_norm": 0.3570592999458313,
      "learning_rate": 0.0002959686919907238,
      "loss": 3.1670196533203123,
      "step": 45650
    },
    {
      "epoch": 0.026905416422400907,
      "grad_norm": 0.3830150067806244,
      "learning_rate": 0.0002959642764416326,
      "loss": 3.1606326293945313,
      "step": 45700
    },
    {
      "epoch": 0.026934853420674868,
      "grad_norm": 0.38765302300453186,
      "learning_rate": 0.0002959598608925414,
      "loss": 3.189225158691406,
      "step": 45750
    },
    {
      "epoch": 0.026964290418948832,
      "grad_norm": 0.4475644826889038,
      "learning_rate": 0.0002959554453434502,
      "loss": 3.142057189941406,
      "step": 45800
    },
    {
      "epoch": 0.026993727417222793,
      "grad_norm": 0.35757917165756226,
      "learning_rate": 0.000295951029794359,
      "loss": 3.162414245605469,
      "step": 45850
    },
    {
      "epoch": 0.027023164415496754,
      "grad_norm": 0.33636152744293213,
      "learning_rate": 0.0002959466142452678,
      "loss": 3.121195983886719,
      "step": 45900
    },
    {
      "epoch": 0.027052601413770715,
      "grad_norm": 0.24081842601299286,
      "learning_rate": 0.00029594219869617664,
      "loss": 3.165047607421875,
      "step": 45950
    },
    {
      "epoch": 0.02708203841204468,
      "grad_norm": 0.47957611083984375,
      "learning_rate": 0.00029593778314708545,
      "loss": 3.19615234375,
      "step": 46000
    },
    {
      "epoch": 0.02711147541031864,
      "grad_norm": 0.4400424361228943,
      "learning_rate": 0.00029593336759799427,
      "loss": 3.1474755859375,
      "step": 46050
    },
    {
      "epoch": 0.0271409124085926,
      "grad_norm": 0.3157220184803009,
      "learning_rate": 0.0002959289520489031,
      "loss": 3.0967413330078126,
      "step": 46100
    },
    {
      "epoch": 0.02717034940686656,
      "grad_norm": 0.3207460939884186,
      "learning_rate": 0.00029592453649981184,
      "loss": 3.2098422241210938,
      "step": 46150
    },
    {
      "epoch": 0.027199786405140525,
      "grad_norm": 0.3640995919704437,
      "learning_rate": 0.0002959201209507207,
      "loss": 3.0722177124023435,
      "step": 46200
    },
    {
      "epoch": 0.027229223403414486,
      "grad_norm": 0.3455788791179657,
      "learning_rate": 0.0002959157054016295,
      "loss": 3.1439898681640623,
      "step": 46250
    },
    {
      "epoch": 0.027258660401688447,
      "grad_norm": 0.47757816314697266,
      "learning_rate": 0.0002959112898525383,
      "loss": 3.136019287109375,
      "step": 46300
    },
    {
      "epoch": 0.027288097399962408,
      "grad_norm": 0.4531172811985016,
      "learning_rate": 0.0002959068743034471,
      "loss": 3.127263488769531,
      "step": 46350
    },
    {
      "epoch": 0.027317534398236372,
      "grad_norm": 0.348557710647583,
      "learning_rate": 0.0002959024587543559,
      "loss": 3.115782470703125,
      "step": 46400
    },
    {
      "epoch": 0.027346971396510333,
      "grad_norm": 0.28779372572898865,
      "learning_rate": 0.00029589804320526473,
      "loss": 3.1235791015625,
      "step": 46450
    },
    {
      "epoch": 0.027376408394784293,
      "grad_norm": 0.3282870650291443,
      "learning_rate": 0.0002958936276561735,
      "loss": 3.0673931884765624,
      "step": 46500
    },
    {
      "epoch": 0.027405845393058254,
      "grad_norm": 0.3718392550945282,
      "learning_rate": 0.00029588921210708236,
      "loss": 3.1579327392578125,
      "step": 46550
    },
    {
      "epoch": 0.02743528239133222,
      "grad_norm": 0.3474173843860626,
      "learning_rate": 0.00029588479655799117,
      "loss": 3.1631640625,
      "step": 46600
    },
    {
      "epoch": 0.02746471938960618,
      "grad_norm": 0.36287301778793335,
      "learning_rate": 0.00029588038100889993,
      "loss": 3.1657595825195313,
      "step": 46650
    },
    {
      "epoch": 0.02749415638788014,
      "grad_norm": 0.2937576472759247,
      "learning_rate": 0.00029587596545980875,
      "loss": 3.1218856811523437,
      "step": 46700
    },
    {
      "epoch": 0.0275235933861541,
      "grad_norm": 0.286822646856308,
      "learning_rate": 0.00029587154991071756,
      "loss": 3.200117492675781,
      "step": 46750
    },
    {
      "epoch": 0.027553030384428065,
      "grad_norm": 0.28982946276664734,
      "learning_rate": 0.0002958671343616264,
      "loss": 3.1205316162109376,
      "step": 46800
    },
    {
      "epoch": 0.027582467382702026,
      "grad_norm": 0.3635452091693878,
      "learning_rate": 0.0002958627188125352,
      "loss": 3.1238372802734373,
      "step": 46850
    },
    {
      "epoch": 0.027611904380975986,
      "grad_norm": 0.2449709177017212,
      "learning_rate": 0.000295858303263444,
      "loss": 3.2334323120117188,
      "step": 46900
    },
    {
      "epoch": 0.027641341379249947,
      "grad_norm": 0.4339551329612732,
      "learning_rate": 0.0002958538877143528,
      "loss": 3.13289306640625,
      "step": 46950
    },
    {
      "epoch": 0.02767077837752391,
      "grad_norm": 0.2827685475349426,
      "learning_rate": 0.00029584947216526163,
      "loss": 3.1517584228515627,
      "step": 47000
    },
    {
      "epoch": 0.027700215375797872,
      "grad_norm": 0.4098069369792938,
      "learning_rate": 0.0002958450566161704,
      "loss": 3.162527770996094,
      "step": 47050
    },
    {
      "epoch": 0.027729652374071833,
      "grad_norm": 0.38044387102127075,
      "learning_rate": 0.0002958406410670792,
      "loss": 3.154069519042969,
      "step": 47100
    },
    {
      "epoch": 0.027759089372345794,
      "grad_norm": 0.5018683671951294,
      "learning_rate": 0.000295836225517988,
      "loss": 3.1280776977539064,
      "step": 47150
    },
    {
      "epoch": 0.027788526370619755,
      "grad_norm": 0.46165037155151367,
      "learning_rate": 0.00029583180996889684,
      "loss": 3.194757385253906,
      "step": 47200
    },
    {
      "epoch": 0.02781796336889372,
      "grad_norm": 0.28209125995635986,
      "learning_rate": 0.00029582739441980565,
      "loss": 3.122397766113281,
      "step": 47250
    },
    {
      "epoch": 0.02784740036716768,
      "grad_norm": 0.2659115493297577,
      "learning_rate": 0.00029582297887071447,
      "loss": 3.1068209838867187,
      "step": 47300
    },
    {
      "epoch": 0.02787683736544164,
      "grad_norm": 0.3520415127277374,
      "learning_rate": 0.0002958185633216233,
      "loss": 3.105587158203125,
      "step": 47350
    },
    {
      "epoch": 0.0279062743637156,
      "grad_norm": 0.38576850295066833,
      "learning_rate": 0.0002958141477725321,
      "loss": 3.2003472900390624,
      "step": 47400
    },
    {
      "epoch": 0.027935711361989565,
      "grad_norm": 0.2902798652648926,
      "learning_rate": 0.00029580973222344086,
      "loss": 3.161643371582031,
      "step": 47450
    },
    {
      "epoch": 0.027965148360263526,
      "grad_norm": 0.35946398973464966,
      "learning_rate": 0.0002958053166743497,
      "loss": 3.1283544921875,
      "step": 47500
    },
    {
      "epoch": 0.027994585358537487,
      "grad_norm": 0.33790016174316406,
      "learning_rate": 0.00029580090112525854,
      "loss": 3.1383416748046873,
      "step": 47550
    },
    {
      "epoch": 0.028024022356811448,
      "grad_norm": 0.42871591448783875,
      "learning_rate": 0.0002957964855761673,
      "loss": 3.2199801635742187,
      "step": 47600
    },
    {
      "epoch": 0.028053459355085412,
      "grad_norm": 0.3488813638687134,
      "learning_rate": 0.0002957920700270761,
      "loss": 3.164528503417969,
      "step": 47650
    },
    {
      "epoch": 0.028082896353359373,
      "grad_norm": 0.39149072766304016,
      "learning_rate": 0.00029578765447798493,
      "loss": 3.1649066162109376,
      "step": 47700
    },
    {
      "epoch": 0.028112333351633333,
      "grad_norm": 0.32711151242256165,
      "learning_rate": 0.00029578323892889374,
      "loss": 3.1814846801757812,
      "step": 47750
    },
    {
      "epoch": 0.028141770349907294,
      "grad_norm": 0.38896259665489197,
      "learning_rate": 0.00029577882337980256,
      "loss": 3.117653503417969,
      "step": 47800
    },
    {
      "epoch": 0.02817120734818126,
      "grad_norm": 0.3864522874355316,
      "learning_rate": 0.00029577440783071137,
      "loss": 3.137496032714844,
      "step": 47850
    },
    {
      "epoch": 0.02820064434645522,
      "grad_norm": 0.31975695490837097,
      "learning_rate": 0.0002957699922816202,
      "loss": 3.161602783203125,
      "step": 47900
    },
    {
      "epoch": 0.02823008134472918,
      "grad_norm": 0.35758647322654724,
      "learning_rate": 0.000295765576732529,
      "loss": 3.1051504516601565,
      "step": 47950
    },
    {
      "epoch": 0.02825951834300314,
      "grad_norm": 0.4807143807411194,
      "learning_rate": 0.00029576116118343776,
      "loss": 3.1229638671875,
      "step": 48000
    },
    {
      "epoch": 0.028288955341277105,
      "grad_norm": 0.3934558928012848,
      "learning_rate": 0.0002957567456343466,
      "loss": 3.167752380371094,
      "step": 48050
    },
    {
      "epoch": 0.028318392339551066,
      "grad_norm": 0.31533294916152954,
      "learning_rate": 0.0002957523300852554,
      "loss": 3.1236029052734375,
      "step": 48100
    },
    {
      "epoch": 0.028347829337825026,
      "grad_norm": 0.3604288399219513,
      "learning_rate": 0.0002957479145361642,
      "loss": 3.149962158203125,
      "step": 48150
    },
    {
      "epoch": 0.028377266336098987,
      "grad_norm": 0.3089168965816498,
      "learning_rate": 0.000295743498987073,
      "loss": 3.077811279296875,
      "step": 48200
    },
    {
      "epoch": 0.02840670333437295,
      "grad_norm": 0.31547290086746216,
      "learning_rate": 0.00029573908343798183,
      "loss": 3.1409677124023436,
      "step": 48250
    },
    {
      "epoch": 0.028436140332646912,
      "grad_norm": 0.40962815284729004,
      "learning_rate": 0.00029573466788889065,
      "loss": 3.1532342529296873,
      "step": 48300
    },
    {
      "epoch": 0.028465577330920873,
      "grad_norm": 0.3702750504016876,
      "learning_rate": 0.0002957302523397994,
      "loss": 3.1379293823242187,
      "step": 48350
    },
    {
      "epoch": 0.028495014329194834,
      "grad_norm": 0.29024842381477356,
      "learning_rate": 0.0002957258367907082,
      "loss": 3.121209716796875,
      "step": 48400
    },
    {
      "epoch": 0.028524451327468798,
      "grad_norm": 0.3680042326450348,
      "learning_rate": 0.00029572142124161704,
      "loss": 3.0851870727539064,
      "step": 48450
    },
    {
      "epoch": 0.02855388832574276,
      "grad_norm": 0.4963594079017639,
      "learning_rate": 0.00029571700569252585,
      "loss": 3.2077157592773435,
      "step": 48500
    },
    {
      "epoch": 0.02858332532401672,
      "grad_norm": 0.36988216638565063,
      "learning_rate": 0.00029571259014343467,
      "loss": 3.1487637329101563,
      "step": 48550
    },
    {
      "epoch": 0.02861276232229068,
      "grad_norm": 0.4131982922554016,
      "learning_rate": 0.0002957081745943435,
      "loss": 3.1522918701171876,
      "step": 48600
    },
    {
      "epoch": 0.028642199320564644,
      "grad_norm": 0.30599445104599,
      "learning_rate": 0.0002957037590452523,
      "loss": 3.1450601196289063,
      "step": 48650
    },
    {
      "epoch": 0.028671636318838605,
      "grad_norm": 0.31610366702079773,
      "learning_rate": 0.0002956993434961611,
      "loss": 3.1487371826171877,
      "step": 48700
    },
    {
      "epoch": 0.028701073317112566,
      "grad_norm": 0.4011955261230469,
      "learning_rate": 0.00029569492794706987,
      "loss": 3.1754876708984376,
      "step": 48750
    },
    {
      "epoch": 0.028730510315386527,
      "grad_norm": 0.40532365441322327,
      "learning_rate": 0.0002956905123979787,
      "loss": 3.196721496582031,
      "step": 48800
    },
    {
      "epoch": 0.02875994731366049,
      "grad_norm": 0.42118945717811584,
      "learning_rate": 0.00029568609684888755,
      "loss": 3.0930279541015624,
      "step": 48850
    },
    {
      "epoch": 0.02878938431193445,
      "grad_norm": 0.3413017988204956,
      "learning_rate": 0.0002956816812997963,
      "loss": 3.168067626953125,
      "step": 48900
    },
    {
      "epoch": 0.028818821310208412,
      "grad_norm": 0.2749626934528351,
      "learning_rate": 0.00029567726575070513,
      "loss": 3.1435061645507814,
      "step": 48950
    },
    {
      "epoch": 0.028848258308482373,
      "grad_norm": 0.37463870644569397,
      "learning_rate": 0.00029567285020161394,
      "loss": 3.1952288818359373,
      "step": 49000
    },
    {
      "epoch": 0.028877695306756334,
      "grad_norm": 0.4485418498516083,
      "learning_rate": 0.00029566843465252276,
      "loss": 3.1457418823242187,
      "step": 49050
    },
    {
      "epoch": 0.028907132305030298,
      "grad_norm": 0.3112456798553467,
      "learning_rate": 0.00029566401910343157,
      "loss": 3.1429379272460936,
      "step": 49100
    },
    {
      "epoch": 0.02893656930330426,
      "grad_norm": 0.30648937821388245,
      "learning_rate": 0.0002956596035543404,
      "loss": 3.110986633300781,
      "step": 49150
    },
    {
      "epoch": 0.02896600630157822,
      "grad_norm": 0.3727593719959259,
      "learning_rate": 0.0002956551880052492,
      "loss": 3.125091857910156,
      "step": 49200
    },
    {
      "epoch": 0.02899544329985218,
      "grad_norm": 0.32527413964271545,
      "learning_rate": 0.000295650772456158,
      "loss": 3.1411199951171875,
      "step": 49250
    },
    {
      "epoch": 0.029024880298126145,
      "grad_norm": 0.5576922297477722,
      "learning_rate": 0.0002956463569070668,
      "loss": 3.093547668457031,
      "step": 49300
    },
    {
      "epoch": 0.029054317296400105,
      "grad_norm": 0.29772675037384033,
      "learning_rate": 0.0002956419413579756,
      "loss": 3.1503118896484374,
      "step": 49350
    },
    {
      "epoch": 0.029083754294674066,
      "grad_norm": 0.32615184783935547,
      "learning_rate": 0.0002956375258088844,
      "loss": 3.153142395019531,
      "step": 49400
    },
    {
      "epoch": 0.029113191292948027,
      "grad_norm": 0.36862096190452576,
      "learning_rate": 0.0002956331102597932,
      "loss": 3.1322540283203124,
      "step": 49450
    },
    {
      "epoch": 0.02914262829122199,
      "grad_norm": 0.39233314990997314,
      "learning_rate": 0.00029562869471070203,
      "loss": 3.1164315795898436,
      "step": 49500
    },
    {
      "epoch": 0.029172065289495952,
      "grad_norm": 0.3334544897079468,
      "learning_rate": 0.00029562427916161085,
      "loss": 3.1120596313476563,
      "step": 49550
    },
    {
      "epoch": 0.029201502287769913,
      "grad_norm": 0.3732755184173584,
      "learning_rate": 0.00029561986361251966,
      "loss": 3.14333740234375,
      "step": 49600
    },
    {
      "epoch": 0.029230939286043874,
      "grad_norm": 0.4033997654914856,
      "learning_rate": 0.0002956154480634285,
      "loss": 3.1742819213867186,
      "step": 49650
    },
    {
      "epoch": 0.029260376284317838,
      "grad_norm": 0.32923829555511475,
      "learning_rate": 0.00029561103251433724,
      "loss": 3.1491485595703126,
      "step": 49700
    },
    {
      "epoch": 0.0292898132825918,
      "grad_norm": 0.3075374364852905,
      "learning_rate": 0.00029560661696524605,
      "loss": 3.093435974121094,
      "step": 49750
    },
    {
      "epoch": 0.02931925028086576,
      "grad_norm": 0.3389895558357239,
      "learning_rate": 0.0002956022014161549,
      "loss": 3.036296081542969,
      "step": 49800
    },
    {
      "epoch": 0.02934868727913972,
      "grad_norm": 0.31929051876068115,
      "learning_rate": 0.0002955977858670637,
      "loss": 3.1331549072265625,
      "step": 49850
    },
    {
      "epoch": 0.029378124277413684,
      "grad_norm": 0.3502655327320099,
      "learning_rate": 0.0002955933703179725,
      "loss": 3.1205038452148437,
      "step": 49900
    },
    {
      "epoch": 0.029407561275687645,
      "grad_norm": 0.36345794796943665,
      "learning_rate": 0.0002955889547688813,
      "loss": 3.139251708984375,
      "step": 49950
    },
    {
      "epoch": 0.029436998273961606,
      "grad_norm": 0.8029529452323914,
      "learning_rate": 0.0002955845392197901,
      "loss": 3.1454083251953127,
      "step": 50000
    },
    {
      "epoch": 0.029466435272235567,
      "grad_norm": 0.45979979634284973,
      "learning_rate": 0.0002955801236706989,
      "loss": 3.1489141845703124,
      "step": 50050
    },
    {
      "epoch": 0.02949587227050953,
      "grad_norm": 0.33692917227745056,
      "learning_rate": 0.0002955757081216077,
      "loss": 3.1364865112304687,
      "step": 50100
    },
    {
      "epoch": 0.02952530926878349,
      "grad_norm": 0.2908169627189636,
      "learning_rate": 0.00029557129257251657,
      "loss": 3.0864877319335937,
      "step": 50150
    },
    {
      "epoch": 0.029554746267057452,
      "grad_norm": 0.4481618404388428,
      "learning_rate": 0.00029556687702342533,
      "loss": 3.117489929199219,
      "step": 50200
    },
    {
      "epoch": 0.029584183265331413,
      "grad_norm": 0.3863079249858856,
      "learning_rate": 0.00029556246147433414,
      "loss": 3.093724365234375,
      "step": 50250
    },
    {
      "epoch": 0.029613620263605377,
      "grad_norm": 0.3971397876739502,
      "learning_rate": 0.00029555804592524296,
      "loss": 3.1477066040039063,
      "step": 50300
    },
    {
      "epoch": 0.029643057261879338,
      "grad_norm": 0.3637728691101074,
      "learning_rate": 0.00029555363037615177,
      "loss": 3.0922012329101562,
      "step": 50350
    },
    {
      "epoch": 0.0296724942601533,
      "grad_norm": 0.3176400363445282,
      "learning_rate": 0.0002955492148270606,
      "loss": 3.1491854858398436,
      "step": 50400
    },
    {
      "epoch": 0.02970193125842726,
      "grad_norm": 0.37296056747436523,
      "learning_rate": 0.0002955447992779694,
      "loss": 3.053348388671875,
      "step": 50450
    },
    {
      "epoch": 0.029731368256701224,
      "grad_norm": 0.3499937653541565,
      "learning_rate": 0.0002955403837288782,
      "loss": 3.0612112426757814,
      "step": 50500
    },
    {
      "epoch": 0.029760805254975185,
      "grad_norm": 0.290060818195343,
      "learning_rate": 0.00029553596817978703,
      "loss": 3.1020465087890625,
      "step": 50550
    },
    {
      "epoch": 0.029790242253249145,
      "grad_norm": 0.3266589641571045,
      "learning_rate": 0.0002955315526306958,
      "loss": 3.141214599609375,
      "step": 50600
    },
    {
      "epoch": 0.029819679251523106,
      "grad_norm": 0.32461926341056824,
      "learning_rate": 0.0002955271370816046,
      "loss": 3.12266845703125,
      "step": 50650
    },
    {
      "epoch": 0.02984911624979707,
      "grad_norm": 0.28531184792518616,
      "learning_rate": 0.0002955227215325134,
      "loss": 3.098194274902344,
      "step": 50700
    },
    {
      "epoch": 0.02987855324807103,
      "grad_norm": 0.3675709366798401,
      "learning_rate": 0.00029551830598342223,
      "loss": 3.1133914184570313,
      "step": 50750
    },
    {
      "epoch": 0.029907990246344992,
      "grad_norm": 0.3599429130554199,
      "learning_rate": 0.00029551389043433105,
      "loss": 3.1060260009765623,
      "step": 50800
    },
    {
      "epoch": 0.029937427244618953,
      "grad_norm": 0.3418704867362976,
      "learning_rate": 0.00029550947488523986,
      "loss": 3.0862640380859374,
      "step": 50850
    },
    {
      "epoch": 0.029966864242892913,
      "grad_norm": 0.31689244508743286,
      "learning_rate": 0.0002955050593361487,
      "loss": 3.11436279296875,
      "step": 50900
    },
    {
      "epoch": 0.029996301241166878,
      "grad_norm": 0.40309491753578186,
      "learning_rate": 0.0002955006437870575,
      "loss": 3.0912689208984374,
      "step": 50950
    },
    {
      "epoch": 0.03002573823944084,
      "grad_norm": 0.2930799126625061,
      "learning_rate": 0.00029549622823796625,
      "loss": 3.057187805175781,
      "step": 51000
    },
    {
      "epoch": 0.0300551752377148,
      "grad_norm": 0.41112062335014343,
      "learning_rate": 0.00029549181268887507,
      "loss": 3.0628628540039062,
      "step": 51050
    },
    {
      "epoch": 0.03008461223598876,
      "grad_norm": 0.3723728656768799,
      "learning_rate": 0.00029548739713978393,
      "loss": 3.13063232421875,
      "step": 51100
    },
    {
      "epoch": 0.030114049234262724,
      "grad_norm": 0.39304110407829285,
      "learning_rate": 0.0002954829815906927,
      "loss": 3.0706436157226564,
      "step": 51150
    },
    {
      "epoch": 0.030143486232536685,
      "grad_norm": 0.30212607979774475,
      "learning_rate": 0.0002954785660416015,
      "loss": 3.1506378173828127,
      "step": 51200
    },
    {
      "epoch": 0.030172923230810646,
      "grad_norm": 0.3334600627422333,
      "learning_rate": 0.0002954741504925103,
      "loss": 3.070340270996094,
      "step": 51250
    },
    {
      "epoch": 0.030202360229084606,
      "grad_norm": 0.5296679735183716,
      "learning_rate": 0.00029546973494341914,
      "loss": 3.0584646606445314,
      "step": 51300
    },
    {
      "epoch": 0.03023179722735857,
      "grad_norm": 0.3373028039932251,
      "learning_rate": 0.00029546531939432795,
      "loss": 3.098596496582031,
      "step": 51350
    },
    {
      "epoch": 0.03026123422563253,
      "grad_norm": 0.376110702753067,
      "learning_rate": 0.0002954609038452367,
      "loss": 3.1733160400390625,
      "step": 51400
    },
    {
      "epoch": 0.030290671223906492,
      "grad_norm": 0.3790414333343506,
      "learning_rate": 0.0002954564882961456,
      "loss": 3.055797119140625,
      "step": 51450
    },
    {
      "epoch": 0.030320108222180453,
      "grad_norm": 0.3399900197982788,
      "learning_rate": 0.0002954520727470544,
      "loss": 3.1229318237304686,
      "step": 51500
    },
    {
      "epoch": 0.030349545220454417,
      "grad_norm": 0.32600316405296326,
      "learning_rate": 0.00029544765719796316,
      "loss": 3.070262756347656,
      "step": 51550
    },
    {
      "epoch": 0.030378982218728378,
      "grad_norm": 0.30688586831092834,
      "learning_rate": 0.00029544324164887197,
      "loss": 3.123429260253906,
      "step": 51600
    },
    {
      "epoch": 0.03040841921700234,
      "grad_norm": 0.33274397253990173,
      "learning_rate": 0.0002954388260997808,
      "loss": 3.091960144042969,
      "step": 51650
    },
    {
      "epoch": 0.0304378562152763,
      "grad_norm": 0.325549840927124,
      "learning_rate": 0.0002954344105506896,
      "loss": 3.1380035400390627,
      "step": 51700
    },
    {
      "epoch": 0.030467293213550264,
      "grad_norm": 0.3637613356113434,
      "learning_rate": 0.0002954299950015984,
      "loss": 3.061991882324219,
      "step": 51750
    },
    {
      "epoch": 0.030496730211824224,
      "grad_norm": 0.3331155478954315,
      "learning_rate": 0.00029542557945250723,
      "loss": 3.1042214965820314,
      "step": 51800
    },
    {
      "epoch": 0.030526167210098185,
      "grad_norm": 0.5314979553222656,
      "learning_rate": 0.00029542116390341604,
      "loss": 3.142459716796875,
      "step": 51850
    },
    {
      "epoch": 0.030555604208372146,
      "grad_norm": 0.2887026071548462,
      "learning_rate": 0.0002954167483543248,
      "loss": 3.06982421875,
      "step": 51900
    },
    {
      "epoch": 0.03058504120664611,
      "grad_norm": 0.3714282214641571,
      "learning_rate": 0.0002954123328052336,
      "loss": 3.125784606933594,
      "step": 51950
    },
    {
      "epoch": 0.03061447820492007,
      "grad_norm": 0.43876373767852783,
      "learning_rate": 0.00029540791725614243,
      "loss": 3.219000244140625,
      "step": 52000
    },
    {
      "epoch": 0.030643915203194032,
      "grad_norm": 0.28773289918899536,
      "learning_rate": 0.00029540350170705125,
      "loss": 3.152445068359375,
      "step": 52050
    },
    {
      "epoch": 0.030673352201467993,
      "grad_norm": 0.40366581082344055,
      "learning_rate": 0.00029539908615796006,
      "loss": 3.0836785888671874,
      "step": 52100
    },
    {
      "epoch": 0.030702789199741957,
      "grad_norm": 0.34363213181495667,
      "learning_rate": 0.0002953946706088689,
      "loss": 3.1106057739257813,
      "step": 52150
    },
    {
      "epoch": 0.030732226198015918,
      "grad_norm": 0.3136003315448761,
      "learning_rate": 0.0002953902550597777,
      "loss": 3.16740234375,
      "step": 52200
    },
    {
      "epoch": 0.03076166319628988,
      "grad_norm": 0.3441563844680786,
      "learning_rate": 0.0002953858395106865,
      "loss": 3.1191925048828124,
      "step": 52250
    },
    {
      "epoch": 0.03079110019456384,
      "grad_norm": 0.433439165353775,
      "learning_rate": 0.00029538142396159526,
      "loss": 3.0960653686523436,
      "step": 52300
    },
    {
      "epoch": 0.030820537192837803,
      "grad_norm": 0.4175854027271271,
      "learning_rate": 0.0002953770084125041,
      "loss": 3.1058285522460936,
      "step": 52350
    },
    {
      "epoch": 0.030849974191111764,
      "grad_norm": 0.3281439244747162,
      "learning_rate": 0.00029537259286341295,
      "loss": 3.0814022827148437,
      "step": 52400
    },
    {
      "epoch": 0.030879411189385725,
      "grad_norm": 0.387580543756485,
      "learning_rate": 0.0002953681773143217,
      "loss": 3.106121826171875,
      "step": 52450
    },
    {
      "epoch": 0.030908848187659686,
      "grad_norm": 0.3115745186805725,
      "learning_rate": 0.0002953637617652305,
      "loss": 3.0780429077148437,
      "step": 52500
    },
    {
      "epoch": 0.03093828518593365,
      "grad_norm": 0.388996958732605,
      "learning_rate": 0.00029535934621613934,
      "loss": 3.115902099609375,
      "step": 52550
    },
    {
      "epoch": 0.03096772218420761,
      "grad_norm": 0.32805803418159485,
      "learning_rate": 0.00029535493066704815,
      "loss": 3.0466293334960937,
      "step": 52600
    },
    {
      "epoch": 0.03099715918248157,
      "grad_norm": 0.30090421438217163,
      "learning_rate": 0.00029535051511795697,
      "loss": 3.122732849121094,
      "step": 52650
    },
    {
      "epoch": 0.031026596180755532,
      "grad_norm": 0.3396115303039551,
      "learning_rate": 0.0002953460995688657,
      "loss": 3.077347412109375,
      "step": 52700
    },
    {
      "epoch": 0.031056033179029493,
      "grad_norm": 0.2237304449081421,
      "learning_rate": 0.0002953416840197746,
      "loss": 3.1370254516601563,
      "step": 52750
    },
    {
      "epoch": 0.031085470177303457,
      "grad_norm": 0.31867435574531555,
      "learning_rate": 0.0002953372684706834,
      "loss": 3.09147705078125,
      "step": 52800
    },
    {
      "epoch": 0.031114907175577418,
      "grad_norm": 0.4212453365325928,
      "learning_rate": 0.00029533285292159217,
      "loss": 3.1452947998046876,
      "step": 52850
    },
    {
      "epoch": 0.03114434417385138,
      "grad_norm": 0.2892448306083679,
      "learning_rate": 0.000295328437372501,
      "loss": 3.1234786987304686,
      "step": 52900
    },
    {
      "epoch": 0.03117378117212534,
      "grad_norm": 0.41096973419189453,
      "learning_rate": 0.0002953240218234098,
      "loss": 3.0912515258789064,
      "step": 52950
    },
    {
      "epoch": 0.031203218170399304,
      "grad_norm": 0.5609623193740845,
      "learning_rate": 0.0002953196062743186,
      "loss": 3.03216552734375,
      "step": 53000
    },
    {
      "epoch": 0.031232655168673264,
      "grad_norm": 0.4838583171367645,
      "learning_rate": 0.00029531519072522743,
      "loss": 3.0332183837890625,
      "step": 53050
    },
    {
      "epoch": 0.031262092166947225,
      "grad_norm": 0.2962011694908142,
      "learning_rate": 0.00029531077517613624,
      "loss": 3.1060174560546874,
      "step": 53100
    },
    {
      "epoch": 0.03129152916522119,
      "grad_norm": 0.41562893986701965,
      "learning_rate": 0.00029530635962704506,
      "loss": 3.1352056884765624,
      "step": 53150
    },
    {
      "epoch": 0.03132096616349515,
      "grad_norm": 0.3613477647304535,
      "learning_rate": 0.00029530194407795387,
      "loss": 3.1372274780273437,
      "step": 53200
    },
    {
      "epoch": 0.03135040316176911,
      "grad_norm": 0.31264689564704895,
      "learning_rate": 0.00029529752852886263,
      "loss": 3.0861447143554686,
      "step": 53250
    },
    {
      "epoch": 0.031379840160043075,
      "grad_norm": 0.30406448245048523,
      "learning_rate": 0.00029529311297977145,
      "loss": 3.092449035644531,
      "step": 53300
    },
    {
      "epoch": 0.03140927715831703,
      "grad_norm": 0.34123295545578003,
      "learning_rate": 0.00029528869743068026,
      "loss": 3.132319030761719,
      "step": 53350
    },
    {
      "epoch": 0.031438714156591,
      "grad_norm": 0.35941964387893677,
      "learning_rate": 0.0002952842818815891,
      "loss": 3.139451904296875,
      "step": 53400
    },
    {
      "epoch": 0.031468151154864954,
      "grad_norm": 0.5264203548431396,
      "learning_rate": 0.0002952798663324979,
      "loss": 3.0871627807617186,
      "step": 53450
    },
    {
      "epoch": 0.03149758815313892,
      "grad_norm": 0.3635786771774292,
      "learning_rate": 0.0002952754507834067,
      "loss": 3.1244940185546874,
      "step": 53500
    },
    {
      "epoch": 0.03152702515141288,
      "grad_norm": 0.37433457374572754,
      "learning_rate": 0.0002952710352343155,
      "loss": 3.1525103759765627,
      "step": 53550
    },
    {
      "epoch": 0.03155646214968684,
      "grad_norm": 0.34269073605537415,
      "learning_rate": 0.0002952666196852243,
      "loss": 3.091374816894531,
      "step": 53600
    },
    {
      "epoch": 0.031585899147960804,
      "grad_norm": 0.27019479870796204,
      "learning_rate": 0.0002952622041361331,
      "loss": 3.1034368896484374,
      "step": 53650
    },
    {
      "epoch": 0.03161533614623477,
      "grad_norm": 0.28589123487472534,
      "learning_rate": 0.00029525778858704196,
      "loss": 3.08233642578125,
      "step": 53700
    },
    {
      "epoch": 0.031644773144508725,
      "grad_norm": 0.35975897312164307,
      "learning_rate": 0.0002952533730379507,
      "loss": 3.082213134765625,
      "step": 53750
    },
    {
      "epoch": 0.03167421014278269,
      "grad_norm": 0.31498920917510986,
      "learning_rate": 0.00029524895748885954,
      "loss": 3.0727099609375,
      "step": 53800
    },
    {
      "epoch": 0.03170364714105665,
      "grad_norm": 0.2657202184200287,
      "learning_rate": 0.00029524454193976835,
      "loss": 3.1103640747070314,
      "step": 53850
    },
    {
      "epoch": 0.03173308413933061,
      "grad_norm": 0.4319733679294586,
      "learning_rate": 0.00029524012639067717,
      "loss": 3.0827285766601564,
      "step": 53900
    },
    {
      "epoch": 0.031762521137604575,
      "grad_norm": 0.45037969946861267,
      "learning_rate": 0.000295235710841586,
      "loss": 3.0481436157226565,
      "step": 53950
    },
    {
      "epoch": 0.03179195813587853,
      "grad_norm": 0.3084319829940796,
      "learning_rate": 0.00029523129529249474,
      "loss": 3.1004351806640624,
      "step": 54000
    },
    {
      "epoch": 0.0318213951341525,
      "grad_norm": 0.49810123443603516,
      "learning_rate": 0.0002952268797434036,
      "loss": 3.102122802734375,
      "step": 54050
    },
    {
      "epoch": 0.03185083213242646,
      "grad_norm": 0.35637006163597107,
      "learning_rate": 0.0002952224641943124,
      "loss": 3.0871356201171873,
      "step": 54100
    },
    {
      "epoch": 0.03188026913070042,
      "grad_norm": 0.24572619795799255,
      "learning_rate": 0.0002952180486452212,
      "loss": 3.0872650146484375,
      "step": 54150
    },
    {
      "epoch": 0.03190970612897438,
      "grad_norm": 0.36404886841773987,
      "learning_rate": 0.00029521363309613,
      "loss": 3.0806393432617187,
      "step": 54200
    },
    {
      "epoch": 0.03193914312724834,
      "grad_norm": 0.3904375731945038,
      "learning_rate": 0.0002952092175470388,
      "loss": 3.079248352050781,
      "step": 54250
    },
    {
      "epoch": 0.031968580125522304,
      "grad_norm": 0.5357195734977722,
      "learning_rate": 0.00029520480199794763,
      "loss": 3.1272201538085938,
      "step": 54300
    },
    {
      "epoch": 0.03199801712379627,
      "grad_norm": 0.29050639271736145,
      "learning_rate": 0.00029520038644885644,
      "loss": 3.03983642578125,
      "step": 54350
    },
    {
      "epoch": 0.032027454122070226,
      "grad_norm": 0.27860090136528015,
      "learning_rate": 0.00029519597089976526,
      "loss": 3.092564697265625,
      "step": 54400
    },
    {
      "epoch": 0.03205689112034419,
      "grad_norm": 0.3423644006252289,
      "learning_rate": 0.00029519155535067407,
      "loss": 3.047486572265625,
      "step": 54450
    },
    {
      "epoch": 0.03208632811861815,
      "grad_norm": 0.3475925624370575,
      "learning_rate": 0.0002951871398015829,
      "loss": 3.0673037719726564,
      "step": 54500
    },
    {
      "epoch": 0.03211576511689211,
      "grad_norm": 0.3894612193107605,
      "learning_rate": 0.00029518272425249165,
      "loss": 3.1072183227539063,
      "step": 54550
    },
    {
      "epoch": 0.032145202115166076,
      "grad_norm": 0.3459846079349518,
      "learning_rate": 0.00029517830870340046,
      "loss": 3.150111083984375,
      "step": 54600
    },
    {
      "epoch": 0.03217463911344003,
      "grad_norm": 0.365872859954834,
      "learning_rate": 0.0002951738931543093,
      "loss": 3.056748352050781,
      "step": 54650
    },
    {
      "epoch": 0.032204076111714,
      "grad_norm": 0.358283132314682,
      "learning_rate": 0.0002951694776052181,
      "loss": 3.041001281738281,
      "step": 54700
    },
    {
      "epoch": 0.03223351310998796,
      "grad_norm": 0.3190099000930786,
      "learning_rate": 0.0002951650620561269,
      "loss": 3.113519592285156,
      "step": 54750
    },
    {
      "epoch": 0.03226295010826192,
      "grad_norm": 0.3329160809516907,
      "learning_rate": 0.0002951606465070357,
      "loss": 3.1277044677734374,
      "step": 54800
    },
    {
      "epoch": 0.03229238710653588,
      "grad_norm": 0.32831257581710815,
      "learning_rate": 0.00029515623095794453,
      "loss": 3.044276123046875,
      "step": 54850
    },
    {
      "epoch": 0.03232182410480984,
      "grad_norm": 0.28304192423820496,
      "learning_rate": 0.00029515181540885335,
      "loss": 3.0971499633789064,
      "step": 54900
    },
    {
      "epoch": 0.032351261103083805,
      "grad_norm": 0.3260906934738159,
      "learning_rate": 0.0002951473998597621,
      "loss": 3.118294677734375,
      "step": 54950
    },
    {
      "epoch": 0.03238069810135777,
      "grad_norm": 0.40868309140205383,
      "learning_rate": 0.0002951429843106709,
      "loss": 3.113529052734375,
      "step": 55000
    },
    {
      "epoch": 0.032410135099631726,
      "grad_norm": 0.3379126191139221,
      "learning_rate": 0.00029513856876157974,
      "loss": 3.0554776000976562,
      "step": 55050
    },
    {
      "epoch": 0.03243957209790569,
      "grad_norm": 0.32060280442237854,
      "learning_rate": 0.00029513415321248855,
      "loss": 3.05168212890625,
      "step": 55100
    },
    {
      "epoch": 0.032469009096179655,
      "grad_norm": 0.4328796863555908,
      "learning_rate": 0.00029512973766339737,
      "loss": 3.086100769042969,
      "step": 55150
    },
    {
      "epoch": 0.03249844609445361,
      "grad_norm": 0.40658456087112427,
      "learning_rate": 0.0002951253221143062,
      "loss": 3.0677859497070314,
      "step": 55200
    },
    {
      "epoch": 0.032527883092727576,
      "grad_norm": 0.2808285653591156,
      "learning_rate": 0.000295120906565215,
      "loss": 3.129729919433594,
      "step": 55250
    },
    {
      "epoch": 0.03255732009100153,
      "grad_norm": 0.324969083070755,
      "learning_rate": 0.00029511649101612375,
      "loss": 3.0893844604492187,
      "step": 55300
    },
    {
      "epoch": 0.0325867570892755,
      "grad_norm": 0.35364148020744324,
      "learning_rate": 0.0002951120754670326,
      "loss": 3.10419189453125,
      "step": 55350
    },
    {
      "epoch": 0.03261619408754946,
      "grad_norm": 0.3343355655670166,
      "learning_rate": 0.00029510765991794144,
      "loss": 3.1022299194335936,
      "step": 55400
    },
    {
      "epoch": 0.03264563108582342,
      "grad_norm": 0.32009056210517883,
      "learning_rate": 0.0002951032443688502,
      "loss": 3.0745733642578124,
      "step": 55450
    },
    {
      "epoch": 0.03267506808409738,
      "grad_norm": 0.34047675132751465,
      "learning_rate": 0.000295098828819759,
      "loss": 3.1177105712890625,
      "step": 55500
    },
    {
      "epoch": 0.03270450508237135,
      "grad_norm": 0.5621446967124939,
      "learning_rate": 0.0002950944132706678,
      "loss": 3.0826409912109374,
      "step": 55550
    },
    {
      "epoch": 0.032733942080645305,
      "grad_norm": 0.4223202168941498,
      "learning_rate": 0.00029508999772157664,
      "loss": 3.0604617309570314,
      "step": 55600
    },
    {
      "epoch": 0.03276337907891927,
      "grad_norm": 0.3531733751296997,
      "learning_rate": 0.00029508558217248546,
      "loss": 3.0902166748046875,
      "step": 55650
    },
    {
      "epoch": 0.032792816077193226,
      "grad_norm": 0.3428526222705841,
      "learning_rate": 0.00029508116662339427,
      "loss": 3.102550048828125,
      "step": 55700
    },
    {
      "epoch": 0.03282225307546719,
      "grad_norm": 0.27442553639411926,
      "learning_rate": 0.0002950767510743031,
      "loss": 3.068128967285156,
      "step": 55750
    },
    {
      "epoch": 0.032851690073741155,
      "grad_norm": 0.2844408452510834,
      "learning_rate": 0.0002950723355252119,
      "loss": 3.0974160766601564,
      "step": 55800
    },
    {
      "epoch": 0.03288112707201511,
      "grad_norm": 0.27269381284713745,
      "learning_rate": 0.00029506791997612066,
      "loss": 3.12382080078125,
      "step": 55850
    },
    {
      "epoch": 0.032910564070289076,
      "grad_norm": 0.5252366065979004,
      "learning_rate": 0.0002950635044270295,
      "loss": 3.109764404296875,
      "step": 55900
    },
    {
      "epoch": 0.03294000106856304,
      "grad_norm": 0.37370988726615906,
      "learning_rate": 0.0002950590888779383,
      "loss": 3.0648663330078123,
      "step": 55950
    },
    {
      "epoch": 0.032969438066837,
      "grad_norm": 0.34394726157188416,
      "learning_rate": 0.0002950546733288471,
      "loss": 3.0893380737304685,
      "step": 56000
    },
    {
      "epoch": 0.03299887506511096,
      "grad_norm": 0.35885581374168396,
      "learning_rate": 0.0002950502577797559,
      "loss": 3.0835638427734375,
      "step": 56050
    },
    {
      "epoch": 0.03302831206338492,
      "grad_norm": 0.42599478363990784,
      "learning_rate": 0.00029504584223066473,
      "loss": 3.0433474731445314,
      "step": 56100
    },
    {
      "epoch": 0.033057749061658884,
      "grad_norm": 0.4606510102748871,
      "learning_rate": 0.00029504142668157355,
      "loss": 3.1198272705078125,
      "step": 56150
    },
    {
      "epoch": 0.03308718605993285,
      "grad_norm": 0.4351559281349182,
      "learning_rate": 0.00029503701113248236,
      "loss": 3.125446472167969,
      "step": 56200
    },
    {
      "epoch": 0.033116623058206805,
      "grad_norm": 0.33799147605895996,
      "learning_rate": 0.0002950325955833911,
      "loss": 3.057550964355469,
      "step": 56250
    },
    {
      "epoch": 0.03314606005648077,
      "grad_norm": 0.33496391773223877,
      "learning_rate": 0.00029502818003429994,
      "loss": 3.1029849243164063,
      "step": 56300
    },
    {
      "epoch": 0.03317549705475473,
      "grad_norm": 0.4359044134616852,
      "learning_rate": 0.0002950237644852088,
      "loss": 3.071239013671875,
      "step": 56350
    },
    {
      "epoch": 0.03320493405302869,
      "grad_norm": 0.4206089973449707,
      "learning_rate": 0.00029501934893611756,
      "loss": 3.0841098022460938,
      "step": 56400
    },
    {
      "epoch": 0.033234371051302655,
      "grad_norm": 0.40799179673194885,
      "learning_rate": 0.0002950149333870264,
      "loss": 3.010738525390625,
      "step": 56450
    },
    {
      "epoch": 0.03326380804957661,
      "grad_norm": 0.3720875084400177,
      "learning_rate": 0.0002950105178379352,
      "loss": 3.115255126953125,
      "step": 56500
    },
    {
      "epoch": 0.03329324504785058,
      "grad_norm": 0.6249273419380188,
      "learning_rate": 0.000295006102288844,
      "loss": 3.0907196044921874,
      "step": 56550
    },
    {
      "epoch": 0.03332268204612454,
      "grad_norm": 0.29867130517959595,
      "learning_rate": 0.0002950016867397528,
      "loss": 3.0853765869140624,
      "step": 56600
    },
    {
      "epoch": 0.0333521190443985,
      "grad_norm": 0.24184280633926392,
      "learning_rate": 0.00029499727119066164,
      "loss": 3.14191650390625,
      "step": 56650
    },
    {
      "epoch": 0.03338155604267246,
      "grad_norm": 0.2753174602985382,
      "learning_rate": 0.00029499285564157045,
      "loss": 3.0470587158203126,
      "step": 56700
    },
    {
      "epoch": 0.03341099304094642,
      "grad_norm": 0.3971617519855499,
      "learning_rate": 0.0002949884400924792,
      "loss": 3.1050164794921873,
      "step": 56750
    },
    {
      "epoch": 0.033440430039220384,
      "grad_norm": 0.35444754362106323,
      "learning_rate": 0.000294984024543388,
      "loss": 3.0601956176757814,
      "step": 56800
    },
    {
      "epoch": 0.03346986703749435,
      "grad_norm": 0.40862220525741577,
      "learning_rate": 0.00029497960899429684,
      "loss": 3.077057189941406,
      "step": 56850
    },
    {
      "epoch": 0.033499304035768306,
      "grad_norm": 0.34963157773017883,
      "learning_rate": 0.00029497519344520566,
      "loss": 3.069934997558594,
      "step": 56900
    },
    {
      "epoch": 0.03352874103404227,
      "grad_norm": 0.4192199110984802,
      "learning_rate": 0.00029497077789611447,
      "loss": 3.0822625732421876,
      "step": 56950
    },
    {
      "epoch": 0.033558178032316234,
      "grad_norm": 0.2705139219760895,
      "learning_rate": 0.0002949663623470233,
      "loss": 3.0902716064453126,
      "step": 57000
    },
    {
      "epoch": 0.03358761503059019,
      "grad_norm": 0.3715530037879944,
      "learning_rate": 0.0002949619467979321,
      "loss": 3.0380062866210937,
      "step": 57050
    },
    {
      "epoch": 0.033617052028864156,
      "grad_norm": 0.2839960753917694,
      "learning_rate": 0.0002949575312488409,
      "loss": 3.057029113769531,
      "step": 57100
    },
    {
      "epoch": 0.03364648902713811,
      "grad_norm": 0.35075050592422485,
      "learning_rate": 0.0002949531156997497,
      "loss": 3.07127685546875,
      "step": 57150
    },
    {
      "epoch": 0.03367592602541208,
      "grad_norm": 0.5025358200073242,
      "learning_rate": 0.0002949487001506585,
      "loss": 3.1131777954101563,
      "step": 57200
    },
    {
      "epoch": 0.03370536302368604,
      "grad_norm": 0.3246264159679413,
      "learning_rate": 0.0002949442846015673,
      "loss": 3.097637023925781,
      "step": 57250
    },
    {
      "epoch": 0.03373480002196,
      "grad_norm": 0.36858588457107544,
      "learning_rate": 0.0002949398690524761,
      "loss": 3.0945028686523437,
      "step": 57300
    },
    {
      "epoch": 0.03376423702023396,
      "grad_norm": 0.3442712128162384,
      "learning_rate": 0.00029493545350338493,
      "loss": 3.10846923828125,
      "step": 57350
    },
    {
      "epoch": 0.03379367401850793,
      "grad_norm": 0.388367623090744,
      "learning_rate": 0.00029493103795429375,
      "loss": 3.027058410644531,
      "step": 57400
    },
    {
      "epoch": 0.033823111016781884,
      "grad_norm": 0.26670828461647034,
      "learning_rate": 0.00029492662240520256,
      "loss": 3.11347900390625,
      "step": 57450
    },
    {
      "epoch": 0.03385254801505585,
      "grad_norm": 0.3216807246208191,
      "learning_rate": 0.0002949222068561114,
      "loss": 3.048636779785156,
      "step": 57500
    },
    {
      "epoch": 0.033881985013329806,
      "grad_norm": 0.29877379536628723,
      "learning_rate": 0.00029491779130702014,
      "loss": 3.019237060546875,
      "step": 57550
    },
    {
      "epoch": 0.03391142201160377,
      "grad_norm": 0.3653230369091034,
      "learning_rate": 0.00029491337575792895,
      "loss": 3.102675476074219,
      "step": 57600
    },
    {
      "epoch": 0.033940859009877734,
      "grad_norm": 0.25685006380081177,
      "learning_rate": 0.0002949089602088378,
      "loss": 3.102961120605469,
      "step": 57650
    },
    {
      "epoch": 0.03397029600815169,
      "grad_norm": 0.3328694999217987,
      "learning_rate": 0.0002949045446597466,
      "loss": 3.099528503417969,
      "step": 57700
    },
    {
      "epoch": 0.033999733006425656,
      "grad_norm": 0.38697561621665955,
      "learning_rate": 0.0002949001291106554,
      "loss": 3.0279901123046873,
      "step": 57750
    },
    {
      "epoch": 0.03402917000469962,
      "grad_norm": 0.3540589511394501,
      "learning_rate": 0.0002948957135615642,
      "loss": 3.0438311767578123,
      "step": 57800
    },
    {
      "epoch": 0.03405860700297358,
      "grad_norm": 0.3198706805706024,
      "learning_rate": 0.000294891298012473,
      "loss": 3.0768682861328127,
      "step": 57850
    },
    {
      "epoch": 0.03408804400124754,
      "grad_norm": 0.38274383544921875,
      "learning_rate": 0.00029488688246338184,
      "loss": 3.0884027099609375,
      "step": 57900
    },
    {
      "epoch": 0.0341174809995215,
      "grad_norm": 0.32881781458854675,
      "learning_rate": 0.00029488246691429065,
      "loss": 3.0998153686523438,
      "step": 57950
    },
    {
      "epoch": 0.03414691799779546,
      "grad_norm": 0.31274962425231934,
      "learning_rate": 0.00029487805136519947,
      "loss": 3.074457702636719,
      "step": 58000
    },
    {
      "epoch": 0.03417635499606943,
      "grad_norm": 0.3402518332004547,
      "learning_rate": 0.0002948736358161083,
      "loss": 3.060108642578125,
      "step": 58050
    },
    {
      "epoch": 0.034205791994343385,
      "grad_norm": 0.4046509861946106,
      "learning_rate": 0.00029486922026701704,
      "loss": 3.0668710327148436,
      "step": 58100
    },
    {
      "epoch": 0.03423522899261735,
      "grad_norm": 0.3534073829650879,
      "learning_rate": 0.00029486480471792585,
      "loss": 3.0704327392578126,
      "step": 58150
    },
    {
      "epoch": 0.034264665990891306,
      "grad_norm": 0.5283153057098389,
      "learning_rate": 0.00029486038916883467,
      "loss": 3.0181573486328124,
      "step": 58200
    },
    {
      "epoch": 0.03429410298916527,
      "grad_norm": 0.2504599690437317,
      "learning_rate": 0.0002948559736197435,
      "loss": 3.0921771240234377,
      "step": 58250
    },
    {
      "epoch": 0.034323539987439235,
      "grad_norm": 0.3494044542312622,
      "learning_rate": 0.0002948515580706523,
      "loss": 3.0527835083007813,
      "step": 58300
    },
    {
      "epoch": 0.03435297698571319,
      "grad_norm": 0.3673362731933594,
      "learning_rate": 0.0002948471425215611,
      "loss": 3.01358154296875,
      "step": 58350
    },
    {
      "epoch": 0.034382413983987156,
      "grad_norm": 0.2928312420845032,
      "learning_rate": 0.00029484272697246993,
      "loss": 3.07281982421875,
      "step": 58400
    },
    {
      "epoch": 0.03441185098226112,
      "grad_norm": 0.4198503792285919,
      "learning_rate": 0.0002948383114233787,
      "loss": 3.0577493286132813,
      "step": 58450
    },
    {
      "epoch": 0.03444128798053508,
      "grad_norm": 0.3403865396976471,
      "learning_rate": 0.0002948338958742875,
      "loss": 3.0801983642578126,
      "step": 58500
    },
    {
      "epoch": 0.03447072497880904,
      "grad_norm": 0.4094002842903137,
      "learning_rate": 0.0002948294803251963,
      "loss": 3.0854925537109374,
      "step": 58550
    },
    {
      "epoch": 0.034500161977083,
      "grad_norm": 0.3234129846096039,
      "learning_rate": 0.00029482506477610513,
      "loss": 3.04945068359375,
      "step": 58600
    },
    {
      "epoch": 0.034529598975356963,
      "grad_norm": 0.3773318827152252,
      "learning_rate": 0.00029482064922701395,
      "loss": 3.03967041015625,
      "step": 58650
    },
    {
      "epoch": 0.03455903597363093,
      "grad_norm": 0.29308196902275085,
      "learning_rate": 0.00029481623367792276,
      "loss": 3.021727294921875,
      "step": 58700
    },
    {
      "epoch": 0.034588472971904885,
      "grad_norm": 0.2855548858642578,
      "learning_rate": 0.0002948118181288316,
      "loss": 3.033852844238281,
      "step": 58750
    },
    {
      "epoch": 0.03461790997017885,
      "grad_norm": 0.295021653175354,
      "learning_rate": 0.0002948074025797404,
      "loss": 3.05652587890625,
      "step": 58800
    },
    {
      "epoch": 0.034647346968452813,
      "grad_norm": 0.3486831784248352,
      "learning_rate": 0.00029480298703064915,
      "loss": 3.086178894042969,
      "step": 58850
    },
    {
      "epoch": 0.03467678396672677,
      "grad_norm": 0.28229820728302,
      "learning_rate": 0.00029479857148155796,
      "loss": 3.0288433837890625,
      "step": 58900
    },
    {
      "epoch": 0.034706220965000735,
      "grad_norm": 0.35058683156967163,
      "learning_rate": 0.00029479415593246683,
      "loss": 3.074605712890625,
      "step": 58950
    },
    {
      "epoch": 0.03473565796327469,
      "grad_norm": 0.2695329785346985,
      "learning_rate": 0.0002947897403833756,
      "loss": 3.085394287109375,
      "step": 59000
    },
    {
      "epoch": 0.034765094961548657,
      "grad_norm": 0.404562771320343,
      "learning_rate": 0.0002947853248342844,
      "loss": 3.0388623046875,
      "step": 59050
    },
    {
      "epoch": 0.03479453195982262,
      "grad_norm": 0.33721381425857544,
      "learning_rate": 0.0002947809092851932,
      "loss": 3.0933596801757814,
      "step": 59100
    },
    {
      "epoch": 0.03482396895809658,
      "grad_norm": 0.3039143681526184,
      "learning_rate": 0.00029477649373610204,
      "loss": 3.0205743408203123,
      "step": 59150
    },
    {
      "epoch": 0.03485340595637054,
      "grad_norm": 0.36498385667800903,
      "learning_rate": 0.00029477207818701085,
      "loss": 3.0723672485351563,
      "step": 59200
    },
    {
      "epoch": 0.034882842954644507,
      "grad_norm": 0.3613220155239105,
      "learning_rate": 0.00029476766263791967,
      "loss": 3.0966119384765625,
      "step": 59250
    },
    {
      "epoch": 0.034912279952918464,
      "grad_norm": 0.38935908675193787,
      "learning_rate": 0.0002947632470888285,
      "loss": 3.0735159301757813,
      "step": 59300
    },
    {
      "epoch": 0.03494171695119243,
      "grad_norm": 0.383461058139801,
      "learning_rate": 0.0002947588315397373,
      "loss": 3.073104248046875,
      "step": 59350
    },
    {
      "epoch": 0.034971153949466385,
      "grad_norm": 0.2955280840396881,
      "learning_rate": 0.00029475441599064605,
      "loss": 3.0789450073242186,
      "step": 59400
    },
    {
      "epoch": 0.03500059094774035,
      "grad_norm": 0.44711557030677795,
      "learning_rate": 0.00029475000044155487,
      "loss": 3.026625671386719,
      "step": 59450
    },
    {
      "epoch": 0.035030027946014314,
      "grad_norm": 0.29472845792770386,
      "learning_rate": 0.0002947455848924637,
      "loss": 3.0029876708984373,
      "step": 59500
    },
    {
      "epoch": 0.03505946494428827,
      "grad_norm": 0.28101933002471924,
      "learning_rate": 0.0002947411693433725,
      "loss": 3.0282656860351564,
      "step": 59550
    },
    {
      "epoch": 0.035088901942562235,
      "grad_norm": 0.375434011220932,
      "learning_rate": 0.0002947367537942813,
      "loss": 3.0418603515625,
      "step": 59600
    },
    {
      "epoch": 0.0351183389408362,
      "grad_norm": 0.3653438687324524,
      "learning_rate": 0.0002947323382451901,
      "loss": 3.0639663696289063,
      "step": 59650
    },
    {
      "epoch": 0.03514777593911016,
      "grad_norm": 0.31800317764282227,
      "learning_rate": 0.00029472792269609894,
      "loss": 3.0432501220703125,
      "step": 59700
    },
    {
      "epoch": 0.03517721293738412,
      "grad_norm": 0.2876017689704895,
      "learning_rate": 0.00029472350714700776,
      "loss": 3.018620910644531,
      "step": 59750
    },
    {
      "epoch": 0.03520664993565808,
      "grad_norm": 0.30855095386505127,
      "learning_rate": 0.0002947190915979165,
      "loss": 3.0691790771484375,
      "step": 59800
    },
    {
      "epoch": 0.03523608693393204,
      "grad_norm": 0.2884214222431183,
      "learning_rate": 0.00029471467604882533,
      "loss": 3.077706604003906,
      "step": 59850
    },
    {
      "epoch": 0.03526552393220601,
      "grad_norm": 0.2842424809932709,
      "learning_rate": 0.0002947102604997342,
      "loss": 3.0838006591796874,
      "step": 59900
    },
    {
      "epoch": 0.035294960930479964,
      "grad_norm": 0.3044009804725647,
      "learning_rate": 0.00029470584495064296,
      "loss": 3.0568218994140626,
      "step": 59950
    },
    {
      "epoch": 0.03532439792875393,
      "grad_norm": Infinity,
      "learning_rate": 0.0002947014294015518,
      "loss": 3.046533508300781,
      "step": 60000
    },
    {
      "epoch": 0.035353834927027886,
      "grad_norm": 0.2357446402311325,
      "learning_rate": 0.0002946970138524606,
      "loss": 3.086975402832031,
      "step": 60050
    },
    {
      "epoch": 0.03538327192530185,
      "grad_norm": 0.33404454588890076,
      "learning_rate": 0.0002946925983033694,
      "loss": 3.095814208984375,
      "step": 60100
    },
    {
      "epoch": 0.035412708923575814,
      "grad_norm": 0.2626489996910095,
      "learning_rate": 0.00029468818275427816,
      "loss": 3.048420104980469,
      "step": 60150
    },
    {
      "epoch": 0.03544214592184977,
      "grad_norm": 0.3368547856807709,
      "learning_rate": 0.000294683767205187,
      "loss": 3.046090393066406,
      "step": 60200
    },
    {
      "epoch": 0.035471582920123736,
      "grad_norm": 0.290681391954422,
      "learning_rate": 0.00029467935165609585,
      "loss": 3.006519470214844,
      "step": 60250
    },
    {
      "epoch": 0.0355010199183977,
      "grad_norm": 0.257733553647995,
      "learning_rate": 0.0002946749361070046,
      "loss": 3.0694674682617187,
      "step": 60300
    },
    {
      "epoch": 0.03553045691667166,
      "grad_norm": 0.3515625596046448,
      "learning_rate": 0.0002946705205579134,
      "loss": 3.050360107421875,
      "step": 60350
    },
    {
      "epoch": 0.03555989391494562,
      "grad_norm": 0.2911955714225769,
      "learning_rate": 0.00029466610500882224,
      "loss": 3.02897705078125,
      "step": 60400
    },
    {
      "epoch": 0.03558933091321958,
      "grad_norm": 0.3891511857509613,
      "learning_rate": 0.00029466168945973105,
      "loss": 3.1205538940429687,
      "step": 60450
    },
    {
      "epoch": 0.03561876791149354,
      "grad_norm": 0.3123391568660736,
      "learning_rate": 0.00029465727391063986,
      "loss": 3.0255856323242187,
      "step": 60500
    },
    {
      "epoch": 0.03564820490976751,
      "grad_norm": 0.335470587015152,
      "learning_rate": 0.0002946528583615487,
      "loss": 3.040260925292969,
      "step": 60550
    },
    {
      "epoch": 0.035677641908041464,
      "grad_norm": 0.3560394048690796,
      "learning_rate": 0.0002946484428124575,
      "loss": 2.980829772949219,
      "step": 60600
    },
    {
      "epoch": 0.03570707890631543,
      "grad_norm": 0.2572697699069977,
      "learning_rate": 0.0002946440272633663,
      "loss": 3.1025067138671876,
      "step": 60650
    },
    {
      "epoch": 0.03573651590458939,
      "grad_norm": 0.33871906995773315,
      "learning_rate": 0.00029463961171427507,
      "loss": 3.080113525390625,
      "step": 60700
    },
    {
      "epoch": 0.03576595290286335,
      "grad_norm": 0.37225279211997986,
      "learning_rate": 0.0002946351961651839,
      "loss": 3.0423370361328126,
      "step": 60750
    },
    {
      "epoch": 0.035795389901137314,
      "grad_norm": 0.260368287563324,
      "learning_rate": 0.0002946307806160927,
      "loss": 3.052308349609375,
      "step": 60800
    },
    {
      "epoch": 0.03582482689941127,
      "grad_norm": 0.3747002184391022,
      "learning_rate": 0.0002946263650670015,
      "loss": 3.055184326171875,
      "step": 60850
    },
    {
      "epoch": 0.035854263897685236,
      "grad_norm": 0.3364167809486389,
      "learning_rate": 0.0002946219495179103,
      "loss": 3.03467529296875,
      "step": 60900
    },
    {
      "epoch": 0.0358837008959592,
      "grad_norm": 0.42385631799697876,
      "learning_rate": 0.00029461753396881914,
      "loss": 3.0529885864257813,
      "step": 60950
    },
    {
      "epoch": 0.03591313789423316,
      "grad_norm": 0.24618953466415405,
      "learning_rate": 0.00029461311841972796,
      "loss": 3.0083935546875,
      "step": 61000
    },
    {
      "epoch": 0.03594257489250712,
      "grad_norm": 0.28359711170196533,
      "learning_rate": 0.00029460870287063677,
      "loss": 3.0347039794921873,
      "step": 61050
    },
    {
      "epoch": 0.035972011890781086,
      "grad_norm": 0.45420578122138977,
      "learning_rate": 0.00029460428732154553,
      "loss": 3.028121032714844,
      "step": 61100
    },
    {
      "epoch": 0.03600144888905504,
      "grad_norm": 0.3112683892250061,
      "learning_rate": 0.00029459987177245434,
      "loss": 3.041036376953125,
      "step": 61150
    },
    {
      "epoch": 0.03603088588732901,
      "grad_norm": 0.2412087321281433,
      "learning_rate": 0.00029459545622336316,
      "loss": 3.0012783813476562,
      "step": 61200
    },
    {
      "epoch": 0.036060322885602965,
      "grad_norm": 0.29504695534706116,
      "learning_rate": 0.000294591040674272,
      "loss": 3.0526199340820312,
      "step": 61250
    },
    {
      "epoch": 0.03608975988387693,
      "grad_norm": 0.3261473774909973,
      "learning_rate": 0.0002945866251251808,
      "loss": 3.0999325561523436,
      "step": 61300
    },
    {
      "epoch": 0.03611919688215089,
      "grad_norm": 0.3132563531398773,
      "learning_rate": 0.0002945822095760896,
      "loss": 3.051993103027344,
      "step": 61350
    },
    {
      "epoch": 0.03614863388042485,
      "grad_norm": 0.3555334508419037,
      "learning_rate": 0.0002945777940269984,
      "loss": 3.014565124511719,
      "step": 61400
    },
    {
      "epoch": 0.036178070878698815,
      "grad_norm": 0.306857705116272,
      "learning_rate": 0.00029457337847790723,
      "loss": 3.001575622558594,
      "step": 61450
    },
    {
      "epoch": 0.03620750787697278,
      "grad_norm": 0.34280824661254883,
      "learning_rate": 0.000294568962928816,
      "loss": 3.0632919311523437,
      "step": 61500
    },
    {
      "epoch": 0.036236944875246736,
      "grad_norm": 0.3671588897705078,
      "learning_rate": 0.00029456454737972486,
      "loss": 3.113761291503906,
      "step": 61550
    },
    {
      "epoch": 0.0362663818735207,
      "grad_norm": 0.2789926528930664,
      "learning_rate": 0.0002945601318306337,
      "loss": 3.103636474609375,
      "step": 61600
    },
    {
      "epoch": 0.03629581887179466,
      "grad_norm": 0.467281311750412,
      "learning_rate": 0.00029455571628154244,
      "loss": 3.058133239746094,
      "step": 61650
    },
    {
      "epoch": 0.03632525587006862,
      "grad_norm": 0.2629607319831848,
      "learning_rate": 0.00029455130073245125,
      "loss": 3.069577941894531,
      "step": 61700
    },
    {
      "epoch": 0.036354692868342586,
      "grad_norm": 0.2776906192302704,
      "learning_rate": 0.00029454688518336006,
      "loss": 3.0603103637695312,
      "step": 61750
    },
    {
      "epoch": 0.036384129866616544,
      "grad_norm": 0.30707359313964844,
      "learning_rate": 0.0002945424696342689,
      "loss": 3.0301858520507814,
      "step": 61800
    },
    {
      "epoch": 0.03641356686489051,
      "grad_norm": 0.29219159483909607,
      "learning_rate": 0.0002945380540851777,
      "loss": 2.997283935546875,
      "step": 61850
    },
    {
      "epoch": 0.036443003863164465,
      "grad_norm": 0.33146852254867554,
      "learning_rate": 0.0002945336385360865,
      "loss": 3.110585021972656,
      "step": 61900
    },
    {
      "epoch": 0.03647244086143843,
      "grad_norm": 0.3259505331516266,
      "learning_rate": 0.0002945292229869953,
      "loss": 3.1217596435546877,
      "step": 61950
    },
    {
      "epoch": 0.036501877859712394,
      "grad_norm": 0.28890669345855713,
      "learning_rate": 0.0002945248074379041,
      "loss": 3.0573846435546876,
      "step": 62000
    },
    {
      "epoch": 0.03653131485798635,
      "grad_norm": 0.2842748165130615,
      "learning_rate": 0.0002945203918888129,
      "loss": 3.0481808471679686,
      "step": 62050
    },
    {
      "epoch": 0.036560751856260315,
      "grad_norm": 0.34645748138427734,
      "learning_rate": 0.0002945159763397217,
      "loss": 3.058802185058594,
      "step": 62100
    },
    {
      "epoch": 0.03659018885453428,
      "grad_norm": 0.30232590436935425,
      "learning_rate": 0.0002945115607906305,
      "loss": 3.021231689453125,
      "step": 62150
    },
    {
      "epoch": 0.03661962585280824,
      "grad_norm": 0.35088667273521423,
      "learning_rate": 0.00029450714524153934,
      "loss": 3.0559786987304687,
      "step": 62200
    },
    {
      "epoch": 0.0366490628510822,
      "grad_norm": 0.30839407444000244,
      "learning_rate": 0.00029450272969244815,
      "loss": 3.0537234497070314,
      "step": 62250
    },
    {
      "epoch": 0.03667849984935616,
      "grad_norm": 0.3233110308647156,
      "learning_rate": 0.00029449831414335697,
      "loss": 3.05907470703125,
      "step": 62300
    },
    {
      "epoch": 0.03670793684763012,
      "grad_norm": 0.3925902843475342,
      "learning_rate": 0.0002944938985942658,
      "loss": 3.0786297607421873,
      "step": 62350
    },
    {
      "epoch": 0.03673737384590409,
      "grad_norm": 0.4088381826877594,
      "learning_rate": 0.00029448948304517454,
      "loss": 3.0704721069335936,
      "step": 62400
    },
    {
      "epoch": 0.036766810844178044,
      "grad_norm": 0.2620042860507965,
      "learning_rate": 0.00029448506749608336,
      "loss": 3.063638000488281,
      "step": 62450
    },
    {
      "epoch": 0.03679624784245201,
      "grad_norm": 0.3871116638183594,
      "learning_rate": 0.0002944806519469922,
      "loss": 3.0529608154296874,
      "step": 62500
    },
    {
      "epoch": 0.03682568484072597,
      "grad_norm": 0.31770750880241394,
      "learning_rate": 0.000294476236397901,
      "loss": 3.0631201171875,
      "step": 62550
    },
    {
      "epoch": 0.03685512183899993,
      "grad_norm": 0.3237428367137909,
      "learning_rate": 0.0002944718208488098,
      "loss": 3.009324951171875,
      "step": 62600
    },
    {
      "epoch": 0.036884558837273894,
      "grad_norm": 0.48745372891426086,
      "learning_rate": 0.0002944674052997186,
      "loss": 3.0388848876953123,
      "step": 62650
    },
    {
      "epoch": 0.03691399583554785,
      "grad_norm": 0.3329834043979645,
      "learning_rate": 0.00029446298975062743,
      "loss": 3.070648193359375,
      "step": 62700
    },
    {
      "epoch": 0.036943432833821815,
      "grad_norm": 0.32407474517822266,
      "learning_rate": 0.00029445857420153625,
      "loss": 3.0938461303710936,
      "step": 62750
    },
    {
      "epoch": 0.03697286983209578,
      "grad_norm": 0.4106646478176117,
      "learning_rate": 0.000294454158652445,
      "loss": 3.0300064086914062,
      "step": 62800
    },
    {
      "epoch": 0.03700230683036974,
      "grad_norm": 0.3037460744380951,
      "learning_rate": 0.0002944497431033539,
      "loss": 3.050010986328125,
      "step": 62850
    },
    {
      "epoch": 0.0370317438286437,
      "grad_norm": 0.46578168869018555,
      "learning_rate": 0.0002944453275542627,
      "loss": 3.066632080078125,
      "step": 62900
    },
    {
      "epoch": 0.037061180826917665,
      "grad_norm": 0.2975713312625885,
      "learning_rate": 0.00029444091200517145,
      "loss": 3.034165954589844,
      "step": 62950
    },
    {
      "epoch": 0.03709061782519162,
      "grad_norm": 0.2725217342376709,
      "learning_rate": 0.00029443649645608026,
      "loss": 3.0418603515625,
      "step": 63000
    },
    {
      "epoch": 0.03712005482346559,
      "grad_norm": 0.2728710174560547,
      "learning_rate": 0.0002944320809069891,
      "loss": 3.0500003051757814,
      "step": 63050
    },
    {
      "epoch": 0.037149491821739544,
      "grad_norm": 0.3130640983581543,
      "learning_rate": 0.0002944276653578979,
      "loss": 3.013915710449219,
      "step": 63100
    },
    {
      "epoch": 0.03717892882001351,
      "grad_norm": 0.3156431317329407,
      "learning_rate": 0.0002944232498088067,
      "loss": 3.0871847534179686,
      "step": 63150
    },
    {
      "epoch": 0.03720836581828747,
      "grad_norm": 0.3096354603767395,
      "learning_rate": 0.0002944188342597155,
      "loss": 3.0840545654296876,
      "step": 63200
    },
    {
      "epoch": 0.03723780281656143,
      "grad_norm": 0.37459954619407654,
      "learning_rate": 0.00029441441871062434,
      "loss": 3.0785122680664063,
      "step": 63250
    },
    {
      "epoch": 0.037267239814835394,
      "grad_norm": 0.2731577455997467,
      "learning_rate": 0.00029441000316153315,
      "loss": 3.0310333251953123,
      "step": 63300
    },
    {
      "epoch": 0.03729667681310936,
      "grad_norm": 0.3533281087875366,
      "learning_rate": 0.0002944055876124419,
      "loss": 3.0408489990234373,
      "step": 63350
    },
    {
      "epoch": 0.037326113811383316,
      "grad_norm": 0.3294720947742462,
      "learning_rate": 0.0002944011720633507,
      "loss": 3.083432922363281,
      "step": 63400
    },
    {
      "epoch": 0.03735555080965728,
      "grad_norm": 0.3637729585170746,
      "learning_rate": 0.00029439675651425954,
      "loss": 3.0745114135742186,
      "step": 63450
    },
    {
      "epoch": 0.03738498780793124,
      "grad_norm": 0.3027518093585968,
      "learning_rate": 0.00029439234096516835,
      "loss": 3.05487548828125,
      "step": 63500
    },
    {
      "epoch": 0.0374144248062052,
      "grad_norm": 0.32498565316200256,
      "learning_rate": 0.00029438792541607717,
      "loss": 3.0107540893554687,
      "step": 63550
    },
    {
      "epoch": 0.037443861804479166,
      "grad_norm": 0.26284077763557434,
      "learning_rate": 0.000294383509866986,
      "loss": 3.0528436279296876,
      "step": 63600
    },
    {
      "epoch": 0.03747329880275312,
      "grad_norm": 0.28458505868911743,
      "learning_rate": 0.0002943790943178948,
      "loss": 3.071527404785156,
      "step": 63650
    },
    {
      "epoch": 0.03750273580102709,
      "grad_norm": 0.3783884048461914,
      "learning_rate": 0.00029437467876880356,
      "loss": 3.046570129394531,
      "step": 63700
    },
    {
      "epoch": 0.037532172799301045,
      "grad_norm": 0.29067668318748474,
      "learning_rate": 0.00029437026321971237,
      "loss": 3.0368106079101564,
      "step": 63750
    },
    {
      "epoch": 0.03756160979757501,
      "grad_norm": 0.33438435196876526,
      "learning_rate": 0.0002943658476706212,
      "loss": 3.0735626220703125,
      "step": 63800
    },
    {
      "epoch": 0.03759104679584897,
      "grad_norm": 0.48806455731391907,
      "learning_rate": 0.00029436143212153,
      "loss": 3.035395202636719,
      "step": 63850
    },
    {
      "epoch": 0.03762048379412293,
      "grad_norm": 0.318143367767334,
      "learning_rate": 0.0002943570165724388,
      "loss": 3.0608349609375,
      "step": 63900
    },
    {
      "epoch": 0.037649920792396895,
      "grad_norm": 0.28331753611564636,
      "learning_rate": 0.00029435260102334763,
      "loss": 3.016904602050781,
      "step": 63950
    },
    {
      "epoch": 0.03767935779067086,
      "grad_norm": 0.3213129937648773,
      "learning_rate": 0.00029434818547425644,
      "loss": 3.0668267822265625,
      "step": 64000
    },
    {
      "epoch": 0.037708794788944816,
      "grad_norm": 0.31765884160995483,
      "learning_rate": 0.00029434376992516526,
      "loss": 3.000462951660156,
      "step": 64050
    },
    {
      "epoch": 0.03773823178721878,
      "grad_norm": 0.4274866580963135,
      "learning_rate": 0.000294339354376074,
      "loss": 3.0121688842773438,
      "step": 64100
    },
    {
      "epoch": 0.03776766878549274,
      "grad_norm": 0.2939354181289673,
      "learning_rate": 0.0002943349388269829,
      "loss": 3.0613363647460936,
      "step": 64150
    },
    {
      "epoch": 0.0377971057837667,
      "grad_norm": 0.2589299976825714,
      "learning_rate": 0.0002943305232778917,
      "loss": 3.0539260864257813,
      "step": 64200
    },
    {
      "epoch": 0.037826542782040666,
      "grad_norm": 0.23596824705600739,
      "learning_rate": 0.00029432610772880046,
      "loss": 3.0562002563476565,
      "step": 64250
    },
    {
      "epoch": 0.03785597978031462,
      "grad_norm": 0.37842506170272827,
      "learning_rate": 0.0002943216921797093,
      "loss": 3.0709542846679687,
      "step": 64300
    },
    {
      "epoch": 0.03788541677858859,
      "grad_norm": 0.3569968640804291,
      "learning_rate": 0.0002943172766306181,
      "loss": 3.0342254638671875,
      "step": 64350
    },
    {
      "epoch": 0.03791485377686255,
      "grad_norm": 0.35310009121894836,
      "learning_rate": 0.0002943128610815269,
      "loss": 3.0795965576171875,
      "step": 64400
    },
    {
      "epoch": 0.03794429077513651,
      "grad_norm": 0.33678919076919556,
      "learning_rate": 0.0002943084455324357,
      "loss": 3.02883056640625,
      "step": 64450
    },
    {
      "epoch": 0.03797372777341047,
      "grad_norm": 0.4190380871295929,
      "learning_rate": 0.00029430402998334454,
      "loss": 3.07998779296875,
      "step": 64500
    },
    {
      "epoch": 0.03800316477168443,
      "grad_norm": 0.3393576145172119,
      "learning_rate": 0.00029429961443425335,
      "loss": 3.037882080078125,
      "step": 64550
    },
    {
      "epoch": 0.038032601769958395,
      "grad_norm": 0.3639167249202728,
      "learning_rate": 0.00029429519888516216,
      "loss": 3.0581094360351564,
      "step": 64600
    },
    {
      "epoch": 0.03806203876823236,
      "grad_norm": 0.23590761423110962,
      "learning_rate": 0.0002942907833360709,
      "loss": 3.0108578491210936,
      "step": 64650
    },
    {
      "epoch": 0.038091475766506316,
      "grad_norm": 0.69399493932724,
      "learning_rate": 0.00029428636778697974,
      "loss": 3.0224224853515627,
      "step": 64700
    },
    {
      "epoch": 0.03812091276478028,
      "grad_norm": 0.28972429037094116,
      "learning_rate": 0.00029428195223788855,
      "loss": 3.0857968139648437,
      "step": 64750
    },
    {
      "epoch": 0.038150349763054245,
      "grad_norm": 0.2526688575744629,
      "learning_rate": 0.00029427753668879737,
      "loss": 3.044835205078125,
      "step": 64800
    },
    {
      "epoch": 0.0381797867613282,
      "grad_norm": 0.3087608218193054,
      "learning_rate": 0.0002942731211397062,
      "loss": 3.0170101928710937,
      "step": 64850
    },
    {
      "epoch": 0.038209223759602166,
      "grad_norm": 0.4164792597293854,
      "learning_rate": 0.000294268705590615,
      "loss": 3.0133001708984377,
      "step": 64900
    },
    {
      "epoch": 0.038238660757876124,
      "grad_norm": 0.30520305037498474,
      "learning_rate": 0.0002942642900415238,
      "loss": 3.0321173095703124,
      "step": 64950
    },
    {
      "epoch": 0.03826809775615009,
      "grad_norm": 0.39662668108940125,
      "learning_rate": 0.0002942598744924326,
      "loss": 3.01680908203125,
      "step": 65000
    },
    {
      "epoch": 0.03829753475442405,
      "grad_norm": 0.349808931350708,
      "learning_rate": 0.0002942554589433414,
      "loss": 3.042023010253906,
      "step": 65050
    },
    {
      "epoch": 0.03832697175269801,
      "grad_norm": 0.40234968066215515,
      "learning_rate": 0.0002942510433942502,
      "loss": 3.010561828613281,
      "step": 65100
    },
    {
      "epoch": 0.038356408750971974,
      "grad_norm": 0.3049967885017395,
      "learning_rate": 0.00029424662784515907,
      "loss": 3.034196472167969,
      "step": 65150
    },
    {
      "epoch": 0.03838584574924594,
      "grad_norm": 0.46047109365463257,
      "learning_rate": 0.00029424221229606783,
      "loss": 3.0274188232421877,
      "step": 65200
    },
    {
      "epoch": 0.038415282747519895,
      "grad_norm": 0.3330150544643402,
      "learning_rate": 0.00029423779674697664,
      "loss": 3.051194763183594,
      "step": 65250
    },
    {
      "epoch": 0.03844471974579386,
      "grad_norm": 0.31014955043792725,
      "learning_rate": 0.00029423338119788546,
      "loss": 3.0657647705078124,
      "step": 65300
    },
    {
      "epoch": 0.03847415674406782,
      "grad_norm": 0.392870157957077,
      "learning_rate": 0.0002942289656487943,
      "loss": 3.0005813598632813,
      "step": 65350
    },
    {
      "epoch": 0.03850359374234178,
      "grad_norm": 0.37230828404426575,
      "learning_rate": 0.00029422455009970303,
      "loss": 3.016052551269531,
      "step": 65400
    },
    {
      "epoch": 0.038533030740615745,
      "grad_norm": 0.3515506386756897,
      "learning_rate": 0.0002942201345506119,
      "loss": 2.9947799682617187,
      "step": 65450
    },
    {
      "epoch": 0.0385624677388897,
      "grad_norm": 0.26900243759155273,
      "learning_rate": 0.0002942157190015207,
      "loss": 3.0141787719726563,
      "step": 65500
    },
    {
      "epoch": 0.03859190473716367,
      "grad_norm": 0.32595980167388916,
      "learning_rate": 0.0002942113034524295,
      "loss": 2.9813525390625,
      "step": 65550
    },
    {
      "epoch": 0.038621341735437624,
      "grad_norm": 0.29014793038368225,
      "learning_rate": 0.0002942068879033383,
      "loss": 3.0459579467773437,
      "step": 65600
    },
    {
      "epoch": 0.03865077873371159,
      "grad_norm": 0.3385797142982483,
      "learning_rate": 0.0002942024723542471,
      "loss": 3.038119201660156,
      "step": 65650
    },
    {
      "epoch": 0.03868021573198555,
      "grad_norm": 0.44001296162605286,
      "learning_rate": 0.0002941980568051559,
      "loss": 3.056650390625,
      "step": 65700
    },
    {
      "epoch": 0.03870965273025951,
      "grad_norm": 0.2928692102432251,
      "learning_rate": 0.00029419364125606474,
      "loss": 3.010270690917969,
      "step": 65750
    },
    {
      "epoch": 0.038739089728533474,
      "grad_norm": 0.31039708852767944,
      "learning_rate": 0.00029418922570697355,
      "loss": 3.02623046875,
      "step": 65800
    },
    {
      "epoch": 0.03876852672680744,
      "grad_norm": 0.33284130692481995,
      "learning_rate": 0.00029418481015788236,
      "loss": 3.030644836425781,
      "step": 65850
    },
    {
      "epoch": 0.038797963725081396,
      "grad_norm": 0.39879724383354187,
      "learning_rate": 0.0002941803946087912,
      "loss": 3.0352774047851563,
      "step": 65900
    },
    {
      "epoch": 0.03882740072335536,
      "grad_norm": 0.3582421541213989,
      "learning_rate": 0.00029417597905969994,
      "loss": 3.0637277221679686,
      "step": 65950
    },
    {
      "epoch": 0.03885683772162932,
      "grad_norm": 0.2480475753545761,
      "learning_rate": 0.00029417156351060875,
      "loss": 3.02767578125,
      "step": 66000
    },
    {
      "epoch": 0.03888627471990328,
      "grad_norm": 0.36179497838020325,
      "learning_rate": 0.00029416714796151757,
      "loss": 2.9851251220703126,
      "step": 66050
    },
    {
      "epoch": 0.038915711718177246,
      "grad_norm": 0.2971908450126648,
      "learning_rate": 0.0002941627324124264,
      "loss": 3.047032470703125,
      "step": 66100
    },
    {
      "epoch": 0.0389451487164512,
      "grad_norm": 0.40516605973243713,
      "learning_rate": 0.0002941583168633352,
      "loss": 3.0108404541015625,
      "step": 66150
    },
    {
      "epoch": 0.03897458571472517,
      "grad_norm": 0.34640687704086304,
      "learning_rate": 0.000294153901314244,
      "loss": 3.0348858642578125,
      "step": 66200
    },
    {
      "epoch": 0.03900402271299913,
      "grad_norm": 0.3448140323162079,
      "learning_rate": 0.0002941494857651528,
      "loss": 3.0594754028320312,
      "step": 66250
    },
    {
      "epoch": 0.03903345971127309,
      "grad_norm": 0.3571757674217224,
      "learning_rate": 0.00029414507021606164,
      "loss": 3.0059573364257814,
      "step": 66300
    },
    {
      "epoch": 0.03906289670954705,
      "grad_norm": 0.35792359709739685,
      "learning_rate": 0.0002941406546669704,
      "loss": 3.0074603271484377,
      "step": 66350
    },
    {
      "epoch": 0.03909233370782101,
      "grad_norm": 0.3045908808708191,
      "learning_rate": 0.0002941362391178792,
      "loss": 3.0191497802734375,
      "step": 66400
    },
    {
      "epoch": 0.039121770706094974,
      "grad_norm": 0.345866858959198,
      "learning_rate": 0.0002941318235687881,
      "loss": 3.0149526977539063,
      "step": 66450
    },
    {
      "epoch": 0.03915120770436894,
      "grad_norm": 0.25107473134994507,
      "learning_rate": 0.00029412740801969684,
      "loss": 3.0253082275390626,
      "step": 66500
    },
    {
      "epoch": 0.039180644702642896,
      "grad_norm": 0.2657244801521301,
      "learning_rate": 0.00029412299247060566,
      "loss": 3.0763674926757814,
      "step": 66550
    },
    {
      "epoch": 0.03921008170091686,
      "grad_norm": 0.3034942150115967,
      "learning_rate": 0.00029411857692151447,
      "loss": 3.0177606201171874,
      "step": 66600
    },
    {
      "epoch": 0.039239518699190824,
      "grad_norm": 0.27323389053344727,
      "learning_rate": 0.0002941141613724233,
      "loss": 2.9817886352539062,
      "step": 66650
    },
    {
      "epoch": 0.03926895569746478,
      "grad_norm": 0.402063250541687,
      "learning_rate": 0.0002941097458233321,
      "loss": 3.029250183105469,
      "step": 66700
    },
    {
      "epoch": 0.039298392695738746,
      "grad_norm": 0.2881503701210022,
      "learning_rate": 0.0002941053302742409,
      "loss": 2.97021484375,
      "step": 66750
    },
    {
      "epoch": 0.0393278296940127,
      "grad_norm": 0.2876952588558197,
      "learning_rate": 0.00029410091472514973,
      "loss": 3.00296142578125,
      "step": 66800
    },
    {
      "epoch": 0.03935726669228667,
      "grad_norm": 0.2826308608055115,
      "learning_rate": 0.00029409649917605855,
      "loss": 3.0248736572265624,
      "step": 66850
    },
    {
      "epoch": 0.03938670369056063,
      "grad_norm": 0.24396555125713348,
      "learning_rate": 0.0002940920836269673,
      "loss": 3.048720397949219,
      "step": 66900
    },
    {
      "epoch": 0.03941614068883459,
      "grad_norm": 0.28055161237716675,
      "learning_rate": 0.0002940876680778761,
      "loss": 3.003580322265625,
      "step": 66950
    },
    {
      "epoch": 0.03944557768710855,
      "grad_norm": 0.4163345396518707,
      "learning_rate": 0.00029408325252878493,
      "loss": 3.02700439453125,
      "step": 67000
    },
    {
      "epoch": 0.03947501468538252,
      "grad_norm": 0.30892303586006165,
      "learning_rate": 0.00029407883697969375,
      "loss": 3.015203552246094,
      "step": 67050
    },
    {
      "epoch": 0.039504451683656475,
      "grad_norm": 0.26241740584373474,
      "learning_rate": 0.00029407442143060256,
      "loss": 2.93031494140625,
      "step": 67100
    },
    {
      "epoch": 0.03953388868193044,
      "grad_norm": 0.30661648511886597,
      "learning_rate": 0.0002940700058815114,
      "loss": 3.0186141967773437,
      "step": 67150
    },
    {
      "epoch": 0.039563325680204396,
      "grad_norm": 0.34826111793518066,
      "learning_rate": 0.0002940655903324202,
      "loss": 3.05658203125,
      "step": 67200
    },
    {
      "epoch": 0.03959276267847836,
      "grad_norm": 0.33226343989372253,
      "learning_rate": 0.00029406117478332895,
      "loss": 3.016123046875,
      "step": 67250
    },
    {
      "epoch": 0.039622199676752325,
      "grad_norm": 0.31179970502853394,
      "learning_rate": 0.00029405675923423777,
      "loss": 3.0571356201171875,
      "step": 67300
    },
    {
      "epoch": 0.03965163667502628,
      "grad_norm": 0.3252367377281189,
      "learning_rate": 0.0002940523436851466,
      "loss": 3.076687927246094,
      "step": 67350
    },
    {
      "epoch": 0.039681073673300246,
      "grad_norm": 0.34948083758354187,
      "learning_rate": 0.0002940479281360554,
      "loss": 3.054732360839844,
      "step": 67400
    },
    {
      "epoch": 0.0397105106715742,
      "grad_norm": 0.3684517741203308,
      "learning_rate": 0.0002940435125869642,
      "loss": 2.9902249145507813,
      "step": 67450
    },
    {
      "epoch": 0.03973994766984817,
      "grad_norm": 0.3036763668060303,
      "learning_rate": 0.000294039097037873,
      "loss": 3.033783264160156,
      "step": 67500
    },
    {
      "epoch": 0.03976938466812213,
      "grad_norm": 0.34210205078125,
      "learning_rate": 0.00029403468148878184,
      "loss": 3.042591552734375,
      "step": 67550
    },
    {
      "epoch": 0.03979882166639609,
      "grad_norm": 0.26261332631111145,
      "learning_rate": 0.00029403026593969065,
      "loss": 2.9406729125976563,
      "step": 67600
    },
    {
      "epoch": 0.03982825866467005,
      "grad_norm": 0.31503063440322876,
      "learning_rate": 0.0002940258503905994,
      "loss": 3.066082763671875,
      "step": 67650
    },
    {
      "epoch": 0.03985769566294402,
      "grad_norm": 0.26078712940216064,
      "learning_rate": 0.00029402143484150823,
      "loss": 3.00061279296875,
      "step": 67700
    },
    {
      "epoch": 0.039887132661217975,
      "grad_norm": 0.2654525339603424,
      "learning_rate": 0.0002940170192924171,
      "loss": 3.0670480346679687,
      "step": 67750
    },
    {
      "epoch": 0.03991656965949194,
      "grad_norm": 0.3821527659893036,
      "learning_rate": 0.00029401260374332586,
      "loss": 3.0226284790039064,
      "step": 67800
    },
    {
      "epoch": 0.039946006657765896,
      "grad_norm": 0.34856703877449036,
      "learning_rate": 0.00029400818819423467,
      "loss": 3.0093756103515625,
      "step": 67850
    },
    {
      "epoch": 0.03997544365603986,
      "grad_norm": 0.3192737400531769,
      "learning_rate": 0.0002940037726451435,
      "loss": 3.03952392578125,
      "step": 67900
    },
    {
      "epoch": 0.040004880654313825,
      "grad_norm": 0.27032271027565,
      "learning_rate": 0.0002939993570960523,
      "loss": 3.0293316650390625,
      "step": 67950
    },
    {
      "epoch": 0.04003431765258778,
      "grad_norm": 0.2674335837364197,
      "learning_rate": 0.0002939949415469611,
      "loss": 2.9594692993164062,
      "step": 68000
    },
    {
      "epoch": 0.040063754650861746,
      "grad_norm": 0.2775218188762665,
      "learning_rate": 0.0002939905259978699,
      "loss": 3.08779052734375,
      "step": 68050
    },
    {
      "epoch": 0.04009319164913571,
      "grad_norm": 0.3415927290916443,
      "learning_rate": 0.00029398611044877874,
      "loss": 3.057966613769531,
      "step": 68100
    },
    {
      "epoch": 0.04012262864740967,
      "grad_norm": 0.3057398200035095,
      "learning_rate": 0.00029398169489968756,
      "loss": 3.0418792724609376,
      "step": 68150
    },
    {
      "epoch": 0.04015206564568363,
      "grad_norm": 0.44394925236701965,
      "learning_rate": 0.0002939772793505963,
      "loss": 3.02694091796875,
      "step": 68200
    },
    {
      "epoch": 0.04018150264395759,
      "grad_norm": 0.34526902437210083,
      "learning_rate": 0.00029397286380150513,
      "loss": 3.0145803833007814,
      "step": 68250
    },
    {
      "epoch": 0.040210939642231554,
      "grad_norm": 0.3161317706108093,
      "learning_rate": 0.00029396844825241395,
      "loss": 3.0417599487304687,
      "step": 68300
    },
    {
      "epoch": 0.04024037664050552,
      "grad_norm": 0.36215898394584656,
      "learning_rate": 0.00029396403270332276,
      "loss": 3.0069790649414063,
      "step": 68350
    },
    {
      "epoch": 0.040269813638779475,
      "grad_norm": 0.2891750931739807,
      "learning_rate": 0.0002939596171542316,
      "loss": 3.0575601196289064,
      "step": 68400
    },
    {
      "epoch": 0.04029925063705344,
      "grad_norm": 0.42896783351898193,
      "learning_rate": 0.0002939552016051404,
      "loss": 2.977552795410156,
      "step": 68450
    },
    {
      "epoch": 0.040328687635327404,
      "grad_norm": 0.2931700348854065,
      "learning_rate": 0.0002939507860560492,
      "loss": 2.970257568359375,
      "step": 68500
    },
    {
      "epoch": 0.04035812463360136,
      "grad_norm": 0.34657377004623413,
      "learning_rate": 0.000293946370506958,
      "loss": 3.0448989868164062,
      "step": 68550
    },
    {
      "epoch": 0.040387561631875325,
      "grad_norm": 0.4644417464733124,
      "learning_rate": 0.0002939419549578668,
      "loss": 3.02500244140625,
      "step": 68600
    },
    {
      "epoch": 0.04041699863014928,
      "grad_norm": 0.2656705379486084,
      "learning_rate": 0.0002939375394087756,
      "loss": 3.0128079223632813,
      "step": 68650
    },
    {
      "epoch": 0.04044643562842325,
      "grad_norm": 0.3767276108264923,
      "learning_rate": 0.0002939331238596844,
      "loss": 3.0267950439453126,
      "step": 68700
    },
    {
      "epoch": 0.04047587262669721,
      "grad_norm": 0.516778826713562,
      "learning_rate": 0.0002939287083105932,
      "loss": 3.0314593505859375,
      "step": 68750
    },
    {
      "epoch": 0.04050530962497117,
      "grad_norm": 0.4280773401260376,
      "learning_rate": 0.00029392429276150204,
      "loss": 3.054854736328125,
      "step": 68800
    },
    {
      "epoch": 0.04053474662324513,
      "grad_norm": 0.22806210815906525,
      "learning_rate": 0.00029391987721241085,
      "loss": 3.0361737060546874,
      "step": 68850
    },
    {
      "epoch": 0.0405641836215191,
      "grad_norm": 0.26806455850601196,
      "learning_rate": 0.00029391546166331967,
      "loss": 3.002540283203125,
      "step": 68900
    },
    {
      "epoch": 0.040593620619793054,
      "grad_norm": 0.24904555082321167,
      "learning_rate": 0.00029391104611422843,
      "loss": 3.02845458984375,
      "step": 68950
    },
    {
      "epoch": 0.04062305761806702,
      "grad_norm": 0.34411320090293884,
      "learning_rate": 0.00029390663056513724,
      "loss": 3.074720458984375,
      "step": 69000
    },
    {
      "epoch": 0.040652494616340976,
      "grad_norm": 0.28868189454078674,
      "learning_rate": 0.0002939022150160461,
      "loss": 2.99505859375,
      "step": 69050
    },
    {
      "epoch": 0.04068193161461494,
      "grad_norm": 0.2834276556968689,
      "learning_rate": 0.00029389779946695487,
      "loss": 2.987850341796875,
      "step": 69100
    },
    {
      "epoch": 0.040711368612888904,
      "grad_norm": 0.2683381140232086,
      "learning_rate": 0.0002938933839178637,
      "loss": 2.9917562866210936,
      "step": 69150
    },
    {
      "epoch": 0.04074080561116286,
      "grad_norm": 0.39286020398139954,
      "learning_rate": 0.0002938889683687725,
      "loss": 3.03931884765625,
      "step": 69200
    },
    {
      "epoch": 0.040770242609436826,
      "grad_norm": 0.3495613932609558,
      "learning_rate": 0.0002938845528196813,
      "loss": 3.024252624511719,
      "step": 69250
    },
    {
      "epoch": 0.04079967960771078,
      "grad_norm": 0.32946065068244934,
      "learning_rate": 0.00029388013727059013,
      "loss": 3.013070373535156,
      "step": 69300
    },
    {
      "epoch": 0.04082911660598475,
      "grad_norm": 0.28699788451194763,
      "learning_rate": 0.0002938757217214989,
      "loss": 3.0064569091796876,
      "step": 69350
    },
    {
      "epoch": 0.04085855360425871,
      "grad_norm": 0.23961147665977478,
      "learning_rate": 0.00029387130617240776,
      "loss": 3.0161456298828124,
      "step": 69400
    },
    {
      "epoch": 0.04088799060253267,
      "grad_norm": 0.3930825889110565,
      "learning_rate": 0.0002938668906233166,
      "loss": 3.00197265625,
      "step": 69450
    },
    {
      "epoch": 0.04091742760080663,
      "grad_norm": 0.33156880736351013,
      "learning_rate": 0.00029386247507422533,
      "loss": 3.0479049682617188,
      "step": 69500
    },
    {
      "epoch": 0.0409468645990806,
      "grad_norm": 0.28612393140792847,
      "learning_rate": 0.00029385805952513415,
      "loss": 3.03264404296875,
      "step": 69550
    },
    {
      "epoch": 0.040976301597354554,
      "grad_norm": 0.3466416895389557,
      "learning_rate": 0.00029385364397604296,
      "loss": 3.0170989990234376,
      "step": 69600
    },
    {
      "epoch": 0.04100573859562852,
      "grad_norm": 0.30626189708709717,
      "learning_rate": 0.0002938492284269518,
      "loss": 3.000002136230469,
      "step": 69650
    },
    {
      "epoch": 0.041035175593902476,
      "grad_norm": 0.3278065323829651,
      "learning_rate": 0.0002938448128778606,
      "loss": 3.0096084594726564,
      "step": 69700
    },
    {
      "epoch": 0.04106461259217644,
      "grad_norm": 0.3058152496814728,
      "learning_rate": 0.0002938403973287694,
      "loss": 2.9851397705078124,
      "step": 69750
    },
    {
      "epoch": 0.041094049590450404,
      "grad_norm": 0.380292683839798,
      "learning_rate": 0.0002938359817796782,
      "loss": 3.022324523925781,
      "step": 69800
    },
    {
      "epoch": 0.04112348658872436,
      "grad_norm": 0.2943824529647827,
      "learning_rate": 0.00029383156623058703,
      "loss": 3.001580810546875,
      "step": 69850
    },
    {
      "epoch": 0.041152923586998326,
      "grad_norm": 0.28487786650657654,
      "learning_rate": 0.0002938271506814958,
      "loss": 3.0060406494140626,
      "step": 69900
    },
    {
      "epoch": 0.04118236058527229,
      "grad_norm": 0.3034692704677582,
      "learning_rate": 0.0002938227351324046,
      "loss": 2.961376953125,
      "step": 69950
    },
    {
      "epoch": 0.04121179758354625,
      "grad_norm": 0.3424625098705292,
      "learning_rate": 0.0002938183195833134,
      "loss": 2.9698135375976564,
      "step": 70000
    },
    {
      "epoch": 0.04124123458182021,
      "grad_norm": 0.2612385153770447,
      "learning_rate": 0.00029381390403422224,
      "loss": 2.997080078125,
      "step": 70050
    },
    {
      "epoch": 0.04127067158009417,
      "grad_norm": 0.2927098870277405,
      "learning_rate": 0.00029380948848513105,
      "loss": 2.9875503540039063,
      "step": 70100
    },
    {
      "epoch": 0.04130010857836813,
      "grad_norm": 0.29350191354751587,
      "learning_rate": 0.00029380507293603987,
      "loss": 2.9955328369140624,
      "step": 70150
    },
    {
      "epoch": 0.0413295455766421,
      "grad_norm": 0.26369452476501465,
      "learning_rate": 0.0002938006573869487,
      "loss": 3.038238830566406,
      "step": 70200
    },
    {
      "epoch": 0.041358982574916055,
      "grad_norm": 0.2703210711479187,
      "learning_rate": 0.0002937962418378575,
      "loss": 2.967362060546875,
      "step": 70250
    },
    {
      "epoch": 0.04138841957319002,
      "grad_norm": 0.27218955755233765,
      "learning_rate": 0.00029379182628876626,
      "loss": 2.9884490966796875,
      "step": 70300
    },
    {
      "epoch": 0.04141785657146398,
      "grad_norm": 0.46223142743110657,
      "learning_rate": 0.0002937874107396751,
      "loss": 3.0802349853515625,
      "step": 70350
    },
    {
      "epoch": 0.04144729356973794,
      "grad_norm": 0.28458502888679504,
      "learning_rate": 0.00029378299519058394,
      "loss": 2.9641851806640624,
      "step": 70400
    },
    {
      "epoch": 0.041476730568011905,
      "grad_norm": 0.23219053447246552,
      "learning_rate": 0.0002937785796414927,
      "loss": 3.035416564941406,
      "step": 70450
    },
    {
      "epoch": 0.04150616756628586,
      "grad_norm": 0.3212012052536011,
      "learning_rate": 0.0002937741640924015,
      "loss": 2.9795672607421877,
      "step": 70500
    },
    {
      "epoch": 0.041535604564559826,
      "grad_norm": 0.2977212965488434,
      "learning_rate": 0.00029376974854331033,
      "loss": 3.026783142089844,
      "step": 70550
    },
    {
      "epoch": 0.04156504156283379,
      "grad_norm": 0.39082273840904236,
      "learning_rate": 0.00029376533299421914,
      "loss": 3.026025695800781,
      "step": 70600
    },
    {
      "epoch": 0.04159447856110775,
      "grad_norm": 0.28042399883270264,
      "learning_rate": 0.0002937609174451279,
      "loss": 2.9984066772460936,
      "step": 70650
    },
    {
      "epoch": 0.04162391555938171,
      "grad_norm": 0.266197144985199,
      "learning_rate": 0.00029375650189603677,
      "loss": 3.013608093261719,
      "step": 70700
    },
    {
      "epoch": 0.041653352557655676,
      "grad_norm": 0.2370111495256424,
      "learning_rate": 0.0002937520863469456,
      "loss": 3.0013946533203124,
      "step": 70750
    },
    {
      "epoch": 0.041682789555929634,
      "grad_norm": 0.3986169695854187,
      "learning_rate": 0.00029374767079785435,
      "loss": 3.0041934204101564,
      "step": 70800
    },
    {
      "epoch": 0.0417122265542036,
      "grad_norm": 0.21865655481815338,
      "learning_rate": 0.00029374325524876316,
      "loss": 2.9999288940429687,
      "step": 70850
    },
    {
      "epoch": 0.041741663552477555,
      "grad_norm": 0.3873344957828522,
      "learning_rate": 0.000293738839699672,
      "loss": 3.0067208862304686,
      "step": 70900
    },
    {
      "epoch": 0.04177110055075152,
      "grad_norm": 0.2265145629644394,
      "learning_rate": 0.0002937344241505808,
      "loss": 2.9990869140625,
      "step": 70950
    },
    {
      "epoch": 0.041800537549025484,
      "grad_norm": 0.29325157403945923,
      "learning_rate": 0.0002937300086014896,
      "loss": 2.985956115722656,
      "step": 71000
    },
    {
      "epoch": 0.04182997454729944,
      "grad_norm": 0.30969175696372986,
      "learning_rate": 0.0002937255930523984,
      "loss": 2.9795242309570313,
      "step": 71050
    },
    {
      "epoch": 0.041859411545573405,
      "grad_norm": 0.31356388330459595,
      "learning_rate": 0.00029372117750330723,
      "loss": 3.040467529296875,
      "step": 71100
    },
    {
      "epoch": 0.04188884854384736,
      "grad_norm": 0.2642512321472168,
      "learning_rate": 0.00029371676195421605,
      "loss": 2.977530212402344,
      "step": 71150
    },
    {
      "epoch": 0.04191828554212133,
      "grad_norm": 0.3605392277240753,
      "learning_rate": 0.0002937123464051248,
      "loss": 2.9563137817382814,
      "step": 71200
    },
    {
      "epoch": 0.04194772254039529,
      "grad_norm": 0.2716870605945587,
      "learning_rate": 0.0002937079308560336,
      "loss": 2.960244140625,
      "step": 71250
    },
    {
      "epoch": 0.04197715953866925,
      "grad_norm": 0.31945645809173584,
      "learning_rate": 0.00029370351530694244,
      "loss": 2.99169189453125,
      "step": 71300
    },
    {
      "epoch": 0.04200659653694321,
      "grad_norm": 0.3093312084674835,
      "learning_rate": 0.00029369909975785125,
      "loss": 2.998623046875,
      "step": 71350
    },
    {
      "epoch": 0.04203603353521718,
      "grad_norm": 0.2947550415992737,
      "learning_rate": 0.00029369468420876007,
      "loss": 2.994556884765625,
      "step": 71400
    },
    {
      "epoch": 0.042065470533491134,
      "grad_norm": 0.36010807752609253,
      "learning_rate": 0.0002936902686596689,
      "loss": 3.0256784057617185,
      "step": 71450
    },
    {
      "epoch": 0.0420949075317651,
      "grad_norm": 0.26789334416389465,
      "learning_rate": 0.0002936858531105777,
      "loss": 3.050470886230469,
      "step": 71500
    },
    {
      "epoch": 0.042124344530039055,
      "grad_norm": 0.3408295810222626,
      "learning_rate": 0.0002936814375614865,
      "loss": 2.8943670654296874,
      "step": 71550
    },
    {
      "epoch": 0.04215378152831302,
      "grad_norm": 0.29996436834335327,
      "learning_rate": 0.00029367702201239527,
      "loss": 3.001092529296875,
      "step": 71600
    },
    {
      "epoch": 0.042183218526586984,
      "grad_norm": 0.29358798265457153,
      "learning_rate": 0.00029367260646330414,
      "loss": 3.039256896972656,
      "step": 71650
    },
    {
      "epoch": 0.04221265552486094,
      "grad_norm": 0.3081859350204468,
      "learning_rate": 0.00029366819091421295,
      "loss": 2.958470458984375,
      "step": 71700
    },
    {
      "epoch": 0.042242092523134905,
      "grad_norm": 0.27917346358299255,
      "learning_rate": 0.0002936637753651217,
      "loss": 2.9860470581054686,
      "step": 71750
    },
    {
      "epoch": 0.04227152952140887,
      "grad_norm": 0.26011982560157776,
      "learning_rate": 0.00029365935981603053,
      "loss": 2.9488134765625,
      "step": 71800
    },
    {
      "epoch": 0.04230096651968283,
      "grad_norm": 0.38340744376182556,
      "learning_rate": 0.00029365494426693934,
      "loss": 3.0155340576171876,
      "step": 71850
    },
    {
      "epoch": 0.04233040351795679,
      "grad_norm": 0.35958972573280334,
      "learning_rate": 0.00029365052871784816,
      "loss": 3.032208251953125,
      "step": 71900
    },
    {
      "epoch": 0.04235984051623075,
      "grad_norm": 0.3300125300884247,
      "learning_rate": 0.00029364611316875697,
      "loss": 2.9323638916015624,
      "step": 71950
    },
    {
      "epoch": 0.04238927751450471,
      "grad_norm": 0.2488812655210495,
      "learning_rate": 0.0002936416976196658,
      "loss": 3.0088546752929686,
      "step": 72000
    },
    {
      "epoch": 0.04241871451277868,
      "grad_norm": 0.30755534768104553,
      "learning_rate": 0.0002936372820705746,
      "loss": 3.0053524780273437,
      "step": 72050
    },
    {
      "epoch": 0.042448151511052634,
      "grad_norm": 0.3060452342033386,
      "learning_rate": 0.0002936328665214834,
      "loss": 2.995467224121094,
      "step": 72100
    },
    {
      "epoch": 0.0424775885093266,
      "grad_norm": 0.39770910143852234,
      "learning_rate": 0.0002936284509723922,
      "loss": 2.9729364013671873,
      "step": 72150
    },
    {
      "epoch": 0.04250702550760056,
      "grad_norm": 0.4746624827384949,
      "learning_rate": 0.000293624035423301,
      "loss": 2.999754638671875,
      "step": 72200
    },
    {
      "epoch": 0.04253646250587452,
      "grad_norm": 0.34196388721466064,
      "learning_rate": 0.0002936196198742098,
      "loss": 2.9672006225585936,
      "step": 72250
    },
    {
      "epoch": 0.042565899504148484,
      "grad_norm": 0.39470869302749634,
      "learning_rate": 0.0002936152043251186,
      "loss": 2.997939453125,
      "step": 72300
    },
    {
      "epoch": 0.04259533650242244,
      "grad_norm": 0.28567811846733093,
      "learning_rate": 0.00029361078877602743,
      "loss": 3.018904724121094,
      "step": 72350
    },
    {
      "epoch": 0.042624773500696406,
      "grad_norm": 0.272731751203537,
      "learning_rate": 0.00029360637322693625,
      "loss": 3.0294949340820314,
      "step": 72400
    },
    {
      "epoch": 0.04265421049897037,
      "grad_norm": 0.39119473099708557,
      "learning_rate": 0.00029360195767784506,
      "loss": 2.9856192016601564,
      "step": 72450
    },
    {
      "epoch": 0.04268364749724433,
      "grad_norm": 0.29839712381362915,
      "learning_rate": 0.0002935975421287538,
      "loss": 3.018202819824219,
      "step": 72500
    },
    {
      "epoch": 0.04271308449551829,
      "grad_norm": 0.4187307357788086,
      "learning_rate": 0.00029359312657966264,
      "loss": 2.98239501953125,
      "step": 72550
    },
    {
      "epoch": 0.04274252149379225,
      "grad_norm": 0.2907007038593292,
      "learning_rate": 0.00029358871103057145,
      "loss": 3.005994873046875,
      "step": 72600
    },
    {
      "epoch": 0.04277195849206621,
      "grad_norm": 0.3380918502807617,
      "learning_rate": 0.00029358429548148027,
      "loss": 3.022466735839844,
      "step": 72650
    },
    {
      "epoch": 0.04280139549034018,
      "grad_norm": 0.2793414294719696,
      "learning_rate": 0.0002935798799323891,
      "loss": 3.015897216796875,
      "step": 72700
    },
    {
      "epoch": 0.042830832488614135,
      "grad_norm": 0.3547948896884918,
      "learning_rate": 0.0002935754643832979,
      "loss": 3.0234600830078127,
      "step": 72750
    },
    {
      "epoch": 0.0428602694868881,
      "grad_norm": 0.22187143564224243,
      "learning_rate": 0.0002935710488342067,
      "loss": 2.9976123046875,
      "step": 72800
    },
    {
      "epoch": 0.04288970648516206,
      "grad_norm": 0.3255198895931244,
      "learning_rate": 0.0002935666332851155,
      "loss": 2.99912841796875,
      "step": 72850
    },
    {
      "epoch": 0.04291914348343602,
      "grad_norm": 0.2809775173664093,
      "learning_rate": 0.0002935622177360243,
      "loss": 3.0889801025390624,
      "step": 72900
    },
    {
      "epoch": 0.042948580481709984,
      "grad_norm": 0.26523661613464355,
      "learning_rate": 0.00029355780218693315,
      "loss": 2.9947564697265623,
      "step": 72950
    },
    {
      "epoch": 0.04297801747998394,
      "grad_norm": 0.3423626720905304,
      "learning_rate": 0.00029355338663784197,
      "loss": 3.0573660278320314,
      "step": 73000
    },
    {
      "epoch": 0.043007454478257906,
      "grad_norm": 0.4677830636501312,
      "learning_rate": 0.00029354897108875073,
      "loss": 2.984145202636719,
      "step": 73050
    },
    {
      "epoch": 0.04303689147653187,
      "grad_norm": 0.32851386070251465,
      "learning_rate": 0.00029354455553965954,
      "loss": 2.9754461669921874,
      "step": 73100
    },
    {
      "epoch": 0.04306632847480583,
      "grad_norm": 0.23142611980438232,
      "learning_rate": 0.00029354013999056836,
      "loss": 3.0073748779296876,
      "step": 73150
    },
    {
      "epoch": 0.04309576547307979,
      "grad_norm": 0.33782634139060974,
      "learning_rate": 0.00029353572444147717,
      "loss": 2.9874285888671874,
      "step": 73200
    },
    {
      "epoch": 0.043125202471353756,
      "grad_norm": 0.2807283103466034,
      "learning_rate": 0.000293531308892386,
      "loss": 3.040721130371094,
      "step": 73250
    },
    {
      "epoch": 0.04315463946962771,
      "grad_norm": 0.36112603545188904,
      "learning_rate": 0.0002935268933432948,
      "loss": 3.0334982299804687,
      "step": 73300
    },
    {
      "epoch": 0.04318407646790168,
      "grad_norm": 0.3012598752975464,
      "learning_rate": 0.0002935224777942036,
      "loss": 3.005351257324219,
      "step": 73350
    },
    {
      "epoch": 0.043213513466175635,
      "grad_norm": 0.2800009250640869,
      "learning_rate": 0.00029351806224511243,
      "loss": 2.9320474243164063,
      "step": 73400
    },
    {
      "epoch": 0.0432429504644496,
      "grad_norm": 0.3120989501476288,
      "learning_rate": 0.0002935136466960212,
      "loss": 3.0144729614257812,
      "step": 73450
    },
    {
      "epoch": 0.04327238746272356,
      "grad_norm": 0.3960503041744232,
      "learning_rate": 0.00029350923114693,
      "loss": 3.0116696166992187,
      "step": 73500
    },
    {
      "epoch": 0.04330182446099752,
      "grad_norm": 0.3386993110179901,
      "learning_rate": 0.0002935048155978388,
      "loss": 2.971710205078125,
      "step": 73550
    },
    {
      "epoch": 0.043331261459271485,
      "grad_norm": 0.2906440794467926,
      "learning_rate": 0.00029350040004874763,
      "loss": 2.9847393798828126,
      "step": 73600
    },
    {
      "epoch": 0.04336069845754545,
      "grad_norm": 0.34912148118019104,
      "learning_rate": 0.00029349598449965645,
      "loss": 3.021273193359375,
      "step": 73650
    },
    {
      "epoch": 0.043390135455819406,
      "grad_norm": 0.3563450574874878,
      "learning_rate": 0.00029349156895056526,
      "loss": 2.9380618286132814,
      "step": 73700
    },
    {
      "epoch": 0.04341957245409337,
      "grad_norm": 0.3351134955883026,
      "learning_rate": 0.0002934871534014741,
      "loss": 3.020028381347656,
      "step": 73750
    },
    {
      "epoch": 0.04344900945236733,
      "grad_norm": 0.3323858082294464,
      "learning_rate": 0.0002934827378523829,
      "loss": 2.99134033203125,
      "step": 73800
    },
    {
      "epoch": 0.04347844645064129,
      "grad_norm": 0.26279717683792114,
      "learning_rate": 0.00029347832230329165,
      "loss": 3.058319396972656,
      "step": 73850
    },
    {
      "epoch": 0.043507883448915256,
      "grad_norm": 0.47315171360969543,
      "learning_rate": 0.00029347390675420047,
      "loss": 2.996357116699219,
      "step": 73900
    },
    {
      "epoch": 0.043537320447189214,
      "grad_norm": 0.28108909726142883,
      "learning_rate": 0.00029346949120510933,
      "loss": 3.021797790527344,
      "step": 73950
    },
    {
      "epoch": 0.04356675744546318,
      "grad_norm": 0.21004638075828552,
      "learning_rate": 0.0002934650756560181,
      "loss": 3.0506527709960936,
      "step": 74000
    },
    {
      "epoch": 0.04359619444373714,
      "grad_norm": 0.23929280042648315,
      "learning_rate": 0.0002934606601069269,
      "loss": 3.0033123779296873,
      "step": 74050
    },
    {
      "epoch": 0.0436256314420111,
      "grad_norm": 0.303564190864563,
      "learning_rate": 0.0002934562445578357,
      "loss": 3.038427734375,
      "step": 74100
    },
    {
      "epoch": 0.043655068440285064,
      "grad_norm": 0.26747608184814453,
      "learning_rate": 0.00029345182900874454,
      "loss": 3.0052359008789065,
      "step": 74150
    },
    {
      "epoch": 0.04368450543855902,
      "grad_norm": 0.3258504271507263,
      "learning_rate": 0.0002934474134596533,
      "loss": 2.975086364746094,
      "step": 74200
    },
    {
      "epoch": 0.043713942436832985,
      "grad_norm": 0.36702287197113037,
      "learning_rate": 0.00029344299791056217,
      "loss": 3.0601901245117187,
      "step": 74250
    },
    {
      "epoch": 0.04374337943510695,
      "grad_norm": 0.25120872259140015,
      "learning_rate": 0.000293438582361471,
      "loss": 2.9766583251953125,
      "step": 74300
    },
    {
      "epoch": 0.04377281643338091,
      "grad_norm": 0.3142986297607422,
      "learning_rate": 0.00029343416681237974,
      "loss": 3.003900451660156,
      "step": 74350
    },
    {
      "epoch": 0.04380225343165487,
      "grad_norm": 0.4683057963848114,
      "learning_rate": 0.00029342975126328856,
      "loss": 2.9696270751953127,
      "step": 74400
    },
    {
      "epoch": 0.04383169042992883,
      "grad_norm": 0.35077014565467834,
      "learning_rate": 0.00029342533571419737,
      "loss": 2.976884460449219,
      "step": 74450
    },
    {
      "epoch": 0.04386112742820279,
      "grad_norm": 0.31220731139183044,
      "learning_rate": 0.0002934209201651062,
      "loss": 2.926393737792969,
      "step": 74500
    },
    {
      "epoch": 0.04389056442647676,
      "grad_norm": 0.3672977685928345,
      "learning_rate": 0.000293416504616015,
      "loss": 3.00499755859375,
      "step": 74550
    },
    {
      "epoch": 0.043920001424750714,
      "grad_norm": 0.44819679856300354,
      "learning_rate": 0.0002934120890669238,
      "loss": 2.985255126953125,
      "step": 74600
    },
    {
      "epoch": 0.04394943842302468,
      "grad_norm": 0.36861079931259155,
      "learning_rate": 0.00029340767351783263,
      "loss": 2.9959271240234373,
      "step": 74650
    },
    {
      "epoch": 0.04397887542129864,
      "grad_norm": 0.28891173005104065,
      "learning_rate": 0.00029340325796874144,
      "loss": 2.95556396484375,
      "step": 74700
    },
    {
      "epoch": 0.0440083124195726,
      "grad_norm": 0.2891867160797119,
      "learning_rate": 0.0002933988424196502,
      "loss": 2.970328674316406,
      "step": 74750
    },
    {
      "epoch": 0.044037749417846564,
      "grad_norm": 0.22075335681438446,
      "learning_rate": 0.000293394426870559,
      "loss": 2.9575164794921873,
      "step": 74800
    },
    {
      "epoch": 0.04406718641612052,
      "grad_norm": 0.3547993004322052,
      "learning_rate": 0.00029339001132146783,
      "loss": 2.9420175170898437,
      "step": 74850
    },
    {
      "epoch": 0.044096623414394485,
      "grad_norm": 0.25040775537490845,
      "learning_rate": 0.00029338559577237665,
      "loss": 3.025487060546875,
      "step": 74900
    },
    {
      "epoch": 0.04412606041266845,
      "grad_norm": 0.2698875069618225,
      "learning_rate": 0.00029338118022328546,
      "loss": 2.98250732421875,
      "step": 74950
    },
    {
      "epoch": 0.04415549741094241,
      "grad_norm": 0.36949434876441956,
      "learning_rate": 0.0002933767646741943,
      "loss": 3.016926574707031,
      "step": 75000
    },
    {
      "epoch": 0.04418493440921637,
      "grad_norm": 0.2557656168937683,
      "learning_rate": 0.0002933723491251031,
      "loss": 2.9817913818359374,
      "step": 75050
    },
    {
      "epoch": 0.044214371407490335,
      "grad_norm": 0.3309698700904846,
      "learning_rate": 0.0002933679335760119,
      "loss": 3.012795104980469,
      "step": 75100
    },
    {
      "epoch": 0.04424380840576429,
      "grad_norm": 0.38079002499580383,
      "learning_rate": 0.00029336351802692067,
      "loss": 3.0126788330078127,
      "step": 75150
    },
    {
      "epoch": 0.04427324540403826,
      "grad_norm": 0.2803564965724945,
      "learning_rate": 0.0002933591024778295,
      "loss": 3.0539703369140625,
      "step": 75200
    },
    {
      "epoch": 0.044302682402312214,
      "grad_norm": 0.3711020350456238,
      "learning_rate": 0.00029335468692873835,
      "loss": 2.9930160522460936,
      "step": 75250
    },
    {
      "epoch": 0.04433211940058618,
      "grad_norm": 0.28908953070640564,
      "learning_rate": 0.0002933502713796471,
      "loss": 3.016504821777344,
      "step": 75300
    },
    {
      "epoch": 0.04436155639886014,
      "grad_norm": 0.28705233335494995,
      "learning_rate": 0.0002933458558305559,
      "loss": 2.9291763305664062,
      "step": 75350
    },
    {
      "epoch": 0.0443909933971341,
      "grad_norm": 0.4106670916080475,
      "learning_rate": 0.00029334144028146474,
      "loss": 2.9765496826171876,
      "step": 75400
    },
    {
      "epoch": 0.044420430395408064,
      "grad_norm": 0.33047720789909363,
      "learning_rate": 0.00029333702473237355,
      "loss": 2.9801727294921876,
      "step": 75450
    },
    {
      "epoch": 0.04444986739368203,
      "grad_norm": 0.262068510055542,
      "learning_rate": 0.00029333260918328237,
      "loss": 2.974722900390625,
      "step": 75500
    },
    {
      "epoch": 0.044479304391955986,
      "grad_norm": 0.41572892665863037,
      "learning_rate": 0.00029332819363419113,
      "loss": 3.0167721557617186,
      "step": 75550
    },
    {
      "epoch": 0.04450874139022995,
      "grad_norm": 0.2542446553707123,
      "learning_rate": 0.0002933237780851,
      "loss": 3.0500848388671873,
      "step": 75600
    },
    {
      "epoch": 0.04453817838850391,
      "grad_norm": 0.301050066947937,
      "learning_rate": 0.0002933193625360088,
      "loss": 2.974074401855469,
      "step": 75650
    },
    {
      "epoch": 0.04456761538677787,
      "grad_norm": 0.33763259649276733,
      "learning_rate": 0.00029331494698691757,
      "loss": 2.998934631347656,
      "step": 75700
    },
    {
      "epoch": 0.044597052385051836,
      "grad_norm": 0.25087401270866394,
      "learning_rate": 0.0002933105314378264,
      "loss": 2.992898254394531,
      "step": 75750
    },
    {
      "epoch": 0.04462648938332579,
      "grad_norm": 0.38863804936408997,
      "learning_rate": 0.0002933061158887352,
      "loss": 2.987515563964844,
      "step": 75800
    },
    {
      "epoch": 0.04465592638159976,
      "grad_norm": 0.46215882897377014,
      "learning_rate": 0.000293301700339644,
      "loss": 3.017305908203125,
      "step": 75850
    },
    {
      "epoch": 0.04468536337987372,
      "grad_norm": 0.21927352249622345,
      "learning_rate": 0.00029329728479055283,
      "loss": 2.9944161987304687,
      "step": 75900
    },
    {
      "epoch": 0.04471480037814768,
      "grad_norm": 0.2361283153295517,
      "learning_rate": 0.00029329286924146164,
      "loss": 3.0062527465820312,
      "step": 75950
    },
    {
      "epoch": 0.04474423737642164,
      "grad_norm": 0.3091279864311218,
      "learning_rate": 0.00029328845369237046,
      "loss": 2.973292236328125,
      "step": 76000
    },
    {
      "epoch": 0.0447736743746956,
      "grad_norm": 0.39961564540863037,
      "learning_rate": 0.0002932840381432792,
      "loss": 3.005071105957031,
      "step": 76050
    },
    {
      "epoch": 0.044803111372969565,
      "grad_norm": 0.25927039980888367,
      "learning_rate": 0.00029327962259418803,
      "loss": 3.001928405761719,
      "step": 76100
    },
    {
      "epoch": 0.04483254837124353,
      "grad_norm": 0.3904722332954407,
      "learning_rate": 0.00029327520704509685,
      "loss": 2.9712554931640627,
      "step": 76150
    },
    {
      "epoch": 0.044861985369517486,
      "grad_norm": 0.37804216146469116,
      "learning_rate": 0.00029327079149600566,
      "loss": 3.0434307861328125,
      "step": 76200
    },
    {
      "epoch": 0.04489142236779145,
      "grad_norm": 0.2361689805984497,
      "learning_rate": 0.0002932663759469145,
      "loss": 2.986275939941406,
      "step": 76250
    },
    {
      "epoch": 0.04492085936606541,
      "grad_norm": 0.3587456941604614,
      "learning_rate": 0.0002932619603978233,
      "loss": 3.0313116455078126,
      "step": 76300
    },
    {
      "epoch": 0.04495029636433937,
      "grad_norm": 0.3424977958202362,
      "learning_rate": 0.0002932575448487321,
      "loss": 3.0419937133789063,
      "step": 76350
    },
    {
      "epoch": 0.044979733362613336,
      "grad_norm": 0.3239906132221222,
      "learning_rate": 0.0002932531292996409,
      "loss": 2.9944476318359374,
      "step": 76400
    },
    {
      "epoch": 0.04500917036088729,
      "grad_norm": 0.3535572588443756,
      "learning_rate": 0.0002932487137505497,
      "loss": 3.0412533569335936,
      "step": 76450
    },
    {
      "epoch": 0.04503860735916126,
      "grad_norm": 0.3070663511753082,
      "learning_rate": 0.0002932442982014585,
      "loss": 2.937154541015625,
      "step": 76500
    },
    {
      "epoch": 0.04506804435743522,
      "grad_norm": 0.43307191133499146,
      "learning_rate": 0.00029323988265236736,
      "loss": 2.9733895874023437,
      "step": 76550
    },
    {
      "epoch": 0.04509748135570918,
      "grad_norm": 0.3441995680332184,
      "learning_rate": 0.0002932354671032761,
      "loss": 3.0130929565429687,
      "step": 76600
    },
    {
      "epoch": 0.04512691835398314,
      "grad_norm": 0.342435747385025,
      "learning_rate": 0.00029323105155418494,
      "loss": 2.9725897216796877,
      "step": 76650
    },
    {
      "epoch": 0.0451563553522571,
      "grad_norm": 0.30592671036720276,
      "learning_rate": 0.00029322663600509375,
      "loss": 2.945675964355469,
      "step": 76700
    },
    {
      "epoch": 0.045185792350531065,
      "grad_norm": 0.2917761504650116,
      "learning_rate": 0.00029322222045600257,
      "loss": 2.937332763671875,
      "step": 76750
    },
    {
      "epoch": 0.04521522934880503,
      "grad_norm": 0.25603505969047546,
      "learning_rate": 0.0002932178049069114,
      "loss": 2.965013427734375,
      "step": 76800
    },
    {
      "epoch": 0.045244666347078986,
      "grad_norm": 0.3016587197780609,
      "learning_rate": 0.00029321338935782014,
      "loss": 2.93090576171875,
      "step": 76850
    },
    {
      "epoch": 0.04527410334535295,
      "grad_norm": 0.2617456614971161,
      "learning_rate": 0.000293208973808729,
      "loss": 2.99254150390625,
      "step": 76900
    },
    {
      "epoch": 0.045303540343626915,
      "grad_norm": 0.29680007696151733,
      "learning_rate": 0.0002932045582596378,
      "loss": 2.9296304321289064,
      "step": 76950
    },
    {
      "epoch": 0.04533297734190087,
      "grad_norm": 0.37218260765075684,
      "learning_rate": 0.0002932001427105466,
      "loss": 2.973033447265625,
      "step": 77000
    },
    {
      "epoch": 0.045362414340174836,
      "grad_norm": 0.30559924244880676,
      "learning_rate": 0.0002931957271614554,
      "loss": 2.9789996337890625,
      "step": 77050
    },
    {
      "epoch": 0.045391851338448794,
      "grad_norm": 0.2979046404361725,
      "learning_rate": 0.0002931913116123642,
      "loss": 2.9189572143554687,
      "step": 77100
    },
    {
      "epoch": 0.04542128833672276,
      "grad_norm": 0.24459974467754364,
      "learning_rate": 0.00029318689606327303,
      "loss": 2.9653741455078126,
      "step": 77150
    },
    {
      "epoch": 0.04545072533499672,
      "grad_norm": 0.24918337166309357,
      "learning_rate": 0.00029318248051418184,
      "loss": 3.0251104736328127,
      "step": 77200
    },
    {
      "epoch": 0.04548016233327068,
      "grad_norm": 0.3018087148666382,
      "learning_rate": 0.00029317806496509066,
      "loss": 3.00870849609375,
      "step": 77250
    },
    {
      "epoch": 0.045509599331544644,
      "grad_norm": 0.3110695779323578,
      "learning_rate": 0.00029317364941599947,
      "loss": 2.969013366699219,
      "step": 77300
    },
    {
      "epoch": 0.04553903632981861,
      "grad_norm": 0.22217288613319397,
      "learning_rate": 0.0002931692338669083,
      "loss": 2.9531686401367185,
      "step": 77350
    },
    {
      "epoch": 0.045568473328092565,
      "grad_norm": 0.33046773076057434,
      "learning_rate": 0.00029316481831781705,
      "loss": 2.9970208740234376,
      "step": 77400
    },
    {
      "epoch": 0.04559791032636653,
      "grad_norm": 0.31086766719818115,
      "learning_rate": 0.00029316040276872586,
      "loss": 2.9982675170898436,
      "step": 77450
    },
    {
      "epoch": 0.04562734732464049,
      "grad_norm": 0.3520906865596771,
      "learning_rate": 0.0002931559872196347,
      "loss": 2.9413543701171876,
      "step": 77500
    },
    {
      "epoch": 0.04565678432291445,
      "grad_norm": 0.32531484961509705,
      "learning_rate": 0.0002931515716705435,
      "loss": 2.956531677246094,
      "step": 77550
    },
    {
      "epoch": 0.045686221321188415,
      "grad_norm": 0.2282513976097107,
      "learning_rate": 0.0002931471561214523,
      "loss": 2.990048828125,
      "step": 77600
    },
    {
      "epoch": 0.04571565831946237,
      "grad_norm": 0.39775121212005615,
      "learning_rate": 0.0002931427405723611,
      "loss": 2.944000244140625,
      "step": 77650
    },
    {
      "epoch": 0.04574509531773634,
      "grad_norm": 0.3927665650844574,
      "learning_rate": 0.00029313832502326993,
      "loss": 2.9724557495117185,
      "step": 77700
    },
    {
      "epoch": 0.0457745323160103,
      "grad_norm": 0.4143678843975067,
      "learning_rate": 0.0002931339094741787,
      "loss": 2.9582040405273435,
      "step": 77750
    },
    {
      "epoch": 0.04580396931428426,
      "grad_norm": 0.315637469291687,
      "learning_rate": 0.0002931294939250875,
      "loss": 3.0244140625,
      "step": 77800
    },
    {
      "epoch": 0.04583340631255822,
      "grad_norm": 0.32850539684295654,
      "learning_rate": 0.0002931250783759964,
      "loss": 2.9709072875976563,
      "step": 77850
    },
    {
      "epoch": 0.04586284331083218,
      "grad_norm": 0.357577919960022,
      "learning_rate": 0.00029312066282690514,
      "loss": 2.97823974609375,
      "step": 77900
    },
    {
      "epoch": 0.045892280309106144,
      "grad_norm": 0.43131357431411743,
      "learning_rate": 0.00029311624727781395,
      "loss": 2.9581417846679687,
      "step": 77950
    },
    {
      "epoch": 0.04592171730738011,
      "grad_norm": 0.2528178095817566,
      "learning_rate": 0.00029311183172872277,
      "loss": 2.9665182495117186,
      "step": 78000
    },
    {
      "epoch": 0.045951154305654066,
      "grad_norm": 0.3109012246131897,
      "learning_rate": 0.0002931074161796316,
      "loss": 2.9998004150390627,
      "step": 78050
    },
    {
      "epoch": 0.04598059130392803,
      "grad_norm": 0.26486527919769287,
      "learning_rate": 0.0002931030006305404,
      "loss": 2.9294912719726565,
      "step": 78100
    },
    {
      "epoch": 0.04601002830220199,
      "grad_norm": 0.4030660092830658,
      "learning_rate": 0.00029309858508144916,
      "loss": 2.93941162109375,
      "step": 78150
    },
    {
      "epoch": 0.04603946530047595,
      "grad_norm": 0.29827579855918884,
      "learning_rate": 0.000293094169532358,
      "loss": 2.943575439453125,
      "step": 78200
    },
    {
      "epoch": 0.046068902298749916,
      "grad_norm": 0.28411224484443665,
      "learning_rate": 0.00029308975398326684,
      "loss": 2.9482513427734376,
      "step": 78250
    },
    {
      "epoch": 0.04609833929702387,
      "grad_norm": 0.3694169819355011,
      "learning_rate": 0.0002930853384341756,
      "loss": 2.9961553955078126,
      "step": 78300
    },
    {
      "epoch": 0.04612777629529784,
      "grad_norm": 0.23741601407527924,
      "learning_rate": 0.0002930809228850844,
      "loss": 2.928175964355469,
      "step": 78350
    },
    {
      "epoch": 0.0461572132935718,
      "grad_norm": 0.2452349066734314,
      "learning_rate": 0.00029307650733599323,
      "loss": 2.9900213623046876,
      "step": 78400
    },
    {
      "epoch": 0.04618665029184576,
      "grad_norm": 0.42046603560447693,
      "learning_rate": 0.00029307209178690204,
      "loss": 2.9856826782226564,
      "step": 78450
    },
    {
      "epoch": 0.04621608729011972,
      "grad_norm": 0.2555469870567322,
      "learning_rate": 0.00029306767623781086,
      "loss": 2.973406982421875,
      "step": 78500
    },
    {
      "epoch": 0.04624552428839368,
      "grad_norm": 0.29096269607543945,
      "learning_rate": 0.00029306326068871967,
      "loss": 2.9920330810546876,
      "step": 78550
    },
    {
      "epoch": 0.046274961286667644,
      "grad_norm": 0.4153391420841217,
      "learning_rate": 0.0002930588451396285,
      "loss": 2.957242126464844,
      "step": 78600
    },
    {
      "epoch": 0.04630439828494161,
      "grad_norm": 0.2661893665790558,
      "learning_rate": 0.0002930544295905373,
      "loss": 3.00689453125,
      "step": 78650
    },
    {
      "epoch": 0.046333835283215566,
      "grad_norm": 0.3118010461330414,
      "learning_rate": 0.00029305001404144606,
      "loss": 2.941624755859375,
      "step": 78700
    },
    {
      "epoch": 0.04636327228148953,
      "grad_norm": 0.30924275517463684,
      "learning_rate": 0.0002930455984923549,
      "loss": 2.9978570556640625,
      "step": 78750
    },
    {
      "epoch": 0.046392709279763494,
      "grad_norm": 0.3121529221534729,
      "learning_rate": 0.0002930411829432637,
      "loss": 2.9376766967773436,
      "step": 78800
    },
    {
      "epoch": 0.04642214627803745,
      "grad_norm": 0.4334242343902588,
      "learning_rate": 0.0002930367673941725,
      "loss": 2.9842318725585937,
      "step": 78850
    },
    {
      "epoch": 0.046451583276311416,
      "grad_norm": 0.3868156969547272,
      "learning_rate": 0.0002930323518450813,
      "loss": 2.9948358154296875,
      "step": 78900
    },
    {
      "epoch": 0.04648102027458537,
      "grad_norm": 0.2415110021829605,
      "learning_rate": 0.00029302793629599013,
      "loss": 2.9591180419921876,
      "step": 78950
    },
    {
      "epoch": 0.04651045727285934,
      "grad_norm": 0.23870296776294708,
      "learning_rate": 0.00029302352074689895,
      "loss": 3.0032745361328126,
      "step": 79000
    },
    {
      "epoch": 0.0465398942711333,
      "grad_norm": 0.40373849868774414,
      "learning_rate": 0.00029301910519780776,
      "loss": 3.0196075439453125,
      "step": 79050
    },
    {
      "epoch": 0.04656933126940726,
      "grad_norm": 0.1967240422964096,
      "learning_rate": 0.0002930146896487165,
      "loss": 2.98166748046875,
      "step": 79100
    },
    {
      "epoch": 0.04659876826768122,
      "grad_norm": 0.26675650477409363,
      "learning_rate": 0.0002930102740996254,
      "loss": 2.967279968261719,
      "step": 79150
    },
    {
      "epoch": 0.04662820526595519,
      "grad_norm": 0.33509448170661926,
      "learning_rate": 0.0002930058585505342,
      "loss": 2.953134765625,
      "step": 79200
    },
    {
      "epoch": 0.046657642264229145,
      "grad_norm": 0.3009587824344635,
      "learning_rate": 0.00029300144300144297,
      "loss": 2.9976190185546874,
      "step": 79250
    },
    {
      "epoch": 0.04668707926250311,
      "grad_norm": 0.42004817724227905,
      "learning_rate": 0.0002929970274523518,
      "loss": 2.998106689453125,
      "step": 79300
    },
    {
      "epoch": 0.046716516260777066,
      "grad_norm": 0.26585450768470764,
      "learning_rate": 0.0002929926119032606,
      "loss": 2.9941986083984373,
      "step": 79350
    },
    {
      "epoch": 0.04674595325905103,
      "grad_norm": 0.4969388246536255,
      "learning_rate": 0.0002929881963541694,
      "loss": 2.984637451171875,
      "step": 79400
    },
    {
      "epoch": 0.046775390257324995,
      "grad_norm": 0.3058989346027374,
      "learning_rate": 0.00029298378080507817,
      "loss": 3.008701171875,
      "step": 79450
    },
    {
      "epoch": 0.04680482725559895,
      "grad_norm": 0.3582475781440735,
      "learning_rate": 0.00029297936525598704,
      "loss": 2.9503436279296875,
      "step": 79500
    },
    {
      "epoch": 0.046834264253872916,
      "grad_norm": 0.3514373302459717,
      "learning_rate": 0.00029297494970689585,
      "loss": 2.8965878295898437,
      "step": 79550
    },
    {
      "epoch": 0.04686370125214688,
      "grad_norm": 0.29862260818481445,
      "learning_rate": 0.0002929705341578046,
      "loss": 2.929866943359375,
      "step": 79600
    },
    {
      "epoch": 0.04689313825042084,
      "grad_norm": 0.31989550590515137,
      "learning_rate": 0.0002929661186087134,
      "loss": 2.9501910400390625,
      "step": 79650
    },
    {
      "epoch": 0.0469225752486948,
      "grad_norm": 0.34409502148628235,
      "learning_rate": 0.00029296170305962224,
      "loss": 2.9328994750976562,
      "step": 79700
    },
    {
      "epoch": 0.04695201224696876,
      "grad_norm": 0.4673641622066498,
      "learning_rate": 0.00029295728751053106,
      "loss": 2.9237164306640624,
      "step": 79750
    },
    {
      "epoch": 0.046981449245242723,
      "grad_norm": 0.3016417920589447,
      "learning_rate": 0.00029295287196143987,
      "loss": 2.9999200439453126,
      "step": 79800
    },
    {
      "epoch": 0.04701088624351669,
      "grad_norm": 0.23751798272132874,
      "learning_rate": 0.0002929484564123487,
      "loss": 2.9345587158203124,
      "step": 79850
    },
    {
      "epoch": 0.047040323241790645,
      "grad_norm": 0.3023880422115326,
      "learning_rate": 0.0002929440408632575,
      "loss": 3.012760009765625,
      "step": 79900
    },
    {
      "epoch": 0.04706976024006461,
      "grad_norm": 0.30775395035743713,
      "learning_rate": 0.0002929396253141663,
      "loss": 2.9729840087890627,
      "step": 79950
    },
    {
      "epoch": 0.047099197238338567,
      "grad_norm": 0.2853832542896271,
      "learning_rate": 0.0002929352097650751,
      "loss": 2.9093313598632813,
      "step": 80000
    },
    {
      "epoch": 0.04712863423661253,
      "grad_norm": 0.4137921929359436,
      "learning_rate": 0.0002929307942159839,
      "loss": 2.9372845458984376,
      "step": 80050
    },
    {
      "epoch": 0.047158071234886495,
      "grad_norm": 0.43859031796455383,
      "learning_rate": 0.0002929263786668927,
      "loss": 2.989783935546875,
      "step": 80100
    },
    {
      "epoch": 0.04718750823316045,
      "grad_norm": 0.2873845100402832,
      "learning_rate": 0.0002929219631178015,
      "loss": 2.9568600463867187,
      "step": 80150
    },
    {
      "epoch": 0.047216945231434417,
      "grad_norm": 0.2333100140094757,
      "learning_rate": 0.00029291754756871033,
      "loss": 2.9419650268554687,
      "step": 80200
    },
    {
      "epoch": 0.04724638222970838,
      "grad_norm": 0.28825393319129944,
      "learning_rate": 0.00029291313201961915,
      "loss": 2.917855224609375,
      "step": 80250
    },
    {
      "epoch": 0.04727581922798234,
      "grad_norm": 0.3679467439651489,
      "learning_rate": 0.00029290871647052796,
      "loss": 3.0024923706054687,
      "step": 80300
    },
    {
      "epoch": 0.0473052562262563,
      "grad_norm": 0.2854974567890167,
      "learning_rate": 0.0002929043009214368,
      "loss": 2.958648681640625,
      "step": 80350
    },
    {
      "epoch": 0.04733469322453026,
      "grad_norm": 0.2517436146736145,
      "learning_rate": 0.00029289988537234554,
      "loss": 2.968162841796875,
      "step": 80400
    },
    {
      "epoch": 0.047364130222804224,
      "grad_norm": 0.31256231665611267,
      "learning_rate": 0.0002928954698232544,
      "loss": 2.944857482910156,
      "step": 80450
    },
    {
      "epoch": 0.04739356722107819,
      "grad_norm": 0.3143554627895355,
      "learning_rate": 0.0002928910542741632,
      "loss": 2.9819097900390625,
      "step": 80500
    },
    {
      "epoch": 0.047423004219352145,
      "grad_norm": 0.2788316011428833,
      "learning_rate": 0.000292886638725072,
      "loss": 2.993269958496094,
      "step": 80550
    },
    {
      "epoch": 0.04745244121762611,
      "grad_norm": 0.40866199135780334,
      "learning_rate": 0.0002928822231759808,
      "loss": 2.9522930908203127,
      "step": 80600
    },
    {
      "epoch": 0.047481878215900074,
      "grad_norm": 0.2969707250595093,
      "learning_rate": 0.0002928778076268896,
      "loss": 2.9012411499023436,
      "step": 80650
    },
    {
      "epoch": 0.04751131521417403,
      "grad_norm": 0.24488116800785065,
      "learning_rate": 0.0002928733920777984,
      "loss": 2.926302490234375,
      "step": 80700
    },
    {
      "epoch": 0.047540752212447995,
      "grad_norm": 0.3316980302333832,
      "learning_rate": 0.00029286897652870724,
      "loss": 2.933970947265625,
      "step": 80750
    },
    {
      "epoch": 0.04757018921072195,
      "grad_norm": 0.3887718915939331,
      "learning_rate": 0.00029286456097961605,
      "loss": 3.024355773925781,
      "step": 80800
    },
    {
      "epoch": 0.04759962620899592,
      "grad_norm": 0.3128609359264374,
      "learning_rate": 0.00029286014543052487,
      "loss": 2.92742431640625,
      "step": 80850
    },
    {
      "epoch": 0.04762906320726988,
      "grad_norm": 0.3241572380065918,
      "learning_rate": 0.0002928557298814337,
      "loss": 2.9725897216796877,
      "step": 80900
    },
    {
      "epoch": 0.04765850020554384,
      "grad_norm": 0.2809228003025055,
      "learning_rate": 0.00029285131433234244,
      "loss": 3.0042388916015623,
      "step": 80950
    },
    {
      "epoch": 0.0476879372038178,
      "grad_norm": 0.28632405400276184,
      "learning_rate": 0.00029284689878325126,
      "loss": 2.9664602661132813,
      "step": 81000
    },
    {
      "epoch": 0.04771737420209177,
      "grad_norm": 0.34363028407096863,
      "learning_rate": 0.00029284248323416007,
      "loss": 2.994759521484375,
      "step": 81050
    },
    {
      "epoch": 0.047746811200365724,
      "grad_norm": 0.40863463282585144,
      "learning_rate": 0.0002928380676850689,
      "loss": 3.0024169921875,
      "step": 81100
    },
    {
      "epoch": 0.04777624819863969,
      "grad_norm": 0.3283088207244873,
      "learning_rate": 0.0002928336521359777,
      "loss": 2.946050109863281,
      "step": 81150
    },
    {
      "epoch": 0.047805685196913646,
      "grad_norm": 0.32985416054725647,
      "learning_rate": 0.0002928292365868865,
      "loss": 2.9418911743164062,
      "step": 81200
    },
    {
      "epoch": 0.04783512219518761,
      "grad_norm": 0.2717657685279846,
      "learning_rate": 0.00029282482103779533,
      "loss": 2.9116250610351564,
      "step": 81250
    },
    {
      "epoch": 0.047864559193461574,
      "grad_norm": 0.42166540026664734,
      "learning_rate": 0.0002928204054887041,
      "loss": 2.8719775390625,
      "step": 81300
    },
    {
      "epoch": 0.04789399619173553,
      "grad_norm": 0.2708142399787903,
      "learning_rate": 0.0002928159899396129,
      "loss": 2.941244812011719,
      "step": 81350
    },
    {
      "epoch": 0.047923433190009496,
      "grad_norm": 0.29358041286468506,
      "learning_rate": 0.0002928115743905217,
      "loss": 2.9954296875,
      "step": 81400
    },
    {
      "epoch": 0.04795287018828346,
      "grad_norm": 0.2816022038459778,
      "learning_rate": 0.00029280715884143053,
      "loss": 2.962860107421875,
      "step": 81450
    },
    {
      "epoch": 0.04798230718655742,
      "grad_norm": 0.32650014758110046,
      "learning_rate": 0.00029280274329233935,
      "loss": 2.9141912841796875,
      "step": 81500
    },
    {
      "epoch": 0.04801174418483138,
      "grad_norm": 0.2580173909664154,
      "learning_rate": 0.00029279832774324816,
      "loss": 3.0138995361328127,
      "step": 81550
    },
    {
      "epoch": 0.04804118118310534,
      "grad_norm": 0.21448089182376862,
      "learning_rate": 0.000292793912194157,
      "loss": 2.9644195556640627,
      "step": 81600
    },
    {
      "epoch": 0.0480706181813793,
      "grad_norm": 0.3973119854927063,
      "learning_rate": 0.0002927894966450658,
      "loss": 2.9648318481445313,
      "step": 81650
    },
    {
      "epoch": 0.04810005517965327,
      "grad_norm": 0.28987178206443787,
      "learning_rate": 0.00029278508109597455,
      "loss": 2.9612738037109376,
      "step": 81700
    },
    {
      "epoch": 0.048129492177927224,
      "grad_norm": 0.2563014030456543,
      "learning_rate": 0.00029278066554688336,
      "loss": 2.944016418457031,
      "step": 81750
    },
    {
      "epoch": 0.04815892917620119,
      "grad_norm": 0.22644810378551483,
      "learning_rate": 0.00029277624999779223,
      "loss": 2.9769744873046875,
      "step": 81800
    },
    {
      "epoch": 0.048188366174475146,
      "grad_norm": 0.2880174219608307,
      "learning_rate": 0.000292771834448701,
      "loss": 2.974004821777344,
      "step": 81850
    },
    {
      "epoch": 0.04821780317274911,
      "grad_norm": 0.3161300718784332,
      "learning_rate": 0.0002927674188996098,
      "loss": 2.9444390869140626,
      "step": 81900
    },
    {
      "epoch": 0.048247240171023074,
      "grad_norm": 0.2476724088191986,
      "learning_rate": 0.0002927630033505186,
      "loss": 2.9837161254882814,
      "step": 81950
    },
    {
      "epoch": 0.04827667716929703,
      "grad_norm": 0.4486261308193207,
      "learning_rate": 0.00029275858780142744,
      "loss": 2.967401428222656,
      "step": 82000
    },
    {
      "epoch": 0.048306114167570996,
      "grad_norm": 0.2794719636440277,
      "learning_rate": 0.00029275417225233625,
      "loss": 2.9778521728515623,
      "step": 82050
    },
    {
      "epoch": 0.04833555116584496,
      "grad_norm": 0.30944472551345825,
      "learning_rate": 0.00029274975670324507,
      "loss": 2.998382568359375,
      "step": 82100
    },
    {
      "epoch": 0.04836498816411892,
      "grad_norm": 0.24193593859672546,
      "learning_rate": 0.0002927453411541539,
      "loss": 2.93600830078125,
      "step": 82150
    },
    {
      "epoch": 0.04839442516239288,
      "grad_norm": 0.33992770314216614,
      "learning_rate": 0.0002927409256050627,
      "loss": 2.926865539550781,
      "step": 82200
    },
    {
      "epoch": 0.04842386216066684,
      "grad_norm": 0.2726367712020874,
      "learning_rate": 0.00029273651005597146,
      "loss": 2.947579345703125,
      "step": 82250
    },
    {
      "epoch": 0.0484532991589408,
      "grad_norm": 0.3763847351074219,
      "learning_rate": 0.00029273209450688027,
      "loss": 2.912745056152344,
      "step": 82300
    },
    {
      "epoch": 0.04848273615721477,
      "grad_norm": 0.2556591033935547,
      "learning_rate": 0.0002927276789577891,
      "loss": 2.9957260131835937,
      "step": 82350
    },
    {
      "epoch": 0.048512173155488725,
      "grad_norm": 0.3187257945537567,
      "learning_rate": 0.0002927232634086979,
      "loss": 2.91492431640625,
      "step": 82400
    },
    {
      "epoch": 0.04854161015376269,
      "grad_norm": 0.3243815302848816,
      "learning_rate": 0.0002927188478596067,
      "loss": 2.9083673095703126,
      "step": 82450
    },
    {
      "epoch": 0.04857104715203665,
      "grad_norm": 0.32996460795402527,
      "learning_rate": 0.00029271443231051553,
      "loss": 2.9864056396484373,
      "step": 82500
    },
    {
      "epoch": 0.04860048415031061,
      "grad_norm": 0.30205851793289185,
      "learning_rate": 0.00029271001676142434,
      "loss": 2.9861544799804687,
      "step": 82550
    },
    {
      "epoch": 0.048629921148584575,
      "grad_norm": 0.2480214238166809,
      "learning_rate": 0.00029270560121233316,
      "loss": 2.969683837890625,
      "step": 82600
    },
    {
      "epoch": 0.04865935814685853,
      "grad_norm": 0.25407764315605164,
      "learning_rate": 0.0002927011856632419,
      "loss": 3.037534484863281,
      "step": 82650
    },
    {
      "epoch": 0.048688795145132496,
      "grad_norm": 0.2632964253425598,
      "learning_rate": 0.00029269677011415073,
      "loss": 2.9775460815429686,
      "step": 82700
    },
    {
      "epoch": 0.04871823214340646,
      "grad_norm": 0.24897678196430206,
      "learning_rate": 0.00029269235456505955,
      "loss": 2.9816357421875,
      "step": 82750
    },
    {
      "epoch": 0.04874766914168042,
      "grad_norm": 0.2605692744255066,
      "learning_rate": 0.00029268793901596836,
      "loss": 3.0142535400390624,
      "step": 82800
    },
    {
      "epoch": 0.04877710613995438,
      "grad_norm": 0.2883355915546417,
      "learning_rate": 0.0002926835234668772,
      "loss": 2.9685308837890627,
      "step": 82850
    },
    {
      "epoch": 0.048806543138228346,
      "grad_norm": 0.26489001512527466,
      "learning_rate": 0.000292679107917786,
      "loss": 2.908830261230469,
      "step": 82900
    },
    {
      "epoch": 0.048835980136502304,
      "grad_norm": 0.36556699872016907,
      "learning_rate": 0.0002926746923686948,
      "loss": 2.9825921630859376,
      "step": 82950
    },
    {
      "epoch": 0.04886541713477627,
      "grad_norm": 0.31108492612838745,
      "learning_rate": 0.00029267027681960356,
      "loss": 2.9773483276367188,
      "step": 83000
    },
    {
      "epoch": 0.048894854133050225,
      "grad_norm": 0.2582406997680664,
      "learning_rate": 0.0002926658612705124,
      "loss": 2.955047607421875,
      "step": 83050
    },
    {
      "epoch": 0.04892429113132419,
      "grad_norm": 0.3360832929611206,
      "learning_rate": 0.00029266144572142125,
      "loss": 2.9778286743164064,
      "step": 83100
    },
    {
      "epoch": 0.048953728129598154,
      "grad_norm": 0.2840680778026581,
      "learning_rate": 0.00029265703017233,
      "loss": 2.9597683715820313,
      "step": 83150
    },
    {
      "epoch": 0.04898316512787211,
      "grad_norm": 0.26842156052589417,
      "learning_rate": 0.0002926526146232388,
      "loss": 2.937070617675781,
      "step": 83200
    },
    {
      "epoch": 0.049012602126146075,
      "grad_norm": 0.24343837797641754,
      "learning_rate": 0.00029264819907414764,
      "loss": 2.988653259277344,
      "step": 83250
    },
    {
      "epoch": 0.04904203912442004,
      "grad_norm": 0.26237455010414124,
      "learning_rate": 0.00029264378352505645,
      "loss": 2.957228698730469,
      "step": 83300
    },
    {
      "epoch": 0.049071476122694,
      "grad_norm": 0.3895970284938812,
      "learning_rate": 0.00029263936797596527,
      "loss": 2.9419122314453126,
      "step": 83350
    },
    {
      "epoch": 0.04910091312096796,
      "grad_norm": 0.2943693697452545,
      "learning_rate": 0.0002926349524268741,
      "loss": 2.977852478027344,
      "step": 83400
    },
    {
      "epoch": 0.04913035011924192,
      "grad_norm": 0.3738647997379303,
      "learning_rate": 0.0002926305368777829,
      "loss": 2.9925506591796873,
      "step": 83450
    },
    {
      "epoch": 0.04915978711751588,
      "grad_norm": 0.42814943194389343,
      "learning_rate": 0.0002926261213286917,
      "loss": 2.9141714477539065,
      "step": 83500
    },
    {
      "epoch": 0.04918922411578985,
      "grad_norm": 0.19101515412330627,
      "learning_rate": 0.00029262170577960047,
      "loss": 2.9390243530273437,
      "step": 83550
    },
    {
      "epoch": 0.049218661114063804,
      "grad_norm": 0.31201308965682983,
      "learning_rate": 0.0002926172902305093,
      "loss": 2.93698974609375,
      "step": 83600
    },
    {
      "epoch": 0.04924809811233777,
      "grad_norm": 0.2756642997264862,
      "learning_rate": 0.0002926128746814181,
      "loss": 2.9218453979492187,
      "step": 83650
    },
    {
      "epoch": 0.049277535110611725,
      "grad_norm": 0.27311795949935913,
      "learning_rate": 0.0002926084591323269,
      "loss": 2.93449951171875,
      "step": 83700
    },
    {
      "epoch": 0.04930697210888569,
      "grad_norm": 0.21413497626781464,
      "learning_rate": 0.0002926040435832357,
      "loss": 3.016335754394531,
      "step": 83750
    },
    {
      "epoch": 0.049336409107159654,
      "grad_norm": 0.29768773913383484,
      "learning_rate": 0.00029259962803414454,
      "loss": 2.9495721435546876,
      "step": 83800
    },
    {
      "epoch": 0.04936584610543361,
      "grad_norm": 0.3443306088447571,
      "learning_rate": 0.00029259521248505336,
      "loss": 2.9328790283203126,
      "step": 83850
    },
    {
      "epoch": 0.049395283103707575,
      "grad_norm": 0.25519657135009766,
      "learning_rate": 0.00029259079693596217,
      "loss": 2.8531866455078125,
      "step": 83900
    },
    {
      "epoch": 0.04942472010198154,
      "grad_norm": 0.20017831027507782,
      "learning_rate": 0.00029258638138687093,
      "loss": 3.042585754394531,
      "step": 83950
    },
    {
      "epoch": 0.0494541571002555,
      "grad_norm": 0.2370505928993225,
      "learning_rate": 0.00029258196583777975,
      "loss": 2.8905209350585936,
      "step": 84000
    },
    {
      "epoch": 0.04948359409852946,
      "grad_norm": 0.3379080891609192,
      "learning_rate": 0.0002925775502886886,
      "loss": 3.0082574462890626,
      "step": 84050
    },
    {
      "epoch": 0.04951303109680342,
      "grad_norm": 0.29552969336509705,
      "learning_rate": 0.0002925731347395974,
      "loss": 2.9605560302734375,
      "step": 84100
    },
    {
      "epoch": 0.04954246809507738,
      "grad_norm": 0.2943541705608368,
      "learning_rate": 0.0002925687191905062,
      "loss": 2.9753768920898436,
      "step": 84150
    },
    {
      "epoch": 0.04957190509335135,
      "grad_norm": 0.3597939908504486,
      "learning_rate": 0.000292564303641415,
      "loss": 2.9936737060546874,
      "step": 84200
    },
    {
      "epoch": 0.049601342091625304,
      "grad_norm": 0.3198048174381256,
      "learning_rate": 0.0002925598880923238,
      "loss": 2.9708514404296875,
      "step": 84250
    },
    {
      "epoch": 0.04963077908989927,
      "grad_norm": 0.3015156090259552,
      "learning_rate": 0.00029255547254323263,
      "loss": 2.9196588134765626,
      "step": 84300
    },
    {
      "epoch": 0.04966021608817323,
      "grad_norm": 0.31474488973617554,
      "learning_rate": 0.0002925510569941414,
      "loss": 2.9856991577148437,
      "step": 84350
    },
    {
      "epoch": 0.04968965308644719,
      "grad_norm": 0.414347380399704,
      "learning_rate": 0.00029254664144505026,
      "loss": 2.9464422607421876,
      "step": 84400
    },
    {
      "epoch": 0.049719090084721154,
      "grad_norm": 0.25020405650138855,
      "learning_rate": 0.000292542225895959,
      "loss": 2.898833923339844,
      "step": 84450
    },
    {
      "epoch": 0.04974852708299511,
      "grad_norm": 0.34470418095588684,
      "learning_rate": 0.00029253781034686784,
      "loss": 2.9942666625976564,
      "step": 84500
    },
    {
      "epoch": 0.049777964081269076,
      "grad_norm": 0.29589521884918213,
      "learning_rate": 0.00029253339479777665,
      "loss": 2.894120788574219,
      "step": 84550
    },
    {
      "epoch": 0.04980740107954304,
      "grad_norm": 0.2800557017326355,
      "learning_rate": 0.00029252897924868546,
      "loss": 2.974844665527344,
      "step": 84600
    },
    {
      "epoch": 0.049836838077817,
      "grad_norm": 0.27978989481925964,
      "learning_rate": 0.0002925245636995943,
      "loss": 2.9438079833984374,
      "step": 84650
    },
    {
      "epoch": 0.04986627507609096,
      "grad_norm": 0.22832803428173065,
      "learning_rate": 0.0002925201481505031,
      "loss": 2.900555419921875,
      "step": 84700
    },
    {
      "epoch": 0.049895712074364926,
      "grad_norm": 0.39695069193840027,
      "learning_rate": 0.0002925157326014119,
      "loss": 2.9414053344726563,
      "step": 84750
    },
    {
      "epoch": 0.04992514907263888,
      "grad_norm": 0.25611230731010437,
      "learning_rate": 0.0002925113170523207,
      "loss": 3.0224798583984374,
      "step": 84800
    },
    {
      "epoch": 0.04995458607091285,
      "grad_norm": 0.2869162857532501,
      "learning_rate": 0.0002925069015032295,
      "loss": 2.9617489624023436,
      "step": 84850
    },
    {
      "epoch": 0.049984023069186805,
      "grad_norm": 0.20533691346645355,
      "learning_rate": 0.0002925024859541383,
      "loss": 2.9758145141601564,
      "step": 84900
    },
    {
      "epoch": 0.05001346006746077,
      "grad_norm": 0.2298811972141266,
      "learning_rate": 0.0002924980704050471,
      "loss": 2.9149267578125,
      "step": 84950
    },
    {
      "epoch": 0.05004289706573473,
      "grad_norm": 0.30331578850746155,
      "learning_rate": 0.0002924936548559559,
      "loss": 2.9144320678710938,
      "step": 85000
    },
    {
      "epoch": 0.05007233406400869,
      "grad_norm": 0.312151700258255,
      "learning_rate": 0.00029248923930686474,
      "loss": 2.9705044555664064,
      "step": 85050
    },
    {
      "epoch": 0.050101771062282655,
      "grad_norm": 0.3626791834831238,
      "learning_rate": 0.00029248482375777356,
      "loss": 2.9501077270507814,
      "step": 85100
    },
    {
      "epoch": 0.05013120806055662,
      "grad_norm": 0.29708653688430786,
      "learning_rate": 0.00029248040820868237,
      "loss": 2.9537448120117187,
      "step": 85150
    },
    {
      "epoch": 0.050160645058830576,
      "grad_norm": 0.27152276039123535,
      "learning_rate": 0.0002924759926595912,
      "loss": 3.0012890625,
      "step": 85200
    },
    {
      "epoch": 0.05019008205710454,
      "grad_norm": 0.35273271799087524,
      "learning_rate": 0.00029247157711049994,
      "loss": 2.9881802368164063,
      "step": 85250
    },
    {
      "epoch": 0.0502195190553785,
      "grad_norm": 0.312522292137146,
      "learning_rate": 0.00029246716156140876,
      "loss": 2.944804992675781,
      "step": 85300
    },
    {
      "epoch": 0.05024895605365246,
      "grad_norm": 0.27220359444618225,
      "learning_rate": 0.00029246274601231763,
      "loss": 2.9434432983398438,
      "step": 85350
    },
    {
      "epoch": 0.050278393051926426,
      "grad_norm": 0.2619008421897888,
      "learning_rate": 0.0002924583304632264,
      "loss": 2.944793395996094,
      "step": 85400
    },
    {
      "epoch": 0.05030783005020038,
      "grad_norm": 0.33514636754989624,
      "learning_rate": 0.0002924539149141352,
      "loss": 2.955124816894531,
      "step": 85450
    },
    {
      "epoch": 0.05033726704847435,
      "grad_norm": 0.4167194366455078,
      "learning_rate": 0.000292449499365044,
      "loss": 2.990211181640625,
      "step": 85500
    },
    {
      "epoch": 0.050366704046748305,
      "grad_norm": 0.22903111577033997,
      "learning_rate": 0.00029244508381595283,
      "loss": 2.948690185546875,
      "step": 85550
    },
    {
      "epoch": 0.05039614104502227,
      "grad_norm": 0.5079652667045593,
      "learning_rate": 0.00029244066826686165,
      "loss": 2.8560031127929686,
      "step": 85600
    },
    {
      "epoch": 0.05042557804329623,
      "grad_norm": 0.2469472438097,
      "learning_rate": 0.0002924362527177704,
      "loss": 2.9811260986328123,
      "step": 85650
    },
    {
      "epoch": 0.05045501504157019,
      "grad_norm": 0.25742775201797485,
      "learning_rate": 0.0002924318371686793,
      "loss": 2.9283062744140627,
      "step": 85700
    },
    {
      "epoch": 0.050484452039844155,
      "grad_norm": 0.22667363286018372,
      "learning_rate": 0.0002924274216195881,
      "loss": 3.01027099609375,
      "step": 85750
    },
    {
      "epoch": 0.05051388903811812,
      "grad_norm": 0.3445165157318115,
      "learning_rate": 0.00029242300607049685,
      "loss": 2.942438049316406,
      "step": 85800
    },
    {
      "epoch": 0.050543326036392076,
      "grad_norm": 0.24224728345870972,
      "learning_rate": 0.00029241859052140566,
      "loss": 2.9340374755859373,
      "step": 85850
    },
    {
      "epoch": 0.05057276303466604,
      "grad_norm": 0.3792750835418701,
      "learning_rate": 0.0002924141749723145,
      "loss": 2.9525750732421874,
      "step": 85900
    },
    {
      "epoch": 0.05060220003294,
      "grad_norm": 0.23307561874389648,
      "learning_rate": 0.0002924097594232233,
      "loss": 2.937311096191406,
      "step": 85950
    },
    {
      "epoch": 0.05063163703121396,
      "grad_norm": 0.3369002938270569,
      "learning_rate": 0.0002924053438741321,
      "loss": 2.9430581665039064,
      "step": 86000
    },
    {
      "epoch": 0.050661074029487926,
      "grad_norm": 0.3130466639995575,
      "learning_rate": 0.0002924009283250409,
      "loss": 2.909884338378906,
      "step": 86050
    },
    {
      "epoch": 0.050690511027761884,
      "grad_norm": 0.2659168541431427,
      "learning_rate": 0.00029239651277594974,
      "loss": 3.0013687133789064,
      "step": 86100
    },
    {
      "epoch": 0.05071994802603585,
      "grad_norm": 0.3349079489707947,
      "learning_rate": 0.0002923920972268585,
      "loss": 2.9793328857421875,
      "step": 86150
    },
    {
      "epoch": 0.05074938502430981,
      "grad_norm": 0.32728347182273865,
      "learning_rate": 0.0002923876816777673,
      "loss": 2.921260986328125,
      "step": 86200
    },
    {
      "epoch": 0.05077882202258377,
      "grad_norm": 0.21240927278995514,
      "learning_rate": 0.0002923832661286761,
      "loss": 2.970738830566406,
      "step": 86250
    },
    {
      "epoch": 0.050808259020857734,
      "grad_norm": 0.2562626302242279,
      "learning_rate": 0.00029237885057958494,
      "loss": 2.9197894287109376,
      "step": 86300
    },
    {
      "epoch": 0.05083769601913169,
      "grad_norm": 0.2946348786354065,
      "learning_rate": 0.00029237443503049375,
      "loss": 2.9017364501953127,
      "step": 86350
    },
    {
      "epoch": 0.050867133017405655,
      "grad_norm": 0.2768012583255768,
      "learning_rate": 0.00029237001948140257,
      "loss": 2.9839459228515626,
      "step": 86400
    },
    {
      "epoch": 0.05089657001567962,
      "grad_norm": 0.4320785701274872,
      "learning_rate": 0.0002923656039323114,
      "loss": 2.9494708251953123,
      "step": 86450
    },
    {
      "epoch": 0.05092600701395358,
      "grad_norm": 0.2522224485874176,
      "learning_rate": 0.0002923611883832202,
      "loss": 2.9537869262695313,
      "step": 86500
    },
    {
      "epoch": 0.05095544401222754,
      "grad_norm": 0.21967104077339172,
      "learning_rate": 0.00029235677283412896,
      "loss": 2.9561062622070313,
      "step": 86550
    },
    {
      "epoch": 0.050984881010501505,
      "grad_norm": 0.33762866258621216,
      "learning_rate": 0.0002923523572850378,
      "loss": 2.9157757568359375,
      "step": 86600
    },
    {
      "epoch": 0.05101431800877546,
      "grad_norm": 0.42869263887405396,
      "learning_rate": 0.00029234794173594664,
      "loss": 2.94093017578125,
      "step": 86650
    },
    {
      "epoch": 0.05104375500704943,
      "grad_norm": 0.3339843153953552,
      "learning_rate": 0.0002923435261868554,
      "loss": 2.9453521728515626,
      "step": 86700
    },
    {
      "epoch": 0.051073192005323384,
      "grad_norm": 0.2426697313785553,
      "learning_rate": 0.0002923391106377642,
      "loss": 2.9700567626953127,
      "step": 86750
    },
    {
      "epoch": 0.05110262900359735,
      "grad_norm": 0.3350450098514557,
      "learning_rate": 0.00029233469508867303,
      "loss": 3.005387878417969,
      "step": 86800
    },
    {
      "epoch": 0.05113206600187131,
      "grad_norm": 0.22678141295909882,
      "learning_rate": 0.00029233027953958185,
      "loss": 2.900238342285156,
      "step": 86850
    },
    {
      "epoch": 0.05116150300014527,
      "grad_norm": 0.246744304895401,
      "learning_rate": 0.00029232586399049066,
      "loss": 2.943045654296875,
      "step": 86900
    },
    {
      "epoch": 0.051190939998419234,
      "grad_norm": 0.3472335934638977,
      "learning_rate": 0.0002923214484413994,
      "loss": 2.9365496826171875,
      "step": 86950
    },
    {
      "epoch": 0.0512203769966932,
      "grad_norm": 0.23234406113624573,
      "learning_rate": 0.0002923170328923083,
      "loss": 2.928248291015625,
      "step": 87000
    },
    {
      "epoch": 0.051249813994967156,
      "grad_norm": 0.3192160129547119,
      "learning_rate": 0.0002923126173432171,
      "loss": 2.9386871337890623,
      "step": 87050
    },
    {
      "epoch": 0.05127925099324112,
      "grad_norm": 0.4706469178199768,
      "learning_rate": 0.00029230820179412586,
      "loss": 2.953164367675781,
      "step": 87100
    },
    {
      "epoch": 0.05130868799151508,
      "grad_norm": 0.3058129549026489,
      "learning_rate": 0.0002923037862450347,
      "loss": 2.966446533203125,
      "step": 87150
    },
    {
      "epoch": 0.05133812498978904,
      "grad_norm": 0.27995187044143677,
      "learning_rate": 0.0002922993706959435,
      "loss": 2.9163787841796873,
      "step": 87200
    },
    {
      "epoch": 0.051367561988063005,
      "grad_norm": 0.2789682447910309,
      "learning_rate": 0.0002922949551468523,
      "loss": 2.93988037109375,
      "step": 87250
    },
    {
      "epoch": 0.05139699898633696,
      "grad_norm": 0.3135078251361847,
      "learning_rate": 0.0002922905395977611,
      "loss": 2.9958511352539063,
      "step": 87300
    },
    {
      "epoch": 0.05142643598461093,
      "grad_norm": 0.29306963086128235,
      "learning_rate": 0.00029228612404866994,
      "loss": 2.9188665771484374,
      "step": 87350
    },
    {
      "epoch": 0.051455872982884884,
      "grad_norm": 0.23034876585006714,
      "learning_rate": 0.00029228170849957875,
      "loss": 2.9100607299804686,
      "step": 87400
    },
    {
      "epoch": 0.05148530998115885,
      "grad_norm": 0.2243877351284027,
      "learning_rate": 0.00029227729295048757,
      "loss": 2.9456195068359374,
      "step": 87450
    },
    {
      "epoch": 0.05151474697943281,
      "grad_norm": 0.29244929552078247,
      "learning_rate": 0.0002922728774013963,
      "loss": 2.8988327026367187,
      "step": 87500
    },
    {
      "epoch": 0.05154418397770677,
      "grad_norm": 0.23942390084266663,
      "learning_rate": 0.00029226846185230514,
      "loss": 2.995950927734375,
      "step": 87550
    },
    {
      "epoch": 0.051573620975980734,
      "grad_norm": 0.2539355456829071,
      "learning_rate": 0.00029226404630321395,
      "loss": 2.948247985839844,
      "step": 87600
    },
    {
      "epoch": 0.0516030579742547,
      "grad_norm": 0.2691596448421478,
      "learning_rate": 0.00029225963075412277,
      "loss": 2.9238662719726562,
      "step": 87650
    },
    {
      "epoch": 0.051632494972528656,
      "grad_norm": 0.25949040055274963,
      "learning_rate": 0.0002922552152050316,
      "loss": 2.992850341796875,
      "step": 87700
    },
    {
      "epoch": 0.05166193197080262,
      "grad_norm": 0.25595927238464355,
      "learning_rate": 0.0002922507996559404,
      "loss": 2.96142578125,
      "step": 87750
    },
    {
      "epoch": 0.05169136896907658,
      "grad_norm": 0.2600817084312439,
      "learning_rate": 0.0002922463841068492,
      "loss": 2.924999084472656,
      "step": 87800
    },
    {
      "epoch": 0.05172080596735054,
      "grad_norm": 0.26991531252861023,
      "learning_rate": 0.00029224196855775797,
      "loss": 2.935982971191406,
      "step": 87850
    },
    {
      "epoch": 0.051750242965624506,
      "grad_norm": 0.36762160062789917,
      "learning_rate": 0.0002922375530086668,
      "loss": 2.9544467163085937,
      "step": 87900
    },
    {
      "epoch": 0.05177967996389846,
      "grad_norm": 0.35206833481788635,
      "learning_rate": 0.0002922331374595756,
      "loss": 2.94233642578125,
      "step": 87950
    },
    {
      "epoch": 0.05180911696217243,
      "grad_norm": 0.3331246078014374,
      "learning_rate": 0.0002922287219104844,
      "loss": 2.94623291015625,
      "step": 88000
    },
    {
      "epoch": 0.05183855396044639,
      "grad_norm": 0.37663733959198,
      "learning_rate": 0.00029222430636139323,
      "loss": 3.005640563964844,
      "step": 88050
    },
    {
      "epoch": 0.05186799095872035,
      "grad_norm": 0.20747141540050507,
      "learning_rate": 0.00029221989081230205,
      "loss": 2.9887371826171876,
      "step": 88100
    },
    {
      "epoch": 0.05189742795699431,
      "grad_norm": 0.3106408715248108,
      "learning_rate": 0.00029221547526321086,
      "loss": 2.887839050292969,
      "step": 88150
    },
    {
      "epoch": 0.05192686495526827,
      "grad_norm": 0.30665263533592224,
      "learning_rate": 0.0002922110597141197,
      "loss": 2.996617431640625,
      "step": 88200
    },
    {
      "epoch": 0.051956301953542235,
      "grad_norm": 0.32487088441848755,
      "learning_rate": 0.00029220664416502843,
      "loss": 2.9024658203125,
      "step": 88250
    },
    {
      "epoch": 0.0519857389518162,
      "grad_norm": 0.24441306293010712,
      "learning_rate": 0.0002922022286159373,
      "loss": 2.939298400878906,
      "step": 88300
    },
    {
      "epoch": 0.052015175950090156,
      "grad_norm": 0.32300248742103577,
      "learning_rate": 0.0002921978130668461,
      "loss": 2.9513092041015625,
      "step": 88350
    },
    {
      "epoch": 0.05204461294836412,
      "grad_norm": 0.2746129035949707,
      "learning_rate": 0.0002921933975177549,
      "loss": 2.8864739990234374,
      "step": 88400
    },
    {
      "epoch": 0.052074049946638085,
      "grad_norm": 0.25321510434150696,
      "learning_rate": 0.0002921889819686637,
      "loss": 2.94298095703125,
      "step": 88450
    },
    {
      "epoch": 0.05210348694491204,
      "grad_norm": 0.3212844431400299,
      "learning_rate": 0.0002921845664195725,
      "loss": 2.9548416137695312,
      "step": 88500
    },
    {
      "epoch": 0.052132923943186006,
      "grad_norm": 0.2471989542245865,
      "learning_rate": 0.0002921801508704813,
      "loss": 2.9961737060546874,
      "step": 88550
    },
    {
      "epoch": 0.05216236094145996,
      "grad_norm": 0.3091031610965729,
      "learning_rate": 0.00029217573532139014,
      "loss": 2.9969122314453127,
      "step": 88600
    },
    {
      "epoch": 0.05219179793973393,
      "grad_norm": 0.30655723810195923,
      "learning_rate": 0.00029217131977229895,
      "loss": 2.9575650024414064,
      "step": 88650
    },
    {
      "epoch": 0.05222123493800789,
      "grad_norm": 0.2866396903991699,
      "learning_rate": 0.00029216690422320776,
      "loss": 2.9047805786132814,
      "step": 88700
    },
    {
      "epoch": 0.05225067193628185,
      "grad_norm": 0.2766111195087433,
      "learning_rate": 0.0002921624886741166,
      "loss": 2.8997100830078124,
      "step": 88750
    },
    {
      "epoch": 0.05228010893455581,
      "grad_norm": 0.29645437002182007,
      "learning_rate": 0.00029215807312502534,
      "loss": 2.9391323852539064,
      "step": 88800
    },
    {
      "epoch": 0.05230954593282978,
      "grad_norm": 0.41252416372299194,
      "learning_rate": 0.00029215365757593415,
      "loss": 2.9579098510742186,
      "step": 88850
    },
    {
      "epoch": 0.052338982931103735,
      "grad_norm": 0.31949400901794434,
      "learning_rate": 0.00029214924202684297,
      "loss": 2.9312783813476564,
      "step": 88900
    },
    {
      "epoch": 0.0523684199293777,
      "grad_norm": 0.2556132972240448,
      "learning_rate": 0.0002921448264777518,
      "loss": 2.9762359619140626,
      "step": 88950
    },
    {
      "epoch": 0.052397856927651656,
      "grad_norm": 0.24869070947170258,
      "learning_rate": 0.0002921404109286606,
      "loss": 2.96077392578125,
      "step": 89000
    },
    {
      "epoch": 0.05242729392592562,
      "grad_norm": 0.23222245275974274,
      "learning_rate": 0.0002921359953795694,
      "loss": 2.939600524902344,
      "step": 89050
    },
    {
      "epoch": 0.052456730924199585,
      "grad_norm": 0.2591666579246521,
      "learning_rate": 0.0002921315798304782,
      "loss": 2.962555236816406,
      "step": 89100
    },
    {
      "epoch": 0.05248616792247354,
      "grad_norm": 0.25837743282318115,
      "learning_rate": 0.00029212716428138704,
      "loss": 2.944260559082031,
      "step": 89150
    },
    {
      "epoch": 0.052515604920747506,
      "grad_norm": 0.3859986960887909,
      "learning_rate": 0.0002921227487322958,
      "loss": 2.9193209838867187,
      "step": 89200
    },
    {
      "epoch": 0.052545041919021464,
      "grad_norm": 0.24505789577960968,
      "learning_rate": 0.0002921183331832046,
      "loss": 2.9465008544921876,
      "step": 89250
    },
    {
      "epoch": 0.05257447891729543,
      "grad_norm": 0.26077038049697876,
      "learning_rate": 0.0002921139176341135,
      "loss": 2.921636962890625,
      "step": 89300
    },
    {
      "epoch": 0.05260391591556939,
      "grad_norm": 0.2892313003540039,
      "learning_rate": 0.00029210950208502224,
      "loss": 2.934742126464844,
      "step": 89350
    },
    {
      "epoch": 0.05263335291384335,
      "grad_norm": 0.29388564825057983,
      "learning_rate": 0.00029210508653593106,
      "loss": 2.9315640258789064,
      "step": 89400
    },
    {
      "epoch": 0.052662789912117314,
      "grad_norm": 0.30763930082321167,
      "learning_rate": 0.0002921006709868399,
      "loss": 2.952646484375,
      "step": 89450
    },
    {
      "epoch": 0.05269222691039128,
      "grad_norm": 0.39084604382514954,
      "learning_rate": 0.0002920962554377487,
      "loss": 2.928721618652344,
      "step": 89500
    },
    {
      "epoch": 0.052721663908665235,
      "grad_norm": 0.3209764063358307,
      "learning_rate": 0.00029209183988865745,
      "loss": 2.9445562744140625,
      "step": 89550
    },
    {
      "epoch": 0.0527511009069392,
      "grad_norm": 0.42515891790390015,
      "learning_rate": 0.0002920874243395663,
      "loss": 2.9504461669921875,
      "step": 89600
    },
    {
      "epoch": 0.05278053790521316,
      "grad_norm": 0.2548574209213257,
      "learning_rate": 0.00029208300879047513,
      "loss": 2.9867984008789064,
      "step": 89650
    },
    {
      "epoch": 0.05280997490348712,
      "grad_norm": 0.32117709517478943,
      "learning_rate": 0.0002920785932413839,
      "loss": 2.985426940917969,
      "step": 89700
    },
    {
      "epoch": 0.052839411901761085,
      "grad_norm": 0.3378082513809204,
      "learning_rate": 0.0002920741776922927,
      "loss": 2.9963568115234374,
      "step": 89750
    },
    {
      "epoch": 0.05286884890003504,
      "grad_norm": 0.2786538600921631,
      "learning_rate": 0.0002920697621432015,
      "loss": 2.970447998046875,
      "step": 89800
    },
    {
      "epoch": 0.05289828589830901,
      "grad_norm": 0.35740339756011963,
      "learning_rate": 0.00029206534659411034,
      "loss": 2.956650390625,
      "step": 89850
    },
    {
      "epoch": 0.05292772289658297,
      "grad_norm": Infinity,
      "learning_rate": 0.00029206093104501915,
      "loss": 2.9240231323242187,
      "step": 89900
    },
    {
      "epoch": 0.05295715989485693,
      "grad_norm": 0.26656511425971985,
      "learning_rate": 0.00029205651549592796,
      "loss": 2.9347805786132812,
      "step": 89950
    },
    {
      "epoch": 0.05298659689313089,
      "grad_norm": 0.4516686499118805,
      "learning_rate": 0.0002920520999468368,
      "loss": 2.9235964965820314,
      "step": 90000
    },
    {
      "epoch": 0.05301603389140485,
      "grad_norm": 0.26306256651878357,
      "learning_rate": 0.0002920476843977456,
      "loss": 2.9584619140625,
      "step": 90050
    },
    {
      "epoch": 0.053045470889678814,
      "grad_norm": 0.3248157203197479,
      "learning_rate": 0.00029204326884865435,
      "loss": 2.981797790527344,
      "step": 90100
    },
    {
      "epoch": 0.05307490788795278,
      "grad_norm": 0.21969804167747498,
      "learning_rate": 0.00029203885329956317,
      "loss": 2.925751953125,
      "step": 90150
    },
    {
      "epoch": 0.053104344886226736,
      "grad_norm": 0.27113398909568787,
      "learning_rate": 0.000292034437750472,
      "loss": 2.8662481689453125,
      "step": 90200
    },
    {
      "epoch": 0.0531337818845007,
      "grad_norm": 0.27786555886268616,
      "learning_rate": 0.0002920300222013808,
      "loss": 2.9723025512695314,
      "step": 90250
    },
    {
      "epoch": 0.053163218882774664,
      "grad_norm": 0.26270046830177307,
      "learning_rate": 0.0002920256066522896,
      "loss": 2.9726702880859377,
      "step": 90300
    },
    {
      "epoch": 0.05319265588104862,
      "grad_norm": 0.22696004807949066,
      "learning_rate": 0.0002920211911031984,
      "loss": 2.899036865234375,
      "step": 90350
    },
    {
      "epoch": 0.053222092879322586,
      "grad_norm": 0.30523061752319336,
      "learning_rate": 0.00029201677555410724,
      "loss": 2.9045648193359375,
      "step": 90400
    },
    {
      "epoch": 0.05325152987759654,
      "grad_norm": 0.26209205389022827,
      "learning_rate": 0.00029201236000501605,
      "loss": 2.9243817138671875,
      "step": 90450
    },
    {
      "epoch": 0.05328096687587051,
      "grad_norm": 0.28337064385414124,
      "learning_rate": 0.0002920079444559248,
      "loss": 2.8657025146484374,
      "step": 90500
    },
    {
      "epoch": 0.05331040387414447,
      "grad_norm": 0.24241769313812256,
      "learning_rate": 0.00029200352890683363,
      "loss": 2.953794250488281,
      "step": 90550
    },
    {
      "epoch": 0.05333984087241843,
      "grad_norm": 0.32688039541244507,
      "learning_rate": 0.0002919991133577425,
      "loss": 2.9025518798828127,
      "step": 90600
    },
    {
      "epoch": 0.05336927787069239,
      "grad_norm": 0.39459100365638733,
      "learning_rate": 0.00029199469780865126,
      "loss": 2.986099853515625,
      "step": 90650
    },
    {
      "epoch": 0.05339871486896636,
      "grad_norm": 0.2655356228351593,
      "learning_rate": 0.0002919902822595601,
      "loss": 2.8599755859375,
      "step": 90700
    },
    {
      "epoch": 0.053428151867240314,
      "grad_norm": 0.28980377316474915,
      "learning_rate": 0.0002919858667104689,
      "loss": 2.9305548095703124,
      "step": 90750
    },
    {
      "epoch": 0.05345758886551428,
      "grad_norm": 0.3074890077114105,
      "learning_rate": 0.0002919814511613777,
      "loss": 2.92408935546875,
      "step": 90800
    },
    {
      "epoch": 0.053487025863788236,
      "grad_norm": 0.2871670126914978,
      "learning_rate": 0.0002919770356122865,
      "loss": 2.924910888671875,
      "step": 90850
    },
    {
      "epoch": 0.0535164628620622,
      "grad_norm": 0.4004918038845062,
      "learning_rate": 0.00029197262006319533,
      "loss": 2.914400634765625,
      "step": 90900
    },
    {
      "epoch": 0.053545899860336164,
      "grad_norm": 0.387874037027359,
      "learning_rate": 0.00029196820451410415,
      "loss": 2.94121337890625,
      "step": 90950
    },
    {
      "epoch": 0.05357533685861012,
      "grad_norm": 0.30118340253829956,
      "learning_rate": 0.00029196378896501296,
      "loss": 2.984588317871094,
      "step": 91000
    },
    {
      "epoch": 0.053604773856884086,
      "grad_norm": 0.2858186364173889,
      "learning_rate": 0.0002919593734159217,
      "loss": 2.9569638061523436,
      "step": 91050
    },
    {
      "epoch": 0.05363421085515804,
      "grad_norm": 0.4211997389793396,
      "learning_rate": 0.00029195495786683053,
      "loss": 2.947149658203125,
      "step": 91100
    },
    {
      "epoch": 0.05366364785343201,
      "grad_norm": 0.2643677890300751,
      "learning_rate": 0.00029195054231773935,
      "loss": 2.9495965576171876,
      "step": 91150
    },
    {
      "epoch": 0.05369308485170597,
      "grad_norm": 0.2486703097820282,
      "learning_rate": 0.00029194612676864816,
      "loss": 2.940249328613281,
      "step": 91200
    },
    {
      "epoch": 0.05372252184997993,
      "grad_norm": 0.26136866211891174,
      "learning_rate": 0.000291941711219557,
      "loss": 2.9572988891601564,
      "step": 91250
    },
    {
      "epoch": 0.05375195884825389,
      "grad_norm": 0.2959657609462738,
      "learning_rate": 0.0002919372956704658,
      "loss": 2.9260769653320313,
      "step": 91300
    },
    {
      "epoch": 0.05378139584652786,
      "grad_norm": 0.31460195779800415,
      "learning_rate": 0.0002919328801213746,
      "loss": 2.927003173828125,
      "step": 91350
    },
    {
      "epoch": 0.053810832844801815,
      "grad_norm": 0.3710183799266815,
      "learning_rate": 0.00029192846457228337,
      "loss": 2.957684326171875,
      "step": 91400
    },
    {
      "epoch": 0.05384026984307578,
      "grad_norm": 0.25648316740989685,
      "learning_rate": 0.0002919240490231922,
      "loss": 2.956380615234375,
      "step": 91450
    },
    {
      "epoch": 0.053869706841349736,
      "grad_norm": 0.25679904222488403,
      "learning_rate": 0.000291919633474101,
      "loss": 2.886138916015625,
      "step": 91500
    },
    {
      "epoch": 0.0538991438396237,
      "grad_norm": 0.25327128171920776,
      "learning_rate": 0.0002919152179250098,
      "loss": 2.981947937011719,
      "step": 91550
    },
    {
      "epoch": 0.053928580837897665,
      "grad_norm": 0.19616246223449707,
      "learning_rate": 0.0002919108023759186,
      "loss": 2.9759487915039062,
      "step": 91600
    },
    {
      "epoch": 0.05395801783617162,
      "grad_norm": 0.21340787410736084,
      "learning_rate": 0.00029190638682682744,
      "loss": 2.925572509765625,
      "step": 91650
    },
    {
      "epoch": 0.053987454834445586,
      "grad_norm": 0.304299920797348,
      "learning_rate": 0.00029190197127773625,
      "loss": 2.9342965698242187,
      "step": 91700
    },
    {
      "epoch": 0.05401689183271955,
      "grad_norm": 0.2772492468357086,
      "learning_rate": 0.00029189755572864507,
      "loss": 2.946837158203125,
      "step": 91750
    },
    {
      "epoch": 0.05404632883099351,
      "grad_norm": 0.35369446873664856,
      "learning_rate": 0.00029189314017955383,
      "loss": 2.982060546875,
      "step": 91800
    },
    {
      "epoch": 0.05407576582926747,
      "grad_norm": 0.26887932419776917,
      "learning_rate": 0.00029188872463046264,
      "loss": 2.8991796875,
      "step": 91850
    },
    {
      "epoch": 0.05410520282754143,
      "grad_norm": 0.24956661462783813,
      "learning_rate": 0.0002918843090813715,
      "loss": 2.8884487915039063,
      "step": 91900
    },
    {
      "epoch": 0.054134639825815394,
      "grad_norm": 0.2308254837989807,
      "learning_rate": 0.00029187989353228027,
      "loss": 2.9433935546875,
      "step": 91950
    },
    {
      "epoch": 0.05416407682408936,
      "grad_norm": 0.4515882730484009,
      "learning_rate": 0.0002918754779831891,
      "loss": 2.884547119140625,
      "step": 92000
    },
    {
      "epoch": 0.054193513822363315,
      "grad_norm": 0.28273168206214905,
      "learning_rate": 0.0002918710624340979,
      "loss": 2.9183694458007814,
      "step": 92050
    },
    {
      "epoch": 0.05422295082063728,
      "grad_norm": 0.24211598932743073,
      "learning_rate": 0.0002918666468850067,
      "loss": 2.92301513671875,
      "step": 92100
    },
    {
      "epoch": 0.054252387818911244,
      "grad_norm": 0.25621163845062256,
      "learning_rate": 0.00029186223133591553,
      "loss": 2.89761962890625,
      "step": 92150
    },
    {
      "epoch": 0.0542818248171852,
      "grad_norm": 0.28832605481147766,
      "learning_rate": 0.00029185781578682434,
      "loss": 2.91585205078125,
      "step": 92200
    },
    {
      "epoch": 0.054311261815459165,
      "grad_norm": 0.2736956477165222,
      "learning_rate": 0.00029185340023773316,
      "loss": 2.9301583862304685,
      "step": 92250
    },
    {
      "epoch": 0.05434069881373312,
      "grad_norm": 0.2639539837837219,
      "learning_rate": 0.000291848984688642,
      "loss": 2.9554486083984375,
      "step": 92300
    },
    {
      "epoch": 0.05437013581200709,
      "grad_norm": 0.24051302671432495,
      "learning_rate": 0.00029184456913955073,
      "loss": 2.884237060546875,
      "step": 92350
    },
    {
      "epoch": 0.05439957281028105,
      "grad_norm": 0.24930138885974884,
      "learning_rate": 0.00029184015359045955,
      "loss": 2.9063204956054687,
      "step": 92400
    },
    {
      "epoch": 0.05442900980855501,
      "grad_norm": 0.31188470125198364,
      "learning_rate": 0.00029183573804136836,
      "loss": 2.924748840332031,
      "step": 92450
    },
    {
      "epoch": 0.05445844680682897,
      "grad_norm": 0.344215452671051,
      "learning_rate": 0.0002918313224922772,
      "loss": 2.89763916015625,
      "step": 92500
    },
    {
      "epoch": 0.05448788380510294,
      "grad_norm": 0.2623194754123688,
      "learning_rate": 0.000291826906943186,
      "loss": 2.9337786865234374,
      "step": 92550
    },
    {
      "epoch": 0.054517320803376894,
      "grad_norm": 0.2372892200946808,
      "learning_rate": 0.0002918224913940948,
      "loss": 3.0001239013671874,
      "step": 92600
    },
    {
      "epoch": 0.05454675780165086,
      "grad_norm": 0.2697855532169342,
      "learning_rate": 0.0002918180758450036,
      "loss": 2.967348937988281,
      "step": 92650
    },
    {
      "epoch": 0.054576194799924815,
      "grad_norm": 0.4307628571987152,
      "learning_rate": 0.00029181366029591244,
      "loss": 2.9137924194335936,
      "step": 92700
    },
    {
      "epoch": 0.05460563179819878,
      "grad_norm": 0.2934668958187103,
      "learning_rate": 0.0002918092447468212,
      "loss": 2.9312973022460938,
      "step": 92750
    },
    {
      "epoch": 0.054635068796472744,
      "grad_norm": 0.3498486578464508,
      "learning_rate": 0.00029180482919773,
      "loss": 2.9344216918945314,
      "step": 92800
    },
    {
      "epoch": 0.0546645057947467,
      "grad_norm": 0.25117963552474976,
      "learning_rate": 0.0002918004136486389,
      "loss": 2.9573358154296874,
      "step": 92850
    },
    {
      "epoch": 0.054693942793020665,
      "grad_norm": 0.302420973777771,
      "learning_rate": 0.00029179599809954764,
      "loss": 2.9258209228515626,
      "step": 92900
    },
    {
      "epoch": 0.05472337979129462,
      "grad_norm": 0.29005810618400574,
      "learning_rate": 0.00029179158255045645,
      "loss": 2.8550140380859377,
      "step": 92950
    },
    {
      "epoch": 0.05475281678956859,
      "grad_norm": 0.29973745346069336,
      "learning_rate": 0.00029178716700136527,
      "loss": 2.932647705078125,
      "step": 93000
    },
    {
      "epoch": 0.05478225378784255,
      "grad_norm": 0.264821320772171,
      "learning_rate": 0.0002917827514522741,
      "loss": 2.9590719604492186,
      "step": 93050
    },
    {
      "epoch": 0.05481169078611651,
      "grad_norm": 0.2389431893825531,
      "learning_rate": 0.00029177833590318284,
      "loss": 2.9279598999023437,
      "step": 93100
    },
    {
      "epoch": 0.05484112778439047,
      "grad_norm": 0.35962745547294617,
      "learning_rate": 0.00029177392035409166,
      "loss": 2.926217346191406,
      "step": 93150
    },
    {
      "epoch": 0.05487056478266444,
      "grad_norm": 0.33367863297462463,
      "learning_rate": 0.0002917695048050005,
      "loss": 2.931853942871094,
      "step": 93200
    },
    {
      "epoch": 0.054900001780938394,
      "grad_norm": 0.3523724675178528,
      "learning_rate": 0.0002917650892559093,
      "loss": 2.9575149536132814,
      "step": 93250
    },
    {
      "epoch": 0.05492943877921236,
      "grad_norm": 0.2586963176727295,
      "learning_rate": 0.0002917606737068181,
      "loss": 2.951387634277344,
      "step": 93300
    },
    {
      "epoch": 0.054958875777486316,
      "grad_norm": 0.2506663501262665,
      "learning_rate": 0.0002917562581577269,
      "loss": 2.935392150878906,
      "step": 93350
    },
    {
      "epoch": 0.05498831277576028,
      "grad_norm": 0.2670501470565796,
      "learning_rate": 0.00029175184260863573,
      "loss": 2.9338873291015624,
      "step": 93400
    },
    {
      "epoch": 0.055017749774034244,
      "grad_norm": 0.25290384888648987,
      "learning_rate": 0.00029174742705954454,
      "loss": 2.913685607910156,
      "step": 93450
    },
    {
      "epoch": 0.0550471867723082,
      "grad_norm": 0.2709985673427582,
      "learning_rate": 0.00029174301151045336,
      "loss": 2.908389892578125,
      "step": 93500
    },
    {
      "epoch": 0.055076623770582166,
      "grad_norm": 0.3638603091239929,
      "learning_rate": 0.0002917385959613622,
      "loss": 2.992056579589844,
      "step": 93550
    },
    {
      "epoch": 0.05510606076885613,
      "grad_norm": 0.2695797085762024,
      "learning_rate": 0.000291734180412271,
      "loss": 2.9759677124023436,
      "step": 93600
    },
    {
      "epoch": 0.05513549776713009,
      "grad_norm": 0.43655550479888916,
      "learning_rate": 0.00029172976486317975,
      "loss": 2.9375177001953126,
      "step": 93650
    },
    {
      "epoch": 0.05516493476540405,
      "grad_norm": 0.260881632566452,
      "learning_rate": 0.00029172534931408856,
      "loss": 2.940554504394531,
      "step": 93700
    },
    {
      "epoch": 0.05519437176367801,
      "grad_norm": 0.2969217300415039,
      "learning_rate": 0.0002917209337649974,
      "loss": 2.9712344360351564,
      "step": 93750
    },
    {
      "epoch": 0.05522380876195197,
      "grad_norm": 0.28551334142684937,
      "learning_rate": 0.0002917165182159062,
      "loss": 2.9339129638671877,
      "step": 93800
    },
    {
      "epoch": 0.05525324576022594,
      "grad_norm": 0.2865505516529083,
      "learning_rate": 0.000291712102666815,
      "loss": 2.9739620971679686,
      "step": 93850
    },
    {
      "epoch": 0.055282682758499894,
      "grad_norm": 0.24665167927742004,
      "learning_rate": 0.0002917076871177238,
      "loss": 2.9297329711914064,
      "step": 93900
    },
    {
      "epoch": 0.05531211975677386,
      "grad_norm": 0.23990610241889954,
      "learning_rate": 0.00029170327156863263,
      "loss": 2.910146484375,
      "step": 93950
    },
    {
      "epoch": 0.05534155675504782,
      "grad_norm": 0.26722779870033264,
      "learning_rate": 0.00029169885601954145,
      "loss": 2.9046279907226564,
      "step": 94000
    },
    {
      "epoch": 0.05537099375332178,
      "grad_norm": 0.25596123933792114,
      "learning_rate": 0.0002916944404704502,
      "loss": 2.95312744140625,
      "step": 94050
    },
    {
      "epoch": 0.055400430751595744,
      "grad_norm": 0.2727380394935608,
      "learning_rate": 0.000291690024921359,
      "loss": 2.936051940917969,
      "step": 94100
    },
    {
      "epoch": 0.0554298677498697,
      "grad_norm": 0.3420192301273346,
      "learning_rate": 0.0002916856093722679,
      "loss": 2.9011553955078124,
      "step": 94150
    },
    {
      "epoch": 0.055459304748143666,
      "grad_norm": 0.225975900888443,
      "learning_rate": 0.00029168119382317665,
      "loss": 2.901436462402344,
      "step": 94200
    },
    {
      "epoch": 0.05548874174641763,
      "grad_norm": 0.2782699465751648,
      "learning_rate": 0.00029167677827408547,
      "loss": 2.9326971435546874,
      "step": 94250
    },
    {
      "epoch": 0.05551817874469159,
      "grad_norm": 0.25087395310401917,
      "learning_rate": 0.0002916723627249943,
      "loss": 2.92835205078125,
      "step": 94300
    },
    {
      "epoch": 0.05554761574296555,
      "grad_norm": 0.2892843186855316,
      "learning_rate": 0.0002916679471759031,
      "loss": 2.909573974609375,
      "step": 94350
    },
    {
      "epoch": 0.05557705274123951,
      "grad_norm": 0.3090110421180725,
      "learning_rate": 0.0002916635316268119,
      "loss": 2.9533087158203126,
      "step": 94400
    },
    {
      "epoch": 0.05560648973951347,
      "grad_norm": 0.2387610524892807,
      "learning_rate": 0.00029165911607772067,
      "loss": 2.9640182495117187,
      "step": 94450
    },
    {
      "epoch": 0.05563592673778744,
      "grad_norm": 0.30772310495376587,
      "learning_rate": 0.00029165470052862954,
      "loss": 2.883038330078125,
      "step": 94500
    },
    {
      "epoch": 0.055665363736061395,
      "grad_norm": 0.27269795536994934,
      "learning_rate": 0.00029165028497953835,
      "loss": 2.90802734375,
      "step": 94550
    },
    {
      "epoch": 0.05569480073433536,
      "grad_norm": 0.3459435999393463,
      "learning_rate": 0.0002916458694304471,
      "loss": 2.9605252075195314,
      "step": 94600
    },
    {
      "epoch": 0.05572423773260932,
      "grad_norm": 0.3549024760723114,
      "learning_rate": 0.00029164145388135593,
      "loss": 2.906956787109375,
      "step": 94650
    },
    {
      "epoch": 0.05575367473088328,
      "grad_norm": 0.3323405683040619,
      "learning_rate": 0.00029163703833226474,
      "loss": 2.9449737548828123,
      "step": 94700
    },
    {
      "epoch": 0.055783111729157245,
      "grad_norm": 0.2931968867778778,
      "learning_rate": 0.00029163262278317356,
      "loss": 2.8860736083984375,
      "step": 94750
    },
    {
      "epoch": 0.0558125487274312,
      "grad_norm": 0.23013806343078613,
      "learning_rate": 0.0002916282072340823,
      "loss": 2.9439166259765623,
      "step": 94800
    },
    {
      "epoch": 0.055841985725705166,
      "grad_norm": 0.28817278146743774,
      "learning_rate": 0.0002916237916849912,
      "loss": 2.901348876953125,
      "step": 94850
    },
    {
      "epoch": 0.05587142272397913,
      "grad_norm": 0.3220099210739136,
      "learning_rate": 0.0002916193761359,
      "loss": 2.937716369628906,
      "step": 94900
    },
    {
      "epoch": 0.05590085972225309,
      "grad_norm": 0.3074128329753876,
      "learning_rate": 0.00029161496058680876,
      "loss": 2.952894287109375,
      "step": 94950
    },
    {
      "epoch": 0.05593029672052705,
      "grad_norm": 0.29753729701042175,
      "learning_rate": 0.0002916105450377176,
      "loss": 2.8969692993164062,
      "step": 95000
    },
    {
      "epoch": 0.055959733718801016,
      "grad_norm": 0.33404800295829773,
      "learning_rate": 0.0002916061294886264,
      "loss": 2.933971252441406,
      "step": 95050
    },
    {
      "epoch": 0.055989170717074974,
      "grad_norm": 0.28240707516670227,
      "learning_rate": 0.0002916017139395352,
      "loss": 2.8682650756835937,
      "step": 95100
    },
    {
      "epoch": 0.05601860771534894,
      "grad_norm": 0.2531830966472626,
      "learning_rate": 0.000291597298390444,
      "loss": 2.9019204711914064,
      "step": 95150
    },
    {
      "epoch": 0.056048044713622895,
      "grad_norm": 0.258685439825058,
      "learning_rate": 0.00029159288284135283,
      "loss": 2.9034002685546874,
      "step": 95200
    },
    {
      "epoch": 0.05607748171189686,
      "grad_norm": 0.3645067512989044,
      "learning_rate": 0.00029158846729226165,
      "loss": 2.9515872192382813,
      "step": 95250
    },
    {
      "epoch": 0.056106918710170824,
      "grad_norm": 0.29123297333717346,
      "learning_rate": 0.00029158405174317046,
      "loss": 2.852480163574219,
      "step": 95300
    },
    {
      "epoch": 0.05613635570844478,
      "grad_norm": 0.38869982957839966,
      "learning_rate": 0.0002915796361940792,
      "loss": 2.8743231201171877,
      "step": 95350
    },
    {
      "epoch": 0.056165792706718745,
      "grad_norm": 0.21799537539482117,
      "learning_rate": 0.00029157522064498804,
      "loss": 2.9302029418945312,
      "step": 95400
    },
    {
      "epoch": 0.05619522970499271,
      "grad_norm": 0.30999961495399475,
      "learning_rate": 0.00029157080509589685,
      "loss": 2.94996826171875,
      "step": 95450
    },
    {
      "epoch": 0.05622466670326667,
      "grad_norm": 0.233930304646492,
      "learning_rate": 0.00029156638954680567,
      "loss": 2.881845703125,
      "step": 95500
    },
    {
      "epoch": 0.05625410370154063,
      "grad_norm": 0.3885948657989502,
      "learning_rate": 0.0002915619739977145,
      "loss": 2.9818429565429687,
      "step": 95550
    },
    {
      "epoch": 0.05628354069981459,
      "grad_norm": 0.25623252987861633,
      "learning_rate": 0.0002915575584486233,
      "loss": 2.947793273925781,
      "step": 95600
    },
    {
      "epoch": 0.05631297769808855,
      "grad_norm": 0.2454005479812622,
      "learning_rate": 0.0002915531428995321,
      "loss": 2.963608703613281,
      "step": 95650
    },
    {
      "epoch": 0.05634241469636252,
      "grad_norm": 0.20989789068698883,
      "learning_rate": 0.0002915487273504409,
      "loss": 2.958271789550781,
      "step": 95700
    },
    {
      "epoch": 0.056371851694636474,
      "grad_norm": 0.29773542284965515,
      "learning_rate": 0.0002915443118013497,
      "loss": 2.880189208984375,
      "step": 95750
    },
    {
      "epoch": 0.05640128869291044,
      "grad_norm": 0.30800727009773254,
      "learning_rate": 0.00029153989625225855,
      "loss": 2.8802105712890627,
      "step": 95800
    },
    {
      "epoch": 0.0564307256911844,
      "grad_norm": 0.26682576537132263,
      "learning_rate": 0.00029153548070316737,
      "loss": 2.9128128051757813,
      "step": 95850
    },
    {
      "epoch": 0.05646016268945836,
      "grad_norm": 0.23872995376586914,
      "learning_rate": 0.00029153106515407613,
      "loss": 2.9186605834960937,
      "step": 95900
    },
    {
      "epoch": 0.056489599687732324,
      "grad_norm": 0.32174408435821533,
      "learning_rate": 0.00029152664960498494,
      "loss": 2.9338278198242187,
      "step": 95950
    },
    {
      "epoch": 0.05651903668600628,
      "grad_norm": 0.23673377931118011,
      "learning_rate": 0.00029152223405589376,
      "loss": 2.903655700683594,
      "step": 96000
    },
    {
      "epoch": 0.056548473684280245,
      "grad_norm": 0.27465513348579407,
      "learning_rate": 0.00029151781850680257,
      "loss": 2.8652609252929686,
      "step": 96050
    },
    {
      "epoch": 0.05657791068255421,
      "grad_norm": 0.32558655738830566,
      "learning_rate": 0.0002915134029577114,
      "loss": 2.896778869628906,
      "step": 96100
    },
    {
      "epoch": 0.05660734768082817,
      "grad_norm": 0.2515106797218323,
      "learning_rate": 0.0002915089874086202,
      "loss": 2.900538024902344,
      "step": 96150
    },
    {
      "epoch": 0.05663678467910213,
      "grad_norm": 0.32516732811927795,
      "learning_rate": 0.000291504571859529,
      "loss": 2.92866943359375,
      "step": 96200
    },
    {
      "epoch": 0.05666622167737609,
      "grad_norm": 0.24348734319210052,
      "learning_rate": 0.00029150015631043783,
      "loss": 2.8744442749023436,
      "step": 96250
    },
    {
      "epoch": 0.05669565867565005,
      "grad_norm": 0.24816389381885529,
      "learning_rate": 0.0002914957407613466,
      "loss": 2.9058438110351563,
      "step": 96300
    },
    {
      "epoch": 0.05672509567392402,
      "grad_norm": 0.4312119781970978,
      "learning_rate": 0.0002914913252122554,
      "loss": 2.9248287963867186,
      "step": 96350
    },
    {
      "epoch": 0.056754532672197974,
      "grad_norm": 0.2893020808696747,
      "learning_rate": 0.0002914869096631642,
      "loss": 2.917902526855469,
      "step": 96400
    },
    {
      "epoch": 0.05678396967047194,
      "grad_norm": 0.3236795663833618,
      "learning_rate": 0.00029148249411407303,
      "loss": 2.9460107421875,
      "step": 96450
    },
    {
      "epoch": 0.0568134066687459,
      "grad_norm": 0.29079514741897583,
      "learning_rate": 0.00029147807856498185,
      "loss": 2.888050231933594,
      "step": 96500
    },
    {
      "epoch": 0.05684284366701986,
      "grad_norm": 0.25691893696784973,
      "learning_rate": 0.00029147366301589066,
      "loss": 2.9057684326171875,
      "step": 96550
    },
    {
      "epoch": 0.056872280665293824,
      "grad_norm": 0.2754230201244354,
      "learning_rate": 0.0002914692474667995,
      "loss": 2.9472647094726563,
      "step": 96600
    },
    {
      "epoch": 0.05690171766356778,
      "grad_norm": 0.30013561248779297,
      "learning_rate": 0.00029146483191770824,
      "loss": 2.906549377441406,
      "step": 96650
    },
    {
      "epoch": 0.056931154661841746,
      "grad_norm": 0.3395778238773346,
      "learning_rate": 0.00029146041636861705,
      "loss": 2.956488037109375,
      "step": 96700
    },
    {
      "epoch": 0.05696059166011571,
      "grad_norm": 0.314247190952301,
      "learning_rate": 0.00029145600081952587,
      "loss": 2.8252719116210936,
      "step": 96750
    },
    {
      "epoch": 0.05699002865838967,
      "grad_norm": 0.3649797737598419,
      "learning_rate": 0.0002914515852704347,
      "loss": 2.8837814331054688,
      "step": 96800
    },
    {
      "epoch": 0.05701946565666363,
      "grad_norm": 0.2492102086544037,
      "learning_rate": 0.0002914471697213435,
      "loss": 2.866067199707031,
      "step": 96850
    },
    {
      "epoch": 0.057048902654937596,
      "grad_norm": 0.27074161171913147,
      "learning_rate": 0.0002914427541722523,
      "loss": 2.931060791015625,
      "step": 96900
    },
    {
      "epoch": 0.05707833965321155,
      "grad_norm": 0.3133189082145691,
      "learning_rate": 0.0002914383386231611,
      "loss": 2.9214837646484373,
      "step": 96950
    },
    {
      "epoch": 0.05710777665148552,
      "grad_norm": 0.2706824243068695,
      "learning_rate": 0.00029143392307406994,
      "loss": 2.9382281494140625,
      "step": 97000
    },
    {
      "epoch": 0.057137213649759475,
      "grad_norm": 0.3144713044166565,
      "learning_rate": 0.0002914295075249787,
      "loss": 2.9911019897460935,
      "step": 97050
    },
    {
      "epoch": 0.05716665064803344,
      "grad_norm": 0.32117271423339844,
      "learning_rate": 0.00029142509197588757,
      "loss": 2.887637023925781,
      "step": 97100
    },
    {
      "epoch": 0.0571960876463074,
      "grad_norm": 0.34102216362953186,
      "learning_rate": 0.0002914206764267964,
      "loss": 2.8541839599609373,
      "step": 97150
    },
    {
      "epoch": 0.05722552464458136,
      "grad_norm": 0.28334006667137146,
      "learning_rate": 0.00029141626087770514,
      "loss": 2.949217529296875,
      "step": 97200
    },
    {
      "epoch": 0.057254961642855325,
      "grad_norm": 0.2650942802429199,
      "learning_rate": 0.00029141184532861396,
      "loss": 2.924463806152344,
      "step": 97250
    },
    {
      "epoch": 0.05728439864112929,
      "grad_norm": 0.3897591531276703,
      "learning_rate": 0.00029140742977952277,
      "loss": 2.8984405517578127,
      "step": 97300
    },
    {
      "epoch": 0.057313835639403246,
      "grad_norm": 0.2848789691925049,
      "learning_rate": 0.0002914030142304316,
      "loss": 2.963315734863281,
      "step": 97350
    },
    {
      "epoch": 0.05734327263767721,
      "grad_norm": 0.25592297315597534,
      "learning_rate": 0.0002913985986813404,
      "loss": 2.895680236816406,
      "step": 97400
    },
    {
      "epoch": 0.05737270963595117,
      "grad_norm": 0.3239132761955261,
      "learning_rate": 0.0002913941831322492,
      "loss": 2.903952331542969,
      "step": 97450
    },
    {
      "epoch": 0.05740214663422513,
      "grad_norm": 0.3244344890117645,
      "learning_rate": 0.00029138976758315803,
      "loss": 2.9029443359375,
      "step": 97500
    },
    {
      "epoch": 0.057431583632499096,
      "grad_norm": 0.2304536998271942,
      "learning_rate": 0.00029138535203406684,
      "loss": 2.8775729370117187,
      "step": 97550
    },
    {
      "epoch": 0.05746102063077305,
      "grad_norm": 0.276327520608902,
      "learning_rate": 0.0002913809364849756,
      "loss": 2.935220947265625,
      "step": 97600
    },
    {
      "epoch": 0.05749045762904702,
      "grad_norm": 0.29436421394348145,
      "learning_rate": 0.0002913765209358844,
      "loss": 2.9374102783203124,
      "step": 97650
    },
    {
      "epoch": 0.05751989462732098,
      "grad_norm": 0.23988719284534454,
      "learning_rate": 0.00029137210538679323,
      "loss": 2.918576965332031,
      "step": 97700
    },
    {
      "epoch": 0.05754933162559494,
      "grad_norm": 0.3047599196434021,
      "learning_rate": 0.00029136768983770205,
      "loss": 2.9384539794921873,
      "step": 97750
    },
    {
      "epoch": 0.0575787686238689,
      "grad_norm": 0.21040789783000946,
      "learning_rate": 0.00029136327428861086,
      "loss": 2.947850341796875,
      "step": 97800
    },
    {
      "epoch": 0.05760820562214286,
      "grad_norm": 0.2634369134902954,
      "learning_rate": 0.0002913588587395197,
      "loss": 2.9718182373046873,
      "step": 97850
    },
    {
      "epoch": 0.057637642620416825,
      "grad_norm": 0.23855134844779968,
      "learning_rate": 0.0002913544431904285,
      "loss": 2.961586608886719,
      "step": 97900
    },
    {
      "epoch": 0.05766707961869079,
      "grad_norm": 0.19655829668045044,
      "learning_rate": 0.0002913500276413373,
      "loss": 2.892228698730469,
      "step": 97950
    },
    {
      "epoch": 0.057696516616964746,
      "grad_norm": 0.2524935007095337,
      "learning_rate": 0.00029134561209224607,
      "loss": 2.8944015502929688,
      "step": 98000
    },
    {
      "epoch": 0.05772595361523871,
      "grad_norm": 0.43016424775123596,
      "learning_rate": 0.0002913411965431549,
      "loss": 2.9359637451171876,
      "step": 98050
    },
    {
      "epoch": 0.05775539061351267,
      "grad_norm": 0.2290690541267395,
      "learning_rate": 0.00029133678099406375,
      "loss": 2.899615478515625,
      "step": 98100
    },
    {
      "epoch": 0.05778482761178663,
      "grad_norm": 0.22039969265460968,
      "learning_rate": 0.0002913323654449725,
      "loss": 2.9213446044921874,
      "step": 98150
    },
    {
      "epoch": 0.057814264610060596,
      "grad_norm": 0.35174819827079773,
      "learning_rate": 0.0002913279498958813,
      "loss": 2.9478955078125,
      "step": 98200
    },
    {
      "epoch": 0.057843701608334554,
      "grad_norm": 0.2176663875579834,
      "learning_rate": 0.00029132353434679014,
      "loss": 2.8740869140625,
      "step": 98250
    },
    {
      "epoch": 0.05787313860660852,
      "grad_norm": 0.25248247385025024,
      "learning_rate": 0.00029131911879769895,
      "loss": 2.9760296630859373,
      "step": 98300
    },
    {
      "epoch": 0.05790257560488248,
      "grad_norm": 0.34346193075180054,
      "learning_rate": 0.0002913147032486077,
      "loss": 2.9683969116210935,
      "step": 98350
    },
    {
      "epoch": 0.05793201260315644,
      "grad_norm": 0.20491580665111542,
      "learning_rate": 0.0002913102876995166,
      "loss": 2.9024032592773437,
      "step": 98400
    },
    {
      "epoch": 0.057961449601430404,
      "grad_norm": 0.20491541922092438,
      "learning_rate": 0.0002913058721504254,
      "loss": 2.8772344970703125,
      "step": 98450
    },
    {
      "epoch": 0.05799088659970436,
      "grad_norm": 0.24902503192424774,
      "learning_rate": 0.00029130145660133416,
      "loss": 2.8977789306640624,
      "step": 98500
    },
    {
      "epoch": 0.058020323597978325,
      "grad_norm": 0.2735876441001892,
      "learning_rate": 0.00029129704105224297,
      "loss": 2.991973571777344,
      "step": 98550
    },
    {
      "epoch": 0.05804976059625229,
      "grad_norm": 0.27182328701019287,
      "learning_rate": 0.0002912926255031518,
      "loss": 2.917396240234375,
      "step": 98600
    },
    {
      "epoch": 0.05807919759452625,
      "grad_norm": 0.25695374608039856,
      "learning_rate": 0.0002912882099540606,
      "loss": 2.9427969360351565,
      "step": 98650
    },
    {
      "epoch": 0.05810863459280021,
      "grad_norm": 0.22839517891407013,
      "learning_rate": 0.0002912837944049694,
      "loss": 2.9366387939453125,
      "step": 98700
    },
    {
      "epoch": 0.058138071591074175,
      "grad_norm": 0.3388182520866394,
      "learning_rate": 0.00029127937885587823,
      "loss": 2.8835491943359375,
      "step": 98750
    },
    {
      "epoch": 0.05816750858934813,
      "grad_norm": 0.3579883575439453,
      "learning_rate": 0.00029127496330678704,
      "loss": 2.8936843872070312,
      "step": 98800
    },
    {
      "epoch": 0.0581969455876221,
      "grad_norm": 0.26625701785087585,
      "learning_rate": 0.00029127054775769586,
      "loss": 2.9416293334960937,
      "step": 98850
    },
    {
      "epoch": 0.058226382585896054,
      "grad_norm": 0.24814482033252716,
      "learning_rate": 0.0002912661322086046,
      "loss": 2.9109576416015623,
      "step": 98900
    },
    {
      "epoch": 0.05825581958417002,
      "grad_norm": 0.26580944657325745,
      "learning_rate": 0.00029126171665951343,
      "loss": 2.917999267578125,
      "step": 98950
    },
    {
      "epoch": 0.05828525658244398,
      "grad_norm": 0.2021879106760025,
      "learning_rate": 0.00029125730111042225,
      "loss": 2.919755554199219,
      "step": 99000
    },
    {
      "epoch": 0.05831469358071794,
      "grad_norm": 0.2493707686662674,
      "learning_rate": 0.00029125288556133106,
      "loss": 2.911775817871094,
      "step": 99050
    },
    {
      "epoch": 0.058344130578991904,
      "grad_norm": 0.3339863717556,
      "learning_rate": 0.0002912484700122399,
      "loss": 2.9458837890625,
      "step": 99100
    },
    {
      "epoch": 0.05837356757726587,
      "grad_norm": 0.2364625781774521,
      "learning_rate": 0.0002912440544631487,
      "loss": 2.867366943359375,
      "step": 99150
    },
    {
      "epoch": 0.058403004575539826,
      "grad_norm": 0.30582910776138306,
      "learning_rate": 0.0002912396389140575,
      "loss": 2.918131103515625,
      "step": 99200
    },
    {
      "epoch": 0.05843244157381379,
      "grad_norm": 0.39197802543640137,
      "learning_rate": 0.0002912352233649663,
      "loss": 2.9510833740234377,
      "step": 99250
    },
    {
      "epoch": 0.05846187857208775,
      "grad_norm": 0.27648425102233887,
      "learning_rate": 0.0002912308078158751,
      "loss": 2.881524658203125,
      "step": 99300
    },
    {
      "epoch": 0.05849131557036171,
      "grad_norm": 0.30569106340408325,
      "learning_rate": 0.0002912263922667839,
      "loss": 2.9449221801757814,
      "step": 99350
    },
    {
      "epoch": 0.058520752568635676,
      "grad_norm": 0.25068342685699463,
      "learning_rate": 0.00029122197671769276,
      "loss": 2.9222561645507814,
      "step": 99400
    },
    {
      "epoch": 0.05855018956690963,
      "grad_norm": 0.2810075283050537,
      "learning_rate": 0.0002912175611686015,
      "loss": 2.8972512817382814,
      "step": 99450
    },
    {
      "epoch": 0.0585796265651836,
      "grad_norm": 0.25244927406311035,
      "learning_rate": 0.00029121314561951034,
      "loss": 2.9634869384765623,
      "step": 99500
    },
    {
      "epoch": 0.05860906356345756,
      "grad_norm": 0.2549196183681488,
      "learning_rate": 0.00029120873007041915,
      "loss": 2.920313720703125,
      "step": 99550
    },
    {
      "epoch": 0.05863850056173152,
      "grad_norm": 0.2057812213897705,
      "learning_rate": 0.00029120431452132797,
      "loss": 2.9074362182617186,
      "step": 99600
    },
    {
      "epoch": 0.05866793756000548,
      "grad_norm": 0.24976307153701782,
      "learning_rate": 0.0002911998989722368,
      "loss": 2.890411071777344,
      "step": 99650
    },
    {
      "epoch": 0.05869737455827944,
      "grad_norm": 0.28493624925613403,
      "learning_rate": 0.0002911954834231456,
      "loss": 2.9206314086914062,
      "step": 99700
    },
    {
      "epoch": 0.058726811556553404,
      "grad_norm": 0.3352918028831482,
      "learning_rate": 0.0002911910678740544,
      "loss": 2.848702087402344,
      "step": 99750
    },
    {
      "epoch": 0.05875624855482737,
      "grad_norm": 0.32755860686302185,
      "learning_rate": 0.0002911866523249632,
      "loss": 2.916754150390625,
      "step": 99800
    },
    {
      "epoch": 0.058785685553101326,
      "grad_norm": 0.3141726553440094,
      "learning_rate": 0.000291182236775872,
      "loss": 2.8943081665039063,
      "step": 99850
    },
    {
      "epoch": 0.05881512255137529,
      "grad_norm": 0.30022692680358887,
      "learning_rate": 0.0002911778212267808,
      "loss": 2.8832421875,
      "step": 99900
    },
    {
      "epoch": 0.05884455954964925,
      "grad_norm": 0.3064238131046295,
      "learning_rate": 0.0002911734056776896,
      "loss": 2.9169888305664062,
      "step": 99950
    },
    {
      "epoch": 0.05887399654792321,
      "grad_norm": 0.32717886567115784,
      "learning_rate": 0.00029116899012859843,
      "loss": 2.8806692504882814,
      "step": 100000
    },
    {
      "epoch": 0.058903433546197176,
      "grad_norm": 0.343678742647171,
      "learning_rate": 0.00029116457457950724,
      "loss": 2.908503112792969,
      "step": 100050
    },
    {
      "epoch": 0.05893287054447113,
      "grad_norm": 0.24260631203651428,
      "learning_rate": 0.00029116015903041606,
      "loss": 2.9449203491210936,
      "step": 100100
    },
    {
      "epoch": 0.0589623075427451,
      "grad_norm": 0.24043351411819458,
      "learning_rate": 0.00029115574348132487,
      "loss": 2.9062875366210936,
      "step": 100150
    },
    {
      "epoch": 0.05899174454101906,
      "grad_norm": 0.277537077665329,
      "learning_rate": 0.00029115132793223363,
      "loss": 2.9006857299804687,
      "step": 100200
    },
    {
      "epoch": 0.05902118153929302,
      "grad_norm": 0.26788073778152466,
      "learning_rate": 0.00029114691238314245,
      "loss": 2.881942138671875,
      "step": 100250
    },
    {
      "epoch": 0.05905061853756698,
      "grad_norm": 0.29988548159599304,
      "learning_rate": 0.00029114249683405126,
      "loss": 2.9126727294921877,
      "step": 100300
    },
    {
      "epoch": 0.05908005553584094,
      "grad_norm": 0.3265853822231293,
      "learning_rate": 0.0002911380812849601,
      "loss": 2.8918057250976563,
      "step": 100350
    },
    {
      "epoch": 0.059109492534114905,
      "grad_norm": 0.31034398078918457,
      "learning_rate": 0.0002911336657358689,
      "loss": 2.8562384033203125,
      "step": 100400
    },
    {
      "epoch": 0.05913892953238887,
      "grad_norm": 0.26982030272483826,
      "learning_rate": 0.0002911292501867777,
      "loss": 2.95739990234375,
      "step": 100450
    },
    {
      "epoch": 0.059168366530662826,
      "grad_norm": 0.282284140586853,
      "learning_rate": 0.0002911248346376865,
      "loss": 2.890999450683594,
      "step": 100500
    },
    {
      "epoch": 0.05919780352893679,
      "grad_norm": 0.24960428476333618,
      "learning_rate": 0.00029112041908859533,
      "loss": 2.9286346435546875,
      "step": 100550
    },
    {
      "epoch": 0.059227240527210755,
      "grad_norm": 0.425866961479187,
      "learning_rate": 0.0002911160035395041,
      "loss": 2.900548400878906,
      "step": 100600
    },
    {
      "epoch": 0.05925667752548471,
      "grad_norm": 0.36173132061958313,
      "learning_rate": 0.0002911115879904129,
      "loss": 2.9286898803710937,
      "step": 100650
    },
    {
      "epoch": 0.059286114523758676,
      "grad_norm": 0.2852208614349365,
      "learning_rate": 0.0002911071724413218,
      "loss": 2.8435214233398436,
      "step": 100700
    },
    {
      "epoch": 0.059315551522032633,
      "grad_norm": 0.2136361300945282,
      "learning_rate": 0.00029110275689223054,
      "loss": 2.912317810058594,
      "step": 100750
    },
    {
      "epoch": 0.0593449885203066,
      "grad_norm": 0.3561578392982483,
      "learning_rate": 0.00029109834134313935,
      "loss": 2.892598876953125,
      "step": 100800
    },
    {
      "epoch": 0.05937442551858056,
      "grad_norm": 0.2790384590625763,
      "learning_rate": 0.00029109392579404817,
      "loss": 2.901575927734375,
      "step": 100850
    },
    {
      "epoch": 0.05940386251685452,
      "grad_norm": 0.2649962306022644,
      "learning_rate": 0.000291089510244957,
      "loss": 2.8840179443359375,
      "step": 100900
    },
    {
      "epoch": 0.059433299515128483,
      "grad_norm": 0.39656296372413635,
      "learning_rate": 0.0002910850946958658,
      "loss": 2.9272970581054687,
      "step": 100950
    },
    {
      "epoch": 0.05946273651340245,
      "grad_norm": 0.3302149474620819,
      "learning_rate": 0.0002910806791467746,
      "loss": 2.9089920043945314,
      "step": 101000
    },
    {
      "epoch": 0.059492173511676405,
      "grad_norm": 0.3341292142868042,
      "learning_rate": 0.0002910762635976834,
      "loss": 2.9458078002929686,
      "step": 101050
    },
    {
      "epoch": 0.05952161050995037,
      "grad_norm": 0.2735714316368103,
      "learning_rate": 0.00029107184804859224,
      "loss": 2.9005746459960937,
      "step": 101100
    },
    {
      "epoch": 0.059551047508224327,
      "grad_norm": 0.23430964350700378,
      "learning_rate": 0.000291067432499501,
      "loss": 2.8833148193359377,
      "step": 101150
    },
    {
      "epoch": 0.05958048450649829,
      "grad_norm": 0.23121774196624756,
      "learning_rate": 0.0002910630169504098,
      "loss": 2.8950115966796877,
      "step": 101200
    },
    {
      "epoch": 0.059609921504772255,
      "grad_norm": 0.2849902808666229,
      "learning_rate": 0.00029105860140131863,
      "loss": 2.925692443847656,
      "step": 101250
    },
    {
      "epoch": 0.05963935850304621,
      "grad_norm": 0.3576222360134125,
      "learning_rate": 0.00029105418585222744,
      "loss": 2.9482049560546875,
      "step": 101300
    },
    {
      "epoch": 0.059668795501320177,
      "grad_norm": 0.2480756938457489,
      "learning_rate": 0.00029104977030313626,
      "loss": 2.8534869384765624,
      "step": 101350
    },
    {
      "epoch": 0.05969823249959414,
      "grad_norm": 0.23178793489933014,
      "learning_rate": 0.00029104535475404507,
      "loss": 2.958385314941406,
      "step": 101400
    },
    {
      "epoch": 0.0597276694978681,
      "grad_norm": 0.2674920856952667,
      "learning_rate": 0.0002910409392049539,
      "loss": 2.845780029296875,
      "step": 101450
    },
    {
      "epoch": 0.05975710649614206,
      "grad_norm": 0.23022252321243286,
      "learning_rate": 0.0002910365236558627,
      "loss": 2.8970550537109374,
      "step": 101500
    },
    {
      "epoch": 0.05978654349441602,
      "grad_norm": 0.2492450326681137,
      "learning_rate": 0.00029103210810677146,
      "loss": 2.9074267578125,
      "step": 101550
    },
    {
      "epoch": 0.059815980492689984,
      "grad_norm": 0.3645123839378357,
      "learning_rate": 0.0002910276925576803,
      "loss": 2.9251113891601563,
      "step": 101600
    },
    {
      "epoch": 0.05984541749096395,
      "grad_norm": 0.25957411527633667,
      "learning_rate": 0.0002910232770085891,
      "loss": 2.8694183349609377,
      "step": 101650
    },
    {
      "epoch": 0.059874854489237905,
      "grad_norm": 0.3129613995552063,
      "learning_rate": 0.0002910188614594979,
      "loss": 2.8838470458984373,
      "step": 101700
    },
    {
      "epoch": 0.05990429148751187,
      "grad_norm": 0.2511197030544281,
      "learning_rate": 0.0002910144459104067,
      "loss": 2.8806277465820314,
      "step": 101750
    },
    {
      "epoch": 0.05993372848578583,
      "grad_norm": 0.2874179780483246,
      "learning_rate": 0.00029101003036131553,
      "loss": 2.9286029052734377,
      "step": 101800
    },
    {
      "epoch": 0.05996316548405979,
      "grad_norm": 0.2753092348575592,
      "learning_rate": 0.00029100561481222435,
      "loss": 2.911233215332031,
      "step": 101850
    },
    {
      "epoch": 0.059992602482333755,
      "grad_norm": 0.3048781454563141,
      "learning_rate": 0.0002910011992631331,
      "loss": 2.9047381591796877,
      "step": 101900
    },
    {
      "epoch": 0.06002203948060771,
      "grad_norm": 0.27063092589378357,
      "learning_rate": 0.0002909967837140419,
      "loss": 2.8787741088867187,
      "step": 101950
    },
    {
      "epoch": 0.06005147647888168,
      "grad_norm": 0.3358362019062042,
      "learning_rate": 0.0002909923681649508,
      "loss": 2.9559146118164064,
      "step": 102000
    },
    {
      "epoch": 0.06008091347715564,
      "grad_norm": 0.35594144463539124,
      "learning_rate": 0.00029098795261585955,
      "loss": 2.880448913574219,
      "step": 102050
    },
    {
      "epoch": 0.0601103504754296,
      "grad_norm": 0.3748849332332611,
      "learning_rate": 0.00029098353706676837,
      "loss": 2.8834829711914063,
      "step": 102100
    },
    {
      "epoch": 0.06013978747370356,
      "grad_norm": 0.23299996554851532,
      "learning_rate": 0.0002909791215176772,
      "loss": 2.829476623535156,
      "step": 102150
    },
    {
      "epoch": 0.06016922447197752,
      "grad_norm": 0.308938592672348,
      "learning_rate": 0.000290974705968586,
      "loss": 2.9282147216796877,
      "step": 102200
    },
    {
      "epoch": 0.060198661470251484,
      "grad_norm": 0.2271452695131302,
      "learning_rate": 0.0002909702904194948,
      "loss": 2.8950924682617187,
      "step": 102250
    },
    {
      "epoch": 0.06022809846852545,
      "grad_norm": 0.3179786801338196,
      "learning_rate": 0.00029096587487040357,
      "loss": 2.9062890625,
      "step": 102300
    },
    {
      "epoch": 0.060257535466799406,
      "grad_norm": 0.26143544912338257,
      "learning_rate": 0.00029096145932131244,
      "loss": 2.913827209472656,
      "step": 102350
    },
    {
      "epoch": 0.06028697246507337,
      "grad_norm": 0.23827321827411652,
      "learning_rate": 0.00029095704377222125,
      "loss": 2.9236605834960936,
      "step": 102400
    },
    {
      "epoch": 0.060316409463347334,
      "grad_norm": 0.28829818964004517,
      "learning_rate": 0.00029095262822313,
      "loss": 2.926838684082031,
      "step": 102450
    },
    {
      "epoch": 0.06034584646162129,
      "grad_norm": 0.28574347496032715,
      "learning_rate": 0.00029094821267403883,
      "loss": 2.933805236816406,
      "step": 102500
    },
    {
      "epoch": 0.060375283459895256,
      "grad_norm": 0.2887125313282013,
      "learning_rate": 0.00029094379712494764,
      "loss": 2.9562725830078125,
      "step": 102550
    },
    {
      "epoch": 0.06040472045816921,
      "grad_norm": 0.38680967688560486,
      "learning_rate": 0.00029093938157585646,
      "loss": 2.870587158203125,
      "step": 102600
    },
    {
      "epoch": 0.06043415745644318,
      "grad_norm": 0.2736600339412689,
      "learning_rate": 0.00029093496602676527,
      "loss": 2.8542416381835936,
      "step": 102650
    },
    {
      "epoch": 0.06046359445471714,
      "grad_norm": 0.25887367129325867,
      "learning_rate": 0.0002909305504776741,
      "loss": 2.9184420776367186,
      "step": 102700
    },
    {
      "epoch": 0.0604930314529911,
      "grad_norm": 0.3196762800216675,
      "learning_rate": 0.0002909261349285829,
      "loss": 2.8767123413085938,
      "step": 102750
    },
    {
      "epoch": 0.06052246845126506,
      "grad_norm": 0.3459981083869934,
      "learning_rate": 0.0002909217193794917,
      "loss": 2.9143212890625,
      "step": 102800
    },
    {
      "epoch": 0.06055190544953903,
      "grad_norm": 0.28222179412841797,
      "learning_rate": 0.0002909173038304005,
      "loss": 2.9013705444335938,
      "step": 102850
    },
    {
      "epoch": 0.060581342447812984,
      "grad_norm": 0.22838880121707916,
      "learning_rate": 0.0002909128882813093,
      "loss": 2.9564154052734377,
      "step": 102900
    },
    {
      "epoch": 0.06061077944608695,
      "grad_norm": 0.1949908435344696,
      "learning_rate": 0.0002909084727322181,
      "loss": 2.883434143066406,
      "step": 102950
    },
    {
      "epoch": 0.060640216444360906,
      "grad_norm": 0.28706973791122437,
      "learning_rate": 0.0002909040571831269,
      "loss": 2.928023376464844,
      "step": 103000
    },
    {
      "epoch": 0.06066965344263487,
      "grad_norm": 0.2914399206638336,
      "learning_rate": 0.00029089964163403573,
      "loss": 2.87707763671875,
      "step": 103050
    },
    {
      "epoch": 0.060699090440908834,
      "grad_norm": 0.23903526365756989,
      "learning_rate": 0.00029089522608494455,
      "loss": 2.8981964111328127,
      "step": 103100
    },
    {
      "epoch": 0.06072852743918279,
      "grad_norm": 0.39911985397338867,
      "learning_rate": 0.00029089081053585336,
      "loss": 2.895768127441406,
      "step": 103150
    },
    {
      "epoch": 0.060757964437456756,
      "grad_norm": 0.2756596505641937,
      "learning_rate": 0.0002908863949867622,
      "loss": 2.905173034667969,
      "step": 103200
    },
    {
      "epoch": 0.06078740143573072,
      "grad_norm": 0.33364245295524597,
      "learning_rate": 0.00029088197943767094,
      "loss": 2.8801199340820314,
      "step": 103250
    },
    {
      "epoch": 0.06081683843400468,
      "grad_norm": 0.3150743842124939,
      "learning_rate": 0.0002908775638885798,
      "loss": 2.9256295776367187,
      "step": 103300
    },
    {
      "epoch": 0.06084627543227864,
      "grad_norm": 0.32949402928352356,
      "learning_rate": 0.0002908731483394886,
      "loss": 2.896568603515625,
      "step": 103350
    },
    {
      "epoch": 0.0608757124305526,
      "grad_norm": 0.24587252736091614,
      "learning_rate": 0.0002908687327903974,
      "loss": 2.9414678955078126,
      "step": 103400
    },
    {
      "epoch": 0.06090514942882656,
      "grad_norm": 0.2950783371925354,
      "learning_rate": 0.0002908643172413062,
      "loss": 2.9000949096679687,
      "step": 103450
    },
    {
      "epoch": 0.06093458642710053,
      "grad_norm": 0.3091264069080353,
      "learning_rate": 0.000290859901692215,
      "loss": 2.8588677978515626,
      "step": 103500
    },
    {
      "epoch": 0.060964023425374485,
      "grad_norm": 0.30744698643684387,
      "learning_rate": 0.0002908554861431238,
      "loss": 2.9057907104492187,
      "step": 103550
    },
    {
      "epoch": 0.06099346042364845,
      "grad_norm": 0.2721624970436096,
      "learning_rate": 0.0002908510705940326,
      "loss": 2.896165771484375,
      "step": 103600
    },
    {
      "epoch": 0.061022897421922406,
      "grad_norm": 0.3084337115287781,
      "learning_rate": 0.00029084665504494145,
      "loss": 2.8924240112304687,
      "step": 103650
    },
    {
      "epoch": 0.06105233442019637,
      "grad_norm": 0.19551952183246613,
      "learning_rate": 0.00029084223949585027,
      "loss": 2.89198974609375,
      "step": 103700
    },
    {
      "epoch": 0.061081771418470335,
      "grad_norm": 0.324750155210495,
      "learning_rate": 0.00029083782394675903,
      "loss": 2.893837890625,
      "step": 103750
    },
    {
      "epoch": 0.06111120841674429,
      "grad_norm": 0.2356337159872055,
      "learning_rate": 0.00029083340839766784,
      "loss": 2.88810546875,
      "step": 103800
    },
    {
      "epoch": 0.061140645415018256,
      "grad_norm": 0.2705807387828827,
      "learning_rate": 0.00029082899284857666,
      "loss": 2.914847412109375,
      "step": 103850
    },
    {
      "epoch": 0.06117008241329222,
      "grad_norm": 0.3094114065170288,
      "learning_rate": 0.00029082457729948547,
      "loss": 2.8655419921875,
      "step": 103900
    },
    {
      "epoch": 0.06119951941156618,
      "grad_norm": 0.21743157505989075,
      "learning_rate": 0.0002908201617503943,
      "loss": 2.9386386108398437,
      "step": 103950
    },
    {
      "epoch": 0.06122895640984014,
      "grad_norm": 0.35700955986976624,
      "learning_rate": 0.0002908157462013031,
      "loss": 2.900682067871094,
      "step": 104000
    },
    {
      "epoch": 0.0612583934081141,
      "grad_norm": 0.26321694254875183,
      "learning_rate": 0.0002908113306522119,
      "loss": 2.9052947998046874,
      "step": 104050
    },
    {
      "epoch": 0.061287830406388064,
      "grad_norm": 0.28436365723609924,
      "learning_rate": 0.00029080691510312073,
      "loss": 2.9137353515625,
      "step": 104100
    },
    {
      "epoch": 0.06131726740466203,
      "grad_norm": 0.21090415120124817,
      "learning_rate": 0.0002908024995540295,
      "loss": 2.8432485961914065,
      "step": 104150
    },
    {
      "epoch": 0.061346704402935985,
      "grad_norm": 0.2544279396533966,
      "learning_rate": 0.0002907980840049383,
      "loss": 2.827618408203125,
      "step": 104200
    },
    {
      "epoch": 0.06137614140120995,
      "grad_norm": 0.23888514935970306,
      "learning_rate": 0.0002907936684558471,
      "loss": 2.8853399658203127,
      "step": 104250
    },
    {
      "epoch": 0.061405578399483914,
      "grad_norm": 0.2416044920682907,
      "learning_rate": 0.00029078925290675593,
      "loss": 2.8581784057617186,
      "step": 104300
    },
    {
      "epoch": 0.06143501539775787,
      "grad_norm": 0.2871928811073303,
      "learning_rate": 0.00029078483735766475,
      "loss": 2.922777099609375,
      "step": 104350
    },
    {
      "epoch": 0.061464452396031835,
      "grad_norm": 0.36042773723602295,
      "learning_rate": 0.00029078042180857356,
      "loss": 2.916822509765625,
      "step": 104400
    },
    {
      "epoch": 0.06149388939430579,
      "grad_norm": 0.2501428723335266,
      "learning_rate": 0.0002907760062594824,
      "loss": 2.8734005737304686,
      "step": 104450
    },
    {
      "epoch": 0.06152332639257976,
      "grad_norm": 0.3886950612068176,
      "learning_rate": 0.0002907715907103912,
      "loss": 2.896537170410156,
      "step": 104500
    },
    {
      "epoch": 0.06155276339085372,
      "grad_norm": 0.21554210782051086,
      "learning_rate": 0.00029076717516129995,
      "loss": 2.894520568847656,
      "step": 104550
    },
    {
      "epoch": 0.06158220038912768,
      "grad_norm": 0.2445795089006424,
      "learning_rate": 0.0002907627596122088,
      "loss": 2.893760986328125,
      "step": 104600
    },
    {
      "epoch": 0.06161163738740164,
      "grad_norm": 0.1898488849401474,
      "learning_rate": 0.00029075834406311763,
      "loss": 2.870807800292969,
      "step": 104650
    },
    {
      "epoch": 0.06164107438567561,
      "grad_norm": 0.2576049268245697,
      "learning_rate": 0.0002907539285140264,
      "loss": 2.898360595703125,
      "step": 104700
    },
    {
      "epoch": 0.061670511383949564,
      "grad_norm": 0.2031017690896988,
      "learning_rate": 0.0002907495129649352,
      "loss": 2.8786212158203126,
      "step": 104750
    },
    {
      "epoch": 0.06169994838222353,
      "grad_norm": 0.3360672891139984,
      "learning_rate": 0.000290745097415844,
      "loss": 2.8886212158203124,
      "step": 104800
    },
    {
      "epoch": 0.061729385380497485,
      "grad_norm": 0.3626479208469391,
      "learning_rate": 0.00029074068186675284,
      "loss": 2.931258239746094,
      "step": 104850
    },
    {
      "epoch": 0.06175882237877145,
      "grad_norm": 0.38852763175964355,
      "learning_rate": 0.00029073626631766165,
      "loss": 2.883630676269531,
      "step": 104900
    },
    {
      "epoch": 0.061788259377045414,
      "grad_norm": 0.2600771188735962,
      "learning_rate": 0.00029073185076857047,
      "loss": 2.9052899169921873,
      "step": 104950
    },
    {
      "epoch": 0.06181769637531937,
      "grad_norm": 0.30892664194107056,
      "learning_rate": 0.0002907274352194793,
      "loss": 2.8991693115234374,
      "step": 105000
    },
    {
      "epoch": 0.061847133373593335,
      "grad_norm": 0.24341440200805664,
      "learning_rate": 0.0002907230196703881,
      "loss": 2.8955291748046874,
      "step": 105050
    },
    {
      "epoch": 0.0618765703718673,
      "grad_norm": 0.23562991619110107,
      "learning_rate": 0.00029071860412129686,
      "loss": 2.93356689453125,
      "step": 105100
    },
    {
      "epoch": 0.06190600737014126,
      "grad_norm": 0.3074955642223358,
      "learning_rate": 0.00029071418857220567,
      "loss": 2.9301626586914065,
      "step": 105150
    },
    {
      "epoch": 0.06193544436841522,
      "grad_norm": 0.29978224635124207,
      "learning_rate": 0.0002907097730231145,
      "loss": 2.882957763671875,
      "step": 105200
    },
    {
      "epoch": 0.06196488136668918,
      "grad_norm": 0.30539247393608093,
      "learning_rate": 0.0002907053574740233,
      "loss": 2.9007968139648437,
      "step": 105250
    },
    {
      "epoch": 0.06199431836496314,
      "grad_norm": 0.24233302474021912,
      "learning_rate": 0.0002907009419249321,
      "loss": 2.910036926269531,
      "step": 105300
    },
    {
      "epoch": 0.06202375536323711,
      "grad_norm": 0.25693029165267944,
      "learning_rate": 0.00029069652637584093,
      "loss": 2.919849548339844,
      "step": 105350
    },
    {
      "epoch": 0.062053192361511064,
      "grad_norm": 0.26417893171310425,
      "learning_rate": 0.00029069211082674974,
      "loss": 2.864848327636719,
      "step": 105400
    },
    {
      "epoch": 0.06208262935978503,
      "grad_norm": 0.2668437957763672,
      "learning_rate": 0.0002906876952776585,
      "loss": 2.87014892578125,
      "step": 105450
    },
    {
      "epoch": 0.062112066358058986,
      "grad_norm": 0.33447328209877014,
      "learning_rate": 0.0002906832797285673,
      "loss": 2.828666076660156,
      "step": 105500
    },
    {
      "epoch": 0.06214150335633295,
      "grad_norm": 0.23564188182353973,
      "learning_rate": 0.00029067886417947613,
      "loss": 2.838035888671875,
      "step": 105550
    },
    {
      "epoch": 0.062170940354606914,
      "grad_norm": 0.26876819133758545,
      "learning_rate": 0.00029067444863038495,
      "loss": 2.8993310546875,
      "step": 105600
    },
    {
      "epoch": 0.06220037735288087,
      "grad_norm": 0.3449220061302185,
      "learning_rate": 0.00029067003308129376,
      "loss": 2.9111004638671876,
      "step": 105650
    },
    {
      "epoch": 0.062229814351154836,
      "grad_norm": 0.261879026889801,
      "learning_rate": 0.0002906656175322026,
      "loss": 2.8191293334960936,
      "step": 105700
    },
    {
      "epoch": 0.0622592513494288,
      "grad_norm": 0.30293500423431396,
      "learning_rate": 0.0002906612019831114,
      "loss": 2.9134356689453127,
      "step": 105750
    },
    {
      "epoch": 0.06228868834770276,
      "grad_norm": 0.3526921570301056,
      "learning_rate": 0.0002906567864340202,
      "loss": 2.8834918212890623,
      "step": 105800
    },
    {
      "epoch": 0.06231812534597672,
      "grad_norm": 0.3682725727558136,
      "learning_rate": 0.00029065237088492896,
      "loss": 2.948492431640625,
      "step": 105850
    },
    {
      "epoch": 0.06234756234425068,
      "grad_norm": 0.24073299765586853,
      "learning_rate": 0.00029064795533583783,
      "loss": 2.9387667846679686,
      "step": 105900
    },
    {
      "epoch": 0.06237699934252464,
      "grad_norm": 0.2765059173107147,
      "learning_rate": 0.00029064353978674665,
      "loss": 2.82505859375,
      "step": 105950
    },
    {
      "epoch": 0.06240643634079861,
      "grad_norm": 0.26394277811050415,
      "learning_rate": 0.0002906391242376554,
      "loss": 2.8679550170898436,
      "step": 106000
    },
    {
      "epoch": 0.062435873339072565,
      "grad_norm": 0.2522808909416199,
      "learning_rate": 0.0002906347086885642,
      "loss": 2.9125384521484374,
      "step": 106050
    },
    {
      "epoch": 0.06246531033734653,
      "grad_norm": 0.3154014050960541,
      "learning_rate": 0.00029063029313947304,
      "loss": 2.9395352172851563,
      "step": 106100
    },
    {
      "epoch": 0.06249474733562049,
      "grad_norm": 0.2939717471599579,
      "learning_rate": 0.00029062587759038185,
      "loss": 2.8959341430664063,
      "step": 106150
    },
    {
      "epoch": 0.06252418433389445,
      "grad_norm": 0.22171235084533691,
      "learning_rate": 0.00029062146204129067,
      "loss": 2.910037841796875,
      "step": 106200
    },
    {
      "epoch": 0.06255362133216841,
      "grad_norm": 0.3787915110588074,
      "learning_rate": 0.0002906170464921995,
      "loss": 2.8675186157226564,
      "step": 106250
    },
    {
      "epoch": 0.06258305833044238,
      "grad_norm": 0.2754770517349243,
      "learning_rate": 0.0002906126309431083,
      "loss": 2.850056457519531,
      "step": 106300
    },
    {
      "epoch": 0.06261249532871634,
      "grad_norm": 0.45920923352241516,
      "learning_rate": 0.0002906082153940171,
      "loss": 2.9185833740234375,
      "step": 106350
    },
    {
      "epoch": 0.0626419323269903,
      "grad_norm": 0.28012815117836,
      "learning_rate": 0.00029060379984492587,
      "loss": 2.7933859252929687,
      "step": 106400
    },
    {
      "epoch": 0.06267136932526426,
      "grad_norm": 0.33645591139793396,
      "learning_rate": 0.0002905993842958347,
      "loss": 2.881284484863281,
      "step": 106450
    },
    {
      "epoch": 0.06270080632353822,
      "grad_norm": 0.19208993017673492,
      "learning_rate": 0.0002905949687467435,
      "loss": 2.9036941528320312,
      "step": 106500
    },
    {
      "epoch": 0.06273024332181218,
      "grad_norm": 0.2503379285335541,
      "learning_rate": 0.0002905905531976523,
      "loss": 2.8778530883789064,
      "step": 106550
    },
    {
      "epoch": 0.06275968032008615,
      "grad_norm": 0.2978924512863159,
      "learning_rate": 0.00029058613764856113,
      "loss": 2.947457275390625,
      "step": 106600
    },
    {
      "epoch": 0.06278911731836011,
      "grad_norm": 0.2522014081478119,
      "learning_rate": 0.00029058172209946994,
      "loss": 2.91978759765625,
      "step": 106650
    },
    {
      "epoch": 0.06281855431663406,
      "grad_norm": 0.3246089518070221,
      "learning_rate": 0.00029057730655037876,
      "loss": 2.9183163452148437,
      "step": 106700
    },
    {
      "epoch": 0.06284799131490804,
      "grad_norm": 0.302788108587265,
      "learning_rate": 0.00029057289100128757,
      "loss": 2.9032998657226563,
      "step": 106750
    },
    {
      "epoch": 0.062877428313182,
      "grad_norm": 0.21364066004753113,
      "learning_rate": 0.00029056847545219633,
      "loss": 2.874453125,
      "step": 106800
    },
    {
      "epoch": 0.06290686531145595,
      "grad_norm": 0.2599111795425415,
      "learning_rate": 0.00029056405990310515,
      "loss": 2.903511962890625,
      "step": 106850
    },
    {
      "epoch": 0.06293630230972991,
      "grad_norm": 0.28297197818756104,
      "learning_rate": 0.000290559644354014,
      "loss": 2.8714306640625,
      "step": 106900
    },
    {
      "epoch": 0.06296573930800388,
      "grad_norm": 0.2777653634548187,
      "learning_rate": 0.0002905552288049228,
      "loss": 2.828352966308594,
      "step": 106950
    },
    {
      "epoch": 0.06299517630627784,
      "grad_norm": 0.23495182394981384,
      "learning_rate": 0.0002905508132558316,
      "loss": 2.914896240234375,
      "step": 107000
    },
    {
      "epoch": 0.0630246133045518,
      "grad_norm": 0.2515745460987091,
      "learning_rate": 0.0002905463977067404,
      "loss": 2.8348907470703124,
      "step": 107050
    },
    {
      "epoch": 0.06305405030282576,
      "grad_norm": 0.35914039611816406,
      "learning_rate": 0.0002905419821576492,
      "loss": 2.848985595703125,
      "step": 107100
    },
    {
      "epoch": 0.06308348730109972,
      "grad_norm": 0.2428610622882843,
      "learning_rate": 0.000290537566608558,
      "loss": 2.878447265625,
      "step": 107150
    },
    {
      "epoch": 0.06311292429937368,
      "grad_norm": 0.2543276846408844,
      "learning_rate": 0.00029053315105946685,
      "loss": 2.8641851806640624,
      "step": 107200
    },
    {
      "epoch": 0.06314236129764765,
      "grad_norm": 0.3342744708061218,
      "learning_rate": 0.00029052873551037566,
      "loss": 2.84156005859375,
      "step": 107250
    },
    {
      "epoch": 0.06317179829592161,
      "grad_norm": 0.32534411549568176,
      "learning_rate": 0.0002905243199612844,
      "loss": 2.95693603515625,
      "step": 107300
    },
    {
      "epoch": 0.06320123529419557,
      "grad_norm": 0.32298779487609863,
      "learning_rate": 0.00029051990441219324,
      "loss": 2.896712341308594,
      "step": 107350
    },
    {
      "epoch": 0.06323067229246954,
      "grad_norm": 0.3192887306213379,
      "learning_rate": 0.00029051548886310205,
      "loss": 2.9458969116210936,
      "step": 107400
    },
    {
      "epoch": 0.0632601092907435,
      "grad_norm": 0.3837229609489441,
      "learning_rate": 0.00029051107331401087,
      "loss": 2.847392272949219,
      "step": 107450
    },
    {
      "epoch": 0.06328954628901745,
      "grad_norm": 0.3108765780925751,
      "learning_rate": 0.0002905066577649197,
      "loss": 2.9249996948242187,
      "step": 107500
    },
    {
      "epoch": 0.06331898328729141,
      "grad_norm": 0.31406891345977783,
      "learning_rate": 0.0002905022422158285,
      "loss": 2.87665283203125,
      "step": 107550
    },
    {
      "epoch": 0.06334842028556538,
      "grad_norm": 0.28193432092666626,
      "learning_rate": 0.0002904978266667373,
      "loss": 2.8562628173828126,
      "step": 107600
    },
    {
      "epoch": 0.06337785728383934,
      "grad_norm": 0.2476724237203598,
      "learning_rate": 0.0002904934111176461,
      "loss": 2.9193231201171876,
      "step": 107650
    },
    {
      "epoch": 0.0634072942821133,
      "grad_norm": 0.2819512188434601,
      "learning_rate": 0.0002904889955685549,
      "loss": 2.81482666015625,
      "step": 107700
    },
    {
      "epoch": 0.06343673128038727,
      "grad_norm": 0.3176180124282837,
      "learning_rate": 0.0002904845800194637,
      "loss": 2.909052429199219,
      "step": 107750
    },
    {
      "epoch": 0.06346616827866122,
      "grad_norm": 0.3948841989040375,
      "learning_rate": 0.0002904801644703725,
      "loss": 2.898951416015625,
      "step": 107800
    },
    {
      "epoch": 0.06349560527693518,
      "grad_norm": 0.27312424778938293,
      "learning_rate": 0.0002904757489212813,
      "loss": 2.9206576538085938,
      "step": 107850
    },
    {
      "epoch": 0.06352504227520915,
      "grad_norm": 0.290961354970932,
      "learning_rate": 0.00029047133337219014,
      "loss": 2.9266622924804686,
      "step": 107900
    },
    {
      "epoch": 0.06355447927348311,
      "grad_norm": 0.26802194118499756,
      "learning_rate": 0.00029046691782309896,
      "loss": 2.830510559082031,
      "step": 107950
    },
    {
      "epoch": 0.06358391627175707,
      "grad_norm": 0.3002345561981201,
      "learning_rate": 0.00029046250227400777,
      "loss": 2.8463259887695314,
      "step": 108000
    },
    {
      "epoch": 0.06361335327003104,
      "grad_norm": 0.21957720816135406,
      "learning_rate": 0.0002904580867249166,
      "loss": 2.9164797973632814,
      "step": 108050
    },
    {
      "epoch": 0.063642790268305,
      "grad_norm": 0.2687874436378479,
      "learning_rate": 0.00029045367117582535,
      "loss": 2.8775665283203127,
      "step": 108100
    },
    {
      "epoch": 0.06367222726657895,
      "grad_norm": 0.22019344568252563,
      "learning_rate": 0.00029044925562673416,
      "loss": 2.8897564697265623,
      "step": 108150
    },
    {
      "epoch": 0.06370166426485292,
      "grad_norm": 0.2843131124973297,
      "learning_rate": 0.00029044484007764303,
      "loss": 2.8706396484375,
      "step": 108200
    },
    {
      "epoch": 0.06373110126312688,
      "grad_norm": 0.18012355268001556,
      "learning_rate": 0.0002904404245285518,
      "loss": 2.8759271240234376,
      "step": 108250
    },
    {
      "epoch": 0.06376053826140084,
      "grad_norm": 0.2409844547510147,
      "learning_rate": 0.0002904360089794606,
      "loss": 2.9346783447265623,
      "step": 108300
    },
    {
      "epoch": 0.0637899752596748,
      "grad_norm": 0.21037611365318298,
      "learning_rate": 0.0002904315934303694,
      "loss": 2.8613186645507813,
      "step": 108350
    },
    {
      "epoch": 0.06381941225794877,
      "grad_norm": 0.30485475063323975,
      "learning_rate": 0.00029042717788127823,
      "loss": 2.8746932983398437,
      "step": 108400
    },
    {
      "epoch": 0.06384884925622272,
      "grad_norm": 0.2768405079841614,
      "learning_rate": 0.00029042276233218705,
      "loss": 2.8329400634765625,
      "step": 108450
    },
    {
      "epoch": 0.06387828625449668,
      "grad_norm": 0.2749389111995697,
      "learning_rate": 0.0002904183467830958,
      "loss": 2.8476742553710936,
      "step": 108500
    },
    {
      "epoch": 0.06390772325277065,
      "grad_norm": 0.2553272843360901,
      "learning_rate": 0.0002904139312340047,
      "loss": 2.9000311279296875,
      "step": 108550
    },
    {
      "epoch": 0.06393716025104461,
      "grad_norm": 0.27529922127723694,
      "learning_rate": 0.0002904095156849135,
      "loss": 2.852520751953125,
      "step": 108600
    },
    {
      "epoch": 0.06396659724931857,
      "grad_norm": 0.2587544620037079,
      "learning_rate": 0.00029040510013582225,
      "loss": 2.875641174316406,
      "step": 108650
    },
    {
      "epoch": 0.06399603424759254,
      "grad_norm": 0.47685784101486206,
      "learning_rate": 0.00029040068458673106,
      "loss": 2.8561859130859375,
      "step": 108700
    },
    {
      "epoch": 0.0640254712458665,
      "grad_norm": 0.2355896681547165,
      "learning_rate": 0.0002903962690376399,
      "loss": 2.8875991821289064,
      "step": 108750
    },
    {
      "epoch": 0.06405490824414045,
      "grad_norm": 0.2798978090286255,
      "learning_rate": 0.0002903918534885487,
      "loss": 2.8935723876953126,
      "step": 108800
    },
    {
      "epoch": 0.06408434524241442,
      "grad_norm": 0.2276436686515808,
      "learning_rate": 0.0002903874379394575,
      "loss": 2.8985830688476564,
      "step": 108850
    },
    {
      "epoch": 0.06411378224068838,
      "grad_norm": 0.2362988442182541,
      "learning_rate": 0.0002903830223903663,
      "loss": 2.8753253173828126,
      "step": 108900
    },
    {
      "epoch": 0.06414321923896234,
      "grad_norm": 0.22385463118553162,
      "learning_rate": 0.00029037860684127514,
      "loss": 2.9256951904296873,
      "step": 108950
    },
    {
      "epoch": 0.0641726562372363,
      "grad_norm": 0.26487329602241516,
      "learning_rate": 0.0002903741912921839,
      "loss": 2.913828125,
      "step": 109000
    },
    {
      "epoch": 0.06420209323551027,
      "grad_norm": 0.3307897746562958,
      "learning_rate": 0.0002903697757430927,
      "loss": 2.8678131103515625,
      "step": 109050
    },
    {
      "epoch": 0.06423153023378422,
      "grad_norm": 0.3305555582046509,
      "learning_rate": 0.0002903653601940015,
      "loss": 2.8846060180664064,
      "step": 109100
    },
    {
      "epoch": 0.06426096723205818,
      "grad_norm": 0.24706366658210754,
      "learning_rate": 0.00029036094464491034,
      "loss": 2.9191400146484376,
      "step": 109150
    },
    {
      "epoch": 0.06429040423033215,
      "grad_norm": 0.22184936702251434,
      "learning_rate": 0.00029035652909581916,
      "loss": 2.8747509765625,
      "step": 109200
    },
    {
      "epoch": 0.06431984122860611,
      "grad_norm": 0.251218318939209,
      "learning_rate": 0.00029035211354672797,
      "loss": 2.8558621215820312,
      "step": 109250
    },
    {
      "epoch": 0.06434927822688007,
      "grad_norm": 0.25734323263168335,
      "learning_rate": 0.0002903476979976368,
      "loss": 2.8295074462890626,
      "step": 109300
    },
    {
      "epoch": 0.06437871522515404,
      "grad_norm": 0.34582504630088806,
      "learning_rate": 0.0002903432824485456,
      "loss": 2.9403350830078123,
      "step": 109350
    },
    {
      "epoch": 0.064408152223428,
      "grad_norm": 0.2267819344997406,
      "learning_rate": 0.00029033886689945436,
      "loss": 2.8924395751953127,
      "step": 109400
    },
    {
      "epoch": 0.06443758922170195,
      "grad_norm": 0.25708669424057007,
      "learning_rate": 0.0002903344513503632,
      "loss": 2.8175299072265627,
      "step": 109450
    },
    {
      "epoch": 0.06446702621997592,
      "grad_norm": 0.2615339756011963,
      "learning_rate": 0.00029033003580127204,
      "loss": 2.887607421875,
      "step": 109500
    },
    {
      "epoch": 0.06449646321824988,
      "grad_norm": 0.2869177460670471,
      "learning_rate": 0.0002903256202521808,
      "loss": 2.8844918823242187,
      "step": 109550
    },
    {
      "epoch": 0.06452590021652384,
      "grad_norm": 0.22131089866161346,
      "learning_rate": 0.0002903212047030896,
      "loss": 2.9199118041992187,
      "step": 109600
    },
    {
      "epoch": 0.06455533721479781,
      "grad_norm": 0.28532668948173523,
      "learning_rate": 0.00029031678915399843,
      "loss": 2.885630798339844,
      "step": 109650
    },
    {
      "epoch": 0.06458477421307177,
      "grad_norm": 0.25108984112739563,
      "learning_rate": 0.00029031237360490725,
      "loss": 2.8407415771484374,
      "step": 109700
    },
    {
      "epoch": 0.06461421121134572,
      "grad_norm": 0.3765522837638855,
      "learning_rate": 0.00029030795805581606,
      "loss": 2.8394342041015626,
      "step": 109750
    },
    {
      "epoch": 0.06464364820961968,
      "grad_norm": 0.22387321293354034,
      "learning_rate": 0.0002903035425067248,
      "loss": 2.8931951904296875,
      "step": 109800
    },
    {
      "epoch": 0.06467308520789365,
      "grad_norm": 0.288616418838501,
      "learning_rate": 0.0002902991269576337,
      "loss": 2.889085693359375,
      "step": 109850
    },
    {
      "epoch": 0.06470252220616761,
      "grad_norm": 0.2895796000957489,
      "learning_rate": 0.0002902947114085425,
      "loss": 2.8911172485351564,
      "step": 109900
    },
    {
      "epoch": 0.06473195920444157,
      "grad_norm": 0.32161492109298706,
      "learning_rate": 0.00029029029585945126,
      "loss": 2.87458251953125,
      "step": 109950
    },
    {
      "epoch": 0.06476139620271554,
      "grad_norm": 0.21102364361286163,
      "learning_rate": 0.0002902858803103601,
      "loss": 2.9161468505859376,
      "step": 110000
    },
    {
      "epoch": 0.0647908332009895,
      "grad_norm": 0.26559963822364807,
      "learning_rate": 0.0002902814647612689,
      "loss": 2.92462646484375,
      "step": 110050
    },
    {
      "epoch": 0.06482027019926345,
      "grad_norm": 0.26561182737350464,
      "learning_rate": 0.0002902770492121777,
      "loss": 2.8767495727539063,
      "step": 110100
    },
    {
      "epoch": 0.06484970719753742,
      "grad_norm": 0.2088671177625656,
      "learning_rate": 0.0002902726336630865,
      "loss": 2.834366455078125,
      "step": 110150
    },
    {
      "epoch": 0.06487914419581138,
      "grad_norm": 0.30123764276504517,
      "learning_rate": 0.00029026821811399534,
      "loss": 2.88776123046875,
      "step": 110200
    },
    {
      "epoch": 0.06490858119408534,
      "grad_norm": 0.3351835608482361,
      "learning_rate": 0.00029026380256490415,
      "loss": 2.868857421875,
      "step": 110250
    },
    {
      "epoch": 0.06493801819235931,
      "grad_norm": 0.22726668417453766,
      "learning_rate": 0.00029025938701581297,
      "loss": 2.9316397094726563,
      "step": 110300
    },
    {
      "epoch": 0.06496745519063327,
      "grad_norm": 0.29100897908210754,
      "learning_rate": 0.0002902549714667217,
      "loss": 2.844149475097656,
      "step": 110350
    },
    {
      "epoch": 0.06499689218890722,
      "grad_norm": 0.2675042748451233,
      "learning_rate": 0.00029025055591763054,
      "loss": 2.899183654785156,
      "step": 110400
    },
    {
      "epoch": 0.0650263291871812,
      "grad_norm": 0.2567143142223358,
      "learning_rate": 0.00029024614036853935,
      "loss": 2.8934817504882813,
      "step": 110450
    },
    {
      "epoch": 0.06505576618545515,
      "grad_norm": 0.2998354434967041,
      "learning_rate": 0.00029024172481944817,
      "loss": 2.9053240966796876,
      "step": 110500
    },
    {
      "epoch": 0.06508520318372911,
      "grad_norm": 0.33098435401916504,
      "learning_rate": 0.000290237309270357,
      "loss": 2.9002755737304686,
      "step": 110550
    },
    {
      "epoch": 0.06511464018200307,
      "grad_norm": 0.23094423115253448,
      "learning_rate": 0.0002902328937212658,
      "loss": 2.895418701171875,
      "step": 110600
    },
    {
      "epoch": 0.06514407718027704,
      "grad_norm": 0.45550796389579773,
      "learning_rate": 0.0002902284781721746,
      "loss": 2.877562255859375,
      "step": 110650
    },
    {
      "epoch": 0.065173514178551,
      "grad_norm": 0.2862653434276581,
      "learning_rate": 0.0002902240626230834,
      "loss": 2.8870364379882814,
      "step": 110700
    },
    {
      "epoch": 0.06520295117682495,
      "grad_norm": 0.3029617965221405,
      "learning_rate": 0.0002902196470739922,
      "loss": 2.8834039306640626,
      "step": 110750
    },
    {
      "epoch": 0.06523238817509892,
      "grad_norm": 0.2550961971282959,
      "learning_rate": 0.00029021523152490106,
      "loss": 2.835882263183594,
      "step": 110800
    },
    {
      "epoch": 0.06526182517337288,
      "grad_norm": 0.22348645329475403,
      "learning_rate": 0.0002902108159758098,
      "loss": 2.8843170166015626,
      "step": 110850
    },
    {
      "epoch": 0.06529126217164684,
      "grad_norm": 0.2671472728252411,
      "learning_rate": 0.00029020640042671863,
      "loss": 2.8716510009765623,
      "step": 110900
    },
    {
      "epoch": 0.06532069916992081,
      "grad_norm": 0.369221568107605,
      "learning_rate": 0.00029020198487762745,
      "loss": 2.8436865234375,
      "step": 110950
    },
    {
      "epoch": 0.06535013616819477,
      "grad_norm": 0.26211515069007874,
      "learning_rate": 0.00029019756932853626,
      "loss": 2.9148681640625,
      "step": 111000
    },
    {
      "epoch": 0.06537957316646872,
      "grad_norm": 0.2697533369064331,
      "learning_rate": 0.0002901931537794451,
      "loss": 2.8887893676757814,
      "step": 111050
    },
    {
      "epoch": 0.0654090101647427,
      "grad_norm": 0.18878044188022614,
      "learning_rate": 0.00029018873823035383,
      "loss": 2.8742974853515624,
      "step": 111100
    },
    {
      "epoch": 0.06543844716301665,
      "grad_norm": 0.3466013967990875,
      "learning_rate": 0.0002901843226812627,
      "loss": 2.8538385009765626,
      "step": 111150
    },
    {
      "epoch": 0.06546788416129061,
      "grad_norm": 0.2431441694498062,
      "learning_rate": 0.0002901799071321715,
      "loss": 2.8876193237304686,
      "step": 111200
    },
    {
      "epoch": 0.06549732115956457,
      "grad_norm": 0.24675752222537994,
      "learning_rate": 0.0002901754915830803,
      "loss": 2.8781753540039063,
      "step": 111250
    },
    {
      "epoch": 0.06552675815783854,
      "grad_norm": 0.38215532898902893,
      "learning_rate": 0.0002901710760339891,
      "loss": 2.8940634155273437,
      "step": 111300
    },
    {
      "epoch": 0.0655561951561125,
      "grad_norm": 0.1993749588727951,
      "learning_rate": 0.0002901666604848979,
      "loss": 2.875889892578125,
      "step": 111350
    },
    {
      "epoch": 0.06558563215438645,
      "grad_norm": 0.3147829473018646,
      "learning_rate": 0.0002901622449358067,
      "loss": 2.9006671142578124,
      "step": 111400
    },
    {
      "epoch": 0.06561506915266042,
      "grad_norm": 0.24800939857959747,
      "learning_rate": 0.00029015782938671554,
      "loss": 2.8823486328125,
      "step": 111450
    },
    {
      "epoch": 0.06564450615093438,
      "grad_norm": 0.2953566312789917,
      "learning_rate": 0.00029015341383762435,
      "loss": 2.915577087402344,
      "step": 111500
    },
    {
      "epoch": 0.06567394314920834,
      "grad_norm": 0.25043821334838867,
      "learning_rate": 0.00029014899828853317,
      "loss": 2.8662042236328125,
      "step": 111550
    },
    {
      "epoch": 0.06570338014748231,
      "grad_norm": 0.2556750774383545,
      "learning_rate": 0.000290144582739442,
      "loss": 2.855448913574219,
      "step": 111600
    },
    {
      "epoch": 0.06573281714575627,
      "grad_norm": 0.317775696516037,
      "learning_rate": 0.00029014016719035074,
      "loss": 2.8965118408203123,
      "step": 111650
    },
    {
      "epoch": 0.06576225414403022,
      "grad_norm": 0.32579997181892395,
      "learning_rate": 0.00029013575164125955,
      "loss": 2.866147155761719,
      "step": 111700
    },
    {
      "epoch": 0.0657916911423042,
      "grad_norm": 0.393545001745224,
      "learning_rate": 0.00029013133609216837,
      "loss": 2.8524554443359373,
      "step": 111750
    },
    {
      "epoch": 0.06582112814057815,
      "grad_norm": 0.3469124436378479,
      "learning_rate": 0.0002901269205430772,
      "loss": 2.8473330688476564,
      "step": 111800
    },
    {
      "epoch": 0.06585056513885211,
      "grad_norm": 0.3147522211074829,
      "learning_rate": 0.000290122504993986,
      "loss": 2.8736062622070313,
      "step": 111850
    },
    {
      "epoch": 0.06588000213712608,
      "grad_norm": 0.2270887792110443,
      "learning_rate": 0.0002901180894448948,
      "loss": 2.8913040161132812,
      "step": 111900
    },
    {
      "epoch": 0.06590943913540004,
      "grad_norm": 0.30551522970199585,
      "learning_rate": 0.0002901136738958036,
      "loss": 2.869385070800781,
      "step": 111950
    },
    {
      "epoch": 0.065938876133674,
      "grad_norm": 0.21986214816570282,
      "learning_rate": 0.00029010925834671244,
      "loss": 2.8925543212890625,
      "step": 112000
    },
    {
      "epoch": 0.06596831313194795,
      "grad_norm": 0.22779053449630737,
      "learning_rate": 0.0002901048427976212,
      "loss": 2.921583557128906,
      "step": 112050
    },
    {
      "epoch": 0.06599775013022192,
      "grad_norm": 0.2972467243671417,
      "learning_rate": 0.00029010042724853007,
      "loss": 2.900799560546875,
      "step": 112100
    },
    {
      "epoch": 0.06602718712849588,
      "grad_norm": 0.2742806673049927,
      "learning_rate": 0.00029009601169943883,
      "loss": 2.8792794799804686,
      "step": 112150
    },
    {
      "epoch": 0.06605662412676984,
      "grad_norm": 0.21894778311252594,
      "learning_rate": 0.00029009159615034765,
      "loss": 2.9265322875976563,
      "step": 112200
    },
    {
      "epoch": 0.06608606112504381,
      "grad_norm": 0.3114735186100006,
      "learning_rate": 0.00029008718060125646,
      "loss": 2.935224914550781,
      "step": 112250
    },
    {
      "epoch": 0.06611549812331777,
      "grad_norm": 0.426044762134552,
      "learning_rate": 0.0002900827650521653,
      "loss": 2.866660461425781,
      "step": 112300
    },
    {
      "epoch": 0.06614493512159172,
      "grad_norm": 0.2730034589767456,
      "learning_rate": 0.0002900783495030741,
      "loss": 2.873021240234375,
      "step": 112350
    },
    {
      "epoch": 0.0661743721198657,
      "grad_norm": 0.2439725250005722,
      "learning_rate": 0.00029007393395398285,
      "loss": 2.851700744628906,
      "step": 112400
    },
    {
      "epoch": 0.06620380911813965,
      "grad_norm": 0.20625387132167816,
      "learning_rate": 0.0002900695184048917,
      "loss": 2.9034521484375,
      "step": 112450
    },
    {
      "epoch": 0.06623324611641361,
      "grad_norm": 0.23443034291267395,
      "learning_rate": 0.00029006510285580053,
      "loss": 2.8645230102539063,
      "step": 112500
    },
    {
      "epoch": 0.06626268311468758,
      "grad_norm": 0.3494531214237213,
      "learning_rate": 0.0002900606873067093,
      "loss": 2.878850402832031,
      "step": 112550
    },
    {
      "epoch": 0.06629212011296154,
      "grad_norm": 0.20173279941082,
      "learning_rate": 0.0002900562717576181,
      "loss": 2.8677618408203127,
      "step": 112600
    },
    {
      "epoch": 0.0663215571112355,
      "grad_norm": 0.3526366949081421,
      "learning_rate": 0.0002900518562085269,
      "loss": 2.915141906738281,
      "step": 112650
    },
    {
      "epoch": 0.06635099410950945,
      "grad_norm": 0.27412402629852295,
      "learning_rate": 0.00029004744065943574,
      "loss": 2.863065185546875,
      "step": 112700
    },
    {
      "epoch": 0.06638043110778342,
      "grad_norm": 0.31756746768951416,
      "learning_rate": 0.00029004302511034455,
      "loss": 2.8955313110351564,
      "step": 112750
    },
    {
      "epoch": 0.06640986810605738,
      "grad_norm": 0.33525487780570984,
      "learning_rate": 0.00029003860956125336,
      "loss": 2.8619540405273436,
      "step": 112800
    },
    {
      "epoch": 0.06643930510433134,
      "grad_norm": 0.2835356891155243,
      "learning_rate": 0.0002900341940121622,
      "loss": 2.8679681396484376,
      "step": 112850
    },
    {
      "epoch": 0.06646874210260531,
      "grad_norm": 0.34606632590293884,
      "learning_rate": 0.000290029778463071,
      "loss": 2.9009054565429686,
      "step": 112900
    },
    {
      "epoch": 0.06649817910087927,
      "grad_norm": 0.43240660429000854,
      "learning_rate": 0.00029002536291397975,
      "loss": 2.87653076171875,
      "step": 112950
    },
    {
      "epoch": 0.06652761609915323,
      "grad_norm": 0.24438327550888062,
      "learning_rate": 0.00029002094736488857,
      "loss": 2.8261148071289064,
      "step": 113000
    },
    {
      "epoch": 0.0665570530974272,
      "grad_norm": 0.2962566018104553,
      "learning_rate": 0.0002900165318157974,
      "loss": 2.786598815917969,
      "step": 113050
    },
    {
      "epoch": 0.06658649009570115,
      "grad_norm": 0.2287105917930603,
      "learning_rate": 0.0002900121162667062,
      "loss": 2.846769104003906,
      "step": 113100
    },
    {
      "epoch": 0.06661592709397511,
      "grad_norm": 0.27770307660102844,
      "learning_rate": 0.000290007700717615,
      "loss": 2.8337591552734374,
      "step": 113150
    },
    {
      "epoch": 0.06664536409224908,
      "grad_norm": 0.23689855635166168,
      "learning_rate": 0.0002900032851685238,
      "loss": 2.8256024169921874,
      "step": 113200
    },
    {
      "epoch": 0.06667480109052304,
      "grad_norm": 0.22097685933113098,
      "learning_rate": 0.00028999886961943264,
      "loss": 2.84720458984375,
      "step": 113250
    },
    {
      "epoch": 0.066704238088797,
      "grad_norm": 0.2510175108909607,
      "learning_rate": 0.00028999445407034146,
      "loss": 2.87729248046875,
      "step": 113300
    },
    {
      "epoch": 0.06673367508707097,
      "grad_norm": 0.4193085730075836,
      "learning_rate": 0.0002899900385212502,
      "loss": 2.8738677978515623,
      "step": 113350
    },
    {
      "epoch": 0.06676311208534493,
      "grad_norm": 0.31912359595298767,
      "learning_rate": 0.0002899856229721591,
      "loss": 2.8979754638671875,
      "step": 113400
    },
    {
      "epoch": 0.06679254908361888,
      "grad_norm": 0.26640456914901733,
      "learning_rate": 0.0002899812074230679,
      "loss": 2.8296710205078126,
      "step": 113450
    },
    {
      "epoch": 0.06682198608189284,
      "grad_norm": 0.33727210760116577,
      "learning_rate": 0.00028997679187397666,
      "loss": 2.85013427734375,
      "step": 113500
    },
    {
      "epoch": 0.06685142308016681,
      "grad_norm": 0.264722615480423,
      "learning_rate": 0.0002899723763248855,
      "loss": 2.8718643188476562,
      "step": 113550
    },
    {
      "epoch": 0.06688086007844077,
      "grad_norm": 0.276334673166275,
      "learning_rate": 0.0002899679607757943,
      "loss": 2.90311767578125,
      "step": 113600
    },
    {
      "epoch": 0.06691029707671473,
      "grad_norm": 0.25847572088241577,
      "learning_rate": 0.0002899635452267031,
      "loss": 2.8705490112304686,
      "step": 113650
    },
    {
      "epoch": 0.0669397340749887,
      "grad_norm": 0.32812413573265076,
      "learning_rate": 0.0002899591296776119,
      "loss": 2.8804913330078126,
      "step": 113700
    },
    {
      "epoch": 0.06696917107326265,
      "grad_norm": 0.23022489249706268,
      "learning_rate": 0.00028995471412852073,
      "loss": 2.888392333984375,
      "step": 113750
    },
    {
      "epoch": 0.06699860807153661,
      "grad_norm": 0.30359914898872375,
      "learning_rate": 0.00028995029857942955,
      "loss": 2.872630615234375,
      "step": 113800
    },
    {
      "epoch": 0.06702804506981058,
      "grad_norm": 0.19582201540470123,
      "learning_rate": 0.0002899458830303383,
      "loss": 2.948680114746094,
      "step": 113850
    },
    {
      "epoch": 0.06705748206808454,
      "grad_norm": 0.2701413333415985,
      "learning_rate": 0.0002899414674812471,
      "loss": 2.9095263671875,
      "step": 113900
    },
    {
      "epoch": 0.0670869190663585,
      "grad_norm": 0.31159284710884094,
      "learning_rate": 0.00028993705193215594,
      "loss": 2.7885931396484374,
      "step": 113950
    },
    {
      "epoch": 0.06711635606463247,
      "grad_norm": 0.2958647310733795,
      "learning_rate": 0.00028993263638306475,
      "loss": 2.8362362670898436,
      "step": 114000
    },
    {
      "epoch": 0.06714579306290643,
      "grad_norm": 0.23138836026191711,
      "learning_rate": 0.00028992822083397356,
      "loss": 2.8879571533203126,
      "step": 114050
    },
    {
      "epoch": 0.06717523006118038,
      "grad_norm": 0.34878599643707275,
      "learning_rate": 0.0002899238052848824,
      "loss": 2.8088442993164064,
      "step": 114100
    },
    {
      "epoch": 0.06720466705945435,
      "grad_norm": 0.28098273277282715,
      "learning_rate": 0.0002899193897357912,
      "loss": 2.777748718261719,
      "step": 114150
    },
    {
      "epoch": 0.06723410405772831,
      "grad_norm": 0.28161197900772095,
      "learning_rate": 0.0002899149741867,
      "loss": 2.870710754394531,
      "step": 114200
    },
    {
      "epoch": 0.06726354105600227,
      "grad_norm": 0.2783256471157074,
      "learning_rate": 0.00028991055863760877,
      "loss": 2.8967645263671873,
      "step": 114250
    },
    {
      "epoch": 0.06729297805427623,
      "grad_norm": 0.317888081073761,
      "learning_rate": 0.0002899061430885176,
      "loss": 2.8446038818359374,
      "step": 114300
    },
    {
      "epoch": 0.0673224150525502,
      "grad_norm": 0.3233780562877655,
      "learning_rate": 0.0002899017275394264,
      "loss": 2.8969155883789064,
      "step": 114350
    },
    {
      "epoch": 0.06735185205082415,
      "grad_norm": 0.20536664128303528,
      "learning_rate": 0.0002898973119903352,
      "loss": 2.8754531860351564,
      "step": 114400
    },
    {
      "epoch": 0.06738128904909811,
      "grad_norm": 0.30006954073905945,
      "learning_rate": 0.000289892896441244,
      "loss": 2.8652969360351563,
      "step": 114450
    },
    {
      "epoch": 0.06741072604737208,
      "grad_norm": 0.24503150582313538,
      "learning_rate": 0.00028988848089215284,
      "loss": 2.918009948730469,
      "step": 114500
    },
    {
      "epoch": 0.06744016304564604,
      "grad_norm": 0.3120179772377014,
      "learning_rate": 0.00028988406534306165,
      "loss": 2.8576736450195312,
      "step": 114550
    },
    {
      "epoch": 0.06746960004392,
      "grad_norm": 0.26069173216819763,
      "learning_rate": 0.00028987964979397047,
      "loss": 2.8954388427734377,
      "step": 114600
    },
    {
      "epoch": 0.06749903704219397,
      "grad_norm": 0.28014951944351196,
      "learning_rate": 0.00028987523424487923,
      "loss": 2.8731149291992186,
      "step": 114650
    },
    {
      "epoch": 0.06752847404046793,
      "grad_norm": 0.29978448152542114,
      "learning_rate": 0.00028987081869578804,
      "loss": 2.8291439819335937,
      "step": 114700
    },
    {
      "epoch": 0.06755791103874188,
      "grad_norm": 0.2412082850933075,
      "learning_rate": 0.0002898664031466969,
      "loss": 2.912606506347656,
      "step": 114750
    },
    {
      "epoch": 0.06758734803701585,
      "grad_norm": 0.2714110314846039,
      "learning_rate": 0.0002898619875976057,
      "loss": 2.8951126098632813,
      "step": 114800
    },
    {
      "epoch": 0.06761678503528981,
      "grad_norm": 0.3872050344944,
      "learning_rate": 0.0002898575720485145,
      "loss": 2.8669003295898436,
      "step": 114850
    },
    {
      "epoch": 0.06764622203356377,
      "grad_norm": 0.23960663378238678,
      "learning_rate": 0.0002898531564994233,
      "loss": 2.83584716796875,
      "step": 114900
    },
    {
      "epoch": 0.06767565903183773,
      "grad_norm": 0.22248612344264984,
      "learning_rate": 0.0002898487409503321,
      "loss": 2.8067916870117187,
      "step": 114950
    },
    {
      "epoch": 0.0677050960301117,
      "grad_norm": 0.2569512128829956,
      "learning_rate": 0.00028984432540124093,
      "loss": 2.897524108886719,
      "step": 115000
    },
    {
      "epoch": 0.06773453302838565,
      "grad_norm": 0.3492935299873352,
      "learning_rate": 0.00028983990985214975,
      "loss": 2.8945217895507813,
      "step": 115050
    },
    {
      "epoch": 0.06776397002665961,
      "grad_norm": 0.31176483631134033,
      "learning_rate": 0.00028983549430305856,
      "loss": 2.888613586425781,
      "step": 115100
    },
    {
      "epoch": 0.06779340702493358,
      "grad_norm": 0.2604803144931793,
      "learning_rate": 0.0002898310787539674,
      "loss": 2.8419976806640626,
      "step": 115150
    },
    {
      "epoch": 0.06782284402320754,
      "grad_norm": 0.3397207260131836,
      "learning_rate": 0.00028982666320487613,
      "loss": 2.879794921875,
      "step": 115200
    },
    {
      "epoch": 0.0678522810214815,
      "grad_norm": 0.25062501430511475,
      "learning_rate": 0.00028982224765578495,
      "loss": 2.8308648681640625,
      "step": 115250
    },
    {
      "epoch": 0.06788171801975547,
      "grad_norm": 0.2922394573688507,
      "learning_rate": 0.00028981783210669376,
      "loss": 2.826538391113281,
      "step": 115300
    },
    {
      "epoch": 0.06791115501802943,
      "grad_norm": 0.2744636833667755,
      "learning_rate": 0.0002898134165576026,
      "loss": 2.920130310058594,
      "step": 115350
    },
    {
      "epoch": 0.06794059201630338,
      "grad_norm": 0.4401750862598419,
      "learning_rate": 0.0002898090010085114,
      "loss": 2.8410296630859375,
      "step": 115400
    },
    {
      "epoch": 0.06797002901457735,
      "grad_norm": 0.21584826707839966,
      "learning_rate": 0.0002898045854594202,
      "loss": 2.918349304199219,
      "step": 115450
    },
    {
      "epoch": 0.06799946601285131,
      "grad_norm": 0.3611242175102234,
      "learning_rate": 0.000289800169910329,
      "loss": 2.886749572753906,
      "step": 115500
    },
    {
      "epoch": 0.06802890301112527,
      "grad_norm": 0.27474355697631836,
      "learning_rate": 0.0002897957543612378,
      "loss": 2.897039489746094,
      "step": 115550
    },
    {
      "epoch": 0.06805834000939924,
      "grad_norm": 0.32512417435646057,
      "learning_rate": 0.0002897913388121466,
      "loss": 2.8217034912109376,
      "step": 115600
    },
    {
      "epoch": 0.0680877770076732,
      "grad_norm": 0.27376168966293335,
      "learning_rate": 0.0002897869232630554,
      "loss": 2.8875067138671877,
      "step": 115650
    },
    {
      "epoch": 0.06811721400594715,
      "grad_norm": 0.3340988755226135,
      "learning_rate": 0.0002897825077139642,
      "loss": 2.9225799560546877,
      "step": 115700
    },
    {
      "epoch": 0.06814665100422111,
      "grad_norm": 0.20359082520008087,
      "learning_rate": 0.00028977809216487304,
      "loss": 2.8438812255859376,
      "step": 115750
    },
    {
      "epoch": 0.06817608800249508,
      "grad_norm": 0.2906690239906311,
      "learning_rate": 0.00028977367661578185,
      "loss": 2.8964239501953126,
      "step": 115800
    },
    {
      "epoch": 0.06820552500076904,
      "grad_norm": 0.3331398665904999,
      "learning_rate": 0.00028976926106669067,
      "loss": 2.8890826416015627,
      "step": 115850
    },
    {
      "epoch": 0.068234961999043,
      "grad_norm": 0.20972785353660583,
      "learning_rate": 0.0002897648455175995,
      "loss": 2.8625637817382814,
      "step": 115900
    },
    {
      "epoch": 0.06826439899731697,
      "grad_norm": 0.3898211717605591,
      "learning_rate": 0.00028976042996850824,
      "loss": 2.8149429321289063,
      "step": 115950
    },
    {
      "epoch": 0.06829383599559093,
      "grad_norm": 0.2694813907146454,
      "learning_rate": 0.00028975601441941706,
      "loss": 2.934507141113281,
      "step": 116000
    },
    {
      "epoch": 0.06832327299386488,
      "grad_norm": 0.2756015956401825,
      "learning_rate": 0.0002897515988703259,
      "loss": 2.902260437011719,
      "step": 116050
    },
    {
      "epoch": 0.06835270999213885,
      "grad_norm": 0.22346080839633942,
      "learning_rate": 0.0002897471833212347,
      "loss": 2.8713433837890623,
      "step": 116100
    },
    {
      "epoch": 0.06838214699041281,
      "grad_norm": 0.21629571914672852,
      "learning_rate": 0.0002897427677721435,
      "loss": 2.8788433837890626,
      "step": 116150
    },
    {
      "epoch": 0.06841158398868677,
      "grad_norm": 0.2403409779071808,
      "learning_rate": 0.0002897383522230523,
      "loss": 2.7790628051757813,
      "step": 116200
    },
    {
      "epoch": 0.06844102098696074,
      "grad_norm": 0.2661104202270508,
      "learning_rate": 0.00028973393667396113,
      "loss": 2.9007183837890627,
      "step": 116250
    },
    {
      "epoch": 0.0684704579852347,
      "grad_norm": 0.28111061453819275,
      "learning_rate": 0.00028972952112486994,
      "loss": 2.890877380371094,
      "step": 116300
    },
    {
      "epoch": 0.06849989498350866,
      "grad_norm": 0.23229998350143433,
      "learning_rate": 0.00028972510557577876,
      "loss": 2.862464599609375,
      "step": 116350
    },
    {
      "epoch": 0.06852933198178261,
      "grad_norm": 0.333209753036499,
      "learning_rate": 0.0002897206900266876,
      "loss": 2.8220809936523437,
      "step": 116400
    },
    {
      "epoch": 0.06855876898005658,
      "grad_norm": 0.21879255771636963,
      "learning_rate": 0.0002897162744775964,
      "loss": 2.837479248046875,
      "step": 116450
    },
    {
      "epoch": 0.06858820597833054,
      "grad_norm": 0.27444833517074585,
      "learning_rate": 0.00028971185892850515,
      "loss": 2.88041748046875,
      "step": 116500
    },
    {
      "epoch": 0.0686176429766045,
      "grad_norm": 0.413615882396698,
      "learning_rate": 0.00028970744337941396,
      "loss": 2.9027960205078127,
      "step": 116550
    },
    {
      "epoch": 0.06864707997487847,
      "grad_norm": 0.43801718950271606,
      "learning_rate": 0.0002897030278303228,
      "loss": 2.8496768188476564,
      "step": 116600
    },
    {
      "epoch": 0.06867651697315243,
      "grad_norm": 0.23024696111679077,
      "learning_rate": 0.0002896986122812316,
      "loss": 2.916062927246094,
      "step": 116650
    },
    {
      "epoch": 0.06870595397142638,
      "grad_norm": 0.2619037330150604,
      "learning_rate": 0.0002896941967321404,
      "loss": 2.8785134887695314,
      "step": 116700
    },
    {
      "epoch": 0.06873539096970036,
      "grad_norm": 0.2850112020969391,
      "learning_rate": 0.0002896897811830492,
      "loss": 2.885296630859375,
      "step": 116750
    },
    {
      "epoch": 0.06876482796797431,
      "grad_norm": 0.26143673062324524,
      "learning_rate": 0.00028968536563395804,
      "loss": 2.7953012084960935,
      "step": 116800
    },
    {
      "epoch": 0.06879426496624827,
      "grad_norm": 0.33160752058029175,
      "learning_rate": 0.00028968095008486685,
      "loss": 2.8313836669921875,
      "step": 116850
    },
    {
      "epoch": 0.06882370196452224,
      "grad_norm": 0.2410001903772354,
      "learning_rate": 0.0002896765345357756,
      "loss": 2.8098760986328126,
      "step": 116900
    },
    {
      "epoch": 0.0688531389627962,
      "grad_norm": 0.35360491275787354,
      "learning_rate": 0.0002896721189866844,
      "loss": 2.8786056518554686,
      "step": 116950
    },
    {
      "epoch": 0.06888257596107016,
      "grad_norm": 0.2145681530237198,
      "learning_rate": 0.0002896677034375933,
      "loss": 2.8802972412109376,
      "step": 117000
    },
    {
      "epoch": 0.06891201295934413,
      "grad_norm": 0.2389291524887085,
      "learning_rate": 0.00028966328788850205,
      "loss": 2.879873046875,
      "step": 117050
    },
    {
      "epoch": 0.06894144995761808,
      "grad_norm": 0.30060791969299316,
      "learning_rate": 0.00028965887233941087,
      "loss": 2.8843218994140627,
      "step": 117100
    },
    {
      "epoch": 0.06897088695589204,
      "grad_norm": 0.3020698130130768,
      "learning_rate": 0.0002896544567903197,
      "loss": 2.87213134765625,
      "step": 117150
    },
    {
      "epoch": 0.069000323954166,
      "grad_norm": 0.2787661850452423,
      "learning_rate": 0.0002896500412412285,
      "loss": 2.8333090209960936,
      "step": 117200
    },
    {
      "epoch": 0.06902976095243997,
      "grad_norm": 0.2389114797115326,
      "learning_rate": 0.00028964562569213726,
      "loss": 2.8571807861328127,
      "step": 117250
    },
    {
      "epoch": 0.06905919795071393,
      "grad_norm": 0.3735071122646332,
      "learning_rate": 0.00028964121014304607,
      "loss": 2.8811306762695312,
      "step": 117300
    },
    {
      "epoch": 0.06908863494898788,
      "grad_norm": 0.26587939262390137,
      "learning_rate": 0.00028963679459395494,
      "loss": 2.870091857910156,
      "step": 117350
    },
    {
      "epoch": 0.06911807194726186,
      "grad_norm": 0.33676400780677795,
      "learning_rate": 0.0002896323790448637,
      "loss": 2.8791372680664065,
      "step": 117400
    },
    {
      "epoch": 0.06914750894553581,
      "grad_norm": 0.2660256028175354,
      "learning_rate": 0.0002896279634957725,
      "loss": 2.8664599609375,
      "step": 117450
    },
    {
      "epoch": 0.06917694594380977,
      "grad_norm": 0.2766222059726715,
      "learning_rate": 0.00028962354794668133,
      "loss": 2.8211709594726564,
      "step": 117500
    },
    {
      "epoch": 0.06920638294208374,
      "grad_norm": 0.28523486852645874,
      "learning_rate": 0.00028961913239759014,
      "loss": 2.85383544921875,
      "step": 117550
    },
    {
      "epoch": 0.0692358199403577,
      "grad_norm": 0.24298636615276337,
      "learning_rate": 0.00028961471684849896,
      "loss": 2.850484619140625,
      "step": 117600
    },
    {
      "epoch": 0.06926525693863166,
      "grad_norm": 0.2612195611000061,
      "learning_rate": 0.0002896103012994078,
      "loss": 2.8322882080078124,
      "step": 117650
    },
    {
      "epoch": 0.06929469393690563,
      "grad_norm": 0.2738446891307831,
      "learning_rate": 0.0002896058857503166,
      "loss": 2.850227966308594,
      "step": 117700
    },
    {
      "epoch": 0.06932413093517958,
      "grad_norm": 0.3086548447608948,
      "learning_rate": 0.0002896014702012254,
      "loss": 2.8736956787109373,
      "step": 117750
    },
    {
      "epoch": 0.06935356793345354,
      "grad_norm": 0.26215389370918274,
      "learning_rate": 0.00028959705465213416,
      "loss": 2.8187680053710937,
      "step": 117800
    },
    {
      "epoch": 0.06938300493172751,
      "grad_norm": 0.28093940019607544,
      "learning_rate": 0.000289592639103043,
      "loss": 2.8270913696289064,
      "step": 117850
    },
    {
      "epoch": 0.06941244193000147,
      "grad_norm": 0.2678487300872803,
      "learning_rate": 0.0002895882235539518,
      "loss": 2.8279693603515623,
      "step": 117900
    },
    {
      "epoch": 0.06944187892827543,
      "grad_norm": 0.22657613456249237,
      "learning_rate": 0.0002895838080048606,
      "loss": 2.857185974121094,
      "step": 117950
    },
    {
      "epoch": 0.06947131592654938,
      "grad_norm": 0.2634674608707428,
      "learning_rate": 0.0002895793924557694,
      "loss": 2.8527822875976563,
      "step": 118000
    },
    {
      "epoch": 0.06950075292482336,
      "grad_norm": 0.19755171239376068,
      "learning_rate": 0.00028957497690667824,
      "loss": 2.8725836181640627,
      "step": 118050
    },
    {
      "epoch": 0.06953018992309731,
      "grad_norm": 0.24207337200641632,
      "learning_rate": 0.00028957056135758705,
      "loss": 2.8813323974609375,
      "step": 118100
    },
    {
      "epoch": 0.06955962692137127,
      "grad_norm": 0.2805010974407196,
      "learning_rate": 0.00028956614580849586,
      "loss": 2.8799273681640627,
      "step": 118150
    },
    {
      "epoch": 0.06958906391964524,
      "grad_norm": 0.2758180499076843,
      "learning_rate": 0.0002895617302594046,
      "loss": 2.8385601806640626,
      "step": 118200
    },
    {
      "epoch": 0.0696185009179192,
      "grad_norm": 0.194618821144104,
      "learning_rate": 0.00028955731471031344,
      "loss": 2.8418392944335937,
      "step": 118250
    },
    {
      "epoch": 0.06964793791619316,
      "grad_norm": 0.24721579253673553,
      "learning_rate": 0.0002895528991612223,
      "loss": 2.889718017578125,
      "step": 118300
    },
    {
      "epoch": 0.06967737491446713,
      "grad_norm": 0.2719581425189972,
      "learning_rate": 0.00028954848361213107,
      "loss": 2.9047665405273437,
      "step": 118350
    },
    {
      "epoch": 0.06970681191274108,
      "grad_norm": 0.3782981336116791,
      "learning_rate": 0.0002895440680630399,
      "loss": 2.8826776123046876,
      "step": 118400
    },
    {
      "epoch": 0.06973624891101504,
      "grad_norm": 0.27168020606040955,
      "learning_rate": 0.0002895396525139487,
      "loss": 2.8658132934570313,
      "step": 118450
    },
    {
      "epoch": 0.06976568590928901,
      "grad_norm": 0.34398967027664185,
      "learning_rate": 0.0002895352369648575,
      "loss": 2.8910589599609375,
      "step": 118500
    },
    {
      "epoch": 0.06979512290756297,
      "grad_norm": 0.20119121670722961,
      "learning_rate": 0.0002895308214157663,
      "loss": 2.8844570922851562,
      "step": 118550
    },
    {
      "epoch": 0.06982455990583693,
      "grad_norm": 0.26281097531318665,
      "learning_rate": 0.0002895264058666751,
      "loss": 2.838390197753906,
      "step": 118600
    },
    {
      "epoch": 0.06985399690411088,
      "grad_norm": 0.23731809854507446,
      "learning_rate": 0.00028952199031758395,
      "loss": 2.870788269042969,
      "step": 118650
    },
    {
      "epoch": 0.06988343390238486,
      "grad_norm": 0.3448312282562256,
      "learning_rate": 0.00028951757476849277,
      "loss": 2.86357421875,
      "step": 118700
    },
    {
      "epoch": 0.06991287090065881,
      "grad_norm": 0.38323384523391724,
      "learning_rate": 0.00028951315921940153,
      "loss": 2.880274658203125,
      "step": 118750
    },
    {
      "epoch": 0.06994230789893277,
      "grad_norm": 0.31275197863578796,
      "learning_rate": 0.00028950874367031034,
      "loss": 2.834150695800781,
      "step": 118800
    },
    {
      "epoch": 0.06997174489720674,
      "grad_norm": 0.28346240520477295,
      "learning_rate": 0.00028950432812121916,
      "loss": 2.8497406005859376,
      "step": 118850
    },
    {
      "epoch": 0.0700011818954807,
      "grad_norm": 0.24078407883644104,
      "learning_rate": 0.000289499912572128,
      "loss": 2.8793734741210937,
      "step": 118900
    },
    {
      "epoch": 0.07003061889375466,
      "grad_norm": 0.2780616283416748,
      "learning_rate": 0.0002894954970230368,
      "loss": 2.894002685546875,
      "step": 118950
    },
    {
      "epoch": 0.07006005589202863,
      "grad_norm": 0.2994701862335205,
      "learning_rate": 0.0002894910814739456,
      "loss": 2.8128411865234373,
      "step": 119000
    },
    {
      "epoch": 0.07008949289030258,
      "grad_norm": 0.2964443862438202,
      "learning_rate": 0.0002894866659248544,
      "loss": 2.8878155517578126,
      "step": 119050
    },
    {
      "epoch": 0.07011892988857654,
      "grad_norm": 0.246500626206398,
      "learning_rate": 0.0002894822503757632,
      "loss": 2.867368469238281,
      "step": 119100
    },
    {
      "epoch": 0.07014836688685051,
      "grad_norm": 0.3012906014919281,
      "learning_rate": 0.000289477834826672,
      "loss": 2.8379248046875,
      "step": 119150
    },
    {
      "epoch": 0.07017780388512447,
      "grad_norm": 0.2942560315132141,
      "learning_rate": 0.0002894734192775808,
      "loss": 2.851900634765625,
      "step": 119200
    },
    {
      "epoch": 0.07020724088339843,
      "grad_norm": 0.24100081622600555,
      "learning_rate": 0.0002894690037284896,
      "loss": 2.84918701171875,
      "step": 119250
    },
    {
      "epoch": 0.0702366778816724,
      "grad_norm": 0.22375395894050598,
      "learning_rate": 0.00028946458817939843,
      "loss": 2.873475341796875,
      "step": 119300
    },
    {
      "epoch": 0.07026611487994636,
      "grad_norm": 0.4105609953403473,
      "learning_rate": 0.00028946017263030725,
      "loss": 2.8683743286132812,
      "step": 119350
    },
    {
      "epoch": 0.07029555187822031,
      "grad_norm": 0.23860882222652435,
      "learning_rate": 0.00028945575708121606,
      "loss": 2.849665222167969,
      "step": 119400
    },
    {
      "epoch": 0.07032498887649427,
      "grad_norm": 0.21775293350219727,
      "learning_rate": 0.0002894513415321249,
      "loss": 2.8792098999023437,
      "step": 119450
    },
    {
      "epoch": 0.07035442587476824,
      "grad_norm": 0.24344699084758759,
      "learning_rate": 0.00028944692598303364,
      "loss": 2.8610272216796875,
      "step": 119500
    },
    {
      "epoch": 0.0703838628730422,
      "grad_norm": 0.35347047448158264,
      "learning_rate": 0.00028944251043394245,
      "loss": 2.8688446044921876,
      "step": 119550
    },
    {
      "epoch": 0.07041329987131616,
      "grad_norm": 0.25674906373023987,
      "learning_rate": 0.0002894380948848513,
      "loss": 2.792940368652344,
      "step": 119600
    },
    {
      "epoch": 0.07044273686959013,
      "grad_norm": 0.24437300860881805,
      "learning_rate": 0.0002894336793357601,
      "loss": 2.844547119140625,
      "step": 119650
    },
    {
      "epoch": 0.07047217386786409,
      "grad_norm": 0.27756768465042114,
      "learning_rate": 0.0002894292637866689,
      "loss": 2.8809478759765623,
      "step": 119700
    },
    {
      "epoch": 0.07050161086613804,
      "grad_norm": 0.2552788555622101,
      "learning_rate": 0.0002894248482375777,
      "loss": 2.889823303222656,
      "step": 119750
    },
    {
      "epoch": 0.07053104786441201,
      "grad_norm": 0.29786714911460876,
      "learning_rate": 0.0002894204326884865,
      "loss": 2.7847021484375,
      "step": 119800
    },
    {
      "epoch": 0.07056048486268597,
      "grad_norm": 0.2623938322067261,
      "learning_rate": 0.00028941601713939534,
      "loss": 2.902796630859375,
      "step": 119850
    },
    {
      "epoch": 0.07058992186095993,
      "grad_norm": 0.2613033354282379,
      "learning_rate": 0.0002894116015903041,
      "loss": 2.879153137207031,
      "step": 119900
    },
    {
      "epoch": 0.0706193588592339,
      "grad_norm": 0.28925570845603943,
      "learning_rate": 0.00028940718604121297,
      "loss": 2.86514892578125,
      "step": 119950
    },
    {
      "epoch": 0.07064879585750786,
      "grad_norm": 0.2633996903896332,
      "learning_rate": 0.0002894027704921218,
      "loss": 2.867372741699219,
      "step": 120000
    },
    {
      "epoch": 0.07067823285578181,
      "grad_norm": 0.22908350825309753,
      "learning_rate": 0.00028939835494303054,
      "loss": 2.9227362060546875,
      "step": 120050
    },
    {
      "epoch": 0.07070766985405577,
      "grad_norm": 0.2635432183742523,
      "learning_rate": 0.00028939393939393936,
      "loss": 2.8600534057617186,
      "step": 120100
    },
    {
      "epoch": 0.07073710685232974,
      "grad_norm": 0.2657284140586853,
      "learning_rate": 0.00028938952384484817,
      "loss": 2.8651272583007814,
      "step": 120150
    },
    {
      "epoch": 0.0707665438506037,
      "grad_norm": 0.24793921411037445,
      "learning_rate": 0.000289385108295757,
      "loss": 2.8570077514648435,
      "step": 120200
    },
    {
      "epoch": 0.07079598084887766,
      "grad_norm": 0.34680846333503723,
      "learning_rate": 0.0002893806927466658,
      "loss": 2.8181326293945315,
      "step": 120250
    },
    {
      "epoch": 0.07082541784715163,
      "grad_norm": 0.388691782951355,
      "learning_rate": 0.0002893762771975746,
      "loss": 2.827430419921875,
      "step": 120300
    },
    {
      "epoch": 0.07085485484542559,
      "grad_norm": 0.27915820479393005,
      "learning_rate": 0.00028937186164848343,
      "loss": 2.8902401733398437,
      "step": 120350
    },
    {
      "epoch": 0.07088429184369954,
      "grad_norm": 0.3311191499233246,
      "learning_rate": 0.00028936744609939224,
      "loss": 2.913979797363281,
      "step": 120400
    },
    {
      "epoch": 0.07091372884197351,
      "grad_norm": 0.2392011135816574,
      "learning_rate": 0.000289363030550301,
      "loss": 2.8645660400390627,
      "step": 120450
    },
    {
      "epoch": 0.07094316584024747,
      "grad_norm": 0.24905885756015778,
      "learning_rate": 0.0002893586150012098,
      "loss": 2.86227783203125,
      "step": 120500
    },
    {
      "epoch": 0.07097260283852143,
      "grad_norm": 0.3091883659362793,
      "learning_rate": 0.00028935419945211863,
      "loss": 2.798553161621094,
      "step": 120550
    },
    {
      "epoch": 0.0710020398367954,
      "grad_norm": 0.34174609184265137,
      "learning_rate": 0.00028934978390302745,
      "loss": 2.8375531005859376,
      "step": 120600
    },
    {
      "epoch": 0.07103147683506936,
      "grad_norm": 0.3038428723812103,
      "learning_rate": 0.00028934536835393626,
      "loss": 2.895174255371094,
      "step": 120650
    },
    {
      "epoch": 0.07106091383334331,
      "grad_norm": 0.238204225897789,
      "learning_rate": 0.0002893409528048451,
      "loss": 2.860135803222656,
      "step": 120700
    },
    {
      "epoch": 0.07109035083161729,
      "grad_norm": 0.22607505321502686,
      "learning_rate": 0.0002893365372557539,
      "loss": 2.8718609619140625,
      "step": 120750
    },
    {
      "epoch": 0.07111978782989124,
      "grad_norm": 0.27519023418426514,
      "learning_rate": 0.00028933212170666265,
      "loss": 2.8670458984375,
      "step": 120800
    },
    {
      "epoch": 0.0711492248281652,
      "grad_norm": 0.2473449856042862,
      "learning_rate": 0.00028932770615757147,
      "loss": 2.86790283203125,
      "step": 120850
    },
    {
      "epoch": 0.07117866182643916,
      "grad_norm": 0.2435062974691391,
      "learning_rate": 0.00028932329060848034,
      "loss": 2.8899710083007815,
      "step": 120900
    },
    {
      "epoch": 0.07120809882471313,
      "grad_norm": 0.25558218359947205,
      "learning_rate": 0.0002893188750593891,
      "loss": 2.8812271118164063,
      "step": 120950
    },
    {
      "epoch": 0.07123753582298709,
      "grad_norm": 0.3371993601322174,
      "learning_rate": 0.0002893144595102979,
      "loss": 2.8616592407226564,
      "step": 121000
    },
    {
      "epoch": 0.07126697282126104,
      "grad_norm": 0.2916070520877838,
      "learning_rate": 0.0002893100439612067,
      "loss": 2.88432861328125,
      "step": 121050
    },
    {
      "epoch": 0.07129640981953501,
      "grad_norm": 0.2981922924518585,
      "learning_rate": 0.00028930562841211554,
      "loss": 2.8135601806640627,
      "step": 121100
    },
    {
      "epoch": 0.07132584681780897,
      "grad_norm": 0.2899923026561737,
      "learning_rate": 0.00028930121286302435,
      "loss": 2.8973333740234377,
      "step": 121150
    },
    {
      "epoch": 0.07135528381608293,
      "grad_norm": 0.2260158658027649,
      "learning_rate": 0.0002892967973139331,
      "loss": 2.831890869140625,
      "step": 121200
    },
    {
      "epoch": 0.0713847208143569,
      "grad_norm": 0.2418341338634491,
      "learning_rate": 0.000289292381764842,
      "loss": 2.901534118652344,
      "step": 121250
    },
    {
      "epoch": 0.07141415781263086,
      "grad_norm": 0.2973328232765198,
      "learning_rate": 0.0002892879662157508,
      "loss": 2.841324462890625,
      "step": 121300
    },
    {
      "epoch": 0.07144359481090481,
      "grad_norm": 0.24257944524288177,
      "learning_rate": 0.00028928355066665956,
      "loss": 2.827840576171875,
      "step": 121350
    },
    {
      "epoch": 0.07147303180917879,
      "grad_norm": 0.23235680162906647,
      "learning_rate": 0.00028927913511756837,
      "loss": 2.8557489013671873,
      "step": 121400
    },
    {
      "epoch": 0.07150246880745274,
      "grad_norm": 0.2912237346172333,
      "learning_rate": 0.0002892747195684772,
      "loss": 2.8828775024414064,
      "step": 121450
    },
    {
      "epoch": 0.0715319058057267,
      "grad_norm": 0.3081805109977722,
      "learning_rate": 0.000289270304019386,
      "loss": 2.8655340576171877,
      "step": 121500
    },
    {
      "epoch": 0.07156134280400067,
      "grad_norm": 0.2799891233444214,
      "learning_rate": 0.0002892658884702948,
      "loss": 2.8949566650390626,
      "step": 121550
    },
    {
      "epoch": 0.07159077980227463,
      "grad_norm": 0.24908915162086487,
      "learning_rate": 0.00028926147292120363,
      "loss": 2.8812405395507814,
      "step": 121600
    },
    {
      "epoch": 0.07162021680054859,
      "grad_norm": 0.28554093837738037,
      "learning_rate": 0.00028925705737211244,
      "loss": 2.8396856689453127,
      "step": 121650
    },
    {
      "epoch": 0.07164965379882254,
      "grad_norm": 0.20806211233139038,
      "learning_rate": 0.00028925264182302126,
      "loss": 2.8478384399414063,
      "step": 121700
    },
    {
      "epoch": 0.07167909079709651,
      "grad_norm": 0.3553103506565094,
      "learning_rate": 0.00028924822627393,
      "loss": 2.896653747558594,
      "step": 121750
    },
    {
      "epoch": 0.07170852779537047,
      "grad_norm": 0.19647401571273804,
      "learning_rate": 0.00028924381072483883,
      "loss": 2.874300537109375,
      "step": 121800
    },
    {
      "epoch": 0.07173796479364443,
      "grad_norm": 0.28816792368888855,
      "learning_rate": 0.00028923939517574765,
      "loss": 2.8134475708007813,
      "step": 121850
    },
    {
      "epoch": 0.0717674017919184,
      "grad_norm": 0.23867583274841309,
      "learning_rate": 0.00028923497962665646,
      "loss": 2.9361538696289062,
      "step": 121900
    },
    {
      "epoch": 0.07179683879019236,
      "grad_norm": 0.30829787254333496,
      "learning_rate": 0.0002892305640775653,
      "loss": 2.8292776489257814,
      "step": 121950
    },
    {
      "epoch": 0.07182627578846631,
      "grad_norm": 0.3072403371334076,
      "learning_rate": 0.0002892261485284741,
      "loss": 2.8628909301757814,
      "step": 122000
    },
    {
      "epoch": 0.07185571278674029,
      "grad_norm": 0.32921382784843445,
      "learning_rate": 0.0002892217329793829,
      "loss": 2.821683349609375,
      "step": 122050
    },
    {
      "epoch": 0.07188514978501424,
      "grad_norm": 0.3130315840244293,
      "learning_rate": 0.0002892173174302917,
      "loss": 2.8113226318359374,
      "step": 122100
    },
    {
      "epoch": 0.0719145867832882,
      "grad_norm": 0.389350950717926,
      "learning_rate": 0.0002892129018812005,
      "loss": 2.8324261474609376,
      "step": 122150
    },
    {
      "epoch": 0.07194402378156217,
      "grad_norm": 0.27225130796432495,
      "learning_rate": 0.0002892084863321093,
      "loss": 2.8632101440429687,
      "step": 122200
    },
    {
      "epoch": 0.07197346077983613,
      "grad_norm": 0.22990313172340393,
      "learning_rate": 0.00028920407078301816,
      "loss": 2.947859802246094,
      "step": 122250
    },
    {
      "epoch": 0.07200289777811009,
      "grad_norm": 0.23364460468292236,
      "learning_rate": 0.0002891996552339269,
      "loss": 2.8518646240234373,
      "step": 122300
    },
    {
      "epoch": 0.07203233477638404,
      "grad_norm": 0.25505390763282776,
      "learning_rate": 0.00028919523968483574,
      "loss": 2.8119207763671876,
      "step": 122350
    },
    {
      "epoch": 0.07206177177465801,
      "grad_norm": 0.23360955715179443,
      "learning_rate": 0.00028919082413574455,
      "loss": 2.8035382080078124,
      "step": 122400
    },
    {
      "epoch": 0.07209120877293197,
      "grad_norm": 0.2584570050239563,
      "learning_rate": 0.00028918640858665337,
      "loss": 2.8273828125,
      "step": 122450
    },
    {
      "epoch": 0.07212064577120593,
      "grad_norm": 0.2888719439506531,
      "learning_rate": 0.00028918199303756213,
      "loss": 2.8736752319335936,
      "step": 122500
    },
    {
      "epoch": 0.0721500827694799,
      "grad_norm": 0.22253751754760742,
      "learning_rate": 0.000289177577488471,
      "loss": 2.880582580566406,
      "step": 122550
    },
    {
      "epoch": 0.07217951976775386,
      "grad_norm": 0.8330067992210388,
      "learning_rate": 0.0002891731619393798,
      "loss": 2.9030050659179687,
      "step": 122600
    },
    {
      "epoch": 0.07220895676602782,
      "grad_norm": 0.37593960762023926,
      "learning_rate": 0.00028916874639028857,
      "loss": 2.8821990966796873,
      "step": 122650
    },
    {
      "epoch": 0.07223839376430179,
      "grad_norm": 0.29836055636405945,
      "learning_rate": 0.0002891643308411974,
      "loss": 2.8055490112304686,
      "step": 122700
    },
    {
      "epoch": 0.07226783076257574,
      "grad_norm": 0.20414574444293976,
      "learning_rate": 0.0002891599152921062,
      "loss": 2.91170166015625,
      "step": 122750
    },
    {
      "epoch": 0.0722972677608497,
      "grad_norm": 0.2668229341506958,
      "learning_rate": 0.000289155499743015,
      "loss": 2.800729064941406,
      "step": 122800
    },
    {
      "epoch": 0.07232670475912367,
      "grad_norm": 0.23036088049411774,
      "learning_rate": 0.00028915108419392383,
      "loss": 2.849445495605469,
      "step": 122850
    },
    {
      "epoch": 0.07235614175739763,
      "grad_norm": 0.2707563638687134,
      "learning_rate": 0.00028914666864483264,
      "loss": 2.787015380859375,
      "step": 122900
    },
    {
      "epoch": 0.07238557875567159,
      "grad_norm": 0.3751641809940338,
      "learning_rate": 0.00028914225309574146,
      "loss": 2.8068228149414063,
      "step": 122950
    },
    {
      "epoch": 0.07241501575394556,
      "grad_norm": 0.2558959722518921,
      "learning_rate": 0.00028913783754665027,
      "loss": 2.8666146850585936,
      "step": 123000
    },
    {
      "epoch": 0.07244445275221952,
      "grad_norm": 0.24316062033176422,
      "learning_rate": 0.00028913342199755903,
      "loss": 2.843439636230469,
      "step": 123050
    },
    {
      "epoch": 0.07247388975049347,
      "grad_norm": 0.242561936378479,
      "learning_rate": 0.00028912900644846785,
      "loss": 2.8266802978515626,
      "step": 123100
    },
    {
      "epoch": 0.07250332674876743,
      "grad_norm": 0.2761647701263428,
      "learning_rate": 0.00028912459089937666,
      "loss": 2.879563293457031,
      "step": 123150
    },
    {
      "epoch": 0.0725327637470414,
      "grad_norm": 0.20979662239551544,
      "learning_rate": 0.0002891201753502855,
      "loss": 2.8128631591796873,
      "step": 123200
    },
    {
      "epoch": 0.07256220074531536,
      "grad_norm": 0.23031483590602875,
      "learning_rate": 0.0002891157598011943,
      "loss": 2.830730285644531,
      "step": 123250
    },
    {
      "epoch": 0.07259163774358932,
      "grad_norm": 0.2662459909915924,
      "learning_rate": 0.0002891113442521031,
      "loss": 2.853585205078125,
      "step": 123300
    },
    {
      "epoch": 0.07262107474186329,
      "grad_norm": 0.22141823172569275,
      "learning_rate": 0.0002891069287030119,
      "loss": 2.853438720703125,
      "step": 123350
    },
    {
      "epoch": 0.07265051174013724,
      "grad_norm": 0.26477885246276855,
      "learning_rate": 0.00028910251315392073,
      "loss": 2.9074777221679686,
      "step": 123400
    },
    {
      "epoch": 0.0726799487384112,
      "grad_norm": 0.22816206514835358,
      "learning_rate": 0.0002890980976048295,
      "loss": 2.804552001953125,
      "step": 123450
    },
    {
      "epoch": 0.07270938573668517,
      "grad_norm": 0.3188365399837494,
      "learning_rate": 0.0002890936820557383,
      "loss": 2.880643615722656,
      "step": 123500
    },
    {
      "epoch": 0.07273882273495913,
      "grad_norm": 0.28226208686828613,
      "learning_rate": 0.0002890892665066472,
      "loss": 2.8263165283203127,
      "step": 123550
    },
    {
      "epoch": 0.07276825973323309,
      "grad_norm": 0.2606245279312134,
      "learning_rate": 0.00028908485095755594,
      "loss": 2.878542785644531,
      "step": 123600
    },
    {
      "epoch": 0.07279769673150706,
      "grad_norm": 0.22438715398311615,
      "learning_rate": 0.00028908043540846475,
      "loss": 2.8915975952148436,
      "step": 123650
    },
    {
      "epoch": 0.07282713372978102,
      "grad_norm": 0.24177086353302002,
      "learning_rate": 0.00028907601985937357,
      "loss": 2.8857345581054688,
      "step": 123700
    },
    {
      "epoch": 0.07285657072805497,
      "grad_norm": 0.22701655328273773,
      "learning_rate": 0.0002890716043102824,
      "loss": 2.8925665283203124,
      "step": 123750
    },
    {
      "epoch": 0.07288600772632893,
      "grad_norm": 0.2921850383281708,
      "learning_rate": 0.0002890671887611912,
      "loss": 2.8622857666015626,
      "step": 123800
    },
    {
      "epoch": 0.0729154447246029,
      "grad_norm": 0.2943372428417206,
      "learning_rate": 0.0002890627732121,
      "loss": 2.866929626464844,
      "step": 123850
    },
    {
      "epoch": 0.07294488172287686,
      "grad_norm": 0.24243415892124176,
      "learning_rate": 0.0002890583576630088,
      "loss": 2.840870666503906,
      "step": 123900
    },
    {
      "epoch": 0.07297431872115082,
      "grad_norm": 0.3342321217060089,
      "learning_rate": 0.00028905394211391764,
      "loss": 2.820020751953125,
      "step": 123950
    },
    {
      "epoch": 0.07300375571942479,
      "grad_norm": 0.28595152497291565,
      "learning_rate": 0.0002890495265648264,
      "loss": 2.878775634765625,
      "step": 124000
    }
  ],
  "logging_steps": 50,
  "max_steps": 3397086,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.184958097063936e+19,
  "train_batch_size": 6,
  "trial_name": null,
  "trial_params": null
}