{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.540606328833813,
  "eval_steps": 500,
  "global_step": 36000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 9.835017580093924e-05,
      "grad_norm": 3.649945020675659,
      "learning_rate": 3.9292730844793716e-08,
      "loss": 1.0527,
      "step": 1
    },
    {
      "epoch": 0.0009835017580093924,
      "grad_norm": 11.982413291931152,
      "learning_rate": 3.9292730844793716e-07,
      "loss": 1.8048,
      "step": 10
    },
    {
      "epoch": 0.0019670035160187847,
      "grad_norm": 3.4894094467163086,
      "learning_rate": 7.858546168958743e-07,
      "loss": 1.7688,
      "step": 20
    },
    {
      "epoch": 0.0029505052740281773,
      "grad_norm": 3.314408779144287,
      "learning_rate": 1.1787819253438115e-06,
      "loss": 1.8063,
      "step": 30
    },
    {
      "epoch": 0.003934007032037569,
      "grad_norm": 3.1056833267211914,
      "learning_rate": 1.5717092337917486e-06,
      "loss": 2.0023,
      "step": 40
    },
    {
      "epoch": 0.004917508790046962,
      "grad_norm": 4.661435604095459,
      "learning_rate": 1.9646365422396858e-06,
      "loss": 2.06,
      "step": 50
    },
    {
      "epoch": 0.005901010548056355,
      "grad_norm": 2.0999605655670166,
      "learning_rate": 2.357563850687623e-06,
      "loss": 1.7816,
      "step": 60
    },
    {
      "epoch": 0.006884512306065747,
      "grad_norm": 5.44950008392334,
      "learning_rate": 2.7504911591355604e-06,
      "loss": 1.789,
      "step": 70
    },
    {
      "epoch": 0.007868014064075139,
      "grad_norm": 5.730320453643799,
      "learning_rate": 3.1434184675834973e-06,
      "loss": 1.5076,
      "step": 80
    },
    {
      "epoch": 0.008851515822084531,
      "grad_norm": 10.488595008850098,
      "learning_rate": 3.5363457760314346e-06,
      "loss": 1.6972,
      "step": 90
    },
    {
      "epoch": 0.009835017580093924,
      "grad_norm": 6.945803642272949,
      "learning_rate": 3.9292730844793715e-06,
      "loss": 1.343,
      "step": 100
    },
    {
      "epoch": 0.010818519338103317,
      "grad_norm": 6.9783830642700195,
      "learning_rate": 4.322200392927308e-06,
      "loss": 1.0908,
      "step": 110
    },
    {
      "epoch": 0.01180202109611271,
      "grad_norm": 2.9272944927215576,
      "learning_rate": 4.715127701375246e-06,
      "loss": 0.8392,
      "step": 120
    },
    {
      "epoch": 0.012785522854122102,
      "grad_norm": 3.644608736038208,
      "learning_rate": 5.108055009823183e-06,
      "loss": 0.6187,
      "step": 130
    },
    {
      "epoch": 0.013769024612131494,
      "grad_norm": 2.0485031604766846,
      "learning_rate": 5.500982318271121e-06,
      "loss": 0.7187,
      "step": 140
    },
    {
      "epoch": 0.014752526370140887,
      "grad_norm": 1.5527009963989258,
      "learning_rate": 5.893909626719058e-06,
      "loss": 0.536,
      "step": 150
    },
    {
      "epoch": 0.015736028128150278,
      "grad_norm": 1.9112921953201294,
      "learning_rate": 6.286836935166995e-06,
      "loss": 0.509,
      "step": 160
    },
    {
      "epoch": 0.01671952988615967,
      "grad_norm": 2.1568541526794434,
      "learning_rate": 6.6797642436149315e-06,
      "loss": 0.4402,
      "step": 170
    },
    {
      "epoch": 0.017703031644169063,
      "grad_norm": 0.6417804956436157,
      "learning_rate": 7.072691552062869e-06,
      "loss": 0.4267,
      "step": 180
    },
    {
      "epoch": 0.018686533402178455,
      "grad_norm": 1.3143049478530884,
      "learning_rate": 7.465618860510806e-06,
      "loss": 0.4131,
      "step": 190
    },
    {
      "epoch": 0.019670035160187848,
      "grad_norm": 1.5052763223648071,
      "learning_rate": 7.858546168958743e-06,
      "loss": 0.4092,
      "step": 200
    },
    {
      "epoch": 0.02065353691819724,
      "grad_norm": 1.0570193529129028,
      "learning_rate": 8.25147347740668e-06,
      "loss": 0.4324,
      "step": 210
    },
    {
      "epoch": 0.021637038676206633,
      "grad_norm": 1.710619330406189,
      "learning_rate": 8.644400785854617e-06,
      "loss": 0.4052,
      "step": 220
    },
    {
      "epoch": 0.022620540434216026,
      "grad_norm": 1.0562747716903687,
      "learning_rate": 9.037328094302554e-06,
      "loss": 0.5058,
      "step": 230
    },
    {
      "epoch": 0.02360404219222542,
      "grad_norm": 1.6072394847869873,
      "learning_rate": 9.430255402750492e-06,
      "loss": 0.5282,
      "step": 240
    },
    {
      "epoch": 0.02458754395023481,
      "grad_norm": 0.5480562448501587,
      "learning_rate": 9.82318271119843e-06,
      "loss": 0.3765,
      "step": 250
    },
    {
      "epoch": 0.025571045708244203,
      "grad_norm": 0.7658659219741821,
      "learning_rate": 1.0216110019646366e-05,
      "loss": 0.3856,
      "step": 260
    },
    {
      "epoch": 0.026554547466253596,
      "grad_norm": 1.19699227809906,
      "learning_rate": 1.0609037328094303e-05,
      "loss": 0.414,
      "step": 270
    },
    {
      "epoch": 0.02753804922426299,
      "grad_norm": 1.9998314380645752,
      "learning_rate": 1.1001964636542242e-05,
      "loss": 0.4558,
      "step": 280
    },
    {
      "epoch": 0.02852155098227238,
      "grad_norm": 0.4776442050933838,
      "learning_rate": 1.1394891944990178e-05,
      "loss": 0.4623,
      "step": 290
    },
    {
      "epoch": 0.029505052740281774,
      "grad_norm": 0.5863878726959229,
      "learning_rate": 1.1787819253438115e-05,
      "loss": 0.317,
      "step": 300
    },
    {
      "epoch": 0.030488554498291166,
      "grad_norm": 0.7499637007713318,
      "learning_rate": 1.2180746561886052e-05,
      "loss": 0.4085,
      "step": 310
    },
    {
      "epoch": 0.031472056256300555,
      "grad_norm": 0.7671499848365784,
      "learning_rate": 1.257367387033399e-05,
      "loss": 0.2961,
      "step": 320
    },
    {
      "epoch": 0.03245555801430995,
      "grad_norm": 0.7907651662826538,
      "learning_rate": 1.2966601178781926e-05,
      "loss": 0.3862,
      "step": 330
    },
    {
      "epoch": 0.03343905977231934,
      "grad_norm": 0.6451826691627502,
      "learning_rate": 1.3359528487229863e-05,
      "loss": 0.4217,
      "step": 340
    },
    {
      "epoch": 0.03442256153032874,
      "grad_norm": 0.42799535393714905,
      "learning_rate": 1.3752455795677802e-05,
      "loss": 0.3331,
      "step": 350
    },
    {
      "epoch": 0.035406063288338126,
      "grad_norm": 0.5752848386764526,
      "learning_rate": 1.4145383104125738e-05,
      "loss": 0.3665,
      "step": 360
    },
    {
      "epoch": 0.03638956504634752,
      "grad_norm": 0.5795350074768066,
      "learning_rate": 1.4538310412573675e-05,
      "loss": 0.3811,
      "step": 370
    },
    {
      "epoch": 0.03737306680435691,
      "grad_norm": 1.5207698345184326,
      "learning_rate": 1.4931237721021612e-05,
      "loss": 0.3787,
      "step": 380
    },
    {
      "epoch": 0.03835656856236631,
      "grad_norm": 1.3819210529327393,
      "learning_rate": 1.532416502946955e-05,
      "loss": 0.4208,
      "step": 390
    },
    {
      "epoch": 0.039340070320375696,
      "grad_norm": 0.7597386837005615,
      "learning_rate": 1.5717092337917486e-05,
      "loss": 0.335,
      "step": 400
    },
    {
      "epoch": 0.04032357207838509,
      "grad_norm": 0.39009803533554077,
      "learning_rate": 1.6110019646365423e-05,
      "loss": 0.3017,
      "step": 410
    },
    {
      "epoch": 0.04130707383639448,
      "grad_norm": 0.6308045983314514,
      "learning_rate": 1.650294695481336e-05,
      "loss": 0.3232,
      "step": 420
    },
    {
      "epoch": 0.04229057559440388,
      "grad_norm": 0.8969710469245911,
      "learning_rate": 1.6895874263261297e-05,
      "loss": 0.3076,
      "step": 430
    },
    {
      "epoch": 0.043274077352413266,
      "grad_norm": 0.6697237491607666,
      "learning_rate": 1.7288801571709234e-05,
      "loss": 0.3273,
      "step": 440
    },
    {
      "epoch": 0.04425757911042266,
      "grad_norm": 0.6771184802055359,
      "learning_rate": 1.768172888015717e-05,
      "loss": 0.3012,
      "step": 450
    },
    {
      "epoch": 0.04524108086843205,
      "grad_norm": 0.32029828429222107,
      "learning_rate": 1.8074656188605107e-05,
      "loss": 0.273,
      "step": 460
    },
    {
      "epoch": 0.04622458262644145,
      "grad_norm": 0.388177752494812,
      "learning_rate": 1.8467583497053048e-05,
      "loss": 0.3686,
      "step": 470
    },
    {
      "epoch": 0.04720808438445084,
      "grad_norm": 1.197744607925415,
      "learning_rate": 1.8860510805500985e-05,
      "loss": 0.3402,
      "step": 480
    },
    {
      "epoch": 0.04819158614246023,
      "grad_norm": 0.7969218492507935,
      "learning_rate": 1.925343811394892e-05,
      "loss": 0.3683,
      "step": 490
    },
    {
      "epoch": 0.04917508790046962,
      "grad_norm": 1.7514092922210693,
      "learning_rate": 1.964636542239686e-05,
      "loss": 0.3258,
      "step": 500
    },
    {
      "epoch": 0.04917508790046962,
      "eval_loss": 0.2306613028049469,
      "eval_runtime": 17.398,
      "eval_samples_per_second": 2.874,
      "eval_steps_per_second": 1.437,
      "step": 500
    },
    {
      "epoch": 0.05015858965847902,
      "grad_norm": 0.8328887820243835,
      "learning_rate": 1.999960259110599e-05,
      "loss": 0.3144,
      "step": 510
    },
    {
      "epoch": 0.05114209141648841,
      "grad_norm": 0.6266557574272156,
      "learning_rate": 1.999562850216588e-05,
      "loss": 0.3344,
      "step": 520
    },
    {
      "epoch": 0.052125593174497796,
      "grad_norm": 0.6225624084472656,
      "learning_rate": 1.999165441322577e-05,
      "loss": 0.2535,
      "step": 530
    },
    {
      "epoch": 0.05310909493250719,
      "grad_norm": 1.2749239206314087,
      "learning_rate": 1.998768032428566e-05,
      "loss": 0.2902,
      "step": 540
    },
    {
      "epoch": 0.05409259669051658,
      "grad_norm": 1.1305932998657227,
      "learning_rate": 1.998370623534555e-05,
      "loss": 0.333,
      "step": 550
    },
    {
      "epoch": 0.05507609844852598,
      "grad_norm": 0.9573833346366882,
      "learning_rate": 1.997973214640544e-05,
      "loss": 0.2581,
      "step": 560
    },
    {
      "epoch": 0.056059600206535366,
      "grad_norm": 0.5128825306892395,
      "learning_rate": 1.997575805746533e-05,
      "loss": 0.2887,
      "step": 570
    },
    {
      "epoch": 0.05704310196454476,
      "grad_norm": 0.887127697467804,
      "learning_rate": 1.9971783968525218e-05,
      "loss": 0.4077,
      "step": 580
    },
    {
      "epoch": 0.05802660372255415,
      "grad_norm": 2.3436291217803955,
      "learning_rate": 1.9967809879585108e-05,
      "loss": 0.2335,
      "step": 590
    },
    {
      "epoch": 0.05901010548056355,
      "grad_norm": 1.4168339967727661,
      "learning_rate": 1.9963835790644998e-05,
      "loss": 0.4166,
      "step": 600
    },
    {
      "epoch": 0.05999360723857294,
      "grad_norm": 0.6777006387710571,
      "learning_rate": 1.9959861701704887e-05,
      "loss": 0.3611,
      "step": 610
    },
    {
      "epoch": 0.06097710899658233,
      "grad_norm": 1.5377174615859985,
      "learning_rate": 1.9955887612764774e-05,
      "loss": 0.3877,
      "step": 620
    },
    {
      "epoch": 0.06196061075459172,
      "grad_norm": 1.4269481897354126,
      "learning_rate": 1.9951913523824663e-05,
      "loss": 0.3423,
      "step": 630
    },
    {
      "epoch": 0.06294411251260111,
      "grad_norm": 0.7028114795684814,
      "learning_rate": 1.9947939434884553e-05,
      "loss": 0.4658,
      "step": 640
    },
    {
      "epoch": 0.06392761427061051,
      "grad_norm": 0.6338093280792236,
      "learning_rate": 1.9943965345944446e-05,
      "loss": 0.3796,
      "step": 650
    },
    {
      "epoch": 0.0649111160286199,
      "grad_norm": 0.8471394777297974,
      "learning_rate": 1.9939991257004336e-05,
      "loss": 0.3291,
      "step": 660
    },
    {
      "epoch": 0.06589461778662929,
      "grad_norm": 1.587775707244873,
      "learning_rate": 1.9936017168064222e-05,
      "loss": 0.367,
      "step": 670
    },
    {
      "epoch": 0.06687811954463868,
      "grad_norm": 0.6585375070571899,
      "learning_rate": 1.993204307912411e-05,
      "loss": 0.3023,
      "step": 680
    },
    {
      "epoch": 0.06786162130264808,
      "grad_norm": 0.8488882184028625,
      "learning_rate": 1.9928068990184e-05,
      "loss": 0.3238,
      "step": 690
    },
    {
      "epoch": 0.06884512306065747,
      "grad_norm": 0.6817759275436401,
      "learning_rate": 1.992409490124389e-05,
      "loss": 0.228,
      "step": 700
    },
    {
      "epoch": 0.06982862481866686,
      "grad_norm": 1.6214189529418945,
      "learning_rate": 1.992012081230378e-05,
      "loss": 0.3284,
      "step": 710
    },
    {
      "epoch": 0.07081212657667625,
      "grad_norm": 1.3868869543075562,
      "learning_rate": 1.991614672336367e-05,
      "loss": 0.3909,
      "step": 720
    },
    {
      "epoch": 0.07179562833468565,
      "grad_norm": 1.374632716178894,
      "learning_rate": 1.991217263442356e-05,
      "loss": 0.3086,
      "step": 730
    },
    {
      "epoch": 0.07277913009269504,
      "grad_norm": 1.1037342548370361,
      "learning_rate": 1.990819854548345e-05,
      "loss": 0.4078,
      "step": 740
    },
    {
      "epoch": 0.07376263185070443,
      "grad_norm": 1.1040786504745483,
      "learning_rate": 1.990422445654334e-05,
      "loss": 0.3532,
      "step": 750
    },
    {
      "epoch": 0.07474613360871382,
      "grad_norm": 1.2953940629959106,
      "learning_rate": 1.990025036760323e-05,
      "loss": 0.3954,
      "step": 760
    },
    {
      "epoch": 0.07572963536672322,
      "grad_norm": 1.4143564701080322,
      "learning_rate": 1.989627627866312e-05,
      "loss": 0.4358,
      "step": 770
    },
    {
      "epoch": 0.07671313712473261,
      "grad_norm": 1.4138742685317993,
      "learning_rate": 1.9892302189723008e-05,
      "loss": 0.3119,
      "step": 780
    },
    {
      "epoch": 0.077696638882742,
      "grad_norm": 0.6395207643508911,
      "learning_rate": 1.9888328100782898e-05,
      "loss": 0.3049,
      "step": 790
    },
    {
      "epoch": 0.07868014064075139,
      "grad_norm": 1.8704555034637451,
      "learning_rate": 1.9884354011842784e-05,
      "loss": 0.368,
      "step": 800
    },
    {
      "epoch": 0.07966364239876078,
      "grad_norm": 0.8261243104934692,
      "learning_rate": 1.9880379922902677e-05,
      "loss": 0.3592,
      "step": 810
    },
    {
      "epoch": 0.08064714415677018,
      "grad_norm": 2.373087167739868,
      "learning_rate": 1.9876405833962567e-05,
      "loss": 0.2079,
      "step": 820
    },
    {
      "epoch": 0.08163064591477957,
      "grad_norm": 0.4923926591873169,
      "learning_rate": 1.9872431745022456e-05,
      "loss": 0.3669,
      "step": 830
    },
    {
      "epoch": 0.08261414767278896,
      "grad_norm": 2.16957688331604,
      "learning_rate": 1.9868457656082346e-05,
      "loss": 0.4158,
      "step": 840
    },
    {
      "epoch": 0.08359764943079835,
      "grad_norm": 1.1719224452972412,
      "learning_rate": 1.9864483567142236e-05,
      "loss": 0.3503,
      "step": 850
    },
    {
      "epoch": 0.08458115118880775,
      "grad_norm": 1.3015501499176025,
      "learning_rate": 1.9860509478202122e-05,
      "loss": 0.3569,
      "step": 860
    },
    {
      "epoch": 0.08556465294681714,
      "grad_norm": 1.0028891563415527,
      "learning_rate": 1.9856535389262012e-05,
      "loss": 0.3994,
      "step": 870
    },
    {
      "epoch": 0.08654815470482653,
      "grad_norm": 1.8605300188064575,
      "learning_rate": 1.98525613003219e-05,
      "loss": 0.382,
      "step": 880
    },
    {
      "epoch": 0.08753165646283592,
      "grad_norm": 0.8204103708267212,
      "learning_rate": 1.9848587211381794e-05,
      "loss": 0.2986,
      "step": 890
    },
    {
      "epoch": 0.08851515822084532,
      "grad_norm": 0.7822405695915222,
      "learning_rate": 1.9844613122441684e-05,
      "loss": 0.231,
      "step": 900
    },
    {
      "epoch": 0.08949865997885471,
      "grad_norm": 0.5924950242042542,
      "learning_rate": 1.984063903350157e-05,
      "loss": 0.3128,
      "step": 910
    },
    {
      "epoch": 0.0904821617368641,
      "grad_norm": 1.5780599117279053,
      "learning_rate": 1.983666494456146e-05,
      "loss": 0.2309,
      "step": 920
    },
    {
      "epoch": 0.09146566349487349,
      "grad_norm": 1.5509686470031738,
      "learning_rate": 1.983269085562135e-05,
      "loss": 0.3227,
      "step": 930
    },
    {
      "epoch": 0.0924491652528829,
      "grad_norm": 1.5279356241226196,
      "learning_rate": 1.982871676668124e-05,
      "loss": 0.3902,
      "step": 940
    },
    {
      "epoch": 0.09343266701089228,
      "grad_norm": 1.8690674304962158,
      "learning_rate": 1.982474267774113e-05,
      "loss": 0.2989,
      "step": 950
    },
    {
      "epoch": 0.09441616876890167,
      "grad_norm": 1.274799108505249,
      "learning_rate": 1.982076858880102e-05,
      "loss": 0.1906,
      "step": 960
    },
    {
      "epoch": 0.09539967052691106,
      "grad_norm": 0.8589666485786438,
      "learning_rate": 1.981679449986091e-05,
      "loss": 0.3481,
      "step": 970
    },
    {
      "epoch": 0.09638317228492047,
      "grad_norm": 1.0235803127288818,
      "learning_rate": 1.9812820410920798e-05,
      "loss": 0.3185,
      "step": 980
    },
    {
      "epoch": 0.09736667404292985,
      "grad_norm": 1.4823609590530396,
      "learning_rate": 1.9808846321980688e-05,
      "loss": 0.4237,
      "step": 990
    },
    {
      "epoch": 0.09835017580093924,
      "grad_norm": 1.3744195699691772,
      "learning_rate": 1.9804872233040577e-05,
      "loss": 0.2549,
      "step": 1000
    },
    {
      "epoch": 0.09835017580093924,
      "eval_loss": 0.20229017734527588,
      "eval_runtime": 16.1384,
      "eval_samples_per_second": 3.098,
      "eval_steps_per_second": 1.549,
      "step": 1000
    },
    {
      "epoch": 0.09933367755894863,
      "grad_norm": 2.249887466430664,
      "learning_rate": 1.9800898144100467e-05,
      "loss": 0.4879,
      "step": 1010
    },
    {
      "epoch": 0.10031717931695804,
      "grad_norm": 1.4358183145523071,
      "learning_rate": 1.9796924055160357e-05,
      "loss": 0.3384,
      "step": 1020
    },
    {
      "epoch": 0.10130068107496742,
      "grad_norm": 0.7850083708763123,
      "learning_rate": 1.9792949966220246e-05,
      "loss": 0.4727,
      "step": 1030
    },
    {
      "epoch": 0.10228418283297681,
      "grad_norm": 1.24857759475708,
      "learning_rate": 1.9788975877280133e-05,
      "loss": 0.2594,
      "step": 1040
    },
    {
      "epoch": 0.1032676845909862,
      "grad_norm": 0.6814334392547607,
      "learning_rate": 1.9785001788340022e-05,
      "loss": 0.3108,
      "step": 1050
    },
    {
      "epoch": 0.10425118634899559,
      "grad_norm": 0.8959739208221436,
      "learning_rate": 1.9781027699399915e-05,
      "loss": 0.3643,
      "step": 1060
    },
    {
      "epoch": 0.105234688107005,
      "grad_norm": 1.1151331663131714,
      "learning_rate": 1.9777053610459805e-05,
      "loss": 0.4537,
      "step": 1070
    },
    {
      "epoch": 0.10621818986501438,
      "grad_norm": 0.7030103802680969,
      "learning_rate": 1.9773079521519695e-05,
      "loss": 0.2592,
      "step": 1080
    },
    {
      "epoch": 0.10720169162302377,
      "grad_norm": 0.9018835425376892,
      "learning_rate": 1.9769105432579584e-05,
      "loss": 0.3518,
      "step": 1090
    },
    {
      "epoch": 0.10818519338103316,
      "grad_norm": 0.46352618932724,
      "learning_rate": 1.976513134363947e-05,
      "loss": 0.3255,
      "step": 1100
    },
    {
      "epoch": 0.10916869513904257,
      "grad_norm": 1.0427846908569336,
      "learning_rate": 1.976115725469936e-05,
      "loss": 0.3765,
      "step": 1110
    },
    {
      "epoch": 0.11015219689705195,
      "grad_norm": 0.5563482642173767,
      "learning_rate": 1.975718316575925e-05,
      "loss": 0.2775,
      "step": 1120
    },
    {
      "epoch": 0.11113569865506134,
      "grad_norm": 1.104644536972046,
      "learning_rate": 1.975320907681914e-05,
      "loss": 0.2524,
      "step": 1130
    },
    {
      "epoch": 0.11211920041307073,
      "grad_norm": 0.8505045175552368,
      "learning_rate": 1.9749234987879033e-05,
      "loss": 0.3474,
      "step": 1140
    },
    {
      "epoch": 0.11310270217108014,
      "grad_norm": 0.6132721304893494,
      "learning_rate": 1.974526089893892e-05,
      "loss": 0.2291,
      "step": 1150
    },
    {
      "epoch": 0.11408620392908952,
      "grad_norm": 1.056565284729004,
      "learning_rate": 1.974128680999881e-05,
      "loss": 0.3622,
      "step": 1160
    },
    {
      "epoch": 0.11506970568709891,
      "grad_norm": 1.2748444080352783,
      "learning_rate": 1.97373127210587e-05,
      "loss": 0.3838,
      "step": 1170
    },
    {
      "epoch": 0.1160532074451083,
      "grad_norm": 0.8781066536903381,
      "learning_rate": 1.9733338632118588e-05,
      "loss": 0.3436,
      "step": 1180
    },
    {
      "epoch": 0.1170367092031177,
      "grad_norm": 1.7326856851577759,
      "learning_rate": 1.9729364543178478e-05,
      "loss": 0.3075,
      "step": 1190
    },
    {
      "epoch": 0.1180202109611271,
      "grad_norm": 1.2784039974212646,
      "learning_rate": 1.9725390454238367e-05,
      "loss": 0.4899,
      "step": 1200
    },
    {
      "epoch": 0.11900371271913648,
      "grad_norm": 2.0357894897460938,
      "learning_rate": 1.9721416365298257e-05,
      "loss": 0.3078,
      "step": 1210
    },
    {
      "epoch": 0.11998721447714587,
      "grad_norm": 0.5481784343719482,
      "learning_rate": 1.9717442276358147e-05,
      "loss": 0.2839,
      "step": 1220
    },
    {
      "epoch": 0.12097071623515528,
      "grad_norm": 1.6970115900039673,
      "learning_rate": 1.9713468187418036e-05,
      "loss": 0.3257,
      "step": 1230
    },
    {
      "epoch": 0.12195421799316467,
      "grad_norm": 1.4209277629852295,
      "learning_rate": 1.9709494098477926e-05,
      "loss": 0.4143,
      "step": 1240
    },
    {
      "epoch": 0.12293771975117405,
      "grad_norm": 1.3740328550338745,
      "learning_rate": 1.9705520009537816e-05,
      "loss": 0.3004,
      "step": 1250
    },
    {
      "epoch": 0.12392122150918344,
      "grad_norm": 1.2223352193832397,
      "learning_rate": 1.9701545920597705e-05,
      "loss": 0.2664,
      "step": 1260
    },
    {
      "epoch": 0.12490472326719285,
      "grad_norm": 1.5225772857666016,
      "learning_rate": 1.9697571831657595e-05,
      "loss": 0.2847,
      "step": 1270
    },
    {
      "epoch": 0.12588822502520222,
      "grad_norm": 1.2129939794540405,
      "learning_rate": 1.969359774271748e-05,
      "loss": 0.3743,
      "step": 1280
    },
    {
      "epoch": 0.12687172678321162,
      "grad_norm": 1.3428657054901123,
      "learning_rate": 1.968962365377737e-05,
      "loss": 0.2888,
      "step": 1290
    },
    {
      "epoch": 0.12785522854122103,
      "grad_norm": 1.0801212787628174,
      "learning_rate": 1.9685649564837264e-05,
      "loss": 0.2564,
      "step": 1300
    },
    {
      "epoch": 0.1288387302992304,
      "grad_norm": 0.9565796256065369,
      "learning_rate": 1.9681675475897154e-05,
      "loss": 0.2988,
      "step": 1310
    },
    {
      "epoch": 0.1298222320572398,
      "grad_norm": 1.2454416751861572,
      "learning_rate": 1.9677701386957043e-05,
      "loss": 0.2863,
      "step": 1320
    },
    {
      "epoch": 0.13080573381524918,
      "grad_norm": 0.6081139445304871,
      "learning_rate": 1.9673727298016933e-05,
      "loss": 0.2704,
      "step": 1330
    },
    {
      "epoch": 0.13178923557325858,
      "grad_norm": 0.44817450642585754,
      "learning_rate": 1.966975320907682e-05,
      "loss": 0.3165,
      "step": 1340
    },
    {
      "epoch": 0.132772737331268,
      "grad_norm": 1.6914284229278564,
      "learning_rate": 1.966577912013671e-05,
      "loss": 0.286,
      "step": 1350
    },
    {
      "epoch": 0.13375623908927736,
      "grad_norm": 0.9408580660820007,
      "learning_rate": 1.96618050311966e-05,
      "loss": 0.2856,
      "step": 1360
    },
    {
      "epoch": 0.13473974084728677,
      "grad_norm": 0.3792386054992676,
      "learning_rate": 1.9657830942256488e-05,
      "loss": 0.3859,
      "step": 1370
    },
    {
      "epoch": 0.13572324260529617,
      "grad_norm": 0.9228833913803101,
      "learning_rate": 1.965385685331638e-05,
      "loss": 0.2832,
      "step": 1380
    },
    {
      "epoch": 0.13670674436330554,
      "grad_norm": 1.593618392944336,
      "learning_rate": 1.9649882764376268e-05,
      "loss": 0.2249,
      "step": 1390
    },
    {
      "epoch": 0.13769024612131495,
      "grad_norm": 1.7434206008911133,
      "learning_rate": 1.9645908675436157e-05,
      "loss": 0.2834,
      "step": 1400
    },
    {
      "epoch": 0.13867374787932432,
      "grad_norm": 1.7918550968170166,
      "learning_rate": 1.9641934586496047e-05,
      "loss": 0.3431,
      "step": 1410
    },
    {
      "epoch": 0.13965724963733372,
      "grad_norm": 1.3712127208709717,
      "learning_rate": 1.9637960497555937e-05,
      "loss": 0.3033,
      "step": 1420
    },
    {
      "epoch": 0.14064075139534313,
      "grad_norm": 2.3793137073516846,
      "learning_rate": 1.9633986408615826e-05,
      "loss": 0.3397,
      "step": 1430
    },
    {
      "epoch": 0.1416242531533525,
      "grad_norm": 0.6082277894020081,
      "learning_rate": 1.9630012319675716e-05,
      "loss": 0.2817,
      "step": 1440
    },
    {
      "epoch": 0.1426077549113619,
      "grad_norm": 1.1378542184829712,
      "learning_rate": 1.9626038230735606e-05,
      "loss": 0.4318,
      "step": 1450
    },
    {
      "epoch": 0.1435912566693713,
      "grad_norm": 0.5686970353126526,
      "learning_rate": 1.9622064141795495e-05,
      "loss": 0.2451,
      "step": 1460
    },
    {
      "epoch": 0.14457475842738068,
      "grad_norm": 0.6625441312789917,
      "learning_rate": 1.9618090052855385e-05,
      "loss": 0.2514,
      "step": 1470
    },
    {
      "epoch": 0.1455582601853901,
      "grad_norm": 0.542182445526123,
      "learning_rate": 1.9614115963915275e-05,
      "loss": 0.3149,
      "step": 1480
    },
    {
      "epoch": 0.14654176194339946,
      "grad_norm": 0.3858255445957184,
      "learning_rate": 1.9610141874975164e-05,
      "loss": 0.2722,
      "step": 1490
    },
    {
      "epoch": 0.14752526370140887,
      "grad_norm": 0.773241400718689,
      "learning_rate": 1.9606167786035054e-05,
      "loss": 0.2475,
      "step": 1500
    },
    {
      "epoch": 0.14752526370140887,
      "eval_loss": 0.19610656797885895,
      "eval_runtime": 19.5568,
      "eval_samples_per_second": 2.557,
      "eval_steps_per_second": 1.278,
      "step": 1500
    },
    {
      "epoch": 0.14850876545941827,
      "grad_norm": 0.9091148376464844,
      "learning_rate": 1.9602193697094944e-05,
      "loss": 0.3697,
      "step": 1510
    },
    {
      "epoch": 0.14949226721742764,
      "grad_norm": 1.2830495834350586,
      "learning_rate": 1.959821960815483e-05,
      "loss": 0.3447,
      "step": 1520
    },
    {
      "epoch": 0.15047576897543705,
      "grad_norm": 1.6588833332061768,
      "learning_rate": 1.959424551921472e-05,
      "loss": 0.2362,
      "step": 1530
    },
    {
      "epoch": 0.15145927073344645,
      "grad_norm": 1.5319061279296875,
      "learning_rate": 1.9590271430274613e-05,
      "loss": 0.2416,
      "step": 1540
    },
    {
      "epoch": 0.15244277249145582,
      "grad_norm": 0.7510396838188171,
      "learning_rate": 1.9586297341334502e-05,
      "loss": 0.2539,
      "step": 1550
    },
    {
      "epoch": 0.15342627424946523,
      "grad_norm": 1.2139660120010376,
      "learning_rate": 1.9582323252394392e-05,
      "loss": 0.3512,
      "step": 1560
    },
    {
      "epoch": 0.1544097760074746,
      "grad_norm": 1.344325304031372,
      "learning_rate": 1.957834916345428e-05,
      "loss": 0.2667,
      "step": 1570
    },
    {
      "epoch": 0.155393277765484,
      "grad_norm": 1.6625233888626099,
      "learning_rate": 1.9574375074514168e-05,
      "loss": 0.2317,
      "step": 1580
    },
    {
      "epoch": 0.1563767795234934,
      "grad_norm": 1.9059685468673706,
      "learning_rate": 1.9570400985574057e-05,
      "loss": 0.2561,
      "step": 1590
    },
    {
      "epoch": 0.15736028128150278,
      "grad_norm": 0.7797982692718506,
      "learning_rate": 1.9566426896633947e-05,
      "loss": 0.3168,
      "step": 1600
    },
    {
      "epoch": 0.1583437830395122,
      "grad_norm": 0.5057277679443359,
      "learning_rate": 1.9562452807693837e-05,
      "loss": 0.2496,
      "step": 1610
    },
    {
      "epoch": 0.15932728479752156,
      "grad_norm": 1.5704916715621948,
      "learning_rate": 1.955847871875373e-05,
      "loss": 0.3539,
      "step": 1620
    },
    {
      "epoch": 0.16031078655553097,
      "grad_norm": 1.0662189722061157,
      "learning_rate": 1.9554504629813616e-05,
      "loss": 0.2697,
      "step": 1630
    },
    {
      "epoch": 0.16129428831354037,
      "grad_norm": 1.3776342868804932,
      "learning_rate": 1.9550530540873506e-05,
      "loss": 0.3037,
      "step": 1640
    },
    {
      "epoch": 0.16227779007154974,
      "grad_norm": 1.1762497425079346,
      "learning_rate": 1.9546556451933395e-05,
      "loss": 0.2372,
      "step": 1650
    },
    {
      "epoch": 0.16326129182955915,
      "grad_norm": 0.9527464509010315,
      "learning_rate": 1.9542582362993285e-05,
      "loss": 0.3551,
      "step": 1660
    },
    {
      "epoch": 0.16424479358756855,
      "grad_norm": 1.6634376049041748,
      "learning_rate": 1.9538608274053175e-05,
      "loss": 0.3816,
      "step": 1670
    },
    {
      "epoch": 0.16522829534557792,
      "grad_norm": 1.6701467037200928,
      "learning_rate": 1.9534634185113064e-05,
      "loss": 0.2199,
      "step": 1680
    },
    {
      "epoch": 0.16621179710358733,
      "grad_norm": 1.0209531784057617,
      "learning_rate": 1.9530660096172954e-05,
      "loss": 0.3532,
      "step": 1690
    },
    {
      "epoch": 0.1671952988615967,
      "grad_norm": 0.8011378049850464,
      "learning_rate": 1.9526686007232844e-05,
      "loss": 0.3637,
      "step": 1700
    },
    {
      "epoch": 0.1681788006196061,
      "grad_norm": 0.581768274307251,
      "learning_rate": 1.9522711918292733e-05,
      "loss": 0.3278,
      "step": 1710
    },
    {
      "epoch": 0.1691623023776155,
      "grad_norm": 1.9274336099624634,
      "learning_rate": 1.9518737829352623e-05,
      "loss": 0.236,
      "step": 1720
    },
    {
      "epoch": 0.17014580413562488,
      "grad_norm": 1.341434359550476,
      "learning_rate": 1.9514763740412513e-05,
      "loss": 0.2646,
      "step": 1730
    },
    {
      "epoch": 0.1711293058936343,
      "grad_norm": 1.8525949716567993,
      "learning_rate": 1.9510789651472402e-05,
      "loss": 0.3428,
      "step": 1740
    },
    {
      "epoch": 0.1721128076516437,
      "grad_norm": 1.5868077278137207,
      "learning_rate": 1.9506815562532292e-05,
      "loss": 0.3762,
      "step": 1750
    },
    {
      "epoch": 0.17309630940965307,
      "grad_norm": 0.9035124778747559,
      "learning_rate": 1.950284147359218e-05,
      "loss": 0.275,
      "step": 1760
    },
    {
      "epoch": 0.17407981116766247,
      "grad_norm": 1.7473918199539185,
      "learning_rate": 1.9498867384652068e-05,
      "loss": 0.2784,
      "step": 1770
    },
    {
      "epoch": 0.17506331292567184,
      "grad_norm": 1.0584453344345093,
      "learning_rate": 1.9494893295711958e-05,
      "loss": 0.3078,
      "step": 1780
    },
    {
      "epoch": 0.17604681468368125,
      "grad_norm": 0.5479601621627808,
      "learning_rate": 1.949091920677185e-05,
      "loss": 0.3793,
      "step": 1790
    },
    {
      "epoch": 0.17703031644169065,
      "grad_norm": 0.8641221523284912,
      "learning_rate": 1.948694511783174e-05,
      "loss": 0.4181,
      "step": 1800
    },
    {
      "epoch": 0.17801381819970002,
      "grad_norm": 0.9480550289154053,
      "learning_rate": 1.948297102889163e-05,
      "loss": 0.4344,
      "step": 1810
    },
    {
      "epoch": 0.17899731995770943,
      "grad_norm": 1.129366397857666,
      "learning_rate": 1.9478996939951516e-05,
      "loss": 0.3321,
      "step": 1820
    },
    {
      "epoch": 0.17998082171571883,
      "grad_norm": 1.2757154703140259,
      "learning_rate": 1.9475022851011406e-05,
      "loss": 0.3701,
      "step": 1830
    },
    {
      "epoch": 0.1809643234737282,
      "grad_norm": 1.2727177143096924,
      "learning_rate": 1.9471048762071296e-05,
      "loss": 0.2436,
      "step": 1840
    },
    {
      "epoch": 0.1819478252317376,
      "grad_norm": 2.6235814094543457,
      "learning_rate": 1.9467074673131185e-05,
      "loss": 0.2863,
      "step": 1850
    },
    {
      "epoch": 0.18293132698974698,
      "grad_norm": 2.299156904220581,
      "learning_rate": 1.9463100584191075e-05,
      "loss": 0.3019,
      "step": 1860
    },
    {
      "epoch": 0.1839148287477564,
      "grad_norm": 0.4710278809070587,
      "learning_rate": 1.9459126495250965e-05,
      "loss": 0.2553,
      "step": 1870
    },
    {
      "epoch": 0.1848983305057658,
      "grad_norm": 0.47247934341430664,
      "learning_rate": 1.9455152406310854e-05,
      "loss": 0.3571,
      "step": 1880
    },
    {
      "epoch": 0.18588183226377517,
      "grad_norm": 0.878594160079956,
      "learning_rate": 1.9451178317370744e-05,
      "loss": 0.1848,
      "step": 1890
    },
    {
      "epoch": 0.18686533402178457,
      "grad_norm": 0.7783956527709961,
      "learning_rate": 1.9447204228430634e-05,
      "loss": 0.2342,
      "step": 1900
    },
    {
      "epoch": 0.18784883577979394,
      "grad_norm": 0.9626131057739258,
      "learning_rate": 1.9443230139490523e-05,
      "loss": 0.2647,
      "step": 1910
    },
    {
      "epoch": 0.18883233753780335,
      "grad_norm": 1.0547096729278564,
      "learning_rate": 1.9439256050550413e-05,
      "loss": 0.2782,
      "step": 1920
    },
    {
      "epoch": 0.18981583929581275,
      "grad_norm": 1.5676674842834473,
      "learning_rate": 1.9435281961610303e-05,
      "loss": 0.3556,
      "step": 1930
    },
    {
      "epoch": 0.19079934105382212,
      "grad_norm": 0.7233918309211731,
      "learning_rate": 1.9431307872670192e-05,
      "loss": 0.2339,
      "step": 1940
    },
    {
      "epoch": 0.19178284281183153,
      "grad_norm": 2.287632465362549,
      "learning_rate": 1.9427333783730082e-05,
      "loss": 0.273,
      "step": 1950
    },
    {
      "epoch": 0.19276634456984093,
      "grad_norm": 0.5366862416267395,
      "learning_rate": 1.942335969478997e-05,
      "loss": 0.2995,
      "step": 1960
    },
    {
      "epoch": 0.1937498463278503,
      "grad_norm": 1.5294963121414185,
      "learning_rate": 1.941938560584986e-05,
      "loss": 0.2755,
      "step": 1970
    },
    {
      "epoch": 0.1947333480858597,
      "grad_norm": 1.1591711044311523,
      "learning_rate": 1.941541151690975e-05,
      "loss": 0.3569,
      "step": 1980
    },
    {
      "epoch": 0.19571684984386908,
      "grad_norm": 1.1852513551712036,
      "learning_rate": 1.941143742796964e-05,
      "loss": 0.3179,
      "step": 1990
    },
    {
      "epoch": 0.1967003516018785,
      "grad_norm": 1.4614847898483276,
      "learning_rate": 1.9407463339029527e-05,
      "loss": 0.1913,
      "step": 2000
    },
    {
      "epoch": 0.1967003516018785,
      "eval_loss": 0.18594306707382202,
      "eval_runtime": 18.3957,
      "eval_samples_per_second": 2.718,
      "eval_steps_per_second": 1.359,
      "step": 2000
    },
    {
      "epoch": 0.1976838533598879,
      "grad_norm": 2.192962884902954,
      "learning_rate": 1.9403489250089417e-05,
      "loss": 0.2024,
      "step": 2010
    },
    {
      "epoch": 0.19866735511789727,
      "grad_norm": 0.4256015419960022,
      "learning_rate": 1.9399515161149306e-05,
      "loss": 0.2221,
      "step": 2020
    },
    {
      "epoch": 0.19965085687590667,
      "grad_norm": 1.269654393196106,
      "learning_rate": 1.93955410722092e-05,
      "loss": 0.2569,
      "step": 2030
    },
    {
      "epoch": 0.20063435863391607,
      "grad_norm": 1.8245949745178223,
      "learning_rate": 1.939156698326909e-05,
      "loss": 0.3545,
      "step": 2040
    },
    {
      "epoch": 0.20161786039192545,
      "grad_norm": 1.3251378536224365,
      "learning_rate": 1.938759289432898e-05,
      "loss": 0.257,
      "step": 2050
    },
    {
      "epoch": 0.20260136214993485,
      "grad_norm": 1.2136321067810059,
      "learning_rate": 1.9383618805388865e-05,
      "loss": 0.2392,
      "step": 2060
    },
    {
      "epoch": 0.20358486390794422,
      "grad_norm": 1.2105964422225952,
      "learning_rate": 1.9379644716448755e-05,
      "loss": 0.4406,
      "step": 2070
    },
    {
      "epoch": 0.20456836566595363,
      "grad_norm": 0.9377761483192444,
      "learning_rate": 1.9375670627508644e-05,
      "loss": 0.2832,
      "step": 2080
    },
    {
      "epoch": 0.20555186742396303,
      "grad_norm": 1.6811449527740479,
      "learning_rate": 1.9371696538568534e-05,
      "loss": 0.3275,
      "step": 2090
    },
    {
      "epoch": 0.2065353691819724,
      "grad_norm": 1.4369457960128784,
      "learning_rate": 1.9367722449628424e-05,
      "loss": 0.3204,
      "step": 2100
    },
    {
      "epoch": 0.2075188709399818,
      "grad_norm": 0.8904029130935669,
      "learning_rate": 1.9363748360688313e-05,
      "loss": 0.2644,
      "step": 2110
    },
    {
      "epoch": 0.20850237269799118,
      "grad_norm": 0.8172305822372437,
      "learning_rate": 1.9359774271748203e-05,
      "loss": 0.2975,
      "step": 2120
    },
    {
      "epoch": 0.2094858744560006,
      "grad_norm": 1.6503185033798218,
      "learning_rate": 1.9355800182808093e-05,
      "loss": 0.2737,
      "step": 2130
    },
    {
      "epoch": 0.21046937621401,
      "grad_norm": 2.1273269653320312,
      "learning_rate": 1.9351826093867982e-05,
      "loss": 0.302,
      "step": 2140
    },
    {
      "epoch": 0.21145287797201937,
      "grad_norm": 1.331870436668396,
      "learning_rate": 1.9347852004927872e-05,
      "loss": 0.3874,
      "step": 2150
    },
    {
      "epoch": 0.21243637973002877,
      "grad_norm": 0.8876633048057556,
      "learning_rate": 1.934387791598776e-05,
      "loss": 0.2727,
      "step": 2160
    },
    {
      "epoch": 0.21341988148803817,
      "grad_norm": 1.1674466133117676,
      "learning_rate": 1.933990382704765e-05,
      "loss": 0.3417,
      "step": 2170
    },
    {
      "epoch": 0.21440338324604755,
      "grad_norm": 1.6989260911941528,
      "learning_rate": 1.933592973810754e-05,
      "loss": 0.387,
      "step": 2180
    },
    {
      "epoch": 0.21538688500405695,
      "grad_norm": 1.3016669750213623,
      "learning_rate": 1.933195564916743e-05,
      "loss": 0.3226,
      "step": 2190
    },
    {
      "epoch": 0.21637038676206632,
      "grad_norm": 0.5776589512825012,
      "learning_rate": 1.932798156022732e-05,
      "loss": 0.2674,
      "step": 2200
    },
    {
      "epoch": 0.21735388852007573,
      "grad_norm": 0.7415688037872314,
      "learning_rate": 1.932400747128721e-05,
      "loss": 0.2484,
      "step": 2210
    },
    {
      "epoch": 0.21833739027808513,
      "grad_norm": 1.439495325088501,
      "learning_rate": 1.93200333823471e-05,
      "loss": 0.3143,
      "step": 2220
    },
    {
      "epoch": 0.2193208920360945,
      "grad_norm": 2.2038726806640625,
      "learning_rate": 1.931605929340699e-05,
      "loss": 0.2642,
      "step": 2230
    },
    {
      "epoch": 0.2203043937941039,
      "grad_norm": 0.8493722081184387,
      "learning_rate": 1.9312085204466875e-05,
      "loss": 0.2897,
      "step": 2240
    },
    {
      "epoch": 0.2212878955521133,
      "grad_norm": 2.087738037109375,
      "learning_rate": 1.9308111115526765e-05,
      "loss": 0.3154,
      "step": 2250
    },
    {
      "epoch": 0.2222713973101227,
      "grad_norm": 1.0796573162078857,
      "learning_rate": 1.9304137026586655e-05,
      "loss": 0.2283,
      "step": 2260
    },
    {
      "epoch": 0.2232548990681321,
      "grad_norm": 1.657051920890808,
      "learning_rate": 1.9300162937646548e-05,
      "loss": 0.2957,
      "step": 2270
    },
    {
      "epoch": 0.22423840082614147,
      "grad_norm": 0.8360177278518677,
      "learning_rate": 1.9296188848706438e-05,
      "loss": 0.2801,
      "step": 2280
    },
    {
      "epoch": 0.22522190258415087,
      "grad_norm": 0.4915439784526825,
      "learning_rate": 1.9292214759766327e-05,
      "loss": 0.2689,
      "step": 2290
    },
    {
      "epoch": 0.22620540434216027,
      "grad_norm": 0.8634927868843079,
      "learning_rate": 1.9288240670826213e-05,
      "loss": 0.2637,
      "step": 2300
    },
    {
      "epoch": 0.22718890610016965,
      "grad_norm": 0.7675266265869141,
      "learning_rate": 1.9284266581886103e-05,
      "loss": 0.3316,
      "step": 2310
    },
    {
      "epoch": 0.22817240785817905,
      "grad_norm": 1.2641220092773438,
      "learning_rate": 1.9280292492945993e-05,
      "loss": 0.3831,
      "step": 2320
    },
    {
      "epoch": 0.22915590961618845,
      "grad_norm": 1.1142165660858154,
      "learning_rate": 1.9276318404005882e-05,
      "loss": 0.324,
      "step": 2330
    },
    {
      "epoch": 0.23013941137419783,
      "grad_norm": 1.5316585302352905,
      "learning_rate": 1.9272344315065772e-05,
      "loss": 0.3329,
      "step": 2340
    },
    {
      "epoch": 0.23112291313220723,
      "grad_norm": 2.2005603313446045,
      "learning_rate": 1.9268370226125662e-05,
      "loss": 0.4578,
      "step": 2350
    },
    {
      "epoch": 0.2321064148902166,
      "grad_norm": 0.6494570374488831,
      "learning_rate": 1.926439613718555e-05,
      "loss": 0.2819,
      "step": 2360
    },
    {
      "epoch": 0.233089916648226,
      "grad_norm": 1.4297327995300293,
      "learning_rate": 1.926042204824544e-05,
      "loss": 0.2775,
      "step": 2370
    },
    {
      "epoch": 0.2340734184062354,
      "grad_norm": 1.6414613723754883,
      "learning_rate": 1.925644795930533e-05,
      "loss": 0.3015,
      "step": 2380
    },
    {
      "epoch": 0.2350569201642448,
      "grad_norm": 1.6433988809585571,
      "learning_rate": 1.925247387036522e-05,
      "loss": 0.2282,
      "step": 2390
    },
    {
      "epoch": 0.2360404219222542,
      "grad_norm": 1.0170117616653442,
      "learning_rate": 1.924849978142511e-05,
      "loss": 0.2157,
      "step": 2400
    },
    {
      "epoch": 0.23702392368026357,
      "grad_norm": 0.700157880783081,
      "learning_rate": 1.9244525692485e-05,
      "loss": 0.3026,
      "step": 2410
    },
    {
      "epoch": 0.23800742543827297,
      "grad_norm": 1.600882887840271,
      "learning_rate": 1.924055160354489e-05,
      "loss": 0.2904,
      "step": 2420
    },
    {
      "epoch": 0.23899092719628237,
      "grad_norm": 1.2791134119033813,
      "learning_rate": 1.923657751460478e-05,
      "loss": 0.1481,
      "step": 2430
    },
    {
      "epoch": 0.23997442895429175,
      "grad_norm": 0.642951488494873,
      "learning_rate": 1.923260342566467e-05,
      "loss": 0.2284,
      "step": 2440
    },
    {
      "epoch": 0.24095793071230115,
      "grad_norm": 1.4791512489318848,
      "learning_rate": 1.922862933672456e-05,
      "loss": 0.4285,
      "step": 2450
    },
    {
      "epoch": 0.24194143247031055,
      "grad_norm": 0.9878162145614624,
      "learning_rate": 1.9224655247784448e-05,
      "loss": 0.2794,
      "step": 2460
    },
    {
      "epoch": 0.24292493422831993,
      "grad_norm": 1.7200347185134888,
      "learning_rate": 1.9220681158844338e-05,
      "loss": 0.2782,
      "step": 2470
    },
    {
      "epoch": 0.24390843598632933,
      "grad_norm": 0.8549171686172485,
      "learning_rate": 1.9216707069904224e-05,
      "loss": 0.3878,
      "step": 2480
    },
    {
      "epoch": 0.2448919377443387,
      "grad_norm": 0.695548951625824,
      "learning_rate": 1.9212732980964114e-05,
      "loss": 0.3212,
      "step": 2490
    },
    {
      "epoch": 0.2458754395023481,
      "grad_norm": 0.5230987668037415,
      "learning_rate": 1.9208758892024003e-05,
      "loss": 0.2651,
      "step": 2500
    },
    {
      "epoch": 0.2458754395023481,
      "eval_loss": 0.1891545057296753,
      "eval_runtime": 15.2446,
      "eval_samples_per_second": 3.28,
      "eval_steps_per_second": 1.64,
      "step": 2500
    },
    {
      "epoch": 0.2468589412603575,
      "grad_norm": 1.109619379043579,
      "learning_rate": 1.9204784803083893e-05,
      "loss": 0.3115,
      "step": 2510
    },
    {
      "epoch": 0.2478424430183669,
      "grad_norm": 1.729632019996643,
      "learning_rate": 1.9200810714143786e-05,
      "loss": 0.3126,
      "step": 2520
    },
    {
      "epoch": 0.2488259447763763,
      "grad_norm": 2.7448902130126953,
      "learning_rate": 1.9196836625203676e-05,
      "loss": 0.3002,
      "step": 2530
    },
    {
      "epoch": 0.2498094465343857,
      "grad_norm": 1.259028434753418,
      "learning_rate": 1.9192862536263562e-05,
      "loss": 0.2244,
      "step": 2540
    },
    {
      "epoch": 0.25079294829239507,
      "grad_norm": 0.6467012763023376,
      "learning_rate": 1.9188888447323452e-05,
      "loss": 0.2983,
      "step": 2550
    },
    {
      "epoch": 0.25177645005040444,
      "grad_norm": 1.3998416662216187,
      "learning_rate": 1.918491435838334e-05,
      "loss": 0.3287,
      "step": 2560
    },
    {
      "epoch": 0.2527599518084139,
      "grad_norm": 1.7507133483886719,
      "learning_rate": 1.918094026944323e-05,
      "loss": 0.286,
      "step": 2570
    },
    {
      "epoch": 0.25374345356642325,
      "grad_norm": 2.0538573265075684,
      "learning_rate": 1.917696618050312e-05,
      "loss": 0.3549,
      "step": 2580
    },
    {
      "epoch": 0.2547269553244326,
      "grad_norm": 0.7538926005363464,
      "learning_rate": 1.917299209156301e-05,
      "loss": 0.309,
      "step": 2590
    },
    {
      "epoch": 0.25571045708244206,
      "grad_norm": 0.7768136858940125,
      "learning_rate": 1.91690180026229e-05,
      "loss": 0.3403,
      "step": 2600
    },
    {
      "epoch": 0.25669395884045143,
      "grad_norm": 1.5854865312576294,
      "learning_rate": 1.916504391368279e-05,
      "loss": 0.3446,
      "step": 2610
    },
    {
      "epoch": 0.2576774605984608,
      "grad_norm": 0.8464900851249695,
      "learning_rate": 1.916106982474268e-05,
      "loss": 0.2657,
      "step": 2620
    },
    {
      "epoch": 0.25866096235647024,
      "grad_norm": 0.8218358755111694,
      "learning_rate": 1.915709573580257e-05,
      "loss": 0.253,
      "step": 2630
    },
    {
      "epoch": 0.2596444641144796,
      "grad_norm": 1.4568731784820557,
      "learning_rate": 1.915312164686246e-05,
      "loss": 0.2408,
      "step": 2640
    },
    {
      "epoch": 0.260627965872489,
      "grad_norm": 1.706364631652832,
      "learning_rate": 1.914914755792235e-05,
      "loss": 0.2368,
      "step": 2650
    },
    {
      "epoch": 0.26161146763049836,
      "grad_norm": 0.566818356513977,
      "learning_rate": 1.9145173468982238e-05,
      "loss": 0.3488,
      "step": 2660
    },
    {
      "epoch": 0.2625949693885078,
      "grad_norm": 0.9257858991622925,
      "learning_rate": 1.9141199380042124e-05,
      "loss": 0.2329,
      "step": 2670
    },
    {
      "epoch": 0.26357847114651717,
      "grad_norm": 1.2466871738433838,
      "learning_rate": 1.9137225291102017e-05,
      "loss": 0.428,
      "step": 2680
    },
    {
      "epoch": 0.26456197290452654,
      "grad_norm": 0.7756865620613098,
      "learning_rate": 1.9133251202161907e-05,
      "loss": 0.3775,
      "step": 2690
    },
    {
      "epoch": 0.265545474662536,
      "grad_norm": 0.9694525003433228,
      "learning_rate": 1.9129277113221797e-05,
      "loss": 0.252,
      "step": 2700
    },
    {
      "epoch": 0.26652897642054535,
      "grad_norm": 1.9040061235427856,
      "learning_rate": 1.9125303024281686e-05,
      "loss": 0.2924,
      "step": 2710
    },
    {
      "epoch": 0.2675124781785547,
      "grad_norm": 1.2722103595733643,
      "learning_rate": 1.9121328935341573e-05,
      "loss": 0.2348,
      "step": 2720
    },
    {
      "epoch": 0.26849597993656416,
      "grad_norm": 0.8166412115097046,
      "learning_rate": 1.9117354846401462e-05,
      "loss": 0.2093,
      "step": 2730
    },
    {
      "epoch": 0.26947948169457353,
      "grad_norm": 0.6143937110900879,
      "learning_rate": 1.9113380757461352e-05,
      "loss": 0.2024,
      "step": 2740
    },
    {
      "epoch": 0.2704629834525829,
      "grad_norm": 1.0523463487625122,
      "learning_rate": 1.910940666852124e-05,
      "loss": 0.3963,
      "step": 2750
    },
    {
      "epoch": 0.27144648521059234,
      "grad_norm": 1.7578366994857788,
      "learning_rate": 1.9105432579581135e-05,
      "loss": 0.3699,
      "step": 2760
    },
    {
      "epoch": 0.2724299869686017,
      "grad_norm": 0.3840101659297943,
      "learning_rate": 1.9101458490641024e-05,
      "loss": 0.3475,
      "step": 2770
    },
    {
      "epoch": 0.2734134887266111,
      "grad_norm": 1.2118396759033203,
      "learning_rate": 1.909748440170091e-05,
      "loss": 0.2155,
      "step": 2780
    },
    {
      "epoch": 0.2743969904846205,
      "grad_norm": 1.1841282844543457,
      "learning_rate": 1.90935103127608e-05,
      "loss": 0.3818,
      "step": 2790
    },
    {
      "epoch": 0.2753804922426299,
      "grad_norm": 0.46255505084991455,
      "learning_rate": 1.908953622382069e-05,
      "loss": 0.3608,
      "step": 2800
    },
    {
      "epoch": 0.27636399400063927,
      "grad_norm": 1.2726696729660034,
      "learning_rate": 1.908556213488058e-05,
      "loss": 0.191,
      "step": 2810
    },
    {
      "epoch": 0.27734749575864864,
      "grad_norm": 2.2195935249328613,
      "learning_rate": 1.908158804594047e-05,
      "loss": 0.2622,
      "step": 2820
    },
    {
      "epoch": 0.2783309975166581,
      "grad_norm": 0.9706044793128967,
      "learning_rate": 1.907761395700036e-05,
      "loss": 0.2811,
      "step": 2830
    },
    {
      "epoch": 0.27931449927466745,
      "grad_norm": 1.3599534034729004,
      "learning_rate": 1.907363986806025e-05,
      "loss": 0.398,
      "step": 2840
    },
    {
      "epoch": 0.2802980010326768,
      "grad_norm": 0.937558114528656,
      "learning_rate": 1.9069665779120138e-05,
      "loss": 0.3319,
      "step": 2850
    },
    {
      "epoch": 0.28128150279068626,
      "grad_norm": 1.9157116413116455,
      "learning_rate": 1.9065691690180028e-05,
      "loss": 0.2936,
      "step": 2860
    },
    {
      "epoch": 0.28226500454869563,
      "grad_norm": 1.3096779584884644,
      "learning_rate": 1.9061717601239918e-05,
      "loss": 0.3654,
      "step": 2870
    },
    {
      "epoch": 0.283248506306705,
      "grad_norm": 1.887032151222229,
      "learning_rate": 1.9057743512299807e-05,
      "loss": 0.364,
      "step": 2880
    },
    {
      "epoch": 0.28423200806471444,
      "grad_norm": 1.9582734107971191,
      "learning_rate": 1.9053769423359697e-05,
      "loss": 0.2359,
      "step": 2890
    },
    {
      "epoch": 0.2852155098227238,
      "grad_norm": 1.3817726373672485,
      "learning_rate": 1.9049795334419587e-05,
      "loss": 0.2605,
      "step": 2900
    },
    {
      "epoch": 0.2861990115807332,
      "grad_norm": 1.1931792497634888,
      "learning_rate": 1.9045821245479473e-05,
      "loss": 0.3133,
      "step": 2910
    },
    {
      "epoch": 0.2871825133387426,
      "grad_norm": 1.1820030212402344,
      "learning_rate": 1.9041847156539366e-05,
      "loss": 0.3351,
      "step": 2920
    },
    {
      "epoch": 0.288166015096752,
      "grad_norm": 1.4011718034744263,
      "learning_rate": 1.9037873067599256e-05,
      "loss": 0.2993,
      "step": 2930
    },
    {
      "epoch": 0.28914951685476137,
      "grad_norm": 1.7731127738952637,
      "learning_rate": 1.9033898978659145e-05,
      "loss": 0.304,
      "step": 2940
    },
    {
      "epoch": 0.29013301861277074,
      "grad_norm": 1.2245922088623047,
      "learning_rate": 1.9029924889719035e-05,
      "loss": 0.3627,
      "step": 2950
    },
    {
      "epoch": 0.2911165203707802,
      "grad_norm": 1.4606093168258667,
      "learning_rate": 1.902595080077892e-05,
      "loss": 0.3513,
      "step": 2960
    },
    {
      "epoch": 0.29210002212878955,
      "grad_norm": 1.4176182746887207,
      "learning_rate": 1.902197671183881e-05,
      "loss": 0.3211,
      "step": 2970
    },
    {
      "epoch": 0.2930835238867989,
      "grad_norm": 0.8106399178504944,
      "learning_rate": 1.90180026228987e-05,
      "loss": 0.3057,
      "step": 2980
    },
    {
      "epoch": 0.29406702564480836,
      "grad_norm": 1.1231039762496948,
      "learning_rate": 1.901402853395859e-05,
      "loss": 0.3033,
      "step": 2990
    },
    {
      "epoch": 0.29505052740281773,
      "grad_norm": 0.3606371581554413,
      "learning_rate": 1.9010054445018483e-05,
      "loss": 0.2825,
      "step": 3000
    },
    {
      "epoch": 0.29505052740281773,
      "eval_loss": 0.18925747275352478,
      "eval_runtime": 18.2369,
      "eval_samples_per_second": 2.742,
      "eval_steps_per_second": 1.371,
      "step": 3000
    },
    {
      "epoch": 0.2960340291608271,
      "grad_norm": 0.6441012620925903,
      "learning_rate": 1.9006080356078373e-05,
      "loss": 0.2882,
      "step": 3010
    },
    {
      "epoch": 0.29701753091883654,
      "grad_norm": 1.2217457294464111,
      "learning_rate": 1.900210626713826e-05,
      "loss": 0.2755,
      "step": 3020
    },
    {
      "epoch": 0.2980010326768459,
      "grad_norm": 1.9655147790908813,
      "learning_rate": 1.899813217819815e-05,
      "loss": 0.2335,
      "step": 3030
    },
    {
      "epoch": 0.2989845344348553,
      "grad_norm": 0.8391982913017273,
      "learning_rate": 1.899415808925804e-05,
      "loss": 0.2076,
      "step": 3040
    },
    {
      "epoch": 0.2999680361928647,
      "grad_norm": 1.5566798448562622,
      "learning_rate": 1.8990184000317928e-05,
      "loss": 0.2677,
      "step": 3050
    },
    {
      "epoch": 0.3009515379508741,
      "grad_norm": 1.858270525932312,
      "learning_rate": 1.8986209911377818e-05,
      "loss": 0.4234,
      "step": 3060
    },
    {
      "epoch": 0.30193503970888347,
      "grad_norm": 1.1900537014007568,
      "learning_rate": 1.8982235822437708e-05,
      "loss": 0.2877,
      "step": 3070
    },
    {
      "epoch": 0.3029185414668929,
      "grad_norm": 2.3156630992889404,
      "learning_rate": 1.8978261733497597e-05,
      "loss": 0.4231,
      "step": 3080
    },
    {
      "epoch": 0.3039020432249023,
      "grad_norm": 1.8294111490249634,
      "learning_rate": 1.8974287644557487e-05,
      "loss": 0.2523,
      "step": 3090
    },
    {
      "epoch": 0.30488554498291165,
      "grad_norm": 1.9268901348114014,
      "learning_rate": 1.8970313555617377e-05,
      "loss": 0.2829,
      "step": 3100
    },
    {
      "epoch": 0.305869046740921,
      "grad_norm": 1.6466792821884155,
      "learning_rate": 1.8966339466677266e-05,
      "loss": 0.2802,
      "step": 3110
    },
    {
      "epoch": 0.30685254849893046,
      "grad_norm": 0.9668562412261963,
      "learning_rate": 1.8962365377737156e-05,
      "loss": 0.2447,
      "step": 3120
    },
    {
      "epoch": 0.30783605025693983,
      "grad_norm": 1.3234363794326782,
      "learning_rate": 1.8958391288797045e-05,
      "loss": 0.3151,
      "step": 3130
    },
    {
      "epoch": 0.3088195520149492,
      "grad_norm": 0.7863141894340515,
      "learning_rate": 1.8954417199856935e-05,
      "loss": 0.2525,
      "step": 3140
    },
    {
      "epoch": 0.30980305377295864,
      "grad_norm": 1.0784416198730469,
      "learning_rate": 1.895044311091682e-05,
      "loss": 0.3044,
      "step": 3150
    },
    {
      "epoch": 0.310786555530968,
      "grad_norm": 1.47318696975708,
      "learning_rate": 1.8946469021976714e-05,
      "loss": 0.2453,
      "step": 3160
    },
    {
      "epoch": 0.3117700572889774,
      "grad_norm": 0.8541170954704285,
      "learning_rate": 1.8942494933036604e-05,
      "loss": 0.2838,
      "step": 3170
    },
    {
      "epoch": 0.3127535590469868,
      "grad_norm": 1.8033347129821777,
      "learning_rate": 1.8938520844096494e-05,
      "loss": 0.2572,
      "step": 3180
    },
    {
      "epoch": 0.3137370608049962,
      "grad_norm": 0.6662768125534058,
      "learning_rate": 1.8934546755156383e-05,
      "loss": 0.2875,
      "step": 3190
    },
    {
      "epoch": 0.31472056256300557,
      "grad_norm": 1.5999656915664673,
      "learning_rate": 1.893057266621627e-05,
      "loss": 0.3892,
      "step": 3200
    },
    {
      "epoch": 0.315704064321015,
      "grad_norm": 1.12599778175354,
      "learning_rate": 1.892659857727616e-05,
      "loss": 0.3817,
      "step": 3210
    },
    {
      "epoch": 0.3166875660790244,
      "grad_norm": 0.5991163849830627,
      "learning_rate": 1.892262448833605e-05,
      "loss": 0.3791,
      "step": 3220
    },
    {
      "epoch": 0.31767106783703375,
      "grad_norm": 2.04829478263855,
      "learning_rate": 1.891865039939594e-05,
      "loss": 0.2998,
      "step": 3230
    },
    {
      "epoch": 0.3186545695950431,
      "grad_norm": 1.1123751401901245,
      "learning_rate": 1.8914676310455832e-05,
      "loss": 0.2266,
      "step": 3240
    },
    {
      "epoch": 0.31963807135305256,
      "grad_norm": 2.765881299972534,
      "learning_rate": 1.891070222151572e-05,
      "loss": 0.3068,
      "step": 3250
    },
    {
      "epoch": 0.32062157311106193,
      "grad_norm": 0.9850254058837891,
      "learning_rate": 1.8906728132575608e-05,
      "loss": 0.1555,
      "step": 3260
    },
    {
      "epoch": 0.3216050748690713,
      "grad_norm": 0.6333345174789429,
      "learning_rate": 1.8902754043635497e-05,
      "loss": 0.2767,
      "step": 3270
    },
    {
      "epoch": 0.32258857662708074,
      "grad_norm": 1.5807948112487793,
      "learning_rate": 1.8898779954695387e-05,
      "loss": 0.4006,
      "step": 3280
    },
    {
      "epoch": 0.3235720783850901,
      "grad_norm": 0.7924546003341675,
      "learning_rate": 1.8894805865755277e-05,
      "loss": 0.3417,
      "step": 3290
    },
    {
      "epoch": 0.3245555801430995,
      "grad_norm": 0.6605357527732849,
      "learning_rate": 1.8890831776815166e-05,
      "loss": 0.2169,
      "step": 3300
    },
    {
      "epoch": 0.3255390819011089,
      "grad_norm": 1.03006112575531,
      "learning_rate": 1.8886857687875056e-05,
      "loss": 0.3255,
      "step": 3310
    },
    {
      "epoch": 0.3265225836591183,
      "grad_norm": 1.244502305984497,
      "learning_rate": 1.8882883598934946e-05,
      "loss": 0.2502,
      "step": 3320
    },
    {
      "epoch": 0.32750608541712767,
      "grad_norm": 2.606926679611206,
      "learning_rate": 1.8878909509994835e-05,
      "loss": 0.344,
      "step": 3330
    },
    {
      "epoch": 0.3284895871751371,
      "grad_norm": 0.2963784337043762,
      "learning_rate": 1.8874935421054725e-05,
      "loss": 0.371,
      "step": 3340
    },
    {
      "epoch": 0.3294730889331465,
      "grad_norm": 1.087789535522461,
      "learning_rate": 1.8870961332114615e-05,
      "loss": 0.2594,
      "step": 3350
    },
    {
      "epoch": 0.33045659069115585,
      "grad_norm": 0.8091050386428833,
      "learning_rate": 1.8866987243174504e-05,
      "loss": 0.3005,
      "step": 3360
    },
    {
      "epoch": 0.3314400924491653,
      "grad_norm": 0.6955603957176208,
      "learning_rate": 1.8863013154234394e-05,
      "loss": 0.2407,
      "step": 3370
    },
    {
      "epoch": 0.33242359420717466,
      "grad_norm": 1.2869349718093872,
      "learning_rate": 1.8859039065294284e-05,
      "loss": 0.2972,
      "step": 3380
    },
    {
      "epoch": 0.33340709596518403,
      "grad_norm": 1.0467299222946167,
      "learning_rate": 1.885506497635417e-05,
      "loss": 0.3742,
      "step": 3390
    },
    {
      "epoch": 0.3343905977231934,
      "grad_norm": 1.2822316884994507,
      "learning_rate": 1.885109088741406e-05,
      "loss": 0.2625,
      "step": 3400
    },
    {
      "epoch": 0.33537409948120284,
      "grad_norm": 1.278634786605835,
      "learning_rate": 1.8847116798473953e-05,
      "loss": 0.3095,
      "step": 3410
    },
    {
      "epoch": 0.3363576012392122,
      "grad_norm": 0.4241173267364502,
      "learning_rate": 1.8843142709533842e-05,
      "loss": 0.3163,
      "step": 3420
    },
    {
      "epoch": 0.3373411029972216,
      "grad_norm": 1.171540379524231,
      "learning_rate": 1.8839168620593732e-05,
      "loss": 0.3011,
      "step": 3430
    },
    {
      "epoch": 0.338324604755231,
      "grad_norm": 1.1309036016464233,
      "learning_rate": 1.883519453165362e-05,
      "loss": 0.3909,
      "step": 3440
    },
    {
      "epoch": 0.3393081065132404,
      "grad_norm": 1.4988631010055542,
      "learning_rate": 1.8831220442713508e-05,
      "loss": 0.3911,
      "step": 3450
    },
    {
      "epoch": 0.34029160827124977,
      "grad_norm": 0.8498252630233765,
      "learning_rate": 1.8827246353773398e-05,
      "loss": 0.2321,
      "step": 3460
    },
    {
      "epoch": 0.3412751100292592,
      "grad_norm": 3.627526044845581,
      "learning_rate": 1.8823272264833287e-05,
      "loss": 0.3124,
      "step": 3470
    },
    {
      "epoch": 0.3422586117872686,
      "grad_norm": 0.695372998714447,
      "learning_rate": 1.8819298175893177e-05,
      "loss": 0.3233,
      "step": 3480
    },
    {
      "epoch": 0.34324211354527795,
      "grad_norm": 0.9676603078842163,
      "learning_rate": 1.881532408695307e-05,
      "loss": 0.2751,
      "step": 3490
    },
    {
      "epoch": 0.3442256153032874,
      "grad_norm": 2.4137444496154785,
      "learning_rate": 1.8811349998012956e-05,
      "loss": 0.239,
      "step": 3500
    },
    {
      "epoch": 0.3442256153032874,
      "eval_loss": 0.18515843152999878,
      "eval_runtime": 18.8557,
      "eval_samples_per_second": 2.652,
      "eval_steps_per_second": 1.326,
      "step": 3500
    },
    {
      "epoch": 0.34520911706129676,
      "grad_norm": 0.8635812401771545,
      "learning_rate": 1.8807375909072846e-05,
      "loss": 0.2727,
      "step": 3510
    },
    {
      "epoch": 0.34619261881930613,
      "grad_norm": 0.8130320906639099,
      "learning_rate": 1.8803401820132736e-05,
      "loss": 0.3389,
      "step": 3520
    },
    {
      "epoch": 0.3471761205773155,
      "grad_norm": 0.8832435011863708,
      "learning_rate": 1.8799427731192625e-05,
      "loss": 0.3392,
      "step": 3530
    },
    {
      "epoch": 0.34815962233532494,
      "grad_norm": 0.41087064146995544,
      "learning_rate": 1.8795453642252515e-05,
      "loss": 0.3043,
      "step": 3540
    },
    {
      "epoch": 0.3491431240933343,
      "grad_norm": 0.865750253200531,
      "learning_rate": 1.8791479553312405e-05,
      "loss": 0.3286,
      "step": 3550
    },
    {
      "epoch": 0.3501266258513437,
      "grad_norm": 2.180868625640869,
      "learning_rate": 1.8787505464372294e-05,
      "loss": 0.2891,
      "step": 3560
    },
    {
      "epoch": 0.3511101276093531,
      "grad_norm": 1.0887324810028076,
      "learning_rate": 1.8783531375432184e-05,
      "loss": 0.3794,
      "step": 3570
    },
    {
      "epoch": 0.3520936293673625,
      "grad_norm": 1.2184492349624634,
      "learning_rate": 1.8779557286492074e-05,
      "loss": 0.2399,
      "step": 3580
    },
    {
      "epoch": 0.35307713112537187,
      "grad_norm": 1.5976436138153076,
      "learning_rate": 1.8775583197551963e-05,
      "loss": 0.321,
      "step": 3590
    },
    {
      "epoch": 0.3540606328833813,
      "grad_norm": 0.7627467513084412,
      "learning_rate": 1.8771609108611853e-05,
      "loss": 0.368,
      "step": 3600
    },
    {
      "epoch": 0.3550441346413907,
      "grad_norm": 0.7713624238967896,
      "learning_rate": 1.8767635019671743e-05,
      "loss": 0.3409,
      "step": 3610
    },
    {
      "epoch": 0.35602763639940005,
      "grad_norm": 0.6090192198753357,
      "learning_rate": 1.8763660930731632e-05,
      "loss": 0.236,
      "step": 3620
    },
    {
      "epoch": 0.3570111381574095,
      "grad_norm": 1.4962300062179565,
      "learning_rate": 1.875968684179152e-05,
      "loss": 0.1852,
      "step": 3630
    },
    {
      "epoch": 0.35799463991541886,
      "grad_norm": 1.274659276008606,
      "learning_rate": 1.8755712752851408e-05,
      "loss": 0.2773,
      "step": 3640
    },
    {
      "epoch": 0.35897814167342823,
      "grad_norm": 1.3212956190109253,
      "learning_rate": 1.87517386639113e-05,
      "loss": 0.221,
      "step": 3650
    },
    {
      "epoch": 0.35996164343143766,
      "grad_norm": 1.0321497917175293,
      "learning_rate": 1.874776457497119e-05,
      "loss": 0.3394,
      "step": 3660
    },
    {
      "epoch": 0.36094514518944704,
      "grad_norm": 0.7226223349571228,
      "learning_rate": 1.874379048603108e-05,
      "loss": 0.246,
      "step": 3670
    },
    {
      "epoch": 0.3619286469474564,
      "grad_norm": 2.584559917449951,
      "learning_rate": 1.8739816397090967e-05,
      "loss": 0.1398,
      "step": 3680
    },
    {
      "epoch": 0.3629121487054658,
      "grad_norm": 0.9170656800270081,
      "learning_rate": 1.8735842308150857e-05,
      "loss": 0.1937,
      "step": 3690
    },
    {
      "epoch": 0.3638956504634752,
      "grad_norm": 1.994689702987671,
      "learning_rate": 1.8731868219210746e-05,
      "loss": 0.3006,
      "step": 3700
    },
    {
      "epoch": 0.3648791522214846,
      "grad_norm": 1.0057986974716187,
      "learning_rate": 1.8727894130270636e-05,
      "loss": 0.2274,
      "step": 3710
    },
    {
      "epoch": 0.36586265397949397,
      "grad_norm": 0.529000461101532,
      "learning_rate": 1.8723920041330526e-05,
      "loss": 0.2544,
      "step": 3720
    },
    {
      "epoch": 0.3668461557375034,
      "grad_norm": 2.1913716793060303,
      "learning_rate": 1.871994595239042e-05,
      "loss": 0.2118,
      "step": 3730
    },
    {
      "epoch": 0.3678296574955128,
      "grad_norm": 1.2990810871124268,
      "learning_rate": 1.8715971863450305e-05,
      "loss": 0.3832,
      "step": 3740
    },
    {
      "epoch": 0.36881315925352215,
      "grad_norm": 1.4489165544509888,
      "learning_rate": 1.8711997774510195e-05,
      "loss": 0.3792,
      "step": 3750
    },
    {
      "epoch": 0.3697966610115316,
      "grad_norm": 1.727668285369873,
      "learning_rate": 1.8708023685570084e-05,
      "loss": 0.3826,
      "step": 3760
    },
    {
      "epoch": 0.37078016276954096,
      "grad_norm": 1.3508563041687012,
      "learning_rate": 1.8704049596629974e-05,
      "loss": 0.3114,
      "step": 3770
    },
    {
      "epoch": 0.37176366452755033,
      "grad_norm": 2.063981533050537,
      "learning_rate": 1.8700075507689864e-05,
      "loss": 0.1861,
      "step": 3780
    },
    {
      "epoch": 0.37274716628555976,
      "grad_norm": 1.491122841835022,
      "learning_rate": 1.8696101418749753e-05,
      "loss": 0.315,
      "step": 3790
    },
    {
      "epoch": 0.37373066804356914,
      "grad_norm": 1.6444443464279175,
      "learning_rate": 1.8692127329809643e-05,
      "loss": 0.3538,
      "step": 3800
    },
    {
      "epoch": 0.3747141698015785,
      "grad_norm": 1.7396812438964844,
      "learning_rate": 1.8688153240869533e-05,
      "loss": 0.2084,
      "step": 3810
    },
    {
      "epoch": 0.3756976715595879,
      "grad_norm": 1.6162070035934448,
      "learning_rate": 1.8684179151929422e-05,
      "loss": 0.3024,
      "step": 3820
    },
    {
      "epoch": 0.3766811733175973,
      "grad_norm": 1.6525781154632568,
      "learning_rate": 1.8680205062989312e-05,
      "loss": 0.3075,
      "step": 3830
    },
    {
      "epoch": 0.3776646750756067,
      "grad_norm": 0.7809441089630127,
      "learning_rate": 1.86762309740492e-05,
      "loss": 0.33,
      "step": 3840
    },
    {
      "epoch": 0.37864817683361607,
      "grad_norm": 1.2998336553573608,
      "learning_rate": 1.867225688510909e-05,
      "loss": 0.2978,
      "step": 3850
    },
    {
      "epoch": 0.3796316785916255,
      "grad_norm": 0.5490562319755554,
      "learning_rate": 1.866828279616898e-05,
      "loss": 0.2625,
      "step": 3860
    },
    {
      "epoch": 0.3806151803496349,
      "grad_norm": 1.808087944984436,
      "learning_rate": 1.8664308707228867e-05,
      "loss": 0.1814,
      "step": 3870
    },
    {
      "epoch": 0.38159868210764425,
      "grad_norm": 1.9151523113250732,
      "learning_rate": 1.8660334618288757e-05,
      "loss": 0.2716,
      "step": 3880
    },
    {
      "epoch": 0.3825821838656537,
      "grad_norm": 1.526478886604309,
      "learning_rate": 1.865636052934865e-05,
      "loss": 0.2755,
      "step": 3890
    },
    {
      "epoch": 0.38356568562366306,
      "grad_norm": 0.8774662017822266,
      "learning_rate": 1.865238644040854e-05,
      "loss": 0.3243,
      "step": 3900
    },
    {
      "epoch": 0.38454918738167243,
      "grad_norm": 1.8767133951187134,
      "learning_rate": 1.864841235146843e-05,
      "loss": 0.3164,
      "step": 3910
    },
    {
      "epoch": 0.38553268913968186,
      "grad_norm": 1.6160063743591309,
      "learning_rate": 1.8644438262528315e-05,
      "loss": 0.34,
      "step": 3920
    },
    {
      "epoch": 0.38651619089769124,
      "grad_norm": 0.7355303764343262,
      "learning_rate": 1.8640464173588205e-05,
      "loss": 0.3179,
      "step": 3930
    },
    {
      "epoch": 0.3874996926557006,
      "grad_norm": 0.8210369348526001,
      "learning_rate": 1.8636490084648095e-05,
      "loss": 0.2245,
      "step": 3940
    },
    {
      "epoch": 0.38848319441371,
      "grad_norm": 2.0436363220214844,
      "learning_rate": 1.8632515995707984e-05,
      "loss": 0.3267,
      "step": 3950
    },
    {
      "epoch": 0.3894666961717194,
      "grad_norm": 1.3316550254821777,
      "learning_rate": 1.8628541906767874e-05,
      "loss": 0.2627,
      "step": 3960
    },
    {
      "epoch": 0.3904501979297288,
      "grad_norm": 2.4543538093566895,
      "learning_rate": 1.8624567817827767e-05,
      "loss": 0.3055,
      "step": 3970
    },
    {
      "epoch": 0.39143369968773817,
      "grad_norm": 1.4763861894607544,
      "learning_rate": 1.8620593728887653e-05,
      "loss": 0.224,
      "step": 3980
    },
    {
      "epoch": 0.3924172014457476,
      "grad_norm": 2.1367290019989014,
      "learning_rate": 1.8616619639947543e-05,
      "loss": 0.2971,
      "step": 3990
    },
    {
      "epoch": 0.393400703203757,
      "grad_norm": 0.5473704934120178,
      "learning_rate": 1.8612645551007433e-05,
      "loss": 0.1911,
      "step": 4000
    },
    {
      "epoch": 0.393400703203757,
      "eval_loss": 0.18661783635616302,
      "eval_runtime": 18.7545,
      "eval_samples_per_second": 2.666,
      "eval_steps_per_second": 1.333,
      "step": 4000
    },
    {
      "epoch": 0.39438420496176635,
      "grad_norm": 0.6994380950927734,
      "learning_rate": 1.8608671462067322e-05,
      "loss": 0.2789,
      "step": 4010
    },
    {
      "epoch": 0.3953677067197758,
      "grad_norm": 0.8082060813903809,
      "learning_rate": 1.8604697373127212e-05,
      "loss": 0.2897,
      "step": 4020
    },
    {
      "epoch": 0.39635120847778516,
      "grad_norm": 1.6137787103652954,
      "learning_rate": 1.8600723284187102e-05,
      "loss": 0.3562,
      "step": 4030
    },
    {
      "epoch": 0.39733471023579453,
      "grad_norm": 0.4196963310241699,
      "learning_rate": 1.859674919524699e-05,
      "loss": 0.1581,
      "step": 4040
    },
    {
      "epoch": 0.39831821199380396,
      "grad_norm": 1.2669254541397095,
      "learning_rate": 1.859277510630688e-05,
      "loss": 0.2467,
      "step": 4050
    },
    {
      "epoch": 0.39930171375181334,
      "grad_norm": 0.7445014715194702,
      "learning_rate": 1.858880101736677e-05,
      "loss": 0.3687,
      "step": 4060
    },
    {
      "epoch": 0.4002852155098227,
      "grad_norm": 1.2076808214187622,
      "learning_rate": 1.858482692842666e-05,
      "loss": 0.2203,
      "step": 4070
    },
    {
      "epoch": 0.40126871726783214,
      "grad_norm": 1.9768540859222412,
      "learning_rate": 1.858085283948655e-05,
      "loss": 0.2688,
      "step": 4080
    },
    {
      "epoch": 0.4022522190258415,
      "grad_norm": 1.1903109550476074,
      "learning_rate": 1.857687875054644e-05,
      "loss": 0.2881,
      "step": 4090
    },
    {
      "epoch": 0.4032357207838509,
      "grad_norm": 1.5276085138320923,
      "learning_rate": 1.857290466160633e-05,
      "loss": 0.2557,
      "step": 4100
    },
    {
      "epoch": 0.40421922254186027,
      "grad_norm": 2.2157998085021973,
      "learning_rate": 1.8568930572666216e-05,
      "loss": 0.298,
      "step": 4110
    },
    {
      "epoch": 0.4052027242998697,
      "grad_norm": 1.0562926530838013,
      "learning_rate": 1.8564956483726105e-05,
      "loss": 0.2623,
      "step": 4120
    },
    {
      "epoch": 0.4061862260578791,
      "grad_norm": 2.135159730911255,
      "learning_rate": 1.8560982394785995e-05,
      "loss": 0.2718,
      "step": 4130
    },
    {
      "epoch": 0.40716972781588845,
      "grad_norm": 1.4359196424484253,
      "learning_rate": 1.8557008305845888e-05,
      "loss": 0.2623,
      "step": 4140
    },
    {
      "epoch": 0.4081532295738979,
      "grad_norm": 0.8158294558525085,
      "learning_rate": 1.8553034216905778e-05,
      "loss": 0.2514,
      "step": 4150
    },
    {
      "epoch": 0.40913673133190726,
      "grad_norm": 1.8092743158340454,
      "learning_rate": 1.8549060127965664e-05,
      "loss": 0.5151,
      "step": 4160
    },
    {
      "epoch": 0.41012023308991663,
      "grad_norm": 1.847460389137268,
      "learning_rate": 1.8545086039025554e-05,
      "loss": 0.1929,
      "step": 4170
    },
    {
      "epoch": 0.41110373484792606,
      "grad_norm": 1.5607435703277588,
      "learning_rate": 1.8541111950085443e-05,
      "loss": 0.2819,
      "step": 4180
    },
    {
      "epoch": 0.41208723660593544,
      "grad_norm": 0.7382811903953552,
      "learning_rate": 1.8537137861145333e-05,
      "loss": 0.2232,
      "step": 4190
    },
    {
      "epoch": 0.4130707383639448,
      "grad_norm": 1.0930057764053345,
      "learning_rate": 1.8533163772205223e-05,
      "loss": 0.2593,
      "step": 4200
    },
    {
      "epoch": 0.41405424012195424,
      "grad_norm": 4.021615505218506,
      "learning_rate": 1.8529189683265112e-05,
      "loss": 0.3129,
      "step": 4210
    },
    {
      "epoch": 0.4150377418799636,
      "grad_norm": 0.39899012446403503,
      "learning_rate": 1.8525215594325002e-05,
      "loss": 0.2396,
      "step": 4220
    },
    {
      "epoch": 0.416021243637973,
      "grad_norm": 1.7783839702606201,
      "learning_rate": 1.852124150538489e-05,
      "loss": 0.2734,
      "step": 4230
    },
    {
      "epoch": 0.41700474539598237,
      "grad_norm": 1.0922350883483887,
      "learning_rate": 1.851726741644478e-05,
      "loss": 0.2591,
      "step": 4240
    },
    {
      "epoch": 0.4179882471539918,
      "grad_norm": 1.4296014308929443,
      "learning_rate": 1.851329332750467e-05,
      "loss": 0.2808,
      "step": 4250
    },
    {
      "epoch": 0.4189717489120012,
      "grad_norm": 1.9061700105667114,
      "learning_rate": 1.850931923856456e-05,
      "loss": 0.3218,
      "step": 4260
    },
    {
      "epoch": 0.41995525067001055,
      "grad_norm": 2.600834608078003,
      "learning_rate": 1.850534514962445e-05,
      "loss": 0.3231,
      "step": 4270
    },
    {
      "epoch": 0.42093875242802,
      "grad_norm": 1.5851668119430542,
      "learning_rate": 1.850137106068434e-05,
      "loss": 0.3265,
      "step": 4280
    },
    {
      "epoch": 0.42192225418602936,
      "grad_norm": 2.13275146484375,
      "learning_rate": 1.849739697174423e-05,
      "loss": 0.2575,
      "step": 4290
    },
    {
      "epoch": 0.42290575594403873,
      "grad_norm": 1.6749345064163208,
      "learning_rate": 1.849342288280412e-05,
      "loss": 0.3271,
      "step": 4300
    },
    {
      "epoch": 0.42388925770204816,
      "grad_norm": 0.5474580526351929,
      "learning_rate": 1.848944879386401e-05,
      "loss": 0.2267,
      "step": 4310
    },
    {
      "epoch": 0.42487275946005754,
      "grad_norm": 0.8282570242881775,
      "learning_rate": 1.84854747049239e-05,
      "loss": 0.321,
      "step": 4320
    },
    {
      "epoch": 0.4258562612180669,
      "grad_norm": 0.44433537125587463,
      "learning_rate": 1.848150061598379e-05,
      "loss": 0.297,
      "step": 4330
    },
    {
      "epoch": 0.42683976297607634,
      "grad_norm": 1.6867755651474,
      "learning_rate": 1.8477526527043678e-05,
      "loss": 0.3186,
      "step": 4340
    },
    {
      "epoch": 0.4278232647340857,
      "grad_norm": 1.451416254043579,
      "learning_rate": 1.8473552438103564e-05,
      "loss": 0.2344,
      "step": 4350
    },
    {
      "epoch": 0.4288067664920951,
      "grad_norm": 0.8302897214889526,
      "learning_rate": 1.8469578349163454e-05,
      "loss": 0.2629,
      "step": 4360
    },
    {
      "epoch": 0.4297902682501045,
      "grad_norm": 1.7050567865371704,
      "learning_rate": 1.8465604260223344e-05,
      "loss": 0.3174,
      "step": 4370
    },
    {
      "epoch": 0.4307737700081139,
      "grad_norm": 1.5906550884246826,
      "learning_rate": 1.8461630171283237e-05,
      "loss": 0.3394,
      "step": 4380
    },
    {
      "epoch": 0.4317572717661233,
      "grad_norm": 0.8430159091949463,
      "learning_rate": 1.8457656082343126e-05,
      "loss": 0.3178,
      "step": 4390
    },
    {
      "epoch": 0.43274077352413265,
      "grad_norm": 0.9703214168548584,
      "learning_rate": 1.8453681993403013e-05,
      "loss": 0.2287,
      "step": 4400
    },
    {
      "epoch": 0.4337242752821421,
      "grad_norm": 0.5622348189353943,
      "learning_rate": 1.8449707904462902e-05,
      "loss": 0.298,
      "step": 4410
    },
    {
      "epoch": 0.43470777704015146,
      "grad_norm": 1.1415741443634033,
      "learning_rate": 1.8445733815522792e-05,
      "loss": 0.1874,
      "step": 4420
    },
    {
      "epoch": 0.43569127879816083,
      "grad_norm": 1.2584352493286133,
      "learning_rate": 1.844175972658268e-05,
      "loss": 0.307,
      "step": 4430
    },
    {
      "epoch": 0.43667478055617026,
      "grad_norm": 1.9526134729385376,
      "learning_rate": 1.843778563764257e-05,
      "loss": 0.1829,
      "step": 4440
    },
    {
      "epoch": 0.43765828231417964,
      "grad_norm": 2.4884955883026123,
      "learning_rate": 1.843381154870246e-05,
      "loss": 0.256,
      "step": 4450
    },
    {
      "epoch": 0.438641784072189,
      "grad_norm": 0.5439223647117615,
      "learning_rate": 1.842983745976235e-05,
      "loss": 0.2438,
      "step": 4460
    },
    {
      "epoch": 0.43962528583019844,
      "grad_norm": 2.7485218048095703,
      "learning_rate": 1.842586337082224e-05,
      "loss": 0.288,
      "step": 4470
    },
    {
      "epoch": 0.4406087875882078,
      "grad_norm": 0.49018505215644836,
      "learning_rate": 1.842188928188213e-05,
      "loss": 0.237,
      "step": 4480
    },
    {
      "epoch": 0.4415922893462172,
      "grad_norm": 2.2588608264923096,
      "learning_rate": 1.841791519294202e-05,
      "loss": 0.2408,
      "step": 4490
    },
    {
      "epoch": 0.4425757911042266,
      "grad_norm": 1.4313758611679077,
      "learning_rate": 1.841394110400191e-05,
      "loss": 0.2917,
      "step": 4500
    },
    {
      "epoch": 0.4425757911042266,
      "eval_loss": 0.17805011570453644,
      "eval_runtime": 18.5759,
      "eval_samples_per_second": 2.692,
      "eval_steps_per_second": 1.346,
      "step": 4500
    },
    {
      "epoch": 0.443559292862236,
      "grad_norm": 1.1929224729537964,
      "learning_rate": 1.84099670150618e-05,
      "loss": 0.2371,
      "step": 4510
    },
    {
      "epoch": 0.4445427946202454,
      "grad_norm": 1.5835803747177124,
      "learning_rate": 1.840599292612169e-05,
      "loss": 0.273,
      "step": 4520
    },
    {
      "epoch": 0.44552629637825475,
      "grad_norm": 1.1826454401016235,
      "learning_rate": 1.8402018837181578e-05,
      "loss": 0.2249,
      "step": 4530
    },
    {
      "epoch": 0.4465097981362642,
      "grad_norm": 0.6066482663154602,
      "learning_rate": 1.8398044748241468e-05,
      "loss": 0.2553,
      "step": 4540
    },
    {
      "epoch": 0.44749329989427356,
      "grad_norm": 0.774614691734314,
      "learning_rate": 1.8394070659301358e-05,
      "loss": 0.1926,
      "step": 4550
    },
    {
      "epoch": 0.44847680165228293,
      "grad_norm": 1.106279969215393,
      "learning_rate": 1.8390096570361247e-05,
      "loss": 0.2636,
      "step": 4560
    },
    {
      "epoch": 0.44946030341029236,
      "grad_norm": 1.2709283828735352,
      "learning_rate": 1.8386122481421137e-05,
      "loss": 0.2895,
      "step": 4570
    },
    {
      "epoch": 0.45044380516830174,
      "grad_norm": 0.7416197657585144,
      "learning_rate": 1.8382148392481027e-05,
      "loss": 0.2984,
      "step": 4580
    },
    {
      "epoch": 0.4514273069263111,
      "grad_norm": 1.7245205640792847,
      "learning_rate": 1.8378174303540913e-05,
      "loss": 0.2593,
      "step": 4590
    },
    {
      "epoch": 0.45241080868432054,
      "grad_norm": 1.1603763103485107,
      "learning_rate": 1.8374200214600802e-05,
      "loss": 0.3668,
      "step": 4600
    },
    {
      "epoch": 0.4533943104423299,
      "grad_norm": 4.167576313018799,
      "learning_rate": 1.8370226125660692e-05,
      "loss": 0.2337,
      "step": 4610
    },
    {
      "epoch": 0.4543778122003393,
      "grad_norm": 2.9600322246551514,
      "learning_rate": 1.8366252036720585e-05,
      "loss": 0.2843,
      "step": 4620
    },
    {
      "epoch": 0.4553613139583487,
      "grad_norm": 0.6669513583183289,
      "learning_rate": 1.8362277947780475e-05,
      "loss": 0.2346,
      "step": 4630
    },
    {
      "epoch": 0.4563448157163581,
      "grad_norm": 1.5543279647827148,
      "learning_rate": 1.835830385884036e-05,
      "loss": 0.2542,
      "step": 4640
    },
    {
      "epoch": 0.4573283174743675,
      "grad_norm": 1.5348154306411743,
      "learning_rate": 1.835432976990025e-05,
      "loss": 0.268,
      "step": 4650
    },
    {
      "epoch": 0.4583118192323769,
      "grad_norm": 2.105473518371582,
      "learning_rate": 1.835035568096014e-05,
      "loss": 0.2747,
      "step": 4660
    },
    {
      "epoch": 0.4592953209903863,
      "grad_norm": 1.7820807695388794,
      "learning_rate": 1.834638159202003e-05,
      "loss": 0.2834,
      "step": 4670
    },
    {
      "epoch": 0.46027882274839566,
      "grad_norm": 1.1379271745681763,
      "learning_rate": 1.834240750307992e-05,
      "loss": 0.2521,
      "step": 4680
    },
    {
      "epoch": 0.46126232450640503,
      "grad_norm": 1.2687854766845703,
      "learning_rate": 1.833843341413981e-05,
      "loss": 0.304,
      "step": 4690
    },
    {
      "epoch": 0.46224582626441446,
      "grad_norm": 0.5946239829063416,
      "learning_rate": 1.83344593251997e-05,
      "loss": 0.3317,
      "step": 4700
    },
    {
      "epoch": 0.46322932802242384,
      "grad_norm": 1.261626958847046,
      "learning_rate": 1.833048523625959e-05,
      "loss": 0.3162,
      "step": 4710
    },
    {
      "epoch": 0.4642128297804332,
      "grad_norm": 2.5874712467193604,
      "learning_rate": 1.832651114731948e-05,
      "loss": 0.2443,
      "step": 4720
    },
    {
      "epoch": 0.46519633153844264,
      "grad_norm": 0.868843138217926,
      "learning_rate": 1.8322537058379368e-05,
      "loss": 0.2325,
      "step": 4730
    },
    {
      "epoch": 0.466179833296452,
      "grad_norm": 3.2576451301574707,
      "learning_rate": 1.8318562969439258e-05,
      "loss": 0.3519,
      "step": 4740
    },
    {
      "epoch": 0.4671633350544614,
      "grad_norm": 1.2640223503112793,
      "learning_rate": 1.8314588880499147e-05,
      "loss": 0.2424,
      "step": 4750
    },
    {
      "epoch": 0.4681468368124708,
      "grad_norm": 2.574295997619629,
      "learning_rate": 1.8310614791559037e-05,
      "loss": 0.32,
      "step": 4760
    },
    {
      "epoch": 0.4691303385704802,
      "grad_norm": 1.2373532056808472,
      "learning_rate": 1.8306640702618927e-05,
      "loss": 0.3316,
      "step": 4770
    },
    {
      "epoch": 0.4701138403284896,
      "grad_norm": 1.2597739696502686,
      "learning_rate": 1.8302666613678816e-05,
      "loss": 0.32,
      "step": 4780
    },
    {
      "epoch": 0.471097342086499,
      "grad_norm": 0.7831315994262695,
      "learning_rate": 1.8298692524738706e-05,
      "loss": 0.2963,
      "step": 4790
    },
    {
      "epoch": 0.4720808438445084,
      "grad_norm": 0.567329466342926,
      "learning_rate": 1.8294718435798596e-05,
      "loss": 0.1686,
      "step": 4800
    },
    {
      "epoch": 0.47306434560251776,
      "grad_norm": 1.0442653894424438,
      "learning_rate": 1.8290744346858485e-05,
      "loss": 0.1783,
      "step": 4810
    },
    {
      "epoch": 0.47404784736052713,
      "grad_norm": 1.4119641780853271,
      "learning_rate": 1.8286770257918375e-05,
      "loss": 0.2443,
      "step": 4820
    },
    {
      "epoch": 0.47503134911853656,
      "grad_norm": 2.7391903400421143,
      "learning_rate": 1.828279616897826e-05,
      "loss": 0.3288,
      "step": 4830
    },
    {
      "epoch": 0.47601485087654594,
      "grad_norm": 2.836953639984131,
      "learning_rate": 1.827882208003815e-05,
      "loss": 0.2084,
      "step": 4840
    },
    {
      "epoch": 0.4769983526345553,
      "grad_norm": 0.7899764180183411,
      "learning_rate": 1.827484799109804e-05,
      "loss": 0.2799,
      "step": 4850
    },
    {
      "epoch": 0.47798185439256474,
      "grad_norm": 1.3200337886810303,
      "learning_rate": 1.8270873902157934e-05,
      "loss": 0.1918,
      "step": 4860
    },
    {
      "epoch": 0.4789653561505741,
      "grad_norm": 0.9252869486808777,
      "learning_rate": 1.8266899813217823e-05,
      "loss": 0.3871,
      "step": 4870
    },
    {
      "epoch": 0.4799488579085835,
      "grad_norm": 0.5087369084358215,
      "learning_rate": 1.826292572427771e-05,
      "loss": 0.2481,
      "step": 4880
    },
    {
      "epoch": 0.4809323596665929,
      "grad_norm": 0.6862156987190247,
      "learning_rate": 1.82589516353376e-05,
      "loss": 0.313,
      "step": 4890
    },
    {
      "epoch": 0.4819158614246023,
      "grad_norm": 2.2030715942382812,
      "learning_rate": 1.825497754639749e-05,
      "loss": 0.2758,
      "step": 4900
    },
    {
      "epoch": 0.4828993631826117,
      "grad_norm": 1.4650092124938965,
      "learning_rate": 1.825100345745738e-05,
      "loss": 0.2371,
      "step": 4910
    },
    {
      "epoch": 0.4838828649406211,
      "grad_norm": 1.1407153606414795,
      "learning_rate": 1.824702936851727e-05,
      "loss": 0.2681,
      "step": 4920
    },
    {
      "epoch": 0.4848663666986305,
      "grad_norm": 0.8102698922157288,
      "learning_rate": 1.8243055279577158e-05,
      "loss": 0.3045,
      "step": 4930
    },
    {
      "epoch": 0.48584986845663986,
      "grad_norm": 2.678874969482422,
      "learning_rate": 1.8239081190637048e-05,
      "loss": 0.296,
      "step": 4940
    },
    {
      "epoch": 0.4868333702146493,
      "grad_norm": 2.3300135135650635,
      "learning_rate": 1.8235107101696937e-05,
      "loss": 0.3185,
      "step": 4950
    },
    {
      "epoch": 0.48781687197265866,
      "grad_norm": 0.7017766833305359,
      "learning_rate": 1.8231133012756827e-05,
      "loss": 0.2241,
      "step": 4960
    },
    {
      "epoch": 0.48880037373066804,
      "grad_norm": 0.9197968244552612,
      "learning_rate": 1.8227158923816717e-05,
      "loss": 0.2869,
      "step": 4970
    },
    {
      "epoch": 0.4897838754886774,
      "grad_norm": 1.4702986478805542,
      "learning_rate": 1.8223184834876606e-05,
      "loss": 0.2739,
      "step": 4980
    },
    {
      "epoch": 0.49076737724668684,
      "grad_norm": 0.45094242691993713,
      "learning_rate": 1.8219210745936496e-05,
      "loss": 0.2225,
      "step": 4990
    },
    {
      "epoch": 0.4917508790046962,
      "grad_norm": 2.3170652389526367,
      "learning_rate": 1.8215236656996386e-05,
      "loss": 0.2958,
      "step": 5000
    },
    {
      "epoch": 0.4917508790046962,
      "eval_loss": 0.17952224612236023,
      "eval_runtime": 16.9014,
      "eval_samples_per_second": 2.958,
      "eval_steps_per_second": 1.479,
      "step": 5000
    },
    {
      "epoch": 0.4927343807627056,
      "grad_norm": 0.6945057511329651,
      "learning_rate": 1.8211262568056275e-05,
      "loss": 0.2397,
      "step": 5010
    },
    {
      "epoch": 0.493717882520715,
      "grad_norm": 1.9027206897735596,
      "learning_rate": 1.820728847911616e-05,
      "loss": 0.2174,
      "step": 5020
    },
    {
      "epoch": 0.4947013842787244,
      "grad_norm": 0.959458589553833,
      "learning_rate": 1.8203314390176055e-05,
      "loss": 0.2665,
      "step": 5030
    },
    {
      "epoch": 0.4956848860367338,
      "grad_norm": 2.319744348526001,
      "learning_rate": 1.8199340301235944e-05,
      "loss": 0.2563,
      "step": 5040
    },
    {
      "epoch": 0.4966683877947432,
      "grad_norm": 2.804980754852295,
      "learning_rate": 1.8195366212295834e-05,
      "loss": 0.2734,
      "step": 5050
    },
    {
      "epoch": 0.4976518895527526,
      "grad_norm": 1.1066958904266357,
      "learning_rate": 1.8191392123355724e-05,
      "loss": 0.3222,
      "step": 5060
    },
    {
      "epoch": 0.49863539131076196,
      "grad_norm": 1.7555015087127686,
      "learning_rate": 1.818741803441561e-05,
      "loss": 0.2398,
      "step": 5070
    },
    {
      "epoch": 0.4996188930687714,
      "grad_norm": 3.245117425918579,
      "learning_rate": 1.81834439454755e-05,
      "loss": 0.176,
      "step": 5080
    },
    {
      "epoch": 0.5006023948267807,
      "grad_norm": 1.5190283060073853,
      "learning_rate": 1.817946985653539e-05,
      "loss": 0.2128,
      "step": 5090
    },
    {
      "epoch": 0.5015858965847901,
      "grad_norm": 1.867402195930481,
      "learning_rate": 1.817549576759528e-05,
      "loss": 0.3745,
      "step": 5100
    },
    {
      "epoch": 0.5025693983427996,
      "grad_norm": 1.7303107976913452,
      "learning_rate": 1.8171521678655172e-05,
      "loss": 0.2556,
      "step": 5110
    },
    {
      "epoch": 0.5035529001008089,
      "grad_norm": 1.0047192573547363,
      "learning_rate": 1.8167547589715058e-05,
      "loss": 0.2547,
      "step": 5120
    },
    {
      "epoch": 0.5045364018588183,
      "grad_norm": 1.7902576923370361,
      "learning_rate": 1.8163573500774948e-05,
      "loss": 0.2026,
      "step": 5130
    },
    {
      "epoch": 0.5055199036168277,
      "grad_norm": 1.2141826152801514,
      "learning_rate": 1.8159599411834838e-05,
      "loss": 0.3075,
      "step": 5140
    },
    {
      "epoch": 0.5065034053748371,
      "grad_norm": 1.0879597663879395,
      "learning_rate": 1.8155625322894727e-05,
      "loss": 0.3035,
      "step": 5150
    },
    {
      "epoch": 0.5074869071328465,
      "grad_norm": 0.6125375032424927,
      "learning_rate": 1.8151651233954617e-05,
      "loss": 0.1975,
      "step": 5160
    },
    {
      "epoch": 0.5084704088908559,
      "grad_norm": 1.361868143081665,
      "learning_rate": 1.8147677145014507e-05,
      "loss": 0.4418,
      "step": 5170
    },
    {
      "epoch": 0.5094539106488652,
      "grad_norm": 2.4404656887054443,
      "learning_rate": 1.8143703056074396e-05,
      "loss": 0.2808,
      "step": 5180
    },
    {
      "epoch": 0.5104374124068747,
      "grad_norm": 3.8543829917907715,
      "learning_rate": 1.8139728967134286e-05,
      "loss": 0.1814,
      "step": 5190
    },
    {
      "epoch": 0.5114209141648841,
      "grad_norm": 1.682686448097229,
      "learning_rate": 1.8135754878194176e-05,
      "loss": 0.2773,
      "step": 5200
    },
    {
      "epoch": 0.5124044159228934,
      "grad_norm": 0.599181056022644,
      "learning_rate": 1.8131780789254065e-05,
      "loss": 0.1515,
      "step": 5210
    },
    {
      "epoch": 0.5133879176809029,
      "grad_norm": 1.6492522954940796,
      "learning_rate": 1.8127806700313955e-05,
      "loss": 0.3122,
      "step": 5220
    },
    {
      "epoch": 0.5143714194389123,
      "grad_norm": 1.381274938583374,
      "learning_rate": 1.8123832611373845e-05,
      "loss": 0.2715,
      "step": 5230
    },
    {
      "epoch": 0.5153549211969216,
      "grad_norm": 3.652162790298462,
      "learning_rate": 1.8119858522433734e-05,
      "loss": 0.361,
      "step": 5240
    },
    {
      "epoch": 0.516338422954931,
      "grad_norm": 0.7934686541557312,
      "learning_rate": 1.8115884433493624e-05,
      "loss": 0.2311,
      "step": 5250
    },
    {
      "epoch": 0.5173219247129405,
      "grad_norm": 1.178621530532837,
      "learning_rate": 1.811191034455351e-05,
      "loss": 0.3368,
      "step": 5260
    },
    {
      "epoch": 0.5183054264709498,
      "grad_norm": 0.9035441279411316,
      "learning_rate": 1.8107936255613403e-05,
      "loss": 0.2595,
      "step": 5270
    },
    {
      "epoch": 0.5192889282289592,
      "grad_norm": 3.2949626445770264,
      "learning_rate": 1.8103962166673293e-05,
      "loss": 0.3289,
      "step": 5280
    },
    {
      "epoch": 0.5202724299869687,
      "grad_norm": 2.0482380390167236,
      "learning_rate": 1.8099988077733183e-05,
      "loss": 0.2056,
      "step": 5290
    },
    {
      "epoch": 0.521255931744978,
      "grad_norm": 0.5833696722984314,
      "learning_rate": 1.8096013988793072e-05,
      "loss": 0.334,
      "step": 5300
    },
    {
      "epoch": 0.5222394335029874,
      "grad_norm": 0.6023494005203247,
      "learning_rate": 1.809203989985296e-05,
      "loss": 0.3494,
      "step": 5310
    },
    {
      "epoch": 0.5232229352609967,
      "grad_norm": 0.8255459666252136,
      "learning_rate": 1.8088065810912848e-05,
      "loss": 0.2537,
      "step": 5320
    },
    {
      "epoch": 0.5242064370190062,
      "grad_norm": 1.5937799215316772,
      "learning_rate": 1.8084091721972738e-05,
      "loss": 0.2484,
      "step": 5330
    },
    {
      "epoch": 0.5251899387770156,
      "grad_norm": 0.9566192030906677,
      "learning_rate": 1.8080117633032628e-05,
      "loss": 0.2696,
      "step": 5340
    },
    {
      "epoch": 0.5261734405350249,
      "grad_norm": 2.102485179901123,
      "learning_rate": 1.807614354409252e-05,
      "loss": 0.3206,
      "step": 5350
    },
    {
      "epoch": 0.5271569422930343,
      "grad_norm": 1.0212339162826538,
      "learning_rate": 1.8072169455152407e-05,
      "loss": 0.2393,
      "step": 5360
    },
    {
      "epoch": 0.5281404440510438,
      "grad_norm": 0.5305758118629456,
      "learning_rate": 1.8068195366212297e-05,
      "loss": 0.2471,
      "step": 5370
    },
    {
      "epoch": 0.5291239458090531,
      "grad_norm": 0.722772479057312,
      "learning_rate": 1.8064221277272186e-05,
      "loss": 0.3649,
      "step": 5380
    },
    {
      "epoch": 0.5301074475670625,
      "grad_norm": 2.7323200702667236,
      "learning_rate": 1.8060247188332076e-05,
      "loss": 0.2667,
      "step": 5390
    },
    {
      "epoch": 0.531090949325072,
      "grad_norm": 0.9532175660133362,
      "learning_rate": 1.8056273099391966e-05,
      "loss": 0.3844,
      "step": 5400
    },
    {
      "epoch": 0.5320744510830813,
      "grad_norm": 1.4109523296356201,
      "learning_rate": 1.8052299010451855e-05,
      "loss": 0.3101,
      "step": 5410
    },
    {
      "epoch": 0.5330579528410907,
      "grad_norm": 1.0147038698196411,
      "learning_rate": 1.8048324921511745e-05,
      "loss": 0.2475,
      "step": 5420
    },
    {
      "epoch": 0.5340414545991001,
      "grad_norm": 1.0827841758728027,
      "learning_rate": 1.8044350832571635e-05,
      "loss": 0.2254,
      "step": 5430
    },
    {
      "epoch": 0.5350249563571094,
      "grad_norm": 0.37398141622543335,
      "learning_rate": 1.8040376743631524e-05,
      "loss": 0.2033,
      "step": 5440
    },
    {
      "epoch": 0.5360084581151189,
      "grad_norm": 0.6746113300323486,
      "learning_rate": 1.8036402654691414e-05,
      "loss": 0.3645,
      "step": 5450
    },
    {
      "epoch": 0.5369919598731283,
      "grad_norm": 1.3214387893676758,
      "learning_rate": 1.8032428565751304e-05,
      "loss": 0.1994,
      "step": 5460
    },
    {
      "epoch": 0.5379754616311376,
      "grad_norm": 2.1967718601226807,
      "learning_rate": 1.8028454476811193e-05,
      "loss": 0.2416,
      "step": 5470
    },
    {
      "epoch": 0.5389589633891471,
      "grad_norm": 2.4310414791107178,
      "learning_rate": 1.8024480387871083e-05,
      "loss": 0.4078,
      "step": 5480
    },
    {
      "epoch": 0.5399424651471565,
      "grad_norm": 0.9052999019622803,
      "learning_rate": 1.8020506298930972e-05,
      "loss": 0.3281,
      "step": 5490
    },
    {
      "epoch": 0.5409259669051658,
      "grad_norm": 2.1289238929748535,
      "learning_rate": 1.801653220999086e-05,
      "loss": 0.252,
      "step": 5500
    },
    {
      "epoch": 0.5409259669051658,
      "eval_loss": 0.18257686495780945,
      "eval_runtime": 16.2375,
      "eval_samples_per_second": 3.079,
      "eval_steps_per_second": 1.54,
      "step": 5500
    },
    {
      "epoch": 0.5419094686631752,
      "grad_norm": 1.422521948814392,
      "learning_rate": 1.8012558121050752e-05,
      "loss": 0.2931,
      "step": 5510
    },
    {
      "epoch": 0.5428929704211847,
      "grad_norm": 0.9348315000534058,
      "learning_rate": 1.800858403211064e-05,
      "loss": 0.2606,
      "step": 5520
    },
    {
      "epoch": 0.543876472179194,
      "grad_norm": 1.98324453830719,
      "learning_rate": 1.800460994317053e-05,
      "loss": 0.194,
      "step": 5530
    },
    {
      "epoch": 0.5448599739372034,
      "grad_norm": 1.9873881340026855,
      "learning_rate": 1.800063585423042e-05,
      "loss": 0.2718,
      "step": 5540
    },
    {
      "epoch": 0.5458434756952129,
      "grad_norm": 0.650570809841156,
      "learning_rate": 1.7996661765290307e-05,
      "loss": 0.1921,
      "step": 5550
    },
    {
      "epoch": 0.5468269774532222,
      "grad_norm": 1.0361882448196411,
      "learning_rate": 1.7992687676350197e-05,
      "loss": 0.1932,
      "step": 5560
    },
    {
      "epoch": 0.5478104792112316,
      "grad_norm": 1.8897120952606201,
      "learning_rate": 1.7988713587410086e-05,
      "loss": 0.254,
      "step": 5570
    },
    {
      "epoch": 0.548793980969241,
      "grad_norm": 1.37480890750885,
      "learning_rate": 1.7984739498469976e-05,
      "loss": 0.2603,
      "step": 5580
    },
    {
      "epoch": 0.5497774827272504,
      "grad_norm": 1.5733188390731812,
      "learning_rate": 1.798076540952987e-05,
      "loss": 0.3241,
      "step": 5590
    },
    {
      "epoch": 0.5507609844852598,
      "grad_norm": 1.8521591424942017,
      "learning_rate": 1.7976791320589755e-05,
      "loss": 0.3151,
      "step": 5600
    },
    {
      "epoch": 0.5517444862432691,
      "grad_norm": 1.0588831901550293,
      "learning_rate": 1.7972817231649645e-05,
      "loss": 0.1527,
      "step": 5610
    },
    {
      "epoch": 0.5527279880012785,
      "grad_norm": 0.48786550760269165,
      "learning_rate": 1.7968843142709535e-05,
      "loss": 0.2082,
      "step": 5620
    },
    {
      "epoch": 0.553711489759288,
      "grad_norm": 1.9939007759094238,
      "learning_rate": 1.7964869053769424e-05,
      "loss": 0.2558,
      "step": 5630
    },
    {
      "epoch": 0.5546949915172973,
      "grad_norm": 1.3654463291168213,
      "learning_rate": 1.7960894964829314e-05,
      "loss": 0.2808,
      "step": 5640
    },
    {
      "epoch": 0.5556784932753067,
      "grad_norm": 1.640133261680603,
      "learning_rate": 1.7956920875889204e-05,
      "loss": 0.3643,
      "step": 5650
    },
    {
      "epoch": 0.5566619950333161,
      "grad_norm": 1.6016790866851807,
      "learning_rate": 1.7952946786949093e-05,
      "loss": 0.2268,
      "step": 5660
    },
    {
      "epoch": 0.5576454967913255,
      "grad_norm": 1.3572251796722412,
      "learning_rate": 1.7948972698008983e-05,
      "loss": 0.3403,
      "step": 5670
    },
    {
      "epoch": 0.5586289985493349,
      "grad_norm": 1.4128433465957642,
      "learning_rate": 1.7944998609068873e-05,
      "loss": 0.332,
      "step": 5680
    },
    {
      "epoch": 0.5596125003073443,
      "grad_norm": 1.1187504529953003,
      "learning_rate": 1.7941024520128762e-05,
      "loss": 0.2393,
      "step": 5690
    },
    {
      "epoch": 0.5605960020653536,
      "grad_norm": 1.141859531402588,
      "learning_rate": 1.7937050431188652e-05,
      "loss": 0.2431,
      "step": 5700
    },
    {
      "epoch": 0.5615795038233631,
      "grad_norm": 0.8871030211448669,
      "learning_rate": 1.7933076342248542e-05,
      "loss": 0.3159,
      "step": 5710
    },
    {
      "epoch": 0.5625630055813725,
      "grad_norm": 1.2970197200775146,
      "learning_rate": 1.792910225330843e-05,
      "loss": 0.2678,
      "step": 5720
    },
    {
      "epoch": 0.5635465073393818,
      "grad_norm": 1.7814615964889526,
      "learning_rate": 1.792512816436832e-05,
      "loss": 0.2298,
      "step": 5730
    },
    {
      "epoch": 0.5645300090973913,
      "grad_norm": 1.900589108467102,
      "learning_rate": 1.7921154075428207e-05,
      "loss": 0.3047,
      "step": 5740
    },
    {
      "epoch": 0.5655135108554007,
      "grad_norm": 0.9044830203056335,
      "learning_rate": 1.7917179986488097e-05,
      "loss": 0.3199,
      "step": 5750
    },
    {
      "epoch": 0.56649701261341,
      "grad_norm": 1.0888880491256714,
      "learning_rate": 1.791320589754799e-05,
      "loss": 0.2272,
      "step": 5760
    },
    {
      "epoch": 0.5674805143714194,
      "grad_norm": 4.100197792053223,
      "learning_rate": 1.790923180860788e-05,
      "loss": 0.322,
      "step": 5770
    },
    {
      "epoch": 0.5684640161294289,
      "grad_norm": 0.7008423209190369,
      "learning_rate": 1.790525771966777e-05,
      "loss": 0.2291,
      "step": 5780
    },
    {
      "epoch": 0.5694475178874382,
      "grad_norm": 1.2595419883728027,
      "learning_rate": 1.7901283630727656e-05,
      "loss": 0.2018,
      "step": 5790
    },
    {
      "epoch": 0.5704310196454476,
      "grad_norm": 1.9947835206985474,
      "learning_rate": 1.7897309541787545e-05,
      "loss": 0.3107,
      "step": 5800
    },
    {
      "epoch": 0.571414521403457,
      "grad_norm": 0.5234070420265198,
      "learning_rate": 1.7893335452847435e-05,
      "loss": 0.2079,
      "step": 5810
    },
    {
      "epoch": 0.5723980231614664,
      "grad_norm": 2.341140031814575,
      "learning_rate": 1.7889361363907325e-05,
      "loss": 0.2779,
      "step": 5820
    },
    {
      "epoch": 0.5733815249194758,
      "grad_norm": 1.9745830297470093,
      "learning_rate": 1.7885387274967214e-05,
      "loss": 0.2857,
      "step": 5830
    },
    {
      "epoch": 0.5743650266774852,
      "grad_norm": 2.5059823989868164,
      "learning_rate": 1.7881413186027104e-05,
      "loss": 0.2123,
      "step": 5840
    },
    {
      "epoch": 0.5753485284354946,
      "grad_norm": 1.8198421001434326,
      "learning_rate": 1.7877439097086994e-05,
      "loss": 0.2633,
      "step": 5850
    },
    {
      "epoch": 0.576332030193504,
      "grad_norm": 1.688056230545044,
      "learning_rate": 1.7873465008146883e-05,
      "loss": 0.2896,
      "step": 5860
    },
    {
      "epoch": 0.5773155319515134,
      "grad_norm": 0.9039472341537476,
      "learning_rate": 1.7869490919206773e-05,
      "loss": 0.2343,
      "step": 5870
    },
    {
      "epoch": 0.5782990337095227,
      "grad_norm": 1.2950299978256226,
      "learning_rate": 1.7865516830266663e-05,
      "loss": 0.2592,
      "step": 5880
    },
    {
      "epoch": 0.5792825354675322,
      "grad_norm": 2.150629997253418,
      "learning_rate": 1.7861542741326552e-05,
      "loss": 0.3204,
      "step": 5890
    },
    {
      "epoch": 0.5802660372255415,
      "grad_norm": 2.5115997791290283,
      "learning_rate": 1.7857568652386442e-05,
      "loss": 0.2205,
      "step": 5900
    },
    {
      "epoch": 0.5812495389835509,
      "grad_norm": 2.0928642749786377,
      "learning_rate": 1.785359456344633e-05,
      "loss": 0.233,
      "step": 5910
    },
    {
      "epoch": 0.5822330407415603,
      "grad_norm": 3.6651504039764404,
      "learning_rate": 1.784962047450622e-05,
      "loss": 0.3126,
      "step": 5920
    },
    {
      "epoch": 0.5832165424995697,
      "grad_norm": 1.2650208473205566,
      "learning_rate": 1.784564638556611e-05,
      "loss": 0.3493,
      "step": 5930
    },
    {
      "epoch": 0.5842000442575791,
      "grad_norm": 2.006185531616211,
      "learning_rate": 1.7841672296626e-05,
      "loss": 0.2449,
      "step": 5940
    },
    {
      "epoch": 0.5851835460155885,
      "grad_norm": 1.3783490657806396,
      "learning_rate": 1.783769820768589e-05,
      "loss": 0.3871,
      "step": 5950
    },
    {
      "epoch": 0.5861670477735978,
      "grad_norm": 2.5335588455200195,
      "learning_rate": 1.783372411874578e-05,
      "loss": 0.3549,
      "step": 5960
    },
    {
      "epoch": 0.5871505495316073,
      "grad_norm": 2.9504337310791016,
      "learning_rate": 1.782975002980567e-05,
      "loss": 0.3744,
      "step": 5970
    },
    {
      "epoch": 0.5881340512896167,
      "grad_norm": 1.9796600341796875,
      "learning_rate": 1.7825775940865556e-05,
      "loss": 0.321,
      "step": 5980
    },
    {
      "epoch": 0.589117553047626,
      "grad_norm": 2.798205614089966,
      "learning_rate": 1.7821801851925446e-05,
      "loss": 0.238,
      "step": 5990
    },
    {
      "epoch": 0.5901010548056355,
      "grad_norm": 1.6959587335586548,
      "learning_rate": 1.781782776298534e-05,
      "loss": 0.3073,
      "step": 6000
    },
    {
      "epoch": 0.5901010548056355,
      "eval_loss": 0.18480060994625092,
      "eval_runtime": 19.2678,
      "eval_samples_per_second": 2.595,
      "eval_steps_per_second": 1.297,
      "step": 6000
    },
    {
      "epoch": 0.5910845565636449,
      "grad_norm": 2.3862385749816895,
      "learning_rate": 1.7813853674045228e-05,
      "loss": 0.3129,
      "step": 6010
    },
    {
      "epoch": 0.5920680583216542,
      "grad_norm": 1.5851318836212158,
      "learning_rate": 1.7809879585105118e-05,
      "loss": 0.2187,
      "step": 6020
    },
    {
      "epoch": 0.5930515600796636,
      "grad_norm": 1.9704920053482056,
      "learning_rate": 1.7805905496165004e-05,
      "loss": 0.2699,
      "step": 6030
    },
    {
      "epoch": 0.5940350618376731,
      "grad_norm": 0.945516049861908,
      "learning_rate": 1.7801931407224894e-05,
      "loss": 0.2358,
      "step": 6040
    },
    {
      "epoch": 0.5950185635956824,
      "grad_norm": 1.1421146392822266,
      "learning_rate": 1.7797957318284784e-05,
      "loss": 0.2533,
      "step": 6050
    },
    {
      "epoch": 0.5960020653536918,
      "grad_norm": 0.9118548035621643,
      "learning_rate": 1.7793983229344673e-05,
      "loss": 0.2338,
      "step": 6060
    },
    {
      "epoch": 0.5969855671117013,
      "grad_norm": 1.8159881830215454,
      "learning_rate": 1.7790009140404563e-05,
      "loss": 0.2702,
      "step": 6070
    },
    {
      "epoch": 0.5979690688697106,
      "grad_norm": 2.794337272644043,
      "learning_rate": 1.7786035051464453e-05,
      "loss": 0.2871,
      "step": 6080
    },
    {
      "epoch": 0.59895257062772,
      "grad_norm": 1.0487737655639648,
      "learning_rate": 1.7782060962524342e-05,
      "loss": 0.2411,
      "step": 6090
    },
    {
      "epoch": 0.5999360723857294,
      "grad_norm": 0.22618043422698975,
      "learning_rate": 1.7778086873584232e-05,
      "loss": 0.2489,
      "step": 6100
    },
    {
      "epoch": 0.6009195741437388,
      "grad_norm": 1.117783546447754,
      "learning_rate": 1.777411278464412e-05,
      "loss": 0.2851,
      "step": 6110
    },
    {
      "epoch": 0.6019030759017482,
      "grad_norm": 0.869219183921814,
      "learning_rate": 1.777013869570401e-05,
      "loss": 0.2425,
      "step": 6120
    },
    {
      "epoch": 0.6028865776597576,
      "grad_norm": 1.8589662313461304,
      "learning_rate": 1.77661646067639e-05,
      "loss": 0.3514,
      "step": 6130
    },
    {
      "epoch": 0.6038700794177669,
      "grad_norm": 0.949151873588562,
      "learning_rate": 1.776219051782379e-05,
      "loss": 0.2864,
      "step": 6140
    },
    {
      "epoch": 0.6048535811757764,
      "grad_norm": 0.8056336045265198,
      "learning_rate": 1.775821642888368e-05,
      "loss": 0.2232,
      "step": 6150
    },
    {
      "epoch": 0.6058370829337858,
      "grad_norm": 1.1552402973175049,
      "learning_rate": 1.775424233994357e-05,
      "loss": 0.2317,
      "step": 6160
    },
    {
      "epoch": 0.6068205846917951,
      "grad_norm": 1.5513499975204468,
      "learning_rate": 1.775026825100346e-05,
      "loss": 0.2486,
      "step": 6170
    },
    {
      "epoch": 0.6078040864498045,
      "grad_norm": 0.8391079306602478,
      "learning_rate": 1.774629416206335e-05,
      "loss": 0.1522,
      "step": 6180
    },
    {
      "epoch": 0.6087875882078139,
      "grad_norm": 4.214269161224365,
      "learning_rate": 1.774232007312324e-05,
      "loss": 0.3992,
      "step": 6190
    },
    {
      "epoch": 0.6097710899658233,
      "grad_norm": 1.27689790725708,
      "learning_rate": 1.773834598418313e-05,
      "loss": 0.2539,
      "step": 6200
    },
    {
      "epoch": 0.6107545917238327,
      "grad_norm": 3.4815011024475098,
      "learning_rate": 1.7734371895243018e-05,
      "loss": 0.3174,
      "step": 6210
    },
    {
      "epoch": 0.611738093481842,
      "grad_norm": 0.5101733803749084,
      "learning_rate": 1.7730397806302904e-05,
      "loss": 0.2232,
      "step": 6220
    },
    {
      "epoch": 0.6127215952398515,
      "grad_norm": 1.796726942062378,
      "learning_rate": 1.7726423717362794e-05,
      "loss": 0.0914,
      "step": 6230
    },
    {
      "epoch": 0.6137050969978609,
      "grad_norm": 1.0965911149978638,
      "learning_rate": 1.7722449628422687e-05,
      "loss": 0.3303,
      "step": 6240
    },
    {
      "epoch": 0.6146885987558702,
      "grad_norm": 1.8056279420852661,
      "learning_rate": 1.7718475539482577e-05,
      "loss": 0.2987,
      "step": 6250
    },
    {
      "epoch": 0.6156721005138797,
      "grad_norm": 3.1960597038269043,
      "learning_rate": 1.7714501450542467e-05,
      "loss": 0.1829,
      "step": 6260
    },
    {
      "epoch": 0.6166556022718891,
      "grad_norm": 0.9879154562950134,
      "learning_rate": 1.7710527361602353e-05,
      "loss": 0.2514,
      "step": 6270
    },
    {
      "epoch": 0.6176391040298984,
      "grad_norm": 1.268825888633728,
      "learning_rate": 1.7706553272662242e-05,
      "loss": 0.3541,
      "step": 6280
    },
    {
      "epoch": 0.6186226057879078,
      "grad_norm": 0.49746277928352356,
      "learning_rate": 1.7702579183722132e-05,
      "loss": 0.35,
      "step": 6290
    },
    {
      "epoch": 0.6196061075459173,
      "grad_norm": 1.2664997577667236,
      "learning_rate": 1.7698605094782022e-05,
      "loss": 0.3302,
      "step": 6300
    },
    {
      "epoch": 0.6205896093039266,
      "grad_norm": 2.224809169769287,
      "learning_rate": 1.769463100584191e-05,
      "loss": 0.265,
      "step": 6310
    },
    {
      "epoch": 0.621573111061936,
      "grad_norm": 0.7661302089691162,
      "learning_rate": 1.76906569169018e-05,
      "loss": 0.163,
      "step": 6320
    },
    {
      "epoch": 0.6225566128199455,
      "grad_norm": 2.224334478378296,
      "learning_rate": 1.768668282796169e-05,
      "loss": 0.2592,
      "step": 6330
    },
    {
      "epoch": 0.6235401145779548,
      "grad_norm": 1.4434432983398438,
      "learning_rate": 1.768270873902158e-05,
      "loss": 0.1518,
      "step": 6340
    },
    {
      "epoch": 0.6245236163359642,
      "grad_norm": 0.9845334887504578,
      "learning_rate": 1.767873465008147e-05,
      "loss": 0.275,
      "step": 6350
    },
    {
      "epoch": 0.6255071180939736,
      "grad_norm": 0.7070971131324768,
      "learning_rate": 1.767476056114136e-05,
      "loss": 0.203,
      "step": 6360
    },
    {
      "epoch": 0.626490619851983,
      "grad_norm": 1.071750521659851,
      "learning_rate": 1.767078647220125e-05,
      "loss": 0.4071,
      "step": 6370
    },
    {
      "epoch": 0.6274741216099924,
      "grad_norm": 0.7713740468025208,
      "learning_rate": 1.766681238326114e-05,
      "loss": 0.2863,
      "step": 6380
    },
    {
      "epoch": 0.6284576233680018,
      "grad_norm": 2.0365185737609863,
      "learning_rate": 1.766283829432103e-05,
      "loss": 0.1939,
      "step": 6390
    },
    {
      "epoch": 0.6294411251260111,
      "grad_norm": 1.9870047569274902,
      "learning_rate": 1.765886420538092e-05,
      "loss": 0.183,
      "step": 6400
    },
    {
      "epoch": 0.6304246268840206,
      "grad_norm": 0.9188165068626404,
      "learning_rate": 1.7654890116440808e-05,
      "loss": 0.2881,
      "step": 6410
    },
    {
      "epoch": 0.63140812864203,
      "grad_norm": 1.0348327159881592,
      "learning_rate": 1.7650916027500698e-05,
      "loss": 0.3133,
      "step": 6420
    },
    {
      "epoch": 0.6323916304000393,
      "grad_norm": 0.5961625576019287,
      "learning_rate": 1.7646941938560587e-05,
      "loss": 0.3248,
      "step": 6430
    },
    {
      "epoch": 0.6333751321580487,
      "grad_norm": 1.9765043258666992,
      "learning_rate": 1.7642967849620477e-05,
      "loss": 0.261,
      "step": 6440
    },
    {
      "epoch": 0.6343586339160582,
      "grad_norm": 1.0637900829315186,
      "learning_rate": 1.7638993760680367e-05,
      "loss": 0.3368,
      "step": 6450
    },
    {
      "epoch": 0.6353421356740675,
      "grad_norm": 1.868017554283142,
      "learning_rate": 1.7635019671740253e-05,
      "loss": 0.303,
      "step": 6460
    },
    {
      "epoch": 0.6363256374320769,
      "grad_norm": 1.0811495780944824,
      "learning_rate": 1.7631045582800143e-05,
      "loss": 0.3333,
      "step": 6470
    },
    {
      "epoch": 0.6373091391900862,
      "grad_norm": 1.6021066904067993,
      "learning_rate": 1.7627071493860036e-05,
      "loss": 0.2341,
      "step": 6480
    },
    {
      "epoch": 0.6382926409480957,
      "grad_norm": 0.8280308246612549,
      "learning_rate": 1.7623097404919925e-05,
      "loss": 0.2931,
      "step": 6490
    },
    {
      "epoch": 0.6392761427061051,
      "grad_norm": 0.4079805016517639,
      "learning_rate": 1.7619123315979815e-05,
      "loss": 0.2615,
      "step": 6500
    },
    {
      "epoch": 0.6392761427061051,
      "eval_loss": 0.16756634414196014,
      "eval_runtime": 21.0382,
      "eval_samples_per_second": 2.377,
      "eval_steps_per_second": 1.188,
      "step": 6500
    },
    {
      "epoch": 0.6402596444641144,
      "grad_norm": 1.9855930805206299,
      "learning_rate": 1.76151492270397e-05,
      "loss": 0.3118,
      "step": 6510
    },
    {
      "epoch": 0.6412431462221239,
      "grad_norm": 2.828648090362549,
      "learning_rate": 1.761117513809959e-05,
      "loss": 0.2595,
      "step": 6520
    },
    {
      "epoch": 0.6422266479801333,
      "grad_norm": 1.2599165439605713,
      "learning_rate": 1.760720104915948e-05,
      "loss": 0.2549,
      "step": 6530
    },
    {
      "epoch": 0.6432101497381426,
      "grad_norm": 2.565375804901123,
      "learning_rate": 1.760322696021937e-05,
      "loss": 0.4036,
      "step": 6540
    },
    {
      "epoch": 0.644193651496152,
      "grad_norm": 1.0683103799819946,
      "learning_rate": 1.759925287127926e-05,
      "loss": 0.2982,
      "step": 6550
    },
    {
      "epoch": 0.6451771532541615,
      "grad_norm": 0.5626850128173828,
      "learning_rate": 1.759527878233915e-05,
      "loss": 0.2816,
      "step": 6560
    },
    {
      "epoch": 0.6461606550121708,
      "grad_norm": 2.2032320499420166,
      "learning_rate": 1.759130469339904e-05,
      "loss": 0.2405,
      "step": 6570
    },
    {
      "epoch": 0.6471441567701802,
      "grad_norm": 1.1257829666137695,
      "learning_rate": 1.758733060445893e-05,
      "loss": 0.3195,
      "step": 6580
    },
    {
      "epoch": 0.6481276585281897,
      "grad_norm": 1.7821928262710571,
      "learning_rate": 1.758335651551882e-05,
      "loss": 0.2793,
      "step": 6590
    },
    {
      "epoch": 0.649111160286199,
      "grad_norm": 1.9086956977844238,
      "learning_rate": 1.757938242657871e-05,
      "loss": 0.2243,
      "step": 6600
    },
    {
      "epoch": 0.6500946620442084,
      "grad_norm": 0.624267041683197,
      "learning_rate": 1.7575408337638598e-05,
      "loss": 0.2452,
      "step": 6610
    },
    {
      "epoch": 0.6510781638022178,
      "grad_norm": 0.9951871037483215,
      "learning_rate": 1.7571434248698488e-05,
      "loss": 0.3167,
      "step": 6620
    },
    {
      "epoch": 0.6520616655602272,
      "grad_norm": 0.869670033454895,
      "learning_rate": 1.7567460159758377e-05,
      "loss": 0.2729,
      "step": 6630
    },
    {
      "epoch": 0.6530451673182366,
      "grad_norm": 2.440999984741211,
      "learning_rate": 1.7563486070818264e-05,
      "loss": 0.2323,
      "step": 6640
    },
    {
      "epoch": 0.654028669076246,
      "grad_norm": 2.6656980514526367,
      "learning_rate": 1.7559511981878157e-05,
      "loss": 0.2467,
      "step": 6650
    },
    {
      "epoch": 0.6550121708342553,
      "grad_norm": 1.7781723737716675,
      "learning_rate": 1.7555537892938046e-05,
      "loss": 0.3012,
      "step": 6660
    },
    {
      "epoch": 0.6559956725922648,
      "grad_norm": 2.4312820434570312,
      "learning_rate": 1.7551563803997936e-05,
      "loss": 0.3306,
      "step": 6670
    },
    {
      "epoch": 0.6569791743502742,
      "grad_norm": 1.759849190711975,
      "learning_rate": 1.7547589715057826e-05,
      "loss": 0.2786,
      "step": 6680
    },
    {
      "epoch": 0.6579626761082835,
      "grad_norm": 3.491177558898926,
      "learning_rate": 1.7543615626117715e-05,
      "loss": 0.2349,
      "step": 6690
    },
    {
      "epoch": 0.658946177866293,
      "grad_norm": 0.7771234512329102,
      "learning_rate": 1.75396415371776e-05,
      "loss": 0.2444,
      "step": 6700
    },
    {
      "epoch": 0.6599296796243024,
      "grad_norm": 0.4858251214027405,
      "learning_rate": 1.753566744823749e-05,
      "loss": 0.3178,
      "step": 6710
    },
    {
      "epoch": 0.6609131813823117,
      "grad_norm": 0.627582848072052,
      "learning_rate": 1.753169335929738e-05,
      "loss": 0.1489,
      "step": 6720
    },
    {
      "epoch": 0.6618966831403211,
      "grad_norm": 0.4715290665626526,
      "learning_rate": 1.7527719270357274e-05,
      "loss": 0.3177,
      "step": 6730
    },
    {
      "epoch": 0.6628801848983306,
      "grad_norm": 2.229329824447632,
      "learning_rate": 1.7523745181417164e-05,
      "loss": 0.3776,
      "step": 6740
    },
    {
      "epoch": 0.6638636866563399,
      "grad_norm": 1.317760944366455,
      "learning_rate": 1.751977109247705e-05,
      "loss": 0.284,
      "step": 6750
    },
    {
      "epoch": 0.6648471884143493,
      "grad_norm": 0.655670166015625,
      "learning_rate": 1.751579700353694e-05,
      "loss": 0.3467,
      "step": 6760
    },
    {
      "epoch": 0.6658306901723586,
      "grad_norm": 0.7603923678398132,
      "learning_rate": 1.751182291459683e-05,
      "loss": 0.2619,
      "step": 6770
    },
    {
      "epoch": 0.6668141919303681,
      "grad_norm": 1.0717917680740356,
      "learning_rate": 1.750784882565672e-05,
      "loss": 0.1964,
      "step": 6780
    },
    {
      "epoch": 0.6677976936883775,
      "grad_norm": 2.8406436443328857,
      "learning_rate": 1.750387473671661e-05,
      "loss": 0.2389,
      "step": 6790
    },
    {
      "epoch": 0.6687811954463868,
      "grad_norm": 1.8166040182113647,
      "learning_rate": 1.7499900647776498e-05,
      "loss": 0.2813,
      "step": 6800
    },
    {
      "epoch": 0.6697646972043962,
      "grad_norm": 0.37853848934173584,
      "learning_rate": 1.7495926558836388e-05,
      "loss": 0.3369,
      "step": 6810
    },
    {
      "epoch": 0.6707481989624057,
      "grad_norm": 1.757017731666565,
      "learning_rate": 1.7491952469896278e-05,
      "loss": 0.1964,
      "step": 6820
    },
    {
      "epoch": 0.671731700720415,
      "grad_norm": 0.8635709881782532,
      "learning_rate": 1.7487978380956167e-05,
      "loss": 0.2284,
      "step": 6830
    },
    {
      "epoch": 0.6727152024784244,
      "grad_norm": 1.8809075355529785,
      "learning_rate": 1.7484004292016057e-05,
      "loss": 0.3185,
      "step": 6840
    },
    {
      "epoch": 0.6736987042364339,
      "grad_norm": 2.1071417331695557,
      "learning_rate": 1.7480030203075947e-05,
      "loss": 0.2415,
      "step": 6850
    },
    {
      "epoch": 0.6746822059944432,
      "grad_norm": 1.8228193521499634,
      "learning_rate": 1.7476056114135836e-05,
      "loss": 0.2727,
      "step": 6860
    },
    {
      "epoch": 0.6756657077524526,
      "grad_norm": 1.8638427257537842,
      "learning_rate": 1.7472082025195726e-05,
      "loss": 0.3047,
      "step": 6870
    },
    {
      "epoch": 0.676649209510462,
      "grad_norm": 0.2602676749229431,
      "learning_rate": 1.7468107936255612e-05,
      "loss": 0.3469,
      "step": 6880
    },
    {
      "epoch": 0.6776327112684714,
      "grad_norm": 0.9284095764160156,
      "learning_rate": 1.7464133847315505e-05,
      "loss": 0.2939,
      "step": 6890
    },
    {
      "epoch": 0.6786162130264808,
      "grad_norm": 2.8218131065368652,
      "learning_rate": 1.7460159758375395e-05,
      "loss": 0.2497,
      "step": 6900
    },
    {
      "epoch": 0.6795997147844902,
      "grad_norm": 1.4904463291168213,
      "learning_rate": 1.7456185669435285e-05,
      "loss": 0.4435,
      "step": 6910
    },
    {
      "epoch": 0.6805832165424995,
      "grad_norm": 0.9148056507110596,
      "learning_rate": 1.7452211580495174e-05,
      "loss": 0.2398,
      "step": 6920
    },
    {
      "epoch": 0.681566718300509,
      "grad_norm": 1.0198630094528198,
      "learning_rate": 1.7448237491555064e-05,
      "loss": 0.3254,
      "step": 6930
    },
    {
      "epoch": 0.6825502200585184,
      "grad_norm": 1.9961555004119873,
      "learning_rate": 1.744426340261495e-05,
      "loss": 0.2865,
      "step": 6940
    },
    {
      "epoch": 0.6835337218165277,
      "grad_norm": 1.7275830507278442,
      "learning_rate": 1.744028931367484e-05,
      "loss": 0.2833,
      "step": 6950
    },
    {
      "epoch": 0.6845172235745371,
      "grad_norm": 1.0084987878799438,
      "learning_rate": 1.743631522473473e-05,
      "loss": 0.283,
      "step": 6960
    },
    {
      "epoch": 0.6855007253325466,
      "grad_norm": 2.165138006210327,
      "learning_rate": 1.7432341135794623e-05,
      "loss": 0.3183,
      "step": 6970
    },
    {
      "epoch": 0.6864842270905559,
      "grad_norm": 1.2501497268676758,
      "learning_rate": 1.7428367046854512e-05,
      "loss": 0.2878,
      "step": 6980
    },
    {
      "epoch": 0.6874677288485653,
      "grad_norm": 1.473513126373291,
      "learning_rate": 1.74243929579144e-05,
      "loss": 0.2705,
      "step": 6990
    },
    {
      "epoch": 0.6884512306065748,
      "grad_norm": 2.8227286338806152,
      "learning_rate": 1.7420418868974288e-05,
      "loss": 0.2592,
      "step": 7000
    },
    {
      "epoch": 0.6884512306065748,
      "eval_loss": 0.17142616212368011,
      "eval_runtime": 17.0015,
      "eval_samples_per_second": 2.941,
      "eval_steps_per_second": 1.47,
      "step": 7000
    },
    {
      "epoch": 0.6894347323645841,
      "grad_norm": 1.9842414855957031,
      "learning_rate": 1.7416444780034178e-05,
      "loss": 0.2982,
      "step": 7010
    },
    {
      "epoch": 0.6904182341225935,
      "grad_norm": 1.5404380559921265,
      "learning_rate": 1.7412470691094067e-05,
      "loss": 0.2632,
      "step": 7020
    },
    {
      "epoch": 0.6914017358806029,
      "grad_norm": 3.1655123233795166,
      "learning_rate": 1.7408496602153957e-05,
      "loss": 0.3676,
      "step": 7030
    },
    {
      "epoch": 0.6923852376386123,
      "grad_norm": 2.0858147144317627,
      "learning_rate": 1.7404522513213847e-05,
      "loss": 0.2786,
      "step": 7040
    },
    {
      "epoch": 0.6933687393966217,
      "grad_norm": 1.6330032348632812,
      "learning_rate": 1.7400548424273736e-05,
      "loss": 0.3458,
      "step": 7050
    },
    {
      "epoch": 0.694352241154631,
      "grad_norm": 1.110235571861267,
      "learning_rate": 1.7396574335333626e-05,
      "loss": 0.2221,
      "step": 7060
    },
    {
      "epoch": 0.6953357429126404,
      "grad_norm": 1.8372716903686523,
      "learning_rate": 1.7392600246393516e-05,
      "loss": 0.3086,
      "step": 7070
    },
    {
      "epoch": 0.6963192446706499,
      "grad_norm": 1.990007996559143,
      "learning_rate": 1.7388626157453405e-05,
      "loss": 0.3037,
      "step": 7080
    },
    {
      "epoch": 0.6973027464286592,
      "grad_norm": 1.34180748462677,
      "learning_rate": 1.7384652068513295e-05,
      "loss": 0.2019,
      "step": 7090
    },
    {
      "epoch": 0.6982862481866686,
      "grad_norm": 1.4167652130126953,
      "learning_rate": 1.7380677979573185e-05,
      "loss": 0.2586,
      "step": 7100
    },
    {
      "epoch": 0.699269749944678,
      "grad_norm": 2.2565560340881348,
      "learning_rate": 1.7376703890633074e-05,
      "loss": 0.2973,
      "step": 7110
    },
    {
      "epoch": 0.7002532517026874,
      "grad_norm": 2.2221834659576416,
      "learning_rate": 1.737272980169296e-05,
      "loss": 0.2,
      "step": 7120
    },
    {
      "epoch": 0.7012367534606968,
      "grad_norm": 1.9819475412368774,
      "learning_rate": 1.7368755712752854e-05,
      "loss": 0.2872,
      "step": 7130
    },
    {
      "epoch": 0.7022202552187062,
      "grad_norm": 2.585352897644043,
      "learning_rate": 1.7364781623812743e-05,
      "loss": 0.2436,
      "step": 7140
    },
    {
      "epoch": 0.7032037569767156,
      "grad_norm": 1.5775268077850342,
      "learning_rate": 1.7360807534872633e-05,
      "loss": 0.2617,
      "step": 7150
    },
    {
      "epoch": 0.704187258734725,
      "grad_norm": 0.8917414546012878,
      "learning_rate": 1.7356833445932523e-05,
      "loss": 0.1701,
      "step": 7160
    },
    {
      "epoch": 0.7051707604927344,
      "grad_norm": 1.4610384702682495,
      "learning_rate": 1.7352859356992412e-05,
      "loss": 0.267,
      "step": 7170
    },
    {
      "epoch": 0.7061542622507437,
      "grad_norm": 1.2714320421218872,
      "learning_rate": 1.73488852680523e-05,
      "loss": 0.2911,
      "step": 7180
    },
    {
      "epoch": 0.7071377640087532,
      "grad_norm": 1.6755682229995728,
      "learning_rate": 1.734491117911219e-05,
      "loss": 0.2565,
      "step": 7190
    },
    {
      "epoch": 0.7081212657667626,
      "grad_norm": 0.6051779985427856,
      "learning_rate": 1.7340937090172078e-05,
      "loss": 0.1993,
      "step": 7200
    },
    {
      "epoch": 0.7091047675247719,
      "grad_norm": 0.8265473246574402,
      "learning_rate": 1.733696300123197e-05,
      "loss": 0.2547,
      "step": 7210
    },
    {
      "epoch": 0.7100882692827813,
      "grad_norm": 1.1671370267868042,
      "learning_rate": 1.733298891229186e-05,
      "loss": 0.2217,
      "step": 7220
    },
    {
      "epoch": 0.7110717710407908,
      "grad_norm": 2.788594961166382,
      "learning_rate": 1.7329014823351747e-05,
      "loss": 0.2691,
      "step": 7230
    },
    {
      "epoch": 0.7120552727988001,
      "grad_norm": 2.4674558639526367,
      "learning_rate": 1.7325040734411637e-05,
      "loss": 0.3065,
      "step": 7240
    },
    {
      "epoch": 0.7130387745568095,
      "grad_norm": 1.8941340446472168,
      "learning_rate": 1.7321066645471526e-05,
      "loss": 0.2059,
      "step": 7250
    },
    {
      "epoch": 0.714022276314819,
      "grad_norm": 1.2387781143188477,
      "learning_rate": 1.7317092556531416e-05,
      "loss": 0.2265,
      "step": 7260
    },
    {
      "epoch": 0.7150057780728283,
      "grad_norm": 2.167513370513916,
      "learning_rate": 1.7313118467591306e-05,
      "loss": 0.3098,
      "step": 7270
    },
    {
      "epoch": 0.7159892798308377,
      "grad_norm": 1.3851529359817505,
      "learning_rate": 1.7309144378651195e-05,
      "loss": 0.2835,
      "step": 7280
    },
    {
      "epoch": 0.7169727815888471,
      "grad_norm": 2.2308197021484375,
      "learning_rate": 1.7305170289711085e-05,
      "loss": 0.2906,
      "step": 7290
    },
    {
      "epoch": 0.7179562833468565,
      "grad_norm": 2.374825954437256,
      "learning_rate": 1.7301196200770975e-05,
      "loss": 0.3629,
      "step": 7300
    },
    {
      "epoch": 0.7189397851048659,
      "grad_norm": 1.3319405317306519,
      "learning_rate": 1.7297222111830864e-05,
      "loss": 0.2441,
      "step": 7310
    },
    {
      "epoch": 0.7199232868628753,
      "grad_norm": 2.5196781158447266,
      "learning_rate": 1.7293248022890754e-05,
      "loss": 0.2844,
      "step": 7320
    },
    {
      "epoch": 0.7209067886208846,
      "grad_norm": 0.5706827640533447,
      "learning_rate": 1.7289273933950644e-05,
      "loss": 0.3429,
      "step": 7330
    },
    {
      "epoch": 0.7218902903788941,
      "grad_norm": 0.982522189617157,
      "learning_rate": 1.7285299845010533e-05,
      "loss": 0.2678,
      "step": 7340
    },
    {
      "epoch": 0.7228737921369034,
      "grad_norm": 1.3022701740264893,
      "learning_rate": 1.7281325756070423e-05,
      "loss": 0.2514,
      "step": 7350
    },
    {
      "epoch": 0.7238572938949128,
      "grad_norm": 2.1632461547851562,
      "learning_rate": 1.727735166713031e-05,
      "loss": 0.2359,
      "step": 7360
    },
    {
      "epoch": 0.7248407956529223,
      "grad_norm": 2.1435694694519043,
      "learning_rate": 1.72733775781902e-05,
      "loss": 0.1974,
      "step": 7370
    },
    {
      "epoch": 0.7258242974109316,
      "grad_norm": 2.0765726566314697,
      "learning_rate": 1.7269403489250092e-05,
      "loss": 0.2958,
      "step": 7380
    },
    {
      "epoch": 0.726807799168941,
      "grad_norm": 0.7352653741836548,
      "learning_rate": 1.726542940030998e-05,
      "loss": 0.3014,
      "step": 7390
    },
    {
      "epoch": 0.7277913009269504,
      "grad_norm": 1.3814529180526733,
      "learning_rate": 1.726145531136987e-05,
      "loss": 0.3172,
      "step": 7400
    },
    {
      "epoch": 0.7287748026849598,
      "grad_norm": 0.7797669768333435,
      "learning_rate": 1.725748122242976e-05,
      "loss": 0.2484,
      "step": 7410
    },
    {
      "epoch": 0.7297583044429692,
      "grad_norm": 1.010933518409729,
      "learning_rate": 1.7253507133489647e-05,
      "loss": 0.2504,
      "step": 7420
    },
    {
      "epoch": 0.7307418062009786,
      "grad_norm": 0.7638196349143982,
      "learning_rate": 1.7249533044549537e-05,
      "loss": 0.2663,
      "step": 7430
    },
    {
      "epoch": 0.7317253079589879,
      "grad_norm": 1.047466516494751,
      "learning_rate": 1.7245558955609427e-05,
      "loss": 0.1963,
      "step": 7440
    },
    {
      "epoch": 0.7327088097169974,
      "grad_norm": 0.6840983033180237,
      "learning_rate": 1.7241584866669316e-05,
      "loss": 0.2936,
      "step": 7450
    },
    {
      "epoch": 0.7336923114750068,
      "grad_norm": 0.9364534020423889,
      "learning_rate": 1.723761077772921e-05,
      "loss": 0.2979,
      "step": 7460
    },
    {
      "epoch": 0.7346758132330161,
      "grad_norm": 1.0363335609436035,
      "learning_rate": 1.7233636688789096e-05,
      "loss": 0.3094,
      "step": 7470
    },
    {
      "epoch": 0.7356593149910255,
      "grad_norm": 1.6205226182937622,
      "learning_rate": 1.7229662599848985e-05,
      "loss": 0.2576,
      "step": 7480
    },
    {
      "epoch": 0.736642816749035,
      "grad_norm": 2.2230446338653564,
      "learning_rate": 1.7225688510908875e-05,
      "loss": 0.4179,
      "step": 7490
    },
    {
      "epoch": 0.7376263185070443,
      "grad_norm": 2.2729926109313965,
      "learning_rate": 1.7221714421968765e-05,
      "loss": 0.2167,
      "step": 7500
    },
    {
      "epoch": 0.7376263185070443,
      "eval_loss": 0.17984023690223694,
      "eval_runtime": 19.5642,
      "eval_samples_per_second": 2.556,
      "eval_steps_per_second": 1.278,
      "step": 7500
    },
    {
      "epoch": 0.7386098202650537,
      "grad_norm": 1.362148404121399,
      "learning_rate": 1.7217740333028654e-05,
      "loss": 0.244,
      "step": 7510
    },
    {
      "epoch": 0.7395933220230632,
      "grad_norm": 3.1289501190185547,
      "learning_rate": 1.7213766244088544e-05,
      "loss": 0.3236,
      "step": 7520
    },
    {
      "epoch": 0.7405768237810725,
      "grad_norm": 1.7559800148010254,
      "learning_rate": 1.7209792155148434e-05,
      "loss": 0.3897,
      "step": 7530
    },
    {
      "epoch": 0.7415603255390819,
      "grad_norm": 1.110167145729065,
      "learning_rate": 1.7205818066208323e-05,
      "loss": 0.3277,
      "step": 7540
    },
    {
      "epoch": 0.7425438272970913,
      "grad_norm": 1.0508075952529907,
      "learning_rate": 1.7201843977268213e-05,
      "loss": 0.3637,
      "step": 7550
    },
    {
      "epoch": 0.7435273290551007,
      "grad_norm": 1.4188263416290283,
      "learning_rate": 1.7197869888328103e-05,
      "loss": 0.2367,
      "step": 7560
    },
    {
      "epoch": 0.7445108308131101,
      "grad_norm": 2.6133649349212646,
      "learning_rate": 1.7193895799387992e-05,
      "loss": 0.207,
      "step": 7570
    },
    {
      "epoch": 0.7454943325711195,
      "grad_norm": 0.7872998118400574,
      "learning_rate": 1.7189921710447882e-05,
      "loss": 0.3007,
      "step": 7580
    },
    {
      "epoch": 0.7464778343291288,
      "grad_norm": 1.397812843322754,
      "learning_rate": 1.718594762150777e-05,
      "loss": 0.213,
      "step": 7590
    },
    {
      "epoch": 0.7474613360871383,
      "grad_norm": 1.035705804824829,
      "learning_rate": 1.7181973532567658e-05,
      "loss": 0.3404,
      "step": 7600
    },
    {
      "epoch": 0.7484448378451477,
      "grad_norm": 1.8265693187713623,
      "learning_rate": 1.7177999443627548e-05,
      "loss": 0.3536,
      "step": 7610
    },
    {
      "epoch": 0.749428339603157,
      "grad_norm": 0.547683596611023,
      "learning_rate": 1.717402535468744e-05,
      "loss": 0.2403,
      "step": 7620
    },
    {
      "epoch": 0.7504118413611665,
      "grad_norm": 1.1466445922851562,
      "learning_rate": 1.717005126574733e-05,
      "loss": 0.284,
      "step": 7630
    },
    {
      "epoch": 0.7513953431191758,
      "grad_norm": 1.1045866012573242,
      "learning_rate": 1.716607717680722e-05,
      "loss": 0.3185,
      "step": 7640
    },
    {
      "epoch": 0.7523788448771852,
      "grad_norm": 0.5115216374397278,
      "learning_rate": 1.716210308786711e-05,
      "loss": 0.395,
      "step": 7650
    },
    {
      "epoch": 0.7533623466351946,
      "grad_norm": 0.9218469858169556,
      "learning_rate": 1.7158128998926996e-05,
      "loss": 0.2298,
      "step": 7660
    },
    {
      "epoch": 0.754345848393204,
      "grad_norm": 0.998163640499115,
      "learning_rate": 1.7154154909986886e-05,
      "loss": 0.2902,
      "step": 7670
    },
    {
      "epoch": 0.7553293501512134,
      "grad_norm": 1.475361704826355,
      "learning_rate": 1.7150180821046775e-05,
      "loss": 0.3588,
      "step": 7680
    },
    {
      "epoch": 0.7563128519092228,
      "grad_norm": 0.8406019806861877,
      "learning_rate": 1.7146206732106665e-05,
      "loss": 0.2207,
      "step": 7690
    },
    {
      "epoch": 0.7572963536672321,
      "grad_norm": 1.2818818092346191,
      "learning_rate": 1.7142232643166558e-05,
      "loss": 0.1986,
      "step": 7700
    },
    {
      "epoch": 0.7582798554252416,
      "grad_norm": 0.9673545956611633,
      "learning_rate": 1.7138258554226444e-05,
      "loss": 0.1709,
      "step": 7710
    },
    {
      "epoch": 0.759263357183251,
      "grad_norm": 0.7116090655326843,
      "learning_rate": 1.7134284465286334e-05,
      "loss": 0.2115,
      "step": 7720
    },
    {
      "epoch": 0.7602468589412603,
      "grad_norm": 1.4711158275604248,
      "learning_rate": 1.7130310376346224e-05,
      "loss": 0.2365,
      "step": 7730
    },
    {
      "epoch": 0.7612303606992697,
      "grad_norm": 0.7362675070762634,
      "learning_rate": 1.7126336287406113e-05,
      "loss": 0.2748,
      "step": 7740
    },
    {
      "epoch": 0.7622138624572792,
      "grad_norm": 1.4853023290634155,
      "learning_rate": 1.7122362198466003e-05,
      "loss": 0.1993,
      "step": 7750
    },
    {
      "epoch": 0.7631973642152885,
      "grad_norm": 4.937195777893066,
      "learning_rate": 1.7118388109525893e-05,
      "loss": 0.3454,
      "step": 7760
    },
    {
      "epoch": 0.7641808659732979,
      "grad_norm": 1.7255171537399292,
      "learning_rate": 1.7114414020585782e-05,
      "loss": 0.3258,
      "step": 7770
    },
    {
      "epoch": 0.7651643677313074,
      "grad_norm": 2.4765617847442627,
      "learning_rate": 1.7110439931645672e-05,
      "loss": 0.3158,
      "step": 7780
    },
    {
      "epoch": 0.7661478694893167,
      "grad_norm": 0.8955258131027222,
      "learning_rate": 1.710646584270556e-05,
      "loss": 0.3209,
      "step": 7790
    },
    {
      "epoch": 0.7671313712473261,
      "grad_norm": 2.678504467010498,
      "learning_rate": 1.710249175376545e-05,
      "loss": 0.2641,
      "step": 7800
    },
    {
      "epoch": 0.7681148730053355,
      "grad_norm": 2.017256736755371,
      "learning_rate": 1.709851766482534e-05,
      "loss": 0.318,
      "step": 7810
    },
    {
      "epoch": 0.7690983747633449,
      "grad_norm": 1.0673086643218994,
      "learning_rate": 1.709454357588523e-05,
      "loss": 0.1347,
      "step": 7820
    },
    {
      "epoch": 0.7700818765213543,
      "grad_norm": 2.0851361751556396,
      "learning_rate": 1.709056948694512e-05,
      "loss": 0.2498,
      "step": 7830
    },
    {
      "epoch": 0.7710653782793637,
      "grad_norm": 2.2323834896087646,
      "learning_rate": 1.7086595398005006e-05,
      "loss": 0.3145,
      "step": 7840
    },
    {
      "epoch": 0.772048880037373,
      "grad_norm": 1.902326226234436,
      "learning_rate": 1.7082621309064896e-05,
      "loss": 0.1788,
      "step": 7850
    },
    {
      "epoch": 0.7730323817953825,
      "grad_norm": 1.267497181892395,
      "learning_rate": 1.707864722012479e-05,
      "loss": 0.2404,
      "step": 7860
    },
    {
      "epoch": 0.7740158835533919,
      "grad_norm": 2.428555727005005,
      "learning_rate": 1.707467313118468e-05,
      "loss": 0.2913,
      "step": 7870
    },
    {
      "epoch": 0.7749993853114012,
      "grad_norm": 1.689639925956726,
      "learning_rate": 1.707069904224457e-05,
      "loss": 0.1982,
      "step": 7880
    },
    {
      "epoch": 0.7759828870694107,
      "grad_norm": 3.5605409145355225,
      "learning_rate": 1.7066724953304458e-05,
      "loss": 0.2529,
      "step": 7890
    },
    {
      "epoch": 0.77696638882742,
      "grad_norm": 6.425320148468018,
      "learning_rate": 1.7062750864364344e-05,
      "loss": 0.3126,
      "step": 7900
    },
    {
      "epoch": 0.7779498905854294,
      "grad_norm": 1.4042447805404663,
      "learning_rate": 1.7058776775424234e-05,
      "loss": 0.2954,
      "step": 7910
    },
    {
      "epoch": 0.7789333923434388,
      "grad_norm": 2.0059497356414795,
      "learning_rate": 1.7054802686484124e-05,
      "loss": 0.2812,
      "step": 7920
    },
    {
      "epoch": 0.7799168941014482,
      "grad_norm": 2.8876333236694336,
      "learning_rate": 1.7050828597544013e-05,
      "loss": 0.3512,
      "step": 7930
    },
    {
      "epoch": 0.7809003958594576,
      "grad_norm": 1.6756712198257446,
      "learning_rate": 1.7046854508603906e-05,
      "loss": 0.3118,
      "step": 7940
    },
    {
      "epoch": 0.781883897617467,
      "grad_norm": 0.9000738263130188,
      "learning_rate": 1.7042880419663793e-05,
      "loss": 0.2629,
      "step": 7950
    },
    {
      "epoch": 0.7828673993754763,
      "grad_norm": 2.579625368118286,
      "learning_rate": 1.7038906330723682e-05,
      "loss": 0.2921,
      "step": 7960
    },
    {
      "epoch": 0.7838509011334858,
      "grad_norm": 0.917523205280304,
      "learning_rate": 1.7034932241783572e-05,
      "loss": 0.1828,
      "step": 7970
    },
    {
      "epoch": 0.7848344028914952,
      "grad_norm": 1.431494951248169,
      "learning_rate": 1.7030958152843462e-05,
      "loss": 0.1177,
      "step": 7980
    },
    {
      "epoch": 0.7858179046495045,
      "grad_norm": 1.6248607635498047,
      "learning_rate": 1.702698406390335e-05,
      "loss": 0.3062,
      "step": 7990
    },
    {
      "epoch": 0.786801406407514,
      "grad_norm": 2.2361698150634766,
      "learning_rate": 1.702300997496324e-05,
      "loss": 0.1935,
      "step": 8000
    },
    {
      "epoch": 0.786801406407514,
      "eval_loss": 0.1714898645877838,
      "eval_runtime": 18.9887,
      "eval_samples_per_second": 2.633,
      "eval_steps_per_second": 1.317,
      "step": 8000
    },
    {
      "epoch": 0.7877849081655234,
      "grad_norm": 0.6476970911026001,
      "learning_rate": 1.701903588602313e-05,
      "loss": 0.3041,
      "step": 8010
    },
    {
      "epoch": 0.7887684099235327,
      "grad_norm": 0.468868225812912,
      "learning_rate": 1.701506179708302e-05,
      "loss": 0.2378,
      "step": 8020
    },
    {
      "epoch": 0.7897519116815421,
      "grad_norm": 1.6067599058151245,
      "learning_rate": 1.701108770814291e-05,
      "loss": 0.2721,
      "step": 8030
    },
    {
      "epoch": 0.7907354134395516,
      "grad_norm": 1.386991262435913,
      "learning_rate": 1.70071136192028e-05,
      "loss": 0.2652,
      "step": 8040
    },
    {
      "epoch": 0.7917189151975609,
      "grad_norm": 1.6014008522033691,
      "learning_rate": 1.700313953026269e-05,
      "loss": 0.2458,
      "step": 8050
    },
    {
      "epoch": 0.7927024169555703,
      "grad_norm": 1.2447532415390015,
      "learning_rate": 1.699916544132258e-05,
      "loss": 0.2731,
      "step": 8060
    },
    {
      "epoch": 0.7936859187135797,
      "grad_norm": 2.875056743621826,
      "learning_rate": 1.699519135238247e-05,
      "loss": 0.3028,
      "step": 8070
    },
    {
      "epoch": 0.7946694204715891,
      "grad_norm": 2.680910587310791,
      "learning_rate": 1.6991217263442355e-05,
      "loss": 0.3273,
      "step": 8080
    },
    {
      "epoch": 0.7956529222295985,
      "grad_norm": 1.944817304611206,
      "learning_rate": 1.6987243174502245e-05,
      "loss": 0.2732,
      "step": 8090
    },
    {
      "epoch": 0.7966364239876079,
      "grad_norm": 3.19771409034729,
      "learning_rate": 1.6983269085562134e-05,
      "loss": 0.1702,
      "step": 8100
    },
    {
      "epoch": 0.7976199257456172,
      "grad_norm": 0.1668514907360077,
      "learning_rate": 1.6979294996622027e-05,
      "loss": 0.3662,
      "step": 8110
    },
    {
      "epoch": 0.7986034275036267,
      "grad_norm": 1.421762466430664,
      "learning_rate": 1.6975320907681917e-05,
      "loss": 0.3515,
      "step": 8120
    },
    {
      "epoch": 0.7995869292616361,
      "grad_norm": 0.5891507267951965,
      "learning_rate": 1.6971346818741807e-05,
      "loss": 0.3687,
      "step": 8130
    },
    {
      "epoch": 0.8005704310196454,
      "grad_norm": 2.5104644298553467,
      "learning_rate": 1.6967372729801693e-05,
      "loss": 0.2561,
      "step": 8140
    },
    {
      "epoch": 0.8015539327776549,
      "grad_norm": 0.637360692024231,
      "learning_rate": 1.6963398640861583e-05,
      "loss": 0.1898,
      "step": 8150
    },
    {
      "epoch": 0.8025374345356643,
      "grad_norm": 0.8222745656967163,
      "learning_rate": 1.6959424551921472e-05,
      "loss": 0.2434,
      "step": 8160
    },
    {
      "epoch": 0.8035209362936736,
      "grad_norm": 1.6066304445266724,
      "learning_rate": 1.6955450462981362e-05,
      "loss": 0.354,
      "step": 8170
    },
    {
      "epoch": 0.804504438051683,
      "grad_norm": 1.689774751663208,
      "learning_rate": 1.695147637404125e-05,
      "loss": 0.3564,
      "step": 8180
    },
    {
      "epoch": 0.8054879398096924,
      "grad_norm": 1.4014936685562134,
      "learning_rate": 1.694750228510114e-05,
      "loss": 0.4074,
      "step": 8190
    },
    {
      "epoch": 0.8064714415677018,
      "grad_norm": 0.9569060802459717,
      "learning_rate": 1.694352819616103e-05,
      "loss": 0.2387,
      "step": 8200
    },
    {
      "epoch": 0.8074549433257112,
      "grad_norm": 0.9100037813186646,
      "learning_rate": 1.693955410722092e-05,
      "loss": 0.2636,
      "step": 8210
    },
    {
      "epoch": 0.8084384450837205,
      "grad_norm": 0.5279045701026917,
      "learning_rate": 1.693558001828081e-05,
      "loss": 0.283,
      "step": 8220
    },
    {
      "epoch": 0.80942194684173,
      "grad_norm": 1.4399425983428955,
      "learning_rate": 1.69316059293407e-05,
      "loss": 0.2874,
      "step": 8230
    },
    {
      "epoch": 0.8104054485997394,
      "grad_norm": 0.5070255994796753,
      "learning_rate": 1.692763184040059e-05,
      "loss": 0.2691,
      "step": 8240
    },
    {
      "epoch": 0.8113889503577487,
      "grad_norm": 1.517592430114746,
      "learning_rate": 1.692365775146048e-05,
      "loss": 0.2214,
      "step": 8250
    },
    {
      "epoch": 0.8123724521157581,
      "grad_norm": 0.8653010129928589,
      "learning_rate": 1.691968366252037e-05,
      "loss": 0.2421,
      "step": 8260
    },
    {
      "epoch": 0.8133559538737676,
      "grad_norm": 0.9418889284133911,
      "learning_rate": 1.691570957358026e-05,
      "loss": 0.2321,
      "step": 8270
    },
    {
      "epoch": 0.8143394556317769,
      "grad_norm": 1.4153378009796143,
      "learning_rate": 1.6911735484640148e-05,
      "loss": 0.2954,
      "step": 8280
    },
    {
      "epoch": 0.8153229573897863,
      "grad_norm": 1.4692529439926147,
      "learning_rate": 1.6907761395700038e-05,
      "loss": 0.3166,
      "step": 8290
    },
    {
      "epoch": 0.8163064591477958,
      "grad_norm": 1.7206356525421143,
      "learning_rate": 1.6903787306759928e-05,
      "loss": 0.2786,
      "step": 8300
    },
    {
      "epoch": 0.8172899609058051,
      "grad_norm": 2.9285550117492676,
      "learning_rate": 1.6899813217819817e-05,
      "loss": 0.2223,
      "step": 8310
    },
    {
      "epoch": 0.8182734626638145,
      "grad_norm": 1.6375441551208496,
      "learning_rate": 1.6895839128879704e-05,
      "loss": 0.4594,
      "step": 8320
    },
    {
      "epoch": 0.8192569644218239,
      "grad_norm": 1.23320734500885,
      "learning_rate": 1.6891865039939593e-05,
      "loss": 0.2349,
      "step": 8330
    },
    {
      "epoch": 0.8202404661798333,
      "grad_norm": 0.9066193699836731,
      "learning_rate": 1.6887890950999483e-05,
      "loss": 0.2678,
      "step": 8340
    },
    {
      "epoch": 0.8212239679378427,
      "grad_norm": 2.4347150325775146,
      "learning_rate": 1.6883916862059376e-05,
      "loss": 0.2813,
      "step": 8350
    },
    {
      "epoch": 0.8222074696958521,
      "grad_norm": 1.2904943227767944,
      "learning_rate": 1.6879942773119266e-05,
      "loss": 0.274,
      "step": 8360
    },
    {
      "epoch": 0.8231909714538614,
      "grad_norm": 0.9572737812995911,
      "learning_rate": 1.6875968684179155e-05,
      "loss": 0.2692,
      "step": 8370
    },
    {
      "epoch": 0.8241744732118709,
      "grad_norm": 1.2107280492782593,
      "learning_rate": 1.687199459523904e-05,
      "loss": 0.2212,
      "step": 8380
    },
    {
      "epoch": 0.8251579749698803,
      "grad_norm": 0.5360173583030701,
      "learning_rate": 1.686802050629893e-05,
      "loss": 0.2378,
      "step": 8390
    },
    {
      "epoch": 0.8261414767278896,
      "grad_norm": 1.0714260339736938,
      "learning_rate": 1.686404641735882e-05,
      "loss": 0.2666,
      "step": 8400
    },
    {
      "epoch": 0.827124978485899,
      "grad_norm": 1.5305095911026,
      "learning_rate": 1.686007232841871e-05,
      "loss": 0.2645,
      "step": 8410
    },
    {
      "epoch": 0.8281084802439085,
      "grad_norm": 1.4608124494552612,
      "learning_rate": 1.68560982394786e-05,
      "loss": 0.374,
      "step": 8420
    },
    {
      "epoch": 0.8290919820019178,
      "grad_norm": 3.1354146003723145,
      "learning_rate": 1.685212415053849e-05,
      "loss": 0.2793,
      "step": 8430
    },
    {
      "epoch": 0.8300754837599272,
      "grad_norm": 0.5149635076522827,
      "learning_rate": 1.684815006159838e-05,
      "loss": 0.3259,
      "step": 8440
    },
    {
      "epoch": 0.8310589855179367,
      "grad_norm": 1.0487557649612427,
      "learning_rate": 1.684417597265827e-05,
      "loss": 0.2507,
      "step": 8450
    },
    {
      "epoch": 0.832042487275946,
      "grad_norm": 3.3058054447174072,
      "learning_rate": 1.684020188371816e-05,
      "loss": 0.3226,
      "step": 8460
    },
    {
      "epoch": 0.8330259890339554,
      "grad_norm": 1.1440839767456055,
      "learning_rate": 1.683622779477805e-05,
      "loss": 0.2999,
      "step": 8470
    },
    {
      "epoch": 0.8340094907919647,
      "grad_norm": 1.0936781167984009,
      "learning_rate": 1.6832253705837938e-05,
      "loss": 0.2374,
      "step": 8480
    },
    {
      "epoch": 0.8349929925499742,
      "grad_norm": 2.1155974864959717,
      "learning_rate": 1.6828279616897828e-05,
      "loss": 0.2307,
      "step": 8490
    },
    {
      "epoch": 0.8359764943079836,
      "grad_norm": 0.9060395359992981,
      "learning_rate": 1.6824305527957718e-05,
      "loss": 0.2469,
      "step": 8500
    },
    {
      "epoch": 0.8359764943079836,
      "eval_loss": 0.1610024869441986,
      "eval_runtime": 16.8772,
      "eval_samples_per_second": 2.963,
      "eval_steps_per_second": 1.481,
      "step": 8500
    },
    {
      "epoch": 0.8369599960659929,
      "grad_norm": 1.9168851375579834,
      "learning_rate": 1.6820331439017607e-05,
      "loss": 0.3033,
      "step": 8510
    },
    {
      "epoch": 0.8379434978240023,
      "grad_norm": 1.245879054069519,
      "learning_rate": 1.6816357350077497e-05,
      "loss": 0.3564,
      "step": 8520
    },
    {
      "epoch": 0.8389269995820118,
      "grad_norm": 2.700672149658203,
      "learning_rate": 1.6812383261137387e-05,
      "loss": 0.2612,
      "step": 8530
    },
    {
      "epoch": 0.8399105013400211,
      "grad_norm": 1.1932783126831055,
      "learning_rate": 1.6808409172197276e-05,
      "loss": 0.254,
      "step": 8540
    },
    {
      "epoch": 0.8408940030980305,
      "grad_norm": 1.1512099504470825,
      "learning_rate": 1.6804435083257166e-05,
      "loss": 0.2405,
      "step": 8550
    },
    {
      "epoch": 0.84187750485604,
      "grad_norm": 0.9437609314918518,
      "learning_rate": 1.6800460994317052e-05,
      "loss": 0.2365,
      "step": 8560
    },
    {
      "epoch": 0.8428610066140493,
      "grad_norm": 2.9077837467193604,
      "learning_rate": 1.6796486905376942e-05,
      "loss": 0.3955,
      "step": 8570
    },
    {
      "epoch": 0.8438445083720587,
      "grad_norm": 2.052456855773926,
      "learning_rate": 1.679251281643683e-05,
      "loss": 0.2618,
      "step": 8580
    },
    {
      "epoch": 0.8448280101300681,
      "grad_norm": 0.6724761128425598,
      "learning_rate": 1.6788538727496725e-05,
      "loss": 0.2363,
      "step": 8590
    },
    {
      "epoch": 0.8458115118880775,
      "grad_norm": 3.1223673820495605,
      "learning_rate": 1.6784564638556614e-05,
      "loss": 0.4068,
      "step": 8600
    },
    {
      "epoch": 0.8467950136460869,
      "grad_norm": 0.6590930819511414,
      "learning_rate": 1.6780590549616504e-05,
      "loss": 0.3605,
      "step": 8610
    },
    {
      "epoch": 0.8477785154040963,
      "grad_norm": 1.586635947227478,
      "learning_rate": 1.677661646067639e-05,
      "loss": 0.3407,
      "step": 8620
    },
    {
      "epoch": 0.8487620171621056,
      "grad_norm": 1.203317403793335,
      "learning_rate": 1.677264237173628e-05,
      "loss": 0.2719,
      "step": 8630
    },
    {
      "epoch": 0.8497455189201151,
      "grad_norm": 1.2657392024993896,
      "learning_rate": 1.676866828279617e-05,
      "loss": 0.2217,
      "step": 8640
    },
    {
      "epoch": 0.8507290206781245,
      "grad_norm": 2.0403079986572266,
      "learning_rate": 1.676469419385606e-05,
      "loss": 0.291,
      "step": 8650
    },
    {
      "epoch": 0.8517125224361338,
      "grad_norm": 1.1299984455108643,
      "learning_rate": 1.676072010491595e-05,
      "loss": 0.2155,
      "step": 8660
    },
    {
      "epoch": 0.8526960241941433,
      "grad_norm": 3.073261260986328,
      "learning_rate": 1.675674601597584e-05,
      "loss": 0.2985,
      "step": 8670
    },
    {
      "epoch": 0.8536795259521527,
      "grad_norm": 2.1931588649749756,
      "learning_rate": 1.6752771927035728e-05,
      "loss": 0.2018,
      "step": 8680
    },
    {
      "epoch": 0.854663027710162,
      "grad_norm": 1.0738897323608398,
      "learning_rate": 1.6748797838095618e-05,
      "loss": 0.2533,
      "step": 8690
    },
    {
      "epoch": 0.8556465294681714,
      "grad_norm": 2.443499803543091,
      "learning_rate": 1.6744823749155507e-05,
      "loss": 0.2284,
      "step": 8700
    },
    {
      "epoch": 0.8566300312261809,
      "grad_norm": 1.1267093420028687,
      "learning_rate": 1.6740849660215397e-05,
      "loss": 0.2141,
      "step": 8710
    },
    {
      "epoch": 0.8576135329841902,
      "grad_norm": 0.6887378096580505,
      "learning_rate": 1.6736875571275287e-05,
      "loss": 0.1738,
      "step": 8720
    },
    {
      "epoch": 0.8585970347421996,
      "grad_norm": 1.0813156366348267,
      "learning_rate": 1.6732901482335176e-05,
      "loss": 0.1604,
      "step": 8730
    },
    {
      "epoch": 0.859580536500209,
      "grad_norm": 2.0310513973236084,
      "learning_rate": 1.6728927393395066e-05,
      "loss": 0.3649,
      "step": 8740
    },
    {
      "epoch": 0.8605640382582184,
      "grad_norm": 1.7131977081298828,
      "learning_rate": 1.6724953304454956e-05,
      "loss": 0.2718,
      "step": 8750
    },
    {
      "epoch": 0.8615475400162278,
      "grad_norm": 1.4873853921890259,
      "learning_rate": 1.6720979215514845e-05,
      "loss": 0.3012,
      "step": 8760
    },
    {
      "epoch": 0.8625310417742371,
      "grad_norm": 2.010711669921875,
      "learning_rate": 1.6717005126574735e-05,
      "loss": 0.3212,
      "step": 8770
    },
    {
      "epoch": 0.8635145435322465,
      "grad_norm": 2.3279435634613037,
      "learning_rate": 1.6713031037634625e-05,
      "loss": 0.2521,
      "step": 8780
    },
    {
      "epoch": 0.864498045290256,
      "grad_norm": 2.6230878829956055,
      "learning_rate": 1.6709056948694514e-05,
      "loss": 0.2162,
      "step": 8790
    },
    {
      "epoch": 0.8654815470482653,
      "grad_norm": 1.1890896558761597,
      "learning_rate": 1.67050828597544e-05,
      "loss": 0.2358,
      "step": 8800
    },
    {
      "epoch": 0.8664650488062747,
      "grad_norm": 0.9919412732124329,
      "learning_rate": 1.670110877081429e-05,
      "loss": 0.3321,
      "step": 8810
    },
    {
      "epoch": 0.8674485505642842,
      "grad_norm": 0.4444110095500946,
      "learning_rate": 1.669713468187418e-05,
      "loss": 0.1969,
      "step": 8820
    },
    {
      "epoch": 0.8684320523222935,
      "grad_norm": 0.6418336629867554,
      "learning_rate": 1.6693160592934073e-05,
      "loss": 0.2933,
      "step": 8830
    },
    {
      "epoch": 0.8694155540803029,
      "grad_norm": 1.0164114236831665,
      "learning_rate": 1.6689186503993963e-05,
      "loss": 0.2688,
      "step": 8840
    },
    {
      "epoch": 0.8703990558383123,
      "grad_norm": 2.1948771476745605,
      "learning_rate": 1.6685212415053852e-05,
      "loss": 0.2807,
      "step": 8850
    },
    {
      "epoch": 0.8713825575963217,
      "grad_norm": 1.2677373886108398,
      "learning_rate": 1.668123832611374e-05,
      "loss": 0.2645,
      "step": 8860
    },
    {
      "epoch": 0.8723660593543311,
      "grad_norm": 2.6350746154785156,
      "learning_rate": 1.667726423717363e-05,
      "loss": 0.2108,
      "step": 8870
    },
    {
      "epoch": 0.8733495611123405,
      "grad_norm": 1.0468815565109253,
      "learning_rate": 1.6673290148233518e-05,
      "loss": 0.2295,
      "step": 8880
    },
    {
      "epoch": 0.8743330628703498,
      "grad_norm": 0.9300365447998047,
      "learning_rate": 1.6669316059293408e-05,
      "loss": 0.198,
      "step": 8890
    },
    {
      "epoch": 0.8753165646283593,
      "grad_norm": 1.8281514644622803,
      "learning_rate": 1.6665341970353297e-05,
      "loss": 0.3443,
      "step": 8900
    },
    {
      "epoch": 0.8763000663863687,
      "grad_norm": 0.6567210555076599,
      "learning_rate": 1.6661367881413187e-05,
      "loss": 0.2858,
      "step": 8910
    },
    {
      "epoch": 0.877283568144378,
      "grad_norm": 6.081761360168457,
      "learning_rate": 1.6657393792473077e-05,
      "loss": 0.3481,
      "step": 8920
    },
    {
      "epoch": 0.8782670699023875,
      "grad_norm": 0.5137054324150085,
      "learning_rate": 1.6653419703532966e-05,
      "loss": 0.1864,
      "step": 8930
    },
    {
      "epoch": 0.8792505716603969,
      "grad_norm": 0.3564288318157196,
      "learning_rate": 1.6649445614592856e-05,
      "loss": 0.2181,
      "step": 8940
    },
    {
      "epoch": 0.8802340734184062,
      "grad_norm": 0.3564998209476471,
      "learning_rate": 1.6645471525652746e-05,
      "loss": 0.1426,
      "step": 8950
    },
    {
      "epoch": 0.8812175751764156,
      "grad_norm": 0.7296929359436035,
      "learning_rate": 1.6641497436712635e-05,
      "loss": 0.36,
      "step": 8960
    },
    {
      "epoch": 0.8822010769344251,
      "grad_norm": 1.9744415283203125,
      "learning_rate": 1.6637523347772525e-05,
      "loss": 0.2383,
      "step": 8970
    },
    {
      "epoch": 0.8831845786924344,
      "grad_norm": 1.7271389961242676,
      "learning_rate": 1.6633549258832415e-05,
      "loss": 0.182,
      "step": 8980
    },
    {
      "epoch": 0.8841680804504438,
      "grad_norm": 1.1774275302886963,
      "learning_rate": 1.66295751698923e-05,
      "loss": 0.2283,
      "step": 8990
    },
    {
      "epoch": 0.8851515822084532,
      "grad_norm": 0.5575772523880005,
      "learning_rate": 1.6625601080952194e-05,
      "loss": 0.3044,
      "step": 9000
    },
    {
      "epoch": 0.8851515822084532,
      "eval_loss": 0.1609308272600174,
      "eval_runtime": 19.1494,
      "eval_samples_per_second": 2.611,
      "eval_steps_per_second": 1.306,
      "step": 9000
    },
    {
      "epoch": 0.8861350839664626,
      "grad_norm": 2.1011977195739746,
      "learning_rate": 1.6621626992012084e-05,
      "loss": 0.264,
      "step": 9010
    },
    {
      "epoch": 0.887118585724472,
      "grad_norm": 1.9053205251693726,
      "learning_rate": 1.6617652903071973e-05,
      "loss": 0.2125,
      "step": 9020
    },
    {
      "epoch": 0.8881020874824814,
      "grad_norm": 1.034562349319458,
      "learning_rate": 1.6613678814131863e-05,
      "loss": 0.1994,
      "step": 9030
    },
    {
      "epoch": 0.8890855892404907,
      "grad_norm": 0.4264748990535736,
      "learning_rate": 1.660970472519175e-05,
      "loss": 0.2756,
      "step": 9040
    },
    {
      "epoch": 0.8900690909985002,
      "grad_norm": 2.293015241622925,
      "learning_rate": 1.660573063625164e-05,
      "loss": 0.2315,
      "step": 9050
    },
    {
      "epoch": 0.8910525927565095,
      "grad_norm": 1.570285439491272,
      "learning_rate": 1.660175654731153e-05,
      "loss": 0.3664,
      "step": 9060
    },
    {
      "epoch": 0.8920360945145189,
      "grad_norm": 0.8685938119888306,
      "learning_rate": 1.6597782458371418e-05,
      "loss": 0.2805,
      "step": 9070
    },
    {
      "epoch": 0.8930195962725284,
      "grad_norm": 1.1349201202392578,
      "learning_rate": 1.659380836943131e-05,
      "loss": 0.3772,
      "step": 9080
    },
    {
      "epoch": 0.8940030980305377,
      "grad_norm": 2.440044403076172,
      "learning_rate": 1.65898342804912e-05,
      "loss": 0.2051,
      "step": 9090
    },
    {
      "epoch": 0.8949865997885471,
      "grad_norm": 0.9280198216438293,
      "learning_rate": 1.6585860191551087e-05,
      "loss": 0.2351,
      "step": 9100
    },
    {
      "epoch": 0.8959701015465565,
      "grad_norm": 0.7486225962638855,
      "learning_rate": 1.6581886102610977e-05,
      "loss": 0.2782,
      "step": 9110
    },
    {
      "epoch": 0.8969536033045659,
      "grad_norm": 1.4163689613342285,
      "learning_rate": 1.6577912013670867e-05,
      "loss": 0.3084,
      "step": 9120
    },
    {
      "epoch": 0.8979371050625753,
      "grad_norm": 0.8505094647407532,
      "learning_rate": 1.6573937924730756e-05,
      "loss": 0.2212,
      "step": 9130
    },
    {
      "epoch": 0.8989206068205847,
      "grad_norm": 1.43965482711792,
      "learning_rate": 1.6569963835790646e-05,
      "loss": 0.2867,
      "step": 9140
    },
    {
      "epoch": 0.899904108578594,
      "grad_norm": 1.275529384613037,
      "learning_rate": 1.6565989746850536e-05,
      "loss": 0.2106,
      "step": 9150
    },
    {
      "epoch": 0.9008876103366035,
      "grad_norm": 1.2573920488357544,
      "learning_rate": 1.6562015657910425e-05,
      "loss": 0.2128,
      "step": 9160
    },
    {
      "epoch": 0.9018711120946129,
      "grad_norm": 1.113728642463684,
      "learning_rate": 1.6558041568970315e-05,
      "loss": 0.3224,
      "step": 9170
    },
    {
      "epoch": 0.9028546138526222,
      "grad_norm": 2.209773063659668,
      "learning_rate": 1.6554067480030205e-05,
      "loss": 0.2121,
      "step": 9180
    },
    {
      "epoch": 0.9038381156106317,
      "grad_norm": 1.6331260204315186,
      "learning_rate": 1.6550093391090094e-05,
      "loss": 0.2636,
      "step": 9190
    },
    {
      "epoch": 0.9048216173686411,
      "grad_norm": 0.5104572176933289,
      "learning_rate": 1.6546119302149984e-05,
      "loss": 0.1854,
      "step": 9200
    },
    {
      "epoch": 0.9058051191266504,
      "grad_norm": 2.398444414138794,
      "learning_rate": 1.6542145213209874e-05,
      "loss": 0.2637,
      "step": 9210
    },
    {
      "epoch": 0.9067886208846598,
      "grad_norm": 2.7371108531951904,
      "learning_rate": 1.6538171124269763e-05,
      "loss": 0.329,
      "step": 9220
    },
    {
      "epoch": 0.9077721226426693,
      "grad_norm": 0.7008442878723145,
      "learning_rate": 1.653419703532965e-05,
      "loss": 0.2298,
      "step": 9230
    },
    {
      "epoch": 0.9087556244006786,
      "grad_norm": 1.478804349899292,
      "learning_rate": 1.6530222946389543e-05,
      "loss": 0.2447,
      "step": 9240
    },
    {
      "epoch": 0.909739126158688,
      "grad_norm": 2.0109310150146484,
      "learning_rate": 1.6526248857449432e-05,
      "loss": 0.3323,
      "step": 9250
    },
    {
      "epoch": 0.9107226279166974,
      "grad_norm": 0.9319548010826111,
      "learning_rate": 1.6522274768509322e-05,
      "loss": 0.3731,
      "step": 9260
    },
    {
      "epoch": 0.9117061296747068,
      "grad_norm": 0.899541974067688,
      "learning_rate": 1.651830067956921e-05,
      "loss": 0.2662,
      "step": 9270
    },
    {
      "epoch": 0.9126896314327162,
      "grad_norm": 2.4901561737060547,
      "learning_rate": 1.6514326590629098e-05,
      "loss": 0.2997,
      "step": 9280
    },
    {
      "epoch": 0.9136731331907256,
      "grad_norm": 2.7003557682037354,
      "learning_rate": 1.6510352501688988e-05,
      "loss": 0.2364,
      "step": 9290
    },
    {
      "epoch": 0.914656634948735,
      "grad_norm": 1.2085120677947998,
      "learning_rate": 1.6506378412748877e-05,
      "loss": 0.2667,
      "step": 9300
    },
    {
      "epoch": 0.9156401367067444,
      "grad_norm": 0.7746391892433167,
      "learning_rate": 1.6502404323808767e-05,
      "loss": 0.3026,
      "step": 9310
    },
    {
      "epoch": 0.9166236384647538,
      "grad_norm": 1.1951857805252075,
      "learning_rate": 1.649843023486866e-05,
      "loss": 0.332,
      "step": 9320
    },
    {
      "epoch": 0.9176071402227631,
      "grad_norm": 0.9990243315696716,
      "learning_rate": 1.649445614592855e-05,
      "loss": 0.2422,
      "step": 9330
    },
    {
      "epoch": 0.9185906419807726,
      "grad_norm": 1.8502593040466309,
      "learning_rate": 1.6490482056988436e-05,
      "loss": 0.3554,
      "step": 9340
    },
    {
      "epoch": 0.9195741437387819,
      "grad_norm": 1.1711889505386353,
      "learning_rate": 1.6486507968048325e-05,
      "loss": 0.2293,
      "step": 9350
    },
    {
      "epoch": 0.9205576454967913,
      "grad_norm": 2.536191463470459,
      "learning_rate": 1.6482533879108215e-05,
      "loss": 0.2577,
      "step": 9360
    },
    {
      "epoch": 0.9215411472548007,
      "grad_norm": 0.34721025824546814,
      "learning_rate": 1.6478559790168105e-05,
      "loss": 0.2612,
      "step": 9370
    },
    {
      "epoch": 0.9225246490128101,
      "grad_norm": 1.5637190341949463,
      "learning_rate": 1.6474585701227994e-05,
      "loss": 0.2713,
      "step": 9380
    },
    {
      "epoch": 0.9235081507708195,
      "grad_norm": 2.7713522911071777,
      "learning_rate": 1.6470611612287884e-05,
      "loss": 0.3551,
      "step": 9390
    },
    {
      "epoch": 0.9244916525288289,
      "grad_norm": 1.4273444414138794,
      "learning_rate": 1.6466637523347774e-05,
      "loss": 0.35,
      "step": 9400
    },
    {
      "epoch": 0.9254751542868382,
      "grad_norm": 2.1872549057006836,
      "learning_rate": 1.6462663434407663e-05,
      "loss": 0.2972,
      "step": 9410
    },
    {
      "epoch": 0.9264586560448477,
      "grad_norm": 1.7454928159713745,
      "learning_rate": 1.6458689345467553e-05,
      "loss": 0.3815,
      "step": 9420
    },
    {
      "epoch": 0.9274421578028571,
      "grad_norm": 1.8231502771377563,
      "learning_rate": 1.6454715256527443e-05,
      "loss": 0.2866,
      "step": 9430
    },
    {
      "epoch": 0.9284256595608664,
      "grad_norm": 1.008145809173584,
      "learning_rate": 1.6450741167587332e-05,
      "loss": 0.1845,
      "step": 9440
    },
    {
      "epoch": 0.9294091613188759,
      "grad_norm": 1.538116216659546,
      "learning_rate": 1.6446767078647222e-05,
      "loss": 0.2184,
      "step": 9450
    },
    {
      "epoch": 0.9303926630768853,
      "grad_norm": 1.898038387298584,
      "learning_rate": 1.6442792989707112e-05,
      "loss": 0.2032,
      "step": 9460
    },
    {
      "epoch": 0.9313761648348946,
      "grad_norm": 0.3986801505088806,
      "learning_rate": 1.6438818900766998e-05,
      "loss": 0.3476,
      "step": 9470
    },
    {
      "epoch": 0.932359666592904,
      "grad_norm": 1.6517155170440674,
      "learning_rate": 1.643484481182689e-05,
      "loss": 0.2283,
      "step": 9480
    },
    {
      "epoch": 0.9333431683509135,
      "grad_norm": 2.5912892818450928,
      "learning_rate": 1.643087072288678e-05,
      "loss": 0.235,
      "step": 9490
    },
    {
      "epoch": 0.9343266701089228,
      "grad_norm": 3.804778814315796,
      "learning_rate": 1.642689663394667e-05,
      "loss": 0.2845,
      "step": 9500
    },
    {
      "epoch": 0.9343266701089228,
      "eval_loss": 0.16946163773536682,
      "eval_runtime": 18.2063,
      "eval_samples_per_second": 2.746,
      "eval_steps_per_second": 1.373,
      "step": 9500
    },
    {
      "epoch": 0.9353101718669322,
      "grad_norm": 2.673412799835205,
      "learning_rate": 1.642292254500656e-05,
      "loss": 0.33,
      "step": 9510
    },
    {
      "epoch": 0.9362936736249416,
      "grad_norm": 0.5579766631126404,
      "learning_rate": 1.6418948456066446e-05,
      "loss": 0.4,
      "step": 9520
    },
    {
      "epoch": 0.937277175382951,
      "grad_norm": 1.3179048299789429,
      "learning_rate": 1.6414974367126336e-05,
      "loss": 0.28,
      "step": 9530
    },
    {
      "epoch": 0.9382606771409604,
      "grad_norm": 1.813857078552246,
      "learning_rate": 1.6411000278186226e-05,
      "loss": 0.2542,
      "step": 9540
    },
    {
      "epoch": 0.9392441788989698,
      "grad_norm": 8.707640647888184,
      "learning_rate": 1.6407026189246115e-05,
      "loss": 0.2239,
      "step": 9550
    },
    {
      "epoch": 0.9402276806569791,
      "grad_norm": 0.7323013544082642,
      "learning_rate": 1.640305210030601e-05,
      "loss": 0.2907,
      "step": 9560
    },
    {
      "epoch": 0.9412111824149886,
      "grad_norm": 0.32580333948135376,
      "learning_rate": 1.6399078011365898e-05,
      "loss": 0.2646,
      "step": 9570
    },
    {
      "epoch": 0.942194684172998,
      "grad_norm": 1.0481038093566895,
      "learning_rate": 1.6395103922425784e-05,
      "loss": 0.2836,
      "step": 9580
    },
    {
      "epoch": 0.9431781859310073,
      "grad_norm": 1.2796183824539185,
      "learning_rate": 1.6391129833485674e-05,
      "loss": 0.2562,
      "step": 9590
    },
    {
      "epoch": 0.9441616876890168,
      "grad_norm": 2.2194364070892334,
      "learning_rate": 1.6387155744545564e-05,
      "loss": 0.3578,
      "step": 9600
    },
    {
      "epoch": 0.9451451894470262,
      "grad_norm": 1.8737331628799438,
      "learning_rate": 1.6383181655605453e-05,
      "loss": 0.1999,
      "step": 9610
    },
    {
      "epoch": 0.9461286912050355,
      "grad_norm": 1.1581262350082397,
      "learning_rate": 1.6379207566665343e-05,
      "loss": 0.1792,
      "step": 9620
    },
    {
      "epoch": 0.9471121929630449,
      "grad_norm": 1.0477012395858765,
      "learning_rate": 1.6375233477725233e-05,
      "loss": 0.2503,
      "step": 9630
    },
    {
      "epoch": 0.9480956947210543,
      "grad_norm": 3.4494800567626953,
      "learning_rate": 1.6371259388785122e-05,
      "loss": 0.2906,
      "step": 9640
    },
    {
      "epoch": 0.9490791964790637,
      "grad_norm": 1.5456995964050293,
      "learning_rate": 1.6367285299845012e-05,
      "loss": 0.2774,
      "step": 9650
    },
    {
      "epoch": 0.9500626982370731,
      "grad_norm": 3.0428876876831055,
      "learning_rate": 1.6363311210904902e-05,
      "loss": 0.219,
      "step": 9660
    },
    {
      "epoch": 0.9510461999950824,
      "grad_norm": 0.7785629630088806,
      "learning_rate": 1.635933712196479e-05,
      "loss": 0.2467,
      "step": 9670
    },
    {
      "epoch": 0.9520297017530919,
      "grad_norm": 2.675586462020874,
      "learning_rate": 1.635536303302468e-05,
      "loss": 0.2585,
      "step": 9680
    },
    {
      "epoch": 0.9530132035111013,
      "grad_norm": 2.9050750732421875,
      "learning_rate": 1.635138894408457e-05,
      "loss": 0.3041,
      "step": 9690
    },
    {
      "epoch": 0.9539967052691106,
      "grad_norm": 1.8580806255340576,
      "learning_rate": 1.634741485514446e-05,
      "loss": 0.3407,
      "step": 9700
    },
    {
      "epoch": 0.95498020702712,
      "grad_norm": 1.26399827003479,
      "learning_rate": 1.6343440766204347e-05,
      "loss": 0.3217,
      "step": 9710
    },
    {
      "epoch": 0.9559637087851295,
      "grad_norm": 2.8637497425079346,
      "learning_rate": 1.6339466677264236e-05,
      "loss": 0.1938,
      "step": 9720
    },
    {
      "epoch": 0.9569472105431388,
      "grad_norm": 1.6564754247665405,
      "learning_rate": 1.633549258832413e-05,
      "loss": 0.2557,
      "step": 9730
    },
    {
      "epoch": 0.9579307123011482,
      "grad_norm": 1.5192463397979736,
      "learning_rate": 1.633151849938402e-05,
      "loss": 0.3262,
      "step": 9740
    },
    {
      "epoch": 0.9589142140591577,
      "grad_norm": 5.074928283691406,
      "learning_rate": 1.632754441044391e-05,
      "loss": 0.2573,
      "step": 9750
    },
    {
      "epoch": 0.959897715817167,
      "grad_norm": 0.720064640045166,
      "learning_rate": 1.6323570321503795e-05,
      "loss": 0.2451,
      "step": 9760
    },
    {
      "epoch": 0.9608812175751764,
      "grad_norm": 2.334301233291626,
      "learning_rate": 1.6319596232563685e-05,
      "loss": 0.2748,
      "step": 9770
    },
    {
      "epoch": 0.9618647193331858,
      "grad_norm": 0.672520101070404,
      "learning_rate": 1.6315622143623574e-05,
      "loss": 0.2755,
      "step": 9780
    },
    {
      "epoch": 0.9628482210911952,
      "grad_norm": 1.1749473810195923,
      "learning_rate": 1.6311648054683464e-05,
      "loss": 0.3088,
      "step": 9790
    },
    {
      "epoch": 0.9638317228492046,
      "grad_norm": 1.1671876907348633,
      "learning_rate": 1.6307673965743354e-05,
      "loss": 0.3445,
      "step": 9800
    },
    {
      "epoch": 0.964815224607214,
      "grad_norm": 1.8781317472457886,
      "learning_rate": 1.6303699876803247e-05,
      "loss": 0.1894,
      "step": 9810
    },
    {
      "epoch": 0.9657987263652233,
      "grad_norm": 2.0575361251831055,
      "learning_rate": 1.6299725787863133e-05,
      "loss": 0.4052,
      "step": 9820
    },
    {
      "epoch": 0.9667822281232328,
      "grad_norm": 1.0505220890045166,
      "learning_rate": 1.6295751698923023e-05,
      "loss": 0.2261,
      "step": 9830
    },
    {
      "epoch": 0.9677657298812422,
      "grad_norm": 1.2270594835281372,
      "learning_rate": 1.6291777609982912e-05,
      "loss": 0.3129,
      "step": 9840
    },
    {
      "epoch": 0.9687492316392515,
      "grad_norm": 1.5712100267410278,
      "learning_rate": 1.6287803521042802e-05,
      "loss": 0.2792,
      "step": 9850
    },
    {
      "epoch": 0.969732733397261,
      "grad_norm": 1.1416723728179932,
      "learning_rate": 1.628382943210269e-05,
      "loss": 0.1883,
      "step": 9860
    },
    {
      "epoch": 0.9707162351552704,
      "grad_norm": 2.3748650550842285,
      "learning_rate": 1.627985534316258e-05,
      "loss": 0.3079,
      "step": 9870
    },
    {
      "epoch": 0.9716997369132797,
      "grad_norm": 0.8768253326416016,
      "learning_rate": 1.627588125422247e-05,
      "loss": 0.3344,
      "step": 9880
    },
    {
      "epoch": 0.9726832386712891,
      "grad_norm": 1.2483183145523071,
      "learning_rate": 1.627190716528236e-05,
      "loss": 0.3065,
      "step": 9890
    },
    {
      "epoch": 0.9736667404292986,
      "grad_norm": 0.6402847766876221,
      "learning_rate": 1.626793307634225e-05,
      "loss": 0.271,
      "step": 9900
    },
    {
      "epoch": 0.9746502421873079,
      "grad_norm": 1.9914218187332153,
      "learning_rate": 1.626395898740214e-05,
      "loss": 0.1675,
      "step": 9910
    },
    {
      "epoch": 0.9756337439453173,
      "grad_norm": 1.2543543577194214,
      "learning_rate": 1.625998489846203e-05,
      "loss": 0.2689,
      "step": 9920
    },
    {
      "epoch": 0.9766172457033266,
      "grad_norm": 1.3631449937820435,
      "learning_rate": 1.625601080952192e-05,
      "loss": 0.2454,
      "step": 9930
    },
    {
      "epoch": 0.9776007474613361,
      "grad_norm": 2.1351165771484375,
      "learning_rate": 1.625203672058181e-05,
      "loss": 0.2253,
      "step": 9940
    },
    {
      "epoch": 0.9785842492193455,
      "grad_norm": 0.6863617300987244,
      "learning_rate": 1.6248062631641695e-05,
      "loss": 0.3127,
      "step": 9950
    },
    {
      "epoch": 0.9795677509773548,
      "grad_norm": 1.2482006549835205,
      "learning_rate": 1.6244088542701585e-05,
      "loss": 0.3138,
      "step": 9960
    },
    {
      "epoch": 0.9805512527353643,
      "grad_norm": 1.801457166671753,
      "learning_rate": 1.6240114453761478e-05,
      "loss": 0.2613,
      "step": 9970
    },
    {
      "epoch": 0.9815347544933737,
      "grad_norm": 2.327338457107544,
      "learning_rate": 1.6236140364821368e-05,
      "loss": 0.1701,
      "step": 9980
    },
    {
      "epoch": 0.982518256251383,
      "grad_norm": 0.7098455429077148,
      "learning_rate": 1.6232166275881257e-05,
      "loss": 0.3116,
      "step": 9990
    },
    {
      "epoch": 0.9835017580093924,
      "grad_norm": 1.502175211906433,
      "learning_rate": 1.6228192186941144e-05,
      "loss": 0.274,
      "step": 10000
    },
    {
      "epoch": 0.9835017580093924,
      "eval_loss": 0.16346381604671478,
      "eval_runtime": 17.3264,
      "eval_samples_per_second": 2.886,
      "eval_steps_per_second": 1.443,
      "step": 10000
    },
    {
      "epoch": 0.9844852597674019,
      "grad_norm": 0.8948222398757935,
      "learning_rate": 1.6224218098001033e-05,
      "loss": 0.1495,
      "step": 10010
    },
    {
      "epoch": 0.9854687615254112,
      "grad_norm": 0.9255267381668091,
      "learning_rate": 1.6220244009060923e-05,
      "loss": 0.3118,
      "step": 10020
    },
    {
      "epoch": 0.9864522632834206,
      "grad_norm": 3.8730742931365967,
      "learning_rate": 1.6216269920120813e-05,
      "loss": 0.3908,
      "step": 10030
    },
    {
      "epoch": 0.98743576504143,
      "grad_norm": 1.2976429462432861,
      "learning_rate": 1.6212295831180702e-05,
      "loss": 0.2071,
      "step": 10040
    },
    {
      "epoch": 0.9884192667994394,
      "grad_norm": 0.9446991682052612,
      "learning_rate": 1.6208321742240595e-05,
      "loss": 0.3687,
      "step": 10050
    },
    {
      "epoch": 0.9894027685574488,
      "grad_norm": 3.9752187728881836,
      "learning_rate": 1.620434765330048e-05,
      "loss": 0.2103,
      "step": 10060
    },
    {
      "epoch": 0.9903862703154582,
      "grad_norm": 2.023470878601074,
      "learning_rate": 1.620037356436037e-05,
      "loss": 0.2543,
      "step": 10070
    },
    {
      "epoch": 0.9913697720734675,
      "grad_norm": 1.043302297592163,
      "learning_rate": 1.619639947542026e-05,
      "loss": 0.2258,
      "step": 10080
    },
    {
      "epoch": 0.992353273831477,
      "grad_norm": 3.0941057205200195,
      "learning_rate": 1.619242538648015e-05,
      "loss": 0.3121,
      "step": 10090
    },
    {
      "epoch": 0.9933367755894864,
      "grad_norm": 1.3577592372894287,
      "learning_rate": 1.618845129754004e-05,
      "loss": 0.1366,
      "step": 10100
    },
    {
      "epoch": 0.9943202773474957,
      "grad_norm": 1.7372103929519653,
      "learning_rate": 1.618447720859993e-05,
      "loss": 0.2481,
      "step": 10110
    },
    {
      "epoch": 0.9953037791055052,
      "grad_norm": 2.0239365100860596,
      "learning_rate": 1.618050311965982e-05,
      "loss": 0.3117,
      "step": 10120
    },
    {
      "epoch": 0.9962872808635146,
      "grad_norm": 2.3687198162078857,
      "learning_rate": 1.617652903071971e-05,
      "loss": 0.3065,
      "step": 10130
    },
    {
      "epoch": 0.9972707826215239,
      "grad_norm": 2.562297821044922,
      "learning_rate": 1.61725549417796e-05,
      "loss": 0.2936,
      "step": 10140
    },
    {
      "epoch": 0.9982542843795333,
      "grad_norm": 1.0834999084472656,
      "learning_rate": 1.616858085283949e-05,
      "loss": 0.2266,
      "step": 10150
    },
    {
      "epoch": 0.9992377861375428,
      "grad_norm": 0.6850488185882568,
      "learning_rate": 1.6164606763899378e-05,
      "loss": 0.3046,
      "step": 10160
    },
    {
      "epoch": 1.000221287895552,
      "grad_norm": 1.008405089378357,
      "learning_rate": 1.6160632674959268e-05,
      "loss": 0.307,
      "step": 10170
    },
    {
      "epoch": 1.0012047896535614,
      "grad_norm": 1.3072564601898193,
      "learning_rate": 1.6156658586019158e-05,
      "loss": 0.2647,
      "step": 10180
    },
    {
      "epoch": 1.002188291411571,
      "grad_norm": 1.2347207069396973,
      "learning_rate": 1.6152684497079044e-05,
      "loss": 0.2033,
      "step": 10190
    },
    {
      "epoch": 1.0031717931695803,
      "grad_norm": 1.1007719039916992,
      "learning_rate": 1.6148710408138933e-05,
      "loss": 0.145,
      "step": 10200
    },
    {
      "epoch": 1.0041552949275896,
      "grad_norm": 0.5624765753746033,
      "learning_rate": 1.6144736319198826e-05,
      "loss": 0.1642,
      "step": 10210
    },
    {
      "epoch": 1.0051387966855991,
      "grad_norm": 2.5909204483032227,
      "learning_rate": 1.6140762230258716e-05,
      "loss": 0.2944,
      "step": 10220
    },
    {
      "epoch": 1.0061222984436085,
      "grad_norm": 1.661992073059082,
      "learning_rate": 1.6136788141318606e-05,
      "loss": 0.194,
      "step": 10230
    },
    {
      "epoch": 1.0071058002016178,
      "grad_norm": 1.6907767057418823,
      "learning_rate": 1.6132814052378492e-05,
      "loss": 0.2617,
      "step": 10240
    },
    {
      "epoch": 1.0080893019596273,
      "grad_norm": 1.1987255811691284,
      "learning_rate": 1.6128839963438382e-05,
      "loss": 0.2384,
      "step": 10250
    },
    {
      "epoch": 1.0090728037176366,
      "grad_norm": 0.7288364171981812,
      "learning_rate": 1.612486587449827e-05,
      "loss": 0.244,
      "step": 10260
    },
    {
      "epoch": 1.010056305475646,
      "grad_norm": 1.0956259965896606,
      "learning_rate": 1.612089178555816e-05,
      "loss": 0.2002,
      "step": 10270
    },
    {
      "epoch": 1.0110398072336555,
      "grad_norm": 1.2467955350875854,
      "learning_rate": 1.611691769661805e-05,
      "loss": 0.1787,
      "step": 10280
    },
    {
      "epoch": 1.0120233089916648,
      "grad_norm": 0.8599863648414612,
      "learning_rate": 1.6112943607677944e-05,
      "loss": 0.2282,
      "step": 10290
    },
    {
      "epoch": 1.0130068107496741,
      "grad_norm": 0.4900651276111603,
      "learning_rate": 1.610896951873783e-05,
      "loss": 0.2283,
      "step": 10300
    },
    {
      "epoch": 1.0139903125076837,
      "grad_norm": 1.299378752708435,
      "learning_rate": 1.610499542979772e-05,
      "loss": 0.2481,
      "step": 10310
    },
    {
      "epoch": 1.014973814265693,
      "grad_norm": 1.2478209733963013,
      "learning_rate": 1.610102134085761e-05,
      "loss": 0.2631,
      "step": 10320
    },
    {
      "epoch": 1.0159573160237023,
      "grad_norm": 1.5120837688446045,
      "learning_rate": 1.60970472519175e-05,
      "loss": 0.2013,
      "step": 10330
    },
    {
      "epoch": 1.0169408177817119,
      "grad_norm": 0.5610188245773315,
      "learning_rate": 1.609307316297739e-05,
      "loss": 0.3882,
      "step": 10340
    },
    {
      "epoch": 1.0179243195397212,
      "grad_norm": 1.6709067821502686,
      "learning_rate": 1.608909907403728e-05,
      "loss": 0.2784,
      "step": 10350
    },
    {
      "epoch": 1.0189078212977305,
      "grad_norm": 0.8348563313484192,
      "learning_rate": 1.6085124985097168e-05,
      "loss": 0.2556,
      "step": 10360
    },
    {
      "epoch": 1.01989132305574,
      "grad_norm": 2.1276087760925293,
      "learning_rate": 1.6081150896157058e-05,
      "loss": 0.2456,
      "step": 10370
    },
    {
      "epoch": 1.0208748248137494,
      "grad_norm": 1.0101721286773682,
      "learning_rate": 1.6077176807216947e-05,
      "loss": 0.2188,
      "step": 10380
    },
    {
      "epoch": 1.0218583265717587,
      "grad_norm": 0.9049150347709656,
      "learning_rate": 1.6073202718276837e-05,
      "loss": 0.1653,
      "step": 10390
    },
    {
      "epoch": 1.0228418283297682,
      "grad_norm": 2.162771224975586,
      "learning_rate": 1.6069228629336727e-05,
      "loss": 0.2223,
      "step": 10400
    },
    {
      "epoch": 1.0238253300877775,
      "grad_norm": 1.646711826324463,
      "learning_rate": 1.6065254540396616e-05,
      "loss": 0.2035,
      "step": 10410
    },
    {
      "epoch": 1.0248088318457869,
      "grad_norm": 0.6415451169013977,
      "learning_rate": 1.6061280451456506e-05,
      "loss": 0.3438,
      "step": 10420
    },
    {
      "epoch": 1.0257923336037964,
      "grad_norm": 0.9765540361404419,
      "learning_rate": 1.6057306362516392e-05,
      "loss": 0.2076,
      "step": 10430
    },
    {
      "epoch": 1.0267758353618057,
      "grad_norm": 0.8718576431274414,
      "learning_rate": 1.6053332273576282e-05,
      "loss": 0.2563,
      "step": 10440
    },
    {
      "epoch": 1.027759337119815,
      "grad_norm": 3.8872036933898926,
      "learning_rate": 1.6049358184636175e-05,
      "loss": 0.2637,
      "step": 10450
    },
    {
      "epoch": 1.0287428388778246,
      "grad_norm": 0.34129858016967773,
      "learning_rate": 1.6045384095696065e-05,
      "loss": 0.2198,
      "step": 10460
    },
    {
      "epoch": 1.029726340635834,
      "grad_norm": 0.7682720422744751,
      "learning_rate": 1.6041410006755954e-05,
      "loss": 0.3239,
      "step": 10470
    },
    {
      "epoch": 1.0307098423938432,
      "grad_norm": 2.776768445968628,
      "learning_rate": 1.603743591781584e-05,
      "loss": 0.1555,
      "step": 10480
    },
    {
      "epoch": 1.0316933441518528,
      "grad_norm": 1.3365027904510498,
      "learning_rate": 1.603346182887573e-05,
      "loss": 0.2112,
      "step": 10490
    },
    {
      "epoch": 1.032676845909862,
      "grad_norm": 1.3282992839813232,
      "learning_rate": 1.602948773993562e-05,
      "loss": 0.195,
      "step": 10500
    },
    {
      "epoch": 1.032676845909862,
      "eval_loss": 0.1750352829694748,
      "eval_runtime": 16.9966,
      "eval_samples_per_second": 2.942,
      "eval_steps_per_second": 1.471,
      "step": 10500
    },
    {
      "epoch": 1.0336603476678714,
      "grad_norm": 1.3822355270385742,
      "learning_rate": 1.602551365099551e-05,
      "loss": 0.1851,
      "step": 10510
    },
    {
      "epoch": 1.034643849425881,
      "grad_norm": 1.0904524326324463,
      "learning_rate": 1.60215395620554e-05,
      "loss": 0.1616,
      "step": 10520
    },
    {
      "epoch": 1.0356273511838903,
      "grad_norm": 0.7074946165084839,
      "learning_rate": 1.601756547311529e-05,
      "loss": 0.2985,
      "step": 10530
    },
    {
      "epoch": 1.0366108529418996,
      "grad_norm": 0.7284005284309387,
      "learning_rate": 1.601359138417518e-05,
      "loss": 0.3503,
      "step": 10540
    },
    {
      "epoch": 1.0375943546999091,
      "grad_norm": 0.3212083578109741,
      "learning_rate": 1.600961729523507e-05,
      "loss": 0.2078,
      "step": 10550
    },
    {
      "epoch": 1.0385778564579184,
      "grad_norm": 1.210832953453064,
      "learning_rate": 1.6005643206294958e-05,
      "loss": 0.242,
      "step": 10560
    },
    {
      "epoch": 1.0395613582159278,
      "grad_norm": 1.1374540328979492,
      "learning_rate": 1.6001669117354848e-05,
      "loss": 0.2138,
      "step": 10570
    },
    {
      "epoch": 1.0405448599739373,
      "grad_norm": 0.281815767288208,
      "learning_rate": 1.5997695028414737e-05,
      "loss": 0.2453,
      "step": 10580
    },
    {
      "epoch": 1.0415283617319466,
      "grad_norm": 0.8695719242095947,
      "learning_rate": 1.5993720939474627e-05,
      "loss": 0.1835,
      "step": 10590
    },
    {
      "epoch": 1.042511863489956,
      "grad_norm": 1.4869102239608765,
      "learning_rate": 1.5989746850534517e-05,
      "loss": 0.2998,
      "step": 10600
    },
    {
      "epoch": 1.0434953652479653,
      "grad_norm": 2.7580795288085938,
      "learning_rate": 1.5985772761594406e-05,
      "loss": 0.3063,
      "step": 10610
    },
    {
      "epoch": 1.0444788670059748,
      "grad_norm": 0.5467950701713562,
      "learning_rate": 1.5981798672654296e-05,
      "loss": 0.2473,
      "step": 10620
    },
    {
      "epoch": 1.0454623687639841,
      "grad_norm": 1.4874897003173828,
      "learning_rate": 1.5977824583714186e-05,
      "loss": 0.2189,
      "step": 10630
    },
    {
      "epoch": 1.0464458705219934,
      "grad_norm": 1.1254550218582153,
      "learning_rate": 1.5973850494774075e-05,
      "loss": 0.2375,
      "step": 10640
    },
    {
      "epoch": 1.047429372280003,
      "grad_norm": 1.4423521757125854,
      "learning_rate": 1.5969876405833965e-05,
      "loss": 0.1031,
      "step": 10650
    },
    {
      "epoch": 1.0484128740380123,
      "grad_norm": 0.7423785328865051,
      "learning_rate": 1.5965902316893855e-05,
      "loss": 0.2293,
      "step": 10660
    },
    {
      "epoch": 1.0493963757960216,
      "grad_norm": 0.626494824886322,
      "learning_rate": 1.596192822795374e-05,
      "loss": 0.1934,
      "step": 10670
    },
    {
      "epoch": 1.0503798775540312,
      "grad_norm": 1.428344488143921,
      "learning_rate": 1.595795413901363e-05,
      "loss": 0.2119,
      "step": 10680
    },
    {
      "epoch": 1.0513633793120405,
      "grad_norm": 0.5278069376945496,
      "learning_rate": 1.595398005007352e-05,
      "loss": 0.3225,
      "step": 10690
    },
    {
      "epoch": 1.0523468810700498,
      "grad_norm": 0.3313775956630707,
      "learning_rate": 1.5950005961133413e-05,
      "loss": 0.1946,
      "step": 10700
    },
    {
      "epoch": 1.0533303828280594,
      "grad_norm": 1.9315088987350464,
      "learning_rate": 1.5946031872193303e-05,
      "loss": 0.2707,
      "step": 10710
    },
    {
      "epoch": 1.0543138845860687,
      "grad_norm": 0.6008973121643066,
      "learning_rate": 1.594205778325319e-05,
      "loss": 0.3136,
      "step": 10720
    },
    {
      "epoch": 1.055297386344078,
      "grad_norm": 0.7295891046524048,
      "learning_rate": 1.593808369431308e-05,
      "loss": 0.2464,
      "step": 10730
    },
    {
      "epoch": 1.0562808881020875,
      "grad_norm": 1.5947253704071045,
      "learning_rate": 1.593410960537297e-05,
      "loss": 0.1516,
      "step": 10740
    },
    {
      "epoch": 1.0572643898600969,
      "grad_norm": 1.8050662279129028,
      "learning_rate": 1.5930135516432858e-05,
      "loss": 0.2702,
      "step": 10750
    },
    {
      "epoch": 1.0582478916181062,
      "grad_norm": 1.135501742362976,
      "learning_rate": 1.5926161427492748e-05,
      "loss": 0.2593,
      "step": 10760
    },
    {
      "epoch": 1.0592313933761157,
      "grad_norm": 1.1788592338562012,
      "learning_rate": 1.5922187338552638e-05,
      "loss": 0.2392,
      "step": 10770
    },
    {
      "epoch": 1.060214895134125,
      "grad_norm": 1.9654382467269897,
      "learning_rate": 1.5918213249612527e-05,
      "loss": 0.2876,
      "step": 10780
    },
    {
      "epoch": 1.0611983968921344,
      "grad_norm": 0.15102195739746094,
      "learning_rate": 1.5914239160672417e-05,
      "loss": 0.244,
      "step": 10790
    },
    {
      "epoch": 1.062181898650144,
      "grad_norm": 0.8354254961013794,
      "learning_rate": 1.5910265071732307e-05,
      "loss": 0.2165,
      "step": 10800
    },
    {
      "epoch": 1.0631654004081532,
      "grad_norm": 0.7330989837646484,
      "learning_rate": 1.5906290982792196e-05,
      "loss": 0.1486,
      "step": 10810
    },
    {
      "epoch": 1.0641489021661625,
      "grad_norm": 2.2787623405456543,
      "learning_rate": 1.5902316893852086e-05,
      "loss": 0.2067,
      "step": 10820
    },
    {
      "epoch": 1.065132403924172,
      "grad_norm": 0.8515722155570984,
      "learning_rate": 1.5898342804911976e-05,
      "loss": 0.2311,
      "step": 10830
    },
    {
      "epoch": 1.0661159056821814,
      "grad_norm": 2.1706154346466064,
      "learning_rate": 1.5894368715971865e-05,
      "loss": 0.1841,
      "step": 10840
    },
    {
      "epoch": 1.0670994074401907,
      "grad_norm": 3.861100673675537,
      "learning_rate": 1.5890394627031755e-05,
      "loss": 0.1702,
      "step": 10850
    },
    {
      "epoch": 1.0680829091982003,
      "grad_norm": 0.5234455466270447,
      "learning_rate": 1.5886420538091645e-05,
      "loss": 0.1791,
      "step": 10860
    },
    {
      "epoch": 1.0690664109562096,
      "grad_norm": 0.9792625904083252,
      "learning_rate": 1.5882446449151534e-05,
      "loss": 0.2743,
      "step": 10870
    },
    {
      "epoch": 1.070049912714219,
      "grad_norm": 1.8640176057815552,
      "learning_rate": 1.5878472360211424e-05,
      "loss": 0.2103,
      "step": 10880
    },
    {
      "epoch": 1.0710334144722284,
      "grad_norm": 2.448150396347046,
      "learning_rate": 1.5874498271271314e-05,
      "loss": 0.3244,
      "step": 10890
    },
    {
      "epoch": 1.0720169162302378,
      "grad_norm": 1.2162164449691772,
      "learning_rate": 1.5870524182331203e-05,
      "loss": 0.3108,
      "step": 10900
    },
    {
      "epoch": 1.073000417988247,
      "grad_norm": 0.895965039730072,
      "learning_rate": 1.586655009339109e-05,
      "loss": 0.2238,
      "step": 10910
    },
    {
      "epoch": 1.0739839197462566,
      "grad_norm": 3.080109119415283,
      "learning_rate": 1.586257600445098e-05,
      "loss": 0.2297,
      "step": 10920
    },
    {
      "epoch": 1.074967421504266,
      "grad_norm": 0.6591482162475586,
      "learning_rate": 1.585860191551087e-05,
      "loss": 0.2226,
      "step": 10930
    },
    {
      "epoch": 1.0759509232622753,
      "grad_norm": 1.4663394689559937,
      "learning_rate": 1.5854627826570762e-05,
      "loss": 0.2442,
      "step": 10940
    },
    {
      "epoch": 1.0769344250202848,
      "grad_norm": 1.6239320039749146,
      "learning_rate": 1.585065373763065e-05,
      "loss": 0.2115,
      "step": 10950
    },
    {
      "epoch": 1.0779179267782941,
      "grad_norm": 0.2651664912700653,
      "learning_rate": 1.5846679648690538e-05,
      "loss": 0.2726,
      "step": 10960
    },
    {
      "epoch": 1.0789014285363034,
      "grad_norm": 0.9693211913108826,
      "learning_rate": 1.5842705559750427e-05,
      "loss": 0.157,
      "step": 10970
    },
    {
      "epoch": 1.079884930294313,
      "grad_norm": 1.8968473672866821,
      "learning_rate": 1.5838731470810317e-05,
      "loss": 0.2652,
      "step": 10980
    },
    {
      "epoch": 1.0808684320523223,
      "grad_norm": 1.7201496362686157,
      "learning_rate": 1.5834757381870207e-05,
      "loss": 0.2528,
      "step": 10990
    },
    {
      "epoch": 1.0818519338103316,
      "grad_norm": 0.6101740598678589,
      "learning_rate": 1.5830783292930096e-05,
      "loss": 0.275,
      "step": 11000
    },
    {
      "epoch": 1.0818519338103316,
      "eval_loss": 0.1684889793395996,
      "eval_runtime": 20.0453,
      "eval_samples_per_second": 2.494,
      "eval_steps_per_second": 1.247,
      "step": 11000
    },
    {
      "epoch": 1.0828354355683412,
      "grad_norm": 1.1838539838790894,
      "learning_rate": 1.5826809203989986e-05,
      "loss": 0.1932,
      "step": 11010
    },
    {
      "epoch": 1.0838189373263505,
      "grad_norm": 1.040086030960083,
      "learning_rate": 1.5822835115049876e-05,
      "loss": 0.2281,
      "step": 11020
    },
    {
      "epoch": 1.0848024390843598,
      "grad_norm": 2.302246332168579,
      "learning_rate": 1.5818861026109765e-05,
      "loss": 0.1934,
      "step": 11030
    },
    {
      "epoch": 1.0857859408423693,
      "grad_norm": 1.5548522472381592,
      "learning_rate": 1.5814886937169655e-05,
      "loss": 0.211,
      "step": 11040
    },
    {
      "epoch": 1.0867694426003787,
      "grad_norm": 1.5086666345596313,
      "learning_rate": 1.5810912848229545e-05,
      "loss": 0.2239,
      "step": 11050
    },
    {
      "epoch": 1.087752944358388,
      "grad_norm": 0.8068061470985413,
      "learning_rate": 1.5806938759289434e-05,
      "loss": 0.2373,
      "step": 11060
    },
    {
      "epoch": 1.0887364461163975,
      "grad_norm": 1.569872260093689,
      "learning_rate": 1.5802964670349324e-05,
      "loss": 0.2313,
      "step": 11070
    },
    {
      "epoch": 1.0897199478744068,
      "grad_norm": 1.8688373565673828,
      "learning_rate": 1.5798990581409214e-05,
      "loss": 0.2964,
      "step": 11080
    },
    {
      "epoch": 1.0907034496324162,
      "grad_norm": 1.7373290061950684,
      "learning_rate": 1.5795016492469103e-05,
      "loss": 0.2808,
      "step": 11090
    },
    {
      "epoch": 1.0916869513904257,
      "grad_norm": 1.3782762289047241,
      "learning_rate": 1.5791042403528993e-05,
      "loss": 0.2017,
      "step": 11100
    },
    {
      "epoch": 1.092670453148435,
      "grad_norm": 0.6687967777252197,
      "learning_rate": 1.5787068314588883e-05,
      "loss": 0.1564,
      "step": 11110
    },
    {
      "epoch": 1.0936539549064443,
      "grad_norm": 1.177499532699585,
      "learning_rate": 1.5783094225648772e-05,
      "loss": 0.2055,
      "step": 11120
    },
    {
      "epoch": 1.094637456664454,
      "grad_norm": 1.7758307456970215,
      "learning_rate": 1.5779120136708662e-05,
      "loss": 0.308,
      "step": 11130
    },
    {
      "epoch": 1.0956209584224632,
      "grad_norm": 0.8646954894065857,
      "learning_rate": 1.5775146047768552e-05,
      "loss": 0.2851,
      "step": 11140
    },
    {
      "epoch": 1.0966044601804725,
      "grad_norm": 1.7536402940750122,
      "learning_rate": 1.5771171958828438e-05,
      "loss": 0.2581,
      "step": 11150
    },
    {
      "epoch": 1.097587961938482,
      "grad_norm": 0.7464534640312195,
      "learning_rate": 1.5767197869888328e-05,
      "loss": 0.2356,
      "step": 11160
    },
    {
      "epoch": 1.0985714636964914,
      "grad_norm": 0.9927958846092224,
      "learning_rate": 1.5763223780948217e-05,
      "loss": 0.2994,
      "step": 11170
    },
    {
      "epoch": 1.0995549654545007,
      "grad_norm": 1.3818191289901733,
      "learning_rate": 1.575924969200811e-05,
      "loss": 0.2317,
      "step": 11180
    },
    {
      "epoch": 1.1005384672125103,
      "grad_norm": 1.2731701135635376,
      "learning_rate": 1.5755275603068e-05,
      "loss": 0.1586,
      "step": 11190
    },
    {
      "epoch": 1.1015219689705196,
      "grad_norm": 0.4445424973964691,
      "learning_rate": 1.5751301514127886e-05,
      "loss": 0.2629,
      "step": 11200
    },
    {
      "epoch": 1.102505470728529,
      "grad_norm": 0.8182106614112854,
      "learning_rate": 1.5747327425187776e-05,
      "loss": 0.2465,
      "step": 11210
    },
    {
      "epoch": 1.1034889724865384,
      "grad_norm": 0.32335585355758667,
      "learning_rate": 1.5743353336247666e-05,
      "loss": 0.175,
      "step": 11220
    },
    {
      "epoch": 1.1044724742445478,
      "grad_norm": 1.8322887420654297,
      "learning_rate": 1.5739379247307555e-05,
      "loss": 0.2475,
      "step": 11230
    },
    {
      "epoch": 1.105455976002557,
      "grad_norm": 0.381082683801651,
      "learning_rate": 1.5735405158367445e-05,
      "loss": 0.2041,
      "step": 11240
    },
    {
      "epoch": 1.1064394777605666,
      "grad_norm": 0.8768240809440613,
      "learning_rate": 1.5731431069427335e-05,
      "loss": 0.2022,
      "step": 11250
    },
    {
      "epoch": 1.107422979518576,
      "grad_norm": 1.1599154472351074,
      "learning_rate": 1.5727456980487224e-05,
      "loss": 0.2321,
      "step": 11260
    },
    {
      "epoch": 1.1084064812765853,
      "grad_norm": 0.9594438076019287,
      "learning_rate": 1.5723482891547114e-05,
      "loss": 0.1651,
      "step": 11270
    },
    {
      "epoch": 1.1093899830345946,
      "grad_norm": 0.7346277236938477,
      "learning_rate": 1.5719508802607004e-05,
      "loss": 0.209,
      "step": 11280
    },
    {
      "epoch": 1.1103734847926041,
      "grad_norm": 1.4004619121551514,
      "learning_rate": 1.5715534713666893e-05,
      "loss": 0.237,
      "step": 11290
    },
    {
      "epoch": 1.1113569865506134,
      "grad_norm": 1.0641740560531616,
      "learning_rate": 1.5711560624726783e-05,
      "loss": 0.2332,
      "step": 11300
    },
    {
      "epoch": 1.1123404883086228,
      "grad_norm": 0.6625562906265259,
      "learning_rate": 1.5707586535786673e-05,
      "loss": 0.1923,
      "step": 11310
    },
    {
      "epoch": 1.1133239900666323,
      "grad_norm": 2.6541385650634766,
      "learning_rate": 1.5703612446846562e-05,
      "loss": 0.2174,
      "step": 11320
    },
    {
      "epoch": 1.1143074918246416,
      "grad_norm": 0.5809294581413269,
      "learning_rate": 1.5699638357906452e-05,
      "loss": 0.254,
      "step": 11330
    },
    {
      "epoch": 1.115290993582651,
      "grad_norm": 0.5939483642578125,
      "learning_rate": 1.5695664268966338e-05,
      "loss": 0.1861,
      "step": 11340
    },
    {
      "epoch": 1.1162744953406605,
      "grad_norm": 0.6164649128913879,
      "learning_rate": 1.569169018002623e-05,
      "loss": 0.2705,
      "step": 11350
    },
    {
      "epoch": 1.1172579970986698,
      "grad_norm": 0.39910444617271423,
      "learning_rate": 1.568771609108612e-05,
      "loss": 0.258,
      "step": 11360
    },
    {
      "epoch": 1.1182414988566791,
      "grad_norm": 0.707711935043335,
      "learning_rate": 1.568374200214601e-05,
      "loss": 0.1839,
      "step": 11370
    },
    {
      "epoch": 1.1192250006146887,
      "grad_norm": 0.9458222985267639,
      "learning_rate": 1.56797679132059e-05,
      "loss": 0.3005,
      "step": 11380
    },
    {
      "epoch": 1.120208502372698,
      "grad_norm": 1.1831637620925903,
      "learning_rate": 1.5675793824265787e-05,
      "loss": 0.3427,
      "step": 11390
    },
    {
      "epoch": 1.1211920041307073,
      "grad_norm": 2.5058209896087646,
      "learning_rate": 1.5671819735325676e-05,
      "loss": 0.2011,
      "step": 11400
    },
    {
      "epoch": 1.1221755058887168,
      "grad_norm": 0.7695888876914978,
      "learning_rate": 1.5667845646385566e-05,
      "loss": 0.1912,
      "step": 11410
    },
    {
      "epoch": 1.1231590076467262,
      "grad_norm": 0.6722692251205444,
      "learning_rate": 1.5663871557445456e-05,
      "loss": 0.2892,
      "step": 11420
    },
    {
      "epoch": 1.1241425094047355,
      "grad_norm": 0.46231839060783386,
      "learning_rate": 1.565989746850535e-05,
      "loss": 0.1979,
      "step": 11430
    },
    {
      "epoch": 1.125126011162745,
      "grad_norm": 1.6147507429122925,
      "learning_rate": 1.5655923379565235e-05,
      "loss": 0.2097,
      "step": 11440
    },
    {
      "epoch": 1.1261095129207543,
      "grad_norm": 0.5903246402740479,
      "learning_rate": 1.5651949290625125e-05,
      "loss": 0.2322,
      "step": 11450
    },
    {
      "epoch": 1.1270930146787637,
      "grad_norm": 0.9642149209976196,
      "learning_rate": 1.5647975201685014e-05,
      "loss": 0.2188,
      "step": 11460
    },
    {
      "epoch": 1.1280765164367732,
      "grad_norm": 0.9721670746803284,
      "learning_rate": 1.5644001112744904e-05,
      "loss": 0.3213,
      "step": 11470
    },
    {
      "epoch": 1.1290600181947825,
      "grad_norm": 0.5057275891304016,
      "learning_rate": 1.5640027023804794e-05,
      "loss": 0.1826,
      "step": 11480
    },
    {
      "epoch": 1.1300435199527918,
      "grad_norm": 0.7329443097114563,
      "learning_rate": 1.5636052934864683e-05,
      "loss": 0.2357,
      "step": 11490
    },
    {
      "epoch": 1.1310270217108014,
      "grad_norm": 0.41778743267059326,
      "learning_rate": 1.5632078845924573e-05,
      "loss": 0.242,
      "step": 11500
    },
    {
      "epoch": 1.1310270217108014,
      "eval_loss": 0.16718797385692596,
      "eval_runtime": 20.7187,
      "eval_samples_per_second": 2.413,
      "eval_steps_per_second": 1.207,
      "step": 11500
    },
    {
      "epoch": 1.1320105234688107,
      "grad_norm": 1.1029562950134277,
      "learning_rate": 1.5628104756984463e-05,
      "loss": 0.1791,
      "step": 11510
    },
    {
      "epoch": 1.13299402522682,
      "grad_norm": 1.1086807250976562,
      "learning_rate": 1.5624130668044352e-05,
      "loss": 0.2324,
      "step": 11520
    },
    {
      "epoch": 1.1339775269848296,
      "grad_norm": 1.1632509231567383,
      "learning_rate": 1.5620156579104242e-05,
      "loss": 0.2047,
      "step": 11530
    },
    {
      "epoch": 1.1349610287428389,
      "grad_norm": 1.6066392660140991,
      "learning_rate": 1.561618249016413e-05,
      "loss": 0.1266,
      "step": 11540
    },
    {
      "epoch": 1.1359445305008482,
      "grad_norm": 1.0749987363815308,
      "learning_rate": 1.561220840122402e-05,
      "loss": 0.2094,
      "step": 11550
    },
    {
      "epoch": 1.1369280322588577,
      "grad_norm": 0.9814296364784241,
      "learning_rate": 1.560823431228391e-05,
      "loss": 0.2656,
      "step": 11560
    },
    {
      "epoch": 1.137911534016867,
      "grad_norm": 2.2510077953338623,
      "learning_rate": 1.56042602233438e-05,
      "loss": 0.3081,
      "step": 11570
    },
    {
      "epoch": 1.1388950357748764,
      "grad_norm": 0.728635847568512,
      "learning_rate": 1.5600286134403687e-05,
      "loss": 0.3033,
      "step": 11580
    },
    {
      "epoch": 1.139878537532886,
      "grad_norm": 1.4196513891220093,
      "learning_rate": 1.559631204546358e-05,
      "loss": 0.2354,
      "step": 11590
    },
    {
      "epoch": 1.1408620392908952,
      "grad_norm": 2.0953516960144043,
      "learning_rate": 1.559233795652347e-05,
      "loss": 0.2404,
      "step": 11600
    },
    {
      "epoch": 1.1418455410489046,
      "grad_norm": 1.5039966106414795,
      "learning_rate": 1.558836386758336e-05,
      "loss": 0.1953,
      "step": 11610
    },
    {
      "epoch": 1.142829042806914,
      "grad_norm": 1.7572118043899536,
      "learning_rate": 1.558438977864325e-05,
      "loss": 0.2377,
      "step": 11620
    },
    {
      "epoch": 1.1438125445649234,
      "grad_norm": 6.402324199676514,
      "learning_rate": 1.5580415689703135e-05,
      "loss": 0.3005,
      "step": 11630
    },
    {
      "epoch": 1.1447960463229327,
      "grad_norm": 1.5109554529190063,
      "learning_rate": 1.5576441600763025e-05,
      "loss": 0.1312,
      "step": 11640
    },
    {
      "epoch": 1.1457795480809423,
      "grad_norm": 0.4686489999294281,
      "learning_rate": 1.5572467511822915e-05,
      "loss": 0.2823,
      "step": 11650
    },
    {
      "epoch": 1.1467630498389516,
      "grad_norm": 3.5495054721832275,
      "learning_rate": 1.5568493422882804e-05,
      "loss": 0.19,
      "step": 11660
    },
    {
      "epoch": 1.147746551596961,
      "grad_norm": 1.636431336402893,
      "learning_rate": 1.5564519333942697e-05,
      "loss": 0.1589,
      "step": 11670
    },
    {
      "epoch": 1.1487300533549702,
      "grad_norm": 0.634340763092041,
      "learning_rate": 1.5560545245002583e-05,
      "loss": 0.1228,
      "step": 11680
    },
    {
      "epoch": 1.1497135551129798,
      "grad_norm": 2.047029972076416,
      "learning_rate": 1.5556571156062473e-05,
      "loss": 0.1921,
      "step": 11690
    },
    {
      "epoch": 1.150697056870989,
      "grad_norm": 1.4832496643066406,
      "learning_rate": 1.5552597067122363e-05,
      "loss": 0.2528,
      "step": 11700
    },
    {
      "epoch": 1.1516805586289984,
      "grad_norm": 1.1205974817276,
      "learning_rate": 1.5548622978182252e-05,
      "loss": 0.1315,
      "step": 11710
    },
    {
      "epoch": 1.152664060387008,
      "grad_norm": 2.786681890487671,
      "learning_rate": 1.5544648889242142e-05,
      "loss": 0.3437,
      "step": 11720
    },
    {
      "epoch": 1.1536475621450173,
      "grad_norm": 2.386262893676758,
      "learning_rate": 1.5540674800302032e-05,
      "loss": 0.194,
      "step": 11730
    },
    {
      "epoch": 1.1546310639030266,
      "grad_norm": 1.2636781930923462,
      "learning_rate": 1.553670071136192e-05,
      "loss": 0.2482,
      "step": 11740
    },
    {
      "epoch": 1.1556145656610362,
      "grad_norm": 1.3071417808532715,
      "learning_rate": 1.553272662242181e-05,
      "loss": 0.2084,
      "step": 11750
    },
    {
      "epoch": 1.1565980674190455,
      "grad_norm": 2.5322470664978027,
      "learning_rate": 1.55287525334817e-05,
      "loss": 0.2909,
      "step": 11760
    },
    {
      "epoch": 1.1575815691770548,
      "grad_norm": 0.9364299178123474,
      "learning_rate": 1.552477844454159e-05,
      "loss": 0.2372,
      "step": 11770
    },
    {
      "epoch": 1.1585650709350643,
      "grad_norm": 1.4364525079727173,
      "learning_rate": 1.552080435560148e-05,
      "loss": 0.3222,
      "step": 11780
    },
    {
      "epoch": 1.1595485726930737,
      "grad_norm": 0.7772247195243835,
      "learning_rate": 1.551683026666137e-05,
      "loss": 0.227,
      "step": 11790
    },
    {
      "epoch": 1.160532074451083,
      "grad_norm": 3.083230972290039,
      "learning_rate": 1.551285617772126e-05,
      "loss": 0.1622,
      "step": 11800
    },
    {
      "epoch": 1.1615155762090925,
      "grad_norm": 0.6908173561096191,
      "learning_rate": 1.550888208878115e-05,
      "loss": 0.2731,
      "step": 11810
    },
    {
      "epoch": 1.1624990779671018,
      "grad_norm": 1.6049355268478394,
      "learning_rate": 1.5504907999841035e-05,
      "loss": 0.2115,
      "step": 11820
    },
    {
      "epoch": 1.1634825797251112,
      "grad_norm": 1.3253471851348877,
      "learning_rate": 1.550093391090093e-05,
      "loss": 0.2486,
      "step": 11830
    },
    {
      "epoch": 1.1644660814831207,
      "grad_norm": 1.939974308013916,
      "learning_rate": 1.5496959821960818e-05,
      "loss": 0.2514,
      "step": 11840
    },
    {
      "epoch": 1.16544958324113,
      "grad_norm": 0.3961417078971863,
      "learning_rate": 1.5492985733020708e-05,
      "loss": 0.1514,
      "step": 11850
    },
    {
      "epoch": 1.1664330849991393,
      "grad_norm": 2.070683002471924,
      "learning_rate": 1.5489011644080597e-05,
      "loss": 0.236,
      "step": 11860
    },
    {
      "epoch": 1.1674165867571489,
      "grad_norm": 2.060924768447876,
      "learning_rate": 1.5485037555140484e-05,
      "loss": 0.2529,
      "step": 11870
    },
    {
      "epoch": 1.1684000885151582,
      "grad_norm": 0.6187996864318848,
      "learning_rate": 1.5481063466200373e-05,
      "loss": 0.4012,
      "step": 11880
    },
    {
      "epoch": 1.1693835902731675,
      "grad_norm": 2.7358853816986084,
      "learning_rate": 1.5477089377260263e-05,
      "loss": 0.299,
      "step": 11890
    },
    {
      "epoch": 1.170367092031177,
      "grad_norm": 1.793433666229248,
      "learning_rate": 1.5473115288320153e-05,
      "loss": 0.3304,
      "step": 11900
    },
    {
      "epoch": 1.1713505937891864,
      "grad_norm": 1.451925277709961,
      "learning_rate": 1.5469141199380046e-05,
      "loss": 0.2606,
      "step": 11910
    },
    {
      "epoch": 1.1723340955471957,
      "grad_norm": 0.632253885269165,
      "learning_rate": 1.5465167110439932e-05,
      "loss": 0.2446,
      "step": 11920
    },
    {
      "epoch": 1.1733175973052052,
      "grad_norm": 1.721764326095581,
      "learning_rate": 1.5461193021499822e-05,
      "loss": 0.3533,
      "step": 11930
    },
    {
      "epoch": 1.1743010990632146,
      "grad_norm": 1.3763706684112549,
      "learning_rate": 1.545721893255971e-05,
      "loss": 0.2624,
      "step": 11940
    },
    {
      "epoch": 1.1752846008212239,
      "grad_norm": 1.520142912864685,
      "learning_rate": 1.54532448436196e-05,
      "loss": 0.159,
      "step": 11950
    },
    {
      "epoch": 1.1762681025792334,
      "grad_norm": 2.0177927017211914,
      "learning_rate": 1.544927075467949e-05,
      "loss": 0.2352,
      "step": 11960
    },
    {
      "epoch": 1.1772516043372427,
      "grad_norm": 1.737296223640442,
      "learning_rate": 1.544529666573938e-05,
      "loss": 0.1823,
      "step": 11970
    },
    {
      "epoch": 1.178235106095252,
      "grad_norm": 1.6785165071487427,
      "learning_rate": 1.544132257679927e-05,
      "loss": 0.1911,
      "step": 11980
    },
    {
      "epoch": 1.1792186078532616,
      "grad_norm": 0.7725062370300293,
      "learning_rate": 1.543734848785916e-05,
      "loss": 0.1469,
      "step": 11990
    },
    {
      "epoch": 1.180202109611271,
      "grad_norm": 1.0442862510681152,
      "learning_rate": 1.543337439891905e-05,
      "loss": 0.28,
      "step": 12000
    },
    {
      "epoch": 1.180202109611271,
      "eval_loss": 0.15947729349136353,
      "eval_runtime": 19.5412,
      "eval_samples_per_second": 2.559,
      "eval_steps_per_second": 1.279,
      "step": 12000
    },
    {
      "epoch": 1.1811856113692802,
      "grad_norm": 0.7599958181381226,
      "learning_rate": 1.542940030997894e-05,
      "loss": 0.2466,
      "step": 12010
    },
    {
      "epoch": 1.1821691131272898,
      "grad_norm": 1.1370816230773926,
      "learning_rate": 1.542542622103883e-05,
      "loss": 0.2842,
      "step": 12020
    },
    {
      "epoch": 1.183152614885299,
      "grad_norm": 1.1675643920898438,
      "learning_rate": 1.542145213209872e-05,
      "loss": 0.2127,
      "step": 12030
    },
    {
      "epoch": 1.1841361166433084,
      "grad_norm": 1.1006205081939697,
      "learning_rate": 1.5417478043158608e-05,
      "loss": 0.2202,
      "step": 12040
    },
    {
      "epoch": 1.185119618401318,
      "grad_norm": 0.6677995324134827,
      "learning_rate": 1.5413503954218498e-05,
      "loss": 0.2997,
      "step": 12050
    },
    {
      "epoch": 1.1861031201593273,
      "grad_norm": 1.038744568824768,
      "learning_rate": 1.5409529865278384e-05,
      "loss": 0.2648,
      "step": 12060
    },
    {
      "epoch": 1.1870866219173366,
      "grad_norm": 2.356663227081299,
      "learning_rate": 1.5405555776338277e-05,
      "loss": 0.1663,
      "step": 12070
    },
    {
      "epoch": 1.1880701236753461,
      "grad_norm": 1.1240571737289429,
      "learning_rate": 1.5401581687398167e-05,
      "loss": 0.2276,
      "step": 12080
    },
    {
      "epoch": 1.1890536254333555,
      "grad_norm": 1.8712751865386963,
      "learning_rate": 1.5397607598458056e-05,
      "loss": 0.2572,
      "step": 12090
    },
    {
      "epoch": 1.1900371271913648,
      "grad_norm": 0.5862861275672913,
      "learning_rate": 1.5393633509517946e-05,
      "loss": 0.1785,
      "step": 12100
    },
    {
      "epoch": 1.1910206289493743,
      "grad_norm": 1.5912821292877197,
      "learning_rate": 1.5389659420577832e-05,
      "loss": 0.3814,
      "step": 12110
    },
    {
      "epoch": 1.1920041307073836,
      "grad_norm": 0.7632641196250916,
      "learning_rate": 1.5385685331637722e-05,
      "loss": 0.1927,
      "step": 12120
    },
    {
      "epoch": 1.192987632465393,
      "grad_norm": 0.7380467057228088,
      "learning_rate": 1.538171124269761e-05,
      "loss": 0.2317,
      "step": 12130
    },
    {
      "epoch": 1.1939711342234025,
      "grad_norm": 1.1679322719573975,
      "learning_rate": 1.53777371537575e-05,
      "loss": 0.2264,
      "step": 12140
    },
    {
      "epoch": 1.1949546359814118,
      "grad_norm": 1.4404914379119873,
      "learning_rate": 1.537376306481739e-05,
      "loss": 0.2184,
      "step": 12150
    },
    {
      "epoch": 1.1959381377394211,
      "grad_norm": 0.8603619337081909,
      "learning_rate": 1.536978897587728e-05,
      "loss": 0.2118,
      "step": 12160
    },
    {
      "epoch": 1.1969216394974307,
      "grad_norm": 1.1067661046981812,
      "learning_rate": 1.536581488693717e-05,
      "loss": 0.3043,
      "step": 12170
    },
    {
      "epoch": 1.19790514125544,
      "grad_norm": 0.4780625104904175,
      "learning_rate": 1.536184079799706e-05,
      "loss": 0.1925,
      "step": 12180
    },
    {
      "epoch": 1.1988886430134493,
      "grad_norm": 1.2900044918060303,
      "learning_rate": 1.535786670905695e-05,
      "loss": 0.2877,
      "step": 12190
    },
    {
      "epoch": 1.1998721447714589,
      "grad_norm": 0.5605069994926453,
      "learning_rate": 1.535389262011684e-05,
      "loss": 0.1849,
      "step": 12200
    },
    {
      "epoch": 1.2008556465294682,
      "grad_norm": 2.1221747398376465,
      "learning_rate": 1.534991853117673e-05,
      "loss": 0.2137,
      "step": 12210
    },
    {
      "epoch": 1.2018391482874775,
      "grad_norm": 2.1045353412628174,
      "learning_rate": 1.534594444223662e-05,
      "loss": 0.197,
      "step": 12220
    },
    {
      "epoch": 1.202822650045487,
      "grad_norm": 1.5403351783752441,
      "learning_rate": 1.5341970353296508e-05,
      "loss": 0.158,
      "step": 12230
    },
    {
      "epoch": 1.2038061518034964,
      "grad_norm": 0.4863685369491577,
      "learning_rate": 1.5337996264356398e-05,
      "loss": 0.2061,
      "step": 12240
    },
    {
      "epoch": 1.2047896535615057,
      "grad_norm": 0.5963559150695801,
      "learning_rate": 1.5334022175416288e-05,
      "loss": 0.2193,
      "step": 12250
    },
    {
      "epoch": 1.2057731553195152,
      "grad_norm": 1.3666846752166748,
      "learning_rate": 1.5330048086476177e-05,
      "loss": 0.2201,
      "step": 12260
    },
    {
      "epoch": 1.2067566570775246,
      "grad_norm": 1.747308373451233,
      "learning_rate": 1.5326073997536067e-05,
      "loss": 0.3071,
      "step": 12270
    },
    {
      "epoch": 1.2077401588355339,
      "grad_norm": 0.7220170497894287,
      "learning_rate": 1.5322099908595957e-05,
      "loss": 0.228,
      "step": 12280
    },
    {
      "epoch": 1.2087236605935434,
      "grad_norm": 1.7711100578308105,
      "learning_rate": 1.5318125819655846e-05,
      "loss": 0.2558,
      "step": 12290
    },
    {
      "epoch": 1.2097071623515527,
      "grad_norm": 1.0758968591690063,
      "learning_rate": 1.5314151730715733e-05,
      "loss": 0.2629,
      "step": 12300
    },
    {
      "epoch": 1.210690664109562,
      "grad_norm": 1.715317964553833,
      "learning_rate": 1.5310177641775622e-05,
      "loss": 0.2422,
      "step": 12310
    },
    {
      "epoch": 1.2116741658675716,
      "grad_norm": 1.1539181470870972,
      "learning_rate": 1.5306203552835515e-05,
      "loss": 0.2245,
      "step": 12320
    },
    {
      "epoch": 1.212657667625581,
      "grad_norm": 3.100677251815796,
      "learning_rate": 1.5302229463895405e-05,
      "loss": 0.3347,
      "step": 12330
    },
    {
      "epoch": 1.2136411693835902,
      "grad_norm": 2.0572402477264404,
      "learning_rate": 1.5298255374955295e-05,
      "loss": 0.2516,
      "step": 12340
    },
    {
      "epoch": 1.2146246711415998,
      "grad_norm": 2.0834171772003174,
      "learning_rate": 1.529428128601518e-05,
      "loss": 0.2509,
      "step": 12350
    },
    {
      "epoch": 1.215608172899609,
      "grad_norm": 0.7151486277580261,
      "learning_rate": 1.529030719707507e-05,
      "loss": 0.1851,
      "step": 12360
    },
    {
      "epoch": 1.2165916746576184,
      "grad_norm": 2.3506407737731934,
      "learning_rate": 1.528633310813496e-05,
      "loss": 0.2926,
      "step": 12370
    },
    {
      "epoch": 1.217575176415628,
      "grad_norm": 2.338649272918701,
      "learning_rate": 1.528235901919485e-05,
      "loss": 0.2026,
      "step": 12380
    },
    {
      "epoch": 1.2185586781736373,
      "grad_norm": 0.8904348611831665,
      "learning_rate": 1.527838493025474e-05,
      "loss": 0.188,
      "step": 12390
    },
    {
      "epoch": 1.2195421799316466,
      "grad_norm": 1.264668345451355,
      "learning_rate": 1.527441084131463e-05,
      "loss": 0.2037,
      "step": 12400
    },
    {
      "epoch": 1.2205256816896561,
      "grad_norm": 1.8992657661437988,
      "learning_rate": 1.527043675237452e-05,
      "loss": 0.2614,
      "step": 12410
    },
    {
      "epoch": 1.2215091834476655,
      "grad_norm": 0.7366829514503479,
      "learning_rate": 1.526646266343441e-05,
      "loss": 0.1882,
      "step": 12420
    },
    {
      "epoch": 1.2224926852056748,
      "grad_norm": 1.5639548301696777,
      "learning_rate": 1.5262488574494298e-05,
      "loss": 0.2037,
      "step": 12430
    },
    {
      "epoch": 1.2234761869636843,
      "grad_norm": 0.48022663593292236,
      "learning_rate": 1.5258514485554188e-05,
      "loss": 0.2173,
      "step": 12440
    },
    {
      "epoch": 1.2244596887216936,
      "grad_norm": 3.2655975818634033,
      "learning_rate": 1.5254540396614078e-05,
      "loss": 0.3362,
      "step": 12450
    },
    {
      "epoch": 1.225443190479703,
      "grad_norm": 0.9768887162208557,
      "learning_rate": 1.5250566307673965e-05,
      "loss": 0.2947,
      "step": 12460
    },
    {
      "epoch": 1.2264266922377123,
      "grad_norm": 0.5320988297462463,
      "learning_rate": 1.5246592218733855e-05,
      "loss": 0.2609,
      "step": 12470
    },
    {
      "epoch": 1.2274101939957218,
      "grad_norm": 3.5792882442474365,
      "learning_rate": 1.5242618129793747e-05,
      "loss": 0.1552,
      "step": 12480
    },
    {
      "epoch": 1.2283936957537311,
      "grad_norm": 0.6869518756866455,
      "learning_rate": 1.5238644040853636e-05,
      "loss": 0.1569,
      "step": 12490
    },
    {
      "epoch": 1.2293771975117405,
      "grad_norm": 0.6907517313957214,
      "learning_rate": 1.5234669951913526e-05,
      "loss": 0.3387,
      "step": 12500
    },
    {
      "epoch": 1.2293771975117405,
      "eval_loss": 0.16012313961982727,
      "eval_runtime": 19.5828,
      "eval_samples_per_second": 2.553,
      "eval_steps_per_second": 1.277,
      "step": 12500
    },
    {
      "epoch": 1.23036069926975,
      "grad_norm": 0.5602425932884216,
      "learning_rate": 1.5230695862973416e-05,
      "loss": 0.395,
      "step": 12510
    },
    {
      "epoch": 1.2313442010277593,
      "grad_norm": 1.9788906574249268,
      "learning_rate": 1.5226721774033303e-05,
      "loss": 0.1828,
      "step": 12520
    },
    {
      "epoch": 1.2323277027857686,
      "grad_norm": 1.832395315170288,
      "learning_rate": 1.5222747685093193e-05,
      "loss": 0.1319,
      "step": 12530
    },
    {
      "epoch": 1.2333112045437782,
      "grad_norm": 0.36584335565567017,
      "learning_rate": 1.5218773596153083e-05,
      "loss": 0.2585,
      "step": 12540
    },
    {
      "epoch": 1.2342947063017875,
      "grad_norm": 1.3098914623260498,
      "learning_rate": 1.5214799507212972e-05,
      "loss": 0.2438,
      "step": 12550
    },
    {
      "epoch": 1.2352782080597968,
      "grad_norm": 4.020171165466309,
      "learning_rate": 1.5210825418272864e-05,
      "loss": 0.2908,
      "step": 12560
    },
    {
      "epoch": 1.2362617098178064,
      "grad_norm": 0.9674311280250549,
      "learning_rate": 1.5206851329332752e-05,
      "loss": 0.2159,
      "step": 12570
    },
    {
      "epoch": 1.2372452115758157,
      "grad_norm": 2.408463478088379,
      "learning_rate": 1.5202877240392641e-05,
      "loss": 0.1815,
      "step": 12580
    },
    {
      "epoch": 1.238228713333825,
      "grad_norm": 0.9522561430931091,
      "learning_rate": 1.5198903151452531e-05,
      "loss": 0.2076,
      "step": 12590
    },
    {
      "epoch": 1.2392122150918345,
      "grad_norm": 0.6239114999771118,
      "learning_rate": 1.519492906251242e-05,
      "loss": 0.208,
      "step": 12600
    },
    {
      "epoch": 1.2401957168498439,
      "grad_norm": 1.2748032808303833,
      "learning_rate": 1.5190954973572309e-05,
      "loss": 0.2277,
      "step": 12610
    },
    {
      "epoch": 1.2411792186078532,
      "grad_norm": 2.681368589401245,
      "learning_rate": 1.5186980884632198e-05,
      "loss": 0.2248,
      "step": 12620
    },
    {
      "epoch": 1.2421627203658627,
      "grad_norm": 0.48817378282546997,
      "learning_rate": 1.5183006795692088e-05,
      "loss": 0.2876,
      "step": 12630
    },
    {
      "epoch": 1.243146222123872,
      "grad_norm": 3.0966594219207764,
      "learning_rate": 1.517903270675198e-05,
      "loss": 0.2638,
      "step": 12640
    },
    {
      "epoch": 1.2441297238818814,
      "grad_norm": 2.382673501968384,
      "learning_rate": 1.5175058617811869e-05,
      "loss": 0.2204,
      "step": 12650
    },
    {
      "epoch": 1.245113225639891,
      "grad_norm": 3.4702672958374023,
      "learning_rate": 1.5171084528871757e-05,
      "loss": 0.2438,
      "step": 12660
    },
    {
      "epoch": 1.2460967273979002,
      "grad_norm": 1.1455655097961426,
      "learning_rate": 1.5167110439931647e-05,
      "loss": 0.323,
      "step": 12670
    },
    {
      "epoch": 1.2470802291559095,
      "grad_norm": 1.0415674448013306,
      "learning_rate": 1.5163136350991536e-05,
      "loss": 0.1862,
      "step": 12680
    },
    {
      "epoch": 1.248063730913919,
      "grad_norm": 0.9128134250640869,
      "learning_rate": 1.5159162262051426e-05,
      "loss": 0.1964,
      "step": 12690
    },
    {
      "epoch": 1.2490472326719284,
      "grad_norm": 1.5137019157409668,
      "learning_rate": 1.5155188173111314e-05,
      "loss": 0.2638,
      "step": 12700
    },
    {
      "epoch": 1.2500307344299377,
      "grad_norm": 2.8884811401367188,
      "learning_rate": 1.5151214084171204e-05,
      "loss": 0.1713,
      "step": 12710
    },
    {
      "epoch": 1.251014236187947,
      "grad_norm": 1.3755849599838257,
      "learning_rate": 1.5147239995231095e-05,
      "loss": 0.2103,
      "step": 12720
    },
    {
      "epoch": 1.2519977379459566,
      "grad_norm": 0.6168149709701538,
      "learning_rate": 1.5143265906290985e-05,
      "loss": 0.3774,
      "step": 12730
    },
    {
      "epoch": 1.252981239703966,
      "grad_norm": 0.5615456700325012,
      "learning_rate": 1.5139291817350874e-05,
      "loss": 0.1406,
      "step": 12740
    },
    {
      "epoch": 1.2539647414619752,
      "grad_norm": 1.0317662954330444,
      "learning_rate": 1.5135317728410764e-05,
      "loss": 0.352,
      "step": 12750
    },
    {
      "epoch": 1.2549482432199848,
      "grad_norm": 1.1228874921798706,
      "learning_rate": 1.5131343639470652e-05,
      "loss": 0.1881,
      "step": 12760
    },
    {
      "epoch": 1.255931744977994,
      "grad_norm": 0.7314227223396301,
      "learning_rate": 1.5127369550530542e-05,
      "loss": 0.3255,
      "step": 12770
    },
    {
      "epoch": 1.2569152467360034,
      "grad_norm": 3.771770715713501,
      "learning_rate": 1.5123395461590431e-05,
      "loss": 0.1557,
      "step": 12780
    },
    {
      "epoch": 1.257898748494013,
      "grad_norm": 4.539660453796387,
      "learning_rate": 1.5119421372650321e-05,
      "loss": 0.1951,
      "step": 12790
    },
    {
      "epoch": 1.2588822502520223,
      "grad_norm": 0.7272465825080872,
      "learning_rate": 1.5115447283710212e-05,
      "loss": 0.2017,
      "step": 12800
    },
    {
      "epoch": 1.2598657520100316,
      "grad_norm": 0.36164748668670654,
      "learning_rate": 1.51114731947701e-05,
      "loss": 0.268,
      "step": 12810
    },
    {
      "epoch": 1.2608492537680411,
      "grad_norm": 0.9491702914237976,
      "learning_rate": 1.510749910582999e-05,
      "loss": 0.3221,
      "step": 12820
    },
    {
      "epoch": 1.2618327555260505,
      "grad_norm": 1.4638797044754028,
      "learning_rate": 1.510352501688988e-05,
      "loss": 0.2655,
      "step": 12830
    },
    {
      "epoch": 1.2628162572840598,
      "grad_norm": 0.7647467851638794,
      "learning_rate": 1.509955092794977e-05,
      "loss": 0.2605,
      "step": 12840
    },
    {
      "epoch": 1.2637997590420693,
      "grad_norm": 1.9903299808502197,
      "learning_rate": 1.5095576839009657e-05,
      "loss": 0.1408,
      "step": 12850
    },
    {
      "epoch": 1.2647832608000786,
      "grad_norm": 0.31834790110588074,
      "learning_rate": 1.5091602750069547e-05,
      "loss": 0.2054,
      "step": 12860
    },
    {
      "epoch": 1.265766762558088,
      "grad_norm": 0.9295620918273926,
      "learning_rate": 1.5087628661129437e-05,
      "loss": 0.2929,
      "step": 12870
    },
    {
      "epoch": 1.2667502643160975,
      "grad_norm": 0.9623826146125793,
      "learning_rate": 1.5083654572189328e-05,
      "loss": 0.1799,
      "step": 12880
    },
    {
      "epoch": 1.2677337660741068,
      "grad_norm": 1.424196720123291,
      "learning_rate": 1.5079680483249218e-05,
      "loss": 0.2967,
      "step": 12890
    },
    {
      "epoch": 1.2687172678321161,
      "grad_norm": 3.3792314529418945,
      "learning_rate": 1.5075706394309106e-05,
      "loss": 0.1802,
      "step": 12900
    },
    {
      "epoch": 1.2697007695901257,
      "grad_norm": 0.9537076950073242,
      "learning_rate": 1.5071732305368995e-05,
      "loss": 0.1846,
      "step": 12910
    },
    {
      "epoch": 1.270684271348135,
      "grad_norm": 1.3461072444915771,
      "learning_rate": 1.5067758216428885e-05,
      "loss": 0.1985,
      "step": 12920
    },
    {
      "epoch": 1.2716677731061443,
      "grad_norm": 3.7908403873443604,
      "learning_rate": 1.5063784127488775e-05,
      "loss": 0.2591,
      "step": 12930
    },
    {
      "epoch": 1.2726512748641539,
      "grad_norm": 2.368834972381592,
      "learning_rate": 1.5059810038548663e-05,
      "loss": 0.2977,
      "step": 12940
    },
    {
      "epoch": 1.2736347766221632,
      "grad_norm": 1.240795612335205,
      "learning_rate": 1.5055835949608552e-05,
      "loss": 0.222,
      "step": 12950
    },
    {
      "epoch": 1.2746182783801725,
      "grad_norm": 0.35216814279556274,
      "learning_rate": 1.5051861860668442e-05,
      "loss": 0.2719,
      "step": 12960
    },
    {
      "epoch": 1.275601780138182,
      "grad_norm": 0.6168338656425476,
      "learning_rate": 1.5047887771728333e-05,
      "loss": 0.3023,
      "step": 12970
    },
    {
      "epoch": 1.2765852818961914,
      "grad_norm": 5.514144420623779,
      "learning_rate": 1.5043913682788223e-05,
      "loss": 0.2775,
      "step": 12980
    },
    {
      "epoch": 1.2775687836542007,
      "grad_norm": 1.363470435142517,
      "learning_rate": 1.5039939593848113e-05,
      "loss": 0.3031,
      "step": 12990
    },
    {
      "epoch": 1.2785522854122102,
      "grad_norm": 1.6486947536468506,
      "learning_rate": 1.5035965504908e-05,
      "loss": 0.1949,
      "step": 13000
    },
    {
      "epoch": 1.2785522854122102,
      "eval_loss": 0.16120639443397522,
      "eval_runtime": 17.3094,
      "eval_samples_per_second": 2.889,
      "eval_steps_per_second": 1.444,
      "step": 13000
    },
    {
      "epoch": 1.2795357871702195,
      "grad_norm": 0.5180084109306335,
      "learning_rate": 1.503199141596789e-05,
      "loss": 0.2341,
      "step": 13010
    },
    {
      "epoch": 1.2805192889282289,
      "grad_norm": 0.9390528798103333,
      "learning_rate": 1.502801732702778e-05,
      "loss": 0.2389,
      "step": 13020
    },
    {
      "epoch": 1.2815027906862384,
      "grad_norm": 2.1658847332000732,
      "learning_rate": 1.502404323808767e-05,
      "loss": 0.3606,
      "step": 13030
    },
    {
      "epoch": 1.2824862924442477,
      "grad_norm": 1.2159450054168701,
      "learning_rate": 1.5020069149147558e-05,
      "loss": 0.3057,
      "step": 13040
    },
    {
      "epoch": 1.283469794202257,
      "grad_norm": 2.1090898513793945,
      "learning_rate": 1.5016095060207449e-05,
      "loss": 0.2236,
      "step": 13050
    },
    {
      "epoch": 1.2844532959602666,
      "grad_norm": 0.6798930168151855,
      "learning_rate": 1.5012120971267339e-05,
      "loss": 0.228,
      "step": 13060
    },
    {
      "epoch": 1.285436797718276,
      "grad_norm": 1.0679746866226196,
      "learning_rate": 1.5008146882327228e-05,
      "loss": 0.238,
      "step": 13070
    },
    {
      "epoch": 1.2864202994762852,
      "grad_norm": 1.5444985628128052,
      "learning_rate": 1.5004172793387118e-05,
      "loss": 0.2334,
      "step": 13080
    },
    {
      "epoch": 1.2874038012342948,
      "grad_norm": 2.783684253692627,
      "learning_rate": 1.5000198704447006e-05,
      "loss": 0.2227,
      "step": 13090
    },
    {
      "epoch": 1.288387302992304,
      "grad_norm": 2.795809745788574,
      "learning_rate": 1.4996224615506896e-05,
      "loss": 0.283,
      "step": 13100
    },
    {
      "epoch": 1.2893708047503134,
      "grad_norm": 0.8713507652282715,
      "learning_rate": 1.4992250526566785e-05,
      "loss": 0.2309,
      "step": 13110
    },
    {
      "epoch": 1.290354306508323,
      "grad_norm": 0.8911621570587158,
      "learning_rate": 1.4988276437626675e-05,
      "loss": 0.2284,
      "step": 13120
    },
    {
      "epoch": 1.2913378082663323,
      "grad_norm": 0.840497612953186,
      "learning_rate": 1.4984302348686566e-05,
      "loss": 0.23,
      "step": 13130
    },
    {
      "epoch": 1.2923213100243416,
      "grad_norm": 1.8198657035827637,
      "learning_rate": 1.4980328259746454e-05,
      "loss": 0.2973,
      "step": 13140
    },
    {
      "epoch": 1.2933048117823511,
      "grad_norm": 2.5215814113616943,
      "learning_rate": 1.4976354170806344e-05,
      "loss": 0.3074,
      "step": 13150
    },
    {
      "epoch": 1.2942883135403604,
      "grad_norm": 1.1055022478103638,
      "learning_rate": 1.4972380081866234e-05,
      "loss": 0.1795,
      "step": 13160
    },
    {
      "epoch": 1.2952718152983698,
      "grad_norm": 1.3594915866851807,
      "learning_rate": 1.4968405992926123e-05,
      "loss": 0.1897,
      "step": 13170
    },
    {
      "epoch": 1.2962553170563793,
      "grad_norm": 0.9490194916725159,
      "learning_rate": 1.4964431903986011e-05,
      "loss": 0.208,
      "step": 13180
    },
    {
      "epoch": 1.2972388188143886,
      "grad_norm": 0.922896146774292,
      "learning_rate": 1.4960457815045901e-05,
      "loss": 0.2214,
      "step": 13190
    },
    {
      "epoch": 1.298222320572398,
      "grad_norm": 1.7510035037994385,
      "learning_rate": 1.495648372610579e-05,
      "loss": 0.1776,
      "step": 13200
    },
    {
      "epoch": 1.2992058223304075,
      "grad_norm": 2.1112654209136963,
      "learning_rate": 1.4952509637165682e-05,
      "loss": 0.2239,
      "step": 13210
    },
    {
      "epoch": 1.3001893240884168,
      "grad_norm": 0.6557284593582153,
      "learning_rate": 1.4948535548225572e-05,
      "loss": 0.1871,
      "step": 13220
    },
    {
      "epoch": 1.3011728258464261,
      "grad_norm": 3.473644733428955,
      "learning_rate": 1.4944561459285461e-05,
      "loss": 0.2617,
      "step": 13230
    },
    {
      "epoch": 1.3021563276044357,
      "grad_norm": 2.699770212173462,
      "learning_rate": 1.494058737034535e-05,
      "loss": 0.228,
      "step": 13240
    },
    {
      "epoch": 1.303139829362445,
      "grad_norm": 1.5528502464294434,
      "learning_rate": 1.4936613281405239e-05,
      "loss": 0.3006,
      "step": 13250
    },
    {
      "epoch": 1.3041233311204543,
      "grad_norm": 1.9565653800964355,
      "learning_rate": 1.4932639192465129e-05,
      "loss": 0.2494,
      "step": 13260
    },
    {
      "epoch": 1.3051068328784639,
      "grad_norm": 0.38119733333587646,
      "learning_rate": 1.4928665103525018e-05,
      "loss": 0.1939,
      "step": 13270
    },
    {
      "epoch": 1.3060903346364732,
      "grad_norm": 2.829664707183838,
      "learning_rate": 1.4924691014584906e-05,
      "loss": 0.166,
      "step": 13280
    },
    {
      "epoch": 1.3070738363944825,
      "grad_norm": 1.1717098951339722,
      "learning_rate": 1.4920716925644797e-05,
      "loss": 0.2323,
      "step": 13290
    },
    {
      "epoch": 1.308057338152492,
      "grad_norm": 1.2100815773010254,
      "learning_rate": 1.4916742836704687e-05,
      "loss": 0.3392,
      "step": 13300
    },
    {
      "epoch": 1.3090408399105014,
      "grad_norm": 1.8113962411880493,
      "learning_rate": 1.4912768747764577e-05,
      "loss": 0.2639,
      "step": 13310
    },
    {
      "epoch": 1.3100243416685107,
      "grad_norm": 3.25687837600708,
      "learning_rate": 1.4908794658824466e-05,
      "loss": 0.2449,
      "step": 13320
    },
    {
      "epoch": 1.3110078434265202,
      "grad_norm": 0.5960381031036377,
      "learning_rate": 1.4904820569884354e-05,
      "loss": 0.2679,
      "step": 13330
    },
    {
      "epoch": 1.3119913451845295,
      "grad_norm": 1.6405853033065796,
      "learning_rate": 1.4900846480944244e-05,
      "loss": 0.1678,
      "step": 13340
    },
    {
      "epoch": 1.3129748469425389,
      "grad_norm": 1.3594471216201782,
      "learning_rate": 1.4896872392004134e-05,
      "loss": 0.2544,
      "step": 13350
    },
    {
      "epoch": 1.3139583487005484,
      "grad_norm": 1.7740055322647095,
      "learning_rate": 1.4892898303064023e-05,
      "loss": 0.2556,
      "step": 13360
    },
    {
      "epoch": 1.3149418504585577,
      "grad_norm": 1.7624731063842773,
      "learning_rate": 1.4888924214123915e-05,
      "loss": 0.3219,
      "step": 13370
    },
    {
      "epoch": 1.315925352216567,
      "grad_norm": 1.0945662260055542,
      "learning_rate": 1.4884950125183803e-05,
      "loss": 0.3301,
      "step": 13380
    },
    {
      "epoch": 1.3169088539745766,
      "grad_norm": 0.43884381651878357,
      "learning_rate": 1.4880976036243692e-05,
      "loss": 0.3388,
      "step": 13390
    },
    {
      "epoch": 1.317892355732586,
      "grad_norm": 0.7153629660606384,
      "learning_rate": 1.4877001947303582e-05,
      "loss": 0.1712,
      "step": 13400
    },
    {
      "epoch": 1.3188758574905952,
      "grad_norm": 2.686660051345825,
      "learning_rate": 1.4873027858363472e-05,
      "loss": 0.1999,
      "step": 13410
    },
    {
      "epoch": 1.3198593592486048,
      "grad_norm": 1.2886070013046265,
      "learning_rate": 1.486905376942336e-05,
      "loss": 0.3052,
      "step": 13420
    },
    {
      "epoch": 1.320842861006614,
      "grad_norm": 1.0095949172973633,
      "learning_rate": 1.486507968048325e-05,
      "loss": 0.2075,
      "step": 13430
    },
    {
      "epoch": 1.3218263627646234,
      "grad_norm": 2.6597988605499268,
      "learning_rate": 1.4861105591543139e-05,
      "loss": 0.1725,
      "step": 13440
    },
    {
      "epoch": 1.322809864522633,
      "grad_norm": 0.8734903335571289,
      "learning_rate": 1.485713150260303e-05,
      "loss": 0.3388,
      "step": 13450
    },
    {
      "epoch": 1.3237933662806423,
      "grad_norm": 0.8990864753723145,
      "learning_rate": 1.485315741366292e-05,
      "loss": 0.1874,
      "step": 13460
    },
    {
      "epoch": 1.3247768680386516,
      "grad_norm": 1.3725947141647339,
      "learning_rate": 1.484918332472281e-05,
      "loss": 0.2593,
      "step": 13470
    },
    {
      "epoch": 1.3257603697966611,
      "grad_norm": 2.846909284591675,
      "learning_rate": 1.4845209235782698e-05,
      "loss": 0.3015,
      "step": 13480
    },
    {
      "epoch": 1.3267438715546704,
      "grad_norm": 0.32054901123046875,
      "learning_rate": 1.4841235146842587e-05,
      "loss": 0.1395,
      "step": 13490
    },
    {
      "epoch": 1.3277273733126798,
      "grad_norm": 1.4633597135543823,
      "learning_rate": 1.4837261057902477e-05,
      "loss": 0.312,
      "step": 13500
    },
    {
      "epoch": 1.3277273733126798,
      "eval_loss": 0.16013847291469574,
      "eval_runtime": 15.9302,
      "eval_samples_per_second": 3.139,
      "eval_steps_per_second": 1.569,
      "step": 13500
    },
    {
      "epoch": 1.3287108750706893,
      "grad_norm": 1.0678695440292358,
      "learning_rate": 1.4833286968962367e-05,
      "loss": 0.153,
      "step": 13510
    },
    {
      "epoch": 1.3296943768286986,
      "grad_norm": 2.600311040878296,
      "learning_rate": 1.4829312880022255e-05,
      "loss": 0.1909,
      "step": 13520
    },
    {
      "epoch": 1.330677878586708,
      "grad_norm": 0.7847926020622253,
      "learning_rate": 1.4825338791082146e-05,
      "loss": 0.3542,
      "step": 13530
    },
    {
      "epoch": 1.3316613803447175,
      "grad_norm": 1.6524434089660645,
      "learning_rate": 1.4821364702142036e-05,
      "loss": 0.1395,
      "step": 13540
    },
    {
      "epoch": 1.3326448821027268,
      "grad_norm": 0.7375239729881287,
      "learning_rate": 1.4817390613201925e-05,
      "loss": 0.2814,
      "step": 13550
    },
    {
      "epoch": 1.3336283838607361,
      "grad_norm": 1.0905476808547974,
      "learning_rate": 1.4813416524261815e-05,
      "loss": 0.1523,
      "step": 13560
    },
    {
      "epoch": 1.3346118856187457,
      "grad_norm": 1.2427202463150024,
      "learning_rate": 1.4809442435321703e-05,
      "loss": 0.2604,
      "step": 13570
    },
    {
      "epoch": 1.335595387376755,
      "grad_norm": 1.518129825592041,
      "learning_rate": 1.4805468346381593e-05,
      "loss": 0.2609,
      "step": 13580
    },
    {
      "epoch": 1.3365788891347643,
      "grad_norm": 0.9168410897254944,
      "learning_rate": 1.4801494257441482e-05,
      "loss": 0.2219,
      "step": 13590
    },
    {
      "epoch": 1.3375623908927738,
      "grad_norm": 0.8411401510238647,
      "learning_rate": 1.4797520168501372e-05,
      "loss": 0.2587,
      "step": 13600
    },
    {
      "epoch": 1.3385458926507832,
      "grad_norm": 2.603593587875366,
      "learning_rate": 1.4793546079561263e-05,
      "loss": 0.1987,
      "step": 13610
    },
    {
      "epoch": 1.3395293944087925,
      "grad_norm": 1.346815824508667,
      "learning_rate": 1.4789571990621151e-05,
      "loss": 0.284,
      "step": 13620
    },
    {
      "epoch": 1.340512896166802,
      "grad_norm": 1.0927194356918335,
      "learning_rate": 1.4785597901681041e-05,
      "loss": 0.2548,
      "step": 13630
    },
    {
      "epoch": 1.3414963979248113,
      "grad_norm": 1.6921230554580688,
      "learning_rate": 1.478162381274093e-05,
      "loss": 0.2953,
      "step": 13640
    },
    {
      "epoch": 1.3424798996828207,
      "grad_norm": 0.9991982579231262,
      "learning_rate": 1.477764972380082e-05,
      "loss": 0.19,
      "step": 13650
    },
    {
      "epoch": 1.3434634014408302,
      "grad_norm": 1.5114970207214355,
      "learning_rate": 1.4773675634860708e-05,
      "loss": 0.2231,
      "step": 13660
    },
    {
      "epoch": 1.3444469031988395,
      "grad_norm": 1.9252091646194458,
      "learning_rate": 1.4769701545920598e-05,
      "loss": 0.2684,
      "step": 13670
    },
    {
      "epoch": 1.3454304049568488,
      "grad_norm": 0.333156019449234,
      "learning_rate": 1.4765727456980488e-05,
      "loss": 0.1842,
      "step": 13680
    },
    {
      "epoch": 1.3464139067148584,
      "grad_norm": 1.4571133852005005,
      "learning_rate": 1.4761753368040377e-05,
      "loss": 0.3031,
      "step": 13690
    },
    {
      "epoch": 1.3473974084728677,
      "grad_norm": 3.225837469100952,
      "learning_rate": 1.4757779279100269e-05,
      "loss": 0.2891,
      "step": 13700
    },
    {
      "epoch": 1.348380910230877,
      "grad_norm": 0.8479882478713989,
      "learning_rate": 1.4753805190160158e-05,
      "loss": 0.2203,
      "step": 13710
    },
    {
      "epoch": 1.3493644119888863,
      "grad_norm": 1.3609168529510498,
      "learning_rate": 1.4749831101220046e-05,
      "loss": 0.3754,
      "step": 13720
    },
    {
      "epoch": 1.350347913746896,
      "grad_norm": 2.286158561706543,
      "learning_rate": 1.4745857012279936e-05,
      "loss": 0.3493,
      "step": 13730
    },
    {
      "epoch": 1.3513314155049052,
      "grad_norm": 0.7514020204544067,
      "learning_rate": 1.4741882923339826e-05,
      "loss": 0.3501,
      "step": 13740
    },
    {
      "epoch": 1.3523149172629145,
      "grad_norm": 1.157344937324524,
      "learning_rate": 1.4737908834399715e-05,
      "loss": 0.3249,
      "step": 13750
    },
    {
      "epoch": 1.353298419020924,
      "grad_norm": 2.1621084213256836,
      "learning_rate": 1.4733934745459603e-05,
      "loss": 0.1955,
      "step": 13760
    },
    {
      "epoch": 1.3542819207789334,
      "grad_norm": 1.8904935121536255,
      "learning_rate": 1.4729960656519493e-05,
      "loss": 0.2264,
      "step": 13770
    },
    {
      "epoch": 1.3552654225369427,
      "grad_norm": 0.9935181736946106,
      "learning_rate": 1.4725986567579384e-05,
      "loss": 0.2989,
      "step": 13780
    },
    {
      "epoch": 1.3562489242949523,
      "grad_norm": 1.2855091094970703,
      "learning_rate": 1.4722012478639274e-05,
      "loss": 0.1557,
      "step": 13790
    },
    {
      "epoch": 1.3572324260529616,
      "grad_norm": 0.6436872482299805,
      "learning_rate": 1.4718038389699164e-05,
      "loss": 0.3041,
      "step": 13800
    },
    {
      "epoch": 1.358215927810971,
      "grad_norm": 2.33740496635437,
      "learning_rate": 1.4714064300759052e-05,
      "loss": 0.3134,
      "step": 13810
    },
    {
      "epoch": 1.3591994295689804,
      "grad_norm": 0.6652860641479492,
      "learning_rate": 1.4710090211818941e-05,
      "loss": 0.232,
      "step": 13820
    },
    {
      "epoch": 1.3601829313269898,
      "grad_norm": 1.8206769227981567,
      "learning_rate": 1.4706116122878831e-05,
      "loss": 0.2781,
      "step": 13830
    },
    {
      "epoch": 1.361166433084999,
      "grad_norm": 2.324544906616211,
      "learning_rate": 1.470214203393872e-05,
      "loss": 0.3082,
      "step": 13840
    },
    {
      "epoch": 1.3621499348430086,
      "grad_norm": 1.1942061185836792,
      "learning_rate": 1.4698167944998609e-05,
      "loss": 0.2314,
      "step": 13850
    },
    {
      "epoch": 1.363133436601018,
      "grad_norm": 0.7377877831459045,
      "learning_rate": 1.46941938560585e-05,
      "loss": 0.1839,
      "step": 13860
    },
    {
      "epoch": 1.3641169383590273,
      "grad_norm": 1.2359178066253662,
      "learning_rate": 1.469021976711839e-05,
      "loss": 0.33,
      "step": 13870
    },
    {
      "epoch": 1.3651004401170366,
      "grad_norm": 0.2554020583629608,
      "learning_rate": 1.468624567817828e-05,
      "loss": 0.2232,
      "step": 13880
    },
    {
      "epoch": 1.3660839418750461,
      "grad_norm": 1.4700244665145874,
      "learning_rate": 1.4682271589238169e-05,
      "loss": 0.2821,
      "step": 13890
    },
    {
      "epoch": 1.3670674436330554,
      "grad_norm": 2.099269390106201,
      "learning_rate": 1.4678297500298057e-05,
      "loss": 0.3381,
      "step": 13900
    },
    {
      "epoch": 1.3680509453910648,
      "grad_norm": 1.11494779586792,
      "learning_rate": 1.4674323411357947e-05,
      "loss": 0.2198,
      "step": 13910
    },
    {
      "epoch": 1.3690344471490743,
      "grad_norm": 0.775420069694519,
      "learning_rate": 1.4670349322417836e-05,
      "loss": 0.2713,
      "step": 13920
    },
    {
      "epoch": 1.3700179489070836,
      "grad_norm": 0.717750072479248,
      "learning_rate": 1.4666375233477726e-05,
      "loss": 0.2709,
      "step": 13930
    },
    {
      "epoch": 1.371001450665093,
      "grad_norm": 4.715325355529785,
      "learning_rate": 1.4662401144537617e-05,
      "loss": 0.3318,
      "step": 13940
    },
    {
      "epoch": 1.3719849524231025,
      "grad_norm": 1.486010193824768,
      "learning_rate": 1.4658427055597507e-05,
      "loss": 0.2406,
      "step": 13950
    },
    {
      "epoch": 1.3729684541811118,
      "grad_norm": 0.7549448609352112,
      "learning_rate": 1.4654452966657395e-05,
      "loss": 0.2218,
      "step": 13960
    },
    {
      "epoch": 1.3739519559391211,
      "grad_norm": 4.172186374664307,
      "learning_rate": 1.4650478877717285e-05,
      "loss": 0.2255,
      "step": 13970
    },
    {
      "epoch": 1.3749354576971307,
      "grad_norm": 1.9817334413528442,
      "learning_rate": 1.4646504788777174e-05,
      "loss": 0.2186,
      "step": 13980
    },
    {
      "epoch": 1.37591895945514,
      "grad_norm": 0.8794751763343811,
      "learning_rate": 1.4642530699837064e-05,
      "loss": 0.225,
      "step": 13990
    },
    {
      "epoch": 1.3769024612131493,
      "grad_norm": 2.93635630607605,
      "learning_rate": 1.4638556610896952e-05,
      "loss": 0.2007,
      "step": 14000
    },
    {
      "epoch": 1.3769024612131493,
      "eval_loss": 0.1604984849691391,
      "eval_runtime": 18.6319,
      "eval_samples_per_second": 2.684,
      "eval_steps_per_second": 1.342,
      "step": 14000
    },
    {
      "epoch": 1.3778859629711588,
      "grad_norm": 0.817743182182312,
      "learning_rate": 1.4634582521956842e-05,
      "loss": 0.3481,
      "step": 14010
    },
    {
      "epoch": 1.3788694647291682,
      "grad_norm": 1.1612430810928345,
      "learning_rate": 1.4630608433016733e-05,
      "loss": 0.2793,
      "step": 14020
    },
    {
      "epoch": 1.3798529664871775,
      "grad_norm": 3.0838816165924072,
      "learning_rate": 1.4626634344076623e-05,
      "loss": 0.2032,
      "step": 14030
    },
    {
      "epoch": 1.380836468245187,
      "grad_norm": 0.568408191204071,
      "learning_rate": 1.4622660255136512e-05,
      "loss": 0.2087,
      "step": 14040
    },
    {
      "epoch": 1.3818199700031963,
      "grad_norm": 3.588900327682495,
      "learning_rate": 1.46186861661964e-05,
      "loss": 0.2822,
      "step": 14050
    },
    {
      "epoch": 1.3828034717612057,
      "grad_norm": 1.1956599950790405,
      "learning_rate": 1.461471207725629e-05,
      "loss": 0.3036,
      "step": 14060
    },
    {
      "epoch": 1.3837869735192152,
      "grad_norm": 0.5779142379760742,
      "learning_rate": 1.461073798831618e-05,
      "loss": 0.2908,
      "step": 14070
    },
    {
      "epoch": 1.3847704752772245,
      "grad_norm": 1.5077542066574097,
      "learning_rate": 1.460676389937607e-05,
      "loss": 0.2289,
      "step": 14080
    },
    {
      "epoch": 1.3857539770352338,
      "grad_norm": 1.0276594161987305,
      "learning_rate": 1.4602789810435957e-05,
      "loss": 0.3104,
      "step": 14090
    },
    {
      "epoch": 1.3867374787932434,
      "grad_norm": 1.3784536123275757,
      "learning_rate": 1.4598815721495848e-05,
      "loss": 0.2864,
      "step": 14100
    },
    {
      "epoch": 1.3877209805512527,
      "grad_norm": 0.7437291145324707,
      "learning_rate": 1.4594841632555738e-05,
      "loss": 0.236,
      "step": 14110
    },
    {
      "epoch": 1.388704482309262,
      "grad_norm": 0.7670959234237671,
      "learning_rate": 1.4590867543615628e-05,
      "loss": 0.273,
      "step": 14120
    },
    {
      "epoch": 1.3896879840672716,
      "grad_norm": 1.2429335117340088,
      "learning_rate": 1.4586893454675517e-05,
      "loss": 0.21,
      "step": 14130
    },
    {
      "epoch": 1.3906714858252809,
      "grad_norm": 0.8416498303413391,
      "learning_rate": 1.4582919365735405e-05,
      "loss": 0.2271,
      "step": 14140
    },
    {
      "epoch": 1.3916549875832902,
      "grad_norm": 0.8700377941131592,
      "learning_rate": 1.4578945276795295e-05,
      "loss": 0.2903,
      "step": 14150
    },
    {
      "epoch": 1.3926384893412997,
      "grad_norm": 1.6111570596694946,
      "learning_rate": 1.4574971187855185e-05,
      "loss": 0.1538,
      "step": 14160
    },
    {
      "epoch": 1.393621991099309,
      "grad_norm": 2.0806915760040283,
      "learning_rate": 1.4570997098915074e-05,
      "loss": 0.2772,
      "step": 14170
    },
    {
      "epoch": 1.3946054928573184,
      "grad_norm": 2.6657345294952393,
      "learning_rate": 1.4567023009974966e-05,
      "loss": 0.2076,
      "step": 14180
    },
    {
      "epoch": 1.395588994615328,
      "grad_norm": 0.41977888345718384,
      "learning_rate": 1.4563048921034855e-05,
      "loss": 0.1928,
      "step": 14190
    },
    {
      "epoch": 1.3965724963733372,
      "grad_norm": 2.8233563899993896,
      "learning_rate": 1.4559074832094743e-05,
      "loss": 0.309,
      "step": 14200
    },
    {
      "epoch": 1.3975559981313466,
      "grad_norm": 0.8026371598243713,
      "learning_rate": 1.4555100743154633e-05,
      "loss": 0.2731,
      "step": 14210
    },
    {
      "epoch": 1.398539499889356,
      "grad_norm": 1.8230639696121216,
      "learning_rate": 1.4551126654214523e-05,
      "loss": 0.1933,
      "step": 14220
    },
    {
      "epoch": 1.3995230016473654,
      "grad_norm": 2.1196181774139404,
      "learning_rate": 1.4547152565274412e-05,
      "loss": 0.2856,
      "step": 14230
    },
    {
      "epoch": 1.4005065034053747,
      "grad_norm": 0.863348662853241,
      "learning_rate": 1.45431784763343e-05,
      "loss": 0.3251,
      "step": 14240
    },
    {
      "epoch": 1.4014900051633843,
      "grad_norm": 0.5216766595840454,
      "learning_rate": 1.453920438739419e-05,
      "loss": 0.292,
      "step": 14250
    },
    {
      "epoch": 1.4024735069213936,
      "grad_norm": 1.193357229232788,
      "learning_rate": 1.4535230298454081e-05,
      "loss": 0.2078,
      "step": 14260
    },
    {
      "epoch": 1.403457008679403,
      "grad_norm": 1.630039095878601,
      "learning_rate": 1.4531256209513971e-05,
      "loss": 0.26,
      "step": 14270
    },
    {
      "epoch": 1.4044405104374125,
      "grad_norm": 1.493647813796997,
      "learning_rate": 1.452728212057386e-05,
      "loss": 0.1423,
      "step": 14280
    },
    {
      "epoch": 1.4054240121954218,
      "grad_norm": 1.684227466583252,
      "learning_rate": 1.4523308031633749e-05,
      "loss": 0.2289,
      "step": 14290
    },
    {
      "epoch": 1.406407513953431,
      "grad_norm": 0.8049089908599854,
      "learning_rate": 1.4519333942693638e-05,
      "loss": 0.285,
      "step": 14300
    },
    {
      "epoch": 1.4073910157114407,
      "grad_norm": 2.782198905944824,
      "learning_rate": 1.4515359853753528e-05,
      "loss": 0.1941,
      "step": 14310
    },
    {
      "epoch": 1.40837451746945,
      "grad_norm": 0.9174168109893799,
      "learning_rate": 1.4511385764813418e-05,
      "loss": 0.2563,
      "step": 14320
    },
    {
      "epoch": 1.4093580192274593,
      "grad_norm": 0.38679203391075134,
      "learning_rate": 1.4507411675873306e-05,
      "loss": 0.2634,
      "step": 14330
    },
    {
      "epoch": 1.4103415209854688,
      "grad_norm": 1.094164490699768,
      "learning_rate": 1.4503437586933197e-05,
      "loss": 0.2494,
      "step": 14340
    },
    {
      "epoch": 1.4113250227434782,
      "grad_norm": 3.6802244186401367,
      "learning_rate": 1.4499463497993087e-05,
      "loss": 0.2378,
      "step": 14350
    },
    {
      "epoch": 1.4123085245014875,
      "grad_norm": 2.8969945907592773,
      "learning_rate": 1.4495489409052976e-05,
      "loss": 0.246,
      "step": 14360
    },
    {
      "epoch": 1.413292026259497,
      "grad_norm": 5.990109443664551,
      "learning_rate": 1.4491515320112866e-05,
      "loss": 0.2624,
      "step": 14370
    },
    {
      "epoch": 1.4142755280175063,
      "grad_norm": 0.3921045958995819,
      "learning_rate": 1.4487541231172754e-05,
      "loss": 0.1834,
      "step": 14380
    },
    {
      "epoch": 1.4152590297755157,
      "grad_norm": 0.8760048747062683,
      "learning_rate": 1.4483567142232644e-05,
      "loss": 0.2399,
      "step": 14390
    },
    {
      "epoch": 1.4162425315335252,
      "grad_norm": 0.5012834668159485,
      "learning_rate": 1.4479593053292533e-05,
      "loss": 0.1621,
      "step": 14400
    },
    {
      "epoch": 1.4172260332915345,
      "grad_norm": 2.5045790672302246,
      "learning_rate": 1.4475618964352423e-05,
      "loss": 0.2474,
      "step": 14410
    },
    {
      "epoch": 1.4182095350495438,
      "grad_norm": 0.9351314902305603,
      "learning_rate": 1.4471644875412314e-05,
      "loss": 0.1777,
      "step": 14420
    },
    {
      "epoch": 1.4191930368075534,
      "grad_norm": 1.5870965719223022,
      "learning_rate": 1.4467670786472204e-05,
      "loss": 0.3198,
      "step": 14430
    },
    {
      "epoch": 1.4201765385655627,
      "grad_norm": 1.0936942100524902,
      "learning_rate": 1.4463696697532092e-05,
      "loss": 0.2152,
      "step": 14440
    },
    {
      "epoch": 1.421160040323572,
      "grad_norm": 0.5950337648391724,
      "learning_rate": 1.4459722608591982e-05,
      "loss": 0.2609,
      "step": 14450
    },
    {
      "epoch": 1.4221435420815816,
      "grad_norm": 0.7351011633872986,
      "learning_rate": 1.4455748519651871e-05,
      "loss": 0.2023,
      "step": 14460
    },
    {
      "epoch": 1.4231270438395909,
      "grad_norm": 1.832258939743042,
      "learning_rate": 1.4451774430711761e-05,
      "loss": 0.2673,
      "step": 14470
    },
    {
      "epoch": 1.4241105455976002,
      "grad_norm": 1.5636357069015503,
      "learning_rate": 1.4447800341771649e-05,
      "loss": 0.2644,
      "step": 14480
    },
    {
      "epoch": 1.4250940473556097,
      "grad_norm": 1.3714380264282227,
      "learning_rate": 1.4443826252831539e-05,
      "loss": 0.2067,
      "step": 14490
    },
    {
      "epoch": 1.426077549113619,
      "grad_norm": 3.591639757156372,
      "learning_rate": 1.4439852163891428e-05,
      "loss": 0.2843,
      "step": 14500
    },
    {
      "epoch": 1.426077549113619,
      "eval_loss": 0.1673988401889801,
      "eval_runtime": 16.9001,
      "eval_samples_per_second": 2.959,
      "eval_steps_per_second": 1.479,
      "step": 14500
    },
    {
      "epoch": 1.4270610508716284,
      "grad_norm": 1.7980598211288452,
      "learning_rate": 1.443587807495132e-05,
      "loss": 0.2681,
      "step": 14510
    },
    {
      "epoch": 1.428044552629638,
      "grad_norm": 1.097467064857483,
      "learning_rate": 1.443190398601121e-05,
      "loss": 0.262,
      "step": 14520
    },
    {
      "epoch": 1.4290280543876472,
      "grad_norm": 3.742469072341919,
      "learning_rate": 1.4427929897071097e-05,
      "loss": 0.1493,
      "step": 14530
    },
    {
      "epoch": 1.4300115561456566,
      "grad_norm": 0.4028743505477905,
      "learning_rate": 1.4423955808130987e-05,
      "loss": 0.3007,
      "step": 14540
    },
    {
      "epoch": 1.430995057903666,
      "grad_norm": 1.2121284008026123,
      "learning_rate": 1.4419981719190877e-05,
      "loss": 0.2974,
      "step": 14550
    },
    {
      "epoch": 1.4319785596616754,
      "grad_norm": 1.3690731525421143,
      "learning_rate": 1.4416007630250766e-05,
      "loss": 0.199,
      "step": 14560
    },
    {
      "epoch": 1.4329620614196847,
      "grad_norm": 1.1825859546661377,
      "learning_rate": 1.4412033541310654e-05,
      "loss": 0.2872,
      "step": 14570
    },
    {
      "epoch": 1.4339455631776943,
      "grad_norm": 2.86308217048645,
      "learning_rate": 1.4408059452370544e-05,
      "loss": 0.2329,
      "step": 14580
    },
    {
      "epoch": 1.4349290649357036,
      "grad_norm": 1.6855119466781616,
      "learning_rate": 1.4404085363430435e-05,
      "loss": 0.2582,
      "step": 14590
    },
    {
      "epoch": 1.435912566693713,
      "grad_norm": 4.255486965179443,
      "learning_rate": 1.4400111274490325e-05,
      "loss": 0.345,
      "step": 14600
    },
    {
      "epoch": 1.4368960684517225,
      "grad_norm": 3.486799478530884,
      "learning_rate": 1.4396137185550215e-05,
      "loss": 0.166,
      "step": 14610
    },
    {
      "epoch": 1.4378795702097318,
      "grad_norm": 0.6261281371116638,
      "learning_rate": 1.4392163096610103e-05,
      "loss": 0.1763,
      "step": 14620
    },
    {
      "epoch": 1.438863071967741,
      "grad_norm": 0.37830328941345215,
      "learning_rate": 1.4388189007669992e-05,
      "loss": 0.2208,
      "step": 14630
    },
    {
      "epoch": 1.4398465737257506,
      "grad_norm": 1.6429592370986938,
      "learning_rate": 1.4384214918729882e-05,
      "loss": 0.2523,
      "step": 14640
    },
    {
      "epoch": 1.44083007548376,
      "grad_norm": 0.754708468914032,
      "learning_rate": 1.4380240829789772e-05,
      "loss": 0.281,
      "step": 14650
    },
    {
      "epoch": 1.4418135772417693,
      "grad_norm": 0.8801028728485107,
      "learning_rate": 1.437626674084966e-05,
      "loss": 0.2863,
      "step": 14660
    },
    {
      "epoch": 1.4427970789997788,
      "grad_norm": 3.150028944015503,
      "learning_rate": 1.4372292651909553e-05,
      "loss": 0.2345,
      "step": 14670
    },
    {
      "epoch": 1.4437805807577881,
      "grad_norm": 0.9762952923774719,
      "learning_rate": 1.436831856296944e-05,
      "loss": 0.2124,
      "step": 14680
    },
    {
      "epoch": 1.4447640825157975,
      "grad_norm": 1.6731175184249878,
      "learning_rate": 1.436434447402933e-05,
      "loss": 0.2326,
      "step": 14690
    },
    {
      "epoch": 1.445747584273807,
      "grad_norm": 2.7856829166412354,
      "learning_rate": 1.436037038508922e-05,
      "loss": 0.2434,
      "step": 14700
    },
    {
      "epoch": 1.4467310860318163,
      "grad_norm": 1.2525384426116943,
      "learning_rate": 1.435639629614911e-05,
      "loss": 0.2459,
      "step": 14710
    },
    {
      "epoch": 1.4477145877898256,
      "grad_norm": 0.6227846741676331,
      "learning_rate": 1.4352422207208998e-05,
      "loss": 0.2067,
      "step": 14720
    },
    {
      "epoch": 1.4486980895478352,
      "grad_norm": 1.1095690727233887,
      "learning_rate": 1.4348448118268887e-05,
      "loss": 0.1717,
      "step": 14730
    },
    {
      "epoch": 1.4496815913058445,
      "grad_norm": 0.937445878982544,
      "learning_rate": 1.4344474029328777e-05,
      "loss": 0.1699,
      "step": 14740
    },
    {
      "epoch": 1.4506650930638538,
      "grad_norm": 0.8587270379066467,
      "learning_rate": 1.4340499940388668e-05,
      "loss": 0.2592,
      "step": 14750
    },
    {
      "epoch": 1.4516485948218634,
      "grad_norm": 1.02701997756958,
      "learning_rate": 1.4336525851448558e-05,
      "loss": 0.2234,
      "step": 14760
    },
    {
      "epoch": 1.4526320965798727,
      "grad_norm": 1.4647469520568848,
      "learning_rate": 1.4332551762508446e-05,
      "loss": 0.2237,
      "step": 14770
    },
    {
      "epoch": 1.453615598337882,
      "grad_norm": 0.657336950302124,
      "learning_rate": 1.4328577673568336e-05,
      "loss": 0.1871,
      "step": 14780
    },
    {
      "epoch": 1.4545991000958916,
      "grad_norm": 1.5731921195983887,
      "learning_rate": 1.4324603584628225e-05,
      "loss": 0.2241,
      "step": 14790
    },
    {
      "epoch": 1.4555826018539009,
      "grad_norm": 2.8147099018096924,
      "learning_rate": 1.4320629495688115e-05,
      "loss": 0.2887,
      "step": 14800
    },
    {
      "epoch": 1.4565661036119102,
      "grad_norm": 1.6455349922180176,
      "learning_rate": 1.4316655406748003e-05,
      "loss": 0.2719,
      "step": 14810
    },
    {
      "epoch": 1.4575496053699197,
      "grad_norm": 1.0510404109954834,
      "learning_rate": 1.4312681317807892e-05,
      "loss": 0.293,
      "step": 14820
    },
    {
      "epoch": 1.458533107127929,
      "grad_norm": 2.582322120666504,
      "learning_rate": 1.4308707228867784e-05,
      "loss": 0.1853,
      "step": 14830
    },
    {
      "epoch": 1.4595166088859384,
      "grad_norm": 1.337494134902954,
      "learning_rate": 1.4304733139927674e-05,
      "loss": 0.3178,
      "step": 14840
    },
    {
      "epoch": 1.460500110643948,
      "grad_norm": 0.4273045063018799,
      "learning_rate": 1.4300759050987563e-05,
      "loss": 0.2021,
      "step": 14850
    },
    {
      "epoch": 1.4614836124019572,
      "grad_norm": 1.0830575227737427,
      "learning_rate": 1.4296784962047451e-05,
      "loss": 0.3078,
      "step": 14860
    },
    {
      "epoch": 1.4624671141599666,
      "grad_norm": 0.7531421184539795,
      "learning_rate": 1.429281087310734e-05,
      "loss": 0.2925,
      "step": 14870
    },
    {
      "epoch": 1.4634506159179759,
      "grad_norm": 1.9662960767745972,
      "learning_rate": 1.428883678416723e-05,
      "loss": 0.2291,
      "step": 14880
    },
    {
      "epoch": 1.4644341176759854,
      "grad_norm": 0.8408644795417786,
      "learning_rate": 1.428486269522712e-05,
      "loss": 0.3016,
      "step": 14890
    },
    {
      "epoch": 1.4654176194339947,
      "grad_norm": 0.5292760729789734,
      "learning_rate": 1.4280888606287008e-05,
      "loss": 0.1698,
      "step": 14900
    },
    {
      "epoch": 1.466401121192004,
      "grad_norm": 1.8510633707046509,
      "learning_rate": 1.4276914517346901e-05,
      "loss": 0.2566,
      "step": 14910
    },
    {
      "epoch": 1.4673846229500136,
      "grad_norm": 0.41463837027549744,
      "learning_rate": 1.4272940428406789e-05,
      "loss": 0.1905,
      "step": 14920
    },
    {
      "epoch": 1.468368124708023,
      "grad_norm": 1.1431360244750977,
      "learning_rate": 1.4268966339466679e-05,
      "loss": 0.1914,
      "step": 14930
    },
    {
      "epoch": 1.4693516264660322,
      "grad_norm": 2.338446617126465,
      "learning_rate": 1.4264992250526568e-05,
      "loss": 0.3817,
      "step": 14940
    },
    {
      "epoch": 1.4703351282240418,
      "grad_norm": 0.3825877606868744,
      "learning_rate": 1.4261018161586458e-05,
      "loss": 0.265,
      "step": 14950
    },
    {
      "epoch": 1.471318629982051,
      "grad_norm": 6.677371025085449,
      "learning_rate": 1.4257044072646346e-05,
      "loss": 0.2294,
      "step": 14960
    },
    {
      "epoch": 1.4723021317400604,
      "grad_norm": 0.9877021312713623,
      "learning_rate": 1.4253069983706236e-05,
      "loss": 0.186,
      "step": 14970
    },
    {
      "epoch": 1.47328563349807,
      "grad_norm": 0.9139722585678101,
      "learning_rate": 1.4249095894766125e-05,
      "loss": 0.2549,
      "step": 14980
    },
    {
      "epoch": 1.4742691352560793,
      "grad_norm": 8.23216438293457,
      "learning_rate": 1.4245121805826017e-05,
      "loss": 0.2506,
      "step": 14990
    },
    {
      "epoch": 1.4752526370140886,
      "grad_norm": 1.6111382246017456,
      "learning_rate": 1.4241147716885906e-05,
      "loss": 0.1843,
      "step": 15000
    },
    {
      "epoch": 1.4752526370140886,
      "eval_loss": 0.15471838414669037,
      "eval_runtime": 19.6276,
      "eval_samples_per_second": 2.547,
      "eval_steps_per_second": 1.274,
      "step": 15000
    },
    {
      "epoch": 1.476236138772098,
      "grad_norm": 4.984705448150635,
      "learning_rate": 1.4237173627945794e-05,
      "loss": 0.2281,
      "step": 15010
    },
    {
      "epoch": 1.4772196405301075,
      "grad_norm": 0.5729973316192627,
      "learning_rate": 1.4233199539005684e-05,
      "loss": 0.1993,
      "step": 15020
    },
    {
      "epoch": 1.4782031422881168,
      "grad_norm": 1.0867680311203003,
      "learning_rate": 1.4229225450065574e-05,
      "loss": 0.2719,
      "step": 15030
    },
    {
      "epoch": 1.479186644046126,
      "grad_norm": 0.7489296793937683,
      "learning_rate": 1.4225251361125463e-05,
      "loss": 0.2843,
      "step": 15040
    },
    {
      "epoch": 1.4801701458041356,
      "grad_norm": 2.038597345352173,
      "learning_rate": 1.4221277272185351e-05,
      "loss": 0.1898,
      "step": 15050
    },
    {
      "epoch": 1.481153647562145,
      "grad_norm": 0.9547818899154663,
      "learning_rate": 1.4217303183245241e-05,
      "loss": 0.2054,
      "step": 15060
    },
    {
      "epoch": 1.4821371493201543,
      "grad_norm": 0.9691654443740845,
      "learning_rate": 1.4213329094305132e-05,
      "loss": 0.2412,
      "step": 15070
    },
    {
      "epoch": 1.4831206510781638,
      "grad_norm": 0.49495363235473633,
      "learning_rate": 1.4209355005365022e-05,
      "loss": 0.2072,
      "step": 15080
    },
    {
      "epoch": 1.4841041528361731,
      "grad_norm": 1.4077001810073853,
      "learning_rate": 1.4205380916424912e-05,
      "loss": 0.3349,
      "step": 15090
    },
    {
      "epoch": 1.4850876545941825,
      "grad_norm": 2.472428560256958,
      "learning_rate": 1.42014068274848e-05,
      "loss": 0.1961,
      "step": 15100
    },
    {
      "epoch": 1.486071156352192,
      "grad_norm": 2.658175230026245,
      "learning_rate": 1.419743273854469e-05,
      "loss": 0.3282,
      "step": 15110
    },
    {
      "epoch": 1.4870546581102013,
      "grad_norm": 1.4328405857086182,
      "learning_rate": 1.4193458649604579e-05,
      "loss": 0.1471,
      "step": 15120
    },
    {
      "epoch": 1.4880381598682106,
      "grad_norm": 8.422746658325195,
      "learning_rate": 1.4189484560664469e-05,
      "loss": 0.3153,
      "step": 15130
    },
    {
      "epoch": 1.4890216616262202,
      "grad_norm": 1.2293131351470947,
      "learning_rate": 1.4185510471724357e-05,
      "loss": 0.1695,
      "step": 15140
    },
    {
      "epoch": 1.4900051633842295,
      "grad_norm": 1.2073103189468384,
      "learning_rate": 1.418153638278425e-05,
      "loss": 0.2651,
      "step": 15150
    },
    {
      "epoch": 1.4909886651422388,
      "grad_norm": 1.6706494092941284,
      "learning_rate": 1.4177562293844138e-05,
      "loss": 0.1824,
      "step": 15160
    },
    {
      "epoch": 1.4919721669002484,
      "grad_norm": 3.6499946117401123,
      "learning_rate": 1.4173588204904027e-05,
      "loss": 0.2299,
      "step": 15170
    },
    {
      "epoch": 1.4929556686582577,
      "grad_norm": 0.6618145704269409,
      "learning_rate": 1.4169614115963917e-05,
      "loss": 0.2916,
      "step": 15180
    },
    {
      "epoch": 1.493939170416267,
      "grad_norm": 1.2981956005096436,
      "learning_rate": 1.4165640027023807e-05,
      "loss": 0.19,
      "step": 15190
    },
    {
      "epoch": 1.4949226721742765,
      "grad_norm": 2.071259021759033,
      "learning_rate": 1.4161665938083695e-05,
      "loss": 0.2375,
      "step": 15200
    },
    {
      "epoch": 1.4959061739322859,
      "grad_norm": 0.239904522895813,
      "learning_rate": 1.4157691849143584e-05,
      "loss": 0.2239,
      "step": 15210
    },
    {
      "epoch": 1.4968896756902952,
      "grad_norm": 1.6176055669784546,
      "learning_rate": 1.4153717760203474e-05,
      "loss": 0.2425,
      "step": 15220
    },
    {
      "epoch": 1.4978731774483047,
      "grad_norm": 1.9910478591918945,
      "learning_rate": 1.4149743671263365e-05,
      "loss": 0.2636,
      "step": 15230
    },
    {
      "epoch": 1.498856679206314,
      "grad_norm": 2.132188081741333,
      "learning_rate": 1.4145769582323255e-05,
      "loss": 0.1761,
      "step": 15240
    },
    {
      "epoch": 1.4998401809643234,
      "grad_norm": 0.621258020401001,
      "learning_rate": 1.4141795493383143e-05,
      "loss": 0.2084,
      "step": 15250
    },
    {
      "epoch": 1.500823682722333,
      "grad_norm": 0.6594855189323425,
      "learning_rate": 1.4137821404443033e-05,
      "loss": 0.131,
      "step": 15260
    },
    {
      "epoch": 1.5018071844803422,
      "grad_norm": 2.1136739253997803,
      "learning_rate": 1.4133847315502922e-05,
      "loss": 0.2408,
      "step": 15270
    },
    {
      "epoch": 1.5027906862383515,
      "grad_norm": 0.18802011013031006,
      "learning_rate": 1.4129873226562812e-05,
      "loss": 0.1964,
      "step": 15280
    },
    {
      "epoch": 1.503774187996361,
      "grad_norm": 1.3580917119979858,
      "learning_rate": 1.41258991376227e-05,
      "loss": 0.3299,
      "step": 15290
    },
    {
      "epoch": 1.5047576897543704,
      "grad_norm": 0.8832782506942749,
      "learning_rate": 1.412192504868259e-05,
      "loss": 0.1468,
      "step": 15300
    },
    {
      "epoch": 1.5057411915123797,
      "grad_norm": 1.706316590309143,
      "learning_rate": 1.411795095974248e-05,
      "loss": 0.2762,
      "step": 15310
    },
    {
      "epoch": 1.5067246932703893,
      "grad_norm": 0.9284764528274536,
      "learning_rate": 1.411397687080237e-05,
      "loss": 0.2199,
      "step": 15320
    },
    {
      "epoch": 1.5077081950283986,
      "grad_norm": 2.1995420455932617,
      "learning_rate": 1.411000278186226e-05,
      "loss": 0.2495,
      "step": 15330
    },
    {
      "epoch": 1.508691696786408,
      "grad_norm": 0.9569781422615051,
      "learning_rate": 1.4106028692922148e-05,
      "loss": 0.2262,
      "step": 15340
    },
    {
      "epoch": 1.5096751985444175,
      "grad_norm": 2.7542948722839355,
      "learning_rate": 1.4102054603982038e-05,
      "loss": 0.3087,
      "step": 15350
    },
    {
      "epoch": 1.5106587003024268,
      "grad_norm": 0.3515753746032715,
      "learning_rate": 1.4098080515041928e-05,
      "loss": 0.1827,
      "step": 15360
    },
    {
      "epoch": 1.511642202060436,
      "grad_norm": 0.8161625862121582,
      "learning_rate": 1.4094106426101817e-05,
      "loss": 0.2475,
      "step": 15370
    },
    {
      "epoch": 1.5126257038184456,
      "grad_norm": 1.6958364248275757,
      "learning_rate": 1.4090132337161705e-05,
      "loss": 0.1875,
      "step": 15380
    },
    {
      "epoch": 1.513609205576455,
      "grad_norm": 2.8651888370513916,
      "learning_rate": 1.4086158248221595e-05,
      "loss": 0.2401,
      "step": 15390
    },
    {
      "epoch": 1.5145927073344643,
      "grad_norm": 0.4766581058502197,
      "learning_rate": 1.4082184159281486e-05,
      "loss": 0.2394,
      "step": 15400
    },
    {
      "epoch": 1.5155762090924738,
      "grad_norm": 2.744847059249878,
      "learning_rate": 1.4078210070341376e-05,
      "loss": 0.2546,
      "step": 15410
    },
    {
      "epoch": 1.5165597108504831,
      "grad_norm": 1.2522701025009155,
      "learning_rate": 1.4074235981401266e-05,
      "loss": 0.2419,
      "step": 15420
    },
    {
      "epoch": 1.5175432126084925,
      "grad_norm": 0.5592695474624634,
      "learning_rate": 1.4070261892461155e-05,
      "loss": 0.2088,
      "step": 15430
    },
    {
      "epoch": 1.518526714366502,
      "grad_norm": 0.7614436745643616,
      "learning_rate": 1.4066287803521043e-05,
      "loss": 0.2493,
      "step": 15440
    },
    {
      "epoch": 1.5195102161245113,
      "grad_norm": 3.61337947845459,
      "learning_rate": 1.4062313714580933e-05,
      "loss": 0.3177,
      "step": 15450
    },
    {
      "epoch": 1.5204937178825206,
      "grad_norm": 1.7841546535491943,
      "learning_rate": 1.4058339625640823e-05,
      "loss": 0.316,
      "step": 15460
    },
    {
      "epoch": 1.5214772196405302,
      "grad_norm": 0.8750028610229492,
      "learning_rate": 1.4054365536700712e-05,
      "loss": 0.2631,
      "step": 15470
    },
    {
      "epoch": 1.5224607213985395,
      "grad_norm": 1.0509700775146484,
      "learning_rate": 1.4050391447760604e-05,
      "loss": 0.1831,
      "step": 15480
    },
    {
      "epoch": 1.5234442231565488,
      "grad_norm": 1.5600563287734985,
      "learning_rate": 1.4046417358820492e-05,
      "loss": 0.1879,
      "step": 15490
    },
    {
      "epoch": 1.5244277249145584,
      "grad_norm": 1.9055111408233643,
      "learning_rate": 1.4042443269880381e-05,
      "loss": 0.2838,
      "step": 15500
    },
    {
      "epoch": 1.5244277249145584,
      "eval_loss": 0.15059475600719452,
      "eval_runtime": 18.3111,
      "eval_samples_per_second": 2.731,
      "eval_steps_per_second": 1.365,
      "step": 15500
    },
    {
      "epoch": 1.5254112266725677,
      "grad_norm": 2.142474412918091,
      "learning_rate": 1.4038469180940271e-05,
      "loss": 0.1434,
      "step": 15510
    },
    {
      "epoch": 1.526394728430577,
      "grad_norm": 1.5137566328048706,
      "learning_rate": 1.403449509200016e-05,
      "loss": 0.1432,
      "step": 15520
    },
    {
      "epoch": 1.5273782301885865,
      "grad_norm": 0.8162533044815063,
      "learning_rate": 1.4030521003060049e-05,
      "loss": 0.1593,
      "step": 15530
    },
    {
      "epoch": 1.5283617319465959,
      "grad_norm": 2.5503768920898438,
      "learning_rate": 1.4026546914119938e-05,
      "loss": 0.2212,
      "step": 15540
    },
    {
      "epoch": 1.5293452337046052,
      "grad_norm": 1.5552313327789307,
      "learning_rate": 1.4022572825179828e-05,
      "loss": 0.2256,
      "step": 15550
    },
    {
      "epoch": 1.5303287354626147,
      "grad_norm": 0.6033895611763,
      "learning_rate": 1.401859873623972e-05,
      "loss": 0.1807,
      "step": 15560
    },
    {
      "epoch": 1.531312237220624,
      "grad_norm": 2.3704938888549805,
      "learning_rate": 1.4014624647299609e-05,
      "loss": 0.336,
      "step": 15570
    },
    {
      "epoch": 1.5322957389786334,
      "grad_norm": 0.3030024468898773,
      "learning_rate": 1.4010650558359497e-05,
      "loss": 0.2108,
      "step": 15580
    },
    {
      "epoch": 1.533279240736643,
      "grad_norm": 0.4000900089740753,
      "learning_rate": 1.4006676469419387e-05,
      "loss": 0.195,
      "step": 15590
    },
    {
      "epoch": 1.5342627424946522,
      "grad_norm": 2.0329573154449463,
      "learning_rate": 1.4002702380479276e-05,
      "loss": 0.2763,
      "step": 15600
    },
    {
      "epoch": 1.5352462442526615,
      "grad_norm": 0.7310929298400879,
      "learning_rate": 1.3998728291539166e-05,
      "loss": 0.2881,
      "step": 15610
    },
    {
      "epoch": 1.536229746010671,
      "grad_norm": 1.0867869853973389,
      "learning_rate": 1.3994754202599054e-05,
      "loss": 0.3376,
      "step": 15620
    },
    {
      "epoch": 1.5372132477686804,
      "grad_norm": 1.6135231256484985,
      "learning_rate": 1.3990780113658943e-05,
      "loss": 0.1918,
      "step": 15630
    },
    {
      "epoch": 1.5381967495266897,
      "grad_norm": 1.075868844985962,
      "learning_rate": 1.3986806024718835e-05,
      "loss": 0.3447,
      "step": 15640
    },
    {
      "epoch": 1.5391802512846993,
      "grad_norm": 0.5639548301696777,
      "learning_rate": 1.3982831935778724e-05,
      "loss": 0.2486,
      "step": 15650
    },
    {
      "epoch": 1.5401637530427086,
      "grad_norm": 0.9546343684196472,
      "learning_rate": 1.3978857846838614e-05,
      "loss": 0.2489,
      "step": 15660
    },
    {
      "epoch": 1.541147254800718,
      "grad_norm": 0.4168679118156433,
      "learning_rate": 1.3974883757898504e-05,
      "loss": 0.2325,
      "step": 15670
    },
    {
      "epoch": 1.5421307565587274,
      "grad_norm": 0.40552496910095215,
      "learning_rate": 1.3970909668958392e-05,
      "loss": 0.2235,
      "step": 15680
    },
    {
      "epoch": 1.5431142583167368,
      "grad_norm": 2.4377217292785645,
      "learning_rate": 1.3966935580018281e-05,
      "loss": 0.2474,
      "step": 15690
    },
    {
      "epoch": 1.544097760074746,
      "grad_norm": 1.058454990386963,
      "learning_rate": 1.3962961491078171e-05,
      "loss": 0.3348,
      "step": 15700
    },
    {
      "epoch": 1.5450812618327556,
      "grad_norm": 2.9664440155029297,
      "learning_rate": 1.395898740213806e-05,
      "loss": 0.2238,
      "step": 15710
    },
    {
      "epoch": 1.546064763590765,
      "grad_norm": 1.4083367586135864,
      "learning_rate": 1.3955013313197952e-05,
      "loss": 0.2907,
      "step": 15720
    },
    {
      "epoch": 1.5470482653487743,
      "grad_norm": 1.7369105815887451,
      "learning_rate": 1.395103922425784e-05,
      "loss": 0.2887,
      "step": 15730
    },
    {
      "epoch": 1.5480317671067838,
      "grad_norm": 1.3579238653182983,
      "learning_rate": 1.394706513531773e-05,
      "loss": 0.3529,
      "step": 15740
    },
    {
      "epoch": 1.5490152688647931,
      "grad_norm": 1.0139018297195435,
      "learning_rate": 1.394309104637762e-05,
      "loss": 0.1786,
      "step": 15750
    },
    {
      "epoch": 1.5499987706228024,
      "grad_norm": 2.8200855255126953,
      "learning_rate": 1.3939116957437509e-05,
      "loss": 0.2144,
      "step": 15760
    },
    {
      "epoch": 1.550982272380812,
      "grad_norm": 1.6889891624450684,
      "learning_rate": 1.3935142868497397e-05,
      "loss": 0.2217,
      "step": 15770
    },
    {
      "epoch": 1.5519657741388213,
      "grad_norm": 0.8183413743972778,
      "learning_rate": 1.3931168779557287e-05,
      "loss": 0.2242,
      "step": 15780
    },
    {
      "epoch": 1.5529492758968306,
      "grad_norm": 1.7975190877914429,
      "learning_rate": 1.3927194690617176e-05,
      "loss": 0.2008,
      "step": 15790
    },
    {
      "epoch": 1.5539327776548402,
      "grad_norm": 3.9404609203338623,
      "learning_rate": 1.3923220601677068e-05,
      "loss": 0.3022,
      "step": 15800
    },
    {
      "epoch": 1.5549162794128495,
      "grad_norm": 1.471698522567749,
      "learning_rate": 1.3919246512736957e-05,
      "loss": 0.2656,
      "step": 15810
    },
    {
      "epoch": 1.5558997811708588,
      "grad_norm": 2.58359956741333,
      "learning_rate": 1.3915272423796845e-05,
      "loss": 0.2049,
      "step": 15820
    },
    {
      "epoch": 1.5568832829288684,
      "grad_norm": 1.1108944416046143,
      "learning_rate": 1.3911298334856735e-05,
      "loss": 0.3153,
      "step": 15830
    },
    {
      "epoch": 1.5578667846868777,
      "grad_norm": 2.1699142456054688,
      "learning_rate": 1.3907324245916625e-05,
      "loss": 0.1621,
      "step": 15840
    },
    {
      "epoch": 1.558850286444887,
      "grad_norm": 4.686306476593018,
      "learning_rate": 1.3903350156976514e-05,
      "loss": 0.2476,
      "step": 15850
    },
    {
      "epoch": 1.5598337882028965,
      "grad_norm": 0.7942647933959961,
      "learning_rate": 1.3899376068036402e-05,
      "loss": 0.315,
      "step": 15860
    },
    {
      "epoch": 1.5608172899609059,
      "grad_norm": 3.1741652488708496,
      "learning_rate": 1.3895401979096292e-05,
      "loss": 0.1179,
      "step": 15870
    },
    {
      "epoch": 1.5618007917189152,
      "grad_norm": 0.33635735511779785,
      "learning_rate": 1.3891427890156183e-05,
      "loss": 0.2017,
      "step": 15880
    },
    {
      "epoch": 1.5627842934769247,
      "grad_norm": 2.7269420623779297,
      "learning_rate": 1.3887453801216073e-05,
      "loss": 0.2831,
      "step": 15890
    },
    {
      "epoch": 1.563767795234934,
      "grad_norm": 0.5738785862922668,
      "learning_rate": 1.3883479712275963e-05,
      "loss": 0.22,
      "step": 15900
    },
    {
      "epoch": 1.5647512969929434,
      "grad_norm": 0.5138858556747437,
      "learning_rate": 1.3879505623335852e-05,
      "loss": 0.253,
      "step": 15910
    },
    {
      "epoch": 1.565734798750953,
      "grad_norm": 0.9820569753646851,
      "learning_rate": 1.387553153439574e-05,
      "loss": 0.2013,
      "step": 15920
    },
    {
      "epoch": 1.5667183005089622,
      "grad_norm": 2.7193431854248047,
      "learning_rate": 1.387155744545563e-05,
      "loss": 0.1752,
      "step": 15930
    },
    {
      "epoch": 1.5677018022669715,
      "grad_norm": 1.6706477403640747,
      "learning_rate": 1.386758335651552e-05,
      "loss": 0.2432,
      "step": 15940
    },
    {
      "epoch": 1.568685304024981,
      "grad_norm": 2.1528844833374023,
      "learning_rate": 1.386360926757541e-05,
      "loss": 0.158,
      "step": 15950
    },
    {
      "epoch": 1.5696688057829902,
      "grad_norm": 2.0076661109924316,
      "learning_rate": 1.38596351786353e-05,
      "loss": 0.3423,
      "step": 15960
    },
    {
      "epoch": 1.5706523075409997,
      "grad_norm": 1.225509524345398,
      "learning_rate": 1.3855661089695189e-05,
      "loss": 0.2511,
      "step": 15970
    },
    {
      "epoch": 1.5716358092990093,
      "grad_norm": 2.767773389816284,
      "learning_rate": 1.3851687000755078e-05,
      "loss": 0.217,
      "step": 15980
    },
    {
      "epoch": 1.5726193110570184,
      "grad_norm": 0.551873505115509,
      "learning_rate": 1.3847712911814968e-05,
      "loss": 0.1632,
      "step": 15990
    },
    {
      "epoch": 1.573602812815028,
      "grad_norm": 1.119555115699768,
      "learning_rate": 1.3843738822874858e-05,
      "loss": 0.2923,
      "step": 16000
    },
    {
      "epoch": 1.573602812815028,
      "eval_loss": 0.1566813588142395,
      "eval_runtime": 20.9844,
      "eval_samples_per_second": 2.383,
      "eval_steps_per_second": 1.191,
      "step": 16000
    },
    {
      "epoch": 1.5745863145730374,
      "grad_norm": 0.5722016096115112,
      "learning_rate": 1.3839764733934746e-05,
      "loss": 0.34,
      "step": 16010
    },
    {
      "epoch": 1.5755698163310465,
      "grad_norm": 2.6336591243743896,
      "learning_rate": 1.3835790644994635e-05,
      "loss": 0.1729,
      "step": 16020
    },
    {
      "epoch": 1.576553318089056,
      "grad_norm": 1.409031629562378,
      "learning_rate": 1.3831816556054525e-05,
      "loss": 0.246,
      "step": 16030
    },
    {
      "epoch": 1.5775368198470656,
      "grad_norm": 0.5853147506713867,
      "learning_rate": 1.3827842467114416e-05,
      "loss": 0.1857,
      "step": 16040
    },
    {
      "epoch": 1.5785203216050747,
      "grad_norm": 0.7682684063911438,
      "learning_rate": 1.3823868378174306e-05,
      "loss": 0.2208,
      "step": 16050
    },
    {
      "epoch": 1.5795038233630843,
      "grad_norm": 0.4091372489929199,
      "learning_rate": 1.3819894289234194e-05,
      "loss": 0.3363,
      "step": 16060
    },
    {
      "epoch": 1.5804873251210938,
      "grad_norm": 0.44479209184646606,
      "learning_rate": 1.3815920200294084e-05,
      "loss": 0.2537,
      "step": 16070
    },
    {
      "epoch": 1.581470826879103,
      "grad_norm": 1.0543726682662964,
      "learning_rate": 1.3811946111353973e-05,
      "loss": 0.2711,
      "step": 16080
    },
    {
      "epoch": 1.5824543286371124,
      "grad_norm": 0.8796023726463318,
      "learning_rate": 1.3807972022413863e-05,
      "loss": 0.2295,
      "step": 16090
    },
    {
      "epoch": 1.583437830395122,
      "grad_norm": 0.8024858832359314,
      "learning_rate": 1.3803997933473751e-05,
      "loss": 0.1683,
      "step": 16100
    },
    {
      "epoch": 1.584421332153131,
      "grad_norm": 1.081507921218872,
      "learning_rate": 1.380002384453364e-05,
      "loss": 0.1549,
      "step": 16110
    },
    {
      "epoch": 1.5854048339111406,
      "grad_norm": 1.3907278776168823,
      "learning_rate": 1.379604975559353e-05,
      "loss": 0.2263,
      "step": 16120
    },
    {
      "epoch": 1.5863883356691502,
      "grad_norm": 1.305191159248352,
      "learning_rate": 1.3792075666653422e-05,
      "loss": 0.1947,
      "step": 16130
    },
    {
      "epoch": 1.5873718374271593,
      "grad_norm": 1.4547632932662964,
      "learning_rate": 1.3788101577713311e-05,
      "loss": 0.204,
      "step": 16140
    },
    {
      "epoch": 1.5883553391851688,
      "grad_norm": 0.6436703205108643,
      "learning_rate": 1.3784127488773201e-05,
      "loss": 0.1092,
      "step": 16150
    },
    {
      "epoch": 1.5893388409431783,
      "grad_norm": 0.6339607834815979,
      "learning_rate": 1.3780153399833089e-05,
      "loss": 0.3019,
      "step": 16160
    },
    {
      "epoch": 1.5903223427011874,
      "grad_norm": 2.8297510147094727,
      "learning_rate": 1.3776179310892979e-05,
      "loss": 0.275,
      "step": 16170
    },
    {
      "epoch": 1.591305844459197,
      "grad_norm": 1.2241812944412231,
      "learning_rate": 1.3772205221952868e-05,
      "loss": 0.2319,
      "step": 16180
    },
    {
      "epoch": 1.5922893462172065,
      "grad_norm": 1.707702875137329,
      "learning_rate": 1.3768231133012758e-05,
      "loss": 0.2802,
      "step": 16190
    },
    {
      "epoch": 1.5932728479752156,
      "grad_norm": 1.518728494644165,
      "learning_rate": 1.3764257044072646e-05,
      "loss": 0.2386,
      "step": 16200
    },
    {
      "epoch": 1.5942563497332252,
      "grad_norm": 0.9995889067649841,
      "learning_rate": 1.3760282955132537e-05,
      "loss": 0.1726,
      "step": 16210
    },
    {
      "epoch": 1.5952398514912347,
      "grad_norm": 1.0202362537384033,
      "learning_rate": 1.3756308866192427e-05,
      "loss": 0.2434,
      "step": 16220
    },
    {
      "epoch": 1.5962233532492438,
      "grad_norm": 0.6778249144554138,
      "learning_rate": 1.3752334777252317e-05,
      "loss": 0.2346,
      "step": 16230
    },
    {
      "epoch": 1.5972068550072533,
      "grad_norm": 2.993239164352417,
      "learning_rate": 1.3748360688312206e-05,
      "loss": 0.1847,
      "step": 16240
    },
    {
      "epoch": 1.5981903567652627,
      "grad_norm": 0.5373549461364746,
      "learning_rate": 1.3744386599372094e-05,
      "loss": 0.2517,
      "step": 16250
    },
    {
      "epoch": 1.599173858523272,
      "grad_norm": 1.3796418905258179,
      "learning_rate": 1.3740412510431984e-05,
      "loss": 0.1498,
      "step": 16260
    },
    {
      "epoch": 1.6001573602812815,
      "grad_norm": 1.9999388456344604,
      "learning_rate": 1.3736438421491874e-05,
      "loss": 0.1945,
      "step": 16270
    },
    {
      "epoch": 1.6011408620392908,
      "grad_norm": 0.6067172884941101,
      "learning_rate": 1.3732464332551763e-05,
      "loss": 0.2447,
      "step": 16280
    },
    {
      "epoch": 1.6021243637973002,
      "grad_norm": 2.563868284225464,
      "learning_rate": 1.3728490243611655e-05,
      "loss": 0.2142,
      "step": 16290
    },
    {
      "epoch": 1.6031078655553097,
      "grad_norm": 2.002779483795166,
      "learning_rate": 1.3724516154671543e-05,
      "loss": 0.2112,
      "step": 16300
    },
    {
      "epoch": 1.604091367313319,
      "grad_norm": 2.3529181480407715,
      "learning_rate": 1.3720542065731432e-05,
      "loss": 0.2633,
      "step": 16310
    },
    {
      "epoch": 1.6050748690713283,
      "grad_norm": 1.228992223739624,
      "learning_rate": 1.3716567976791322e-05,
      "loss": 0.1728,
      "step": 16320
    },
    {
      "epoch": 1.606058370829338,
      "grad_norm": 1.3093105554580688,
      "learning_rate": 1.3712593887851212e-05,
      "loss": 0.1955,
      "step": 16330
    },
    {
      "epoch": 1.6070418725873472,
      "grad_norm": 0.8239392638206482,
      "learning_rate": 1.37086197989111e-05,
      "loss": 0.235,
      "step": 16340
    },
    {
      "epoch": 1.6080253743453565,
      "grad_norm": 1.030205488204956,
      "learning_rate": 1.370464570997099e-05,
      "loss": 0.2271,
      "step": 16350
    },
    {
      "epoch": 1.609008876103366,
      "grad_norm": 0.7238038778305054,
      "learning_rate": 1.3700671621030879e-05,
      "loss": 0.1725,
      "step": 16360
    },
    {
      "epoch": 1.6099923778613754,
      "grad_norm": 0.4815444350242615,
      "learning_rate": 1.369669753209077e-05,
      "loss": 0.2095,
      "step": 16370
    },
    {
      "epoch": 1.6109758796193847,
      "grad_norm": 1.8351434469223022,
      "learning_rate": 1.369272344315066e-05,
      "loss": 0.1796,
      "step": 16380
    },
    {
      "epoch": 1.6119593813773943,
      "grad_norm": 2.1774728298187256,
      "learning_rate": 1.368874935421055e-05,
      "loss": 0.2293,
      "step": 16390
    },
    {
      "epoch": 1.6129428831354036,
      "grad_norm": 0.7291221022605896,
      "learning_rate": 1.3684775265270437e-05,
      "loss": 0.1647,
      "step": 16400
    },
    {
      "epoch": 1.613926384893413,
      "grad_norm": 0.8646822571754456,
      "learning_rate": 1.3680801176330327e-05,
      "loss": 0.2258,
      "step": 16410
    },
    {
      "epoch": 1.6149098866514224,
      "grad_norm": 0.8485159873962402,
      "learning_rate": 1.3676827087390217e-05,
      "loss": 0.2776,
      "step": 16420
    },
    {
      "epoch": 1.6158933884094318,
      "grad_norm": 2.74750018119812,
      "learning_rate": 1.3672852998450106e-05,
      "loss": 0.3018,
      "step": 16430
    },
    {
      "epoch": 1.616876890167441,
      "grad_norm": 1.1387678384780884,
      "learning_rate": 1.3668878909509994e-05,
      "loss": 0.2192,
      "step": 16440
    },
    {
      "epoch": 1.6178603919254506,
      "grad_norm": 0.5467789173126221,
      "learning_rate": 1.3664904820569886e-05,
      "loss": 0.2944,
      "step": 16450
    },
    {
      "epoch": 1.61884389368346,
      "grad_norm": 3.122748851776123,
      "learning_rate": 1.3660930731629775e-05,
      "loss": 0.3162,
      "step": 16460
    },
    {
      "epoch": 1.6198273954414693,
      "grad_norm": 0.3885956108570099,
      "learning_rate": 1.3656956642689665e-05,
      "loss": 0.2615,
      "step": 16470
    },
    {
      "epoch": 1.6208108971994788,
      "grad_norm": 1.7819777727127075,
      "learning_rate": 1.3652982553749555e-05,
      "loss": 0.277,
      "step": 16480
    },
    {
      "epoch": 1.6217943989574881,
      "grad_norm": 1.2958201169967651,
      "learning_rate": 1.3649008464809443e-05,
      "loss": 0.2059,
      "step": 16490
    },
    {
      "epoch": 1.6227779007154974,
      "grad_norm": 1.3207638263702393,
      "learning_rate": 1.3645034375869332e-05,
      "loss": 0.1881,
      "step": 16500
    },
    {
      "epoch": 1.6227779007154974,
      "eval_loss": 0.1571766883134842,
      "eval_runtime": 16.1033,
      "eval_samples_per_second": 3.105,
      "eval_steps_per_second": 1.552,
      "step": 16500
    },
    {
      "epoch": 1.623761402473507,
      "grad_norm": 1.1741260290145874,
      "learning_rate": 1.3641060286929222e-05,
      "loss": 0.3229,
      "step": 16510
    },
    {
      "epoch": 1.6247449042315163,
      "grad_norm": 2.915156364440918,
      "learning_rate": 1.3637086197989112e-05,
      "loss": 0.1764,
      "step": 16520
    },
    {
      "epoch": 1.6257284059895256,
      "grad_norm": 1.0990588665008545,
      "learning_rate": 1.3633112109049003e-05,
      "loss": 0.3262,
      "step": 16530
    },
    {
      "epoch": 1.6267119077475352,
      "grad_norm": 1.178744912147522,
      "learning_rate": 1.3629138020108891e-05,
      "loss": 0.1941,
      "step": 16540
    },
    {
      "epoch": 1.6276954095055445,
      "grad_norm": 1.4768882989883423,
      "learning_rate": 1.362516393116878e-05,
      "loss": 0.2371,
      "step": 16550
    },
    {
      "epoch": 1.6286789112635538,
      "grad_norm": 0.5752533674240112,
      "learning_rate": 1.362118984222867e-05,
      "loss": 0.2492,
      "step": 16560
    },
    {
      "epoch": 1.6296624130215633,
      "grad_norm": 2.9447243213653564,
      "learning_rate": 1.361721575328856e-05,
      "loss": 0.1487,
      "step": 16570
    },
    {
      "epoch": 1.6306459147795727,
      "grad_norm": 1.0001193284988403,
      "learning_rate": 1.3613241664348448e-05,
      "loss": 0.3564,
      "step": 16580
    },
    {
      "epoch": 1.631629416537582,
      "grad_norm": 1.4185397624969482,
      "learning_rate": 1.3609267575408338e-05,
      "loss": 0.2264,
      "step": 16590
    },
    {
      "epoch": 1.6326129182955915,
      "grad_norm": 2.0910747051239014,
      "learning_rate": 1.3605293486468227e-05,
      "loss": 0.1866,
      "step": 16600
    },
    {
      "epoch": 1.6335964200536008,
      "grad_norm": 4.188348293304443,
      "learning_rate": 1.3601319397528119e-05,
      "loss": 0.1481,
      "step": 16610
    },
    {
      "epoch": 1.6345799218116102,
      "grad_norm": 2.285092830657959,
      "learning_rate": 1.3597345308588008e-05,
      "loss": 0.2241,
      "step": 16620
    },
    {
      "epoch": 1.6355634235696197,
      "grad_norm": 0.2983485162258148,
      "learning_rate": 1.3593371219647898e-05,
      "loss": 0.1065,
      "step": 16630
    },
    {
      "epoch": 1.636546925327629,
      "grad_norm": 1.9504050016403198,
      "learning_rate": 1.3589397130707786e-05,
      "loss": 0.1945,
      "step": 16640
    },
    {
      "epoch": 1.6375304270856383,
      "grad_norm": 1.3591973781585693,
      "learning_rate": 1.3585423041767676e-05,
      "loss": 0.26,
      "step": 16650
    },
    {
      "epoch": 1.6385139288436479,
      "grad_norm": 3.9365203380584717,
      "learning_rate": 1.3581448952827565e-05,
      "loss": 0.1286,
      "step": 16660
    },
    {
      "epoch": 1.6394974306016572,
      "grad_norm": 1.2122793197631836,
      "learning_rate": 1.3577474863887455e-05,
      "loss": 0.2407,
      "step": 16670
    },
    {
      "epoch": 1.6404809323596665,
      "grad_norm": 0.9854438304901123,
      "learning_rate": 1.3573500774947343e-05,
      "loss": 0.297,
      "step": 16680
    },
    {
      "epoch": 1.641464434117676,
      "grad_norm": 1.231350302696228,
      "learning_rate": 1.3569526686007234e-05,
      "loss": 0.2017,
      "step": 16690
    },
    {
      "epoch": 1.6424479358756854,
      "grad_norm": 1.2629377841949463,
      "learning_rate": 1.3565552597067124e-05,
      "loss": 0.2411,
      "step": 16700
    },
    {
      "epoch": 1.6434314376336947,
      "grad_norm": 0.4649147391319275,
      "learning_rate": 1.3561578508127014e-05,
      "loss": 0.2721,
      "step": 16710
    },
    {
      "epoch": 1.6444149393917042,
      "grad_norm": 1.1152622699737549,
      "learning_rate": 1.3557604419186903e-05,
      "loss": 0.2632,
      "step": 16720
    },
    {
      "epoch": 1.6453984411497136,
      "grad_norm": 1.209902286529541,
      "learning_rate": 1.3553630330246791e-05,
      "loss": 0.2572,
      "step": 16730
    },
    {
      "epoch": 1.6463819429077229,
      "grad_norm": 2.158609628677368,
      "learning_rate": 1.3549656241306681e-05,
      "loss": 0.2976,
      "step": 16740
    },
    {
      "epoch": 1.6473654446657324,
      "grad_norm": 0.4169705808162689,
      "learning_rate": 1.354568215236657e-05,
      "loss": 0.2098,
      "step": 16750
    },
    {
      "epoch": 1.6483489464237417,
      "grad_norm": 1.0553632974624634,
      "learning_rate": 1.354170806342646e-05,
      "loss": 0.1597,
      "step": 16760
    },
    {
      "epoch": 1.649332448181751,
      "grad_norm": 0.7214099168777466,
      "learning_rate": 1.3537733974486352e-05,
      "loss": 0.2487,
      "step": 16770
    },
    {
      "epoch": 1.6503159499397606,
      "grad_norm": 1.0439274311065674,
      "learning_rate": 1.353375988554624e-05,
      "loss": 0.3632,
      "step": 16780
    },
    {
      "epoch": 1.65129945169777,
      "grad_norm": 0.5576213598251343,
      "learning_rate": 1.352978579660613e-05,
      "loss": 0.1836,
      "step": 16790
    },
    {
      "epoch": 1.6522829534557792,
      "grad_norm": 1.6675044298171997,
      "learning_rate": 1.3525811707666019e-05,
      "loss": 0.2255,
      "step": 16800
    },
    {
      "epoch": 1.6532664552137888,
      "grad_norm": 0.6567625403404236,
      "learning_rate": 1.3521837618725909e-05,
      "loss": 0.2699,
      "step": 16810
    },
    {
      "epoch": 1.654249956971798,
      "grad_norm": 0.30426979064941406,
      "learning_rate": 1.3517863529785797e-05,
      "loss": 0.2056,
      "step": 16820
    },
    {
      "epoch": 1.6552334587298074,
      "grad_norm": 0.6561160683631897,
      "learning_rate": 1.3513889440845686e-05,
      "loss": 0.2077,
      "step": 16830
    },
    {
      "epoch": 1.656216960487817,
      "grad_norm": 0.47024333477020264,
      "learning_rate": 1.3509915351905576e-05,
      "loss": 0.307,
      "step": 16840
    },
    {
      "epoch": 1.6572004622458263,
      "grad_norm": 2.3148345947265625,
      "learning_rate": 1.3505941262965467e-05,
      "loss": 0.2145,
      "step": 16850
    },
    {
      "epoch": 1.6581839640038356,
      "grad_norm": 1.8510692119598389,
      "learning_rate": 1.3501967174025357e-05,
      "loss": 0.2056,
      "step": 16860
    },
    {
      "epoch": 1.6591674657618452,
      "grad_norm": 5.483177661895752,
      "learning_rate": 1.3497993085085247e-05,
      "loss": 0.1826,
      "step": 16870
    },
    {
      "epoch": 1.6601509675198545,
      "grad_norm": 0.8653757572174072,
      "learning_rate": 1.3494018996145135e-05,
      "loss": 0.2596,
      "step": 16880
    },
    {
      "epoch": 1.6611344692778638,
      "grad_norm": 0.8157027363777161,
      "learning_rate": 1.3490044907205024e-05,
      "loss": 0.2814,
      "step": 16890
    },
    {
      "epoch": 1.6621179710358733,
      "grad_norm": 1.144384503364563,
      "learning_rate": 1.3486070818264914e-05,
      "loss": 0.3275,
      "step": 16900
    },
    {
      "epoch": 1.6631014727938827,
      "grad_norm": 3.486330509185791,
      "learning_rate": 1.3482096729324804e-05,
      "loss": 0.2449,
      "step": 16910
    },
    {
      "epoch": 1.664084974551892,
      "grad_norm": 0.7469828724861145,
      "learning_rate": 1.3478122640384692e-05,
      "loss": 0.2374,
      "step": 16920
    },
    {
      "epoch": 1.6650684763099015,
      "grad_norm": 1.1899664402008057,
      "learning_rate": 1.3474148551444581e-05,
      "loss": 0.1911,
      "step": 16930
    },
    {
      "epoch": 1.6660519780679108,
      "grad_norm": 2.471895694732666,
      "learning_rate": 1.3470174462504473e-05,
      "loss": 0.1957,
      "step": 16940
    },
    {
      "epoch": 1.6670354798259202,
      "grad_norm": 1.0681343078613281,
      "learning_rate": 1.3466200373564362e-05,
      "loss": 0.3053,
      "step": 16950
    },
    {
      "epoch": 1.6680189815839297,
      "grad_norm": 2.1743273735046387,
      "learning_rate": 1.3462226284624252e-05,
      "loss": 0.227,
      "step": 16960
    },
    {
      "epoch": 1.669002483341939,
      "grad_norm": 6.08627462387085,
      "learning_rate": 1.345825219568414e-05,
      "loss": 0.2416,
      "step": 16970
    },
    {
      "epoch": 1.6699859850999483,
      "grad_norm": 0.654421329498291,
      "learning_rate": 1.345427810674403e-05,
      "loss": 0.2075,
      "step": 16980
    },
    {
      "epoch": 1.6709694868579579,
      "grad_norm": 0.7476296424865723,
      "learning_rate": 1.345030401780392e-05,
      "loss": 0.1306,
      "step": 16990
    },
    {
      "epoch": 1.6719529886159672,
      "grad_norm": 0.9483456611633301,
      "learning_rate": 1.3446329928863809e-05,
      "loss": 0.3279,
      "step": 17000
    },
    {
      "epoch": 1.6719529886159672,
      "eval_loss": 0.15918737649917603,
      "eval_runtime": 18.2512,
      "eval_samples_per_second": 2.74,
      "eval_steps_per_second": 1.37,
      "step": 17000
    },
    {
      "epoch": 1.6729364903739765,
      "grad_norm": 4.026605606079102,
      "learning_rate": 1.3442355839923697e-05,
      "loss": 0.2603,
      "step": 17010
    },
    {
      "epoch": 1.673919992131986,
      "grad_norm": 1.619104266166687,
      "learning_rate": 1.3438381750983588e-05,
      "loss": 0.2716,
      "step": 17020
    },
    {
      "epoch": 1.6749034938899954,
      "grad_norm": 1.8711702823638916,
      "learning_rate": 1.3434407662043478e-05,
      "loss": 0.2267,
      "step": 17030
    },
    {
      "epoch": 1.6758869956480047,
      "grad_norm": 2.9381566047668457,
      "learning_rate": 1.3430433573103368e-05,
      "loss": 0.2086,
      "step": 17040
    },
    {
      "epoch": 1.6768704974060142,
      "grad_norm": 0.5107515454292297,
      "learning_rate": 1.3426459484163257e-05,
      "loss": 0.1975,
      "step": 17050
    },
    {
      "epoch": 1.6778539991640236,
      "grad_norm": 3.57466197013855,
      "learning_rate": 1.3422485395223145e-05,
      "loss": 0.2691,
      "step": 17060
    },
    {
      "epoch": 1.6788375009220329,
      "grad_norm": 1.05047607421875,
      "learning_rate": 1.3418511306283035e-05,
      "loss": 0.2854,
      "step": 17070
    },
    {
      "epoch": 1.6798210026800424,
      "grad_norm": 1.1247413158416748,
      "learning_rate": 1.3414537217342925e-05,
      "loss": 0.2301,
      "step": 17080
    },
    {
      "epoch": 1.6808045044380515,
      "grad_norm": 3.0839147567749023,
      "learning_rate": 1.3410563128402814e-05,
      "loss": 0.1918,
      "step": 17090
    },
    {
      "epoch": 1.681788006196061,
      "grad_norm": 0.8100689053535461,
      "learning_rate": 1.3406589039462706e-05,
      "loss": 0.307,
      "step": 17100
    },
    {
      "epoch": 1.6827715079540706,
      "grad_norm": 1.0803511142730713,
      "learning_rate": 1.3402614950522595e-05,
      "loss": 0.2355,
      "step": 17110
    },
    {
      "epoch": 1.6837550097120797,
      "grad_norm": 1.2122515439987183,
      "learning_rate": 1.3398640861582483e-05,
      "loss": 0.2274,
      "step": 17120
    },
    {
      "epoch": 1.6847385114700892,
      "grad_norm": 0.907745897769928,
      "learning_rate": 1.3394666772642373e-05,
      "loss": 0.1913,
      "step": 17130
    },
    {
      "epoch": 1.6857220132280988,
      "grad_norm": 1.1718151569366455,
      "learning_rate": 1.3390692683702263e-05,
      "loss": 0.2933,
      "step": 17140
    },
    {
      "epoch": 1.6867055149861079,
      "grad_norm": 0.6171371340751648,
      "learning_rate": 1.3386718594762152e-05,
      "loss": 0.2956,
      "step": 17150
    },
    {
      "epoch": 1.6876890167441174,
      "grad_norm": 1.1524289846420288,
      "learning_rate": 1.338274450582204e-05,
      "loss": 0.2185,
      "step": 17160
    },
    {
      "epoch": 1.688672518502127,
      "grad_norm": 0.5245811939239502,
      "learning_rate": 1.337877041688193e-05,
      "loss": 0.2736,
      "step": 17170
    },
    {
      "epoch": 1.689656020260136,
      "grad_norm": 4.1112470626831055,
      "learning_rate": 1.3374796327941821e-05,
      "loss": 0.2162,
      "step": 17180
    },
    {
      "epoch": 1.6906395220181456,
      "grad_norm": 0.631099283695221,
      "learning_rate": 1.337082223900171e-05,
      "loss": 0.3262,
      "step": 17190
    },
    {
      "epoch": 1.6916230237761551,
      "grad_norm": 0.7816455364227295,
      "learning_rate": 1.33668481500616e-05,
      "loss": 0.2709,
      "step": 17200
    },
    {
      "epoch": 1.6926065255341642,
      "grad_norm": 0.8032369017601013,
      "learning_rate": 1.3362874061121488e-05,
      "loss": 0.1441,
      "step": 17210
    },
    {
      "epoch": 1.6935900272921738,
      "grad_norm": 1.126566767692566,
      "learning_rate": 1.3358899972181378e-05,
      "loss": 0.1997,
      "step": 17220
    },
    {
      "epoch": 1.6945735290501833,
      "grad_norm": 0.9024196267127991,
      "learning_rate": 1.3354925883241268e-05,
      "loss": 0.2046,
      "step": 17230
    },
    {
      "epoch": 1.6955570308081924,
      "grad_norm": 1.4062788486480713,
      "learning_rate": 1.3350951794301157e-05,
      "loss": 0.3639,
      "step": 17240
    },
    {
      "epoch": 1.696540532566202,
      "grad_norm": 2.7009520530700684,
      "learning_rate": 1.3346977705361045e-05,
      "loss": 0.2141,
      "step": 17250
    },
    {
      "epoch": 1.6975240343242115,
      "grad_norm": 1.0145210027694702,
      "learning_rate": 1.3343003616420937e-05,
      "loss": 0.3401,
      "step": 17260
    },
    {
      "epoch": 1.6985075360822206,
      "grad_norm": 0.8520883321762085,
      "learning_rate": 1.3339029527480826e-05,
      "loss": 0.2577,
      "step": 17270
    },
    {
      "epoch": 1.6994910378402301,
      "grad_norm": 0.5788146257400513,
      "learning_rate": 1.3335055438540716e-05,
      "loss": 0.2418,
      "step": 17280
    },
    {
      "epoch": 1.7004745395982397,
      "grad_norm": 1.2726303339004517,
      "learning_rate": 1.3331081349600606e-05,
      "loss": 0.2434,
      "step": 17290
    },
    {
      "epoch": 1.7014580413562488,
      "grad_norm": 1.0344926118850708,
      "learning_rate": 1.3327107260660494e-05,
      "loss": 0.1686,
      "step": 17300
    },
    {
      "epoch": 1.7024415431142583,
      "grad_norm": 1.0950030088424683,
      "learning_rate": 1.3323133171720383e-05,
      "loss": 0.3009,
      "step": 17310
    },
    {
      "epoch": 1.7034250448722679,
      "grad_norm": 0.7477503418922424,
      "learning_rate": 1.3319159082780273e-05,
      "loss": 0.1959,
      "step": 17320
    },
    {
      "epoch": 1.704408546630277,
      "grad_norm": 0.9627076983451843,
      "learning_rate": 1.3315184993840163e-05,
      "loss": 0.2229,
      "step": 17330
    },
    {
      "epoch": 1.7053920483882865,
      "grad_norm": 2.0675864219665527,
      "learning_rate": 1.3311210904900054e-05,
      "loss": 0.1794,
      "step": 17340
    },
    {
      "epoch": 1.706375550146296,
      "grad_norm": 0.5318220257759094,
      "learning_rate": 1.3307236815959944e-05,
      "loss": 0.1252,
      "step": 17350
    },
    {
      "epoch": 1.7073590519043051,
      "grad_norm": 1.1632235050201416,
      "learning_rate": 1.3303262727019832e-05,
      "loss": 0.2574,
      "step": 17360
    },
    {
      "epoch": 1.7083425536623147,
      "grad_norm": 0.5292351245880127,
      "learning_rate": 1.3299288638079721e-05,
      "loss": 0.2551,
      "step": 17370
    },
    {
      "epoch": 1.709326055420324,
      "grad_norm": 0.38190001249313354,
      "learning_rate": 1.3295314549139611e-05,
      "loss": 0.2473,
      "step": 17380
    },
    {
      "epoch": 1.7103095571783333,
      "grad_norm": 5.364218711853027,
      "learning_rate": 1.32913404601995e-05,
      "loss": 0.3242,
      "step": 17390
    },
    {
      "epoch": 1.7112930589363429,
      "grad_norm": 1.7767460346221924,
      "learning_rate": 1.3287366371259389e-05,
      "loss": 0.2854,
      "step": 17400
    },
    {
      "epoch": 1.7122765606943522,
      "grad_norm": 0.7090295553207397,
      "learning_rate": 1.3283392282319278e-05,
      "loss": 0.1963,
      "step": 17410
    },
    {
      "epoch": 1.7132600624523615,
      "grad_norm": 0.6008509993553162,
      "learning_rate": 1.327941819337917e-05,
      "loss": 0.3349,
      "step": 17420
    },
    {
      "epoch": 1.714243564210371,
      "grad_norm": 1.0188231468200684,
      "learning_rate": 1.327544410443906e-05,
      "loss": 0.2329,
      "step": 17430
    },
    {
      "epoch": 1.7152270659683804,
      "grad_norm": 2.4888973236083984,
      "learning_rate": 1.3271470015498949e-05,
      "loss": 0.2454,
      "step": 17440
    },
    {
      "epoch": 1.7162105677263897,
      "grad_norm": 1.0194109678268433,
      "learning_rate": 1.3267495926558837e-05,
      "loss": 0.1806,
      "step": 17450
    },
    {
      "epoch": 1.7171940694843992,
      "grad_norm": 1.178078532218933,
      "learning_rate": 1.3263521837618727e-05,
      "loss": 0.3933,
      "step": 17460
    },
    {
      "epoch": 1.7181775712424086,
      "grad_norm": 1.6931121349334717,
      "learning_rate": 1.3259547748678616e-05,
      "loss": 0.2191,
      "step": 17470
    },
    {
      "epoch": 1.7191610730004179,
      "grad_norm": 2.874523639678955,
      "learning_rate": 1.3255573659738506e-05,
      "loss": 0.3308,
      "step": 17480
    },
    {
      "epoch": 1.7201445747584274,
      "grad_norm": 3.4492263793945312,
      "learning_rate": 1.3251599570798394e-05,
      "loss": 0.1748,
      "step": 17490
    },
    {
      "epoch": 1.7211280765164367,
      "grad_norm": 1.1551506519317627,
      "learning_rate": 1.3247625481858285e-05,
      "loss": 0.2132,
      "step": 17500
    },
    {
      "epoch": 1.7211280765164367,
      "eval_loss": 0.15584313869476318,
      "eval_runtime": 17.0322,
      "eval_samples_per_second": 2.936,
      "eval_steps_per_second": 1.468,
      "step": 17500
    },
    {
      "epoch": 1.722111578274446,
      "grad_norm": 3.204711437225342,
      "learning_rate": 1.3243651392918175e-05,
      "loss": 0.2223,
      "step": 17510
    },
    {
      "epoch": 1.7230950800324556,
      "grad_norm": 1.4167150259017944,
      "learning_rate": 1.3239677303978065e-05,
      "loss": 0.2441,
      "step": 17520
    },
    {
      "epoch": 1.724078581790465,
      "grad_norm": 2.277780294418335,
      "learning_rate": 1.3235703215037954e-05,
      "loss": 0.2682,
      "step": 17530
    },
    {
      "epoch": 1.7250620835484742,
      "grad_norm": 0.6956846714019775,
      "learning_rate": 1.3231729126097842e-05,
      "loss": 0.16,
      "step": 17540
    },
    {
      "epoch": 1.7260455853064838,
      "grad_norm": 0.8625393509864807,
      "learning_rate": 1.3227755037157732e-05,
      "loss": 0.2059,
      "step": 17550
    },
    {
      "epoch": 1.727029087064493,
      "grad_norm": 0.696932852268219,
      "learning_rate": 1.3223780948217622e-05,
      "loss": 0.3554,
      "step": 17560
    },
    {
      "epoch": 1.7280125888225024,
      "grad_norm": 3.09187388420105,
      "learning_rate": 1.3219806859277511e-05,
      "loss": 0.197,
      "step": 17570
    },
    {
      "epoch": 1.728996090580512,
      "grad_norm": 2.1331708431243896,
      "learning_rate": 1.3215832770337403e-05,
      "loss": 0.2262,
      "step": 17580
    },
    {
      "epoch": 1.7299795923385213,
      "grad_norm": 2.1550347805023193,
      "learning_rate": 1.3211858681397292e-05,
      "loss": 0.3214,
      "step": 17590
    },
    {
      "epoch": 1.7309630940965306,
      "grad_norm": 3.0045039653778076,
      "learning_rate": 1.320788459245718e-05,
      "loss": 0.1884,
      "step": 17600
    },
    {
      "epoch": 1.7319465958545401,
      "grad_norm": 1.457900047302246,
      "learning_rate": 1.320391050351707e-05,
      "loss": 0.2306,
      "step": 17610
    },
    {
      "epoch": 1.7329300976125495,
      "grad_norm": 1.152354121208191,
      "learning_rate": 1.319993641457696e-05,
      "loss": 0.2643,
      "step": 17620
    },
    {
      "epoch": 1.7339135993705588,
      "grad_norm": 0.5737890601158142,
      "learning_rate": 1.319596232563685e-05,
      "loss": 0.2697,
      "step": 17630
    },
    {
      "epoch": 1.7348971011285683,
      "grad_norm": 0.5850633978843689,
      "learning_rate": 1.3191988236696737e-05,
      "loss": 0.2399,
      "step": 17640
    },
    {
      "epoch": 1.7358806028865776,
      "grad_norm": 2.2772579193115234,
      "learning_rate": 1.3188014147756627e-05,
      "loss": 0.1175,
      "step": 17650
    },
    {
      "epoch": 1.736864104644587,
      "grad_norm": 2.7282495498657227,
      "learning_rate": 1.3184040058816518e-05,
      "loss": 0.2286,
      "step": 17660
    },
    {
      "epoch": 1.7378476064025965,
      "grad_norm": 1.0198181867599487,
      "learning_rate": 1.3180065969876408e-05,
      "loss": 0.2325,
      "step": 17670
    },
    {
      "epoch": 1.7388311081606058,
      "grad_norm": 1.8783186674118042,
      "learning_rate": 1.3176091880936298e-05,
      "loss": 0.1183,
      "step": 17680
    },
    {
      "epoch": 1.7398146099186151,
      "grad_norm": 2.2488327026367188,
      "learning_rate": 1.3172117791996186e-05,
      "loss": 0.2618,
      "step": 17690
    },
    {
      "epoch": 1.7407981116766247,
      "grad_norm": 5.103484630584717,
      "learning_rate": 1.3168143703056075e-05,
      "loss": 0.2685,
      "step": 17700
    },
    {
      "epoch": 1.741781613434634,
      "grad_norm": 0.5979586839675903,
      "learning_rate": 1.3164169614115965e-05,
      "loss": 0.2041,
      "step": 17710
    },
    {
      "epoch": 1.7427651151926433,
      "grad_norm": 3.733778238296509,
      "learning_rate": 1.3160195525175855e-05,
      "loss": 0.1391,
      "step": 17720
    },
    {
      "epoch": 1.7437486169506529,
      "grad_norm": 1.909714698791504,
      "learning_rate": 1.3156221436235743e-05,
      "loss": 0.235,
      "step": 17730
    },
    {
      "epoch": 1.7447321187086622,
      "grad_norm": 1.2753373384475708,
      "learning_rate": 1.3152247347295632e-05,
      "loss": 0.2916,
      "step": 17740
    },
    {
      "epoch": 1.7457156204666715,
      "grad_norm": 1.276990294456482,
      "learning_rate": 1.3148273258355524e-05,
      "loss": 0.2657,
      "step": 17750
    },
    {
      "epoch": 1.746699122224681,
      "grad_norm": 2.39896297454834,
      "learning_rate": 1.3144299169415413e-05,
      "loss": 0.2799,
      "step": 17760
    },
    {
      "epoch": 1.7476826239826904,
      "grad_norm": 0.5444474220275879,
      "learning_rate": 1.3140325080475303e-05,
      "loss": 0.1874,
      "step": 17770
    },
    {
      "epoch": 1.7486661257406997,
      "grad_norm": 1.2919403314590454,
      "learning_rate": 1.3136350991535191e-05,
      "loss": 0.2485,
      "step": 17780
    },
    {
      "epoch": 1.7496496274987092,
      "grad_norm": 0.9817249774932861,
      "learning_rate": 1.313237690259508e-05,
      "loss": 0.1711,
      "step": 17790
    },
    {
      "epoch": 1.7506331292567185,
      "grad_norm": 0.4421529769897461,
      "learning_rate": 1.312840281365497e-05,
      "loss": 0.1969,
      "step": 17800
    },
    {
      "epoch": 1.7516166310147279,
      "grad_norm": 1.9339884519577026,
      "learning_rate": 1.312442872471486e-05,
      "loss": 0.3163,
      "step": 17810
    },
    {
      "epoch": 1.7526001327727374,
      "grad_norm": 0.7091568112373352,
      "learning_rate": 1.3120454635774748e-05,
      "loss": 0.2134,
      "step": 17820
    },
    {
      "epoch": 1.7535836345307467,
      "grad_norm": 1.4017484188079834,
      "learning_rate": 1.3116480546834641e-05,
      "loss": 0.206,
      "step": 17830
    },
    {
      "epoch": 1.754567136288756,
      "grad_norm": 0.38947057723999023,
      "learning_rate": 1.3112506457894529e-05,
      "loss": 0.3262,
      "step": 17840
    },
    {
      "epoch": 1.7555506380467656,
      "grad_norm": 1.5356085300445557,
      "learning_rate": 1.3108532368954419e-05,
      "loss": 0.3551,
      "step": 17850
    },
    {
      "epoch": 1.756534139804775,
      "grad_norm": 2.8606903553009033,
      "learning_rate": 1.3104558280014308e-05,
      "loss": 0.2159,
      "step": 17860
    },
    {
      "epoch": 1.7575176415627842,
      "grad_norm": 2.0851094722747803,
      "learning_rate": 1.3100584191074198e-05,
      "loss": 0.1928,
      "step": 17870
    },
    {
      "epoch": 1.7585011433207938,
      "grad_norm": 0.6435067057609558,
      "learning_rate": 1.3096610102134086e-05,
      "loss": 0.2425,
      "step": 17880
    },
    {
      "epoch": 1.759484645078803,
      "grad_norm": 0.6339534521102905,
      "learning_rate": 1.3092636013193976e-05,
      "loss": 0.1548,
      "step": 17890
    },
    {
      "epoch": 1.7604681468368124,
      "grad_norm": 2.1207547187805176,
      "learning_rate": 1.3088661924253865e-05,
      "loss": 0.2594,
      "step": 17900
    },
    {
      "epoch": 1.761451648594822,
      "grad_norm": 0.6143303513526917,
      "learning_rate": 1.3084687835313757e-05,
      "loss": 0.2998,
      "step": 17910
    },
    {
      "epoch": 1.7624351503528313,
      "grad_norm": 0.7414078712463379,
      "learning_rate": 1.3080713746373646e-05,
      "loss": 0.1891,
      "step": 17920
    },
    {
      "epoch": 1.7634186521108406,
      "grad_norm": 1.9617842435836792,
      "learning_rate": 1.3076739657433534e-05,
      "loss": 0.297,
      "step": 17930
    },
    {
      "epoch": 1.7644021538688501,
      "grad_norm": 2.322709321975708,
      "learning_rate": 1.3072765568493424e-05,
      "loss": 0.2052,
      "step": 17940
    },
    {
      "epoch": 1.7653856556268595,
      "grad_norm": 1.813644528388977,
      "learning_rate": 1.3068791479553314e-05,
      "loss": 0.2478,
      "step": 17950
    },
    {
      "epoch": 1.7663691573848688,
      "grad_norm": 1.4270058870315552,
      "learning_rate": 1.3064817390613203e-05,
      "loss": 0.124,
      "step": 17960
    },
    {
      "epoch": 1.7673526591428783,
      "grad_norm": 2.472296953201294,
      "learning_rate": 1.3060843301673091e-05,
      "loss": 0.2331,
      "step": 17970
    },
    {
      "epoch": 1.7683361609008876,
      "grad_norm": 2.430737257003784,
      "learning_rate": 1.305686921273298e-05,
      "loss": 0.2501,
      "step": 17980
    },
    {
      "epoch": 1.769319662658897,
      "grad_norm": 0.6140730977058411,
      "learning_rate": 1.3052895123792872e-05,
      "loss": 0.1823,
      "step": 17990
    },
    {
      "epoch": 1.7703031644169065,
      "grad_norm": 4.935455799102783,
      "learning_rate": 1.3048921034852762e-05,
      "loss": 0.2335,
      "step": 18000
    },
    {
      "epoch": 1.7703031644169065,
      "eval_loss": 0.16023403406143188,
      "eval_runtime": 18.1415,
      "eval_samples_per_second": 2.756,
      "eval_steps_per_second": 1.378,
      "step": 18000
    },
    {
      "epoch": 1.7712866661749158,
      "grad_norm": 3.046405076980591,
      "learning_rate": 1.3044946945912651e-05,
      "loss": 0.2097,
      "step": 18010
    },
    {
      "epoch": 1.7722701679329251,
      "grad_norm": 0.8612200021743774,
      "learning_rate": 1.304097285697254e-05,
      "loss": 0.2561,
      "step": 18020
    },
    {
      "epoch": 1.7732536696909347,
      "grad_norm": 1.075387954711914,
      "learning_rate": 1.3036998768032429e-05,
      "loss": 0.2052,
      "step": 18030
    },
    {
      "epoch": 1.774237171448944,
      "grad_norm": 0.7561732530593872,
      "learning_rate": 1.3033024679092319e-05,
      "loss": 0.2157,
      "step": 18040
    },
    {
      "epoch": 1.7752206732069533,
      "grad_norm": 0.7665427923202515,
      "learning_rate": 1.3029050590152208e-05,
      "loss": 0.1883,
      "step": 18050
    },
    {
      "epoch": 1.7762041749649629,
      "grad_norm": 1.8395278453826904,
      "learning_rate": 1.3025076501212096e-05,
      "loss": 0.2507,
      "step": 18060
    },
    {
      "epoch": 1.7771876767229722,
      "grad_norm": 1.2933135032653809,
      "learning_rate": 1.302110241227199e-05,
      "loss": 0.307,
      "step": 18070
    },
    {
      "epoch": 1.7781711784809815,
      "grad_norm": 0.7886130213737488,
      "learning_rate": 1.3017128323331877e-05,
      "loss": 0.2525,
      "step": 18080
    },
    {
      "epoch": 1.779154680238991,
      "grad_norm": 1.028622031211853,
      "learning_rate": 1.3013154234391767e-05,
      "loss": 0.2672,
      "step": 18090
    },
    {
      "epoch": 1.7801381819970004,
      "grad_norm": 0.8745284080505371,
      "learning_rate": 1.3009180145451657e-05,
      "loss": 0.1684,
      "step": 18100
    },
    {
      "epoch": 1.7811216837550097,
      "grad_norm": 0.5309194922447205,
      "learning_rate": 1.3005206056511546e-05,
      "loss": 0.2063,
      "step": 18110
    },
    {
      "epoch": 1.7821051855130192,
      "grad_norm": 1.1381081342697144,
      "learning_rate": 1.3001231967571434e-05,
      "loss": 0.3476,
      "step": 18120
    },
    {
      "epoch": 1.7830886872710285,
      "grad_norm": 0.6130464673042297,
      "learning_rate": 1.2997257878631324e-05,
      "loss": 0.2196,
      "step": 18130
    },
    {
      "epoch": 1.7840721890290379,
      "grad_norm": 0.6083819270133972,
      "learning_rate": 1.2993283789691214e-05,
      "loss": 0.2292,
      "step": 18140
    },
    {
      "epoch": 1.7850556907870474,
      "grad_norm": 0.9610211253166199,
      "learning_rate": 1.2989309700751105e-05,
      "loss": 0.3196,
      "step": 18150
    },
    {
      "epoch": 1.7860391925450567,
      "grad_norm": 4.28626823425293,
      "learning_rate": 1.2985335611810995e-05,
      "loss": 0.1689,
      "step": 18160
    },
    {
      "epoch": 1.787022694303066,
      "grad_norm": 2.249110698699951,
      "learning_rate": 1.2981361522870883e-05,
      "loss": 0.2129,
      "step": 18170
    },
    {
      "epoch": 1.7880061960610756,
      "grad_norm": 0.44461897015571594,
      "learning_rate": 1.2977387433930772e-05,
      "loss": 0.2095,
      "step": 18180
    },
    {
      "epoch": 1.788989697819085,
      "grad_norm": 0.26538947224617004,
      "learning_rate": 1.2973413344990662e-05,
      "loss": 0.2116,
      "step": 18190
    },
    {
      "epoch": 1.7899731995770942,
      "grad_norm": 2.8765032291412354,
      "learning_rate": 1.2969439256050552e-05,
      "loss": 0.2552,
      "step": 18200
    },
    {
      "epoch": 1.7909567013351038,
      "grad_norm": 1.751949667930603,
      "learning_rate": 1.296546516711044e-05,
      "loss": 0.2782,
      "step": 18210
    },
    {
      "epoch": 1.791940203093113,
      "grad_norm": 1.121596336364746,
      "learning_rate": 1.296149107817033e-05,
      "loss": 0.1649,
      "step": 18220
    },
    {
      "epoch": 1.7929237048511224,
      "grad_norm": 1.0512248277664185,
      "learning_rate": 1.295751698923022e-05,
      "loss": 0.3041,
      "step": 18230
    },
    {
      "epoch": 1.793907206609132,
      "grad_norm": 2.1734120845794678,
      "learning_rate": 1.295354290029011e-05,
      "loss": 0.1779,
      "step": 18240
    },
    {
      "epoch": 1.794890708367141,
      "grad_norm": 1.8944523334503174,
      "learning_rate": 1.294956881135e-05,
      "loss": 0.2143,
      "step": 18250
    },
    {
      "epoch": 1.7958742101251506,
      "grad_norm": 4.825395584106445,
      "learning_rate": 1.2945594722409888e-05,
      "loss": 0.2896,
      "step": 18260
    },
    {
      "epoch": 1.7968577118831601,
      "grad_norm": 0.7082745432853699,
      "learning_rate": 1.2941620633469778e-05,
      "loss": 0.234,
      "step": 18270
    },
    {
      "epoch": 1.7978412136411692,
      "grad_norm": 0.8548723459243774,
      "learning_rate": 1.2937646544529667e-05,
      "loss": 0.1127,
      "step": 18280
    },
    {
      "epoch": 1.7988247153991788,
      "grad_norm": 3.628645181655884,
      "learning_rate": 1.2933672455589557e-05,
      "loss": 0.2571,
      "step": 18290
    },
    {
      "epoch": 1.7998082171571883,
      "grad_norm": 0.721328616142273,
      "learning_rate": 1.2929698366649445e-05,
      "loss": 0.3004,
      "step": 18300
    },
    {
      "epoch": 1.8007917189151974,
      "grad_norm": 1.983008623123169,
      "learning_rate": 1.2925724277709338e-05,
      "loss": 0.1744,
      "step": 18310
    },
    {
      "epoch": 1.801775220673207,
      "grad_norm": 1.1194294691085815,
      "learning_rate": 1.2921750188769226e-05,
      "loss": 0.242,
      "step": 18320
    },
    {
      "epoch": 1.8027587224312165,
      "grad_norm": 0.8498597741127014,
      "learning_rate": 1.2917776099829116e-05,
      "loss": 0.1855,
      "step": 18330
    },
    {
      "epoch": 1.8037422241892256,
      "grad_norm": 1.695459246635437,
      "learning_rate": 1.2913802010889005e-05,
      "loss": 0.2498,
      "step": 18340
    },
    {
      "epoch": 1.8047257259472351,
      "grad_norm": 1.5076769590377808,
      "learning_rate": 1.2909827921948895e-05,
      "loss": 0.1511,
      "step": 18350
    },
    {
      "epoch": 1.8057092277052447,
      "grad_norm": 1.4327051639556885,
      "learning_rate": 1.2905853833008783e-05,
      "loss": 0.1891,
      "step": 18360
    },
    {
      "epoch": 1.8066927294632538,
      "grad_norm": 1.8368661403656006,
      "learning_rate": 1.2901879744068673e-05,
      "loss": 0.2104,
      "step": 18370
    },
    {
      "epoch": 1.8076762312212633,
      "grad_norm": 0.7648179531097412,
      "learning_rate": 1.2897905655128562e-05,
      "loss": 0.2628,
      "step": 18380
    },
    {
      "epoch": 1.8086597329792728,
      "grad_norm": 1.534253478050232,
      "learning_rate": 1.2893931566188454e-05,
      "loss": 0.1789,
      "step": 18390
    },
    {
      "epoch": 1.809643234737282,
      "grad_norm": 1.4189066886901855,
      "learning_rate": 1.2889957477248343e-05,
      "loss": 0.2721,
      "step": 18400
    },
    {
      "epoch": 1.8106267364952915,
      "grad_norm": 2.705341100692749,
      "learning_rate": 1.2885983388308231e-05,
      "loss": 0.2601,
      "step": 18410
    },
    {
      "epoch": 1.811610238253301,
      "grad_norm": 0.8574623465538025,
      "learning_rate": 1.2882009299368121e-05,
      "loss": 0.2281,
      "step": 18420
    },
    {
      "epoch": 1.8125937400113101,
      "grad_norm": 0.766758143901825,
      "learning_rate": 1.287803521042801e-05,
      "loss": 0.2576,
      "step": 18430
    },
    {
      "epoch": 1.8135772417693197,
      "grad_norm": 1.3519665002822876,
      "learning_rate": 1.28740611214879e-05,
      "loss": 0.153,
      "step": 18440
    },
    {
      "epoch": 1.8145607435273292,
      "grad_norm": 0.4131558835506439,
      "learning_rate": 1.2870087032547788e-05,
      "loss": 0.2223,
      "step": 18450
    },
    {
      "epoch": 1.8155442452853383,
      "grad_norm": 2.183131456375122,
      "learning_rate": 1.2866112943607678e-05,
      "loss": 0.1797,
      "step": 18460
    },
    {
      "epoch": 1.8165277470433479,
      "grad_norm": 1.3639419078826904,
      "learning_rate": 1.286213885466757e-05,
      "loss": 0.2461,
      "step": 18470
    },
    {
      "epoch": 1.8175112488013574,
      "grad_norm": 1.5011903047561646,
      "learning_rate": 1.2858164765727459e-05,
      "loss": 0.1996,
      "step": 18480
    },
    {
      "epoch": 1.8184947505593665,
      "grad_norm": 1.1669235229492188,
      "learning_rate": 1.2854190676787349e-05,
      "loss": 0.3128,
      "step": 18490
    },
    {
      "epoch": 1.819478252317376,
      "grad_norm": 1.762707233428955,
      "learning_rate": 1.2850216587847237e-05,
      "loss": 0.1563,
      "step": 18500
    },
    {
      "epoch": 1.819478252317376,
      "eval_loss": 0.12783123552799225,
      "eval_runtime": 16.797,
      "eval_samples_per_second": 2.977,
      "eval_steps_per_second": 1.488,
      "step": 18500
    },
    {
      "epoch": 1.8204617540753856,
      "grad_norm": 3.2158849239349365,
      "learning_rate": 1.2846242498907126e-05,
      "loss": 0.2463,
      "step": 18510
    },
    {
      "epoch": 1.8214452558333947,
      "grad_norm": 0.165601447224617,
      "learning_rate": 1.2842268409967016e-05,
      "loss": 0.1599,
      "step": 18520
    },
    {
      "epoch": 1.8224287575914042,
      "grad_norm": 1.2064436674118042,
      "learning_rate": 1.2838294321026906e-05,
      "loss": 0.251,
      "step": 18530
    },
    {
      "epoch": 1.8234122593494135,
      "grad_norm": 1.1310498714447021,
      "learning_rate": 1.2834320232086794e-05,
      "loss": 0.2921,
      "step": 18540
    },
    {
      "epoch": 1.8243957611074229,
      "grad_norm": 1.2284313440322876,
      "learning_rate": 1.2830346143146683e-05,
      "loss": 0.2508,
      "step": 18550
    },
    {
      "epoch": 1.8253792628654324,
      "grad_norm": 0.17183829843997955,
      "learning_rate": 1.2826372054206575e-05,
      "loss": 0.21,
      "step": 18560
    },
    {
      "epoch": 1.8263627646234417,
      "grad_norm": 1.699674129486084,
      "learning_rate": 1.2822397965266464e-05,
      "loss": 0.2536,
      "step": 18570
    },
    {
      "epoch": 1.827346266381451,
      "grad_norm": 1.4755967855453491,
      "learning_rate": 1.2818423876326354e-05,
      "loss": 0.2296,
      "step": 18580
    },
    {
      "epoch": 1.8283297681394606,
      "grad_norm": 0.8879458904266357,
      "learning_rate": 1.2814449787386244e-05,
      "loss": 0.2436,
      "step": 18590
    },
    {
      "epoch": 1.82931326989747,
      "grad_norm": 0.39650049805641174,
      "learning_rate": 1.2810475698446132e-05,
      "loss": 0.19,
      "step": 18600
    },
    {
      "epoch": 1.8302967716554792,
      "grad_norm": 3.2891576290130615,
      "learning_rate": 1.2806501609506021e-05,
      "loss": 0.2146,
      "step": 18610
    },
    {
      "epoch": 1.8312802734134888,
      "grad_norm": 0.9302894473075867,
      "learning_rate": 1.2802527520565911e-05,
      "loss": 0.2911,
      "step": 18620
    },
    {
      "epoch": 1.832263775171498,
      "grad_norm": 0.9274962544441223,
      "learning_rate": 1.27985534316258e-05,
      "loss": 0.1712,
      "step": 18630
    },
    {
      "epoch": 1.8332472769295074,
      "grad_norm": 0.8566771149635315,
      "learning_rate": 1.2794579342685692e-05,
      "loss": 0.1546,
      "step": 18640
    },
    {
      "epoch": 1.834230778687517,
      "grad_norm": 2.141951322555542,
      "learning_rate": 1.279060525374558e-05,
      "loss": 0.3158,
      "step": 18650
    },
    {
      "epoch": 1.8352142804455263,
      "grad_norm": 0.6342948079109192,
      "learning_rate": 1.278663116480547e-05,
      "loss": 0.3165,
      "step": 18660
    },
    {
      "epoch": 1.8361977822035356,
      "grad_norm": 0.9059627056121826,
      "learning_rate": 1.278265707586536e-05,
      "loss": 0.2401,
      "step": 18670
    },
    {
      "epoch": 1.8371812839615451,
      "grad_norm": 2.2364370822906494,
      "learning_rate": 1.2778682986925249e-05,
      "loss": 0.1763,
      "step": 18680
    },
    {
      "epoch": 1.8381647857195544,
      "grad_norm": 1.2001042366027832,
      "learning_rate": 1.2774708897985137e-05,
      "loss": 0.1694,
      "step": 18690
    },
    {
      "epoch": 1.8391482874775638,
      "grad_norm": 1.6993484497070312,
      "learning_rate": 1.2770734809045027e-05,
      "loss": 0.2158,
      "step": 18700
    },
    {
      "epoch": 1.8401317892355733,
      "grad_norm": 1.790419578552246,
      "learning_rate": 1.2766760720104916e-05,
      "loss": 0.1804,
      "step": 18710
    },
    {
      "epoch": 1.8411152909935826,
      "grad_norm": 0.9894189834594727,
      "learning_rate": 1.2762786631164808e-05,
      "loss": 0.203,
      "step": 18720
    },
    {
      "epoch": 1.842098792751592,
      "grad_norm": 0.6162654161453247,
      "learning_rate": 1.2758812542224697e-05,
      "loss": 0.2229,
      "step": 18730
    },
    {
      "epoch": 1.8430822945096015,
      "grad_norm": 0.6175249814987183,
      "learning_rate": 1.2754838453284585e-05,
      "loss": 0.2179,
      "step": 18740
    },
    {
      "epoch": 1.8440657962676108,
      "grad_norm": 0.8340634703636169,
      "learning_rate": 1.2750864364344475e-05,
      "loss": 0.2586,
      "step": 18750
    },
    {
      "epoch": 1.8450492980256201,
      "grad_norm": 1.8784235715866089,
      "learning_rate": 1.2746890275404364e-05,
      "loss": 0.1957,
      "step": 18760
    },
    {
      "epoch": 1.8460327997836297,
      "grad_norm": 1.9351670742034912,
      "learning_rate": 1.2742916186464254e-05,
      "loss": 0.2829,
      "step": 18770
    },
    {
      "epoch": 1.847016301541639,
      "grad_norm": 0.6201217770576477,
      "learning_rate": 1.2738942097524142e-05,
      "loss": 0.2467,
      "step": 18780
    },
    {
      "epoch": 1.8479998032996483,
      "grad_norm": 0.5133494734764099,
      "learning_rate": 1.2734968008584032e-05,
      "loss": 0.2822,
      "step": 18790
    },
    {
      "epoch": 1.8489833050576578,
      "grad_norm": 1.161572813987732,
      "learning_rate": 1.2730993919643923e-05,
      "loss": 0.2605,
      "step": 18800
    },
    {
      "epoch": 1.8499668068156672,
      "grad_norm": 2.1235814094543457,
      "learning_rate": 1.2727019830703813e-05,
      "loss": 0.2341,
      "step": 18810
    },
    {
      "epoch": 1.8509503085736765,
      "grad_norm": 3.7511794567108154,
      "learning_rate": 1.2723045741763702e-05,
      "loss": 0.2416,
      "step": 18820
    },
    {
      "epoch": 1.851933810331686,
      "grad_norm": 1.0420117378234863,
      "learning_rate": 1.2719071652823592e-05,
      "loss": 0.2211,
      "step": 18830
    },
    {
      "epoch": 1.8529173120896953,
      "grad_norm": 1.2071075439453125,
      "learning_rate": 1.271509756388348e-05,
      "loss": 0.2008,
      "step": 18840
    },
    {
      "epoch": 1.8539008138477047,
      "grad_norm": 1.5324718952178955,
      "learning_rate": 1.271112347494337e-05,
      "loss": 0.2727,
      "step": 18850
    },
    {
      "epoch": 1.8548843156057142,
      "grad_norm": 2.7754740715026855,
      "learning_rate": 1.270714938600326e-05,
      "loss": 0.3221,
      "step": 18860
    },
    {
      "epoch": 1.8558678173637235,
      "grad_norm": 0.812025249004364,
      "learning_rate": 1.2703175297063149e-05,
      "loss": 0.2665,
      "step": 18870
    },
    {
      "epoch": 1.8568513191217328,
      "grad_norm": 0.7611910104751587,
      "learning_rate": 1.269920120812304e-05,
      "loss": 0.2205,
      "step": 18880
    },
    {
      "epoch": 1.8578348208797424,
      "grad_norm": 1.793511986732483,
      "learning_rate": 1.2695227119182928e-05,
      "loss": 0.1648,
      "step": 18890
    },
    {
      "epoch": 1.8588183226377517,
      "grad_norm": 1.9614449739456177,
      "learning_rate": 1.2691253030242818e-05,
      "loss": 0.2559,
      "step": 18900
    },
    {
      "epoch": 1.859801824395761,
      "grad_norm": 2.4327352046966553,
      "learning_rate": 1.2687278941302708e-05,
      "loss": 0.2072,
      "step": 18910
    },
    {
      "epoch": 1.8607853261537706,
      "grad_norm": 0.6133950352668762,
      "learning_rate": 1.2683304852362597e-05,
      "loss": 0.1654,
      "step": 18920
    },
    {
      "epoch": 1.8617688279117799,
      "grad_norm": 1.5375863313674927,
      "learning_rate": 1.2679330763422485e-05,
      "loss": 0.1405,
      "step": 18930
    },
    {
      "epoch": 1.8627523296697892,
      "grad_norm": 1.1150537729263306,
      "learning_rate": 1.2675356674482375e-05,
      "loss": 0.2805,
      "step": 18940
    },
    {
      "epoch": 1.8637358314277988,
      "grad_norm": 0.5071201920509338,
      "learning_rate": 1.2671382585542265e-05,
      "loss": 0.1458,
      "step": 18950
    },
    {
      "epoch": 1.864719333185808,
      "grad_norm": 0.7502576112747192,
      "learning_rate": 1.2667408496602156e-05,
      "loss": 0.1965,
      "step": 18960
    },
    {
      "epoch": 1.8657028349438174,
      "grad_norm": 0.3296958804130554,
      "learning_rate": 1.2663434407662046e-05,
      "loss": 0.2225,
      "step": 18970
    },
    {
      "epoch": 1.866686336701827,
      "grad_norm": 1.3702101707458496,
      "learning_rate": 1.2659460318721934e-05,
      "loss": 0.1995,
      "step": 18980
    },
    {
      "epoch": 1.8676698384598363,
      "grad_norm": 3.903212070465088,
      "learning_rate": 1.2655486229781823e-05,
      "loss": 0.1307,
      "step": 18990
    },
    {
      "epoch": 1.8686533402178456,
      "grad_norm": 3.3656933307647705,
      "learning_rate": 1.2651512140841713e-05,
      "loss": 0.2051,
      "step": 19000
    },
    {
      "epoch": 1.8686533402178456,
      "eval_loss": 0.1276712268590927,
      "eval_runtime": 17.9095,
      "eval_samples_per_second": 2.792,
      "eval_steps_per_second": 1.396,
      "step": 19000
    },
    {
      "epoch": 1.8696368419758551,
      "grad_norm": 5.251664161682129,
      "learning_rate": 1.2647538051901603e-05,
      "loss": 0.169,
      "step": 19010
    },
    {
      "epoch": 1.8706203437338644,
      "grad_norm": 1.5694376230239868,
      "learning_rate": 1.264356396296149e-05,
      "loss": 0.2338,
      "step": 19020
    },
    {
      "epoch": 1.8716038454918738,
      "grad_norm": 1.6634079217910767,
      "learning_rate": 1.263958987402138e-05,
      "loss": 0.205,
      "step": 19030
    },
    {
      "epoch": 1.8725873472498833,
      "grad_norm": 0.9844359755516052,
      "learning_rate": 1.2635615785081272e-05,
      "loss": 0.1634,
      "step": 19040
    },
    {
      "epoch": 1.8735708490078926,
      "grad_norm": 3.8357009887695312,
      "learning_rate": 1.2631641696141161e-05,
      "loss": 0.3183,
      "step": 19050
    },
    {
      "epoch": 1.874554350765902,
      "grad_norm": 4.7126946449279785,
      "learning_rate": 1.2627667607201051e-05,
      "loss": 0.2224,
      "step": 19060
    },
    {
      "epoch": 1.8755378525239115,
      "grad_norm": 1.908003568649292,
      "learning_rate": 1.262369351826094e-05,
      "loss": 0.1636,
      "step": 19070
    },
    {
      "epoch": 1.8765213542819208,
      "grad_norm": 0.8002133965492249,
      "learning_rate": 1.2619719429320829e-05,
      "loss": 0.1864,
      "step": 19080
    },
    {
      "epoch": 1.8775048560399301,
      "grad_norm": 1.1409190893173218,
      "learning_rate": 1.2615745340380718e-05,
      "loss": 0.2992,
      "step": 19090
    },
    {
      "epoch": 1.8784883577979397,
      "grad_norm": 0.7023701071739197,
      "learning_rate": 1.2611771251440608e-05,
      "loss": 0.2487,
      "step": 19100
    },
    {
      "epoch": 1.879471859555949,
      "grad_norm": 0.6862336993217468,
      "learning_rate": 1.2607797162500498e-05,
      "loss": 0.2431,
      "step": 19110
    },
    {
      "epoch": 1.8804553613139583,
      "grad_norm": 0.65170818567276,
      "learning_rate": 1.2603823073560389e-05,
      "loss": 0.2289,
      "step": 19120
    },
    {
      "epoch": 1.8814388630719678,
      "grad_norm": 0.6964781880378723,
      "learning_rate": 1.2599848984620277e-05,
      "loss": 0.2642,
      "step": 19130
    },
    {
      "epoch": 1.8824223648299772,
      "grad_norm": 0.9222815632820129,
      "learning_rate": 1.2595874895680167e-05,
      "loss": 0.2591,
      "step": 19140
    },
    {
      "epoch": 1.8834058665879865,
      "grad_norm": 1.1235778331756592,
      "learning_rate": 1.2591900806740056e-05,
      "loss": 0.2364,
      "step": 19150
    },
    {
      "epoch": 1.884389368345996,
      "grad_norm": 1.0253043174743652,
      "learning_rate": 1.2587926717799946e-05,
      "loss": 0.1966,
      "step": 19160
    },
    {
      "epoch": 1.8853728701040053,
      "grad_norm": 2.779137134552002,
      "learning_rate": 1.2583952628859834e-05,
      "loss": 0.1278,
      "step": 19170
    },
    {
      "epoch": 1.8863563718620147,
      "grad_norm": 1.0971506834030151,
      "learning_rate": 1.2579978539919724e-05,
      "loss": 0.1466,
      "step": 19180
    },
    {
      "epoch": 1.8873398736200242,
      "grad_norm": 1.736945390701294,
      "learning_rate": 1.2576004450979613e-05,
      "loss": 0.1307,
      "step": 19190
    },
    {
      "epoch": 1.8883233753780335,
      "grad_norm": 2.4305453300476074,
      "learning_rate": 1.2572030362039505e-05,
      "loss": 0.2525,
      "step": 19200
    },
    {
      "epoch": 1.8893068771360428,
      "grad_norm": 0.7563501596450806,
      "learning_rate": 1.2568056273099394e-05,
      "loss": 0.2186,
      "step": 19210
    },
    {
      "epoch": 1.8902903788940524,
      "grad_norm": 2.958002805709839,
      "learning_rate": 1.2564082184159282e-05,
      "loss": 0.2845,
      "step": 19220
    },
    {
      "epoch": 1.8912738806520617,
      "grad_norm": 2.1089866161346436,
      "learning_rate": 1.2560108095219172e-05,
      "loss": 0.2058,
      "step": 19230
    },
    {
      "epoch": 1.892257382410071,
      "grad_norm": 1.3916528224945068,
      "learning_rate": 1.2556134006279062e-05,
      "loss": 0.2509,
      "step": 19240
    },
    {
      "epoch": 1.8932408841680806,
      "grad_norm": 0.8218703866004944,
      "learning_rate": 1.2552159917338951e-05,
      "loss": 0.2344,
      "step": 19250
    },
    {
      "epoch": 1.8942243859260899,
      "grad_norm": 1.5939429998397827,
      "learning_rate": 1.254818582839884e-05,
      "loss": 0.1323,
      "step": 19260
    },
    {
      "epoch": 1.8952078876840992,
      "grad_norm": 1.384416103363037,
      "learning_rate": 1.2544211739458729e-05,
      "loss": 0.3115,
      "step": 19270
    },
    {
      "epoch": 1.8961913894421087,
      "grad_norm": 1.6949342489242554,
      "learning_rate": 1.254023765051862e-05,
      "loss": 0.1937,
      "step": 19280
    },
    {
      "epoch": 1.897174891200118,
      "grad_norm": 0.34746313095092773,
      "learning_rate": 1.253626356157851e-05,
      "loss": 0.2859,
      "step": 19290
    },
    {
      "epoch": 1.8981583929581274,
      "grad_norm": 0.28613096475601196,
      "learning_rate": 1.25322894726384e-05,
      "loss": 0.3169,
      "step": 19300
    },
    {
      "epoch": 1.899141894716137,
      "grad_norm": 1.5283467769622803,
      "learning_rate": 1.252831538369829e-05,
      "loss": 0.1637,
      "step": 19310
    },
    {
      "epoch": 1.9001253964741462,
      "grad_norm": 3.342489719390869,
      "learning_rate": 1.2524341294758177e-05,
      "loss": 0.2558,
      "step": 19320
    },
    {
      "epoch": 1.9011088982321556,
      "grad_norm": 0.7206294536590576,
      "learning_rate": 1.2520367205818067e-05,
      "loss": 0.3589,
      "step": 19330
    },
    {
      "epoch": 1.902092399990165,
      "grad_norm": 0.2409156858921051,
      "learning_rate": 1.2516393116877957e-05,
      "loss": 0.1457,
      "step": 19340
    },
    {
      "epoch": 1.9030759017481744,
      "grad_norm": 1.1184440851211548,
      "learning_rate": 1.2512419027937846e-05,
      "loss": 0.226,
      "step": 19350
    },
    {
      "epoch": 1.9040594035061837,
      "grad_norm": 1.8674637079238892,
      "learning_rate": 1.2508444938997734e-05,
      "loss": 0.2031,
      "step": 19360
    },
    {
      "epoch": 1.9050429052641933,
      "grad_norm": 0.9980869293212891,
      "learning_rate": 1.2504470850057626e-05,
      "loss": 0.2601,
      "step": 19370
    },
    {
      "epoch": 1.9060264070222026,
      "grad_norm": 1.2215583324432373,
      "learning_rate": 1.2500496761117515e-05,
      "loss": 0.1995,
      "step": 19380
    },
    {
      "epoch": 1.907009908780212,
      "grad_norm": 0.635554850101471,
      "learning_rate": 1.2496522672177405e-05,
      "loss": 0.2687,
      "step": 19390
    },
    {
      "epoch": 1.9079934105382215,
      "grad_norm": 2.1318042278289795,
      "learning_rate": 1.2492548583237295e-05,
      "loss": 0.2183,
      "step": 19400
    },
    {
      "epoch": 1.9089769122962306,
      "grad_norm": 0.2807762622833252,
      "learning_rate": 1.2488574494297183e-05,
      "loss": 0.2251,
      "step": 19410
    },
    {
      "epoch": 1.90996041405424,
      "grad_norm": 2.432757616043091,
      "learning_rate": 1.2484600405357072e-05,
      "loss": 0.1734,
      "step": 19420
    },
    {
      "epoch": 1.9109439158122496,
      "grad_norm": 1.1255124807357788,
      "learning_rate": 1.2480626316416962e-05,
      "loss": 0.2215,
      "step": 19430
    },
    {
      "epoch": 1.9119274175702587,
      "grad_norm": 1.1864866018295288,
      "learning_rate": 1.2476652227476852e-05,
      "loss": 0.2485,
      "step": 19440
    },
    {
      "epoch": 1.9129109193282683,
      "grad_norm": 0.8781852126121521,
      "learning_rate": 1.2472678138536743e-05,
      "loss": 0.2639,
      "step": 19450
    },
    {
      "epoch": 1.9138944210862778,
      "grad_norm": 0.8348128199577332,
      "learning_rate": 1.2468704049596631e-05,
      "loss": 0.2793,
      "step": 19460
    },
    {
      "epoch": 1.914877922844287,
      "grad_norm": 0.407959520816803,
      "learning_rate": 1.246472996065652e-05,
      "loss": 0.2604,
      "step": 19470
    },
    {
      "epoch": 1.9158614246022965,
      "grad_norm": 2.0708372592926025,
      "learning_rate": 1.246075587171641e-05,
      "loss": 0.3622,
      "step": 19480
    },
    {
      "epoch": 1.916844926360306,
      "grad_norm": 1.2978930473327637,
      "learning_rate": 1.24567817827763e-05,
      "loss": 0.3135,
      "step": 19490
    },
    {
      "epoch": 1.917828428118315,
      "grad_norm": 2.307457208633423,
      "learning_rate": 1.2452807693836188e-05,
      "loss": 0.2206,
      "step": 19500
    },
    {
      "epoch": 1.917828428118315,
      "eval_loss": 0.13785213232040405,
      "eval_runtime": 18.2171,
      "eval_samples_per_second": 2.745,
      "eval_steps_per_second": 1.372,
      "step": 19500
    },
    {
      "epoch": 1.9188119298763247,
      "grad_norm": 0.5169830322265625,
      "learning_rate": 1.2448833604896077e-05,
      "loss": 0.273,
      "step": 19510
    },
    {
      "epoch": 1.9197954316343342,
      "grad_norm": 0.7869198322296143,
      "learning_rate": 1.2444859515955967e-05,
      "loss": 0.1729,
      "step": 19520
    },
    {
      "epoch": 1.9207789333923433,
      "grad_norm": 1.5760027170181274,
      "learning_rate": 1.2440885427015859e-05,
      "loss": 0.1508,
      "step": 19530
    },
    {
      "epoch": 1.9217624351503528,
      "grad_norm": 2.80485200881958,
      "learning_rate": 1.2436911338075748e-05,
      "loss": 0.3853,
      "step": 19540
    },
    {
      "epoch": 1.9227459369083624,
      "grad_norm": 1.0592259168624878,
      "learning_rate": 1.2432937249135638e-05,
      "loss": 0.2399,
      "step": 19550
    },
    {
      "epoch": 1.9237294386663715,
      "grad_norm": 0.8268088698387146,
      "learning_rate": 1.2428963160195526e-05,
      "loss": 0.2155,
      "step": 19560
    },
    {
      "epoch": 1.924712940424381,
      "grad_norm": 1.1707844734191895,
      "learning_rate": 1.2424989071255415e-05,
      "loss": 0.2499,
      "step": 19570
    },
    {
      "epoch": 1.9256964421823906,
      "grad_norm": 1.9008899927139282,
      "learning_rate": 1.2421014982315305e-05,
      "loss": 0.2489,
      "step": 19580
    },
    {
      "epoch": 1.9266799439403997,
      "grad_norm": 0.991098940372467,
      "learning_rate": 1.2417040893375195e-05,
      "loss": 0.2994,
      "step": 19590
    },
    {
      "epoch": 1.9276634456984092,
      "grad_norm": 1.5907323360443115,
      "learning_rate": 1.2413066804435083e-05,
      "loss": 0.2677,
      "step": 19600
    },
    {
      "epoch": 1.9286469474564187,
      "grad_norm": 0.3907619118690491,
      "learning_rate": 1.2409092715494974e-05,
      "loss": 0.259,
      "step": 19610
    },
    {
      "epoch": 1.9296304492144278,
      "grad_norm": 0.6614681482315063,
      "learning_rate": 1.2405118626554864e-05,
      "loss": 0.2999,
      "step": 19620
    },
    {
      "epoch": 1.9306139509724374,
      "grad_norm": 1.7540950775146484,
      "learning_rate": 1.2401144537614753e-05,
      "loss": 0.3011,
      "step": 19630
    },
    {
      "epoch": 1.931597452730447,
      "grad_norm": 2.1542980670928955,
      "learning_rate": 1.2397170448674643e-05,
      "loss": 0.1974,
      "step": 19640
    },
    {
      "epoch": 1.932580954488456,
      "grad_norm": 0.1936643123626709,
      "learning_rate": 1.2393196359734531e-05,
      "loss": 0.1426,
      "step": 19650
    },
    {
      "epoch": 1.9335644562464656,
      "grad_norm": 1.4605008363723755,
      "learning_rate": 1.238922227079442e-05,
      "loss": 0.2313,
      "step": 19660
    },
    {
      "epoch": 1.934547958004475,
      "grad_norm": 0.7753693461418152,
      "learning_rate": 1.238524818185431e-05,
      "loss": 0.302,
      "step": 19670
    },
    {
      "epoch": 1.9355314597624842,
      "grad_norm": 1.482616662979126,
      "learning_rate": 1.23812740929142e-05,
      "loss": 0.3001,
      "step": 19680
    },
    {
      "epoch": 1.9365149615204937,
      "grad_norm": 0.818903386592865,
      "learning_rate": 1.2377300003974091e-05,
      "loss": 0.2544,
      "step": 19690
    },
    {
      "epoch": 1.937498463278503,
      "grad_norm": 0.44427138566970825,
      "learning_rate": 1.237332591503398e-05,
      "loss": 0.2739,
      "step": 19700
    },
    {
      "epoch": 1.9384819650365124,
      "grad_norm": 1.0438803434371948,
      "learning_rate": 1.2369351826093869e-05,
      "loss": 0.23,
      "step": 19710
    },
    {
      "epoch": 1.939465466794522,
      "grad_norm": 2.2877156734466553,
      "learning_rate": 1.2365377737153759e-05,
      "loss": 0.2936,
      "step": 19720
    },
    {
      "epoch": 1.9404489685525312,
      "grad_norm": 0.29466554522514343,
      "learning_rate": 1.2361403648213648e-05,
      "loss": 0.2685,
      "step": 19730
    },
    {
      "epoch": 1.9414324703105406,
      "grad_norm": 1.4913973808288574,
      "learning_rate": 1.2357429559273536e-05,
      "loss": 0.2509,
      "step": 19740
    },
    {
      "epoch": 1.94241597206855,
      "grad_norm": 1.4097399711608887,
      "learning_rate": 1.2353455470333426e-05,
      "loss": 0.1971,
      "step": 19750
    },
    {
      "epoch": 1.9433994738265594,
      "grad_norm": 1.3867621421813965,
      "learning_rate": 1.2349481381393316e-05,
      "loss": 0.2442,
      "step": 19760
    },
    {
      "epoch": 1.9443829755845687,
      "grad_norm": 1.646846890449524,
      "learning_rate": 1.2345507292453207e-05,
      "loss": 0.1876,
      "step": 19770
    },
    {
      "epoch": 1.9453664773425783,
      "grad_norm": 1.509426474571228,
      "learning_rate": 1.2341533203513097e-05,
      "loss": 0.32,
      "step": 19780
    },
    {
      "epoch": 1.9463499791005876,
      "grad_norm": 2.4654855728149414,
      "learning_rate": 1.2337559114572986e-05,
      "loss": 0.2244,
      "step": 19790
    },
    {
      "epoch": 1.947333480858597,
      "grad_norm": 1.9394328594207764,
      "learning_rate": 1.2333585025632874e-05,
      "loss": 0.1677,
      "step": 19800
    },
    {
      "epoch": 1.9483169826166065,
      "grad_norm": 2.2644760608673096,
      "learning_rate": 1.2329610936692764e-05,
      "loss": 0.304,
      "step": 19810
    },
    {
      "epoch": 1.9493004843746158,
      "grad_norm": 0.9643498659133911,
      "learning_rate": 1.2325636847752654e-05,
      "loss": 0.2295,
      "step": 19820
    },
    {
      "epoch": 1.950283986132625,
      "grad_norm": 0.6572133898735046,
      "learning_rate": 1.2321662758812543e-05,
      "loss": 0.2946,
      "step": 19830
    },
    {
      "epoch": 1.9512674878906346,
      "grad_norm": 0.9991806745529175,
      "learning_rate": 1.2317688669872431e-05,
      "loss": 0.1694,
      "step": 19840
    },
    {
      "epoch": 1.952250989648644,
      "grad_norm": 1.4036468267440796,
      "learning_rate": 1.2313714580932323e-05,
      "loss": 0.2091,
      "step": 19850
    },
    {
      "epoch": 1.9532344914066533,
      "grad_norm": 0.4763045012950897,
      "learning_rate": 1.2309740491992212e-05,
      "loss": 0.2523,
      "step": 19860
    },
    {
      "epoch": 1.9542179931646628,
      "grad_norm": 2.5620388984680176,
      "learning_rate": 1.2305766403052102e-05,
      "loss": 0.2626,
      "step": 19870
    },
    {
      "epoch": 1.9552014949226721,
      "grad_norm": 1.4312139749526978,
      "learning_rate": 1.2301792314111992e-05,
      "loss": 0.323,
      "step": 19880
    },
    {
      "epoch": 1.9561849966806815,
      "grad_norm": 0.5666785836219788,
      "learning_rate": 1.229781822517188e-05,
      "loss": 0.216,
      "step": 19890
    },
    {
      "epoch": 1.957168498438691,
      "grad_norm": 2.9127252101898193,
      "learning_rate": 1.229384413623177e-05,
      "loss": 0.3055,
      "step": 19900
    },
    {
      "epoch": 1.9581520001967003,
      "grad_norm": 0.6122851967811584,
      "learning_rate": 1.2289870047291659e-05,
      "loss": 0.2277,
      "step": 19910
    },
    {
      "epoch": 1.9591355019547096,
      "grad_norm": 0.7845405340194702,
      "learning_rate": 1.2285895958351549e-05,
      "loss": 0.185,
      "step": 19920
    },
    {
      "epoch": 1.9601190037127192,
      "grad_norm": 1.7378236055374146,
      "learning_rate": 1.228192186941144e-05,
      "loss": 0.2031,
      "step": 19930
    },
    {
      "epoch": 1.9611025054707285,
      "grad_norm": 1.2192821502685547,
      "learning_rate": 1.2277947780471328e-05,
      "loss": 0.1736,
      "step": 19940
    },
    {
      "epoch": 1.9620860072287378,
      "grad_norm": 0.7177460193634033,
      "learning_rate": 1.2273973691531218e-05,
      "loss": 0.2042,
      "step": 19950
    },
    {
      "epoch": 1.9630695089867474,
      "grad_norm": 0.33699867129325867,
      "learning_rate": 1.2269999602591107e-05,
      "loss": 0.1934,
      "step": 19960
    },
    {
      "epoch": 1.9640530107447567,
      "grad_norm": 0.47641661763191223,
      "learning_rate": 1.2266025513650997e-05,
      "loss": 0.233,
      "step": 19970
    },
    {
      "epoch": 1.965036512502766,
      "grad_norm": 0.7735099792480469,
      "learning_rate": 1.2262051424710885e-05,
      "loss": 0.2167,
      "step": 19980
    },
    {
      "epoch": 1.9660200142607756,
      "grad_norm": 0.8787487149238586,
      "learning_rate": 1.2258077335770775e-05,
      "loss": 0.2523,
      "step": 19990
    },
    {
      "epoch": 1.9670035160187849,
      "grad_norm": 0.5472764372825623,
      "learning_rate": 1.2254103246830664e-05,
      "loss": 0.1696,
      "step": 20000
    },
    {
      "epoch": 1.9670035160187849,
      "eval_loss": 0.1349218338727951,
      "eval_runtime": 16.6238,
      "eval_samples_per_second": 3.008,
      "eval_steps_per_second": 1.504,
      "step": 20000
    },
    {
      "epoch": 1.9679870177767942,
      "grad_norm": 1.3282941579818726,
      "learning_rate": 1.2250129157890556e-05,
      "loss": 0.2454,
      "step": 20010
    },
    {
      "epoch": 1.9689705195348037,
      "grad_norm": 0.5796499252319336,
      "learning_rate": 1.2246155068950445e-05,
      "loss": 0.2414,
      "step": 20020
    },
    {
      "epoch": 1.969954021292813,
      "grad_norm": 0.5740196704864502,
      "learning_rate": 1.2242180980010335e-05,
      "loss": 0.2331,
      "step": 20030
    },
    {
      "epoch": 1.9709375230508224,
      "grad_norm": 1.5052549839019775,
      "learning_rate": 1.2238206891070223e-05,
      "loss": 0.1921,
      "step": 20040
    },
    {
      "epoch": 1.971921024808832,
      "grad_norm": 1.2629809379577637,
      "learning_rate": 1.2234232802130113e-05,
      "loss": 0.198,
      "step": 20050
    },
    {
      "epoch": 1.9729045265668412,
      "grad_norm": 1.0493677854537964,
      "learning_rate": 1.2230258713190002e-05,
      "loss": 0.2535,
      "step": 20060
    },
    {
      "epoch": 1.9738880283248506,
      "grad_norm": 1.112351417541504,
      "learning_rate": 1.2226284624249892e-05,
      "loss": 0.2749,
      "step": 20070
    },
    {
      "epoch": 1.97487153008286,
      "grad_norm": 0.5583861470222473,
      "learning_rate": 1.222231053530978e-05,
      "loss": 0.2528,
      "step": 20080
    },
    {
      "epoch": 1.9758550318408694,
      "grad_norm": 1.9555214643478394,
      "learning_rate": 1.221833644636967e-05,
      "loss": 0.1862,
      "step": 20090
    },
    {
      "epoch": 1.9768385335988787,
      "grad_norm": 0.671154260635376,
      "learning_rate": 1.2214362357429561e-05,
      "loss": 0.1837,
      "step": 20100
    },
    {
      "epoch": 1.9778220353568883,
      "grad_norm": 1.681707501411438,
      "learning_rate": 1.221038826848945e-05,
      "loss": 0.2118,
      "step": 20110
    },
    {
      "epoch": 1.9788055371148976,
      "grad_norm": 0.433243989944458,
      "learning_rate": 1.220641417954934e-05,
      "loss": 0.3268,
      "step": 20120
    },
    {
      "epoch": 1.979789038872907,
      "grad_norm": 1.1356432437896729,
      "learning_rate": 1.2202440090609228e-05,
      "loss": 0.2099,
      "step": 20130
    },
    {
      "epoch": 1.9807725406309165,
      "grad_norm": 0.8848307728767395,
      "learning_rate": 1.2198466001669118e-05,
      "loss": 0.2387,
      "step": 20140
    },
    {
      "epoch": 1.9817560423889258,
      "grad_norm": 1.2508822679519653,
      "learning_rate": 1.2194491912729008e-05,
      "loss": 0.2348,
      "step": 20150
    },
    {
      "epoch": 1.982739544146935,
      "grad_norm": 1.3926819562911987,
      "learning_rate": 1.2190517823788897e-05,
      "loss": 0.1377,
      "step": 20160
    },
    {
      "epoch": 1.9837230459049446,
      "grad_norm": 1.2700151205062866,
      "learning_rate": 1.2186543734848785e-05,
      "loss": 0.217,
      "step": 20170
    },
    {
      "epoch": 1.984706547662954,
      "grad_norm": 1.615131139755249,
      "learning_rate": 1.2182569645908677e-05,
      "loss": 0.2022,
      "step": 20180
    },
    {
      "epoch": 1.9856900494209633,
      "grad_norm": 0.827870786190033,
      "learning_rate": 1.2178595556968566e-05,
      "loss": 0.3,
      "step": 20190
    },
    {
      "epoch": 1.9866735511789728,
      "grad_norm": 0.9399486780166626,
      "learning_rate": 1.2174621468028456e-05,
      "loss": 0.2433,
      "step": 20200
    },
    {
      "epoch": 1.9876570529369821,
      "grad_norm": 0.638542652130127,
      "learning_rate": 1.2170647379088346e-05,
      "loss": 0.2699,
      "step": 20210
    },
    {
      "epoch": 1.9886405546949915,
      "grad_norm": 0.666766881942749,
      "learning_rate": 1.2166673290148234e-05,
      "loss": 0.2616,
      "step": 20220
    },
    {
      "epoch": 1.989624056453001,
      "grad_norm": 1.5862071514129639,
      "learning_rate": 1.2162699201208123e-05,
      "loss": 0.189,
      "step": 20230
    },
    {
      "epoch": 1.9906075582110103,
      "grad_norm": 1.093273639678955,
      "learning_rate": 1.2158725112268013e-05,
      "loss": 0.3595,
      "step": 20240
    },
    {
      "epoch": 1.9915910599690196,
      "grad_norm": 0.5610359907150269,
      "learning_rate": 1.2154751023327903e-05,
      "loss": 0.2155,
      "step": 20250
    },
    {
      "epoch": 1.9925745617270292,
      "grad_norm": 0.38059747219085693,
      "learning_rate": 1.2150776934387794e-05,
      "loss": 0.1547,
      "step": 20260
    },
    {
      "epoch": 1.9935580634850385,
      "grad_norm": 0.5128862261772156,
      "learning_rate": 1.2146802845447684e-05,
      "loss": 0.2517,
      "step": 20270
    },
    {
      "epoch": 1.9945415652430478,
      "grad_norm": 0.22182230651378632,
      "learning_rate": 1.2142828756507572e-05,
      "loss": 0.167,
      "step": 20280
    },
    {
      "epoch": 1.9955250670010574,
      "grad_norm": 0.8813309073448181,
      "learning_rate": 1.2138854667567461e-05,
      "loss": 0.3163,
      "step": 20290
    },
    {
      "epoch": 1.9965085687590667,
      "grad_norm": 0.5298721194267273,
      "learning_rate": 1.213488057862735e-05,
      "loss": 0.2425,
      "step": 20300
    },
    {
      "epoch": 1.997492070517076,
      "grad_norm": 0.42940881848335266,
      "learning_rate": 1.213090648968724e-05,
      "loss": 0.1336,
      "step": 20310
    },
    {
      "epoch": 1.9984755722750855,
      "grad_norm": 0.486102819442749,
      "learning_rate": 1.2126932400747128e-05,
      "loss": 0.3274,
      "step": 20320
    },
    {
      "epoch": 1.9994590740330949,
      "grad_norm": 1.8005328178405762,
      "learning_rate": 1.2122958311807018e-05,
      "loss": 0.2202,
      "step": 20330
    },
    {
      "epoch": 2.000442575791104,
      "grad_norm": 1.0286778211593628,
      "learning_rate": 1.211898422286691e-05,
      "loss": 0.1598,
      "step": 20340
    },
    {
      "epoch": 2.0014260775491137,
      "grad_norm": 0.1806839257478714,
      "learning_rate": 1.21150101339268e-05,
      "loss": 0.1884,
      "step": 20350
    },
    {
      "epoch": 2.002409579307123,
      "grad_norm": 0.7888972163200378,
      "learning_rate": 1.2111036044986689e-05,
      "loss": 0.1484,
      "step": 20360
    },
    {
      "epoch": 2.0033930810651324,
      "grad_norm": 0.4760337769985199,
      "learning_rate": 1.2107061956046577e-05,
      "loss": 0.1861,
      "step": 20370
    },
    {
      "epoch": 2.004376582823142,
      "grad_norm": 1.431585431098938,
      "learning_rate": 1.2103087867106466e-05,
      "loss": 0.2105,
      "step": 20380
    },
    {
      "epoch": 2.005360084581151,
      "grad_norm": 1.5212070941925049,
      "learning_rate": 1.2099113778166356e-05,
      "loss": 0.2318,
      "step": 20390
    },
    {
      "epoch": 2.0063435863391605,
      "grad_norm": 0.7492786645889282,
      "learning_rate": 1.2095139689226246e-05,
      "loss": 0.2358,
      "step": 20400
    },
    {
      "epoch": 2.00732708809717,
      "grad_norm": 1.1958400011062622,
      "learning_rate": 1.2091165600286134e-05,
      "loss": 0.2445,
      "step": 20410
    },
    {
      "epoch": 2.008310589855179,
      "grad_norm": 1.0838582515716553,
      "learning_rate": 1.2087191511346025e-05,
      "loss": 0.1547,
      "step": 20420
    },
    {
      "epoch": 2.0092940916131887,
      "grad_norm": 2.8028171062469482,
      "learning_rate": 1.2083217422405915e-05,
      "loss": 0.1836,
      "step": 20430
    },
    {
      "epoch": 2.0102775933711983,
      "grad_norm": 0.11121039092540741,
      "learning_rate": 1.2079243333465804e-05,
      "loss": 0.2085,
      "step": 20440
    },
    {
      "epoch": 2.0112610951292074,
      "grad_norm": 0.6349248886108398,
      "learning_rate": 1.2075269244525694e-05,
      "loss": 0.1738,
      "step": 20450
    },
    {
      "epoch": 2.012244596887217,
      "grad_norm": 2.560642957687378,
      "learning_rate": 1.2071295155585582e-05,
      "loss": 0.2848,
      "step": 20460
    },
    {
      "epoch": 2.0132280986452264,
      "grad_norm": 0.6541864275932312,
      "learning_rate": 1.2067321066645472e-05,
      "loss": 0.2546,
      "step": 20470
    },
    {
      "epoch": 2.0142116004032355,
      "grad_norm": 0.8187896609306335,
      "learning_rate": 1.2063346977705361e-05,
      "loss": 0.189,
      "step": 20480
    },
    {
      "epoch": 2.015195102161245,
      "grad_norm": 2.4058218002319336,
      "learning_rate": 1.2059372888765251e-05,
      "loss": 0.181,
      "step": 20490
    },
    {
      "epoch": 2.0161786039192546,
      "grad_norm": 0.8954346776008606,
      "learning_rate": 1.2055398799825142e-05,
      "loss": 0.238,
      "step": 20500
    },
    {
      "epoch": 2.0161786039192546,
      "eval_loss": 0.13197824358940125,
      "eval_runtime": 18.561,
      "eval_samples_per_second": 2.694,
      "eval_steps_per_second": 1.347,
      "step": 20500
    },
    {
      "epoch": 2.0171621056772637,
      "grad_norm": 1.251834511756897,
      "learning_rate": 1.2051424710885032e-05,
      "loss": 0.2142,
      "step": 20510
    },
    {
      "epoch": 2.0181456074352733,
      "grad_norm": 1.2214981317520142,
      "learning_rate": 1.204745062194492e-05,
      "loss": 0.2327,
      "step": 20520
    },
    {
      "epoch": 2.019129109193283,
      "grad_norm": 1.1138148307800293,
      "learning_rate": 1.204347653300481e-05,
      "loss": 0.241,
      "step": 20530
    },
    {
      "epoch": 2.020112610951292,
      "grad_norm": 1.0483516454696655,
      "learning_rate": 1.20395024440647e-05,
      "loss": 0.2041,
      "step": 20540
    },
    {
      "epoch": 2.0210961127093015,
      "grad_norm": 0.6449570655822754,
      "learning_rate": 1.2035528355124589e-05,
      "loss": 0.204,
      "step": 20550
    },
    {
      "epoch": 2.022079614467311,
      "grad_norm": 0.3355649411678314,
      "learning_rate": 1.2031554266184477e-05,
      "loss": 0.236,
      "step": 20560
    },
    {
      "epoch": 2.02306311622532,
      "grad_norm": 1.418927788734436,
      "learning_rate": 1.2027580177244367e-05,
      "loss": 0.2231,
      "step": 20570
    },
    {
      "epoch": 2.0240466179833296,
      "grad_norm": 1.9252437353134155,
      "learning_rate": 1.2023606088304258e-05,
      "loss": 0.1752,
      "step": 20580
    },
    {
      "epoch": 2.025030119741339,
      "grad_norm": 0.509128749370575,
      "learning_rate": 1.2019631999364148e-05,
      "loss": 0.2209,
      "step": 20590
    },
    {
      "epoch": 2.0260136214993483,
      "grad_norm": 1.4364620447158813,
      "learning_rate": 1.2015657910424037e-05,
      "loss": 0.2018,
      "step": 20600
    },
    {
      "epoch": 2.026997123257358,
      "grad_norm": 0.6519067287445068,
      "learning_rate": 1.2011683821483925e-05,
      "loss": 0.2391,
      "step": 20610
    },
    {
      "epoch": 2.0279806250153674,
      "grad_norm": 0.9485163688659668,
      "learning_rate": 1.2007709732543815e-05,
      "loss": 0.1662,
      "step": 20620
    },
    {
      "epoch": 2.0289641267733765,
      "grad_norm": 0.6042230129241943,
      "learning_rate": 1.2003735643603705e-05,
      "loss": 0.1974,
      "step": 20630
    },
    {
      "epoch": 2.029947628531386,
      "grad_norm": 0.9156279563903809,
      "learning_rate": 1.1999761554663594e-05,
      "loss": 0.2873,
      "step": 20640
    },
    {
      "epoch": 2.0309311302893955,
      "grad_norm": 1.7737699747085571,
      "learning_rate": 1.1995787465723482e-05,
      "loss": 0.2187,
      "step": 20650
    },
    {
      "epoch": 2.0319146320474046,
      "grad_norm": 1.2266044616699219,
      "learning_rate": 1.1991813376783374e-05,
      "loss": 0.214,
      "step": 20660
    },
    {
      "epoch": 2.032898133805414,
      "grad_norm": 1.0398406982421875,
      "learning_rate": 1.1987839287843263e-05,
      "loss": 0.1384,
      "step": 20670
    },
    {
      "epoch": 2.0338816355634237,
      "grad_norm": 0.5955280065536499,
      "learning_rate": 1.1983865198903153e-05,
      "loss": 0.3007,
      "step": 20680
    },
    {
      "epoch": 2.034865137321433,
      "grad_norm": 2.1822447776794434,
      "learning_rate": 1.1979891109963043e-05,
      "loss": 0.2657,
      "step": 20690
    },
    {
      "epoch": 2.0358486390794424,
      "grad_norm": 0.9914331436157227,
      "learning_rate": 1.197591702102293e-05,
      "loss": 0.1636,
      "step": 20700
    },
    {
      "epoch": 2.036832140837452,
      "grad_norm": 0.28163185715675354,
      "learning_rate": 1.197194293208282e-05,
      "loss": 0.1733,
      "step": 20710
    },
    {
      "epoch": 2.037815642595461,
      "grad_norm": 1.0898809432983398,
      "learning_rate": 1.196796884314271e-05,
      "loss": 0.128,
      "step": 20720
    },
    {
      "epoch": 2.0387991443534705,
      "grad_norm": 0.48158377408981323,
      "learning_rate": 1.19639947542026e-05,
      "loss": 0.3055,
      "step": 20730
    },
    {
      "epoch": 2.03978264611148,
      "grad_norm": 0.6361274123191833,
      "learning_rate": 1.1960020665262491e-05,
      "loss": 0.2653,
      "step": 20740
    },
    {
      "epoch": 2.040766147869489,
      "grad_norm": 0.229063481092453,
      "learning_rate": 1.195604657632238e-05,
      "loss": 0.2892,
      "step": 20750
    },
    {
      "epoch": 2.0417496496274987,
      "grad_norm": 0.7816784381866455,
      "learning_rate": 1.1952072487382269e-05,
      "loss": 0.2426,
      "step": 20760
    },
    {
      "epoch": 2.0427331513855083,
      "grad_norm": 1.1882227659225464,
      "learning_rate": 1.1948098398442158e-05,
      "loss": 0.2364,
      "step": 20770
    },
    {
      "epoch": 2.0437166531435174,
      "grad_norm": 1.2545922994613647,
      "learning_rate": 1.1944124309502048e-05,
      "loss": 0.2674,
      "step": 20780
    },
    {
      "epoch": 2.044700154901527,
      "grad_norm": 1.2698802947998047,
      "learning_rate": 1.1940150220561938e-05,
      "loss": 0.2482,
      "step": 20790
    },
    {
      "epoch": 2.0456836566595364,
      "grad_norm": 1.024978756904602,
      "learning_rate": 1.1936176131621826e-05,
      "loss": 0.3956,
      "step": 20800
    },
    {
      "epoch": 2.0466671584175455,
      "grad_norm": 0.33348727226257324,
      "learning_rate": 1.1932202042681715e-05,
      "loss": 0.2243,
      "step": 20810
    },
    {
      "epoch": 2.047650660175555,
      "grad_norm": 2.104206085205078,
      "learning_rate": 1.1928227953741607e-05,
      "loss": 0.2308,
      "step": 20820
    },
    {
      "epoch": 2.0486341619335646,
      "grad_norm": 0.27045074105262756,
      "learning_rate": 1.1924253864801496e-05,
      "loss": 0.2153,
      "step": 20830
    },
    {
      "epoch": 2.0496176636915737,
      "grad_norm": 1.4179080724716187,
      "learning_rate": 1.1920279775861386e-05,
      "loss": 0.164,
      "step": 20840
    },
    {
      "epoch": 2.0506011654495833,
      "grad_norm": 0.8275901079177856,
      "learning_rate": 1.1916305686921274e-05,
      "loss": 0.3085,
      "step": 20850
    },
    {
      "epoch": 2.051584667207593,
      "grad_norm": 0.627906858921051,
      "learning_rate": 1.1912331597981164e-05,
      "loss": 0.1773,
      "step": 20860
    },
    {
      "epoch": 2.052568168965602,
      "grad_norm": 0.8649151921272278,
      "learning_rate": 1.1908357509041053e-05,
      "loss": 0.1718,
      "step": 20870
    },
    {
      "epoch": 2.0535516707236114,
      "grad_norm": 0.573418378829956,
      "learning_rate": 1.1904383420100943e-05,
      "loss": 0.2993,
      "step": 20880
    },
    {
      "epoch": 2.054535172481621,
      "grad_norm": 1.994998812675476,
      "learning_rate": 1.1900409331160831e-05,
      "loss": 0.2411,
      "step": 20890
    },
    {
      "epoch": 2.05551867423963,
      "grad_norm": 0.6588026285171509,
      "learning_rate": 1.189643524222072e-05,
      "loss": 0.1765,
      "step": 20900
    },
    {
      "epoch": 2.0565021759976396,
      "grad_norm": 1.3483917713165283,
      "learning_rate": 1.1892461153280612e-05,
      "loss": 0.3692,
      "step": 20910
    },
    {
      "epoch": 2.057485677755649,
      "grad_norm": 1.204897165298462,
      "learning_rate": 1.1888487064340502e-05,
      "loss": 0.1944,
      "step": 20920
    },
    {
      "epoch": 2.0584691795136583,
      "grad_norm": 0.5633821487426758,
      "learning_rate": 1.1884512975400391e-05,
      "loss": 0.2193,
      "step": 20930
    },
    {
      "epoch": 2.059452681271668,
      "grad_norm": 0.7229523658752441,
      "learning_rate": 1.188053888646028e-05,
      "loss": 0.1945,
      "step": 20940
    },
    {
      "epoch": 2.0604361830296773,
      "grad_norm": 3.430799961090088,
      "learning_rate": 1.1876564797520169e-05,
      "loss": 0.131,
      "step": 20950
    },
    {
      "epoch": 2.0614196847876864,
      "grad_norm": 0.41817110776901245,
      "learning_rate": 1.1872590708580059e-05,
      "loss": 0.3194,
      "step": 20960
    },
    {
      "epoch": 2.062403186545696,
      "grad_norm": 0.5513402223587036,
      "learning_rate": 1.1868616619639948e-05,
      "loss": 0.2289,
      "step": 20970
    },
    {
      "epoch": 2.0633866883037055,
      "grad_norm": 0.4411904513835907,
      "learning_rate": 1.1864642530699836e-05,
      "loss": 0.2082,
      "step": 20980
    },
    {
      "epoch": 2.0643701900617146,
      "grad_norm": 0.8698786497116089,
      "learning_rate": 1.186066844175973e-05,
      "loss": 0.199,
      "step": 20990
    },
    {
      "epoch": 2.065353691819724,
      "grad_norm": 0.9931325912475586,
      "learning_rate": 1.1856694352819617e-05,
      "loss": 0.1714,
      "step": 21000
    },
    {
      "epoch": 2.065353691819724,
      "eval_loss": 0.12633630633354187,
      "eval_runtime": 19.4678,
      "eval_samples_per_second": 2.568,
      "eval_steps_per_second": 1.284,
      "step": 21000
    },
    {
      "epoch": 2.0663371935777337,
      "grad_norm": 0.9795089960098267,
      "learning_rate": 1.1852720263879507e-05,
      "loss": 0.207,
      "step": 21010
    },
    {
      "epoch": 2.067320695335743,
      "grad_norm": 1.8761729001998901,
      "learning_rate": 1.1848746174939397e-05,
      "loss": 0.1822,
      "step": 21020
    },
    {
      "epoch": 2.0683041970937524,
      "grad_norm": 0.8495930433273315,
      "learning_rate": 1.1844772085999286e-05,
      "loss": 0.1868,
      "step": 21030
    },
    {
      "epoch": 2.069287698851762,
      "grad_norm": 2.16050386428833,
      "learning_rate": 1.1840797997059174e-05,
      "loss": 0.1159,
      "step": 21040
    },
    {
      "epoch": 2.070271200609771,
      "grad_norm": 0.6666444540023804,
      "learning_rate": 1.1836823908119064e-05,
      "loss": 0.1658,
      "step": 21050
    },
    {
      "epoch": 2.0712547023677805,
      "grad_norm": 0.551084041595459,
      "learning_rate": 1.1832849819178954e-05,
      "loss": 0.179,
      "step": 21060
    },
    {
      "epoch": 2.07223820412579,
      "grad_norm": 0.5672010183334351,
      "learning_rate": 1.1828875730238845e-05,
      "loss": 0.155,
      "step": 21070
    },
    {
      "epoch": 2.073221705883799,
      "grad_norm": 0.702887237071991,
      "learning_rate": 1.1824901641298735e-05,
      "loss": 0.2037,
      "step": 21080
    },
    {
      "epoch": 2.0742052076418087,
      "grad_norm": 0.5012316107749939,
      "learning_rate": 1.1820927552358622e-05,
      "loss": 0.2157,
      "step": 21090
    },
    {
      "epoch": 2.0751887093998183,
      "grad_norm": 1.0360255241394043,
      "learning_rate": 1.1816953463418512e-05,
      "loss": 0.259,
      "step": 21100
    },
    {
      "epoch": 2.0761722111578274,
      "grad_norm": 2.1426212787628174,
      "learning_rate": 1.1812979374478402e-05,
      "loss": 0.2645,
      "step": 21110
    },
    {
      "epoch": 2.077155712915837,
      "grad_norm": 0.9235216975212097,
      "learning_rate": 1.1809005285538291e-05,
      "loss": 0.1871,
      "step": 21120
    },
    {
      "epoch": 2.0781392146738464,
      "grad_norm": 2.114724636077881,
      "learning_rate": 1.180503119659818e-05,
      "loss": 0.2081,
      "step": 21130
    },
    {
      "epoch": 2.0791227164318555,
      "grad_norm": 0.5473008751869202,
      "learning_rate": 1.1801057107658069e-05,
      "loss": 0.2544,
      "step": 21140
    },
    {
      "epoch": 2.080106218189865,
      "grad_norm": 0.6295793652534485,
      "learning_rate": 1.179708301871796e-05,
      "loss": 0.2624,
      "step": 21150
    },
    {
      "epoch": 2.0810897199478746,
      "grad_norm": 1.548294186592102,
      "learning_rate": 1.179310892977785e-05,
      "loss": 0.2725,
      "step": 21160
    },
    {
      "epoch": 2.0820732217058837,
      "grad_norm": 1.4170395135879517,
      "learning_rate": 1.178913484083774e-05,
      "loss": 0.2087,
      "step": 21170
    },
    {
      "epoch": 2.0830567234638933,
      "grad_norm": 1.2778583765029907,
      "learning_rate": 1.1785160751897628e-05,
      "loss": 0.1622,
      "step": 21180
    },
    {
      "epoch": 2.084040225221903,
      "grad_norm": 0.39172619581222534,
      "learning_rate": 1.1781186662957517e-05,
      "loss": 0.0954,
      "step": 21190
    },
    {
      "epoch": 2.085023726979912,
      "grad_norm": 2.5332343578338623,
      "learning_rate": 1.1777212574017407e-05,
      "loss": 0.2202,
      "step": 21200
    },
    {
      "epoch": 2.0860072287379214,
      "grad_norm": 0.7690456509590149,
      "learning_rate": 1.1773238485077297e-05,
      "loss": 0.2473,
      "step": 21210
    },
    {
      "epoch": 2.0869907304959305,
      "grad_norm": 0.6942778825759888,
      "learning_rate": 1.1769264396137185e-05,
      "loss": 0.249,
      "step": 21220
    },
    {
      "epoch": 2.08797423225394,
      "grad_norm": 3.2177066802978516,
      "learning_rate": 1.1765290307197078e-05,
      "loss": 0.181,
      "step": 21230
    },
    {
      "epoch": 2.0889577340119496,
      "grad_norm": 0.6732898354530334,
      "learning_rate": 1.1761316218256966e-05,
      "loss": 0.1209,
      "step": 21240
    },
    {
      "epoch": 2.0899412357699587,
      "grad_norm": 1.3652567863464355,
      "learning_rate": 1.1757342129316855e-05,
      "loss": 0.1869,
      "step": 21250
    },
    {
      "epoch": 2.0909247375279683,
      "grad_norm": 1.0012937784194946,
      "learning_rate": 1.1753368040376745e-05,
      "loss": 0.231,
      "step": 21260
    },
    {
      "epoch": 2.091908239285978,
      "grad_norm": 0.48299527168273926,
      "learning_rate": 1.1749393951436635e-05,
      "loss": 0.2221,
      "step": 21270
    },
    {
      "epoch": 2.092891741043987,
      "grad_norm": 0.378261536359787,
      "learning_rate": 1.1745419862496523e-05,
      "loss": 0.166,
      "step": 21280
    },
    {
      "epoch": 2.0938752428019964,
      "grad_norm": 0.937616765499115,
      "learning_rate": 1.1741445773556412e-05,
      "loss": 0.1401,
      "step": 21290
    },
    {
      "epoch": 2.094858744560006,
      "grad_norm": 1.5174037218093872,
      "learning_rate": 1.1737471684616302e-05,
      "loss": 0.2337,
      "step": 21300
    },
    {
      "epoch": 2.095842246318015,
      "grad_norm": 1.0154869556427002,
      "learning_rate": 1.1733497595676193e-05,
      "loss": 0.1803,
      "step": 21310
    },
    {
      "epoch": 2.0968257480760246,
      "grad_norm": 1.1122788190841675,
      "learning_rate": 1.1729523506736083e-05,
      "loss": 0.2925,
      "step": 21320
    },
    {
      "epoch": 2.097809249834034,
      "grad_norm": 0.744653582572937,
      "learning_rate": 1.1725549417795971e-05,
      "loss": 0.1841,
      "step": 21330
    },
    {
      "epoch": 2.0987927515920433,
      "grad_norm": 1.2522811889648438,
      "learning_rate": 1.172157532885586e-05,
      "loss": 0.2478,
      "step": 21340
    },
    {
      "epoch": 2.099776253350053,
      "grad_norm": 2.0904698371887207,
      "learning_rate": 1.171760123991575e-05,
      "loss": 0.1925,
      "step": 21350
    },
    {
      "epoch": 2.1007597551080623,
      "grad_norm": 0.4512757360935211,
      "learning_rate": 1.171362715097564e-05,
      "loss": 0.3095,
      "step": 21360
    },
    {
      "epoch": 2.1017432568660714,
      "grad_norm": 0.8685400485992432,
      "learning_rate": 1.1709653062035528e-05,
      "loss": 0.2012,
      "step": 21370
    },
    {
      "epoch": 2.102726758624081,
      "grad_norm": 1.649245262145996,
      "learning_rate": 1.1705678973095418e-05,
      "loss": 0.0771,
      "step": 21380
    },
    {
      "epoch": 2.1037102603820905,
      "grad_norm": 2.175687789916992,
      "learning_rate": 1.1701704884155309e-05,
      "loss": 0.2466,
      "step": 21390
    },
    {
      "epoch": 2.1046937621400996,
      "grad_norm": 0.8011319637298584,
      "learning_rate": 1.1697730795215199e-05,
      "loss": 0.2011,
      "step": 21400
    },
    {
      "epoch": 2.105677263898109,
      "grad_norm": 0.9571422934532166,
      "learning_rate": 1.1693756706275088e-05,
      "loss": 0.2437,
      "step": 21410
    },
    {
      "epoch": 2.1066607656561187,
      "grad_norm": 0.3107330799102783,
      "learning_rate": 1.1689782617334976e-05,
      "loss": 0.2645,
      "step": 21420
    },
    {
      "epoch": 2.107644267414128,
      "grad_norm": 0.8494362235069275,
      "learning_rate": 1.1685808528394866e-05,
      "loss": 0.2183,
      "step": 21430
    },
    {
      "epoch": 2.1086277691721373,
      "grad_norm": 0.581502377986908,
      "learning_rate": 1.1681834439454756e-05,
      "loss": 0.1728,
      "step": 21440
    },
    {
      "epoch": 2.109611270930147,
      "grad_norm": 1.8993754386901855,
      "learning_rate": 1.1677860350514645e-05,
      "loss": 0.1762,
      "step": 21450
    },
    {
      "epoch": 2.110594772688156,
      "grad_norm": 2.3973135948181152,
      "learning_rate": 1.1673886261574533e-05,
      "loss": 0.1829,
      "step": 21460
    },
    {
      "epoch": 2.1115782744461655,
      "grad_norm": 1.4782636165618896,
      "learning_rate": 1.1669912172634426e-05,
      "loss": 0.2052,
      "step": 21470
    },
    {
      "epoch": 2.112561776204175,
      "grad_norm": 1.6421269178390503,
      "learning_rate": 1.1665938083694314e-05,
      "loss": 0.2584,
      "step": 21480
    },
    {
      "epoch": 2.113545277962184,
      "grad_norm": 0.34578168392181396,
      "learning_rate": 1.1661963994754204e-05,
      "loss": 0.2365,
      "step": 21490
    },
    {
      "epoch": 2.1145287797201937,
      "grad_norm": 0.6166962385177612,
      "learning_rate": 1.1657989905814094e-05,
      "loss": 0.2209,
      "step": 21500
    },
    {
      "epoch": 2.1145287797201937,
      "eval_loss": 0.12207618355751038,
      "eval_runtime": 19.3719,
      "eval_samples_per_second": 2.581,
      "eval_steps_per_second": 1.291,
      "step": 21500
    },
    {
      "epoch": 2.1155122814782032,
      "grad_norm": 1.2740017175674438,
      "learning_rate": 1.1654015816873983e-05,
      "loss": 0.1723,
      "step": 21510
    },
    {
      "epoch": 2.1164957832362123,
      "grad_norm": 1.4409515857696533,
      "learning_rate": 1.1650041727933871e-05,
      "loss": 0.1431,
      "step": 21520
    },
    {
      "epoch": 2.117479284994222,
      "grad_norm": 0.5763534307479858,
      "learning_rate": 1.1646067638993761e-05,
      "loss": 0.2046,
      "step": 21530
    },
    {
      "epoch": 2.1184627867522314,
      "grad_norm": 1.4775296449661255,
      "learning_rate": 1.164209355005365e-05,
      "loss": 0.2873,
      "step": 21540
    },
    {
      "epoch": 2.1194462885102405,
      "grad_norm": 1.134960412979126,
      "learning_rate": 1.1638119461113542e-05,
      "loss": 0.2338,
      "step": 21550
    },
    {
      "epoch": 2.12042979026825,
      "grad_norm": 1.5541033744812012,
      "learning_rate": 1.1634145372173432e-05,
      "loss": 0.2024,
      "step": 21560
    },
    {
      "epoch": 2.1214132920262596,
      "grad_norm": 0.3558676838874817,
      "learning_rate": 1.163017128323332e-05,
      "loss": 0.2318,
      "step": 21570
    },
    {
      "epoch": 2.1223967937842687,
      "grad_norm": 0.7510431408882141,
      "learning_rate": 1.162619719429321e-05,
      "loss": 0.2736,
      "step": 21580
    },
    {
      "epoch": 2.1233802955422783,
      "grad_norm": 0.7550353407859802,
      "learning_rate": 1.1622223105353099e-05,
      "loss": 0.1648,
      "step": 21590
    },
    {
      "epoch": 2.124363797300288,
      "grad_norm": 2.832073450088501,
      "learning_rate": 1.1618249016412989e-05,
      "loss": 0.1297,
      "step": 21600
    },
    {
      "epoch": 2.125347299058297,
      "grad_norm": 1.1138674020767212,
      "learning_rate": 1.1614274927472877e-05,
      "loss": 0.1841,
      "step": 21610
    },
    {
      "epoch": 2.1263308008163064,
      "grad_norm": 0.8778446912765503,
      "learning_rate": 1.1610300838532766e-05,
      "loss": 0.2238,
      "step": 21620
    },
    {
      "epoch": 2.127314302574316,
      "grad_norm": 1.0984127521514893,
      "learning_rate": 1.1606326749592658e-05,
      "loss": 0.1756,
      "step": 21630
    },
    {
      "epoch": 2.128297804332325,
      "grad_norm": 1.374091386795044,
      "learning_rate": 1.1602352660652547e-05,
      "loss": 0.1335,
      "step": 21640
    },
    {
      "epoch": 2.1292813060903346,
      "grad_norm": 0.6483007669448853,
      "learning_rate": 1.1598378571712437e-05,
      "loss": 0.2623,
      "step": 21650
    },
    {
      "epoch": 2.130264807848344,
      "grad_norm": 0.8100119233131409,
      "learning_rate": 1.1594404482772325e-05,
      "loss": 0.1401,
      "step": 21660
    },
    {
      "epoch": 2.1312483096063533,
      "grad_norm": 0.619308352470398,
      "learning_rate": 1.1590430393832215e-05,
      "loss": 0.2544,
      "step": 21670
    },
    {
      "epoch": 2.132231811364363,
      "grad_norm": 0.19790104031562805,
      "learning_rate": 1.1586456304892104e-05,
      "loss": 0.2049,
      "step": 21680
    },
    {
      "epoch": 2.1332153131223723,
      "grad_norm": 0.6911652684211731,
      "learning_rate": 1.1582482215951994e-05,
      "loss": 0.2325,
      "step": 21690
    },
    {
      "epoch": 2.1341988148803814,
      "grad_norm": 0.9950113296508789,
      "learning_rate": 1.1578508127011882e-05,
      "loss": 0.2132,
      "step": 21700
    },
    {
      "epoch": 2.135182316638391,
      "grad_norm": 1.3821048736572266,
      "learning_rate": 1.1574534038071772e-05,
      "loss": 0.2346,
      "step": 21710
    },
    {
      "epoch": 2.1361658183964005,
      "grad_norm": 0.7430508732795715,
      "learning_rate": 1.1570559949131663e-05,
      "loss": 0.2575,
      "step": 21720
    },
    {
      "epoch": 2.1371493201544096,
      "grad_norm": 2.4440536499023438,
      "learning_rate": 1.1566585860191553e-05,
      "loss": 0.2214,
      "step": 21730
    },
    {
      "epoch": 2.138132821912419,
      "grad_norm": 4.348588466644287,
      "learning_rate": 1.1562611771251442e-05,
      "loss": 0.1869,
      "step": 21740
    },
    {
      "epoch": 2.1391163236704287,
      "grad_norm": 0.4643959403038025,
      "learning_rate": 1.1558637682311332e-05,
      "loss": 0.2129,
      "step": 21750
    },
    {
      "epoch": 2.140099825428438,
      "grad_norm": 1.3620189428329468,
      "learning_rate": 1.155466359337122e-05,
      "loss": 0.2255,
      "step": 21760
    },
    {
      "epoch": 2.1410833271864473,
      "grad_norm": 1.957170844078064,
      "learning_rate": 1.155068950443111e-05,
      "loss": 0.1781,
      "step": 21770
    },
    {
      "epoch": 2.142066828944457,
      "grad_norm": 1.7803796529769897,
      "learning_rate": 1.1546715415491e-05,
      "loss": 0.3,
      "step": 21780
    },
    {
      "epoch": 2.143050330702466,
      "grad_norm": 0.6519695520401001,
      "learning_rate": 1.1542741326550889e-05,
      "loss": 0.1946,
      "step": 21790
    },
    {
      "epoch": 2.1440338324604755,
      "grad_norm": 0.34365659952163696,
      "learning_rate": 1.153876723761078e-05,
      "loss": 0.3447,
      "step": 21800
    },
    {
      "epoch": 2.145017334218485,
      "grad_norm": 1.302046298980713,
      "learning_rate": 1.1534793148670668e-05,
      "loss": 0.3466,
      "step": 21810
    },
    {
      "epoch": 2.146000835976494,
      "grad_norm": 0.7232846617698669,
      "learning_rate": 1.1530819059730558e-05,
      "loss": 0.2998,
      "step": 21820
    },
    {
      "epoch": 2.1469843377345037,
      "grad_norm": 1.635127305984497,
      "learning_rate": 1.1526844970790448e-05,
      "loss": 0.1282,
      "step": 21830
    },
    {
      "epoch": 2.1479678394925132,
      "grad_norm": 1.3019146919250488,
      "learning_rate": 1.1522870881850337e-05,
      "loss": 0.2473,
      "step": 21840
    },
    {
      "epoch": 2.1489513412505223,
      "grad_norm": 0.8861162066459656,
      "learning_rate": 1.1518896792910225e-05,
      "loss": 0.1754,
      "step": 21850
    },
    {
      "epoch": 2.149934843008532,
      "grad_norm": 1.6959495544433594,
      "learning_rate": 1.1514922703970115e-05,
      "loss": 0.1867,
      "step": 21860
    },
    {
      "epoch": 2.1509183447665414,
      "grad_norm": 0.5939361453056335,
      "learning_rate": 1.1510948615030004e-05,
      "loss": 0.2335,
      "step": 21870
    },
    {
      "epoch": 2.1519018465245505,
      "grad_norm": 0.9883837699890137,
      "learning_rate": 1.1506974526089896e-05,
      "loss": 0.1448,
      "step": 21880
    },
    {
      "epoch": 2.15288534828256,
      "grad_norm": 0.3421158194541931,
      "learning_rate": 1.1503000437149786e-05,
      "loss": 0.1313,
      "step": 21890
    },
    {
      "epoch": 2.1538688500405696,
      "grad_norm": 0.8726436495780945,
      "learning_rate": 1.1499026348209673e-05,
      "loss": 0.1874,
      "step": 21900
    },
    {
      "epoch": 2.1548523517985787,
      "grad_norm": 1.3549585342407227,
      "learning_rate": 1.1495052259269563e-05,
      "loss": 0.1903,
      "step": 21910
    },
    {
      "epoch": 2.1558358535565882,
      "grad_norm": 1.9957268238067627,
      "learning_rate": 1.1491078170329453e-05,
      "loss": 0.2602,
      "step": 21920
    },
    {
      "epoch": 2.156819355314598,
      "grad_norm": 0.9010176658630371,
      "learning_rate": 1.1487104081389342e-05,
      "loss": 0.2442,
      "step": 21930
    },
    {
      "epoch": 2.157802857072607,
      "grad_norm": 0.6982910633087158,
      "learning_rate": 1.148312999244923e-05,
      "loss": 0.1611,
      "step": 21940
    },
    {
      "epoch": 2.1587863588306164,
      "grad_norm": 0.8347655534744263,
      "learning_rate": 1.147915590350912e-05,
      "loss": 0.2313,
      "step": 21950
    },
    {
      "epoch": 2.159769860588626,
      "grad_norm": 3.3864970207214355,
      "learning_rate": 1.1475181814569011e-05,
      "loss": 0.2354,
      "step": 21960
    },
    {
      "epoch": 2.160753362346635,
      "grad_norm": 2.0524072647094727,
      "learning_rate": 1.1471207725628901e-05,
      "loss": 0.303,
      "step": 21970
    },
    {
      "epoch": 2.1617368641046446,
      "grad_norm": 0.5293691158294678,
      "learning_rate": 1.146723363668879e-05,
      "loss": 0.1661,
      "step": 21980
    },
    {
      "epoch": 2.162720365862654,
      "grad_norm": 0.9710631966590881,
      "learning_rate": 1.146325954774868e-05,
      "loss": 0.1995,
      "step": 21990
    },
    {
      "epoch": 2.1637038676206632,
      "grad_norm": 4.298491477966309,
      "learning_rate": 1.1459285458808568e-05,
      "loss": 0.126,
      "step": 22000
    },
    {
      "epoch": 2.1637038676206632,
      "eval_loss": 0.12008929252624512,
      "eval_runtime": 22.1618,
      "eval_samples_per_second": 2.256,
      "eval_steps_per_second": 1.128,
      "step": 22000
    },
    {
      "epoch": 2.164687369378673,
      "grad_norm": 1.042594313621521,
      "learning_rate": 1.1455311369868458e-05,
      "loss": 0.0853,
      "step": 22010
    },
    {
      "epoch": 2.1656708711366823,
      "grad_norm": 0.550438404083252,
      "learning_rate": 1.1451337280928348e-05,
      "loss": 0.2328,
      "step": 22020
    },
    {
      "epoch": 2.1666543728946914,
      "grad_norm": 2.2264037132263184,
      "learning_rate": 1.1447363191988237e-05,
      "loss": 0.2319,
      "step": 22030
    },
    {
      "epoch": 2.167637874652701,
      "grad_norm": 0.7744798064231873,
      "learning_rate": 1.1443389103048129e-05,
      "loss": 0.1784,
      "step": 22040
    },
    {
      "epoch": 2.1686213764107105,
      "grad_norm": 0.25838974118232727,
      "learning_rate": 1.1439415014108017e-05,
      "loss": 0.1522,
      "step": 22050
    },
    {
      "epoch": 2.1696048781687196,
      "grad_norm": 1.460982084274292,
      "learning_rate": 1.1435440925167906e-05,
      "loss": 0.1469,
      "step": 22060
    },
    {
      "epoch": 2.170588379926729,
      "grad_norm": 0.9990485310554504,
      "learning_rate": 1.1431466836227796e-05,
      "loss": 0.2747,
      "step": 22070
    },
    {
      "epoch": 2.1715718816847387,
      "grad_norm": 1.4448250532150269,
      "learning_rate": 1.1427492747287686e-05,
      "loss": 0.1212,
      "step": 22080
    },
    {
      "epoch": 2.172555383442748,
      "grad_norm": 2.9366111755371094,
      "learning_rate": 1.1423518658347574e-05,
      "loss": 0.2416,
      "step": 22090
    },
    {
      "epoch": 2.1735388852007573,
      "grad_norm": 1.9704415798187256,
      "learning_rate": 1.1419544569407463e-05,
      "loss": 0.2473,
      "step": 22100
    },
    {
      "epoch": 2.174522386958767,
      "grad_norm": 0.8068645000457764,
      "learning_rate": 1.1415570480467353e-05,
      "loss": 0.216,
      "step": 22110
    },
    {
      "epoch": 2.175505888716776,
      "grad_norm": 1.2421001195907593,
      "learning_rate": 1.1411596391527244e-05,
      "loss": 0.2233,
      "step": 22120
    },
    {
      "epoch": 2.1764893904747855,
      "grad_norm": 1.0471240282058716,
      "learning_rate": 1.1407622302587134e-05,
      "loss": 0.2244,
      "step": 22130
    },
    {
      "epoch": 2.177472892232795,
      "grad_norm": 0.49878430366516113,
      "learning_rate": 1.1403648213647022e-05,
      "loss": 0.23,
      "step": 22140
    },
    {
      "epoch": 2.178456393990804,
      "grad_norm": 1.0834639072418213,
      "learning_rate": 1.1399674124706912e-05,
      "loss": 0.1489,
      "step": 22150
    },
    {
      "epoch": 2.1794398957488137,
      "grad_norm": 1.5779012441635132,
      "learning_rate": 1.1395700035766801e-05,
      "loss": 0.2697,
      "step": 22160
    },
    {
      "epoch": 2.1804233975068232,
      "grad_norm": 4.6256279945373535,
      "learning_rate": 1.1391725946826691e-05,
      "loss": 0.1769,
      "step": 22170
    },
    {
      "epoch": 2.1814068992648323,
      "grad_norm": 2.0662331581115723,
      "learning_rate": 1.1387751857886579e-05,
      "loss": 0.213,
      "step": 22180
    },
    {
      "epoch": 2.182390401022842,
      "grad_norm": 0.6344393491744995,
      "learning_rate": 1.1383777768946469e-05,
      "loss": 0.3054,
      "step": 22190
    },
    {
      "epoch": 2.1833739027808514,
      "grad_norm": 1.839341640472412,
      "learning_rate": 1.137980368000636e-05,
      "loss": 0.1929,
      "step": 22200
    },
    {
      "epoch": 2.1843574045388605,
      "grad_norm": 2.015293836593628,
      "learning_rate": 1.137582959106625e-05,
      "loss": 0.2403,
      "step": 22210
    },
    {
      "epoch": 2.18534090629687,
      "grad_norm": 0.6259804964065552,
      "learning_rate": 1.137185550212614e-05,
      "loss": 0.2405,
      "step": 22220
    },
    {
      "epoch": 2.1863244080548796,
      "grad_norm": 0.1938798427581787,
      "learning_rate": 1.1367881413186029e-05,
      "loss": 0.1978,
      "step": 22230
    },
    {
      "epoch": 2.1873079098128887,
      "grad_norm": 0.419461190700531,
      "learning_rate": 1.1363907324245917e-05,
      "loss": 0.2166,
      "step": 22240
    },
    {
      "epoch": 2.1882914115708982,
      "grad_norm": 1.0594192743301392,
      "learning_rate": 1.1359933235305807e-05,
      "loss": 0.287,
      "step": 22250
    },
    {
      "epoch": 2.189274913328908,
      "grad_norm": 1.3059344291687012,
      "learning_rate": 1.1355959146365696e-05,
      "loss": 0.2602,
      "step": 22260
    },
    {
      "epoch": 2.190258415086917,
      "grad_norm": 0.9751050472259521,
      "learning_rate": 1.1351985057425586e-05,
      "loss": 0.1706,
      "step": 22270
    },
    {
      "epoch": 2.1912419168449264,
      "grad_norm": 0.6977351903915405,
      "learning_rate": 1.1348010968485477e-05,
      "loss": 0.2563,
      "step": 22280
    },
    {
      "epoch": 2.192225418602936,
      "grad_norm": 0.644167959690094,
      "learning_rate": 1.1344036879545365e-05,
      "loss": 0.1823,
      "step": 22290
    },
    {
      "epoch": 2.193208920360945,
      "grad_norm": 0.7303948998451233,
      "learning_rate": 1.1340062790605255e-05,
      "loss": 0.24,
      "step": 22300
    },
    {
      "epoch": 2.1941924221189546,
      "grad_norm": 1.255916953086853,
      "learning_rate": 1.1336088701665145e-05,
      "loss": 0.2991,
      "step": 22310
    },
    {
      "epoch": 2.195175923876964,
      "grad_norm": 1.121828317642212,
      "learning_rate": 1.1332114612725034e-05,
      "loss": 0.1636,
      "step": 22320
    },
    {
      "epoch": 2.1961594256349732,
      "grad_norm": 0.6688191890716553,
      "learning_rate": 1.1328140523784922e-05,
      "loss": 0.2715,
      "step": 22330
    },
    {
      "epoch": 2.197142927392983,
      "grad_norm": 1.378141164779663,
      "learning_rate": 1.1324166434844812e-05,
      "loss": 0.1942,
      "step": 22340
    },
    {
      "epoch": 2.1981264291509923,
      "grad_norm": 0.4261777698993683,
      "learning_rate": 1.1320192345904702e-05,
      "loss": 0.2135,
      "step": 22350
    },
    {
      "epoch": 2.1991099309090014,
      "grad_norm": 0.3720337450504303,
      "learning_rate": 1.1316218256964593e-05,
      "loss": 0.262,
      "step": 22360
    },
    {
      "epoch": 2.200093432667011,
      "grad_norm": 0.6999090909957886,
      "learning_rate": 1.1312244168024483e-05,
      "loss": 0.1933,
      "step": 22370
    },
    {
      "epoch": 2.2010769344250205,
      "grad_norm": 0.9601486921310425,
      "learning_rate": 1.130827007908437e-05,
      "loss": 0.3139,
      "step": 22380
    },
    {
      "epoch": 2.2020604361830296,
      "grad_norm": 0.9890156984329224,
      "learning_rate": 1.130429599014426e-05,
      "loss": 0.217,
      "step": 22390
    },
    {
      "epoch": 2.203043937941039,
      "grad_norm": 1.0042225122451782,
      "learning_rate": 1.130032190120415e-05,
      "loss": 0.29,
      "step": 22400
    },
    {
      "epoch": 2.2040274396990487,
      "grad_norm": 0.8810020089149475,
      "learning_rate": 1.129634781226404e-05,
      "loss": 0.2314,
      "step": 22410
    },
    {
      "epoch": 2.205010941457058,
      "grad_norm": 3.3902359008789062,
      "learning_rate": 1.1292373723323928e-05,
      "loss": 0.3176,
      "step": 22420
    },
    {
      "epoch": 2.2059944432150673,
      "grad_norm": 0.7750841379165649,
      "learning_rate": 1.1288399634383817e-05,
      "loss": 0.2395,
      "step": 22430
    },
    {
      "epoch": 2.206977944973077,
      "grad_norm": 0.47941309213638306,
      "learning_rate": 1.1284425545443709e-05,
      "loss": 0.2485,
      "step": 22440
    },
    {
      "epoch": 2.207961446731086,
      "grad_norm": 0.6404972076416016,
      "learning_rate": 1.1280451456503598e-05,
      "loss": 0.193,
      "step": 22450
    },
    {
      "epoch": 2.2089449484890955,
      "grad_norm": 0.5544617176055908,
      "learning_rate": 1.1276477367563488e-05,
      "loss": 0.2392,
      "step": 22460
    },
    {
      "epoch": 2.209928450247105,
      "grad_norm": 0.6468740701675415,
      "learning_rate": 1.1272503278623378e-05,
      "loss": 0.1146,
      "step": 22470
    },
    {
      "epoch": 2.210911952005114,
      "grad_norm": 1.14541494846344,
      "learning_rate": 1.1268529189683266e-05,
      "loss": 0.1455,
      "step": 22480
    },
    {
      "epoch": 2.2118954537631237,
      "grad_norm": 1.0405633449554443,
      "learning_rate": 1.1264555100743155e-05,
      "loss": 0.2719,
      "step": 22490
    },
    {
      "epoch": 2.2128789555211332,
      "grad_norm": 0.6183965802192688,
      "learning_rate": 1.1260581011803045e-05,
      "loss": 0.3315,
      "step": 22500
    },
    {
      "epoch": 2.2128789555211332,
      "eval_loss": 0.12000677734613419,
      "eval_runtime": 16.723,
      "eval_samples_per_second": 2.99,
      "eval_steps_per_second": 1.495,
      "step": 22500
    },
    {
      "epoch": 2.2138624572791423,
      "grad_norm": 0.9129937887191772,
      "learning_rate": 1.1256606922862935e-05,
      "loss": 0.2667,
      "step": 22510
    },
    {
      "epoch": 2.214845959037152,
      "grad_norm": 0.31970441341400146,
      "learning_rate": 1.1252632833922823e-05,
      "loss": 0.0376,
      "step": 22520
    },
    {
      "epoch": 2.215829460795161,
      "grad_norm": 0.9021411538124084,
      "learning_rate": 1.1248658744982714e-05,
      "loss": 0.2351,
      "step": 22530
    },
    {
      "epoch": 2.2168129625531705,
      "grad_norm": 2.631866455078125,
      "learning_rate": 1.1244684656042604e-05,
      "loss": 0.2439,
      "step": 22540
    },
    {
      "epoch": 2.21779646431118,
      "grad_norm": 0.9322601556777954,
      "learning_rate": 1.1240710567102493e-05,
      "loss": 0.2018,
      "step": 22550
    },
    {
      "epoch": 2.218779966069189,
      "grad_norm": 0.3407883942127228,
      "learning_rate": 1.1236736478162383e-05,
      "loss": 0.1916,
      "step": 22560
    },
    {
      "epoch": 2.2197634678271987,
      "grad_norm": 3.0421650409698486,
      "learning_rate": 1.1232762389222271e-05,
      "loss": 0.1731,
      "step": 22570
    },
    {
      "epoch": 2.2207469695852082,
      "grad_norm": 1.467153787612915,
      "learning_rate": 1.122878830028216e-05,
      "loss": 0.2077,
      "step": 22580
    },
    {
      "epoch": 2.2217304713432173,
      "grad_norm": 2.4003944396972656,
      "learning_rate": 1.122481421134205e-05,
      "loss": 0.2574,
      "step": 22590
    },
    {
      "epoch": 2.222713973101227,
      "grad_norm": 0.5600699782371521,
      "learning_rate": 1.122084012240194e-05,
      "loss": 0.2947,
      "step": 22600
    },
    {
      "epoch": 2.2236974748592364,
      "grad_norm": 0.11170317977666855,
      "learning_rate": 1.1216866033461831e-05,
      "loss": 0.2289,
      "step": 22610
    },
    {
      "epoch": 2.2246809766172455,
      "grad_norm": 0.7878932952880859,
      "learning_rate": 1.121289194452172e-05,
      "loss": 0.2115,
      "step": 22620
    },
    {
      "epoch": 2.225664478375255,
      "grad_norm": 1.297023892402649,
      "learning_rate": 1.1208917855581609e-05,
      "loss": 0.315,
      "step": 22630
    },
    {
      "epoch": 2.2266479801332646,
      "grad_norm": 0.8216570615768433,
      "learning_rate": 1.1204943766641499e-05,
      "loss": 0.2609,
      "step": 22640
    },
    {
      "epoch": 2.2276314818912737,
      "grad_norm": 1.1088916063308716,
      "learning_rate": 1.1200969677701388e-05,
      "loss": 0.1976,
      "step": 22650
    },
    {
      "epoch": 2.2286149836492832,
      "grad_norm": 3.3971316814422607,
      "learning_rate": 1.1196995588761276e-05,
      "loss": 0.1781,
      "step": 22660
    },
    {
      "epoch": 2.2295984854072928,
      "grad_norm": 2.520642042160034,
      "learning_rate": 1.1193021499821166e-05,
      "loss": 0.217,
      "step": 22670
    },
    {
      "epoch": 2.230581987165302,
      "grad_norm": 0.9775090217590332,
      "learning_rate": 1.1189047410881055e-05,
      "loss": 0.1775,
      "step": 22680
    },
    {
      "epoch": 2.2315654889233114,
      "grad_norm": 0.7490184307098389,
      "learning_rate": 1.1185073321940947e-05,
      "loss": 0.1408,
      "step": 22690
    },
    {
      "epoch": 2.232548990681321,
      "grad_norm": 3.4259045124053955,
      "learning_rate": 1.1181099233000837e-05,
      "loss": 0.1902,
      "step": 22700
    },
    {
      "epoch": 2.23353249243933,
      "grad_norm": 2.450721502304077,
      "learning_rate": 1.1177125144060726e-05,
      "loss": 0.2047,
      "step": 22710
    },
    {
      "epoch": 2.2345159941973396,
      "grad_norm": 0.7697030305862427,
      "learning_rate": 1.1173151055120614e-05,
      "loss": 0.2477,
      "step": 22720
    },
    {
      "epoch": 2.235499495955349,
      "grad_norm": 1.226370930671692,
      "learning_rate": 1.1169176966180504e-05,
      "loss": 0.1745,
      "step": 22730
    },
    {
      "epoch": 2.2364829977133582,
      "grad_norm": 1.002942681312561,
      "learning_rate": 1.1165202877240393e-05,
      "loss": 0.1485,
      "step": 22740
    },
    {
      "epoch": 2.2374664994713678,
      "grad_norm": 0.6475685834884644,
      "learning_rate": 1.1161228788300283e-05,
      "loss": 0.2848,
      "step": 22750
    },
    {
      "epoch": 2.2384500012293773,
      "grad_norm": 1.8169118165969849,
      "learning_rate": 1.1157254699360171e-05,
      "loss": 0.2519,
      "step": 22760
    },
    {
      "epoch": 2.2394335029873864,
      "grad_norm": 1.4185065031051636,
      "learning_rate": 1.1153280610420062e-05,
      "loss": 0.1619,
      "step": 22770
    },
    {
      "epoch": 2.240417004745396,
      "grad_norm": 0.21674694120883942,
      "learning_rate": 1.1149306521479952e-05,
      "loss": 0.2106,
      "step": 22780
    },
    {
      "epoch": 2.2414005065034055,
      "grad_norm": 1.7250815629959106,
      "learning_rate": 1.1145332432539842e-05,
      "loss": 0.2229,
      "step": 22790
    },
    {
      "epoch": 2.2423840082614146,
      "grad_norm": 0.9569165706634521,
      "learning_rate": 1.1141358343599731e-05,
      "loss": 0.2533,
      "step": 22800
    },
    {
      "epoch": 2.243367510019424,
      "grad_norm": 1.190146803855896,
      "learning_rate": 1.113738425465962e-05,
      "loss": 0.2204,
      "step": 22810
    },
    {
      "epoch": 2.2443510117774337,
      "grad_norm": 2.0811798572540283,
      "learning_rate": 1.1133410165719509e-05,
      "loss": 0.2003,
      "step": 22820
    },
    {
      "epoch": 2.2453345135354428,
      "grad_norm": 0.4453285038471222,
      "learning_rate": 1.1129436076779399e-05,
      "loss": 0.1955,
      "step": 22830
    },
    {
      "epoch": 2.2463180152934523,
      "grad_norm": 0.8300244808197021,
      "learning_rate": 1.1125461987839288e-05,
      "loss": 0.1842,
      "step": 22840
    },
    {
      "epoch": 2.247301517051462,
      "grad_norm": 1.4930312633514404,
      "learning_rate": 1.112148789889918e-05,
      "loss": 0.2175,
      "step": 22850
    },
    {
      "epoch": 2.248285018809471,
      "grad_norm": 0.8433923125267029,
      "learning_rate": 1.1117513809959068e-05,
      "loss": 0.2786,
      "step": 22860
    },
    {
      "epoch": 2.2492685205674805,
      "grad_norm": 0.6747217178344727,
      "learning_rate": 1.1113539721018957e-05,
      "loss": 0.2544,
      "step": 22870
    },
    {
      "epoch": 2.25025202232549,
      "grad_norm": 1.8815783262252808,
      "learning_rate": 1.1109565632078847e-05,
      "loss": 0.1326,
      "step": 22880
    },
    {
      "epoch": 2.251235524083499,
      "grad_norm": 1.1676626205444336,
      "learning_rate": 1.1105591543138737e-05,
      "loss": 0.1685,
      "step": 22890
    },
    {
      "epoch": 2.2522190258415087,
      "grad_norm": 0.9920549392700195,
      "learning_rate": 1.1101617454198625e-05,
      "loss": 0.1928,
      "step": 22900
    },
    {
      "epoch": 2.2532025275995182,
      "grad_norm": 0.5512763857841492,
      "learning_rate": 1.1097643365258514e-05,
      "loss": 0.1643,
      "step": 22910
    },
    {
      "epoch": 2.2541860293575273,
      "grad_norm": 3.966198444366455,
      "learning_rate": 1.1093669276318404e-05,
      "loss": 0.1933,
      "step": 22920
    },
    {
      "epoch": 2.255169531115537,
      "grad_norm": 1.7686697244644165,
      "learning_rate": 1.1089695187378295e-05,
      "loss": 0.1953,
      "step": 22930
    },
    {
      "epoch": 2.2561530328735464,
      "grad_norm": 1.5562306642532349,
      "learning_rate": 1.1085721098438185e-05,
      "loss": 0.1546,
      "step": 22940
    },
    {
      "epoch": 2.2571365346315555,
      "grad_norm": 1.2184745073318481,
      "learning_rate": 1.1081747009498075e-05,
      "loss": 0.1791,
      "step": 22950
    },
    {
      "epoch": 2.258120036389565,
      "grad_norm": 2.333268642425537,
      "learning_rate": 1.1077772920557963e-05,
      "loss": 0.2178,
      "step": 22960
    },
    {
      "epoch": 2.2591035381475746,
      "grad_norm": 1.0209434032440186,
      "learning_rate": 1.1073798831617852e-05,
      "loss": 0.2372,
      "step": 22970
    },
    {
      "epoch": 2.2600870399055837,
      "grad_norm": 0.5352984070777893,
      "learning_rate": 1.1069824742677742e-05,
      "loss": 0.2114,
      "step": 22980
    },
    {
      "epoch": 2.2610705416635932,
      "grad_norm": 0.33463746309280396,
      "learning_rate": 1.1065850653737632e-05,
      "loss": 0.1422,
      "step": 22990
    },
    {
      "epoch": 2.2620540434216028,
      "grad_norm": 1.2513502836227417,
      "learning_rate": 1.106187656479752e-05,
      "loss": 0.2779,
      "step": 23000
    },
    {
      "epoch": 2.2620540434216028,
      "eval_loss": 0.12226882576942444,
      "eval_runtime": 19.3685,
      "eval_samples_per_second": 2.582,
      "eval_steps_per_second": 1.291,
      "step": 23000
    },
    {
      "epoch": 2.263037545179612,
      "grad_norm": 1.8305885791778564,
      "learning_rate": 1.1057902475857411e-05,
      "loss": 0.1699,
      "step": 23010
    },
    {
      "epoch": 2.2640210469376214,
      "grad_norm": 0.9677567481994629,
      "learning_rate": 1.10539283869173e-05,
      "loss": 0.1796,
      "step": 23020
    },
    {
      "epoch": 2.265004548695631,
      "grad_norm": 0.470826119184494,
      "learning_rate": 1.104995429797719e-05,
      "loss": 0.1581,
      "step": 23030
    },
    {
      "epoch": 2.26598805045364,
      "grad_norm": 0.6805540919303894,
      "learning_rate": 1.104598020903708e-05,
      "loss": 0.2149,
      "step": 23040
    },
    {
      "epoch": 2.2669715522116496,
      "grad_norm": 1.141538381576538,
      "learning_rate": 1.1042006120096968e-05,
      "loss": 0.2186,
      "step": 23050
    },
    {
      "epoch": 2.267955053969659,
      "grad_norm": 0.43806010484695435,
      "learning_rate": 1.1038032031156858e-05,
      "loss": 0.2429,
      "step": 23060
    },
    {
      "epoch": 2.2689385557276682,
      "grad_norm": 0.746082067489624,
      "learning_rate": 1.1034057942216747e-05,
      "loss": 0.2457,
      "step": 23070
    },
    {
      "epoch": 2.2699220574856778,
      "grad_norm": 1.439133882522583,
      "learning_rate": 1.1030083853276637e-05,
      "loss": 0.1903,
      "step": 23080
    },
    {
      "epoch": 2.2709055592436873,
      "grad_norm": 0.9891164302825928,
      "learning_rate": 1.1026109764336528e-05,
      "loss": 0.3032,
      "step": 23090
    },
    {
      "epoch": 2.2718890610016964,
      "grad_norm": 0.4387306869029999,
      "learning_rate": 1.1022135675396416e-05,
      "loss": 0.2149,
      "step": 23100
    },
    {
      "epoch": 2.272872562759706,
      "grad_norm": 0.6124128103256226,
      "learning_rate": 1.1018161586456306e-05,
      "loss": 0.1631,
      "step": 23110
    },
    {
      "epoch": 2.2738560645177155,
      "grad_norm": 0.19758909940719604,
      "learning_rate": 1.1014187497516196e-05,
      "loss": 0.1581,
      "step": 23120
    },
    {
      "epoch": 2.2748395662757246,
      "grad_norm": 1.1576956510543823,
      "learning_rate": 1.1010213408576085e-05,
      "loss": 0.1836,
      "step": 23130
    },
    {
      "epoch": 2.275823068033734,
      "grad_norm": 1.2810118198394775,
      "learning_rate": 1.1006239319635973e-05,
      "loss": 0.218,
      "step": 23140
    },
    {
      "epoch": 2.2768065697917437,
      "grad_norm": 2.966301918029785,
      "learning_rate": 1.1002265230695863e-05,
      "loss": 0.2945,
      "step": 23150
    },
    {
      "epoch": 2.2777900715497528,
      "grad_norm": 0.5344410538673401,
      "learning_rate": 1.0998291141755753e-05,
      "loss": 0.2595,
      "step": 23160
    },
    {
      "epoch": 2.2787735733077623,
      "grad_norm": 1.2781692743301392,
      "learning_rate": 1.0994317052815644e-05,
      "loss": 0.1449,
      "step": 23170
    },
    {
      "epoch": 2.279757075065772,
      "grad_norm": 0.9891101717948914,
      "learning_rate": 1.0990342963875534e-05,
      "loss": 0.2795,
      "step": 23180
    },
    {
      "epoch": 2.280740576823781,
      "grad_norm": 8.33986759185791,
      "learning_rate": 1.0986368874935423e-05,
      "loss": 0.1602,
      "step": 23190
    },
    {
      "epoch": 2.2817240785817905,
      "grad_norm": 2.056553602218628,
      "learning_rate": 1.0982394785995311e-05,
      "loss": 0.2283,
      "step": 23200
    },
    {
      "epoch": 2.2827075803398,
      "grad_norm": 0.5050723552703857,
      "learning_rate": 1.0978420697055201e-05,
      "loss": 0.1988,
      "step": 23210
    },
    {
      "epoch": 2.283691082097809,
      "grad_norm": 0.78752601146698,
      "learning_rate": 1.097444660811509e-05,
      "loss": 0.1589,
      "step": 23220
    },
    {
      "epoch": 2.2846745838558187,
      "grad_norm": 1.4518001079559326,
      "learning_rate": 1.097047251917498e-05,
      "loss": 0.1503,
      "step": 23230
    },
    {
      "epoch": 2.285658085613828,
      "grad_norm": 1.8836430311203003,
      "learning_rate": 1.0966498430234868e-05,
      "loss": 0.2008,
      "step": 23240
    },
    {
      "epoch": 2.2866415873718373,
      "grad_norm": 1.1787099838256836,
      "learning_rate": 1.096252434129476e-05,
      "loss": 0.2306,
      "step": 23250
    },
    {
      "epoch": 2.287625089129847,
      "grad_norm": 1.4644756317138672,
      "learning_rate": 1.095855025235465e-05,
      "loss": 0.2355,
      "step": 23260
    },
    {
      "epoch": 2.2886085908878564,
      "grad_norm": 0.6272462606430054,
      "learning_rate": 1.0954576163414539e-05,
      "loss": 0.2345,
      "step": 23270
    },
    {
      "epoch": 2.2895920926458655,
      "grad_norm": 0.36970436573028564,
      "learning_rate": 1.0950602074474429e-05,
      "loss": 0.2368,
      "step": 23280
    },
    {
      "epoch": 2.290575594403875,
      "grad_norm": 1.6562615633010864,
      "learning_rate": 1.0946627985534317e-05,
      "loss": 0.2508,
      "step": 23290
    },
    {
      "epoch": 2.2915590961618846,
      "grad_norm": 0.40460827946662903,
      "learning_rate": 1.0942653896594206e-05,
      "loss": 0.1283,
      "step": 23300
    },
    {
      "epoch": 2.2925425979198937,
      "grad_norm": 0.5433436632156372,
      "learning_rate": 1.0938679807654096e-05,
      "loss": 0.2428,
      "step": 23310
    },
    {
      "epoch": 2.293526099677903,
      "grad_norm": 1.0284267663955688,
      "learning_rate": 1.0934705718713986e-05,
      "loss": 0.2417,
      "step": 23320
    },
    {
      "epoch": 2.2945096014359128,
      "grad_norm": 0.09268907457590103,
      "learning_rate": 1.0930731629773874e-05,
      "loss": 0.1898,
      "step": 23330
    },
    {
      "epoch": 2.295493103193922,
      "grad_norm": 2.299131155014038,
      "learning_rate": 1.0926757540833765e-05,
      "loss": 0.1991,
      "step": 23340
    },
    {
      "epoch": 2.2964766049519314,
      "grad_norm": 0.7069013118743896,
      "learning_rate": 1.0922783451893655e-05,
      "loss": 0.2412,
      "step": 23350
    },
    {
      "epoch": 2.2974601067099405,
      "grad_norm": 4.045040607452393,
      "learning_rate": 1.0918809362953544e-05,
      "loss": 0.1945,
      "step": 23360
    },
    {
      "epoch": 2.29844360846795,
      "grad_norm": 2.47917103767395,
      "learning_rate": 1.0914835274013434e-05,
      "loss": 0.2878,
      "step": 23370
    },
    {
      "epoch": 2.2994271102259596,
      "grad_norm": 1.6581519842147827,
      "learning_rate": 1.0910861185073322e-05,
      "loss": 0.1854,
      "step": 23380
    },
    {
      "epoch": 2.3004106119839687,
      "grad_norm": 4.0485687255859375,
      "learning_rate": 1.0906887096133212e-05,
      "loss": 0.1432,
      "step": 23390
    },
    {
      "epoch": 2.301394113741978,
      "grad_norm": 1.495073676109314,
      "learning_rate": 1.0902913007193101e-05,
      "loss": 0.3493,
      "step": 23400
    },
    {
      "epoch": 2.3023776154999878,
      "grad_norm": 1.586604118347168,
      "learning_rate": 1.089893891825299e-05,
      "loss": 0.1864,
      "step": 23410
    },
    {
      "epoch": 2.303361117257997,
      "grad_norm": 0.8354415893554688,
      "learning_rate": 1.0894964829312882e-05,
      "loss": 0.2969,
      "step": 23420
    },
    {
      "epoch": 2.3043446190160064,
      "grad_norm": 0.16059598326683044,
      "learning_rate": 1.0890990740372772e-05,
      "loss": 0.2807,
      "step": 23430
    },
    {
      "epoch": 2.305328120774016,
      "grad_norm": 1.2058663368225098,
      "learning_rate": 1.088701665143266e-05,
      "loss": 0.1765,
      "step": 23440
    },
    {
      "epoch": 2.306311622532025,
      "grad_norm": 0.7288966178894043,
      "learning_rate": 1.088304256249255e-05,
      "loss": 0.2272,
      "step": 23450
    },
    {
      "epoch": 2.3072951242900346,
      "grad_norm": 0.13346998393535614,
      "learning_rate": 1.087906847355244e-05,
      "loss": 0.2205,
      "step": 23460
    },
    {
      "epoch": 2.308278626048044,
      "grad_norm": 2.4363770484924316,
      "learning_rate": 1.0875094384612329e-05,
      "loss": 0.1356,
      "step": 23470
    },
    {
      "epoch": 2.309262127806053,
      "grad_norm": 0.868334949016571,
      "learning_rate": 1.0871120295672217e-05,
      "loss": 0.1763,
      "step": 23480
    },
    {
      "epoch": 2.3102456295640628,
      "grad_norm": 1.074905276298523,
      "learning_rate": 1.0867146206732106e-05,
      "loss": 0.1909,
      "step": 23490
    },
    {
      "epoch": 2.3112291313220723,
      "grad_norm": 1.5321420431137085,
      "learning_rate": 1.0863172117791998e-05,
      "loss": 0.1557,
      "step": 23500
    },
    {
      "epoch": 2.3112291313220723,
      "eval_loss": 0.12389986217021942,
      "eval_runtime": 18.2263,
      "eval_samples_per_second": 2.743,
      "eval_steps_per_second": 1.372,
      "step": 23500
    },
    {
      "epoch": 2.3122126330800814,
      "grad_norm": 0.3417229652404785,
      "learning_rate": 1.0859198028851887e-05,
      "loss": 0.0932,
      "step": 23510
    },
    {
      "epoch": 2.313196134838091,
      "grad_norm": 1.651539921760559,
      "learning_rate": 1.0855223939911777e-05,
      "loss": 0.2012,
      "step": 23520
    },
    {
      "epoch": 2.3141796365961005,
      "grad_norm": 1.1157879829406738,
      "learning_rate": 1.0851249850971665e-05,
      "loss": 0.1719,
      "step": 23530
    },
    {
      "epoch": 2.3151631383541096,
      "grad_norm": 1.6907389163970947,
      "learning_rate": 1.0847275762031555e-05,
      "loss": 0.21,
      "step": 23540
    },
    {
      "epoch": 2.316146640112119,
      "grad_norm": 0.9047884345054626,
      "learning_rate": 1.0843301673091444e-05,
      "loss": 0.2338,
      "step": 23550
    },
    {
      "epoch": 2.3171301418701287,
      "grad_norm": 0.9360957145690918,
      "learning_rate": 1.0839327584151334e-05,
      "loss": 0.1679,
      "step": 23560
    },
    {
      "epoch": 2.3181136436281378,
      "grad_norm": 0.7328473925590515,
      "learning_rate": 1.0835353495211222e-05,
      "loss": 0.2031,
      "step": 23570
    },
    {
      "epoch": 2.3190971453861473,
      "grad_norm": 1.0778604745864868,
      "learning_rate": 1.0831379406271113e-05,
      "loss": 0.1219,
      "step": 23580
    },
    {
      "epoch": 2.320080647144157,
      "grad_norm": 1.1837531328201294,
      "learning_rate": 1.0827405317331003e-05,
      "loss": 0.2164,
      "step": 23590
    },
    {
      "epoch": 2.321064148902166,
      "grad_norm": 0.9001725912094116,
      "learning_rate": 1.0823431228390893e-05,
      "loss": 0.2259,
      "step": 23600
    },
    {
      "epoch": 2.3220476506601755,
      "grad_norm": 1.0018750429153442,
      "learning_rate": 1.0819457139450782e-05,
      "loss": 0.2306,
      "step": 23610
    },
    {
      "epoch": 2.323031152418185,
      "grad_norm": 1.3852676153182983,
      "learning_rate": 1.081548305051067e-05,
      "loss": 0.2403,
      "step": 23620
    },
    {
      "epoch": 2.324014654176194,
      "grad_norm": 2.191063404083252,
      "learning_rate": 1.081150896157056e-05,
      "loss": 0.2695,
      "step": 23630
    },
    {
      "epoch": 2.3249981559342037,
      "grad_norm": 6.096485137939453,
      "learning_rate": 1.080753487263045e-05,
      "loss": 0.1783,
      "step": 23640
    },
    {
      "epoch": 2.325981657692213,
      "grad_norm": 1.0244005918502808,
      "learning_rate": 1.080356078369034e-05,
      "loss": 0.2792,
      "step": 23650
    },
    {
      "epoch": 2.3269651594502223,
      "grad_norm": 0.5512336492538452,
      "learning_rate": 1.079958669475023e-05,
      "loss": 0.2371,
      "step": 23660
    },
    {
      "epoch": 2.327948661208232,
      "grad_norm": 0.7230295538902283,
      "learning_rate": 1.079561260581012e-05,
      "loss": 0.1569,
      "step": 23670
    },
    {
      "epoch": 2.3289321629662414,
      "grad_norm": 0.17145349085330963,
      "learning_rate": 1.0791638516870008e-05,
      "loss": 0.1774,
      "step": 23680
    },
    {
      "epoch": 2.3299156647242505,
      "grad_norm": 1.245060920715332,
      "learning_rate": 1.0787664427929898e-05,
      "loss": 0.1762,
      "step": 23690
    },
    {
      "epoch": 2.33089916648226,
      "grad_norm": 0.9476994276046753,
      "learning_rate": 1.0783690338989788e-05,
      "loss": 0.2464,
      "step": 23700
    },
    {
      "epoch": 2.3318826682402696,
      "grad_norm": 0.9879484176635742,
      "learning_rate": 1.0779716250049677e-05,
      "loss": 0.2248,
      "step": 23710
    },
    {
      "epoch": 2.3328661699982787,
      "grad_norm": 0.7402428388595581,
      "learning_rate": 1.0775742161109565e-05,
      "loss": 0.2144,
      "step": 23720
    },
    {
      "epoch": 2.333849671756288,
      "grad_norm": 2.64520263671875,
      "learning_rate": 1.0771768072169455e-05,
      "loss": 0.2686,
      "step": 23730
    },
    {
      "epoch": 2.3348331735142978,
      "grad_norm": 0.9090415239334106,
      "learning_rate": 1.0767793983229346e-05,
      "loss": 0.3184,
      "step": 23740
    },
    {
      "epoch": 2.335816675272307,
      "grad_norm": 1.2032514810562134,
      "learning_rate": 1.0763819894289236e-05,
      "loss": 0.2231,
      "step": 23750
    },
    {
      "epoch": 2.3368001770303164,
      "grad_norm": 1.3694044351577759,
      "learning_rate": 1.0759845805349126e-05,
      "loss": 0.2158,
      "step": 23760
    },
    {
      "epoch": 2.337783678788326,
      "grad_norm": 0.6778462529182434,
      "learning_rate": 1.0755871716409014e-05,
      "loss": 0.23,
      "step": 23770
    },
    {
      "epoch": 2.338767180546335,
      "grad_norm": 0.8439954519271851,
      "learning_rate": 1.0751897627468903e-05,
      "loss": 0.1731,
      "step": 23780
    },
    {
      "epoch": 2.3397506823043446,
      "grad_norm": 0.3221350312232971,
      "learning_rate": 1.0747923538528793e-05,
      "loss": 0.1623,
      "step": 23790
    },
    {
      "epoch": 2.340734184062354,
      "grad_norm": 1.6610746383666992,
      "learning_rate": 1.0743949449588683e-05,
      "loss": 0.2062,
      "step": 23800
    },
    {
      "epoch": 2.341717685820363,
      "grad_norm": 0.34602779150009155,
      "learning_rate": 1.073997536064857e-05,
      "loss": 0.1302,
      "step": 23810
    },
    {
      "epoch": 2.3427011875783728,
      "grad_norm": 0.7501428723335266,
      "learning_rate": 1.0736001271708462e-05,
      "loss": 0.1755,
      "step": 23820
    },
    {
      "epoch": 2.3436846893363823,
      "grad_norm": 1.4959636926651,
      "learning_rate": 1.0732027182768352e-05,
      "loss": 0.3559,
      "step": 23830
    },
    {
      "epoch": 2.3446681910943914,
      "grad_norm": 0.8919134140014648,
      "learning_rate": 1.0728053093828241e-05,
      "loss": 0.1897,
      "step": 23840
    },
    {
      "epoch": 2.345651692852401,
      "grad_norm": 1.2190852165222168,
      "learning_rate": 1.0724079004888131e-05,
      "loss": 0.2574,
      "step": 23850
    },
    {
      "epoch": 2.3466351946104105,
      "grad_norm": 1.1519238948822021,
      "learning_rate": 1.0720104915948019e-05,
      "loss": 0.1902,
      "step": 23860
    },
    {
      "epoch": 2.3476186963684196,
      "grad_norm": 0.5527150630950928,
      "learning_rate": 1.0716130827007909e-05,
      "loss": 0.1775,
      "step": 23870
    },
    {
      "epoch": 2.348602198126429,
      "grad_norm": 0.6248764991760254,
      "learning_rate": 1.0712156738067798e-05,
      "loss": 0.1542,
      "step": 23880
    },
    {
      "epoch": 2.3495856998844387,
      "grad_norm": 0.46714580059051514,
      "learning_rate": 1.0708182649127688e-05,
      "loss": 0.2932,
      "step": 23890
    },
    {
      "epoch": 2.3505692016424478,
      "grad_norm": 0.5677179098129272,
      "learning_rate": 1.070420856018758e-05,
      "loss": 0.2823,
      "step": 23900
    },
    {
      "epoch": 2.3515527034004573,
      "grad_norm": 0.47147136926651,
      "learning_rate": 1.0700234471247469e-05,
      "loss": 0.1622,
      "step": 23910
    },
    {
      "epoch": 2.352536205158467,
      "grad_norm": 1.528113603591919,
      "learning_rate": 1.0696260382307357e-05,
      "loss": 0.2353,
      "step": 23920
    },
    {
      "epoch": 2.353519706916476,
      "grad_norm": 1.0832566022872925,
      "learning_rate": 1.0692286293367247e-05,
      "loss": 0.1539,
      "step": 23930
    },
    {
      "epoch": 2.3545032086744855,
      "grad_norm": 2.5159566402435303,
      "learning_rate": 1.0688312204427136e-05,
      "loss": 0.222,
      "step": 23940
    },
    {
      "epoch": 2.355486710432495,
      "grad_norm": 0.45711979269981384,
      "learning_rate": 1.0684338115487026e-05,
      "loss": 0.2243,
      "step": 23950
    },
    {
      "epoch": 2.356470212190504,
      "grad_norm": 0.749158501625061,
      "learning_rate": 1.0680364026546914e-05,
      "loss": 0.1643,
      "step": 23960
    },
    {
      "epoch": 2.3574537139485137,
      "grad_norm": 0.9952805042266846,
      "learning_rate": 1.0676389937606804e-05,
      "loss": 0.168,
      "step": 23970
    },
    {
      "epoch": 2.358437215706523,
      "grad_norm": 2.383165121078491,
      "learning_rate": 1.0672415848666695e-05,
      "loss": 0.196,
      "step": 23980
    },
    {
      "epoch": 2.3594207174645323,
      "grad_norm": 1.5254125595092773,
      "learning_rate": 1.0668441759726585e-05,
      "loss": 0.1929,
      "step": 23990
    },
    {
      "epoch": 2.360404219222542,
      "grad_norm": 1.1301990747451782,
      "learning_rate": 1.0664467670786474e-05,
      "loss": 0.202,
      "step": 24000
    },
    {
      "epoch": 2.360404219222542,
      "eval_loss": 0.12581513822078705,
      "eval_runtime": 20.8817,
      "eval_samples_per_second": 2.394,
      "eval_steps_per_second": 1.197,
      "step": 24000
    },
    {
      "epoch": 2.3613877209805514,
      "grad_norm": 0.46260926127433777,
      "learning_rate": 1.0660493581846362e-05,
      "loss": 0.1141,
      "step": 24010
    },
    {
      "epoch": 2.3623712227385605,
      "grad_norm": 0.6546036005020142,
      "learning_rate": 1.0656519492906252e-05,
      "loss": 0.2127,
      "step": 24020
    },
    {
      "epoch": 2.36335472449657,
      "grad_norm": 0.46774768829345703,
      "learning_rate": 1.0652545403966142e-05,
      "loss": 0.264,
      "step": 24030
    },
    {
      "epoch": 2.3643382262545796,
      "grad_norm": 0.9853132367134094,
      "learning_rate": 1.0648571315026031e-05,
      "loss": 0.1887,
      "step": 24040
    },
    {
      "epoch": 2.3653217280125887,
      "grad_norm": 0.7118104100227356,
      "learning_rate": 1.064459722608592e-05,
      "loss": 0.2641,
      "step": 24050
    },
    {
      "epoch": 2.366305229770598,
      "grad_norm": 2.5412228107452393,
      "learning_rate": 1.064062313714581e-05,
      "loss": 0.2682,
      "step": 24060
    },
    {
      "epoch": 2.3672887315286077,
      "grad_norm": 0.961894154548645,
      "learning_rate": 1.06366490482057e-05,
      "loss": 0.1522,
      "step": 24070
    },
    {
      "epoch": 2.368272233286617,
      "grad_norm": 0.3436634838581085,
      "learning_rate": 1.063267495926559e-05,
      "loss": 0.1739,
      "step": 24080
    },
    {
      "epoch": 2.3692557350446264,
      "grad_norm": 0.6866682171821594,
      "learning_rate": 1.062870087032548e-05,
      "loss": 0.191,
      "step": 24090
    },
    {
      "epoch": 2.370239236802636,
      "grad_norm": 1.4524731636047363,
      "learning_rate": 1.0624726781385368e-05,
      "loss": 0.1792,
      "step": 24100
    },
    {
      "epoch": 2.371222738560645,
      "grad_norm": 1.2320064306259155,
      "learning_rate": 1.0620752692445257e-05,
      "loss": 0.2205,
      "step": 24110
    },
    {
      "epoch": 2.3722062403186546,
      "grad_norm": 1.1350302696228027,
      "learning_rate": 1.0616778603505147e-05,
      "loss": 0.2501,
      "step": 24120
    },
    {
      "epoch": 2.373189742076664,
      "grad_norm": 3.394150495529175,
      "learning_rate": 1.0612804514565037e-05,
      "loss": 0.1806,
      "step": 24130
    },
    {
      "epoch": 2.374173243834673,
      "grad_norm": 0.8134811520576477,
      "learning_rate": 1.0608830425624925e-05,
      "loss": 0.255,
      "step": 24140
    },
    {
      "epoch": 2.3751567455926827,
      "grad_norm": 0.6805379986763,
      "learning_rate": 1.0604856336684818e-05,
      "loss": 0.1241,
      "step": 24150
    },
    {
      "epoch": 2.3761402473506923,
      "grad_norm": 0.5397747159004211,
      "learning_rate": 1.0600882247744706e-05,
      "loss": 0.2283,
      "step": 24160
    },
    {
      "epoch": 2.3771237491087014,
      "grad_norm": 0.2824234664440155,
      "learning_rate": 1.0596908158804595e-05,
      "loss": 0.2521,
      "step": 24170
    },
    {
      "epoch": 2.378107250866711,
      "grad_norm": 2.4280147552490234,
      "learning_rate": 1.0592934069864485e-05,
      "loss": 0.2096,
      "step": 24180
    },
    {
      "epoch": 2.3790907526247205,
      "grad_norm": 0.49068471789360046,
      "learning_rate": 1.0588959980924375e-05,
      "loss": 0.2259,
      "step": 24190
    },
    {
      "epoch": 2.3800742543827296,
      "grad_norm": 1.3167073726654053,
      "learning_rate": 1.0584985891984262e-05,
      "loss": 0.1058,
      "step": 24200
    },
    {
      "epoch": 2.381057756140739,
      "grad_norm": 0.7268620729446411,
      "learning_rate": 1.0581011803044152e-05,
      "loss": 0.1939,
      "step": 24210
    },
    {
      "epoch": 2.3820412578987487,
      "grad_norm": 0.5210294127464294,
      "learning_rate": 1.0577037714104042e-05,
      "loss": 0.1967,
      "step": 24220
    },
    {
      "epoch": 2.3830247596567578,
      "grad_norm": 2.116398811340332,
      "learning_rate": 1.0573063625163933e-05,
      "loss": 0.242,
      "step": 24230
    },
    {
      "epoch": 2.3840082614147673,
      "grad_norm": 0.8582640886306763,
      "learning_rate": 1.0569089536223823e-05,
      "loss": 0.1509,
      "step": 24240
    },
    {
      "epoch": 2.384991763172777,
      "grad_norm": 0.9974415898323059,
      "learning_rate": 1.056511544728371e-05,
      "loss": 0.2076,
      "step": 24250
    },
    {
      "epoch": 2.385975264930786,
      "grad_norm": 1.5299766063690186,
      "learning_rate": 1.05611413583436e-05,
      "loss": 0.2533,
      "step": 24260
    },
    {
      "epoch": 2.3869587666887955,
      "grad_norm": 0.9930517673492432,
      "learning_rate": 1.055716726940349e-05,
      "loss": 0.1992,
      "step": 24270
    },
    {
      "epoch": 2.387942268446805,
      "grad_norm": 0.5722897052764893,
      "learning_rate": 1.055319318046338e-05,
      "loss": 0.1644,
      "step": 24280
    },
    {
      "epoch": 2.388925770204814,
      "grad_norm": 1.3460361957550049,
      "learning_rate": 1.0549219091523268e-05,
      "loss": 0.2251,
      "step": 24290
    },
    {
      "epoch": 2.3899092719628237,
      "grad_norm": 0.635595977306366,
      "learning_rate": 1.0545245002583157e-05,
      "loss": 0.2137,
      "step": 24300
    },
    {
      "epoch": 2.390892773720833,
      "grad_norm": 2.7226004600524902,
      "learning_rate": 1.0541270913643049e-05,
      "loss": 0.1122,
      "step": 24310
    },
    {
      "epoch": 2.3918762754788423,
      "grad_norm": 1.5168453454971313,
      "learning_rate": 1.0537296824702938e-05,
      "loss": 0.1795,
      "step": 24320
    },
    {
      "epoch": 2.392859777236852,
      "grad_norm": 0.39879947900772095,
      "learning_rate": 1.0533322735762828e-05,
      "loss": 0.295,
      "step": 24330
    },
    {
      "epoch": 2.3938432789948614,
      "grad_norm": 0.5160859227180481,
      "learning_rate": 1.0529348646822716e-05,
      "loss": 0.1735,
      "step": 24340
    },
    {
      "epoch": 2.3948267807528705,
      "grad_norm": 1.0614607334136963,
      "learning_rate": 1.0525374557882606e-05,
      "loss": 0.2118,
      "step": 24350
    },
    {
      "epoch": 2.39581028251088,
      "grad_norm": 1.070089340209961,
      "learning_rate": 1.0521400468942495e-05,
      "loss": 0.2928,
      "step": 24360
    },
    {
      "epoch": 2.3967937842688896,
      "grad_norm": 0.6737170815467834,
      "learning_rate": 1.0517426380002385e-05,
      "loss": 0.2404,
      "step": 24370
    },
    {
      "epoch": 2.3977772860268987,
      "grad_norm": 1.382066011428833,
      "learning_rate": 1.0513452291062273e-05,
      "loss": 0.3195,
      "step": 24380
    },
    {
      "epoch": 2.398760787784908,
      "grad_norm": 1.577677607536316,
      "learning_rate": 1.0509478202122166e-05,
      "loss": 0.2037,
      "step": 24390
    },
    {
      "epoch": 2.3997442895429177,
      "grad_norm": 0.7233482003211975,
      "learning_rate": 1.0505504113182054e-05,
      "loss": 0.2168,
      "step": 24400
    },
    {
      "epoch": 2.400727791300927,
      "grad_norm": 1.9310663938522339,
      "learning_rate": 1.0501530024241944e-05,
      "loss": 0.1892,
      "step": 24410
    },
    {
      "epoch": 2.4017112930589364,
      "grad_norm": 0.4090423882007599,
      "learning_rate": 1.0497555935301833e-05,
      "loss": 0.2314,
      "step": 24420
    },
    {
      "epoch": 2.402694794816946,
      "grad_norm": 0.6759452819824219,
      "learning_rate": 1.0493581846361723e-05,
      "loss": 0.2013,
      "step": 24430
    },
    {
      "epoch": 2.403678296574955,
      "grad_norm": 1.5449095964431763,
      "learning_rate": 1.0489607757421611e-05,
      "loss": 0.2065,
      "step": 24440
    },
    {
      "epoch": 2.4046617983329646,
      "grad_norm": 0.8562779426574707,
      "learning_rate": 1.04856336684815e-05,
      "loss": 0.0849,
      "step": 24450
    },
    {
      "epoch": 2.405645300090974,
      "grad_norm": 2.506950855255127,
      "learning_rate": 1.048165957954139e-05,
      "loss": 0.2349,
      "step": 24460
    },
    {
      "epoch": 2.406628801848983,
      "grad_norm": 0.3711108863353729,
      "learning_rate": 1.0477685490601282e-05,
      "loss": 0.1949,
      "step": 24470
    },
    {
      "epoch": 2.4076123036069927,
      "grad_norm": 0.5963961482048035,
      "learning_rate": 1.0473711401661171e-05,
      "loss": 0.3249,
      "step": 24480
    },
    {
      "epoch": 2.4085958053650023,
      "grad_norm": 0.1989096999168396,
      "learning_rate": 1.046973731272106e-05,
      "loss": 0.1634,
      "step": 24490
    },
    {
      "epoch": 2.4095793071230114,
      "grad_norm": 1.052170753479004,
      "learning_rate": 1.0465763223780949e-05,
      "loss": 0.1794,
      "step": 24500
    },
    {
      "epoch": 2.4095793071230114,
      "eval_loss": 0.12587225437164307,
      "eval_runtime": 16.8866,
      "eval_samples_per_second": 2.961,
      "eval_steps_per_second": 1.48,
      "step": 24500
    },
    {
      "epoch": 2.410562808881021,
      "grad_norm": 0.8879571557044983,
      "learning_rate": 1.0461789134840839e-05,
      "loss": 0.2561,
      "step": 24510
    },
    {
      "epoch": 2.4115463106390305,
      "grad_norm": 0.6819064617156982,
      "learning_rate": 1.0457815045900728e-05,
      "loss": 0.2363,
      "step": 24520
    },
    {
      "epoch": 2.4125298123970396,
      "grad_norm": 0.4369518458843231,
      "learning_rate": 1.0453840956960616e-05,
      "loss": 0.2366,
      "step": 24530
    },
    {
      "epoch": 2.413513314155049,
      "grad_norm": 0.3565225303173065,
      "learning_rate": 1.0449866868020506e-05,
      "loss": 0.2178,
      "step": 24540
    },
    {
      "epoch": 2.4144968159130586,
      "grad_norm": 0.8098437190055847,
      "learning_rate": 1.0445892779080397e-05,
      "loss": 0.278,
      "step": 24550
    },
    {
      "epoch": 2.4154803176710677,
      "grad_norm": 0.873010516166687,
      "learning_rate": 1.0441918690140287e-05,
      "loss": 0.314,
      "step": 24560
    },
    {
      "epoch": 2.4164638194290773,
      "grad_norm": 1.4115465879440308,
      "learning_rate": 1.0437944601200177e-05,
      "loss": 0.2881,
      "step": 24570
    },
    {
      "epoch": 2.417447321187087,
      "grad_norm": 1.7016613483428955,
      "learning_rate": 1.0433970512260065e-05,
      "loss": 0.3005,
      "step": 24580
    },
    {
      "epoch": 2.418430822945096,
      "grad_norm": 0.9918956756591797,
      "learning_rate": 1.0429996423319954e-05,
      "loss": 0.2868,
      "step": 24590
    },
    {
      "epoch": 2.4194143247031055,
      "grad_norm": 0.4580109417438507,
      "learning_rate": 1.0426022334379844e-05,
      "loss": 0.192,
      "step": 24600
    },
    {
      "epoch": 2.420397826461115,
      "grad_norm": 0.97365802526474,
      "learning_rate": 1.0422048245439734e-05,
      "loss": 0.2039,
      "step": 24610
    },
    {
      "epoch": 2.421381328219124,
      "grad_norm": 1.3537076711654663,
      "learning_rate": 1.0418074156499622e-05,
      "loss": 0.1981,
      "step": 24620
    },
    {
      "epoch": 2.4223648299771336,
      "grad_norm": 0.6715251803398132,
      "learning_rate": 1.0414100067559515e-05,
      "loss": 0.2591,
      "step": 24630
    },
    {
      "epoch": 2.423348331735143,
      "grad_norm": 1.817028284072876,
      "learning_rate": 1.0410125978619403e-05,
      "loss": 0.181,
      "step": 24640
    },
    {
      "epoch": 2.4243318334931523,
      "grad_norm": 1.0385745763778687,
      "learning_rate": 1.0406151889679292e-05,
      "loss": 0.1753,
      "step": 24650
    },
    {
      "epoch": 2.425315335251162,
      "grad_norm": 0.4937601089477539,
      "learning_rate": 1.0402177800739182e-05,
      "loss": 0.1725,
      "step": 24660
    },
    {
      "epoch": 2.4262988370091714,
      "grad_norm": 1.1675567626953125,
      "learning_rate": 1.0398203711799072e-05,
      "loss": 0.2863,
      "step": 24670
    },
    {
      "epoch": 2.4272823387671805,
      "grad_norm": 0.7065426111221313,
      "learning_rate": 1.039422962285896e-05,
      "loss": 0.2022,
      "step": 24680
    },
    {
      "epoch": 2.42826584052519,
      "grad_norm": 0.770263671875,
      "learning_rate": 1.039025553391885e-05,
      "loss": 0.2201,
      "step": 24690
    },
    {
      "epoch": 2.4292493422831996,
      "grad_norm": 0.46510329842567444,
      "learning_rate": 1.0386281444978739e-05,
      "loss": 0.2175,
      "step": 24700
    },
    {
      "epoch": 2.4302328440412087,
      "grad_norm": 0.845753014087677,
      "learning_rate": 1.038230735603863e-05,
      "loss": 0.2049,
      "step": 24710
    },
    {
      "epoch": 2.431216345799218,
      "grad_norm": 1.2518194913864136,
      "learning_rate": 1.037833326709852e-05,
      "loss": 0.2833,
      "step": 24720
    },
    {
      "epoch": 2.4321998475572277,
      "grad_norm": 1.889909029006958,
      "learning_rate": 1.0374359178158408e-05,
      "loss": 0.1633,
      "step": 24730
    },
    {
      "epoch": 2.433183349315237,
      "grad_norm": 0.6773773431777954,
      "learning_rate": 1.0370385089218298e-05,
      "loss": 0.2636,
      "step": 24740
    },
    {
      "epoch": 2.4341668510732464,
      "grad_norm": 1.3301688432693481,
      "learning_rate": 1.0366411000278187e-05,
      "loss": 0.1885,
      "step": 24750
    },
    {
      "epoch": 2.435150352831256,
      "grad_norm": 0.4370648264884949,
      "learning_rate": 1.0362436911338077e-05,
      "loss": 0.2366,
      "step": 24760
    },
    {
      "epoch": 2.436133854589265,
      "grad_norm": 1.1172949075698853,
      "learning_rate": 1.0358462822397965e-05,
      "loss": 0.2925,
      "step": 24770
    },
    {
      "epoch": 2.4371173563472746,
      "grad_norm": 0.37216636538505554,
      "learning_rate": 1.0354488733457855e-05,
      "loss": 0.2844,
      "step": 24780
    },
    {
      "epoch": 2.438100858105284,
      "grad_norm": 0.6310292482376099,
      "learning_rate": 1.0350514644517746e-05,
      "loss": 0.2222,
      "step": 24790
    },
    {
      "epoch": 2.439084359863293,
      "grad_norm": 1.4703803062438965,
      "learning_rate": 1.0346540555577636e-05,
      "loss": 0.1935,
      "step": 24800
    },
    {
      "epoch": 2.4400678616213027,
      "grad_norm": 0.8147374987602234,
      "learning_rate": 1.0342566466637525e-05,
      "loss": 0.1447,
      "step": 24810
    },
    {
      "epoch": 2.4410513633793123,
      "grad_norm": 0.6662246584892273,
      "learning_rate": 1.0338592377697413e-05,
      "loss": 0.1878,
      "step": 24820
    },
    {
      "epoch": 2.4420348651373214,
      "grad_norm": 0.7072861194610596,
      "learning_rate": 1.0334618288757303e-05,
      "loss": 0.3015,
      "step": 24830
    },
    {
      "epoch": 2.443018366895331,
      "grad_norm": 1.876746416091919,
      "learning_rate": 1.0330644199817193e-05,
      "loss": 0.3142,
      "step": 24840
    },
    {
      "epoch": 2.4440018686533405,
      "grad_norm": 0.9753235578536987,
      "learning_rate": 1.0326670110877082e-05,
      "loss": 0.2081,
      "step": 24850
    },
    {
      "epoch": 2.4449853704113496,
      "grad_norm": 1.0954594612121582,
      "learning_rate": 1.032269602193697e-05,
      "loss": 0.3415,
      "step": 24860
    },
    {
      "epoch": 2.445968872169359,
      "grad_norm": 1.0371835231781006,
      "learning_rate": 1.0318721932996863e-05,
      "loss": 0.23,
      "step": 24870
    },
    {
      "epoch": 2.4469523739273686,
      "grad_norm": 1.1946139335632324,
      "learning_rate": 1.0314747844056751e-05,
      "loss": 0.1863,
      "step": 24880
    },
    {
      "epoch": 2.4479358756853777,
      "grad_norm": 0.19204482436180115,
      "learning_rate": 1.0310773755116641e-05,
      "loss": 0.2889,
      "step": 24890
    },
    {
      "epoch": 2.4489193774433873,
      "grad_norm": 1.4059542417526245,
      "learning_rate": 1.030679966617653e-05,
      "loss": 0.2667,
      "step": 24900
    },
    {
      "epoch": 2.449902879201397,
      "grad_norm": 0.19449950754642487,
      "learning_rate": 1.030282557723642e-05,
      "loss": 0.2618,
      "step": 24910
    },
    {
      "epoch": 2.450886380959406,
      "grad_norm": 0.6006851196289062,
      "learning_rate": 1.0298851488296308e-05,
      "loss": 0.2489,
      "step": 24920
    },
    {
      "epoch": 2.4518698827174155,
      "grad_norm": 1.953939437866211,
      "learning_rate": 1.0294877399356198e-05,
      "loss": 0.226,
      "step": 24930
    },
    {
      "epoch": 2.4528533844754246,
      "grad_norm": 1.157392144203186,
      "learning_rate": 1.0290903310416088e-05,
      "loss": 0.1784,
      "step": 24940
    },
    {
      "epoch": 2.453836886233434,
      "grad_norm": 0.37648680806159973,
      "learning_rate": 1.0286929221475977e-05,
      "loss": 0.2387,
      "step": 24950
    },
    {
      "epoch": 2.4548203879914436,
      "grad_norm": 1.1377531290054321,
      "learning_rate": 1.0282955132535869e-05,
      "loss": 0.1905,
      "step": 24960
    },
    {
      "epoch": 2.4558038897494527,
      "grad_norm": 1.6910361051559448,
      "learning_rate": 1.0278981043595757e-05,
      "loss": 0.1878,
      "step": 24970
    },
    {
      "epoch": 2.4567873915074623,
      "grad_norm": 2.3196709156036377,
      "learning_rate": 1.0275006954655646e-05,
      "loss": 0.2849,
      "step": 24980
    },
    {
      "epoch": 2.457770893265472,
      "grad_norm": 1.3127903938293457,
      "learning_rate": 1.0271032865715536e-05,
      "loss": 0.2607,
      "step": 24990
    },
    {
      "epoch": 2.458754395023481,
      "grad_norm": 1.0793159008026123,
      "learning_rate": 1.0267058776775426e-05,
      "loss": 0.2561,
      "step": 25000
    },
    {
      "epoch": 2.458754395023481,
      "eval_loss": 0.12595096230506897,
      "eval_runtime": 18.2255,
      "eval_samples_per_second": 2.743,
      "eval_steps_per_second": 1.372,
      "step": 25000
    },
    {
      "epoch": 2.4597378967814905,
      "grad_norm": 1.686716914176941,
      "learning_rate": 1.0263084687835313e-05,
      "loss": 0.175,
      "step": 25010
    },
    {
      "epoch": 2.4607213985395,
      "grad_norm": 1.6246962547302246,
      "learning_rate": 1.0259110598895203e-05,
      "loss": 0.2054,
      "step": 25020
    },
    {
      "epoch": 2.461704900297509,
      "grad_norm": 0.8834644556045532,
      "learning_rate": 1.0255136509955093e-05,
      "loss": 0.264,
      "step": 25030
    },
    {
      "epoch": 2.4626884020555186,
      "grad_norm": 2.999387264251709,
      "learning_rate": 1.0251162421014984e-05,
      "loss": 0.1614,
      "step": 25040
    },
    {
      "epoch": 2.463671903813528,
      "grad_norm": 0.9904057383537292,
      "learning_rate": 1.0247188332074874e-05,
      "loss": 0.177,
      "step": 25050
    },
    {
      "epoch": 2.4646554055715373,
      "grad_norm": 0.770376980304718,
      "learning_rate": 1.0243214243134762e-05,
      "loss": 0.2024,
      "step": 25060
    },
    {
      "epoch": 2.465638907329547,
      "grad_norm": 2.4670302867889404,
      "learning_rate": 1.0239240154194651e-05,
      "loss": 0.0944,
      "step": 25070
    },
    {
      "epoch": 2.4666224090875564,
      "grad_norm": 0.5328916907310486,
      "learning_rate": 1.0235266065254541e-05,
      "loss": 0.2334,
      "step": 25080
    },
    {
      "epoch": 2.4676059108455655,
      "grad_norm": 0.724310576915741,
      "learning_rate": 1.023129197631443e-05,
      "loss": 0.1495,
      "step": 25090
    },
    {
      "epoch": 2.468589412603575,
      "grad_norm": 0.9128016233444214,
      "learning_rate": 1.0227317887374319e-05,
      "loss": 0.159,
      "step": 25100
    },
    {
      "epoch": 2.4695729143615845,
      "grad_norm": 0.7053813934326172,
      "learning_rate": 1.0223343798434208e-05,
      "loss": 0.3856,
      "step": 25110
    },
    {
      "epoch": 2.4705564161195936,
      "grad_norm": 1.1937212944030762,
      "learning_rate": 1.02193697094941e-05,
      "loss": 0.17,
      "step": 25120
    },
    {
      "epoch": 2.471539917877603,
      "grad_norm": 1.3782202005386353,
      "learning_rate": 1.021539562055399e-05,
      "loss": 0.1572,
      "step": 25130
    },
    {
      "epoch": 2.4725234196356127,
      "grad_norm": 0.8486477136611938,
      "learning_rate": 1.0211421531613879e-05,
      "loss": 0.1794,
      "step": 25140
    },
    {
      "epoch": 2.473506921393622,
      "grad_norm": 0.7491905093193054,
      "learning_rate": 1.0207447442673769e-05,
      "loss": 0.1787,
      "step": 25150
    },
    {
      "epoch": 2.4744904231516314,
      "grad_norm": 0.6677191853523254,
      "learning_rate": 1.0203473353733657e-05,
      "loss": 0.1964,
      "step": 25160
    },
    {
      "epoch": 2.475473924909641,
      "grad_norm": 1.6847872734069824,
      "learning_rate": 1.0199499264793546e-05,
      "loss": 0.1782,
      "step": 25170
    },
    {
      "epoch": 2.47645742666765,
      "grad_norm": 1.3266242742538452,
      "learning_rate": 1.0195525175853436e-05,
      "loss": 0.1698,
      "step": 25180
    },
    {
      "epoch": 2.4774409284256595,
      "grad_norm": 0.22463712096214294,
      "learning_rate": 1.0191551086913326e-05,
      "loss": 0.1967,
      "step": 25190
    },
    {
      "epoch": 2.478424430183669,
      "grad_norm": 1.7434459924697876,
      "learning_rate": 1.0187576997973217e-05,
      "loss": 0.3137,
      "step": 25200
    },
    {
      "epoch": 2.479407931941678,
      "grad_norm": 2.9534683227539062,
      "learning_rate": 1.0183602909033105e-05,
      "loss": 0.1166,
      "step": 25210
    },
    {
      "epoch": 2.4803914336996877,
      "grad_norm": 0.9041188359260559,
      "learning_rate": 1.0179628820092995e-05,
      "loss": 0.2327,
      "step": 25220
    },
    {
      "epoch": 2.4813749354576973,
      "grad_norm": 2.4157469272613525,
      "learning_rate": 1.0175654731152884e-05,
      "loss": 0.2136,
      "step": 25230
    },
    {
      "epoch": 2.4823584372157064,
      "grad_norm": 3.838491201400757,
      "learning_rate": 1.0171680642212774e-05,
      "loss": 0.3159,
      "step": 25240
    },
    {
      "epoch": 2.483341938973716,
      "grad_norm": 2.9579169750213623,
      "learning_rate": 1.0167706553272662e-05,
      "loss": 0.2254,
      "step": 25250
    },
    {
      "epoch": 2.4843254407317255,
      "grad_norm": 2.0895047187805176,
      "learning_rate": 1.0163732464332552e-05,
      "loss": 0.2235,
      "step": 25260
    },
    {
      "epoch": 2.4853089424897346,
      "grad_norm": 0.3538540303707123,
      "learning_rate": 1.0159758375392441e-05,
      "loss": 0.254,
      "step": 25270
    },
    {
      "epoch": 2.486292444247744,
      "grad_norm": 0.8005692958831787,
      "learning_rate": 1.0155784286452333e-05,
      "loss": 0.2868,
      "step": 25280
    },
    {
      "epoch": 2.4872759460057536,
      "grad_norm": 0.44420817494392395,
      "learning_rate": 1.0151810197512222e-05,
      "loss": 0.1653,
      "step": 25290
    },
    {
      "epoch": 2.4882594477637627,
      "grad_norm": 0.8710966110229492,
      "learning_rate": 1.014783610857211e-05,
      "loss": 0.1818,
      "step": 25300
    },
    {
      "epoch": 2.4892429495217723,
      "grad_norm": 2.267538070678711,
      "learning_rate": 1.0143862019632e-05,
      "loss": 0.149,
      "step": 25310
    },
    {
      "epoch": 2.490226451279782,
      "grad_norm": 0.5567754507064819,
      "learning_rate": 1.013988793069189e-05,
      "loss": 0.1821,
      "step": 25320
    },
    {
      "epoch": 2.491209953037791,
      "grad_norm": 0.5361708998680115,
      "learning_rate": 1.013591384175178e-05,
      "loss": 0.1408,
      "step": 25330
    },
    {
      "epoch": 2.4921934547958005,
      "grad_norm": 1.3161852359771729,
      "learning_rate": 1.0131939752811667e-05,
      "loss": 0.1775,
      "step": 25340
    },
    {
      "epoch": 2.49317695655381,
      "grad_norm": 1.63074791431427,
      "learning_rate": 1.0127965663871557e-05,
      "loss": 0.1754,
      "step": 25350
    },
    {
      "epoch": 2.494160458311819,
      "grad_norm": 0.591672956943512,
      "learning_rate": 1.0123991574931448e-05,
      "loss": 0.2224,
      "step": 25360
    },
    {
      "epoch": 2.4951439600698286,
      "grad_norm": 0.8783940672874451,
      "learning_rate": 1.0120017485991338e-05,
      "loss": 0.1685,
      "step": 25370
    },
    {
      "epoch": 2.496127461827838,
      "grad_norm": 1.294005036354065,
      "learning_rate": 1.0116043397051228e-05,
      "loss": 0.2783,
      "step": 25380
    },
    {
      "epoch": 2.4971109635858473,
      "grad_norm": 1.8267871141433716,
      "learning_rate": 1.0112069308111117e-05,
      "loss": 0.2339,
      "step": 25390
    },
    {
      "epoch": 2.498094465343857,
      "grad_norm": 1.058579683303833,
      "learning_rate": 1.0108095219171005e-05,
      "loss": 0.1437,
      "step": 25400
    },
    {
      "epoch": 2.4990779671018664,
      "grad_norm": 0.35047388076782227,
      "learning_rate": 1.0104121130230895e-05,
      "loss": 0.1035,
      "step": 25410
    },
    {
      "epoch": 2.5000614688598755,
      "grad_norm": 0.5837266445159912,
      "learning_rate": 1.0100147041290785e-05,
      "loss": 0.1586,
      "step": 25420
    },
    {
      "epoch": 2.501044970617885,
      "grad_norm": 0.7025007009506226,
      "learning_rate": 1.0096172952350674e-05,
      "loss": 0.2133,
      "step": 25430
    },
    {
      "epoch": 2.502028472375894,
      "grad_norm": 1.1473788022994995,
      "learning_rate": 1.0092198863410566e-05,
      "loss": 0.1158,
      "step": 25440
    },
    {
      "epoch": 2.5030119741339036,
      "grad_norm": 1.7159006595611572,
      "learning_rate": 1.0088224774470454e-05,
      "loss": 0.2806,
      "step": 25450
    },
    {
      "epoch": 2.503995475891913,
      "grad_norm": 2.7162020206451416,
      "learning_rate": 1.0084250685530343e-05,
      "loss": 0.2626,
      "step": 25460
    },
    {
      "epoch": 2.5049789776499223,
      "grad_norm": 0.5796257257461548,
      "learning_rate": 1.0080276596590233e-05,
      "loss": 0.2202,
      "step": 25470
    },
    {
      "epoch": 2.505962479407932,
      "grad_norm": 0.4555465579032898,
      "learning_rate": 1.0076302507650123e-05,
      "loss": 0.1799,
      "step": 25480
    },
    {
      "epoch": 2.5069459811659414,
      "grad_norm": 0.17807602882385254,
      "learning_rate": 1.007232841871001e-05,
      "loss": 0.2468,
      "step": 25490
    },
    {
      "epoch": 2.5079294829239505,
      "grad_norm": 1.4680837392807007,
      "learning_rate": 1.00683543297699e-05,
      "loss": 0.2256,
      "step": 25500
    },
    {
      "epoch": 2.5079294829239505,
      "eval_loss": 0.12521879374980927,
      "eval_runtime": 18.9765,
      "eval_samples_per_second": 2.635,
      "eval_steps_per_second": 1.317,
      "step": 25500
    },
    {
      "epoch": 2.50891298468196,
      "grad_norm": 0.8120176792144775,
      "learning_rate": 1.006438024082979e-05,
      "loss": 0.2246,
      "step": 25510
    },
    {
      "epoch": 2.5098964864399695,
      "grad_norm": 1.7058813571929932,
      "learning_rate": 1.0060406151889681e-05,
      "loss": 0.1898,
      "step": 25520
    },
    {
      "epoch": 2.5108799881979786,
      "grad_norm": 2.02591609954834,
      "learning_rate": 1.0056432062949571e-05,
      "loss": 0.1844,
      "step": 25530
    },
    {
      "epoch": 2.511863489955988,
      "grad_norm": 0.452986478805542,
      "learning_rate": 1.0052457974009459e-05,
      "loss": 0.1278,
      "step": 25540
    },
    {
      "epoch": 2.5128469917139977,
      "grad_norm": 0.6622622609138489,
      "learning_rate": 1.0048483885069349e-05,
      "loss": 0.2361,
      "step": 25550
    },
    {
      "epoch": 2.513830493472007,
      "grad_norm": 2.510401487350464,
      "learning_rate": 1.0044509796129238e-05,
      "loss": 0.2983,
      "step": 25560
    },
    {
      "epoch": 2.5148139952300164,
      "grad_norm": 1.1683661937713623,
      "learning_rate": 1.0040535707189128e-05,
      "loss": 0.2436,
      "step": 25570
    },
    {
      "epoch": 2.515797496988026,
      "grad_norm": 1.2521823644638062,
      "learning_rate": 1.0036561618249016e-05,
      "loss": 0.145,
      "step": 25580
    },
    {
      "epoch": 2.516780998746035,
      "grad_norm": 1.279665470123291,
      "learning_rate": 1.0032587529308906e-05,
      "loss": 0.1753,
      "step": 25590
    },
    {
      "epoch": 2.5177645005040445,
      "grad_norm": 0.5141640901565552,
      "learning_rate": 1.0028613440368797e-05,
      "loss": 0.2364,
      "step": 25600
    },
    {
      "epoch": 2.518748002262054,
      "grad_norm": 0.45487838983535767,
      "learning_rate": 1.0024639351428687e-05,
      "loss": 0.1719,
      "step": 25610
    },
    {
      "epoch": 2.519731504020063,
      "grad_norm": 0.6889397501945496,
      "learning_rate": 1.0020665262488576e-05,
      "loss": 0.2684,
      "step": 25620
    },
    {
      "epoch": 2.5207150057780727,
      "grad_norm": 1.1064757108688354,
      "learning_rate": 1.0016691173548466e-05,
      "loss": 0.2345,
      "step": 25630
    },
    {
      "epoch": 2.5216985075360823,
      "grad_norm": 1.3356391191482544,
      "learning_rate": 1.0012717084608354e-05,
      "loss": 0.1524,
      "step": 25640
    },
    {
      "epoch": 2.5226820092940914,
      "grad_norm": 1.0441110134124756,
      "learning_rate": 1.0008742995668244e-05,
      "loss": 0.1907,
      "step": 25650
    },
    {
      "epoch": 2.523665511052101,
      "grad_norm": 0.581211268901825,
      "learning_rate": 1.0004768906728133e-05,
      "loss": 0.1763,
      "step": 25660
    },
    {
      "epoch": 2.5246490128101104,
      "grad_norm": 2.296384811401367,
      "learning_rate": 1.0000794817788023e-05,
      "loss": 0.291,
      "step": 25670
    },
    {
      "epoch": 2.5256325145681195,
      "grad_norm": 0.5643019080162048,
      "learning_rate": 9.996820728847913e-06,
      "loss": 0.2914,
      "step": 25680
    },
    {
      "epoch": 2.526616016326129,
      "grad_norm": 1.7198952436447144,
      "learning_rate": 9.992846639907802e-06,
      "loss": 0.1675,
      "step": 25690
    },
    {
      "epoch": 2.5275995180841386,
      "grad_norm": 0.4589093029499054,
      "learning_rate": 9.98887255096769e-06,
      "loss": 0.2152,
      "step": 25700
    },
    {
      "epoch": 2.5285830198421477,
      "grad_norm": 1.0957061052322388,
      "learning_rate": 9.984898462027582e-06,
      "loss": 0.1656,
      "step": 25710
    },
    {
      "epoch": 2.5295665216001573,
      "grad_norm": 1.6807781457901,
      "learning_rate": 9.980924373087471e-06,
      "loss": 0.2362,
      "step": 25720
    },
    {
      "epoch": 2.530550023358167,
      "grad_norm": 0.4425702691078186,
      "learning_rate": 9.97695028414736e-06,
      "loss": 0.2482,
      "step": 25730
    },
    {
      "epoch": 2.531533525116176,
      "grad_norm": 0.26340290904045105,
      "learning_rate": 9.972976195207249e-06,
      "loss": 0.1555,
      "step": 25740
    },
    {
      "epoch": 2.5325170268741855,
      "grad_norm": 0.6684683561325073,
      "learning_rate": 9.96900210626714e-06,
      "loss": 0.2472,
      "step": 25750
    },
    {
      "epoch": 2.533500528632195,
      "grad_norm": 2.2530484199523926,
      "learning_rate": 9.965028017327028e-06,
      "loss": 0.1788,
      "step": 25760
    },
    {
      "epoch": 2.534484030390204,
      "grad_norm": 1.1162317991256714,
      "learning_rate": 9.961053928386918e-06,
      "loss": 0.2171,
      "step": 25770
    },
    {
      "epoch": 2.5354675321482136,
      "grad_norm": 0.19642123579978943,
      "learning_rate": 9.957079839446808e-06,
      "loss": 0.1548,
      "step": 25780
    },
    {
      "epoch": 2.536451033906223,
      "grad_norm": 1.437812328338623,
      "learning_rate": 9.953105750506697e-06,
      "loss": 0.3077,
      "step": 25790
    },
    {
      "epoch": 2.5374345356642323,
      "grad_norm": 0.3670991361141205,
      "learning_rate": 9.949131661566587e-06,
      "loss": 0.228,
      "step": 25800
    },
    {
      "epoch": 2.538418037422242,
      "grad_norm": 0.945225715637207,
      "learning_rate": 9.945157572626476e-06,
      "loss": 0.2034,
      "step": 25810
    },
    {
      "epoch": 2.5394015391802514,
      "grad_norm": 1.6273648738861084,
      "learning_rate": 9.941183483686364e-06,
      "loss": 0.1443,
      "step": 25820
    },
    {
      "epoch": 2.5403850409382605,
      "grad_norm": 0.4249015152454376,
      "learning_rate": 9.937209394746256e-06,
      "loss": 0.0696,
      "step": 25830
    },
    {
      "epoch": 2.54136854269627,
      "grad_norm": 1.2602684497833252,
      "learning_rate": 9.933235305806145e-06,
      "loss": 0.2626,
      "step": 25840
    },
    {
      "epoch": 2.5423520444542795,
      "grad_norm": 0.5752518773078918,
      "learning_rate": 9.929261216866033e-06,
      "loss": 0.2839,
      "step": 25850
    },
    {
      "epoch": 2.5433355462122886,
      "grad_norm": 1.0904006958007812,
      "learning_rate": 9.925287127925923e-06,
      "loss": 0.1481,
      "step": 25860
    },
    {
      "epoch": 2.544319047970298,
      "grad_norm": 0.8281343579292297,
      "learning_rate": 9.921313038985814e-06,
      "loss": 0.2193,
      "step": 25870
    },
    {
      "epoch": 2.5453025497283077,
      "grad_norm": 0.5674927830696106,
      "learning_rate": 9.917338950045702e-06,
      "loss": 0.224,
      "step": 25880
    },
    {
      "epoch": 2.546286051486317,
      "grad_norm": 0.3176279664039612,
      "learning_rate": 9.913364861105592e-06,
      "loss": 0.2513,
      "step": 25890
    },
    {
      "epoch": 2.5472695532443264,
      "grad_norm": 1.221111536026001,
      "learning_rate": 9.909390772165482e-06,
      "loss": 0.1335,
      "step": 25900
    },
    {
      "epoch": 2.548253055002336,
      "grad_norm": 0.720691978931427,
      "learning_rate": 9.905416683225371e-06,
      "loss": 0.1934,
      "step": 25910
    },
    {
      "epoch": 2.549236556760345,
      "grad_norm": 2.970773696899414,
      "learning_rate": 9.901442594285261e-06,
      "loss": 0.2014,
      "step": 25920
    },
    {
      "epoch": 2.5502200585183545,
      "grad_norm": 3.3352341651916504,
      "learning_rate": 9.89746850534515e-06,
      "loss": 0.1885,
      "step": 25930
    },
    {
      "epoch": 2.551203560276364,
      "grad_norm": 1.7905553579330444,
      "learning_rate": 9.893494416405039e-06,
      "loss": 0.3061,
      "step": 25940
    },
    {
      "epoch": 2.552187062034373,
      "grad_norm": 1.478422999382019,
      "learning_rate": 9.88952032746493e-06,
      "loss": 0.1855,
      "step": 25950
    },
    {
      "epoch": 2.5531705637923827,
      "grad_norm": 0.7994058132171631,
      "learning_rate": 9.88554623852482e-06,
      "loss": 0.1579,
      "step": 25960
    },
    {
      "epoch": 2.5541540655503923,
      "grad_norm": 2.159670114517212,
      "learning_rate": 9.881572149584708e-06,
      "loss": 0.1243,
      "step": 25970
    },
    {
      "epoch": 2.5551375673084014,
      "grad_norm": 1.8579719066619873,
      "learning_rate": 9.877598060644597e-06,
      "loss": 0.3046,
      "step": 25980
    },
    {
      "epoch": 2.556121069066411,
      "grad_norm": 2.223876714706421,
      "learning_rate": 9.873623971704489e-06,
      "loss": 0.1892,
      "step": 25990
    },
    {
      "epoch": 2.5571045708244204,
      "grad_norm": 0.31044161319732666,
      "learning_rate": 9.869649882764377e-06,
      "loss": 0.166,
      "step": 26000
    },
    {
      "epoch": 2.5571045708244204,
      "eval_loss": 0.12297271937131882,
      "eval_runtime": 18.1156,
      "eval_samples_per_second": 2.76,
      "eval_steps_per_second": 1.38,
      "step": 26000
    },
    {
      "epoch": 2.5580880725824295,
      "grad_norm": 1.37006676197052,
      "learning_rate": 9.865675793824266e-06,
      "loss": 0.319,
      "step": 26010
    },
    {
      "epoch": 2.559071574340439,
      "grad_norm": 1.233155369758606,
      "learning_rate": 9.861701704884156e-06,
      "loss": 0.2236,
      "step": 26020
    },
    {
      "epoch": 2.5600550760984486,
      "grad_norm": 1.331598162651062,
      "learning_rate": 9.857727615944046e-06,
      "loss": 0.2089,
      "step": 26030
    },
    {
      "epoch": 2.5610385778564577,
      "grad_norm": 0.8588127493858337,
      "learning_rate": 9.853753527003935e-06,
      "loss": 0.2139,
      "step": 26040
    },
    {
      "epoch": 2.5620220796144673,
      "grad_norm": 0.8715842366218567,
      "learning_rate": 9.849779438063825e-06,
      "loss": 0.1915,
      "step": 26050
    },
    {
      "epoch": 2.563005581372477,
      "grad_norm": 2.4266014099121094,
      "learning_rate": 9.845805349123713e-06,
      "loss": 0.2881,
      "step": 26060
    },
    {
      "epoch": 2.563989083130486,
      "grad_norm": 0.435737282037735,
      "learning_rate": 9.841831260183604e-06,
      "loss": 0.2017,
      "step": 26070
    },
    {
      "epoch": 2.5649725848884954,
      "grad_norm": 0.880113959312439,
      "learning_rate": 9.837857171243494e-06,
      "loss": 0.2318,
      "step": 26080
    },
    {
      "epoch": 2.565956086646505,
      "grad_norm": 0.9128925204277039,
      "learning_rate": 9.833883082303382e-06,
      "loss": 0.1924,
      "step": 26090
    },
    {
      "epoch": 2.566939588404514,
      "grad_norm": 0.4470633268356323,
      "learning_rate": 9.829908993363272e-06,
      "loss": 0.1752,
      "step": 26100
    },
    {
      "epoch": 2.5679230901625236,
      "grad_norm": 0.5922941565513611,
      "learning_rate": 9.825934904423163e-06,
      "loss": 0.1703,
      "step": 26110
    },
    {
      "epoch": 2.568906591920533,
      "grad_norm": 0.4576959013938904,
      "learning_rate": 9.821960815483051e-06,
      "loss": 0.2354,
      "step": 26120
    },
    {
      "epoch": 2.5698900936785423,
      "grad_norm": 1.6791609525680542,
      "learning_rate": 9.81798672654294e-06,
      "loss": 0.1089,
      "step": 26130
    },
    {
      "epoch": 2.570873595436552,
      "grad_norm": 1.9039769172668457,
      "learning_rate": 9.81401263760283e-06,
      "loss": 0.2101,
      "step": 26140
    },
    {
      "epoch": 2.5718570971945613,
      "grad_norm": 0.5855951905250549,
      "learning_rate": 9.81003854866272e-06,
      "loss": 0.2774,
      "step": 26150
    },
    {
      "epoch": 2.5728405989525704,
      "grad_norm": 0.6082465052604675,
      "learning_rate": 9.80606445972261e-06,
      "loss": 0.1697,
      "step": 26160
    },
    {
      "epoch": 2.57382410071058,
      "grad_norm": 1.0260789394378662,
      "learning_rate": 9.8020903707825e-06,
      "loss": 0.2786,
      "step": 26170
    },
    {
      "epoch": 2.5748076024685895,
      "grad_norm": 4.773690700531006,
      "learning_rate": 9.798116281842387e-06,
      "loss": 0.2392,
      "step": 26180
    },
    {
      "epoch": 2.5757911042265986,
      "grad_norm": 1.1605236530303955,
      "learning_rate": 9.794142192902279e-06,
      "loss": 0.2532,
      "step": 26190
    },
    {
      "epoch": 2.576774605984608,
      "grad_norm": 0.7416682243347168,
      "learning_rate": 9.790168103962168e-06,
      "loss": 0.259,
      "step": 26200
    },
    {
      "epoch": 2.5777581077426177,
      "grad_norm": 0.3520563542842865,
      "learning_rate": 9.786194015022056e-06,
      "loss": 0.2625,
      "step": 26210
    },
    {
      "epoch": 2.578741609500627,
      "grad_norm": 0.9534961581230164,
      "learning_rate": 9.782219926081946e-06,
      "loss": 0.3266,
      "step": 26220
    },
    {
      "epoch": 2.5797251112586363,
      "grad_norm": 1.174358606338501,
      "learning_rate": 9.778245837141837e-06,
      "loss": 0.2664,
      "step": 26230
    },
    {
      "epoch": 2.580708613016646,
      "grad_norm": 1.092491626739502,
      "learning_rate": 9.774271748201725e-06,
      "loss": 0.1844,
      "step": 26240
    },
    {
      "epoch": 2.581692114774655,
      "grad_norm": 0.3747186064720154,
      "learning_rate": 9.770297659261615e-06,
      "loss": 0.2445,
      "step": 26250
    },
    {
      "epoch": 2.5826756165326645,
      "grad_norm": 0.16160008311271667,
      "learning_rate": 9.766323570321505e-06,
      "loss": 0.1839,
      "step": 26260
    },
    {
      "epoch": 2.583659118290674,
      "grad_norm": 0.3357384204864502,
      "learning_rate": 9.762349481381394e-06,
      "loss": 0.2154,
      "step": 26270
    },
    {
      "epoch": 2.584642620048683,
      "grad_norm": 1.854767918586731,
      "learning_rate": 9.758375392441284e-06,
      "loss": 0.2245,
      "step": 26280
    },
    {
      "epoch": 2.5856261218066927,
      "grad_norm": 2.4164562225341797,
      "learning_rate": 9.754401303501174e-06,
      "loss": 0.1707,
      "step": 26290
    },
    {
      "epoch": 2.5866096235647023,
      "grad_norm": 0.44090378284454346,
      "learning_rate": 9.750427214561062e-06,
      "loss": 0.2257,
      "step": 26300
    },
    {
      "epoch": 2.5875931253227114,
      "grad_norm": 0.8784712553024292,
      "learning_rate": 9.746453125620953e-06,
      "loss": 0.221,
      "step": 26310
    },
    {
      "epoch": 2.588576627080721,
      "grad_norm": 1.543505311012268,
      "learning_rate": 9.742479036680843e-06,
      "loss": 0.2557,
      "step": 26320
    },
    {
      "epoch": 2.5895601288387304,
      "grad_norm": 3.15250825881958,
      "learning_rate": 9.73850494774073e-06,
      "loss": 0.1896,
      "step": 26330
    },
    {
      "epoch": 2.5905436305967395,
      "grad_norm": 1.497484803199768,
      "learning_rate": 9.73453085880062e-06,
      "loss": 0.2873,
      "step": 26340
    },
    {
      "epoch": 2.591527132354749,
      "grad_norm": 1.0310429334640503,
      "learning_rate": 9.730556769860512e-06,
      "loss": 0.2716,
      "step": 26350
    },
    {
      "epoch": 2.5925106341127586,
      "grad_norm": 1.2639644145965576,
      "learning_rate": 9.7265826809204e-06,
      "loss": 0.1936,
      "step": 26360
    },
    {
      "epoch": 2.5934941358707677,
      "grad_norm": 0.6588050723075867,
      "learning_rate": 9.72260859198029e-06,
      "loss": 0.2893,
      "step": 26370
    },
    {
      "epoch": 2.5944776376287773,
      "grad_norm": 0.252906858921051,
      "learning_rate": 9.718634503040179e-06,
      "loss": 0.2306,
      "step": 26380
    },
    {
      "epoch": 2.595461139386787,
      "grad_norm": 1.3904304504394531,
      "learning_rate": 9.714660414100069e-06,
      "loss": 0.125,
      "step": 26390
    },
    {
      "epoch": 2.596444641144796,
      "grad_norm": 2.613572835922241,
      "learning_rate": 9.710686325159958e-06,
      "loss": 0.3689,
      "step": 26400
    },
    {
      "epoch": 2.5974281429028054,
      "grad_norm": 0.6632049679756165,
      "learning_rate": 9.706712236219848e-06,
      "loss": 0.1896,
      "step": 26410
    },
    {
      "epoch": 2.598411644660815,
      "grad_norm": 2.895082950592041,
      "learning_rate": 9.702738147279736e-06,
      "loss": 0.2164,
      "step": 26420
    },
    {
      "epoch": 2.599395146418824,
      "grad_norm": 0.6230047345161438,
      "learning_rate": 9.698764058339627e-06,
      "loss": 0.1824,
      "step": 26430
    },
    {
      "epoch": 2.6003786481768336,
      "grad_norm": 0.6954286694526672,
      "learning_rate": 9.694789969399517e-06,
      "loss": 0.2352,
      "step": 26440
    },
    {
      "epoch": 2.601362149934843,
      "grad_norm": 0.4880354106426239,
      "learning_rate": 9.690815880459405e-06,
      "loss": 0.1676,
      "step": 26450
    },
    {
      "epoch": 2.6023456516928523,
      "grad_norm": 0.8374060392379761,
      "learning_rate": 9.686841791519295e-06,
      "loss": 0.1599,
      "step": 26460
    },
    {
      "epoch": 2.603329153450862,
      "grad_norm": 1.3754957914352417,
      "learning_rate": 9.682867702579186e-06,
      "loss": 0.1887,
      "step": 26470
    },
    {
      "epoch": 2.6043126552088713,
      "grad_norm": 1.3043149709701538,
      "learning_rate": 9.678893613639074e-06,
      "loss": 0.2618,
      "step": 26480
    },
    {
      "epoch": 2.6052961569668804,
      "grad_norm": 0.8774238228797913,
      "learning_rate": 9.674919524698964e-06,
      "loss": 0.1974,
      "step": 26490
    },
    {
      "epoch": 2.60627965872489,
      "grad_norm": 2.9066178798675537,
      "learning_rate": 9.670945435758853e-06,
      "loss": 0.1928,
      "step": 26500
    },
    {
      "epoch": 2.60627965872489,
      "eval_loss": 0.12362838536500931,
      "eval_runtime": 16.6693,
      "eval_samples_per_second": 3.0,
      "eval_steps_per_second": 1.5,
      "step": 26500
    },
    {
      "epoch": 2.6072631604828995,
      "grad_norm": 1.5432078838348389,
      "learning_rate": 9.666971346818743e-06,
      "loss": 0.2182,
      "step": 26510
    },
    {
      "epoch": 2.6082466622409086,
      "grad_norm": 0.46753302216529846,
      "learning_rate": 9.662997257878633e-06,
      "loss": 0.2579,
      "step": 26520
    },
    {
      "epoch": 2.609230163998918,
      "grad_norm": 0.6570789217948914,
      "learning_rate": 9.659023168938522e-06,
      "loss": 0.2037,
      "step": 26530
    },
    {
      "epoch": 2.6102136657569277,
      "grad_norm": 0.64068204164505,
      "learning_rate": 9.65504907999841e-06,
      "loss": 0.1439,
      "step": 26540
    },
    {
      "epoch": 2.611197167514937,
      "grad_norm": 0.6301237344741821,
      "learning_rate": 9.6510749910583e-06,
      "loss": 0.1773,
      "step": 26550
    },
    {
      "epoch": 2.6121806692729463,
      "grad_norm": 0.8529449105262756,
      "learning_rate": 9.647100902118191e-06,
      "loss": 0.0962,
      "step": 26560
    },
    {
      "epoch": 2.613164171030956,
      "grad_norm": 1.1946789026260376,
      "learning_rate": 9.64312681317808e-06,
      "loss": 0.2033,
      "step": 26570
    },
    {
      "epoch": 2.614147672788965,
      "grad_norm": 0.19759801030158997,
      "learning_rate": 9.639152724237969e-06,
      "loss": 0.2209,
      "step": 26580
    },
    {
      "epoch": 2.6151311745469745,
      "grad_norm": 5.525092124938965,
      "learning_rate": 9.635178635297858e-06,
      "loss": 0.2819,
      "step": 26590
    },
    {
      "epoch": 2.616114676304984,
      "grad_norm": 2.7954213619232178,
      "learning_rate": 9.631204546357748e-06,
      "loss": 0.2135,
      "step": 26600
    },
    {
      "epoch": 2.617098178062993,
      "grad_norm": 0.28763845562934875,
      "learning_rate": 9.627230457417638e-06,
      "loss": 0.1211,
      "step": 26610
    },
    {
      "epoch": 2.6180816798210027,
      "grad_norm": 0.9569519758224487,
      "learning_rate": 9.623256368477527e-06,
      "loss": 0.2459,
      "step": 26620
    },
    {
      "epoch": 2.6190651815790122,
      "grad_norm": 0.3743652105331421,
      "learning_rate": 9.619282279537417e-06,
      "loss": 0.2082,
      "step": 26630
    },
    {
      "epoch": 2.6200486833370213,
      "grad_norm": 0.7890741229057312,
      "learning_rate": 9.615308190597307e-06,
      "loss": 0.1258,
      "step": 26640
    },
    {
      "epoch": 2.621032185095031,
      "grad_norm": 0.48824724555015564,
      "learning_rate": 9.611334101657196e-06,
      "loss": 0.2455,
      "step": 26650
    },
    {
      "epoch": 2.6220156868530404,
      "grad_norm": 0.6401808261871338,
      "learning_rate": 9.607360012717084e-06,
      "loss": 0.1541,
      "step": 26660
    },
    {
      "epoch": 2.6229991886110495,
      "grad_norm": 0.747993528842926,
      "learning_rate": 9.603385923776974e-06,
      "loss": 0.2794,
      "step": 26670
    },
    {
      "epoch": 2.623982690369059,
      "grad_norm": 0.5995237231254578,
      "learning_rate": 9.599411834836865e-06,
      "loss": 0.2389,
      "step": 26680
    },
    {
      "epoch": 2.6249661921270686,
      "grad_norm": 1.6733250617980957,
      "learning_rate": 9.595437745896753e-06,
      "loss": 0.1522,
      "step": 26690
    },
    {
      "epoch": 2.6259496938850777,
      "grad_norm": 2.178950071334839,
      "learning_rate": 9.591463656956643e-06,
      "loss": 0.1937,
      "step": 26700
    },
    {
      "epoch": 2.6269331956430872,
      "grad_norm": 1.5474810600280762,
      "learning_rate": 9.587489568016533e-06,
      "loss": 0.2734,
      "step": 26710
    },
    {
      "epoch": 2.627916697401097,
      "grad_norm": 1.2631654739379883,
      "learning_rate": 9.583515479076422e-06,
      "loss": 0.2732,
      "step": 26720
    },
    {
      "epoch": 2.628900199159106,
      "grad_norm": 0.35731109976768494,
      "learning_rate": 9.579541390136312e-06,
      "loss": 0.1897,
      "step": 26730
    },
    {
      "epoch": 2.6298837009171154,
      "grad_norm": 0.9533439874649048,
      "learning_rate": 9.575567301196202e-06,
      "loss": 0.1651,
      "step": 26740
    },
    {
      "epoch": 2.630867202675125,
      "grad_norm": 1.1866984367370605,
      "learning_rate": 9.571593212256091e-06,
      "loss": 0.1431,
      "step": 26750
    },
    {
      "epoch": 2.631850704433134,
      "grad_norm": 1.5564192533493042,
      "learning_rate": 9.567619123315981e-06,
      "loss": 0.2232,
      "step": 26760
    },
    {
      "epoch": 2.6328342061911436,
      "grad_norm": 1.2694995403289795,
      "learning_rate": 9.56364503437587e-06,
      "loss": 0.1206,
      "step": 26770
    },
    {
      "epoch": 2.633817707949153,
      "grad_norm": 0.9269311428070068,
      "learning_rate": 9.559670945435759e-06,
      "loss": 0.2766,
      "step": 26780
    },
    {
      "epoch": 2.6348012097071623,
      "grad_norm": 4.775659084320068,
      "learning_rate": 9.555696856495648e-06,
      "loss": 0.149,
      "step": 26790
    },
    {
      "epoch": 2.635784711465172,
      "grad_norm": 1.256229281425476,
      "learning_rate": 9.55172276755554e-06,
      "loss": 0.2743,
      "step": 26800
    },
    {
      "epoch": 2.6367682132231813,
      "grad_norm": 0.9255625009536743,
      "learning_rate": 9.547748678615428e-06,
      "loss": 0.2241,
      "step": 26810
    },
    {
      "epoch": 2.6377517149811904,
      "grad_norm": 1.4762382507324219,
      "learning_rate": 9.543774589675317e-06,
      "loss": 0.2827,
      "step": 26820
    },
    {
      "epoch": 2.6387352167392,
      "grad_norm": 1.1053285598754883,
      "learning_rate": 9.539800500735207e-06,
      "loss": 0.1927,
      "step": 26830
    },
    {
      "epoch": 2.6397187184972095,
      "grad_norm": 0.5553393959999084,
      "learning_rate": 9.535826411795097e-06,
      "loss": 0.2214,
      "step": 26840
    },
    {
      "epoch": 2.6407022202552186,
      "grad_norm": 1.2319566011428833,
      "learning_rate": 9.531852322854986e-06,
      "loss": 0.2522,
      "step": 26850
    },
    {
      "epoch": 2.641685722013228,
      "grad_norm": 2.134190320968628,
      "learning_rate": 9.527878233914876e-06,
      "loss": 0.3017,
      "step": 26860
    },
    {
      "epoch": 2.6426692237712377,
      "grad_norm": 0.22002668678760529,
      "learning_rate": 9.523904144974766e-06,
      "loss": 0.1739,
      "step": 26870
    },
    {
      "epoch": 2.643652725529247,
      "grad_norm": 0.21295656263828278,
      "learning_rate": 9.519930056034655e-06,
      "loss": 0.2597,
      "step": 26880
    },
    {
      "epoch": 2.6446362272872563,
      "grad_norm": 1.9469270706176758,
      "learning_rate": 9.515955967094545e-06,
      "loss": 0.1453,
      "step": 26890
    },
    {
      "epoch": 2.645619729045266,
      "grad_norm": 0.90763258934021,
      "learning_rate": 9.511981878154433e-06,
      "loss": 0.2524,
      "step": 26900
    },
    {
      "epoch": 2.646603230803275,
      "grad_norm": 1.8186426162719727,
      "learning_rate": 9.508007789214323e-06,
      "loss": 0.2746,
      "step": 26910
    },
    {
      "epoch": 2.6475867325612845,
      "grad_norm": 1.2653254270553589,
      "learning_rate": 9.504033700274214e-06,
      "loss": 0.2519,
      "step": 26920
    },
    {
      "epoch": 2.648570234319294,
      "grad_norm": 1.2931697368621826,
      "learning_rate": 9.500059611334102e-06,
      "loss": 0.1913,
      "step": 26930
    },
    {
      "epoch": 2.649553736077303,
      "grad_norm": 0.5535080432891846,
      "learning_rate": 9.496085522393992e-06,
      "loss": 0.209,
      "step": 26940
    },
    {
      "epoch": 2.6505372378353127,
      "grad_norm": 0.6815732717514038,
      "learning_rate": 9.492111433453881e-06,
      "loss": 0.1653,
      "step": 26950
    },
    {
      "epoch": 2.6515207395933222,
      "grad_norm": 0.4440705180168152,
      "learning_rate": 9.488137344513771e-06,
      "loss": 0.1697,
      "step": 26960
    },
    {
      "epoch": 2.6525042413513313,
      "grad_norm": 0.5131239891052246,
      "learning_rate": 9.48416325557366e-06,
      "loss": 0.1631,
      "step": 26970
    },
    {
      "epoch": 2.653487743109341,
      "grad_norm": 0.3114909529685974,
      "learning_rate": 9.48018916663355e-06,
      "loss": 0.1937,
      "step": 26980
    },
    {
      "epoch": 2.6544712448673504,
      "grad_norm": 1.17475426197052,
      "learning_rate": 9.47621507769344e-06,
      "loss": 0.1177,
      "step": 26990
    },
    {
      "epoch": 2.6554547466253595,
      "grad_norm": 2.7243964672088623,
      "learning_rate": 9.47224098875333e-06,
      "loss": 0.2956,
      "step": 27000
    },
    {
      "epoch": 2.6554547466253595,
      "eval_loss": 0.12182632088661194,
      "eval_runtime": 18.2151,
      "eval_samples_per_second": 2.745,
      "eval_steps_per_second": 1.372,
      "step": 27000
    },
    {
      "epoch": 2.656438248383369,
      "grad_norm": 1.5404561758041382,
      "learning_rate": 9.46826689981322e-06,
      "loss": 0.2371,
      "step": 27010
    },
    {
      "epoch": 2.6574217501413786,
      "grad_norm": 0.9724194407463074,
      "learning_rate": 9.464292810873107e-06,
      "loss": 0.1877,
      "step": 27020
    },
    {
      "epoch": 2.6584052518993877,
      "grad_norm": 1.9115068912506104,
      "learning_rate": 9.460318721932997e-06,
      "loss": 0.203,
      "step": 27030
    },
    {
      "epoch": 2.6593887536573972,
      "grad_norm": 1.4371966123580933,
      "learning_rate": 9.456344632992888e-06,
      "loss": 0.2567,
      "step": 27040
    },
    {
      "epoch": 2.660372255415407,
      "grad_norm": 1.023058295249939,
      "learning_rate": 9.452370544052776e-06,
      "loss": 0.1497,
      "step": 27050
    },
    {
      "epoch": 2.661355757173416,
      "grad_norm": 2.9103660583496094,
      "learning_rate": 9.448396455112666e-06,
      "loss": 0.1619,
      "step": 27060
    },
    {
      "epoch": 2.6623392589314254,
      "grad_norm": 0.5257496237754822,
      "learning_rate": 9.444422366172556e-06,
      "loss": 0.1668,
      "step": 27070
    },
    {
      "epoch": 2.663322760689435,
      "grad_norm": 0.3518742620944977,
      "learning_rate": 9.440448277232445e-06,
      "loss": 0.2312,
      "step": 27080
    },
    {
      "epoch": 2.664306262447444,
      "grad_norm": 0.6371358633041382,
      "learning_rate": 9.436474188292335e-06,
      "loss": 0.138,
      "step": 27090
    },
    {
      "epoch": 2.6652897642054536,
      "grad_norm": 1.0923123359680176,
      "learning_rate": 9.432500099352225e-06,
      "loss": 0.1703,
      "step": 27100
    },
    {
      "epoch": 2.666273265963463,
      "grad_norm": 0.8021963238716125,
      "learning_rate": 9.428526010412114e-06,
      "loss": 0.2727,
      "step": 27110
    },
    {
      "epoch": 2.6672567677214722,
      "grad_norm": 1.092530369758606,
      "learning_rate": 9.424551921472004e-06,
      "loss": 0.3269,
      "step": 27120
    },
    {
      "epoch": 2.668240269479482,
      "grad_norm": 2.668480634689331,
      "learning_rate": 9.420577832531894e-06,
      "loss": 0.2236,
      "step": 27130
    },
    {
      "epoch": 2.6692237712374913,
      "grad_norm": 2.236377239227295,
      "learning_rate": 9.416603743591782e-06,
      "loss": 0.2983,
      "step": 27140
    },
    {
      "epoch": 2.6702072729955004,
      "grad_norm": 1.4047900438308716,
      "learning_rate": 9.412629654651671e-06,
      "loss": 0.2974,
      "step": 27150
    },
    {
      "epoch": 2.67119077475351,
      "grad_norm": 0.9929019808769226,
      "learning_rate": 9.408655565711563e-06,
      "loss": 0.1797,
      "step": 27160
    },
    {
      "epoch": 2.6721742765115195,
      "grad_norm": 0.6414074897766113,
      "learning_rate": 9.40468147677145e-06,
      "loss": 0.1573,
      "step": 27170
    },
    {
      "epoch": 2.6731577782695286,
      "grad_norm": 0.6707764863967896,
      "learning_rate": 9.40070738783134e-06,
      "loss": 0.1535,
      "step": 27180
    },
    {
      "epoch": 2.674141280027538,
      "grad_norm": 1.877697467803955,
      "learning_rate": 9.39673329889123e-06,
      "loss": 0.2274,
      "step": 27190
    },
    {
      "epoch": 2.6751247817855477,
      "grad_norm": 0.78323894739151,
      "learning_rate": 9.39275920995112e-06,
      "loss": 0.295,
      "step": 27200
    },
    {
      "epoch": 2.676108283543557,
      "grad_norm": 0.5460190773010254,
      "learning_rate": 9.38878512101101e-06,
      "loss": 0.2139,
      "step": 27210
    },
    {
      "epoch": 2.6770917853015663,
      "grad_norm": 1.3874690532684326,
      "learning_rate": 9.384811032070899e-06,
      "loss": 0.276,
      "step": 27220
    },
    {
      "epoch": 2.678075287059576,
      "grad_norm": 0.6501830220222473,
      "learning_rate": 9.380836943130789e-06,
      "loss": 0.2158,
      "step": 27230
    },
    {
      "epoch": 2.679058788817585,
      "grad_norm": 1.5910675525665283,
      "learning_rate": 9.376862854190678e-06,
      "loss": 0.2682,
      "step": 27240
    },
    {
      "epoch": 2.6800422905755945,
      "grad_norm": 1.3960938453674316,
      "learning_rate": 9.372888765250568e-06,
      "loss": 0.3153,
      "step": 27250
    },
    {
      "epoch": 2.681025792333604,
      "grad_norm": 0.7218887805938721,
      "learning_rate": 9.368914676310456e-06,
      "loss": 0.1431,
      "step": 27260
    },
    {
      "epoch": 2.682009294091613,
      "grad_norm": 0.6101197600364685,
      "learning_rate": 9.364940587370346e-06,
      "loss": 0.2299,
      "step": 27270
    },
    {
      "epoch": 2.6829927958496227,
      "grad_norm": 1.9330745935440063,
      "learning_rate": 9.360966498430235e-06,
      "loss": 0.1978,
      "step": 27280
    },
    {
      "epoch": 2.6839762976076322,
      "grad_norm": 0.9924318194389343,
      "learning_rate": 9.356992409490125e-06,
      "loss": 0.2722,
      "step": 27290
    },
    {
      "epoch": 2.6849597993656413,
      "grad_norm": 0.8466578722000122,
      "learning_rate": 9.353018320550015e-06,
      "loss": 0.3312,
      "step": 27300
    },
    {
      "epoch": 2.685943301123651,
      "grad_norm": 0.32302579283714294,
      "learning_rate": 9.349044231609904e-06,
      "loss": 0.267,
      "step": 27310
    },
    {
      "epoch": 2.6869268028816604,
      "grad_norm": 0.049537353217601776,
      "learning_rate": 9.345070142669794e-06,
      "loss": 0.1174,
      "step": 27320
    },
    {
      "epoch": 2.6879103046396695,
      "grad_norm": 1.555120587348938,
      "learning_rate": 9.341096053729684e-06,
      "loss": 0.1646,
      "step": 27330
    },
    {
      "epoch": 2.688893806397679,
      "grad_norm": 1.5903842449188232,
      "learning_rate": 9.337121964789573e-06,
      "loss": 0.2165,
      "step": 27340
    },
    {
      "epoch": 2.6898773081556886,
      "grad_norm": 1.6580713987350464,
      "learning_rate": 9.333147875849463e-06,
      "loss": 0.1619,
      "step": 27350
    },
    {
      "epoch": 2.6908608099136977,
      "grad_norm": 1.3948620557785034,
      "learning_rate": 9.32917378690935e-06,
      "loss": 0.1949,
      "step": 27360
    },
    {
      "epoch": 2.6918443116717072,
      "grad_norm": 1.0545307397842407,
      "learning_rate": 9.325199697969242e-06,
      "loss": 0.2992,
      "step": 27370
    },
    {
      "epoch": 2.6928278134297168,
      "grad_norm": 0.4301310181617737,
      "learning_rate": 9.32122560902913e-06,
      "loss": 0.1718,
      "step": 27380
    },
    {
      "epoch": 2.693811315187726,
      "grad_norm": 0.8084037899971008,
      "learning_rate": 9.31725152008902e-06,
      "loss": 0.2577,
      "step": 27390
    },
    {
      "epoch": 2.6947948169457354,
      "grad_norm": 0.48154643177986145,
      "learning_rate": 9.31327743114891e-06,
      "loss": 0.1771,
      "step": 27400
    },
    {
      "epoch": 2.695778318703745,
      "grad_norm": 1.021758794784546,
      "learning_rate": 9.309303342208799e-06,
      "loss": 0.2348,
      "step": 27410
    },
    {
      "epoch": 2.696761820461754,
      "grad_norm": 0.9262601137161255,
      "learning_rate": 9.305329253268689e-06,
      "loss": 0.2558,
      "step": 27420
    },
    {
      "epoch": 2.6977453222197636,
      "grad_norm": 0.8262643218040466,
      "learning_rate": 9.301355164328578e-06,
      "loss": 0.2068,
      "step": 27430
    },
    {
      "epoch": 2.6987288239777727,
      "grad_norm": 1.3612849712371826,
      "learning_rate": 9.297381075388468e-06,
      "loss": 0.21,
      "step": 27440
    },
    {
      "epoch": 2.6997123257357822,
      "grad_norm": 0.6201552748680115,
      "learning_rate": 9.293406986448358e-06,
      "loss": 0.1185,
      "step": 27450
    },
    {
      "epoch": 2.700695827493792,
      "grad_norm": 1.0904179811477661,
      "learning_rate": 9.289432897508247e-06,
      "loss": 0.2703,
      "step": 27460
    },
    {
      "epoch": 2.701679329251801,
      "grad_norm": 1.0167815685272217,
      "learning_rate": 9.285458808568137e-06,
      "loss": 0.1428,
      "step": 27470
    },
    {
      "epoch": 2.7026628310098104,
      "grad_norm": 1.3628720045089722,
      "learning_rate": 9.281484719628025e-06,
      "loss": 0.2142,
      "step": 27480
    },
    {
      "epoch": 2.70364633276782,
      "grad_norm": 0.7690005898475647,
      "learning_rate": 9.277510630687916e-06,
      "loss": 0.1802,
      "step": 27490
    },
    {
      "epoch": 2.704629834525829,
      "grad_norm": 0.9450478553771973,
      "learning_rate": 9.273536541747804e-06,
      "loss": 0.2007,
      "step": 27500
    },
    {
      "epoch": 2.704629834525829,
      "eval_loss": 0.11772012710571289,
      "eval_runtime": 18.9836,
      "eval_samples_per_second": 2.634,
      "eval_steps_per_second": 1.317,
      "step": 27500
    },
    {
      "epoch": 2.7056133362838386,
      "grad_norm": 11.136773109436035,
      "learning_rate": 9.269562452807694e-06,
      "loss": 0.1494,
      "step": 27510
    },
    {
      "epoch": 2.706596838041848,
      "grad_norm": 1.8330661058425903,
      "learning_rate": 9.265588363867584e-06,
      "loss": 0.2296,
      "step": 27520
    },
    {
      "epoch": 2.7075803397998572,
      "grad_norm": 0.5862045884132385,
      "learning_rate": 9.261614274927473e-06,
      "loss": 0.1731,
      "step": 27530
    },
    {
      "epoch": 2.708563841557867,
      "grad_norm": 1.7953245639801025,
      "learning_rate": 9.257640185987363e-06,
      "loss": 0.2048,
      "step": 27540
    },
    {
      "epoch": 2.7095473433158763,
      "grad_norm": 0.5008899569511414,
      "learning_rate": 9.253666097047253e-06,
      "loss": 0.2088,
      "step": 27550
    },
    {
      "epoch": 2.7105308450738854,
      "grad_norm": 1.6881611347198486,
      "learning_rate": 9.249692008107142e-06,
      "loss": 0.2589,
      "step": 27560
    },
    {
      "epoch": 2.711514346831895,
      "grad_norm": 2.1870365142822266,
      "learning_rate": 9.245717919167032e-06,
      "loss": 0.1861,
      "step": 27570
    },
    {
      "epoch": 2.7124978485899045,
      "grad_norm": 0.7728579640388489,
      "learning_rate": 9.241743830226922e-06,
      "loss": 0.2845,
      "step": 27580
    },
    {
      "epoch": 2.7134813503479136,
      "grad_norm": 0.534707248210907,
      "learning_rate": 9.237769741286811e-06,
      "loss": 0.138,
      "step": 27590
    },
    {
      "epoch": 2.714464852105923,
      "grad_norm": 0.7656206488609314,
      "learning_rate": 9.2337956523467e-06,
      "loss": 0.1957,
      "step": 27600
    },
    {
      "epoch": 2.7154483538639327,
      "grad_norm": 1.1382602453231812,
      "learning_rate": 9.22982156340659e-06,
      "loss": 0.1841,
      "step": 27610
    },
    {
      "epoch": 2.716431855621942,
      "grad_norm": 0.698901355266571,
      "learning_rate": 9.225847474466479e-06,
      "loss": 0.1717,
      "step": 27620
    },
    {
      "epoch": 2.7174153573799513,
      "grad_norm": 1.258329153060913,
      "learning_rate": 9.221873385526368e-06,
      "loss": 0.1288,
      "step": 27630
    },
    {
      "epoch": 2.718398859137961,
      "grad_norm": 0.5331089496612549,
      "learning_rate": 9.217899296586258e-06,
      "loss": 0.224,
      "step": 27640
    },
    {
      "epoch": 2.71938236089597,
      "grad_norm": 0.28912073373794556,
      "learning_rate": 9.213925207646148e-06,
      "loss": 0.2152,
      "step": 27650
    },
    {
      "epoch": 2.7203658626539795,
      "grad_norm": 0.4922606945037842,
      "learning_rate": 9.209951118706037e-06,
      "loss": 0.1785,
      "step": 27660
    },
    {
      "epoch": 2.721349364411989,
      "grad_norm": 0.6296001076698303,
      "learning_rate": 9.205977029765927e-06,
      "loss": 0.1462,
      "step": 27670
    },
    {
      "epoch": 2.722332866169998,
      "grad_norm": 0.6022535562515259,
      "learning_rate": 9.202002940825817e-06,
      "loss": 0.1855,
      "step": 27680
    },
    {
      "epoch": 2.7233163679280077,
      "grad_norm": 1.7227388620376587,
      "learning_rate": 9.198028851885706e-06,
      "loss": 0.1792,
      "step": 27690
    },
    {
      "epoch": 2.7242998696860172,
      "grad_norm": 0.5050339698791504,
      "learning_rate": 9.194054762945596e-06,
      "loss": 0.226,
      "step": 27700
    },
    {
      "epoch": 2.7252833714440263,
      "grad_norm": 0.4780879020690918,
      "learning_rate": 9.190080674005486e-06,
      "loss": 0.2302,
      "step": 27710
    },
    {
      "epoch": 2.726266873202036,
      "grad_norm": 0.1593957096338272,
      "learning_rate": 9.186106585065374e-06,
      "loss": 0.1973,
      "step": 27720
    },
    {
      "epoch": 2.727250374960045,
      "grad_norm": 0.3137963116168976,
      "learning_rate": 9.182132496125265e-06,
      "loss": 0.1865,
      "step": 27730
    },
    {
      "epoch": 2.7282338767180545,
      "grad_norm": 3.1869826316833496,
      "learning_rate": 9.178158407185153e-06,
      "loss": 0.2086,
      "step": 27740
    },
    {
      "epoch": 2.729217378476064,
      "grad_norm": 1.0982507467269897,
      "learning_rate": 9.174184318245043e-06,
      "loss": 0.2532,
      "step": 27750
    },
    {
      "epoch": 2.730200880234073,
      "grad_norm": 2.14682674407959,
      "learning_rate": 9.170210229304932e-06,
      "loss": 0.1477,
      "step": 27760
    },
    {
      "epoch": 2.7311843819920827,
      "grad_norm": 1.5476717948913574,
      "learning_rate": 9.166236140364822e-06,
      "loss": 0.1635,
      "step": 27770
    },
    {
      "epoch": 2.7321678837500922,
      "grad_norm": 1.0337785482406616,
      "learning_rate": 9.162262051424712e-06,
      "loss": 0.2289,
      "step": 27780
    },
    {
      "epoch": 2.7331513855081013,
      "grad_norm": 1.0297173261642456,
      "learning_rate": 9.158287962484601e-06,
      "loss": 0.303,
      "step": 27790
    },
    {
      "epoch": 2.734134887266111,
      "grad_norm": 0.8042473196983337,
      "learning_rate": 9.154313873544491e-06,
      "loss": 0.1754,
      "step": 27800
    },
    {
      "epoch": 2.7351183890241204,
      "grad_norm": 1.1096657514572144,
      "learning_rate": 9.15033978460438e-06,
      "loss": 0.2365,
      "step": 27810
    },
    {
      "epoch": 2.7361018907821295,
      "grad_norm": 0.8313658237457275,
      "learning_rate": 9.14636569566427e-06,
      "loss": 0.1505,
      "step": 27820
    },
    {
      "epoch": 2.737085392540139,
      "grad_norm": 0.7719443440437317,
      "learning_rate": 9.14239160672416e-06,
      "loss": 0.1858,
      "step": 27830
    },
    {
      "epoch": 2.7380688942981486,
      "grad_norm": 0.19261518120765686,
      "learning_rate": 9.138417517784048e-06,
      "loss": 0.2037,
      "step": 27840
    },
    {
      "epoch": 2.7390523960561577,
      "grad_norm": 0.7655139565467834,
      "learning_rate": 9.13444342884394e-06,
      "loss": 0.274,
      "step": 27850
    },
    {
      "epoch": 2.7400358978141672,
      "grad_norm": 0.6452736258506775,
      "learning_rate": 9.130469339903827e-06,
      "loss": 0.2163,
      "step": 27860
    },
    {
      "epoch": 2.7410193995721768,
      "grad_norm": 1.4266693592071533,
      "learning_rate": 9.126495250963717e-06,
      "loss": 0.1944,
      "step": 27870
    },
    {
      "epoch": 2.742002901330186,
      "grad_norm": 1.0404728651046753,
      "learning_rate": 9.122521162023607e-06,
      "loss": 0.2191,
      "step": 27880
    },
    {
      "epoch": 2.7429864030881954,
      "grad_norm": 1.3982700109481812,
      "learning_rate": 9.118547073083496e-06,
      "loss": 0.1796,
      "step": 27890
    },
    {
      "epoch": 2.743969904846205,
      "grad_norm": 1.3127501010894775,
      "learning_rate": 9.114572984143386e-06,
      "loss": 0.2794,
      "step": 27900
    },
    {
      "epoch": 2.744953406604214,
      "grad_norm": 0.7358602285385132,
      "learning_rate": 9.110598895203276e-06,
      "loss": 0.3201,
      "step": 27910
    },
    {
      "epoch": 2.7459369083622236,
      "grad_norm": 1.1799973249435425,
      "learning_rate": 9.106624806263165e-06,
      "loss": 0.2373,
      "step": 27920
    },
    {
      "epoch": 2.746920410120233,
      "grad_norm": 0.52203768491745,
      "learning_rate": 9.102650717323055e-06,
      "loss": 0.1438,
      "step": 27930
    },
    {
      "epoch": 2.7479039118782422,
      "grad_norm": 1.5550827980041504,
      "learning_rate": 9.098676628382945e-06,
      "loss": 0.1915,
      "step": 27940
    },
    {
      "epoch": 2.7488874136362518,
      "grad_norm": 1.1797600984573364,
      "learning_rate": 9.094702539442834e-06,
      "loss": 0.1632,
      "step": 27950
    },
    {
      "epoch": 2.7498709153942613,
      "grad_norm": 0.9325649738311768,
      "learning_rate": 9.090728450502722e-06,
      "loss": 0.1879,
      "step": 27960
    },
    {
      "epoch": 2.7508544171522704,
      "grad_norm": 1.2596094608306885,
      "learning_rate": 9.086754361562614e-06,
      "loss": 0.1654,
      "step": 27970
    },
    {
      "epoch": 2.75183791891028,
      "grad_norm": 0.39892297983169556,
      "learning_rate": 9.082780272622502e-06,
      "loss": 0.2034,
      "step": 27980
    },
    {
      "epoch": 2.7528214206682895,
      "grad_norm": 1.4378695487976074,
      "learning_rate": 9.078806183682391e-06,
      "loss": 0.2561,
      "step": 27990
    },
    {
      "epoch": 2.7538049224262986,
      "grad_norm": 1.9730724096298218,
      "learning_rate": 9.074832094742281e-06,
      "loss": 0.2129,
      "step": 28000
    },
    {
      "epoch": 2.7538049224262986,
      "eval_loss": 0.12233906984329224,
      "eval_runtime": 18.4862,
      "eval_samples_per_second": 2.705,
      "eval_steps_per_second": 1.352,
      "step": 28000
    },
    {
      "epoch": 2.754788424184308,
      "grad_norm": 1.2149182558059692,
      "learning_rate": 9.07085800580217e-06,
      "loss": 0.1312,
      "step": 28010
    },
    {
      "epoch": 2.7557719259423177,
      "grad_norm": 1.0150631666183472,
      "learning_rate": 9.06688391686206e-06,
      "loss": 0.287,
      "step": 28020
    },
    {
      "epoch": 2.7567554277003268,
      "grad_norm": 0.542308509349823,
      "learning_rate": 9.06290982792195e-06,
      "loss": 0.1862,
      "step": 28030
    },
    {
      "epoch": 2.7577389294583363,
      "grad_norm": 0.3378434479236603,
      "learning_rate": 9.05893573898184e-06,
      "loss": 0.2743,
      "step": 28040
    },
    {
      "epoch": 2.758722431216346,
      "grad_norm": 1.0021579265594482,
      "learning_rate": 9.05496165004173e-06,
      "loss": 0.1584,
      "step": 28050
    },
    {
      "epoch": 2.759705932974355,
      "grad_norm": 0.13331718742847443,
      "learning_rate": 9.050987561101619e-06,
      "loss": 0.1916,
      "step": 28060
    },
    {
      "epoch": 2.7606894347323645,
      "grad_norm": 2.321328639984131,
      "learning_rate": 9.047013472161509e-06,
      "loss": 0.175,
      "step": 28070
    },
    {
      "epoch": 2.761672936490374,
      "grad_norm": 0.6371372938156128,
      "learning_rate": 9.043039383221397e-06,
      "loss": 0.174,
      "step": 28080
    },
    {
      "epoch": 2.762656438248383,
      "grad_norm": 2.9457879066467285,
      "learning_rate": 9.039065294281286e-06,
      "loss": 0.2511,
      "step": 28090
    },
    {
      "epoch": 2.7636399400063927,
      "grad_norm": 0.9190905690193176,
      "learning_rate": 9.035091205341176e-06,
      "loss": 0.1659,
      "step": 28100
    },
    {
      "epoch": 2.7646234417644022,
      "grad_norm": 1.509871482849121,
      "learning_rate": 9.031117116401066e-06,
      "loss": 0.1308,
      "step": 28110
    },
    {
      "epoch": 2.7656069435224113,
      "grad_norm": 0.33418288826942444,
      "learning_rate": 9.027143027460955e-06,
      "loss": 0.1684,
      "step": 28120
    },
    {
      "epoch": 2.766590445280421,
      "grad_norm": 1.953349232673645,
      "learning_rate": 9.023168938520845e-06,
      "loss": 0.1792,
      "step": 28130
    },
    {
      "epoch": 2.7675739470384304,
      "grad_norm": 0.6116806864738464,
      "learning_rate": 9.019194849580735e-06,
      "loss": 0.2809,
      "step": 28140
    },
    {
      "epoch": 2.7685574487964395,
      "grad_norm": 0.8784261345863342,
      "learning_rate": 9.015220760640624e-06,
      "loss": 0.1352,
      "step": 28150
    },
    {
      "epoch": 2.769540950554449,
      "grad_norm": 1.326364278793335,
      "learning_rate": 9.011246671700514e-06,
      "loss": 0.2668,
      "step": 28160
    },
    {
      "epoch": 2.7705244523124586,
      "grad_norm": 0.5899569392204285,
      "learning_rate": 9.007272582760402e-06,
      "loss": 0.3221,
      "step": 28170
    },
    {
      "epoch": 2.7715079540704677,
      "grad_norm": 0.5236835479736328,
      "learning_rate": 9.003298493820293e-06,
      "loss": 0.2541,
      "step": 28180
    },
    {
      "epoch": 2.7724914558284772,
      "grad_norm": 1.8051170110702515,
      "learning_rate": 8.999324404880183e-06,
      "loss": 0.2342,
      "step": 28190
    },
    {
      "epoch": 2.7734749575864868,
      "grad_norm": 0.22561007738113403,
      "learning_rate": 8.99535031594007e-06,
      "loss": 0.2332,
      "step": 28200
    },
    {
      "epoch": 2.774458459344496,
      "grad_norm": 0.8784757256507874,
      "learning_rate": 8.99137622699996e-06,
      "loss": 0.1558,
      "step": 28210
    },
    {
      "epoch": 2.7754419611025054,
      "grad_norm": 0.8380150198936462,
      "learning_rate": 8.98740213805985e-06,
      "loss": 0.224,
      "step": 28220
    },
    {
      "epoch": 2.776425462860515,
      "grad_norm": 0.8617953658103943,
      "learning_rate": 8.98342804911974e-06,
      "loss": 0.2752,
      "step": 28230
    },
    {
      "epoch": 2.777408964618524,
      "grad_norm": 1.576338291168213,
      "learning_rate": 8.97945396017963e-06,
      "loss": 0.1994,
      "step": 28240
    },
    {
      "epoch": 2.7783924663765336,
      "grad_norm": 0.359642893075943,
      "learning_rate": 8.975479871239519e-06,
      "loss": 0.1899,
      "step": 28250
    },
    {
      "epoch": 2.779375968134543,
      "grad_norm": 1.1645811796188354,
      "learning_rate": 8.971505782299409e-06,
      "loss": 0.2148,
      "step": 28260
    },
    {
      "epoch": 2.7803594698925522,
      "grad_norm": 1.116481900215149,
      "learning_rate": 8.967531693359298e-06,
      "loss": 0.2202,
      "step": 28270
    },
    {
      "epoch": 2.7813429716505618,
      "grad_norm": 0.5225256681442261,
      "learning_rate": 8.963557604419188e-06,
      "loss": 0.1449,
      "step": 28280
    },
    {
      "epoch": 2.7823264734085713,
      "grad_norm": 0.782061755657196,
      "learning_rate": 8.959583515479076e-06,
      "loss": 0.2793,
      "step": 28290
    },
    {
      "epoch": 2.7833099751665804,
      "grad_norm": 0.6921595931053162,
      "learning_rate": 8.955609426538967e-06,
      "loss": 0.12,
      "step": 28300
    },
    {
      "epoch": 2.78429347692459,
      "grad_norm": 0.6155844926834106,
      "learning_rate": 8.951635337598857e-06,
      "loss": 0.187,
      "step": 28310
    },
    {
      "epoch": 2.7852769786825995,
      "grad_norm": 0.61292564868927,
      "learning_rate": 8.947661248658745e-06,
      "loss": 0.1792,
      "step": 28320
    },
    {
      "epoch": 2.7862604804406086,
      "grad_norm": 5.320784091949463,
      "learning_rate": 8.943687159718635e-06,
      "loss": 0.1971,
      "step": 28330
    },
    {
      "epoch": 2.787243982198618,
      "grad_norm": 1.5784276723861694,
      "learning_rate": 8.939713070778524e-06,
      "loss": 0.3297,
      "step": 28340
    },
    {
      "epoch": 2.7882274839566277,
      "grad_norm": 0.4494692087173462,
      "learning_rate": 8.935738981838414e-06,
      "loss": 0.1365,
      "step": 28350
    },
    {
      "epoch": 2.7892109857146368,
      "grad_norm": 0.41863226890563965,
      "learning_rate": 8.931764892898304e-06,
      "loss": 0.1865,
      "step": 28360
    },
    {
      "epoch": 2.7901944874726463,
      "grad_norm": 0.506903886795044,
      "learning_rate": 8.927790803958193e-06,
      "loss": 0.2627,
      "step": 28370
    },
    {
      "epoch": 2.791177989230656,
      "grad_norm": 1.0100220441818237,
      "learning_rate": 8.923816715018083e-06,
      "loss": 0.2728,
      "step": 28380
    },
    {
      "epoch": 2.792161490988665,
      "grad_norm": 0.788859486579895,
      "learning_rate": 8.919842626077973e-06,
      "loss": 0.2081,
      "step": 28390
    },
    {
      "epoch": 2.7931449927466745,
      "grad_norm": 0.8792150616645813,
      "learning_rate": 8.915868537137862e-06,
      "loss": 0.2177,
      "step": 28400
    },
    {
      "epoch": 2.794128494504684,
      "grad_norm": 1.0238473415374756,
      "learning_rate": 8.91189444819775e-06,
      "loss": 0.2368,
      "step": 28410
    },
    {
      "epoch": 2.795111996262693,
      "grad_norm": 2.283637046813965,
      "learning_rate": 8.907920359257642e-06,
      "loss": 0.2786,
      "step": 28420
    },
    {
      "epoch": 2.7960954980207027,
      "grad_norm": 0.11248058825731277,
      "learning_rate": 8.903946270317531e-06,
      "loss": 0.1421,
      "step": 28430
    },
    {
      "epoch": 2.797078999778712,
      "grad_norm": 2.3209190368652344,
      "learning_rate": 8.89997218137742e-06,
      "loss": 0.1536,
      "step": 28440
    },
    {
      "epoch": 2.7980625015367213,
      "grad_norm": 0.4981313943862915,
      "learning_rate": 8.895998092437309e-06,
      "loss": 0.1468,
      "step": 28450
    },
    {
      "epoch": 2.799046003294731,
      "grad_norm": 0.35236284136772156,
      "learning_rate": 8.892024003497199e-06,
      "loss": 0.2641,
      "step": 28460
    },
    {
      "epoch": 2.8000295050527404,
      "grad_norm": 4.2277703285217285,
      "learning_rate": 8.888049914557088e-06,
      "loss": 0.1715,
      "step": 28470
    },
    {
      "epoch": 2.8010130068107495,
      "grad_norm": 0.7296504378318787,
      "learning_rate": 8.884075825616978e-06,
      "loss": 0.1647,
      "step": 28480
    },
    {
      "epoch": 2.801996508568759,
      "grad_norm": 2.090623140335083,
      "learning_rate": 8.880101736676868e-06,
      "loss": 0.1697,
      "step": 28490
    },
    {
      "epoch": 2.8029800103267686,
      "grad_norm": 0.14077714085578918,
      "learning_rate": 8.876127647736757e-06,
      "loss": 0.1929,
      "step": 28500
    },
    {
      "epoch": 2.8029800103267686,
      "eval_loss": 0.12228866666555405,
      "eval_runtime": 20.9307,
      "eval_samples_per_second": 2.389,
      "eval_steps_per_second": 1.194,
      "step": 28500
    },
    {
      "epoch": 2.8039635120847777,
      "grad_norm": 2.1589229106903076,
      "learning_rate": 8.872153558796647e-06,
      "loss": 0.2132,
      "step": 28510
    },
    {
      "epoch": 2.804947013842787,
      "grad_norm": 0.2110229879617691,
      "learning_rate": 8.868179469856537e-06,
      "loss": 0.1616,
      "step": 28520
    },
    {
      "epoch": 2.8059305156007968,
      "grad_norm": 1.5847959518432617,
      "learning_rate": 8.864205380916425e-06,
      "loss": 0.1359,
      "step": 28530
    },
    {
      "epoch": 2.806914017358806,
      "grad_norm": 0.6458785533905029,
      "learning_rate": 8.860231291976316e-06,
      "loss": 0.1785,
      "step": 28540
    },
    {
      "epoch": 2.8078975191168154,
      "grad_norm": 0.5321483016014099,
      "learning_rate": 8.856257203036206e-06,
      "loss": 0.2318,
      "step": 28550
    },
    {
      "epoch": 2.808881020874825,
      "grad_norm": 1.2807509899139404,
      "learning_rate": 8.852283114096094e-06,
      "loss": 0.1583,
      "step": 28560
    },
    {
      "epoch": 2.809864522632834,
      "grad_norm": 0.766925573348999,
      "learning_rate": 8.848309025155983e-06,
      "loss": 0.2679,
      "step": 28570
    },
    {
      "epoch": 2.8108480243908436,
      "grad_norm": 0.6170240044593811,
      "learning_rate": 8.844334936215873e-06,
      "loss": 0.2493,
      "step": 28580
    },
    {
      "epoch": 2.811831526148853,
      "grad_norm": 1.8137459754943848,
      "learning_rate": 8.840360847275763e-06,
      "loss": 0.2313,
      "step": 28590
    },
    {
      "epoch": 2.812815027906862,
      "grad_norm": 1.1472567319869995,
      "learning_rate": 8.836386758335652e-06,
      "loss": 0.1704,
      "step": 28600
    },
    {
      "epoch": 2.8137985296648718,
      "grad_norm": 2.9038665294647217,
      "learning_rate": 8.832412669395542e-06,
      "loss": 0.2366,
      "step": 28610
    },
    {
      "epoch": 2.8147820314228813,
      "grad_norm": 1.5422487258911133,
      "learning_rate": 8.828438580455432e-06,
      "loss": 0.0977,
      "step": 28620
    },
    {
      "epoch": 2.8157655331808904,
      "grad_norm": 1.3508249521255493,
      "learning_rate": 8.824464491515321e-06,
      "loss": 0.2698,
      "step": 28630
    },
    {
      "epoch": 2.8167490349389,
      "grad_norm": 0.6246904134750366,
      "learning_rate": 8.820490402575211e-06,
      "loss": 0.1542,
      "step": 28640
    },
    {
      "epoch": 2.8177325366969095,
      "grad_norm": 0.2572641968727112,
      "learning_rate": 8.816516313635099e-06,
      "loss": 0.2488,
      "step": 28650
    },
    {
      "epoch": 2.8187160384549186,
      "grad_norm": 0.16635353863239288,
      "learning_rate": 8.81254222469499e-06,
      "loss": 0.2405,
      "step": 28660
    },
    {
      "epoch": 2.819699540212928,
      "grad_norm": 1.9076553583145142,
      "learning_rate": 8.80856813575488e-06,
      "loss": 0.329,
      "step": 28670
    },
    {
      "epoch": 2.8206830419709377,
      "grad_norm": 0.9525811672210693,
      "learning_rate": 8.804594046814768e-06,
      "loss": 0.1924,
      "step": 28680
    },
    {
      "epoch": 2.8216665437289468,
      "grad_norm": 1.3024401664733887,
      "learning_rate": 8.800619957874658e-06,
      "loss": 0.2242,
      "step": 28690
    },
    {
      "epoch": 2.8226500454869563,
      "grad_norm": 0.6237722635269165,
      "learning_rate": 8.796645868934547e-06,
      "loss": 0.2294,
      "step": 28700
    },
    {
      "epoch": 2.823633547244966,
      "grad_norm": 0.5489286184310913,
      "learning_rate": 8.792671779994437e-06,
      "loss": 0.1815,
      "step": 28710
    },
    {
      "epoch": 2.824617049002975,
      "grad_norm": 0.7488086819648743,
      "learning_rate": 8.788697691054327e-06,
      "loss": 0.1495,
      "step": 28720
    },
    {
      "epoch": 2.8256005507609845,
      "grad_norm": 1.6106791496276855,
      "learning_rate": 8.784723602114216e-06,
      "loss": 0.2004,
      "step": 28730
    },
    {
      "epoch": 2.826584052518994,
      "grad_norm": 1.1848503351211548,
      "learning_rate": 8.780749513174106e-06,
      "loss": 0.3045,
      "step": 28740
    },
    {
      "epoch": 2.827567554277003,
      "grad_norm": 1.018410563468933,
      "learning_rate": 8.776775424233996e-06,
      "loss": 0.2271,
      "step": 28750
    },
    {
      "epoch": 2.8285510560350127,
      "grad_norm": 0.7283086776733398,
      "learning_rate": 8.772801335293885e-06,
      "loss": 0.3003,
      "step": 28760
    },
    {
      "epoch": 2.829534557793022,
      "grad_norm": 0.8939152359962463,
      "learning_rate": 8.768827246353773e-06,
      "loss": 0.2051,
      "step": 28770
    },
    {
      "epoch": 2.8305180595510313,
      "grad_norm": 1.9083774089813232,
      "learning_rate": 8.764853157413665e-06,
      "loss": 0.3075,
      "step": 28780
    },
    {
      "epoch": 2.831501561309041,
      "grad_norm": 1.497853398323059,
      "learning_rate": 8.760879068473554e-06,
      "loss": 0.1993,
      "step": 28790
    },
    {
      "epoch": 2.8324850630670504,
      "grad_norm": 0.5038604736328125,
      "learning_rate": 8.756904979533442e-06,
      "loss": 0.1835,
      "step": 28800
    },
    {
      "epoch": 2.8334685648250595,
      "grad_norm": 0.49522608518600464,
      "learning_rate": 8.752930890593332e-06,
      "loss": 0.2387,
      "step": 28810
    },
    {
      "epoch": 2.834452066583069,
      "grad_norm": 0.2055412083864212,
      "learning_rate": 8.748956801653222e-06,
      "loss": 0.1647,
      "step": 28820
    },
    {
      "epoch": 2.8354355683410786,
      "grad_norm": 0.1830047070980072,
      "learning_rate": 8.744982712713111e-06,
      "loss": 0.1638,
      "step": 28830
    },
    {
      "epoch": 2.8364190700990877,
      "grad_norm": 2.257323741912842,
      "learning_rate": 8.741008623773001e-06,
      "loss": 0.1678,
      "step": 28840
    },
    {
      "epoch": 2.837402571857097,
      "grad_norm": 1.6618006229400635,
      "learning_rate": 8.73703453483289e-06,
      "loss": 0.2156,
      "step": 28850
    },
    {
      "epoch": 2.8383860736151068,
      "grad_norm": 1.9732855558395386,
      "learning_rate": 8.73306044589278e-06,
      "loss": 0.1709,
      "step": 28860
    },
    {
      "epoch": 2.839369575373116,
      "grad_norm": 0.7155810594558716,
      "learning_rate": 8.72908635695267e-06,
      "loss": 0.1784,
      "step": 28870
    },
    {
      "epoch": 2.8403530771311254,
      "grad_norm": 0.24890826642513275,
      "learning_rate": 8.72511226801256e-06,
      "loss": 0.2786,
      "step": 28880
    },
    {
      "epoch": 2.841336578889135,
      "grad_norm": 1.0565311908721924,
      "learning_rate": 8.721138179072448e-06,
      "loss": 0.2135,
      "step": 28890
    },
    {
      "epoch": 2.842320080647144,
      "grad_norm": 0.3702998161315918,
      "learning_rate": 8.717164090132337e-06,
      "loss": 0.2111,
      "step": 28900
    },
    {
      "epoch": 2.8433035824051536,
      "grad_norm": 0.5531200766563416,
      "learning_rate": 8.713190001192229e-06,
      "loss": 0.2772,
      "step": 28910
    },
    {
      "epoch": 2.844287084163163,
      "grad_norm": 1.44815194606781,
      "learning_rate": 8.709215912252116e-06,
      "loss": 0.2009,
      "step": 28920
    },
    {
      "epoch": 2.845270585921172,
      "grad_norm": 0.26301082968711853,
      "learning_rate": 8.705241823312006e-06,
      "loss": 0.2195,
      "step": 28930
    },
    {
      "epoch": 2.8462540876791818,
      "grad_norm": 5.985122203826904,
      "learning_rate": 8.701267734371896e-06,
      "loss": 0.2069,
      "step": 28940
    },
    {
      "epoch": 2.8472375894371913,
      "grad_norm": 0.3828321099281311,
      "learning_rate": 8.697293645431785e-06,
      "loss": 0.1972,
      "step": 28950
    },
    {
      "epoch": 2.8482210911952004,
      "grad_norm": 1.7138978242874146,
      "learning_rate": 8.693319556491675e-06,
      "loss": 0.1522,
      "step": 28960
    },
    {
      "epoch": 2.84920459295321,
      "grad_norm": 1.8047984838485718,
      "learning_rate": 8.689345467551565e-06,
      "loss": 0.1605,
      "step": 28970
    },
    {
      "epoch": 2.8501880947112195,
      "grad_norm": 1.2494192123413086,
      "learning_rate": 8.685371378611453e-06,
      "loss": 0.2755,
      "step": 28980
    },
    {
      "epoch": 2.8511715964692286,
      "grad_norm": 0.6676988005638123,
      "learning_rate": 8.681397289671344e-06,
      "loss": 0.2336,
      "step": 28990
    },
    {
      "epoch": 2.852155098227238,
      "grad_norm": 1.2020059823989868,
      "learning_rate": 8.677423200731234e-06,
      "loss": 0.2366,
      "step": 29000
    },
    {
      "epoch": 2.852155098227238,
      "eval_loss": 0.12494421750307083,
      "eval_runtime": 17.4081,
      "eval_samples_per_second": 2.872,
      "eval_steps_per_second": 1.436,
      "step": 29000
    },
    {
      "epoch": 2.8531385999852477,
      "grad_norm": 1.4287397861480713,
      "learning_rate": 8.673449111791122e-06,
      "loss": 0.2116,
      "step": 29010
    },
    {
      "epoch": 2.8541221017432568,
      "grad_norm": 0.8067907691001892,
      "learning_rate": 8.669475022851011e-06,
      "loss": 0.1835,
      "step": 29020
    },
    {
      "epoch": 2.8551056035012663,
      "grad_norm": 0.2560260593891144,
      "learning_rate": 8.665500933910903e-06,
      "loss": 0.2746,
      "step": 29030
    },
    {
      "epoch": 2.856089105259276,
      "grad_norm": 1.1014164686203003,
      "learning_rate": 8.66152684497079e-06,
      "loss": 0.1783,
      "step": 29040
    },
    {
      "epoch": 2.857072607017285,
      "grad_norm": 1.2261013984680176,
      "learning_rate": 8.65755275603068e-06,
      "loss": 0.2009,
      "step": 29050
    },
    {
      "epoch": 2.8580561087752945,
      "grad_norm": 0.8522171974182129,
      "learning_rate": 8.65357866709057e-06,
      "loss": 0.2382,
      "step": 29060
    },
    {
      "epoch": 2.859039610533304,
      "grad_norm": 1.2073774337768555,
      "learning_rate": 8.64960457815046e-06,
      "loss": 0.1162,
      "step": 29070
    },
    {
      "epoch": 2.860023112291313,
      "grad_norm": 3.7264292240142822,
      "learning_rate": 8.64563048921035e-06,
      "loss": 0.2941,
      "step": 29080
    },
    {
      "epoch": 2.8610066140493227,
      "grad_norm": 0.8291231989860535,
      "learning_rate": 8.641656400270239e-06,
      "loss": 0.24,
      "step": 29090
    },
    {
      "epoch": 2.861990115807332,
      "grad_norm": 0.7799243927001953,
      "learning_rate": 8.637682311330127e-06,
      "loss": 0.2807,
      "step": 29100
    },
    {
      "epoch": 2.8629736175653413,
      "grad_norm": 1.201129674911499,
      "learning_rate": 8.633708222390018e-06,
      "loss": 0.2662,
      "step": 29110
    },
    {
      "epoch": 2.863957119323351,
      "grad_norm": 1.764297604560852,
      "learning_rate": 8.629734133449908e-06,
      "loss": 0.1968,
      "step": 29120
    },
    {
      "epoch": 2.8649406210813604,
      "grad_norm": 0.7513929009437561,
      "learning_rate": 8.625760044509796e-06,
      "loss": 0.149,
      "step": 29130
    },
    {
      "epoch": 2.8659241228393695,
      "grad_norm": 0.11042553931474686,
      "learning_rate": 8.621785955569686e-06,
      "loss": 0.1718,
      "step": 29140
    },
    {
      "epoch": 2.866907624597379,
      "grad_norm": 0.49265816807746887,
      "learning_rate": 8.617811866629577e-06,
      "loss": 0.2166,
      "step": 29150
    },
    {
      "epoch": 2.8678911263553886,
      "grad_norm": 0.6122161746025085,
      "learning_rate": 8.613837777689465e-06,
      "loss": 0.1956,
      "step": 29160
    },
    {
      "epoch": 2.8688746281133977,
      "grad_norm": 0.8903423547744751,
      "learning_rate": 8.609863688749355e-06,
      "loss": 0.2179,
      "step": 29170
    },
    {
      "epoch": 2.869858129871407,
      "grad_norm": 1.8324393033981323,
      "learning_rate": 8.605889599809244e-06,
      "loss": 0.2135,
      "step": 29180
    },
    {
      "epoch": 2.8708416316294167,
      "grad_norm": 0.9282931685447693,
      "learning_rate": 8.601915510869134e-06,
      "loss": 0.1615,
      "step": 29190
    },
    {
      "epoch": 2.871825133387426,
      "grad_norm": 0.577796220779419,
      "learning_rate": 8.597941421929024e-06,
      "loss": 0.2672,
      "step": 29200
    },
    {
      "epoch": 2.8728086351454354,
      "grad_norm": 0.5581834316253662,
      "learning_rate": 8.593967332988913e-06,
      "loss": 0.1615,
      "step": 29210
    },
    {
      "epoch": 2.873792136903445,
      "grad_norm": 1.03250253200531,
      "learning_rate": 8.589993244048801e-06,
      "loss": 0.1836,
      "step": 29220
    },
    {
      "epoch": 2.874775638661454,
      "grad_norm": 0.719618558883667,
      "learning_rate": 8.586019155108693e-06,
      "loss": 0.3187,
      "step": 29230
    },
    {
      "epoch": 2.8757591404194636,
      "grad_norm": 0.5816129446029663,
      "learning_rate": 8.582045066168582e-06,
      "loss": 0.1661,
      "step": 29240
    },
    {
      "epoch": 2.876742642177473,
      "grad_norm": 1.870931625366211,
      "learning_rate": 8.57807097722847e-06,
      "loss": 0.1292,
      "step": 29250
    },
    {
      "epoch": 2.877726143935482,
      "grad_norm": 0.6622405052185059,
      "learning_rate": 8.57409688828836e-06,
      "loss": 0.3036,
      "step": 29260
    },
    {
      "epoch": 2.8787096456934917,
      "grad_norm": 1.67839777469635,
      "learning_rate": 8.570122799348251e-06,
      "loss": 0.1944,
      "step": 29270
    },
    {
      "epoch": 2.8796931474515013,
      "grad_norm": 0.7301672697067261,
      "learning_rate": 8.56614871040814e-06,
      "loss": 0.3032,
      "step": 29280
    },
    {
      "epoch": 2.8806766492095104,
      "grad_norm": 1.228724479675293,
      "learning_rate": 8.562174621468029e-06,
      "loss": 0.2329,
      "step": 29290
    },
    {
      "epoch": 2.88166015096752,
      "grad_norm": 0.5540966987609863,
      "learning_rate": 8.558200532527919e-06,
      "loss": 0.2967,
      "step": 29300
    },
    {
      "epoch": 2.8826436527255295,
      "grad_norm": 0.32900357246398926,
      "learning_rate": 8.554226443587808e-06,
      "loss": 0.2202,
      "step": 29310
    },
    {
      "epoch": 2.8836271544835386,
      "grad_norm": 0.3154147267341614,
      "learning_rate": 8.550252354647698e-06,
      "loss": 0.2517,
      "step": 29320
    },
    {
      "epoch": 2.884610656241548,
      "grad_norm": 0.9213051795959473,
      "learning_rate": 8.546278265707588e-06,
      "loss": 0.2602,
      "step": 29330
    },
    {
      "epoch": 2.8855941579995577,
      "grad_norm": 0.6817721724510193,
      "learning_rate": 8.542304176767476e-06,
      "loss": 0.2992,
      "step": 29340
    },
    {
      "epoch": 2.8865776597575667,
      "grad_norm": 1.0806705951690674,
      "learning_rate": 8.538330087827367e-06,
      "loss": 0.1915,
      "step": 29350
    },
    {
      "epoch": 2.8875611615155763,
      "grad_norm": 0.8594111800193787,
      "learning_rate": 8.534355998887257e-06,
      "loss": 0.2021,
      "step": 29360
    },
    {
      "epoch": 2.888544663273586,
      "grad_norm": 0.2863863408565521,
      "learning_rate": 8.530381909947145e-06,
      "loss": 0.225,
      "step": 29370
    },
    {
      "epoch": 2.889528165031595,
      "grad_norm": 1.9358649253845215,
      "learning_rate": 8.526407821007034e-06,
      "loss": 0.2276,
      "step": 29380
    },
    {
      "epoch": 2.8905116667896045,
      "grad_norm": 0.4354264438152313,
      "learning_rate": 8.522433732066926e-06,
      "loss": 0.1729,
      "step": 29390
    },
    {
      "epoch": 2.891495168547614,
      "grad_norm": 2.1319658756256104,
      "learning_rate": 8.518459643126814e-06,
      "loss": 0.1549,
      "step": 29400
    },
    {
      "epoch": 2.892478670305623,
      "grad_norm": 3.555476188659668,
      "learning_rate": 8.514485554186703e-06,
      "loss": 0.1514,
      "step": 29410
    },
    {
      "epoch": 2.8934621720636327,
      "grad_norm": 0.7322472333908081,
      "learning_rate": 8.510511465246593e-06,
      "loss": 0.182,
      "step": 29420
    },
    {
      "epoch": 2.894445673821642,
      "grad_norm": 1.2825742959976196,
      "learning_rate": 8.506537376306483e-06,
      "loss": 0.2376,
      "step": 29430
    },
    {
      "epoch": 2.8954291755796513,
      "grad_norm": 1.2198630571365356,
      "learning_rate": 8.502563287366372e-06,
      "loss": 0.2935,
      "step": 29440
    },
    {
      "epoch": 2.896412677337661,
      "grad_norm": 0.12802644073963165,
      "learning_rate": 8.498589198426262e-06,
      "loss": 0.154,
      "step": 29450
    },
    {
      "epoch": 2.8973961790956704,
      "grad_norm": 1.0437980890274048,
      "learning_rate": 8.49461510948615e-06,
      "loss": 0.2054,
      "step": 29460
    },
    {
      "epoch": 2.8983796808536795,
      "grad_norm": 2.8305885791778564,
      "learning_rate": 8.490641020546041e-06,
      "loss": 0.2329,
      "step": 29470
    },
    {
      "epoch": 2.899363182611689,
      "grad_norm": 0.8940551280975342,
      "learning_rate": 8.486666931605931e-06,
      "loss": 0.1698,
      "step": 29480
    },
    {
      "epoch": 2.9003466843696986,
      "grad_norm": 1.2706600427627563,
      "learning_rate": 8.482692842665819e-06,
      "loss": 0.1952,
      "step": 29490
    },
    {
      "epoch": 2.9013301861277077,
      "grad_norm": 1.8138004541397095,
      "learning_rate": 8.478718753725709e-06,
      "loss": 0.2928,
      "step": 29500
    },
    {
      "epoch": 2.9013301861277077,
      "eval_loss": 0.12123703211545944,
      "eval_runtime": 17.8571,
      "eval_samples_per_second": 2.8,
      "eval_steps_per_second": 1.4,
      "step": 29500
    },
    {
      "epoch": 2.902313687885717,
      "grad_norm": 1.4499008655548096,
      "learning_rate": 8.4747446647856e-06,
      "loss": 0.211,
      "step": 29510
    },
    {
      "epoch": 2.9032971896437267,
      "grad_norm": 2.5677297115325928,
      "learning_rate": 8.470770575845488e-06,
      "loss": 0.2401,
      "step": 29520
    },
    {
      "epoch": 2.904280691401736,
      "grad_norm": 1.0985254049301147,
      "learning_rate": 8.466796486905378e-06,
      "loss": 0.1441,
      "step": 29530
    },
    {
      "epoch": 2.9052641931597454,
      "grad_norm": 1.7599356174468994,
      "learning_rate": 8.462822397965267e-06,
      "loss": 0.164,
      "step": 29540
    },
    {
      "epoch": 2.906247694917755,
      "grad_norm": 0.7811633944511414,
      "learning_rate": 8.458848309025157e-06,
      "loss": 0.2528,
      "step": 29550
    },
    {
      "epoch": 2.907231196675764,
      "grad_norm": 0.5994932055473328,
      "learning_rate": 8.454874220085047e-06,
      "loss": 0.3667,
      "step": 29560
    },
    {
      "epoch": 2.9082146984337736,
      "grad_norm": 0.3563966453075409,
      "learning_rate": 8.450900131144936e-06,
      "loss": 0.1834,
      "step": 29570
    },
    {
      "epoch": 2.909198200191783,
      "grad_norm": 3.5165040493011475,
      "learning_rate": 8.446926042204824e-06,
      "loss": 0.3832,
      "step": 29580
    },
    {
      "epoch": 2.910181701949792,
      "grad_norm": 2.4149057865142822,
      "learning_rate": 8.442951953264716e-06,
      "loss": 0.2664,
      "step": 29590
    },
    {
      "epoch": 2.9111652037078017,
      "grad_norm": 1.1386704444885254,
      "learning_rate": 8.438977864324605e-06,
      "loss": 0.1699,
      "step": 29600
    },
    {
      "epoch": 2.9121487054658113,
      "grad_norm": 3.486772298812866,
      "learning_rate": 8.435003775384493e-06,
      "loss": 0.2228,
      "step": 29610
    },
    {
      "epoch": 2.9131322072238204,
      "grad_norm": 0.9155600666999817,
      "learning_rate": 8.431029686444383e-06,
      "loss": 0.2772,
      "step": 29620
    },
    {
      "epoch": 2.91411570898183,
      "grad_norm": 0.644264817237854,
      "learning_rate": 8.427055597504274e-06,
      "loss": 0.2006,
      "step": 29630
    },
    {
      "epoch": 2.9150992107398395,
      "grad_norm": 0.813632607460022,
      "learning_rate": 8.423081508564162e-06,
      "loss": 0.2644,
      "step": 29640
    },
    {
      "epoch": 2.9160827124978486,
      "grad_norm": 1.260005235671997,
      "learning_rate": 8.419107419624052e-06,
      "loss": 0.132,
      "step": 29650
    },
    {
      "epoch": 2.917066214255858,
      "grad_norm": 2.324798107147217,
      "learning_rate": 8.415133330683942e-06,
      "loss": 0.2186,
      "step": 29660
    },
    {
      "epoch": 2.9180497160138676,
      "grad_norm": 0.573977530002594,
      "learning_rate": 8.411159241743831e-06,
      "loss": 0.184,
      "step": 29670
    },
    {
      "epoch": 2.9190332177718767,
      "grad_norm": 0.9063766598701477,
      "learning_rate": 8.407185152803721e-06,
      "loss": 0.1659,
      "step": 29680
    },
    {
      "epoch": 2.9200167195298863,
      "grad_norm": 0.5683373808860779,
      "learning_rate": 8.40321106386361e-06,
      "loss": 0.281,
      "step": 29690
    },
    {
      "epoch": 2.921000221287896,
      "grad_norm": 2.299201011657715,
      "learning_rate": 8.399236974923498e-06,
      "loss": 0.3064,
      "step": 29700
    },
    {
      "epoch": 2.921983723045905,
      "grad_norm": 1.1093686819076538,
      "learning_rate": 8.395262885983388e-06,
      "loss": 0.2682,
      "step": 29710
    },
    {
      "epoch": 2.9229672248039145,
      "grad_norm": 0.9980324506759644,
      "learning_rate": 8.39128879704328e-06,
      "loss": 0.1977,
      "step": 29720
    },
    {
      "epoch": 2.9239507265619236,
      "grad_norm": 0.5987958312034607,
      "learning_rate": 8.387314708103167e-06,
      "loss": 0.0887,
      "step": 29730
    },
    {
      "epoch": 2.924934228319933,
      "grad_norm": 1.2359176874160767,
      "learning_rate": 8.383340619163057e-06,
      "loss": 0.2705,
      "step": 29740
    },
    {
      "epoch": 2.9259177300779426,
      "grad_norm": 0.11103789508342743,
      "learning_rate": 8.379366530222947e-06,
      "loss": 0.1573,
      "step": 29750
    },
    {
      "epoch": 2.9269012318359517,
      "grad_norm": 1.2775187492370605,
      "learning_rate": 8.375392441282836e-06,
      "loss": 0.1728,
      "step": 29760
    },
    {
      "epoch": 2.9278847335939613,
      "grad_norm": 0.3069576323032379,
      "learning_rate": 8.371418352342726e-06,
      "loss": 0.2152,
      "step": 29770
    },
    {
      "epoch": 2.928868235351971,
      "grad_norm": 0.18963612616062164,
      "learning_rate": 8.367444263402616e-06,
      "loss": 0.1034,
      "step": 29780
    },
    {
      "epoch": 2.92985173710998,
      "grad_norm": 4.958619594573975,
      "learning_rate": 8.363470174462505e-06,
      "loss": 0.2122,
      "step": 29790
    },
    {
      "epoch": 2.9308352388679895,
      "grad_norm": 1.2674598693847656,
      "learning_rate": 8.359496085522395e-06,
      "loss": 0.1352,
      "step": 29800
    },
    {
      "epoch": 2.931818740625999,
      "grad_norm": 5.292276382446289,
      "learning_rate": 8.355521996582285e-06,
      "loss": 0.1786,
      "step": 29810
    },
    {
      "epoch": 2.932802242384008,
      "grad_norm": 5.913046360015869,
      "learning_rate": 8.351547907642173e-06,
      "loss": 0.2726,
      "step": 29820
    },
    {
      "epoch": 2.9337857441420176,
      "grad_norm": 1.4404199123382568,
      "learning_rate": 8.347573818702062e-06,
      "loss": 0.1817,
      "step": 29830
    },
    {
      "epoch": 2.934769245900027,
      "grad_norm": 0.4281212389469147,
      "learning_rate": 8.343599729761954e-06,
      "loss": 0.3099,
      "step": 29840
    },
    {
      "epoch": 2.9357527476580363,
      "grad_norm": 0.5780307054519653,
      "learning_rate": 8.339625640821842e-06,
      "loss": 0.1779,
      "step": 29850
    },
    {
      "epoch": 2.936736249416046,
      "grad_norm": 0.4119488000869751,
      "learning_rate": 8.335651551881731e-06,
      "loss": 0.2055,
      "step": 29860
    },
    {
      "epoch": 2.9377197511740554,
      "grad_norm": 2.0947678089141846,
      "learning_rate": 8.331677462941621e-06,
      "loss": 0.2263,
      "step": 29870
    },
    {
      "epoch": 2.9387032529320645,
      "grad_norm": 3.156257152557373,
      "learning_rate": 8.32770337400151e-06,
      "loss": 0.2569,
      "step": 29880
    },
    {
      "epoch": 2.939686754690074,
      "grad_norm": 1.3036227226257324,
      "learning_rate": 8.3237292850614e-06,
      "loss": 0.1654,
      "step": 29890
    },
    {
      "epoch": 2.9406702564480836,
      "grad_norm": 1.7258992195129395,
      "learning_rate": 8.31975519612129e-06,
      "loss": 0.2657,
      "step": 29900
    },
    {
      "epoch": 2.9416537582060927,
      "grad_norm": 0.7538513541221619,
      "learning_rate": 8.31578110718118e-06,
      "loss": 0.1335,
      "step": 29910
    },
    {
      "epoch": 2.942637259964102,
      "grad_norm": 0.5512486696243286,
      "learning_rate": 8.31180701824107e-06,
      "loss": 0.2118,
      "step": 29920
    },
    {
      "epoch": 2.9436207617221117,
      "grad_norm": 1.1244486570358276,
      "learning_rate": 8.307832929300959e-06,
      "loss": 0.1494,
      "step": 29930
    },
    {
      "epoch": 2.944604263480121,
      "grad_norm": 2.096811056137085,
      "learning_rate": 8.303858840360847e-06,
      "loss": 0.1748,
      "step": 29940
    },
    {
      "epoch": 2.9455877652381304,
      "grad_norm": 0.6595863103866577,
      "learning_rate": 8.299884751420737e-06,
      "loss": 0.2772,
      "step": 29950
    },
    {
      "epoch": 2.94657126699614,
      "grad_norm": 0.380372554063797,
      "learning_rate": 8.295910662480628e-06,
      "loss": 0.2749,
      "step": 29960
    },
    {
      "epoch": 2.947554768754149,
      "grad_norm": 1.3502016067504883,
      "learning_rate": 8.291936573540516e-06,
      "loss": 0.1691,
      "step": 29970
    },
    {
      "epoch": 2.9485382705121586,
      "grad_norm": 0.8366619348526001,
      "learning_rate": 8.287962484600406e-06,
      "loss": 0.1547,
      "step": 29980
    },
    {
      "epoch": 2.949521772270168,
      "grad_norm": 0.2709037959575653,
      "learning_rate": 8.283988395660295e-06,
      "loss": 0.156,
      "step": 29990
    },
    {
      "epoch": 2.950505274028177,
      "grad_norm": 0.7497245073318481,
      "learning_rate": 8.280014306720185e-06,
      "loss": 0.2401,
      "step": 30000
    },
    {
      "epoch": 2.950505274028177,
      "eval_loss": 0.12118250876665115,
      "eval_runtime": 20.871,
      "eval_samples_per_second": 2.396,
      "eval_steps_per_second": 1.198,
      "step": 30000
    },
    {
      "epoch": 2.9514887757861867,
      "grad_norm": 0.6676973700523376,
      "learning_rate": 8.276040217780075e-06,
      "loss": 0.2575,
      "step": 30010
    },
    {
      "epoch": 2.952472277544196,
      "grad_norm": 1.6974183320999146,
      "learning_rate": 8.272066128839964e-06,
      "loss": 0.2304,
      "step": 30020
    },
    {
      "epoch": 2.9534557793022054,
      "grad_norm": 1.2901015281677246,
      "learning_rate": 8.268092039899854e-06,
      "loss": 0.1907,
      "step": 30030
    },
    {
      "epoch": 2.954439281060215,
      "grad_norm": 2.2742185592651367,
      "learning_rate": 8.264117950959744e-06,
      "loss": 0.1403,
      "step": 30040
    },
    {
      "epoch": 2.955422782818224,
      "grad_norm": 0.535651445388794,
      "learning_rate": 8.260143862019633e-06,
      "loss": 0.2052,
      "step": 30050
    },
    {
      "epoch": 2.9564062845762336,
      "grad_norm": 1.6494063138961792,
      "learning_rate": 8.256169773079521e-06,
      "loss": 0.1801,
      "step": 30060
    },
    {
      "epoch": 2.957389786334243,
      "grad_norm": 2.04034161567688,
      "learning_rate": 8.252195684139411e-06,
      "loss": 0.2169,
      "step": 30070
    },
    {
      "epoch": 2.958373288092252,
      "grad_norm": 0.3735710680484772,
      "learning_rate": 8.248221595199302e-06,
      "loss": 0.1573,
      "step": 30080
    },
    {
      "epoch": 2.9593567898502617,
      "grad_norm": 0.9145519137382507,
      "learning_rate": 8.24424750625919e-06,
      "loss": 0.2631,
      "step": 30090
    },
    {
      "epoch": 2.9603402916082713,
      "grad_norm": 0.36900460720062256,
      "learning_rate": 8.24027341731908e-06,
      "loss": 0.1587,
      "step": 30100
    },
    {
      "epoch": 2.9613237933662804,
      "grad_norm": 1.695147156715393,
      "learning_rate": 8.23629932837897e-06,
      "loss": 0.2926,
      "step": 30110
    },
    {
      "epoch": 2.96230729512429,
      "grad_norm": 1.1021901369094849,
      "learning_rate": 8.23232523943886e-06,
      "loss": 0.1658,
      "step": 30120
    },
    {
      "epoch": 2.9632907968822995,
      "grad_norm": 2.1732630729675293,
      "learning_rate": 8.228351150498749e-06,
      "loss": 0.1942,
      "step": 30130
    },
    {
      "epoch": 2.9642742986403086,
      "grad_norm": 2.214512586593628,
      "learning_rate": 8.224377061558639e-06,
      "loss": 0.1876,
      "step": 30140
    },
    {
      "epoch": 2.965257800398318,
      "grad_norm": 0.7325847148895264,
      "learning_rate": 8.220402972618528e-06,
      "loss": 0.1788,
      "step": 30150
    },
    {
      "epoch": 2.9662413021563276,
      "grad_norm": 0.7839856743812561,
      "learning_rate": 8.216428883678418e-06,
      "loss": 0.195,
      "step": 30160
    },
    {
      "epoch": 2.9672248039143367,
      "grad_norm": 1.3128502368927002,
      "learning_rate": 8.212454794738308e-06,
      "loss": 0.3257,
      "step": 30170
    },
    {
      "epoch": 2.9682083056723463,
      "grad_norm": 0.9301101565361023,
      "learning_rate": 8.208480705798196e-06,
      "loss": 0.3217,
      "step": 30180
    },
    {
      "epoch": 2.969191807430356,
      "grad_norm": 1.5134687423706055,
      "learning_rate": 8.204506616858085e-06,
      "loss": 0.2426,
      "step": 30190
    },
    {
      "epoch": 2.970175309188365,
      "grad_norm": 0.48188215494155884,
      "learning_rate": 8.200532527917977e-06,
      "loss": 0.2246,
      "step": 30200
    },
    {
      "epoch": 2.9711588109463745,
      "grad_norm": 1.5620793104171753,
      "learning_rate": 8.196558438977865e-06,
      "loss": 0.2173,
      "step": 30210
    },
    {
      "epoch": 2.972142312704384,
      "grad_norm": 0.5938036441802979,
      "learning_rate": 8.192584350037754e-06,
      "loss": 0.1634,
      "step": 30220
    },
    {
      "epoch": 2.973125814462393,
      "grad_norm": 1.9721848964691162,
      "learning_rate": 8.188610261097644e-06,
      "loss": 0.1698,
      "step": 30230
    },
    {
      "epoch": 2.9741093162204026,
      "grad_norm": 1.3379566669464111,
      "learning_rate": 8.184636172157534e-06,
      "loss": 0.2143,
      "step": 30240
    },
    {
      "epoch": 2.975092817978412,
      "grad_norm": 2.263864517211914,
      "learning_rate": 8.180662083217423e-06,
      "loss": 0.1574,
      "step": 30250
    },
    {
      "epoch": 2.9760763197364213,
      "grad_norm": 1.6023887395858765,
      "learning_rate": 8.176687994277313e-06,
      "loss": 0.1478,
      "step": 30260
    },
    {
      "epoch": 2.977059821494431,
      "grad_norm": 0.4679398536682129,
      "learning_rate": 8.172713905337203e-06,
      "loss": 0.1792,
      "step": 30270
    },
    {
      "epoch": 2.9780433232524404,
      "grad_norm": 0.6919494271278381,
      "learning_rate": 8.168739816397092e-06,
      "loss": 0.2046,
      "step": 30280
    },
    {
      "epoch": 2.9790268250104495,
      "grad_norm": 0.637020468711853,
      "learning_rate": 8.164765727456982e-06,
      "loss": 0.1326,
      "step": 30290
    },
    {
      "epoch": 2.980010326768459,
      "grad_norm": 0.6778492331504822,
      "learning_rate": 8.16079163851687e-06,
      "loss": 0.1781,
      "step": 30300
    },
    {
      "epoch": 2.9809938285264685,
      "grad_norm": 1.518211007118225,
      "learning_rate": 8.15681754957676e-06,
      "loss": 0.2089,
      "step": 30310
    },
    {
      "epoch": 2.9819773302844776,
      "grad_norm": 0.9797000885009766,
      "learning_rate": 8.152843460636651e-06,
      "loss": 0.2146,
      "step": 30320
    },
    {
      "epoch": 2.982960832042487,
      "grad_norm": 1.643019199371338,
      "learning_rate": 8.148869371696539e-06,
      "loss": 0.3223,
      "step": 30330
    },
    {
      "epoch": 2.9839443338004967,
      "grad_norm": 1.33889639377594,
      "learning_rate": 8.144895282756429e-06,
      "loss": 0.1224,
      "step": 30340
    },
    {
      "epoch": 2.984927835558506,
      "grad_norm": 1.031048059463501,
      "learning_rate": 8.140921193816318e-06,
      "loss": 0.2173,
      "step": 30350
    },
    {
      "epoch": 2.9859113373165154,
      "grad_norm": 0.3998883068561554,
      "learning_rate": 8.136947104876208e-06,
      "loss": 0.1328,
      "step": 30360
    },
    {
      "epoch": 2.986894839074525,
      "grad_norm": 0.7423034906387329,
      "learning_rate": 8.132973015936098e-06,
      "loss": 0.2294,
      "step": 30370
    },
    {
      "epoch": 2.987878340832534,
      "grad_norm": 1.018033504486084,
      "learning_rate": 8.128998926995987e-06,
      "loss": 0.2343,
      "step": 30380
    },
    {
      "epoch": 2.9888618425905435,
      "grad_norm": 0.8161916732788086,
      "learning_rate": 8.125024838055877e-06,
      "loss": 0.3269,
      "step": 30390
    },
    {
      "epoch": 2.989845344348553,
      "grad_norm": 0.3255046308040619,
      "learning_rate": 8.121050749115767e-06,
      "loss": 0.2651,
      "step": 30400
    },
    {
      "epoch": 2.990828846106562,
      "grad_norm": 0.3091247081756592,
      "learning_rate": 8.117076660175656e-06,
      "loss": 0.1769,
      "step": 30410
    },
    {
      "epoch": 2.9918123478645717,
      "grad_norm": 0.1740453690290451,
      "learning_rate": 8.113102571235544e-06,
      "loss": 0.1343,
      "step": 30420
    },
    {
      "epoch": 2.9927958496225813,
      "grad_norm": 0.5746241807937622,
      "learning_rate": 8.109128482295434e-06,
      "loss": 0.2753,
      "step": 30430
    },
    {
      "epoch": 2.9937793513805904,
      "grad_norm": 0.605728805065155,
      "learning_rate": 8.105154393355325e-06,
      "loss": 0.2022,
      "step": 30440
    },
    {
      "epoch": 2.9947628531386,
      "grad_norm": 0.6295777559280396,
      "learning_rate": 8.101180304415213e-06,
      "loss": 0.1666,
      "step": 30450
    },
    {
      "epoch": 2.9957463548966095,
      "grad_norm": 0.697580873966217,
      "learning_rate": 8.097206215475103e-06,
      "loss": 0.1339,
      "step": 30460
    },
    {
      "epoch": 2.9967298566546186,
      "grad_norm": 0.7362121939659119,
      "learning_rate": 8.093232126534993e-06,
      "loss": 0.2896,
      "step": 30470
    },
    {
      "epoch": 2.997713358412628,
      "grad_norm": 0.18402792513370514,
      "learning_rate": 8.089258037594882e-06,
      "loss": 0.1232,
      "step": 30480
    },
    {
      "epoch": 2.9986968601706376,
      "grad_norm": 1.2359050512313843,
      "learning_rate": 8.085283948654772e-06,
      "loss": 0.2461,
      "step": 30490
    },
    {
      "epoch": 2.9996803619286467,
      "grad_norm": 0.38628000020980835,
      "learning_rate": 8.081309859714662e-06,
      "loss": 0.1896,
      "step": 30500
    },
    {
      "epoch": 2.9996803619286467,
      "eval_loss": 0.1228804811835289,
      "eval_runtime": 17.7254,
      "eval_samples_per_second": 2.821,
      "eval_steps_per_second": 1.41,
      "step": 30500
    },
    {
      "epoch": 3.0006638636866563,
      "grad_norm": 1.5436768531799316,
      "learning_rate": 8.077335770774551e-06,
      "loss": 0.235,
      "step": 30510
    },
    {
      "epoch": 3.001647365444666,
      "grad_norm": 0.49330833554267883,
      "learning_rate": 8.073361681834439e-06,
      "loss": 0.1425,
      "step": 30520
    },
    {
      "epoch": 3.002630867202675,
      "grad_norm": 0.7838341593742371,
      "learning_rate": 8.06938759289433e-06,
      "loss": 0.1806,
      "step": 30530
    },
    {
      "epoch": 3.0036143689606845,
      "grad_norm": 0.5121620893478394,
      "learning_rate": 8.065413503954218e-06,
      "loss": 0.1986,
      "step": 30540
    },
    {
      "epoch": 3.004597870718694,
      "grad_norm": 1.0584651231765747,
      "learning_rate": 8.061439415014108e-06,
      "loss": 0.2191,
      "step": 30550
    },
    {
      "epoch": 3.005581372476703,
      "grad_norm": 2.7569994926452637,
      "learning_rate": 8.057465326073998e-06,
      "loss": 0.1805,
      "step": 30560
    },
    {
      "epoch": 3.0065648742347126,
      "grad_norm": 1.0442713499069214,
      "learning_rate": 8.053491237133887e-06,
      "loss": 0.3119,
      "step": 30570
    },
    {
      "epoch": 3.007548375992722,
      "grad_norm": 0.8963764309883118,
      "learning_rate": 8.049517148193777e-06,
      "loss": 0.1806,
      "step": 30580
    },
    {
      "epoch": 3.0085318777507313,
      "grad_norm": 1.6401909589767456,
      "learning_rate": 8.045543059253667e-06,
      "loss": 0.2019,
      "step": 30590
    },
    {
      "epoch": 3.009515379508741,
      "grad_norm": 0.45249977707862854,
      "learning_rate": 8.041568970313556e-06,
      "loss": 0.1886,
      "step": 30600
    },
    {
      "epoch": 3.0104988812667504,
      "grad_norm": 0.792337954044342,
      "learning_rate": 8.037594881373446e-06,
      "loss": 0.2098,
      "step": 30610
    },
    {
      "epoch": 3.0114823830247595,
      "grad_norm": 0.5009030103683472,
      "learning_rate": 8.033620792433336e-06,
      "loss": 0.2774,
      "step": 30620
    },
    {
      "epoch": 3.012465884782769,
      "grad_norm": 1.3158507347106934,
      "learning_rate": 8.029646703493225e-06,
      "loss": 0.1768,
      "step": 30630
    },
    {
      "epoch": 3.0134493865407785,
      "grad_norm": 0.5353335738182068,
      "learning_rate": 8.025672614553113e-06,
      "loss": 0.3349,
      "step": 30640
    },
    {
      "epoch": 3.0144328882987876,
      "grad_norm": 0.9227167367935181,
      "learning_rate": 8.021698525613005e-06,
      "loss": 0.2027,
      "step": 30650
    },
    {
      "epoch": 3.015416390056797,
      "grad_norm": 1.4296085834503174,
      "learning_rate": 8.017724436672893e-06,
      "loss": 0.1577,
      "step": 30660
    },
    {
      "epoch": 3.0163998918148067,
      "grad_norm": 1.0844863653182983,
      "learning_rate": 8.013750347732782e-06,
      "loss": 0.1652,
      "step": 30670
    },
    {
      "epoch": 3.017383393572816,
      "grad_norm": 0.8566285371780396,
      "learning_rate": 8.009776258792672e-06,
      "loss": 0.2387,
      "step": 30680
    },
    {
      "epoch": 3.0183668953308254,
      "grad_norm": 0.4086707830429077,
      "learning_rate": 8.005802169852562e-06,
      "loss": 0.2151,
      "step": 30690
    },
    {
      "epoch": 3.019350397088835,
      "grad_norm": 0.4113174080848694,
      "learning_rate": 8.001828080912451e-06,
      "loss": 0.1869,
      "step": 30700
    },
    {
      "epoch": 3.020333898846844,
      "grad_norm": 0.35407760739326477,
      "learning_rate": 7.997853991972341e-06,
      "loss": 0.1761,
      "step": 30710
    },
    {
      "epoch": 3.0213174006048535,
      "grad_norm": 0.7007184624671936,
      "learning_rate": 7.99387990303223e-06,
      "loss": 0.2157,
      "step": 30720
    },
    {
      "epoch": 3.022300902362863,
      "grad_norm": 0.623153567314148,
      "learning_rate": 7.98990581409212e-06,
      "loss": 0.2404,
      "step": 30730
    },
    {
      "epoch": 3.023284404120872,
      "grad_norm": 0.4026603698730469,
      "learning_rate": 7.98593172515201e-06,
      "loss": 0.2679,
      "step": 30740
    },
    {
      "epoch": 3.0242679058788817,
      "grad_norm": 0.36850112676620483,
      "learning_rate": 7.9819576362119e-06,
      "loss": 0.1855,
      "step": 30750
    },
    {
      "epoch": 3.0252514076368913,
      "grad_norm": 0.7698309421539307,
      "learning_rate": 7.977983547271788e-06,
      "loss": 0.2256,
      "step": 30760
    },
    {
      "epoch": 3.0262349093949004,
      "grad_norm": 2.0218682289123535,
      "learning_rate": 7.974009458331679e-06,
      "loss": 0.1395,
      "step": 30770
    },
    {
      "epoch": 3.02721841115291,
      "grad_norm": 0.40838736295700073,
      "learning_rate": 7.970035369391567e-06,
      "loss": 0.1798,
      "step": 30780
    },
    {
      "epoch": 3.0282019129109194,
      "grad_norm": 0.21284779906272888,
      "learning_rate": 7.966061280451457e-06,
      "loss": 0.1356,
      "step": 30790
    },
    {
      "epoch": 3.0291854146689285,
      "grad_norm": 0.8340510725975037,
      "learning_rate": 7.962087191511346e-06,
      "loss": 0.287,
      "step": 30800
    },
    {
      "epoch": 3.030168916426938,
      "grad_norm": 0.8605313301086426,
      "learning_rate": 7.958113102571236e-06,
      "loss": 0.1405,
      "step": 30810
    },
    {
      "epoch": 3.0311524181849476,
      "grad_norm": 1.0435175895690918,
      "learning_rate": 7.954139013631126e-06,
      "loss": 0.2611,
      "step": 30820
    },
    {
      "epoch": 3.0321359199429567,
      "grad_norm": 0.8442772626876831,
      "learning_rate": 7.950164924691015e-06,
      "loss": 0.2721,
      "step": 30830
    },
    {
      "epoch": 3.0331194217009663,
      "grad_norm": 1.565529704093933,
      "learning_rate": 7.946190835750905e-06,
      "loss": 0.1798,
      "step": 30840
    },
    {
      "epoch": 3.034102923458976,
      "grad_norm": 0.5709851384162903,
      "learning_rate": 7.942216746810795e-06,
      "loss": 0.1672,
      "step": 30850
    },
    {
      "epoch": 3.035086425216985,
      "grad_norm": 0.4600856304168701,
      "learning_rate": 7.938242657870684e-06,
      "loss": 0.2025,
      "step": 30860
    },
    {
      "epoch": 3.0360699269749944,
      "grad_norm": 0.6264340877532959,
      "learning_rate": 7.934268568930574e-06,
      "loss": 0.1344,
      "step": 30870
    },
    {
      "epoch": 3.037053428733004,
      "grad_norm": 1.8087103366851807,
      "learning_rate": 7.930294479990462e-06,
      "loss": 0.2786,
      "step": 30880
    },
    {
      "epoch": 3.038036930491013,
      "grad_norm": 3.1393840312957764,
      "learning_rate": 7.926320391050353e-06,
      "loss": 0.2822,
      "step": 30890
    },
    {
      "epoch": 3.0390204322490226,
      "grad_norm": 0.3457108438014984,
      "learning_rate": 7.922346302110241e-06,
      "loss": 0.2511,
      "step": 30900
    },
    {
      "epoch": 3.040003934007032,
      "grad_norm": 0.5391051769256592,
      "learning_rate": 7.918372213170131e-06,
      "loss": 0.1501,
      "step": 30910
    },
    {
      "epoch": 3.0409874357650413,
      "grad_norm": 0.30457252264022827,
      "learning_rate": 7.91439812423002e-06,
      "loss": 0.1377,
      "step": 30920
    },
    {
      "epoch": 3.041970937523051,
      "grad_norm": 0.6017821431159973,
      "learning_rate": 7.91042403528991e-06,
      "loss": 0.2246,
      "step": 30930
    },
    {
      "epoch": 3.0429544392810604,
      "grad_norm": 1.6097629070281982,
      "learning_rate": 7.9064499463498e-06,
      "loss": 0.2797,
      "step": 30940
    },
    {
      "epoch": 3.0439379410390695,
      "grad_norm": 1.6000497341156006,
      "learning_rate": 7.90247585740969e-06,
      "loss": 0.2186,
      "step": 30950
    },
    {
      "epoch": 3.044921442797079,
      "grad_norm": 0.5865890383720398,
      "learning_rate": 7.89850176846958e-06,
      "loss": 0.1514,
      "step": 30960
    },
    {
      "epoch": 3.0459049445550885,
      "grad_norm": 1.072675108909607,
      "learning_rate": 7.894527679529469e-06,
      "loss": 0.2242,
      "step": 30970
    },
    {
      "epoch": 3.0468884463130976,
      "grad_norm": 0.5994366407394409,
      "learning_rate": 7.890553590589359e-06,
      "loss": 0.2024,
      "step": 30980
    },
    {
      "epoch": 3.047871948071107,
      "grad_norm": 0.466334730386734,
      "learning_rate": 7.886579501649248e-06,
      "loss": 0.1856,
      "step": 30990
    },
    {
      "epoch": 3.0488554498291167,
      "grad_norm": 3.52630615234375,
      "learning_rate": 7.882605412709136e-06,
      "loss": 0.1959,
      "step": 31000
    },
    {
      "epoch": 3.0488554498291167,
      "eval_loss": 0.11854755133390427,
      "eval_runtime": 16.899,
      "eval_samples_per_second": 2.959,
      "eval_steps_per_second": 1.479,
      "step": 31000
    },
    {
      "epoch": 3.049838951587126,
      "grad_norm": 1.0482897758483887,
      "learning_rate": 7.878631323769028e-06,
      "loss": 0.2095,
      "step": 31010
    },
    {
      "epoch": 3.0508224533451354,
      "grad_norm": 1.1640293598175049,
      "learning_rate": 7.874657234828916e-06,
      "loss": 0.2858,
      "step": 31020
    },
    {
      "epoch": 3.051805955103145,
      "grad_norm": 0.5311436653137207,
      "learning_rate": 7.870683145888805e-06,
      "loss": 0.1531,
      "step": 31030
    },
    {
      "epoch": 3.052789456861154,
      "grad_norm": 0.36933091282844543,
      "learning_rate": 7.866709056948695e-06,
      "loss": 0.2279,
      "step": 31040
    },
    {
      "epoch": 3.0537729586191635,
      "grad_norm": 0.9239271879196167,
      "learning_rate": 7.862734968008585e-06,
      "loss": 0.1999,
      "step": 31050
    },
    {
      "epoch": 3.054756460377173,
      "grad_norm": 0.23407602310180664,
      "learning_rate": 7.858760879068474e-06,
      "loss": 0.2136,
      "step": 31060
    },
    {
      "epoch": 3.055739962135182,
      "grad_norm": 0.6105637550354004,
      "learning_rate": 7.854786790128364e-06,
      "loss": 0.2174,
      "step": 31070
    },
    {
      "epoch": 3.0567234638931917,
      "grad_norm": 0.5166285634040833,
      "learning_rate": 7.850812701188254e-06,
      "loss": 0.182,
      "step": 31080
    },
    {
      "epoch": 3.0577069656512013,
      "grad_norm": 1.0382561683654785,
      "learning_rate": 7.846838612248143e-06,
      "loss": 0.2865,
      "step": 31090
    },
    {
      "epoch": 3.0586904674092104,
      "grad_norm": 0.04259077087044716,
      "learning_rate": 7.842864523308033e-06,
      "loss": 0.2629,
      "step": 31100
    },
    {
      "epoch": 3.05967396916722,
      "grad_norm": 0.3611224889755249,
      "learning_rate": 7.838890434367923e-06,
      "loss": 0.1513,
      "step": 31110
    },
    {
      "epoch": 3.0606574709252294,
      "grad_norm": 1.518552303314209,
      "learning_rate": 7.83491634542781e-06,
      "loss": 0.203,
      "step": 31120
    },
    {
      "epoch": 3.0616409726832385,
      "grad_norm": 0.2685289680957794,
      "learning_rate": 7.830942256487702e-06,
      "loss": 0.1509,
      "step": 31130
    },
    {
      "epoch": 3.062624474441248,
      "grad_norm": 0.5196729898452759,
      "learning_rate": 7.82696816754759e-06,
      "loss": 0.171,
      "step": 31140
    },
    {
      "epoch": 3.0636079761992576,
      "grad_norm": 0.7608104944229126,
      "learning_rate": 7.82299407860748e-06,
      "loss": 0.2358,
      "step": 31150
    },
    {
      "epoch": 3.0645914779572667,
      "grad_norm": 0.5399700403213501,
      "learning_rate": 7.81901998966737e-06,
      "loss": 0.1741,
      "step": 31160
    },
    {
      "epoch": 3.0655749797152763,
      "grad_norm": 0.7594351172447205,
      "learning_rate": 7.815045900727259e-06,
      "loss": 0.1544,
      "step": 31170
    },
    {
      "epoch": 3.066558481473286,
      "grad_norm": 0.9618496298789978,
      "learning_rate": 7.811071811787149e-06,
      "loss": 0.1786,
      "step": 31180
    },
    {
      "epoch": 3.067541983231295,
      "grad_norm": 0.7477512359619141,
      "learning_rate": 7.807097722847038e-06,
      "loss": 0.1425,
      "step": 31190
    },
    {
      "epoch": 3.0685254849893044,
      "grad_norm": 0.30735233426094055,
      "learning_rate": 7.803123633906928e-06,
      "loss": 0.2515,
      "step": 31200
    },
    {
      "epoch": 3.069508986747314,
      "grad_norm": 0.758304238319397,
      "learning_rate": 7.799149544966818e-06,
      "loss": 0.1892,
      "step": 31210
    },
    {
      "epoch": 3.070492488505323,
      "grad_norm": 0.4117341637611389,
      "learning_rate": 7.795175456026707e-06,
      "loss": 0.1546,
      "step": 31220
    },
    {
      "epoch": 3.0714759902633326,
      "grad_norm": 1.254814863204956,
      "learning_rate": 7.791201367086597e-06,
      "loss": 0.1797,
      "step": 31230
    },
    {
      "epoch": 3.072459492021342,
      "grad_norm": 1.452399492263794,
      "learning_rate": 7.787227278146485e-06,
      "loss": 0.2549,
      "step": 31240
    },
    {
      "epoch": 3.0734429937793513,
      "grad_norm": 0.6258145570755005,
      "learning_rate": 7.783253189206376e-06,
      "loss": 0.1528,
      "step": 31250
    },
    {
      "epoch": 3.074426495537361,
      "grad_norm": 1.303449273109436,
      "learning_rate": 7.779279100266264e-06,
      "loss": 0.1748,
      "step": 31260
    },
    {
      "epoch": 3.0754099972953703,
      "grad_norm": 1.0126678943634033,
      "learning_rate": 7.775305011326154e-06,
      "loss": 0.2219,
      "step": 31270
    },
    {
      "epoch": 3.0763934990533794,
      "grad_norm": 0.4630603790283203,
      "learning_rate": 7.771330922386043e-06,
      "loss": 0.1699,
      "step": 31280
    },
    {
      "epoch": 3.077377000811389,
      "grad_norm": 1.1373541355133057,
      "learning_rate": 7.767356833445933e-06,
      "loss": 0.2667,
      "step": 31290
    },
    {
      "epoch": 3.0783605025693985,
      "grad_norm": 3.100982427597046,
      "learning_rate": 7.763382744505823e-06,
      "loss": 0.1706,
      "step": 31300
    },
    {
      "epoch": 3.0793440043274076,
      "grad_norm": 0.4552260637283325,
      "learning_rate": 7.759408655565712e-06,
      "loss": 0.1831,
      "step": 31310
    },
    {
      "epoch": 3.080327506085417,
      "grad_norm": 2.0742850303649902,
      "learning_rate": 7.755434566625602e-06,
      "loss": 0.1867,
      "step": 31320
    },
    {
      "epoch": 3.0813110078434267,
      "grad_norm": 0.5648025870323181,
      "learning_rate": 7.75146047768549e-06,
      "loss": 0.179,
      "step": 31330
    },
    {
      "epoch": 3.082294509601436,
      "grad_norm": 0.28768691420555115,
      "learning_rate": 7.747486388745381e-06,
      "loss": 0.1708,
      "step": 31340
    },
    {
      "epoch": 3.0832780113594453,
      "grad_norm": 0.522784411907196,
      "learning_rate": 7.743512299805271e-06,
      "loss": 0.192,
      "step": 31350
    },
    {
      "epoch": 3.084261513117455,
      "grad_norm": 0.7258384227752686,
      "learning_rate": 7.739538210865159e-06,
      "loss": 0.2074,
      "step": 31360
    },
    {
      "epoch": 3.085245014875464,
      "grad_norm": 1.0599561929702759,
      "learning_rate": 7.735564121925049e-06,
      "loss": 0.1306,
      "step": 31370
    },
    {
      "epoch": 3.0862285166334735,
      "grad_norm": 0.6479315161705017,
      "learning_rate": 7.731590032984938e-06,
      "loss": 0.3329,
      "step": 31380
    },
    {
      "epoch": 3.087212018391483,
      "grad_norm": 1.6357970237731934,
      "learning_rate": 7.727615944044828e-06,
      "loss": 0.2368,
      "step": 31390
    },
    {
      "epoch": 3.088195520149492,
      "grad_norm": 1.8428471088409424,
      "learning_rate": 7.723641855104718e-06,
      "loss": 0.1648,
      "step": 31400
    },
    {
      "epoch": 3.0891790219075017,
      "grad_norm": 0.43346959352493286,
      "learning_rate": 7.719667766164607e-06,
      "loss": 0.1035,
      "step": 31410
    },
    {
      "epoch": 3.0901625236655113,
      "grad_norm": 1.0585253238677979,
      "learning_rate": 7.715693677224497e-06,
      "loss": 0.1864,
      "step": 31420
    },
    {
      "epoch": 3.0911460254235203,
      "grad_norm": 0.7270784974098206,
      "learning_rate": 7.711719588284387e-06,
      "loss": 0.1715,
      "step": 31430
    },
    {
      "epoch": 3.09212952718153,
      "grad_norm": 0.19803878664970398,
      "learning_rate": 7.707745499344276e-06,
      "loss": 0.1547,
      "step": 31440
    },
    {
      "epoch": 3.0931130289395394,
      "grad_norm": 0.36690932512283325,
      "learning_rate": 7.703771410404164e-06,
      "loss": 0.2647,
      "step": 31450
    },
    {
      "epoch": 3.0940965306975485,
      "grad_norm": 0.42811933159828186,
      "learning_rate": 7.699797321464056e-06,
      "loss": 0.2242,
      "step": 31460
    },
    {
      "epoch": 3.095080032455558,
      "grad_norm": 0.3193596303462982,
      "learning_rate": 7.695823232523945e-06,
      "loss": 0.1819,
      "step": 31470
    },
    {
      "epoch": 3.0960635342135676,
      "grad_norm": 0.5629414319992065,
      "learning_rate": 7.691849143583833e-06,
      "loss": 0.2528,
      "step": 31480
    },
    {
      "epoch": 3.0970470359715767,
      "grad_norm": 0.21826787292957306,
      "learning_rate": 7.687875054643723e-06,
      "loss": 0.1463,
      "step": 31490
    },
    {
      "epoch": 3.0980305377295863,
      "grad_norm": 1.1460973024368286,
      "learning_rate": 7.683900965703613e-06,
      "loss": 0.1669,
      "step": 31500
    },
    {
      "epoch": 3.0980305377295863,
      "eval_loss": 0.1169167309999466,
      "eval_runtime": 18.3179,
      "eval_samples_per_second": 2.73,
      "eval_steps_per_second": 1.365,
      "step": 31500
    },
    {
      "epoch": 3.099014039487596,
      "grad_norm": 1.028279423713684,
      "learning_rate": 7.679926876763502e-06,
      "loss": 0.2178,
      "step": 31510
    },
    {
      "epoch": 3.099997541245605,
      "grad_norm": 0.3888244330883026,
      "learning_rate": 7.675952787823392e-06,
      "loss": 0.1574,
      "step": 31520
    },
    {
      "epoch": 3.1009810430036144,
      "grad_norm": 0.4609070420265198,
      "learning_rate": 7.671978698883282e-06,
      "loss": 0.1974,
      "step": 31530
    },
    {
      "epoch": 3.101964544761624,
      "grad_norm": 0.3917083740234375,
      "learning_rate": 7.668004609943171e-06,
      "loss": 0.2735,
      "step": 31540
    },
    {
      "epoch": 3.102948046519633,
      "grad_norm": 1.1195194721221924,
      "learning_rate": 7.664030521003061e-06,
      "loss": 0.1768,
      "step": 31550
    },
    {
      "epoch": 3.1039315482776426,
      "grad_norm": 0.25413259863853455,
      "learning_rate": 7.66005643206295e-06,
      "loss": 0.187,
      "step": 31560
    },
    {
      "epoch": 3.104915050035652,
      "grad_norm": 0.1355896294116974,
      "learning_rate": 7.656082343122839e-06,
      "loss": 0.0985,
      "step": 31570
    },
    {
      "epoch": 3.1058985517936613,
      "grad_norm": 0.3639681339263916,
      "learning_rate": 7.65210825418273e-06,
      "loss": 0.2879,
      "step": 31580
    },
    {
      "epoch": 3.106882053551671,
      "grad_norm": 0.13450194895267487,
      "learning_rate": 7.64813416524262e-06,
      "loss": 0.1446,
      "step": 31590
    },
    {
      "epoch": 3.1078655553096803,
      "grad_norm": 30.96489715576172,
      "learning_rate": 7.644160076302508e-06,
      "loss": 0.1567,
      "step": 31600
    },
    {
      "epoch": 3.1088490570676894,
      "grad_norm": 1.1467623710632324,
      "learning_rate": 7.640185987362397e-06,
      "loss": 0.1917,
      "step": 31610
    },
    {
      "epoch": 3.109832558825699,
      "grad_norm": 2.060237407684326,
      "learning_rate": 7.636211898422287e-06,
      "loss": 0.1424,
      "step": 31620
    },
    {
      "epoch": 3.1108160605837085,
      "grad_norm": 0.38937437534332275,
      "learning_rate": 7.632237809482177e-06,
      "loss": 0.1967,
      "step": 31630
    },
    {
      "epoch": 3.1117995623417176,
      "grad_norm": 0.9515398740768433,
      "learning_rate": 7.628263720542066e-06,
      "loss": 0.1264,
      "step": 31640
    },
    {
      "epoch": 3.112783064099727,
      "grad_norm": 0.6877108216285706,
      "learning_rate": 7.624289631601955e-06,
      "loss": 0.1955,
      "step": 31650
    },
    {
      "epoch": 3.1137665658577363,
      "grad_norm": 0.5784842371940613,
      "learning_rate": 7.620315542661846e-06,
      "loss": 0.1444,
      "step": 31660
    },
    {
      "epoch": 3.114750067615746,
      "grad_norm": 1.5829799175262451,
      "learning_rate": 7.616341453721735e-06,
      "loss": 0.2132,
      "step": 31670
    },
    {
      "epoch": 3.1157335693737553,
      "grad_norm": 0.7843524217605591,
      "learning_rate": 7.612367364781624e-06,
      "loss": 0.1878,
      "step": 31680
    },
    {
      "epoch": 3.1167170711317644,
      "grad_norm": 1.2282633781433105,
      "learning_rate": 7.608393275841514e-06,
      "loss": 0.2217,
      "step": 31690
    },
    {
      "epoch": 3.117700572889774,
      "grad_norm": 0.3797919750213623,
      "learning_rate": 7.6044191869014035e-06,
      "loss": 0.119,
      "step": 31700
    },
    {
      "epoch": 3.1186840746477835,
      "grad_norm": 2.6364216804504395,
      "learning_rate": 7.600445097961293e-06,
      "loss": 0.2002,
      "step": 31710
    },
    {
      "epoch": 3.1196675764057926,
      "grad_norm": 1.5436702966690063,
      "learning_rate": 7.596471009021183e-06,
      "loss": 0.1477,
      "step": 31720
    },
    {
      "epoch": 3.120651078163802,
      "grad_norm": 0.4043786823749542,
      "learning_rate": 7.592496920081072e-06,
      "loss": 0.2011,
      "step": 31730
    },
    {
      "epoch": 3.1216345799218117,
      "grad_norm": 0.8174799084663391,
      "learning_rate": 7.588522831140962e-06,
      "loss": 0.1601,
      "step": 31740
    },
    {
      "epoch": 3.122618081679821,
      "grad_norm": 0.34667864441871643,
      "learning_rate": 7.584548742200851e-06,
      "loss": 0.1948,
      "step": 31750
    },
    {
      "epoch": 3.1236015834378303,
      "grad_norm": 0.3857724070549011,
      "learning_rate": 7.580574653260741e-06,
      "loss": 0.118,
      "step": 31760
    },
    {
      "epoch": 3.12458508519584,
      "grad_norm": 1.493760347366333,
      "learning_rate": 7.5766005643206294e-06,
      "loss": 0.2103,
      "step": 31770
    },
    {
      "epoch": 3.125568586953849,
      "grad_norm": 1.1172994375228882,
      "learning_rate": 7.57262647538052e-06,
      "loss": 0.1778,
      "step": 31780
    },
    {
      "epoch": 3.1265520887118585,
      "grad_norm": 0.6166003346443176,
      "learning_rate": 7.56865238644041e-06,
      "loss": 0.152,
      "step": 31790
    },
    {
      "epoch": 3.127535590469868,
      "grad_norm": 0.5556543469429016,
      "learning_rate": 7.5646782975002984e-06,
      "loss": 0.2136,
      "step": 31800
    },
    {
      "epoch": 3.128519092227877,
      "grad_norm": 1.0468389987945557,
      "learning_rate": 7.560704208560188e-06,
      "loss": 0.2625,
      "step": 31810
    },
    {
      "epoch": 3.1295025939858867,
      "grad_norm": 0.9141632914543152,
      "learning_rate": 7.556730119620078e-06,
      "loss": 0.1199,
      "step": 31820
    },
    {
      "epoch": 3.1304860957438962,
      "grad_norm": 1.8059278726577759,
      "learning_rate": 7.5527560306799674e-06,
      "loss": 0.1591,
      "step": 31830
    },
    {
      "epoch": 3.1314695975019053,
      "grad_norm": 0.9291358590126038,
      "learning_rate": 7.548781941739857e-06,
      "loss": 0.1537,
      "step": 31840
    },
    {
      "epoch": 3.132453099259915,
      "grad_norm": 0.43296074867248535,
      "learning_rate": 7.544807852799746e-06,
      "loss": 0.1286,
      "step": 31850
    },
    {
      "epoch": 3.1334366010179244,
      "grad_norm": 0.7689695954322815,
      "learning_rate": 7.5408337638596364e-06,
      "loss": 0.1265,
      "step": 31860
    },
    {
      "epoch": 3.1344201027759335,
      "grad_norm": 1.3215502500534058,
      "learning_rate": 7.536859674919525e-06,
      "loss": 0.1528,
      "step": 31870
    },
    {
      "epoch": 3.135403604533943,
      "grad_norm": 0.22979998588562012,
      "learning_rate": 7.532885585979415e-06,
      "loss": 0.2016,
      "step": 31880
    },
    {
      "epoch": 3.1363871062919526,
      "grad_norm": 1.030752182006836,
      "learning_rate": 7.528911497039304e-06,
      "loss": 0.1128,
      "step": 31890
    },
    {
      "epoch": 3.1373706080499617,
      "grad_norm": 2.3236682415008545,
      "learning_rate": 7.524937408099194e-06,
      "loss": 0.1911,
      "step": 31900
    },
    {
      "epoch": 3.1383541098079712,
      "grad_norm": 0.23360274732112885,
      "learning_rate": 7.520963319159084e-06,
      "loss": 0.1864,
      "step": 31910
    },
    {
      "epoch": 3.139337611565981,
      "grad_norm": 1.6743212938308716,
      "learning_rate": 7.516989230218973e-06,
      "loss": 0.1219,
      "step": 31920
    },
    {
      "epoch": 3.14032111332399,
      "grad_norm": 0.8622710704803467,
      "learning_rate": 7.513015141278862e-06,
      "loss": 0.2372,
      "step": 31930
    },
    {
      "epoch": 3.1413046150819994,
      "grad_norm": 0.4087425172328949,
      "learning_rate": 7.509041052338752e-06,
      "loss": 0.1996,
      "step": 31940
    },
    {
      "epoch": 3.142288116840009,
      "grad_norm": 1.8322125673294067,
      "learning_rate": 7.505066963398642e-06,
      "loss": 0.2281,
      "step": 31950
    },
    {
      "epoch": 3.143271618598018,
      "grad_norm": 2.583505392074585,
      "learning_rate": 7.501092874458531e-06,
      "loss": 0.1595,
      "step": 31960
    },
    {
      "epoch": 3.1442551203560276,
      "grad_norm": 0.5924628376960754,
      "learning_rate": 7.49711878551842e-06,
      "loss": 0.1901,
      "step": 31970
    },
    {
      "epoch": 3.145238622114037,
      "grad_norm": 1.7466529607772827,
      "learning_rate": 7.493144696578311e-06,
      "loss": 0.1545,
      "step": 31980
    },
    {
      "epoch": 3.1462221238720463,
      "grad_norm": 1.0587272644042969,
      "learning_rate": 7.4891706076381995e-06,
      "loss": 0.1402,
      "step": 31990
    },
    {
      "epoch": 3.147205625630056,
      "grad_norm": 1.4738565683364868,
      "learning_rate": 7.485196518698089e-06,
      "loss": 0.2065,
      "step": 32000
    },
    {
      "epoch": 3.147205625630056,
      "eval_loss": 0.11695942282676697,
      "eval_runtime": 16.6238,
      "eval_samples_per_second": 3.008,
      "eval_steps_per_second": 1.504,
      "step": 32000
    },
    {
      "epoch": 3.1481891273880653,
      "grad_norm": 1.245446801185608,
      "learning_rate": 7.481222429757978e-06,
      "loss": 0.1854,
      "step": 32010
    },
    {
      "epoch": 3.1491726291460744,
      "grad_norm": 1.756117343902588,
      "learning_rate": 7.4772483408178685e-06,
      "loss": 0.2727,
      "step": 32020
    },
    {
      "epoch": 3.150156130904084,
      "grad_norm": 0.3403267562389374,
      "learning_rate": 7.473274251877758e-06,
      "loss": 0.2097,
      "step": 32030
    },
    {
      "epoch": 3.1511396326620935,
      "grad_norm": 0.45989128947257996,
      "learning_rate": 7.469300162937647e-06,
      "loss": 0.1552,
      "step": 32040
    },
    {
      "epoch": 3.1521231344201026,
      "grad_norm": 1.0242866277694702,
      "learning_rate": 7.465326073997537e-06,
      "loss": 0.2169,
      "step": 32050
    },
    {
      "epoch": 3.153106636178112,
      "grad_norm": 0.25825977325439453,
      "learning_rate": 7.461351985057426e-06,
      "loss": 0.1225,
      "step": 32060
    },
    {
      "epoch": 3.1540901379361217,
      "grad_norm": 1.4743444919586182,
      "learning_rate": 7.457377896117316e-06,
      "loss": 0.2738,
      "step": 32070
    },
    {
      "epoch": 3.155073639694131,
      "grad_norm": 1.8272091150283813,
      "learning_rate": 7.453403807177206e-06,
      "loss": 0.2369,
      "step": 32080
    },
    {
      "epoch": 3.1560571414521403,
      "grad_norm": 0.8361017107963562,
      "learning_rate": 7.4494297182370945e-06,
      "loss": 0.1716,
      "step": 32090
    },
    {
      "epoch": 3.15704064321015,
      "grad_norm": 0.7447807788848877,
      "learning_rate": 7.445455629296984e-06,
      "loss": 0.1722,
      "step": 32100
    },
    {
      "epoch": 3.158024144968159,
      "grad_norm": 3.4965121746063232,
      "learning_rate": 7.441481540356874e-06,
      "loss": 0.2161,
      "step": 32110
    },
    {
      "epoch": 3.1590076467261685,
      "grad_norm": 2.7191264629364014,
      "learning_rate": 7.4375074514167635e-06,
      "loss": 0.1452,
      "step": 32120
    },
    {
      "epoch": 3.159991148484178,
      "grad_norm": 1.3699681758880615,
      "learning_rate": 7.433533362476652e-06,
      "loss": 0.2319,
      "step": 32130
    },
    {
      "epoch": 3.160974650242187,
      "grad_norm": 0.4341868460178375,
      "learning_rate": 7.429559273536542e-06,
      "loss": 0.1768,
      "step": 32140
    },
    {
      "epoch": 3.1619581520001967,
      "grad_norm": 0.7707095146179199,
      "learning_rate": 7.4255851845964325e-06,
      "loss": 0.1715,
      "step": 32150
    },
    {
      "epoch": 3.1629416537582062,
      "grad_norm": 0.3822473883628845,
      "learning_rate": 7.421611095656321e-06,
      "loss": 0.1659,
      "step": 32160
    },
    {
      "epoch": 3.1639251555162153,
      "grad_norm": 1.2275961637496948,
      "learning_rate": 7.417637006716211e-06,
      "loss": 0.2094,
      "step": 32170
    },
    {
      "epoch": 3.164908657274225,
      "grad_norm": 1.0981230735778809,
      "learning_rate": 7.4136629177761e-06,
      "loss": 0.1301,
      "step": 32180
    },
    {
      "epoch": 3.1658921590322344,
      "grad_norm": 0.4137146770954132,
      "learning_rate": 7.40968882883599e-06,
      "loss": 0.1893,
      "step": 32190
    },
    {
      "epoch": 3.1668756607902435,
      "grad_norm": 0.7313489317893982,
      "learning_rate": 7.40571473989588e-06,
      "loss": 0.2226,
      "step": 32200
    },
    {
      "epoch": 3.167859162548253,
      "grad_norm": 0.5336371064186096,
      "learning_rate": 7.401740650955769e-06,
      "loss": 0.1276,
      "step": 32210
    },
    {
      "epoch": 3.1688426643062626,
      "grad_norm": 0.16569511592388153,
      "learning_rate": 7.3977665620156584e-06,
      "loss": 0.1372,
      "step": 32220
    },
    {
      "epoch": 3.1698261660642717,
      "grad_norm": 0.27265745401382446,
      "learning_rate": 7.393792473075548e-06,
      "loss": 0.3227,
      "step": 32230
    },
    {
      "epoch": 3.1708096678222812,
      "grad_norm": 1.2933647632598877,
      "learning_rate": 7.389818384135438e-06,
      "loss": 0.163,
      "step": 32240
    },
    {
      "epoch": 3.171793169580291,
      "grad_norm": 2.2621102333068848,
      "learning_rate": 7.3858442951953266e-06,
      "loss": 0.1644,
      "step": 32250
    },
    {
      "epoch": 3.1727766713383,
      "grad_norm": 0.5217112898826599,
      "learning_rate": 7.381870206255216e-06,
      "loss": 0.2593,
      "step": 32260
    },
    {
      "epoch": 3.1737601730963094,
      "grad_norm": 0.8894764184951782,
      "learning_rate": 7.377896117315107e-06,
      "loss": 0.1204,
      "step": 32270
    },
    {
      "epoch": 3.174743674854319,
      "grad_norm": 1.873124361038208,
      "learning_rate": 7.3739220283749956e-06,
      "loss": 0.2319,
      "step": 32280
    },
    {
      "epoch": 3.175727176612328,
      "grad_norm": 0.8186569213867188,
      "learning_rate": 7.369947939434885e-06,
      "loss": 0.1459,
      "step": 32290
    },
    {
      "epoch": 3.1767106783703376,
      "grad_norm": 0.42005041241645813,
      "learning_rate": 7.365973850494774e-06,
      "loss": 0.159,
      "step": 32300
    },
    {
      "epoch": 3.177694180128347,
      "grad_norm": 1.3418946266174316,
      "learning_rate": 7.3619997615546646e-06,
      "loss": 0.2799,
      "step": 32310
    },
    {
      "epoch": 3.1786776818863562,
      "grad_norm": 1.168924331665039,
      "learning_rate": 7.358025672614554e-06,
      "loss": 0.2939,
      "step": 32320
    },
    {
      "epoch": 3.179661183644366,
      "grad_norm": 0.21440809965133667,
      "learning_rate": 7.354051583674443e-06,
      "loss": 0.0732,
      "step": 32330
    },
    {
      "epoch": 3.1806446854023753,
      "grad_norm": 1.0417389869689941,
      "learning_rate": 7.350077494734333e-06,
      "loss": 0.1937,
      "step": 32340
    },
    {
      "epoch": 3.1816281871603844,
      "grad_norm": 0.9816319346427917,
      "learning_rate": 7.346103405794222e-06,
      "loss": 0.1935,
      "step": 32350
    },
    {
      "epoch": 3.182611688918394,
      "grad_norm": 1.8569186925888062,
      "learning_rate": 7.342129316854112e-06,
      "loss": 0.169,
      "step": 32360
    },
    {
      "epoch": 3.1835951906764035,
      "grad_norm": 1.4092768430709839,
      "learning_rate": 7.338155227914001e-06,
      "loss": 0.2134,
      "step": 32370
    },
    {
      "epoch": 3.1845786924344126,
      "grad_norm": 0.5531893372535706,
      "learning_rate": 7.3341811389738905e-06,
      "loss": 0.2163,
      "step": 32380
    },
    {
      "epoch": 3.185562194192422,
      "grad_norm": 1.8517935276031494,
      "learning_rate": 7.330207050033781e-06,
      "loss": 0.1831,
      "step": 32390
    },
    {
      "epoch": 3.1865456959504317,
      "grad_norm": 1.3389599323272705,
      "learning_rate": 7.32623296109367e-06,
      "loss": 0.1613,
      "step": 32400
    },
    {
      "epoch": 3.187529197708441,
      "grad_norm": 0.3436429798603058,
      "learning_rate": 7.3222588721535595e-06,
      "loss": 0.2205,
      "step": 32410
    },
    {
      "epoch": 3.1885126994664503,
      "grad_norm": 1.0781066417694092,
      "learning_rate": 7.318284783213448e-06,
      "loss": 0.121,
      "step": 32420
    },
    {
      "epoch": 3.18949620122446,
      "grad_norm": 0.9889883995056152,
      "learning_rate": 7.314310694273339e-06,
      "loss": 0.1683,
      "step": 32430
    },
    {
      "epoch": 3.190479702982469,
      "grad_norm": 2.1204090118408203,
      "learning_rate": 7.310336605333228e-06,
      "loss": 0.2273,
      "step": 32440
    },
    {
      "epoch": 3.1914632047404785,
      "grad_norm": 0.8797752857208252,
      "learning_rate": 7.306362516393117e-06,
      "loss": 0.1523,
      "step": 32450
    },
    {
      "epoch": 3.192446706498488,
      "grad_norm": 1.7521774768829346,
      "learning_rate": 7.302388427453007e-06,
      "loss": 0.2407,
      "step": 32460
    },
    {
      "epoch": 3.193430208256497,
      "grad_norm": 1.2867456674575806,
      "learning_rate": 7.298414338512897e-06,
      "loss": 0.176,
      "step": 32470
    },
    {
      "epoch": 3.1944137100145067,
      "grad_norm": 0.11294372379779816,
      "learning_rate": 7.294440249572786e-06,
      "loss": 0.1571,
      "step": 32480
    },
    {
      "epoch": 3.1953972117725162,
      "grad_norm": 1.2366126775741577,
      "learning_rate": 7.290466160632675e-06,
      "loss": 0.1813,
      "step": 32490
    },
    {
      "epoch": 3.1963807135305253,
      "grad_norm": 0.6433688402175903,
      "learning_rate": 7.286492071692565e-06,
      "loss": 0.1792,
      "step": 32500
    },
    {
      "epoch": 3.1963807135305253,
      "eval_loss": 0.11758752912282944,
      "eval_runtime": 20.8811,
      "eval_samples_per_second": 2.395,
      "eval_steps_per_second": 1.197,
      "step": 32500
    },
    {
      "epoch": 3.197364215288535,
      "grad_norm": 1.1521432399749756,
      "learning_rate": 7.282517982752455e-06,
      "loss": 0.2708,
      "step": 32510
    },
    {
      "epoch": 3.1983477170465444,
      "grad_norm": 0.4797680377960205,
      "learning_rate": 7.278543893812344e-06,
      "loss": 0.3051,
      "step": 32520
    },
    {
      "epoch": 3.1993312188045535,
      "grad_norm": 1.5195132493972778,
      "learning_rate": 7.274569804872234e-06,
      "loss": 0.1316,
      "step": 32530
    },
    {
      "epoch": 3.200314720562563,
      "grad_norm": 0.17302559316158295,
      "learning_rate": 7.270595715932123e-06,
      "loss": 0.1391,
      "step": 32540
    },
    {
      "epoch": 3.2012982223205726,
      "grad_norm": 0.9784829616546631,
      "learning_rate": 7.266621626992013e-06,
      "loss": 0.2081,
      "step": 32550
    },
    {
      "epoch": 3.2022817240785817,
      "grad_norm": 0.8939286470413208,
      "learning_rate": 7.262647538051902e-06,
      "loss": 0.1981,
      "step": 32560
    },
    {
      "epoch": 3.2032652258365912,
      "grad_norm": 0.6267346143722534,
      "learning_rate": 7.258673449111792e-06,
      "loss": 0.2875,
      "step": 32570
    },
    {
      "epoch": 3.2042487275946008,
      "grad_norm": 0.15202075242996216,
      "learning_rate": 7.254699360171681e-06,
      "loss": 0.2477,
      "step": 32580
    },
    {
      "epoch": 3.20523222935261,
      "grad_norm": 0.5904789566993713,
      "learning_rate": 7.250725271231571e-06,
      "loss": 0.1588,
      "step": 32590
    },
    {
      "epoch": 3.2062157311106194,
      "grad_norm": 0.4800198972225189,
      "learning_rate": 7.246751182291461e-06,
      "loss": 0.0764,
      "step": 32600
    },
    {
      "epoch": 3.207199232868629,
      "grad_norm": 0.41733959317207336,
      "learning_rate": 7.242777093351349e-06,
      "loss": 0.2756,
      "step": 32610
    },
    {
      "epoch": 3.208182734626638,
      "grad_norm": 0.09101834148168564,
      "learning_rate": 7.238803004411239e-06,
      "loss": 0.1507,
      "step": 32620
    },
    {
      "epoch": 3.2091662363846476,
      "grad_norm": 0.7980632781982422,
      "learning_rate": 7.23482891547113e-06,
      "loss": 0.1985,
      "step": 32630
    },
    {
      "epoch": 3.2101497381426567,
      "grad_norm": 1.3122634887695312,
      "learning_rate": 7.230854826531018e-06,
      "loss": 0.1364,
      "step": 32640
    },
    {
      "epoch": 3.2111332399006662,
      "grad_norm": 1.0960237979888916,
      "learning_rate": 7.226880737590908e-06,
      "loss": 0.1429,
      "step": 32650
    },
    {
      "epoch": 3.212116741658676,
      "grad_norm": 0.1762075573205948,
      "learning_rate": 7.222906648650797e-06,
      "loss": 0.1854,
      "step": 32660
    },
    {
      "epoch": 3.213100243416685,
      "grad_norm": 0.3500872254371643,
      "learning_rate": 7.218932559710687e-06,
      "loss": 0.2285,
      "step": 32670
    },
    {
      "epoch": 3.2140837451746944,
      "grad_norm": 0.602541446685791,
      "learning_rate": 7.214958470770576e-06,
      "loss": 0.2755,
      "step": 32680
    },
    {
      "epoch": 3.215067246932704,
      "grad_norm": 0.34568265080451965,
      "learning_rate": 7.210984381830466e-06,
      "loss": 0.1966,
      "step": 32690
    },
    {
      "epoch": 3.216050748690713,
      "grad_norm": 0.6943805813789368,
      "learning_rate": 7.2070102928903556e-06,
      "loss": 0.2801,
      "step": 32700
    },
    {
      "epoch": 3.2170342504487226,
      "grad_norm": 1.8825277090072632,
      "learning_rate": 7.203036203950245e-06,
      "loss": 0.2246,
      "step": 32710
    },
    {
      "epoch": 3.218017752206732,
      "grad_norm": 1.251085877418518,
      "learning_rate": 7.199062115010135e-06,
      "loss": 0.2346,
      "step": 32720
    },
    {
      "epoch": 3.2190012539647412,
      "grad_norm": 1.6574790477752686,
      "learning_rate": 7.195088026070024e-06,
      "loss": 0.2262,
      "step": 32730
    },
    {
      "epoch": 3.219984755722751,
      "grad_norm": 0.347114622592926,
      "learning_rate": 7.191113937129913e-06,
      "loss": 0.1595,
      "step": 32740
    },
    {
      "epoch": 3.2209682574807603,
      "grad_norm": 1.9753527641296387,
      "learning_rate": 7.187139848189804e-06,
      "loss": 0.2737,
      "step": 32750
    },
    {
      "epoch": 3.2219517592387694,
      "grad_norm": 0.15236753225326538,
      "learning_rate": 7.183165759249693e-06,
      "loss": 0.1592,
      "step": 32760
    },
    {
      "epoch": 3.222935260996779,
      "grad_norm": 0.43684929609298706,
      "learning_rate": 7.179191670309582e-06,
      "loss": 0.1877,
      "step": 32770
    },
    {
      "epoch": 3.2239187627547885,
      "grad_norm": 0.6682396531105042,
      "learning_rate": 7.175217581369471e-06,
      "loss": 0.1321,
      "step": 32780
    },
    {
      "epoch": 3.2249022645127976,
      "grad_norm": 0.430677205324173,
      "learning_rate": 7.171243492429362e-06,
      "loss": 0.2208,
      "step": 32790
    },
    {
      "epoch": 3.225885766270807,
      "grad_norm": 1.3447238206863403,
      "learning_rate": 7.1672694034892505e-06,
      "loss": 0.2048,
      "step": 32800
    },
    {
      "epoch": 3.2268692680288167,
      "grad_norm": 0.4871419072151184,
      "learning_rate": 7.16329531454914e-06,
      "loss": 0.1191,
      "step": 32810
    },
    {
      "epoch": 3.227852769786826,
      "grad_norm": 0.9699082970619202,
      "learning_rate": 7.15932122560903e-06,
      "loss": 0.2104,
      "step": 32820
    },
    {
      "epoch": 3.2288362715448353,
      "grad_norm": 0.5603752136230469,
      "learning_rate": 7.1553471366689195e-06,
      "loss": 0.177,
      "step": 32830
    },
    {
      "epoch": 3.229819773302845,
      "grad_norm": 0.5944997668266296,
      "learning_rate": 7.151373047728809e-06,
      "loss": 0.1743,
      "step": 32840
    },
    {
      "epoch": 3.230803275060854,
      "grad_norm": 1.1690326929092407,
      "learning_rate": 7.147398958788698e-06,
      "loss": 0.2163,
      "step": 32850
    },
    {
      "epoch": 3.2317867768188635,
      "grad_norm": 0.5509839057922363,
      "learning_rate": 7.143424869848588e-06,
      "loss": 0.2276,
      "step": 32860
    },
    {
      "epoch": 3.232770278576873,
      "grad_norm": 1.3236446380615234,
      "learning_rate": 7.1394507809084765e-06,
      "loss": 0.1761,
      "step": 32870
    },
    {
      "epoch": 3.233753780334882,
      "grad_norm": 0.7778728604316711,
      "learning_rate": 7.135476691968367e-06,
      "loss": 0.2241,
      "step": 32880
    },
    {
      "epoch": 3.2347372820928917,
      "grad_norm": 0.4267124831676483,
      "learning_rate": 7.131502603028257e-06,
      "loss": 0.1965,
      "step": 32890
    },
    {
      "epoch": 3.2357207838509012,
      "grad_norm": 1.4709194898605347,
      "learning_rate": 7.1275285140881455e-06,
      "loss": 0.1544,
      "step": 32900
    },
    {
      "epoch": 3.2367042856089103,
      "grad_norm": 0.23302286863327026,
      "learning_rate": 7.123554425148035e-06,
      "loss": 0.1595,
      "step": 32910
    },
    {
      "epoch": 3.23768778736692,
      "grad_norm": 0.6105418801307678,
      "learning_rate": 7.119580336207925e-06,
      "loss": 0.1827,
      "step": 32920
    },
    {
      "epoch": 3.2386712891249294,
      "grad_norm": 0.40103980898857117,
      "learning_rate": 7.1156062472678145e-06,
      "loss": 0.377,
      "step": 32930
    },
    {
      "epoch": 3.2396547908829385,
      "grad_norm": 0.8761731386184692,
      "learning_rate": 7.111632158327704e-06,
      "loss": 0.2274,
      "step": 32940
    },
    {
      "epoch": 3.240638292640948,
      "grad_norm": 0.181829035282135,
      "learning_rate": 7.107658069387593e-06,
      "loss": 0.2274,
      "step": 32950
    },
    {
      "epoch": 3.2416217943989576,
      "grad_norm": 1.0195627212524414,
      "learning_rate": 7.1036839804474835e-06,
      "loss": 0.3015,
      "step": 32960
    },
    {
      "epoch": 3.2426052961569667,
      "grad_norm": 1.2234985828399658,
      "learning_rate": 7.099709891507372e-06,
      "loss": 0.1759,
      "step": 32970
    },
    {
      "epoch": 3.2435887979149762,
      "grad_norm": 2.0877585411071777,
      "learning_rate": 7.095735802567262e-06,
      "loss": 0.1551,
      "step": 32980
    },
    {
      "epoch": 3.2445722996729858,
      "grad_norm": 1.9813555479049683,
      "learning_rate": 7.091761713627151e-06,
      "loss": 0.1677,
      "step": 32990
    },
    {
      "epoch": 3.245555801430995,
      "grad_norm": 1.2899110317230225,
      "learning_rate": 7.087787624687041e-06,
      "loss": 0.2256,
      "step": 33000
    },
    {
      "epoch": 3.245555801430995,
      "eval_loss": 0.11789312213659286,
      "eval_runtime": 20.8963,
      "eval_samples_per_second": 2.393,
      "eval_steps_per_second": 1.196,
      "step": 33000
    },
    {
      "epoch": 3.2465393031890044,
      "grad_norm": 0.9657261967658997,
      "learning_rate": 7.083813535746931e-06,
      "loss": 0.2636,
      "step": 33010
    },
    {
      "epoch": 3.247522804947014,
      "grad_norm": 1.7438515424728394,
      "learning_rate": 7.07983944680682e-06,
      "loss": 0.2387,
      "step": 33020
    },
    {
      "epoch": 3.248506306705023,
      "grad_norm": 0.4442780613899231,
      "learning_rate": 7.075865357866709e-06,
      "loss": 0.274,
      "step": 33030
    },
    {
      "epoch": 3.2494898084630326,
      "grad_norm": 0.63568115234375,
      "learning_rate": 7.071891268926599e-06,
      "loss": 0.2495,
      "step": 33040
    },
    {
      "epoch": 3.250473310221042,
      "grad_norm": 0.8722742795944214,
      "learning_rate": 7.067917179986489e-06,
      "loss": 0.168,
      "step": 33050
    },
    {
      "epoch": 3.2514568119790512,
      "grad_norm": 1.4411827325820923,
      "learning_rate": 7.063943091046378e-06,
      "loss": 0.2814,
      "step": 33060
    },
    {
      "epoch": 3.2524403137370608,
      "grad_norm": 0.23856908082962036,
      "learning_rate": 7.059969002106267e-06,
      "loss": 0.2155,
      "step": 33070
    },
    {
      "epoch": 3.2534238154950703,
      "grad_norm": 0.4681982696056366,
      "learning_rate": 7.055994913166158e-06,
      "loss": 0.1963,
      "step": 33080
    },
    {
      "epoch": 3.2544073172530794,
      "grad_norm": 0.9917487502098083,
      "learning_rate": 7.0520208242260466e-06,
      "loss": 0.2566,
      "step": 33090
    },
    {
      "epoch": 3.255390819011089,
      "grad_norm": 0.6508780717849731,
      "learning_rate": 7.048046735285936e-06,
      "loss": 0.2226,
      "step": 33100
    },
    {
      "epoch": 3.2563743207690985,
      "grad_norm": 1.2814282178878784,
      "learning_rate": 7.044072646345825e-06,
      "loss": 0.2213,
      "step": 33110
    },
    {
      "epoch": 3.2573578225271076,
      "grad_norm": 1.1933754682540894,
      "learning_rate": 7.0400985574057155e-06,
      "loss": 0.2197,
      "step": 33120
    },
    {
      "epoch": 3.258341324285117,
      "grad_norm": 0.5371510982513428,
      "learning_rate": 7.036124468465605e-06,
      "loss": 0.2021,
      "step": 33130
    },
    {
      "epoch": 3.2593248260431267,
      "grad_norm": 0.8904058337211609,
      "learning_rate": 7.032150379525494e-06,
      "loss": 0.1948,
      "step": 33140
    },
    {
      "epoch": 3.2603083278011358,
      "grad_norm": 0.3558623790740967,
      "learning_rate": 7.028176290585384e-06,
      "loss": 0.2308,
      "step": 33150
    },
    {
      "epoch": 3.2612918295591453,
      "grad_norm": 0.20771129429340363,
      "learning_rate": 7.024202201645273e-06,
      "loss": 0.0968,
      "step": 33160
    },
    {
      "epoch": 3.262275331317155,
      "grad_norm": 0.3460518419742584,
      "learning_rate": 7.020228112705163e-06,
      "loss": 0.1776,
      "step": 33170
    },
    {
      "epoch": 3.263258833075164,
      "grad_norm": 0.3401819169521332,
      "learning_rate": 7.016254023765053e-06,
      "loss": 0.1276,
      "step": 33180
    },
    {
      "epoch": 3.2642423348331735,
      "grad_norm": 0.787639856338501,
      "learning_rate": 7.0122799348249415e-06,
      "loss": 0.2096,
      "step": 33190
    },
    {
      "epoch": 3.265225836591183,
      "grad_norm": 0.21232843399047852,
      "learning_rate": 7.008305845884832e-06,
      "loss": 0.1858,
      "step": 33200
    },
    {
      "epoch": 3.266209338349192,
      "grad_norm": 0.18599532544612885,
      "learning_rate": 7.004331756944721e-06,
      "loss": 0.1921,
      "step": 33210
    },
    {
      "epoch": 3.2671928401072017,
      "grad_norm": 0.44295501708984375,
      "learning_rate": 7.0003576680046105e-06,
      "loss": 0.126,
      "step": 33220
    },
    {
      "epoch": 3.268176341865211,
      "grad_norm": 0.9706286787986755,
      "learning_rate": 6.996383579064499e-06,
      "loss": 0.2774,
      "step": 33230
    },
    {
      "epoch": 3.2691598436232203,
      "grad_norm": 0.16358977556228638,
      "learning_rate": 6.99240949012439e-06,
      "loss": 0.2533,
      "step": 33240
    },
    {
      "epoch": 3.27014334538123,
      "grad_norm": 2.576416492462158,
      "learning_rate": 6.9884354011842795e-06,
      "loss": 0.2855,
      "step": 33250
    },
    {
      "epoch": 3.2711268471392394,
      "grad_norm": 1.0210784673690796,
      "learning_rate": 6.984461312244168e-06,
      "loss": 0.2505,
      "step": 33260
    },
    {
      "epoch": 3.2721103488972485,
      "grad_norm": 0.8797509670257568,
      "learning_rate": 6.980487223304058e-06,
      "loss": 0.2141,
      "step": 33270
    },
    {
      "epoch": 3.273093850655258,
      "grad_norm": 0.30824482440948486,
      "learning_rate": 6.976513134363948e-06,
      "loss": 0.1742,
      "step": 33280
    },
    {
      "epoch": 3.2740773524132676,
      "grad_norm": 0.39702358841896057,
      "learning_rate": 6.972539045423837e-06,
      "loss": 0.1995,
      "step": 33290
    },
    {
      "epoch": 3.2750608541712767,
      "grad_norm": 2.781437873840332,
      "learning_rate": 6.968564956483727e-06,
      "loss": 0.1352,
      "step": 33300
    },
    {
      "epoch": 3.2760443559292862,
      "grad_norm": 1.4445058107376099,
      "learning_rate": 6.964590867543616e-06,
      "loss": 0.1443,
      "step": 33310
    },
    {
      "epoch": 3.2770278576872958,
      "grad_norm": 0.854621410369873,
      "learning_rate": 6.960616778603506e-06,
      "loss": 0.2027,
      "step": 33320
    },
    {
      "epoch": 3.278011359445305,
      "grad_norm": 1.5687012672424316,
      "learning_rate": 6.956642689663395e-06,
      "loss": 0.2498,
      "step": 33330
    },
    {
      "epoch": 3.2789948612033144,
      "grad_norm": 1.041850209236145,
      "learning_rate": 6.952668600723285e-06,
      "loss": 0.1666,
      "step": 33340
    },
    {
      "epoch": 3.279978362961324,
      "grad_norm": 0.35938170552253723,
      "learning_rate": 6.948694511783174e-06,
      "loss": 0.2129,
      "step": 33350
    },
    {
      "epoch": 3.280961864719333,
      "grad_norm": 0.10888555645942688,
      "learning_rate": 6.944720422843064e-06,
      "loss": 0.1266,
      "step": 33360
    },
    {
      "epoch": 3.2819453664773426,
      "grad_norm": 0.6250441670417786,
      "learning_rate": 6.940746333902954e-06,
      "loss": 0.1325,
      "step": 33370
    },
    {
      "epoch": 3.282928868235352,
      "grad_norm": 1.226815938949585,
      "learning_rate": 6.936772244962843e-06,
      "loss": 0.2172,
      "step": 33380
    },
    {
      "epoch": 3.2839123699933612,
      "grad_norm": 0.16180641949176788,
      "learning_rate": 6.932798156022732e-06,
      "loss": 0.1938,
      "step": 33390
    },
    {
      "epoch": 3.2848958717513708,
      "grad_norm": 0.600370466709137,
      "learning_rate": 6.928824067082622e-06,
      "loss": 0.1631,
      "step": 33400
    },
    {
      "epoch": 3.2858793735093803,
      "grad_norm": 1.6209992170333862,
      "learning_rate": 6.924849978142512e-06,
      "loss": 0.1728,
      "step": 33410
    },
    {
      "epoch": 3.2868628752673894,
      "grad_norm": 1.65892493724823,
      "learning_rate": 6.9208758892024e-06,
      "loss": 0.1649,
      "step": 33420
    },
    {
      "epoch": 3.287846377025399,
      "grad_norm": 0.6686263084411621,
      "learning_rate": 6.91690180026229e-06,
      "loss": 0.3185,
      "step": 33430
    },
    {
      "epoch": 3.2888298787834085,
      "grad_norm": 0.32287952303886414,
      "learning_rate": 6.912927711322181e-06,
      "loss": 0.2165,
      "step": 33440
    },
    {
      "epoch": 3.2898133805414176,
      "grad_norm": 0.7468112111091614,
      "learning_rate": 6.908953622382069e-06,
      "loss": 0.2343,
      "step": 33450
    },
    {
      "epoch": 3.290796882299427,
      "grad_norm": 0.6859676241874695,
      "learning_rate": 6.904979533441959e-06,
      "loss": 0.2155,
      "step": 33460
    },
    {
      "epoch": 3.2917803840574367,
      "grad_norm": 2.146181106567383,
      "learning_rate": 6.901005444501848e-06,
      "loss": 0.214,
      "step": 33470
    },
    {
      "epoch": 3.2927638858154458,
      "grad_norm": 1.0984450578689575,
      "learning_rate": 6.897031355561738e-06,
      "loss": 0.2845,
      "step": 33480
    },
    {
      "epoch": 3.2937473875734553,
      "grad_norm": 0.38485586643218994,
      "learning_rate": 6.893057266621628e-06,
      "loss": 0.2578,
      "step": 33490
    },
    {
      "epoch": 3.294730889331465,
      "grad_norm": 0.7894379496574402,
      "learning_rate": 6.889083177681517e-06,
      "loss": 0.1563,
      "step": 33500
    },
    {
      "epoch": 3.294730889331465,
      "eval_loss": 0.117005854845047,
      "eval_runtime": 16.1268,
      "eval_samples_per_second": 3.1,
      "eval_steps_per_second": 1.55,
      "step": 33500
    },
    {
      "epoch": 3.295714391089474,
      "grad_norm": 0.5310940742492676,
      "learning_rate": 6.8851090887414065e-06,
      "loss": 0.2821,
      "step": 33510
    },
    {
      "epoch": 3.2966978928474835,
      "grad_norm": 1.7750648260116577,
      "learning_rate": 6.881134999801296e-06,
      "loss": 0.2697,
      "step": 33520
    },
    {
      "epoch": 3.297681394605493,
      "grad_norm": 1.573096513748169,
      "learning_rate": 6.877160910861186e-06,
      "loss": 0.2191,
      "step": 33530
    },
    {
      "epoch": 3.298664896363502,
      "grad_norm": 0.6690852642059326,
      "learning_rate": 6.873186821921075e-06,
      "loss": 0.2901,
      "step": 33540
    },
    {
      "epoch": 3.2996483981215117,
      "grad_norm": 1.3802883625030518,
      "learning_rate": 6.869212732980964e-06,
      "loss": 0.2022,
      "step": 33550
    },
    {
      "epoch": 3.300631899879521,
      "grad_norm": 0.8098796606063843,
      "learning_rate": 6.865238644040855e-06,
      "loss": 0.2498,
      "step": 33560
    },
    {
      "epoch": 3.3016154016375303,
      "grad_norm": 1.7303550243377686,
      "learning_rate": 6.861264555100744e-06,
      "loss": 0.2695,
      "step": 33570
    },
    {
      "epoch": 3.30259890339554,
      "grad_norm": 0.9760664105415344,
      "learning_rate": 6.857290466160633e-06,
      "loss": 0.1864,
      "step": 33580
    },
    {
      "epoch": 3.3035824051535494,
      "grad_norm": 0.6733797788619995,
      "learning_rate": 6.853316377220522e-06,
      "loss": 0.1678,
      "step": 33590
    },
    {
      "epoch": 3.3045659069115585,
      "grad_norm": 1.4998847246170044,
      "learning_rate": 6.849342288280413e-06,
      "loss": 0.182,
      "step": 33600
    },
    {
      "epoch": 3.305549408669568,
      "grad_norm": 0.3813591003417969,
      "learning_rate": 6.845368199340302e-06,
      "loss": 0.1512,
      "step": 33610
    },
    {
      "epoch": 3.3065329104275776,
      "grad_norm": 0.8506468534469604,
      "learning_rate": 6.841394110400191e-06,
      "loss": 0.1841,
      "step": 33620
    },
    {
      "epoch": 3.3075164121855867,
      "grad_norm": 0.5800037980079651,
      "learning_rate": 6.837420021460081e-06,
      "loss": 0.1648,
      "step": 33630
    },
    {
      "epoch": 3.308499913943596,
      "grad_norm": 0.4694751799106598,
      "learning_rate": 6.8334459325199705e-06,
      "loss": 0.1034,
      "step": 33640
    },
    {
      "epoch": 3.3094834157016058,
      "grad_norm": 1.381948471069336,
      "learning_rate": 6.82947184357986e-06,
      "loss": 0.1966,
      "step": 33650
    },
    {
      "epoch": 3.310466917459615,
      "grad_norm": 0.9570664167404175,
      "learning_rate": 6.825497754639749e-06,
      "loss": 0.1416,
      "step": 33660
    },
    {
      "epoch": 3.3114504192176244,
      "grad_norm": 1.3214004039764404,
      "learning_rate": 6.821523665699639e-06,
      "loss": 0.2701,
      "step": 33670
    },
    {
      "epoch": 3.312433920975634,
      "grad_norm": 1.189394235610962,
      "learning_rate": 6.817549576759528e-06,
      "loss": 0.322,
      "step": 33680
    },
    {
      "epoch": 3.313417422733643,
      "grad_norm": 0.7238168716430664,
      "learning_rate": 6.813575487819418e-06,
      "loss": 0.1982,
      "step": 33690
    },
    {
      "epoch": 3.3144009244916526,
      "grad_norm": 0.3914591372013092,
      "learning_rate": 6.809601398879308e-06,
      "loss": 0.173,
      "step": 33700
    },
    {
      "epoch": 3.315384426249662,
      "grad_norm": 0.491703063249588,
      "learning_rate": 6.8056273099391965e-06,
      "loss": 0.2151,
      "step": 33710
    },
    {
      "epoch": 3.316367928007671,
      "grad_norm": 1.746738076210022,
      "learning_rate": 6.801653220999086e-06,
      "loss": 0.2345,
      "step": 33720
    },
    {
      "epoch": 3.3173514297656808,
      "grad_norm": 1.026167869567871,
      "learning_rate": 6.797679132058977e-06,
      "loss": 0.2082,
      "step": 33730
    },
    {
      "epoch": 3.3183349315236903,
      "grad_norm": 1.7243843078613281,
      "learning_rate": 6.7937050431188654e-06,
      "loss": 0.2422,
      "step": 33740
    },
    {
      "epoch": 3.3193184332816994,
      "grad_norm": 2.0129072666168213,
      "learning_rate": 6.789730954178755e-06,
      "loss": 0.2314,
      "step": 33750
    },
    {
      "epoch": 3.320301935039709,
      "grad_norm": 1.2712618112564087,
      "learning_rate": 6.785756865238644e-06,
      "loss": 0.2042,
      "step": 33760
    },
    {
      "epoch": 3.3212854367977185,
      "grad_norm": 0.9794314503669739,
      "learning_rate": 6.7817827762985344e-06,
      "loss": 0.2315,
      "step": 33770
    },
    {
      "epoch": 3.3222689385557276,
      "grad_norm": 0.66265869140625,
      "learning_rate": 6.777808687358423e-06,
      "loss": 0.1541,
      "step": 33780
    },
    {
      "epoch": 3.323252440313737,
      "grad_norm": 0.7355524301528931,
      "learning_rate": 6.773834598418313e-06,
      "loss": 0.1472,
      "step": 33790
    },
    {
      "epoch": 3.3242359420717467,
      "grad_norm": 0.4393594563007355,
      "learning_rate": 6.769860509478203e-06,
      "loss": 0.2169,
      "step": 33800
    },
    {
      "epoch": 3.3252194438297558,
      "grad_norm": 0.4566223621368408,
      "learning_rate": 6.765886420538092e-06,
      "loss": 0.2252,
      "step": 33810
    },
    {
      "epoch": 3.3262029455877653,
      "grad_norm": 0.6023932099342346,
      "learning_rate": 6.761912331597982e-06,
      "loss": 0.129,
      "step": 33820
    },
    {
      "epoch": 3.327186447345775,
      "grad_norm": 0.10506648570299149,
      "learning_rate": 6.757938242657871e-06,
      "loss": 0.1242,
      "step": 33830
    },
    {
      "epoch": 3.328169949103784,
      "grad_norm": 1.2360717058181763,
      "learning_rate": 6.75396415371776e-06,
      "loss": 0.2681,
      "step": 33840
    },
    {
      "epoch": 3.3291534508617935,
      "grad_norm": 0.9308496713638306,
      "learning_rate": 6.749990064777651e-06,
      "loss": 0.1293,
      "step": 33850
    },
    {
      "epoch": 3.330136952619803,
      "grad_norm": 0.4173755645751953,
      "learning_rate": 6.74601597583754e-06,
      "loss": 0.2014,
      "step": 33860
    },
    {
      "epoch": 3.331120454377812,
      "grad_norm": 4.138999938964844,
      "learning_rate": 6.742041886897429e-06,
      "loss": 0.2215,
      "step": 33870
    },
    {
      "epoch": 3.3321039561358217,
      "grad_norm": 0.23688192665576935,
      "learning_rate": 6.738067797957318e-06,
      "loss": 0.245,
      "step": 33880
    },
    {
      "epoch": 3.333087457893831,
      "grad_norm": 0.947866678237915,
      "learning_rate": 6.734093709017209e-06,
      "loss": 0.2418,
      "step": 33890
    },
    {
      "epoch": 3.3340709596518403,
      "grad_norm": 1.1110749244689941,
      "learning_rate": 6.7301196200770975e-06,
      "loss": 0.2828,
      "step": 33900
    },
    {
      "epoch": 3.33505446140985,
      "grad_norm": 1.053632140159607,
      "learning_rate": 6.726145531136987e-06,
      "loss": 0.1867,
      "step": 33910
    },
    {
      "epoch": 3.3360379631678594,
      "grad_norm": 0.5738980174064636,
      "learning_rate": 6.722171442196877e-06,
      "loss": 0.1942,
      "step": 33920
    },
    {
      "epoch": 3.3370214649258685,
      "grad_norm": 0.07501865923404694,
      "learning_rate": 6.7181973532567665e-06,
      "loss": 0.1547,
      "step": 33930
    },
    {
      "epoch": 3.338004966683878,
      "grad_norm": 0.46395015716552734,
      "learning_rate": 6.714223264316656e-06,
      "loss": 0.2232,
      "step": 33940
    },
    {
      "epoch": 3.3389884684418876,
      "grad_norm": 0.21362808346748352,
      "learning_rate": 6.710249175376545e-06,
      "loss": 0.2594,
      "step": 33950
    },
    {
      "epoch": 3.3399719701998967,
      "grad_norm": 1.4730660915374756,
      "learning_rate": 6.706275086436435e-06,
      "loss": 0.1954,
      "step": 33960
    },
    {
      "epoch": 3.340955471957906,
      "grad_norm": 1.2356129884719849,
      "learning_rate": 6.702300997496325e-06,
      "loss": 0.2185,
      "step": 33970
    },
    {
      "epoch": 3.3419389737159158,
      "grad_norm": 1.5479633808135986,
      "learning_rate": 6.698326908556214e-06,
      "loss": 0.2533,
      "step": 33980
    },
    {
      "epoch": 3.342922475473925,
      "grad_norm": 1.2569940090179443,
      "learning_rate": 6.694352819616104e-06,
      "loss": 0.2133,
      "step": 33990
    },
    {
      "epoch": 3.3439059772319344,
      "grad_norm": 0.516693115234375,
      "learning_rate": 6.6903787306759925e-06,
      "loss": 0.2356,
      "step": 34000
    },
    {
      "epoch": 3.3439059772319344,
      "eval_loss": 0.11839893460273743,
      "eval_runtime": 17.6612,
      "eval_samples_per_second": 2.831,
      "eval_steps_per_second": 1.416,
      "step": 34000
    },
    {
      "epoch": 3.344889478989944,
      "grad_norm": 0.4600711762905121,
      "learning_rate": 6.686404641735883e-06,
      "loss": 0.1115,
      "step": 34010
    },
    {
      "epoch": 3.345872980747953,
      "grad_norm": 0.40922555327415466,
      "learning_rate": 6.682430552795772e-06,
      "loss": 0.2602,
      "step": 34020
    },
    {
      "epoch": 3.3468564825059626,
      "grad_norm": 0.4795723259449005,
      "learning_rate": 6.6784564638556615e-06,
      "loss": 0.1489,
      "step": 34030
    },
    {
      "epoch": 3.347839984263972,
      "grad_norm": 0.5202747583389282,
      "learning_rate": 6.674482374915551e-06,
      "loss": 0.1751,
      "step": 34040
    },
    {
      "epoch": 3.348823486021981,
      "grad_norm": 0.8103330135345459,
      "learning_rate": 6.670508285975441e-06,
      "loss": 0.1866,
      "step": 34050
    },
    {
      "epoch": 3.3498069877799908,
      "grad_norm": 0.6577498316764832,
      "learning_rate": 6.6665341970353305e-06,
      "loss": 0.1945,
      "step": 34060
    },
    {
      "epoch": 3.3507904895380003,
      "grad_norm": 0.5543909668922424,
      "learning_rate": 6.662560108095219e-06,
      "loss": 0.0748,
      "step": 34070
    },
    {
      "epoch": 3.3517739912960094,
      "grad_norm": 0.3950757682323456,
      "learning_rate": 6.658586019155109e-06,
      "loss": 0.2112,
      "step": 34080
    },
    {
      "epoch": 3.352757493054019,
      "grad_norm": 0.6226676106452942,
      "learning_rate": 6.6546119302149995e-06,
      "loss": 0.2478,
      "step": 34090
    },
    {
      "epoch": 3.3537409948120285,
      "grad_norm": 0.7202355265617371,
      "learning_rate": 6.650637841274888e-06,
      "loss": 0.2961,
      "step": 34100
    },
    {
      "epoch": 3.3547244965700376,
      "grad_norm": 2.388300657272339,
      "learning_rate": 6.646663752334778e-06,
      "loss": 0.183,
      "step": 34110
    },
    {
      "epoch": 3.355707998328047,
      "grad_norm": 0.7886066436767578,
      "learning_rate": 6.642689663394667e-06,
      "loss": 0.2087,
      "step": 34120
    },
    {
      "epoch": 3.356691500086056,
      "grad_norm": 0.3806540369987488,
      "learning_rate": 6.638715574454557e-06,
      "loss": 0.1824,
      "step": 34130
    },
    {
      "epoch": 3.3576750018440658,
      "grad_norm": 2.428232431411743,
      "learning_rate": 6.634741485514446e-06,
      "loss": 0.2822,
      "step": 34140
    },
    {
      "epoch": 3.3586585036020753,
      "grad_norm": 0.47671598196029663,
      "learning_rate": 6.630767396574336e-06,
      "loss": 0.2101,
      "step": 34150
    },
    {
      "epoch": 3.3596420053600844,
      "grad_norm": 1.0764350891113281,
      "learning_rate": 6.6267933076342254e-06,
      "loss": 0.2439,
      "step": 34160
    },
    {
      "epoch": 3.360625507118094,
      "grad_norm": 0.19563129544258118,
      "learning_rate": 6.622819218694115e-06,
      "loss": 0.2158,
      "step": 34170
    },
    {
      "epoch": 3.3616090088761035,
      "grad_norm": 0.4823220372200012,
      "learning_rate": 6.618845129754005e-06,
      "loss": 0.1742,
      "step": 34180
    },
    {
      "epoch": 3.3625925106341126,
      "grad_norm": 1.3417798280715942,
      "learning_rate": 6.614871040813894e-06,
      "loss": 0.2537,
      "step": 34190
    },
    {
      "epoch": 3.363576012392122,
      "grad_norm": 1.0107239484786987,
      "learning_rate": 6.610896951873783e-06,
      "loss": 0.176,
      "step": 34200
    },
    {
      "epoch": 3.3645595141501317,
      "grad_norm": 0.7224860191345215,
      "learning_rate": 6.606922862933674e-06,
      "loss": 0.2095,
      "step": 34210
    },
    {
      "epoch": 3.3655430159081408,
      "grad_norm": 0.8758500218391418,
      "learning_rate": 6.602948773993563e-06,
      "loss": 0.1393,
      "step": 34220
    },
    {
      "epoch": 3.3665265176661503,
      "grad_norm": 0.7904019355773926,
      "learning_rate": 6.598974685053452e-06,
      "loss": 0.1872,
      "step": 34230
    },
    {
      "epoch": 3.36751001942416,
      "grad_norm": 1.1942181587219238,
      "learning_rate": 6.595000596113341e-06,
      "loss": 0.2817,
      "step": 34240
    },
    {
      "epoch": 3.368493521182169,
      "grad_norm": 0.9850848913192749,
      "learning_rate": 6.5910265071732316e-06,
      "loss": 0.177,
      "step": 34250
    },
    {
      "epoch": 3.3694770229401785,
      "grad_norm": 1.3652971982955933,
      "learning_rate": 6.58705241823312e-06,
      "loss": 0.232,
      "step": 34260
    },
    {
      "epoch": 3.370460524698188,
      "grad_norm": 0.37273120880126953,
      "learning_rate": 6.58307832929301e-06,
      "loss": 0.1928,
      "step": 34270
    },
    {
      "epoch": 3.371444026456197,
      "grad_norm": 1.5884888172149658,
      "learning_rate": 6.5791042403529e-06,
      "loss": 0.0956,
      "step": 34280
    },
    {
      "epoch": 3.3724275282142067,
      "grad_norm": 1.9908277988433838,
      "learning_rate": 6.575130151412789e-06,
      "loss": 0.2205,
      "step": 34290
    },
    {
      "epoch": 3.373411029972216,
      "grad_norm": 0.3747865855693817,
      "learning_rate": 6.571156062472679e-06,
      "loss": 0.1879,
      "step": 34300
    },
    {
      "epoch": 3.3743945317302253,
      "grad_norm": 0.24655213952064514,
      "learning_rate": 6.567181973532568e-06,
      "loss": 0.2002,
      "step": 34310
    },
    {
      "epoch": 3.375378033488235,
      "grad_norm": 0.8035260438919067,
      "learning_rate": 6.5632078845924575e-06,
      "loss": 0.2027,
      "step": 34320
    },
    {
      "epoch": 3.3763615352462444,
      "grad_norm": 0.25017067790031433,
      "learning_rate": 6.559233795652348e-06,
      "loss": 0.2793,
      "step": 34330
    },
    {
      "epoch": 3.3773450370042535,
      "grad_norm": 0.13351485133171082,
      "learning_rate": 6.555259706712237e-06,
      "loss": 0.1746,
      "step": 34340
    },
    {
      "epoch": 3.378328538762263,
      "grad_norm": 0.4170202314853668,
      "learning_rate": 6.5512856177721265e-06,
      "loss": 0.1552,
      "step": 34350
    },
    {
      "epoch": 3.3793120405202726,
      "grad_norm": 0.25968262553215027,
      "learning_rate": 6.547311528832015e-06,
      "loss": 0.1619,
      "step": 34360
    },
    {
      "epoch": 3.3802955422782817,
      "grad_norm": 2.428452253341675,
      "learning_rate": 6.543337439891906e-06,
      "loss": 0.1165,
      "step": 34370
    },
    {
      "epoch": 3.381279044036291,
      "grad_norm": 1.7165864706039429,
      "learning_rate": 6.539363350951795e-06,
      "loss": 0.1918,
      "step": 34380
    },
    {
      "epoch": 3.3822625457943007,
      "grad_norm": 0.634429931640625,
      "learning_rate": 6.535389262011684e-06,
      "loss": 0.1812,
      "step": 34390
    },
    {
      "epoch": 3.38324604755231,
      "grad_norm": 1.2559707164764404,
      "learning_rate": 6.531415173071573e-06,
      "loss": 0.3113,
      "step": 34400
    },
    {
      "epoch": 3.3842295493103194,
      "grad_norm": 0.8117685914039612,
      "learning_rate": 6.527441084131464e-06,
      "loss": 0.1822,
      "step": 34410
    },
    {
      "epoch": 3.385213051068329,
      "grad_norm": 1.5536867380142212,
      "learning_rate": 6.523466995191353e-06,
      "loss": 0.1113,
      "step": 34420
    },
    {
      "epoch": 3.386196552826338,
      "grad_norm": 1.0924869775772095,
      "learning_rate": 6.519492906251242e-06,
      "loss": 0.2487,
      "step": 34430
    },
    {
      "epoch": 3.3871800545843476,
      "grad_norm": 0.7709922790527344,
      "learning_rate": 6.515518817311132e-06,
      "loss": 0.2193,
      "step": 34440
    },
    {
      "epoch": 3.388163556342357,
      "grad_norm": 1.841170310974121,
      "learning_rate": 6.511544728371022e-06,
      "loss": 0.1149,
      "step": 34450
    },
    {
      "epoch": 3.389147058100366,
      "grad_norm": 0.4802115559577942,
      "learning_rate": 6.507570639430911e-06,
      "loss": 0.2525,
      "step": 34460
    },
    {
      "epoch": 3.3901305598583757,
      "grad_norm": 0.5731269717216492,
      "learning_rate": 6.503596550490801e-06,
      "loss": 0.1602,
      "step": 34470
    },
    {
      "epoch": 3.3911140616163853,
      "grad_norm": 0.488396018743515,
      "learning_rate": 6.49962246155069e-06,
      "loss": 0.1671,
      "step": 34480
    },
    {
      "epoch": 3.3920975633743944,
      "grad_norm": 0.9870609045028687,
      "learning_rate": 6.495648372610579e-06,
      "loss": 0.17,
      "step": 34490
    },
    {
      "epoch": 3.393081065132404,
      "grad_norm": 0.36211255192756653,
      "learning_rate": 6.491674283670469e-06,
      "loss": 0.2352,
      "step": 34500
    },
    {
      "epoch": 3.393081065132404,
      "eval_loss": 0.11643476784229279,
      "eval_runtime": 16.9274,
      "eval_samples_per_second": 2.954,
      "eval_steps_per_second": 1.477,
      "step": 34500
    },
    {
      "epoch": 3.3940645668904135,
      "grad_norm": 0.9025580286979675,
      "learning_rate": 6.487700194730359e-06,
      "loss": 0.2056,
      "step": 34510
    },
    {
      "epoch": 3.3950480686484226,
      "grad_norm": 1.125788688659668,
      "learning_rate": 6.4837261057902474e-06,
      "loss": 0.1919,
      "step": 34520
    },
    {
      "epoch": 3.396031570406432,
      "grad_norm": 0.9839863181114197,
      "learning_rate": 6.479752016850137e-06,
      "loss": 0.1315,
      "step": 34530
    },
    {
      "epoch": 3.3970150721644417,
      "grad_norm": 0.07589655369520187,
      "learning_rate": 6.475777927910028e-06,
      "loss": 0.1146,
      "step": 34540
    },
    {
      "epoch": 3.3979985739224507,
      "grad_norm": 2.114081859588623,
      "learning_rate": 6.4718038389699164e-06,
      "loss": 0.116,
      "step": 34550
    },
    {
      "epoch": 3.3989820756804603,
      "grad_norm": 0.7543653845787048,
      "learning_rate": 6.467829750029806e-06,
      "loss": 0.167,
      "step": 34560
    },
    {
      "epoch": 3.39996557743847,
      "grad_norm": 0.4026939272880554,
      "learning_rate": 6.463855661089695e-06,
      "loss": 0.181,
      "step": 34570
    },
    {
      "epoch": 3.400949079196479,
      "grad_norm": 2.032933473587036,
      "learning_rate": 6.4598815721495854e-06,
      "loss": 0.0675,
      "step": 34580
    },
    {
      "epoch": 3.4019325809544885,
      "grad_norm": 1.10984206199646,
      "learning_rate": 6.455907483209475e-06,
      "loss": 0.23,
      "step": 34590
    },
    {
      "epoch": 3.402916082712498,
      "grad_norm": 0.7900817394256592,
      "learning_rate": 6.451933394269364e-06,
      "loss": 0.188,
      "step": 34600
    },
    {
      "epoch": 3.403899584470507,
      "grad_norm": 0.9765810370445251,
      "learning_rate": 6.4479593053292536e-06,
      "loss": 0.1501,
      "step": 34610
    },
    {
      "epoch": 3.4048830862285167,
      "grad_norm": 0.9752830862998962,
      "learning_rate": 6.443985216389143e-06,
      "loss": 0.2133,
      "step": 34620
    },
    {
      "epoch": 3.405866587986526,
      "grad_norm": 2.466362237930298,
      "learning_rate": 6.440011127449033e-06,
      "loss": 0.2199,
      "step": 34630
    },
    {
      "epoch": 3.4068500897445353,
      "grad_norm": 1.1002997159957886,
      "learning_rate": 6.436037038508922e-06,
      "loss": 0.2043,
      "step": 34640
    },
    {
      "epoch": 3.407833591502545,
      "grad_norm": 2.216463327407837,
      "learning_rate": 6.432062949568811e-06,
      "loss": 0.2458,
      "step": 34650
    },
    {
      "epoch": 3.4088170932605544,
      "grad_norm": 0.41991540789604187,
      "learning_rate": 6.428088860628702e-06,
      "loss": 0.1945,
      "step": 34660
    },
    {
      "epoch": 3.4098005950185635,
      "grad_norm": 0.6682888865470886,
      "learning_rate": 6.424114771688591e-06,
      "loss": 0.218,
      "step": 34670
    },
    {
      "epoch": 3.410784096776573,
      "grad_norm": 1.0558764934539795,
      "learning_rate": 6.42014068274848e-06,
      "loss": 0.1781,
      "step": 34680
    },
    {
      "epoch": 3.4117675985345826,
      "grad_norm": 0.5618786811828613,
      "learning_rate": 6.416166593808369e-06,
      "loss": 0.2492,
      "step": 34690
    },
    {
      "epoch": 3.4127511002925917,
      "grad_norm": 0.3969118893146515,
      "learning_rate": 6.41219250486826e-06,
      "loss": 0.1723,
      "step": 34700
    },
    {
      "epoch": 3.413734602050601,
      "grad_norm": 0.7788593173027039,
      "learning_rate": 6.408218415928149e-06,
      "loss": 0.1819,
      "step": 34710
    },
    {
      "epoch": 3.4147181038086107,
      "grad_norm": 0.6339419484138489,
      "learning_rate": 6.404244326988038e-06,
      "loss": 0.1933,
      "step": 34720
    },
    {
      "epoch": 3.41570160556662,
      "grad_norm": 3.0357329845428467,
      "learning_rate": 6.400270238047928e-06,
      "loss": 0.2384,
      "step": 34730
    },
    {
      "epoch": 3.4166851073246294,
      "grad_norm": 0.7546097040176392,
      "learning_rate": 6.3962961491078175e-06,
      "loss": 0.1562,
      "step": 34740
    },
    {
      "epoch": 3.417668609082639,
      "grad_norm": 0.2636072635650635,
      "learning_rate": 6.392322060167707e-06,
      "loss": 0.1887,
      "step": 34750
    },
    {
      "epoch": 3.418652110840648,
      "grad_norm": 0.1599857211112976,
      "learning_rate": 6.388347971227596e-06,
      "loss": 0.1095,
      "step": 34760
    },
    {
      "epoch": 3.4196356125986576,
      "grad_norm": 0.12882690131664276,
      "learning_rate": 6.384373882287486e-06,
      "loss": 0.1667,
      "step": 34770
    },
    {
      "epoch": 3.420619114356667,
      "grad_norm": 1.473196268081665,
      "learning_rate": 6.380399793347376e-06,
      "loss": 0.2108,
      "step": 34780
    },
    {
      "epoch": 3.421602616114676,
      "grad_norm": 0.6438500881195068,
      "learning_rate": 6.376425704407265e-06,
      "loss": 0.2418,
      "step": 34790
    },
    {
      "epoch": 3.4225861178726857,
      "grad_norm": 1.5048677921295166,
      "learning_rate": 6.372451615467155e-06,
      "loss": 0.2191,
      "step": 34800
    },
    {
      "epoch": 3.4235696196306953,
      "grad_norm": 1.208437442779541,
      "learning_rate": 6.3684775265270435e-06,
      "loss": 0.1866,
      "step": 34810
    },
    {
      "epoch": 3.4245531213887044,
      "grad_norm": 0.35769960284233093,
      "learning_rate": 6.364503437586934e-06,
      "loss": 0.1976,
      "step": 34820
    },
    {
      "epoch": 3.425536623146714,
      "grad_norm": 0.31596699357032776,
      "learning_rate": 6.360529348646824e-06,
      "loss": 0.1637,
      "step": 34830
    },
    {
      "epoch": 3.426520124904723,
      "grad_norm": 0.6313391327857971,
      "learning_rate": 6.3565552597067125e-06,
      "loss": 0.1783,
      "step": 34840
    },
    {
      "epoch": 3.4275036266627326,
      "grad_norm": 0.5374031662940979,
      "learning_rate": 6.352581170766602e-06,
      "loss": 0.2105,
      "step": 34850
    },
    {
      "epoch": 3.428487128420742,
      "grad_norm": 1.1502714157104492,
      "learning_rate": 6.348607081826492e-06,
      "loss": 0.1998,
      "step": 34860
    },
    {
      "epoch": 3.429470630178751,
      "grad_norm": 0.2589196562767029,
      "learning_rate": 6.3446329928863815e-06,
      "loss": 0.2485,
      "step": 34870
    },
    {
      "epoch": 3.4304541319367607,
      "grad_norm": 0.5239776968955994,
      "learning_rate": 6.34065890394627e-06,
      "loss": 0.2309,
      "step": 34880
    },
    {
      "epoch": 3.4314376336947703,
      "grad_norm": 0.4215831458568573,
      "learning_rate": 6.33668481500616e-06,
      "loss": 0.2086,
      "step": 34890
    },
    {
      "epoch": 3.4324211354527794,
      "grad_norm": 0.2826988697052002,
      "learning_rate": 6.3327107260660505e-06,
      "loss": 0.1729,
      "step": 34900
    },
    {
      "epoch": 3.433404637210789,
      "grad_norm": 1.7079347372055054,
      "learning_rate": 6.328736637125939e-06,
      "loss": 0.224,
      "step": 34910
    },
    {
      "epoch": 3.4343881389687985,
      "grad_norm": 2.6849637031555176,
      "learning_rate": 6.324762548185829e-06,
      "loss": 0.2203,
      "step": 34920
    },
    {
      "epoch": 3.4353716407268076,
      "grad_norm": 0.39513906836509705,
      "learning_rate": 6.320788459245718e-06,
      "loss": 0.2422,
      "step": 34930
    },
    {
      "epoch": 3.436355142484817,
      "grad_norm": 1.4758193492889404,
      "learning_rate": 6.316814370305608e-06,
      "loss": 0.1015,
      "step": 34940
    },
    {
      "epoch": 3.4373386442428266,
      "grad_norm": 0.9000169038772583,
      "learning_rate": 6.312840281365498e-06,
      "loss": 0.1771,
      "step": 34950
    },
    {
      "epoch": 3.4383221460008357,
      "grad_norm": 1.5330618619918823,
      "learning_rate": 6.308866192425387e-06,
      "loss": 0.2195,
      "step": 34960
    },
    {
      "epoch": 3.4393056477588453,
      "grad_norm": 0.6581733226776123,
      "learning_rate": 6.304892103485276e-06,
      "loss": 0.2737,
      "step": 34970
    },
    {
      "epoch": 3.440289149516855,
      "grad_norm": 0.494530588388443,
      "learning_rate": 6.300918014545166e-06,
      "loss": 0.2535,
      "step": 34980
    },
    {
      "epoch": 3.441272651274864,
      "grad_norm": 2.8313052654266357,
      "learning_rate": 6.296943925605056e-06,
      "loss": 0.2243,
      "step": 34990
    },
    {
      "epoch": 3.4422561530328735,
      "grad_norm": 0.35256826877593994,
      "learning_rate": 6.2929698366649446e-06,
      "loss": 0.1598,
      "step": 35000
    },
    {
      "epoch": 3.4422561530328735,
      "eval_loss": 0.11682377755641937,
      "eval_runtime": 19.7716,
      "eval_samples_per_second": 2.529,
      "eval_steps_per_second": 1.264,
      "step": 35000
    },
    {
      "epoch": 3.443239654790883,
      "grad_norm": 0.9331240057945251,
      "learning_rate": 6.288995747724834e-06,
      "loss": 0.2248,
      "step": 35010
    },
    {
      "epoch": 3.444223156548892,
      "grad_norm": 0.9498600363731384,
      "learning_rate": 6.285021658784725e-06,
      "loss": 0.2281,
      "step": 35020
    },
    {
      "epoch": 3.4452066583069016,
      "grad_norm": 1.1861779689788818,
      "learning_rate": 6.2810475698446136e-06,
      "loss": 0.2029,
      "step": 35030
    },
    {
      "epoch": 3.446190160064911,
      "grad_norm": 0.6265971064567566,
      "learning_rate": 6.277073480904503e-06,
      "loss": 0.1566,
      "step": 35040
    },
    {
      "epoch": 3.4471736618229203,
      "grad_norm": 0.1985936313867569,
      "learning_rate": 6.273099391964392e-06,
      "loss": 0.1737,
      "step": 35050
    },
    {
      "epoch": 3.44815716358093,
      "grad_norm": 0.2556389272212982,
      "learning_rate": 6.2691253030242826e-06,
      "loss": 0.1608,
      "step": 35060
    },
    {
      "epoch": 3.4491406653389394,
      "grad_norm": 0.5306591987609863,
      "learning_rate": 6.265151214084172e-06,
      "loss": 0.2038,
      "step": 35070
    },
    {
      "epoch": 3.4501241670969485,
      "grad_norm": 0.7322729825973511,
      "learning_rate": 6.261177125144061e-06,
      "loss": 0.2146,
      "step": 35080
    },
    {
      "epoch": 3.451107668854958,
      "grad_norm": 1.0988006591796875,
      "learning_rate": 6.257203036203951e-06,
      "loss": 0.2701,
      "step": 35090
    },
    {
      "epoch": 3.4520911706129676,
      "grad_norm": 0.382568359375,
      "learning_rate": 6.25322894726384e-06,
      "loss": 0.1788,
      "step": 35100
    },
    {
      "epoch": 3.4530746723709766,
      "grad_norm": 1.1923779249191284,
      "learning_rate": 6.24925485832373e-06,
      "loss": 0.2115,
      "step": 35110
    },
    {
      "epoch": 3.454058174128986,
      "grad_norm": 0.3326428234577179,
      "learning_rate": 6.245280769383619e-06,
      "loss": 0.1822,
      "step": 35120
    },
    {
      "epoch": 3.4550416758869957,
      "grad_norm": 0.5573846101760864,
      "learning_rate": 6.2413066804435085e-06,
      "loss": 0.2395,
      "step": 35130
    },
    {
      "epoch": 3.456025177645005,
      "grad_norm": 0.848188042640686,
      "learning_rate": 6.237332591503399e-06,
      "loss": 0.1655,
      "step": 35140
    },
    {
      "epoch": 3.4570086794030144,
      "grad_norm": 0.7337232232093811,
      "learning_rate": 6.233358502563288e-06,
      "loss": 0.2413,
      "step": 35150
    },
    {
      "epoch": 3.457992181161024,
      "grad_norm": 0.5480116605758667,
      "learning_rate": 6.2293844136231775e-06,
      "loss": 0.1676,
      "step": 35160
    },
    {
      "epoch": 3.458975682919033,
      "grad_norm": 0.938613772392273,
      "learning_rate": 6.225410324683066e-06,
      "loss": 0.1292,
      "step": 35170
    },
    {
      "epoch": 3.4599591846770426,
      "grad_norm": 0.7171252369880676,
      "learning_rate": 6.221436235742957e-06,
      "loss": 0.2464,
      "step": 35180
    },
    {
      "epoch": 3.460942686435052,
      "grad_norm": 0.6132894158363342,
      "learning_rate": 6.2174621468028465e-06,
      "loss": 0.2727,
      "step": 35190
    },
    {
      "epoch": 3.461926188193061,
      "grad_norm": 4.07462215423584,
      "learning_rate": 6.213488057862735e-06,
      "loss": 0.1068,
      "step": 35200
    },
    {
      "epoch": 3.4629096899510707,
      "grad_norm": 0.724358856678009,
      "learning_rate": 6.209513968922625e-06,
      "loss": 0.1658,
      "step": 35210
    },
    {
      "epoch": 3.4638931917090803,
      "grad_norm": 0.09326642751693726,
      "learning_rate": 6.205539879982515e-06,
      "loss": 0.0746,
      "step": 35220
    },
    {
      "epoch": 3.4648766934670894,
      "grad_norm": 4.169434070587158,
      "learning_rate": 6.201565791042404e-06,
      "loss": 0.173,
      "step": 35230
    },
    {
      "epoch": 3.465860195225099,
      "grad_norm": 1.4873085021972656,
      "learning_rate": 6.197591702102293e-06,
      "loss": 0.2532,
      "step": 35240
    },
    {
      "epoch": 3.4668436969831085,
      "grad_norm": 0.22203786671161652,
      "learning_rate": 6.193617613162183e-06,
      "loss": 0.1647,
      "step": 35250
    },
    {
      "epoch": 3.4678271987411176,
      "grad_norm": 0.8890891075134277,
      "learning_rate": 6.1896435242220725e-06,
      "loss": 0.1263,
      "step": 35260
    },
    {
      "epoch": 3.468810700499127,
      "grad_norm": 1.1431549787521362,
      "learning_rate": 6.185669435281962e-06,
      "loss": 0.2693,
      "step": 35270
    },
    {
      "epoch": 3.4697942022571366,
      "grad_norm": 1.133975625038147,
      "learning_rate": 6.181695346341852e-06,
      "loss": 0.2801,
      "step": 35280
    },
    {
      "epoch": 3.4707777040151457,
      "grad_norm": 0.5976429581642151,
      "learning_rate": 6.177721257401741e-06,
      "loss": 0.1671,
      "step": 35290
    },
    {
      "epoch": 3.4717612057731553,
      "grad_norm": 1.3519940376281738,
      "learning_rate": 6.17374716846163e-06,
      "loss": 0.3204,
      "step": 35300
    },
    {
      "epoch": 3.472744707531165,
      "grad_norm": 0.4874255359172821,
      "learning_rate": 6.169773079521521e-06,
      "loss": 0.1915,
      "step": 35310
    },
    {
      "epoch": 3.473728209289174,
      "grad_norm": 2.0294809341430664,
      "learning_rate": 6.16579899058141e-06,
      "loss": 0.1796,
      "step": 35320
    },
    {
      "epoch": 3.4747117110471835,
      "grad_norm": 0.573607861995697,
      "learning_rate": 6.161824901641299e-06,
      "loss": 0.1704,
      "step": 35330
    },
    {
      "epoch": 3.475695212805193,
      "grad_norm": 1.4688594341278076,
      "learning_rate": 6.157850812701188e-06,
      "loss": 0.2388,
      "step": 35340
    },
    {
      "epoch": 3.476678714563202,
      "grad_norm": 1.3949116468429565,
      "learning_rate": 6.153876723761079e-06,
      "loss": 0.2454,
      "step": 35350
    },
    {
      "epoch": 3.4776622163212116,
      "grad_norm": 1.074675440788269,
      "learning_rate": 6.149902634820967e-06,
      "loss": 0.1916,
      "step": 35360
    },
    {
      "epoch": 3.478645718079221,
      "grad_norm": 1.8467960357666016,
      "learning_rate": 6.145928545880857e-06,
      "loss": 0.2278,
      "step": 35370
    },
    {
      "epoch": 3.4796292198372303,
      "grad_norm": 0.5333929061889648,
      "learning_rate": 6.141954456940746e-06,
      "loss": 0.2142,
      "step": 35380
    },
    {
      "epoch": 3.48061272159524,
      "grad_norm": 0.37479594349861145,
      "learning_rate": 6.137980368000636e-06,
      "loss": 0.188,
      "step": 35390
    },
    {
      "epoch": 3.4815962233532494,
      "grad_norm": 0.4170948266983032,
      "learning_rate": 6.134006279060526e-06,
      "loss": 0.104,
      "step": 35400
    },
    {
      "epoch": 3.4825797251112585,
      "grad_norm": 1.5275107622146606,
      "learning_rate": 6.130032190120415e-06,
      "loss": 0.2214,
      "step": 35410
    },
    {
      "epoch": 3.483563226869268,
      "grad_norm": 1.325067162513733,
      "learning_rate": 6.1260581011803046e-06,
      "loss": 0.2264,
      "step": 35420
    },
    {
      "epoch": 3.4845467286272775,
      "grad_norm": 1.20541250705719,
      "learning_rate": 6.122084012240195e-06,
      "loss": 0.2887,
      "step": 35430
    },
    {
      "epoch": 3.4855302303852866,
      "grad_norm": 0.8210470676422119,
      "learning_rate": 6.118109923300084e-06,
      "loss": 0.1476,
      "step": 35440
    },
    {
      "epoch": 3.486513732143296,
      "grad_norm": 1.8403013944625854,
      "learning_rate": 6.1141358343599736e-06,
      "loss": 0.157,
      "step": 35450
    },
    {
      "epoch": 3.4874972339013057,
      "grad_norm": 0.9750491976737976,
      "learning_rate": 6.110161745419862e-06,
      "loss": 0.2567,
      "step": 35460
    },
    {
      "epoch": 3.488480735659315,
      "grad_norm": 0.4504006505012512,
      "learning_rate": 6.106187656479753e-06,
      "loss": 0.2618,
      "step": 35470
    },
    {
      "epoch": 3.4894642374173244,
      "grad_norm": 0.8415562510490417,
      "learning_rate": 6.102213567539642e-06,
      "loss": 0.1635,
      "step": 35480
    },
    {
      "epoch": 3.490447739175334,
      "grad_norm": 2.1266133785247803,
      "learning_rate": 6.098239478599531e-06,
      "loss": 0.2345,
      "step": 35490
    },
    {
      "epoch": 3.491431240933343,
      "grad_norm": 1.2690298557281494,
      "learning_rate": 6.09426538965942e-06,
      "loss": 0.1781,
      "step": 35500
    },
    {
      "epoch": 3.491431240933343,
      "eval_loss": 0.11791013926267624,
      "eval_runtime": 17.481,
      "eval_samples_per_second": 2.86,
      "eval_steps_per_second": 1.43,
      "step": 35500
    },
    {
      "epoch": 3.4924147426913525,
      "grad_norm": 1.0220181941986084,
      "learning_rate": 6.090291300719311e-06,
      "loss": 0.0995,
      "step": 35510
    },
    {
      "epoch": 3.493398244449362,
      "grad_norm": 0.8852249979972839,
      "learning_rate": 6.0863172117792e-06,
      "loss": 0.2229,
      "step": 35520
    },
    {
      "epoch": 3.494381746207371,
      "grad_norm": 0.6984592080116272,
      "learning_rate": 6.082343122839089e-06,
      "loss": 0.2568,
      "step": 35530
    },
    {
      "epoch": 3.4953652479653807,
      "grad_norm": 0.2515038549900055,
      "learning_rate": 6.078369033898979e-06,
      "loss": 0.2064,
      "step": 35540
    },
    {
      "epoch": 3.4963487497233903,
      "grad_norm": 0.8225836753845215,
      "learning_rate": 6.074394944958869e-06,
      "loss": 0.2517,
      "step": 35550
    },
    {
      "epoch": 3.4973322514813994,
      "grad_norm": 0.27752599120140076,
      "learning_rate": 6.070420856018758e-06,
      "loss": 0.0464,
      "step": 35560
    },
    {
      "epoch": 3.498315753239409,
      "grad_norm": 0.038399215787649155,
      "learning_rate": 6.066446767078648e-06,
      "loss": 0.1506,
      "step": 35570
    },
    {
      "epoch": 3.4992992549974185,
      "grad_norm": 0.7299541234970093,
      "learning_rate": 6.062472678138537e-06,
      "loss": 0.2598,
      "step": 35580
    },
    {
      "epoch": 3.5002827567554275,
      "grad_norm": 1.7209151983261108,
      "learning_rate": 6.058498589198427e-06,
      "loss": 0.188,
      "step": 35590
    },
    {
      "epoch": 3.501266258513437,
      "grad_norm": 0.3954479694366455,
      "learning_rate": 6.054524500258316e-06,
      "loss": 0.1885,
      "step": 35600
    },
    {
      "epoch": 3.5022497602714466,
      "grad_norm": 1.3695343732833862,
      "learning_rate": 6.050550411318206e-06,
      "loss": 0.0878,
      "step": 35610
    },
    {
      "epoch": 3.5032332620294557,
      "grad_norm": 0.9690384864807129,
      "learning_rate": 6.0465763223780945e-06,
      "loss": 0.1552,
      "step": 35620
    },
    {
      "epoch": 3.5042167637874653,
      "grad_norm": 1.019343614578247,
      "learning_rate": 6.042602233437985e-06,
      "loss": 0.2125,
      "step": 35630
    },
    {
      "epoch": 3.505200265545475,
      "grad_norm": 0.5192074179649353,
      "learning_rate": 6.038628144497875e-06,
      "loss": 0.1729,
      "step": 35640
    },
    {
      "epoch": 3.506183767303484,
      "grad_norm": 1.5759379863739014,
      "learning_rate": 6.0346540555577635e-06,
      "loss": 0.1651,
      "step": 35650
    },
    {
      "epoch": 3.5071672690614935,
      "grad_norm": 1.7547839879989624,
      "learning_rate": 6.030679966617653e-06,
      "loss": 0.2143,
      "step": 35660
    },
    {
      "epoch": 3.508150770819503,
      "grad_norm": 0.5516292452812195,
      "learning_rate": 6.026705877677544e-06,
      "loss": 0.1183,
      "step": 35670
    },
    {
      "epoch": 3.509134272577512,
      "grad_norm": 0.2886405885219574,
      "learning_rate": 6.0227317887374325e-06,
      "loss": 0.1418,
      "step": 35680
    },
    {
      "epoch": 3.5101177743355216,
      "grad_norm": 1.5895148515701294,
      "learning_rate": 6.018757699797322e-06,
      "loss": 0.277,
      "step": 35690
    },
    {
      "epoch": 3.511101276093531,
      "grad_norm": 1.234676718711853,
      "learning_rate": 6.014783610857211e-06,
      "loss": 0.2684,
      "step": 35700
    },
    {
      "epoch": 3.5120847778515403,
      "grad_norm": 0.45959922671318054,
      "learning_rate": 6.0108095219171014e-06,
      "loss": 0.1182,
      "step": 35710
    },
    {
      "epoch": 3.51306827960955,
      "grad_norm": 0.6054401993751526,
      "learning_rate": 6.00683543297699e-06,
      "loss": 0.2596,
      "step": 35720
    },
    {
      "epoch": 3.5140517813675594,
      "grad_norm": 0.4694426357746124,
      "learning_rate": 6.00286134403688e-06,
      "loss": 0.1923,
      "step": 35730
    },
    {
      "epoch": 3.5150352831255685,
      "grad_norm": 0.8458042144775391,
      "learning_rate": 5.998887255096769e-06,
      "loss": 0.2215,
      "step": 35740
    },
    {
      "epoch": 3.516018784883578,
      "grad_norm": 0.3480573892593384,
      "learning_rate": 5.994913166156659e-06,
      "loss": 0.2045,
      "step": 35750
    },
    {
      "epoch": 3.5170022866415875,
      "grad_norm": 1.7498723268508911,
      "learning_rate": 5.990939077216549e-06,
      "loss": 0.1557,
      "step": 35760
    },
    {
      "epoch": 3.5179857883995966,
      "grad_norm": 0.4557783305644989,
      "learning_rate": 5.986964988276438e-06,
      "loss": 0.0803,
      "step": 35770
    },
    {
      "epoch": 3.518969290157606,
      "grad_norm": 0.8801820278167725,
      "learning_rate": 5.982990899336327e-06,
      "loss": 0.2201,
      "step": 35780
    },
    {
      "epoch": 3.5199527919156157,
      "grad_norm": 0.32409054040908813,
      "learning_rate": 5.979016810396218e-06,
      "loss": 0.146,
      "step": 35790
    },
    {
      "epoch": 3.520936293673625,
      "grad_norm": 2.258561372756958,
      "learning_rate": 5.975042721456107e-06,
      "loss": 0.2379,
      "step": 35800
    },
    {
      "epoch": 3.5219197954316344,
      "grad_norm": 0.1241031289100647,
      "learning_rate": 5.971068632515996e-06,
      "loss": 0.1933,
      "step": 35810
    },
    {
      "epoch": 3.522903297189644,
      "grad_norm": 0.32828909158706665,
      "learning_rate": 5.967094543575885e-06,
      "loss": 0.1732,
      "step": 35820
    },
    {
      "epoch": 3.523886798947653,
      "grad_norm": 1.036968469619751,
      "learning_rate": 5.963120454635776e-06,
      "loss": 0.1615,
      "step": 35830
    },
    {
      "epoch": 3.5248703007056625,
      "grad_norm": 0.6440073847770691,
      "learning_rate": 5.9591463656956646e-06,
      "loss": 0.1422,
      "step": 35840
    },
    {
      "epoch": 3.525853802463672,
      "grad_norm": 0.7314437031745911,
      "learning_rate": 5.955172276755554e-06,
      "loss": 0.1484,
      "step": 35850
    },
    {
      "epoch": 3.526837304221681,
      "grad_norm": 1.1431103944778442,
      "learning_rate": 5.951198187815443e-06,
      "loss": 0.2525,
      "step": 35860
    },
    {
      "epoch": 3.5278208059796907,
      "grad_norm": 0.30557236075401306,
      "learning_rate": 5.9472240988753335e-06,
      "loss": 0.1921,
      "step": 35870
    },
    {
      "epoch": 3.5288043077377003,
      "grad_norm": 1.6383107900619507,
      "learning_rate": 5.943250009935223e-06,
      "loss": 0.2466,
      "step": 35880
    },
    {
      "epoch": 3.5297878094957094,
      "grad_norm": 0.8749147057533264,
      "learning_rate": 5.939275920995112e-06,
      "loss": 0.2489,
      "step": 35890
    },
    {
      "epoch": 3.530771311253719,
      "grad_norm": 0.33233484625816345,
      "learning_rate": 5.935301832055002e-06,
      "loss": 0.1653,
      "step": 35900
    },
    {
      "epoch": 3.5317548130117284,
      "grad_norm": 1.1335556507110596,
      "learning_rate": 5.931327743114892e-06,
      "loss": 0.2507,
      "step": 35910
    },
    {
      "epoch": 3.5327383147697375,
      "grad_norm": 0.05900925025343895,
      "learning_rate": 5.927353654174781e-06,
      "loss": 0.2375,
      "step": 35920
    },
    {
      "epoch": 3.533721816527747,
      "grad_norm": 2.2414448261260986,
      "learning_rate": 5.923379565234671e-06,
      "loss": 0.243,
      "step": 35930
    },
    {
      "epoch": 3.5347053182857566,
      "grad_norm": 1.2612745761871338,
      "learning_rate": 5.9194054762945595e-06,
      "loss": 0.2604,
      "step": 35940
    },
    {
      "epoch": 3.5356888200437657,
      "grad_norm": 0.41445934772491455,
      "learning_rate": 5.91543138735445e-06,
      "loss": 0.2468,
      "step": 35950
    },
    {
      "epoch": 3.5366723218017753,
      "grad_norm": 0.3327304422855377,
      "learning_rate": 5.911457298414339e-06,
      "loss": 0.3412,
      "step": 35960
    },
    {
      "epoch": 3.537655823559785,
      "grad_norm": 1.1569575071334839,
      "learning_rate": 5.9074832094742285e-06,
      "loss": 0.2789,
      "step": 35970
    },
    {
      "epoch": 3.538639325317794,
      "grad_norm": 1.7389458417892456,
      "learning_rate": 5.903509120534117e-06,
      "loss": 0.1788,
      "step": 35980
    },
    {
      "epoch": 3.5396228270758034,
      "grad_norm": 1.139573574066162,
      "learning_rate": 5.899535031594008e-06,
      "loss": 0.208,
      "step": 35990
    },
    {
      "epoch": 3.540606328833813,
      "grad_norm": 0.546160101890564,
      "learning_rate": 5.8955609426538975e-06,
      "loss": 0.2703,
      "step": 36000
    },
    {
      "epoch": 3.540606328833813,
      "eval_loss": 0.11898215115070343,
      "eval_runtime": 18.2129,
      "eval_samples_per_second": 2.745,
      "eval_steps_per_second": 1.373,
      "step": 36000
    }
  ],
  "logging_steps": 10,
  "max_steps": 50835,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 2000,
  "total_flos": 3.4553845225556214e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}