{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 9480,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0010548523206751054,
      "grad_norm": 1.1361546516418457,
      "learning_rate": 0.00015789473684210527,
      "loss": 7.6497,
      "step": 10
    },
    {
      "epoch": 0.002109704641350211,
      "grad_norm": 0.908168613910675,
      "learning_rate": 0.00031578947368421053,
      "loss": 7.2777,
      "step": 20
    },
    {
      "epoch": 0.0031645569620253164,
      "grad_norm": 0.8880953192710876,
      "learning_rate": 0.00047368421052631577,
      "loss": 6.528,
      "step": 30
    },
    {
      "epoch": 0.004219409282700422,
      "grad_norm": 0.6118085980415344,
      "learning_rate": 0.0006315789473684211,
      "loss": 5.8181,
      "step": 40
    },
    {
      "epoch": 0.005274261603375527,
      "grad_norm": 0.33708152174949646,
      "learning_rate": 0.0007894736842105263,
      "loss": 5.3712,
      "step": 50
    },
    {
      "epoch": 0.006329113924050633,
      "grad_norm": 0.24235396087169647,
      "learning_rate": 0.0009473684210526315,
      "loss": 4.9669,
      "step": 60
    },
    {
      "epoch": 0.007383966244725738,
      "grad_norm": 0.17883867025375366,
      "learning_rate": 0.0011052631578947368,
      "loss": 4.6049,
      "step": 70
    },
    {
      "epoch": 0.008438818565400843,
      "grad_norm": 0.34269165992736816,
      "learning_rate": 0.0012631578947368421,
      "loss": 4.3748,
      "step": 80
    },
    {
      "epoch": 0.00949367088607595,
      "grad_norm": 0.25559672713279724,
      "learning_rate": 0.0014210526315789472,
      "loss": 4.203,
      "step": 90
    },
    {
      "epoch": 0.010548523206751054,
      "grad_norm": 0.2559225261211395,
      "learning_rate": 0.0014999989494847376,
      "loss": 4.0774,
      "step": 100
    },
    {
      "epoch": 0.011603375527426161,
      "grad_norm": 0.24822884798049927,
      "learning_rate": 0.0014999905453802946,
      "loss": 3.9568,
      "step": 110
    },
    {
      "epoch": 0.012658227848101266,
      "grad_norm": 0.4745056927204132,
      "learning_rate": 0.0014999737372655805,
      "loss": 3.8672,
      "step": 120
    },
    {
      "epoch": 0.013713080168776372,
      "grad_norm": 0.2829824984073639,
      "learning_rate": 0.0014999485253289388,
      "loss": 3.7831,
      "step": 130
    },
    {
      "epoch": 0.014767932489451477,
      "grad_norm": 0.3547530174255371,
      "learning_rate": 0.0014999149098528814,
      "loss": 3.7153,
      "step": 140
    },
    {
      "epoch": 0.015822784810126583,
      "grad_norm": 0.32553762197494507,
      "learning_rate": 0.0014998728912140862,
      "loss": 3.6379,
      "step": 150
    },
    {
      "epoch": 0.016877637130801686,
      "grad_norm": 0.29201358556747437,
      "learning_rate": 0.0014998224698833922,
      "loss": 3.5747,
      "step": 160
    },
    {
      "epoch": 0.017932489451476793,
      "grad_norm": 0.3410353660583496,
      "learning_rate": 0.0014997636464257956,
      "loss": 3.5088,
      "step": 170
    },
    {
      "epoch": 0.0189873417721519,
      "grad_norm": 0.5075426697731018,
      "learning_rate": 0.0014996964215004416,
      "loss": 3.473,
      "step": 180
    },
    {
      "epoch": 0.020042194092827006,
      "grad_norm": 0.3382231593132019,
      "learning_rate": 0.0014996207958606182,
      "loss": 3.4133,
      "step": 190
    },
    {
      "epoch": 0.02109704641350211,
      "grad_norm": 0.36965426802635193,
      "learning_rate": 0.001499536770353748,
      "loss": 3.3602,
      "step": 200
    },
    {
      "epoch": 0.022151898734177215,
      "grad_norm": 0.49166449904441833,
      "learning_rate": 0.0014994443459213774,
      "loss": 3.3143,
      "step": 210
    },
    {
      "epoch": 0.023206751054852322,
      "grad_norm": 0.34380143880844116,
      "learning_rate": 0.001499343523599168,
      "loss": 3.2807,
      "step": 220
    },
    {
      "epoch": 0.024261603375527425,
      "grad_norm": 0.5123799443244934,
      "learning_rate": 0.0014992343045168823,
      "loss": 3.2242,
      "step": 230
    },
    {
      "epoch": 0.02531645569620253,
      "grad_norm": 0.4475407004356384,
      "learning_rate": 0.0014991166898983739,
      "loss": 3.1804,
      "step": 240
    },
    {
      "epoch": 0.026371308016877638,
      "grad_norm": 0.6833670735359192,
      "learning_rate": 0.001498990681061572,
      "loss": 3.1403,
      "step": 250
    },
    {
      "epoch": 0.027426160337552744,
      "grad_norm": 0.5144447088241577,
      "learning_rate": 0.001498856279418467,
      "loss": 3.096,
      "step": 260
    },
    {
      "epoch": 0.028481012658227847,
      "grad_norm": 0.42934003472328186,
      "learning_rate": 0.0014987134864750948,
      "loss": 3.0594,
      "step": 270
    },
    {
      "epoch": 0.029535864978902954,
      "grad_norm": 0.5111614465713501,
      "learning_rate": 0.0014985623038315206,
      "loss": 3.0117,
      "step": 280
    },
    {
      "epoch": 0.03059071729957806,
      "grad_norm": 0.5440753102302551,
      "learning_rate": 0.0014984027331818193,
      "loss": 2.9799,
      "step": 290
    },
    {
      "epoch": 0.03164556962025317,
      "grad_norm": 0.5430408716201782,
      "learning_rate": 0.0014982347763140584,
      "loss": 2.9458,
      "step": 300
    },
    {
      "epoch": 0.03270042194092827,
      "grad_norm": 0.8532850742340088,
      "learning_rate": 0.0014980584351102762,
      "loss": 2.9085,
      "step": 310
    },
    {
      "epoch": 0.03375527426160337,
      "grad_norm": 0.48789384961128235,
      "learning_rate": 0.001497873711546462,
      "loss": 2.8788,
      "step": 320
    },
    {
      "epoch": 0.03481012658227848,
      "grad_norm": 0.46921592950820923,
      "learning_rate": 0.0014976806076925334,
      "loss": 2.8617,
      "step": 330
    },
    {
      "epoch": 0.035864978902953586,
      "grad_norm": 0.40585583448410034,
      "learning_rate": 0.0014974791257123137,
      "loss": 2.8296,
      "step": 340
    },
    {
      "epoch": 0.03691983122362869,
      "grad_norm": 0.8285185694694519,
      "learning_rate": 0.001497269267863507,
      "loss": 2.7758,
      "step": 350
    },
    {
      "epoch": 0.0379746835443038,
      "grad_norm": 0.5533679127693176,
      "learning_rate": 0.0014970510364976724,
      "loss": 2.7506,
      "step": 360
    },
    {
      "epoch": 0.039029535864978905,
      "grad_norm": 0.8483739495277405,
      "learning_rate": 0.0014968244340601996,
      "loss": 2.736,
      "step": 370
    },
    {
      "epoch": 0.04008438818565401,
      "grad_norm": 0.7598241567611694,
      "learning_rate": 0.001496589463090279,
      "loss": 2.7141,
      "step": 380
    },
    {
      "epoch": 0.04113924050632911,
      "grad_norm": 0.738063633441925,
      "learning_rate": 0.001496346126220875,
      "loss": 2.6895,
      "step": 390
    },
    {
      "epoch": 0.04219409282700422,
      "grad_norm": 0.6263159513473511,
      "learning_rate": 0.0014960944261786966,
      "loss": 2.6634,
      "step": 400
    },
    {
      "epoch": 0.043248945147679324,
      "grad_norm": 0.4596900939941406,
      "learning_rate": 0.0014958343657841655,
      "loss": 2.6408,
      "step": 410
    },
    {
      "epoch": 0.04430379746835443,
      "grad_norm": 0.4793330729007721,
      "learning_rate": 0.001495565947951385,
      "loss": 2.6084,
      "step": 420
    },
    {
      "epoch": 0.04535864978902954,
      "grad_norm": 0.7548810243606567,
      "learning_rate": 0.0014952891756881085,
      "loss": 2.5838,
      "step": 430
    },
    {
      "epoch": 0.046413502109704644,
      "grad_norm": 0.8852179050445557,
      "learning_rate": 0.0014950040520957037,
      "loss": 2.5606,
      "step": 440
    },
    {
      "epoch": 0.04746835443037975,
      "grad_norm": 0.5558875203132629,
      "learning_rate": 0.0014947105803691204,
      "loss": 2.5508,
      "step": 450
    },
    {
      "epoch": 0.04852320675105485,
      "grad_norm": 0.4711481034755707,
      "learning_rate": 0.0014944087637968522,
      "loss": 2.5245,
      "step": 460
    },
    {
      "epoch": 0.049578059071729956,
      "grad_norm": 0.510650098323822,
      "learning_rate": 0.0014940986057609012,
      "loss": 2.4975,
      "step": 470
    },
    {
      "epoch": 0.05063291139240506,
      "grad_norm": 0.8742207884788513,
      "learning_rate": 0.0014937801097367396,
      "loss": 2.499,
      "step": 480
    },
    {
      "epoch": 0.05168776371308017,
      "grad_norm": 0.7497596144676208,
      "learning_rate": 0.001493453279293271,
      "loss": 2.4649,
      "step": 490
    },
    {
      "epoch": 0.052742616033755275,
      "grad_norm": 0.4843575954437256,
      "learning_rate": 0.0014931181180927902,
      "loss": 2.4477,
      "step": 500
    },
    {
      "epoch": 0.05379746835443038,
      "grad_norm": 0.7055267095565796,
      "learning_rate": 0.001492774629890942,
      "loss": 2.4372,
      "step": 510
    },
    {
      "epoch": 0.05485232067510549,
      "grad_norm": 1.0991405248641968,
      "learning_rate": 0.001492422818536679,
      "loss": 2.4239,
      "step": 520
    },
    {
      "epoch": 0.05590717299578059,
      "grad_norm": 0.6026334762573242,
      "learning_rate": 0.00149206268797222,
      "loss": 2.4042,
      "step": 530
    },
    {
      "epoch": 0.056962025316455694,
      "grad_norm": 0.9058926701545715,
      "learning_rate": 0.0014916942422330032,
      "loss": 2.3768,
      "step": 540
    },
    {
      "epoch": 0.0580168776371308,
      "grad_norm": 0.5138038992881775,
      "learning_rate": 0.001491317485447643,
      "loss": 2.3767,
      "step": 550
    },
    {
      "epoch": 0.05907172995780591,
      "grad_norm": 0.5587601065635681,
      "learning_rate": 0.0014909324218378838,
      "loss": 2.3387,
      "step": 560
    },
    {
      "epoch": 0.060126582278481014,
      "grad_norm": 0.6543309092521667,
      "learning_rate": 0.0014905390557185508,
      "loss": 2.3423,
      "step": 570
    },
    {
      "epoch": 0.06118143459915612,
      "grad_norm": 0.6026796102523804,
      "learning_rate": 0.0014901373914975036,
      "loss": 2.3378,
      "step": 580
    },
    {
      "epoch": 0.06223628691983123,
      "grad_norm": 0.8843022584915161,
      "learning_rate": 0.0014897274336755856,
      "loss": 2.3148,
      "step": 590
    },
    {
      "epoch": 0.06329113924050633,
      "grad_norm": 0.7628885507583618,
      "learning_rate": 0.001489309186846575,
      "loss": 2.2917,
      "step": 600
    },
    {
      "epoch": 0.06434599156118144,
      "grad_norm": 0.5476912260055542,
      "learning_rate": 0.0014888826556971313,
      "loss": 2.2892,
      "step": 610
    },
    {
      "epoch": 0.06540084388185655,
      "grad_norm": 0.42697277665138245,
      "learning_rate": 0.0014884478450067444,
      "loss": 2.2752,
      "step": 620
    },
    {
      "epoch": 0.06645569620253164,
      "grad_norm": 0.4697224795818329,
      "learning_rate": 0.0014880047596476807,
      "loss": 2.2573,
      "step": 630
    },
    {
      "epoch": 0.06751054852320675,
      "grad_norm": 0.4894421398639679,
      "learning_rate": 0.0014875534045849274,
      "loss": 2.2651,
      "step": 640
    },
    {
      "epoch": 0.06856540084388185,
      "grad_norm": 0.6345561742782593,
      "learning_rate": 0.0014870937848761388,
      "loss": 2.2555,
      "step": 650
    },
    {
      "epoch": 0.06962025316455696,
      "grad_norm": 0.49453651905059814,
      "learning_rate": 0.001486625905671578,
      "loss": 2.2441,
      "step": 660
    },
    {
      "epoch": 0.07067510548523206,
      "grad_norm": 0.6118682026863098,
      "learning_rate": 0.00148614977221406,
      "loss": 2.2208,
      "step": 670
    },
    {
      "epoch": 0.07172995780590717,
      "grad_norm": 0.4619838297367096,
      "learning_rate": 0.0014856653898388927,
      "loss": 2.2133,
      "step": 680
    },
    {
      "epoch": 0.07278481012658228,
      "grad_norm": 0.5305221676826477,
      "learning_rate": 0.001485172763973817,
      "loss": 2.2258,
      "step": 690
    },
    {
      "epoch": 0.07383966244725738,
      "grad_norm": 0.47047242522239685,
      "learning_rate": 0.0014846719001389466,
      "loss": 2.1992,
      "step": 700
    },
    {
      "epoch": 0.07489451476793249,
      "grad_norm": 1.0235542058944702,
      "learning_rate": 0.001484162803946705,
      "loss": 2.1917,
      "step": 710
    },
    {
      "epoch": 0.0759493670886076,
      "grad_norm": 0.56936115026474,
      "learning_rate": 0.0014836454811017635,
      "loss": 2.1703,
      "step": 720
    },
    {
      "epoch": 0.0770042194092827,
      "grad_norm": 0.6495599746704102,
      "learning_rate": 0.0014831199374009778,
      "loss": 2.1679,
      "step": 730
    },
    {
      "epoch": 0.07805907172995781,
      "grad_norm": 0.49974289536476135,
      "learning_rate": 0.0014825861787333208,
      "loss": 2.1746,
      "step": 740
    },
    {
      "epoch": 0.07911392405063292,
      "grad_norm": 0.5245474576950073,
      "learning_rate": 0.0014820442110798197,
      "loss": 2.1507,
      "step": 750
    },
    {
      "epoch": 0.08016877637130802,
      "grad_norm": 0.5185036063194275,
      "learning_rate": 0.0014814940405134865,
      "loss": 2.1499,
      "step": 760
    },
    {
      "epoch": 0.08122362869198312,
      "grad_norm": 0.5844479203224182,
      "learning_rate": 0.001480935673199251,
      "loss": 2.1403,
      "step": 770
    },
    {
      "epoch": 0.08227848101265822,
      "grad_norm": 0.48124414682388306,
      "learning_rate": 0.0014803691153938915,
      "loss": 2.1389,
      "step": 780
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 0.7114575505256653,
      "learning_rate": 0.0014797943734459653,
      "loss": 2.1425,
      "step": 790
    },
    {
      "epoch": 0.08438818565400844,
      "grad_norm": 0.4797729551792145,
      "learning_rate": 0.001479211453795736,
      "loss": 2.1121,
      "step": 800
    },
    {
      "epoch": 0.08544303797468354,
      "grad_norm": 0.4670267105102539,
      "learning_rate": 0.0014786203629751033,
      "loss": 2.0947,
      "step": 810
    },
    {
      "epoch": 0.08649789029535865,
      "grad_norm": 0.447757363319397,
      "learning_rate": 0.0014780211076075279,
      "loss": 2.1016,
      "step": 820
    },
    {
      "epoch": 0.08755274261603375,
      "grad_norm": 0.5141759514808655,
      "learning_rate": 0.0014774136944079594,
      "loss": 2.1068,
      "step": 830
    },
    {
      "epoch": 0.08860759493670886,
      "grad_norm": 0.7842695713043213,
      "learning_rate": 0.0014767981301827592,
      "loss": 2.0852,
      "step": 840
    },
    {
      "epoch": 0.08966244725738397,
      "grad_norm": 0.48477619886398315,
      "learning_rate": 0.0014761744218296249,
      "loss": 2.0852,
      "step": 850
    },
    {
      "epoch": 0.09071729957805907,
      "grad_norm": 0.4969441294670105,
      "learning_rate": 0.001475542576337513,
      "loss": 2.0816,
      "step": 860
    },
    {
      "epoch": 0.09177215189873418,
      "grad_norm": 0.6644434332847595,
      "learning_rate": 0.001474902600786561,
      "loss": 2.0758,
      "step": 870
    },
    {
      "epoch": 0.09282700421940929,
      "grad_norm": 0.5549584627151489,
      "learning_rate": 0.0014742545023480075,
      "loss": 2.0705,
      "step": 880
    },
    {
      "epoch": 0.0938818565400844,
      "grad_norm": 0.44507768750190735,
      "learning_rate": 0.0014735982882841117,
      "loss": 2.0548,
      "step": 890
    },
    {
      "epoch": 0.0949367088607595,
      "grad_norm": 0.5654860138893127,
      "learning_rate": 0.0014729339659480727,
      "loss": 2.0569,
      "step": 900
    },
    {
      "epoch": 0.09599156118143459,
      "grad_norm": 0.4581911563873291,
      "learning_rate": 0.0014722615427839468,
      "loss": 2.0635,
      "step": 910
    },
    {
      "epoch": 0.0970464135021097,
      "grad_norm": 0.6122888922691345,
      "learning_rate": 0.0014715810263265633,
      "loss": 2.0475,
      "step": 920
    },
    {
      "epoch": 0.0981012658227848,
      "grad_norm": 0.7882494330406189,
      "learning_rate": 0.0014708924242014423,
      "loss": 2.0373,
      "step": 930
    },
    {
      "epoch": 0.09915611814345991,
      "grad_norm": 0.7913827896118164,
      "learning_rate": 0.0014701957441247064,
      "loss": 2.0423,
      "step": 940
    },
    {
      "epoch": 0.10021097046413502,
      "grad_norm": 0.6759977340698242,
      "learning_rate": 0.0014694909939029959,
      "loss": 2.024,
      "step": 950
    },
    {
      "epoch": 0.10126582278481013,
      "grad_norm": 0.4497184753417969,
      "learning_rate": 0.0014687781814333814,
      "loss": 2.0273,
      "step": 960
    },
    {
      "epoch": 0.10232067510548523,
      "grad_norm": 0.9482643008232117,
      "learning_rate": 0.0014680573147032746,
      "loss": 2.0239,
      "step": 970
    },
    {
      "epoch": 0.10337552742616034,
      "grad_norm": 0.5979325175285339,
      "learning_rate": 0.0014673284017903392,
      "loss": 2.0078,
      "step": 980
    },
    {
      "epoch": 0.10443037974683544,
      "grad_norm": 0.4397555887699127,
      "learning_rate": 0.0014665914508624,
      "loss": 2.0031,
      "step": 990
    },
    {
      "epoch": 0.10548523206751055,
      "grad_norm": 0.9002691507339478,
      "learning_rate": 0.0014658464701773526,
      "loss": 2.0123,
      "step": 1000
    },
    {
      "epoch": 0.10654008438818566,
      "grad_norm": 0.9311442375183105,
      "learning_rate": 0.0014650934680830688,
      "loss": 2.0061,
      "step": 1010
    },
    {
      "epoch": 0.10759493670886076,
      "grad_norm": 0.6002566814422607,
      "learning_rate": 0.0014643324530173051,
      "loss": 1.9991,
      "step": 1020
    },
    {
      "epoch": 0.10864978902953587,
      "grad_norm": 0.6356918811798096,
      "learning_rate": 0.0014635634335076067,
      "loss": 1.989,
      "step": 1030
    },
    {
      "epoch": 0.10970464135021098,
      "grad_norm": 0.6467710733413696,
      "learning_rate": 0.001462786418171213,
      "loss": 1.985,
      "step": 1040
    },
    {
      "epoch": 0.11075949367088607,
      "grad_norm": 1.0886985063552856,
      "learning_rate": 0.0014620014157149597,
      "loss": 1.9903,
      "step": 1050
    },
    {
      "epoch": 0.11181434599156118,
      "grad_norm": 0.7051685452461243,
      "learning_rate": 0.001461208434935183,
      "loss": 1.9818,
      "step": 1060
    },
    {
      "epoch": 0.11286919831223628,
      "grad_norm": 0.4662516713142395,
      "learning_rate": 0.0014604074847176197,
      "loss": 1.9642,
      "step": 1070
    },
    {
      "epoch": 0.11392405063291139,
      "grad_norm": 0.515107274055481,
      "learning_rate": 0.0014595985740373082,
      "loss": 1.9665,
      "step": 1080
    },
    {
      "epoch": 0.1149789029535865,
      "grad_norm": 0.566654622554779,
      "learning_rate": 0.0014587817119584873,
      "loss": 1.9749,
      "step": 1090
    },
    {
      "epoch": 0.1160337552742616,
      "grad_norm": 0.5290713310241699,
      "learning_rate": 0.001457956907634496,
      "loss": 1.9587,
      "step": 1100
    },
    {
      "epoch": 0.11708860759493671,
      "grad_norm": 0.5899307727813721,
      "learning_rate": 0.0014571241703076692,
      "loss": 1.9641,
      "step": 1110
    },
    {
      "epoch": 0.11814345991561181,
      "grad_norm": 0.4900335967540741,
      "learning_rate": 0.0014562835093092348,
      "loss": 1.9549,
      "step": 1120
    },
    {
      "epoch": 0.11919831223628692,
      "grad_norm": 1.0779500007629395,
      "learning_rate": 0.0014554349340592104,
      "loss": 1.9446,
      "step": 1130
    },
    {
      "epoch": 0.12025316455696203,
      "grad_norm": 0.7757524251937866,
      "learning_rate": 0.001454578454066296,
      "loss": 1.9566,
      "step": 1140
    },
    {
      "epoch": 0.12130801687763713,
      "grad_norm": 0.7656589150428772,
      "learning_rate": 0.0014537140789277678,
      "loss": 1.94,
      "step": 1150
    },
    {
      "epoch": 0.12236286919831224,
      "grad_norm": 0.5552901029586792,
      "learning_rate": 0.0014528418183293716,
      "loss": 1.9486,
      "step": 1160
    },
    {
      "epoch": 0.12341772151898735,
      "grad_norm": 0.546013593673706,
      "learning_rate": 0.001451961682045213,
      "loss": 1.9385,
      "step": 1170
    },
    {
      "epoch": 0.12447257383966245,
      "grad_norm": 0.4671438932418823,
      "learning_rate": 0.001451073679937649,
      "loss": 1.9205,
      "step": 1180
    },
    {
      "epoch": 0.12552742616033755,
      "grad_norm": 0.9370329976081848,
      "learning_rate": 0.0014501778219571766,
      "loss": 1.9231,
      "step": 1190
    },
    {
      "epoch": 0.12658227848101267,
      "grad_norm": 0.4605046510696411,
      "learning_rate": 0.0014492741181423225,
      "loss": 1.9354,
      "step": 1200
    },
    {
      "epoch": 0.12763713080168776,
      "grad_norm": 0.4588429927825928,
      "learning_rate": 0.0014483625786195285,
      "loss": 1.93,
      "step": 1210
    },
    {
      "epoch": 0.12869198312236288,
      "grad_norm": 0.8734654784202576,
      "learning_rate": 0.0014474432136030405,
      "loss": 1.9162,
      "step": 1220
    },
    {
      "epoch": 0.12974683544303797,
      "grad_norm": 0.8143278956413269,
      "learning_rate": 0.0014465160333947923,
      "loss": 1.9052,
      "step": 1230
    },
    {
      "epoch": 0.1308016877637131,
      "grad_norm": 0.5360792875289917,
      "learning_rate": 0.0014455810483842908,
      "loss": 1.919,
      "step": 1240
    },
    {
      "epoch": 0.13185654008438819,
      "grad_norm": 0.6346741914749146,
      "learning_rate": 0.0014446382690484997,
      "loss": 1.9187,
      "step": 1250
    },
    {
      "epoch": 0.13291139240506328,
      "grad_norm": 0.774360179901123,
      "learning_rate": 0.0014436877059517215,
      "loss": 1.9048,
      "step": 1260
    },
    {
      "epoch": 0.1339662447257384,
      "grad_norm": 0.5729191899299622,
      "learning_rate": 0.0014427293697454803,
      "loss": 1.9038,
      "step": 1270
    },
    {
      "epoch": 0.1350210970464135,
      "grad_norm": 0.48271700739860535,
      "learning_rate": 0.001441763271168401,
      "loss": 1.9108,
      "step": 1280
    },
    {
      "epoch": 0.1360759493670886,
      "grad_norm": 0.5652008056640625,
      "learning_rate": 0.00144078942104609,
      "loss": 1.8999,
      "step": 1290
    },
    {
      "epoch": 0.1371308016877637,
      "grad_norm": 0.5593475103378296,
      "learning_rate": 0.001439807830291013,
      "loss": 1.8875,
      "step": 1300
    },
    {
      "epoch": 0.13818565400843882,
      "grad_norm": 0.4523153305053711,
      "learning_rate": 0.0014388185099023744,
      "loss": 1.8913,
      "step": 1310
    },
    {
      "epoch": 0.13924050632911392,
      "grad_norm": 0.9170459508895874,
      "learning_rate": 0.0014378214709659916,
      "loss": 1.8905,
      "step": 1320
    },
    {
      "epoch": 0.14029535864978904,
      "grad_norm": 0.7784819602966309,
      "learning_rate": 0.0014368167246541733,
      "loss": 1.8853,
      "step": 1330
    },
    {
      "epoch": 0.14135021097046413,
      "grad_norm": 0.5773763060569763,
      "learning_rate": 0.0014358042822255918,
      "loss": 1.8851,
      "step": 1340
    },
    {
      "epoch": 0.14240506329113925,
      "grad_norm": 0.8327829837799072,
      "learning_rate": 0.0014347841550251597,
      "loss": 1.8909,
      "step": 1350
    },
    {
      "epoch": 0.14345991561181434,
      "grad_norm": 0.6381543874740601,
      "learning_rate": 0.0014337563544838997,
      "loss": 1.8805,
      "step": 1360
    },
    {
      "epoch": 0.14451476793248946,
      "grad_norm": 0.5755815505981445,
      "learning_rate": 0.001432720892118819,
      "loss": 1.8754,
      "step": 1370
    },
    {
      "epoch": 0.14556962025316456,
      "grad_norm": 0.6887590885162354,
      "learning_rate": 0.0014316777795327794,
      "loss": 1.8649,
      "step": 1380
    },
    {
      "epoch": 0.14662447257383968,
      "grad_norm": 0.5535954833030701,
      "learning_rate": 0.001430627028414366,
      "loss": 1.8785,
      "step": 1390
    },
    {
      "epoch": 0.14767932489451477,
      "grad_norm": 0.5511561036109924,
      "learning_rate": 0.0014295686505377586,
      "loss": 1.8601,
      "step": 1400
    },
    {
      "epoch": 0.14873417721518986,
      "grad_norm": 0.6784546375274658,
      "learning_rate": 0.0014285026577625982,
      "loss": 1.8624,
      "step": 1410
    },
    {
      "epoch": 0.14978902953586498,
      "grad_norm": 0.46250203251838684,
      "learning_rate": 0.0014274290620338542,
      "loss": 1.8663,
      "step": 1420
    },
    {
      "epoch": 0.15084388185654007,
      "grad_norm": 0.7906670570373535,
      "learning_rate": 0.0014263478753816906,
      "loss": 1.8584,
      "step": 1430
    },
    {
      "epoch": 0.1518987341772152,
      "grad_norm": 0.5146434903144836,
      "learning_rate": 0.0014252591099213326,
      "loss": 1.8554,
      "step": 1440
    },
    {
      "epoch": 0.1529535864978903,
      "grad_norm": 1.0950006246566772,
      "learning_rate": 0.001424162777852928,
      "loss": 1.8564,
      "step": 1450
    },
    {
      "epoch": 0.1540084388185654,
      "grad_norm": 0.5953426361083984,
      "learning_rate": 0.0014230588914614134,
      "loss": 1.8485,
      "step": 1460
    },
    {
      "epoch": 0.1550632911392405,
      "grad_norm": 0.5193158388137817,
      "learning_rate": 0.0014219474631163745,
      "loss": 1.8437,
      "step": 1470
    },
    {
      "epoch": 0.15611814345991562,
      "grad_norm": 0.6556932330131531,
      "learning_rate": 0.001420828505271909,
      "loss": 1.8488,
      "step": 1480
    },
    {
      "epoch": 0.1571729957805907,
      "grad_norm": 0.5962861776351929,
      "learning_rate": 0.0014197020304664856,
      "loss": 1.8471,
      "step": 1490
    },
    {
      "epoch": 0.15822784810126583,
      "grad_norm": 0.46357786655426025,
      "learning_rate": 0.0014185680513228048,
      "loss": 1.8467,
      "step": 1500
    },
    {
      "epoch": 0.15928270042194093,
      "grad_norm": 0.4353756010532379,
      "learning_rate": 0.0014174265805476564,
      "loss": 1.841,
      "step": 1510
    },
    {
      "epoch": 0.16033755274261605,
      "grad_norm": 0.4645099639892578,
      "learning_rate": 0.0014162776309317778,
      "loss": 1.8449,
      "step": 1520
    },
    {
      "epoch": 0.16139240506329114,
      "grad_norm": 0.5141316652297974,
      "learning_rate": 0.0014151212153497108,
      "loss": 1.8296,
      "step": 1530
    },
    {
      "epoch": 0.16244725738396623,
      "grad_norm": 0.6246196031570435,
      "learning_rate": 0.0014139573467596561,
      "loss": 1.8191,
      "step": 1540
    },
    {
      "epoch": 0.16350210970464135,
      "grad_norm": 0.5215326547622681,
      "learning_rate": 0.00141278603820333,
      "loss": 1.8196,
      "step": 1550
    },
    {
      "epoch": 0.16455696202531644,
      "grad_norm": 0.5152313113212585,
      "learning_rate": 0.0014116073028058165,
      "loss": 1.8215,
      "step": 1560
    },
    {
      "epoch": 0.16561181434599156,
      "grad_norm": 0.5511691570281982,
      "learning_rate": 0.0014104211537754217,
      "loss": 1.811,
      "step": 1570
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 0.6224981546401978,
      "learning_rate": 0.001409227604403524,
      "loss": 1.8241,
      "step": 1580
    },
    {
      "epoch": 0.16772151898734178,
      "grad_norm": 0.5345869064331055,
      "learning_rate": 0.0014080266680644277,
      "loss": 1.8219,
      "step": 1590
    },
    {
      "epoch": 0.16877637130801687,
      "grad_norm": 0.5171884298324585,
      "learning_rate": 0.0014068183582152103,
      "loss": 1.8211,
      "step": 1600
    },
    {
      "epoch": 0.169831223628692,
      "grad_norm": 0.548721432685852,
      "learning_rate": 0.001405602688395574,
      "loss": 1.8192,
      "step": 1610
    },
    {
      "epoch": 0.17088607594936708,
      "grad_norm": 0.7188496589660645,
      "learning_rate": 0.0014043796722276924,
      "loss": 1.7978,
      "step": 1620
    },
    {
      "epoch": 0.1719409282700422,
      "grad_norm": 1.0115805864334106,
      "learning_rate": 0.0014031493234160591,
      "loss": 1.8066,
      "step": 1630
    },
    {
      "epoch": 0.1729957805907173,
      "grad_norm": 0.9728096723556519,
      "learning_rate": 0.0014019116557473332,
      "loss": 1.8252,
      "step": 1640
    },
    {
      "epoch": 0.17405063291139242,
      "grad_norm": 0.5120858550071716,
      "learning_rate": 0.0014006666830901854,
      "loss": 1.8007,
      "step": 1650
    },
    {
      "epoch": 0.1751054852320675,
      "grad_norm": 0.6093056201934814,
      "learning_rate": 0.001399414419395142,
      "loss": 1.794,
      "step": 1660
    },
    {
      "epoch": 0.17616033755274263,
      "grad_norm": 0.7036264538764954,
      "learning_rate": 0.0013981548786944293,
      "loss": 1.8054,
      "step": 1670
    },
    {
      "epoch": 0.17721518987341772,
      "grad_norm": 0.49233219027519226,
      "learning_rate": 0.0013968880751018158,
      "loss": 1.7936,
      "step": 1680
    },
    {
      "epoch": 0.17827004219409281,
      "grad_norm": 0.48756009340286255,
      "learning_rate": 0.0013956140228124545,
      "loss": 1.7855,
      "step": 1690
    },
    {
      "epoch": 0.17932489451476794,
      "grad_norm": 0.8501176238059998,
      "learning_rate": 0.0013943327361027231,
      "loss": 1.8065,
      "step": 1700
    },
    {
      "epoch": 0.18037974683544303,
      "grad_norm": 0.4327588975429535,
      "learning_rate": 0.0013930442293300649,
      "loss": 1.7812,
      "step": 1710
    },
    {
      "epoch": 0.18143459915611815,
      "grad_norm": 0.5454190373420715,
      "learning_rate": 0.0013917485169328279,
      "loss": 1.7814,
      "step": 1720
    },
    {
      "epoch": 0.18248945147679324,
      "grad_norm": 0.5129348039627075,
      "learning_rate": 0.0013904456134301016,
      "loss": 1.7804,
      "step": 1730
    },
    {
      "epoch": 0.18354430379746836,
      "grad_norm": 0.5207455158233643,
      "learning_rate": 0.0013891355334215562,
      "loss": 1.7872,
      "step": 1740
    },
    {
      "epoch": 0.18459915611814345,
      "grad_norm": 0.6900990605354309,
      "learning_rate": 0.0013878182915872776,
      "loss": 1.808,
      "step": 1750
    },
    {
      "epoch": 0.18565400843881857,
      "grad_norm": 0.44003963470458984,
      "learning_rate": 0.001386493902687604,
      "loss": 1.7829,
      "step": 1760
    },
    {
      "epoch": 0.18670886075949367,
      "grad_norm": 0.43902549147605896,
      "learning_rate": 0.00138516238156296,
      "loss": 1.7846,
      "step": 1770
    },
    {
      "epoch": 0.1877637130801688,
      "grad_norm": 0.5925940275192261,
      "learning_rate": 0.0013838237431336895,
      "loss": 1.7847,
      "step": 1780
    },
    {
      "epoch": 0.18881856540084388,
      "grad_norm": 0.5813276767730713,
      "learning_rate": 0.0013824780023998899,
      "loss": 1.7833,
      "step": 1790
    },
    {
      "epoch": 0.189873417721519,
      "grad_norm": 0.9307818412780762,
      "learning_rate": 0.0013811251744412431,
      "loss": 1.7816,
      "step": 1800
    },
    {
      "epoch": 0.1909282700421941,
      "grad_norm": 0.7600083351135254,
      "learning_rate": 0.0013797652744168473,
      "loss": 1.7716,
      "step": 1810
    },
    {
      "epoch": 0.19198312236286919,
      "grad_norm": 0.5846881866455078,
      "learning_rate": 0.0013783983175650457,
      "loss": 1.7686,
      "step": 1820
    },
    {
      "epoch": 0.1930379746835443,
      "grad_norm": 0.5641414523124695,
      "learning_rate": 0.0013770243192032581,
      "loss": 1.7659,
      "step": 1830
    },
    {
      "epoch": 0.1940928270042194,
      "grad_norm": 0.5104548931121826,
      "learning_rate": 0.0013756432947278064,
      "loss": 1.7637,
      "step": 1840
    },
    {
      "epoch": 0.19514767932489452,
      "grad_norm": 0.7146143317222595,
      "learning_rate": 0.0013742552596137444,
      "loss": 1.7683,
      "step": 1850
    },
    {
      "epoch": 0.1962025316455696,
      "grad_norm": 0.6184601783752441,
      "learning_rate": 0.0013728602294146833,
      "loss": 1.7711,
      "step": 1860
    },
    {
      "epoch": 0.19725738396624473,
      "grad_norm": 0.8336368799209595,
      "learning_rate": 0.0013714582197626175,
      "loss": 1.7746,
      "step": 1870
    },
    {
      "epoch": 0.19831223628691982,
      "grad_norm": 0.46143412590026855,
      "learning_rate": 0.0013700492463677501,
      "loss": 1.7564,
      "step": 1880
    },
    {
      "epoch": 0.19936708860759494,
      "grad_norm": 0.532975971698761,
      "learning_rate": 0.0013686333250183154,
      "loss": 1.7602,
      "step": 1890
    },
    {
      "epoch": 0.20042194092827004,
      "grad_norm": 0.79651939868927,
      "learning_rate": 0.001367210471580404,
      "loss": 1.7577,
      "step": 1900
    },
    {
      "epoch": 0.20147679324894516,
      "grad_norm": 0.48664984107017517,
      "learning_rate": 0.0013657807019977835,
      "loss": 1.758,
      "step": 1910
    },
    {
      "epoch": 0.20253164556962025,
      "grad_norm": 0.6820895671844482,
      "learning_rate": 0.0013643440322917198,
      "loss": 1.7565,
      "step": 1920
    },
    {
      "epoch": 0.20358649789029537,
      "grad_norm": 0.9615606665611267,
      "learning_rate": 0.0013629004785607989,
      "loss": 1.7572,
      "step": 1930
    },
    {
      "epoch": 0.20464135021097046,
      "grad_norm": 0.4524003565311432,
      "learning_rate": 0.0013614500569807445,
      "loss": 1.7452,
      "step": 1940
    },
    {
      "epoch": 0.20569620253164558,
      "grad_norm": 0.49840429425239563,
      "learning_rate": 0.0013599927838042394,
      "loss": 1.7489,
      "step": 1950
    },
    {
      "epoch": 0.20675105485232068,
      "grad_norm": 1.0147011280059814,
      "learning_rate": 0.0013585286753607408,
      "loss": 1.7504,
      "step": 1960
    },
    {
      "epoch": 0.20780590717299577,
      "grad_norm": 0.483129620552063,
      "learning_rate": 0.0013570577480562986,
      "loss": 1.7596,
      "step": 1970
    },
    {
      "epoch": 0.2088607594936709,
      "grad_norm": 0.5301021337509155,
      "learning_rate": 0.0013555800183733717,
      "loss": 1.7408,
      "step": 1980
    },
    {
      "epoch": 0.20991561181434598,
      "grad_norm": 0.5274137258529663,
      "learning_rate": 0.0013540955028706425,
      "loss": 1.7372,
      "step": 1990
    },
    {
      "epoch": 0.2109704641350211,
      "grad_norm": 0.5680530071258545,
      "learning_rate": 0.0013526042181828324,
      "loss": 1.7373,
      "step": 2000
    },
    {
      "epoch": 0.2120253164556962,
      "grad_norm": 0.69745272397995,
      "learning_rate": 0.0013511061810205143,
      "loss": 1.744,
      "step": 2010
    },
    {
      "epoch": 0.21308016877637131,
      "grad_norm": 0.775166392326355,
      "learning_rate": 0.001349601408169926,
      "loss": 1.7526,
      "step": 2020
    },
    {
      "epoch": 0.2141350210970464,
      "grad_norm": 0.5244166254997253,
      "learning_rate": 0.0013480899164927823,
      "loss": 1.7365,
      "step": 2030
    },
    {
      "epoch": 0.21518987341772153,
      "grad_norm": 0.5409502983093262,
      "learning_rate": 0.0013465717229260853,
      "loss": 1.7379,
      "step": 2040
    },
    {
      "epoch": 0.21624472573839662,
      "grad_norm": 0.5058172941207886,
      "learning_rate": 0.001345046844481935,
      "loss": 1.7349,
      "step": 2050
    },
    {
      "epoch": 0.21729957805907174,
      "grad_norm": 0.5646553039550781,
      "learning_rate": 0.0013435152982473396,
      "loss": 1.7276,
      "step": 2060
    },
    {
      "epoch": 0.21835443037974683,
      "grad_norm": 0.7753597497940063,
      "learning_rate": 0.0013419771013840217,
      "loss": 1.729,
      "step": 2070
    },
    {
      "epoch": 0.21940928270042195,
      "grad_norm": 0.7563711404800415,
      "learning_rate": 0.001340432271128229,
      "loss": 1.733,
      "step": 2080
    },
    {
      "epoch": 0.22046413502109705,
      "grad_norm": 0.6744264960289001,
      "learning_rate": 0.0013388808247905381,
      "loss": 1.7233,
      "step": 2090
    },
    {
      "epoch": 0.22151898734177214,
      "grad_norm": 0.5431970357894897,
      "learning_rate": 0.0013373227797556634,
      "loss": 1.7336,
      "step": 2100
    },
    {
      "epoch": 0.22257383966244726,
      "grad_norm": 0.5262954831123352,
      "learning_rate": 0.00133575815348226,
      "loss": 1.7207,
      "step": 2110
    },
    {
      "epoch": 0.22362869198312235,
      "grad_norm": 0.448763906955719,
      "learning_rate": 0.0013341869635027292,
      "loss": 1.7204,
      "step": 2120
    },
    {
      "epoch": 0.22468354430379747,
      "grad_norm": 0.5699217915534973,
      "learning_rate": 0.001332609227423022,
      "loss": 1.7231,
      "step": 2130
    },
    {
      "epoch": 0.22573839662447256,
      "grad_norm": 0.8906753659248352,
      "learning_rate": 0.0013310249629224417,
      "loss": 1.7173,
      "step": 2140
    },
    {
      "epoch": 0.22679324894514769,
      "grad_norm": 0.4379415810108185,
      "learning_rate": 0.0013294341877534454,
      "loss": 1.7231,
      "step": 2150
    },
    {
      "epoch": 0.22784810126582278,
      "grad_norm": 0.4739811420440674,
      "learning_rate": 0.0013278369197414458,
      "loss": 1.7344,
      "step": 2160
    },
    {
      "epoch": 0.2289029535864979,
      "grad_norm": 0.4546789824962616,
      "learning_rate": 0.0013262331767846104,
      "loss": 1.7187,
      "step": 2170
    },
    {
      "epoch": 0.229957805907173,
      "grad_norm": 0.47932279109954834,
      "learning_rate": 0.0013246229768536628,
      "loss": 1.7022,
      "step": 2180
    },
    {
      "epoch": 0.2310126582278481,
      "grad_norm": 0.5030372142791748,
      "learning_rate": 0.001323006337991679,
      "loss": 1.7183,
      "step": 2190
    },
    {
      "epoch": 0.2320675105485232,
      "grad_norm": 0.5990695357322693,
      "learning_rate": 0.0013213832783138873,
      "loss": 1.7157,
      "step": 2200
    },
    {
      "epoch": 0.23312236286919832,
      "grad_norm": 0.6022274494171143,
      "learning_rate": 0.0013197538160074633,
      "loss": 1.7025,
      "step": 2210
    },
    {
      "epoch": 0.23417721518987342,
      "grad_norm": 0.7774659395217896,
      "learning_rate": 0.0013181179693313283,
      "loss": 1.7191,
      "step": 2220
    },
    {
      "epoch": 0.23523206751054854,
      "grad_norm": 0.863082766532898,
      "learning_rate": 0.0013164757566159428,
      "loss": 1.7104,
      "step": 2230
    },
    {
      "epoch": 0.23628691983122363,
      "grad_norm": 0.8202386498451233,
      "learning_rate": 0.001314827196263102,
      "loss": 1.7061,
      "step": 2240
    },
    {
      "epoch": 0.23734177215189872,
      "grad_norm": 0.9174307584762573,
      "learning_rate": 0.0013131723067457302,
      "loss": 1.7049,
      "step": 2250
    },
    {
      "epoch": 0.23839662447257384,
      "grad_norm": 0.5396402478218079,
      "learning_rate": 0.0013115111066076721,
      "loss": 1.6945,
      "step": 2260
    },
    {
      "epoch": 0.23945147679324894,
      "grad_norm": 0.541278064250946,
      "learning_rate": 0.0013098436144634862,
      "loss": 1.7272,
      "step": 2270
    },
    {
      "epoch": 0.24050632911392406,
      "grad_norm": 0.6107963919639587,
      "learning_rate": 0.0013081698489982364,
      "loss": 1.7114,
      "step": 2280
    },
    {
      "epoch": 0.24156118143459915,
      "grad_norm": 0.6348549127578735,
      "learning_rate": 0.001306489828967282,
      "loss": 1.7009,
      "step": 2290
    },
    {
      "epoch": 0.24261603375527427,
      "grad_norm": 0.5911619663238525,
      "learning_rate": 0.0013048035731960679,
      "loss": 1.6979,
      "step": 2300
    },
    {
      "epoch": 0.24367088607594936,
      "grad_norm": 0.5971229076385498,
      "learning_rate": 0.0013031111005799133,
      "loss": 1.6995,
      "step": 2310
    },
    {
      "epoch": 0.24472573839662448,
      "grad_norm": 0.4567926824092865,
      "learning_rate": 0.0013014124300838004,
      "loss": 1.7054,
      "step": 2320
    },
    {
      "epoch": 0.24578059071729957,
      "grad_norm": 0.4713601768016815,
      "learning_rate": 0.0012997075807421612,
      "loss": 1.6876,
      "step": 2330
    },
    {
      "epoch": 0.2468354430379747,
      "grad_norm": 0.5646883249282837,
      "learning_rate": 0.0012979965716586653,
      "loss": 1.686,
      "step": 2340
    },
    {
      "epoch": 0.2478902953586498,
      "grad_norm": 0.9336428642272949,
      "learning_rate": 0.0012962794220060048,
      "loss": 1.6851,
      "step": 2350
    },
    {
      "epoch": 0.2489451476793249,
      "grad_norm": 0.5376665592193604,
      "learning_rate": 0.0012945561510256801,
      "loss": 1.6934,
      "step": 2360
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8239842653274536,
      "learning_rate": 0.001292826778027784,
      "loss": 1.6911,
      "step": 2370
    },
    {
      "epoch": 0.2510548523206751,
      "grad_norm": 0.7588313817977905,
      "learning_rate": 0.0012910913223907856,
      "loss": 1.6883,
      "step": 2380
    },
    {
      "epoch": 0.2521097046413502,
      "grad_norm": 0.8761573433876038,
      "learning_rate": 0.0012893498035613123,
      "loss": 1.6887,
      "step": 2390
    },
    {
      "epoch": 0.25316455696202533,
      "grad_norm": 0.7116152048110962,
      "learning_rate": 0.001287602241053933,
      "loss": 1.6884,
      "step": 2400
    },
    {
      "epoch": 0.2542194092827004,
      "grad_norm": 0.4818405210971832,
      "learning_rate": 0.0012858486544509392,
      "loss": 1.6917,
      "step": 2410
    },
    {
      "epoch": 0.2552742616033755,
      "grad_norm": 0.7458882331848145,
      "learning_rate": 0.0012840890634021249,
      "loss": 1.6826,
      "step": 2420
    },
    {
      "epoch": 0.2563291139240506,
      "grad_norm": 0.6218952536582947,
      "learning_rate": 0.0012823234876245667,
      "loss": 1.6809,
      "step": 2430
    },
    {
      "epoch": 0.25738396624472576,
      "grad_norm": 0.8475202322006226,
      "learning_rate": 0.0012805519469024035,
      "loss": 1.6909,
      "step": 2440
    },
    {
      "epoch": 0.25843881856540085,
      "grad_norm": 0.6821154356002808,
      "learning_rate": 0.0012787744610866143,
      "loss": 1.6719,
      "step": 2450
    },
    {
      "epoch": 0.25949367088607594,
      "grad_norm": 0.5371242165565491,
      "learning_rate": 0.0012769910500947954,
      "loss": 1.6878,
      "step": 2460
    },
    {
      "epoch": 0.26054852320675104,
      "grad_norm": 0.5225316882133484,
      "learning_rate": 0.0012752017339109376,
      "loss": 1.6784,
      "step": 2470
    },
    {
      "epoch": 0.2616033755274262,
      "grad_norm": 0.5158612728118896,
      "learning_rate": 0.0012734065325852029,
      "loss": 1.6814,
      "step": 2480
    },
    {
      "epoch": 0.2626582278481013,
      "grad_norm": 0.5762125849723816,
      "learning_rate": 0.0012716054662336987,
      "loss": 1.6703,
      "step": 2490
    },
    {
      "epoch": 0.26371308016877637,
      "grad_norm": 0.4870578348636627,
      "learning_rate": 0.001269798555038252,
      "loss": 1.6748,
      "step": 2500
    },
    {
      "epoch": 0.26476793248945146,
      "grad_norm": 0.47660568356513977,
      "learning_rate": 0.0012679858192461864,
      "loss": 1.6742,
      "step": 2510
    },
    {
      "epoch": 0.26582278481012656,
      "grad_norm": 0.6045103073120117,
      "learning_rate": 0.0012661672791700906,
      "loss": 1.6789,
      "step": 2520
    },
    {
      "epoch": 0.2668776371308017,
      "grad_norm": 0.6843808889389038,
      "learning_rate": 0.0012643429551875945,
      "loss": 1.6716,
      "step": 2530
    },
    {
      "epoch": 0.2679324894514768,
      "grad_norm": 0.5017849802970886,
      "learning_rate": 0.0012625128677411388,
      "loss": 1.6752,
      "step": 2540
    },
    {
      "epoch": 0.2689873417721519,
      "grad_norm": 0.47625529766082764,
      "learning_rate": 0.0012606770373377475,
      "loss": 1.6705,
      "step": 2550
    },
    {
      "epoch": 0.270042194092827,
      "grad_norm": 0.5121369957923889,
      "learning_rate": 0.0012588354845487959,
      "loss": 1.6759,
      "step": 2560
    },
    {
      "epoch": 0.27109704641350213,
      "grad_norm": 0.6447870135307312,
      "learning_rate": 0.001256988230009783,
      "loss": 1.6679,
      "step": 2570
    },
    {
      "epoch": 0.2721518987341772,
      "grad_norm": 0.46388521790504456,
      "learning_rate": 0.0012551352944200976,
      "loss": 1.6661,
      "step": 2580
    },
    {
      "epoch": 0.2732067510548523,
      "grad_norm": 0.6746838688850403,
      "learning_rate": 0.0012532766985427874,
      "loss": 1.6722,
      "step": 2590
    },
    {
      "epoch": 0.2742616033755274,
      "grad_norm": 0.5236440300941467,
      "learning_rate": 0.0012514124632043272,
      "loss": 1.6728,
      "step": 2600
    },
    {
      "epoch": 0.27531645569620256,
      "grad_norm": 0.6455425024032593,
      "learning_rate": 0.0012495426092943842,
      "loss": 1.6698,
      "step": 2610
    },
    {
      "epoch": 0.27637130801687765,
      "grad_norm": 0.8995099067687988,
      "learning_rate": 0.0012476671577655845,
      "loss": 1.6676,
      "step": 2620
    },
    {
      "epoch": 0.27742616033755274,
      "grad_norm": 1.1469542980194092,
      "learning_rate": 0.0012457861296332774,
      "loss": 1.6603,
      "step": 2630
    },
    {
      "epoch": 0.27848101265822783,
      "grad_norm": 0.7620439529418945,
      "learning_rate": 0.001243899545975303,
      "loss": 1.6716,
      "step": 2640
    },
    {
      "epoch": 0.2795358649789029,
      "grad_norm": 0.5949313640594482,
      "learning_rate": 0.0012420074279317515,
      "loss": 1.6606,
      "step": 2650
    },
    {
      "epoch": 0.2805907172995781,
      "grad_norm": 0.4743097424507141,
      "learning_rate": 0.0012401097967047298,
      "loss": 1.654,
      "step": 2660
    },
    {
      "epoch": 0.28164556962025317,
      "grad_norm": 0.47151774168014526,
      "learning_rate": 0.001238206673558122,
      "loss": 1.6552,
      "step": 2670
    },
    {
      "epoch": 0.28270042194092826,
      "grad_norm": 0.5160755515098572,
      "learning_rate": 0.0012362980798173526,
      "loss": 1.6547,
      "step": 2680
    },
    {
      "epoch": 0.28375527426160335,
      "grad_norm": 0.5096105337142944,
      "learning_rate": 0.0012343840368691462,
      "loss": 1.6557,
      "step": 2690
    },
    {
      "epoch": 0.2848101265822785,
      "grad_norm": 0.6104464530944824,
      "learning_rate": 0.0012324645661612886,
      "loss": 1.6492,
      "step": 2700
    },
    {
      "epoch": 0.2858649789029536,
      "grad_norm": 0.855845034122467,
      "learning_rate": 0.0012305396892023867,
      "loss": 1.6585,
      "step": 2710
    },
    {
      "epoch": 0.2869198312236287,
      "grad_norm": 0.424182653427124,
      "learning_rate": 0.0012286094275616264,
      "loss": 1.6553,
      "step": 2720
    },
    {
      "epoch": 0.2879746835443038,
      "grad_norm": 0.5019987225532532,
      "learning_rate": 0.0012266738028685318,
      "loss": 1.6343,
      "step": 2730
    },
    {
      "epoch": 0.2890295358649789,
      "grad_norm": 1.1053872108459473,
      "learning_rate": 0.001224732836812723,
      "loss": 1.6509,
      "step": 2740
    },
    {
      "epoch": 0.290084388185654,
      "grad_norm": 0.7308480143547058,
      "learning_rate": 0.0012227865511436724,
      "loss": 1.6543,
      "step": 2750
    },
    {
      "epoch": 0.2911392405063291,
      "grad_norm": 0.451983779668808,
      "learning_rate": 0.001220834967670461,
      "loss": 1.6637,
      "step": 2760
    },
    {
      "epoch": 0.2921940928270042,
      "grad_norm": 0.4617938995361328,
      "learning_rate": 0.0012188781082615346,
      "loss": 1.6566,
      "step": 2770
    },
    {
      "epoch": 0.29324894514767935,
      "grad_norm": 0.5226640105247498,
      "learning_rate": 0.0012169159948444588,
      "loss": 1.6494,
      "step": 2780
    },
    {
      "epoch": 0.29430379746835444,
      "grad_norm": 0.6613280177116394,
      "learning_rate": 0.001214948649405672,
      "loss": 1.6504,
      "step": 2790
    },
    {
      "epoch": 0.29535864978902954,
      "grad_norm": 0.5911545753479004,
      "learning_rate": 0.0012129760939902407,
      "loss": 1.6513,
      "step": 2800
    },
    {
      "epoch": 0.29641350210970463,
      "grad_norm": 0.53304123878479,
      "learning_rate": 0.0012109983507016114,
      "loss": 1.6499,
      "step": 2810
    },
    {
      "epoch": 0.2974683544303797,
      "grad_norm": 0.6235332489013672,
      "learning_rate": 0.0012090154417013636,
      "loss": 1.6537,
      "step": 2820
    },
    {
      "epoch": 0.29852320675105487,
      "grad_norm": 1.1931818723678589,
      "learning_rate": 0.0012070273892089605,
      "loss": 1.6297,
      "step": 2830
    },
    {
      "epoch": 0.29957805907172996,
      "grad_norm": 0.9852569699287415,
      "learning_rate": 0.0012050342155015012,
      "loss": 1.6325,
      "step": 2840
    },
    {
      "epoch": 0.30063291139240506,
      "grad_norm": 0.5030422806739807,
      "learning_rate": 0.0012030359429134707,
      "loss": 1.6376,
      "step": 2850
    },
    {
      "epoch": 0.30168776371308015,
      "grad_norm": 0.6530547142028809,
      "learning_rate": 0.0012010325938364883,
      "loss": 1.6404,
      "step": 2860
    },
    {
      "epoch": 0.3027426160337553,
      "grad_norm": 1.1742103099822998,
      "learning_rate": 0.0011990241907190592,
      "loss": 1.6437,
      "step": 2870
    },
    {
      "epoch": 0.3037974683544304,
      "grad_norm": 0.574685275554657,
      "learning_rate": 0.001197010756066321,
      "loss": 1.626,
      "step": 2880
    },
    {
      "epoch": 0.3048523206751055,
      "grad_norm": 0.5830341577529907,
      "learning_rate": 0.0011949923124397917,
      "loss": 1.6341,
      "step": 2890
    },
    {
      "epoch": 0.3059071729957806,
      "grad_norm": 0.484795480966568,
      "learning_rate": 0.001192968882457118,
      "loss": 1.6394,
      "step": 2900
    },
    {
      "epoch": 0.3069620253164557,
      "grad_norm": 1.126344919204712,
      "learning_rate": 0.001190940488791821,
      "loss": 1.6474,
      "step": 2910
    },
    {
      "epoch": 0.3080168776371308,
      "grad_norm": 0.5768723487854004,
      "learning_rate": 0.0011889071541730419,
      "loss": 1.6351,
      "step": 2920
    },
    {
      "epoch": 0.3090717299578059,
      "grad_norm": 0.5678271651268005,
      "learning_rate": 0.001186868901385288,
      "loss": 1.6264,
      "step": 2930
    },
    {
      "epoch": 0.310126582278481,
      "grad_norm": 0.49463793635368347,
      "learning_rate": 0.001184825753268177,
      "loss": 1.6308,
      "step": 2940
    },
    {
      "epoch": 0.3111814345991561,
      "grad_norm": 0.9706162810325623,
      "learning_rate": 0.0011827777327161814,
      "loss": 1.6408,
      "step": 2950
    },
    {
      "epoch": 0.31223628691983124,
      "grad_norm": 0.9023301601409912,
      "learning_rate": 0.0011807248626783714,
      "loss": 1.6304,
      "step": 2960
    },
    {
      "epoch": 0.31329113924050633,
      "grad_norm": 0.5531619191169739,
      "learning_rate": 0.0011786671661581584,
      "loss": 1.6191,
      "step": 2970
    },
    {
      "epoch": 0.3143459915611814,
      "grad_norm": 0.5522653460502625,
      "learning_rate": 0.001176604666213036,
      "loss": 1.6213,
      "step": 2980
    },
    {
      "epoch": 0.3154008438818565,
      "grad_norm": 0.713241457939148,
      "learning_rate": 0.0011745373859543236,
      "loss": 1.639,
      "step": 2990
    },
    {
      "epoch": 0.31645569620253167,
      "grad_norm": 0.4535176753997803,
      "learning_rate": 0.0011724653485469063,
      "loss": 1.6168,
      "step": 3000
    },
    {
      "epoch": 0.31751054852320676,
      "grad_norm": 0.6761485934257507,
      "learning_rate": 0.0011703885772089743,
      "loss": 1.6242,
      "step": 3010
    },
    {
      "epoch": 0.31856540084388185,
      "grad_norm": 0.970213770866394,
      "learning_rate": 0.0011683070952117646,
      "loss": 1.624,
      "step": 3020
    },
    {
      "epoch": 0.31962025316455694,
      "grad_norm": 0.5876306295394897,
      "learning_rate": 0.0011662209258792998,
      "loss": 1.6137,
      "step": 3030
    },
    {
      "epoch": 0.3206751054852321,
      "grad_norm": 0.5551946759223938,
      "learning_rate": 0.0011641300925881257,
      "loss": 1.6243,
      "step": 3040
    },
    {
      "epoch": 0.3217299578059072,
      "grad_norm": 0.5187597870826721,
      "learning_rate": 0.0011620346187670501,
      "loss": 1.6191,
      "step": 3050
    },
    {
      "epoch": 0.3227848101265823,
      "grad_norm": 0.5157437920570374,
      "learning_rate": 0.0011599345278968806,
      "loss": 1.6302,
      "step": 3060
    },
    {
      "epoch": 0.32383966244725737,
      "grad_norm": 0.47498974204063416,
      "learning_rate": 0.0011578298435101604,
      "loss": 1.6186,
      "step": 3070
    },
    {
      "epoch": 0.32489451476793246,
      "grad_norm": 1.1298749446868896,
      "learning_rate": 0.0011557205891909062,
      "loss": 1.607,
      "step": 3080
    },
    {
      "epoch": 0.3259493670886076,
      "grad_norm": 0.46439072489738464,
      "learning_rate": 0.0011536067885743423,
      "loss": 1.6239,
      "step": 3090
    },
    {
      "epoch": 0.3270042194092827,
      "grad_norm": 0.5722910165786743,
      "learning_rate": 0.001151488465346637,
      "loss": 1.6094,
      "step": 3100
    },
    {
      "epoch": 0.3280590717299578,
      "grad_norm": 0.6602784395217896,
      "learning_rate": 0.0011493656432446362,
      "loss": 1.6138,
      "step": 3110
    },
    {
      "epoch": 0.3291139240506329,
      "grad_norm": 0.49073052406311035,
      "learning_rate": 0.0011472383460555983,
      "loss": 1.6163,
      "step": 3120
    },
    {
      "epoch": 0.33016877637130804,
      "grad_norm": 0.6607328653335571,
      "learning_rate": 0.001145106597616927,
      "loss": 1.6252,
      "step": 3130
    },
    {
      "epoch": 0.33122362869198313,
      "grad_norm": 0.9633255004882812,
      "learning_rate": 0.001142970421815904,
      "loss": 1.6057,
      "step": 3140
    },
    {
      "epoch": 0.3322784810126582,
      "grad_norm": 0.437183678150177,
      "learning_rate": 0.0011408298425894226,
      "loss": 1.6151,
      "step": 3150
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.46504244208335876,
      "learning_rate": 0.0011386848839237186,
      "loss": 1.6137,
      "step": 3160
    },
    {
      "epoch": 0.33438818565400846,
      "grad_norm": 0.5309893488883972,
      "learning_rate": 0.0011365355698541005,
      "loss": 1.6169,
      "step": 3170
    },
    {
      "epoch": 0.33544303797468356,
      "grad_norm": 0.5407806038856506,
      "learning_rate": 0.0011343819244646824,
      "loss": 1.6134,
      "step": 3180
    },
    {
      "epoch": 0.33649789029535865,
      "grad_norm": 0.4959893524646759,
      "learning_rate": 0.001132223971888112,
      "loss": 1.612,
      "step": 3190
    },
    {
      "epoch": 0.33755274261603374,
      "grad_norm": 0.6287081837654114,
      "learning_rate": 0.0011300617363053024,
      "loss": 1.6025,
      "step": 3200
    },
    {
      "epoch": 0.33860759493670883,
      "grad_norm": 0.7256467342376709,
      "learning_rate": 0.0011278952419451586,
      "loss": 1.6203,
      "step": 3210
    },
    {
      "epoch": 0.339662447257384,
      "grad_norm": 0.5274771451950073,
      "learning_rate": 0.0011257245130843077,
      "loss": 1.611,
      "step": 3220
    },
    {
      "epoch": 0.3407172995780591,
      "grad_norm": 0.6178369522094727,
      "learning_rate": 0.0011235495740468265,
      "loss": 1.6042,
      "step": 3230
    },
    {
      "epoch": 0.34177215189873417,
      "grad_norm": 0.4739209711551666,
      "learning_rate": 0.0011213704492039694,
      "loss": 1.5843,
      "step": 3240
    },
    {
      "epoch": 0.34282700421940926,
      "grad_norm": 0.4660987854003906,
      "learning_rate": 0.001119187162973894,
      "loss": 1.6075,
      "step": 3250
    },
    {
      "epoch": 0.3438818565400844,
      "grad_norm": 0.4897654056549072,
      "learning_rate": 0.001116999739821388,
      "loss": 1.5964,
      "step": 3260
    },
    {
      "epoch": 0.3449367088607595,
      "grad_norm": 0.5139145851135254,
      "learning_rate": 0.0011148082042575968,
      "loss": 1.6137,
      "step": 3270
    },
    {
      "epoch": 0.3459915611814346,
      "grad_norm": 0.6091996431350708,
      "learning_rate": 0.0011126125808397461,
      "loss": 1.6136,
      "step": 3280
    },
    {
      "epoch": 0.3470464135021097,
      "grad_norm": 0.4915454387664795,
      "learning_rate": 0.0011104128941708683,
      "loss": 1.5932,
      "step": 3290
    },
    {
      "epoch": 0.34810126582278483,
      "grad_norm": 0.8600702881813049,
      "learning_rate": 0.001108209168899527,
      "loss": 1.6074,
      "step": 3300
    },
    {
      "epoch": 0.3491561181434599,
      "grad_norm": 0.6012086868286133,
      "learning_rate": 0.0011060014297195396,
      "loss": 1.6054,
      "step": 3310
    },
    {
      "epoch": 0.350210970464135,
      "grad_norm": 0.7052494883537292,
      "learning_rate": 0.0011037897013697015,
      "loss": 1.6105,
      "step": 3320
    },
    {
      "epoch": 0.3512658227848101,
      "grad_norm": 0.6381321549415588,
      "learning_rate": 0.0011015740086335092,
      "loss": 1.6029,
      "step": 3330
    },
    {
      "epoch": 0.35232067510548526,
      "grad_norm": 0.5391980409622192,
      "learning_rate": 0.0010993543763388814,
      "loss": 1.6018,
      "step": 3340
    },
    {
      "epoch": 0.35337552742616035,
      "grad_norm": 0.47316044569015503,
      "learning_rate": 0.0010971308293578814,
      "loss": 1.5906,
      "step": 3350
    },
    {
      "epoch": 0.35443037974683544,
      "grad_norm": 0.4353296160697937,
      "learning_rate": 0.0010949033926064397,
      "loss": 1.5893,
      "step": 3360
    },
    {
      "epoch": 0.35548523206751054,
      "grad_norm": 0.4708898365497589,
      "learning_rate": 0.0010926720910440725,
      "loss": 1.6018,
      "step": 3370
    },
    {
      "epoch": 0.35654008438818563,
      "grad_norm": 0.5538137555122375,
      "learning_rate": 0.001090436949673603,
      "loss": 1.5953,
      "step": 3380
    },
    {
      "epoch": 0.3575949367088608,
      "grad_norm": 0.4721989035606384,
      "learning_rate": 0.0010881979935408815,
      "loss": 1.5876,
      "step": 3390
    },
    {
      "epoch": 0.35864978902953587,
      "grad_norm": 0.766671895980835,
      "learning_rate": 0.0010859552477345052,
      "loss": 1.6006,
      "step": 3400
    },
    {
      "epoch": 0.35970464135021096,
      "grad_norm": 0.5119830369949341,
      "learning_rate": 0.001083708737385536,
      "loss": 1.5961,
      "step": 3410
    },
    {
      "epoch": 0.36075949367088606,
      "grad_norm": 0.5428466796875,
      "learning_rate": 0.0010814584876672187,
      "loss": 1.5788,
      "step": 3420
    },
    {
      "epoch": 0.3618143459915612,
      "grad_norm": 0.6007835865020752,
      "learning_rate": 0.0010792045237947008,
      "loss": 1.5947,
      "step": 3430
    },
    {
      "epoch": 0.3628691983122363,
      "grad_norm": 0.6104546785354614,
      "learning_rate": 0.0010769468710247478,
      "loss": 1.59,
      "step": 3440
    },
    {
      "epoch": 0.3639240506329114,
      "grad_norm": 0.4558543264865875,
      "learning_rate": 0.0010746855546554612,
      "loss": 1.5862,
      "step": 3450
    },
    {
      "epoch": 0.3649789029535865,
      "grad_norm": 0.49403393268585205,
      "learning_rate": 0.0010724206000259954,
      "loss": 1.5795,
      "step": 3460
    },
    {
      "epoch": 0.36603375527426163,
      "grad_norm": 0.547333836555481,
      "learning_rate": 0.0010701520325162727,
      "loss": 1.5939,
      "step": 3470
    },
    {
      "epoch": 0.3670886075949367,
      "grad_norm": 0.4920315444469452,
      "learning_rate": 0.0010678798775467001,
      "loss": 1.5966,
      "step": 3480
    },
    {
      "epoch": 0.3681434599156118,
      "grad_norm": 1.1386079788208008,
      "learning_rate": 0.0010656041605778832,
      "loss": 1.5955,
      "step": 3490
    },
    {
      "epoch": 0.3691983122362869,
      "grad_norm": 0.5450473427772522,
      "learning_rate": 0.001063324907110342,
      "loss": 1.5729,
      "step": 3500
    },
    {
      "epoch": 0.370253164556962,
      "grad_norm": 0.5650832653045654,
      "learning_rate": 0.0010610421426842241,
      "loss": 1.5917,
      "step": 3510
    },
    {
      "epoch": 0.37130801687763715,
      "grad_norm": 0.5141381621360779,
      "learning_rate": 0.00105875589287902,
      "loss": 1.5911,
      "step": 3520
    },
    {
      "epoch": 0.37236286919831224,
      "grad_norm": 0.910280168056488,
      "learning_rate": 0.0010564661833132752,
      "loss": 1.5994,
      "step": 3530
    },
    {
      "epoch": 0.37341772151898733,
      "grad_norm": 0.9338151812553406,
      "learning_rate": 0.001054173039644303,
      "loss": 1.5933,
      "step": 3540
    },
    {
      "epoch": 0.3744725738396624,
      "grad_norm": 0.5695720911026001,
      "learning_rate": 0.0010518764875678981,
      "loss": 1.5809,
      "step": 3550
    },
    {
      "epoch": 0.3755274261603376,
      "grad_norm": 0.6266768574714661,
      "learning_rate": 0.001049576552818048,
      "loss": 1.5838,
      "step": 3560
    },
    {
      "epoch": 0.37658227848101267,
      "grad_norm": 0.6025685667991638,
      "learning_rate": 0.0010472732611666448,
      "loss": 1.5797,
      "step": 3570
    },
    {
      "epoch": 0.37763713080168776,
      "grad_norm": 0.6261733770370483,
      "learning_rate": 0.0010449666384231954,
      "loss": 1.5781,
      "step": 3580
    },
    {
      "epoch": 0.37869198312236285,
      "grad_norm": 0.48340892791748047,
      "learning_rate": 0.0010426567104345346,
      "loss": 1.5916,
      "step": 3590
    },
    {
      "epoch": 0.379746835443038,
      "grad_norm": 0.49008363485336304,
      "learning_rate": 0.0010403435030845332,
      "loss": 1.574,
      "step": 3600
    },
    {
      "epoch": 0.3808016877637131,
      "grad_norm": 1.1611214876174927,
      "learning_rate": 0.0010380270422938093,
      "loss": 1.5838,
      "step": 3610
    },
    {
      "epoch": 0.3818565400843882,
      "grad_norm": 0.9247241616249084,
      "learning_rate": 0.0010357073540194362,
      "loss": 1.573,
      "step": 3620
    },
    {
      "epoch": 0.3829113924050633,
      "grad_norm": 0.6713014245033264,
      "learning_rate": 0.001033384464254655,
      "loss": 1.5775,
      "step": 3630
    },
    {
      "epoch": 0.38396624472573837,
      "grad_norm": 0.5715181827545166,
      "learning_rate": 0.001031058399028579,
      "loss": 1.5749,
      "step": 3640
    },
    {
      "epoch": 0.3850210970464135,
      "grad_norm": 0.5150052309036255,
      "learning_rate": 0.001028729184405905,
      "loss": 1.5846,
      "step": 3650
    },
    {
      "epoch": 0.3860759493670886,
      "grad_norm": 0.5131828784942627,
      "learning_rate": 0.0010263968464866201,
      "loss": 1.5758,
      "step": 3660
    },
    {
      "epoch": 0.3871308016877637,
      "grad_norm": 0.5704460740089417,
      "learning_rate": 0.0010240614114057098,
      "loss": 1.5706,
      "step": 3670
    },
    {
      "epoch": 0.3881856540084388,
      "grad_norm": 0.535331130027771,
      "learning_rate": 0.001021722905332864,
      "loss": 1.5751,
      "step": 3680
    },
    {
      "epoch": 0.38924050632911394,
      "grad_norm": 0.6319772601127625,
      "learning_rate": 0.0010193813544721855,
      "loss": 1.5818,
      "step": 3690
    },
    {
      "epoch": 0.39029535864978904,
      "grad_norm": 0.8062748312950134,
      "learning_rate": 0.001017036785061895,
      "loss": 1.586,
      "step": 3700
    },
    {
      "epoch": 0.39135021097046413,
      "grad_norm": 0.49930211901664734,
      "learning_rate": 0.0010146892233740376,
      "loss": 1.5723,
      "step": 3710
    },
    {
      "epoch": 0.3924050632911392,
      "grad_norm": 0.4763012230396271,
      "learning_rate": 0.0010123386957141883,
      "loss": 1.5535,
      "step": 3720
    },
    {
      "epoch": 0.39345991561181437,
      "grad_norm": 0.86577969789505,
      "learning_rate": 0.0010099852284211573,
      "loss": 1.5716,
      "step": 3730
    },
    {
      "epoch": 0.39451476793248946,
      "grad_norm": 0.5172073841094971,
      "learning_rate": 0.0010076288478666944,
      "loss": 1.5707,
      "step": 3740
    },
    {
      "epoch": 0.39556962025316456,
      "grad_norm": 0.5007083415985107,
      "learning_rate": 0.0010052695804551946,
      "loss": 1.5751,
      "step": 3750
    },
    {
      "epoch": 0.39662447257383965,
      "grad_norm": 0.6111957430839539,
      "learning_rate": 0.0010029074526234014,
      "loss": 1.5652,
      "step": 3760
    },
    {
      "epoch": 0.39767932489451474,
      "grad_norm": 0.6261873841285706,
      "learning_rate": 0.0010005424908401104,
      "loss": 1.5806,
      "step": 3770
    },
    {
      "epoch": 0.3987341772151899,
      "grad_norm": 0.5061927437782288,
      "learning_rate": 0.0009981747216058728,
      "loss": 1.5665,
      "step": 3780
    },
    {
      "epoch": 0.399789029535865,
      "grad_norm": 0.49627190828323364,
      "learning_rate": 0.0009958041714526998,
      "loss": 1.5696,
      "step": 3790
    },
    {
      "epoch": 0.4008438818565401,
      "grad_norm": 0.6029568314552307,
      "learning_rate": 0.0009934308669437627,
      "loss": 1.5745,
      "step": 3800
    },
    {
      "epoch": 0.40189873417721517,
      "grad_norm": 0.47557276487350464,
      "learning_rate": 0.0009910548346730972,
      "loss": 1.5624,
      "step": 3810
    },
    {
      "epoch": 0.4029535864978903,
      "grad_norm": 0.5486270785331726,
      "learning_rate": 0.0009886761012653062,
      "loss": 1.552,
      "step": 3820
    },
    {
      "epoch": 0.4040084388185654,
      "grad_norm": 0.6182465553283691,
      "learning_rate": 0.000986294693375258,
      "loss": 1.5581,
      "step": 3830
    },
    {
      "epoch": 0.4050632911392405,
      "grad_norm": 0.47843965888023376,
      "learning_rate": 0.000983910637687791,
      "loss": 1.574,
      "step": 3840
    },
    {
      "epoch": 0.4061181434599156,
      "grad_norm": 0.5410100221633911,
      "learning_rate": 0.0009815239609174138,
      "loss": 1.5578,
      "step": 3850
    },
    {
      "epoch": 0.40717299578059074,
      "grad_norm": 0.9322824478149414,
      "learning_rate": 0.0009791346898080043,
      "loss": 1.5709,
      "step": 3860
    },
    {
      "epoch": 0.40822784810126583,
      "grad_norm": 0.5444523692131042,
      "learning_rate": 0.0009767428511325122,
      "loss": 1.5607,
      "step": 3870
    },
    {
      "epoch": 0.4092827004219409,
      "grad_norm": 0.48717835545539856,
      "learning_rate": 0.0009743484716926576,
      "loss": 1.5495,
      "step": 3880
    },
    {
      "epoch": 0.410337552742616,
      "grad_norm": 0.6598711013793945,
      "learning_rate": 0.0009719515783186319,
      "loss": 1.5493,
      "step": 3890
    },
    {
      "epoch": 0.41139240506329117,
      "grad_norm": 0.6580759882926941,
      "learning_rate": 0.0009695521978687951,
      "loss": 1.5531,
      "step": 3900
    },
    {
      "epoch": 0.41244725738396626,
      "grad_norm": 0.5247423648834229,
      "learning_rate": 0.0009671503572293767,
      "loss": 1.5563,
      "step": 3910
    },
    {
      "epoch": 0.41350210970464135,
      "grad_norm": 0.5178701877593994,
      "learning_rate": 0.0009647460833141742,
      "loss": 1.5511,
      "step": 3920
    },
    {
      "epoch": 0.41455696202531644,
      "grad_norm": 0.5494204759597778,
      "learning_rate": 0.0009623394030642507,
      "loss": 1.5573,
      "step": 3930
    },
    {
      "epoch": 0.41561181434599154,
      "grad_norm": 0.7308394908905029,
      "learning_rate": 0.0009599303434476334,
      "loss": 1.5499,
      "step": 3940
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 1.116909146308899,
      "learning_rate": 0.0009575189314590118,
      "loss": 1.5572,
      "step": 3950
    },
    {
      "epoch": 0.4177215189873418,
      "grad_norm": 0.5184165835380554,
      "learning_rate": 0.0009551051941194346,
      "loss": 1.5589,
      "step": 3960
    },
    {
      "epoch": 0.41877637130801687,
      "grad_norm": 0.6424084305763245,
      "learning_rate": 0.0009526891584760071,
      "loss": 1.5447,
      "step": 3970
    },
    {
      "epoch": 0.41983122362869196,
      "grad_norm": 0.46564581990242004,
      "learning_rate": 0.0009502708516015889,
      "loss": 1.5557,
      "step": 3980
    },
    {
      "epoch": 0.4208860759493671,
      "grad_norm": 0.5686947703361511,
      "learning_rate": 0.0009478503005944888,
      "loss": 1.5433,
      "step": 3990
    },
    {
      "epoch": 0.4219409282700422,
      "grad_norm": 0.5193183422088623,
      "learning_rate": 0.0009454275325781632,
      "loss": 1.5623,
      "step": 4000
    },
    {
      "epoch": 0.4229957805907173,
      "grad_norm": 0.5051531791687012,
      "learning_rate": 0.0009430025747009104,
      "loss": 1.5481,
      "step": 4010
    },
    {
      "epoch": 0.4240506329113924,
      "grad_norm": 0.48578837513923645,
      "learning_rate": 0.0009405754541355677,
      "loss": 1.5489,
      "step": 4020
    },
    {
      "epoch": 0.42510548523206754,
      "grad_norm": 0.48167306184768677,
      "learning_rate": 0.0009381461980792061,
      "loss": 1.5411,
      "step": 4030
    },
    {
      "epoch": 0.42616033755274263,
      "grad_norm": 0.6252226233482361,
      "learning_rate": 0.0009357148337528256,
      "loss": 1.5525,
      "step": 4040
    },
    {
      "epoch": 0.4272151898734177,
      "grad_norm": 0.6204479336738586,
      "learning_rate": 0.0009332813884010511,
      "loss": 1.5524,
      "step": 4050
    },
    {
      "epoch": 0.4282700421940928,
      "grad_norm": 0.4954337775707245,
      "learning_rate": 0.0009308458892918259,
      "loss": 1.5567,
      "step": 4060
    },
    {
      "epoch": 0.4293248945147679,
      "grad_norm": 0.625588595867157,
      "learning_rate": 0.0009284083637161064,
      "loss": 1.5512,
      "step": 4070
    },
    {
      "epoch": 0.43037974683544306,
      "grad_norm": 0.5604734420776367,
      "learning_rate": 0.0009259688389875574,
      "loss": 1.5569,
      "step": 4080
    },
    {
      "epoch": 0.43143459915611815,
      "grad_norm": 0.521360456943512,
      "learning_rate": 0.0009235273424422442,
      "loss": 1.5519,
      "step": 4090
    },
    {
      "epoch": 0.43248945147679324,
      "grad_norm": 0.4531897008419037,
      "learning_rate": 0.0009210839014383282,
      "loss": 1.5469,
      "step": 4100
    },
    {
      "epoch": 0.43354430379746833,
      "grad_norm": 0.47998490929603577,
      "learning_rate": 0.0009186385433557584,
      "loss": 1.5399,
      "step": 4110
    },
    {
      "epoch": 0.4345991561181435,
      "grad_norm": 0.8988192081451416,
      "learning_rate": 0.0009161912955959668,
      "loss": 1.5473,
      "step": 4120
    },
    {
      "epoch": 0.4356540084388186,
      "grad_norm": 0.5566585063934326,
      "learning_rate": 0.000913742185581559,
      "loss": 1.5383,
      "step": 4130
    },
    {
      "epoch": 0.43670886075949367,
      "grad_norm": 0.6921567320823669,
      "learning_rate": 0.0009112912407560086,
      "loss": 1.5394,
      "step": 4140
    },
    {
      "epoch": 0.43776371308016876,
      "grad_norm": 0.5351143479347229,
      "learning_rate": 0.0009088384885833495,
      "loss": 1.5453,
      "step": 4150
    },
    {
      "epoch": 0.4388185654008439,
      "grad_norm": 0.5496554374694824,
      "learning_rate": 0.000906383956547867,
      "loss": 1.5375,
      "step": 4160
    },
    {
      "epoch": 0.439873417721519,
      "grad_norm": 0.47633862495422363,
      "learning_rate": 0.0009039276721537915,
      "loss": 1.5394,
      "step": 4170
    },
    {
      "epoch": 0.4409282700421941,
      "grad_norm": 0.5072779059410095,
      "learning_rate": 0.0009014696629249886,
      "loss": 1.5341,
      "step": 4180
    },
    {
      "epoch": 0.4419831223628692,
      "grad_norm": 0.7401310205459595,
      "learning_rate": 0.0008990099564046522,
      "loss": 1.5441,
      "step": 4190
    },
    {
      "epoch": 0.4430379746835443,
      "grad_norm": 0.5734917521476746,
      "learning_rate": 0.0008965485801549946,
      "loss": 1.5435,
      "step": 4200
    },
    {
      "epoch": 0.4440928270042194,
      "grad_norm": 0.4642139971256256,
      "learning_rate": 0.000894085561756939,
      "loss": 1.5331,
      "step": 4210
    },
    {
      "epoch": 0.4451476793248945,
      "grad_norm": 0.7488202452659607,
      "learning_rate": 0.0008916209288098088,
      "loss": 1.5402,
      "step": 4220
    },
    {
      "epoch": 0.4462025316455696,
      "grad_norm": 0.5966841578483582,
      "learning_rate": 0.0008891547089310198,
      "loss": 1.5368,
      "step": 4230
    },
    {
      "epoch": 0.4472573839662447,
      "grad_norm": 0.6754875183105469,
      "learning_rate": 0.0008866869297557699,
      "loss": 1.5364,
      "step": 4240
    },
    {
      "epoch": 0.44831223628691985,
      "grad_norm": 0.5848161578178406,
      "learning_rate": 0.0008842176189367299,
      "loss": 1.5403,
      "step": 4250
    },
    {
      "epoch": 0.44936708860759494,
      "grad_norm": 0.7788020372390747,
      "learning_rate": 0.0008817468041437329,
      "loss": 1.5455,
      "step": 4260
    },
    {
      "epoch": 0.45042194092827004,
      "grad_norm": 0.46313008666038513,
      "learning_rate": 0.0008792745130634654,
      "loss": 1.5328,
      "step": 4270
    },
    {
      "epoch": 0.45147679324894513,
      "grad_norm": 0.5829064846038818,
      "learning_rate": 0.0008768007733991561,
      "loss": 1.5294,
      "step": 4280
    },
    {
      "epoch": 0.4525316455696203,
      "grad_norm": 0.4733564257621765,
      "learning_rate": 0.0008743256128702658,
      "loss": 1.5399,
      "step": 4290
    },
    {
      "epoch": 0.45358649789029537,
      "grad_norm": 0.6273133158683777,
      "learning_rate": 0.0008718490592121768,
      "loss": 1.5368,
      "step": 4300
    },
    {
      "epoch": 0.45464135021097046,
      "grad_norm": 0.45090365409851074,
      "learning_rate": 0.0008693711401758822,
      "loss": 1.5354,
      "step": 4310
    },
    {
      "epoch": 0.45569620253164556,
      "grad_norm": 1.128738522529602,
      "learning_rate": 0.0008668918835276747,
      "loss": 1.5302,
      "step": 4320
    },
    {
      "epoch": 0.45675105485232065,
      "grad_norm": 0.5392206311225891,
      "learning_rate": 0.0008644113170488355,
      "loss": 1.5375,
      "step": 4330
    },
    {
      "epoch": 0.4578059071729958,
      "grad_norm": 0.6845002174377441,
      "learning_rate": 0.0008619294685353235,
      "loss": 1.5274,
      "step": 4340
    },
    {
      "epoch": 0.4588607594936709,
      "grad_norm": 0.8332899808883667,
      "learning_rate": 0.0008594463657974627,
      "loss": 1.5286,
      "step": 4350
    },
    {
      "epoch": 0.459915611814346,
      "grad_norm": 0.7652190327644348,
      "learning_rate": 0.0008569620366596322,
      "loss": 1.5463,
      "step": 4360
    },
    {
      "epoch": 0.4609704641350211,
      "grad_norm": 0.6484683752059937,
      "learning_rate": 0.000854476508959953,
      "loss": 1.5245,
      "step": 4370
    },
    {
      "epoch": 0.4620253164556962,
      "grad_norm": 0.519527792930603,
      "learning_rate": 0.0008519898105499762,
      "loss": 1.5303,
      "step": 4380
    },
    {
      "epoch": 0.4630801687763713,
      "grad_norm": 0.4845265746116638,
      "learning_rate": 0.0008495019692943721,
      "loss": 1.5235,
      "step": 4390
    },
    {
      "epoch": 0.4641350210970464,
      "grad_norm": 0.48636671900749207,
      "learning_rate": 0.0008470130130706166,
      "loss": 1.529,
      "step": 4400
    },
    {
      "epoch": 0.4651898734177215,
      "grad_norm": 0.8035237193107605,
      "learning_rate": 0.0008445229697686795,
      "loss": 1.5423,
      "step": 4410
    },
    {
      "epoch": 0.46624472573839665,
      "grad_norm": 0.5630937814712524,
      "learning_rate": 0.0008420318672907119,
      "loss": 1.5273,
      "step": 4420
    },
    {
      "epoch": 0.46729957805907174,
      "grad_norm": 0.7922337055206299,
      "learning_rate": 0.0008395397335507334,
      "loss": 1.5282,
      "step": 4430
    },
    {
      "epoch": 0.46835443037974683,
      "grad_norm": 0.48462677001953125,
      "learning_rate": 0.0008370465964743196,
      "loss": 1.526,
      "step": 4440
    },
    {
      "epoch": 0.4694092827004219,
      "grad_norm": 0.49322086572647095,
      "learning_rate": 0.0008345524839982886,
      "loss": 1.533,
      "step": 4450
    },
    {
      "epoch": 0.4704641350210971,
      "grad_norm": 0.47821831703186035,
      "learning_rate": 0.0008320574240703886,
      "loss": 1.5191,
      "step": 4460
    },
    {
      "epoch": 0.47151898734177217,
      "grad_norm": 0.4673146903514862,
      "learning_rate": 0.0008295614446489842,
      "loss": 1.5279,
      "step": 4470
    },
    {
      "epoch": 0.47257383966244726,
      "grad_norm": 0.5318535566329956,
      "learning_rate": 0.0008270645737027441,
      "loss": 1.531,
      "step": 4480
    },
    {
      "epoch": 0.47362869198312235,
      "grad_norm": 0.7749707698822021,
      "learning_rate": 0.0008245668392103259,
      "loss": 1.5219,
      "step": 4490
    },
    {
      "epoch": 0.47468354430379744,
      "grad_norm": 0.7400705218315125,
      "learning_rate": 0.0008220682691600645,
      "loss": 1.5198,
      "step": 4500
    },
    {
      "epoch": 0.4757383966244726,
      "grad_norm": 0.4604952335357666,
      "learning_rate": 0.0008195688915496571,
      "loss": 1.5253,
      "step": 4510
    },
    {
      "epoch": 0.4767932489451477,
      "grad_norm": 0.6762343049049377,
      "learning_rate": 0.0008170687343858506,
      "loss": 1.5081,
      "step": 4520
    },
    {
      "epoch": 0.4778481012658228,
      "grad_norm": 0.4940911531448364,
      "learning_rate": 0.0008145678256841265,
      "loss": 1.5261,
      "step": 4530
    },
    {
      "epoch": 0.47890295358649787,
      "grad_norm": 0.5079637169837952,
      "learning_rate": 0.0008120661934683879,
      "loss": 1.5261,
      "step": 4540
    },
    {
      "epoch": 0.479957805907173,
      "grad_norm": 0.6899538636207581,
      "learning_rate": 0.0008095638657706456,
      "loss": 1.524,
      "step": 4550
    },
    {
      "epoch": 0.4810126582278481,
      "grad_norm": 0.7249855995178223,
      "learning_rate": 0.000807060870630703,
      "loss": 1.5169,
      "step": 4560
    },
    {
      "epoch": 0.4820675105485232,
      "grad_norm": 0.5128796100616455,
      "learning_rate": 0.000804557236095843,
      "loss": 1.4958,
      "step": 4570
    },
    {
      "epoch": 0.4831223628691983,
      "grad_norm": 0.5689982771873474,
      "learning_rate": 0.0008020529902205129,
      "loss": 1.5212,
      "step": 4580
    },
    {
      "epoch": 0.48417721518987344,
      "grad_norm": 0.48652711510658264,
      "learning_rate": 0.0007995481610660108,
      "loss": 1.5055,
      "step": 4590
    },
    {
      "epoch": 0.48523206751054854,
      "grad_norm": 0.6480730175971985,
      "learning_rate": 0.0007970427767001702,
      "loss": 1.523,
      "step": 4600
    },
    {
      "epoch": 0.48628691983122363,
      "grad_norm": 0.6398733854293823,
      "learning_rate": 0.0007945368651970464,
      "loss": 1.525,
      "step": 4610
    },
    {
      "epoch": 0.4873417721518987,
      "grad_norm": 0.4856146574020386,
      "learning_rate": 0.0007920304546366013,
      "loss": 1.5048,
      "step": 4620
    },
    {
      "epoch": 0.4883966244725738,
      "grad_norm": 0.5048703551292419,
      "learning_rate": 0.000789523573104389,
      "loss": 1.519,
      "step": 4630
    },
    {
      "epoch": 0.48945147679324896,
      "grad_norm": 0.4923306107521057,
      "learning_rate": 0.0007870162486912414,
      "loss": 1.5219,
      "step": 4640
    },
    {
      "epoch": 0.49050632911392406,
      "grad_norm": 0.502138078212738,
      "learning_rate": 0.0007845085094929527,
      "loss": 1.523,
      "step": 4650
    },
    {
      "epoch": 0.49156118143459915,
      "grad_norm": 0.5037270188331604,
      "learning_rate": 0.0007820003836099649,
      "loss": 1.5126,
      "step": 4660
    },
    {
      "epoch": 0.49261603375527424,
      "grad_norm": 0.5258697271347046,
      "learning_rate": 0.0007794918991470537,
      "loss": 1.5105,
      "step": 4670
    },
    {
      "epoch": 0.4936708860759494,
      "grad_norm": 0.5824125409126282,
      "learning_rate": 0.0007769830842130119,
      "loss": 1.507,
      "step": 4680
    },
    {
      "epoch": 0.4947257383966245,
      "grad_norm": 0.6700602173805237,
      "learning_rate": 0.0007744739669203361,
      "loss": 1.505,
      "step": 4690
    },
    {
      "epoch": 0.4957805907172996,
      "grad_norm": 0.5182927846908569,
      "learning_rate": 0.0007719645753849108,
      "loss": 1.5141,
      "step": 4700
    },
    {
      "epoch": 0.49683544303797467,
      "grad_norm": 0.463286429643631,
      "learning_rate": 0.0007694549377256932,
      "loss": 1.5154,
      "step": 4710
    },
    {
      "epoch": 0.4978902953586498,
      "grad_norm": 0.9340118169784546,
      "learning_rate": 0.0007669450820643987,
      "loss": 1.5101,
      "step": 4720
    },
    {
      "epoch": 0.4989451476793249,
      "grad_norm": 0.5239022970199585,
      "learning_rate": 0.0007644350365251855,
      "loss": 1.5085,
      "step": 4730
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.531623363494873,
      "learning_rate": 0.0007619248292343399,
      "loss": 1.5059,
      "step": 4740
    },
    {
      "epoch": 0.5010548523206751,
      "grad_norm": 0.5866503715515137,
      "learning_rate": 0.0007594144883199599,
      "loss": 1.5241,
      "step": 4750
    },
    {
      "epoch": 0.5021097046413502,
      "grad_norm": 0.551724910736084,
      "learning_rate": 0.0007569040419116413,
      "loss": 1.508,
      "step": 4760
    },
    {
      "epoch": 0.5031645569620253,
      "grad_norm": 0.5427130460739136,
      "learning_rate": 0.000754393518140162,
      "loss": 1.5053,
      "step": 4770
    },
    {
      "epoch": 0.5042194092827004,
      "grad_norm": 0.4989750385284424,
      "learning_rate": 0.0007518829451371665,
      "loss": 1.5182,
      "step": 4780
    },
    {
      "epoch": 0.5052742616033755,
      "grad_norm": 0.470165878534317,
      "learning_rate": 0.0007493723510348516,
      "loss": 1.504,
      "step": 4790
    },
    {
      "epoch": 0.5063291139240507,
      "grad_norm": 0.7496804594993591,
      "learning_rate": 0.0007468617639656496,
      "loss": 1.5051,
      "step": 4800
    },
    {
      "epoch": 0.5073839662447257,
      "grad_norm": 0.5750884413719177,
      "learning_rate": 0.0007443512120619144,
      "loss": 1.5032,
      "step": 4810
    },
    {
      "epoch": 0.5084388185654009,
      "grad_norm": 0.5860179662704468,
      "learning_rate": 0.0007418407234556067,
      "loss": 1.5243,
      "step": 4820
    },
    {
      "epoch": 0.509493670886076,
      "grad_norm": 0.581038773059845,
      "learning_rate": 0.0007393303262779767,
      "loss": 1.5039,
      "step": 4830
    },
    {
      "epoch": 0.510548523206751,
      "grad_norm": 0.47394222021102905,
      "learning_rate": 0.0007368200486592507,
      "loss": 1.4983,
      "step": 4840
    },
    {
      "epoch": 0.5116033755274262,
      "grad_norm": 0.5796689391136169,
      "learning_rate": 0.0007343099187283149,
      "loss": 1.515,
      "step": 4850
    },
    {
      "epoch": 0.5126582278481012,
      "grad_norm": 0.6724705696105957,
      "learning_rate": 0.0007317999646124011,
      "loss": 1.5008,
      "step": 4860
    },
    {
      "epoch": 0.5137130801687764,
      "grad_norm": 0.6664035320281982,
      "learning_rate": 0.0007292902144367704,
      "loss": 1.493,
      "step": 4870
    },
    {
      "epoch": 0.5147679324894515,
      "grad_norm": 0.9945948719978333,
      "learning_rate": 0.0007267806963243995,
      "loss": 1.5157,
      "step": 4880
    },
    {
      "epoch": 0.5158227848101266,
      "grad_norm": 0.8466383218765259,
      "learning_rate": 0.0007242714383956639,
      "loss": 1.5047,
      "step": 4890
    },
    {
      "epoch": 0.5168776371308017,
      "grad_norm": 0.48031893372535706,
      "learning_rate": 0.000721762468768024,
      "loss": 1.5102,
      "step": 4900
    },
    {
      "epoch": 0.5179324894514767,
      "grad_norm": 0.5278339385986328,
      "learning_rate": 0.0007192538155557094,
      "loss": 1.5052,
      "step": 4910
    },
    {
      "epoch": 0.5189873417721519,
      "grad_norm": 0.7497695088386536,
      "learning_rate": 0.0007167455068694046,
      "loss": 1.4956,
      "step": 4920
    },
    {
      "epoch": 0.520042194092827,
      "grad_norm": 0.7196361422538757,
      "learning_rate": 0.000714237570815933,
      "loss": 1.5035,
      "step": 4930
    },
    {
      "epoch": 0.5210970464135021,
      "grad_norm": 0.49846696853637695,
      "learning_rate": 0.0007117300354979423,
      "loss": 1.5002,
      "step": 4940
    },
    {
      "epoch": 0.5221518987341772,
      "grad_norm": 0.5527176856994629,
      "learning_rate": 0.000709222929013591,
      "loss": 1.5091,
      "step": 4950
    },
    {
      "epoch": 0.5232067510548524,
      "grad_norm": 0.5904789566993713,
      "learning_rate": 0.0007067162794562309,
      "loss": 1.4891,
      "step": 4960
    },
    {
      "epoch": 0.5242616033755274,
      "grad_norm": 0.7089250683784485,
      "learning_rate": 0.0007042101149140943,
      "loss": 1.4979,
      "step": 4970
    },
    {
      "epoch": 0.5253164556962026,
      "grad_norm": 0.5322713851928711,
      "learning_rate": 0.0007017044634699787,
      "loss": 1.4927,
      "step": 4980
    },
    {
      "epoch": 0.5263713080168776,
      "grad_norm": 0.6182569861412048,
      "learning_rate": 0.0006991993532009319,
      "loss": 1.4916,
      "step": 4990
    },
    {
      "epoch": 0.5274261603375527,
      "grad_norm": 0.5848808884620667,
      "learning_rate": 0.0006966948121779378,
      "loss": 1.5061,
      "step": 5000
    },
    {
      "epoch": 0.5284810126582279,
      "grad_norm": 0.4910695552825928,
      "learning_rate": 0.000694190868465601,
      "loss": 1.4983,
      "step": 5010
    },
    {
      "epoch": 0.5295358649789029,
      "grad_norm": 0.5071682929992676,
      "learning_rate": 0.0006916875501218343,
      "loss": 1.4931,
      "step": 5020
    },
    {
      "epoch": 0.5305907172995781,
      "grad_norm": 0.4766605496406555,
      "learning_rate": 0.0006891848851975416,
      "loss": 1.4752,
      "step": 5030
    },
    {
      "epoch": 0.5316455696202531,
      "grad_norm": 0.45955702662467957,
      "learning_rate": 0.0006866829017363054,
      "loss": 1.5053,
      "step": 5040
    },
    {
      "epoch": 0.5327004219409283,
      "grad_norm": 0.501193642616272,
      "learning_rate": 0.0006841816277740722,
      "loss": 1.4923,
      "step": 5050
    },
    {
      "epoch": 0.5337552742616034,
      "grad_norm": 0.6511161923408508,
      "learning_rate": 0.0006816810913388379,
      "loss": 1.5066,
      "step": 5060
    },
    {
      "epoch": 0.5348101265822784,
      "grad_norm": 0.5442824363708496,
      "learning_rate": 0.0006791813204503342,
      "loss": 1.4916,
      "step": 5070
    },
    {
      "epoch": 0.5358649789029536,
      "grad_norm": 0.5160462856292725,
      "learning_rate": 0.0006766823431197147,
      "loss": 1.5036,
      "step": 5080
    },
    {
      "epoch": 0.5369198312236287,
      "grad_norm": 0.5459616780281067,
      "learning_rate": 0.0006741841873492406,
      "loss": 1.4942,
      "step": 5090
    },
    {
      "epoch": 0.5379746835443038,
      "grad_norm": 1.007577896118164,
      "learning_rate": 0.0006716868811319671,
      "loss": 1.4822,
      "step": 5100
    },
    {
      "epoch": 0.5390295358649789,
      "grad_norm": 0.9401910901069641,
      "learning_rate": 0.0006691904524514297,
      "loss": 1.4961,
      "step": 5110
    },
    {
      "epoch": 0.540084388185654,
      "grad_norm": 0.8771179914474487,
      "learning_rate": 0.0006666949292813306,
      "loss": 1.4882,
      "step": 5120
    },
    {
      "epoch": 0.5411392405063291,
      "grad_norm": 0.70003342628479,
      "learning_rate": 0.0006642003395852258,
      "loss": 1.4858,
      "step": 5130
    },
    {
      "epoch": 0.5421940928270043,
      "grad_norm": 0.5451027154922485,
      "learning_rate": 0.0006617067113162103,
      "loss": 1.5005,
      "step": 5140
    },
    {
      "epoch": 0.5432489451476793,
      "grad_norm": 0.4791337251663208,
      "learning_rate": 0.0006592140724166073,
      "loss": 1.4988,
      "step": 5150
    },
    {
      "epoch": 0.5443037974683544,
      "grad_norm": 0.5083988308906555,
      "learning_rate": 0.0006567224508176523,
      "loss": 1.4902,
      "step": 5160
    },
    {
      "epoch": 0.5453586497890295,
      "grad_norm": 0.5682942271232605,
      "learning_rate": 0.0006542318744391821,
      "loss": 1.4896,
      "step": 5170
    },
    {
      "epoch": 0.5464135021097046,
      "grad_norm": 0.6583775877952576,
      "learning_rate": 0.0006517423711893209,
      "loss": 1.5011,
      "step": 5180
    },
    {
      "epoch": 0.5474683544303798,
      "grad_norm": 0.5080510377883911,
      "learning_rate": 0.0006492539689641685,
      "loss": 1.4832,
      "step": 5190
    },
    {
      "epoch": 0.5485232067510548,
      "grad_norm": 0.8768990635871887,
      "learning_rate": 0.0006467666956474865,
      "loss": 1.4948,
      "step": 5200
    },
    {
      "epoch": 0.54957805907173,
      "grad_norm": 0.7748043537139893,
      "learning_rate": 0.0006442805791103873,
      "loss": 1.4788,
      "step": 5210
    },
    {
      "epoch": 0.5506329113924051,
      "grad_norm": 0.45537272095680237,
      "learning_rate": 0.0006417956472110205,
      "loss": 1.4894,
      "step": 5220
    },
    {
      "epoch": 0.5516877637130801,
      "grad_norm": 0.7094871401786804,
      "learning_rate": 0.0006393119277942614,
      "loss": 1.4837,
      "step": 5230
    },
    {
      "epoch": 0.5527426160337553,
      "grad_norm": 0.5354472398757935,
      "learning_rate": 0.0006368294486913987,
      "loss": 1.491,
      "step": 5240
    },
    {
      "epoch": 0.5537974683544303,
      "grad_norm": 0.4849853217601776,
      "learning_rate": 0.0006343482377198232,
      "loss": 1.4821,
      "step": 5250
    },
    {
      "epoch": 0.5548523206751055,
      "grad_norm": 0.4794313609600067,
      "learning_rate": 0.0006318683226827151,
      "loss": 1.4897,
      "step": 5260
    },
    {
      "epoch": 0.5559071729957806,
      "grad_norm": 0.6370975375175476,
      "learning_rate": 0.0006293897313687331,
      "loss": 1.4967,
      "step": 5270
    },
    {
      "epoch": 0.5569620253164557,
      "grad_norm": 0.6387209892272949,
      "learning_rate": 0.0006269124915517037,
      "loss": 1.486,
      "step": 5280
    },
    {
      "epoch": 0.5580168776371308,
      "grad_norm": 0.5333986282348633,
      "learning_rate": 0.0006244366309903084,
      "loss": 1.4849,
      "step": 5290
    },
    {
      "epoch": 0.5590717299578059,
      "grad_norm": 0.5619298815727234,
      "learning_rate": 0.0006219621774277737,
      "loss": 1.4897,
      "step": 5300
    },
    {
      "epoch": 0.560126582278481,
      "grad_norm": 0.5637062191963196,
      "learning_rate": 0.00061948915859156,
      "loss": 1.4763,
      "step": 5310
    },
    {
      "epoch": 0.5611814345991561,
      "grad_norm": 0.5141758918762207,
      "learning_rate": 0.0006170176021930509,
      "loss": 1.4759,
      "step": 5320
    },
    {
      "epoch": 0.5622362869198312,
      "grad_norm": 0.490723192691803,
      "learning_rate": 0.0006145475359272424,
      "loss": 1.484,
      "step": 5330
    },
    {
      "epoch": 0.5632911392405063,
      "grad_norm": 0.5165253281593323,
      "learning_rate": 0.0006120789874724336,
      "loss": 1.48,
      "step": 5340
    },
    {
      "epoch": 0.5643459915611815,
      "grad_norm": 0.5606610178947449,
      "learning_rate": 0.0006096119844899151,
      "loss": 1.4716,
      "step": 5350
    },
    {
      "epoch": 0.5654008438818565,
      "grad_norm": 0.4859774708747864,
      "learning_rate": 0.0006071465546236601,
      "loss": 1.481,
      "step": 5360
    },
    {
      "epoch": 0.5664556962025317,
      "grad_norm": 0.6043248772621155,
      "learning_rate": 0.0006046827255000135,
      "loss": 1.4837,
      "step": 5370
    },
    {
      "epoch": 0.5675105485232067,
      "grad_norm": 0.5425053238868713,
      "learning_rate": 0.0006022205247273845,
      "loss": 1.4712,
      "step": 5380
    },
    {
      "epoch": 0.5685654008438819,
      "grad_norm": 0.5926060080528259,
      "learning_rate": 0.0005997599798959343,
      "loss": 1.4746,
      "step": 5390
    },
    {
      "epoch": 0.569620253164557,
      "grad_norm": 0.5091972947120667,
      "learning_rate": 0.0005973011185772694,
      "loss": 1.4755,
      "step": 5400
    },
    {
      "epoch": 0.570675105485232,
      "grad_norm": 0.4799787104129791,
      "learning_rate": 0.0005948439683241318,
      "loss": 1.4782,
      "step": 5410
    },
    {
      "epoch": 0.5717299578059072,
      "grad_norm": 0.46803295612335205,
      "learning_rate": 0.0005923885566700896,
      "loss": 1.473,
      "step": 5420
    },
    {
      "epoch": 0.5727848101265823,
      "grad_norm": 0.5348048210144043,
      "learning_rate": 0.0005899349111292293,
      "loss": 1.4695,
      "step": 5430
    },
    {
      "epoch": 0.5738396624472574,
      "grad_norm": 0.5067234635353088,
      "learning_rate": 0.0005874830591958474,
      "loss": 1.469,
      "step": 5440
    },
    {
      "epoch": 0.5748945147679325,
      "grad_norm": 0.5505863428115845,
      "learning_rate": 0.000585033028344142,
      "loss": 1.4743,
      "step": 5450
    },
    {
      "epoch": 0.5759493670886076,
      "grad_norm": 0.49517422914505005,
      "learning_rate": 0.0005825848460279048,
      "loss": 1.4754,
      "step": 5460
    },
    {
      "epoch": 0.5770042194092827,
      "grad_norm": 0.5773853659629822,
      "learning_rate": 0.0005801385396802146,
      "loss": 1.477,
      "step": 5470
    },
    {
      "epoch": 0.5780590717299579,
      "grad_norm": 0.7006990313529968,
      "learning_rate": 0.0005776941367131282,
      "loss": 1.4793,
      "step": 5480
    },
    {
      "epoch": 0.5791139240506329,
      "grad_norm": 0.7622592449188232,
      "learning_rate": 0.0005752516645173745,
      "loss": 1.482,
      "step": 5490
    },
    {
      "epoch": 0.580168776371308,
      "grad_norm": 0.528407871723175,
      "learning_rate": 0.0005728111504620472,
      "loss": 1.4717,
      "step": 5500
    },
    {
      "epoch": 0.5812236286919831,
      "grad_norm": 0.48450398445129395,
      "learning_rate": 0.0005703726218942976,
      "loss": 1.4793,
      "step": 5510
    },
    {
      "epoch": 0.5822784810126582,
      "grad_norm": 0.6169544458389282,
      "learning_rate": 0.0005679361061390295,
      "loss": 1.4735,
      "step": 5520
    },
    {
      "epoch": 0.5833333333333334,
      "grad_norm": 0.5328255891799927,
      "learning_rate": 0.0005655016304985908,
      "loss": 1.4632,
      "step": 5530
    },
    {
      "epoch": 0.5843881856540084,
      "grad_norm": 0.5129302740097046,
      "learning_rate": 0.0005630692222524709,
      "loss": 1.464,
      "step": 5540
    },
    {
      "epoch": 0.5854430379746836,
      "grad_norm": 0.5741155743598938,
      "learning_rate": 0.0005606389086569911,
      "loss": 1.4801,
      "step": 5550
    },
    {
      "epoch": 0.5864978902953587,
      "grad_norm": 0.48890742659568787,
      "learning_rate": 0.0005582107169450023,
      "loss": 1.4838,
      "step": 5560
    },
    {
      "epoch": 0.5875527426160337,
      "grad_norm": 0.5709759593009949,
      "learning_rate": 0.0005557846743255783,
      "loss": 1.4668,
      "step": 5570
    },
    {
      "epoch": 0.5886075949367089,
      "grad_norm": 0.5346205830574036,
      "learning_rate": 0.0005533608079837109,
      "loss": 1.4657,
      "step": 5580
    },
    {
      "epoch": 0.5896624472573839,
      "grad_norm": 0.6232300996780396,
      "learning_rate": 0.0005509391450800061,
      "loss": 1.4701,
      "step": 5590
    },
    {
      "epoch": 0.5907172995780591,
      "grad_norm": 0.5745095014572144,
      "learning_rate": 0.0005485197127503795,
      "loss": 1.4647,
      "step": 5600
    },
    {
      "epoch": 0.5917721518987342,
      "grad_norm": 0.58150714635849,
      "learning_rate": 0.0005461025381057516,
      "loss": 1.4695,
      "step": 5610
    },
    {
      "epoch": 0.5928270042194093,
      "grad_norm": 0.9152214527130127,
      "learning_rate": 0.0005436876482317444,
      "loss": 1.4783,
      "step": 5620
    },
    {
      "epoch": 0.5938818565400844,
      "grad_norm": 0.49668022990226746,
      "learning_rate": 0.0005412750701883782,
      "loss": 1.4727,
      "step": 5630
    },
    {
      "epoch": 0.5949367088607594,
      "grad_norm": 0.5094896554946899,
      "learning_rate": 0.0005388648310097682,
      "loss": 1.4798,
      "step": 5640
    },
    {
      "epoch": 0.5959915611814346,
      "grad_norm": 0.6089109182357788,
      "learning_rate": 0.000536456957703821,
      "loss": 1.4738,
      "step": 5650
    },
    {
      "epoch": 0.5970464135021097,
      "grad_norm": 0.6121323108673096,
      "learning_rate": 0.0005340514772519324,
      "loss": 1.4677,
      "step": 5660
    },
    {
      "epoch": 0.5981012658227848,
      "grad_norm": 0.51023930311203,
      "learning_rate": 0.0005316484166086863,
      "loss": 1.4761,
      "step": 5670
    },
    {
      "epoch": 0.5991561181434599,
      "grad_norm": 0.4947879910469055,
      "learning_rate": 0.00052924780270155,
      "loss": 1.478,
      "step": 5680
    },
    {
      "epoch": 0.6002109704641351,
      "grad_norm": 0.6367293000221252,
      "learning_rate": 0.0005268496624305747,
      "loss": 1.4539,
      "step": 5690
    },
    {
      "epoch": 0.6012658227848101,
      "grad_norm": 0.512224018573761,
      "learning_rate": 0.0005244540226680931,
      "loss": 1.4594,
      "step": 5700
    },
    {
      "epoch": 0.6023206751054853,
      "grad_norm": 0.5635572075843811,
      "learning_rate": 0.0005220609102584185,
      "loss": 1.4645,
      "step": 5710
    },
    {
      "epoch": 0.6033755274261603,
      "grad_norm": 0.5069919228553772,
      "learning_rate": 0.0005196703520175437,
      "loss": 1.4672,
      "step": 5720
    },
    {
      "epoch": 0.6044303797468354,
      "grad_norm": 0.5056061744689941,
      "learning_rate": 0.0005172823747328415,
      "loss": 1.4643,
      "step": 5730
    },
    {
      "epoch": 0.6054852320675106,
      "grad_norm": 0.5348817110061646,
      "learning_rate": 0.0005148970051627632,
      "loss": 1.462,
      "step": 5740
    },
    {
      "epoch": 0.6065400843881856,
      "grad_norm": 0.5328285098075867,
      "learning_rate": 0.0005125142700365394,
      "loss": 1.437,
      "step": 5750
    },
    {
      "epoch": 0.6075949367088608,
      "grad_norm": 0.48856961727142334,
      "learning_rate": 0.000510134196053881,
      "loss": 1.4457,
      "step": 5760
    },
    {
      "epoch": 0.6086497890295358,
      "grad_norm": 0.6688364148139954,
      "learning_rate": 0.0005077568098846789,
      "loss": 1.4514,
      "step": 5770
    },
    {
      "epoch": 0.609704641350211,
      "grad_norm": 0.47691768407821655,
      "learning_rate": 0.000505382138168706,
      "loss": 1.4585,
      "step": 5780
    },
    {
      "epoch": 0.6107594936708861,
      "grad_norm": 0.4656788408756256,
      "learning_rate": 0.0005030102075153181,
      "loss": 1.4551,
      "step": 5790
    },
    {
      "epoch": 0.6118143459915611,
      "grad_norm": 0.5498069524765015,
      "learning_rate": 0.0005006410445031569,
      "loss": 1.4583,
      "step": 5800
    },
    {
      "epoch": 0.6128691983122363,
      "grad_norm": 0.632783830165863,
      "learning_rate": 0.0004982746756798507,
      "loss": 1.4639,
      "step": 5810
    },
    {
      "epoch": 0.6139240506329114,
      "grad_norm": 0.5301288962364197,
      "learning_rate": 0.0004959111275617174,
      "loss": 1.4778,
      "step": 5820
    },
    {
      "epoch": 0.6149789029535865,
      "grad_norm": 0.4994509220123291,
      "learning_rate": 0.0004935504266334677,
      "loss": 1.4593,
      "step": 5830
    },
    {
      "epoch": 0.6160337552742616,
      "grad_norm": 0.5345269441604614,
      "learning_rate": 0.0004911925993479085,
      "loss": 1.4705,
      "step": 5840
    },
    {
      "epoch": 0.6170886075949367,
      "grad_norm": 0.5767863988876343,
      "learning_rate": 0.0004888376721256456,
      "loss": 1.4752,
      "step": 5850
    },
    {
      "epoch": 0.6181434599156118,
      "grad_norm": 0.7049380540847778,
      "learning_rate": 0.00048648567135478805,
      "loss": 1.4632,
      "step": 5860
    },
    {
      "epoch": 0.619198312236287,
      "grad_norm": 0.49564504623413086,
      "learning_rate": 0.0004841366233906538,
      "loss": 1.4551,
      "step": 5870
    },
    {
      "epoch": 0.620253164556962,
      "grad_norm": 0.4838204085826874,
      "learning_rate": 0.0004817905545554717,
      "loss": 1.4668,
      "step": 5880
    },
    {
      "epoch": 0.6213080168776371,
      "grad_norm": 0.7612228393554688,
      "learning_rate": 0.00047944749113808884,
      "loss": 1.4552,
      "step": 5890
    },
    {
      "epoch": 0.6223628691983122,
      "grad_norm": 0.4583361744880676,
      "learning_rate": 0.00047710745939367474,
      "loss": 1.4561,
      "step": 5900
    },
    {
      "epoch": 0.6234177215189873,
      "grad_norm": 0.5316163301467896,
      "learning_rate": 0.0004747704855434278,
      "loss": 1.4491,
      "step": 5910
    },
    {
      "epoch": 0.6244725738396625,
      "grad_norm": 0.4847448468208313,
      "learning_rate": 0.0004724365957742809,
      "loss": 1.4476,
      "step": 5920
    },
    {
      "epoch": 0.6255274261603375,
      "grad_norm": 1.0385831594467163,
      "learning_rate": 0.00047010581623860883,
      "loss": 1.4501,
      "step": 5930
    },
    {
      "epoch": 0.6265822784810127,
      "grad_norm": 0.829448401927948,
      "learning_rate": 0.0004677781730539342,
      "loss": 1.4587,
      "step": 5940
    },
    {
      "epoch": 0.6276371308016878,
      "grad_norm": 0.5561820268630981,
      "learning_rate": 0.0004654536923026356,
      "loss": 1.4479,
      "step": 5950
    },
    {
      "epoch": 0.6286919831223629,
      "grad_norm": 0.5981544852256775,
      "learning_rate": 0.00046313240003165466,
      "loss": 1.4546,
      "step": 5960
    },
    {
      "epoch": 0.629746835443038,
      "grad_norm": 0.4896555244922638,
      "learning_rate": 0.0004608143222522048,
      "loss": 1.4547,
      "step": 5970
    },
    {
      "epoch": 0.630801687763713,
      "grad_norm": 0.5490660071372986,
      "learning_rate": 0.0004584994849394795,
      "loss": 1.4384,
      "step": 5980
    },
    {
      "epoch": 0.6318565400843882,
      "grad_norm": 0.634878933429718,
      "learning_rate": 0.0004561879140323607,
      "loss": 1.4575,
      "step": 5990
    },
    {
      "epoch": 0.6329113924050633,
      "grad_norm": 0.5208130478858948,
      "learning_rate": 0.0004538796354331298,
      "loss": 1.4629,
      "step": 6000
    },
    {
      "epoch": 0.6339662447257384,
      "grad_norm": 0.48796504735946655,
      "learning_rate": 0.0004515746750071754,
      "loss": 1.4502,
      "step": 6010
    },
    {
      "epoch": 0.6350210970464135,
      "grad_norm": 0.7172197699546814,
      "learning_rate": 0.0004492730585827046,
      "loss": 1.4566,
      "step": 6020
    },
    {
      "epoch": 0.6360759493670886,
      "grad_norm": 0.5025409460067749,
      "learning_rate": 0.0004469748119504529,
      "loss": 1.4372,
      "step": 6030
    },
    {
      "epoch": 0.6371308016877637,
      "grad_norm": 0.545889675617218,
      "learning_rate": 0.0004446799608633964,
      "loss": 1.4583,
      "step": 6040
    },
    {
      "epoch": 0.6381856540084389,
      "grad_norm": 0.4706031382083893,
      "learning_rate": 0.00044238853103646154,
      "loss": 1.4407,
      "step": 6050
    },
    {
      "epoch": 0.6392405063291139,
      "grad_norm": 0.5463115572929382,
      "learning_rate": 0.00044010054814623925,
      "loss": 1.4597,
      "step": 6060
    },
    {
      "epoch": 0.640295358649789,
      "grad_norm": 0.515608012676239,
      "learning_rate": 0.0004378160378306944,
      "loss": 1.4488,
      "step": 6070
    },
    {
      "epoch": 0.6413502109704642,
      "grad_norm": 0.49624624848365784,
      "learning_rate": 0.00043553502568888095,
      "loss": 1.4455,
      "step": 6080
    },
    {
      "epoch": 0.6424050632911392,
      "grad_norm": 0.5415939688682556,
      "learning_rate": 0.0004332575372806534,
      "loss": 1.4438,
      "step": 6090
    },
    {
      "epoch": 0.6434599156118144,
      "grad_norm": 0.49219149351119995,
      "learning_rate": 0.00043098359812638145,
      "loss": 1.4373,
      "step": 6100
    },
    {
      "epoch": 0.6445147679324894,
      "grad_norm": 0.4915757477283478,
      "learning_rate": 0.00042871323370666383,
      "loss": 1.4512,
      "step": 6110
    },
    {
      "epoch": 0.6455696202531646,
      "grad_norm": 0.5928741097450256,
      "learning_rate": 0.0004264464694620421,
      "loss": 1.4478,
      "step": 6120
    },
    {
      "epoch": 0.6466244725738397,
      "grad_norm": 0.48575419187545776,
      "learning_rate": 0.000424183330792717,
      "loss": 1.4447,
      "step": 6130
    },
    {
      "epoch": 0.6476793248945147,
      "grad_norm": 0.507031261920929,
      "learning_rate": 0.0004219238430582621,
      "loss": 1.4469,
      "step": 6140
    },
    {
      "epoch": 0.6487341772151899,
      "grad_norm": 0.4940553903579712,
      "learning_rate": 0.0004196680315773408,
      "loss": 1.4555,
      "step": 6150
    },
    {
      "epoch": 0.6497890295358649,
      "grad_norm": 0.4866984486579895,
      "learning_rate": 0.00041741592162742214,
      "loss": 1.4506,
      "step": 6160
    },
    {
      "epoch": 0.6508438818565401,
      "grad_norm": 0.5252698659896851,
      "learning_rate": 0.0004151675384444978,
      "loss": 1.4264,
      "step": 6170
    },
    {
      "epoch": 0.6518987341772152,
      "grad_norm": 0.5911986231803894,
      "learning_rate": 0.00041292290722279914,
      "loss": 1.4483,
      "step": 6180
    },
    {
      "epoch": 0.6529535864978903,
      "grad_norm": 0.7079945206642151,
      "learning_rate": 0.00041068205311451517,
      "loss": 1.4541,
      "step": 6190
    },
    {
      "epoch": 0.6540084388185654,
      "grad_norm": 0.5665788054466248,
      "learning_rate": 0.00040844500122951026,
      "loss": 1.4467,
      "step": 6200
    },
    {
      "epoch": 0.6550632911392406,
      "grad_norm": 1.0949095487594604,
      "learning_rate": 0.00040621177663504313,
      "loss": 1.4421,
      "step": 6210
    },
    {
      "epoch": 0.6561181434599156,
      "grad_norm": 0.7985562086105347,
      "learning_rate": 0.00040398240435548583,
      "loss": 1.4485,
      "step": 6220
    },
    {
      "epoch": 0.6571729957805907,
      "grad_norm": 0.4975112974643707,
      "learning_rate": 0.00040175690937204324,
      "loss": 1.4378,
      "step": 6230
    },
    {
      "epoch": 0.6582278481012658,
      "grad_norm": 0.48909488320350647,
      "learning_rate": 0.00039953531662247343,
      "loss": 1.4427,
      "step": 6240
    },
    {
      "epoch": 0.6592827004219409,
      "grad_norm": 0.7844357490539551,
      "learning_rate": 0.0003973176510008075,
      "loss": 1.4488,
      "step": 6250
    },
    {
      "epoch": 0.6603375527426161,
      "grad_norm": 0.5498688220977783,
      "learning_rate": 0.00039510393735707233,
      "loss": 1.4412,
      "step": 6260
    },
    {
      "epoch": 0.6613924050632911,
      "grad_norm": 0.507627010345459,
      "learning_rate": 0.00039289420049700986,
      "loss": 1.4504,
      "step": 6270
    },
    {
      "epoch": 0.6624472573839663,
      "grad_norm": 0.5005232095718384,
      "learning_rate": 0.0003906884651818006,
      "loss": 1.4465,
      "step": 6280
    },
    {
      "epoch": 0.6635021097046413,
      "grad_norm": 0.5058642029762268,
      "learning_rate": 0.00038848675612778577,
      "loss": 1.4441,
      "step": 6290
    },
    {
      "epoch": 0.6645569620253164,
      "grad_norm": 0.6778172850608826,
      "learning_rate": 0.00038628909800619046,
      "loss": 1.4385,
      "step": 6300
    },
    {
      "epoch": 0.6656118143459916,
      "grad_norm": 0.49329179525375366,
      "learning_rate": 0.0003840955154428467,
      "loss": 1.4373,
      "step": 6310
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.8938937783241272,
      "learning_rate": 0.00038190603301791864,
      "loss": 1.4336,
      "step": 6320
    },
    {
      "epoch": 0.6677215189873418,
      "grad_norm": 0.5498088002204895,
      "learning_rate": 0.0003797206752656258,
      "loss": 1.4439,
      "step": 6330
    },
    {
      "epoch": 0.6687763713080169,
      "grad_norm": 0.5045040845870972,
      "learning_rate": 0.0003775394666739688,
      "loss": 1.4363,
      "step": 6340
    },
    {
      "epoch": 0.669831223628692,
      "grad_norm": 0.6053415536880493,
      "learning_rate": 0.00037536243168445507,
      "loss": 1.4424,
      "step": 6350
    },
    {
      "epoch": 0.6708860759493671,
      "grad_norm": 0.49733415246009827,
      "learning_rate": 0.0003731895946918246,
      "loss": 1.4403,
      "step": 6360
    },
    {
      "epoch": 0.6719409282700421,
      "grad_norm": 0.5580061078071594,
      "learning_rate": 0.0003710209800437769,
      "loss": 1.4208,
      "step": 6370
    },
    {
      "epoch": 0.6729957805907173,
      "grad_norm": 0.5783162713050842,
      "learning_rate": 0.00036885661204069767,
      "loss": 1.4273,
      "step": 6380
    },
    {
      "epoch": 0.6740506329113924,
      "grad_norm": 0.5714283585548401,
      "learning_rate": 0.0003666965149353878,
      "loss": 1.4445,
      "step": 6390
    },
    {
      "epoch": 0.6751054852320675,
      "grad_norm": 0.5972985625267029,
      "learning_rate": 0.0003645407129327898,
      "loss": 1.4412,
      "step": 6400
    },
    {
      "epoch": 0.6761603375527426,
      "grad_norm": 0.4876542389392853,
      "learning_rate": 0.00036238923018971783,
      "loss": 1.4358,
      "step": 6410
    },
    {
      "epoch": 0.6772151898734177,
      "grad_norm": 0.5544925928115845,
      "learning_rate": 0.0003602420908145865,
      "loss": 1.428,
      "step": 6420
    },
    {
      "epoch": 0.6782700421940928,
      "grad_norm": 0.6289705038070679,
      "learning_rate": 0.00035809931886714093,
      "loss": 1.437,
      "step": 6430
    },
    {
      "epoch": 0.679324894514768,
      "grad_norm": 0.5447190999984741,
      "learning_rate": 0.00035596093835818683,
      "loss": 1.4212,
      "step": 6440
    },
    {
      "epoch": 0.680379746835443,
      "grad_norm": 0.6949787139892578,
      "learning_rate": 0.00035382697324932245,
      "loss": 1.4266,
      "step": 6450
    },
    {
      "epoch": 0.6814345991561181,
      "grad_norm": 0.8234121203422546,
      "learning_rate": 0.00035169744745266866,
      "loss": 1.44,
      "step": 6460
    },
    {
      "epoch": 0.6824894514767933,
      "grad_norm": 0.6068360209465027,
      "learning_rate": 0.0003495723848306017,
      "loss": 1.4319,
      "step": 6470
    },
    {
      "epoch": 0.6835443037974683,
      "grad_norm": 0.6669226884841919,
      "learning_rate": 0.0003474518091954859,
      "loss": 1.4407,
      "step": 6480
    },
    {
      "epoch": 0.6845991561181435,
      "grad_norm": 0.5997419953346252,
      "learning_rate": 0.0003453357443094068,
      "loss": 1.4333,
      "step": 6490
    },
    {
      "epoch": 0.6856540084388185,
      "grad_norm": 0.5909753441810608,
      "learning_rate": 0.00034322421388390456,
      "loss": 1.4512,
      "step": 6500
    },
    {
      "epoch": 0.6867088607594937,
      "grad_norm": 0.694766104221344,
      "learning_rate": 0.0003411172415797087,
      "loss": 1.4497,
      "step": 6510
    },
    {
      "epoch": 0.6877637130801688,
      "grad_norm": 0.5000135898590088,
      "learning_rate": 0.0003390148510064727,
      "loss": 1.4329,
      "step": 6520
    },
    {
      "epoch": 0.6888185654008439,
      "grad_norm": 0.5609813928604126,
      "learning_rate": 0.0003369170657225094,
      "loss": 1.4239,
      "step": 6530
    },
    {
      "epoch": 0.689873417721519,
      "grad_norm": 0.5548239350318909,
      "learning_rate": 0.0003348239092345275,
      "loss": 1.435,
      "step": 6540
    },
    {
      "epoch": 0.6909282700421941,
      "grad_norm": 0.6655601263046265,
      "learning_rate": 0.0003327354049973672,
      "loss": 1.4325,
      "step": 6550
    },
    {
      "epoch": 0.6919831223628692,
      "grad_norm": 0.9289987683296204,
      "learning_rate": 0.00033065157641373847,
      "loss": 1.4305,
      "step": 6560
    },
    {
      "epoch": 0.6930379746835443,
      "grad_norm": 0.7574074864387512,
      "learning_rate": 0.0003285724468339576,
      "loss": 1.4293,
      "step": 6570
    },
    {
      "epoch": 0.6940928270042194,
      "grad_norm": 0.6782986521720886,
      "learning_rate": 0.00032649803955568755,
      "loss": 1.4279,
      "step": 6580
    },
    {
      "epoch": 0.6951476793248945,
      "grad_norm": 0.7088582515716553,
      "learning_rate": 0.00032442837782367434,
      "loss": 1.431,
      "step": 6590
    },
    {
      "epoch": 0.6962025316455697,
      "grad_norm": 0.5320621132850647,
      "learning_rate": 0.0003223634848294883,
      "loss": 1.4373,
      "step": 6600
    },
    {
      "epoch": 0.6972573839662447,
      "grad_norm": 0.5401026010513306,
      "learning_rate": 0.00032030338371126374,
      "loss": 1.4244,
      "step": 6610
    },
    {
      "epoch": 0.6983122362869199,
      "grad_norm": 0.5319799184799194,
      "learning_rate": 0.0003182480975534395,
      "loss": 1.4287,
      "step": 6620
    },
    {
      "epoch": 0.6993670886075949,
      "grad_norm": 0.5029904842376709,
      "learning_rate": 0.00031619764938650057,
      "loss": 1.4247,
      "step": 6630
    },
    {
      "epoch": 0.70042194092827,
      "grad_norm": 0.5175420045852661,
      "learning_rate": 0.0003141520621867197,
      "loss": 1.4276,
      "step": 6640
    },
    {
      "epoch": 0.7014767932489452,
      "grad_norm": 0.611236572265625,
      "learning_rate": 0.00031211135887590074,
      "loss": 1.4243,
      "step": 6650
    },
    {
      "epoch": 0.7025316455696202,
      "grad_norm": 0.5167410969734192,
      "learning_rate": 0.0003100755623211205,
      "loss": 1.4261,
      "step": 6660
    },
    {
      "epoch": 0.7035864978902954,
      "grad_norm": 0.47252219915390015,
      "learning_rate": 0.0003080446953344735,
      "loss": 1.4251,
      "step": 6670
    },
    {
      "epoch": 0.7046413502109705,
      "grad_norm": 0.5216288566589355,
      "learning_rate": 0.00030601878067281575,
      "loss": 1.4296,
      "step": 6680
    },
    {
      "epoch": 0.7056962025316456,
      "grad_norm": 0.554894745349884,
      "learning_rate": 0.00030399784103751044,
      "loss": 1.4218,
      "step": 6690
    },
    {
      "epoch": 0.7067510548523207,
      "grad_norm": 0.5862643122673035,
      "learning_rate": 0.000301981899074173,
      "loss": 1.4237,
      "step": 6700
    },
    {
      "epoch": 0.7078059071729957,
      "grad_norm": 0.5114737153053284,
      "learning_rate": 0.0002999709773724171,
      "loss": 1.4253,
      "step": 6710
    },
    {
      "epoch": 0.7088607594936709,
      "grad_norm": 0.8156765103340149,
      "learning_rate": 0.00029796509846560294,
      "loss": 1.4159,
      "step": 6720
    },
    {
      "epoch": 0.709915611814346,
      "grad_norm": 0.4957371950149536,
      "learning_rate": 0.0002959642848305828,
      "loss": 1.424,
      "step": 6730
    },
    {
      "epoch": 0.7109704641350211,
      "grad_norm": 0.6011149883270264,
      "learning_rate": 0.00029396855888745045,
      "loss": 1.4256,
      "step": 6740
    },
    {
      "epoch": 0.7120253164556962,
      "grad_norm": 0.49337509274482727,
      "learning_rate": 0.0002919779429992895,
      "loss": 1.4291,
      "step": 6750
    },
    {
      "epoch": 0.7130801687763713,
      "grad_norm": 0.5435550212860107,
      "learning_rate": 0.0002899924594719231,
      "loss": 1.4248,
      "step": 6760
    },
    {
      "epoch": 0.7141350210970464,
      "grad_norm": 0.5038012862205505,
      "learning_rate": 0.00028801213055366335,
      "loss": 1.4277,
      "step": 6770
    },
    {
      "epoch": 0.7151898734177216,
      "grad_norm": 0.5369750261306763,
      "learning_rate": 0.00028603697843506315,
      "loss": 1.4218,
      "step": 6780
    },
    {
      "epoch": 0.7162447257383966,
      "grad_norm": 0.5309962034225464,
      "learning_rate": 0.0002840670252486662,
      "loss": 1.4316,
      "step": 6790
    },
    {
      "epoch": 0.7172995780590717,
      "grad_norm": 0.49596792459487915,
      "learning_rate": 0.00028210229306876,
      "loss": 1.4156,
      "step": 6800
    },
    {
      "epoch": 0.7183544303797469,
      "grad_norm": 0.6298664808273315,
      "learning_rate": 0.0002801428039111279,
      "loss": 1.4216,
      "step": 6810
    },
    {
      "epoch": 0.7194092827004219,
      "grad_norm": 0.4989502727985382,
      "learning_rate": 0.00027818857973280274,
      "loss": 1.4235,
      "step": 6820
    },
    {
      "epoch": 0.7204641350210971,
      "grad_norm": 0.6086215376853943,
      "learning_rate": 0.0002762396424318206,
      "loss": 1.4186,
      "step": 6830
    },
    {
      "epoch": 0.7215189873417721,
      "grad_norm": 0.504088819026947,
      "learning_rate": 0.00027429601384697526,
      "loss": 1.4104,
      "step": 6840
    },
    {
      "epoch": 0.7225738396624473,
      "grad_norm": 0.517050564289093,
      "learning_rate": 0.00027235771575757466,
      "loss": 1.4184,
      "step": 6850
    },
    {
      "epoch": 0.7236286919831224,
      "grad_norm": 0.6705247759819031,
      "learning_rate": 0.0002704247698831951,
      "loss": 1.4165,
      "step": 6860
    },
    {
      "epoch": 0.7246835443037974,
      "grad_norm": 0.5113469958305359,
      "learning_rate": 0.0002684971978834389,
      "loss": 1.4125,
      "step": 6870
    },
    {
      "epoch": 0.7257383966244726,
      "grad_norm": 0.5772517323493958,
      "learning_rate": 0.0002665750213576914,
      "loss": 1.4188,
      "step": 6880
    },
    {
      "epoch": 0.7267932489451476,
      "grad_norm": 0.6171859502792358,
      "learning_rate": 0.0002646582618448794,
      "loss": 1.4113,
      "step": 6890
    },
    {
      "epoch": 0.7278481012658228,
      "grad_norm": 0.5865874290466309,
      "learning_rate": 0.00026274694082322896,
      "loss": 1.4122,
      "step": 6900
    },
    {
      "epoch": 0.7289029535864979,
      "grad_norm": 0.48441967368125916,
      "learning_rate": 0.0002608410797100255,
      "loss": 1.4378,
      "step": 6910
    },
    {
      "epoch": 0.729957805907173,
      "grad_norm": 0.6300725340843201,
      "learning_rate": 0.0002589406998613733,
      "loss": 1.4158,
      "step": 6920
    },
    {
      "epoch": 0.7310126582278481,
      "grad_norm": 0.5188561677932739,
      "learning_rate": 0.0002570458225719567,
      "loss": 1.4208,
      "step": 6930
    },
    {
      "epoch": 0.7320675105485233,
      "grad_norm": 0.512396514415741,
      "learning_rate": 0.00025515646907480074,
      "loss": 1.4254,
      "step": 6940
    },
    {
      "epoch": 0.7331223628691983,
      "grad_norm": 0.5857425332069397,
      "learning_rate": 0.00025327266054103395,
      "loss": 1.4213,
      "step": 6950
    },
    {
      "epoch": 0.7341772151898734,
      "grad_norm": 0.47508955001831055,
      "learning_rate": 0.0002513944180796509,
      "loss": 1.4111,
      "step": 6960
    },
    {
      "epoch": 0.7352320675105485,
      "grad_norm": 0.48741745948791504,
      "learning_rate": 0.0002495217627372752,
      "loss": 1.4074,
      "step": 6970
    },
    {
      "epoch": 0.7362869198312236,
      "grad_norm": 0.4814155697822571,
      "learning_rate": 0.0002476547154979248,
      "loss": 1.4098,
      "step": 6980
    },
    {
      "epoch": 0.7373417721518988,
      "grad_norm": 0.5336549878120422,
      "learning_rate": 0.00024579329728277534,
      "loss": 1.4127,
      "step": 6990
    },
    {
      "epoch": 0.7383966244725738,
      "grad_norm": 0.5137410163879395,
      "learning_rate": 0.00024393752894992708,
      "loss": 1.4231,
      "step": 7000
    },
    {
      "epoch": 0.739451476793249,
      "grad_norm": 0.47526130080223083,
      "learning_rate": 0.00024208743129417004,
      "loss": 1.3962,
      "step": 7010
    },
    {
      "epoch": 0.740506329113924,
      "grad_norm": 0.8842220902442932,
      "learning_rate": 0.00024024302504675206,
      "loss": 1.4106,
      "step": 7020
    },
    {
      "epoch": 0.7415611814345991,
      "grad_norm": 0.5335461497306824,
      "learning_rate": 0.0002384043308751454,
      "loss": 1.4144,
      "step": 7030
    },
    {
      "epoch": 0.7426160337552743,
      "grad_norm": 0.5184959769248962,
      "learning_rate": 0.00023657136938281653,
      "loss": 1.4188,
      "step": 7040
    },
    {
      "epoch": 0.7436708860759493,
      "grad_norm": 0.4702299237251282,
      "learning_rate": 0.00023474416110899377,
      "loss": 1.407,
      "step": 7050
    },
    {
      "epoch": 0.7447257383966245,
      "grad_norm": 0.4945663511753082,
      "learning_rate": 0.00023292272652843807,
      "loss": 1.4127,
      "step": 7060
    },
    {
      "epoch": 0.7457805907172996,
      "grad_norm": 0.4893150329589844,
      "learning_rate": 0.00023110708605121317,
      "loss": 1.4206,
      "step": 7070
    },
    {
      "epoch": 0.7468354430379747,
      "grad_norm": 0.5535306334495544,
      "learning_rate": 0.00022929726002245728,
      "loss": 1.4136,
      "step": 7080
    },
    {
      "epoch": 0.7478902953586498,
      "grad_norm": 0.502993106842041,
      "learning_rate": 0.00022749326872215472,
      "loss": 1.4129,
      "step": 7090
    },
    {
      "epoch": 0.7489451476793249,
      "grad_norm": 0.5133369565010071,
      "learning_rate": 0.0002256951323649087,
      "loss": 1.4093,
      "step": 7100
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6289997696876526,
      "learning_rate": 0.00022390287109971547,
      "loss": 1.4204,
      "step": 7110
    },
    {
      "epoch": 0.7510548523206751,
      "grad_norm": 0.605743408203125,
      "learning_rate": 0.00022211650500973746,
      "loss": 1.4171,
      "step": 7120
    },
    {
      "epoch": 0.7521097046413502,
      "grad_norm": 0.55927574634552,
      "learning_rate": 0.0002203360541120789,
      "loss": 1.4226,
      "step": 7130
    },
    {
      "epoch": 0.7531645569620253,
      "grad_norm": 0.4868223965167999,
      "learning_rate": 0.00021856153835756164,
      "loss": 1.4065,
      "step": 7140
    },
    {
      "epoch": 0.7542194092827004,
      "grad_norm": 0.7868787050247192,
      "learning_rate": 0.00021679297763050104,
      "loss": 1.4025,
      "step": 7150
    },
    {
      "epoch": 0.7552742616033755,
      "grad_norm": 0.5276433229446411,
      "learning_rate": 0.0002150303917484834,
      "loss": 1.4139,
      "step": 7160
    },
    {
      "epoch": 0.7563291139240507,
      "grad_norm": 0.8709946274757385,
      "learning_rate": 0.0002132738004621446,
      "loss": 1.4194,
      "step": 7170
    },
    {
      "epoch": 0.7573839662447257,
      "grad_norm": 0.4975385367870331,
      "learning_rate": 0.00021152322345494763,
      "loss": 1.4005,
      "step": 7180
    },
    {
      "epoch": 0.7584388185654009,
      "grad_norm": 0.782618522644043,
      "learning_rate": 0.00020977868034296253,
      "loss": 1.4013,
      "step": 7190
    },
    {
      "epoch": 0.759493670886076,
      "grad_norm": 0.6031981706619263,
      "learning_rate": 0.00020804019067464667,
      "loss": 1.4081,
      "step": 7200
    },
    {
      "epoch": 0.760548523206751,
      "grad_norm": 0.47875404357910156,
      "learning_rate": 0.00020630777393062575,
      "loss": 1.4112,
      "step": 7210
    },
    {
      "epoch": 0.7616033755274262,
      "grad_norm": 0.5568411350250244,
      "learning_rate": 0.00020458144952347523,
      "loss": 1.4234,
      "step": 7220
    },
    {
      "epoch": 0.7626582278481012,
      "grad_norm": 0.4957244098186493,
      "learning_rate": 0.00020286123679750314,
      "loss": 1.4181,
      "step": 7230
    },
    {
      "epoch": 0.7637130801687764,
      "grad_norm": 0.5013930797576904,
      "learning_rate": 0.00020114715502853292,
      "loss": 1.417,
      "step": 7240
    },
    {
      "epoch": 0.7647679324894515,
      "grad_norm": 0.742794930934906,
      "learning_rate": 0.0001994392234236878,
      "loss": 1.4109,
      "step": 7250
    },
    {
      "epoch": 0.7658227848101266,
      "grad_norm": 0.5725691914558411,
      "learning_rate": 0.0001977374611211754,
      "loss": 1.4096,
      "step": 7260
    },
    {
      "epoch": 0.7668776371308017,
      "grad_norm": 0.5211153030395508,
      "learning_rate": 0.00019604188719007313,
      "loss": 1.4062,
      "step": 7270
    },
    {
      "epoch": 0.7679324894514767,
      "grad_norm": 0.5076520442962646,
      "learning_rate": 0.00019435252063011504,
      "loss": 1.4084,
      "step": 7280
    },
    {
      "epoch": 0.7689873417721519,
      "grad_norm": 0.5510384440422058,
      "learning_rate": 0.0001926693803714779,
      "loss": 1.4192,
      "step": 7290
    },
    {
      "epoch": 0.770042194092827,
      "grad_norm": 0.5101417303085327,
      "learning_rate": 0.00019099248527457068,
      "loss": 1.4254,
      "step": 7300
    },
    {
      "epoch": 0.7710970464135021,
      "grad_norm": 0.4743573069572449,
      "learning_rate": 0.0001893218541298216,
      "loss": 1.4035,
      "step": 7310
    },
    {
      "epoch": 0.7721518987341772,
      "grad_norm": 0.6846520900726318,
      "learning_rate": 0.00018765750565746827,
      "loss": 1.3998,
      "step": 7320
    },
    {
      "epoch": 0.7732067510548524,
      "grad_norm": 0.5854378938674927,
      "learning_rate": 0.00018599945850734812,
      "loss": 1.4054,
      "step": 7330
    },
    {
      "epoch": 0.7742616033755274,
      "grad_norm": 0.7829344272613525,
      "learning_rate": 0.00018434773125868895,
      "loss": 1.4029,
      "step": 7340
    },
    {
      "epoch": 0.7753164556962026,
      "grad_norm": 0.49994421005249023,
      "learning_rate": 0.00018270234241990108,
      "loss": 1.3987,
      "step": 7350
    },
    {
      "epoch": 0.7763713080168776,
      "grad_norm": 0.5216307640075684,
      "learning_rate": 0.0001810633104283698,
      "loss": 1.4063,
      "step": 7360
    },
    {
      "epoch": 0.7774261603375527,
      "grad_norm": 0.4907887279987335,
      "learning_rate": 0.0001794306536502492,
      "loss": 1.4049,
      "step": 7370
    },
    {
      "epoch": 0.7784810126582279,
      "grad_norm": 0.5378735065460205,
      "learning_rate": 0.0001778043903802555,
      "loss": 1.3946,
      "step": 7380
    },
    {
      "epoch": 0.7795358649789029,
      "grad_norm": 0.48792096972465515,
      "learning_rate": 0.0001761845388414627,
      "loss": 1.4063,
      "step": 7390
    },
    {
      "epoch": 0.7805907172995781,
      "grad_norm": 0.5851631164550781,
      "learning_rate": 0.00017457111718509831,
      "loss": 1.4133,
      "step": 7400
    },
    {
      "epoch": 0.7816455696202531,
      "grad_norm": 0.5007026195526123,
      "learning_rate": 0.00017296414349033976,
      "loss": 1.3926,
      "step": 7410
    },
    {
      "epoch": 0.7827004219409283,
      "grad_norm": 0.5991578698158264,
      "learning_rate": 0.00017136363576411172,
      "loss": 1.3987,
      "step": 7420
    },
    {
      "epoch": 0.7837552742616034,
      "grad_norm": 0.5213151574134827,
      "learning_rate": 0.00016976961194088526,
      "loss": 1.3861,
      "step": 7430
    },
    {
      "epoch": 0.7848101265822784,
      "grad_norm": 0.5028907060623169,
      "learning_rate": 0.00016818208988247533,
      "loss": 1.3906,
      "step": 7440
    },
    {
      "epoch": 0.7858649789029536,
      "grad_norm": 0.5429515242576599,
      "learning_rate": 0.0001666010873778419,
      "loss": 1.3933,
      "step": 7450
    },
    {
      "epoch": 0.7869198312236287,
      "grad_norm": 0.501041054725647,
      "learning_rate": 0.00016502662214289,
      "loss": 1.4043,
      "step": 7460
    },
    {
      "epoch": 0.7879746835443038,
      "grad_norm": 0.5604754090309143,
      "learning_rate": 0.00016345871182027124,
      "loss": 1.3939,
      "step": 7470
    },
    {
      "epoch": 0.7890295358649789,
      "grad_norm": 0.7888116240501404,
      "learning_rate": 0.00016189737397918653,
      "loss": 1.4003,
      "step": 7480
    },
    {
      "epoch": 0.790084388185654,
      "grad_norm": 0.4839627146720886,
      "learning_rate": 0.0001603426261151884,
      "loss": 1.4093,
      "step": 7490
    },
    {
      "epoch": 0.7911392405063291,
      "grad_norm": 0.49811026453971863,
      "learning_rate": 0.00015879448564998648,
      "loss": 1.4049,
      "step": 7500
    },
    {
      "epoch": 0.7921940928270043,
      "grad_norm": 0.47760209441185,
      "learning_rate": 0.0001572529699312501,
      "loss": 1.4091,
      "step": 7510
    },
    {
      "epoch": 0.7932489451476793,
      "grad_norm": 0.5026172995567322,
      "learning_rate": 0.0001557180962324158,
      "loss": 1.3937,
      "step": 7520
    },
    {
      "epoch": 0.7943037974683544,
      "grad_norm": 0.5351853966712952,
      "learning_rate": 0.00015418988175249282,
      "loss": 1.3954,
      "step": 7530
    },
    {
      "epoch": 0.7953586497890295,
      "grad_norm": 0.48242607712745667,
      "learning_rate": 0.00015266834361587063,
      "loss": 1.3946,
      "step": 7540
    },
    {
      "epoch": 0.7964135021097046,
      "grad_norm": 0.48497846722602844,
      "learning_rate": 0.00015115349887212678,
      "loss": 1.3898,
      "step": 7550
    },
    {
      "epoch": 0.7974683544303798,
      "grad_norm": 0.579359769821167,
      "learning_rate": 0.00014964536449583657,
      "loss": 1.3918,
      "step": 7560
    },
    {
      "epoch": 0.7985232067510548,
      "grad_norm": 0.5208415389060974,
      "learning_rate": 0.00014814395738638195,
      "loss": 1.405,
      "step": 7570
    },
    {
      "epoch": 0.79957805907173,
      "grad_norm": 0.5188043713569641,
      "learning_rate": 0.00014664929436776278,
      "loss": 1.4002,
      "step": 7580
    },
    {
      "epoch": 0.8006329113924051,
      "grad_norm": 0.49395835399627686,
      "learning_rate": 0.00014516139218840788,
      "loss": 1.3942,
      "step": 7590
    },
    {
      "epoch": 0.8016877637130801,
      "grad_norm": 0.4898037910461426,
      "learning_rate": 0.00014368026752098782,
      "loss": 1.394,
      "step": 7600
    },
    {
      "epoch": 0.8027426160337553,
      "grad_norm": 0.5861129760742188,
      "learning_rate": 0.00014220593696222768,
      "loss": 1.4089,
      "step": 7610
    },
    {
      "epoch": 0.8037974683544303,
      "grad_norm": 0.49235138297080994,
      "learning_rate": 0.00014073841703272092,
      "loss": 1.3795,
      "step": 7620
    },
    {
      "epoch": 0.8048523206751055,
      "grad_norm": 0.49516093730926514,
      "learning_rate": 0.00013927772417674558,
      "loss": 1.4054,
      "step": 7630
    },
    {
      "epoch": 0.8059071729957806,
      "grad_norm": 0.48757612705230713,
      "learning_rate": 0.00013782387476207788,
      "loss": 1.3962,
      "step": 7640
    },
    {
      "epoch": 0.8069620253164557,
      "grad_norm": 0.47646474838256836,
      "learning_rate": 0.00013637688507981064,
      "loss": 1.4117,
      "step": 7650
    },
    {
      "epoch": 0.8080168776371308,
      "grad_norm": 0.4907473027706146,
      "learning_rate": 0.0001349367713441697,
      "loss": 1.3903,
      "step": 7660
    },
    {
      "epoch": 0.8090717299578059,
      "grad_norm": 0.47686466574668884,
      "learning_rate": 0.0001335035496923326,
      "loss": 1.3957,
      "step": 7670
    },
    {
      "epoch": 0.810126582278481,
      "grad_norm": 0.5995292067527771,
      "learning_rate": 0.0001320772361842478,
      "loss": 1.4121,
      "step": 7680
    },
    {
      "epoch": 0.8111814345991561,
      "grad_norm": 0.46882882714271545,
      "learning_rate": 0.00013065784680245442,
      "loss": 1.3854,
      "step": 7690
    },
    {
      "epoch": 0.8122362869198312,
      "grad_norm": 0.4745437800884247,
      "learning_rate": 0.00012924539745190402,
      "loss": 1.3952,
      "step": 7700
    },
    {
      "epoch": 0.8132911392405063,
      "grad_norm": 0.5046836137771606,
      "learning_rate": 0.0001278399039597809,
      "loss": 1.3924,
      "step": 7710
    },
    {
      "epoch": 0.8143459915611815,
      "grad_norm": 0.5603382587432861,
      "learning_rate": 0.0001264413820753261,
      "loss": 1.3962,
      "step": 7720
    },
    {
      "epoch": 0.8154008438818565,
      "grad_norm": 0.5876381397247314,
      "learning_rate": 0.00012504984746966003,
      "loss": 1.4011,
      "step": 7730
    },
    {
      "epoch": 0.8164556962025317,
      "grad_norm": 0.4754839241504669,
      "learning_rate": 0.00012366531573560754,
      "loss": 1.4011,
      "step": 7740
    },
    {
      "epoch": 0.8175105485232067,
      "grad_norm": 0.48102545738220215,
      "learning_rate": 0.00012228780238752264,
      "loss": 1.3875,
      "step": 7750
    },
    {
      "epoch": 0.8185654008438819,
      "grad_norm": 0.634244978427887,
      "learning_rate": 0.00012091732286111514,
      "loss": 1.388,
      "step": 7760
    },
    {
      "epoch": 0.819620253164557,
      "grad_norm": 0.5213733911514282,
      "learning_rate": 0.00011955389251327737,
      "loss": 1.3971,
      "step": 7770
    },
    {
      "epoch": 0.820675105485232,
      "grad_norm": 0.6531879901885986,
      "learning_rate": 0.00011819752662191197,
      "loss": 1.3887,
      "step": 7780
    },
    {
      "epoch": 0.8217299578059072,
      "grad_norm": 0.539393961429596,
      "learning_rate": 0.00011684824038576115,
      "loss": 1.4077,
      "step": 7790
    },
    {
      "epoch": 0.8227848101265823,
      "grad_norm": 0.47519391775131226,
      "learning_rate": 0.00011550604892423593,
      "loss": 1.3923,
      "step": 7800
    },
    {
      "epoch": 0.8238396624472574,
      "grad_norm": 0.5234396457672119,
      "learning_rate": 0.0001141709672772471,
      "loss": 1.3932,
      "step": 7810
    },
    {
      "epoch": 0.8248945147679325,
      "grad_norm": 0.48334819078445435,
      "learning_rate": 0.00011284301040503625,
      "loss": 1.3999,
      "step": 7820
    },
    {
      "epoch": 0.8259493670886076,
      "grad_norm": 0.4964877665042877,
      "learning_rate": 0.0001115221931880088,
      "loss": 1.3906,
      "step": 7830
    },
    {
      "epoch": 0.8270042194092827,
      "grad_norm": 0.499929815530777,
      "learning_rate": 0.00011020853042656648,
      "loss": 1.3781,
      "step": 7840
    },
    {
      "epoch": 0.8280590717299579,
      "grad_norm": 0.47219541668891907,
      "learning_rate": 0.000108902036840942,
      "loss": 1.3864,
      "step": 7850
    },
    {
      "epoch": 0.8291139240506329,
      "grad_norm": 0.5601077675819397,
      "learning_rate": 0.00010760272707103389,
      "loss": 1.3749,
      "step": 7860
    },
    {
      "epoch": 0.830168776371308,
      "grad_norm": 0.5578227639198303,
      "learning_rate": 0.00010631061567624259,
      "loss": 1.3898,
      "step": 7870
    },
    {
      "epoch": 0.8312236286919831,
      "grad_norm": 0.5889272093772888,
      "learning_rate": 0.00010502571713530706,
      "loss": 1.4064,
      "step": 7880
    },
    {
      "epoch": 0.8322784810126582,
      "grad_norm": 0.49493587017059326,
      "learning_rate": 0.00010374804584614308,
      "loss": 1.402,
      "step": 7890
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 0.4760115146636963,
      "learning_rate": 0.00010247761612568129,
      "loss": 1.3848,
      "step": 7900
    },
    {
      "epoch": 0.8343881856540084,
      "grad_norm": 0.7417068481445312,
      "learning_rate": 0.0001012144422097069,
      "loss": 1.393,
      "step": 7910
    },
    {
      "epoch": 0.8354430379746836,
      "grad_norm": 0.49573588371276855,
      "learning_rate": 9.995853825270052e-05,
      "loss": 1.392,
      "step": 7920
    },
    {
      "epoch": 0.8364978902953587,
      "grad_norm": 0.5548797249794006,
      "learning_rate": 9.870991832767919e-05,
      "loss": 1.4008,
      "step": 7930
    },
    {
      "epoch": 0.8375527426160337,
      "grad_norm": 0.4964168965816498,
      "learning_rate": 9.746859642603884e-05,
      "loss": 1.3946,
      "step": 7940
    },
    {
      "epoch": 0.8386075949367089,
      "grad_norm": 0.47221243381500244,
      "learning_rate": 9.623458645739755e-05,
      "loss": 1.396,
      "step": 7950
    },
    {
      "epoch": 0.8396624472573839,
      "grad_norm": 0.4813820421695709,
      "learning_rate": 9.50079022494395e-05,
      "loss": 1.3842,
      "step": 7960
    },
    {
      "epoch": 0.8407172995780591,
      "grad_norm": 0.5066344141960144,
      "learning_rate": 9.378855754776028e-05,
      "loss": 1.3839,
      "step": 7970
    },
    {
      "epoch": 0.8417721518987342,
      "grad_norm": 0.4867765009403229,
      "learning_rate": 9.257656601571266e-05,
      "loss": 1.3828,
      "step": 7980
    },
    {
      "epoch": 0.8428270042194093,
      "grad_norm": 0.5355703234672546,
      "learning_rate": 9.137194123425349e-05,
      "loss": 1.3944,
      "step": 7990
    },
    {
      "epoch": 0.8438818565400844,
      "grad_norm": 0.5081289410591125,
      "learning_rate": 9.017469670179168e-05,
      "loss": 1.3902,
      "step": 8000
    },
    {
      "epoch": 0.8449367088607594,
      "grad_norm": 0.4760856032371521,
      "learning_rate": 8.898484583403668e-05,
      "loss": 1.394,
      "step": 8010
    },
    {
      "epoch": 0.8459915611814346,
      "grad_norm": 0.4752003848552704,
      "learning_rate": 8.780240196384873e-05,
      "loss": 1.3862,
      "step": 8020
    },
    {
      "epoch": 0.8470464135021097,
      "grad_norm": 0.48132503032684326,
      "learning_rate": 8.662737834108861e-05,
      "loss": 1.3867,
      "step": 8030
    },
    {
      "epoch": 0.8481012658227848,
      "grad_norm": 0.48374220728874207,
      "learning_rate": 8.545978813246987e-05,
      "loss": 1.4,
      "step": 8040
    },
    {
      "epoch": 0.8491561181434599,
      "grad_norm": 0.49021220207214355,
      "learning_rate": 8.429964442141072e-05,
      "loss": 1.3774,
      "step": 8050
    },
    {
      "epoch": 0.8502109704641351,
      "grad_norm": 0.5363766551017761,
      "learning_rate": 8.314696020788806e-05,
      "loss": 1.3857,
      "step": 8060
    },
    {
      "epoch": 0.8512658227848101,
      "grad_norm": 0.5091484785079956,
      "learning_rate": 8.200174840829136e-05,
      "loss": 1.394,
      "step": 8070
    },
    {
      "epoch": 0.8523206751054853,
      "grad_norm": 0.48978227376937866,
      "learning_rate": 8.08640218552778e-05,
      "loss": 1.3967,
      "step": 8080
    },
    {
      "epoch": 0.8533755274261603,
      "grad_norm": 0.48981067538261414,
      "learning_rate": 7.973379329762925e-05,
      "loss": 1.3842,
      "step": 8090
    },
    {
      "epoch": 0.8544303797468354,
      "grad_norm": 0.5715848803520203,
      "learning_rate": 7.861107540010845e-05,
      "loss": 1.3759,
      "step": 8100
    },
    {
      "epoch": 0.8554852320675106,
      "grad_norm": 0.4878443777561188,
      "learning_rate": 7.749588074331762e-05,
      "loss": 1.3957,
      "step": 8110
    },
    {
      "epoch": 0.8565400843881856,
      "grad_norm": 0.495976060628891,
      "learning_rate": 7.63882218235575e-05,
      "loss": 1.3845,
      "step": 8120
    },
    {
      "epoch": 0.8575949367088608,
      "grad_norm": 0.48120617866516113,
      "learning_rate": 7.528811105268699e-05,
      "loss": 1.3869,
      "step": 8130
    },
    {
      "epoch": 0.8586497890295358,
      "grad_norm": 0.4755679965019226,
      "learning_rate": 7.41955607579845e-05,
      "loss": 1.3815,
      "step": 8140
    },
    {
      "epoch": 0.859704641350211,
      "grad_norm": 0.4891829490661621,
      "learning_rate": 7.311058318200969e-05,
      "loss": 1.3787,
      "step": 8150
    },
    {
      "epoch": 0.8607594936708861,
      "grad_norm": 0.5429104566574097,
      "learning_rate": 7.203319048246599e-05,
      "loss": 1.393,
      "step": 8160
    },
    {
      "epoch": 0.8618143459915611,
      "grad_norm": 0.5007178783416748,
      "learning_rate": 7.096339473206471e-05,
      "loss": 1.3828,
      "step": 8170
    },
    {
      "epoch": 0.8628691983122363,
      "grad_norm": 0.504002034664154,
      "learning_rate": 6.990120791838953e-05,
      "loss": 1.3925,
      "step": 8180
    },
    {
      "epoch": 0.8639240506329114,
      "grad_norm": 0.6871325969696045,
      "learning_rate": 6.884664194376233e-05,
      "loss": 1.3841,
      "step": 8190
    },
    {
      "epoch": 0.8649789029535865,
      "grad_norm": 0.49031054973602295,
      "learning_rate": 6.779970862510989e-05,
      "loss": 1.3939,
      "step": 8200
    },
    {
      "epoch": 0.8660337552742616,
      "grad_norm": 0.5044110417366028,
      "learning_rate": 6.676041969383107e-05,
      "loss": 1.3852,
      "step": 8210
    },
    {
      "epoch": 0.8670886075949367,
      "grad_norm": 0.5476711988449097,
      "learning_rate": 6.572878679566605e-05,
      "loss": 1.3932,
      "step": 8220
    },
    {
      "epoch": 0.8681434599156118,
      "grad_norm": 0.5918845534324646,
      "learning_rate": 6.470482149056509e-05,
      "loss": 1.3891,
      "step": 8230
    },
    {
      "epoch": 0.869198312236287,
      "grad_norm": 0.4630289673805237,
      "learning_rate": 6.368853525255942e-05,
      "loss": 1.3872,
      "step": 8240
    },
    {
      "epoch": 0.870253164556962,
      "grad_norm": 0.5118920207023621,
      "learning_rate": 6.267993946963249e-05,
      "loss": 1.4014,
      "step": 8250
    },
    {
      "epoch": 0.8713080168776371,
      "grad_norm": 0.47048085927963257,
      "learning_rate": 6.167904544359265e-05,
      "loss": 1.3924,
      "step": 8260
    },
    {
      "epoch": 0.8723628691983122,
      "grad_norm": 0.48211798071861267,
      "learning_rate": 6.068586438994617e-05,
      "loss": 1.389,
      "step": 8270
    },
    {
      "epoch": 0.8734177215189873,
      "grad_norm": 0.4633164405822754,
      "learning_rate": 5.970040743777161e-05,
      "loss": 1.3709,
      "step": 8280
    },
    {
      "epoch": 0.8744725738396625,
      "grad_norm": 0.4633491039276123,
      "learning_rate": 5.8722685629595454e-05,
      "loss": 1.3727,
      "step": 8290
    },
    {
      "epoch": 0.8755274261603375,
      "grad_norm": 0.46621644496917725,
      "learning_rate": 5.7752709921267855e-05,
      "loss": 1.3976,
      "step": 8300
    },
    {
      "epoch": 0.8765822784810127,
      "grad_norm": 0.5643050074577332,
      "learning_rate": 5.6790491181840294e-05,
      "loss": 1.376,
      "step": 8310
    },
    {
      "epoch": 0.8776371308016878,
      "grad_norm": 0.5608354210853577,
      "learning_rate": 5.583604019344354e-05,
      "loss": 1.3941,
      "step": 8320
    },
    {
      "epoch": 0.8786919831223629,
      "grad_norm": 0.5177557468414307,
      "learning_rate": 5.4889367651167007e-05,
      "loss": 1.3904,
      "step": 8330
    },
    {
      "epoch": 0.879746835443038,
      "grad_norm": 0.4648410677909851,
      "learning_rate": 5.3950484162938714e-05,
      "loss": 1.3766,
      "step": 8340
    },
    {
      "epoch": 0.880801687763713,
      "grad_norm": 0.47074922919273376,
      "learning_rate": 5.3019400249406686e-05,
      "loss": 1.3796,
      "step": 8350
    },
    {
      "epoch": 0.8818565400843882,
      "grad_norm": 0.4819316864013672,
      "learning_rate": 5.209612634382077e-05,
      "loss": 1.378,
      "step": 8360
    },
    {
      "epoch": 0.8829113924050633,
      "grad_norm": 0.5172094106674194,
      "learning_rate": 5.118067279191599e-05,
      "loss": 1.385,
      "step": 8370
    },
    {
      "epoch": 0.8839662447257384,
      "grad_norm": 0.46945855021476746,
      "learning_rate": 5.0273049851796205e-05,
      "loss": 1.3923,
      "step": 8380
    },
    {
      "epoch": 0.8850210970464135,
      "grad_norm": 0.47022953629493713,
      "learning_rate": 4.9373267693819805e-05,
      "loss": 1.381,
      "step": 8390
    },
    {
      "epoch": 0.8860759493670886,
      "grad_norm": 0.4940224587917328,
      "learning_rate": 4.848133640048513e-05,
      "loss": 1.381,
      "step": 8400
    },
    {
      "epoch": 0.8871308016877637,
      "grad_norm": 0.5979254245758057,
      "learning_rate": 4.75972659663178e-05,
      "loss": 1.4,
      "step": 8410
    },
    {
      "epoch": 0.8881856540084389,
      "grad_norm": 0.48716863989830017,
      "learning_rate": 4.672106629775882e-05,
      "loss": 1.3876,
      "step": 8420
    },
    {
      "epoch": 0.8892405063291139,
      "grad_norm": 0.4779558777809143,
      "learning_rate": 4.585274721305333e-05,
      "loss": 1.3826,
      "step": 8430
    },
    {
      "epoch": 0.890295358649789,
      "grad_norm": 0.48549434542655945,
      "learning_rate": 4.4992318442140575e-05,
      "loss": 1.3785,
      "step": 8440
    },
    {
      "epoch": 0.8913502109704642,
      "grad_norm": 0.5071036219596863,
      "learning_rate": 4.413978962654508e-05,
      "loss": 1.3876,
      "step": 8450
    },
    {
      "epoch": 0.8924050632911392,
      "grad_norm": 0.4944823980331421,
      "learning_rate": 4.3295170319268554e-05,
      "loss": 1.3839,
      "step": 8460
    },
    {
      "epoch": 0.8934599156118144,
      "grad_norm": 0.4969879984855652,
      "learning_rate": 4.245846998468261e-05,
      "loss": 1.3809,
      "step": 8470
    },
    {
      "epoch": 0.8945147679324894,
      "grad_norm": 0.47751787304878235,
      "learning_rate": 4.16296979984232e-05,
      "loss": 1.3748,
      "step": 8480
    },
    {
      "epoch": 0.8955696202531646,
      "grad_norm": 0.5304287672042847,
      "learning_rate": 4.080886364728506e-05,
      "loss": 1.3839,
      "step": 8490
    },
    {
      "epoch": 0.8966244725738397,
      "grad_norm": 0.48898303508758545,
      "learning_rate": 3.999597612911793e-05,
      "loss": 1.3719,
      "step": 8500
    },
    {
      "epoch": 0.8976793248945147,
      "grad_norm": 0.4708358347415924,
      "learning_rate": 3.9191044552723345e-05,
      "loss": 1.385,
      "step": 8510
    },
    {
      "epoch": 0.8987341772151899,
      "grad_norm": 0.4926777482032776,
      "learning_rate": 3.839407793775268e-05,
      "loss": 1.3802,
      "step": 8520
    },
    {
      "epoch": 0.8997890295358649,
      "grad_norm": 0.4793872833251953,
      "learning_rate": 3.760508521460584e-05,
      "loss": 1.3966,
      "step": 8530
    },
    {
      "epoch": 0.9008438818565401,
      "grad_norm": 0.4818679988384247,
      "learning_rate": 3.682407522433173e-05,
      "loss": 1.3911,
      "step": 8540
    },
    {
      "epoch": 0.9018987341772152,
      "grad_norm": 0.48742154240608215,
      "learning_rate": 3.605105671852854e-05,
      "loss": 1.3805,
      "step": 8550
    },
    {
      "epoch": 0.9029535864978903,
      "grad_norm": 0.4817183017730713,
      "learning_rate": 3.528603835924626e-05,
      "loss": 1.3743,
      "step": 8560
    },
    {
      "epoch": 0.9040084388185654,
      "grad_norm": 0.5547499656677246,
      "learning_rate": 3.4529028718888935e-05,
      "loss": 1.3931,
      "step": 8570
    },
    {
      "epoch": 0.9050632911392406,
      "grad_norm": 0.5378564596176147,
      "learning_rate": 3.378003628011938e-05,
      "loss": 1.3838,
      "step": 8580
    },
    {
      "epoch": 0.9061181434599156,
      "grad_norm": 0.648749589920044,
      "learning_rate": 3.303906943576346e-05,
      "loss": 1.387,
      "step": 8590
    },
    {
      "epoch": 0.9071729957805907,
      "grad_norm": 0.5041891932487488,
      "learning_rate": 3.230613648871661e-05,
      "loss": 1.3801,
      "step": 8600
    },
    {
      "epoch": 0.9082278481012658,
      "grad_norm": 0.5527295470237732,
      "learning_rate": 3.158124565185022e-05,
      "loss": 1.3801,
      "step": 8610
    },
    {
      "epoch": 0.9092827004219409,
      "grad_norm": 0.47023510932922363,
      "learning_rate": 3.086440504792026e-05,
      "loss": 1.378,
      "step": 8620
    },
    {
      "epoch": 0.9103375527426161,
      "grad_norm": 0.5023501515388489,
      "learning_rate": 3.015562270947553e-05,
      "loss": 1.3898,
      "step": 8630
    },
    {
      "epoch": 0.9113924050632911,
      "grad_norm": 0.5792856216430664,
      "learning_rate": 2.945490657876837e-05,
      "loss": 1.3669,
      "step": 8640
    },
    {
      "epoch": 0.9124472573839663,
      "grad_norm": 0.48649540543556213,
      "learning_rate": 2.8762264507665113e-05,
      "loss": 1.3743,
      "step": 8650
    },
    {
      "epoch": 0.9135021097046413,
      "grad_norm": 0.5258405804634094,
      "learning_rate": 2.807770425755829e-05,
      "loss": 1.3763,
      "step": 8660
    },
    {
      "epoch": 0.9145569620253164,
      "grad_norm": 0.4875074028968811,
      "learning_rate": 2.7401233499279866e-05,
      "loss": 1.382,
      "step": 8670
    },
    {
      "epoch": 0.9156118143459916,
      "grad_norm": 0.47191861271858215,
      "learning_rate": 2.6732859813014987e-05,
      "loss": 1.3905,
      "step": 8680
    },
    {
      "epoch": 0.9166666666666666,
      "grad_norm": 0.49229896068573,
      "learning_rate": 2.607259068821721e-05,
      "loss": 1.3829,
      "step": 8690
    },
    {
      "epoch": 0.9177215189873418,
      "grad_norm": 0.46301206946372986,
      "learning_rate": 2.5420433523524493e-05,
      "loss": 1.382,
      "step": 8700
    },
    {
      "epoch": 0.9187763713080169,
      "grad_norm": 0.4987659454345703,
      "learning_rate": 2.4776395626676162e-05,
      "loss": 1.38,
      "step": 8710
    },
    {
      "epoch": 0.919831223628692,
      "grad_norm": 0.5030317306518555,
      "learning_rate": 2.414048421443141e-05,
      "loss": 1.3761,
      "step": 8720
    },
    {
      "epoch": 0.9208860759493671,
      "grad_norm": 0.4857030212879181,
      "learning_rate": 2.3512706412488012e-05,
      "loss": 1.3881,
      "step": 8730
    },
    {
      "epoch": 0.9219409282700421,
      "grad_norm": 0.4724518656730652,
      "learning_rate": 2.2893069255402993e-05,
      "loss": 1.3764,
      "step": 8740
    },
    {
      "epoch": 0.9229957805907173,
      "grad_norm": 0.4726200997829437,
      "learning_rate": 2.2281579686513176e-05,
      "loss": 1.3732,
      "step": 8750
    },
    {
      "epoch": 0.9240506329113924,
      "grad_norm": 0.5828139185905457,
      "learning_rate": 2.1678244557857663e-05,
      "loss": 1.3622,
      "step": 8760
    },
    {
      "epoch": 0.9251054852320675,
      "grad_norm": 0.4617006778717041,
      "learning_rate": 2.1083070630101232e-05,
      "loss": 1.3738,
      "step": 8770
    },
    {
      "epoch": 0.9261603375527426,
      "grad_norm": 0.4608205258846283,
      "learning_rate": 2.0496064572458395e-05,
      "loss": 1.3848,
      "step": 8780
    },
    {
      "epoch": 0.9272151898734177,
      "grad_norm": 0.46868443489074707,
      "learning_rate": 1.991723296261863e-05,
      "loss": 1.3765,
      "step": 8790
    },
    {
      "epoch": 0.9282700421940928,
      "grad_norm": 0.4702487289905548,
      "learning_rate": 1.9346582286672686e-05,
      "loss": 1.3744,
      "step": 8800
    },
    {
      "epoch": 0.929324894514768,
      "grad_norm": 0.46541425585746765,
      "learning_rate": 1.878411893904014e-05,
      "loss": 1.3908,
      "step": 8810
    },
    {
      "epoch": 0.930379746835443,
      "grad_norm": 0.467693030834198,
      "learning_rate": 1.822984922239737e-05,
      "loss": 1.3794,
      "step": 8820
    },
    {
      "epoch": 0.9314345991561181,
      "grad_norm": 0.48465168476104736,
      "learning_rate": 1.7683779347607286e-05,
      "loss": 1.3927,
      "step": 8830
    },
    {
      "epoch": 0.9324894514767933,
      "grad_norm": 0.4728999733924866,
      "learning_rate": 1.714591543364938e-05,
      "loss": 1.3811,
      "step": 8840
    },
    {
      "epoch": 0.9335443037974683,
      "grad_norm": 0.5732617974281311,
      "learning_rate": 1.6616263507551437e-05,
      "loss": 1.3769,
      "step": 8850
    },
    {
      "epoch": 0.9345991561181435,
      "grad_norm": 0.5440745949745178,
      "learning_rate": 1.609482950432195e-05,
      "loss": 1.3782,
      "step": 8860
    },
    {
      "epoch": 0.9356540084388185,
      "grad_norm": 0.47789183259010315,
      "learning_rate": 1.5581619266883563e-05,
      "loss": 1.3821,
      "step": 8870
    },
    {
      "epoch": 0.9367088607594937,
      "grad_norm": 0.49290019273757935,
      "learning_rate": 1.5076638546007548e-05,
      "loss": 1.3851,
      "step": 8880
    },
    {
      "epoch": 0.9377637130801688,
      "grad_norm": 0.4680052697658539,
      "learning_rate": 1.457989300024945e-05,
      "loss": 1.3843,
      "step": 8890
    },
    {
      "epoch": 0.9388185654008439,
      "grad_norm": 0.4927351176738739,
      "learning_rate": 1.4091388195885625e-05,
      "loss": 1.3744,
      "step": 8900
    },
    {
      "epoch": 0.939873417721519,
      "grad_norm": 0.5219542980194092,
      "learning_rate": 1.3611129606851041e-05,
      "loss": 1.3729,
      "step": 8910
    },
    {
      "epoch": 0.9409282700421941,
      "grad_norm": 0.459949254989624,
      "learning_rate": 1.313912261467759e-05,
      "loss": 1.381,
      "step": 8920
    },
    {
      "epoch": 0.9419831223628692,
      "grad_norm": 0.4824971854686737,
      "learning_rate": 1.267537250843412e-05,
      "loss": 1.3766,
      "step": 8930
    },
    {
      "epoch": 0.9430379746835443,
      "grad_norm": 0.4652012288570404,
      "learning_rate": 1.2219884484667071e-05,
      "loss": 1.378,
      "step": 8940
    },
    {
      "epoch": 0.9440928270042194,
      "grad_norm": 0.45781955122947693,
      "learning_rate": 1.1772663647341947e-05,
      "loss": 1.3801,
      "step": 8950
    },
    {
      "epoch": 0.9451476793248945,
      "grad_norm": 0.4768141508102417,
      "learning_rate": 1.1333715007786932e-05,
      "loss": 1.384,
      "step": 8960
    },
    {
      "epoch": 0.9462025316455697,
      "grad_norm": 0.48779961466789246,
      "learning_rate": 1.0903043484635694e-05,
      "loss": 1.3817,
      "step": 8970
    },
    {
      "epoch": 0.9472573839662447,
      "grad_norm": 0.5315613150596619,
      "learning_rate": 1.0480653903772924e-05,
      "loss": 1.3762,
      "step": 8980
    },
    {
      "epoch": 0.9483122362869199,
      "grad_norm": 0.4680369794368744,
      "learning_rate": 1.0066550998280132e-05,
      "loss": 1.3828,
      "step": 8990
    },
    {
      "epoch": 0.9493670886075949,
      "grad_norm": 0.4570881128311157,
      "learning_rate": 9.660739408382608e-06,
      "loss": 1.3783,
      "step": 9000
    },
    {
      "epoch": 0.95042194092827,
      "grad_norm": 0.4943011701107025,
      "learning_rate": 9.26322368139737e-06,
      "loss": 1.3737,
      "step": 9010
    },
    {
      "epoch": 0.9514767932489452,
      "grad_norm": 0.4810599982738495,
      "learning_rate": 8.874008271682222e-06,
      "loss": 1.3761,
      "step": 9020
    },
    {
      "epoch": 0.9525316455696202,
      "grad_norm": 0.4643784463405609,
      "learning_rate": 8.493097540585775e-06,
      "loss": 1.3928,
      "step": 9030
    },
    {
      "epoch": 0.9535864978902954,
      "grad_norm": 0.4680810570716858,
      "learning_rate": 8.120495756399005e-06,
      "loss": 1.371,
      "step": 9040
    },
    {
      "epoch": 0.9546413502109705,
      "grad_norm": 0.47044894099235535,
      "learning_rate": 7.756207094306605e-06,
      "loss": 1.3828,
      "step": 9050
    },
    {
      "epoch": 0.9556962025316456,
      "grad_norm": 0.46625006198883057,
      "learning_rate": 7.400235636340957e-06,
      "loss": 1.3823,
      "step": 9060
    },
    {
      "epoch": 0.9567510548523207,
      "grad_norm": 0.4695632755756378,
      "learning_rate": 7.0525853713362395e-06,
      "loss": 1.3848,
      "step": 9070
    },
    {
      "epoch": 0.9578059071729957,
      "grad_norm": 0.5059646964073181,
      "learning_rate": 6.71326019488322e-06,
      "loss": 1.3835,
      "step": 9080
    },
    {
      "epoch": 0.9588607594936709,
      "grad_norm": 0.45554956793785095,
      "learning_rate": 6.3822639092862846e-06,
      "loss": 1.3841,
      "step": 9090
    },
    {
      "epoch": 0.959915611814346,
      "grad_norm": 0.46608665585517883,
      "learning_rate": 6.059600223520478e-06,
      "loss": 1.3633,
      "step": 9100
    },
    {
      "epoch": 0.9609704641350211,
      "grad_norm": 0.4648936986923218,
      "learning_rate": 5.745272753189784e-06,
      "loss": 1.3776,
      "step": 9110
    },
    {
      "epoch": 0.9620253164556962,
      "grad_norm": 0.4707690477371216,
      "learning_rate": 5.439285020487156e-06,
      "loss": 1.3837,
      "step": 9120
    },
    {
      "epoch": 0.9630801687763713,
      "grad_norm": 0.4885684847831726,
      "learning_rate": 5.141640454154467e-06,
      "loss": 1.3691,
      "step": 9130
    },
    {
      "epoch": 0.9641350210970464,
      "grad_norm": 0.47174009680747986,
      "learning_rate": 4.852342389444458e-06,
      "loss": 1.3909,
      "step": 9140
    },
    {
      "epoch": 0.9651898734177216,
      "grad_norm": 0.48270609974861145,
      "learning_rate": 4.571394068083185e-06,
      "loss": 1.3775,
      "step": 9150
    },
    {
      "epoch": 0.9662447257383966,
      "grad_norm": 0.4584430456161499,
      "learning_rate": 4.298798638233709e-06,
      "loss": 1.3858,
      "step": 9160
    },
    {
      "epoch": 0.9672995780590717,
      "grad_norm": 0.4954625368118286,
      "learning_rate": 4.034559154461049e-06,
      "loss": 1.3832,
      "step": 9170
    },
    {
      "epoch": 0.9683544303797469,
      "grad_norm": 0.45214712619781494,
      "learning_rate": 3.7786785776976198e-06,
      "loss": 1.3774,
      "step": 9180
    },
    {
      "epoch": 0.9694092827004219,
      "grad_norm": 0.46736812591552734,
      "learning_rate": 3.5311597752100964e-06,
      "loss": 1.3735,
      "step": 9190
    },
    {
      "epoch": 0.9704641350210971,
      "grad_norm": 0.4651713967323303,
      "learning_rate": 3.2920055205676867e-06,
      "loss": 1.3762,
      "step": 9200
    },
    {
      "epoch": 0.9715189873417721,
      "grad_norm": 0.46557387709617615,
      "learning_rate": 3.06121849361049e-06,
      "loss": 1.3785,
      "step": 9210
    },
    {
      "epoch": 0.9725738396624473,
      "grad_norm": 0.47982534766197205,
      "learning_rate": 2.838801280419856e-06,
      "loss": 1.3752,
      "step": 9220
    },
    {
      "epoch": 0.9736286919831224,
      "grad_norm": 0.46053677797317505,
      "learning_rate": 2.624756373289322e-06,
      "loss": 1.3676,
      "step": 9230
    },
    {
      "epoch": 0.9746835443037974,
      "grad_norm": 0.4777507483959198,
      "learning_rate": 2.419086170696472e-06,
      "loss": 1.3632,
      "step": 9240
    },
    {
      "epoch": 0.9757383966244726,
      "grad_norm": 0.45830753445625305,
      "learning_rate": 2.2217929772764545e-06,
      "loss": 1.3738,
      "step": 9250
    },
    {
      "epoch": 0.9767932489451476,
      "grad_norm": 0.460542768239975,
      "learning_rate": 2.0328790037957568e-06,
      "loss": 1.372,
      "step": 9260
    },
    {
      "epoch": 0.9778481012658228,
      "grad_norm": 0.45632460713386536,
      "learning_rate": 1.8523463671278052e-06,
      "loss": 1.3822,
      "step": 9270
    },
    {
      "epoch": 0.9789029535864979,
      "grad_norm": 0.46475285291671753,
      "learning_rate": 1.6801970902288188e-06,
      "loss": 1.3698,
      "step": 9280
    },
    {
      "epoch": 0.979957805907173,
      "grad_norm": 0.4633490741252899,
      "learning_rate": 1.5164331021155774e-06,
      "loss": 1.3782,
      "step": 9290
    },
    {
      "epoch": 0.9810126582278481,
      "grad_norm": 0.4666520357131958,
      "learning_rate": 1.3610562378435221e-06,
      "loss": 1.3754,
      "step": 9300
    },
    {
      "epoch": 0.9820675105485233,
      "grad_norm": 0.48110079765319824,
      "learning_rate": 1.2140682384862712e-06,
      "loss": 1.3764,
      "step": 9310
    },
    {
      "epoch": 0.9831223628691983,
      "grad_norm": 0.4777778089046478,
      "learning_rate": 1.0754707511161365e-06,
      "loss": 1.3712,
      "step": 9320
    },
    {
      "epoch": 0.9841772151898734,
      "grad_norm": 0.46745941042900085,
      "learning_rate": 9.452653287856383e-07,
      "loss": 1.3867,
      "step": 9330
    },
    {
      "epoch": 0.9852320675105485,
      "grad_norm": 0.4661557972431183,
      "learning_rate": 8.234534305101015e-07,
      "loss": 1.3805,
      "step": 9340
    },
    {
      "epoch": 0.9862869198312236,
      "grad_norm": 0.4665854573249817,
      "learning_rate": 7.100364212513367e-07,
      "loss": 1.3928,
      "step": 9350
    },
    {
      "epoch": 0.9873417721518988,
      "grad_norm": 0.47986653447151184,
      "learning_rate": 6.050155719023176e-07,
      "loss": 1.3842,
      "step": 9360
    },
    {
      "epoch": 0.9883966244725738,
      "grad_norm": 0.4748641550540924,
      "learning_rate": 5.08392059272944e-07,
      "loss": 1.3801,
      "step": 9370
    },
    {
      "epoch": 0.989451476793249,
      "grad_norm": 0.48322656750679016,
      "learning_rate": 4.2016696607680147e-07,
      "loss": 1.3675,
      "step": 9380
    },
    {
      "epoch": 0.990506329113924,
      "grad_norm": 0.477649986743927,
      "learning_rate": 3.4034128091917085e-07,
      "loss": 1.3728,
      "step": 9390
    },
    {
      "epoch": 0.9915611814345991,
      "grad_norm": 0.47399818897247314,
      "learning_rate": 2.689158982859541e-07,
      "loss": 1.389,
      "step": 9400
    },
    {
      "epoch": 0.9926160337552743,
      "grad_norm": 0.49044132232666016,
      "learning_rate": 2.05891618533266e-07,
      "loss": 1.3623,
      "step": 9410
    },
    {
      "epoch": 0.9936708860759493,
      "grad_norm": 0.47090208530426025,
      "learning_rate": 1.5126914787894074e-07,
      "loss": 1.3804,
      "step": 9420
    },
    {
      "epoch": 0.9947257383966245,
      "grad_norm": 0.4704456329345703,
      "learning_rate": 1.0504909839462173e-07,
      "loss": 1.3817,
      "step": 9430
    },
    {
      "epoch": 0.9957805907172996,
      "grad_norm": 0.4629286229610443,
      "learning_rate": 6.723198799826746e-08,
      "loss": 1.3774,
      "step": 9440
    },
    {
      "epoch": 0.9968354430379747,
      "grad_norm": 0.46754178404808044,
      "learning_rate": 3.781824044932214e-08,
      "loss": 1.3866,
      "step": 9450
    },
    {
      "epoch": 0.9978902953586498,
      "grad_norm": 0.46161213517189026,
      "learning_rate": 1.6808185342970238e-08,
      "loss": 1.3682,
      "step": 9460
    },
    {
      "epoch": 0.9989451476793249,
      "grad_norm": 0.4677896797657013,
      "learning_rate": 4.202058107305451e-09,
      "loss": 1.3826,
      "step": 9470
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.379104733467102,
      "learning_rate": 0.0,
      "loss": 1.3731,
      "step": 9480
    }
  ],
  "logging_steps": 10,
  "max_steps": 9480,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.839757817279693e+16,
  "train_batch_size": 1024,
  "trial_name": null,
  "trial_params": null
}