{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "eval_steps": 500,
  "global_step": 24996,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.008001280204832774,
      "grad_norm": 0.9324016571044922,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 2.0562,
      "step": 50
    },
    {
      "epoch": 0.016002560409665547,
      "grad_norm": 0.4582217037677765,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.9037,
      "step": 100
    },
    {
      "epoch": 0.02400384061449832,
      "grad_norm": 0.6721668243408203,
      "learning_rate": 6e-06,
      "loss": 1.6957,
      "step": 150
    },
    {
      "epoch": 0.032005120819331094,
      "grad_norm": 0.6238133907318115,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.5897,
      "step": 200
    },
    {
      "epoch": 0.04000640102416387,
      "grad_norm": 0.6062737703323364,
      "learning_rate": 1e-05,
      "loss": 1.5664,
      "step": 250
    },
    {
      "epoch": 0.04800768122899664,
      "grad_norm": 0.6185649633407593,
      "learning_rate": 1.2e-05,
      "loss": 1.5315,
      "step": 300
    },
    {
      "epoch": 0.05600896143382941,
      "grad_norm": 0.6351732611656189,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 1.4996,
      "step": 350
    },
    {
      "epoch": 0.06401024163866219,
      "grad_norm": 0.6717058420181274,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.4784,
      "step": 400
    },
    {
      "epoch": 0.07201152184349496,
      "grad_norm": 0.6665578484535217,
      "learning_rate": 1.8e-05,
      "loss": 1.4795,
      "step": 450
    },
    {
      "epoch": 0.08001280204832774,
      "grad_norm": 0.5694233775138855,
      "learning_rate": 2e-05,
      "loss": 1.4583,
      "step": 500
    },
    {
      "epoch": 0.0880140822531605,
      "grad_norm": 0.6044241786003113,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 1.4392,
      "step": 550
    },
    {
      "epoch": 0.09601536245799328,
      "grad_norm": 0.6153579354286194,
      "learning_rate": 2.4e-05,
      "loss": 1.436,
      "step": 600
    },
    {
      "epoch": 0.10401664266282605,
      "grad_norm": 0.5287560224533081,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 1.4244,
      "step": 650
    },
    {
      "epoch": 0.11201792286765883,
      "grad_norm": 0.5597755908966064,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 1.4369,
      "step": 700
    },
    {
      "epoch": 0.1200192030724916,
      "grad_norm": 0.4981745183467865,
      "learning_rate": 3e-05,
      "loss": 1.3981,
      "step": 750
    },
    {
      "epoch": 0.12802048327732438,
      "grad_norm": 0.48743560910224915,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 1.4009,
      "step": 800
    },
    {
      "epoch": 0.13602176348215714,
      "grad_norm": 0.5637604594230652,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 1.4028,
      "step": 850
    },
    {
      "epoch": 0.14402304368698993,
      "grad_norm": 0.470486581325531,
      "learning_rate": 3.6e-05,
      "loss": 1.3938,
      "step": 900
    },
    {
      "epoch": 0.1520243238918227,
      "grad_norm": 0.5075104236602783,
      "learning_rate": 3.8e-05,
      "loss": 1.3859,
      "step": 950
    },
    {
      "epoch": 0.16002560409665548,
      "grad_norm": 0.4708271324634552,
      "learning_rate": 4e-05,
      "loss": 1.3865,
      "step": 1000
    },
    {
      "epoch": 0.16802688430148824,
      "grad_norm": 0.3994545638561249,
      "learning_rate": 4.2e-05,
      "loss": 1.4072,
      "step": 1050
    },
    {
      "epoch": 0.176028164506321,
      "grad_norm": 0.4936317801475525,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 1.3828,
      "step": 1100
    },
    {
      "epoch": 0.1840294447111538,
      "grad_norm": 0.3775925934314728,
      "learning_rate": 4.600000000000001e-05,
      "loss": 1.4046,
      "step": 1150
    },
    {
      "epoch": 0.19203072491598655,
      "grad_norm": 0.45916563272476196,
      "learning_rate": 4.8e-05,
      "loss": 1.3725,
      "step": 1200
    },
    {
      "epoch": 0.20003200512081934,
      "grad_norm": 0.42247775197029114,
      "learning_rate": 5e-05,
      "loss": 1.371,
      "step": 1250
    },
    {
      "epoch": 0.2080332853256521,
      "grad_norm": 0.4264281392097473,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 1.3729,
      "step": 1300
    },
    {
      "epoch": 0.2160345655304849,
      "grad_norm": 0.3741685152053833,
      "learning_rate": 5.4000000000000005e-05,
      "loss": 1.3621,
      "step": 1350
    },
    {
      "epoch": 0.22403584573531765,
      "grad_norm": 0.392640084028244,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 1.3557,
      "step": 1400
    },
    {
      "epoch": 0.2320371259401504,
      "grad_norm": 0.3837718367576599,
      "learning_rate": 5.8e-05,
      "loss": 1.3599,
      "step": 1450
    },
    {
      "epoch": 0.2400384061449832,
      "grad_norm": 0.36931145191192627,
      "learning_rate": 6e-05,
      "loss": 1.3527,
      "step": 1500
    },
    {
      "epoch": 0.24803968634981596,
      "grad_norm": 0.4106380343437195,
      "learning_rate": 6.2e-05,
      "loss": 1.3531,
      "step": 1550
    },
    {
      "epoch": 0.25604096655464875,
      "grad_norm": 0.41261059045791626,
      "learning_rate": 6.400000000000001e-05,
      "loss": 1.3722,
      "step": 1600
    },
    {
      "epoch": 0.2640422467594815,
      "grad_norm": 0.3928689658641815,
      "learning_rate": 6.6e-05,
      "loss": 1.3423,
      "step": 1650
    },
    {
      "epoch": 0.2720435269643143,
      "grad_norm": 0.4072926640510559,
      "learning_rate": 6.800000000000001e-05,
      "loss": 1.3458,
      "step": 1700
    },
    {
      "epoch": 0.28004480716914704,
      "grad_norm": 0.3700864315032959,
      "learning_rate": 7e-05,
      "loss": 1.3566,
      "step": 1750
    },
    {
      "epoch": 0.28804608737397985,
      "grad_norm": 0.36805135011672974,
      "learning_rate": 7.2e-05,
      "loss": 1.3124,
      "step": 1800
    },
    {
      "epoch": 0.2960473675788126,
      "grad_norm": 0.39832988381385803,
      "learning_rate": 7.4e-05,
      "loss": 1.3408,
      "step": 1850
    },
    {
      "epoch": 0.3040486477836454,
      "grad_norm": 0.4116155505180359,
      "learning_rate": 7.6e-05,
      "loss": 1.3259,
      "step": 1900
    },
    {
      "epoch": 0.31204992798847814,
      "grad_norm": 0.3844102919101715,
      "learning_rate": 7.800000000000001e-05,
      "loss": 1.3223,
      "step": 1950
    },
    {
      "epoch": 0.32005120819331095,
      "grad_norm": 0.36435940861701965,
      "learning_rate": 8e-05,
      "loss": 1.3493,
      "step": 2000
    },
    {
      "epoch": 0.3280524883981437,
      "grad_norm": 0.40303346514701843,
      "learning_rate": 8.2e-05,
      "loss": 1.313,
      "step": 2050
    },
    {
      "epoch": 0.3360537686029765,
      "grad_norm": 0.3959302604198456,
      "learning_rate": 8.4e-05,
      "loss": 1.3116,
      "step": 2100
    },
    {
      "epoch": 0.34405504880780924,
      "grad_norm": 0.42148756980895996,
      "learning_rate": 8.6e-05,
      "loss": 1.3172,
      "step": 2150
    },
    {
      "epoch": 0.352056329012642,
      "grad_norm": 0.35398295521736145,
      "learning_rate": 8.800000000000001e-05,
      "loss": 1.3123,
      "step": 2200
    },
    {
      "epoch": 0.3600576092174748,
      "grad_norm": 0.32611459493637085,
      "learning_rate": 9e-05,
      "loss": 1.3291,
      "step": 2250
    },
    {
      "epoch": 0.3680588894223076,
      "grad_norm": 0.3089679777622223,
      "learning_rate": 9.200000000000001e-05,
      "loss": 1.3162,
      "step": 2300
    },
    {
      "epoch": 0.37606016962714034,
      "grad_norm": 0.38052898645401,
      "learning_rate": 9.4e-05,
      "loss": 1.2888,
      "step": 2350
    },
    {
      "epoch": 0.3840614498319731,
      "grad_norm": 0.40810805559158325,
      "learning_rate": 9.6e-05,
      "loss": 1.292,
      "step": 2400
    },
    {
      "epoch": 0.39206273003680586,
      "grad_norm": 0.3573181927204132,
      "learning_rate": 9.8e-05,
      "loss": 1.3094,
      "step": 2450
    },
    {
      "epoch": 0.4000640102416387,
      "grad_norm": 0.3486098349094391,
      "learning_rate": 0.0001,
      "loss": 1.2908,
      "step": 2500
    },
    {
      "epoch": 0.40806529044647144,
      "grad_norm": 0.4237920045852661,
      "learning_rate": 9.999878110192404e-05,
      "loss": 1.2868,
      "step": 2550
    },
    {
      "epoch": 0.4160665706513042,
      "grad_norm": 0.3493655025959015,
      "learning_rate": 9.999512446712467e-05,
      "loss": 1.3221,
      "step": 2600
    },
    {
      "epoch": 0.42406785085613696,
      "grad_norm": 0.3716041147708893,
      "learning_rate": 9.998903027388448e-05,
      "loss": 1.2728,
      "step": 2650
    },
    {
      "epoch": 0.4320691310609698,
      "grad_norm": 0.43094709515571594,
      "learning_rate": 9.998049881933149e-05,
      "loss": 1.2864,
      "step": 2700
    },
    {
      "epoch": 0.44007041126580254,
      "grad_norm": 0.379719078540802,
      "learning_rate": 9.996953051942463e-05,
      "loss": 1.2956,
      "step": 2750
    },
    {
      "epoch": 0.4480716914706353,
      "grad_norm": 0.45424267649650574,
      "learning_rate": 9.995612590893352e-05,
      "loss": 1.2927,
      "step": 2800
    },
    {
      "epoch": 0.45607297167546806,
      "grad_norm": 0.34709858894348145,
      "learning_rate": 9.99402856414123e-05,
      "loss": 1.2801,
      "step": 2850
    },
    {
      "epoch": 0.4640742518803008,
      "grad_norm": 0.3956676423549652,
      "learning_rate": 9.992201048916783e-05,
      "loss": 1.2869,
      "step": 2900
    },
    {
      "epoch": 0.47207553208513364,
      "grad_norm": 0.3462512493133545,
      "learning_rate": 9.990130134322201e-05,
      "loss": 1.268,
      "step": 2950
    },
    {
      "epoch": 0.4800768122899664,
      "grad_norm": 0.3835497796535492,
      "learning_rate": 9.987815921326842e-05,
      "loss": 1.254,
      "step": 3000
    },
    {
      "epoch": 0.48807809249479917,
      "grad_norm": 0.37989285588264465,
      "learning_rate": 9.98525852276229e-05,
      "loss": 1.2869,
      "step": 3050
    },
    {
      "epoch": 0.4960793726996319,
      "grad_norm": 0.3443322777748108,
      "learning_rate": 9.982458063316877e-05,
      "loss": 1.2933,
      "step": 3100
    },
    {
      "epoch": 0.5040806529044647,
      "grad_norm": 0.32426556944847107,
      "learning_rate": 9.979414679529588e-05,
      "loss": 1.2756,
      "step": 3150
    },
    {
      "epoch": 0.5120819331092975,
      "grad_norm": 0.4362022578716278,
      "learning_rate": 9.976128519783405e-05,
      "loss": 1.251,
      "step": 3200
    },
    {
      "epoch": 0.5200832133141302,
      "grad_norm": 0.3969637155532837,
      "learning_rate": 9.972599744298086e-05,
      "loss": 1.2593,
      "step": 3250
    },
    {
      "epoch": 0.528084493518963,
      "grad_norm": 0.41199642419815063,
      "learning_rate": 9.96882852512233e-05,
      "loss": 1.2484,
      "step": 3300
    },
    {
      "epoch": 0.5360857737237958,
      "grad_norm": 0.3282572329044342,
      "learning_rate": 9.964815046125412e-05,
      "loss": 1.2559,
      "step": 3350
    },
    {
      "epoch": 0.5440870539286286,
      "grad_norm": 0.4450869858264923,
      "learning_rate": 9.960559502988208e-05,
      "loss": 1.2488,
      "step": 3400
    },
    {
      "epoch": 0.5520883341334614,
      "grad_norm": 0.356630802154541,
      "learning_rate": 9.956062103193648e-05,
      "loss": 1.2688,
      "step": 3450
    },
    {
      "epoch": 0.5600896143382941,
      "grad_norm": 0.35085442662239075,
      "learning_rate": 9.95132306601661e-05,
      "loss": 1.2398,
      "step": 3500
    },
    {
      "epoch": 0.5680908945431269,
      "grad_norm": 0.4403521716594696,
      "learning_rate": 9.946342622513229e-05,
      "loss": 1.2442,
      "step": 3550
    },
    {
      "epoch": 0.5760921747479597,
      "grad_norm": 0.4456493854522705,
      "learning_rate": 9.941121015509621e-05,
      "loss": 1.2413,
      "step": 3600
    },
    {
      "epoch": 0.5840934549527924,
      "grad_norm": 0.4137299954891205,
      "learning_rate": 9.935658499590059e-05,
      "loss": 1.2546,
      "step": 3650
    },
    {
      "epoch": 0.5920947351576252,
      "grad_norm": 0.3572884798049927,
      "learning_rate": 9.929955341084547e-05,
      "loss": 1.2533,
      "step": 3700
    },
    {
      "epoch": 0.600096015362458,
      "grad_norm": 0.42966127395629883,
      "learning_rate": 9.924011818055843e-05,
      "loss": 1.2389,
      "step": 3750
    },
    {
      "epoch": 0.6080972955672908,
      "grad_norm": 0.4403945803642273,
      "learning_rate": 9.917828220285896e-05,
      "loss": 1.2198,
      "step": 3800
    },
    {
      "epoch": 0.6160985757721236,
      "grad_norm": 0.4667624235153198,
      "learning_rate": 9.911404849261726e-05,
      "loss": 1.2417,
      "step": 3850
    },
    {
      "epoch": 0.6240998559769563,
      "grad_norm": 0.6005256772041321,
      "learning_rate": 9.904742018160715e-05,
      "loss": 1.2006,
      "step": 3900
    },
    {
      "epoch": 0.6321011361817891,
      "grad_norm": 0.3864199221134186,
      "learning_rate": 9.897840051835343e-05,
      "loss": 1.2164,
      "step": 3950
    },
    {
      "epoch": 0.6401024163866219,
      "grad_norm": 0.5411766171455383,
      "learning_rate": 9.89069928679735e-05,
      "loss": 1.2187,
      "step": 4000
    },
    {
      "epoch": 0.6481036965914546,
      "grad_norm": 0.38500306010246277,
      "learning_rate": 9.883320071201325e-05,
      "loss": 1.2363,
      "step": 4050
    },
    {
      "epoch": 0.6561049767962874,
      "grad_norm": 0.4081592559814453,
      "learning_rate": 9.875702764827737e-05,
      "loss": 1.2247,
      "step": 4100
    },
    {
      "epoch": 0.6641062570011201,
      "grad_norm": 0.43678033351898193,
      "learning_rate": 9.86784773906539e-05,
      "loss": 1.2323,
      "step": 4150
    },
    {
      "epoch": 0.672107537205953,
      "grad_norm": 0.4319170117378235,
      "learning_rate": 9.859755376893314e-05,
      "loss": 1.1906,
      "step": 4200
    },
    {
      "epoch": 0.6801088174107858,
      "grad_norm": 0.4729995131492615,
      "learning_rate": 9.851426072862098e-05,
      "loss": 1.2098,
      "step": 4250
    },
    {
      "epoch": 0.6881100976156185,
      "grad_norm": 0.4233373701572418,
      "learning_rate": 9.842860233074645e-05,
      "loss": 1.204,
      "step": 4300
    },
    {
      "epoch": 0.6961113778204513,
      "grad_norm": 0.48369497060775757,
      "learning_rate": 9.834058275166383e-05,
      "loss": 1.2046,
      "step": 4350
    },
    {
      "epoch": 0.704112658025284,
      "grad_norm": 0.36634331941604614,
      "learning_rate": 9.825020628284896e-05,
      "loss": 1.2117,
      "step": 4400
    },
    {
      "epoch": 0.7121139382301168,
      "grad_norm": 0.41213107109069824,
      "learning_rate": 9.815747733068995e-05,
      "loss": 1.1977,
      "step": 4450
    },
    {
      "epoch": 0.7201152184349496,
      "grad_norm": 0.37744736671447754,
      "learning_rate": 9.80624004162725e-05,
      "loss": 1.2372,
      "step": 4500
    },
    {
      "epoch": 0.7281164986397823,
      "grad_norm": 0.49204540252685547,
      "learning_rate": 9.79649801751593e-05,
      "loss": 1.2191,
      "step": 4550
    },
    {
      "epoch": 0.7361177788446152,
      "grad_norm": 0.42844048142433167,
      "learning_rate": 9.786522135716415e-05,
      "loss": 1.2114,
      "step": 4600
    },
    {
      "epoch": 0.7441190590494479,
      "grad_norm": 0.5176280736923218,
      "learning_rate": 9.776312882612028e-05,
      "loss": 1.2041,
      "step": 4650
    },
    {
      "epoch": 0.7521203392542807,
      "grad_norm": 0.44180744886398315,
      "learning_rate": 9.765870755964331e-05,
      "loss": 1.2096,
      "step": 4700
    },
    {
      "epoch": 0.7601216194591135,
      "grad_norm": 0.4751160442829132,
      "learning_rate": 9.755196264888846e-05,
      "loss": 1.1928,
      "step": 4750
    },
    {
      "epoch": 0.7681228996639462,
      "grad_norm": 0.4147936999797821,
      "learning_rate": 9.744289929830236e-05,
      "loss": 1.2183,
      "step": 4800
    },
    {
      "epoch": 0.776124179868779,
      "grad_norm": 0.45143744349479675,
      "learning_rate": 9.733152282536936e-05,
      "loss": 1.1798,
      "step": 4850
    },
    {
      "epoch": 0.7841254600736117,
      "grad_norm": 0.46376484632492065,
      "learning_rate": 9.721783866035218e-05,
      "loss": 1.1871,
      "step": 4900
    },
    {
      "epoch": 0.7921267402784445,
      "grad_norm": 0.4516090154647827,
      "learning_rate": 9.710185234602727e-05,
      "loss": 1.1901,
      "step": 4950
    },
    {
      "epoch": 0.8001280204832774,
      "grad_norm": 0.4624463617801666,
      "learning_rate": 9.698356953741438e-05,
      "loss": 1.1848,
      "step": 5000
    },
    {
      "epoch": 0.8081293006881101,
      "grad_norm": 0.41177043318748474,
      "learning_rate": 9.686299600150108e-05,
      "loss": 1.1934,
      "step": 5050
    },
    {
      "epoch": 0.8161305808929429,
      "grad_norm": 0.46030393242836,
      "learning_rate": 9.674013761696136e-05,
      "loss": 1.1695,
      "step": 5100
    },
    {
      "epoch": 0.8241318610977757,
      "grad_norm": 0.48872360587120056,
      "learning_rate": 9.661500037386921e-05,
      "loss": 1.192,
      "step": 5150
    },
    {
      "epoch": 0.8321331413026084,
      "grad_norm": 0.36913803219795227,
      "learning_rate": 9.648759037340638e-05,
      "loss": 1.1624,
      "step": 5200
    },
    {
      "epoch": 0.8401344215074412,
      "grad_norm": 0.5374457836151123,
      "learning_rate": 9.635791382756508e-05,
      "loss": 1.1847,
      "step": 5250
    },
    {
      "epoch": 0.8481357017122739,
      "grad_norm": 0.43082547187805176,
      "learning_rate": 9.622597705884496e-05,
      "loss": 1.1798,
      "step": 5300
    },
    {
      "epoch": 0.8561369819171067,
      "grad_norm": 0.4757367968559265,
      "learning_rate": 9.609178649994499e-05,
      "loss": 1.185,
      "step": 5350
    },
    {
      "epoch": 0.8641382621219396,
      "grad_norm": 0.41058582067489624,
      "learning_rate": 9.595534869344974e-05,
      "loss": 1.18,
      "step": 5400
    },
    {
      "epoch": 0.8721395423267723,
      "grad_norm": 0.4382982552051544,
      "learning_rate": 9.581667029151039e-05,
      "loss": 1.1977,
      "step": 5450
    },
    {
      "epoch": 0.8801408225316051,
      "grad_norm": 0.44662296772003174,
      "learning_rate": 9.567575805552042e-05,
      "loss": 1.1841,
      "step": 5500
    },
    {
      "epoch": 0.8881421027364378,
      "grad_norm": 0.4811069071292877,
      "learning_rate": 9.553261885578597e-05,
      "loss": 1.1765,
      "step": 5550
    },
    {
      "epoch": 0.8961433829412706,
      "grad_norm": 0.4336834251880646,
      "learning_rate": 9.538725967119087e-05,
      "loss": 1.1879,
      "step": 5600
    },
    {
      "epoch": 0.9041446631461034,
      "grad_norm": 0.44644805788993835,
      "learning_rate": 9.523968758885631e-05,
      "loss": 1.2058,
      "step": 5650
    },
    {
      "epoch": 0.9121459433509361,
      "grad_norm": 0.4499703347682953,
      "learning_rate": 9.508990980379537e-05,
      "loss": 1.1397,
      "step": 5700
    },
    {
      "epoch": 0.920147223555769,
      "grad_norm": 0.4322624206542969,
      "learning_rate": 9.493793361856226e-05,
      "loss": 1.1396,
      "step": 5750
    },
    {
      "epoch": 0.9281485037606017,
      "grad_norm": 0.43696922063827515,
      "learning_rate": 9.478376644289613e-05,
      "loss": 1.1698,
      "step": 5800
    },
    {
      "epoch": 0.9361497839654345,
      "grad_norm": 0.4423019587993622,
      "learning_rate": 9.462741579335992e-05,
      "loss": 1.1365,
      "step": 5850
    },
    {
      "epoch": 0.9441510641702673,
      "grad_norm": 0.4975908696651459,
      "learning_rate": 9.44688892929739e-05,
      "loss": 1.1775,
      "step": 5900
    },
    {
      "epoch": 0.9521523443751,
      "grad_norm": 0.49656620621681213,
      "learning_rate": 9.430819467084392e-05,
      "loss": 1.141,
      "step": 5950
    },
    {
      "epoch": 0.9601536245799328,
      "grad_norm": 0.4862751066684723,
      "learning_rate": 9.414533976178458e-05,
      "loss": 1.144,
      "step": 6000
    },
    {
      "epoch": 0.9681549047847655,
      "grad_norm": 0.5158004760742188,
      "learning_rate": 9.398033250593733e-05,
      "loss": 1.1603,
      "step": 6050
    },
    {
      "epoch": 0.9761561849895983,
      "grad_norm": 0.4406384229660034,
      "learning_rate": 9.381318094838321e-05,
      "loss": 1.1351,
      "step": 6100
    },
    {
      "epoch": 0.9841574651944311,
      "grad_norm": 0.4819541573524475,
      "learning_rate": 9.36438932387507e-05,
      "loss": 1.172,
      "step": 6150
    },
    {
      "epoch": 0.9921587453992639,
      "grad_norm": 0.4627222418785095,
      "learning_rate": 9.347247763081835e-05,
      "loss": 1.1587,
      "step": 6200
    },
    {
      "epoch": 1.0001600256040966,
      "grad_norm": 0.4336191713809967,
      "learning_rate": 9.329894248211233e-05,
      "loss": 1.1324,
      "step": 6250
    },
    {
      "epoch": 1.0081613058089294,
      "grad_norm": 0.5482844114303589,
      "learning_rate": 9.312329625349902e-05,
      "loss": 1.0647,
      "step": 6300
    },
    {
      "epoch": 1.0161625860137622,
      "grad_norm": 0.48142924904823303,
      "learning_rate": 9.294554750877242e-05,
      "loss": 1.0681,
      "step": 6350
    },
    {
      "epoch": 1.024163866218595,
      "grad_norm": 0.45703792572021484,
      "learning_rate": 9.276570491423664e-05,
      "loss": 1.0956,
      "step": 6400
    },
    {
      "epoch": 1.0321651464234278,
      "grad_norm": 0.5305693745613098,
      "learning_rate": 9.258377723828338e-05,
      "loss": 1.0476,
      "step": 6450
    },
    {
      "epoch": 1.0401664266282604,
      "grad_norm": 0.5061540603637695,
      "learning_rate": 9.23997733509644e-05,
      "loss": 1.0613,
      "step": 6500
    },
    {
      "epoch": 1.0481677068330932,
      "grad_norm": 0.4534345865249634,
      "learning_rate": 9.221370222355908e-05,
      "loss": 1.0604,
      "step": 6550
    },
    {
      "epoch": 1.056168987037926,
      "grad_norm": 0.5327995419502258,
      "learning_rate": 9.202557292813697e-05,
      "loss": 1.0396,
      "step": 6600
    },
    {
      "epoch": 1.0641702672427589,
      "grad_norm": 0.5349058508872986,
      "learning_rate": 9.183539463711555e-05,
      "loss": 1.0909,
      "step": 6650
    },
    {
      "epoch": 1.0721715474475917,
      "grad_norm": 0.5274565815925598,
      "learning_rate": 9.16431766228129e-05,
      "loss": 1.0539,
      "step": 6700
    },
    {
      "epoch": 1.0801728276524245,
      "grad_norm": 0.5548765659332275,
      "learning_rate": 9.144892825699577e-05,
      "loss": 1.0438,
      "step": 6750
    },
    {
      "epoch": 1.088174107857257,
      "grad_norm": 0.5511574745178223,
      "learning_rate": 9.125265901042251e-05,
      "loss": 1.091,
      "step": 6800
    },
    {
      "epoch": 1.09617538806209,
      "grad_norm": 0.5760679841041565,
      "learning_rate": 9.105437845238141e-05,
      "loss": 1.0522,
      "step": 6850
    },
    {
      "epoch": 1.1041766682669227,
      "grad_norm": 0.6349498629570007,
      "learning_rate": 9.085409625022411e-05,
      "loss": 1.0592,
      "step": 6900
    },
    {
      "epoch": 1.1121779484717556,
      "grad_norm": 0.6243582963943481,
      "learning_rate": 9.065182216889423e-05,
      "loss": 1.0683,
      "step": 6950
    },
    {
      "epoch": 1.1201792286765881,
      "grad_norm": 0.5389096140861511,
      "learning_rate": 9.04475660704513e-05,
      "loss": 1.0549,
      "step": 7000
    },
    {
      "epoch": 1.128180508881421,
      "grad_norm": 0.4826034605503082,
      "learning_rate": 9.024133791358995e-05,
      "loss": 1.078,
      "step": 7050
    },
    {
      "epoch": 1.1361817890862538,
      "grad_norm": 0.517662525177002,
      "learning_rate": 9.003314775315433e-05,
      "loss": 1.0857,
      "step": 7100
    },
    {
      "epoch": 1.1441830692910866,
      "grad_norm": 0.5285195112228394,
      "learning_rate": 8.982300573964788e-05,
      "loss": 1.0449,
      "step": 7150
    },
    {
      "epoch": 1.1521843494959194,
      "grad_norm": 0.47838062047958374,
      "learning_rate": 8.961092211873841e-05,
      "loss": 1.035,
      "step": 7200
    },
    {
      "epoch": 1.1601856297007522,
      "grad_norm": 0.5411238670349121,
      "learning_rate": 8.939690723075865e-05,
      "loss": 1.058,
      "step": 7250
    },
    {
      "epoch": 1.1681869099055848,
      "grad_norm": 0.49349942803382874,
      "learning_rate": 8.9180971510202e-05,
      "loss": 1.0567,
      "step": 7300
    },
    {
      "epoch": 1.1761881901104176,
      "grad_norm": 0.5397252440452576,
      "learning_rate": 8.896312548521382e-05,
      "loss": 1.0842,
      "step": 7350
    },
    {
      "epoch": 1.1841894703152505,
      "grad_norm": 0.5732976794242859,
      "learning_rate": 8.874337977707817e-05,
      "loss": 1.0581,
      "step": 7400
    },
    {
      "epoch": 1.1921907505200833,
      "grad_norm": 0.4691354036331177,
      "learning_rate": 8.852174509969984e-05,
      "loss": 1.0205,
      "step": 7450
    },
    {
      "epoch": 1.2001920307249159,
      "grad_norm": 0.6526204943656921,
      "learning_rate": 8.829823225908214e-05,
      "loss": 1.0387,
      "step": 7500
    },
    {
      "epoch": 1.2081933109297487,
      "grad_norm": 0.46403875946998596,
      "learning_rate": 8.80728521527999e-05,
      "loss": 1.0554,
      "step": 7550
    },
    {
      "epoch": 1.2161945911345815,
      "grad_norm": 0.462835431098938,
      "learning_rate": 8.784561576946824e-05,
      "loss": 1.0503,
      "step": 7600
    },
    {
      "epoch": 1.2241958713394143,
      "grad_norm": 0.5204878449440002,
      "learning_rate": 8.76165341882068e-05,
      "loss": 1.0815,
      "step": 7650
    },
    {
      "epoch": 1.2321971515442471,
      "grad_norm": 0.5061752796173096,
      "learning_rate": 8.738561857809949e-05,
      "loss": 1.0534,
      "step": 7700
    },
    {
      "epoch": 1.24019843174908,
      "grad_norm": 0.5491166710853577,
      "learning_rate": 8.715288019765006e-05,
      "loss": 1.0625,
      "step": 7750
    },
    {
      "epoch": 1.2481997119539125,
      "grad_norm": 0.4824821650981903,
      "learning_rate": 8.691833039423305e-05,
      "loss": 1.0473,
      "step": 7800
    },
    {
      "epoch": 1.2562009921587454,
      "grad_norm": 0.5545431971549988,
      "learning_rate": 8.668198060354065e-05,
      "loss": 1.0432,
      "step": 7850
    },
    {
      "epoch": 1.2642022723635782,
      "grad_norm": 0.5635075569152832,
      "learning_rate": 8.644384234902503e-05,
      "loss": 1.0473,
      "step": 7900
    },
    {
      "epoch": 1.272203552568411,
      "grad_norm": 0.49760839343070984,
      "learning_rate": 8.620392724133661e-05,
      "loss": 1.0616,
      "step": 7950
    },
    {
      "epoch": 1.2802048327732436,
      "grad_norm": 0.6228436231613159,
      "learning_rate": 8.596224697775795e-05,
      "loss": 1.0722,
      "step": 8000
    },
    {
      "epoch": 1.2882061129780764,
      "grad_norm": 0.5831848978996277,
      "learning_rate": 8.571881334163334e-05,
      "loss": 1.0581,
      "step": 8050
    },
    {
      "epoch": 1.2962073931829092,
      "grad_norm": 0.4925413131713867,
      "learning_rate": 8.547363820179442e-05,
      "loss": 1.0429,
      "step": 8100
    },
    {
      "epoch": 1.304208673387742,
      "grad_norm": 0.5401794910430908,
      "learning_rate": 8.522673351198143e-05,
      "loss": 1.0366,
      "step": 8150
    },
    {
      "epoch": 1.3122099535925749,
      "grad_norm": 0.5081512928009033,
      "learning_rate": 8.497811131026046e-05,
      "loss": 1.0464,
      "step": 8200
    },
    {
      "epoch": 1.3202112337974077,
      "grad_norm": 0.5704318881034851,
      "learning_rate": 8.472778371843641e-05,
      "loss": 1.0509,
      "step": 8250
    },
    {
      "epoch": 1.3282125140022405,
      "grad_norm": 0.505493700504303,
      "learning_rate": 8.447576294146209e-05,
      "loss": 1.0713,
      "step": 8300
    },
    {
      "epoch": 1.336213794207073,
      "grad_norm": 0.5492855906486511,
      "learning_rate": 8.422206126684311e-05,
      "loss": 1.0385,
      "step": 8350
    },
    {
      "epoch": 1.344215074411906,
      "grad_norm": 0.4849604666233063,
      "learning_rate": 8.39666910640388e-05,
      "loss": 1.0422,
      "step": 8400
    },
    {
      "epoch": 1.3522163546167387,
      "grad_norm": 0.5335291624069214,
      "learning_rate": 8.370966478385907e-05,
      "loss": 1.0235,
      "step": 8450
    },
    {
      "epoch": 1.3602176348215713,
      "grad_norm": 0.5804494619369507,
      "learning_rate": 8.345099495785753e-05,
      "loss": 1.0069,
      "step": 8500
    },
    {
      "epoch": 1.3682189150264041,
      "grad_norm": 0.5363398194313049,
      "learning_rate": 8.319069419772023e-05,
      "loss": 1.0402,
      "step": 8550
    },
    {
      "epoch": 1.376220195231237,
      "grad_norm": 0.6018884181976318,
      "learning_rate": 8.292877519465105e-05,
      "loss": 1.031,
      "step": 8600
    },
    {
      "epoch": 1.3842214754360698,
      "grad_norm": 0.5088986158370972,
      "learning_rate": 8.266525071875273e-05,
      "loss": 1.0349,
      "step": 8650
    },
    {
      "epoch": 1.3922227556409026,
      "grad_norm": 0.575141191482544,
      "learning_rate": 8.240013361840436e-05,
      "loss": 1.019,
      "step": 8700
    },
    {
      "epoch": 1.4002240358457354,
      "grad_norm": 0.5023386478424072,
      "learning_rate": 8.21334368196348e-05,
      "loss": 1.0428,
      "step": 8750
    },
    {
      "epoch": 1.4082253160505682,
      "grad_norm": 0.5338187217712402,
      "learning_rate": 8.186517332549276e-05,
      "loss": 1.0188,
      "step": 8800
    },
    {
      "epoch": 1.4162265962554008,
      "grad_norm": 0.5415542721748352,
      "learning_rate": 8.159535621541243e-05,
      "loss": 1.0172,
      "step": 8850
    },
    {
      "epoch": 1.4242278764602336,
      "grad_norm": 0.4934610426425934,
      "learning_rate": 8.13239986445761e-05,
      "loss": 1.0249,
      "step": 8900
    },
    {
      "epoch": 1.4322291566650664,
      "grad_norm": 0.5592052936553955,
      "learning_rate": 8.10511138432726e-05,
      "loss": 1.0295,
      "step": 8950
    },
    {
      "epoch": 1.4402304368698993,
      "grad_norm": 0.5904792547225952,
      "learning_rate": 8.07767151162523e-05,
      "loss": 1.0427,
      "step": 9000
    },
    {
      "epoch": 1.4482317170747319,
      "grad_norm": 0.5952058434486389,
      "learning_rate": 8.050081584207843e-05,
      "loss": 1.0504,
      "step": 9050
    },
    {
      "epoch": 1.4562329972795647,
      "grad_norm": 0.49933668971061707,
      "learning_rate": 8.022342947247474e-05,
      "loss": 1.0391,
      "step": 9100
    },
    {
      "epoch": 1.4642342774843975,
      "grad_norm": 0.5853399634361267,
      "learning_rate": 7.994456953166973e-05,
      "loss": 1.0013,
      "step": 9150
    },
    {
      "epoch": 1.4722355576892303,
      "grad_norm": 0.6290839910507202,
      "learning_rate": 7.966424961573722e-05,
      "loss": 1.0159,
      "step": 9200
    },
    {
      "epoch": 1.4802368378940631,
      "grad_norm": 0.6153242588043213,
      "learning_rate": 7.938248339193346e-05,
      "loss": 1.0312,
      "step": 9250
    },
    {
      "epoch": 1.488238118098896,
      "grad_norm": 0.5723234415054321,
      "learning_rate": 7.909928459803077e-05,
      "loss": 1.0198,
      "step": 9300
    },
    {
      "epoch": 1.4962393983037285,
      "grad_norm": 0.5766580700874329,
      "learning_rate": 7.881466704164774e-05,
      "loss": 0.9915,
      "step": 9350
    },
    {
      "epoch": 1.5042406785085614,
      "grad_norm": 0.6211101412773132,
      "learning_rate": 7.852864459957605e-05,
      "loss": 1.0324,
      "step": 9400
    },
    {
      "epoch": 1.5122419587133942,
      "grad_norm": 0.5500127673149109,
      "learning_rate": 7.824123121710387e-05,
      "loss": 1.0316,
      "step": 9450
    },
    {
      "epoch": 1.5202432389182268,
      "grad_norm": 0.6087145209312439,
      "learning_rate": 7.795244090733597e-05,
      "loss": 1.0334,
      "step": 9500
    },
    {
      "epoch": 1.5282445191230596,
      "grad_norm": 0.5456790924072266,
      "learning_rate": 7.766228775051045e-05,
      "loss": 1.0243,
      "step": 9550
    },
    {
      "epoch": 1.5362457993278924,
      "grad_norm": 0.613003671169281,
      "learning_rate": 7.737078589331229e-05,
      "loss": 0.9954,
      "step": 9600
    },
    {
      "epoch": 1.5442470795327252,
      "grad_norm": 0.5732637047767639,
      "learning_rate": 7.707794954818363e-05,
      "loss": 1.0172,
      "step": 9650
    },
    {
      "epoch": 1.552248359737558,
      "grad_norm": 0.5933035612106323,
      "learning_rate": 7.678379299263076e-05,
      "loss": 1.0049,
      "step": 9700
    },
    {
      "epoch": 1.5602496399423909,
      "grad_norm": 0.5576400756835938,
      "learning_rate": 7.648833056852807e-05,
      "loss": 0.9918,
      "step": 9750
    },
    {
      "epoch": 1.5682509201472237,
      "grad_norm": 0.4915982782840729,
      "learning_rate": 7.619157668141877e-05,
      "loss": 0.9949,
      "step": 9800
    },
    {
      "epoch": 1.5762522003520565,
      "grad_norm": 0.5492275357246399,
      "learning_rate": 7.58935457998125e-05,
      "loss": 1.0233,
      "step": 9850
    },
    {
      "epoch": 1.584253480556889,
      "grad_norm": 0.5720072984695435,
      "learning_rate": 7.559425245448006e-05,
      "loss": 1.053,
      "step": 9900
    },
    {
      "epoch": 1.592254760761722,
      "grad_norm": 0.5385164022445679,
      "learning_rate": 7.529371123774471e-05,
      "loss": 1.0329,
      "step": 9950
    },
    {
      "epoch": 1.6002560409665545,
      "grad_norm": 0.5508076548576355,
      "learning_rate": 7.49919368027709e-05,
      "loss": 1.0156,
      "step": 10000
    },
    {
      "epoch": 1.6082573211713873,
      "grad_norm": 0.5111932754516602,
      "learning_rate": 7.468894386284976e-05,
      "loss": 1.0061,
      "step": 10050
    },
    {
      "epoch": 1.6162586013762201,
      "grad_norm": 0.5652381777763367,
      "learning_rate": 7.438474719068173e-05,
      "loss": 0.999,
      "step": 10100
    },
    {
      "epoch": 1.624259881581053,
      "grad_norm": 0.5519747138023376,
      "learning_rate": 7.407936161765637e-05,
      "loss": 1.0129,
      "step": 10150
    },
    {
      "epoch": 1.6322611617858858,
      "grad_norm": 0.5380906462669373,
      "learning_rate": 7.377280203312917e-05,
      "loss": 0.9991,
      "step": 10200
    },
    {
      "epoch": 1.6402624419907186,
      "grad_norm": 0.635221540927887,
      "learning_rate": 7.346508338369563e-05,
      "loss": 0.9889,
      "step": 10250
    },
    {
      "epoch": 1.6482637221955514,
      "grad_norm": 0.4963364005088806,
      "learning_rate": 7.315622067246254e-05,
      "loss": 0.9808,
      "step": 10300
    },
    {
      "epoch": 1.6562650024003842,
      "grad_norm": 0.5576931238174438,
      "learning_rate": 7.28462289583165e-05,
      "loss": 1.0122,
      "step": 10350
    },
    {
      "epoch": 1.6642662826052168,
      "grad_norm": 0.5357556343078613,
      "learning_rate": 7.253512335518964e-05,
      "loss": 1.0193,
      "step": 10400
    },
    {
      "epoch": 1.6722675628100496,
      "grad_norm": 0.5757668018341064,
      "learning_rate": 7.22229190313228e-05,
      "loss": 1.0032,
      "step": 10450
    },
    {
      "epoch": 1.6802688430148822,
      "grad_norm": 0.5368041396141052,
      "learning_rate": 7.1909631208526e-05,
      "loss": 1.0159,
      "step": 10500
    },
    {
      "epoch": 1.688270123219715,
      "grad_norm": 0.5617921948432922,
      "learning_rate": 7.15952751614362e-05,
      "loss": 1.0178,
      "step": 10550
    },
    {
      "epoch": 1.6962714034245479,
      "grad_norm": 0.6092725396156311,
      "learning_rate": 7.127986621677263e-05,
      "loss": 1.0112,
      "step": 10600
    },
    {
      "epoch": 1.7042726836293807,
      "grad_norm": 0.6592617034912109,
      "learning_rate": 7.096341975258953e-05,
      "loss": 0.9944,
      "step": 10650
    },
    {
      "epoch": 1.7122739638342135,
      "grad_norm": 0.5574333667755127,
      "learning_rate": 7.064595119752636e-05,
      "loss": 0.9896,
      "step": 10700
    },
    {
      "epoch": 1.7202752440390463,
      "grad_norm": 0.5796050429344177,
      "learning_rate": 7.032747603005552e-05,
      "loss": 1.0099,
      "step": 10750
    },
    {
      "epoch": 1.7282765242438791,
      "grad_norm": 0.5359869599342346,
      "learning_rate": 7.00080097777278e-05,
      "loss": 0.976,
      "step": 10800
    },
    {
      "epoch": 1.736277804448712,
      "grad_norm": 0.533762514591217,
      "learning_rate": 6.968756801641523e-05,
      "loss": 1.011,
      "step": 10850
    },
    {
      "epoch": 1.7442790846535445,
      "grad_norm": 0.6239184141159058,
      "learning_rate": 6.936616636955164e-05,
      "loss": 0.9971,
      "step": 10900
    },
    {
      "epoch": 1.7522803648583773,
      "grad_norm": 0.6219179034233093,
      "learning_rate": 6.904382050737096e-05,
      "loss": 1.0195,
      "step": 10950
    },
    {
      "epoch": 1.7602816450632102,
      "grad_norm": 0.6290085315704346,
      "learning_rate": 6.872054614614328e-05,
      "loss": 1.0393,
      "step": 11000
    },
    {
      "epoch": 1.7682829252680428,
      "grad_norm": 0.6031250953674316,
      "learning_rate": 6.839635904740846e-05,
      "loss": 1.013,
      "step": 11050
    },
    {
      "epoch": 1.7762842054728756,
      "grad_norm": 0.5307977199554443,
      "learning_rate": 6.807127501720773e-05,
      "loss": 0.9924,
      "step": 11100
    },
    {
      "epoch": 1.7842854856777084,
      "grad_norm": 0.593413233757019,
      "learning_rate": 6.774530990531308e-05,
      "loss": 0.9848,
      "step": 11150
    },
    {
      "epoch": 1.7922867658825412,
      "grad_norm": 0.5930994749069214,
      "learning_rate": 6.741847960445437e-05,
      "loss": 0.9891,
      "step": 11200
    },
    {
      "epoch": 1.800288046087374,
      "grad_norm": 0.5843376517295837,
      "learning_rate": 6.709080004954465e-05,
      "loss": 1.0075,
      "step": 11250
    },
    {
      "epoch": 1.8082893262922068,
      "grad_norm": 0.6507195830345154,
      "learning_rate": 6.676228721690301e-05,
      "loss": 0.9872,
      "step": 11300
    },
    {
      "epoch": 1.8162906064970397,
      "grad_norm": 0.502108097076416,
      "learning_rate": 6.64329571234759e-05,
      "loss": 0.9898,
      "step": 11350
    },
    {
      "epoch": 1.8242918867018723,
      "grad_norm": 0.6289730072021484,
      "learning_rate": 6.610282582605597e-05,
      "loss": 0.9886,
      "step": 11400
    },
    {
      "epoch": 1.832293166906705,
      "grad_norm": 0.5781500935554504,
      "learning_rate": 6.577190942049937e-05,
      "loss": 1.0211,
      "step": 11450
    },
    {
      "epoch": 1.840294447111538,
      "grad_norm": 0.6279985308647156,
      "learning_rate": 6.544022404094092e-05,
      "loss": 1.0123,
      "step": 11500
    },
    {
      "epoch": 1.8482957273163705,
      "grad_norm": 0.6612405180931091,
      "learning_rate": 6.51077858590074e-05,
      "loss": 0.9978,
      "step": 11550
    },
    {
      "epoch": 1.8562970075212033,
      "grad_norm": 0.6027297973632812,
      "learning_rate": 6.477461108302927e-05,
      "loss": 0.9786,
      "step": 11600
    },
    {
      "epoch": 1.8642982877260361,
      "grad_norm": 0.6454290151596069,
      "learning_rate": 6.444071595725026e-05,
      "loss": 0.9692,
      "step": 11650
    },
    {
      "epoch": 1.872299567930869,
      "grad_norm": 0.5385527014732361,
      "learning_rate": 6.410611676103542e-05,
      "loss": 0.9789,
      "step": 11700
    },
    {
      "epoch": 1.8803008481357018,
      "grad_norm": 0.5324643850326538,
      "learning_rate": 6.37708298080774e-05,
      "loss": 0.9892,
      "step": 11750
    },
    {
      "epoch": 1.8883021283405346,
      "grad_norm": 0.5693587064743042,
      "learning_rate": 6.343487144560108e-05,
      "loss": 0.9881,
      "step": 11800
    },
    {
      "epoch": 1.8963034085453674,
      "grad_norm": 0.6578769683837891,
      "learning_rate": 6.309825805356654e-05,
      "loss": 0.9871,
      "step": 11850
    },
    {
      "epoch": 1.9043046887502,
      "grad_norm": 0.5471733212471008,
      "learning_rate": 6.276100604387039e-05,
      "loss": 1.0013,
      "step": 11900
    },
    {
      "epoch": 1.9123059689550328,
      "grad_norm": 0.6353397369384766,
      "learning_rate": 6.24231318595457e-05,
      "loss": 0.9653,
      "step": 11950
    },
    {
      "epoch": 1.9203072491598656,
      "grad_norm": 0.5183307528495789,
      "learning_rate": 6.208465197396013e-05,
      "loss": 0.9833,
      "step": 12000
    },
    {
      "epoch": 1.9283085293646982,
      "grad_norm": 0.539452850818634,
      "learning_rate": 6.1745582890013e-05,
      "loss": 0.9686,
      "step": 12050
    },
    {
      "epoch": 1.936309809569531,
      "grad_norm": 0.681061863899231,
      "learning_rate": 6.140594113933042e-05,
      "loss": 0.9814,
      "step": 12100
    },
    {
      "epoch": 1.9443110897743638,
      "grad_norm": 0.6232041716575623,
      "learning_rate": 6.106574328145949e-05,
      "loss": 0.9858,
      "step": 12150
    },
    {
      "epoch": 1.9523123699791967,
      "grad_norm": 0.5435238480567932,
      "learning_rate": 6.072500590306076e-05,
      "loss": 0.9921,
      "step": 12200
    },
    {
      "epoch": 1.9603136501840295,
      "grad_norm": 0.6599912047386169,
      "learning_rate": 6.038374561709964e-05,
      "loss": 0.9627,
      "step": 12250
    },
    {
      "epoch": 1.9683149303888623,
      "grad_norm": 0.6491255760192871,
      "learning_rate": 6.0041979062036356e-05,
      "loss": 1.0074,
      "step": 12300
    },
    {
      "epoch": 1.976316210593695,
      "grad_norm": 0.545872688293457,
      "learning_rate": 5.969972290101478e-05,
      "loss": 0.9769,
      "step": 12350
    },
    {
      "epoch": 1.984317490798528,
      "grad_norm": 0.5526396632194519,
      "learning_rate": 5.935699382104994e-05,
      "loss": 0.9853,
      "step": 12400
    },
    {
      "epoch": 1.9923187710033605,
      "grad_norm": 0.5161740183830261,
      "learning_rate": 5.901380853221448e-05,
      "loss": 1.0122,
      "step": 12450
    },
    {
      "epoch": 2.000320051208193,
      "grad_norm": 0.501007080078125,
      "learning_rate": 5.867018376682396e-05,
      "loss": 1.0062,
      "step": 12500
    },
    {
      "epoch": 2.008321331413026,
      "grad_norm": 0.6925072073936462,
      "learning_rate": 5.832613627862097e-05,
      "loss": 0.8524,
      "step": 12550
    },
    {
      "epoch": 2.0163226116178588,
      "grad_norm": 0.6264834403991699,
      "learning_rate": 5.798168284195837e-05,
      "loss": 0.8474,
      "step": 12600
    },
    {
      "epoch": 2.0243238918226916,
      "grad_norm": 0.543258786201477,
      "learning_rate": 5.7636840250981405e-05,
      "loss": 0.819,
      "step": 12650
    },
    {
      "epoch": 2.0323251720275244,
      "grad_norm": 0.5597353577613831,
      "learning_rate": 5.729162531880892e-05,
      "loss": 0.8635,
      "step": 12700
    },
    {
      "epoch": 2.040326452232357,
      "grad_norm": 0.8554184436798096,
      "learning_rate": 5.694605487671357e-05,
      "loss": 0.8283,
      "step": 12750
    },
    {
      "epoch": 2.04832773243719,
      "grad_norm": 0.5572032332420349,
      "learning_rate": 5.6600145773301206e-05,
      "loss": 0.8744,
      "step": 12800
    },
    {
      "epoch": 2.056329012642023,
      "grad_norm": 0.6553796529769897,
      "learning_rate": 5.625391487368949e-05,
      "loss": 0.8572,
      "step": 12850
    },
    {
      "epoch": 2.0643302928468557,
      "grad_norm": 0.6666269302368164,
      "learning_rate": 5.590737905868551e-05,
      "loss": 0.8625,
      "step": 12900
    },
    {
      "epoch": 2.0723315730516885,
      "grad_norm": 0.6420098543167114,
      "learning_rate": 5.5560555223962784e-05,
      "loss": 0.8571,
      "step": 12950
    },
    {
      "epoch": 2.080332853256521,
      "grad_norm": 0.6100978255271912,
      "learning_rate": 5.5213460279237496e-05,
      "loss": 0.8579,
      "step": 13000
    },
    {
      "epoch": 2.0883341334613537,
      "grad_norm": 0.5606010556221008,
      "learning_rate": 5.4866111147444085e-05,
      "loss": 0.8715,
      "step": 13050
    },
    {
      "epoch": 2.0963354136661865,
      "grad_norm": 0.6860626339912415,
      "learning_rate": 5.4518524763910076e-05,
      "loss": 0.8513,
      "step": 13100
    },
    {
      "epoch": 2.1043366938710193,
      "grad_norm": 0.7012873291969299,
      "learning_rate": 5.4170718075530404e-05,
      "loss": 0.8454,
      "step": 13150
    },
    {
      "epoch": 2.112337974075852,
      "grad_norm": 0.6773386597633362,
      "learning_rate": 5.382270803994126e-05,
      "loss": 0.881,
      "step": 13200
    },
    {
      "epoch": 2.120339254280685,
      "grad_norm": 0.7174808382987976,
      "learning_rate": 5.347451162469309e-05,
      "loss": 0.858,
      "step": 13250
    },
    {
      "epoch": 2.1283405344855177,
      "grad_norm": 0.6617708206176758,
      "learning_rate": 5.3126145806423575e-05,
      "loss": 0.8531,
      "step": 13300
    },
    {
      "epoch": 2.1363418146903506,
      "grad_norm": 0.7019938826560974,
      "learning_rate": 5.277762757002971e-05,
      "loss": 0.8665,
      "step": 13350
    },
    {
      "epoch": 2.1443430948951834,
      "grad_norm": 0.6727155447006226,
      "learning_rate": 5.2428973907839804e-05,
      "loss": 0.8528,
      "step": 13400
    },
    {
      "epoch": 2.152344375100016,
      "grad_norm": 0.6648008227348328,
      "learning_rate": 5.208020181878499e-05,
      "loss": 0.8399,
      "step": 13450
    },
    {
      "epoch": 2.160345655304849,
      "grad_norm": 0.6274258494377136,
      "learning_rate": 5.173132830757039e-05,
      "loss": 0.8662,
      "step": 13500
    },
    {
      "epoch": 2.1683469355096814,
      "grad_norm": 0.6117835640907288,
      "learning_rate": 5.138237038384608e-05,
      "loss": 0.8503,
      "step": 13550
    },
    {
      "epoch": 2.176348215714514,
      "grad_norm": 0.6691561341285706,
      "learning_rate": 5.103334506137772e-05,
      "loss": 0.8645,
      "step": 13600
    },
    {
      "epoch": 2.184349495919347,
      "grad_norm": 0.5877872109413147,
      "learning_rate": 5.0684269357217084e-05,
      "loss": 0.8555,
      "step": 13650
    },
    {
      "epoch": 2.19235077612418,
      "grad_norm": 0.717145562171936,
      "learning_rate": 5.033516029087231e-05,
      "loss": 0.8799,
      "step": 13700
    },
    {
      "epoch": 2.2003520563290127,
      "grad_norm": 0.6587306261062622,
      "learning_rate": 4.99860348834782e-05,
      "loss": 0.88,
      "step": 13750
    },
    {
      "epoch": 2.2083533365338455,
      "grad_norm": 0.6128435134887695,
      "learning_rate": 4.963691015696624e-05,
      "loss": 0.853,
      "step": 13800
    },
    {
      "epoch": 2.2163546167386783,
      "grad_norm": 0.5783178210258484,
      "learning_rate": 4.928780313323473e-05,
      "loss": 0.8863,
      "step": 13850
    },
    {
      "epoch": 2.224355896943511,
      "grad_norm": 0.6364635825157166,
      "learning_rate": 4.893873083331882e-05,
      "loss": 0.8378,
      "step": 13900
    },
    {
      "epoch": 2.232357177148344,
      "grad_norm": 0.7266676425933838,
      "learning_rate": 4.8589710276560744e-05,
      "loss": 0.8647,
      "step": 13950
    },
    {
      "epoch": 2.2403584573531763,
      "grad_norm": 0.6661261916160583,
      "learning_rate": 4.8240758479779884e-05,
      "loss": 0.835,
      "step": 14000
    },
    {
      "epoch": 2.248359737558009,
      "grad_norm": 0.7814577221870422,
      "learning_rate": 4.789189245644318e-05,
      "loss": 0.8379,
      "step": 14050
    },
    {
      "epoch": 2.256361017762842,
      "grad_norm": 0.6463228464126587,
      "learning_rate": 4.754312921583564e-05,
      "loss": 0.8628,
      "step": 14100
    },
    {
      "epoch": 2.2643622979676747,
      "grad_norm": 0.7221981883049011,
      "learning_rate": 4.719448576223096e-05,
      "loss": 0.8565,
      "step": 14150
    },
    {
      "epoch": 2.2723635781725076,
      "grad_norm": 0.6834704279899597,
      "learning_rate": 4.684597909406253e-05,
      "loss": 0.8776,
      "step": 14200
    },
    {
      "epoch": 2.2803648583773404,
      "grad_norm": 0.6285929083824158,
      "learning_rate": 4.649762620309466e-05,
      "loss": 0.8356,
      "step": 14250
    },
    {
      "epoch": 2.288366138582173,
      "grad_norm": 0.6223785281181335,
      "learning_rate": 4.61494440735941e-05,
      "loss": 0.834,
      "step": 14300
    },
    {
      "epoch": 2.296367418787006,
      "grad_norm": 0.7635972499847412,
      "learning_rate": 4.580144968150192e-05,
      "loss": 0.8603,
      "step": 14350
    },
    {
      "epoch": 2.304368698991839,
      "grad_norm": 0.6658472418785095,
      "learning_rate": 4.5453659993605944e-05,
      "loss": 0.8393,
      "step": 14400
    },
    {
      "epoch": 2.3123699791966716,
      "grad_norm": 0.6321608424186707,
      "learning_rate": 4.510609196671345e-05,
      "loss": 0.8416,
      "step": 14450
    },
    {
      "epoch": 2.3203712594015045,
      "grad_norm": 0.6526052355766296,
      "learning_rate": 4.4758762546824364e-05,
      "loss": 0.8696,
      "step": 14500
    },
    {
      "epoch": 2.328372539606337,
      "grad_norm": 0.6651396751403809,
      "learning_rate": 4.441168866830516e-05,
      "loss": 0.8619,
      "step": 14550
    },
    {
      "epoch": 2.3363738198111696,
      "grad_norm": 0.6737195253372192,
      "learning_rate": 4.406488725306317e-05,
      "loss": 0.8447,
      "step": 14600
    },
    {
      "epoch": 2.3443751000160025,
      "grad_norm": 0.5813534259796143,
      "learning_rate": 4.3718375209721505e-05,
      "loss": 0.8468,
      "step": 14650
    },
    {
      "epoch": 2.3523763802208353,
      "grad_norm": 0.7838544249534607,
      "learning_rate": 4.337216943279464e-05,
      "loss": 0.8344,
      "step": 14700
    },
    {
      "epoch": 2.360377660425668,
      "grad_norm": 0.6356272101402283,
      "learning_rate": 4.3026286801864854e-05,
      "loss": 0.8699,
      "step": 14750
    },
    {
      "epoch": 2.368378940630501,
      "grad_norm": 0.6201728582382202,
      "learning_rate": 4.268074418075903e-05,
      "loss": 0.8515,
      "step": 14800
    },
    {
      "epoch": 2.3763802208353337,
      "grad_norm": 0.6424387693405151,
      "learning_rate": 4.233555841672663e-05,
      "loss": 0.8429,
      "step": 14850
    },
    {
      "epoch": 2.3843815010401666,
      "grad_norm": 0.6518192291259766,
      "learning_rate": 4.1990746339618216e-05,
      "loss": 0.8666,
      "step": 14900
    },
    {
      "epoch": 2.3923827812449994,
      "grad_norm": 0.7236935496330261,
      "learning_rate": 4.164632476106484e-05,
      "loss": 0.8478,
      "step": 14950
    },
    {
      "epoch": 2.4003840614498317,
      "grad_norm": 0.692077100276947,
      "learning_rate": 4.13023104736585e-05,
      "loss": 0.8756,
      "step": 15000
    },
    {
      "epoch": 2.4083853416546646,
      "grad_norm": 0.6063551306724548,
      "learning_rate": 4.095872025013333e-05,
      "loss": 0.8221,
      "step": 15050
    },
    {
      "epoch": 2.4163866218594974,
      "grad_norm": 0.5967947840690613,
      "learning_rate": 4.0615570842547815e-05,
      "loss": 0.8731,
      "step": 15100
    },
    {
      "epoch": 2.42438790206433,
      "grad_norm": 0.7256248593330383,
      "learning_rate": 4.0272878981468045e-05,
      "loss": 0.8614,
      "step": 15150
    },
    {
      "epoch": 2.432389182269163,
      "grad_norm": 0.6148253083229065,
      "learning_rate": 3.9930661375152045e-05,
      "loss": 0.8337,
      "step": 15200
    },
    {
      "epoch": 2.440390462473996,
      "grad_norm": 0.6977733373641968,
      "learning_rate": 3.958893470873509e-05,
      "loss": 0.8235,
      "step": 15250
    },
    {
      "epoch": 2.4483917426788286,
      "grad_norm": 0.6713569164276123,
      "learning_rate": 3.924771564341621e-05,
      "loss": 0.8301,
      "step": 15300
    },
    {
      "epoch": 2.4563930228836615,
      "grad_norm": 0.637150228023529,
      "learning_rate": 3.890702081564593e-05,
      "loss": 0.8355,
      "step": 15350
    },
    {
      "epoch": 2.4643943030884943,
      "grad_norm": 0.6515012383460999,
      "learning_rate": 3.856686683631502e-05,
      "loss": 0.8547,
      "step": 15400
    },
    {
      "epoch": 2.472395583293327,
      "grad_norm": 0.6041115522384644,
      "learning_rate": 3.822727028994471e-05,
      "loss": 0.8421,
      "step": 15450
    },
    {
      "epoch": 2.48039686349816,
      "grad_norm": 0.6472122073173523,
      "learning_rate": 3.788824773387811e-05,
      "loss": 0.838,
      "step": 15500
    },
    {
      "epoch": 2.4883981437029927,
      "grad_norm": 0.7198679447174072,
      "learning_rate": 3.754981569747287e-05,
      "loss": 0.8213,
      "step": 15550
    },
    {
      "epoch": 2.496399423907825,
      "grad_norm": 0.6885024309158325,
      "learning_rate": 3.721199068129527e-05,
      "loss": 0.8354,
      "step": 15600
    },
    {
      "epoch": 2.504400704112658,
      "grad_norm": 0.6485710740089417,
      "learning_rate": 3.6874789156315835e-05,
      "loss": 0.8633,
      "step": 15650
    },
    {
      "epoch": 2.5124019843174907,
      "grad_norm": 0.6844043135643005,
      "learning_rate": 3.6538227563106164e-05,
      "loss": 0.8105,
      "step": 15700
    },
    {
      "epoch": 2.5204032645223235,
      "grad_norm": 0.8702924847602844,
      "learning_rate": 3.6202322311037374e-05,
      "loss": 0.8121,
      "step": 15750
    },
    {
      "epoch": 2.5284045447271564,
      "grad_norm": 0.7209028005599976,
      "learning_rate": 3.586708977748012e-05,
      "loss": 0.8758,
      "step": 15800
    },
    {
      "epoch": 2.536405824931989,
      "grad_norm": 0.6202672123908997,
      "learning_rate": 3.553254630700597e-05,
      "loss": 0.8474,
      "step": 15850
    },
    {
      "epoch": 2.544407105136822,
      "grad_norm": 0.6669834852218628,
      "learning_rate": 3.5198708210590626e-05,
      "loss": 0.8482,
      "step": 15900
    },
    {
      "epoch": 2.552408385341655,
      "grad_norm": 0.6584005951881409,
      "learning_rate": 3.486559176481865e-05,
      "loss": 0.8416,
      "step": 15950
    },
    {
      "epoch": 2.560409665546487,
      "grad_norm": 0.5252745747566223,
      "learning_rate": 3.4533213211089825e-05,
      "loss": 0.8401,
      "step": 16000
    },
    {
      "epoch": 2.56841094575132,
      "grad_norm": 0.6504681706428528,
      "learning_rate": 3.4201588754827314e-05,
      "loss": 0.8486,
      "step": 16050
    },
    {
      "epoch": 2.576412225956153,
      "grad_norm": 0.7693225741386414,
      "learning_rate": 3.387073456468761e-05,
      "loss": 0.845,
      "step": 16100
    },
    {
      "epoch": 2.5844135061609856,
      "grad_norm": 0.6965412497520447,
      "learning_rate": 3.354066677177214e-05,
      "loss": 0.8568,
      "step": 16150
    },
    {
      "epoch": 2.5924147863658185,
      "grad_norm": 0.6364871859550476,
      "learning_rate": 3.321140146884081e-05,
      "loss": 0.8693,
      "step": 16200
    },
    {
      "epoch": 2.6004160665706513,
      "grad_norm": 0.7464070916175842,
      "learning_rate": 3.288295470952737e-05,
      "loss": 0.8566,
      "step": 16250
    },
    {
      "epoch": 2.608417346775484,
      "grad_norm": 0.6804706454277039,
      "learning_rate": 3.255534250755678e-05,
      "loss": 0.8708,
      "step": 16300
    },
    {
      "epoch": 2.616418626980317,
      "grad_norm": 0.6775168776512146,
      "learning_rate": 3.222858083596431e-05,
      "loss": 0.8558,
      "step": 16350
    },
    {
      "epoch": 2.6244199071851497,
      "grad_norm": 0.7037110924720764,
      "learning_rate": 3.190268562631688e-05,
      "loss": 0.8313,
      "step": 16400
    },
    {
      "epoch": 2.6324211873899825,
      "grad_norm": 0.5500233173370361,
      "learning_rate": 3.157767276793626e-05,
      "loss": 0.7998,
      "step": 16450
    },
    {
      "epoch": 2.6404224675948154,
      "grad_norm": 0.7002319097518921,
      "learning_rate": 3.125355810712435e-05,
      "loss": 0.8247,
      "step": 16500
    },
    {
      "epoch": 2.648423747799648,
      "grad_norm": 0.7857663631439209,
      "learning_rate": 3.093035744639061e-05,
      "loss": 0.8318,
      "step": 16550
    },
    {
      "epoch": 2.656425028004481,
      "grad_norm": 0.7866588830947876,
      "learning_rate": 3.06080865436816e-05,
      "loss": 0.814,
      "step": 16600
    },
    {
      "epoch": 2.6644263082093134,
      "grad_norm": 0.6732119917869568,
      "learning_rate": 3.0286761111612626e-05,
      "loss": 0.8527,
      "step": 16650
    },
    {
      "epoch": 2.672427588414146,
      "grad_norm": 0.6847143769264221,
      "learning_rate": 2.9966396816701725e-05,
      "loss": 0.8012,
      "step": 16700
    },
    {
      "epoch": 2.680428868618979,
      "grad_norm": 0.6862397193908691,
      "learning_rate": 2.964700927860581e-05,
      "loss": 0.817,
      "step": 16750
    },
    {
      "epoch": 2.688430148823812,
      "grad_norm": 0.6373656988143921,
      "learning_rate": 2.9328614069359128e-05,
      "loss": 0.8625,
      "step": 16800
    },
    {
      "epoch": 2.6964314290286446,
      "grad_norm": 0.7099259495735168,
      "learning_rate": 2.9011226712613937e-05,
      "loss": 0.8445,
      "step": 16850
    },
    {
      "epoch": 2.7044327092334774,
      "grad_norm": 0.6120832562446594,
      "learning_rate": 2.8694862682883866e-05,
      "loss": 0.7982,
      "step": 16900
    },
    {
      "epoch": 2.7124339894383103,
      "grad_norm": 0.8442961573600769,
      "learning_rate": 2.8379537404789124e-05,
      "loss": 0.8307,
      "step": 16950
    },
    {
      "epoch": 2.7204352696431426,
      "grad_norm": 0.6093111634254456,
      "learning_rate": 2.8065266252304712e-05,
      "loss": 0.7811,
      "step": 17000
    },
    {
      "epoch": 2.7284365498479755,
      "grad_norm": 0.6981213092803955,
      "learning_rate": 2.775206454801079e-05,
      "loss": 0.8533,
      "step": 17050
    },
    {
      "epoch": 2.7364378300528083,
      "grad_norm": 0.696796178817749,
      "learning_rate": 2.7439947562345546e-05,
      "loss": 0.8182,
      "step": 17100
    },
    {
      "epoch": 2.744439110257641,
      "grad_norm": 0.6803203225135803,
      "learning_rate": 2.7128930512860658e-05,
      "loss": 0.8391,
      "step": 17150
    },
    {
      "epoch": 2.752440390462474,
      "grad_norm": 0.6710325479507446,
      "learning_rate": 2.6819028563479505e-05,
      "loss": 0.8229,
      "step": 17200
    },
    {
      "epoch": 2.7604416706673067,
      "grad_norm": 0.7354364395141602,
      "learning_rate": 2.6510256823757667e-05,
      "loss": 0.8121,
      "step": 17250
    },
    {
      "epoch": 2.7684429508721395,
      "grad_norm": 0.7637900710105896,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.84,
      "step": 17300
    },
    {
      "epoch": 2.7764442310769724,
      "grad_norm": 0.587841808795929,
      "learning_rate": 2.589616413525824e-05,
      "loss": 0.8117,
      "step": 17350
    },
    {
      "epoch": 2.784445511281805,
      "grad_norm": 0.667797327041626,
      "learning_rate": 2.559087312713651e-05,
      "loss": 0.8131,
      "step": 17400
    },
    {
      "epoch": 2.792446791486638,
      "grad_norm": 0.5983754992485046,
      "learning_rate": 2.5286772208526027e-05,
      "loss": 0.8309,
      "step": 17450
    },
    {
      "epoch": 2.800448071691471,
      "grad_norm": 0.7884683012962341,
      "learning_rate": 2.4983876206147776e-05,
      "loss": 0.8208,
      "step": 17500
    },
    {
      "epoch": 2.8084493518963036,
      "grad_norm": 0.6957002878189087,
      "learning_rate": 2.4682199887975938e-05,
      "loss": 0.8699,
      "step": 17550
    },
    {
      "epoch": 2.8164506321011364,
      "grad_norm": 0.5889870524406433,
      "learning_rate": 2.438175796251786e-05,
      "loss": 0.8282,
      "step": 17600
    },
    {
      "epoch": 2.824451912305969,
      "grad_norm": 0.7357444167137146,
      "learning_rate": 2.4082565078096935e-05,
      "loss": 0.8025,
      "step": 17650
    },
    {
      "epoch": 2.8324531925108016,
      "grad_norm": 0.8791075348854065,
      "learning_rate": 2.3784635822138424e-05,
      "loss": 0.8338,
      "step": 17700
    },
    {
      "epoch": 2.8404544727156344,
      "grad_norm": 0.6486666202545166,
      "learning_rate": 2.348798472045819e-05,
      "loss": 0.8221,
      "step": 17750
    },
    {
      "epoch": 2.8484557529204673,
      "grad_norm": 0.6927144527435303,
      "learning_rate": 2.3192626236554516e-05,
      "loss": 0.8249,
      "step": 17800
    },
    {
      "epoch": 2.8564570331253,
      "grad_norm": 0.7874731421470642,
      "learning_rate": 2.2898574770902914e-05,
      "loss": 0.8267,
      "step": 17850
    },
    {
      "epoch": 2.864458313330133,
      "grad_norm": 0.6948744654655457,
      "learning_rate": 2.260584466025401e-05,
      "loss": 0.8139,
      "step": 17900
    },
    {
      "epoch": 2.8724595935349657,
      "grad_norm": 0.6847252249717712,
      "learning_rate": 2.231445017693454e-05,
      "loss": 0.8432,
      "step": 17950
    },
    {
      "epoch": 2.8804608737397985,
      "grad_norm": 0.6781948804855347,
      "learning_rate": 2.2024405528151547e-05,
      "loss": 0.8298,
      "step": 18000
    },
    {
      "epoch": 2.888462153944631,
      "grad_norm": 0.6336878538131714,
      "learning_rate": 2.1735724855299567e-05,
      "loss": 0.81,
      "step": 18050
    },
    {
      "epoch": 2.8964634341494637,
      "grad_norm": 0.6111786365509033,
      "learning_rate": 2.1448422233271254e-05,
      "loss": 0.7977,
      "step": 18100
    },
    {
      "epoch": 2.9044647143542965,
      "grad_norm": 0.7334992289543152,
      "learning_rate": 2.116251166977118e-05,
      "loss": 0.8221,
      "step": 18150
    },
    {
      "epoch": 2.9124659945591294,
      "grad_norm": 0.8408239483833313,
      "learning_rate": 2.0878007104632775e-05,
      "loss": 0.8597,
      "step": 18200
    },
    {
      "epoch": 2.920467274763962,
      "grad_norm": 0.6892300248146057,
      "learning_rate": 2.059492240913866e-05,
      "loss": 0.8088,
      "step": 18250
    },
    {
      "epoch": 2.928468554968795,
      "grad_norm": 0.5887216925621033,
      "learning_rate": 2.0313271385344522e-05,
      "loss": 0.8207,
      "step": 18300
    },
    {
      "epoch": 2.936469835173628,
      "grad_norm": 0.6477887630462646,
      "learning_rate": 2.0033067765406004e-05,
      "loss": 0.839,
      "step": 18350
    },
    {
      "epoch": 2.9444711153784606,
      "grad_norm": 0.7348781228065491,
      "learning_rate": 1.9754325210909174e-05,
      "loss": 0.8412,
      "step": 18400
    },
    {
      "epoch": 2.9524723955832934,
      "grad_norm": 0.6994746923446655,
      "learning_rate": 1.947705731220462e-05,
      "loss": 0.8069,
      "step": 18450
    },
    {
      "epoch": 2.9604736757881263,
      "grad_norm": 0.6778689026832581,
      "learning_rate": 1.920127758774466e-05,
      "loss": 0.8119,
      "step": 18500
    },
    {
      "epoch": 2.968474955992959,
      "grad_norm": 0.7225350737571716,
      "learning_rate": 1.8926999483424313e-05,
      "loss": 0.8504,
      "step": 18550
    },
    {
      "epoch": 2.976476236197792,
      "grad_norm": 0.8543927073478699,
      "learning_rate": 1.8654236371925727e-05,
      "loss": 0.8124,
      "step": 18600
    },
    {
      "epoch": 2.9844775164026243,
      "grad_norm": 0.6733004450798035,
      "learning_rate": 1.8383001552066164e-05,
      "loss": 0.8299,
      "step": 18650
    },
    {
      "epoch": 2.992478796607457,
      "grad_norm": 0.7595524787902832,
      "learning_rate": 1.8113308248149635e-05,
      "loss": 0.8177,
      "step": 18700
    },
    {
      "epoch": 3.00048007681229,
      "grad_norm": 0.6430355310440063,
      "learning_rate": 1.784516960932211e-05,
      "loss": 0.8311,
      "step": 18750
    },
    {
      "epoch": 3.0084813570171227,
      "grad_norm": 0.6314593553543091,
      "learning_rate": 1.7578598708930428e-05,
      "loss": 0.7241,
      "step": 18800
    },
    {
      "epoch": 3.0164826372219555,
      "grad_norm": 0.6709548234939575,
      "learning_rate": 1.7313608543884868e-05,
      "loss": 0.7459,
      "step": 18850
    },
    {
      "epoch": 3.0244839174267883,
      "grad_norm": 0.6981055736541748,
      "learning_rate": 1.7050212034025575e-05,
      "loss": 0.7399,
      "step": 18900
    },
    {
      "epoch": 3.032485197631621,
      "grad_norm": 0.7344752550125122,
      "learning_rate": 1.6788422021492472e-05,
      "loss": 0.7197,
      "step": 18950
    },
    {
      "epoch": 3.040486477836454,
      "grad_norm": 0.7181177139282227,
      "learning_rate": 1.6528251270099256e-05,
      "loss": 0.7161,
      "step": 19000
    },
    {
      "epoch": 3.048487758041287,
      "grad_norm": 0.5588538646697998,
      "learning_rate": 1.6269712464711105e-05,
      "loss": 0.7488,
      "step": 19050
    },
    {
      "epoch": 3.056489038246119,
      "grad_norm": 0.7480596303939819,
      "learning_rate": 1.6012818210626075e-05,
      "loss": 0.712,
      "step": 19100
    },
    {
      "epoch": 3.064490318450952,
      "grad_norm": 0.5872476696968079,
      "learning_rate": 1.5757581032960638e-05,
      "loss": 0.7154,
      "step": 19150
    },
    {
      "epoch": 3.072491598655785,
      "grad_norm": 0.6517592072486877,
      "learning_rate": 1.550401337603902e-05,
      "loss": 0.7134,
      "step": 19200
    },
    {
      "epoch": 3.0804928788606176,
      "grad_norm": 0.6740624308586121,
      "learning_rate": 1.5252127602786397e-05,
      "loss": 0.7446,
      "step": 19250
    },
    {
      "epoch": 3.0884941590654504,
      "grad_norm": 0.6558243632316589,
      "learning_rate": 1.5001935994126104e-05,
      "loss": 0.7354,
      "step": 19300
    },
    {
      "epoch": 3.0964954392702833,
      "grad_norm": 0.6562448740005493,
      "learning_rate": 1.4753450748380953e-05,
      "loss": 0.7035,
      "step": 19350
    },
    {
      "epoch": 3.104496719475116,
      "grad_norm": 0.7197704315185547,
      "learning_rate": 1.4506683980678503e-05,
      "loss": 0.7313,
      "step": 19400
    },
    {
      "epoch": 3.112497999679949,
      "grad_norm": 0.7322149872779846,
      "learning_rate": 1.4261647722360278e-05,
      "loss": 0.7357,
      "step": 19450
    },
    {
      "epoch": 3.1204992798847817,
      "grad_norm": 0.7477754354476929,
      "learning_rate": 1.4018353920395193e-05,
      "loss": 0.7334,
      "step": 19500
    },
    {
      "epoch": 3.1285005600896145,
      "grad_norm": 0.5789663791656494,
      "learning_rate": 1.3776814436797176e-05,
      "loss": 0.7435,
      "step": 19550
    },
    {
      "epoch": 3.1365018402944473,
      "grad_norm": 0.6711790561676025,
      "learning_rate": 1.3537041048046695e-05,
      "loss": 0.713,
      "step": 19600
    },
    {
      "epoch": 3.1445031204992797,
      "grad_norm": 0.6365526914596558,
      "learning_rate": 1.32990454445166e-05,
      "loss": 0.7679,
      "step": 19650
    },
    {
      "epoch": 3.1525044007041125,
      "grad_norm": 0.6711440682411194,
      "learning_rate": 1.3062839229902263e-05,
      "loss": 0.7199,
      "step": 19700
    },
    {
      "epoch": 3.1605056809089453,
      "grad_norm": 0.7575156688690186,
      "learning_rate": 1.2828433920655687e-05,
      "loss": 0.7052,
      "step": 19750
    },
    {
      "epoch": 3.168506961113778,
      "grad_norm": 0.6248161792755127,
      "learning_rate": 1.2595840945424093e-05,
      "loss": 0.7129,
      "step": 19800
    },
    {
      "epoch": 3.176508241318611,
      "grad_norm": 0.7342621684074402,
      "learning_rate": 1.2365071644492682e-05,
      "loss": 0.7187,
      "step": 19850
    },
    {
      "epoch": 3.184509521523444,
      "grad_norm": 0.6555837988853455,
      "learning_rate": 1.2136137269231723e-05,
      "loss": 0.7378,
      "step": 19900
    },
    {
      "epoch": 3.1925108017282766,
      "grad_norm": 0.7092923521995544,
      "learning_rate": 1.1909048981547998e-05,
      "loss": 0.7259,
      "step": 19950
    },
    {
      "epoch": 3.2005120819331094,
      "grad_norm": 0.6612015962600708,
      "learning_rate": 1.1683817853340595e-05,
      "loss": 0.7168,
      "step": 20000
    },
    {
      "epoch": 3.2085133621379422,
      "grad_norm": 0.7098072171211243,
      "learning_rate": 1.1460454865961056e-05,
      "loss": 0.7226,
      "step": 20050
    },
    {
      "epoch": 3.2165146423427746,
      "grad_norm": 0.7153401970863342,
      "learning_rate": 1.1238970909677993e-05,
      "loss": 0.7304,
      "step": 20100
    },
    {
      "epoch": 3.2245159225476074,
      "grad_norm": 0.6184985637664795,
      "learning_rate": 1.1019376783146174e-05,
      "loss": 0.7113,
      "step": 20150
    },
    {
      "epoch": 3.2325172027524403,
      "grad_norm": 0.658746063709259,
      "learning_rate": 1.080168319287989e-05,
      "loss": 0.7545,
      "step": 20200
    },
    {
      "epoch": 3.240518482957273,
      "grad_norm": 0.6574704647064209,
      "learning_rate": 1.0585900752731077e-05,
      "loss": 0.7103,
      "step": 20250
    },
    {
      "epoch": 3.248519763162106,
      "grad_norm": 0.7436890006065369,
      "learning_rate": 1.0372039983371818e-05,
      "loss": 0.7335,
      "step": 20300
    },
    {
      "epoch": 3.2565210433669387,
      "grad_norm": 0.7010487914085388,
      "learning_rate": 1.0160111311781284e-05,
      "loss": 0.7463,
      "step": 20350
    },
    {
      "epoch": 3.2645223235717715,
      "grad_norm": 0.7422741651535034,
      "learning_rate": 9.950125070737476e-06,
      "loss": 0.7253,
      "step": 20400
    },
    {
      "epoch": 3.2725236037766043,
      "grad_norm": 0.6463279724121094,
      "learning_rate": 9.742091498313426e-06,
      "loss": 0.7334,
      "step": 20450
    },
    {
      "epoch": 3.280524883981437,
      "grad_norm": 0.7407013177871704,
      "learning_rate": 9.536020737377993e-06,
      "loss": 0.7104,
      "step": 20500
    },
    {
      "epoch": 3.28852616418627,
      "grad_norm": 0.6367645263671875,
      "learning_rate": 9.331922835101282e-06,
      "loss": 0.7105,
      "step": 20550
    },
    {
      "epoch": 3.296527444391103,
      "grad_norm": 0.7848844528198242,
      "learning_rate": 9.129807742464957e-06,
      "loss": 0.7155,
      "step": 20600
    },
    {
      "epoch": 3.304528724595935,
      "grad_norm": 0.6976688504219055,
      "learning_rate": 8.929685313776903e-06,
      "loss": 0.7363,
      "step": 20650
    },
    {
      "epoch": 3.312530004800768,
      "grad_norm": 0.7478333711624146,
      "learning_rate": 8.731565306190852e-06,
      "loss": 0.7612,
      "step": 20700
    },
    {
      "epoch": 3.320531285005601,
      "grad_norm": 0.6891793012619019,
      "learning_rate": 8.535457379230649e-06,
      "loss": 0.7221,
      "step": 20750
    },
    {
      "epoch": 3.3285325652104336,
      "grad_norm": 0.6285429000854492,
      "learning_rate": 8.341371094319289e-06,
      "loss": 0.7554,
      "step": 20800
    },
    {
      "epoch": 3.3365338454152664,
      "grad_norm": 0.7216114401817322,
      "learning_rate": 8.149315914312733e-06,
      "loss": 0.7297,
      "step": 20850
    },
    {
      "epoch": 3.3445351256200992,
      "grad_norm": 0.6681531071662903,
      "learning_rate": 7.959301203038566e-06,
      "loss": 0.7283,
      "step": 20900
    },
    {
      "epoch": 3.352536405824932,
      "grad_norm": 0.6501981019973755,
      "learning_rate": 7.771336224839425e-06,
      "loss": 0.723,
      "step": 20950
    },
    {
      "epoch": 3.360537686029765,
      "grad_norm": 0.7048876285552979,
      "learning_rate": 7.585430144121319e-06,
      "loss": 0.7239,
      "step": 21000
    },
    {
      "epoch": 3.3685389662345977,
      "grad_norm": 0.6659078001976013,
      "learning_rate": 7.401592024906812e-06,
      "loss": 0.7417,
      "step": 21050
    },
    {
      "epoch": 3.37654024643943,
      "grad_norm": 0.632502019405365,
      "learning_rate": 7.219830830393093e-06,
      "loss": 0.7115,
      "step": 21100
    },
    {
      "epoch": 3.384541526644263,
      "grad_norm": 0.6977764964103699,
      "learning_rate": 7.040155422514977e-06,
      "loss": 0.6854,
      "step": 21150
    },
    {
      "epoch": 3.3925428068490957,
      "grad_norm": 0.6409167051315308,
      "learning_rate": 6.862574561512825e-06,
      "loss": 0.7016,
      "step": 21200
    },
    {
      "epoch": 3.4005440870539285,
      "grad_norm": 0.6645964980125427,
      "learning_rate": 6.6870969055054246e-06,
      "loss": 0.7224,
      "step": 21250
    },
    {
      "epoch": 3.4085453672587613,
      "grad_norm": 0.6525962352752686,
      "learning_rate": 6.51373101006787e-06,
      "loss": 0.7248,
      "step": 21300
    },
    {
      "epoch": 3.416546647463594,
      "grad_norm": 0.7221346497535706,
      "learning_rate": 6.3424853278144015e-06,
      "loss": 0.7159,
      "step": 21350
    },
    {
      "epoch": 3.424547927668427,
      "grad_norm": 0.7742597460746765,
      "learning_rate": 6.173368207986358e-06,
      "loss": 0.7158,
      "step": 21400
    },
    {
      "epoch": 3.43254920787326,
      "grad_norm": 0.7201706171035767,
      "learning_rate": 6.00638789604499e-06,
      "loss": 0.7351,
      "step": 21450
    },
    {
      "epoch": 3.4405504880780926,
      "grad_norm": 0.6694920063018799,
      "learning_rate": 5.841552533269534e-06,
      "loss": 0.7313,
      "step": 21500
    },
    {
      "epoch": 3.4485517682829254,
      "grad_norm": 0.773014485836029,
      "learning_rate": 5.6788701563602665e-06,
      "loss": 0.7526,
      "step": 21550
    },
    {
      "epoch": 3.4565530484877582,
      "grad_norm": 0.602942168712616,
      "learning_rate": 5.518348697046644e-06,
      "loss": 0.7304,
      "step": 21600
    },
    {
      "epoch": 3.464554328692591,
      "grad_norm": 0.7288883924484253,
      "learning_rate": 5.359995981700544e-06,
      "loss": 0.7086,
      "step": 21650
    },
    {
      "epoch": 3.4725556088974234,
      "grad_norm": 0.8319910168647766,
      "learning_rate": 5.203819730954806e-06,
      "loss": 0.7356,
      "step": 21700
    },
    {
      "epoch": 3.4805568891022562,
      "grad_norm": 0.7911546230316162,
      "learning_rate": 5.049827559326686e-06,
      "loss": 0.7192,
      "step": 21750
    },
    {
      "epoch": 3.488558169307089,
      "grad_norm": 0.704937756061554,
      "learning_rate": 4.898026974846631e-06,
      "loss": 0.7209,
      "step": 21800
    },
    {
      "epoch": 3.496559449511922,
      "grad_norm": 0.7878553867340088,
      "learning_rate": 4.748425378692278e-06,
      "loss": 0.7449,
      "step": 21850
    },
    {
      "epoch": 3.5045607297167547,
      "grad_norm": 0.7807219624519348,
      "learning_rate": 4.601030064827527e-06,
      "loss": 0.722,
      "step": 21900
    },
    {
      "epoch": 3.5125620099215875,
      "grad_norm": 0.7362565994262695,
      "learning_rate": 4.455848219646957e-06,
      "loss": 0.7069,
      "step": 21950
    },
    {
      "epoch": 3.5205632901264203,
      "grad_norm": 0.6918688416481018,
      "learning_rate": 4.3128869216254366e-06,
      "loss": 0.7198,
      "step": 22000
    },
    {
      "epoch": 3.528564570331253,
      "grad_norm": 0.6748279929161072,
      "learning_rate": 4.172153140973012e-06,
      "loss": 0.7429,
      "step": 22050
    },
    {
      "epoch": 3.5365658505360855,
      "grad_norm": 0.7261555194854736,
      "learning_rate": 4.033653739295062e-06,
      "loss": 0.7453,
      "step": 22100
    },
    {
      "epoch": 3.5445671307409183,
      "grad_norm": 0.7590264081954956,
      "learning_rate": 3.897395469257759e-06,
      "loss": 0.7435,
      "step": 22150
    },
    {
      "epoch": 3.552568410945751,
      "grad_norm": 0.6835177540779114,
      "learning_rate": 3.7633849742588213e-06,
      "loss": 0.7305,
      "step": 22200
    },
    {
      "epoch": 3.560569691150584,
      "grad_norm": 0.7218915224075317,
      "learning_rate": 3.6316287881036306e-06,
      "loss": 0.7357,
      "step": 22250
    },
    {
      "epoch": 3.568570971355417,
      "grad_norm": 0.7100145220756531,
      "learning_rate": 3.502133334686669e-06,
      "loss": 0.7244,
      "step": 22300
    },
    {
      "epoch": 3.5765722515602496,
      "grad_norm": 0.6982618570327759,
      "learning_rate": 3.374904927678285e-06,
      "loss": 0.7172,
      "step": 22350
    },
    {
      "epoch": 3.5845735317650824,
      "grad_norm": 0.6808996200561523,
      "learning_rate": 3.2499497702169035e-06,
      "loss": 0.7112,
      "step": 22400
    },
    {
      "epoch": 3.5925748119699152,
      "grad_norm": 0.6580341458320618,
      "learning_rate": 3.1272739546065746e-06,
      "loss": 0.7375,
      "step": 22450
    },
    {
      "epoch": 3.600576092174748,
      "grad_norm": 0.7925476431846619,
      "learning_rate": 3.0068834620199103e-06,
      "loss": 0.7351,
      "step": 22500
    },
    {
      "epoch": 3.608577372379581,
      "grad_norm": 0.6711053848266602,
      "learning_rate": 2.888784162206504e-06,
      "loss": 0.7281,
      "step": 22550
    },
    {
      "epoch": 3.6165786525844137,
      "grad_norm": 0.6397190690040588,
      "learning_rate": 2.7729818132067264e-06,
      "loss": 0.7277,
      "step": 22600
    },
    {
      "epoch": 3.6245799327892465,
      "grad_norm": 0.6673882007598877,
      "learning_rate": 2.6594820610710057e-06,
      "loss": 0.7045,
      "step": 22650
    },
    {
      "epoch": 3.6325812129940793,
      "grad_norm": 0.6923602223396301,
      "learning_rate": 2.5482904395845107e-06,
      "loss": 0.7435,
      "step": 22700
    },
    {
      "epoch": 3.6405824931989117,
      "grad_norm": 0.7044751048088074,
      "learning_rate": 2.439412369997374e-06,
      "loss": 0.7135,
      "step": 22750
    },
    {
      "epoch": 3.6485837734037445,
      "grad_norm": 0.7120568156242371,
      "learning_rate": 2.33285316076039e-06,
      "loss": 0.7115,
      "step": 22800
    },
    {
      "epoch": 3.6565850536085773,
      "grad_norm": 0.6994568705558777,
      "learning_rate": 2.22861800726617e-06,
      "loss": 0.7212,
      "step": 22850
    },
    {
      "epoch": 3.66458633381341,
      "grad_norm": 0.7410824298858643,
      "learning_rate": 2.1267119915958088e-06,
      "loss": 0.6973,
      "step": 22900
    },
    {
      "epoch": 3.672587614018243,
      "grad_norm": 0.7019298076629639,
      "learning_rate": 2.0271400822711894e-06,
      "loss": 0.724,
      "step": 22950
    },
    {
      "epoch": 3.680588894223076,
      "grad_norm": 0.7187603116035461,
      "learning_rate": 1.929907134012654e-06,
      "loss": 0.699,
      "step": 23000
    },
    {
      "epoch": 3.6885901744279086,
      "grad_norm": 0.7453562617301941,
      "learning_rate": 1.8350178875023272e-06,
      "loss": 0.7243,
      "step": 23050
    },
    {
      "epoch": 3.696591454632741,
      "grad_norm": 0.8272750973701477,
      "learning_rate": 1.7424769691530284e-06,
      "loss": 0.7178,
      "step": 23100
    },
    {
      "epoch": 3.704592734837574,
      "grad_norm": 0.6875506043434143,
      "learning_rate": 1.6522888908826473e-06,
      "loss": 0.7328,
      "step": 23150
    },
    {
      "epoch": 3.7125940150424066,
      "grad_norm": 0.6446670889854431,
      "learning_rate": 1.5644580498941851e-06,
      "loss": 0.7121,
      "step": 23200
    },
    {
      "epoch": 3.7205952952472394,
      "grad_norm": 0.6092432141304016,
      "learning_rate": 1.478988728461367e-06,
      "loss": 0.7045,
      "step": 23250
    },
    {
      "epoch": 3.7285965754520722,
      "grad_norm": 0.6800952553749084,
      "learning_rate": 1.3958850937198453e-06,
      "loss": 0.7275,
      "step": 23300
    },
    {
      "epoch": 3.736597855656905,
      "grad_norm": 0.684956431388855,
      "learning_rate": 1.315151197464043e-06,
      "loss": 0.7272,
      "step": 23350
    },
    {
      "epoch": 3.744599135861738,
      "grad_norm": 0.7142546772956848,
      "learning_rate": 1.236790975949592e-06,
      "loss": 0.7213,
      "step": 23400
    },
    {
      "epoch": 3.7526004160665707,
      "grad_norm": 0.7309266328811646,
      "learning_rate": 1.1608082497014228e-06,
      "loss": 0.7212,
      "step": 23450
    },
    {
      "epoch": 3.7606016962714035,
      "grad_norm": 0.7835919260978699,
      "learning_rate": 1.087206723327483e-06,
      "loss": 0.7411,
      "step": 23500
    },
    {
      "epoch": 3.7686029764762363,
      "grad_norm": 0.6577751040458679,
      "learning_rate": 1.0159899853381394e-06,
      "loss": 0.7104,
      "step": 23550
    },
    {
      "epoch": 3.776604256681069,
      "grad_norm": 0.7286819219589233,
      "learning_rate": 9.471615079711838e-07,
      "loss": 0.7284,
      "step": 23600
    },
    {
      "epoch": 3.784605536885902,
      "grad_norm": 0.7102928757667542,
      "learning_rate": 8.807246470225517e-07,
      "loss": 0.7415,
      "step": 23650
    },
    {
      "epoch": 3.7926068170907348,
      "grad_norm": 0.6017107963562012,
      "learning_rate": 8.166826416827422e-07,
      "loss": 0.7063,
      "step": 23700
    },
    {
      "epoch": 3.800608097295567,
      "grad_norm": 0.7728732824325562,
      "learning_rate": 7.550386143788224e-07,
      "loss": 0.6959,
      "step": 23750
    },
    {
      "epoch": 3.8086093775004,
      "grad_norm": 0.7075885534286499,
      "learning_rate": 6.95795570622243e-07,
      "loss": 0.6977,
      "step": 23800
    },
    {
      "epoch": 3.816610657705233,
      "grad_norm": 0.7628594636917114,
      "learning_rate": 6.389563988622948e-07,
      "loss": 0.7189,
      "step": 23850
    },
    {
      "epoch": 3.8246119379100656,
      "grad_norm": 0.6857520341873169,
      "learning_rate": 5.845238703452604e-07,
      "loss": 0.7281,
      "step": 23900
    },
    {
      "epoch": 3.8326132181148984,
      "grad_norm": 0.676717221736908,
      "learning_rate": 5.325006389793053e-07,
      "loss": 0.7275,
      "step": 23950
    },
    {
      "epoch": 3.8406144983197312,
      "grad_norm": 0.6261276602745056,
      "learning_rate": 4.828892412050978e-07,
      "loss": 0.7162,
      "step": 24000
    },
    {
      "epoch": 3.848615778524564,
      "grad_norm": 0.761638343334198,
      "learning_rate": 4.356920958721311e-07,
      "loss": 0.7405,
      "step": 24050
    },
    {
      "epoch": 3.856617058729397,
      "grad_norm": 0.697512686252594,
      "learning_rate": 3.909115041207889e-07,
      "loss": 0.7156,
      "step": 24100
    },
    {
      "epoch": 3.8646183389342292,
      "grad_norm": 0.6379271149635315,
      "learning_rate": 3.4854964927015235e-07,
      "loss": 0.7443,
      "step": 24150
    },
    {
      "epoch": 3.872619619139062,
      "grad_norm": 0.7762777209281921,
      "learning_rate": 3.086085967115626e-07,
      "loss": 0.7186,
      "step": 24200
    },
    {
      "epoch": 3.880620899343895,
      "grad_norm": 0.6799700260162354,
      "learning_rate": 2.7109029380790186e-07,
      "loss": 0.6928,
      "step": 24250
    },
    {
      "epoch": 3.8886221795487277,
      "grad_norm": 0.8152979016304016,
      "learning_rate": 2.3599656979866325e-07,
      "loss": 0.7145,
      "step": 24300
    },
    {
      "epoch": 3.8966234597535605,
      "grad_norm": 0.6055078506469727,
      "learning_rate": 2.0332913571074476e-07,
      "loss": 0.7105,
      "step": 24350
    },
    {
      "epoch": 3.9046247399583933,
      "grad_norm": 0.7122170329093933,
      "learning_rate": 1.7308958427505462e-07,
      "loss": 0.7322,
      "step": 24400
    },
    {
      "epoch": 3.912626020163226,
      "grad_norm": 0.7174406051635742,
      "learning_rate": 1.4527938984883471e-07,
      "loss": 0.7294,
      "step": 24450
    },
    {
      "epoch": 3.920627300368059,
      "grad_norm": 0.7419302463531494,
      "learning_rate": 1.1989990834378462e-07,
      "loss": 0.7272,
      "step": 24500
    },
    {
      "epoch": 3.9286285805728918,
      "grad_norm": 0.7208146452903748,
      "learning_rate": 9.695237715994788e-08,
      "loss": 0.7066,
      "step": 24550
    },
    {
      "epoch": 3.9366298607777246,
      "grad_norm": 0.8183807730674744,
      "learning_rate": 7.6437915125388e-08,
      "loss": 0.721,
      "step": 24600
    },
    {
      "epoch": 3.9446311409825574,
      "grad_norm": 0.7065560817718506,
      "learning_rate": 5.835752244164883e-08,
      "loss": 0.737,
      "step": 24650
    },
    {
      "epoch": 3.95263242118739,
      "grad_norm": 0.7247999310493469,
      "learning_rate": 4.2712080634949024e-08,
      "loss": 0.7346,
      "step": 24700
    },
    {
      "epoch": 3.9606337013922226,
      "grad_norm": 0.6564909219741821,
      "learning_rate": 2.9502352513255394e-08,
      "loss": 0.7305,
      "step": 24750
    },
    {
      "epoch": 3.9686349815970554,
      "grad_norm": 0.7844890356063843,
      "learning_rate": 1.8728982129051497e-08,
      "loss": 0.7076,
      "step": 24800
    },
    {
      "epoch": 3.9766362618018882,
      "grad_norm": 0.7181993722915649,
      "learning_rate": 1.0392494747957227e-08,
      "loss": 0.7439,
      "step": 24850
    },
    {
      "epoch": 3.984637542006721,
      "grad_norm": 0.6785070300102234,
      "learning_rate": 4.493296823104842e-09,
      "loss": 0.7222,
      "step": 24900
    },
    {
      "epoch": 3.992638822211554,
      "grad_norm": 0.6972596645355225,
      "learning_rate": 1.0316759753381534e-09,
      "loss": 0.7258,
      "step": 24950
    }
  ],
  "logging_steps": 50,
  "max_steps": 24996,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.928976828545164e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}