{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 9.999379998759997,
  "eval_steps": 500,
  "global_step": 40320,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00248000496000992,
      "grad_norm": 0.7131296396255493,
      "learning_rate": 0.00019999996964510877,
      "loss": 2.4732,
      "step": 10
    },
    {
      "epoch": 0.00496000992001984,
      "grad_norm": 0.6677972078323364,
      "learning_rate": 0.00019999987858045355,
      "loss": 2.2955,
      "step": 20
    },
    {
      "epoch": 0.00744001488002976,
      "grad_norm": 0.5878852605819702,
      "learning_rate": 0.0001999997268060896,
      "loss": 2.2142,
      "step": 30
    },
    {
      "epoch": 0.00992001984003968,
      "grad_norm": 0.5556073784828186,
      "learning_rate": 0.00019999951432210904,
      "loss": 2.1472,
      "step": 40
    },
    {
      "epoch": 0.0124000248000496,
      "grad_norm": 0.5539277195930481,
      "learning_rate": 0.00019999924112864092,
      "loss": 2.1705,
      "step": 50
    },
    {
      "epoch": 0.01488002976005952,
      "grad_norm": 0.5625210404396057,
      "learning_rate": 0.00019999890722585106,
      "loss": 2.1509,
      "step": 60
    },
    {
      "epoch": 0.01736003472006944,
      "grad_norm": 0.5463761687278748,
      "learning_rate": 0.00019999851261394218,
      "loss": 2.1414,
      "step": 70
    },
    {
      "epoch": 0.01984003968007936,
      "grad_norm": 0.515055775642395,
      "learning_rate": 0.00019999805729315381,
      "loss": 2.1695,
      "step": 80
    },
    {
      "epoch": 0.02232004464008928,
      "grad_norm": 0.5034283995628357,
      "learning_rate": 0.00019999754126376248,
      "loss": 2.13,
      "step": 90
    },
    {
      "epoch": 0.0248000496000992,
      "grad_norm": 0.6123887896537781,
      "learning_rate": 0.00019999696452608135,
      "loss": 2.1778,
      "step": 100
    },
    {
      "epoch": 0.02728005456010912,
      "grad_norm": 0.5200942158699036,
      "learning_rate": 0.00019999632708046064,
      "loss": 2.141,
      "step": 110
    },
    {
      "epoch": 0.02976005952011904,
      "grad_norm": 0.5305571556091309,
      "learning_rate": 0.0001999956289272873,
      "loss": 2.1276,
      "step": 120
    },
    {
      "epoch": 0.03224006448012896,
      "grad_norm": 0.5080032348632812,
      "learning_rate": 0.0001999948700669852,
      "loss": 2.1138,
      "step": 130
    },
    {
      "epoch": 0.03472006944013888,
      "grad_norm": 0.5774938464164734,
      "learning_rate": 0.00019999405050001497,
      "loss": 2.155,
      "step": 140
    },
    {
      "epoch": 0.0372000744001488,
      "grad_norm": 0.5072618722915649,
      "learning_rate": 0.0001999931702268743,
      "loss": 2.1443,
      "step": 150
    },
    {
      "epoch": 0.03968007936015872,
      "grad_norm": 0.4849551320075989,
      "learning_rate": 0.0001999922292480975,
      "loss": 2.1275,
      "step": 160
    },
    {
      "epoch": 0.04216008432016864,
      "grad_norm": 0.5276527404785156,
      "learning_rate": 0.00019999122756425585,
      "loss": 2.1001,
      "step": 170
    },
    {
      "epoch": 0.04464008928017856,
      "grad_norm": 0.46588486433029175,
      "learning_rate": 0.00019999016517595753,
      "loss": 2.0843,
      "step": 180
    },
    {
      "epoch": 0.04712009424018848,
      "grad_norm": 0.528197169303894,
      "learning_rate": 0.00019998904208384742,
      "loss": 2.0699,
      "step": 190
    },
    {
      "epoch": 0.0496000992001984,
      "grad_norm": 0.5729973912239075,
      "learning_rate": 0.00019998785828860742,
      "loss": 2.0668,
      "step": 200
    },
    {
      "epoch": 0.05208010416020832,
      "grad_norm": 0.5057010650634766,
      "learning_rate": 0.0001999866137909562,
      "loss": 2.08,
      "step": 210
    },
    {
      "epoch": 0.05456010912021824,
      "grad_norm": 0.5133991837501526,
      "learning_rate": 0.00019998530859164927,
      "loss": 2.0329,
      "step": 220
    },
    {
      "epoch": 0.05704011408022816,
      "grad_norm": 0.5246201157569885,
      "learning_rate": 0.000199983942691479,
      "loss": 2.0358,
      "step": 230
    },
    {
      "epoch": 0.05952011904023808,
      "grad_norm": 0.556400716304779,
      "learning_rate": 0.00019998251609127464,
      "loss": 2.0534,
      "step": 240
    },
    {
      "epoch": 0.062000124000248,
      "grad_norm": 0.5591869950294495,
      "learning_rate": 0.0001999810287919023,
      "loss": 2.0162,
      "step": 250
    },
    {
      "epoch": 0.06448012896025793,
      "grad_norm": 0.5347974896430969,
      "learning_rate": 0.00019997948079426492,
      "loss": 2.0735,
      "step": 260
    },
    {
      "epoch": 0.06696013392026784,
      "grad_norm": 0.5143352150917053,
      "learning_rate": 0.00019997787209930222,
      "loss": 2.0251,
      "step": 270
    },
    {
      "epoch": 0.06944013888027777,
      "grad_norm": 0.5554919838905334,
      "learning_rate": 0.00019997620270799092,
      "loss": 2.1207,
      "step": 280
    },
    {
      "epoch": 0.07192014384028768,
      "grad_norm": 0.5315896272659302,
      "learning_rate": 0.00019997447262134444,
      "loss": 2.0936,
      "step": 290
    },
    {
      "epoch": 0.0744001488002976,
      "grad_norm": 0.5028589367866516,
      "learning_rate": 0.00019997268184041315,
      "loss": 2.048,
      "step": 300
    },
    {
      "epoch": 0.07688015376030752,
      "grad_norm": 0.4932255148887634,
      "learning_rate": 0.00019997083036628422,
      "loss": 2.0523,
      "step": 310
    },
    {
      "epoch": 0.07936015872031744,
      "grad_norm": 0.5493288040161133,
      "learning_rate": 0.00019996891820008164,
      "loss": 2.0368,
      "step": 320
    },
    {
      "epoch": 0.08184016368032736,
      "grad_norm": 0.51752769947052,
      "learning_rate": 0.0001999669453429663,
      "loss": 2.0567,
      "step": 330
    },
    {
      "epoch": 0.08432016864033728,
      "grad_norm": 0.5354537963867188,
      "learning_rate": 0.00019996491179613597,
      "loss": 2.0354,
      "step": 340
    },
    {
      "epoch": 0.0868001736003472,
      "grad_norm": 0.536638081073761,
      "learning_rate": 0.00019996281756082517,
      "loss": 2.0324,
      "step": 350
    },
    {
      "epoch": 0.08928017856035712,
      "grad_norm": 0.5761566758155823,
      "learning_rate": 0.00019996066263830531,
      "loss": 2.1082,
      "step": 360
    },
    {
      "epoch": 0.09176018352036704,
      "grad_norm": 0.553530216217041,
      "learning_rate": 0.00019995844702988463,
      "loss": 2.0161,
      "step": 370
    },
    {
      "epoch": 0.09424018848037696,
      "grad_norm": 0.5311240553855896,
      "learning_rate": 0.0001999561707369082,
      "loss": 2.0226,
      "step": 380
    },
    {
      "epoch": 0.09672019344038688,
      "grad_norm": 0.546971321105957,
      "learning_rate": 0.000199953833760758,
      "loss": 2.0872,
      "step": 390
    },
    {
      "epoch": 0.0992001984003968,
      "grad_norm": 0.5235766172409058,
      "learning_rate": 0.00019995143610285277,
      "loss": 2.0279,
      "step": 400
    },
    {
      "epoch": 0.10168020336040672,
      "grad_norm": 0.5465070009231567,
      "learning_rate": 0.00019994897776464817,
      "loss": 2.0212,
      "step": 410
    },
    {
      "epoch": 0.10416020832041664,
      "grad_norm": 0.5408602356910706,
      "learning_rate": 0.00019994645874763658,
      "loss": 2.0009,
      "step": 420
    },
    {
      "epoch": 0.10664021328042655,
      "grad_norm": 0.5485517382621765,
      "learning_rate": 0.00019994387905334735,
      "loss": 2.0427,
      "step": 430
    },
    {
      "epoch": 0.10912021824043648,
      "grad_norm": 0.5387979745864868,
      "learning_rate": 0.00019994123868334655,
      "loss": 2.0833,
      "step": 440
    },
    {
      "epoch": 0.1116002232004464,
      "grad_norm": 0.5328435301780701,
      "learning_rate": 0.00019993853763923723,
      "loss": 2.0441,
      "step": 450
    },
    {
      "epoch": 0.11408022816045632,
      "grad_norm": 0.5278816819190979,
      "learning_rate": 0.0001999357759226591,
      "loss": 2.0305,
      "step": 460
    },
    {
      "epoch": 0.11656023312046625,
      "grad_norm": 0.5591515898704529,
      "learning_rate": 0.00019993295353528884,
      "loss": 2.0095,
      "step": 470
    },
    {
      "epoch": 0.11904023808047616,
      "grad_norm": 0.536952555179596,
      "learning_rate": 0.00019993007047883988,
      "loss": 2.0396,
      "step": 480
    },
    {
      "epoch": 0.12152024304048609,
      "grad_norm": 0.5500532984733582,
      "learning_rate": 0.00019992712675506253,
      "loss": 2.1116,
      "step": 490
    },
    {
      "epoch": 0.124000248000496,
      "grad_norm": 0.5213359594345093,
      "learning_rate": 0.00019992412236574395,
      "loss": 2.03,
      "step": 500
    },
    {
      "epoch": 0.1264802529605059,
      "grad_norm": 0.5710284113883972,
      "learning_rate": 0.00019992105731270806,
      "loss": 2.0542,
      "step": 510
    },
    {
      "epoch": 0.12896025792051585,
      "grad_norm": 0.5325077772140503,
      "learning_rate": 0.00019991793159781569,
      "loss": 2.0765,
      "step": 520
    },
    {
      "epoch": 0.13144026288052577,
      "grad_norm": 0.5777318477630615,
      "learning_rate": 0.00019991474522296443,
      "loss": 1.9969,
      "step": 530
    },
    {
      "epoch": 0.13392026784053568,
      "grad_norm": 0.5040794610977173,
      "learning_rate": 0.0001999114981900887,
      "loss": 1.9853,
      "step": 540
    },
    {
      "epoch": 0.1364002728005456,
      "grad_norm": 0.5974857807159424,
      "learning_rate": 0.0001999081905011598,
      "loss": 2.0857,
      "step": 550
    },
    {
      "epoch": 0.13888027776055553,
      "grad_norm": 0.5376875996589661,
      "learning_rate": 0.0001999048221581858,
      "loss": 2.0042,
      "step": 560
    },
    {
      "epoch": 0.14136028272056544,
      "grad_norm": 0.5114587545394897,
      "learning_rate": 0.00019990139316321162,
      "loss": 2.0838,
      "step": 570
    },
    {
      "epoch": 0.14384028768057536,
      "grad_norm": 0.554987907409668,
      "learning_rate": 0.00019989790351831896,
      "loss": 2.0281,
      "step": 580
    },
    {
      "epoch": 0.14632029264058527,
      "grad_norm": 0.5457165837287903,
      "learning_rate": 0.00019989435322562645,
      "loss": 2.0423,
      "step": 590
    },
    {
      "epoch": 0.1488002976005952,
      "grad_norm": 0.5625684857368469,
      "learning_rate": 0.0001998907422872894,
      "loss": 2.0691,
      "step": 600
    },
    {
      "epoch": 0.15128030256060512,
      "grad_norm": 0.5338137149810791,
      "learning_rate": 0.00019988707070550007,
      "loss": 2.0549,
      "step": 610
    },
    {
      "epoch": 0.15376030752061504,
      "grad_norm": 0.5166420936584473,
      "learning_rate": 0.0001998833384824874,
      "loss": 1.9901,
      "step": 620
    },
    {
      "epoch": 0.15624031248062495,
      "grad_norm": 0.5435931086540222,
      "learning_rate": 0.00019987954562051725,
      "loss": 2.0217,
      "step": 630
    },
    {
      "epoch": 0.1587203174406349,
      "grad_norm": 0.5430518984794617,
      "learning_rate": 0.00019987569212189224,
      "loss": 1.9782,
      "step": 640
    },
    {
      "epoch": 0.1612003224006448,
      "grad_norm": 0.5538361668586731,
      "learning_rate": 0.00019987177798895184,
      "loss": 2.0747,
      "step": 650
    },
    {
      "epoch": 0.16368032736065471,
      "grad_norm": 0.5739989280700684,
      "learning_rate": 0.0001998678032240723,
      "loss": 1.9879,
      "step": 660
    },
    {
      "epoch": 0.16616033232066463,
      "grad_norm": 0.581364631652832,
      "learning_rate": 0.0001998637678296667,
      "loss": 2.0,
      "step": 670
    },
    {
      "epoch": 0.16864033728067457,
      "grad_norm": 0.5130109786987305,
      "learning_rate": 0.0001998596718081849,
      "loss": 2.082,
      "step": 680
    },
    {
      "epoch": 0.17112034224068448,
      "grad_norm": 0.5570313334465027,
      "learning_rate": 0.00019985551516211363,
      "loss": 2.0369,
      "step": 690
    },
    {
      "epoch": 0.1736003472006944,
      "grad_norm": 0.5379239916801453,
      "learning_rate": 0.00019985129789397633,
      "loss": 2.0345,
      "step": 700
    },
    {
      "epoch": 0.17608035216070433,
      "grad_norm": 0.5915724635124207,
      "learning_rate": 0.0001998470200063333,
      "loss": 1.9996,
      "step": 710
    },
    {
      "epoch": 0.17856035712071425,
      "grad_norm": 0.574665904045105,
      "learning_rate": 0.00019984268150178167,
      "loss": 1.9805,
      "step": 720
    },
    {
      "epoch": 0.18104036208072416,
      "grad_norm": 0.5390008687973022,
      "learning_rate": 0.0001998382823829553,
      "loss": 1.9605,
      "step": 730
    },
    {
      "epoch": 0.18352036704073407,
      "grad_norm": 0.5619198083877563,
      "learning_rate": 0.00019983382265252491,
      "loss": 2.0536,
      "step": 740
    },
    {
      "epoch": 0.186000372000744,
      "grad_norm": 0.5629109144210815,
      "learning_rate": 0.000199829302313198,
      "loss": 1.9862,
      "step": 750
    },
    {
      "epoch": 0.18848037696075393,
      "grad_norm": 0.5493379235267639,
      "learning_rate": 0.00019982472136771882,
      "loss": 1.9932,
      "step": 760
    },
    {
      "epoch": 0.19096038192076384,
      "grad_norm": 0.5466563701629639,
      "learning_rate": 0.00019982007981886847,
      "loss": 2.0434,
      "step": 770
    },
    {
      "epoch": 0.19344038688077375,
      "grad_norm": 0.5508472919464111,
      "learning_rate": 0.00019981537766946484,
      "loss": 2.0423,
      "step": 780
    },
    {
      "epoch": 0.1959203918407837,
      "grad_norm": 0.5475140810012817,
      "learning_rate": 0.00019981061492236262,
      "loss": 1.9899,
      "step": 790
    },
    {
      "epoch": 0.1984003968007936,
      "grad_norm": 0.5537005066871643,
      "learning_rate": 0.0001998057915804532,
      "loss": 1.9943,
      "step": 800
    },
    {
      "epoch": 0.20088040176080352,
      "grad_norm": 0.5661734342575073,
      "learning_rate": 0.00019980090764666484,
      "loss": 2.0474,
      "step": 810
    },
    {
      "epoch": 0.20336040672081343,
      "grad_norm": 0.5358636975288391,
      "learning_rate": 0.0001997959631239626,
      "loss": 1.9881,
      "step": 820
    },
    {
      "epoch": 0.20584041168082337,
      "grad_norm": 0.5859567523002625,
      "learning_rate": 0.00019979095801534821,
      "loss": 2.0211,
      "step": 830
    },
    {
      "epoch": 0.20832041664083328,
      "grad_norm": 0.5483726263046265,
      "learning_rate": 0.00019978589232386035,
      "loss": 2.0127,
      "step": 840
    },
    {
      "epoch": 0.2108004216008432,
      "grad_norm": 0.5464625954627991,
      "learning_rate": 0.00019978076605257436,
      "loss": 2.0091,
      "step": 850
    },
    {
      "epoch": 0.2132804265608531,
      "grad_norm": 0.5446053147315979,
      "learning_rate": 0.00019977557920460237,
      "loss": 1.9911,
      "step": 860
    },
    {
      "epoch": 0.21576043152086305,
      "grad_norm": 0.5547277331352234,
      "learning_rate": 0.00019977033178309332,
      "loss": 2.0402,
      "step": 870
    },
    {
      "epoch": 0.21824043648087296,
      "grad_norm": 0.5449572801589966,
      "learning_rate": 0.0001997650237912329,
      "loss": 2.0366,
      "step": 880
    },
    {
      "epoch": 0.22072044144088288,
      "grad_norm": 0.5232194066047668,
      "learning_rate": 0.00019975965523224358,
      "loss": 1.9832,
      "step": 890
    },
    {
      "epoch": 0.2232004464008928,
      "grad_norm": 0.5563146471977234,
      "learning_rate": 0.00019975422610938462,
      "loss": 1.9306,
      "step": 900
    },
    {
      "epoch": 0.22568045136090273,
      "grad_norm": 0.5598681569099426,
      "learning_rate": 0.000199748736425952,
      "loss": 1.8938,
      "step": 910
    },
    {
      "epoch": 0.22816045632091264,
      "grad_norm": 0.5681236386299133,
      "learning_rate": 0.00019974318618527849,
      "loss": 1.9727,
      "step": 920
    },
    {
      "epoch": 0.23064046128092255,
      "grad_norm": 0.5816565155982971,
      "learning_rate": 0.00019973757539073366,
      "loss": 1.9525,
      "step": 930
    },
    {
      "epoch": 0.2331204662409325,
      "grad_norm": 0.5877417922019958,
      "learning_rate": 0.0001997319040457238,
      "loss": 2.0935,
      "step": 940
    },
    {
      "epoch": 0.2356004712009424,
      "grad_norm": 0.5637423992156982,
      "learning_rate": 0.000199726172153692,
      "loss": 2.0272,
      "step": 950
    },
    {
      "epoch": 0.23808047616095232,
      "grad_norm": 0.6446289420127869,
      "learning_rate": 0.00019972037971811802,
      "loss": 2.0328,
      "step": 960
    },
    {
      "epoch": 0.24056048112096223,
      "grad_norm": 0.5652419328689575,
      "learning_rate": 0.00019971452674251848,
      "loss": 2.014,
      "step": 970
    },
    {
      "epoch": 0.24304048608097217,
      "grad_norm": 0.5587862730026245,
      "learning_rate": 0.00019970861323044667,
      "loss": 2.0134,
      "step": 980
    },
    {
      "epoch": 0.2455204910409821,
      "grad_norm": 0.5309275984764099,
      "learning_rate": 0.00019970263918549274,
      "loss": 2.0183,
      "step": 990
    },
    {
      "epoch": 0.248000496000992,
      "grad_norm": 0.5930149555206299,
      "learning_rate": 0.00019969660461128342,
      "loss": 1.9548,
      "step": 1000
    },
    {
      "epoch": 0.25048050096100194,
      "grad_norm": 0.59003746509552,
      "learning_rate": 0.00019969050951148233,
      "loss": 1.9657,
      "step": 1010
    },
    {
      "epoch": 0.2529605059210118,
      "grad_norm": 0.5892198085784912,
      "learning_rate": 0.00019968435388978984,
      "loss": 1.9756,
      "step": 1020
    },
    {
      "epoch": 0.25544051088102177,
      "grad_norm": 0.57949298620224,
      "learning_rate": 0.00019967813774994294,
      "loss": 1.9637,
      "step": 1030
    },
    {
      "epoch": 0.2579205158410317,
      "grad_norm": 0.556459367275238,
      "learning_rate": 0.00019967186109571552,
      "loss": 1.991,
      "step": 1040
    },
    {
      "epoch": 0.2604005208010416,
      "grad_norm": 0.5568014979362488,
      "learning_rate": 0.00019966552393091804,
      "loss": 1.9393,
      "step": 1050
    },
    {
      "epoch": 0.26288052576105153,
      "grad_norm": 0.5933910012245178,
      "learning_rate": 0.00019965912625939783,
      "loss": 1.9554,
      "step": 1060
    },
    {
      "epoch": 0.2653605307210614,
      "grad_norm": 0.5912190079689026,
      "learning_rate": 0.00019965266808503887,
      "loss": 1.9702,
      "step": 1070
    },
    {
      "epoch": 0.26784053568107136,
      "grad_norm": 0.5662450194358826,
      "learning_rate": 0.00019964614941176195,
      "loss": 2.0206,
      "step": 1080
    },
    {
      "epoch": 0.2703205406410813,
      "grad_norm": 0.5403727889060974,
      "learning_rate": 0.00019963957024352446,
      "loss": 2.0103,
      "step": 1090
    },
    {
      "epoch": 0.2728005456010912,
      "grad_norm": 0.5555917024612427,
      "learning_rate": 0.00019963293058432066,
      "loss": 1.875,
      "step": 1100
    },
    {
      "epoch": 0.2752805505611011,
      "grad_norm": 0.5734595060348511,
      "learning_rate": 0.0001996262304381815,
      "loss": 2.0566,
      "step": 1110
    },
    {
      "epoch": 0.27776055552111106,
      "grad_norm": 0.5790075659751892,
      "learning_rate": 0.00019961946980917456,
      "loss": 2.0087,
      "step": 1120
    },
    {
      "epoch": 0.28024056048112095,
      "grad_norm": 0.5560197234153748,
      "learning_rate": 0.00019961264870140422,
      "loss": 1.967,
      "step": 1130
    },
    {
      "epoch": 0.2827205654411309,
      "grad_norm": 0.5903038382530212,
      "learning_rate": 0.0001996057671190116,
      "loss": 1.9977,
      "step": 1140
    },
    {
      "epoch": 0.28520057040114083,
      "grad_norm": 0.5051490068435669,
      "learning_rate": 0.00019959882506617445,
      "loss": 1.9385,
      "step": 1150
    },
    {
      "epoch": 0.2876805753611507,
      "grad_norm": 0.5680397152900696,
      "learning_rate": 0.0001995918225471073,
      "loss": 1.9929,
      "step": 1160
    },
    {
      "epoch": 0.29016058032116065,
      "grad_norm": 0.5738064050674438,
      "learning_rate": 0.00019958475956606133,
      "loss": 2.0118,
      "step": 1170
    },
    {
      "epoch": 0.29264058528117054,
      "grad_norm": 0.5676369667053223,
      "learning_rate": 0.0001995776361273245,
      "loss": 1.9608,
      "step": 1180
    },
    {
      "epoch": 0.2951205902411805,
      "grad_norm": 0.5738691687583923,
      "learning_rate": 0.0001995704522352214,
      "loss": 1.9918,
      "step": 1190
    },
    {
      "epoch": 0.2976005952011904,
      "grad_norm": 0.5853675007820129,
      "learning_rate": 0.00019956320789411337,
      "loss": 1.9739,
      "step": 1200
    },
    {
      "epoch": 0.3000806001612003,
      "grad_norm": 0.5621643662452698,
      "learning_rate": 0.00019955590310839847,
      "loss": 1.9861,
      "step": 1210
    },
    {
      "epoch": 0.30256060512121025,
      "grad_norm": 0.5738389492034912,
      "learning_rate": 0.00019954853788251137,
      "loss": 1.9618,
      "step": 1220
    },
    {
      "epoch": 0.3050406100812202,
      "grad_norm": 0.5730957984924316,
      "learning_rate": 0.0001995411122209235,
      "loss": 1.9789,
      "step": 1230
    },
    {
      "epoch": 0.30752061504123007,
      "grad_norm": 0.5502608418464661,
      "learning_rate": 0.00019953362612814296,
      "loss": 1.9983,
      "step": 1240
    },
    {
      "epoch": 0.31000062000124,
      "grad_norm": 0.6123182773590088,
      "learning_rate": 0.00019952607960871453,
      "loss": 2.025,
      "step": 1250
    },
    {
      "epoch": 0.3124806249612499,
      "grad_norm": 0.562450647354126,
      "learning_rate": 0.0001995184726672197,
      "loss": 1.9941,
      "step": 1260
    },
    {
      "epoch": 0.31496062992125984,
      "grad_norm": 0.5684399008750916,
      "learning_rate": 0.00019951080530827663,
      "loss": 1.9974,
      "step": 1270
    },
    {
      "epoch": 0.3174406348812698,
      "grad_norm": 0.5505920052528381,
      "learning_rate": 0.00019950307753654017,
      "loss": 2.0099,
      "step": 1280
    },
    {
      "epoch": 0.31992063984127966,
      "grad_norm": 0.5620154738426208,
      "learning_rate": 0.0001994952893567018,
      "loss": 1.97,
      "step": 1290
    },
    {
      "epoch": 0.3224006448012896,
      "grad_norm": 0.5569179058074951,
      "learning_rate": 0.00019948744077348968,
      "loss": 1.9819,
      "step": 1300
    },
    {
      "epoch": 0.32488064976129954,
      "grad_norm": 0.6155531406402588,
      "learning_rate": 0.00019947953179166877,
      "loss": 1.9119,
      "step": 1310
    },
    {
      "epoch": 0.32736065472130943,
      "grad_norm": 0.5746706128120422,
      "learning_rate": 0.0001994715624160405,
      "loss": 1.925,
      "step": 1320
    },
    {
      "epoch": 0.32984065968131937,
      "grad_norm": 0.5678473711013794,
      "learning_rate": 0.00019946353265144312,
      "loss": 1.9679,
      "step": 1330
    },
    {
      "epoch": 0.33232066464132926,
      "grad_norm": 0.5691818594932556,
      "learning_rate": 0.00019945544250275145,
      "loss": 1.9778,
      "step": 1340
    },
    {
      "epoch": 0.3348006696013392,
      "grad_norm": 0.5543385744094849,
      "learning_rate": 0.000199447291974877,
      "loss": 1.9585,
      "step": 1350
    },
    {
      "epoch": 0.33728067456134914,
      "grad_norm": 0.5823257565498352,
      "learning_rate": 0.00019943908107276798,
      "loss": 1.9624,
      "step": 1360
    },
    {
      "epoch": 0.339760679521359,
      "grad_norm": 0.5965515971183777,
      "learning_rate": 0.00019943080980140918,
      "loss": 2.0268,
      "step": 1370
    },
    {
      "epoch": 0.34224068448136896,
      "grad_norm": 0.6067745685577393,
      "learning_rate": 0.00019942247816582206,
      "loss": 1.9783,
      "step": 1380
    },
    {
      "epoch": 0.3447206894413789,
      "grad_norm": 0.5427114963531494,
      "learning_rate": 0.00019941408617106472,
      "loss": 1.9887,
      "step": 1390
    },
    {
      "epoch": 0.3472006944013888,
      "grad_norm": 0.5312680602073669,
      "learning_rate": 0.00019940563382223197,
      "loss": 1.998,
      "step": 1400
    },
    {
      "epoch": 0.34968069936139873,
      "grad_norm": 0.6098049283027649,
      "learning_rate": 0.0001993971211244552,
      "loss": 1.9817,
      "step": 1410
    },
    {
      "epoch": 0.35216070432140867,
      "grad_norm": 0.5767588019371033,
      "learning_rate": 0.00019938854808290244,
      "loss": 1.951,
      "step": 1420
    },
    {
      "epoch": 0.35464070928141855,
      "grad_norm": 0.5756887197494507,
      "learning_rate": 0.00019937991470277835,
      "loss": 1.9462,
      "step": 1430
    },
    {
      "epoch": 0.3571207142414285,
      "grad_norm": 0.6078529357910156,
      "learning_rate": 0.00019937122098932428,
      "loss": 1.9716,
      "step": 1440
    },
    {
      "epoch": 0.3596007192014384,
      "grad_norm": 0.5578196048736572,
      "learning_rate": 0.0001993624669478181,
      "loss": 1.9097,
      "step": 1450
    },
    {
      "epoch": 0.3620807241614483,
      "grad_norm": 0.5579915642738342,
      "learning_rate": 0.00019935365258357446,
      "loss": 1.9615,
      "step": 1460
    },
    {
      "epoch": 0.36456072912145826,
      "grad_norm": 0.5643064975738525,
      "learning_rate": 0.00019934477790194445,
      "loss": 1.9567,
      "step": 1470
    },
    {
      "epoch": 0.36704073408146815,
      "grad_norm": 0.5760548114776611,
      "learning_rate": 0.0001993358429083159,
      "loss": 1.9476,
      "step": 1480
    },
    {
      "epoch": 0.3695207390414781,
      "grad_norm": 0.5931412577629089,
      "learning_rate": 0.00019932684760811328,
      "loss": 1.984,
      "step": 1490
    },
    {
      "epoch": 0.372000744001488,
      "grad_norm": 0.5657218098640442,
      "learning_rate": 0.00019931779200679754,
      "loss": 1.9508,
      "step": 1500
    },
    {
      "epoch": 0.3744807489614979,
      "grad_norm": 0.5403887629508972,
      "learning_rate": 0.00019930867610986637,
      "loss": 1.9676,
      "step": 1510
    },
    {
      "epoch": 0.37696075392150785,
      "grad_norm": 0.5199195742607117,
      "learning_rate": 0.00019929949992285396,
      "loss": 1.9848,
      "step": 1520
    },
    {
      "epoch": 0.37944075888151774,
      "grad_norm": 0.5714607834815979,
      "learning_rate": 0.00019929026345133122,
      "loss": 1.9478,
      "step": 1530
    },
    {
      "epoch": 0.3819207638415277,
      "grad_norm": 0.6201160550117493,
      "learning_rate": 0.0001992809667009055,
      "loss": 1.9482,
      "step": 1540
    },
    {
      "epoch": 0.3844007688015376,
      "grad_norm": 0.5621275305747986,
      "learning_rate": 0.00019927160967722092,
      "loss": 2.0238,
      "step": 1550
    },
    {
      "epoch": 0.3868807737615475,
      "grad_norm": 0.5677786469459534,
      "learning_rate": 0.0001992621923859581,
      "loss": 1.9221,
      "step": 1560
    },
    {
      "epoch": 0.38936077872155744,
      "grad_norm": 0.6040672063827515,
      "learning_rate": 0.0001992527148328342,
      "loss": 1.9526,
      "step": 1570
    },
    {
      "epoch": 0.3918407836815674,
      "grad_norm": 0.6181504130363464,
      "learning_rate": 0.0001992431770236031,
      "loss": 1.9546,
      "step": 1580
    },
    {
      "epoch": 0.39432078864157727,
      "grad_norm": 0.5984564423561096,
      "learning_rate": 0.0001992335789640551,
      "loss": 1.9511,
      "step": 1590
    },
    {
      "epoch": 0.3968007936015872,
      "grad_norm": 0.578197717666626,
      "learning_rate": 0.00019922392066001722,
      "loss": 1.9471,
      "step": 1600
    },
    {
      "epoch": 0.39928079856159715,
      "grad_norm": 0.5593079328536987,
      "learning_rate": 0.00019921420211735299,
      "loss": 1.9363,
      "step": 1610
    },
    {
      "epoch": 0.40176080352160704,
      "grad_norm": 0.5923002362251282,
      "learning_rate": 0.00019920442334196248,
      "loss": 2.0088,
      "step": 1620
    },
    {
      "epoch": 0.404240808481617,
      "grad_norm": 0.6107719540596008,
      "learning_rate": 0.0001991945843397824,
      "loss": 1.9541,
      "step": 1630
    },
    {
      "epoch": 0.40672081344162686,
      "grad_norm": 0.6352981925010681,
      "learning_rate": 0.00019918468511678596,
      "loss": 1.986,
      "step": 1640
    },
    {
      "epoch": 0.4092008184016368,
      "grad_norm": 0.5772244334220886,
      "learning_rate": 0.00019917472567898298,
      "loss": 1.9746,
      "step": 1650
    },
    {
      "epoch": 0.41168082336164674,
      "grad_norm": 0.6013234853744507,
      "learning_rate": 0.0001991647060324198,
      "loss": 1.9267,
      "step": 1660
    },
    {
      "epoch": 0.4141608283216566,
      "grad_norm": 0.5733570456504822,
      "learning_rate": 0.00019915462618317936,
      "loss": 1.9582,
      "step": 1670
    },
    {
      "epoch": 0.41664083328166657,
      "grad_norm": 0.6049069166183472,
      "learning_rate": 0.00019914448613738106,
      "loss": 1.9562,
      "step": 1680
    },
    {
      "epoch": 0.4191208382416765,
      "grad_norm": 0.5593000650405884,
      "learning_rate": 0.00019913428590118094,
      "loss": 1.9757,
      "step": 1690
    },
    {
      "epoch": 0.4216008432016864,
      "grad_norm": 0.5543436408042908,
      "learning_rate": 0.0001991240254807715,
      "loss": 1.9765,
      "step": 1700
    },
    {
      "epoch": 0.42408084816169633,
      "grad_norm": 0.5699513554573059,
      "learning_rate": 0.00019911370488238182,
      "loss": 1.9864,
      "step": 1710
    },
    {
      "epoch": 0.4265608531217062,
      "grad_norm": 0.6728619933128357,
      "learning_rate": 0.00019910332411227758,
      "loss": 1.9937,
      "step": 1720
    },
    {
      "epoch": 0.42904085808171616,
      "grad_norm": 0.5685863494873047,
      "learning_rate": 0.0001990928831767609,
      "loss": 1.941,
      "step": 1730
    },
    {
      "epoch": 0.4315208630417261,
      "grad_norm": 0.6041725873947144,
      "learning_rate": 0.00019908238208217038,
      "loss": 1.9517,
      "step": 1740
    },
    {
      "epoch": 0.434000868001736,
      "grad_norm": 0.570541501045227,
      "learning_rate": 0.00019907182083488129,
      "loss": 1.974,
      "step": 1750
    },
    {
      "epoch": 0.4364808729617459,
      "grad_norm": 0.5913785696029663,
      "learning_rate": 0.0001990611994413053,
      "loss": 1.9703,
      "step": 1760
    },
    {
      "epoch": 0.43896087792175587,
      "grad_norm": 0.5988500714302063,
      "learning_rate": 0.00019905051790789063,
      "loss": 1.9865,
      "step": 1770
    },
    {
      "epoch": 0.44144088288176575,
      "grad_norm": 0.67069411277771,
      "learning_rate": 0.00019903977624112206,
      "loss": 1.9606,
      "step": 1780
    },
    {
      "epoch": 0.4439208878417757,
      "grad_norm": 0.5693567991256714,
      "learning_rate": 0.00019902897444752077,
      "loss": 1.9973,
      "step": 1790
    },
    {
      "epoch": 0.4464008928017856,
      "grad_norm": 0.5602978467941284,
      "learning_rate": 0.00019901811253364456,
      "loss": 1.98,
      "step": 1800
    },
    {
      "epoch": 0.4488808977617955,
      "grad_norm": 0.5763017535209656,
      "learning_rate": 0.00019900719050608764,
      "loss": 1.9746,
      "step": 1810
    },
    {
      "epoch": 0.45136090272180546,
      "grad_norm": 0.6153312921524048,
      "learning_rate": 0.00019899620837148077,
      "loss": 1.9261,
      "step": 1820
    },
    {
      "epoch": 0.45384090768181534,
      "grad_norm": 0.5938084721565247,
      "learning_rate": 0.0001989851661364912,
      "loss": 1.9688,
      "step": 1830
    },
    {
      "epoch": 0.4563209126418253,
      "grad_norm": 0.9307805895805359,
      "learning_rate": 0.00019897406380782261,
      "loss": 1.963,
      "step": 1840
    },
    {
      "epoch": 0.4588009176018352,
      "grad_norm": 0.6255986094474792,
      "learning_rate": 0.0001989629013922152,
      "loss": 1.9374,
      "step": 1850
    },
    {
      "epoch": 0.4612809225618451,
      "grad_norm": 0.5881803631782532,
      "learning_rate": 0.00019895167889644565,
      "loss": 1.9721,
      "step": 1860
    },
    {
      "epoch": 0.46376092752185505,
      "grad_norm": 0.5993257761001587,
      "learning_rate": 0.00019894039632732712,
      "loss": 1.8728,
      "step": 1870
    },
    {
      "epoch": 0.466240932481865,
      "grad_norm": 0.6063452959060669,
      "learning_rate": 0.00019892905369170926,
      "loss": 1.929,
      "step": 1880
    },
    {
      "epoch": 0.4687209374418749,
      "grad_norm": 0.6029859185218811,
      "learning_rate": 0.0001989176509964781,
      "loss": 1.9544,
      "step": 1890
    },
    {
      "epoch": 0.4712009424018848,
      "grad_norm": 0.5588030815124512,
      "learning_rate": 0.00019890618824855625,
      "loss": 1.9094,
      "step": 1900
    },
    {
      "epoch": 0.4736809473618947,
      "grad_norm": 0.617078959941864,
      "learning_rate": 0.00019889466545490269,
      "loss": 1.9294,
      "step": 1910
    },
    {
      "epoch": 0.47616095232190464,
      "grad_norm": 0.6039510369300842,
      "learning_rate": 0.00019888308262251285,
      "loss": 2.0049,
      "step": 1920
    },
    {
      "epoch": 0.4786409572819146,
      "grad_norm": 0.5853057503700256,
      "learning_rate": 0.00019887143975841874,
      "loss": 1.9758,
      "step": 1930
    },
    {
      "epoch": 0.48112096224192447,
      "grad_norm": 0.5796060562133789,
      "learning_rate": 0.0001988597368696886,
      "loss": 1.9671,
      "step": 1940
    },
    {
      "epoch": 0.4836009672019344,
      "grad_norm": 0.5971837043762207,
      "learning_rate": 0.0001988479739634273,
      "loss": 1.9011,
      "step": 1950
    },
    {
      "epoch": 0.48608097216194435,
      "grad_norm": 0.5895419716835022,
      "learning_rate": 0.00019883615104677608,
      "loss": 1.9518,
      "step": 1960
    },
    {
      "epoch": 0.48856097712195423,
      "grad_norm": 0.5794113874435425,
      "learning_rate": 0.00019882426812691258,
      "loss": 1.9927,
      "step": 1970
    },
    {
      "epoch": 0.4910409820819642,
      "grad_norm": 0.5955272316932678,
      "learning_rate": 0.00019881232521105089,
      "loss": 1.9457,
      "step": 1980
    },
    {
      "epoch": 0.49352098704197406,
      "grad_norm": 0.5644362568855286,
      "learning_rate": 0.00019880032230644155,
      "loss": 1.8852,
      "step": 1990
    },
    {
      "epoch": 0.496000992001984,
      "grad_norm": 0.5644434690475464,
      "learning_rate": 0.00019878825942037148,
      "loss": 1.9762,
      "step": 2000
    },
    {
      "epoch": 0.49848099696199394,
      "grad_norm": 0.5878923535346985,
      "learning_rate": 0.00019877613656016404,
      "loss": 1.8868,
      "step": 2010
    },
    {
      "epoch": 0.5009610019220039,
      "grad_norm": 0.5987402200698853,
      "learning_rate": 0.000198763953733179,
      "loss": 1.9813,
      "step": 2020
    },
    {
      "epoch": 0.5034410068820138,
      "grad_norm": 0.543759822845459,
      "learning_rate": 0.00019875171094681248,
      "loss": 1.9633,
      "step": 2030
    },
    {
      "epoch": 0.5059210118420236,
      "grad_norm": 0.6039268374443054,
      "learning_rate": 0.00019873940820849714,
      "loss": 1.9667,
      "step": 2040
    },
    {
      "epoch": 0.5084010168020336,
      "grad_norm": 0.5571461319923401,
      "learning_rate": 0.00019872704552570188,
      "loss": 1.9474,
      "step": 2050
    },
    {
      "epoch": 0.5108810217620435,
      "grad_norm": 0.5795441269874573,
      "learning_rate": 0.00019871462290593206,
      "loss": 1.9691,
      "step": 2060
    },
    {
      "epoch": 0.5133610267220534,
      "grad_norm": 0.5893127918243408,
      "learning_rate": 0.00019870214035672942,
      "loss": 1.9224,
      "step": 2070
    },
    {
      "epoch": 0.5158410316820634,
      "grad_norm": 0.6003038883209229,
      "learning_rate": 0.00019868959788567212,
      "loss": 1.9734,
      "step": 2080
    },
    {
      "epoch": 0.5183210366420733,
      "grad_norm": 0.6020908355712891,
      "learning_rate": 0.00019867699550037466,
      "loss": 1.963,
      "step": 2090
    },
    {
      "epoch": 0.5208010416020832,
      "grad_norm": 0.5741065144538879,
      "learning_rate": 0.0001986643332084879,
      "loss": 1.9665,
      "step": 2100
    },
    {
      "epoch": 0.5232810465620932,
      "grad_norm": 0.5555444955825806,
      "learning_rate": 0.0001986516110176991,
      "loss": 1.9648,
      "step": 2110
    },
    {
      "epoch": 0.5257610515221031,
      "grad_norm": 0.563761293888092,
      "learning_rate": 0.00019863882893573188,
      "loss": 1.9516,
      "step": 2120
    },
    {
      "epoch": 0.528241056482113,
      "grad_norm": 0.6273584961891174,
      "learning_rate": 0.00019862598697034623,
      "loss": 1.9514,
      "step": 2130
    },
    {
      "epoch": 0.5307210614421228,
      "grad_norm": 0.5726107954978943,
      "learning_rate": 0.00019861308512933844,
      "loss": 1.9276,
      "step": 2140
    },
    {
      "epoch": 0.5332010664021328,
      "grad_norm": 0.5795220732688904,
      "learning_rate": 0.00019860012342054123,
      "loss": 1.9434,
      "step": 2150
    },
    {
      "epoch": 0.5356810713621427,
      "grad_norm": 0.6007318496704102,
      "learning_rate": 0.0001985871018518236,
      "loss": 1.9402,
      "step": 2160
    },
    {
      "epoch": 0.5381610763221526,
      "grad_norm": 0.6281079053878784,
      "learning_rate": 0.0001985740204310909,
      "loss": 1.9485,
      "step": 2170
    },
    {
      "epoch": 0.5406410812821626,
      "grad_norm": 0.5630472898483276,
      "learning_rate": 0.00019856087916628488,
      "loss": 1.9213,
      "step": 2180
    },
    {
      "epoch": 0.5431210862421725,
      "grad_norm": 0.5595583915710449,
      "learning_rate": 0.00019854767806538353,
      "loss": 1.9632,
      "step": 2190
    },
    {
      "epoch": 0.5456010912021824,
      "grad_norm": 0.5649421215057373,
      "learning_rate": 0.0001985344171364012,
      "loss": 1.9448,
      "step": 2200
    },
    {
      "epoch": 0.5480810961621924,
      "grad_norm": 0.6145544052124023,
      "learning_rate": 0.00019852109638738866,
      "loss": 1.9589,
      "step": 2210
    },
    {
      "epoch": 0.5505611011222022,
      "grad_norm": 0.6031001806259155,
      "learning_rate": 0.0001985077158264328,
      "loss": 1.9201,
      "step": 2220
    },
    {
      "epoch": 0.5530411060822121,
      "grad_norm": 0.6062312722206116,
      "learning_rate": 0.000198494275461657,
      "loss": 1.9224,
      "step": 2230
    },
    {
      "epoch": 0.5555211110422221,
      "grad_norm": 0.5883815288543701,
      "learning_rate": 0.00019848077530122083,
      "loss": 1.9887,
      "step": 2240
    },
    {
      "epoch": 0.558001116002232,
      "grad_norm": 0.5546337962150574,
      "learning_rate": 0.00019846721535332021,
      "loss": 1.8561,
      "step": 2250
    },
    {
      "epoch": 0.5604811209622419,
      "grad_norm": 0.5851352214813232,
      "learning_rate": 0.0001984535956261874,
      "loss": 1.9077,
      "step": 2260
    },
    {
      "epoch": 0.5629611259222519,
      "grad_norm": 0.9459104537963867,
      "learning_rate": 0.00019843991612809088,
      "loss": 1.9316,
      "step": 2270
    },
    {
      "epoch": 0.5654411308822618,
      "grad_norm": 0.5813398957252502,
      "learning_rate": 0.00019842617686733545,
      "loss": 1.9449,
      "step": 2280
    },
    {
      "epoch": 0.5679211358422717,
      "grad_norm": 0.6058304905891418,
      "learning_rate": 0.00019841237785226214,
      "loss": 1.9019,
      "step": 2290
    },
    {
      "epoch": 0.5704011408022817,
      "grad_norm": 0.5609736442565918,
      "learning_rate": 0.00019839851909124839,
      "loss": 1.9281,
      "step": 2300
    },
    {
      "epoch": 0.5728811457622915,
      "grad_norm": 0.6068705916404724,
      "learning_rate": 0.00019838460059270775,
      "loss": 1.9794,
      "step": 2310
    },
    {
      "epoch": 0.5753611507223014,
      "grad_norm": 0.6005422472953796,
      "learning_rate": 0.00019837062236509014,
      "loss": 1.9014,
      "step": 2320
    },
    {
      "epoch": 0.5778411556823113,
      "grad_norm": 0.5585853457450867,
      "learning_rate": 0.0001983565844168817,
      "loss": 1.9306,
      "step": 2330
    },
    {
      "epoch": 0.5803211606423213,
      "grad_norm": 0.5718742609024048,
      "learning_rate": 0.00019834248675660486,
      "loss": 1.9956,
      "step": 2340
    },
    {
      "epoch": 0.5828011656023312,
      "grad_norm": 0.6195574998855591,
      "learning_rate": 0.00019832832939281825,
      "loss": 1.9504,
      "step": 2350
    },
    {
      "epoch": 0.5852811705623411,
      "grad_norm": 0.5761587023735046,
      "learning_rate": 0.00019831411233411678,
      "loss": 1.9483,
      "step": 2360
    },
    {
      "epoch": 0.5877611755223511,
      "grad_norm": 0.5685802698135376,
      "learning_rate": 0.00019829983558913164,
      "loss": 1.8909,
      "step": 2370
    },
    {
      "epoch": 0.590241180482361,
      "grad_norm": 0.5675533413887024,
      "learning_rate": 0.0001982854991665301,
      "loss": 1.9896,
      "step": 2380
    },
    {
      "epoch": 0.5927211854423708,
      "grad_norm": 0.6346814036369324,
      "learning_rate": 0.00019827110307501592,
      "loss": 1.962,
      "step": 2390
    },
    {
      "epoch": 0.5952011904023808,
      "grad_norm": 0.6148980855941772,
      "learning_rate": 0.00019825664732332884,
      "loss": 1.9433,
      "step": 2400
    },
    {
      "epoch": 0.5976811953623907,
      "grad_norm": 0.547666609287262,
      "learning_rate": 0.00019824213192024491,
      "loss": 1.9084,
      "step": 2410
    },
    {
      "epoch": 0.6001612003224006,
      "grad_norm": 0.5827407240867615,
      "learning_rate": 0.00019822755687457647,
      "loss": 1.942,
      "step": 2420
    },
    {
      "epoch": 0.6026412052824106,
      "grad_norm": 0.5866121649742126,
      "learning_rate": 0.00019821292219517192,
      "loss": 1.9391,
      "step": 2430
    },
    {
      "epoch": 0.6051212102424205,
      "grad_norm": 0.5981879830360413,
      "learning_rate": 0.00019819822789091598,
      "loss": 1.8883,
      "step": 2440
    },
    {
      "epoch": 0.6076012152024304,
      "grad_norm": 0.6277055144309998,
      "learning_rate": 0.00019818347397072955,
      "loss": 1.8806,
      "step": 2450
    },
    {
      "epoch": 0.6100812201624404,
      "grad_norm": 0.5904166102409363,
      "learning_rate": 0.00019816866044356966,
      "loss": 1.9563,
      "step": 2460
    },
    {
      "epoch": 0.6125612251224503,
      "grad_norm": 0.5788779258728027,
      "learning_rate": 0.0001981537873184296,
      "loss": 1.9807,
      "step": 2470
    },
    {
      "epoch": 0.6150412300824601,
      "grad_norm": 0.6119228601455688,
      "learning_rate": 0.00019813885460433879,
      "loss": 1.942,
      "step": 2480
    },
    {
      "epoch": 0.6175212350424701,
      "grad_norm": 0.5789150595664978,
      "learning_rate": 0.00019812386231036287,
      "loss": 1.9688,
      "step": 2490
    },
    {
      "epoch": 0.62000124000248,
      "grad_norm": 0.6484764218330383,
      "learning_rate": 0.0001981088104456036,
      "loss": 1.9767,
      "step": 2500
    },
    {
      "epoch": 0.6224812449624899,
      "grad_norm": 0.5884684920310974,
      "learning_rate": 0.00019809369901919896,
      "loss": 1.9175,
      "step": 2510
    },
    {
      "epoch": 0.6249612499224998,
      "grad_norm": 0.615318238735199,
      "learning_rate": 0.00019807852804032305,
      "loss": 1.9431,
      "step": 2520
    },
    {
      "epoch": 0.6274412548825098,
      "grad_norm": 0.606164813041687,
      "learning_rate": 0.00019806329751818614,
      "loss": 1.9414,
      "step": 2530
    },
    {
      "epoch": 0.6299212598425197,
      "grad_norm": 0.6334375143051147,
      "learning_rate": 0.0001980480074620347,
      "loss": 1.913,
      "step": 2540
    },
    {
      "epoch": 0.6324012648025296,
      "grad_norm": 0.6048720479011536,
      "learning_rate": 0.0001980326578811512,
      "loss": 1.9082,
      "step": 2550
    },
    {
      "epoch": 0.6348812697625396,
      "grad_norm": 0.5949190258979797,
      "learning_rate": 0.00019801724878485438,
      "loss": 1.8712,
      "step": 2560
    },
    {
      "epoch": 0.6373612747225494,
      "grad_norm": 0.6880390048027039,
      "learning_rate": 0.00019800178018249912,
      "loss": 2.0191,
      "step": 2570
    },
    {
      "epoch": 0.6398412796825593,
      "grad_norm": 0.6032581329345703,
      "learning_rate": 0.00019798625208347626,
      "loss": 1.9965,
      "step": 2580
    },
    {
      "epoch": 0.6423212846425693,
      "grad_norm": 0.5974915027618408,
      "learning_rate": 0.00019797066449721295,
      "loss": 1.9477,
      "step": 2590
    },
    {
      "epoch": 0.6448012896025792,
      "grad_norm": 0.5914649367332458,
      "learning_rate": 0.0001979550174331724,
      "loss": 1.8974,
      "step": 2600
    },
    {
      "epoch": 0.6472812945625891,
      "grad_norm": 0.6158638000488281,
      "learning_rate": 0.0001979393109008538,
      "loss": 1.9837,
      "step": 2610
    },
    {
      "epoch": 0.6497612995225991,
      "grad_norm": 0.6009888052940369,
      "learning_rate": 0.0001979235449097927,
      "loss": 1.936,
      "step": 2620
    },
    {
      "epoch": 0.652241304482609,
      "grad_norm": 0.6107520461082458,
      "learning_rate": 0.00019790771946956052,
      "loss": 1.9553,
      "step": 2630
    },
    {
      "epoch": 0.6547213094426189,
      "grad_norm": 0.6255472898483276,
      "learning_rate": 0.00019789183458976484,
      "loss": 1.934,
      "step": 2640
    },
    {
      "epoch": 0.6572013144026289,
      "grad_norm": 0.6186276078224182,
      "learning_rate": 0.00019787589028004936,
      "loss": 1.952,
      "step": 2650
    },
    {
      "epoch": 0.6596813193626387,
      "grad_norm": 0.5936594605445862,
      "learning_rate": 0.00019785988655009385,
      "loss": 1.9244,
      "step": 2660
    },
    {
      "epoch": 0.6621613243226486,
      "grad_norm": 0.6173788905143738,
      "learning_rate": 0.0001978438234096141,
      "loss": 1.9429,
      "step": 2670
    },
    {
      "epoch": 0.6646413292826585,
      "grad_norm": 0.5978009104728699,
      "learning_rate": 0.000197827700868362,
      "loss": 1.9723,
      "step": 2680
    },
    {
      "epoch": 0.6671213342426685,
      "grad_norm": 0.6397370100021362,
      "learning_rate": 0.00019781151893612556,
      "loss": 1.9479,
      "step": 2690
    },
    {
      "epoch": 0.6696013392026784,
      "grad_norm": 0.5891425013542175,
      "learning_rate": 0.00019779527762272877,
      "loss": 1.9741,
      "step": 2700
    },
    {
      "epoch": 0.6720813441626883,
      "grad_norm": 0.5772106051445007,
      "learning_rate": 0.0001977789769380317,
      "loss": 1.9108,
      "step": 2710
    },
    {
      "epoch": 0.6745613491226983,
      "grad_norm": 0.6096621155738831,
      "learning_rate": 0.00019776261689193048,
      "loss": 1.9042,
      "step": 2720
    },
    {
      "epoch": 0.6770413540827082,
      "grad_norm": 0.5689034461975098,
      "learning_rate": 0.0001977461974943572,
      "loss": 1.9529,
      "step": 2730
    },
    {
      "epoch": 0.679521359042718,
      "grad_norm": 0.7213597893714905,
      "learning_rate": 0.00019772971875528007,
      "loss": 1.9048,
      "step": 2740
    },
    {
      "epoch": 0.682001364002728,
      "grad_norm": 0.5970284938812256,
      "learning_rate": 0.00019771318068470333,
      "loss": 1.8852,
      "step": 2750
    },
    {
      "epoch": 0.6844813689627379,
      "grad_norm": 0.6357975006103516,
      "learning_rate": 0.00019769658329266718,
      "loss": 1.947,
      "step": 2760
    },
    {
      "epoch": 0.6869613739227478,
      "grad_norm": 0.615311324596405,
      "learning_rate": 0.00019767992658924785,
      "loss": 1.9314,
      "step": 2770
    },
    {
      "epoch": 0.6894413788827578,
      "grad_norm": 0.5649407505989075,
      "learning_rate": 0.0001976632105845576,
      "loss": 1.877,
      "step": 2780
    },
    {
      "epoch": 0.6919213838427677,
      "grad_norm": 0.597957968711853,
      "learning_rate": 0.0001976464352887447,
      "loss": 1.904,
      "step": 2790
    },
    {
      "epoch": 0.6944013888027776,
      "grad_norm": 0.6019652485847473,
      "learning_rate": 0.00019762960071199333,
      "loss": 1.9559,
      "step": 2800
    },
    {
      "epoch": 0.6968813937627876,
      "grad_norm": 0.5853824019432068,
      "learning_rate": 0.0001976127068645238,
      "loss": 1.8575,
      "step": 2810
    },
    {
      "epoch": 0.6993613987227975,
      "grad_norm": 0.6015722751617432,
      "learning_rate": 0.0001975957537565923,
      "loss": 1.9112,
      "step": 2820
    },
    {
      "epoch": 0.7018414036828073,
      "grad_norm": 0.5531386733055115,
      "learning_rate": 0.00019757874139849103,
      "loss": 1.8914,
      "step": 2830
    },
    {
      "epoch": 0.7043214086428173,
      "grad_norm": 0.6410971283912659,
      "learning_rate": 0.00019756166980054813,
      "loss": 1.9788,
      "step": 2840
    },
    {
      "epoch": 0.7068014136028272,
      "grad_norm": 0.6276296377182007,
      "learning_rate": 0.00019754453897312776,
      "loss": 1.928,
      "step": 2850
    },
    {
      "epoch": 0.7092814185628371,
      "grad_norm": 0.6744928956031799,
      "learning_rate": 0.00019752734892663,
      "loss": 1.9618,
      "step": 2860
    },
    {
      "epoch": 0.711761423522847,
      "grad_norm": 0.6264201998710632,
      "learning_rate": 0.00019751009967149087,
      "loss": 1.9456,
      "step": 2870
    },
    {
      "epoch": 0.714241428482857,
      "grad_norm": 0.6102432012557983,
      "learning_rate": 0.00019749279121818235,
      "loss": 1.9352,
      "step": 2880
    },
    {
      "epoch": 0.7167214334428669,
      "grad_norm": 0.6014752388000488,
      "learning_rate": 0.00019747542357721244,
      "loss": 1.9279,
      "step": 2890
    },
    {
      "epoch": 0.7192014384028768,
      "grad_norm": 0.6810415983200073,
      "learning_rate": 0.0001974579967591249,
      "loss": 1.9062,
      "step": 2900
    },
    {
      "epoch": 0.7216814433628868,
      "grad_norm": 0.6644706726074219,
      "learning_rate": 0.00019744051077449958,
      "loss": 1.9392,
      "step": 2910
    },
    {
      "epoch": 0.7241614483228966,
      "grad_norm": 0.6256020069122314,
      "learning_rate": 0.00019742296563395216,
      "loss": 2.0146,
      "step": 2920
    },
    {
      "epoch": 0.7266414532829065,
      "grad_norm": 0.6017519235610962,
      "learning_rate": 0.00019740536134813422,
      "loss": 1.9183,
      "step": 2930
    },
    {
      "epoch": 0.7291214582429165,
      "grad_norm": 0.6205018758773804,
      "learning_rate": 0.00019738769792773336,
      "loss": 1.951,
      "step": 2940
    },
    {
      "epoch": 0.7316014632029264,
      "grad_norm": 0.6017516851425171,
      "learning_rate": 0.00019736997538347296,
      "loss": 1.9254,
      "step": 2950
    },
    {
      "epoch": 0.7340814681629363,
      "grad_norm": 0.5861731767654419,
      "learning_rate": 0.00019735219372611233,
      "loss": 1.9687,
      "step": 2960
    },
    {
      "epoch": 0.7365614731229463,
      "grad_norm": 0.6167269349098206,
      "learning_rate": 0.0001973343529664467,
      "loss": 1.9662,
      "step": 2970
    },
    {
      "epoch": 0.7390414780829562,
      "grad_norm": 0.5838399529457092,
      "learning_rate": 0.00019731645311530716,
      "loss": 1.913,
      "step": 2980
    },
    {
      "epoch": 0.7415214830429661,
      "grad_norm": 0.6024439334869385,
      "learning_rate": 0.0001972984941835606,
      "loss": 1.8726,
      "step": 2990
    },
    {
      "epoch": 0.744001488002976,
      "grad_norm": 0.6209374666213989,
      "learning_rate": 0.00019728047618210995,
      "loss": 1.9045,
      "step": 3000
    },
    {
      "epoch": 0.7464814929629859,
      "grad_norm": 0.6270463466644287,
      "learning_rate": 0.00019726239912189382,
      "loss": 1.8652,
      "step": 3010
    },
    {
      "epoch": 0.7489614979229958,
      "grad_norm": 0.5481278896331787,
      "learning_rate": 0.0001972442630138868,
      "loss": 1.9435,
      "step": 3020
    },
    {
      "epoch": 0.7514415028830058,
      "grad_norm": 0.5757542252540588,
      "learning_rate": 0.00019722606786909928,
      "loss": 1.9238,
      "step": 3030
    },
    {
      "epoch": 0.7539215078430157,
      "grad_norm": 0.6415989995002747,
      "learning_rate": 0.00019720781369857746,
      "loss": 1.9384,
      "step": 3040
    },
    {
      "epoch": 0.7564015128030256,
      "grad_norm": 0.5898995995521545,
      "learning_rate": 0.00019718950051340347,
      "loss": 1.9314,
      "step": 3050
    },
    {
      "epoch": 0.7588815177630355,
      "grad_norm": 0.5926699042320251,
      "learning_rate": 0.0001971711283246951,
      "loss": 1.9161,
      "step": 3060
    },
    {
      "epoch": 0.7613615227230455,
      "grad_norm": 0.624092698097229,
      "learning_rate": 0.0001971526971436062,
      "loss": 1.9285,
      "step": 3070
    },
    {
      "epoch": 0.7638415276830554,
      "grad_norm": 0.606102705001831,
      "learning_rate": 0.00019713420698132614,
      "loss": 1.9185,
      "step": 3080
    },
    {
      "epoch": 0.7663215326430652,
      "grad_norm": 0.6747568845748901,
      "learning_rate": 0.0001971156578490804,
      "loss": 1.9096,
      "step": 3090
    },
    {
      "epoch": 0.7688015376030752,
      "grad_norm": 0.6271367073059082,
      "learning_rate": 0.00019709704975813007,
      "loss": 1.9465,
      "step": 3100
    },
    {
      "epoch": 0.7712815425630851,
      "grad_norm": 0.6246865391731262,
      "learning_rate": 0.00019707838271977205,
      "loss": 1.8818,
      "step": 3110
    },
    {
      "epoch": 0.773761547523095,
      "grad_norm": 0.6215932369232178,
      "learning_rate": 0.0001970596567453391,
      "loss": 1.9701,
      "step": 3120
    },
    {
      "epoch": 0.776241552483105,
      "grad_norm": 0.6191825866699219,
      "learning_rate": 0.0001970408718461997,
      "loss": 1.9358,
      "step": 3130
    },
    {
      "epoch": 0.7787215574431149,
      "grad_norm": 0.6205334663391113,
      "learning_rate": 0.0001970220280337581,
      "loss": 1.8874,
      "step": 3140
    },
    {
      "epoch": 0.7812015624031248,
      "grad_norm": 0.5906861424446106,
      "learning_rate": 0.00019700312531945442,
      "loss": 1.8725,
      "step": 3150
    },
    {
      "epoch": 0.7836815673631348,
      "grad_norm": 0.6469618082046509,
      "learning_rate": 0.00019698416371476433,
      "loss": 1.9795,
      "step": 3160
    },
    {
      "epoch": 0.7861615723231447,
      "grad_norm": 0.6400570273399353,
      "learning_rate": 0.0001969651432311995,
      "loss": 1.9573,
      "step": 3170
    },
    {
      "epoch": 0.7886415772831545,
      "grad_norm": 0.56891930103302,
      "learning_rate": 0.00019694606388030714,
      "loss": 1.9016,
      "step": 3180
    },
    {
      "epoch": 0.7911215822431645,
      "grad_norm": 0.6157445907592773,
      "learning_rate": 0.00019692692567367033,
      "loss": 1.866,
      "step": 3190
    },
    {
      "epoch": 0.7936015872031744,
      "grad_norm": 0.5788578391075134,
      "learning_rate": 0.0001969077286229078,
      "loss": 1.8527,
      "step": 3200
    },
    {
      "epoch": 0.7960815921631843,
      "grad_norm": 0.59492427110672,
      "learning_rate": 0.00019688847273967408,
      "loss": 1.885,
      "step": 3210
    },
    {
      "epoch": 0.7985615971231943,
      "grad_norm": 0.6037836074829102,
      "learning_rate": 0.00019686915803565934,
      "loss": 1.9793,
      "step": 3220
    },
    {
      "epoch": 0.8010416020832042,
      "grad_norm": 0.6103031039237976,
      "learning_rate": 0.0001968497845225895,
      "loss": 1.8906,
      "step": 3230
    },
    {
      "epoch": 0.8035216070432141,
      "grad_norm": 0.5790871381759644,
      "learning_rate": 0.00019683035221222618,
      "loss": 1.9323,
      "step": 3240
    },
    {
      "epoch": 0.806001612003224,
      "grad_norm": 0.6135836243629456,
      "learning_rate": 0.00019681086111636667,
      "loss": 1.9197,
      "step": 3250
    },
    {
      "epoch": 0.808481616963234,
      "grad_norm": 0.5828390717506409,
      "learning_rate": 0.00019679131124684403,
      "loss": 1.9198,
      "step": 3260
    },
    {
      "epoch": 0.8109616219232438,
      "grad_norm": 0.6304100751876831,
      "learning_rate": 0.0001967717026155269,
      "loss": 1.9025,
      "step": 3270
    },
    {
      "epoch": 0.8134416268832537,
      "grad_norm": 0.5930795073509216,
      "learning_rate": 0.00019675203523431966,
      "loss": 1.9309,
      "step": 3280
    },
    {
      "epoch": 0.8159216318432637,
      "grad_norm": 0.6989979147911072,
      "learning_rate": 0.00019673230911516226,
      "loss": 1.956,
      "step": 3290
    },
    {
      "epoch": 0.8184016368032736,
      "grad_norm": 0.7460166811943054,
      "learning_rate": 0.0001967125242700305,
      "loss": 1.9574,
      "step": 3300
    },
    {
      "epoch": 0.8208816417632835,
      "grad_norm": 0.6681553721427917,
      "learning_rate": 0.00019669268071093566,
      "loss": 1.9416,
      "step": 3310
    },
    {
      "epoch": 0.8233616467232935,
      "grad_norm": 0.644540548324585,
      "learning_rate": 0.00019667277844992475,
      "loss": 1.8949,
      "step": 3320
    },
    {
      "epoch": 0.8258416516833034,
      "grad_norm": 0.5839378833770752,
      "learning_rate": 0.00019665281749908033,
      "loss": 1.9243,
      "step": 3330
    },
    {
      "epoch": 0.8283216566433133,
      "grad_norm": 0.6076035499572754,
      "learning_rate": 0.00019663279787052068,
      "loss": 1.92,
      "step": 3340
    },
    {
      "epoch": 0.8308016616033232,
      "grad_norm": 0.6083244681358337,
      "learning_rate": 0.0001966127195763997,
      "loss": 1.9054,
      "step": 3350
    },
    {
      "epoch": 0.8332816665633331,
      "grad_norm": 0.6293017268180847,
      "learning_rate": 0.00019659258262890683,
      "loss": 1.9715,
      "step": 3360
    },
    {
      "epoch": 0.835761671523343,
      "grad_norm": 0.6333664059638977,
      "learning_rate": 0.00019657238704026721,
      "loss": 1.9388,
      "step": 3370
    },
    {
      "epoch": 0.838241676483353,
      "grad_norm": 0.6385533809661865,
      "learning_rate": 0.00019655213282274153,
      "loss": 1.8607,
      "step": 3380
    },
    {
      "epoch": 0.8407216814433629,
      "grad_norm": 0.5880801677703857,
      "learning_rate": 0.00019653181998862605,
      "loss": 1.9777,
      "step": 3390
    },
    {
      "epoch": 0.8432016864033728,
      "grad_norm": 0.610713541507721,
      "learning_rate": 0.00019651144855025265,
      "loss": 1.9246,
      "step": 3400
    },
    {
      "epoch": 0.8456816913633828,
      "grad_norm": 0.6361165046691895,
      "learning_rate": 0.00019649101851998885,
      "loss": 1.9277,
      "step": 3410
    },
    {
      "epoch": 0.8481616963233927,
      "grad_norm": 0.5994930863380432,
      "learning_rate": 0.0001964705299102376,
      "loss": 1.8723,
      "step": 3420
    },
    {
      "epoch": 0.8506417012834026,
      "grad_norm": 0.5819071531295776,
      "learning_rate": 0.00019644998273343753,
      "loss": 1.8956,
      "step": 3430
    },
    {
      "epoch": 0.8531217062434124,
      "grad_norm": 0.6169791221618652,
      "learning_rate": 0.0001964293770020628,
      "loss": 1.9188,
      "step": 3440
    },
    {
      "epoch": 0.8556017112034224,
      "grad_norm": 0.9334343671798706,
      "learning_rate": 0.00019640871272862302,
      "loss": 1.9583,
      "step": 3450
    },
    {
      "epoch": 0.8580817161634323,
      "grad_norm": 0.6008388996124268,
      "learning_rate": 0.00019638798992566354,
      "loss": 1.9483,
      "step": 3460
    },
    {
      "epoch": 0.8605617211234422,
      "grad_norm": 0.642737627029419,
      "learning_rate": 0.00019636720860576507,
      "loss": 1.9526,
      "step": 3470
    },
    {
      "epoch": 0.8630417260834522,
      "grad_norm": 0.6243256330490112,
      "learning_rate": 0.0001963463687815439,
      "loss": 1.923,
      "step": 3480
    },
    {
      "epoch": 0.8655217310434621,
      "grad_norm": 0.5904022455215454,
      "learning_rate": 0.00019632547046565186,
      "loss": 1.8666,
      "step": 3490
    },
    {
      "epoch": 0.868001736003472,
      "grad_norm": 0.6636185646057129,
      "learning_rate": 0.00019630451367077628,
      "loss": 1.8892,
      "step": 3500
    },
    {
      "epoch": 0.870481740963482,
      "grad_norm": 0.5875678062438965,
      "learning_rate": 0.00019628349840963995,
      "loss": 1.9218,
      "step": 3510
    },
    {
      "epoch": 0.8729617459234918,
      "grad_norm": 0.6118807792663574,
      "learning_rate": 0.0001962624246950012,
      "loss": 1.8806,
      "step": 3520
    },
    {
      "epoch": 0.8754417508835017,
      "grad_norm": 0.6047744750976562,
      "learning_rate": 0.00019624129253965387,
      "loss": 1.949,
      "step": 3530
    },
    {
      "epoch": 0.8779217558435117,
      "grad_norm": 0.6061970591545105,
      "learning_rate": 0.0001962201019564272,
      "loss": 1.9103,
      "step": 3540
    },
    {
      "epoch": 0.8804017608035216,
      "grad_norm": 0.6186763048171997,
      "learning_rate": 0.000196198852958186,
      "loss": 1.9533,
      "step": 3550
    },
    {
      "epoch": 0.8828817657635315,
      "grad_norm": 0.602182924747467,
      "learning_rate": 0.00019617754555783043,
      "loss": 1.8873,
      "step": 3560
    },
    {
      "epoch": 0.8853617707235415,
      "grad_norm": 0.6514226794242859,
      "learning_rate": 0.0001961561797682962,
      "loss": 1.9642,
      "step": 3570
    },
    {
      "epoch": 0.8878417756835514,
      "grad_norm": 0.6289548277854919,
      "learning_rate": 0.00019613475560255442,
      "loss": 1.8861,
      "step": 3580
    },
    {
      "epoch": 0.8903217806435613,
      "grad_norm": 0.6488407850265503,
      "learning_rate": 0.0001961132730736117,
      "loss": 1.8255,
      "step": 3590
    },
    {
      "epoch": 0.8928017856035712,
      "grad_norm": 0.6131303906440735,
      "learning_rate": 0.00019609173219450998,
      "loss": 1.905,
      "step": 3600
    },
    {
      "epoch": 0.8952817905635811,
      "grad_norm": 0.6090990304946899,
      "learning_rate": 0.00019607013297832669,
      "loss": 1.8714,
      "step": 3610
    },
    {
      "epoch": 0.897761795523591,
      "grad_norm": 0.6134405732154846,
      "learning_rate": 0.00019604847543817466,
      "loss": 1.9458,
      "step": 3620
    },
    {
      "epoch": 0.9002418004836009,
      "grad_norm": 0.6179640889167786,
      "learning_rate": 0.00019602675958720222,
      "loss": 1.9406,
      "step": 3630
    },
    {
      "epoch": 0.9027218054436109,
      "grad_norm": 0.6475931406021118,
      "learning_rate": 0.0001960049854385929,
      "loss": 1.9175,
      "step": 3640
    },
    {
      "epoch": 0.9052018104036208,
      "grad_norm": 0.6346822381019592,
      "learning_rate": 0.00019598315300556576,
      "loss": 1.859,
      "step": 3650
    },
    {
      "epoch": 0.9076818153636307,
      "grad_norm": 0.6485849618911743,
      "learning_rate": 0.0001959612623013753,
      "loss": 1.9199,
      "step": 3660
    },
    {
      "epoch": 0.9101618203236407,
      "grad_norm": 0.6322528719902039,
      "learning_rate": 0.00019593931333931127,
      "loss": 1.9031,
      "step": 3670
    },
    {
      "epoch": 0.9126418252836506,
      "grad_norm": 0.5780066251754761,
      "learning_rate": 0.0001959173061326988,
      "loss": 1.9051,
      "step": 3680
    },
    {
      "epoch": 0.9151218302436605,
      "grad_norm": 0.5900623202323914,
      "learning_rate": 0.0001958952406948985,
      "loss": 1.8254,
      "step": 3690
    },
    {
      "epoch": 0.9176018352036704,
      "grad_norm": 0.7210493087768555,
      "learning_rate": 0.00019587311703930614,
      "loss": 1.891,
      "step": 3700
    },
    {
      "epoch": 0.9200818401636803,
      "grad_norm": 0.5721245408058167,
      "learning_rate": 0.00019585093517935305,
      "loss": 1.8755,
      "step": 3710
    },
    {
      "epoch": 0.9225618451236902,
      "grad_norm": 0.5767670273780823,
      "learning_rate": 0.00019582869512850575,
      "loss": 1.9101,
      "step": 3720
    },
    {
      "epoch": 0.9250418500837002,
      "grad_norm": 0.591179370880127,
      "learning_rate": 0.0001958063969002661,
      "loss": 1.903,
      "step": 3730
    },
    {
      "epoch": 0.9275218550437101,
      "grad_norm": 0.6186399459838867,
      "learning_rate": 0.00019578404050817134,
      "loss": 1.9523,
      "step": 3740
    },
    {
      "epoch": 0.93000186000372,
      "grad_norm": 0.5627548694610596,
      "learning_rate": 0.00019576162596579398,
      "loss": 1.9311,
      "step": 3750
    },
    {
      "epoch": 0.93248186496373,
      "grad_norm": 0.6412395238876343,
      "learning_rate": 0.0001957391532867418,
      "loss": 1.9825,
      "step": 3760
    },
    {
      "epoch": 0.9349618699237399,
      "grad_norm": 0.6057267189025879,
      "learning_rate": 0.000195716622484658,
      "loss": 1.9442,
      "step": 3770
    },
    {
      "epoch": 0.9374418748837497,
      "grad_norm": 0.6053009629249573,
      "learning_rate": 0.0001956940335732209,
      "loss": 1.9398,
      "step": 3780
    },
    {
      "epoch": 0.9399218798437596,
      "grad_norm": 0.6406477093696594,
      "learning_rate": 0.0001956713865661442,
      "loss": 1.8631,
      "step": 3790
    },
    {
      "epoch": 0.9424018848037696,
      "grad_norm": 0.6801642775535583,
      "learning_rate": 0.0001956486814771769,
      "loss": 1.9436,
      "step": 3800
    },
    {
      "epoch": 0.9448818897637795,
      "grad_norm": 0.7259861826896667,
      "learning_rate": 0.00019562591832010312,
      "loss": 1.854,
      "step": 3810
    },
    {
      "epoch": 0.9473618947237894,
      "grad_norm": 0.6290436387062073,
      "learning_rate": 0.00019560309710874238,
      "loss": 1.8815,
      "step": 3820
    },
    {
      "epoch": 0.9498418996837994,
      "grad_norm": 0.6102296710014343,
      "learning_rate": 0.0001955802178569494,
      "loss": 1.9118,
      "step": 3830
    },
    {
      "epoch": 0.9523219046438093,
      "grad_norm": 0.6027212738990784,
      "learning_rate": 0.0001955572805786141,
      "loss": 1.9514,
      "step": 3840
    },
    {
      "epoch": 0.9548019096038192,
      "grad_norm": 1.153975486755371,
      "learning_rate": 0.00019553428528766163,
      "loss": 1.8342,
      "step": 3850
    },
    {
      "epoch": 0.9572819145638292,
      "grad_norm": 0.6031201481819153,
      "learning_rate": 0.0001955112319980524,
      "loss": 1.9253,
      "step": 3860
    },
    {
      "epoch": 0.959761919523839,
      "grad_norm": 0.6483965516090393,
      "learning_rate": 0.00019548812072378205,
      "loss": 1.8936,
      "step": 3870
    },
    {
      "epoch": 0.9622419244838489,
      "grad_norm": 0.6587181687355042,
      "learning_rate": 0.00019546495147888132,
      "loss": 1.9135,
      "step": 3880
    },
    {
      "epoch": 0.9647219294438589,
      "grad_norm": 0.5917319655418396,
      "learning_rate": 0.00019544172427741628,
      "loss": 1.8195,
      "step": 3890
    },
    {
      "epoch": 0.9672019344038688,
      "grad_norm": 0.6090488433837891,
      "learning_rate": 0.00019541843913348805,
      "loss": 1.8985,
      "step": 3900
    },
    {
      "epoch": 0.9696819393638787,
      "grad_norm": 0.5770931243896484,
      "learning_rate": 0.000195395096061233,
      "loss": 1.8363,
      "step": 3910
    },
    {
      "epoch": 0.9721619443238887,
      "grad_norm": 0.6690269112586975,
      "learning_rate": 0.0001953716950748227,
      "loss": 1.9192,
      "step": 3920
    },
    {
      "epoch": 0.9746419492838986,
      "grad_norm": 0.6674990653991699,
      "learning_rate": 0.0001953482361884638,
      "loss": 1.914,
      "step": 3930
    },
    {
      "epoch": 0.9771219542439085,
      "grad_norm": 0.6217818856239319,
      "learning_rate": 0.00019532471941639815,
      "loss": 1.9067,
      "step": 3940
    },
    {
      "epoch": 0.9796019592039185,
      "grad_norm": 0.569581151008606,
      "learning_rate": 0.00019530114477290272,
      "loss": 1.9381,
      "step": 3950
    },
    {
      "epoch": 0.9820819641639283,
      "grad_norm": 0.6703535318374634,
      "learning_rate": 0.00019527751227228963,
      "loss": 1.9191,
      "step": 3960
    },
    {
      "epoch": 0.9845619691239382,
      "grad_norm": 0.60328209400177,
      "learning_rate": 0.00019525382192890614,
      "loss": 1.9204,
      "step": 3970
    },
    {
      "epoch": 0.9870419740839481,
      "grad_norm": 0.6037915349006653,
      "learning_rate": 0.0001952300737571346,
      "loss": 1.8657,
      "step": 3980
    },
    {
      "epoch": 0.9895219790439581,
      "grad_norm": 0.6142215728759766,
      "learning_rate": 0.00019520626777139245,
      "loss": 1.891,
      "step": 3990
    },
    {
      "epoch": 0.992001984003968,
      "grad_norm": 0.6303610801696777,
      "learning_rate": 0.00019518240398613227,
      "loss": 1.9321,
      "step": 4000
    },
    {
      "epoch": 0.9944819889639779,
      "grad_norm": 0.6166995763778687,
      "learning_rate": 0.00019515848241584167,
      "loss": 1.8985,
      "step": 4010
    },
    {
      "epoch": 0.9969619939239879,
      "grad_norm": 0.6104696989059448,
      "learning_rate": 0.00019513450307504348,
      "loss": 1.8478,
      "step": 4020
    },
    {
      "epoch": 0.9994419988839978,
      "grad_norm": 0.6207946538925171,
      "learning_rate": 0.0001951104659782954,
      "loss": 1.9168,
      "step": 4030
    },
    {
      "epoch": 1.0019220038440078,
      "grad_norm": 0.6029482483863831,
      "learning_rate": 0.00019508637114019038,
      "loss": 1.7681,
      "step": 4040
    },
    {
      "epoch": 1.0044020088040175,
      "grad_norm": 0.6340354084968567,
      "learning_rate": 0.0001950622185753563,
      "loss": 1.816,
      "step": 4050
    },
    {
      "epoch": 1.0068820137640275,
      "grad_norm": 0.6285779476165771,
      "learning_rate": 0.0001950380082984561,
      "loss": 1.7906,
      "step": 4060
    },
    {
      "epoch": 1.0093620187240375,
      "grad_norm": 0.6094495058059692,
      "learning_rate": 0.0001950137403241879,
      "loss": 1.7904,
      "step": 4070
    },
    {
      "epoch": 1.0118420236840473,
      "grad_norm": 0.7566218376159668,
      "learning_rate": 0.0001949894146672846,
      "loss": 1.8968,
      "step": 4080
    },
    {
      "epoch": 1.0143220286440573,
      "grad_norm": 0.6424439549446106,
      "learning_rate": 0.00019496503134251435,
      "loss": 1.7873,
      "step": 4090
    },
    {
      "epoch": 1.0168020336040673,
      "grad_norm": 0.6438541412353516,
      "learning_rate": 0.00019494059036468016,
      "loss": 1.8155,
      "step": 4100
    },
    {
      "epoch": 1.019282038564077,
      "grad_norm": 0.621476948261261,
      "learning_rate": 0.0001949160917486201,
      "loss": 1.8174,
      "step": 4110
    },
    {
      "epoch": 1.021762043524087,
      "grad_norm": 0.6082315444946289,
      "learning_rate": 0.00019489153550920728,
      "loss": 1.774,
      "step": 4120
    },
    {
      "epoch": 1.024242048484097,
      "grad_norm": 0.652134895324707,
      "learning_rate": 0.00019486692166134964,
      "loss": 1.909,
      "step": 4130
    },
    {
      "epoch": 1.0267220534441068,
      "grad_norm": 0.6727526783943176,
      "learning_rate": 0.0001948422502199903,
      "loss": 1.8005,
      "step": 4140
    },
    {
      "epoch": 1.0292020584041168,
      "grad_norm": 0.6741703748703003,
      "learning_rate": 0.00019481752120010715,
      "loss": 1.8373,
      "step": 4150
    },
    {
      "epoch": 1.0316820633641268,
      "grad_norm": 0.6614888906478882,
      "learning_rate": 0.0001947927346167132,
      "loss": 1.8277,
      "step": 4160
    },
    {
      "epoch": 1.0341620683241366,
      "grad_norm": 0.6599487066268921,
      "learning_rate": 0.00019476789048485625,
      "loss": 1.8234,
      "step": 4170
    },
    {
      "epoch": 1.0366420732841466,
      "grad_norm": 0.6961712837219238,
      "learning_rate": 0.00019474298881961918,
      "loss": 1.8076,
      "step": 4180
    },
    {
      "epoch": 1.0391220782441566,
      "grad_norm": 0.6359785795211792,
      "learning_rate": 0.00019471802963611973,
      "loss": 1.8765,
      "step": 4190
    },
    {
      "epoch": 1.0416020832041664,
      "grad_norm": 0.6598185896873474,
      "learning_rate": 0.0001946930129495106,
      "loss": 1.8113,
      "step": 4200
    },
    {
      "epoch": 1.0440820881641764,
      "grad_norm": 0.6477219462394714,
      "learning_rate": 0.00019466793877497923,
      "loss": 1.7794,
      "step": 4210
    },
    {
      "epoch": 1.0465620931241864,
      "grad_norm": 0.6349904537200928,
      "learning_rate": 0.00019464280712774826,
      "loss": 1.8253,
      "step": 4220
    },
    {
      "epoch": 1.0490420980841961,
      "grad_norm": 0.7010254263877869,
      "learning_rate": 0.00019461761802307495,
      "loss": 1.8362,
      "step": 4230
    },
    {
      "epoch": 1.0515221030442061,
      "grad_norm": 0.6054363250732422,
      "learning_rate": 0.0001945923714762516,
      "loss": 1.8158,
      "step": 4240
    },
    {
      "epoch": 1.054002108004216,
      "grad_norm": 0.6621227264404297,
      "learning_rate": 0.00019456706750260532,
      "loss": 1.7927,
      "step": 4250
    },
    {
      "epoch": 1.056482112964226,
      "grad_norm": 0.656527042388916,
      "learning_rate": 0.0001945417061174981,
      "loss": 1.8297,
      "step": 4260
    },
    {
      "epoch": 1.058962117924236,
      "grad_norm": 0.6720926761627197,
      "learning_rate": 0.0001945162873363268,
      "loss": 1.8212,
      "step": 4270
    },
    {
      "epoch": 1.0614421228842457,
      "grad_norm": 0.670739471912384,
      "learning_rate": 0.00019449081117452302,
      "loss": 1.8129,
      "step": 4280
    },
    {
      "epoch": 1.0639221278442557,
      "grad_norm": 0.6356748938560486,
      "learning_rate": 0.00019446527764755342,
      "loss": 1.8215,
      "step": 4290
    },
    {
      "epoch": 1.0664021328042657,
      "grad_norm": 0.6663000583648682,
      "learning_rate": 0.00019443968677091927,
      "loss": 1.782,
      "step": 4300
    },
    {
      "epoch": 1.0688821377642754,
      "grad_norm": 0.6405335068702698,
      "learning_rate": 0.0001944140385601567,
      "loss": 1.7916,
      "step": 4310
    },
    {
      "epoch": 1.0713621427242854,
      "grad_norm": 0.6829027533531189,
      "learning_rate": 0.00019438833303083678,
      "loss": 1.8293,
      "step": 4320
    },
    {
      "epoch": 1.0738421476842954,
      "grad_norm": 0.6544651389122009,
      "learning_rate": 0.00019436257019856518,
      "loss": 1.8509,
      "step": 4330
    },
    {
      "epoch": 1.0763221526443052,
      "grad_norm": 0.6550015211105347,
      "learning_rate": 0.00019433675007898255,
      "loss": 1.8203,
      "step": 4340
    },
    {
      "epoch": 1.0788021576043152,
      "grad_norm": 0.6609222888946533,
      "learning_rate": 0.00019431087268776412,
      "loss": 1.7975,
      "step": 4350
    },
    {
      "epoch": 1.0812821625643252,
      "grad_norm": 0.6910188794136047,
      "learning_rate": 0.0001942849380406201,
      "loss": 1.8557,
      "step": 4360
    },
    {
      "epoch": 1.083762167524335,
      "grad_norm": 0.6951532959938049,
      "learning_rate": 0.00019425894615329536,
      "loss": 1.7861,
      "step": 4370
    },
    {
      "epoch": 1.086242172484345,
      "grad_norm": 0.6598848104476929,
      "learning_rate": 0.00019423289704156944,
      "loss": 1.7694,
      "step": 4380
    },
    {
      "epoch": 1.088722177444355,
      "grad_norm": 0.7436254024505615,
      "learning_rate": 0.00019420679072125677,
      "loss": 1.7986,
      "step": 4390
    },
    {
      "epoch": 1.0912021824043647,
      "grad_norm": 0.6893884539604187,
      "learning_rate": 0.00019418062720820637,
      "loss": 1.7821,
      "step": 4400
    },
    {
      "epoch": 1.0936821873643747,
      "grad_norm": 0.713741660118103,
      "learning_rate": 0.00019415440651830208,
      "loss": 1.8634,
      "step": 4410
    },
    {
      "epoch": 1.0961621923243847,
      "grad_norm": 0.700987696647644,
      "learning_rate": 0.00019412812866746246,
      "loss": 1.7938,
      "step": 4420
    },
    {
      "epoch": 1.0986421972843945,
      "grad_norm": 0.7220914363861084,
      "learning_rate": 0.00019410179367164071,
      "loss": 1.8701,
      "step": 4430
    },
    {
      "epoch": 1.1011222022444045,
      "grad_norm": 0.7219149470329285,
      "learning_rate": 0.00019407540154682472,
      "loss": 1.807,
      "step": 4440
    },
    {
      "epoch": 1.1036022072044145,
      "grad_norm": 0.7215644717216492,
      "learning_rate": 0.00019404895230903718,
      "loss": 1.7848,
      "step": 4450
    },
    {
      "epoch": 1.1060822121644243,
      "grad_norm": 0.6691747307777405,
      "learning_rate": 0.00019402244597433525,
      "loss": 1.8294,
      "step": 4460
    },
    {
      "epoch": 1.1085622171244343,
      "grad_norm": 0.6951769590377808,
      "learning_rate": 0.00019399588255881095,
      "loss": 1.7931,
      "step": 4470
    },
    {
      "epoch": 1.1110422220844443,
      "grad_norm": 0.8239799737930298,
      "learning_rate": 0.00019396926207859084,
      "loss": 1.7905,
      "step": 4480
    },
    {
      "epoch": 1.113522227044454,
      "grad_norm": 0.68569016456604,
      "learning_rate": 0.00019394258454983617,
      "loss": 1.774,
      "step": 4490
    },
    {
      "epoch": 1.116002232004464,
      "grad_norm": 0.7133092880249023,
      "learning_rate": 0.0001939158499887428,
      "loss": 1.8231,
      "step": 4500
    },
    {
      "epoch": 1.118482236964474,
      "grad_norm": 0.7494200468063354,
      "learning_rate": 0.0001938890584115412,
      "loss": 1.7985,
      "step": 4510
    },
    {
      "epoch": 1.1209622419244838,
      "grad_norm": 0.6811527609825134,
      "learning_rate": 0.00019386220983449653,
      "loss": 1.8308,
      "step": 4520
    },
    {
      "epoch": 1.1234422468844938,
      "grad_norm": 0.7043130993843079,
      "learning_rate": 0.00019383530427390845,
      "loss": 1.7974,
      "step": 4530
    },
    {
      "epoch": 1.1259222518445038,
      "grad_norm": 0.6619195938110352,
      "learning_rate": 0.00019380834174611132,
      "loss": 1.7768,
      "step": 4540
    },
    {
      "epoch": 1.1284022568045136,
      "grad_norm": 0.6526765823364258,
      "learning_rate": 0.00019378132226747398,
      "loss": 1.7887,
      "step": 4550
    },
    {
      "epoch": 1.1308822617645236,
      "grad_norm": 0.6597820520401001,
      "learning_rate": 0.00019375424585439994,
      "loss": 1.7811,
      "step": 4560
    },
    {
      "epoch": 1.1333622667245336,
      "grad_norm": 0.6860714554786682,
      "learning_rate": 0.00019372711252332717,
      "loss": 1.8105,
      "step": 4570
    },
    {
      "epoch": 1.1358422716845433,
      "grad_norm": 0.72899329662323,
      "learning_rate": 0.00019369992229072836,
      "loss": 1.8321,
      "step": 4580
    },
    {
      "epoch": 1.1383222766445533,
      "grad_norm": 0.6844159960746765,
      "learning_rate": 0.00019367267517311057,
      "loss": 1.7651,
      "step": 4590
    },
    {
      "epoch": 1.140802281604563,
      "grad_norm": 0.6703256964683533,
      "learning_rate": 0.00019364537118701542,
      "loss": 1.7685,
      "step": 4600
    },
    {
      "epoch": 1.143282286564573,
      "grad_norm": 0.6494504809379578,
      "learning_rate": 0.00019361801034901922,
      "loss": 1.8628,
      "step": 4610
    },
    {
      "epoch": 1.145762291524583,
      "grad_norm": 0.6616336107254028,
      "learning_rate": 0.0001935905926757326,
      "loss": 1.7785,
      "step": 4620
    },
    {
      "epoch": 1.148242296484593,
      "grad_norm": 0.6509258151054382,
      "learning_rate": 0.00019356311818380075,
      "loss": 1.8705,
      "step": 4630
    },
    {
      "epoch": 1.1507223014446029,
      "grad_norm": 0.711273193359375,
      "learning_rate": 0.0001935355868899034,
      "loss": 1.8168,
      "step": 4640
    },
    {
      "epoch": 1.1532023064046129,
      "grad_norm": 0.6760425567626953,
      "learning_rate": 0.0001935079988107548,
      "loss": 1.8339,
      "step": 4650
    },
    {
      "epoch": 1.1556823113646226,
      "grad_norm": 0.6510275602340698,
      "learning_rate": 0.0001934803539631035,
      "loss": 1.8179,
      "step": 4660
    },
    {
      "epoch": 1.1581623163246326,
      "grad_norm": 0.6667962074279785,
      "learning_rate": 0.00019345265236373274,
      "loss": 1.8456,
      "step": 4670
    },
    {
      "epoch": 1.1606423212846426,
      "grad_norm": 0.6934930086135864,
      "learning_rate": 0.00019342489402945998,
      "loss": 1.8093,
      "step": 4680
    },
    {
      "epoch": 1.1631223262446524,
      "grad_norm": 0.6169160604476929,
      "learning_rate": 0.00019339707897713738,
      "loss": 1.793,
      "step": 4690
    },
    {
      "epoch": 1.1656023312046624,
      "grad_norm": 0.6679814457893372,
      "learning_rate": 0.00019336920722365125,
      "loss": 1.8465,
      "step": 4700
    },
    {
      "epoch": 1.1680823361646724,
      "grad_norm": 0.7247648239135742,
      "learning_rate": 0.0001933412787859226,
      "loss": 1.8768,
      "step": 4710
    },
    {
      "epoch": 1.1705623411246822,
      "grad_norm": 0.7149258255958557,
      "learning_rate": 0.00019331329368090666,
      "loss": 1.8649,
      "step": 4720
    },
    {
      "epoch": 1.1730423460846922,
      "grad_norm": 0.6846023797988892,
      "learning_rate": 0.00019328525192559312,
      "loss": 1.8084,
      "step": 4730
    },
    {
      "epoch": 1.1755223510447022,
      "grad_norm": 0.6747673153877258,
      "learning_rate": 0.00019325715353700612,
      "loss": 1.8186,
      "step": 4740
    },
    {
      "epoch": 1.178002356004712,
      "grad_norm": 0.6865365505218506,
      "learning_rate": 0.0001932289985322041,
      "loss": 1.8744,
      "step": 4750
    },
    {
      "epoch": 1.180482360964722,
      "grad_norm": 0.7503153085708618,
      "learning_rate": 0.00019320078692827987,
      "loss": 1.8562,
      "step": 4760
    },
    {
      "epoch": 1.182962365924732,
      "grad_norm": 0.6764112114906311,
      "learning_rate": 0.00019317251874236067,
      "loss": 1.8092,
      "step": 4770
    },
    {
      "epoch": 1.1854423708847417,
      "grad_norm": 0.6778523921966553,
      "learning_rate": 0.00019314419399160803,
      "loss": 1.847,
      "step": 4780
    },
    {
      "epoch": 1.1879223758447517,
      "grad_norm": 0.6746340394020081,
      "learning_rate": 0.0001931158126932179,
      "loss": 1.7979,
      "step": 4790
    },
    {
      "epoch": 1.1904023808047617,
      "grad_norm": 0.7232058644294739,
      "learning_rate": 0.00019308737486442045,
      "loss": 1.875,
      "step": 4800
    },
    {
      "epoch": 1.1928823857647715,
      "grad_norm": 0.7230646014213562,
      "learning_rate": 0.00019305888052248023,
      "loss": 1.8013,
      "step": 4810
    },
    {
      "epoch": 1.1953623907247815,
      "grad_norm": 0.6311892867088318,
      "learning_rate": 0.0001930303296846961,
      "loss": 1.8306,
      "step": 4820
    },
    {
      "epoch": 1.1978423956847914,
      "grad_norm": 0.70650315284729,
      "learning_rate": 0.00019300172236840122,
      "loss": 1.8035,
      "step": 4830
    },
    {
      "epoch": 1.2003224006448012,
      "grad_norm": 0.6846140623092651,
      "learning_rate": 0.00019297305859096304,
      "loss": 1.8046,
      "step": 4840
    },
    {
      "epoch": 1.2028024056048112,
      "grad_norm": 0.6740015149116516,
      "learning_rate": 0.00019294433836978322,
      "loss": 1.8784,
      "step": 4850
    },
    {
      "epoch": 1.2052824105648212,
      "grad_norm": 0.7008079886436462,
      "learning_rate": 0.00019291556172229785,
      "loss": 1.8449,
      "step": 4860
    },
    {
      "epoch": 1.207762415524831,
      "grad_norm": 0.6753029823303223,
      "learning_rate": 0.00019288672866597705,
      "loss": 1.7802,
      "step": 4870
    },
    {
      "epoch": 1.210242420484841,
      "grad_norm": 0.7091749310493469,
      "learning_rate": 0.00019285783921832537,
      "loss": 1.8042,
      "step": 4880
    },
    {
      "epoch": 1.212722425444851,
      "grad_norm": 0.7150561809539795,
      "learning_rate": 0.00019282889339688152,
      "loss": 1.9043,
      "step": 4890
    },
    {
      "epoch": 1.2152024304048608,
      "grad_norm": 0.8080011606216431,
      "learning_rate": 0.00019279989121921847,
      "loss": 1.8447,
      "step": 4900
    },
    {
      "epoch": 1.2176824353648708,
      "grad_norm": 0.705141544342041,
      "learning_rate": 0.00019277083270294334,
      "loss": 1.7947,
      "step": 4910
    },
    {
      "epoch": 1.2201624403248807,
      "grad_norm": 0.6963218450546265,
      "learning_rate": 0.0001927417178656975,
      "loss": 1.7765,
      "step": 4920
    },
    {
      "epoch": 1.2226424452848905,
      "grad_norm": 0.6910169124603271,
      "learning_rate": 0.00019271254672515654,
      "loss": 1.7994,
      "step": 4930
    },
    {
      "epoch": 1.2251224502449005,
      "grad_norm": 0.6857121586799622,
      "learning_rate": 0.00019268331929903014,
      "loss": 1.8188,
      "step": 4940
    },
    {
      "epoch": 1.2276024552049103,
      "grad_norm": 0.6874817609786987,
      "learning_rate": 0.00019265403560506222,
      "loss": 1.8254,
      "step": 4950
    },
    {
      "epoch": 1.2300824601649203,
      "grad_norm": 0.7109423875808716,
      "learning_rate": 0.0001926246956610309,
      "loss": 1.7868,
      "step": 4960
    },
    {
      "epoch": 1.2325624651249303,
      "grad_norm": 0.6798704266548157,
      "learning_rate": 0.00019259529948474833,
      "loss": 1.813,
      "step": 4970
    },
    {
      "epoch": 1.2350424700849403,
      "grad_norm": 0.7131743431091309,
      "learning_rate": 0.00019256584709406092,
      "loss": 1.7819,
      "step": 4980
    },
    {
      "epoch": 1.23752247504495,
      "grad_norm": 0.7012653946876526,
      "learning_rate": 0.00019253633850684909,
      "loss": 1.7884,
      "step": 4990
    },
    {
      "epoch": 1.24000248000496,
      "grad_norm": 0.6879581809043884,
      "learning_rate": 0.0001925067737410275,
      "loss": 1.8382,
      "step": 5000
    },
    {
      "epoch": 1.2424824849649698,
      "grad_norm": 0.6491703987121582,
      "learning_rate": 0.00019247715281454483,
      "loss": 1.8413,
      "step": 5010
    },
    {
      "epoch": 1.2449624899249798,
      "grad_norm": 0.6270185112953186,
      "learning_rate": 0.00019244747574538384,
      "loss": 1.7905,
      "step": 5020
    },
    {
      "epoch": 1.2474424948849898,
      "grad_norm": 0.6901357769966125,
      "learning_rate": 0.00019241774255156148,
      "loss": 1.7767,
      "step": 5030
    },
    {
      "epoch": 1.2499224998449998,
      "grad_norm": 0.6535638570785522,
      "learning_rate": 0.0001923879532511287,
      "loss": 1.7687,
      "step": 5040
    },
    {
      "epoch": 1.2524025048050096,
      "grad_norm": 0.7104007005691528,
      "learning_rate": 0.00019235810786217048,
      "loss": 1.7414,
      "step": 5050
    },
    {
      "epoch": 1.2548825097650196,
      "grad_norm": 0.7325509190559387,
      "learning_rate": 0.0001923282064028059,
      "loss": 1.8172,
      "step": 5060
    },
    {
      "epoch": 1.2573625147250294,
      "grad_norm": 0.6885420083999634,
      "learning_rate": 0.0001922982488911881,
      "loss": 1.7878,
      "step": 5070
    },
    {
      "epoch": 1.2598425196850394,
      "grad_norm": 0.7148260474205017,
      "learning_rate": 0.00019226823534550418,
      "loss": 1.8057,
      "step": 5080
    },
    {
      "epoch": 1.2623225246450493,
      "grad_norm": 0.6735394597053528,
      "learning_rate": 0.00019223816578397535,
      "loss": 1.8795,
      "step": 5090
    },
    {
      "epoch": 1.2648025296050593,
      "grad_norm": 0.7113747000694275,
      "learning_rate": 0.00019220804022485673,
      "loss": 1.8209,
      "step": 5100
    },
    {
      "epoch": 1.2672825345650691,
      "grad_norm": 0.667385458946228,
      "learning_rate": 0.00019217785868643747,
      "loss": 1.8239,
      "step": 5110
    },
    {
      "epoch": 1.2697625395250791,
      "grad_norm": 0.7234987020492554,
      "learning_rate": 0.00019214762118704076,
      "loss": 1.779,
      "step": 5120
    },
    {
      "epoch": 1.2722425444850889,
      "grad_norm": 0.6842116713523865,
      "learning_rate": 0.00019211732774502372,
      "loss": 1.7783,
      "step": 5130
    },
    {
      "epoch": 1.2747225494450989,
      "grad_norm": 0.7074073553085327,
      "learning_rate": 0.0001920869783787774,
      "loss": 1.7763,
      "step": 5140
    },
    {
      "epoch": 1.2772025544051089,
      "grad_norm": 0.6838715076446533,
      "learning_rate": 0.00019205657310672686,
      "loss": 1.7461,
      "step": 5150
    },
    {
      "epoch": 1.2796825593651187,
      "grad_norm": 0.6861848831176758,
      "learning_rate": 0.00019202611194733108,
      "loss": 1.7965,
      "step": 5160
    },
    {
      "epoch": 1.2821625643251287,
      "grad_norm": 0.7556761503219604,
      "learning_rate": 0.0001919955949190829,
      "loss": 1.8114,
      "step": 5170
    },
    {
      "epoch": 1.2846425692851386,
      "grad_norm": 0.6757796406745911,
      "learning_rate": 0.00019196502204050922,
      "loss": 1.8087,
      "step": 5180
    },
    {
      "epoch": 1.2871225742451484,
      "grad_norm": 0.7062699794769287,
      "learning_rate": 0.00019193439333017076,
      "loss": 1.8071,
      "step": 5190
    },
    {
      "epoch": 1.2896025792051584,
      "grad_norm": 0.6406750679016113,
      "learning_rate": 0.00019190370880666207,
      "loss": 1.8025,
      "step": 5200
    },
    {
      "epoch": 1.2920825841651684,
      "grad_norm": 0.6705310940742493,
      "learning_rate": 0.00019187296848861176,
      "loss": 1.8674,
      "step": 5210
    },
    {
      "epoch": 1.2945625891251782,
      "grad_norm": 0.6770936846733093,
      "learning_rate": 0.00019184217239468212,
      "loss": 1.8778,
      "step": 5220
    },
    {
      "epoch": 1.2970425940851882,
      "grad_norm": 0.7093015313148499,
      "learning_rate": 0.00019181132054356943,
      "loss": 1.8537,
      "step": 5230
    },
    {
      "epoch": 1.299522599045198,
      "grad_norm": 0.6702137589454651,
      "learning_rate": 0.00019178041295400382,
      "loss": 1.8195,
      "step": 5240
    },
    {
      "epoch": 1.302002604005208,
      "grad_norm": 0.7661167979240417,
      "learning_rate": 0.00019174944964474912,
      "loss": 1.7991,
      "step": 5250
    },
    {
      "epoch": 1.304482608965218,
      "grad_norm": 0.6782104969024658,
      "learning_rate": 0.0001917184306346032,
      "loss": 1.8504,
      "step": 5260
    },
    {
      "epoch": 1.306962613925228,
      "grad_norm": 0.727779746055603,
      "learning_rate": 0.00019168735594239758,
      "loss": 1.8796,
      "step": 5270
    },
    {
      "epoch": 1.3094426188852377,
      "grad_norm": 0.665866494178772,
      "learning_rate": 0.00019165622558699763,
      "loss": 1.8183,
      "step": 5280
    },
    {
      "epoch": 1.3119226238452477,
      "grad_norm": 0.6575871706008911,
      "learning_rate": 0.00019162503958730255,
      "loss": 1.7832,
      "step": 5290
    },
    {
      "epoch": 1.3144026288052575,
      "grad_norm": 0.6850647926330566,
      "learning_rate": 0.00019159379796224524,
      "loss": 1.8544,
      "step": 5300
    },
    {
      "epoch": 1.3168826337652675,
      "grad_norm": 0.6749188303947449,
      "learning_rate": 0.0001915625007307925,
      "loss": 1.805,
      "step": 5310
    },
    {
      "epoch": 1.3193626387252775,
      "grad_norm": 0.6634512543678284,
      "learning_rate": 0.00019153114791194473,
      "loss": 1.8462,
      "step": 5320
    },
    {
      "epoch": 1.3218426436852875,
      "grad_norm": 0.6670765280723572,
      "learning_rate": 0.0001914997395247362,
      "loss": 1.8249,
      "step": 5330
    },
    {
      "epoch": 1.3243226486452973,
      "grad_norm": 0.6884291768074036,
      "learning_rate": 0.00019146827558823488,
      "loss": 1.8354,
      "step": 5340
    },
    {
      "epoch": 1.3268026536053072,
      "grad_norm": 0.7221989035606384,
      "learning_rate": 0.00019143675612154245,
      "loss": 1.7939,
      "step": 5350
    },
    {
      "epoch": 1.329282658565317,
      "grad_norm": 0.6698116660118103,
      "learning_rate": 0.00019140518114379434,
      "loss": 1.8188,
      "step": 5360
    },
    {
      "epoch": 1.331762663525327,
      "grad_norm": 0.7133959531784058,
      "learning_rate": 0.00019137355067415956,
      "loss": 1.8009,
      "step": 5370
    },
    {
      "epoch": 1.334242668485337,
      "grad_norm": 0.6672759056091309,
      "learning_rate": 0.000191341864731841,
      "loss": 1.7672,
      "step": 5380
    },
    {
      "epoch": 1.336722673445347,
      "grad_norm": 0.6286668181419373,
      "learning_rate": 0.00019131012333607506,
      "loss": 1.8311,
      "step": 5390
    },
    {
      "epoch": 1.3392026784053568,
      "grad_norm": 0.6913756728172302,
      "learning_rate": 0.00019127832650613189,
      "loss": 1.8319,
      "step": 5400
    },
    {
      "epoch": 1.3416826833653668,
      "grad_norm": 0.7492555379867554,
      "learning_rate": 0.0001912464742613153,
      "loss": 1.8288,
      "step": 5410
    },
    {
      "epoch": 1.3441626883253766,
      "grad_norm": 0.7322856187820435,
      "learning_rate": 0.00019121456662096268,
      "loss": 1.773,
      "step": 5420
    },
    {
      "epoch": 1.3466426932853865,
      "grad_norm": 0.6672503352165222,
      "learning_rate": 0.00019118260360444515,
      "loss": 1.8674,
      "step": 5430
    },
    {
      "epoch": 1.3491226982453965,
      "grad_norm": 0.73240065574646,
      "learning_rate": 0.00019115058523116733,
      "loss": 1.8439,
      "step": 5440
    },
    {
      "epoch": 1.3516027032054065,
      "grad_norm": 0.6532642245292664,
      "learning_rate": 0.0001911185115205675,
      "loss": 1.8261,
      "step": 5450
    },
    {
      "epoch": 1.3540827081654163,
      "grad_norm": 0.7535740733146667,
      "learning_rate": 0.00019108638249211758,
      "loss": 1.8175,
      "step": 5460
    },
    {
      "epoch": 1.3565627131254263,
      "grad_norm": 0.6851351261138916,
      "learning_rate": 0.000191054198165323,
      "loss": 1.7735,
      "step": 5470
    },
    {
      "epoch": 1.359042718085436,
      "grad_norm": 0.7507172226905823,
      "learning_rate": 0.00019102195855972287,
      "loss": 1.8291,
      "step": 5480
    },
    {
      "epoch": 1.361522723045446,
      "grad_norm": 0.6462118625640869,
      "learning_rate": 0.00019098966369488965,
      "loss": 1.8241,
      "step": 5490
    },
    {
      "epoch": 1.364002728005456,
      "grad_norm": 0.7370539307594299,
      "learning_rate": 0.0001909573135904296,
      "loss": 1.8278,
      "step": 5500
    },
    {
      "epoch": 1.3664827329654659,
      "grad_norm": 0.6809933185577393,
      "learning_rate": 0.00019092490826598234,
      "loss": 1.8321,
      "step": 5510
    },
    {
      "epoch": 1.3689627379254758,
      "grad_norm": 0.6588488221168518,
      "learning_rate": 0.0001908924477412211,
      "loss": 1.763,
      "step": 5520
    },
    {
      "epoch": 1.3714427428854858,
      "grad_norm": 0.7076486349105835,
      "learning_rate": 0.0001908599320358526,
      "loss": 1.8042,
      "step": 5530
    },
    {
      "epoch": 1.3739227478454956,
      "grad_norm": 0.6753181219100952,
      "learning_rate": 0.00019082736116961697,
      "loss": 1.8308,
      "step": 5540
    },
    {
      "epoch": 1.3764027528055056,
      "grad_norm": 0.6668750643730164,
      "learning_rate": 0.00019079473516228802,
      "loss": 1.8132,
      "step": 5550
    },
    {
      "epoch": 1.3788827577655156,
      "grad_norm": 0.7110525369644165,
      "learning_rate": 0.00019076205403367285,
      "loss": 1.8847,
      "step": 5560
    },
    {
      "epoch": 1.3813627627255254,
      "grad_norm": 1.0364247560501099,
      "learning_rate": 0.00019072931780361217,
      "loss": 1.8394,
      "step": 5570
    },
    {
      "epoch": 1.3838427676855354,
      "grad_norm": 0.7233348488807678,
      "learning_rate": 0.00019069652649198005,
      "loss": 1.8338,
      "step": 5580
    },
    {
      "epoch": 1.3863227726455452,
      "grad_norm": 0.8322575688362122,
      "learning_rate": 0.00019066368011868401,
      "loss": 1.7853,
      "step": 5590
    },
    {
      "epoch": 1.3888027776055551,
      "grad_norm": 0.6443812251091003,
      "learning_rate": 0.000190630778703665,
      "loss": 1.8372,
      "step": 5600
    },
    {
      "epoch": 1.3912827825655651,
      "grad_norm": 0.8004515171051025,
      "learning_rate": 0.00019059782226689743,
      "loss": 1.8277,
      "step": 5610
    },
    {
      "epoch": 1.3937627875255751,
      "grad_norm": 0.7190122008323669,
      "learning_rate": 0.00019056481082838905,
      "loss": 1.8278,
      "step": 5620
    },
    {
      "epoch": 1.396242792485585,
      "grad_norm": 0.6689620018005371,
      "learning_rate": 0.00019053174440818106,
      "loss": 1.863,
      "step": 5630
    },
    {
      "epoch": 1.398722797445595,
      "grad_norm": 0.6717751622200012,
      "learning_rate": 0.000190498623026348,
      "loss": 1.7858,
      "step": 5640
    },
    {
      "epoch": 1.4012028024056047,
      "grad_norm": 0.6776305437088013,
      "learning_rate": 0.00019046544670299777,
      "loss": 1.8747,
      "step": 5650
    },
    {
      "epoch": 1.4036828073656147,
      "grad_norm": 0.6982606053352356,
      "learning_rate": 0.0001904322154582717,
      "loss": 1.793,
      "step": 5660
    },
    {
      "epoch": 1.4061628123256247,
      "grad_norm": 0.7391812205314636,
      "learning_rate": 0.00019039892931234435,
      "loss": 1.8151,
      "step": 5670
    },
    {
      "epoch": 1.4086428172856347,
      "grad_norm": 0.7182745933532715,
      "learning_rate": 0.00019036558828542367,
      "loss": 1.7983,
      "step": 5680
    },
    {
      "epoch": 1.4111228222456444,
      "grad_norm": 0.7192206978797913,
      "learning_rate": 0.00019033219239775095,
      "loss": 1.7845,
      "step": 5690
    },
    {
      "epoch": 1.4136028272056544,
      "grad_norm": 0.7159098386764526,
      "learning_rate": 0.00019029874166960075,
      "loss": 1.7565,
      "step": 5700
    },
    {
      "epoch": 1.4160828321656642,
      "grad_norm": 0.7049682140350342,
      "learning_rate": 0.00019026523612128096,
      "loss": 1.872,
      "step": 5710
    },
    {
      "epoch": 1.4185628371256742,
      "grad_norm": 0.6793913841247559,
      "learning_rate": 0.00019023167577313264,
      "loss": 1.8182,
      "step": 5720
    },
    {
      "epoch": 1.4210428420856842,
      "grad_norm": 0.7400872707366943,
      "learning_rate": 0.00019019806064553033,
      "loss": 1.8298,
      "step": 5730
    },
    {
      "epoch": 1.4235228470456942,
      "grad_norm": 0.6520139575004578,
      "learning_rate": 0.00019016439075888162,
      "loss": 1.8588,
      "step": 5740
    },
    {
      "epoch": 1.426002852005704,
      "grad_norm": 0.6989015936851501,
      "learning_rate": 0.00019013066613362742,
      "loss": 1.7805,
      "step": 5750
    },
    {
      "epoch": 1.428482856965714,
      "grad_norm": 0.6831812262535095,
      "learning_rate": 0.0001900968867902419,
      "loss": 1.8152,
      "step": 5760
    },
    {
      "epoch": 1.4309628619257238,
      "grad_norm": 0.7414357662200928,
      "learning_rate": 0.00019006305274923247,
      "loss": 1.767,
      "step": 5770
    },
    {
      "epoch": 1.4334428668857337,
      "grad_norm": 0.6898692846298218,
      "learning_rate": 0.00019002916403113962,
      "loss": 1.7556,
      "step": 5780
    },
    {
      "epoch": 1.4359228718457437,
      "grad_norm": 0.6934981942176819,
      "learning_rate": 0.00018999522065653714,
      "loss": 1.8059,
      "step": 5790
    },
    {
      "epoch": 1.4384028768057537,
      "grad_norm": 0.7213485240936279,
      "learning_rate": 0.00018996122264603203,
      "loss": 1.7809,
      "step": 5800
    },
    {
      "epoch": 1.4408828817657635,
      "grad_norm": 0.6892271637916565,
      "learning_rate": 0.00018992717002026433,
      "loss": 1.8452,
      "step": 5810
    },
    {
      "epoch": 1.4433628867257735,
      "grad_norm": 0.7585854530334473,
      "learning_rate": 0.00018989306279990737,
      "loss": 1.8235,
      "step": 5820
    },
    {
      "epoch": 1.4458428916857833,
      "grad_norm": 0.6786388158798218,
      "learning_rate": 0.00018985890100566753,
      "loss": 1.8044,
      "step": 5830
    },
    {
      "epoch": 1.4483228966457933,
      "grad_norm": 0.6582244038581848,
      "learning_rate": 0.00018982468465828442,
      "loss": 1.8139,
      "step": 5840
    },
    {
      "epoch": 1.4508029016058033,
      "grad_norm": 0.6639909744262695,
      "learning_rate": 0.00018979041377853068,
      "loss": 1.8363,
      "step": 5850
    },
    {
      "epoch": 1.453282906565813,
      "grad_norm": 0.728732168674469,
      "learning_rate": 0.0001897560883872121,
      "loss": 1.8377,
      "step": 5860
    },
    {
      "epoch": 1.455762911525823,
      "grad_norm": 0.7115894556045532,
      "learning_rate": 0.00018972170850516747,
      "loss": 1.8009,
      "step": 5870
    },
    {
      "epoch": 1.458242916485833,
      "grad_norm": 0.7272158861160278,
      "learning_rate": 0.00018968727415326884,
      "loss": 1.8232,
      "step": 5880
    },
    {
      "epoch": 1.4607229214458428,
      "grad_norm": 0.6763330101966858,
      "learning_rate": 0.0001896527853524212,
      "loss": 1.8064,
      "step": 5890
    },
    {
      "epoch": 1.4632029264058528,
      "grad_norm": 0.6789953708648682,
      "learning_rate": 0.0001896182421235626,
      "loss": 1.8653,
      "step": 5900
    },
    {
      "epoch": 1.4656829313658628,
      "grad_norm": 0.7505097389221191,
      "learning_rate": 0.0001895836444876642,
      "loss": 1.8533,
      "step": 5910
    },
    {
      "epoch": 1.4681629363258726,
      "grad_norm": 0.7787876129150391,
      "learning_rate": 0.0001895489924657301,
      "loss": 1.7521,
      "step": 5920
    },
    {
      "epoch": 1.4706429412858826,
      "grad_norm": 0.7812589406967163,
      "learning_rate": 0.00018951428607879753,
      "loss": 1.8075,
      "step": 5930
    },
    {
      "epoch": 1.4731229462458926,
      "grad_norm": 0.695969820022583,
      "learning_rate": 0.00018947952534793661,
      "loss": 1.8117,
      "step": 5940
    },
    {
      "epoch": 1.4756029512059023,
      "grad_norm": 0.680429995059967,
      "learning_rate": 0.00018944471029425054,
      "loss": 1.8116,
      "step": 5950
    },
    {
      "epoch": 1.4780829561659123,
      "grad_norm": 0.6775833964347839,
      "learning_rate": 0.00018940984093887542,
      "loss": 1.7969,
      "step": 5960
    },
    {
      "epoch": 1.4805629611259223,
      "grad_norm": 0.6806792616844177,
      "learning_rate": 0.0001893749173029804,
      "loss": 1.8317,
      "step": 5970
    },
    {
      "epoch": 1.4830429660859321,
      "grad_norm": 0.7450188398361206,
      "learning_rate": 0.00018933993940776752,
      "loss": 1.7515,
      "step": 5980
    },
    {
      "epoch": 1.485522971045942,
      "grad_norm": 0.6525201201438904,
      "learning_rate": 0.00018930490727447178,
      "loss": 1.813,
      "step": 5990
    },
    {
      "epoch": 1.4880029760059519,
      "grad_norm": 0.6384407877922058,
      "learning_rate": 0.00018926982092436117,
      "loss": 1.7391,
      "step": 6000
    },
    {
      "epoch": 1.4904829809659619,
      "grad_norm": 0.706158459186554,
      "learning_rate": 0.00018923468037873646,
      "loss": 1.8431,
      "step": 6010
    },
    {
      "epoch": 1.4929629859259719,
      "grad_norm": 0.6749498248100281,
      "learning_rate": 0.00018919948565893142,
      "loss": 1.7907,
      "step": 6020
    },
    {
      "epoch": 1.4954429908859819,
      "grad_norm": 0.7001073360443115,
      "learning_rate": 0.00018916423678631272,
      "loss": 1.7967,
      "step": 6030
    },
    {
      "epoch": 1.4979229958459916,
      "grad_norm": 0.6850577592849731,
      "learning_rate": 0.00018912893378227985,
      "loss": 1.8312,
      "step": 6040
    },
    {
      "epoch": 1.5004030008060016,
      "grad_norm": 0.7084920406341553,
      "learning_rate": 0.00018909357666826518,
      "loss": 1.7268,
      "step": 6050
    },
    {
      "epoch": 1.5028830057660114,
      "grad_norm": 0.701327919960022,
      "learning_rate": 0.00018905816546573397,
      "loss": 1.8329,
      "step": 6060
    },
    {
      "epoch": 1.5053630107260214,
      "grad_norm": 0.7240101099014282,
      "learning_rate": 0.00018902270019618422,
      "loss": 1.7809,
      "step": 6070
    },
    {
      "epoch": 1.5078430156860314,
      "grad_norm": 0.6603263020515442,
      "learning_rate": 0.0001889871808811469,
      "loss": 1.8182,
      "step": 6080
    },
    {
      "epoch": 1.5103230206460414,
      "grad_norm": 0.7466841340065002,
      "learning_rate": 0.00018895160754218561,
      "loss": 1.8238,
      "step": 6090
    },
    {
      "epoch": 1.5128030256060512,
      "grad_norm": 0.7204167246818542,
      "learning_rate": 0.00018891598020089693,
      "loss": 1.8201,
      "step": 6100
    },
    {
      "epoch": 1.5152830305660612,
      "grad_norm": 0.6652133464813232,
      "learning_rate": 0.00018888029887891014,
      "loss": 1.8013,
      "step": 6110
    },
    {
      "epoch": 1.517763035526071,
      "grad_norm": 0.6607919335365295,
      "learning_rate": 0.00018884456359788724,
      "loss": 1.7745,
      "step": 6120
    },
    {
      "epoch": 1.520243040486081,
      "grad_norm": 0.7172577381134033,
      "learning_rate": 0.0001888087743795231,
      "loss": 1.8148,
      "step": 6130
    },
    {
      "epoch": 1.522723045446091,
      "grad_norm": 0.7216603755950928,
      "learning_rate": 0.0001887729312455452,
      "loss": 1.7853,
      "step": 6140
    },
    {
      "epoch": 1.525203050406101,
      "grad_norm": 0.7555466294288635,
      "learning_rate": 0.0001887370342177139,
      "loss": 1.7995,
      "step": 6150
    },
    {
      "epoch": 1.5276830553661107,
      "grad_norm": 0.683175265789032,
      "learning_rate": 0.00018870108331782217,
      "loss": 1.8145,
      "step": 6160
    },
    {
      "epoch": 1.5301630603261207,
      "grad_norm": 0.6748200058937073,
      "learning_rate": 0.00018866507856769575,
      "loss": 1.8319,
      "step": 6170
    },
    {
      "epoch": 1.5326430652861305,
      "grad_norm": 0.8977167010307312,
      "learning_rate": 0.00018862901998919305,
      "loss": 1.8121,
      "step": 6180
    },
    {
      "epoch": 1.5351230702461405,
      "grad_norm": 0.732475757598877,
      "learning_rate": 0.0001885929076042051,
      "loss": 1.8169,
      "step": 6190
    },
    {
      "epoch": 1.5376030752061505,
      "grad_norm": 0.6811177134513855,
      "learning_rate": 0.00018855674143465566,
      "loss": 1.8847,
      "step": 6200
    },
    {
      "epoch": 1.5400830801661605,
      "grad_norm": 0.7238559722900391,
      "learning_rate": 0.00018852052150250122,
      "loss": 1.8222,
      "step": 6210
    },
    {
      "epoch": 1.5425630851261702,
      "grad_norm": 0.748440146446228,
      "learning_rate": 0.00018848424782973074,
      "loss": 1.8437,
      "step": 6220
    },
    {
      "epoch": 1.54504309008618,
      "grad_norm": 0.6953113079071045,
      "learning_rate": 0.00018844792043836589,
      "loss": 1.7924,
      "step": 6230
    },
    {
      "epoch": 1.54752309504619,
      "grad_norm": 0.7056795358657837,
      "learning_rate": 0.00018841153935046098,
      "loss": 1.8471,
      "step": 6240
    },
    {
      "epoch": 1.5500031000062,
      "grad_norm": 0.7412731647491455,
      "learning_rate": 0.0001883751045881029,
      "loss": 1.749,
      "step": 6250
    },
    {
      "epoch": 1.55248310496621,
      "grad_norm": 0.7239865660667419,
      "learning_rate": 0.00018833861617341107,
      "loss": 1.8143,
      "step": 6260
    },
    {
      "epoch": 1.55496310992622,
      "grad_norm": 0.708827555179596,
      "learning_rate": 0.00018830207412853756,
      "loss": 1.8153,
      "step": 6270
    },
    {
      "epoch": 1.5574431148862298,
      "grad_norm": 0.6861132383346558,
      "learning_rate": 0.00018826547847566693,
      "loss": 1.8874,
      "step": 6280
    },
    {
      "epoch": 1.5599231198462395,
      "grad_norm": 0.6977689862251282,
      "learning_rate": 0.00018822882923701638,
      "loss": 1.8395,
      "step": 6290
    },
    {
      "epoch": 1.5624031248062495,
      "grad_norm": 0.6514766216278076,
      "learning_rate": 0.0001881921264348355,
      "loss": 1.7926,
      "step": 6300
    },
    {
      "epoch": 1.5648831297662595,
      "grad_norm": 0.6926382780075073,
      "learning_rate": 0.00018815537009140657,
      "loss": 1.7719,
      "step": 6310
    },
    {
      "epoch": 1.5673631347262695,
      "grad_norm": 0.7346818447113037,
      "learning_rate": 0.00018811856022904423,
      "loss": 1.819,
      "step": 6320
    },
    {
      "epoch": 1.5698431396862795,
      "grad_norm": 0.690940260887146,
      "learning_rate": 0.00018808169687009572,
      "loss": 1.8205,
      "step": 6330
    },
    {
      "epoch": 1.5723231446462893,
      "grad_norm": 0.6950753331184387,
      "learning_rate": 0.00018804478003694067,
      "loss": 1.7586,
      "step": 6340
    },
    {
      "epoch": 1.574803149606299,
      "grad_norm": 0.6781951189041138,
      "learning_rate": 0.00018800780975199115,
      "loss": 1.8386,
      "step": 6350
    },
    {
      "epoch": 1.577283154566309,
      "grad_norm": 0.7350020408630371,
      "learning_rate": 0.00018797078603769184,
      "loss": 1.894,
      "step": 6360
    },
    {
      "epoch": 1.579763159526319,
      "grad_norm": 0.7319828867912292,
      "learning_rate": 0.00018793370891651972,
      "loss": 1.8185,
      "step": 6370
    },
    {
      "epoch": 1.582243164486329,
      "grad_norm": 0.6983395218849182,
      "learning_rate": 0.0001878965784109842,
      "loss": 1.8203,
      "step": 6380
    },
    {
      "epoch": 1.5847231694463388,
      "grad_norm": 0.6651434898376465,
      "learning_rate": 0.00018785939454362717,
      "loss": 1.7963,
      "step": 6390
    },
    {
      "epoch": 1.5872031744063488,
      "grad_norm": 0.6983841061592102,
      "learning_rate": 0.00018782215733702286,
      "loss": 1.8204,
      "step": 6400
    },
    {
      "epoch": 1.5896831793663586,
      "grad_norm": 0.7472341656684875,
      "learning_rate": 0.0001877848668137779,
      "loss": 1.8136,
      "step": 6410
    },
    {
      "epoch": 1.5921631843263686,
      "grad_norm": 0.7456172108650208,
      "learning_rate": 0.00018774752299653127,
      "loss": 1.863,
      "step": 6420
    },
    {
      "epoch": 1.5946431892863786,
      "grad_norm": 0.6850917935371399,
      "learning_rate": 0.00018771012590795434,
      "loss": 1.7964,
      "step": 6430
    },
    {
      "epoch": 1.5971231942463886,
      "grad_norm": 0.6574013233184814,
      "learning_rate": 0.0001876726755707508,
      "loss": 1.8635,
      "step": 6440
    },
    {
      "epoch": 1.5996031992063984,
      "grad_norm": 0.7072936296463013,
      "learning_rate": 0.00018763517200765664,
      "loss": 1.8926,
      "step": 6450
    },
    {
      "epoch": 1.6020832041664084,
      "grad_norm": 0.6977803707122803,
      "learning_rate": 0.00018759761524144022,
      "loss": 1.8293,
      "step": 6460
    },
    {
      "epoch": 1.6045632091264181,
      "grad_norm": 0.7164632678031921,
      "learning_rate": 0.00018756000529490216,
      "loss": 1.7896,
      "step": 6470
    },
    {
      "epoch": 1.6070432140864281,
      "grad_norm": 0.7482587695121765,
      "learning_rate": 0.00018752234219087538,
      "loss": 1.8141,
      "step": 6480
    },
    {
      "epoch": 1.6095232190464381,
      "grad_norm": 0.6833829879760742,
      "learning_rate": 0.00018748462595222508,
      "loss": 1.7785,
      "step": 6490
    },
    {
      "epoch": 1.6120032240064481,
      "grad_norm": 0.7720143795013428,
      "learning_rate": 0.00018744685660184866,
      "loss": 1.7982,
      "step": 6500
    },
    {
      "epoch": 1.614483228966458,
      "grad_norm": 0.6675408482551575,
      "learning_rate": 0.0001874090341626759,
      "loss": 1.7722,
      "step": 6510
    },
    {
      "epoch": 1.616963233926468,
      "grad_norm": 0.6653440594673157,
      "learning_rate": 0.00018737115865766863,
      "loss": 1.8461,
      "step": 6520
    },
    {
      "epoch": 1.6194432388864777,
      "grad_norm": 0.7201128602027893,
      "learning_rate": 0.00018733323010982104,
      "loss": 1.8344,
      "step": 6530
    },
    {
      "epoch": 1.6219232438464877,
      "grad_norm": 0.7022073864936829,
      "learning_rate": 0.00018729524854215943,
      "loss": 1.8033,
      "step": 6540
    },
    {
      "epoch": 1.6244032488064977,
      "grad_norm": 0.6909770965576172,
      "learning_rate": 0.0001872572139777424,
      "loss": 1.7777,
      "step": 6550
    },
    {
      "epoch": 1.6268832537665077,
      "grad_norm": 0.6666807532310486,
      "learning_rate": 0.00018721912643966055,
      "loss": 1.7996,
      "step": 6560
    },
    {
      "epoch": 1.6293632587265174,
      "grad_norm": 0.7246910929679871,
      "learning_rate": 0.0001871809859510368,
      "loss": 1.81,
      "step": 6570
    },
    {
      "epoch": 1.6318432636865272,
      "grad_norm": 0.6781728863716125,
      "learning_rate": 0.00018714279253502616,
      "loss": 1.7576,
      "step": 6580
    },
    {
      "epoch": 1.6343232686465372,
      "grad_norm": 0.6999672651290894,
      "learning_rate": 0.00018710454621481576,
      "loss": 1.8126,
      "step": 6590
    },
    {
      "epoch": 1.6368032736065472,
      "grad_norm": 0.703766405582428,
      "learning_rate": 0.00018706624701362483,
      "loss": 1.8109,
      "step": 6600
    },
    {
      "epoch": 1.6392832785665572,
      "grad_norm": 0.7479855418205261,
      "learning_rate": 0.00018702789495470478,
      "loss": 1.7835,
      "step": 6610
    },
    {
      "epoch": 1.6417632835265672,
      "grad_norm": 0.7109487056732178,
      "learning_rate": 0.00018698949006133904,
      "loss": 1.8078,
      "step": 6620
    },
    {
      "epoch": 1.644243288486577,
      "grad_norm": 0.7246449589729309,
      "learning_rate": 0.00018695103235684312,
      "loss": 1.7999,
      "step": 6630
    },
    {
      "epoch": 1.6467232934465867,
      "grad_norm": 0.6888371706008911,
      "learning_rate": 0.00018691252186456465,
      "loss": 1.8556,
      "step": 6640
    },
    {
      "epoch": 1.6492032984065967,
      "grad_norm": 0.6614171862602234,
      "learning_rate": 0.00018687395860788323,
      "loss": 1.8626,
      "step": 6650
    },
    {
      "epoch": 1.6516833033666067,
      "grad_norm": 0.6451037526130676,
      "learning_rate": 0.00018683534261021057,
      "loss": 1.8507,
      "step": 6660
    },
    {
      "epoch": 1.6541633083266167,
      "grad_norm": 0.7350413203239441,
      "learning_rate": 0.0001867966738949903,
      "loss": 1.8001,
      "step": 6670
    },
    {
      "epoch": 1.6566433132866267,
      "grad_norm": 0.761847972869873,
      "learning_rate": 0.00018675795248569813,
      "loss": 1.7653,
      "step": 6680
    },
    {
      "epoch": 1.6591233182466365,
      "grad_norm": 0.6977943778038025,
      "learning_rate": 0.0001867191784058418,
      "loss": 1.7868,
      "step": 6690
    },
    {
      "epoch": 1.6616033232066463,
      "grad_norm": 0.6676158308982849,
      "learning_rate": 0.00018668035167896088,
      "loss": 1.854,
      "step": 6700
    },
    {
      "epoch": 1.6640833281666563,
      "grad_norm": 0.7719148397445679,
      "learning_rate": 0.00018664147232862706,
      "loss": 1.7553,
      "step": 6710
    },
    {
      "epoch": 1.6665633331266663,
      "grad_norm": 0.6652414202690125,
      "learning_rate": 0.00018660254037844388,
      "loss": 1.8375,
      "step": 6720
    },
    {
      "epoch": 1.6690433380866763,
      "grad_norm": 0.7422439455986023,
      "learning_rate": 0.00018656355585204684,
      "loss": 1.8659,
      "step": 6730
    },
    {
      "epoch": 1.671523343046686,
      "grad_norm": 0.6858110427856445,
      "learning_rate": 0.00018652451877310336,
      "loss": 1.8021,
      "step": 6740
    },
    {
      "epoch": 1.674003348006696,
      "grad_norm": 0.6965064406394958,
      "learning_rate": 0.0001864854291653128,
      "loss": 1.8454,
      "step": 6750
    },
    {
      "epoch": 1.6764833529667058,
      "grad_norm": 0.7218753695487976,
      "learning_rate": 0.00018644628705240636,
      "loss": 1.8381,
      "step": 6760
    },
    {
      "epoch": 1.6789633579267158,
      "grad_norm": 0.6853536367416382,
      "learning_rate": 0.0001864070924581471,
      "loss": 1.7834,
      "step": 6770
    },
    {
      "epoch": 1.6814433628867258,
      "grad_norm": 0.7122074365615845,
      "learning_rate": 0.00018636784540633,
      "loss": 1.765,
      "step": 6780
    },
    {
      "epoch": 1.6839233678467358,
      "grad_norm": 0.7236666083335876,
      "learning_rate": 0.00018632854592078184,
      "loss": 1.8046,
      "step": 6790
    },
    {
      "epoch": 1.6864033728067456,
      "grad_norm": 0.6996268033981323,
      "learning_rate": 0.0001862891940253613,
      "loss": 1.8089,
      "step": 6800
    },
    {
      "epoch": 1.6888833777667556,
      "grad_norm": 0.7394871711730957,
      "learning_rate": 0.0001862497897439588,
      "loss": 1.7953,
      "step": 6810
    },
    {
      "epoch": 1.6913633827267653,
      "grad_norm": 0.6939400434494019,
      "learning_rate": 0.00018621033310049656,
      "loss": 1.8292,
      "step": 6820
    },
    {
      "epoch": 1.6938433876867753,
      "grad_norm": 0.6782166957855225,
      "learning_rate": 0.00018617082411892866,
      "loss": 1.8001,
      "step": 6830
    },
    {
      "epoch": 1.6963233926467853,
      "grad_norm": 0.7154285907745361,
      "learning_rate": 0.00018613126282324092,
      "loss": 1.7969,
      "step": 6840
    },
    {
      "epoch": 1.6988033976067953,
      "grad_norm": 0.759038507938385,
      "learning_rate": 0.00018609164923745087,
      "loss": 1.7746,
      "step": 6850
    },
    {
      "epoch": 1.701283402566805,
      "grad_norm": 0.6908287405967712,
      "learning_rate": 0.0001860519833856079,
      "loss": 1.7682,
      "step": 6860
    },
    {
      "epoch": 1.703763407526815,
      "grad_norm": 0.7424526214599609,
      "learning_rate": 0.00018601226529179306,
      "loss": 1.8196,
      "step": 6870
    },
    {
      "epoch": 1.7062434124868249,
      "grad_norm": 0.7291129231452942,
      "learning_rate": 0.00018597249498011903,
      "loss": 1.8359,
      "step": 6880
    },
    {
      "epoch": 1.7087234174468349,
      "grad_norm": 0.7176723480224609,
      "learning_rate": 0.0001859326724747304,
      "loss": 1.8016,
      "step": 6890
    },
    {
      "epoch": 1.7112034224068449,
      "grad_norm": 0.7186075448989868,
      "learning_rate": 0.0001858927977998032,
      "loss": 1.7719,
      "step": 6900
    },
    {
      "epoch": 1.7136834273668549,
      "grad_norm": 0.6708627939224243,
      "learning_rate": 0.00018585287097954534,
      "loss": 1.8005,
      "step": 6910
    },
    {
      "epoch": 1.7161634323268646,
      "grad_norm": 0.6831984519958496,
      "learning_rate": 0.0001858128920381963,
      "loss": 1.7962,
      "step": 6920
    },
    {
      "epoch": 1.7186434372868744,
      "grad_norm": 0.7038664221763611,
      "learning_rate": 0.00018577286100002723,
      "loss": 1.761,
      "step": 6930
    },
    {
      "epoch": 1.7211234422468844,
      "grad_norm": 0.693360447883606,
      "learning_rate": 0.00018573277788934084,
      "loss": 1.7751,
      "step": 6940
    },
    {
      "epoch": 1.7236034472068944,
      "grad_norm": 0.7013189792633057,
      "learning_rate": 0.00018569264273047146,
      "loss": 1.8005,
      "step": 6950
    },
    {
      "epoch": 1.7260834521669044,
      "grad_norm": 0.8188039064407349,
      "learning_rate": 0.00018565245554778517,
      "loss": 1.8367,
      "step": 6960
    },
    {
      "epoch": 1.7285634571269144,
      "grad_norm": 0.7016475796699524,
      "learning_rate": 0.00018561221636567943,
      "loss": 1.7609,
      "step": 6970
    },
    {
      "epoch": 1.7310434620869242,
      "grad_norm": 0.7186803221702576,
      "learning_rate": 0.0001855719252085834,
      "loss": 1.8397,
      "step": 6980
    },
    {
      "epoch": 1.733523467046934,
      "grad_norm": 0.7351244688034058,
      "learning_rate": 0.00018553158210095772,
      "loss": 1.7872,
      "step": 6990
    },
    {
      "epoch": 1.736003472006944,
      "grad_norm": 0.7762275338172913,
      "learning_rate": 0.00018549118706729468,
      "loss": 1.8339,
      "step": 7000
    },
    {
      "epoch": 1.738483476966954,
      "grad_norm": 0.7965319156646729,
      "learning_rate": 0.00018545074013211793,
      "loss": 1.7642,
      "step": 7010
    },
    {
      "epoch": 1.740963481926964,
      "grad_norm": 0.6989258527755737,
      "learning_rate": 0.00018541024131998274,
      "loss": 1.7861,
      "step": 7020
    },
    {
      "epoch": 1.743443486886974,
      "grad_norm": 0.7284238338470459,
      "learning_rate": 0.0001853696906554759,
      "loss": 1.7762,
      "step": 7030
    },
    {
      "epoch": 1.7459234918469837,
      "grad_norm": 0.6847079396247864,
      "learning_rate": 0.00018532908816321558,
      "loss": 1.8342,
      "step": 7040
    },
    {
      "epoch": 1.7484034968069935,
      "grad_norm": 0.6576705574989319,
      "learning_rate": 0.00018528843386785145,
      "loss": 1.815,
      "step": 7050
    },
    {
      "epoch": 1.7508835017670035,
      "grad_norm": 0.7153476476669312,
      "learning_rate": 0.0001852477277940647,
      "loss": 1.7593,
      "step": 7060
    },
    {
      "epoch": 1.7533635067270135,
      "grad_norm": 0.6890503168106079,
      "learning_rate": 0.00018520696996656788,
      "loss": 1.7855,
      "step": 7070
    },
    {
      "epoch": 1.7558435116870235,
      "grad_norm": 0.7169741988182068,
      "learning_rate": 0.00018516616041010494,
      "loss": 1.7918,
      "step": 7080
    },
    {
      "epoch": 1.7583235166470335,
      "grad_norm": 0.7051769495010376,
      "learning_rate": 0.00018512529914945135,
      "loss": 1.783,
      "step": 7090
    },
    {
      "epoch": 1.7608035216070432,
      "grad_norm": 0.7099260091781616,
      "learning_rate": 0.00018508438620941378,
      "loss": 1.859,
      "step": 7100
    },
    {
      "epoch": 1.763283526567053,
      "grad_norm": 0.733961820602417,
      "learning_rate": 0.00018504342161483047,
      "loss": 1.8326,
      "step": 7110
    },
    {
      "epoch": 1.765763531527063,
      "grad_norm": 0.8449649810791016,
      "learning_rate": 0.0001850024053905709,
      "loss": 1.7666,
      "step": 7120
    },
    {
      "epoch": 1.768243536487073,
      "grad_norm": 0.6947612166404724,
      "learning_rate": 0.00018496133756153602,
      "loss": 1.7784,
      "step": 7130
    },
    {
      "epoch": 1.770723541447083,
      "grad_norm": 0.8235635161399841,
      "learning_rate": 0.0001849202181526579,
      "loss": 1.7541,
      "step": 7140
    },
    {
      "epoch": 1.7732035464070928,
      "grad_norm": 0.7722574472427368,
      "learning_rate": 0.0001848790471889001,
      "loss": 1.8518,
      "step": 7150
    },
    {
      "epoch": 1.7756835513671028,
      "grad_norm": 0.7052320241928101,
      "learning_rate": 0.00018483782469525738,
      "loss": 1.7835,
      "step": 7160
    },
    {
      "epoch": 1.7781635563271125,
      "grad_norm": 0.7227842807769775,
      "learning_rate": 0.0001847965506967559,
      "loss": 1.788,
      "step": 7170
    },
    {
      "epoch": 1.7806435612871225,
      "grad_norm": 0.713796854019165,
      "learning_rate": 0.00018475522521845295,
      "loss": 1.8347,
      "step": 7180
    },
    {
      "epoch": 1.7831235662471325,
      "grad_norm": 0.6927456855773926,
      "learning_rate": 0.00018471384828543718,
      "loss": 1.787,
      "step": 7190
    },
    {
      "epoch": 1.7856035712071425,
      "grad_norm": 0.7317062616348267,
      "learning_rate": 0.00018467241992282843,
      "loss": 1.8165,
      "step": 7200
    },
    {
      "epoch": 1.7880835761671523,
      "grad_norm": 0.7170538306236267,
      "learning_rate": 0.00018463094015577775,
      "loss": 1.7905,
      "step": 7210
    },
    {
      "epoch": 1.7905635811271623,
      "grad_norm": 0.6815361380577087,
      "learning_rate": 0.00018458940900946739,
      "loss": 1.8998,
      "step": 7220
    },
    {
      "epoch": 1.793043586087172,
      "grad_norm": 0.7838888764381409,
      "learning_rate": 0.0001845478265091109,
      "loss": 1.7994,
      "step": 7230
    },
    {
      "epoch": 1.795523591047182,
      "grad_norm": 0.7330991625785828,
      "learning_rate": 0.00018450619267995283,
      "loss": 1.8262,
      "step": 7240
    },
    {
      "epoch": 1.798003596007192,
      "grad_norm": 0.7867507934570312,
      "learning_rate": 0.00018446450754726907,
      "loss": 1.7856,
      "step": 7250
    },
    {
      "epoch": 1.800483600967202,
      "grad_norm": 0.6976866126060486,
      "learning_rate": 0.00018442277113636654,
      "loss": 1.837,
      "step": 7260
    },
    {
      "epoch": 1.8029636059272118,
      "grad_norm": 0.7053040266036987,
      "learning_rate": 0.0001843809834725833,
      "loss": 1.8075,
      "step": 7270
    },
    {
      "epoch": 1.8054436108872218,
      "grad_norm": 0.7264782190322876,
      "learning_rate": 0.0001843391445812886,
      "loss": 1.8392,
      "step": 7280
    },
    {
      "epoch": 1.8079236158472316,
      "grad_norm": 0.7384432554244995,
      "learning_rate": 0.00018429725448788266,
      "loss": 1.7901,
      "step": 7290
    },
    {
      "epoch": 1.8104036208072416,
      "grad_norm": 0.7113457918167114,
      "learning_rate": 0.00018425531321779697,
      "loss": 1.859,
      "step": 7300
    },
    {
      "epoch": 1.8128836257672516,
      "grad_norm": 0.693682074546814,
      "learning_rate": 0.0001842133207964939,
      "loss": 1.8757,
      "step": 7310
    },
    {
      "epoch": 1.8153636307272616,
      "grad_norm": 0.696943998336792,
      "learning_rate": 0.00018417127724946702,
      "loss": 1.7946,
      "step": 7320
    },
    {
      "epoch": 1.8178436356872714,
      "grad_norm": 0.6696971654891968,
      "learning_rate": 0.0001841291826022408,
      "loss": 1.7775,
      "step": 7330
    },
    {
      "epoch": 1.8203236406472811,
      "grad_norm": 0.6955931782722473,
      "learning_rate": 0.00018408703688037087,
      "loss": 1.7338,
      "step": 7340
    },
    {
      "epoch": 1.8228036456072911,
      "grad_norm": 0.8137204051017761,
      "learning_rate": 0.0001840448401094438,
      "loss": 1.8533,
      "step": 7350
    },
    {
      "epoch": 1.8252836505673011,
      "grad_norm": 0.6938469409942627,
      "learning_rate": 0.00018400259231507717,
      "loss": 1.818,
      "step": 7360
    },
    {
      "epoch": 1.8277636555273111,
      "grad_norm": 0.7239459156990051,
      "learning_rate": 0.00018396029352291945,
      "loss": 1.7605,
      "step": 7370
    },
    {
      "epoch": 1.8302436604873211,
      "grad_norm": 0.7191905975341797,
      "learning_rate": 0.00018391794375865024,
      "loss": 1.8303,
      "step": 7380
    },
    {
      "epoch": 1.832723665447331,
      "grad_norm": 0.7267665266990662,
      "learning_rate": 0.0001838755430479799,
      "loss": 1.818,
      "step": 7390
    },
    {
      "epoch": 1.8352036704073407,
      "grad_norm": 0.748570442199707,
      "learning_rate": 0.0001838330914166499,
      "loss": 1.7961,
      "step": 7400
    },
    {
      "epoch": 1.8376836753673507,
      "grad_norm": 0.7724071741104126,
      "learning_rate": 0.00018379058889043246,
      "loss": 1.8515,
      "step": 7410
    },
    {
      "epoch": 1.8401636803273607,
      "grad_norm": 0.750237762928009,
      "learning_rate": 0.0001837480354951308,
      "loss": 1.867,
      "step": 7420
    },
    {
      "epoch": 1.8426436852873707,
      "grad_norm": 0.675631582736969,
      "learning_rate": 0.000183705431256579,
      "loss": 1.7754,
      "step": 7430
    },
    {
      "epoch": 1.8451236902473807,
      "grad_norm": 0.7144594192504883,
      "learning_rate": 0.00018366277620064197,
      "loss": 1.8192,
      "step": 7440
    },
    {
      "epoch": 1.8476036952073904,
      "grad_norm": 0.7249010801315308,
      "learning_rate": 0.00018362007035321556,
      "loss": 1.8547,
      "step": 7450
    },
    {
      "epoch": 1.8500837001674002,
      "grad_norm": 0.7054729461669922,
      "learning_rate": 0.00018357731374022635,
      "loss": 1.8139,
      "step": 7460
    },
    {
      "epoch": 1.8525637051274102,
      "grad_norm": 0.741942286491394,
      "learning_rate": 0.00018353450638763179,
      "loss": 1.7698,
      "step": 7470
    },
    {
      "epoch": 1.8550437100874202,
      "grad_norm": 0.7423719763755798,
      "learning_rate": 0.00018349164832142013,
      "loss": 1.7934,
      "step": 7480
    },
    {
      "epoch": 1.8575237150474302,
      "grad_norm": 0.8158997893333435,
      "learning_rate": 0.00018344873956761045,
      "loss": 1.893,
      "step": 7490
    },
    {
      "epoch": 1.86000372000744,
      "grad_norm": 0.7317871451377869,
      "learning_rate": 0.0001834057801522525,
      "loss": 1.8123,
      "step": 7500
    },
    {
      "epoch": 1.86248372496745,
      "grad_norm": 0.7026477456092834,
      "learning_rate": 0.0001833627701014269,
      "loss": 1.8213,
      "step": 7510
    },
    {
      "epoch": 1.8649637299274597,
      "grad_norm": 0.7163852453231812,
      "learning_rate": 0.0001833197094412449,
      "loss": 1.7753,
      "step": 7520
    },
    {
      "epoch": 1.8674437348874697,
      "grad_norm": 0.7176632285118103,
      "learning_rate": 0.00018327659819784858,
      "loss": 1.8194,
      "step": 7530
    },
    {
      "epoch": 1.8699237398474797,
      "grad_norm": 0.7520141005516052,
      "learning_rate": 0.00018323343639741068,
      "loss": 1.7382,
      "step": 7540
    },
    {
      "epoch": 1.8724037448074897,
      "grad_norm": 0.710927426815033,
      "learning_rate": 0.00018319022406613464,
      "loss": 1.8185,
      "step": 7550
    },
    {
      "epoch": 1.8748837497674995,
      "grad_norm": 0.6701392531394958,
      "learning_rate": 0.00018314696123025454,
      "loss": 1.7982,
      "step": 7560
    },
    {
      "epoch": 1.8773637547275095,
      "grad_norm": 0.7263465523719788,
      "learning_rate": 0.00018310364791603517,
      "loss": 1.781,
      "step": 7570
    },
    {
      "epoch": 1.8798437596875193,
      "grad_norm": 0.7514358758926392,
      "learning_rate": 0.00018306028414977193,
      "loss": 1.8347,
      "step": 7580
    },
    {
      "epoch": 1.8823237646475293,
      "grad_norm": 0.7695625424385071,
      "learning_rate": 0.00018301686995779093,
      "loss": 1.7763,
      "step": 7590
    },
    {
      "epoch": 1.8848037696075393,
      "grad_norm": 0.7509194016456604,
      "learning_rate": 0.00018297340536644875,
      "loss": 1.796,
      "step": 7600
    },
    {
      "epoch": 1.8872837745675493,
      "grad_norm": 0.6772938370704651,
      "learning_rate": 0.00018292989040213272,
      "loss": 1.7439,
      "step": 7610
    },
    {
      "epoch": 1.889763779527559,
      "grad_norm": 0.6906134486198425,
      "learning_rate": 0.00018288632509126066,
      "loss": 1.8171,
      "step": 7620
    },
    {
      "epoch": 1.892243784487569,
      "grad_norm": 0.6892842054367065,
      "learning_rate": 0.00018284270946028092,
      "loss": 1.7133,
      "step": 7630
    },
    {
      "epoch": 1.8947237894475788,
      "grad_norm": 0.7177791595458984,
      "learning_rate": 0.00018279904353567253,
      "loss": 1.7848,
      "step": 7640
    },
    {
      "epoch": 1.8972037944075888,
      "grad_norm": 0.7404572367668152,
      "learning_rate": 0.0001827553273439449,
      "loss": 1.8338,
      "step": 7650
    },
    {
      "epoch": 1.8996837993675988,
      "grad_norm": 0.6802579164505005,
      "learning_rate": 0.00018271156091163813,
      "loss": 1.8355,
      "step": 7660
    },
    {
      "epoch": 1.9021638043276088,
      "grad_norm": 0.6848039031028748,
      "learning_rate": 0.00018266774426532268,
      "loss": 1.795,
      "step": 7670
    },
    {
      "epoch": 1.9046438092876186,
      "grad_norm": 0.7783024907112122,
      "learning_rate": 0.0001826238774315995,
      "loss": 1.8309,
      "step": 7680
    },
    {
      "epoch": 1.9071238142476283,
      "grad_norm": 0.7328544855117798,
      "learning_rate": 0.00018257996043710012,
      "loss": 1.8229,
      "step": 7690
    },
    {
      "epoch": 1.9096038192076383,
      "grad_norm": 0.7623286843299866,
      "learning_rate": 0.00018253599330848638,
      "loss": 1.8428,
      "step": 7700
    },
    {
      "epoch": 1.9120838241676483,
      "grad_norm": 0.7214025855064392,
      "learning_rate": 0.0001824919760724507,
      "loss": 1.798,
      "step": 7710
    },
    {
      "epoch": 1.9145638291276583,
      "grad_norm": 0.734126091003418,
      "learning_rate": 0.0001824479087557158,
      "loss": 1.7948,
      "step": 7720
    },
    {
      "epoch": 1.9170438340876683,
      "grad_norm": 0.7752106785774231,
      "learning_rate": 0.00018240379138503488,
      "loss": 1.8656,
      "step": 7730
    },
    {
      "epoch": 1.919523839047678,
      "grad_norm": 0.6882955431938171,
      "learning_rate": 0.00018235962398719147,
      "loss": 1.7882,
      "step": 7740
    },
    {
      "epoch": 1.9220038440076879,
      "grad_norm": 0.7176995873451233,
      "learning_rate": 0.00018231540658899956,
      "loss": 1.7967,
      "step": 7750
    },
    {
      "epoch": 1.9244838489676979,
      "grad_norm": 0.6762322187423706,
      "learning_rate": 0.00018227113921730334,
      "loss": 1.7541,
      "step": 7760
    },
    {
      "epoch": 1.9269638539277079,
      "grad_norm": 0.6771329045295715,
      "learning_rate": 0.00018222682189897752,
      "loss": 1.8081,
      "step": 7770
    },
    {
      "epoch": 1.9294438588877179,
      "grad_norm": 0.6885058283805847,
      "learning_rate": 0.00018218245466092702,
      "loss": 1.8035,
      "step": 7780
    },
    {
      "epoch": 1.9319238638477279,
      "grad_norm": 0.6635348200798035,
      "learning_rate": 0.00018213803753008707,
      "loss": 1.8155,
      "step": 7790
    },
    {
      "epoch": 1.9344038688077376,
      "grad_norm": 0.7170267105102539,
      "learning_rate": 0.00018209357053342325,
      "loss": 1.7877,
      "step": 7800
    },
    {
      "epoch": 1.9368838737677474,
      "grad_norm": 0.7271670699119568,
      "learning_rate": 0.00018204905369793135,
      "loss": 1.8503,
      "step": 7810
    },
    {
      "epoch": 1.9393638787277574,
      "grad_norm": 0.7565329670906067,
      "learning_rate": 0.00018200448705063747,
      "loss": 1.7848,
      "step": 7820
    },
    {
      "epoch": 1.9418438836877674,
      "grad_norm": 0.7370867133140564,
      "learning_rate": 0.0001819598706185979,
      "loss": 1.7872,
      "step": 7830
    },
    {
      "epoch": 1.9443238886477774,
      "grad_norm": 0.7043139934539795,
      "learning_rate": 0.0001819152044288992,
      "loss": 1.764,
      "step": 7840
    },
    {
      "epoch": 1.9468038936077872,
      "grad_norm": 0.7746196985244751,
      "learning_rate": 0.00018187048850865808,
      "loss": 1.7954,
      "step": 7850
    },
    {
      "epoch": 1.9492838985677972,
      "grad_norm": 0.7171877026557922,
      "learning_rate": 0.00018182572288502152,
      "loss": 1.7933,
      "step": 7860
    },
    {
      "epoch": 1.951763903527807,
      "grad_norm": 0.667573094367981,
      "learning_rate": 0.00018178090758516662,
      "loss": 1.7796,
      "step": 7870
    },
    {
      "epoch": 1.954243908487817,
      "grad_norm": 0.7113328576087952,
      "learning_rate": 0.00018173604263630063,
      "loss": 1.8426,
      "step": 7880
    },
    {
      "epoch": 1.956723913447827,
      "grad_norm": 0.7328044772148132,
      "learning_rate": 0.000181691128065661,
      "loss": 1.8522,
      "step": 7890
    },
    {
      "epoch": 1.959203918407837,
      "grad_norm": 0.6921177506446838,
      "learning_rate": 0.00018164616390051522,
      "loss": 1.837,
      "step": 7900
    },
    {
      "epoch": 1.9616839233678467,
      "grad_norm": 0.7000945806503296,
      "learning_rate": 0.000181601150168161,
      "loss": 1.7452,
      "step": 7910
    },
    {
      "epoch": 1.9641639283278567,
      "grad_norm": 0.835935652256012,
      "learning_rate": 0.00018155608689592604,
      "loss": 1.8445,
      "step": 7920
    },
    {
      "epoch": 1.9666439332878665,
      "grad_norm": 0.7225850820541382,
      "learning_rate": 0.00018151097411116813,
      "loss": 1.786,
      "step": 7930
    },
    {
      "epoch": 1.9691239382478765,
      "grad_norm": 0.6941862106323242,
      "learning_rate": 0.00018146581184127518,
      "loss": 1.7506,
      "step": 7940
    },
    {
      "epoch": 1.9716039432078865,
      "grad_norm": 0.7091786861419678,
      "learning_rate": 0.00018142060011366512,
      "loss": 1.8744,
      "step": 7950
    },
    {
      "epoch": 1.9740839481678965,
      "grad_norm": 0.6893685460090637,
      "learning_rate": 0.00018137533895578583,
      "loss": 1.7464,
      "step": 7960
    },
    {
      "epoch": 1.9765639531279062,
      "grad_norm": 0.6945198774337769,
      "learning_rate": 0.00018133002839511533,
      "loss": 1.8894,
      "step": 7970
    },
    {
      "epoch": 1.9790439580879162,
      "grad_norm": 0.7510820031166077,
      "learning_rate": 0.00018128466845916154,
      "loss": 1.8154,
      "step": 7980
    },
    {
      "epoch": 1.981523963047926,
      "grad_norm": 0.7420031428337097,
      "learning_rate": 0.00018123925917546237,
      "loss": 1.8146,
      "step": 7990
    },
    {
      "epoch": 1.984003968007936,
      "grad_norm": 0.7060868740081787,
      "learning_rate": 0.00018119380057158568,
      "loss": 1.7797,
      "step": 8000
    },
    {
      "epoch": 1.986483972967946,
      "grad_norm": 0.7086240649223328,
      "learning_rate": 0.00018114829267512928,
      "loss": 1.7934,
      "step": 8010
    },
    {
      "epoch": 1.988963977927956,
      "grad_norm": 0.7174015045166016,
      "learning_rate": 0.000181102735513721,
      "loss": 1.7562,
      "step": 8020
    },
    {
      "epoch": 1.9914439828879658,
      "grad_norm": 0.6231516599655151,
      "learning_rate": 0.00018105712911501838,
      "loss": 1.7924,
      "step": 8030
    },
    {
      "epoch": 1.9939239878479755,
      "grad_norm": 0.7719596028327942,
      "learning_rate": 0.00018101147350670907,
      "loss": 1.8747,
      "step": 8040
    },
    {
      "epoch": 1.9964039928079855,
      "grad_norm": 0.749664843082428,
      "learning_rate": 0.00018096576871651042,
      "loss": 1.8553,
      "step": 8050
    },
    {
      "epoch": 1.9988839977679955,
      "grad_norm": 0.7149576544761658,
      "learning_rate": 0.00018092001477216973,
      "loss": 1.7741,
      "step": 8060
    },
    {
      "epoch": 2.0013640027280055,
      "grad_norm": 0.7196642160415649,
      "learning_rate": 0.00018087421170146412,
      "loss": 1.7028,
      "step": 8070
    },
    {
      "epoch": 2.0038440076880155,
      "grad_norm": 0.8466257452964783,
      "learning_rate": 0.00018082835953220054,
      "loss": 1.7266,
      "step": 8080
    },
    {
      "epoch": 2.0063240126480255,
      "grad_norm": 0.6822969913482666,
      "learning_rate": 0.00018078245829221575,
      "loss": 1.6703,
      "step": 8090
    },
    {
      "epoch": 2.008804017608035,
      "grad_norm": 0.8267595171928406,
      "learning_rate": 0.00018073650800937624,
      "loss": 1.7133,
      "step": 8100
    },
    {
      "epoch": 2.011284022568045,
      "grad_norm": 0.7269774675369263,
      "learning_rate": 0.0001806905087115784,
      "loss": 1.7366,
      "step": 8110
    },
    {
      "epoch": 2.013764027528055,
      "grad_norm": 0.7767934799194336,
      "learning_rate": 0.00018064446042674828,
      "loss": 1.7637,
      "step": 8120
    },
    {
      "epoch": 2.016244032488065,
      "grad_norm": 0.7064205408096313,
      "learning_rate": 0.00018059836318284168,
      "loss": 1.6629,
      "step": 8130
    },
    {
      "epoch": 2.018724037448075,
      "grad_norm": 0.7273749709129333,
      "learning_rate": 0.0001805522170078441,
      "loss": 1.674,
      "step": 8140
    },
    {
      "epoch": 2.0212040424080846,
      "grad_norm": 0.7249225974082947,
      "learning_rate": 0.00018050602192977086,
      "loss": 1.6633,
      "step": 8150
    },
    {
      "epoch": 2.0236840473680946,
      "grad_norm": 0.746360719203949,
      "learning_rate": 0.00018045977797666684,
      "loss": 1.7123,
      "step": 8160
    },
    {
      "epoch": 2.0261640523281046,
      "grad_norm": 0.714149534702301,
      "learning_rate": 0.00018041348517660666,
      "loss": 1.6978,
      "step": 8170
    },
    {
      "epoch": 2.0286440572881146,
      "grad_norm": 0.719533383846283,
      "learning_rate": 0.00018036714355769458,
      "loss": 1.7,
      "step": 8180
    },
    {
      "epoch": 2.0311240622481246,
      "grad_norm": 0.7731805443763733,
      "learning_rate": 0.00018032075314806448,
      "loss": 1.7369,
      "step": 8190
    },
    {
      "epoch": 2.0336040672081346,
      "grad_norm": 0.739291250705719,
      "learning_rate": 0.00018027431397587992,
      "loss": 1.737,
      "step": 8200
    },
    {
      "epoch": 2.036084072168144,
      "grad_norm": 0.792762041091919,
      "learning_rate": 0.00018022782606933398,
      "loss": 1.7221,
      "step": 8210
    },
    {
      "epoch": 2.038564077128154,
      "grad_norm": 0.8123416304588318,
      "learning_rate": 0.00018018128945664937,
      "loss": 1.6844,
      "step": 8220
    },
    {
      "epoch": 2.041044082088164,
      "grad_norm": 0.7434293031692505,
      "learning_rate": 0.00018013470416607837,
      "loss": 1.7407,
      "step": 8230
    },
    {
      "epoch": 2.043524087048174,
      "grad_norm": 0.6932642459869385,
      "learning_rate": 0.00018008807022590283,
      "loss": 1.7084,
      "step": 8240
    },
    {
      "epoch": 2.046004092008184,
      "grad_norm": 0.7205861210823059,
      "learning_rate": 0.00018004138766443406,
      "loss": 1.6771,
      "step": 8250
    },
    {
      "epoch": 2.048484096968194,
      "grad_norm": 0.828132688999176,
      "learning_rate": 0.00017999465651001296,
      "loss": 1.7001,
      "step": 8260
    },
    {
      "epoch": 2.0509641019282037,
      "grad_norm": 0.8889887928962708,
      "learning_rate": 0.00017994787679100998,
      "loss": 1.7458,
      "step": 8270
    },
    {
      "epoch": 2.0534441068882137,
      "grad_norm": 0.7904499173164368,
      "learning_rate": 0.00017990104853582493,
      "loss": 1.6571,
      "step": 8280
    },
    {
      "epoch": 2.0559241118482237,
      "grad_norm": 0.8607293367385864,
      "learning_rate": 0.00017985417177288714,
      "loss": 1.7113,
      "step": 8290
    },
    {
      "epoch": 2.0584041168082337,
      "grad_norm": 0.7840188145637512,
      "learning_rate": 0.00017980724653065538,
      "loss": 1.7306,
      "step": 8300
    },
    {
      "epoch": 2.0608841217682436,
      "grad_norm": 0.7568309903144836,
      "learning_rate": 0.00017976027283761787,
      "loss": 1.7568,
      "step": 8310
    },
    {
      "epoch": 2.0633641267282536,
      "grad_norm": 0.7962266206741333,
      "learning_rate": 0.00017971325072229226,
      "loss": 1.7057,
      "step": 8320
    },
    {
      "epoch": 2.065844131688263,
      "grad_norm": 0.7668977379798889,
      "learning_rate": 0.00017966618021322557,
      "loss": 1.7208,
      "step": 8330
    },
    {
      "epoch": 2.068324136648273,
      "grad_norm": 0.797734797000885,
      "learning_rate": 0.0001796190613389942,
      "loss": 1.7364,
      "step": 8340
    },
    {
      "epoch": 2.070804141608283,
      "grad_norm": 0.8187463879585266,
      "learning_rate": 0.00017957189412820386,
      "loss": 1.6687,
      "step": 8350
    },
    {
      "epoch": 2.073284146568293,
      "grad_norm": 0.7944939732551575,
      "learning_rate": 0.00017952467860948973,
      "loss": 1.6799,
      "step": 8360
    },
    {
      "epoch": 2.075764151528303,
      "grad_norm": 0.7479957342147827,
      "learning_rate": 0.0001794774148115163,
      "loss": 1.6735,
      "step": 8370
    },
    {
      "epoch": 2.078244156488313,
      "grad_norm": 0.8243913054466248,
      "learning_rate": 0.00017943010276297717,
      "loss": 1.7604,
      "step": 8380
    },
    {
      "epoch": 2.0807241614483227,
      "grad_norm": 0.7490248680114746,
      "learning_rate": 0.00017938274249259547,
      "loss": 1.6228,
      "step": 8390
    },
    {
      "epoch": 2.0832041664083327,
      "grad_norm": 0.7714105844497681,
      "learning_rate": 0.00017933533402912354,
      "loss": 1.6705,
      "step": 8400
    },
    {
      "epoch": 2.0856841713683427,
      "grad_norm": 0.8160969018936157,
      "learning_rate": 0.00017928787740134288,
      "loss": 1.707,
      "step": 8410
    },
    {
      "epoch": 2.0881641763283527,
      "grad_norm": 0.7864915728569031,
      "learning_rate": 0.0001792403726380644,
      "loss": 1.6826,
      "step": 8420
    },
    {
      "epoch": 2.0906441812883627,
      "grad_norm": 0.7798502445220947,
      "learning_rate": 0.00017919281976812803,
      "loss": 1.6958,
      "step": 8430
    },
    {
      "epoch": 2.0931241862483727,
      "grad_norm": 0.8461658954620361,
      "learning_rate": 0.0001791452188204031,
      "loss": 1.6742,
      "step": 8440
    },
    {
      "epoch": 2.0956041912083823,
      "grad_norm": 0.7308477759361267,
      "learning_rate": 0.000179097569823788,
      "loss": 1.706,
      "step": 8450
    },
    {
      "epoch": 2.0980841961683923,
      "grad_norm": 0.9373131990432739,
      "learning_rate": 0.00017904987280721035,
      "loss": 1.676,
      "step": 8460
    },
    {
      "epoch": 2.1005642011284023,
      "grad_norm": 0.7441970705986023,
      "learning_rate": 0.0001790021277996269,
      "loss": 1.6748,
      "step": 8470
    },
    {
      "epoch": 2.1030442060884122,
      "grad_norm": 0.7834221720695496,
      "learning_rate": 0.00017895433483002354,
      "loss": 1.6826,
      "step": 8480
    },
    {
      "epoch": 2.1055242110484222,
      "grad_norm": 0.8050951361656189,
      "learning_rate": 0.0001789064939274153,
      "loss": 1.7652,
      "step": 8490
    },
    {
      "epoch": 2.108004216008432,
      "grad_norm": 0.807235836982727,
      "learning_rate": 0.00017885860512084623,
      "loss": 1.7278,
      "step": 8500
    },
    {
      "epoch": 2.110484220968442,
      "grad_norm": 0.7822979688644409,
      "learning_rate": 0.00017881066843938956,
      "loss": 1.703,
      "step": 8510
    },
    {
      "epoch": 2.112964225928452,
      "grad_norm": 0.8085262179374695,
      "learning_rate": 0.00017876268391214754,
      "loss": 1.712,
      "step": 8520
    },
    {
      "epoch": 2.115444230888462,
      "grad_norm": 0.7531107664108276,
      "learning_rate": 0.00017871465156825148,
      "loss": 1.7009,
      "step": 8530
    },
    {
      "epoch": 2.117924235848472,
      "grad_norm": 0.7848300337791443,
      "learning_rate": 0.00017866657143686168,
      "loss": 1.7549,
      "step": 8540
    },
    {
      "epoch": 2.1204042408084818,
      "grad_norm": 0.7618765830993652,
      "learning_rate": 0.00017861844354716757,
      "loss": 1.7733,
      "step": 8550
    },
    {
      "epoch": 2.1228842457684913,
      "grad_norm": 0.8201743960380554,
      "learning_rate": 0.00017857026792838737,
      "loss": 1.7583,
      "step": 8560
    },
    {
      "epoch": 2.1253642507285013,
      "grad_norm": 0.7396107912063599,
      "learning_rate": 0.00017852204460976846,
      "loss": 1.7147,
      "step": 8570
    },
    {
      "epoch": 2.1278442556885113,
      "grad_norm": 0.7462581396102905,
      "learning_rate": 0.00017847377362058712,
      "loss": 1.7035,
      "step": 8580
    },
    {
      "epoch": 2.1303242606485213,
      "grad_norm": 0.7520790696144104,
      "learning_rate": 0.00017842545499014852,
      "loss": 1.6986,
      "step": 8590
    },
    {
      "epoch": 2.1328042656085313,
      "grad_norm": 0.7530083060264587,
      "learning_rate": 0.00017837708874778683,
      "loss": 1.753,
      "step": 8600
    },
    {
      "epoch": 2.1352842705685413,
      "grad_norm": 0.8094317317008972,
      "learning_rate": 0.00017832867492286505,
      "loss": 1.7664,
      "step": 8610
    },
    {
      "epoch": 2.137764275528551,
      "grad_norm": 0.7181040048599243,
      "learning_rate": 0.00017828021354477516,
      "loss": 1.6667,
      "step": 8620
    },
    {
      "epoch": 2.140244280488561,
      "grad_norm": 0.8137990832328796,
      "learning_rate": 0.0001782317046429379,
      "loss": 1.7005,
      "step": 8630
    },
    {
      "epoch": 2.142724285448571,
      "grad_norm": 0.792864203453064,
      "learning_rate": 0.000178183148246803,
      "loss": 1.6757,
      "step": 8640
    },
    {
      "epoch": 2.145204290408581,
      "grad_norm": 0.7756101489067078,
      "learning_rate": 0.00017813454438584886,
      "loss": 1.7149,
      "step": 8650
    },
    {
      "epoch": 2.147684295368591,
      "grad_norm": 0.8082769513130188,
      "learning_rate": 0.00017808589308958284,
      "loss": 1.6721,
      "step": 8660
    },
    {
      "epoch": 2.150164300328601,
      "grad_norm": 0.8983075022697449,
      "learning_rate": 0.00017803719438754095,
      "loss": 1.7542,
      "step": 8670
    },
    {
      "epoch": 2.1526443052886104,
      "grad_norm": 0.8210811018943787,
      "learning_rate": 0.00017798844830928817,
      "loss": 1.6946,
      "step": 8680
    },
    {
      "epoch": 2.1551243102486204,
      "grad_norm": 0.7229629755020142,
      "learning_rate": 0.0001779396548844181,
      "loss": 1.6686,
      "step": 8690
    },
    {
      "epoch": 2.1576043152086304,
      "grad_norm": 0.7622604370117188,
      "learning_rate": 0.00017789081414255312,
      "loss": 1.7335,
      "step": 8700
    },
    {
      "epoch": 2.1600843201686404,
      "grad_norm": 0.771666944026947,
      "learning_rate": 0.0001778419261133443,
      "loss": 1.6763,
      "step": 8710
    },
    {
      "epoch": 2.1625643251286504,
      "grad_norm": 0.8421552777290344,
      "learning_rate": 0.00017779299082647148,
      "loss": 1.741,
      "step": 8720
    },
    {
      "epoch": 2.1650443300886604,
      "grad_norm": 0.7799654006958008,
      "learning_rate": 0.00017774400831164323,
      "loss": 1.6901,
      "step": 8730
    },
    {
      "epoch": 2.16752433504867,
      "grad_norm": 0.7670300602912903,
      "learning_rate": 0.00017769497859859663,
      "loss": 1.7533,
      "step": 8740
    },
    {
      "epoch": 2.17000434000868,
      "grad_norm": 0.8089250922203064,
      "learning_rate": 0.0001776459017170976,
      "loss": 1.7927,
      "step": 8750
    },
    {
      "epoch": 2.17248434496869,
      "grad_norm": 0.7334123253822327,
      "learning_rate": 0.0001775967776969405,
      "loss": 1.7462,
      "step": 8760
    },
    {
      "epoch": 2.1749643499287,
      "grad_norm": 0.7781856060028076,
      "learning_rate": 0.0001775476065679485,
      "loss": 1.6699,
      "step": 8770
    },
    {
      "epoch": 2.17744435488871,
      "grad_norm": 0.7868925929069519,
      "learning_rate": 0.00017749838835997328,
      "loss": 1.712,
      "step": 8780
    },
    {
      "epoch": 2.1799243598487195,
      "grad_norm": 0.6954182386398315,
      "learning_rate": 0.0001774491231028951,
      "loss": 1.7006,
      "step": 8790
    },
    {
      "epoch": 2.1824043648087295,
      "grad_norm": 0.7828925251960754,
      "learning_rate": 0.00017739981082662276,
      "loss": 1.707,
      "step": 8800
    },
    {
      "epoch": 2.1848843697687395,
      "grad_norm": 0.8318265676498413,
      "learning_rate": 0.00017735045156109367,
      "loss": 1.7669,
      "step": 8810
    },
    {
      "epoch": 2.1873643747287494,
      "grad_norm": 0.844477653503418,
      "learning_rate": 0.0001773010453362737,
      "loss": 1.6718,
      "step": 8820
    },
    {
      "epoch": 2.1898443796887594,
      "grad_norm": 0.8850805759429932,
      "learning_rate": 0.0001772515921821573,
      "loss": 1.7371,
      "step": 8830
    },
    {
      "epoch": 2.1923243846487694,
      "grad_norm": 0.7837343811988831,
      "learning_rate": 0.00017720209212876737,
      "loss": 1.6949,
      "step": 8840
    },
    {
      "epoch": 2.1948043896087794,
      "grad_norm": 0.7297570109367371,
      "learning_rate": 0.00017715254520615523,
      "loss": 1.6592,
      "step": 8850
    },
    {
      "epoch": 2.197284394568789,
      "grad_norm": 0.8145079612731934,
      "learning_rate": 0.00017710295144440078,
      "loss": 1.7381,
      "step": 8860
    },
    {
      "epoch": 2.199764399528799,
      "grad_norm": 0.8012209534645081,
      "learning_rate": 0.00017705331087361224,
      "loss": 1.7826,
      "step": 8870
    },
    {
      "epoch": 2.202244404488809,
      "grad_norm": 0.7602465152740479,
      "learning_rate": 0.0001770036235239263,
      "loss": 1.7411,
      "step": 8880
    },
    {
      "epoch": 2.204724409448819,
      "grad_norm": 0.7813806533813477,
      "learning_rate": 0.00017695388942550805,
      "loss": 1.7317,
      "step": 8890
    },
    {
      "epoch": 2.207204414408829,
      "grad_norm": 0.8720101714134216,
      "learning_rate": 0.00017690410860855094,
      "loss": 1.66,
      "step": 8900
    },
    {
      "epoch": 2.2096844193688385,
      "grad_norm": 0.8317455649375916,
      "learning_rate": 0.00017685428110327683,
      "loss": 1.7734,
      "step": 8910
    },
    {
      "epoch": 2.2121644243288485,
      "grad_norm": 0.7609864473342896,
      "learning_rate": 0.00017680440693993585,
      "loss": 1.6641,
      "step": 8920
    },
    {
      "epoch": 2.2146444292888585,
      "grad_norm": 0.78978431224823,
      "learning_rate": 0.0001767544861488065,
      "loss": 1.6812,
      "step": 8930
    },
    {
      "epoch": 2.2171244342488685,
      "grad_norm": 0.7690613269805908,
      "learning_rate": 0.00017670451876019563,
      "loss": 1.7145,
      "step": 8940
    },
    {
      "epoch": 2.2196044392088785,
      "grad_norm": 0.7488145232200623,
      "learning_rate": 0.00017665450480443826,
      "loss": 1.7223,
      "step": 8950
    },
    {
      "epoch": 2.2220844441688885,
      "grad_norm": 0.766918420791626,
      "learning_rate": 0.0001766044443118978,
      "loss": 1.7422,
      "step": 8960
    },
    {
      "epoch": 2.224564449128898,
      "grad_norm": 0.8061556220054626,
      "learning_rate": 0.00017655433731296588,
      "loss": 1.7434,
      "step": 8970
    },
    {
      "epoch": 2.227044454088908,
      "grad_norm": 0.7330317497253418,
      "learning_rate": 0.00017650418383806233,
      "loss": 1.7188,
      "step": 8980
    },
    {
      "epoch": 2.229524459048918,
      "grad_norm": 0.822127640247345,
      "learning_rate": 0.00017645398391763516,
      "loss": 1.6602,
      "step": 8990
    },
    {
      "epoch": 2.232004464008928,
      "grad_norm": 0.8580800294876099,
      "learning_rate": 0.00017640373758216077,
      "loss": 1.7452,
      "step": 9000
    },
    {
      "epoch": 2.234484468968938,
      "grad_norm": 0.7564178705215454,
      "learning_rate": 0.00017635344486214346,
      "loss": 1.7382,
      "step": 9010
    },
    {
      "epoch": 2.236964473928948,
      "grad_norm": 0.7477293610572815,
      "learning_rate": 0.0001763031057881159,
      "loss": 1.7573,
      "step": 9020
    },
    {
      "epoch": 2.2394444788889576,
      "grad_norm": 0.7672526836395264,
      "learning_rate": 0.00017625272039063883,
      "loss": 1.7245,
      "step": 9030
    },
    {
      "epoch": 2.2419244838489676,
      "grad_norm": 0.8399450778961182,
      "learning_rate": 0.00017620228870030108,
      "loss": 1.6886,
      "step": 9040
    },
    {
      "epoch": 2.2444044888089776,
      "grad_norm": 0.8164432644844055,
      "learning_rate": 0.00017615181074771968,
      "loss": 1.6718,
      "step": 9050
    },
    {
      "epoch": 2.2468844937689876,
      "grad_norm": 0.77805495262146,
      "learning_rate": 0.00017610128656353962,
      "loss": 1.653,
      "step": 9060
    },
    {
      "epoch": 2.2493644987289976,
      "grad_norm": 0.817104697227478,
      "learning_rate": 0.00017605071617843407,
      "loss": 1.7316,
      "step": 9070
    },
    {
      "epoch": 2.2518445036890076,
      "grad_norm": 0.7542750835418701,
      "learning_rate": 0.00017600009962310417,
      "loss": 1.732,
      "step": 9080
    },
    {
      "epoch": 2.254324508649017,
      "grad_norm": 0.8358319997787476,
      "learning_rate": 0.00017594943692827915,
      "loss": 1.7192,
      "step": 9090
    },
    {
      "epoch": 2.256804513609027,
      "grad_norm": 0.7867502570152283,
      "learning_rate": 0.0001758987281247162,
      "loss": 1.7219,
      "step": 9100
    },
    {
      "epoch": 2.259284518569037,
      "grad_norm": 0.7503636479377747,
      "learning_rate": 0.00017584797324320052,
      "loss": 1.722,
      "step": 9110
    },
    {
      "epoch": 2.261764523529047,
      "grad_norm": 0.7286480069160461,
      "learning_rate": 0.0001757971723145453,
      "loss": 1.7113,
      "step": 9120
    },
    {
      "epoch": 2.264244528489057,
      "grad_norm": 0.9005977511405945,
      "learning_rate": 0.00017574632536959168,
      "loss": 1.7213,
      "step": 9130
    },
    {
      "epoch": 2.266724533449067,
      "grad_norm": 0.8166096806526184,
      "learning_rate": 0.00017569543243920873,
      "loss": 1.6483,
      "step": 9140
    },
    {
      "epoch": 2.2692045384090767,
      "grad_norm": 0.8187897801399231,
      "learning_rate": 0.0001756444935542934,
      "loss": 1.6985,
      "step": 9150
    },
    {
      "epoch": 2.2716845433690867,
      "grad_norm": 0.7724904417991638,
      "learning_rate": 0.00017559350874577065,
      "loss": 1.7356,
      "step": 9160
    },
    {
      "epoch": 2.2741645483290966,
      "grad_norm": 0.8105554580688477,
      "learning_rate": 0.00017554247804459316,
      "loss": 1.6206,
      "step": 9170
    },
    {
      "epoch": 2.2766445532891066,
      "grad_norm": 0.7334214448928833,
      "learning_rate": 0.0001754914014817416,
      "loss": 1.6594,
      "step": 9180
    },
    {
      "epoch": 2.2791245582491166,
      "grad_norm": 0.8149214386940002,
      "learning_rate": 0.00017544027908822444,
      "loss": 1.6853,
      "step": 9190
    },
    {
      "epoch": 2.281604563209126,
      "grad_norm": 0.7332905530929565,
      "learning_rate": 0.000175389110895078,
      "loss": 1.7218,
      "step": 9200
    },
    {
      "epoch": 2.284084568169136,
      "grad_norm": 0.743989884853363,
      "learning_rate": 0.00017533789693336633,
      "loss": 1.7004,
      "step": 9210
    },
    {
      "epoch": 2.286564573129146,
      "grad_norm": 0.7727570533752441,
      "learning_rate": 0.00017528663723418136,
      "loss": 1.6864,
      "step": 9220
    },
    {
      "epoch": 2.289044578089156,
      "grad_norm": 0.8129294514656067,
      "learning_rate": 0.0001752353318286427,
      "loss": 1.6946,
      "step": 9230
    },
    {
      "epoch": 2.291524583049166,
      "grad_norm": 0.9174445867538452,
      "learning_rate": 0.00017518398074789775,
      "loss": 1.7763,
      "step": 9240
    },
    {
      "epoch": 2.294004588009176,
      "grad_norm": 0.7691509127616882,
      "learning_rate": 0.00017513258402312168,
      "loss": 1.7373,
      "step": 9250
    },
    {
      "epoch": 2.296484592969186,
      "grad_norm": 0.7788468599319458,
      "learning_rate": 0.0001750811416855173,
      "loss": 1.6936,
      "step": 9260
    },
    {
      "epoch": 2.2989645979291957,
      "grad_norm": 0.7824164628982544,
      "learning_rate": 0.00017502965376631514,
      "loss": 1.743,
      "step": 9270
    },
    {
      "epoch": 2.3014446028892057,
      "grad_norm": 0.7765858173370361,
      "learning_rate": 0.00017497812029677344,
      "loss": 1.7117,
      "step": 9280
    },
    {
      "epoch": 2.3039246078492157,
      "grad_norm": 0.7741357088088989,
      "learning_rate": 0.000174926541308178,
      "loss": 1.7012,
      "step": 9290
    },
    {
      "epoch": 2.3064046128092257,
      "grad_norm": 0.8304239511489868,
      "learning_rate": 0.00017487491683184235,
      "loss": 1.6477,
      "step": 9300
    },
    {
      "epoch": 2.3088846177692357,
      "grad_norm": 0.9060750603675842,
      "learning_rate": 0.00017482324689910758,
      "loss": 1.7314,
      "step": 9310
    },
    {
      "epoch": 2.3113646227292453,
      "grad_norm": 0.7856857180595398,
      "learning_rate": 0.00017477153154134243,
      "loss": 1.7262,
      "step": 9320
    },
    {
      "epoch": 2.3138446276892553,
      "grad_norm": 0.7838661074638367,
      "learning_rate": 0.00017471977078994315,
      "loss": 1.7106,
      "step": 9330
    },
    {
      "epoch": 2.3163246326492652,
      "grad_norm": 0.7551782727241516,
      "learning_rate": 0.00017466796467633354,
      "loss": 1.767,
      "step": 9340
    },
    {
      "epoch": 2.3188046376092752,
      "grad_norm": 0.7626734375953674,
      "learning_rate": 0.00017461611323196506,
      "loss": 1.6888,
      "step": 9350
    },
    {
      "epoch": 2.3212846425692852,
      "grad_norm": 0.7644707560539246,
      "learning_rate": 0.00017456421648831655,
      "loss": 1.6997,
      "step": 9360
    },
    {
      "epoch": 2.3237646475292952,
      "grad_norm": 0.8020896911621094,
      "learning_rate": 0.00017451227447689446,
      "loss": 1.7241,
      "step": 9370
    },
    {
      "epoch": 2.326244652489305,
      "grad_norm": 0.8020631670951843,
      "learning_rate": 0.00017446028722923265,
      "loss": 1.689,
      "step": 9380
    },
    {
      "epoch": 2.328724657449315,
      "grad_norm": 0.8963233232498169,
      "learning_rate": 0.00017440825477689243,
      "loss": 1.6784,
      "step": 9390
    },
    {
      "epoch": 2.331204662409325,
      "grad_norm": 0.7478158473968506,
      "learning_rate": 0.00017435617715146263,
      "loss": 1.7703,
      "step": 9400
    },
    {
      "epoch": 2.3336846673693348,
      "grad_norm": 0.8088776469230652,
      "learning_rate": 0.0001743040543845594,
      "loss": 1.6268,
      "step": 9410
    },
    {
      "epoch": 2.3361646723293448,
      "grad_norm": 0.780013382434845,
      "learning_rate": 0.00017425188650782648,
      "loss": 1.6768,
      "step": 9420
    },
    {
      "epoch": 2.3386446772893548,
      "grad_norm": 0.7265764474868774,
      "learning_rate": 0.00017419967355293477,
      "loss": 1.7766,
      "step": 9430
    },
    {
      "epoch": 2.3411246822493643,
      "grad_norm": 0.8079309463500977,
      "learning_rate": 0.00017414741555158266,
      "loss": 1.7487,
      "step": 9440
    },
    {
      "epoch": 2.3436046872093743,
      "grad_norm": 0.7889912724494934,
      "learning_rate": 0.00017409511253549593,
      "loss": 1.7093,
      "step": 9450
    },
    {
      "epoch": 2.3460846921693843,
      "grad_norm": 0.8098340034484863,
      "learning_rate": 0.00017404276453642753,
      "loss": 1.6022,
      "step": 9460
    },
    {
      "epoch": 2.3485646971293943,
      "grad_norm": 0.7909839749336243,
      "learning_rate": 0.0001739903715861579,
      "loss": 1.757,
      "step": 9470
    },
    {
      "epoch": 2.3510447020894043,
      "grad_norm": 0.8547044396400452,
      "learning_rate": 0.00017393793371649462,
      "loss": 1.7027,
      "step": 9480
    },
    {
      "epoch": 2.353524707049414,
      "grad_norm": 0.7982743382453918,
      "learning_rate": 0.00017388545095927267,
      "loss": 1.7402,
      "step": 9490
    },
    {
      "epoch": 2.356004712009424,
      "grad_norm": 0.8027835488319397,
      "learning_rate": 0.0001738329233463542,
      "loss": 1.6997,
      "step": 9500
    },
    {
      "epoch": 2.358484716969434,
      "grad_norm": 0.9833059310913086,
      "learning_rate": 0.00017378035090962856,
      "loss": 1.6549,
      "step": 9510
    },
    {
      "epoch": 2.360964721929444,
      "grad_norm": 0.7589620351791382,
      "learning_rate": 0.0001737277336810124,
      "loss": 1.6797,
      "step": 9520
    },
    {
      "epoch": 2.363444726889454,
      "grad_norm": 0.8295498490333557,
      "learning_rate": 0.00017367507169244956,
      "loss": 1.8149,
      "step": 9530
    },
    {
      "epoch": 2.365924731849464,
      "grad_norm": 0.8063281178474426,
      "learning_rate": 0.00017362236497591094,
      "loss": 1.7239,
      "step": 9540
    },
    {
      "epoch": 2.368404736809474,
      "grad_norm": 0.747158944606781,
      "learning_rate": 0.00017356961356339477,
      "loss": 1.7549,
      "step": 9550
    },
    {
      "epoch": 2.3708847417694834,
      "grad_norm": 0.8062665462493896,
      "learning_rate": 0.0001735168174869262,
      "loss": 1.7706,
      "step": 9560
    },
    {
      "epoch": 2.3733647467294934,
      "grad_norm": 0.8283780813217163,
      "learning_rate": 0.0001734639767785577,
      "loss": 1.6389,
      "step": 9570
    },
    {
      "epoch": 2.3758447516895034,
      "grad_norm": 0.7413634061813354,
      "learning_rate": 0.00017341109147036874,
      "loss": 1.7445,
      "step": 9580
    },
    {
      "epoch": 2.3783247566495134,
      "grad_norm": 0.8234320878982544,
      "learning_rate": 0.00017335816159446584,
      "loss": 1.6611,
      "step": 9590
    },
    {
      "epoch": 2.3808047616095234,
      "grad_norm": 0.7659749984741211,
      "learning_rate": 0.00017330518718298264,
      "loss": 1.6859,
      "step": 9600
    },
    {
      "epoch": 2.383284766569533,
      "grad_norm": 0.8007732033729553,
      "learning_rate": 0.0001732521682680798,
      "loss": 1.7249,
      "step": 9610
    },
    {
      "epoch": 2.385764771529543,
      "grad_norm": 0.7565622329711914,
      "learning_rate": 0.00017319910488194492,
      "loss": 1.7031,
      "step": 9620
    },
    {
      "epoch": 2.388244776489553,
      "grad_norm": 0.7995969653129578,
      "learning_rate": 0.00017314599705679277,
      "loss": 1.7743,
      "step": 9630
    },
    {
      "epoch": 2.390724781449563,
      "grad_norm": 0.7712788581848145,
      "learning_rate": 0.00017309284482486495,
      "loss": 1.7556,
      "step": 9640
    },
    {
      "epoch": 2.393204786409573,
      "grad_norm": 0.846427857875824,
      "learning_rate": 0.00017303964821843002,
      "loss": 1.7141,
      "step": 9650
    },
    {
      "epoch": 2.395684791369583,
      "grad_norm": 0.7683253884315491,
      "learning_rate": 0.00017298640726978357,
      "loss": 1.7192,
      "step": 9660
    },
    {
      "epoch": 2.398164796329593,
      "grad_norm": 0.797095537185669,
      "learning_rate": 0.00017293312201124806,
      "loss": 1.7032,
      "step": 9670
    },
    {
      "epoch": 2.4006448012896024,
      "grad_norm": 0.7995537519454956,
      "learning_rate": 0.00017287979247517286,
      "loss": 1.7381,
      "step": 9680
    },
    {
      "epoch": 2.4031248062496124,
      "grad_norm": 0.8047399520874023,
      "learning_rate": 0.00017282641869393418,
      "loss": 1.7271,
      "step": 9690
    },
    {
      "epoch": 2.4056048112096224,
      "grad_norm": 0.7926669120788574,
      "learning_rate": 0.00017277300069993515,
      "loss": 1.7047,
      "step": 9700
    },
    {
      "epoch": 2.4080848161696324,
      "grad_norm": 0.7393646240234375,
      "learning_rate": 0.00017271953852560575,
      "loss": 1.7405,
      "step": 9710
    },
    {
      "epoch": 2.4105648211296424,
      "grad_norm": 0.7627696990966797,
      "learning_rate": 0.0001726660322034027,
      "loss": 1.6729,
      "step": 9720
    },
    {
      "epoch": 2.413044826089652,
      "grad_norm": 0.794486403465271,
      "learning_rate": 0.00017261248176580958,
      "loss": 1.6695,
      "step": 9730
    },
    {
      "epoch": 2.415524831049662,
      "grad_norm": 0.8272292017936707,
      "learning_rate": 0.0001725588872453368,
      "loss": 1.7694,
      "step": 9740
    },
    {
      "epoch": 2.418004836009672,
      "grad_norm": 0.7959834337234497,
      "learning_rate": 0.0001725052486745214,
      "loss": 1.6821,
      "step": 9750
    },
    {
      "epoch": 2.420484840969682,
      "grad_norm": 0.8304336071014404,
      "learning_rate": 0.00017245156608592727,
      "loss": 1.702,
      "step": 9760
    },
    {
      "epoch": 2.422964845929692,
      "grad_norm": 0.7784603238105774,
      "learning_rate": 0.00017239783951214502,
      "loss": 1.7451,
      "step": 9770
    },
    {
      "epoch": 2.425444850889702,
      "grad_norm": 0.8049282431602478,
      "learning_rate": 0.00017234406898579188,
      "loss": 1.7577,
      "step": 9780
    },
    {
      "epoch": 2.4279248558497115,
      "grad_norm": 0.7931556105613708,
      "learning_rate": 0.00017229025453951183,
      "loss": 1.705,
      "step": 9790
    },
    {
      "epoch": 2.4304048608097215,
      "grad_norm": 0.7905335426330566,
      "learning_rate": 0.00017223639620597556,
      "loss": 1.7001,
      "step": 9800
    },
    {
      "epoch": 2.4328848657697315,
      "grad_norm": 0.785211443901062,
      "learning_rate": 0.00017218249401788031,
      "loss": 1.7204,
      "step": 9810
    },
    {
      "epoch": 2.4353648707297415,
      "grad_norm": 0.7745071649551392,
      "learning_rate": 0.00017212854800794996,
      "loss": 1.7019,
      "step": 9820
    },
    {
      "epoch": 2.4378448756897515,
      "grad_norm": 0.7807512879371643,
      "learning_rate": 0.00017207455820893505,
      "loss": 1.7346,
      "step": 9830
    },
    {
      "epoch": 2.4403248806497615,
      "grad_norm": 0.7244405746459961,
      "learning_rate": 0.00017202052465361268,
      "loss": 1.7758,
      "step": 9840
    },
    {
      "epoch": 2.442804885609771,
      "grad_norm": 0.7513141632080078,
      "learning_rate": 0.00017196644737478645,
      "loss": 1.7316,
      "step": 9850
    },
    {
      "epoch": 2.445284890569781,
      "grad_norm": 0.7699015736579895,
      "learning_rate": 0.0001719123264052866,
      "loss": 1.7933,
      "step": 9860
    },
    {
      "epoch": 2.447764895529791,
      "grad_norm": 0.7975505590438843,
      "learning_rate": 0.0001718581617779698,
      "loss": 1.6899,
      "step": 9870
    },
    {
      "epoch": 2.450244900489801,
      "grad_norm": 0.7789102792739868,
      "learning_rate": 0.0001718039535257194,
      "loss": 1.7527,
      "step": 9880
    },
    {
      "epoch": 2.452724905449811,
      "grad_norm": 0.8182029128074646,
      "learning_rate": 0.00017174970168144495,
      "loss": 1.7485,
      "step": 9890
    },
    {
      "epoch": 2.4552049104098206,
      "grad_norm": 0.7718999981880188,
      "learning_rate": 0.00017169540627808274,
      "loss": 1.7086,
      "step": 9900
    },
    {
      "epoch": 2.4576849153698306,
      "grad_norm": 0.7620275616645813,
      "learning_rate": 0.00017164106734859536,
      "loss": 1.7243,
      "step": 9910
    },
    {
      "epoch": 2.4601649203298406,
      "grad_norm": 0.8015621304512024,
      "learning_rate": 0.00017158668492597186,
      "loss": 1.6978,
      "step": 9920
    },
    {
      "epoch": 2.4626449252898506,
      "grad_norm": 0.7627947330474854,
      "learning_rate": 0.00017153225904322766,
      "loss": 1.6888,
      "step": 9930
    },
    {
      "epoch": 2.4651249302498606,
      "grad_norm": 0.738109827041626,
      "learning_rate": 0.00017147778973340466,
      "loss": 1.7365,
      "step": 9940
    },
    {
      "epoch": 2.4676049352098706,
      "grad_norm": 0.7651798129081726,
      "learning_rate": 0.000171423277029571,
      "loss": 1.7147,
      "step": 9950
    },
    {
      "epoch": 2.4700849401698806,
      "grad_norm": 0.8257949352264404,
      "learning_rate": 0.0001713687209648212,
      "loss": 1.704,
      "step": 9960
    },
    {
      "epoch": 2.47256494512989,
      "grad_norm": 0.7816750407218933,
      "learning_rate": 0.00017131412157227625,
      "loss": 1.7019,
      "step": 9970
    },
    {
      "epoch": 2.4750449500899,
      "grad_norm": 0.8303975462913513,
      "learning_rate": 0.0001712594788850832,
      "loss": 1.7225,
      "step": 9980
    },
    {
      "epoch": 2.47752495504991,
      "grad_norm": 0.8580611944198608,
      "learning_rate": 0.00017120479293641555,
      "loss": 1.7306,
      "step": 9990
    },
    {
      "epoch": 2.48000496000992,
      "grad_norm": 0.7864581346511841,
      "learning_rate": 0.00017115006375947303,
      "loss": 1.6757,
      "step": 10000
    },
    {
      "epoch": 2.48248496496993,
      "grad_norm": 0.7828495502471924,
      "learning_rate": 0.00017109529138748157,
      "loss": 1.6971,
      "step": 10010
    },
    {
      "epoch": 2.4849649699299396,
      "grad_norm": 0.7776135206222534,
      "learning_rate": 0.00017104047585369343,
      "loss": 1.7392,
      "step": 10020
    },
    {
      "epoch": 2.4874449748899496,
      "grad_norm": 0.8050247430801392,
      "learning_rate": 0.00017098561719138694,
      "loss": 1.7947,
      "step": 10030
    },
    {
      "epoch": 2.4899249798499596,
      "grad_norm": 0.7671525478363037,
      "learning_rate": 0.00017093071543386668,
      "loss": 1.7134,
      "step": 10040
    },
    {
      "epoch": 2.4924049848099696,
      "grad_norm": 0.824383020401001,
      "learning_rate": 0.00017087577061446337,
      "loss": 1.7234,
      "step": 10050
    },
    {
      "epoch": 2.4948849897699796,
      "grad_norm": 0.74275803565979,
      "learning_rate": 0.00017082078276653393,
      "loss": 1.7163,
      "step": 10060
    },
    {
      "epoch": 2.4973649947299896,
      "grad_norm": 0.7931815981864929,
      "learning_rate": 0.00017076575192346137,
      "loss": 1.7078,
      "step": 10070
    },
    {
      "epoch": 2.4998449996899996,
      "grad_norm": 0.7598164081573486,
      "learning_rate": 0.00017071067811865476,
      "loss": 1.7395,
      "step": 10080
    },
    {
      "epoch": 2.502325004650009,
      "grad_norm": 0.8434139490127563,
      "learning_rate": 0.0001706555613855493,
      "loss": 1.6828,
      "step": 10090
    },
    {
      "epoch": 2.504805009610019,
      "grad_norm": 0.740534245967865,
      "learning_rate": 0.00017060040175760624,
      "loss": 1.75,
      "step": 10100
    },
    {
      "epoch": 2.507285014570029,
      "grad_norm": 0.8243404626846313,
      "learning_rate": 0.0001705451992683129,
      "loss": 1.6813,
      "step": 10110
    },
    {
      "epoch": 2.509765019530039,
      "grad_norm": 0.8587687611579895,
      "learning_rate": 0.00017048995395118252,
      "loss": 1.6835,
      "step": 10120
    },
    {
      "epoch": 2.512245024490049,
      "grad_norm": 0.7877902388572693,
      "learning_rate": 0.00017043466583975447,
      "loss": 1.6899,
      "step": 10130
    },
    {
      "epoch": 2.5147250294500587,
      "grad_norm": 0.8048937916755676,
      "learning_rate": 0.00017037933496759404,
      "loss": 1.7113,
      "step": 10140
    },
    {
      "epoch": 2.5172050344100687,
      "grad_norm": 0.8216809630393982,
      "learning_rate": 0.00017032396136829246,
      "loss": 1.7106,
      "step": 10150
    },
    {
      "epoch": 2.5196850393700787,
      "grad_norm": 0.8559943437576294,
      "learning_rate": 0.00017026854507546692,
      "loss": 1.7434,
      "step": 10160
    },
    {
      "epoch": 2.5221650443300887,
      "grad_norm": 0.8496769070625305,
      "learning_rate": 0.00017021308612276054,
      "loss": 1.7544,
      "step": 10170
    },
    {
      "epoch": 2.5246450492900987,
      "grad_norm": 0.7176957130432129,
      "learning_rate": 0.00017015758454384233,
      "loss": 1.6993,
      "step": 10180
    },
    {
      "epoch": 2.5271250542501082,
      "grad_norm": 0.788009524345398,
      "learning_rate": 0.00017010204037240718,
      "loss": 1.806,
      "step": 10190
    },
    {
      "epoch": 2.5296050592101187,
      "grad_norm": 0.7690684795379639,
      "learning_rate": 0.00017004645364217583,
      "loss": 1.7622,
      "step": 10200
    },
    {
      "epoch": 2.5320850641701282,
      "grad_norm": 0.8179535269737244,
      "learning_rate": 0.0001699908243868949,
      "loss": 1.7159,
      "step": 10210
    },
    {
      "epoch": 2.5345650691301382,
      "grad_norm": 0.8218116760253906,
      "learning_rate": 0.00016993515264033672,
      "loss": 1.7361,
      "step": 10220
    },
    {
      "epoch": 2.5370450740901482,
      "grad_norm": 0.8178150057792664,
      "learning_rate": 0.00016987943843629953,
      "loss": 1.7788,
      "step": 10230
    },
    {
      "epoch": 2.5395250790501582,
      "grad_norm": 0.7793101668357849,
      "learning_rate": 0.00016982368180860728,
      "loss": 1.7611,
      "step": 10240
    },
    {
      "epoch": 2.5420050840101682,
      "grad_norm": 0.8476938605308533,
      "learning_rate": 0.00016976788279110975,
      "loss": 1.7419,
      "step": 10250
    },
    {
      "epoch": 2.5444850889701778,
      "grad_norm": 0.8483054041862488,
      "learning_rate": 0.00016971204141768233,
      "loss": 1.7384,
      "step": 10260
    },
    {
      "epoch": 2.5469650939301878,
      "grad_norm": 0.7705952525138855,
      "learning_rate": 0.00016965615772222625,
      "loss": 1.7456,
      "step": 10270
    },
    {
      "epoch": 2.5494450988901978,
      "grad_norm": 0.8043239712715149,
      "learning_rate": 0.00016960023173866835,
      "loss": 1.747,
      "step": 10280
    },
    {
      "epoch": 2.5519251038502078,
      "grad_norm": 0.8316717147827148,
      "learning_rate": 0.00016954426350096116,
      "loss": 1.7278,
      "step": 10290
    },
    {
      "epoch": 2.5544051088102178,
      "grad_norm": 0.7975428104400635,
      "learning_rate": 0.00016948825304308293,
      "loss": 1.7185,
      "step": 10300
    },
    {
      "epoch": 2.5568851137702273,
      "grad_norm": 0.8466578722000122,
      "learning_rate": 0.00016943220039903744,
      "loss": 1.7281,
      "step": 10310
    },
    {
      "epoch": 2.5593651187302373,
      "grad_norm": 0.784413754940033,
      "learning_rate": 0.00016937610560285418,
      "loss": 1.7667,
      "step": 10320
    },
    {
      "epoch": 2.5618451236902473,
      "grad_norm": 0.7354859113693237,
      "learning_rate": 0.0001693199686885881,
      "loss": 1.6981,
      "step": 10330
    },
    {
      "epoch": 2.5643251286502573,
      "grad_norm": 0.830909013748169,
      "learning_rate": 0.00016926378969031987,
      "loss": 1.7656,
      "step": 10340
    },
    {
      "epoch": 2.5668051336102673,
      "grad_norm": 0.7383814454078674,
      "learning_rate": 0.00016920756864215557,
      "loss": 1.7628,
      "step": 10350
    },
    {
      "epoch": 2.5692851385702773,
      "grad_norm": 0.7582780122756958,
      "learning_rate": 0.00016915130557822695,
      "loss": 1.7609,
      "step": 10360
    },
    {
      "epoch": 2.5717651435302873,
      "grad_norm": 0.8444397449493408,
      "learning_rate": 0.00016909500053269113,
      "loss": 1.7922,
      "step": 10370
    },
    {
      "epoch": 2.574245148490297,
      "grad_norm": 0.8196545243263245,
      "learning_rate": 0.00016903865353973084,
      "loss": 1.7598,
      "step": 10380
    },
    {
      "epoch": 2.576725153450307,
      "grad_norm": 0.8253687024116516,
      "learning_rate": 0.00016898226463355417,
      "loss": 1.7142,
      "step": 10390
    },
    {
      "epoch": 2.579205158410317,
      "grad_norm": 0.93372642993927,
      "learning_rate": 0.0001689258338483947,
      "loss": 1.7048,
      "step": 10400
    },
    {
      "epoch": 2.581685163370327,
      "grad_norm": 0.8133403658866882,
      "learning_rate": 0.0001688693612185115,
      "loss": 1.7031,
      "step": 10410
    },
    {
      "epoch": 2.584165168330337,
      "grad_norm": 0.7580595016479492,
      "learning_rate": 0.00016881284677818892,
      "loss": 1.7277,
      "step": 10420
    },
    {
      "epoch": 2.5866451732903464,
      "grad_norm": 0.7730793952941895,
      "learning_rate": 0.00016875629056173675,
      "loss": 1.7465,
      "step": 10430
    },
    {
      "epoch": 2.5891251782503564,
      "grad_norm": 0.8139658570289612,
      "learning_rate": 0.00016869969260349018,
      "loss": 1.7008,
      "step": 10440
    },
    {
      "epoch": 2.5916051832103664,
      "grad_norm": 0.7631027102470398,
      "learning_rate": 0.00016864305293780966,
      "loss": 1.8093,
      "step": 10450
    },
    {
      "epoch": 2.5940851881703764,
      "grad_norm": 0.7781238555908203,
      "learning_rate": 0.00016858637159908107,
      "loss": 1.7592,
      "step": 10460
    },
    {
      "epoch": 2.5965651931303864,
      "grad_norm": 0.7929747104644775,
      "learning_rate": 0.00016852964862171553,
      "loss": 1.6993,
      "step": 10470
    },
    {
      "epoch": 2.599045198090396,
      "grad_norm": 0.828659176826477,
      "learning_rate": 0.00016847288404014935,
      "loss": 1.6902,
      "step": 10480
    },
    {
      "epoch": 2.6015252030504064,
      "grad_norm": 0.8091562986373901,
      "learning_rate": 0.00016841607788884425,
      "loss": 1.7654,
      "step": 10490
    },
    {
      "epoch": 2.604005208010416,
      "grad_norm": 0.7710474729537964,
      "learning_rate": 0.00016835923020228712,
      "loss": 1.7146,
      "step": 10500
    },
    {
      "epoch": 2.606485212970426,
      "grad_norm": 0.8594821095466614,
      "learning_rate": 0.00016830234101499007,
      "loss": 1.7195,
      "step": 10510
    },
    {
      "epoch": 2.608965217930436,
      "grad_norm": 0.7700628638267517,
      "learning_rate": 0.00016824541036149037,
      "loss": 1.6829,
      "step": 10520
    },
    {
      "epoch": 2.611445222890446,
      "grad_norm": 0.852709174156189,
      "learning_rate": 0.0001681884382763505,
      "loss": 1.7721,
      "step": 10530
    },
    {
      "epoch": 2.613925227850456,
      "grad_norm": 0.8819437623023987,
      "learning_rate": 0.00016813142479415812,
      "loss": 1.6806,
      "step": 10540
    },
    {
      "epoch": 2.6164052328104654,
      "grad_norm": 0.8042049407958984,
      "learning_rate": 0.000168074369949526,
      "loss": 1.712,
      "step": 10550
    },
    {
      "epoch": 2.6188852377704754,
      "grad_norm": 0.8049604296684265,
      "learning_rate": 0.00016801727377709194,
      "loss": 1.7204,
      "step": 10560
    },
    {
      "epoch": 2.6213652427304854,
      "grad_norm": 0.8116569519042969,
      "learning_rate": 0.00016796013631151897,
      "loss": 1.684,
      "step": 10570
    },
    {
      "epoch": 2.6238452476904954,
      "grad_norm": 0.718235969543457,
      "learning_rate": 0.0001679029575874951,
      "loss": 1.7264,
      "step": 10580
    },
    {
      "epoch": 2.6263252526505054,
      "grad_norm": 0.9108774065971375,
      "learning_rate": 0.0001678457376397334,
      "loss": 1.7857,
      "step": 10590
    },
    {
      "epoch": 2.628805257610515,
      "grad_norm": 0.7946862578392029,
      "learning_rate": 0.00016778847650297197,
      "loss": 1.6848,
      "step": 10600
    },
    {
      "epoch": 2.6312852625705254,
      "grad_norm": 0.8489319682121277,
      "learning_rate": 0.00016773117421197396,
      "loss": 1.7319,
      "step": 10610
    },
    {
      "epoch": 2.633765267530535,
      "grad_norm": 0.7626186609268188,
      "learning_rate": 0.00016767383080152742,
      "loss": 1.7554,
      "step": 10620
    },
    {
      "epoch": 2.636245272490545,
      "grad_norm": 0.8287044763565063,
      "learning_rate": 0.00016761644630644546,
      "loss": 1.7215,
      "step": 10630
    },
    {
      "epoch": 2.638725277450555,
      "grad_norm": 0.769493043422699,
      "learning_rate": 0.00016755902076156604,
      "loss": 1.7282,
      "step": 10640
    },
    {
      "epoch": 2.641205282410565,
      "grad_norm": 0.770527184009552,
      "learning_rate": 0.00016750155420175208,
      "loss": 1.7322,
      "step": 10650
    },
    {
      "epoch": 2.643685287370575,
      "grad_norm": 0.8423058986663818,
      "learning_rate": 0.00016744404666189144,
      "loss": 1.7149,
      "step": 10660
    },
    {
      "epoch": 2.6461652923305845,
      "grad_norm": 0.8349190354347229,
      "learning_rate": 0.0001673864981768968,
      "loss": 1.743,
      "step": 10670
    },
    {
      "epoch": 2.6486452972905945,
      "grad_norm": 0.8283718824386597,
      "learning_rate": 0.0001673289087817057,
      "loss": 1.7103,
      "step": 10680
    },
    {
      "epoch": 2.6511253022506045,
      "grad_norm": 0.7439224123954773,
      "learning_rate": 0.00016727127851128056,
      "loss": 1.7867,
      "step": 10690
    },
    {
      "epoch": 2.6536053072106145,
      "grad_norm": 0.8084254264831543,
      "learning_rate": 0.00016721360740060864,
      "loss": 1.7138,
      "step": 10700
    },
    {
      "epoch": 2.6560853121706245,
      "grad_norm": 0.7427732944488525,
      "learning_rate": 0.00016715589548470185,
      "loss": 1.7044,
      "step": 10710
    },
    {
      "epoch": 2.658565317130634,
      "grad_norm": 0.7997560501098633,
      "learning_rate": 0.00016709814279859702,
      "loss": 1.6955,
      "step": 10720
    },
    {
      "epoch": 2.661045322090644,
      "grad_norm": 0.8720448017120361,
      "learning_rate": 0.0001670403493773557,
      "loss": 1.7229,
      "step": 10730
    },
    {
      "epoch": 2.663525327050654,
      "grad_norm": 0.8524209856987,
      "learning_rate": 0.0001669825152560641,
      "loss": 1.7147,
      "step": 10740
    },
    {
      "epoch": 2.666005332010664,
      "grad_norm": 0.7974799871444702,
      "learning_rate": 0.00016692464046983322,
      "loss": 1.7396,
      "step": 10750
    },
    {
      "epoch": 2.668485336970674,
      "grad_norm": 0.7441201210021973,
      "learning_rate": 0.0001668667250537987,
      "loss": 1.7124,
      "step": 10760
    },
    {
      "epoch": 2.670965341930684,
      "grad_norm": 0.8221575617790222,
      "learning_rate": 0.0001668087690431209,
      "loss": 1.7115,
      "step": 10770
    },
    {
      "epoch": 2.673445346890694,
      "grad_norm": 0.799111545085907,
      "learning_rate": 0.00016675077247298474,
      "loss": 1.727,
      "step": 10780
    },
    {
      "epoch": 2.6759253518507036,
      "grad_norm": 0.8170874714851379,
      "learning_rate": 0.00016669273537859986,
      "loss": 1.6521,
      "step": 10790
    },
    {
      "epoch": 2.6784053568107136,
      "grad_norm": 0.8116423487663269,
      "learning_rate": 0.0001666346577952004,
      "loss": 1.6758,
      "step": 10800
    },
    {
      "epoch": 2.6808853617707236,
      "grad_norm": 0.7921115159988403,
      "learning_rate": 0.00016657653975804516,
      "loss": 1.6891,
      "step": 10810
    },
    {
      "epoch": 2.6833653667307336,
      "grad_norm": 0.8088225722312927,
      "learning_rate": 0.0001665183813024175,
      "loss": 1.7268,
      "step": 10820
    },
    {
      "epoch": 2.6858453716907436,
      "grad_norm": 0.8141648173332214,
      "learning_rate": 0.00016646018246362527,
      "loss": 1.7644,
      "step": 10830
    },
    {
      "epoch": 2.688325376650753,
      "grad_norm": 0.8276552557945251,
      "learning_rate": 0.00016640194327700086,
      "loss": 1.7645,
      "step": 10840
    },
    {
      "epoch": 2.690805381610763,
      "grad_norm": 0.7764387130737305,
      "learning_rate": 0.00016634366377790114,
      "loss": 1.6971,
      "step": 10850
    },
    {
      "epoch": 2.693285386570773,
      "grad_norm": 0.9159328937530518,
      "learning_rate": 0.00016628534400170745,
      "loss": 1.6994,
      "step": 10860
    },
    {
      "epoch": 2.695765391530783,
      "grad_norm": 0.8896689414978027,
      "learning_rate": 0.00016622698398382566,
      "loss": 1.7405,
      "step": 10870
    },
    {
      "epoch": 2.698245396490793,
      "grad_norm": 0.7660658359527588,
      "learning_rate": 0.00016616858375968595,
      "loss": 1.7525,
      "step": 10880
    },
    {
      "epoch": 2.7007254014508026,
      "grad_norm": 0.7988965511322021,
      "learning_rate": 0.00016611014336474303,
      "loss": 1.7569,
      "step": 10890
    },
    {
      "epoch": 2.703205406410813,
      "grad_norm": 0.7468309998512268,
      "learning_rate": 0.00016605166283447585,
      "loss": 1.7449,
      "step": 10900
    },
    {
      "epoch": 2.7056854113708226,
      "grad_norm": 0.7522343993186951,
      "learning_rate": 0.00016599314220438794,
      "loss": 1.7165,
      "step": 10910
    },
    {
      "epoch": 2.7081654163308326,
      "grad_norm": 0.8288154006004333,
      "learning_rate": 0.00016593458151000688,
      "loss": 1.7,
      "step": 10920
    },
    {
      "epoch": 2.7106454212908426,
      "grad_norm": 0.7801446914672852,
      "learning_rate": 0.0001658759807868849,
      "loss": 1.6664,
      "step": 10930
    },
    {
      "epoch": 2.7131254262508526,
      "grad_norm": 0.8585086464881897,
      "learning_rate": 0.00016581734007059827,
      "loss": 1.7183,
      "step": 10940
    },
    {
      "epoch": 2.7156054312108626,
      "grad_norm": 0.818763017654419,
      "learning_rate": 0.00016575865939674772,
      "loss": 1.7009,
      "step": 10950
    },
    {
      "epoch": 2.718085436170872,
      "grad_norm": 0.7211923003196716,
      "learning_rate": 0.00016569993880095806,
      "loss": 1.6735,
      "step": 10960
    },
    {
      "epoch": 2.720565441130882,
      "grad_norm": 0.7824010252952576,
      "learning_rate": 0.00016564117831887852,
      "loss": 1.6974,
      "step": 10970
    },
    {
      "epoch": 2.723045446090892,
      "grad_norm": 0.8001604080200195,
      "learning_rate": 0.00016558237798618245,
      "loss": 1.6868,
      "step": 10980
    },
    {
      "epoch": 2.725525451050902,
      "grad_norm": 0.7506614327430725,
      "learning_rate": 0.00016552353783856732,
      "loss": 1.6618,
      "step": 10990
    },
    {
      "epoch": 2.728005456010912,
      "grad_norm": 0.774455189704895,
      "learning_rate": 0.00016546465791175496,
      "loss": 1.7212,
      "step": 11000
    },
    {
      "epoch": 2.7304854609709217,
      "grad_norm": 0.7329303622245789,
      "learning_rate": 0.00016540573824149122,
      "loss": 1.7457,
      "step": 11010
    },
    {
      "epoch": 2.7329654659309317,
      "grad_norm": 0.7539793252944946,
      "learning_rate": 0.00016534677886354606,
      "loss": 1.7019,
      "step": 11020
    },
    {
      "epoch": 2.7354454708909417,
      "grad_norm": 0.8947096467018127,
      "learning_rate": 0.00016528777981371365,
      "loss": 1.7468,
      "step": 11030
    },
    {
      "epoch": 2.7379254758509517,
      "grad_norm": 0.7925174832344055,
      "learning_rate": 0.00016522874112781213,
      "loss": 1.702,
      "step": 11040
    },
    {
      "epoch": 2.7404054808109617,
      "grad_norm": 0.7956472635269165,
      "learning_rate": 0.00016516966284168378,
      "loss": 1.7347,
      "step": 11050
    },
    {
      "epoch": 2.7428854857709717,
      "grad_norm": 0.7949102520942688,
      "learning_rate": 0.0001651105449911949,
      "loss": 1.7275,
      "step": 11060
    },
    {
      "epoch": 2.7453654907309817,
      "grad_norm": 0.8082578182220459,
      "learning_rate": 0.00016505138761223584,
      "loss": 1.7234,
      "step": 11070
    },
    {
      "epoch": 2.7478454956909912,
      "grad_norm": 0.7731408476829529,
      "learning_rate": 0.00016499219074072086,
      "loss": 1.7376,
      "step": 11080
    },
    {
      "epoch": 2.7503255006510012,
      "grad_norm": 0.8738985061645508,
      "learning_rate": 0.00016493295441258828,
      "loss": 1.8141,
      "step": 11090
    },
    {
      "epoch": 2.7528055056110112,
      "grad_norm": 0.7660444974899292,
      "learning_rate": 0.00016487367866380037,
      "loss": 1.6608,
      "step": 11100
    },
    {
      "epoch": 2.7552855105710212,
      "grad_norm": 0.8163630366325378,
      "learning_rate": 0.00016481436353034325,
      "loss": 1.7539,
      "step": 11110
    },
    {
      "epoch": 2.757765515531031,
      "grad_norm": 0.8122231960296631,
      "learning_rate": 0.00016475500904822706,
      "loss": 1.714,
      "step": 11120
    },
    {
      "epoch": 2.7602455204910408,
      "grad_norm": 0.7877174615859985,
      "learning_rate": 0.00016469561525348577,
      "loss": 1.7644,
      "step": 11130
    },
    {
      "epoch": 2.7627255254510508,
      "grad_norm": 0.8161805272102356,
      "learning_rate": 0.00016463618218217718,
      "loss": 1.6885,
      "step": 11140
    },
    {
      "epoch": 2.7652055304110608,
      "grad_norm": 0.7412929534912109,
      "learning_rate": 0.00016457670987038298,
      "loss": 1.6591,
      "step": 11150
    },
    {
      "epoch": 2.7676855353710708,
      "grad_norm": 0.8340774774551392,
      "learning_rate": 0.00016451719835420877,
      "loss": 1.65,
      "step": 11160
    },
    {
      "epoch": 2.7701655403310808,
      "grad_norm": 0.8444342613220215,
      "learning_rate": 0.00016445764766978379,
      "loss": 1.7755,
      "step": 11170
    },
    {
      "epoch": 2.7726455452910903,
      "grad_norm": 0.9029738306999207,
      "learning_rate": 0.00016439805785326112,
      "loss": 1.688,
      "step": 11180
    },
    {
      "epoch": 2.7751255502511007,
      "grad_norm": 0.8989600539207458,
      "learning_rate": 0.00016433842894081764,
      "loss": 1.7138,
      "step": 11190
    },
    {
      "epoch": 2.7776055552111103,
      "grad_norm": 0.8586159348487854,
      "learning_rate": 0.00016427876096865394,
      "loss": 1.7075,
      "step": 11200
    },
    {
      "epoch": 2.7800855601711203,
      "grad_norm": 0.7890956401824951,
      "learning_rate": 0.0001642190539729943,
      "loss": 1.7397,
      "step": 11210
    },
    {
      "epoch": 2.7825655651311303,
      "grad_norm": 0.8047048449516296,
      "learning_rate": 0.00016415930799008668,
      "loss": 1.7553,
      "step": 11220
    },
    {
      "epoch": 2.7850455700911403,
      "grad_norm": 0.777072548866272,
      "learning_rate": 0.0001640995230562028,
      "loss": 1.767,
      "step": 11230
    },
    {
      "epoch": 2.7875255750511503,
      "grad_norm": 0.7799883484840393,
      "learning_rate": 0.00016403969920763788,
      "loss": 1.6753,
      "step": 11240
    },
    {
      "epoch": 2.79000558001116,
      "grad_norm": 0.8299691081047058,
      "learning_rate": 0.00016397983648071095,
      "loss": 1.7417,
      "step": 11250
    },
    {
      "epoch": 2.79248558497117,
      "grad_norm": 0.7810803651809692,
      "learning_rate": 0.00016391993491176446,
      "loss": 1.7456,
      "step": 11260
    },
    {
      "epoch": 2.79496558993118,
      "grad_norm": 0.7570012807846069,
      "learning_rate": 0.00016385999453716454,
      "loss": 1.6847,
      "step": 11270
    },
    {
      "epoch": 2.79744559489119,
      "grad_norm": 0.7709137201309204,
      "learning_rate": 0.0001638000153933009,
      "loss": 1.7746,
      "step": 11280
    },
    {
      "epoch": 2.7999255998512,
      "grad_norm": 0.9917616248130798,
      "learning_rate": 0.00016373999751658665,
      "loss": 1.7129,
      "step": 11290
    },
    {
      "epoch": 2.8024056048112094,
      "grad_norm": 0.8247378468513489,
      "learning_rate": 0.00016367994094345864,
      "loss": 1.7178,
      "step": 11300
    },
    {
      "epoch": 2.80488560977122,
      "grad_norm": 0.8372973203659058,
      "learning_rate": 0.000163619845710377,
      "loss": 1.6927,
      "step": 11310
    },
    {
      "epoch": 2.8073656147312294,
      "grad_norm": 0.8099014759063721,
      "learning_rate": 0.00016355971185382545,
      "loss": 1.7366,
      "step": 11320
    },
    {
      "epoch": 2.8098456196912394,
      "grad_norm": 0.7249281406402588,
      "learning_rate": 0.00016349953941031112,
      "loss": 1.7108,
      "step": 11330
    },
    {
      "epoch": 2.8123256246512494,
      "grad_norm": 0.8013809323310852,
      "learning_rate": 0.00016343932841636456,
      "loss": 1.7111,
      "step": 11340
    },
    {
      "epoch": 2.8148056296112594,
      "grad_norm": 0.8951674103736877,
      "learning_rate": 0.00016337907890853973,
      "loss": 1.7322,
      "step": 11350
    },
    {
      "epoch": 2.8172856345712693,
      "grad_norm": 0.7558394074440002,
      "learning_rate": 0.000163318790923414,
      "loss": 1.8479,
      "step": 11360
    },
    {
      "epoch": 2.819765639531279,
      "grad_norm": 0.7628364562988281,
      "learning_rate": 0.00016325846449758805,
      "loss": 1.7534,
      "step": 11370
    },
    {
      "epoch": 2.822245644491289,
      "grad_norm": 0.8447631597518921,
      "learning_rate": 0.0001631980996676859,
      "loss": 1.7115,
      "step": 11380
    },
    {
      "epoch": 2.824725649451299,
      "grad_norm": 0.7334152460098267,
      "learning_rate": 0.00016313769647035496,
      "loss": 1.7544,
      "step": 11390
    },
    {
      "epoch": 2.827205654411309,
      "grad_norm": 0.8144456148147583,
      "learning_rate": 0.00016307725494226584,
      "loss": 1.6882,
      "step": 11400
    },
    {
      "epoch": 2.829685659371319,
      "grad_norm": 0.8416145443916321,
      "learning_rate": 0.00016301677512011248,
      "loss": 1.7176,
      "step": 11410
    },
    {
      "epoch": 2.8321656643313284,
      "grad_norm": 0.8626994490623474,
      "learning_rate": 0.00016295625704061204,
      "loss": 1.7225,
      "step": 11420
    },
    {
      "epoch": 2.8346456692913384,
      "grad_norm": 0.8473905920982361,
      "learning_rate": 0.00016289570074050493,
      "loss": 1.6883,
      "step": 11430
    },
    {
      "epoch": 2.8371256742513484,
      "grad_norm": 0.7284490466117859,
      "learning_rate": 0.00016283510625655472,
      "loss": 1.6837,
      "step": 11440
    },
    {
      "epoch": 2.8396056792113584,
      "grad_norm": 0.8658609986305237,
      "learning_rate": 0.00016277447362554823,
      "loss": 1.7612,
      "step": 11450
    },
    {
      "epoch": 2.8420856841713684,
      "grad_norm": 0.7673606872558594,
      "learning_rate": 0.00016271380288429533,
      "loss": 1.7423,
      "step": 11460
    },
    {
      "epoch": 2.8445656891313784,
      "grad_norm": 0.8407950401306152,
      "learning_rate": 0.0001626530940696292,
      "loss": 1.7509,
      "step": 11470
    },
    {
      "epoch": 2.8470456940913884,
      "grad_norm": 0.8333432078361511,
      "learning_rate": 0.00016259234721840591,
      "loss": 1.6779,
      "step": 11480
    },
    {
      "epoch": 2.849525699051398,
      "grad_norm": 0.7581879496574402,
      "learning_rate": 0.00016253156236750485,
      "loss": 1.7077,
      "step": 11490
    },
    {
      "epoch": 2.852005704011408,
      "grad_norm": 0.7580318450927734,
      "learning_rate": 0.0001624707395538283,
      "loss": 1.7709,
      "step": 11500
    },
    {
      "epoch": 2.854485708971418,
      "grad_norm": 0.7712346911430359,
      "learning_rate": 0.00016240987881430163,
      "loss": 1.7841,
      "step": 11510
    },
    {
      "epoch": 2.856965713931428,
      "grad_norm": 0.7301474809646606,
      "learning_rate": 0.00016234898018587337,
      "loss": 1.6845,
      "step": 11520
    },
    {
      "epoch": 2.859445718891438,
      "grad_norm": 0.7858560681343079,
      "learning_rate": 0.00016228804370551486,
      "loss": 1.7104,
      "step": 11530
    },
    {
      "epoch": 2.8619257238514475,
      "grad_norm": 0.7861492037773132,
      "learning_rate": 0.00016222706941022055,
      "loss": 1.7292,
      "step": 11540
    },
    {
      "epoch": 2.8644057288114575,
      "grad_norm": 0.8644739389419556,
      "learning_rate": 0.00016216605733700775,
      "loss": 1.7029,
      "step": 11550
    },
    {
      "epoch": 2.8668857337714675,
      "grad_norm": 0.801563560962677,
      "learning_rate": 0.0001621050075229168,
      "loss": 1.7611,
      "step": 11560
    },
    {
      "epoch": 2.8693657387314775,
      "grad_norm": 0.768110454082489,
      "learning_rate": 0.00016204392000501093,
      "loss": 1.7673,
      "step": 11570
    },
    {
      "epoch": 2.8718457436914875,
      "grad_norm": 0.8002123832702637,
      "learning_rate": 0.00016198279482037618,
      "loss": 1.7176,
      "step": 11580
    },
    {
      "epoch": 2.874325748651497,
      "grad_norm": 0.8760389685630798,
      "learning_rate": 0.00016192163200612155,
      "loss": 1.734,
      "step": 11590
    },
    {
      "epoch": 2.8768057536115075,
      "grad_norm": 0.7727819681167603,
      "learning_rate": 0.00016186043159937882,
      "loss": 1.7203,
      "step": 11600
    },
    {
      "epoch": 2.879285758571517,
      "grad_norm": 0.7959619164466858,
      "learning_rate": 0.00016179919363730266,
      "loss": 1.788,
      "step": 11610
    },
    {
      "epoch": 2.881765763531527,
      "grad_norm": 0.8371580839157104,
      "learning_rate": 0.00016173791815707051,
      "loss": 1.725,
      "step": 11620
    },
    {
      "epoch": 2.884245768491537,
      "grad_norm": 0.7960420846939087,
      "learning_rate": 0.00016167660519588256,
      "loss": 1.7278,
      "step": 11630
    },
    {
      "epoch": 2.886725773451547,
      "grad_norm": 0.7642151117324829,
      "learning_rate": 0.00016161525479096178,
      "loss": 1.7459,
      "step": 11640
    },
    {
      "epoch": 2.889205778411557,
      "grad_norm": 0.8472262620925903,
      "learning_rate": 0.00016155386697955387,
      "loss": 1.7484,
      "step": 11650
    },
    {
      "epoch": 2.8916857833715666,
      "grad_norm": 0.7980145812034607,
      "learning_rate": 0.0001614924417989272,
      "loss": 1.7002,
      "step": 11660
    },
    {
      "epoch": 2.8941657883315766,
      "grad_norm": 0.888417661190033,
      "learning_rate": 0.0001614309792863729,
      "loss": 1.6796,
      "step": 11670
    },
    {
      "epoch": 2.8966457932915866,
      "grad_norm": 0.8283897042274475,
      "learning_rate": 0.00016136947947920476,
      "loss": 1.7446,
      "step": 11680
    },
    {
      "epoch": 2.8991257982515966,
      "grad_norm": 0.7665685415267944,
      "learning_rate": 0.0001613079424147591,
      "loss": 1.8176,
      "step": 11690
    },
    {
      "epoch": 2.9016058032116065,
      "grad_norm": 0.7955074310302734,
      "learning_rate": 0.00016124636813039502,
      "loss": 1.7458,
      "step": 11700
    },
    {
      "epoch": 2.904085808171616,
      "grad_norm": 0.7898004651069641,
      "learning_rate": 0.00016118475666349408,
      "loss": 1.6708,
      "step": 11710
    },
    {
      "epoch": 2.906565813131626,
      "grad_norm": 0.8602806329727173,
      "learning_rate": 0.0001611231080514605,
      "loss": 1.7355,
      "step": 11720
    },
    {
      "epoch": 2.909045818091636,
      "grad_norm": 0.7743834853172302,
      "learning_rate": 0.00016106142233172097,
      "loss": 1.719,
      "step": 11730
    },
    {
      "epoch": 2.911525823051646,
      "grad_norm": 0.8028146624565125,
      "learning_rate": 0.0001609996995417248,
      "loss": 1.7753,
      "step": 11740
    },
    {
      "epoch": 2.914005828011656,
      "grad_norm": 0.7885788679122925,
      "learning_rate": 0.00016093793971894374,
      "loss": 1.73,
      "step": 11750
    },
    {
      "epoch": 2.916485832971666,
      "grad_norm": 1.0257158279418945,
      "learning_rate": 0.00016087614290087208,
      "loss": 1.7334,
      "step": 11760
    },
    {
      "epoch": 2.918965837931676,
      "grad_norm": 0.8035752177238464,
      "learning_rate": 0.00016081430912502648,
      "loss": 1.7246,
      "step": 11770
    },
    {
      "epoch": 2.9214458428916856,
      "grad_norm": 0.8140546083450317,
      "learning_rate": 0.00016075243842894615,
      "loss": 1.7036,
      "step": 11780
    },
    {
      "epoch": 2.9239258478516956,
      "grad_norm": 0.829397976398468,
      "learning_rate": 0.00016069053085019256,
      "loss": 1.7177,
      "step": 11790
    },
    {
      "epoch": 2.9264058528117056,
      "grad_norm": 0.8287886381149292,
      "learning_rate": 0.0001606285864263498,
      "loss": 1.68,
      "step": 11800
    },
    {
      "epoch": 2.9288858577717156,
      "grad_norm": 0.8033359050750732,
      "learning_rate": 0.0001605666051950241,
      "loss": 1.7039,
      "step": 11810
    },
    {
      "epoch": 2.9313658627317256,
      "grad_norm": 0.7789167761802673,
      "learning_rate": 0.00016050458719384412,
      "loss": 1.7292,
      "step": 11820
    },
    {
      "epoch": 2.933845867691735,
      "grad_norm": 0.7982509136199951,
      "learning_rate": 0.0001604425324604609,
      "loss": 1.7573,
      "step": 11830
    },
    {
      "epoch": 2.936325872651745,
      "grad_norm": 0.8048049807548523,
      "learning_rate": 0.00016038044103254775,
      "loss": 1.7235,
      "step": 11840
    },
    {
      "epoch": 2.938805877611755,
      "grad_norm": 0.8201325535774231,
      "learning_rate": 0.0001603183129478002,
      "loss": 1.8011,
      "step": 11850
    },
    {
      "epoch": 2.941285882571765,
      "grad_norm": 0.7818323373794556,
      "learning_rate": 0.00016025614824393607,
      "loss": 1.6917,
      "step": 11860
    },
    {
      "epoch": 2.943765887531775,
      "grad_norm": 0.8268362283706665,
      "learning_rate": 0.00016019394695869545,
      "loss": 1.7088,
      "step": 11870
    },
    {
      "epoch": 2.946245892491785,
      "grad_norm": 0.7949564456939697,
      "learning_rate": 0.00016013170912984058,
      "loss": 1.7006,
      "step": 11880
    },
    {
      "epoch": 2.948725897451795,
      "grad_norm": 0.8073463439941406,
      "learning_rate": 0.00016006943479515595,
      "loss": 1.6718,
      "step": 11890
    },
    {
      "epoch": 2.9512059024118047,
      "grad_norm": 0.7876682877540588,
      "learning_rate": 0.0001600071239924481,
      "loss": 1.7143,
      "step": 11900
    },
    {
      "epoch": 2.9536859073718147,
      "grad_norm": 0.8282910585403442,
      "learning_rate": 0.00015994477675954584,
      "loss": 1.73,
      "step": 11910
    },
    {
      "epoch": 2.9561659123318247,
      "grad_norm": 0.7657924890518188,
      "learning_rate": 0.00015988239313430004,
      "loss": 1.7406,
      "step": 11920
    },
    {
      "epoch": 2.9586459172918347,
      "grad_norm": 0.8015568256378174,
      "learning_rate": 0.00015981997315458362,
      "loss": 1.7657,
      "step": 11930
    },
    {
      "epoch": 2.9611259222518447,
      "grad_norm": 0.8141234517097473,
      "learning_rate": 0.00015975751685829166,
      "loss": 1.6911,
      "step": 11940
    },
    {
      "epoch": 2.9636059272118542,
      "grad_norm": 0.7737033367156982,
      "learning_rate": 0.00015969502428334122,
      "loss": 1.723,
      "step": 11950
    },
    {
      "epoch": 2.9660859321718642,
      "grad_norm": 0.7937765121459961,
      "learning_rate": 0.00015963249546767144,
      "loss": 1.7345,
      "step": 11960
    },
    {
      "epoch": 2.968565937131874,
      "grad_norm": 0.7321860790252686,
      "learning_rate": 0.00015956993044924334,
      "loss": 1.7622,
      "step": 11970
    },
    {
      "epoch": 2.971045942091884,
      "grad_norm": 0.7269527912139893,
      "learning_rate": 0.0001595073292660401,
      "loss": 1.7536,
      "step": 11980
    },
    {
      "epoch": 2.973525947051894,
      "grad_norm": 0.7339338064193726,
      "learning_rate": 0.00015944469195606673,
      "loss": 1.7122,
      "step": 11990
    },
    {
      "epoch": 2.9760059520119038,
      "grad_norm": 0.7892552614212036,
      "learning_rate": 0.00015938201855735014,
      "loss": 1.7115,
      "step": 12000
    },
    {
      "epoch": 2.978485956971914,
      "grad_norm": 0.7363120317459106,
      "learning_rate": 0.00015931930910793938,
      "loss": 1.6846,
      "step": 12010
    },
    {
      "epoch": 2.9809659619319238,
      "grad_norm": 0.8925533890724182,
      "learning_rate": 0.00015925656364590505,
      "loss": 1.7447,
      "step": 12020
    },
    {
      "epoch": 2.9834459668919338,
      "grad_norm": 0.8083146214485168,
      "learning_rate": 0.00015919378220933985,
      "loss": 1.7259,
      "step": 12030
    },
    {
      "epoch": 2.9859259718519437,
      "grad_norm": 0.7717840671539307,
      "learning_rate": 0.00015913096483635824,
      "loss": 1.713,
      "step": 12040
    },
    {
      "epoch": 2.9884059768119537,
      "grad_norm": 0.7873446345329285,
      "learning_rate": 0.00015906811156509655,
      "loss": 1.6912,
      "step": 12050
    },
    {
      "epoch": 2.9908859817719637,
      "grad_norm": 0.8484950065612793,
      "learning_rate": 0.00015900522243371282,
      "loss": 1.698,
      "step": 12060
    },
    {
      "epoch": 2.9933659867319733,
      "grad_norm": 0.8505200743675232,
      "learning_rate": 0.00015894229748038695,
      "loss": 1.7299,
      "step": 12070
    },
    {
      "epoch": 2.9958459916919833,
      "grad_norm": 0.7281607389450073,
      "learning_rate": 0.00015887933674332046,
      "loss": 1.6884,
      "step": 12080
    },
    {
      "epoch": 2.9983259966519933,
      "grad_norm": 0.7407076954841614,
      "learning_rate": 0.00015881634026073676,
      "loss": 1.6939,
      "step": 12090
    },
    {
      "epoch": 3.0008060016120033,
      "grad_norm": 0.8531381487846375,
      "learning_rate": 0.00015875330807088087,
      "loss": 1.7431,
      "step": 12100
    },
    {
      "epoch": 3.0032860065720133,
      "grad_norm": 0.7499832510948181,
      "learning_rate": 0.00015869024021201948,
      "loss": 1.6307,
      "step": 12110
    },
    {
      "epoch": 3.0057660115320233,
      "grad_norm": 0.7780420184135437,
      "learning_rate": 0.00015862713672244093,
      "loss": 1.5834,
      "step": 12120
    },
    {
      "epoch": 3.008246016492033,
      "grad_norm": 0.8454210758209229,
      "learning_rate": 0.00015856399764045518,
      "loss": 1.605,
      "step": 12130
    },
    {
      "epoch": 3.010726021452043,
      "grad_norm": 0.8125146627426147,
      "learning_rate": 0.00015850082300439394,
      "loss": 1.6195,
      "step": 12140
    },
    {
      "epoch": 3.013206026412053,
      "grad_norm": 0.8854175806045532,
      "learning_rate": 0.00015843761285261028,
      "loss": 1.6233,
      "step": 12150
    },
    {
      "epoch": 3.015686031372063,
      "grad_norm": 0.856911301612854,
      "learning_rate": 0.000158374367223479,
      "loss": 1.5868,
      "step": 12160
    },
    {
      "epoch": 3.018166036332073,
      "grad_norm": 0.8482590317726135,
      "learning_rate": 0.00015831108615539638,
      "loss": 1.59,
      "step": 12170
    },
    {
      "epoch": 3.020646041292083,
      "grad_norm": 0.8031054139137268,
      "learning_rate": 0.00015824776968678024,
      "loss": 1.5753,
      "step": 12180
    },
    {
      "epoch": 3.0231260462520924,
      "grad_norm": 0.8475634455680847,
      "learning_rate": 0.0001581844178560698,
      "loss": 1.6346,
      "step": 12190
    },
    {
      "epoch": 3.0256060512121024,
      "grad_norm": 0.886826753616333,
      "learning_rate": 0.0001581210307017259,
      "loss": 1.6747,
      "step": 12200
    },
    {
      "epoch": 3.0280860561721124,
      "grad_norm": 0.8088538646697998,
      "learning_rate": 0.0001580576082622307,
      "loss": 1.553,
      "step": 12210
    },
    {
      "epoch": 3.0305660611321223,
      "grad_norm": 0.7956720590591431,
      "learning_rate": 0.00015799415057608785,
      "loss": 1.5824,
      "step": 12220
    },
    {
      "epoch": 3.0330460660921323,
      "grad_norm": 0.8272199630737305,
      "learning_rate": 0.00015793065768182233,
      "loss": 1.6197,
      "step": 12230
    },
    {
      "epoch": 3.035526071052142,
      "grad_norm": 0.8243331909179688,
      "learning_rate": 0.0001578671296179806,
      "loss": 1.5601,
      "step": 12240
    },
    {
      "epoch": 3.038006076012152,
      "grad_norm": 0.9027004241943359,
      "learning_rate": 0.00015780356642313036,
      "loss": 1.6884,
      "step": 12250
    },
    {
      "epoch": 3.040486080972162,
      "grad_norm": 0.8033971190452576,
      "learning_rate": 0.00015773996813586068,
      "loss": 1.59,
      "step": 12260
    },
    {
      "epoch": 3.042966085932172,
      "grad_norm": 0.8007137179374695,
      "learning_rate": 0.00015767633479478197,
      "loss": 1.5946,
      "step": 12270
    },
    {
      "epoch": 3.045446090892182,
      "grad_norm": 0.943191409111023,
      "learning_rate": 0.00015761266643852587,
      "loss": 1.62,
      "step": 12280
    },
    {
      "epoch": 3.047926095852192,
      "grad_norm": 0.8191910982131958,
      "learning_rate": 0.00015754896310574536,
      "loss": 1.6174,
      "step": 12290
    },
    {
      "epoch": 3.0504061008122014,
      "grad_norm": 0.8275129795074463,
      "learning_rate": 0.00015748522483511452,
      "loss": 1.679,
      "step": 12300
    },
    {
      "epoch": 3.0528861057722114,
      "grad_norm": 0.8390007615089417,
      "learning_rate": 0.00015742145166532876,
      "loss": 1.5712,
      "step": 12310
    },
    {
      "epoch": 3.0553661107322214,
      "grad_norm": 0.8464158773422241,
      "learning_rate": 0.0001573576436351046,
      "loss": 1.6154,
      "step": 12320
    },
    {
      "epoch": 3.0578461156922314,
      "grad_norm": 0.8567997813224792,
      "learning_rate": 0.0001572938007831798,
      "loss": 1.6661,
      "step": 12330
    },
    {
      "epoch": 3.0603261206522414,
      "grad_norm": 0.9141848087310791,
      "learning_rate": 0.00015722992314831319,
      "loss": 1.6535,
      "step": 12340
    },
    {
      "epoch": 3.0628061256122514,
      "grad_norm": 0.7453342080116272,
      "learning_rate": 0.00015716601076928475,
      "loss": 1.6644,
      "step": 12350
    },
    {
      "epoch": 3.065286130572261,
      "grad_norm": 0.7754812836647034,
      "learning_rate": 0.00015710206368489552,
      "loss": 1.6265,
      "step": 12360
    },
    {
      "epoch": 3.067766135532271,
      "grad_norm": 0.8418968915939331,
      "learning_rate": 0.00015703808193396766,
      "loss": 1.6306,
      "step": 12370
    },
    {
      "epoch": 3.070246140492281,
      "grad_norm": 0.844227135181427,
      "learning_rate": 0.00015697406555534437,
      "loss": 1.6495,
      "step": 12380
    },
    {
      "epoch": 3.072726145452291,
      "grad_norm": 0.814213752746582,
      "learning_rate": 0.00015691001458788983,
      "loss": 1.6155,
      "step": 12390
    },
    {
      "epoch": 3.075206150412301,
      "grad_norm": 0.8387388586997986,
      "learning_rate": 0.00015684592907048926,
      "loss": 1.6246,
      "step": 12400
    },
    {
      "epoch": 3.077686155372311,
      "grad_norm": 0.7879986763000488,
      "learning_rate": 0.00015678180904204878,
      "loss": 1.648,
      "step": 12410
    },
    {
      "epoch": 3.0801661603323205,
      "grad_norm": 0.8591954112052917,
      "learning_rate": 0.00015671765454149559,
      "loss": 1.5457,
      "step": 12420
    },
    {
      "epoch": 3.0826461652923305,
      "grad_norm": 0.8503502607345581,
      "learning_rate": 0.00015665346560777774,
      "loss": 1.609,
      "step": 12430
    },
    {
      "epoch": 3.0851261702523405,
      "grad_norm": 0.8228851556777954,
      "learning_rate": 0.00015658924227986414,
      "loss": 1.6144,
      "step": 12440
    },
    {
      "epoch": 3.0876061752123505,
      "grad_norm": 0.8595585227012634,
      "learning_rate": 0.00015652498459674467,
      "loss": 1.629,
      "step": 12450
    },
    {
      "epoch": 3.0900861801723605,
      "grad_norm": 0.8532552719116211,
      "learning_rate": 0.00015646069259743007,
      "loss": 1.6055,
      "step": 12460
    },
    {
      "epoch": 3.0925661851323705,
      "grad_norm": 0.8255425095558167,
      "learning_rate": 0.0001563963663209518,
      "loss": 1.6725,
      "step": 12470
    },
    {
      "epoch": 3.09504619009238,
      "grad_norm": 0.8189271688461304,
      "learning_rate": 0.0001563320058063622,
      "loss": 1.6298,
      "step": 12480
    },
    {
      "epoch": 3.09752619505239,
      "grad_norm": 0.7872645854949951,
      "learning_rate": 0.0001562676110927345,
      "loss": 1.604,
      "step": 12490
    },
    {
      "epoch": 3.1000062000124,
      "grad_norm": 0.8693544864654541,
      "learning_rate": 0.00015620318221916247,
      "loss": 1.5533,
      "step": 12500
    },
    {
      "epoch": 3.10248620497241,
      "grad_norm": 0.8340660929679871,
      "learning_rate": 0.0001561387192247608,
      "loss": 1.61,
      "step": 12510
    },
    {
      "epoch": 3.10496620993242,
      "grad_norm": 0.8567219972610474,
      "learning_rate": 0.0001560742221486648,
      "loss": 1.6695,
      "step": 12520
    },
    {
      "epoch": 3.1074462148924296,
      "grad_norm": 0.846973717212677,
      "learning_rate": 0.00015600969103003056,
      "loss": 1.6645,
      "step": 12530
    },
    {
      "epoch": 3.1099262198524396,
      "grad_norm": 0.8784561157226562,
      "learning_rate": 0.00015594512590803473,
      "loss": 1.6265,
      "step": 12540
    },
    {
      "epoch": 3.1124062248124496,
      "grad_norm": 0.7776249647140503,
      "learning_rate": 0.00015588052682187468,
      "loss": 1.5806,
      "step": 12550
    },
    {
      "epoch": 3.1148862297724595,
      "grad_norm": 0.8816384077072144,
      "learning_rate": 0.0001558158938107684,
      "loss": 1.6276,
      "step": 12560
    },
    {
      "epoch": 3.1173662347324695,
      "grad_norm": 0.8353577256202698,
      "learning_rate": 0.0001557512269139544,
      "loss": 1.6276,
      "step": 12570
    },
    {
      "epoch": 3.1198462396924795,
      "grad_norm": 0.8443197011947632,
      "learning_rate": 0.00015568652617069183,
      "loss": 1.5794,
      "step": 12580
    },
    {
      "epoch": 3.122326244652489,
      "grad_norm": 0.9142757654190063,
      "learning_rate": 0.00015562179162026038,
      "loss": 1.6802,
      "step": 12590
    },
    {
      "epoch": 3.124806249612499,
      "grad_norm": 0.8508244156837463,
      "learning_rate": 0.00015555702330196023,
      "loss": 1.6376,
      "step": 12600
    },
    {
      "epoch": 3.127286254572509,
      "grad_norm": 0.9337193369865417,
      "learning_rate": 0.00015549222125511213,
      "loss": 1.6247,
      "step": 12610
    },
    {
      "epoch": 3.129766259532519,
      "grad_norm": 0.8202626705169678,
      "learning_rate": 0.0001554273855190572,
      "loss": 1.6098,
      "step": 12620
    },
    {
      "epoch": 3.132246264492529,
      "grad_norm": 0.8608548045158386,
      "learning_rate": 0.00015536251613315714,
      "loss": 1.6122,
      "step": 12630
    },
    {
      "epoch": 3.134726269452539,
      "grad_norm": 0.8218183517456055,
      "learning_rate": 0.00015529761313679393,
      "loss": 1.6801,
      "step": 12640
    },
    {
      "epoch": 3.1372062744125486,
      "grad_norm": 0.8353672027587891,
      "learning_rate": 0.00015523267656937018,
      "loss": 1.6018,
      "step": 12650
    },
    {
      "epoch": 3.1396862793725586,
      "grad_norm": 0.8429569005966187,
      "learning_rate": 0.00015516770647030858,
      "loss": 1.6067,
      "step": 12660
    },
    {
      "epoch": 3.1421662843325686,
      "grad_norm": 0.8685290813446045,
      "learning_rate": 0.0001551027028790524,
      "loss": 1.6714,
      "step": 12670
    },
    {
      "epoch": 3.1446462892925786,
      "grad_norm": 0.963028073310852,
      "learning_rate": 0.00015503766583506524,
      "loss": 1.6326,
      "step": 12680
    },
    {
      "epoch": 3.1471262942525886,
      "grad_norm": 0.8299250602722168,
      "learning_rate": 0.00015497259537783085,
      "loss": 1.6665,
      "step": 12690
    },
    {
      "epoch": 3.1496062992125986,
      "grad_norm": 0.8954488635063171,
      "learning_rate": 0.0001549074915468534,
      "loss": 1.6717,
      "step": 12700
    },
    {
      "epoch": 3.152086304172608,
      "grad_norm": 0.8646858334541321,
      "learning_rate": 0.00015484235438165727,
      "loss": 1.6755,
      "step": 12710
    },
    {
      "epoch": 3.154566309132618,
      "grad_norm": 0.855476975440979,
      "learning_rate": 0.00015477718392178716,
      "loss": 1.5664,
      "step": 12720
    },
    {
      "epoch": 3.157046314092628,
      "grad_norm": 0.7919172048568726,
      "learning_rate": 0.00015471198020680781,
      "loss": 1.5712,
      "step": 12730
    },
    {
      "epoch": 3.159526319052638,
      "grad_norm": 0.8448411226272583,
      "learning_rate": 0.00015464674327630434,
      "loss": 1.5894,
      "step": 12740
    },
    {
      "epoch": 3.162006324012648,
      "grad_norm": 0.8859073519706726,
      "learning_rate": 0.00015458147316988193,
      "loss": 1.6677,
      "step": 12750
    },
    {
      "epoch": 3.164486328972658,
      "grad_norm": 0.8675896525382996,
      "learning_rate": 0.0001545161699271659,
      "loss": 1.6389,
      "step": 12760
    },
    {
      "epoch": 3.1669663339326677,
      "grad_norm": 0.8519783020019531,
      "learning_rate": 0.0001544508335878017,
      "loss": 1.6636,
      "step": 12770
    },
    {
      "epoch": 3.1694463388926777,
      "grad_norm": 0.8615175485610962,
      "learning_rate": 0.00015438546419145488,
      "loss": 1.5697,
      "step": 12780
    },
    {
      "epoch": 3.1719263438526877,
      "grad_norm": 0.8288758397102356,
      "learning_rate": 0.00015432006177781108,
      "loss": 1.5953,
      "step": 12790
    },
    {
      "epoch": 3.1744063488126977,
      "grad_norm": 0.8835858106613159,
      "learning_rate": 0.00015425462638657595,
      "loss": 1.6528,
      "step": 12800
    },
    {
      "epoch": 3.1768863537727077,
      "grad_norm": 0.8913437128067017,
      "learning_rate": 0.00015418915805747517,
      "loss": 1.6063,
      "step": 12810
    },
    {
      "epoch": 3.1793663587327177,
      "grad_norm": 0.8508837819099426,
      "learning_rate": 0.00015412365683025448,
      "loss": 1.6699,
      "step": 12820
    },
    {
      "epoch": 3.181846363692727,
      "grad_norm": 0.7990511655807495,
      "learning_rate": 0.0001540581227446794,
      "loss": 1.6572,
      "step": 12830
    },
    {
      "epoch": 3.184326368652737,
      "grad_norm": 0.857603907585144,
      "learning_rate": 0.00015399255584053567,
      "loss": 1.6515,
      "step": 12840
    },
    {
      "epoch": 3.186806373612747,
      "grad_norm": 0.8712793588638306,
      "learning_rate": 0.00015392695615762874,
      "loss": 1.696,
      "step": 12850
    },
    {
      "epoch": 3.189286378572757,
      "grad_norm": 0.8377465009689331,
      "learning_rate": 0.00015386132373578404,
      "loss": 1.6259,
      "step": 12860
    },
    {
      "epoch": 3.191766383532767,
      "grad_norm": 0.8834864497184753,
      "learning_rate": 0.00015379565861484688,
      "loss": 1.7125,
      "step": 12870
    },
    {
      "epoch": 3.194246388492777,
      "grad_norm": 0.8212023377418518,
      "learning_rate": 0.0001537299608346824,
      "loss": 1.6291,
      "step": 12880
    },
    {
      "epoch": 3.1967263934527868,
      "grad_norm": 0.8849619626998901,
      "learning_rate": 0.0001536642304351756,
      "loss": 1.708,
      "step": 12890
    },
    {
      "epoch": 3.1992063984127967,
      "grad_norm": 0.8786877393722534,
      "learning_rate": 0.00015359846745623126,
      "loss": 1.6772,
      "step": 12900
    },
    {
      "epoch": 3.2016864033728067,
      "grad_norm": 0.9455540776252747,
      "learning_rate": 0.00015353267193777393,
      "loss": 1.6662,
      "step": 12910
    },
    {
      "epoch": 3.2041664083328167,
      "grad_norm": 0.8279966115951538,
      "learning_rate": 0.00015346684391974794,
      "loss": 1.6842,
      "step": 12920
    },
    {
      "epoch": 3.2066464132928267,
      "grad_norm": 0.8265210390090942,
      "learning_rate": 0.00015340098344211733,
      "loss": 1.6751,
      "step": 12930
    },
    {
      "epoch": 3.2091264182528363,
      "grad_norm": 0.8862226009368896,
      "learning_rate": 0.00015333509054486584,
      "loss": 1.6636,
      "step": 12940
    },
    {
      "epoch": 3.2116064232128463,
      "grad_norm": 0.9301692843437195,
      "learning_rate": 0.00015326916526799692,
      "loss": 1.587,
      "step": 12950
    },
    {
      "epoch": 3.2140864281728563,
      "grad_norm": 0.9067266583442688,
      "learning_rate": 0.00015320320765153367,
      "loss": 1.6828,
      "step": 12960
    },
    {
      "epoch": 3.2165664331328663,
      "grad_norm": 0.8438600301742554,
      "learning_rate": 0.00015313721773551876,
      "loss": 1.6409,
      "step": 12970
    },
    {
      "epoch": 3.2190464380928763,
      "grad_norm": 0.890546977519989,
      "learning_rate": 0.00015307119556001465,
      "loss": 1.6323,
      "step": 12980
    },
    {
      "epoch": 3.2215264430528863,
      "grad_norm": 0.7987525463104248,
      "learning_rate": 0.0001530051411651031,
      "loss": 1.6684,
      "step": 12990
    },
    {
      "epoch": 3.224006448012896,
      "grad_norm": 0.9028956890106201,
      "learning_rate": 0.00015293905459088568,
      "loss": 1.6021,
      "step": 13000
    },
    {
      "epoch": 3.226486452972906,
      "grad_norm": 0.8655980229377747,
      "learning_rate": 0.00015287293587748342,
      "loss": 1.6022,
      "step": 13010
    },
    {
      "epoch": 3.228966457932916,
      "grad_norm": 0.8287554383277893,
      "learning_rate": 0.0001528067850650368,
      "loss": 1.6954,
      "step": 13020
    },
    {
      "epoch": 3.231446462892926,
      "grad_norm": 0.8922123312950134,
      "learning_rate": 0.00015274060219370585,
      "loss": 1.6897,
      "step": 13030
    },
    {
      "epoch": 3.233926467852936,
      "grad_norm": 0.8276147246360779,
      "learning_rate": 0.0001526743873036701,
      "loss": 1.6058,
      "step": 13040
    },
    {
      "epoch": 3.236406472812946,
      "grad_norm": 0.8847476840019226,
      "learning_rate": 0.00015260814043512836,
      "loss": 1.5953,
      "step": 13050
    },
    {
      "epoch": 3.2388864777729554,
      "grad_norm": 0.8755683898925781,
      "learning_rate": 0.00015254186162829903,
      "loss": 1.6701,
      "step": 13060
    },
    {
      "epoch": 3.2413664827329653,
      "grad_norm": 0.8448535203933716,
      "learning_rate": 0.0001524755509234198,
      "loss": 1.7325,
      "step": 13070
    },
    {
      "epoch": 3.2438464876929753,
      "grad_norm": 0.8898816108703613,
      "learning_rate": 0.00015240920836074776,
      "loss": 1.642,
      "step": 13080
    },
    {
      "epoch": 3.2463264926529853,
      "grad_norm": 0.8643108010292053,
      "learning_rate": 0.0001523428339805594,
      "loss": 1.6345,
      "step": 13090
    },
    {
      "epoch": 3.2488064976129953,
      "grad_norm": 0.8506256937980652,
      "learning_rate": 0.00015227642782315038,
      "loss": 1.6628,
      "step": 13100
    },
    {
      "epoch": 3.2512865025730053,
      "grad_norm": 0.949025571346283,
      "learning_rate": 0.00015220998992883573,
      "loss": 1.6418,
      "step": 13110
    },
    {
      "epoch": 3.253766507533015,
      "grad_norm": 0.8501860499382019,
      "learning_rate": 0.0001521435203379498,
      "loss": 1.6433,
      "step": 13120
    },
    {
      "epoch": 3.256246512493025,
      "grad_norm": 0.8821784853935242,
      "learning_rate": 0.00015207701909084612,
      "loss": 1.6308,
      "step": 13130
    },
    {
      "epoch": 3.258726517453035,
      "grad_norm": 0.8270078897476196,
      "learning_rate": 0.00015201048622789747,
      "loss": 1.6803,
      "step": 13140
    },
    {
      "epoch": 3.261206522413045,
      "grad_norm": 0.9211381673812866,
      "learning_rate": 0.00015194392178949576,
      "loss": 1.6086,
      "step": 13150
    },
    {
      "epoch": 3.263686527373055,
      "grad_norm": 0.90462327003479,
      "learning_rate": 0.00015187732581605217,
      "loss": 1.5827,
      "step": 13160
    },
    {
      "epoch": 3.266166532333065,
      "grad_norm": 0.8179154992103577,
      "learning_rate": 0.00015181069834799691,
      "loss": 1.6422,
      "step": 13170
    },
    {
      "epoch": 3.2686465372930744,
      "grad_norm": 0.8857276439666748,
      "learning_rate": 0.00015174403942577942,
      "loss": 1.6346,
      "step": 13180
    },
    {
      "epoch": 3.2711265422530844,
      "grad_norm": 0.8045947551727295,
      "learning_rate": 0.00015167734908986814,
      "loss": 1.636,
      "step": 13190
    },
    {
      "epoch": 3.2736065472130944,
      "grad_norm": 0.8784744739532471,
      "learning_rate": 0.00015161062738075067,
      "loss": 1.5687,
      "step": 13200
    },
    {
      "epoch": 3.2760865521731044,
      "grad_norm": 0.7755175828933716,
      "learning_rate": 0.0001515438743389336,
      "loss": 1.6515,
      "step": 13210
    },
    {
      "epoch": 3.2785665571331144,
      "grad_norm": 0.8107953667640686,
      "learning_rate": 0.00015147709000494256,
      "loss": 1.6565,
      "step": 13220
    },
    {
      "epoch": 3.281046562093124,
      "grad_norm": 0.8617423176765442,
      "learning_rate": 0.00015141027441932216,
      "loss": 1.6476,
      "step": 13230
    },
    {
      "epoch": 3.283526567053134,
      "grad_norm": 1.0366381406784058,
      "learning_rate": 0.00015134342762263605,
      "loss": 1.6506,
      "step": 13240
    },
    {
      "epoch": 3.286006572013144,
      "grad_norm": 0.8942514657974243,
      "learning_rate": 0.00015127654965546672,
      "loss": 1.6069,
      "step": 13250
    },
    {
      "epoch": 3.288486576973154,
      "grad_norm": 0.8343939185142517,
      "learning_rate": 0.00015120964055841563,
      "loss": 1.639,
      "step": 13260
    },
    {
      "epoch": 3.290966581933164,
      "grad_norm": 1.0147429704666138,
      "learning_rate": 0.00015114270037210316,
      "loss": 1.6268,
      "step": 13270
    },
    {
      "epoch": 3.293446586893174,
      "grad_norm": 0.862209141254425,
      "learning_rate": 0.00015107572913716858,
      "loss": 1.6769,
      "step": 13280
    },
    {
      "epoch": 3.295926591853184,
      "grad_norm": 0.8578622937202454,
      "learning_rate": 0.00015100872689426993,
      "loss": 1.6625,
      "step": 13290
    },
    {
      "epoch": 3.2984065968131935,
      "grad_norm": 0.8519478440284729,
      "learning_rate": 0.0001509416936840842,
      "loss": 1.6312,
      "step": 13300
    },
    {
      "epoch": 3.3008866017732035,
      "grad_norm": 0.9173608422279358,
      "learning_rate": 0.00015087462954730703,
      "loss": 1.6838,
      "step": 13310
    },
    {
      "epoch": 3.3033666067332135,
      "grad_norm": 0.8745505213737488,
      "learning_rate": 0.00015080753452465296,
      "loss": 1.6099,
      "step": 13320
    },
    {
      "epoch": 3.3058466116932235,
      "grad_norm": 0.8400318622589111,
      "learning_rate": 0.00015074040865685522,
      "loss": 1.7042,
      "step": 13330
    },
    {
      "epoch": 3.3083266166532335,
      "grad_norm": 0.8654444813728333,
      "learning_rate": 0.00015067325198466576,
      "loss": 1.6304,
      "step": 13340
    },
    {
      "epoch": 3.310806621613243,
      "grad_norm": 0.981468915939331,
      "learning_rate": 0.00015060606454885526,
      "loss": 1.67,
      "step": 13350
    },
    {
      "epoch": 3.313286626573253,
      "grad_norm": 0.8678485751152039,
      "learning_rate": 0.0001505388463902131,
      "loss": 1.6023,
      "step": 13360
    },
    {
      "epoch": 3.315766631533263,
      "grad_norm": 0.852952778339386,
      "learning_rate": 0.0001504715975495472,
      "loss": 1.6585,
      "step": 13370
    },
    {
      "epoch": 3.318246636493273,
      "grad_norm": 0.8926320672035217,
      "learning_rate": 0.00015040431806768428,
      "loss": 1.6037,
      "step": 13380
    },
    {
      "epoch": 3.320726641453283,
      "grad_norm": 0.9940789341926575,
      "learning_rate": 0.00015033700798546953,
      "loss": 1.6151,
      "step": 13390
    },
    {
      "epoch": 3.323206646413293,
      "grad_norm": 0.880584180355072,
      "learning_rate": 0.0001502696673437667,
      "loss": 1.6386,
      "step": 13400
    },
    {
      "epoch": 3.325686651373303,
      "grad_norm": 0.8516528010368347,
      "learning_rate": 0.0001502022961834582,
      "loss": 1.6993,
      "step": 13410
    },
    {
      "epoch": 3.3281666563333125,
      "grad_norm": 0.7992837429046631,
      "learning_rate": 0.00015013489454544494,
      "loss": 1.6166,
      "step": 13420
    },
    {
      "epoch": 3.3306466612933225,
      "grad_norm": 0.848483681678772,
      "learning_rate": 0.00015006746247064624,
      "loss": 1.6442,
      "step": 13430
    },
    {
      "epoch": 3.3331266662533325,
      "grad_norm": 0.8658037781715393,
      "learning_rate": 0.00015000000000000001,
      "loss": 1.6639,
      "step": 13440
    },
    {
      "epoch": 3.3356066712133425,
      "grad_norm": 0.8861033320426941,
      "learning_rate": 0.00014993250717446256,
      "loss": 1.6886,
      "step": 13450
    },
    {
      "epoch": 3.3380866761733525,
      "grad_norm": 0.8217837810516357,
      "learning_rate": 0.00014986498403500863,
      "loss": 1.6541,
      "step": 13460
    },
    {
      "epoch": 3.340566681133362,
      "grad_norm": 0.8470338582992554,
      "learning_rate": 0.0001497974306226314,
      "loss": 1.5988,
      "step": 13470
    },
    {
      "epoch": 3.343046686093372,
      "grad_norm": 0.9460231065750122,
      "learning_rate": 0.0001497298469783424,
      "loss": 1.5984,
      "step": 13480
    },
    {
      "epoch": 3.345526691053382,
      "grad_norm": 0.83684903383255,
      "learning_rate": 0.0001496622331431714,
      "loss": 1.6649,
      "step": 13490
    },
    {
      "epoch": 3.348006696013392,
      "grad_norm": 0.8802300691604614,
      "learning_rate": 0.0001495945891581668,
      "loss": 1.5957,
      "step": 13500
    },
    {
      "epoch": 3.350486700973402,
      "grad_norm": 0.8913207054138184,
      "learning_rate": 0.00014952691506439495,
      "loss": 1.6574,
      "step": 13510
    },
    {
      "epoch": 3.3529667059334116,
      "grad_norm": 0.9203808903694153,
      "learning_rate": 0.00014945921090294076,
      "loss": 1.6445,
      "step": 13520
    },
    {
      "epoch": 3.3554467108934216,
      "grad_norm": 0.8547185063362122,
      "learning_rate": 0.0001493914767149072,
      "loss": 1.6289,
      "step": 13530
    },
    {
      "epoch": 3.3579267158534316,
      "grad_norm": 0.9169796705245972,
      "learning_rate": 0.0001493237125414156,
      "loss": 1.6561,
      "step": 13540
    },
    {
      "epoch": 3.3604067208134416,
      "grad_norm": 0.8416388034820557,
      "learning_rate": 0.00014925591842360542,
      "loss": 1.6602,
      "step": 13550
    },
    {
      "epoch": 3.3628867257734516,
      "grad_norm": 0.8726803660392761,
      "learning_rate": 0.00014918809440263436,
      "loss": 1.6112,
      "step": 13560
    },
    {
      "epoch": 3.3653667307334616,
      "grad_norm": 0.8233663439750671,
      "learning_rate": 0.00014912024051967816,
      "loss": 1.6279,
      "step": 13570
    },
    {
      "epoch": 3.3678467356934716,
      "grad_norm": 0.8870377540588379,
      "learning_rate": 0.0001490523568159308,
      "loss": 1.5935,
      "step": 13580
    },
    {
      "epoch": 3.370326740653481,
      "grad_norm": 0.8920510411262512,
      "learning_rate": 0.00014898444333260436,
      "loss": 1.6159,
      "step": 13590
    },
    {
      "epoch": 3.372806745613491,
      "grad_norm": 0.87026447057724,
      "learning_rate": 0.00014891650011092896,
      "loss": 1.7169,
      "step": 13600
    },
    {
      "epoch": 3.375286750573501,
      "grad_norm": 0.8275659680366516,
      "learning_rate": 0.0001488485271921527,
      "loss": 1.6774,
      "step": 13610
    },
    {
      "epoch": 3.377766755533511,
      "grad_norm": 0.8684106469154358,
      "learning_rate": 0.0001487805246175419,
      "loss": 1.6216,
      "step": 13620
    },
    {
      "epoch": 3.380246760493521,
      "grad_norm": 0.8039277195930481,
      "learning_rate": 0.00014871249242838074,
      "loss": 1.5738,
      "step": 13630
    },
    {
      "epoch": 3.3827267654535307,
      "grad_norm": 0.9385021924972534,
      "learning_rate": 0.00014864443066597139,
      "loss": 1.6307,
      "step": 13640
    },
    {
      "epoch": 3.3852067704135407,
      "grad_norm": 0.9041079878807068,
      "learning_rate": 0.00014857633937163402,
      "loss": 1.6933,
      "step": 13650
    },
    {
      "epoch": 3.3876867753735507,
      "grad_norm": 0.8626042604446411,
      "learning_rate": 0.00014850821858670667,
      "loss": 1.6563,
      "step": 13660
    },
    {
      "epoch": 3.3901667803335607,
      "grad_norm": 0.8708447217941284,
      "learning_rate": 0.0001484400683525454,
      "loss": 1.6932,
      "step": 13670
    },
    {
      "epoch": 3.3926467852935707,
      "grad_norm": 0.8160665035247803,
      "learning_rate": 0.000148371888710524,
      "loss": 1.6541,
      "step": 13680
    },
    {
      "epoch": 3.3951267902535807,
      "grad_norm": 0.8544750213623047,
      "learning_rate": 0.00014830367970203415,
      "loss": 1.6279,
      "step": 13690
    },
    {
      "epoch": 3.3976067952135907,
      "grad_norm": 0.8258888125419617,
      "learning_rate": 0.00014823544136848552,
      "loss": 1.6896,
      "step": 13700
    },
    {
      "epoch": 3.4000868001736,
      "grad_norm": 0.8572070598602295,
      "learning_rate": 0.00014816717375130533,
      "loss": 1.6395,
      "step": 13710
    },
    {
      "epoch": 3.40256680513361,
      "grad_norm": 0.8532230854034424,
      "learning_rate": 0.00014809887689193877,
      "loss": 1.6723,
      "step": 13720
    },
    {
      "epoch": 3.40504681009362,
      "grad_norm": 0.918097198009491,
      "learning_rate": 0.0001480305508318487,
      "loss": 1.7022,
      "step": 13730
    },
    {
      "epoch": 3.40752681505363,
      "grad_norm": 0.909914493560791,
      "learning_rate": 0.00014796219561251568,
      "loss": 1.6047,
      "step": 13740
    },
    {
      "epoch": 3.41000682001364,
      "grad_norm": 0.8644607067108154,
      "learning_rate": 0.00014789381127543807,
      "loss": 1.6696,
      "step": 13750
    },
    {
      "epoch": 3.4124868249736497,
      "grad_norm": 0.8513312339782715,
      "learning_rate": 0.00014782539786213183,
      "loss": 1.6378,
      "step": 13760
    },
    {
      "epoch": 3.4149668299336597,
      "grad_norm": 0.8677627444267273,
      "learning_rate": 0.00014775695541413062,
      "loss": 1.6185,
      "step": 13770
    },
    {
      "epoch": 3.4174468348936697,
      "grad_norm": 0.8117983341217041,
      "learning_rate": 0.00014768848397298562,
      "loss": 1.6657,
      "step": 13780
    },
    {
      "epoch": 3.4199268398536797,
      "grad_norm": 0.8320634961128235,
      "learning_rate": 0.0001476199835802658,
      "loss": 1.6567,
      "step": 13790
    },
    {
      "epoch": 3.4224068448136897,
      "grad_norm": 0.8326737880706787,
      "learning_rate": 0.00014755145427755754,
      "loss": 1.7125,
      "step": 13800
    },
    {
      "epoch": 3.4248868497736997,
      "grad_norm": 0.860938549041748,
      "learning_rate": 0.00014748289610646482,
      "loss": 1.5955,
      "step": 13810
    },
    {
      "epoch": 3.4273668547337093,
      "grad_norm": 0.8838903307914734,
      "learning_rate": 0.0001474143091086092,
      "loss": 1.6864,
      "step": 13820
    },
    {
      "epoch": 3.4298468596937193,
      "grad_norm": 0.9615389108657837,
      "learning_rate": 0.00014734569332562966,
      "loss": 1.6906,
      "step": 13830
    },
    {
      "epoch": 3.4323268646537293,
      "grad_norm": 0.8392242193222046,
      "learning_rate": 0.0001472770487991827,
      "loss": 1.653,
      "step": 13840
    },
    {
      "epoch": 3.4348068696137393,
      "grad_norm": 0.8907983303070068,
      "learning_rate": 0.00014720837557094228,
      "loss": 1.6522,
      "step": 13850
    },
    {
      "epoch": 3.4372868745737493,
      "grad_norm": 0.9216994643211365,
      "learning_rate": 0.0001471396736825998,
      "loss": 1.6204,
      "step": 13860
    },
    {
      "epoch": 3.4397668795337593,
      "grad_norm": 0.9157097935676575,
      "learning_rate": 0.00014707094317586392,
      "loss": 1.6045,
      "step": 13870
    },
    {
      "epoch": 3.442246884493769,
      "grad_norm": 0.861574113368988,
      "learning_rate": 0.00014700218409246087,
      "loss": 1.7071,
      "step": 13880
    },
    {
      "epoch": 3.444726889453779,
      "grad_norm": 0.9511358141899109,
      "learning_rate": 0.0001469333964741341,
      "loss": 1.6046,
      "step": 13890
    },
    {
      "epoch": 3.447206894413789,
      "grad_norm": 0.8342337608337402,
      "learning_rate": 0.00014686458036264447,
      "loss": 1.6776,
      "step": 13900
    },
    {
      "epoch": 3.449686899373799,
      "grad_norm": 0.8049719333648682,
      "learning_rate": 0.00014679573579977004,
      "loss": 1.641,
      "step": 13910
    },
    {
      "epoch": 3.452166904333809,
      "grad_norm": 0.8005982041358948,
      "learning_rate": 0.0001467268628273062,
      "loss": 1.6753,
      "step": 13920
    },
    {
      "epoch": 3.4546469092938183,
      "grad_norm": 0.8201278448104858,
      "learning_rate": 0.0001466579614870656,
      "loss": 1.6484,
      "step": 13930
    },
    {
      "epoch": 3.4571269142538283,
      "grad_norm": 0.8669260144233704,
      "learning_rate": 0.00014658903182087813,
      "loss": 1.5958,
      "step": 13940
    },
    {
      "epoch": 3.4596069192138383,
      "grad_norm": 0.9065529704093933,
      "learning_rate": 0.00014652007387059077,
      "loss": 1.6607,
      "step": 13950
    },
    {
      "epoch": 3.4620869241738483,
      "grad_norm": 0.8734999299049377,
      "learning_rate": 0.00014645108767806777,
      "loss": 1.6161,
      "step": 13960
    },
    {
      "epoch": 3.4645669291338583,
      "grad_norm": 0.8499678373336792,
      "learning_rate": 0.00014638207328519052,
      "loss": 1.6,
      "step": 13970
    },
    {
      "epoch": 3.4670469340938683,
      "grad_norm": 0.828967809677124,
      "learning_rate": 0.00014631303073385746,
      "loss": 1.5823,
      "step": 13980
    },
    {
      "epoch": 3.4695269390538783,
      "grad_norm": 0.9271993637084961,
      "learning_rate": 0.0001462439600659842,
      "loss": 1.6964,
      "step": 13990
    },
    {
      "epoch": 3.472006944013888,
      "grad_norm": 0.8467398881912231,
      "learning_rate": 0.00014617486132350343,
      "loss": 1.6077,
      "step": 14000
    },
    {
      "epoch": 3.474486948973898,
      "grad_norm": 0.8329309821128845,
      "learning_rate": 0.00014610573454836476,
      "loss": 1.663,
      "step": 14010
    },
    {
      "epoch": 3.476966953933908,
      "grad_norm": 0.8573420643806458,
      "learning_rate": 0.00014603657978253497,
      "loss": 1.6057,
      "step": 14020
    },
    {
      "epoch": 3.479446958893918,
      "grad_norm": 0.901738703250885,
      "learning_rate": 0.00014596739706799775,
      "loss": 1.6707,
      "step": 14030
    },
    {
      "epoch": 3.481926963853928,
      "grad_norm": 0.8534144163131714,
      "learning_rate": 0.00014589818644675378,
      "loss": 1.5936,
      "step": 14040
    },
    {
      "epoch": 3.4844069688139374,
      "grad_norm": 1.1463650465011597,
      "learning_rate": 0.00014582894796082068,
      "loss": 1.5724,
      "step": 14050
    },
    {
      "epoch": 3.4868869737739474,
      "grad_norm": 0.8025235533714294,
      "learning_rate": 0.00014575968165223298,
      "loss": 1.642,
      "step": 14060
    },
    {
      "epoch": 3.4893669787339574,
      "grad_norm": 0.8681792616844177,
      "learning_rate": 0.00014569038756304207,
      "loss": 1.6466,
      "step": 14070
    },
    {
      "epoch": 3.4918469836939674,
      "grad_norm": 0.9857133030891418,
      "learning_rate": 0.0001456210657353163,
      "loss": 1.6794,
      "step": 14080
    },
    {
      "epoch": 3.4943269886539774,
      "grad_norm": 0.9428927302360535,
      "learning_rate": 0.0001455517162111408,
      "loss": 1.6231,
      "step": 14090
    },
    {
      "epoch": 3.4968069936139874,
      "grad_norm": 0.8979089856147766,
      "learning_rate": 0.00014548233903261746,
      "loss": 1.6431,
      "step": 14100
    },
    {
      "epoch": 3.4992869985739974,
      "grad_norm": 0.9292755126953125,
      "learning_rate": 0.00014541293424186503,
      "loss": 1.6476,
      "step": 14110
    },
    {
      "epoch": 3.501767003534007,
      "grad_norm": 0.9158943295478821,
      "learning_rate": 0.00014534350188101904,
      "loss": 1.6577,
      "step": 14120
    },
    {
      "epoch": 3.504247008494017,
      "grad_norm": 0.8280609250068665,
      "learning_rate": 0.00014527404199223172,
      "loss": 1.6099,
      "step": 14130
    },
    {
      "epoch": 3.506727013454027,
      "grad_norm": 0.9120189547538757,
      "learning_rate": 0.000145204554617672,
      "loss": 1.6737,
      "step": 14140
    },
    {
      "epoch": 3.509207018414037,
      "grad_norm": 0.7935187816619873,
      "learning_rate": 0.0001451350397995255,
      "loss": 1.6808,
      "step": 14150
    },
    {
      "epoch": 3.511687023374047,
      "grad_norm": 0.8127133250236511,
      "learning_rate": 0.00014506549757999454,
      "loss": 1.6635,
      "step": 14160
    },
    {
      "epoch": 3.5141670283340565,
      "grad_norm": 0.8754492402076721,
      "learning_rate": 0.00014499592800129804,
      "loss": 1.6821,
      "step": 14170
    },
    {
      "epoch": 3.5166470332940665,
      "grad_norm": 0.940440833568573,
      "learning_rate": 0.00014492633110567155,
      "loss": 1.6923,
      "step": 14180
    },
    {
      "epoch": 3.5191270382540765,
      "grad_norm": 0.8897917866706848,
      "learning_rate": 0.00014485670693536716,
      "loss": 1.5622,
      "step": 14190
    },
    {
      "epoch": 3.5216070432140865,
      "grad_norm": 0.8506290912628174,
      "learning_rate": 0.00014478705553265362,
      "loss": 1.5945,
      "step": 14200
    },
    {
      "epoch": 3.5240870481740965,
      "grad_norm": 0.8483827114105225,
      "learning_rate": 0.0001447173769398161,
      "loss": 1.6169,
      "step": 14210
    },
    {
      "epoch": 3.526567053134106,
      "grad_norm": 0.8673563003540039,
      "learning_rate": 0.00014464767119915629,
      "loss": 1.6915,
      "step": 14220
    },
    {
      "epoch": 3.5290470580941165,
      "grad_norm": 0.8081308007240295,
      "learning_rate": 0.00014457793835299246,
      "loss": 1.6491,
      "step": 14230
    },
    {
      "epoch": 3.531527063054126,
      "grad_norm": 0.9240626692771912,
      "learning_rate": 0.00014450817844365921,
      "loss": 1.6974,
      "step": 14240
    },
    {
      "epoch": 3.534007068014136,
      "grad_norm": 0.8092256784439087,
      "learning_rate": 0.00014443839151350772,
      "loss": 1.6547,
      "step": 14250
    },
    {
      "epoch": 3.536487072974146,
      "grad_norm": 0.8489810824394226,
      "learning_rate": 0.0001443685776049054,
      "loss": 1.6383,
      "step": 14260
    },
    {
      "epoch": 3.538967077934156,
      "grad_norm": 0.9102332592010498,
      "learning_rate": 0.0001442987367602361,
      "loss": 1.6713,
      "step": 14270
    },
    {
      "epoch": 3.541447082894166,
      "grad_norm": 0.8811954855918884,
      "learning_rate": 0.00014422886902190014,
      "loss": 1.628,
      "step": 14280
    },
    {
      "epoch": 3.5439270878541755,
      "grad_norm": 0.9573650360107422,
      "learning_rate": 0.000144158974432314,
      "loss": 1.6651,
      "step": 14290
    },
    {
      "epoch": 3.5464070928141855,
      "grad_norm": 0.9757610559463501,
      "learning_rate": 0.00014408905303391054,
      "loss": 1.6547,
      "step": 14300
    },
    {
      "epoch": 3.5488870977741955,
      "grad_norm": 0.8832597732543945,
      "learning_rate": 0.00014401910486913893,
      "loss": 1.705,
      "step": 14310
    },
    {
      "epoch": 3.5513671027342055,
      "grad_norm": 0.781204104423523,
      "learning_rate": 0.0001439491299804645,
      "loss": 1.6595,
      "step": 14320
    },
    {
      "epoch": 3.5538471076942155,
      "grad_norm": 0.809134840965271,
      "learning_rate": 0.00014387912841036887,
      "loss": 1.6322,
      "step": 14330
    },
    {
      "epoch": 3.556327112654225,
      "grad_norm": 0.9049520492553711,
      "learning_rate": 0.00014380910020134988,
      "loss": 1.6494,
      "step": 14340
    },
    {
      "epoch": 3.558807117614235,
      "grad_norm": 0.8458988666534424,
      "learning_rate": 0.00014373904539592144,
      "loss": 1.6275,
      "step": 14350
    },
    {
      "epoch": 3.561287122574245,
      "grad_norm": 0.9220494031906128,
      "learning_rate": 0.0001436689640366137,
      "loss": 1.6152,
      "step": 14360
    },
    {
      "epoch": 3.563767127534255,
      "grad_norm": 0.8901864290237427,
      "learning_rate": 0.0001435988561659729,
      "loss": 1.6828,
      "step": 14370
    },
    {
      "epoch": 3.566247132494265,
      "grad_norm": 0.9905135631561279,
      "learning_rate": 0.00014352872182656139,
      "loss": 1.6756,
      "step": 14380
    },
    {
      "epoch": 3.568727137454275,
      "grad_norm": 0.8881046175956726,
      "learning_rate": 0.00014345856106095752,
      "loss": 1.6471,
      "step": 14390
    },
    {
      "epoch": 3.571207142414285,
      "grad_norm": 0.8560605645179749,
      "learning_rate": 0.00014338837391175582,
      "loss": 1.6544,
      "step": 14400
    },
    {
      "epoch": 3.5736871473742946,
      "grad_norm": 0.8330683708190918,
      "learning_rate": 0.0001433181604215667,
      "loss": 1.6811,
      "step": 14410
    },
    {
      "epoch": 3.5761671523343046,
      "grad_norm": 0.7894644737243652,
      "learning_rate": 0.00014324792063301662,
      "loss": 1.6553,
      "step": 14420
    },
    {
      "epoch": 3.5786471572943146,
      "grad_norm": 0.8979082703590393,
      "learning_rate": 0.00014317765458874803,
      "loss": 1.6732,
      "step": 14430
    },
    {
      "epoch": 3.5811271622543246,
      "grad_norm": 0.8586720824241638,
      "learning_rate": 0.00014310736233141925,
      "loss": 1.6311,
      "step": 14440
    },
    {
      "epoch": 3.5836071672143346,
      "grad_norm": 0.9059020280838013,
      "learning_rate": 0.00014303704390370464,
      "loss": 1.6694,
      "step": 14450
    },
    {
      "epoch": 3.586087172174344,
      "grad_norm": 0.8023223876953125,
      "learning_rate": 0.00014296669934829424,
      "loss": 1.6898,
      "step": 14460
    },
    {
      "epoch": 3.588567177134354,
      "grad_norm": 0.8363684415817261,
      "learning_rate": 0.00014289632870789418,
      "loss": 1.6077,
      "step": 14470
    },
    {
      "epoch": 3.591047182094364,
      "grad_norm": 0.9061567187309265,
      "learning_rate": 0.00014282593202522627,
      "loss": 1.699,
      "step": 14480
    },
    {
      "epoch": 3.593527187054374,
      "grad_norm": 0.8779317736625671,
      "learning_rate": 0.00014275550934302823,
      "loss": 1.6464,
      "step": 14490
    },
    {
      "epoch": 3.596007192014384,
      "grad_norm": 0.9246005415916443,
      "learning_rate": 0.00014268506070405344,
      "loss": 1.5933,
      "step": 14500
    },
    {
      "epoch": 3.5984871969743937,
      "grad_norm": 0.8338398933410645,
      "learning_rate": 0.0001426145861510712,
      "loss": 1.6311,
      "step": 14510
    },
    {
      "epoch": 3.600967201934404,
      "grad_norm": 0.8370964527130127,
      "learning_rate": 0.00014254408572686642,
      "loss": 1.6231,
      "step": 14520
    },
    {
      "epoch": 3.6034472068944137,
      "grad_norm": 0.8520059585571289,
      "learning_rate": 0.00014247355947423973,
      "loss": 1.7129,
      "step": 14530
    },
    {
      "epoch": 3.6059272118544237,
      "grad_norm": 0.8423281908035278,
      "learning_rate": 0.0001424030074360075,
      "loss": 1.6319,
      "step": 14540
    },
    {
      "epoch": 3.6084072168144337,
      "grad_norm": 1.005537748336792,
      "learning_rate": 0.00014233242965500171,
      "loss": 1.6302,
      "step": 14550
    },
    {
      "epoch": 3.6108872217744437,
      "grad_norm": 0.8900558352470398,
      "learning_rate": 0.00014226182617406996,
      "loss": 1.5912,
      "step": 14560
    },
    {
      "epoch": 3.6133672267344537,
      "grad_norm": 0.8765195608139038,
      "learning_rate": 0.00014219119703607548,
      "loss": 1.627,
      "step": 14570
    },
    {
      "epoch": 3.615847231694463,
      "grad_norm": 0.7887884974479675,
      "learning_rate": 0.0001421205422838971,
      "loss": 1.6835,
      "step": 14580
    },
    {
      "epoch": 3.618327236654473,
      "grad_norm": 0.9221144914627075,
      "learning_rate": 0.0001420498619604291,
      "loss": 1.6755,
      "step": 14590
    },
    {
      "epoch": 3.620807241614483,
      "grad_norm": 0.8162357211112976,
      "learning_rate": 0.00014197915610858144,
      "loss": 1.6397,
      "step": 14600
    },
    {
      "epoch": 3.623287246574493,
      "grad_norm": 0.8468723297119141,
      "learning_rate": 0.0001419084247712794,
      "loss": 1.6348,
      "step": 14610
    },
    {
      "epoch": 3.625767251534503,
      "grad_norm": 0.891946017742157,
      "learning_rate": 0.00014183766799146384,
      "loss": 1.6595,
      "step": 14620
    },
    {
      "epoch": 3.6282472564945127,
      "grad_norm": 0.9180444478988647,
      "learning_rate": 0.0001417668858120911,
      "loss": 1.6729,
      "step": 14630
    },
    {
      "epoch": 3.630727261454523,
      "grad_norm": 0.9543078541755676,
      "learning_rate": 0.00014169607827613283,
      "loss": 1.5954,
      "step": 14640
    },
    {
      "epoch": 3.6332072664145327,
      "grad_norm": 0.8497787117958069,
      "learning_rate": 0.00014162524542657616,
      "loss": 1.645,
      "step": 14650
    },
    {
      "epoch": 3.6356872713745427,
      "grad_norm": 0.8668899536132812,
      "learning_rate": 0.00014155438730642354,
      "loss": 1.6452,
      "step": 14660
    },
    {
      "epoch": 3.6381672763345527,
      "grad_norm": 0.901395320892334,
      "learning_rate": 0.00014148350395869278,
      "loss": 1.6671,
      "step": 14670
    },
    {
      "epoch": 3.6406472812945627,
      "grad_norm": 1.0655369758605957,
      "learning_rate": 0.00014141259542641704,
      "loss": 1.6761,
      "step": 14680
    },
    {
      "epoch": 3.6431272862545727,
      "grad_norm": 0.8188184499740601,
      "learning_rate": 0.00014134166175264475,
      "loss": 1.6836,
      "step": 14690
    },
    {
      "epoch": 3.6456072912145823,
      "grad_norm": 0.8401299715042114,
      "learning_rate": 0.00014127070298043947,
      "loss": 1.6563,
      "step": 14700
    },
    {
      "epoch": 3.6480872961745923,
      "grad_norm": 0.8492573499679565,
      "learning_rate": 0.00014119971915288026,
      "loss": 1.6963,
      "step": 14710
    },
    {
      "epoch": 3.6505673011346023,
      "grad_norm": 0.8795192837715149,
      "learning_rate": 0.00014112871031306119,
      "loss": 1.6859,
      "step": 14720
    },
    {
      "epoch": 3.6530473060946123,
      "grad_norm": 0.870771050453186,
      "learning_rate": 0.00014105767650409153,
      "loss": 1.6251,
      "step": 14730
    },
    {
      "epoch": 3.6555273110546223,
      "grad_norm": 0.8470786809921265,
      "learning_rate": 0.00014098661776909582,
      "loss": 1.6641,
      "step": 14740
    },
    {
      "epoch": 3.658007316014632,
      "grad_norm": 0.873631477355957,
      "learning_rate": 0.0001409155341512136,
      "loss": 1.6614,
      "step": 14750
    },
    {
      "epoch": 3.660487320974642,
      "grad_norm": 0.8350811004638672,
      "learning_rate": 0.00014084442569359964,
      "loss": 1.6961,
      "step": 14760
    },
    {
      "epoch": 3.662967325934652,
      "grad_norm": 0.8719115853309631,
      "learning_rate": 0.00014077329243942369,
      "loss": 1.6678,
      "step": 14770
    },
    {
      "epoch": 3.665447330894662,
      "grad_norm": 0.8512424230575562,
      "learning_rate": 0.00014070213443187062,
      "loss": 1.6482,
      "step": 14780
    },
    {
      "epoch": 3.667927335854672,
      "grad_norm": 0.9083479046821594,
      "learning_rate": 0.00014063095171414024,
      "loss": 1.6229,
      "step": 14790
    },
    {
      "epoch": 3.670407340814682,
      "grad_norm": 0.9024813771247864,
      "learning_rate": 0.00014055974432944753,
      "loss": 1.6918,
      "step": 14800
    },
    {
      "epoch": 3.672887345774692,
      "grad_norm": 0.8541541695594788,
      "learning_rate": 0.00014048851232102222,
      "loss": 1.6573,
      "step": 14810
    },
    {
      "epoch": 3.6753673507347013,
      "grad_norm": 0.8761372566223145,
      "learning_rate": 0.0001404172557321092,
      "loss": 1.5982,
      "step": 14820
    },
    {
      "epoch": 3.6778473556947113,
      "grad_norm": 0.8903871774673462,
      "learning_rate": 0.00014034597460596817,
      "loss": 1.6667,
      "step": 14830
    },
    {
      "epoch": 3.6803273606547213,
      "grad_norm": 0.9093607664108276,
      "learning_rate": 0.00014027466898587374,
      "loss": 1.5902,
      "step": 14840
    },
    {
      "epoch": 3.6828073656147313,
      "grad_norm": 0.9126347303390503,
      "learning_rate": 0.00014020333891511536,
      "loss": 1.6916,
      "step": 14850
    },
    {
      "epoch": 3.6852873705747413,
      "grad_norm": 0.8325662612915039,
      "learning_rate": 0.0001401319844369974,
      "loss": 1.5759,
      "step": 14860
    },
    {
      "epoch": 3.687767375534751,
      "grad_norm": 0.8358848690986633,
      "learning_rate": 0.000140060605594839,
      "loss": 1.6168,
      "step": 14870
    },
    {
      "epoch": 3.690247380494761,
      "grad_norm": 0.8498730063438416,
      "learning_rate": 0.00013998920243197407,
      "loss": 1.6593,
      "step": 14880
    },
    {
      "epoch": 3.692727385454771,
      "grad_norm": 0.846355676651001,
      "learning_rate": 0.0001399177749917514,
      "loss": 1.662,
      "step": 14890
    },
    {
      "epoch": 3.695207390414781,
      "grad_norm": 0.8313154578208923,
      "learning_rate": 0.00013984632331753437,
      "loss": 1.707,
      "step": 14900
    },
    {
      "epoch": 3.697687395374791,
      "grad_norm": 0.9697901606559753,
      "learning_rate": 0.00013977484745270112,
      "loss": 1.6901,
      "step": 14910
    },
    {
      "epoch": 3.7001674003348004,
      "grad_norm": 0.8515355587005615,
      "learning_rate": 0.0001397033474406445,
      "loss": 1.6474,
      "step": 14920
    },
    {
      "epoch": 3.702647405294811,
      "grad_norm": 0.9480315446853638,
      "learning_rate": 0.00013963182332477208,
      "loss": 1.6305,
      "step": 14930
    },
    {
      "epoch": 3.7051274102548204,
      "grad_norm": 0.8484178185462952,
      "learning_rate": 0.0001395602751485059,
      "loss": 1.6325,
      "step": 14940
    },
    {
      "epoch": 3.7076074152148304,
      "grad_norm": 0.9017485976219177,
      "learning_rate": 0.00013948870295528275,
      "loss": 1.6506,
      "step": 14950
    },
    {
      "epoch": 3.7100874201748404,
      "grad_norm": 0.9036349058151245,
      "learning_rate": 0.00013941710678855396,
      "loss": 1.6943,
      "step": 14960
    },
    {
      "epoch": 3.7125674251348504,
      "grad_norm": 0.848396360874176,
      "learning_rate": 0.00013934548669178534,
      "loss": 1.6135,
      "step": 14970
    },
    {
      "epoch": 3.7150474300948604,
      "grad_norm": 0.8705415725708008,
      "learning_rate": 0.0001392738427084574,
      "loss": 1.6606,
      "step": 14980
    },
    {
      "epoch": 3.71752743505487,
      "grad_norm": 0.9253519177436829,
      "learning_rate": 0.00013920217488206502,
      "loss": 1.6466,
      "step": 14990
    },
    {
      "epoch": 3.72000744001488,
      "grad_norm": 0.876288890838623,
      "learning_rate": 0.0001391304832561175,
      "loss": 1.7095,
      "step": 15000
    },
    {
      "epoch": 3.72248744497489,
      "grad_norm": 0.8728715181350708,
      "learning_rate": 0.00013905876787413877,
      "loss": 1.6841,
      "step": 15010
    },
    {
      "epoch": 3.7249674499349,
      "grad_norm": 0.8103494048118591,
      "learning_rate": 0.000138987028779667,
      "loss": 1.6046,
      "step": 15020
    },
    {
      "epoch": 3.72744745489491,
      "grad_norm": 0.8583076596260071,
      "learning_rate": 0.0001389152660162549,
      "loss": 1.6191,
      "step": 15030
    },
    {
      "epoch": 3.7299274598549195,
      "grad_norm": 0.9058374762535095,
      "learning_rate": 0.00013884347962746948,
      "loss": 1.7004,
      "step": 15040
    },
    {
      "epoch": 3.7324074648149295,
      "grad_norm": 0.884195864200592,
      "learning_rate": 0.00013877166965689205,
      "loss": 1.6381,
      "step": 15050
    },
    {
      "epoch": 3.7348874697749395,
      "grad_norm": 0.8272615075111389,
      "learning_rate": 0.00013869983614811838,
      "loss": 1.6541,
      "step": 15060
    },
    {
      "epoch": 3.7373674747349495,
      "grad_norm": 0.8531262874603271,
      "learning_rate": 0.00013862797914475833,
      "loss": 1.6763,
      "step": 15070
    },
    {
      "epoch": 3.7398474796949595,
      "grad_norm": 0.8404091000556946,
      "learning_rate": 0.0001385560986904362,
      "loss": 1.6767,
      "step": 15080
    },
    {
      "epoch": 3.7423274846549694,
      "grad_norm": 0.8463233113288879,
      "learning_rate": 0.00013848419482879041,
      "loss": 1.6518,
      "step": 15090
    },
    {
      "epoch": 3.7448074896149794,
      "grad_norm": 1.0089993476867676,
      "learning_rate": 0.0001384122676034737,
      "loss": 1.6555,
      "step": 15100
    },
    {
      "epoch": 3.747287494574989,
      "grad_norm": 0.9475327134132385,
      "learning_rate": 0.00013834031705815288,
      "loss": 1.6226,
      "step": 15110
    },
    {
      "epoch": 3.749767499534999,
      "grad_norm": 0.8604915142059326,
      "learning_rate": 0.000138268343236509,
      "loss": 1.6499,
      "step": 15120
    },
    {
      "epoch": 3.752247504495009,
      "grad_norm": 0.8939393162727356,
      "learning_rate": 0.0001381963461822372,
      "loss": 1.6966,
      "step": 15130
    },
    {
      "epoch": 3.754727509455019,
      "grad_norm": 0.9622209668159485,
      "learning_rate": 0.0001381243259390467,
      "loss": 1.6915,
      "step": 15140
    },
    {
      "epoch": 3.757207514415029,
      "grad_norm": 0.8684049248695374,
      "learning_rate": 0.00013805228255066087,
      "loss": 1.6985,
      "step": 15150
    },
    {
      "epoch": 3.7596875193750385,
      "grad_norm": 0.8036013245582581,
      "learning_rate": 0.00013798021606081714,
      "loss": 1.6954,
      "step": 15160
    },
    {
      "epoch": 3.7621675243350485,
      "grad_norm": 0.8598616123199463,
      "learning_rate": 0.0001379081265132668,
      "loss": 1.7063,
      "step": 15170
    },
    {
      "epoch": 3.7646475292950585,
      "grad_norm": 0.9074985384941101,
      "learning_rate": 0.00013783601395177538,
      "loss": 1.6476,
      "step": 15180
    },
    {
      "epoch": 3.7671275342550685,
      "grad_norm": 0.8250356912612915,
      "learning_rate": 0.00013776387842012218,
      "loss": 1.6387,
      "step": 15190
    },
    {
      "epoch": 3.7696075392150785,
      "grad_norm": 0.9312306642532349,
      "learning_rate": 0.00013769171996210052,
      "loss": 1.6544,
      "step": 15200
    },
    {
      "epoch": 3.7720875441750885,
      "grad_norm": 0.8160702586174011,
      "learning_rate": 0.00013761953862151773,
      "loss": 1.6221,
      "step": 15210
    },
    {
      "epoch": 3.7745675491350985,
      "grad_norm": 0.9043015837669373,
      "learning_rate": 0.00013754733444219487,
      "loss": 1.621,
      "step": 15220
    },
    {
      "epoch": 3.777047554095108,
      "grad_norm": 0.909737229347229,
      "learning_rate": 0.00013747510746796695,
      "loss": 1.6104,
      "step": 15230
    },
    {
      "epoch": 3.779527559055118,
      "grad_norm": 0.8393818736076355,
      "learning_rate": 0.00013740285774268283,
      "loss": 1.6653,
      "step": 15240
    },
    {
      "epoch": 3.782007564015128,
      "grad_norm": 0.819762647151947,
      "learning_rate": 0.0001373305853102051,
      "loss": 1.6314,
      "step": 15250
    },
    {
      "epoch": 3.784487568975138,
      "grad_norm": 0.8263760805130005,
      "learning_rate": 0.0001372582902144103,
      "loss": 1.5974,
      "step": 15260
    },
    {
      "epoch": 3.786967573935148,
      "grad_norm": 0.8518638014793396,
      "learning_rate": 0.00013718597249918855,
      "loss": 1.5851,
      "step": 15270
    },
    {
      "epoch": 3.7894475788951576,
      "grad_norm": 0.8781617879867554,
      "learning_rate": 0.00013711363220844379,
      "loss": 1.6306,
      "step": 15280
    },
    {
      "epoch": 3.7919275838551676,
      "grad_norm": 0.8384988903999329,
      "learning_rate": 0.0001370412693860937,
      "loss": 1.6278,
      "step": 15290
    },
    {
      "epoch": 3.7944075888151776,
      "grad_norm": 0.8870386481285095,
      "learning_rate": 0.00013696888407606952,
      "loss": 1.6205,
      "step": 15300
    },
    {
      "epoch": 3.7968875937751876,
      "grad_norm": 0.8454505801200867,
      "learning_rate": 0.00013689647632231624,
      "loss": 1.6872,
      "step": 15310
    },
    {
      "epoch": 3.7993675987351976,
      "grad_norm": 0.852705717086792,
      "learning_rate": 0.00013682404616879247,
      "loss": 1.6451,
      "step": 15320
    },
    {
      "epoch": 3.801847603695207,
      "grad_norm": 0.9460528492927551,
      "learning_rate": 0.00013675159365947036,
      "loss": 1.749,
      "step": 15330
    },
    {
      "epoch": 3.8043276086552176,
      "grad_norm": 0.8498387336730957,
      "learning_rate": 0.00013667911883833572,
      "loss": 1.6879,
      "step": 15340
    },
    {
      "epoch": 3.806807613615227,
      "grad_norm": 0.8632598519325256,
      "learning_rate": 0.00013660662174938784,
      "loss": 1.6336,
      "step": 15350
    },
    {
      "epoch": 3.809287618575237,
      "grad_norm": 0.8933224081993103,
      "learning_rate": 0.00013653410243663952,
      "loss": 1.6859,
      "step": 15360
    },
    {
      "epoch": 3.811767623535247,
      "grad_norm": 0.8379520773887634,
      "learning_rate": 0.00013646156094411708,
      "loss": 1.7119,
      "step": 15370
    },
    {
      "epoch": 3.814247628495257,
      "grad_norm": 0.9459693431854248,
      "learning_rate": 0.00013638899731586036,
      "loss": 1.6674,
      "step": 15380
    },
    {
      "epoch": 3.816727633455267,
      "grad_norm": 0.8293569684028625,
      "learning_rate": 0.00013631641159592253,
      "loss": 1.6721,
      "step": 15390
    },
    {
      "epoch": 3.8192076384152767,
      "grad_norm": 0.8475781083106995,
      "learning_rate": 0.00013624380382837016,
      "loss": 1.6964,
      "step": 15400
    },
    {
      "epoch": 3.8216876433752867,
      "grad_norm": 0.87235426902771,
      "learning_rate": 0.0001361711740572834,
      "loss": 1.6833,
      "step": 15410
    },
    {
      "epoch": 3.8241676483352967,
      "grad_norm": 0.8890097141265869,
      "learning_rate": 0.00013609852232675558,
      "loss": 1.6289,
      "step": 15420
    },
    {
      "epoch": 3.8266476532953067,
      "grad_norm": 0.8693356513977051,
      "learning_rate": 0.00013602584868089334,
      "loss": 1.6395,
      "step": 15430
    },
    {
      "epoch": 3.8291276582553166,
      "grad_norm": 0.8777374029159546,
      "learning_rate": 0.00013595315316381677,
      "loss": 1.6592,
      "step": 15440
    },
    {
      "epoch": 3.831607663215326,
      "grad_norm": 0.8257191181182861,
      "learning_rate": 0.00013588043581965912,
      "loss": 1.6437,
      "step": 15450
    },
    {
      "epoch": 3.834087668175336,
      "grad_norm": 0.8561131954193115,
      "learning_rate": 0.00013580769669256694,
      "loss": 1.7308,
      "step": 15460
    },
    {
      "epoch": 3.836567673135346,
      "grad_norm": 0.8346956372261047,
      "learning_rate": 0.00013573493582670003,
      "loss": 1.6415,
      "step": 15470
    },
    {
      "epoch": 3.839047678095356,
      "grad_norm": 0.8684515357017517,
      "learning_rate": 0.0001356621532662313,
      "loss": 1.5912,
      "step": 15480
    },
    {
      "epoch": 3.841527683055366,
      "grad_norm": 0.8844882845878601,
      "learning_rate": 0.00013558934905534692,
      "loss": 1.5855,
      "step": 15490
    },
    {
      "epoch": 3.844007688015376,
      "grad_norm": 0.8354759216308594,
      "learning_rate": 0.00013551652323824618,
      "loss": 1.6571,
      "step": 15500
    },
    {
      "epoch": 3.846487692975386,
      "grad_norm": 0.8265485763549805,
      "learning_rate": 0.00013544367585914142,
      "loss": 1.5934,
      "step": 15510
    },
    {
      "epoch": 3.8489676979353957,
      "grad_norm": 0.8156879544258118,
      "learning_rate": 0.00013537080696225814,
      "loss": 1.6905,
      "step": 15520
    },
    {
      "epoch": 3.8514477028954057,
      "grad_norm": 0.8880894184112549,
      "learning_rate": 0.00013529791659183492,
      "loss": 1.6486,
      "step": 15530
    },
    {
      "epoch": 3.8539277078554157,
      "grad_norm": 0.8683051466941833,
      "learning_rate": 0.00013522500479212337,
      "loss": 1.6602,
      "step": 15540
    },
    {
      "epoch": 3.8564077128154257,
      "grad_norm": 0.8110718131065369,
      "learning_rate": 0.000135152071607388,
      "loss": 1.6421,
      "step": 15550
    },
    {
      "epoch": 3.8588877177754357,
      "grad_norm": 0.8966351747512817,
      "learning_rate": 0.00013507911708190645,
      "loss": 1.6323,
      "step": 15560
    },
    {
      "epoch": 3.8613677227354453,
      "grad_norm": 0.8900407552719116,
      "learning_rate": 0.00013500614125996923,
      "loss": 1.6449,
      "step": 15570
    },
    {
      "epoch": 3.8638477276954553,
      "grad_norm": 0.8171928524971008,
      "learning_rate": 0.00013493314418587982,
      "loss": 1.7337,
      "step": 15580
    },
    {
      "epoch": 3.8663277326554653,
      "grad_norm": 0.8650508522987366,
      "learning_rate": 0.00013486012590395457,
      "loss": 1.6864,
      "step": 15590
    },
    {
      "epoch": 3.8688077376154753,
      "grad_norm": 0.8077772259712219,
      "learning_rate": 0.00013478708645852272,
      "loss": 1.659,
      "step": 15600
    },
    {
      "epoch": 3.8712877425754852,
      "grad_norm": 0.9582907557487488,
      "learning_rate": 0.00013471402589392638,
      "loss": 1.7033,
      "step": 15610
    },
    {
      "epoch": 3.873767747535495,
      "grad_norm": 0.897682249546051,
      "learning_rate": 0.00013464094425452044,
      "loss": 1.7148,
      "step": 15620
    },
    {
      "epoch": 3.8762477524955052,
      "grad_norm": 0.8741872310638428,
      "learning_rate": 0.0001345678415846726,
      "loss": 1.6151,
      "step": 15630
    },
    {
      "epoch": 3.878727757455515,
      "grad_norm": 0.8569635152816772,
      "learning_rate": 0.00013449471792876334,
      "loss": 1.6303,
      "step": 15640
    },
    {
      "epoch": 3.881207762415525,
      "grad_norm": 0.9225397706031799,
      "learning_rate": 0.00013442157333118588,
      "loss": 1.6697,
      "step": 15650
    },
    {
      "epoch": 3.883687767375535,
      "grad_norm": 0.8708016872406006,
      "learning_rate": 0.0001343484078363461,
      "loss": 1.6731,
      "step": 15660
    },
    {
      "epoch": 3.886167772335545,
      "grad_norm": 0.8538072109222412,
      "learning_rate": 0.0001342752214886627,
      "loss": 1.6666,
      "step": 15670
    },
    {
      "epoch": 3.8886477772955548,
      "grad_norm": 0.8768512606620789,
      "learning_rate": 0.00013420201433256689,
      "loss": 1.6598,
      "step": 15680
    },
    {
      "epoch": 3.8911277822555643,
      "grad_norm": 0.8463048338890076,
      "learning_rate": 0.00013412878641250257,
      "loss": 1.6111,
      "step": 15690
    },
    {
      "epoch": 3.8936077872155743,
      "grad_norm": 0.9186829328536987,
      "learning_rate": 0.00013405553777292626,
      "loss": 1.6806,
      "step": 15700
    },
    {
      "epoch": 3.8960877921755843,
      "grad_norm": 0.8851925730705261,
      "learning_rate": 0.0001339822684583071,
      "loss": 1.6311,
      "step": 15710
    },
    {
      "epoch": 3.8985677971355943,
      "grad_norm": 0.9072033762931824,
      "learning_rate": 0.00013390897851312667,
      "loss": 1.6812,
      "step": 15720
    },
    {
      "epoch": 3.9010478020956043,
      "grad_norm": 0.8682987093925476,
      "learning_rate": 0.00013383566798187916,
      "loss": 1.7077,
      "step": 15730
    },
    {
      "epoch": 3.903527807055614,
      "grad_norm": 0.9126836657524109,
      "learning_rate": 0.00013376233690907125,
      "loss": 1.6453,
      "step": 15740
    },
    {
      "epoch": 3.9060078120156243,
      "grad_norm": 0.8302199840545654,
      "learning_rate": 0.000133688985339222,
      "loss": 1.6918,
      "step": 15750
    },
    {
      "epoch": 3.908487816975634,
      "grad_norm": 0.8872604966163635,
      "learning_rate": 0.0001336156133168631,
      "loss": 1.6749,
      "step": 15760
    },
    {
      "epoch": 3.910967821935644,
      "grad_norm": 0.8368204832077026,
      "learning_rate": 0.00013354222088653847,
      "loss": 1.6291,
      "step": 15770
    },
    {
      "epoch": 3.913447826895654,
      "grad_norm": 0.8654637336730957,
      "learning_rate": 0.0001334688080928045,
      "loss": 1.6324,
      "step": 15780
    },
    {
      "epoch": 3.915927831855664,
      "grad_norm": 0.8717195391654968,
      "learning_rate": 0.00013339537498022996,
      "loss": 1.6609,
      "step": 15790
    },
    {
      "epoch": 3.918407836815674,
      "grad_norm": 1.0029350519180298,
      "learning_rate": 0.00013332192159339594,
      "loss": 1.5961,
      "step": 15800
    },
    {
      "epoch": 3.9208878417756834,
      "grad_norm": 0.8780935406684875,
      "learning_rate": 0.0001332484479768958,
      "loss": 1.6268,
      "step": 15810
    },
    {
      "epoch": 3.9233678467356934,
      "grad_norm": 0.8545321822166443,
      "learning_rate": 0.00013317495417533524,
      "loss": 1.6405,
      "step": 15820
    },
    {
      "epoch": 3.9258478516957034,
      "grad_norm": 0.8589227199554443,
      "learning_rate": 0.00013310144023333213,
      "loss": 1.6694,
      "step": 15830
    },
    {
      "epoch": 3.9283278566557134,
      "grad_norm": 0.820460855960846,
      "learning_rate": 0.00013302790619551674,
      "loss": 1.6752,
      "step": 15840
    },
    {
      "epoch": 3.9308078616157234,
      "grad_norm": 0.9134910106658936,
      "learning_rate": 0.0001329543521065313,
      "loss": 1.6383,
      "step": 15850
    },
    {
      "epoch": 3.933287866575733,
      "grad_norm": 0.8335714340209961,
      "learning_rate": 0.0001328807780110304,
      "loss": 1.6459,
      "step": 15860
    },
    {
      "epoch": 3.935767871535743,
      "grad_norm": 0.8406058549880981,
      "learning_rate": 0.00013280718395368072,
      "loss": 1.6956,
      "step": 15870
    },
    {
      "epoch": 3.938247876495753,
      "grad_norm": 0.7940702438354492,
      "learning_rate": 0.00013273356997916104,
      "loss": 1.6769,
      "step": 15880
    },
    {
      "epoch": 3.940727881455763,
      "grad_norm": 0.9058804512023926,
      "learning_rate": 0.00013265993613216224,
      "loss": 1.6029,
      "step": 15890
    },
    {
      "epoch": 3.943207886415773,
      "grad_norm": 0.8984003663063049,
      "learning_rate": 0.00013258628245738725,
      "loss": 1.6405,
      "step": 15900
    },
    {
      "epoch": 3.945687891375783,
      "grad_norm": 0.8421421051025391,
      "learning_rate": 0.0001325126089995511,
      "loss": 1.6769,
      "step": 15910
    },
    {
      "epoch": 3.948167896335793,
      "grad_norm": 0.9049249887466431,
      "learning_rate": 0.00013243891580338072,
      "loss": 1.7335,
      "step": 15920
    },
    {
      "epoch": 3.9506479012958025,
      "grad_norm": 0.9383307099342346,
      "learning_rate": 0.00013236520291361515,
      "loss": 1.6907,
      "step": 15930
    },
    {
      "epoch": 3.9531279062558125,
      "grad_norm": 0.8742726445198059,
      "learning_rate": 0.00013229147037500534,
      "loss": 1.688,
      "step": 15940
    },
    {
      "epoch": 3.9556079112158224,
      "grad_norm": 0.8476269245147705,
      "learning_rate": 0.00013221771823231405,
      "loss": 1.6047,
      "step": 15950
    },
    {
      "epoch": 3.9580879161758324,
      "grad_norm": 0.9126971364021301,
      "learning_rate": 0.00013214394653031616,
      "loss": 1.6945,
      "step": 15960
    },
    {
      "epoch": 3.9605679211358424,
      "grad_norm": 0.9117973446846008,
      "learning_rate": 0.00013207015531379828,
      "loss": 1.6821,
      "step": 15970
    },
    {
      "epoch": 3.963047926095852,
      "grad_norm": 0.8037660121917725,
      "learning_rate": 0.00013199634462755886,
      "loss": 1.5992,
      "step": 15980
    },
    {
      "epoch": 3.965527931055862,
      "grad_norm": 0.8710912466049194,
      "learning_rate": 0.0001319225145164082,
      "loss": 1.6744,
      "step": 15990
    },
    {
      "epoch": 3.968007936015872,
      "grad_norm": 0.8851284980773926,
      "learning_rate": 0.00013184866502516845,
      "loss": 1.6502,
      "step": 16000
    },
    {
      "epoch": 3.970487940975882,
      "grad_norm": 0.8333200216293335,
      "learning_rate": 0.00013177479619867345,
      "loss": 1.6874,
      "step": 16010
    },
    {
      "epoch": 3.972967945935892,
      "grad_norm": 0.8478822708129883,
      "learning_rate": 0.00013170090808176883,
      "loss": 1.6114,
      "step": 16020
    },
    {
      "epoch": 3.9754479508959015,
      "grad_norm": 0.8424682021141052,
      "learning_rate": 0.00013162700071931184,
      "loss": 1.6576,
      "step": 16030
    },
    {
      "epoch": 3.977927955855912,
      "grad_norm": 0.8300601840019226,
      "learning_rate": 0.00013155307415617154,
      "loss": 1.5744,
      "step": 16040
    },
    {
      "epoch": 3.9804079608159215,
      "grad_norm": 0.8115140199661255,
      "learning_rate": 0.0001314791284372286,
      "loss": 1.6163,
      "step": 16050
    },
    {
      "epoch": 3.9828879657759315,
      "grad_norm": 0.895980715751648,
      "learning_rate": 0.00013140516360737523,
      "loss": 1.5989,
      "step": 16060
    },
    {
      "epoch": 3.9853679707359415,
      "grad_norm": 0.879489004611969,
      "learning_rate": 0.00013133117971151533,
      "loss": 1.6623,
      "step": 16070
    },
    {
      "epoch": 3.9878479756959515,
      "grad_norm": 0.8440234661102295,
      "learning_rate": 0.00013125717679456447,
      "loss": 1.7006,
      "step": 16080
    },
    {
      "epoch": 3.9903279806559615,
      "grad_norm": 0.8498426675796509,
      "learning_rate": 0.00013118315490144954,
      "loss": 1.6202,
      "step": 16090
    },
    {
      "epoch": 3.992807985615971,
      "grad_norm": 0.9186023473739624,
      "learning_rate": 0.00013110911407710908,
      "loss": 1.6572,
      "step": 16100
    },
    {
      "epoch": 3.995287990575981,
      "grad_norm": 0.858972430229187,
      "learning_rate": 0.00013103505436649318,
      "loss": 1.6199,
      "step": 16110
    },
    {
      "epoch": 3.997767995535991,
      "grad_norm": 0.9124534130096436,
      "learning_rate": 0.0001309609758145633,
      "loss": 1.6951,
      "step": 16120
    },
    {
      "epoch": 4.000248000496001,
      "grad_norm": 0.9108834862709045,
      "learning_rate": 0.00013088687846629234,
      "loss": 1.6633,
      "step": 16130
    },
    {
      "epoch": 4.002728005456011,
      "grad_norm": 0.8702978491783142,
      "learning_rate": 0.00013081276236666468,
      "loss": 1.5845,
      "step": 16140
    },
    {
      "epoch": 4.005208010416021,
      "grad_norm": 0.8718990683555603,
      "learning_rate": 0.00013073862756067604,
      "loss": 1.562,
      "step": 16150
    },
    {
      "epoch": 4.007688015376031,
      "grad_norm": 0.8283565044403076,
      "learning_rate": 0.00013066447409333345,
      "loss": 1.5418,
      "step": 16160
    },
    {
      "epoch": 4.010168020336041,
      "grad_norm": 0.833257794380188,
      "learning_rate": 0.00013059030200965536,
      "loss": 1.5805,
      "step": 16170
    },
    {
      "epoch": 4.012648025296051,
      "grad_norm": 0.8424034714698792,
      "learning_rate": 0.00013051611135467145,
      "loss": 1.5345,
      "step": 16180
    },
    {
      "epoch": 4.015128030256061,
      "grad_norm": 0.9445743560791016,
      "learning_rate": 0.00013044190217342277,
      "loss": 1.5442,
      "step": 16190
    },
    {
      "epoch": 4.01760803521607,
      "grad_norm": 0.9044946432113647,
      "learning_rate": 0.00013036767451096148,
      "loss": 1.5643,
      "step": 16200
    },
    {
      "epoch": 4.020088040176081,
      "grad_norm": 0.8936172127723694,
      "learning_rate": 0.00013029342841235105,
      "loss": 1.5514,
      "step": 16210
    },
    {
      "epoch": 4.02256804513609,
      "grad_norm": 0.8858177661895752,
      "learning_rate": 0.00013021916392266618,
      "loss": 1.5782,
      "step": 16220
    },
    {
      "epoch": 4.025048050096101,
      "grad_norm": 0.8533552885055542,
      "learning_rate": 0.0001301448810869926,
      "loss": 1.5399,
      "step": 16230
    },
    {
      "epoch": 4.02752805505611,
      "grad_norm": 0.8473833203315735,
      "learning_rate": 0.00013007057995042732,
      "loss": 1.54,
      "step": 16240
    },
    {
      "epoch": 4.03000806001612,
      "grad_norm": 0.972991943359375,
      "learning_rate": 0.00012999626055807838,
      "loss": 1.555,
      "step": 16250
    },
    {
      "epoch": 4.03248806497613,
      "grad_norm": 0.9364944100379944,
      "learning_rate": 0.00012992192295506488,
      "loss": 1.5752,
      "step": 16260
    },
    {
      "epoch": 4.03496806993614,
      "grad_norm": 0.852487325668335,
      "learning_rate": 0.00012984756718651704,
      "loss": 1.5312,
      "step": 16270
    },
    {
      "epoch": 4.03744807489615,
      "grad_norm": 0.9019510746002197,
      "learning_rate": 0.00012977319329757615,
      "loss": 1.5375,
      "step": 16280
    },
    {
      "epoch": 4.03992807985616,
      "grad_norm": 0.8527230024337769,
      "learning_rate": 0.00012969880133339437,
      "loss": 1.5458,
      "step": 16290
    },
    {
      "epoch": 4.042408084816169,
      "grad_norm": 0.9539262056350708,
      "learning_rate": 0.00012962439133913488,
      "loss": 1.5066,
      "step": 16300
    },
    {
      "epoch": 4.04488808977618,
      "grad_norm": 0.9539197087287903,
      "learning_rate": 0.0001295499633599719,
      "loss": 1.5595,
      "step": 16310
    },
    {
      "epoch": 4.047368094736189,
      "grad_norm": 0.913061261177063,
      "learning_rate": 0.00012947551744109043,
      "loss": 1.5516,
      "step": 16320
    },
    {
      "epoch": 4.0498480996962,
      "grad_norm": 0.9694399833679199,
      "learning_rate": 0.00012940105362768643,
      "loss": 1.5797,
      "step": 16330
    },
    {
      "epoch": 4.052328104656209,
      "grad_norm": 0.8864725232124329,
      "learning_rate": 0.0001293265719649668,
      "loss": 1.5784,
      "step": 16340
    },
    {
      "epoch": 4.05480810961622,
      "grad_norm": 0.9271555542945862,
      "learning_rate": 0.0001292520724981491,
      "loss": 1.5129,
      "step": 16350
    },
    {
      "epoch": 4.057288114576229,
      "grad_norm": 0.9082526564598083,
      "learning_rate": 0.00012917755527246179,
      "loss": 1.5535,
      "step": 16360
    },
    {
      "epoch": 4.059768119536239,
      "grad_norm": 0.8983141183853149,
      "learning_rate": 0.0001291030203331442,
      "loss": 1.4634,
      "step": 16370
    },
    {
      "epoch": 4.062248124496249,
      "grad_norm": 0.8280670046806335,
      "learning_rate": 0.00012902846772544624,
      "loss": 1.5698,
      "step": 16380
    },
    {
      "epoch": 4.064728129456259,
      "grad_norm": 0.8899021148681641,
      "learning_rate": 0.00012895389749462873,
      "loss": 1.576,
      "step": 16390
    },
    {
      "epoch": 4.067208134416269,
      "grad_norm": 0.9653387665748596,
      "learning_rate": 0.000128879309685963,
      "loss": 1.5106,
      "step": 16400
    },
    {
      "epoch": 4.069688139376279,
      "grad_norm": 0.8746720552444458,
      "learning_rate": 0.00012880470434473116,
      "loss": 1.6016,
      "step": 16410
    },
    {
      "epoch": 4.072168144336288,
      "grad_norm": 0.9301053285598755,
      "learning_rate": 0.00012873008151622606,
      "loss": 1.6109,
      "step": 16420
    },
    {
      "epoch": 4.074648149296299,
      "grad_norm": 0.8845133781433105,
      "learning_rate": 0.00012865544124575096,
      "loss": 1.5611,
      "step": 16430
    },
    {
      "epoch": 4.077128154256308,
      "grad_norm": 0.9510053396224976,
      "learning_rate": 0.0001285807835786198,
      "loss": 1.5434,
      "step": 16440
    },
    {
      "epoch": 4.079608159216319,
      "grad_norm": 0.8830054998397827,
      "learning_rate": 0.0001285061085601571,
      "loss": 1.5926,
      "step": 16450
    },
    {
      "epoch": 4.082088164176328,
      "grad_norm": 0.855392336845398,
      "learning_rate": 0.00012843141623569793,
      "loss": 1.61,
      "step": 16460
    },
    {
      "epoch": 4.084568169136339,
      "grad_norm": 0.8734385967254639,
      "learning_rate": 0.00012835670665058778,
      "loss": 1.6126,
      "step": 16470
    },
    {
      "epoch": 4.087048174096348,
      "grad_norm": 0.8623180985450745,
      "learning_rate": 0.00012828197985018276,
      "loss": 1.5737,
      "step": 16480
    },
    {
      "epoch": 4.089528179056358,
      "grad_norm": 0.8639189600944519,
      "learning_rate": 0.0001282072358798493,
      "loss": 1.598,
      "step": 16490
    },
    {
      "epoch": 4.092008184016368,
      "grad_norm": 0.8384979963302612,
      "learning_rate": 0.00012813247478496429,
      "loss": 1.5695,
      "step": 16500
    },
    {
      "epoch": 4.094488188976378,
      "grad_norm": 0.9530404210090637,
      "learning_rate": 0.00012805769661091505,
      "loss": 1.5871,
      "step": 16510
    },
    {
      "epoch": 4.096968193936388,
      "grad_norm": 1.038830280303955,
      "learning_rate": 0.00012798290140309923,
      "loss": 1.5774,
      "step": 16520
    },
    {
      "epoch": 4.099448198896398,
      "grad_norm": 0.9369829893112183,
      "learning_rate": 0.00012790808920692484,
      "loss": 1.5159,
      "step": 16530
    },
    {
      "epoch": 4.101928203856407,
      "grad_norm": 0.9211897253990173,
      "learning_rate": 0.00012783326006781022,
      "loss": 1.5691,
      "step": 16540
    },
    {
      "epoch": 4.104408208816418,
      "grad_norm": 0.9339462518692017,
      "learning_rate": 0.000127758414031184,
      "loss": 1.578,
      "step": 16550
    },
    {
      "epoch": 4.106888213776427,
      "grad_norm": 0.9960739612579346,
      "learning_rate": 0.00012768355114248494,
      "loss": 1.5432,
      "step": 16560
    },
    {
      "epoch": 4.109368218736438,
      "grad_norm": 0.8537956476211548,
      "learning_rate": 0.00012760867144716228,
      "loss": 1.5483,
      "step": 16570
    },
    {
      "epoch": 4.111848223696447,
      "grad_norm": 0.883642852306366,
      "learning_rate": 0.00012753377499067522,
      "loss": 1.5365,
      "step": 16580
    },
    {
      "epoch": 4.114328228656458,
      "grad_norm": 0.9451387524604797,
      "learning_rate": 0.00012745886181849325,
      "loss": 1.6028,
      "step": 16590
    },
    {
      "epoch": 4.116808233616467,
      "grad_norm": 0.9538312554359436,
      "learning_rate": 0.00012738393197609602,
      "loss": 1.6162,
      "step": 16600
    },
    {
      "epoch": 4.119288238576477,
      "grad_norm": 0.8514898419380188,
      "learning_rate": 0.00012730898550897325,
      "loss": 1.5132,
      "step": 16610
    },
    {
      "epoch": 4.121768243536487,
      "grad_norm": 0.9028187394142151,
      "learning_rate": 0.00012723402246262483,
      "loss": 1.5294,
      "step": 16620
    },
    {
      "epoch": 4.124248248496497,
      "grad_norm": 0.8952030539512634,
      "learning_rate": 0.0001271590428825606,
      "loss": 1.5304,
      "step": 16630
    },
    {
      "epoch": 4.126728253456507,
      "grad_norm": 0.9488352537155151,
      "learning_rate": 0.00012708404681430053,
      "loss": 1.5866,
      "step": 16640
    },
    {
      "epoch": 4.129208258416517,
      "grad_norm": 0.9322410821914673,
      "learning_rate": 0.00012700903430337457,
      "loss": 1.5799,
      "step": 16650
    },
    {
      "epoch": 4.131688263376526,
      "grad_norm": 0.9060160517692566,
      "learning_rate": 0.00012693400539532263,
      "loss": 1.6343,
      "step": 16660
    },
    {
      "epoch": 4.134168268336537,
      "grad_norm": 0.9226081967353821,
      "learning_rate": 0.0001268589601356946,
      "loss": 1.5547,
      "step": 16670
    },
    {
      "epoch": 4.136648273296546,
      "grad_norm": 0.9264705181121826,
      "learning_rate": 0.00012678389857005034,
      "loss": 1.6106,
      "step": 16680
    },
    {
      "epoch": 4.139128278256557,
      "grad_norm": 0.9333235621452332,
      "learning_rate": 0.0001267088207439595,
      "loss": 1.5993,
      "step": 16690
    },
    {
      "epoch": 4.141608283216566,
      "grad_norm": 0.9364868998527527,
      "learning_rate": 0.0001266337267030017,
      "loss": 1.5662,
      "step": 16700
    },
    {
      "epoch": 4.144088288176576,
      "grad_norm": 0.8981188535690308,
      "learning_rate": 0.00012655861649276635,
      "loss": 1.5942,
      "step": 16710
    },
    {
      "epoch": 4.146568293136586,
      "grad_norm": 0.916425347328186,
      "learning_rate": 0.00012648349015885273,
      "loss": 1.6121,
      "step": 16720
    },
    {
      "epoch": 4.149048298096596,
      "grad_norm": 0.8678816556930542,
      "learning_rate": 0.00012640834774686985,
      "loss": 1.521,
      "step": 16730
    },
    {
      "epoch": 4.151528303056606,
      "grad_norm": 0.8749919533729553,
      "learning_rate": 0.00012633318930243648,
      "loss": 1.5951,
      "step": 16740
    },
    {
      "epoch": 4.154008308016616,
      "grad_norm": 0.8698053956031799,
      "learning_rate": 0.00012625801487118121,
      "loss": 1.5455,
      "step": 16750
    },
    {
      "epoch": 4.156488312976626,
      "grad_norm": 0.9387723803520203,
      "learning_rate": 0.00012618282449874222,
      "loss": 1.5829,
      "step": 16760
    },
    {
      "epoch": 4.158968317936636,
      "grad_norm": 0.8954786062240601,
      "learning_rate": 0.00012610761823076745,
      "loss": 1.5551,
      "step": 16770
    },
    {
      "epoch": 4.1614483228966455,
      "grad_norm": 0.9120689630508423,
      "learning_rate": 0.00012603239611291443,
      "loss": 1.6123,
      "step": 16780
    },
    {
      "epoch": 4.163928327856656,
      "grad_norm": 0.9066893458366394,
      "learning_rate": 0.0001259571581908504,
      "loss": 1.6861,
      "step": 16790
    },
    {
      "epoch": 4.1664083328166654,
      "grad_norm": 0.9290622472763062,
      "learning_rate": 0.00012588190451025207,
      "loss": 1.5792,
      "step": 16800
    },
    {
      "epoch": 4.168888337776676,
      "grad_norm": 0.9743390679359436,
      "learning_rate": 0.00012580663511680586,
      "loss": 1.5547,
      "step": 16810
    },
    {
      "epoch": 4.171368342736685,
      "grad_norm": 0.8968302607536316,
      "learning_rate": 0.00012573135005620757,
      "loss": 1.5424,
      "step": 16820
    },
    {
      "epoch": 4.173848347696695,
      "grad_norm": 0.8571000099182129,
      "learning_rate": 0.00012565604937416267,
      "loss": 1.5457,
      "step": 16830
    },
    {
      "epoch": 4.176328352656705,
      "grad_norm": 0.9518805146217346,
      "learning_rate": 0.00012558073311638604,
      "loss": 1.5602,
      "step": 16840
    },
    {
      "epoch": 4.178808357616715,
      "grad_norm": 1.0199187994003296,
      "learning_rate": 0.0001255054013286019,
      "loss": 1.5634,
      "step": 16850
    },
    {
      "epoch": 4.181288362576725,
      "grad_norm": 0.9404400587081909,
      "learning_rate": 0.00012543005405654418,
      "loss": 1.5479,
      "step": 16860
    },
    {
      "epoch": 4.183768367536735,
      "grad_norm": 0.9390067458152771,
      "learning_rate": 0.00012535469134595595,
      "loss": 1.5106,
      "step": 16870
    },
    {
      "epoch": 4.186248372496745,
      "grad_norm": 0.9579924941062927,
      "learning_rate": 0.00012527931324258975,
      "loss": 1.5743,
      "step": 16880
    },
    {
      "epoch": 4.188728377456755,
      "grad_norm": 0.9165119528770447,
      "learning_rate": 0.0001252039197922075,
      "loss": 1.5944,
      "step": 16890
    },
    {
      "epoch": 4.1912083824167645,
      "grad_norm": 0.9501695036888123,
      "learning_rate": 0.00012512851104058036,
      "loss": 1.6268,
      "step": 16900
    },
    {
      "epoch": 4.193688387376775,
      "grad_norm": 0.8611862659454346,
      "learning_rate": 0.00012505308703348884,
      "loss": 1.5173,
      "step": 16910
    },
    {
      "epoch": 4.1961683923367845,
      "grad_norm": 0.9533225893974304,
      "learning_rate": 0.0001249776478167227,
      "loss": 1.5802,
      "step": 16920
    },
    {
      "epoch": 4.198648397296795,
      "grad_norm": 0.8947030305862427,
      "learning_rate": 0.0001249021934360809,
      "loss": 1.5093,
      "step": 16930
    },
    {
      "epoch": 4.2011284022568045,
      "grad_norm": 0.886284351348877,
      "learning_rate": 0.00012482672393737164,
      "loss": 1.5132,
      "step": 16940
    },
    {
      "epoch": 4.203608407216814,
      "grad_norm": 0.9174060225486755,
      "learning_rate": 0.00012475123936641228,
      "loss": 1.5967,
      "step": 16950
    },
    {
      "epoch": 4.2060884121768245,
      "grad_norm": 0.9190420508384705,
      "learning_rate": 0.00012467573976902935,
      "loss": 1.5946,
      "step": 16960
    },
    {
      "epoch": 4.208568417136834,
      "grad_norm": 0.8581568598747253,
      "learning_rate": 0.00012460022519105853,
      "loss": 1.5895,
      "step": 16970
    },
    {
      "epoch": 4.2110484220968445,
      "grad_norm": 0.9697762727737427,
      "learning_rate": 0.00012452469567834448,
      "loss": 1.5264,
      "step": 16980
    },
    {
      "epoch": 4.213528427056854,
      "grad_norm": 0.8641528487205505,
      "learning_rate": 0.00012444915127674103,
      "loss": 1.5314,
      "step": 16990
    },
    {
      "epoch": 4.216008432016864,
      "grad_norm": 0.8917909264564514,
      "learning_rate": 0.00012437359203211108,
      "loss": 1.5553,
      "step": 17000
    },
    {
      "epoch": 4.218488436976874,
      "grad_norm": 0.8812795877456665,
      "learning_rate": 0.0001242980179903264,
      "loss": 1.5835,
      "step": 17010
    },
    {
      "epoch": 4.220968441936884,
      "grad_norm": 0.9558031558990479,
      "learning_rate": 0.00012422242919726785,
      "loss": 1.6167,
      "step": 17020
    },
    {
      "epoch": 4.223448446896894,
      "grad_norm": 0.9204807877540588,
      "learning_rate": 0.00012414682569882526,
      "loss": 1.5671,
      "step": 17030
    },
    {
      "epoch": 4.225928451856904,
      "grad_norm": 0.9132686257362366,
      "learning_rate": 0.00012407120754089732,
      "loss": 1.5554,
      "step": 17040
    },
    {
      "epoch": 4.228408456816914,
      "grad_norm": 0.823129415512085,
      "learning_rate": 0.00012399557476939162,
      "loss": 1.604,
      "step": 17050
    },
    {
      "epoch": 4.230888461776924,
      "grad_norm": 0.889136552810669,
      "learning_rate": 0.00012391992743022472,
      "loss": 1.5277,
      "step": 17060
    },
    {
      "epoch": 4.233368466736933,
      "grad_norm": 0.9668729901313782,
      "learning_rate": 0.0001238442655693219,
      "loss": 1.5865,
      "step": 17070
    },
    {
      "epoch": 4.235848471696944,
      "grad_norm": 0.8794295191764832,
      "learning_rate": 0.00012376858923261733,
      "loss": 1.6106,
      "step": 17080
    },
    {
      "epoch": 4.238328476656953,
      "grad_norm": 0.884594202041626,
      "learning_rate": 0.00012369289846605395,
      "loss": 1.65,
      "step": 17090
    },
    {
      "epoch": 4.2408084816169636,
      "grad_norm": 0.9115503430366516,
      "learning_rate": 0.00012361719331558345,
      "loss": 1.5805,
      "step": 17100
    },
    {
      "epoch": 4.243288486576973,
      "grad_norm": 0.9228629469871521,
      "learning_rate": 0.00012354147382716627,
      "loss": 1.5496,
      "step": 17110
    },
    {
      "epoch": 4.245768491536983,
      "grad_norm": 0.9323840737342834,
      "learning_rate": 0.00012346574004677154,
      "loss": 1.5651,
      "step": 17120
    },
    {
      "epoch": 4.248248496496993,
      "grad_norm": 0.9578195810317993,
      "learning_rate": 0.00012338999202037712,
      "loss": 1.5877,
      "step": 17130
    },
    {
      "epoch": 4.250728501457003,
      "grad_norm": 0.9142360687255859,
      "learning_rate": 0.00012331422979396935,
      "loss": 1.6024,
      "step": 17140
    },
    {
      "epoch": 4.253208506417013,
      "grad_norm": 0.8800964951515198,
      "learning_rate": 0.0001232384534135435,
      "loss": 1.5427,
      "step": 17150
    },
    {
      "epoch": 4.255688511377023,
      "grad_norm": 0.9429461359977722,
      "learning_rate": 0.00012316266292510306,
      "loss": 1.5537,
      "step": 17160
    },
    {
      "epoch": 4.258168516337033,
      "grad_norm": 0.9197307229042053,
      "learning_rate": 0.00012308685837466037,
      "loss": 1.5751,
      "step": 17170
    },
    {
      "epoch": 4.260648521297043,
      "grad_norm": 0.936857283115387,
      "learning_rate": 0.00012301103980823618,
      "loss": 1.5294,
      "step": 17180
    },
    {
      "epoch": 4.263128526257052,
      "grad_norm": 0.8872460126876831,
      "learning_rate": 0.0001229352072718598,
      "loss": 1.5525,
      "step": 17190
    },
    {
      "epoch": 4.265608531217063,
      "grad_norm": 0.9296783804893494,
      "learning_rate": 0.00012285936081156897,
      "loss": 1.6134,
      "step": 17200
    },
    {
      "epoch": 4.268088536177072,
      "grad_norm": 0.9605028629302979,
      "learning_rate": 0.00012278350047340993,
      "loss": 1.5943,
      "step": 17210
    },
    {
      "epoch": 4.270568541137083,
      "grad_norm": 0.9095436930656433,
      "learning_rate": 0.00012270762630343734,
      "loss": 1.5653,
      "step": 17220
    },
    {
      "epoch": 4.273048546097092,
      "grad_norm": 0.8624875545501709,
      "learning_rate": 0.00012263173834771418,
      "loss": 1.5652,
      "step": 17230
    },
    {
      "epoch": 4.275528551057102,
      "grad_norm": 0.8745260238647461,
      "learning_rate": 0.00012255583665231196,
      "loss": 1.5874,
      "step": 17240
    },
    {
      "epoch": 4.278008556017112,
      "grad_norm": 0.9086858034133911,
      "learning_rate": 0.00012247992126331034,
      "loss": 1.6238,
      "step": 17250
    },
    {
      "epoch": 4.280488560977122,
      "grad_norm": 0.8863484859466553,
      "learning_rate": 0.00012240399222679746,
      "loss": 1.5528,
      "step": 17260
    },
    {
      "epoch": 4.282968565937132,
      "grad_norm": 0.9518700838088989,
      "learning_rate": 0.00012232804958886963,
      "loss": 1.6185,
      "step": 17270
    },
    {
      "epoch": 4.285448570897142,
      "grad_norm": 0.9303643107414246,
      "learning_rate": 0.00012225209339563145,
      "loss": 1.5571,
      "step": 17280
    },
    {
      "epoch": 4.287928575857151,
      "grad_norm": 0.8627997636795044,
      "learning_rate": 0.00012217612369319579,
      "loss": 1.5377,
      "step": 17290
    },
    {
      "epoch": 4.290408580817162,
      "grad_norm": 0.9821747541427612,
      "learning_rate": 0.00012210014052768371,
      "loss": 1.5882,
      "step": 17300
    },
    {
      "epoch": 4.292888585777171,
      "grad_norm": 0.9349398612976074,
      "learning_rate": 0.00012202414394522436,
      "loss": 1.5169,
      "step": 17310
    },
    {
      "epoch": 4.295368590737182,
      "grad_norm": 0.8663684725761414,
      "learning_rate": 0.00012194813399195518,
      "loss": 1.5934,
      "step": 17320
    },
    {
      "epoch": 4.297848595697191,
      "grad_norm": 0.9309785962104797,
      "learning_rate": 0.00012187211071402154,
      "loss": 1.578,
      "step": 17330
    },
    {
      "epoch": 4.300328600657202,
      "grad_norm": 0.9538763165473938,
      "learning_rate": 0.0001217960741575771,
      "loss": 1.593,
      "step": 17340
    },
    {
      "epoch": 4.302808605617211,
      "grad_norm": 0.8893478512763977,
      "learning_rate": 0.00012172002436878347,
      "loss": 1.5809,
      "step": 17350
    },
    {
      "epoch": 4.305288610577221,
      "grad_norm": 0.9751872420310974,
      "learning_rate": 0.00012164396139381029,
      "loss": 1.6679,
      "step": 17360
    },
    {
      "epoch": 4.307768615537231,
      "grad_norm": 0.908886730670929,
      "learning_rate": 0.00012156788527883522,
      "loss": 1.5627,
      "step": 17370
    },
    {
      "epoch": 4.310248620497241,
      "grad_norm": 0.8988890051841736,
      "learning_rate": 0.00012149179607004396,
      "loss": 1.6336,
      "step": 17380
    },
    {
      "epoch": 4.312728625457251,
      "grad_norm": 0.936722993850708,
      "learning_rate": 0.00012141569381363004,
      "loss": 1.6238,
      "step": 17390
    },
    {
      "epoch": 4.315208630417261,
      "grad_norm": 0.9053364992141724,
      "learning_rate": 0.00012133957855579501,
      "loss": 1.585,
      "step": 17400
    },
    {
      "epoch": 4.317688635377271,
      "grad_norm": 0.8379243016242981,
      "learning_rate": 0.00012126345034274824,
      "loss": 1.6305,
      "step": 17410
    },
    {
      "epoch": 4.320168640337281,
      "grad_norm": 1.0416642427444458,
      "learning_rate": 0.00012118730922070707,
      "loss": 1.5581,
      "step": 17420
    },
    {
      "epoch": 4.32264864529729,
      "grad_norm": 0.9125607013702393,
      "learning_rate": 0.00012111115523589651,
      "loss": 1.5019,
      "step": 17430
    },
    {
      "epoch": 4.325128650257301,
      "grad_norm": 0.9234462976455688,
      "learning_rate": 0.00012103498843454959,
      "loss": 1.5838,
      "step": 17440
    },
    {
      "epoch": 4.32760865521731,
      "grad_norm": 0.9076301455497742,
      "learning_rate": 0.00012095880886290691,
      "loss": 1.6045,
      "step": 17450
    },
    {
      "epoch": 4.330088660177321,
      "grad_norm": 0.8853614330291748,
      "learning_rate": 0.000120882616567217,
      "loss": 1.6001,
      "step": 17460
    },
    {
      "epoch": 4.33256866513733,
      "grad_norm": 0.9661962389945984,
      "learning_rate": 0.00012080641159373598,
      "loss": 1.6082,
      "step": 17470
    },
    {
      "epoch": 4.33504867009734,
      "grad_norm": 0.9738268852233887,
      "learning_rate": 0.00012073019398872778,
      "loss": 1.6306,
      "step": 17480
    },
    {
      "epoch": 4.33752867505735,
      "grad_norm": 0.9750413298606873,
      "learning_rate": 0.00012065396379846387,
      "loss": 1.5809,
      "step": 17490
    },
    {
      "epoch": 4.34000868001736,
      "grad_norm": 0.9576679468154907,
      "learning_rate": 0.00012057772106922349,
      "loss": 1.5498,
      "step": 17500
    },
    {
      "epoch": 4.34248868497737,
      "grad_norm": 0.9164273738861084,
      "learning_rate": 0.00012050146584729343,
      "loss": 1.6447,
      "step": 17510
    },
    {
      "epoch": 4.34496868993738,
      "grad_norm": 0.9457810521125793,
      "learning_rate": 0.00012042519817896804,
      "loss": 1.615,
      "step": 17520
    },
    {
      "epoch": 4.347448694897389,
      "grad_norm": 0.925794243812561,
      "learning_rate": 0.00012034891811054929,
      "loss": 1.5567,
      "step": 17530
    },
    {
      "epoch": 4.3499286998574,
      "grad_norm": 0.8849574327468872,
      "learning_rate": 0.00012027262568834658,
      "loss": 1.5638,
      "step": 17540
    },
    {
      "epoch": 4.352408704817409,
      "grad_norm": 0.851435661315918,
      "learning_rate": 0.00012019632095867697,
      "loss": 1.5958,
      "step": 17550
    },
    {
      "epoch": 4.35488870977742,
      "grad_norm": 0.9845481514930725,
      "learning_rate": 0.00012012000396786485,
      "loss": 1.6096,
      "step": 17560
    },
    {
      "epoch": 4.357368714737429,
      "grad_norm": 0.9225103259086609,
      "learning_rate": 0.00012004367476224206,
      "loss": 1.5877,
      "step": 17570
    },
    {
      "epoch": 4.359848719697439,
      "grad_norm": 0.8676236867904663,
      "learning_rate": 0.00011996733338814794,
      "loss": 1.5761,
      "step": 17580
    },
    {
      "epoch": 4.362328724657449,
      "grad_norm": 1.0130139589309692,
      "learning_rate": 0.00011989097989192917,
      "loss": 1.6344,
      "step": 17590
    },
    {
      "epoch": 4.364808729617459,
      "grad_norm": 0.9105481505393982,
      "learning_rate": 0.00011981461431993977,
      "loss": 1.5979,
      "step": 17600
    },
    {
      "epoch": 4.367288734577469,
      "grad_norm": 0.8567654490470886,
      "learning_rate": 0.00011973823671854112,
      "loss": 1.5648,
      "step": 17610
    },
    {
      "epoch": 4.369768739537479,
      "grad_norm": 0.8724489808082581,
      "learning_rate": 0.00011966184713410191,
      "loss": 1.5438,
      "step": 17620
    },
    {
      "epoch": 4.372248744497489,
      "grad_norm": 0.9080607891082764,
      "learning_rate": 0.00011958544561299807,
      "loss": 1.5705,
      "step": 17630
    },
    {
      "epoch": 4.374728749457499,
      "grad_norm": 0.9310619831085205,
      "learning_rate": 0.00011950903220161285,
      "loss": 1.5964,
      "step": 17640
    },
    {
      "epoch": 4.3772087544175085,
      "grad_norm": 0.9245249032974243,
      "learning_rate": 0.0001194326069463366,
      "loss": 1.6202,
      "step": 17650
    },
    {
      "epoch": 4.379688759377519,
      "grad_norm": 0.8819152116775513,
      "learning_rate": 0.00011935616989356692,
      "loss": 1.5045,
      "step": 17660
    },
    {
      "epoch": 4.382168764337528,
      "grad_norm": 0.9895660877227783,
      "learning_rate": 0.00011927972108970864,
      "loss": 1.5295,
      "step": 17670
    },
    {
      "epoch": 4.384648769297539,
      "grad_norm": 0.923691987991333,
      "learning_rate": 0.00011920326058117364,
      "loss": 1.6531,
      "step": 17680
    },
    {
      "epoch": 4.387128774257548,
      "grad_norm": 0.9826307892799377,
      "learning_rate": 0.00011912678841438087,
      "loss": 1.6019,
      "step": 17690
    },
    {
      "epoch": 4.389608779217559,
      "grad_norm": 0.8674072027206421,
      "learning_rate": 0.0001190503046357565,
      "loss": 1.5409,
      "step": 17700
    },
    {
      "epoch": 4.392088784177568,
      "grad_norm": 0.9578332304954529,
      "learning_rate": 0.00011897380929173364,
      "loss": 1.5687,
      "step": 17710
    },
    {
      "epoch": 4.394568789137578,
      "grad_norm": 0.9147858619689941,
      "learning_rate": 0.00011889730242875243,
      "loss": 1.5923,
      "step": 17720
    },
    {
      "epoch": 4.397048794097588,
      "grad_norm": 0.8567659854888916,
      "learning_rate": 0.00011882078409326002,
      "loss": 1.5652,
      "step": 17730
    },
    {
      "epoch": 4.399528799057598,
      "grad_norm": 0.9948070645332336,
      "learning_rate": 0.00011874425433171055,
      "loss": 1.6276,
      "step": 17740
    },
    {
      "epoch": 4.402008804017608,
      "grad_norm": 0.9563866257667542,
      "learning_rate": 0.00011866771319056504,
      "loss": 1.5963,
      "step": 17750
    },
    {
      "epoch": 4.404488808977618,
      "grad_norm": 0.8760507702827454,
      "learning_rate": 0.00011859116071629149,
      "loss": 1.6185,
      "step": 17760
    },
    {
      "epoch": 4.4069688139376275,
      "grad_norm": 0.9775795340538025,
      "learning_rate": 0.00011851459695536467,
      "loss": 1.6253,
      "step": 17770
    },
    {
      "epoch": 4.409448818897638,
      "grad_norm": 0.9185107350349426,
      "learning_rate": 0.00011843802195426635,
      "loss": 1.6111,
      "step": 17780
    },
    {
      "epoch": 4.4119288238576475,
      "grad_norm": 0.8888441920280457,
      "learning_rate": 0.00011836143575948501,
      "loss": 1.5946,
      "step": 17790
    },
    {
      "epoch": 4.414408828817658,
      "grad_norm": 0.976132333278656,
      "learning_rate": 0.00011828483841751596,
      "loss": 1.6458,
      "step": 17800
    },
    {
      "epoch": 4.4168888337776675,
      "grad_norm": 0.8497588634490967,
      "learning_rate": 0.0001182082299748613,
      "loss": 1.5704,
      "step": 17810
    },
    {
      "epoch": 4.419368838737677,
      "grad_norm": 0.8618581295013428,
      "learning_rate": 0.00011813161047802985,
      "loss": 1.5761,
      "step": 17820
    },
    {
      "epoch": 4.4218488436976875,
      "grad_norm": 0.8814425468444824,
      "learning_rate": 0.0001180549799735371,
      "loss": 1.5556,
      "step": 17830
    },
    {
      "epoch": 4.424328848657697,
      "grad_norm": 0.9566665291786194,
      "learning_rate": 0.00011797833850790527,
      "loss": 1.6298,
      "step": 17840
    },
    {
      "epoch": 4.4268088536177075,
      "grad_norm": 0.9071756601333618,
      "learning_rate": 0.0001179016861276633,
      "loss": 1.5629,
      "step": 17850
    },
    {
      "epoch": 4.429288858577717,
      "grad_norm": 0.8935988545417786,
      "learning_rate": 0.00011782502287934657,
      "loss": 1.5592,
      "step": 17860
    },
    {
      "epoch": 4.4317688635377275,
      "grad_norm": 0.9023693799972534,
      "learning_rate": 0.00011774834880949728,
      "loss": 1.6003,
      "step": 17870
    },
    {
      "epoch": 4.434248868497737,
      "grad_norm": 0.90963214635849,
      "learning_rate": 0.00011767166396466403,
      "loss": 1.6143,
      "step": 17880
    },
    {
      "epoch": 4.436728873457747,
      "grad_norm": 0.9300664663314819,
      "learning_rate": 0.000117594968391402,
      "loss": 1.4829,
      "step": 17890
    },
    {
      "epoch": 4.439208878417757,
      "grad_norm": 0.9050344824790955,
      "learning_rate": 0.00011751826213627297,
      "loss": 1.5639,
      "step": 17900
    },
    {
      "epoch": 4.441688883377767,
      "grad_norm": 0.9257349371910095,
      "learning_rate": 0.0001174415452458451,
      "loss": 1.5882,
      "step": 17910
    },
    {
      "epoch": 4.444168888337777,
      "grad_norm": 0.8953958749771118,
      "learning_rate": 0.00011736481776669306,
      "loss": 1.4813,
      "step": 17920
    },
    {
      "epoch": 4.446648893297787,
      "grad_norm": 0.8810930848121643,
      "learning_rate": 0.00011728807974539793,
      "loss": 1.6231,
      "step": 17930
    },
    {
      "epoch": 4.449128898257796,
      "grad_norm": 0.8710618019104004,
      "learning_rate": 0.0001172113312285472,
      "loss": 1.5967,
      "step": 17940
    },
    {
      "epoch": 4.451608903217807,
      "grad_norm": 1.0707768201828003,
      "learning_rate": 0.00011713457226273472,
      "loss": 1.5689,
      "step": 17950
    },
    {
      "epoch": 4.454088908177816,
      "grad_norm": 1.0256376266479492,
      "learning_rate": 0.00011705780289456068,
      "loss": 1.5905,
      "step": 17960
    },
    {
      "epoch": 4.4565689131378265,
      "grad_norm": 0.9599533677101135,
      "learning_rate": 0.00011698102317063164,
      "loss": 1.6043,
      "step": 17970
    },
    {
      "epoch": 4.459048918097836,
      "grad_norm": 0.9416408538818359,
      "learning_rate": 0.00011690423313756035,
      "loss": 1.6063,
      "step": 17980
    },
    {
      "epoch": 4.4615289230578465,
      "grad_norm": 0.9655886888504028,
      "learning_rate": 0.00011682743284196593,
      "loss": 1.5265,
      "step": 17990
    },
    {
      "epoch": 4.464008928017856,
      "grad_norm": 0.9687894582748413,
      "learning_rate": 0.00011675062233047364,
      "loss": 1.633,
      "step": 18000
    },
    {
      "epoch": 4.466488932977866,
      "grad_norm": 0.8962520956993103,
      "learning_rate": 0.00011667380164971495,
      "loss": 1.5742,
      "step": 18010
    },
    {
      "epoch": 4.468968937937876,
      "grad_norm": 1.0037689208984375,
      "learning_rate": 0.00011659697084632759,
      "loss": 1.5413,
      "step": 18020
    },
    {
      "epoch": 4.471448942897886,
      "grad_norm": 0.8900139927864075,
      "learning_rate": 0.00011652012996695533,
      "loss": 1.5923,
      "step": 18030
    },
    {
      "epoch": 4.473928947857896,
      "grad_norm": 0.9598802924156189,
      "learning_rate": 0.00011644327905824808,
      "loss": 1.6259,
      "step": 18040
    },
    {
      "epoch": 4.476408952817906,
      "grad_norm": 0.8544650077819824,
      "learning_rate": 0.00011636641816686191,
      "loss": 1.5554,
      "step": 18050
    },
    {
      "epoch": 4.478888957777915,
      "grad_norm": 0.9171319007873535,
      "learning_rate": 0.0001162895473394589,
      "loss": 1.5885,
      "step": 18060
    },
    {
      "epoch": 4.481368962737926,
      "grad_norm": 0.8878941535949707,
      "learning_rate": 0.00011621266662270709,
      "loss": 1.6038,
      "step": 18070
    },
    {
      "epoch": 4.483848967697935,
      "grad_norm": 0.895150363445282,
      "learning_rate": 0.00011613577606328068,
      "loss": 1.5593,
      "step": 18080
    },
    {
      "epoch": 4.486328972657946,
      "grad_norm": 0.8921257257461548,
      "learning_rate": 0.00011605887570785972,
      "loss": 1.6331,
      "step": 18090
    },
    {
      "epoch": 4.488808977617955,
      "grad_norm": 0.8955435752868652,
      "learning_rate": 0.00011598196560313024,
      "loss": 1.6114,
      "step": 18100
    },
    {
      "epoch": 4.491288982577965,
      "grad_norm": 0.9963474273681641,
      "learning_rate": 0.00011590504579578423,
      "loss": 1.6417,
      "step": 18110
    },
    {
      "epoch": 4.493768987537975,
      "grad_norm": 1.062529444694519,
      "learning_rate": 0.0001158281163325195,
      "loss": 1.6089,
      "step": 18120
    },
    {
      "epoch": 4.496248992497985,
      "grad_norm": 1.002374291419983,
      "learning_rate": 0.00011575117726003979,
      "loss": 1.5972,
      "step": 18130
    },
    {
      "epoch": 4.498728997457995,
      "grad_norm": 0.9317733645439148,
      "learning_rate": 0.00011567422862505464,
      "loss": 1.6363,
      "step": 18140
    },
    {
      "epoch": 4.501209002418005,
      "grad_norm": 0.8911933302879333,
      "learning_rate": 0.00011559727047427938,
      "loss": 1.5452,
      "step": 18150
    },
    {
      "epoch": 4.503689007378015,
      "grad_norm": 1.041032075881958,
      "learning_rate": 0.00011552030285443515,
      "loss": 1.5243,
      "step": 18160
    },
    {
      "epoch": 4.506169012338025,
      "grad_norm": 0.9466549158096313,
      "learning_rate": 0.00011544332581224884,
      "loss": 1.5773,
      "step": 18170
    },
    {
      "epoch": 4.508649017298034,
      "grad_norm": 0.9585476517677307,
      "learning_rate": 0.000115366339394453,
      "loss": 1.5939,
      "step": 18180
    },
    {
      "epoch": 4.511129022258045,
      "grad_norm": 0.8418630361557007,
      "learning_rate": 0.00011528934364778598,
      "loss": 1.6288,
      "step": 18190
    },
    {
      "epoch": 4.513609027218054,
      "grad_norm": 0.8050907850265503,
      "learning_rate": 0.00011521233861899167,
      "loss": 1.6518,
      "step": 18200
    },
    {
      "epoch": 4.516089032178065,
      "grad_norm": 0.9089813232421875,
      "learning_rate": 0.00011513532435481968,
      "loss": 1.6014,
      "step": 18210
    },
    {
      "epoch": 4.518569037138074,
      "grad_norm": 0.9099536538124084,
      "learning_rate": 0.00011505830090202523,
      "loss": 1.5566,
      "step": 18220
    },
    {
      "epoch": 4.521049042098085,
      "grad_norm": 0.9340983033180237,
      "learning_rate": 0.00011498126830736904,
      "loss": 1.5523,
      "step": 18230
    },
    {
      "epoch": 4.523529047058094,
      "grad_norm": 0.8862547278404236,
      "learning_rate": 0.00011490422661761744,
      "loss": 1.575,
      "step": 18240
    },
    {
      "epoch": 4.526009052018104,
      "grad_norm": 0.9429985284805298,
      "learning_rate": 0.00011482717587954232,
      "loss": 1.6204,
      "step": 18250
    },
    {
      "epoch": 4.528489056978114,
      "grad_norm": 0.9843054413795471,
      "learning_rate": 0.00011475011613992097,
      "loss": 1.5703,
      "step": 18260
    },
    {
      "epoch": 4.530969061938124,
      "grad_norm": 0.9080626964569092,
      "learning_rate": 0.00011467304744553618,
      "loss": 1.5923,
      "step": 18270
    },
    {
      "epoch": 4.533449066898134,
      "grad_norm": 0.8902535438537598,
      "learning_rate": 0.00011459596984317622,
      "loss": 1.5964,
      "step": 18280
    },
    {
      "epoch": 4.535929071858144,
      "grad_norm": 0.9213718175888062,
      "learning_rate": 0.00011451888337963475,
      "loss": 1.5722,
      "step": 18290
    },
    {
      "epoch": 4.538409076818153,
      "grad_norm": 0.8145584464073181,
      "learning_rate": 0.00011444178810171073,
      "loss": 1.6199,
      "step": 18300
    },
    {
      "epoch": 4.540889081778164,
      "grad_norm": 0.9201323390007019,
      "learning_rate": 0.00011436468405620859,
      "loss": 1.5207,
      "step": 18310
    },
    {
      "epoch": 4.543369086738173,
      "grad_norm": 0.8831585645675659,
      "learning_rate": 0.00011428757128993802,
      "loss": 1.5228,
      "step": 18320
    },
    {
      "epoch": 4.545849091698184,
      "grad_norm": 0.8730006814002991,
      "learning_rate": 0.00011421044984971399,
      "loss": 1.583,
      "step": 18330
    },
    {
      "epoch": 4.548329096658193,
      "grad_norm": 0.9736560583114624,
      "learning_rate": 0.00011413331978235678,
      "loss": 1.5421,
      "step": 18340
    },
    {
      "epoch": 4.550809101618203,
      "grad_norm": 0.9348646402359009,
      "learning_rate": 0.00011405618113469185,
      "loss": 1.6008,
      "step": 18350
    },
    {
      "epoch": 4.553289106578213,
      "grad_norm": 0.9965288043022156,
      "learning_rate": 0.00011397903395354996,
      "loss": 1.5903,
      "step": 18360
    },
    {
      "epoch": 4.555769111538223,
      "grad_norm": 0.9236303567886353,
      "learning_rate": 0.00011390187828576697,
      "loss": 1.6122,
      "step": 18370
    },
    {
      "epoch": 4.558249116498233,
      "grad_norm": 0.8591942191123962,
      "learning_rate": 0.0001138247141781839,
      "loss": 1.5646,
      "step": 18380
    },
    {
      "epoch": 4.560729121458243,
      "grad_norm": 0.9395301342010498,
      "learning_rate": 0.00011374754167764696,
      "loss": 1.5958,
      "step": 18390
    },
    {
      "epoch": 4.563209126418252,
      "grad_norm": 0.9221425652503967,
      "learning_rate": 0.00011367036083100735,
      "loss": 1.5468,
      "step": 18400
    },
    {
      "epoch": 4.565689131378263,
      "grad_norm": 0.9344799518585205,
      "learning_rate": 0.00011359317168512142,
      "loss": 1.6367,
      "step": 18410
    },
    {
      "epoch": 4.568169136338272,
      "grad_norm": 1.4354673624038696,
      "learning_rate": 0.00011351597428685055,
      "loss": 1.5494,
      "step": 18420
    },
    {
      "epoch": 4.570649141298283,
      "grad_norm": 0.9043157696723938,
      "learning_rate": 0.00011343876868306109,
      "loss": 1.5759,
      "step": 18430
    },
    {
      "epoch": 4.573129146258292,
      "grad_norm": 1.0231481790542603,
      "learning_rate": 0.00011336155492062439,
      "loss": 1.5701,
      "step": 18440
    },
    {
      "epoch": 4.575609151218303,
      "grad_norm": 0.9277271032333374,
      "learning_rate": 0.00011328433304641679,
      "loss": 1.6045,
      "step": 18450
    },
    {
      "epoch": 4.578089156178312,
      "grad_norm": 0.9549906253814697,
      "learning_rate": 0.0001132071031073195,
      "loss": 1.5703,
      "step": 18460
    },
    {
      "epoch": 4.580569161138322,
      "grad_norm": 0.8800386190414429,
      "learning_rate": 0.00011312986515021862,
      "loss": 1.598,
      "step": 18470
    },
    {
      "epoch": 4.583049166098332,
      "grad_norm": 0.9046245217323303,
      "learning_rate": 0.00011305261922200519,
      "loss": 1.563,
      "step": 18480
    },
    {
      "epoch": 4.585529171058342,
      "grad_norm": 0.9251183867454529,
      "learning_rate": 0.000112975365369575,
      "loss": 1.5971,
      "step": 18490
    },
    {
      "epoch": 4.588009176018352,
      "grad_norm": 0.969687283039093,
      "learning_rate": 0.00011289810363982875,
      "loss": 1.6287,
      "step": 18500
    },
    {
      "epoch": 4.590489180978362,
      "grad_norm": 0.8724190592765808,
      "learning_rate": 0.00011282083407967183,
      "loss": 1.5623,
      "step": 18510
    },
    {
      "epoch": 4.592969185938372,
      "grad_norm": 0.8921563625335693,
      "learning_rate": 0.00011274355673601444,
      "loss": 1.5819,
      "step": 18520
    },
    {
      "epoch": 4.595449190898382,
      "grad_norm": 1.0023810863494873,
      "learning_rate": 0.00011266627165577148,
      "loss": 1.5978,
      "step": 18530
    },
    {
      "epoch": 4.597929195858391,
      "grad_norm": 0.9489151835441589,
      "learning_rate": 0.00011258897888586255,
      "loss": 1.5348,
      "step": 18540
    },
    {
      "epoch": 4.600409200818402,
      "grad_norm": 0.9250749945640564,
      "learning_rate": 0.00011251167847321193,
      "loss": 1.5839,
      "step": 18550
    },
    {
      "epoch": 4.602889205778411,
      "grad_norm": 0.9519440531730652,
      "learning_rate": 0.00011243437046474853,
      "loss": 1.6104,
      "step": 18560
    },
    {
      "epoch": 4.605369210738422,
      "grad_norm": 0.9788894653320312,
      "learning_rate": 0.00011235705490740589,
      "loss": 1.5947,
      "step": 18570
    },
    {
      "epoch": 4.607849215698431,
      "grad_norm": 0.8717751502990723,
      "learning_rate": 0.00011227973184812206,
      "loss": 1.5585,
      "step": 18580
    },
    {
      "epoch": 4.610329220658441,
      "grad_norm": 0.9002294540405273,
      "learning_rate": 0.00011220240133383974,
      "loss": 1.5863,
      "step": 18590
    },
    {
      "epoch": 4.612809225618451,
      "grad_norm": 0.9561557769775391,
      "learning_rate": 0.00011212506341150615,
      "loss": 1.6116,
      "step": 18600
    },
    {
      "epoch": 4.615289230578461,
      "grad_norm": 0.8865216374397278,
      "learning_rate": 0.00011204771812807295,
      "loss": 1.6194,
      "step": 18610
    },
    {
      "epoch": 4.617769235538471,
      "grad_norm": 0.9733191728591919,
      "learning_rate": 0.00011197036553049625,
      "loss": 1.6004,
      "step": 18620
    },
    {
      "epoch": 4.620249240498481,
      "grad_norm": 0.9483121633529663,
      "learning_rate": 0.0001118930056657367,
      "loss": 1.6101,
      "step": 18630
    },
    {
      "epoch": 4.6227292454584905,
      "grad_norm": 0.8942649364471436,
      "learning_rate": 0.0001118156385807593,
      "loss": 1.6269,
      "step": 18640
    },
    {
      "epoch": 4.625209250418501,
      "grad_norm": 0.9124876856803894,
      "learning_rate": 0.00011173826432253338,
      "loss": 1.6196,
      "step": 18650
    },
    {
      "epoch": 4.6276892553785105,
      "grad_norm": 0.9813550114631653,
      "learning_rate": 0.00011166088293803276,
      "loss": 1.6095,
      "step": 18660
    },
    {
      "epoch": 4.630169260338521,
      "grad_norm": 0.8427887558937073,
      "learning_rate": 0.00011158349447423549,
      "loss": 1.6013,
      "step": 18670
    },
    {
      "epoch": 4.6326492652985305,
      "grad_norm": 0.9709888696670532,
      "learning_rate": 0.00011150609897812387,
      "loss": 1.5761,
      "step": 18680
    },
    {
      "epoch": 4.63512927025854,
      "grad_norm": 0.9658144116401672,
      "learning_rate": 0.00011142869649668466,
      "loss": 1.5706,
      "step": 18690
    },
    {
      "epoch": 4.6376092752185505,
      "grad_norm": 0.920049786567688,
      "learning_rate": 0.0001113512870769086,
      "loss": 1.5534,
      "step": 18700
    },
    {
      "epoch": 4.64008928017856,
      "grad_norm": 0.9348146319389343,
      "learning_rate": 0.00011127387076579092,
      "loss": 1.6135,
      "step": 18710
    },
    {
      "epoch": 4.6425692851385705,
      "grad_norm": 1.042134404182434,
      "learning_rate": 0.00011119644761033078,
      "loss": 1.5355,
      "step": 18720
    },
    {
      "epoch": 4.64504929009858,
      "grad_norm": 0.9624781012535095,
      "learning_rate": 0.00011111901765753168,
      "loss": 1.5915,
      "step": 18730
    },
    {
      "epoch": 4.6475292950585905,
      "grad_norm": 0.9615873098373413,
      "learning_rate": 0.00011104158095440113,
      "loss": 1.5729,
      "step": 18740
    },
    {
      "epoch": 4.6500093000186,
      "grad_norm": 0.9324778318405151,
      "learning_rate": 0.00011096413754795083,
      "loss": 1.606,
      "step": 18750
    },
    {
      "epoch": 4.65248930497861,
      "grad_norm": 0.921046793460846,
      "learning_rate": 0.00011088668748519647,
      "loss": 1.6428,
      "step": 18760
    },
    {
      "epoch": 4.65496930993862,
      "grad_norm": 1.008867859840393,
      "learning_rate": 0.00011080923081315781,
      "loss": 1.6306,
      "step": 18770
    },
    {
      "epoch": 4.65744931489863,
      "grad_norm": 0.8651713132858276,
      "learning_rate": 0.00011073176757885866,
      "loss": 1.599,
      "step": 18780
    },
    {
      "epoch": 4.65992931985864,
      "grad_norm": 0.9174283742904663,
      "learning_rate": 0.00011065429782932673,
      "loss": 1.5038,
      "step": 18790
    },
    {
      "epoch": 4.66240932481865,
      "grad_norm": 0.8761575818061829,
      "learning_rate": 0.00011057682161159379,
      "loss": 1.5958,
      "step": 18800
    },
    {
      "epoch": 4.66488932977866,
      "grad_norm": 0.9216639399528503,
      "learning_rate": 0.00011049933897269547,
      "loss": 1.5961,
      "step": 18810
    },
    {
      "epoch": 4.6673693347386696,
      "grad_norm": 0.9265581965446472,
      "learning_rate": 0.00011042184995967127,
      "loss": 1.6,
      "step": 18820
    },
    {
      "epoch": 4.669849339698679,
      "grad_norm": 0.9563526511192322,
      "learning_rate": 0.00011034435461956466,
      "loss": 1.5949,
      "step": 18830
    },
    {
      "epoch": 4.6723293446586895,
      "grad_norm": 0.8865848183631897,
      "learning_rate": 0.00011026685299942285,
      "loss": 1.6383,
      "step": 18840
    },
    {
      "epoch": 4.674809349618699,
      "grad_norm": 0.955040693283081,
      "learning_rate": 0.00011018934514629693,
      "loss": 1.5657,
      "step": 18850
    },
    {
      "epoch": 4.6772893545787095,
      "grad_norm": 0.9501464366912842,
      "learning_rate": 0.00011011183110724172,
      "loss": 1.6361,
      "step": 18860
    },
    {
      "epoch": 4.679769359538719,
      "grad_norm": 0.9693083167076111,
      "learning_rate": 0.00011003431092931589,
      "loss": 1.5893,
      "step": 18870
    },
    {
      "epoch": 4.682249364498729,
      "grad_norm": 0.893928587436676,
      "learning_rate": 0.00010995678465958168,
      "loss": 1.5705,
      "step": 18880
    },
    {
      "epoch": 4.684729369458739,
      "grad_norm": 0.9571881890296936,
      "learning_rate": 0.00010987925234510519,
      "loss": 1.5811,
      "step": 18890
    },
    {
      "epoch": 4.687209374418749,
      "grad_norm": 0.9786219596862793,
      "learning_rate": 0.0001098017140329561,
      "loss": 1.5305,
      "step": 18900
    },
    {
      "epoch": 4.689689379378759,
      "grad_norm": 0.972812294960022,
      "learning_rate": 0.0001097241697702077,
      "loss": 1.5485,
      "step": 18910
    },
    {
      "epoch": 4.692169384338769,
      "grad_norm": 0.929424524307251,
      "learning_rate": 0.00010964661960393703,
      "loss": 1.57,
      "step": 18920
    },
    {
      "epoch": 4.694649389298778,
      "grad_norm": 0.9796428084373474,
      "learning_rate": 0.00010956906358122456,
      "loss": 1.5994,
      "step": 18930
    },
    {
      "epoch": 4.697129394258789,
      "grad_norm": 0.9514657258987427,
      "learning_rate": 0.00010949150174915442,
      "loss": 1.6026,
      "step": 18940
    },
    {
      "epoch": 4.699609399218798,
      "grad_norm": 0.9882819056510925,
      "learning_rate": 0.00010941393415481423,
      "loss": 1.6254,
      "step": 18950
    },
    {
      "epoch": 4.702089404178809,
      "grad_norm": 0.8934924006462097,
      "learning_rate": 0.00010933636084529506,
      "loss": 1.5688,
      "step": 18960
    },
    {
      "epoch": 4.704569409138818,
      "grad_norm": 0.904549241065979,
      "learning_rate": 0.00010925878186769158,
      "loss": 1.5813,
      "step": 18970
    },
    {
      "epoch": 4.707049414098828,
      "grad_norm": 0.9113224744796753,
      "learning_rate": 0.00010918119726910174,
      "loss": 1.6022,
      "step": 18980
    },
    {
      "epoch": 4.709529419058838,
      "grad_norm": 0.9225690364837646,
      "learning_rate": 0.00010910360709662701,
      "loss": 1.5954,
      "step": 18990
    },
    {
      "epoch": 4.712009424018848,
      "grad_norm": 0.8976655602455139,
      "learning_rate": 0.00010902601139737225,
      "loss": 1.5457,
      "step": 19000
    },
    {
      "epoch": 4.714489428978858,
      "grad_norm": 0.9754193425178528,
      "learning_rate": 0.0001089484102184456,
      "loss": 1.5562,
      "step": 19010
    },
    {
      "epoch": 4.716969433938868,
      "grad_norm": 0.9099609851837158,
      "learning_rate": 0.00010887080360695854,
      "loss": 1.5565,
      "step": 19020
    },
    {
      "epoch": 4.719449438898878,
      "grad_norm": 0.9398124814033508,
      "learning_rate": 0.00010879319161002594,
      "loss": 1.611,
      "step": 19030
    },
    {
      "epoch": 4.721929443858888,
      "grad_norm": 0.9173715710639954,
      "learning_rate": 0.00010871557427476583,
      "loss": 1.5626,
      "step": 19040
    },
    {
      "epoch": 4.724409448818898,
      "grad_norm": 0.9978727698326111,
      "learning_rate": 0.00010863795164829955,
      "loss": 1.6209,
      "step": 19050
    },
    {
      "epoch": 4.726889453778908,
      "grad_norm": 0.9299024939537048,
      "learning_rate": 0.0001085603237777516,
      "loss": 1.5473,
      "step": 19060
    },
    {
      "epoch": 4.729369458738917,
      "grad_norm": 0.938007116317749,
      "learning_rate": 0.00010848269071024975,
      "loss": 1.6074,
      "step": 19070
    },
    {
      "epoch": 4.731849463698928,
      "grad_norm": 0.9142996072769165,
      "learning_rate": 0.00010840505249292476,
      "loss": 1.5746,
      "step": 19080
    },
    {
      "epoch": 4.734329468658937,
      "grad_norm": 0.8934489488601685,
      "learning_rate": 0.00010832740917291073,
      "loss": 1.6187,
      "step": 19090
    },
    {
      "epoch": 4.736809473618948,
      "grad_norm": 0.9229673743247986,
      "learning_rate": 0.00010824976079734472,
      "loss": 1.602,
      "step": 19100
    },
    {
      "epoch": 4.739289478578957,
      "grad_norm": 0.90119469165802,
      "learning_rate": 0.00010817210741336684,
      "loss": 1.6417,
      "step": 19110
    },
    {
      "epoch": 4.741769483538967,
      "grad_norm": 0.8615995645523071,
      "learning_rate": 0.00010809444906812033,
      "loss": 1.6084,
      "step": 19120
    },
    {
      "epoch": 4.744249488498977,
      "grad_norm": 0.9701696634292603,
      "learning_rate": 0.00010801678580875143,
      "loss": 1.5592,
      "step": 19130
    },
    {
      "epoch": 4.746729493458987,
      "grad_norm": 0.977692723274231,
      "learning_rate": 0.00010793911768240929,
      "loss": 1.5653,
      "step": 19140
    },
    {
      "epoch": 4.749209498418997,
      "grad_norm": 0.8544299006462097,
      "learning_rate": 0.00010786144473624608,
      "loss": 1.5722,
      "step": 19150
    },
    {
      "epoch": 4.751689503379007,
      "grad_norm": 0.9128935933113098,
      "learning_rate": 0.00010778376701741689,
      "loss": 1.5968,
      "step": 19160
    },
    {
      "epoch": 4.754169508339016,
      "grad_norm": 0.9116058349609375,
      "learning_rate": 0.00010770608457307965,
      "loss": 1.6172,
      "step": 19170
    },
    {
      "epoch": 4.756649513299027,
      "grad_norm": 0.9581488966941833,
      "learning_rate": 0.00010762839745039526,
      "loss": 1.5649,
      "step": 19180
    },
    {
      "epoch": 4.759129518259036,
      "grad_norm": 1.0204472541809082,
      "learning_rate": 0.00010755070569652735,
      "loss": 1.532,
      "step": 19190
    },
    {
      "epoch": 4.761609523219047,
      "grad_norm": 0.9427098035812378,
      "learning_rate": 0.00010747300935864243,
      "loss": 1.6342,
      "step": 19200
    },
    {
      "epoch": 4.764089528179056,
      "grad_norm": 0.8668059706687927,
      "learning_rate": 0.00010739530848390981,
      "loss": 1.5539,
      "step": 19210
    },
    {
      "epoch": 4.766569533139066,
      "grad_norm": 0.9295413494110107,
      "learning_rate": 0.0001073176031195015,
      "loss": 1.5857,
      "step": 19220
    },
    {
      "epoch": 4.769049538099076,
      "grad_norm": 0.92892986536026,
      "learning_rate": 0.00010723989331259223,
      "loss": 1.6505,
      "step": 19230
    },
    {
      "epoch": 4.771529543059086,
      "grad_norm": 0.9731125235557556,
      "learning_rate": 0.00010716217911035951,
      "loss": 1.6123,
      "step": 19240
    },
    {
      "epoch": 4.774009548019096,
      "grad_norm": 0.9232926964759827,
      "learning_rate": 0.00010708446055998343,
      "loss": 1.6521,
      "step": 19250
    },
    {
      "epoch": 4.776489552979106,
      "grad_norm": 0.9706169366836548,
      "learning_rate": 0.00010700673770864673,
      "loss": 1.6323,
      "step": 19260
    },
    {
      "epoch": 4.778969557939116,
      "grad_norm": 0.9299634099006653,
      "learning_rate": 0.00010692901060353483,
      "loss": 1.5784,
      "step": 19270
    },
    {
      "epoch": 4.781449562899126,
      "grad_norm": 0.8783867955207825,
      "learning_rate": 0.00010685127929183567,
      "loss": 1.6084,
      "step": 19280
    },
    {
      "epoch": 4.783929567859135,
      "grad_norm": 0.8926278948783875,
      "learning_rate": 0.00010677354382073976,
      "loss": 1.5303,
      "step": 19290
    },
    {
      "epoch": 4.786409572819146,
      "grad_norm": 0.9633662700653076,
      "learning_rate": 0.00010669580423744015,
      "loss": 1.5834,
      "step": 19300
    },
    {
      "epoch": 4.788889577779155,
      "grad_norm": 0.8755930066108704,
      "learning_rate": 0.0001066180605891323,
      "loss": 1.5767,
      "step": 19310
    },
    {
      "epoch": 4.791369582739166,
      "grad_norm": 0.912855327129364,
      "learning_rate": 0.00010654031292301432,
      "loss": 1.5674,
      "step": 19320
    },
    {
      "epoch": 4.793849587699175,
      "grad_norm": 0.9317062497138977,
      "learning_rate": 0.00010646256128628657,
      "loss": 1.5992,
      "step": 19330
    },
    {
      "epoch": 4.796329592659186,
      "grad_norm": 0.9335405230522156,
      "learning_rate": 0.0001063848057261519,
      "loss": 1.576,
      "step": 19340
    },
    {
      "epoch": 4.798809597619195,
      "grad_norm": 0.9386693239212036,
      "learning_rate": 0.0001063070462898156,
      "loss": 1.6253,
      "step": 19350
    },
    {
      "epoch": 4.801289602579205,
      "grad_norm": 0.9020708203315735,
      "learning_rate": 0.00010622928302448523,
      "loss": 1.5704,
      "step": 19360
    },
    {
      "epoch": 4.803769607539215,
      "grad_norm": 0.9368181228637695,
      "learning_rate": 0.00010615151597737065,
      "loss": 1.6198,
      "step": 19370
    },
    {
      "epoch": 4.806249612499225,
      "grad_norm": 0.8870874643325806,
      "learning_rate": 0.00010607374519568412,
      "loss": 1.6187,
      "step": 19380
    },
    {
      "epoch": 4.808729617459235,
      "grad_norm": 0.8933711647987366,
      "learning_rate": 0.00010599597072664012,
      "loss": 1.5932,
      "step": 19390
    },
    {
      "epoch": 4.811209622419245,
      "grad_norm": 0.9829477667808533,
      "learning_rate": 0.00010591819261745528,
      "loss": 1.6349,
      "step": 19400
    },
    {
      "epoch": 4.813689627379254,
      "grad_norm": 0.9438134431838989,
      "learning_rate": 0.00010584041091534863,
      "loss": 1.5833,
      "step": 19410
    },
    {
      "epoch": 4.816169632339265,
      "grad_norm": 0.9365798234939575,
      "learning_rate": 0.00010576262566754121,
      "loss": 1.5672,
      "step": 19420
    },
    {
      "epoch": 4.818649637299274,
      "grad_norm": 0.8850331902503967,
      "learning_rate": 0.00010568483692125624,
      "loss": 1.5688,
      "step": 19430
    },
    {
      "epoch": 4.821129642259285,
      "grad_norm": 0.988837480545044,
      "learning_rate": 0.00010560704472371919,
      "loss": 1.5419,
      "step": 19440
    },
    {
      "epoch": 4.823609647219294,
      "grad_norm": 0.9462050795555115,
      "learning_rate": 0.00010552924912215747,
      "loss": 1.5893,
      "step": 19450
    },
    {
      "epoch": 4.826089652179304,
      "grad_norm": 0.9274313449859619,
      "learning_rate": 0.00010545145016380065,
      "loss": 1.5462,
      "step": 19460
    },
    {
      "epoch": 4.828569657139314,
      "grad_norm": 0.8957258462905884,
      "learning_rate": 0.00010537364789588029,
      "loss": 1.6319,
      "step": 19470
    },
    {
      "epoch": 4.831049662099324,
      "grad_norm": 0.9592501521110535,
      "learning_rate": 0.00010529584236562995,
      "loss": 1.5837,
      "step": 19480
    },
    {
      "epoch": 4.833529667059334,
      "grad_norm": 0.8686869144439697,
      "learning_rate": 0.00010521803362028529,
      "loss": 1.5705,
      "step": 19490
    },
    {
      "epoch": 4.836009672019344,
      "grad_norm": 0.8717570304870605,
      "learning_rate": 0.00010514022170708374,
      "loss": 1.606,
      "step": 19500
    },
    {
      "epoch": 4.8384896769793535,
      "grad_norm": 0.9135420322418213,
      "learning_rate": 0.00010506240667326477,
      "loss": 1.633,
      "step": 19510
    },
    {
      "epoch": 4.840969681939364,
      "grad_norm": 0.9622677564620972,
      "learning_rate": 0.00010498458856606972,
      "loss": 1.6093,
      "step": 19520
    },
    {
      "epoch": 4.8434496868993735,
      "grad_norm": 0.9665126204490662,
      "learning_rate": 0.00010490676743274181,
      "loss": 1.5636,
      "step": 19530
    },
    {
      "epoch": 4.845929691859384,
      "grad_norm": 0.971638023853302,
      "learning_rate": 0.00010482894332052607,
      "loss": 1.631,
      "step": 19540
    },
    {
      "epoch": 4.8484096968193935,
      "grad_norm": 0.956616997718811,
      "learning_rate": 0.00010475111627666934,
      "loss": 1.61,
      "step": 19550
    },
    {
      "epoch": 4.850889701779404,
      "grad_norm": 0.9185702800750732,
      "learning_rate": 0.00010467328634842024,
      "loss": 1.5968,
      "step": 19560
    },
    {
      "epoch": 4.8533697067394135,
      "grad_norm": 0.910837709903717,
      "learning_rate": 0.00010459545358302918,
      "loss": 1.5601,
      "step": 19570
    },
    {
      "epoch": 4.855849711699423,
      "grad_norm": 0.885400116443634,
      "learning_rate": 0.00010451761802774824,
      "loss": 1.6183,
      "step": 19580
    },
    {
      "epoch": 4.8583297166594335,
      "grad_norm": 0.893608808517456,
      "learning_rate": 0.00010443977972983126,
      "loss": 1.5548,
      "step": 19590
    },
    {
      "epoch": 4.860809721619443,
      "grad_norm": 0.8942586779594421,
      "learning_rate": 0.00010436193873653361,
      "loss": 1.6634,
      "step": 19600
    },
    {
      "epoch": 4.8632897265794535,
      "grad_norm": 0.9296571016311646,
      "learning_rate": 0.00010428409509511248,
      "loss": 1.5669,
      "step": 19610
    },
    {
      "epoch": 4.865769731539463,
      "grad_norm": 0.946972668170929,
      "learning_rate": 0.00010420624885282653,
      "loss": 1.5418,
      "step": 19620
    },
    {
      "epoch": 4.8682497364994735,
      "grad_norm": 0.8497663140296936,
      "learning_rate": 0.00010412840005693603,
      "loss": 1.5891,
      "step": 19630
    },
    {
      "epoch": 4.870729741459483,
      "grad_norm": 0.8625304102897644,
      "learning_rate": 0.00010405054875470286,
      "loss": 1.5861,
      "step": 19640
    },
    {
      "epoch": 4.873209746419493,
      "grad_norm": 0.9989964962005615,
      "learning_rate": 0.00010397269499339035,
      "loss": 1.6085,
      "step": 19650
    },
    {
      "epoch": 4.875689751379503,
      "grad_norm": 0.9507265090942383,
      "learning_rate": 0.00010389483882026334,
      "loss": 1.573,
      "step": 19660
    },
    {
      "epoch": 4.8781697563395126,
      "grad_norm": 0.9283966422080994,
      "learning_rate": 0.00010381698028258817,
      "loss": 1.5744,
      "step": 19670
    },
    {
      "epoch": 4.880649761299523,
      "grad_norm": 0.9498452544212341,
      "learning_rate": 0.0001037391194276326,
      "loss": 1.63,
      "step": 19680
    },
    {
      "epoch": 4.8831297662595325,
      "grad_norm": 0.8902698159217834,
      "learning_rate": 0.0001036612563026657,
      "loss": 1.604,
      "step": 19690
    },
    {
      "epoch": 4.885609771219542,
      "grad_norm": 0.9303712248802185,
      "learning_rate": 0.0001035833909549581,
      "loss": 1.569,
      "step": 19700
    },
    {
      "epoch": 4.8880897761795525,
      "grad_norm": 0.9185006618499756,
      "learning_rate": 0.00010350552343178163,
      "loss": 1.5671,
      "step": 19710
    },
    {
      "epoch": 4.890569781139562,
      "grad_norm": 0.9281051754951477,
      "learning_rate": 0.00010342765378040953,
      "loss": 1.6518,
      "step": 19720
    },
    {
      "epoch": 4.8930497860995725,
      "grad_norm": 0.894377589225769,
      "learning_rate": 0.00010334978204811629,
      "loss": 1.5639,
      "step": 19730
    },
    {
      "epoch": 4.895529791059582,
      "grad_norm": 0.9298607110977173,
      "learning_rate": 0.00010327190828217763,
      "loss": 1.5616,
      "step": 19740
    },
    {
      "epoch": 4.898009796019592,
      "grad_norm": 0.9328896999359131,
      "learning_rate": 0.00010319403252987059,
      "loss": 1.5928,
      "step": 19750
    },
    {
      "epoch": 4.900489800979602,
      "grad_norm": 0.8884739875793457,
      "learning_rate": 0.00010311615483847332,
      "loss": 1.5494,
      "step": 19760
    },
    {
      "epoch": 4.902969805939612,
      "grad_norm": 0.9351387619972229,
      "learning_rate": 0.00010303827525526523,
      "loss": 1.5713,
      "step": 19770
    },
    {
      "epoch": 4.905449810899622,
      "grad_norm": 0.9865672588348389,
      "learning_rate": 0.00010296039382752687,
      "loss": 1.6145,
      "step": 19780
    },
    {
      "epoch": 4.907929815859632,
      "grad_norm": 0.9271677732467651,
      "learning_rate": 0.00010288251060253986,
      "loss": 1.5772,
      "step": 19790
    },
    {
      "epoch": 4.910409820819641,
      "grad_norm": 0.911442220211029,
      "learning_rate": 0.0001028046256275869,
      "loss": 1.5749,
      "step": 19800
    },
    {
      "epoch": 4.912889825779652,
      "grad_norm": 0.9066318869590759,
      "learning_rate": 0.00010272673894995187,
      "loss": 1.5969,
      "step": 19810
    },
    {
      "epoch": 4.915369830739661,
      "grad_norm": 1.0287082195281982,
      "learning_rate": 0.00010264885061691954,
      "loss": 1.4949,
      "step": 19820
    },
    {
      "epoch": 4.917849835699672,
      "grad_norm": 0.9573404788970947,
      "learning_rate": 0.00010257096067577572,
      "loss": 1.6417,
      "step": 19830
    },
    {
      "epoch": 4.920329840659681,
      "grad_norm": 0.9665036201477051,
      "learning_rate": 0.0001024930691738073,
      "loss": 1.5804,
      "step": 19840
    },
    {
      "epoch": 4.922809845619692,
      "grad_norm": 0.9561458826065063,
      "learning_rate": 0.000102415176158302,
      "loss": 1.5897,
      "step": 19850
    },
    {
      "epoch": 4.925289850579701,
      "grad_norm": 0.9099255800247192,
      "learning_rate": 0.0001023372816765485,
      "loss": 1.549,
      "step": 19860
    },
    {
      "epoch": 4.927769855539711,
      "grad_norm": 0.9153082966804504,
      "learning_rate": 0.0001022593857758364,
      "loss": 1.5885,
      "step": 19870
    },
    {
      "epoch": 4.930249860499721,
      "grad_norm": 0.8561480045318604,
      "learning_rate": 0.00010218148850345613,
      "loss": 1.5734,
      "step": 19880
    },
    {
      "epoch": 4.932729865459731,
      "grad_norm": 0.9554060101509094,
      "learning_rate": 0.00010210358990669888,
      "loss": 1.6141,
      "step": 19890
    },
    {
      "epoch": 4.935209870419741,
      "grad_norm": 0.941474199295044,
      "learning_rate": 0.00010202569003285682,
      "loss": 1.5803,
      "step": 19900
    },
    {
      "epoch": 4.937689875379751,
      "grad_norm": 0.8871884346008301,
      "learning_rate": 0.00010194778892922274,
      "loss": 1.5363,
      "step": 19910
    },
    {
      "epoch": 4.940169880339761,
      "grad_norm": 0.9098392128944397,
      "learning_rate": 0.00010186988664309023,
      "loss": 1.6852,
      "step": 19920
    },
    {
      "epoch": 4.942649885299771,
      "grad_norm": 0.9218733310699463,
      "learning_rate": 0.00010179198322175361,
      "loss": 1.5865,
      "step": 19930
    },
    {
      "epoch": 4.94512989025978,
      "grad_norm": 0.9652133584022522,
      "learning_rate": 0.00010171407871250789,
      "loss": 1.6027,
      "step": 19940
    },
    {
      "epoch": 4.947609895219791,
      "grad_norm": 0.8366691470146179,
      "learning_rate": 0.00010163617316264869,
      "loss": 1.5675,
      "step": 19950
    },
    {
      "epoch": 4.9500899001798,
      "grad_norm": 0.9406192898750305,
      "learning_rate": 0.00010155826661947231,
      "loss": 1.6375,
      "step": 19960
    },
    {
      "epoch": 4.952569905139811,
      "grad_norm": 0.9226821064949036,
      "learning_rate": 0.00010148035913027569,
      "loss": 1.6013,
      "step": 19970
    },
    {
      "epoch": 4.95504991009982,
      "grad_norm": 0.9119101762771606,
      "learning_rate": 0.00010140245074235624,
      "loss": 1.6033,
      "step": 19980
    },
    {
      "epoch": 4.95752991505983,
      "grad_norm": 0.9745951890945435,
      "learning_rate": 0.00010132454150301201,
      "loss": 1.566,
      "step": 19990
    },
    {
      "epoch": 4.96000992001984,
      "grad_norm": 0.8798539638519287,
      "learning_rate": 0.00010124663145954152,
      "loss": 1.5833,
      "step": 20000
    },
    {
      "epoch": 4.96248992497985,
      "grad_norm": 0.9379274249076843,
      "learning_rate": 0.00010116872065924376,
      "loss": 1.5883,
      "step": 20010
    },
    {
      "epoch": 4.96496992993986,
      "grad_norm": 0.9305238127708435,
      "learning_rate": 0.00010109080914941824,
      "loss": 1.6406,
      "step": 20020
    },
    {
      "epoch": 4.96744993489987,
      "grad_norm": 0.9678001999855042,
      "learning_rate": 0.00010101289697736486,
      "loss": 1.5633,
      "step": 20030
    },
    {
      "epoch": 4.969929939859879,
      "grad_norm": 1.002220630645752,
      "learning_rate": 0.00010093498419038394,
      "loss": 1.5972,
      "step": 20040
    },
    {
      "epoch": 4.97240994481989,
      "grad_norm": 0.9045056700706482,
      "learning_rate": 0.00010085707083577613,
      "loss": 1.5854,
      "step": 20050
    },
    {
      "epoch": 4.974889949779899,
      "grad_norm": 0.8465107679367065,
      "learning_rate": 0.00010077915696084248,
      "loss": 1.6154,
      "step": 20060
    },
    {
      "epoch": 4.97736995473991,
      "grad_norm": 0.8597650527954102,
      "learning_rate": 0.00010070124261288436,
      "loss": 1.5694,
      "step": 20070
    },
    {
      "epoch": 4.979849959699919,
      "grad_norm": 0.9025321006774902,
      "learning_rate": 0.00010062332783920336,
      "loss": 1.6415,
      "step": 20080
    },
    {
      "epoch": 4.982329964659929,
      "grad_norm": 0.8586678504943848,
      "learning_rate": 0.00010054541268710138,
      "loss": 1.5581,
      "step": 20090
    },
    {
      "epoch": 4.984809969619939,
      "grad_norm": 0.866348147392273,
      "learning_rate": 0.00010046749720388055,
      "loss": 1.6083,
      "step": 20100
    },
    {
      "epoch": 4.987289974579949,
      "grad_norm": 0.9521636962890625,
      "learning_rate": 0.0001003895814368432,
      "loss": 1.5893,
      "step": 20110
    },
    {
      "epoch": 4.989769979539959,
      "grad_norm": 0.9525920748710632,
      "learning_rate": 0.00010031166543329178,
      "loss": 1.5775,
      "step": 20120
    },
    {
      "epoch": 4.992249984499969,
      "grad_norm": 0.9372658133506775,
      "learning_rate": 0.00010023374924052897,
      "loss": 1.6016,
      "step": 20130
    },
    {
      "epoch": 4.994729989459979,
      "grad_norm": 0.9438722729682922,
      "learning_rate": 0.0001001558329058575,
      "loss": 1.5807,
      "step": 20140
    },
    {
      "epoch": 4.997209994419989,
      "grad_norm": 0.9148738980293274,
      "learning_rate": 0.00010007791647658022,
      "loss": 1.66,
      "step": 20150
    },
    {
      "epoch": 4.999689999379999,
      "grad_norm": 0.9490894675254822,
      "learning_rate": 0.0001,
      "loss": 1.5962,
      "step": 20160
    },
    {
      "epoch": 5.002170004340009,
      "grad_norm": 0.9742351770401001,
      "learning_rate": 9.992208352341978e-05,
      "loss": 1.5214,
      "step": 20170
    },
    {
      "epoch": 5.004650009300018,
      "grad_norm": 0.9827502369880676,
      "learning_rate": 9.98441670941425e-05,
      "loss": 1.4653,
      "step": 20180
    },
    {
      "epoch": 5.007130014260029,
      "grad_norm": 0.9274985790252686,
      "learning_rate": 9.976625075947103e-05,
      "loss": 1.5592,
      "step": 20190
    },
    {
      "epoch": 5.009610019220038,
      "grad_norm": 0.9367074370384216,
      "learning_rate": 9.968833456670824e-05,
      "loss": 1.5098,
      "step": 20200
    },
    {
      "epoch": 5.012090024180049,
      "grad_norm": 0.9240421056747437,
      "learning_rate": 9.96104185631568e-05,
      "loss": 1.5163,
      "step": 20210
    },
    {
      "epoch": 5.014570029140058,
      "grad_norm": 0.9595016241073608,
      "learning_rate": 9.953250279611946e-05,
      "loss": 1.4962,
      "step": 20220
    },
    {
      "epoch": 5.017050034100068,
      "grad_norm": 0.9109339118003845,
      "learning_rate": 9.945458731289862e-05,
      "loss": 1.526,
      "step": 20230
    },
    {
      "epoch": 5.019530039060078,
      "grad_norm": 0.8929713368415833,
      "learning_rate": 9.937667216079665e-05,
      "loss": 1.5212,
      "step": 20240
    },
    {
      "epoch": 5.022010044020088,
      "grad_norm": 0.9416952133178711,
      "learning_rate": 9.929875738711564e-05,
      "loss": 1.5036,
      "step": 20250
    },
    {
      "epoch": 5.024490048980098,
      "grad_norm": 0.9692783951759338,
      "learning_rate": 9.922084303915753e-05,
      "loss": 1.4676,
      "step": 20260
    },
    {
      "epoch": 5.026970053940108,
      "grad_norm": 1.0281188488006592,
      "learning_rate": 9.914292916422387e-05,
      "loss": 1.5252,
      "step": 20270
    },
    {
      "epoch": 5.029450058900117,
      "grad_norm": 0.9119085669517517,
      "learning_rate": 9.90650158096161e-05,
      "loss": 1.5494,
      "step": 20280
    },
    {
      "epoch": 5.031930063860128,
      "grad_norm": 0.8939598798751831,
      "learning_rate": 9.898710302263515e-05,
      "loss": 1.4699,
      "step": 20290
    },
    {
      "epoch": 5.034410068820137,
      "grad_norm": 0.8867928981781006,
      "learning_rate": 9.890919085058178e-05,
      "loss": 1.4507,
      "step": 20300
    },
    {
      "epoch": 5.036890073780148,
      "grad_norm": 0.9023534655570984,
      "learning_rate": 9.883127934075624e-05,
      "loss": 1.487,
      "step": 20310
    },
    {
      "epoch": 5.039370078740157,
      "grad_norm": 0.8967320322990417,
      "learning_rate": 9.875336854045851e-05,
      "loss": 1.5008,
      "step": 20320
    },
    {
      "epoch": 5.041850083700167,
      "grad_norm": 0.9554635882377625,
      "learning_rate": 9.867545849698799e-05,
      "loss": 1.5245,
      "step": 20330
    },
    {
      "epoch": 5.044330088660177,
      "grad_norm": 0.9490965604782104,
      "learning_rate": 9.859754925764378e-05,
      "loss": 1.4932,
      "step": 20340
    },
    {
      "epoch": 5.046810093620187,
      "grad_norm": 0.9715158939361572,
      "learning_rate": 9.851964086972432e-05,
      "loss": 1.4802,
      "step": 20350
    },
    {
      "epoch": 5.049290098580197,
      "grad_norm": 1.0020184516906738,
      "learning_rate": 9.84417333805277e-05,
      "loss": 1.5061,
      "step": 20360
    },
    {
      "epoch": 5.051770103540207,
      "grad_norm": 0.9794074892997742,
      "learning_rate": 9.836382683735132e-05,
      "loss": 1.4768,
      "step": 20370
    },
    {
      "epoch": 5.054250108500217,
      "grad_norm": 0.9327630400657654,
      "learning_rate": 9.828592128749215e-05,
      "loss": 1.5129,
      "step": 20380
    },
    {
      "epoch": 5.056730113460227,
      "grad_norm": 0.9747712016105652,
      "learning_rate": 9.820801677824639e-05,
      "loss": 1.5984,
      "step": 20390
    },
    {
      "epoch": 5.0592101184202365,
      "grad_norm": 0.9542945623397827,
      "learning_rate": 9.813011335690981e-05,
      "loss": 1.521,
      "step": 20400
    },
    {
      "epoch": 5.061690123380247,
      "grad_norm": 0.933824896812439,
      "learning_rate": 9.805221107077726e-05,
      "loss": 1.4929,
      "step": 20410
    },
    {
      "epoch": 5.0641701283402565,
      "grad_norm": 0.9382598996162415,
      "learning_rate": 9.79743099671432e-05,
      "loss": 1.54,
      "step": 20420
    },
    {
      "epoch": 5.066650133300267,
      "grad_norm": 0.9482308030128479,
      "learning_rate": 9.789641009330111e-05,
      "loss": 1.53,
      "step": 20430
    },
    {
      "epoch": 5.0691301382602765,
      "grad_norm": 0.9552004337310791,
      "learning_rate": 9.78185114965439e-05,
      "loss": 1.4699,
      "step": 20440
    },
    {
      "epoch": 5.071610143220286,
      "grad_norm": 0.9345937967300415,
      "learning_rate": 9.774061422416359e-05,
      "loss": 1.5226,
      "step": 20450
    },
    {
      "epoch": 5.0740901481802965,
      "grad_norm": 0.878754734992981,
      "learning_rate": 9.766271832345152e-05,
      "loss": 1.546,
      "step": 20460
    },
    {
      "epoch": 5.076570153140306,
      "grad_norm": 0.9064426422119141,
      "learning_rate": 9.7584823841698e-05,
      "loss": 1.4589,
      "step": 20470
    },
    {
      "epoch": 5.0790501581003165,
      "grad_norm": 0.9096664786338806,
      "learning_rate": 9.750693082619273e-05,
      "loss": 1.4949,
      "step": 20480
    },
    {
      "epoch": 5.081530163060326,
      "grad_norm": 0.9073386192321777,
      "learning_rate": 9.74290393242243e-05,
      "loss": 1.4912,
      "step": 20490
    },
    {
      "epoch": 5.0840101680203365,
      "grad_norm": 0.982649564743042,
      "learning_rate": 9.735114938308051e-05,
      "loss": 1.4881,
      "step": 20500
    },
    {
      "epoch": 5.086490172980346,
      "grad_norm": 0.9858987927436829,
      "learning_rate": 9.727326105004817e-05,
      "loss": 1.4769,
      "step": 20510
    },
    {
      "epoch": 5.0889701779403556,
      "grad_norm": 1.029187560081482,
      "learning_rate": 9.719537437241312e-05,
      "loss": 1.4935,
      "step": 20520
    },
    {
      "epoch": 5.091450182900366,
      "grad_norm": 0.9769375920295715,
      "learning_rate": 9.711748939746016e-05,
      "loss": 1.4886,
      "step": 20530
    },
    {
      "epoch": 5.0939301878603755,
      "grad_norm": 1.0216981172561646,
      "learning_rate": 9.703960617247317e-05,
      "loss": 1.537,
      "step": 20540
    },
    {
      "epoch": 5.096410192820386,
      "grad_norm": 1.0032145977020264,
      "learning_rate": 9.696172474473479e-05,
      "loss": 1.5073,
      "step": 20550
    },
    {
      "epoch": 5.0988901977803955,
      "grad_norm": 0.9975841641426086,
      "learning_rate": 9.688384516152672e-05,
      "loss": 1.4811,
      "step": 20560
    },
    {
      "epoch": 5.101370202740405,
      "grad_norm": 0.9111374020576477,
      "learning_rate": 9.680596747012945e-05,
      "loss": 1.5231,
      "step": 20570
    },
    {
      "epoch": 5.1038502077004155,
      "grad_norm": 0.9400519728660583,
      "learning_rate": 9.67280917178224e-05,
      "loss": 1.5583,
      "step": 20580
    },
    {
      "epoch": 5.106330212660425,
      "grad_norm": 0.9704780578613281,
      "learning_rate": 9.665021795188373e-05,
      "loss": 1.4422,
      "step": 20590
    },
    {
      "epoch": 5.1088102176204355,
      "grad_norm": 0.9567640423774719,
      "learning_rate": 9.657234621959051e-05,
      "loss": 1.5286,
      "step": 20600
    },
    {
      "epoch": 5.111290222580445,
      "grad_norm": 0.9019995927810669,
      "learning_rate": 9.649447656821839e-05,
      "loss": 1.5245,
      "step": 20610
    },
    {
      "epoch": 5.1137702275404555,
      "grad_norm": 0.9860844016075134,
      "learning_rate": 9.641660904504195e-05,
      "loss": 1.5136,
      "step": 20620
    },
    {
      "epoch": 5.116250232500465,
      "grad_norm": 0.9451846480369568,
      "learning_rate": 9.633874369733432e-05,
      "loss": 1.5181,
      "step": 20630
    },
    {
      "epoch": 5.118730237460475,
      "grad_norm": 0.9941641688346863,
      "learning_rate": 9.626088057236745e-05,
      "loss": 1.4943,
      "step": 20640
    },
    {
      "epoch": 5.121210242420485,
      "grad_norm": 0.9477522969245911,
      "learning_rate": 9.618301971741184e-05,
      "loss": 1.5175,
      "step": 20650
    },
    {
      "epoch": 5.123690247380495,
      "grad_norm": 0.9193263053894043,
      "learning_rate": 9.61051611797367e-05,
      "loss": 1.5415,
      "step": 20660
    },
    {
      "epoch": 5.126170252340505,
      "grad_norm": 0.9805010557174683,
      "learning_rate": 9.602730500660967e-05,
      "loss": 1.5124,
      "step": 20670
    },
    {
      "epoch": 5.128650257300515,
      "grad_norm": 1.0175899267196655,
      "learning_rate": 9.594945124529719e-05,
      "loss": 1.5062,
      "step": 20680
    },
    {
      "epoch": 5.131130262260524,
      "grad_norm": 0.9650846719741821,
      "learning_rate": 9.5871599943064e-05,
      "loss": 1.5027,
      "step": 20690
    },
    {
      "epoch": 5.133610267220535,
      "grad_norm": 0.977501392364502,
      "learning_rate": 9.579375114717351e-05,
      "loss": 1.5471,
      "step": 20700
    },
    {
      "epoch": 5.136090272180544,
      "grad_norm": 0.9525604248046875,
      "learning_rate": 9.571590490488755e-05,
      "loss": 1.5443,
      "step": 20710
    },
    {
      "epoch": 5.138570277140555,
      "grad_norm": 0.9992299675941467,
      "learning_rate": 9.563806126346642e-05,
      "loss": 1.4349,
      "step": 20720
    },
    {
      "epoch": 5.141050282100564,
      "grad_norm": 0.9223850965499878,
      "learning_rate": 9.556022027016878e-05,
      "loss": 1.4943,
      "step": 20730
    },
    {
      "epoch": 5.143530287060574,
      "grad_norm": 0.9668163657188416,
      "learning_rate": 9.54823819722518e-05,
      "loss": 1.5274,
      "step": 20740
    },
    {
      "epoch": 5.146010292020584,
      "grad_norm": 0.9030009508132935,
      "learning_rate": 9.540454641697084e-05,
      "loss": 1.5134,
      "step": 20750
    },
    {
      "epoch": 5.148490296980594,
      "grad_norm": 0.9662803411483765,
      "learning_rate": 9.53267136515798e-05,
      "loss": 1.4816,
      "step": 20760
    },
    {
      "epoch": 5.150970301940604,
      "grad_norm": 1.0170722007751465,
      "learning_rate": 9.52488837233307e-05,
      "loss": 1.4449,
      "step": 20770
    },
    {
      "epoch": 5.153450306900614,
      "grad_norm": 0.9316909313201904,
      "learning_rate": 9.517105667947397e-05,
      "loss": 1.5169,
      "step": 20780
    },
    {
      "epoch": 5.155930311860624,
      "grad_norm": 0.9504343867301941,
      "learning_rate": 9.509323256725821e-05,
      "loss": 1.4605,
      "step": 20790
    },
    {
      "epoch": 5.158410316820634,
      "grad_norm": 0.9367119669914246,
      "learning_rate": 9.501541143393028e-05,
      "loss": 1.5345,
      "step": 20800
    },
    {
      "epoch": 5.160890321780643,
      "grad_norm": 1.2000889778137207,
      "learning_rate": 9.493759332673527e-05,
      "loss": 1.4495,
      "step": 20810
    },
    {
      "epoch": 5.163370326740654,
      "grad_norm": 0.9850399494171143,
      "learning_rate": 9.485977829291627e-05,
      "loss": 1.5541,
      "step": 20820
    },
    {
      "epoch": 5.165850331700663,
      "grad_norm": 0.9986492395401001,
      "learning_rate": 9.478196637971475e-05,
      "loss": 1.5413,
      "step": 20830
    },
    {
      "epoch": 5.168330336660674,
      "grad_norm": 0.9224624037742615,
      "learning_rate": 9.470415763437004e-05,
      "loss": 1.5242,
      "step": 20840
    },
    {
      "epoch": 5.170810341620683,
      "grad_norm": 0.9644695520401001,
      "learning_rate": 9.462635210411974e-05,
      "loss": 1.5029,
      "step": 20850
    },
    {
      "epoch": 5.173290346580693,
      "grad_norm": 0.9682588577270508,
      "learning_rate": 9.454854983619936e-05,
      "loss": 1.5529,
      "step": 20860
    },
    {
      "epoch": 5.175770351540703,
      "grad_norm": 0.9783344864845276,
      "learning_rate": 9.447075087784254e-05,
      "loss": 1.5205,
      "step": 20870
    },
    {
      "epoch": 5.178250356500713,
      "grad_norm": 0.9680948853492737,
      "learning_rate": 9.439295527628081e-05,
      "loss": 1.5127,
      "step": 20880
    },
    {
      "epoch": 5.180730361460723,
      "grad_norm": 0.9525998830795288,
      "learning_rate": 9.431516307874377e-05,
      "loss": 1.4869,
      "step": 20890
    },
    {
      "epoch": 5.183210366420733,
      "grad_norm": 0.9906119704246521,
      "learning_rate": 9.423737433245883e-05,
      "loss": 1.4938,
      "step": 20900
    },
    {
      "epoch": 5.185690371380743,
      "grad_norm": 0.9663385152816772,
      "learning_rate": 9.41595890846514e-05,
      "loss": 1.4462,
      "step": 20910
    },
    {
      "epoch": 5.188170376340753,
      "grad_norm": 0.9746707081794739,
      "learning_rate": 9.408180738254471e-05,
      "loss": 1.4846,
      "step": 20920
    },
    {
      "epoch": 5.190650381300762,
      "grad_norm": 0.9378529191017151,
      "learning_rate": 9.400402927335992e-05,
      "loss": 1.5291,
      "step": 20930
    },
    {
      "epoch": 5.193130386260773,
      "grad_norm": 0.9330746531486511,
      "learning_rate": 9.392625480431587e-05,
      "loss": 1.5204,
      "step": 20940
    },
    {
      "epoch": 5.195610391220782,
      "grad_norm": 1.0342406034469604,
      "learning_rate": 9.384848402262938e-05,
      "loss": 1.5185,
      "step": 20950
    },
    {
      "epoch": 5.198090396180793,
      "grad_norm": 0.9911377429962158,
      "learning_rate": 9.37707169755148e-05,
      "loss": 1.5061,
      "step": 20960
    },
    {
      "epoch": 5.200570401140802,
      "grad_norm": 0.9533239603042603,
      "learning_rate": 9.369295371018442e-05,
      "loss": 1.5666,
      "step": 20970
    },
    {
      "epoch": 5.203050406100812,
      "grad_norm": 0.9491400122642517,
      "learning_rate": 9.361519427384808e-05,
      "loss": 1.5549,
      "step": 20980
    },
    {
      "epoch": 5.205530411060822,
      "grad_norm": 0.938381016254425,
      "learning_rate": 9.353743871371345e-05,
      "loss": 1.4659,
      "step": 20990
    },
    {
      "epoch": 5.208010416020832,
      "grad_norm": 0.9344185590744019,
      "learning_rate": 9.345968707698569e-05,
      "loss": 1.5561,
      "step": 21000
    },
    {
      "epoch": 5.210490420980842,
      "grad_norm": 0.9927399754524231,
      "learning_rate": 9.338193941086772e-05,
      "loss": 1.5511,
      "step": 21010
    },
    {
      "epoch": 5.212970425940852,
      "grad_norm": 0.9011238217353821,
      "learning_rate": 9.330419576255986e-05,
      "loss": 1.4998,
      "step": 21020
    },
    {
      "epoch": 5.215450430900862,
      "grad_norm": 0.9369461536407471,
      "learning_rate": 9.322645617926026e-05,
      "loss": 1.5338,
      "step": 21030
    },
    {
      "epoch": 5.217930435860872,
      "grad_norm": 0.9484929442405701,
      "learning_rate": 9.314872070816434e-05,
      "loss": 1.4942,
      "step": 21040
    },
    {
      "epoch": 5.220410440820881,
      "grad_norm": 1.034021258354187,
      "learning_rate": 9.307098939646518e-05,
      "loss": 1.563,
      "step": 21050
    },
    {
      "epoch": 5.222890445780892,
      "grad_norm": 1.0780504941940308,
      "learning_rate": 9.299326229135326e-05,
      "loss": 1.4832,
      "step": 21060
    },
    {
      "epoch": 5.225370450740901,
      "grad_norm": 0.9114693403244019,
      "learning_rate": 9.29155394400166e-05,
      "loss": 1.5379,
      "step": 21070
    },
    {
      "epoch": 5.227850455700912,
      "grad_norm": 0.9753614068031311,
      "learning_rate": 9.283782088964049e-05,
      "loss": 1.5208,
      "step": 21080
    },
    {
      "epoch": 5.230330460660921,
      "grad_norm": 0.9457783102989197,
      "learning_rate": 9.27601066874078e-05,
      "loss": 1.5519,
      "step": 21090
    },
    {
      "epoch": 5.232810465620931,
      "grad_norm": 0.9436642527580261,
      "learning_rate": 9.268239688049854e-05,
      "loss": 1.5891,
      "step": 21100
    },
    {
      "epoch": 5.235290470580941,
      "grad_norm": 0.9453377723693848,
      "learning_rate": 9.260469151609021e-05,
      "loss": 1.4666,
      "step": 21110
    },
    {
      "epoch": 5.237770475540951,
      "grad_norm": 0.9990126490592957,
      "learning_rate": 9.252699064135758e-05,
      "loss": 1.547,
      "step": 21120
    },
    {
      "epoch": 5.240250480500961,
      "grad_norm": 0.9898304343223572,
      "learning_rate": 9.244929430347269e-05,
      "loss": 1.5728,
      "step": 21130
    },
    {
      "epoch": 5.242730485460971,
      "grad_norm": 0.8843860626220703,
      "learning_rate": 9.237160254960476e-05,
      "loss": 1.5918,
      "step": 21140
    },
    {
      "epoch": 5.24521049042098,
      "grad_norm": 1.0320794582366943,
      "learning_rate": 9.229391542692039e-05,
      "loss": 1.5393,
      "step": 21150
    },
    {
      "epoch": 5.247690495380991,
      "grad_norm": 1.0260951519012451,
      "learning_rate": 9.221623298258315e-05,
      "loss": 1.5119,
      "step": 21160
    },
    {
      "epoch": 5.250170500341,
      "grad_norm": 1.0024783611297607,
      "learning_rate": 9.213855526375395e-05,
      "loss": 1.5714,
      "step": 21170
    },
    {
      "epoch": 5.252650505301011,
      "grad_norm": 0.9587418437004089,
      "learning_rate": 9.206088231759073e-05,
      "loss": 1.5112,
      "step": 21180
    },
    {
      "epoch": 5.25513051026102,
      "grad_norm": 0.9575029015541077,
      "learning_rate": 9.19832141912486e-05,
      "loss": 1.5,
      "step": 21190
    },
    {
      "epoch": 5.257610515221031,
      "grad_norm": 0.9021822810173035,
      "learning_rate": 9.190555093187967e-05,
      "loss": 1.5117,
      "step": 21200
    },
    {
      "epoch": 5.26009052018104,
      "grad_norm": 1.0176606178283691,
      "learning_rate": 9.182789258663321e-05,
      "loss": 1.5093,
      "step": 21210
    },
    {
      "epoch": 5.26257052514105,
      "grad_norm": 0.8936800360679626,
      "learning_rate": 9.17502392026553e-05,
      "loss": 1.4963,
      "step": 21220
    },
    {
      "epoch": 5.26505053010106,
      "grad_norm": 0.9021929502487183,
      "learning_rate": 9.16725908270893e-05,
      "loss": 1.4675,
      "step": 21230
    },
    {
      "epoch": 5.26753053506107,
      "grad_norm": 0.9234278798103333,
      "learning_rate": 9.159494750707526e-05,
      "loss": 1.5487,
      "step": 21240
    },
    {
      "epoch": 5.27001054002108,
      "grad_norm": 0.9856491088867188,
      "learning_rate": 9.15173092897503e-05,
      "loss": 1.4821,
      "step": 21250
    },
    {
      "epoch": 5.27249054498109,
      "grad_norm": 0.9914385080337524,
      "learning_rate": 9.143967622224842e-05,
      "loss": 1.5057,
      "step": 21260
    },
    {
      "epoch": 5.2749705499410995,
      "grad_norm": 0.9577724933624268,
      "learning_rate": 9.13620483517005e-05,
      "loss": 1.5988,
      "step": 21270
    },
    {
      "epoch": 5.27745055490111,
      "grad_norm": 0.9790413975715637,
      "learning_rate": 9.128442572523417e-05,
      "loss": 1.5017,
      "step": 21280
    },
    {
      "epoch": 5.2799305598611195,
      "grad_norm": 0.9989108443260193,
      "learning_rate": 9.120680838997411e-05,
      "loss": 1.5679,
      "step": 21290
    },
    {
      "epoch": 5.28241056482113,
      "grad_norm": 0.9500874280929565,
      "learning_rate": 9.112919639304148e-05,
      "loss": 1.5154,
      "step": 21300
    },
    {
      "epoch": 5.2848905697811395,
      "grad_norm": 1.0164166688919067,
      "learning_rate": 9.105158978155445e-05,
      "loss": 1.53,
      "step": 21310
    },
    {
      "epoch": 5.28737057474115,
      "grad_norm": 0.9272359609603882,
      "learning_rate": 9.097398860262776e-05,
      "loss": 1.4727,
      "step": 21320
    },
    {
      "epoch": 5.2898505797011595,
      "grad_norm": 1.0567892789840698,
      "learning_rate": 9.0896392903373e-05,
      "loss": 1.5458,
      "step": 21330
    },
    {
      "epoch": 5.292330584661169,
      "grad_norm": 0.9932803511619568,
      "learning_rate": 9.081880273089827e-05,
      "loss": 1.5074,
      "step": 21340
    },
    {
      "epoch": 5.2948105896211795,
      "grad_norm": 0.9956728219985962,
      "learning_rate": 9.074121813230847e-05,
      "loss": 1.4898,
      "step": 21350
    },
    {
      "epoch": 5.297290594581189,
      "grad_norm": 1.0154134035110474,
      "learning_rate": 9.066363915470495e-05,
      "loss": 1.5296,
      "step": 21360
    },
    {
      "epoch": 5.2997705995411994,
      "grad_norm": 1.0684008598327637,
      "learning_rate": 9.05860658451858e-05,
      "loss": 1.4418,
      "step": 21370
    },
    {
      "epoch": 5.302250604501209,
      "grad_norm": 0.9834532141685486,
      "learning_rate": 9.050849825084559e-05,
      "loss": 1.4744,
      "step": 21380
    },
    {
      "epoch": 5.3047306094612185,
      "grad_norm": 0.9449496269226074,
      "learning_rate": 9.043093641877547e-05,
      "loss": 1.5429,
      "step": 21390
    },
    {
      "epoch": 5.307210614421229,
      "grad_norm": 0.8877017498016357,
      "learning_rate": 9.0353380396063e-05,
      "loss": 1.4816,
      "step": 21400
    },
    {
      "epoch": 5.3096906193812385,
      "grad_norm": 0.9575859308242798,
      "learning_rate": 9.027583022979234e-05,
      "loss": 1.5607,
      "step": 21410
    },
    {
      "epoch": 5.312170624341249,
      "grad_norm": 0.954052746295929,
      "learning_rate": 9.019828596704394e-05,
      "loss": 1.5308,
      "step": 21420
    },
    {
      "epoch": 5.3146506293012585,
      "grad_norm": 0.9780444502830505,
      "learning_rate": 9.012074765489482e-05,
      "loss": 1.5246,
      "step": 21430
    },
    {
      "epoch": 5.317130634261268,
      "grad_norm": 0.9581279754638672,
      "learning_rate": 9.004321534041835e-05,
      "loss": 1.5133,
      "step": 21440
    },
    {
      "epoch": 5.3196106392212785,
      "grad_norm": 0.9770628213882446,
      "learning_rate": 8.996568907068415e-05,
      "loss": 1.5733,
      "step": 21450
    },
    {
      "epoch": 5.322090644181288,
      "grad_norm": 0.9912844896316528,
      "learning_rate": 8.988816889275829e-05,
      "loss": 1.5536,
      "step": 21460
    },
    {
      "epoch": 5.3245706491412985,
      "grad_norm": 1.015771746635437,
      "learning_rate": 8.981065485370308e-05,
      "loss": 1.5477,
      "step": 21470
    },
    {
      "epoch": 5.327050654101308,
      "grad_norm": 0.9286255240440369,
      "learning_rate": 8.973314700057717e-05,
      "loss": 1.534,
      "step": 21480
    },
    {
      "epoch": 5.3295306590613185,
      "grad_norm": 0.9836567640304565,
      "learning_rate": 8.965564538043535e-05,
      "loss": 1.4998,
      "step": 21490
    },
    {
      "epoch": 5.332010664021328,
      "grad_norm": 0.9883909821510315,
      "learning_rate": 8.957815004032876e-05,
      "loss": 1.5322,
      "step": 21500
    },
    {
      "epoch": 5.334490668981338,
      "grad_norm": 0.9234176278114319,
      "learning_rate": 8.950066102730456e-05,
      "loss": 1.5265,
      "step": 21510
    },
    {
      "epoch": 5.336970673941348,
      "grad_norm": 1.0736953020095825,
      "learning_rate": 8.942317838840623e-05,
      "loss": 1.5229,
      "step": 21520
    },
    {
      "epoch": 5.339450678901358,
      "grad_norm": 1.023149847984314,
      "learning_rate": 8.934570217067327e-05,
      "loss": 1.5687,
      "step": 21530
    },
    {
      "epoch": 5.341930683861368,
      "grad_norm": 1.065730094909668,
      "learning_rate": 8.926823242114136e-05,
      "loss": 1.493,
      "step": 21540
    },
    {
      "epoch": 5.344410688821378,
      "grad_norm": 0.9680074453353882,
      "learning_rate": 8.91907691868422e-05,
      "loss": 1.5618,
      "step": 21550
    },
    {
      "epoch": 5.346890693781387,
      "grad_norm": 0.9289672374725342,
      "learning_rate": 8.911331251480357e-05,
      "loss": 1.5213,
      "step": 21560
    },
    {
      "epoch": 5.349370698741398,
      "grad_norm": 0.9402804374694824,
      "learning_rate": 8.903586245204917e-05,
      "loss": 1.5191,
      "step": 21570
    },
    {
      "epoch": 5.351850703701407,
      "grad_norm": 0.9461135268211365,
      "learning_rate": 8.895841904559888e-05,
      "loss": 1.4828,
      "step": 21580
    },
    {
      "epoch": 5.354330708661418,
      "grad_norm": 0.9638580679893494,
      "learning_rate": 8.888098234246832e-05,
      "loss": 1.4709,
      "step": 21590
    },
    {
      "epoch": 5.356810713621427,
      "grad_norm": 0.986600935459137,
      "learning_rate": 8.880355238966923e-05,
      "loss": 1.5509,
      "step": 21600
    },
    {
      "epoch": 5.359290718581438,
      "grad_norm": 0.9543759822845459,
      "learning_rate": 8.872612923420909e-05,
      "loss": 1.5371,
      "step": 21610
    },
    {
      "epoch": 5.361770723541447,
      "grad_norm": 1.040656328201294,
      "learning_rate": 8.86487129230914e-05,
      "loss": 1.5651,
      "step": 21620
    },
    {
      "epoch": 5.364250728501457,
      "grad_norm": 0.995304524898529,
      "learning_rate": 8.857130350331535e-05,
      "loss": 1.5591,
      "step": 21630
    },
    {
      "epoch": 5.366730733461467,
      "grad_norm": 0.9509426355361938,
      "learning_rate": 8.849390102187614e-05,
      "loss": 1.4872,
      "step": 21640
    },
    {
      "epoch": 5.369210738421477,
      "grad_norm": 0.9640476703643799,
      "learning_rate": 8.841650552576453e-05,
      "loss": 1.5134,
      "step": 21650
    },
    {
      "epoch": 5.371690743381487,
      "grad_norm": 0.9115557074546814,
      "learning_rate": 8.833911706196725e-05,
      "loss": 1.4922,
      "step": 21660
    },
    {
      "epoch": 5.374170748341497,
      "grad_norm": 0.9620742797851562,
      "learning_rate": 8.826173567746662e-05,
      "loss": 1.5666,
      "step": 21670
    },
    {
      "epoch": 5.376650753301506,
      "grad_norm": 0.9360364079475403,
      "learning_rate": 8.818436141924072e-05,
      "loss": 1.5095,
      "step": 21680
    },
    {
      "epoch": 5.379130758261517,
      "grad_norm": 0.9433877468109131,
      "learning_rate": 8.81069943342633e-05,
      "loss": 1.4912,
      "step": 21690
    },
    {
      "epoch": 5.381610763221526,
      "grad_norm": 0.9260382652282715,
      "learning_rate": 8.802963446950377e-05,
      "loss": 1.4416,
      "step": 21700
    },
    {
      "epoch": 5.384090768181537,
      "grad_norm": 1.1015149354934692,
      "learning_rate": 8.795228187192707e-05,
      "loss": 1.5746,
      "step": 21710
    },
    {
      "epoch": 5.386570773141546,
      "grad_norm": 0.9676945209503174,
      "learning_rate": 8.787493658849386e-05,
      "loss": 1.6083,
      "step": 21720
    },
    {
      "epoch": 5.389050778101556,
      "grad_norm": 0.9367676973342896,
      "learning_rate": 8.779759866616025e-05,
      "loss": 1.5372,
      "step": 21730
    },
    {
      "epoch": 5.391530783061566,
      "grad_norm": 0.9828059673309326,
      "learning_rate": 8.772026815187796e-05,
      "loss": 1.6202,
      "step": 21740
    },
    {
      "epoch": 5.394010788021576,
      "grad_norm": 1.0444278717041016,
      "learning_rate": 8.764294509259414e-05,
      "loss": 1.6003,
      "step": 21750
    },
    {
      "epoch": 5.396490792981586,
      "grad_norm": 0.9950647354125977,
      "learning_rate": 8.756562953525152e-05,
      "loss": 1.5197,
      "step": 21760
    },
    {
      "epoch": 5.398970797941596,
      "grad_norm": 0.9401552677154541,
      "learning_rate": 8.74883215267881e-05,
      "loss": 1.5367,
      "step": 21770
    },
    {
      "epoch": 5.401450802901606,
      "grad_norm": 0.9246699213981628,
      "learning_rate": 8.741102111413748e-05,
      "loss": 1.4873,
      "step": 21780
    },
    {
      "epoch": 5.403930807861616,
      "grad_norm": 1.1276699304580688,
      "learning_rate": 8.733372834422854e-05,
      "loss": 1.5845,
      "step": 21790
    },
    {
      "epoch": 5.406410812821625,
      "grad_norm": 0.9638933539390564,
      "learning_rate": 8.725644326398558e-05,
      "loss": 1.5181,
      "step": 21800
    },
    {
      "epoch": 5.408890817781636,
      "grad_norm": 0.9737806916236877,
      "learning_rate": 8.717916592032818e-05,
      "loss": 1.5819,
      "step": 21810
    },
    {
      "epoch": 5.411370822741645,
      "grad_norm": 0.9329572916030884,
      "learning_rate": 8.71018963601713e-05,
      "loss": 1.5322,
      "step": 21820
    },
    {
      "epoch": 5.413850827701656,
      "grad_norm": 1.0153096914291382,
      "learning_rate": 8.7024634630425e-05,
      "loss": 1.5332,
      "step": 21830
    },
    {
      "epoch": 5.416330832661665,
      "grad_norm": 1.0339939594268799,
      "learning_rate": 8.694738077799488e-05,
      "loss": 1.5554,
      "step": 21840
    },
    {
      "epoch": 5.418810837621676,
      "grad_norm": 0.9483512043952942,
      "learning_rate": 8.687013484978142e-05,
      "loss": 1.4977,
      "step": 21850
    },
    {
      "epoch": 5.421290842581685,
      "grad_norm": 1.008466362953186,
      "learning_rate": 8.679289689268056e-05,
      "loss": 1.5357,
      "step": 21860
    },
    {
      "epoch": 5.423770847541695,
      "grad_norm": 0.9744418263435364,
      "learning_rate": 8.671566695358324e-05,
      "loss": 1.5586,
      "step": 21870
    },
    {
      "epoch": 5.426250852501705,
      "grad_norm": 0.9113421440124512,
      "learning_rate": 8.663844507937562e-05,
      "loss": 1.5162,
      "step": 21880
    },
    {
      "epoch": 5.428730857461715,
      "grad_norm": 1.041822910308838,
      "learning_rate": 8.656123131693892e-05,
      "loss": 1.5326,
      "step": 21890
    },
    {
      "epoch": 5.431210862421725,
      "grad_norm": 1.0268244743347168,
      "learning_rate": 8.648402571314949e-05,
      "loss": 1.4946,
      "step": 21900
    },
    {
      "epoch": 5.433690867381735,
      "grad_norm": 0.9706840515136719,
      "learning_rate": 8.64068283148786e-05,
      "loss": 1.5051,
      "step": 21910
    },
    {
      "epoch": 5.436170872341744,
      "grad_norm": 0.8814148902893066,
      "learning_rate": 8.632963916899268e-05,
      "loss": 1.5925,
      "step": 21920
    },
    {
      "epoch": 5.438650877301755,
      "grad_norm": 0.9990885853767395,
      "learning_rate": 8.625245832235307e-05,
      "loss": 1.5186,
      "step": 21930
    },
    {
      "epoch": 5.441130882261764,
      "grad_norm": 0.9785584211349487,
      "learning_rate": 8.617528582181612e-05,
      "loss": 1.4405,
      "step": 21940
    },
    {
      "epoch": 5.443610887221775,
      "grad_norm": 1.027880072593689,
      "learning_rate": 8.609812171423305e-05,
      "loss": 1.5838,
      "step": 21950
    },
    {
      "epoch": 5.446090892181784,
      "grad_norm": 0.9656705856323242,
      "learning_rate": 8.602096604645009e-05,
      "loss": 1.5129,
      "step": 21960
    },
    {
      "epoch": 5.448570897141794,
      "grad_norm": 0.9562718272209167,
      "learning_rate": 8.594381886530818e-05,
      "loss": 1.5247,
      "step": 21970
    },
    {
      "epoch": 5.451050902101804,
      "grad_norm": 0.9530388116836548,
      "learning_rate": 8.586668021764329e-05,
      "loss": 1.5454,
      "step": 21980
    },
    {
      "epoch": 5.453530907061814,
      "grad_norm": 1.0748162269592285,
      "learning_rate": 8.578955015028605e-05,
      "loss": 1.5,
      "step": 21990
    },
    {
      "epoch": 5.456010912021824,
      "grad_norm": 1.067697286605835,
      "learning_rate": 8.571242871006202e-05,
      "loss": 1.5529,
      "step": 22000
    },
    {
      "epoch": 5.458490916981834,
      "grad_norm": 0.9875476360321045,
      "learning_rate": 8.563531594379142e-05,
      "loss": 1.5706,
      "step": 22010
    },
    {
      "epoch": 5.460970921941843,
      "grad_norm": 1.0351884365081787,
      "learning_rate": 8.555821189828932e-05,
      "loss": 1.6127,
      "step": 22020
    },
    {
      "epoch": 5.463450926901854,
      "grad_norm": 0.9771510362625122,
      "learning_rate": 8.548111662036528e-05,
      "loss": 1.5345,
      "step": 22030
    },
    {
      "epoch": 5.465930931861863,
      "grad_norm": 0.9876402020454407,
      "learning_rate": 8.540403015682381e-05,
      "loss": 1.456,
      "step": 22040
    },
    {
      "epoch": 5.468410936821874,
      "grad_norm": 1.109188199043274,
      "learning_rate": 8.532695255446383e-05,
      "loss": 1.5097,
      "step": 22050
    },
    {
      "epoch": 5.470890941781883,
      "grad_norm": 0.9162520170211792,
      "learning_rate": 8.524988386007905e-05,
      "loss": 1.481,
      "step": 22060
    },
    {
      "epoch": 5.473370946741894,
      "grad_norm": 0.9496723413467407,
      "learning_rate": 8.517282412045771e-05,
      "loss": 1.5208,
      "step": 22070
    },
    {
      "epoch": 5.475850951701903,
      "grad_norm": 1.0151824951171875,
      "learning_rate": 8.509577338238255e-05,
      "loss": 1.5203,
      "step": 22080
    },
    {
      "epoch": 5.478330956661913,
      "grad_norm": 1.076178789138794,
      "learning_rate": 8.501873169263097e-05,
      "loss": 1.4665,
      "step": 22090
    },
    {
      "epoch": 5.480810961621923,
      "grad_norm": 0.9714109301567078,
      "learning_rate": 8.494169909797478e-05,
      "loss": 1.5494,
      "step": 22100
    },
    {
      "epoch": 5.483290966581933,
      "grad_norm": 0.9446751475334167,
      "learning_rate": 8.486467564518034e-05,
      "loss": 1.4615,
      "step": 22110
    },
    {
      "epoch": 5.485770971541943,
      "grad_norm": 0.9611682295799255,
      "learning_rate": 8.478766138100834e-05,
      "loss": 1.575,
      "step": 22120
    },
    {
      "epoch": 5.488250976501953,
      "grad_norm": 0.9424753785133362,
      "learning_rate": 8.471065635221404e-05,
      "loss": 1.4713,
      "step": 22130
    },
    {
      "epoch": 5.490730981461963,
      "grad_norm": 0.9659827947616577,
      "learning_rate": 8.463366060554698e-05,
      "loss": 1.5298,
      "step": 22140
    },
    {
      "epoch": 5.493210986421973,
      "grad_norm": 0.9294324517250061,
      "learning_rate": 8.455667418775117e-05,
      "loss": 1.5346,
      "step": 22150
    },
    {
      "epoch": 5.4956909913819825,
      "grad_norm": 0.9476675391197205,
      "learning_rate": 8.447969714556484e-05,
      "loss": 1.5124,
      "step": 22160
    },
    {
      "epoch": 5.498170996341993,
      "grad_norm": 0.947379469871521,
      "learning_rate": 8.440272952572064e-05,
      "loss": 1.5919,
      "step": 22170
    },
    {
      "epoch": 5.5006510013020025,
      "grad_norm": 0.9382186532020569,
      "learning_rate": 8.432577137494537e-05,
      "loss": 1.4837,
      "step": 22180
    },
    {
      "epoch": 5.503131006262013,
      "grad_norm": 0.927191436290741,
      "learning_rate": 8.424882273996024e-05,
      "loss": 1.5346,
      "step": 22190
    },
    {
      "epoch": 5.5056110112220225,
      "grad_norm": 0.949855387210846,
      "learning_rate": 8.417188366748052e-05,
      "loss": 1.4853,
      "step": 22200
    },
    {
      "epoch": 5.508091016182032,
      "grad_norm": 1.0020701885223389,
      "learning_rate": 8.40949542042158e-05,
      "loss": 1.5281,
      "step": 22210
    },
    {
      "epoch": 5.5105710211420424,
      "grad_norm": 1.042439341545105,
      "learning_rate": 8.401803439686977e-05,
      "loss": 1.5147,
      "step": 22220
    },
    {
      "epoch": 5.513051026102052,
      "grad_norm": 0.9486450552940369,
      "learning_rate": 8.39411242921403e-05,
      "loss": 1.5571,
      "step": 22230
    },
    {
      "epoch": 5.515531031062062,
      "grad_norm": 0.924824595451355,
      "learning_rate": 8.386422393671933e-05,
      "loss": 1.5526,
      "step": 22240
    },
    {
      "epoch": 5.518011036022072,
      "grad_norm": 0.9190182685852051,
      "learning_rate": 8.378733337729294e-05,
      "loss": 1.4867,
      "step": 22250
    },
    {
      "epoch": 5.5204910409820815,
      "grad_norm": 1.0011485815048218,
      "learning_rate": 8.371045266054114e-05,
      "loss": 1.5246,
      "step": 22260
    },
    {
      "epoch": 5.522971045942092,
      "grad_norm": 0.9253968596458435,
      "learning_rate": 8.36335818331381e-05,
      "loss": 1.5336,
      "step": 22270
    },
    {
      "epoch": 5.5254510509021015,
      "grad_norm": 0.9603173136711121,
      "learning_rate": 8.35567209417519e-05,
      "loss": 1.5006,
      "step": 22280
    },
    {
      "epoch": 5.527931055862112,
      "grad_norm": 1.0172189474105835,
      "learning_rate": 8.347987003304469e-05,
      "loss": 1.4873,
      "step": 22290
    },
    {
      "epoch": 5.5304110608221215,
      "grad_norm": 0.9984214305877686,
      "learning_rate": 8.340302915367241e-05,
      "loss": 1.5289,
      "step": 22300
    },
    {
      "epoch": 5.532891065782131,
      "grad_norm": 0.943496584892273,
      "learning_rate": 8.332619835028506e-05,
      "loss": 1.5037,
      "step": 22310
    },
    {
      "epoch": 5.5353710707421415,
      "grad_norm": 0.9430272579193115,
      "learning_rate": 8.324937766952638e-05,
      "loss": 1.5323,
      "step": 22320
    },
    {
      "epoch": 5.537851075702151,
      "grad_norm": 0.9519038200378418,
      "learning_rate": 8.317256715803407e-05,
      "loss": 1.5355,
      "step": 22330
    },
    {
      "epoch": 5.5403310806621615,
      "grad_norm": 1.0110712051391602,
      "learning_rate": 8.309576686243964e-05,
      "loss": 1.5617,
      "step": 22340
    },
    {
      "epoch": 5.542811085622171,
      "grad_norm": 1.0450818538665771,
      "learning_rate": 8.301897682936838e-05,
      "loss": 1.4925,
      "step": 22350
    },
    {
      "epoch": 5.5452910905821815,
      "grad_norm": 1.0279377698898315,
      "learning_rate": 8.294219710543932e-05,
      "loss": 1.5183,
      "step": 22360
    },
    {
      "epoch": 5.547771095542191,
      "grad_norm": 1.0317164659500122,
      "learning_rate": 8.286542773726534e-05,
      "loss": 1.5277,
      "step": 22370
    },
    {
      "epoch": 5.5502511005022015,
      "grad_norm": 0.9843716025352478,
      "learning_rate": 8.278866877145282e-05,
      "loss": 1.5274,
      "step": 22380
    },
    {
      "epoch": 5.552731105462211,
      "grad_norm": 0.9920136332511902,
      "learning_rate": 8.271192025460211e-05,
      "loss": 1.5119,
      "step": 22390
    },
    {
      "epoch": 5.555211110422221,
      "grad_norm": 0.9626421928405762,
      "learning_rate": 8.263518223330697e-05,
      "loss": 1.4893,
      "step": 22400
    },
    {
      "epoch": 5.557691115382231,
      "grad_norm": 0.930276095867157,
      "learning_rate": 8.255845475415494e-05,
      "loss": 1.569,
      "step": 22410
    },
    {
      "epoch": 5.560171120342241,
      "grad_norm": 0.9820217490196228,
      "learning_rate": 8.248173786372705e-05,
      "loss": 1.5701,
      "step": 22420
    },
    {
      "epoch": 5.562651125302251,
      "grad_norm": 0.9253460764884949,
      "learning_rate": 8.240503160859804e-05,
      "loss": 1.4679,
      "step": 22430
    },
    {
      "epoch": 5.565131130262261,
      "grad_norm": 0.947089433670044,
      "learning_rate": 8.2328336035336e-05,
      "loss": 1.4811,
      "step": 22440
    },
    {
      "epoch": 5.56761113522227,
      "grad_norm": 0.9547702670097351,
      "learning_rate": 8.225165119050277e-05,
      "loss": 1.5287,
      "step": 22450
    },
    {
      "epoch": 5.570091140182281,
      "grad_norm": 0.9524838924407959,
      "learning_rate": 8.217497712065345e-05,
      "loss": 1.5594,
      "step": 22460
    },
    {
      "epoch": 5.57257114514229,
      "grad_norm": 0.9504765868186951,
      "learning_rate": 8.209831387233676e-05,
      "loss": 1.5289,
      "step": 22470
    },
    {
      "epoch": 5.575051150102301,
      "grad_norm": 1.5735292434692383,
      "learning_rate": 8.202166149209474e-05,
      "loss": 1.5328,
      "step": 22480
    },
    {
      "epoch": 5.57753115506231,
      "grad_norm": 0.8840258121490479,
      "learning_rate": 8.194502002646294e-05,
      "loss": 1.5362,
      "step": 22490
    },
    {
      "epoch": 5.58001116002232,
      "grad_norm": 0.886695146560669,
      "learning_rate": 8.186838952197018e-05,
      "loss": 1.4982,
      "step": 22500
    },
    {
      "epoch": 5.58249116498233,
      "grad_norm": 1.0302002429962158,
      "learning_rate": 8.179177002513873e-05,
      "loss": 1.5819,
      "step": 22510
    },
    {
      "epoch": 5.58497116994234,
      "grad_norm": 0.9544497132301331,
      "learning_rate": 8.171516158248406e-05,
      "loss": 1.5219,
      "step": 22520
    },
    {
      "epoch": 5.58745117490235,
      "grad_norm": 0.9738973379135132,
      "learning_rate": 8.163856424051502e-05,
      "loss": 1.5627,
      "step": 22530
    },
    {
      "epoch": 5.58993117986236,
      "grad_norm": 1.01767897605896,
      "learning_rate": 8.156197804573366e-05,
      "loss": 1.5328,
      "step": 22540
    },
    {
      "epoch": 5.592411184822369,
      "grad_norm": 0.9286201596260071,
      "learning_rate": 8.148540304463536e-05,
      "loss": 1.5356,
      "step": 22550
    },
    {
      "epoch": 5.59489118978238,
      "grad_norm": 1.0125569105148315,
      "learning_rate": 8.140883928370855e-05,
      "loss": 1.5648,
      "step": 22560
    },
    {
      "epoch": 5.597371194742389,
      "grad_norm": 0.9836313724517822,
      "learning_rate": 8.133228680943501e-05,
      "loss": 1.5341,
      "step": 22570
    },
    {
      "epoch": 5.5998511997024,
      "grad_norm": 0.9896276593208313,
      "learning_rate": 8.125574566828946e-05,
      "loss": 1.5399,
      "step": 22580
    },
    {
      "epoch": 5.602331204662409,
      "grad_norm": 0.9839266538619995,
      "learning_rate": 8.117921590674001e-05,
      "loss": 1.5382,
      "step": 22590
    },
    {
      "epoch": 5.60481120962242,
      "grad_norm": 0.9478524923324585,
      "learning_rate": 8.11026975712476e-05,
      "loss": 1.4931,
      "step": 22600
    },
    {
      "epoch": 5.607291214582429,
      "grad_norm": 0.9081190228462219,
      "learning_rate": 8.102619070826639e-05,
      "loss": 1.5147,
      "step": 22610
    },
    {
      "epoch": 5.609771219542439,
      "grad_norm": 0.9841187000274658,
      "learning_rate": 8.094969536424351e-05,
      "loss": 1.5404,
      "step": 22620
    },
    {
      "epoch": 5.612251224502449,
      "grad_norm": 0.991550624370575,
      "learning_rate": 8.087321158561917e-05,
      "loss": 1.5206,
      "step": 22630
    },
    {
      "epoch": 5.614731229462459,
      "grad_norm": 0.9358656406402588,
      "learning_rate": 8.07967394188264e-05,
      "loss": 1.5052,
      "step": 22640
    },
    {
      "epoch": 5.617211234422469,
      "grad_norm": 0.9936838150024414,
      "learning_rate": 8.07202789102914e-05,
      "loss": 1.5927,
      "step": 22650
    },
    {
      "epoch": 5.619691239382479,
      "grad_norm": 0.9371949434280396,
      "learning_rate": 8.06438301064331e-05,
      "loss": 1.5075,
      "step": 22660
    },
    {
      "epoch": 5.622171244342489,
      "grad_norm": 1.0312095880508423,
      "learning_rate": 8.056739305366346e-05,
      "loss": 1.5319,
      "step": 22670
    },
    {
      "epoch": 5.624651249302499,
      "grad_norm": 0.9725061655044556,
      "learning_rate": 8.049096779838719e-05,
      "loss": 1.5334,
      "step": 22680
    },
    {
      "epoch": 5.627131254262508,
      "grad_norm": 0.9553120732307434,
      "learning_rate": 8.041455438700191e-05,
      "loss": 1.5311,
      "step": 22690
    },
    {
      "epoch": 5.629611259222519,
      "grad_norm": 0.9907311201095581,
      "learning_rate": 8.03381528658981e-05,
      "loss": 1.5381,
      "step": 22700
    },
    {
      "epoch": 5.632091264182528,
      "grad_norm": 0.9710587859153748,
      "learning_rate": 8.026176328145887e-05,
      "loss": 1.4894,
      "step": 22710
    },
    {
      "epoch": 5.634571269142539,
      "grad_norm": 0.8912402391433716,
      "learning_rate": 8.018538568006027e-05,
      "loss": 1.5291,
      "step": 22720
    },
    {
      "epoch": 5.637051274102548,
      "grad_norm": 0.9066137075424194,
      "learning_rate": 8.010902010807084e-05,
      "loss": 1.5306,
      "step": 22730
    },
    {
      "epoch": 5.639531279062558,
      "grad_norm": 0.9754036068916321,
      "learning_rate": 8.003266661185209e-05,
      "loss": 1.5521,
      "step": 22740
    },
    {
      "epoch": 5.642011284022568,
      "grad_norm": 0.9809487462043762,
      "learning_rate": 7.995632523775795e-05,
      "loss": 1.5178,
      "step": 22750
    },
    {
      "epoch": 5.644491288982578,
      "grad_norm": 0.9499403238296509,
      "learning_rate": 7.987999603213519e-05,
      "loss": 1.4584,
      "step": 22760
    },
    {
      "epoch": 5.646971293942588,
      "grad_norm": 1.0232181549072266,
      "learning_rate": 7.980367904132302e-05,
      "loss": 1.5577,
      "step": 22770
    },
    {
      "epoch": 5.649451298902598,
      "grad_norm": 1.1556051969528198,
      "learning_rate": 7.972737431165343e-05,
      "loss": 1.5877,
      "step": 22780
    },
    {
      "epoch": 5.651931303862607,
      "grad_norm": 0.929179310798645,
      "learning_rate": 7.965108188945072e-05,
      "loss": 1.5101,
      "step": 22790
    },
    {
      "epoch": 5.654411308822618,
      "grad_norm": 0.928312361240387,
      "learning_rate": 7.957480182103198e-05,
      "loss": 1.5602,
      "step": 22800
    },
    {
      "epoch": 5.656891313782627,
      "grad_norm": 0.9117720127105713,
      "learning_rate": 7.949853415270658e-05,
      "loss": 1.5158,
      "step": 22810
    },
    {
      "epoch": 5.659371318742638,
      "grad_norm": 0.9475207924842834,
      "learning_rate": 7.942227893077652e-05,
      "loss": 1.4721,
      "step": 22820
    },
    {
      "epoch": 5.661851323702647,
      "grad_norm": 1.001230239868164,
      "learning_rate": 7.934603620153614e-05,
      "loss": 1.535,
      "step": 22830
    },
    {
      "epoch": 5.664331328662657,
      "grad_norm": 0.9647219777107239,
      "learning_rate": 7.926980601127225e-05,
      "loss": 1.5649,
      "step": 22840
    },
    {
      "epoch": 5.666811333622667,
      "grad_norm": 0.9498812556266785,
      "learning_rate": 7.9193588406264e-05,
      "loss": 1.5056,
      "step": 22850
    },
    {
      "epoch": 5.669291338582677,
      "grad_norm": 0.9786957502365112,
      "learning_rate": 7.911738343278304e-05,
      "loss": 1.5892,
      "step": 22860
    },
    {
      "epoch": 5.671771343542687,
      "grad_norm": 0.980808436870575,
      "learning_rate": 7.90411911370931e-05,
      "loss": 1.5487,
      "step": 22870
    },
    {
      "epoch": 5.674251348502697,
      "grad_norm": 1.0204776525497437,
      "learning_rate": 7.896501156545045e-05,
      "loss": 1.5215,
      "step": 22880
    },
    {
      "epoch": 5.676731353462707,
      "grad_norm": 0.9597605466842651,
      "learning_rate": 7.888884476410348e-05,
      "loss": 1.5125,
      "step": 22890
    },
    {
      "epoch": 5.679211358422717,
      "grad_norm": 0.9643506407737732,
      "learning_rate": 7.881269077929297e-05,
      "loss": 1.5202,
      "step": 22900
    },
    {
      "epoch": 5.681691363382726,
      "grad_norm": 0.9709390997886658,
      "learning_rate": 7.873654965725175e-05,
      "loss": 1.5712,
      "step": 22910
    },
    {
      "epoch": 5.684171368342737,
      "grad_norm": 0.9741882681846619,
      "learning_rate": 7.866042144420502e-05,
      "loss": 1.5791,
      "step": 22920
    },
    {
      "epoch": 5.686651373302746,
      "grad_norm": 0.9291117787361145,
      "learning_rate": 7.858430618636995e-05,
      "loss": 1.5708,
      "step": 22930
    },
    {
      "epoch": 5.689131378262757,
      "grad_norm": 0.9480016231536865,
      "learning_rate": 7.850820392995605e-05,
      "loss": 1.544,
      "step": 22940
    },
    {
      "epoch": 5.691611383222766,
      "grad_norm": 0.9766435027122498,
      "learning_rate": 7.843211472116476e-05,
      "loss": 1.5171,
      "step": 22950
    },
    {
      "epoch": 5.694091388182777,
      "grad_norm": 1.0347306728363037,
      "learning_rate": 7.835603860618972e-05,
      "loss": 1.5346,
      "step": 22960
    },
    {
      "epoch": 5.696571393142786,
      "grad_norm": 1.0211946964263916,
      "learning_rate": 7.827997563121654e-05,
      "loss": 1.5056,
      "step": 22970
    },
    {
      "epoch": 5.699051398102796,
      "grad_norm": 1.0085428953170776,
      "learning_rate": 7.820392584242293e-05,
      "loss": 1.5418,
      "step": 22980
    },
    {
      "epoch": 5.701531403062806,
      "grad_norm": 0.8925297260284424,
      "learning_rate": 7.812788928597845e-05,
      "loss": 1.5284,
      "step": 22990
    },
    {
      "epoch": 5.704011408022816,
      "grad_norm": 0.9773134589195251,
      "learning_rate": 7.805186600804489e-05,
      "loss": 1.5926,
      "step": 23000
    },
    {
      "epoch": 5.706491412982826,
      "grad_norm": 1.0136878490447998,
      "learning_rate": 7.797585605477566e-05,
      "loss": 1.5515,
      "step": 23010
    },
    {
      "epoch": 5.708971417942836,
      "grad_norm": 0.9715712666511536,
      "learning_rate": 7.789985947231632e-05,
      "loss": 1.5485,
      "step": 23020
    },
    {
      "epoch": 5.7114514229028455,
      "grad_norm": 0.9595378637313843,
      "learning_rate": 7.782387630680421e-05,
      "loss": 1.4779,
      "step": 23030
    },
    {
      "epoch": 5.713931427862856,
      "grad_norm": 0.9107018709182739,
      "learning_rate": 7.774790660436858e-05,
      "loss": 1.4954,
      "step": 23040
    },
    {
      "epoch": 5.7164114328228655,
      "grad_norm": 1.0032902956008911,
      "learning_rate": 7.767195041113039e-05,
      "loss": 1.5585,
      "step": 23050
    },
    {
      "epoch": 5.718891437782876,
      "grad_norm": 0.9346843957901001,
      "learning_rate": 7.759600777320258e-05,
      "loss": 1.5281,
      "step": 23060
    },
    {
      "epoch": 5.7213714427428854,
      "grad_norm": 0.9562672972679138,
      "learning_rate": 7.752007873668967e-05,
      "loss": 1.4774,
      "step": 23070
    },
    {
      "epoch": 5.723851447702895,
      "grad_norm": 0.9957661628723145,
      "learning_rate": 7.744416334768808e-05,
      "loss": 1.5524,
      "step": 23080
    },
    {
      "epoch": 5.726331452662905,
      "grad_norm": 1.002894401550293,
      "learning_rate": 7.736826165228581e-05,
      "loss": 1.485,
      "step": 23090
    },
    {
      "epoch": 5.728811457622915,
      "grad_norm": 0.9445948600769043,
      "learning_rate": 7.729237369656269e-05,
      "loss": 1.5815,
      "step": 23100
    },
    {
      "epoch": 5.731291462582925,
      "grad_norm": 0.977097749710083,
      "learning_rate": 7.721649952659007e-05,
      "loss": 1.5471,
      "step": 23110
    },
    {
      "epoch": 5.733771467542935,
      "grad_norm": 0.8809939026832581,
      "learning_rate": 7.714063918843106e-05,
      "loss": 1.4514,
      "step": 23120
    },
    {
      "epoch": 5.7362514725029445,
      "grad_norm": 0.9751113057136536,
      "learning_rate": 7.706479272814023e-05,
      "loss": 1.4972,
      "step": 23130
    },
    {
      "epoch": 5.738731477462955,
      "grad_norm": 0.9898456335067749,
      "learning_rate": 7.698896019176386e-05,
      "loss": 1.5328,
      "step": 23140
    },
    {
      "epoch": 5.7412114824229645,
      "grad_norm": 0.9402456879615784,
      "learning_rate": 7.691314162533967e-05,
      "loss": 1.5631,
      "step": 23150
    },
    {
      "epoch": 5.743691487382975,
      "grad_norm": 1.0204651355743408,
      "learning_rate": 7.683733707489699e-05,
      "loss": 1.5,
      "step": 23160
    },
    {
      "epoch": 5.7461714923429845,
      "grad_norm": 0.995430052280426,
      "learning_rate": 7.676154658645656e-05,
      "loss": 1.5573,
      "step": 23170
    },
    {
      "epoch": 5.748651497302995,
      "grad_norm": 1.022791862487793,
      "learning_rate": 7.668577020603068e-05,
      "loss": 1.5647,
      "step": 23180
    },
    {
      "epoch": 5.7511315022630045,
      "grad_norm": 0.98724764585495,
      "learning_rate": 7.661000797962292e-05,
      "loss": 1.5313,
      "step": 23190
    },
    {
      "epoch": 5.753611507223014,
      "grad_norm": 1.041318655014038,
      "learning_rate": 7.653425995322851e-05,
      "loss": 1.5309,
      "step": 23200
    },
    {
      "epoch": 5.7560915121830245,
      "grad_norm": 1.0121514797210693,
      "learning_rate": 7.645852617283377e-05,
      "loss": 1.5208,
      "step": 23210
    },
    {
      "epoch": 5.758571517143034,
      "grad_norm": 0.9732059836387634,
      "learning_rate": 7.63828066844166e-05,
      "loss": 1.5598,
      "step": 23220
    },
    {
      "epoch": 5.7610515221030445,
      "grad_norm": 0.9636842608451843,
      "learning_rate": 7.630710153394607e-05,
      "loss": 1.5274,
      "step": 23230
    },
    {
      "epoch": 5.763531527063054,
      "grad_norm": 0.9335260987281799,
      "learning_rate": 7.623141076738271e-05,
      "loss": 1.514,
      "step": 23240
    },
    {
      "epoch": 5.7660115320230645,
      "grad_norm": 0.977904736995697,
      "learning_rate": 7.615573443067812e-05,
      "loss": 1.487,
      "step": 23250
    },
    {
      "epoch": 5.768491536983074,
      "grad_norm": 0.9485705494880676,
      "learning_rate": 7.608007256977533e-05,
      "loss": 1.5116,
      "step": 23260
    },
    {
      "epoch": 5.770971541943084,
      "grad_norm": 0.8961159586906433,
      "learning_rate": 7.60044252306084e-05,
      "loss": 1.4846,
      "step": 23270
    },
    {
      "epoch": 5.773451546903094,
      "grad_norm": 0.9879873394966125,
      "learning_rate": 7.592879245910273e-05,
      "loss": 1.6182,
      "step": 23280
    },
    {
      "epoch": 5.775931551863104,
      "grad_norm": 0.9821351766586304,
      "learning_rate": 7.585317430117475e-05,
      "loss": 1.5422,
      "step": 23290
    },
    {
      "epoch": 5.778411556823114,
      "grad_norm": 0.9630663394927979,
      "learning_rate": 7.577757080273216e-05,
      "loss": 1.5499,
      "step": 23300
    },
    {
      "epoch": 5.780891561783124,
      "grad_norm": 0.9344465136528015,
      "learning_rate": 7.570198200967362e-05,
      "loss": 1.5832,
      "step": 23310
    },
    {
      "epoch": 5.783371566743133,
      "grad_norm": 0.9779849052429199,
      "learning_rate": 7.562640796788892e-05,
      "loss": 1.5778,
      "step": 23320
    },
    {
      "epoch": 5.785851571703144,
      "grad_norm": 0.9390009641647339,
      "learning_rate": 7.555084872325897e-05,
      "loss": 1.4917,
      "step": 23330
    },
    {
      "epoch": 5.788331576663153,
      "grad_norm": 1.118727207183838,
      "learning_rate": 7.54753043216555e-05,
      "loss": 1.5914,
      "step": 23340
    },
    {
      "epoch": 5.790811581623164,
      "grad_norm": 1.0065404176712036,
      "learning_rate": 7.53997748089415e-05,
      "loss": 1.5559,
      "step": 23350
    },
    {
      "epoch": 5.793291586583173,
      "grad_norm": 0.9158838987350464,
      "learning_rate": 7.532426023097063e-05,
      "loss": 1.6127,
      "step": 23360
    },
    {
      "epoch": 5.795771591543183,
      "grad_norm": 0.9717867970466614,
      "learning_rate": 7.524876063358773e-05,
      "loss": 1.5145,
      "step": 23370
    },
    {
      "epoch": 5.798251596503193,
      "grad_norm": 0.9658146500587463,
      "learning_rate": 7.517327606262836e-05,
      "loss": 1.5459,
      "step": 23380
    },
    {
      "epoch": 5.800731601463203,
      "grad_norm": 0.9799458384513855,
      "learning_rate": 7.509780656391912e-05,
      "loss": 1.5463,
      "step": 23390
    },
    {
      "epoch": 5.803211606423213,
      "grad_norm": 0.919108510017395,
      "learning_rate": 7.502235218327731e-05,
      "loss": 1.5224,
      "step": 23400
    },
    {
      "epoch": 5.805691611383223,
      "grad_norm": 1.0490554571151733,
      "learning_rate": 7.494691296651118e-05,
      "loss": 1.5558,
      "step": 23410
    },
    {
      "epoch": 5.808171616343232,
      "grad_norm": 0.9717322587966919,
      "learning_rate": 7.487148895941965e-05,
      "loss": 1.5053,
      "step": 23420
    },
    {
      "epoch": 5.810651621303243,
      "grad_norm": 0.9787877798080444,
      "learning_rate": 7.479608020779252e-05,
      "loss": 1.5595,
      "step": 23430
    },
    {
      "epoch": 5.813131626263252,
      "grad_norm": 1.015682339668274,
      "learning_rate": 7.472068675741025e-05,
      "loss": 1.5414,
      "step": 23440
    },
    {
      "epoch": 5.815611631223263,
      "grad_norm": 0.9951199889183044,
      "learning_rate": 7.464530865404407e-05,
      "loss": 1.5741,
      "step": 23450
    },
    {
      "epoch": 5.818091636183272,
      "grad_norm": 0.9637252688407898,
      "learning_rate": 7.45699459434558e-05,
      "loss": 1.5815,
      "step": 23460
    },
    {
      "epoch": 5.820571641143283,
      "grad_norm": 0.9071375131607056,
      "learning_rate": 7.44945986713981e-05,
      "loss": 1.5273,
      "step": 23470
    },
    {
      "epoch": 5.823051646103292,
      "grad_norm": 0.9421333074569702,
      "learning_rate": 7.4419266883614e-05,
      "loss": 1.5034,
      "step": 23480
    },
    {
      "epoch": 5.825531651063303,
      "grad_norm": 1.0461348295211792,
      "learning_rate": 7.434395062583734e-05,
      "loss": 1.5687,
      "step": 23490
    },
    {
      "epoch": 5.828011656023312,
      "grad_norm": 0.9921796321868896,
      "learning_rate": 7.426864994379243e-05,
      "loss": 1.4845,
      "step": 23500
    },
    {
      "epoch": 5.830491660983322,
      "grad_norm": 0.9265093207359314,
      "learning_rate": 7.419336488319417e-05,
      "loss": 1.6003,
      "step": 23510
    },
    {
      "epoch": 5.832971665943332,
      "grad_norm": 0.908584475517273,
      "learning_rate": 7.411809548974792e-05,
      "loss": 1.5234,
      "step": 23520
    },
    {
      "epoch": 5.835451670903342,
      "grad_norm": 0.9192255139350891,
      "learning_rate": 7.404284180914964e-05,
      "loss": 1.5112,
      "step": 23530
    },
    {
      "epoch": 5.837931675863352,
      "grad_norm": 0.9425198435783386,
      "learning_rate": 7.396760388708555e-05,
      "loss": 1.5185,
      "step": 23540
    },
    {
      "epoch": 5.840411680823362,
      "grad_norm": 0.9518033862113953,
      "learning_rate": 7.389238176923258e-05,
      "loss": 1.5166,
      "step": 23550
    },
    {
      "epoch": 5.842891685783371,
      "grad_norm": 0.9426175355911255,
      "learning_rate": 7.38171755012578e-05,
      "loss": 1.5618,
      "step": 23560
    },
    {
      "epoch": 5.845371690743382,
      "grad_norm": 0.9501382112503052,
      "learning_rate": 7.374198512881881e-05,
      "loss": 1.5021,
      "step": 23570
    },
    {
      "epoch": 5.847851695703391,
      "grad_norm": 0.94672030210495,
      "learning_rate": 7.366681069756352e-05,
      "loss": 1.5153,
      "step": 23580
    },
    {
      "epoch": 5.850331700663402,
      "grad_norm": 0.9547986388206482,
      "learning_rate": 7.359165225313019e-05,
      "loss": 1.5246,
      "step": 23590
    },
    {
      "epoch": 5.852811705623411,
      "grad_norm": 0.9420895576477051,
      "learning_rate": 7.351650984114728e-05,
      "loss": 1.5534,
      "step": 23600
    },
    {
      "epoch": 5.855291710583421,
      "grad_norm": 0.9720684289932251,
      "learning_rate": 7.344138350723369e-05,
      "loss": 1.5364,
      "step": 23610
    },
    {
      "epoch": 5.857771715543431,
      "grad_norm": 0.9347152709960938,
      "learning_rate": 7.336627329699833e-05,
      "loss": 1.5349,
      "step": 23620
    },
    {
      "epoch": 5.860251720503441,
      "grad_norm": 0.9553906917572021,
      "learning_rate": 7.329117925604054e-05,
      "loss": 1.4873,
      "step": 23630
    },
    {
      "epoch": 5.862731725463451,
      "grad_norm": 0.9500386714935303,
      "learning_rate": 7.32161014299497e-05,
      "loss": 1.5453,
      "step": 23640
    },
    {
      "epoch": 5.865211730423461,
      "grad_norm": 0.92239910364151,
      "learning_rate": 7.314103986430543e-05,
      "loss": 1.4598,
      "step": 23650
    },
    {
      "epoch": 5.86769173538347,
      "grad_norm": 0.9757826924324036,
      "learning_rate": 7.30659946046774e-05,
      "loss": 1.5205,
      "step": 23660
    },
    {
      "epoch": 5.870171740343481,
      "grad_norm": 0.9532195925712585,
      "learning_rate": 7.29909656966255e-05,
      "loss": 1.545,
      "step": 23670
    },
    {
      "epoch": 5.87265174530349,
      "grad_norm": 0.8907900452613831,
      "learning_rate": 7.291595318569951e-05,
      "loss": 1.4541,
      "step": 23680
    },
    {
      "epoch": 5.875131750263501,
      "grad_norm": 1.0454277992248535,
      "learning_rate": 7.284095711743944e-05,
      "loss": 1.5117,
      "step": 23690
    },
    {
      "epoch": 5.87761175522351,
      "grad_norm": 0.9614602327346802,
      "learning_rate": 7.27659775373752e-05,
      "loss": 1.5193,
      "step": 23700
    },
    {
      "epoch": 5.880091760183521,
      "grad_norm": 0.923225462436676,
      "learning_rate": 7.269101449102676e-05,
      "loss": 1.5317,
      "step": 23710
    },
    {
      "epoch": 5.88257176514353,
      "grad_norm": 0.940643310546875,
      "learning_rate": 7.2616068023904e-05,
      "loss": 1.494,
      "step": 23720
    },
    {
      "epoch": 5.88505177010354,
      "grad_norm": 0.9426398873329163,
      "learning_rate": 7.25411381815068e-05,
      "loss": 1.4917,
      "step": 23730
    },
    {
      "epoch": 5.88753177506355,
      "grad_norm": 0.8780081272125244,
      "learning_rate": 7.24662250093248e-05,
      "loss": 1.4823,
      "step": 23740
    },
    {
      "epoch": 5.89001178002356,
      "grad_norm": 0.962202250957489,
      "learning_rate": 7.239132855283777e-05,
      "loss": 1.557,
      "step": 23750
    },
    {
      "epoch": 5.89249178498357,
      "grad_norm": 1.0483089685440063,
      "learning_rate": 7.231644885751507e-05,
      "loss": 1.4819,
      "step": 23760
    },
    {
      "epoch": 5.89497178994358,
      "grad_norm": 0.9742222428321838,
      "learning_rate": 7.224158596881606e-05,
      "loss": 1.5385,
      "step": 23770
    },
    {
      "epoch": 5.89745179490359,
      "grad_norm": 1.055155634880066,
      "learning_rate": 7.216673993218979e-05,
      "loss": 1.5791,
      "step": 23780
    },
    {
      "epoch": 5.8999317998636,
      "grad_norm": 1.012840986251831,
      "learning_rate": 7.209191079307521e-05,
      "loss": 1.5722,
      "step": 23790
    },
    {
      "epoch": 5.902411804823609,
      "grad_norm": 1.5436009168624878,
      "learning_rate": 7.20170985969008e-05,
      "loss": 1.5834,
      "step": 23800
    },
    {
      "epoch": 5.90489180978362,
      "grad_norm": 0.9777030944824219,
      "learning_rate": 7.194230338908501e-05,
      "loss": 1.5576,
      "step": 23810
    },
    {
      "epoch": 5.907371814743629,
      "grad_norm": 1.1380470991134644,
      "learning_rate": 7.186752521503575e-05,
      "loss": 1.6199,
      "step": 23820
    },
    {
      "epoch": 5.90985181970364,
      "grad_norm": 0.9923495650291443,
      "learning_rate": 7.179276412015075e-05,
      "loss": 1.5589,
      "step": 23830
    },
    {
      "epoch": 5.912331824663649,
      "grad_norm": 1.0055482387542725,
      "learning_rate": 7.171802014981726e-05,
      "loss": 1.5621,
      "step": 23840
    },
    {
      "epoch": 5.914811829623659,
      "grad_norm": 0.960023820400238,
      "learning_rate": 7.164329334941224e-05,
      "loss": 1.545,
      "step": 23850
    },
    {
      "epoch": 5.917291834583669,
      "grad_norm": 0.9554881453514099,
      "learning_rate": 7.15685837643021e-05,
      "loss": 1.4998,
      "step": 23860
    },
    {
      "epoch": 5.919771839543679,
      "grad_norm": 0.9166724681854248,
      "learning_rate": 7.149389143984295e-05,
      "loss": 1.5257,
      "step": 23870
    },
    {
      "epoch": 5.922251844503689,
      "grad_norm": 0.9697259664535522,
      "learning_rate": 7.141921642138025e-05,
      "loss": 1.5697,
      "step": 23880
    },
    {
      "epoch": 5.924731849463699,
      "grad_norm": 0.9463170766830444,
      "learning_rate": 7.13445587542491e-05,
      "loss": 1.447,
      "step": 23890
    },
    {
      "epoch": 5.9272118544237085,
      "grad_norm": 0.9994683861732483,
      "learning_rate": 7.126991848377395e-05,
      "loss": 1.5664,
      "step": 23900
    },
    {
      "epoch": 5.929691859383719,
      "grad_norm": 0.9542251229286194,
      "learning_rate": 7.119529565526883e-05,
      "loss": 1.5363,
      "step": 23910
    },
    {
      "epoch": 5.9321718643437285,
      "grad_norm": 0.8708572387695312,
      "learning_rate": 7.112069031403704e-05,
      "loss": 1.5512,
      "step": 23920
    },
    {
      "epoch": 5.934651869303739,
      "grad_norm": 0.9620928168296814,
      "learning_rate": 7.104610250537133e-05,
      "loss": 1.5337,
      "step": 23930
    },
    {
      "epoch": 5.937131874263748,
      "grad_norm": 0.9948837161064148,
      "learning_rate": 7.097153227455379e-05,
      "loss": 1.5175,
      "step": 23940
    },
    {
      "epoch": 5.939611879223758,
      "grad_norm": 0.9141833782196045,
      "learning_rate": 7.089697966685581e-05,
      "loss": 1.5532,
      "step": 23950
    },
    {
      "epoch": 5.942091884183768,
      "grad_norm": 0.9160381555557251,
      "learning_rate": 7.082244472753822e-05,
      "loss": 1.5984,
      "step": 23960
    },
    {
      "epoch": 5.944571889143778,
      "grad_norm": 1.0294259786605835,
      "learning_rate": 7.074792750185094e-05,
      "loss": 1.5502,
      "step": 23970
    },
    {
      "epoch": 5.947051894103788,
      "grad_norm": 1.0058735609054565,
      "learning_rate": 7.067342803503323e-05,
      "loss": 1.5325,
      "step": 23980
    },
    {
      "epoch": 5.949531899063798,
      "grad_norm": 0.9840842485427856,
      "learning_rate": 7.059894637231354e-05,
      "loss": 1.5092,
      "step": 23990
    },
    {
      "epoch": 5.952011904023808,
      "grad_norm": 0.9545488953590393,
      "learning_rate": 7.052448255890957e-05,
      "loss": 1.5778,
      "step": 24000
    },
    {
      "epoch": 5.954491908983818,
      "grad_norm": 0.9650373458862305,
      "learning_rate": 7.045003664002809e-05,
      "loss": 1.497,
      "step": 24010
    },
    {
      "epoch": 5.9569719139438275,
      "grad_norm": 0.971540629863739,
      "learning_rate": 7.037560866086513e-05,
      "loss": 1.5427,
      "step": 24020
    },
    {
      "epoch": 5.959451918903838,
      "grad_norm": 0.9746721982955933,
      "learning_rate": 7.030119866660564e-05,
      "loss": 1.5155,
      "step": 24030
    },
    {
      "epoch": 5.9619319238638475,
      "grad_norm": 0.9826546907424927,
      "learning_rate": 7.022680670242387e-05,
      "loss": 1.5574,
      "step": 24040
    },
    {
      "epoch": 5.964411928823858,
      "grad_norm": 1.0468549728393555,
      "learning_rate": 7.015243281348293e-05,
      "loss": 1.5316,
      "step": 24050
    },
    {
      "epoch": 5.9668919337838675,
      "grad_norm": 1.0100024938583374,
      "learning_rate": 7.007807704493514e-05,
      "loss": 1.5668,
      "step": 24060
    },
    {
      "epoch": 5.969371938743878,
      "grad_norm": 0.9497383832931519,
      "learning_rate": 7.000373944192165e-05,
      "loss": 1.4998,
      "step": 24070
    },
    {
      "epoch": 5.9718519437038875,
      "grad_norm": 0.9240361452102661,
      "learning_rate": 6.992942004957271e-05,
      "loss": 1.5402,
      "step": 24080
    },
    {
      "epoch": 5.974331948663897,
      "grad_norm": 1.0378366708755493,
      "learning_rate": 6.985511891300737e-05,
      "loss": 1.5714,
      "step": 24090
    },
    {
      "epoch": 5.9768119536239075,
      "grad_norm": 1.0124924182891846,
      "learning_rate": 6.978083607733384e-05,
      "loss": 1.5335,
      "step": 24100
    },
    {
      "epoch": 5.979291958583917,
      "grad_norm": 0.9290186166763306,
      "learning_rate": 6.970657158764894e-05,
      "loss": 1.4215,
      "step": 24110
    },
    {
      "epoch": 5.9817719635439275,
      "grad_norm": 0.8907943964004517,
      "learning_rate": 6.963232548903853e-05,
      "loss": 1.5144,
      "step": 24120
    },
    {
      "epoch": 5.984251968503937,
      "grad_norm": 0.9668370485305786,
      "learning_rate": 6.955809782657724e-05,
      "loss": 1.5497,
      "step": 24130
    },
    {
      "epoch": 5.986731973463947,
      "grad_norm": 0.956092119216919,
      "learning_rate": 6.948388864532857e-05,
      "loss": 1.6132,
      "step": 24140
    },
    {
      "epoch": 5.989211978423957,
      "grad_norm": 1.0998320579528809,
      "learning_rate": 6.940969799034465e-05,
      "loss": 1.5673,
      "step": 24150
    },
    {
      "epoch": 5.991691983383967,
      "grad_norm": 1.0120083093643188,
      "learning_rate": 6.933552590666659e-05,
      "loss": 1.5247,
      "step": 24160
    },
    {
      "epoch": 5.994171988343977,
      "grad_norm": 0.9429424405097961,
      "learning_rate": 6.926137243932398e-05,
      "loss": 1.5167,
      "step": 24170
    },
    {
      "epoch": 5.996651993303987,
      "grad_norm": 0.9629198908805847,
      "learning_rate": 6.918723763333532e-05,
      "loss": 1.4815,
      "step": 24180
    },
    {
      "epoch": 5.999131998263996,
      "grad_norm": 1.0192433595657349,
      "learning_rate": 6.911312153370765e-05,
      "loss": 1.4817,
      "step": 24190
    },
    {
      "epoch": 6.001612003224007,
      "grad_norm": 0.9021772742271423,
      "learning_rate": 6.903902418543671e-05,
      "loss": 1.5574,
      "step": 24200
    },
    {
      "epoch": 6.004092008184016,
      "grad_norm": 0.9498673677444458,
      "learning_rate": 6.89649456335068e-05,
      "loss": 1.4026,
      "step": 24210
    },
    {
      "epoch": 6.006572013144027,
      "grad_norm": 1.02396559715271,
      "learning_rate": 6.889088592289093e-05,
      "loss": 1.4367,
      "step": 24220
    },
    {
      "epoch": 6.009052018104036,
      "grad_norm": 0.9916176199913025,
      "learning_rate": 6.881684509855049e-05,
      "loss": 1.5412,
      "step": 24230
    },
    {
      "epoch": 6.0115320230640465,
      "grad_norm": 0.9803375005722046,
      "learning_rate": 6.874282320543557e-05,
      "loss": 1.4064,
      "step": 24240
    },
    {
      "epoch": 6.014012028024056,
      "grad_norm": 1.0052950382232666,
      "learning_rate": 6.866882028848465e-05,
      "loss": 1.5316,
      "step": 24250
    },
    {
      "epoch": 6.016492032984066,
      "grad_norm": 0.9743315577507019,
      "learning_rate": 6.859483639262481e-05,
      "loss": 1.4378,
      "step": 24260
    },
    {
      "epoch": 6.018972037944076,
      "grad_norm": 0.9998024106025696,
      "learning_rate": 6.852087156277143e-05,
      "loss": 1.5104,
      "step": 24270
    },
    {
      "epoch": 6.021452042904086,
      "grad_norm": 1.0471398830413818,
      "learning_rate": 6.844692584382848e-05,
      "loss": 1.5089,
      "step": 24280
    },
    {
      "epoch": 6.023932047864096,
      "grad_norm": 1.002640724182129,
      "learning_rate": 6.837299928068817e-05,
      "loss": 1.4579,
      "step": 24290
    },
    {
      "epoch": 6.026412052824106,
      "grad_norm": 0.9999975562095642,
      "learning_rate": 6.829909191823121e-05,
      "loss": 1.4601,
      "step": 24300
    },
    {
      "epoch": 6.028892057784115,
      "grad_norm": 0.9147332906723022,
      "learning_rate": 6.822520380132655e-05,
      "loss": 1.4445,
      "step": 24310
    },
    {
      "epoch": 6.031372062744126,
      "grad_norm": 0.9814079999923706,
      "learning_rate": 6.815133497483157e-05,
      "loss": 1.4039,
      "step": 24320
    },
    {
      "epoch": 6.033852067704135,
      "grad_norm": 0.9648290872573853,
      "learning_rate": 6.807748548359181e-05,
      "loss": 1.4357,
      "step": 24330
    },
    {
      "epoch": 6.036332072664146,
      "grad_norm": 1.0147061347961426,
      "learning_rate": 6.80036553724412e-05,
      "loss": 1.4742,
      "step": 24340
    },
    {
      "epoch": 6.038812077624155,
      "grad_norm": 0.9684031009674072,
      "learning_rate": 6.792984468620174e-05,
      "loss": 1.4793,
      "step": 24350
    },
    {
      "epoch": 6.041292082584166,
      "grad_norm": 1.0336523056030273,
      "learning_rate": 6.785605346968386e-05,
      "loss": 1.483,
      "step": 24360
    },
    {
      "epoch": 6.043772087544175,
      "grad_norm": 0.947838306427002,
      "learning_rate": 6.778228176768595e-05,
      "loss": 1.4407,
      "step": 24370
    },
    {
      "epoch": 6.046252092504185,
      "grad_norm": 0.9754374623298645,
      "learning_rate": 6.770852962499471e-05,
      "loss": 1.4824,
      "step": 24380
    },
    {
      "epoch": 6.048732097464195,
      "grad_norm": 1.0519452095031738,
      "learning_rate": 6.763479708638485e-05,
      "loss": 1.5115,
      "step": 24390
    },
    {
      "epoch": 6.051212102424205,
      "grad_norm": 0.9581342339515686,
      "learning_rate": 6.756108419661931e-05,
      "loss": 1.4636,
      "step": 24400
    },
    {
      "epoch": 6.053692107384215,
      "grad_norm": 1.0844944715499878,
      "learning_rate": 6.748739100044892e-05,
      "loss": 1.5126,
      "step": 24410
    },
    {
      "epoch": 6.056172112344225,
      "grad_norm": 0.9884443283081055,
      "learning_rate": 6.741371754261278e-05,
      "loss": 1.4207,
      "step": 24420
    },
    {
      "epoch": 6.058652117304234,
      "grad_norm": 0.9929337501525879,
      "learning_rate": 6.73400638678378e-05,
      "loss": 1.4127,
      "step": 24430
    },
    {
      "epoch": 6.061132122264245,
      "grad_norm": 1.0066174268722534,
      "learning_rate": 6.7266430020839e-05,
      "loss": 1.4225,
      "step": 24440
    },
    {
      "epoch": 6.063612127224254,
      "grad_norm": 1.0377559661865234,
      "learning_rate": 6.71928160463193e-05,
      "loss": 1.4664,
      "step": 24450
    },
    {
      "epoch": 6.066092132184265,
      "grad_norm": 0.9668002128601074,
      "learning_rate": 6.711922198896962e-05,
      "loss": 1.4753,
      "step": 24460
    },
    {
      "epoch": 6.068572137144274,
      "grad_norm": 1.0058457851409912,
      "learning_rate": 6.704564789346873e-05,
      "loss": 1.5021,
      "step": 24470
    },
    {
      "epoch": 6.071052142104284,
      "grad_norm": 0.9895146489143372,
      "learning_rate": 6.697209380448333e-05,
      "loss": 1.4896,
      "step": 24480
    },
    {
      "epoch": 6.073532147064294,
      "grad_norm": 1.07613205909729,
      "learning_rate": 6.689855976666788e-05,
      "loss": 1.4871,
      "step": 24490
    },
    {
      "epoch": 6.076012152024304,
      "grad_norm": 1.0612536668777466,
      "learning_rate": 6.682504582466482e-05,
      "loss": 1.5251,
      "step": 24500
    },
    {
      "epoch": 6.078492156984314,
      "grad_norm": 0.9371278285980225,
      "learning_rate": 6.675155202310424e-05,
      "loss": 1.4509,
      "step": 24510
    },
    {
      "epoch": 6.080972161944324,
      "grad_norm": 0.9730448126792908,
      "learning_rate": 6.66780784066041e-05,
      "loss": 1.4651,
      "step": 24520
    },
    {
      "epoch": 6.083452166904334,
      "grad_norm": 1.00874662399292,
      "learning_rate": 6.660462501977004e-05,
      "loss": 1.4882,
      "step": 24530
    },
    {
      "epoch": 6.085932171864344,
      "grad_norm": 1.0154139995574951,
      "learning_rate": 6.653119190719554e-05,
      "loss": 1.4462,
      "step": 24540
    },
    {
      "epoch": 6.088412176824353,
      "grad_norm": 1.04769766330719,
      "learning_rate": 6.645777911346155e-05,
      "loss": 1.435,
      "step": 24550
    },
    {
      "epoch": 6.090892181784364,
      "grad_norm": 1.0066906213760376,
      "learning_rate": 6.638438668313695e-05,
      "loss": 1.4698,
      "step": 24560
    },
    {
      "epoch": 6.093372186744373,
      "grad_norm": 1.0486011505126953,
      "learning_rate": 6.6311014660778e-05,
      "loss": 1.4811,
      "step": 24570
    },
    {
      "epoch": 6.095852191704384,
      "grad_norm": 1.110256552696228,
      "learning_rate": 6.623766309092879e-05,
      "loss": 1.4431,
      "step": 24580
    },
    {
      "epoch": 6.098332196664393,
      "grad_norm": 1.0279182195663452,
      "learning_rate": 6.616433201812085e-05,
      "loss": 1.4469,
      "step": 24590
    },
    {
      "epoch": 6.100812201624403,
      "grad_norm": 1.0100728273391724,
      "learning_rate": 6.609102148687333e-05,
      "loss": 1.4593,
      "step": 24600
    },
    {
      "epoch": 6.103292206584413,
      "grad_norm": 0.9717130661010742,
      "learning_rate": 6.60177315416929e-05,
      "loss": 1.4699,
      "step": 24610
    },
    {
      "epoch": 6.105772211544423,
      "grad_norm": 1.0001333951950073,
      "learning_rate": 6.59444622270737e-05,
      "loss": 1.4175,
      "step": 24620
    },
    {
      "epoch": 6.108252216504433,
      "grad_norm": 0.9675756692886353,
      "learning_rate": 6.587121358749745e-05,
      "loss": 1.4331,
      "step": 24630
    },
    {
      "epoch": 6.110732221464443,
      "grad_norm": 0.9195937514305115,
      "learning_rate": 6.579798566743314e-05,
      "loss": 1.483,
      "step": 24640
    },
    {
      "epoch": 6.113212226424453,
      "grad_norm": 0.9295458197593689,
      "learning_rate": 6.572477851133733e-05,
      "loss": 1.4546,
      "step": 24650
    },
    {
      "epoch": 6.115692231384463,
      "grad_norm": 1.1020594835281372,
      "learning_rate": 6.565159216365389e-05,
      "loss": 1.5096,
      "step": 24660
    },
    {
      "epoch": 6.118172236344472,
      "grad_norm": 0.9957802295684814,
      "learning_rate": 6.557842666881416e-05,
      "loss": 1.4511,
      "step": 24670
    },
    {
      "epoch": 6.120652241304483,
      "grad_norm": 1.0554035902023315,
      "learning_rate": 6.550528207123668e-05,
      "loss": 1.4479,
      "step": 24680
    },
    {
      "epoch": 6.123132246264492,
      "grad_norm": 0.9450856447219849,
      "learning_rate": 6.543215841532744e-05,
      "loss": 1.3555,
      "step": 24690
    },
    {
      "epoch": 6.125612251224503,
      "grad_norm": 0.9754582643508911,
      "learning_rate": 6.535905574547956e-05,
      "loss": 1.5365,
      "step": 24700
    },
    {
      "epoch": 6.128092256184512,
      "grad_norm": 0.9962130784988403,
      "learning_rate": 6.528597410607364e-05,
      "loss": 1.5361,
      "step": 24710
    },
    {
      "epoch": 6.130572261144522,
      "grad_norm": 1.0566751956939697,
      "learning_rate": 6.521291354147727e-05,
      "loss": 1.4661,
      "step": 24720
    },
    {
      "epoch": 6.133052266104532,
      "grad_norm": 1.0275439023971558,
      "learning_rate": 6.513987409604545e-05,
      "loss": 1.42,
      "step": 24730
    },
    {
      "epoch": 6.135532271064542,
      "grad_norm": 1.0249825716018677,
      "learning_rate": 6.506685581412018e-05,
      "loss": 1.4582,
      "step": 24740
    },
    {
      "epoch": 6.138012276024552,
      "grad_norm": 0.9934401512145996,
      "learning_rate": 6.499385874003076e-05,
      "loss": 1.436,
      "step": 24750
    },
    {
      "epoch": 6.140492280984562,
      "grad_norm": 0.9783113598823547,
      "learning_rate": 6.492088291809354e-05,
      "loss": 1.5651,
      "step": 24760
    },
    {
      "epoch": 6.1429722859445715,
      "grad_norm": 0.9646732807159424,
      "learning_rate": 6.484792839261203e-05,
      "loss": 1.4806,
      "step": 24770
    },
    {
      "epoch": 6.145452290904582,
      "grad_norm": 0.9748082756996155,
      "learning_rate": 6.477499520787665e-05,
      "loss": 1.4407,
      "step": 24780
    },
    {
      "epoch": 6.1479322958645914,
      "grad_norm": 0.9589222073554993,
      "learning_rate": 6.470208340816508e-05,
      "loss": 1.4854,
      "step": 24790
    },
    {
      "epoch": 6.150412300824602,
      "grad_norm": 1.0084720849990845,
      "learning_rate": 6.462919303774186e-05,
      "loss": 1.4739,
      "step": 24800
    },
    {
      "epoch": 6.152892305784611,
      "grad_norm": 1.0057191848754883,
      "learning_rate": 6.455632414085861e-05,
      "loss": 1.4447,
      "step": 24810
    },
    {
      "epoch": 6.155372310744622,
      "grad_norm": 0.9664227366447449,
      "learning_rate": 6.448347676175384e-05,
      "loss": 1.5124,
      "step": 24820
    },
    {
      "epoch": 6.157852315704631,
      "grad_norm": 1.0164328813552856,
      "learning_rate": 6.441065094465311e-05,
      "loss": 1.5045,
      "step": 24830
    },
    {
      "epoch": 6.160332320664641,
      "grad_norm": 1.020713448524475,
      "learning_rate": 6.43378467337687e-05,
      "loss": 1.4937,
      "step": 24840
    },
    {
      "epoch": 6.162812325624651,
      "grad_norm": 0.9904918074607849,
      "learning_rate": 6.42650641733e-05,
      "loss": 1.4852,
      "step": 24850
    },
    {
      "epoch": 6.165292330584661,
      "grad_norm": 1.080602765083313,
      "learning_rate": 6.419230330743305e-05,
      "loss": 1.4141,
      "step": 24860
    },
    {
      "epoch": 6.167772335544671,
      "grad_norm": 1.0138860940933228,
      "learning_rate": 6.41195641803409e-05,
      "loss": 1.5157,
      "step": 24870
    },
    {
      "epoch": 6.170252340504681,
      "grad_norm": 0.9689585566520691,
      "learning_rate": 6.404684683618325e-05,
      "loss": 1.5049,
      "step": 24880
    },
    {
      "epoch": 6.1727323454646905,
      "grad_norm": 1.0130665302276611,
      "learning_rate": 6.397415131910671e-05,
      "loss": 1.4695,
      "step": 24890
    },
    {
      "epoch": 6.175212350424701,
      "grad_norm": 0.97395259141922,
      "learning_rate": 6.390147767324445e-05,
      "loss": 1.4501,
      "step": 24900
    },
    {
      "epoch": 6.1776923553847105,
      "grad_norm": 0.9632650017738342,
      "learning_rate": 6.382882594271663e-05,
      "loss": 1.4742,
      "step": 24910
    },
    {
      "epoch": 6.180172360344721,
      "grad_norm": 1.001616358757019,
      "learning_rate": 6.375619617162985e-05,
      "loss": 1.4871,
      "step": 24920
    },
    {
      "epoch": 6.1826523653047305,
      "grad_norm": 1.0696035623550415,
      "learning_rate": 6.368358840407753e-05,
      "loss": 1.5157,
      "step": 24930
    },
    {
      "epoch": 6.185132370264741,
      "grad_norm": 0.9896767735481262,
      "learning_rate": 6.361100268413967e-05,
      "loss": 1.3934,
      "step": 24940
    },
    {
      "epoch": 6.1876123752247505,
      "grad_norm": 0.9482393264770508,
      "learning_rate": 6.353843905588295e-05,
      "loss": 1.467,
      "step": 24950
    },
    {
      "epoch": 6.19009238018476,
      "grad_norm": 0.9930483102798462,
      "learning_rate": 6.34658975633605e-05,
      "loss": 1.4932,
      "step": 24960
    },
    {
      "epoch": 6.1925723851447705,
      "grad_norm": 1.052764654159546,
      "learning_rate": 6.33933782506122e-05,
      "loss": 1.4872,
      "step": 24970
    },
    {
      "epoch": 6.19505239010478,
      "grad_norm": 1.0381749868392944,
      "learning_rate": 6.33208811616643e-05,
      "loss": 1.4503,
      "step": 24980
    },
    {
      "epoch": 6.1975323950647905,
      "grad_norm": 1.0356924533843994,
      "learning_rate": 6.324840634052967e-05,
      "loss": 1.4349,
      "step": 24990
    },
    {
      "epoch": 6.2000124000248,
      "grad_norm": 0.9789268970489502,
      "learning_rate": 6.317595383120756e-05,
      "loss": 1.5251,
      "step": 25000
    },
    {
      "epoch": 6.20249240498481,
      "grad_norm": 0.9749928712844849,
      "learning_rate": 6.31035236776838e-05,
      "loss": 1.4467,
      "step": 25010
    },
    {
      "epoch": 6.20497240994482,
      "grad_norm": 0.978130042552948,
      "learning_rate": 6.30311159239305e-05,
      "loss": 1.5107,
      "step": 25020
    },
    {
      "epoch": 6.20745241490483,
      "grad_norm": 1.008919358253479,
      "learning_rate": 6.295873061390636e-05,
      "loss": 1.4652,
      "step": 25030
    },
    {
      "epoch": 6.20993241986484,
      "grad_norm": 0.9471341967582703,
      "learning_rate": 6.288636779155621e-05,
      "loss": 1.4217,
      "step": 25040
    },
    {
      "epoch": 6.21241242482485,
      "grad_norm": 1.0002670288085938,
      "learning_rate": 6.281402750081147e-05,
      "loss": 1.5057,
      "step": 25050
    },
    {
      "epoch": 6.214892429784859,
      "grad_norm": 0.9456087946891785,
      "learning_rate": 6.27417097855897e-05,
      "loss": 1.4608,
      "step": 25060
    },
    {
      "epoch": 6.21737243474487,
      "grad_norm": 0.9659121632575989,
      "learning_rate": 6.26694146897949e-05,
      "loss": 1.4419,
      "step": 25070
    },
    {
      "epoch": 6.219852439704879,
      "grad_norm": 0.9993132948875427,
      "learning_rate": 6.259714225731718e-05,
      "loss": 1.4938,
      "step": 25080
    },
    {
      "epoch": 6.2223324446648896,
      "grad_norm": 0.9439984560012817,
      "learning_rate": 6.252489253203309e-05,
      "loss": 1.4579,
      "step": 25090
    },
    {
      "epoch": 6.224812449624899,
      "grad_norm": 1.018552303314209,
      "learning_rate": 6.245266555780513e-05,
      "loss": 1.4627,
      "step": 25100
    },
    {
      "epoch": 6.2272924545849095,
      "grad_norm": 0.9674966931343079,
      "learning_rate": 6.23804613784823e-05,
      "loss": 1.4592,
      "step": 25110
    },
    {
      "epoch": 6.229772459544919,
      "grad_norm": 1.0720467567443848,
      "learning_rate": 6.230828003789949e-05,
      "loss": 1.518,
      "step": 25120
    },
    {
      "epoch": 6.232252464504929,
      "grad_norm": 0.9193775653839111,
      "learning_rate": 6.223612157987786e-05,
      "loss": 1.4278,
      "step": 25130
    },
    {
      "epoch": 6.234732469464939,
      "grad_norm": 1.0076197385787964,
      "learning_rate": 6.216398604822466e-05,
      "loss": 1.4888,
      "step": 25140
    },
    {
      "epoch": 6.237212474424949,
      "grad_norm": 0.9544000625610352,
      "learning_rate": 6.209187348673325e-05,
      "loss": 1.4576,
      "step": 25150
    },
    {
      "epoch": 6.239692479384959,
      "grad_norm": 0.9171335101127625,
      "learning_rate": 6.20197839391829e-05,
      "loss": 1.4899,
      "step": 25160
    },
    {
      "epoch": 6.242172484344969,
      "grad_norm": 1.0323432683944702,
      "learning_rate": 6.194771744933916e-05,
      "loss": 1.5335,
      "step": 25170
    },
    {
      "epoch": 6.244652489304978,
      "grad_norm": 0.9690384864807129,
      "learning_rate": 6.187567406095333e-05,
      "loss": 1.4252,
      "step": 25180
    },
    {
      "epoch": 6.247132494264989,
      "grad_norm": 1.0386059284210205,
      "learning_rate": 6.180365381776285e-05,
      "loss": 1.5064,
      "step": 25190
    },
    {
      "epoch": 6.249612499224998,
      "grad_norm": 0.9646037220954895,
      "learning_rate": 6.173165676349103e-05,
      "loss": 1.3969,
      "step": 25200
    },
    {
      "epoch": 6.252092504185009,
      "grad_norm": 1.0465009212493896,
      "learning_rate": 6.165968294184713e-05,
      "loss": 1.4816,
      "step": 25210
    },
    {
      "epoch": 6.254572509145018,
      "grad_norm": 1.0445176362991333,
      "learning_rate": 6.158773239652633e-05,
      "loss": 1.4505,
      "step": 25220
    },
    {
      "epoch": 6.257052514105029,
      "grad_norm": 0.9644328951835632,
      "learning_rate": 6.151580517120958e-05,
      "loss": 1.5001,
      "step": 25230
    },
    {
      "epoch": 6.259532519065038,
      "grad_norm": 1.0155954360961914,
      "learning_rate": 6.144390130956384e-05,
      "loss": 1.4321,
      "step": 25240
    },
    {
      "epoch": 6.262012524025048,
      "grad_norm": 0.985386848449707,
      "learning_rate": 6.137202085524167e-05,
      "loss": 1.5258,
      "step": 25250
    },
    {
      "epoch": 6.264492528985058,
      "grad_norm": 0.9713849425315857,
      "learning_rate": 6.130016385188166e-05,
      "loss": 1.5211,
      "step": 25260
    },
    {
      "epoch": 6.266972533945068,
      "grad_norm": 1.054305911064148,
      "learning_rate": 6.122833034310793e-05,
      "loss": 1.4022,
      "step": 25270
    },
    {
      "epoch": 6.269452538905078,
      "grad_norm": 0.8938084244728088,
      "learning_rate": 6.115652037253053e-05,
      "loss": 1.395,
      "step": 25280
    },
    {
      "epoch": 6.271932543865088,
      "grad_norm": 0.9827483892440796,
      "learning_rate": 6.10847339837451e-05,
      "loss": 1.5041,
      "step": 25290
    },
    {
      "epoch": 6.274412548825097,
      "grad_norm": 1.0114959478378296,
      "learning_rate": 6.1012971220333016e-05,
      "loss": 1.4406,
      "step": 25300
    },
    {
      "epoch": 6.276892553785108,
      "grad_norm": 0.9963436126708984,
      "learning_rate": 6.0941232125861235e-05,
      "loss": 1.4858,
      "step": 25310
    },
    {
      "epoch": 6.279372558745117,
      "grad_norm": 0.9555656313896179,
      "learning_rate": 6.0869516743882516e-05,
      "loss": 1.4382,
      "step": 25320
    },
    {
      "epoch": 6.281852563705128,
      "grad_norm": 0.9409336447715759,
      "learning_rate": 6.079782511793502e-05,
      "loss": 1.4517,
      "step": 25330
    },
    {
      "epoch": 6.284332568665137,
      "grad_norm": 0.9207855463027954,
      "learning_rate": 6.0726157291542605e-05,
      "loss": 1.4633,
      "step": 25340
    },
    {
      "epoch": 6.286812573625147,
      "grad_norm": 1.043115496635437,
      "learning_rate": 6.0654513308214645e-05,
      "loss": 1.4632,
      "step": 25350
    },
    {
      "epoch": 6.289292578585157,
      "grad_norm": 0.9693536758422852,
      "learning_rate": 6.058289321144608e-05,
      "loss": 1.4656,
      "step": 25360
    },
    {
      "epoch": 6.291772583545167,
      "grad_norm": 1.041846752166748,
      "learning_rate": 6.0511297044717264e-05,
      "loss": 1.5087,
      "step": 25370
    },
    {
      "epoch": 6.294252588505177,
      "grad_norm": 0.9643771052360535,
      "learning_rate": 6.043972485149414e-05,
      "loss": 1.4855,
      "step": 25380
    },
    {
      "epoch": 6.296732593465187,
      "grad_norm": 1.0329574346542358,
      "learning_rate": 6.0368176675227964e-05,
      "loss": 1.4842,
      "step": 25390
    },
    {
      "epoch": 6.299212598425197,
      "grad_norm": 1.0316472053527832,
      "learning_rate": 6.02966525593555e-05,
      "loss": 1.5037,
      "step": 25400
    },
    {
      "epoch": 6.301692603385207,
      "grad_norm": 0.9890727400779724,
      "learning_rate": 6.02251525472989e-05,
      "loss": 1.4633,
      "step": 25410
    },
    {
      "epoch": 6.304172608345216,
      "grad_norm": 0.9751850366592407,
      "learning_rate": 6.0153676682465654e-05,
      "loss": 1.4507,
      "step": 25420
    },
    {
      "epoch": 6.306652613305227,
      "grad_norm": 0.8939128518104553,
      "learning_rate": 6.00822250082486e-05,
      "loss": 1.4735,
      "step": 25430
    },
    {
      "epoch": 6.309132618265236,
      "grad_norm": 0.9928026795387268,
      "learning_rate": 6.001079756802592e-05,
      "loss": 1.4833,
      "step": 25440
    },
    {
      "epoch": 6.311612623225247,
      "grad_norm": 1.0220106840133667,
      "learning_rate": 5.9939394405160986e-05,
      "loss": 1.4725,
      "step": 25450
    },
    {
      "epoch": 6.314092628185256,
      "grad_norm": 1.026551604270935,
      "learning_rate": 5.9868015563002625e-05,
      "loss": 1.4409,
      "step": 25460
    },
    {
      "epoch": 6.316572633145267,
      "grad_norm": 0.993131697177887,
      "learning_rate": 5.9796661084884644e-05,
      "loss": 1.5148,
      "step": 25470
    },
    {
      "epoch": 6.319052638105276,
      "grad_norm": 1.0696313381195068,
      "learning_rate": 5.9725331014126294e-05,
      "loss": 1.477,
      "step": 25480
    },
    {
      "epoch": 6.321532643065286,
      "grad_norm": 0.9706017374992371,
      "learning_rate": 5.965402539403182e-05,
      "loss": 1.5219,
      "step": 25490
    },
    {
      "epoch": 6.324012648025296,
      "grad_norm": 1.0308212041854858,
      "learning_rate": 5.9582744267890814e-05,
      "loss": 1.5055,
      "step": 25500
    },
    {
      "epoch": 6.326492652985306,
      "grad_norm": 0.9662330746650696,
      "learning_rate": 5.951148767897776e-05,
      "loss": 1.4692,
      "step": 25510
    },
    {
      "epoch": 6.328972657945316,
      "grad_norm": 1.0245823860168457,
      "learning_rate": 5.944025567055251e-05,
      "loss": 1.4989,
      "step": 25520
    },
    {
      "epoch": 6.331452662905326,
      "grad_norm": 0.990571141242981,
      "learning_rate": 5.936904828585976e-05,
      "loss": 1.4467,
      "step": 25530
    },
    {
      "epoch": 6.333932667865335,
      "grad_norm": 1.0541971921920776,
      "learning_rate": 5.929786556812943e-05,
      "loss": 1.4546,
      "step": 25540
    },
    {
      "epoch": 6.336412672825346,
      "grad_norm": 0.9833094477653503,
      "learning_rate": 5.922670756057633e-05,
      "loss": 1.4629,
      "step": 25550
    },
    {
      "epoch": 6.338892677785355,
      "grad_norm": 0.9783018231391907,
      "learning_rate": 5.9155574306400395e-05,
      "loss": 1.4717,
      "step": 25560
    },
    {
      "epoch": 6.341372682745366,
      "grad_norm": 0.9881133437156677,
      "learning_rate": 5.908446584878642e-05,
      "loss": 1.4355,
      "step": 25570
    },
    {
      "epoch": 6.343852687705375,
      "grad_norm": 1.0198750495910645,
      "learning_rate": 5.901338223090425e-05,
      "loss": 1.4109,
      "step": 25580
    },
    {
      "epoch": 6.346332692665385,
      "grad_norm": 1.0592362880706787,
      "learning_rate": 5.89423234959085e-05,
      "loss": 1.499,
      "step": 25590
    },
    {
      "epoch": 6.348812697625395,
      "grad_norm": 1.0093663930892944,
      "learning_rate": 5.887128968693887e-05,
      "loss": 1.4483,
      "step": 25600
    },
    {
      "epoch": 6.351292702585405,
      "grad_norm": 1.007263422012329,
      "learning_rate": 5.880028084711976e-05,
      "loss": 1.4667,
      "step": 25610
    },
    {
      "epoch": 6.353772707545415,
      "grad_norm": 1.0381349325180054,
      "learning_rate": 5.872929701956054e-05,
      "loss": 1.5031,
      "step": 25620
    },
    {
      "epoch": 6.356252712505425,
      "grad_norm": 1.0256422758102417,
      "learning_rate": 5.8658338247355296e-05,
      "loss": 1.5182,
      "step": 25630
    },
    {
      "epoch": 6.358732717465435,
      "grad_norm": 1.0254257917404175,
      "learning_rate": 5.858740457358298e-05,
      "loss": 1.4856,
      "step": 25640
    },
    {
      "epoch": 6.361212722425445,
      "grad_norm": 0.9955101013183594,
      "learning_rate": 5.851649604130723e-05,
      "loss": 1.444,
      "step": 25650
    },
    {
      "epoch": 6.363692727385454,
      "grad_norm": 1.0298657417297363,
      "learning_rate": 5.8445612693576504e-05,
      "loss": 1.5016,
      "step": 25660
    },
    {
      "epoch": 6.366172732345465,
      "grad_norm": 1.0355310440063477,
      "learning_rate": 5.837475457342385e-05,
      "loss": 1.4489,
      "step": 25670
    },
    {
      "epoch": 6.368652737305474,
      "grad_norm": 0.996111273765564,
      "learning_rate": 5.830392172386723e-05,
      "loss": 1.3925,
      "step": 25680
    },
    {
      "epoch": 6.371132742265485,
      "grad_norm": 0.9839708805084229,
      "learning_rate": 5.8233114187908935e-05,
      "loss": 1.4616,
      "step": 25690
    },
    {
      "epoch": 6.373612747225494,
      "grad_norm": 1.061486840248108,
      "learning_rate": 5.816233200853618e-05,
      "loss": 1.4766,
      "step": 25700
    },
    {
      "epoch": 6.376092752185504,
      "grad_norm": 0.9838336110115051,
      "learning_rate": 5.809157522872065e-05,
      "loss": 1.4451,
      "step": 25710
    },
    {
      "epoch": 6.378572757145514,
      "grad_norm": 0.9915831089019775,
      "learning_rate": 5.802084389141862e-05,
      "loss": 1.5084,
      "step": 25720
    },
    {
      "epoch": 6.381052762105524,
      "grad_norm": 1.028008222579956,
      "learning_rate": 5.79501380395709e-05,
      "loss": 1.4982,
      "step": 25730
    },
    {
      "epoch": 6.383532767065534,
      "grad_norm": 1.0214972496032715,
      "learning_rate": 5.787945771610296e-05,
      "loss": 1.5264,
      "step": 25740
    },
    {
      "epoch": 6.386012772025544,
      "grad_norm": 1.0997072458267212,
      "learning_rate": 5.7808802963924524e-05,
      "loss": 1.4507,
      "step": 25750
    },
    {
      "epoch": 6.388492776985554,
      "grad_norm": 1.0471376180648804,
      "learning_rate": 5.773817382593008e-05,
      "loss": 1.4611,
      "step": 25760
    },
    {
      "epoch": 6.390972781945564,
      "grad_norm": 0.9893296957015991,
      "learning_rate": 5.7667570344998343e-05,
      "loss": 1.5209,
      "step": 25770
    },
    {
      "epoch": 6.3934527869055735,
      "grad_norm": 0.9793508052825928,
      "learning_rate": 5.759699256399255e-05,
      "loss": 1.5272,
      "step": 25780
    },
    {
      "epoch": 6.395932791865584,
      "grad_norm": 1.0266550779342651,
      "learning_rate": 5.7526440525760284e-05,
      "loss": 1.5027,
      "step": 25790
    },
    {
      "epoch": 6.3984127968255935,
      "grad_norm": 0.9966225624084473,
      "learning_rate": 5.7455914273133646e-05,
      "loss": 1.4147,
      "step": 25800
    },
    {
      "epoch": 6.400892801785604,
      "grad_norm": 0.9305916428565979,
      "learning_rate": 5.7385413848928814e-05,
      "loss": 1.4721,
      "step": 25810
    },
    {
      "epoch": 6.4033728067456135,
      "grad_norm": 0.9645832777023315,
      "learning_rate": 5.731493929594658e-05,
      "loss": 1.4901,
      "step": 25820
    },
    {
      "epoch": 6.405852811705623,
      "grad_norm": 1.1395509243011475,
      "learning_rate": 5.7244490656971815e-05,
      "loss": 1.5164,
      "step": 25830
    },
    {
      "epoch": 6.4083328166656335,
      "grad_norm": 1.0750949382781982,
      "learning_rate": 5.717406797477372e-05,
      "loss": 1.4958,
      "step": 25840
    },
    {
      "epoch": 6.410812821625643,
      "grad_norm": 0.987261950969696,
      "learning_rate": 5.710367129210583e-05,
      "loss": 1.4071,
      "step": 25850
    },
    {
      "epoch": 6.4132928265856535,
      "grad_norm": 1.0140446424484253,
      "learning_rate": 5.703330065170577e-05,
      "loss": 1.4673,
      "step": 25860
    },
    {
      "epoch": 6.415772831545663,
      "grad_norm": 0.9446235299110413,
      "learning_rate": 5.6962956096295404e-05,
      "loss": 1.512,
      "step": 25870
    },
    {
      "epoch": 6.418252836505673,
      "grad_norm": 1.0067250728607178,
      "learning_rate": 5.6892637668580716e-05,
      "loss": 1.5007,
      "step": 25880
    },
    {
      "epoch": 6.420732841465683,
      "grad_norm": 0.998441219329834,
      "learning_rate": 5.6822345411251995e-05,
      "loss": 1.4786,
      "step": 25890
    },
    {
      "epoch": 6.423212846425693,
      "grad_norm": 1.0318422317504883,
      "learning_rate": 5.675207936698337e-05,
      "loss": 1.4718,
      "step": 25900
    },
    {
      "epoch": 6.425692851385703,
      "grad_norm": 1.0681356191635132,
      "learning_rate": 5.6681839578433314e-05,
      "loss": 1.4734,
      "step": 25910
    },
    {
      "epoch": 6.428172856345713,
      "grad_norm": 1.0494177341461182,
      "learning_rate": 5.6611626088244194e-05,
      "loss": 1.4677,
      "step": 25920
    },
    {
      "epoch": 6.430652861305723,
      "grad_norm": 0.9486759901046753,
      "learning_rate": 5.654143893904249e-05,
      "loss": 1.4856,
      "step": 25930
    },
    {
      "epoch": 6.4331328662657326,
      "grad_norm": 1.0628020763397217,
      "learning_rate": 5.647127817343861e-05,
      "loss": 1.4124,
      "step": 25940
    },
    {
      "epoch": 6.435612871225742,
      "grad_norm": 0.9767960906028748,
      "learning_rate": 5.640114383402714e-05,
      "loss": 1.4999,
      "step": 25950
    },
    {
      "epoch": 6.4380928761857525,
      "grad_norm": 0.9298993945121765,
      "learning_rate": 5.6331035963386315e-05,
      "loss": 1.445,
      "step": 25960
    },
    {
      "epoch": 6.440572881145762,
      "grad_norm": 1.0534563064575195,
      "learning_rate": 5.6260954604078585e-05,
      "loss": 1.5237,
      "step": 25970
    },
    {
      "epoch": 6.4430528861057725,
      "grad_norm": 0.9476912617683411,
      "learning_rate": 5.6190899798650156e-05,
      "loss": 1.4964,
      "step": 25980
    },
    {
      "epoch": 6.445532891065782,
      "grad_norm": 1.005138635635376,
      "learning_rate": 5.6120871589631155e-05,
      "loss": 1.4278,
      "step": 25990
    },
    {
      "epoch": 6.448012896025792,
      "grad_norm": 1.0275954008102417,
      "learning_rate": 5.6050870019535494e-05,
      "loss": 1.4616,
      "step": 26000
    },
    {
      "epoch": 6.450492900985802,
      "grad_norm": 1.0758296251296997,
      "learning_rate": 5.598089513086108e-05,
      "loss": 1.4958,
      "step": 26010
    },
    {
      "epoch": 6.452972905945812,
      "grad_norm": 0.9819092154502869,
      "learning_rate": 5.591094696608946e-05,
      "loss": 1.5186,
      "step": 26020
    },
    {
      "epoch": 6.455452910905822,
      "grad_norm": 1.0485687255859375,
      "learning_rate": 5.5841025567686026e-05,
      "loss": 1.4086,
      "step": 26030
    },
    {
      "epoch": 6.457932915865832,
      "grad_norm": 1.0583301782608032,
      "learning_rate": 5.577113097809989e-05,
      "loss": 1.5041,
      "step": 26040
    },
    {
      "epoch": 6.460412920825842,
      "grad_norm": 1.018678069114685,
      "learning_rate": 5.570126323976391e-05,
      "loss": 1.4579,
      "step": 26050
    },
    {
      "epoch": 6.462892925785852,
      "grad_norm": 1.063525915145874,
      "learning_rate": 5.563142239509461e-05,
      "loss": 1.477,
      "step": 26060
    },
    {
      "epoch": 6.465372930745861,
      "grad_norm": 0.9450671076774597,
      "learning_rate": 5.556160848649228e-05,
      "loss": 1.5033,
      "step": 26070
    },
    {
      "epoch": 6.467852935705872,
      "grad_norm": 1.0004218816757202,
      "learning_rate": 5.549182155634076e-05,
      "loss": 1.4485,
      "step": 26080
    },
    {
      "epoch": 6.470332940665881,
      "grad_norm": 1.0630313158035278,
      "learning_rate": 5.542206164700754e-05,
      "loss": 1.4359,
      "step": 26090
    },
    {
      "epoch": 6.472812945625892,
      "grad_norm": 1.0103967189788818,
      "learning_rate": 5.5352328800843724e-05,
      "loss": 1.4517,
      "step": 26100
    },
    {
      "epoch": 6.475292950585901,
      "grad_norm": 1.0282511711120605,
      "learning_rate": 5.528262306018395e-05,
      "loss": 1.4638,
      "step": 26110
    },
    {
      "epoch": 6.477772955545911,
      "grad_norm": 1.0227611064910889,
      "learning_rate": 5.521294446734637e-05,
      "loss": 1.4756,
      "step": 26120
    },
    {
      "epoch": 6.480252960505921,
      "grad_norm": 0.9566037058830261,
      "learning_rate": 5.514329306463284e-05,
      "loss": 1.4416,
      "step": 26130
    },
    {
      "epoch": 6.482732965465931,
      "grad_norm": 1.0739294290542603,
      "learning_rate": 5.5073668894328475e-05,
      "loss": 1.5094,
      "step": 26140
    },
    {
      "epoch": 6.485212970425941,
      "grad_norm": 0.9938433766365051,
      "learning_rate": 5.500407199870199e-05,
      "loss": 1.4879,
      "step": 26150
    },
    {
      "epoch": 6.487692975385951,
      "grad_norm": 0.9702029228210449,
      "learning_rate": 5.493450242000546e-05,
      "loss": 1.5264,
      "step": 26160
    },
    {
      "epoch": 6.49017298034596,
      "grad_norm": 0.9910847544670105,
      "learning_rate": 5.486496020047456e-05,
      "loss": 1.4107,
      "step": 26170
    },
    {
      "epoch": 6.492652985305971,
      "grad_norm": 0.9760767817497253,
      "learning_rate": 5.4795445382328037e-05,
      "loss": 1.4871,
      "step": 26180
    },
    {
      "epoch": 6.49513299026598,
      "grad_norm": 0.9848397374153137,
      "learning_rate": 5.4725958007768304e-05,
      "loss": 1.4719,
      "step": 26190
    },
    {
      "epoch": 6.497612995225991,
      "grad_norm": 0.9554773569107056,
      "learning_rate": 5.465649811898098e-05,
      "loss": 1.4414,
      "step": 26200
    },
    {
      "epoch": 6.500093000186,
      "grad_norm": 0.9741032719612122,
      "learning_rate": 5.458706575813499e-05,
      "loss": 1.4004,
      "step": 26210
    },
    {
      "epoch": 6.502573005146011,
      "grad_norm": 0.9804884791374207,
      "learning_rate": 5.451766096738256e-05,
      "loss": 1.4557,
      "step": 26220
    },
    {
      "epoch": 6.50505301010602,
      "grad_norm": 0.9539157152175903,
      "learning_rate": 5.4448283788859265e-05,
      "loss": 1.4847,
      "step": 26230
    },
    {
      "epoch": 6.50753301506603,
      "grad_norm": 1.0004538297653198,
      "learning_rate": 5.43789342646837e-05,
      "loss": 1.4715,
      "step": 26240
    },
    {
      "epoch": 6.51001302002604,
      "grad_norm": 0.9888569712638855,
      "learning_rate": 5.4309612436957937e-05,
      "loss": 1.4583,
      "step": 26250
    },
    {
      "epoch": 6.51249302498605,
      "grad_norm": 1.0565658807754517,
      "learning_rate": 5.424031834776706e-05,
      "loss": 1.5336,
      "step": 26260
    },
    {
      "epoch": 6.51497302994606,
      "grad_norm": 0.9787483215332031,
      "learning_rate": 5.417105203917936e-05,
      "loss": 1.4092,
      "step": 26270
    },
    {
      "epoch": 6.51745303490607,
      "grad_norm": 0.95024573802948,
      "learning_rate": 5.410181355324622e-05,
      "loss": 1.408,
      "step": 26280
    },
    {
      "epoch": 6.51993303986608,
      "grad_norm": 1.0401235818862915,
      "learning_rate": 5.40326029320023e-05,
      "loss": 1.5157,
      "step": 26290
    },
    {
      "epoch": 6.52241304482609,
      "grad_norm": 0.9956122040748596,
      "learning_rate": 5.396342021746506e-05,
      "loss": 1.5031,
      "step": 26300
    },
    {
      "epoch": 6.524893049786099,
      "grad_norm": 1.0054196119308472,
      "learning_rate": 5.389426545163527e-05,
      "loss": 1.5093,
      "step": 26310
    },
    {
      "epoch": 6.52737305474611,
      "grad_norm": 0.9857822060585022,
      "learning_rate": 5.382513867649663e-05,
      "loss": 1.5204,
      "step": 26320
    },
    {
      "epoch": 6.529853059706119,
      "grad_norm": 1.0408052206039429,
      "learning_rate": 5.3756039934015836e-05,
      "loss": 1.4885,
      "step": 26330
    },
    {
      "epoch": 6.53233306466613,
      "grad_norm": 0.9730471968650818,
      "learning_rate": 5.368696926614255e-05,
      "loss": 1.4119,
      "step": 26340
    },
    {
      "epoch": 6.534813069626139,
      "grad_norm": 0.949488639831543,
      "learning_rate": 5.3617926714809544e-05,
      "loss": 1.4324,
      "step": 26350
    },
    {
      "epoch": 6.537293074586149,
      "grad_norm": 0.9932969808578491,
      "learning_rate": 5.3548912321932244e-05,
      "loss": 1.4441,
      "step": 26360
    },
    {
      "epoch": 6.539773079546159,
      "grad_norm": 1.0820375680923462,
      "learning_rate": 5.347992612940926e-05,
      "loss": 1.4928,
      "step": 26370
    },
    {
      "epoch": 6.542253084506169,
      "grad_norm": 1.0019023418426514,
      "learning_rate": 5.34109681791219e-05,
      "loss": 1.5409,
      "step": 26380
    },
    {
      "epoch": 6.544733089466179,
      "grad_norm": 1.0442291498184204,
      "learning_rate": 5.3342038512934424e-05,
      "loss": 1.4661,
      "step": 26390
    },
    {
      "epoch": 6.547213094426189,
      "grad_norm": 1.1074390411376953,
      "learning_rate": 5.32731371726938e-05,
      "loss": 1.4568,
      "step": 26400
    },
    {
      "epoch": 6.549693099386198,
      "grad_norm": 0.9921213984489441,
      "learning_rate": 5.320426420023001e-05,
      "loss": 1.4836,
      "step": 26410
    },
    {
      "epoch": 6.552173104346209,
      "grad_norm": 1.0667870044708252,
      "learning_rate": 5.313541963735556e-05,
      "loss": 1.4685,
      "step": 26420
    },
    {
      "epoch": 6.554653109306218,
      "grad_norm": 0.9324483871459961,
      "learning_rate": 5.3066603525865924e-05,
      "loss": 1.4437,
      "step": 26430
    },
    {
      "epoch": 6.557133114266229,
      "grad_norm": 1.0055015087127686,
      "learning_rate": 5.2997815907539164e-05,
      "loss": 1.4864,
      "step": 26440
    },
    {
      "epoch": 6.559613119226238,
      "grad_norm": 1.0197795629501343,
      "learning_rate": 5.292905682413612e-05,
      "loss": 1.4771,
      "step": 26450
    },
    {
      "epoch": 6.562093124186248,
      "grad_norm": 1.0034635066986084,
      "learning_rate": 5.286032631740023e-05,
      "loss": 1.5043,
      "step": 26460
    },
    {
      "epoch": 6.564573129146258,
      "grad_norm": 1.0399538278579712,
      "learning_rate": 5.279162442905772e-05,
      "loss": 1.4874,
      "step": 26470
    },
    {
      "epoch": 6.567053134106268,
      "grad_norm": 1.0726194381713867,
      "learning_rate": 5.272295120081732e-05,
      "loss": 1.4854,
      "step": 26480
    },
    {
      "epoch": 6.569533139066278,
      "grad_norm": 0.9627480506896973,
      "learning_rate": 5.265430667437033e-05,
      "loss": 1.4264,
      "step": 26490
    },
    {
      "epoch": 6.572013144026288,
      "grad_norm": 0.9933289885520935,
      "learning_rate": 5.258569089139085e-05,
      "loss": 1.4828,
      "step": 26500
    },
    {
      "epoch": 6.574493148986298,
      "grad_norm": 1.0292378664016724,
      "learning_rate": 5.251710389353519e-05,
      "loss": 1.5436,
      "step": 26510
    },
    {
      "epoch": 6.576973153946308,
      "grad_norm": 0.9901852011680603,
      "learning_rate": 5.2448545722442486e-05,
      "loss": 1.4329,
      "step": 26520
    },
    {
      "epoch": 6.579453158906317,
      "grad_norm": 1.0028679370880127,
      "learning_rate": 5.238001641973422e-05,
      "loss": 1.4741,
      "step": 26530
    },
    {
      "epoch": 6.581933163866328,
      "grad_norm": 1.047088384628296,
      "learning_rate": 5.2311516027014394e-05,
      "loss": 1.4756,
      "step": 26540
    },
    {
      "epoch": 6.584413168826337,
      "grad_norm": 1.0282377004623413,
      "learning_rate": 5.2243044585869394e-05,
      "loss": 1.4281,
      "step": 26550
    },
    {
      "epoch": 6.586893173786348,
      "grad_norm": 0.9611074924468994,
      "learning_rate": 5.217460213786821e-05,
      "loss": 1.4849,
      "step": 26560
    },
    {
      "epoch": 6.589373178746357,
      "grad_norm": 0.9675851464271545,
      "learning_rate": 5.2106188724561944e-05,
      "loss": 1.4897,
      "step": 26570
    },
    {
      "epoch": 6.591853183706368,
      "grad_norm": 1.0171164274215698,
      "learning_rate": 5.203780438748433e-05,
      "loss": 1.4731,
      "step": 26580
    },
    {
      "epoch": 6.594333188666377,
      "grad_norm": 0.9999147653579712,
      "learning_rate": 5.196944916815134e-05,
      "loss": 1.5018,
      "step": 26590
    },
    {
      "epoch": 6.596813193626387,
      "grad_norm": 1.0138448476791382,
      "learning_rate": 5.190112310806126e-05,
      "loss": 1.5076,
      "step": 26600
    },
    {
      "epoch": 6.599293198586397,
      "grad_norm": 0.9509438872337341,
      "learning_rate": 5.1832826248694654e-05,
      "loss": 1.4431,
      "step": 26610
    },
    {
      "epoch": 6.601773203546407,
      "grad_norm": 0.9981142282485962,
      "learning_rate": 5.176455863151448e-05,
      "loss": 1.4173,
      "step": 26620
    },
    {
      "epoch": 6.604253208506417,
      "grad_norm": 0.9768161177635193,
      "learning_rate": 5.1696320297965825e-05,
      "loss": 1.4908,
      "step": 26630
    },
    {
      "epoch": 6.606733213466427,
      "grad_norm": 1.0115209817886353,
      "learning_rate": 5.162811128947602e-05,
      "loss": 1.5331,
      "step": 26640
    },
    {
      "epoch": 6.6092132184264365,
      "grad_norm": 1.0393757820129395,
      "learning_rate": 5.155993164745463e-05,
      "loss": 1.4673,
      "step": 26650
    },
    {
      "epoch": 6.611693223386447,
      "grad_norm": 1.0088094472885132,
      "learning_rate": 5.149178141329334e-05,
      "loss": 1.4769,
      "step": 26660
    },
    {
      "epoch": 6.6141732283464565,
      "grad_norm": 1.0713369846343994,
      "learning_rate": 5.142366062836599e-05,
      "loss": 1.4762,
      "step": 26670
    },
    {
      "epoch": 6.616653233306467,
      "grad_norm": 0.9907805919647217,
      "learning_rate": 5.135556933402862e-05,
      "loss": 1.5143,
      "step": 26680
    },
    {
      "epoch": 6.6191332382664765,
      "grad_norm": 1.0409246683120728,
      "learning_rate": 5.1287507571619275e-05,
      "loss": 1.4116,
      "step": 26690
    },
    {
      "epoch": 6.621613243226486,
      "grad_norm": 1.1151947975158691,
      "learning_rate": 5.12194753824581e-05,
      "loss": 1.4693,
      "step": 26700
    },
    {
      "epoch": 6.6240932481864965,
      "grad_norm": 1.0386748313903809,
      "learning_rate": 5.115147280784727e-05,
      "loss": 1.5714,
      "step": 26710
    },
    {
      "epoch": 6.626573253146506,
      "grad_norm": 1.0036121606826782,
      "learning_rate": 5.108349988907111e-05,
      "loss": 1.505,
      "step": 26720
    },
    {
      "epoch": 6.6290532581065165,
      "grad_norm": 1.053237795829773,
      "learning_rate": 5.1015556667395636e-05,
      "loss": 1.475,
      "step": 26730
    },
    {
      "epoch": 6.631533263066526,
      "grad_norm": 0.996638834476471,
      "learning_rate": 5.09476431840692e-05,
      "loss": 1.5783,
      "step": 26740
    },
    {
      "epoch": 6.634013268026536,
      "grad_norm": 0.9656083583831787,
      "learning_rate": 5.0879759480321866e-05,
      "loss": 1.4629,
      "step": 26750
    },
    {
      "epoch": 6.636493272986546,
      "grad_norm": 0.9791948199272156,
      "learning_rate": 5.0811905597365684e-05,
      "loss": 1.4609,
      "step": 26760
    },
    {
      "epoch": 6.638973277946556,
      "grad_norm": 1.0298975706100464,
      "learning_rate": 5.0744081576394566e-05,
      "loss": 1.4529,
      "step": 26770
    },
    {
      "epoch": 6.641453282906566,
      "grad_norm": 1.0415596961975098,
      "learning_rate": 5.0676287458584436e-05,
      "loss": 1.4407,
      "step": 26780
    },
    {
      "epoch": 6.6439332878665756,
      "grad_norm": 1.0678547620773315,
      "learning_rate": 5.06085232850928e-05,
      "loss": 1.4811,
      "step": 26790
    },
    {
      "epoch": 6.646413292826586,
      "grad_norm": 0.9788858890533447,
      "learning_rate": 5.054078909705926e-05,
      "loss": 1.4804,
      "step": 26800
    },
    {
      "epoch": 6.6488932977865955,
      "grad_norm": 1.0442765951156616,
      "learning_rate": 5.047308493560506e-05,
      "loss": 1.5068,
      "step": 26810
    },
    {
      "epoch": 6.651373302746606,
      "grad_norm": 0.9830688238143921,
      "learning_rate": 5.0405410841833253e-05,
      "loss": 1.4581,
      "step": 26820
    },
    {
      "epoch": 6.6538533077066155,
      "grad_norm": 0.9420977234840393,
      "learning_rate": 5.033776685682858e-05,
      "loss": 1.4918,
      "step": 26830
    },
    {
      "epoch": 6.656333312666625,
      "grad_norm": 1.0370205640792847,
      "learning_rate": 5.0270153021657674e-05,
      "loss": 1.5327,
      "step": 26840
    },
    {
      "epoch": 6.6588133176266355,
      "grad_norm": 0.9369286894798279,
      "learning_rate": 5.0202569377368616e-05,
      "loss": 1.4392,
      "step": 26850
    },
    {
      "epoch": 6.661293322586645,
      "grad_norm": 0.9928011298179626,
      "learning_rate": 5.0135015964991373e-05,
      "loss": 1.453,
      "step": 26860
    },
    {
      "epoch": 6.6637733275466555,
      "grad_norm": 1.0058857202529907,
      "learning_rate": 5.0067492825537466e-05,
      "loss": 1.4793,
      "step": 26870
    },
    {
      "epoch": 6.666253332506665,
      "grad_norm": 0.9954456686973572,
      "learning_rate": 5.000000000000002e-05,
      "loss": 1.4629,
      "step": 26880
    },
    {
      "epoch": 6.668733337466675,
      "grad_norm": 0.9622682332992554,
      "learning_rate": 4.993253752935376e-05,
      "loss": 1.5008,
      "step": 26890
    },
    {
      "epoch": 6.671213342426685,
      "grad_norm": 1.0237582921981812,
      "learning_rate": 4.9865105454555113e-05,
      "loss": 1.5273,
      "step": 26900
    },
    {
      "epoch": 6.673693347386695,
      "grad_norm": 1.0076124668121338,
      "learning_rate": 4.979770381654181e-05,
      "loss": 1.4979,
      "step": 26910
    },
    {
      "epoch": 6.676173352346705,
      "grad_norm": 1.091383457183838,
      "learning_rate": 4.973033265623332e-05,
      "loss": 1.48,
      "step": 26920
    },
    {
      "epoch": 6.678653357306715,
      "grad_norm": 1.0162371397018433,
      "learning_rate": 4.966299201453052e-05,
      "loss": 1.4391,
      "step": 26930
    },
    {
      "epoch": 6.681133362266724,
      "grad_norm": 1.0265648365020752,
      "learning_rate": 4.959568193231575e-05,
      "loss": 1.468,
      "step": 26940
    },
    {
      "epoch": 6.683613367226735,
      "grad_norm": 1.0596145391464233,
      "learning_rate": 4.952840245045278e-05,
      "loss": 1.4158,
      "step": 26950
    },
    {
      "epoch": 6.686093372186744,
      "grad_norm": 1.010204792022705,
      "learning_rate": 4.946115360978696e-05,
      "loss": 1.4245,
      "step": 26960
    },
    {
      "epoch": 6.688573377146755,
      "grad_norm": 1.0146294832229614,
      "learning_rate": 4.939393545114475e-05,
      "loss": 1.4877,
      "step": 26970
    },
    {
      "epoch": 6.691053382106764,
      "grad_norm": 1.0546412467956543,
      "learning_rate": 4.9326748015334266e-05,
      "loss": 1.4588,
      "step": 26980
    },
    {
      "epoch": 6.693533387066774,
      "grad_norm": 0.9743219017982483,
      "learning_rate": 4.925959134314483e-05,
      "loss": 1.4712,
      "step": 26990
    },
    {
      "epoch": 6.696013392026784,
      "grad_norm": 1.0180968046188354,
      "learning_rate": 4.919246547534708e-05,
      "loss": 1.4652,
      "step": 27000
    },
    {
      "epoch": 6.698493396986794,
      "grad_norm": 1.0588595867156982,
      "learning_rate": 4.912537045269298e-05,
      "loss": 1.5079,
      "step": 27010
    },
    {
      "epoch": 6.700973401946804,
      "grad_norm": 0.9891775846481323,
      "learning_rate": 4.9058306315915826e-05,
      "loss": 1.4951,
      "step": 27020
    },
    {
      "epoch": 6.703453406906814,
      "grad_norm": 1.017204999923706,
      "learning_rate": 4.899127310573009e-05,
      "loss": 1.4805,
      "step": 27030
    },
    {
      "epoch": 6.705933411866823,
      "grad_norm": 1.0103859901428223,
      "learning_rate": 4.892427086283147e-05,
      "loss": 1.4185,
      "step": 27040
    },
    {
      "epoch": 6.708413416826834,
      "grad_norm": 0.9853695631027222,
      "learning_rate": 4.885729962789688e-05,
      "loss": 1.4916,
      "step": 27050
    },
    {
      "epoch": 6.710893421786843,
      "grad_norm": 0.973719596862793,
      "learning_rate": 4.879035944158443e-05,
      "loss": 1.5332,
      "step": 27060
    },
    {
      "epoch": 6.713373426746854,
      "grad_norm": 0.9500641822814941,
      "learning_rate": 4.872345034453332e-05,
      "loss": 1.4972,
      "step": 27070
    },
    {
      "epoch": 6.715853431706863,
      "grad_norm": 0.9965008497238159,
      "learning_rate": 4.8656572377363974e-05,
      "loss": 1.4828,
      "step": 27080
    },
    {
      "epoch": 6.718333436666874,
      "grad_norm": 1.011154294013977,
      "learning_rate": 4.8589725580677835e-05,
      "loss": 1.4871,
      "step": 27090
    },
    {
      "epoch": 6.720813441626883,
      "grad_norm": 1.0305109024047852,
      "learning_rate": 4.852290999505742e-05,
      "loss": 1.4241,
      "step": 27100
    },
    {
      "epoch": 6.723293446586894,
      "grad_norm": 1.0206142663955688,
      "learning_rate": 4.845612566106643e-05,
      "loss": 1.4975,
      "step": 27110
    },
    {
      "epoch": 6.725773451546903,
      "grad_norm": 1.0499778985977173,
      "learning_rate": 4.8389372619249326e-05,
      "loss": 1.4872,
      "step": 27120
    },
    {
      "epoch": 6.728253456506913,
      "grad_norm": 1.0595289468765259,
      "learning_rate": 4.8322650910131874e-05,
      "loss": 1.5371,
      "step": 27130
    },
    {
      "epoch": 6.730733461466923,
      "grad_norm": 1.0716408491134644,
      "learning_rate": 4.8255960574220614e-05,
      "loss": 1.5045,
      "step": 27140
    },
    {
      "epoch": 6.733213466426933,
      "grad_norm": 1.0024359226226807,
      "learning_rate": 4.818930165200313e-05,
      "loss": 1.4937,
      "step": 27150
    },
    {
      "epoch": 6.735693471386943,
      "grad_norm": 0.9272238612174988,
      "learning_rate": 4.8122674183947836e-05,
      "loss": 1.4816,
      "step": 27160
    },
    {
      "epoch": 6.738173476346953,
      "grad_norm": 1.0077641010284424,
      "learning_rate": 4.805607821050423e-05,
      "loss": 1.4926,
      "step": 27170
    },
    {
      "epoch": 6.740653481306962,
      "grad_norm": 1.0041226148605347,
      "learning_rate": 4.7989513772102537e-05,
      "loss": 1.5087,
      "step": 27180
    },
    {
      "epoch": 6.743133486266973,
      "grad_norm": 0.9688929915428162,
      "learning_rate": 4.792298090915388e-05,
      "loss": 1.4847,
      "step": 27190
    },
    {
      "epoch": 6.745613491226982,
      "grad_norm": 1.0488873720169067,
      "learning_rate": 4.78564796620502e-05,
      "loss": 1.4595,
      "step": 27200
    },
    {
      "epoch": 6.748093496186993,
      "grad_norm": 1.0456444025039673,
      "learning_rate": 4.7790010071164295e-05,
      "loss": 1.5137,
      "step": 27210
    },
    {
      "epoch": 6.750573501147002,
      "grad_norm": 1.0180890560150146,
      "learning_rate": 4.772357217684964e-05,
      "loss": 1.4553,
      "step": 27220
    },
    {
      "epoch": 6.753053506107012,
      "grad_norm": 0.9567440152168274,
      "learning_rate": 4.7657166019440614e-05,
      "loss": 1.4726,
      "step": 27230
    },
    {
      "epoch": 6.755533511067022,
      "grad_norm": 1.0050458908081055,
      "learning_rate": 4.759079163925223e-05,
      "loss": 1.4967,
      "step": 27240
    },
    {
      "epoch": 6.758013516027032,
      "grad_norm": 0.9629562497138977,
      "learning_rate": 4.7524449076580224e-05,
      "loss": 1.4792,
      "step": 27250
    },
    {
      "epoch": 6.760493520987042,
      "grad_norm": 1.0333480834960938,
      "learning_rate": 4.745813837170101e-05,
      "loss": 1.4672,
      "step": 27260
    },
    {
      "epoch": 6.762973525947052,
      "grad_norm": 1.03449547290802,
      "learning_rate": 4.739185956487169e-05,
      "loss": 1.5072,
      "step": 27270
    },
    {
      "epoch": 6.765453530907061,
      "grad_norm": 0.96829754114151,
      "learning_rate": 4.732561269632992e-05,
      "loss": 1.4646,
      "step": 27280
    },
    {
      "epoch": 6.767933535867072,
      "grad_norm": 1.003289818763733,
      "learning_rate": 4.725939780629415e-05,
      "loss": 1.4566,
      "step": 27290
    },
    {
      "epoch": 6.770413540827081,
      "grad_norm": 0.9847796559333801,
      "learning_rate": 4.7193214934963206e-05,
      "loss": 1.4673,
      "step": 27300
    },
    {
      "epoch": 6.772893545787092,
      "grad_norm": 0.8907151818275452,
      "learning_rate": 4.7127064122516593e-05,
      "loss": 1.4547,
      "step": 27310
    },
    {
      "epoch": 6.775373550747101,
      "grad_norm": 1.001332402229309,
      "learning_rate": 4.706094540911429e-05,
      "loss": 1.4919,
      "step": 27320
    },
    {
      "epoch": 6.777853555707112,
      "grad_norm": 1.0360084772109985,
      "learning_rate": 4.699485883489693e-05,
      "loss": 1.5503,
      "step": 27330
    },
    {
      "epoch": 6.780333560667121,
      "grad_norm": 1.0425145626068115,
      "learning_rate": 4.692880443998538e-05,
      "loss": 1.4477,
      "step": 27340
    },
    {
      "epoch": 6.782813565627131,
      "grad_norm": 0.9380311965942383,
      "learning_rate": 4.686278226448122e-05,
      "loss": 1.4753,
      "step": 27350
    },
    {
      "epoch": 6.785293570587141,
      "grad_norm": 0.9759949445724487,
      "learning_rate": 4.6796792348466356e-05,
      "loss": 1.5504,
      "step": 27360
    },
    {
      "epoch": 6.787773575547151,
      "grad_norm": 0.9939712882041931,
      "learning_rate": 4.6730834732003104e-05,
      "loss": 1.471,
      "step": 27370
    },
    {
      "epoch": 6.790253580507161,
      "grad_norm": 1.0386635065078735,
      "learning_rate": 4.666490945513415e-05,
      "loss": 1.4876,
      "step": 27380
    },
    {
      "epoch": 6.792733585467171,
      "grad_norm": 1.0397119522094727,
      "learning_rate": 4.6599016557882715e-05,
      "loss": 1.4365,
      "step": 27390
    },
    {
      "epoch": 6.795213590427181,
      "grad_norm": 1.0358515977859497,
      "learning_rate": 4.653315608025207e-05,
      "loss": 1.5524,
      "step": 27400
    },
    {
      "epoch": 6.797693595387191,
      "grad_norm": 1.0351741313934326,
      "learning_rate": 4.646732806222608e-05,
      "loss": 1.4834,
      "step": 27410
    },
    {
      "epoch": 6.8001736003472,
      "grad_norm": 0.9974995255470276,
      "learning_rate": 4.6401532543768766e-05,
      "loss": 1.4947,
      "step": 27420
    },
    {
      "epoch": 6.802653605307211,
      "grad_norm": 1.01504385471344,
      "learning_rate": 4.6335769564824425e-05,
      "loss": 1.4457,
      "step": 27430
    },
    {
      "epoch": 6.80513361026722,
      "grad_norm": 0.9690166115760803,
      "learning_rate": 4.6270039165317605e-05,
      "loss": 1.4985,
      "step": 27440
    },
    {
      "epoch": 6.807613615227231,
      "grad_norm": 1.0365806818008423,
      "learning_rate": 4.6204341385153186e-05,
      "loss": 1.5117,
      "step": 27450
    },
    {
      "epoch": 6.81009362018724,
      "grad_norm": 1.0430920124053955,
      "learning_rate": 4.6138676264216e-05,
      "loss": 1.4598,
      "step": 27460
    },
    {
      "epoch": 6.81257362514725,
      "grad_norm": 1.0349470376968384,
      "learning_rate": 4.60730438423713e-05,
      "loss": 1.4375,
      "step": 27470
    },
    {
      "epoch": 6.81505363010726,
      "grad_norm": 0.9820765852928162,
      "learning_rate": 4.600744415946438e-05,
      "loss": 1.4243,
      "step": 27480
    },
    {
      "epoch": 6.81753363506727,
      "grad_norm": 1.0257947444915771,
      "learning_rate": 4.594187725532063e-05,
      "loss": 1.4724,
      "step": 27490
    },
    {
      "epoch": 6.82001364002728,
      "grad_norm": 1.0491620302200317,
      "learning_rate": 4.587634316974555e-05,
      "loss": 1.3997,
      "step": 27500
    },
    {
      "epoch": 6.82249364498729,
      "grad_norm": 1.043031096458435,
      "learning_rate": 4.5810841942524864e-05,
      "loss": 1.509,
      "step": 27510
    },
    {
      "epoch": 6.8249736499472995,
      "grad_norm": 1.0190041065216064,
      "learning_rate": 4.574537361342407e-05,
      "loss": 1.4837,
      "step": 27520
    },
    {
      "epoch": 6.82745365490731,
      "grad_norm": 1.0018768310546875,
      "learning_rate": 4.5679938222188946e-05,
      "loss": 1.4611,
      "step": 27530
    },
    {
      "epoch": 6.8299336598673195,
      "grad_norm": 1.0004897117614746,
      "learning_rate": 4.561453580854516e-05,
      "loss": 1.4628,
      "step": 27540
    },
    {
      "epoch": 6.83241366482733,
      "grad_norm": 0.979097843170166,
      "learning_rate": 4.554916641219835e-05,
      "loss": 1.4714,
      "step": 27550
    },
    {
      "epoch": 6.8348936697873395,
      "grad_norm": 1.035117268562317,
      "learning_rate": 4.548383007283412e-05,
      "loss": 1.4824,
      "step": 27560
    },
    {
      "epoch": 6.837373674747349,
      "grad_norm": 0.9248478412628174,
      "learning_rate": 4.541852683011812e-05,
      "loss": 1.4493,
      "step": 27570
    },
    {
      "epoch": 6.8398536797073595,
      "grad_norm": 1.0277906656265259,
      "learning_rate": 4.535325672369567e-05,
      "loss": 1.4597,
      "step": 27580
    },
    {
      "epoch": 6.842333684667369,
      "grad_norm": 1.085394024848938,
      "learning_rate": 4.52880197931922e-05,
      "loss": 1.3996,
      "step": 27590
    },
    {
      "epoch": 6.8448136896273795,
      "grad_norm": 0.9753702878952026,
      "learning_rate": 4.522281607821288e-05,
      "loss": 1.4157,
      "step": 27600
    },
    {
      "epoch": 6.847293694587389,
      "grad_norm": 1.005562663078308,
      "learning_rate": 4.515764561834276e-05,
      "loss": 1.519,
      "step": 27610
    },
    {
      "epoch": 6.8497736995473995,
      "grad_norm": 0.9952786564826965,
      "learning_rate": 4.509250845314662e-05,
      "loss": 1.5127,
      "step": 27620
    },
    {
      "epoch": 6.852253704507409,
      "grad_norm": 0.9692873358726501,
      "learning_rate": 4.502740462216919e-05,
      "loss": 1.509,
      "step": 27630
    },
    {
      "epoch": 6.854733709467419,
      "grad_norm": 0.954460084438324,
      "learning_rate": 4.4962334164934806e-05,
      "loss": 1.4875,
      "step": 27640
    },
    {
      "epoch": 6.857213714427429,
      "grad_norm": 1.0443848371505737,
      "learning_rate": 4.4897297120947624e-05,
      "loss": 1.4615,
      "step": 27650
    },
    {
      "epoch": 6.8596937193874385,
      "grad_norm": 1.0167874097824097,
      "learning_rate": 4.483229352969146e-05,
      "loss": 1.459,
      "step": 27660
    },
    {
      "epoch": 6.862173724347449,
      "grad_norm": 1.013476848602295,
      "learning_rate": 4.4767323430629894e-05,
      "loss": 1.5109,
      "step": 27670
    },
    {
      "epoch": 6.8646537293074585,
      "grad_norm": 0.9592860341072083,
      "learning_rate": 4.470238686320606e-05,
      "loss": 1.4595,
      "step": 27680
    },
    {
      "epoch": 6.867133734267469,
      "grad_norm": 1.0391489267349243,
      "learning_rate": 4.463748386684289e-05,
      "loss": 1.4536,
      "step": 27690
    },
    {
      "epoch": 6.8696137392274785,
      "grad_norm": 1.093828797340393,
      "learning_rate": 4.457261448094282e-05,
      "loss": 1.5159,
      "step": 27700
    },
    {
      "epoch": 6.872093744187488,
      "grad_norm": 0.9685669541358948,
      "learning_rate": 4.450777874488792e-05,
      "loss": 1.4964,
      "step": 27710
    },
    {
      "epoch": 6.8745737491474985,
      "grad_norm": 0.9868682026863098,
      "learning_rate": 4.444297669803981e-05,
      "loss": 1.4689,
      "step": 27720
    },
    {
      "epoch": 6.877053754107508,
      "grad_norm": 0.94803786277771,
      "learning_rate": 4.437820837973964e-05,
      "loss": 1.5238,
      "step": 27730
    },
    {
      "epoch": 6.8795337590675185,
      "grad_norm": 1.0392026901245117,
      "learning_rate": 4.431347382930818e-05,
      "loss": 1.5276,
      "step": 27740
    },
    {
      "epoch": 6.882013764027528,
      "grad_norm": 0.980580747127533,
      "learning_rate": 4.424877308604563e-05,
      "loss": 1.4614,
      "step": 27750
    },
    {
      "epoch": 6.884493768987538,
      "grad_norm": 0.9709716439247131,
      "learning_rate": 4.4184106189231625e-05,
      "loss": 1.5172,
      "step": 27760
    },
    {
      "epoch": 6.886973773947548,
      "grad_norm": 0.9925903081893921,
      "learning_rate": 4.411947317812529e-05,
      "loss": 1.5287,
      "step": 27770
    },
    {
      "epoch": 6.889453778907558,
      "grad_norm": 1.0866552591323853,
      "learning_rate": 4.405487409196526e-05,
      "loss": 1.4618,
      "step": 27780
    },
    {
      "epoch": 6.891933783867568,
      "grad_norm": 1.0447962284088135,
      "learning_rate": 4.399030896996945e-05,
      "loss": 1.4656,
      "step": 27790
    },
    {
      "epoch": 6.894413788827578,
      "grad_norm": 1.1322532892227173,
      "learning_rate": 4.3925777851335206e-05,
      "loss": 1.5223,
      "step": 27800
    },
    {
      "epoch": 6.896893793787587,
      "grad_norm": 1.0157110691070557,
      "learning_rate": 4.386128077523923e-05,
      "loss": 1.4398,
      "step": 27810
    },
    {
      "epoch": 6.899373798747598,
      "grad_norm": 1.016741156578064,
      "learning_rate": 4.3796817780837574e-05,
      "loss": 1.4614,
      "step": 27820
    },
    {
      "epoch": 6.901853803707607,
      "grad_norm": 1.0230491161346436,
      "learning_rate": 4.373238890726553e-05,
      "loss": 1.4625,
      "step": 27830
    },
    {
      "epoch": 6.904333808667618,
      "grad_norm": 0.9990814924240112,
      "learning_rate": 4.3667994193637796e-05,
      "loss": 1.5499,
      "step": 27840
    },
    {
      "epoch": 6.906813813627627,
      "grad_norm": 0.9694319367408752,
      "learning_rate": 4.360363367904823e-05,
      "loss": 1.5205,
      "step": 27850
    },
    {
      "epoch": 6.909293818587637,
      "grad_norm": 1.0149998664855957,
      "learning_rate": 4.353930740256996e-05,
      "loss": 1.4735,
      "step": 27860
    },
    {
      "epoch": 6.911773823547647,
      "grad_norm": 1.0618832111358643,
      "learning_rate": 4.3475015403255304e-05,
      "loss": 1.5441,
      "step": 27870
    },
    {
      "epoch": 6.914253828507657,
      "grad_norm": 0.9641765356063843,
      "learning_rate": 4.341075772013589e-05,
      "loss": 1.4671,
      "step": 27880
    },
    {
      "epoch": 6.916733833467667,
      "grad_norm": 1.0613869428634644,
      "learning_rate": 4.3346534392222274e-05,
      "loss": 1.5015,
      "step": 27890
    },
    {
      "epoch": 6.919213838427677,
      "grad_norm": 1.0948052406311035,
      "learning_rate": 4.328234545850442e-05,
      "loss": 1.4689,
      "step": 27900
    },
    {
      "epoch": 6.921693843387687,
      "grad_norm": 1.0151530504226685,
      "learning_rate": 4.321819095795123e-05,
      "loss": 1.4932,
      "step": 27910
    },
    {
      "epoch": 6.924173848347697,
      "grad_norm": 1.0254912376403809,
      "learning_rate": 4.315407092951078e-05,
      "loss": 1.4806,
      "step": 27920
    },
    {
      "epoch": 6.926653853307707,
      "grad_norm": 1.0084120035171509,
      "learning_rate": 4.308998541211015e-05,
      "loss": 1.5049,
      "step": 27930
    },
    {
      "epoch": 6.929133858267717,
      "grad_norm": 0.9881705045700073,
      "learning_rate": 4.3025934444655655e-05,
      "loss": 1.4716,
      "step": 27940
    },
    {
      "epoch": 6.931613863227726,
      "grad_norm": 1.069356083869934,
      "learning_rate": 4.2961918066032326e-05,
      "loss": 1.544,
      "step": 27950
    },
    {
      "epoch": 6.934093868187737,
      "grad_norm": 0.9830098748207092,
      "learning_rate": 4.289793631510448e-05,
      "loss": 1.4613,
      "step": 27960
    },
    {
      "epoch": 6.936573873147746,
      "grad_norm": 0.971098005771637,
      "learning_rate": 4.2833989230715275e-05,
      "loss": 1.4965,
      "step": 27970
    },
    {
      "epoch": 6.939053878107757,
      "grad_norm": 0.9721178412437439,
      "learning_rate": 4.2770076851686824e-05,
      "loss": 1.4694,
      "step": 27980
    },
    {
      "epoch": 6.941533883067766,
      "grad_norm": 1.0568654537200928,
      "learning_rate": 4.2706199216820185e-05,
      "loss": 1.5414,
      "step": 27990
    },
    {
      "epoch": 6.944013888027776,
      "grad_norm": 0.9629611372947693,
      "learning_rate": 4.264235636489542e-05,
      "loss": 1.4599,
      "step": 28000
    },
    {
      "epoch": 6.946493892987786,
      "grad_norm": 0.9821149706840515,
      "learning_rate": 4.257854833467124e-05,
      "loss": 1.4858,
      "step": 28010
    },
    {
      "epoch": 6.948973897947796,
      "grad_norm": 0.9563151597976685,
      "learning_rate": 4.2514775164885487e-05,
      "loss": 1.4879,
      "step": 28020
    },
    {
      "epoch": 6.951453902907806,
      "grad_norm": 1.0238558053970337,
      "learning_rate": 4.245103689425467e-05,
      "loss": 1.5114,
      "step": 28030
    },
    {
      "epoch": 6.953933907867816,
      "grad_norm": 1.0556073188781738,
      "learning_rate": 4.238733356147414e-05,
      "loss": 1.4774,
      "step": 28040
    },
    {
      "epoch": 6.956413912827825,
      "grad_norm": 1.0477728843688965,
      "learning_rate": 4.232366520521803e-05,
      "loss": 1.4576,
      "step": 28050
    },
    {
      "epoch": 6.958893917787836,
      "grad_norm": 1.0674879550933838,
      "learning_rate": 4.226003186413937e-05,
      "loss": 1.4974,
      "step": 28060
    },
    {
      "epoch": 6.961373922747845,
      "grad_norm": 1.0021710395812988,
      "learning_rate": 4.219643357686967e-05,
      "loss": 1.4779,
      "step": 28070
    },
    {
      "epoch": 6.963853927707856,
      "grad_norm": 1.0351998805999756,
      "learning_rate": 4.213287038201943e-05,
      "loss": 1.4574,
      "step": 28080
    },
    {
      "epoch": 6.966333932667865,
      "grad_norm": 0.9760614633560181,
      "learning_rate": 4.206934231817768e-05,
      "loss": 1.5576,
      "step": 28090
    },
    {
      "epoch": 6.968813937627875,
      "grad_norm": 0.9329560399055481,
      "learning_rate": 4.200584942391219e-05,
      "loss": 1.4752,
      "step": 28100
    },
    {
      "epoch": 6.971293942587885,
      "grad_norm": 0.914653480052948,
      "learning_rate": 4.194239173776929e-05,
      "loss": 1.4993,
      "step": 28110
    },
    {
      "epoch": 6.973773947547895,
      "grad_norm": 0.9832881689071655,
      "learning_rate": 4.187896929827414e-05,
      "loss": 1.4533,
      "step": 28120
    },
    {
      "epoch": 6.976253952507905,
      "grad_norm": 0.9643089175224304,
      "learning_rate": 4.18155821439302e-05,
      "loss": 1.4188,
      "step": 28130
    },
    {
      "epoch": 6.978733957467915,
      "grad_norm": 1.0578384399414062,
      "learning_rate": 4.17522303132198e-05,
      "loss": 1.4795,
      "step": 28140
    },
    {
      "epoch": 6.981213962427924,
      "grad_norm": 1.0531888008117676,
      "learning_rate": 4.168891384460365e-05,
      "loss": 1.4477,
      "step": 28150
    },
    {
      "epoch": 6.983693967387935,
      "grad_norm": 1.0598180294036865,
      "learning_rate": 4.1625632776521037e-05,
      "loss": 1.5376,
      "step": 28160
    },
    {
      "epoch": 6.986173972347944,
      "grad_norm": 0.9898632764816284,
      "learning_rate": 4.1562387147389735e-05,
      "loss": 1.5151,
      "step": 28170
    },
    {
      "epoch": 6.988653977307955,
      "grad_norm": 1.0068705081939697,
      "learning_rate": 4.14991769956061e-05,
      "loss": 1.4815,
      "step": 28180
    },
    {
      "epoch": 6.991133982267964,
      "grad_norm": 1.069098949432373,
      "learning_rate": 4.143600235954483e-05,
      "loss": 1.5614,
      "step": 28190
    },
    {
      "epoch": 6.993613987227975,
      "grad_norm": 0.9832902550697327,
      "learning_rate": 4.137286327755913e-05,
      "loss": 1.4777,
      "step": 28200
    },
    {
      "epoch": 6.996093992187984,
      "grad_norm": 1.042362093925476,
      "learning_rate": 4.1309759787980565e-05,
      "loss": 1.4705,
      "step": 28210
    },
    {
      "epoch": 6.998573997147995,
      "grad_norm": 0.9542574286460876,
      "learning_rate": 4.124669192911916e-05,
      "loss": 1.5014,
      "step": 28220
    },
    {
      "epoch": 7.001054002108004,
      "grad_norm": 0.9545601010322571,
      "learning_rate": 4.118365973926322e-05,
      "loss": 1.4232,
      "step": 28230
    },
    {
      "epoch": 7.003534007068014,
      "grad_norm": 0.9939719438552856,
      "learning_rate": 4.112066325667954e-05,
      "loss": 1.3614,
      "step": 28240
    },
    {
      "epoch": 7.006014012028024,
      "grad_norm": 0.9878036379814148,
      "learning_rate": 4.1057702519613084e-05,
      "loss": 1.4261,
      "step": 28250
    },
    {
      "epoch": 7.008494016988034,
      "grad_norm": 1.09098482131958,
      "learning_rate": 4.0994777566287204e-05,
      "loss": 1.4501,
      "step": 28260
    },
    {
      "epoch": 7.010974021948044,
      "grad_norm": 0.9886322617530823,
      "learning_rate": 4.0931888434903484e-05,
      "loss": 1.4651,
      "step": 28270
    },
    {
      "epoch": 7.013454026908054,
      "grad_norm": 1.0460489988327026,
      "learning_rate": 4.086903516364179e-05,
      "loss": 1.4458,
      "step": 28280
    },
    {
      "epoch": 7.015934031868063,
      "grad_norm": 0.9749070405960083,
      "learning_rate": 4.0806217790660176e-05,
      "loss": 1.3642,
      "step": 28290
    },
    {
      "epoch": 7.018414036828074,
      "grad_norm": 1.0082083940505981,
      "learning_rate": 4.074343635409499e-05,
      "loss": 1.3828,
      "step": 28300
    },
    {
      "epoch": 7.020894041788083,
      "grad_norm": 1.0243860483169556,
      "learning_rate": 4.068069089206067e-05,
      "loss": 1.4127,
      "step": 28310
    },
    {
      "epoch": 7.023374046748094,
      "grad_norm": 0.9525448679924011,
      "learning_rate": 4.0617981442649855e-05,
      "loss": 1.4274,
      "step": 28320
    },
    {
      "epoch": 7.025854051708103,
      "grad_norm": 1.0825144052505493,
      "learning_rate": 4.055530804393329e-05,
      "loss": 1.5011,
      "step": 28330
    },
    {
      "epoch": 7.028334056668113,
      "grad_norm": 0.9980800747871399,
      "learning_rate": 4.049267073395995e-05,
      "loss": 1.3625,
      "step": 28340
    },
    {
      "epoch": 7.030814061628123,
      "grad_norm": 1.0209801197052002,
      "learning_rate": 4.0430069550756665e-05,
      "loss": 1.4427,
      "step": 28350
    },
    {
      "epoch": 7.033294066588133,
      "grad_norm": 0.9010273218154907,
      "learning_rate": 4.03675045323286e-05,
      "loss": 1.3683,
      "step": 28360
    },
    {
      "epoch": 7.035774071548143,
      "grad_norm": 1.0834155082702637,
      "learning_rate": 4.03049757166588e-05,
      "loss": 1.4283,
      "step": 28370
    },
    {
      "epoch": 7.038254076508153,
      "grad_norm": 0.9353013634681702,
      "learning_rate": 4.0242483141708334e-05,
      "loss": 1.407,
      "step": 28380
    },
    {
      "epoch": 7.040734081468163,
      "grad_norm": 1.0415675640106201,
      "learning_rate": 4.018002684541638e-05,
      "loss": 1.4974,
      "step": 28390
    },
    {
      "epoch": 7.043214086428173,
      "grad_norm": 0.987139880657196,
      "learning_rate": 4.0117606865699975e-05,
      "loss": 1.4595,
      "step": 28400
    },
    {
      "epoch": 7.0456940913881825,
      "grad_norm": 0.9571609497070312,
      "learning_rate": 4.0055223240454167e-05,
      "loss": 1.4665,
      "step": 28410
    },
    {
      "epoch": 7.048174096348193,
      "grad_norm": 1.0244213342666626,
      "learning_rate": 3.999287600755192e-05,
      "loss": 1.3859,
      "step": 28420
    },
    {
      "epoch": 7.0506541013082025,
      "grad_norm": 1.0276232957839966,
      "learning_rate": 3.9930565204844094e-05,
      "loss": 1.3819,
      "step": 28430
    },
    {
      "epoch": 7.053134106268213,
      "grad_norm": 0.9392919540405273,
      "learning_rate": 3.9868290870159405e-05,
      "loss": 1.364,
      "step": 28440
    },
    {
      "epoch": 7.0556141112282225,
      "grad_norm": 1.0440912246704102,
      "learning_rate": 3.980605304130456e-05,
      "loss": 1.4126,
      "step": 28450
    },
    {
      "epoch": 7.058094116188232,
      "grad_norm": 1.0756579637527466,
      "learning_rate": 3.974385175606393e-05,
      "loss": 1.4206,
      "step": 28460
    },
    {
      "epoch": 7.0605741211482425,
      "grad_norm": 1.0449657440185547,
      "learning_rate": 3.968168705219982e-05,
      "loss": 1.3799,
      "step": 28470
    },
    {
      "epoch": 7.063054126108252,
      "grad_norm": 1.0110985040664673,
      "learning_rate": 3.961955896745224e-05,
      "loss": 1.3762,
      "step": 28480
    },
    {
      "epoch": 7.0655341310682624,
      "grad_norm": 1.0575342178344727,
      "learning_rate": 3.9557467539539115e-05,
      "loss": 1.4568,
      "step": 28490
    },
    {
      "epoch": 7.068014136028272,
      "grad_norm": 1.0499767065048218,
      "learning_rate": 3.9495412806155883e-05,
      "loss": 1.4177,
      "step": 28500
    },
    {
      "epoch": 7.0704941409882816,
      "grad_norm": 1.0368854999542236,
      "learning_rate": 3.943339480497593e-05,
      "loss": 1.3894,
      "step": 28510
    },
    {
      "epoch": 7.072974145948292,
      "grad_norm": 1.0026966333389282,
      "learning_rate": 3.937141357365023e-05,
      "loss": 1.4974,
      "step": 28520
    },
    {
      "epoch": 7.0754541509083015,
      "grad_norm": 1.004560112953186,
      "learning_rate": 3.930946914980744e-05,
      "loss": 1.4786,
      "step": 28530
    },
    {
      "epoch": 7.077934155868312,
      "grad_norm": 0.9743759036064148,
      "learning_rate": 3.924756157105387e-05,
      "loss": 1.4094,
      "step": 28540
    },
    {
      "epoch": 7.0804141608283215,
      "grad_norm": 1.0317790508270264,
      "learning_rate": 3.9185690874973555e-05,
      "loss": 1.4224,
      "step": 28550
    },
    {
      "epoch": 7.082894165788332,
      "grad_norm": 1.181426763534546,
      "learning_rate": 3.9123857099127936e-05,
      "loss": 1.4553,
      "step": 28560
    },
    {
      "epoch": 7.0853741707483415,
      "grad_norm": 0.9656166434288025,
      "learning_rate": 3.906206028105627e-05,
      "loss": 1.4207,
      "step": 28570
    },
    {
      "epoch": 7.087854175708351,
      "grad_norm": 1.0018301010131836,
      "learning_rate": 3.9000300458275216e-05,
      "loss": 1.4154,
      "step": 28580
    },
    {
      "epoch": 7.0903341806683615,
      "grad_norm": 1.0536874532699585,
      "learning_rate": 3.8938577668279055e-05,
      "loss": 1.3925,
      "step": 28590
    },
    {
      "epoch": 7.092814185628371,
      "grad_norm": 1.0299558639526367,
      "learning_rate": 3.887689194853951e-05,
      "loss": 1.3927,
      "step": 28600
    },
    {
      "epoch": 7.0952941905883815,
      "grad_norm": 1.037526249885559,
      "learning_rate": 3.8815243336505954e-05,
      "loss": 1.4441,
      "step": 28610
    },
    {
      "epoch": 7.097774195548391,
      "grad_norm": 1.0067542791366577,
      "learning_rate": 3.875363186960499e-05,
      "loss": 1.4132,
      "step": 28620
    },
    {
      "epoch": 7.100254200508401,
      "grad_norm": 1.088285207748413,
      "learning_rate": 3.8692057585240905e-05,
      "loss": 1.4681,
      "step": 28630
    },
    {
      "epoch": 7.102734205468411,
      "grad_norm": 1.0379809141159058,
      "learning_rate": 3.863052052079528e-05,
      "loss": 1.4277,
      "step": 28640
    },
    {
      "epoch": 7.105214210428421,
      "grad_norm": 1.0791525840759277,
      "learning_rate": 3.8569020713627126e-05,
      "loss": 1.483,
      "step": 28650
    },
    {
      "epoch": 7.107694215388431,
      "grad_norm": 1.0711592435836792,
      "learning_rate": 3.850755820107281e-05,
      "loss": 1.4331,
      "step": 28660
    },
    {
      "epoch": 7.110174220348441,
      "grad_norm": 0.9801573157310486,
      "learning_rate": 3.8446133020446204e-05,
      "loss": 1.4416,
      "step": 28670
    },
    {
      "epoch": 7.112654225308451,
      "grad_norm": 1.5557962656021118,
      "learning_rate": 3.838474520903825e-05,
      "loss": 1.4729,
      "step": 28680
    },
    {
      "epoch": 7.115134230268461,
      "grad_norm": 1.0418564081192017,
      "learning_rate": 3.832339480411747e-05,
      "loss": 1.438,
      "step": 28690
    },
    {
      "epoch": 7.11761423522847,
      "grad_norm": 0.9985926747322083,
      "learning_rate": 3.826208184292952e-05,
      "loss": 1.4208,
      "step": 28700
    },
    {
      "epoch": 7.120094240188481,
      "grad_norm": 1.0571954250335693,
      "learning_rate": 3.820080636269737e-05,
      "loss": 1.4457,
      "step": 28710
    },
    {
      "epoch": 7.12257424514849,
      "grad_norm": 0.9708046913146973,
      "learning_rate": 3.813956840062118e-05,
      "loss": 1.3994,
      "step": 28720
    },
    {
      "epoch": 7.125054250108501,
      "grad_norm": 1.052629828453064,
      "learning_rate": 3.807836799387852e-05,
      "loss": 1.4197,
      "step": 28730
    },
    {
      "epoch": 7.12753425506851,
      "grad_norm": 1.021501898765564,
      "learning_rate": 3.801720517962386e-05,
      "loss": 1.3936,
      "step": 28740
    },
    {
      "epoch": 7.13001426002852,
      "grad_norm": 0.9857499003410339,
      "learning_rate": 3.7956079994989115e-05,
      "loss": 1.4273,
      "step": 28750
    },
    {
      "epoch": 7.13249426498853,
      "grad_norm": 0.98251873254776,
      "learning_rate": 3.7894992477083224e-05,
      "loss": 1.3984,
      "step": 28760
    },
    {
      "epoch": 7.13497426994854,
      "grad_norm": 1.0209919214248657,
      "learning_rate": 3.783394266299228e-05,
      "loss": 1.419,
      "step": 28770
    },
    {
      "epoch": 7.13745427490855,
      "grad_norm": 1.0622920989990234,
      "learning_rate": 3.7772930589779474e-05,
      "loss": 1.4213,
      "step": 28780
    },
    {
      "epoch": 7.13993427986856,
      "grad_norm": 1.0227277278900146,
      "learning_rate": 3.771195629448515e-05,
      "loss": 1.4288,
      "step": 28790
    },
    {
      "epoch": 7.14241428482857,
      "grad_norm": 1.0391366481781006,
      "learning_rate": 3.7651019814126654e-05,
      "loss": 1.4327,
      "step": 28800
    },
    {
      "epoch": 7.14489428978858,
      "grad_norm": 1.063780426979065,
      "learning_rate": 3.759012118569838e-05,
      "loss": 1.4384,
      "step": 28810
    },
    {
      "epoch": 7.147374294748589,
      "grad_norm": 0.9601597785949707,
      "learning_rate": 3.7529260446171756e-05,
      "loss": 1.3743,
      "step": 28820
    },
    {
      "epoch": 7.1498542997086,
      "grad_norm": 1.036189079284668,
      "learning_rate": 3.746843763249521e-05,
      "loss": 1.4047,
      "step": 28830
    },
    {
      "epoch": 7.152334304668609,
      "grad_norm": 1.0754128694534302,
      "learning_rate": 3.7407652781594095e-05,
      "loss": 1.4245,
      "step": 28840
    },
    {
      "epoch": 7.15481430962862,
      "grad_norm": 1.0699996948242188,
      "learning_rate": 3.734690593037083e-05,
      "loss": 1.4439,
      "step": 28850
    },
    {
      "epoch": 7.157294314588629,
      "grad_norm": 1.0184645652770996,
      "learning_rate": 3.728619711570468e-05,
      "loss": 1.3803,
      "step": 28860
    },
    {
      "epoch": 7.159774319548639,
      "grad_norm": 1.082371473312378,
      "learning_rate": 3.7225526374451815e-05,
      "loss": 1.4388,
      "step": 28870
    },
    {
      "epoch": 7.162254324508649,
      "grad_norm": 1.0035544633865356,
      "learning_rate": 3.716489374344527e-05,
      "loss": 1.4139,
      "step": 28880
    },
    {
      "epoch": 7.164734329468659,
      "grad_norm": 1.1102402210235596,
      "learning_rate": 3.710429925949511e-05,
      "loss": 1.4408,
      "step": 28890
    },
    {
      "epoch": 7.167214334428669,
      "grad_norm": 1.0354952812194824,
      "learning_rate": 3.704374295938797e-05,
      "loss": 1.4069,
      "step": 28900
    },
    {
      "epoch": 7.169694339388679,
      "grad_norm": 1.008049726486206,
      "learning_rate": 3.698322487988755e-05,
      "loss": 1.4966,
      "step": 28910
    },
    {
      "epoch": 7.172174344348688,
      "grad_norm": 1.06466805934906,
      "learning_rate": 3.692274505773419e-05,
      "loss": 1.4678,
      "step": 28920
    },
    {
      "epoch": 7.174654349308699,
      "grad_norm": 1.097727656364441,
      "learning_rate": 3.686230352964508e-05,
      "loss": 1.4584,
      "step": 28930
    },
    {
      "epoch": 7.177134354268708,
      "grad_norm": 0.9777037501335144,
      "learning_rate": 3.680190033231411e-05,
      "loss": 1.3806,
      "step": 28940
    },
    {
      "epoch": 7.179614359228719,
      "grad_norm": 1.014937400817871,
      "learning_rate": 3.674153550241202e-05,
      "loss": 1.3776,
      "step": 28950
    },
    {
      "epoch": 7.182094364188728,
      "grad_norm": 1.1064553260803223,
      "learning_rate": 3.668120907658603e-05,
      "loss": 1.4125,
      "step": 28960
    },
    {
      "epoch": 7.184574369148739,
      "grad_norm": 1.0743889808654785,
      "learning_rate": 3.662092109146029e-05,
      "loss": 1.3742,
      "step": 28970
    },
    {
      "epoch": 7.187054374108748,
      "grad_norm": 1.1073992252349854,
      "learning_rate": 3.6560671583635467e-05,
      "loss": 1.3365,
      "step": 28980
    },
    {
      "epoch": 7.189534379068758,
      "grad_norm": 0.9936246275901794,
      "learning_rate": 3.650046058968888e-05,
      "loss": 1.3753,
      "step": 28990
    },
    {
      "epoch": 7.192014384028768,
      "grad_norm": 1.1510530710220337,
      "learning_rate": 3.644028814617454e-05,
      "loss": 1.4367,
      "step": 29000
    },
    {
      "epoch": 7.194494388988778,
      "grad_norm": 1.0159631967544556,
      "learning_rate": 3.6380154289623e-05,
      "loss": 1.3983,
      "step": 29010
    },
    {
      "epoch": 7.196974393948788,
      "grad_norm": 1.0250071287155151,
      "learning_rate": 3.632005905654138e-05,
      "loss": 1.4273,
      "step": 29020
    },
    {
      "epoch": 7.199454398908798,
      "grad_norm": 1.075727939605713,
      "learning_rate": 3.6260002483413325e-05,
      "loss": 1.4309,
      "step": 29030
    },
    {
      "epoch": 7.201934403868807,
      "grad_norm": 1.1067150831222534,
      "learning_rate": 3.6199984606699155e-05,
      "loss": 1.3985,
      "step": 29040
    },
    {
      "epoch": 7.204414408828818,
      "grad_norm": 1.0309680700302124,
      "learning_rate": 3.614000546283547e-05,
      "loss": 1.4569,
      "step": 29050
    },
    {
      "epoch": 7.206894413788827,
      "grad_norm": 1.014217734336853,
      "learning_rate": 3.608006508823556e-05,
      "loss": 1.4271,
      "step": 29060
    },
    {
      "epoch": 7.209374418748838,
      "grad_norm": 1.09121835231781,
      "learning_rate": 3.6020163519289076e-05,
      "loss": 1.4005,
      "step": 29070
    },
    {
      "epoch": 7.211854423708847,
      "grad_norm": 1.0779592990875244,
      "learning_rate": 3.596030079236212e-05,
      "loss": 1.4582,
      "step": 29080
    },
    {
      "epoch": 7.214334428668858,
      "grad_norm": 1.1280418634414673,
      "learning_rate": 3.59004769437972e-05,
      "loss": 1.4004,
      "step": 29090
    },
    {
      "epoch": 7.216814433628867,
      "grad_norm": 1.0576897859573364,
      "learning_rate": 3.5840692009913344e-05,
      "loss": 1.4606,
      "step": 29100
    },
    {
      "epoch": 7.219294438588877,
      "grad_norm": 0.9991346001625061,
      "learning_rate": 3.5780946027005715e-05,
      "loss": 1.4001,
      "step": 29110
    },
    {
      "epoch": 7.221774443548887,
      "grad_norm": 1.076958179473877,
      "learning_rate": 3.5721239031346066e-05,
      "loss": 1.4134,
      "step": 29120
    },
    {
      "epoch": 7.224254448508897,
      "grad_norm": 1.0537842512130737,
      "learning_rate": 3.5661571059182366e-05,
      "loss": 1.4734,
      "step": 29130
    },
    {
      "epoch": 7.226734453468907,
      "grad_norm": 1.103651762008667,
      "learning_rate": 3.56019421467389e-05,
      "loss": 1.4437,
      "step": 29140
    },
    {
      "epoch": 7.229214458428917,
      "grad_norm": 1.057870864868164,
      "learning_rate": 3.554235233021621e-05,
      "loss": 1.4043,
      "step": 29150
    },
    {
      "epoch": 7.231694463388926,
      "grad_norm": 1.0721110105514526,
      "learning_rate": 3.548280164579126e-05,
      "loss": 1.414,
      "step": 29160
    },
    {
      "epoch": 7.234174468348937,
      "grad_norm": 1.1894139051437378,
      "learning_rate": 3.5423290129617e-05,
      "loss": 1.3853,
      "step": 29170
    },
    {
      "epoch": 7.236654473308946,
      "grad_norm": 1.1009711027145386,
      "learning_rate": 3.5363817817822855e-05,
      "loss": 1.4837,
      "step": 29180
    },
    {
      "epoch": 7.239134478268957,
      "grad_norm": 1.0206698179244995,
      "learning_rate": 3.530438474651428e-05,
      "loss": 1.3513,
      "step": 29190
    },
    {
      "epoch": 7.241614483228966,
      "grad_norm": 1.024185061454773,
      "learning_rate": 3.5244990951772974e-05,
      "loss": 1.4392,
      "step": 29200
    },
    {
      "epoch": 7.244094488188976,
      "grad_norm": 1.1004998683929443,
      "learning_rate": 3.5185636469656746e-05,
      "loss": 1.4252,
      "step": 29210
    },
    {
      "epoch": 7.246574493148986,
      "grad_norm": 0.9977849721908569,
      "learning_rate": 3.5126321336199676e-05,
      "loss": 1.4368,
      "step": 29220
    },
    {
      "epoch": 7.249054498108996,
      "grad_norm": 1.039672613143921,
      "learning_rate": 3.5067045587411716e-05,
      "loss": 1.4761,
      "step": 29230
    },
    {
      "epoch": 7.251534503069006,
      "grad_norm": 0.9992042183876038,
      "learning_rate": 3.500780925927915e-05,
      "loss": 1.4159,
      "step": 29240
    },
    {
      "epoch": 7.254014508029016,
      "grad_norm": 0.9692917466163635,
      "learning_rate": 3.494861238776418e-05,
      "loss": 1.4956,
      "step": 29250
    },
    {
      "epoch": 7.256494512989026,
      "grad_norm": 1.065436840057373,
      "learning_rate": 3.4889455008805106e-05,
      "loss": 1.4487,
      "step": 29260
    },
    {
      "epoch": 7.258974517949036,
      "grad_norm": 1.0740379095077515,
      "learning_rate": 3.4830337158316216e-05,
      "loss": 1.4561,
      "step": 29270
    },
    {
      "epoch": 7.2614545229090455,
      "grad_norm": 1.0296998023986816,
      "learning_rate": 3.477125887218792e-05,
      "loss": 1.4657,
      "step": 29280
    },
    {
      "epoch": 7.263934527869056,
      "grad_norm": 1.0426205396652222,
      "learning_rate": 3.471222018628638e-05,
      "loss": 1.4201,
      "step": 29290
    },
    {
      "epoch": 7.2664145328290655,
      "grad_norm": 1.0202819108963013,
      "learning_rate": 3.4653221136453965e-05,
      "loss": 1.4121,
      "step": 29300
    },
    {
      "epoch": 7.268894537789076,
      "grad_norm": 1.0543783903121948,
      "learning_rate": 3.459426175850882e-05,
      "loss": 1.3914,
      "step": 29310
    },
    {
      "epoch": 7.2713745427490855,
      "grad_norm": 1.0744569301605225,
      "learning_rate": 3.453534208824507e-05,
      "loss": 1.4069,
      "step": 29320
    },
    {
      "epoch": 7.273854547709095,
      "grad_norm": 0.9860808253288269,
      "learning_rate": 3.447646216143268e-05,
      "loss": 1.3699,
      "step": 29330
    },
    {
      "epoch": 7.2763345526691054,
      "grad_norm": 1.036865234375,
      "learning_rate": 3.4417622013817595e-05,
      "loss": 1.4141,
      "step": 29340
    },
    {
      "epoch": 7.278814557629115,
      "grad_norm": 1.0248973369598389,
      "learning_rate": 3.43588216811215e-05,
      "loss": 1.4487,
      "step": 29350
    },
    {
      "epoch": 7.281294562589125,
      "grad_norm": 1.03071928024292,
      "learning_rate": 3.430006119904197e-05,
      "loss": 1.4971,
      "step": 29360
    },
    {
      "epoch": 7.283774567549135,
      "grad_norm": 1.1092323064804077,
      "learning_rate": 3.424134060325233e-05,
      "loss": 1.3836,
      "step": 29370
    },
    {
      "epoch": 7.286254572509145,
      "grad_norm": 1.075713872909546,
      "learning_rate": 3.418265992940175e-05,
      "loss": 1.3099,
      "step": 29380
    },
    {
      "epoch": 7.288734577469155,
      "grad_norm": 1.04108726978302,
      "learning_rate": 3.412401921311511e-05,
      "loss": 1.4589,
      "step": 29390
    },
    {
      "epoch": 7.2912145824291645,
      "grad_norm": 0.9846664071083069,
      "learning_rate": 3.406541848999312e-05,
      "loss": 1.3991,
      "step": 29400
    },
    {
      "epoch": 7.293694587389175,
      "grad_norm": 1.0906990766525269,
      "learning_rate": 3.4006857795612115e-05,
      "loss": 1.4416,
      "step": 29410
    },
    {
      "epoch": 7.2961745923491845,
      "grad_norm": 0.999430239200592,
      "learning_rate": 3.394833716552416e-05,
      "loss": 1.4222,
      "step": 29420
    },
    {
      "epoch": 7.298654597309195,
      "grad_norm": 1.1052192449569702,
      "learning_rate": 3.388985663525702e-05,
      "loss": 1.5021,
      "step": 29430
    },
    {
      "epoch": 7.3011346022692045,
      "grad_norm": 0.9980838894844055,
      "learning_rate": 3.383141624031408e-05,
      "loss": 1.3648,
      "step": 29440
    },
    {
      "epoch": 7.303614607229214,
      "grad_norm": 1.0518790483474731,
      "learning_rate": 3.3773016016174366e-05,
      "loss": 1.4328,
      "step": 29450
    },
    {
      "epoch": 7.3060946121892245,
      "grad_norm": 1.0774750709533691,
      "learning_rate": 3.371465599829257e-05,
      "loss": 1.4478,
      "step": 29460
    },
    {
      "epoch": 7.308574617149234,
      "grad_norm": 1.0285404920578003,
      "learning_rate": 3.365633622209891e-05,
      "loss": 1.3548,
      "step": 29470
    },
    {
      "epoch": 7.3110546221092445,
      "grad_norm": 1.0568714141845703,
      "learning_rate": 3.359805672299918e-05,
      "loss": 1.4449,
      "step": 29480
    },
    {
      "epoch": 7.313534627069254,
      "grad_norm": 1.0464444160461426,
      "learning_rate": 3.353981753637473e-05,
      "loss": 1.3494,
      "step": 29490
    },
    {
      "epoch": 7.316014632029264,
      "grad_norm": 1.117560625076294,
      "learning_rate": 3.3481618697582526e-05,
      "loss": 1.4988,
      "step": 29500
    },
    {
      "epoch": 7.318494636989274,
      "grad_norm": 0.9742351174354553,
      "learning_rate": 3.342346024195483e-05,
      "loss": 1.4054,
      "step": 29510
    },
    {
      "epoch": 7.320974641949284,
      "grad_norm": 1.0255587100982666,
      "learning_rate": 3.336534220479961e-05,
      "loss": 1.4537,
      "step": 29520
    },
    {
      "epoch": 7.323454646909294,
      "grad_norm": 0.9579190015792847,
      "learning_rate": 3.330726462140017e-05,
      "loss": 1.407,
      "step": 29530
    },
    {
      "epoch": 7.325934651869304,
      "grad_norm": 1.1523399353027344,
      "learning_rate": 3.324922752701528e-05,
      "loss": 1.3922,
      "step": 29540
    },
    {
      "epoch": 7.328414656829314,
      "grad_norm": 1.0771806240081787,
      "learning_rate": 3.3191230956879104e-05,
      "loss": 1.4265,
      "step": 29550
    },
    {
      "epoch": 7.330894661789324,
      "grad_norm": 1.105000376701355,
      "learning_rate": 3.3133274946201334e-05,
      "loss": 1.4276,
      "step": 29560
    },
    {
      "epoch": 7.333374666749333,
      "grad_norm": 1.0332446098327637,
      "learning_rate": 3.307535953016679e-05,
      "loss": 1.4173,
      "step": 29570
    },
    {
      "epoch": 7.335854671709344,
      "grad_norm": 1.025925874710083,
      "learning_rate": 3.301748474393592e-05,
      "loss": 1.4031,
      "step": 29580
    },
    {
      "epoch": 7.338334676669353,
      "grad_norm": 0.991253674030304,
      "learning_rate": 3.295965062264434e-05,
      "loss": 1.4386,
      "step": 29590
    },
    {
      "epoch": 7.340814681629364,
      "grad_norm": 1.0498908758163452,
      "learning_rate": 3.290185720140301e-05,
      "loss": 1.4447,
      "step": 29600
    },
    {
      "epoch": 7.343294686589373,
      "grad_norm": 1.1031231880187988,
      "learning_rate": 3.2844104515298155e-05,
      "loss": 1.4773,
      "step": 29610
    },
    {
      "epoch": 7.345774691549384,
      "grad_norm": 1.076324701309204,
      "learning_rate": 3.278639259939138e-05,
      "loss": 1.4199,
      "step": 29620
    },
    {
      "epoch": 7.348254696509393,
      "grad_norm": 1.0134528875350952,
      "learning_rate": 3.272872148871944e-05,
      "loss": 1.4166,
      "step": 29630
    },
    {
      "epoch": 7.350734701469403,
      "grad_norm": 1.0191160440444946,
      "learning_rate": 3.2671091218294284e-05,
      "loss": 1.3918,
      "step": 29640
    },
    {
      "epoch": 7.353214706429413,
      "grad_norm": 1.0939322710037231,
      "learning_rate": 3.261350182310324e-05,
      "loss": 1.4418,
      "step": 29650
    },
    {
      "epoch": 7.355694711389423,
      "grad_norm": 1.0721194744110107,
      "learning_rate": 3.255595333810857e-05,
      "loss": 1.4617,
      "step": 29660
    },
    {
      "epoch": 7.358174716349433,
      "grad_norm": 1.0546571016311646,
      "learning_rate": 3.2498445798247926e-05,
      "loss": 1.4557,
      "step": 29670
    },
    {
      "epoch": 7.360654721309443,
      "grad_norm": 0.9861288666725159,
      "learning_rate": 3.244097923843398e-05,
      "loss": 1.4372,
      "step": 29680
    },
    {
      "epoch": 7.363134726269452,
      "grad_norm": 0.9901741743087769,
      "learning_rate": 3.238355369355456e-05,
      "loss": 1.499,
      "step": 29690
    },
    {
      "epoch": 7.365614731229463,
      "grad_norm": 1.1363624334335327,
      "learning_rate": 3.2326169198472556e-05,
      "loss": 1.4539,
      "step": 29700
    },
    {
      "epoch": 7.368094736189472,
      "grad_norm": 1.0030924081802368,
      "learning_rate": 3.226882578802607e-05,
      "loss": 1.4246,
      "step": 29710
    },
    {
      "epoch": 7.370574741149483,
      "grad_norm": 1.200905680656433,
      "learning_rate": 3.2211523497028015e-05,
      "loss": 1.4518,
      "step": 29720
    },
    {
      "epoch": 7.373054746109492,
      "grad_norm": 1.0047590732574463,
      "learning_rate": 3.2154262360266607e-05,
      "loss": 1.437,
      "step": 29730
    },
    {
      "epoch": 7.375534751069502,
      "grad_norm": 1.0192116498947144,
      "learning_rate": 3.209704241250492e-05,
      "loss": 1.3976,
      "step": 29740
    },
    {
      "epoch": 7.378014756029512,
      "grad_norm": 1.0943766832351685,
      "learning_rate": 3.2039863688481055e-05,
      "loss": 1.4706,
      "step": 29750
    },
    {
      "epoch": 7.380494760989522,
      "grad_norm": 0.9721190929412842,
      "learning_rate": 3.198272622290804e-05,
      "loss": 1.4195,
      "step": 29760
    },
    {
      "epoch": 7.382974765949532,
      "grad_norm": 0.9810141324996948,
      "learning_rate": 3.192563005047404e-05,
      "loss": 1.4659,
      "step": 29770
    },
    {
      "epoch": 7.385454770909542,
      "grad_norm": 1.0326452255249023,
      "learning_rate": 3.186857520584187e-05,
      "loss": 1.4838,
      "step": 29780
    },
    {
      "epoch": 7.387934775869551,
      "grad_norm": 1.0693917274475098,
      "learning_rate": 3.18115617236495e-05,
      "loss": 1.4175,
      "step": 29790
    },
    {
      "epoch": 7.390414780829562,
      "grad_norm": 0.9504159092903137,
      "learning_rate": 3.1754589638509644e-05,
      "loss": 1.389,
      "step": 29800
    },
    {
      "epoch": 7.392894785789571,
      "grad_norm": 0.9387778639793396,
      "learning_rate": 3.169765898500995e-05,
      "loss": 1.4293,
      "step": 29810
    },
    {
      "epoch": 7.395374790749582,
      "grad_norm": 1.0295659303665161,
      "learning_rate": 3.164076979771287e-05,
      "loss": 1.4512,
      "step": 29820
    },
    {
      "epoch": 7.397854795709591,
      "grad_norm": 1.0724929571151733,
      "learning_rate": 3.1583922111155776e-05,
      "loss": 1.4934,
      "step": 29830
    },
    {
      "epoch": 7.400334800669602,
      "grad_norm": 1.1034655570983887,
      "learning_rate": 3.152711595985065e-05,
      "loss": 1.4806,
      "step": 29840
    },
    {
      "epoch": 7.402814805629611,
      "grad_norm": 1.030177116394043,
      "learning_rate": 3.14703513782845e-05,
      "loss": 1.4063,
      "step": 29850
    },
    {
      "epoch": 7.405294810589621,
      "grad_norm": 1.000606656074524,
      "learning_rate": 3.1413628400918926e-05,
      "loss": 1.4082,
      "step": 29860
    },
    {
      "epoch": 7.407774815549631,
      "grad_norm": 1.1129791736602783,
      "learning_rate": 3.1356947062190345e-05,
      "loss": 1.5047,
      "step": 29870
    },
    {
      "epoch": 7.410254820509641,
      "grad_norm": 0.9959533214569092,
      "learning_rate": 3.130030739650983e-05,
      "loss": 1.467,
      "step": 29880
    },
    {
      "epoch": 7.412734825469651,
      "grad_norm": 0.9653241038322449,
      "learning_rate": 3.1243709438263255e-05,
      "loss": 1.4495,
      "step": 29890
    },
    {
      "epoch": 7.415214830429661,
      "grad_norm": 1.1216167211532593,
      "learning_rate": 3.1187153221811105e-05,
      "loss": 1.4328,
      "step": 29900
    },
    {
      "epoch": 7.417694835389671,
      "grad_norm": 1.0418484210968018,
      "learning_rate": 3.113063878148852e-05,
      "loss": 1.3936,
      "step": 29910
    },
    {
      "epoch": 7.420174840349681,
      "grad_norm": 0.994735836982727,
      "learning_rate": 3.10741661516053e-05,
      "loss": 1.4255,
      "step": 29920
    },
    {
      "epoch": 7.42265484530969,
      "grad_norm": 1.0060540437698364,
      "learning_rate": 3.1017735366445855e-05,
      "loss": 1.4469,
      "step": 29930
    },
    {
      "epoch": 7.425134850269701,
      "grad_norm": 1.046750783920288,
      "learning_rate": 3.096134646026917e-05,
      "loss": 1.4321,
      "step": 29940
    },
    {
      "epoch": 7.42761485522971,
      "grad_norm": 1.0146149396896362,
      "learning_rate": 3.090499946730887e-05,
      "loss": 1.4625,
      "step": 29950
    },
    {
      "epoch": 7.430094860189721,
      "grad_norm": 0.9463431239128113,
      "learning_rate": 3.0848694421773075e-05,
      "loss": 1.4214,
      "step": 29960
    },
    {
      "epoch": 7.43257486514973,
      "grad_norm": 1.04469633102417,
      "learning_rate": 3.079243135784444e-05,
      "loss": 1.4258,
      "step": 29970
    },
    {
      "epoch": 7.43505487010974,
      "grad_norm": 1.025856852531433,
      "learning_rate": 3.073621030968017e-05,
      "loss": 1.527,
      "step": 29980
    },
    {
      "epoch": 7.43753487506975,
      "grad_norm": 0.9592713713645935,
      "learning_rate": 3.068003131141193e-05,
      "loss": 1.467,
      "step": 29990
    },
    {
      "epoch": 7.44001488002976,
      "grad_norm": 1.049975872039795,
      "learning_rate": 3.062389439714584e-05,
      "loss": 1.4338,
      "step": 30000
    },
    {
      "epoch": 7.44249488498977,
      "grad_norm": 1.0892877578735352,
      "learning_rate": 3.056779960096256e-05,
      "loss": 1.4806,
      "step": 30010
    },
    {
      "epoch": 7.44497488994978,
      "grad_norm": 1.0573172569274902,
      "learning_rate": 3.0511746956917085e-05,
      "loss": 1.4326,
      "step": 30020
    },
    {
      "epoch": 7.447454894909789,
      "grad_norm": 0.9932147860527039,
      "learning_rate": 3.0455736499038845e-05,
      "loss": 1.447,
      "step": 30030
    },
    {
      "epoch": 7.4499348998698,
      "grad_norm": 1.0222856998443604,
      "learning_rate": 3.0399768261331662e-05,
      "loss": 1.4302,
      "step": 30040
    },
    {
      "epoch": 7.452414904829809,
      "grad_norm": 1.116368055343628,
      "learning_rate": 3.0343842277773793e-05,
      "loss": 1.4759,
      "step": 30050
    },
    {
      "epoch": 7.45489490978982,
      "grad_norm": 0.9921149015426636,
      "learning_rate": 3.0287958582317676e-05,
      "loss": 1.3803,
      "step": 30060
    },
    {
      "epoch": 7.457374914749829,
      "grad_norm": 1.152550220489502,
      "learning_rate": 3.0232117208890276e-05,
      "loss": 1.4771,
      "step": 30070
    },
    {
      "epoch": 7.45985491970984,
      "grad_norm": 0.9995396137237549,
      "learning_rate": 3.0176318191392726e-05,
      "loss": 1.4224,
      "step": 30080
    },
    {
      "epoch": 7.462334924669849,
      "grad_norm": 1.037562608718872,
      "learning_rate": 3.01205615637005e-05,
      "loss": 1.499,
      "step": 30090
    },
    {
      "epoch": 7.464814929629859,
      "grad_norm": 1.0492740869522095,
      "learning_rate": 3.0064847359663284e-05,
      "loss": 1.4988,
      "step": 30100
    },
    {
      "epoch": 7.467294934589869,
      "grad_norm": 1.0149956941604614,
      "learning_rate": 3.0009175613105147e-05,
      "loss": 1.4181,
      "step": 30110
    },
    {
      "epoch": 7.469774939549879,
      "grad_norm": 0.9838835000991821,
      "learning_rate": 2.995354635782417e-05,
      "loss": 1.4438,
      "step": 30120
    },
    {
      "epoch": 7.472254944509889,
      "grad_norm": 1.0166876316070557,
      "learning_rate": 2.989795962759283e-05,
      "loss": 1.4298,
      "step": 30130
    },
    {
      "epoch": 7.474734949469899,
      "grad_norm": 1.0134931802749634,
      "learning_rate": 2.984241545615768e-05,
      "loss": 1.4095,
      "step": 30140
    },
    {
      "epoch": 7.4772149544299085,
      "grad_norm": 1.0498225688934326,
      "learning_rate": 2.9786913877239487e-05,
      "loss": 1.413,
      "step": 30150
    },
    {
      "epoch": 7.479694959389919,
      "grad_norm": 1.09099280834198,
      "learning_rate": 2.9731454924533086e-05,
      "loss": 1.4469,
      "step": 30160
    },
    {
      "epoch": 7.4821749643499285,
      "grad_norm": 0.9902943968772888,
      "learning_rate": 2.9676038631707593e-05,
      "loss": 1.421,
      "step": 30170
    },
    {
      "epoch": 7.484654969309939,
      "grad_norm": 1.0532760620117188,
      "learning_rate": 2.962066503240598e-05,
      "loss": 1.4642,
      "step": 30180
    },
    {
      "epoch": 7.4871349742699485,
      "grad_norm": 0.9733676314353943,
      "learning_rate": 2.9565334160245538e-05,
      "loss": 1.4042,
      "step": 30190
    },
    {
      "epoch": 7.489614979229959,
      "grad_norm": 1.0446985960006714,
      "learning_rate": 2.9510046048817497e-05,
      "loss": 1.4819,
      "step": 30200
    },
    {
      "epoch": 7.492094984189968,
      "grad_norm": 1.1242960691452026,
      "learning_rate": 2.945480073168714e-05,
      "loss": 1.473,
      "step": 30210
    },
    {
      "epoch": 7.494574989149978,
      "grad_norm": 0.9344368577003479,
      "learning_rate": 2.9399598242393755e-05,
      "loss": 1.4249,
      "step": 30220
    },
    {
      "epoch": 7.497054994109988,
      "grad_norm": 1.0807608366012573,
      "learning_rate": 2.934443861445073e-05,
      "loss": 1.475,
      "step": 30230
    },
    {
      "epoch": 7.499534999069998,
      "grad_norm": 1.0247372388839722,
      "learning_rate": 2.9289321881345254e-05,
      "loss": 1.4569,
      "step": 30240
    },
    {
      "epoch": 7.502015004030008,
      "grad_norm": 0.9852992296218872,
      "learning_rate": 2.923424807653863e-05,
      "loss": 1.4907,
      "step": 30250
    },
    {
      "epoch": 7.504495008990018,
      "grad_norm": 1.0893725156784058,
      "learning_rate": 2.9179217233466095e-05,
      "loss": 1.409,
      "step": 30260
    },
    {
      "epoch": 7.5069750139500275,
      "grad_norm": 1.003062129020691,
      "learning_rate": 2.912422938553664e-05,
      "loss": 1.4518,
      "step": 30270
    },
    {
      "epoch": 7.509455018910038,
      "grad_norm": 1.0472670793533325,
      "learning_rate": 2.906928456613336e-05,
      "loss": 1.4337,
      "step": 30280
    },
    {
      "epoch": 7.5119350238700475,
      "grad_norm": 1.045111894607544,
      "learning_rate": 2.9014382808613093e-05,
      "loss": 1.4733,
      "step": 30290
    },
    {
      "epoch": 7.514415028830058,
      "grad_norm": 1.0085594654083252,
      "learning_rate": 2.8959524146306603e-05,
      "loss": 1.4333,
      "step": 30300
    },
    {
      "epoch": 7.5168950337900675,
      "grad_norm": 1.0071136951446533,
      "learning_rate": 2.8904708612518404e-05,
      "loss": 1.4055,
      "step": 30310
    },
    {
      "epoch": 7.519375038750077,
      "grad_norm": 1.0234194993972778,
      "learning_rate": 2.8849936240527008e-05,
      "loss": 1.448,
      "step": 30320
    },
    {
      "epoch": 7.5218550437100875,
      "grad_norm": 1.0164966583251953,
      "learning_rate": 2.879520706358446e-05,
      "loss": 1.4866,
      "step": 30330
    },
    {
      "epoch": 7.524335048670097,
      "grad_norm": 0.995505690574646,
      "learning_rate": 2.874052111491682e-05,
      "loss": 1.4713,
      "step": 30340
    },
    {
      "epoch": 7.5268150536301075,
      "grad_norm": 1.1226674318313599,
      "learning_rate": 2.8685878427723777e-05,
      "loss": 1.4977,
      "step": 30350
    },
    {
      "epoch": 7.529295058590117,
      "grad_norm": 1.1339809894561768,
      "learning_rate": 2.8631279035178793e-05,
      "loss": 1.4647,
      "step": 30360
    },
    {
      "epoch": 7.531775063550127,
      "grad_norm": 1.034926176071167,
      "learning_rate": 2.8576722970429016e-05,
      "loss": 1.4658,
      "step": 30370
    },
    {
      "epoch": 7.534255068510137,
      "grad_norm": 0.9799590706825256,
      "learning_rate": 2.8522210266595384e-05,
      "loss": 1.4583,
      "step": 30380
    },
    {
      "epoch": 7.536735073470147,
      "grad_norm": 1.005009651184082,
      "learning_rate": 2.846774095677234e-05,
      "loss": 1.5044,
      "step": 30390
    },
    {
      "epoch": 7.539215078430157,
      "grad_norm": 0.9981338977813721,
      "learning_rate": 2.8413315074028158e-05,
      "loss": 1.4273,
      "step": 30400
    },
    {
      "epoch": 7.541695083390167,
      "grad_norm": 1.0441522598266602,
      "learning_rate": 2.835893265140466e-05,
      "loss": 1.4125,
      "step": 30410
    },
    {
      "epoch": 7.544175088350177,
      "grad_norm": 1.0379878282546997,
      "learning_rate": 2.8304593721917285e-05,
      "loss": 1.4184,
      "step": 30420
    },
    {
      "epoch": 7.546655093310187,
      "grad_norm": 1.0805764198303223,
      "learning_rate": 2.8250298318555045e-05,
      "loss": 1.4633,
      "step": 30430
    },
    {
      "epoch": 7.549135098270197,
      "grad_norm": 1.0463159084320068,
      "learning_rate": 2.8196046474280667e-05,
      "loss": 1.5028,
      "step": 30440
    },
    {
      "epoch": 7.551615103230207,
      "grad_norm": 0.99134761095047,
      "learning_rate": 2.814183822203019e-05,
      "loss": 1.3978,
      "step": 30450
    },
    {
      "epoch": 7.554095108190216,
      "grad_norm": 0.9535025358200073,
      "learning_rate": 2.8087673594713438e-05,
      "loss": 1.4794,
      "step": 30460
    },
    {
      "epoch": 7.556575113150227,
      "grad_norm": 0.9892309904098511,
      "learning_rate": 2.803355262521359e-05,
      "loss": 1.3823,
      "step": 30470
    },
    {
      "epoch": 7.559055118110236,
      "grad_norm": 1.0789638757705688,
      "learning_rate": 2.797947534638736e-05,
      "loss": 1.3913,
      "step": 30480
    },
    {
      "epoch": 7.5615351230702466,
      "grad_norm": 0.9983831644058228,
      "learning_rate": 2.7925441791064945e-05,
      "loss": 1.4396,
      "step": 30490
    },
    {
      "epoch": 7.564015128030256,
      "grad_norm": 1.0396370887756348,
      "learning_rate": 2.7871451992050034e-05,
      "loss": 1.4401,
      "step": 30500
    },
    {
      "epoch": 7.566495132990266,
      "grad_norm": 1.1130553483963013,
      "learning_rate": 2.7817505982119706e-05,
      "loss": 1.4458,
      "step": 30510
    },
    {
      "epoch": 7.568975137950276,
      "grad_norm": 1.0616374015808105,
      "learning_rate": 2.776360379402445e-05,
      "loss": 1.434,
      "step": 30520
    },
    {
      "epoch": 7.571455142910286,
      "grad_norm": 1.1154457330703735,
      "learning_rate": 2.7709745460488177e-05,
      "loss": 1.4303,
      "step": 30530
    },
    {
      "epoch": 7.573935147870296,
      "grad_norm": 1.104222297668457,
      "learning_rate": 2.765593101420816e-05,
      "loss": 1.4674,
      "step": 30540
    },
    {
      "epoch": 7.576415152830306,
      "grad_norm": 0.9552186727523804,
      "learning_rate": 2.7602160487855e-05,
      "loss": 1.4183,
      "step": 30550
    },
    {
      "epoch": 7.578895157790315,
      "grad_norm": 1.0074481964111328,
      "learning_rate": 2.7548433914072734e-05,
      "loss": 1.4802,
      "step": 30560
    },
    {
      "epoch": 7.581375162750326,
      "grad_norm": 1.0582637786865234,
      "learning_rate": 2.749475132547862e-05,
      "loss": 1.4327,
      "step": 30570
    },
    {
      "epoch": 7.583855167710335,
      "grad_norm": 1.156577706336975,
      "learning_rate": 2.7441112754663222e-05,
      "loss": 1.463,
      "step": 30580
    },
    {
      "epoch": 7.586335172670346,
      "grad_norm": 1.0950535535812378,
      "learning_rate": 2.738751823419041e-05,
      "loss": 1.4868,
      "step": 30590
    },
    {
      "epoch": 7.588815177630355,
      "grad_norm": 1.0288358926773071,
      "learning_rate": 2.7333967796597315e-05,
      "loss": 1.4739,
      "step": 30600
    },
    {
      "epoch": 7.591295182590365,
      "grad_norm": 0.9875478148460388,
      "learning_rate": 2.7280461474394247e-05,
      "loss": 1.4421,
      "step": 30610
    },
    {
      "epoch": 7.593775187550375,
      "grad_norm": 0.9938637018203735,
      "learning_rate": 2.7226999300064836e-05,
      "loss": 1.4035,
      "step": 30620
    },
    {
      "epoch": 7.596255192510385,
      "grad_norm": 0.9095118045806885,
      "learning_rate": 2.7173581306065833e-05,
      "loss": 1.4271,
      "step": 30630
    },
    {
      "epoch": 7.598735197470395,
      "grad_norm": 0.9821234345436096,
      "learning_rate": 2.7120207524827168e-05,
      "loss": 1.4309,
      "step": 30640
    },
    {
      "epoch": 7.601215202430405,
      "grad_norm": 1.0130341053009033,
      "learning_rate": 2.7066877988751938e-05,
      "loss": 1.4619,
      "step": 30650
    },
    {
      "epoch": 7.603695207390415,
      "grad_norm": 1.0122448205947876,
      "learning_rate": 2.7013592730216465e-05,
      "loss": 1.4547,
      "step": 30660
    },
    {
      "epoch": 7.606175212350425,
      "grad_norm": 1.0752753019332886,
      "learning_rate": 2.6960351781569992e-05,
      "loss": 1.3997,
      "step": 30670
    },
    {
      "epoch": 7.608655217310434,
      "grad_norm": 1.0285594463348389,
      "learning_rate": 2.690715517513508e-05,
      "loss": 1.4347,
      "step": 30680
    },
    {
      "epoch": 7.611135222270445,
      "grad_norm": 1.0548043251037598,
      "learning_rate": 2.6854002943207246e-05,
      "loss": 1.4606,
      "step": 30690
    },
    {
      "epoch": 7.613615227230454,
      "grad_norm": 1.1300123929977417,
      "learning_rate": 2.680089511805508e-05,
      "loss": 1.4533,
      "step": 30700
    },
    {
      "epoch": 7.616095232190465,
      "grad_norm": 0.9323368668556213,
      "learning_rate": 2.6747831731920214e-05,
      "loss": 1.3743,
      "step": 30710
    },
    {
      "epoch": 7.618575237150474,
      "grad_norm": 1.109718918800354,
      "learning_rate": 2.669481281701739e-05,
      "loss": 1.4525,
      "step": 30720
    },
    {
      "epoch": 7.621055242110485,
      "grad_norm": 1.0525087118148804,
      "learning_rate": 2.664183840553417e-05,
      "loss": 1.4265,
      "step": 30730
    },
    {
      "epoch": 7.623535247070494,
      "grad_norm": 0.99526447057724,
      "learning_rate": 2.6588908529631285e-05,
      "loss": 1.4605,
      "step": 30740
    },
    {
      "epoch": 7.626015252030504,
      "grad_norm": 1.0765104293823242,
      "learning_rate": 2.6536023221442318e-05,
      "loss": 1.5179,
      "step": 30750
    },
    {
      "epoch": 7.628495256990514,
      "grad_norm": 1.057417631149292,
      "learning_rate": 2.6483182513073835e-05,
      "loss": 1.5021,
      "step": 30760
    },
    {
      "epoch": 7.630975261950524,
      "grad_norm": 1.0523009300231934,
      "learning_rate": 2.643038643660526e-05,
      "loss": 1.4259,
      "step": 30770
    },
    {
      "epoch": 7.633455266910534,
      "grad_norm": 1.0764293670654297,
      "learning_rate": 2.6377635024089087e-05,
      "loss": 1.4572,
      "step": 30780
    },
    {
      "epoch": 7.635935271870544,
      "grad_norm": 0.9474040865898132,
      "learning_rate": 2.6324928307550466e-05,
      "loss": 1.4008,
      "step": 30790
    },
    {
      "epoch": 7.638415276830553,
      "grad_norm": 1.0371317863464355,
      "learning_rate": 2.6272266318987603e-05,
      "loss": 1.4276,
      "step": 30800
    },
    {
      "epoch": 7.640895281790564,
      "grad_norm": 0.966749370098114,
      "learning_rate": 2.6219649090371466e-05,
      "loss": 1.4593,
      "step": 30810
    },
    {
      "epoch": 7.643375286750573,
      "grad_norm": 1.0057498216629028,
      "learning_rate": 2.616707665364584e-05,
      "loss": 1.4272,
      "step": 30820
    },
    {
      "epoch": 7.645855291710584,
      "grad_norm": 1.0172383785247803,
      "learning_rate": 2.6114549040727332e-05,
      "loss": 1.4493,
      "step": 30830
    },
    {
      "epoch": 7.648335296670593,
      "grad_norm": 0.9629692435264587,
      "learning_rate": 2.6062066283505403e-05,
      "loss": 1.417,
      "step": 30840
    },
    {
      "epoch": 7.650815301630603,
      "grad_norm": 1.0054121017456055,
      "learning_rate": 2.6009628413842123e-05,
      "loss": 1.4327,
      "step": 30850
    },
    {
      "epoch": 7.653295306590613,
      "grad_norm": 1.0924439430236816,
      "learning_rate": 2.5957235463572484e-05,
      "loss": 1.4825,
      "step": 30860
    },
    {
      "epoch": 7.655775311550623,
      "grad_norm": 1.0084000825881958,
      "learning_rate": 2.5904887464504114e-05,
      "loss": 1.4974,
      "step": 30870
    },
    {
      "epoch": 7.658255316510633,
      "grad_norm": 1.0829739570617676,
      "learning_rate": 2.5852584448417328e-05,
      "loss": 1.3917,
      "step": 30880
    },
    {
      "epoch": 7.660735321470643,
      "grad_norm": 0.9569641351699829,
      "learning_rate": 2.580032644706524e-05,
      "loss": 1.4001,
      "step": 30890
    },
    {
      "epoch": 7.663215326430652,
      "grad_norm": 1.0015615224838257,
      "learning_rate": 2.5748113492173543e-05,
      "loss": 1.4117,
      "step": 30900
    },
    {
      "epoch": 7.665695331390663,
      "grad_norm": 0.9971137046813965,
      "learning_rate": 2.5695945615440596e-05,
      "loss": 1.435,
      "step": 30910
    },
    {
      "epoch": 7.668175336350672,
      "grad_norm": 1.0712945461273193,
      "learning_rate": 2.564382284853738e-05,
      "loss": 1.4487,
      "step": 30920
    },
    {
      "epoch": 7.670655341310683,
      "grad_norm": 1.0099509954452515,
      "learning_rate": 2.559174522310761e-05,
      "loss": 1.4838,
      "step": 30930
    },
    {
      "epoch": 7.673135346270692,
      "grad_norm": 1.0410501956939697,
      "learning_rate": 2.5539712770767376e-05,
      "loss": 1.3944,
      "step": 30940
    },
    {
      "epoch": 7.675615351230703,
      "grad_norm": 1.0309990644454956,
      "learning_rate": 2.5487725523105544e-05,
      "loss": 1.3999,
      "step": 30950
    },
    {
      "epoch": 7.678095356190712,
      "grad_norm": 1.100633144378662,
      "learning_rate": 2.5435783511683443e-05,
      "loss": 1.4263,
      "step": 30960
    },
    {
      "epoch": 7.680575361150722,
      "grad_norm": 1.046830415725708,
      "learning_rate": 2.538388676803496e-05,
      "loss": 1.4382,
      "step": 30970
    },
    {
      "epoch": 7.683055366110732,
      "grad_norm": 1.0287320613861084,
      "learning_rate": 2.5332035323666447e-05,
      "loss": 1.4257,
      "step": 30980
    },
    {
      "epoch": 7.685535371070742,
      "grad_norm": 1.0454462766647339,
      "learning_rate": 2.5280229210056895e-05,
      "loss": 1.4422,
      "step": 30990
    },
    {
      "epoch": 7.688015376030752,
      "grad_norm": 1.068619728088379,
      "learning_rate": 2.5228468458657584e-05,
      "loss": 1.4173,
      "step": 31000
    },
    {
      "epoch": 7.690495380990762,
      "grad_norm": 1.054749608039856,
      "learning_rate": 2.5176753100892426e-05,
      "loss": 1.4263,
      "step": 31010
    },
    {
      "epoch": 7.692975385950772,
      "grad_norm": 1.009238839149475,
      "learning_rate": 2.512508316815767e-05,
      "loss": 1.4302,
      "step": 31020
    },
    {
      "epoch": 7.695455390910782,
      "grad_norm": 1.03357994556427,
      "learning_rate": 2.507345869182203e-05,
      "loss": 1.38,
      "step": 31030
    },
    {
      "epoch": 7.6979353958707915,
      "grad_norm": 1.0613250732421875,
      "learning_rate": 2.502187970322657e-05,
      "loss": 1.38,
      "step": 31040
    },
    {
      "epoch": 7.700415400830802,
      "grad_norm": 1.0838817358016968,
      "learning_rate": 2.497034623368486e-05,
      "loss": 1.4417,
      "step": 31050
    },
    {
      "epoch": 7.7028954057908114,
      "grad_norm": 1.040667176246643,
      "learning_rate": 2.4918858314482718e-05,
      "loss": 1.4634,
      "step": 31060
    },
    {
      "epoch": 7.705375410750822,
      "grad_norm": 1.1136744022369385,
      "learning_rate": 2.4867415976878337e-05,
      "loss": 1.4517,
      "step": 31070
    },
    {
      "epoch": 7.707855415710831,
      "grad_norm": 1.0463638305664062,
      "learning_rate": 2.4816019252102273e-05,
      "loss": 1.4042,
      "step": 31080
    },
    {
      "epoch": 7.710335420670841,
      "grad_norm": 0.9802637696266174,
      "learning_rate": 2.4764668171357343e-05,
      "loss": 1.3391,
      "step": 31090
    },
    {
      "epoch": 7.712815425630851,
      "grad_norm": 1.0218875408172607,
      "learning_rate": 2.4713362765818658e-05,
      "loss": 1.5236,
      "step": 31100
    },
    {
      "epoch": 7.715295430590861,
      "grad_norm": 1.0065280199050903,
      "learning_rate": 2.466210306663368e-05,
      "loss": 1.4363,
      "step": 31110
    },
    {
      "epoch": 7.717775435550871,
      "grad_norm": 1.0505353212356567,
      "learning_rate": 2.461088910492202e-05,
      "loss": 1.4227,
      "step": 31120
    },
    {
      "epoch": 7.720255440510881,
      "grad_norm": 1.0982780456542969,
      "learning_rate": 2.4559720911775564e-05,
      "loss": 1.4404,
      "step": 31130
    },
    {
      "epoch": 7.7227354454708905,
      "grad_norm": 1.084692358970642,
      "learning_rate": 2.450859851825842e-05,
      "loss": 1.4475,
      "step": 31140
    },
    {
      "epoch": 7.725215450430901,
      "grad_norm": 0.9937741160392761,
      "learning_rate": 2.445752195540687e-05,
      "loss": 1.4108,
      "step": 31150
    },
    {
      "epoch": 7.7276954553909105,
      "grad_norm": 1.0374422073364258,
      "learning_rate": 2.440649125422937e-05,
      "loss": 1.4338,
      "step": 31160
    },
    {
      "epoch": 7.730175460350921,
      "grad_norm": 1.0560063123703003,
      "learning_rate": 2.43555064457066e-05,
      "loss": 1.4561,
      "step": 31170
    },
    {
      "epoch": 7.7326554653109305,
      "grad_norm": 1.0604078769683838,
      "learning_rate": 2.4304567560791293e-05,
      "loss": 1.4373,
      "step": 31180
    },
    {
      "epoch": 7.73513547027094,
      "grad_norm": 1.0357143878936768,
      "learning_rate": 2.425367463040834e-05,
      "loss": 1.4736,
      "step": 31190
    },
    {
      "epoch": 7.7376154752309505,
      "grad_norm": 1.0199404954910278,
      "learning_rate": 2.420282768545469e-05,
      "loss": 1.4684,
      "step": 31200
    },
    {
      "epoch": 7.74009548019096,
      "grad_norm": 1.1008087396621704,
      "learning_rate": 2.4152026756799506e-05,
      "loss": 1.4925,
      "step": 31210
    },
    {
      "epoch": 7.7425754851509705,
      "grad_norm": 0.9964824318885803,
      "learning_rate": 2.4101271875283817e-05,
      "loss": 1.4495,
      "step": 31220
    },
    {
      "epoch": 7.74505549011098,
      "grad_norm": 1.0641846656799316,
      "learning_rate": 2.4050563071720867e-05,
      "loss": 1.4402,
      "step": 31230
    },
    {
      "epoch": 7.7475354950709905,
      "grad_norm": 1.1189098358154297,
      "learning_rate": 2.3999900376895845e-05,
      "loss": 1.4009,
      "step": 31240
    },
    {
      "epoch": 7.750015500031,
      "grad_norm": 1.0134433507919312,
      "learning_rate": 2.394928382156596e-05,
      "loss": 1.4331,
      "step": 31250
    },
    {
      "epoch": 7.7524955049910105,
      "grad_norm": 1.0753346681594849,
      "learning_rate": 2.3898713436460375e-05,
      "loss": 1.4288,
      "step": 31260
    },
    {
      "epoch": 7.75497550995102,
      "grad_norm": 1.0120718479156494,
      "learning_rate": 2.384818925228036e-05,
      "loss": 1.5383,
      "step": 31270
    },
    {
      "epoch": 7.75745551491103,
      "grad_norm": 1.036631464958191,
      "learning_rate": 2.3797711299698923e-05,
      "loss": 1.4406,
      "step": 31280
    },
    {
      "epoch": 7.75993551987104,
      "grad_norm": 1.0271037817001343,
      "learning_rate": 2.3747279609361196e-05,
      "loss": 1.5359,
      "step": 31290
    },
    {
      "epoch": 7.76241552483105,
      "grad_norm": 1.0515282154083252,
      "learning_rate": 2.3696894211884123e-05,
      "loss": 1.4693,
      "step": 31300
    },
    {
      "epoch": 7.76489552979106,
      "grad_norm": 1.0361008644104004,
      "learning_rate": 2.3646555137856576e-05,
      "loss": 1.4372,
      "step": 31310
    },
    {
      "epoch": 7.76737553475107,
      "grad_norm": 1.026922583580017,
      "learning_rate": 2.3596262417839255e-05,
      "loss": 1.432,
      "step": 31320
    },
    {
      "epoch": 7.769855539711079,
      "grad_norm": 1.0789787769317627,
      "learning_rate": 2.3546016082364852e-05,
      "loss": 1.388,
      "step": 31330
    },
    {
      "epoch": 7.77233554467109,
      "grad_norm": 0.9373264908790588,
      "learning_rate": 2.34958161619377e-05,
      "loss": 1.3714,
      "step": 31340
    },
    {
      "epoch": 7.774815549631099,
      "grad_norm": 0.9983879327774048,
      "learning_rate": 2.3445662687034144e-05,
      "loss": 1.4818,
      "step": 31350
    },
    {
      "epoch": 7.7772955545911095,
      "grad_norm": 1.1489768028259277,
      "learning_rate": 2.339555568810221e-05,
      "loss": 1.4882,
      "step": 31360
    },
    {
      "epoch": 7.779775559551119,
      "grad_norm": 1.3525007963180542,
      "learning_rate": 2.3345495195561762e-05,
      "loss": 1.5085,
      "step": 31370
    },
    {
      "epoch": 7.782255564511129,
      "grad_norm": 0.9903725981712341,
      "learning_rate": 2.3295481239804384e-05,
      "loss": 1.4513,
      "step": 31380
    },
    {
      "epoch": 7.784735569471139,
      "grad_norm": 0.9812037944793701,
      "learning_rate": 2.3245513851193534e-05,
      "loss": 1.4733,
      "step": 31390
    },
    {
      "epoch": 7.787215574431149,
      "grad_norm": 1.1686617136001587,
      "learning_rate": 2.319559306006417e-05,
      "loss": 1.4414,
      "step": 31400
    },
    {
      "epoch": 7.789695579391159,
      "grad_norm": 1.058807373046875,
      "learning_rate": 2.31457188967232e-05,
      "loss": 1.4858,
      "step": 31410
    },
    {
      "epoch": 7.792175584351169,
      "grad_norm": 1.070768117904663,
      "learning_rate": 2.3095891391449086e-05,
      "loss": 1.4402,
      "step": 31420
    },
    {
      "epoch": 7.794655589311178,
      "grad_norm": 1.0862927436828613,
      "learning_rate": 2.3046110574491985e-05,
      "loss": 1.4959,
      "step": 31430
    },
    {
      "epoch": 7.797135594271189,
      "grad_norm": 0.9778184294700623,
      "learning_rate": 2.2996376476073723e-05,
      "loss": 1.3462,
      "step": 31440
    },
    {
      "epoch": 7.799615599231198,
      "grad_norm": 1.089483618736267,
      "learning_rate": 2.294668912638781e-05,
      "loss": 1.482,
      "step": 31450
    },
    {
      "epoch": 7.802095604191209,
      "grad_norm": 1.0856155157089233,
      "learning_rate": 2.2897048555599253e-05,
      "loss": 1.4108,
      "step": 31460
    },
    {
      "epoch": 7.804575609151218,
      "grad_norm": 1.0545732975006104,
      "learning_rate": 2.2847454793844793e-05,
      "loss": 1.3956,
      "step": 31470
    },
    {
      "epoch": 7.807055614111228,
      "grad_norm": 1.056648850440979,
      "learning_rate": 2.279790787123267e-05,
      "loss": 1.4274,
      "step": 31480
    },
    {
      "epoch": 7.809535619071238,
      "grad_norm": 1.0907402038574219,
      "learning_rate": 2.274840781784273e-05,
      "loss": 1.4366,
      "step": 31490
    },
    {
      "epoch": 7.812015624031248,
      "grad_norm": 1.0658317804336548,
      "learning_rate": 2.26989546637263e-05,
      "loss": 1.456,
      "step": 31500
    },
    {
      "epoch": 7.814495628991258,
      "grad_norm": 0.9488449096679688,
      "learning_rate": 2.2649548438906353e-05,
      "loss": 1.4676,
      "step": 31510
    },
    {
      "epoch": 7.816975633951268,
      "grad_norm": 1.0413172245025635,
      "learning_rate": 2.260018917337726e-05,
      "loss": 1.506,
      "step": 31520
    },
    {
      "epoch": 7.819455638911278,
      "grad_norm": 1.0351229906082153,
      "learning_rate": 2.2550876897104913e-05,
      "loss": 1.4968,
      "step": 31530
    },
    {
      "epoch": 7.821935643871288,
      "grad_norm": 0.9866580963134766,
      "learning_rate": 2.2501611640026743e-05,
      "loss": 1.3963,
      "step": 31540
    },
    {
      "epoch": 7.824415648831298,
      "grad_norm": 1.0516616106033325,
      "learning_rate": 2.24523934320515e-05,
      "loss": 1.4826,
      "step": 31550
    },
    {
      "epoch": 7.826895653791308,
      "grad_norm": 0.9789044857025146,
      "learning_rate": 2.240322230305951e-05,
      "loss": 1.4426,
      "step": 31560
    },
    {
      "epoch": 7.829375658751317,
      "grad_norm": 1.0472681522369385,
      "learning_rate": 2.2354098282902446e-05,
      "loss": 1.4116,
      "step": 31570
    },
    {
      "epoch": 7.831855663711328,
      "grad_norm": 1.0835953950881958,
      "learning_rate": 2.2305021401403382e-05,
      "loss": 1.4867,
      "step": 31580
    },
    {
      "epoch": 7.834335668671337,
      "grad_norm": 1.1364092826843262,
      "learning_rate": 2.225599168835677e-05,
      "loss": 1.4628,
      "step": 31590
    },
    {
      "epoch": 7.836815673631348,
      "grad_norm": 0.95635586977005,
      "learning_rate": 2.2207009173528527e-05,
      "loss": 1.4281,
      "step": 31600
    },
    {
      "epoch": 7.839295678591357,
      "grad_norm": 1.0518940687179565,
      "learning_rate": 2.2158073886655705e-05,
      "loss": 1.4702,
      "step": 31610
    },
    {
      "epoch": 7.841775683551367,
      "grad_norm": 1.1285464763641357,
      "learning_rate": 2.2109185857446903e-05,
      "loss": 1.4512,
      "step": 31620
    },
    {
      "epoch": 7.844255688511377,
      "grad_norm": 1.1680338382720947,
      "learning_rate": 2.2060345115581914e-05,
      "loss": 1.4784,
      "step": 31630
    },
    {
      "epoch": 7.846735693471387,
      "grad_norm": 1.005530834197998,
      "learning_rate": 2.201155169071184e-05,
      "loss": 1.414,
      "step": 31640
    },
    {
      "epoch": 7.849215698431397,
      "grad_norm": 1.0277854204177856,
      "learning_rate": 2.1962805612459024e-05,
      "loss": 1.4305,
      "step": 31650
    },
    {
      "epoch": 7.851695703391407,
      "grad_norm": 0.9508567452430725,
      "learning_rate": 2.1914106910417186e-05,
      "loss": 1.4075,
      "step": 31660
    },
    {
      "epoch": 7.854175708351416,
      "grad_norm": 0.9974478483200073,
      "learning_rate": 2.1865455614151155e-05,
      "loss": 1.4251,
      "step": 31670
    },
    {
      "epoch": 7.856655713311427,
      "grad_norm": 1.0750144720077515,
      "learning_rate": 2.181685175319702e-05,
      "loss": 1.3923,
      "step": 31680
    },
    {
      "epoch": 7.859135718271436,
      "grad_norm": 0.9576234817504883,
      "learning_rate": 2.1768295357062107e-05,
      "loss": 1.448,
      "step": 31690
    },
    {
      "epoch": 7.861615723231447,
      "grad_norm": 1.0810132026672363,
      "learning_rate": 2.1719786455224878e-05,
      "loss": 1.4417,
      "step": 31700
    },
    {
      "epoch": 7.864095728191456,
      "grad_norm": 1.040737271308899,
      "learning_rate": 2.1671325077134963e-05,
      "loss": 1.425,
      "step": 31710
    },
    {
      "epoch": 7.866575733151466,
      "grad_norm": 1.0674554109573364,
      "learning_rate": 2.1622911252213197e-05,
      "loss": 1.4129,
      "step": 31720
    },
    {
      "epoch": 7.869055738111476,
      "grad_norm": 1.0067627429962158,
      "learning_rate": 2.15745450098515e-05,
      "loss": 1.418,
      "step": 31730
    },
    {
      "epoch": 7.871535743071486,
      "grad_norm": 1.0110206604003906,
      "learning_rate": 2.1526226379412906e-05,
      "loss": 1.4551,
      "step": 31740
    },
    {
      "epoch": 7.874015748031496,
      "grad_norm": 1.0148226022720337,
      "learning_rate": 2.147795539023153e-05,
      "loss": 1.4012,
      "step": 31750
    },
    {
      "epoch": 7.876495752991506,
      "grad_norm": 1.0703665018081665,
      "learning_rate": 2.142973207161265e-05,
      "loss": 1.3901,
      "step": 31760
    },
    {
      "epoch": 7.878975757951516,
      "grad_norm": 1.0552300214767456,
      "learning_rate": 2.138155645283244e-05,
      "loss": 1.481,
      "step": 31770
    },
    {
      "epoch": 7.881455762911526,
      "grad_norm": 1.0581234693527222,
      "learning_rate": 2.1333428563138303e-05,
      "loss": 1.3928,
      "step": 31780
    },
    {
      "epoch": 7.883935767871535,
      "grad_norm": 1.0785514116287231,
      "learning_rate": 2.128534843174853e-05,
      "loss": 1.4399,
      "step": 31790
    },
    {
      "epoch": 7.886415772831546,
      "grad_norm": 1.0023902654647827,
      "learning_rate": 2.1237316087852466e-05,
      "loss": 1.4149,
      "step": 31800
    },
    {
      "epoch": 7.888895777791555,
      "grad_norm": 1.0342328548431396,
      "learning_rate": 2.1189331560610435e-05,
      "loss": 1.4297,
      "step": 31810
    },
    {
      "epoch": 7.891375782751566,
      "grad_norm": 1.1631563901901245,
      "learning_rate": 2.114139487915381e-05,
      "loss": 1.5044,
      "step": 31820
    },
    {
      "epoch": 7.893855787711575,
      "grad_norm": 0.9945201277732849,
      "learning_rate": 2.1093506072584724e-05,
      "loss": 1.4679,
      "step": 31830
    },
    {
      "epoch": 7.896335792671586,
      "grad_norm": 1.0690093040466309,
      "learning_rate": 2.1045665169976468e-05,
      "loss": 1.4696,
      "step": 31840
    },
    {
      "epoch": 7.898815797631595,
      "grad_norm": 1.1721092462539673,
      "learning_rate": 2.0997872200373116e-05,
      "loss": 1.4174,
      "step": 31850
    },
    {
      "epoch": 7.901295802591605,
      "grad_norm": 0.9855526685714722,
      "learning_rate": 2.095012719278966e-05,
      "loss": 1.4592,
      "step": 31860
    },
    {
      "epoch": 7.903775807551615,
      "grad_norm": 1.0813654661178589,
      "learning_rate": 2.0902430176211995e-05,
      "loss": 1.4446,
      "step": 31870
    },
    {
      "epoch": 7.906255812511625,
      "grad_norm": 1.001652479171753,
      "learning_rate": 2.0854781179596938e-05,
      "loss": 1.3965,
      "step": 31880
    },
    {
      "epoch": 7.908735817471635,
      "grad_norm": 1.0926847457885742,
      "learning_rate": 2.080718023187198e-05,
      "loss": 1.3916,
      "step": 31890
    },
    {
      "epoch": 7.911215822431645,
      "grad_norm": 0.972129762172699,
      "learning_rate": 2.075962736193564e-05,
      "loss": 1.4492,
      "step": 31900
    },
    {
      "epoch": 7.9136958273916544,
      "grad_norm": 1.0363141298294067,
      "learning_rate": 2.0712122598657135e-05,
      "loss": 1.5137,
      "step": 31910
    },
    {
      "epoch": 7.916175832351665,
      "grad_norm": 1.038306713104248,
      "learning_rate": 2.0664665970876496e-05,
      "loss": 1.4088,
      "step": 31920
    },
    {
      "epoch": 7.918655837311674,
      "grad_norm": 1.120137333869934,
      "learning_rate": 2.0617257507404532e-05,
      "loss": 1.4371,
      "step": 31930
    },
    {
      "epoch": 7.921135842271685,
      "grad_norm": 1.0201654434204102,
      "learning_rate": 2.056989723702287e-05,
      "loss": 1.4596,
      "step": 31940
    },
    {
      "epoch": 7.923615847231694,
      "grad_norm": 1.0573019981384277,
      "learning_rate": 2.0522585188483745e-05,
      "loss": 1.4672,
      "step": 31950
    },
    {
      "epoch": 7.926095852191704,
      "grad_norm": 1.104080080986023,
      "learning_rate": 2.047532139051026e-05,
      "loss": 1.4372,
      "step": 31960
    },
    {
      "epoch": 7.928575857151714,
      "grad_norm": 0.9933714866638184,
      "learning_rate": 2.0428105871796154e-05,
      "loss": 1.4312,
      "step": 31970
    },
    {
      "epoch": 7.931055862111724,
      "grad_norm": 1.007702350616455,
      "learning_rate": 2.0380938661005854e-05,
      "loss": 1.4008,
      "step": 31980
    },
    {
      "epoch": 7.933535867071734,
      "grad_norm": 1.020708680152893,
      "learning_rate": 2.0333819786774444e-05,
      "loss": 1.4521,
      "step": 31990
    },
    {
      "epoch": 7.936015872031744,
      "grad_norm": 1.0443121194839478,
      "learning_rate": 2.0286749277707782e-05,
      "loss": 1.4368,
      "step": 32000
    },
    {
      "epoch": 7.9384958769917535,
      "grad_norm": 1.0342239141464233,
      "learning_rate": 2.0239727162382148e-05,
      "loss": 1.4525,
      "step": 32010
    },
    {
      "epoch": 7.940975881951764,
      "grad_norm": 1.0420253276824951,
      "learning_rate": 2.019275346934466e-05,
      "loss": 1.4377,
      "step": 32020
    },
    {
      "epoch": 7.9434558869117735,
      "grad_norm": 0.9696030616760254,
      "learning_rate": 2.0145828227112917e-05,
      "loss": 1.4235,
      "step": 32030
    },
    {
      "epoch": 7.945935891871784,
      "grad_norm": 1.0450141429901123,
      "learning_rate": 2.009895146417512e-05,
      "loss": 1.4557,
      "step": 32040
    },
    {
      "epoch": 7.9484158968317935,
      "grad_norm": 1.0528062582015991,
      "learning_rate": 2.0052123208990025e-05,
      "loss": 1.4333,
      "step": 32050
    },
    {
      "epoch": 7.950895901791804,
      "grad_norm": 1.0280351638793945,
      "learning_rate": 2.000534348998704e-05,
      "loss": 1.4952,
      "step": 32060
    },
    {
      "epoch": 7.9533759067518135,
      "grad_norm": 1.0444345474243164,
      "learning_rate": 1.9958612335565975e-05,
      "loss": 1.424,
      "step": 32070
    },
    {
      "epoch": 7.955855911711823,
      "grad_norm": 1.0241684913635254,
      "learning_rate": 1.9911929774097215e-05,
      "loss": 1.4075,
      "step": 32080
    },
    {
      "epoch": 7.9583359166718335,
      "grad_norm": 1.0345664024353027,
      "learning_rate": 1.986529583392167e-05,
      "loss": 1.4826,
      "step": 32090
    },
    {
      "epoch": 7.960815921631843,
      "grad_norm": 0.9985606074333191,
      "learning_rate": 1.981871054335067e-05,
      "loss": 1.4267,
      "step": 32100
    },
    {
      "epoch": 7.9632959265918535,
      "grad_norm": 1.0109633207321167,
      "learning_rate": 1.977217393066604e-05,
      "loss": 1.3878,
      "step": 32110
    },
    {
      "epoch": 7.965775931551863,
      "grad_norm": 1.0159655809402466,
      "learning_rate": 1.9725686024120095e-05,
      "loss": 1.4437,
      "step": 32120
    },
    {
      "epoch": 7.9682559365118735,
      "grad_norm": 1.05583918094635,
      "learning_rate": 1.967924685193552e-05,
      "loss": 1.3912,
      "step": 32130
    },
    {
      "epoch": 7.970735941471883,
      "grad_norm": 1.030286431312561,
      "learning_rate": 1.9632856442305415e-05,
      "loss": 1.4066,
      "step": 32140
    },
    {
      "epoch": 7.973215946431893,
      "grad_norm": 1.0742601156234741,
      "learning_rate": 1.9586514823393363e-05,
      "loss": 1.4134,
      "step": 32150
    },
    {
      "epoch": 7.975695951391903,
      "grad_norm": 0.9984321594238281,
      "learning_rate": 1.9540222023333166e-05,
      "loss": 1.5042,
      "step": 32160
    },
    {
      "epoch": 7.978175956351913,
      "grad_norm": 1.0197498798370361,
      "learning_rate": 1.949397807022916e-05,
      "loss": 1.4358,
      "step": 32170
    },
    {
      "epoch": 7.980655961311923,
      "grad_norm": 1.0034178495407104,
      "learning_rate": 1.9447782992155916e-05,
      "loss": 1.4705,
      "step": 32180
    },
    {
      "epoch": 7.983135966271933,
      "grad_norm": 1.04500150680542,
      "learning_rate": 1.9401636817158365e-05,
      "loss": 1.3831,
      "step": 32190
    },
    {
      "epoch": 7.985615971231942,
      "grad_norm": 1.114035725593567,
      "learning_rate": 1.9355539573251734e-05,
      "loss": 1.4582,
      "step": 32200
    },
    {
      "epoch": 7.9880959761919526,
      "grad_norm": 1.0428582429885864,
      "learning_rate": 1.93094912884216e-05,
      "loss": 1.42,
      "step": 32210
    },
    {
      "epoch": 7.990575981151962,
      "grad_norm": 1.0457404851913452,
      "learning_rate": 1.926349199062376e-05,
      "loss": 1.468,
      "step": 32220
    },
    {
      "epoch": 7.9930559861119725,
      "grad_norm": 1.023844599723816,
      "learning_rate": 1.921754170778428e-05,
      "loss": 1.4377,
      "step": 32230
    },
    {
      "epoch": 7.995535991071982,
      "grad_norm": 1.0685179233551025,
      "learning_rate": 1.917164046779948e-05,
      "loss": 1.4771,
      "step": 32240
    },
    {
      "epoch": 7.998015996031992,
      "grad_norm": 1.034908652305603,
      "learning_rate": 1.9125788298535908e-05,
      "loss": 1.4559,
      "step": 32250
    },
    {
      "epoch": 8.000496000992001,
      "grad_norm": 1.0679064989089966,
      "learning_rate": 1.9079985227830276e-05,
      "loss": 1.4761,
      "step": 32260
    },
    {
      "epoch": 8.002976005952013,
      "grad_norm": 1.0527770519256592,
      "learning_rate": 1.903423128348959e-05,
      "loss": 1.3527,
      "step": 32270
    },
    {
      "epoch": 8.005456010912022,
      "grad_norm": 0.9856534600257874,
      "learning_rate": 1.898852649329095e-05,
      "loss": 1.411,
      "step": 32280
    },
    {
      "epoch": 8.007936015872032,
      "grad_norm": 1.0472939014434814,
      "learning_rate": 1.894287088498162e-05,
      "loss": 1.3859,
      "step": 32290
    },
    {
      "epoch": 8.010416020832041,
      "grad_norm": 1.021368145942688,
      "learning_rate": 1.889726448627904e-05,
      "loss": 1.4096,
      "step": 32300
    },
    {
      "epoch": 8.01289602579205,
      "grad_norm": 1.0529119968414307,
      "learning_rate": 1.885170732487074e-05,
      "loss": 1.3803,
      "step": 32310
    },
    {
      "epoch": 8.015376030752062,
      "grad_norm": 1.087838053703308,
      "learning_rate": 1.880619942841435e-05,
      "loss": 1.4731,
      "step": 32320
    },
    {
      "epoch": 8.017856035712072,
      "grad_norm": 0.9879887700080872,
      "learning_rate": 1.876074082453766e-05,
      "loss": 1.3251,
      "step": 32330
    },
    {
      "epoch": 8.020336040672081,
      "grad_norm": 1.0613797903060913,
      "learning_rate": 1.8715331540838487e-05,
      "loss": 1.4409,
      "step": 32340
    },
    {
      "epoch": 8.02281604563209,
      "grad_norm": 1.0037634372711182,
      "learning_rate": 1.8669971604884683e-05,
      "loss": 1.4334,
      "step": 32350
    },
    {
      "epoch": 8.025296050592102,
      "grad_norm": 1.1097238063812256,
      "learning_rate": 1.8624661044214152e-05,
      "loss": 1.3923,
      "step": 32360
    },
    {
      "epoch": 8.027776055552112,
      "grad_norm": 1.0040903091430664,
      "learning_rate": 1.8579399886334914e-05,
      "loss": 1.3527,
      "step": 32370
    },
    {
      "epoch": 8.030256060512121,
      "grad_norm": 0.9767634272575378,
      "learning_rate": 1.853418815872482e-05,
      "loss": 1.3936,
      "step": 32380
    },
    {
      "epoch": 8.03273606547213,
      "grad_norm": 0.9867961406707764,
      "learning_rate": 1.848902588883188e-05,
      "loss": 1.4003,
      "step": 32390
    },
    {
      "epoch": 8.03521607043214,
      "grad_norm": 1.016300082206726,
      "learning_rate": 1.8443913104073983e-05,
      "loss": 1.3895,
      "step": 32400
    },
    {
      "epoch": 8.037696075392152,
      "grad_norm": 1.042302131652832,
      "learning_rate": 1.8398849831839014e-05,
      "loss": 1.3428,
      "step": 32410
    },
    {
      "epoch": 8.040176080352161,
      "grad_norm": 0.99931401014328,
      "learning_rate": 1.8353836099484767e-05,
      "loss": 1.407,
      "step": 32420
    },
    {
      "epoch": 8.04265608531217,
      "grad_norm": 1.0488470792770386,
      "learning_rate": 1.8308871934339033e-05,
      "loss": 1.4504,
      "step": 32430
    },
    {
      "epoch": 8.04513609027218,
      "grad_norm": 1.020246982574463,
      "learning_rate": 1.826395736369937e-05,
      "loss": 1.4135,
      "step": 32440
    },
    {
      "epoch": 8.04761609523219,
      "grad_norm": 1.080391526222229,
      "learning_rate": 1.8219092414833394e-05,
      "loss": 1.4268,
      "step": 32450
    },
    {
      "epoch": 8.050096100192201,
      "grad_norm": 0.9757678508758545,
      "learning_rate": 1.81742771149785e-05,
      "loss": 1.433,
      "step": 32460
    },
    {
      "epoch": 8.05257610515221,
      "grad_norm": 1.076442003250122,
      "learning_rate": 1.8129511491341946e-05,
      "loss": 1.4138,
      "step": 32470
    },
    {
      "epoch": 8.05505611011222,
      "grad_norm": 1.091922402381897,
      "learning_rate": 1.808479557110081e-05,
      "loss": 1.4215,
      "step": 32480
    },
    {
      "epoch": 8.05753611507223,
      "grad_norm": 1.0268551111221313,
      "learning_rate": 1.8040129381402137e-05,
      "loss": 1.4107,
      "step": 32490
    },
    {
      "epoch": 8.06001612003224,
      "grad_norm": 1.0977098941802979,
      "learning_rate": 1.7995512949362547e-05,
      "loss": 1.4162,
      "step": 32500
    },
    {
      "epoch": 8.06249612499225,
      "grad_norm": 1.0292160511016846,
      "learning_rate": 1.795094630206866e-05,
      "loss": 1.3961,
      "step": 32510
    },
    {
      "epoch": 8.06497612995226,
      "grad_norm": 1.0747003555297852,
      "learning_rate": 1.7906429466576767e-05,
      "loss": 1.3931,
      "step": 32520
    },
    {
      "epoch": 8.06745613491227,
      "grad_norm": 0.9828006029129028,
      "learning_rate": 1.7861962469912953e-05,
      "loss": 1.3724,
      "step": 32530
    },
    {
      "epoch": 8.06993613987228,
      "grad_norm": 1.0837029218673706,
      "learning_rate": 1.7817545339072994e-05,
      "loss": 1.3638,
      "step": 32540
    },
    {
      "epoch": 8.072416144832289,
      "grad_norm": 1.0823603868484497,
      "learning_rate": 1.7773178101022514e-05,
      "loss": 1.44,
      "step": 32550
    },
    {
      "epoch": 8.0748961497923,
      "grad_norm": 1.0632768869400024,
      "learning_rate": 1.7728860782696664e-05,
      "loss": 1.4125,
      "step": 32560
    },
    {
      "epoch": 8.07737615475231,
      "grad_norm": 1.0829638242721558,
      "learning_rate": 1.7684593411000472e-05,
      "loss": 1.4523,
      "step": 32570
    },
    {
      "epoch": 8.07985615971232,
      "grad_norm": 1.0273957252502441,
      "learning_rate": 1.7640376012808536e-05,
      "loss": 1.4002,
      "step": 32580
    },
    {
      "epoch": 8.082336164672329,
      "grad_norm": 0.9694962501525879,
      "learning_rate": 1.759620861496515e-05,
      "loss": 1.44,
      "step": 32590
    },
    {
      "epoch": 8.084816169632338,
      "grad_norm": 0.9955787062644958,
      "learning_rate": 1.7552091244284197e-05,
      "loss": 1.3595,
      "step": 32600
    },
    {
      "epoch": 8.08729617459235,
      "grad_norm": 1.0302938222885132,
      "learning_rate": 1.7508023927549333e-05,
      "loss": 1.4095,
      "step": 32610
    },
    {
      "epoch": 8.08977617955236,
      "grad_norm": 1.0442672967910767,
      "learning_rate": 1.7464006691513623e-05,
      "loss": 1.4573,
      "step": 32620
    },
    {
      "epoch": 8.092256184512369,
      "grad_norm": 1.0897337198257446,
      "learning_rate": 1.7420039562899915e-05,
      "loss": 1.4222,
      "step": 32630
    },
    {
      "epoch": 8.094736189472378,
      "grad_norm": 1.081861138343811,
      "learning_rate": 1.7376122568400532e-05,
      "loss": 1.4007,
      "step": 32640
    },
    {
      "epoch": 8.09721619443239,
      "grad_norm": 0.9963325262069702,
      "learning_rate": 1.733225573467737e-05,
      "loss": 1.3782,
      "step": 32650
    },
    {
      "epoch": 8.0996961993924,
      "grad_norm": 1.0090794563293457,
      "learning_rate": 1.7288439088361886e-05,
      "loss": 1.4332,
      "step": 32660
    },
    {
      "epoch": 8.102176204352409,
      "grad_norm": 0.9832302927970886,
      "learning_rate": 1.7244672656055106e-05,
      "loss": 1.4046,
      "step": 32670
    },
    {
      "epoch": 8.104656209312418,
      "grad_norm": 1.0199371576309204,
      "learning_rate": 1.720095646432751e-05,
      "loss": 1.3729,
      "step": 32680
    },
    {
      "epoch": 8.107136214272428,
      "grad_norm": 1.0891382694244385,
      "learning_rate": 1.7157290539719106e-05,
      "loss": 1.4103,
      "step": 32690
    },
    {
      "epoch": 8.10961621923244,
      "grad_norm": 1.03241765499115,
      "learning_rate": 1.7113674908739396e-05,
      "loss": 1.4102,
      "step": 32700
    },
    {
      "epoch": 8.112096224192449,
      "grad_norm": 0.988692581653595,
      "learning_rate": 1.7070109597867312e-05,
      "loss": 1.4142,
      "step": 32710
    },
    {
      "epoch": 8.114576229152458,
      "grad_norm": 0.9958691596984863,
      "learning_rate": 1.702659463355125e-05,
      "loss": 1.374,
      "step": 32720
    },
    {
      "epoch": 8.117056234112468,
      "grad_norm": 1.058645248413086,
      "learning_rate": 1.6983130042209094e-05,
      "loss": 1.3947,
      "step": 32730
    },
    {
      "epoch": 8.119536239072477,
      "grad_norm": 1.0806529521942139,
      "learning_rate": 1.693971585022808e-05,
      "loss": 1.3713,
      "step": 32740
    },
    {
      "epoch": 8.122016244032489,
      "grad_norm": 1.0624850988388062,
      "learning_rate": 1.689635208396486e-05,
      "loss": 1.4297,
      "step": 32750
    },
    {
      "epoch": 8.124496248992498,
      "grad_norm": 1.0744974613189697,
      "learning_rate": 1.6853038769745467e-05,
      "loss": 1.3951,
      "step": 32760
    },
    {
      "epoch": 8.126976253952508,
      "grad_norm": 1.0920403003692627,
      "learning_rate": 1.680977593386538e-05,
      "loss": 1.3603,
      "step": 32770
    },
    {
      "epoch": 8.129456258912517,
      "grad_norm": 1.0960181951522827,
      "learning_rate": 1.6766563602589324e-05,
      "loss": 1.4094,
      "step": 32780
    },
    {
      "epoch": 8.131936263872527,
      "grad_norm": 1.077284336090088,
      "learning_rate": 1.672340180215143e-05,
      "loss": 1.3769,
      "step": 32790
    },
    {
      "epoch": 8.134416268832538,
      "grad_norm": 1.00383460521698,
      "learning_rate": 1.668029055875512e-05,
      "loss": 1.3663,
      "step": 32800
    },
    {
      "epoch": 8.136896273792548,
      "grad_norm": 1.0465641021728516,
      "learning_rate": 1.6637229898573113e-05,
      "loss": 1.3224,
      "step": 32810
    },
    {
      "epoch": 8.139376278752557,
      "grad_norm": 1.0937128067016602,
      "learning_rate": 1.6594219847747504e-05,
      "loss": 1.3983,
      "step": 32820
    },
    {
      "epoch": 8.141856283712567,
      "grad_norm": 1.0818277597427368,
      "learning_rate": 1.655126043238957e-05,
      "loss": 1.4798,
      "step": 32830
    },
    {
      "epoch": 8.144336288672577,
      "grad_norm": 1.0414772033691406,
      "learning_rate": 1.6508351678579882e-05,
      "loss": 1.3921,
      "step": 32840
    },
    {
      "epoch": 8.146816293632588,
      "grad_norm": 1.0060216188430786,
      "learning_rate": 1.6465493612368233e-05,
      "loss": 1.4231,
      "step": 32850
    },
    {
      "epoch": 8.149296298592597,
      "grad_norm": 1.0131288766860962,
      "learning_rate": 1.6422686259773678e-05,
      "loss": 1.3954,
      "step": 32860
    },
    {
      "epoch": 8.151776303552607,
      "grad_norm": 1.0088540315628052,
      "learning_rate": 1.6379929646784443e-05,
      "loss": 1.4203,
      "step": 32870
    },
    {
      "epoch": 8.154256308512617,
      "grad_norm": 1.0268006324768066,
      "learning_rate": 1.6337223799358026e-05,
      "loss": 1.4408,
      "step": 32880
    },
    {
      "epoch": 8.156736313472628,
      "grad_norm": 1.0829941034317017,
      "learning_rate": 1.6294568743421025e-05,
      "loss": 1.4245,
      "step": 32890
    },
    {
      "epoch": 8.159216318432637,
      "grad_norm": 1.097237229347229,
      "learning_rate": 1.6251964504869222e-05,
      "loss": 1.4472,
      "step": 32900
    },
    {
      "epoch": 8.161696323392647,
      "grad_norm": 1.127237319946289,
      "learning_rate": 1.620941110956754e-05,
      "loss": 1.4938,
      "step": 32910
    },
    {
      "epoch": 8.164176328352656,
      "grad_norm": 1.1016814708709717,
      "learning_rate": 1.616690858335014e-05,
      "loss": 1.3906,
      "step": 32920
    },
    {
      "epoch": 8.166656333312666,
      "grad_norm": 1.079646348953247,
      "learning_rate": 1.6124456952020093e-05,
      "loss": 1.4228,
      "step": 32930
    },
    {
      "epoch": 8.169136338272677,
      "grad_norm": 1.043107032775879,
      "learning_rate": 1.6082056241349786e-05,
      "loss": 1.34,
      "step": 32940
    },
    {
      "epoch": 8.171616343232687,
      "grad_norm": 1.044143795967102,
      "learning_rate": 1.6039706477080562e-05,
      "loss": 1.3492,
      "step": 32950
    },
    {
      "epoch": 8.174096348192696,
      "grad_norm": 1.0955991744995117,
      "learning_rate": 1.5997407684922862e-05,
      "loss": 1.4136,
      "step": 32960
    },
    {
      "epoch": 8.176576353152706,
      "grad_norm": 0.9900245666503906,
      "learning_rate": 1.595515989055618e-05,
      "loss": 1.3789,
      "step": 32970
    },
    {
      "epoch": 8.179056358112716,
      "grad_norm": 0.9846022725105286,
      "learning_rate": 1.5912963119629144e-05,
      "loss": 1.3932,
      "step": 32980
    },
    {
      "epoch": 8.181536363072727,
      "grad_norm": 1.0475287437438965,
      "learning_rate": 1.5870817397759198e-05,
      "loss": 1.3597,
      "step": 32990
    },
    {
      "epoch": 8.184016368032736,
      "grad_norm": 1.1756962537765503,
      "learning_rate": 1.5828722750533008e-05,
      "loss": 1.4158,
      "step": 33000
    },
    {
      "epoch": 8.186496372992746,
      "grad_norm": 1.143308401107788,
      "learning_rate": 1.5786679203506105e-05,
      "loss": 1.3885,
      "step": 33010
    },
    {
      "epoch": 8.188976377952756,
      "grad_norm": 1.0628608465194702,
      "learning_rate": 1.5744686782203055e-05,
      "loss": 1.4024,
      "step": 33020
    },
    {
      "epoch": 8.191456382912765,
      "grad_norm": 1.023871660232544,
      "learning_rate": 1.5702745512117324e-05,
      "loss": 1.4226,
      "step": 33030
    },
    {
      "epoch": 8.193936387872776,
      "grad_norm": 1.109716534614563,
      "learning_rate": 1.566085541871145e-05,
      "loss": 1.4659,
      "step": 33040
    },
    {
      "epoch": 8.196416392832786,
      "grad_norm": 1.1114068031311035,
      "learning_rate": 1.5619016527416707e-05,
      "loss": 1.3765,
      "step": 33050
    },
    {
      "epoch": 8.198896397792796,
      "grad_norm": 1.0773273706436157,
      "learning_rate": 1.5577228863633486e-05,
      "loss": 1.3958,
      "step": 33060
    },
    {
      "epoch": 8.201376402752805,
      "grad_norm": 0.9811004996299744,
      "learning_rate": 1.5535492452730947e-05,
      "loss": 1.4084,
      "step": 33070
    },
    {
      "epoch": 8.203856407712815,
      "grad_norm": 1.0447224378585815,
      "learning_rate": 1.549380732004718e-05,
      "loss": 1.4256,
      "step": 33080
    },
    {
      "epoch": 8.206336412672826,
      "grad_norm": 1.0115346908569336,
      "learning_rate": 1.545217349088911e-05,
      "loss": 1.4162,
      "step": 33090
    },
    {
      "epoch": 8.208816417632836,
      "grad_norm": 1.0720356702804565,
      "learning_rate": 1.5410590990532625e-05,
      "loss": 1.484,
      "step": 33100
    },
    {
      "epoch": 8.211296422592845,
      "grad_norm": 1.0902684926986694,
      "learning_rate": 1.5369059844222278e-05,
      "loss": 1.3935,
      "step": 33110
    },
    {
      "epoch": 8.213776427552855,
      "grad_norm": 1.0302410125732422,
      "learning_rate": 1.5327580077171587e-05,
      "loss": 1.3723,
      "step": 33120
    },
    {
      "epoch": 8.216256432512864,
      "grad_norm": 1.01740562915802,
      "learning_rate": 1.5286151714562824e-05,
      "loss": 1.472,
      "step": 33130
    },
    {
      "epoch": 8.218736437472876,
      "grad_norm": 1.0869848728179932,
      "learning_rate": 1.5244774781547055e-05,
      "loss": 1.4054,
      "step": 33140
    },
    {
      "epoch": 8.221216442432885,
      "grad_norm": 1.1105198860168457,
      "learning_rate": 1.5203449303244099e-05,
      "loss": 1.4238,
      "step": 33150
    },
    {
      "epoch": 8.223696447392895,
      "grad_norm": 1.002714991569519,
      "learning_rate": 1.5162175304742632e-05,
      "loss": 1.4326,
      "step": 33160
    },
    {
      "epoch": 8.226176452352904,
      "grad_norm": 1.0198575258255005,
      "learning_rate": 1.512095281109992e-05,
      "loss": 1.4443,
      "step": 33170
    },
    {
      "epoch": 8.228656457312916,
      "grad_norm": 0.9816635251045227,
      "learning_rate": 1.5079781847342123e-05,
      "loss": 1.4123,
      "step": 33180
    },
    {
      "epoch": 8.231136462272925,
      "grad_norm": 1.0508211851119995,
      "learning_rate": 1.5038662438464001e-05,
      "loss": 1.4052,
      "step": 33190
    },
    {
      "epoch": 8.233616467232935,
      "grad_norm": 1.0052751302719116,
      "learning_rate": 1.4997594609429088e-05,
      "loss": 1.3232,
      "step": 33200
    },
    {
      "epoch": 8.236096472192944,
      "grad_norm": 1.048970341682434,
      "learning_rate": 1.4956578385169528e-05,
      "loss": 1.4079,
      "step": 33210
    },
    {
      "epoch": 8.238576477152954,
      "grad_norm": 1.1010568141937256,
      "learning_rate": 1.4915613790586225e-05,
      "loss": 1.3763,
      "step": 33220
    },
    {
      "epoch": 8.241056482112965,
      "grad_norm": 1.0694609880447388,
      "learning_rate": 1.4874700850548695e-05,
      "loss": 1.3859,
      "step": 33230
    },
    {
      "epoch": 8.243536487072975,
      "grad_norm": 1.1031140089035034,
      "learning_rate": 1.4833839589895071e-05,
      "loss": 1.358,
      "step": 33240
    },
    {
      "epoch": 8.246016492032984,
      "grad_norm": 1.1470513343811035,
      "learning_rate": 1.4793030033432142e-05,
      "loss": 1.4303,
      "step": 33250
    },
    {
      "epoch": 8.248496496992994,
      "grad_norm": 1.1016775369644165,
      "learning_rate": 1.4752272205935313e-05,
      "loss": 1.491,
      "step": 33260
    },
    {
      "epoch": 8.250976501953003,
      "grad_norm": 1.0250742435455322,
      "learning_rate": 1.4711566132148536e-05,
      "loss": 1.3813,
      "step": 33270
    },
    {
      "epoch": 8.253456506913015,
      "grad_norm": 1.0153090953826904,
      "learning_rate": 1.467091183678444e-05,
      "loss": 1.4163,
      "step": 33280
    },
    {
      "epoch": 8.255936511873024,
      "grad_norm": 1.034895420074463,
      "learning_rate": 1.4630309344524118e-05,
      "loss": 1.4455,
      "step": 33290
    },
    {
      "epoch": 8.258416516833034,
      "grad_norm": 0.9518606662750244,
      "learning_rate": 1.4589758680017263e-05,
      "loss": 1.4113,
      "step": 33300
    },
    {
      "epoch": 8.260896521793043,
      "grad_norm": 1.0583425760269165,
      "learning_rate": 1.4549259867882092e-05,
      "loss": 1.4058,
      "step": 33310
    },
    {
      "epoch": 8.263376526753053,
      "grad_norm": 1.0160610675811768,
      "learning_rate": 1.4508812932705363e-05,
      "loss": 1.3968,
      "step": 33320
    },
    {
      "epoch": 8.265856531713064,
      "grad_norm": 1.0411465167999268,
      "learning_rate": 1.4468417899042275e-05,
      "loss": 1.4378,
      "step": 33330
    },
    {
      "epoch": 8.268336536673074,
      "grad_norm": 1.0812145471572876,
      "learning_rate": 1.4428074791416624e-05,
      "loss": 1.3769,
      "step": 33340
    },
    {
      "epoch": 8.270816541633083,
      "grad_norm": 1.0007779598236084,
      "learning_rate": 1.4387783634320595e-05,
      "loss": 1.3606,
      "step": 33350
    },
    {
      "epoch": 8.273296546593093,
      "grad_norm": 1.0087769031524658,
      "learning_rate": 1.4347544452214868e-05,
      "loss": 1.3872,
      "step": 33360
    },
    {
      "epoch": 8.275776551553102,
      "grad_norm": 1.1092948913574219,
      "learning_rate": 1.4307357269528542e-05,
      "loss": 1.4013,
      "step": 33370
    },
    {
      "epoch": 8.278256556513114,
      "grad_norm": 1.0083212852478027,
      "learning_rate": 1.4267222110659218e-05,
      "loss": 1.4702,
      "step": 33380
    },
    {
      "epoch": 8.280736561473123,
      "grad_norm": 1.1285642385482788,
      "learning_rate": 1.42271389999728e-05,
      "loss": 1.4447,
      "step": 33390
    },
    {
      "epoch": 8.283216566433133,
      "grad_norm": 1.034905195236206,
      "learning_rate": 1.4187107961803703e-05,
      "loss": 1.4137,
      "step": 33400
    },
    {
      "epoch": 8.285696571393142,
      "grad_norm": 1.0655313730239868,
      "learning_rate": 1.4147129020454663e-05,
      "loss": 1.396,
      "step": 33410
    },
    {
      "epoch": 8.288176576353152,
      "grad_norm": 1.0830758810043335,
      "learning_rate": 1.4107202200196801e-05,
      "loss": 1.4299,
      "step": 33420
    },
    {
      "epoch": 8.290656581313163,
      "grad_norm": 1.1033685207366943,
      "learning_rate": 1.4067327525269636e-05,
      "loss": 1.3837,
      "step": 33430
    },
    {
      "epoch": 8.293136586273173,
      "grad_norm": 1.054334282875061,
      "learning_rate": 1.402750501988097e-05,
      "loss": 1.4024,
      "step": 33440
    },
    {
      "epoch": 8.295616591233182,
      "grad_norm": 1.046523094177246,
      "learning_rate": 1.398773470820698e-05,
      "loss": 1.3954,
      "step": 33450
    },
    {
      "epoch": 8.298096596193192,
      "grad_norm": 1.0344880819320679,
      "learning_rate": 1.3948016614392112e-05,
      "loss": 1.4287,
      "step": 33460
    },
    {
      "epoch": 8.300576601153203,
      "grad_norm": 1.0231508016586304,
      "learning_rate": 1.3908350762549138e-05,
      "loss": 1.381,
      "step": 33470
    },
    {
      "epoch": 8.303056606113213,
      "grad_norm": 1.1426787376403809,
      "learning_rate": 1.3868737176759106e-05,
      "loss": 1.3853,
      "step": 33480
    },
    {
      "epoch": 8.305536611073222,
      "grad_norm": 1.0430539846420288,
      "learning_rate": 1.3829175881071354e-05,
      "loss": 1.4398,
      "step": 33490
    },
    {
      "epoch": 8.308016616033232,
      "grad_norm": 1.0005583763122559,
      "learning_rate": 1.3789666899503462e-05,
      "loss": 1.4349,
      "step": 33500
    },
    {
      "epoch": 8.310496620993241,
      "grad_norm": 1.0415821075439453,
      "learning_rate": 1.3750210256041241e-05,
      "loss": 1.3943,
      "step": 33510
    },
    {
      "epoch": 8.312976625953253,
      "grad_norm": 1.0108755826950073,
      "learning_rate": 1.3710805974638696e-05,
      "loss": 1.3861,
      "step": 33520
    },
    {
      "epoch": 8.315456630913262,
      "grad_norm": 1.02773916721344,
      "learning_rate": 1.367145407921817e-05,
      "loss": 1.4026,
      "step": 33530
    },
    {
      "epoch": 8.317936635873272,
      "grad_norm": 1.1380155086517334,
      "learning_rate": 1.363215459367001e-05,
      "loss": 1.3809,
      "step": 33540
    },
    {
      "epoch": 8.320416640833281,
      "grad_norm": 1.0224312543869019,
      "learning_rate": 1.3592907541852906e-05,
      "loss": 1.4431,
      "step": 33550
    },
    {
      "epoch": 8.322896645793291,
      "grad_norm": 1.0225660800933838,
      "learning_rate": 1.3553712947593656e-05,
      "loss": 1.4379,
      "step": 33560
    },
    {
      "epoch": 8.325376650753302,
      "grad_norm": 1.1233614683151245,
      "learning_rate": 1.3514570834687202e-05,
      "loss": 1.4483,
      "step": 33570
    },
    {
      "epoch": 8.327856655713312,
      "grad_norm": 1.0314141511917114,
      "learning_rate": 1.3475481226896624e-05,
      "loss": 1.3876,
      "step": 33580
    },
    {
      "epoch": 8.330336660673321,
      "grad_norm": 1.0322446823120117,
      "learning_rate": 1.3436444147953187e-05,
      "loss": 1.3916,
      "step": 33590
    },
    {
      "epoch": 8.332816665633331,
      "grad_norm": 1.1228183507919312,
      "learning_rate": 1.339745962155613e-05,
      "loss": 1.4225,
      "step": 33600
    },
    {
      "epoch": 8.33529667059334,
      "grad_norm": 0.96470046043396,
      "learning_rate": 1.3358527671372956e-05,
      "loss": 1.3723,
      "step": 33610
    },
    {
      "epoch": 8.337776675553352,
      "grad_norm": 1.0331604480743408,
      "learning_rate": 1.3319648321039136e-05,
      "loss": 1.3941,
      "step": 33620
    },
    {
      "epoch": 8.340256680513361,
      "grad_norm": 1.0940603017807007,
      "learning_rate": 1.3280821594158243e-05,
      "loss": 1.4059,
      "step": 33630
    },
    {
      "epoch": 8.34273668547337,
      "grad_norm": 0.9891999959945679,
      "learning_rate": 1.3242047514301858e-05,
      "loss": 1.3959,
      "step": 33640
    },
    {
      "epoch": 8.34521669043338,
      "grad_norm": 1.0726075172424316,
      "learning_rate": 1.320332610500974e-05,
      "loss": 1.4053,
      "step": 33650
    },
    {
      "epoch": 8.34769669539339,
      "grad_norm": 1.1014840602874756,
      "learning_rate": 1.3164657389789458e-05,
      "loss": 1.3337,
      "step": 33660
    },
    {
      "epoch": 8.350176700353401,
      "grad_norm": 1.047423005104065,
      "learning_rate": 1.3126041392116772e-05,
      "loss": 1.3769,
      "step": 33670
    },
    {
      "epoch": 8.35265670531341,
      "grad_norm": 1.0521049499511719,
      "learning_rate": 1.308747813543536e-05,
      "loss": 1.4227,
      "step": 33680
    },
    {
      "epoch": 8.35513671027342,
      "grad_norm": 0.9757634997367859,
      "learning_rate": 1.3048967643156884e-05,
      "loss": 1.3741,
      "step": 33690
    },
    {
      "epoch": 8.35761671523343,
      "grad_norm": 1.0512996912002563,
      "learning_rate": 1.3010509938660965e-05,
      "loss": 1.413,
      "step": 33700
    },
    {
      "epoch": 8.36009672019344,
      "grad_norm": 1.1035054922103882,
      "learning_rate": 1.2972105045295247e-05,
      "loss": 1.4607,
      "step": 33710
    },
    {
      "epoch": 8.36257672515345,
      "grad_norm": 1.0338871479034424,
      "learning_rate": 1.293375298637518e-05,
      "loss": 1.4175,
      "step": 33720
    },
    {
      "epoch": 8.36505673011346,
      "grad_norm": 1.0466989278793335,
      "learning_rate": 1.2895453785184275e-05,
      "loss": 1.3691,
      "step": 33730
    },
    {
      "epoch": 8.36753673507347,
      "grad_norm": 1.0674350261688232,
      "learning_rate": 1.2857207464973875e-05,
      "loss": 1.4637,
      "step": 33740
    },
    {
      "epoch": 8.37001674003348,
      "grad_norm": 1.09450101852417,
      "learning_rate": 1.281901404896323e-05,
      "loss": 1.3582,
      "step": 33750
    },
    {
      "epoch": 8.37249674499349,
      "grad_norm": 1.075958251953125,
      "learning_rate": 1.2780873560339468e-05,
      "loss": 1.4301,
      "step": 33760
    },
    {
      "epoch": 8.3749767499535,
      "grad_norm": 1.1581681966781616,
      "learning_rate": 1.2742786022257659e-05,
      "loss": 1.4585,
      "step": 33770
    },
    {
      "epoch": 8.37745675491351,
      "grad_norm": 1.0443540811538696,
      "learning_rate": 1.270475145784057e-05,
      "loss": 1.411,
      "step": 33780
    },
    {
      "epoch": 8.37993675987352,
      "grad_norm": 1.0078887939453125,
      "learning_rate": 1.2666769890178975e-05,
      "loss": 1.3823,
      "step": 33790
    },
    {
      "epoch": 8.382416764833529,
      "grad_norm": 1.0253844261169434,
      "learning_rate": 1.2628841342331388e-05,
      "loss": 1.4208,
      "step": 33800
    },
    {
      "epoch": 8.38489676979354,
      "grad_norm": 0.9839961528778076,
      "learning_rate": 1.2590965837324131e-05,
      "loss": 1.4099,
      "step": 33810
    },
    {
      "epoch": 8.38737677475355,
      "grad_norm": 1.0767897367477417,
      "learning_rate": 1.2553143398151324e-05,
      "loss": 1.4056,
      "step": 33820
    },
    {
      "epoch": 8.38985677971356,
      "grad_norm": 1.0594230890274048,
      "learning_rate": 1.2515374047774941e-05,
      "loss": 1.4008,
      "step": 33830
    },
    {
      "epoch": 8.392336784673569,
      "grad_norm": 1.0357154607772827,
      "learning_rate": 1.2477657809124631e-05,
      "loss": 1.3909,
      "step": 33840
    },
    {
      "epoch": 8.394816789633579,
      "grad_norm": 1.07870614528656,
      "learning_rate": 1.2439994705097869e-05,
      "loss": 1.439,
      "step": 33850
    },
    {
      "epoch": 8.39729679459359,
      "grad_norm": 1.042245626449585,
      "learning_rate": 1.2402384758559815e-05,
      "loss": 1.4808,
      "step": 33860
    },
    {
      "epoch": 8.3997767995536,
      "grad_norm": 1.1021331548690796,
      "learning_rate": 1.2364827992343397e-05,
      "loss": 1.401,
      "step": 33870
    },
    {
      "epoch": 8.402256804513609,
      "grad_norm": 1.0259779691696167,
      "learning_rate": 1.2327324429249232e-05,
      "loss": 1.4524,
      "step": 33880
    },
    {
      "epoch": 8.404736809473619,
      "grad_norm": 1.0595521926879883,
      "learning_rate": 1.2289874092045684e-05,
      "loss": 1.475,
      "step": 33890
    },
    {
      "epoch": 8.407216814433628,
      "grad_norm": 1.1025539636611938,
      "learning_rate": 1.2252477003468743e-05,
      "loss": 1.5016,
      "step": 33900
    },
    {
      "epoch": 8.40969681939364,
      "grad_norm": 1.0166747570037842,
      "learning_rate": 1.2215133186222128e-05,
      "loss": 1.3846,
      "step": 33910
    },
    {
      "epoch": 8.412176824353649,
      "grad_norm": 0.9710475206375122,
      "learning_rate": 1.2177842662977135e-05,
      "loss": 1.3827,
      "step": 33920
    },
    {
      "epoch": 8.414656829313659,
      "grad_norm": 1.0323448181152344,
      "learning_rate": 1.2140605456372855e-05,
      "loss": 1.4043,
      "step": 33930
    },
    {
      "epoch": 8.417136834273668,
      "grad_norm": 1.0237551927566528,
      "learning_rate": 1.2103421589015806e-05,
      "loss": 1.3894,
      "step": 33940
    },
    {
      "epoch": 8.419616839233678,
      "grad_norm": 1.030096173286438,
      "learning_rate": 1.2066291083480296e-05,
      "loss": 1.3108,
      "step": 33950
    },
    {
      "epoch": 8.422096844193689,
      "grad_norm": 1.0543568134307861,
      "learning_rate": 1.2029213962308172e-05,
      "loss": 1.385,
      "step": 33960
    },
    {
      "epoch": 8.424576849153699,
      "grad_norm": 0.9940221905708313,
      "learning_rate": 1.1992190248008861e-05,
      "loss": 1.3932,
      "step": 33970
    },
    {
      "epoch": 8.427056854113708,
      "grad_norm": 1.0462923049926758,
      "learning_rate": 1.195521996305936e-05,
      "loss": 1.3871,
      "step": 33980
    },
    {
      "epoch": 8.429536859073718,
      "grad_norm": 1.0675877332687378,
      "learning_rate": 1.1918303129904317e-05,
      "loss": 1.3765,
      "step": 33990
    },
    {
      "epoch": 8.432016864033727,
      "grad_norm": 1.0424073934555054,
      "learning_rate": 1.188143977095576e-05,
      "loss": 1.4018,
      "step": 34000
    },
    {
      "epoch": 8.434496868993739,
      "grad_norm": 1.0719586610794067,
      "learning_rate": 1.1844629908593441e-05,
      "loss": 1.406,
      "step": 34010
    },
    {
      "epoch": 8.436976873953748,
      "grad_norm": 1.080687165260315,
      "learning_rate": 1.1807873565164506e-05,
      "loss": 1.446,
      "step": 34020
    },
    {
      "epoch": 8.439456878913758,
      "grad_norm": 1.0027790069580078,
      "learning_rate": 1.1771170762983641e-05,
      "loss": 1.4801,
      "step": 34030
    },
    {
      "epoch": 8.441936883873767,
      "grad_norm": 1.3560655117034912,
      "learning_rate": 1.1734521524333086e-05,
      "loss": 1.4105,
      "step": 34040
    },
    {
      "epoch": 8.444416888833779,
      "grad_norm": 1.0444145202636719,
      "learning_rate": 1.1697925871462467e-05,
      "loss": 1.4516,
      "step": 34050
    },
    {
      "epoch": 8.446896893793788,
      "grad_norm": 1.1016205549240112,
      "learning_rate": 1.1661383826588957e-05,
      "loss": 1.4534,
      "step": 34060
    },
    {
      "epoch": 8.449376898753798,
      "grad_norm": 1.0291553735733032,
      "learning_rate": 1.1624895411897107e-05,
      "loss": 1.3971,
      "step": 34070
    },
    {
      "epoch": 8.451856903713807,
      "grad_norm": 1.014909029006958,
      "learning_rate": 1.1588460649539035e-05,
      "loss": 1.4349,
      "step": 34080
    },
    {
      "epoch": 8.454336908673817,
      "grad_norm": 1.1258878707885742,
      "learning_rate": 1.155207956163411e-05,
      "loss": 1.4256,
      "step": 34090
    },
    {
      "epoch": 8.456816913633828,
      "grad_norm": 1.0256129503250122,
      "learning_rate": 1.1515752170269278e-05,
      "loss": 1.3964,
      "step": 34100
    },
    {
      "epoch": 8.459296918593838,
      "grad_norm": 0.9788112044334412,
      "learning_rate": 1.1479478497498797e-05,
      "loss": 1.4095,
      "step": 34110
    },
    {
      "epoch": 8.461776923553847,
      "grad_norm": 1.0624443292617798,
      "learning_rate": 1.1443258565344329e-05,
      "loss": 1.4325,
      "step": 34120
    },
    {
      "epoch": 8.464256928513857,
      "grad_norm": 1.0203397274017334,
      "learning_rate": 1.1407092395794906e-05,
      "loss": 1.4189,
      "step": 34130
    },
    {
      "epoch": 8.466736933473866,
      "grad_norm": 1.0989471673965454,
      "learning_rate": 1.1370980010806997e-05,
      "loss": 1.418,
      "step": 34140
    },
    {
      "epoch": 8.469216938433878,
      "grad_norm": 1.0702160596847534,
      "learning_rate": 1.1334921432304257e-05,
      "loss": 1.3736,
      "step": 34150
    },
    {
      "epoch": 8.471696943393887,
      "grad_norm": 1.0528777837753296,
      "learning_rate": 1.129891668217783e-05,
      "loss": 1.3751,
      "step": 34160
    },
    {
      "epoch": 8.474176948353897,
      "grad_norm": 0.9908372163772583,
      "learning_rate": 1.1262965782286116e-05,
      "loss": 1.4225,
      "step": 34170
    },
    {
      "epoch": 8.476656953313906,
      "grad_norm": 1.0847793817520142,
      "learning_rate": 1.1227068754454816e-05,
      "loss": 1.3987,
      "step": 34180
    },
    {
      "epoch": 8.479136958273916,
      "grad_norm": 1.0748522281646729,
      "learning_rate": 1.119122562047692e-05,
      "loss": 1.392,
      "step": 34190
    },
    {
      "epoch": 8.481616963233927,
      "grad_norm": 1.1547174453735352,
      "learning_rate": 1.1155436402112785e-05,
      "loss": 1.4226,
      "step": 34200
    },
    {
      "epoch": 8.484096968193937,
      "grad_norm": 0.9955134391784668,
      "learning_rate": 1.1119701121089876e-05,
      "loss": 1.4093,
      "step": 34210
    },
    {
      "epoch": 8.486576973153946,
      "grad_norm": 0.9794337153434753,
      "learning_rate": 1.1084019799103074e-05,
      "loss": 1.3061,
      "step": 34220
    },
    {
      "epoch": 8.489056978113956,
      "grad_norm": 1.0419962406158447,
      "learning_rate": 1.1048392457814405e-05,
      "loss": 1.4271,
      "step": 34230
    },
    {
      "epoch": 8.491536983073965,
      "grad_norm": 1.068056344985962,
      "learning_rate": 1.1012819118853147e-05,
      "loss": 1.4002,
      "step": 34240
    },
    {
      "epoch": 8.494016988033977,
      "grad_norm": 1.0033705234527588,
      "learning_rate": 1.0977299803815788e-05,
      "loss": 1.3555,
      "step": 34250
    },
    {
      "epoch": 8.496496992993986,
      "grad_norm": 1.0239999294281006,
      "learning_rate": 1.0941834534266083e-05,
      "loss": 1.4237,
      "step": 34260
    },
    {
      "epoch": 8.498976997953996,
      "grad_norm": 1.0543749332427979,
      "learning_rate": 1.0906423331734839e-05,
      "loss": 1.4626,
      "step": 34270
    },
    {
      "epoch": 8.501457002914005,
      "grad_norm": 0.9943702816963196,
      "learning_rate": 1.0871066217720172e-05,
      "loss": 1.388,
      "step": 34280
    },
    {
      "epoch": 8.503937007874015,
      "grad_norm": 1.0968214273452759,
      "learning_rate": 1.08357632136873e-05,
      "loss": 1.4635,
      "step": 34290
    },
    {
      "epoch": 8.506417012834026,
      "grad_norm": 1.0669578313827515,
      "learning_rate": 1.080051434106859e-05,
      "loss": 1.3861,
      "step": 34300
    },
    {
      "epoch": 8.508897017794036,
      "grad_norm": 1.0519565343856812,
      "learning_rate": 1.0765319621263548e-05,
      "loss": 1.4476,
      "step": 34310
    },
    {
      "epoch": 8.511377022754045,
      "grad_norm": 1.0022358894348145,
      "learning_rate": 1.0730179075638868e-05,
      "loss": 1.4748,
      "step": 34320
    },
    {
      "epoch": 8.513857027714055,
      "grad_norm": 1.0759047269821167,
      "learning_rate": 1.069509272552821e-05,
      "loss": 1.4024,
      "step": 34330
    },
    {
      "epoch": 8.516337032674066,
      "grad_norm": 1.0005208253860474,
      "learning_rate": 1.0660060592232491e-05,
      "loss": 1.4147,
      "step": 34340
    },
    {
      "epoch": 8.518817037634076,
      "grad_norm": 1.0109820365905762,
      "learning_rate": 1.0625082697019628e-05,
      "loss": 1.3991,
      "step": 34350
    },
    {
      "epoch": 8.521297042594085,
      "grad_norm": 1.0770360231399536,
      "learning_rate": 1.0590159061124605e-05,
      "loss": 1.4605,
      "step": 34360
    },
    {
      "epoch": 8.523777047554095,
      "grad_norm": 1.0791715383529663,
      "learning_rate": 1.0555289705749483e-05,
      "loss": 1.4202,
      "step": 34370
    },
    {
      "epoch": 8.526257052514104,
      "grad_norm": 1.031917929649353,
      "learning_rate": 1.0520474652063394e-05,
      "loss": 1.4212,
      "step": 34380
    },
    {
      "epoch": 8.528737057474116,
      "grad_norm": 1.0541309118270874,
      "learning_rate": 1.0485713921202489e-05,
      "loss": 1.4647,
      "step": 34390
    },
    {
      "epoch": 8.531217062434125,
      "grad_norm": 1.1268516778945923,
      "learning_rate": 1.0451007534269907e-05,
      "loss": 1.3857,
      "step": 34400
    },
    {
      "epoch": 8.533697067394135,
      "grad_norm": 1.0549654960632324,
      "learning_rate": 1.041635551233583e-05,
      "loss": 1.4261,
      "step": 34410
    },
    {
      "epoch": 8.536177072354144,
      "grad_norm": 1.158935785293579,
      "learning_rate": 1.0381757876437425e-05,
      "loss": 1.4153,
      "step": 34420
    },
    {
      "epoch": 8.538657077314154,
      "grad_norm": 1.0003379583358765,
      "learning_rate": 1.0347214647578818e-05,
      "loss": 1.4511,
      "step": 34430
    },
    {
      "epoch": 8.541137082274165,
      "grad_norm": 1.0294896364212036,
      "learning_rate": 1.0312725846731175e-05,
      "loss": 1.3688,
      "step": 34440
    },
    {
      "epoch": 8.543617087234175,
      "grad_norm": 1.0496424436569214,
      "learning_rate": 1.0278291494832538e-05,
      "loss": 1.3769,
      "step": 34450
    },
    {
      "epoch": 8.546097092194184,
      "grad_norm": 1.0207414627075195,
      "learning_rate": 1.0243911612787948e-05,
      "loss": 1.4401,
      "step": 34460
    },
    {
      "epoch": 8.548577097154194,
      "grad_norm": 1.054727554321289,
      "learning_rate": 1.0209586221469335e-05,
      "loss": 1.4387,
      "step": 34470
    },
    {
      "epoch": 8.551057102114203,
      "grad_norm": 1.0674917697906494,
      "learning_rate": 1.0175315341715597e-05,
      "loss": 1.4405,
      "step": 34480
    },
    {
      "epoch": 8.553537107074215,
      "grad_norm": 1.025215744972229,
      "learning_rate": 1.0141098994332454e-05,
      "loss": 1.4608,
      "step": 34490
    },
    {
      "epoch": 8.556017112034224,
      "grad_norm": 1.0221035480499268,
      "learning_rate": 1.0106937200092648e-05,
      "loss": 1.4309,
      "step": 34500
    },
    {
      "epoch": 8.558497116994234,
      "grad_norm": 1.072180151939392,
      "learning_rate": 1.0072829979735699e-05,
      "loss": 1.4098,
      "step": 34510
    },
    {
      "epoch": 8.560977121954243,
      "grad_norm": 1.0292712450027466,
      "learning_rate": 1.003877735396801e-05,
      "loss": 1.3924,
      "step": 34520
    },
    {
      "epoch": 8.563457126914255,
      "grad_norm": 1.0132172107696533,
      "learning_rate": 1.0004779343462867e-05,
      "loss": 1.4269,
      "step": 34530
    },
    {
      "epoch": 8.565937131874264,
      "grad_norm": 1.067787766456604,
      "learning_rate": 9.970835968860414e-06,
      "loss": 1.4548,
      "step": 34540
    },
    {
      "epoch": 8.568417136834274,
      "grad_norm": 1.0257725715637207,
      "learning_rate": 9.936947250767558e-06,
      "loss": 1.4267,
      "step": 34550
    },
    {
      "epoch": 8.570897141794283,
      "grad_norm": 1.1108858585357666,
      "learning_rate": 9.903113209758096e-06,
      "loss": 1.3806,
      "step": 34560
    },
    {
      "epoch": 8.573377146754293,
      "grad_norm": 1.0043811798095703,
      "learning_rate": 9.869333866372587e-06,
      "loss": 1.4635,
      "step": 34570
    },
    {
      "epoch": 8.575857151714303,
      "grad_norm": 0.9807415008544922,
      "learning_rate": 9.835609241118404e-06,
      "loss": 1.4018,
      "step": 34580
    },
    {
      "epoch": 8.578337156674314,
      "grad_norm": 1.0493428707122803,
      "learning_rate": 9.80193935446967e-06,
      "loss": 1.4173,
      "step": 34590
    },
    {
      "epoch": 8.580817161634323,
      "grad_norm": 1.2100938558578491,
      "learning_rate": 9.768324226867353e-06,
      "loss": 1.4688,
      "step": 34600
    },
    {
      "epoch": 8.583297166594333,
      "grad_norm": 1.0792169570922852,
      "learning_rate": 9.734763878719067e-06,
      "loss": 1.3982,
      "step": 34610
    },
    {
      "epoch": 8.585777171554343,
      "grad_norm": 1.0124642848968506,
      "learning_rate": 9.701258330399255e-06,
      "loss": 1.4573,
      "step": 34620
    },
    {
      "epoch": 8.588257176514354,
      "grad_norm": 1.063339114189148,
      "learning_rate": 9.667807602249057e-06,
      "loss": 1.4195,
      "step": 34630
    },
    {
      "epoch": 8.590737181474363,
      "grad_norm": 1.03120756149292,
      "learning_rate": 9.634411714576353e-06,
      "loss": 1.4046,
      "step": 34640
    },
    {
      "epoch": 8.593217186434373,
      "grad_norm": 1.0576833486557007,
      "learning_rate": 9.601070687655667e-06,
      "loss": 1.4082,
      "step": 34650
    },
    {
      "epoch": 8.595697191394382,
      "grad_norm": 1.1593866348266602,
      "learning_rate": 9.56778454172832e-06,
      "loss": 1.4311,
      "step": 34660
    },
    {
      "epoch": 8.598177196354392,
      "grad_norm": 1.015978217124939,
      "learning_rate": 9.534553297002225e-06,
      "loss": 1.4241,
      "step": 34670
    },
    {
      "epoch": 8.600657201314403,
      "grad_norm": 1.0379754304885864,
      "learning_rate": 9.501376973651999e-06,
      "loss": 1.409,
      "step": 34680
    },
    {
      "epoch": 8.603137206274413,
      "grad_norm": 1.065412998199463,
      "learning_rate": 9.468255591818953e-06,
      "loss": 1.4255,
      "step": 34690
    },
    {
      "epoch": 8.605617211234422,
      "grad_norm": 1.0218896865844727,
      "learning_rate": 9.435189171610948e-06,
      "loss": 1.3884,
      "step": 34700
    },
    {
      "epoch": 8.608097216194432,
      "grad_norm": 1.1107914447784424,
      "learning_rate": 9.402177733102579e-06,
      "loss": 1.3714,
      "step": 34710
    },
    {
      "epoch": 8.610577221154442,
      "grad_norm": 1.1440714597702026,
      "learning_rate": 9.369221296335006e-06,
      "loss": 1.4462,
      "step": 34720
    },
    {
      "epoch": 8.613057226114453,
      "grad_norm": 0.9951843023300171,
      "learning_rate": 9.336319881316014e-06,
      "loss": 1.4014,
      "step": 34730
    },
    {
      "epoch": 8.615537231074462,
      "grad_norm": 1.0567536354064941,
      "learning_rate": 9.303473508019944e-06,
      "loss": 1.4029,
      "step": 34740
    },
    {
      "epoch": 8.618017236034472,
      "grad_norm": 1.1181285381317139,
      "learning_rate": 9.270682196387836e-06,
      "loss": 1.4163,
      "step": 34750
    },
    {
      "epoch": 8.620497240994482,
      "grad_norm": 1.0565319061279297,
      "learning_rate": 9.237945966327133e-06,
      "loss": 1.3815,
      "step": 34760
    },
    {
      "epoch": 8.622977245954491,
      "grad_norm": 1.017255187034607,
      "learning_rate": 9.205264837711991e-06,
      "loss": 1.4408,
      "step": 34770
    },
    {
      "epoch": 8.625457250914502,
      "grad_norm": 1.1201671361923218,
      "learning_rate": 9.17263883038304e-06,
      "loss": 1.426,
      "step": 34780
    },
    {
      "epoch": 8.627937255874512,
      "grad_norm": 1.0846129655838013,
      "learning_rate": 9.140067964147447e-06,
      "loss": 1.4052,
      "step": 34790
    },
    {
      "epoch": 8.630417260834522,
      "grad_norm": 1.1419168710708618,
      "learning_rate": 9.107552258778907e-06,
      "loss": 1.4286,
      "step": 34800
    },
    {
      "epoch": 8.632897265794531,
      "grad_norm": 1.0810461044311523,
      "learning_rate": 9.075091734017682e-06,
      "loss": 1.4161,
      "step": 34810
    },
    {
      "epoch": 8.635377270754542,
      "grad_norm": 1.0595791339874268,
      "learning_rate": 9.04268640957041e-06,
      "loss": 1.3417,
      "step": 34820
    },
    {
      "epoch": 8.637857275714552,
      "grad_norm": 1.0288753509521484,
      "learning_rate": 9.010336305110345e-06,
      "loss": 1.3905,
      "step": 34830
    },
    {
      "epoch": 8.640337280674562,
      "grad_norm": 1.0574288368225098,
      "learning_rate": 8.978041440277163e-06,
      "loss": 1.4734,
      "step": 34840
    },
    {
      "epoch": 8.642817285634571,
      "grad_norm": 1.0483571290969849,
      "learning_rate": 8.945801834676992e-06,
      "loss": 1.3855,
      "step": 34850
    },
    {
      "epoch": 8.64529729059458,
      "grad_norm": 1.087746262550354,
      "learning_rate": 8.91361750788241e-06,
      "loss": 1.3799,
      "step": 34860
    },
    {
      "epoch": 8.64777729555459,
      "grad_norm": 1.0188192129135132,
      "learning_rate": 8.88148847943251e-06,
      "loss": 1.3831,
      "step": 34870
    },
    {
      "epoch": 8.650257300514602,
      "grad_norm": 1.0634173154830933,
      "learning_rate": 8.849414768832687e-06,
      "loss": 1.4252,
      "step": 34880
    },
    {
      "epoch": 8.652737305474611,
      "grad_norm": 0.9798762798309326,
      "learning_rate": 8.817396395554866e-06,
      "loss": 1.3535,
      "step": 34890
    },
    {
      "epoch": 8.65521731043462,
      "grad_norm": 1.0635244846343994,
      "learning_rate": 8.78543337903731e-06,
      "loss": 1.3672,
      "step": 34900
    },
    {
      "epoch": 8.65769731539463,
      "grad_norm": 1.0241966247558594,
      "learning_rate": 8.75352573868471e-06,
      "loss": 1.4077,
      "step": 34910
    },
    {
      "epoch": 8.660177320354641,
      "grad_norm": 1.1106797456741333,
      "learning_rate": 8.72167349386811e-06,
      "loss": 1.4681,
      "step": 34920
    },
    {
      "epoch": 8.662657325314651,
      "grad_norm": 1.0830975770950317,
      "learning_rate": 8.689876663924957e-06,
      "loss": 1.4076,
      "step": 34930
    },
    {
      "epoch": 8.66513733027466,
      "grad_norm": 1.0571904182434082,
      "learning_rate": 8.658135268159018e-06,
      "loss": 1.4542,
      "step": 34940
    },
    {
      "epoch": 8.66761733523467,
      "grad_norm": 1.0933750867843628,
      "learning_rate": 8.626449325840447e-06,
      "loss": 1.4156,
      "step": 34950
    },
    {
      "epoch": 8.67009734019468,
      "grad_norm": 1.0800130367279053,
      "learning_rate": 8.5948188562057e-06,
      "loss": 1.3905,
      "step": 34960
    },
    {
      "epoch": 8.672577345154691,
      "grad_norm": 0.9996431469917297,
      "learning_rate": 8.56324387845756e-06,
      "loss": 1.5192,
      "step": 34970
    },
    {
      "epoch": 8.6750573501147,
      "grad_norm": 1.0539569854736328,
      "learning_rate": 8.531724411765119e-06,
      "loss": 1.3983,
      "step": 34980
    },
    {
      "epoch": 8.67753735507471,
      "grad_norm": 1.0754151344299316,
      "learning_rate": 8.5002604752638e-06,
      "loss": 1.4119,
      "step": 34990
    },
    {
      "epoch": 8.68001736003472,
      "grad_norm": 1.1121124029159546,
      "learning_rate": 8.46885208805529e-06,
      "loss": 1.4655,
      "step": 35000
    },
    {
      "epoch": 8.68249736499473,
      "grad_norm": 1.0148959159851074,
      "learning_rate": 8.437499269207539e-06,
      "loss": 1.3647,
      "step": 35010
    },
    {
      "epoch": 8.68497736995474,
      "grad_norm": 1.091356873512268,
      "learning_rate": 8.406202037754774e-06,
      "loss": 1.4008,
      "step": 35020
    },
    {
      "epoch": 8.68745737491475,
      "grad_norm": 1.0934149026870728,
      "learning_rate": 8.374960412697485e-06,
      "loss": 1.4645,
      "step": 35030
    },
    {
      "epoch": 8.68993737987476,
      "grad_norm": 1.1903691291809082,
      "learning_rate": 8.343774413002381e-06,
      "loss": 1.4172,
      "step": 35040
    },
    {
      "epoch": 8.69241738483477,
      "grad_norm": 1.0634151697158813,
      "learning_rate": 8.312644057602436e-06,
      "loss": 1.4175,
      "step": 35050
    },
    {
      "epoch": 8.694897389794779,
      "grad_norm": 1.0880907773971558,
      "learning_rate": 8.281569365396812e-06,
      "loss": 1.4677,
      "step": 35060
    },
    {
      "epoch": 8.69737739475479,
      "grad_norm": 1.0232330560684204,
      "learning_rate": 8.250550355250875e-06,
      "loss": 1.459,
      "step": 35070
    },
    {
      "epoch": 8.6998573997148,
      "grad_norm": 1.0494637489318848,
      "learning_rate": 8.219587045996203e-06,
      "loss": 1.4531,
      "step": 35080
    },
    {
      "epoch": 8.70233740467481,
      "grad_norm": 1.0898422002792358,
      "learning_rate": 8.18867945643058e-06,
      "loss": 1.385,
      "step": 35090
    },
    {
      "epoch": 8.704817409634819,
      "grad_norm": 1.0978202819824219,
      "learning_rate": 8.157827605317892e-06,
      "loss": 1.3983,
      "step": 35100
    },
    {
      "epoch": 8.70729741459483,
      "grad_norm": 1.0695819854736328,
      "learning_rate": 8.127031511388273e-06,
      "loss": 1.444,
      "step": 35110
    },
    {
      "epoch": 8.70977741955484,
      "grad_norm": 1.1173222064971924,
      "learning_rate": 8.096291193337934e-06,
      "loss": 1.3457,
      "step": 35120
    },
    {
      "epoch": 8.71225742451485,
      "grad_norm": 1.062896966934204,
      "learning_rate": 8.065606669829273e-06,
      "loss": 1.463,
      "step": 35130
    },
    {
      "epoch": 8.714737429474859,
      "grad_norm": 1.0228677988052368,
      "learning_rate": 8.034977959490775e-06,
      "loss": 1.4077,
      "step": 35140
    },
    {
      "epoch": 8.717217434434868,
      "grad_norm": 1.1282140016555786,
      "learning_rate": 8.00440508091711e-06,
      "loss": 1.4278,
      "step": 35150
    },
    {
      "epoch": 8.719697439394878,
      "grad_norm": 1.1063153743743896,
      "learning_rate": 7.973888052668943e-06,
      "loss": 1.3547,
      "step": 35160
    },
    {
      "epoch": 8.72217744435489,
      "grad_norm": 1.0718064308166504,
      "learning_rate": 7.943426893273142e-06,
      "loss": 1.3838,
      "step": 35170
    },
    {
      "epoch": 8.724657449314899,
      "grad_norm": 0.9802948832511902,
      "learning_rate": 7.913021621222604e-06,
      "loss": 1.4587,
      "step": 35180
    },
    {
      "epoch": 8.727137454274908,
      "grad_norm": 1.1451220512390137,
      "learning_rate": 7.882672254976297e-06,
      "loss": 1.3912,
      "step": 35190
    },
    {
      "epoch": 8.729617459234918,
      "grad_norm": 1.0279031991958618,
      "learning_rate": 7.852378812959227e-06,
      "loss": 1.4077,
      "step": 35200
    },
    {
      "epoch": 8.73209746419493,
      "grad_norm": 1.0391227006912231,
      "learning_rate": 7.822141313562547e-06,
      "loss": 1.4476,
      "step": 35210
    },
    {
      "epoch": 8.734577469154939,
      "grad_norm": 1.1570180654525757,
      "learning_rate": 7.791959775143298e-06,
      "loss": 1.3834,
      "step": 35220
    },
    {
      "epoch": 8.737057474114948,
      "grad_norm": 1.0537258386611938,
      "learning_rate": 7.761834216024678e-06,
      "loss": 1.3613,
      "step": 35230
    },
    {
      "epoch": 8.739537479074958,
      "grad_norm": 0.986975371837616,
      "learning_rate": 7.731764654495832e-06,
      "loss": 1.4096,
      "step": 35240
    },
    {
      "epoch": 8.742017484034967,
      "grad_norm": 1.1508692502975464,
      "learning_rate": 7.701751108811938e-06,
      "loss": 1.4927,
      "step": 35250
    },
    {
      "epoch": 8.744497488994979,
      "grad_norm": 1.0710281133651733,
      "learning_rate": 7.67179359719411e-06,
      "loss": 1.3809,
      "step": 35260
    },
    {
      "epoch": 8.746977493954988,
      "grad_norm": 1.0391231775283813,
      "learning_rate": 7.641892137829566e-06,
      "loss": 1.3815,
      "step": 35270
    },
    {
      "epoch": 8.749457498914998,
      "grad_norm": 1.0123393535614014,
      "learning_rate": 7.612046748871327e-06,
      "loss": 1.3923,
      "step": 35280
    },
    {
      "epoch": 8.751937503875007,
      "grad_norm": 1.0555802583694458,
      "learning_rate": 7.582257448438512e-06,
      "loss": 1.382,
      "step": 35290
    },
    {
      "epoch": 8.754417508835017,
      "grad_norm": 1.1561102867126465,
      "learning_rate": 7.552524254616178e-06,
      "loss": 1.3976,
      "step": 35300
    },
    {
      "epoch": 8.756897513795028,
      "grad_norm": 1.1613948345184326,
      "learning_rate": 7.522847185455195e-06,
      "loss": 1.4498,
      "step": 35310
    },
    {
      "epoch": 8.759377518755038,
      "grad_norm": 1.0044676065444946,
      "learning_rate": 7.493226258972519e-06,
      "loss": 1.3991,
      "step": 35320
    },
    {
      "epoch": 8.761857523715047,
      "grad_norm": 1.0604032278060913,
      "learning_rate": 7.4636614931509304e-06,
      "loss": 1.3658,
      "step": 35330
    },
    {
      "epoch": 8.764337528675057,
      "grad_norm": 1.0889891386032104,
      "learning_rate": 7.434152905939118e-06,
      "loss": 1.4178,
      "step": 35340
    },
    {
      "epoch": 8.766817533635066,
      "grad_norm": 1.1055028438568115,
      "learning_rate": 7.404700515251672e-06,
      "loss": 1.4237,
      "step": 35350
    },
    {
      "epoch": 8.769297538595078,
      "grad_norm": 1.104724645614624,
      "learning_rate": 7.375304338969136e-06,
      "loss": 1.4087,
      "step": 35360
    },
    {
      "epoch": 8.771777543555087,
      "grad_norm": 1.016101598739624,
      "learning_rate": 7.345964394937787e-06,
      "loss": 1.3964,
      "step": 35370
    },
    {
      "epoch": 8.774257548515097,
      "grad_norm": 1.0164542198181152,
      "learning_rate": 7.316680700969891e-06,
      "loss": 1.4447,
      "step": 35380
    },
    {
      "epoch": 8.776737553475106,
      "grad_norm": 0.9957998991012573,
      "learning_rate": 7.287453274843492e-06,
      "loss": 1.3885,
      "step": 35390
    },
    {
      "epoch": 8.779217558435118,
      "grad_norm": 1.13089919090271,
      "learning_rate": 7.258282134302519e-06,
      "loss": 1.4144,
      "step": 35400
    },
    {
      "epoch": 8.781697563395127,
      "grad_norm": 1.068381428718567,
      "learning_rate": 7.229167297056671e-06,
      "loss": 1.3725,
      "step": 35410
    },
    {
      "epoch": 8.784177568355137,
      "grad_norm": 1.0708982944488525,
      "learning_rate": 7.200108780781556e-06,
      "loss": 1.4271,
      "step": 35420
    },
    {
      "epoch": 8.786657573315146,
      "grad_norm": 1.0746763944625854,
      "learning_rate": 7.171106603118482e-06,
      "loss": 1.401,
      "step": 35430
    },
    {
      "epoch": 8.789137578275156,
      "grad_norm": 1.074387550354004,
      "learning_rate": 7.142160781674645e-06,
      "loss": 1.4837,
      "step": 35440
    },
    {
      "epoch": 8.791617583235167,
      "grad_norm": 1.0946547985076904,
      "learning_rate": 7.1132713340229755e-06,
      "loss": 1.3906,
      "step": 35450
    },
    {
      "epoch": 8.794097588195177,
      "grad_norm": 1.0786830186843872,
      "learning_rate": 7.084438277702188e-06,
      "loss": 1.3938,
      "step": 35460
    },
    {
      "epoch": 8.796577593155186,
      "grad_norm": 1.033219575881958,
      "learning_rate": 7.05566163021677e-06,
      "loss": 1.3912,
      "step": 35470
    },
    {
      "epoch": 8.799057598115196,
      "grad_norm": 1.1158262491226196,
      "learning_rate": 7.026941409036991e-06,
      "loss": 1.459,
      "step": 35480
    },
    {
      "epoch": 8.801537603075205,
      "grad_norm": 0.9801075458526611,
      "learning_rate": 6.998277631598793e-06,
      "loss": 1.4496,
      "step": 35490
    },
    {
      "epoch": 8.804017608035217,
      "grad_norm": 1.01754891872406,
      "learning_rate": 6.969670315303911e-06,
      "loss": 1.5087,
      "step": 35500
    },
    {
      "epoch": 8.806497612995226,
      "grad_norm": 1.0917232036590576,
      "learning_rate": 6.941119477519798e-06,
      "loss": 1.3484,
      "step": 35510
    },
    {
      "epoch": 8.808977617955236,
      "grad_norm": 0.9824590086936951,
      "learning_rate": 6.9126251355795864e-06,
      "loss": 1.4118,
      "step": 35520
    },
    {
      "epoch": 8.811457622915245,
      "grad_norm": 1.0820149183273315,
      "learning_rate": 6.884187306782119e-06,
      "loss": 1.4271,
      "step": 35530
    },
    {
      "epoch": 8.813937627875255,
      "grad_norm": 1.0591022968292236,
      "learning_rate": 6.855806008391974e-06,
      "loss": 1.3741,
      "step": 35540
    },
    {
      "epoch": 8.816417632835266,
      "grad_norm": 0.9803717732429504,
      "learning_rate": 6.827481257639345e-06,
      "loss": 1.4386,
      "step": 35550
    },
    {
      "epoch": 8.818897637795276,
      "grad_norm": 1.0050376653671265,
      "learning_rate": 6.7992130717201564e-06,
      "loss": 1.36,
      "step": 35560
    },
    {
      "epoch": 8.821377642755285,
      "grad_norm": 1.0928151607513428,
      "learning_rate": 6.771001467795934e-06,
      "loss": 1.3858,
      "step": 35570
    },
    {
      "epoch": 8.823857647715295,
      "grad_norm": 1.05537748336792,
      "learning_rate": 6.742846462993901e-06,
      "loss": 1.3562,
      "step": 35580
    },
    {
      "epoch": 8.826337652675305,
      "grad_norm": 1.044603943824768,
      "learning_rate": 6.714748074406874e-06,
      "loss": 1.4484,
      "step": 35590
    },
    {
      "epoch": 8.828817657635316,
      "grad_norm": 1.0399260520935059,
      "learning_rate": 6.68670631909335e-06,
      "loss": 1.3723,
      "step": 35600
    },
    {
      "epoch": 8.831297662595325,
      "grad_norm": 1.0280603170394897,
      "learning_rate": 6.658721214077412e-06,
      "loss": 1.4067,
      "step": 35610
    },
    {
      "epoch": 8.833777667555335,
      "grad_norm": 1.0593305826187134,
      "learning_rate": 6.630792776348749e-06,
      "loss": 1.3742,
      "step": 35620
    },
    {
      "epoch": 8.836257672515345,
      "grad_norm": 1.0912466049194336,
      "learning_rate": 6.6029210228626626e-06,
      "loss": 1.4118,
      "step": 35630
    },
    {
      "epoch": 8.838737677475354,
      "grad_norm": 1.05013108253479,
      "learning_rate": 6.5751059705400295e-06,
      "loss": 1.4158,
      "step": 35640
    },
    {
      "epoch": 8.841217682435365,
      "grad_norm": 0.9891907572746277,
      "learning_rate": 6.547347636267287e-06,
      "loss": 1.4039,
      "step": 35650
    },
    {
      "epoch": 8.843697687395375,
      "grad_norm": 1.0187528133392334,
      "learning_rate": 6.519646036896499e-06,
      "loss": 1.3678,
      "step": 35660
    },
    {
      "epoch": 8.846177692355385,
      "grad_norm": 1.0717898607254028,
      "learning_rate": 6.492001189245223e-06,
      "loss": 1.4115,
      "step": 35670
    },
    {
      "epoch": 8.848657697315394,
      "grad_norm": 1.029092788696289,
      "learning_rate": 6.464413110096601e-06,
      "loss": 1.3866,
      "step": 35680
    },
    {
      "epoch": 8.851137702275405,
      "grad_norm": 1.0228272676467896,
      "learning_rate": 6.436881816199258e-06,
      "loss": 1.373,
      "step": 35690
    },
    {
      "epoch": 8.853617707235415,
      "grad_norm": 1.0140708684921265,
      "learning_rate": 6.409407324267447e-06,
      "loss": 1.3329,
      "step": 35700
    },
    {
      "epoch": 8.856097712195425,
      "grad_norm": 0.9810483455657959,
      "learning_rate": 6.381989650980802e-06,
      "loss": 1.4641,
      "step": 35710
    },
    {
      "epoch": 8.858577717155434,
      "grad_norm": 1.127550721168518,
      "learning_rate": 6.354628812984576e-06,
      "loss": 1.4714,
      "step": 35720
    },
    {
      "epoch": 8.861057722115444,
      "grad_norm": 1.0540467500686646,
      "learning_rate": 6.327324826889469e-06,
      "loss": 1.3891,
      "step": 35730
    },
    {
      "epoch": 8.863537727075455,
      "grad_norm": 1.0507473945617676,
      "learning_rate": 6.3000777092716566e-06,
      "loss": 1.3931,
      "step": 35740
    },
    {
      "epoch": 8.866017732035465,
      "grad_norm": 1.0516712665557861,
      "learning_rate": 6.272887476672806e-06,
      "loss": 1.4759,
      "step": 35750
    },
    {
      "epoch": 8.868497736995474,
      "grad_norm": 1.0787099599838257,
      "learning_rate": 6.245754145600091e-06,
      "loss": 1.3865,
      "step": 35760
    },
    {
      "epoch": 8.870977741955484,
      "grad_norm": 1.0408920049667358,
      "learning_rate": 6.218677732526035e-06,
      "loss": 1.4059,
      "step": 35770
    },
    {
      "epoch": 8.873457746915493,
      "grad_norm": 1.0530054569244385,
      "learning_rate": 6.191658253888699e-06,
      "loss": 1.3508,
      "step": 35780
    },
    {
      "epoch": 8.875937751875504,
      "grad_norm": 1.031050443649292,
      "learning_rate": 6.164695726091563e-06,
      "loss": 1.4165,
      "step": 35790
    },
    {
      "epoch": 8.878417756835514,
      "grad_norm": 1.0282436609268188,
      "learning_rate": 6.137790165503499e-06,
      "loss": 1.4261,
      "step": 35800
    },
    {
      "epoch": 8.880897761795524,
      "grad_norm": 1.0038362741470337,
      "learning_rate": 6.110941588458807e-06,
      "loss": 1.4105,
      "step": 35810
    },
    {
      "epoch": 8.883377766755533,
      "grad_norm": 1.0419639348983765,
      "learning_rate": 6.084150011257239e-06,
      "loss": 1.3816,
      "step": 35820
    },
    {
      "epoch": 8.885857771715543,
      "grad_norm": 1.1044398546218872,
      "learning_rate": 6.057415450163839e-06,
      "loss": 1.3929,
      "step": 35830
    },
    {
      "epoch": 8.888337776675554,
      "grad_norm": 1.10807466506958,
      "learning_rate": 6.030737921409169e-06,
      "loss": 1.3409,
      "step": 35840
    },
    {
      "epoch": 8.890817781635564,
      "grad_norm": 1.1144291162490845,
      "learning_rate": 6.004117441189061e-06,
      "loss": 1.4778,
      "step": 35850
    },
    {
      "epoch": 8.893297786595573,
      "grad_norm": 0.9802186489105225,
      "learning_rate": 5.977554025664755e-06,
      "loss": 1.3595,
      "step": 35860
    },
    {
      "epoch": 8.895777791555583,
      "grad_norm": 1.0574133396148682,
      "learning_rate": 5.951047690962841e-06,
      "loss": 1.4007,
      "step": 35870
    },
    {
      "epoch": 8.898257796515592,
      "grad_norm": 0.9653461575508118,
      "learning_rate": 5.9245984531752784e-06,
      "loss": 1.4166,
      "step": 35880
    },
    {
      "epoch": 8.900737801475604,
      "grad_norm": 1.0629173517227173,
      "learning_rate": 5.898206328359302e-06,
      "loss": 1.3397,
      "step": 35890
    },
    {
      "epoch": 8.903217806435613,
      "grad_norm": 1.0503249168395996,
      "learning_rate": 5.8718713325375485e-06,
      "loss": 1.4268,
      "step": 35900
    },
    {
      "epoch": 8.905697811395623,
      "grad_norm": 1.0458600521087646,
      "learning_rate": 5.8455934816979305e-06,
      "loss": 1.4417,
      "step": 35910
    },
    {
      "epoch": 8.908177816355632,
      "grad_norm": 1.0539132356643677,
      "learning_rate": 5.8193727917936536e-06,
      "loss": 1.3868,
      "step": 35920
    },
    {
      "epoch": 8.910657821315642,
      "grad_norm": 1.0538716316223145,
      "learning_rate": 5.793209278743261e-06,
      "loss": 1.4121,
      "step": 35930
    },
    {
      "epoch": 8.913137826275653,
      "grad_norm": 1.0634170770645142,
      "learning_rate": 5.76710295843057e-06,
      "loss": 1.3727,
      "step": 35940
    },
    {
      "epoch": 8.915617831235663,
      "grad_norm": 1.0303020477294922,
      "learning_rate": 5.741053846704658e-06,
      "loss": 1.4504,
      "step": 35950
    },
    {
      "epoch": 8.918097836195672,
      "grad_norm": 1.0691232681274414,
      "learning_rate": 5.715061959379875e-06,
      "loss": 1.3289,
      "step": 35960
    },
    {
      "epoch": 8.920577841155682,
      "grad_norm": 1.0344038009643555,
      "learning_rate": 5.689127312235876e-06,
      "loss": 1.4227,
      "step": 35970
    },
    {
      "epoch": 8.923057846115693,
      "grad_norm": 1.0390594005584717,
      "learning_rate": 5.663249921017477e-06,
      "loss": 1.3365,
      "step": 35980
    },
    {
      "epoch": 8.925537851075703,
      "grad_norm": 1.0197139978408813,
      "learning_rate": 5.637429801434824e-06,
      "loss": 1.3722,
      "step": 35990
    },
    {
      "epoch": 8.928017856035712,
      "grad_norm": 1.0387476682662964,
      "learning_rate": 5.611666969163243e-06,
      "loss": 1.3939,
      "step": 36000
    },
    {
      "epoch": 8.930497860995722,
      "grad_norm": 1.1280559301376343,
      "learning_rate": 5.585961439843302e-06,
      "loss": 1.3973,
      "step": 36010
    },
    {
      "epoch": 8.932977865955731,
      "grad_norm": 1.0584766864776611,
      "learning_rate": 5.560313229080749e-06,
      "loss": 1.4572,
      "step": 36020
    },
    {
      "epoch": 8.935457870915743,
      "grad_norm": 1.0336962938308716,
      "learning_rate": 5.534722352446598e-06,
      "loss": 1.3458,
      "step": 36030
    },
    {
      "epoch": 8.937937875875752,
      "grad_norm": 1.0421390533447266,
      "learning_rate": 5.509188825476963e-06,
      "loss": 1.366,
      "step": 36040
    },
    {
      "epoch": 8.940417880835762,
      "grad_norm": 1.0823450088500977,
      "learning_rate": 5.483712663673224e-06,
      "loss": 1.4318,
      "step": 36050
    },
    {
      "epoch": 8.942897885795771,
      "grad_norm": 0.9930027723312378,
      "learning_rate": 5.458293882501908e-06,
      "loss": 1.445,
      "step": 36060
    },
    {
      "epoch": 8.94537789075578,
      "grad_norm": 0.9809021353721619,
      "learning_rate": 5.432932497394694e-06,
      "loss": 1.3888,
      "step": 36070
    },
    {
      "epoch": 8.947857895715792,
      "grad_norm": 1.074016809463501,
      "learning_rate": 5.4076285237483984e-06,
      "loss": 1.4226,
      "step": 36080
    },
    {
      "epoch": 8.950337900675802,
      "grad_norm": 1.1049420833587646,
      "learning_rate": 5.382381976925044e-06,
      "loss": 1.3478,
      "step": 36090
    },
    {
      "epoch": 8.952817905635811,
      "grad_norm": 1.0886642932891846,
      "learning_rate": 5.357192872251748e-06,
      "loss": 1.3751,
      "step": 36100
    },
    {
      "epoch": 8.95529791059582,
      "grad_norm": 0.975545346736908,
      "learning_rate": 5.332061225020757e-06,
      "loss": 1.5169,
      "step": 36110
    },
    {
      "epoch": 8.95777791555583,
      "grad_norm": 1.1444214582443237,
      "learning_rate": 5.306987050489442e-06,
      "loss": 1.3858,
      "step": 36120
    },
    {
      "epoch": 8.960257920515842,
      "grad_norm": 1.0759073495864868,
      "learning_rate": 5.281970363880262e-06,
      "loss": 1.3536,
      "step": 36130
    },
    {
      "epoch": 8.962737925475851,
      "grad_norm": 1.1276054382324219,
      "learning_rate": 5.2570111803808e-06,
      "loss": 1.4528,
      "step": 36140
    },
    {
      "epoch": 8.96521793043586,
      "grad_norm": 1.0630743503570557,
      "learning_rate": 5.232109515143746e-06,
      "loss": 1.4241,
      "step": 36150
    },
    {
      "epoch": 8.96769793539587,
      "grad_norm": 1.0743145942687988,
      "learning_rate": 5.20726538328683e-06,
      "loss": 1.4425,
      "step": 36160
    },
    {
      "epoch": 8.97017794035588,
      "grad_norm": 1.0920917987823486,
      "learning_rate": 5.182478799892865e-06,
      "loss": 1.4461,
      "step": 36170
    },
    {
      "epoch": 8.972657945315891,
      "grad_norm": 1.0802286863327026,
      "learning_rate": 5.157749780009735e-06,
      "loss": 1.4009,
      "step": 36180
    },
    {
      "epoch": 8.9751379502759,
      "grad_norm": 1.0261247158050537,
      "learning_rate": 5.1330783386503765e-06,
      "loss": 1.4202,
      "step": 36190
    },
    {
      "epoch": 8.97761795523591,
      "grad_norm": 1.0045171976089478,
      "learning_rate": 5.1084644907927526e-06,
      "loss": 1.4174,
      "step": 36200
    },
    {
      "epoch": 8.98009796019592,
      "grad_norm": 1.1183091402053833,
      "learning_rate": 5.08390825137991e-06,
      "loss": 1.4168,
      "step": 36210
    },
    {
      "epoch": 8.98257796515593,
      "grad_norm": 1.0777711868286133,
      "learning_rate": 5.059409635319867e-06,
      "loss": 1.3862,
      "step": 36220
    },
    {
      "epoch": 8.98505797011594,
      "grad_norm": 1.0636361837387085,
      "learning_rate": 5.034968657485673e-06,
      "loss": 1.4373,
      "step": 36230
    },
    {
      "epoch": 8.98753797507595,
      "grad_norm": 1.0551748275756836,
      "learning_rate": 5.0105853327154e-06,
      "loss": 1.3829,
      "step": 36240
    },
    {
      "epoch": 8.99001798003596,
      "grad_norm": 1.0396533012390137,
      "learning_rate": 4.986259675812133e-06,
      "loss": 1.4231,
      "step": 36250
    },
    {
      "epoch": 8.99249798499597,
      "grad_norm": 1.0980195999145508,
      "learning_rate": 4.961991701543889e-06,
      "loss": 1.4699,
      "step": 36260
    },
    {
      "epoch": 8.99497798995598,
      "grad_norm": 1.020045280456543,
      "learning_rate": 4.937781424643728e-06,
      "loss": 1.3617,
      "step": 36270
    },
    {
      "epoch": 8.99745799491599,
      "grad_norm": 1.0332356691360474,
      "learning_rate": 4.913628859809638e-06,
      "loss": 1.3875,
      "step": 36280
    },
    {
      "epoch": 8.999937999876,
      "grad_norm": 1.0232480764389038,
      "learning_rate": 4.889534021704611e-06,
      "loss": 1.401,
      "step": 36290
    },
    {
      "epoch": 9.00241800483601,
      "grad_norm": 1.0589970350265503,
      "learning_rate": 4.865496924956536e-06,
      "loss": 1.4866,
      "step": 36300
    },
    {
      "epoch": 9.004898009796019,
      "grad_norm": 1.0917316675186157,
      "learning_rate": 4.84151758415834e-06,
      "loss": 1.4285,
      "step": 36310
    },
    {
      "epoch": 9.00737801475603,
      "grad_norm": 0.9955704212188721,
      "learning_rate": 4.817596013867764e-06,
      "loss": 1.3862,
      "step": 36320
    },
    {
      "epoch": 9.00985801971604,
      "grad_norm": 1.0633028745651245,
      "learning_rate": 4.7937322286075725e-06,
      "loss": 1.4126,
      "step": 36330
    },
    {
      "epoch": 9.01233802467605,
      "grad_norm": 1.0090926885604858,
      "learning_rate": 4.7699262428654305e-06,
      "loss": 1.411,
      "step": 36340
    },
    {
      "epoch": 9.014818029636059,
      "grad_norm": 1.0382198095321655,
      "learning_rate": 4.746178071093876e-06,
      "loss": 1.4622,
      "step": 36350
    },
    {
      "epoch": 9.017298034596068,
      "grad_norm": 1.0646711587905884,
      "learning_rate": 4.722487727710368e-06,
      "loss": 1.3662,
      "step": 36360
    },
    {
      "epoch": 9.01977803955608,
      "grad_norm": 1.0893924236297607,
      "learning_rate": 4.698855227097298e-06,
      "loss": 1.4297,
      "step": 36370
    },
    {
      "epoch": 9.02225804451609,
      "grad_norm": 1.06475830078125,
      "learning_rate": 4.675280583601871e-06,
      "loss": 1.4096,
      "step": 36380
    },
    {
      "epoch": 9.024738049476099,
      "grad_norm": 0.9807024598121643,
      "learning_rate": 4.6517638115362136e-06,
      "loss": 1.3532,
      "step": 36390
    },
    {
      "epoch": 9.027218054436108,
      "grad_norm": 1.0107083320617676,
      "learning_rate": 4.628304925177318e-06,
      "loss": 1.4175,
      "step": 36400
    },
    {
      "epoch": 9.029698059396118,
      "grad_norm": 1.0552297830581665,
      "learning_rate": 4.604903938767002e-06,
      "loss": 1.4115,
      "step": 36410
    },
    {
      "epoch": 9.03217806435613,
      "grad_norm": 1.0633686780929565,
      "learning_rate": 4.581560866511958e-06,
      "loss": 1.3529,
      "step": 36420
    },
    {
      "epoch": 9.034658069316139,
      "grad_norm": 1.0776898860931396,
      "learning_rate": 4.558275722583749e-06,
      "loss": 1.418,
      "step": 36430
    },
    {
      "epoch": 9.037138074276148,
      "grad_norm": 1.032347321510315,
      "learning_rate": 4.5350485211186675e-06,
      "loss": 1.4072,
      "step": 36440
    },
    {
      "epoch": 9.039618079236158,
      "grad_norm": 1.0418621301651,
      "learning_rate": 4.5118792762179676e-06,
      "loss": 1.3685,
      "step": 36450
    },
    {
      "epoch": 9.042098084196168,
      "grad_norm": 0.9533702731132507,
      "learning_rate": 4.4887680019475965e-06,
      "loss": 1.4239,
      "step": 36460
    },
    {
      "epoch": 9.044578089156179,
      "grad_norm": 1.0281661748886108,
      "learning_rate": 4.465714712338398e-06,
      "loss": 1.358,
      "step": 36470
    },
    {
      "epoch": 9.047058094116188,
      "grad_norm": 1.0689723491668701,
      "learning_rate": 4.442719421385922e-06,
      "loss": 1.4248,
      "step": 36480
    },
    {
      "epoch": 9.049538099076198,
      "grad_norm": 1.0380345582962036,
      "learning_rate": 4.419782143050633e-06,
      "loss": 1.4749,
      "step": 36490
    },
    {
      "epoch": 9.052018104036208,
      "grad_norm": 1.0402356386184692,
      "learning_rate": 4.396902891257626e-06,
      "loss": 1.3932,
      "step": 36500
    },
    {
      "epoch": 9.054498108996219,
      "grad_norm": 1.0118567943572998,
      "learning_rate": 4.374081679896902e-06,
      "loss": 1.3717,
      "step": 36510
    },
    {
      "epoch": 9.056978113956228,
      "grad_norm": 1.0979758501052856,
      "learning_rate": 4.351318522823133e-06,
      "loss": 1.3867,
      "step": 36520
    },
    {
      "epoch": 9.059458118916238,
      "grad_norm": 0.9887623190879822,
      "learning_rate": 4.32861343385581e-06,
      "loss": 1.2905,
      "step": 36530
    },
    {
      "epoch": 9.061938123876248,
      "grad_norm": 1.1138519048690796,
      "learning_rate": 4.305966426779118e-06,
      "loss": 1.4328,
      "step": 36540
    },
    {
      "epoch": 9.064418128836257,
      "grad_norm": 1.0642791986465454,
      "learning_rate": 4.2833775153420245e-06,
      "loss": 1.3976,
      "step": 36550
    },
    {
      "epoch": 9.066898133796268,
      "grad_norm": 0.9739475846290588,
      "learning_rate": 4.260846713258193e-06,
      "loss": 1.4238,
      "step": 36560
    },
    {
      "epoch": 9.069378138756278,
      "grad_norm": 1.1281789541244507,
      "learning_rate": 4.238374034206028e-06,
      "loss": 1.4407,
      "step": 36570
    },
    {
      "epoch": 9.071858143716288,
      "grad_norm": 1.0865991115570068,
      "learning_rate": 4.215959491828681e-06,
      "loss": 1.3598,
      "step": 36580
    },
    {
      "epoch": 9.074338148676297,
      "grad_norm": 1.1083083152770996,
      "learning_rate": 4.1936030997339e-06,
      "loss": 1.4213,
      "step": 36590
    },
    {
      "epoch": 9.076818153636307,
      "grad_norm": 1.1080453395843506,
      "learning_rate": 4.171304871494264e-06,
      "loss": 1.4361,
      "step": 36600
    },
    {
      "epoch": 9.079298158596318,
      "grad_norm": 1.1190769672393799,
      "learning_rate": 4.149064820646953e-06,
      "loss": 1.4086,
      "step": 36610
    },
    {
      "epoch": 9.081778163556327,
      "grad_norm": 1.0619183778762817,
      "learning_rate": 4.126882960693868e-06,
      "loss": 1.4103,
      "step": 36620
    },
    {
      "epoch": 9.084258168516337,
      "grad_norm": 1.069216012954712,
      "learning_rate": 4.104759305101525e-06,
      "loss": 1.3868,
      "step": 36630
    },
    {
      "epoch": 9.086738173476347,
      "grad_norm": 1.0546233654022217,
      "learning_rate": 4.082693867301224e-06,
      "loss": 1.3681,
      "step": 36640
    },
    {
      "epoch": 9.089218178436356,
      "grad_norm": 1.0817266702651978,
      "learning_rate": 4.060686660688762e-06,
      "loss": 1.3694,
      "step": 36650
    },
    {
      "epoch": 9.091698183396367,
      "grad_norm": 1.005463719367981,
      "learning_rate": 4.0387376986247174e-06,
      "loss": 1.4149,
      "step": 36660
    },
    {
      "epoch": 9.094178188356377,
      "grad_norm": 1.0748347043991089,
      "learning_rate": 4.016846994434243e-06,
      "loss": 1.4455,
      "step": 36670
    },
    {
      "epoch": 9.096658193316387,
      "grad_norm": 1.0070174932479858,
      "learning_rate": 3.99501456140714e-06,
      "loss": 1.3758,
      "step": 36680
    },
    {
      "epoch": 9.099138198276396,
      "grad_norm": 1.0930790901184082,
      "learning_rate": 3.973240412797819e-06,
      "loss": 1.3981,
      "step": 36690
    },
    {
      "epoch": 9.101618203236406,
      "grad_norm": 0.9995220303535461,
      "learning_rate": 3.951524561825337e-06,
      "loss": 1.3719,
      "step": 36700
    },
    {
      "epoch": 9.104098208196417,
      "grad_norm": 0.9938735365867615,
      "learning_rate": 3.929867021673328e-06,
      "loss": 1.4177,
      "step": 36710
    },
    {
      "epoch": 9.106578213156427,
      "grad_norm": 1.043028712272644,
      "learning_rate": 3.908267805490051e-06,
      "loss": 1.3514,
      "step": 36720
    },
    {
      "epoch": 9.109058218116436,
      "grad_norm": 1.078033208847046,
      "learning_rate": 3.886726926388329e-06,
      "loss": 1.4582,
      "step": 36730
    },
    {
      "epoch": 9.111538223076446,
      "grad_norm": 0.9632874131202698,
      "learning_rate": 3.865244397445577e-06,
      "loss": 1.3785,
      "step": 36740
    },
    {
      "epoch": 9.114018228036455,
      "grad_norm": 1.1127055883407593,
      "learning_rate": 3.8438202317037986e-06,
      "loss": 1.3847,
      "step": 36750
    },
    {
      "epoch": 9.116498232996467,
      "grad_norm": 1.082115888595581,
      "learning_rate": 3.822454442169576e-06,
      "loss": 1.3243,
      "step": 36760
    },
    {
      "epoch": 9.118978237956476,
      "grad_norm": 1.1298651695251465,
      "learning_rate": 3.801147041814013e-06,
      "loss": 1.3943,
      "step": 36770
    },
    {
      "epoch": 9.121458242916486,
      "grad_norm": 1.0635226964950562,
      "learning_rate": 3.779898043572794e-06,
      "loss": 1.3355,
      "step": 36780
    },
    {
      "epoch": 9.123938247876495,
      "grad_norm": 1.1357967853546143,
      "learning_rate": 3.7587074603461227e-06,
      "loss": 1.4085,
      "step": 36790
    },
    {
      "epoch": 9.126418252836507,
      "grad_norm": 1.184017539024353,
      "learning_rate": 3.7375753049987973e-06,
      "loss": 1.4233,
      "step": 36800
    },
    {
      "epoch": 9.128898257796516,
      "grad_norm": 1.040810227394104,
      "learning_rate": 3.7165015903600555e-06,
      "loss": 1.4008,
      "step": 36810
    },
    {
      "epoch": 9.131378262756526,
      "grad_norm": 1.0753012895584106,
      "learning_rate": 3.6954863292237297e-06,
      "loss": 1.4354,
      "step": 36820
    },
    {
      "epoch": 9.133858267716535,
      "grad_norm": 0.9710498452186584,
      "learning_rate": 3.674529534348137e-06,
      "loss": 1.3908,
      "step": 36830
    },
    {
      "epoch": 9.136338272676545,
      "grad_norm": 1.0646116733551025,
      "learning_rate": 3.6536312184560993e-06,
      "loss": 1.4451,
      "step": 36840
    },
    {
      "epoch": 9.138818277636556,
      "grad_norm": 1.073171615600586,
      "learning_rate": 3.6327913942349357e-06,
      "loss": 1.3632,
      "step": 36850
    },
    {
      "epoch": 9.141298282596566,
      "grad_norm": 0.9970207214355469,
      "learning_rate": 3.6120100743364692e-06,
      "loss": 1.3601,
      "step": 36860
    },
    {
      "epoch": 9.143778287556575,
      "grad_norm": 1.032759666442871,
      "learning_rate": 3.5912872713769753e-06,
      "loss": 1.4208,
      "step": 36870
    },
    {
      "epoch": 9.146258292516585,
      "grad_norm": 1.0733903646469116,
      "learning_rate": 3.570622997937234e-06,
      "loss": 1.407,
      "step": 36880
    },
    {
      "epoch": 9.148738297476594,
      "grad_norm": 1.0444880723953247,
      "learning_rate": 3.550017266562489e-06,
      "loss": 1.4379,
      "step": 36890
    },
    {
      "epoch": 9.151218302436606,
      "grad_norm": 1.0926703214645386,
      "learning_rate": 3.529470089762421e-06,
      "loss": 1.4767,
      "step": 36900
    },
    {
      "epoch": 9.153698307396615,
      "grad_norm": 1.046228289604187,
      "learning_rate": 3.5089814800111643e-06,
      "loss": 1.4199,
      "step": 36910
    },
    {
      "epoch": 9.156178312356625,
      "grad_norm": 1.09345543384552,
      "learning_rate": 3.488551449747357e-06,
      "loss": 1.4341,
      "step": 36920
    },
    {
      "epoch": 9.158658317316634,
      "grad_norm": 0.9652727842330933,
      "learning_rate": 3.4681800113739783e-06,
      "loss": 1.3601,
      "step": 36930
    },
    {
      "epoch": 9.161138322276644,
      "grad_norm": 1.0334725379943848,
      "learning_rate": 3.4478671772585012e-06,
      "loss": 1.3328,
      "step": 36940
    },
    {
      "epoch": 9.163618327236655,
      "grad_norm": 1.0892077684402466,
      "learning_rate": 3.4276129597328064e-06,
      "loss": 1.4341,
      "step": 36950
    },
    {
      "epoch": 9.166098332196665,
      "grad_norm": 1.0149070024490356,
      "learning_rate": 3.40741737109318e-06,
      "loss": 1.4402,
      "step": 36960
    },
    {
      "epoch": 9.168578337156674,
      "grad_norm": 1.1309213638305664,
      "learning_rate": 3.3872804236003143e-06,
      "loss": 1.4286,
      "step": 36970
    },
    {
      "epoch": 9.171058342116684,
      "grad_norm": 1.0236012935638428,
      "learning_rate": 3.3672021294793297e-06,
      "loss": 1.3888,
      "step": 36980
    },
    {
      "epoch": 9.173538347076693,
      "grad_norm": 1.181558609008789,
      "learning_rate": 3.3471825009196766e-06,
      "loss": 1.3897,
      "step": 36990
    },
    {
      "epoch": 9.176018352036705,
      "grad_norm": 1.0703715085983276,
      "learning_rate": 3.3272215500752658e-06,
      "loss": 1.4069,
      "step": 37000
    },
    {
      "epoch": 9.178498356996714,
      "grad_norm": 1.0862619876861572,
      "learning_rate": 3.307319289064337e-06,
      "loss": 1.4278,
      "step": 37010
    },
    {
      "epoch": 9.180978361956724,
      "grad_norm": 1.0121700763702393,
      "learning_rate": 3.2874757299694913e-06,
      "loss": 1.3398,
      "step": 37020
    },
    {
      "epoch": 9.183458366916733,
      "grad_norm": 1.0737900733947754,
      "learning_rate": 3.267690884837726e-06,
      "loss": 1.3361,
      "step": 37030
    },
    {
      "epoch": 9.185938371876743,
      "grad_norm": 1.0572049617767334,
      "learning_rate": 3.247964765680389e-06,
      "loss": 1.3766,
      "step": 37040
    },
    {
      "epoch": 9.188418376836754,
      "grad_norm": 1.0468642711639404,
      "learning_rate": 3.2282973844731224e-06,
      "loss": 1.3749,
      "step": 37050
    },
    {
      "epoch": 9.190898381796764,
      "grad_norm": 1.0476771593093872,
      "learning_rate": 3.2086887531559973e-06,
      "loss": 1.3749,
      "step": 37060
    },
    {
      "epoch": 9.193378386756773,
      "grad_norm": 1.0572019815444946,
      "learning_rate": 3.189138883633347e-06,
      "loss": 1.3872,
      "step": 37070
    },
    {
      "epoch": 9.195858391716783,
      "grad_norm": 1.0561579465866089,
      "learning_rate": 3.169647787773866e-06,
      "loss": 1.3807,
      "step": 37080
    },
    {
      "epoch": 9.198338396676794,
      "grad_norm": 1.0616618394851685,
      "learning_rate": 3.150215477410534e-06,
      "loss": 1.4009,
      "step": 37090
    },
    {
      "epoch": 9.200818401636804,
      "grad_norm": 1.012891173362732,
      "learning_rate": 3.130841964340692e-06,
      "loss": 1.353,
      "step": 37100
    },
    {
      "epoch": 9.203298406596813,
      "grad_norm": 1.1254786252975464,
      "learning_rate": 3.1115272603259416e-06,
      "loss": 1.391,
      "step": 37110
    },
    {
      "epoch": 9.205778411556823,
      "grad_norm": 1.0111192464828491,
      "learning_rate": 3.092271377092215e-06,
      "loss": 1.3792,
      "step": 37120
    },
    {
      "epoch": 9.208258416516832,
      "grad_norm": 1.0976078510284424,
      "learning_rate": 3.0730743263296945e-06,
      "loss": 1.3778,
      "step": 37130
    },
    {
      "epoch": 9.210738421476844,
      "grad_norm": 1.0674266815185547,
      "learning_rate": 3.0539361196928682e-06,
      "loss": 1.3974,
      "step": 37140
    },
    {
      "epoch": 9.213218426436853,
      "grad_norm": 1.074644684791565,
      "learning_rate": 3.034856768800509e-06,
      "loss": 1.4434,
      "step": 37150
    },
    {
      "epoch": 9.215698431396863,
      "grad_norm": 1.0942853689193726,
      "learning_rate": 3.0158362852356626e-06,
      "loss": 1.4539,
      "step": 37160
    },
    {
      "epoch": 9.218178436356872,
      "grad_norm": 1.0741316080093384,
      "learning_rate": 2.996874680545603e-06,
      "loss": 1.3604,
      "step": 37170
    },
    {
      "epoch": 9.220658441316882,
      "grad_norm": 1.0775376558303833,
      "learning_rate": 2.977971966241877e-06,
      "loss": 1.4091,
      "step": 37180
    },
    {
      "epoch": 9.223138446276893,
      "grad_norm": 1.051910400390625,
      "learning_rate": 2.9591281538003167e-06,
      "loss": 1.3887,
      "step": 37190
    },
    {
      "epoch": 9.225618451236903,
      "grad_norm": 0.9799169898033142,
      "learning_rate": 2.9403432546609043e-06,
      "loss": 1.3789,
      "step": 37200
    },
    {
      "epoch": 9.228098456196912,
      "grad_norm": 1.0778720378875732,
      "learning_rate": 2.9216172802279507e-06,
      "loss": 1.3752,
      "step": 37210
    },
    {
      "epoch": 9.230578461156922,
      "grad_norm": 1.0666970014572144,
      "learning_rate": 2.9029502418699505e-06,
      "loss": 1.3777,
      "step": 37220
    },
    {
      "epoch": 9.233058466116931,
      "grad_norm": 1.0108171701431274,
      "learning_rate": 2.8843421509196057e-06,
      "loss": 1.3827,
      "step": 37230
    },
    {
      "epoch": 9.235538471076943,
      "grad_norm": 1.0825223922729492,
      "learning_rate": 2.8657930186738567e-06,
      "loss": 1.3398,
      "step": 37240
    },
    {
      "epoch": 9.238018476036952,
      "grad_norm": 1.0724096298217773,
      "learning_rate": 2.8473028563938407e-06,
      "loss": 1.3843,
      "step": 37250
    },
    {
      "epoch": 9.240498480996962,
      "grad_norm": 1.083712100982666,
      "learning_rate": 2.8288716753049005e-06,
      "loss": 1.4467,
      "step": 37260
    },
    {
      "epoch": 9.242978485956971,
      "grad_norm": 1.1053098440170288,
      "learning_rate": 2.810499486596563e-06,
      "loss": 1.4234,
      "step": 37270
    },
    {
      "epoch": 9.245458490916981,
      "grad_norm": 1.0849559307098389,
      "learning_rate": 2.7921863014225503e-06,
      "loss": 1.3446,
      "step": 37280
    },
    {
      "epoch": 9.247938495876992,
      "grad_norm": 1.0642001628875732,
      "learning_rate": 2.773932130900736e-06,
      "loss": 1.3362,
      "step": 37290
    },
    {
      "epoch": 9.250418500837002,
      "grad_norm": 1.1150140762329102,
      "learning_rate": 2.7557369861132e-06,
      "loss": 1.4089,
      "step": 37300
    },
    {
      "epoch": 9.252898505797011,
      "grad_norm": 1.0477197170257568,
      "learning_rate": 2.7376008781061835e-06,
      "loss": 1.4083,
      "step": 37310
    },
    {
      "epoch": 9.255378510757021,
      "grad_norm": 1.0632885694503784,
      "learning_rate": 2.7195238178900684e-06,
      "loss": 1.396,
      "step": 37320
    },
    {
      "epoch": 9.257858515717032,
      "grad_norm": 1.0281466245651245,
      "learning_rate": 2.701505816439409e-06,
      "loss": 1.4331,
      "step": 37330
    },
    {
      "epoch": 9.260338520677042,
      "grad_norm": 0.9859536290168762,
      "learning_rate": 2.683546884692878e-06,
      "loss": 1.3713,
      "step": 37340
    },
    {
      "epoch": 9.262818525637051,
      "grad_norm": 1.028248906135559,
      "learning_rate": 2.6656470335533092e-06,
      "loss": 1.4008,
      "step": 37350
    },
    {
      "epoch": 9.265298530597061,
      "grad_norm": 1.0284212827682495,
      "learning_rate": 2.647806273887665e-06,
      "loss": 1.3948,
      "step": 37360
    },
    {
      "epoch": 9.26777853555707,
      "grad_norm": 1.0324586629867554,
      "learning_rate": 2.630024616527038e-06,
      "loss": 1.3841,
      "step": 37370
    },
    {
      "epoch": 9.270258540517082,
      "grad_norm": 1.093057632446289,
      "learning_rate": 2.612302072266637e-06,
      "loss": 1.4453,
      "step": 37380
    },
    {
      "epoch": 9.272738545477091,
      "grad_norm": 1.0433088541030884,
      "learning_rate": 2.594638651865766e-06,
      "loss": 1.4044,
      "step": 37390
    },
    {
      "epoch": 9.275218550437101,
      "grad_norm": 1.1759510040283203,
      "learning_rate": 2.577034366047848e-06,
      "loss": 1.5191,
      "step": 37400
    },
    {
      "epoch": 9.27769855539711,
      "grad_norm": 0.9739142060279846,
      "learning_rate": 2.5594892255004334e-06,
      "loss": 1.378,
      "step": 37410
    },
    {
      "epoch": 9.28017856035712,
      "grad_norm": 1.0965644121170044,
      "learning_rate": 2.5420032408751014e-06,
      "loss": 1.4267,
      "step": 37420
    },
    {
      "epoch": 9.282658565317131,
      "grad_norm": 1.0043507814407349,
      "learning_rate": 2.5245764227875722e-06,
      "loss": 1.4089,
      "step": 37430
    },
    {
      "epoch": 9.285138570277141,
      "grad_norm": 1.0738314390182495,
      "learning_rate": 2.5072087818176382e-06,
      "loss": 1.3901,
      "step": 37440
    },
    {
      "epoch": 9.28761857523715,
      "grad_norm": 1.0553945302963257,
      "learning_rate": 2.4899003285091536e-06,
      "loss": 1.4073,
      "step": 37450
    },
    {
      "epoch": 9.29009858019716,
      "grad_norm": 1.0140724182128906,
      "learning_rate": 2.4726510733700246e-06,
      "loss": 1.3535,
      "step": 37460
    },
    {
      "epoch": 9.29257858515717,
      "grad_norm": 0.9664640426635742,
      "learning_rate": 2.4554610268722634e-06,
      "loss": 1.4218,
      "step": 37470
    },
    {
      "epoch": 9.295058590117181,
      "grad_norm": 1.074286937713623,
      "learning_rate": 2.438330199451877e-06,
      "loss": 1.354,
      "step": 37480
    },
    {
      "epoch": 9.29753859507719,
      "grad_norm": 1.097893476486206,
      "learning_rate": 2.421258601508991e-06,
      "loss": 1.3618,
      "step": 37490
    },
    {
      "epoch": 9.3000186000372,
      "grad_norm": 0.9826263785362244,
      "learning_rate": 2.404246243407704e-06,
      "loss": 1.341,
      "step": 37500
    },
    {
      "epoch": 9.30249860499721,
      "grad_norm": 1.022720217704773,
      "learning_rate": 2.3872931354761983e-06,
      "loss": 1.3872,
      "step": 37510
    },
    {
      "epoch": 9.30497860995722,
      "grad_norm": 1.042087435722351,
      "learning_rate": 2.3703992880066638e-06,
      "loss": 1.3552,
      "step": 37520
    },
    {
      "epoch": 9.30745861491723,
      "grad_norm": 1.06119966506958,
      "learning_rate": 2.3535647112553294e-06,
      "loss": 1.3845,
      "step": 37530
    },
    {
      "epoch": 9.30993861987724,
      "grad_norm": 1.0763623714447021,
      "learning_rate": 2.3367894154424087e-06,
      "loss": 1.3719,
      "step": 37540
    },
    {
      "epoch": 9.31241862483725,
      "grad_norm": 1.0972094535827637,
      "learning_rate": 2.320073410752155e-06,
      "loss": 1.3893,
      "step": 37550
    },
    {
      "epoch": 9.31489862979726,
      "grad_norm": 1.0489115715026855,
      "learning_rate": 2.3034167073328284e-06,
      "loss": 1.3093,
      "step": 37560
    },
    {
      "epoch": 9.317378634757269,
      "grad_norm": 1.042915940284729,
      "learning_rate": 2.2868193152966734e-06,
      "loss": 1.3327,
      "step": 37570
    },
    {
      "epoch": 9.31985863971728,
      "grad_norm": 1.0937442779541016,
      "learning_rate": 2.2702812447199185e-06,
      "loss": 1.4009,
      "step": 37580
    },
    {
      "epoch": 9.32233864467729,
      "grad_norm": 1.0158263444900513,
      "learning_rate": 2.2538025056428214e-06,
      "loss": 1.4095,
      "step": 37590
    },
    {
      "epoch": 9.3248186496373,
      "grad_norm": 1.0596281290054321,
      "learning_rate": 2.237383108069546e-06,
      "loss": 1.3821,
      "step": 37600
    },
    {
      "epoch": 9.327298654597309,
      "grad_norm": 1.0023540258407593,
      "learning_rate": 2.2210230619683083e-06,
      "loss": 1.4085,
      "step": 37610
    },
    {
      "epoch": 9.32977865955732,
      "grad_norm": 1.0688737630844116,
      "learning_rate": 2.20472237727124e-06,
      "loss": 1.3995,
      "step": 37620
    },
    {
      "epoch": 9.33225866451733,
      "grad_norm": 1.1530110836029053,
      "learning_rate": 2.188481063874448e-06,
      "loss": 1.4486,
      "step": 37630
    },
    {
      "epoch": 9.334738669477339,
      "grad_norm": 1.109735131263733,
      "learning_rate": 2.1722991316380003e-06,
      "loss": 1.4128,
      "step": 37640
    },
    {
      "epoch": 9.337218674437349,
      "grad_norm": 1.0421795845031738,
      "learning_rate": 2.1561765903859275e-06,
      "loss": 1.475,
      "step": 37650
    },
    {
      "epoch": 9.339698679397358,
      "grad_norm": 1.0335205793380737,
      "learning_rate": 2.140113449906167e-06,
      "loss": 1.3753,
      "step": 37660
    },
    {
      "epoch": 9.34217868435737,
      "grad_norm": 0.9735974669456482,
      "learning_rate": 2.12410971995064e-06,
      "loss": 1.3865,
      "step": 37670
    },
    {
      "epoch": 9.344658689317379,
      "grad_norm": 1.105315089225769,
      "learning_rate": 2.1081654102351635e-06,
      "loss": 1.4218,
      "step": 37680
    },
    {
      "epoch": 9.347138694277389,
      "grad_norm": 1.0441200733184814,
      "learning_rate": 2.0922805304394832e-06,
      "loss": 1.3724,
      "step": 37690
    },
    {
      "epoch": 9.349618699237398,
      "grad_norm": 1.056541919708252,
      "learning_rate": 2.0764550902072854e-06,
      "loss": 1.3433,
      "step": 37700
    },
    {
      "epoch": 9.352098704197408,
      "grad_norm": 1.038950800895691,
      "learning_rate": 2.0606890991461737e-06,
      "loss": 1.3924,
      "step": 37710
    },
    {
      "epoch": 9.354578709157419,
      "grad_norm": 0.9514971375465393,
      "learning_rate": 2.0449825668276244e-06,
      "loss": 1.4014,
      "step": 37720
    },
    {
      "epoch": 9.357058714117429,
      "grad_norm": 1.1039893627166748,
      "learning_rate": 2.0293355027870554e-06,
      "loss": 1.4024,
      "step": 37730
    },
    {
      "epoch": 9.359538719077438,
      "grad_norm": 1.0915133953094482,
      "learning_rate": 2.013747916523756e-06,
      "loss": 1.4493,
      "step": 37740
    },
    {
      "epoch": 9.362018724037448,
      "grad_norm": 1.0225774049758911,
      "learning_rate": 1.998219817500924e-06,
      "loss": 1.4054,
      "step": 37750
    },
    {
      "epoch": 9.364498728997457,
      "grad_norm": 1.0327415466308594,
      "learning_rate": 1.9827512151456173e-06,
      "loss": 1.3863,
      "step": 37760
    },
    {
      "epoch": 9.366978733957469,
      "grad_norm": 1.0993423461914062,
      "learning_rate": 1.9673421188488137e-06,
      "loss": 1.407,
      "step": 37770
    },
    {
      "epoch": 9.369458738917478,
      "grad_norm": 1.0667732954025269,
      "learning_rate": 1.951992537965319e-06,
      "loss": 1.3781,
      "step": 37780
    },
    {
      "epoch": 9.371938743877488,
      "grad_norm": 1.065179467201233,
      "learning_rate": 1.936702481813857e-06,
      "loss": 1.3953,
      "step": 37790
    },
    {
      "epoch": 9.374418748837497,
      "grad_norm": 1.0528008937835693,
      "learning_rate": 1.921471959676957e-06,
      "loss": 1.3768,
      "step": 37800
    },
    {
      "epoch": 9.376898753797507,
      "grad_norm": 1.0844627618789673,
      "learning_rate": 1.9063009808010569e-06,
      "loss": 1.4497,
      "step": 37810
    },
    {
      "epoch": 9.379378758757518,
      "grad_norm": 1.1173009872436523,
      "learning_rate": 1.891189554396422e-06,
      "loss": 1.3281,
      "step": 37820
    },
    {
      "epoch": 9.381858763717528,
      "grad_norm": 1.0677911043167114,
      "learning_rate": 1.8761376896371697e-06,
      "loss": 1.4197,
      "step": 37830
    },
    {
      "epoch": 9.384338768677537,
      "grad_norm": 1.0368330478668213,
      "learning_rate": 1.8611453956612347e-06,
      "loss": 1.4121,
      "step": 37840
    },
    {
      "epoch": 9.386818773637547,
      "grad_norm": 1.025154948234558,
      "learning_rate": 1.846212681570425e-06,
      "loss": 1.3365,
      "step": 37850
    },
    {
      "epoch": 9.389298778597556,
      "grad_norm": 1.0957260131835938,
      "learning_rate": 1.831339556430356e-06,
      "loss": 1.4194,
      "step": 37860
    },
    {
      "epoch": 9.391778783557568,
      "grad_norm": 1.013924479484558,
      "learning_rate": 1.8165260292704711e-06,
      "loss": 1.3767,
      "step": 37870
    },
    {
      "epoch": 9.394258788517577,
      "grad_norm": 0.9932572841644287,
      "learning_rate": 1.8017721090840322e-06,
      "loss": 1.3851,
      "step": 37880
    },
    {
      "epoch": 9.396738793477587,
      "grad_norm": 1.075989842414856,
      "learning_rate": 1.7870778048280966e-06,
      "loss": 1.4236,
      "step": 37890
    },
    {
      "epoch": 9.399218798437596,
      "grad_norm": 1.0282961130142212,
      "learning_rate": 1.7724431254235618e-06,
      "loss": 1.3502,
      "step": 37900
    },
    {
      "epoch": 9.401698803397608,
      "grad_norm": 1.0960266590118408,
      "learning_rate": 1.7578680797550872e-06,
      "loss": 1.3932,
      "step": 37910
    },
    {
      "epoch": 9.404178808357617,
      "grad_norm": 1.0175817012786865,
      "learning_rate": 1.7433526766711728e-06,
      "loss": 1.3902,
      "step": 37920
    },
    {
      "epoch": 9.406658813317627,
      "grad_norm": 1.090591311454773,
      "learning_rate": 1.7288969249840915e-06,
      "loss": 1.4048,
      "step": 37930
    },
    {
      "epoch": 9.409138818277636,
      "grad_norm": 1.071060299873352,
      "learning_rate": 1.7145008334698898e-06,
      "loss": 1.3387,
      "step": 37940
    },
    {
      "epoch": 9.411618823237646,
      "grad_norm": 1.0652714967727661,
      "learning_rate": 1.7001644108683878e-06,
      "loss": 1.2656,
      "step": 37950
    },
    {
      "epoch": 9.414098828197657,
      "grad_norm": 1.066692590713501,
      "learning_rate": 1.6858876658832233e-06,
      "loss": 1.4334,
      "step": 37960
    },
    {
      "epoch": 9.416578833157667,
      "grad_norm": 1.169921875,
      "learning_rate": 1.6716706071817633e-06,
      "loss": 1.3824,
      "step": 37970
    },
    {
      "epoch": 9.419058838117676,
      "grad_norm": 0.9965091943740845,
      "learning_rate": 1.657513243395159e-06,
      "loss": 1.4842,
      "step": 37980
    },
    {
      "epoch": 9.421538843077686,
      "grad_norm": 1.088774561882019,
      "learning_rate": 1.6434155831183018e-06,
      "loss": 1.4068,
      "step": 37990
    },
    {
      "epoch": 9.424018848037695,
      "grad_norm": 1.0832850933074951,
      "learning_rate": 1.6293776349098677e-06,
      "loss": 1.4314,
      "step": 38000
    },
    {
      "epoch": 9.426498852997707,
      "grad_norm": 1.0543954372406006,
      "learning_rate": 1.6153994072922506e-06,
      "loss": 1.4288,
      "step": 38010
    },
    {
      "epoch": 9.428978857957716,
      "grad_norm": 0.9991889595985413,
      "learning_rate": 1.6014809087516291e-06,
      "loss": 1.4025,
      "step": 38020
    },
    {
      "epoch": 9.431458862917726,
      "grad_norm": 1.0605100393295288,
      "learning_rate": 1.587622147737855e-06,
      "loss": 1.3978,
      "step": 38030
    },
    {
      "epoch": 9.433938867877735,
      "grad_norm": 1.0560320615768433,
      "learning_rate": 1.5738231326645758e-06,
      "loss": 1.4002,
      "step": 38040
    },
    {
      "epoch": 9.436418872837745,
      "grad_norm": 0.9635500311851501,
      "learning_rate": 1.5600838719091348e-06,
      "loss": 1.392,
      "step": 38050
    },
    {
      "epoch": 9.438898877797756,
      "grad_norm": 1.1183180809020996,
      "learning_rate": 1.5464043738126045e-06,
      "loss": 1.4105,
      "step": 38060
    },
    {
      "epoch": 9.441378882757766,
      "grad_norm": 1.0084953308105469,
      "learning_rate": 1.5327846466797856e-06,
      "loss": 1.3811,
      "step": 38070
    },
    {
      "epoch": 9.443858887717775,
      "grad_norm": 1.0079189538955688,
      "learning_rate": 1.5192246987791981e-06,
      "loss": 1.384,
      "step": 38080
    },
    {
      "epoch": 9.446338892677785,
      "grad_norm": 1.1129961013793945,
      "learning_rate": 1.505724538343023e-06,
      "loss": 1.37,
      "step": 38090
    },
    {
      "epoch": 9.448818897637794,
      "grad_norm": 1.0100232362747192,
      "learning_rate": 1.4922841735672156e-06,
      "loss": 1.3868,
      "step": 38100
    },
    {
      "epoch": 9.451298902597806,
      "grad_norm": 1.0200841426849365,
      "learning_rate": 1.4789036126113597e-06,
      "loss": 1.3904,
      "step": 38110
    },
    {
      "epoch": 9.453778907557815,
      "grad_norm": 1.0760002136230469,
      "learning_rate": 1.465582863598791e-06,
      "loss": 1.35,
      "step": 38120
    },
    {
      "epoch": 9.456258912517825,
      "grad_norm": 0.9961051344871521,
      "learning_rate": 1.4523219346164852e-06,
      "loss": 1.3225,
      "step": 38130
    },
    {
      "epoch": 9.458738917477834,
      "grad_norm": 1.0708684921264648,
      "learning_rate": 1.4391208337151463e-06,
      "loss": 1.3798,
      "step": 38140
    },
    {
      "epoch": 9.461218922437844,
      "grad_norm": 1.0574396848678589,
      "learning_rate": 1.4259795689090972e-06,
      "loss": 1.417,
      "step": 38150
    },
    {
      "epoch": 9.463698927397855,
      "grad_norm": 1.1168031692504883,
      "learning_rate": 1.4128981481764115e-06,
      "loss": 1.4336,
      "step": 38160
    },
    {
      "epoch": 9.466178932357865,
      "grad_norm": 1.0774537324905396,
      "learning_rate": 1.399876579458781e-06,
      "loss": 1.4477,
      "step": 38170
    },
    {
      "epoch": 9.468658937317874,
      "grad_norm": 1.075043797492981,
      "learning_rate": 1.3869148706615598e-06,
      "loss": 1.4456,
      "step": 38180
    },
    {
      "epoch": 9.471138942277884,
      "grad_norm": 1.0636255741119385,
      "learning_rate": 1.3740130296537646e-06,
      "loss": 1.3887,
      "step": 38190
    },
    {
      "epoch": 9.473618947237895,
      "grad_norm": 1.0367157459259033,
      "learning_rate": 1.3611710642681076e-06,
      "loss": 1.4313,
      "step": 38200
    },
    {
      "epoch": 9.476098952197905,
      "grad_norm": 1.1512954235076904,
      "learning_rate": 1.3483889823008965e-06,
      "loss": 1.3543,
      "step": 38210
    },
    {
      "epoch": 9.478578957157914,
      "grad_norm": 1.047049880027771,
      "learning_rate": 1.3356667915121025e-06,
      "loss": 1.3982,
      "step": 38220
    },
    {
      "epoch": 9.481058962117924,
      "grad_norm": 1.0412118434906006,
      "learning_rate": 1.3230044996253466e-06,
      "loss": 1.4216,
      "step": 38230
    },
    {
      "epoch": 9.483538967077934,
      "grad_norm": 1.0545369386672974,
      "learning_rate": 1.3104021143278911e-06,
      "loss": 1.3765,
      "step": 38240
    },
    {
      "epoch": 9.486018972037945,
      "grad_norm": 1.1080199480056763,
      "learning_rate": 1.2978596432705825e-06,
      "loss": 1.4318,
      "step": 38250
    },
    {
      "epoch": 9.488498976997954,
      "grad_norm": 1.0659513473510742,
      "learning_rate": 1.2853770940679633e-06,
      "loss": 1.3932,
      "step": 38260
    },
    {
      "epoch": 9.490978981957964,
      "grad_norm": 1.0260642766952515,
      "learning_rate": 1.2729544742981492e-06,
      "loss": 1.4366,
      "step": 38270
    },
    {
      "epoch": 9.493458986917974,
      "grad_norm": 1.0209259986877441,
      "learning_rate": 1.2605917915028742e-06,
      "loss": 1.3627,
      "step": 38280
    },
    {
      "epoch": 9.495938991877983,
      "grad_norm": 1.1583101749420166,
      "learning_rate": 1.2482890531875125e-06,
      "loss": 1.4544,
      "step": 38290
    },
    {
      "epoch": 9.498418996837994,
      "grad_norm": 1.0427088737487793,
      "learning_rate": 1.2360462668210227e-06,
      "loss": 1.3453,
      "step": 38300
    },
    {
      "epoch": 9.500899001798004,
      "grad_norm": 1.0863418579101562,
      "learning_rate": 1.2238634398359706e-06,
      "loss": 1.3817,
      "step": 38310
    },
    {
      "epoch": 9.503379006758014,
      "grad_norm": 1.1137123107910156,
      "learning_rate": 1.2117405796285286e-06,
      "loss": 1.3923,
      "step": 38320
    },
    {
      "epoch": 9.505859011718023,
      "grad_norm": 1.0869648456573486,
      "learning_rate": 1.1996776935584653e-06,
      "loss": 1.3358,
      "step": 38330
    },
    {
      "epoch": 9.508339016678033,
      "grad_norm": 1.069647192955017,
      "learning_rate": 1.1876747889491223e-06,
      "loss": 1.4135,
      "step": 38340
    },
    {
      "epoch": 9.510819021638044,
      "grad_norm": 1.0883690118789673,
      "learning_rate": 1.1757318730874379e-06,
      "loss": 1.3843,
      "step": 38350
    },
    {
      "epoch": 9.513299026598053,
      "grad_norm": 1.023659586906433,
      "learning_rate": 1.163848953223934e-06,
      "loss": 1.4235,
      "step": 38360
    },
    {
      "epoch": 9.515779031558063,
      "grad_norm": 1.1154769659042358,
      "learning_rate": 1.1520260365726954e-06,
      "loss": 1.4074,
      "step": 38370
    },
    {
      "epoch": 9.518259036518073,
      "grad_norm": 1.0261458158493042,
      "learning_rate": 1.1402631303114032e-06,
      "loss": 1.4057,
      "step": 38380
    },
    {
      "epoch": 9.520739041478082,
      "grad_norm": 1.0308420658111572,
      "learning_rate": 1.128560241581289e-06,
      "loss": 1.4169,
      "step": 38390
    },
    {
      "epoch": 9.523219046438093,
      "grad_norm": 1.0664455890655518,
      "learning_rate": 1.1169173774871478e-06,
      "loss": 1.4453,
      "step": 38400
    },
    {
      "epoch": 9.525699051398103,
      "grad_norm": 1.019059419631958,
      "learning_rate": 1.1053345450973362e-06,
      "loss": 1.4105,
      "step": 38410
    },
    {
      "epoch": 9.528179056358113,
      "grad_norm": 1.0536880493164062,
      "learning_rate": 1.093811751443774e-06,
      "loss": 1.4134,
      "step": 38420
    },
    {
      "epoch": 9.530659061318122,
      "grad_norm": 1.056960940361023,
      "learning_rate": 1.0823490035218987e-06,
      "loss": 1.4193,
      "step": 38430
    },
    {
      "epoch": 9.533139066278132,
      "grad_norm": 1.0073970556259155,
      "learning_rate": 1.0709463082907545e-06,
      "loss": 1.4197,
      "step": 38440
    },
    {
      "epoch": 9.535619071238143,
      "grad_norm": 1.1760201454162598,
      "learning_rate": 1.0596036726728819e-06,
      "loss": 1.378,
      "step": 38450
    },
    {
      "epoch": 9.538099076198153,
      "grad_norm": 1.0135118961334229,
      "learning_rate": 1.0483211035543504e-06,
      "loss": 1.4017,
      "step": 38460
    },
    {
      "epoch": 9.540579081158162,
      "grad_norm": 1.0354691743850708,
      "learning_rate": 1.0370986077848145e-06,
      "loss": 1.4306,
      "step": 38470
    },
    {
      "epoch": 9.543059086118172,
      "grad_norm": 1.0982271432876587,
      "learning_rate": 1.0259361921774013e-06,
      "loss": 1.4364,
      "step": 38480
    },
    {
      "epoch": 9.545539091078183,
      "grad_norm": 1.0671141147613525,
      "learning_rate": 1.0148338635088018e-06,
      "loss": 1.3727,
      "step": 38490
    },
    {
      "epoch": 9.548019096038193,
      "grad_norm": 1.1574429273605347,
      "learning_rate": 1.003791628519213e-06,
      "loss": 1.4063,
      "step": 38500
    },
    {
      "epoch": 9.550499100998202,
      "grad_norm": 1.0300366878509521,
      "learning_rate": 9.9280949391235e-07,
      "loss": 1.4205,
      "step": 38510
    },
    {
      "epoch": 9.552979105958212,
      "grad_norm": 1.1044412851333618,
      "learning_rate": 9.818874663554357e-07,
      "loss": 1.401,
      "step": 38520
    },
    {
      "epoch": 9.555459110918221,
      "grad_norm": 1.1097668409347534,
      "learning_rate": 9.710255524792322e-07,
      "loss": 1.4522,
      "step": 38530
    },
    {
      "epoch": 9.557939115878233,
      "grad_norm": 0.9702837467193604,
      "learning_rate": 9.60223758877965e-07,
      "loss": 1.3913,
      "step": 38540
    },
    {
      "epoch": 9.560419120838242,
      "grad_norm": 1.0994696617126465,
      "learning_rate": 9.494820921093772e-07,
      "loss": 1.4454,
      "step": 38550
    },
    {
      "epoch": 9.562899125798252,
      "grad_norm": 1.0229690074920654,
      "learning_rate": 9.388005586947191e-07,
      "loss": 1.3578,
      "step": 38560
    },
    {
      "epoch": 9.565379130758261,
      "grad_norm": 1.0709983110427856,
      "learning_rate": 9.281791651187366e-07,
      "loss": 1.4071,
      "step": 38570
    },
    {
      "epoch": 9.56785913571827,
      "grad_norm": 1.0604861974716187,
      "learning_rate": 9.176179178296385e-07,
      "loss": 1.4025,
      "step": 38580
    },
    {
      "epoch": 9.570339140678282,
      "grad_norm": 0.9834784865379333,
      "learning_rate": 9.071168232391292e-07,
      "loss": 1.3858,
      "step": 38590
    },
    {
      "epoch": 9.572819145638292,
      "grad_norm": 1.063391089439392,
      "learning_rate": 8.966758877224201e-07,
      "loss": 1.4405,
      "step": 38600
    },
    {
      "epoch": 9.575299150598301,
      "grad_norm": 1.0823516845703125,
      "learning_rate": 8.862951176181744e-07,
      "loss": 1.3624,
      "step": 38610
    },
    {
      "epoch": 9.57777915555831,
      "grad_norm": 0.991945743560791,
      "learning_rate": 8.759745192285285e-07,
      "loss": 1.3768,
      "step": 38620
    },
    {
      "epoch": 9.58025916051832,
      "grad_norm": 1.09054696559906,
      "learning_rate": 8.657140988191037e-07,
      "loss": 1.4156,
      "step": 38630
    },
    {
      "epoch": 9.582739165478332,
      "grad_norm": 0.9679309129714966,
      "learning_rate": 8.555138626189618e-07,
      "loss": 1.329,
      "step": 38640
    },
    {
      "epoch": 9.585219170438341,
      "grad_norm": 1.0478179454803467,
      "learning_rate": 8.453738168206604e-07,
      "loss": 1.4078,
      "step": 38650
    },
    {
      "epoch": 9.58769917539835,
      "grad_norm": 1.062233567237854,
      "learning_rate": 8.352939675801975e-07,
      "loss": 1.4452,
      "step": 38660
    },
    {
      "epoch": 9.59017918035836,
      "grad_norm": 1.1104792356491089,
      "learning_rate": 8.252743210170222e-07,
      "loss": 1.391,
      "step": 38670
    },
    {
      "epoch": 9.59265918531837,
      "grad_norm": 1.1112204790115356,
      "learning_rate": 8.153148832140466e-07,
      "loss": 1.349,
      "step": 38680
    },
    {
      "epoch": 9.595139190278381,
      "grad_norm": 1.0226380825042725,
      "learning_rate": 8.054156602176233e-07,
      "loss": 1.4159,
      "step": 38690
    },
    {
      "epoch": 9.59761919523839,
      "grad_norm": 1.1281567811965942,
      "learning_rate": 7.955766580375335e-07,
      "loss": 1.407,
      "step": 38700
    },
    {
      "epoch": 9.6000992001984,
      "grad_norm": 1.0717427730560303,
      "learning_rate": 7.857978826470325e-07,
      "loss": 1.4154,
      "step": 38710
    },
    {
      "epoch": 9.60257920515841,
      "grad_norm": 1.0515215396881104,
      "learning_rate": 7.760793399827937e-07,
      "loss": 1.4024,
      "step": 38720
    },
    {
      "epoch": 9.60505921011842,
      "grad_norm": 1.0398309230804443,
      "learning_rate": 7.664210359449086e-07,
      "loss": 1.3898,
      "step": 38730
    },
    {
      "epoch": 9.60753921507843,
      "grad_norm": 1.1013152599334717,
      "learning_rate": 7.568229763969204e-07,
      "loss": 1.4075,
      "step": 38740
    },
    {
      "epoch": 9.61001922003844,
      "grad_norm": 1.0536187887191772,
      "learning_rate": 7.47285167165801e-07,
      "loss": 1.3514,
      "step": 38750
    },
    {
      "epoch": 9.61249922499845,
      "grad_norm": 1.0260318517684937,
      "learning_rate": 7.378076140419187e-07,
      "loss": 1.3927,
      "step": 38760
    },
    {
      "epoch": 9.61497922995846,
      "grad_norm": 1.0289767980575562,
      "learning_rate": 7.283903227790823e-07,
      "loss": 1.4149,
      "step": 38770
    },
    {
      "epoch": 9.61745923491847,
      "grad_norm": 1.11062753200531,
      "learning_rate": 7.19033299094496e-07,
      "loss": 1.3977,
      "step": 38780
    },
    {
      "epoch": 9.61993923987848,
      "grad_norm": 1.0465054512023926,
      "learning_rate": 7.097365486688157e-07,
      "loss": 1.4034,
      "step": 38790
    },
    {
      "epoch": 9.62241924483849,
      "grad_norm": 1.0386594533920288,
      "learning_rate": 7.00500077146038e-07,
      "loss": 1.3796,
      "step": 38800
    },
    {
      "epoch": 9.6248992497985,
      "grad_norm": 0.9937693476676941,
      "learning_rate": 6.913238901336549e-07,
      "loss": 1.3505,
      "step": 38810
    },
    {
      "epoch": 9.627379254758509,
      "grad_norm": 1.0787221193313599,
      "learning_rate": 6.822079932024661e-07,
      "loss": 1.4293,
      "step": 38820
    },
    {
      "epoch": 9.62985925971852,
      "grad_norm": 1.031509280204773,
      "learning_rate": 6.731523918867333e-07,
      "loss": 1.4084,
      "step": 38830
    },
    {
      "epoch": 9.63233926467853,
      "grad_norm": 1.0514459609985352,
      "learning_rate": 6.641570916840922e-07,
      "loss": 1.3825,
      "step": 38840
    },
    {
      "epoch": 9.63481926963854,
      "grad_norm": 1.0696451663970947,
      "learning_rate": 6.552220980555635e-07,
      "loss": 1.3867,
      "step": 38850
    },
    {
      "epoch": 9.637299274598549,
      "grad_norm": 1.0945204496383667,
      "learning_rate": 6.463474164255634e-07,
      "loss": 1.3632,
      "step": 38860
    },
    {
      "epoch": 9.639779279558558,
      "grad_norm": 1.0794506072998047,
      "learning_rate": 6.375330521818934e-07,
      "loss": 1.3657,
      "step": 38870
    },
    {
      "epoch": 9.64225928451857,
      "grad_norm": 1.0652384757995605,
      "learning_rate": 6.287790106757396e-07,
      "loss": 1.3519,
      "step": 38880
    },
    {
      "epoch": 9.64473928947858,
      "grad_norm": 0.9793534874916077,
      "learning_rate": 6.20085297221662e-07,
      "loss": 1.3986,
      "step": 38890
    },
    {
      "epoch": 9.647219294438589,
      "grad_norm": 1.1023114919662476,
      "learning_rate": 6.11451917097583e-07,
      "loss": 1.4228,
      "step": 38900
    },
    {
      "epoch": 9.649699299398598,
      "grad_norm": 1.117197871208191,
      "learning_rate": 6.028788755448211e-07,
      "loss": 1.4341,
      "step": 38910
    },
    {
      "epoch": 9.652179304358608,
      "grad_norm": 1.0328142642974854,
      "learning_rate": 5.943661777680354e-07,
      "loss": 1.4006,
      "step": 38920
    },
    {
      "epoch": 9.65465930931862,
      "grad_norm": 1.154806137084961,
      "learning_rate": 5.859138289352917e-07,
      "loss": 1.3936,
      "step": 38930
    },
    {
      "epoch": 9.657139314278629,
      "grad_norm": 1.0782853364944458,
      "learning_rate": 5.775218341779742e-07,
      "loss": 1.3953,
      "step": 38940
    },
    {
      "epoch": 9.659619319238638,
      "grad_norm": 0.9896028637886047,
      "learning_rate": 5.69190198590841e-07,
      "loss": 1.44,
      "step": 38950
    },
    {
      "epoch": 9.662099324198648,
      "grad_norm": 1.1631059646606445,
      "learning_rate": 5.609189272320237e-07,
      "loss": 1.3734,
      "step": 38960
    },
    {
      "epoch": 9.66457932915866,
      "grad_norm": 1.133226752281189,
      "learning_rate": 5.527080251229833e-07,
      "loss": 1.3878,
      "step": 38970
    },
    {
      "epoch": 9.667059334118669,
      "grad_norm": 1.1199979782104492,
      "learning_rate": 5.445574972485434e-07,
      "loss": 1.4203,
      "step": 38980
    },
    {
      "epoch": 9.669539339078678,
      "grad_norm": 1.095866084098816,
      "learning_rate": 5.364673485568794e-07,
      "loss": 1.3982,
      "step": 38990
    },
    {
      "epoch": 9.672019344038688,
      "grad_norm": 0.9899543523788452,
      "learning_rate": 5.284375839594957e-07,
      "loss": 1.3403,
      "step": 39000
    },
    {
      "epoch": 9.674499348998697,
      "grad_norm": 1.0644506216049194,
      "learning_rate": 5.204682083312484e-07,
      "loss": 1.4253,
      "step": 39010
    },
    {
      "epoch": 9.676979353958707,
      "grad_norm": 1.0117524862289429,
      "learning_rate": 5.125592265103118e-07,
      "loss": 1.4065,
      "step": 39020
    },
    {
      "epoch": 9.679459358918718,
      "grad_norm": 1.099260687828064,
      "learning_rate": 5.04710643298234e-07,
      "loss": 1.3436,
      "step": 39030
    },
    {
      "epoch": 9.681939363878728,
      "grad_norm": 1.042979121208191,
      "learning_rate": 4.969224634598591e-07,
      "loss": 1.4066,
      "step": 39040
    },
    {
      "epoch": 9.684419368838737,
      "grad_norm": 1.043346881866455,
      "learning_rate": 4.891946917233825e-07,
      "loss": 1.3617,
      "step": 39050
    },
    {
      "epoch": 9.686899373798747,
      "grad_norm": 1.0590424537658691,
      "learning_rate": 4.815273327803182e-07,
      "loss": 1.4236,
      "step": 39060
    },
    {
      "epoch": 9.689379378758758,
      "grad_norm": 1.0161995887756348,
      "learning_rate": 4.7392039128548725e-07,
      "loss": 1.3855,
      "step": 39070
    },
    {
      "epoch": 9.691859383718768,
      "grad_norm": 1.0723992586135864,
      "learning_rate": 4.6637387185706206e-07,
      "loss": 1.415,
      "step": 39080
    },
    {
      "epoch": 9.694339388678777,
      "grad_norm": 1.0755267143249512,
      "learning_rate": 4.5888777907652227e-07,
      "loss": 1.4233,
      "step": 39090
    },
    {
      "epoch": 9.696819393638787,
      "grad_norm": 1.0378844738006592,
      "learning_rate": 4.514621174886435e-07,
      "loss": 1.4311,
      "step": 39100
    },
    {
      "epoch": 9.699299398598797,
      "grad_norm": 0.9756118059158325,
      "learning_rate": 4.440968916015309e-07,
      "loss": 1.3828,
      "step": 39110
    },
    {
      "epoch": 9.701779403558808,
      "grad_norm": 1.0769411325454712,
      "learning_rate": 4.3679210588661866e-07,
      "loss": 1.439,
      "step": 39120
    },
    {
      "epoch": 9.704259408518817,
      "grad_norm": 1.0276910066604614,
      "learning_rate": 4.2954776477860393e-07,
      "loss": 1.3836,
      "step": 39130
    },
    {
      "epoch": 9.706739413478827,
      "grad_norm": 1.0128233432769775,
      "learning_rate": 4.22363872675513e-07,
      "loss": 1.3781,
      "step": 39140
    },
    {
      "epoch": 9.709219418438837,
      "grad_norm": 1.0176806449890137,
      "learning_rate": 4.152404339386795e-07,
      "loss": 1.4247,
      "step": 39150
    },
    {
      "epoch": 9.711699423398846,
      "grad_norm": 1.0066100358963013,
      "learning_rate": 4.081774528927218e-07,
      "loss": 1.3822,
      "step": 39160
    },
    {
      "epoch": 9.714179428358857,
      "grad_norm": 1.0204678773880005,
      "learning_rate": 4.011749338255544e-07,
      "loss": 1.4417,
      "step": 39170
    },
    {
      "epoch": 9.716659433318867,
      "grad_norm": 1.1039975881576538,
      "learning_rate": 3.942328809884099e-07,
      "loss": 1.3699,
      "step": 39180
    },
    {
      "epoch": 9.719139438278876,
      "grad_norm": 1.1025723218917847,
      "learning_rate": 3.873512985957728e-07,
      "loss": 1.4301,
      "step": 39190
    },
    {
      "epoch": 9.721619443238886,
      "grad_norm": 1.0049769878387451,
      "learning_rate": 3.805301908254455e-07,
      "loss": 1.3407,
      "step": 39200
    },
    {
      "epoch": 9.724099448198896,
      "grad_norm": 1.0349602699279785,
      "learning_rate": 3.737695618185155e-07,
      "loss": 1.3534,
      "step": 39210
    },
    {
      "epoch": 9.726579453158907,
      "grad_norm": 0.9897842407226562,
      "learning_rate": 3.6706941567933304e-07,
      "loss": 1.4331,
      "step": 39220
    },
    {
      "epoch": 9.729059458118916,
      "grad_norm": 1.050890564918518,
      "learning_rate": 3.6042975647554433e-07,
      "loss": 1.3952,
      "step": 39230
    },
    {
      "epoch": 9.731539463078926,
      "grad_norm": 1.050793170928955,
      "learning_rate": 3.5385058823809156e-07,
      "loss": 1.3724,
      "step": 39240
    },
    {
      "epoch": 9.734019468038936,
      "grad_norm": 1.1213107109069824,
      "learning_rate": 3.4733191496113536e-07,
      "loss": 1.3988,
      "step": 39250
    },
    {
      "epoch": 9.736499472998947,
      "grad_norm": 1.0219743251800537,
      "learning_rate": 3.4087374060218777e-07,
      "loss": 1.3956,
      "step": 39260
    },
    {
      "epoch": 9.738979477958956,
      "grad_norm": 1.0561084747314453,
      "learning_rate": 3.3447606908196817e-07,
      "loss": 1.4519,
      "step": 39270
    },
    {
      "epoch": 9.741459482918966,
      "grad_norm": 1.0141974687576294,
      "learning_rate": 3.281389042844918e-07,
      "loss": 1.4234,
      "step": 39280
    },
    {
      "epoch": 9.743939487878976,
      "grad_norm": 1.0682047605514526,
      "learning_rate": 3.2186225005704784e-07,
      "loss": 1.3464,
      "step": 39290
    },
    {
      "epoch": 9.746419492838985,
      "grad_norm": 1.016051173210144,
      "learning_rate": 3.156461102101771e-07,
      "loss": 1.3778,
      "step": 39300
    },
    {
      "epoch": 9.748899497798995,
      "grad_norm": 1.0662356615066528,
      "learning_rate": 3.0949048851767194e-07,
      "loss": 1.4213,
      "step": 39310
    },
    {
      "epoch": 9.751379502759006,
      "grad_norm": 1.0327116250991821,
      "learning_rate": 3.0339538871660965e-07,
      "loss": 1.3685,
      "step": 39320
    },
    {
      "epoch": 9.753859507719016,
      "grad_norm": 1.0809624195098877,
      "learning_rate": 2.9736081450730815e-07,
      "loss": 1.3436,
      "step": 39330
    },
    {
      "epoch": 9.756339512679025,
      "grad_norm": 1.104588508605957,
      "learning_rate": 2.9138676955333676e-07,
      "loss": 1.4181,
      "step": 39340
    },
    {
      "epoch": 9.758819517639035,
      "grad_norm": 1.0799038410186768,
      "learning_rate": 2.8547325748153885e-07,
      "loss": 1.3374,
      "step": 39350
    },
    {
      "epoch": 9.761299522599046,
      "grad_norm": 1.0426992177963257,
      "learning_rate": 2.7962028188198706e-07,
      "loss": 1.3495,
      "step": 39360
    },
    {
      "epoch": 9.763779527559056,
      "grad_norm": 1.0788627862930298,
      "learning_rate": 2.738278463080057e-07,
      "loss": 1.3959,
      "step": 39370
    },
    {
      "epoch": 9.766259532519065,
      "grad_norm": 1.0900304317474365,
      "learning_rate": 2.6809595427618183e-07,
      "loss": 1.4163,
      "step": 39380
    },
    {
      "epoch": 9.768739537479075,
      "grad_norm": 1.0566004514694214,
      "learning_rate": 2.624246092663318e-07,
      "loss": 1.3932,
      "step": 39390
    },
    {
      "epoch": 9.771219542439084,
      "grad_norm": 0.9841712713241577,
      "learning_rate": 2.5681381472151267e-07,
      "loss": 1.4414,
      "step": 39400
    },
    {
      "epoch": 9.773699547399096,
      "grad_norm": 1.0372689962387085,
      "learning_rate": 2.5126357404802183e-07,
      "loss": 1.3482,
      "step": 39410
    },
    {
      "epoch": 9.776179552359105,
      "grad_norm": 1.058873176574707,
      "learning_rate": 2.457738906153972e-07,
      "loss": 1.3571,
      "step": 39420
    },
    {
      "epoch": 9.778659557319115,
      "grad_norm": 1.0845701694488525,
      "learning_rate": 2.4034476775642856e-07,
      "loss": 1.3415,
      "step": 39430
    },
    {
      "epoch": 9.781139562279124,
      "grad_norm": 1.0763607025146484,
      "learning_rate": 2.3497620876711257e-07,
      "loss": 1.3309,
      "step": 39440
    },
    {
      "epoch": 9.783619567239134,
      "grad_norm": 1.013174057006836,
      "learning_rate": 2.2966821690669771e-07,
      "loss": 1.3979,
      "step": 39450
    },
    {
      "epoch": 9.786099572199145,
      "grad_norm": 1.1713688373565674,
      "learning_rate": 2.244207953976396e-07,
      "loss": 1.3639,
      "step": 39460
    },
    {
      "epoch": 9.788579577159155,
      "grad_norm": 1.0831536054611206,
      "learning_rate": 2.1923394742564553e-07,
      "loss": 1.4203,
      "step": 39470
    },
    {
      "epoch": 9.791059582119164,
      "grad_norm": 1.056071162223816,
      "learning_rate": 2.141076761396521e-07,
      "loss": 1.4416,
      "step": 39480
    },
    {
      "epoch": 9.793539587079174,
      "grad_norm": 1.087546706199646,
      "learning_rate": 2.0904198465178105e-07,
      "loss": 1.4253,
      "step": 39490
    },
    {
      "epoch": 9.796019592039183,
      "grad_norm": 1.104124903678894,
      "learning_rate": 2.0403687603742783e-07,
      "loss": 1.3681,
      "step": 39500
    },
    {
      "epoch": 9.798499596999195,
      "grad_norm": 1.0968530178070068,
      "learning_rate": 1.9909235333517295e-07,
      "loss": 1.4218,
      "step": 39510
    },
    {
      "epoch": 9.800979601959204,
      "grad_norm": 1.0319266319274902,
      "learning_rate": 1.9420841954681525e-07,
      "loss": 1.4737,
      "step": 39520
    },
    {
      "epoch": 9.803459606919214,
      "grad_norm": 1.1220463514328003,
      "learning_rate": 1.89385077637394e-07,
      "loss": 1.3877,
      "step": 39530
    },
    {
      "epoch": 9.805939611879223,
      "grad_norm": 1.0038042068481445,
      "learning_rate": 1.8462233053514467e-07,
      "loss": 1.3989,
      "step": 39540
    },
    {
      "epoch": 9.808419616839235,
      "grad_norm": 1.0417869091033936,
      "learning_rate": 1.7992018113152098e-07,
      "loss": 1.415,
      "step": 39550
    },
    {
      "epoch": 9.810899621799244,
      "grad_norm": 1.080538272857666,
      "learning_rate": 1.752786322811839e-07,
      "loss": 1.3935,
      "step": 39560
    },
    {
      "epoch": 9.813379626759254,
      "grad_norm": 1.0630862712860107,
      "learning_rate": 1.706976868020127e-07,
      "loss": 1.3429,
      "step": 39570
    },
    {
      "epoch": 9.815859631719263,
      "grad_norm": 1.1879829168319702,
      "learning_rate": 1.6617734747509383e-07,
      "loss": 1.3798,
      "step": 39580
    },
    {
      "epoch": 9.818339636679273,
      "grad_norm": 1.0357943773269653,
      "learning_rate": 1.6171761704469878e-07,
      "loss": 1.4031,
      "step": 39590
    },
    {
      "epoch": 9.820819641639282,
      "grad_norm": 1.0229921340942383,
      "learning_rate": 1.5731849821833954e-07,
      "loss": 1.3625,
      "step": 39600
    },
    {
      "epoch": 9.823299646599294,
      "grad_norm": 1.0634421110153198,
      "learning_rate": 1.5297999366671312e-07,
      "loss": 1.3808,
      "step": 39610
    },
    {
      "epoch": 9.825779651559303,
      "grad_norm": 1.0519945621490479,
      "learning_rate": 1.487021060236904e-07,
      "loss": 1.3687,
      "step": 39620
    },
    {
      "epoch": 9.828259656519313,
      "grad_norm": 1.0606703758239746,
      "learning_rate": 1.444848378863828e-07,
      "loss": 1.3501,
      "step": 39630
    },
    {
      "epoch": 9.830739661479322,
      "grad_norm": 1.1069635152816772,
      "learning_rate": 1.403281918150978e-07,
      "loss": 1.416,
      "step": 39640
    },
    {
      "epoch": 9.833219666439334,
      "grad_norm": 1.057266354560852,
      "learning_rate": 1.3623217033330583e-07,
      "loss": 1.3404,
      "step": 39650
    },
    {
      "epoch": 9.835699671399343,
      "grad_norm": 1.0354572534561157,
      "learning_rate": 1.3219677592770652e-07,
      "loss": 1.3787,
      "step": 39660
    },
    {
      "epoch": 9.838179676359353,
      "grad_norm": 1.0284291505813599,
      "learning_rate": 1.2822201104816246e-07,
      "loss": 1.3968,
      "step": 39670
    },
    {
      "epoch": 9.840659681319362,
      "grad_norm": 1.1412193775177002,
      "learning_rate": 1.2430787810776555e-07,
      "loss": 1.3948,
      "step": 39680
    },
    {
      "epoch": 9.843139686279372,
      "grad_norm": 1.154058814048767,
      "learning_rate": 1.204543794827595e-07,
      "loss": 1.3643,
      "step": 39690
    },
    {
      "epoch": 9.845619691239383,
      "grad_norm": 1.053155541419983,
      "learning_rate": 1.1666151751260623e-07,
      "loss": 1.3957,
      "step": 39700
    },
    {
      "epoch": 9.848099696199393,
      "grad_norm": 1.0365995168685913,
      "learning_rate": 1.1292929449994161e-07,
      "loss": 1.3524,
      "step": 39710
    },
    {
      "epoch": 9.850579701159402,
      "grad_norm": 1.0026479959487915,
      "learning_rate": 1.0925771271058649e-07,
      "loss": 1.406,
      "step": 39720
    },
    {
      "epoch": 9.853059706119412,
      "grad_norm": 1.0312763452529907,
      "learning_rate": 1.0564677437355786e-07,
      "loss": 1.3892,
      "step": 39730
    },
    {
      "epoch": 9.855539711079421,
      "grad_norm": 1.0404436588287354,
      "learning_rate": 1.0209648168103547e-07,
      "loss": 1.442,
      "step": 39740
    },
    {
      "epoch": 9.858019716039433,
      "grad_norm": 1.1056069135665894,
      "learning_rate": 9.860683678840632e-08,
      "loss": 1.4487,
      "step": 39750
    },
    {
      "epoch": 9.860499720999442,
      "grad_norm": 1.0739343166351318,
      "learning_rate": 9.517784181422019e-08,
      "loss": 1.4275,
      "step": 39760
    },
    {
      "epoch": 9.862979725959452,
      "grad_norm": 1.0319334268569946,
      "learning_rate": 9.180949884022294e-08,
      "loss": 1.3365,
      "step": 39770
    },
    {
      "epoch": 9.865459730919461,
      "grad_norm": 1.0002667903900146,
      "learning_rate": 8.850180991131219e-08,
      "loss": 1.4084,
      "step": 39780
    },
    {
      "epoch": 9.867939735879471,
      "grad_norm": 1.0114107131958008,
      "learning_rate": 8.525477703558161e-08,
      "loss": 1.3684,
      "step": 39790
    },
    {
      "epoch": 9.870419740839482,
      "grad_norm": 1.0454875230789185,
      "learning_rate": 8.206840218430989e-08,
      "loss": 1.3802,
      "step": 39800
    },
    {
      "epoch": 9.872899745799492,
      "grad_norm": 1.1178107261657715,
      "learning_rate": 7.894268729192744e-08,
      "loss": 1.3888,
      "step": 39810
    },
    {
      "epoch": 9.875379750759501,
      "grad_norm": 1.1164809465408325,
      "learning_rate": 7.587763425604965e-08,
      "loss": 1.3838,
      "step": 39820
    },
    {
      "epoch": 9.877859755719511,
      "grad_norm": 1.0479484796524048,
      "learning_rate": 7.287324493747693e-08,
      "loss": 1.401,
      "step": 39830
    },
    {
      "epoch": 9.880339760679522,
      "grad_norm": 1.1312552690505981,
      "learning_rate": 6.992952116013918e-08,
      "loss": 1.3807,
      "step": 39840
    },
    {
      "epoch": 9.882819765639532,
      "grad_norm": 1.045345664024353,
      "learning_rate": 6.70464647111957e-08,
      "loss": 1.3687,
      "step": 39850
    },
    {
      "epoch": 9.885299770599541,
      "grad_norm": 1.0771009922027588,
      "learning_rate": 6.422407734092416e-08,
      "loss": 1.4415,
      "step": 39860
    },
    {
      "epoch": 9.887779775559551,
      "grad_norm": 1.0385736227035522,
      "learning_rate": 6.146236076279843e-08,
      "loss": 1.3948,
      "step": 39870
    },
    {
      "epoch": 9.89025978051956,
      "grad_norm": 1.1350256204605103,
      "learning_rate": 5.8761316653455076e-08,
      "loss": 1.3736,
      "step": 39880
    },
    {
      "epoch": 9.892739785479572,
      "grad_norm": 0.9889593720436096,
      "learning_rate": 5.612094665267131e-08,
      "loss": 1.3958,
      "step": 39890
    },
    {
      "epoch": 9.895219790439581,
      "grad_norm": 0.9963741302490234,
      "learning_rate": 5.354125236343155e-08,
      "loss": 1.4073,
      "step": 39900
    },
    {
      "epoch": 9.89769979539959,
      "grad_norm": 0.9665492177009583,
      "learning_rate": 5.102223535186079e-08,
      "loss": 1.3998,
      "step": 39910
    },
    {
      "epoch": 9.9001798003596,
      "grad_norm": 1.0084357261657715,
      "learning_rate": 4.856389714723575e-08,
      "loss": 1.4046,
      "step": 39920
    },
    {
      "epoch": 9.90265980531961,
      "grad_norm": 1.0370889902114868,
      "learning_rate": 4.616623924201813e-08,
      "loss": 1.4053,
      "step": 39930
    },
    {
      "epoch": 9.905139810279621,
      "grad_norm": 0.9871793985366821,
      "learning_rate": 4.3829263091810235e-08,
      "loss": 1.4169,
      "step": 39940
    },
    {
      "epoch": 9.90761981523963,
      "grad_norm": 1.0563249588012695,
      "learning_rate": 4.1552970115399384e-08,
      "loss": 1.3651,
      "step": 39950
    },
    {
      "epoch": 9.91009982019964,
      "grad_norm": 1.017209768295288,
      "learning_rate": 3.933736169471347e-08,
      "loss": 1.3772,
      "step": 39960
    },
    {
      "epoch": 9.91257982515965,
      "grad_norm": 1.0138312578201294,
      "learning_rate": 3.7182439174832106e-08,
      "loss": 1.4503,
      "step": 39970
    },
    {
      "epoch": 9.91505983011966,
      "grad_norm": 1.13675057888031,
      "learning_rate": 3.5088203864031e-08,
      "loss": 1.4347,
      "step": 39980
    },
    {
      "epoch": 9.91753983507967,
      "grad_norm": 1.0499796867370605,
      "learning_rate": 3.305465703368205e-08,
      "loss": 1.4008,
      "step": 39990
    },
    {
      "epoch": 9.92001984003968,
      "grad_norm": 1.042291522026062,
      "learning_rate": 3.1081799918375454e-08,
      "loss": 1.4099,
      "step": 40000
    },
    {
      "epoch": 9.92249984499969,
      "grad_norm": 1.1130834817886353,
      "learning_rate": 2.916963371580872e-08,
      "loss": 1.4223,
      "step": 40010
    },
    {
      "epoch": 9.9249798499597,
      "grad_norm": 1.0592601299285889,
      "learning_rate": 2.7318159586864344e-08,
      "loss": 1.3967,
      "step": 40020
    },
    {
      "epoch": 9.927459854919709,
      "grad_norm": 1.015381097793579,
      "learning_rate": 2.5527378655565427e-08,
      "loss": 1.3518,
      "step": 40030
    },
    {
      "epoch": 9.92993985987972,
      "grad_norm": 1.0501749515533447,
      "learning_rate": 2.379729200908676e-08,
      "loss": 1.3567,
      "step": 40040
    },
    {
      "epoch": 9.93241986483973,
      "grad_norm": 1.1451592445373535,
      "learning_rate": 2.2127900697777036e-08,
      "loss": 1.3506,
      "step": 40050
    },
    {
      "epoch": 9.93489986979974,
      "grad_norm": 1.1427299976348877,
      "learning_rate": 2.0519205735103353e-08,
      "loss": 1.3914,
      "step": 40060
    },
    {
      "epoch": 9.937379874759749,
      "grad_norm": 1.0566608905792236,
      "learning_rate": 1.897120809769559e-08,
      "loss": 1.4098,
      "step": 40070
    },
    {
      "epoch": 9.939859879719759,
      "grad_norm": 1.0093517303466797,
      "learning_rate": 1.7483908725357545e-08,
      "loss": 1.4334,
      "step": 40080
    },
    {
      "epoch": 9.94233988467977,
      "grad_norm": 1.0438250303268433,
      "learning_rate": 1.6057308521022498e-08,
      "loss": 1.3979,
      "step": 40090
    },
    {
      "epoch": 9.94481988963978,
      "grad_norm": 1.0427708625793457,
      "learning_rate": 1.4691408350764324e-08,
      "loss": 1.4356,
      "step": 40100
    },
    {
      "epoch": 9.947299894599789,
      "grad_norm": 1.1051512956619263,
      "learning_rate": 1.3386209043819708e-08,
      "loss": 1.3958,
      "step": 40110
    },
    {
      "epoch": 9.949779899559799,
      "grad_norm": 1.0951145887374878,
      "learning_rate": 1.2141711392588129e-08,
      "loss": 1.3822,
      "step": 40120
    },
    {
      "epoch": 9.95225990451981,
      "grad_norm": 1.099120855331421,
      "learning_rate": 1.0957916152587456e-08,
      "loss": 1.4432,
      "step": 40130
    },
    {
      "epoch": 9.95473990947982,
      "grad_norm": 1.1092506647109985,
      "learning_rate": 9.834824042498358e-09,
      "loss": 1.426,
      "step": 40140
    },
    {
      "epoch": 9.957219914439829,
      "grad_norm": 1.007125973701477,
      "learning_rate": 8.772435744153206e-09,
      "loss": 1.3725,
      "step": 40150
    },
    {
      "epoch": 9.959699919399839,
      "grad_norm": 1.0477224588394165,
      "learning_rate": 7.770751902513862e-09,
      "loss": 1.4066,
      "step": 40160
    },
    {
      "epoch": 9.962179924359848,
      "grad_norm": 1.011325716972351,
      "learning_rate": 6.829773125716088e-09,
      "loss": 1.3982,
      "step": 40170
    },
    {
      "epoch": 9.96465992931986,
      "grad_norm": 1.067104458808899,
      "learning_rate": 5.949499985025142e-09,
      "loss": 1.3502,
      "step": 40180
    },
    {
      "epoch": 9.967139934279869,
      "grad_norm": 1.0777113437652588,
      "learning_rate": 5.129933014835775e-09,
      "loss": 1.3502,
      "step": 40190
    },
    {
      "epoch": 9.969619939239879,
      "grad_norm": 1.0474942922592163,
      "learning_rate": 4.371072712727742e-09,
      "loss": 1.4274,
      "step": 40200
    },
    {
      "epoch": 9.972099944199888,
      "grad_norm": 1.0896284580230713,
      "learning_rate": 3.6729195393769843e-09,
      "loss": 1.4181,
      "step": 40210
    },
    {
      "epoch": 9.974579949159898,
      "grad_norm": 1.1275458335876465,
      "learning_rate": 3.0354739186555512e-09,
      "loss": 1.4095,
      "step": 40220
    },
    {
      "epoch": 9.977059954119909,
      "grad_norm": 1.0350884199142456,
      "learning_rate": 2.45873623754278e-09,
      "loss": 1.3916,
      "step": 40230
    },
    {
      "epoch": 9.979539959079919,
      "grad_norm": 1.0565330982208252,
      "learning_rate": 1.9427068461808083e-09,
      "loss": 1.4198,
      "step": 40240
    },
    {
      "epoch": 9.982019964039928,
      "grad_norm": 1.0948325395584106,
      "learning_rate": 1.4873860578412668e-09,
      "loss": 1.3361,
      "step": 40250
    },
    {
      "epoch": 9.984499968999938,
      "grad_norm": 1.0515812635421753,
      "learning_rate": 1.0927741489585862e-09,
      "loss": 1.4043,
      "step": 40260
    },
    {
      "epoch": 9.986979973959947,
      "grad_norm": 1.0280400514602661,
      "learning_rate": 7.588713590966911e-10,
      "loss": 1.3602,
      "step": 40270
    },
    {
      "epoch": 9.989459978919959,
      "grad_norm": 1.0958027839660645,
      "learning_rate": 4.856778909601012e-10,
      "loss": 1.4598,
      "step": 40280
    },
    {
      "epoch": 9.991939983879968,
      "grad_norm": 1.048204779624939,
      "learning_rate": 2.731939104161363e-10,
      "loss": 1.3762,
      "step": 40290
    },
    {
      "epoch": 9.994419988839978,
      "grad_norm": 1.0323026180267334,
      "learning_rate": 1.2141954645050747e-10,
      "loss": 1.399,
      "step": 40300
    },
    {
      "epoch": 9.996899993799987,
      "grad_norm": 1.0778690576553345,
      "learning_rate": 3.035489122282797e-11,
      "loss": 1.3887,
      "step": 40310
    },
    {
      "epoch": 9.999379998759997,
      "grad_norm": 1.0634547472000122,
      "learning_rate": 0.0,
      "loss": 1.3647,
      "step": 40320
    },
    {
      "epoch": 9.999379998759997,
      "step": 40320,
      "total_flos": 2.647899280244736e+18,
      "train_loss": 1.5969342465911593,
      "train_runtime": 104090.1136,
      "train_samples_per_second": 1.55,
      "train_steps_per_second": 0.387
    }
  ],
  "logging_steps": 10,
  "max_steps": 40320,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 2.647899280244736e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}