{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 2179,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0004589261128958238,
      "grad_norm": 0.24931495636064183,
      "learning_rate": 9.174311926605506e-07,
      "loss": 1.1206,
      "step": 1
    },
    {
      "epoch": 0.002294630564479119,
      "grad_norm": 0.26866738165199944,
      "learning_rate": 4.587155963302753e-06,
      "loss": 1.1242,
      "step": 5
    },
    {
      "epoch": 0.004589261128958238,
      "grad_norm": 0.2706384220584976,
      "learning_rate": 9.174311926605506e-06,
      "loss": 1.1504,
      "step": 10
    },
    {
      "epoch": 0.006883891693437357,
      "grad_norm": 0.22472049733931831,
      "learning_rate": 1.3761467889908258e-05,
      "loss": 1.1168,
      "step": 15
    },
    {
      "epoch": 0.009178522257916476,
      "grad_norm": 0.20645077000731837,
      "learning_rate": 1.834862385321101e-05,
      "loss": 1.0668,
      "step": 20
    },
    {
      "epoch": 0.011473152822395595,
      "grad_norm": 0.17022030208338307,
      "learning_rate": 2.2935779816513765e-05,
      "loss": 1.0672,
      "step": 25
    },
    {
      "epoch": 0.013767783386874713,
      "grad_norm": 0.15571207235273762,
      "learning_rate": 2.7522935779816515e-05,
      "loss": 1.0782,
      "step": 30
    },
    {
      "epoch": 0.016062413951353834,
      "grad_norm": 0.1146172388400534,
      "learning_rate": 3.211009174311927e-05,
      "loss": 1.0178,
      "step": 35
    },
    {
      "epoch": 0.018357044515832952,
      "grad_norm": 0.10569957850527,
      "learning_rate": 3.669724770642202e-05,
      "loss": 1.0347,
      "step": 40
    },
    {
      "epoch": 0.02065167508031207,
      "grad_norm": 0.10758552876940583,
      "learning_rate": 4.1284403669724776e-05,
      "loss": 1.0464,
      "step": 45
    },
    {
      "epoch": 0.02294630564479119,
      "grad_norm": 0.12022433488838774,
      "learning_rate": 4.587155963302753e-05,
      "loss": 0.9938,
      "step": 50
    },
    {
      "epoch": 0.025240936209270308,
      "grad_norm": 0.12413949810603703,
      "learning_rate": 5.0458715596330276e-05,
      "loss": 1.0116,
      "step": 55
    },
    {
      "epoch": 0.027535566773749427,
      "grad_norm": 0.11089749594093384,
      "learning_rate": 5.504587155963303e-05,
      "loss": 1.0099,
      "step": 60
    },
    {
      "epoch": 0.029830197338228545,
      "grad_norm": 0.11116787947934133,
      "learning_rate": 5.9633027522935784e-05,
      "loss": 1.0123,
      "step": 65
    },
    {
      "epoch": 0.03212482790270767,
      "grad_norm": 0.1045508356848914,
      "learning_rate": 6.422018348623854e-05,
      "loss": 0.9922,
      "step": 70
    },
    {
      "epoch": 0.03441945846718678,
      "grad_norm": 0.11419989748358071,
      "learning_rate": 6.880733944954129e-05,
      "loss": 1.0221,
      "step": 75
    },
    {
      "epoch": 0.036714089031665904,
      "grad_norm": 0.11920093762065691,
      "learning_rate": 7.339449541284404e-05,
      "loss": 0.9903,
      "step": 80
    },
    {
      "epoch": 0.03900871959614502,
      "grad_norm": 0.1205539032721293,
      "learning_rate": 7.79816513761468e-05,
      "loss": 0.9892,
      "step": 85
    },
    {
      "epoch": 0.04130335016062414,
      "grad_norm": 0.1153297215401342,
      "learning_rate": 8.256880733944955e-05,
      "loss": 0.9801,
      "step": 90
    },
    {
      "epoch": 0.04359798072510326,
      "grad_norm": 0.11168014813048414,
      "learning_rate": 8.715596330275229e-05,
      "loss": 0.9927,
      "step": 95
    },
    {
      "epoch": 0.04589261128958238,
      "grad_norm": 0.11609074160199676,
      "learning_rate": 9.174311926605506e-05,
      "loss": 0.9794,
      "step": 100
    },
    {
      "epoch": 0.048187241854061494,
      "grad_norm": 0.11950825569103571,
      "learning_rate": 9.63302752293578e-05,
      "loss": 0.9818,
      "step": 105
    },
    {
      "epoch": 0.050481872418540616,
      "grad_norm": 0.11880725497478824,
      "learning_rate": 0.00010091743119266055,
      "loss": 0.9986,
      "step": 110
    },
    {
      "epoch": 0.05277650298301973,
      "grad_norm": 0.11245311133827594,
      "learning_rate": 0.00010550458715596329,
      "loss": 1.0082,
      "step": 115
    },
    {
      "epoch": 0.05507113354749885,
      "grad_norm": 0.11463279657110968,
      "learning_rate": 0.00011009174311926606,
      "loss": 0.9701,
      "step": 120
    },
    {
      "epoch": 0.05736576411197797,
      "grad_norm": 0.1147080076997952,
      "learning_rate": 0.00011467889908256881,
      "loss": 0.9561,
      "step": 125
    },
    {
      "epoch": 0.05966039467645709,
      "grad_norm": 0.10731908295172292,
      "learning_rate": 0.00011926605504587157,
      "loss": 0.9806,
      "step": 130
    },
    {
      "epoch": 0.06195502524093621,
      "grad_norm": 0.11772654963107822,
      "learning_rate": 0.00012385321100917432,
      "loss": 0.9781,
      "step": 135
    },
    {
      "epoch": 0.06424965580541533,
      "grad_norm": 0.10616965717110965,
      "learning_rate": 0.00012844036697247707,
      "loss": 0.9575,
      "step": 140
    },
    {
      "epoch": 0.06654428636989444,
      "grad_norm": 0.11403517689802216,
      "learning_rate": 0.00013302752293577983,
      "loss": 0.9967,
      "step": 145
    },
    {
      "epoch": 0.06883891693437356,
      "grad_norm": 0.12124685445241375,
      "learning_rate": 0.00013761467889908258,
      "loss": 0.962,
      "step": 150
    },
    {
      "epoch": 0.07113354749885269,
      "grad_norm": 0.11004755929686598,
      "learning_rate": 0.0001422018348623853,
      "loss": 0.9458,
      "step": 155
    },
    {
      "epoch": 0.07342817806333181,
      "grad_norm": 0.11198046801850037,
      "learning_rate": 0.0001467889908256881,
      "loss": 0.9853,
      "step": 160
    },
    {
      "epoch": 0.07572280862781092,
      "grad_norm": 0.10851191328572089,
      "learning_rate": 0.00015137614678899084,
      "loss": 0.9599,
      "step": 165
    },
    {
      "epoch": 0.07801743919229004,
      "grad_norm": 0.1027573560837855,
      "learning_rate": 0.0001559633027522936,
      "loss": 0.9705,
      "step": 170
    },
    {
      "epoch": 0.08031206975676916,
      "grad_norm": 0.09904618883874605,
      "learning_rate": 0.00016055045871559632,
      "loss": 0.9607,
      "step": 175
    },
    {
      "epoch": 0.08260670032124828,
      "grad_norm": 0.10960846096179629,
      "learning_rate": 0.0001651376146788991,
      "loss": 1.0047,
      "step": 180
    },
    {
      "epoch": 0.08490133088572739,
      "grad_norm": 0.10544676605226945,
      "learning_rate": 0.00016972477064220186,
      "loss": 0.9894,
      "step": 185
    },
    {
      "epoch": 0.08719596145020651,
      "grad_norm": 0.10229698437978632,
      "learning_rate": 0.00017431192660550458,
      "loss": 0.9764,
      "step": 190
    },
    {
      "epoch": 0.08949059201468564,
      "grad_norm": 0.09568938580281933,
      "learning_rate": 0.00017889908256880734,
      "loss": 0.9618,
      "step": 195
    },
    {
      "epoch": 0.09178522257916476,
      "grad_norm": 0.10024788325345245,
      "learning_rate": 0.00018348623853211012,
      "loss": 0.953,
      "step": 200
    },
    {
      "epoch": 0.09407985314364388,
      "grad_norm": 0.09773114551334808,
      "learning_rate": 0.00018807339449541284,
      "loss": 0.9775,
      "step": 205
    },
    {
      "epoch": 0.09637448370812299,
      "grad_norm": 0.103651995811559,
      "learning_rate": 0.0001926605504587156,
      "loss": 0.9621,
      "step": 210
    },
    {
      "epoch": 0.09866911427260211,
      "grad_norm": 0.10031363164526642,
      "learning_rate": 0.00019724770642201835,
      "loss": 0.9771,
      "step": 215
    },
    {
      "epoch": 0.10096374483708123,
      "grad_norm": 0.09400862481398278,
      "learning_rate": 0.00019999948669655127,
      "loss": 0.9685,
      "step": 220
    },
    {
      "epoch": 0.10325837540156035,
      "grad_norm": 0.09148587705607365,
      "learning_rate": 0.00019999371209327089,
      "loss": 0.9472,
      "step": 225
    },
    {
      "epoch": 0.10555300596603946,
      "grad_norm": 0.1000742259516268,
      "learning_rate": 0.00019998152162914806,
      "loss": 0.9601,
      "step": 230
    },
    {
      "epoch": 0.10784763653051858,
      "grad_norm": 0.099276599123678,
      "learning_rate": 0.00019996291608635525,
      "loss": 0.9597,
      "step": 235
    },
    {
      "epoch": 0.1101422670949977,
      "grad_norm": 0.09100711111755377,
      "learning_rate": 0.00019993789665867314,
      "loss": 0.9427,
      "step": 240
    },
    {
      "epoch": 0.11243689765947683,
      "grad_norm": 0.09143444839325382,
      "learning_rate": 0.00019990646495141444,
      "loss": 0.9521,
      "step": 245
    },
    {
      "epoch": 0.11473152822395594,
      "grad_norm": 0.09469721895906218,
      "learning_rate": 0.0001998686229813205,
      "loss": 0.9642,
      "step": 250
    },
    {
      "epoch": 0.11702615878843506,
      "grad_norm": 0.09530573435158946,
      "learning_rate": 0.00019982437317643217,
      "loss": 0.9475,
      "step": 255
    },
    {
      "epoch": 0.11932078935291418,
      "grad_norm": 0.09537227174416016,
      "learning_rate": 0.0001997737183759338,
      "loss": 0.9596,
      "step": 260
    },
    {
      "epoch": 0.1216154199173933,
      "grad_norm": 0.09276269113129716,
      "learning_rate": 0.00019971666182997137,
      "loss": 0.9264,
      "step": 265
    },
    {
      "epoch": 0.12391005048187242,
      "grad_norm": 0.09602318064837084,
      "learning_rate": 0.00019965320719944366,
      "loss": 0.9388,
      "step": 270
    },
    {
      "epoch": 0.12620468104635155,
      "grad_norm": 0.08888849048667083,
      "learning_rate": 0.00019958335855576738,
      "loss": 0.9631,
      "step": 275
    },
    {
      "epoch": 0.12849931161083067,
      "grad_norm": 0.08994328067689368,
      "learning_rate": 0.00019950712038061617,
      "loss": 0.9481,
      "step": 280
    },
    {
      "epoch": 0.13079394217530976,
      "grad_norm": 0.0897263586822677,
      "learning_rate": 0.00019942449756563279,
      "loss": 0.9566,
      "step": 285
    },
    {
      "epoch": 0.13308857273978889,
      "grad_norm": 0.08889749103510053,
      "learning_rate": 0.00019933549541211552,
      "loss": 0.965,
      "step": 290
    },
    {
      "epoch": 0.135383203304268,
      "grad_norm": 0.09008585580167996,
      "learning_rate": 0.00019924011963067765,
      "loss": 0.9489,
      "step": 295
    },
    {
      "epoch": 0.13767783386874713,
      "grad_norm": 0.08978334358331166,
      "learning_rate": 0.00019913837634088144,
      "loss": 0.9591,
      "step": 300
    },
    {
      "epoch": 0.13997246443322625,
      "grad_norm": 0.08869912275141656,
      "learning_rate": 0.00019903027207084523,
      "loss": 0.9647,
      "step": 305
    },
    {
      "epoch": 0.14226709499770537,
      "grad_norm": 0.08650660717454736,
      "learning_rate": 0.00019891581375682473,
      "loss": 0.9877,
      "step": 310
    },
    {
      "epoch": 0.1445617255621845,
      "grad_norm": 0.08750656685685948,
      "learning_rate": 0.00019879500874276786,
      "loss": 0.9654,
      "step": 315
    },
    {
      "epoch": 0.14685635612666362,
      "grad_norm": 0.08617919321649241,
      "learning_rate": 0.00019866786477984357,
      "loss": 0.965,
      "step": 320
    },
    {
      "epoch": 0.14915098669114274,
      "grad_norm": 0.08483226196137736,
      "learning_rate": 0.0001985343900259446,
      "loss": 0.9677,
      "step": 325
    },
    {
      "epoch": 0.15144561725562183,
      "grad_norm": 0.09712002013315797,
      "learning_rate": 0.0001983945930451639,
      "loss": 0.9845,
      "step": 330
    },
    {
      "epoch": 0.15374024782010096,
      "grad_norm": 0.08834848603530239,
      "learning_rate": 0.0001982484828072452,
      "loss": 0.9551,
      "step": 335
    },
    {
      "epoch": 0.15603487838458008,
      "grad_norm": 0.08706421027052569,
      "learning_rate": 0.00019809606868700756,
      "loss": 0.9725,
      "step": 340
    },
    {
      "epoch": 0.1583295089490592,
      "grad_norm": 0.08948100664780789,
      "learning_rate": 0.00019793736046374373,
      "loss": 0.9537,
      "step": 345
    },
    {
      "epoch": 0.16062413951353832,
      "grad_norm": 0.08854849269535629,
      "learning_rate": 0.0001977723683205928,
      "loss": 0.9826,
      "step": 350
    },
    {
      "epoch": 0.16291877007801744,
      "grad_norm": 0.09035742208671649,
      "learning_rate": 0.00019760110284388667,
      "loss": 0.9685,
      "step": 355
    },
    {
      "epoch": 0.16521340064249657,
      "grad_norm": 0.2078369241555783,
      "learning_rate": 0.00019742357502247103,
      "loss": 0.953,
      "step": 360
    },
    {
      "epoch": 0.1675080312069757,
      "grad_norm": 0.08496387083805744,
      "learning_rate": 0.00019723979624700004,
      "loss": 0.9551,
      "step": 365
    },
    {
      "epoch": 0.16980266177145478,
      "grad_norm": 0.09478873291884332,
      "learning_rate": 0.0001970497783092057,
      "loss": 0.9744,
      "step": 370
    },
    {
      "epoch": 0.1720972923359339,
      "grad_norm": 0.08984634771176177,
      "learning_rate": 0.00019685353340114103,
      "loss": 0.9564,
      "step": 375
    },
    {
      "epoch": 0.17439192290041303,
      "grad_norm": 0.09065621901432767,
      "learning_rate": 0.00019665107411439805,
      "loss": 0.9807,
      "step": 380
    },
    {
      "epoch": 0.17668655346489215,
      "grad_norm": 0.08667656345676594,
      "learning_rate": 0.00019644241343929966,
      "loss": 0.9426,
      "step": 385
    },
    {
      "epoch": 0.17898118402937127,
      "grad_norm": 0.09114781451954795,
      "learning_rate": 0.00019622756476406626,
      "loss": 0.9551,
      "step": 390
    },
    {
      "epoch": 0.1812758145938504,
      "grad_norm": 0.09037110088641298,
      "learning_rate": 0.00019600654187395663,
      "loss": 0.9345,
      "step": 395
    },
    {
      "epoch": 0.18357044515832951,
      "grad_norm": 0.09611180100762262,
      "learning_rate": 0.00019577935895038361,
      "loss": 0.9565,
      "step": 400
    },
    {
      "epoch": 0.18586507572280864,
      "grad_norm": 0.09346105169502975,
      "learning_rate": 0.00019554603057000397,
      "loss": 0.9607,
      "step": 405
    },
    {
      "epoch": 0.18815970628728776,
      "grad_norm": 0.08669880613701339,
      "learning_rate": 0.0001953065717037832,
      "loss": 0.9434,
      "step": 410
    },
    {
      "epoch": 0.19045433685176685,
      "grad_norm": 0.09074513235696971,
      "learning_rate": 0.00019506099771603513,
      "loss": 0.9539,
      "step": 415
    },
    {
      "epoch": 0.19274896741624598,
      "grad_norm": 0.08770031036399964,
      "learning_rate": 0.00019480932436343582,
      "loss": 0.9466,
      "step": 420
    },
    {
      "epoch": 0.1950435979807251,
      "grad_norm": 0.08915513286930843,
      "learning_rate": 0.00019455156779401265,
      "loss": 0.9628,
      "step": 425
    },
    {
      "epoch": 0.19733822854520422,
      "grad_norm": 0.08790684691806365,
      "learning_rate": 0.00019428774454610843,
      "loss": 0.9598,
      "step": 430
    },
    {
      "epoch": 0.19963285910968334,
      "grad_norm": 0.08895484944044536,
      "learning_rate": 0.00019401787154731993,
      "loss": 0.9553,
      "step": 435
    },
    {
      "epoch": 0.20192748967416246,
      "grad_norm": 0.08798273269859104,
      "learning_rate": 0.0001937419661134121,
      "loss": 0.9796,
      "step": 440
    },
    {
      "epoch": 0.20422212023864159,
      "grad_norm": 0.08777210421997016,
      "learning_rate": 0.00019346004594720669,
      "loss": 0.9495,
      "step": 445
    },
    {
      "epoch": 0.2065167508031207,
      "grad_norm": 0.08564705845641755,
      "learning_rate": 0.0001931721291374467,
      "loss": 0.9501,
      "step": 450
    },
    {
      "epoch": 0.20881138136759983,
      "grad_norm": 0.09474670359706922,
      "learning_rate": 0.00019287823415763553,
      "loss": 0.9559,
      "step": 455
    },
    {
      "epoch": 0.21110601193207892,
      "grad_norm": 0.08777147268242179,
      "learning_rate": 0.00019257837986485187,
      "loss": 0.9828,
      "step": 460
    },
    {
      "epoch": 0.21340064249655805,
      "grad_norm": 0.09274202152499614,
      "learning_rate": 0.0001922725854985396,
      "loss": 0.943,
      "step": 465
    },
    {
      "epoch": 0.21569527306103717,
      "grad_norm": 0.08906570334946029,
      "learning_rate": 0.0001919608706792735,
      "loss": 0.963,
      "step": 470
    },
    {
      "epoch": 0.2179899036255163,
      "grad_norm": 0.09431879186087927,
      "learning_rate": 0.00019164325540750016,
      "loss": 0.976,
      "step": 475
    },
    {
      "epoch": 0.2202845341899954,
      "grad_norm": 0.0880885330404675,
      "learning_rate": 0.00019131976006225488,
      "loss": 0.934,
      "step": 480
    },
    {
      "epoch": 0.22257916475447453,
      "grad_norm": 0.09001487242711557,
      "learning_rate": 0.00019099040539985394,
      "loss": 0.9608,
      "step": 485
    },
    {
      "epoch": 0.22487379531895366,
      "grad_norm": 0.08632949578437538,
      "learning_rate": 0.00019065521255256298,
      "loss": 0.9489,
      "step": 490
    },
    {
      "epoch": 0.22716842588343278,
      "grad_norm": 0.08860042188251825,
      "learning_rate": 0.00019031420302724096,
      "loss": 0.9539,
      "step": 495
    },
    {
      "epoch": 0.22946305644791187,
      "grad_norm": 0.09204792889480558,
      "learning_rate": 0.00018996739870396026,
      "loss": 0.9455,
      "step": 500
    },
    {
      "epoch": 0.231757687012391,
      "grad_norm": 0.0907813137121107,
      "learning_rate": 0.00018961482183460282,
      "loss": 0.9455,
      "step": 505
    },
    {
      "epoch": 0.23405231757687012,
      "grad_norm": 0.09123445749023676,
      "learning_rate": 0.00018925649504143244,
      "loss": 0.9386,
      "step": 510
    },
    {
      "epoch": 0.23634694814134924,
      "grad_norm": 0.08933477090645768,
      "learning_rate": 0.0001888924413156432,
      "loss": 0.9566,
      "step": 515
    },
    {
      "epoch": 0.23864157870582836,
      "grad_norm": 0.08928138581489246,
      "learning_rate": 0.0001885226840158843,
      "loss": 0.9617,
      "step": 520
    },
    {
      "epoch": 0.24093620927030748,
      "grad_norm": 0.1274731933313709,
      "learning_rate": 0.00018814724686676133,
      "loss": 0.9596,
      "step": 525
    },
    {
      "epoch": 0.2432308398347866,
      "grad_norm": 0.0905076097730417,
      "learning_rate": 0.00018776615395731398,
      "loss": 0.9438,
      "step": 530
    },
    {
      "epoch": 0.24552547039926573,
      "grad_norm": 0.08684198828453624,
      "learning_rate": 0.0001873794297394706,
      "loss": 0.9601,
      "step": 535
    },
    {
      "epoch": 0.24782010096374485,
      "grad_norm": 0.0888276936189942,
      "learning_rate": 0.00018698709902647902,
      "loss": 0.9276,
      "step": 540
    },
    {
      "epoch": 0.25011473152822394,
      "grad_norm": 0.09323508497961805,
      "learning_rate": 0.00018658918699131468,
      "loss": 0.9767,
      "step": 545
    },
    {
      "epoch": 0.2524093620927031,
      "grad_norm": 0.0881344870799144,
      "learning_rate": 0.00018618571916506545,
      "loss": 0.9604,
      "step": 550
    },
    {
      "epoch": 0.2547039926571822,
      "grad_norm": 0.08872071593430497,
      "learning_rate": 0.00018577672143529336,
      "loss": 0.9587,
      "step": 555
    },
    {
      "epoch": 0.25699862322166134,
      "grad_norm": 0.08503719828368154,
      "learning_rate": 0.00018536222004437368,
      "loss": 0.9589,
      "step": 560
    },
    {
      "epoch": 0.25929325378614043,
      "grad_norm": 0.08844948896045982,
      "learning_rate": 0.0001849422415878112,
      "loss": 0.9611,
      "step": 565
    },
    {
      "epoch": 0.2615878843506195,
      "grad_norm": 0.08833201695867715,
      "learning_rate": 0.00018451681301253362,
      "loss": 0.9619,
      "step": 570
    },
    {
      "epoch": 0.2638825149150987,
      "grad_norm": 0.08789348466222431,
      "learning_rate": 0.00018408596161516267,
      "loss": 0.9732,
      "step": 575
    },
    {
      "epoch": 0.26617714547957777,
      "grad_norm": 0.09007398961819005,
      "learning_rate": 0.00018364971504026273,
      "loss": 0.9546,
      "step": 580
    },
    {
      "epoch": 0.2684717760440569,
      "grad_norm": 0.09389996335115901,
      "learning_rate": 0.00018320810127856705,
      "loss": 0.9394,
      "step": 585
    },
    {
      "epoch": 0.270766406608536,
      "grad_norm": 0.08615315931672284,
      "learning_rate": 0.00018276114866518168,
      "loss": 0.956,
      "step": 590
    },
    {
      "epoch": 0.27306103717301516,
      "grad_norm": 0.08933240638639854,
      "learning_rate": 0.00018230888587776755,
      "loss": 0.9688,
      "step": 595
    },
    {
      "epoch": 0.27535566773749426,
      "grad_norm": 0.09191231486736685,
      "learning_rate": 0.00018185134193470043,
      "loss": 0.9504,
      "step": 600
    },
    {
      "epoch": 0.2776502983019734,
      "grad_norm": 0.08712371583759838,
      "learning_rate": 0.00018138854619320893,
      "loss": 0.9682,
      "step": 605
    },
    {
      "epoch": 0.2799449288664525,
      "grad_norm": 0.08901746734004592,
      "learning_rate": 0.00018092052834749094,
      "loss": 0.9402,
      "step": 610
    },
    {
      "epoch": 0.2822395594309316,
      "grad_norm": 0.08840589666910355,
      "learning_rate": 0.0001804473184268084,
      "loss": 0.9768,
      "step": 615
    },
    {
      "epoch": 0.28453418999541075,
      "grad_norm": 0.08481006585653182,
      "learning_rate": 0.0001799689467935604,
      "loss": 0.9567,
      "step": 620
    },
    {
      "epoch": 0.28682882055988984,
      "grad_norm": 0.08497095117377013,
      "learning_rate": 0.00017948544414133534,
      "loss": 0.9502,
      "step": 625
    },
    {
      "epoch": 0.289123451124369,
      "grad_norm": 0.08990180476167892,
      "learning_rate": 0.00017899684149294117,
      "loss": 0.9736,
      "step": 630
    },
    {
      "epoch": 0.2914180816888481,
      "grad_norm": 0.08740257994079272,
      "learning_rate": 0.00017850317019841514,
      "loss": 0.9452,
      "step": 635
    },
    {
      "epoch": 0.29371271225332723,
      "grad_norm": 0.09168725176770938,
      "learning_rate": 0.00017800446193301227,
      "loss": 0.9584,
      "step": 640
    },
    {
      "epoch": 0.29600734281780633,
      "grad_norm": 0.0854935785988436,
      "learning_rate": 0.00017750074869517284,
      "loss": 0.9498,
      "step": 645
    },
    {
      "epoch": 0.2983019733822855,
      "grad_norm": 0.08713641806146565,
      "learning_rate": 0.00017699206280446953,
      "loss": 0.9781,
      "step": 650
    },
    {
      "epoch": 0.3005966039467646,
      "grad_norm": 0.08670609625106108,
      "learning_rate": 0.00017647843689953352,
      "loss": 0.9436,
      "step": 655
    },
    {
      "epoch": 0.30289123451124367,
      "grad_norm": 0.08793355472500575,
      "learning_rate": 0.00017595990393596026,
      "loss": 0.9526,
      "step": 660
    },
    {
      "epoch": 0.3051858650757228,
      "grad_norm": 0.09193899666963108,
      "learning_rate": 0.0001754364971841952,
      "loss": 0.9558,
      "step": 665
    },
    {
      "epoch": 0.3074804956402019,
      "grad_norm": 0.08520469387571077,
      "learning_rate": 0.0001749082502273988,
      "loss": 0.9384,
      "step": 670
    },
    {
      "epoch": 0.30977512620468106,
      "grad_norm": 0.08632463244243371,
      "learning_rate": 0.00017437519695929194,
      "loss": 0.9349,
      "step": 675
    },
    {
      "epoch": 0.31206975676916016,
      "grad_norm": 0.08898391879883893,
      "learning_rate": 0.0001738373715819811,
      "loss": 0.9851,
      "step": 680
    },
    {
      "epoch": 0.3143643873336393,
      "grad_norm": 0.08743602323721172,
      "learning_rate": 0.00017329480860376391,
      "loss": 0.9684,
      "step": 685
    },
    {
      "epoch": 0.3166590178981184,
      "grad_norm": 0.0884272891415175,
      "learning_rate": 0.00017274754283691504,
      "loss": 0.9513,
      "step": 690
    },
    {
      "epoch": 0.3189536484625975,
      "grad_norm": 0.0829396476234397,
      "learning_rate": 0.00017219560939545246,
      "loss": 0.9227,
      "step": 695
    },
    {
      "epoch": 0.32124827902707664,
      "grad_norm": 0.08812333013019032,
      "learning_rate": 0.00017163904369288445,
      "loss": 0.9266,
      "step": 700
    },
    {
      "epoch": 0.32354290959155574,
      "grad_norm": 0.08790955029633435,
      "learning_rate": 0.00017107788143993743,
      "loss": 0.9645,
      "step": 705
    },
    {
      "epoch": 0.3258375401560349,
      "grad_norm": 0.08662253079337641,
      "learning_rate": 0.00017051215864226469,
      "loss": 0.9636,
      "step": 710
    },
    {
      "epoch": 0.328132170720514,
      "grad_norm": 0.08662254032383067,
      "learning_rate": 0.0001699419115981361,
      "loss": 0.9461,
      "step": 715
    },
    {
      "epoch": 0.33042680128499313,
      "grad_norm": 0.08795758448530268,
      "learning_rate": 0.00016936717689610903,
      "loss": 0.933,
      "step": 720
    },
    {
      "epoch": 0.3327214318494722,
      "grad_norm": 0.08428350308511785,
      "learning_rate": 0.00016878799141268106,
      "loss": 0.9368,
      "step": 725
    },
    {
      "epoch": 0.3350160624139514,
      "grad_norm": 0.08616657609346429,
      "learning_rate": 0.00016820439230992343,
      "loss": 0.9621,
      "step": 730
    },
    {
      "epoch": 0.33731069297843047,
      "grad_norm": 0.08578062404208871,
      "learning_rate": 0.00016761641703309704,
      "loss": 0.9503,
      "step": 735
    },
    {
      "epoch": 0.33960532354290957,
      "grad_norm": 0.08875908176879958,
      "learning_rate": 0.00016702410330824962,
      "loss": 0.9377,
      "step": 740
    },
    {
      "epoch": 0.3418999541073887,
      "grad_norm": 0.08739702470553666,
      "learning_rate": 0.00016642748913979513,
      "loss": 0.9604,
      "step": 745
    },
    {
      "epoch": 0.3441945846718678,
      "grad_norm": 0.08662264283634206,
      "learning_rate": 0.00016582661280807552,
      "loss": 0.9765,
      "step": 750
    },
    {
      "epoch": 0.34648921523634696,
      "grad_norm": 0.08704277951624015,
      "learning_rate": 0.0001652215128669042,
      "loss": 0.962,
      "step": 755
    },
    {
      "epoch": 0.34878384580082605,
      "grad_norm": 0.08627254433727607,
      "learning_rate": 0.00016461222814109268,
      "loss": 0.9503,
      "step": 760
    },
    {
      "epoch": 0.3510784763653052,
      "grad_norm": 0.08968065039630777,
      "learning_rate": 0.00016399879772395915,
      "loss": 0.9694,
      "step": 765
    },
    {
      "epoch": 0.3533731069297843,
      "grad_norm": 0.08992120522770167,
      "learning_rate": 0.00016338126097482057,
      "loss": 0.9425,
      "step": 770
    },
    {
      "epoch": 0.35566773749426345,
      "grad_norm": 0.08636178728280854,
      "learning_rate": 0.0001627596575164668,
      "loss": 0.9486,
      "step": 775
    },
    {
      "epoch": 0.35796236805874254,
      "grad_norm": 0.08955961975585672,
      "learning_rate": 0.0001621340272326185,
      "loss": 0.9545,
      "step": 780
    },
    {
      "epoch": 0.36025699862322164,
      "grad_norm": 0.09696892159638587,
      "learning_rate": 0.00016150441026536823,
      "loss": 0.9541,
      "step": 785
    },
    {
      "epoch": 0.3625516291877008,
      "grad_norm": 0.0931806547196916,
      "learning_rate": 0.00016087084701260466,
      "loss": 0.9338,
      "step": 790
    },
    {
      "epoch": 0.3648462597521799,
      "grad_norm": 0.08965151017485885,
      "learning_rate": 0.00016023337812542048,
      "loss": 0.9337,
      "step": 795
    },
    {
      "epoch": 0.36714089031665903,
      "grad_norm": 0.08623125405498211,
      "learning_rate": 0.00015959204450550428,
      "loss": 0.9602,
      "step": 800
    },
    {
      "epoch": 0.3694355208811381,
      "grad_norm": 0.08928739936481382,
      "learning_rate": 0.00015894688730251614,
      "loss": 0.9453,
      "step": 805
    },
    {
      "epoch": 0.3717301514456173,
      "grad_norm": 0.08612949012161293,
      "learning_rate": 0.0001582979479114472,
      "loss": 0.9454,
      "step": 810
    },
    {
      "epoch": 0.37402478201009637,
      "grad_norm": 0.08768044431530787,
      "learning_rate": 0.000157645267969964,
      "loss": 0.9429,
      "step": 815
    },
    {
      "epoch": 0.3763194125745755,
      "grad_norm": 0.08907644899471728,
      "learning_rate": 0.0001569888893557365,
      "loss": 0.9459,
      "step": 820
    },
    {
      "epoch": 0.3786140431390546,
      "grad_norm": 0.08749252420425255,
      "learning_rate": 0.00015632885418375136,
      "loss": 0.9585,
      "step": 825
    },
    {
      "epoch": 0.3809086737035337,
      "grad_norm": 0.08988996586262392,
      "learning_rate": 0.00015566520480360957,
      "loss": 0.9663,
      "step": 830
    },
    {
      "epoch": 0.38320330426801286,
      "grad_norm": 0.08784510387671604,
      "learning_rate": 0.0001549979837968094,
      "loss": 0.9562,
      "step": 835
    },
    {
      "epoch": 0.38549793483249195,
      "grad_norm": 0.0863284429643144,
      "learning_rate": 0.00015432723397401405,
      "loss": 0.9507,
      "step": 840
    },
    {
      "epoch": 0.3877925653969711,
      "grad_norm": 0.08716881362564995,
      "learning_rate": 0.00015365299837230483,
      "loss": 0.9436,
      "step": 845
    },
    {
      "epoch": 0.3900871959614502,
      "grad_norm": 0.09125631926968461,
      "learning_rate": 0.00015297532025241992,
      "loss": 0.9326,
      "step": 850
    },
    {
      "epoch": 0.39238182652592934,
      "grad_norm": 0.08538626816667327,
      "learning_rate": 0.00015229424309597852,
      "loss": 0.9458,
      "step": 855
    },
    {
      "epoch": 0.39467645709040844,
      "grad_norm": 0.09072041145807094,
      "learning_rate": 0.00015160981060269107,
      "loss": 0.9893,
      "step": 860
    },
    {
      "epoch": 0.3969710876548876,
      "grad_norm": 0.0856815427930852,
      "learning_rate": 0.00015092206668755517,
      "loss": 0.9249,
      "step": 865
    },
    {
      "epoch": 0.3992657182193667,
      "grad_norm": 0.09070757596561506,
      "learning_rate": 0.00015023105547803806,
      "loss": 0.9618,
      "step": 870
    },
    {
      "epoch": 0.4015603487838458,
      "grad_norm": 0.08729844078361124,
      "learning_rate": 0.00014953682131124528,
      "loss": 0.9617,
      "step": 875
    },
    {
      "epoch": 0.4038549793483249,
      "grad_norm": 0.08727964740364995,
      "learning_rate": 0.0001488394087310757,
      "loss": 0.9347,
      "step": 880
    },
    {
      "epoch": 0.406149609912804,
      "grad_norm": 0.08472455072190438,
      "learning_rate": 0.00014813886248536375,
      "loss": 0.9407,
      "step": 885
    },
    {
      "epoch": 0.40844424047728317,
      "grad_norm": 0.08766621771656237,
      "learning_rate": 0.00014743522752300793,
      "loss": 0.9483,
      "step": 890
    },
    {
      "epoch": 0.41073887104176227,
      "grad_norm": 0.09000327025386383,
      "learning_rate": 0.00014672854899108718,
      "loss": 0.951,
      "step": 895
    },
    {
      "epoch": 0.4130335016062414,
      "grad_norm": 0.08539102749160071,
      "learning_rate": 0.00014601887223196372,
      "loss": 0.9439,
      "step": 900
    },
    {
      "epoch": 0.4153281321707205,
      "grad_norm": 0.08868715226055852,
      "learning_rate": 0.00014530624278037406,
      "loss": 0.9511,
      "step": 905
    },
    {
      "epoch": 0.41762276273519966,
      "grad_norm": 0.08710262112773155,
      "learning_rate": 0.0001445907063605072,
      "loss": 0.941,
      "step": 910
    },
    {
      "epoch": 0.41991739329967875,
      "grad_norm": 0.09153632726865457,
      "learning_rate": 0.00014387230888307097,
      "loss": 0.935,
      "step": 915
    },
    {
      "epoch": 0.42221202386415785,
      "grad_norm": 0.08807434203845403,
      "learning_rate": 0.0001431510964423462,
      "loss": 0.9618,
      "step": 920
    },
    {
      "epoch": 0.424506654428637,
      "grad_norm": 0.08735092148978792,
      "learning_rate": 0.0001424271153132291,
      "loss": 0.9697,
      "step": 925
    },
    {
      "epoch": 0.4268012849931161,
      "grad_norm": 0.08909879626027925,
      "learning_rate": 0.00014170041194826248,
      "loss": 0.958,
      "step": 930
    },
    {
      "epoch": 0.42909591555759524,
      "grad_norm": 0.08915959240120495,
      "learning_rate": 0.0001409710329746547,
      "loss": 0.9423,
      "step": 935
    },
    {
      "epoch": 0.43139054612207434,
      "grad_norm": 0.08869687131713624,
      "learning_rate": 0.0001402390251912885,
      "loss": 0.9633,
      "step": 940
    },
    {
      "epoch": 0.4336851766865535,
      "grad_norm": 0.0881776336437586,
      "learning_rate": 0.00013950443556571778,
      "loss": 0.9548,
      "step": 945
    },
    {
      "epoch": 0.4359798072510326,
      "grad_norm": 0.09157715163747275,
      "learning_rate": 0.0001387673112311545,
      "loss": 0.9284,
      "step": 950
    },
    {
      "epoch": 0.4382744378155117,
      "grad_norm": 0.08636314894179382,
      "learning_rate": 0.00013802769948344405,
      "loss": 0.975,
      "step": 955
    },
    {
      "epoch": 0.4405690683799908,
      "grad_norm": 0.08623157739774547,
      "learning_rate": 0.00013728564777803088,
      "loss": 0.9482,
      "step": 960
    },
    {
      "epoch": 0.4428636989444699,
      "grad_norm": 0.0922928322489806,
      "learning_rate": 0.0001365412037269136,
      "loss": 0.9359,
      "step": 965
    },
    {
      "epoch": 0.44515832950894907,
      "grad_norm": 0.08868079382090867,
      "learning_rate": 0.00013579441509559004,
      "loss": 0.9612,
      "step": 970
    },
    {
      "epoch": 0.44745296007342816,
      "grad_norm": 0.08869466662618647,
      "learning_rate": 0.0001350453297999925,
      "loss": 0.9547,
      "step": 975
    },
    {
      "epoch": 0.4497475906379073,
      "grad_norm": 0.08994785511659796,
      "learning_rate": 0.00013429399590341324,
      "loss": 0.9294,
      "step": 980
    },
    {
      "epoch": 0.4520422212023864,
      "grad_norm": 0.08806701296884258,
      "learning_rate": 0.00013354046161342085,
      "loss": 0.9401,
      "step": 985
    },
    {
      "epoch": 0.45433685176686556,
      "grad_norm": 0.08441148992282059,
      "learning_rate": 0.0001327847752787669,
      "loss": 0.942,
      "step": 990
    },
    {
      "epoch": 0.45663148233134465,
      "grad_norm": 0.08891059913792923,
      "learning_rate": 0.00013202698538628376,
      "loss": 0.9729,
      "step": 995
    },
    {
      "epoch": 0.45892611289582375,
      "grad_norm": 0.11772466765874874,
      "learning_rate": 0.00013126714055777377,
      "loss": 0.947,
      "step": 1000
    },
    {
      "epoch": 0.4612207434603029,
      "grad_norm": 0.08689924689552052,
      "learning_rate": 0.0001305052895468893,
      "loss": 0.9459,
      "step": 1005
    },
    {
      "epoch": 0.463515374024782,
      "grad_norm": 0.08769974762625138,
      "learning_rate": 0.00012974148123600477,
      "loss": 0.9256,
      "step": 1010
    },
    {
      "epoch": 0.46581000458926114,
      "grad_norm": 0.08612855828244712,
      "learning_rate": 0.00012897576463307997,
      "loss": 0.9506,
      "step": 1015
    },
    {
      "epoch": 0.46810463515374023,
      "grad_norm": 0.08750477475758126,
      "learning_rate": 0.00012820818886851598,
      "loss": 0.9601,
      "step": 1020
    },
    {
      "epoch": 0.4703992657182194,
      "grad_norm": 0.08517063132537664,
      "learning_rate": 0.00012743880319200242,
      "loss": 0.9678,
      "step": 1025
    },
    {
      "epoch": 0.4726938962826985,
      "grad_norm": 0.08974912157026026,
      "learning_rate": 0.00012666765696935773,
      "loss": 0.9448,
      "step": 1030
    },
    {
      "epoch": 0.47498852684717763,
      "grad_norm": 0.08899338914543631,
      "learning_rate": 0.0001258947996793616,
      "loss": 0.9585,
      "step": 1035
    },
    {
      "epoch": 0.4772831574116567,
      "grad_norm": 0.08848650886885345,
      "learning_rate": 0.00012512028091058044,
      "loss": 0.9526,
      "step": 1040
    },
    {
      "epoch": 0.4795777879761358,
      "grad_norm": 0.08563509332051665,
      "learning_rate": 0.00012434415035818535,
      "loss": 0.941,
      "step": 1045
    },
    {
      "epoch": 0.48187241854061497,
      "grad_norm": 0.08828671373838473,
      "learning_rate": 0.00012356645782076383,
      "loss": 0.9486,
      "step": 1050
    },
    {
      "epoch": 0.48416704910509406,
      "grad_norm": 0.08751318444476795,
      "learning_rate": 0.00012278725319712447,
      "loss": 0.9617,
      "step": 1055
    },
    {
      "epoch": 0.4864616796695732,
      "grad_norm": 0.08565511827085842,
      "learning_rate": 0.00012200658648309531,
      "loss": 0.9741,
      "step": 1060
    },
    {
      "epoch": 0.4887563102340523,
      "grad_norm": 0.085745763510799,
      "learning_rate": 0.00012122450776831594,
      "loss": 0.9353,
      "step": 1065
    },
    {
      "epoch": 0.49105094079853145,
      "grad_norm": 0.08682740835362232,
      "learning_rate": 0.00012044106723302364,
      "loss": 0.9276,
      "step": 1070
    },
    {
      "epoch": 0.49334557136301055,
      "grad_norm": 0.08892634949977557,
      "learning_rate": 0.00011965631514483375,
      "loss": 0.9241,
      "step": 1075
    },
    {
      "epoch": 0.4956402019274897,
      "grad_norm": 0.08726407422876443,
      "learning_rate": 0.00011887030185551426,
      "loss": 0.9563,
      "step": 1080
    },
    {
      "epoch": 0.4979348324919688,
      "grad_norm": 0.08689400522409962,
      "learning_rate": 0.00011808307779775518,
      "loss": 0.9622,
      "step": 1085
    },
    {
      "epoch": 0.5002294630564479,
      "grad_norm": 0.08645950072651667,
      "learning_rate": 0.00011729469348193262,
      "loss": 0.9346,
      "step": 1090
    },
    {
      "epoch": 0.502524093620927,
      "grad_norm": 0.0869325513465856,
      "learning_rate": 0.00011650519949286795,
      "loss": 0.934,
      "step": 1095
    },
    {
      "epoch": 0.5048187241854062,
      "grad_norm": 0.08804258211646028,
      "learning_rate": 0.00011571464648658201,
      "loss": 0.9204,
      "step": 1100
    },
    {
      "epoch": 0.5071133547498853,
      "grad_norm": 0.08652733765274413,
      "learning_rate": 0.00011492308518704506,
      "loss": 0.9355,
      "step": 1105
    },
    {
      "epoch": 0.5094079853143644,
      "grad_norm": 0.08731670771279552,
      "learning_rate": 0.00011413056638292215,
      "loss": 0.9376,
      "step": 1110
    },
    {
      "epoch": 0.5117026158788435,
      "grad_norm": 0.0876288621678255,
      "learning_rate": 0.00011333714092431423,
      "loss": 0.9434,
      "step": 1115
    },
    {
      "epoch": 0.5139972464433227,
      "grad_norm": 0.08842497197396523,
      "learning_rate": 0.00011254285971949574,
      "loss": 0.9466,
      "step": 1120
    },
    {
      "epoch": 0.5162918770078018,
      "grad_norm": 0.08833124212553112,
      "learning_rate": 0.00011174777373164798,
      "loss": 0.9424,
      "step": 1125
    },
    {
      "epoch": 0.5185865075722809,
      "grad_norm": 0.08843181606902133,
      "learning_rate": 0.0001109519339755893,
      "loss": 0.9532,
      "step": 1130
    },
    {
      "epoch": 0.52088113813676,
      "grad_norm": 0.0913721070189569,
      "learning_rate": 0.00011015539151450172,
      "loss": 0.9568,
      "step": 1135
    },
    {
      "epoch": 0.523175768701239,
      "grad_norm": 0.090062313572945,
      "learning_rate": 0.00010935819745665477,
      "loss": 0.9365,
      "step": 1140
    },
    {
      "epoch": 0.5254703992657183,
      "grad_norm": 0.08718333029104089,
      "learning_rate": 0.00010856040295212612,
      "loss": 0.9609,
      "step": 1145
    },
    {
      "epoch": 0.5277650298301974,
      "grad_norm": 0.08762141136978818,
      "learning_rate": 0.00010776205918951969,
      "loss": 0.947,
      "step": 1150
    },
    {
      "epoch": 0.5300596603946764,
      "grad_norm": 0.09288381777131122,
      "learning_rate": 0.00010696321739268121,
      "loss": 0.95,
      "step": 1155
    },
    {
      "epoch": 0.5323542909591555,
      "grad_norm": 0.08690475691948375,
      "learning_rate": 0.00010616392881741166,
      "loss": 0.9259,
      "step": 1160
    },
    {
      "epoch": 0.5346489215236347,
      "grad_norm": 0.08761053817275535,
      "learning_rate": 0.00010536424474817847,
      "loss": 0.9257,
      "step": 1165
    },
    {
      "epoch": 0.5369435520881138,
      "grad_norm": 0.08580794325150348,
      "learning_rate": 0.00010456421649482502,
      "loss": 0.9491,
      "step": 1170
    },
    {
      "epoch": 0.5392381826525929,
      "grad_norm": 0.08875803785646835,
      "learning_rate": 0.0001037638953892784,
      "loss": 0.9611,
      "step": 1175
    },
    {
      "epoch": 0.541532813217072,
      "grad_norm": 0.08822711832526892,
      "learning_rate": 0.00010296333278225599,
      "loss": 0.9466,
      "step": 1180
    },
    {
      "epoch": 0.5438274437815511,
      "grad_norm": 0.08973140268678324,
      "learning_rate": 0.00010216258003997043,
      "loss": 0.9707,
      "step": 1185
    },
    {
      "epoch": 0.5461220743460303,
      "grad_norm": 0.08753068797299891,
      "learning_rate": 0.00010136168854083402,
      "loss": 0.9042,
      "step": 1190
    },
    {
      "epoch": 0.5484167049105094,
      "grad_norm": 0.08919024988233984,
      "learning_rate": 0.00010056070967216198,
      "loss": 0.9366,
      "step": 1195
    },
    {
      "epoch": 0.5507113354749885,
      "grad_norm": 0.088157407581,
      "learning_rate": 9.975969482687547e-05,
      "loss": 0.9258,
      "step": 1200
    },
    {
      "epoch": 0.5530059660394676,
      "grad_norm": 0.08612555956203556,
      "learning_rate": 9.8958695400204e-05,
      "loss": 0.9428,
      "step": 1205
    },
    {
      "epoch": 0.5553005966039468,
      "grad_norm": 0.08863199154938804,
      "learning_rate": 9.815776278638771e-05,
      "loss": 0.9659,
      "step": 1210
    },
    {
      "epoch": 0.5575952271684259,
      "grad_norm": 0.0877396299101558,
      "learning_rate": 9.735694837537993e-05,
      "loss": 0.9146,
      "step": 1215
    },
    {
      "epoch": 0.559889857732905,
      "grad_norm": 0.10016922577996923,
      "learning_rate": 9.655630354954974e-05,
      "loss": 0.9344,
      "step": 1220
    },
    {
      "epoch": 0.5621844882973841,
      "grad_norm": 0.08776498802340123,
      "learning_rate": 9.57558796803852e-05,
      "loss": 0.9508,
      "step": 1225
    },
    {
      "epoch": 0.5644791188618632,
      "grad_norm": 0.08576773127748236,
      "learning_rate": 9.495572812519718e-05,
      "loss": 0.9367,
      "step": 1230
    },
    {
      "epoch": 0.5667737494263424,
      "grad_norm": 0.090860993764857,
      "learning_rate": 9.415590022382419e-05,
      "loss": 0.9482,
      "step": 1235
    },
    {
      "epoch": 0.5690683799908215,
      "grad_norm": 0.08739072969654303,
      "learning_rate": 9.33564472953383e-05,
      "loss": 0.9296,
      "step": 1240
    },
    {
      "epoch": 0.5713630105553006,
      "grad_norm": 0.0861680762249506,
      "learning_rate": 9.255742063475228e-05,
      "loss": 0.9474,
      "step": 1245
    },
    {
      "epoch": 0.5736576411197797,
      "grad_norm": 0.09031291699160855,
      "learning_rate": 9.175887150972841e-05,
      "loss": 0.9347,
      "step": 1250
    },
    {
      "epoch": 0.5759522716842589,
      "grad_norm": 0.08782934008281447,
      "learning_rate": 9.096085115728902e-05,
      "loss": 0.9254,
      "step": 1255
    },
    {
      "epoch": 0.578246902248738,
      "grad_norm": 0.08672572395996989,
      "learning_rate": 9.016341078052908e-05,
      "loss": 0.924,
      "step": 1260
    },
    {
      "epoch": 0.5805415328132171,
      "grad_norm": 0.08848493722986495,
      "learning_rate": 8.936660154533069e-05,
      "loss": 0.9398,
      "step": 1265
    },
    {
      "epoch": 0.5828361633776962,
      "grad_norm": 0.08795360842845397,
      "learning_rate": 8.857047457708023e-05,
      "loss": 0.9408,
      "step": 1270
    },
    {
      "epoch": 0.5851307939421753,
      "grad_norm": 0.0860614623181383,
      "learning_rate": 8.777508095738818e-05,
      "loss": 0.9334,
      "step": 1275
    },
    {
      "epoch": 0.5874254245066545,
      "grad_norm": 0.09151871492029366,
      "learning_rate": 8.698047172081128e-05,
      "loss": 0.9372,
      "step": 1280
    },
    {
      "epoch": 0.5897200550711336,
      "grad_norm": 0.08719515886726915,
      "learning_rate": 8.618669785157825e-05,
      "loss": 0.9346,
      "step": 1285
    },
    {
      "epoch": 0.5920146856356127,
      "grad_norm": 0.08602811823889805,
      "learning_rate": 8.539381028031837e-05,
      "loss": 0.95,
      "step": 1290
    },
    {
      "epoch": 0.5943093162000918,
      "grad_norm": 0.09010753852396704,
      "learning_rate": 8.460185988079379e-05,
      "loss": 0.9623,
      "step": 1295
    },
    {
      "epoch": 0.596603946764571,
      "grad_norm": 0.08768581543675588,
      "learning_rate": 8.381089746663517e-05,
      "loss": 0.9391,
      "step": 1300
    },
    {
      "epoch": 0.59889857732905,
      "grad_norm": 0.08992597459848777,
      "learning_rate": 8.302097378808146e-05,
      "loss": 0.9436,
      "step": 1305
    },
    {
      "epoch": 0.6011932078935291,
      "grad_norm": 0.08897797699662331,
      "learning_rate": 8.223213952872353e-05,
      "loss": 0.9466,
      "step": 1310
    },
    {
      "epoch": 0.6034878384580082,
      "grad_norm": 0.08711970422648517,
      "learning_rate": 8.144444530225236e-05,
      "loss": 0.9602,
      "step": 1315
    },
    {
      "epoch": 0.6057824690224873,
      "grad_norm": 0.08744571704998887,
      "learning_rate": 8.065794164921128e-05,
      "loss": 0.9609,
      "step": 1320
    },
    {
      "epoch": 0.6080770995869665,
      "grad_norm": 0.08732342917773199,
      "learning_rate": 7.987267903375332e-05,
      "loss": 0.9388,
      "step": 1325
    },
    {
      "epoch": 0.6103717301514456,
      "grad_norm": 0.08634671243153437,
      "learning_rate": 7.90887078404033e-05,
      "loss": 0.9379,
      "step": 1330
    },
    {
      "epoch": 0.6126663607159247,
      "grad_norm": 0.08792026326901815,
      "learning_rate": 7.830607837082493e-05,
      "loss": 0.9364,
      "step": 1335
    },
    {
      "epoch": 0.6149609912804038,
      "grad_norm": 0.08657044899326832,
      "learning_rate": 7.75248408405934e-05,
      "loss": 0.9502,
      "step": 1340
    },
    {
      "epoch": 0.617255621844883,
      "grad_norm": 0.08924319703176266,
      "learning_rate": 7.674504537597336e-05,
      "loss": 0.9431,
      "step": 1345
    },
    {
      "epoch": 0.6195502524093621,
      "grad_norm": 0.08777369904115767,
      "learning_rate": 7.596674201070282e-05,
      "loss": 0.9507,
      "step": 1350
    },
    {
      "epoch": 0.6218448829738412,
      "grad_norm": 0.08822604258128913,
      "learning_rate": 7.518998068278266e-05,
      "loss": 0.9389,
      "step": 1355
    },
    {
      "epoch": 0.6241395135383203,
      "grad_norm": 0.0872584357855511,
      "learning_rate": 7.441481123127257e-05,
      "loss": 0.9272,
      "step": 1360
    },
    {
      "epoch": 0.6264341441027994,
      "grad_norm": 0.09420031727135263,
      "learning_rate": 7.364128339309326e-05,
      "loss": 0.9496,
      "step": 1365
    },
    {
      "epoch": 0.6287287746672786,
      "grad_norm": 0.09178449177995368,
      "learning_rate": 7.28694467998352e-05,
      "loss": 0.9272,
      "step": 1370
    },
    {
      "epoch": 0.6310234052317577,
      "grad_norm": 0.08547328194451309,
      "learning_rate": 7.209935097457413e-05,
      "loss": 0.943,
      "step": 1375
    },
    {
      "epoch": 0.6333180357962368,
      "grad_norm": 0.08867995753938929,
      "learning_rate": 7.133104532869342e-05,
      "loss": 0.963,
      "step": 1380
    },
    {
      "epoch": 0.6356126663607159,
      "grad_norm": 0.08966519220883923,
      "learning_rate": 7.056457915871399e-05,
      "loss": 0.9317,
      "step": 1385
    },
    {
      "epoch": 0.637907296925195,
      "grad_norm": 0.08882768063994861,
      "learning_rate": 6.980000164313093e-05,
      "loss": 0.9226,
      "step": 1390
    },
    {
      "epoch": 0.6402019274896742,
      "grad_norm": 0.08837784094850205,
      "learning_rate": 6.903736183925835e-05,
      "loss": 0.939,
      "step": 1395
    },
    {
      "epoch": 0.6424965580541533,
      "grad_norm": 0.08678715695619918,
      "learning_rate": 6.827670868008171e-05,
      "loss": 0.9574,
      "step": 1400
    },
    {
      "epoch": 0.6447911886186324,
      "grad_norm": 0.08687361153125922,
      "learning_rate": 6.751809097111798e-05,
      "loss": 0.9608,
      "step": 1405
    },
    {
      "epoch": 0.6470858191831115,
      "grad_norm": 0.09043663200344924,
      "learning_rate": 6.676155738728438e-05,
      "loss": 0.9454,
      "step": 1410
    },
    {
      "epoch": 0.6493804497475907,
      "grad_norm": 0.08587170447948225,
      "learning_rate": 6.600715646977502e-05,
      "loss": 0.9436,
      "step": 1415
    },
    {
      "epoch": 0.6516750803120698,
      "grad_norm": 0.08747748944349737,
      "learning_rate": 6.525493662294668e-05,
      "loss": 0.9433,
      "step": 1420
    },
    {
      "epoch": 0.6539697108765489,
      "grad_norm": 0.08644951356394669,
      "learning_rate": 6.450494611121273e-05,
      "loss": 0.9544,
      "step": 1425
    },
    {
      "epoch": 0.656264341441028,
      "grad_norm": 0.08616266377880767,
      "learning_rate": 6.375723305594658e-05,
      "loss": 0.9502,
      "step": 1430
    },
    {
      "epoch": 0.6585589720055071,
      "grad_norm": 0.08828640675025567,
      "learning_rate": 6.301184543239398e-05,
      "loss": 0.9397,
      "step": 1435
    },
    {
      "epoch": 0.6608536025699863,
      "grad_norm": 0.08766906251082068,
      "learning_rate": 6.226883106659485e-05,
      "loss": 0.966,
      "step": 1440
    },
    {
      "epoch": 0.6631482331344654,
      "grad_norm": 0.0901196932930806,
      "learning_rate": 6.152823763231463e-05,
      "loss": 0.9307,
      "step": 1445
    },
    {
      "epoch": 0.6654428636989445,
      "grad_norm": 0.08653082781266369,
      "learning_rate": 6.079011264798534e-05,
      "loss": 0.9362,
      "step": 1450
    },
    {
      "epoch": 0.6677374942634235,
      "grad_norm": 0.08851540125345415,
      "learning_rate": 6.005450347365687e-05,
      "loss": 0.9575,
      "step": 1455
    },
    {
      "epoch": 0.6700321248279028,
      "grad_norm": 0.0898855821818352,
      "learning_rate": 5.932145730795793e-05,
      "loss": 0.94,
      "step": 1460
    },
    {
      "epoch": 0.6723267553923818,
      "grad_norm": 0.08793133426546608,
      "learning_rate": 5.8591021185067876e-05,
      "loss": 0.9307,
      "step": 1465
    },
    {
      "epoch": 0.6746213859568609,
      "grad_norm": 0.09185643968705434,
      "learning_rate": 5.786324197169887e-05,
      "loss": 0.9603,
      "step": 1470
    },
    {
      "epoch": 0.67691601652134,
      "grad_norm": 0.08642240102898958,
      "learning_rate": 5.7138166364088705e-05,
      "loss": 0.9417,
      "step": 1475
    },
    {
      "epoch": 0.6792106470858191,
      "grad_norm": 0.08511965941194902,
      "learning_rate": 5.641584088500461e-05,
      "loss": 0.94,
      "step": 1480
    },
    {
      "epoch": 0.6815052776502983,
      "grad_norm": 0.08723628225521773,
      "learning_rate": 5.569631188075841e-05,
      "loss": 0.9248,
      "step": 1485
    },
    {
      "epoch": 0.6837999082147774,
      "grad_norm": 0.08633274736360919,
      "learning_rate": 5.497962551823266e-05,
      "loss": 0.9469,
      "step": 1490
    },
    {
      "epoch": 0.6860945387792565,
      "grad_norm": 0.09266736313451518,
      "learning_rate": 5.4265827781918576e-05,
      "loss": 0.9692,
      "step": 1495
    },
    {
      "epoch": 0.6883891693437356,
      "grad_norm": 0.08989480494070151,
      "learning_rate": 5.355496447096533e-05,
      "loss": 0.969,
      "step": 1500
    },
    {
      "epoch": 0.6906837999082148,
      "grad_norm": 0.08776725254212203,
      "learning_rate": 5.284708119624173e-05,
      "loss": 0.9655,
      "step": 1505
    },
    {
      "epoch": 0.6929784304726939,
      "grad_norm": 0.0858923659739521,
      "learning_rate": 5.214222337740962e-05,
      "loss": 0.9364,
      "step": 1510
    },
    {
      "epoch": 0.695273061037173,
      "grad_norm": 0.08991893131765896,
      "learning_rate": 5.144043624000944e-05,
      "loss": 0.8978,
      "step": 1515
    },
    {
      "epoch": 0.6975676916016521,
      "grad_norm": 0.08502238183879142,
      "learning_rate": 5.0741764812558724e-05,
      "loss": 0.9402,
      "step": 1520
    },
    {
      "epoch": 0.6998623221661312,
      "grad_norm": 0.08668719280033117,
      "learning_rate": 5.00462539236628e-05,
      "loss": 0.9471,
      "step": 1525
    },
    {
      "epoch": 0.7021569527306104,
      "grad_norm": 0.08462698467850949,
      "learning_rate": 4.935394819913849e-05,
      "loss": 0.9249,
      "step": 1530
    },
    {
      "epoch": 0.7044515832950895,
      "grad_norm": 0.08607092895236686,
      "learning_rate": 4.8664892059150723e-05,
      "loss": 0.9602,
      "step": 1535
    },
    {
      "epoch": 0.7067462138595686,
      "grad_norm": 0.08935744366472125,
      "learning_rate": 4.7979129715362626e-05,
      "loss": 0.9798,
      "step": 1540
    },
    {
      "epoch": 0.7090408444240477,
      "grad_norm": 0.0854376252471685,
      "learning_rate": 4.72967051680985e-05,
      "loss": 0.9393,
      "step": 1545
    },
    {
      "epoch": 0.7113354749885269,
      "grad_norm": 0.08935720877957573,
      "learning_rate": 4.661766220352097e-05,
      "loss": 0.9224,
      "step": 1550
    },
    {
      "epoch": 0.713630105553006,
      "grad_norm": 0.08985254696056548,
      "learning_rate": 4.5942044390821214e-05,
      "loss": 0.9561,
      "step": 1555
    },
    {
      "epoch": 0.7159247361174851,
      "grad_norm": 0.08908651127378205,
      "learning_rate": 4.526989507942374e-05,
      "loss": 0.9339,
      "step": 1560
    },
    {
      "epoch": 0.7182193666819642,
      "grad_norm": 0.08630595159367531,
      "learning_rate": 4.460125739620479e-05,
      "loss": 0.9338,
      "step": 1565
    },
    {
      "epoch": 0.7205139972464433,
      "grad_norm": 0.08857577901710953,
      "learning_rate": 4.393617424272527e-05,
      "loss": 0.9493,
      "step": 1570
    },
    {
      "epoch": 0.7228086278109225,
      "grad_norm": 0.08887546613261192,
      "learning_rate": 4.3274688292478106e-05,
      "loss": 0.9469,
      "step": 1575
    },
    {
      "epoch": 0.7251032583754016,
      "grad_norm": 0.08719401842713113,
      "learning_rate": 4.261684198815003e-05,
      "loss": 0.9444,
      "step": 1580
    },
    {
      "epoch": 0.7273978889398807,
      "grad_norm": 0.08619407541853812,
      "learning_rate": 4.1962677538898645e-05,
      "loss": 0.944,
      "step": 1585
    },
    {
      "epoch": 0.7296925195043598,
      "grad_norm": 0.08443943693203552,
      "learning_rate": 4.131223691764383e-05,
      "loss": 0.9225,
      "step": 1590
    },
    {
      "epoch": 0.731987150068839,
      "grad_norm": 0.08565149963984416,
      "learning_rate": 4.0665561858374934e-05,
      "loss": 0.9289,
      "step": 1595
    },
    {
      "epoch": 0.7342817806333181,
      "grad_norm": 0.0878443494214115,
      "learning_rate": 4.0022693853472884e-05,
      "loss": 0.9111,
      "step": 1600
    },
    {
      "epoch": 0.7365764111977972,
      "grad_norm": 0.08904194669918195,
      "learning_rate": 3.938367415104793e-05,
      "loss": 0.9531,
      "step": 1605
    },
    {
      "epoch": 0.7388710417622762,
      "grad_norm": 0.08893651456090547,
      "learning_rate": 3.8748543752293e-05,
      "loss": 0.9584,
      "step": 1610
    },
    {
      "epoch": 0.7411656723267553,
      "grad_norm": 0.08897315588461706,
      "learning_rate": 3.8117343408853127e-05,
      "loss": 0.9319,
      "step": 1615
    },
    {
      "epoch": 0.7434603028912345,
      "grad_norm": 0.08616983463133208,
      "learning_rate": 3.7490113620210485e-05,
      "loss": 0.9146,
      "step": 1620
    },
    {
      "epoch": 0.7457549334557136,
      "grad_norm": 0.0873842160083806,
      "learning_rate": 3.686689463108608e-05,
      "loss": 0.9431,
      "step": 1625
    },
    {
      "epoch": 0.7480495640201927,
      "grad_norm": 0.08712124349433248,
      "learning_rate": 3.624772642885734e-05,
      "loss": 0.923,
      "step": 1630
    },
    {
      "epoch": 0.7503441945846718,
      "grad_norm": 0.08857054628390904,
      "learning_rate": 3.563264874099258e-05,
      "loss": 0.9537,
      "step": 1635
    },
    {
      "epoch": 0.752638825149151,
      "grad_norm": 0.08430268133869642,
      "learning_rate": 3.502170103250177e-05,
      "loss": 0.9093,
      "step": 1640
    },
    {
      "epoch": 0.7549334557136301,
      "grad_norm": 0.08778596880519368,
      "learning_rate": 3.441492250340461e-05,
      "loss": 0.9361,
      "step": 1645
    },
    {
      "epoch": 0.7572280862781092,
      "grad_norm": 0.08734144942203491,
      "learning_rate": 3.381235208621522e-05,
      "loss": 0.9259,
      "step": 1650
    },
    {
      "epoch": 0.7595227168425883,
      "grad_norm": 0.08734310507948465,
      "learning_rate": 3.3214028443444036e-05,
      "loss": 0.9354,
      "step": 1655
    },
    {
      "epoch": 0.7618173474070674,
      "grad_norm": 0.08820927771902966,
      "learning_rate": 3.2619989965117356e-05,
      "loss": 0.9134,
      "step": 1660
    },
    {
      "epoch": 0.7641119779715466,
      "grad_norm": 0.0890839295904084,
      "learning_rate": 3.2030274766313865e-05,
      "loss": 0.9326,
      "step": 1665
    },
    {
      "epoch": 0.7664066085360257,
      "grad_norm": 0.0930961939643943,
      "learning_rate": 3.1444920684719394e-05,
      "loss": 0.9347,
      "step": 1670
    },
    {
      "epoch": 0.7687012391005048,
      "grad_norm": 0.0870955697496051,
      "learning_rate": 3.086396527819876e-05,
      "loss": 0.9264,
      "step": 1675
    },
    {
      "epoch": 0.7709958696649839,
      "grad_norm": 0.0899287105378734,
      "learning_rate": 3.028744582238633e-05,
      "loss": 0.9407,
      "step": 1680
    },
    {
      "epoch": 0.7732905002294631,
      "grad_norm": 0.09021085870835642,
      "learning_rate": 2.9715399308294e-05,
      "loss": 0.9652,
      "step": 1685
    },
    {
      "epoch": 0.7755851307939422,
      "grad_norm": 0.08809579986676752,
      "learning_rate": 2.914786243993808e-05,
      "loss": 0.9271,
      "step": 1690
    },
    {
      "epoch": 0.7778797613584213,
      "grad_norm": 0.08806930004292979,
      "learning_rate": 2.8584871631983888e-05,
      "loss": 0.9353,
      "step": 1695
    },
    {
      "epoch": 0.7801743919229004,
      "grad_norm": 0.08874560139452363,
      "learning_rate": 2.8026463007409664e-05,
      "loss": 0.9364,
      "step": 1700
    },
    {
      "epoch": 0.7824690224873795,
      "grad_norm": 0.09095899541023127,
      "learning_rate": 2.747267239518857e-05,
      "loss": 0.9432,
      "step": 1705
    },
    {
      "epoch": 0.7847636530518587,
      "grad_norm": 0.08597962745032185,
      "learning_rate": 2.6923535327989924e-05,
      "loss": 0.9365,
      "step": 1710
    },
    {
      "epoch": 0.7870582836163378,
      "grad_norm": 0.08728499968023844,
      "learning_rate": 2.637908703989924e-05,
      "loss": 0.9546,
      "step": 1715
    },
    {
      "epoch": 0.7893529141808169,
      "grad_norm": 0.0865310427946889,
      "learning_rate": 2.5839362464157634e-05,
      "loss": 0.9444,
      "step": 1720
    },
    {
      "epoch": 0.791647544745296,
      "grad_norm": 0.08952067073529965,
      "learning_rate": 2.5304396230920345e-05,
      "loss": 0.9551,
      "step": 1725
    },
    {
      "epoch": 0.7939421753097752,
      "grad_norm": 0.08590487479214523,
      "learning_rate": 2.477422266503473e-05,
      "loss": 0.9416,
      "step": 1730
    },
    {
      "epoch": 0.7962368058742543,
      "grad_norm": 0.08843022208186156,
      "learning_rate": 2.4248875783837987e-05,
      "loss": 0.9401,
      "step": 1735
    },
    {
      "epoch": 0.7985314364387334,
      "grad_norm": 0.08728520520911832,
      "learning_rate": 2.3728389294974475e-05,
      "loss": 0.954,
      "step": 1740
    },
    {
      "epoch": 0.8008260670032125,
      "grad_norm": 0.09169106802431096,
      "learning_rate": 2.3212796594232943e-05,
      "loss": 0.9592,
      "step": 1745
    },
    {
      "epoch": 0.8031206975676916,
      "grad_norm": 0.08931921878894941,
      "learning_rate": 2.2702130763403673e-05,
      "loss": 0.9406,
      "step": 1750
    },
    {
      "epoch": 0.8054153281321708,
      "grad_norm": 0.08764598773379657,
      "learning_rate": 2.2196424568156073e-05,
      "loss": 0.9384,
      "step": 1755
    },
    {
      "epoch": 0.8077099586966499,
      "grad_norm": 0.08855507556833625,
      "learning_rate": 2.1695710455936115e-05,
      "loss": 0.9433,
      "step": 1760
    },
    {
      "epoch": 0.810004589261129,
      "grad_norm": 0.08532871067750841,
      "learning_rate": 2.1200020553884604e-05,
      "loss": 0.9346,
      "step": 1765
    },
    {
      "epoch": 0.812299219825608,
      "grad_norm": 0.08966754551907985,
      "learning_rate": 2.070938666677573e-05,
      "loss": 0.9508,
      "step": 1770
    },
    {
      "epoch": 0.8145938503900872,
      "grad_norm": 0.09025156831747091,
      "learning_rate": 2.0223840274976413e-05,
      "loss": 0.9809,
      "step": 1775
    },
    {
      "epoch": 0.8168884809545663,
      "grad_norm": 0.08854265504832948,
      "learning_rate": 1.974341253242635e-05,
      "loss": 0.9191,
      "step": 1780
    },
    {
      "epoch": 0.8191831115190454,
      "grad_norm": 0.08884601227475634,
      "learning_rate": 1.9268134264639272e-05,
      "loss": 0.9446,
      "step": 1785
    },
    {
      "epoch": 0.8214777420835245,
      "grad_norm": 0.0900014724971187,
      "learning_rate": 1.879803596672497e-05,
      "loss": 0.9569,
      "step": 1790
    },
    {
      "epoch": 0.8237723726480036,
      "grad_norm": 0.08586515698037704,
      "learning_rate": 1.8333147801432614e-05,
      "loss": 0.9237,
      "step": 1795
    },
    {
      "epoch": 0.8260670032124828,
      "grad_norm": 0.08704559914739361,
      "learning_rate": 1.7873499597215604e-05,
      "loss": 0.9146,
      "step": 1800
    },
    {
      "epoch": 0.8283616337769619,
      "grad_norm": 0.08957226377233579,
      "learning_rate": 1.741912084631746e-05,
      "loss": 0.9495,
      "step": 1805
    },
    {
      "epoch": 0.830656264341441,
      "grad_norm": 0.0862080402115559,
      "learning_rate": 1.697004070287982e-05,
      "loss": 0.9089,
      "step": 1810
    },
    {
      "epoch": 0.8329508949059201,
      "grad_norm": 0.08632960573186395,
      "learning_rate": 1.6526287981071475e-05,
      "loss": 0.9157,
      "step": 1815
    },
    {
      "epoch": 0.8352455254703993,
      "grad_norm": 0.08440907866960998,
      "learning_rate": 1.608789115323993e-05,
      "loss": 0.9658,
      "step": 1820
    },
    {
      "epoch": 0.8375401560348784,
      "grad_norm": 0.08731523817505203,
      "learning_rate": 1.5654878348084244e-05,
      "loss": 0.9551,
      "step": 1825
    },
    {
      "epoch": 0.8398347865993575,
      "grad_norm": 0.08915491610885863,
      "learning_rate": 1.5227277348850466e-05,
      "loss": 0.9446,
      "step": 1830
    },
    {
      "epoch": 0.8421294171638366,
      "grad_norm": 0.0869732185667195,
      "learning_rate": 1.4805115591548746e-05,
      "loss": 0.9225,
      "step": 1835
    },
    {
      "epoch": 0.8444240477283157,
      "grad_norm": 0.08670718205091706,
      "learning_rate": 1.4388420163193217e-05,
      "loss": 0.9261,
      "step": 1840
    },
    {
      "epoch": 0.8467186782927949,
      "grad_norm": 0.09137885764266358,
      "learning_rate": 1.3977217800063846e-05,
      "loss": 0.9469,
      "step": 1845
    },
    {
      "epoch": 0.849013308857274,
      "grad_norm": 0.08818902846204707,
      "learning_rate": 1.3571534885991044e-05,
      "loss": 0.9192,
      "step": 1850
    },
    {
      "epoch": 0.8513079394217531,
      "grad_norm": 0.08515505179393418,
      "learning_rate": 1.3171397450662715e-05,
      "loss": 0.9322,
      "step": 1855
    },
    {
      "epoch": 0.8536025699862322,
      "grad_norm": 0.09793678826450831,
      "learning_rate": 1.277683116795425e-05,
      "loss": 0.9209,
      "step": 1860
    },
    {
      "epoch": 0.8558972005507114,
      "grad_norm": 0.08921313107573868,
      "learning_rate": 1.2387861354281194e-05,
      "loss": 0.9459,
      "step": 1865
    },
    {
      "epoch": 0.8581918311151905,
      "grad_norm": 0.08675456932143413,
      "learning_rate": 1.2004512966974745e-05,
      "loss": 0.9341,
      "step": 1870
    },
    {
      "epoch": 0.8604864616796696,
      "grad_norm": 0.08853376406641464,
      "learning_rate": 1.162681060268065e-05,
      "loss": 0.9358,
      "step": 1875
    },
    {
      "epoch": 0.8627810922441487,
      "grad_norm": 0.08772170537544985,
      "learning_rate": 1.1254778495780748e-05,
      "loss": 0.9462,
      "step": 1880
    },
    {
      "epoch": 0.8650757228086278,
      "grad_norm": 0.087438966238015,
      "learning_rate": 1.0888440516838373e-05,
      "loss": 0.9371,
      "step": 1885
    },
    {
      "epoch": 0.867370353373107,
      "grad_norm": 0.08774570416471039,
      "learning_rate": 1.0527820171066372e-05,
      "loss": 0.9616,
      "step": 1890
    },
    {
      "epoch": 0.8696649839375861,
      "grad_norm": 0.09079641508572181,
      "learning_rate": 1.0172940596819259e-05,
      "loss": 0.9203,
      "step": 1895
    },
    {
      "epoch": 0.8719596145020652,
      "grad_norm": 0.08682703034420731,
      "learning_rate": 9.823824564108408e-06,
      "loss": 0.9572,
      "step": 1900
    },
    {
      "epoch": 0.8742542450665443,
      "grad_norm": 0.08884587029741915,
      "learning_rate": 9.480494473141188e-06,
      "loss": 0.9565,
      "step": 1905
    },
    {
      "epoch": 0.8765488756310233,
      "grad_norm": 0.08810947120591099,
      "learning_rate": 9.142972352883594e-06,
      "loss": 0.9452,
      "step": 1910
    },
    {
      "epoch": 0.8788435061955026,
      "grad_norm": 0.08431726449593704,
      "learning_rate": 8.811279859646915e-06,
      "loss": 0.9275,
      "step": 1915
    },
    {
      "epoch": 0.8811381367599816,
      "grad_norm": 0.08783013214009486,
      "learning_rate": 8.485438275698154e-06,
      "loss": 0.9344,
      "step": 1920
    },
    {
      "epoch": 0.8834327673244607,
      "grad_norm": 0.0877811337071556,
      "learning_rate": 8.165468507894514e-06,
      "loss": 0.9422,
      "step": 1925
    },
    {
      "epoch": 0.8857273978889398,
      "grad_norm": 0.08918928300947922,
      "learning_rate": 7.851391086341952e-06,
      "loss": 0.923,
      "step": 1930
    },
    {
      "epoch": 0.888022028453419,
      "grad_norm": 0.08684462019085216,
      "learning_rate": 7.543226163077899e-06,
      "loss": 0.9501,
      "step": 1935
    },
    {
      "epoch": 0.8903166590178981,
      "grad_norm": 0.08897290874527142,
      "learning_rate": 7.240993510778305e-06,
      "loss": 0.9468,
      "step": 1940
    },
    {
      "epoch": 0.8926112895823772,
      "grad_norm": 0.08848785350377304,
      "learning_rate": 6.9447125214888835e-06,
      "loss": 0.9407,
      "step": 1945
    },
    {
      "epoch": 0.8949059201468563,
      "grad_norm": 0.08850978272970857,
      "learning_rate": 6.65440220538096e-06,
      "loss": 0.9467,
      "step": 1950
    },
    {
      "epoch": 0.8972005507113354,
      "grad_norm": 0.08764525689057708,
      "learning_rate": 6.370081189531707e-06,
      "loss": 0.9183,
      "step": 1955
    },
    {
      "epoch": 0.8994951812758146,
      "grad_norm": 0.09111687142640715,
      "learning_rate": 6.091767716728924e-06,
      "loss": 0.9337,
      "step": 1960
    },
    {
      "epoch": 0.9017898118402937,
      "grad_norm": 0.09082638866765445,
      "learning_rate": 5.819479644300563e-06,
      "loss": 0.9311,
      "step": 1965
    },
    {
      "epoch": 0.9040844424047728,
      "grad_norm": 0.08768468656217991,
      "learning_rate": 5.553234442969013e-06,
      "loss": 0.9382,
      "step": 1970
    },
    {
      "epoch": 0.9063790729692519,
      "grad_norm": 0.08591220062356768,
      "learning_rate": 5.293049195730038e-06,
      "loss": 0.9363,
      "step": 1975
    },
    {
      "epoch": 0.9086737035337311,
      "grad_norm": 0.0892059058046249,
      "learning_rate": 5.038940596756747e-06,
      "loss": 0.9562,
      "step": 1980
    },
    {
      "epoch": 0.9109683340982102,
      "grad_norm": 0.08874849077210194,
      "learning_rate": 4.790924950328435e-06,
      "loss": 0.9528,
      "step": 1985
    },
    {
      "epoch": 0.9132629646626893,
      "grad_norm": 0.08444810522115238,
      "learning_rate": 4.5490181697844916e-06,
      "loss": 0.9037,
      "step": 1990
    },
    {
      "epoch": 0.9155575952271684,
      "grad_norm": 0.0860541394012721,
      "learning_rate": 4.313235776503244e-06,
      "loss": 0.9153,
      "step": 1995
    },
    {
      "epoch": 0.9178522257916475,
      "grad_norm": 0.08677442991318424,
      "learning_rate": 4.08359289890623e-06,
      "loss": 0.9402,
      "step": 2000
    },
    {
      "epoch": 0.9201468563561267,
      "grad_norm": 0.09013856091786,
      "learning_rate": 3.860104271487397e-06,
      "loss": 0.9529,
      "step": 2005
    },
    {
      "epoch": 0.9224414869206058,
      "grad_norm": 0.08509499257002519,
      "learning_rate": 3.6427842338677352e-06,
      "loss": 0.9323,
      "step": 2010
    },
    {
      "epoch": 0.9247361174850849,
      "grad_norm": 0.08719834616969914,
      "learning_rate": 3.4316467298752263e-06,
      "loss": 0.9392,
      "step": 2015
    },
    {
      "epoch": 0.927030748049564,
      "grad_norm": 0.08697447693662008,
      "learning_rate": 3.226705306650113e-06,
      "loss": 0.9446,
      "step": 2020
    },
    {
      "epoch": 0.9293253786140432,
      "grad_norm": 0.08969207347043825,
      "learning_rate": 3.0279731137757948e-06,
      "loss": 0.9119,
      "step": 2025
    },
    {
      "epoch": 0.9316200091785223,
      "grad_norm": 0.08694060973009726,
      "learning_rate": 2.835462902434971e-06,
      "loss": 0.9081,
      "step": 2030
    },
    {
      "epoch": 0.9339146397430014,
      "grad_norm": 0.08665729431202235,
      "learning_rate": 2.649187024591604e-06,
      "loss": 0.9491,
      "step": 2035
    },
    {
      "epoch": 0.9362092703074805,
      "grad_norm": 0.08529555353847522,
      "learning_rate": 2.4691574321983215e-06,
      "loss": 0.9545,
      "step": 2040
    },
    {
      "epoch": 0.9385039008719596,
      "grad_norm": 0.08638294723774989,
      "learning_rate": 2.2953856764295624e-06,
      "loss": 0.9433,
      "step": 2045
    },
    {
      "epoch": 0.9407985314364388,
      "grad_norm": 0.08509058036503529,
      "learning_rate": 2.1278829069404484e-06,
      "loss": 0.8975,
      "step": 2050
    },
    {
      "epoch": 0.9430931620009179,
      "grad_norm": 0.09005898438040956,
      "learning_rate": 1.966659871151366e-06,
      "loss": 0.9441,
      "step": 2055
    },
    {
      "epoch": 0.945387792565397,
      "grad_norm": 0.0913580806589797,
      "learning_rate": 1.811726913558387e-06,
      "loss": 0.9383,
      "step": 2060
    },
    {
      "epoch": 0.947682423129876,
      "grad_norm": 0.08996121201187086,
      "learning_rate": 1.6630939750695519e-06,
      "loss": 0.9266,
      "step": 2065
    },
    {
      "epoch": 0.9499770536943553,
      "grad_norm": 0.08735684631954084,
      "learning_rate": 1.5207705923670157e-06,
      "loss": 0.9374,
      "step": 2070
    },
    {
      "epoch": 0.9522716842588343,
      "grad_norm": 0.08782321872428674,
      "learning_rate": 1.3847658972951483e-06,
      "loss": 0.9393,
      "step": 2075
    },
    {
      "epoch": 0.9545663148233134,
      "grad_norm": 0.08638359940423465,
      "learning_rate": 1.2550886162746467e-06,
      "loss": 0.9542,
      "step": 2080
    },
    {
      "epoch": 0.9568609453877925,
      "grad_norm": 0.08816158923371066,
      "learning_rate": 1.1317470697425837e-06,
      "loss": 0.9553,
      "step": 2085
    },
    {
      "epoch": 0.9591555759522716,
      "grad_norm": 0.0869294153247488,
      "learning_rate": 1.0147491716185676e-06,
      "loss": 0.9338,
      "step": 2090
    },
    {
      "epoch": 0.9614502065167508,
      "grad_norm": 0.08669044776728387,
      "learning_rate": 9.04102428796949e-07,
      "loss": 0.9598,
      "step": 2095
    },
    {
      "epoch": 0.9637448370812299,
      "grad_norm": 0.08623247272425665,
      "learning_rate": 7.99813940665195e-07,
      "loss": 0.9373,
      "step": 2100
    },
    {
      "epoch": 0.966039467645709,
      "grad_norm": 0.08874393697081741,
      "learning_rate": 7.018903986483083e-07,
      "loss": 0.9415,
      "step": 2105
    },
    {
      "epoch": 0.9683340982101881,
      "grad_norm": 0.08826549132097718,
      "learning_rate": 6.103380857795604e-07,
      "loss": 0.9509,
      "step": 2110
    },
    {
      "epoch": 0.9706287287746673,
      "grad_norm": 0.08811680099621978,
      "learning_rate": 5.251628762972916e-07,
      "loss": 0.9307,
      "step": 2115
    },
    {
      "epoch": 0.9729233593391464,
      "grad_norm": 0.0855595252734913,
      "learning_rate": 4.463702352680787e-07,
      "loss": 0.9272,
      "step": 2120
    },
    {
      "epoch": 0.9752179899036255,
      "grad_norm": 0.086175580567214,
      "learning_rate": 3.7396521823600537e-07,
      "loss": 0.9209,
      "step": 2125
    },
    {
      "epoch": 0.9775126204681046,
      "grad_norm": 0.08552971051081726,
      "learning_rate": 3.079524708983095e-07,
      "loss": 0.9297,
      "step": 2130
    },
    {
      "epoch": 0.9798072510325837,
      "grad_norm": 0.0887704339359113,
      "learning_rate": 2.483362288073443e-07,
      "loss": 0.9283,
      "step": 2135
    },
    {
      "epoch": 0.9821018815970629,
      "grad_norm": 0.08671129007392188,
      "learning_rate": 1.9512031709874035e-07,
      "loss": 0.9332,
      "step": 2140
    },
    {
      "epoch": 0.984396512161542,
      "grad_norm": 0.08984732497930137,
      "learning_rate": 1.4830815024606815e-07,
      "loss": 0.9247,
      "step": 2145
    },
    {
      "epoch": 0.9866911427260211,
      "grad_norm": 0.0894694714684205,
      "learning_rate": 1.0790273184164701e-07,
      "loss": 0.9074,
      "step": 2150
    },
    {
      "epoch": 0.9889857732905002,
      "grad_norm": 0.0890952939852077,
      "learning_rate": 7.390665440393241e-08,
      "loss": 0.9366,
      "step": 2155
    },
    {
      "epoch": 0.9912804038549794,
      "grad_norm": 0.08892399654551203,
      "learning_rate": 4.632209921107134e-08,
      "loss": 0.9486,
      "step": 2160
    },
    {
      "epoch": 0.9935750344194585,
      "grad_norm": 0.09036079277558923,
      "learning_rate": 2.5150836161058622e-08,
      "loss": 0.9386,
      "step": 2165
    },
    {
      "epoch": 0.9958696649839376,
      "grad_norm": 0.0874025371093686,
      "learning_rate": 1.03942236580723e-08,
      "loss": 0.9373,
      "step": 2170
    },
    {
      "epoch": 0.9981642955484167,
      "grad_norm": 0.0950153209645722,
      "learning_rate": 2.053208525365502e-09,
      "loss": 0.9454,
      "step": 2175
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.943856418132782,
      "eval_runtime": 822.8687,
      "eval_samples_per_second": 18.753,
      "eval_steps_per_second": 0.294,
      "step": 2179
    },
    {
      "epoch": 1.0,
      "step": 2179,
      "total_flos": 3.4589037545127936e+16,
      "train_loss": 0.9514529290906111,
      "train_runtime": 33671.5063,
      "train_samples_per_second": 4.141,
      "train_steps_per_second": 0.065
    }
  ],
  "logging_steps": 5,
  "max_steps": 2179,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.4589037545127936e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}