{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "eval_steps": 500,
  "global_step": 37420,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002672367717797969,
      "grad_norm": 1.0593788623809814,
      "learning_rate": 0.0001999999647578693,
      "loss": 2.8682,
      "step": 10
    },
    {
      "epoch": 0.005344735435595938,
      "grad_norm": 1.3729766607284546,
      "learning_rate": 0.00019999985903150193,
      "loss": 2.4555,
      "step": 20
    },
    {
      "epoch": 0.008017103153393906,
      "grad_norm": 0.9050101041793823,
      "learning_rate": 0.00019999968282097247,
      "loss": 2.2431,
      "step": 30
    },
    {
      "epoch": 0.010689470871191877,
      "grad_norm": 0.8476489186286926,
      "learning_rate": 0.0001999994361264051,
      "loss": 2.2865,
      "step": 40
    },
    {
      "epoch": 0.013361838588989846,
      "grad_norm": 0.8859209418296814,
      "learning_rate": 0.00019999911894797372,
      "loss": 2.2338,
      "step": 50
    },
    {
      "epoch": 0.016034206306787813,
      "grad_norm": 0.8001280426979065,
      "learning_rate": 0.00019999873128590187,
      "loss": 2.2895,
      "step": 60
    },
    {
      "epoch": 0.01870657402458578,
      "grad_norm": 0.8251445293426514,
      "learning_rate": 0.00019999827314046284,
      "loss": 2.2898,
      "step": 70
    },
    {
      "epoch": 0.021378941742383754,
      "grad_norm": 0.7365292310714722,
      "learning_rate": 0.00019999774451197945,
      "loss": 2.2032,
      "step": 80
    },
    {
      "epoch": 0.024051309460181722,
      "grad_norm": 0.8399436473846436,
      "learning_rate": 0.0001999971454008244,
      "loss": 2.1982,
      "step": 90
    },
    {
      "epoch": 0.02672367717797969,
      "grad_norm": 0.7834980487823486,
      "learning_rate": 0.00019999647580741989,
      "loss": 2.2182,
      "step": 100
    },
    {
      "epoch": 0.02939604489577766,
      "grad_norm": 0.7640682458877563,
      "learning_rate": 0.00019999573573223795,
      "loss": 2.2249,
      "step": 110
    },
    {
      "epoch": 0.032068412613575625,
      "grad_norm": 0.8014008402824402,
      "learning_rate": 0.0001999949251758002,
      "loss": 2.2053,
      "step": 120
    },
    {
      "epoch": 0.034740780331373594,
      "grad_norm": 0.7757932543754578,
      "learning_rate": 0.0001999940441386779,
      "loss": 2.2097,
      "step": 130
    },
    {
      "epoch": 0.03741314804917156,
      "grad_norm": 0.6968387961387634,
      "learning_rate": 0.0001999930926214921,
      "loss": 2.2222,
      "step": 140
    },
    {
      "epoch": 0.04008551576696953,
      "grad_norm": 0.7413656711578369,
      "learning_rate": 0.00019999207062491346,
      "loss": 2.2071,
      "step": 150
    },
    {
      "epoch": 0.04275788348476751,
      "grad_norm": 0.9104848504066467,
      "learning_rate": 0.0001999909781496623,
      "loss": 2.2083,
      "step": 160
    },
    {
      "epoch": 0.045430251202565476,
      "grad_norm": 0.7201218605041504,
      "learning_rate": 0.00019998981519650868,
      "loss": 2.1978,
      "step": 170
    },
    {
      "epoch": 0.048102618920363445,
      "grad_norm": 0.8535075783729553,
      "learning_rate": 0.00019998858176627227,
      "loss": 2.1774,
      "step": 180
    },
    {
      "epoch": 0.050774986638161414,
      "grad_norm": 0.7436001896858215,
      "learning_rate": 0.00019998727785982247,
      "loss": 2.0783,
      "step": 190
    },
    {
      "epoch": 0.05344735435595938,
      "grad_norm": 0.7691600918769836,
      "learning_rate": 0.00019998590347807826,
      "loss": 2.2117,
      "step": 200
    },
    {
      "epoch": 0.05611972207375735,
      "grad_norm": 0.7730023860931396,
      "learning_rate": 0.00019998445862200845,
      "loss": 2.1707,
      "step": 210
    },
    {
      "epoch": 0.05879208979155532,
      "grad_norm": 0.7378381490707397,
      "learning_rate": 0.00019998294329263142,
      "loss": 2.1539,
      "step": 220
    },
    {
      "epoch": 0.06146445750935329,
      "grad_norm": 0.7207351922988892,
      "learning_rate": 0.0001999813574910152,
      "loss": 2.1684,
      "step": 230
    },
    {
      "epoch": 0.06413682522715125,
      "grad_norm": 0.7563502788543701,
      "learning_rate": 0.00019997970121827754,
      "loss": 2.21,
      "step": 240
    },
    {
      "epoch": 0.06680919294494922,
      "grad_norm": 0.7375833988189697,
      "learning_rate": 0.0001999779744755859,
      "loss": 2.1732,
      "step": 250
    },
    {
      "epoch": 0.06948156066274719,
      "grad_norm": 0.8896276950836182,
      "learning_rate": 0.0001999761772641573,
      "loss": 2.1718,
      "step": 260
    },
    {
      "epoch": 0.07215392838054516,
      "grad_norm": 0.7230550050735474,
      "learning_rate": 0.0001999743095852585,
      "loss": 2.1987,
      "step": 270
    },
    {
      "epoch": 0.07482629609834313,
      "grad_norm": 0.7491215467453003,
      "learning_rate": 0.00019997237144020596,
      "loss": 2.1536,
      "step": 280
    },
    {
      "epoch": 0.0774986638161411,
      "grad_norm": 0.8419792056083679,
      "learning_rate": 0.00019997036283036574,
      "loss": 2.1556,
      "step": 290
    },
    {
      "epoch": 0.08017103153393906,
      "grad_norm": 0.751606822013855,
      "learning_rate": 0.00019996828375715356,
      "loss": 2.1717,
      "step": 300
    },
    {
      "epoch": 0.08284339925173703,
      "grad_norm": 0.8259275555610657,
      "learning_rate": 0.00019996613422203492,
      "loss": 2.1644,
      "step": 310
    },
    {
      "epoch": 0.08551576696953501,
      "grad_norm": 0.8105437755584717,
      "learning_rate": 0.00019996391422652488,
      "loss": 2.0957,
      "step": 320
    },
    {
      "epoch": 0.08818813468733298,
      "grad_norm": 0.7292518615722656,
      "learning_rate": 0.0001999616237721881,
      "loss": 2.1461,
      "step": 330
    },
    {
      "epoch": 0.09086050240513095,
      "grad_norm": 0.7527144551277161,
      "learning_rate": 0.0001999592628606391,
      "loss": 2.1472,
      "step": 340
    },
    {
      "epoch": 0.09353287012292892,
      "grad_norm": 0.7687304615974426,
      "learning_rate": 0.00019995683149354193,
      "loss": 2.224,
      "step": 350
    },
    {
      "epoch": 0.09620523784072689,
      "grad_norm": 0.7979860305786133,
      "learning_rate": 0.00019995432967261025,
      "loss": 2.1426,
      "step": 360
    },
    {
      "epoch": 0.09887760555852486,
      "grad_norm": 0.7962672114372253,
      "learning_rate": 0.00019995175739960752,
      "loss": 2.1492,
      "step": 370
    },
    {
      "epoch": 0.10154997327632283,
      "grad_norm": 0.7849389314651489,
      "learning_rate": 0.00019994911467634675,
      "loss": 2.0579,
      "step": 380
    },
    {
      "epoch": 0.1042223409941208,
      "grad_norm": 0.8130772113800049,
      "learning_rate": 0.0001999464015046907,
      "loss": 2.1402,
      "step": 390
    },
    {
      "epoch": 0.10689470871191876,
      "grad_norm": 0.7432055473327637,
      "learning_rate": 0.00019994361788655164,
      "loss": 2.1381,
      "step": 400
    },
    {
      "epoch": 0.10956707642971673,
      "grad_norm": 0.7954995632171631,
      "learning_rate": 0.00019994076382389166,
      "loss": 2.1832,
      "step": 410
    },
    {
      "epoch": 0.1122394441475147,
      "grad_norm": 0.8514792323112488,
      "learning_rate": 0.0001999378393187224,
      "loss": 2.2132,
      "step": 420
    },
    {
      "epoch": 0.11491181186531267,
      "grad_norm": 0.9865955114364624,
      "learning_rate": 0.00019993484437310518,
      "loss": 2.1417,
      "step": 430
    },
    {
      "epoch": 0.11758417958311064,
      "grad_norm": 0.7414752244949341,
      "learning_rate": 0.00019993177898915094,
      "loss": 2.1338,
      "step": 440
    },
    {
      "epoch": 0.12025654730090861,
      "grad_norm": 0.6959917545318604,
      "learning_rate": 0.00019992864316902034,
      "loss": 2.1111,
      "step": 450
    },
    {
      "epoch": 0.12292891501870658,
      "grad_norm": 0.7759518623352051,
      "learning_rate": 0.0001999254369149236,
      "loss": 2.0755,
      "step": 460
    },
    {
      "epoch": 0.12560128273650453,
      "grad_norm": 0.8887970447540283,
      "learning_rate": 0.0001999221602291206,
      "loss": 2.1154,
      "step": 470
    },
    {
      "epoch": 0.1282736504543025,
      "grad_norm": 0.8119891285896301,
      "learning_rate": 0.00019991881311392097,
      "loss": 2.1146,
      "step": 480
    },
    {
      "epoch": 0.13094601817210047,
      "grad_norm": 0.8034600615501404,
      "learning_rate": 0.0001999153955716838,
      "loss": 2.138,
      "step": 490
    },
    {
      "epoch": 0.13361838588989844,
      "grad_norm": 0.8797882795333862,
      "learning_rate": 0.000199911907604818,
      "loss": 2.15,
      "step": 500
    },
    {
      "epoch": 0.1362907536076964,
      "grad_norm": 0.8310526609420776,
      "learning_rate": 0.00019990834921578205,
      "loss": 2.0907,
      "step": 510
    },
    {
      "epoch": 0.13896312132549438,
      "grad_norm": 0.8153840899467468,
      "learning_rate": 0.000199904720407084,
      "loss": 2.1459,
      "step": 520
    },
    {
      "epoch": 0.14163548904329235,
      "grad_norm": 0.7525952458381653,
      "learning_rate": 0.00019990102118128155,
      "loss": 2.2412,
      "step": 530
    },
    {
      "epoch": 0.14430785676109031,
      "grad_norm": 0.8331490159034729,
      "learning_rate": 0.00019989725154098218,
      "loss": 2.1192,
      "step": 540
    },
    {
      "epoch": 0.14698022447888828,
      "grad_norm": 0.780968189239502,
      "learning_rate": 0.00019989341148884282,
      "loss": 2.0778,
      "step": 550
    },
    {
      "epoch": 0.14965259219668625,
      "grad_norm": 0.8250514268875122,
      "learning_rate": 0.00019988950102757012,
      "loss": 2.179,
      "step": 560
    },
    {
      "epoch": 0.15232495991448422,
      "grad_norm": 0.8055257201194763,
      "learning_rate": 0.00019988552015992036,
      "loss": 2.0793,
      "step": 570
    },
    {
      "epoch": 0.1549973276322822,
      "grad_norm": 0.9461053013801575,
      "learning_rate": 0.00019988146888869939,
      "loss": 2.0553,
      "step": 580
    },
    {
      "epoch": 0.15766969535008016,
      "grad_norm": 0.9230166077613831,
      "learning_rate": 0.00019987734721676272,
      "loss": 2.1883,
      "step": 590
    },
    {
      "epoch": 0.16034206306787813,
      "grad_norm": 0.8047350645065308,
      "learning_rate": 0.00019987315514701553,
      "loss": 2.1516,
      "step": 600
    },
    {
      "epoch": 0.1630144307856761,
      "grad_norm": 0.8617376685142517,
      "learning_rate": 0.00019986889268241252,
      "loss": 2.0262,
      "step": 610
    },
    {
      "epoch": 0.16568679850347406,
      "grad_norm": 0.8188526630401611,
      "learning_rate": 0.00019986455982595803,
      "loss": 2.106,
      "step": 620
    },
    {
      "epoch": 0.16835916622127206,
      "grad_norm": 0.7814268469810486,
      "learning_rate": 0.0001998601565807061,
      "loss": 2.0266,
      "step": 630
    },
    {
      "epoch": 0.17103153393907003,
      "grad_norm": 0.8137741088867188,
      "learning_rate": 0.00019985568294976036,
      "loss": 2.0728,
      "step": 640
    },
    {
      "epoch": 0.173703901656868,
      "grad_norm": 0.74834805727005,
      "learning_rate": 0.0001998511389362739,
      "loss": 2.1,
      "step": 650
    },
    {
      "epoch": 0.17637626937466597,
      "grad_norm": 0.841573178768158,
      "learning_rate": 0.00019984652454344966,
      "loss": 2.1448,
      "step": 660
    },
    {
      "epoch": 0.17904863709246394,
      "grad_norm": 0.8152615427970886,
      "learning_rate": 0.00019984183977453995,
      "loss": 2.1404,
      "step": 670
    },
    {
      "epoch": 0.1817210048102619,
      "grad_norm": 0.8331522941589355,
      "learning_rate": 0.00019983708463284686,
      "loss": 2.054,
      "step": 680
    },
    {
      "epoch": 0.18439337252805987,
      "grad_norm": 0.9017409086227417,
      "learning_rate": 0.000199832259121722,
      "loss": 2.1148,
      "step": 690
    },
    {
      "epoch": 0.18706574024585784,
      "grad_norm": 0.8750881552696228,
      "learning_rate": 0.00019982736324456658,
      "loss": 2.0815,
      "step": 700
    },
    {
      "epoch": 0.1897381079636558,
      "grad_norm": 0.8703632950782776,
      "learning_rate": 0.00019982239700483146,
      "loss": 2.0799,
      "step": 710
    },
    {
      "epoch": 0.19241047568145378,
      "grad_norm": 0.8342357873916626,
      "learning_rate": 0.00019981736040601704,
      "loss": 2.0941,
      "step": 720
    },
    {
      "epoch": 0.19508284339925175,
      "grad_norm": 0.8456534147262573,
      "learning_rate": 0.0001998122534516733,
      "loss": 2.091,
      "step": 730
    },
    {
      "epoch": 0.19775521111704972,
      "grad_norm": 0.7912267446517944,
      "learning_rate": 0.0001998070761453999,
      "loss": 2.1353,
      "step": 740
    },
    {
      "epoch": 0.20042757883484769,
      "grad_norm": 0.7755369544029236,
      "learning_rate": 0.00019980182849084594,
      "loss": 2.1272,
      "step": 750
    },
    {
      "epoch": 0.20309994655264565,
      "grad_norm": 0.7691362500190735,
      "learning_rate": 0.00019979651049171026,
      "loss": 2.1243,
      "step": 760
    },
    {
      "epoch": 0.20577231427044362,
      "grad_norm": 0.8000863194465637,
      "learning_rate": 0.0001997911221517412,
      "loss": 2.0803,
      "step": 770
    },
    {
      "epoch": 0.2084446819882416,
      "grad_norm": 0.7349149584770203,
      "learning_rate": 0.0001997856634747367,
      "loss": 2.1411,
      "step": 780
    },
    {
      "epoch": 0.21111704970603956,
      "grad_norm": 0.865577220916748,
      "learning_rate": 0.00019978013446454425,
      "loss": 2.16,
      "step": 790
    },
    {
      "epoch": 0.21378941742383753,
      "grad_norm": 0.7996314167976379,
      "learning_rate": 0.00019977453512506094,
      "loss": 2.0745,
      "step": 800
    },
    {
      "epoch": 0.2164617851416355,
      "grad_norm": 0.8223113417625427,
      "learning_rate": 0.0001997688654602334,
      "loss": 2.1165,
      "step": 810
    },
    {
      "epoch": 0.21913415285943347,
      "grad_norm": 0.7781673669815063,
      "learning_rate": 0.00019976312547405784,
      "loss": 2.1064,
      "step": 820
    },
    {
      "epoch": 0.22180652057723144,
      "grad_norm": 0.9179145097732544,
      "learning_rate": 0.00019975731517058012,
      "loss": 2.1021,
      "step": 830
    },
    {
      "epoch": 0.2244788882950294,
      "grad_norm": 0.8308878540992737,
      "learning_rate": 0.00019975143455389555,
      "loss": 2.0775,
      "step": 840
    },
    {
      "epoch": 0.22715125601282737,
      "grad_norm": 0.7753422260284424,
      "learning_rate": 0.000199745483628149,
      "loss": 2.0697,
      "step": 850
    },
    {
      "epoch": 0.22982362373062534,
      "grad_norm": 0.8096175193786621,
      "learning_rate": 0.000199739462397535,
      "loss": 2.014,
      "step": 860
    },
    {
      "epoch": 0.2324959914484233,
      "grad_norm": 0.8164053559303284,
      "learning_rate": 0.00019973337086629753,
      "loss": 2.1292,
      "step": 870
    },
    {
      "epoch": 0.23516835916622128,
      "grad_norm": 0.7342932820320129,
      "learning_rate": 0.00019972720903873017,
      "loss": 2.0944,
      "step": 880
    },
    {
      "epoch": 0.23784072688401925,
      "grad_norm": 0.8686122298240662,
      "learning_rate": 0.00019972097691917603,
      "loss": 2.1152,
      "step": 890
    },
    {
      "epoch": 0.24051309460181722,
      "grad_norm": 0.8342360258102417,
      "learning_rate": 0.0001997146745120278,
      "loss": 2.0581,
      "step": 900
    },
    {
      "epoch": 0.2431854623196152,
      "grad_norm": 0.7059487104415894,
      "learning_rate": 0.00019970830182172768,
      "loss": 2.0178,
      "step": 910
    },
    {
      "epoch": 0.24585783003741316,
      "grad_norm": 0.7941276431083679,
      "learning_rate": 0.0001997018588527674,
      "loss": 2.0689,
      "step": 920
    },
    {
      "epoch": 0.24853019775521112,
      "grad_norm": 0.7752018570899963,
      "learning_rate": 0.00019969534560968823,
      "loss": 2.0976,
      "step": 930
    },
    {
      "epoch": 0.25120256547300907,
      "grad_norm": 0.7754221558570862,
      "learning_rate": 0.000199688762097081,
      "loss": 2.0676,
      "step": 940
    },
    {
      "epoch": 0.25387493319080706,
      "grad_norm": 0.7400916814804077,
      "learning_rate": 0.00019968210831958604,
      "loss": 2.0806,
      "step": 950
    },
    {
      "epoch": 0.256547300908605,
      "grad_norm": 0.8396859765052795,
      "learning_rate": 0.00019967538428189321,
      "loss": 2.0666,
      "step": 960
    },
    {
      "epoch": 0.259219668626403,
      "grad_norm": 0.821546196937561,
      "learning_rate": 0.00019966858998874193,
      "loss": 2.0362,
      "step": 970
    },
    {
      "epoch": 0.26189203634420094,
      "grad_norm": 0.7961469888687134,
      "learning_rate": 0.00019966172544492108,
      "loss": 2.0702,
      "step": 980
    },
    {
      "epoch": 0.26456440406199894,
      "grad_norm": 0.8565064072608948,
      "learning_rate": 0.00019965479065526908,
      "loss": 2.0715,
      "step": 990
    },
    {
      "epoch": 0.2672367717797969,
      "grad_norm": 0.9843941926956177,
      "learning_rate": 0.00019964778562467386,
      "loss": 2.1114,
      "step": 1000
    },
    {
      "epoch": 0.2699091394975949,
      "grad_norm": 0.8522785305976868,
      "learning_rate": 0.0001996407103580729,
      "loss": 2.102,
      "step": 1010
    },
    {
      "epoch": 0.2725815072153928,
      "grad_norm": 0.8632455468177795,
      "learning_rate": 0.00019963356486045312,
      "loss": 2.075,
      "step": 1020
    },
    {
      "epoch": 0.2752538749331908,
      "grad_norm": 0.8559175729751587,
      "learning_rate": 0.000199626349136851,
      "loss": 2.0732,
      "step": 1030
    },
    {
      "epoch": 0.27792624265098875,
      "grad_norm": 0.8057096600532532,
      "learning_rate": 0.00019961906319235242,
      "loss": 2.0666,
      "step": 1040
    },
    {
      "epoch": 0.28059861036878675,
      "grad_norm": 0.766559362411499,
      "learning_rate": 0.0001996117070320929,
      "loss": 2.0428,
      "step": 1050
    },
    {
      "epoch": 0.2832709780865847,
      "grad_norm": 0.8320493102073669,
      "learning_rate": 0.00019960428066125735,
      "loss": 2.1302,
      "step": 1060
    },
    {
      "epoch": 0.2859433458043827,
      "grad_norm": 0.8726279735565186,
      "learning_rate": 0.0001995967840850802,
      "loss": 2.1308,
      "step": 1070
    },
    {
      "epoch": 0.28861571352218063,
      "grad_norm": 0.8588752150535583,
      "learning_rate": 0.0001995892173088453,
      "loss": 2.0735,
      "step": 1080
    },
    {
      "epoch": 0.2912880812399786,
      "grad_norm": 0.8037392497062683,
      "learning_rate": 0.00019958158033788612,
      "loss": 2.0189,
      "step": 1090
    },
    {
      "epoch": 0.29396044895777657,
      "grad_norm": 0.8673548698425293,
      "learning_rate": 0.0001995738731775855,
      "loss": 2.0375,
      "step": 1100
    },
    {
      "epoch": 0.29663281667557456,
      "grad_norm": 0.7821346521377563,
      "learning_rate": 0.00019956609583337572,
      "loss": 2.074,
      "step": 1110
    },
    {
      "epoch": 0.2993051843933725,
      "grad_norm": 0.908275306224823,
      "learning_rate": 0.00019955824831073863,
      "loss": 2.1195,
      "step": 1120
    },
    {
      "epoch": 0.3019775521111705,
      "grad_norm": 1.0341707468032837,
      "learning_rate": 0.0001995503306152055,
      "loss": 2.0508,
      "step": 1130
    },
    {
      "epoch": 0.30464991982896844,
      "grad_norm": 0.8817709684371948,
      "learning_rate": 0.00019954234275235705,
      "loss": 2.0335,
      "step": 1140
    },
    {
      "epoch": 0.30732228754676644,
      "grad_norm": 0.8848737478256226,
      "learning_rate": 0.00019953428472782348,
      "loss": 2.0893,
      "step": 1150
    },
    {
      "epoch": 0.3099946552645644,
      "grad_norm": 0.834466814994812,
      "learning_rate": 0.0001995261565472844,
      "loss": 2.0991,
      "step": 1160
    },
    {
      "epoch": 0.3126670229823624,
      "grad_norm": 0.8102088570594788,
      "learning_rate": 0.00019951795821646892,
      "loss": 2.0823,
      "step": 1170
    },
    {
      "epoch": 0.3153393907001603,
      "grad_norm": 0.8358621001243591,
      "learning_rate": 0.0001995096897411556,
      "loss": 2.0049,
      "step": 1180
    },
    {
      "epoch": 0.3180117584179583,
      "grad_norm": 0.8446478843688965,
      "learning_rate": 0.00019950135112717235,
      "loss": 2.1207,
      "step": 1190
    },
    {
      "epoch": 0.32068412613575625,
      "grad_norm": 0.8990554809570312,
      "learning_rate": 0.0001994929423803966,
      "loss": 2.043,
      "step": 1200
    },
    {
      "epoch": 0.32335649385355425,
      "grad_norm": 0.8363642692565918,
      "learning_rate": 0.0001994844635067552,
      "loss": 2.0611,
      "step": 1210
    },
    {
      "epoch": 0.3260288615713522,
      "grad_norm": 0.7991921901702881,
      "learning_rate": 0.00019947591451222448,
      "loss": 2.0634,
      "step": 1220
    },
    {
      "epoch": 0.3287012292891502,
      "grad_norm": 0.8177972435951233,
      "learning_rate": 0.00019946729540283,
      "loss": 2.0359,
      "step": 1230
    },
    {
      "epoch": 0.33137359700694813,
      "grad_norm": 0.8667873740196228,
      "learning_rate": 0.00019945860618464703,
      "loss": 1.9758,
      "step": 1240
    },
    {
      "epoch": 0.3340459647247461,
      "grad_norm": 0.8804141283035278,
      "learning_rate": 0.00019944984686380001,
      "loss": 2.0795,
      "step": 1250
    },
    {
      "epoch": 0.3367183324425441,
      "grad_norm": 0.8883786797523499,
      "learning_rate": 0.00019944101744646288,
      "loss": 2.0324,
      "step": 1260
    },
    {
      "epoch": 0.33939070016034206,
      "grad_norm": 0.8005781173706055,
      "learning_rate": 0.00019943211793885902,
      "loss": 2.0907,
      "step": 1270
    },
    {
      "epoch": 0.34206306787814006,
      "grad_norm": 0.7767745852470398,
      "learning_rate": 0.00019942314834726118,
      "loss": 2.0345,
      "step": 1280
    },
    {
      "epoch": 0.344735435595938,
      "grad_norm": 0.8179928064346313,
      "learning_rate": 0.00019941410867799152,
      "loss": 2.0417,
      "step": 1290
    },
    {
      "epoch": 0.347407803313736,
      "grad_norm": 0.782202959060669,
      "learning_rate": 0.00019940499893742155,
      "loss": 2.0555,
      "step": 1300
    },
    {
      "epoch": 0.35008017103153394,
      "grad_norm": 0.7710146307945251,
      "learning_rate": 0.0001993958191319722,
      "loss": 2.0209,
      "step": 1310
    },
    {
      "epoch": 0.35275253874933193,
      "grad_norm": 0.8291559815406799,
      "learning_rate": 0.00019938656926811384,
      "loss": 2.143,
      "step": 1320
    },
    {
      "epoch": 0.3554249064671299,
      "grad_norm": 0.8180106282234192,
      "learning_rate": 0.0001993772493523661,
      "loss": 2.0363,
      "step": 1330
    },
    {
      "epoch": 0.35809727418492787,
      "grad_norm": 0.8292155265808105,
      "learning_rate": 0.0001993678593912981,
      "loss": 2.0649,
      "step": 1340
    },
    {
      "epoch": 0.3607696419027258,
      "grad_norm": 0.8484436869621277,
      "learning_rate": 0.00019935839939152832,
      "loss": 2.0399,
      "step": 1350
    },
    {
      "epoch": 0.3634420096205238,
      "grad_norm": 0.8078696727752686,
      "learning_rate": 0.0001993488693597245,
      "loss": 2.0051,
      "step": 1360
    },
    {
      "epoch": 0.36611437733832175,
      "grad_norm": 0.8645321726799011,
      "learning_rate": 0.00019933926930260385,
      "loss": 2.1165,
      "step": 1370
    },
    {
      "epoch": 0.36878674505611975,
      "grad_norm": 0.8084302544593811,
      "learning_rate": 0.00019932959922693287,
      "loss": 2.0621,
      "step": 1380
    },
    {
      "epoch": 0.3714591127739177,
      "grad_norm": 0.8464387655258179,
      "learning_rate": 0.00019931985913952747,
      "loss": 2.0896,
      "step": 1390
    },
    {
      "epoch": 0.3741314804917157,
      "grad_norm": 1.1601299047470093,
      "learning_rate": 0.00019931004904725285,
      "loss": 2.0684,
      "step": 1400
    },
    {
      "epoch": 0.3768038482095136,
      "grad_norm": 0.8167917728424072,
      "learning_rate": 0.0001993001689570236,
      "loss": 2.0864,
      "step": 1410
    },
    {
      "epoch": 0.3794762159273116,
      "grad_norm": 0.8336923718452454,
      "learning_rate": 0.00019929021887580368,
      "loss": 2.1179,
      "step": 1420
    },
    {
      "epoch": 0.38214858364510956,
      "grad_norm": 0.8231195211410522,
      "learning_rate": 0.00019928019881060625,
      "loss": 2.0374,
      "step": 1430
    },
    {
      "epoch": 0.38482095136290756,
      "grad_norm": 0.9005337953567505,
      "learning_rate": 0.00019927010876849387,
      "loss": 2.0469,
      "step": 1440
    },
    {
      "epoch": 0.3874933190807055,
      "grad_norm": 0.7575165629386902,
      "learning_rate": 0.0001992599487565785,
      "loss": 2.0316,
      "step": 1450
    },
    {
      "epoch": 0.3901656867985035,
      "grad_norm": 0.84959477186203,
      "learning_rate": 0.00019924971878202135,
      "loss": 2.064,
      "step": 1460
    },
    {
      "epoch": 0.39283805451630144,
      "grad_norm": 0.9098086357116699,
      "learning_rate": 0.00019923941885203287,
      "loss": 2.1038,
      "step": 1470
    },
    {
      "epoch": 0.39551042223409943,
      "grad_norm": 0.8135308623313904,
      "learning_rate": 0.00019922904897387293,
      "loss": 2.0406,
      "step": 1480
    },
    {
      "epoch": 0.3981827899518974,
      "grad_norm": 0.822067141532898,
      "learning_rate": 0.0001992186091548507,
      "loss": 2.0401,
      "step": 1490
    },
    {
      "epoch": 0.40085515766969537,
      "grad_norm": 0.8210283517837524,
      "learning_rate": 0.00019920809940232454,
      "loss": 2.0821,
      "step": 1500
    },
    {
      "epoch": 0.4035275253874933,
      "grad_norm": 1.090965747833252,
      "learning_rate": 0.00019919751972370222,
      "loss": 2.0746,
      "step": 1510
    },
    {
      "epoch": 0.4061998931052913,
      "grad_norm": 0.808088481426239,
      "learning_rate": 0.00019918687012644072,
      "loss": 2.0348,
      "step": 1520
    },
    {
      "epoch": 0.40887226082308925,
      "grad_norm": 0.8605575561523438,
      "learning_rate": 0.00019917615061804637,
      "loss": 2.0399,
      "step": 1530
    },
    {
      "epoch": 0.41154462854088725,
      "grad_norm": 0.8858651518821716,
      "learning_rate": 0.0001991653612060747,
      "loss": 2.0639,
      "step": 1540
    },
    {
      "epoch": 0.4142169962586852,
      "grad_norm": 0.847783625125885,
      "learning_rate": 0.00019915450189813053,
      "loss": 2.0702,
      "step": 1550
    },
    {
      "epoch": 0.4168893639764832,
      "grad_norm": 0.8495000004768372,
      "learning_rate": 0.000199143572701868,
      "loss": 2.0471,
      "step": 1560
    },
    {
      "epoch": 0.4195617316942811,
      "grad_norm": 0.8186051845550537,
      "learning_rate": 0.00019913257362499047,
      "loss": 2.0661,
      "step": 1570
    },
    {
      "epoch": 0.4222340994120791,
      "grad_norm": 0.8053857684135437,
      "learning_rate": 0.00019912150467525056,
      "loss": 2.0013,
      "step": 1580
    },
    {
      "epoch": 0.42490646712987706,
      "grad_norm": 0.8610458374023438,
      "learning_rate": 0.0001991103658604501,
      "loss": 2.0594,
      "step": 1590
    },
    {
      "epoch": 0.42757883484767506,
      "grad_norm": 0.9696066975593567,
      "learning_rate": 0.00019909915718844024,
      "loss": 2.0695,
      "step": 1600
    },
    {
      "epoch": 0.430251202565473,
      "grad_norm": 0.7572137713432312,
      "learning_rate": 0.00019908787866712132,
      "loss": 2.0141,
      "step": 1610
    },
    {
      "epoch": 0.432923570283271,
      "grad_norm": 0.809211254119873,
      "learning_rate": 0.00019907653030444293,
      "loss": 2.051,
      "step": 1620
    },
    {
      "epoch": 0.43559593800106894,
      "grad_norm": 0.8706380724906921,
      "learning_rate": 0.00019906511210840386,
      "loss": 2.076,
      "step": 1630
    },
    {
      "epoch": 0.43826830571886694,
      "grad_norm": 0.8855727910995483,
      "learning_rate": 0.00019905362408705214,
      "loss": 2.0107,
      "step": 1640
    },
    {
      "epoch": 0.4409406734366649,
      "grad_norm": 0.82815021276474,
      "learning_rate": 0.00019904206624848507,
      "loss": 1.9842,
      "step": 1650
    },
    {
      "epoch": 0.4436130411544629,
      "grad_norm": 0.8440461754798889,
      "learning_rate": 0.00019903043860084905,
      "loss": 2.0796,
      "step": 1660
    },
    {
      "epoch": 0.4462854088722608,
      "grad_norm": 0.963351309299469,
      "learning_rate": 0.00019901874115233975,
      "loss": 2.0181,
      "step": 1670
    },
    {
      "epoch": 0.4489577765900588,
      "grad_norm": 0.8633173704147339,
      "learning_rate": 0.000199006973911202,
      "loss": 2.0045,
      "step": 1680
    },
    {
      "epoch": 0.45163014430785675,
      "grad_norm": 0.8765243291854858,
      "learning_rate": 0.00019899513688572998,
      "loss": 2.1057,
      "step": 1690
    },
    {
      "epoch": 0.45430251202565475,
      "grad_norm": 0.8254087567329407,
      "learning_rate": 0.00019898323008426679,
      "loss": 2.0332,
      "step": 1700
    },
    {
      "epoch": 0.4569748797434527,
      "grad_norm": 0.8703109622001648,
      "learning_rate": 0.00019897125351520488,
      "loss": 2.0668,
      "step": 1710
    },
    {
      "epoch": 0.4596472474612507,
      "grad_norm": 0.9613226652145386,
      "learning_rate": 0.0001989592071869859,
      "loss": 2.0436,
      "step": 1720
    },
    {
      "epoch": 0.4623196151790486,
      "grad_norm": 0.844394862651825,
      "learning_rate": 0.00019894709110810057,
      "loss": 2.0382,
      "step": 1730
    },
    {
      "epoch": 0.4649919828968466,
      "grad_norm": 0.8338039517402649,
      "learning_rate": 0.00019893490528708884,
      "loss": 2.0148,
      "step": 1740
    },
    {
      "epoch": 0.46766435061464456,
      "grad_norm": 0.9704885482788086,
      "learning_rate": 0.0001989226497325398,
      "loss": 2.0642,
      "step": 1750
    },
    {
      "epoch": 0.47033671833244256,
      "grad_norm": 0.8037359118461609,
      "learning_rate": 0.00019891032445309165,
      "loss": 2.0764,
      "step": 1760
    },
    {
      "epoch": 0.4730090860502405,
      "grad_norm": 0.8697875142097473,
      "learning_rate": 0.00019889792945743183,
      "loss": 2.0119,
      "step": 1770
    },
    {
      "epoch": 0.4756814537680385,
      "grad_norm": 0.9716849327087402,
      "learning_rate": 0.0001988854647542968,
      "loss": 2.0506,
      "step": 1780
    },
    {
      "epoch": 0.47835382148583644,
      "grad_norm": 0.8439099788665771,
      "learning_rate": 0.00019887293035247226,
      "loss": 1.9582,
      "step": 1790
    },
    {
      "epoch": 0.48102618920363444,
      "grad_norm": 0.9845560193061829,
      "learning_rate": 0.00019886032626079296,
      "loss": 1.9794,
      "step": 1800
    },
    {
      "epoch": 0.4836985569214324,
      "grad_norm": 0.8697616457939148,
      "learning_rate": 0.00019884765248814282,
      "loss": 2.0714,
      "step": 1810
    },
    {
      "epoch": 0.4863709246392304,
      "grad_norm": 0.8646136522293091,
      "learning_rate": 0.00019883490904345485,
      "loss": 1.9833,
      "step": 1820
    },
    {
      "epoch": 0.4890432923570283,
      "grad_norm": 0.7895427942276001,
      "learning_rate": 0.00019882209593571117,
      "loss": 1.9911,
      "step": 1830
    },
    {
      "epoch": 0.4917156600748263,
      "grad_norm": 1.0419542789459229,
      "learning_rate": 0.000198809213173943,
      "loss": 2.0412,
      "step": 1840
    },
    {
      "epoch": 0.49438802779262425,
      "grad_norm": 0.8819153308868408,
      "learning_rate": 0.0001987962607672307,
      "loss": 2.0194,
      "step": 1850
    },
    {
      "epoch": 0.49706039551042225,
      "grad_norm": 0.8423511981964111,
      "learning_rate": 0.00019878323872470364,
      "loss": 2.0493,
      "step": 1860
    },
    {
      "epoch": 0.4997327632282202,
      "grad_norm": 0.9035742878913879,
      "learning_rate": 0.00019877014705554027,
      "loss": 2.0242,
      "step": 1870
    },
    {
      "epoch": 0.5024051309460181,
      "grad_norm": 0.7854070663452148,
      "learning_rate": 0.0001987569857689682,
      "loss": 2.016,
      "step": 1880
    },
    {
      "epoch": 0.5050774986638161,
      "grad_norm": 1.0013773441314697,
      "learning_rate": 0.00019874375487426412,
      "loss": 2.0343,
      "step": 1890
    },
    {
      "epoch": 0.5077498663816141,
      "grad_norm": 0.8524296283721924,
      "learning_rate": 0.00019873045438075366,
      "loss": 2.0316,
      "step": 1900
    },
    {
      "epoch": 0.5104222340994121,
      "grad_norm": 0.8658786416053772,
      "learning_rate": 0.00019871708429781157,
      "loss": 2.0874,
      "step": 1910
    },
    {
      "epoch": 0.51309460181721,
      "grad_norm": 0.8632771968841553,
      "learning_rate": 0.00019870364463486168,
      "loss": 1.9716,
      "step": 1920
    },
    {
      "epoch": 0.515766969535008,
      "grad_norm": 0.8511568307876587,
      "learning_rate": 0.00019869013540137683,
      "loss": 1.9946,
      "step": 1930
    },
    {
      "epoch": 0.518439337252806,
      "grad_norm": 0.8525965809822083,
      "learning_rate": 0.00019867655660687895,
      "loss": 1.9906,
      "step": 1940
    },
    {
      "epoch": 0.521111704970604,
      "grad_norm": 0.8269341588020325,
      "learning_rate": 0.00019866290826093885,
      "loss": 2.0097,
      "step": 1950
    },
    {
      "epoch": 0.5237840726884019,
      "grad_norm": 0.8655617833137512,
      "learning_rate": 0.00019864919037317652,
      "loss": 2.1078,
      "step": 1960
    },
    {
      "epoch": 0.5264564404061999,
      "grad_norm": 0.8016558289527893,
      "learning_rate": 0.00019863540295326094,
      "loss": 1.9657,
      "step": 1970
    },
    {
      "epoch": 0.5291288081239979,
      "grad_norm": 0.8438629508018494,
      "learning_rate": 0.00019862154601091005,
      "loss": 2.0081,
      "step": 1980
    },
    {
      "epoch": 0.5318011758417959,
      "grad_norm": 0.8288686275482178,
      "learning_rate": 0.0001986076195558908,
      "loss": 2.0081,
      "step": 1990
    },
    {
      "epoch": 0.5344735435595938,
      "grad_norm": 0.9016776084899902,
      "learning_rate": 0.00019859362359801917,
      "loss": 2.0128,
      "step": 2000
    },
    {
      "epoch": 0.5371459112773918,
      "grad_norm": 0.9300397634506226,
      "learning_rate": 0.0001985795581471601,
      "loss": 2.0566,
      "step": 2010
    },
    {
      "epoch": 0.5398182789951897,
      "grad_norm": 1.0317420959472656,
      "learning_rate": 0.0001985654232132275,
      "loss": 1.9624,
      "step": 2020
    },
    {
      "epoch": 0.5424906467129877,
      "grad_norm": 1.061338186264038,
      "learning_rate": 0.00019855121880618432,
      "loss": 2.1136,
      "step": 2030
    },
    {
      "epoch": 0.5451630144307856,
      "grad_norm": 0.9200817346572876,
      "learning_rate": 0.00019853694493604236,
      "loss": 2.0427,
      "step": 2040
    },
    {
      "epoch": 0.5478353821485836,
      "grad_norm": 0.925785481929779,
      "learning_rate": 0.00019852260161286252,
      "loss": 2.0383,
      "step": 2050
    },
    {
      "epoch": 0.5505077498663816,
      "grad_norm": 0.8511183857917786,
      "learning_rate": 0.00019850818884675452,
      "loss": 2.0463,
      "step": 2060
    },
    {
      "epoch": 0.5531801175841796,
      "grad_norm": 0.9213443398475647,
      "learning_rate": 0.00019849370664787717,
      "loss": 1.9861,
      "step": 2070
    },
    {
      "epoch": 0.5558524853019775,
      "grad_norm": 0.9268140196800232,
      "learning_rate": 0.00019847915502643808,
      "loss": 2.0883,
      "step": 2080
    },
    {
      "epoch": 0.5585248530197755,
      "grad_norm": 0.9962113499641418,
      "learning_rate": 0.00019846453399269387,
      "loss": 2.0392,
      "step": 2090
    },
    {
      "epoch": 0.5611972207375735,
      "grad_norm": 0.9202669262886047,
      "learning_rate": 0.00019844984355695008,
      "loss": 2.0181,
      "step": 2100
    },
    {
      "epoch": 0.5638695884553715,
      "grad_norm": 0.8881505727767944,
      "learning_rate": 0.00019843508372956116,
      "loss": 1.9855,
      "step": 2110
    },
    {
      "epoch": 0.5665419561731694,
      "grad_norm": 0.8822938799858093,
      "learning_rate": 0.00019842025452093044,
      "loss": 2.0401,
      "step": 2120
    },
    {
      "epoch": 0.5692143238909674,
      "grad_norm": 0.8415065407752991,
      "learning_rate": 0.00019840535594151023,
      "loss": 2.0081,
      "step": 2130
    },
    {
      "epoch": 0.5718866916087654,
      "grad_norm": 0.8376311659812927,
      "learning_rate": 0.0001983903880018016,
      "loss": 2.0858,
      "step": 2140
    },
    {
      "epoch": 0.5745590593265634,
      "grad_norm": 0.889597475528717,
      "learning_rate": 0.00019837535071235462,
      "loss": 2.0322,
      "step": 2150
    },
    {
      "epoch": 0.5772314270443613,
      "grad_norm": 0.8720742464065552,
      "learning_rate": 0.00019836024408376824,
      "loss": 1.9494,
      "step": 2160
    },
    {
      "epoch": 0.5799037947621593,
      "grad_norm": 0.8727856278419495,
      "learning_rate": 0.00019834506812669027,
      "loss": 2.073,
      "step": 2170
    },
    {
      "epoch": 0.5825761624799572,
      "grad_norm": 0.8532377481460571,
      "learning_rate": 0.00019832982285181733,
      "loss": 1.9606,
      "step": 2180
    },
    {
      "epoch": 0.5852485301977552,
      "grad_norm": 0.8545679450035095,
      "learning_rate": 0.00019831450826989495,
      "loss": 1.9868,
      "step": 2190
    },
    {
      "epoch": 0.5879208979155531,
      "grad_norm": 0.9709376096725464,
      "learning_rate": 0.00019829912439171753,
      "loss": 2.0459,
      "step": 2200
    },
    {
      "epoch": 0.5905932656333511,
      "grad_norm": 0.9187166094779968,
      "learning_rate": 0.00019828367122812823,
      "loss": 2.0411,
      "step": 2210
    },
    {
      "epoch": 0.5932656333511491,
      "grad_norm": 0.8447383642196655,
      "learning_rate": 0.0001982681487900191,
      "loss": 2.0448,
      "step": 2220
    },
    {
      "epoch": 0.5959380010689471,
      "grad_norm": 0.8754927515983582,
      "learning_rate": 0.00019825255708833108,
      "loss": 1.987,
      "step": 2230
    },
    {
      "epoch": 0.598610368786745,
      "grad_norm": 0.8265763521194458,
      "learning_rate": 0.00019823689613405383,
      "loss": 1.9984,
      "step": 2240
    },
    {
      "epoch": 0.601282736504543,
      "grad_norm": 0.948645830154419,
      "learning_rate": 0.00019822116593822586,
      "loss": 2.0691,
      "step": 2250
    },
    {
      "epoch": 0.603955104222341,
      "grad_norm": 0.8917912244796753,
      "learning_rate": 0.00019820536651193445,
      "loss": 2.0276,
      "step": 2260
    },
    {
      "epoch": 0.606627471940139,
      "grad_norm": 0.8749749064445496,
      "learning_rate": 0.00019818949786631573,
      "loss": 1.991,
      "step": 2270
    },
    {
      "epoch": 0.6092998396579369,
      "grad_norm": 0.8689747452735901,
      "learning_rate": 0.00019817356001255461,
      "loss": 1.979,
      "step": 2280
    },
    {
      "epoch": 0.6119722073757349,
      "grad_norm": 0.8354358673095703,
      "learning_rate": 0.0001981575529618848,
      "loss": 2.0286,
      "step": 2290
    },
    {
      "epoch": 0.6146445750935329,
      "grad_norm": 0.8581618070602417,
      "learning_rate": 0.0001981414767255887,
      "loss": 2.0085,
      "step": 2300
    },
    {
      "epoch": 0.6173169428113309,
      "grad_norm": 0.8814964890480042,
      "learning_rate": 0.0001981253313149975,
      "loss": 1.9892,
      "step": 2310
    },
    {
      "epoch": 0.6199893105291288,
      "grad_norm": 0.8790181875228882,
      "learning_rate": 0.00019810911674149125,
      "loss": 1.9553,
      "step": 2320
    },
    {
      "epoch": 0.6226616782469268,
      "grad_norm": 0.8602410554885864,
      "learning_rate": 0.00019809283301649866,
      "loss": 1.9178,
      "step": 2330
    },
    {
      "epoch": 0.6253340459647247,
      "grad_norm": 0.8457080125808716,
      "learning_rate": 0.0001980764801514971,
      "loss": 2.0387,
      "step": 2340
    },
    {
      "epoch": 0.6280064136825227,
      "grad_norm": 0.868516206741333,
      "learning_rate": 0.0001980600581580129,
      "loss": 2.0005,
      "step": 2350
    },
    {
      "epoch": 0.6306787814003206,
      "grad_norm": 0.9111327528953552,
      "learning_rate": 0.00019804356704762087,
      "loss": 2.0489,
      "step": 2360
    },
    {
      "epoch": 0.6333511491181186,
      "grad_norm": 0.8195019364356995,
      "learning_rate": 0.00019802700683194474,
      "loss": 1.9894,
      "step": 2370
    },
    {
      "epoch": 0.6360235168359166,
      "grad_norm": 0.914009153842926,
      "learning_rate": 0.00019801037752265675,
      "loss": 1.9797,
      "step": 2380
    },
    {
      "epoch": 0.6386958845537146,
      "grad_norm": 0.8143549561500549,
      "learning_rate": 0.00019799367913147806,
      "loss": 1.9689,
      "step": 2390
    },
    {
      "epoch": 0.6413682522715125,
      "grad_norm": 0.9965906143188477,
      "learning_rate": 0.00019797691167017833,
      "loss": 1.9581,
      "step": 2400
    },
    {
      "epoch": 0.6440406199893105,
      "grad_norm": 0.8981281518936157,
      "learning_rate": 0.000197960075150576,
      "loss": 2.0586,
      "step": 2410
    },
    {
      "epoch": 0.6467129877071085,
      "grad_norm": 0.911881685256958,
      "learning_rate": 0.00019794316958453818,
      "loss": 2.0806,
      "step": 2420
    },
    {
      "epoch": 0.6493853554249065,
      "grad_norm": 0.9170183539390564,
      "learning_rate": 0.00019792619498398064,
      "loss": 2.0259,
      "step": 2430
    },
    {
      "epoch": 0.6520577231427044,
      "grad_norm": 0.8699791431427002,
      "learning_rate": 0.00019790915136086776,
      "loss": 1.9954,
      "step": 2440
    },
    {
      "epoch": 0.6547300908605024,
      "grad_norm": 0.8370604515075684,
      "learning_rate": 0.00019789203872721265,
      "loss": 1.9813,
      "step": 2450
    },
    {
      "epoch": 0.6574024585783004,
      "grad_norm": 0.9251917600631714,
      "learning_rate": 0.000197874857095077,
      "loss": 1.9787,
      "step": 2460
    },
    {
      "epoch": 0.6600748262960984,
      "grad_norm": 0.8766528367996216,
      "learning_rate": 0.00019785760647657118,
      "loss": 2.037,
      "step": 2470
    },
    {
      "epoch": 0.6627471940138963,
      "grad_norm": 0.9043455719947815,
      "learning_rate": 0.00019784028688385415,
      "loss": 1.981,
      "step": 2480
    },
    {
      "epoch": 0.6654195617316943,
      "grad_norm": 0.9129969477653503,
      "learning_rate": 0.00019782289832913347,
      "loss": 1.9867,
      "step": 2490
    },
    {
      "epoch": 0.6680919294494923,
      "grad_norm": 0.9942403435707092,
      "learning_rate": 0.0001978054408246654,
      "loss": 2.057,
      "step": 2500
    },
    {
      "epoch": 0.6707642971672902,
      "grad_norm": 0.8948236107826233,
      "learning_rate": 0.00019778791438275465,
      "loss": 2.0288,
      "step": 2510
    },
    {
      "epoch": 0.6734366648850882,
      "grad_norm": 0.8486409187316895,
      "learning_rate": 0.0001977703190157547,
      "loss": 1.9786,
      "step": 2520
    },
    {
      "epoch": 0.6761090326028861,
      "grad_norm": 0.8692457675933838,
      "learning_rate": 0.00019775265473606743,
      "loss": 1.995,
      "step": 2530
    },
    {
      "epoch": 0.6787814003206841,
      "grad_norm": 0.8827071785926819,
      "learning_rate": 0.00019773492155614338,
      "loss": 2.0023,
      "step": 2540
    },
    {
      "epoch": 0.6814537680384821,
      "grad_norm": 0.9078000783920288,
      "learning_rate": 0.00019771711948848172,
      "loss": 1.9479,
      "step": 2550
    },
    {
      "epoch": 0.6841261357562801,
      "grad_norm": 0.8222280144691467,
      "learning_rate": 0.00019769924854563006,
      "loss": 1.9448,
      "step": 2560
    },
    {
      "epoch": 0.686798503474078,
      "grad_norm": 0.8683822154998779,
      "learning_rate": 0.00019768130874018463,
      "loss": 2.0054,
      "step": 2570
    },
    {
      "epoch": 0.689470871191876,
      "grad_norm": 0.9791079163551331,
      "learning_rate": 0.0001976633000847901,
      "loss": 2.0422,
      "step": 2580
    },
    {
      "epoch": 0.692143238909674,
      "grad_norm": 0.860660970211029,
      "learning_rate": 0.0001976452225921398,
      "loss": 2.0139,
      "step": 2590
    },
    {
      "epoch": 0.694815606627472,
      "grad_norm": 1.0318732261657715,
      "learning_rate": 0.0001976270762749755,
      "loss": 2.0283,
      "step": 2600
    },
    {
      "epoch": 0.6974879743452699,
      "grad_norm": 0.9548642039299011,
      "learning_rate": 0.00019760886114608749,
      "loss": 2.0183,
      "step": 2610
    },
    {
      "epoch": 0.7001603420630679,
      "grad_norm": 0.885148823261261,
      "learning_rate": 0.0001975905772183146,
      "loss": 2.006,
      "step": 2620
    },
    {
      "epoch": 0.7028327097808659,
      "grad_norm": 0.9038057327270508,
      "learning_rate": 0.00019757222450454408,
      "loss": 1.999,
      "step": 2630
    },
    {
      "epoch": 0.7055050774986639,
      "grad_norm": 0.889042317867279,
      "learning_rate": 0.00019755380301771176,
      "loss": 1.9462,
      "step": 2640
    },
    {
      "epoch": 0.7081774452164618,
      "grad_norm": 0.8829699754714966,
      "learning_rate": 0.0001975353127708018,
      "loss": 1.9913,
      "step": 2650
    },
    {
      "epoch": 0.7108498129342598,
      "grad_norm": 0.8977982401847839,
      "learning_rate": 0.00019751675377684702,
      "loss": 2.0038,
      "step": 2660
    },
    {
      "epoch": 0.7135221806520577,
      "grad_norm": 0.9305534958839417,
      "learning_rate": 0.0001974981260489285,
      "loss": 2.0484,
      "step": 2670
    },
    {
      "epoch": 0.7161945483698557,
      "grad_norm": 0.9134384989738464,
      "learning_rate": 0.00019747942960017588,
      "loss": 2.028,
      "step": 2680
    },
    {
      "epoch": 0.7188669160876536,
      "grad_norm": 0.9414751529693604,
      "learning_rate": 0.00019746066444376728,
      "loss": 2.0499,
      "step": 2690
    },
    {
      "epoch": 0.7215392838054516,
      "grad_norm": 0.9050728678703308,
      "learning_rate": 0.0001974418305929291,
      "loss": 2.0301,
      "step": 2700
    },
    {
      "epoch": 0.7242116515232496,
      "grad_norm": 0.8627575635910034,
      "learning_rate": 0.00019742292806093628,
      "loss": 1.9909,
      "step": 2710
    },
    {
      "epoch": 0.7268840192410476,
      "grad_norm": 0.9992572069168091,
      "learning_rate": 0.0001974039568611121,
      "loss": 2.0404,
      "step": 2720
    },
    {
      "epoch": 0.7295563869588455,
      "grad_norm": 0.8972439765930176,
      "learning_rate": 0.00019738491700682832,
      "loss": 1.9952,
      "step": 2730
    },
    {
      "epoch": 0.7322287546766435,
      "grad_norm": 0.8587448596954346,
      "learning_rate": 0.00019736580851150497,
      "loss": 1.9673,
      "step": 2740
    },
    {
      "epoch": 0.7349011223944415,
      "grad_norm": 0.883135199546814,
      "learning_rate": 0.00019734663138861057,
      "loss": 2.0555,
      "step": 2750
    },
    {
      "epoch": 0.7375734901122395,
      "grad_norm": 0.8620017170906067,
      "learning_rate": 0.00019732738565166202,
      "loss": 1.9871,
      "step": 2760
    },
    {
      "epoch": 0.7402458578300374,
      "grad_norm": 0.933148205280304,
      "learning_rate": 0.00019730807131422448,
      "loss": 1.9907,
      "step": 2770
    },
    {
      "epoch": 0.7429182255478354,
      "grad_norm": 0.9321724772453308,
      "learning_rate": 0.00019728868838991153,
      "loss": 1.9875,
      "step": 2780
    },
    {
      "epoch": 0.7455905932656334,
      "grad_norm": 1.0499948263168335,
      "learning_rate": 0.00019726923689238505,
      "loss": 2.0677,
      "step": 2790
    },
    {
      "epoch": 0.7482629609834314,
      "grad_norm": 0.9769105911254883,
      "learning_rate": 0.00019724971683535535,
      "loss": 2.0078,
      "step": 2800
    },
    {
      "epoch": 0.7509353287012293,
      "grad_norm": 0.9284023642539978,
      "learning_rate": 0.00019723012823258095,
      "loss": 1.9492,
      "step": 2810
    },
    {
      "epoch": 0.7536076964190273,
      "grad_norm": 0.8626641035079956,
      "learning_rate": 0.00019721047109786874,
      "loss": 1.9734,
      "step": 2820
    },
    {
      "epoch": 0.7562800641368252,
      "grad_norm": 0.8522788286209106,
      "learning_rate": 0.00019719074544507391,
      "loss": 1.9692,
      "step": 2830
    },
    {
      "epoch": 0.7589524318546232,
      "grad_norm": 0.8608167767524719,
      "learning_rate": 0.00019717095128809998,
      "loss": 1.9476,
      "step": 2840
    },
    {
      "epoch": 0.7616247995724211,
      "grad_norm": 0.9356365203857422,
      "learning_rate": 0.00019715108864089861,
      "loss": 1.9983,
      "step": 2850
    },
    {
      "epoch": 0.7642971672902191,
      "grad_norm": 0.8763227462768555,
      "learning_rate": 0.00019713115751746995,
      "loss": 1.9729,
      "step": 2860
    },
    {
      "epoch": 0.7669695350080171,
      "grad_norm": 0.9147051572799683,
      "learning_rate": 0.00019711115793186225,
      "loss": 2.009,
      "step": 2870
    },
    {
      "epoch": 0.7696419027258151,
      "grad_norm": 0.8648636937141418,
      "learning_rate": 0.00019709108989817212,
      "loss": 2.0437,
      "step": 2880
    },
    {
      "epoch": 0.772314270443613,
      "grad_norm": 1.0553714036941528,
      "learning_rate": 0.0001970709534305443,
      "loss": 2.002,
      "step": 2890
    },
    {
      "epoch": 0.774986638161411,
      "grad_norm": 0.9300885796546936,
      "learning_rate": 0.00019705074854317186,
      "loss": 1.9963,
      "step": 2900
    },
    {
      "epoch": 0.777659005879209,
      "grad_norm": 0.9071952700614929,
      "learning_rate": 0.00019703047525029604,
      "loss": 1.9934,
      "step": 2910
    },
    {
      "epoch": 0.780331373597007,
      "grad_norm": 0.9207800626754761,
      "learning_rate": 0.00019701013356620637,
      "loss": 2.0097,
      "step": 2920
    },
    {
      "epoch": 0.7830037413148049,
      "grad_norm": 0.9175180196762085,
      "learning_rate": 0.0001969897235052405,
      "loss": 2.0086,
      "step": 2930
    },
    {
      "epoch": 0.7856761090326029,
      "grad_norm": 0.8758548498153687,
      "learning_rate": 0.00019696924508178434,
      "loss": 2.0164,
      "step": 2940
    },
    {
      "epoch": 0.7883484767504009,
      "grad_norm": 0.979289174079895,
      "learning_rate": 0.00019694869831027192,
      "loss": 2.0434,
      "step": 2950
    },
    {
      "epoch": 0.7910208444681989,
      "grad_norm": 0.9176928400993347,
      "learning_rate": 0.0001969280832051855,
      "loss": 2.0593,
      "step": 2960
    },
    {
      "epoch": 0.7936932121859968,
      "grad_norm": 0.857001781463623,
      "learning_rate": 0.00019690739978105547,
      "loss": 1.9793,
      "step": 2970
    },
    {
      "epoch": 0.7963655799037948,
      "grad_norm": 0.9963621497154236,
      "learning_rate": 0.00019688664805246042,
      "loss": 2.029,
      "step": 2980
    },
    {
      "epoch": 0.7990379476215927,
      "grad_norm": 0.9742750525474548,
      "learning_rate": 0.000196865828034027,
      "loss": 2.0583,
      "step": 2990
    },
    {
      "epoch": 0.8017103153393907,
      "grad_norm": 0.9546931982040405,
      "learning_rate": 0.0001968449397404301,
      "loss": 2.0153,
      "step": 3000
    },
    {
      "epoch": 0.8043826830571886,
      "grad_norm": 0.9610984921455383,
      "learning_rate": 0.0001968239831863927,
      "loss": 1.9752,
      "step": 3010
    },
    {
      "epoch": 0.8070550507749866,
      "grad_norm": 0.8832826614379883,
      "learning_rate": 0.0001968029583866858,
      "loss": 1.9657,
      "step": 3020
    },
    {
      "epoch": 0.8097274184927846,
      "grad_norm": 0.8802211284637451,
      "learning_rate": 0.0001967818653561286,
      "loss": 2.0418,
      "step": 3030
    },
    {
      "epoch": 0.8123997862105826,
      "grad_norm": 0.9195637106895447,
      "learning_rate": 0.00019676070410958838,
      "loss": 1.9739,
      "step": 3040
    },
    {
      "epoch": 0.8150721539283805,
      "grad_norm": 1.000476598739624,
      "learning_rate": 0.00019673947466198048,
      "loss": 2.0173,
      "step": 3050
    },
    {
      "epoch": 0.8177445216461785,
      "grad_norm": 0.9482691287994385,
      "learning_rate": 0.00019671817702826832,
      "loss": 2.0132,
      "step": 3060
    },
    {
      "epoch": 0.8204168893639765,
      "grad_norm": 0.8577678799629211,
      "learning_rate": 0.00019669681122346338,
      "loss": 1.9925,
      "step": 3070
    },
    {
      "epoch": 0.8230892570817745,
      "grad_norm": 0.849473237991333,
      "learning_rate": 0.0001966753772626252,
      "loss": 2.0145,
      "step": 3080
    },
    {
      "epoch": 0.8257616247995724,
      "grad_norm": 0.8931000232696533,
      "learning_rate": 0.00019665387516086132,
      "loss": 2.0173,
      "step": 3090
    },
    {
      "epoch": 0.8284339925173704,
      "grad_norm": 0.8839173913002014,
      "learning_rate": 0.00019663230493332736,
      "loss": 1.9789,
      "step": 3100
    },
    {
      "epoch": 0.8311063602351684,
      "grad_norm": 0.8746767044067383,
      "learning_rate": 0.00019661066659522694,
      "loss": 1.9989,
      "step": 3110
    },
    {
      "epoch": 0.8337787279529664,
      "grad_norm": 0.9150108098983765,
      "learning_rate": 0.00019658896016181167,
      "loss": 1.9957,
      "step": 3120
    },
    {
      "epoch": 0.8364510956707643,
      "grad_norm": 0.929069459438324,
      "learning_rate": 0.0001965671856483812,
      "loss": 1.9754,
      "step": 3130
    },
    {
      "epoch": 0.8391234633885623,
      "grad_norm": 0.9361787438392639,
      "learning_rate": 0.00019654534307028307,
      "loss": 1.9889,
      "step": 3140
    },
    {
      "epoch": 0.8417958311063602,
      "grad_norm": 0.8942627906799316,
      "learning_rate": 0.0001965234324429129,
      "loss": 1.992,
      "step": 3150
    },
    {
      "epoch": 0.8444681988241582,
      "grad_norm": 0.9483137130737305,
      "learning_rate": 0.00019650145378171427,
      "loss": 1.9743,
      "step": 3160
    },
    {
      "epoch": 0.8471405665419561,
      "grad_norm": 0.9091803431510925,
      "learning_rate": 0.00019647940710217863,
      "loss": 1.9508,
      "step": 3170
    },
    {
      "epoch": 0.8498129342597541,
      "grad_norm": 0.961330771446228,
      "learning_rate": 0.00019645729241984542,
      "loss": 2.0216,
      "step": 3180
    },
    {
      "epoch": 0.8524853019775521,
      "grad_norm": 0.9919024109840393,
      "learning_rate": 0.000196435109750302,
      "loss": 1.9786,
      "step": 3190
    },
    {
      "epoch": 0.8551576696953501,
      "grad_norm": 0.8831027746200562,
      "learning_rate": 0.0001964128591091837,
      "loss": 1.975,
      "step": 3200
    },
    {
      "epoch": 0.857830037413148,
      "grad_norm": 0.8697969317436218,
      "learning_rate": 0.0001963905405121737,
      "loss": 1.9896,
      "step": 3210
    },
    {
      "epoch": 0.860502405130946,
      "grad_norm": 1.0299196243286133,
      "learning_rate": 0.00019636815397500312,
      "loss": 2.0273,
      "step": 3220
    },
    {
      "epoch": 0.863174772848744,
      "grad_norm": 0.9206592440605164,
      "learning_rate": 0.00019634569951345088,
      "loss": 1.9528,
      "step": 3230
    },
    {
      "epoch": 0.865847140566542,
      "grad_norm": 0.9223202466964722,
      "learning_rate": 0.00019632317714334395,
      "loss": 2.0344,
      "step": 3240
    },
    {
      "epoch": 0.8685195082843399,
      "grad_norm": 1.019234299659729,
      "learning_rate": 0.00019630058688055693,
      "loss": 1.9809,
      "step": 3250
    },
    {
      "epoch": 0.8711918760021379,
      "grad_norm": 0.8873318433761597,
      "learning_rate": 0.0001962779287410125,
      "loss": 1.959,
      "step": 3260
    },
    {
      "epoch": 0.8738642437199359,
      "grad_norm": 0.9181541800498962,
      "learning_rate": 0.00019625520274068104,
      "loss": 1.9607,
      "step": 3270
    },
    {
      "epoch": 0.8765366114377339,
      "grad_norm": 0.9755869507789612,
      "learning_rate": 0.0001962324088955808,
      "loss": 2.0121,
      "step": 3280
    },
    {
      "epoch": 0.8792089791555318,
      "grad_norm": 0.8667630553245544,
      "learning_rate": 0.00019620954722177785,
      "loss": 2.0633,
      "step": 3290
    },
    {
      "epoch": 0.8818813468733298,
      "grad_norm": 0.9352260231971741,
      "learning_rate": 0.00019618661773538606,
      "loss": 1.9954,
      "step": 3300
    },
    {
      "epoch": 0.8845537145911277,
      "grad_norm": 1.1069834232330322,
      "learning_rate": 0.0001961636204525672,
      "loss": 1.9668,
      "step": 3310
    },
    {
      "epoch": 0.8872260823089257,
      "grad_norm": 0.8498321771621704,
      "learning_rate": 0.00019614055538953062,
      "loss": 2.0374,
      "step": 3320
    },
    {
      "epoch": 0.8898984500267236,
      "grad_norm": 0.9243105053901672,
      "learning_rate": 0.0001961174225625336,
      "loss": 1.9106,
      "step": 3330
    },
    {
      "epoch": 0.8925708177445216,
      "grad_norm": 0.9041611552238464,
      "learning_rate": 0.00019609422198788119,
      "loss": 1.9412,
      "step": 3340
    },
    {
      "epoch": 0.8952431854623196,
      "grad_norm": 0.9980326294898987,
      "learning_rate": 0.00019607095368192609,
      "loss": 1.9816,
      "step": 3350
    },
    {
      "epoch": 0.8979155531801176,
      "grad_norm": 0.8987212777137756,
      "learning_rate": 0.00019604761766106878,
      "loss": 2.0317,
      "step": 3360
    },
    {
      "epoch": 0.9005879208979155,
      "grad_norm": 0.9415543675422668,
      "learning_rate": 0.00019602421394175753,
      "loss": 2.0433,
      "step": 3370
    },
    {
      "epoch": 0.9032602886157135,
      "grad_norm": 0.8867407441139221,
      "learning_rate": 0.00019600074254048826,
      "loss": 1.9821,
      "step": 3380
    },
    {
      "epoch": 0.9059326563335115,
      "grad_norm": 1.0224419832229614,
      "learning_rate": 0.0001959772034738046,
      "loss": 2.0231,
      "step": 3390
    },
    {
      "epoch": 0.9086050240513095,
      "grad_norm": 0.9306578040122986,
      "learning_rate": 0.0001959535967582979,
      "loss": 2.0645,
      "step": 3400
    },
    {
      "epoch": 0.9112773917691074,
      "grad_norm": 0.9539933204650879,
      "learning_rate": 0.00019592992241060717,
      "loss": 2.023,
      "step": 3410
    },
    {
      "epoch": 0.9139497594869054,
      "grad_norm": 1.0526412725448608,
      "learning_rate": 0.0001959061804474191,
      "loss": 2.0413,
      "step": 3420
    },
    {
      "epoch": 0.9166221272047034,
      "grad_norm": 0.9722763299942017,
      "learning_rate": 0.00019588237088546807,
      "loss": 2.0235,
      "step": 3430
    },
    {
      "epoch": 0.9192944949225014,
      "grad_norm": 0.9531739354133606,
      "learning_rate": 0.00019585849374153603,
      "loss": 1.9189,
      "step": 3440
    },
    {
      "epoch": 0.9219668626402993,
      "grad_norm": 0.8986770510673523,
      "learning_rate": 0.00019583454903245265,
      "loss": 2.0764,
      "step": 3450
    },
    {
      "epoch": 0.9246392303580973,
      "grad_norm": 0.8783401846885681,
      "learning_rate": 0.0001958105367750951,
      "loss": 1.9981,
      "step": 3460
    },
    {
      "epoch": 0.9273115980758952,
      "grad_norm": 0.8921948075294495,
      "learning_rate": 0.00019578645698638836,
      "loss": 1.9624,
      "step": 3470
    },
    {
      "epoch": 0.9299839657936932,
      "grad_norm": 0.9205301403999329,
      "learning_rate": 0.0001957623096833048,
      "loss": 1.954,
      "step": 3480
    },
    {
      "epoch": 0.9326563335114911,
      "grad_norm": 0.9025654196739197,
      "learning_rate": 0.00019573809488286452,
      "loss": 2.0084,
      "step": 3490
    },
    {
      "epoch": 0.9353287012292891,
      "grad_norm": 0.9316665530204773,
      "learning_rate": 0.00019571381260213508,
      "loss": 1.9663,
      "step": 3500
    },
    {
      "epoch": 0.9380010689470871,
      "grad_norm": 1.025264024734497,
      "learning_rate": 0.00019568946285823175,
      "loss": 2.001,
      "step": 3510
    },
    {
      "epoch": 0.9406734366648851,
      "grad_norm": 0.8893492817878723,
      "learning_rate": 0.0001956650456683172,
      "loss": 1.9919,
      "step": 3520
    },
    {
      "epoch": 0.943345804382683,
      "grad_norm": 0.9078524708747864,
      "learning_rate": 0.00019564056104960176,
      "loss": 2.0113,
      "step": 3530
    },
    {
      "epoch": 0.946018172100481,
      "grad_norm": 0.9093054533004761,
      "learning_rate": 0.0001956160090193432,
      "loss": 2.0191,
      "step": 3540
    },
    {
      "epoch": 0.948690539818279,
      "grad_norm": 0.9037675857543945,
      "learning_rate": 0.0001955913895948468,
      "loss": 1.9785,
      "step": 3550
    },
    {
      "epoch": 0.951362907536077,
      "grad_norm": 0.914854884147644,
      "learning_rate": 0.00019556670279346548,
      "loss": 2.0078,
      "step": 3560
    },
    {
      "epoch": 0.9540352752538749,
      "grad_norm": 0.9203145503997803,
      "learning_rate": 0.00019554194863259948,
      "loss": 1.9819,
      "step": 3570
    },
    {
      "epoch": 0.9567076429716729,
      "grad_norm": 0.9281837344169617,
      "learning_rate": 0.00019551712712969656,
      "loss": 1.981,
      "step": 3580
    },
    {
      "epoch": 0.9593800106894709,
      "grad_norm": 0.9327614903450012,
      "learning_rate": 0.00019549223830225203,
      "loss": 1.9994,
      "step": 3590
    },
    {
      "epoch": 0.9620523784072689,
      "grad_norm": 0.9100453853607178,
      "learning_rate": 0.0001954672821678086,
      "loss": 1.9513,
      "step": 3600
    },
    {
      "epoch": 0.9647247461250668,
      "grad_norm": 0.8855226635932922,
      "learning_rate": 0.0001954422587439564,
      "loss": 1.9675,
      "step": 3610
    },
    {
      "epoch": 0.9673971138428648,
      "grad_norm": 0.91499924659729,
      "learning_rate": 0.00019541716804833296,
      "loss": 1.9912,
      "step": 3620
    },
    {
      "epoch": 0.9700694815606627,
      "grad_norm": 0.9307335615158081,
      "learning_rate": 0.00019539201009862332,
      "loss": 1.9813,
      "step": 3630
    },
    {
      "epoch": 0.9727418492784607,
      "grad_norm": 1.0112236738204956,
      "learning_rate": 0.00019536678491255987,
      "loss": 1.9543,
      "step": 3640
    },
    {
      "epoch": 0.9754142169962586,
      "grad_norm": 0.939149022102356,
      "learning_rate": 0.00019534149250792238,
      "loss": 1.9789,
      "step": 3650
    },
    {
      "epoch": 0.9780865847140566,
      "grad_norm": 0.8903077840805054,
      "learning_rate": 0.00019531613290253805,
      "loss": 2.0364,
      "step": 3660
    },
    {
      "epoch": 0.9807589524318546,
      "grad_norm": 0.8920482397079468,
      "learning_rate": 0.00019529070611428137,
      "loss": 1.9463,
      "step": 3670
    },
    {
      "epoch": 0.9834313201496526,
      "grad_norm": 0.947891354560852,
      "learning_rate": 0.00019526521216107427,
      "loss": 1.9562,
      "step": 3680
    },
    {
      "epoch": 0.9861036878674505,
      "grad_norm": 0.9681838154792786,
      "learning_rate": 0.00019523965106088594,
      "loss": 2.0359,
      "step": 3690
    },
    {
      "epoch": 0.9887760555852485,
      "grad_norm": 0.9280781149864197,
      "learning_rate": 0.0001952140228317329,
      "loss": 1.9686,
      "step": 3700
    },
    {
      "epoch": 0.9914484233030465,
      "grad_norm": 0.9600924253463745,
      "learning_rate": 0.00019518832749167912,
      "loss": 1.9946,
      "step": 3710
    },
    {
      "epoch": 0.9941207910208445,
      "grad_norm": 0.8943674564361572,
      "learning_rate": 0.00019516256505883566,
      "loss": 1.9827,
      "step": 3720
    },
    {
      "epoch": 0.9967931587386424,
      "grad_norm": 0.9458594918251038,
      "learning_rate": 0.00019513673555136105,
      "loss": 1.9716,
      "step": 3730
    },
    {
      "epoch": 0.9994655264564404,
      "grad_norm": 0.9723443984985352,
      "learning_rate": 0.00019511083898746101,
      "loss": 1.9845,
      "step": 3740
    },
    {
      "epoch": 1.0021378941742385,
      "grad_norm": 0.9325452446937561,
      "learning_rate": 0.00019508487538538852,
      "loss": 1.8939,
      "step": 3750
    },
    {
      "epoch": 1.0048102618920363,
      "grad_norm": 0.9777805805206299,
      "learning_rate": 0.00019505884476344384,
      "loss": 1.8208,
      "step": 3760
    },
    {
      "epoch": 1.0074826296098343,
      "grad_norm": 1.0973187685012817,
      "learning_rate": 0.0001950327471399745,
      "loss": 1.8432,
      "step": 3770
    },
    {
      "epoch": 1.0101549973276323,
      "grad_norm": 1.0199981927871704,
      "learning_rate": 0.00019500658253337517,
      "loss": 1.8238,
      "step": 3780
    },
    {
      "epoch": 1.0128273650454303,
      "grad_norm": 1.0170180797576904,
      "learning_rate": 0.00019498035096208782,
      "loss": 1.8515,
      "step": 3790
    },
    {
      "epoch": 1.0154997327632282,
      "grad_norm": 0.9383100867271423,
      "learning_rate": 0.00019495405244460154,
      "loss": 1.8544,
      "step": 3800
    },
    {
      "epoch": 1.0181721004810262,
      "grad_norm": 0.9334129095077515,
      "learning_rate": 0.0001949276869994527,
      "loss": 1.8378,
      "step": 3810
    },
    {
      "epoch": 1.0208444681988242,
      "grad_norm": 1.0014691352844238,
      "learning_rate": 0.0001949012546452247,
      "loss": 1.778,
      "step": 3820
    },
    {
      "epoch": 1.0235168359166222,
      "grad_norm": 1.0079625844955444,
      "learning_rate": 0.00019487475540054828,
      "loss": 1.8765,
      "step": 3830
    },
    {
      "epoch": 1.02618920363442,
      "grad_norm": 0.9731796383857727,
      "learning_rate": 0.0001948481892841012,
      "loss": 1.8032,
      "step": 3840
    },
    {
      "epoch": 1.028861571352218,
      "grad_norm": 0.9402268528938293,
      "learning_rate": 0.00019482155631460836,
      "loss": 1.8097,
      "step": 3850
    },
    {
      "epoch": 1.031533939070016,
      "grad_norm": 1.013541340827942,
      "learning_rate": 0.00019479485651084186,
      "loss": 1.8851,
      "step": 3860
    },
    {
      "epoch": 1.034206306787814,
      "grad_norm": 1.0363236665725708,
      "learning_rate": 0.00019476808989162087,
      "loss": 1.7967,
      "step": 3870
    },
    {
      "epoch": 1.036878674505612,
      "grad_norm": 1.1777698993682861,
      "learning_rate": 0.00019474125647581157,
      "loss": 1.8259,
      "step": 3880
    },
    {
      "epoch": 1.03955104222341,
      "grad_norm": 0.9758262038230896,
      "learning_rate": 0.0001947143562823274,
      "loss": 1.8336,
      "step": 3890
    },
    {
      "epoch": 1.042223409941208,
      "grad_norm": 1.039447546005249,
      "learning_rate": 0.00019468738933012866,
      "loss": 1.8978,
      "step": 3900
    },
    {
      "epoch": 1.044895777659006,
      "grad_norm": 1.0558720827102661,
      "learning_rate": 0.00019466035563822285,
      "loss": 1.8205,
      "step": 3910
    },
    {
      "epoch": 1.0475681453768038,
      "grad_norm": 0.9531826972961426,
      "learning_rate": 0.0001946332552256645,
      "loss": 1.8302,
      "step": 3920
    },
    {
      "epoch": 1.0502405130946018,
      "grad_norm": 1.007906198501587,
      "learning_rate": 0.00019460608811155508,
      "loss": 1.8116,
      "step": 3930
    },
    {
      "epoch": 1.0529128808123998,
      "grad_norm": 1.0126383304595947,
      "learning_rate": 0.00019457885431504318,
      "loss": 1.8165,
      "step": 3940
    },
    {
      "epoch": 1.0555852485301978,
      "grad_norm": 1.0046895742416382,
      "learning_rate": 0.0001945515538553243,
      "loss": 1.86,
      "step": 3950
    },
    {
      "epoch": 1.0582576162479957,
      "grad_norm": 1.0600786209106445,
      "learning_rate": 0.00019452418675164096,
      "loss": 1.9054,
      "step": 3960
    },
    {
      "epoch": 1.0609299839657937,
      "grad_norm": 1.1300631761550903,
      "learning_rate": 0.0001944967530232827,
      "loss": 1.8817,
      "step": 3970
    },
    {
      "epoch": 1.0636023516835917,
      "grad_norm": 1.0602167844772339,
      "learning_rate": 0.00019446925268958598,
      "loss": 1.8615,
      "step": 3980
    },
    {
      "epoch": 1.0662747194013895,
      "grad_norm": 0.9714751839637756,
      "learning_rate": 0.00019444168576993417,
      "loss": 1.8389,
      "step": 3990
    },
    {
      "epoch": 1.0689470871191875,
      "grad_norm": 1.0327038764953613,
      "learning_rate": 0.00019441405228375763,
      "loss": 1.8058,
      "step": 4000
    },
    {
      "epoch": 1.0716194548369855,
      "grad_norm": 0.9951843619346619,
      "learning_rate": 0.00019438635225053362,
      "loss": 1.8708,
      "step": 4010
    },
    {
      "epoch": 1.0742918225547835,
      "grad_norm": 1.0652300119400024,
      "learning_rate": 0.00019435858568978633,
      "loss": 1.881,
      "step": 4020
    },
    {
      "epoch": 1.0769641902725815,
      "grad_norm": 1.0392547845840454,
      "learning_rate": 0.00019433075262108678,
      "loss": 1.8336,
      "step": 4030
    },
    {
      "epoch": 1.0796365579903795,
      "grad_norm": 1.0417035818099976,
      "learning_rate": 0.0001943028530640529,
      "loss": 1.8565,
      "step": 4040
    },
    {
      "epoch": 1.0823089257081775,
      "grad_norm": 0.9951125979423523,
      "learning_rate": 0.00019427488703834952,
      "loss": 1.8354,
      "step": 4050
    },
    {
      "epoch": 1.0849812934259755,
      "grad_norm": 1.0386244058609009,
      "learning_rate": 0.00019424685456368823,
      "loss": 1.838,
      "step": 4060
    },
    {
      "epoch": 1.0876536611437735,
      "grad_norm": 1.0159943103790283,
      "learning_rate": 0.00019421875565982756,
      "loss": 1.7713,
      "step": 4070
    },
    {
      "epoch": 1.0903260288615713,
      "grad_norm": 1.042913556098938,
      "learning_rate": 0.0001941905903465728,
      "loss": 1.8475,
      "step": 4080
    },
    {
      "epoch": 1.0929983965793693,
      "grad_norm": 0.9615035653114319,
      "learning_rate": 0.0001941623586437761,
      "loss": 1.8218,
      "step": 4090
    },
    {
      "epoch": 1.0956707642971673,
      "grad_norm": 1.1015328168869019,
      "learning_rate": 0.0001941340605713363,
      "loss": 1.8404,
      "step": 4100
    },
    {
      "epoch": 1.0983431320149653,
      "grad_norm": 1.030737280845642,
      "learning_rate": 0.0001941056961491991,
      "loss": 1.7929,
      "step": 4110
    },
    {
      "epoch": 1.1010154997327632,
      "grad_norm": 1.2163360118865967,
      "learning_rate": 0.00019407726539735705,
      "loss": 1.7742,
      "step": 4120
    },
    {
      "epoch": 1.1036878674505612,
      "grad_norm": 1.073459267616272,
      "learning_rate": 0.00019404876833584923,
      "loss": 1.8787,
      "step": 4130
    },
    {
      "epoch": 1.1063602351683592,
      "grad_norm": 1.069594144821167,
      "learning_rate": 0.0001940202049847616,
      "loss": 1.7977,
      "step": 4140
    },
    {
      "epoch": 1.1090326028861572,
      "grad_norm": 0.9366630911827087,
      "learning_rate": 0.00019399157536422693,
      "loss": 1.8075,
      "step": 4150
    },
    {
      "epoch": 1.111704970603955,
      "grad_norm": 1.013414978981018,
      "learning_rate": 0.00019396287949442447,
      "loss": 1.8798,
      "step": 4160
    },
    {
      "epoch": 1.114377338321753,
      "grad_norm": 0.9827794432640076,
      "learning_rate": 0.0001939341173955804,
      "loss": 1.8713,
      "step": 4170
    },
    {
      "epoch": 1.117049706039551,
      "grad_norm": 1.013396143913269,
      "learning_rate": 0.00019390528908796737,
      "loss": 1.8524,
      "step": 4180
    },
    {
      "epoch": 1.119722073757349,
      "grad_norm": 0.9573310613632202,
      "learning_rate": 0.00019387639459190488,
      "loss": 1.8323,
      "step": 4190
    },
    {
      "epoch": 1.122394441475147,
      "grad_norm": 1.0018446445465088,
      "learning_rate": 0.00019384743392775893,
      "loss": 1.7816,
      "step": 4200
    },
    {
      "epoch": 1.125066809192945,
      "grad_norm": 1.1065547466278076,
      "learning_rate": 0.00019381840711594234,
      "loss": 1.8001,
      "step": 4210
    },
    {
      "epoch": 1.127739176910743,
      "grad_norm": 1.0700459480285645,
      "learning_rate": 0.00019378931417691434,
      "loss": 1.8595,
      "step": 4220
    },
    {
      "epoch": 1.130411544628541,
      "grad_norm": 1.0059701204299927,
      "learning_rate": 0.00019376015513118092,
      "loss": 1.8351,
      "step": 4230
    },
    {
      "epoch": 1.1330839123463388,
      "grad_norm": 0.9777234792709351,
      "learning_rate": 0.00019373092999929462,
      "loss": 1.8352,
      "step": 4240
    },
    {
      "epoch": 1.1357562800641368,
      "grad_norm": 1.075574517250061,
      "learning_rate": 0.00019370163880185452,
      "loss": 1.8722,
      "step": 4250
    },
    {
      "epoch": 1.1384286477819348,
      "grad_norm": 1.0689491033554077,
      "learning_rate": 0.00019367228155950633,
      "loss": 1.8386,
      "step": 4260
    },
    {
      "epoch": 1.1411010154997328,
      "grad_norm": 1.088172197341919,
      "learning_rate": 0.00019364285829294228,
      "loss": 1.8065,
      "step": 4270
    },
    {
      "epoch": 1.1437733832175307,
      "grad_norm": 1.137303113937378,
      "learning_rate": 0.0001936133690229012,
      "loss": 1.9046,
      "step": 4280
    },
    {
      "epoch": 1.1464457509353287,
      "grad_norm": 1.072802186012268,
      "learning_rate": 0.0001935838137701683,
      "loss": 1.8399,
      "step": 4290
    },
    {
      "epoch": 1.1491181186531267,
      "grad_norm": 1.100389838218689,
      "learning_rate": 0.0001935541925555754,
      "loss": 1.8745,
      "step": 4300
    },
    {
      "epoch": 1.1517904863709245,
      "grad_norm": 1.0402289628982544,
      "learning_rate": 0.00019352450540000083,
      "loss": 1.8369,
      "step": 4310
    },
    {
      "epoch": 1.1544628540887225,
      "grad_norm": 1.0470116138458252,
      "learning_rate": 0.00019349475232436936,
      "loss": 1.8598,
      "step": 4320
    },
    {
      "epoch": 1.1571352218065205,
      "grad_norm": 1.1584430932998657,
      "learning_rate": 0.00019346493334965217,
      "loss": 1.8547,
      "step": 4330
    },
    {
      "epoch": 1.1598075895243185,
      "grad_norm": 1.0952426195144653,
      "learning_rate": 0.000193435048496867,
      "loss": 1.8756,
      "step": 4340
    },
    {
      "epoch": 1.1624799572421165,
      "grad_norm": 1.0330133438110352,
      "learning_rate": 0.00019340509778707792,
      "loss": 1.8375,
      "step": 4350
    },
    {
      "epoch": 1.1651523249599145,
      "grad_norm": 1.1242432594299316,
      "learning_rate": 0.00019337508124139553,
      "loss": 1.7711,
      "step": 4360
    },
    {
      "epoch": 1.1678246926777125,
      "grad_norm": 1.0563093423843384,
      "learning_rate": 0.00019334499888097674,
      "loss": 1.8461,
      "step": 4370
    },
    {
      "epoch": 1.1704970603955105,
      "grad_norm": 1.044492244720459,
      "learning_rate": 0.00019331485072702484,
      "loss": 1.9135,
      "step": 4380
    },
    {
      "epoch": 1.1731694281133085,
      "grad_norm": 0.9759335517883301,
      "learning_rate": 0.0001932846368007896,
      "loss": 1.8698,
      "step": 4390
    },
    {
      "epoch": 1.1758417958311065,
      "grad_norm": 1.0697232484817505,
      "learning_rate": 0.00019325435712356704,
      "loss": 1.8769,
      "step": 4400
    },
    {
      "epoch": 1.1785141635489043,
      "grad_norm": 1.0401313304901123,
      "learning_rate": 0.00019322401171669958,
      "loss": 1.9235,
      "step": 4410
    },
    {
      "epoch": 1.1811865312667023,
      "grad_norm": 1.1742342710494995,
      "learning_rate": 0.00019319360060157594,
      "loss": 1.7966,
      "step": 4420
    },
    {
      "epoch": 1.1838588989845003,
      "grad_norm": 1.048688292503357,
      "learning_rate": 0.00019316312379963118,
      "loss": 1.7849,
      "step": 4430
    },
    {
      "epoch": 1.1865312667022982,
      "grad_norm": 1.037684679031372,
      "learning_rate": 0.0001931325813323467,
      "loss": 1.8176,
      "step": 4440
    },
    {
      "epoch": 1.1892036344200962,
      "grad_norm": 1.0703048706054688,
      "learning_rate": 0.00019310197322125006,
      "loss": 1.8587,
      "step": 4450
    },
    {
      "epoch": 1.1918760021378942,
      "grad_norm": 0.9846740365028381,
      "learning_rate": 0.0001930712994879152,
      "loss": 1.8726,
      "step": 4460
    },
    {
      "epoch": 1.1945483698556922,
      "grad_norm": 1.0052661895751953,
      "learning_rate": 0.0001930405601539622,
      "loss": 1.8343,
      "step": 4470
    },
    {
      "epoch": 1.19722073757349,
      "grad_norm": 1.0184731483459473,
      "learning_rate": 0.00019300975524105758,
      "loss": 1.8112,
      "step": 4480
    },
    {
      "epoch": 1.199893105291288,
      "grad_norm": 1.0682815313339233,
      "learning_rate": 0.00019297888477091388,
      "loss": 1.8566,
      "step": 4490
    },
    {
      "epoch": 1.202565473009086,
      "grad_norm": 0.9575110077857971,
      "learning_rate": 0.00019294794876528992,
      "loss": 1.8807,
      "step": 4500
    },
    {
      "epoch": 1.205237840726884,
      "grad_norm": 1.0943394899368286,
      "learning_rate": 0.00019291694724599072,
      "loss": 1.8857,
      "step": 4510
    },
    {
      "epoch": 1.207910208444682,
      "grad_norm": 1.1048181056976318,
      "learning_rate": 0.00019288588023486748,
      "loss": 1.8038,
      "step": 4520
    },
    {
      "epoch": 1.21058257616248,
      "grad_norm": 1.0477944612503052,
      "learning_rate": 0.00019285474775381759,
      "loss": 1.8832,
      "step": 4530
    },
    {
      "epoch": 1.213254943880278,
      "grad_norm": 1.0154651403427124,
      "learning_rate": 0.00019282354982478445,
      "loss": 1.8237,
      "step": 4540
    },
    {
      "epoch": 1.215927311598076,
      "grad_norm": 1.1119800806045532,
      "learning_rate": 0.00019279228646975778,
      "loss": 1.8808,
      "step": 4550
    },
    {
      "epoch": 1.218599679315874,
      "grad_norm": 1.0663464069366455,
      "learning_rate": 0.0001927609577107733,
      "loss": 1.8468,
      "step": 4560
    },
    {
      "epoch": 1.2212720470336718,
      "grad_norm": 1.007313847541809,
      "learning_rate": 0.00019272956356991284,
      "loss": 1.7839,
      "step": 4570
    },
    {
      "epoch": 1.2239444147514698,
      "grad_norm": 1.0278741121292114,
      "learning_rate": 0.00019269810406930438,
      "loss": 1.8177,
      "step": 4580
    },
    {
      "epoch": 1.2266167824692678,
      "grad_norm": 1.0416043996810913,
      "learning_rate": 0.00019266657923112184,
      "loss": 1.8579,
      "step": 4590
    },
    {
      "epoch": 1.2292891501870657,
      "grad_norm": 1.0626152753829956,
      "learning_rate": 0.0001926349890775853,
      "loss": 1.8608,
      "step": 4600
    },
    {
      "epoch": 1.2319615179048637,
      "grad_norm": 1.106374740600586,
      "learning_rate": 0.00019260333363096087,
      "loss": 1.8846,
      "step": 4610
    },
    {
      "epoch": 1.2346338856226617,
      "grad_norm": 1.1124606132507324,
      "learning_rate": 0.00019257161291356064,
      "loss": 1.8111,
      "step": 4620
    },
    {
      "epoch": 1.2373062533404597,
      "grad_norm": 1.074202299118042,
      "learning_rate": 0.0001925398269477427,
      "loss": 1.9275,
      "step": 4630
    },
    {
      "epoch": 1.2399786210582575,
      "grad_norm": 0.9949113726615906,
      "learning_rate": 0.0001925079757559112,
      "loss": 1.8803,
      "step": 4640
    },
    {
      "epoch": 1.2426509887760555,
      "grad_norm": 1.1150023937225342,
      "learning_rate": 0.00019247605936051617,
      "loss": 1.8573,
      "step": 4650
    },
    {
      "epoch": 1.2453233564938535,
      "grad_norm": 1.0155227184295654,
      "learning_rate": 0.00019244407778405372,
      "loss": 1.8604,
      "step": 4660
    },
    {
      "epoch": 1.2479957242116515,
      "grad_norm": 1.0073447227478027,
      "learning_rate": 0.0001924120310490657,
      "loss": 1.8634,
      "step": 4670
    },
    {
      "epoch": 1.2506680919294495,
      "grad_norm": 1.0564298629760742,
      "learning_rate": 0.00019237991917814013,
      "loss": 1.8315,
      "step": 4680
    },
    {
      "epoch": 1.2533404596472475,
      "grad_norm": 1.0739296674728394,
      "learning_rate": 0.00019234774219391082,
      "loss": 1.8478,
      "step": 4690
    },
    {
      "epoch": 1.2560128273650455,
      "grad_norm": 1.1494308710098267,
      "learning_rate": 0.0001923155001190574,
      "loss": 1.8246,
      "step": 4700
    },
    {
      "epoch": 1.2586851950828435,
      "grad_norm": 1.0917733907699585,
      "learning_rate": 0.0001922831929763055,
      "loss": 1.8558,
      "step": 4710
    },
    {
      "epoch": 1.2613575628006415,
      "grad_norm": 1.0394524335861206,
      "learning_rate": 0.0001922508207884266,
      "loss": 1.8838,
      "step": 4720
    },
    {
      "epoch": 1.2640299305184393,
      "grad_norm": 1.042967438697815,
      "learning_rate": 0.000192218383578238,
      "loss": 1.8862,
      "step": 4730
    },
    {
      "epoch": 1.2667022982362373,
      "grad_norm": 1.2198184728622437,
      "learning_rate": 0.00019218588136860274,
      "loss": 1.9148,
      "step": 4740
    },
    {
      "epoch": 1.2693746659540353,
      "grad_norm": 1.0503358840942383,
      "learning_rate": 0.00019215331418242988,
      "loss": 1.7926,
      "step": 4750
    },
    {
      "epoch": 1.2720470336718332,
      "grad_norm": 1.0391160249710083,
      "learning_rate": 0.00019212068204267407,
      "loss": 1.8469,
      "step": 4760
    },
    {
      "epoch": 1.2747194013896312,
      "grad_norm": 1.2048615217208862,
      "learning_rate": 0.00019208798497233586,
      "loss": 1.9291,
      "step": 4770
    },
    {
      "epoch": 1.2773917691074292,
      "grad_norm": 1.1003245115280151,
      "learning_rate": 0.00019205522299446158,
      "loss": 1.8386,
      "step": 4780
    },
    {
      "epoch": 1.280064136825227,
      "grad_norm": 1.0906586647033691,
      "learning_rate": 0.0001920223961321432,
      "loss": 1.8626,
      "step": 4790
    },
    {
      "epoch": 1.282736504543025,
      "grad_norm": 1.0515018701553345,
      "learning_rate": 0.00019198950440851853,
      "loss": 1.8482,
      "step": 4800
    },
    {
      "epoch": 1.285408872260823,
      "grad_norm": 1.0668543577194214,
      "learning_rate": 0.00019195654784677106,
      "loss": 1.8351,
      "step": 4810
    },
    {
      "epoch": 1.288081239978621,
      "grad_norm": 1.0039379596710205,
      "learning_rate": 0.00019192352647012995,
      "loss": 1.8304,
      "step": 4820
    },
    {
      "epoch": 1.290753607696419,
      "grad_norm": 1.0450323820114136,
      "learning_rate": 0.00019189044030187017,
      "loss": 1.8702,
      "step": 4830
    },
    {
      "epoch": 1.293425975414217,
      "grad_norm": 1.1013343334197998,
      "learning_rate": 0.0001918572893653121,
      "loss": 1.8401,
      "step": 4840
    },
    {
      "epoch": 1.296098343132015,
      "grad_norm": 1.148041009902954,
      "learning_rate": 0.00019182407368382206,
      "loss": 1.9516,
      "step": 4850
    },
    {
      "epoch": 1.298770710849813,
      "grad_norm": 1.0496457815170288,
      "learning_rate": 0.00019179079328081184,
      "loss": 1.8738,
      "step": 4860
    },
    {
      "epoch": 1.301443078567611,
      "grad_norm": 1.0360918045043945,
      "learning_rate": 0.00019175744817973887,
      "loss": 1.8597,
      "step": 4870
    },
    {
      "epoch": 1.304115446285409,
      "grad_norm": 1.1782798767089844,
      "learning_rate": 0.00019172403840410625,
      "loss": 1.8483,
      "step": 4880
    },
    {
      "epoch": 1.3067878140032068,
      "grad_norm": 1.0753065347671509,
      "learning_rate": 0.00019169056397746255,
      "loss": 1.8699,
      "step": 4890
    },
    {
      "epoch": 1.3094601817210048,
      "grad_norm": 1.0449695587158203,
      "learning_rate": 0.000191657024923402,
      "loss": 1.8936,
      "step": 4900
    },
    {
      "epoch": 1.3121325494388028,
      "grad_norm": 1.0549265146255493,
      "learning_rate": 0.00019162342126556435,
      "loss": 1.8174,
      "step": 4910
    },
    {
      "epoch": 1.3148049171566007,
      "grad_norm": 1.0759707689285278,
      "learning_rate": 0.0001915897530276349,
      "loss": 1.8916,
      "step": 4920
    },
    {
      "epoch": 1.3174772848743987,
      "grad_norm": 1.0495866537094116,
      "learning_rate": 0.00019155602023334444,
      "loss": 1.829,
      "step": 4930
    },
    {
      "epoch": 1.3201496525921967,
      "grad_norm": 1.112349271774292,
      "learning_rate": 0.0001915222229064693,
      "loss": 1.8413,
      "step": 4940
    },
    {
      "epoch": 1.3228220203099945,
      "grad_norm": 1.011779546737671,
      "learning_rate": 0.00019148836107083127,
      "loss": 1.9259,
      "step": 4950
    },
    {
      "epoch": 1.3254943880277925,
      "grad_norm": 1.1030875444412231,
      "learning_rate": 0.0001914544347502976,
      "loss": 1.8494,
      "step": 4960
    },
    {
      "epoch": 1.3281667557455905,
      "grad_norm": 1.1434781551361084,
      "learning_rate": 0.00019142044396878104,
      "loss": 1.9277,
      "step": 4970
    },
    {
      "epoch": 1.3308391234633885,
      "grad_norm": 1.0343862771987915,
      "learning_rate": 0.00019138638875023972,
      "loss": 1.8284,
      "step": 4980
    },
    {
      "epoch": 1.3335114911811865,
      "grad_norm": 1.0611770153045654,
      "learning_rate": 0.00019135226911867718,
      "loss": 1.8642,
      "step": 4990
    },
    {
      "epoch": 1.3361838588989845,
      "grad_norm": 0.9910778999328613,
      "learning_rate": 0.00019131808509814245,
      "loss": 1.8258,
      "step": 5000
    },
    {
      "epoch": 1.3388562266167825,
      "grad_norm": 1.0867102146148682,
      "learning_rate": 0.00019128383671272988,
      "loss": 1.8897,
      "step": 5010
    },
    {
      "epoch": 1.3415285943345805,
      "grad_norm": 1.124879002571106,
      "learning_rate": 0.0001912495239865791,
      "loss": 1.8994,
      "step": 5020
    },
    {
      "epoch": 1.3442009620523785,
      "grad_norm": 1.1122723817825317,
      "learning_rate": 0.0001912151469438753,
      "loss": 1.9291,
      "step": 5030
    },
    {
      "epoch": 1.3468733297701765,
      "grad_norm": 1.0620670318603516,
      "learning_rate": 0.00019118070560884885,
      "loss": 1.8314,
      "step": 5040
    },
    {
      "epoch": 1.3495456974879745,
      "grad_norm": 1.0552860498428345,
      "learning_rate": 0.0001911462000057754,
      "loss": 1.7935,
      "step": 5050
    },
    {
      "epoch": 1.3522180652057723,
      "grad_norm": 1.136927843093872,
      "learning_rate": 0.00019111163015897607,
      "loss": 1.8598,
      "step": 5060
    },
    {
      "epoch": 1.3548904329235703,
      "grad_norm": 1.0880241394042969,
      "learning_rate": 0.0001910769960928171,
      "loss": 1.8243,
      "step": 5070
    },
    {
      "epoch": 1.3575628006413683,
      "grad_norm": 1.0525450706481934,
      "learning_rate": 0.00019104229783171005,
      "loss": 1.8203,
      "step": 5080
    },
    {
      "epoch": 1.3602351683591662,
      "grad_norm": 1.1240369081497192,
      "learning_rate": 0.0001910075354001118,
      "loss": 1.8641,
      "step": 5090
    },
    {
      "epoch": 1.3629075360769642,
      "grad_norm": 1.0436363220214844,
      "learning_rate": 0.0001909727088225243,
      "loss": 1.8671,
      "step": 5100
    },
    {
      "epoch": 1.3655799037947622,
      "grad_norm": 1.040597677230835,
      "learning_rate": 0.00019093781812349486,
      "loss": 1.8052,
      "step": 5110
    },
    {
      "epoch": 1.36825227151256,
      "grad_norm": 0.9839260578155518,
      "learning_rate": 0.00019090286332761592,
      "loss": 1.8177,
      "step": 5120
    },
    {
      "epoch": 1.370924639230358,
      "grad_norm": 1.1703975200653076,
      "learning_rate": 0.00019086784445952515,
      "loss": 1.8334,
      "step": 5130
    },
    {
      "epoch": 1.373597006948156,
      "grad_norm": 1.0526046752929688,
      "learning_rate": 0.0001908327615439053,
      "loss": 1.7837,
      "step": 5140
    },
    {
      "epoch": 1.376269374665954,
      "grad_norm": 1.1150929927825928,
      "learning_rate": 0.00019079761460548427,
      "loss": 1.8717,
      "step": 5150
    },
    {
      "epoch": 1.378941742383752,
      "grad_norm": 1.211668848991394,
      "learning_rate": 0.00019076240366903518,
      "loss": 1.8667,
      "step": 5160
    },
    {
      "epoch": 1.38161411010155,
      "grad_norm": 1.0549869537353516,
      "learning_rate": 0.0001907271287593762,
      "loss": 1.8025,
      "step": 5170
    },
    {
      "epoch": 1.384286477819348,
      "grad_norm": 1.2277272939682007,
      "learning_rate": 0.00019069178990137054,
      "loss": 1.8574,
      "step": 5180
    },
    {
      "epoch": 1.386958845537146,
      "grad_norm": 1.0361557006835938,
      "learning_rate": 0.00019065638711992654,
      "loss": 1.8309,
      "step": 5190
    },
    {
      "epoch": 1.389631213254944,
      "grad_norm": 1.081335425376892,
      "learning_rate": 0.00019062092043999767,
      "loss": 1.8703,
      "step": 5200
    },
    {
      "epoch": 1.392303580972742,
      "grad_norm": 1.1670414209365845,
      "learning_rate": 0.00019058538988658223,
      "loss": 1.9426,
      "step": 5210
    },
    {
      "epoch": 1.3949759486905398,
      "grad_norm": 1.1812220811843872,
      "learning_rate": 0.00019054979548472378,
      "loss": 1.8306,
      "step": 5220
    },
    {
      "epoch": 1.3976483164083378,
      "grad_norm": 1.1016520261764526,
      "learning_rate": 0.0001905141372595107,
      "loss": 1.8894,
      "step": 5230
    },
    {
      "epoch": 1.4003206841261358,
      "grad_norm": 1.078292727470398,
      "learning_rate": 0.00019047841523607648,
      "loss": 1.822,
      "step": 5240
    },
    {
      "epoch": 1.4029930518439337,
      "grad_norm": 1.0356316566467285,
      "learning_rate": 0.00019044262943959949,
      "loss": 1.8315,
      "step": 5250
    },
    {
      "epoch": 1.4056654195617317,
      "grad_norm": 1.0294471979141235,
      "learning_rate": 0.00019040677989530305,
      "loss": 1.8875,
      "step": 5260
    },
    {
      "epoch": 1.4083377872795297,
      "grad_norm": 1.043329119682312,
      "learning_rate": 0.0001903708666284555,
      "loss": 1.8544,
      "step": 5270
    },
    {
      "epoch": 1.4110101549973275,
      "grad_norm": 1.1735234260559082,
      "learning_rate": 0.00019033488966437004,
      "loss": 1.8749,
      "step": 5280
    },
    {
      "epoch": 1.4136825227151255,
      "grad_norm": 1.0429078340530396,
      "learning_rate": 0.00019029884902840478,
      "loss": 1.8523,
      "step": 5290
    },
    {
      "epoch": 1.4163548904329235,
      "grad_norm": 1.1267577409744263,
      "learning_rate": 0.00019026274474596267,
      "loss": 1.8811,
      "step": 5300
    },
    {
      "epoch": 1.4190272581507215,
      "grad_norm": 1.157902717590332,
      "learning_rate": 0.00019022657684249152,
      "loss": 1.8161,
      "step": 5310
    },
    {
      "epoch": 1.4216996258685195,
      "grad_norm": 1.0437076091766357,
      "learning_rate": 0.00019019034534348408,
      "loss": 1.8793,
      "step": 5320
    },
    {
      "epoch": 1.4243719935863175,
      "grad_norm": 0.9943877458572388,
      "learning_rate": 0.00019015405027447777,
      "loss": 1.8072,
      "step": 5330
    },
    {
      "epoch": 1.4270443613041155,
      "grad_norm": 1.1403636932373047,
      "learning_rate": 0.00019011769166105496,
      "loss": 1.9485,
      "step": 5340
    },
    {
      "epoch": 1.4297167290219135,
      "grad_norm": 1.0246118307113647,
      "learning_rate": 0.00019008126952884272,
      "loss": 1.9087,
      "step": 5350
    },
    {
      "epoch": 1.4323890967397115,
      "grad_norm": 1.1130748987197876,
      "learning_rate": 0.00019004478390351296,
      "loss": 1.8049,
      "step": 5360
    },
    {
      "epoch": 1.4350614644575095,
      "grad_norm": 1.1769757270812988,
      "learning_rate": 0.00019000823481078228,
      "loss": 1.8204,
      "step": 5370
    },
    {
      "epoch": 1.4377338321753073,
      "grad_norm": 1.2379077672958374,
      "learning_rate": 0.00018997162227641203,
      "loss": 1.8298,
      "step": 5380
    },
    {
      "epoch": 1.4404061998931053,
      "grad_norm": 1.1044251918792725,
      "learning_rate": 0.0001899349463262083,
      "loss": 1.9136,
      "step": 5390
    },
    {
      "epoch": 1.4430785676109033,
      "grad_norm": 1.0281884670257568,
      "learning_rate": 0.00018989820698602183,
      "loss": 1.846,
      "step": 5400
    },
    {
      "epoch": 1.4457509353287012,
      "grad_norm": 1.1197541952133179,
      "learning_rate": 0.0001898614042817481,
      "loss": 1.8016,
      "step": 5410
    },
    {
      "epoch": 1.4484233030464992,
      "grad_norm": 1.0418063402175903,
      "learning_rate": 0.00018982453823932722,
      "loss": 1.8255,
      "step": 5420
    },
    {
      "epoch": 1.4510956707642972,
      "grad_norm": 1.0735664367675781,
      "learning_rate": 0.00018978760888474397,
      "loss": 1.8143,
      "step": 5430
    },
    {
      "epoch": 1.453768038482095,
      "grad_norm": 1.1447256803512573,
      "learning_rate": 0.00018975061624402768,
      "loss": 1.8316,
      "step": 5440
    },
    {
      "epoch": 1.456440406199893,
      "grad_norm": 1.2337983846664429,
      "learning_rate": 0.0001897135603432524,
      "loss": 1.8917,
      "step": 5450
    },
    {
      "epoch": 1.459112773917691,
      "grad_norm": 1.085819959640503,
      "learning_rate": 0.00018967644120853667,
      "loss": 1.8752,
      "step": 5460
    },
    {
      "epoch": 1.461785141635489,
      "grad_norm": 1.1256166696548462,
      "learning_rate": 0.00018963925886604366,
      "loss": 1.9224,
      "step": 5470
    },
    {
      "epoch": 1.464457509353287,
      "grad_norm": 1.051095724105835,
      "learning_rate": 0.00018960201334198104,
      "loss": 1.7973,
      "step": 5480
    },
    {
      "epoch": 1.467129877071085,
      "grad_norm": 1.1790239810943604,
      "learning_rate": 0.00018956470466260106,
      "loss": 1.8763,
      "step": 5490
    },
    {
      "epoch": 1.469802244788883,
      "grad_norm": 1.0808390378952026,
      "learning_rate": 0.0001895273328542005,
      "loss": 1.8333,
      "step": 5500
    },
    {
      "epoch": 1.472474612506681,
      "grad_norm": 1.176282525062561,
      "learning_rate": 0.0001894898979431205,
      "loss": 1.8483,
      "step": 5510
    },
    {
      "epoch": 1.475146980224479,
      "grad_norm": 1.1892223358154297,
      "learning_rate": 0.0001894523999557469,
      "loss": 1.8969,
      "step": 5520
    },
    {
      "epoch": 1.477819347942277,
      "grad_norm": 1.2082583904266357,
      "learning_rate": 0.00018941483891850983,
      "loss": 1.9246,
      "step": 5530
    },
    {
      "epoch": 1.4804917156600748,
      "grad_norm": 1.0708314180374146,
      "learning_rate": 0.00018937721485788388,
      "loss": 1.8795,
      "step": 5540
    },
    {
      "epoch": 1.4831640833778728,
      "grad_norm": 1.1821789741516113,
      "learning_rate": 0.00018933952780038815,
      "loss": 1.8206,
      "step": 5550
    },
    {
      "epoch": 1.4858364510956708,
      "grad_norm": 1.065481424331665,
      "learning_rate": 0.00018930177777258604,
      "loss": 1.8698,
      "step": 5560
    },
    {
      "epoch": 1.4885088188134687,
      "grad_norm": 1.157880425453186,
      "learning_rate": 0.00018926396480108538,
      "loss": 1.8601,
      "step": 5570
    },
    {
      "epoch": 1.4911811865312667,
      "grad_norm": 1.1690573692321777,
      "learning_rate": 0.00018922608891253843,
      "loss": 1.8147,
      "step": 5580
    },
    {
      "epoch": 1.4938535542490647,
      "grad_norm": 1.0936037302017212,
      "learning_rate": 0.00018918815013364162,
      "loss": 1.8651,
      "step": 5590
    },
    {
      "epoch": 1.4965259219668625,
      "grad_norm": 1.087705135345459,
      "learning_rate": 0.0001891501484911359,
      "loss": 1.8107,
      "step": 5600
    },
    {
      "epoch": 1.4991982896846605,
      "grad_norm": 1.171329140663147,
      "learning_rate": 0.00018911208401180642,
      "loss": 1.9208,
      "step": 5610
    },
    {
      "epoch": 1.5018706574024585,
      "grad_norm": 1.0548722743988037,
      "learning_rate": 0.00018907395672248267,
      "loss": 1.8568,
      "step": 5620
    },
    {
      "epoch": 1.5045430251202565,
      "grad_norm": 1.0758652687072754,
      "learning_rate": 0.00018903576665003832,
      "loss": 1.8277,
      "step": 5630
    },
    {
      "epoch": 1.5072153928380545,
      "grad_norm": 1.0779211521148682,
      "learning_rate": 0.00018899751382139143,
      "loss": 1.8602,
      "step": 5640
    },
    {
      "epoch": 1.5098877605558525,
      "grad_norm": 1.063739538192749,
      "learning_rate": 0.0001889591982635042,
      "loss": 1.8677,
      "step": 5650
    },
    {
      "epoch": 1.5125601282736505,
      "grad_norm": 1.0633587837219238,
      "learning_rate": 0.0001889208200033831,
      "loss": 1.8956,
      "step": 5660
    },
    {
      "epoch": 1.5152324959914485,
      "grad_norm": 1.100932240486145,
      "learning_rate": 0.0001888823790680787,
      "loss": 1.816,
      "step": 5670
    },
    {
      "epoch": 1.5179048637092465,
      "grad_norm": 1.0497140884399414,
      "learning_rate": 0.00018884387548468587,
      "loss": 1.883,
      "step": 5680
    },
    {
      "epoch": 1.5205772314270445,
      "grad_norm": 1.1753191947937012,
      "learning_rate": 0.00018880530928034353,
      "loss": 1.8177,
      "step": 5690
    },
    {
      "epoch": 1.5232495991448425,
      "grad_norm": 1.0253231525421143,
      "learning_rate": 0.00018876668048223478,
      "loss": 1.8721,
      "step": 5700
    },
    {
      "epoch": 1.5259219668626403,
      "grad_norm": 1.1670130491256714,
      "learning_rate": 0.0001887279891175869,
      "loss": 1.8963,
      "step": 5710
    },
    {
      "epoch": 1.5285943345804383,
      "grad_norm": 1.018936038017273,
      "learning_rate": 0.00018868923521367118,
      "loss": 1.8449,
      "step": 5720
    },
    {
      "epoch": 1.5312667022982362,
      "grad_norm": 1.0974923372268677,
      "learning_rate": 0.000188650418797803,
      "loss": 1.9107,
      "step": 5730
    },
    {
      "epoch": 1.5339390700160342,
      "grad_norm": 1.0885361433029175,
      "learning_rate": 0.00018861153989734185,
      "loss": 1.8306,
      "step": 5740
    },
    {
      "epoch": 1.536611437733832,
      "grad_norm": 1.2174198627471924,
      "learning_rate": 0.00018857259853969125,
      "loss": 1.8493,
      "step": 5750
    },
    {
      "epoch": 1.53928380545163,
      "grad_norm": 1.1191515922546387,
      "learning_rate": 0.0001885335947522987,
      "loss": 1.7669,
      "step": 5760
    },
    {
      "epoch": 1.541956173169428,
      "grad_norm": 1.0823826789855957,
      "learning_rate": 0.00018849452856265574,
      "loss": 1.8432,
      "step": 5770
    },
    {
      "epoch": 1.544628540887226,
      "grad_norm": 1.1358907222747803,
      "learning_rate": 0.00018845539999829788,
      "loss": 1.8698,
      "step": 5780
    },
    {
      "epoch": 1.547300908605024,
      "grad_norm": 1.1223632097244263,
      "learning_rate": 0.00018841620908680463,
      "loss": 1.8296,
      "step": 5790
    },
    {
      "epoch": 1.549973276322822,
      "grad_norm": 1.072277545928955,
      "learning_rate": 0.00018837695585579938,
      "loss": 1.8513,
      "step": 5800
    },
    {
      "epoch": 1.55264564404062,
      "grad_norm": 1.0672662258148193,
      "learning_rate": 0.00018833764033294946,
      "loss": 1.8718,
      "step": 5810
    },
    {
      "epoch": 1.555318011758418,
      "grad_norm": 1.0297456979751587,
      "learning_rate": 0.00018829826254596621,
      "loss": 1.8659,
      "step": 5820
    },
    {
      "epoch": 1.557990379476216,
      "grad_norm": 1.0709869861602783,
      "learning_rate": 0.00018825882252260467,
      "loss": 1.7967,
      "step": 5830
    },
    {
      "epoch": 1.560662747194014,
      "grad_norm": 1.1051650047302246,
      "learning_rate": 0.0001882193202906639,
      "loss": 1.838,
      "step": 5840
    },
    {
      "epoch": 1.563335114911812,
      "grad_norm": 1.1122859716415405,
      "learning_rate": 0.0001881797558779868,
      "loss": 1.8849,
      "step": 5850
    },
    {
      "epoch": 1.56600748262961,
      "grad_norm": 1.1311308145523071,
      "learning_rate": 0.00018814012931246,
      "loss": 1.871,
      "step": 5860
    },
    {
      "epoch": 1.5686798503474078,
      "grad_norm": 1.0583769083023071,
      "learning_rate": 0.00018810044062201396,
      "loss": 1.8633,
      "step": 5870
    },
    {
      "epoch": 1.5713522180652058,
      "grad_norm": 1.008792519569397,
      "learning_rate": 0.00018806068983462303,
      "loss": 1.9072,
      "step": 5880
    },
    {
      "epoch": 1.5740245857830037,
      "grad_norm": 1.4484059810638428,
      "learning_rate": 0.00018802087697830522,
      "loss": 1.8135,
      "step": 5890
    },
    {
      "epoch": 1.5766969535008017,
      "grad_norm": 1.1241083145141602,
      "learning_rate": 0.00018798100208112234,
      "loss": 1.7813,
      "step": 5900
    },
    {
      "epoch": 1.5793693212185995,
      "grad_norm": 1.1186484098434448,
      "learning_rate": 0.00018794106517117993,
      "loss": 1.8517,
      "step": 5910
    },
    {
      "epoch": 1.5820416889363975,
      "grad_norm": 1.09034264087677,
      "learning_rate": 0.0001879010662766272,
      "loss": 1.8193,
      "step": 5920
    },
    {
      "epoch": 1.5847140566541955,
      "grad_norm": 1.064842700958252,
      "learning_rate": 0.0001878610054256571,
      "loss": 1.8343,
      "step": 5930
    },
    {
      "epoch": 1.5873864243719935,
      "grad_norm": 1.1788545846939087,
      "learning_rate": 0.00018782088264650618,
      "loss": 1.8777,
      "step": 5940
    },
    {
      "epoch": 1.5900587920897915,
      "grad_norm": 1.1238136291503906,
      "learning_rate": 0.00018778069796745474,
      "loss": 1.8223,
      "step": 5950
    },
    {
      "epoch": 1.5927311598075895,
      "grad_norm": 1.1837947368621826,
      "learning_rate": 0.00018774045141682662,
      "loss": 1.8243,
      "step": 5960
    },
    {
      "epoch": 1.5954035275253875,
      "grad_norm": 1.0124212503433228,
      "learning_rate": 0.00018770014302298932,
      "loss": 1.7579,
      "step": 5970
    },
    {
      "epoch": 1.5980758952431855,
      "grad_norm": 1.0607494115829468,
      "learning_rate": 0.00018765977281435392,
      "loss": 1.885,
      "step": 5980
    },
    {
      "epoch": 1.6007482629609835,
      "grad_norm": 1.0666025876998901,
      "learning_rate": 0.00018761934081937504,
      "loss": 1.9027,
      "step": 5990
    },
    {
      "epoch": 1.6034206306787815,
      "grad_norm": 1.1201056241989136,
      "learning_rate": 0.00018757884706655087,
      "loss": 1.8567,
      "step": 6000
    },
    {
      "epoch": 1.6060929983965795,
      "grad_norm": 1.068117380142212,
      "learning_rate": 0.00018753829158442314,
      "loss": 1.8151,
      "step": 6010
    },
    {
      "epoch": 1.6087653661143775,
      "grad_norm": 1.1987791061401367,
      "learning_rate": 0.0001874976744015771,
      "loss": 1.9186,
      "step": 6020
    },
    {
      "epoch": 1.6114377338321753,
      "grad_norm": 1.2409660816192627,
      "learning_rate": 0.00018745699554664148,
      "loss": 1.869,
      "step": 6030
    },
    {
      "epoch": 1.6141101015499733,
      "grad_norm": 1.100341558456421,
      "learning_rate": 0.00018741625504828843,
      "loss": 1.8966,
      "step": 6040
    },
    {
      "epoch": 1.6167824692677712,
      "grad_norm": 1.1203241348266602,
      "learning_rate": 0.00018737545293523363,
      "loss": 1.7927,
      "step": 6050
    },
    {
      "epoch": 1.6194548369855692,
      "grad_norm": 1.1305533647537231,
      "learning_rate": 0.0001873345892362361,
      "loss": 1.7844,
      "step": 6060
    },
    {
      "epoch": 1.622127204703367,
      "grad_norm": 1.1537564992904663,
      "learning_rate": 0.00018729366398009836,
      "loss": 1.928,
      "step": 6070
    },
    {
      "epoch": 1.624799572421165,
      "grad_norm": 1.112511396408081,
      "learning_rate": 0.00018725267719566627,
      "loss": 1.7971,
      "step": 6080
    },
    {
      "epoch": 1.627471940138963,
      "grad_norm": 1.2046387195587158,
      "learning_rate": 0.00018721162891182903,
      "loss": 1.8557,
      "step": 6090
    },
    {
      "epoch": 1.630144307856761,
      "grad_norm": 1.129660964012146,
      "learning_rate": 0.00018717051915751926,
      "loss": 1.8919,
      "step": 6100
    },
    {
      "epoch": 1.632816675574559,
      "grad_norm": 1.0578484535217285,
      "learning_rate": 0.00018712934796171285,
      "loss": 1.8957,
      "step": 6110
    },
    {
      "epoch": 1.635489043292357,
      "grad_norm": 1.1723358631134033,
      "learning_rate": 0.00018708811535342898,
      "loss": 1.9637,
      "step": 6120
    },
    {
      "epoch": 1.638161411010155,
      "grad_norm": 0.9757607579231262,
      "learning_rate": 0.0001870468213617302,
      "loss": 1.8283,
      "step": 6130
    },
    {
      "epoch": 1.640833778727953,
      "grad_norm": 1.2090801000595093,
      "learning_rate": 0.00018700546601572223,
      "loss": 1.8349,
      "step": 6140
    },
    {
      "epoch": 1.643506146445751,
      "grad_norm": 1.1274935007095337,
      "learning_rate": 0.00018696404934455415,
      "loss": 1.9076,
      "step": 6150
    },
    {
      "epoch": 1.646178514163549,
      "grad_norm": 1.1622339487075806,
      "learning_rate": 0.00018692257137741813,
      "loss": 1.9078,
      "step": 6160
    },
    {
      "epoch": 1.648850881881347,
      "grad_norm": 1.0929253101348877,
      "learning_rate": 0.00018688103214354964,
      "loss": 1.8216,
      "step": 6170
    },
    {
      "epoch": 1.651523249599145,
      "grad_norm": 1.1966350078582764,
      "learning_rate": 0.0001868394316722273,
      "loss": 1.8581,
      "step": 6180
    },
    {
      "epoch": 1.6541956173169428,
      "grad_norm": 1.1206058263778687,
      "learning_rate": 0.00018679776999277285,
      "loss": 1.9012,
      "step": 6190
    },
    {
      "epoch": 1.6568679850347408,
      "grad_norm": 1.1007308959960938,
      "learning_rate": 0.00018675604713455125,
      "loss": 1.8622,
      "step": 6200
    },
    {
      "epoch": 1.6595403527525387,
      "grad_norm": 1.0664573907852173,
      "learning_rate": 0.00018671426312697059,
      "loss": 1.8181,
      "step": 6210
    },
    {
      "epoch": 1.6622127204703367,
      "grad_norm": 1.0425337553024292,
      "learning_rate": 0.00018667241799948196,
      "loss": 1.8398,
      "step": 6220
    },
    {
      "epoch": 1.6648850881881345,
      "grad_norm": 1.132024884223938,
      "learning_rate": 0.00018663051178157965,
      "loss": 1.8853,
      "step": 6230
    },
    {
      "epoch": 1.6675574559059325,
      "grad_norm": 1.1238131523132324,
      "learning_rate": 0.00018658854450280085,
      "loss": 1.8376,
      "step": 6240
    },
    {
      "epoch": 1.6702298236237305,
      "grad_norm": 1.1062246561050415,
      "learning_rate": 0.00018654651619272596,
      "loss": 1.8883,
      "step": 6250
    },
    {
      "epoch": 1.6729021913415285,
      "grad_norm": 1.114938497543335,
      "learning_rate": 0.0001865044268809783,
      "loss": 1.8638,
      "step": 6260
    },
    {
      "epoch": 1.6755745590593265,
      "grad_norm": 1.092271089553833,
      "learning_rate": 0.0001864622765972243,
      "loss": 1.8254,
      "step": 6270
    },
    {
      "epoch": 1.6782469267771245,
      "grad_norm": 1.424766182899475,
      "learning_rate": 0.0001864200653711731,
      "loss": 1.8943,
      "step": 6280
    },
    {
      "epoch": 1.6809192944949225,
      "grad_norm": 1.1574349403381348,
      "learning_rate": 0.0001863777932325771,
      "loss": 1.9185,
      "step": 6290
    },
    {
      "epoch": 1.6835916622127205,
      "grad_norm": 1.229568600654602,
      "learning_rate": 0.00018633546021123147,
      "loss": 1.8111,
      "step": 6300
    },
    {
      "epoch": 1.6862640299305185,
      "grad_norm": 1.1014653444290161,
      "learning_rate": 0.00018629306633697433,
      "loss": 1.8791,
      "step": 6310
    },
    {
      "epoch": 1.6889363976483165,
      "grad_norm": 1.1644926071166992,
      "learning_rate": 0.0001862506116396867,
      "loss": 1.8766,
      "step": 6320
    },
    {
      "epoch": 1.6916087653661145,
      "grad_norm": 1.1016088724136353,
      "learning_rate": 0.00018620809614929243,
      "loss": 1.8591,
      "step": 6330
    },
    {
      "epoch": 1.6942811330839125,
      "grad_norm": 1.0894542932510376,
      "learning_rate": 0.00018616551989575827,
      "loss": 1.9127,
      "step": 6340
    },
    {
      "epoch": 1.6969535008017105,
      "grad_norm": 1.157467246055603,
      "learning_rate": 0.00018612288290909378,
      "loss": 1.8365,
      "step": 6350
    },
    {
      "epoch": 1.6996258685195083,
      "grad_norm": 1.1508746147155762,
      "learning_rate": 0.0001860801852193513,
      "loss": 1.8592,
      "step": 6360
    },
    {
      "epoch": 1.7022982362373063,
      "grad_norm": 1.2317012548446655,
      "learning_rate": 0.00018603742685662606,
      "loss": 1.867,
      "step": 6370
    },
    {
      "epoch": 1.7049706039551042,
      "grad_norm": 1.1163196563720703,
      "learning_rate": 0.00018599460785105587,
      "loss": 1.9096,
      "step": 6380
    },
    {
      "epoch": 1.7076429716729022,
      "grad_norm": 1.0866283178329468,
      "learning_rate": 0.00018595172823282147,
      "loss": 1.904,
      "step": 6390
    },
    {
      "epoch": 1.7103153393907,
      "grad_norm": 1.0890675783157349,
      "learning_rate": 0.00018590878803214617,
      "loss": 1.8342,
      "step": 6400
    },
    {
      "epoch": 1.712987707108498,
      "grad_norm": 1.1632815599441528,
      "learning_rate": 0.0001858657872792961,
      "loss": 1.806,
      "step": 6410
    },
    {
      "epoch": 1.715660074826296,
      "grad_norm": 1.287760853767395,
      "learning_rate": 0.00018582272600458003,
      "loss": 1.8504,
      "step": 6420
    },
    {
      "epoch": 1.718332442544094,
      "grad_norm": 1.004636287689209,
      "learning_rate": 0.00018577960423834936,
      "loss": 1.9219,
      "step": 6430
    },
    {
      "epoch": 1.721004810261892,
      "grad_norm": 1.1246731281280518,
      "learning_rate": 0.00018573642201099815,
      "loss": 1.8198,
      "step": 6440
    },
    {
      "epoch": 1.72367717797969,
      "grad_norm": 1.2009623050689697,
      "learning_rate": 0.00018569317935296308,
      "loss": 1.823,
      "step": 6450
    },
    {
      "epoch": 1.726349545697488,
      "grad_norm": 1.147096037864685,
      "learning_rate": 0.00018564987629472347,
      "loss": 1.8457,
      "step": 6460
    },
    {
      "epoch": 1.729021913415286,
      "grad_norm": 1.1151899099349976,
      "learning_rate": 0.00018560651286680103,
      "loss": 1.8493,
      "step": 6470
    },
    {
      "epoch": 1.731694281133084,
      "grad_norm": 1.1140600442886353,
      "learning_rate": 0.00018556308909976028,
      "loss": 1.9133,
      "step": 6480
    },
    {
      "epoch": 1.734366648850882,
      "grad_norm": 1.0503792762756348,
      "learning_rate": 0.00018551960502420807,
      "loss": 1.8455,
      "step": 6490
    },
    {
      "epoch": 1.73703901656868,
      "grad_norm": 1.1020917892456055,
      "learning_rate": 0.0001854760606707939,
      "loss": 1.7742,
      "step": 6500
    },
    {
      "epoch": 1.739711384286478,
      "grad_norm": 1.0384830236434937,
      "learning_rate": 0.0001854324560702096,
      "loss": 1.9241,
      "step": 6510
    },
    {
      "epoch": 1.7423837520042758,
      "grad_norm": 1.0608861446380615,
      "learning_rate": 0.00018538879125318958,
      "loss": 1.8665,
      "step": 6520
    },
    {
      "epoch": 1.7450561197220738,
      "grad_norm": 1.1713470220565796,
      "learning_rate": 0.00018534506625051067,
      "loss": 1.7895,
      "step": 6530
    },
    {
      "epoch": 1.7477284874398717,
      "grad_norm": 1.0618896484375,
      "learning_rate": 0.0001853012810929921,
      "loss": 1.8779,
      "step": 6540
    },
    {
      "epoch": 1.7504008551576697,
      "grad_norm": 1.171653151512146,
      "learning_rate": 0.0001852574358114956,
      "loss": 1.8838,
      "step": 6550
    },
    {
      "epoch": 1.7530732228754675,
      "grad_norm": 1.2031880617141724,
      "learning_rate": 0.0001852135304369251,
      "loss": 1.9069,
      "step": 6560
    },
    {
      "epoch": 1.7557455905932655,
      "grad_norm": 1.048595666885376,
      "learning_rate": 0.00018516956500022697,
      "loss": 1.8772,
      "step": 6570
    },
    {
      "epoch": 1.7584179583110635,
      "grad_norm": 1.0669938325881958,
      "learning_rate": 0.00018512553953238997,
      "loss": 1.8379,
      "step": 6580
    },
    {
      "epoch": 1.7610903260288615,
      "grad_norm": 1.0816853046417236,
      "learning_rate": 0.0001850814540644451,
      "loss": 1.8266,
      "step": 6590
    },
    {
      "epoch": 1.7637626937466595,
      "grad_norm": 1.0739929676055908,
      "learning_rate": 0.00018503730862746574,
      "loss": 1.8376,
      "step": 6600
    },
    {
      "epoch": 1.7664350614644575,
      "grad_norm": 1.144262671470642,
      "learning_rate": 0.0001849931032525674,
      "loss": 1.854,
      "step": 6610
    },
    {
      "epoch": 1.7691074291822555,
      "grad_norm": 1.1057103872299194,
      "learning_rate": 0.00018494883797090793,
      "loss": 1.9118,
      "step": 6620
    },
    {
      "epoch": 1.7717797969000535,
      "grad_norm": 1.1595027446746826,
      "learning_rate": 0.0001849045128136874,
      "loss": 1.8516,
      "step": 6630
    },
    {
      "epoch": 1.7744521646178515,
      "grad_norm": 1.1505125761032104,
      "learning_rate": 0.00018486012781214806,
      "loss": 1.8469,
      "step": 6640
    },
    {
      "epoch": 1.7771245323356495,
      "grad_norm": 1.2194557189941406,
      "learning_rate": 0.00018481568299757436,
      "loss": 1.8613,
      "step": 6650
    },
    {
      "epoch": 1.7797969000534475,
      "grad_norm": 1.0633267164230347,
      "learning_rate": 0.0001847711784012929,
      "loss": 1.8683,
      "step": 6660
    },
    {
      "epoch": 1.7824692677712455,
      "grad_norm": 1.0982544422149658,
      "learning_rate": 0.0001847266140546724,
      "loss": 1.8655,
      "step": 6670
    },
    {
      "epoch": 1.7851416354890433,
      "grad_norm": 1.1030447483062744,
      "learning_rate": 0.0001846819899891237,
      "loss": 1.7817,
      "step": 6680
    },
    {
      "epoch": 1.7878140032068413,
      "grad_norm": 1.08018159866333,
      "learning_rate": 0.0001846373062360998,
      "loss": 1.8533,
      "step": 6690
    },
    {
      "epoch": 1.7904863709246392,
      "grad_norm": 1.2506588697433472,
      "learning_rate": 0.00018459256282709566,
      "loss": 1.8723,
      "step": 6700
    },
    {
      "epoch": 1.7931587386424372,
      "grad_norm": 1.0448452234268188,
      "learning_rate": 0.00018454775979364837,
      "loss": 1.7822,
      "step": 6710
    },
    {
      "epoch": 1.795831106360235,
      "grad_norm": 1.2007098197937012,
      "learning_rate": 0.000184502897167337,
      "loss": 1.8141,
      "step": 6720
    },
    {
      "epoch": 1.798503474078033,
      "grad_norm": 1.0861321687698364,
      "learning_rate": 0.00018445797497978266,
      "loss": 1.8231,
      "step": 6730
    },
    {
      "epoch": 1.801175841795831,
      "grad_norm": 1.1015934944152832,
      "learning_rate": 0.00018441299326264838,
      "loss": 1.8457,
      "step": 6740
    },
    {
      "epoch": 1.803848209513629,
      "grad_norm": 1.0648318529129028,
      "learning_rate": 0.00018436795204763922,
      "loss": 1.843,
      "step": 6750
    },
    {
      "epoch": 1.806520577231427,
      "grad_norm": 1.1390804052352905,
      "learning_rate": 0.00018432285136650215,
      "loss": 1.8092,
      "step": 6760
    },
    {
      "epoch": 1.809192944949225,
      "grad_norm": 1.3198267221450806,
      "learning_rate": 0.00018427769125102606,
      "loss": 1.8894,
      "step": 6770
    },
    {
      "epoch": 1.811865312667023,
      "grad_norm": 1.1119917631149292,
      "learning_rate": 0.00018423247173304173,
      "loss": 1.8626,
      "step": 6780
    },
    {
      "epoch": 1.814537680384821,
      "grad_norm": 1.1425055265426636,
      "learning_rate": 0.00018418719284442176,
      "loss": 1.9254,
      "step": 6790
    },
    {
      "epoch": 1.817210048102619,
      "grad_norm": 1.2299972772598267,
      "learning_rate": 0.00018414185461708066,
      "loss": 1.8001,
      "step": 6800
    },
    {
      "epoch": 1.819882415820417,
      "grad_norm": 1.1462756395339966,
      "learning_rate": 0.00018409645708297476,
      "loss": 1.7814,
      "step": 6810
    },
    {
      "epoch": 1.822554783538215,
      "grad_norm": 1.1987062692642212,
      "learning_rate": 0.00018405100027410218,
      "loss": 1.8232,
      "step": 6820
    },
    {
      "epoch": 1.825227151256013,
      "grad_norm": 1.0986485481262207,
      "learning_rate": 0.00018400548422250278,
      "loss": 1.796,
      "step": 6830
    },
    {
      "epoch": 1.8278995189738108,
      "grad_norm": 1.223432183265686,
      "learning_rate": 0.00018395990896025825,
      "loss": 1.8445,
      "step": 6840
    },
    {
      "epoch": 1.8305718866916088,
      "grad_norm": 1.1204471588134766,
      "learning_rate": 0.000183914274519492,
      "loss": 1.7464,
      "step": 6850
    },
    {
      "epoch": 1.8332442544094067,
      "grad_norm": 1.1691845655441284,
      "learning_rate": 0.00018386858093236906,
      "loss": 1.8865,
      "step": 6860
    },
    {
      "epoch": 1.8359166221272047,
      "grad_norm": 1.2536617517471313,
      "learning_rate": 0.00018382282823109624,
      "loss": 1.8843,
      "step": 6870
    },
    {
      "epoch": 1.8385889898450025,
      "grad_norm": 1.1956285238265991,
      "learning_rate": 0.000183777016447922,
      "loss": 1.898,
      "step": 6880
    },
    {
      "epoch": 1.8412613575628005,
      "grad_norm": 1.0576280355453491,
      "learning_rate": 0.00018373114561513645,
      "loss": 1.8173,
      "step": 6890
    },
    {
      "epoch": 1.8439337252805985,
      "grad_norm": 1.1451317071914673,
      "learning_rate": 0.00018368521576507128,
      "loss": 1.8351,
      "step": 6900
    },
    {
      "epoch": 1.8466060929983965,
      "grad_norm": 1.1110111474990845,
      "learning_rate": 0.00018363922693009988,
      "loss": 1.8865,
      "step": 6910
    },
    {
      "epoch": 1.8492784607161945,
      "grad_norm": 1.1600598096847534,
      "learning_rate": 0.00018359317914263702,
      "loss": 1.839,
      "step": 6920
    },
    {
      "epoch": 1.8519508284339925,
      "grad_norm": 1.1117746829986572,
      "learning_rate": 0.00018354707243513926,
      "loss": 1.8191,
      "step": 6930
    },
    {
      "epoch": 1.8546231961517905,
      "grad_norm": 1.1108200550079346,
      "learning_rate": 0.0001835009068401045,
      "loss": 1.8204,
      "step": 6940
    },
    {
      "epoch": 1.8572955638695885,
      "grad_norm": 1.1949942111968994,
      "learning_rate": 0.00018345468239007223,
      "loss": 1.868,
      "step": 6950
    },
    {
      "epoch": 1.8599679315873865,
      "grad_norm": 1.1258621215820312,
      "learning_rate": 0.00018340839911762342,
      "loss": 1.8301,
      "step": 6960
    },
    {
      "epoch": 1.8626402993051845,
      "grad_norm": 1.1038228273391724,
      "learning_rate": 0.0001833620570553805,
      "loss": 1.855,
      "step": 6970
    },
    {
      "epoch": 1.8653126670229825,
      "grad_norm": 1.0757447481155396,
      "learning_rate": 0.0001833156562360073,
      "loss": 1.8098,
      "step": 6980
    },
    {
      "epoch": 1.8679850347407805,
      "grad_norm": 1.1476863622665405,
      "learning_rate": 0.00018326919669220913,
      "loss": 1.8136,
      "step": 6990
    },
    {
      "epoch": 1.8706574024585783,
      "grad_norm": 1.1160389184951782,
      "learning_rate": 0.00018322267845673266,
      "loss": 1.8543,
      "step": 7000
    },
    {
      "epoch": 1.8733297701763763,
      "grad_norm": 1.2721679210662842,
      "learning_rate": 0.00018317610156236592,
      "loss": 1.8373,
      "step": 7010
    },
    {
      "epoch": 1.8760021378941742,
      "grad_norm": 1.2080637216567993,
      "learning_rate": 0.00018312946604193825,
      "loss": 1.884,
      "step": 7020
    },
    {
      "epoch": 1.8786745056119722,
      "grad_norm": 1.0939440727233887,
      "learning_rate": 0.00018308277192832038,
      "loss": 1.8406,
      "step": 7030
    },
    {
      "epoch": 1.88134687332977,
      "grad_norm": 1.1047283411026,
      "learning_rate": 0.00018303601925442436,
      "loss": 1.8284,
      "step": 7040
    },
    {
      "epoch": 1.884019241047568,
      "grad_norm": 1.0662176609039307,
      "learning_rate": 0.00018298920805320337,
      "loss": 1.8719,
      "step": 7050
    },
    {
      "epoch": 1.886691608765366,
      "grad_norm": 1.106064796447754,
      "learning_rate": 0.00018294233835765202,
      "loss": 1.832,
      "step": 7060
    },
    {
      "epoch": 1.889363976483164,
      "grad_norm": 1.2257906198501587,
      "learning_rate": 0.00018289541020080602,
      "loss": 1.8272,
      "step": 7070
    },
    {
      "epoch": 1.892036344200962,
      "grad_norm": 1.1595290899276733,
      "learning_rate": 0.00018284842361574236,
      "loss": 1.8415,
      "step": 7080
    },
    {
      "epoch": 1.89470871191876,
      "grad_norm": 1.028878927230835,
      "learning_rate": 0.00018280137863557918,
      "loss": 1.8264,
      "step": 7090
    },
    {
      "epoch": 1.897381079636558,
      "grad_norm": 1.1327093839645386,
      "learning_rate": 0.0001827542752934758,
      "loss": 1.7834,
      "step": 7100
    },
    {
      "epoch": 1.900053447354356,
      "grad_norm": 1.1777125597000122,
      "learning_rate": 0.00018270711362263262,
      "loss": 1.8784,
      "step": 7110
    },
    {
      "epoch": 1.902725815072154,
      "grad_norm": 1.0794941186904907,
      "learning_rate": 0.00018265989365629125,
      "loss": 1.9329,
      "step": 7120
    },
    {
      "epoch": 1.905398182789952,
      "grad_norm": 1.4766443967819214,
      "learning_rate": 0.0001826126154277343,
      "loss": 1.9082,
      "step": 7130
    },
    {
      "epoch": 1.90807055050775,
      "grad_norm": 1.0953209400177002,
      "learning_rate": 0.00018256527897028547,
      "loss": 1.9247,
      "step": 7140
    },
    {
      "epoch": 1.910742918225548,
      "grad_norm": 1.0441701412200928,
      "learning_rate": 0.00018251788431730955,
      "loss": 1.8696,
      "step": 7150
    },
    {
      "epoch": 1.9134152859433458,
      "grad_norm": 1.1116993427276611,
      "learning_rate": 0.0001824704315022123,
      "loss": 1.8222,
      "step": 7160
    },
    {
      "epoch": 1.9160876536611438,
      "grad_norm": 1.1172654628753662,
      "learning_rate": 0.00018242292055844047,
      "loss": 1.9006,
      "step": 7170
    },
    {
      "epoch": 1.9187600213789417,
      "grad_norm": 1.1256300210952759,
      "learning_rate": 0.0001823753515194818,
      "loss": 1.8,
      "step": 7180
    },
    {
      "epoch": 1.9214323890967397,
      "grad_norm": 1.1486270427703857,
      "learning_rate": 0.00018232772441886501,
      "loss": 1.9045,
      "step": 7190
    },
    {
      "epoch": 1.9241047568145375,
      "grad_norm": 1.1202726364135742,
      "learning_rate": 0.00018228003929015966,
      "loss": 1.7617,
      "step": 7200
    },
    {
      "epoch": 1.9267771245323355,
      "grad_norm": 1.1803336143493652,
      "learning_rate": 0.0001822322961669763,
      "loss": 1.8798,
      "step": 7210
    },
    {
      "epoch": 1.9294494922501335,
      "grad_norm": 1.1806000471115112,
      "learning_rate": 0.0001821844950829663,
      "loss": 1.8395,
      "step": 7220
    },
    {
      "epoch": 1.9321218599679315,
      "grad_norm": 1.2114359140396118,
      "learning_rate": 0.0001821366360718219,
      "loss": 1.8914,
      "step": 7230
    },
    {
      "epoch": 1.9347942276857295,
      "grad_norm": 1.1424024105072021,
      "learning_rate": 0.00018208871916727617,
      "loss": 1.8854,
      "step": 7240
    },
    {
      "epoch": 1.9374665954035275,
      "grad_norm": 1.1180267333984375,
      "learning_rate": 0.000182040744403103,
      "loss": 1.8004,
      "step": 7250
    },
    {
      "epoch": 1.9401389631213255,
      "grad_norm": 1.235189437866211,
      "learning_rate": 0.000181992711813117,
      "loss": 1.8273,
      "step": 7260
    },
    {
      "epoch": 1.9428113308391235,
      "grad_norm": 1.2162269353866577,
      "learning_rate": 0.00018194462143117366,
      "loss": 1.8169,
      "step": 7270
    },
    {
      "epoch": 1.9454836985569215,
      "grad_norm": 1.2939962148666382,
      "learning_rate": 0.0001818964732911691,
      "loss": 1.8331,
      "step": 7280
    },
    {
      "epoch": 1.9481560662747195,
      "grad_norm": 1.1024796962738037,
      "learning_rate": 0.00018184826742704017,
      "loss": 1.85,
      "step": 7290
    },
    {
      "epoch": 1.9508284339925175,
      "grad_norm": 1.2357715368270874,
      "learning_rate": 0.00018180000387276441,
      "loss": 1.8573,
      "step": 7300
    },
    {
      "epoch": 1.9535008017103155,
      "grad_norm": 1.202744960784912,
      "learning_rate": 0.00018175168266236006,
      "loss": 1.7975,
      "step": 7310
    },
    {
      "epoch": 1.9561731694281133,
      "grad_norm": 1.0578547716140747,
      "learning_rate": 0.00018170330382988594,
      "loss": 1.8337,
      "step": 7320
    },
    {
      "epoch": 1.9588455371459113,
      "grad_norm": 1.1348929405212402,
      "learning_rate": 0.0001816548674094415,
      "loss": 1.9244,
      "step": 7330
    },
    {
      "epoch": 1.9615179048637092,
      "grad_norm": 1.1298471689224243,
      "learning_rate": 0.00018160637343516686,
      "loss": 1.7958,
      "step": 7340
    },
    {
      "epoch": 1.9641902725815072,
      "grad_norm": 1.12726628780365,
      "learning_rate": 0.0001815578219412426,
      "loss": 1.8162,
      "step": 7350
    },
    {
      "epoch": 1.966862640299305,
      "grad_norm": 1.2801421880722046,
      "learning_rate": 0.00018150921296188982,
      "loss": 1.7888,
      "step": 7360
    },
    {
      "epoch": 1.969535008017103,
      "grad_norm": 1.093096137046814,
      "learning_rate": 0.00018146054653137033,
      "loss": 1.9479,
      "step": 7370
    },
    {
      "epoch": 1.972207375734901,
      "grad_norm": 1.0839108228683472,
      "learning_rate": 0.0001814118226839862,
      "loss": 1.8465,
      "step": 7380
    },
    {
      "epoch": 1.974879743452699,
      "grad_norm": 1.2250748872756958,
      "learning_rate": 0.0001813630414540801,
      "loss": 1.7888,
      "step": 7390
    },
    {
      "epoch": 1.977552111170497,
      "grad_norm": 1.1333938837051392,
      "learning_rate": 0.00018131420287603516,
      "loss": 1.8968,
      "step": 7400
    },
    {
      "epoch": 1.980224478888295,
      "grad_norm": 1.0982754230499268,
      "learning_rate": 0.0001812653069842748,
      "loss": 1.8264,
      "step": 7410
    },
    {
      "epoch": 1.982896846606093,
      "grad_norm": 1.1105761528015137,
      "learning_rate": 0.00018121635381326305,
      "loss": 1.9022,
      "step": 7420
    },
    {
      "epoch": 1.985569214323891,
      "grad_norm": 1.217867136001587,
      "learning_rate": 0.00018116734339750407,
      "loss": 1.8284,
      "step": 7430
    },
    {
      "epoch": 1.988241582041689,
      "grad_norm": 1.0955591201782227,
      "learning_rate": 0.00018111827577154256,
      "loss": 1.7938,
      "step": 7440
    },
    {
      "epoch": 1.990913949759487,
      "grad_norm": 1.2205541133880615,
      "learning_rate": 0.00018106915096996348,
      "loss": 1.8832,
      "step": 7450
    },
    {
      "epoch": 1.993586317477285,
      "grad_norm": 1.1036415100097656,
      "learning_rate": 0.00018101996902739206,
      "loss": 1.8848,
      "step": 7460
    },
    {
      "epoch": 1.996258685195083,
      "grad_norm": 1.194022297859192,
      "learning_rate": 0.00018097072997849385,
      "loss": 1.8513,
      "step": 7470
    },
    {
      "epoch": 1.9989310529128808,
      "grad_norm": 1.1090449094772339,
      "learning_rate": 0.0001809214338579746,
      "loss": 1.8482,
      "step": 7480
    },
    {
      "epoch": 2.001603420630679,
      "grad_norm": 1.0925272703170776,
      "learning_rate": 0.0001808720807005803,
      "loss": 1.8398,
      "step": 7490
    },
    {
      "epoch": 2.004275788348477,
      "grad_norm": 1.2713534832000732,
      "learning_rate": 0.00018082267054109723,
      "loss": 1.6291,
      "step": 7500
    },
    {
      "epoch": 2.0069481560662745,
      "grad_norm": 1.154667615890503,
      "learning_rate": 0.0001807732034143517,
      "loss": 1.6611,
      "step": 7510
    },
    {
      "epoch": 2.0096205237840725,
      "grad_norm": 1.2427644729614258,
      "learning_rate": 0.00018072367935521028,
      "loss": 1.6878,
      "step": 7520
    },
    {
      "epoch": 2.0122928915018705,
      "grad_norm": 1.2123231887817383,
      "learning_rate": 0.0001806740983985797,
      "loss": 1.659,
      "step": 7530
    },
    {
      "epoch": 2.0149652592196685,
      "grad_norm": 1.1265623569488525,
      "learning_rate": 0.00018062446057940662,
      "loss": 1.6901,
      "step": 7540
    },
    {
      "epoch": 2.0176376269374665,
      "grad_norm": 1.1744213104248047,
      "learning_rate": 0.00018057476593267793,
      "loss": 1.7088,
      "step": 7550
    },
    {
      "epoch": 2.0203099946552645,
      "grad_norm": 1.2280677556991577,
      "learning_rate": 0.00018052501449342055,
      "loss": 1.6548,
      "step": 7560
    },
    {
      "epoch": 2.0229823623730625,
      "grad_norm": 1.3028384447097778,
      "learning_rate": 0.00018047520629670144,
      "loss": 1.6767,
      "step": 7570
    },
    {
      "epoch": 2.0256547300908605,
      "grad_norm": 1.2041252851486206,
      "learning_rate": 0.0001804253413776275,
      "loss": 1.636,
      "step": 7580
    },
    {
      "epoch": 2.0283270978086585,
      "grad_norm": 1.2943586111068726,
      "learning_rate": 0.00018037541977134567,
      "loss": 1.6053,
      "step": 7590
    },
    {
      "epoch": 2.0309994655264565,
      "grad_norm": 1.2200071811676025,
      "learning_rate": 0.0001803254415130428,
      "loss": 1.6664,
      "step": 7600
    },
    {
      "epoch": 2.0336718332442545,
      "grad_norm": 1.2025283575057983,
      "learning_rate": 0.00018027540663794573,
      "loss": 1.6928,
      "step": 7610
    },
    {
      "epoch": 2.0363442009620525,
      "grad_norm": 1.2312372922897339,
      "learning_rate": 0.00018022531518132112,
      "loss": 1.7203,
      "step": 7620
    },
    {
      "epoch": 2.0390165686798505,
      "grad_norm": 1.2150242328643799,
      "learning_rate": 0.00018017516717847563,
      "loss": 1.7097,
      "step": 7630
    },
    {
      "epoch": 2.0416889363976485,
      "grad_norm": 1.2572401762008667,
      "learning_rate": 0.00018012496266475565,
      "loss": 1.6523,
      "step": 7640
    },
    {
      "epoch": 2.0443613041154465,
      "grad_norm": 1.3172168731689453,
      "learning_rate": 0.0001800747016755475,
      "loss": 1.6943,
      "step": 7650
    },
    {
      "epoch": 2.0470336718332445,
      "grad_norm": 1.5790674686431885,
      "learning_rate": 0.00018002438424627725,
      "loss": 1.6717,
      "step": 7660
    },
    {
      "epoch": 2.049706039551042,
      "grad_norm": 1.2752028703689575,
      "learning_rate": 0.0001799740104124108,
      "loss": 1.6941,
      "step": 7670
    },
    {
      "epoch": 2.05237840726884,
      "grad_norm": 1.2330474853515625,
      "learning_rate": 0.00017992358020945372,
      "loss": 1.6986,
      "step": 7680
    },
    {
      "epoch": 2.055050774986638,
      "grad_norm": 1.2675464153289795,
      "learning_rate": 0.0001798730936729514,
      "loss": 1.6786,
      "step": 7690
    },
    {
      "epoch": 2.057723142704436,
      "grad_norm": 1.1552021503448486,
      "learning_rate": 0.0001798225508384889,
      "loss": 1.6453,
      "step": 7700
    },
    {
      "epoch": 2.060395510422234,
      "grad_norm": 1.3470126390457153,
      "learning_rate": 0.00017977195174169098,
      "loss": 1.6547,
      "step": 7710
    },
    {
      "epoch": 2.063067878140032,
      "grad_norm": 1.253616213798523,
      "learning_rate": 0.00017972129641822198,
      "loss": 1.6228,
      "step": 7720
    },
    {
      "epoch": 2.06574024585783,
      "grad_norm": 1.179351806640625,
      "learning_rate": 0.000179670584903786,
      "loss": 1.7211,
      "step": 7730
    },
    {
      "epoch": 2.068412613575628,
      "grad_norm": 1.2924188375473022,
      "learning_rate": 0.00017961981723412662,
      "loss": 1.7298,
      "step": 7740
    },
    {
      "epoch": 2.071084981293426,
      "grad_norm": 1.1614151000976562,
      "learning_rate": 0.0001795689934450271,
      "loss": 1.629,
      "step": 7750
    },
    {
      "epoch": 2.073757349011224,
      "grad_norm": 1.3283143043518066,
      "learning_rate": 0.0001795181135723102,
      "loss": 1.621,
      "step": 7760
    },
    {
      "epoch": 2.076429716729022,
      "grad_norm": 1.2223650217056274,
      "learning_rate": 0.00017946717765183822,
      "loss": 1.6813,
      "step": 7770
    },
    {
      "epoch": 2.07910208444682,
      "grad_norm": 1.2945345640182495,
      "learning_rate": 0.00017941618571951296,
      "loss": 1.6779,
      "step": 7780
    },
    {
      "epoch": 2.081774452164618,
      "grad_norm": 1.2835251092910767,
      "learning_rate": 0.0001793651378112757,
      "loss": 1.7069,
      "step": 7790
    },
    {
      "epoch": 2.084446819882416,
      "grad_norm": 1.2451834678649902,
      "learning_rate": 0.0001793140339631072,
      "loss": 1.6638,
      "step": 7800
    },
    {
      "epoch": 2.087119187600214,
      "grad_norm": 1.2750883102416992,
      "learning_rate": 0.00017926287421102764,
      "loss": 1.718,
      "step": 7810
    },
    {
      "epoch": 2.089791555318012,
      "grad_norm": 1.2635786533355713,
      "learning_rate": 0.00017921165859109656,
      "loss": 1.7728,
      "step": 7820
    },
    {
      "epoch": 2.0924639230358095,
      "grad_norm": 1.2947158813476562,
      "learning_rate": 0.00017916038713941296,
      "loss": 1.6969,
      "step": 7830
    },
    {
      "epoch": 2.0951362907536075,
      "grad_norm": 1.2207590341567993,
      "learning_rate": 0.00017910905989211505,
      "loss": 1.7142,
      "step": 7840
    },
    {
      "epoch": 2.0978086584714055,
      "grad_norm": 1.2223358154296875,
      "learning_rate": 0.0001790576768853806,
      "loss": 1.6306,
      "step": 7850
    },
    {
      "epoch": 2.1004810261892035,
      "grad_norm": 1.2628051042556763,
      "learning_rate": 0.00017900623815542642,
      "loss": 1.6523,
      "step": 7860
    },
    {
      "epoch": 2.1031533939070015,
      "grad_norm": 1.3147523403167725,
      "learning_rate": 0.0001789547437385088,
      "loss": 1.682,
      "step": 7870
    },
    {
      "epoch": 2.1058257616247995,
      "grad_norm": 1.2582192420959473,
      "learning_rate": 0.0001789031936709231,
      "loss": 1.687,
      "step": 7880
    },
    {
      "epoch": 2.1084981293425975,
      "grad_norm": 1.2919081449508667,
      "learning_rate": 0.00017885158798900414,
      "loss": 1.6614,
      "step": 7890
    },
    {
      "epoch": 2.1111704970603955,
      "grad_norm": 1.2467252016067505,
      "learning_rate": 0.00017879992672912572,
      "loss": 1.749,
      "step": 7900
    },
    {
      "epoch": 2.1138428647781935,
      "grad_norm": 1.2039240598678589,
      "learning_rate": 0.00017874820992770088,
      "loss": 1.6873,
      "step": 7910
    },
    {
      "epoch": 2.1165152324959915,
      "grad_norm": 1.3247950077056885,
      "learning_rate": 0.00017869643762118188,
      "loss": 1.7158,
      "step": 7920
    },
    {
      "epoch": 2.1191876002137895,
      "grad_norm": 1.228939414024353,
      "learning_rate": 0.00017864460984606,
      "loss": 1.6909,
      "step": 7930
    },
    {
      "epoch": 2.1218599679315875,
      "grad_norm": 1.5371471643447876,
      "learning_rate": 0.0001785927266388657,
      "loss": 1.7049,
      "step": 7940
    },
    {
      "epoch": 2.1245323356493855,
      "grad_norm": 1.2502973079681396,
      "learning_rate": 0.00017854078803616853,
      "loss": 1.6828,
      "step": 7950
    },
    {
      "epoch": 2.1272047033671835,
      "grad_norm": 1.2730250358581543,
      "learning_rate": 0.00017848879407457687,
      "loss": 1.6686,
      "step": 7960
    },
    {
      "epoch": 2.1298770710849815,
      "grad_norm": 1.2667866945266724,
      "learning_rate": 0.0001784367447907384,
      "loss": 1.6014,
      "step": 7970
    },
    {
      "epoch": 2.132549438802779,
      "grad_norm": 1.2797930240631104,
      "learning_rate": 0.00017838464022133967,
      "loss": 1.7063,
      "step": 7980
    },
    {
      "epoch": 2.135221806520577,
      "grad_norm": 1.3010532855987549,
      "learning_rate": 0.00017833248040310617,
      "loss": 1.7519,
      "step": 7990
    },
    {
      "epoch": 2.137894174238375,
      "grad_norm": 1.3024216890335083,
      "learning_rate": 0.00017828026537280235,
      "loss": 1.689,
      "step": 8000
    },
    {
      "epoch": 2.140566541956173,
      "grad_norm": 1.3460325002670288,
      "learning_rate": 0.00017822799516723165,
      "loss": 1.6936,
      "step": 8010
    },
    {
      "epoch": 2.143238909673971,
      "grad_norm": 1.1167391538619995,
      "learning_rate": 0.00017817566982323627,
      "loss": 1.6171,
      "step": 8020
    },
    {
      "epoch": 2.145911277391769,
      "grad_norm": 1.421852707862854,
      "learning_rate": 0.0001781232893776974,
      "loss": 1.7152,
      "step": 8030
    },
    {
      "epoch": 2.148583645109567,
      "grad_norm": 1.273163080215454,
      "learning_rate": 0.00017807085386753495,
      "loss": 1.6343,
      "step": 8040
    },
    {
      "epoch": 2.151256012827365,
      "grad_norm": 1.3871347904205322,
      "learning_rate": 0.00017801836332970777,
      "loss": 1.6637,
      "step": 8050
    },
    {
      "epoch": 2.153928380545163,
      "grad_norm": 1.3075594902038574,
      "learning_rate": 0.00017796581780121334,
      "loss": 1.7416,
      "step": 8060
    },
    {
      "epoch": 2.156600748262961,
      "grad_norm": 1.2139554023742676,
      "learning_rate": 0.00017791321731908805,
      "loss": 1.7024,
      "step": 8070
    },
    {
      "epoch": 2.159273115980759,
      "grad_norm": 1.2737025022506714,
      "learning_rate": 0.00017786056192040697,
      "loss": 1.686,
      "step": 8080
    },
    {
      "epoch": 2.161945483698557,
      "grad_norm": 1.3183096647262573,
      "learning_rate": 0.00017780785164228386,
      "loss": 1.7213,
      "step": 8090
    },
    {
      "epoch": 2.164617851416355,
      "grad_norm": 1.2604891061782837,
      "learning_rate": 0.0001777550865218711,
      "loss": 1.6772,
      "step": 8100
    },
    {
      "epoch": 2.167290219134153,
      "grad_norm": 1.2124100923538208,
      "learning_rate": 0.00017770226659635994,
      "loss": 1.7333,
      "step": 8110
    },
    {
      "epoch": 2.169962586851951,
      "grad_norm": 1.2802506685256958,
      "learning_rate": 0.00017764939190297998,
      "loss": 1.7097,
      "step": 8120
    },
    {
      "epoch": 2.172634954569749,
      "grad_norm": 1.3564809560775757,
      "learning_rate": 0.0001775964624789996,
      "loss": 1.6261,
      "step": 8130
    },
    {
      "epoch": 2.175307322287547,
      "grad_norm": 1.2205817699432373,
      "learning_rate": 0.00017754347836172575,
      "loss": 1.6072,
      "step": 8140
    },
    {
      "epoch": 2.177979690005345,
      "grad_norm": 1.2707934379577637,
      "learning_rate": 0.00017749043958850385,
      "loss": 1.6535,
      "step": 8150
    },
    {
      "epoch": 2.1806520577231425,
      "grad_norm": 1.285674810409546,
      "learning_rate": 0.00017743734619671787,
      "loss": 1.6612,
      "step": 8160
    },
    {
      "epoch": 2.1833244254409405,
      "grad_norm": 1.2037670612335205,
      "learning_rate": 0.00017738419822379035,
      "loss": 1.6488,
      "step": 8170
    },
    {
      "epoch": 2.1859967931587385,
      "grad_norm": 1.215549349784851,
      "learning_rate": 0.00017733099570718224,
      "loss": 1.6912,
      "step": 8180
    },
    {
      "epoch": 2.1886691608765365,
      "grad_norm": 1.3087904453277588,
      "learning_rate": 0.0001772777386843929,
      "loss": 1.6609,
      "step": 8190
    },
    {
      "epoch": 2.1913415285943345,
      "grad_norm": 1.2005733251571655,
      "learning_rate": 0.00017722442719296017,
      "loss": 1.6959,
      "step": 8200
    },
    {
      "epoch": 2.1940138963121325,
      "grad_norm": 1.394248366355896,
      "learning_rate": 0.00017717106127046028,
      "loss": 1.6696,
      "step": 8210
    },
    {
      "epoch": 2.1966862640299305,
      "grad_norm": 1.2202259302139282,
      "learning_rate": 0.0001771176409545078,
      "loss": 1.6734,
      "step": 8220
    },
    {
      "epoch": 2.1993586317477285,
      "grad_norm": 1.2602444887161255,
      "learning_rate": 0.00017706416628275563,
      "loss": 1.7377,
      "step": 8230
    },
    {
      "epoch": 2.2020309994655265,
      "grad_norm": 1.4346433877944946,
      "learning_rate": 0.00017701063729289498,
      "loss": 1.6947,
      "step": 8240
    },
    {
      "epoch": 2.2047033671833245,
      "grad_norm": 1.2007207870483398,
      "learning_rate": 0.0001769570540226554,
      "loss": 1.7211,
      "step": 8250
    },
    {
      "epoch": 2.2073757349011225,
      "grad_norm": 1.3080480098724365,
      "learning_rate": 0.00017690341650980463,
      "loss": 1.7328,
      "step": 8260
    },
    {
      "epoch": 2.2100481026189205,
      "grad_norm": 1.2944198846817017,
      "learning_rate": 0.0001768497247921487,
      "loss": 1.6602,
      "step": 8270
    },
    {
      "epoch": 2.2127204703367185,
      "grad_norm": 1.3239096403121948,
      "learning_rate": 0.00017679597890753182,
      "loss": 1.6563,
      "step": 8280
    },
    {
      "epoch": 2.2153928380545165,
      "grad_norm": 1.3229306936264038,
      "learning_rate": 0.00017674217889383637,
      "loss": 1.7382,
      "step": 8290
    },
    {
      "epoch": 2.2180652057723145,
      "grad_norm": 1.3248566389083862,
      "learning_rate": 0.0001766883247889829,
      "loss": 1.7465,
      "step": 8300
    },
    {
      "epoch": 2.220737573490112,
      "grad_norm": 1.284016728401184,
      "learning_rate": 0.0001766344166309301,
      "loss": 1.7106,
      "step": 8310
    },
    {
      "epoch": 2.22340994120791,
      "grad_norm": 1.2725125551223755,
      "learning_rate": 0.00017658045445767466,
      "loss": 1.7608,
      "step": 8320
    },
    {
      "epoch": 2.226082308925708,
      "grad_norm": 1.3566988706588745,
      "learning_rate": 0.0001765264383072515,
      "loss": 1.7268,
      "step": 8330
    },
    {
      "epoch": 2.228754676643506,
      "grad_norm": 1.265575885772705,
      "learning_rate": 0.00017647236821773349,
      "loss": 1.722,
      "step": 8340
    },
    {
      "epoch": 2.231427044361304,
      "grad_norm": 1.2469218969345093,
      "learning_rate": 0.00017641824422723147,
      "loss": 1.6961,
      "step": 8350
    },
    {
      "epoch": 2.234099412079102,
      "grad_norm": 1.4362605810165405,
      "learning_rate": 0.00017636406637389443,
      "loss": 1.7367,
      "step": 8360
    },
    {
      "epoch": 2.2367717797969,
      "grad_norm": 1.2980564832687378,
      "learning_rate": 0.00017630983469590916,
      "loss": 1.6679,
      "step": 8370
    },
    {
      "epoch": 2.239444147514698,
      "grad_norm": 1.2054814100265503,
      "learning_rate": 0.00017625554923150047,
      "loss": 1.7167,
      "step": 8380
    },
    {
      "epoch": 2.242116515232496,
      "grad_norm": 1.2891802787780762,
      "learning_rate": 0.0001762012100189311,
      "loss": 1.6553,
      "step": 8390
    },
    {
      "epoch": 2.244788882950294,
      "grad_norm": 1.2684916257858276,
      "learning_rate": 0.00017614681709650162,
      "loss": 1.7258,
      "step": 8400
    },
    {
      "epoch": 2.247461250668092,
      "grad_norm": 1.2523155212402344,
      "learning_rate": 0.00017609237050255047,
      "loss": 1.8091,
      "step": 8410
    },
    {
      "epoch": 2.25013361838589,
      "grad_norm": 1.2900711297988892,
      "learning_rate": 0.00017603787027545392,
      "loss": 1.6807,
      "step": 8420
    },
    {
      "epoch": 2.252805986103688,
      "grad_norm": 1.3863922357559204,
      "learning_rate": 0.0001759833164536261,
      "loss": 1.7239,
      "step": 8430
    },
    {
      "epoch": 2.255478353821486,
      "grad_norm": 1.3520922660827637,
      "learning_rate": 0.00017592870907551887,
      "loss": 1.7375,
      "step": 8440
    },
    {
      "epoch": 2.258150721539284,
      "grad_norm": 1.2624331712722778,
      "learning_rate": 0.00017587404817962177,
      "loss": 1.706,
      "step": 8450
    },
    {
      "epoch": 2.260823089257082,
      "grad_norm": 1.2753965854644775,
      "learning_rate": 0.00017581933380446217,
      "loss": 1.693,
      "step": 8460
    },
    {
      "epoch": 2.26349545697488,
      "grad_norm": 1.2787902355194092,
      "learning_rate": 0.00017576456598860512,
      "loss": 1.6783,
      "step": 8470
    },
    {
      "epoch": 2.2661678246926775,
      "grad_norm": 1.255979299545288,
      "learning_rate": 0.00017570974477065324,
      "loss": 1.7037,
      "step": 8480
    },
    {
      "epoch": 2.2688401924104755,
      "grad_norm": 1.220231533050537,
      "learning_rate": 0.00017565487018924694,
      "loss": 1.7603,
      "step": 8490
    },
    {
      "epoch": 2.2715125601282735,
      "grad_norm": 1.3431339263916016,
      "learning_rate": 0.0001755999422830641,
      "loss": 1.6591,
      "step": 8500
    },
    {
      "epoch": 2.2741849278460715,
      "grad_norm": 1.2896251678466797,
      "learning_rate": 0.0001755449610908203,
      "loss": 1.6617,
      "step": 8510
    },
    {
      "epoch": 2.2768572955638695,
      "grad_norm": 1.2404873371124268,
      "learning_rate": 0.00017548992665126858,
      "loss": 1.7057,
      "step": 8520
    },
    {
      "epoch": 2.2795296632816675,
      "grad_norm": 1.3518658876419067,
      "learning_rate": 0.0001754348390031996,
      "loss": 1.682,
      "step": 8530
    },
    {
      "epoch": 2.2822020309994655,
      "grad_norm": 1.28294837474823,
      "learning_rate": 0.00017537969818544143,
      "loss": 1.6905,
      "step": 8540
    },
    {
      "epoch": 2.2848743987172635,
      "grad_norm": 1.3236899375915527,
      "learning_rate": 0.00017532450423685974,
      "loss": 1.7282,
      "step": 8550
    },
    {
      "epoch": 2.2875467664350615,
      "grad_norm": 1.3238036632537842,
      "learning_rate": 0.0001752692571963575,
      "loss": 1.6797,
      "step": 8560
    },
    {
      "epoch": 2.2902191341528595,
      "grad_norm": 1.2107789516448975,
      "learning_rate": 0.00017521395710287524,
      "loss": 1.6792,
      "step": 8570
    },
    {
      "epoch": 2.2928915018706575,
      "grad_norm": 1.1869723796844482,
      "learning_rate": 0.0001751586039953908,
      "loss": 1.7645,
      "step": 8580
    },
    {
      "epoch": 2.2955638695884555,
      "grad_norm": 1.2540340423583984,
      "learning_rate": 0.0001751031979129194,
      "loss": 1.7228,
      "step": 8590
    },
    {
      "epoch": 2.2982362373062535,
      "grad_norm": 1.3475019931793213,
      "learning_rate": 0.00017504773889451361,
      "loss": 1.7466,
      "step": 8600
    },
    {
      "epoch": 2.3009086050240515,
      "grad_norm": 1.367012619972229,
      "learning_rate": 0.0001749922269792633,
      "loss": 1.6538,
      "step": 8610
    },
    {
      "epoch": 2.303580972741849,
      "grad_norm": 1.3030816316604614,
      "learning_rate": 0.00017493666220629568,
      "loss": 1.7625,
      "step": 8620
    },
    {
      "epoch": 2.306253340459647,
      "grad_norm": 1.3685482740402222,
      "learning_rate": 0.0001748810446147751,
      "loss": 1.6994,
      "step": 8630
    },
    {
      "epoch": 2.308925708177445,
      "grad_norm": 1.2512644529342651,
      "learning_rate": 0.00017482537424390332,
      "loss": 1.6751,
      "step": 8640
    },
    {
      "epoch": 2.311598075895243,
      "grad_norm": 1.2357844114303589,
      "learning_rate": 0.00017476965113291906,
      "loss": 1.7458,
      "step": 8650
    },
    {
      "epoch": 2.314270443613041,
      "grad_norm": 1.305551290512085,
      "learning_rate": 0.0001747138753210984,
      "loss": 1.6777,
      "step": 8660
    },
    {
      "epoch": 2.316942811330839,
      "grad_norm": 1.8365817070007324,
      "learning_rate": 0.0001746580468477545,
      "loss": 1.6625,
      "step": 8670
    },
    {
      "epoch": 2.319615179048637,
      "grad_norm": 1.3910210132598877,
      "learning_rate": 0.0001746021657522377,
      "loss": 1.687,
      "step": 8680
    },
    {
      "epoch": 2.322287546766435,
      "grad_norm": 1.301171898841858,
      "learning_rate": 0.0001745462320739353,
      "loss": 1.706,
      "step": 8690
    },
    {
      "epoch": 2.324959914484233,
      "grad_norm": 1.1999925374984741,
      "learning_rate": 0.00017449024585227177,
      "loss": 1.7317,
      "step": 8700
    },
    {
      "epoch": 2.327632282202031,
      "grad_norm": 1.381013035774231,
      "learning_rate": 0.0001744342071267086,
      "loss": 1.695,
      "step": 8710
    },
    {
      "epoch": 2.330304649919829,
      "grad_norm": 1.274725317955017,
      "learning_rate": 0.00017437811593674427,
      "loss": 1.7112,
      "step": 8720
    },
    {
      "epoch": 2.332977017637627,
      "grad_norm": 1.2251441478729248,
      "learning_rate": 0.00017432197232191424,
      "loss": 1.7284,
      "step": 8730
    },
    {
      "epoch": 2.335649385355425,
      "grad_norm": 1.3940762281417847,
      "learning_rate": 0.0001742657763217909,
      "loss": 1.6257,
      "step": 8740
    },
    {
      "epoch": 2.338321753073223,
      "grad_norm": 1.2540252208709717,
      "learning_rate": 0.00017420952797598358,
      "loss": 1.7471,
      "step": 8750
    },
    {
      "epoch": 2.340994120791021,
      "grad_norm": 1.3477139472961426,
      "learning_rate": 0.00017415322732413854,
      "loss": 1.7856,
      "step": 8760
    },
    {
      "epoch": 2.343666488508819,
      "grad_norm": 1.329470157623291,
      "learning_rate": 0.0001740968744059389,
      "loss": 1.6957,
      "step": 8770
    },
    {
      "epoch": 2.346338856226617,
      "grad_norm": 1.2522149085998535,
      "learning_rate": 0.00017404046926110454,
      "loss": 1.7184,
      "step": 8780
    },
    {
      "epoch": 2.349011223944415,
      "grad_norm": 1.3245595693588257,
      "learning_rate": 0.00017398401192939225,
      "loss": 1.6724,
      "step": 8790
    },
    {
      "epoch": 2.351683591662213,
      "grad_norm": 1.2924376726150513,
      "learning_rate": 0.00017392750245059552,
      "loss": 1.6874,
      "step": 8800
    },
    {
      "epoch": 2.3543559593800105,
      "grad_norm": 1.359859585762024,
      "learning_rate": 0.0001738709408645447,
      "loss": 1.7426,
      "step": 8810
    },
    {
      "epoch": 2.3570283270978085,
      "grad_norm": 1.3060246706008911,
      "learning_rate": 0.00017381432721110676,
      "loss": 1.6724,
      "step": 8820
    },
    {
      "epoch": 2.3597006948156065,
      "grad_norm": 1.378063440322876,
      "learning_rate": 0.00017375766153018542,
      "loss": 1.7137,
      "step": 8830
    },
    {
      "epoch": 2.3623730625334045,
      "grad_norm": 1.2369308471679688,
      "learning_rate": 0.0001737009438617211,
      "loss": 1.7285,
      "step": 8840
    },
    {
      "epoch": 2.3650454302512025,
      "grad_norm": 1.3088947534561157,
      "learning_rate": 0.0001736441742456908,
      "loss": 1.7183,
      "step": 8850
    },
    {
      "epoch": 2.3677177979690005,
      "grad_norm": 1.389485478401184,
      "learning_rate": 0.00017358735272210817,
      "loss": 1.6696,
      "step": 8860
    },
    {
      "epoch": 2.3703901656867985,
      "grad_norm": 1.2320945262908936,
      "learning_rate": 0.00017353047933102347,
      "loss": 1.6916,
      "step": 8870
    },
    {
      "epoch": 2.3730625334045965,
      "grad_norm": 1.2729374170303345,
      "learning_rate": 0.00017347355411252341,
      "loss": 1.7538,
      "step": 8880
    },
    {
      "epoch": 2.3757349011223945,
      "grad_norm": 1.2985988855361938,
      "learning_rate": 0.0001734165771067314,
      "loss": 1.7286,
      "step": 8890
    },
    {
      "epoch": 2.3784072688401925,
      "grad_norm": 1.290063500404358,
      "learning_rate": 0.0001733595483538072,
      "loss": 1.6714,
      "step": 8900
    },
    {
      "epoch": 2.3810796365579905,
      "grad_norm": 1.323641061782837,
      "learning_rate": 0.00017330246789394715,
      "loss": 1.6183,
      "step": 8910
    },
    {
      "epoch": 2.3837520042757885,
      "grad_norm": 1.1713404655456543,
      "learning_rate": 0.00017324533576738396,
      "loss": 1.6931,
      "step": 8920
    },
    {
      "epoch": 2.3864243719935865,
      "grad_norm": 1.326493263244629,
      "learning_rate": 0.00017318815201438682,
      "loss": 1.7018,
      "step": 8930
    },
    {
      "epoch": 2.3890967397113845,
      "grad_norm": 1.3071362972259521,
      "learning_rate": 0.00017313091667526121,
      "loss": 1.7383,
      "step": 8940
    },
    {
      "epoch": 2.391769107429182,
      "grad_norm": 1.309009075164795,
      "learning_rate": 0.00017307362979034912,
      "loss": 1.7412,
      "step": 8950
    },
    {
      "epoch": 2.39444147514698,
      "grad_norm": 1.3492226600646973,
      "learning_rate": 0.00017301629140002874,
      "loss": 1.7089,
      "step": 8960
    },
    {
      "epoch": 2.397113842864778,
      "grad_norm": 1.1894094944000244,
      "learning_rate": 0.0001729589015447146,
      "loss": 1.6837,
      "step": 8970
    },
    {
      "epoch": 2.399786210582576,
      "grad_norm": 1.2730973958969116,
      "learning_rate": 0.00017290146026485758,
      "loss": 1.6845,
      "step": 8980
    },
    {
      "epoch": 2.402458578300374,
      "grad_norm": 1.4169113636016846,
      "learning_rate": 0.00017284396760094465,
      "loss": 1.7025,
      "step": 8990
    },
    {
      "epoch": 2.405130946018172,
      "grad_norm": 1.3806253671646118,
      "learning_rate": 0.00017278642359349914,
      "loss": 1.6783,
      "step": 9000
    },
    {
      "epoch": 2.40780331373597,
      "grad_norm": 1.2125340700149536,
      "learning_rate": 0.00017272882828308053,
      "loss": 1.7031,
      "step": 9010
    },
    {
      "epoch": 2.410475681453768,
      "grad_norm": 1.2701557874679565,
      "learning_rate": 0.00017267118171028443,
      "loss": 1.7034,
      "step": 9020
    },
    {
      "epoch": 2.413148049171566,
      "grad_norm": 1.3081810474395752,
      "learning_rate": 0.00017261348391574258,
      "loss": 1.7071,
      "step": 9030
    },
    {
      "epoch": 2.415820416889364,
      "grad_norm": 1.255834937095642,
      "learning_rate": 0.0001725557349401229,
      "loss": 1.7474,
      "step": 9040
    },
    {
      "epoch": 2.418492784607162,
      "grad_norm": 1.499633550643921,
      "learning_rate": 0.0001724979348241293,
      "loss": 1.6629,
      "step": 9050
    },
    {
      "epoch": 2.42116515232496,
      "grad_norm": 1.3313318490982056,
      "learning_rate": 0.00017244008360850174,
      "loss": 1.7318,
      "step": 9060
    },
    {
      "epoch": 2.423837520042758,
      "grad_norm": 1.2691699266433716,
      "learning_rate": 0.00017238218133401625,
      "loss": 1.7433,
      "step": 9070
    },
    {
      "epoch": 2.426509887760556,
      "grad_norm": 1.3339135646820068,
      "learning_rate": 0.00017232422804148482,
      "loss": 1.6833,
      "step": 9080
    },
    {
      "epoch": 2.429182255478354,
      "grad_norm": 1.2071961164474487,
      "learning_rate": 0.0001722662237717554,
      "loss": 1.7443,
      "step": 9090
    },
    {
      "epoch": 2.431854623196152,
      "grad_norm": 1.2049094438552856,
      "learning_rate": 0.00017220816856571187,
      "loss": 1.6856,
      "step": 9100
    },
    {
      "epoch": 2.43452699091395,
      "grad_norm": 1.3422529697418213,
      "learning_rate": 0.000172150062464274,
      "loss": 1.7147,
      "step": 9110
    },
    {
      "epoch": 2.437199358631748,
      "grad_norm": 1.2806687355041504,
      "learning_rate": 0.00017209190550839746,
      "loss": 1.8169,
      "step": 9120
    },
    {
      "epoch": 2.4398717263495455,
      "grad_norm": 1.3218368291854858,
      "learning_rate": 0.00017203369773907375,
      "loss": 1.7606,
      "step": 9130
    },
    {
      "epoch": 2.4425440940673435,
      "grad_norm": 1.264967441558838,
      "learning_rate": 0.0001719754391973302,
      "loss": 1.7328,
      "step": 9140
    },
    {
      "epoch": 2.4452164617851415,
      "grad_norm": 1.3537585735321045,
      "learning_rate": 0.00017191712992422985,
      "loss": 1.7355,
      "step": 9150
    },
    {
      "epoch": 2.4478888295029395,
      "grad_norm": 1.3413550853729248,
      "learning_rate": 0.00017185876996087164,
      "loss": 1.6849,
      "step": 9160
    },
    {
      "epoch": 2.4505611972207375,
      "grad_norm": 1.3472052812576294,
      "learning_rate": 0.0001718003593483901,
      "loss": 1.7102,
      "step": 9170
    },
    {
      "epoch": 2.4532335649385355,
      "grad_norm": 1.3312984704971313,
      "learning_rate": 0.00017174189812795557,
      "loss": 1.802,
      "step": 9180
    },
    {
      "epoch": 2.4559059326563335,
      "grad_norm": 1.3904449939727783,
      "learning_rate": 0.000171683386340774,
      "loss": 1.7174,
      "step": 9190
    },
    {
      "epoch": 2.4585783003741315,
      "grad_norm": 1.4721754789352417,
      "learning_rate": 0.00017162482402808694,
      "loss": 1.7527,
      "step": 9200
    },
    {
      "epoch": 2.4612506680919295,
      "grad_norm": 1.2413989305496216,
      "learning_rate": 0.00017156621123117166,
      "loss": 1.7112,
      "step": 9210
    },
    {
      "epoch": 2.4639230358097275,
      "grad_norm": 1.3577944040298462,
      "learning_rate": 0.00017150754799134094,
      "loss": 1.7467,
      "step": 9220
    },
    {
      "epoch": 2.4665954035275255,
      "grad_norm": 1.378377079963684,
      "learning_rate": 0.00017144883434994313,
      "loss": 1.7219,
      "step": 9230
    },
    {
      "epoch": 2.4692677712453235,
      "grad_norm": 1.3738240003585815,
      "learning_rate": 0.0001713900703483621,
      "loss": 1.7736,
      "step": 9240
    },
    {
      "epoch": 2.4719401389631215,
      "grad_norm": 1.3451578617095947,
      "learning_rate": 0.00017133125602801722,
      "loss": 1.7312,
      "step": 9250
    },
    {
      "epoch": 2.4746125066809195,
      "grad_norm": 1.3625116348266602,
      "learning_rate": 0.0001712723914303633,
      "loss": 1.7264,
      "step": 9260
    },
    {
      "epoch": 2.477284874398717,
      "grad_norm": 1.248052716255188,
      "learning_rate": 0.0001712134765968907,
      "loss": 1.7155,
      "step": 9270
    },
    {
      "epoch": 2.479957242116515,
      "grad_norm": 1.4670758247375488,
      "learning_rate": 0.00017115451156912506,
      "loss": 1.6817,
      "step": 9280
    },
    {
      "epoch": 2.482629609834313,
      "grad_norm": 1.3838512897491455,
      "learning_rate": 0.0001710954963886274,
      "loss": 1.7273,
      "step": 9290
    },
    {
      "epoch": 2.485301977552111,
      "grad_norm": 1.2552757263183594,
      "learning_rate": 0.0001710364310969942,
      "loss": 1.7415,
      "step": 9300
    },
    {
      "epoch": 2.487974345269909,
      "grad_norm": 1.2451655864715576,
      "learning_rate": 0.0001709773157358572,
      "loss": 1.7138,
      "step": 9310
    },
    {
      "epoch": 2.490646712987707,
      "grad_norm": 1.3128032684326172,
      "learning_rate": 0.0001709181503468834,
      "loss": 1.7685,
      "step": 9320
    },
    {
      "epoch": 2.493319080705505,
      "grad_norm": 1.2741421461105347,
      "learning_rate": 0.00017085893497177502,
      "loss": 1.7489,
      "step": 9330
    },
    {
      "epoch": 2.495991448423303,
      "grad_norm": 1.3218576908111572,
      "learning_rate": 0.0001707996696522697,
      "loss": 1.7889,
      "step": 9340
    },
    {
      "epoch": 2.498663816141101,
      "grad_norm": 1.350038766860962,
      "learning_rate": 0.00017074035443014006,
      "loss": 1.7364,
      "step": 9350
    },
    {
      "epoch": 2.501336183858899,
      "grad_norm": 1.3044747114181519,
      "learning_rate": 0.0001706809893471941,
      "loss": 1.7715,
      "step": 9360
    },
    {
      "epoch": 2.504008551576697,
      "grad_norm": 1.3677699565887451,
      "learning_rate": 0.00017062157444527477,
      "loss": 1.7649,
      "step": 9370
    },
    {
      "epoch": 2.506680919294495,
      "grad_norm": 1.347490668296814,
      "learning_rate": 0.00017056210976626023,
      "loss": 1.7158,
      "step": 9380
    },
    {
      "epoch": 2.509353287012293,
      "grad_norm": 1.3666515350341797,
      "learning_rate": 0.00017050259535206382,
      "loss": 1.7001,
      "step": 9390
    },
    {
      "epoch": 2.512025654730091,
      "grad_norm": 1.380007266998291,
      "learning_rate": 0.00017044303124463372,
      "loss": 1.7905,
      "step": 9400
    },
    {
      "epoch": 2.514698022447889,
      "grad_norm": 1.2397706508636475,
      "learning_rate": 0.00017038341748595328,
      "loss": 1.7075,
      "step": 9410
    },
    {
      "epoch": 2.517370390165687,
      "grad_norm": 1.390709400177002,
      "learning_rate": 0.00017032375411804086,
      "loss": 1.7399,
      "step": 9420
    },
    {
      "epoch": 2.520042757883485,
      "grad_norm": 1.31925368309021,
      "learning_rate": 0.0001702640411829497,
      "loss": 1.7753,
      "step": 9430
    },
    {
      "epoch": 2.522715125601283,
      "grad_norm": 1.3578137159347534,
      "learning_rate": 0.00017020427872276806,
      "loss": 1.6879,
      "step": 9440
    },
    {
      "epoch": 2.525387493319081,
      "grad_norm": 1.3528056144714355,
      "learning_rate": 0.00017014446677961903,
      "loss": 1.7204,
      "step": 9450
    },
    {
      "epoch": 2.5280598610368785,
      "grad_norm": 1.3264003992080688,
      "learning_rate": 0.0001700846053956606,
      "loss": 1.6537,
      "step": 9460
    },
    {
      "epoch": 2.5307322287546765,
      "grad_norm": 1.285840392112732,
      "learning_rate": 0.00017002469461308573,
      "loss": 1.7186,
      "step": 9470
    },
    {
      "epoch": 2.5334045964724745,
      "grad_norm": 1.2316747903823853,
      "learning_rate": 0.00016996473447412197,
      "loss": 1.7763,
      "step": 9480
    },
    {
      "epoch": 2.5360769641902725,
      "grad_norm": 1.3861089944839478,
      "learning_rate": 0.00016990472502103186,
      "loss": 1.7207,
      "step": 9490
    },
    {
      "epoch": 2.5387493319080705,
      "grad_norm": 1.3715717792510986,
      "learning_rate": 0.00016984466629611254,
      "loss": 1.7232,
      "step": 9500
    },
    {
      "epoch": 2.5414216996258685,
      "grad_norm": 1.5018290281295776,
      "learning_rate": 0.00016978455834169606,
      "loss": 1.7424,
      "step": 9510
    },
    {
      "epoch": 2.5440940673436665,
      "grad_norm": 1.26276695728302,
      "learning_rate": 0.000169724401200149,
      "loss": 1.7195,
      "step": 9520
    },
    {
      "epoch": 2.5467664350614645,
      "grad_norm": 1.4196491241455078,
      "learning_rate": 0.00016966419491387267,
      "loss": 1.788,
      "step": 9530
    },
    {
      "epoch": 2.5494388027792625,
      "grad_norm": 1.3643399477005005,
      "learning_rate": 0.0001696039395253031,
      "loss": 1.7935,
      "step": 9540
    },
    {
      "epoch": 2.5521111704970605,
      "grad_norm": 1.230284333229065,
      "learning_rate": 0.00016954363507691076,
      "loss": 1.6665,
      "step": 9550
    },
    {
      "epoch": 2.5547835382148585,
      "grad_norm": 1.2991338968276978,
      "learning_rate": 0.00016948328161120086,
      "loss": 1.7156,
      "step": 9560
    },
    {
      "epoch": 2.5574559059326565,
      "grad_norm": 1.5775761604309082,
      "learning_rate": 0.00016942287917071308,
      "loss": 1.6749,
      "step": 9570
    },
    {
      "epoch": 2.560128273650454,
      "grad_norm": 1.3274940252304077,
      "learning_rate": 0.00016936242779802164,
      "loss": 1.7407,
      "step": 9580
    },
    {
      "epoch": 2.562800641368252,
      "grad_norm": 1.2171717882156372,
      "learning_rate": 0.0001693019275357352,
      "loss": 1.7929,
      "step": 9590
    },
    {
      "epoch": 2.56547300908605,
      "grad_norm": 1.3335037231445312,
      "learning_rate": 0.00016924137842649702,
      "loss": 1.7101,
      "step": 9600
    },
    {
      "epoch": 2.568145376803848,
      "grad_norm": 1.2525278329849243,
      "learning_rate": 0.00016918078051298457,
      "loss": 1.7031,
      "step": 9610
    },
    {
      "epoch": 2.570817744521646,
      "grad_norm": 1.2409635782241821,
      "learning_rate": 0.0001691201338379099,
      "loss": 1.7306,
      "step": 9620
    },
    {
      "epoch": 2.573490112239444,
      "grad_norm": 1.2902002334594727,
      "learning_rate": 0.0001690594384440194,
      "loss": 1.7211,
      "step": 9630
    },
    {
      "epoch": 2.576162479957242,
      "grad_norm": 1.3910725116729736,
      "learning_rate": 0.00016899869437409372,
      "loss": 1.7647,
      "step": 9640
    },
    {
      "epoch": 2.57883484767504,
      "grad_norm": 1.4463739395141602,
      "learning_rate": 0.0001689379016709479,
      "loss": 1.7101,
      "step": 9650
    },
    {
      "epoch": 2.581507215392838,
      "grad_norm": 1.390336036682129,
      "learning_rate": 0.0001688770603774312,
      "loss": 1.6808,
      "step": 9660
    },
    {
      "epoch": 2.584179583110636,
      "grad_norm": 1.2885065078735352,
      "learning_rate": 0.0001688161705364272,
      "loss": 1.7203,
      "step": 9670
    },
    {
      "epoch": 2.586851950828434,
      "grad_norm": 1.3509775400161743,
      "learning_rate": 0.00016875523219085363,
      "loss": 1.6905,
      "step": 9680
    },
    {
      "epoch": 2.589524318546232,
      "grad_norm": 1.29179048538208,
      "learning_rate": 0.0001686942453836624,
      "loss": 1.7753,
      "step": 9690
    },
    {
      "epoch": 2.59219668626403,
      "grad_norm": 1.3519556522369385,
      "learning_rate": 0.00016863321015783968,
      "loss": 1.743,
      "step": 9700
    },
    {
      "epoch": 2.594869053981828,
      "grad_norm": 1.3056449890136719,
      "learning_rate": 0.00016857212655640565,
      "loss": 1.7583,
      "step": 9710
    },
    {
      "epoch": 2.597541421699626,
      "grad_norm": 1.325150728225708,
      "learning_rate": 0.00016851099462241467,
      "loss": 1.7167,
      "step": 9720
    },
    {
      "epoch": 2.600213789417424,
      "grad_norm": 1.3329095840454102,
      "learning_rate": 0.00016844981439895505,
      "loss": 1.7032,
      "step": 9730
    },
    {
      "epoch": 2.602886157135222,
      "grad_norm": 1.1164215803146362,
      "learning_rate": 0.00016838858592914933,
      "loss": 1.6624,
      "step": 9740
    },
    {
      "epoch": 2.60555852485302,
      "grad_norm": 1.32456636428833,
      "learning_rate": 0.00016832730925615388,
      "loss": 1.7574,
      "step": 9750
    },
    {
      "epoch": 2.608230892570818,
      "grad_norm": 1.4649831056594849,
      "learning_rate": 0.0001682659844231591,
      "loss": 1.7048,
      "step": 9760
    },
    {
      "epoch": 2.610903260288616,
      "grad_norm": 1.3625433444976807,
      "learning_rate": 0.00016820461147338942,
      "loss": 1.7026,
      "step": 9770
    },
    {
      "epoch": 2.6135756280064135,
      "grad_norm": 1.3138645887374878,
      "learning_rate": 0.000168143190450103,
      "loss": 1.7745,
      "step": 9780
    },
    {
      "epoch": 2.6162479957242115,
      "grad_norm": 1.2378613948822021,
      "learning_rate": 0.00016808172139659209,
      "loss": 1.6838,
      "step": 9790
    },
    {
      "epoch": 2.6189203634420095,
      "grad_norm": 1.6066712141036987,
      "learning_rate": 0.00016802020435618267,
      "loss": 1.7235,
      "step": 9800
    },
    {
      "epoch": 2.6215927311598075,
      "grad_norm": 1.3208084106445312,
      "learning_rate": 0.00016795863937223453,
      "loss": 1.6804,
      "step": 9810
    },
    {
      "epoch": 2.6242650988776055,
      "grad_norm": 1.2170714139938354,
      "learning_rate": 0.00016789702648814135,
      "loss": 1.7057,
      "step": 9820
    },
    {
      "epoch": 2.6269374665954035,
      "grad_norm": 1.3493729829788208,
      "learning_rate": 0.00016783536574733047,
      "loss": 1.6769,
      "step": 9830
    },
    {
      "epoch": 2.6296098343132015,
      "grad_norm": 1.3735666275024414,
      "learning_rate": 0.00016777365719326305,
      "loss": 1.7213,
      "step": 9840
    },
    {
      "epoch": 2.6322822020309995,
      "grad_norm": 1.315576434135437,
      "learning_rate": 0.0001677119008694339,
      "loss": 1.7572,
      "step": 9850
    },
    {
      "epoch": 2.6349545697487975,
      "grad_norm": 1.2990847826004028,
      "learning_rate": 0.00016765009681937149,
      "loss": 1.741,
      "step": 9860
    },
    {
      "epoch": 2.6376269374665955,
      "grad_norm": 1.2594319581985474,
      "learning_rate": 0.00016758824508663795,
      "loss": 1.718,
      "step": 9870
    },
    {
      "epoch": 2.6402993051843935,
      "grad_norm": 1.3621447086334229,
      "learning_rate": 0.00016752634571482903,
      "loss": 1.6852,
      "step": 9880
    },
    {
      "epoch": 2.6429716729021915,
      "grad_norm": 1.4120498895645142,
      "learning_rate": 0.00016746439874757406,
      "loss": 1.7506,
      "step": 9890
    },
    {
      "epoch": 2.645644040619989,
      "grad_norm": 1.1625776290893555,
      "learning_rate": 0.00016740240422853582,
      "loss": 1.7878,
      "step": 9900
    },
    {
      "epoch": 2.648316408337787,
      "grad_norm": 1.3444859981536865,
      "learning_rate": 0.00016734036220141077,
      "loss": 1.711,
      "step": 9910
    },
    {
      "epoch": 2.650988776055585,
      "grad_norm": 1.290184736251831,
      "learning_rate": 0.0001672782727099288,
      "loss": 1.6725,
      "step": 9920
    },
    {
      "epoch": 2.653661143773383,
      "grad_norm": 1.3869785070419312,
      "learning_rate": 0.00016721613579785317,
      "loss": 1.7347,
      "step": 9930
    },
    {
      "epoch": 2.656333511491181,
      "grad_norm": 1.3091959953308105,
      "learning_rate": 0.00016715395150898065,
      "loss": 1.7358,
      "step": 9940
    },
    {
      "epoch": 2.659005879208979,
      "grad_norm": 1.4104623794555664,
      "learning_rate": 0.00016709171988714134,
      "loss": 1.7717,
      "step": 9950
    },
    {
      "epoch": 2.661678246926777,
      "grad_norm": 1.270210862159729,
      "learning_rate": 0.00016702944097619877,
      "loss": 1.7246,
      "step": 9960
    },
    {
      "epoch": 2.664350614644575,
      "grad_norm": 1.3403785228729248,
      "learning_rate": 0.00016696711482004975,
      "loss": 1.7422,
      "step": 9970
    },
    {
      "epoch": 2.667022982362373,
      "grad_norm": 1.3155403137207031,
      "learning_rate": 0.00016690474146262445,
      "loss": 1.7393,
      "step": 9980
    },
    {
      "epoch": 2.669695350080171,
      "grad_norm": 1.2800922393798828,
      "learning_rate": 0.00016684232094788621,
      "loss": 1.7407,
      "step": 9990
    },
    {
      "epoch": 2.672367717797969,
      "grad_norm": 1.3908976316452026,
      "learning_rate": 0.00016677985331983174,
      "loss": 1.7189,
      "step": 10000
    },
    {
      "epoch": 2.675040085515767,
      "grad_norm": 1.28983736038208,
      "learning_rate": 0.00016671733862249087,
      "loss": 1.6803,
      "step": 10010
    },
    {
      "epoch": 2.677712453233565,
      "grad_norm": 1.2946239709854126,
      "learning_rate": 0.00016665477689992657,
      "loss": 1.6751,
      "step": 10020
    },
    {
      "epoch": 2.680384820951363,
      "grad_norm": 1.310894250869751,
      "learning_rate": 0.000166592168196235,
      "loss": 1.7841,
      "step": 10030
    },
    {
      "epoch": 2.683057188669161,
      "grad_norm": 1.3421567678451538,
      "learning_rate": 0.00016652951255554556,
      "loss": 1.7249,
      "step": 10040
    },
    {
      "epoch": 2.685729556386959,
      "grad_norm": 1.2867803573608398,
      "learning_rate": 0.00016646681002202053,
      "loss": 1.7287,
      "step": 10050
    },
    {
      "epoch": 2.688401924104757,
      "grad_norm": 1.28252112865448,
      "learning_rate": 0.00016640406063985532,
      "loss": 1.7645,
      "step": 10060
    },
    {
      "epoch": 2.691074291822555,
      "grad_norm": 1.3790788650512695,
      "learning_rate": 0.00016634126445327837,
      "loss": 1.6884,
      "step": 10070
    },
    {
      "epoch": 2.693746659540353,
      "grad_norm": 1.3141494989395142,
      "learning_rate": 0.00016627842150655116,
      "loss": 1.7077,
      "step": 10080
    },
    {
      "epoch": 2.696419027258151,
      "grad_norm": 1.4122681617736816,
      "learning_rate": 0.00016621553184396797,
      "loss": 1.7857,
      "step": 10090
    },
    {
      "epoch": 2.699091394975949,
      "grad_norm": 1.212904930114746,
      "learning_rate": 0.00016615259550985628,
      "loss": 1.6774,
      "step": 10100
    },
    {
      "epoch": 2.7017637626937465,
      "grad_norm": 1.3065979480743408,
      "learning_rate": 0.00016608961254857614,
      "loss": 1.7002,
      "step": 10110
    },
    {
      "epoch": 2.7044361304115445,
      "grad_norm": 1.3611199855804443,
      "learning_rate": 0.00016602658300452072,
      "loss": 1.7818,
      "step": 10120
    },
    {
      "epoch": 2.7071084981293425,
      "grad_norm": 1.3592082262039185,
      "learning_rate": 0.0001659635069221159,
      "loss": 1.6907,
      "step": 10130
    },
    {
      "epoch": 2.7097808658471405,
      "grad_norm": 1.3596382141113281,
      "learning_rate": 0.00016590038434582038,
      "loss": 1.7305,
      "step": 10140
    },
    {
      "epoch": 2.7124532335649385,
      "grad_norm": 1.226095199584961,
      "learning_rate": 0.00016583721532012567,
      "loss": 1.7599,
      "step": 10150
    },
    {
      "epoch": 2.7151256012827365,
      "grad_norm": 1.2843650579452515,
      "learning_rate": 0.00016577399988955595,
      "loss": 1.6564,
      "step": 10160
    },
    {
      "epoch": 2.7177979690005345,
      "grad_norm": 1.2752437591552734,
      "learning_rate": 0.00016571073809866816,
      "loss": 1.7466,
      "step": 10170
    },
    {
      "epoch": 2.7204703367183325,
      "grad_norm": 1.2111462354660034,
      "learning_rate": 0.00016564742999205194,
      "loss": 1.7266,
      "step": 10180
    },
    {
      "epoch": 2.7231427044361305,
      "grad_norm": 1.2403128147125244,
      "learning_rate": 0.00016558407561432953,
      "loss": 1.741,
      "step": 10190
    },
    {
      "epoch": 2.7258150721539285,
      "grad_norm": 1.2999945878982544,
      "learning_rate": 0.0001655206750101558,
      "loss": 1.7946,
      "step": 10200
    },
    {
      "epoch": 2.7284874398717265,
      "grad_norm": 1.2286380529403687,
      "learning_rate": 0.00016545722822421813,
      "loss": 1.7352,
      "step": 10210
    },
    {
      "epoch": 2.7311598075895245,
      "grad_norm": 1.4907112121582031,
      "learning_rate": 0.0001653937353012366,
      "loss": 1.6907,
      "step": 10220
    },
    {
      "epoch": 2.733832175307322,
      "grad_norm": 1.3053269386291504,
      "learning_rate": 0.00016533019628596368,
      "loss": 1.7078,
      "step": 10230
    },
    {
      "epoch": 2.73650454302512,
      "grad_norm": 1.367099642753601,
      "learning_rate": 0.00016526661122318444,
      "loss": 1.704,
      "step": 10240
    },
    {
      "epoch": 2.739176910742918,
      "grad_norm": 1.4227197170257568,
      "learning_rate": 0.00016520298015771626,
      "loss": 1.7224,
      "step": 10250
    },
    {
      "epoch": 2.741849278460716,
      "grad_norm": 1.3679789304733276,
      "learning_rate": 0.0001651393031344091,
      "loss": 1.7932,
      "step": 10260
    },
    {
      "epoch": 2.744521646178514,
      "grad_norm": 1.3766450881958008,
      "learning_rate": 0.00016507558019814516,
      "loss": 1.6428,
      "step": 10270
    },
    {
      "epoch": 2.747194013896312,
      "grad_norm": 1.4413139820098877,
      "learning_rate": 0.00016501181139383917,
      "loss": 1.6424,
      "step": 10280
    },
    {
      "epoch": 2.74986638161411,
      "grad_norm": 1.2802283763885498,
      "learning_rate": 0.00016494799676643803,
      "loss": 1.7878,
      "step": 10290
    },
    {
      "epoch": 2.752538749331908,
      "grad_norm": 1.3058050870895386,
      "learning_rate": 0.000164884136360921,
      "loss": 1.7315,
      "step": 10300
    },
    {
      "epoch": 2.755211117049706,
      "grad_norm": 1.2209022045135498,
      "learning_rate": 0.00016482023022229968,
      "loss": 1.7429,
      "step": 10310
    },
    {
      "epoch": 2.757883484767504,
      "grad_norm": 1.3601993322372437,
      "learning_rate": 0.0001647562783956178,
      "loss": 1.8099,
      "step": 10320
    },
    {
      "epoch": 2.760555852485302,
      "grad_norm": 1.326979398727417,
      "learning_rate": 0.00016469228092595137,
      "loss": 1.7229,
      "step": 10330
    },
    {
      "epoch": 2.7632282202031,
      "grad_norm": 1.2629777193069458,
      "learning_rate": 0.00016462823785840848,
      "loss": 1.8033,
      "step": 10340
    },
    {
      "epoch": 2.765900587920898,
      "grad_norm": 1.466591477394104,
      "learning_rate": 0.00016456414923812944,
      "loss": 1.7294,
      "step": 10350
    },
    {
      "epoch": 2.768572955638696,
      "grad_norm": 1.3726177215576172,
      "learning_rate": 0.00016450001511028665,
      "loss": 1.7337,
      "step": 10360
    },
    {
      "epoch": 2.771245323356494,
      "grad_norm": 1.3623446226119995,
      "learning_rate": 0.00016443583552008456,
      "loss": 1.6891,
      "step": 10370
    },
    {
      "epoch": 2.773917691074292,
      "grad_norm": 1.264294981956482,
      "learning_rate": 0.0001643716105127597,
      "loss": 1.6886,
      "step": 10380
    },
    {
      "epoch": 2.77659005879209,
      "grad_norm": 1.3817180395126343,
      "learning_rate": 0.00016430734013358055,
      "loss": 1.7559,
      "step": 10390
    },
    {
      "epoch": 2.779262426509888,
      "grad_norm": 1.3496696949005127,
      "learning_rate": 0.00016424302442784764,
      "loss": 1.6906,
      "step": 10400
    },
    {
      "epoch": 2.781934794227686,
      "grad_norm": 1.4543319940567017,
      "learning_rate": 0.00016417866344089346,
      "loss": 1.7465,
      "step": 10410
    },
    {
      "epoch": 2.784607161945484,
      "grad_norm": 1.3868132829666138,
      "learning_rate": 0.00016411425721808226,
      "loss": 1.7322,
      "step": 10420
    },
    {
      "epoch": 2.7872795296632815,
      "grad_norm": 1.2394338846206665,
      "learning_rate": 0.00016404980580481042,
      "loss": 1.7826,
      "step": 10430
    },
    {
      "epoch": 2.7899518973810795,
      "grad_norm": 1.2655469179153442,
      "learning_rate": 0.00016398530924650602,
      "loss": 1.6935,
      "step": 10440
    },
    {
      "epoch": 2.7926242650988775,
      "grad_norm": 1.3503713607788086,
      "learning_rate": 0.0001639207675886289,
      "loss": 1.7606,
      "step": 10450
    },
    {
      "epoch": 2.7952966328166755,
      "grad_norm": 1.290440559387207,
      "learning_rate": 0.00016385618087667086,
      "loss": 1.7148,
      "step": 10460
    },
    {
      "epoch": 2.7979690005344735,
      "grad_norm": 1.3137966394424438,
      "learning_rate": 0.0001637915491561553,
      "loss": 1.7471,
      "step": 10470
    },
    {
      "epoch": 2.8006413682522715,
      "grad_norm": 1.3273319005966187,
      "learning_rate": 0.00016372687247263746,
      "loss": 1.739,
      "step": 10480
    },
    {
      "epoch": 2.8033137359700695,
      "grad_norm": 1.428066372871399,
      "learning_rate": 0.00016366215087170417,
      "loss": 1.7515,
      "step": 10490
    },
    {
      "epoch": 2.8059861036878675,
      "grad_norm": 1.3554637432098389,
      "learning_rate": 0.00016359738439897402,
      "loss": 1.7182,
      "step": 10500
    },
    {
      "epoch": 2.8086584714056655,
      "grad_norm": 1.374074101448059,
      "learning_rate": 0.0001635325731000972,
      "loss": 1.7339,
      "step": 10510
    },
    {
      "epoch": 2.8113308391234635,
      "grad_norm": 1.3375787734985352,
      "learning_rate": 0.00016346771702075542,
      "loss": 1.6936,
      "step": 10520
    },
    {
      "epoch": 2.8140032068412615,
      "grad_norm": 1.3112101554870605,
      "learning_rate": 0.00016340281620666204,
      "loss": 1.6997,
      "step": 10530
    },
    {
      "epoch": 2.8166755745590595,
      "grad_norm": 1.3300225734710693,
      "learning_rate": 0.0001633378707035619,
      "loss": 1.677,
      "step": 10540
    },
    {
      "epoch": 2.819347942276857,
      "grad_norm": 1.3386642932891846,
      "learning_rate": 0.00016327288055723136,
      "loss": 1.7043,
      "step": 10550
    },
    {
      "epoch": 2.822020309994655,
      "grad_norm": 1.2374159097671509,
      "learning_rate": 0.00016320784581347827,
      "loss": 1.7164,
      "step": 10560
    },
    {
      "epoch": 2.824692677712453,
      "grad_norm": 1.4122684001922607,
      "learning_rate": 0.00016314276651814184,
      "loss": 1.7111,
      "step": 10570
    },
    {
      "epoch": 2.827365045430251,
      "grad_norm": 1.2185890674591064,
      "learning_rate": 0.00016307764271709281,
      "loss": 1.7596,
      "step": 10580
    },
    {
      "epoch": 2.830037413148049,
      "grad_norm": 1.3645682334899902,
      "learning_rate": 0.00016301247445623312,
      "loss": 1.6581,
      "step": 10590
    },
    {
      "epoch": 2.832709780865847,
      "grad_norm": 1.2973945140838623,
      "learning_rate": 0.0001629472617814962,
      "loss": 1.7805,
      "step": 10600
    },
    {
      "epoch": 2.835382148583645,
      "grad_norm": 1.334075689315796,
      "learning_rate": 0.00016288200473884667,
      "loss": 1.806,
      "step": 10610
    },
    {
      "epoch": 2.838054516301443,
      "grad_norm": 1.2988075017929077,
      "learning_rate": 0.00016281670337428054,
      "loss": 1.6806,
      "step": 10620
    },
    {
      "epoch": 2.840726884019241,
      "grad_norm": 1.379639744758606,
      "learning_rate": 0.00016275135773382494,
      "loss": 1.6805,
      "step": 10630
    },
    {
      "epoch": 2.843399251737039,
      "grad_norm": 1.4113703966140747,
      "learning_rate": 0.00016268596786353832,
      "loss": 1.7433,
      "step": 10640
    },
    {
      "epoch": 2.846071619454837,
      "grad_norm": 1.301796555519104,
      "learning_rate": 0.0001626205338095102,
      "loss": 1.7348,
      "step": 10650
    },
    {
      "epoch": 2.848743987172635,
      "grad_norm": 1.2673159837722778,
      "learning_rate": 0.00016255505561786127,
      "loss": 1.7428,
      "step": 10660
    },
    {
      "epoch": 2.851416354890433,
      "grad_norm": 1.3145058155059814,
      "learning_rate": 0.0001624895333347434,
      "loss": 1.7835,
      "step": 10670
    },
    {
      "epoch": 2.854088722608231,
      "grad_norm": 1.3564109802246094,
      "learning_rate": 0.00016242396700633945,
      "loss": 1.7186,
      "step": 10680
    },
    {
      "epoch": 2.856761090326029,
      "grad_norm": 1.294067144393921,
      "learning_rate": 0.0001623583566788634,
      "loss": 1.7799,
      "step": 10690
    },
    {
      "epoch": 2.859433458043827,
      "grad_norm": 1.3368945121765137,
      "learning_rate": 0.00016229270239856018,
      "loss": 1.8102,
      "step": 10700
    },
    {
      "epoch": 2.862105825761625,
      "grad_norm": 1.353661060333252,
      "learning_rate": 0.0001622270042117057,
      "loss": 1.7566,
      "step": 10710
    },
    {
      "epoch": 2.864778193479423,
      "grad_norm": 1.3188835382461548,
      "learning_rate": 0.0001621612621646069,
      "loss": 1.7552,
      "step": 10720
    },
    {
      "epoch": 2.867450561197221,
      "grad_norm": 1.3148976564407349,
      "learning_rate": 0.00016209547630360155,
      "loss": 1.7441,
      "step": 10730
    },
    {
      "epoch": 2.870122928915019,
      "grad_norm": 1.2544513940811157,
      "learning_rate": 0.0001620296466750583,
      "loss": 1.7836,
      "step": 10740
    },
    {
      "epoch": 2.8727952966328165,
      "grad_norm": 1.249030590057373,
      "learning_rate": 0.0001619637733253767,
      "loss": 1.7693,
      "step": 10750
    },
    {
      "epoch": 2.8754676643506145,
      "grad_norm": 1.238337516784668,
      "learning_rate": 0.00016189785630098716,
      "loss": 1.7058,
      "step": 10760
    },
    {
      "epoch": 2.8781400320684125,
      "grad_norm": 1.3537184000015259,
      "learning_rate": 0.0001618318956483507,
      "loss": 1.7217,
      "step": 10770
    },
    {
      "epoch": 2.8808123997862105,
      "grad_norm": 1.2351958751678467,
      "learning_rate": 0.00016176589141395924,
      "loss": 1.7181,
      "step": 10780
    },
    {
      "epoch": 2.8834847675040085,
      "grad_norm": 1.3087337017059326,
      "learning_rate": 0.00016169984364433537,
      "loss": 1.7465,
      "step": 10790
    },
    {
      "epoch": 2.8861571352218065,
      "grad_norm": 1.2616442441940308,
      "learning_rate": 0.0001616337523860324,
      "loss": 1.7807,
      "step": 10800
    },
    {
      "epoch": 2.8888295029396045,
      "grad_norm": 1.2221183776855469,
      "learning_rate": 0.00016156761768563427,
      "loss": 1.6754,
      "step": 10810
    },
    {
      "epoch": 2.8915018706574025,
      "grad_norm": 1.399147391319275,
      "learning_rate": 0.00016150143958975548,
      "loss": 1.7301,
      "step": 10820
    },
    {
      "epoch": 2.8941742383752005,
      "grad_norm": 1.2853881120681763,
      "learning_rate": 0.00016143521814504118,
      "loss": 1.7429,
      "step": 10830
    },
    {
      "epoch": 2.8968466060929985,
      "grad_norm": 1.3160619735717773,
      "learning_rate": 0.00016136895339816714,
      "loss": 1.7433,
      "step": 10840
    },
    {
      "epoch": 2.8995189738107965,
      "grad_norm": 1.3412164449691772,
      "learning_rate": 0.0001613026453958395,
      "loss": 1.6782,
      "step": 10850
    },
    {
      "epoch": 2.9021913415285945,
      "grad_norm": 1.538735032081604,
      "learning_rate": 0.00016123629418479505,
      "loss": 1.7102,
      "step": 10860
    },
    {
      "epoch": 2.904863709246392,
      "grad_norm": 1.3501207828521729,
      "learning_rate": 0.00016116989981180085,
      "loss": 1.7055,
      "step": 10870
    },
    {
      "epoch": 2.90753607696419,
      "grad_norm": 1.1933869123458862,
      "learning_rate": 0.00016110346232365452,
      "loss": 1.745,
      "step": 10880
    },
    {
      "epoch": 2.910208444681988,
      "grad_norm": 1.3364607095718384,
      "learning_rate": 0.00016103698176718405,
      "loss": 1.7083,
      "step": 10890
    },
    {
      "epoch": 2.912880812399786,
      "grad_norm": 1.2646677494049072,
      "learning_rate": 0.00016097045818924777,
      "loss": 1.7698,
      "step": 10900
    },
    {
      "epoch": 2.915553180117584,
      "grad_norm": 1.337547779083252,
      "learning_rate": 0.00016090389163673434,
      "loss": 1.7008,
      "step": 10910
    },
    {
      "epoch": 2.918225547835382,
      "grad_norm": 1.3661710023880005,
      "learning_rate": 0.00016083728215656267,
      "loss": 1.7276,
      "step": 10920
    },
    {
      "epoch": 2.92089791555318,
      "grad_norm": 1.3688632249832153,
      "learning_rate": 0.00016077062979568196,
      "loss": 1.6726,
      "step": 10930
    },
    {
      "epoch": 2.923570283270978,
      "grad_norm": 1.3069576025009155,
      "learning_rate": 0.0001607039346010717,
      "loss": 1.7281,
      "step": 10940
    },
    {
      "epoch": 2.926242650988776,
      "grad_norm": 1.3173613548278809,
      "learning_rate": 0.00016063719661974142,
      "loss": 1.7207,
      "step": 10950
    },
    {
      "epoch": 2.928915018706574,
      "grad_norm": 1.3447896242141724,
      "learning_rate": 0.00016057041589873094,
      "loss": 1.7599,
      "step": 10960
    },
    {
      "epoch": 2.931587386424372,
      "grad_norm": 1.4540988206863403,
      "learning_rate": 0.00016050359248511015,
      "loss": 1.661,
      "step": 10970
    },
    {
      "epoch": 2.93425975414217,
      "grad_norm": 1.2639456987380981,
      "learning_rate": 0.00016043672642597901,
      "loss": 1.64,
      "step": 10980
    },
    {
      "epoch": 2.936932121859968,
      "grad_norm": 1.3014018535614014,
      "learning_rate": 0.00016036981776846763,
      "loss": 1.6938,
      "step": 10990
    },
    {
      "epoch": 2.939604489577766,
      "grad_norm": 1.3340500593185425,
      "learning_rate": 0.00016030286655973605,
      "loss": 1.7315,
      "step": 11000
    },
    {
      "epoch": 2.942276857295564,
      "grad_norm": 1.3241417407989502,
      "learning_rate": 0.0001602358728469743,
      "loss": 1.7756,
      "step": 11010
    },
    {
      "epoch": 2.944949225013362,
      "grad_norm": 1.4531340599060059,
      "learning_rate": 0.00016016883667740247,
      "loss": 1.764,
      "step": 11020
    },
    {
      "epoch": 2.94762159273116,
      "grad_norm": 1.360221266746521,
      "learning_rate": 0.00016010175809827047,
      "loss": 1.7168,
      "step": 11030
    },
    {
      "epoch": 2.950293960448958,
      "grad_norm": 1.288360595703125,
      "learning_rate": 0.00016003463715685812,
      "loss": 1.7297,
      "step": 11040
    },
    {
      "epoch": 2.952966328166756,
      "grad_norm": 1.3397499322891235,
      "learning_rate": 0.00015996747390047519,
      "loss": 1.7434,
      "step": 11050
    },
    {
      "epoch": 2.955638695884554,
      "grad_norm": 1.356138825416565,
      "learning_rate": 0.00015990026837646115,
      "loss": 1.7398,
      "step": 11060
    },
    {
      "epoch": 2.9583110636023515,
      "grad_norm": 1.2560276985168457,
      "learning_rate": 0.00015983302063218532,
      "loss": 1.8056,
      "step": 11070
    },
    {
      "epoch": 2.9609834313201495,
      "grad_norm": 1.293848991394043,
      "learning_rate": 0.00015976573071504682,
      "loss": 1.7335,
      "step": 11080
    },
    {
      "epoch": 2.9636557990379475,
      "grad_norm": 1.3045257329940796,
      "learning_rate": 0.00015969839867247435,
      "loss": 1.7713,
      "step": 11090
    },
    {
      "epoch": 2.9663281667557455,
      "grad_norm": 1.3143424987792969,
      "learning_rate": 0.00015963102455192652,
      "loss": 1.7589,
      "step": 11100
    },
    {
      "epoch": 2.9690005344735435,
      "grad_norm": 1.3248462677001953,
      "learning_rate": 0.00015956360840089142,
      "loss": 1.7045,
      "step": 11110
    },
    {
      "epoch": 2.9716729021913415,
      "grad_norm": 1.3211400508880615,
      "learning_rate": 0.00015949615026688684,
      "loss": 1.7464,
      "step": 11120
    },
    {
      "epoch": 2.9743452699091395,
      "grad_norm": 1.4016636610031128,
      "learning_rate": 0.00015942865019746014,
      "loss": 1.701,
      "step": 11130
    },
    {
      "epoch": 2.9770176376269375,
      "grad_norm": 1.300790786743164,
      "learning_rate": 0.00015936110824018825,
      "loss": 1.7487,
      "step": 11140
    },
    {
      "epoch": 2.9796900053447355,
      "grad_norm": 1.4114218950271606,
      "learning_rate": 0.00015929352444267762,
      "loss": 1.6506,
      "step": 11150
    },
    {
      "epoch": 2.9823623730625335,
      "grad_norm": 1.2759777307510376,
      "learning_rate": 0.0001592258988525642,
      "loss": 1.7607,
      "step": 11160
    },
    {
      "epoch": 2.9850347407803315,
      "grad_norm": 1.2985563278198242,
      "learning_rate": 0.0001591582315175134,
      "loss": 1.7276,
      "step": 11170
    },
    {
      "epoch": 2.9877071084981295,
      "grad_norm": 1.3185759782791138,
      "learning_rate": 0.00015909052248521996,
      "loss": 1.7113,
      "step": 11180
    },
    {
      "epoch": 2.990379476215927,
      "grad_norm": 1.4013677835464478,
      "learning_rate": 0.0001590227718034082,
      "loss": 1.7043,
      "step": 11190
    },
    {
      "epoch": 2.993051843933725,
      "grad_norm": 1.3771142959594727,
      "learning_rate": 0.00015895497951983162,
      "loss": 1.7113,
      "step": 11200
    },
    {
      "epoch": 2.995724211651523,
      "grad_norm": 1.4461156129837036,
      "learning_rate": 0.00015888714568227313,
      "loss": 1.689,
      "step": 11210
    },
    {
      "epoch": 2.998396579369321,
      "grad_norm": 1.3167692422866821,
      "learning_rate": 0.0001588192703385449,
      "loss": 1.7975,
      "step": 11220
    },
    {
      "epoch": 3.001068947087119,
      "grad_norm": 1.246795415878296,
      "learning_rate": 0.0001587513535364884,
      "loss": 1.6658,
      "step": 11230
    },
    {
      "epoch": 3.003741314804917,
      "grad_norm": 1.34390389919281,
      "learning_rate": 0.00015868339532397425,
      "loss": 1.5372,
      "step": 11240
    },
    {
      "epoch": 3.006413682522715,
      "grad_norm": 1.3438529968261719,
      "learning_rate": 0.00015861539574890226,
      "loss": 1.6494,
      "step": 11250
    },
    {
      "epoch": 3.009086050240513,
      "grad_norm": 1.4086190462112427,
      "learning_rate": 0.00015854735485920152,
      "loss": 1.5712,
      "step": 11260
    },
    {
      "epoch": 3.011758417958311,
      "grad_norm": 1.4233211278915405,
      "learning_rate": 0.0001584792727028301,
      "loss": 1.5603,
      "step": 11270
    },
    {
      "epoch": 3.014430785676109,
      "grad_norm": 1.5091338157653809,
      "learning_rate": 0.0001584111493277752,
      "loss": 1.5916,
      "step": 11280
    },
    {
      "epoch": 3.017103153393907,
      "grad_norm": 1.4274498224258423,
      "learning_rate": 0.0001583429847820531,
      "loss": 1.5357,
      "step": 11290
    },
    {
      "epoch": 3.019775521111705,
      "grad_norm": 1.9058419466018677,
      "learning_rate": 0.000158274779113709,
      "loss": 1.5132,
      "step": 11300
    },
    {
      "epoch": 3.022447888829503,
      "grad_norm": 1.4838693141937256,
      "learning_rate": 0.00015820653237081728,
      "loss": 1.5447,
      "step": 11310
    },
    {
      "epoch": 3.025120256547301,
      "grad_norm": 1.347912311553955,
      "learning_rate": 0.0001581382446014811,
      "loss": 1.5856,
      "step": 11320
    },
    {
      "epoch": 3.027792624265099,
      "grad_norm": 1.5340098142623901,
      "learning_rate": 0.00015806991585383257,
      "loss": 1.5716,
      "step": 11330
    },
    {
      "epoch": 3.030464991982897,
      "grad_norm": 1.4221084117889404,
      "learning_rate": 0.00015800154617603269,
      "loss": 1.4937,
      "step": 11340
    },
    {
      "epoch": 3.033137359700695,
      "grad_norm": 1.35970938205719,
      "learning_rate": 0.00015793313561627135,
      "loss": 1.4901,
      "step": 11350
    },
    {
      "epoch": 3.035809727418493,
      "grad_norm": 1.438905954360962,
      "learning_rate": 0.00015786468422276728,
      "loss": 1.5713,
      "step": 11360
    },
    {
      "epoch": 3.038482095136291,
      "grad_norm": 1.491072177886963,
      "learning_rate": 0.00015779619204376786,
      "loss": 1.6207,
      "step": 11370
    },
    {
      "epoch": 3.0411544628540885,
      "grad_norm": 1.4626355171203613,
      "learning_rate": 0.0001577276591275493,
      "loss": 1.4127,
      "step": 11380
    },
    {
      "epoch": 3.0438268305718865,
      "grad_norm": 1.398729681968689,
      "learning_rate": 0.00015765908552241658,
      "loss": 1.5265,
      "step": 11390
    },
    {
      "epoch": 3.0464991982896845,
      "grad_norm": 1.3984798192977905,
      "learning_rate": 0.00015759047127670325,
      "loss": 1.656,
      "step": 11400
    },
    {
      "epoch": 3.0491715660074825,
      "grad_norm": 1.559378981590271,
      "learning_rate": 0.00015752181643877155,
      "loss": 1.5467,
      "step": 11410
    },
    {
      "epoch": 3.0518439337252805,
      "grad_norm": 1.4860576391220093,
      "learning_rate": 0.00015745312105701237,
      "loss": 1.5323,
      "step": 11420
    },
    {
      "epoch": 3.0545163014430785,
      "grad_norm": 1.4080946445465088,
      "learning_rate": 0.00015738438517984513,
      "loss": 1.5656,
      "step": 11430
    },
    {
      "epoch": 3.0571886691608765,
      "grad_norm": 1.4330800771713257,
      "learning_rate": 0.00015731560885571782,
      "loss": 1.6246,
      "step": 11440
    },
    {
      "epoch": 3.0598610368786745,
      "grad_norm": 1.4812606573104858,
      "learning_rate": 0.00015724679213310685,
      "loss": 1.6035,
      "step": 11450
    },
    {
      "epoch": 3.0625334045964725,
      "grad_norm": 1.4903162717819214,
      "learning_rate": 0.00015717793506051726,
      "loss": 1.6082,
      "step": 11460
    },
    {
      "epoch": 3.0652057723142705,
      "grad_norm": 1.451586365699768,
      "learning_rate": 0.0001571090376864824,
      "loss": 1.5459,
      "step": 11470
    },
    {
      "epoch": 3.0678781400320685,
      "grad_norm": 1.4366097450256348,
      "learning_rate": 0.00015704010005956415,
      "loss": 1.596,
      "step": 11480
    },
    {
      "epoch": 3.0705505077498665,
      "grad_norm": 1.4396469593048096,
      "learning_rate": 0.0001569711222283526,
      "loss": 1.5666,
      "step": 11490
    },
    {
      "epoch": 3.0732228754676645,
      "grad_norm": 1.4093619585037231,
      "learning_rate": 0.0001569021042414663,
      "loss": 1.5452,
      "step": 11500
    },
    {
      "epoch": 3.0758952431854625,
      "grad_norm": 1.4156750440597534,
      "learning_rate": 0.00015683304614755208,
      "loss": 1.6241,
      "step": 11510
    },
    {
      "epoch": 3.0785676109032605,
      "grad_norm": 1.5277137756347656,
      "learning_rate": 0.00015676394799528503,
      "loss": 1.5817,
      "step": 11520
    },
    {
      "epoch": 3.081239978621058,
      "grad_norm": 1.405486822128296,
      "learning_rate": 0.00015669480983336842,
      "loss": 1.5807,
      "step": 11530
    },
    {
      "epoch": 3.083912346338856,
      "grad_norm": 1.6159441471099854,
      "learning_rate": 0.00015662563171053382,
      "loss": 1.5476,
      "step": 11540
    },
    {
      "epoch": 3.086584714056654,
      "grad_norm": 1.4131660461425781,
      "learning_rate": 0.0001565564136755409,
      "loss": 1.575,
      "step": 11550
    },
    {
      "epoch": 3.089257081774452,
      "grad_norm": 1.4706740379333496,
      "learning_rate": 0.00015648715577717753,
      "loss": 1.5856,
      "step": 11560
    },
    {
      "epoch": 3.09192944949225,
      "grad_norm": 1.5544694662094116,
      "learning_rate": 0.00015641785806425957,
      "loss": 1.5473,
      "step": 11570
    },
    {
      "epoch": 3.094601817210048,
      "grad_norm": 1.465657114982605,
      "learning_rate": 0.00015634852058563103,
      "loss": 1.5396,
      "step": 11580
    },
    {
      "epoch": 3.097274184927846,
      "grad_norm": 1.37485933303833,
      "learning_rate": 0.00015627914339016386,
      "loss": 1.5261,
      "step": 11590
    },
    {
      "epoch": 3.099946552645644,
      "grad_norm": 1.4936021566390991,
      "learning_rate": 0.00015620972652675817,
      "loss": 1.5454,
      "step": 11600
    },
    {
      "epoch": 3.102618920363442,
      "grad_norm": 1.3922851085662842,
      "learning_rate": 0.00015614027004434187,
      "loss": 1.5764,
      "step": 11610
    },
    {
      "epoch": 3.10529128808124,
      "grad_norm": 1.4229692220687866,
      "learning_rate": 0.0001560707739918708,
      "loss": 1.5583,
      "step": 11620
    },
    {
      "epoch": 3.107963655799038,
      "grad_norm": 1.4172704219818115,
      "learning_rate": 0.0001560012384183288,
      "loss": 1.5571,
      "step": 11630
    },
    {
      "epoch": 3.110636023516836,
      "grad_norm": 1.5273038148880005,
      "learning_rate": 0.0001559316633727275,
      "loss": 1.5861,
      "step": 11640
    },
    {
      "epoch": 3.113308391234634,
      "grad_norm": 1.4015402793884277,
      "learning_rate": 0.00015586204890410637,
      "loss": 1.5473,
      "step": 11650
    },
    {
      "epoch": 3.115980758952432,
      "grad_norm": 1.5113670825958252,
      "learning_rate": 0.0001557923950615326,
      "loss": 1.6162,
      "step": 11660
    },
    {
      "epoch": 3.11865312667023,
      "grad_norm": 1.4393116235733032,
      "learning_rate": 0.0001557227018941012,
      "loss": 1.6128,
      "step": 11670
    },
    {
      "epoch": 3.121325494388028,
      "grad_norm": 1.5794994831085205,
      "learning_rate": 0.00015565296945093491,
      "loss": 1.5698,
      "step": 11680
    },
    {
      "epoch": 3.123997862105826,
      "grad_norm": 1.395870566368103,
      "learning_rate": 0.0001555831977811842,
      "loss": 1.6017,
      "step": 11690
    },
    {
      "epoch": 3.1266702298236235,
      "grad_norm": 1.6304802894592285,
      "learning_rate": 0.00015551338693402692,
      "loss": 1.583,
      "step": 11700
    },
    {
      "epoch": 3.1293425975414215,
      "grad_norm": 1.4940779209136963,
      "learning_rate": 0.00015544353695866892,
      "loss": 1.5248,
      "step": 11710
    },
    {
      "epoch": 3.1320149652592195,
      "grad_norm": 1.4320130348205566,
      "learning_rate": 0.0001553736479043433,
      "loss": 1.5636,
      "step": 11720
    },
    {
      "epoch": 3.1346873329770175,
      "grad_norm": 1.405187964439392,
      "learning_rate": 0.00015530371982031097,
      "loss": 1.5589,
      "step": 11730
    },
    {
      "epoch": 3.1373597006948155,
      "grad_norm": 1.5311636924743652,
      "learning_rate": 0.00015523375275586015,
      "loss": 1.4999,
      "step": 11740
    },
    {
      "epoch": 3.1400320684126135,
      "grad_norm": 1.614567756652832,
      "learning_rate": 0.0001551637467603066,
      "loss": 1.5888,
      "step": 11750
    },
    {
      "epoch": 3.1427044361304115,
      "grad_norm": 1.5212658643722534,
      "learning_rate": 0.0001550937018829936,
      "loss": 1.6661,
      "step": 11760
    },
    {
      "epoch": 3.1453768038482095,
      "grad_norm": 1.5886166095733643,
      "learning_rate": 0.00015502361817329167,
      "loss": 1.5447,
      "step": 11770
    },
    {
      "epoch": 3.1480491715660075,
      "grad_norm": 1.5095372200012207,
      "learning_rate": 0.00015495349568059888,
      "loss": 1.5925,
      "step": 11780
    },
    {
      "epoch": 3.1507215392838055,
      "grad_norm": 1.6610372066497803,
      "learning_rate": 0.00015488333445434047,
      "loss": 1.5864,
      "step": 11790
    },
    {
      "epoch": 3.1533939070016035,
      "grad_norm": 1.493183970451355,
      "learning_rate": 0.0001548131345439691,
      "loss": 1.5641,
      "step": 11800
    },
    {
      "epoch": 3.1560662747194015,
      "grad_norm": 1.4638108015060425,
      "learning_rate": 0.00015474289599896467,
      "loss": 1.5694,
      "step": 11810
    },
    {
      "epoch": 3.1587386424371995,
      "grad_norm": 1.4211833477020264,
      "learning_rate": 0.0001546726188688343,
      "loss": 1.6706,
      "step": 11820
    },
    {
      "epoch": 3.1614110101549975,
      "grad_norm": 1.4523656368255615,
      "learning_rate": 0.0001546023032031123,
      "loss": 1.6533,
      "step": 11830
    },
    {
      "epoch": 3.1640833778727955,
      "grad_norm": 1.50687575340271,
      "learning_rate": 0.00015453194905136014,
      "loss": 1.5646,
      "step": 11840
    },
    {
      "epoch": 3.1667557455905935,
      "grad_norm": 1.4127225875854492,
      "learning_rate": 0.00015446155646316645,
      "loss": 1.5041,
      "step": 11850
    },
    {
      "epoch": 3.169428113308391,
      "grad_norm": 1.570069432258606,
      "learning_rate": 0.00015439112548814687,
      "loss": 1.571,
      "step": 11860
    },
    {
      "epoch": 3.172100481026189,
      "grad_norm": 1.3592219352722168,
      "learning_rate": 0.0001543206561759442,
      "loss": 1.6437,
      "step": 11870
    },
    {
      "epoch": 3.174772848743987,
      "grad_norm": 1.427656888961792,
      "learning_rate": 0.0001542501485762282,
      "loss": 1.5855,
      "step": 11880
    },
    {
      "epoch": 3.177445216461785,
      "grad_norm": 1.4631425142288208,
      "learning_rate": 0.00015417960273869565,
      "loss": 1.6285,
      "step": 11890
    },
    {
      "epoch": 3.180117584179583,
      "grad_norm": 1.522918701171875,
      "learning_rate": 0.00015410901871307022,
      "loss": 1.5572,
      "step": 11900
    },
    {
      "epoch": 3.182789951897381,
      "grad_norm": 1.433417797088623,
      "learning_rate": 0.0001540383965491026,
      "loss": 1.5254,
      "step": 11910
    },
    {
      "epoch": 3.185462319615179,
      "grad_norm": 1.5127241611480713,
      "learning_rate": 0.00015396773629657026,
      "loss": 1.6015,
      "step": 11920
    },
    {
      "epoch": 3.188134687332977,
      "grad_norm": 1.500370979309082,
      "learning_rate": 0.00015389703800527756,
      "loss": 1.5406,
      "step": 11930
    },
    {
      "epoch": 3.190807055050775,
      "grad_norm": 1.4559721946716309,
      "learning_rate": 0.00015382630172505564,
      "loss": 1.574,
      "step": 11940
    },
    {
      "epoch": 3.193479422768573,
      "grad_norm": 1.5691871643066406,
      "learning_rate": 0.00015375552750576248,
      "loss": 1.5687,
      "step": 11950
    },
    {
      "epoch": 3.196151790486371,
      "grad_norm": 2.147005319595337,
      "learning_rate": 0.00015368471539728277,
      "loss": 1.6227,
      "step": 11960
    },
    {
      "epoch": 3.198824158204169,
      "grad_norm": 1.44551420211792,
      "learning_rate": 0.00015361386544952791,
      "loss": 1.6064,
      "step": 11970
    },
    {
      "epoch": 3.201496525921967,
      "grad_norm": 1.506608486175537,
      "learning_rate": 0.00015354297771243594,
      "loss": 1.5777,
      "step": 11980
    },
    {
      "epoch": 3.204168893639765,
      "grad_norm": 1.6357446908950806,
      "learning_rate": 0.00015347205223597155,
      "loss": 1.5225,
      "step": 11990
    },
    {
      "epoch": 3.206841261357563,
      "grad_norm": 1.5898518562316895,
      "learning_rate": 0.00015340108907012606,
      "loss": 1.5726,
      "step": 12000
    },
    {
      "epoch": 3.209513629075361,
      "grad_norm": 1.3775098323822021,
      "learning_rate": 0.00015333008826491735,
      "loss": 1.4897,
      "step": 12010
    },
    {
      "epoch": 3.212185996793159,
      "grad_norm": 1.532679796218872,
      "learning_rate": 0.00015325904987038974,
      "loss": 1.621,
      "step": 12020
    },
    {
      "epoch": 3.2148583645109565,
      "grad_norm": 1.4138544797897339,
      "learning_rate": 0.0001531879739366142,
      "loss": 1.6127,
      "step": 12030
    },
    {
      "epoch": 3.2175307322287545,
      "grad_norm": 1.481067419052124,
      "learning_rate": 0.00015311686051368805,
      "loss": 1.5911,
      "step": 12040
    },
    {
      "epoch": 3.2202030999465525,
      "grad_norm": 1.4436348676681519,
      "learning_rate": 0.00015304570965173504,
      "loss": 1.5967,
      "step": 12050
    },
    {
      "epoch": 3.2228754676643505,
      "grad_norm": 1.6716563701629639,
      "learning_rate": 0.0001529745214009053,
      "loss": 1.5598,
      "step": 12060
    },
    {
      "epoch": 3.2255478353821485,
      "grad_norm": 1.4454665184020996,
      "learning_rate": 0.00015290329581137545,
      "loss": 1.5709,
      "step": 12070
    },
    {
      "epoch": 3.2282202030999465,
      "grad_norm": 1.512351393699646,
      "learning_rate": 0.00015283203293334826,
      "loss": 1.6501,
      "step": 12080
    },
    {
      "epoch": 3.2308925708177445,
      "grad_norm": 1.4209846258163452,
      "learning_rate": 0.0001527607328170528,
      "loss": 1.6023,
      "step": 12090
    },
    {
      "epoch": 3.2335649385355425,
      "grad_norm": 1.5086133480072021,
      "learning_rate": 0.0001526893955127445,
      "loss": 1.561,
      "step": 12100
    },
    {
      "epoch": 3.2362373062533405,
      "grad_norm": 1.4856555461883545,
      "learning_rate": 0.0001526180210707049,
      "loss": 1.642,
      "step": 12110
    },
    {
      "epoch": 3.2389096739711385,
      "grad_norm": 1.5898791551589966,
      "learning_rate": 0.00015254660954124174,
      "loss": 1.6302,
      "step": 12120
    },
    {
      "epoch": 3.2415820416889365,
      "grad_norm": 1.4988596439361572,
      "learning_rate": 0.00015247516097468893,
      "loss": 1.5737,
      "step": 12130
    },
    {
      "epoch": 3.2442544094067345,
      "grad_norm": 1.62245512008667,
      "learning_rate": 0.00015240367542140642,
      "loss": 1.5576,
      "step": 12140
    },
    {
      "epoch": 3.2469267771245325,
      "grad_norm": 1.5361400842666626,
      "learning_rate": 0.00015233215293178036,
      "loss": 1.6377,
      "step": 12150
    },
    {
      "epoch": 3.2495991448423305,
      "grad_norm": 1.448956847190857,
      "learning_rate": 0.00015226059355622276,
      "loss": 1.6119,
      "step": 12160
    },
    {
      "epoch": 3.252271512560128,
      "grad_norm": 1.6497639417648315,
      "learning_rate": 0.00015218899734517177,
      "loss": 1.5715,
      "step": 12170
    },
    {
      "epoch": 3.254943880277926,
      "grad_norm": 1.4061638116836548,
      "learning_rate": 0.00015211736434909146,
      "loss": 1.601,
      "step": 12180
    },
    {
      "epoch": 3.257616247995724,
      "grad_norm": 1.408265233039856,
      "learning_rate": 0.00015204569461847175,
      "loss": 1.581,
      "step": 12190
    },
    {
      "epoch": 3.260288615713522,
      "grad_norm": 1.5317623615264893,
      "learning_rate": 0.0001519739882038286,
      "loss": 1.6239,
      "step": 12200
    },
    {
      "epoch": 3.26296098343132,
      "grad_norm": 1.4291706085205078,
      "learning_rate": 0.0001519022451557037,
      "loss": 1.5913,
      "step": 12210
    },
    {
      "epoch": 3.265633351149118,
      "grad_norm": 1.454978346824646,
      "learning_rate": 0.0001518304655246646,
      "loss": 1.5593,
      "step": 12220
    },
    {
      "epoch": 3.268305718866916,
      "grad_norm": 1.5863350629806519,
      "learning_rate": 0.00015175864936130466,
      "loss": 1.5995,
      "step": 12230
    },
    {
      "epoch": 3.270978086584714,
      "grad_norm": 1.6901382207870483,
      "learning_rate": 0.00015168679671624303,
      "loss": 1.5685,
      "step": 12240
    },
    {
      "epoch": 3.273650454302512,
      "grad_norm": 1.4609698057174683,
      "learning_rate": 0.0001516149076401244,
      "loss": 1.5684,
      "step": 12250
    },
    {
      "epoch": 3.27632282202031,
      "grad_norm": 1.5242557525634766,
      "learning_rate": 0.00015154298218361935,
      "loss": 1.5681,
      "step": 12260
    },
    {
      "epoch": 3.278995189738108,
      "grad_norm": 1.4671461582183838,
      "learning_rate": 0.00015147102039742396,
      "loss": 1.6675,
      "step": 12270
    },
    {
      "epoch": 3.281667557455906,
      "grad_norm": 1.7080386877059937,
      "learning_rate": 0.00015139902233226,
      "loss": 1.5686,
      "step": 12280
    },
    {
      "epoch": 3.284339925173704,
      "grad_norm": 1.4940818548202515,
      "learning_rate": 0.0001513269880388747,
      "loss": 1.6327,
      "step": 12290
    },
    {
      "epoch": 3.287012292891502,
      "grad_norm": 1.4788765907287598,
      "learning_rate": 0.00015125491756804097,
      "loss": 1.6308,
      "step": 12300
    },
    {
      "epoch": 3.2896846606093,
      "grad_norm": 1.609084963798523,
      "learning_rate": 0.00015118281097055712,
      "loss": 1.5448,
      "step": 12310
    },
    {
      "epoch": 3.292357028327098,
      "grad_norm": 1.4446818828582764,
      "learning_rate": 0.000151110668297247,
      "loss": 1.5586,
      "step": 12320
    },
    {
      "epoch": 3.295029396044896,
      "grad_norm": 1.4681098461151123,
      "learning_rate": 0.00015103848959895978,
      "loss": 1.6612,
      "step": 12330
    },
    {
      "epoch": 3.297701763762694,
      "grad_norm": 1.649408221244812,
      "learning_rate": 0.00015096627492657008,
      "loss": 1.5422,
      "step": 12340
    },
    {
      "epoch": 3.3003741314804915,
      "grad_norm": 1.343523383140564,
      "learning_rate": 0.0001508940243309779,
      "loss": 1.5831,
      "step": 12350
    },
    {
      "epoch": 3.3030464991982895,
      "grad_norm": 1.6102079153060913,
      "learning_rate": 0.00015082173786310857,
      "loss": 1.6207,
      "step": 12360
    },
    {
      "epoch": 3.3057188669160875,
      "grad_norm": 1.4566102027893066,
      "learning_rate": 0.00015074941557391265,
      "loss": 1.6514,
      "step": 12370
    },
    {
      "epoch": 3.3083912346338855,
      "grad_norm": 1.4235904216766357,
      "learning_rate": 0.00015067705751436593,
      "loss": 1.6241,
      "step": 12380
    },
    {
      "epoch": 3.3110636023516835,
      "grad_norm": 1.3748440742492676,
      "learning_rate": 0.00015060466373546954,
      "loss": 1.6408,
      "step": 12390
    },
    {
      "epoch": 3.3137359700694815,
      "grad_norm": 1.5889629125595093,
      "learning_rate": 0.0001505322342882496,
      "loss": 1.4889,
      "step": 12400
    },
    {
      "epoch": 3.3164083377872795,
      "grad_norm": 1.455016851425171,
      "learning_rate": 0.00015045976922375756,
      "loss": 1.5592,
      "step": 12410
    },
    {
      "epoch": 3.3190807055050775,
      "grad_norm": 1.4698208570480347,
      "learning_rate": 0.0001503872685930698,
      "loss": 1.6376,
      "step": 12420
    },
    {
      "epoch": 3.3217530732228755,
      "grad_norm": 1.435241937637329,
      "learning_rate": 0.00015031473244728793,
      "loss": 1.6346,
      "step": 12430
    },
    {
      "epoch": 3.3244254409406735,
      "grad_norm": 1.4528361558914185,
      "learning_rate": 0.0001502421608375385,
      "loss": 1.546,
      "step": 12440
    },
    {
      "epoch": 3.3270978086584715,
      "grad_norm": 1.5781196355819702,
      "learning_rate": 0.00015016955381497304,
      "loss": 1.5376,
      "step": 12450
    },
    {
      "epoch": 3.3297701763762695,
      "grad_norm": 1.466631531715393,
      "learning_rate": 0.0001500969114307681,
      "loss": 1.6496,
      "step": 12460
    },
    {
      "epoch": 3.3324425440940675,
      "grad_norm": 1.4700963497161865,
      "learning_rate": 0.00015002423373612514,
      "loss": 1.5212,
      "step": 12470
    },
    {
      "epoch": 3.3351149118118655,
      "grad_norm": 1.5303999185562134,
      "learning_rate": 0.00014995152078227045,
      "loss": 1.615,
      "step": 12480
    },
    {
      "epoch": 3.3377872795296635,
      "grad_norm": 1.5123428106307983,
      "learning_rate": 0.00014987877262045526,
      "loss": 1.6091,
      "step": 12490
    },
    {
      "epoch": 3.340459647247461,
      "grad_norm": 1.481543779373169,
      "learning_rate": 0.00014980598930195555,
      "loss": 1.5989,
      "step": 12500
    },
    {
      "epoch": 3.343132014965259,
      "grad_norm": 1.5110771656036377,
      "learning_rate": 0.00014973317087807212,
      "loss": 1.6372,
      "step": 12510
    },
    {
      "epoch": 3.345804382683057,
      "grad_norm": 1.4414563179016113,
      "learning_rate": 0.00014966031740013048,
      "loss": 1.6062,
      "step": 12520
    },
    {
      "epoch": 3.348476750400855,
      "grad_norm": 1.542820692062378,
      "learning_rate": 0.00014958742891948092,
      "loss": 1.632,
      "step": 12530
    },
    {
      "epoch": 3.351149118118653,
      "grad_norm": 1.557808518409729,
      "learning_rate": 0.00014951450548749827,
      "loss": 1.6191,
      "step": 12540
    },
    {
      "epoch": 3.353821485836451,
      "grad_norm": 1.4685314893722534,
      "learning_rate": 0.00014944154715558215,
      "loss": 1.6241,
      "step": 12550
    },
    {
      "epoch": 3.356493853554249,
      "grad_norm": 1.5369318723678589,
      "learning_rate": 0.0001493685539751566,
      "loss": 1.5917,
      "step": 12560
    },
    {
      "epoch": 3.359166221272047,
      "grad_norm": 1.484265685081482,
      "learning_rate": 0.00014929552599767044,
      "loss": 1.5976,
      "step": 12570
    },
    {
      "epoch": 3.361838588989845,
      "grad_norm": 1.4627116918563843,
      "learning_rate": 0.00014922246327459684,
      "loss": 1.5765,
      "step": 12580
    },
    {
      "epoch": 3.364510956707643,
      "grad_norm": 1.5683720111846924,
      "learning_rate": 0.0001491493658574335,
      "loss": 1.6757,
      "step": 12590
    },
    {
      "epoch": 3.367183324425441,
      "grad_norm": 1.498773217201233,
      "learning_rate": 0.00014907623379770263,
      "loss": 1.5233,
      "step": 12600
    },
    {
      "epoch": 3.369855692143239,
      "grad_norm": 1.517832636833191,
      "learning_rate": 0.00014900306714695082,
      "loss": 1.6862,
      "step": 12610
    },
    {
      "epoch": 3.372528059861037,
      "grad_norm": 1.419629693031311,
      "learning_rate": 0.00014892986595674903,
      "loss": 1.5996,
      "step": 12620
    },
    {
      "epoch": 3.375200427578835,
      "grad_norm": 1.5775679349899292,
      "learning_rate": 0.0001488566302786926,
      "loss": 1.5628,
      "step": 12630
    },
    {
      "epoch": 3.377872795296633,
      "grad_norm": 1.6346704959869385,
      "learning_rate": 0.00014878336016440114,
      "loss": 1.6387,
      "step": 12640
    },
    {
      "epoch": 3.380545163014431,
      "grad_norm": 1.4277915954589844,
      "learning_rate": 0.00014871005566551853,
      "loss": 1.6293,
      "step": 12650
    },
    {
      "epoch": 3.383217530732229,
      "grad_norm": 1.3590073585510254,
      "learning_rate": 0.00014863671683371297,
      "loss": 1.5521,
      "step": 12660
    },
    {
      "epoch": 3.385889898450027,
      "grad_norm": 1.6856508255004883,
      "learning_rate": 0.0001485633437206767,
      "loss": 1.6168,
      "step": 12670
    },
    {
      "epoch": 3.3885622661678245,
      "grad_norm": 1.4158910512924194,
      "learning_rate": 0.0001484899363781263,
      "loss": 1.5644,
      "step": 12680
    },
    {
      "epoch": 3.3912346338856225,
      "grad_norm": 1.4972407817840576,
      "learning_rate": 0.00014841649485780234,
      "loss": 1.6706,
      "step": 12690
    },
    {
      "epoch": 3.3939070016034205,
      "grad_norm": 1.4520623683929443,
      "learning_rate": 0.00014834301921146955,
      "loss": 1.6666,
      "step": 12700
    },
    {
      "epoch": 3.3965793693212185,
      "grad_norm": 1.3872112035751343,
      "learning_rate": 0.0001482695094909167,
      "loss": 1.5605,
      "step": 12710
    },
    {
      "epoch": 3.3992517370390165,
      "grad_norm": 1.4994367361068726,
      "learning_rate": 0.00014819596574795657,
      "loss": 1.5943,
      "step": 12720
    },
    {
      "epoch": 3.4019241047568145,
      "grad_norm": 1.5546753406524658,
      "learning_rate": 0.00014812238803442598,
      "loss": 1.6166,
      "step": 12730
    },
    {
      "epoch": 3.4045964724746125,
      "grad_norm": 1.5407750606536865,
      "learning_rate": 0.00014804877640218553,
      "loss": 1.6469,
      "step": 12740
    },
    {
      "epoch": 3.4072688401924105,
      "grad_norm": 1.43174409866333,
      "learning_rate": 0.00014797513090311986,
      "loss": 1.6406,
      "step": 12750
    },
    {
      "epoch": 3.4099412079102085,
      "grad_norm": 1.5952435731887817,
      "learning_rate": 0.00014790145158913756,
      "loss": 1.6278,
      "step": 12760
    },
    {
      "epoch": 3.4126135756280065,
      "grad_norm": 1.6369998455047607,
      "learning_rate": 0.00014782773851217086,
      "loss": 1.5891,
      "step": 12770
    },
    {
      "epoch": 3.4152859433458045,
      "grad_norm": 1.4865986108779907,
      "learning_rate": 0.00014775399172417588,
      "loss": 1.5867,
      "step": 12780
    },
    {
      "epoch": 3.4179583110636025,
      "grad_norm": 1.5650240182876587,
      "learning_rate": 0.0001476802112771325,
      "loss": 1.6074,
      "step": 12790
    },
    {
      "epoch": 3.4206306787814005,
      "grad_norm": 1.4547903537750244,
      "learning_rate": 0.00014760639722304436,
      "loss": 1.6048,
      "step": 12800
    },
    {
      "epoch": 3.4233030464991985,
      "grad_norm": 1.4975963830947876,
      "learning_rate": 0.00014753254961393872,
      "loss": 1.5787,
      "step": 12810
    },
    {
      "epoch": 3.425975414216996,
      "grad_norm": 1.6132854223251343,
      "learning_rate": 0.00014745866850186654,
      "loss": 1.6289,
      "step": 12820
    },
    {
      "epoch": 3.428647781934794,
      "grad_norm": 1.3988043069839478,
      "learning_rate": 0.00014738475393890235,
      "loss": 1.6492,
      "step": 12830
    },
    {
      "epoch": 3.431320149652592,
      "grad_norm": 1.5922125577926636,
      "learning_rate": 0.00014731080597714432,
      "loss": 1.6155,
      "step": 12840
    },
    {
      "epoch": 3.43399251737039,
      "grad_norm": 1.4675812721252441,
      "learning_rate": 0.0001472368246687141,
      "loss": 1.6425,
      "step": 12850
    },
    {
      "epoch": 3.436664885088188,
      "grad_norm": 1.4547349214553833,
      "learning_rate": 0.00014716281006575685,
      "loss": 1.6382,
      "step": 12860
    },
    {
      "epoch": 3.439337252805986,
      "grad_norm": 1.455011248588562,
      "learning_rate": 0.00014708876222044122,
      "loss": 1.6046,
      "step": 12870
    },
    {
      "epoch": 3.442009620523784,
      "grad_norm": 1.469198226928711,
      "learning_rate": 0.00014701468118495935,
      "loss": 1.6146,
      "step": 12880
    },
    {
      "epoch": 3.444681988241582,
      "grad_norm": 1.5194551944732666,
      "learning_rate": 0.0001469405670115267,
      "loss": 1.59,
      "step": 12890
    },
    {
      "epoch": 3.44735435595938,
      "grad_norm": 1.4832775592803955,
      "learning_rate": 0.000146866419752382,
      "loss": 1.5836,
      "step": 12900
    },
    {
      "epoch": 3.450026723677178,
      "grad_norm": 1.5915850400924683,
      "learning_rate": 0.00014679223945978748,
      "loss": 1.6516,
      "step": 12910
    },
    {
      "epoch": 3.452699091394976,
      "grad_norm": 1.58090341091156,
      "learning_rate": 0.00014671802618602855,
      "loss": 1.6118,
      "step": 12920
    },
    {
      "epoch": 3.455371459112774,
      "grad_norm": 1.323840856552124,
      "learning_rate": 0.00014664377998341391,
      "loss": 1.57,
      "step": 12930
    },
    {
      "epoch": 3.458043826830572,
      "grad_norm": 1.7080026865005493,
      "learning_rate": 0.00014656950090427543,
      "loss": 1.5692,
      "step": 12940
    },
    {
      "epoch": 3.46071619454837,
      "grad_norm": 1.403754472732544,
      "learning_rate": 0.00014649518900096815,
      "loss": 1.5972,
      "step": 12950
    },
    {
      "epoch": 3.463388562266168,
      "grad_norm": 1.4818941354751587,
      "learning_rate": 0.00014642084432587028,
      "loss": 1.6036,
      "step": 12960
    },
    {
      "epoch": 3.466060929983966,
      "grad_norm": 1.4573633670806885,
      "learning_rate": 0.00014634646693138315,
      "loss": 1.626,
      "step": 12970
    },
    {
      "epoch": 3.468733297701764,
      "grad_norm": 1.5643197298049927,
      "learning_rate": 0.00014627205686993107,
      "loss": 1.622,
      "step": 12980
    },
    {
      "epoch": 3.471405665419562,
      "grad_norm": 1.3899352550506592,
      "learning_rate": 0.0001461976141939615,
      "loss": 1.6618,
      "step": 12990
    },
    {
      "epoch": 3.4740780331373595,
      "grad_norm": 1.5522507429122925,
      "learning_rate": 0.0001461231389559447,
      "loss": 1.5913,
      "step": 13000
    },
    {
      "epoch": 3.4767504008551575,
      "grad_norm": 1.4575941562652588,
      "learning_rate": 0.00014604863120837403,
      "loss": 1.6049,
      "step": 13010
    },
    {
      "epoch": 3.4794227685729555,
      "grad_norm": 1.5314279794692993,
      "learning_rate": 0.00014597409100376577,
      "loss": 1.6405,
      "step": 13020
    },
    {
      "epoch": 3.4820951362907535,
      "grad_norm": 1.4816808700561523,
      "learning_rate": 0.000145899518394659,
      "loss": 1.5926,
      "step": 13030
    },
    {
      "epoch": 3.4847675040085515,
      "grad_norm": 1.5141303539276123,
      "learning_rate": 0.00014582491343361563,
      "loss": 1.5615,
      "step": 13040
    },
    {
      "epoch": 3.4874398717263495,
      "grad_norm": 1.5823702812194824,
      "learning_rate": 0.0001457502761732205,
      "loss": 1.6383,
      "step": 13050
    },
    {
      "epoch": 3.4901122394441475,
      "grad_norm": 1.444441795349121,
      "learning_rate": 0.0001456756066660811,
      "loss": 1.5515,
      "step": 13060
    },
    {
      "epoch": 3.4927846071619455,
      "grad_norm": 1.5254098176956177,
      "learning_rate": 0.00014560090496482765,
      "loss": 1.6191,
      "step": 13070
    },
    {
      "epoch": 3.4954569748797435,
      "grad_norm": 1.6265779733657837,
      "learning_rate": 0.00014552617112211308,
      "loss": 1.5522,
      "step": 13080
    },
    {
      "epoch": 3.4981293425975415,
      "grad_norm": 1.4543230533599854,
      "learning_rate": 0.00014545140519061308,
      "loss": 1.5356,
      "step": 13090
    },
    {
      "epoch": 3.5008017103153395,
      "grad_norm": 1.4384841918945312,
      "learning_rate": 0.0001453766072230258,
      "loss": 1.514,
      "step": 13100
    },
    {
      "epoch": 3.5034740780331375,
      "grad_norm": 1.6042413711547852,
      "learning_rate": 0.000145301777272072,
      "loss": 1.6351,
      "step": 13110
    },
    {
      "epoch": 3.5061464457509355,
      "grad_norm": 1.4373027086257935,
      "learning_rate": 0.00014522691539049508,
      "loss": 1.6038,
      "step": 13120
    },
    {
      "epoch": 3.508818813468733,
      "grad_norm": 1.5652923583984375,
      "learning_rate": 0.00014515202163106088,
      "loss": 1.6038,
      "step": 13130
    },
    {
      "epoch": 3.511491181186531,
      "grad_norm": 1.4025564193725586,
      "learning_rate": 0.0001450770960465577,
      "loss": 1.5811,
      "step": 13140
    },
    {
      "epoch": 3.514163548904329,
      "grad_norm": 1.440600037574768,
      "learning_rate": 0.00014500213868979625,
      "loss": 1.5859,
      "step": 13150
    },
    {
      "epoch": 3.516835916622127,
      "grad_norm": 1.5395610332489014,
      "learning_rate": 0.00014492714961360977,
      "loss": 1.6319,
      "step": 13160
    },
    {
      "epoch": 3.519508284339925,
      "grad_norm": 1.471100926399231,
      "learning_rate": 0.00014485212887085363,
      "loss": 1.5959,
      "step": 13170
    },
    {
      "epoch": 3.522180652057723,
      "grad_norm": 1.5483871698379517,
      "learning_rate": 0.00014477707651440572,
      "loss": 1.5763,
      "step": 13180
    },
    {
      "epoch": 3.524853019775521,
      "grad_norm": 1.4020414352416992,
      "learning_rate": 0.00014470199259716612,
      "loss": 1.6219,
      "step": 13190
    },
    {
      "epoch": 3.527525387493319,
      "grad_norm": 1.3737884759902954,
      "learning_rate": 0.00014462687717205718,
      "loss": 1.6411,
      "step": 13200
    },
    {
      "epoch": 3.530197755211117,
      "grad_norm": 1.392750859260559,
      "learning_rate": 0.00014455173029202345,
      "loss": 1.6043,
      "step": 13210
    },
    {
      "epoch": 3.532870122928915,
      "grad_norm": 1.5100160837173462,
      "learning_rate": 0.00014447655201003164,
      "loss": 1.621,
      "step": 13220
    },
    {
      "epoch": 3.535542490646713,
      "grad_norm": 1.4548851251602173,
      "learning_rate": 0.00014440134237907065,
      "loss": 1.6193,
      "step": 13230
    },
    {
      "epoch": 3.538214858364511,
      "grad_norm": 1.6431235074996948,
      "learning_rate": 0.00014432610145215138,
      "loss": 1.6293,
      "step": 13240
    },
    {
      "epoch": 3.540887226082309,
      "grad_norm": 1.6478720903396606,
      "learning_rate": 0.0001442508292823069,
      "loss": 1.6586,
      "step": 13250
    },
    {
      "epoch": 3.543559593800107,
      "grad_norm": 1.408124327659607,
      "learning_rate": 0.00014417552592259217,
      "loss": 1.6555,
      "step": 13260
    },
    {
      "epoch": 3.546231961517905,
      "grad_norm": 1.6119136810302734,
      "learning_rate": 0.00014410019142608427,
      "loss": 1.7086,
      "step": 13270
    },
    {
      "epoch": 3.548904329235703,
      "grad_norm": 1.6675114631652832,
      "learning_rate": 0.0001440248258458821,
      "loss": 1.5894,
      "step": 13280
    },
    {
      "epoch": 3.551576696953501,
      "grad_norm": 1.4569134712219238,
      "learning_rate": 0.00014394942923510662,
      "loss": 1.5982,
      "step": 13290
    },
    {
      "epoch": 3.554249064671299,
      "grad_norm": 1.5121387243270874,
      "learning_rate": 0.0001438740016469005,
      "loss": 1.6715,
      "step": 13300
    },
    {
      "epoch": 3.556921432389097,
      "grad_norm": 1.4674450159072876,
      "learning_rate": 0.00014379854313442835,
      "loss": 1.5981,
      "step": 13310
    },
    {
      "epoch": 3.559593800106895,
      "grad_norm": 1.4840753078460693,
      "learning_rate": 0.00014372305375087653,
      "loss": 1.6548,
      "step": 13320
    },
    {
      "epoch": 3.5622661678246925,
      "grad_norm": 1.6692088842391968,
      "learning_rate": 0.0001436475335494532,
      "loss": 1.7056,
      "step": 13330
    },
    {
      "epoch": 3.5649385355424905,
      "grad_norm": 1.4009300470352173,
      "learning_rate": 0.0001435719825833882,
      "loss": 1.6539,
      "step": 13340
    },
    {
      "epoch": 3.5676109032602885,
      "grad_norm": 1.411410927772522,
      "learning_rate": 0.00014349640090593308,
      "loss": 1.5697,
      "step": 13350
    },
    {
      "epoch": 3.5702832709780865,
      "grad_norm": 1.4743436574935913,
      "learning_rate": 0.000143420788570361,
      "loss": 1.5776,
      "step": 13360
    },
    {
      "epoch": 3.5729556386958845,
      "grad_norm": 1.4356075525283813,
      "learning_rate": 0.00014334514562996682,
      "loss": 1.5673,
      "step": 13370
    },
    {
      "epoch": 3.5756280064136825,
      "grad_norm": 1.4502203464508057,
      "learning_rate": 0.00014326947213806677,
      "loss": 1.5098,
      "step": 13380
    },
    {
      "epoch": 3.5783003741314805,
      "grad_norm": 1.4926453828811646,
      "learning_rate": 0.0001431937681479989,
      "loss": 1.6879,
      "step": 13390
    },
    {
      "epoch": 3.5809727418492785,
      "grad_norm": 1.4945552349090576,
      "learning_rate": 0.00014311803371312257,
      "loss": 1.5889,
      "step": 13400
    },
    {
      "epoch": 3.5836451095670765,
      "grad_norm": 1.5944979190826416,
      "learning_rate": 0.0001430422688868186,
      "loss": 1.6233,
      "step": 13410
    },
    {
      "epoch": 3.5863174772848745,
      "grad_norm": 1.4976544380187988,
      "learning_rate": 0.00014296647372248928,
      "loss": 1.5894,
      "step": 13420
    },
    {
      "epoch": 3.5889898450026725,
      "grad_norm": 1.353517770767212,
      "learning_rate": 0.0001428906482735583,
      "loss": 1.6334,
      "step": 13430
    },
    {
      "epoch": 3.5916622127204705,
      "grad_norm": 1.5863207578659058,
      "learning_rate": 0.00014281479259347062,
      "loss": 1.6212,
      "step": 13440
    },
    {
      "epoch": 3.594334580438268,
      "grad_norm": 1.4160728454589844,
      "learning_rate": 0.0001427389067356926,
      "loss": 1.5923,
      "step": 13450
    },
    {
      "epoch": 3.597006948156066,
      "grad_norm": 1.4250168800354004,
      "learning_rate": 0.00014266299075371182,
      "loss": 1.6324,
      "step": 13460
    },
    {
      "epoch": 3.599679315873864,
      "grad_norm": 1.418119192123413,
      "learning_rate": 0.00014258704470103704,
      "loss": 1.6647,
      "step": 13470
    },
    {
      "epoch": 3.602351683591662,
      "grad_norm": 1.6640735864639282,
      "learning_rate": 0.00014251106863119839,
      "loss": 1.5728,
      "step": 13480
    },
    {
      "epoch": 3.60502405130946,
      "grad_norm": 1.7105728387832642,
      "learning_rate": 0.00014243506259774695,
      "loss": 1.5729,
      "step": 13490
    },
    {
      "epoch": 3.607696419027258,
      "grad_norm": 1.4894355535507202,
      "learning_rate": 0.00014235902665425503,
      "loss": 1.6117,
      "step": 13500
    },
    {
      "epoch": 3.610368786745056,
      "grad_norm": 1.5188177824020386,
      "learning_rate": 0.00014228296085431598,
      "loss": 1.6428,
      "step": 13510
    },
    {
      "epoch": 3.613041154462854,
      "grad_norm": 1.4988083839416504,
      "learning_rate": 0.0001422068652515443,
      "loss": 1.6426,
      "step": 13520
    },
    {
      "epoch": 3.615713522180652,
      "grad_norm": 1.4839959144592285,
      "learning_rate": 0.00014213073989957533,
      "loss": 1.5665,
      "step": 13530
    },
    {
      "epoch": 3.61838588989845,
      "grad_norm": 1.4822163581848145,
      "learning_rate": 0.0001420545848520655,
      "loss": 1.6193,
      "step": 13540
    },
    {
      "epoch": 3.621058257616248,
      "grad_norm": 1.5349560976028442,
      "learning_rate": 0.00014197840016269211,
      "loss": 1.565,
      "step": 13550
    },
    {
      "epoch": 3.623730625334046,
      "grad_norm": 1.4710910320281982,
      "learning_rate": 0.00014190218588515342,
      "loss": 1.5949,
      "step": 13560
    },
    {
      "epoch": 3.626402993051844,
      "grad_norm": 1.5057989358901978,
      "learning_rate": 0.0001418259420731685,
      "loss": 1.6028,
      "step": 13570
    },
    {
      "epoch": 3.629075360769642,
      "grad_norm": 1.491237998008728,
      "learning_rate": 0.00014174966878047714,
      "loss": 1.7041,
      "step": 13580
    },
    {
      "epoch": 3.63174772848744,
      "grad_norm": 1.5255022048950195,
      "learning_rate": 0.0001416733660608401,
      "loss": 1.5834,
      "step": 13590
    },
    {
      "epoch": 3.634420096205238,
      "grad_norm": 1.4522159099578857,
      "learning_rate": 0.00014159703396803876,
      "loss": 1.6645,
      "step": 13600
    },
    {
      "epoch": 3.637092463923036,
      "grad_norm": 1.438297986984253,
      "learning_rate": 0.00014152067255587526,
      "loss": 1.5807,
      "step": 13610
    },
    {
      "epoch": 3.639764831640834,
      "grad_norm": 1.5111329555511475,
      "learning_rate": 0.00014144428187817234,
      "loss": 1.5751,
      "step": 13620
    },
    {
      "epoch": 3.642437199358632,
      "grad_norm": 1.5307103395462036,
      "learning_rate": 0.0001413678619887734,
      "loss": 1.6652,
      "step": 13630
    },
    {
      "epoch": 3.64510956707643,
      "grad_norm": 1.4978127479553223,
      "learning_rate": 0.0001412914129415425,
      "loss": 1.6253,
      "step": 13640
    },
    {
      "epoch": 3.6477819347942275,
      "grad_norm": 1.4403648376464844,
      "learning_rate": 0.0001412149347903641,
      "loss": 1.5672,
      "step": 13650
    },
    {
      "epoch": 3.6504543025120255,
      "grad_norm": 1.7367560863494873,
      "learning_rate": 0.0001411384275891433,
      "loss": 1.6484,
      "step": 13660
    },
    {
      "epoch": 3.6531266702298235,
      "grad_norm": 1.4811859130859375,
      "learning_rate": 0.00014106189139180566,
      "loss": 1.695,
      "step": 13670
    },
    {
      "epoch": 3.6557990379476215,
      "grad_norm": 1.4210959672927856,
      "learning_rate": 0.0001409853262522971,
      "loss": 1.6099,
      "step": 13680
    },
    {
      "epoch": 3.6584714056654195,
      "grad_norm": 1.5033975839614868,
      "learning_rate": 0.00014090873222458408,
      "loss": 1.6033,
      "step": 13690
    },
    {
      "epoch": 3.6611437733832175,
      "grad_norm": 1.6621878147125244,
      "learning_rate": 0.00014083210936265323,
      "loss": 1.6461,
      "step": 13700
    },
    {
      "epoch": 3.6638161411010155,
      "grad_norm": 1.5738985538482666,
      "learning_rate": 0.0001407554577205117,
      "loss": 1.6294,
      "step": 13710
    },
    {
      "epoch": 3.6664885088188135,
      "grad_norm": 1.5229177474975586,
      "learning_rate": 0.0001406787773521867,
      "loss": 1.5999,
      "step": 13720
    },
    {
      "epoch": 3.6691608765366115,
      "grad_norm": 1.506054401397705,
      "learning_rate": 0.000140602068311726,
      "loss": 1.5744,
      "step": 13730
    },
    {
      "epoch": 3.6718332442544095,
      "grad_norm": 1.5278657674789429,
      "learning_rate": 0.00014052533065319727,
      "loss": 1.645,
      "step": 13740
    },
    {
      "epoch": 3.6745056119722075,
      "grad_norm": 1.3956048488616943,
      "learning_rate": 0.00014044856443068852,
      "loss": 1.6006,
      "step": 13750
    },
    {
      "epoch": 3.6771779796900055,
      "grad_norm": 1.4022210836410522,
      "learning_rate": 0.00014037176969830789,
      "loss": 1.5826,
      "step": 13760
    },
    {
      "epoch": 3.6798503474078035,
      "grad_norm": 1.491211175918579,
      "learning_rate": 0.00014029494651018353,
      "loss": 1.6265,
      "step": 13770
    },
    {
      "epoch": 3.682522715125601,
      "grad_norm": 1.6359549760818481,
      "learning_rate": 0.0001402180949204637,
      "loss": 1.5713,
      "step": 13780
    },
    {
      "epoch": 3.685195082843399,
      "grad_norm": 1.328031063079834,
      "learning_rate": 0.0001401412149833167,
      "loss": 1.6198,
      "step": 13790
    },
    {
      "epoch": 3.687867450561197,
      "grad_norm": 1.4872769117355347,
      "learning_rate": 0.00014006430675293077,
      "loss": 1.7155,
      "step": 13800
    },
    {
      "epoch": 3.690539818278995,
      "grad_norm": 1.4812895059585571,
      "learning_rate": 0.00013998737028351414,
      "loss": 1.5703,
      "step": 13810
    },
    {
      "epoch": 3.693212185996793,
      "grad_norm": 1.5284749269485474,
      "learning_rate": 0.00013991040562929486,
      "loss": 1.6417,
      "step": 13820
    },
    {
      "epoch": 3.695884553714591,
      "grad_norm": 1.4381073713302612,
      "learning_rate": 0.00013983341284452093,
      "loss": 1.6485,
      "step": 13830
    },
    {
      "epoch": 3.698556921432389,
      "grad_norm": 1.5141340494155884,
      "learning_rate": 0.00013975639198346014,
      "loss": 1.6118,
      "step": 13840
    },
    {
      "epoch": 3.701229289150187,
      "grad_norm": 1.5383323431015015,
      "learning_rate": 0.0001396793431004001,
      "loss": 1.6584,
      "step": 13850
    },
    {
      "epoch": 3.703901656867985,
      "grad_norm": 1.5171921253204346,
      "learning_rate": 0.0001396022662496481,
      "loss": 1.6573,
      "step": 13860
    },
    {
      "epoch": 3.706574024585783,
      "grad_norm": 1.4966976642608643,
      "learning_rate": 0.0001395251614855312,
      "loss": 1.6719,
      "step": 13870
    },
    {
      "epoch": 3.709246392303581,
      "grad_norm": 1.4085315465927124,
      "learning_rate": 0.00013944802886239617,
      "loss": 1.5832,
      "step": 13880
    },
    {
      "epoch": 3.711918760021379,
      "grad_norm": 1.5608574151992798,
      "learning_rate": 0.00013937086843460932,
      "loss": 1.6529,
      "step": 13890
    },
    {
      "epoch": 3.714591127739177,
      "grad_norm": 1.4574732780456543,
      "learning_rate": 0.00013929368025655662,
      "loss": 1.5927,
      "step": 13900
    },
    {
      "epoch": 3.717263495456975,
      "grad_norm": 1.6187883615493774,
      "learning_rate": 0.00013921646438264356,
      "loss": 1.6188,
      "step": 13910
    },
    {
      "epoch": 3.719935863174773,
      "grad_norm": 1.416913628578186,
      "learning_rate": 0.00013913922086729525,
      "loss": 1.6252,
      "step": 13920
    },
    {
      "epoch": 3.722608230892571,
      "grad_norm": 1.4689831733703613,
      "learning_rate": 0.00013906194976495616,
      "loss": 1.6159,
      "step": 13930
    },
    {
      "epoch": 3.725280598610369,
      "grad_norm": 1.452686071395874,
      "learning_rate": 0.00013898465113009026,
      "loss": 1.5895,
      "step": 13940
    },
    {
      "epoch": 3.727952966328167,
      "grad_norm": 1.4064078330993652,
      "learning_rate": 0.0001389073250171809,
      "loss": 1.6013,
      "step": 13950
    },
    {
      "epoch": 3.730625334045965,
      "grad_norm": 1.5133086442947388,
      "learning_rate": 0.00013882997148073086,
      "loss": 1.6002,
      "step": 13960
    },
    {
      "epoch": 3.733297701763763,
      "grad_norm": 1.4783916473388672,
      "learning_rate": 0.00013875259057526224,
      "loss": 1.6582,
      "step": 13970
    },
    {
      "epoch": 3.7359700694815605,
      "grad_norm": 1.5024664402008057,
      "learning_rate": 0.0001386751823553163,
      "loss": 1.6967,
      "step": 13980
    },
    {
      "epoch": 3.7386424371993585,
      "grad_norm": 1.5081539154052734,
      "learning_rate": 0.00013859774687545372,
      "loss": 1.6207,
      "step": 13990
    },
    {
      "epoch": 3.7413148049171565,
      "grad_norm": 1.4246116876602173,
      "learning_rate": 0.00013852028419025433,
      "loss": 1.6206,
      "step": 14000
    },
    {
      "epoch": 3.7439871726349545,
      "grad_norm": 1.4178860187530518,
      "learning_rate": 0.00013844279435431712,
      "loss": 1.5968,
      "step": 14010
    },
    {
      "epoch": 3.7466595403527525,
      "grad_norm": 1.4298408031463623,
      "learning_rate": 0.0001383652774222602,
      "loss": 1.6642,
      "step": 14020
    },
    {
      "epoch": 3.7493319080705505,
      "grad_norm": 1.5386979579925537,
      "learning_rate": 0.00013828773344872082,
      "loss": 1.6282,
      "step": 14030
    },
    {
      "epoch": 3.7520042757883485,
      "grad_norm": 1.481295108795166,
      "learning_rate": 0.0001382101624883553,
      "loss": 1.6478,
      "step": 14040
    },
    {
      "epoch": 3.7546766435061465,
      "grad_norm": 1.4294100999832153,
      "learning_rate": 0.00013813256459583898,
      "loss": 1.6048,
      "step": 14050
    },
    {
      "epoch": 3.7573490112239445,
      "grad_norm": 1.4785627126693726,
      "learning_rate": 0.00013805493982586611,
      "loss": 1.6811,
      "step": 14060
    },
    {
      "epoch": 3.7600213789417425,
      "grad_norm": 1.4725021123886108,
      "learning_rate": 0.00013797728823314992,
      "loss": 1.661,
      "step": 14070
    },
    {
      "epoch": 3.7626937466595405,
      "grad_norm": 1.6216282844543457,
      "learning_rate": 0.00013789960987242264,
      "loss": 1.6109,
      "step": 14080
    },
    {
      "epoch": 3.7653661143773385,
      "grad_norm": 1.6100808382034302,
      "learning_rate": 0.00013782190479843522,
      "loss": 1.6065,
      "step": 14090
    },
    {
      "epoch": 3.768038482095136,
      "grad_norm": 1.464152216911316,
      "learning_rate": 0.00013774417306595756,
      "loss": 1.6429,
      "step": 14100
    },
    {
      "epoch": 3.770710849812934,
      "grad_norm": 1.5179142951965332,
      "learning_rate": 0.0001376664147297782,
      "loss": 1.5804,
      "step": 14110
    },
    {
      "epoch": 3.773383217530732,
      "grad_norm": 1.4735183715820312,
      "learning_rate": 0.00013758862984470467,
      "loss": 1.6397,
      "step": 14120
    },
    {
      "epoch": 3.77605558524853,
      "grad_norm": 1.6094744205474854,
      "learning_rate": 0.000137510818465563,
      "loss": 1.6399,
      "step": 14130
    },
    {
      "epoch": 3.778727952966328,
      "grad_norm": 1.3748862743377686,
      "learning_rate": 0.0001374329806471979,
      "loss": 1.5782,
      "step": 14140
    },
    {
      "epoch": 3.781400320684126,
      "grad_norm": 1.5021231174468994,
      "learning_rate": 0.00013735511644447288,
      "loss": 1.6185,
      "step": 14150
    },
    {
      "epoch": 3.784072688401924,
      "grad_norm": 1.4829655885696411,
      "learning_rate": 0.0001372772259122699,
      "loss": 1.564,
      "step": 14160
    },
    {
      "epoch": 3.786745056119722,
      "grad_norm": 1.4910268783569336,
      "learning_rate": 0.0001371993091054896,
      "loss": 1.5831,
      "step": 14170
    },
    {
      "epoch": 3.78941742383752,
      "grad_norm": 1.4876508712768555,
      "learning_rate": 0.00013712136607905095,
      "loss": 1.6153,
      "step": 14180
    },
    {
      "epoch": 3.792089791555318,
      "grad_norm": 1.646272897720337,
      "learning_rate": 0.00013704339688789158,
      "loss": 1.6347,
      "step": 14190
    },
    {
      "epoch": 3.794762159273116,
      "grad_norm": 1.5087896585464478,
      "learning_rate": 0.0001369654015869675,
      "loss": 1.6687,
      "step": 14200
    },
    {
      "epoch": 3.797434526990914,
      "grad_norm": 1.557450294494629,
      "learning_rate": 0.00013688738023125315,
      "loss": 1.6415,
      "step": 14210
    },
    {
      "epoch": 3.800106894708712,
      "grad_norm": 1.4506268501281738,
      "learning_rate": 0.0001368093328757412,
      "loss": 1.5379,
      "step": 14220
    },
    {
      "epoch": 3.80277926242651,
      "grad_norm": 1.3444366455078125,
      "learning_rate": 0.00013673125957544286,
      "loss": 1.6135,
      "step": 14230
    },
    {
      "epoch": 3.805451630144308,
      "grad_norm": 1.3975130319595337,
      "learning_rate": 0.0001366531603853875,
      "loss": 1.6524,
      "step": 14240
    },
    {
      "epoch": 3.808123997862106,
      "grad_norm": 1.4078421592712402,
      "learning_rate": 0.00013657503536062271,
      "loss": 1.685,
      "step": 14250
    },
    {
      "epoch": 3.810796365579904,
      "grad_norm": 1.5333263874053955,
      "learning_rate": 0.00013649688455621438,
      "loss": 1.6449,
      "step": 14260
    },
    {
      "epoch": 3.813468733297702,
      "grad_norm": 1.5773098468780518,
      "learning_rate": 0.00013641870802724655,
      "loss": 1.5746,
      "step": 14270
    },
    {
      "epoch": 3.8161411010155,
      "grad_norm": 1.3503774404525757,
      "learning_rate": 0.00013634050582882128,
      "loss": 1.5675,
      "step": 14280
    },
    {
      "epoch": 3.818813468733298,
      "grad_norm": 1.4837092161178589,
      "learning_rate": 0.00013626227801605888,
      "loss": 1.677,
      "step": 14290
    },
    {
      "epoch": 3.8214858364510955,
      "grad_norm": 1.3931994438171387,
      "learning_rate": 0.00013618402464409763,
      "loss": 1.5816,
      "step": 14300
    },
    {
      "epoch": 3.8241582041688935,
      "grad_norm": 1.601622223854065,
      "learning_rate": 0.00013610574576809382,
      "loss": 1.5674,
      "step": 14310
    },
    {
      "epoch": 3.8268305718866915,
      "grad_norm": 1.4701857566833496,
      "learning_rate": 0.00013602744144322178,
      "loss": 1.6683,
      "step": 14320
    },
    {
      "epoch": 3.8295029396044895,
      "grad_norm": 1.4149401187896729,
      "learning_rate": 0.00013594911172467374,
      "loss": 1.6973,
      "step": 14330
    },
    {
      "epoch": 3.8321753073222875,
      "grad_norm": 1.4293116331100464,
      "learning_rate": 0.00013587075666765975,
      "loss": 1.6284,
      "step": 14340
    },
    {
      "epoch": 3.8348476750400855,
      "grad_norm": 1.3892247676849365,
      "learning_rate": 0.00013579237632740788,
      "loss": 1.6965,
      "step": 14350
    },
    {
      "epoch": 3.8375200427578835,
      "grad_norm": 1.5815978050231934,
      "learning_rate": 0.00013571397075916384,
      "loss": 1.5702,
      "step": 14360
    },
    {
      "epoch": 3.8401924104756815,
      "grad_norm": 1.4898732900619507,
      "learning_rate": 0.00013563554001819135,
      "loss": 1.5876,
      "step": 14370
    },
    {
      "epoch": 3.8428647781934795,
      "grad_norm": 1.4967395067214966,
      "learning_rate": 0.00013555708415977162,
      "loss": 1.6004,
      "step": 14380
    },
    {
      "epoch": 3.8455371459112775,
      "grad_norm": 1.3928459882736206,
      "learning_rate": 0.00013547860323920373,
      "loss": 1.6381,
      "step": 14390
    },
    {
      "epoch": 3.8482095136290755,
      "grad_norm": 1.4960623979568481,
      "learning_rate": 0.00013540009731180435,
      "loss": 1.6381,
      "step": 14400
    },
    {
      "epoch": 3.8508818813468735,
      "grad_norm": 1.4093217849731445,
      "learning_rate": 0.00013532156643290784,
      "loss": 1.6366,
      "step": 14410
    },
    {
      "epoch": 3.853554249064671,
      "grad_norm": 1.447858214378357,
      "learning_rate": 0.00013524301065786606,
      "loss": 1.6263,
      "step": 14420
    },
    {
      "epoch": 3.856226616782469,
      "grad_norm": 1.536370873451233,
      "learning_rate": 0.00013516443004204853,
      "loss": 1.741,
      "step": 14430
    },
    {
      "epoch": 3.858898984500267,
      "grad_norm": 1.43016517162323,
      "learning_rate": 0.0001350858246408422,
      "loss": 1.6889,
      "step": 14440
    },
    {
      "epoch": 3.861571352218065,
      "grad_norm": 1.4906656742095947,
      "learning_rate": 0.00013500719450965147,
      "loss": 1.5983,
      "step": 14450
    },
    {
      "epoch": 3.864243719935863,
      "grad_norm": 1.5198860168457031,
      "learning_rate": 0.00013492853970389825,
      "loss": 1.675,
      "step": 14460
    },
    {
      "epoch": 3.866916087653661,
      "grad_norm": 1.5043227672576904,
      "learning_rate": 0.00013484986027902175,
      "loss": 1.6472,
      "step": 14470
    },
    {
      "epoch": 3.869588455371459,
      "grad_norm": 1.4240082502365112,
      "learning_rate": 0.00013477115629047863,
      "loss": 1.6388,
      "step": 14480
    },
    {
      "epoch": 3.872260823089257,
      "grad_norm": 1.4360988140106201,
      "learning_rate": 0.00013469242779374282,
      "loss": 1.6289,
      "step": 14490
    },
    {
      "epoch": 3.874933190807055,
      "grad_norm": 1.422788143157959,
      "learning_rate": 0.00013461367484430547,
      "loss": 1.5923,
      "step": 14500
    },
    {
      "epoch": 3.877605558524853,
      "grad_norm": 1.4307031631469727,
      "learning_rate": 0.00013453489749767504,
      "loss": 1.6479,
      "step": 14510
    },
    {
      "epoch": 3.880277926242651,
      "grad_norm": 1.4923681020736694,
      "learning_rate": 0.00013445609580937714,
      "loss": 1.6039,
      "step": 14520
    },
    {
      "epoch": 3.882950293960449,
      "grad_norm": 1.5482001304626465,
      "learning_rate": 0.00013437726983495462,
      "loss": 1.6784,
      "step": 14530
    },
    {
      "epoch": 3.885622661678247,
      "grad_norm": 1.447283387184143,
      "learning_rate": 0.0001342984196299673,
      "loss": 1.5706,
      "step": 14540
    },
    {
      "epoch": 3.888295029396045,
      "grad_norm": 1.4632364511489868,
      "learning_rate": 0.00013421954524999225,
      "loss": 1.6004,
      "step": 14550
    },
    {
      "epoch": 3.890967397113843,
      "grad_norm": 1.7275340557098389,
      "learning_rate": 0.00013414064675062342,
      "loss": 1.647,
      "step": 14560
    },
    {
      "epoch": 3.893639764831641,
      "grad_norm": 1.4087649583816528,
      "learning_rate": 0.00013406172418747191,
      "loss": 1.6432,
      "step": 14570
    },
    {
      "epoch": 3.896312132549439,
      "grad_norm": 1.3862338066101074,
      "learning_rate": 0.00013398277761616565,
      "loss": 1.6009,
      "step": 14580
    },
    {
      "epoch": 3.898984500267237,
      "grad_norm": 1.5114810466766357,
      "learning_rate": 0.00013390380709234955,
      "loss": 1.5917,
      "step": 14590
    },
    {
      "epoch": 3.901656867985035,
      "grad_norm": 1.423058032989502,
      "learning_rate": 0.0001338248126716854,
      "loss": 1.6769,
      "step": 14600
    },
    {
      "epoch": 3.904329235702833,
      "grad_norm": 1.4373550415039062,
      "learning_rate": 0.00013374579440985192,
      "loss": 1.6135,
      "step": 14610
    },
    {
      "epoch": 3.9070016034206305,
      "grad_norm": 1.6480869054794312,
      "learning_rate": 0.0001336667523625444,
      "loss": 1.5833,
      "step": 14620
    },
    {
      "epoch": 3.9096739711384285,
      "grad_norm": 1.372557520866394,
      "learning_rate": 0.00013358768658547517,
      "loss": 1.6504,
      "step": 14630
    },
    {
      "epoch": 3.9123463388562265,
      "grad_norm": 1.424797534942627,
      "learning_rate": 0.00013350859713437305,
      "loss": 1.6278,
      "step": 14640
    },
    {
      "epoch": 3.9150187065740245,
      "grad_norm": 1.443160057067871,
      "learning_rate": 0.00013342948406498377,
      "loss": 1.6827,
      "step": 14650
    },
    {
      "epoch": 3.9176910742918225,
      "grad_norm": 1.5389927625656128,
      "learning_rate": 0.0001333503474330695,
      "loss": 1.6948,
      "step": 14660
    },
    {
      "epoch": 3.9203634420096205,
      "grad_norm": 1.5167105197906494,
      "learning_rate": 0.00013327118729440914,
      "loss": 1.6306,
      "step": 14670
    },
    {
      "epoch": 3.9230358097274185,
      "grad_norm": 1.5564993619918823,
      "learning_rate": 0.00013319200370479813,
      "loss": 1.527,
      "step": 14680
    },
    {
      "epoch": 3.9257081774452165,
      "grad_norm": 1.3414462804794312,
      "learning_rate": 0.0001331127967200485,
      "loss": 1.6159,
      "step": 14690
    },
    {
      "epoch": 3.9283805451630145,
      "grad_norm": 1.637369155883789,
      "learning_rate": 0.0001330335663959886,
      "loss": 1.5953,
      "step": 14700
    },
    {
      "epoch": 3.9310529128808125,
      "grad_norm": 1.5273038148880005,
      "learning_rate": 0.00013295431278846339,
      "loss": 1.6665,
      "step": 14710
    },
    {
      "epoch": 3.9337252805986105,
      "grad_norm": 1.38784658908844,
      "learning_rate": 0.0001328750359533342,
      "loss": 1.6488,
      "step": 14720
    },
    {
      "epoch": 3.9363976483164085,
      "grad_norm": 1.5924921035766602,
      "learning_rate": 0.00013279573594647873,
      "loss": 1.617,
      "step": 14730
    },
    {
      "epoch": 3.939070016034206,
      "grad_norm": 1.6329870223999023,
      "learning_rate": 0.000132716412823791,
      "loss": 1.5561,
      "step": 14740
    },
    {
      "epoch": 3.941742383752004,
      "grad_norm": 1.403551697731018,
      "learning_rate": 0.00013263706664118127,
      "loss": 1.6606,
      "step": 14750
    },
    {
      "epoch": 3.944414751469802,
      "grad_norm": 1.438462734222412,
      "learning_rate": 0.00013255769745457616,
      "loss": 1.6405,
      "step": 14760
    },
    {
      "epoch": 3.9470871191876,
      "grad_norm": 1.7514116764068604,
      "learning_rate": 0.00013247830531991846,
      "loss": 1.6367,
      "step": 14770
    },
    {
      "epoch": 3.949759486905398,
      "grad_norm": 1.4804353713989258,
      "learning_rate": 0.0001323988902931671,
      "loss": 1.6431,
      "step": 14780
    },
    {
      "epoch": 3.952431854623196,
      "grad_norm": 1.52631676197052,
      "learning_rate": 0.00013231945243029724,
      "loss": 1.6394,
      "step": 14790
    },
    {
      "epoch": 3.955104222340994,
      "grad_norm": 1.5103986263275146,
      "learning_rate": 0.00013223999178730002,
      "loss": 1.6736,
      "step": 14800
    },
    {
      "epoch": 3.957776590058792,
      "grad_norm": 1.4855271577835083,
      "learning_rate": 0.00013216050842018267,
      "loss": 1.6506,
      "step": 14810
    },
    {
      "epoch": 3.96044895777659,
      "grad_norm": 1.7139270305633545,
      "learning_rate": 0.0001320810023849685,
      "loss": 1.6167,
      "step": 14820
    },
    {
      "epoch": 3.963121325494388,
      "grad_norm": 1.4682615995407104,
      "learning_rate": 0.0001320014737376967,
      "loss": 1.6672,
      "step": 14830
    },
    {
      "epoch": 3.965793693212186,
      "grad_norm": 1.5905976295471191,
      "learning_rate": 0.00013192192253442248,
      "loss": 1.635,
      "step": 14840
    },
    {
      "epoch": 3.968466060929984,
      "grad_norm": 1.48699152469635,
      "learning_rate": 0.00013184234883121694,
      "loss": 1.638,
      "step": 14850
    },
    {
      "epoch": 3.971138428647782,
      "grad_norm": 1.4956756830215454,
      "learning_rate": 0.00013176275268416698,
      "loss": 1.6672,
      "step": 14860
    },
    {
      "epoch": 3.97381079636558,
      "grad_norm": 1.6715842485427856,
      "learning_rate": 0.00013168313414937537,
      "loss": 1.6804,
      "step": 14870
    },
    {
      "epoch": 3.976483164083378,
      "grad_norm": 1.3169384002685547,
      "learning_rate": 0.00013160349328296061,
      "loss": 1.6972,
      "step": 14880
    },
    {
      "epoch": 3.979155531801176,
      "grad_norm": 1.6806117296218872,
      "learning_rate": 0.00013152383014105706,
      "loss": 1.6283,
      "step": 14890
    },
    {
      "epoch": 3.981827899518974,
      "grad_norm": 1.4955217838287354,
      "learning_rate": 0.00013144414477981467,
      "loss": 1.5623,
      "step": 14900
    },
    {
      "epoch": 3.984500267236772,
      "grad_norm": 1.4891384840011597,
      "learning_rate": 0.000131364437255399,
      "loss": 1.6444,
      "step": 14910
    },
    {
      "epoch": 3.98717263495457,
      "grad_norm": 1.5744833946228027,
      "learning_rate": 0.0001312847076239914,
      "loss": 1.6152,
      "step": 14920
    },
    {
      "epoch": 3.989845002672368,
      "grad_norm": 1.3829927444458008,
      "learning_rate": 0.00013120495594178867,
      "loss": 1.6469,
      "step": 14930
    },
    {
      "epoch": 3.9925173703901655,
      "grad_norm": 1.3851839303970337,
      "learning_rate": 0.00013112518226500322,
      "loss": 1.6489,
      "step": 14940
    },
    {
      "epoch": 3.9951897381079635,
      "grad_norm": 1.3766522407531738,
      "learning_rate": 0.0001310453866498629,
      "loss": 1.6432,
      "step": 14950
    },
    {
      "epoch": 3.9978621058257615,
      "grad_norm": 1.4151705503463745,
      "learning_rate": 0.00013096556915261109,
      "loss": 1.5371,
      "step": 14960
    },
    {
      "epoch": 4.00053447354356,
      "grad_norm": 1.4186418056488037,
      "learning_rate": 0.00013088572982950658,
      "loss": 1.6407,
      "step": 14970
    },
    {
      "epoch": 4.003206841261358,
      "grad_norm": 1.7752667665481567,
      "learning_rate": 0.00013080586873682345,
      "loss": 1.5291,
      "step": 14980
    },
    {
      "epoch": 4.005879208979156,
      "grad_norm": 1.5340584516525269,
      "learning_rate": 0.0001307259859308513,
      "loss": 1.483,
      "step": 14990
    },
    {
      "epoch": 4.008551576696954,
      "grad_norm": 1.703906774520874,
      "learning_rate": 0.00013064608146789487,
      "loss": 1.376,
      "step": 15000
    },
    {
      "epoch": 4.011223944414751,
      "grad_norm": 1.5323362350463867,
      "learning_rate": 0.00013056615540427426,
      "loss": 1.5141,
      "step": 15010
    },
    {
      "epoch": 4.013896312132549,
      "grad_norm": 1.5463310480117798,
      "learning_rate": 0.00013048620779632474,
      "loss": 1.4556,
      "step": 15020
    },
    {
      "epoch": 4.016568679850347,
      "grad_norm": 1.6218100786209106,
      "learning_rate": 0.0001304062387003968,
      "loss": 1.3566,
      "step": 15030
    },
    {
      "epoch": 4.019241047568145,
      "grad_norm": 1.5737180709838867,
      "learning_rate": 0.00013032624817285607,
      "loss": 1.4114,
      "step": 15040
    },
    {
      "epoch": 4.021913415285943,
      "grad_norm": 1.626188039779663,
      "learning_rate": 0.00013024623627008335,
      "loss": 1.4558,
      "step": 15050
    },
    {
      "epoch": 4.024585783003741,
      "grad_norm": 1.8518829345703125,
      "learning_rate": 0.0001301662030484743,
      "loss": 1.4462,
      "step": 15060
    },
    {
      "epoch": 4.027258150721539,
      "grad_norm": 1.687781572341919,
      "learning_rate": 0.00013008614856443984,
      "loss": 1.4625,
      "step": 15070
    },
    {
      "epoch": 4.029930518439337,
      "grad_norm": 1.7048814296722412,
      "learning_rate": 0.00013000607287440578,
      "loss": 1.5125,
      "step": 15080
    },
    {
      "epoch": 4.032602886157135,
      "grad_norm": 1.6215217113494873,
      "learning_rate": 0.00012992597603481287,
      "loss": 1.4001,
      "step": 15090
    },
    {
      "epoch": 4.035275253874933,
      "grad_norm": 1.71282958984375,
      "learning_rate": 0.00012984585810211675,
      "loss": 1.4038,
      "step": 15100
    },
    {
      "epoch": 4.037947621592731,
      "grad_norm": 1.5622607469558716,
      "learning_rate": 0.00012976571913278793,
      "loss": 1.3764,
      "step": 15110
    },
    {
      "epoch": 4.040619989310529,
      "grad_norm": 1.5343408584594727,
      "learning_rate": 0.00012968555918331184,
      "loss": 1.449,
      "step": 15120
    },
    {
      "epoch": 4.043292357028327,
      "grad_norm": 1.6914194822311401,
      "learning_rate": 0.00012960537831018863,
      "loss": 1.4227,
      "step": 15130
    },
    {
      "epoch": 4.045964724746125,
      "grad_norm": 1.560380458831787,
      "learning_rate": 0.00012952517656993312,
      "loss": 1.4633,
      "step": 15140
    },
    {
      "epoch": 4.048637092463923,
      "grad_norm": 1.57892644405365,
      "learning_rate": 0.00012944495401907496,
      "loss": 1.4584,
      "step": 15150
    },
    {
      "epoch": 4.051309460181721,
      "grad_norm": 1.5968350172042847,
      "learning_rate": 0.00012936471071415845,
      "loss": 1.4349,
      "step": 15160
    },
    {
      "epoch": 4.053981827899519,
      "grad_norm": 1.717559576034546,
      "learning_rate": 0.00012928444671174246,
      "loss": 1.471,
      "step": 15170
    },
    {
      "epoch": 4.056654195617317,
      "grad_norm": 1.6292623281478882,
      "learning_rate": 0.00012920416206840048,
      "loss": 1.5261,
      "step": 15180
    },
    {
      "epoch": 4.059326563335115,
      "grad_norm": 1.6947565078735352,
      "learning_rate": 0.00012912385684072053,
      "loss": 1.4252,
      "step": 15190
    },
    {
      "epoch": 4.061998931052913,
      "grad_norm": 1.6276607513427734,
      "learning_rate": 0.00012904353108530517,
      "loss": 1.4703,
      "step": 15200
    },
    {
      "epoch": 4.064671298770711,
      "grad_norm": 1.6942775249481201,
      "learning_rate": 0.00012896318485877147,
      "loss": 1.5128,
      "step": 15210
    },
    {
      "epoch": 4.067343666488509,
      "grad_norm": 1.716341495513916,
      "learning_rate": 0.00012888281821775077,
      "loss": 1.5169,
      "step": 15220
    },
    {
      "epoch": 4.070016034206307,
      "grad_norm": 1.8272207975387573,
      "learning_rate": 0.00012880243121888898,
      "loss": 1.4746,
      "step": 15230
    },
    {
      "epoch": 4.072688401924105,
      "grad_norm": 1.6025763750076294,
      "learning_rate": 0.00012872202391884624,
      "loss": 1.4425,
      "step": 15240
    },
    {
      "epoch": 4.075360769641903,
      "grad_norm": 1.7037829160690308,
      "learning_rate": 0.00012864159637429712,
      "loss": 1.4231,
      "step": 15250
    },
    {
      "epoch": 4.078033137359701,
      "grad_norm": 1.6595052480697632,
      "learning_rate": 0.00012856114864193027,
      "loss": 1.5113,
      "step": 15260
    },
    {
      "epoch": 4.080705505077499,
      "grad_norm": 1.6561627388000488,
      "learning_rate": 0.00012848068077844874,
      "loss": 1.5073,
      "step": 15270
    },
    {
      "epoch": 4.083377872795297,
      "grad_norm": 1.6377439498901367,
      "learning_rate": 0.00012840019284056972,
      "loss": 1.5123,
      "step": 15280
    },
    {
      "epoch": 4.086050240513095,
      "grad_norm": 1.4633361101150513,
      "learning_rate": 0.00012831968488502454,
      "loss": 1.4576,
      "step": 15290
    },
    {
      "epoch": 4.088722608230893,
      "grad_norm": 1.716450572013855,
      "learning_rate": 0.00012823915696855858,
      "loss": 1.4567,
      "step": 15300
    },
    {
      "epoch": 4.091394975948691,
      "grad_norm": 1.7311489582061768,
      "learning_rate": 0.00012815860914793144,
      "loss": 1.3882,
      "step": 15310
    },
    {
      "epoch": 4.094067343666489,
      "grad_norm": 1.7296059131622314,
      "learning_rate": 0.00012807804147991654,
      "loss": 1.4566,
      "step": 15320
    },
    {
      "epoch": 4.096739711384286,
      "grad_norm": 1.5362845659255981,
      "learning_rate": 0.0001279974540213015,
      "loss": 1.3942,
      "step": 15330
    },
    {
      "epoch": 4.099412079102084,
      "grad_norm": 1.4967952966690063,
      "learning_rate": 0.00012791684682888777,
      "loss": 1.4921,
      "step": 15340
    },
    {
      "epoch": 4.102084446819882,
      "grad_norm": 1.711885690689087,
      "learning_rate": 0.00012783621995949077,
      "loss": 1.467,
      "step": 15350
    },
    {
      "epoch": 4.10475681453768,
      "grad_norm": 1.605841875076294,
      "learning_rate": 0.00012775557346993965,
      "loss": 1.4144,
      "step": 15360
    },
    {
      "epoch": 4.107429182255478,
      "grad_norm": 1.652406930923462,
      "learning_rate": 0.00012767490741707758,
      "loss": 1.4517,
      "step": 15370
    },
    {
      "epoch": 4.110101549973276,
      "grad_norm": 1.6053305864334106,
      "learning_rate": 0.0001275942218577614,
      "loss": 1.4935,
      "step": 15380
    },
    {
      "epoch": 4.112773917691074,
      "grad_norm": 1.6521443128585815,
      "learning_rate": 0.00012751351684886172,
      "loss": 1.4737,
      "step": 15390
    },
    {
      "epoch": 4.115446285408872,
      "grad_norm": 1.6109788417816162,
      "learning_rate": 0.0001274327924472629,
      "loss": 1.5309,
      "step": 15400
    },
    {
      "epoch": 4.11811865312667,
      "grad_norm": 1.7511069774627686,
      "learning_rate": 0.00012735204870986292,
      "loss": 1.449,
      "step": 15410
    },
    {
      "epoch": 4.120791020844468,
      "grad_norm": 1.600865364074707,
      "learning_rate": 0.00012727128569357343,
      "loss": 1.5098,
      "step": 15420
    },
    {
      "epoch": 4.123463388562266,
      "grad_norm": 1.8842839002609253,
      "learning_rate": 0.0001271905034553196,
      "loss": 1.4775,
      "step": 15430
    },
    {
      "epoch": 4.126135756280064,
      "grad_norm": 1.5898122787475586,
      "learning_rate": 0.00012710970205204023,
      "loss": 1.4892,
      "step": 15440
    },
    {
      "epoch": 4.128808123997862,
      "grad_norm": 1.571650743484497,
      "learning_rate": 0.00012702888154068762,
      "loss": 1.4514,
      "step": 15450
    },
    {
      "epoch": 4.13148049171566,
      "grad_norm": 1.657235860824585,
      "learning_rate": 0.00012694804197822742,
      "loss": 1.4535,
      "step": 15460
    },
    {
      "epoch": 4.134152859433458,
      "grad_norm": 1.6750986576080322,
      "learning_rate": 0.00012686718342163887,
      "loss": 1.4641,
      "step": 15470
    },
    {
      "epoch": 4.136825227151256,
      "grad_norm": 1.49517023563385,
      "learning_rate": 0.00012678630592791453,
      "loss": 1.4382,
      "step": 15480
    },
    {
      "epoch": 4.139497594869054,
      "grad_norm": 1.6910736560821533,
      "learning_rate": 0.00012670540955406025,
      "loss": 1.4938,
      "step": 15490
    },
    {
      "epoch": 4.142169962586852,
      "grad_norm": 1.5621329545974731,
      "learning_rate": 0.00012662449435709528,
      "loss": 1.4439,
      "step": 15500
    },
    {
      "epoch": 4.14484233030465,
      "grad_norm": 1.5824596881866455,
      "learning_rate": 0.00012654356039405214,
      "loss": 1.4253,
      "step": 15510
    },
    {
      "epoch": 4.147514698022448,
      "grad_norm": 1.777213454246521,
      "learning_rate": 0.00012646260772197646,
      "loss": 1.4832,
      "step": 15520
    },
    {
      "epoch": 4.150187065740246,
      "grad_norm": 1.6388694047927856,
      "learning_rate": 0.00012638163639792715,
      "loss": 1.5706,
      "step": 15530
    },
    {
      "epoch": 4.152859433458044,
      "grad_norm": 1.7183681726455688,
      "learning_rate": 0.0001263006464789763,
      "loss": 1.477,
      "step": 15540
    },
    {
      "epoch": 4.155531801175842,
      "grad_norm": 1.6367969512939453,
      "learning_rate": 0.00012621963802220896,
      "loss": 1.5303,
      "step": 15550
    },
    {
      "epoch": 4.15820416889364,
      "grad_norm": 1.7480429410934448,
      "learning_rate": 0.00012613861108472344,
      "loss": 1.3677,
      "step": 15560
    },
    {
      "epoch": 4.160876536611438,
      "grad_norm": 1.5859206914901733,
      "learning_rate": 0.00012605756572363095,
      "loss": 1.4675,
      "step": 15570
    },
    {
      "epoch": 4.163548904329236,
      "grad_norm": 1.7138357162475586,
      "learning_rate": 0.00012597650199605567,
      "loss": 1.4469,
      "step": 15580
    },
    {
      "epoch": 4.166221272047034,
      "grad_norm": 1.7989431619644165,
      "learning_rate": 0.00012589541995913482,
      "loss": 1.4298,
      "step": 15590
    },
    {
      "epoch": 4.168893639764832,
      "grad_norm": 1.6668016910552979,
      "learning_rate": 0.00012581431967001845,
      "loss": 1.4636,
      "step": 15600
    },
    {
      "epoch": 4.17156600748263,
      "grad_norm": 1.6256632804870605,
      "learning_rate": 0.00012573320118586956,
      "loss": 1.5082,
      "step": 15610
    },
    {
      "epoch": 4.174238375200428,
      "grad_norm": 1.6245003938674927,
      "learning_rate": 0.0001256520645638638,
      "loss": 1.5126,
      "step": 15620
    },
    {
      "epoch": 4.176910742918226,
      "grad_norm": 1.5841891765594482,
      "learning_rate": 0.0001255709098611898,
      "loss": 1.5234,
      "step": 15630
    },
    {
      "epoch": 4.179583110636024,
      "grad_norm": 1.6938221454620361,
      "learning_rate": 0.00012548973713504883,
      "loss": 1.513,
      "step": 15640
    },
    {
      "epoch": 4.182255478353822,
      "grad_norm": 1.8873212337493896,
      "learning_rate": 0.0001254085464426549,
      "loss": 1.5582,
      "step": 15650
    },
    {
      "epoch": 4.184927846071619,
      "grad_norm": 1.723900556564331,
      "learning_rate": 0.0001253273378412347,
      "loss": 1.4572,
      "step": 15660
    },
    {
      "epoch": 4.187600213789417,
      "grad_norm": 1.6873656511306763,
      "learning_rate": 0.0001252461113880274,
      "loss": 1.4842,
      "step": 15670
    },
    {
      "epoch": 4.190272581507215,
      "grad_norm": 1.7792420387268066,
      "learning_rate": 0.000125164867140285,
      "loss": 1.5768,
      "step": 15680
    },
    {
      "epoch": 4.192944949225013,
      "grad_norm": 1.718456745147705,
      "learning_rate": 0.00012508360515527182,
      "loss": 1.5009,
      "step": 15690
    },
    {
      "epoch": 4.195617316942811,
      "grad_norm": 1.7421867847442627,
      "learning_rate": 0.0001250023254902648,
      "loss": 1.4752,
      "step": 15700
    },
    {
      "epoch": 4.198289684660609,
      "grad_norm": 1.6288411617279053,
      "learning_rate": 0.00012492102820255334,
      "loss": 1.4887,
      "step": 15710
    },
    {
      "epoch": 4.200962052378407,
      "grad_norm": 1.5560775995254517,
      "learning_rate": 0.00012483971334943916,
      "loss": 1.5078,
      "step": 15720
    },
    {
      "epoch": 4.203634420096205,
      "grad_norm": 1.8239293098449707,
      "learning_rate": 0.0001247583809882365,
      "loss": 1.4998,
      "step": 15730
    },
    {
      "epoch": 4.206306787814003,
      "grad_norm": 1.5634268522262573,
      "learning_rate": 0.00012467703117627184,
      "loss": 1.4761,
      "step": 15740
    },
    {
      "epoch": 4.208979155531801,
      "grad_norm": 1.6948027610778809,
      "learning_rate": 0.000124595663970884,
      "loss": 1.4968,
      "step": 15750
    },
    {
      "epoch": 4.211651523249599,
      "grad_norm": 1.6174324750900269,
      "learning_rate": 0.00012451427942942404,
      "loss": 1.4994,
      "step": 15760
    },
    {
      "epoch": 4.214323890967397,
      "grad_norm": 1.7737600803375244,
      "learning_rate": 0.00012443287760925532,
      "loss": 1.3916,
      "step": 15770
    },
    {
      "epoch": 4.216996258685195,
      "grad_norm": 1.6498640775680542,
      "learning_rate": 0.00012435145856775324,
      "loss": 1.4857,
      "step": 15780
    },
    {
      "epoch": 4.219668626402993,
      "grad_norm": 1.6566587686538696,
      "learning_rate": 0.0001242700223623054,
      "loss": 1.5357,
      "step": 15790
    },
    {
      "epoch": 4.222340994120791,
      "grad_norm": 1.753976821899414,
      "learning_rate": 0.00012418856905031158,
      "loss": 1.4841,
      "step": 15800
    },
    {
      "epoch": 4.225013361838589,
      "grad_norm": 1.6846094131469727,
      "learning_rate": 0.00012410709868918352,
      "loss": 1.4906,
      "step": 15810
    },
    {
      "epoch": 4.227685729556387,
      "grad_norm": 1.6260082721710205,
      "learning_rate": 0.00012402561133634493,
      "loss": 1.5374,
      "step": 15820
    },
    {
      "epoch": 4.230358097274185,
      "grad_norm": 1.675095796585083,
      "learning_rate": 0.00012394410704923167,
      "loss": 1.5468,
      "step": 15830
    },
    {
      "epoch": 4.233030464991983,
      "grad_norm": 1.6872878074645996,
      "learning_rate": 0.00012386258588529137,
      "loss": 1.528,
      "step": 15840
    },
    {
      "epoch": 4.235702832709781,
      "grad_norm": 1.683913230895996,
      "learning_rate": 0.00012378104790198368,
      "loss": 1.5133,
      "step": 15850
    },
    {
      "epoch": 4.238375200427579,
      "grad_norm": 1.6218829154968262,
      "learning_rate": 0.00012369949315678,
      "loss": 1.4151,
      "step": 15860
    },
    {
      "epoch": 4.241047568145377,
      "grad_norm": 1.6738678216934204,
      "learning_rate": 0.00012361792170716355,
      "loss": 1.4968,
      "step": 15870
    },
    {
      "epoch": 4.243719935863175,
      "grad_norm": 1.6918460130691528,
      "learning_rate": 0.00012353633361062943,
      "loss": 1.4572,
      "step": 15880
    },
    {
      "epoch": 4.246392303580973,
      "grad_norm": 1.720625638961792,
      "learning_rate": 0.00012345472892468445,
      "loss": 1.5016,
      "step": 15890
    },
    {
      "epoch": 4.249064671298771,
      "grad_norm": 1.7121583223342896,
      "learning_rate": 0.00012337310770684697,
      "loss": 1.5141,
      "step": 15900
    },
    {
      "epoch": 4.251737039016569,
      "grad_norm": 1.6069037914276123,
      "learning_rate": 0.0001232914700146471,
      "loss": 1.4759,
      "step": 15910
    },
    {
      "epoch": 4.254409406734367,
      "grad_norm": 1.7477941513061523,
      "learning_rate": 0.00012320981590562666,
      "loss": 1.4975,
      "step": 15920
    },
    {
      "epoch": 4.257081774452165,
      "grad_norm": 1.6874366998672485,
      "learning_rate": 0.0001231281454373389,
      "loss": 1.4427,
      "step": 15930
    },
    {
      "epoch": 4.259754142169963,
      "grad_norm": 1.6226550340652466,
      "learning_rate": 0.0001230464586673486,
      "loss": 1.4729,
      "step": 15940
    },
    {
      "epoch": 4.262426509887761,
      "grad_norm": 1.574035882949829,
      "learning_rate": 0.0001229647556532321,
      "loss": 1.4162,
      "step": 15950
    },
    {
      "epoch": 4.265098877605558,
      "grad_norm": 1.7540584802627563,
      "learning_rate": 0.0001228830364525772,
      "loss": 1.5068,
      "step": 15960
    },
    {
      "epoch": 4.267771245323356,
      "grad_norm": 1.5365920066833496,
      "learning_rate": 0.0001228013011229831,
      "loss": 1.5271,
      "step": 15970
    },
    {
      "epoch": 4.270443613041154,
      "grad_norm": 1.6674189567565918,
      "learning_rate": 0.00012271954972206025,
      "loss": 1.4932,
      "step": 15980
    },
    {
      "epoch": 4.273115980758952,
      "grad_norm": 1.8557534217834473,
      "learning_rate": 0.00012263778230743057,
      "loss": 1.3994,
      "step": 15990
    },
    {
      "epoch": 4.27578834847675,
      "grad_norm": 1.6800191402435303,
      "learning_rate": 0.00012255599893672722,
      "loss": 1.5023,
      "step": 16000
    },
    {
      "epoch": 4.278460716194548,
      "grad_norm": 1.7056418657302856,
      "learning_rate": 0.00012247419966759463,
      "loss": 1.5013,
      "step": 16010
    },
    {
      "epoch": 4.281133083912346,
      "grad_norm": 1.7103627920150757,
      "learning_rate": 0.00012239238455768835,
      "loss": 1.5767,
      "step": 16020
    },
    {
      "epoch": 4.283805451630144,
      "grad_norm": 1.6329611539840698,
      "learning_rate": 0.0001223105536646752,
      "loss": 1.5312,
      "step": 16030
    },
    {
      "epoch": 4.286477819347942,
      "grad_norm": 1.7396610975265503,
      "learning_rate": 0.00012222870704623308,
      "loss": 1.5558,
      "step": 16040
    },
    {
      "epoch": 4.28915018706574,
      "grad_norm": 1.755860686302185,
      "learning_rate": 0.00012214684476005098,
      "loss": 1.4537,
      "step": 16050
    },
    {
      "epoch": 4.291822554783538,
      "grad_norm": 1.6175237894058228,
      "learning_rate": 0.00012206496686382891,
      "loss": 1.4715,
      "step": 16060
    },
    {
      "epoch": 4.294494922501336,
      "grad_norm": 1.626566767692566,
      "learning_rate": 0.00012198307341527789,
      "loss": 1.4918,
      "step": 16070
    },
    {
      "epoch": 4.297167290219134,
      "grad_norm": 1.6249829530715942,
      "learning_rate": 0.00012190116447211995,
      "loss": 1.5604,
      "step": 16080
    },
    {
      "epoch": 4.299839657936932,
      "grad_norm": 1.971280813217163,
      "learning_rate": 0.00012181924009208794,
      "loss": 1.4767,
      "step": 16090
    },
    {
      "epoch": 4.30251202565473,
      "grad_norm": 1.6820917129516602,
      "learning_rate": 0.00012173730033292571,
      "loss": 1.4629,
      "step": 16100
    },
    {
      "epoch": 4.305184393372528,
      "grad_norm": 1.669192910194397,
      "learning_rate": 0.00012165534525238789,
      "loss": 1.4328,
      "step": 16110
    },
    {
      "epoch": 4.307856761090326,
      "grad_norm": 1.638898253440857,
      "learning_rate": 0.0001215733749082399,
      "loss": 1.4909,
      "step": 16120
    },
    {
      "epoch": 4.310529128808124,
      "grad_norm": 1.646751880645752,
      "learning_rate": 0.00012149138935825792,
      "loss": 1.4865,
      "step": 16130
    },
    {
      "epoch": 4.313201496525922,
      "grad_norm": 1.757340908050537,
      "learning_rate": 0.00012140938866022888,
      "loss": 1.4045,
      "step": 16140
    },
    {
      "epoch": 4.31587386424372,
      "grad_norm": 1.6209243535995483,
      "learning_rate": 0.00012132737287195036,
      "loss": 1.4834,
      "step": 16150
    },
    {
      "epoch": 4.318546231961518,
      "grad_norm": 1.7393606901168823,
      "learning_rate": 0.00012124534205123057,
      "loss": 1.5536,
      "step": 16160
    },
    {
      "epoch": 4.321218599679316,
      "grad_norm": 1.7389954328536987,
      "learning_rate": 0.00012116329625588836,
      "loss": 1.5147,
      "step": 16170
    },
    {
      "epoch": 4.323890967397114,
      "grad_norm": 1.6584761142730713,
      "learning_rate": 0.00012108123554375306,
      "loss": 1.5098,
      "step": 16180
    },
    {
      "epoch": 4.326563335114912,
      "grad_norm": 1.5832411050796509,
      "learning_rate": 0.0001209991599726646,
      "loss": 1.4712,
      "step": 16190
    },
    {
      "epoch": 4.32923570283271,
      "grad_norm": 1.6773817539215088,
      "learning_rate": 0.00012091706960047329,
      "loss": 1.5145,
      "step": 16200
    },
    {
      "epoch": 4.331908070550508,
      "grad_norm": 1.6375465393066406,
      "learning_rate": 0.00012083496448503999,
      "loss": 1.4662,
      "step": 16210
    },
    {
      "epoch": 4.334580438268306,
      "grad_norm": 1.6932520866394043,
      "learning_rate": 0.00012075284468423582,
      "loss": 1.5005,
      "step": 16220
    },
    {
      "epoch": 4.337252805986104,
      "grad_norm": 1.6839436292648315,
      "learning_rate": 0.00012067071025594234,
      "loss": 1.5137,
      "step": 16230
    },
    {
      "epoch": 4.339925173703902,
      "grad_norm": 1.9438105821609497,
      "learning_rate": 0.00012058856125805142,
      "loss": 1.5039,
      "step": 16240
    },
    {
      "epoch": 4.3425975414217,
      "grad_norm": 1.6157655715942383,
      "learning_rate": 0.00012050639774846515,
      "loss": 1.4942,
      "step": 16250
    },
    {
      "epoch": 4.345269909139498,
      "grad_norm": 1.721543788909912,
      "learning_rate": 0.00012042421978509586,
      "loss": 1.5226,
      "step": 16260
    },
    {
      "epoch": 4.347942276857296,
      "grad_norm": 1.6050511598587036,
      "learning_rate": 0.00012034202742586611,
      "loss": 1.5069,
      "step": 16270
    },
    {
      "epoch": 4.350614644575094,
      "grad_norm": 1.7026658058166504,
      "learning_rate": 0.00012025982072870853,
      "loss": 1.5875,
      "step": 16280
    },
    {
      "epoch": 4.353287012292892,
      "grad_norm": 1.7754448652267456,
      "learning_rate": 0.000120177599751566,
      "loss": 1.5895,
      "step": 16290
    },
    {
      "epoch": 4.35595938001069,
      "grad_norm": 1.7176076173782349,
      "learning_rate": 0.00012009536455239123,
      "loss": 1.5354,
      "step": 16300
    },
    {
      "epoch": 4.358631747728487,
      "grad_norm": 1.6290333271026611,
      "learning_rate": 0.00012001311518914719,
      "loss": 1.4675,
      "step": 16310
    },
    {
      "epoch": 4.361304115446285,
      "grad_norm": 1.5428414344787598,
      "learning_rate": 0.00011993085171980671,
      "loss": 1.5288,
      "step": 16320
    },
    {
      "epoch": 4.363976483164083,
      "grad_norm": 1.5654804706573486,
      "learning_rate": 0.00011984857420235263,
      "loss": 1.4936,
      "step": 16330
    },
    {
      "epoch": 4.366648850881881,
      "grad_norm": 1.573370099067688,
      "learning_rate": 0.00011976628269477759,
      "loss": 1.4682,
      "step": 16340
    },
    {
      "epoch": 4.369321218599679,
      "grad_norm": 1.6399669647216797,
      "learning_rate": 0.00011968397725508416,
      "loss": 1.5217,
      "step": 16350
    },
    {
      "epoch": 4.371993586317477,
      "grad_norm": 1.6087729930877686,
      "learning_rate": 0.00011960165794128473,
      "loss": 1.462,
      "step": 16360
    },
    {
      "epoch": 4.374665954035275,
      "grad_norm": 1.5650618076324463,
      "learning_rate": 0.0001195193248114015,
      "loss": 1.4895,
      "step": 16370
    },
    {
      "epoch": 4.377338321753073,
      "grad_norm": 1.6232175827026367,
      "learning_rate": 0.00011943697792346629,
      "loss": 1.4741,
      "step": 16380
    },
    {
      "epoch": 4.380010689470871,
      "grad_norm": 1.6423583030700684,
      "learning_rate": 0.00011935461733552075,
      "loss": 1.5686,
      "step": 16390
    },
    {
      "epoch": 4.382683057188669,
      "grad_norm": 1.905605673789978,
      "learning_rate": 0.00011927224310561613,
      "loss": 1.5631,
      "step": 16400
    },
    {
      "epoch": 4.385355424906467,
      "grad_norm": 1.5372464656829834,
      "learning_rate": 0.00011918985529181333,
      "loss": 1.5095,
      "step": 16410
    },
    {
      "epoch": 4.388027792624265,
      "grad_norm": 1.6715388298034668,
      "learning_rate": 0.0001191074539521827,
      "loss": 1.4995,
      "step": 16420
    },
    {
      "epoch": 4.390700160342063,
      "grad_norm": 1.7411582469940186,
      "learning_rate": 0.00011902503914480428,
      "loss": 1.4601,
      "step": 16430
    },
    {
      "epoch": 4.393372528059861,
      "grad_norm": 1.6783335208892822,
      "learning_rate": 0.00011894261092776752,
      "loss": 1.6192,
      "step": 16440
    },
    {
      "epoch": 4.396044895777659,
      "grad_norm": 1.5689774751663208,
      "learning_rate": 0.00011886016935917136,
      "loss": 1.5356,
      "step": 16450
    },
    {
      "epoch": 4.398717263495457,
      "grad_norm": 1.7067049741744995,
      "learning_rate": 0.00011877771449712414,
      "loss": 1.5107,
      "step": 16460
    },
    {
      "epoch": 4.401389631213255,
      "grad_norm": 1.7084238529205322,
      "learning_rate": 0.00011869524639974349,
      "loss": 1.5365,
      "step": 16470
    },
    {
      "epoch": 4.404061998931053,
      "grad_norm": 1.608877420425415,
      "learning_rate": 0.0001186127651251565,
      "loss": 1.5091,
      "step": 16480
    },
    {
      "epoch": 4.406734366648851,
      "grad_norm": 1.7160688638687134,
      "learning_rate": 0.0001185302707314995,
      "loss": 1.5292,
      "step": 16490
    },
    {
      "epoch": 4.409406734366649,
      "grad_norm": 1.552913784980774,
      "learning_rate": 0.000118447763276918,
      "loss": 1.5217,
      "step": 16500
    },
    {
      "epoch": 4.412079102084447,
      "grad_norm": 1.9324225187301636,
      "learning_rate": 0.00011836524281956678,
      "loss": 1.5129,
      "step": 16510
    },
    {
      "epoch": 4.414751469802245,
      "grad_norm": 1.7439939975738525,
      "learning_rate": 0.00011828270941760981,
      "loss": 1.4675,
      "step": 16520
    },
    {
      "epoch": 4.417423837520043,
      "grad_norm": 1.7479952573776245,
      "learning_rate": 0.00011820016312922015,
      "loss": 1.5736,
      "step": 16530
    },
    {
      "epoch": 4.420096205237841,
      "grad_norm": 1.6934566497802734,
      "learning_rate": 0.0001181176040125799,
      "loss": 1.5506,
      "step": 16540
    },
    {
      "epoch": 4.422768572955639,
      "grad_norm": 1.6633288860321045,
      "learning_rate": 0.00011803503212588027,
      "loss": 1.5214,
      "step": 16550
    },
    {
      "epoch": 4.425440940673437,
      "grad_norm": 1.7800343036651611,
      "learning_rate": 0.00011795244752732145,
      "loss": 1.4409,
      "step": 16560
    },
    {
      "epoch": 4.428113308391235,
      "grad_norm": 1.7088680267333984,
      "learning_rate": 0.00011786985027511259,
      "loss": 1.4628,
      "step": 16570
    },
    {
      "epoch": 4.430785676109033,
      "grad_norm": 1.6736171245574951,
      "learning_rate": 0.00011778724042747172,
      "loss": 1.4649,
      "step": 16580
    },
    {
      "epoch": 4.433458043826831,
      "grad_norm": 1.578446865081787,
      "learning_rate": 0.00011770461804262578,
      "loss": 1.4735,
      "step": 16590
    },
    {
      "epoch": 4.436130411544629,
      "grad_norm": 1.6248250007629395,
      "learning_rate": 0.00011762198317881059,
      "loss": 1.5512,
      "step": 16600
    },
    {
      "epoch": 4.438802779262426,
      "grad_norm": 1.6464059352874756,
      "learning_rate": 0.00011753933589427073,
      "loss": 1.5397,
      "step": 16610
    },
    {
      "epoch": 4.441475146980224,
      "grad_norm": 1.5229616165161133,
      "learning_rate": 0.00011745667624725945,
      "loss": 1.5168,
      "step": 16620
    },
    {
      "epoch": 4.444147514698022,
      "grad_norm": 1.659442663192749,
      "learning_rate": 0.00011737400429603888,
      "loss": 1.5667,
      "step": 16630
    },
    {
      "epoch": 4.44681988241582,
      "grad_norm": 1.4989534616470337,
      "learning_rate": 0.00011729132009887967,
      "loss": 1.5182,
      "step": 16640
    },
    {
      "epoch": 4.449492250133618,
      "grad_norm": 1.6228270530700684,
      "learning_rate": 0.00011720862371406123,
      "loss": 1.5465,
      "step": 16650
    },
    {
      "epoch": 4.452164617851416,
      "grad_norm": 1.6722487211227417,
      "learning_rate": 0.00011712591519987141,
      "loss": 1.4773,
      "step": 16660
    },
    {
      "epoch": 4.454836985569214,
      "grad_norm": 1.5463364124298096,
      "learning_rate": 0.00011704319461460675,
      "loss": 1.5465,
      "step": 16670
    },
    {
      "epoch": 4.457509353287012,
      "grad_norm": 1.6173169612884521,
      "learning_rate": 0.00011696046201657222,
      "loss": 1.5153,
      "step": 16680
    },
    {
      "epoch": 4.46018172100481,
      "grad_norm": 1.6405445337295532,
      "learning_rate": 0.00011687771746408134,
      "loss": 1.5327,
      "step": 16690
    },
    {
      "epoch": 4.462854088722608,
      "grad_norm": 1.771985411643982,
      "learning_rate": 0.00011679496101545593,
      "loss": 1.5819,
      "step": 16700
    },
    {
      "epoch": 4.465526456440406,
      "grad_norm": 1.4997650384902954,
      "learning_rate": 0.00011671219272902629,
      "loss": 1.589,
      "step": 16710
    },
    {
      "epoch": 4.468198824158204,
      "grad_norm": 1.574514627456665,
      "learning_rate": 0.00011662941266313101,
      "loss": 1.5826,
      "step": 16720
    },
    {
      "epoch": 4.470871191876002,
      "grad_norm": 1.5952101945877075,
      "learning_rate": 0.00011654662087611707,
      "loss": 1.5426,
      "step": 16730
    },
    {
      "epoch": 4.4735435595938,
      "grad_norm": 1.6968765258789062,
      "learning_rate": 0.00011646381742633959,
      "loss": 1.5159,
      "step": 16740
    },
    {
      "epoch": 4.476215927311598,
      "grad_norm": 1.6814135313034058,
      "learning_rate": 0.00011638100237216197,
      "loss": 1.4628,
      "step": 16750
    },
    {
      "epoch": 4.478888295029396,
      "grad_norm": 1.6689420938491821,
      "learning_rate": 0.0001162981757719558,
      "loss": 1.4763,
      "step": 16760
    },
    {
      "epoch": 4.481560662747194,
      "grad_norm": 1.742651343345642,
      "learning_rate": 0.00011621533768410087,
      "loss": 1.4145,
      "step": 16770
    },
    {
      "epoch": 4.484233030464992,
      "grad_norm": 1.593358039855957,
      "learning_rate": 0.00011613248816698488,
      "loss": 1.5209,
      "step": 16780
    },
    {
      "epoch": 4.48690539818279,
      "grad_norm": 1.813662052154541,
      "learning_rate": 0.00011604962727900375,
      "loss": 1.492,
      "step": 16790
    },
    {
      "epoch": 4.489577765900588,
      "grad_norm": 1.4908579587936401,
      "learning_rate": 0.00011596675507856138,
      "loss": 1.5273,
      "step": 16800
    },
    {
      "epoch": 4.492250133618386,
      "grad_norm": 1.5751073360443115,
      "learning_rate": 0.0001158838716240696,
      "loss": 1.5151,
      "step": 16810
    },
    {
      "epoch": 4.494922501336184,
      "grad_norm": 1.8383218050003052,
      "learning_rate": 0.00011580097697394823,
      "loss": 1.5644,
      "step": 16820
    },
    {
      "epoch": 4.497594869053982,
      "grad_norm": 1.5725160837173462,
      "learning_rate": 0.00011571807118662492,
      "loss": 1.5439,
      "step": 16830
    },
    {
      "epoch": 4.50026723677178,
      "grad_norm": 1.8009822368621826,
      "learning_rate": 0.00011563515432053522,
      "loss": 1.4875,
      "step": 16840
    },
    {
      "epoch": 4.502939604489578,
      "grad_norm": 1.4893590211868286,
      "learning_rate": 0.00011555222643412249,
      "loss": 1.4788,
      "step": 16850
    },
    {
      "epoch": 4.505611972207376,
      "grad_norm": 1.7862300872802734,
      "learning_rate": 0.00011546928758583777,
      "loss": 1.4898,
      "step": 16860
    },
    {
      "epoch": 4.508284339925174,
      "grad_norm": 1.8359789848327637,
      "learning_rate": 0.00011538633783413997,
      "loss": 1.5299,
      "step": 16870
    },
    {
      "epoch": 4.510956707642972,
      "grad_norm": 1.6124062538146973,
      "learning_rate": 0.00011530337723749559,
      "loss": 1.5457,
      "step": 16880
    },
    {
      "epoch": 4.51362907536077,
      "grad_norm": 1.6466174125671387,
      "learning_rate": 0.00011522040585437879,
      "loss": 1.4771,
      "step": 16890
    },
    {
      "epoch": 4.516301443078568,
      "grad_norm": 1.9064375162124634,
      "learning_rate": 0.0001151374237432713,
      "loss": 1.5142,
      "step": 16900
    },
    {
      "epoch": 4.518973810796366,
      "grad_norm": 1.6905407905578613,
      "learning_rate": 0.00011505443096266249,
      "loss": 1.5282,
      "step": 16910
    },
    {
      "epoch": 4.521646178514164,
      "grad_norm": 1.6533970832824707,
      "learning_rate": 0.0001149714275710492,
      "loss": 1.5009,
      "step": 16920
    },
    {
      "epoch": 4.524318546231962,
      "grad_norm": 1.6736377477645874,
      "learning_rate": 0.00011488841362693576,
      "loss": 1.5616,
      "step": 16930
    },
    {
      "epoch": 4.52699091394976,
      "grad_norm": 1.7624212503433228,
      "learning_rate": 0.00011480538918883395,
      "loss": 1.5338,
      "step": 16940
    },
    {
      "epoch": 4.529663281667558,
      "grad_norm": 1.6372759342193604,
      "learning_rate": 0.00011472235431526288,
      "loss": 1.594,
      "step": 16950
    },
    {
      "epoch": 4.532335649385355,
      "grad_norm": 1.683350920677185,
      "learning_rate": 0.0001146393090647491,
      "loss": 1.534,
      "step": 16960
    },
    {
      "epoch": 4.535008017103153,
      "grad_norm": 1.6282875537872314,
      "learning_rate": 0.00011455625349582648,
      "loss": 1.5676,
      "step": 16970
    },
    {
      "epoch": 4.537680384820951,
      "grad_norm": 1.816609263420105,
      "learning_rate": 0.00011447318766703606,
      "loss": 1.4262,
      "step": 16980
    },
    {
      "epoch": 4.540352752538749,
      "grad_norm": 1.822721242904663,
      "learning_rate": 0.00011439011163692623,
      "loss": 1.5414,
      "step": 16990
    },
    {
      "epoch": 4.543025120256547,
      "grad_norm": 1.5583539009094238,
      "learning_rate": 0.00011430702546405246,
      "loss": 1.4702,
      "step": 17000
    },
    {
      "epoch": 4.545697487974345,
      "grad_norm": 1.673153042793274,
      "learning_rate": 0.00011422392920697747,
      "loss": 1.5011,
      "step": 17010
    },
    {
      "epoch": 4.548369855692143,
      "grad_norm": 1.710513949394226,
      "learning_rate": 0.00011414082292427102,
      "loss": 1.5034,
      "step": 17020
    },
    {
      "epoch": 4.551042223409941,
      "grad_norm": 1.7203640937805176,
      "learning_rate": 0.00011405770667450996,
      "loss": 1.5271,
      "step": 17030
    },
    {
      "epoch": 4.553714591127739,
      "grad_norm": 1.5803667306900024,
      "learning_rate": 0.00011397458051627816,
      "loss": 1.5608,
      "step": 17040
    },
    {
      "epoch": 4.556386958845537,
      "grad_norm": 1.6926594972610474,
      "learning_rate": 0.00011389144450816653,
      "loss": 1.5609,
      "step": 17050
    },
    {
      "epoch": 4.559059326563335,
      "grad_norm": 1.5539196729660034,
      "learning_rate": 0.00011380829870877281,
      "loss": 1.5476,
      "step": 17060
    },
    {
      "epoch": 4.561731694281133,
      "grad_norm": 1.6851810216903687,
      "learning_rate": 0.00011372514317670171,
      "loss": 1.4733,
      "step": 17070
    },
    {
      "epoch": 4.564404061998931,
      "grad_norm": 1.6946110725402832,
      "learning_rate": 0.00011364197797056482,
      "loss": 1.5316,
      "step": 17080
    },
    {
      "epoch": 4.567076429716729,
      "grad_norm": 1.5859239101409912,
      "learning_rate": 0.00011355880314898053,
      "loss": 1.5671,
      "step": 17090
    },
    {
      "epoch": 4.569748797434527,
      "grad_norm": 1.6291956901550293,
      "learning_rate": 0.00011347561877057396,
      "loss": 1.5179,
      "step": 17100
    },
    {
      "epoch": 4.572421165152325,
      "grad_norm": 1.643264651298523,
      "learning_rate": 0.00011339242489397699,
      "loss": 1.5202,
      "step": 17110
    },
    {
      "epoch": 4.575093532870123,
      "grad_norm": 1.6885887384414673,
      "learning_rate": 0.00011330922157782828,
      "loss": 1.505,
      "step": 17120
    },
    {
      "epoch": 4.577765900587921,
      "grad_norm": 1.5524781942367554,
      "learning_rate": 0.00011322600888077303,
      "loss": 1.5031,
      "step": 17130
    },
    {
      "epoch": 4.580438268305719,
      "grad_norm": 1.680329442024231,
      "learning_rate": 0.0001131427868614631,
      "loss": 1.4235,
      "step": 17140
    },
    {
      "epoch": 4.583110636023517,
      "grad_norm": 1.6419285535812378,
      "learning_rate": 0.00011305955557855689,
      "loss": 1.5403,
      "step": 17150
    },
    {
      "epoch": 4.585783003741315,
      "grad_norm": 1.6222217082977295,
      "learning_rate": 0.00011297631509071942,
      "loss": 1.5205,
      "step": 17160
    },
    {
      "epoch": 4.588455371459113,
      "grad_norm": 1.6735419034957886,
      "learning_rate": 0.00011289306545662206,
      "loss": 1.6031,
      "step": 17170
    },
    {
      "epoch": 4.591127739176911,
      "grad_norm": 1.6568113565444946,
      "learning_rate": 0.00011280980673494278,
      "loss": 1.5723,
      "step": 17180
    },
    {
      "epoch": 4.593800106894709,
      "grad_norm": 1.6171001195907593,
      "learning_rate": 0.00011272653898436578,
      "loss": 1.4823,
      "step": 17190
    },
    {
      "epoch": 4.596472474612507,
      "grad_norm": 1.6146831512451172,
      "learning_rate": 0.00011264326226358177,
      "loss": 1.5652,
      "step": 17200
    },
    {
      "epoch": 4.599144842330305,
      "grad_norm": 1.5950102806091309,
      "learning_rate": 0.00011255997663128774,
      "loss": 1.465,
      "step": 17210
    },
    {
      "epoch": 4.601817210048103,
      "grad_norm": 1.6530253887176514,
      "learning_rate": 0.00011247668214618693,
      "loss": 1.4929,
      "step": 17220
    },
    {
      "epoch": 4.604489577765901,
      "grad_norm": 1.644036889076233,
      "learning_rate": 0.00011239337886698887,
      "loss": 1.4702,
      "step": 17230
    },
    {
      "epoch": 4.607161945483698,
      "grad_norm": 1.6774314641952515,
      "learning_rate": 0.00011231006685240923,
      "loss": 1.5021,
      "step": 17240
    },
    {
      "epoch": 4.609834313201496,
      "grad_norm": 1.683488130569458,
      "learning_rate": 0.0001122267461611699,
      "loss": 1.5395,
      "step": 17250
    },
    {
      "epoch": 4.612506680919294,
      "grad_norm": 1.850111722946167,
      "learning_rate": 0.00011214341685199883,
      "loss": 1.5249,
      "step": 17260
    },
    {
      "epoch": 4.615179048637092,
      "grad_norm": 1.778827428817749,
      "learning_rate": 0.00011206007898363007,
      "loss": 1.4886,
      "step": 17270
    },
    {
      "epoch": 4.61785141635489,
      "grad_norm": 1.686880350112915,
      "learning_rate": 0.0001119767326148037,
      "loss": 1.4967,
      "step": 17280
    },
    {
      "epoch": 4.620523784072688,
      "grad_norm": 1.7504363059997559,
      "learning_rate": 0.00011189337780426583,
      "loss": 1.5259,
      "step": 17290
    },
    {
      "epoch": 4.623196151790486,
      "grad_norm": 1.7304437160491943,
      "learning_rate": 0.00011181001461076843,
      "loss": 1.6043,
      "step": 17300
    },
    {
      "epoch": 4.625868519508284,
      "grad_norm": 1.5700219869613647,
      "learning_rate": 0.00011172664309306943,
      "loss": 1.4881,
      "step": 17310
    },
    {
      "epoch": 4.628540887226082,
      "grad_norm": 1.7662694454193115,
      "learning_rate": 0.00011164326330993268,
      "loss": 1.5357,
      "step": 17320
    },
    {
      "epoch": 4.63121325494388,
      "grad_norm": 1.661639928817749,
      "learning_rate": 0.00011155987532012779,
      "loss": 1.5531,
      "step": 17330
    },
    {
      "epoch": 4.633885622661678,
      "grad_norm": 1.7782138586044312,
      "learning_rate": 0.00011147647918243013,
      "loss": 1.5453,
      "step": 17340
    },
    {
      "epoch": 4.636557990379476,
      "grad_norm": 1.8270082473754883,
      "learning_rate": 0.0001113930749556209,
      "loss": 1.5348,
      "step": 17350
    },
    {
      "epoch": 4.639230358097274,
      "grad_norm": 1.7289249897003174,
      "learning_rate": 0.00011130966269848691,
      "loss": 1.5346,
      "step": 17360
    },
    {
      "epoch": 4.641902725815072,
      "grad_norm": 1.5998317003250122,
      "learning_rate": 0.00011122624246982072,
      "loss": 1.522,
      "step": 17370
    },
    {
      "epoch": 4.64457509353287,
      "grad_norm": 1.8454827070236206,
      "learning_rate": 0.0001111428143284204,
      "loss": 1.5138,
      "step": 17380
    },
    {
      "epoch": 4.647247461250668,
      "grad_norm": 1.699512243270874,
      "learning_rate": 0.00011105937833308968,
      "loss": 1.5008,
      "step": 17390
    },
    {
      "epoch": 4.649919828968466,
      "grad_norm": 1.6202223300933838,
      "learning_rate": 0.00011097593454263784,
      "loss": 1.4387,
      "step": 17400
    },
    {
      "epoch": 4.652592196686264,
      "grad_norm": 1.6622673273086548,
      "learning_rate": 0.00011089248301587962,
      "loss": 1.4613,
      "step": 17410
    },
    {
      "epoch": 4.655264564404062,
      "grad_norm": 1.6387486457824707,
      "learning_rate": 0.00011080902381163519,
      "loss": 1.5228,
      "step": 17420
    },
    {
      "epoch": 4.65793693212186,
      "grad_norm": 1.6574199199676514,
      "learning_rate": 0.00011072555698873013,
      "loss": 1.5268,
      "step": 17430
    },
    {
      "epoch": 4.660609299839658,
      "grad_norm": 1.704759955406189,
      "learning_rate": 0.00011064208260599542,
      "loss": 1.5739,
      "step": 17440
    },
    {
      "epoch": 4.663281667557456,
      "grad_norm": 2.1619670391082764,
      "learning_rate": 0.00011055860072226743,
      "loss": 1.5523,
      "step": 17450
    },
    {
      "epoch": 4.665954035275254,
      "grad_norm": 1.734423041343689,
      "learning_rate": 0.00011047511139638767,
      "loss": 1.5457,
      "step": 17460
    },
    {
      "epoch": 4.668626402993052,
      "grad_norm": 1.6960095167160034,
      "learning_rate": 0.000110391614687203,
      "loss": 1.5086,
      "step": 17470
    },
    {
      "epoch": 4.67129877071085,
      "grad_norm": 1.6343101263046265,
      "learning_rate": 0.00011030811065356548,
      "loss": 1.5039,
      "step": 17480
    },
    {
      "epoch": 4.673971138428648,
      "grad_norm": 1.7738159894943237,
      "learning_rate": 0.00011022459935433231,
      "loss": 1.5295,
      "step": 17490
    },
    {
      "epoch": 4.676643506146446,
      "grad_norm": 1.6277084350585938,
      "learning_rate": 0.00011014108084836576,
      "loss": 1.5171,
      "step": 17500
    },
    {
      "epoch": 4.679315873864244,
      "grad_norm": 1.579516887664795,
      "learning_rate": 0.00011005755519453328,
      "loss": 1.5303,
      "step": 17510
    },
    {
      "epoch": 4.681988241582042,
      "grad_norm": 1.6907621622085571,
      "learning_rate": 0.00010997402245170734,
      "loss": 1.4913,
      "step": 17520
    },
    {
      "epoch": 4.68466060929984,
      "grad_norm": 1.5866539478302002,
      "learning_rate": 0.00010989048267876529,
      "loss": 1.5551,
      "step": 17530
    },
    {
      "epoch": 4.687332977017638,
      "grad_norm": 1.681728482246399,
      "learning_rate": 0.00010980693593458962,
      "loss": 1.502,
      "step": 17540
    },
    {
      "epoch": 4.690005344735436,
      "grad_norm": 1.8035271167755127,
      "learning_rate": 0.00010972338227806754,
      "loss": 1.4714,
      "step": 17550
    },
    {
      "epoch": 4.692677712453234,
      "grad_norm": 1.6702238321304321,
      "learning_rate": 0.00010963982176809126,
      "loss": 1.5588,
      "step": 17560
    },
    {
      "epoch": 4.695350080171032,
      "grad_norm": 1.6105695962905884,
      "learning_rate": 0.00010955625446355784,
      "loss": 1.6007,
      "step": 17570
    },
    {
      "epoch": 4.69802244788883,
      "grad_norm": 1.6845314502716064,
      "learning_rate": 0.000109472680423369,
      "loss": 1.5524,
      "step": 17580
    },
    {
      "epoch": 4.700694815606628,
      "grad_norm": 1.7203104496002197,
      "learning_rate": 0.00010938909970643134,
      "loss": 1.5257,
      "step": 17590
    },
    {
      "epoch": 4.703367183324426,
      "grad_norm": 1.7428655624389648,
      "learning_rate": 0.00010930551237165609,
      "loss": 1.5282,
      "step": 17600
    },
    {
      "epoch": 4.706039551042223,
      "grad_norm": 1.62808358669281,
      "learning_rate": 0.00010922191847795919,
      "loss": 1.5423,
      "step": 17610
    },
    {
      "epoch": 4.708711918760021,
      "grad_norm": 1.7725077867507935,
      "learning_rate": 0.00010913831808426113,
      "loss": 1.4918,
      "step": 17620
    },
    {
      "epoch": 4.711384286477819,
      "grad_norm": 1.71577787399292,
      "learning_rate": 0.00010905471124948707,
      "loss": 1.4809,
      "step": 17630
    },
    {
      "epoch": 4.714056654195617,
      "grad_norm": 1.6842659711837769,
      "learning_rate": 0.00010897109803256667,
      "loss": 1.5723,
      "step": 17640
    },
    {
      "epoch": 4.716729021913415,
      "grad_norm": 1.6058441400527954,
      "learning_rate": 0.00010888747849243406,
      "loss": 1.4779,
      "step": 17650
    },
    {
      "epoch": 4.719401389631213,
      "grad_norm": 1.6452794075012207,
      "learning_rate": 0.00010880385268802789,
      "loss": 1.5023,
      "step": 17660
    },
    {
      "epoch": 4.722073757349011,
      "grad_norm": 1.5528175830841064,
      "learning_rate": 0.00010872022067829115,
      "loss": 1.5177,
      "step": 17670
    },
    {
      "epoch": 4.724746125066809,
      "grad_norm": 1.8851854801177979,
      "learning_rate": 0.00010863658252217125,
      "loss": 1.5116,
      "step": 17680
    },
    {
      "epoch": 4.727418492784607,
      "grad_norm": 1.567244052886963,
      "learning_rate": 0.00010855293827862002,
      "loss": 1.5115,
      "step": 17690
    },
    {
      "epoch": 4.730090860502405,
      "grad_norm": 1.7209495306015015,
      "learning_rate": 0.00010846928800659333,
      "loss": 1.5098,
      "step": 17700
    },
    {
      "epoch": 4.732763228220203,
      "grad_norm": 1.6979482173919678,
      "learning_rate": 0.00010838563176505158,
      "loss": 1.5321,
      "step": 17710
    },
    {
      "epoch": 4.735435595938001,
      "grad_norm": 1.5735869407653809,
      "learning_rate": 0.0001083019696129592,
      "loss": 1.5287,
      "step": 17720
    },
    {
      "epoch": 4.738107963655799,
      "grad_norm": 1.691196084022522,
      "learning_rate": 0.00010821830160928486,
      "loss": 1.5122,
      "step": 17730
    },
    {
      "epoch": 4.740780331373597,
      "grad_norm": 1.6603938341140747,
      "learning_rate": 0.00010813462781300133,
      "loss": 1.4964,
      "step": 17740
    },
    {
      "epoch": 4.743452699091395,
      "grad_norm": 1.6050552129745483,
      "learning_rate": 0.00010805094828308544,
      "loss": 1.5483,
      "step": 17750
    },
    {
      "epoch": 4.746125066809193,
      "grad_norm": 1.6820952892303467,
      "learning_rate": 0.0001079672630785181,
      "loss": 1.5085,
      "step": 17760
    },
    {
      "epoch": 4.748797434526991,
      "grad_norm": 1.7059487104415894,
      "learning_rate": 0.00010788357225828427,
      "loss": 1.6139,
      "step": 17770
    },
    {
      "epoch": 4.751469802244789,
      "grad_norm": 1.7276294231414795,
      "learning_rate": 0.00010779987588137272,
      "loss": 1.4786,
      "step": 17780
    },
    {
      "epoch": 4.754142169962587,
      "grad_norm": 1.722737193107605,
      "learning_rate": 0.00010771617400677624,
      "loss": 1.5177,
      "step": 17790
    },
    {
      "epoch": 4.756814537680385,
      "grad_norm": 1.7700971364974976,
      "learning_rate": 0.00010763246669349154,
      "loss": 1.4759,
      "step": 17800
    },
    {
      "epoch": 4.759486905398183,
      "grad_norm": 1.8075792789459229,
      "learning_rate": 0.00010754875400051907,
      "loss": 1.4808,
      "step": 17810
    },
    {
      "epoch": 4.762159273115981,
      "grad_norm": 1.7821334600448608,
      "learning_rate": 0.00010746503598686305,
      "loss": 1.5287,
      "step": 17820
    },
    {
      "epoch": 4.764831640833779,
      "grad_norm": 1.6413185596466064,
      "learning_rate": 0.00010738131271153156,
      "loss": 1.4688,
      "step": 17830
    },
    {
      "epoch": 4.767504008551577,
      "grad_norm": 1.663130760192871,
      "learning_rate": 0.00010729758423353633,
      "loss": 1.5225,
      "step": 17840
    },
    {
      "epoch": 4.770176376269375,
      "grad_norm": 1.6460022926330566,
      "learning_rate": 0.00010721385061189275,
      "loss": 1.4966,
      "step": 17850
    },
    {
      "epoch": 4.772848743987173,
      "grad_norm": 1.723634123802185,
      "learning_rate": 0.00010713011190561983,
      "loss": 1.4607,
      "step": 17860
    },
    {
      "epoch": 4.775521111704971,
      "grad_norm": 1.6349022388458252,
      "learning_rate": 0.00010704636817374018,
      "loss": 1.4627,
      "step": 17870
    },
    {
      "epoch": 4.778193479422769,
      "grad_norm": 1.7217038869857788,
      "learning_rate": 0.00010696261947527999,
      "loss": 1.5699,
      "step": 17880
    },
    {
      "epoch": 4.780865847140566,
      "grad_norm": 1.7527496814727783,
      "learning_rate": 0.00010687886586926889,
      "loss": 1.5585,
      "step": 17890
    },
    {
      "epoch": 4.783538214858364,
      "grad_norm": 1.7060827016830444,
      "learning_rate": 0.00010679510741473997,
      "loss": 1.524,
      "step": 17900
    },
    {
      "epoch": 4.786210582576162,
      "grad_norm": 1.7142846584320068,
      "learning_rate": 0.00010671134417072979,
      "loss": 1.5867,
      "step": 17910
    },
    {
      "epoch": 4.78888295029396,
      "grad_norm": 1.581526517868042,
      "learning_rate": 0.00010662757619627821,
      "loss": 1.5185,
      "step": 17920
    },
    {
      "epoch": 4.791555318011758,
      "grad_norm": 1.7608598470687866,
      "learning_rate": 0.00010654380355042852,
      "loss": 1.5467,
      "step": 17930
    },
    {
      "epoch": 4.794227685729556,
      "grad_norm": 1.634638786315918,
      "learning_rate": 0.0001064600262922272,
      "loss": 1.5749,
      "step": 17940
    },
    {
      "epoch": 4.796900053447354,
      "grad_norm": 1.8929290771484375,
      "learning_rate": 0.00010637624448072408,
      "loss": 1.5064,
      "step": 17950
    },
    {
      "epoch": 4.799572421165152,
      "grad_norm": 1.6078649759292603,
      "learning_rate": 0.00010629245817497212,
      "loss": 1.5267,
      "step": 17960
    },
    {
      "epoch": 4.80224478888295,
      "grad_norm": 1.4439537525177002,
      "learning_rate": 0.00010620866743402749,
      "loss": 1.5473,
      "step": 17970
    },
    {
      "epoch": 4.804917156600748,
      "grad_norm": 1.642608642578125,
      "learning_rate": 0.00010612487231694946,
      "loss": 1.5439,
      "step": 17980
    },
    {
      "epoch": 4.807589524318546,
      "grad_norm": 1.7744070291519165,
      "learning_rate": 0.0001060410728828004,
      "loss": 1.4915,
      "step": 17990
    },
    {
      "epoch": 4.810261892036344,
      "grad_norm": 1.6814850568771362,
      "learning_rate": 0.00010595726919064574,
      "loss": 1.4888,
      "step": 18000
    },
    {
      "epoch": 4.812934259754142,
      "grad_norm": 1.6025012731552124,
      "learning_rate": 0.00010587346129955393,
      "loss": 1.5233,
      "step": 18010
    },
    {
      "epoch": 4.81560662747194,
      "grad_norm": 1.5257656574249268,
      "learning_rate": 0.00010578964926859627,
      "loss": 1.5285,
      "step": 18020
    },
    {
      "epoch": 4.818278995189738,
      "grad_norm": 1.627409815788269,
      "learning_rate": 0.00010570583315684708,
      "loss": 1.4537,
      "step": 18030
    },
    {
      "epoch": 4.820951362907536,
      "grad_norm": 1.6643608808517456,
      "learning_rate": 0.0001056220130233835,
      "loss": 1.6133,
      "step": 18040
    },
    {
      "epoch": 4.823623730625334,
      "grad_norm": 1.6544005870819092,
      "learning_rate": 0.00010553818892728562,
      "loss": 1.4923,
      "step": 18050
    },
    {
      "epoch": 4.826296098343132,
      "grad_norm": 1.677706003189087,
      "learning_rate": 0.00010545436092763614,
      "loss": 1.4986,
      "step": 18060
    },
    {
      "epoch": 4.82896846606093,
      "grad_norm": 1.5974152088165283,
      "learning_rate": 0.00010537052908352065,
      "loss": 1.5083,
      "step": 18070
    },
    {
      "epoch": 4.831640833778728,
      "grad_norm": 1.6557848453521729,
      "learning_rate": 0.00010528669345402742,
      "loss": 1.5287,
      "step": 18080
    },
    {
      "epoch": 4.834313201496526,
      "grad_norm": 1.7581212520599365,
      "learning_rate": 0.00010520285409824733,
      "loss": 1.5609,
      "step": 18090
    },
    {
      "epoch": 4.836985569214324,
      "grad_norm": 1.649861216545105,
      "learning_rate": 0.00010511901107527398,
      "loss": 1.5856,
      "step": 18100
    },
    {
      "epoch": 4.839657936932122,
      "grad_norm": 1.6427505016326904,
      "learning_rate": 0.00010503516444420345,
      "loss": 1.5213,
      "step": 18110
    },
    {
      "epoch": 4.84233030464992,
      "grad_norm": 1.7283744812011719,
      "learning_rate": 0.00010495131426413444,
      "loss": 1.5068,
      "step": 18120
    },
    {
      "epoch": 4.845002672367718,
      "grad_norm": 1.759529948234558,
      "learning_rate": 0.00010486746059416816,
      "loss": 1.4968,
      "step": 18130
    },
    {
      "epoch": 4.847675040085516,
      "grad_norm": 1.6878550052642822,
      "learning_rate": 0.00010478360349340823,
      "loss": 1.5685,
      "step": 18140
    },
    {
      "epoch": 4.850347407803314,
      "grad_norm": 1.6848292350769043,
      "learning_rate": 0.00010469974302096068,
      "loss": 1.5831,
      "step": 18150
    },
    {
      "epoch": 4.853019775521112,
      "grad_norm": 1.613651156425476,
      "learning_rate": 0.00010461587923593399,
      "loss": 1.5108,
      "step": 18160
    },
    {
      "epoch": 4.85569214323891,
      "grad_norm": 1.599210500717163,
      "learning_rate": 0.00010453201219743894,
      "loss": 1.4874,
      "step": 18170
    },
    {
      "epoch": 4.858364510956708,
      "grad_norm": 1.4573466777801514,
      "learning_rate": 0.0001044481419645885,
      "loss": 1.4947,
      "step": 18180
    },
    {
      "epoch": 4.861036878674506,
      "grad_norm": 1.7144567966461182,
      "learning_rate": 0.00010436426859649808,
      "loss": 1.5119,
      "step": 18190
    },
    {
      "epoch": 4.863709246392304,
      "grad_norm": 1.716574788093567,
      "learning_rate": 0.00010428039215228516,
      "loss": 1.5,
      "step": 18200
    },
    {
      "epoch": 4.866381614110102,
      "grad_norm": 1.664913535118103,
      "learning_rate": 0.00010419651269106947,
      "loss": 1.4667,
      "step": 18210
    },
    {
      "epoch": 4.8690539818279,
      "grad_norm": 1.7342499494552612,
      "learning_rate": 0.0001041126302719728,
      "loss": 1.5026,
      "step": 18220
    },
    {
      "epoch": 4.871726349545698,
      "grad_norm": 1.7548737525939941,
      "learning_rate": 0.00010402874495411902,
      "loss": 1.5029,
      "step": 18230
    },
    {
      "epoch": 4.874398717263496,
      "grad_norm": 1.5733733177185059,
      "learning_rate": 0.00010394485679663412,
      "loss": 1.5932,
      "step": 18240
    },
    {
      "epoch": 4.877071084981293,
      "grad_norm": 1.4801501035690308,
      "learning_rate": 0.00010386096585864609,
      "loss": 1.5043,
      "step": 18250
    },
    {
      "epoch": 4.879743452699091,
      "grad_norm": 1.5585309267044067,
      "learning_rate": 0.00010377707219928474,
      "loss": 1.5304,
      "step": 18260
    },
    {
      "epoch": 4.882415820416889,
      "grad_norm": 1.6687904596328735,
      "learning_rate": 0.00010369317587768199,
      "loss": 1.5446,
      "step": 18270
    },
    {
      "epoch": 4.885088188134687,
      "grad_norm": 1.7294179201126099,
      "learning_rate": 0.00010360927695297146,
      "loss": 1.5587,
      "step": 18280
    },
    {
      "epoch": 4.887760555852485,
      "grad_norm": 1.885548710823059,
      "learning_rate": 0.00010352537548428876,
      "loss": 1.498,
      "step": 18290
    },
    {
      "epoch": 4.890432923570283,
      "grad_norm": 1.6851317882537842,
      "learning_rate": 0.00010344147153077115,
      "loss": 1.5331,
      "step": 18300
    },
    {
      "epoch": 4.893105291288081,
      "grad_norm": 1.45623779296875,
      "learning_rate": 0.00010335756515155775,
      "loss": 1.499,
      "step": 18310
    },
    {
      "epoch": 4.895777659005879,
      "grad_norm": 1.7638206481933594,
      "learning_rate": 0.00010327365640578937,
      "loss": 1.514,
      "step": 18320
    },
    {
      "epoch": 4.898450026723677,
      "grad_norm": 1.7740068435668945,
      "learning_rate": 0.00010318974535260844,
      "loss": 1.4728,
      "step": 18330
    },
    {
      "epoch": 4.901122394441475,
      "grad_norm": 1.7067680358886719,
      "learning_rate": 0.00010310583205115905,
      "loss": 1.5582,
      "step": 18340
    },
    {
      "epoch": 4.903794762159273,
      "grad_norm": 1.7819015979766846,
      "learning_rate": 0.00010302191656058686,
      "loss": 1.538,
      "step": 18350
    },
    {
      "epoch": 4.906467129877071,
      "grad_norm": 1.7383277416229248,
      "learning_rate": 0.00010293799894003912,
      "loss": 1.559,
      "step": 18360
    },
    {
      "epoch": 4.909139497594869,
      "grad_norm": 1.5966684818267822,
      "learning_rate": 0.00010285407924866454,
      "loss": 1.5746,
      "step": 18370
    },
    {
      "epoch": 4.911811865312667,
      "grad_norm": 1.6552889347076416,
      "learning_rate": 0.00010277015754561328,
      "loss": 1.539,
      "step": 18380
    },
    {
      "epoch": 4.914484233030465,
      "grad_norm": 1.4528568983078003,
      "learning_rate": 0.00010268623389003691,
      "loss": 1.5546,
      "step": 18390
    },
    {
      "epoch": 4.917156600748263,
      "grad_norm": 1.4890824556350708,
      "learning_rate": 0.00010260230834108841,
      "loss": 1.5143,
      "step": 18400
    },
    {
      "epoch": 4.919828968466061,
      "grad_norm": 1.7056405544281006,
      "learning_rate": 0.00010251838095792215,
      "loss": 1.4425,
      "step": 18410
    },
    {
      "epoch": 4.922501336183859,
      "grad_norm": 1.6442804336547852,
      "learning_rate": 0.00010243445179969362,
      "loss": 1.6035,
      "step": 18420
    },
    {
      "epoch": 4.925173703901657,
      "grad_norm": 1.7363308668136597,
      "learning_rate": 0.00010235052092555972,
      "loss": 1.55,
      "step": 18430
    },
    {
      "epoch": 4.927846071619455,
      "grad_norm": 1.6665538549423218,
      "learning_rate": 0.00010226658839467855,
      "loss": 1.5404,
      "step": 18440
    },
    {
      "epoch": 4.930518439337253,
      "grad_norm": 1.6387377977371216,
      "learning_rate": 0.00010218265426620925,
      "loss": 1.497,
      "step": 18450
    },
    {
      "epoch": 4.933190807055051,
      "grad_norm": 1.5948083400726318,
      "learning_rate": 0.00010209871859931224,
      "loss": 1.4706,
      "step": 18460
    },
    {
      "epoch": 4.935863174772849,
      "grad_norm": 1.6902433633804321,
      "learning_rate": 0.00010201478145314888,
      "loss": 1.5247,
      "step": 18470
    },
    {
      "epoch": 4.938535542490647,
      "grad_norm": 1.7885191440582275,
      "learning_rate": 0.00010193084288688173,
      "loss": 1.5034,
      "step": 18480
    },
    {
      "epoch": 4.941207910208445,
      "grad_norm": 1.6652768850326538,
      "learning_rate": 0.00010184690295967422,
      "loss": 1.5281,
      "step": 18490
    },
    {
      "epoch": 4.943880277926243,
      "grad_norm": 1.7208900451660156,
      "learning_rate": 0.00010176296173069079,
      "loss": 1.4823,
      "step": 18500
    },
    {
      "epoch": 4.946552645644041,
      "grad_norm": 1.4993844032287598,
      "learning_rate": 0.00010167901925909681,
      "loss": 1.5828,
      "step": 18510
    },
    {
      "epoch": 4.949225013361839,
      "grad_norm": 1.6090344190597534,
      "learning_rate": 0.0001015950756040585,
      "loss": 1.5345,
      "step": 18520
    },
    {
      "epoch": 4.951897381079636,
      "grad_norm": 1.9685152769088745,
      "learning_rate": 0.00010151113082474295,
      "loss": 1.5182,
      "step": 18530
    },
    {
      "epoch": 4.954569748797434,
      "grad_norm": 1.5380126237869263,
      "learning_rate": 0.00010142718498031799,
      "loss": 1.5448,
      "step": 18540
    },
    {
      "epoch": 4.957242116515232,
      "grad_norm": 1.6140483617782593,
      "learning_rate": 0.00010134323812995221,
      "loss": 1.5863,
      "step": 18550
    },
    {
      "epoch": 4.95991448423303,
      "grad_norm": 1.865526556968689,
      "learning_rate": 0.00010125929033281499,
      "loss": 1.5463,
      "step": 18560
    },
    {
      "epoch": 4.962586851950828,
      "grad_norm": 1.67505943775177,
      "learning_rate": 0.00010117534164807628,
      "loss": 1.4375,
      "step": 18570
    },
    {
      "epoch": 4.965259219668626,
      "grad_norm": 1.5862981081008911,
      "learning_rate": 0.00010109139213490666,
      "loss": 1.5158,
      "step": 18580
    },
    {
      "epoch": 4.967931587386424,
      "grad_norm": 1.5448644161224365,
      "learning_rate": 0.00010100744185247737,
      "loss": 1.5918,
      "step": 18590
    },
    {
      "epoch": 4.970603955104222,
      "grad_norm": 1.7706352472305298,
      "learning_rate": 0.00010092349085996011,
      "loss": 1.4799,
      "step": 18600
    },
    {
      "epoch": 4.97327632282202,
      "grad_norm": 1.5707390308380127,
      "learning_rate": 0.00010083953921652718,
      "loss": 1.5097,
      "step": 18610
    },
    {
      "epoch": 4.975948690539818,
      "grad_norm": 1.6571247577667236,
      "learning_rate": 0.00010075558698135121,
      "loss": 1.5722,
      "step": 18620
    },
    {
      "epoch": 4.978621058257616,
      "grad_norm": 1.6878763437271118,
      "learning_rate": 0.00010067163421360535,
      "loss": 1.4935,
      "step": 18630
    },
    {
      "epoch": 4.981293425975414,
      "grad_norm": 1.627952218055725,
      "learning_rate": 0.00010058768097246307,
      "loss": 1.4999,
      "step": 18640
    },
    {
      "epoch": 4.983965793693212,
      "grad_norm": 1.6279401779174805,
      "learning_rate": 0.0001005037273170982,
      "loss": 1.496,
      "step": 18650
    },
    {
      "epoch": 4.98663816141101,
      "grad_norm": 1.705061674118042,
      "learning_rate": 0.00010041977330668484,
      "loss": 1.438,
      "step": 18660
    },
    {
      "epoch": 4.989310529128808,
      "grad_norm": 1.5573621988296509,
      "learning_rate": 0.00010033581900039735,
      "loss": 1.4353,
      "step": 18670
    },
    {
      "epoch": 4.991982896846606,
      "grad_norm": 1.6259018182754517,
      "learning_rate": 0.00010025186445741033,
      "loss": 1.5293,
      "step": 18680
    },
    {
      "epoch": 4.994655264564404,
      "grad_norm": 1.663500189781189,
      "learning_rate": 0.00010016790973689853,
      "loss": 1.5599,
      "step": 18690
    },
    {
      "epoch": 4.997327632282202,
      "grad_norm": 1.6640194654464722,
      "learning_rate": 0.00010008395489803677,
      "loss": 1.5134,
      "step": 18700
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.5554745197296143,
      "learning_rate": 0.0001,
      "loss": 1.5918,
      "step": 18710
    },
    {
      "epoch": 5.002672367717798,
      "grad_norm": 1.7414125204086304,
      "learning_rate": 9.991604510196329e-05,
      "loss": 1.3742,
      "step": 18720
    },
    {
      "epoch": 5.005344735435596,
      "grad_norm": 1.72783625125885,
      "learning_rate": 9.983209026310149e-05,
      "loss": 1.3763,
      "step": 18730
    },
    {
      "epoch": 5.008017103153394,
      "grad_norm": 2.173156261444092,
      "learning_rate": 9.974813554258969e-05,
      "loss": 1.4874,
      "step": 18740
    },
    {
      "epoch": 5.010689470871192,
      "grad_norm": 1.8801006078720093,
      "learning_rate": 9.966418099960263e-05,
      "loss": 1.3718,
      "step": 18750
    },
    {
      "epoch": 5.01336183858899,
      "grad_norm": 1.846749186515808,
      "learning_rate": 9.95802266933152e-05,
      "loss": 1.4114,
      "step": 18760
    },
    {
      "epoch": 5.016034206306788,
      "grad_norm": 1.8073856830596924,
      "learning_rate": 9.949627268290185e-05,
      "loss": 1.3464,
      "step": 18770
    },
    {
      "epoch": 5.018706574024586,
      "grad_norm": 1.755150318145752,
      "learning_rate": 9.941231902753696e-05,
      "loss": 1.3483,
      "step": 18780
    },
    {
      "epoch": 5.021378941742384,
      "grad_norm": 1.779171347618103,
      "learning_rate": 9.932836578639466e-05,
      "loss": 1.3527,
      "step": 18790
    },
    {
      "epoch": 5.024051309460182,
      "grad_norm": 1.8565480709075928,
      "learning_rate": 9.924441301864881e-05,
      "loss": 1.3768,
      "step": 18800
    },
    {
      "epoch": 5.02672367717798,
      "grad_norm": 1.936449408531189,
      "learning_rate": 9.916046078347283e-05,
      "loss": 1.4007,
      "step": 18810
    },
    {
      "epoch": 5.029396044895778,
      "grad_norm": 1.9122281074523926,
      "learning_rate": 9.90765091400399e-05,
      "loss": 1.3241,
      "step": 18820
    },
    {
      "epoch": 5.032068412613576,
      "grad_norm": 1.670769214630127,
      "learning_rate": 9.899255814752263e-05,
      "loss": 1.3464,
      "step": 18830
    },
    {
      "epoch": 5.034740780331374,
      "grad_norm": 1.7659333944320679,
      "learning_rate": 9.890860786509337e-05,
      "loss": 1.3799,
      "step": 18840
    },
    {
      "epoch": 5.037413148049172,
      "grad_norm": 1.6679108142852783,
      "learning_rate": 9.882465835192377e-05,
      "loss": 1.365,
      "step": 18850
    },
    {
      "epoch": 5.04008551576697,
      "grad_norm": 1.6594696044921875,
      "learning_rate": 9.874070966718502e-05,
      "loss": 1.3707,
      "step": 18860
    },
    {
      "epoch": 5.042757883484768,
      "grad_norm": 1.8254822492599487,
      "learning_rate": 9.86567618700478e-05,
      "loss": 1.3905,
      "step": 18870
    },
    {
      "epoch": 5.045430251202565,
      "grad_norm": 1.815039038658142,
      "learning_rate": 9.857281501968206e-05,
      "loss": 1.4385,
      "step": 18880
    },
    {
      "epoch": 5.048102618920363,
      "grad_norm": 1.7348051071166992,
      "learning_rate": 9.848886917525707e-05,
      "loss": 1.3823,
      "step": 18890
    },
    {
      "epoch": 5.050774986638161,
      "grad_norm": 1.7569689750671387,
      "learning_rate": 9.840492439594152e-05,
      "loss": 1.3298,
      "step": 18900
    },
    {
      "epoch": 5.053447354355959,
      "grad_norm": 1.709765911102295,
      "learning_rate": 9.83209807409032e-05,
      "loss": 1.3331,
      "step": 18910
    },
    {
      "epoch": 5.056119722073757,
      "grad_norm": 1.85061514377594,
      "learning_rate": 9.823703826930922e-05,
      "loss": 1.4231,
      "step": 18920
    },
    {
      "epoch": 5.058792089791555,
      "grad_norm": 1.7782377004623413,
      "learning_rate": 9.815309704032583e-05,
      "loss": 1.3423,
      "step": 18930
    },
    {
      "epoch": 5.061464457509353,
      "grad_norm": 1.7693212032318115,
      "learning_rate": 9.806915711311828e-05,
      "loss": 1.3175,
      "step": 18940
    },
    {
      "epoch": 5.064136825227151,
      "grad_norm": 1.6982011795043945,
      "learning_rate": 9.798521854685113e-05,
      "loss": 1.3129,
      "step": 18950
    },
    {
      "epoch": 5.066809192944949,
      "grad_norm": 1.6515928506851196,
      "learning_rate": 9.790128140068782e-05,
      "loss": 1.3634,
      "step": 18960
    },
    {
      "epoch": 5.069481560662747,
      "grad_norm": 1.8829503059387207,
      "learning_rate": 9.781734573379077e-05,
      "loss": 1.3536,
      "step": 18970
    },
    {
      "epoch": 5.072153928380545,
      "grad_norm": 1.8078536987304688,
      "learning_rate": 9.773341160532148e-05,
      "loss": 1.2813,
      "step": 18980
    },
    {
      "epoch": 5.074826296098343,
      "grad_norm": 1.822912335395813,
      "learning_rate": 9.764947907444027e-05,
      "loss": 1.3712,
      "step": 18990
    },
    {
      "epoch": 5.077498663816141,
      "grad_norm": 1.8065963983535767,
      "learning_rate": 9.75655482003064e-05,
      "loss": 1.3676,
      "step": 19000
    },
    {
      "epoch": 5.080171031533939,
      "grad_norm": 1.9224086999893188,
      "learning_rate": 9.74816190420779e-05,
      "loss": 1.3197,
      "step": 19010
    },
    {
      "epoch": 5.082843399251737,
      "grad_norm": 1.7556531429290771,
      "learning_rate": 9.739769165891158e-05,
      "loss": 1.332,
      "step": 19020
    },
    {
      "epoch": 5.085515766969535,
      "grad_norm": 2.011889696121216,
      "learning_rate": 9.731376610996313e-05,
      "loss": 1.3945,
      "step": 19030
    },
    {
      "epoch": 5.088188134687333,
      "grad_norm": 1.82138192653656,
      "learning_rate": 9.722984245438678e-05,
      "loss": 1.4567,
      "step": 19040
    },
    {
      "epoch": 5.090860502405131,
      "grad_norm": 1.7994951009750366,
      "learning_rate": 9.714592075133548e-05,
      "loss": 1.3876,
      "step": 19050
    },
    {
      "epoch": 5.093532870122929,
      "grad_norm": 1.8251056671142578,
      "learning_rate": 9.706200105996089e-05,
      "loss": 1.3479,
      "step": 19060
    },
    {
      "epoch": 5.096205237840727,
      "grad_norm": 1.8530081510543823,
      "learning_rate": 9.697808343941313e-05,
      "loss": 1.4082,
      "step": 19070
    },
    {
      "epoch": 5.098877605558525,
      "grad_norm": 1.7855250835418701,
      "learning_rate": 9.689416794884098e-05,
      "loss": 1.3922,
      "step": 19080
    },
    {
      "epoch": 5.101549973276323,
      "grad_norm": 1.8199368715286255,
      "learning_rate": 9.681025464739161e-05,
      "loss": 1.4061,
      "step": 19090
    },
    {
      "epoch": 5.104222340994121,
      "grad_norm": 1.7470518350601196,
      "learning_rate": 9.672634359421064e-05,
      "loss": 1.3625,
      "step": 19100
    },
    {
      "epoch": 5.106894708711919,
      "grad_norm": 1.8709263801574707,
      "learning_rate": 9.664243484844226e-05,
      "loss": 1.4037,
      "step": 19110
    },
    {
      "epoch": 5.109567076429717,
      "grad_norm": 1.6874449253082275,
      "learning_rate": 9.655852846922889e-05,
      "loss": 1.3718,
      "step": 19120
    },
    {
      "epoch": 5.112239444147515,
      "grad_norm": 1.8321990966796875,
      "learning_rate": 9.647462451571127e-05,
      "loss": 1.3687,
      "step": 19130
    },
    {
      "epoch": 5.114911811865313,
      "grad_norm": 1.8551712036132812,
      "learning_rate": 9.639072304702856e-05,
      "loss": 1.461,
      "step": 19140
    },
    {
      "epoch": 5.117584179583111,
      "grad_norm": 1.9754506349563599,
      "learning_rate": 9.630682412231803e-05,
      "loss": 1.3965,
      "step": 19150
    },
    {
      "epoch": 5.120256547300909,
      "grad_norm": 1.7540966272354126,
      "learning_rate": 9.622292780071527e-05,
      "loss": 1.38,
      "step": 19160
    },
    {
      "epoch": 5.122928915018707,
      "grad_norm": 1.7249715328216553,
      "learning_rate": 9.613903414135394e-05,
      "loss": 1.3296,
      "step": 19170
    },
    {
      "epoch": 5.125601282736505,
      "grad_norm": 1.878151774406433,
      "learning_rate": 9.605514320336586e-05,
      "loss": 1.353,
      "step": 19180
    },
    {
      "epoch": 5.128273650454302,
      "grad_norm": 2.0917043685913086,
      "learning_rate": 9.597125504588099e-05,
      "loss": 1.4847,
      "step": 19190
    },
    {
      "epoch": 5.1309460181721,
      "grad_norm": 1.8457001447677612,
      "learning_rate": 9.588736972802726e-05,
      "loss": 1.4028,
      "step": 19200
    },
    {
      "epoch": 5.133618385889898,
      "grad_norm": 1.8538477420806885,
      "learning_rate": 9.580348730893054e-05,
      "loss": 1.4034,
      "step": 19210
    },
    {
      "epoch": 5.136290753607696,
      "grad_norm": 1.855852484703064,
      "learning_rate": 9.571960784771485e-05,
      "loss": 1.3972,
      "step": 19220
    },
    {
      "epoch": 5.138963121325494,
      "grad_norm": 1.7284225225448608,
      "learning_rate": 9.563573140350192e-05,
      "loss": 1.4122,
      "step": 19230
    },
    {
      "epoch": 5.141635489043292,
      "grad_norm": 1.8227636814117432,
      "learning_rate": 9.555185803541151e-05,
      "loss": 1.4259,
      "step": 19240
    },
    {
      "epoch": 5.14430785676109,
      "grad_norm": 1.6609549522399902,
      "learning_rate": 9.546798780256111e-05,
      "loss": 1.321,
      "step": 19250
    },
    {
      "epoch": 5.146980224478888,
      "grad_norm": 1.7828965187072754,
      "learning_rate": 9.5384120764066e-05,
      "loss": 1.4045,
      "step": 19260
    },
    {
      "epoch": 5.149652592196686,
      "grad_norm": 1.7603487968444824,
      "learning_rate": 9.530025697903934e-05,
      "loss": 1.3366,
      "step": 19270
    },
    {
      "epoch": 5.152324959914484,
      "grad_norm": 1.9478005170822144,
      "learning_rate": 9.521639650659182e-05,
      "loss": 1.3052,
      "step": 19280
    },
    {
      "epoch": 5.154997327632282,
      "grad_norm": 1.7383671998977661,
      "learning_rate": 9.513253940583186e-05,
      "loss": 1.48,
      "step": 19290
    },
    {
      "epoch": 5.15766969535008,
      "grad_norm": 1.86366868019104,
      "learning_rate": 9.504868573586558e-05,
      "loss": 1.4074,
      "step": 19300
    },
    {
      "epoch": 5.160342063067878,
      "grad_norm": 1.9816826581954956,
      "learning_rate": 9.496483555579658e-05,
      "loss": 1.3471,
      "step": 19310
    },
    {
      "epoch": 5.163014430785676,
      "grad_norm": 1.709796667098999,
      "learning_rate": 9.488098892472606e-05,
      "loss": 1.3413,
      "step": 19320
    },
    {
      "epoch": 5.165686798503474,
      "grad_norm": 1.8086800575256348,
      "learning_rate": 9.47971459017527e-05,
      "loss": 1.3767,
      "step": 19330
    },
    {
      "epoch": 5.168359166221272,
      "grad_norm": 1.8949642181396484,
      "learning_rate": 9.47133065459726e-05,
      "loss": 1.4077,
      "step": 19340
    },
    {
      "epoch": 5.17103153393907,
      "grad_norm": 1.7108567953109741,
      "learning_rate": 9.462947091647937e-05,
      "loss": 1.345,
      "step": 19350
    },
    {
      "epoch": 5.173703901656868,
      "grad_norm": 1.874477744102478,
      "learning_rate": 9.454563907236387e-05,
      "loss": 1.4406,
      "step": 19360
    },
    {
      "epoch": 5.176376269374666,
      "grad_norm": 2.145576238632202,
      "learning_rate": 9.44618110727144e-05,
      "loss": 1.373,
      "step": 19370
    },
    {
      "epoch": 5.179048637092464,
      "grad_norm": 1.8650093078613281,
      "learning_rate": 9.43779869766165e-05,
      "loss": 1.3812,
      "step": 19380
    },
    {
      "epoch": 5.181721004810262,
      "grad_norm": 1.8273286819458008,
      "learning_rate": 9.429416684315293e-05,
      "loss": 1.3712,
      "step": 19390
    },
    {
      "epoch": 5.18439337252806,
      "grad_norm": 1.8486456871032715,
      "learning_rate": 9.421035073140377e-05,
      "loss": 1.3746,
      "step": 19400
    },
    {
      "epoch": 5.187065740245858,
      "grad_norm": 1.9003373384475708,
      "learning_rate": 9.412653870044612e-05,
      "loss": 1.3877,
      "step": 19410
    },
    {
      "epoch": 5.189738107963656,
      "grad_norm": 2.0229334831237793,
      "learning_rate": 9.404273080935425e-05,
      "loss": 1.3323,
      "step": 19420
    },
    {
      "epoch": 5.192410475681454,
      "grad_norm": 1.6970276832580566,
      "learning_rate": 9.395892711719962e-05,
      "loss": 1.4055,
      "step": 19430
    },
    {
      "epoch": 5.195082843399252,
      "grad_norm": 1.801120400428772,
      "learning_rate": 9.387512768305059e-05,
      "loss": 1.376,
      "step": 19440
    },
    {
      "epoch": 5.19775521111705,
      "grad_norm": 1.8643522262573242,
      "learning_rate": 9.379133256597255e-05,
      "loss": 1.4128,
      "step": 19450
    },
    {
      "epoch": 5.200427578834848,
      "grad_norm": 1.7856494188308716,
      "learning_rate": 9.370754182502792e-05,
      "loss": 1.3659,
      "step": 19460
    },
    {
      "epoch": 5.203099946552646,
      "grad_norm": 1.9588515758514404,
      "learning_rate": 9.362375551927593e-05,
      "loss": 1.3671,
      "step": 19470
    },
    {
      "epoch": 5.205772314270444,
      "grad_norm": 1.8798905611038208,
      "learning_rate": 9.353997370777282e-05,
      "loss": 1.3511,
      "step": 19480
    },
    {
      "epoch": 5.208444681988242,
      "grad_norm": 1.9828431606292725,
      "learning_rate": 9.345619644957152e-05,
      "loss": 1.388,
      "step": 19490
    },
    {
      "epoch": 5.21111704970604,
      "grad_norm": 1.9904956817626953,
      "learning_rate": 9.337242380372181e-05,
      "loss": 1.3294,
      "step": 19500
    },
    {
      "epoch": 5.213789417423838,
      "grad_norm": 1.6589508056640625,
      "learning_rate": 9.328865582927024e-05,
      "loss": 1.3557,
      "step": 19510
    },
    {
      "epoch": 5.216461785141636,
      "grad_norm": 1.7825688123703003,
      "learning_rate": 9.320489258526007e-05,
      "loss": 1.3411,
      "step": 19520
    },
    {
      "epoch": 5.219134152859433,
      "grad_norm": 1.8600260019302368,
      "learning_rate": 9.312113413073113e-05,
      "loss": 1.3333,
      "step": 19530
    },
    {
      "epoch": 5.221806520577231,
      "grad_norm": 1.7367174625396729,
      "learning_rate": 9.303738052472003e-05,
      "loss": 1.3826,
      "step": 19540
    },
    {
      "epoch": 5.224478888295029,
      "grad_norm": 1.841676950454712,
      "learning_rate": 9.29536318262598e-05,
      "loss": 1.4269,
      "step": 19550
    },
    {
      "epoch": 5.227151256012827,
      "grad_norm": 1.6773523092269897,
      "learning_rate": 9.286988809438019e-05,
      "loss": 1.3821,
      "step": 19560
    },
    {
      "epoch": 5.229823623730625,
      "grad_norm": 1.700252890586853,
      "learning_rate": 9.278614938810729e-05,
      "loss": 1.37,
      "step": 19570
    },
    {
      "epoch": 5.232495991448423,
      "grad_norm": 2.0488553047180176,
      "learning_rate": 9.270241576646368e-05,
      "loss": 1.3832,
      "step": 19580
    },
    {
      "epoch": 5.235168359166221,
      "grad_norm": 1.926261305809021,
      "learning_rate": 9.261868728846845e-05,
      "loss": 1.3946,
      "step": 19590
    },
    {
      "epoch": 5.237840726884019,
      "grad_norm": 1.8437085151672363,
      "learning_rate": 9.2534964013137e-05,
      "loss": 1.3816,
      "step": 19600
    },
    {
      "epoch": 5.240513094601817,
      "grad_norm": 1.9436142444610596,
      "learning_rate": 9.245124599948096e-05,
      "loss": 1.3795,
      "step": 19610
    },
    {
      "epoch": 5.243185462319615,
      "grad_norm": 1.788064956665039,
      "learning_rate": 9.236753330650848e-05,
      "loss": 1.4214,
      "step": 19620
    },
    {
      "epoch": 5.245857830037413,
      "grad_norm": 1.8409327268600464,
      "learning_rate": 9.228382599322374e-05,
      "loss": 1.4142,
      "step": 19630
    },
    {
      "epoch": 5.248530197755211,
      "grad_norm": 1.763501763343811,
      "learning_rate": 9.22001241186273e-05,
      "loss": 1.3932,
      "step": 19640
    },
    {
      "epoch": 5.251202565473009,
      "grad_norm": 1.7721229791641235,
      "learning_rate": 9.211642774171578e-05,
      "loss": 1.317,
      "step": 19650
    },
    {
      "epoch": 5.253874933190807,
      "grad_norm": 1.8915736675262451,
      "learning_rate": 9.203273692148189e-05,
      "loss": 1.3491,
      "step": 19660
    },
    {
      "epoch": 5.256547300908605,
      "grad_norm": 1.7938752174377441,
      "learning_rate": 9.194905171691459e-05,
      "loss": 1.3978,
      "step": 19670
    },
    {
      "epoch": 5.259219668626403,
      "grad_norm": 1.881883144378662,
      "learning_rate": 9.186537218699872e-05,
      "loss": 1.4447,
      "step": 19680
    },
    {
      "epoch": 5.261892036344201,
      "grad_norm": 1.88962721824646,
      "learning_rate": 9.178169839071515e-05,
      "loss": 1.3895,
      "step": 19690
    },
    {
      "epoch": 5.264564404061999,
      "grad_norm": 1.8478615283966064,
      "learning_rate": 9.169803038704083e-05,
      "loss": 1.4221,
      "step": 19700
    },
    {
      "epoch": 5.267236771779797,
      "grad_norm": 1.848584771156311,
      "learning_rate": 9.161436823494843e-05,
      "loss": 1.4194,
      "step": 19710
    },
    {
      "epoch": 5.269909139497595,
      "grad_norm": 1.8469613790512085,
      "learning_rate": 9.15307119934067e-05,
      "loss": 1.346,
      "step": 19720
    },
    {
      "epoch": 5.272581507215393,
      "grad_norm": 1.7804778814315796,
      "learning_rate": 9.144706172138002e-05,
      "loss": 1.4627,
      "step": 19730
    },
    {
      "epoch": 5.275253874933191,
      "grad_norm": 1.8468170166015625,
      "learning_rate": 9.136341747782874e-05,
      "loss": 1.3987,
      "step": 19740
    },
    {
      "epoch": 5.277926242650989,
      "grad_norm": 1.8156429529190063,
      "learning_rate": 9.127977932170888e-05,
      "loss": 1.4457,
      "step": 19750
    },
    {
      "epoch": 5.280598610368787,
      "grad_norm": 1.8687725067138672,
      "learning_rate": 9.119614731197216e-05,
      "loss": 1.3922,
      "step": 19760
    },
    {
      "epoch": 5.283270978086585,
      "grad_norm": 1.7553884983062744,
      "learning_rate": 9.111252150756595e-05,
      "loss": 1.471,
      "step": 19770
    },
    {
      "epoch": 5.285943345804383,
      "grad_norm": 1.8026105165481567,
      "learning_rate": 9.102890196743337e-05,
      "loss": 1.4102,
      "step": 19780
    },
    {
      "epoch": 5.288615713522181,
      "grad_norm": 1.931124210357666,
      "learning_rate": 9.094528875051292e-05,
      "loss": 1.3855,
      "step": 19790
    },
    {
      "epoch": 5.291288081239979,
      "grad_norm": 1.6505416631698608,
      "learning_rate": 9.086168191573888e-05,
      "loss": 1.4462,
      "step": 19800
    },
    {
      "epoch": 5.293960448957777,
      "grad_norm": 1.8449018001556396,
      "learning_rate": 9.077808152204084e-05,
      "loss": 1.4887,
      "step": 19810
    },
    {
      "epoch": 5.296632816675575,
      "grad_norm": 1.9283699989318848,
      "learning_rate": 9.069448762834391e-05,
      "loss": 1.4667,
      "step": 19820
    },
    {
      "epoch": 5.299305184393372,
      "grad_norm": 1.7096954584121704,
      "learning_rate": 9.061090029356867e-05,
      "loss": 1.37,
      "step": 19830
    },
    {
      "epoch": 5.30197755211117,
      "grad_norm": 1.9460853338241577,
      "learning_rate": 9.052731957663104e-05,
      "loss": 1.4034,
      "step": 19840
    },
    {
      "epoch": 5.304649919828968,
      "grad_norm": 1.6662943363189697,
      "learning_rate": 9.044374553644218e-05,
      "loss": 1.4375,
      "step": 19850
    },
    {
      "epoch": 5.307322287546766,
      "grad_norm": 1.8387529850006104,
      "learning_rate": 9.036017823190876e-05,
      "loss": 1.4105,
      "step": 19860
    },
    {
      "epoch": 5.309994655264564,
      "grad_norm": 1.7618277072906494,
      "learning_rate": 9.027661772193249e-05,
      "loss": 1.4752,
      "step": 19870
    },
    {
      "epoch": 5.312667022982362,
      "grad_norm": 1.6710059642791748,
      "learning_rate": 9.019306406541042e-05,
      "loss": 1.3728,
      "step": 19880
    },
    {
      "epoch": 5.31533939070016,
      "grad_norm": 1.8021734952926636,
      "learning_rate": 9.010951732123473e-05,
      "loss": 1.37,
      "step": 19890
    },
    {
      "epoch": 5.318011758417958,
      "grad_norm": 1.6048088073730469,
      "learning_rate": 9.002597754829269e-05,
      "loss": 1.3938,
      "step": 19900
    },
    {
      "epoch": 5.320684126135756,
      "grad_norm": 1.799424409866333,
      "learning_rate": 8.994244480546672e-05,
      "loss": 1.4425,
      "step": 19910
    },
    {
      "epoch": 5.323356493853554,
      "grad_norm": 1.723467469215393,
      "learning_rate": 8.985891915163425e-05,
      "loss": 1.4524,
      "step": 19920
    },
    {
      "epoch": 5.326028861571352,
      "grad_norm": 1.785679817199707,
      "learning_rate": 8.977540064566771e-05,
      "loss": 1.4493,
      "step": 19930
    },
    {
      "epoch": 5.32870122928915,
      "grad_norm": 1.9639917612075806,
      "learning_rate": 8.969188934643454e-05,
      "loss": 1.3669,
      "step": 19940
    },
    {
      "epoch": 5.331373597006948,
      "grad_norm": 1.80316960811615,
      "learning_rate": 8.960838531279698e-05,
      "loss": 1.4181,
      "step": 19950
    },
    {
      "epoch": 5.334045964724746,
      "grad_norm": 1.7789998054504395,
      "learning_rate": 8.952488860361234e-05,
      "loss": 1.3567,
      "step": 19960
    },
    {
      "epoch": 5.336718332442544,
      "grad_norm": 1.8293192386627197,
      "learning_rate": 8.944139927773261e-05,
      "loss": 1.4086,
      "step": 19970
    },
    {
      "epoch": 5.339390700160342,
      "grad_norm": 1.614897608757019,
      "learning_rate": 8.935791739400457e-05,
      "loss": 1.3897,
      "step": 19980
    },
    {
      "epoch": 5.34206306787814,
      "grad_norm": 1.9090148210525513,
      "learning_rate": 8.92744430112699e-05,
      "loss": 1.4291,
      "step": 19990
    },
    {
      "epoch": 5.344735435595938,
      "grad_norm": 1.67301607131958,
      "learning_rate": 8.919097618836488e-05,
      "loss": 1.3847,
      "step": 20000
    },
    {
      "epoch": 5.347407803313736,
      "grad_norm": 1.8355300426483154,
      "learning_rate": 8.91075169841204e-05,
      "loss": 1.454,
      "step": 20010
    },
    {
      "epoch": 5.350080171031534,
      "grad_norm": 1.7994592189788818,
      "learning_rate": 8.902406545736217e-05,
      "loss": 1.3414,
      "step": 20020
    },
    {
      "epoch": 5.352752538749332,
      "grad_norm": 1.747182846069336,
      "learning_rate": 8.89406216669103e-05,
      "loss": 1.4433,
      "step": 20030
    },
    {
      "epoch": 5.35542490646713,
      "grad_norm": 1.8212345838546753,
      "learning_rate": 8.885718567157964e-05,
      "loss": 1.3703,
      "step": 20040
    },
    {
      "epoch": 5.358097274184928,
      "grad_norm": 1.795236349105835,
      "learning_rate": 8.877375753017933e-05,
      "loss": 1.3588,
      "step": 20050
    },
    {
      "epoch": 5.360769641902726,
      "grad_norm": 2.3064260482788086,
      "learning_rate": 8.869033730151312e-05,
      "loss": 1.4676,
      "step": 20060
    },
    {
      "epoch": 5.363442009620524,
      "grad_norm": 1.7423838376998901,
      "learning_rate": 8.860692504437912e-05,
      "loss": 1.4386,
      "step": 20070
    },
    {
      "epoch": 5.366114377338322,
      "grad_norm": 1.7394424676895142,
      "learning_rate": 8.852352081756989e-05,
      "loss": 1.3764,
      "step": 20080
    },
    {
      "epoch": 5.36878674505612,
      "grad_norm": 1.929689645767212,
      "learning_rate": 8.844012467987222e-05,
      "loss": 1.4234,
      "step": 20090
    },
    {
      "epoch": 5.371459112773918,
      "grad_norm": 1.8902708292007446,
      "learning_rate": 8.835673669006733e-05,
      "loss": 1.4606,
      "step": 20100
    },
    {
      "epoch": 5.374131480491716,
      "grad_norm": 1.847443699836731,
      "learning_rate": 8.827335690693055e-05,
      "loss": 1.4488,
      "step": 20110
    },
    {
      "epoch": 5.376803848209514,
      "grad_norm": 1.915730357170105,
      "learning_rate": 8.81899853892316e-05,
      "loss": 1.4001,
      "step": 20120
    },
    {
      "epoch": 5.379476215927312,
      "grad_norm": 1.954857587814331,
      "learning_rate": 8.810662219573421e-05,
      "loss": 1.3996,
      "step": 20130
    },
    {
      "epoch": 5.38214858364511,
      "grad_norm": 1.909435749053955,
      "learning_rate": 8.80232673851963e-05,
      "loss": 1.4051,
      "step": 20140
    },
    {
      "epoch": 5.384820951362908,
      "grad_norm": 1.8227602243423462,
      "learning_rate": 8.793992101636995e-05,
      "loss": 1.3539,
      "step": 20150
    },
    {
      "epoch": 5.387493319080706,
      "grad_norm": 1.9692318439483643,
      "learning_rate": 8.785658314800121e-05,
      "loss": 1.3321,
      "step": 20160
    },
    {
      "epoch": 5.390165686798504,
      "grad_norm": 2.007525682449341,
      "learning_rate": 8.777325383883011e-05,
      "loss": 1.4246,
      "step": 20170
    },
    {
      "epoch": 5.392838054516301,
      "grad_norm": 1.845348596572876,
      "learning_rate": 8.76899331475908e-05,
      "loss": 1.3951,
      "step": 20180
    },
    {
      "epoch": 5.395510422234099,
      "grad_norm": 1.9953432083129883,
      "learning_rate": 8.760662113301113e-05,
      "loss": 1.4315,
      "step": 20190
    },
    {
      "epoch": 5.398182789951897,
      "grad_norm": 1.837344765663147,
      "learning_rate": 8.752331785381308e-05,
      "loss": 1.4457,
      "step": 20200
    },
    {
      "epoch": 5.400855157669695,
      "grad_norm": 1.7662973403930664,
      "learning_rate": 8.74400233687123e-05,
      "loss": 1.4232,
      "step": 20210
    },
    {
      "epoch": 5.403527525387493,
      "grad_norm": 1.7298915386199951,
      "learning_rate": 8.735673773641824e-05,
      "loss": 1.3671,
      "step": 20220
    },
    {
      "epoch": 5.406199893105291,
      "grad_norm": 1.8851754665374756,
      "learning_rate": 8.727346101563425e-05,
      "loss": 1.4451,
      "step": 20230
    },
    {
      "epoch": 5.408872260823089,
      "grad_norm": 1.8406022787094116,
      "learning_rate": 8.719019326505729e-05,
      "loss": 1.3841,
      "step": 20240
    },
    {
      "epoch": 5.411544628540887,
      "grad_norm": 1.753535270690918,
      "learning_rate": 8.710693454337796e-05,
      "loss": 1.3557,
      "step": 20250
    },
    {
      "epoch": 5.414216996258685,
      "grad_norm": 1.8125604391098022,
      "learning_rate": 8.70236849092806e-05,
      "loss": 1.4351,
      "step": 20260
    },
    {
      "epoch": 5.416889363976483,
      "grad_norm": 1.7749202251434326,
      "learning_rate": 8.694044442144312e-05,
      "loss": 1.4501,
      "step": 20270
    },
    {
      "epoch": 5.419561731694281,
      "grad_norm": 1.9773153066635132,
      "learning_rate": 8.685721313853693e-05,
      "loss": 1.4406,
      "step": 20280
    },
    {
      "epoch": 5.422234099412079,
      "grad_norm": 1.7915995121002197,
      "learning_rate": 8.677399111922702e-05,
      "loss": 1.4535,
      "step": 20290
    },
    {
      "epoch": 5.424906467129877,
      "grad_norm": 1.9801911115646362,
      "learning_rate": 8.669077842217173e-05,
      "loss": 1.3725,
      "step": 20300
    },
    {
      "epoch": 5.427578834847675,
      "grad_norm": 1.9179128408432007,
      "learning_rate": 8.660757510602302e-05,
      "loss": 1.4339,
      "step": 20310
    },
    {
      "epoch": 5.430251202565473,
      "grad_norm": 1.912291169166565,
      "learning_rate": 8.652438122942609e-05,
      "loss": 1.4284,
      "step": 20320
    },
    {
      "epoch": 5.432923570283271,
      "grad_norm": 1.7995072603225708,
      "learning_rate": 8.64411968510195e-05,
      "loss": 1.4228,
      "step": 20330
    },
    {
      "epoch": 5.435595938001069,
      "grad_norm": 1.821509838104248,
      "learning_rate": 8.63580220294352e-05,
      "loss": 1.5033,
      "step": 20340
    },
    {
      "epoch": 5.438268305718867,
      "grad_norm": 1.7498754262924194,
      "learning_rate": 8.627485682329827e-05,
      "loss": 1.4062,
      "step": 20350
    },
    {
      "epoch": 5.440940673436665,
      "grad_norm": 1.8918187618255615,
      "learning_rate": 8.619170129122721e-05,
      "loss": 1.5136,
      "step": 20360
    },
    {
      "epoch": 5.443613041154463,
      "grad_norm": 1.8670575618743896,
      "learning_rate": 8.61085554918335e-05,
      "loss": 1.3779,
      "step": 20370
    },
    {
      "epoch": 5.446285408872261,
      "grad_norm": 1.6510086059570312,
      "learning_rate": 8.602541948372182e-05,
      "loss": 1.4161,
      "step": 20380
    },
    {
      "epoch": 5.448957776590059,
      "grad_norm": 1.811910629272461,
      "learning_rate": 8.594229332549006e-05,
      "loss": 1.443,
      "step": 20390
    },
    {
      "epoch": 5.451630144307857,
      "grad_norm": 1.8337417840957642,
      "learning_rate": 8.585917707572903e-05,
      "loss": 1.4084,
      "step": 20400
    },
    {
      "epoch": 5.454302512025655,
      "grad_norm": 1.7589662075042725,
      "learning_rate": 8.577607079302254e-05,
      "loss": 1.4467,
      "step": 20410
    },
    {
      "epoch": 5.456974879743453,
      "grad_norm": 1.909666895866394,
      "learning_rate": 8.569297453594755e-05,
      "loss": 1.3712,
      "step": 20420
    },
    {
      "epoch": 5.459647247461251,
      "grad_norm": 1.9083164930343628,
      "learning_rate": 8.56098883630738e-05,
      "loss": 1.4333,
      "step": 20430
    },
    {
      "epoch": 5.462319615179049,
      "grad_norm": 1.8496124744415283,
      "learning_rate": 8.552681233296395e-05,
      "loss": 1.4009,
      "step": 20440
    },
    {
      "epoch": 5.464991982896847,
      "grad_norm": 1.6915220022201538,
      "learning_rate": 8.544374650417354e-05,
      "loss": 1.4183,
      "step": 20450
    },
    {
      "epoch": 5.467664350614645,
      "grad_norm": 1.8739166259765625,
      "learning_rate": 8.536069093525089e-05,
      "loss": 1.4258,
      "step": 20460
    },
    {
      "epoch": 5.470336718332443,
      "grad_norm": 1.8223084211349487,
      "learning_rate": 8.527764568473714e-05,
      "loss": 1.4589,
      "step": 20470
    },
    {
      "epoch": 5.47300908605024,
      "grad_norm": 1.802821159362793,
      "learning_rate": 8.51946108111661e-05,
      "loss": 1.4588,
      "step": 20480
    },
    {
      "epoch": 5.475681453768038,
      "grad_norm": 1.7672052383422852,
      "learning_rate": 8.511158637306425e-05,
      "loss": 1.4324,
      "step": 20490
    },
    {
      "epoch": 5.478353821485836,
      "grad_norm": 2.055225372314453,
      "learning_rate": 8.502857242895082e-05,
      "loss": 1.3586,
      "step": 20500
    },
    {
      "epoch": 5.481026189203634,
      "grad_norm": 1.851271152496338,
      "learning_rate": 8.494556903733751e-05,
      "loss": 1.4696,
      "step": 20510
    },
    {
      "epoch": 5.483698556921432,
      "grad_norm": 1.7848930358886719,
      "learning_rate": 8.486257625672872e-05,
      "loss": 1.4534,
      "step": 20520
    },
    {
      "epoch": 5.48637092463923,
      "grad_norm": 1.7438911199569702,
      "learning_rate": 8.477959414562126e-05,
      "loss": 1.4173,
      "step": 20530
    },
    {
      "epoch": 5.489043292357028,
      "grad_norm": 1.926193118095398,
      "learning_rate": 8.469662276250442e-05,
      "loss": 1.4543,
      "step": 20540
    },
    {
      "epoch": 5.491715660074826,
      "grad_norm": 1.714095115661621,
      "learning_rate": 8.461366216586005e-05,
      "loss": 1.4284,
      "step": 20550
    },
    {
      "epoch": 5.494388027792624,
      "grad_norm": 1.7335968017578125,
      "learning_rate": 8.453071241416225e-05,
      "loss": 1.3828,
      "step": 20560
    },
    {
      "epoch": 5.497060395510422,
      "grad_norm": 1.8304640054702759,
      "learning_rate": 8.444777356587755e-05,
      "loss": 1.3795,
      "step": 20570
    },
    {
      "epoch": 5.49973276322822,
      "grad_norm": 1.7875304222106934,
      "learning_rate": 8.43648456794648e-05,
      "loss": 1.4115,
      "step": 20580
    },
    {
      "epoch": 5.502405130946018,
      "grad_norm": 1.8970807790756226,
      "learning_rate": 8.42819288133751e-05,
      "loss": 1.4616,
      "step": 20590
    },
    {
      "epoch": 5.505077498663816,
      "grad_norm": 1.857749581336975,
      "learning_rate": 8.41990230260518e-05,
      "loss": 1.3808,
      "step": 20600
    },
    {
      "epoch": 5.507749866381614,
      "grad_norm": 1.6583001613616943,
      "learning_rate": 8.411612837593043e-05,
      "loss": 1.3328,
      "step": 20610
    },
    {
      "epoch": 5.510422234099412,
      "grad_norm": 1.8332334756851196,
      "learning_rate": 8.403324492143865e-05,
      "loss": 1.4356,
      "step": 20620
    },
    {
      "epoch": 5.51309460181721,
      "grad_norm": 1.8376007080078125,
      "learning_rate": 8.395037272099629e-05,
      "loss": 1.4134,
      "step": 20630
    },
    {
      "epoch": 5.515766969535008,
      "grad_norm": 1.9075250625610352,
      "learning_rate": 8.386751183301515e-05,
      "loss": 1.4126,
      "step": 20640
    },
    {
      "epoch": 5.518439337252806,
      "grad_norm": 1.7450436353683472,
      "learning_rate": 8.378466231589917e-05,
      "loss": 1.3828,
      "step": 20650
    },
    {
      "epoch": 5.521111704970604,
      "grad_norm": 1.9347198009490967,
      "learning_rate": 8.37018242280442e-05,
      "loss": 1.4048,
      "step": 20660
    },
    {
      "epoch": 5.523784072688402,
      "grad_norm": 1.8395016193389893,
      "learning_rate": 8.361899762783803e-05,
      "loss": 1.4849,
      "step": 20670
    },
    {
      "epoch": 5.5264564404062,
      "grad_norm": 1.7988110780715942,
      "learning_rate": 8.353618257366044e-05,
      "loss": 1.3643,
      "step": 20680
    },
    {
      "epoch": 5.529128808123998,
      "grad_norm": 1.9901680946350098,
      "learning_rate": 8.345337912388297e-05,
      "loss": 1.442,
      "step": 20690
    },
    {
      "epoch": 5.531801175841796,
      "grad_norm": 1.6740586757659912,
      "learning_rate": 8.337058733686898e-05,
      "loss": 1.4137,
      "step": 20700
    },
    {
      "epoch": 5.534473543559594,
      "grad_norm": 1.8861411809921265,
      "learning_rate": 8.328780727097373e-05,
      "loss": 1.4486,
      "step": 20710
    },
    {
      "epoch": 5.537145911277392,
      "grad_norm": 1.8824735879898071,
      "learning_rate": 8.320503898454411e-05,
      "loss": 1.4204,
      "step": 20720
    },
    {
      "epoch": 5.53981827899519,
      "grad_norm": 1.8825196027755737,
      "learning_rate": 8.312228253591867e-05,
      "loss": 1.4858,
      "step": 20730
    },
    {
      "epoch": 5.542490646712988,
      "grad_norm": 1.7106484174728394,
      "learning_rate": 8.303953798342779e-05,
      "loss": 1.4151,
      "step": 20740
    },
    {
      "epoch": 5.545163014430786,
      "grad_norm": 1.8421587944030762,
      "learning_rate": 8.295680538539326e-05,
      "loss": 1.4265,
      "step": 20750
    },
    {
      "epoch": 5.547835382148584,
      "grad_norm": 2.018005609512329,
      "learning_rate": 8.287408480012861e-05,
      "loss": 1.4118,
      "step": 20760
    },
    {
      "epoch": 5.550507749866382,
      "grad_norm": 1.9154627323150635,
      "learning_rate": 8.279137628593882e-05,
      "loss": 1.3656,
      "step": 20770
    },
    {
      "epoch": 5.55318011758418,
      "grad_norm": 1.7975690364837646,
      "learning_rate": 8.270867990112034e-05,
      "loss": 1.4457,
      "step": 20780
    },
    {
      "epoch": 5.555852485301978,
      "grad_norm": 1.9669737815856934,
      "learning_rate": 8.262599570396115e-05,
      "loss": 1.4032,
      "step": 20790
    },
    {
      "epoch": 5.558524853019776,
      "grad_norm": 1.916625738143921,
      "learning_rate": 8.254332375274056e-05,
      "loss": 1.4549,
      "step": 20800
    },
    {
      "epoch": 5.561197220737574,
      "grad_norm": 1.8926728963851929,
      "learning_rate": 8.24606641057293e-05,
      "loss": 1.3828,
      "step": 20810
    },
    {
      "epoch": 5.563869588455372,
      "grad_norm": 1.8983536958694458,
      "learning_rate": 8.237801682118942e-05,
      "loss": 1.3588,
      "step": 20820
    },
    {
      "epoch": 5.566541956173169,
      "grad_norm": 2.161370038986206,
      "learning_rate": 8.22953819573742e-05,
      "loss": 1.2994,
      "step": 20830
    },
    {
      "epoch": 5.569214323890967,
      "grad_norm": 1.795060396194458,
      "learning_rate": 8.22127595725283e-05,
      "loss": 1.4147,
      "step": 20840
    },
    {
      "epoch": 5.571886691608765,
      "grad_norm": 1.8414568901062012,
      "learning_rate": 8.213014972488745e-05,
      "loss": 1.4071,
      "step": 20850
    },
    {
      "epoch": 5.574559059326563,
      "grad_norm": 1.9023696184158325,
      "learning_rate": 8.204755247267855e-05,
      "loss": 1.4326,
      "step": 20860
    },
    {
      "epoch": 5.577231427044361,
      "grad_norm": 1.855142593383789,
      "learning_rate": 8.196496787411975e-05,
      "loss": 1.378,
      "step": 20870
    },
    {
      "epoch": 5.579903794762159,
      "grad_norm": 1.7736339569091797,
      "learning_rate": 8.188239598742014e-05,
      "loss": 1.4636,
      "step": 20880
    },
    {
      "epoch": 5.582576162479957,
      "grad_norm": 1.7798281908035278,
      "learning_rate": 8.179983687077986e-05,
      "loss": 1.4085,
      "step": 20890
    },
    {
      "epoch": 5.585248530197755,
      "grad_norm": 1.8400295972824097,
      "learning_rate": 8.17172905823902e-05,
      "loss": 1.4173,
      "step": 20900
    },
    {
      "epoch": 5.587920897915553,
      "grad_norm": 1.8477122783660889,
      "learning_rate": 8.16347571804332e-05,
      "loss": 1.3912,
      "step": 20910
    },
    {
      "epoch": 5.590593265633351,
      "grad_norm": 1.8249187469482422,
      "learning_rate": 8.155223672308204e-05,
      "loss": 1.3408,
      "step": 20920
    },
    {
      "epoch": 5.593265633351149,
      "grad_norm": 1.8549121618270874,
      "learning_rate": 8.146972926850054e-05,
      "loss": 1.4286,
      "step": 20930
    },
    {
      "epoch": 5.595938001068947,
      "grad_norm": 2.0361337661743164,
      "learning_rate": 8.13872348748435e-05,
      "loss": 1.4204,
      "step": 20940
    },
    {
      "epoch": 5.598610368786745,
      "grad_norm": 1.8047282695770264,
      "learning_rate": 8.130475360025653e-05,
      "loss": 1.4394,
      "step": 20950
    },
    {
      "epoch": 5.601282736504543,
      "grad_norm": 1.8704631328582764,
      "learning_rate": 8.122228550287591e-05,
      "loss": 1.4417,
      "step": 20960
    },
    {
      "epoch": 5.603955104222341,
      "grad_norm": 1.8467167615890503,
      "learning_rate": 8.113983064082866e-05,
      "loss": 1.3722,
      "step": 20970
    },
    {
      "epoch": 5.606627471940139,
      "grad_norm": 1.8335866928100586,
      "learning_rate": 8.105738907223249e-05,
      "loss": 1.414,
      "step": 20980
    },
    {
      "epoch": 5.609299839657937,
      "grad_norm": 1.8786503076553345,
      "learning_rate": 8.097496085519573e-05,
      "loss": 1.3668,
      "step": 20990
    },
    {
      "epoch": 5.611972207375735,
      "grad_norm": 1.8728291988372803,
      "learning_rate": 8.089254604781732e-05,
      "loss": 1.3863,
      "step": 21000
    },
    {
      "epoch": 5.614644575093533,
      "grad_norm": 1.7736185789108276,
      "learning_rate": 8.081014470818672e-05,
      "loss": 1.4979,
      "step": 21010
    },
    {
      "epoch": 5.617316942811331,
      "grad_norm": 1.698287844657898,
      "learning_rate": 8.072775689438385e-05,
      "loss": 1.4372,
      "step": 21020
    },
    {
      "epoch": 5.619989310529129,
      "grad_norm": 1.6462870836257935,
      "learning_rate": 8.064538266447926e-05,
      "loss": 1.4553,
      "step": 21030
    },
    {
      "epoch": 5.622661678246927,
      "grad_norm": 1.702731728553772,
      "learning_rate": 8.056302207653375e-05,
      "loss": 1.3848,
      "step": 21040
    },
    {
      "epoch": 5.625334045964725,
      "grad_norm": 1.9015578031539917,
      "learning_rate": 8.048067518859853e-05,
      "loss": 1.4118,
      "step": 21050
    },
    {
      "epoch": 5.628006413682523,
      "grad_norm": 2.0413286685943604,
      "learning_rate": 8.03983420587153e-05,
      "loss": 1.3706,
      "step": 21060
    },
    {
      "epoch": 5.630678781400321,
      "grad_norm": 2.0313758850097656,
      "learning_rate": 8.031602274491585e-05,
      "loss": 1.4304,
      "step": 21070
    },
    {
      "epoch": 5.633351149118119,
      "grad_norm": 1.720688819885254,
      "learning_rate": 8.023371730522244e-05,
      "loss": 1.4069,
      "step": 21080
    },
    {
      "epoch": 5.636023516835917,
      "grad_norm": 1.7646926641464233,
      "learning_rate": 8.015142579764741e-05,
      "loss": 1.4524,
      "step": 21090
    },
    {
      "epoch": 5.638695884553715,
      "grad_norm": 1.887972354888916,
      "learning_rate": 8.006914828019327e-05,
      "loss": 1.3912,
      "step": 21100
    },
    {
      "epoch": 5.641368252271512,
      "grad_norm": 1.9093656539916992,
      "learning_rate": 7.998688481085283e-05,
      "loss": 1.3113,
      "step": 21110
    },
    {
      "epoch": 5.64404061998931,
      "grad_norm": 1.8109502792358398,
      "learning_rate": 7.99046354476088e-05,
      "loss": 1.4354,
      "step": 21120
    },
    {
      "epoch": 5.646712987707108,
      "grad_norm": 1.6826542615890503,
      "learning_rate": 7.982240024843404e-05,
      "loss": 1.3939,
      "step": 21130
    },
    {
      "epoch": 5.649385355424906,
      "grad_norm": 1.8867690563201904,
      "learning_rate": 7.974017927129148e-05,
      "loss": 1.4446,
      "step": 21140
    },
    {
      "epoch": 5.652057723142704,
      "grad_norm": 1.8560004234313965,
      "learning_rate": 7.965797257413391e-05,
      "loss": 1.4355,
      "step": 21150
    },
    {
      "epoch": 5.654730090860502,
      "grad_norm": 1.6699858903884888,
      "learning_rate": 7.957578021490415e-05,
      "loss": 1.4192,
      "step": 21160
    },
    {
      "epoch": 5.6574024585783,
      "grad_norm": 1.786773920059204,
      "learning_rate": 7.949360225153488e-05,
      "loss": 1.3611,
      "step": 21170
    },
    {
      "epoch": 5.660074826296098,
      "grad_norm": 1.9619256258010864,
      "learning_rate": 7.941143874194859e-05,
      "loss": 1.3829,
      "step": 21180
    },
    {
      "epoch": 5.662747194013896,
      "grad_norm": 1.8596181869506836,
      "learning_rate": 7.932928974405767e-05,
      "loss": 1.4219,
      "step": 21190
    },
    {
      "epoch": 5.665419561731694,
      "grad_norm": 1.9420338869094849,
      "learning_rate": 7.92471553157642e-05,
      "loss": 1.4303,
      "step": 21200
    },
    {
      "epoch": 5.668091929449492,
      "grad_norm": 1.885595440864563,
      "learning_rate": 7.916503551496002e-05,
      "loss": 1.4115,
      "step": 21210
    },
    {
      "epoch": 5.67076429716729,
      "grad_norm": 2.14754319190979,
      "learning_rate": 7.908293039952672e-05,
      "loss": 1.4569,
      "step": 21220
    },
    {
      "epoch": 5.673436664885088,
      "grad_norm": 1.840187907218933,
      "learning_rate": 7.900084002733541e-05,
      "loss": 1.3816,
      "step": 21230
    },
    {
      "epoch": 5.676109032602886,
      "grad_norm": 1.805789828300476,
      "learning_rate": 7.891876445624695e-05,
      "loss": 1.4556,
      "step": 21240
    },
    {
      "epoch": 5.678781400320684,
      "grad_norm": 1.8917306661605835,
      "learning_rate": 7.883670374411168e-05,
      "loss": 1.3914,
      "step": 21250
    },
    {
      "epoch": 5.681453768038482,
      "grad_norm": 1.7766830921173096,
      "learning_rate": 7.875465794876944e-05,
      "loss": 1.4106,
      "step": 21260
    },
    {
      "epoch": 5.68412613575628,
      "grad_norm": 1.8623842000961304,
      "learning_rate": 7.867262712804967e-05,
      "loss": 1.4727,
      "step": 21270
    },
    {
      "epoch": 5.686798503474078,
      "grad_norm": 1.8976304531097412,
      "learning_rate": 7.859061133977116e-05,
      "loss": 1.3634,
      "step": 21280
    },
    {
      "epoch": 5.689470871191876,
      "grad_norm": 1.7975999116897583,
      "learning_rate": 7.850861064174209e-05,
      "loss": 1.4084,
      "step": 21290
    },
    {
      "epoch": 5.692143238909674,
      "grad_norm": 1.9109054803848267,
      "learning_rate": 7.842662509176013e-05,
      "loss": 1.3689,
      "step": 21300
    },
    {
      "epoch": 5.694815606627472,
      "grad_norm": 2.0148255825042725,
      "learning_rate": 7.834465474761212e-05,
      "loss": 1.3981,
      "step": 21310
    },
    {
      "epoch": 5.69748797434527,
      "grad_norm": 1.7822149991989136,
      "learning_rate": 7.82626996670743e-05,
      "loss": 1.4759,
      "step": 21320
    },
    {
      "epoch": 5.700160342063068,
      "grad_norm": 1.698251485824585,
      "learning_rate": 7.818075990791208e-05,
      "loss": 1.4854,
      "step": 21330
    },
    {
      "epoch": 5.702832709780866,
      "grad_norm": 1.8046317100524902,
      "learning_rate": 7.809883552788009e-05,
      "loss": 1.4243,
      "step": 21340
    },
    {
      "epoch": 5.705505077498664,
      "grad_norm": 1.8524879217147827,
      "learning_rate": 7.801692658472214e-05,
      "loss": 1.3466,
      "step": 21350
    },
    {
      "epoch": 5.708177445216462,
      "grad_norm": 1.8356361389160156,
      "learning_rate": 7.793503313617113e-05,
      "loss": 1.4532,
      "step": 21360
    },
    {
      "epoch": 5.71084981293426,
      "grad_norm": 1.8731732368469238,
      "learning_rate": 7.785315523994903e-05,
      "loss": 1.4338,
      "step": 21370
    },
    {
      "epoch": 5.713522180652058,
      "grad_norm": 1.968556523323059,
      "learning_rate": 7.777129295376693e-05,
      "loss": 1.4749,
      "step": 21380
    },
    {
      "epoch": 5.716194548369856,
      "grad_norm": 1.7161849737167358,
      "learning_rate": 7.768944633532477e-05,
      "loss": 1.4179,
      "step": 21390
    },
    {
      "epoch": 5.718866916087654,
      "grad_norm": 1.7846759557724,
      "learning_rate": 7.760761544231166e-05,
      "loss": 1.4557,
      "step": 21400
    },
    {
      "epoch": 5.721539283805452,
      "grad_norm": 1.7804423570632935,
      "learning_rate": 7.752580033240542e-05,
      "loss": 1.4131,
      "step": 21410
    },
    {
      "epoch": 5.72421165152325,
      "grad_norm": 1.953589677810669,
      "learning_rate": 7.744400106327279e-05,
      "loss": 1.3795,
      "step": 21420
    },
    {
      "epoch": 5.726884019241048,
      "grad_norm": 1.726546049118042,
      "learning_rate": 7.736221769256946e-05,
      "loss": 1.4868,
      "step": 21430
    },
    {
      "epoch": 5.729556386958846,
      "grad_norm": 1.8707987070083618,
      "learning_rate": 7.72804502779398e-05,
      "loss": 1.4808,
      "step": 21440
    },
    {
      "epoch": 5.732228754676644,
      "grad_norm": 1.7484774589538574,
      "learning_rate": 7.719869887701694e-05,
      "loss": 1.4293,
      "step": 21450
    },
    {
      "epoch": 5.734901122394442,
      "grad_norm": 1.8914085626602173,
      "learning_rate": 7.711696354742281e-05,
      "loss": 1.4495,
      "step": 21460
    },
    {
      "epoch": 5.73757349011224,
      "grad_norm": 1.8179551362991333,
      "learning_rate": 7.70352443467679e-05,
      "loss": 1.4383,
      "step": 21470
    },
    {
      "epoch": 5.740245857830037,
      "grad_norm": 1.9113681316375732,
      "learning_rate": 7.695354133265144e-05,
      "loss": 1.4195,
      "step": 21480
    },
    {
      "epoch": 5.742918225547835,
      "grad_norm": 1.7981265783309937,
      "learning_rate": 7.687185456266116e-05,
      "loss": 1.4797,
      "step": 21490
    },
    {
      "epoch": 5.745590593265633,
      "grad_norm": 1.8957091569900513,
      "learning_rate": 7.679018409437335e-05,
      "loss": 1.4402,
      "step": 21500
    },
    {
      "epoch": 5.748262960983431,
      "grad_norm": 1.728651762008667,
      "learning_rate": 7.67085299853529e-05,
      "loss": 1.4046,
      "step": 21510
    },
    {
      "epoch": 5.750935328701229,
      "grad_norm": 1.767903447151184,
      "learning_rate": 7.662689229315307e-05,
      "loss": 1.3694,
      "step": 21520
    },
    {
      "epoch": 5.753607696419027,
      "grad_norm": 1.8116743564605713,
      "learning_rate": 7.654527107531559e-05,
      "loss": 1.3906,
      "step": 21530
    },
    {
      "epoch": 5.756280064136825,
      "grad_norm": 1.7834914922714233,
      "learning_rate": 7.646366638937057e-05,
      "loss": 1.463,
      "step": 21540
    },
    {
      "epoch": 5.758952431854623,
      "grad_norm": 1.8045434951782227,
      "learning_rate": 7.638207829283645e-05,
      "loss": 1.4808,
      "step": 21550
    },
    {
      "epoch": 5.761624799572421,
      "grad_norm": 1.7416683435440063,
      "learning_rate": 7.630050684322005e-05,
      "loss": 1.3901,
      "step": 21560
    },
    {
      "epoch": 5.764297167290219,
      "grad_norm": 1.8306225538253784,
      "learning_rate": 7.621895209801636e-05,
      "loss": 1.4188,
      "step": 21570
    },
    {
      "epoch": 5.766969535008017,
      "grad_norm": 1.7175840139389038,
      "learning_rate": 7.613741411470862e-05,
      "loss": 1.3847,
      "step": 21580
    },
    {
      "epoch": 5.769641902725815,
      "grad_norm": 1.8680164813995361,
      "learning_rate": 7.605589295076834e-05,
      "loss": 1.4709,
      "step": 21590
    },
    {
      "epoch": 5.772314270443613,
      "grad_norm": 1.7645819187164307,
      "learning_rate": 7.59743886636551e-05,
      "loss": 1.4534,
      "step": 21600
    },
    {
      "epoch": 5.774986638161411,
      "grad_norm": 1.9910109043121338,
      "learning_rate": 7.58929013108165e-05,
      "loss": 1.4604,
      "step": 21610
    },
    {
      "epoch": 5.777659005879209,
      "grad_norm": 1.7553080320358276,
      "learning_rate": 7.581143094968845e-05,
      "loss": 1.4155,
      "step": 21620
    },
    {
      "epoch": 5.780331373597007,
      "grad_norm": 1.9407318830490112,
      "learning_rate": 7.572997763769459e-05,
      "loss": 1.3913,
      "step": 21630
    },
    {
      "epoch": 5.783003741314805,
      "grad_norm": 2.0086443424224854,
      "learning_rate": 7.564854143224679e-05,
      "loss": 1.3741,
      "step": 21640
    },
    {
      "epoch": 5.785676109032603,
      "grad_norm": 1.8153927326202393,
      "learning_rate": 7.556712239074473e-05,
      "loss": 1.4104,
      "step": 21650
    },
    {
      "epoch": 5.788348476750401,
      "grad_norm": 1.8651721477508545,
      "learning_rate": 7.548572057057594e-05,
      "loss": 1.4595,
      "step": 21660
    },
    {
      "epoch": 5.791020844468199,
      "grad_norm": 1.723351240158081,
      "learning_rate": 7.540433602911603e-05,
      "loss": 1.4139,
      "step": 21670
    },
    {
      "epoch": 5.793693212185997,
      "grad_norm": 1.7649847269058228,
      "learning_rate": 7.53229688237282e-05,
      "loss": 1.4655,
      "step": 21680
    },
    {
      "epoch": 5.796365579903795,
      "grad_norm": 1.9785157442092896,
      "learning_rate": 7.524161901176353e-05,
      "loss": 1.4393,
      "step": 21690
    },
    {
      "epoch": 5.799037947621593,
      "grad_norm": 1.8101013898849487,
      "learning_rate": 7.516028665056086e-05,
      "loss": 1.3584,
      "step": 21700
    },
    {
      "epoch": 5.801710315339391,
      "grad_norm": 1.8289343118667603,
      "learning_rate": 7.50789717974467e-05,
      "loss": 1.4208,
      "step": 21710
    },
    {
      "epoch": 5.804382683057189,
      "grad_norm": 1.940071702003479,
      "learning_rate": 7.499767450973523e-05,
      "loss": 1.4662,
      "step": 21720
    },
    {
      "epoch": 5.807055050774987,
      "grad_norm": 1.7949780225753784,
      "learning_rate": 7.491639484472819e-05,
      "loss": 1.4615,
      "step": 21730
    },
    {
      "epoch": 5.809727418492785,
      "grad_norm": 1.6472865343093872,
      "learning_rate": 7.483513285971502e-05,
      "loss": 1.4643,
      "step": 21740
    },
    {
      "epoch": 5.812399786210583,
      "grad_norm": 1.8350540399551392,
      "learning_rate": 7.475388861197261e-05,
      "loss": 1.4522,
      "step": 21750
    },
    {
      "epoch": 5.81507215392838,
      "grad_norm": 1.730674386024475,
      "learning_rate": 7.467266215876537e-05,
      "loss": 1.4097,
      "step": 21760
    },
    {
      "epoch": 5.817744521646178,
      "grad_norm": 1.818445086479187,
      "learning_rate": 7.459145355734509e-05,
      "loss": 1.4022,
      "step": 21770
    },
    {
      "epoch": 5.820416889363976,
      "grad_norm": 1.796509027481079,
      "learning_rate": 7.451026286495118e-05,
      "loss": 1.4705,
      "step": 21780
    },
    {
      "epoch": 5.823089257081774,
      "grad_norm": 1.96515691280365,
      "learning_rate": 7.442909013881018e-05,
      "loss": 1.3849,
      "step": 21790
    },
    {
      "epoch": 5.825761624799572,
      "grad_norm": 1.7800146341323853,
      "learning_rate": 7.434793543613621e-05,
      "loss": 1.4661,
      "step": 21800
    },
    {
      "epoch": 5.82843399251737,
      "grad_norm": 1.7168282270431519,
      "learning_rate": 7.42667988141305e-05,
      "loss": 1.3941,
      "step": 21810
    },
    {
      "epoch": 5.831106360235168,
      "grad_norm": 1.8376736640930176,
      "learning_rate": 7.418568032998153e-05,
      "loss": 1.4191,
      "step": 21820
    },
    {
      "epoch": 5.833778727952966,
      "grad_norm": 1.7231342792510986,
      "learning_rate": 7.410458004086519e-05,
      "loss": 1.4534,
      "step": 21830
    },
    {
      "epoch": 5.836451095670764,
      "grad_norm": 1.7365481853485107,
      "learning_rate": 7.402349800394436e-05,
      "loss": 1.4341,
      "step": 21840
    },
    {
      "epoch": 5.839123463388562,
      "grad_norm": 1.8495562076568604,
      "learning_rate": 7.394243427636906e-05,
      "loss": 1.444,
      "step": 21850
    },
    {
      "epoch": 5.84179583110636,
      "grad_norm": 1.8117125034332275,
      "learning_rate": 7.386138891527659e-05,
      "loss": 1.4369,
      "step": 21860
    },
    {
      "epoch": 5.844468198824158,
      "grad_norm": 2.011932611465454,
      "learning_rate": 7.378036197779105e-05,
      "loss": 1.4503,
      "step": 21870
    },
    {
      "epoch": 5.847140566541956,
      "grad_norm": 1.7977993488311768,
      "learning_rate": 7.369935352102373e-05,
      "loss": 1.4153,
      "step": 21880
    },
    {
      "epoch": 5.849812934259754,
      "grad_norm": 1.7696391344070435,
      "learning_rate": 7.361836360207287e-05,
      "loss": 1.3521,
      "step": 21890
    },
    {
      "epoch": 5.852485301977552,
      "grad_norm": 1.8803189992904663,
      "learning_rate": 7.353739227802357e-05,
      "loss": 1.3855,
      "step": 21900
    },
    {
      "epoch": 5.85515766969535,
      "grad_norm": 1.787385106086731,
      "learning_rate": 7.34564396059479e-05,
      "loss": 1.3908,
      "step": 21910
    },
    {
      "epoch": 5.857830037413148,
      "grad_norm": 1.916798710823059,
      "learning_rate": 7.337550564290473e-05,
      "loss": 1.4026,
      "step": 21920
    },
    {
      "epoch": 5.860502405130946,
      "grad_norm": 1.7575953006744385,
      "learning_rate": 7.329459044593976e-05,
      "loss": 1.4585,
      "step": 21930
    },
    {
      "epoch": 5.863174772848744,
      "grad_norm": 1.8872182369232178,
      "learning_rate": 7.321369407208551e-05,
      "loss": 1.3945,
      "step": 21940
    },
    {
      "epoch": 5.865847140566542,
      "grad_norm": 1.7646726369857788,
      "learning_rate": 7.313281657836113e-05,
      "loss": 1.3848,
      "step": 21950
    },
    {
      "epoch": 5.86851950828434,
      "grad_norm": 1.8913259506225586,
      "learning_rate": 7.30519580217726e-05,
      "loss": 1.4149,
      "step": 21960
    },
    {
      "epoch": 5.871191876002138,
      "grad_norm": 1.8760039806365967,
      "learning_rate": 7.297111845931243e-05,
      "loss": 1.4296,
      "step": 21970
    },
    {
      "epoch": 5.873864243719936,
      "grad_norm": 1.732088327407837,
      "learning_rate": 7.289029794795976e-05,
      "loss": 1.4447,
      "step": 21980
    },
    {
      "epoch": 5.876536611437734,
      "grad_norm": 1.9041616916656494,
      "learning_rate": 7.280949654468043e-05,
      "loss": 1.4114,
      "step": 21990
    },
    {
      "epoch": 5.879208979155532,
      "grad_norm": 1.7664214372634888,
      "learning_rate": 7.272871430642663e-05,
      "loss": 1.3673,
      "step": 22000
    },
    {
      "epoch": 5.88188134687333,
      "grad_norm": 1.9334638118743896,
      "learning_rate": 7.264795129013709e-05,
      "loss": 1.5063,
      "step": 22010
    },
    {
      "epoch": 5.884553714591128,
      "grad_norm": 1.80722177028656,
      "learning_rate": 7.256720755273713e-05,
      "loss": 1.4412,
      "step": 22020
    },
    {
      "epoch": 5.887226082308926,
      "grad_norm": 1.869354248046875,
      "learning_rate": 7.248648315113829e-05,
      "loss": 1.4711,
      "step": 22030
    },
    {
      "epoch": 5.889898450026724,
      "grad_norm": 1.9294512271881104,
      "learning_rate": 7.240577814223863e-05,
      "loss": 1.4012,
      "step": 22040
    },
    {
      "epoch": 5.892570817744522,
      "grad_norm": 1.8781780004501343,
      "learning_rate": 7.232509258292246e-05,
      "loss": 1.438,
      "step": 22050
    },
    {
      "epoch": 5.89524318546232,
      "grad_norm": 1.8945399522781372,
      "learning_rate": 7.224442653006038e-05,
      "loss": 1.3915,
      "step": 22060
    },
    {
      "epoch": 5.897915553180118,
      "grad_norm": 2.3173465728759766,
      "learning_rate": 7.216378004050928e-05,
      "loss": 1.4527,
      "step": 22070
    },
    {
      "epoch": 5.900587920897916,
      "grad_norm": 1.8504656553268433,
      "learning_rate": 7.208315317111224e-05,
      "loss": 1.4334,
      "step": 22080
    },
    {
      "epoch": 5.903260288615714,
      "grad_norm": 1.7004691362380981,
      "learning_rate": 7.20025459786985e-05,
      "loss": 1.3564,
      "step": 22090
    },
    {
      "epoch": 5.905932656333512,
      "grad_norm": 1.9632539749145508,
      "learning_rate": 7.192195852008347e-05,
      "loss": 1.4859,
      "step": 22100
    },
    {
      "epoch": 5.90860502405131,
      "grad_norm": 1.7879068851470947,
      "learning_rate": 7.184139085206858e-05,
      "loss": 1.4729,
      "step": 22110
    },
    {
      "epoch": 5.911277391769107,
      "grad_norm": 1.8004533052444458,
      "learning_rate": 7.176084303144141e-05,
      "loss": 1.4285,
      "step": 22120
    },
    {
      "epoch": 5.913949759486905,
      "grad_norm": 1.7685760259628296,
      "learning_rate": 7.16803151149755e-05,
      "loss": 1.3682,
      "step": 22130
    },
    {
      "epoch": 5.916622127204703,
      "grad_norm": 1.9934276342391968,
      "learning_rate": 7.159980715943027e-05,
      "loss": 1.411,
      "step": 22140
    },
    {
      "epoch": 5.919294494922501,
      "grad_norm": 1.8428826332092285,
      "learning_rate": 7.151931922155126e-05,
      "loss": 1.4323,
      "step": 22150
    },
    {
      "epoch": 5.921966862640299,
      "grad_norm": 1.8446909189224243,
      "learning_rate": 7.143885135806975e-05,
      "loss": 1.3668,
      "step": 22160
    },
    {
      "epoch": 5.924639230358097,
      "grad_norm": 1.707067847251892,
      "learning_rate": 7.13584036257029e-05,
      "loss": 1.4099,
      "step": 22170
    },
    {
      "epoch": 5.927311598075895,
      "grad_norm": 1.795223355293274,
      "learning_rate": 7.127797608115376e-05,
      "loss": 1.414,
      "step": 22180
    },
    {
      "epoch": 5.929983965793693,
      "grad_norm": 1.6973556280136108,
      "learning_rate": 7.119756878111104e-05,
      "loss": 1.3696,
      "step": 22190
    },
    {
      "epoch": 5.932656333511491,
      "grad_norm": 1.8875623941421509,
      "learning_rate": 7.111718178224925e-05,
      "loss": 1.4816,
      "step": 22200
    },
    {
      "epoch": 5.935328701229289,
      "grad_norm": 2.0357954502105713,
      "learning_rate": 7.10368151412286e-05,
      "loss": 1.4494,
      "step": 22210
    },
    {
      "epoch": 5.938001068947087,
      "grad_norm": 1.8732619285583496,
      "learning_rate": 7.095646891469482e-05,
      "loss": 1.4331,
      "step": 22220
    },
    {
      "epoch": 5.940673436664885,
      "grad_norm": 1.720133662223816,
      "learning_rate": 7.08761431592795e-05,
      "loss": 1.4076,
      "step": 22230
    },
    {
      "epoch": 5.943345804382683,
      "grad_norm": 1.7815426588058472,
      "learning_rate": 7.079583793159957e-05,
      "loss": 1.455,
      "step": 22240
    },
    {
      "epoch": 5.946018172100481,
      "grad_norm": 1.8538612127304077,
      "learning_rate": 7.071555328825758e-05,
      "loss": 1.406,
      "step": 22250
    },
    {
      "epoch": 5.948690539818279,
      "grad_norm": 1.8338390588760376,
      "learning_rate": 7.063528928584158e-05,
      "loss": 1.3684,
      "step": 22260
    },
    {
      "epoch": 5.951362907536077,
      "grad_norm": 1.7937397956848145,
      "learning_rate": 7.055504598092503e-05,
      "loss": 1.4068,
      "step": 22270
    },
    {
      "epoch": 5.954035275253875,
      "grad_norm": 1.9398363828659058,
      "learning_rate": 7.04748234300669e-05,
      "loss": 1.4004,
      "step": 22280
    },
    {
      "epoch": 5.956707642971673,
      "grad_norm": 1.991209626197815,
      "learning_rate": 7.039462168981144e-05,
      "loss": 1.3988,
      "step": 22290
    },
    {
      "epoch": 5.959380010689471,
      "grad_norm": 1.8283241987228394,
      "learning_rate": 7.031444081668817e-05,
      "loss": 1.4362,
      "step": 22300
    },
    {
      "epoch": 5.962052378407269,
      "grad_norm": 1.9161646366119385,
      "learning_rate": 7.023428086721209e-05,
      "loss": 1.437,
      "step": 22310
    },
    {
      "epoch": 5.964724746125067,
      "grad_norm": 1.9440922737121582,
      "learning_rate": 7.015414189788332e-05,
      "loss": 1.4421,
      "step": 22320
    },
    {
      "epoch": 5.967397113842865,
      "grad_norm": 1.8406633138656616,
      "learning_rate": 7.007402396518717e-05,
      "loss": 1.386,
      "step": 22330
    },
    {
      "epoch": 5.970069481560663,
      "grad_norm": 1.7498799562454224,
      "learning_rate": 6.999392712559425e-05,
      "loss": 1.5282,
      "step": 22340
    },
    {
      "epoch": 5.972741849278461,
      "grad_norm": 1.7741248607635498,
      "learning_rate": 6.991385143556015e-05,
      "loss": 1.4413,
      "step": 22350
    },
    {
      "epoch": 5.975414216996259,
      "grad_norm": 1.7782055139541626,
      "learning_rate": 6.983379695152571e-05,
      "loss": 1.4095,
      "step": 22360
    },
    {
      "epoch": 5.978086584714057,
      "grad_norm": 1.7061249017715454,
      "learning_rate": 6.97537637299167e-05,
      "loss": 1.4772,
      "step": 22370
    },
    {
      "epoch": 5.980758952431855,
      "grad_norm": 1.7291443347930908,
      "learning_rate": 6.967375182714391e-05,
      "loss": 1.4935,
      "step": 22380
    },
    {
      "epoch": 5.983431320149653,
      "grad_norm": 2.1783764362335205,
      "learning_rate": 6.95937612996032e-05,
      "loss": 1.4364,
      "step": 22390
    },
    {
      "epoch": 5.98610368786745,
      "grad_norm": 1.6925691366195679,
      "learning_rate": 6.951379220367528e-05,
      "loss": 1.4147,
      "step": 22400
    },
    {
      "epoch": 5.988776055585248,
      "grad_norm": 1.7771650552749634,
      "learning_rate": 6.943384459572575e-05,
      "loss": 1.4636,
      "step": 22410
    },
    {
      "epoch": 5.991448423303046,
      "grad_norm": 1.9734643697738647,
      "learning_rate": 6.935391853210515e-05,
      "loss": 1.4018,
      "step": 22420
    },
    {
      "epoch": 5.994120791020844,
      "grad_norm": 1.6722396612167358,
      "learning_rate": 6.927401406914872e-05,
      "loss": 1.407,
      "step": 22430
    },
    {
      "epoch": 5.996793158738642,
      "grad_norm": 1.7412807941436768,
      "learning_rate": 6.919413126317656e-05,
      "loss": 1.4692,
      "step": 22440
    },
    {
      "epoch": 5.99946552645644,
      "grad_norm": 1.8932757377624512,
      "learning_rate": 6.911427017049346e-05,
      "loss": 1.3839,
      "step": 22450
    },
    {
      "epoch": 6.002137894174238,
      "grad_norm": 1.9046579599380493,
      "learning_rate": 6.903443084738892e-05,
      "loss": 1.3296,
      "step": 22460
    },
    {
      "epoch": 6.004810261892036,
      "grad_norm": 1.9861325025558472,
      "learning_rate": 6.895461335013712e-05,
      "loss": 1.3221,
      "step": 22470
    },
    {
      "epoch": 6.007482629609834,
      "grad_norm": 1.8595702648162842,
      "learning_rate": 6.887481773499683e-05,
      "loss": 1.2499,
      "step": 22480
    },
    {
      "epoch": 6.010154997327632,
      "grad_norm": 2.0673961639404297,
      "learning_rate": 6.879504405821134e-05,
      "loss": 1.3126,
      "step": 22490
    },
    {
      "epoch": 6.01282736504543,
      "grad_norm": 1.9374237060546875,
      "learning_rate": 6.871529237600864e-05,
      "loss": 1.2919,
      "step": 22500
    },
    {
      "epoch": 6.015499732763228,
      "grad_norm": 1.9081655740737915,
      "learning_rate": 6.8635562744601e-05,
      "loss": 1.2276,
      "step": 22510
    },
    {
      "epoch": 6.018172100481026,
      "grad_norm": 2.1691958904266357,
      "learning_rate": 6.855585522018537e-05,
      "loss": 1.3396,
      "step": 22520
    },
    {
      "epoch": 6.020844468198824,
      "grad_norm": 2.029203414916992,
      "learning_rate": 6.847616985894296e-05,
      "loss": 1.3081,
      "step": 22530
    },
    {
      "epoch": 6.023516835916622,
      "grad_norm": 1.9637291431427002,
      "learning_rate": 6.839650671703937e-05,
      "loss": 1.2681,
      "step": 22540
    },
    {
      "epoch": 6.02618920363442,
      "grad_norm": 2.052896499633789,
      "learning_rate": 6.831686585062467e-05,
      "loss": 1.2409,
      "step": 22550
    },
    {
      "epoch": 6.028861571352218,
      "grad_norm": 1.8653088808059692,
      "learning_rate": 6.823724731583307e-05,
      "loss": 1.2823,
      "step": 22560
    },
    {
      "epoch": 6.031533939070016,
      "grad_norm": 1.7768802642822266,
      "learning_rate": 6.815765116878307e-05,
      "loss": 1.322,
      "step": 22570
    },
    {
      "epoch": 6.034206306787814,
      "grad_norm": 2.0534133911132812,
      "learning_rate": 6.807807746557754e-05,
      "loss": 1.2878,
      "step": 22580
    },
    {
      "epoch": 6.036878674505612,
      "grad_norm": 1.9299341440200806,
      "learning_rate": 6.799852626230332e-05,
      "loss": 1.2697,
      "step": 22590
    },
    {
      "epoch": 6.03955104222341,
      "grad_norm": 1.8582969903945923,
      "learning_rate": 6.791899761503153e-05,
      "loss": 1.3149,
      "step": 22600
    },
    {
      "epoch": 6.042223409941208,
      "grad_norm": 1.945414423942566,
      "learning_rate": 6.783949157981735e-05,
      "loss": 1.3075,
      "step": 22610
    },
    {
      "epoch": 6.044895777659006,
      "grad_norm": 1.971627950668335,
      "learning_rate": 6.776000821270001e-05,
      "loss": 1.2919,
      "step": 22620
    },
    {
      "epoch": 6.047568145376804,
      "grad_norm": 1.818709135055542,
      "learning_rate": 6.768054756970279e-05,
      "loss": 1.2471,
      "step": 22630
    },
    {
      "epoch": 6.050240513094602,
      "grad_norm": 1.9002695083618164,
      "learning_rate": 6.760110970683288e-05,
      "loss": 1.202,
      "step": 22640
    },
    {
      "epoch": 6.0529128808124,
      "grad_norm": 1.9322364330291748,
      "learning_rate": 6.752169468008153e-05,
      "loss": 1.2837,
      "step": 22650
    },
    {
      "epoch": 6.055585248530198,
      "grad_norm": 1.9187424182891846,
      "learning_rate": 6.744230254542386e-05,
      "loss": 1.2427,
      "step": 22660
    },
    {
      "epoch": 6.058257616247996,
      "grad_norm": 1.9976487159729004,
      "learning_rate": 6.736293335881874e-05,
      "loss": 1.2721,
      "step": 22670
    },
    {
      "epoch": 6.060929983965794,
      "grad_norm": 1.9472683668136597,
      "learning_rate": 6.728358717620903e-05,
      "loss": 1.3635,
      "step": 22680
    },
    {
      "epoch": 6.063602351683592,
      "grad_norm": 1.8906091451644897,
      "learning_rate": 6.720426405352129e-05,
      "loss": 1.2698,
      "step": 22690
    },
    {
      "epoch": 6.06627471940139,
      "grad_norm": 2.0350935459136963,
      "learning_rate": 6.71249640466658e-05,
      "loss": 1.2213,
      "step": 22700
    },
    {
      "epoch": 6.068947087119188,
      "grad_norm": 1.9788485765457153,
      "learning_rate": 6.704568721153662e-05,
      "loss": 1.2135,
      "step": 22710
    },
    {
      "epoch": 6.071619454836986,
      "grad_norm": 2.0295591354370117,
      "learning_rate": 6.696643360401144e-05,
      "loss": 1.2765,
      "step": 22720
    },
    {
      "epoch": 6.074291822554784,
      "grad_norm": 1.9607418775558472,
      "learning_rate": 6.688720327995153e-05,
      "loss": 1.2846,
      "step": 22730
    },
    {
      "epoch": 6.076964190272582,
      "grad_norm": 2.017274856567383,
      "learning_rate": 6.680799629520187e-05,
      "loss": 1.3168,
      "step": 22740
    },
    {
      "epoch": 6.079636557990379,
      "grad_norm": 1.886898398399353,
      "learning_rate": 6.672881270559086e-05,
      "loss": 1.3313,
      "step": 22750
    },
    {
      "epoch": 6.082308925708177,
      "grad_norm": 1.9333875179290771,
      "learning_rate": 6.664965256693052e-05,
      "loss": 1.2698,
      "step": 22760
    },
    {
      "epoch": 6.084981293425975,
      "grad_norm": 1.9045292139053345,
      "learning_rate": 6.657051593501629e-05,
      "loss": 1.2806,
      "step": 22770
    },
    {
      "epoch": 6.087653661143773,
      "grad_norm": 1.9250677824020386,
      "learning_rate": 6.649140286562696e-05,
      "loss": 1.2848,
      "step": 22780
    },
    {
      "epoch": 6.090326028861571,
      "grad_norm": 1.9216147661209106,
      "learning_rate": 6.641231341452487e-05,
      "loss": 1.2887,
      "step": 22790
    },
    {
      "epoch": 6.092998396579369,
      "grad_norm": 1.9174996614456177,
      "learning_rate": 6.633324763745563e-05,
      "loss": 1.3431,
      "step": 22800
    },
    {
      "epoch": 6.095670764297167,
      "grad_norm": 1.8849976062774658,
      "learning_rate": 6.625420559014812e-05,
      "loss": 1.315,
      "step": 22810
    },
    {
      "epoch": 6.098343132014965,
      "grad_norm": 1.927781343460083,
      "learning_rate": 6.61751873283146e-05,
      "loss": 1.2483,
      "step": 22820
    },
    {
      "epoch": 6.101015499732763,
      "grad_norm": 1.9553791284561157,
      "learning_rate": 6.609619290765045e-05,
      "loss": 1.1845,
      "step": 22830
    },
    {
      "epoch": 6.103687867450561,
      "grad_norm": 2.073179244995117,
      "learning_rate": 6.601722238383437e-05,
      "loss": 1.2956,
      "step": 22840
    },
    {
      "epoch": 6.106360235168359,
      "grad_norm": 1.7539079189300537,
      "learning_rate": 6.593827581252812e-05,
      "loss": 1.2073,
      "step": 22850
    },
    {
      "epoch": 6.109032602886157,
      "grad_norm": 2.009506940841675,
      "learning_rate": 6.585935324937656e-05,
      "loss": 1.3296,
      "step": 22860
    },
    {
      "epoch": 6.111704970603955,
      "grad_norm": 1.9820828437805176,
      "learning_rate": 6.578045475000778e-05,
      "loss": 1.2751,
      "step": 22870
    },
    {
      "epoch": 6.114377338321753,
      "grad_norm": 2.021517515182495,
      "learning_rate": 6.570158037003273e-05,
      "loss": 1.3532,
      "step": 22880
    },
    {
      "epoch": 6.117049706039551,
      "grad_norm": 2.0773496627807617,
      "learning_rate": 6.562273016504541e-05,
      "loss": 1.3231,
      "step": 22890
    },
    {
      "epoch": 6.119722073757349,
      "grad_norm": 2.197002410888672,
      "learning_rate": 6.554390419062288e-05,
      "loss": 1.2714,
      "step": 22900
    },
    {
      "epoch": 6.122394441475147,
      "grad_norm": 1.852329134941101,
      "learning_rate": 6.546510250232498e-05,
      "loss": 1.3469,
      "step": 22910
    },
    {
      "epoch": 6.125066809192945,
      "grad_norm": 1.8561760187149048,
      "learning_rate": 6.538632515569457e-05,
      "loss": 1.3371,
      "step": 22920
    },
    {
      "epoch": 6.127739176910743,
      "grad_norm": 1.9889413118362427,
      "learning_rate": 6.530757220625723e-05,
      "loss": 1.2938,
      "step": 22930
    },
    {
      "epoch": 6.130411544628541,
      "grad_norm": 1.8642148971557617,
      "learning_rate": 6.522884370952136e-05,
      "loss": 1.2297,
      "step": 22940
    },
    {
      "epoch": 6.133083912346339,
      "grad_norm": 2.2356655597686768,
      "learning_rate": 6.515013972097825e-05,
      "loss": 1.2885,
      "step": 22950
    },
    {
      "epoch": 6.135756280064137,
      "grad_norm": 1.9197638034820557,
      "learning_rate": 6.507146029610179e-05,
      "loss": 1.2557,
      "step": 22960
    },
    {
      "epoch": 6.138428647781935,
      "grad_norm": 2.064286231994629,
      "learning_rate": 6.499280549034855e-05,
      "loss": 1.3098,
      "step": 22970
    },
    {
      "epoch": 6.141101015499733,
      "grad_norm": 1.8423385620117188,
      "learning_rate": 6.491417535915783e-05,
      "loss": 1.2579,
      "step": 22980
    },
    {
      "epoch": 6.143773383217531,
      "grad_norm": 1.8803256750106812,
      "learning_rate": 6.483556995795147e-05,
      "loss": 1.3517,
      "step": 22990
    },
    {
      "epoch": 6.146445750935329,
      "grad_norm": 2.0200650691986084,
      "learning_rate": 6.475698934213395e-05,
      "loss": 1.2232,
      "step": 23000
    },
    {
      "epoch": 6.149118118653127,
      "grad_norm": 2.00144100189209,
      "learning_rate": 6.467843356709221e-05,
      "loss": 1.3506,
      "step": 23010
    },
    {
      "epoch": 6.151790486370925,
      "grad_norm": 1.9935312271118164,
      "learning_rate": 6.459990268819567e-05,
      "loss": 1.244,
      "step": 23020
    },
    {
      "epoch": 6.154462854088723,
      "grad_norm": 1.9040898084640503,
      "learning_rate": 6.452139676079631e-05,
      "loss": 1.3473,
      "step": 23030
    },
    {
      "epoch": 6.157135221806521,
      "grad_norm": 2.008814811706543,
      "learning_rate": 6.444291584022843e-05,
      "loss": 1.2708,
      "step": 23040
    },
    {
      "epoch": 6.159807589524319,
      "grad_norm": 1.8682596683502197,
      "learning_rate": 6.436445998180866e-05,
      "loss": 1.2746,
      "step": 23050
    },
    {
      "epoch": 6.162479957242116,
      "grad_norm": 1.8754842281341553,
      "learning_rate": 6.428602924083615e-05,
      "loss": 1.248,
      "step": 23060
    },
    {
      "epoch": 6.165152324959914,
      "grad_norm": 2.0143139362335205,
      "learning_rate": 6.420762367259213e-05,
      "loss": 1.3027,
      "step": 23070
    },
    {
      "epoch": 6.167824692677712,
      "grad_norm": 1.872719645500183,
      "learning_rate": 6.412924333234026e-05,
      "loss": 1.2795,
      "step": 23080
    },
    {
      "epoch": 6.17049706039551,
      "grad_norm": 2.1277658939361572,
      "learning_rate": 6.40508882753263e-05,
      "loss": 1.2405,
      "step": 23090
    },
    {
      "epoch": 6.173169428113308,
      "grad_norm": 1.7047022581100464,
      "learning_rate": 6.397255855677821e-05,
      "loss": 1.2908,
      "step": 23100
    },
    {
      "epoch": 6.175841795831106,
      "grad_norm": 1.8312227725982666,
      "learning_rate": 6.389425423190619e-05,
      "loss": 1.3096,
      "step": 23110
    },
    {
      "epoch": 6.178514163548904,
      "grad_norm": 2.1834187507629395,
      "learning_rate": 6.381597535590242e-05,
      "loss": 1.2477,
      "step": 23120
    },
    {
      "epoch": 6.181186531266702,
      "grad_norm": 2.458266496658325,
      "learning_rate": 6.373772198394114e-05,
      "loss": 1.2699,
      "step": 23130
    },
    {
      "epoch": 6.1838588989845,
      "grad_norm": 2.0120556354522705,
      "learning_rate": 6.365949417117876e-05,
      "loss": 1.2854,
      "step": 23140
    },
    {
      "epoch": 6.186531266702298,
      "grad_norm": 2.13199782371521,
      "learning_rate": 6.358129197275348e-05,
      "loss": 1.3501,
      "step": 23150
    },
    {
      "epoch": 6.189203634420096,
      "grad_norm": 2.2219533920288086,
      "learning_rate": 6.350311544378564e-05,
      "loss": 1.2776,
      "step": 23160
    },
    {
      "epoch": 6.191876002137894,
      "grad_norm": 2.0801823139190674,
      "learning_rate": 6.342496463937731e-05,
      "loss": 1.3194,
      "step": 23170
    },
    {
      "epoch": 6.194548369855692,
      "grad_norm": 1.946763515472412,
      "learning_rate": 6.334683961461252e-05,
      "loss": 1.3422,
      "step": 23180
    },
    {
      "epoch": 6.19722073757349,
      "grad_norm": 1.9578254222869873,
      "learning_rate": 6.326874042455715e-05,
      "loss": 1.3403,
      "step": 23190
    },
    {
      "epoch": 6.199893105291288,
      "grad_norm": 1.91990327835083,
      "learning_rate": 6.319066712425883e-05,
      "loss": 1.2727,
      "step": 23200
    },
    {
      "epoch": 6.202565473009086,
      "grad_norm": 1.915351152420044,
      "learning_rate": 6.311261976874689e-05,
      "loss": 1.2846,
      "step": 23210
    },
    {
      "epoch": 6.205237840726884,
      "grad_norm": 1.8709057569503784,
      "learning_rate": 6.303459841303253e-05,
      "loss": 1.2895,
      "step": 23220
    },
    {
      "epoch": 6.207910208444682,
      "grad_norm": 1.9091672897338867,
      "learning_rate": 6.295660311210843e-05,
      "loss": 1.295,
      "step": 23230
    },
    {
      "epoch": 6.21058257616248,
      "grad_norm": 1.864214539527893,
      "learning_rate": 6.287863392094908e-05,
      "loss": 1.3519,
      "step": 23240
    },
    {
      "epoch": 6.213254943880278,
      "grad_norm": 2.014496088027954,
      "learning_rate": 6.280069089451046e-05,
      "loss": 1.3719,
      "step": 23250
    },
    {
      "epoch": 6.215927311598076,
      "grad_norm": 2.009049892425537,
      "learning_rate": 6.272277408773009e-05,
      "loss": 1.2893,
      "step": 23260
    },
    {
      "epoch": 6.218599679315874,
      "grad_norm": 1.9774397611618042,
      "learning_rate": 6.264488355552714e-05,
      "loss": 1.2778,
      "step": 23270
    },
    {
      "epoch": 6.221272047033672,
      "grad_norm": 2.062709093093872,
      "learning_rate": 6.256701935280214e-05,
      "loss": 1.3393,
      "step": 23280
    },
    {
      "epoch": 6.22394441475147,
      "grad_norm": 2.0006885528564453,
      "learning_rate": 6.248918153443705e-05,
      "loss": 1.3484,
      "step": 23290
    },
    {
      "epoch": 6.226616782469268,
      "grad_norm": 2.1000020503997803,
      "learning_rate": 6.241137015529537e-05,
      "loss": 1.3144,
      "step": 23300
    },
    {
      "epoch": 6.229289150187066,
      "grad_norm": 1.9920177459716797,
      "learning_rate": 6.233358527022177e-05,
      "loss": 1.3562,
      "step": 23310
    },
    {
      "epoch": 6.231961517904864,
      "grad_norm": 2.069359540939331,
      "learning_rate": 6.225582693404249e-05,
      "loss": 1.3087,
      "step": 23320
    },
    {
      "epoch": 6.234633885622662,
      "grad_norm": 1.9647448062896729,
      "learning_rate": 6.217809520156481e-05,
      "loss": 1.2882,
      "step": 23330
    },
    {
      "epoch": 6.23730625334046,
      "grad_norm": 1.8878511190414429,
      "learning_rate": 6.210039012757738e-05,
      "loss": 1.3485,
      "step": 23340
    },
    {
      "epoch": 6.239978621058258,
      "grad_norm": 1.9968006610870361,
      "learning_rate": 6.20227117668501e-05,
      "loss": 1.3109,
      "step": 23350
    },
    {
      "epoch": 6.242650988776056,
      "grad_norm": 1.8300390243530273,
      "learning_rate": 6.194506017413392e-05,
      "loss": 1.3033,
      "step": 23360
    },
    {
      "epoch": 6.245323356493854,
      "grad_norm": 1.9156193733215332,
      "learning_rate": 6.186743540416104e-05,
      "loss": 1.252,
      "step": 23370
    },
    {
      "epoch": 6.247995724211652,
      "grad_norm": 2.013760805130005,
      "learning_rate": 6.178983751164471e-05,
      "loss": 1.3234,
      "step": 23380
    },
    {
      "epoch": 6.25066809192945,
      "grad_norm": 1.9752377271652222,
      "learning_rate": 6.171226655127916e-05,
      "loss": 1.333,
      "step": 23390
    },
    {
      "epoch": 6.253340459647247,
      "grad_norm": 1.9661108255386353,
      "learning_rate": 6.163472257773982e-05,
      "loss": 1.2996,
      "step": 23400
    },
    {
      "epoch": 6.256012827365045,
      "grad_norm": 1.981852412223816,
      "learning_rate": 6.155720564568293e-05,
      "loss": 1.3432,
      "step": 23410
    },
    {
      "epoch": 6.258685195082843,
      "grad_norm": 2.1153745651245117,
      "learning_rate": 6.147971580974568e-05,
      "loss": 1.3372,
      "step": 23420
    },
    {
      "epoch": 6.261357562800641,
      "grad_norm": 2.1223621368408203,
      "learning_rate": 6.140225312454629e-05,
      "loss": 1.319,
      "step": 23430
    },
    {
      "epoch": 6.264029930518439,
      "grad_norm": 1.8627023696899414,
      "learning_rate": 6.132481764468373e-05,
      "loss": 1.3023,
      "step": 23440
    },
    {
      "epoch": 6.266702298236237,
      "grad_norm": 1.9366716146469116,
      "learning_rate": 6.124740942473778e-05,
      "loss": 1.2827,
      "step": 23450
    },
    {
      "epoch": 6.269374665954035,
      "grad_norm": 2.0096912384033203,
      "learning_rate": 6.117002851926914e-05,
      "loss": 1.3612,
      "step": 23460
    },
    {
      "epoch": 6.272047033671833,
      "grad_norm": 2.025266170501709,
      "learning_rate": 6.109267498281909e-05,
      "loss": 1.3672,
      "step": 23470
    },
    {
      "epoch": 6.274719401389631,
      "grad_norm": 2.2905125617980957,
      "learning_rate": 6.101534886990976e-05,
      "loss": 1.303,
      "step": 23480
    },
    {
      "epoch": 6.277391769107429,
      "grad_norm": 1.912898063659668,
      "learning_rate": 6.0938050235043886e-05,
      "loss": 1.2745,
      "step": 23490
    },
    {
      "epoch": 6.280064136825227,
      "grad_norm": 1.9694336652755737,
      "learning_rate": 6.086077913270476e-05,
      "loss": 1.4124,
      "step": 23500
    },
    {
      "epoch": 6.282736504543025,
      "grad_norm": 2.0188982486724854,
      "learning_rate": 6.0783535617356434e-05,
      "loss": 1.3102,
      "step": 23510
    },
    {
      "epoch": 6.285408872260823,
      "grad_norm": 2.0003912448883057,
      "learning_rate": 6.070631974344343e-05,
      "loss": 1.307,
      "step": 23520
    },
    {
      "epoch": 6.288081239978621,
      "grad_norm": 1.8425689935684204,
      "learning_rate": 6.0629131565390715e-05,
      "loss": 1.2732,
      "step": 23530
    },
    {
      "epoch": 6.290753607696419,
      "grad_norm": 1.9907017946243286,
      "learning_rate": 6.055197113760385e-05,
      "loss": 1.3561,
      "step": 23540
    },
    {
      "epoch": 6.293425975414217,
      "grad_norm": 1.8586206436157227,
      "learning_rate": 6.04748385144688e-05,
      "loss": 1.3121,
      "step": 23550
    },
    {
      "epoch": 6.296098343132015,
      "grad_norm": 1.839750051498413,
      "learning_rate": 6.0397733750351925e-05,
      "loss": 1.3793,
      "step": 23560
    },
    {
      "epoch": 6.298770710849813,
      "grad_norm": 1.9381271600723267,
      "learning_rate": 6.0320656899599934e-05,
      "loss": 1.3002,
      "step": 23570
    },
    {
      "epoch": 6.301443078567611,
      "grad_norm": 1.955628514289856,
      "learning_rate": 6.024360801653985e-05,
      "loss": 1.2719,
      "step": 23580
    },
    {
      "epoch": 6.304115446285409,
      "grad_norm": 1.8711755275726318,
      "learning_rate": 6.016658715547909e-05,
      "loss": 1.3457,
      "step": 23590
    },
    {
      "epoch": 6.306787814003207,
      "grad_norm": 1.9570006132125854,
      "learning_rate": 6.008959437070518e-05,
      "loss": 1.3249,
      "step": 23600
    },
    {
      "epoch": 6.309460181721005,
      "grad_norm": 2.015333652496338,
      "learning_rate": 6.001262971648589e-05,
      "loss": 1.2877,
      "step": 23610
    },
    {
      "epoch": 6.312132549438803,
      "grad_norm": 2.026175022125244,
      "learning_rate": 5.993569324706925e-05,
      "loss": 1.29,
      "step": 23620
    },
    {
      "epoch": 6.314804917156601,
      "grad_norm": 1.9941869974136353,
      "learning_rate": 5.9858785016683306e-05,
      "loss": 1.3553,
      "step": 23630
    },
    {
      "epoch": 6.317477284874399,
      "grad_norm": 1.9505996704101562,
      "learning_rate": 5.9781905079536324e-05,
      "loss": 1.369,
      "step": 23640
    },
    {
      "epoch": 6.320149652592197,
      "grad_norm": 2.053548812866211,
      "learning_rate": 5.970505348981652e-05,
      "loss": 1.3175,
      "step": 23650
    },
    {
      "epoch": 6.322822020309995,
      "grad_norm": 1.9613093137741089,
      "learning_rate": 5.962823030169213e-05,
      "loss": 1.335,
      "step": 23660
    },
    {
      "epoch": 6.325494388027793,
      "grad_norm": 2.036151647567749,
      "learning_rate": 5.955143556931149e-05,
      "loss": 1.3233,
      "step": 23670
    },
    {
      "epoch": 6.328166755745591,
      "grad_norm": 2.0008156299591064,
      "learning_rate": 5.947466934680277e-05,
      "loss": 1.274,
      "step": 23680
    },
    {
      "epoch": 6.330839123463389,
      "grad_norm": 1.8678631782531738,
      "learning_rate": 5.9397931688274023e-05,
      "loss": 1.2367,
      "step": 23690
    },
    {
      "epoch": 6.333511491181187,
      "grad_norm": 1.9700605869293213,
      "learning_rate": 5.93212226478133e-05,
      "loss": 1.3093,
      "step": 23700
    },
    {
      "epoch": 6.336183858898984,
      "grad_norm": 1.8679239749908447,
      "learning_rate": 5.9244542279488346e-05,
      "loss": 1.326,
      "step": 23710
    },
    {
      "epoch": 6.338856226616782,
      "grad_norm": 1.9873723983764648,
      "learning_rate": 5.91678906373468e-05,
      "loss": 1.2827,
      "step": 23720
    },
    {
      "epoch": 6.34152859433458,
      "grad_norm": 1.9908934831619263,
      "learning_rate": 5.909126777541596e-05,
      "loss": 1.3772,
      "step": 23730
    },
    {
      "epoch": 6.344200962052378,
      "grad_norm": 1.9041279554367065,
      "learning_rate": 5.9014673747702886e-05,
      "loss": 1.3175,
      "step": 23740
    },
    {
      "epoch": 6.346873329770176,
      "grad_norm": 2.04715895652771,
      "learning_rate": 5.8938108608194355e-05,
      "loss": 1.3249,
      "step": 23750
    },
    {
      "epoch": 6.349545697487974,
      "grad_norm": 1.8260263204574585,
      "learning_rate": 5.886157241085672e-05,
      "loss": 1.3476,
      "step": 23760
    },
    {
      "epoch": 6.352218065205772,
      "grad_norm": 1.9012395143508911,
      "learning_rate": 5.8785065209635916e-05,
      "loss": 1.3144,
      "step": 23770
    },
    {
      "epoch": 6.35489043292357,
      "grad_norm": 1.9461390972137451,
      "learning_rate": 5.8708587058457545e-05,
      "loss": 1.2653,
      "step": 23780
    },
    {
      "epoch": 6.357562800641368,
      "grad_norm": 2.0778017044067383,
      "learning_rate": 5.863213801122659e-05,
      "loss": 1.287,
      "step": 23790
    },
    {
      "epoch": 6.360235168359166,
      "grad_norm": 1.932028889656067,
      "learning_rate": 5.855571812182769e-05,
      "loss": 1.2953,
      "step": 23800
    },
    {
      "epoch": 6.362907536076964,
      "grad_norm": 2.24345326423645,
      "learning_rate": 5.847932744412477e-05,
      "loss": 1.3015,
      "step": 23810
    },
    {
      "epoch": 6.365579903794762,
      "grad_norm": 1.9071317911148071,
      "learning_rate": 5.840296603196124e-05,
      "loss": 1.3571,
      "step": 23820
    },
    {
      "epoch": 6.36825227151256,
      "grad_norm": 1.827386736869812,
      "learning_rate": 5.832663393915991e-05,
      "loss": 1.2797,
      "step": 23830
    },
    {
      "epoch": 6.370924639230358,
      "grad_norm": 1.8914283514022827,
      "learning_rate": 5.8250331219522925e-05,
      "loss": 1.313,
      "step": 23840
    },
    {
      "epoch": 6.373597006948156,
      "grad_norm": 1.8569040298461914,
      "learning_rate": 5.817405792683155e-05,
      "loss": 1.3113,
      "step": 23850
    },
    {
      "epoch": 6.376269374665954,
      "grad_norm": 1.8708868026733398,
      "learning_rate": 5.809781411484662e-05,
      "loss": 1.2918,
      "step": 23860
    },
    {
      "epoch": 6.378941742383752,
      "grad_norm": 1.834140419960022,
      "learning_rate": 5.802159983730786e-05,
      "loss": 1.3398,
      "step": 23870
    },
    {
      "epoch": 6.38161411010155,
      "grad_norm": 1.9090319871902466,
      "learning_rate": 5.7945415147934525e-05,
      "loss": 1.336,
      "step": 23880
    },
    {
      "epoch": 6.384286477819348,
      "grad_norm": 2.0437538623809814,
      "learning_rate": 5.786926010042468e-05,
      "loss": 1.3584,
      "step": 23890
    },
    {
      "epoch": 6.386958845537146,
      "grad_norm": 2.0860626697540283,
      "learning_rate": 5.7793134748455714e-05,
      "loss": 1.3766,
      "step": 23900
    },
    {
      "epoch": 6.389631213254944,
      "grad_norm": 1.9386303424835205,
      "learning_rate": 5.7717039145684005e-05,
      "loss": 1.2482,
      "step": 23910
    },
    {
      "epoch": 6.392303580972742,
      "grad_norm": 2.092529773712158,
      "learning_rate": 5.764097334574503e-05,
      "loss": 1.336,
      "step": 23920
    },
    {
      "epoch": 6.39497594869054,
      "grad_norm": 2.0294992923736572,
      "learning_rate": 5.756493740225306e-05,
      "loss": 1.3181,
      "step": 23930
    },
    {
      "epoch": 6.397648316408338,
      "grad_norm": 1.9609752893447876,
      "learning_rate": 5.7488931368801644e-05,
      "loss": 1.336,
      "step": 23940
    },
    {
      "epoch": 6.400320684126136,
      "grad_norm": 2.144275426864624,
      "learning_rate": 5.741295529896295e-05,
      "loss": 1.3315,
      "step": 23950
    },
    {
      "epoch": 6.402993051843934,
      "grad_norm": 2.0646297931671143,
      "learning_rate": 5.733700924628821e-05,
      "loss": 1.3433,
      "step": 23960
    },
    {
      "epoch": 6.405665419561732,
      "grad_norm": 1.8610888719558716,
      "learning_rate": 5.7261093264307416e-05,
      "loss": 1.2768,
      "step": 23970
    },
    {
      "epoch": 6.40833778727953,
      "grad_norm": 1.857261061668396,
      "learning_rate": 5.7185207406529386e-05,
      "loss": 1.2985,
      "step": 23980
    },
    {
      "epoch": 6.411010154997328,
      "grad_norm": 1.9301776885986328,
      "learning_rate": 5.710935172644175e-05,
      "loss": 1.3038,
      "step": 23990
    },
    {
      "epoch": 6.413682522715126,
      "grad_norm": 2.1339030265808105,
      "learning_rate": 5.7033526277510755e-05,
      "loss": 1.2281,
      "step": 24000
    },
    {
      "epoch": 6.416354890432924,
      "grad_norm": 1.8557411432266235,
      "learning_rate": 5.6957731113181433e-05,
      "loss": 1.3243,
      "step": 24010
    },
    {
      "epoch": 6.419027258150722,
      "grad_norm": 1.8736084699630737,
      "learning_rate": 5.688196628687745e-05,
      "loss": 1.3724,
      "step": 24020
    },
    {
      "epoch": 6.42169962586852,
      "grad_norm": 2.0215272903442383,
      "learning_rate": 5.6806231852001104e-05,
      "loss": 1.2903,
      "step": 24030
    },
    {
      "epoch": 6.424371993586318,
      "grad_norm": 2.0285580158233643,
      "learning_rate": 5.673052786193323e-05,
      "loss": 1.4089,
      "step": 24040
    },
    {
      "epoch": 6.427044361304115,
      "grad_norm": 1.9792866706848145,
      "learning_rate": 5.665485437003326e-05,
      "loss": 1.3206,
      "step": 24050
    },
    {
      "epoch": 6.429716729021913,
      "grad_norm": 2.078568935394287,
      "learning_rate": 5.6579211429639e-05,
      "loss": 1.3477,
      "step": 24060
    },
    {
      "epoch": 6.432389096739711,
      "grad_norm": 1.9913173913955688,
      "learning_rate": 5.650359909406695e-05,
      "loss": 1.3042,
      "step": 24070
    },
    {
      "epoch": 6.435061464457509,
      "grad_norm": 1.9369879961013794,
      "learning_rate": 5.642801741661181e-05,
      "loss": 1.36,
      "step": 24080
    },
    {
      "epoch": 6.437733832175307,
      "grad_norm": 1.9501110315322876,
      "learning_rate": 5.6352466450546806e-05,
      "loss": 1.4001,
      "step": 24090
    },
    {
      "epoch": 6.440406199893105,
      "grad_norm": 1.8964717388153076,
      "learning_rate": 5.627694624912347e-05,
      "loss": 1.3047,
      "step": 24100
    },
    {
      "epoch": 6.443078567610903,
      "grad_norm": 2.044865846633911,
      "learning_rate": 5.620145686557165e-05,
      "loss": 1.3673,
      "step": 24110
    },
    {
      "epoch": 6.445750935328701,
      "grad_norm": 2.040405750274658,
      "learning_rate": 5.612599835309949e-05,
      "loss": 1.3436,
      "step": 24120
    },
    {
      "epoch": 6.448423303046499,
      "grad_norm": 1.9377208948135376,
      "learning_rate": 5.605057076489342e-05,
      "loss": 1.348,
      "step": 24130
    },
    {
      "epoch": 6.451095670764297,
      "grad_norm": 1.9589042663574219,
      "learning_rate": 5.597517415411792e-05,
      "loss": 1.2912,
      "step": 24140
    },
    {
      "epoch": 6.453768038482095,
      "grad_norm": 1.9442590475082397,
      "learning_rate": 5.589980857391577e-05,
      "loss": 1.2581,
      "step": 24150
    },
    {
      "epoch": 6.456440406199893,
      "grad_norm": 1.9758940935134888,
      "learning_rate": 5.582447407740785e-05,
      "loss": 1.3421,
      "step": 24160
    },
    {
      "epoch": 6.459112773917691,
      "grad_norm": 2.0502991676330566,
      "learning_rate": 5.574917071769313e-05,
      "loss": 1.2519,
      "step": 24170
    },
    {
      "epoch": 6.461785141635489,
      "grad_norm": 2.0679097175598145,
      "learning_rate": 5.5673898547848656e-05,
      "loss": 1.3707,
      "step": 24180
    },
    {
      "epoch": 6.464457509353287,
      "grad_norm": 2.0154318809509277,
      "learning_rate": 5.559865762092935e-05,
      "loss": 1.2755,
      "step": 24190
    },
    {
      "epoch": 6.467129877071085,
      "grad_norm": 2.023235559463501,
      "learning_rate": 5.552344798996838e-05,
      "loss": 1.3422,
      "step": 24200
    },
    {
      "epoch": 6.469802244788883,
      "grad_norm": 1.8802125453948975,
      "learning_rate": 5.544826970797658e-05,
      "loss": 1.2408,
      "step": 24210
    },
    {
      "epoch": 6.472474612506681,
      "grad_norm": 2.142982006072998,
      "learning_rate": 5.5373122827942844e-05,
      "loss": 1.3135,
      "step": 24220
    },
    {
      "epoch": 6.475146980224479,
      "grad_norm": 1.9517136812210083,
      "learning_rate": 5.5298007402833905e-05,
      "loss": 1.2831,
      "step": 24230
    },
    {
      "epoch": 6.477819347942277,
      "grad_norm": 1.7909530401229858,
      "learning_rate": 5.522292348559434e-05,
      "loss": 1.2253,
      "step": 24240
    },
    {
      "epoch": 6.480491715660075,
      "grad_norm": 1.970603585243225,
      "learning_rate": 5.5147871129146377e-05,
      "loss": 1.3396,
      "step": 24250
    },
    {
      "epoch": 6.483164083377873,
      "grad_norm": 1.9476258754730225,
      "learning_rate": 5.5072850386390283e-05,
      "loss": 1.3097,
      "step": 24260
    },
    {
      "epoch": 6.485836451095671,
      "grad_norm": 1.9174326658248901,
      "learning_rate": 5.49978613102037e-05,
      "loss": 1.2603,
      "step": 24270
    },
    {
      "epoch": 6.488508818813469,
      "grad_norm": 1.8014626502990723,
      "learning_rate": 5.492290395344232e-05,
      "loss": 1.3148,
      "step": 24280
    },
    {
      "epoch": 6.491181186531267,
      "grad_norm": 2.1735575199127197,
      "learning_rate": 5.4847978368939126e-05,
      "loss": 1.3427,
      "step": 24290
    },
    {
      "epoch": 6.493853554249065,
      "grad_norm": 1.9877837896347046,
      "learning_rate": 5.477308460950491e-05,
      "loss": 1.2973,
      "step": 24300
    },
    {
      "epoch": 6.496525921966863,
      "grad_norm": 1.9472131729125977,
      "learning_rate": 5.4698222727927997e-05,
      "loss": 1.353,
      "step": 24310
    },
    {
      "epoch": 6.499198289684661,
      "grad_norm": 1.8456213474273682,
      "learning_rate": 5.462339277697426e-05,
      "loss": 1.3024,
      "step": 24320
    },
    {
      "epoch": 6.501870657402459,
      "grad_norm": 1.9842456579208374,
      "learning_rate": 5.454859480938692e-05,
      "loss": 1.3093,
      "step": 24330
    },
    {
      "epoch": 6.504543025120256,
      "grad_norm": 1.9452012777328491,
      "learning_rate": 5.4473828877886925e-05,
      "loss": 1.2748,
      "step": 24340
    },
    {
      "epoch": 6.507215392838054,
      "grad_norm": 2.105241537094116,
      "learning_rate": 5.43990950351724e-05,
      "loss": 1.3571,
      "step": 24350
    },
    {
      "epoch": 6.509887760555852,
      "grad_norm": 2.0901341438293457,
      "learning_rate": 5.432439333391895e-05,
      "loss": 1.4147,
      "step": 24360
    },
    {
      "epoch": 6.51256012827365,
      "grad_norm": 2.008657693862915,
      "learning_rate": 5.4249723826779556e-05,
      "loss": 1.397,
      "step": 24370
    },
    {
      "epoch": 6.515232495991448,
      "grad_norm": 2.063403606414795,
      "learning_rate": 5.417508656638437e-05,
      "loss": 1.3559,
      "step": 24380
    },
    {
      "epoch": 6.517904863709246,
      "grad_norm": 2.0258562564849854,
      "learning_rate": 5.410048160534107e-05,
      "loss": 1.2944,
      "step": 24390
    },
    {
      "epoch": 6.520577231427044,
      "grad_norm": 1.9289439916610718,
      "learning_rate": 5.402590899623427e-05,
      "loss": 1.2823,
      "step": 24400
    },
    {
      "epoch": 6.523249599144842,
      "grad_norm": 2.064502716064453,
      "learning_rate": 5.395136879162599e-05,
      "loss": 1.3447,
      "step": 24410
    },
    {
      "epoch": 6.52592196686264,
      "grad_norm": 2.0392673015594482,
      "learning_rate": 5.3876861044055335e-05,
      "loss": 1.3871,
      "step": 24420
    },
    {
      "epoch": 6.528594334580438,
      "grad_norm": 1.8871740102767944,
      "learning_rate": 5.3802385806038535e-05,
      "loss": 1.2774,
      "step": 24430
    },
    {
      "epoch": 6.531266702298236,
      "grad_norm": 2.1257073879241943,
      "learning_rate": 5.3727943130068904e-05,
      "loss": 1.317,
      "step": 24440
    },
    {
      "epoch": 6.533939070016034,
      "grad_norm": 2.031369924545288,
      "learning_rate": 5.3653533068616866e-05,
      "loss": 1.2801,
      "step": 24450
    },
    {
      "epoch": 6.536611437733832,
      "grad_norm": 1.995954990386963,
      "learning_rate": 5.357915567412968e-05,
      "loss": 1.2943,
      "step": 24460
    },
    {
      "epoch": 6.53928380545163,
      "grad_norm": 1.9537169933319092,
      "learning_rate": 5.350481099903186e-05,
      "loss": 1.3021,
      "step": 24470
    },
    {
      "epoch": 6.541956173169428,
      "grad_norm": 1.887035846710205,
      "learning_rate": 5.3430499095724596e-05,
      "loss": 1.3474,
      "step": 24480
    },
    {
      "epoch": 6.544628540887226,
      "grad_norm": 1.9988093376159668,
      "learning_rate": 5.33562200165861e-05,
      "loss": 1.3242,
      "step": 24490
    },
    {
      "epoch": 6.547300908605024,
      "grad_norm": 1.9960969686508179,
      "learning_rate": 5.3281973813971444e-05,
      "loss": 1.3295,
      "step": 24500
    },
    {
      "epoch": 6.549973276322822,
      "grad_norm": 1.8658087253570557,
      "learning_rate": 5.320776054021253e-05,
      "loss": 1.3153,
      "step": 24510
    },
    {
      "epoch": 6.55264564404062,
      "grad_norm": 2.133042573928833,
      "learning_rate": 5.3133580247618e-05,
      "loss": 1.3661,
      "step": 24520
    },
    {
      "epoch": 6.555318011758418,
      "grad_norm": 1.898963451385498,
      "learning_rate": 5.305943298847336e-05,
      "loss": 1.381,
      "step": 24530
    },
    {
      "epoch": 6.557990379476216,
      "grad_norm": 2.0984890460968018,
      "learning_rate": 5.2985318815040653e-05,
      "loss": 1.3343,
      "step": 24540
    },
    {
      "epoch": 6.560662747194014,
      "grad_norm": 2.0512940883636475,
      "learning_rate": 5.2911237779558774e-05,
      "loss": 1.3343,
      "step": 24550
    },
    {
      "epoch": 6.563335114911812,
      "grad_norm": 2.725210428237915,
      "learning_rate": 5.283718993424317e-05,
      "loss": 1.3337,
      "step": 24560
    },
    {
      "epoch": 6.56600748262961,
      "grad_norm": 1.88321053981781,
      "learning_rate": 5.276317533128593e-05,
      "loss": 1.3841,
      "step": 24570
    },
    {
      "epoch": 6.568679850347408,
      "grad_norm": 2.130901336669922,
      "learning_rate": 5.268919402285573e-05,
      "loss": 1.3608,
      "step": 24580
    },
    {
      "epoch": 6.571352218065206,
      "grad_norm": 1.9648263454437256,
      "learning_rate": 5.261524606109764e-05,
      "loss": 1.316,
      "step": 24590
    },
    {
      "epoch": 6.574024585783004,
      "grad_norm": 1.9932940006256104,
      "learning_rate": 5.254133149813349e-05,
      "loss": 1.3343,
      "step": 24600
    },
    {
      "epoch": 6.576696953500802,
      "grad_norm": 2.0183959007263184,
      "learning_rate": 5.24674503860613e-05,
      "loss": 1.3807,
      "step": 24610
    },
    {
      "epoch": 6.5793693212186,
      "grad_norm": 1.8899110555648804,
      "learning_rate": 5.239360277695565e-05,
      "loss": 1.3112,
      "step": 24620
    },
    {
      "epoch": 6.582041688936398,
      "grad_norm": 2.061530113220215,
      "learning_rate": 5.23197887228675e-05,
      "loss": 1.3094,
      "step": 24630
    },
    {
      "epoch": 6.584714056654196,
      "grad_norm": 1.839147686958313,
      "learning_rate": 5.224600827582418e-05,
      "loss": 1.3192,
      "step": 24640
    },
    {
      "epoch": 6.587386424371994,
      "grad_norm": 1.986692190170288,
      "learning_rate": 5.2172261487829146e-05,
      "loss": 1.3544,
      "step": 24650
    },
    {
      "epoch": 6.590058792089792,
      "grad_norm": 1.9184010028839111,
      "learning_rate": 5.2098548410862466e-05,
      "loss": 1.3775,
      "step": 24660
    },
    {
      "epoch": 6.59273115980759,
      "grad_norm": 1.9746228456497192,
      "learning_rate": 5.202486909688009e-05,
      "loss": 1.3051,
      "step": 24670
    },
    {
      "epoch": 6.595403527525388,
      "grad_norm": 1.9751840829849243,
      "learning_rate": 5.19512235978145e-05,
      "loss": 1.3443,
      "step": 24680
    },
    {
      "epoch": 6.598075895243186,
      "grad_norm": 1.950573444366455,
      "learning_rate": 5.187761196557406e-05,
      "loss": 1.34,
      "step": 24690
    },
    {
      "epoch": 6.600748262960983,
      "grad_norm": 1.9397246837615967,
      "learning_rate": 5.180403425204342e-05,
      "loss": 1.2836,
      "step": 24700
    },
    {
      "epoch": 6.603420630678781,
      "grad_norm": 2.12785267829895,
      "learning_rate": 5.1730490509083296e-05,
      "loss": 1.3503,
      "step": 24710
    },
    {
      "epoch": 6.606092998396579,
      "grad_norm": 2.105128526687622,
      "learning_rate": 5.1656980788530484e-05,
      "loss": 1.3326,
      "step": 24720
    },
    {
      "epoch": 6.608765366114377,
      "grad_norm": 2.083244800567627,
      "learning_rate": 5.1583505142197694e-05,
      "loss": 1.3435,
      "step": 24730
    },
    {
      "epoch": 6.611437733832175,
      "grad_norm": 1.8611396551132202,
      "learning_rate": 5.151006362187374e-05,
      "loss": 1.38,
      "step": 24740
    },
    {
      "epoch": 6.614110101549973,
      "grad_norm": 2.061445951461792,
      "learning_rate": 5.143665627932331e-05,
      "loss": 1.2961,
      "step": 24750
    },
    {
      "epoch": 6.616782469267771,
      "grad_norm": 1.8973666429519653,
      "learning_rate": 5.136328316628706e-05,
      "loss": 1.3606,
      "step": 24760
    },
    {
      "epoch": 6.619454836985569,
      "grad_norm": 1.97553551197052,
      "learning_rate": 5.12899443344815e-05,
      "loss": 1.3811,
      "step": 24770
    },
    {
      "epoch": 6.622127204703367,
      "grad_norm": 1.9667669534683228,
      "learning_rate": 5.1216639835598856e-05,
      "loss": 1.3449,
      "step": 24780
    },
    {
      "epoch": 6.624799572421165,
      "grad_norm": 2.018714666366577,
      "learning_rate": 5.1143369721307424e-05,
      "loss": 1.2524,
      "step": 24790
    },
    {
      "epoch": 6.627471940138963,
      "grad_norm": 1.939206838607788,
      "learning_rate": 5.1070134043250983e-05,
      "loss": 1.3711,
      "step": 24800
    },
    {
      "epoch": 6.630144307856761,
      "grad_norm": 1.9365715980529785,
      "learning_rate": 5.0996932853049186e-05,
      "loss": 1.3017,
      "step": 24810
    },
    {
      "epoch": 6.632816675574559,
      "grad_norm": 2.0131990909576416,
      "learning_rate": 5.0923766202297376e-05,
      "loss": 1.3143,
      "step": 24820
    },
    {
      "epoch": 6.635489043292357,
      "grad_norm": 1.902485728263855,
      "learning_rate": 5.08506341425665e-05,
      "loss": 1.2985,
      "step": 24830
    },
    {
      "epoch": 6.638161411010155,
      "grad_norm": 2.0457568168640137,
      "learning_rate": 5.0777536725403175e-05,
      "loss": 1.2853,
      "step": 24840
    },
    {
      "epoch": 6.640833778727953,
      "grad_norm": 2.0161006450653076,
      "learning_rate": 5.070447400232958e-05,
      "loss": 1.2634,
      "step": 24850
    },
    {
      "epoch": 6.643506146445751,
      "grad_norm": 2.0461578369140625,
      "learning_rate": 5.0631446024843365e-05,
      "loss": 1.2905,
      "step": 24860
    },
    {
      "epoch": 6.646178514163549,
      "grad_norm": 1.9626697301864624,
      "learning_rate": 5.0558452844417895e-05,
      "loss": 1.3102,
      "step": 24870
    },
    {
      "epoch": 6.648850881881347,
      "grad_norm": 1.8565678596496582,
      "learning_rate": 5.048549451250173e-05,
      "loss": 1.2997,
      "step": 24880
    },
    {
      "epoch": 6.651523249599145,
      "grad_norm": 2.02064847946167,
      "learning_rate": 5.041257108051909e-05,
      "loss": 1.3787,
      "step": 24890
    },
    {
      "epoch": 6.654195617316943,
      "grad_norm": 2.0025739669799805,
      "learning_rate": 5.033968259986951e-05,
      "loss": 1.3616,
      "step": 24900
    },
    {
      "epoch": 6.656867985034741,
      "grad_norm": 1.908174991607666,
      "learning_rate": 5.026682912192788e-05,
      "loss": 1.3133,
      "step": 24910
    },
    {
      "epoch": 6.659540352752539,
      "grad_norm": 2.0004167556762695,
      "learning_rate": 5.019401069804448e-05,
      "loss": 1.3082,
      "step": 24920
    },
    {
      "epoch": 6.662212720470337,
      "grad_norm": 1.8949096202850342,
      "learning_rate": 5.0121227379544775e-05,
      "loss": 1.3418,
      "step": 24930
    },
    {
      "epoch": 6.664885088188135,
      "grad_norm": 2.18448543548584,
      "learning_rate": 5.004847921772957e-05,
      "loss": 1.3453,
      "step": 24940
    },
    {
      "epoch": 6.667557455905933,
      "grad_norm": 1.9995133876800537,
      "learning_rate": 4.997576626387489e-05,
      "loss": 1.3099,
      "step": 24950
    },
    {
      "epoch": 6.670229823623731,
      "grad_norm": 2.020721673965454,
      "learning_rate": 4.9903088569231904e-05,
      "loss": 1.3469,
      "step": 24960
    },
    {
      "epoch": 6.672902191341529,
      "grad_norm": 1.9147567749023438,
      "learning_rate": 4.983044618502696e-05,
      "loss": 1.3453,
      "step": 24970
    },
    {
      "epoch": 6.675574559059327,
      "grad_norm": 1.9366708993911743,
      "learning_rate": 4.975783916246154e-05,
      "loss": 1.2894,
      "step": 24980
    },
    {
      "epoch": 6.678246926777124,
      "grad_norm": 2.050410509109497,
      "learning_rate": 4.968526755271206e-05,
      "loss": 1.3283,
      "step": 24990
    },
    {
      "epoch": 6.680919294494922,
      "grad_norm": 1.9889848232269287,
      "learning_rate": 4.961273140693021e-05,
      "loss": 1.3503,
      "step": 25000
    },
    {
      "epoch": 6.68359166221272,
      "grad_norm": 2.047287940979004,
      "learning_rate": 4.9540230776242466e-05,
      "loss": 1.3312,
      "step": 25010
    },
    {
      "epoch": 6.686264029930518,
      "grad_norm": 1.980771541595459,
      "learning_rate": 4.9467765711750414e-05,
      "loss": 1.2966,
      "step": 25020
    },
    {
      "epoch": 6.688936397648316,
      "grad_norm": 1.8416842222213745,
      "learning_rate": 4.9395336264530486e-05,
      "loss": 1.3168,
      "step": 25030
    },
    {
      "epoch": 6.691608765366114,
      "grad_norm": 2.0462563037872314,
      "learning_rate": 4.932294248563406e-05,
      "loss": 1.3618,
      "step": 25040
    },
    {
      "epoch": 6.694281133083912,
      "grad_norm": 1.9609215259552002,
      "learning_rate": 4.9250584426087356e-05,
      "loss": 1.3455,
      "step": 25050
    },
    {
      "epoch": 6.69695350080171,
      "grad_norm": 2.044280767440796,
      "learning_rate": 4.9178262136891454e-05,
      "loss": 1.3548,
      "step": 25060
    },
    {
      "epoch": 6.699625868519508,
      "grad_norm": 1.970746636390686,
      "learning_rate": 4.910597566902211e-05,
      "loss": 1.3309,
      "step": 25070
    },
    {
      "epoch": 6.702298236237306,
      "grad_norm": 1.9606901407241821,
      "learning_rate": 4.903372507342995e-05,
      "loss": 1.2857,
      "step": 25080
    },
    {
      "epoch": 6.704970603955104,
      "grad_norm": 2.2049036026000977,
      "learning_rate": 4.8961510401040254e-05,
      "loss": 1.3342,
      "step": 25090
    },
    {
      "epoch": 6.707642971672902,
      "grad_norm": 1.8935282230377197,
      "learning_rate": 4.8889331702753025e-05,
      "loss": 1.3692,
      "step": 25100
    },
    {
      "epoch": 6.7103153393907,
      "grad_norm": 2.032672643661499,
      "learning_rate": 4.881718902944291e-05,
      "loss": 1.3425,
      "step": 25110
    },
    {
      "epoch": 6.712987707108498,
      "grad_norm": 1.9839675426483154,
      "learning_rate": 4.874508243195902e-05,
      "loss": 1.3862,
      "step": 25120
    },
    {
      "epoch": 6.715660074826296,
      "grad_norm": 1.9736772775650024,
      "learning_rate": 4.8673011961125334e-05,
      "loss": 1.3463,
      "step": 25130
    },
    {
      "epoch": 6.718332442544094,
      "grad_norm": 1.9014322757720947,
      "learning_rate": 4.8600977667740055e-05,
      "loss": 1.2962,
      "step": 25140
    },
    {
      "epoch": 6.721004810261892,
      "grad_norm": 2.1234326362609863,
      "learning_rate": 4.8528979602576066e-05,
      "loss": 1.3125,
      "step": 25150
    },
    {
      "epoch": 6.72367717797969,
      "grad_norm": 2.13374400138855,
      "learning_rate": 4.8457017816380655e-05,
      "loss": 1.4241,
      "step": 25160
    },
    {
      "epoch": 6.726349545697488,
      "grad_norm": 1.9276955127716064,
      "learning_rate": 4.8385092359875626e-05,
      "loss": 1.2602,
      "step": 25170
    },
    {
      "epoch": 6.729021913415286,
      "grad_norm": 1.8928720951080322,
      "learning_rate": 4.831320328375698e-05,
      "loss": 1.2977,
      "step": 25180
    },
    {
      "epoch": 6.731694281133084,
      "grad_norm": 1.9410635232925415,
      "learning_rate": 4.8241350638695346e-05,
      "loss": 1.3074,
      "step": 25190
    },
    {
      "epoch": 6.734366648850882,
      "grad_norm": 1.9313756227493286,
      "learning_rate": 4.8169534475335374e-05,
      "loss": 1.3694,
      "step": 25200
    },
    {
      "epoch": 6.73703901656868,
      "grad_norm": 2.0795538425445557,
      "learning_rate": 4.8097754844296326e-05,
      "loss": 1.3431,
      "step": 25210
    },
    {
      "epoch": 6.739711384286478,
      "grad_norm": 2.1617908477783203,
      "learning_rate": 4.8026011796171425e-05,
      "loss": 1.3635,
      "step": 25220
    },
    {
      "epoch": 6.742383752004276,
      "grad_norm": 2.0728282928466797,
      "learning_rate": 4.7954305381528254e-05,
      "loss": 1.285,
      "step": 25230
    },
    {
      "epoch": 6.745056119722074,
      "grad_norm": 1.8272680044174194,
      "learning_rate": 4.788263565090856e-05,
      "loss": 1.3446,
      "step": 25240
    },
    {
      "epoch": 6.747728487439872,
      "grad_norm": 1.9283318519592285,
      "learning_rate": 4.781100265482826e-05,
      "loss": 1.3083,
      "step": 25250
    },
    {
      "epoch": 6.75040085515767,
      "grad_norm": 1.8441859483718872,
      "learning_rate": 4.773940644377727e-05,
      "loss": 1.3354,
      "step": 25260
    },
    {
      "epoch": 6.753073222875468,
      "grad_norm": 1.9824219942092896,
      "learning_rate": 4.7667847068219674e-05,
      "loss": 1.2434,
      "step": 25270
    },
    {
      "epoch": 6.755745590593266,
      "grad_norm": 1.8814092874526978,
      "learning_rate": 4.7596324578593586e-05,
      "loss": 1.3441,
      "step": 25280
    },
    {
      "epoch": 6.758417958311064,
      "grad_norm": 1.9747974872589111,
      "learning_rate": 4.752483902531111e-05,
      "loss": 1.3623,
      "step": 25290
    },
    {
      "epoch": 6.761090326028862,
      "grad_norm": 1.963364601135254,
      "learning_rate": 4.745339045875832e-05,
      "loss": 1.2794,
      "step": 25300
    },
    {
      "epoch": 6.76376269374666,
      "grad_norm": 2.001263380050659,
      "learning_rate": 4.738197892929512e-05,
      "loss": 1.3283,
      "step": 25310
    },
    {
      "epoch": 6.766435061464458,
      "grad_norm": 1.976517677307129,
      "learning_rate": 4.731060448725554e-05,
      "loss": 1.344,
      "step": 25320
    },
    {
      "epoch": 6.769107429182256,
      "grad_norm": 2.0967068672180176,
      "learning_rate": 4.723926718294722e-05,
      "loss": 1.2342,
      "step": 25330
    },
    {
      "epoch": 6.771779796900054,
      "grad_norm": 2.0544328689575195,
      "learning_rate": 4.716796706665177e-05,
      "loss": 1.3306,
      "step": 25340
    },
    {
      "epoch": 6.774452164617851,
      "grad_norm": 1.9995408058166504,
      "learning_rate": 4.7096704188624564e-05,
      "loss": 1.2832,
      "step": 25350
    },
    {
      "epoch": 6.777124532335649,
      "grad_norm": 1.9461665153503418,
      "learning_rate": 4.7025478599094686e-05,
      "loss": 1.3502,
      "step": 25360
    },
    {
      "epoch": 6.779796900053447,
      "grad_norm": 2.1463234424591064,
      "learning_rate": 4.695429034826499e-05,
      "loss": 1.2909,
      "step": 25370
    },
    {
      "epoch": 6.782469267771245,
      "grad_norm": 1.9402949810028076,
      "learning_rate": 4.6883139486312e-05,
      "loss": 1.3051,
      "step": 25380
    },
    {
      "epoch": 6.785141635489043,
      "grad_norm": 1.8647191524505615,
      "learning_rate": 4.6812026063385796e-05,
      "loss": 1.2959,
      "step": 25390
    },
    {
      "epoch": 6.787814003206841,
      "grad_norm": 1.8827990293502808,
      "learning_rate": 4.674095012961027e-05,
      "loss": 1.2996,
      "step": 25400
    },
    {
      "epoch": 6.790486370924639,
      "grad_norm": 2.082275152206421,
      "learning_rate": 4.6669911735082685e-05,
      "loss": 1.3208,
      "step": 25410
    },
    {
      "epoch": 6.793158738642437,
      "grad_norm": 1.8540116548538208,
      "learning_rate": 4.659891092987395e-05,
      "loss": 1.3016,
      "step": 25420
    },
    {
      "epoch": 6.795831106360235,
      "grad_norm": 1.945229411125183,
      "learning_rate": 4.6527947764028456e-05,
      "loss": 1.2816,
      "step": 25430
    },
    {
      "epoch": 6.798503474078033,
      "grad_norm": 1.9631634950637817,
      "learning_rate": 4.6457022287564066e-05,
      "loss": 1.3821,
      "step": 25440
    },
    {
      "epoch": 6.801175841795831,
      "grad_norm": 2.0663106441497803,
      "learning_rate": 4.638613455047213e-05,
      "loss": 1.2343,
      "step": 25450
    },
    {
      "epoch": 6.803848209513629,
      "grad_norm": 2.1802597045898438,
      "learning_rate": 4.631528460271725e-05,
      "loss": 1.2804,
      "step": 25460
    },
    {
      "epoch": 6.806520577231427,
      "grad_norm": 2.0285112857818604,
      "learning_rate": 4.6244472494237536e-05,
      "loss": 1.3734,
      "step": 25470
    },
    {
      "epoch": 6.809192944949225,
      "grad_norm": 1.9220365285873413,
      "learning_rate": 4.617369827494439e-05,
      "loss": 1.302,
      "step": 25480
    },
    {
      "epoch": 6.811865312667023,
      "grad_norm": 2.2240166664123535,
      "learning_rate": 4.6102961994722515e-05,
      "loss": 1.3799,
      "step": 25490
    },
    {
      "epoch": 6.814537680384821,
      "grad_norm": 1.9817068576812744,
      "learning_rate": 4.603226370342977e-05,
      "loss": 1.3467,
      "step": 25500
    },
    {
      "epoch": 6.817210048102619,
      "grad_norm": 1.7919260263442993,
      "learning_rate": 4.5961603450897436e-05,
      "loss": 1.4048,
      "step": 25510
    },
    {
      "epoch": 6.819882415820417,
      "grad_norm": 1.9146183729171753,
      "learning_rate": 4.5890981286929754e-05,
      "loss": 1.2824,
      "step": 25520
    },
    {
      "epoch": 6.822554783538215,
      "grad_norm": 1.9657856225967407,
      "learning_rate": 4.582039726130437e-05,
      "loss": 1.3488,
      "step": 25530
    },
    {
      "epoch": 6.825227151256013,
      "grad_norm": 1.9412111043930054,
      "learning_rate": 4.57498514237718e-05,
      "loss": 1.3343,
      "step": 25540
    },
    {
      "epoch": 6.827899518973811,
      "grad_norm": 2.0994606018066406,
      "learning_rate": 4.567934382405581e-05,
      "loss": 1.3324,
      "step": 25550
    },
    {
      "epoch": 6.830571886691609,
      "grad_norm": 1.8755228519439697,
      "learning_rate": 4.560887451185314e-05,
      "loss": 1.2411,
      "step": 25560
    },
    {
      "epoch": 6.833244254409407,
      "grad_norm": 1.7629376649856567,
      "learning_rate": 4.553844353683361e-05,
      "loss": 1.2474,
      "step": 25570
    },
    {
      "epoch": 6.835916622127205,
      "grad_norm": 1.887208104133606,
      "learning_rate": 4.546805094863985e-05,
      "loss": 1.3752,
      "step": 25580
    },
    {
      "epoch": 6.838588989845003,
      "grad_norm": 2.057021141052246,
      "learning_rate": 4.5397696796887726e-05,
      "loss": 1.327,
      "step": 25590
    },
    {
      "epoch": 6.841261357562801,
      "grad_norm": 1.8888603448867798,
      "learning_rate": 4.532738113116571e-05,
      "loss": 1.3035,
      "step": 25600
    },
    {
      "epoch": 6.843933725280599,
      "grad_norm": 1.9359172582626343,
      "learning_rate": 4.5257104001035333e-05,
      "loss": 1.3673,
      "step": 25610
    },
    {
      "epoch": 6.846606092998397,
      "grad_norm": 2.076350450515747,
      "learning_rate": 4.5186865456030914e-05,
      "loss": 1.3122,
      "step": 25620
    },
    {
      "epoch": 6.849278460716194,
      "grad_norm": 1.9578807353973389,
      "learning_rate": 4.511666554565955e-05,
      "loss": 1.3545,
      "step": 25630
    },
    {
      "epoch": 6.851950828433992,
      "grad_norm": 2.000336170196533,
      "learning_rate": 4.5046504319401183e-05,
      "loss": 1.316,
      "step": 25640
    },
    {
      "epoch": 6.85462319615179,
      "grad_norm": 2.111220359802246,
      "learning_rate": 4.497638182670836e-05,
      "loss": 1.3444,
      "step": 25650
    },
    {
      "epoch": 6.857295563869588,
      "grad_norm": 2.0362842082977295,
      "learning_rate": 4.490629811700644e-05,
      "loss": 1.3057,
      "step": 25660
    },
    {
      "epoch": 6.859967931587386,
      "grad_norm": 2.01241135597229,
      "learning_rate": 4.483625323969339e-05,
      "loss": 1.3195,
      "step": 25670
    },
    {
      "epoch": 6.862640299305184,
      "grad_norm": 1.8604832887649536,
      "learning_rate": 4.476624724413986e-05,
      "loss": 1.3234,
      "step": 25680
    },
    {
      "epoch": 6.865312667022982,
      "grad_norm": 1.9640719890594482,
      "learning_rate": 4.4696280179689034e-05,
      "loss": 1.3018,
      "step": 25690
    },
    {
      "epoch": 6.86798503474078,
      "grad_norm": 1.862758994102478,
      "learning_rate": 4.462635209565671e-05,
      "loss": 1.3515,
      "step": 25700
    },
    {
      "epoch": 6.870657402458578,
      "grad_norm": 1.9486703872680664,
      "learning_rate": 4.4556463041331086e-05,
      "loss": 1.2896,
      "step": 25710
    },
    {
      "epoch": 6.873329770176376,
      "grad_norm": 1.8705099821090698,
      "learning_rate": 4.44866130659731e-05,
      "loss": 1.3557,
      "step": 25720
    },
    {
      "epoch": 6.876002137894174,
      "grad_norm": 1.9109236001968384,
      "learning_rate": 4.441680221881587e-05,
      "loss": 1.3565,
      "step": 25730
    },
    {
      "epoch": 6.878674505611972,
      "grad_norm": 1.9683568477630615,
      "learning_rate": 4.434703054906508e-05,
      "loss": 1.3075,
      "step": 25740
    },
    {
      "epoch": 6.88134687332977,
      "grad_norm": 1.9412583112716675,
      "learning_rate": 4.4277298105898815e-05,
      "loss": 1.3709,
      "step": 25750
    },
    {
      "epoch": 6.884019241047568,
      "grad_norm": 1.8737123012542725,
      "learning_rate": 4.420760493846743e-05,
      "loss": 1.3315,
      "step": 25760
    },
    {
      "epoch": 6.886691608765366,
      "grad_norm": 1.9197075366973877,
      "learning_rate": 4.413795109589366e-05,
      "loss": 1.282,
      "step": 25770
    },
    {
      "epoch": 6.889363976483164,
      "grad_norm": 2.134570837020874,
      "learning_rate": 4.406833662727253e-05,
      "loss": 1.3583,
      "step": 25780
    },
    {
      "epoch": 6.892036344200962,
      "grad_norm": 2.059075355529785,
      "learning_rate": 4.399876158167122e-05,
      "loss": 1.4106,
      "step": 25790
    },
    {
      "epoch": 6.89470871191876,
      "grad_norm": 1.904106616973877,
      "learning_rate": 4.392922600812922e-05,
      "loss": 1.2815,
      "step": 25800
    },
    {
      "epoch": 6.897381079636558,
      "grad_norm": 1.9103493690490723,
      "learning_rate": 4.385972995565817e-05,
      "loss": 1.3453,
      "step": 25810
    },
    {
      "epoch": 6.900053447354356,
      "grad_norm": 1.8709713220596313,
      "learning_rate": 4.379027347324184e-05,
      "loss": 1.3045,
      "step": 25820
    },
    {
      "epoch": 6.902725815072154,
      "grad_norm": 1.9322621822357178,
      "learning_rate": 4.372085660983616e-05,
      "loss": 1.3484,
      "step": 25830
    },
    {
      "epoch": 6.905398182789952,
      "grad_norm": 2.026193380355835,
      "learning_rate": 4.365147941436899e-05,
      "loss": 1.3053,
      "step": 25840
    },
    {
      "epoch": 6.90807055050775,
      "grad_norm": 1.9680495262145996,
      "learning_rate": 4.358214193574046e-05,
      "loss": 1.3385,
      "step": 25850
    },
    {
      "epoch": 6.910742918225548,
      "grad_norm": 1.9501479864120483,
      "learning_rate": 4.351284422282249e-05,
      "loss": 1.3452,
      "step": 25860
    },
    {
      "epoch": 6.913415285943346,
      "grad_norm": 1.7781267166137695,
      "learning_rate": 4.3443586324459095e-05,
      "loss": 1.3682,
      "step": 25870
    },
    {
      "epoch": 6.916087653661144,
      "grad_norm": 2.174943685531616,
      "learning_rate": 4.337436828946618e-05,
      "loss": 1.3665,
      "step": 25880
    },
    {
      "epoch": 6.918760021378942,
      "grad_norm": 1.9405536651611328,
      "learning_rate": 4.330519016663163e-05,
      "loss": 1.4252,
      "step": 25890
    },
    {
      "epoch": 6.92143238909674,
      "grad_norm": 2.011246681213379,
      "learning_rate": 4.323605200471499e-05,
      "loss": 1.359,
      "step": 25900
    },
    {
      "epoch": 6.924104756814538,
      "grad_norm": 1.920982837677002,
      "learning_rate": 4.3166953852447954e-05,
      "loss": 1.3207,
      "step": 25910
    },
    {
      "epoch": 6.926777124532336,
      "grad_norm": 2.0366029739379883,
      "learning_rate": 4.309789575853368e-05,
      "loss": 1.3285,
      "step": 25920
    },
    {
      "epoch": 6.929449492250134,
      "grad_norm": 1.91665518283844,
      "learning_rate": 4.302887777164742e-05,
      "loss": 1.3189,
      "step": 25930
    },
    {
      "epoch": 6.932121859967932,
      "grad_norm": 1.9528197050094604,
      "learning_rate": 4.2959899940435866e-05,
      "loss": 1.2788,
      "step": 25940
    },
    {
      "epoch": 6.93479422768573,
      "grad_norm": 2.000671625137329,
      "learning_rate": 4.289096231351759e-05,
      "loss": 1.3488,
      "step": 25950
    },
    {
      "epoch": 6.937466595403528,
      "grad_norm": 2.094839096069336,
      "learning_rate": 4.2822064939482744e-05,
      "loss": 1.3504,
      "step": 25960
    },
    {
      "epoch": 6.940138963121326,
      "grad_norm": 2.1478819847106934,
      "learning_rate": 4.275320786689319e-05,
      "loss": 1.3497,
      "step": 25970
    },
    {
      "epoch": 6.942811330839124,
      "grad_norm": 2.1537442207336426,
      "learning_rate": 4.268439114428223e-05,
      "loss": 1.3983,
      "step": 25980
    },
    {
      "epoch": 6.945483698556921,
      "grad_norm": 1.9925214052200317,
      "learning_rate": 4.2615614820154895e-05,
      "loss": 1.3543,
      "step": 25990
    },
    {
      "epoch": 6.948156066274719,
      "grad_norm": 1.9166977405548096,
      "learning_rate": 4.254687894298764e-05,
      "loss": 1.3626,
      "step": 26000
    },
    {
      "epoch": 6.950828433992517,
      "grad_norm": 2.103987216949463,
      "learning_rate": 4.2478183561228455e-05,
      "loss": 1.3837,
      "step": 26010
    },
    {
      "epoch": 6.953500801710315,
      "grad_norm": 2.0219504833221436,
      "learning_rate": 4.24095287232968e-05,
      "loss": 1.4031,
      "step": 26020
    },
    {
      "epoch": 6.956173169428113,
      "grad_norm": 1.9439438581466675,
      "learning_rate": 4.2340914477583425e-05,
      "loss": 1.2934,
      "step": 26030
    },
    {
      "epoch": 6.958845537145911,
      "grad_norm": 1.9404563903808594,
      "learning_rate": 4.227234087245072e-05,
      "loss": 1.3562,
      "step": 26040
    },
    {
      "epoch": 6.961517904863709,
      "grad_norm": 2.0004537105560303,
      "learning_rate": 4.220380795623218e-05,
      "loss": 1.3195,
      "step": 26050
    },
    {
      "epoch": 6.964190272581507,
      "grad_norm": 1.8503367900848389,
      "learning_rate": 4.213531577723274e-05,
      "loss": 1.3032,
      "step": 26060
    },
    {
      "epoch": 6.966862640299305,
      "grad_norm": 2.080974817276001,
      "learning_rate": 4.206686438372864e-05,
      "loss": 1.3404,
      "step": 26070
    },
    {
      "epoch": 6.969535008017103,
      "grad_norm": 1.9929003715515137,
      "learning_rate": 4.199845382396732e-05,
      "loss": 1.3262,
      "step": 26080
    },
    {
      "epoch": 6.972207375734901,
      "grad_norm": 2.0957374572753906,
      "learning_rate": 4.193008414616745e-05,
      "loss": 1.4116,
      "step": 26090
    },
    {
      "epoch": 6.974879743452699,
      "grad_norm": 2.079894542694092,
      "learning_rate": 4.186175539851894e-05,
      "loss": 1.3242,
      "step": 26100
    },
    {
      "epoch": 6.977552111170497,
      "grad_norm": 2.051224946975708,
      "learning_rate": 4.17934676291827e-05,
      "loss": 1.3347,
      "step": 26110
    },
    {
      "epoch": 6.980224478888295,
      "grad_norm": 1.7990056276321411,
      "learning_rate": 4.172522088629099e-05,
      "loss": 1.3434,
      "step": 26120
    },
    {
      "epoch": 6.982896846606093,
      "grad_norm": 1.9451942443847656,
      "learning_rate": 4.1657015217946936e-05,
      "loss": 1.3354,
      "step": 26130
    },
    {
      "epoch": 6.985569214323891,
      "grad_norm": 2.1389999389648438,
      "learning_rate": 4.158885067222481e-05,
      "loss": 1.2951,
      "step": 26140
    },
    {
      "epoch": 6.988241582041689,
      "grad_norm": 2.0456199645996094,
      "learning_rate": 4.1520727297169906e-05,
      "loss": 1.2959,
      "step": 26150
    },
    {
      "epoch": 6.990913949759487,
      "grad_norm": 2.0102691650390625,
      "learning_rate": 4.145264514079847e-05,
      "loss": 1.3544,
      "step": 26160
    },
    {
      "epoch": 6.993586317477285,
      "grad_norm": 1.9765325784683228,
      "learning_rate": 4.138460425109775e-05,
      "loss": 1.3281,
      "step": 26170
    },
    {
      "epoch": 6.996258685195083,
      "grad_norm": 1.9393184185028076,
      "learning_rate": 4.131660467602579e-05,
      "loss": 1.2823,
      "step": 26180
    },
    {
      "epoch": 6.998931052912881,
      "grad_norm": 2.0244810581207275,
      "learning_rate": 4.124864646351163e-05,
      "loss": 1.349,
      "step": 26190
    },
    {
      "epoch": 7.001603420630679,
      "grad_norm": 1.9014036655426025,
      "learning_rate": 4.1180729661455106e-05,
      "loss": 1.2706,
      "step": 26200
    },
    {
      "epoch": 7.004275788348477,
      "grad_norm": 1.9684710502624512,
      "learning_rate": 4.111285431772692e-05,
      "loss": 1.2202,
      "step": 26210
    },
    {
      "epoch": 7.006948156066275,
      "grad_norm": 2.188354969024658,
      "learning_rate": 4.1045020480168394e-05,
      "loss": 1.2392,
      "step": 26220
    },
    {
      "epoch": 7.009620523784073,
      "grad_norm": 1.9602842330932617,
      "learning_rate": 4.097722819659184e-05,
      "loss": 1.2321,
      "step": 26230
    },
    {
      "epoch": 7.012292891501871,
      "grad_norm": 2.304478406906128,
      "learning_rate": 4.090947751478003e-05,
      "loss": 1.1661,
      "step": 26240
    },
    {
      "epoch": 7.014965259219669,
      "grad_norm": 2.202989339828491,
      "learning_rate": 4.0841768482486655e-05,
      "loss": 1.1744,
      "step": 26250
    },
    {
      "epoch": 7.017637626937467,
      "grad_norm": 2.0549769401550293,
      "learning_rate": 4.077410114743582e-05,
      "loss": 1.2174,
      "step": 26260
    },
    {
      "epoch": 7.020309994655265,
      "grad_norm": 2.0904242992401123,
      "learning_rate": 4.070647555732239e-05,
      "loss": 1.2886,
      "step": 26270
    },
    {
      "epoch": 7.022982362373063,
      "grad_norm": 1.912893533706665,
      "learning_rate": 4.0638891759811746e-05,
      "loss": 1.1457,
      "step": 26280
    },
    {
      "epoch": 7.025654730090861,
      "grad_norm": 2.1165356636047363,
      "learning_rate": 4.0571349802539895e-05,
      "loss": 1.3009,
      "step": 26290
    },
    {
      "epoch": 7.028327097808658,
      "grad_norm": 2.0290679931640625,
      "learning_rate": 4.050384973311315e-05,
      "loss": 1.2149,
      "step": 26300
    },
    {
      "epoch": 7.030999465526456,
      "grad_norm": 2.025852680206299,
      "learning_rate": 4.0436391599108606e-05,
      "loss": 1.1835,
      "step": 26310
    },
    {
      "epoch": 7.033671833244254,
      "grad_norm": 2.161137342453003,
      "learning_rate": 4.03689754480735e-05,
      "loss": 1.2131,
      "step": 26320
    },
    {
      "epoch": 7.036344200962052,
      "grad_norm": 2.2476398944854736,
      "learning_rate": 4.030160132752565e-05,
      "loss": 1.1689,
      "step": 26330
    },
    {
      "epoch": 7.03901656867985,
      "grad_norm": 2.1565744876861572,
      "learning_rate": 4.023426928495322e-05,
      "loss": 1.2858,
      "step": 26340
    },
    {
      "epoch": 7.041688936397648,
      "grad_norm": 2.135911703109741,
      "learning_rate": 4.0166979367814685e-05,
      "loss": 1.1995,
      "step": 26350
    },
    {
      "epoch": 7.044361304115446,
      "grad_norm": 1.8626652956008911,
      "learning_rate": 4.009973162353888e-05,
      "loss": 1.1992,
      "step": 26360
    },
    {
      "epoch": 7.047033671833244,
      "grad_norm": 2.130089044570923,
      "learning_rate": 4.003252609952484e-05,
      "loss": 1.2277,
      "step": 26370
    },
    {
      "epoch": 7.049706039551042,
      "grad_norm": 2.1491618156433105,
      "learning_rate": 3.9965362843141886e-05,
      "loss": 1.2943,
      "step": 26380
    },
    {
      "epoch": 7.05237840726884,
      "grad_norm": 2.036555528640747,
      "learning_rate": 3.989824190172955e-05,
      "loss": 1.1872,
      "step": 26390
    },
    {
      "epoch": 7.055050774986638,
      "grad_norm": 2.0331013202667236,
      "learning_rate": 3.9831163322597544e-05,
      "loss": 1.1884,
      "step": 26400
    },
    {
      "epoch": 7.057723142704436,
      "grad_norm": 2.068228006362915,
      "learning_rate": 3.97641271530257e-05,
      "loss": 1.2011,
      "step": 26410
    },
    {
      "epoch": 7.060395510422234,
      "grad_norm": 2.1616146564483643,
      "learning_rate": 3.9697133440264e-05,
      "loss": 1.1771,
      "step": 26420
    },
    {
      "epoch": 7.063067878140032,
      "grad_norm": 2.004011392593384,
      "learning_rate": 3.9630182231532366e-05,
      "loss": 1.2046,
      "step": 26430
    },
    {
      "epoch": 7.06574024585783,
      "grad_norm": 2.036231279373169,
      "learning_rate": 3.956327357402101e-05,
      "loss": 1.2836,
      "step": 26440
    },
    {
      "epoch": 7.068412613575628,
      "grad_norm": 2.101543426513672,
      "learning_rate": 3.949640751488989e-05,
      "loss": 1.2117,
      "step": 26450
    },
    {
      "epoch": 7.071084981293426,
      "grad_norm": 1.9944366216659546,
      "learning_rate": 3.942958410126909e-05,
      "loss": 1.1968,
      "step": 26460
    },
    {
      "epoch": 7.073757349011224,
      "grad_norm": 1.9979811906814575,
      "learning_rate": 3.9362803380258606e-05,
      "loss": 1.1528,
      "step": 26470
    },
    {
      "epoch": 7.076429716729022,
      "grad_norm": 2.1518075466156006,
      "learning_rate": 3.929606539892833e-05,
      "loss": 1.2141,
      "step": 26480
    },
    {
      "epoch": 7.07910208444682,
      "grad_norm": 2.061533212661743,
      "learning_rate": 3.922937020431803e-05,
      "loss": 1.2192,
      "step": 26490
    },
    {
      "epoch": 7.081774452164618,
      "grad_norm": 2.0257511138916016,
      "learning_rate": 3.916271784343737e-05,
      "loss": 1.1701,
      "step": 26500
    },
    {
      "epoch": 7.084446819882416,
      "grad_norm": 2.084320545196533,
      "learning_rate": 3.9096108363265694e-05,
      "loss": 1.2416,
      "step": 26510
    },
    {
      "epoch": 7.087119187600214,
      "grad_norm": 2.0941121578216553,
      "learning_rate": 3.9029541810752244e-05,
      "loss": 1.208,
      "step": 26520
    },
    {
      "epoch": 7.089791555318012,
      "grad_norm": 2.087214708328247,
      "learning_rate": 3.896301823281595e-05,
      "loss": 1.1721,
      "step": 26530
    },
    {
      "epoch": 7.09246392303581,
      "grad_norm": 2.1198551654815674,
      "learning_rate": 3.889653767634549e-05,
      "loss": 1.2566,
      "step": 26540
    },
    {
      "epoch": 7.095136290753608,
      "grad_norm": 1.9783620834350586,
      "learning_rate": 3.88301001881992e-05,
      "loss": 1.2379,
      "step": 26550
    },
    {
      "epoch": 7.097808658471406,
      "grad_norm": 2.0319607257843018,
      "learning_rate": 3.8763705815204976e-05,
      "loss": 1.224,
      "step": 26560
    },
    {
      "epoch": 7.100481026189204,
      "grad_norm": 2.227588415145874,
      "learning_rate": 3.869735460416051e-05,
      "loss": 1.2774,
      "step": 26570
    },
    {
      "epoch": 7.103153393907002,
      "grad_norm": 2.153714895248413,
      "learning_rate": 3.8631046601832876e-05,
      "loss": 1.2179,
      "step": 26580
    },
    {
      "epoch": 7.1058257616248,
      "grad_norm": 2.2052743434906006,
      "learning_rate": 3.856478185495882e-05,
      "loss": 1.2417,
      "step": 26590
    },
    {
      "epoch": 7.108498129342598,
      "grad_norm": 2.112349271774292,
      "learning_rate": 3.8498560410244546e-05,
      "loss": 1.2785,
      "step": 26600
    },
    {
      "epoch": 7.111170497060396,
      "grad_norm": 2.0284695625305176,
      "learning_rate": 3.84323823143658e-05,
      "loss": 1.2482,
      "step": 26610
    },
    {
      "epoch": 7.113842864778193,
      "grad_norm": 2.106285810470581,
      "learning_rate": 3.83662476139676e-05,
      "loss": 1.274,
      "step": 26620
    },
    {
      "epoch": 7.116515232495991,
      "grad_norm": 2.0881786346435547,
      "learning_rate": 3.8300156355664665e-05,
      "loss": 1.2292,
      "step": 26630
    },
    {
      "epoch": 7.119187600213789,
      "grad_norm": 2.2239317893981934,
      "learning_rate": 3.823410858604076e-05,
      "loss": 1.212,
      "step": 26640
    },
    {
      "epoch": 7.121859967931587,
      "grad_norm": 2.2325310707092285,
      "learning_rate": 3.8168104351649335e-05,
      "loss": 1.2368,
      "step": 26650
    },
    {
      "epoch": 7.124532335649385,
      "grad_norm": 2.174346685409546,
      "learning_rate": 3.810214369901287e-05,
      "loss": 1.1748,
      "step": 26660
    },
    {
      "epoch": 7.127204703367183,
      "grad_norm": 2.0972397327423096,
      "learning_rate": 3.803622667462328e-05,
      "loss": 1.1974,
      "step": 26670
    },
    {
      "epoch": 7.129877071084981,
      "grad_norm": 2.162646770477295,
      "learning_rate": 3.797035332494169e-05,
      "loss": 1.1844,
      "step": 26680
    },
    {
      "epoch": 7.132549438802779,
      "grad_norm": 2.159243583679199,
      "learning_rate": 3.7904523696398484e-05,
      "loss": 1.202,
      "step": 26690
    },
    {
      "epoch": 7.135221806520577,
      "grad_norm": 2.0711281299591064,
      "learning_rate": 3.7838737835393124e-05,
      "loss": 1.2173,
      "step": 26700
    },
    {
      "epoch": 7.137894174238375,
      "grad_norm": 2.230268955230713,
      "learning_rate": 3.777299578829431e-05,
      "loss": 1.2559,
      "step": 26710
    },
    {
      "epoch": 7.140566541956173,
      "grad_norm": 1.9708648920059204,
      "learning_rate": 3.770729760143985e-05,
      "loss": 1.2635,
      "step": 26720
    },
    {
      "epoch": 7.143238909673971,
      "grad_norm": 2.09283447265625,
      "learning_rate": 3.7641643321136624e-05,
      "loss": 1.2281,
      "step": 26730
    },
    {
      "epoch": 7.145911277391769,
      "grad_norm": 2.0920164585113525,
      "learning_rate": 3.757603299366059e-05,
      "loss": 1.2649,
      "step": 26740
    },
    {
      "epoch": 7.148583645109567,
      "grad_norm": 2.066572666168213,
      "learning_rate": 3.7510466665256614e-05,
      "loss": 1.2286,
      "step": 26750
    },
    {
      "epoch": 7.151256012827365,
      "grad_norm": 1.872041940689087,
      "learning_rate": 3.744494438213877e-05,
      "loss": 1.2269,
      "step": 26760
    },
    {
      "epoch": 7.153928380545163,
      "grad_norm": 2.0074431896209717,
      "learning_rate": 3.737946619048985e-05,
      "loss": 1.2338,
      "step": 26770
    },
    {
      "epoch": 7.156600748262961,
      "grad_norm": 2.1136064529418945,
      "learning_rate": 3.731403213646171e-05,
      "loss": 1.1935,
      "step": 26780
    },
    {
      "epoch": 7.159273115980759,
      "grad_norm": 2.053072452545166,
      "learning_rate": 3.724864226617506e-05,
      "loss": 1.221,
      "step": 26790
    },
    {
      "epoch": 7.161945483698557,
      "grad_norm": 2.150113344192505,
      "learning_rate": 3.718329662571946e-05,
      "loss": 1.1816,
      "step": 26800
    },
    {
      "epoch": 7.164617851416355,
      "grad_norm": 2.078615427017212,
      "learning_rate": 3.711799526115333e-05,
      "loss": 1.2467,
      "step": 26810
    },
    {
      "epoch": 7.167290219134153,
      "grad_norm": 2.1886465549468994,
      "learning_rate": 3.705273821850385e-05,
      "loss": 1.2066,
      "step": 26820
    },
    {
      "epoch": 7.169962586851951,
      "grad_norm": 2.0835061073303223,
      "learning_rate": 3.698752554376689e-05,
      "loss": 1.2217,
      "step": 26830
    },
    {
      "epoch": 7.172634954569749,
      "grad_norm": 2.081408977508545,
      "learning_rate": 3.6922357282907236e-05,
      "loss": 1.2074,
      "step": 26840
    },
    {
      "epoch": 7.175307322287547,
      "grad_norm": 1.9500755071640015,
      "learning_rate": 3.685723348185818e-05,
      "loss": 1.2378,
      "step": 26850
    },
    {
      "epoch": 7.177979690005345,
      "grad_norm": 2.18097186088562,
      "learning_rate": 3.679215418652177e-05,
      "loss": 1.2477,
      "step": 26860
    },
    {
      "epoch": 7.180652057723143,
      "grad_norm": 2.0728347301483154,
      "learning_rate": 3.672711944276865e-05,
      "loss": 1.1835,
      "step": 26870
    },
    {
      "epoch": 7.183324425440941,
      "grad_norm": 2.0779342651367188,
      "learning_rate": 3.666212929643812e-05,
      "loss": 1.2418,
      "step": 26880
    },
    {
      "epoch": 7.185996793158739,
      "grad_norm": 2.126870632171631,
      "learning_rate": 3.659718379333801e-05,
      "loss": 1.1285,
      "step": 26890
    },
    {
      "epoch": 7.188669160876537,
      "grad_norm": 2.122117280960083,
      "learning_rate": 3.653228297924461e-05,
      "loss": 1.1866,
      "step": 26900
    },
    {
      "epoch": 7.191341528594335,
      "grad_norm": 2.13063645362854,
      "learning_rate": 3.6467426899902826e-05,
      "loss": 1.1705,
      "step": 26910
    },
    {
      "epoch": 7.194013896312133,
      "grad_norm": 2.143397092819214,
      "learning_rate": 3.640261560102598e-05,
      "loss": 1.249,
      "step": 26920
    },
    {
      "epoch": 7.196686264029931,
      "grad_norm": 2.073452949523926,
      "learning_rate": 3.633784912829586e-05,
      "loss": 1.2529,
      "step": 26930
    },
    {
      "epoch": 7.199358631747728,
      "grad_norm": 2.1552131175994873,
      "learning_rate": 3.627312752736256e-05,
      "loss": 1.2631,
      "step": 26940
    },
    {
      "epoch": 7.202030999465526,
      "grad_norm": 2.0443172454833984,
      "learning_rate": 3.620845084384474e-05,
      "loss": 1.2495,
      "step": 26950
    },
    {
      "epoch": 7.204703367183324,
      "grad_norm": 1.8864065408706665,
      "learning_rate": 3.614381912332916e-05,
      "loss": 1.2517,
      "step": 26960
    },
    {
      "epoch": 7.207375734901122,
      "grad_norm": 2.0248312950134277,
      "learning_rate": 3.6079232411371125e-05,
      "loss": 1.2414,
      "step": 26970
    },
    {
      "epoch": 7.21004810261892,
      "grad_norm": 2.1586263179779053,
      "learning_rate": 3.601469075349402e-05,
      "loss": 1.2887,
      "step": 26980
    },
    {
      "epoch": 7.212720470336718,
      "grad_norm": 2.0618584156036377,
      "learning_rate": 3.595019419518958e-05,
      "loss": 1.179,
      "step": 26990
    },
    {
      "epoch": 7.215392838054516,
      "grad_norm": 2.2867074012756348,
      "learning_rate": 3.588574278191773e-05,
      "loss": 1.1725,
      "step": 27000
    },
    {
      "epoch": 7.218065205772314,
      "grad_norm": 2.165614366531372,
      "learning_rate": 3.58213365591066e-05,
      "loss": 1.2002,
      "step": 27010
    },
    {
      "epoch": 7.220737573490112,
      "grad_norm": 2.111840009689331,
      "learning_rate": 3.575697557215236e-05,
      "loss": 1.2551,
      "step": 27020
    },
    {
      "epoch": 7.22340994120791,
      "grad_norm": 2.1707651615142822,
      "learning_rate": 3.5692659866419484e-05,
      "loss": 1.3149,
      "step": 27030
    },
    {
      "epoch": 7.226082308925708,
      "grad_norm": 2.069751501083374,
      "learning_rate": 3.5628389487240344e-05,
      "loss": 1.2253,
      "step": 27040
    },
    {
      "epoch": 7.228754676643506,
      "grad_norm": 2.1931421756744385,
      "learning_rate": 3.556416447991546e-05,
      "loss": 1.2033,
      "step": 27050
    },
    {
      "epoch": 7.231427044361304,
      "grad_norm": 2.147146463394165,
      "learning_rate": 3.5499984889713364e-05,
      "loss": 1.2643,
      "step": 27060
    },
    {
      "epoch": 7.234099412079102,
      "grad_norm": 2.0687849521636963,
      "learning_rate": 3.543585076187057e-05,
      "loss": 1.2699,
      "step": 27070
    },
    {
      "epoch": 7.2367717797969,
      "grad_norm": 2.0323586463928223,
      "learning_rate": 3.537176214159156e-05,
      "loss": 1.1776,
      "step": 27080
    },
    {
      "epoch": 7.239444147514698,
      "grad_norm": 2.0149714946746826,
      "learning_rate": 3.530771907404867e-05,
      "loss": 1.2115,
      "step": 27090
    },
    {
      "epoch": 7.242116515232496,
      "grad_norm": 2.117044448852539,
      "learning_rate": 3.524372160438222e-05,
      "loss": 1.2277,
      "step": 27100
    },
    {
      "epoch": 7.244788882950294,
      "grad_norm": 2.203836441040039,
      "learning_rate": 3.517976977770033e-05,
      "loss": 1.2374,
      "step": 27110
    },
    {
      "epoch": 7.247461250668092,
      "grad_norm": 2.0977160930633545,
      "learning_rate": 3.511586363907902e-05,
      "loss": 1.2262,
      "step": 27120
    },
    {
      "epoch": 7.25013361838589,
      "grad_norm": 2.137202501296997,
      "learning_rate": 3.5052003233562e-05,
      "loss": 1.2438,
      "step": 27130
    },
    {
      "epoch": 7.252805986103688,
      "grad_norm": 2.1474907398223877,
      "learning_rate": 3.49881886061609e-05,
      "loss": 1.2149,
      "step": 27140
    },
    {
      "epoch": 7.255478353821486,
      "grad_norm": 2.2300004959106445,
      "learning_rate": 3.492441980185484e-05,
      "loss": 1.227,
      "step": 27150
    },
    {
      "epoch": 7.258150721539284,
      "grad_norm": 2.0743207931518555,
      "learning_rate": 3.486069686559095e-05,
      "loss": 1.2462,
      "step": 27160
    },
    {
      "epoch": 7.260823089257082,
      "grad_norm": 2.0569257736206055,
      "learning_rate": 3.479701984228375e-05,
      "loss": 1.2546,
      "step": 27170
    },
    {
      "epoch": 7.26349545697488,
      "grad_norm": 2.0144920349121094,
      "learning_rate": 3.473338877681558e-05,
      "loss": 1.3128,
      "step": 27180
    },
    {
      "epoch": 7.266167824692678,
      "grad_norm": 2.3156790733337402,
      "learning_rate": 3.4669803714036306e-05,
      "loss": 1.2507,
      "step": 27190
    },
    {
      "epoch": 7.268840192410476,
      "grad_norm": 2.3155198097229004,
      "learning_rate": 3.4606264698763404e-05,
      "loss": 1.2753,
      "step": 27200
    },
    {
      "epoch": 7.271512560128274,
      "grad_norm": 2.1147971153259277,
      "learning_rate": 3.454277177578187e-05,
      "loss": 1.2602,
      "step": 27210
    },
    {
      "epoch": 7.274184927846072,
      "grad_norm": 2.0096116065979004,
      "learning_rate": 3.447932498984425e-05,
      "loss": 1.1753,
      "step": 27220
    },
    {
      "epoch": 7.27685729556387,
      "grad_norm": 2.1410534381866455,
      "learning_rate": 3.4415924385670495e-05,
      "loss": 1.2732,
      "step": 27230
    },
    {
      "epoch": 7.279529663281668,
      "grad_norm": 2.093339204788208,
      "learning_rate": 3.435257000794807e-05,
      "loss": 1.1837,
      "step": 27240
    },
    {
      "epoch": 7.282202030999466,
      "grad_norm": 2.1783902645111084,
      "learning_rate": 3.428926190133185e-05,
      "loss": 1.255,
      "step": 27250
    },
    {
      "epoch": 7.284874398717264,
      "grad_norm": 1.9138840436935425,
      "learning_rate": 3.4226000110444064e-05,
      "loss": 1.2411,
      "step": 27260
    },
    {
      "epoch": 7.287546766435061,
      "grad_norm": 1.9429867267608643,
      "learning_rate": 3.416278467987439e-05,
      "loss": 1.186,
      "step": 27270
    },
    {
      "epoch": 7.290219134152859,
      "grad_norm": 2.091928005218506,
      "learning_rate": 3.409961565417963e-05,
      "loss": 1.2449,
      "step": 27280
    },
    {
      "epoch": 7.292891501870657,
      "grad_norm": 1.927122712135315,
      "learning_rate": 3.403649307788414e-05,
      "loss": 1.2616,
      "step": 27290
    },
    {
      "epoch": 7.295563869588455,
      "grad_norm": 2.201514482498169,
      "learning_rate": 3.39734169954793e-05,
      "loss": 1.2214,
      "step": 27300
    },
    {
      "epoch": 7.298236237306253,
      "grad_norm": 2.097895622253418,
      "learning_rate": 3.3910387451423876e-05,
      "loss": 1.2455,
      "step": 27310
    },
    {
      "epoch": 7.300908605024051,
      "grad_norm": 2.2148094177246094,
      "learning_rate": 3.3847404490143755e-05,
      "loss": 1.2837,
      "step": 27320
    },
    {
      "epoch": 7.303580972741849,
      "grad_norm": 2.0491528511047363,
      "learning_rate": 3.378446815603205e-05,
      "loss": 1.3052,
      "step": 27330
    },
    {
      "epoch": 7.306253340459647,
      "grad_norm": 2.043707847595215,
      "learning_rate": 3.372157849344887e-05,
      "loss": 1.1766,
      "step": 27340
    },
    {
      "epoch": 7.308925708177445,
      "grad_norm": 2.0265252590179443,
      "learning_rate": 3.3658735546721664e-05,
      "loss": 1.3366,
      "step": 27350
    },
    {
      "epoch": 7.311598075895243,
      "grad_norm": 2.085498571395874,
      "learning_rate": 3.359593936014469e-05,
      "loss": 1.1958,
      "step": 27360
    },
    {
      "epoch": 7.314270443613041,
      "grad_norm": 2.19191837310791,
      "learning_rate": 3.353318997797951e-05,
      "loss": 1.2541,
      "step": 27370
    },
    {
      "epoch": 7.316942811330839,
      "grad_norm": 2.31546688079834,
      "learning_rate": 3.3470487444454446e-05,
      "loss": 1.2631,
      "step": 27380
    },
    {
      "epoch": 7.319615179048637,
      "grad_norm": 1.9966678619384766,
      "learning_rate": 3.340783180376498e-05,
      "loss": 1.2182,
      "step": 27390
    },
    {
      "epoch": 7.322287546766435,
      "grad_norm": 2.0418057441711426,
      "learning_rate": 3.334522310007345e-05,
      "loss": 1.1623,
      "step": 27400
    },
    {
      "epoch": 7.324959914484233,
      "grad_norm": 2.1081957817077637,
      "learning_rate": 3.328266137750918e-05,
      "loss": 1.2262,
      "step": 27410
    },
    {
      "epoch": 7.327632282202031,
      "grad_norm": 2.1674318313598633,
      "learning_rate": 3.3220146680168287e-05,
      "loss": 1.288,
      "step": 27420
    },
    {
      "epoch": 7.330304649919829,
      "grad_norm": 2.1051673889160156,
      "learning_rate": 3.31576790521138e-05,
      "loss": 1.2202,
      "step": 27430
    },
    {
      "epoch": 7.332977017637627,
      "grad_norm": 2.1748886108398438,
      "learning_rate": 3.309525853737557e-05,
      "loss": 1.3121,
      "step": 27440
    },
    {
      "epoch": 7.335649385355425,
      "grad_norm": 2.2035019397735596,
      "learning_rate": 3.303288517995026e-05,
      "loss": 1.1804,
      "step": 27450
    },
    {
      "epoch": 7.338321753073223,
      "grad_norm": 2.1828949451446533,
      "learning_rate": 3.297055902380129e-05,
      "loss": 1.2538,
      "step": 27460
    },
    {
      "epoch": 7.340994120791021,
      "grad_norm": 1.8600070476531982,
      "learning_rate": 3.290828011285868e-05,
      "loss": 1.2314,
      "step": 27470
    },
    {
      "epoch": 7.343666488508819,
      "grad_norm": 2.1104629039764404,
      "learning_rate": 3.28460484910194e-05,
      "loss": 1.2091,
      "step": 27480
    },
    {
      "epoch": 7.346338856226617,
      "grad_norm": 1.9043546915054321,
      "learning_rate": 3.278386420214685e-05,
      "loss": 1.1755,
      "step": 27490
    },
    {
      "epoch": 7.349011223944415,
      "grad_norm": 2.1234591007232666,
      "learning_rate": 3.272172729007121e-05,
      "loss": 1.2054,
      "step": 27500
    },
    {
      "epoch": 7.351683591662213,
      "grad_norm": 2.143435478210449,
      "learning_rate": 3.265963779858922e-05,
      "loss": 1.3058,
      "step": 27510
    },
    {
      "epoch": 7.354355959380011,
      "grad_norm": 2.012240171432495,
      "learning_rate": 3.2597595771464185e-05,
      "loss": 1.1978,
      "step": 27520
    },
    {
      "epoch": 7.357028327097809,
      "grad_norm": 2.1521317958831787,
      "learning_rate": 3.253560125242597e-05,
      "loss": 1.1987,
      "step": 27530
    },
    {
      "epoch": 7.359700694815607,
      "grad_norm": 2.0875651836395264,
      "learning_rate": 3.2473654285171004e-05,
      "loss": 1.2867,
      "step": 27540
    },
    {
      "epoch": 7.362373062533405,
      "grad_norm": 2.0559356212615967,
      "learning_rate": 3.241175491336206e-05,
      "loss": 1.2875,
      "step": 27550
    },
    {
      "epoch": 7.365045430251203,
      "grad_norm": 2.1649205684661865,
      "learning_rate": 3.234990318062855e-05,
      "loss": 1.2321,
      "step": 27560
    },
    {
      "epoch": 7.367717797969001,
      "grad_norm": 2.1024770736694336,
      "learning_rate": 3.228809913056613e-05,
      "loss": 1.2764,
      "step": 27570
    },
    {
      "epoch": 7.370390165686798,
      "grad_norm": 2.1040284633636475,
      "learning_rate": 3.222634280673696e-05,
      "loss": 1.2064,
      "step": 27580
    },
    {
      "epoch": 7.373062533404596,
      "grad_norm": 2.161027431488037,
      "learning_rate": 3.2164634252669535e-05,
      "loss": 1.2076,
      "step": 27590
    },
    {
      "epoch": 7.375734901122394,
      "grad_norm": 1.9908994436264038,
      "learning_rate": 3.210297351185867e-05,
      "loss": 1.2856,
      "step": 27600
    },
    {
      "epoch": 7.378407268840192,
      "grad_norm": 2.054513454437256,
      "learning_rate": 3.2041360627765504e-05,
      "loss": 1.2207,
      "step": 27610
    },
    {
      "epoch": 7.38107963655799,
      "grad_norm": 2.0925278663635254,
      "learning_rate": 3.197979564381738e-05,
      "loss": 1.2605,
      "step": 27620
    },
    {
      "epoch": 7.383752004275788,
      "grad_norm": 2.0387942790985107,
      "learning_rate": 3.191827860340794e-05,
      "loss": 1.1624,
      "step": 27630
    },
    {
      "epoch": 7.386424371993586,
      "grad_norm": 1.9733607769012451,
      "learning_rate": 3.1856809549897e-05,
      "loss": 1.2643,
      "step": 27640
    },
    {
      "epoch": 7.389096739711384,
      "grad_norm": 2.050025463104248,
      "learning_rate": 3.1795388526610633e-05,
      "loss": 1.2614,
      "step": 27650
    },
    {
      "epoch": 7.391769107429182,
      "grad_norm": 2.247288227081299,
      "learning_rate": 3.1734015576840895e-05,
      "loss": 1.16,
      "step": 27660
    },
    {
      "epoch": 7.39444147514698,
      "grad_norm": 2.1705844402313232,
      "learning_rate": 3.167269074384616e-05,
      "loss": 1.2709,
      "step": 27670
    },
    {
      "epoch": 7.397113842864778,
      "grad_norm": 2.0863277912139893,
      "learning_rate": 3.1611414070850666e-05,
      "loss": 1.2264,
      "step": 27680
    },
    {
      "epoch": 7.399786210582576,
      "grad_norm": 2.080504894256592,
      "learning_rate": 3.155018560104496e-05,
      "loss": 1.1547,
      "step": 27690
    },
    {
      "epoch": 7.402458578300374,
      "grad_norm": 2.1259548664093018,
      "learning_rate": 3.148900537758537e-05,
      "loss": 1.2138,
      "step": 27700
    },
    {
      "epoch": 7.405130946018172,
      "grad_norm": 2.1991636753082275,
      "learning_rate": 3.142787344359436e-05,
      "loss": 1.2567,
      "step": 27710
    },
    {
      "epoch": 7.40780331373597,
      "grad_norm": 2.322218418121338,
      "learning_rate": 3.136678984216033e-05,
      "loss": 1.323,
      "step": 27720
    },
    {
      "epoch": 7.410475681453768,
      "grad_norm": 2.108767032623291,
      "learning_rate": 3.1305754616337624e-05,
      "loss": 1.2391,
      "step": 27730
    },
    {
      "epoch": 7.413148049171566,
      "grad_norm": 2.124277353286743,
      "learning_rate": 3.124476780914638e-05,
      "loss": 1.2679,
      "step": 27740
    },
    {
      "epoch": 7.415820416889364,
      "grad_norm": 2.0178637504577637,
      "learning_rate": 3.118382946357282e-05,
      "loss": 1.2728,
      "step": 27750
    },
    {
      "epoch": 7.418492784607162,
      "grad_norm": 2.0202298164367676,
      "learning_rate": 3.1122939622568816e-05,
      "loss": 1.242,
      "step": 27760
    },
    {
      "epoch": 7.42116515232496,
      "grad_norm": 2.028144598007202,
      "learning_rate": 3.1062098329052124e-05,
      "loss": 1.2401,
      "step": 27770
    },
    {
      "epoch": 7.423837520042758,
      "grad_norm": 1.9863171577453613,
      "learning_rate": 3.100130562590631e-05,
      "loss": 1.2009,
      "step": 27780
    },
    {
      "epoch": 7.426509887760556,
      "grad_norm": 2.0276663303375244,
      "learning_rate": 3.094056155598063e-05,
      "loss": 1.2562,
      "step": 27790
    },
    {
      "epoch": 7.429182255478354,
      "grad_norm": 2.038090467453003,
      "learning_rate": 3.0879866162090145e-05,
      "loss": 1.2114,
      "step": 27800
    },
    {
      "epoch": 7.431854623196152,
      "grad_norm": 1.9752060174942017,
      "learning_rate": 3.081921948701548e-05,
      "loss": 1.2638,
      "step": 27810
    },
    {
      "epoch": 7.43452699091395,
      "grad_norm": 2.0179131031036377,
      "learning_rate": 3.075862157350304e-05,
      "loss": 1.2521,
      "step": 27820
    },
    {
      "epoch": 7.437199358631748,
      "grad_norm": 2.139413356781006,
      "learning_rate": 3.06980724642648e-05,
      "loss": 1.192,
      "step": 27830
    },
    {
      "epoch": 7.439871726349546,
      "grad_norm": 2.0572116374969482,
      "learning_rate": 3.0637572201978384e-05,
      "loss": 1.2547,
      "step": 27840
    },
    {
      "epoch": 7.442544094067344,
      "grad_norm": 2.1968374252319336,
      "learning_rate": 3.057712082928692e-05,
      "loss": 1.2545,
      "step": 27850
    },
    {
      "epoch": 7.445216461785142,
      "grad_norm": 2.1138153076171875,
      "learning_rate": 3.051671838879917e-05,
      "loss": 1.2324,
      "step": 27860
    },
    {
      "epoch": 7.44788882950294,
      "grad_norm": 1.9967312812805176,
      "learning_rate": 3.0456364923089232e-05,
      "loss": 1.224,
      "step": 27870
    },
    {
      "epoch": 7.450561197220738,
      "grad_norm": 2.2048964500427246,
      "learning_rate": 3.039606047469694e-05,
      "loss": 1.2165,
      "step": 27880
    },
    {
      "epoch": 7.453233564938536,
      "grad_norm": 2.059983730316162,
      "learning_rate": 3.0335805086127332e-05,
      "loss": 1.2961,
      "step": 27890
    },
    {
      "epoch": 7.455905932656334,
      "grad_norm": 1.978102445602417,
      "learning_rate": 3.0275598799851023e-05,
      "loss": 1.2161,
      "step": 27900
    },
    {
      "epoch": 7.458578300374132,
      "grad_norm": 2.2445929050445557,
      "learning_rate": 3.0215441658303955e-05,
      "loss": 1.3037,
      "step": 27910
    },
    {
      "epoch": 7.461250668091929,
      "grad_norm": 2.0911028385162354,
      "learning_rate": 3.0155333703887455e-05,
      "loss": 1.2744,
      "step": 27920
    },
    {
      "epoch": 7.463923035809727,
      "grad_norm": 2.2235841751098633,
      "learning_rate": 3.009527497896817e-05,
      "loss": 1.2155,
      "step": 27930
    },
    {
      "epoch": 7.466595403527525,
      "grad_norm": 2.1373989582061768,
      "learning_rate": 3.0035265525878065e-05,
      "loss": 1.1749,
      "step": 27940
    },
    {
      "epoch": 7.469267771245323,
      "grad_norm": 2.1481776237487793,
      "learning_rate": 2.997530538691431e-05,
      "loss": 1.2451,
      "step": 27950
    },
    {
      "epoch": 7.471940138963121,
      "grad_norm": 2.2615137100219727,
      "learning_rate": 2.99153946043394e-05,
      "loss": 1.2551,
      "step": 27960
    },
    {
      "epoch": 7.474612506680919,
      "grad_norm": 2.2867887020111084,
      "learning_rate": 2.9855533220380994e-05,
      "loss": 1.2871,
      "step": 27970
    },
    {
      "epoch": 7.477284874398717,
      "grad_norm": 2.131208658218384,
      "learning_rate": 2.9795721277231957e-05,
      "loss": 1.2874,
      "step": 27980
    },
    {
      "epoch": 7.479957242116515,
      "grad_norm": 2.4127533435821533,
      "learning_rate": 2.9735958817050335e-05,
      "loss": 1.2257,
      "step": 27990
    },
    {
      "epoch": 7.482629609834313,
      "grad_norm": 2.0967767238616943,
      "learning_rate": 2.967624588195914e-05,
      "loss": 1.2808,
      "step": 28000
    },
    {
      "epoch": 7.485301977552111,
      "grad_norm": 2.086289405822754,
      "learning_rate": 2.9616582514046743e-05,
      "loss": 1.1906,
      "step": 28010
    },
    {
      "epoch": 7.487974345269909,
      "grad_norm": 1.8830486536026,
      "learning_rate": 2.955696875536631e-05,
      "loss": 1.2248,
      "step": 28020
    },
    {
      "epoch": 7.490646712987707,
      "grad_norm": 2.2826850414276123,
      "learning_rate": 2.949740464793621e-05,
      "loss": 1.2467,
      "step": 28030
    },
    {
      "epoch": 7.493319080705505,
      "grad_norm": 2.0717697143554688,
      "learning_rate": 2.9437890233739762e-05,
      "loss": 1.2972,
      "step": 28040
    },
    {
      "epoch": 7.495991448423303,
      "grad_norm": 2.160871744155884,
      "learning_rate": 2.9378425554725274e-05,
      "loss": 1.1881,
      "step": 28050
    },
    {
      "epoch": 7.498663816141101,
      "grad_norm": 2.1282119750976562,
      "learning_rate": 2.9319010652805913e-05,
      "loss": 1.2222,
      "step": 28060
    },
    {
      "epoch": 7.501336183858899,
      "grad_norm": 2.094926118850708,
      "learning_rate": 2.925964556985995e-05,
      "loss": 1.2999,
      "step": 28070
    },
    {
      "epoch": 7.504008551576697,
      "grad_norm": 2.021768093109131,
      "learning_rate": 2.920033034773031e-05,
      "loss": 1.2802,
      "step": 28080
    },
    {
      "epoch": 7.506680919294495,
      "grad_norm": 2.03518009185791,
      "learning_rate": 2.914106502822499e-05,
      "loss": 1.2134,
      "step": 28090
    },
    {
      "epoch": 7.509353287012293,
      "grad_norm": 2.202364206314087,
      "learning_rate": 2.9081849653116644e-05,
      "loss": 1.2066,
      "step": 28100
    },
    {
      "epoch": 7.512025654730091,
      "grad_norm": 2.1216659545898438,
      "learning_rate": 2.902268426414282e-05,
      "loss": 1.1717,
      "step": 28110
    },
    {
      "epoch": 7.514698022447889,
      "grad_norm": 2.0889415740966797,
      "learning_rate": 2.896356890300579e-05,
      "loss": 1.2003,
      "step": 28120
    },
    {
      "epoch": 7.517370390165687,
      "grad_norm": 2.193819284439087,
      "learning_rate": 2.8904503611372612e-05,
      "loss": 1.2026,
      "step": 28130
    },
    {
      "epoch": 7.520042757883485,
      "grad_norm": 2.163248062133789,
      "learning_rate": 2.8845488430874944e-05,
      "loss": 1.2555,
      "step": 28140
    },
    {
      "epoch": 7.522715125601283,
      "grad_norm": 2.0415689945220947,
      "learning_rate": 2.878652340310932e-05,
      "loss": 1.2142,
      "step": 28150
    },
    {
      "epoch": 7.525387493319081,
      "grad_norm": 1.9740504026412964,
      "learning_rate": 2.87276085696367e-05,
      "loss": 1.23,
      "step": 28160
    },
    {
      "epoch": 7.528059861036879,
      "grad_norm": 2.0317816734313965,
      "learning_rate": 2.8668743971982816e-05,
      "loss": 1.2271,
      "step": 28170
    },
    {
      "epoch": 7.530732228754677,
      "grad_norm": 2.0579888820648193,
      "learning_rate": 2.8609929651637925e-05,
      "loss": 1.2266,
      "step": 28180
    },
    {
      "epoch": 7.533404596472475,
      "grad_norm": 2.053981065750122,
      "learning_rate": 2.855116565005689e-05,
      "loss": 1.1658,
      "step": 28190
    },
    {
      "epoch": 7.536076964190273,
      "grad_norm": 2.0381927490234375,
      "learning_rate": 2.8492452008659086e-05,
      "loss": 1.2244,
      "step": 28200
    },
    {
      "epoch": 7.53874933190807,
      "grad_norm": 2.3309037685394287,
      "learning_rate": 2.8433788768828363e-05,
      "loss": 1.2272,
      "step": 28210
    },
    {
      "epoch": 7.541421699625868,
      "grad_norm": 1.9979252815246582,
      "learning_rate": 2.8375175971913072e-05,
      "loss": 1.1987,
      "step": 28220
    },
    {
      "epoch": 7.544094067343666,
      "grad_norm": 2.2014684677124023,
      "learning_rate": 2.8316613659226022e-05,
      "loss": 1.202,
      "step": 28230
    },
    {
      "epoch": 7.546766435061464,
      "grad_norm": 2.075988531112671,
      "learning_rate": 2.825810187204443e-05,
      "loss": 1.239,
      "step": 28240
    },
    {
      "epoch": 7.549438802779262,
      "grad_norm": 2.1374521255493164,
      "learning_rate": 2.8199640651609883e-05,
      "loss": 1.2426,
      "step": 28250
    },
    {
      "epoch": 7.55211117049706,
      "grad_norm": 2.334420680999756,
      "learning_rate": 2.8141230039128395e-05,
      "loss": 1.2005,
      "step": 28260
    },
    {
      "epoch": 7.554783538214858,
      "grad_norm": 2.030236005783081,
      "learning_rate": 2.808287007577014e-05,
      "loss": 1.2058,
      "step": 28270
    },
    {
      "epoch": 7.557455905932656,
      "grad_norm": 2.1864736080169678,
      "learning_rate": 2.802456080266984e-05,
      "loss": 1.2464,
      "step": 28280
    },
    {
      "epoch": 7.560128273650454,
      "grad_norm": 2.1882996559143066,
      "learning_rate": 2.796630226092627e-05,
      "loss": 1.272,
      "step": 28290
    },
    {
      "epoch": 7.562800641368252,
      "grad_norm": 2.258795976638794,
      "learning_rate": 2.7908094491602556e-05,
      "loss": 1.2588,
      "step": 28300
    },
    {
      "epoch": 7.56547300908605,
      "grad_norm": 2.1145286560058594,
      "learning_rate": 2.784993753572601e-05,
      "loss": 1.2809,
      "step": 28310
    },
    {
      "epoch": 7.568145376803848,
      "grad_norm": 2.2491462230682373,
      "learning_rate": 2.779183143428814e-05,
      "loss": 1.2408,
      "step": 28320
    },
    {
      "epoch": 7.570817744521646,
      "grad_norm": 2.005479097366333,
      "learning_rate": 2.773377622824459e-05,
      "loss": 1.2191,
      "step": 28330
    },
    {
      "epoch": 7.573490112239444,
      "grad_norm": 2.2879302501678467,
      "learning_rate": 2.7675771958515196e-05,
      "loss": 1.284,
      "step": 28340
    },
    {
      "epoch": 7.576162479957242,
      "grad_norm": 2.1377029418945312,
      "learning_rate": 2.7617818665983762e-05,
      "loss": 1.1812,
      "step": 28350
    },
    {
      "epoch": 7.57883484767504,
      "grad_norm": 2.1379306316375732,
      "learning_rate": 2.7559916391498276e-05,
      "loss": 1.3124,
      "step": 28360
    },
    {
      "epoch": 7.581507215392838,
      "grad_norm": 2.143080472946167,
      "learning_rate": 2.7502065175870718e-05,
      "loss": 1.184,
      "step": 28370
    },
    {
      "epoch": 7.584179583110636,
      "grad_norm": 2.1253340244293213,
      "learning_rate": 2.7444265059877106e-05,
      "loss": 1.1638,
      "step": 28380
    },
    {
      "epoch": 7.586851950828434,
      "grad_norm": 2.110478162765503,
      "learning_rate": 2.738651608425744e-05,
      "loss": 1.183,
      "step": 28390
    },
    {
      "epoch": 7.589524318546232,
      "grad_norm": 1.9841487407684326,
      "learning_rate": 2.7328818289715576e-05,
      "loss": 1.2086,
      "step": 28400
    },
    {
      "epoch": 7.59219668626403,
      "grad_norm": 2.1448872089385986,
      "learning_rate": 2.7271171716919496e-05,
      "loss": 1.3087,
      "step": 28410
    },
    {
      "epoch": 7.594869053981828,
      "grad_norm": 2.0087194442749023,
      "learning_rate": 2.7213576406500873e-05,
      "loss": 1.2426,
      "step": 28420
    },
    {
      "epoch": 7.597541421699626,
      "grad_norm": 2.205631971359253,
      "learning_rate": 2.7156032399055375e-05,
      "loss": 1.2845,
      "step": 28430
    },
    {
      "epoch": 7.600213789417424,
      "grad_norm": 2.080989122390747,
      "learning_rate": 2.7098539735142448e-05,
      "loss": 1.2443,
      "step": 28440
    },
    {
      "epoch": 7.602886157135222,
      "grad_norm": 2.1031556129455566,
      "learning_rate": 2.704109845528542e-05,
      "loss": 1.2825,
      "step": 28450
    },
    {
      "epoch": 7.60555852485302,
      "grad_norm": 2.1243724822998047,
      "learning_rate": 2.6983708599971257e-05,
      "loss": 1.2246,
      "step": 28460
    },
    {
      "epoch": 7.608230892570818,
      "grad_norm": 2.1016321182250977,
      "learning_rate": 2.6926370209650898e-05,
      "loss": 1.2626,
      "step": 28470
    },
    {
      "epoch": 7.610903260288616,
      "grad_norm": 2.1604385375976562,
      "learning_rate": 2.6869083324738764e-05,
      "loss": 1.2269,
      "step": 28480
    },
    {
      "epoch": 7.613575628006414,
      "grad_norm": 2.0251622200012207,
      "learning_rate": 2.6811847985613204e-05,
      "loss": 1.1951,
      "step": 28490
    },
    {
      "epoch": 7.616247995724212,
      "grad_norm": 2.0509791374206543,
      "learning_rate": 2.6754664232616045e-05,
      "loss": 1.2121,
      "step": 28500
    },
    {
      "epoch": 7.61892036344201,
      "grad_norm": 2.0376598834991455,
      "learning_rate": 2.669753210605286e-05,
      "loss": 1.1605,
      "step": 28510
    },
    {
      "epoch": 7.621592731159808,
      "grad_norm": 2.168583631515503,
      "learning_rate": 2.66404516461928e-05,
      "loss": 1.2685,
      "step": 28520
    },
    {
      "epoch": 7.624265098877606,
      "grad_norm": 2.124511480331421,
      "learning_rate": 2.6583422893268638e-05,
      "loss": 1.2421,
      "step": 28530
    },
    {
      "epoch": 7.626937466595404,
      "grad_norm": 2.171947479248047,
      "learning_rate": 2.6526445887476613e-05,
      "loss": 1.288,
      "step": 28540
    },
    {
      "epoch": 7.629609834313202,
      "grad_norm": 2.1840152740478516,
      "learning_rate": 2.646952066897658e-05,
      "loss": 1.2568,
      "step": 28550
    },
    {
      "epoch": 7.632282202031,
      "grad_norm": 2.0040221214294434,
      "learning_rate": 2.6412647277891846e-05,
      "loss": 1.2027,
      "step": 28560
    },
    {
      "epoch": 7.634954569748797,
      "grad_norm": 2.1620066165924072,
      "learning_rate": 2.6355825754309217e-05,
      "loss": 1.2359,
      "step": 28570
    },
    {
      "epoch": 7.637626937466595,
      "grad_norm": 2.356335401535034,
      "learning_rate": 2.6299056138278945e-05,
      "loss": 1.2326,
      "step": 28580
    },
    {
      "epoch": 7.640299305184393,
      "grad_norm": 2.1007184982299805,
      "learning_rate": 2.6242338469814575e-05,
      "loss": 1.2403,
      "step": 28590
    },
    {
      "epoch": 7.642971672902191,
      "grad_norm": 2.003970146179199,
      "learning_rate": 2.618567278889328e-05,
      "loss": 1.3214,
      "step": 28600
    },
    {
      "epoch": 7.645644040619989,
      "grad_norm": 2.0494182109832764,
      "learning_rate": 2.6129059135455335e-05,
      "loss": 1.2185,
      "step": 28610
    },
    {
      "epoch": 7.648316408337787,
      "grad_norm": 2.0207862854003906,
      "learning_rate": 2.6072497549404506e-05,
      "loss": 1.244,
      "step": 28620
    },
    {
      "epoch": 7.650988776055585,
      "grad_norm": 1.914623737335205,
      "learning_rate": 2.601598807060779e-05,
      "loss": 1.2497,
      "step": 28630
    },
    {
      "epoch": 7.653661143773383,
      "grad_norm": 2.183497190475464,
      "learning_rate": 2.5959530738895486e-05,
      "loss": 1.1827,
      "step": 28640
    },
    {
      "epoch": 7.656333511491181,
      "grad_norm": 2.0640687942504883,
      "learning_rate": 2.5903125594061128e-05,
      "loss": 1.3111,
      "step": 28650
    },
    {
      "epoch": 7.659005879208979,
      "grad_norm": 2.200634241104126,
      "learning_rate": 2.5846772675861485e-05,
      "loss": 1.2258,
      "step": 28660
    },
    {
      "epoch": 7.661678246926777,
      "grad_norm": 2.038461685180664,
      "learning_rate": 2.579047202401642e-05,
      "loss": 1.289,
      "step": 28670
    },
    {
      "epoch": 7.664350614644575,
      "grad_norm": 1.9681406021118164,
      "learning_rate": 2.5734223678209135e-05,
      "loss": 1.3086,
      "step": 28680
    },
    {
      "epoch": 7.667022982362373,
      "grad_norm": 2.1332638263702393,
      "learning_rate": 2.5678027678085793e-05,
      "loss": 1.3003,
      "step": 28690
    },
    {
      "epoch": 7.669695350080171,
      "grad_norm": 2.178931713104248,
      "learning_rate": 2.5621884063255742e-05,
      "loss": 1.2336,
      "step": 28700
    },
    {
      "epoch": 7.672367717797969,
      "grad_norm": 2.1359126567840576,
      "learning_rate": 2.5565792873291395e-05,
      "loss": 1.2421,
      "step": 28710
    },
    {
      "epoch": 7.675040085515767,
      "grad_norm": 2.1914706230163574,
      "learning_rate": 2.5509754147728226e-05,
      "loss": 1.2467,
      "step": 28720
    },
    {
      "epoch": 7.677712453233565,
      "grad_norm": 2.121519088745117,
      "learning_rate": 2.545376792606473e-05,
      "loss": 1.2759,
      "step": 28730
    },
    {
      "epoch": 7.680384820951363,
      "grad_norm": 2.1237680912017822,
      "learning_rate": 2.539783424776233e-05,
      "loss": 1.2678,
      "step": 28740
    },
    {
      "epoch": 7.683057188669161,
      "grad_norm": 2.0890979766845703,
      "learning_rate": 2.5341953152245502e-05,
      "loss": 1.2993,
      "step": 28750
    },
    {
      "epoch": 7.685729556386959,
      "grad_norm": 2.0381410121917725,
      "learning_rate": 2.5286124678901624e-05,
      "loss": 1.2441,
      "step": 28760
    },
    {
      "epoch": 7.688401924104757,
      "grad_norm": 2.1491963863372803,
      "learning_rate": 2.5230348867081e-05,
      "loss": 1.2417,
      "step": 28770
    },
    {
      "epoch": 7.691074291822555,
      "grad_norm": 2.4194345474243164,
      "learning_rate": 2.5174625756096716e-05,
      "loss": 1.1738,
      "step": 28780
    },
    {
      "epoch": 7.693746659540353,
      "grad_norm": 2.290524959564209,
      "learning_rate": 2.5118955385224907e-05,
      "loss": 1.2388,
      "step": 28790
    },
    {
      "epoch": 7.696419027258151,
      "grad_norm": 1.9745029211044312,
      "learning_rate": 2.506333779370432e-05,
      "loss": 1.2206,
      "step": 28800
    },
    {
      "epoch": 7.699091394975949,
      "grad_norm": 2.3277316093444824,
      "learning_rate": 2.5007773020736712e-05,
      "loss": 1.2415,
      "step": 28810
    },
    {
      "epoch": 7.701763762693747,
      "grad_norm": 2.0364115238189697,
      "learning_rate": 2.495226110548642e-05,
      "loss": 1.1968,
      "step": 28820
    },
    {
      "epoch": 7.704436130411545,
      "grad_norm": 2.0834743976593018,
      "learning_rate": 2.489680208708063e-05,
      "loss": 1.2205,
      "step": 28830
    },
    {
      "epoch": 7.707108498129343,
      "grad_norm": 2.029390573501587,
      "learning_rate": 2.484139600460922e-05,
      "loss": 1.2304,
      "step": 28840
    },
    {
      "epoch": 7.709780865847141,
      "grad_norm": 2.1209352016448975,
      "learning_rate": 2.4786042897124807e-05,
      "loss": 1.2894,
      "step": 28850
    },
    {
      "epoch": 7.712453233564938,
      "grad_norm": 2.161252498626709,
      "learning_rate": 2.4730742803642504e-05,
      "loss": 1.2743,
      "step": 28860
    },
    {
      "epoch": 7.715125601282736,
      "grad_norm": 2.150007963180542,
      "learning_rate": 2.4675495763140298e-05,
      "loss": 1.2482,
      "step": 28870
    },
    {
      "epoch": 7.717797969000534,
      "grad_norm": 1.979568600654602,
      "learning_rate": 2.462030181455859e-05,
      "loss": 1.3048,
      "step": 28880
    },
    {
      "epoch": 7.720470336718332,
      "grad_norm": 2.0801188945770264,
      "learning_rate": 2.4565160996800428e-05,
      "loss": 1.2267,
      "step": 28890
    },
    {
      "epoch": 7.72314270443613,
      "grad_norm": 2.001863718032837,
      "learning_rate": 2.4510073348731434e-05,
      "loss": 1.2412,
      "step": 28900
    },
    {
      "epoch": 7.725815072153928,
      "grad_norm": 2.247753381729126,
      "learning_rate": 2.4455038909179717e-05,
      "loss": 1.2977,
      "step": 28910
    },
    {
      "epoch": 7.728487439871726,
      "grad_norm": 1.9314312934875488,
      "learning_rate": 2.440005771693593e-05,
      "loss": 1.1992,
      "step": 28920
    },
    {
      "epoch": 7.731159807589524,
      "grad_norm": 2.0058915615081787,
      "learning_rate": 2.4345129810753087e-05,
      "loss": 1.2119,
      "step": 28930
    },
    {
      "epoch": 7.733832175307322,
      "grad_norm": 2.0803093910217285,
      "learning_rate": 2.429025522934677e-05,
      "loss": 1.263,
      "step": 28940
    },
    {
      "epoch": 7.73650454302512,
      "grad_norm": 2.030503273010254,
      "learning_rate": 2.423543401139491e-05,
      "loss": 1.1688,
      "step": 28950
    },
    {
      "epoch": 7.739176910742918,
      "grad_norm": 2.104008913040161,
      "learning_rate": 2.4180666195537838e-05,
      "loss": 1.1846,
      "step": 28960
    },
    {
      "epoch": 7.741849278460716,
      "grad_norm": 2.021791696548462,
      "learning_rate": 2.4125951820378235e-05,
      "loss": 1.2346,
      "step": 28970
    },
    {
      "epoch": 7.744521646178514,
      "grad_norm": 2.2479772567749023,
      "learning_rate": 2.407129092448117e-05,
      "loss": 1.2751,
      "step": 28980
    },
    {
      "epoch": 7.747194013896312,
      "grad_norm": 2.1344785690307617,
      "learning_rate": 2.4016683546373886e-05,
      "loss": 1.2018,
      "step": 28990
    },
    {
      "epoch": 7.74986638161411,
      "grad_norm": 2.1255135536193848,
      "learning_rate": 2.396212972454609e-05,
      "loss": 1.2589,
      "step": 29000
    },
    {
      "epoch": 7.752538749331908,
      "grad_norm": 2.1533989906311035,
      "learning_rate": 2.390762949744956e-05,
      "loss": 1.2402,
      "step": 29010
    },
    {
      "epoch": 7.755211117049706,
      "grad_norm": 2.156620740890503,
      "learning_rate": 2.3853182903498404e-05,
      "loss": 1.2419,
      "step": 29020
    },
    {
      "epoch": 7.757883484767504,
      "grad_norm": 1.9978028535842896,
      "learning_rate": 2.3798789981068914e-05,
      "loss": 1.2721,
      "step": 29030
    },
    {
      "epoch": 7.760555852485302,
      "grad_norm": 2.0848798751831055,
      "learning_rate": 2.3744450768499538e-05,
      "loss": 1.2698,
      "step": 29040
    },
    {
      "epoch": 7.7632282202031,
      "grad_norm": 2.175245761871338,
      "learning_rate": 2.369016530409085e-05,
      "loss": 1.2468,
      "step": 29050
    },
    {
      "epoch": 7.765900587920898,
      "grad_norm": 2.1696841716766357,
      "learning_rate": 2.363593362610561e-05,
      "loss": 1.2172,
      "step": 29060
    },
    {
      "epoch": 7.768572955638696,
      "grad_norm": 1.9714295864105225,
      "learning_rate": 2.3581755772768543e-05,
      "loss": 1.3093,
      "step": 29070
    },
    {
      "epoch": 7.771245323356494,
      "grad_norm": 2.104121208190918,
      "learning_rate": 2.3527631782266544e-05,
      "loss": 1.2086,
      "step": 29080
    },
    {
      "epoch": 7.773917691074292,
      "grad_norm": 2.087700366973877,
      "learning_rate": 2.3473561692748513e-05,
      "loss": 1.2256,
      "step": 29090
    },
    {
      "epoch": 7.77659005879209,
      "grad_norm": 2.0084388256073,
      "learning_rate": 2.3419545542325338e-05,
      "loss": 1.2699,
      "step": 29100
    },
    {
      "epoch": 7.779262426509888,
      "grad_norm": 2.2062480449676514,
      "learning_rate": 2.3365583369069943e-05,
      "loss": 1.2715,
      "step": 29110
    },
    {
      "epoch": 7.781934794227686,
      "grad_norm": 2.1122920513153076,
      "learning_rate": 2.331167521101708e-05,
      "loss": 1.2422,
      "step": 29120
    },
    {
      "epoch": 7.784607161945484,
      "grad_norm": 2.2014482021331787,
      "learning_rate": 2.3257821106163634e-05,
      "loss": 1.2173,
      "step": 29130
    },
    {
      "epoch": 7.787279529663282,
      "grad_norm": 2.2499282360076904,
      "learning_rate": 2.3204021092468186e-05,
      "loss": 1.2048,
      "step": 29140
    },
    {
      "epoch": 7.78995189738108,
      "grad_norm": 2.075568199157715,
      "learning_rate": 2.3150275207851303e-05,
      "loss": 1.2639,
      "step": 29150
    },
    {
      "epoch": 7.792624265098878,
      "grad_norm": 2.0956461429595947,
      "learning_rate": 2.3096583490195377e-05,
      "loss": 1.2731,
      "step": 29160
    },
    {
      "epoch": 7.795296632816676,
      "grad_norm": 2.10376238822937,
      "learning_rate": 2.3042945977344642e-05,
      "loss": 1.2319,
      "step": 29170
    },
    {
      "epoch": 7.797969000534474,
      "grad_norm": 2.21205472946167,
      "learning_rate": 2.2989362707105032e-05,
      "loss": 1.3066,
      "step": 29180
    },
    {
      "epoch": 7.800641368252272,
      "grad_norm": 2.1282801628112793,
      "learning_rate": 2.2935833717244415e-05,
      "loss": 1.2209,
      "step": 29190
    },
    {
      "epoch": 7.80331373597007,
      "grad_norm": 2.08268666267395,
      "learning_rate": 2.2882359045492196e-05,
      "loss": 1.3034,
      "step": 29200
    },
    {
      "epoch": 7.805986103687868,
      "grad_norm": 2.090839385986328,
      "learning_rate": 2.2828938729539728e-05,
      "loss": 1.2736,
      "step": 29210
    },
    {
      "epoch": 7.808658471405665,
      "grad_norm": 2.1238749027252197,
      "learning_rate": 2.2775572807039825e-05,
      "loss": 1.2777,
      "step": 29220
    },
    {
      "epoch": 7.811330839123463,
      "grad_norm": 2.1232802867889404,
      "learning_rate": 2.2722261315607106e-05,
      "loss": 1.2004,
      "step": 29230
    },
    {
      "epoch": 7.814003206841261,
      "grad_norm": 2.057396650314331,
      "learning_rate": 2.2669004292817762e-05,
      "loss": 1.2156,
      "step": 29240
    },
    {
      "epoch": 7.816675574559059,
      "grad_norm": 2.0630247592926025,
      "learning_rate": 2.2615801776209667e-05,
      "loss": 1.1776,
      "step": 29250
    },
    {
      "epoch": 7.819347942276857,
      "grad_norm": 2.2614002227783203,
      "learning_rate": 2.2562653803282142e-05,
      "loss": 1.2957,
      "step": 29260
    },
    {
      "epoch": 7.822020309994655,
      "grad_norm": 2.082691192626953,
      "learning_rate": 2.2509560411496188e-05,
      "loss": 1.2261,
      "step": 29270
    },
    {
      "epoch": 7.824692677712453,
      "grad_norm": 1.936834692955017,
      "learning_rate": 2.2456521638274276e-05,
      "loss": 1.2929,
      "step": 29280
    },
    {
      "epoch": 7.827365045430251,
      "grad_norm": 2.020348310470581,
      "learning_rate": 2.2403537521000406e-05,
      "loss": 1.2206,
      "step": 29290
    },
    {
      "epoch": 7.830037413148049,
      "grad_norm": 2.078829288482666,
      "learning_rate": 2.2350608097020053e-05,
      "loss": 1.2926,
      "step": 29300
    },
    {
      "epoch": 7.832709780865847,
      "grad_norm": 2.0276589393615723,
      "learning_rate": 2.2297733403640075e-05,
      "loss": 1.2183,
      "step": 29310
    },
    {
      "epoch": 7.835382148583645,
      "grad_norm": 2.114861488342285,
      "learning_rate": 2.2244913478128892e-05,
      "loss": 1.2716,
      "step": 29320
    },
    {
      "epoch": 7.838054516301443,
      "grad_norm": 2.007420301437378,
      "learning_rate": 2.2192148357716168e-05,
      "loss": 1.195,
      "step": 29330
    },
    {
      "epoch": 7.840726884019241,
      "grad_norm": 2.0975234508514404,
      "learning_rate": 2.2139438079593034e-05,
      "loss": 1.2404,
      "step": 29340
    },
    {
      "epoch": 7.843399251737039,
      "grad_norm": 2.0706093311309814,
      "learning_rate": 2.2086782680911945e-05,
      "loss": 1.2111,
      "step": 29350
    },
    {
      "epoch": 7.846071619454837,
      "grad_norm": 2.136385679244995,
      "learning_rate": 2.2034182198786666e-05,
      "loss": 1.1672,
      "step": 29360
    },
    {
      "epoch": 7.848743987172635,
      "grad_norm": 2.143287420272827,
      "learning_rate": 2.198163667029226e-05,
      "loss": 1.3171,
      "step": 29370
    },
    {
      "epoch": 7.851416354890433,
      "grad_norm": 1.9991885423660278,
      "learning_rate": 2.1929146132465073e-05,
      "loss": 1.285,
      "step": 29380
    },
    {
      "epoch": 7.854088722608231,
      "grad_norm": 2.1113333702087402,
      "learning_rate": 2.1876710622302598e-05,
      "loss": 1.2908,
      "step": 29390
    },
    {
      "epoch": 7.856761090326029,
      "grad_norm": 2.063178539276123,
      "learning_rate": 2.182433017676374e-05,
      "loss": 1.2944,
      "step": 29400
    },
    {
      "epoch": 7.859433458043827,
      "grad_norm": 2.074934244155884,
      "learning_rate": 2.1772004832768365e-05,
      "loss": 1.2717,
      "step": 29410
    },
    {
      "epoch": 7.862105825761625,
      "grad_norm": 2.060929775238037,
      "learning_rate": 2.1719734627197643e-05,
      "loss": 1.2044,
      "step": 29420
    },
    {
      "epoch": 7.864778193479423,
      "grad_norm": 2.197664737701416,
      "learning_rate": 2.1667519596893836e-05,
      "loss": 1.3214,
      "step": 29430
    },
    {
      "epoch": 7.867450561197221,
      "grad_norm": 2.159766435623169,
      "learning_rate": 2.1615359778660327e-05,
      "loss": 1.2499,
      "step": 29440
    },
    {
      "epoch": 7.870122928915019,
      "grad_norm": 2.094172716140747,
      "learning_rate": 2.156325520926161e-05,
      "loss": 1.2652,
      "step": 29450
    },
    {
      "epoch": 7.872795296632817,
      "grad_norm": 2.1401398181915283,
      "learning_rate": 2.151120592542315e-05,
      "loss": 1.3186,
      "step": 29460
    },
    {
      "epoch": 7.875467664350615,
      "grad_norm": 2.1363375186920166,
      "learning_rate": 2.1459211963831526e-05,
      "loss": 1.2029,
      "step": 29470
    },
    {
      "epoch": 7.878140032068413,
      "grad_norm": 1.9351859092712402,
      "learning_rate": 2.1407273361134293e-05,
      "loss": 1.2713,
      "step": 29480
    },
    {
      "epoch": 7.880812399786211,
      "grad_norm": 2.3286333084106445,
      "learning_rate": 2.1355390153940023e-05,
      "loss": 1.2721,
      "step": 29490
    },
    {
      "epoch": 7.883484767504008,
      "grad_norm": 2.230107545852661,
      "learning_rate": 2.130356237881813e-05,
      "loss": 1.3185,
      "step": 29500
    },
    {
      "epoch": 7.886157135221806,
      "grad_norm": 2.3039634227752686,
      "learning_rate": 2.125179007229915e-05,
      "loss": 1.2812,
      "step": 29510
    },
    {
      "epoch": 7.888829502939604,
      "grad_norm": 2.1231534481048584,
      "learning_rate": 2.120007327087429e-05,
      "loss": 1.2437,
      "step": 29520
    },
    {
      "epoch": 7.891501870657402,
      "grad_norm": 2.1449410915374756,
      "learning_rate": 2.114841201099588e-05,
      "loss": 1.2607,
      "step": 29530
    },
    {
      "epoch": 7.8941742383752,
      "grad_norm": 2.2946348190307617,
      "learning_rate": 2.1096806329076892e-05,
      "loss": 1.3148,
      "step": 29540
    },
    {
      "epoch": 7.896846606092998,
      "grad_norm": 2.03314208984375,
      "learning_rate": 2.104525626149123e-05,
      "loss": 1.1725,
      "step": 29550
    },
    {
      "epoch": 7.899518973810796,
      "grad_norm": 2.103083848953247,
      "learning_rate": 2.099376184457358e-05,
      "loss": 1.2777,
      "step": 29560
    },
    {
      "epoch": 7.902191341528594,
      "grad_norm": 2.1012446880340576,
      "learning_rate": 2.0942323114619432e-05,
      "loss": 1.3446,
      "step": 29570
    },
    {
      "epoch": 7.904863709246392,
      "grad_norm": 2.1700327396392822,
      "learning_rate": 2.0890940107884916e-05,
      "loss": 1.2592,
      "step": 29580
    },
    {
      "epoch": 7.90753607696419,
      "grad_norm": 2.192653179168701,
      "learning_rate": 2.0839612860587076e-05,
      "loss": 1.2216,
      "step": 29590
    },
    {
      "epoch": 7.910208444681988,
      "grad_norm": 2.11545467376709,
      "learning_rate": 2.0788341408903445e-05,
      "loss": 1.173,
      "step": 29600
    },
    {
      "epoch": 7.912880812399786,
      "grad_norm": 2.073657512664795,
      "learning_rate": 2.0737125788972366e-05,
      "loss": 1.1833,
      "step": 29610
    },
    {
      "epoch": 7.915553180117584,
      "grad_norm": 2.1943564414978027,
      "learning_rate": 2.06859660368928e-05,
      "loss": 1.2882,
      "step": 29620
    },
    {
      "epoch": 7.918225547835382,
      "grad_norm": 2.3015801906585693,
      "learning_rate": 2.06348621887243e-05,
      "loss": 1.2154,
      "step": 29630
    },
    {
      "epoch": 7.92089791555318,
      "grad_norm": 2.050428867340088,
      "learning_rate": 2.058381428048708e-05,
      "loss": 1.1586,
      "step": 29640
    },
    {
      "epoch": 7.923570283270978,
      "grad_norm": 2.131366491317749,
      "learning_rate": 2.053282234816182e-05,
      "loss": 1.2523,
      "step": 29650
    },
    {
      "epoch": 7.926242650988776,
      "grad_norm": 2.0764682292938232,
      "learning_rate": 2.048188642768982e-05,
      "loss": 1.218,
      "step": 29660
    },
    {
      "epoch": 7.928915018706574,
      "grad_norm": 2.240628719329834,
      "learning_rate": 2.043100655497291e-05,
      "loss": 1.2841,
      "step": 29670
    },
    {
      "epoch": 7.931587386424372,
      "grad_norm": 2.1216185092926025,
      "learning_rate": 2.038018276587339e-05,
      "loss": 1.2165,
      "step": 29680
    },
    {
      "epoch": 7.93425975414217,
      "grad_norm": 2.0891330242156982,
      "learning_rate": 2.0329415096214022e-05,
      "loss": 1.2904,
      "step": 29690
    },
    {
      "epoch": 7.936932121859968,
      "grad_norm": 2.0506865978240967,
      "learning_rate": 2.0278703581778045e-05,
      "loss": 1.3114,
      "step": 29700
    },
    {
      "epoch": 7.939604489577766,
      "grad_norm": 2.0114119052886963,
      "learning_rate": 2.022804825830904e-05,
      "loss": 1.2361,
      "step": 29710
    },
    {
      "epoch": 7.942276857295564,
      "grad_norm": 2.024193286895752,
      "learning_rate": 2.017744916151112e-05,
      "loss": 1.1984,
      "step": 29720
    },
    {
      "epoch": 7.944949225013362,
      "grad_norm": 2.2009425163269043,
      "learning_rate": 2.012690632704861e-05,
      "loss": 1.2247,
      "step": 29730
    },
    {
      "epoch": 7.94762159273116,
      "grad_norm": 2.092966079711914,
      "learning_rate": 2.0076419790546297e-05,
      "loss": 1.2011,
      "step": 29740
    },
    {
      "epoch": 7.950293960448958,
      "grad_norm": 2.0474660396575928,
      "learning_rate": 2.0025989587589213e-05,
      "loss": 1.3308,
      "step": 29750
    },
    {
      "epoch": 7.952966328166756,
      "grad_norm": 2.188830852508545,
      "learning_rate": 1.9975615753722744e-05,
      "loss": 1.2081,
      "step": 29760
    },
    {
      "epoch": 7.955638695884554,
      "grad_norm": 2.28385066986084,
      "learning_rate": 1.992529832445249e-05,
      "loss": 1.2707,
      "step": 29770
    },
    {
      "epoch": 7.958311063602352,
      "grad_norm": 2.2341206073760986,
      "learning_rate": 1.987503733524437e-05,
      "loss": 1.2414,
      "step": 29780
    },
    {
      "epoch": 7.96098343132015,
      "grad_norm": 2.103367805480957,
      "learning_rate": 1.98248328215244e-05,
      "loss": 1.1736,
      "step": 29790
    },
    {
      "epoch": 7.963655799037948,
      "grad_norm": 2.0253405570983887,
      "learning_rate": 1.977468481867889e-05,
      "loss": 1.27,
      "step": 29800
    },
    {
      "epoch": 7.966328166755746,
      "grad_norm": 2.1806020736694336,
      "learning_rate": 1.97245933620543e-05,
      "loss": 1.2888,
      "step": 29810
    },
    {
      "epoch": 7.969000534473544,
      "grad_norm": 2.115906238555908,
      "learning_rate": 1.9674558486957207e-05,
      "loss": 1.2915,
      "step": 29820
    },
    {
      "epoch": 7.971672902191342,
      "grad_norm": 2.048633575439453,
      "learning_rate": 1.9624580228654366e-05,
      "loss": 1.2526,
      "step": 29830
    },
    {
      "epoch": 7.97434526990914,
      "grad_norm": 2.162379264831543,
      "learning_rate": 1.95746586223725e-05,
      "loss": 1.3142,
      "step": 29840
    },
    {
      "epoch": 7.977017637626938,
      "grad_norm": 1.999262809753418,
      "learning_rate": 1.9524793703298572e-05,
      "loss": 1.2922,
      "step": 29850
    },
    {
      "epoch": 7.979690005344735,
      "grad_norm": 2.1235110759735107,
      "learning_rate": 1.9474985506579446e-05,
      "loss": 1.2304,
      "step": 29860
    },
    {
      "epoch": 7.982362373062533,
      "grad_norm": 2.2172183990478516,
      "learning_rate": 1.9425234067322085e-05,
      "loss": 1.24,
      "step": 29870
    },
    {
      "epoch": 7.985034740780331,
      "grad_norm": 2.0238895416259766,
      "learning_rate": 1.93755394205934e-05,
      "loss": 1.2907,
      "step": 29880
    },
    {
      "epoch": 7.987707108498129,
      "grad_norm": 2.0927395820617676,
      "learning_rate": 1.932590160142036e-05,
      "loss": 1.2551,
      "step": 29890
    },
    {
      "epoch": 7.990379476215927,
      "grad_norm": 2.1852126121520996,
      "learning_rate": 1.92763206447897e-05,
      "loss": 1.2659,
      "step": 29900
    },
    {
      "epoch": 7.993051843933725,
      "grad_norm": 2.173623561859131,
      "learning_rate": 1.922679658564832e-05,
      "loss": 1.3331,
      "step": 29910
    },
    {
      "epoch": 7.995724211651523,
      "grad_norm": 2.218395471572876,
      "learning_rate": 1.9177329458902772e-05,
      "loss": 1.3252,
      "step": 29920
    },
    {
      "epoch": 7.998396579369321,
      "grad_norm": 2.0829696655273438,
      "learning_rate": 1.9127919299419705e-05,
      "loss": 1.206,
      "step": 29930
    },
    {
      "epoch": 8.00106894708712,
      "grad_norm": 1.966801404953003,
      "learning_rate": 1.9078566142025424e-05,
      "loss": 1.2192,
      "step": 29940
    },
    {
      "epoch": 8.003741314804918,
      "grad_norm": 1.886599063873291,
      "learning_rate": 1.9029270021506162e-05,
      "loss": 1.1609,
      "step": 29950
    },
    {
      "epoch": 8.006413682522716,
      "grad_norm": 2.034529685974121,
      "learning_rate": 1.8980030972607933e-05,
      "loss": 1.2364,
      "step": 29960
    },
    {
      "epoch": 8.009086050240514,
      "grad_norm": 1.9105424880981445,
      "learning_rate": 1.8930849030036536e-05,
      "loss": 1.1542,
      "step": 29970
    },
    {
      "epoch": 8.011758417958312,
      "grad_norm": 2.1083168983459473,
      "learning_rate": 1.8881724228457442e-05,
      "loss": 1.0939,
      "step": 29980
    },
    {
      "epoch": 8.01443078567611,
      "grad_norm": 2.141692876815796,
      "learning_rate": 1.8832656602495934e-05,
      "loss": 1.1055,
      "step": 29990
    },
    {
      "epoch": 8.017103153393908,
      "grad_norm": 2.1567461490631104,
      "learning_rate": 1.8783646186736982e-05,
      "loss": 1.1455,
      "step": 30000
    },
    {
      "epoch": 8.019775521111704,
      "grad_norm": 2.061882495880127,
      "learning_rate": 1.8734693015725203e-05,
      "loss": 1.1745,
      "step": 30010
    },
    {
      "epoch": 8.022447888829502,
      "grad_norm": 2.381556987762451,
      "learning_rate": 1.8685797123964886e-05,
      "loss": 1.2167,
      "step": 30020
    },
    {
      "epoch": 8.0251202565473,
      "grad_norm": 2.1993541717529297,
      "learning_rate": 1.8636958545919904e-05,
      "loss": 1.1297,
      "step": 30030
    },
    {
      "epoch": 8.027792624265098,
      "grad_norm": 2.132571220397949,
      "learning_rate": 1.8588177316013833e-05,
      "loss": 1.1577,
      "step": 30040
    },
    {
      "epoch": 8.030464991982896,
      "grad_norm": 2.038928508758545,
      "learning_rate": 1.8539453468629698e-05,
      "loss": 1.1506,
      "step": 30050
    },
    {
      "epoch": 8.033137359700694,
      "grad_norm": 2.096400737762451,
      "learning_rate": 1.8490787038110167e-05,
      "loss": 1.0896,
      "step": 30060
    },
    {
      "epoch": 8.035809727418492,
      "grad_norm": 2.067260980606079,
      "learning_rate": 1.844217805875743e-05,
      "loss": 1.1404,
      "step": 30070
    },
    {
      "epoch": 8.03848209513629,
      "grad_norm": 2.1161482334136963,
      "learning_rate": 1.8393626564833144e-05,
      "loss": 1.1602,
      "step": 30080
    },
    {
      "epoch": 8.041154462854088,
      "grad_norm": 2.1654980182647705,
      "learning_rate": 1.834513259055849e-05,
      "loss": 1.1428,
      "step": 30090
    },
    {
      "epoch": 8.043826830571886,
      "grad_norm": 2.059821605682373,
      "learning_rate": 1.8296696170114092e-05,
      "loss": 1.1078,
      "step": 30100
    },
    {
      "epoch": 8.046499198289684,
      "grad_norm": 2.2347376346588135,
      "learning_rate": 1.8248317337639943e-05,
      "loss": 1.098,
      "step": 30110
    },
    {
      "epoch": 8.049171566007482,
      "grad_norm": 2.23544979095459,
      "learning_rate": 1.819999612723561e-05,
      "loss": 1.1647,
      "step": 30120
    },
    {
      "epoch": 8.05184393372528,
      "grad_norm": 2.2311031818389893,
      "learning_rate": 1.8151732572959857e-05,
      "loss": 1.172,
      "step": 30130
    },
    {
      "epoch": 8.054516301443078,
      "grad_norm": 2.089353561401367,
      "learning_rate": 1.810352670883092e-05,
      "loss": 1.1726,
      "step": 30140
    },
    {
      "epoch": 8.057188669160876,
      "grad_norm": 2.0155904293060303,
      "learning_rate": 1.805537856882634e-05,
      "loss": 1.1595,
      "step": 30150
    },
    {
      "epoch": 8.059861036878674,
      "grad_norm": 2.144171953201294,
      "learning_rate": 1.8007288186882987e-05,
      "loss": 1.2483,
      "step": 30160
    },
    {
      "epoch": 8.062533404596472,
      "grad_norm": 1.969314455986023,
      "learning_rate": 1.795925559689704e-05,
      "loss": 1.13,
      "step": 30170
    },
    {
      "epoch": 8.06520577231427,
      "grad_norm": 2.1316988468170166,
      "learning_rate": 1.7911280832723865e-05,
      "loss": 1.1821,
      "step": 30180
    },
    {
      "epoch": 8.067878140032068,
      "grad_norm": 2.348546266555786,
      "learning_rate": 1.7863363928178122e-05,
      "loss": 1.1506,
      "step": 30190
    },
    {
      "epoch": 8.070550507749866,
      "grad_norm": 2.159191846847534,
      "learning_rate": 1.7815504917033722e-05,
      "loss": 1.1026,
      "step": 30200
    },
    {
      "epoch": 8.073222875467664,
      "grad_norm": 2.0628116130828857,
      "learning_rate": 1.776770383302374e-05,
      "loss": 1.2313,
      "step": 30210
    },
    {
      "epoch": 8.075895243185462,
      "grad_norm": 2.302281379699707,
      "learning_rate": 1.7719960709840342e-05,
      "loss": 1.1445,
      "step": 30220
    },
    {
      "epoch": 8.07856761090326,
      "grad_norm": 2.1654512882232666,
      "learning_rate": 1.7672275581135023e-05,
      "loss": 1.1616,
      "step": 30230
    },
    {
      "epoch": 8.081239978621058,
      "grad_norm": 2.1467254161834717,
      "learning_rate": 1.762464848051819e-05,
      "loss": 1.1584,
      "step": 30240
    },
    {
      "epoch": 8.083912346338856,
      "grad_norm": 2.3476336002349854,
      "learning_rate": 1.7577079441559553e-05,
      "loss": 1.2594,
      "step": 30250
    },
    {
      "epoch": 8.086584714056654,
      "grad_norm": 2.1335368156433105,
      "learning_rate": 1.752956849778772e-05,
      "loss": 1.212,
      "step": 30260
    },
    {
      "epoch": 8.089257081774452,
      "grad_norm": 2.0671567916870117,
      "learning_rate": 1.7482115682690457e-05,
      "loss": 1.2279,
      "step": 30270
    },
    {
      "epoch": 8.09192944949225,
      "grad_norm": 2.1876256465911865,
      "learning_rate": 1.743472102971454e-05,
      "loss": 1.1069,
      "step": 30280
    },
    {
      "epoch": 8.094601817210048,
      "grad_norm": 2.2041354179382324,
      "learning_rate": 1.738738457226574e-05,
      "loss": 1.1003,
      "step": 30290
    },
    {
      "epoch": 8.097274184927846,
      "grad_norm": 2.3087782859802246,
      "learning_rate": 1.734010634370876e-05,
      "loss": 1.2044,
      "step": 30300
    },
    {
      "epoch": 8.099946552645644,
      "grad_norm": 2.107633352279663,
      "learning_rate": 1.72928863773674e-05,
      "loss": 1.1731,
      "step": 30310
    },
    {
      "epoch": 8.102618920363442,
      "grad_norm": 2.0872414112091064,
      "learning_rate": 1.7245724706524234e-05,
      "loss": 1.1967,
      "step": 30320
    },
    {
      "epoch": 8.10529128808124,
      "grad_norm": 2.263470411300659,
      "learning_rate": 1.719862136442083e-05,
      "loss": 1.226,
      "step": 30330
    },
    {
      "epoch": 8.107963655799038,
      "grad_norm": 2.2201919555664062,
      "learning_rate": 1.715157638425765e-05,
      "loss": 1.1702,
      "step": 30340
    },
    {
      "epoch": 8.110636023516836,
      "grad_norm": 2.167698383331299,
      "learning_rate": 1.7104589799193982e-05,
      "loss": 1.2308,
      "step": 30350
    },
    {
      "epoch": 8.113308391234634,
      "grad_norm": 2.2280988693237305,
      "learning_rate": 1.705766164234801e-05,
      "loss": 1.1297,
      "step": 30360
    },
    {
      "epoch": 8.115980758952432,
      "grad_norm": 2.1165075302124023,
      "learning_rate": 1.7010791946796632e-05,
      "loss": 1.154,
      "step": 30370
    },
    {
      "epoch": 8.11865312667023,
      "grad_norm": 2.2603657245635986,
      "learning_rate": 1.6963980745575657e-05,
      "loss": 1.1188,
      "step": 30380
    },
    {
      "epoch": 8.121325494388028,
      "grad_norm": 2.1390349864959717,
      "learning_rate": 1.6917228071679602e-05,
      "loss": 1.1859,
      "step": 30390
    },
    {
      "epoch": 8.123997862105826,
      "grad_norm": 2.039499521255493,
      "learning_rate": 1.6870533958061752e-05,
      "loss": 1.1416,
      "step": 30400
    },
    {
      "epoch": 8.126670229823624,
      "grad_norm": 2.24080228805542,
      "learning_rate": 1.6823898437634088e-05,
      "loss": 1.1811,
      "step": 30410
    },
    {
      "epoch": 8.129342597541422,
      "grad_norm": 2.186419725418091,
      "learning_rate": 1.6777321543267356e-05,
      "loss": 1.1423,
      "step": 30420
    },
    {
      "epoch": 8.13201496525922,
      "grad_norm": 2.129504680633545,
      "learning_rate": 1.6730803307790842e-05,
      "loss": 1.1633,
      "step": 30430
    },
    {
      "epoch": 8.134687332977018,
      "grad_norm": 2.101668119430542,
      "learning_rate": 1.6684343763992703e-05,
      "loss": 1.2498,
      "step": 30440
    },
    {
      "epoch": 8.137359700694816,
      "grad_norm": 2.193866729736328,
      "learning_rate": 1.6637942944619522e-05,
      "loss": 1.1965,
      "step": 30450
    },
    {
      "epoch": 8.140032068412614,
      "grad_norm": 2.162787437438965,
      "learning_rate": 1.6591600882376602e-05,
      "loss": 1.207,
      "step": 30460
    },
    {
      "epoch": 8.142704436130412,
      "grad_norm": 2.0925650596618652,
      "learning_rate": 1.654531760992779e-05,
      "loss": 1.183,
      "step": 30470
    },
    {
      "epoch": 8.14537680384821,
      "grad_norm": 2.38498854637146,
      "learning_rate": 1.6499093159895518e-05,
      "loss": 1.2175,
      "step": 30480
    },
    {
      "epoch": 8.148049171566008,
      "grad_norm": 2.302579641342163,
      "learning_rate": 1.6452927564860743e-05,
      "loss": 1.1527,
      "step": 30490
    },
    {
      "epoch": 8.150721539283806,
      "grad_norm": 2.1436870098114014,
      "learning_rate": 1.640682085736298e-05,
      "loss": 1.2166,
      "step": 30500
    },
    {
      "epoch": 8.153393907001604,
      "grad_norm": 2.1942169666290283,
      "learning_rate": 1.6360773069900148e-05,
      "loss": 1.1716,
      "step": 30510
    },
    {
      "epoch": 8.156066274719402,
      "grad_norm": 2.1793153285980225,
      "learning_rate": 1.6314784234928702e-05,
      "loss": 1.1684,
      "step": 30520
    },
    {
      "epoch": 8.1587386424372,
      "grad_norm": 2.14441180229187,
      "learning_rate": 1.626885438486355e-05,
      "loss": 1.1195,
      "step": 30530
    },
    {
      "epoch": 8.161411010154998,
      "grad_norm": 2.1185004711151123,
      "learning_rate": 1.6222983552078007e-05,
      "loss": 1.2215,
      "step": 30540
    },
    {
      "epoch": 8.164083377872796,
      "grad_norm": 2.123196840286255,
      "learning_rate": 1.61771717689038e-05,
      "loss": 1.1832,
      "step": 30550
    },
    {
      "epoch": 8.166755745590594,
      "grad_norm": 2.186194896697998,
      "learning_rate": 1.613141906763096e-05,
      "loss": 1.2083,
      "step": 30560
    },
    {
      "epoch": 8.169428113308392,
      "grad_norm": 2.193206787109375,
      "learning_rate": 1.608572548050803e-05,
      "loss": 1.0979,
      "step": 30570
    },
    {
      "epoch": 8.17210048102619,
      "grad_norm": 2.2169249057769775,
      "learning_rate": 1.6040091039741744e-05,
      "loss": 1.12,
      "step": 30580
    },
    {
      "epoch": 8.174772848743988,
      "grad_norm": 2.133208990097046,
      "learning_rate": 1.5994515777497223e-05,
      "loss": 1.1334,
      "step": 30590
    },
    {
      "epoch": 8.177445216461786,
      "grad_norm": 2.180858850479126,
      "learning_rate": 1.594899972589783e-05,
      "loss": 1.1644,
      "step": 30600
    },
    {
      "epoch": 8.180117584179584,
      "grad_norm": 1.9578543901443481,
      "learning_rate": 1.590354291702526e-05,
      "loss": 1.1888,
      "step": 30610
    },
    {
      "epoch": 8.182789951897382,
      "grad_norm": 2.098051071166992,
      "learning_rate": 1.5858145382919333e-05,
      "loss": 1.1086,
      "step": 30620
    },
    {
      "epoch": 8.18546231961518,
      "grad_norm": 2.2391555309295654,
      "learning_rate": 1.581280715557827e-05,
      "loss": 1.1452,
      "step": 30630
    },
    {
      "epoch": 8.188134687332978,
      "grad_norm": 2.376694917678833,
      "learning_rate": 1.5767528266958277e-05,
      "loss": 1.1841,
      "step": 30640
    },
    {
      "epoch": 8.190807055050776,
      "grad_norm": 2.172715902328491,
      "learning_rate": 1.5722308748973947e-05,
      "loss": 1.1453,
      "step": 30650
    },
    {
      "epoch": 8.193479422768572,
      "grad_norm": 2.26023268699646,
      "learning_rate": 1.5677148633497853e-05,
      "loss": 1.2372,
      "step": 30660
    },
    {
      "epoch": 8.19615179048637,
      "grad_norm": 2.113327980041504,
      "learning_rate": 1.5632047952360783e-05,
      "loss": 1.2091,
      "step": 30670
    },
    {
      "epoch": 8.198824158204168,
      "grad_norm": 2.140690326690674,
      "learning_rate": 1.5587006737351628e-05,
      "loss": 1.1531,
      "step": 30680
    },
    {
      "epoch": 8.201496525921966,
      "grad_norm": 2.1371402740478516,
      "learning_rate": 1.5542025020217375e-05,
      "loss": 1.2126,
      "step": 30690
    },
    {
      "epoch": 8.204168893639764,
      "grad_norm": 2.1615569591522217,
      "learning_rate": 1.5497102832663023e-05,
      "loss": 1.2039,
      "step": 30700
    },
    {
      "epoch": 8.206841261357562,
      "grad_norm": 2.0711669921875,
      "learning_rate": 1.5452240206351643e-05,
      "loss": 1.2236,
      "step": 30710
    },
    {
      "epoch": 8.20951362907536,
      "grad_norm": 2.1842689514160156,
      "learning_rate": 1.5407437172904336e-05,
      "loss": 1.1873,
      "step": 30720
    },
    {
      "epoch": 8.212185996793158,
      "grad_norm": 2.288241147994995,
      "learning_rate": 1.5362693763900193e-05,
      "loss": 1.204,
      "step": 30730
    },
    {
      "epoch": 8.214858364510956,
      "grad_norm": 2.0649006366729736,
      "learning_rate": 1.5318010010876305e-05,
      "loss": 1.1229,
      "step": 30740
    },
    {
      "epoch": 8.217530732228754,
      "grad_norm": 2.0768380165100098,
      "learning_rate": 1.52733859453276e-05,
      "loss": 1.1974,
      "step": 30750
    },
    {
      "epoch": 8.220203099946552,
      "grad_norm": 1.9523011445999146,
      "learning_rate": 1.5228821598707122e-05,
      "loss": 1.1893,
      "step": 30760
    },
    {
      "epoch": 8.22287546766435,
      "grad_norm": 2.1099400520324707,
      "learning_rate": 1.5184317002425653e-05,
      "loss": 1.1913,
      "step": 30770
    },
    {
      "epoch": 8.225547835382148,
      "grad_norm": 2.0837039947509766,
      "learning_rate": 1.5139872187851944e-05,
      "loss": 1.1638,
      "step": 30780
    },
    {
      "epoch": 8.228220203099946,
      "grad_norm": 2.160473585128784,
      "learning_rate": 1.5095487186312607e-05,
      "loss": 1.1237,
      "step": 30790
    },
    {
      "epoch": 8.230892570817744,
      "grad_norm": 2.131002426147461,
      "learning_rate": 1.505116202909207e-05,
      "loss": 1.2002,
      "step": 30800
    },
    {
      "epoch": 8.233564938535542,
      "grad_norm": 2.072420835494995,
      "learning_rate": 1.5006896747432609e-05,
      "loss": 1.1823,
      "step": 30810
    },
    {
      "epoch": 8.23623730625334,
      "grad_norm": 2.480342149734497,
      "learning_rate": 1.4962691372534288e-05,
      "loss": 1.1554,
      "step": 30820
    },
    {
      "epoch": 8.238909673971138,
      "grad_norm": 2.637852668762207,
      "learning_rate": 1.4918545935554872e-05,
      "loss": 1.171,
      "step": 30830
    },
    {
      "epoch": 8.241582041688936,
      "grad_norm": 1.9761701822280884,
      "learning_rate": 1.4874460467610041e-05,
      "loss": 1.1365,
      "step": 30840
    },
    {
      "epoch": 8.244254409406734,
      "grad_norm": 2.2092714309692383,
      "learning_rate": 1.4830434999773046e-05,
      "loss": 1.2147,
      "step": 30850
    },
    {
      "epoch": 8.246926777124532,
      "grad_norm": 2.3081250190734863,
      "learning_rate": 1.4786469563074933e-05,
      "loss": 1.1673,
      "step": 30860
    },
    {
      "epoch": 8.24959914484233,
      "grad_norm": 2.143738031387329,
      "learning_rate": 1.4742564188504426e-05,
      "loss": 1.1376,
      "step": 30870
    },
    {
      "epoch": 8.252271512560128,
      "grad_norm": 2.171787977218628,
      "learning_rate": 1.469871890700788e-05,
      "loss": 1.1716,
      "step": 30880
    },
    {
      "epoch": 8.254943880277926,
      "grad_norm": 2.045050859451294,
      "learning_rate": 1.4654933749489363e-05,
      "loss": 1.1857,
      "step": 30890
    },
    {
      "epoch": 8.257616247995724,
      "grad_norm": 2.2319345474243164,
      "learning_rate": 1.4611208746810456e-05,
      "loss": 1.1791,
      "step": 30900
    },
    {
      "epoch": 8.260288615713522,
      "grad_norm": 2.127666711807251,
      "learning_rate": 1.456754392979044e-05,
      "loss": 1.1681,
      "step": 30910
    },
    {
      "epoch": 8.26296098343132,
      "grad_norm": 2.217803955078125,
      "learning_rate": 1.4523939329206138e-05,
      "loss": 1.2134,
      "step": 30920
    },
    {
      "epoch": 8.265633351149118,
      "grad_norm": 2.1986429691314697,
      "learning_rate": 1.4480394975791955e-05,
      "loss": 1.1816,
      "step": 30930
    },
    {
      "epoch": 8.268305718866916,
      "grad_norm": 2.1153335571289062,
      "learning_rate": 1.4436910900239731e-05,
      "loss": 1.1473,
      "step": 30940
    },
    {
      "epoch": 8.270978086584714,
      "grad_norm": 2.03336763381958,
      "learning_rate": 1.439348713319898e-05,
      "loss": 1.1628,
      "step": 30950
    },
    {
      "epoch": 8.273650454302512,
      "grad_norm": 2.1557438373565674,
      "learning_rate": 1.4350123705276552e-05,
      "loss": 1.1765,
      "step": 30960
    },
    {
      "epoch": 8.27632282202031,
      "grad_norm": 2.291888952255249,
      "learning_rate": 1.4306820647036923e-05,
      "loss": 1.1561,
      "step": 30970
    },
    {
      "epoch": 8.278995189738108,
      "grad_norm": 2.04992413520813,
      "learning_rate": 1.4263577989001863e-05,
      "loss": 1.1502,
      "step": 30980
    },
    {
      "epoch": 8.281667557455906,
      "grad_norm": 2.212059259414673,
      "learning_rate": 1.422039576165065e-05,
      "loss": 1.2022,
      "step": 30990
    },
    {
      "epoch": 8.284339925173704,
      "grad_norm": 2.0827481746673584,
      "learning_rate": 1.4177273995419982e-05,
      "loss": 1.2266,
      "step": 31000
    },
    {
      "epoch": 8.287012292891502,
      "grad_norm": 2.099534273147583,
      "learning_rate": 1.4134212720703931e-05,
      "loss": 1.2009,
      "step": 31010
    },
    {
      "epoch": 8.2896846606093,
      "grad_norm": 2.334784984588623,
      "learning_rate": 1.4091211967853845e-05,
      "loss": 1.1279,
      "step": 31020
    },
    {
      "epoch": 8.292357028327098,
      "grad_norm": 2.196622848510742,
      "learning_rate": 1.4048271767178579e-05,
      "loss": 1.1313,
      "step": 31030
    },
    {
      "epoch": 8.295029396044896,
      "grad_norm": 2.226259231567383,
      "learning_rate": 1.4005392148944152e-05,
      "loss": 1.1444,
      "step": 31040
    },
    {
      "epoch": 8.297701763762694,
      "grad_norm": 1.986608624458313,
      "learning_rate": 1.3962573143373969e-05,
      "loss": 1.1371,
      "step": 31050
    },
    {
      "epoch": 8.300374131480492,
      "grad_norm": 2.1444942951202393,
      "learning_rate": 1.3919814780648688e-05,
      "loss": 1.2145,
      "step": 31060
    },
    {
      "epoch": 8.30304649919829,
      "grad_norm": 2.1759116649627686,
      "learning_rate": 1.3877117090906222e-05,
      "loss": 1.222,
      "step": 31070
    },
    {
      "epoch": 8.305718866916088,
      "grad_norm": 2.197798252105713,
      "learning_rate": 1.3834480104241754e-05,
      "loss": 1.1764,
      "step": 31080
    },
    {
      "epoch": 8.308391234633886,
      "grad_norm": 2.1161584854125977,
      "learning_rate": 1.3791903850707589e-05,
      "loss": 1.1809,
      "step": 31090
    },
    {
      "epoch": 8.311063602351684,
      "grad_norm": 2.106163501739502,
      "learning_rate": 1.374938836031332e-05,
      "loss": 1.1629,
      "step": 31100
    },
    {
      "epoch": 8.313735970069482,
      "grad_norm": 2.179640293121338,
      "learning_rate": 1.3706933663025678e-05,
      "loss": 1.187,
      "step": 31110
    },
    {
      "epoch": 8.31640833778728,
      "grad_norm": 2.1718761920928955,
      "learning_rate": 1.3664539788768539e-05,
      "loss": 1.1902,
      "step": 31120
    },
    {
      "epoch": 8.319080705505078,
      "grad_norm": 2.2546775341033936,
      "learning_rate": 1.3622206767422907e-05,
      "loss": 1.171,
      "step": 31130
    },
    {
      "epoch": 8.321753073222876,
      "grad_norm": 2.2284255027770996,
      "learning_rate": 1.357993462882693e-05,
      "loss": 1.1958,
      "step": 31140
    },
    {
      "epoch": 8.324425440940674,
      "grad_norm": 2.2016327381134033,
      "learning_rate": 1.3537723402775737e-05,
      "loss": 1.1807,
      "step": 31150
    },
    {
      "epoch": 8.327097808658472,
      "grad_norm": 2.0302164554595947,
      "learning_rate": 1.3495573119021698e-05,
      "loss": 1.1936,
      "step": 31160
    },
    {
      "epoch": 8.32977017637627,
      "grad_norm": 2.3190088272094727,
      "learning_rate": 1.3453483807274048e-05,
      "loss": 1.1601,
      "step": 31170
    },
    {
      "epoch": 8.332442544094068,
      "grad_norm": 2.2674684524536133,
      "learning_rate": 1.3411455497199165e-05,
      "loss": 1.2164,
      "step": 31180
    },
    {
      "epoch": 8.335114911811866,
      "grad_norm": 2.414410352706909,
      "learning_rate": 1.336948821842039e-05,
      "loss": 1.1185,
      "step": 31190
    },
    {
      "epoch": 8.337787279529664,
      "grad_norm": 1.936843991279602,
      "learning_rate": 1.3327582000518035e-05,
      "loss": 1.1742,
      "step": 31200
    },
    {
      "epoch": 8.340459647247462,
      "grad_norm": 2.2939600944519043,
      "learning_rate": 1.3285736873029409e-05,
      "loss": 1.1519,
      "step": 31210
    },
    {
      "epoch": 8.34313201496526,
      "grad_norm": 2.153366804122925,
      "learning_rate": 1.324395286544876e-05,
      "loss": 1.2037,
      "step": 31220
    },
    {
      "epoch": 8.345804382683058,
      "grad_norm": 2.1847779750823975,
      "learning_rate": 1.3202230007227179e-05,
      "loss": 1.189,
      "step": 31230
    },
    {
      "epoch": 8.348476750400856,
      "grad_norm": 2.138488531112671,
      "learning_rate": 1.3160568327772738e-05,
      "loss": 1.2171,
      "step": 31240
    },
    {
      "epoch": 8.351149118118654,
      "grad_norm": 2.1546854972839355,
      "learning_rate": 1.3118967856450382e-05,
      "loss": 1.2126,
      "step": 31250
    },
    {
      "epoch": 8.353821485836452,
      "grad_norm": 2.1969501972198486,
      "learning_rate": 1.3077428622581877e-05,
      "loss": 1.1706,
      "step": 31260
    },
    {
      "epoch": 8.35649385355425,
      "grad_norm": 2.181748628616333,
      "learning_rate": 1.3035950655445872e-05,
      "loss": 1.1448,
      "step": 31270
    },
    {
      "epoch": 8.359166221272048,
      "grad_norm": 2.1777846813201904,
      "learning_rate": 1.2994533984277758e-05,
      "loss": 1.1601,
      "step": 31280
    },
    {
      "epoch": 8.361838588989844,
      "grad_norm": 2.219191789627075,
      "learning_rate": 1.2953178638269825e-05,
      "loss": 1.3012,
      "step": 31290
    },
    {
      "epoch": 8.364510956707644,
      "grad_norm": 2.1889684200286865,
      "learning_rate": 1.2911884646571037e-05,
      "loss": 1.1901,
      "step": 31300
    },
    {
      "epoch": 8.36718332442544,
      "grad_norm": 2.0811409950256348,
      "learning_rate": 1.287065203828719e-05,
      "loss": 1.1899,
      "step": 31310
    },
    {
      "epoch": 8.369855692143238,
      "grad_norm": 2.2898550033569336,
      "learning_rate": 1.2829480842480769e-05,
      "loss": 1.1818,
      "step": 31320
    },
    {
      "epoch": 8.372528059861036,
      "grad_norm": 2.0709502696990967,
      "learning_rate": 1.278837108817098e-05,
      "loss": 1.2489,
      "step": 31330
    },
    {
      "epoch": 8.375200427578834,
      "grad_norm": 2.123767614364624,
      "learning_rate": 1.274732280433375e-05,
      "loss": 1.1974,
      "step": 31340
    },
    {
      "epoch": 8.377872795296632,
      "grad_norm": 2.071045160293579,
      "learning_rate": 1.2706336019901665e-05,
      "loss": 1.1965,
      "step": 31350
    },
    {
      "epoch": 8.38054516301443,
      "grad_norm": 2.184329032897949,
      "learning_rate": 1.2665410763763896e-05,
      "loss": 1.2172,
      "step": 31360
    },
    {
      "epoch": 8.383217530732228,
      "grad_norm": 2.3401215076446533,
      "learning_rate": 1.2624547064766402e-05,
      "loss": 1.1899,
      "step": 31370
    },
    {
      "epoch": 8.385889898450026,
      "grad_norm": 2.155318260192871,
      "learning_rate": 1.2583744951711584e-05,
      "loss": 1.1792,
      "step": 31380
    },
    {
      "epoch": 8.388562266167824,
      "grad_norm": 2.20011305809021,
      "learning_rate": 1.2543004453358531e-05,
      "loss": 1.2066,
      "step": 31390
    },
    {
      "epoch": 8.391234633885622,
      "grad_norm": 2.0160932540893555,
      "learning_rate": 1.2502325598422893e-05,
      "loss": 1.2222,
      "step": 31400
    },
    {
      "epoch": 8.39390700160342,
      "grad_norm": 2.0931403636932373,
      "learning_rate": 1.2461708415576856e-05,
      "loss": 1.2384,
      "step": 31410
    },
    {
      "epoch": 8.396579369321218,
      "grad_norm": 2.1289875507354736,
      "learning_rate": 1.2421152933449165e-05,
      "loss": 1.1575,
      "step": 31420
    },
    {
      "epoch": 8.399251737039016,
      "grad_norm": 2.1781466007232666,
      "learning_rate": 1.2380659180624998e-05,
      "loss": 1.191,
      "step": 31430
    },
    {
      "epoch": 8.401924104756814,
      "grad_norm": 2.1087586879730225,
      "learning_rate": 1.2340227185646113e-05,
      "loss": 1.1978,
      "step": 31440
    },
    {
      "epoch": 8.404596472474612,
      "grad_norm": 2.1136343479156494,
      "learning_rate": 1.2299856977010694e-05,
      "loss": 1.1861,
      "step": 31450
    },
    {
      "epoch": 8.40726884019241,
      "grad_norm": 2.284985303878784,
      "learning_rate": 1.2259548583173409e-05,
      "loss": 1.1511,
      "step": 31460
    },
    {
      "epoch": 8.409941207910208,
      "grad_norm": 2.1557559967041016,
      "learning_rate": 1.2219302032545266e-05,
      "loss": 1.1971,
      "step": 31470
    },
    {
      "epoch": 8.412613575628006,
      "grad_norm": 2.25907826423645,
      "learning_rate": 1.217911735349384e-05,
      "loss": 1.2251,
      "step": 31480
    },
    {
      "epoch": 8.415285943345804,
      "grad_norm": 2.07696533203125,
      "learning_rate": 1.213899457434291e-05,
      "loss": 1.1539,
      "step": 31490
    },
    {
      "epoch": 8.417958311063602,
      "grad_norm": 2.1086347103118896,
      "learning_rate": 1.2098933723372819e-05,
      "loss": 1.1878,
      "step": 31500
    },
    {
      "epoch": 8.4206306787814,
      "grad_norm": 2.1808483600616455,
      "learning_rate": 1.2058934828820079e-05,
      "loss": 1.154,
      "step": 31510
    },
    {
      "epoch": 8.423303046499198,
      "grad_norm": 2.208563804626465,
      "learning_rate": 1.2018997918877662e-05,
      "loss": 1.1957,
      "step": 31520
    },
    {
      "epoch": 8.425975414216996,
      "grad_norm": 2.0855329036712646,
      "learning_rate": 1.1979123021694782e-05,
      "loss": 1.1778,
      "step": 31530
    },
    {
      "epoch": 8.428647781934794,
      "grad_norm": 2.3629298210144043,
      "learning_rate": 1.1939310165376993e-05,
      "loss": 1.1381,
      "step": 31540
    },
    {
      "epoch": 8.431320149652592,
      "grad_norm": 2.230238914489746,
      "learning_rate": 1.1899559377986035e-05,
      "loss": 1.163,
      "step": 31550
    },
    {
      "epoch": 8.43399251737039,
      "grad_norm": 2.171351671218872,
      "learning_rate": 1.1859870687540043e-05,
      "loss": 1.2073,
      "step": 31560
    },
    {
      "epoch": 8.436664885088188,
      "grad_norm": 2.2594099044799805,
      "learning_rate": 1.1820244122013213e-05,
      "loss": 1.1977,
      "step": 31570
    },
    {
      "epoch": 8.439337252805986,
      "grad_norm": 2.267430067062378,
      "learning_rate": 1.1780679709336085e-05,
      "loss": 1.2261,
      "step": 31580
    },
    {
      "epoch": 8.442009620523784,
      "grad_norm": 2.2696616649627686,
      "learning_rate": 1.1741177477395337e-05,
      "loss": 1.1302,
      "step": 31590
    },
    {
      "epoch": 8.444681988241582,
      "grad_norm": 2.0815224647521973,
      "learning_rate": 1.1701737454033812e-05,
      "loss": 1.1985,
      "step": 31600
    },
    {
      "epoch": 8.44735435595938,
      "grad_norm": 2.148184299468994,
      "learning_rate": 1.166235966705056e-05,
      "loss": 1.1627,
      "step": 31610
    },
    {
      "epoch": 8.450026723677178,
      "grad_norm": 2.28935170173645,
      "learning_rate": 1.1623044144200657e-05,
      "loss": 1.1723,
      "step": 31620
    },
    {
      "epoch": 8.452699091394976,
      "grad_norm": 2.2608673572540283,
      "learning_rate": 1.1583790913195402e-05,
      "loss": 1.2414,
      "step": 31630
    },
    {
      "epoch": 8.455371459112774,
      "grad_norm": 2.1188554763793945,
      "learning_rate": 1.1544600001702133e-05,
      "loss": 1.2185,
      "step": 31640
    },
    {
      "epoch": 8.458043826830572,
      "grad_norm": 2.191650629043579,
      "learning_rate": 1.1505471437344272e-05,
      "loss": 1.1641,
      "step": 31650
    },
    {
      "epoch": 8.46071619454837,
      "grad_norm": 2.0754432678222656,
      "learning_rate": 1.1466405247701307e-05,
      "loss": 1.2299,
      "step": 31660
    },
    {
      "epoch": 8.463388562266168,
      "grad_norm": 2.1514980792999268,
      "learning_rate": 1.1427401460308774e-05,
      "loss": 1.2263,
      "step": 31670
    },
    {
      "epoch": 8.466060929983966,
      "grad_norm": 2.1117684841156006,
      "learning_rate": 1.1388460102658138e-05,
      "loss": 1.1282,
      "step": 31680
    },
    {
      "epoch": 8.468733297701764,
      "grad_norm": 2.0499234199523926,
      "learning_rate": 1.1349581202197012e-05,
      "loss": 1.1178,
      "step": 31690
    },
    {
      "epoch": 8.471405665419562,
      "grad_norm": 2.1124424934387207,
      "learning_rate": 1.1310764786328842e-05,
      "loss": 1.1398,
      "step": 31700
    },
    {
      "epoch": 8.47407803313736,
      "grad_norm": 2.0329463481903076,
      "learning_rate": 1.1272010882413109e-05,
      "loss": 1.1365,
      "step": 31710
    },
    {
      "epoch": 8.476750400855158,
      "grad_norm": 2.228235960006714,
      "learning_rate": 1.1233319517765218e-05,
      "loss": 1.2108,
      "step": 31720
    },
    {
      "epoch": 8.479422768572956,
      "grad_norm": 2.252894878387451,
      "learning_rate": 1.1194690719656486e-05,
      "loss": 1.1789,
      "step": 31730
    },
    {
      "epoch": 8.482095136290754,
      "grad_norm": 1.9969401359558105,
      "learning_rate": 1.1156124515314147e-05,
      "loss": 1.1997,
      "step": 31740
    },
    {
      "epoch": 8.484767504008552,
      "grad_norm": 2.015465497970581,
      "learning_rate": 1.1117620931921313e-05,
      "loss": 1.1992,
      "step": 31750
    },
    {
      "epoch": 8.48743987172635,
      "grad_norm": 2.102015733718872,
      "learning_rate": 1.1079179996616906e-05,
      "loss": 1.2278,
      "step": 31760
    },
    {
      "epoch": 8.490112239444148,
      "grad_norm": 2.2371857166290283,
      "learning_rate": 1.1040801736495799e-05,
      "loss": 1.164,
      "step": 31770
    },
    {
      "epoch": 8.492784607161946,
      "grad_norm": 2.1743972301483154,
      "learning_rate": 1.1002486178608573e-05,
      "loss": 1.1726,
      "step": 31780
    },
    {
      "epoch": 8.495456974879744,
      "grad_norm": 2.2615981101989746,
      "learning_rate": 1.0964233349961684e-05,
      "loss": 1.1675,
      "step": 31790
    },
    {
      "epoch": 8.498129342597542,
      "grad_norm": 2.159367561340332,
      "learning_rate": 1.0926043277517373e-05,
      "loss": 1.1977,
      "step": 31800
    },
    {
      "epoch": 8.50080171031534,
      "grad_norm": 2.0810561180114746,
      "learning_rate": 1.0887915988193575e-05,
      "loss": 1.1934,
      "step": 31810
    },
    {
      "epoch": 8.503474078033138,
      "grad_norm": 2.270254611968994,
      "learning_rate": 1.0849851508864117e-05,
      "loss": 1.1449,
      "step": 31820
    },
    {
      "epoch": 8.506146445750936,
      "grad_norm": 2.281573534011841,
      "learning_rate": 1.0811849866358393e-05,
      "loss": 1.1947,
      "step": 31830
    },
    {
      "epoch": 8.508818813468734,
      "grad_norm": 2.147365093231201,
      "learning_rate": 1.0773911087461597e-05,
      "loss": 1.1604,
      "step": 31840
    },
    {
      "epoch": 8.511491181186532,
      "grad_norm": 2.095155954360962,
      "learning_rate": 1.0736035198914606e-05,
      "loss": 1.1658,
      "step": 31850
    },
    {
      "epoch": 8.51416354890433,
      "grad_norm": 2.161773920059204,
      "learning_rate": 1.069822222741398e-05,
      "loss": 1.123,
      "step": 31860
    },
    {
      "epoch": 8.516835916622128,
      "grad_norm": 2.12282133102417,
      "learning_rate": 1.066047219961186e-05,
      "loss": 1.2099,
      "step": 31870
    },
    {
      "epoch": 8.519508284339926,
      "grad_norm": 2.1173250675201416,
      "learning_rate": 1.0622785142116131e-05,
      "loss": 1.1547,
      "step": 31880
    },
    {
      "epoch": 8.522180652057724,
      "grad_norm": 2.0824410915374756,
      "learning_rate": 1.058516108149018e-05,
      "loss": 1.1521,
      "step": 31890
    },
    {
      "epoch": 8.524853019775522,
      "grad_norm": 2.1164462566375732,
      "learning_rate": 1.0547600044253115e-05,
      "loss": 1.2265,
      "step": 31900
    },
    {
      "epoch": 8.52752538749332,
      "grad_norm": 2.152815818786621,
      "learning_rate": 1.05101020568795e-05,
      "loss": 1.2396,
      "step": 31910
    },
    {
      "epoch": 8.530197755211116,
      "grad_norm": 2.2693629264831543,
      "learning_rate": 1.0472667145799543e-05,
      "loss": 1.1862,
      "step": 31920
    },
    {
      "epoch": 8.532870122928916,
      "grad_norm": 2.1222035884857178,
      "learning_rate": 1.0435295337398943e-05,
      "loss": 1.1686,
      "step": 31930
    },
    {
      "epoch": 8.535542490646712,
      "grad_norm": 2.096386432647705,
      "learning_rate": 1.0397986658018988e-05,
      "loss": 1.2282,
      "step": 31940
    },
    {
      "epoch": 8.538214858364512,
      "grad_norm": 2.029423475265503,
      "learning_rate": 1.0360741133956352e-05,
      "loss": 1.1486,
      "step": 31950
    },
    {
      "epoch": 8.540887226082308,
      "grad_norm": 2.0768489837646484,
      "learning_rate": 1.032355879146334e-05,
      "loss": 1.1587,
      "step": 31960
    },
    {
      "epoch": 8.543559593800106,
      "grad_norm": 2.2596304416656494,
      "learning_rate": 1.0286439656747614e-05,
      "loss": 1.1796,
      "step": 31970
    },
    {
      "epoch": 8.546231961517904,
      "grad_norm": 2.1683359146118164,
      "learning_rate": 1.0249383755972319e-05,
      "loss": 1.1875,
      "step": 31980
    },
    {
      "epoch": 8.548904329235702,
      "grad_norm": 1.9842551946640015,
      "learning_rate": 1.0212391115256049e-05,
      "loss": 1.1538,
      "step": 31990
    },
    {
      "epoch": 8.5515766969535,
      "grad_norm": 2.1803505420684814,
      "learning_rate": 1.017546176067279e-05,
      "loss": 1.2272,
      "step": 32000
    },
    {
      "epoch": 8.554249064671298,
      "grad_norm": 2.23764705657959,
      "learning_rate": 1.0138595718251931e-05,
      "loss": 1.1616,
      "step": 32010
    },
    {
      "epoch": 8.556921432389096,
      "grad_norm": 2.2283926010131836,
      "learning_rate": 1.0101793013978212e-05,
      "loss": 1.2217,
      "step": 32020
    },
    {
      "epoch": 8.559593800106894,
      "grad_norm": 2.2320849895477295,
      "learning_rate": 1.0065053673791748e-05,
      "loss": 1.1324,
      "step": 32030
    },
    {
      "epoch": 8.562266167824692,
      "grad_norm": 2.1736326217651367,
      "learning_rate": 1.0028377723588001e-05,
      "loss": 1.1829,
      "step": 32040
    },
    {
      "epoch": 8.56493853554249,
      "grad_norm": 2.1591625213623047,
      "learning_rate": 9.991765189217739e-06,
      "loss": 1.1342,
      "step": 32050
    },
    {
      "epoch": 8.567610903260288,
      "grad_norm": 2.165165662765503,
      "learning_rate": 9.955216096487052e-06,
      "loss": 1.248,
      "step": 32060
    },
    {
      "epoch": 8.570283270978086,
      "grad_norm": 2.13219952583313,
      "learning_rate": 9.918730471157289e-06,
      "loss": 1.2312,
      "step": 32070
    },
    {
      "epoch": 8.572955638695884,
      "grad_norm": 2.083112955093384,
      "learning_rate": 9.882308338945045e-06,
      "loss": 1.1996,
      "step": 32080
    },
    {
      "epoch": 8.575628006413682,
      "grad_norm": 2.139697790145874,
      "learning_rate": 9.845949725522252e-06,
      "loss": 1.2121,
      "step": 32090
    },
    {
      "epoch": 8.57830037413148,
      "grad_norm": 2.1033358573913574,
      "learning_rate": 9.809654656515955e-06,
      "loss": 1.1711,
      "step": 32100
    },
    {
      "epoch": 8.580972741849278,
      "grad_norm": 2.103965997695923,
      "learning_rate": 9.773423157508489e-06,
      "loss": 1.2165,
      "step": 32110
    },
    {
      "epoch": 8.583645109567076,
      "grad_norm": 2.147108554840088,
      "learning_rate": 9.737255254037347e-06,
      "loss": 1.2132,
      "step": 32120
    },
    {
      "epoch": 8.586317477284874,
      "grad_norm": 2.2541441917419434,
      "learning_rate": 9.701150971595218e-06,
      "loss": 1.231,
      "step": 32130
    },
    {
      "epoch": 8.588989845002672,
      "grad_norm": 2.222564697265625,
      "learning_rate": 9.665110335629935e-06,
      "loss": 1.1506,
      "step": 32140
    },
    {
      "epoch": 8.59166221272047,
      "grad_norm": 2.3519718647003174,
      "learning_rate": 9.629133371544496e-06,
      "loss": 1.2058,
      "step": 32150
    },
    {
      "epoch": 8.594334580438268,
      "grad_norm": 2.1793272495269775,
      "learning_rate": 9.593220104696965e-06,
      "loss": 1.1983,
      "step": 32160
    },
    {
      "epoch": 8.597006948156066,
      "grad_norm": 2.234360933303833,
      "learning_rate": 9.557370560400548e-06,
      "loss": 1.1081,
      "step": 32170
    },
    {
      "epoch": 8.599679315873864,
      "grad_norm": 2.244149684906006,
      "learning_rate": 9.521584763923552e-06,
      "loss": 1.1293,
      "step": 32180
    },
    {
      "epoch": 8.602351683591662,
      "grad_norm": 2.090513229370117,
      "learning_rate": 9.485862740489304e-06,
      "loss": 1.1629,
      "step": 32190
    },
    {
      "epoch": 8.60502405130946,
      "grad_norm": 2.1351866722106934,
      "learning_rate": 9.450204515276251e-06,
      "loss": 1.1316,
      "step": 32200
    },
    {
      "epoch": 8.607696419027258,
      "grad_norm": 2.268317699432373,
      "learning_rate": 9.414610113417766e-06,
      "loss": 1.1728,
      "step": 32210
    },
    {
      "epoch": 8.610368786745056,
      "grad_norm": 2.255051851272583,
      "learning_rate": 9.37907956000237e-06,
      "loss": 1.2554,
      "step": 32220
    },
    {
      "epoch": 8.613041154462854,
      "grad_norm": 2.1467652320861816,
      "learning_rate": 9.34361288007346e-06,
      "loss": 1.1468,
      "step": 32230
    },
    {
      "epoch": 8.615713522180652,
      "grad_norm": 2.307605266571045,
      "learning_rate": 9.308210098629488e-06,
      "loss": 1.1892,
      "step": 32240
    },
    {
      "epoch": 8.61838588989845,
      "grad_norm": 2.0895166397094727,
      "learning_rate": 9.272871240623827e-06,
      "loss": 1.1816,
      "step": 32250
    },
    {
      "epoch": 8.621058257616248,
      "grad_norm": 2.0934860706329346,
      "learning_rate": 9.237596330964838e-06,
      "loss": 1.1487,
      "step": 32260
    },
    {
      "epoch": 8.623730625334046,
      "grad_norm": 2.2088074684143066,
      "learning_rate": 9.202385394515733e-06,
      "loss": 1.2107,
      "step": 32270
    },
    {
      "epoch": 8.626402993051844,
      "grad_norm": 2.1643431186676025,
      "learning_rate": 9.16723845609474e-06,
      "loss": 1.1752,
      "step": 32280
    },
    {
      "epoch": 8.629075360769642,
      "grad_norm": 2.221729040145874,
      "learning_rate": 9.132155540474851e-06,
      "loss": 1.169,
      "step": 32290
    },
    {
      "epoch": 8.63174772848744,
      "grad_norm": 2.102905035018921,
      "learning_rate": 9.097136672384078e-06,
      "loss": 1.1952,
      "step": 32300
    },
    {
      "epoch": 8.634420096205238,
      "grad_norm": 2.2897191047668457,
      "learning_rate": 9.06218187650515e-06,
      "loss": 1.1807,
      "step": 32310
    },
    {
      "epoch": 8.637092463923036,
      "grad_norm": 2.1335794925689697,
      "learning_rate": 9.027291177475716e-06,
      "loss": 1.2587,
      "step": 32320
    },
    {
      "epoch": 8.639764831640834,
      "grad_norm": 2.032491683959961,
      "learning_rate": 8.992464599888229e-06,
      "loss": 1.171,
      "step": 32330
    },
    {
      "epoch": 8.642437199358632,
      "grad_norm": 2.24336314201355,
      "learning_rate": 8.95770216828996e-06,
      "loss": 1.2135,
      "step": 32340
    },
    {
      "epoch": 8.64510956707643,
      "grad_norm": 2.183116912841797,
      "learning_rate": 8.923003907182926e-06,
      "loss": 1.147,
      "step": 32350
    },
    {
      "epoch": 8.647781934794228,
      "grad_norm": 2.1955769062042236,
      "learning_rate": 8.888369841023946e-06,
      "loss": 1.1265,
      "step": 32360
    },
    {
      "epoch": 8.650454302512026,
      "grad_norm": 2.3789520263671875,
      "learning_rate": 8.853799994224599e-06,
      "loss": 1.1907,
      "step": 32370
    },
    {
      "epoch": 8.653126670229824,
      "grad_norm": 2.164802074432373,
      "learning_rate": 8.819294391151168e-06,
      "loss": 1.2056,
      "step": 32380
    },
    {
      "epoch": 8.655799037947622,
      "grad_norm": 2.162881851196289,
      "learning_rate": 8.784853056124709e-06,
      "loss": 1.0927,
      "step": 32390
    },
    {
      "epoch": 8.65847140566542,
      "grad_norm": 2.1710457801818848,
      "learning_rate": 8.750476013420883e-06,
      "loss": 1.2124,
      "step": 32400
    },
    {
      "epoch": 8.661143773383218,
      "grad_norm": 2.2281243801116943,
      "learning_rate": 8.716163287270152e-06,
      "loss": 1.1615,
      "step": 32410
    },
    {
      "epoch": 8.663816141101016,
      "grad_norm": 2.126492500305176,
      "learning_rate": 8.681914901857557e-06,
      "loss": 1.1706,
      "step": 32420
    },
    {
      "epoch": 8.666488508818814,
      "grad_norm": 2.3496429920196533,
      "learning_rate": 8.647730881322813e-06,
      "loss": 1.2675,
      "step": 32430
    },
    {
      "epoch": 8.669160876536612,
      "grad_norm": 2.2455477714538574,
      "learning_rate": 8.6136112497603e-06,
      "loss": 1.1817,
      "step": 32440
    },
    {
      "epoch": 8.67183324425441,
      "grad_norm": 2.2622618675231934,
      "learning_rate": 8.579556031218971e-06,
      "loss": 1.1119,
      "step": 32450
    },
    {
      "epoch": 8.674505611972208,
      "grad_norm": 2.265237331390381,
      "learning_rate": 8.545565249702403e-06,
      "loss": 1.1981,
      "step": 32460
    },
    {
      "epoch": 8.677177979690006,
      "grad_norm": 2.200049877166748,
      "learning_rate": 8.511638929168764e-06,
      "loss": 1.2551,
      "step": 32470
    },
    {
      "epoch": 8.679850347407804,
      "grad_norm": 2.108189344406128,
      "learning_rate": 8.477777093530704e-06,
      "loss": 1.1802,
      "step": 32480
    },
    {
      "epoch": 8.682522715125602,
      "grad_norm": 2.2164433002471924,
      "learning_rate": 8.443979766655585e-06,
      "loss": 1.2256,
      "step": 32490
    },
    {
      "epoch": 8.6851950828434,
      "grad_norm": 2.0696816444396973,
      "learning_rate": 8.410246972365131e-06,
      "loss": 1.1808,
      "step": 32500
    },
    {
      "epoch": 8.687867450561198,
      "grad_norm": 2.179058790206909,
      "learning_rate": 8.37657873443567e-06,
      "loss": 1.1595,
      "step": 32510
    },
    {
      "epoch": 8.690539818278996,
      "grad_norm": 2.022372245788574,
      "learning_rate": 8.342975076598014e-06,
      "loss": 1.1624,
      "step": 32520
    },
    {
      "epoch": 8.693212185996794,
      "grad_norm": 2.078016519546509,
      "learning_rate": 8.309436022537465e-06,
      "loss": 1.2175,
      "step": 32530
    },
    {
      "epoch": 8.695884553714592,
      "grad_norm": 2.112997055053711,
      "learning_rate": 8.275961595893777e-06,
      "loss": 1.2474,
      "step": 32540
    },
    {
      "epoch": 8.69855692143239,
      "grad_norm": 2.2092154026031494,
      "learning_rate": 8.242551820261125e-06,
      "loss": 1.2105,
      "step": 32550
    },
    {
      "epoch": 8.701229289150188,
      "grad_norm": 2.2042713165283203,
      "learning_rate": 8.209206719188168e-06,
      "loss": 1.2052,
      "step": 32560
    },
    {
      "epoch": 8.703901656867984,
      "grad_norm": 2.080826759338379,
      "learning_rate": 8.17592631617794e-06,
      "loss": 1.1409,
      "step": 32570
    },
    {
      "epoch": 8.706574024585784,
      "grad_norm": 2.205794095993042,
      "learning_rate": 8.142710634687911e-06,
      "loss": 1.1895,
      "step": 32580
    },
    {
      "epoch": 8.70924639230358,
      "grad_norm": 2.2292401790618896,
      "learning_rate": 8.109559698129853e-06,
      "loss": 1.1653,
      "step": 32590
    },
    {
      "epoch": 8.71191876002138,
      "grad_norm": 2.2298977375030518,
      "learning_rate": 8.076473529870043e-06,
      "loss": 1.2368,
      "step": 32600
    },
    {
      "epoch": 8.714591127739176,
      "grad_norm": 2.172769546508789,
      "learning_rate": 8.043452153228937e-06,
      "loss": 1.1762,
      "step": 32610
    },
    {
      "epoch": 8.717263495456974,
      "grad_norm": 2.3154499530792236,
      "learning_rate": 8.010495591481481e-06,
      "loss": 1.2175,
      "step": 32620
    },
    {
      "epoch": 8.719935863174772,
      "grad_norm": 2.349804401397705,
      "learning_rate": 7.97760386785682e-06,
      "loss": 1.1963,
      "step": 32630
    },
    {
      "epoch": 8.72260823089257,
      "grad_norm": 2.2397024631500244,
      "learning_rate": 7.944777005538451e-06,
      "loss": 1.184,
      "step": 32640
    },
    {
      "epoch": 8.725280598610368,
      "grad_norm": 2.1825761795043945,
      "learning_rate": 7.912015027664144e-06,
      "loss": 1.1751,
      "step": 32650
    },
    {
      "epoch": 8.727952966328166,
      "grad_norm": 2.283320188522339,
      "learning_rate": 7.87931795732596e-06,
      "loss": 1.2155,
      "step": 32660
    },
    {
      "epoch": 8.730625334045964,
      "grad_norm": 2.1549315452575684,
      "learning_rate": 7.846685817570142e-06,
      "loss": 1.1893,
      "step": 32670
    },
    {
      "epoch": 8.733297701763762,
      "grad_norm": 2.342792272567749,
      "learning_rate": 7.814118631397271e-06,
      "loss": 1.193,
      "step": 32680
    },
    {
      "epoch": 8.73597006948156,
      "grad_norm": 2.0795183181762695,
      "learning_rate": 7.781616421762028e-06,
      "loss": 1.161,
      "step": 32690
    },
    {
      "epoch": 8.738642437199358,
      "grad_norm": 2.1530849933624268,
      "learning_rate": 7.749179211573398e-06,
      "loss": 1.2372,
      "step": 32700
    },
    {
      "epoch": 8.741314804917156,
      "grad_norm": 2.294534206390381,
      "learning_rate": 7.716807023694494e-06,
      "loss": 1.2193,
      "step": 32710
    },
    {
      "epoch": 8.743987172634954,
      "grad_norm": 2.171541213989258,
      "learning_rate": 7.684499880942608e-06,
      "loss": 1.2213,
      "step": 32720
    },
    {
      "epoch": 8.746659540352752,
      "grad_norm": 2.1561813354492188,
      "learning_rate": 7.652257806089213e-06,
      "loss": 1.1438,
      "step": 32730
    },
    {
      "epoch": 8.74933190807055,
      "grad_norm": 2.322664260864258,
      "learning_rate": 7.620080821859876e-06,
      "loss": 1.1736,
      "step": 32740
    },
    {
      "epoch": 8.752004275788348,
      "grad_norm": 1.9789481163024902,
      "learning_rate": 7.587968950934299e-06,
      "loss": 1.2061,
      "step": 32750
    },
    {
      "epoch": 8.754676643506146,
      "grad_norm": 1.9960687160491943,
      "learning_rate": 7.55592221594632e-06,
      "loss": 1.134,
      "step": 32760
    },
    {
      "epoch": 8.757349011223944,
      "grad_norm": 2.099275588989258,
      "learning_rate": 7.523940639483828e-06,
      "loss": 1.2045,
      "step": 32770
    },
    {
      "epoch": 8.760021378941742,
      "grad_norm": 2.18982195854187,
      "learning_rate": 7.492024244088813e-06,
      "loss": 1.2143,
      "step": 32780
    },
    {
      "epoch": 8.76269374665954,
      "grad_norm": 2.0969889163970947,
      "learning_rate": 7.460173052257314e-06,
      "loss": 1.1896,
      "step": 32790
    },
    {
      "epoch": 8.765366114377338,
      "grad_norm": 2.050344467163086,
      "learning_rate": 7.42838708643937e-06,
      "loss": 1.1792,
      "step": 32800
    },
    {
      "epoch": 8.768038482095136,
      "grad_norm": 2.1144275665283203,
      "learning_rate": 7.396666369039152e-06,
      "loss": 1.225,
      "step": 32810
    },
    {
      "epoch": 8.770710849812934,
      "grad_norm": 2.132254123687744,
      "learning_rate": 7.365010922414706e-06,
      "loss": 1.2049,
      "step": 32820
    },
    {
      "epoch": 8.773383217530732,
      "grad_norm": 2.2373416423797607,
      "learning_rate": 7.333420768878174e-06,
      "loss": 1.2315,
      "step": 32830
    },
    {
      "epoch": 8.77605558524853,
      "grad_norm": 2.2836244106292725,
      "learning_rate": 7.301895930695635e-06,
      "loss": 1.2383,
      "step": 32840
    },
    {
      "epoch": 8.778727952966328,
      "grad_norm": 2.31648588180542,
      "learning_rate": 7.270436430087146e-06,
      "loss": 1.2515,
      "step": 32850
    },
    {
      "epoch": 8.781400320684126,
      "grad_norm": 2.178370952606201,
      "learning_rate": 7.239042289226694e-06,
      "loss": 1.18,
      "step": 32860
    },
    {
      "epoch": 8.784072688401924,
      "grad_norm": 2.1737213134765625,
      "learning_rate": 7.207713530242233e-06,
      "loss": 1.2426,
      "step": 32870
    },
    {
      "epoch": 8.786745056119722,
      "grad_norm": 2.117929220199585,
      "learning_rate": 7.176450175215555e-06,
      "loss": 1.1848,
      "step": 32880
    },
    {
      "epoch": 8.78941742383752,
      "grad_norm": 2.199974775314331,
      "learning_rate": 7.145252246182443e-06,
      "loss": 1.1761,
      "step": 32890
    },
    {
      "epoch": 8.792089791555318,
      "grad_norm": 2.190769910812378,
      "learning_rate": 7.114119765132521e-06,
      "loss": 1.1209,
      "step": 32900
    },
    {
      "epoch": 8.794762159273116,
      "grad_norm": 2.011537790298462,
      "learning_rate": 7.08305275400929e-06,
      "loss": 1.2093,
      "step": 32910
    },
    {
      "epoch": 8.797434526990914,
      "grad_norm": 2.2044901847839355,
      "learning_rate": 7.052051234710111e-06,
      "loss": 1.246,
      "step": 32920
    },
    {
      "epoch": 8.800106894708712,
      "grad_norm": 2.1224358081817627,
      "learning_rate": 7.0211152290861285e-06,
      "loss": 1.1448,
      "step": 32930
    },
    {
      "epoch": 8.80277926242651,
      "grad_norm": 2.1358206272125244,
      "learning_rate": 6.990244758942432e-06,
      "loss": 1.2381,
      "step": 32940
    },
    {
      "epoch": 8.805451630144308,
      "grad_norm": 2.263728380203247,
      "learning_rate": 6.959439846037796e-06,
      "loss": 1.1022,
      "step": 32950
    },
    {
      "epoch": 8.808123997862106,
      "grad_norm": 2.27034068107605,
      "learning_rate": 6.92870051208484e-06,
      "loss": 1.1611,
      "step": 32960
    },
    {
      "epoch": 8.810796365579904,
      "grad_norm": 2.3890955448150635,
      "learning_rate": 6.8980267787499556e-06,
      "loss": 1.2175,
      "step": 32970
    },
    {
      "epoch": 8.813468733297702,
      "grad_norm": 2.088869571685791,
      "learning_rate": 6.867418667653325e-06,
      "loss": 1.2053,
      "step": 32980
    },
    {
      "epoch": 8.8161411010155,
      "grad_norm": 2.095961809158325,
      "learning_rate": 6.836876200368802e-06,
      "loss": 1.1894,
      "step": 32990
    },
    {
      "epoch": 8.818813468733298,
      "grad_norm": 2.2726595401763916,
      "learning_rate": 6.806399398424079e-06,
      "loss": 1.2248,
      "step": 33000
    },
    {
      "epoch": 8.821485836451096,
      "grad_norm": 2.1984798908233643,
      "learning_rate": 6.775988283300427e-06,
      "loss": 1.1446,
      "step": 33010
    },
    {
      "epoch": 8.824158204168894,
      "grad_norm": 2.169297456741333,
      "learning_rate": 6.745642876432978e-06,
      "loss": 1.219,
      "step": 33020
    },
    {
      "epoch": 8.826830571886692,
      "grad_norm": 2.158966302871704,
      "learning_rate": 6.715363199210423e-06,
      "loss": 1.2234,
      "step": 33030
    },
    {
      "epoch": 8.82950293960449,
      "grad_norm": 2.1342318058013916,
      "learning_rate": 6.685149272975167e-06,
      "loss": 1.2455,
      "step": 33040
    },
    {
      "epoch": 8.832175307322288,
      "grad_norm": 2.495408773422241,
      "learning_rate": 6.655001119023285e-06,
      "loss": 1.1708,
      "step": 33050
    },
    {
      "epoch": 8.834847675040086,
      "grad_norm": 2.3328559398651123,
      "learning_rate": 6.624918758604492e-06,
      "loss": 1.2229,
      "step": 33060
    },
    {
      "epoch": 8.837520042757884,
      "grad_norm": 2.2289695739746094,
      "learning_rate": 6.594902212922083e-06,
      "loss": 1.1357,
      "step": 33070
    },
    {
      "epoch": 8.840192410475682,
      "grad_norm": 2.0448060035705566,
      "learning_rate": 6.564951503133032e-06,
      "loss": 1.1718,
      "step": 33080
    },
    {
      "epoch": 8.84286477819348,
      "grad_norm": 2.258244037628174,
      "learning_rate": 6.535066650347854e-06,
      "loss": 1.1893,
      "step": 33090
    },
    {
      "epoch": 8.845537145911278,
      "grad_norm": 2.1701242923736572,
      "learning_rate": 6.505247675630665e-06,
      "loss": 1.2069,
      "step": 33100
    },
    {
      "epoch": 8.848209513629076,
      "grad_norm": 2.173630952835083,
      "learning_rate": 6.475494599999188e-06,
      "loss": 1.251,
      "step": 33110
    },
    {
      "epoch": 8.850881881346874,
      "grad_norm": 2.2525391578674316,
      "learning_rate": 6.445807444424601e-06,
      "loss": 1.2198,
      "step": 33120
    },
    {
      "epoch": 8.853554249064672,
      "grad_norm": 2.163273334503174,
      "learning_rate": 6.416186229831722e-06,
      "loss": 1.1957,
      "step": 33130
    },
    {
      "epoch": 8.85622661678247,
      "grad_norm": 2.2199954986572266,
      "learning_rate": 6.386630977098828e-06,
      "loss": 1.1306,
      "step": 33140
    },
    {
      "epoch": 8.858898984500268,
      "grad_norm": 2.177464723587036,
      "learning_rate": 6.357141707057712e-06,
      "loss": 1.2086,
      "step": 33150
    },
    {
      "epoch": 8.861571352218066,
      "grad_norm": 2.008685350418091,
      "learning_rate": 6.327718440493679e-06,
      "loss": 1.2155,
      "step": 33160
    },
    {
      "epoch": 8.864243719935864,
      "grad_norm": 2.207615613937378,
      "learning_rate": 6.298361198145497e-06,
      "loss": 1.2484,
      "step": 33170
    },
    {
      "epoch": 8.866916087653662,
      "grad_norm": 2.3766024112701416,
      "learning_rate": 6.269070000705413e-06,
      "loss": 1.1647,
      "step": 33180
    },
    {
      "epoch": 8.86958845537146,
      "grad_norm": 2.1413516998291016,
      "learning_rate": 6.239844868819111e-06,
      "loss": 1.1999,
      "step": 33190
    },
    {
      "epoch": 8.872260823089258,
      "grad_norm": 2.293713331222534,
      "learning_rate": 6.210685823085671e-06,
      "loss": 1.1767,
      "step": 33200
    },
    {
      "epoch": 8.874933190807056,
      "grad_norm": 2.025331497192383,
      "learning_rate": 6.181592884057686e-06,
      "loss": 1.1347,
      "step": 33210
    },
    {
      "epoch": 8.877605558524852,
      "grad_norm": 2.303579568862915,
      "learning_rate": 6.152566072241061e-06,
      "loss": 1.1863,
      "step": 33220
    },
    {
      "epoch": 8.880277926242652,
      "grad_norm": 2.131896495819092,
      "learning_rate": 6.123605408095146e-06,
      "loss": 1.1295,
      "step": 33230
    },
    {
      "epoch": 8.882950293960448,
      "grad_norm": 2.34211802482605,
      "learning_rate": 6.094710912032642e-06,
      "loss": 1.1889,
      "step": 33240
    },
    {
      "epoch": 8.885622661678246,
      "grad_norm": 2.198781728744507,
      "learning_rate": 6.065882604419637e-06,
      "loss": 1.2392,
      "step": 33250
    },
    {
      "epoch": 8.888295029396044,
      "grad_norm": 2.364373207092285,
      "learning_rate": 6.037120505575544e-06,
      "loss": 1.2118,
      "step": 33260
    },
    {
      "epoch": 8.890967397113842,
      "grad_norm": 2.2449607849121094,
      "learning_rate": 6.008424635773102e-06,
      "loss": 1.1974,
      "step": 33270
    },
    {
      "epoch": 8.89363976483164,
      "grad_norm": 2.028512716293335,
      "learning_rate": 5.979795015238398e-06,
      "loss": 1.2539,
      "step": 33280
    },
    {
      "epoch": 8.896312132549438,
      "grad_norm": 2.094280242919922,
      "learning_rate": 5.951231664150803e-06,
      "loss": 1.1834,
      "step": 33290
    },
    {
      "epoch": 8.898984500267236,
      "grad_norm": 2.253570556640625,
      "learning_rate": 5.922734602642999e-06,
      "loss": 1.2057,
      "step": 33300
    },
    {
      "epoch": 8.901656867985034,
      "grad_norm": 2.342904567718506,
      "learning_rate": 5.894303850800886e-06,
      "loss": 1.2225,
      "step": 33310
    },
    {
      "epoch": 8.904329235702832,
      "grad_norm": 2.143791675567627,
      "learning_rate": 5.865939428663725e-06,
      "loss": 1.1796,
      "step": 33320
    },
    {
      "epoch": 8.90700160342063,
      "grad_norm": 2.2972323894500732,
      "learning_rate": 5.837641356223911e-06,
      "loss": 1.1537,
      "step": 33330
    },
    {
      "epoch": 8.909673971138428,
      "grad_norm": 2.1303181648254395,
      "learning_rate": 5.8094096534272e-06,
      "loss": 1.1488,
      "step": 33340
    },
    {
      "epoch": 8.912346338856226,
      "grad_norm": 2.275083065032959,
      "learning_rate": 5.7812443401724535e-06,
      "loss": 1.1373,
      "step": 33350
    },
    {
      "epoch": 8.915018706574024,
      "grad_norm": 2.0265774726867676,
      "learning_rate": 5.75314543631178e-06,
      "loss": 1.1475,
      "step": 33360
    },
    {
      "epoch": 8.917691074291822,
      "grad_norm": 2.5210697650909424,
      "learning_rate": 5.725112961650514e-06,
      "loss": 1.1884,
      "step": 33370
    },
    {
      "epoch": 8.92036344200962,
      "grad_norm": 2.2452547550201416,
      "learning_rate": 5.697146935947128e-06,
      "loss": 1.2297,
      "step": 33380
    },
    {
      "epoch": 8.923035809727418,
      "grad_norm": 2.1646676063537598,
      "learning_rate": 5.669247378913234e-06,
      "loss": 1.1503,
      "step": 33390
    },
    {
      "epoch": 8.925708177445216,
      "grad_norm": 2.314223527908325,
      "learning_rate": 5.641414310213688e-06,
      "loss": 1.1617,
      "step": 33400
    },
    {
      "epoch": 8.928380545163014,
      "grad_norm": 2.2160964012145996,
      "learning_rate": 5.613647749466377e-06,
      "loss": 1.1467,
      "step": 33410
    },
    {
      "epoch": 8.931052912880812,
      "grad_norm": 2.2023355960845947,
      "learning_rate": 5.585947716242379e-06,
      "loss": 1.1815,
      "step": 33420
    },
    {
      "epoch": 8.93372528059861,
      "grad_norm": 2.236827850341797,
      "learning_rate": 5.558314230065842e-06,
      "loss": 1.1166,
      "step": 33430
    },
    {
      "epoch": 8.936397648316408,
      "grad_norm": 2.114706039428711,
      "learning_rate": 5.530747310414042e-06,
      "loss": 1.1916,
      "step": 33440
    },
    {
      "epoch": 8.939070016034206,
      "grad_norm": 2.270576238632202,
      "learning_rate": 5.5032469767173245e-06,
      "loss": 1.2367,
      "step": 33450
    },
    {
      "epoch": 8.941742383752004,
      "grad_norm": 2.239755868911743,
      "learning_rate": 5.475813248359063e-06,
      "loss": 1.1749,
      "step": 33460
    },
    {
      "epoch": 8.944414751469802,
      "grad_norm": 2.1854758262634277,
      "learning_rate": 5.448446144675745e-06,
      "loss": 1.1235,
      "step": 33470
    },
    {
      "epoch": 8.9470871191876,
      "grad_norm": 2.4273905754089355,
      "learning_rate": 5.421145684956852e-06,
      "loss": 1.1714,
      "step": 33480
    },
    {
      "epoch": 8.949759486905398,
      "grad_norm": 2.011171817779541,
      "learning_rate": 5.3939118884449355e-06,
      "loss": 1.0991,
      "step": 33490
    },
    {
      "epoch": 8.952431854623196,
      "grad_norm": 2.3507046699523926,
      "learning_rate": 5.366744774335519e-06,
      "loss": 1.1888,
      "step": 33500
    },
    {
      "epoch": 8.955104222340994,
      "grad_norm": 2.2492406368255615,
      "learning_rate": 5.339644361777163e-06,
      "loss": 1.1963,
      "step": 33510
    },
    {
      "epoch": 8.957776590058792,
      "grad_norm": 2.262593984603882,
      "learning_rate": 5.312610669871354e-06,
      "loss": 1.1745,
      "step": 33520
    },
    {
      "epoch": 8.96044895777659,
      "grad_norm": 2.1749863624572754,
      "learning_rate": 5.285643717672628e-06,
      "loss": 1.2471,
      "step": 33530
    },
    {
      "epoch": 8.963121325494388,
      "grad_norm": 2.265610933303833,
      "learning_rate": 5.258743524188425e-06,
      "loss": 1.1776,
      "step": 33540
    },
    {
      "epoch": 8.965793693212186,
      "grad_norm": 2.293970823287964,
      "learning_rate": 5.231910108379145e-06,
      "loss": 1.1752,
      "step": 33550
    },
    {
      "epoch": 8.968466060929984,
      "grad_norm": 2.2802977561950684,
      "learning_rate": 5.205143489158138e-06,
      "loss": 1.1997,
      "step": 33560
    },
    {
      "epoch": 8.971138428647782,
      "grad_norm": 2.1155900955200195,
      "learning_rate": 5.178443685391643e-06,
      "loss": 1.2017,
      "step": 33570
    },
    {
      "epoch": 8.97381079636558,
      "grad_norm": 2.154723644256592,
      "learning_rate": 5.151810715898819e-06,
      "loss": 1.2119,
      "step": 33580
    },
    {
      "epoch": 8.976483164083378,
      "grad_norm": 2.347046375274658,
      "learning_rate": 5.12524459945174e-06,
      "loss": 1.1224,
      "step": 33590
    },
    {
      "epoch": 8.979155531801176,
      "grad_norm": 2.1975371837615967,
      "learning_rate": 5.098745354775303e-06,
      "loss": 1.1923,
      "step": 33600
    },
    {
      "epoch": 8.981827899518974,
      "grad_norm": 2.0898139476776123,
      "learning_rate": 5.072313000547335e-06,
      "loss": 1.1291,
      "step": 33610
    },
    {
      "epoch": 8.984500267236772,
      "grad_norm": 2.156890630722046,
      "learning_rate": 5.045947555398467e-06,
      "loss": 1.2336,
      "step": 33620
    },
    {
      "epoch": 8.98717263495457,
      "grad_norm": 2.219177722930908,
      "learning_rate": 5.0196490379121885e-06,
      "loss": 1.2069,
      "step": 33630
    },
    {
      "epoch": 8.989845002672368,
      "grad_norm": 2.0741376876831055,
      "learning_rate": 4.993417466624839e-06,
      "loss": 1.1385,
      "step": 33640
    },
    {
      "epoch": 8.992517370390166,
      "grad_norm": 2.0923755168914795,
      "learning_rate": 4.967252860025506e-06,
      "loss": 1.186,
      "step": 33650
    },
    {
      "epoch": 8.995189738107964,
      "grad_norm": 2.1709890365600586,
      "learning_rate": 4.941155236556161e-06,
      "loss": 1.2221,
      "step": 33660
    },
    {
      "epoch": 8.997862105825762,
      "grad_norm": 2.1641249656677246,
      "learning_rate": 4.9151246146115e-06,
      "loss": 1.1887,
      "step": 33670
    },
    {
      "epoch": 9.00053447354356,
      "grad_norm": 2.1118524074554443,
      "learning_rate": 4.889161012539012e-06,
      "loss": 1.0973,
      "step": 33680
    },
    {
      "epoch": 9.003206841261358,
      "grad_norm": 2.1524829864501953,
      "learning_rate": 4.86326444863896e-06,
      "loss": 1.2212,
      "step": 33690
    },
    {
      "epoch": 9.005879208979156,
      "grad_norm": 2.078460693359375,
      "learning_rate": 4.837434941164354e-06,
      "loss": 1.1211,
      "step": 33700
    },
    {
      "epoch": 9.008551576696954,
      "grad_norm": 2.1573588848114014,
      "learning_rate": 4.8116725083208966e-06,
      "loss": 1.1962,
      "step": 33710
    },
    {
      "epoch": 9.011223944414752,
      "grad_norm": 2.1500027179718018,
      "learning_rate": 4.785977168267097e-06,
      "loss": 1.0888,
      "step": 33720
    },
    {
      "epoch": 9.01389631213255,
      "grad_norm": 2.233577251434326,
      "learning_rate": 4.760348939114068e-06,
      "loss": 1.147,
      "step": 33730
    },
    {
      "epoch": 9.016568679850348,
      "grad_norm": 2.1687660217285156,
      "learning_rate": 4.734787838925736e-06,
      "loss": 1.1486,
      "step": 33740
    },
    {
      "epoch": 9.019241047568146,
      "grad_norm": 2.109074592590332,
      "learning_rate": 4.709293885718624e-06,
      "loss": 1.1148,
      "step": 33750
    },
    {
      "epoch": 9.021913415285944,
      "grad_norm": 1.9952311515808105,
      "learning_rate": 4.683867097461947e-06,
      "loss": 1.113,
      "step": 33760
    },
    {
      "epoch": 9.024585783003742,
      "grad_norm": 2.157128095626831,
      "learning_rate": 4.658507492077613e-06,
      "loss": 1.119,
      "step": 33770
    },
    {
      "epoch": 9.02725815072154,
      "grad_norm": 2.2225191593170166,
      "learning_rate": 4.633215087440157e-06,
      "loss": 1.1487,
      "step": 33780
    },
    {
      "epoch": 9.029930518439338,
      "grad_norm": 2.1577110290527344,
      "learning_rate": 4.607989901376708e-06,
      "loss": 1.1143,
      "step": 33790
    },
    {
      "epoch": 9.032602886157136,
      "grad_norm": 2.2246758937835693,
      "learning_rate": 4.5828319516670655e-06,
      "loss": 1.1549,
      "step": 33800
    },
    {
      "epoch": 9.035275253874934,
      "grad_norm": 2.27089524269104,
      "learning_rate": 4.557741256043646e-06,
      "loss": 1.1351,
      "step": 33810
    },
    {
      "epoch": 9.037947621592732,
      "grad_norm": 2.040964126586914,
      "learning_rate": 4.5327178321914135e-06,
      "loss": 1.1093,
      "step": 33820
    },
    {
      "epoch": 9.04061998931053,
      "grad_norm": 2.337263345718384,
      "learning_rate": 4.507761697747981e-06,
      "loss": 1.1322,
      "step": 33830
    },
    {
      "epoch": 9.043292357028328,
      "grad_norm": 2.1500494480133057,
      "learning_rate": 4.482872870303434e-06,
      "loss": 1.1621,
      "step": 33840
    },
    {
      "epoch": 9.045964724746126,
      "grad_norm": 2.226703643798828,
      "learning_rate": 4.458051367400551e-06,
      "loss": 1.1566,
      "step": 33850
    },
    {
      "epoch": 9.048637092463924,
      "grad_norm": 2.3370070457458496,
      "learning_rate": 4.433297206534537e-06,
      "loss": 1.1253,
      "step": 33860
    },
    {
      "epoch": 9.051309460181722,
      "grad_norm": 2.151838541030884,
      "learning_rate": 4.408610405153191e-06,
      "loss": 1.1143,
      "step": 33870
    },
    {
      "epoch": 9.053981827899518,
      "grad_norm": 2.4844799041748047,
      "learning_rate": 4.383990980656816e-06,
      "loss": 1.2107,
      "step": 33880
    },
    {
      "epoch": 9.056654195617316,
      "grad_norm": 2.3924453258514404,
      "learning_rate": 4.359438950398242e-06,
      "loss": 1.1648,
      "step": 33890
    },
    {
      "epoch": 9.059326563335114,
      "grad_norm": 2.0562124252319336,
      "learning_rate": 4.3349543316827836e-06,
      "loss": 1.087,
      "step": 33900
    },
    {
      "epoch": 9.061998931052912,
      "grad_norm": 2.1791796684265137,
      "learning_rate": 4.310537141768267e-06,
      "loss": 1.1827,
      "step": 33910
    },
    {
      "epoch": 9.06467129877071,
      "grad_norm": 2.0512163639068604,
      "learning_rate": 4.286187397864905e-06,
      "loss": 1.2029,
      "step": 33920
    },
    {
      "epoch": 9.067343666488508,
      "grad_norm": 2.287886381149292,
      "learning_rate": 4.261905117135512e-06,
      "loss": 1.1932,
      "step": 33930
    },
    {
      "epoch": 9.070016034206306,
      "grad_norm": 2.1602370738983154,
      "learning_rate": 4.237690316695208e-06,
      "loss": 1.1276,
      "step": 33940
    },
    {
      "epoch": 9.072688401924104,
      "grad_norm": 2.3687829971313477,
      "learning_rate": 4.213543013611654e-06,
      "loss": 1.1071,
      "step": 33950
    },
    {
      "epoch": 9.075360769641902,
      "grad_norm": 2.4160573482513428,
      "learning_rate": 4.1894632249048996e-06,
      "loss": 1.1945,
      "step": 33960
    },
    {
      "epoch": 9.0780331373597,
      "grad_norm": 2.2797768115997314,
      "learning_rate": 4.16545096754738e-06,
      "loss": 1.1039,
      "step": 33970
    },
    {
      "epoch": 9.080705505077498,
      "grad_norm": 2.1881837844848633,
      "learning_rate": 4.14150625846399e-06,
      "loss": 1.0811,
      "step": 33980
    },
    {
      "epoch": 9.083377872795296,
      "grad_norm": 2.1956918239593506,
      "learning_rate": 4.117629114531951e-06,
      "loss": 1.1245,
      "step": 33990
    },
    {
      "epoch": 9.086050240513094,
      "grad_norm": 2.291219711303711,
      "learning_rate": 4.093819552580902e-06,
      "loss": 1.218,
      "step": 34000
    },
    {
      "epoch": 9.088722608230892,
      "grad_norm": 2.258108615875244,
      "learning_rate": 4.0700775893928515e-06,
      "loss": 1.2022,
      "step": 34010
    },
    {
      "epoch": 9.09139497594869,
      "grad_norm": 2.1044251918792725,
      "learning_rate": 4.0464032417021345e-06,
      "loss": 1.1566,
      "step": 34020
    },
    {
      "epoch": 9.094067343666488,
      "grad_norm": 2.324979782104492,
      "learning_rate": 4.022796526195416e-06,
      "loss": 1.1808,
      "step": 34030
    },
    {
      "epoch": 9.096739711384286,
      "grad_norm": 2.1966042518615723,
      "learning_rate": 3.999257459511774e-06,
      "loss": 1.1668,
      "step": 34040
    },
    {
      "epoch": 9.099412079102084,
      "grad_norm": 2.2304556369781494,
      "learning_rate": 3.975786058242481e-06,
      "loss": 1.1108,
      "step": 34050
    },
    {
      "epoch": 9.102084446819882,
      "grad_norm": 2.3114962577819824,
      "learning_rate": 3.952382338931238e-06,
      "loss": 1.1604,
      "step": 34060
    },
    {
      "epoch": 9.10475681453768,
      "grad_norm": 2.3876705169677734,
      "learning_rate": 3.929046318073937e-06,
      "loss": 1.1378,
      "step": 34070
    },
    {
      "epoch": 9.107429182255478,
      "grad_norm": 2.2183847427368164,
      "learning_rate": 3.905778012118821e-06,
      "loss": 1.1565,
      "step": 34080
    },
    {
      "epoch": 9.110101549973276,
      "grad_norm": 2.278012752532959,
      "learning_rate": 3.882577437466395e-06,
      "loss": 1.1854,
      "step": 34090
    },
    {
      "epoch": 9.112773917691074,
      "grad_norm": 2.2312142848968506,
      "learning_rate": 3.859444610469398e-06,
      "loss": 1.1526,
      "step": 34100
    },
    {
      "epoch": 9.115446285408872,
      "grad_norm": 2.153632402420044,
      "learning_rate": 3.836379547432823e-06,
      "loss": 1.1506,
      "step": 34110
    },
    {
      "epoch": 9.11811865312667,
      "grad_norm": 2.2880208492279053,
      "learning_rate": 3.8133822646139293e-06,
      "loss": 1.1634,
      "step": 34120
    },
    {
      "epoch": 9.120791020844468,
      "grad_norm": 2.2320468425750732,
      "learning_rate": 3.7904527782221754e-06,
      "loss": 1.1958,
      "step": 34130
    },
    {
      "epoch": 9.123463388562266,
      "grad_norm": 2.315049648284912,
      "learning_rate": 3.7675911044192327e-06,
      "loss": 1.136,
      "step": 34140
    },
    {
      "epoch": 9.126135756280064,
      "grad_norm": 2.208770990371704,
      "learning_rate": 3.7447972593189884e-06,
      "loss": 1.1423,
      "step": 34150
    },
    {
      "epoch": 9.128808123997862,
      "grad_norm": 2.2792012691497803,
      "learning_rate": 3.722071258987514e-06,
      "loss": 1.1834,
      "step": 34160
    },
    {
      "epoch": 9.13148049171566,
      "grad_norm": 2.2620413303375244,
      "learning_rate": 3.6994131194430846e-06,
      "loss": 1.1729,
      "step": 34170
    },
    {
      "epoch": 9.134152859433458,
      "grad_norm": 2.3530848026275635,
      "learning_rate": 3.6768228566560812e-06,
      "loss": 1.1507,
      "step": 34180
    },
    {
      "epoch": 9.136825227151256,
      "grad_norm": 2.227083683013916,
      "learning_rate": 3.6543004865491113e-06,
      "loss": 1.2029,
      "step": 34190
    },
    {
      "epoch": 9.139497594869054,
      "grad_norm": 2.284367561340332,
      "learning_rate": 3.6318460249968987e-06,
      "loss": 1.0919,
      "step": 34200
    },
    {
      "epoch": 9.142169962586852,
      "grad_norm": 2.215280294418335,
      "learning_rate": 3.6094594878262942e-06,
      "loss": 1.2249,
      "step": 34210
    },
    {
      "epoch": 9.14484233030465,
      "grad_norm": 2.3096508979797363,
      "learning_rate": 3.5871408908162986e-06,
      "loss": 1.1523,
      "step": 34220
    },
    {
      "epoch": 9.147514698022448,
      "grad_norm": 2.094024896621704,
      "learning_rate": 3.564890249698005e-06,
      "loss": 1.1408,
      "step": 34230
    },
    {
      "epoch": 9.150187065740246,
      "grad_norm": 2.1089279651641846,
      "learning_rate": 3.5427075801545915e-06,
      "loss": 1.2022,
      "step": 34240
    },
    {
      "epoch": 9.152859433458044,
      "grad_norm": 2.203495740890503,
      "learning_rate": 3.520592897821395e-06,
      "loss": 1.1853,
      "step": 34250
    },
    {
      "epoch": 9.155531801175842,
      "grad_norm": 2.32757830619812,
      "learning_rate": 3.4985462182857366e-06,
      "loss": 1.1672,
      "step": 34260
    },
    {
      "epoch": 9.15820416889364,
      "grad_norm": 2.2531614303588867,
      "learning_rate": 3.476567557087085e-06,
      "loss": 1.131,
      "step": 34270
    },
    {
      "epoch": 9.160876536611438,
      "grad_norm": 2.1226727962493896,
      "learning_rate": 3.4546569297169372e-06,
      "loss": 1.129,
      "step": 34280
    },
    {
      "epoch": 9.163548904329236,
      "grad_norm": 2.287727117538452,
      "learning_rate": 3.432814351618818e-06,
      "loss": 1.2042,
      "step": 34290
    },
    {
      "epoch": 9.166221272047034,
      "grad_norm": 2.398174524307251,
      "learning_rate": 3.4110398381883237e-06,
      "loss": 1.1827,
      "step": 34300
    },
    {
      "epoch": 9.168893639764832,
      "grad_norm": 2.2923951148986816,
      "learning_rate": 3.3893334047730653e-06,
      "loss": 1.1051,
      "step": 34310
    },
    {
      "epoch": 9.17156600748263,
      "grad_norm": 2.2338578701019287,
      "learning_rate": 3.3676950666726493e-06,
      "loss": 1.1521,
      "step": 34320
    },
    {
      "epoch": 9.174238375200428,
      "grad_norm": 2.015608787536621,
      "learning_rate": 3.346124839138687e-06,
      "loss": 1.2117,
      "step": 34330
    },
    {
      "epoch": 9.176910742918226,
      "grad_norm": 2.1158697605133057,
      "learning_rate": 3.3246227373748164e-06,
      "loss": 1.1213,
      "step": 34340
    },
    {
      "epoch": 9.179583110636024,
      "grad_norm": 2.2086422443389893,
      "learning_rate": 3.3031887765366254e-06,
      "loss": 1.1276,
      "step": 34350
    },
    {
      "epoch": 9.182255478353822,
      "grad_norm": 2.3656766414642334,
      "learning_rate": 3.2818229717316964e-06,
      "loss": 1.1717,
      "step": 34360
    },
    {
      "epoch": 9.18492784607162,
      "grad_norm": 2.2100043296813965,
      "learning_rate": 3.260525338019527e-06,
      "loss": 1.2049,
      "step": 34370
    },
    {
      "epoch": 9.187600213789418,
      "grad_norm": 2.2874975204467773,
      "learning_rate": 3.239295890411631e-06,
      "loss": 1.1129,
      "step": 34380
    },
    {
      "epoch": 9.190272581507216,
      "grad_norm": 2.130533218383789,
      "learning_rate": 3.2181346438714065e-06,
      "loss": 1.0837,
      "step": 34390
    },
    {
      "epoch": 9.192944949225014,
      "grad_norm": 2.237393379211426,
      "learning_rate": 3.1970416133142112e-06,
      "loss": 1.1709,
      "step": 34400
    },
    {
      "epoch": 9.195617316942812,
      "grad_norm": 2.257607936859131,
      "learning_rate": 3.1760168136073074e-06,
      "loss": 1.1382,
      "step": 34410
    },
    {
      "epoch": 9.19828968466061,
      "grad_norm": 2.200047492980957,
      "learning_rate": 3.1550602595698863e-06,
      "loss": 1.1314,
      "step": 34420
    },
    {
      "epoch": 9.200962052378408,
      "grad_norm": 2.082170248031616,
      "learning_rate": 3.1341719659729764e-06,
      "loss": 1.1748,
      "step": 34430
    },
    {
      "epoch": 9.203634420096206,
      "grad_norm": 2.2494056224823,
      "learning_rate": 3.1133519475396e-06,
      "loss": 1.1469,
      "step": 34440
    },
    {
      "epoch": 9.206306787814004,
      "grad_norm": 2.122774839401245,
      "learning_rate": 3.0926002189445302e-06,
      "loss": 1.124,
      "step": 34450
    },
    {
      "epoch": 9.208979155531802,
      "grad_norm": 2.2540276050567627,
      "learning_rate": 3.071916794814522e-06,
      "loss": 1.1909,
      "step": 34460
    },
    {
      "epoch": 9.2116515232496,
      "grad_norm": 2.2521321773529053,
      "learning_rate": 3.051301689728103e-06,
      "loss": 1.2312,
      "step": 34470
    },
    {
      "epoch": 9.214323890967398,
      "grad_norm": 2.194652557373047,
      "learning_rate": 3.0307549182156834e-06,
      "loss": 1.0393,
      "step": 34480
    },
    {
      "epoch": 9.216996258685196,
      "grad_norm": 2.143259286880493,
      "learning_rate": 3.010276494759512e-06,
      "loss": 1.0846,
      "step": 34490
    },
    {
      "epoch": 9.219668626402994,
      "grad_norm": 2.321810245513916,
      "learning_rate": 2.9898664337936644e-06,
      "loss": 1.1721,
      "step": 34500
    },
    {
      "epoch": 9.222340994120792,
      "grad_norm": 2.3118536472320557,
      "learning_rate": 2.969524749703978e-06,
      "loss": 1.1513,
      "step": 34510
    },
    {
      "epoch": 9.225013361838588,
      "grad_norm": 2.189937114715576,
      "learning_rate": 2.949251456828184e-06,
      "loss": 1.0958,
      "step": 34520
    },
    {
      "epoch": 9.227685729556386,
      "grad_norm": 2.18630313873291,
      "learning_rate": 2.9290465694557402e-06,
      "loss": 1.1264,
      "step": 34530
    },
    {
      "epoch": 9.230358097274184,
      "grad_norm": 2.2235612869262695,
      "learning_rate": 2.9089101018279107e-06,
      "loss": 1.0883,
      "step": 34540
    },
    {
      "epoch": 9.233030464991982,
      "grad_norm": 2.0618996620178223,
      "learning_rate": 2.8888420681377647e-06,
      "loss": 1.0961,
      "step": 34550
    },
    {
      "epoch": 9.23570283270978,
      "grad_norm": 2.2291855812072754,
      "learning_rate": 2.8688424825300543e-06,
      "loss": 1.1772,
      "step": 34560
    },
    {
      "epoch": 9.238375200427578,
      "grad_norm": 2.2478034496307373,
      "learning_rate": 2.8489113591013917e-06,
      "loss": 1.1674,
      "step": 34570
    },
    {
      "epoch": 9.241047568145376,
      "grad_norm": 2.2389354705810547,
      "learning_rate": 2.829048711900062e-06,
      "loss": 1.2088,
      "step": 34580
    },
    {
      "epoch": 9.243719935863174,
      "grad_norm": 2.271409034729004,
      "learning_rate": 2.8092545549260997e-06,
      "loss": 1.1155,
      "step": 34590
    },
    {
      "epoch": 9.246392303580972,
      "grad_norm": 2.2548491954803467,
      "learning_rate": 2.7895289021312666e-06,
      "loss": 1.1604,
      "step": 34600
    },
    {
      "epoch": 9.24906467129877,
      "grad_norm": 2.3220374584198,
      "learning_rate": 2.7698717674190633e-06,
      "loss": 1.1154,
      "step": 34610
    },
    {
      "epoch": 9.251737039016568,
      "grad_norm": 2.199079990386963,
      "learning_rate": 2.7502831646446625e-06,
      "loss": 1.1318,
      "step": 34620
    },
    {
      "epoch": 9.254409406734366,
      "grad_norm": 2.409346580505371,
      "learning_rate": 2.7307631076149644e-06,
      "loss": 1.1433,
      "step": 34630
    },
    {
      "epoch": 9.257081774452164,
      "grad_norm": 2.2329723834991455,
      "learning_rate": 2.711311610088485e-06,
      "loss": 1.1395,
      "step": 34640
    },
    {
      "epoch": 9.259754142169962,
      "grad_norm": 2.2747905254364014,
      "learning_rate": 2.6919286857755355e-06,
      "loss": 1.1135,
      "step": 34650
    },
    {
      "epoch": 9.26242650988776,
      "grad_norm": 2.1231606006622314,
      "learning_rate": 2.6726143483379873e-06,
      "loss": 1.1629,
      "step": 34660
    },
    {
      "epoch": 9.265098877605558,
      "grad_norm": 2.458559989929199,
      "learning_rate": 2.6533686113894175e-06,
      "loss": 1.1092,
      "step": 34670
    },
    {
      "epoch": 9.267771245323356,
      "grad_norm": 2.1900291442871094,
      "learning_rate": 2.6341914884950414e-06,
      "loss": 1.1332,
      "step": 34680
    },
    {
      "epoch": 9.270443613041154,
      "grad_norm": 2.2296600341796875,
      "learning_rate": 2.6150829931717137e-06,
      "loss": 1.1402,
      "step": 34690
    },
    {
      "epoch": 9.273115980758952,
      "grad_norm": 2.1724588871002197,
      "learning_rate": 2.596043138887927e-06,
      "loss": 1.157,
      "step": 34700
    },
    {
      "epoch": 9.27578834847675,
      "grad_norm": 2.071732997894287,
      "learning_rate": 2.5770719390637467e-06,
      "loss": 1.1205,
      "step": 34710
    },
    {
      "epoch": 9.278460716194548,
      "grad_norm": 2.1276488304138184,
      "learning_rate": 2.5581694070709204e-06,
      "loss": 1.1443,
      "step": 34720
    },
    {
      "epoch": 9.281133083912346,
      "grad_norm": 2.182807445526123,
      "learning_rate": 2.539335556232736e-06,
      "loss": 1.0873,
      "step": 34730
    },
    {
      "epoch": 9.283805451630144,
      "grad_norm": 2.1881489753723145,
      "learning_rate": 2.520570399824118e-06,
      "loss": 1.1431,
      "step": 34740
    },
    {
      "epoch": 9.286477819347942,
      "grad_norm": 2.231375217437744,
      "learning_rate": 2.5018739510715093e-06,
      "loss": 1.2077,
      "step": 34750
    },
    {
      "epoch": 9.28915018706574,
      "grad_norm": 2.198413610458374,
      "learning_rate": 2.483246223153013e-06,
      "loss": 1.1648,
      "step": 34760
    },
    {
      "epoch": 9.291822554783538,
      "grad_norm": 2.266705274581909,
      "learning_rate": 2.4646872291981926e-06,
      "loss": 1.1607,
      "step": 34770
    },
    {
      "epoch": 9.294494922501336,
      "grad_norm": 2.1397669315338135,
      "learning_rate": 2.4461969822882735e-06,
      "loss": 1.0897,
      "step": 34780
    },
    {
      "epoch": 9.297167290219134,
      "grad_norm": 2.2377161979675293,
      "learning_rate": 2.4277754954559197e-06,
      "loss": 1.1907,
      "step": 34790
    },
    {
      "epoch": 9.299839657936932,
      "grad_norm": 2.3266890048980713,
      "learning_rate": 2.4094227816854107e-06,
      "loss": 1.1884,
      "step": 34800
    },
    {
      "epoch": 9.30251202565473,
      "grad_norm": 2.2268149852752686,
      "learning_rate": 2.3911388539125112e-06,
      "loss": 1.2367,
      "step": 34810
    },
    {
      "epoch": 9.305184393372528,
      "grad_norm": 2.183056116104126,
      "learning_rate": 2.372923725024523e-06,
      "loss": 1.1417,
      "step": 34820
    },
    {
      "epoch": 9.307856761090326,
      "grad_norm": 2.2253060340881348,
      "learning_rate": 2.354777407860209e-06,
      "loss": 1.1454,
      "step": 34830
    },
    {
      "epoch": 9.310529128808124,
      "grad_norm": 2.1809163093566895,
      "learning_rate": 2.3366999152099166e-06,
      "loss": 1.1244,
      "step": 34840
    },
    {
      "epoch": 9.313201496525922,
      "grad_norm": 2.2653419971466064,
      "learning_rate": 2.3186912598154086e-06,
      "loss": 1.1478,
      "step": 34850
    },
    {
      "epoch": 9.31587386424372,
      "grad_norm": 2.2308356761932373,
      "learning_rate": 2.300751454369943e-06,
      "loss": 1.1471,
      "step": 34860
    },
    {
      "epoch": 9.318546231961518,
      "grad_norm": 2.16866397857666,
      "learning_rate": 2.2828805115182816e-06,
      "loss": 1.1804,
      "step": 34870
    },
    {
      "epoch": 9.321218599679316,
      "grad_norm": 2.1032817363739014,
      "learning_rate": 2.2650784438566052e-06,
      "loss": 1.159,
      "step": 34880
    },
    {
      "epoch": 9.323890967397114,
      "grad_norm": 2.2278592586517334,
      "learning_rate": 2.2473452639325986e-06,
      "loss": 1.23,
      "step": 34890
    },
    {
      "epoch": 9.326563335114912,
      "grad_norm": 2.134263515472412,
      "learning_rate": 2.2296809842453194e-06,
      "loss": 1.1183,
      "step": 34900
    },
    {
      "epoch": 9.32923570283271,
      "grad_norm": 2.315495014190674,
      "learning_rate": 2.2120856172453408e-06,
      "loss": 1.1538,
      "step": 34910
    },
    {
      "epoch": 9.331908070550508,
      "grad_norm": 2.3744022846221924,
      "learning_rate": 2.1945591753346207e-06,
      "loss": 1.0838,
      "step": 34920
    },
    {
      "epoch": 9.334580438268306,
      "grad_norm": 2.1596031188964844,
      "learning_rate": 2.1771016708665217e-06,
      "loss": 1.1911,
      "step": 34930
    },
    {
      "epoch": 9.337252805986104,
      "grad_norm": 2.2654311656951904,
      "learning_rate": 2.159713116145867e-06,
      "loss": 1.1786,
      "step": 34940
    },
    {
      "epoch": 9.339925173703902,
      "grad_norm": 2.3472368717193604,
      "learning_rate": 2.1423935234288315e-06,
      "loss": 1.1545,
      "step": 34950
    },
    {
      "epoch": 9.3425975414217,
      "grad_norm": 2.106304168701172,
      "learning_rate": 2.1251429049230054e-06,
      "loss": 1.1079,
      "step": 34960
    },
    {
      "epoch": 9.345269909139498,
      "grad_norm": 2.091672897338867,
      "learning_rate": 2.107961272787362e-06,
      "loss": 1.2343,
      "step": 34970
    },
    {
      "epoch": 9.347942276857296,
      "grad_norm": 1.9937862157821655,
      "learning_rate": 2.090848639132248e-06,
      "loss": 1.162,
      "step": 34980
    },
    {
      "epoch": 9.350614644575094,
      "grad_norm": 2.2787647247314453,
      "learning_rate": 2.0738050160193813e-06,
      "loss": 1.119,
      "step": 34990
    },
    {
      "epoch": 9.353287012292892,
      "grad_norm": 2.159125566482544,
      "learning_rate": 2.056830415461819e-06,
      "loss": 1.1531,
      "step": 35000
    },
    {
      "epoch": 9.35595938001069,
      "grad_norm": 2.2107110023498535,
      "learning_rate": 2.039924849423991e-06,
      "loss": 1.1955,
      "step": 35010
    },
    {
      "epoch": 9.358631747728488,
      "grad_norm": 2.2851288318634033,
      "learning_rate": 2.0230883298216762e-06,
      "loss": 1.1427,
      "step": 35020
    },
    {
      "epoch": 9.361304115446286,
      "grad_norm": 2.156013250350952,
      "learning_rate": 2.0063208685219494e-06,
      "loss": 1.1963,
      "step": 35030
    },
    {
      "epoch": 9.363976483164084,
      "grad_norm": 2.153143882751465,
      "learning_rate": 1.9896224773432335e-06,
      "loss": 1.1267,
      "step": 35040
    },
    {
      "epoch": 9.366648850881882,
      "grad_norm": 2.4556376934051514,
      "learning_rate": 1.9729931680552814e-06,
      "loss": 1.0745,
      "step": 35050
    },
    {
      "epoch": 9.36932121859968,
      "grad_norm": 2.2956066131591797,
      "learning_rate": 1.9564329523791283e-06,
      "loss": 1.077,
      "step": 35060
    },
    {
      "epoch": 9.371993586317478,
      "grad_norm": 1.9940216541290283,
      "learning_rate": 1.939941841987114e-06,
      "loss": 1.1743,
      "step": 35070
    },
    {
      "epoch": 9.374665954035276,
      "grad_norm": 2.188178062438965,
      "learning_rate": 1.9235198485028973e-06,
      "loss": 1.1612,
      "step": 35080
    },
    {
      "epoch": 9.377338321753074,
      "grad_norm": 2.148359537124634,
      "learning_rate": 1.9071669835013627e-06,
      "loss": 1.1766,
      "step": 35090
    },
    {
      "epoch": 9.380010689470872,
      "grad_norm": 2.1279687881469727,
      "learning_rate": 1.8908832585087466e-06,
      "loss": 1.1796,
      "step": 35100
    },
    {
      "epoch": 9.38268305718867,
      "grad_norm": 2.182737350463867,
      "learning_rate": 1.8746686850024897e-06,
      "loss": 1.0833,
      "step": 35110
    },
    {
      "epoch": 9.385355424906468,
      "grad_norm": 2.1898319721221924,
      "learning_rate": 1.8585232744113169e-06,
      "loss": 1.2201,
      "step": 35120
    },
    {
      "epoch": 9.388027792624266,
      "grad_norm": 2.154783010482788,
      "learning_rate": 1.8424470381152137e-06,
      "loss": 1.2013,
      "step": 35130
    },
    {
      "epoch": 9.390700160342064,
      "grad_norm": 2.2670187950134277,
      "learning_rate": 1.826439987445383e-06,
      "loss": 1.1101,
      "step": 35140
    },
    {
      "epoch": 9.393372528059862,
      "grad_norm": 2.1478779315948486,
      "learning_rate": 1.8105021336842665e-06,
      "loss": 1.1717,
      "step": 35150
    },
    {
      "epoch": 9.39604489577766,
      "grad_norm": 2.2820193767547607,
      "learning_rate": 1.7946334880655668e-06,
      "loss": 1.1958,
      "step": 35160
    },
    {
      "epoch": 9.398717263495456,
      "grad_norm": 2.1999762058258057,
      "learning_rate": 1.7788340617741595e-06,
      "loss": 1.1903,
      "step": 35170
    },
    {
      "epoch": 9.401389631213254,
      "grad_norm": 2.127655506134033,
      "learning_rate": 1.7631038659461807e-06,
      "loss": 1.1321,
      "step": 35180
    },
    {
      "epoch": 9.404061998931052,
      "grad_norm": 2.36655855178833,
      "learning_rate": 1.7474429116689173e-06,
      "loss": 1.1444,
      "step": 35190
    },
    {
      "epoch": 9.40673436664885,
      "grad_norm": 2.2247183322906494,
      "learning_rate": 1.7318512099808949e-06,
      "loss": 1.1389,
      "step": 35200
    },
    {
      "epoch": 9.409406734366648,
      "grad_norm": 2.115464687347412,
      "learning_rate": 1.716328771871789e-06,
      "loss": 1.1312,
      "step": 35210
    },
    {
      "epoch": 9.412079102084446,
      "grad_norm": 2.1210875511169434,
      "learning_rate": 1.7008756082825038e-06,
      "loss": 1.2016,
      "step": 35220
    },
    {
      "epoch": 9.414751469802244,
      "grad_norm": 2.1347267627716064,
      "learning_rate": 1.68549173010506e-06,
      "loss": 1.1493,
      "step": 35230
    },
    {
      "epoch": 9.417423837520042,
      "grad_norm": 2.0214762687683105,
      "learning_rate": 1.6701771481826834e-06,
      "loss": 1.0916,
      "step": 35240
    },
    {
      "epoch": 9.42009620523784,
      "grad_norm": 2.2505288124084473,
      "learning_rate": 1.6549318733097395e-06,
      "loss": 1.1825,
      "step": 35250
    },
    {
      "epoch": 9.422768572955638,
      "grad_norm": 2.238849401473999,
      "learning_rate": 1.6397559162317555e-06,
      "loss": 1.1622,
      "step": 35260
    },
    {
      "epoch": 9.425440940673436,
      "grad_norm": 2.153085947036743,
      "learning_rate": 1.6246492876453968e-06,
      "loss": 1.1791,
      "step": 35270
    },
    {
      "epoch": 9.428113308391234,
      "grad_norm": 2.3326046466827393,
      "learning_rate": 1.6096119981984237e-06,
      "loss": 1.1357,
      "step": 35280
    },
    {
      "epoch": 9.430785676109032,
      "grad_norm": 2.0615546703338623,
      "learning_rate": 1.5946440584898136e-06,
      "loss": 1.1313,
      "step": 35290
    },
    {
      "epoch": 9.43345804382683,
      "grad_norm": 2.2044572830200195,
      "learning_rate": 1.5797454790695609e-06,
      "loss": 1.211,
      "step": 35300
    },
    {
      "epoch": 9.436130411544628,
      "grad_norm": 2.285611867904663,
      "learning_rate": 1.564916270438843e-06,
      "loss": 1.2329,
      "step": 35310
    },
    {
      "epoch": 9.438802779262426,
      "grad_norm": 2.199610710144043,
      "learning_rate": 1.5501564430499104e-06,
      "loss": 1.1352,
      "step": 35320
    },
    {
      "epoch": 9.441475146980224,
      "grad_norm": 2.1760666370391846,
      "learning_rate": 1.53546600730613e-06,
      "loss": 1.1469,
      "step": 35330
    },
    {
      "epoch": 9.444147514698022,
      "grad_norm": 2.078403949737549,
      "learning_rate": 1.5208449735619302e-06,
      "loss": 1.1517,
      "step": 35340
    },
    {
      "epoch": 9.44681988241582,
      "grad_norm": 2.2231712341308594,
      "learning_rate": 1.506293352122856e-06,
      "loss": 1.1508,
      "step": 35350
    },
    {
      "epoch": 9.449492250133618,
      "grad_norm": 2.1282825469970703,
      "learning_rate": 1.4918111532454815e-06,
      "loss": 1.1357,
      "step": 35360
    },
    {
      "epoch": 9.452164617851416,
      "grad_norm": 2.149660348892212,
      "learning_rate": 1.4773983871375186e-06,
      "loss": 1.1193,
      "step": 35370
    },
    {
      "epoch": 9.454836985569214,
      "grad_norm": 2.314764976501465,
      "learning_rate": 1.4630550639576635e-06,
      "loss": 1.1809,
      "step": 35380
    },
    {
      "epoch": 9.457509353287012,
      "grad_norm": 2.2163586616516113,
      "learning_rate": 1.448781193815718e-06,
      "loss": 1.1336,
      "step": 35390
    },
    {
      "epoch": 9.46018172100481,
      "grad_norm": 2.264066696166992,
      "learning_rate": 1.4345767867725123e-06,
      "loss": 1.1527,
      "step": 35400
    },
    {
      "epoch": 9.462854088722608,
      "grad_norm": 2.1492066383361816,
      "learning_rate": 1.4204418528399155e-06,
      "loss": 1.1342,
      "step": 35410
    },
    {
      "epoch": 9.465526456440406,
      "grad_norm": 2.2571280002593994,
      "learning_rate": 1.4063764019808468e-06,
      "loss": 1.1842,
      "step": 35420
    },
    {
      "epoch": 9.468198824158204,
      "grad_norm": 2.078209638595581,
      "learning_rate": 1.3923804441092092e-06,
      "loss": 1.2063,
      "step": 35430
    },
    {
      "epoch": 9.470871191876002,
      "grad_norm": 2.1398227214813232,
      "learning_rate": 1.3784539890899563e-06,
      "loss": 1.0818,
      "step": 35440
    },
    {
      "epoch": 9.4735435595938,
      "grad_norm": 2.125175952911377,
      "learning_rate": 1.364597046739058e-06,
      "loss": 1.1194,
      "step": 35450
    },
    {
      "epoch": 9.476215927311598,
      "grad_norm": 2.2943966388702393,
      "learning_rate": 1.3508096268234793e-06,
      "loss": 1.1576,
      "step": 35460
    },
    {
      "epoch": 9.478888295029396,
      "grad_norm": 2.25311017036438,
      "learning_rate": 1.3370917390611692e-06,
      "loss": 1.162,
      "step": 35470
    },
    {
      "epoch": 9.481560662747194,
      "grad_norm": 2.2590370178222656,
      "learning_rate": 1.3234433931210932e-06,
      "loss": 1.1264,
      "step": 35480
    },
    {
      "epoch": 9.484233030464992,
      "grad_norm": 2.213134288787842,
      "learning_rate": 1.3098645986231673e-06,
      "loss": 1.1393,
      "step": 35490
    },
    {
      "epoch": 9.48690539818279,
      "grad_norm": 2.335373878479004,
      "learning_rate": 1.2963553651383243e-06,
      "loss": 1.1824,
      "step": 35500
    },
    {
      "epoch": 9.489577765900588,
      "grad_norm": 2.319972515106201,
      "learning_rate": 1.2829157021884363e-06,
      "loss": 1.1844,
      "step": 35510
    },
    {
      "epoch": 9.492250133618386,
      "grad_norm": 2.00939679145813,
      "learning_rate": 1.2695456192463596e-06,
      "loss": 1.0984,
      "step": 35520
    },
    {
      "epoch": 9.494922501336184,
      "grad_norm": 2.128068685531616,
      "learning_rate": 1.2562451257358887e-06,
      "loss": 1.1354,
      "step": 35530
    },
    {
      "epoch": 9.497594869053982,
      "grad_norm": 2.1643667221069336,
      "learning_rate": 1.2430142310317805e-06,
      "loss": 1.1354,
      "step": 35540
    },
    {
      "epoch": 9.50026723677178,
      "grad_norm": 2.1348328590393066,
      "learning_rate": 1.2298529444597306e-06,
      "loss": 1.179,
      "step": 35550
    },
    {
      "epoch": 9.502939604489578,
      "grad_norm": 2.144758939743042,
      "learning_rate": 1.2167612752963965e-06,
      "loss": 1.1704,
      "step": 35560
    },
    {
      "epoch": 9.505611972207376,
      "grad_norm": 1.9720749855041504,
      "learning_rate": 1.2037392327693075e-06,
      "loss": 1.1452,
      "step": 35570
    },
    {
      "epoch": 9.508284339925174,
      "grad_norm": 2.160682439804077,
      "learning_rate": 1.1907868260569888e-06,
      "loss": 1.2285,
      "step": 35580
    },
    {
      "epoch": 9.510956707642972,
      "grad_norm": 2.20426869392395,
      "learning_rate": 1.177904064288826e-06,
      "loss": 1.1771,
      "step": 35590
    },
    {
      "epoch": 9.51362907536077,
      "grad_norm": 2.1965248584747314,
      "learning_rate": 1.1650909565451562e-06,
      "loss": 1.1335,
      "step": 35600
    },
    {
      "epoch": 9.516301443078568,
      "grad_norm": 2.211765766143799,
      "learning_rate": 1.1523475118571768e-06,
      "loss": 1.1642,
      "step": 35610
    },
    {
      "epoch": 9.518973810796366,
      "grad_norm": 2.3630828857421875,
      "learning_rate": 1.1396737392070479e-06,
      "loss": 1.161,
      "step": 35620
    },
    {
      "epoch": 9.521646178514164,
      "grad_norm": 2.196058511734009,
      "learning_rate": 1.1270696475277565e-06,
      "loss": 1.2022,
      "step": 35630
    },
    {
      "epoch": 9.524318546231962,
      "grad_norm": 2.264314889907837,
      "learning_rate": 1.1145352457032077e-06,
      "loss": 1.1206,
      "step": 35640
    },
    {
      "epoch": 9.52699091394976,
      "grad_norm": 2.4490818977355957,
      "learning_rate": 1.1020705425681898e-06,
      "loss": 1.1671,
      "step": 35650
    },
    {
      "epoch": 9.529663281667558,
      "grad_norm": 2.1642322540283203,
      "learning_rate": 1.089675546908353e-06,
      "loss": 1.1532,
      "step": 35660
    },
    {
      "epoch": 9.532335649385356,
      "grad_norm": 2.2017457485198975,
      "learning_rate": 1.0773502674602199e-06,
      "loss": 1.2436,
      "step": 35670
    },
    {
      "epoch": 9.535008017103154,
      "grad_norm": 2.204679250717163,
      "learning_rate": 1.0650947129111521e-06,
      "loss": 1.1109,
      "step": 35680
    },
    {
      "epoch": 9.537680384820952,
      "grad_norm": 2.31233811378479,
      "learning_rate": 1.052908891899429e-06,
      "loss": 1.1487,
      "step": 35690
    },
    {
      "epoch": 9.54035275253875,
      "grad_norm": 2.1574692726135254,
      "learning_rate": 1.040792813014102e-06,
      "loss": 1.2181,
      "step": 35700
    },
    {
      "epoch": 9.543025120256548,
      "grad_norm": 2.236241340637207,
      "learning_rate": 1.0287464847951177e-06,
      "loss": 1.1333,
      "step": 35710
    },
    {
      "epoch": 9.545697487974346,
      "grad_norm": 2.260523557662964,
      "learning_rate": 1.016769915733229e-06,
      "loss": 1.1369,
      "step": 35720
    },
    {
      "epoch": 9.548369855692144,
      "grad_norm": 2.384070873260498,
      "learning_rate": 1.0048631142700383e-06,
      "loss": 1.1293,
      "step": 35730
    },
    {
      "epoch": 9.551042223409942,
      "grad_norm": 2.217766761779785,
      "learning_rate": 9.93026088797977e-07,
      "loss": 1.165,
      "step": 35740
    },
    {
      "epoch": 9.55371459112774,
      "grad_norm": 2.254021644592285,
      "learning_rate": 9.812588476602712e-07,
      "loss": 1.1192,
      "step": 35750
    },
    {
      "epoch": 9.556386958845538,
      "grad_norm": 2.1699182987213135,
      "learning_rate": 9.695613991509644e-07,
      "loss": 1.1211,
      "step": 35760
    },
    {
      "epoch": 9.559059326563336,
      "grad_norm": 2.119053363800049,
      "learning_rate": 9.57933751514939e-07,
      "loss": 1.1655,
      "step": 35770
    },
    {
      "epoch": 9.561731694281134,
      "grad_norm": 2.1433210372924805,
      "learning_rate": 9.4637591294785e-07,
      "loss": 1.1766,
      "step": 35780
    },
    {
      "epoch": 9.564404061998932,
      "grad_norm": 2.2094874382019043,
      "learning_rate": 9.348878915961479e-07,
      "loss": 1.1003,
      "step": 35790
    },
    {
      "epoch": 9.567076429716728,
      "grad_norm": 2.401143789291382,
      "learning_rate": 9.234696955570776e-07,
      "loss": 1.1256,
      "step": 35800
    },
    {
      "epoch": 9.569748797434528,
      "grad_norm": 2.1601922512054443,
      "learning_rate": 9.121213328786793e-07,
      "loss": 1.1402,
      "step": 35810
    },
    {
      "epoch": 9.572421165152324,
      "grad_norm": 2.0809834003448486,
      "learning_rate": 9.008428115597766e-07,
      "loss": 1.1373,
      "step": 35820
    },
    {
      "epoch": 9.575093532870122,
      "grad_norm": 2.3695173263549805,
      "learning_rate": 8.896341395499108e-07,
      "loss": 1.1881,
      "step": 35830
    },
    {
      "epoch": 9.57776590058792,
      "grad_norm": 2.322136163711548,
      "learning_rate": 8.784953247494732e-07,
      "loss": 1.148,
      "step": 35840
    },
    {
      "epoch": 9.580438268305718,
      "grad_norm": 2.2366907596588135,
      "learning_rate": 8.674263750095502e-07,
      "loss": 1.1159,
      "step": 35850
    },
    {
      "epoch": 9.583110636023516,
      "grad_norm": 2.4116995334625244,
      "learning_rate": 8.564272981320121e-07,
      "loss": 1.2513,
      "step": 35860
    },
    {
      "epoch": 9.585783003741314,
      "grad_norm": 2.315009832382202,
      "learning_rate": 8.454981018694796e-07,
      "loss": 1.188,
      "step": 35870
    },
    {
      "epoch": 9.588455371459112,
      "grad_norm": 2.3434865474700928,
      "learning_rate": 8.346387939253352e-07,
      "loss": 1.1815,
      "step": 35880
    },
    {
      "epoch": 9.59112773917691,
      "grad_norm": 2.3306496143341064,
      "learning_rate": 8.238493819536452e-07,
      "loss": 1.1481,
      "step": 35890
    },
    {
      "epoch": 9.593800106894708,
      "grad_norm": 2.046522617340088,
      "learning_rate": 8.131298735592818e-07,
      "loss": 1.1876,
      "step": 35900
    },
    {
      "epoch": 9.596472474612506,
      "grad_norm": 2.22929310798645,
      "learning_rate": 8.024802762977901e-07,
      "loss": 1.2076,
      "step": 35910
    },
    {
      "epoch": 9.599144842330304,
      "grad_norm": 2.208465337753296,
      "learning_rate": 7.919005976754656e-07,
      "loss": 1.1427,
      "step": 35920
    },
    {
      "epoch": 9.601817210048102,
      "grad_norm": 2.2475030422210693,
      "learning_rate": 7.813908451493102e-07,
      "loss": 1.3049,
      "step": 35930
    },
    {
      "epoch": 9.6044895777659,
      "grad_norm": 2.2951815128326416,
      "learning_rate": 7.709510261270647e-07,
      "loss": 1.1495,
      "step": 35940
    },
    {
      "epoch": 9.607161945483698,
      "grad_norm": 2.2295608520507812,
      "learning_rate": 7.605811479671321e-07,
      "loss": 1.1845,
      "step": 35950
    },
    {
      "epoch": 9.609834313201496,
      "grad_norm": 2.262756109237671,
      "learning_rate": 7.502812179786767e-07,
      "loss": 1.1801,
      "step": 35960
    },
    {
      "epoch": 9.612506680919294,
      "grad_norm": 2.2262392044067383,
      "learning_rate": 7.400512434214912e-07,
      "loss": 1.2225,
      "step": 35970
    },
    {
      "epoch": 9.615179048637092,
      "grad_norm": 2.133723497390747,
      "learning_rate": 7.298912315061302e-07,
      "loss": 1.0648,
      "step": 35980
    },
    {
      "epoch": 9.61785141635489,
      "grad_norm": 2.137983798980713,
      "learning_rate": 7.198011893937761e-07,
      "loss": 1.1054,
      "step": 35990
    },
    {
      "epoch": 9.620523784072688,
      "grad_norm": 2.0405776500701904,
      "learning_rate": 7.097811241963403e-07,
      "loss": 1.1628,
      "step": 36000
    },
    {
      "epoch": 9.623196151790486,
      "grad_norm": 2.210630178451538,
      "learning_rate": 6.998310429763843e-07,
      "loss": 1.137,
      "step": 36010
    },
    {
      "epoch": 9.625868519508284,
      "grad_norm": 2.083991289138794,
      "learning_rate": 6.899509527471537e-07,
      "loss": 1.1504,
      "step": 36020
    },
    {
      "epoch": 9.628540887226082,
      "grad_norm": 2.0966291427612305,
      "learning_rate": 6.801408604725446e-07,
      "loss": 1.1015,
      "step": 36030
    },
    {
      "epoch": 9.63121325494388,
      "grad_norm": 2.172142505645752,
      "learning_rate": 6.70400773067148e-07,
      "loss": 1.1719,
      "step": 36040
    },
    {
      "epoch": 9.633885622661678,
      "grad_norm": 2.178936719894409,
      "learning_rate": 6.607306973961724e-07,
      "loss": 1.1747,
      "step": 36050
    },
    {
      "epoch": 9.636557990379476,
      "grad_norm": 2.1253082752227783,
      "learning_rate": 6.511306402755101e-07,
      "loss": 1.1207,
      "step": 36060
    },
    {
      "epoch": 9.639230358097274,
      "grad_norm": 2.2981467247009277,
      "learning_rate": 6.41600608471693e-07,
      "loss": 1.0727,
      "step": 36070
    },
    {
      "epoch": 9.641902725815072,
      "grad_norm": 2.133519172668457,
      "learning_rate": 6.321406087018811e-07,
      "loss": 1.1964,
      "step": 36080
    },
    {
      "epoch": 9.64457509353287,
      "grad_norm": 2.155704975128174,
      "learning_rate": 6.227506476338963e-07,
      "loss": 1.1886,
      "step": 36090
    },
    {
      "epoch": 9.647247461250668,
      "grad_norm": 2.2018682956695557,
      "learning_rate": 6.134307318861776e-07,
      "loss": 1.2259,
      "step": 36100
    },
    {
      "epoch": 9.649919828968466,
      "grad_norm": 2.2575747966766357,
      "learning_rate": 6.041808680278038e-07,
      "loss": 1.1502,
      "step": 36110
    },
    {
      "epoch": 9.652592196686264,
      "grad_norm": 2.2044577598571777,
      "learning_rate": 5.950010625784707e-07,
      "loss": 1.1595,
      "step": 36120
    },
    {
      "epoch": 9.655264564404062,
      "grad_norm": 2.363152265548706,
      "learning_rate": 5.858913220085027e-07,
      "loss": 1.1607,
      "step": 36130
    },
    {
      "epoch": 9.65793693212186,
      "grad_norm": 2.114800214767456,
      "learning_rate": 5.76851652738819e-07,
      "loss": 1.2111,
      "step": 36140
    },
    {
      "epoch": 9.660609299839658,
      "grad_norm": 2.2297019958496094,
      "learning_rate": 5.678820611409785e-07,
      "loss": 1.1889,
      "step": 36150
    },
    {
      "epoch": 9.663281667557456,
      "grad_norm": 2.206164598464966,
      "learning_rate": 5.58982553537124e-07,
      "loss": 1.1724,
      "step": 36160
    },
    {
      "epoch": 9.665954035275254,
      "grad_norm": 2.4453980922698975,
      "learning_rate": 5.501531362000046e-07,
      "loss": 1.1992,
      "step": 36170
    },
    {
      "epoch": 9.668626402993052,
      "grad_norm": 2.4066483974456787,
      "learning_rate": 5.413938153529752e-07,
      "loss": 1.2186,
      "step": 36180
    },
    {
      "epoch": 9.67129877071085,
      "grad_norm": 2.136568784713745,
      "learning_rate": 5.327045971699862e-07,
      "loss": 1.173,
      "step": 36190
    },
    {
      "epoch": 9.673971138428648,
      "grad_norm": 2.1053264141082764,
      "learning_rate": 5.240854877755497e-07,
      "loss": 1.1664,
      "step": 36200
    },
    {
      "epoch": 9.676643506146446,
      "grad_norm": 2.2664544582366943,
      "learning_rate": 5.155364932447837e-07,
      "loss": 1.2088,
      "step": 36210
    },
    {
      "epoch": 9.679315873864244,
      "grad_norm": 2.215672254562378,
      "learning_rate": 5.070576196034127e-07,
      "loss": 1.1669,
      "step": 36220
    },
    {
      "epoch": 9.681988241582042,
      "grad_norm": 2.0596985816955566,
      "learning_rate": 4.986488728276673e-07,
      "loss": 1.1515,
      "step": 36230
    },
    {
      "epoch": 9.68466060929984,
      "grad_norm": 2.085028886795044,
      "learning_rate": 4.903102588444175e-07,
      "loss": 1.15,
      "step": 36240
    },
    {
      "epoch": 9.687332977017638,
      "grad_norm": 2.119126081466675,
      "learning_rate": 4.820417835310731e-07,
      "loss": 1.1824,
      "step": 36250
    },
    {
      "epoch": 9.690005344735436,
      "grad_norm": 2.221674919128418,
      "learning_rate": 4.738434527155944e-07,
      "loss": 1.1601,
      "step": 36260
    },
    {
      "epoch": 9.692677712453234,
      "grad_norm": 2.2603559494018555,
      "learning_rate": 4.6571527217652566e-07,
      "loss": 1.1711,
      "step": 36270
    },
    {
      "epoch": 9.695350080171032,
      "grad_norm": 2.2842955589294434,
      "learning_rate": 4.576572476429508e-07,
      "loss": 1.1509,
      "step": 36280
    },
    {
      "epoch": 9.69802244788883,
      "grad_norm": 2.3427512645721436,
      "learning_rate": 4.4966938479449326e-07,
      "loss": 1.1184,
      "step": 36290
    },
    {
      "epoch": 9.700694815606628,
      "grad_norm": 2.1294307708740234,
      "learning_rate": 4.417516892613716e-07,
      "loss": 1.1445,
      "step": 36300
    },
    {
      "epoch": 9.703367183324426,
      "grad_norm": 2.2346582412719727,
      "learning_rate": 4.3390416662428823e-07,
      "loss": 1.1325,
      "step": 36310
    },
    {
      "epoch": 9.706039551042224,
      "grad_norm": 2.1415388584136963,
      "learning_rate": 4.2612682241452986e-07,
      "loss": 1.0958,
      "step": 36320
    },
    {
      "epoch": 9.708711918760022,
      "grad_norm": 2.217890977859497,
      "learning_rate": 4.1841966211388914e-07,
      "loss": 1.1301,
      "step": 36330
    },
    {
      "epoch": 9.71138428647782,
      "grad_norm": 2.219364881515503,
      "learning_rate": 4.107826911546986e-07,
      "loss": 1.1017,
      "step": 36340
    },
    {
      "epoch": 9.714056654195618,
      "grad_norm": 2.173316717147827,
      "learning_rate": 4.0321591491983e-07,
      "loss": 1.159,
      "step": 36350
    },
    {
      "epoch": 9.716729021913416,
      "grad_norm": 2.2166037559509277,
      "learning_rate": 3.9571933874266163e-07,
      "loss": 1.1681,
      "step": 36360
    },
    {
      "epoch": 9.719401389631214,
      "grad_norm": 2.2103919982910156,
      "learning_rate": 3.882929679071112e-07,
      "loss": 1.1577,
      "step": 36370
    },
    {
      "epoch": 9.722073757349012,
      "grad_norm": 2.0758190155029297,
      "learning_rate": 3.8093680764758054e-07,
      "loss": 1.2246,
      "step": 36380
    },
    {
      "epoch": 9.72474612506681,
      "grad_norm": 2.353356122970581,
      "learning_rate": 3.736508631490332e-07,
      "loss": 1.1729,
      "step": 36390
    },
    {
      "epoch": 9.727418492784608,
      "grad_norm": 1.9840309619903564,
      "learning_rate": 3.664351395468835e-07,
      "loss": 1.068,
      "step": 36400
    },
    {
      "epoch": 9.730090860502406,
      "grad_norm": 2.2292327880859375,
      "learning_rate": 3.592896419271075e-07,
      "loss": 1.1288,
      "step": 36410
    },
    {
      "epoch": 9.732763228220204,
      "grad_norm": 2.07846999168396,
      "learning_rate": 3.522143753261431e-07,
      "loss": 1.1392,
      "step": 36420
    },
    {
      "epoch": 9.735435595938002,
      "grad_norm": 2.6675517559051514,
      "learning_rate": 3.452093447309346e-07,
      "loss": 1.1,
      "step": 36430
    },
    {
      "epoch": 9.7381079636558,
      "grad_norm": 2.240673303604126,
      "learning_rate": 3.3827455507894345e-07,
      "loss": 1.1353,
      "step": 36440
    },
    {
      "epoch": 9.740780331373596,
      "grad_norm": 2.311814785003662,
      "learning_rate": 3.31410011258082e-07,
      "loss": 1.1452,
      "step": 36450
    },
    {
      "epoch": 9.743452699091396,
      "grad_norm": 2.222055673599243,
      "learning_rate": 3.2461571810679104e-07,
      "loss": 1.1698,
      "step": 36460
    },
    {
      "epoch": 9.746125066809192,
      "grad_norm": 2.24259614944458,
      "learning_rate": 3.178916804139731e-07,
      "loss": 1.1597,
      "step": 36470
    },
    {
      "epoch": 9.74879743452699,
      "grad_norm": 2.257625102996826,
      "learning_rate": 3.112379029190149e-07,
      "loss": 1.1344,
      "step": 36480
    },
    {
      "epoch": 9.751469802244788,
      "grad_norm": 2.3131158351898193,
      "learning_rate": 3.04654390311776e-07,
      "loss": 1.1931,
      "step": 36490
    },
    {
      "epoch": 9.754142169962586,
      "grad_norm": 2.151228427886963,
      "learning_rate": 2.981411472326112e-07,
      "loss": 1.2228,
      "step": 36500
    },
    {
      "epoch": 9.756814537680384,
      "grad_norm": 2.3429644107818604,
      "learning_rate": 2.91698178272326e-07,
      "loss": 1.1212,
      "step": 36510
    },
    {
      "epoch": 9.759486905398182,
      "grad_norm": 2.1811740398406982,
      "learning_rate": 2.853254879721989e-07,
      "loss": 1.1278,
      "step": 36520
    },
    {
      "epoch": 9.76215927311598,
      "grad_norm": 2.123075008392334,
      "learning_rate": 2.790230808239591e-07,
      "loss": 1.1174,
      "step": 36530
    },
    {
      "epoch": 9.764831640833778,
      "grad_norm": 2.2247934341430664,
      "learning_rate": 2.7279096126984203e-07,
      "loss": 1.1534,
      "step": 36540
    },
    {
      "epoch": 9.767504008551576,
      "grad_norm": 2.342512845993042,
      "learning_rate": 2.6662913370248956e-07,
      "loss": 1.1981,
      "step": 36550
    },
    {
      "epoch": 9.770176376269374,
      "grad_norm": 2.2304458618164062,
      "learning_rate": 2.6053760246501634e-07,
      "loss": 1.1064,
      "step": 36560
    },
    {
      "epoch": 9.772848743987172,
      "grad_norm": 2.1226627826690674,
      "learning_rate": 2.545163718510102e-07,
      "loss": 1.1067,
      "step": 36570
    },
    {
      "epoch": 9.77552111170497,
      "grad_norm": 2.2388312816619873,
      "learning_rate": 2.485654461044873e-07,
      "loss": 1.1208,
      "step": 36580
    },
    {
      "epoch": 9.778193479422768,
      "grad_norm": 2.459993839263916,
      "learning_rate": 2.4268482941989244e-07,
      "loss": 1.1374,
      "step": 36590
    },
    {
      "epoch": 9.780865847140566,
      "grad_norm": 2.284939765930176,
      "learning_rate": 2.3687452594216563e-07,
      "loss": 1.1587,
      "step": 36600
    },
    {
      "epoch": 9.783538214858364,
      "grad_norm": 2.184542179107666,
      "learning_rate": 2.3113453976663092e-07,
      "loss": 1.1153,
      "step": 36610
    },
    {
      "epoch": 9.786210582576162,
      "grad_norm": 2.275195360183716,
      "learning_rate": 2.254648749390964e-07,
      "loss": 1.205,
      "step": 36620
    },
    {
      "epoch": 9.78888295029396,
      "grad_norm": 2.168830394744873,
      "learning_rate": 2.198655354557766e-07,
      "loss": 1.0811,
      "step": 36630
    },
    {
      "epoch": 9.791555318011758,
      "grad_norm": 2.296387195587158,
      "learning_rate": 2.143365252633145e-07,
      "loss": 1.1222,
      "step": 36640
    },
    {
      "epoch": 9.794227685729556,
      "grad_norm": 2.32746958732605,
      "learning_rate": 2.0887784825880385e-07,
      "loss": 1.087,
      "step": 36650
    },
    {
      "epoch": 9.796900053447354,
      "grad_norm": 2.0658833980560303,
      "learning_rate": 2.034895082897448e-07,
      "loss": 1.0945,
      "step": 36660
    },
    {
      "epoch": 9.799572421165152,
      "grad_norm": 2.0493061542510986,
      "learning_rate": 1.981715091540659e-07,
      "loss": 1.1167,
      "step": 36670
    },
    {
      "epoch": 9.80224478888295,
      "grad_norm": 2.1636769771575928,
      "learning_rate": 1.929238546001355e-07,
      "loss": 1.1381,
      "step": 36680
    },
    {
      "epoch": 9.804917156600748,
      "grad_norm": 2.239959239959717,
      "learning_rate": 1.8774654832671712e-07,
      "loss": 1.1333,
      "step": 36690
    },
    {
      "epoch": 9.807589524318546,
      "grad_norm": 2.1792783737182617,
      "learning_rate": 1.8263959398298058e-07,
      "loss": 1.1398,
      "step": 36700
    },
    {
      "epoch": 9.810261892036344,
      "grad_norm": 2.1390347480773926,
      "learning_rate": 1.776029951685465e-07,
      "loss": 1.096,
      "step": 36710
    },
    {
      "epoch": 9.812934259754142,
      "grad_norm": 2.2639973163604736,
      "learning_rate": 1.726367554334196e-07,
      "loss": 1.2209,
      "step": 36720
    },
    {
      "epoch": 9.81560662747194,
      "grad_norm": 2.128241539001465,
      "learning_rate": 1.6774087827801098e-07,
      "loss": 1.1625,
      "step": 36730
    },
    {
      "epoch": 9.818278995189738,
      "grad_norm": 2.096963405609131,
      "learning_rate": 1.6291536715314915e-07,
      "loss": 1.1966,
      "step": 36740
    },
    {
      "epoch": 9.820951362907536,
      "grad_norm": 2.1843740940093994,
      "learning_rate": 1.5816022546005782e-07,
      "loss": 1.1901,
      "step": 36750
    },
    {
      "epoch": 9.823623730625334,
      "grad_norm": 2.144131660461426,
      "learning_rate": 1.534754565503671e-07,
      "loss": 1.1724,
      "step": 36760
    },
    {
      "epoch": 9.826296098343132,
      "grad_norm": 2.0758657455444336,
      "learning_rate": 1.4886106372609122e-07,
      "loss": 1.1841,
      "step": 36770
    },
    {
      "epoch": 9.82896846606093,
      "grad_norm": 2.2896933555603027,
      "learning_rate": 1.443170502396507e-07,
      "loss": 1.1162,
      "step": 36780
    },
    {
      "epoch": 9.831640833778728,
      "grad_norm": 2.102187395095825,
      "learning_rate": 1.3984341929388356e-07,
      "loss": 1.1621,
      "step": 36790
    },
    {
      "epoch": 9.834313201496526,
      "grad_norm": 1.845706820487976,
      "learning_rate": 1.3544017404196751e-07,
      "loss": 1.0704,
      "step": 36800
    },
    {
      "epoch": 9.836985569214324,
      "grad_norm": 2.300069570541382,
      "learning_rate": 1.3110731758750882e-07,
      "loss": 1.1541,
      "step": 36810
    },
    {
      "epoch": 9.839657936932122,
      "grad_norm": 2.137861728668213,
      "learning_rate": 1.2684485298449789e-07,
      "loss": 1.1428,
      "step": 36820
    },
    {
      "epoch": 9.84233030464992,
      "grad_norm": 2.2372660636901855,
      "learning_rate": 1.2265278323728703e-07,
      "loss": 1.1281,
      "step": 36830
    },
    {
      "epoch": 9.845002672367718,
      "grad_norm": 2.1376864910125732,
      "learning_rate": 1.1853111130063487e-07,
      "loss": 1.1876,
      "step": 36840
    },
    {
      "epoch": 9.847675040085516,
      "grad_norm": 2.1345643997192383,
      "learning_rate": 1.1447984007966206e-07,
      "loss": 1.1304,
      "step": 36850
    },
    {
      "epoch": 9.850347407803314,
      "grad_norm": 2.3281424045562744,
      "learning_rate": 1.1049897242988438e-07,
      "loss": 1.1783,
      "step": 36860
    },
    {
      "epoch": 9.853019775521112,
      "grad_norm": 2.2077133655548096,
      "learning_rate": 1.0658851115719071e-07,
      "loss": 1.101,
      "step": 36870
    },
    {
      "epoch": 9.85569214323891,
      "grad_norm": 2.10617995262146,
      "learning_rate": 1.0274845901783182e-07,
      "loss": 1.1355,
      "step": 36880
    },
    {
      "epoch": 9.858364510956708,
      "grad_norm": 2.1513755321502686,
      "learning_rate": 9.897881871844262e-08,
      "loss": 1.1561,
      "step": 36890
    },
    {
      "epoch": 9.861036878674506,
      "grad_norm": 2.424665689468384,
      "learning_rate": 9.527959291603105e-08,
      "loss": 1.2086,
      "step": 36900
    },
    {
      "epoch": 9.863709246392304,
      "grad_norm": 2.4013102054595947,
      "learning_rate": 9.165078421795593e-08,
      "loss": 1.2428,
      "step": 36910
    },
    {
      "epoch": 9.866381614110102,
      "grad_norm": 2.4641964435577393,
      "learning_rate": 8.809239518198231e-08,
      "loss": 1.1453,
      "step": 36920
    },
    {
      "epoch": 9.8690539818279,
      "grad_norm": 2.3074302673339844,
      "learning_rate": 8.460442831619286e-08,
      "loss": 1.1758,
      "step": 36930
    },
    {
      "epoch": 9.871726349545698,
      "grad_norm": 2.1441409587860107,
      "learning_rate": 8.118688607905433e-08,
      "loss": 1.1102,
      "step": 36940
    },
    {
      "epoch": 9.874398717263496,
      "grad_norm": 2.108098268508911,
      "learning_rate": 7.783977087940652e-08,
      "loss": 1.1661,
      "step": 36950
    },
    {
      "epoch": 9.877071084981294,
      "grad_norm": 2.226764678955078,
      "learning_rate": 7.456308507642896e-08,
      "loss": 1.119,
      "step": 36960
    },
    {
      "epoch": 9.879743452699092,
      "grad_norm": 2.0593080520629883,
      "learning_rate": 7.135683097968527e-08,
      "loss": 1.0746,
      "step": 36970
    },
    {
      "epoch": 9.88241582041689,
      "grad_norm": 2.4087886810302734,
      "learning_rate": 6.822101084906774e-08,
      "loss": 1.1997,
      "step": 36980
    },
    {
      "epoch": 9.885088188134688,
      "grad_norm": 2.1702661514282227,
      "learning_rate": 6.515562689483056e-08,
      "loss": 1.1641,
      "step": 36990
    },
    {
      "epoch": 9.887760555852486,
      "grad_norm": 2.217583179473877,
      "learning_rate": 6.2160681277601e-08,
      "loss": 1.1708,
      "step": 37000
    },
    {
      "epoch": 9.890432923570284,
      "grad_norm": 2.226370096206665,
      "learning_rate": 5.923617610833487e-08,
      "loss": 1.2202,
      "step": 37010
    },
    {
      "epoch": 9.893105291288082,
      "grad_norm": 2.27646803855896,
      "learning_rate": 5.638211344836108e-08,
      "loss": 1.1542,
      "step": 37020
    },
    {
      "epoch": 9.89577765900588,
      "grad_norm": 2.170738697052002,
      "learning_rate": 5.359849530932603e-08,
      "loss": 1.1387,
      "step": 37030
    },
    {
      "epoch": 9.898450026723678,
      "grad_norm": 2.182894468307495,
      "learning_rate": 5.088532365324916e-08,
      "loss": 1.1364,
      "step": 37040
    },
    {
      "epoch": 9.901122394441476,
      "grad_norm": 2.218874216079712,
      "learning_rate": 4.824260039250072e-08,
      "loss": 1.1479,
      "step": 37050
    },
    {
      "epoch": 9.903794762159274,
      "grad_norm": 1.9596279859542847,
      "learning_rate": 4.567032738976851e-08,
      "loss": 1.143,
      "step": 37060
    },
    {
      "epoch": 9.906467129877072,
      "grad_norm": 2.1424190998077393,
      "learning_rate": 4.316850645810222e-08,
      "loss": 1.1494,
      "step": 37070
    },
    {
      "epoch": 9.909139497594868,
      "grad_norm": 2.3702521324157715,
      "learning_rate": 4.07371393609024e-08,
      "loss": 1.1363,
      "step": 37080
    },
    {
      "epoch": 9.911811865312668,
      "grad_norm": 2.1741983890533447,
      "learning_rate": 3.83762278118982e-08,
      "loss": 1.1109,
      "step": 37090
    },
    {
      "epoch": 9.914484233030464,
      "grad_norm": 2.1364736557006836,
      "learning_rate": 3.6085773475147413e-08,
      "loss": 1.1601,
      "step": 37100
    },
    {
      "epoch": 9.917156600748264,
      "grad_norm": 2.38227915763855,
      "learning_rate": 3.386577796508084e-08,
      "loss": 1.2247,
      "step": 37110
    },
    {
      "epoch": 9.91982896846606,
      "grad_norm": 2.1709582805633545,
      "learning_rate": 3.171624284642461e-08,
      "loss": 1.1875,
      "step": 37120
    },
    {
      "epoch": 9.922501336183858,
      "grad_norm": 2.435558319091797,
      "learning_rate": 2.9637169634277872e-08,
      "loss": 1.1001,
      "step": 37130
    },
    {
      "epoch": 9.925173703901656,
      "grad_norm": 2.0772573947906494,
      "learning_rate": 2.7628559794057317e-08,
      "loss": 1.138,
      "step": 37140
    },
    {
      "epoch": 9.927846071619454,
      "grad_norm": 2.2174742221832275,
      "learning_rate": 2.569041474150824e-08,
      "loss": 1.1254,
      "step": 37150
    },
    {
      "epoch": 9.930518439337252,
      "grad_norm": 2.2710330486297607,
      "learning_rate": 2.3822735842726763e-08,
      "loss": 1.1296,
      "step": 37160
    },
    {
      "epoch": 9.93319080705505,
      "grad_norm": 2.1777970790863037,
      "learning_rate": 2.202552441412653e-08,
      "loss": 1.1039,
      "step": 37170
    },
    {
      "epoch": 9.935863174772848,
      "grad_norm": 2.194032669067383,
      "learning_rate": 2.0298781722460912e-08,
      "loss": 1.1462,
      "step": 37180
    },
    {
      "epoch": 9.938535542490646,
      "grad_norm": 2.1591246128082275,
      "learning_rate": 1.86425089848119e-08,
      "loss": 1.1566,
      "step": 37190
    },
    {
      "epoch": 9.941207910208444,
      "grad_norm": 2.110175132751465,
      "learning_rate": 1.705670736860121e-08,
      "loss": 1.138,
      "step": 37200
    },
    {
      "epoch": 9.943880277926242,
      "grad_norm": 2.1057093143463135,
      "learning_rate": 1.5541377991545868e-08,
      "loss": 1.175,
      "step": 37210
    },
    {
      "epoch": 9.94655264564404,
      "grad_norm": 2.1288132667541504,
      "learning_rate": 1.409652192173594e-08,
      "loss": 1.187,
      "step": 37220
    },
    {
      "epoch": 9.949225013361838,
      "grad_norm": 2.1421315670013428,
      "learning_rate": 1.2722140177567899e-08,
      "loss": 1.1366,
      "step": 37230
    },
    {
      "epoch": 9.951897381079636,
      "grad_norm": 2.2259390354156494,
      "learning_rate": 1.1418233727744642e-08,
      "loss": 1.1012,
      "step": 37240
    },
    {
      "epoch": 9.954569748797434,
      "grad_norm": 2.059415578842163,
      "learning_rate": 1.0184803491342099e-08,
      "loss": 1.1875,
      "step": 37250
    },
    {
      "epoch": 9.957242116515232,
      "grad_norm": 2.463301420211792,
      "learning_rate": 9.021850337709303e-09,
      "loss": 1.0934,
      "step": 37260
    },
    {
      "epoch": 9.95991448423303,
      "grad_norm": 2.4153430461883545,
      "learning_rate": 7.929375086557223e-09,
      "loss": 1.0634,
      "step": 37270
    },
    {
      "epoch": 9.962586851950828,
      "grad_norm": 2.280428886413574,
      "learning_rate": 6.907378507903239e-09,
      "loss": 1.1093,
      "step": 37280
    },
    {
      "epoch": 9.965259219668626,
      "grad_norm": 2.283224582672119,
      "learning_rate": 5.9558613221044615e-09,
      "loss": 1.1276,
      "step": 37290
    },
    {
      "epoch": 9.967931587386424,
      "grad_norm": 2.3668198585510254,
      "learning_rate": 5.074824199824413e-09,
      "loss": 1.1984,
      "step": 37300
    },
    {
      "epoch": 9.970603955104222,
      "grad_norm": 2.254871368408203,
      "learning_rate": 4.264267762044139e-09,
      "loss": 1.1044,
      "step": 37310
    },
    {
      "epoch": 9.97327632282202,
      "grad_norm": 2.074324369430542,
      "learning_rate": 3.524192580106611e-09,
      "loss": 1.145,
      "step": 37320
    },
    {
      "epoch": 9.975948690539818,
      "grad_norm": 2.271371603012085,
      "learning_rate": 2.8545991756168122e-09,
      "loss": 1.0873,
      "step": 37330
    },
    {
      "epoch": 9.978621058257616,
      "grad_norm": 2.232445001602173,
      "learning_rate": 2.255488020552754e-09,
      "loss": 1.1409,
      "step": 37340
    },
    {
      "epoch": 9.981293425975414,
      "grad_norm": 2.2682485580444336,
      "learning_rate": 1.726859537198866e-09,
      "loss": 1.1406,
      "step": 37350
    },
    {
      "epoch": 9.983965793693212,
      "grad_norm": 2.0237770080566406,
      "learning_rate": 1.2687140981348933e-09,
      "loss": 1.1185,
      "step": 37360
    },
    {
      "epoch": 9.98663816141101,
      "grad_norm": 2.0874645709991455,
      "learning_rate": 8.810520262914068e-10,
      "loss": 1.1659,
      "step": 37370
    },
    {
      "epoch": 9.989310529128808,
      "grad_norm": 2.3018386363983154,
      "learning_rate": 5.638735949053953e-10,
      "loss": 1.1434,
      "step": 37380
    },
    {
      "epoch": 9.991982896846606,
      "grad_norm": 2.2468602657318115,
      "learning_rate": 3.171790275424691e-10,
      "loss": 1.1011,
      "step": 37390
    },
    {
      "epoch": 9.994655264564404,
      "grad_norm": 2.2814855575561523,
      "learning_rate": 1.409684980857584e-10,
      "loss": 1.1761,
      "step": 37400
    },
    {
      "epoch": 9.997327632282202,
      "grad_norm": 2.0475192070007324,
      "learning_rate": 3.524213073591298e-11,
      "loss": 1.1567,
      "step": 37410
    },
    {
      "epoch": 10.0,
      "grad_norm": 2.189579963684082,
      "learning_rate": 0.0,
      "loss": 1.1511,
      "step": 37420
    },
    {
      "epoch": 10.0,
      "step": 37420,
      "total_flos": 1.840643671200891e+18,
      "train_loss": 1.5030417529482818,
      "train_runtime": 53386.1968,
      "train_samples_per_second": 2.804,
      "train_steps_per_second": 0.701
    }
  ],
  "logging_steps": 10,
  "max_steps": 37420,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 1.840643671200891e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}