{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9999002576676919,
  "eval_steps": 10000,
  "global_step": 30076,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.006649488820546921,
      "grad_norm": 0.2602408528327942,
      "learning_rate": 3.324468085106383e-06,
      "loss": 0.818,
      "step": 100
    },
    {
      "epoch": 0.013298977641093841,
      "grad_norm": 0.14098776876926422,
      "learning_rate": 6.648936170212766e-06,
      "loss": 0.5412,
      "step": 200
    },
    {
      "epoch": 0.019948466461640762,
      "grad_norm": 0.4028046429157257,
      "learning_rate": 9.973404255319148e-06,
      "loss": 0.5292,
      "step": 300
    },
    {
      "epoch": 0.026597955282187682,
      "grad_norm": 0.15940147638320923,
      "learning_rate": 1.3297872340425532e-05,
      "loss": 0.5203,
      "step": 400
    },
    {
      "epoch": 0.0332474441027346,
      "grad_norm": 0.32640570402145386,
      "learning_rate": 1.6622340425531915e-05,
      "loss": 0.5078,
      "step": 500
    },
    {
      "epoch": 0.039896932923281524,
      "grad_norm": 0.24910107254981995,
      "learning_rate": 1.9946808510638297e-05,
      "loss": 0.4875,
      "step": 600
    },
    {
      "epoch": 0.04654642174382844,
      "grad_norm": 0.43389075994491577,
      "learning_rate": 2.3271276595744682e-05,
      "loss": 0.4978,
      "step": 700
    },
    {
      "epoch": 0.053195910564375365,
      "grad_norm": 0.2782938778400421,
      "learning_rate": 2.6595744680851064e-05,
      "loss": 0.4807,
      "step": 800
    },
    {
      "epoch": 0.05984539938492228,
      "grad_norm": 0.3213005065917969,
      "learning_rate": 2.9920212765957452e-05,
      "loss": 0.5007,
      "step": 900
    },
    {
      "epoch": 0.0664948882054692,
      "grad_norm": 0.7565635442733765,
      "learning_rate": 3.324468085106383e-05,
      "loss": 0.4754,
      "step": 1000
    },
    {
      "epoch": 0.07314437702601613,
      "grad_norm": 0.5631841421127319,
      "learning_rate": 3.6569148936170215e-05,
      "loss": 0.4943,
      "step": 1100
    },
    {
      "epoch": 0.07979386584656305,
      "grad_norm": 0.40041548013687134,
      "learning_rate": 3.9893617021276594e-05,
      "loss": 0.4928,
      "step": 1200
    },
    {
      "epoch": 0.08644335466710996,
      "grad_norm": 0.43981432914733887,
      "learning_rate": 4.3218085106382986e-05,
      "loss": 0.4829,
      "step": 1300
    },
    {
      "epoch": 0.09309284348765688,
      "grad_norm": 0.49957290291786194,
      "learning_rate": 4.6542553191489364e-05,
      "loss": 0.4797,
      "step": 1400
    },
    {
      "epoch": 0.09974233230820381,
      "grad_norm": 1.0348299741744995,
      "learning_rate": 4.986702127659575e-05,
      "loss": 0.5052,
      "step": 1500
    },
    {
      "epoch": 0.10639182112875073,
      "grad_norm": 0.7030391097068787,
      "learning_rate": 5.319148936170213e-05,
      "loss": 0.4693,
      "step": 1600
    },
    {
      "epoch": 0.11304130994929765,
      "grad_norm": 0.7447443008422852,
      "learning_rate": 5.651595744680851e-05,
      "loss": 0.4767,
      "step": 1700
    },
    {
      "epoch": 0.11969079876984456,
      "grad_norm": 0.5579438209533691,
      "learning_rate": 5.9840425531914904e-05,
      "loss": 0.4823,
      "step": 1800
    },
    {
      "epoch": 0.12634028759039148,
      "grad_norm": 0.5548936724662781,
      "learning_rate": 6.316489361702128e-05,
      "loss": 0.4631,
      "step": 1900
    },
    {
      "epoch": 0.1329897764109384,
      "grad_norm": 0.9034917950630188,
      "learning_rate": 6.648936170212766e-05,
      "loss": 0.5277,
      "step": 2000
    },
    {
      "epoch": 0.13963926523148534,
      "grad_norm": 0.7092494964599609,
      "learning_rate": 6.981382978723405e-05,
      "loss": 0.4813,
      "step": 2100
    },
    {
      "epoch": 0.14628875405203226,
      "grad_norm": 1.0150775909423828,
      "learning_rate": 7.313829787234043e-05,
      "loss": 0.5026,
      "step": 2200
    },
    {
      "epoch": 0.15293824287257918,
      "grad_norm": 1.1101422309875488,
      "learning_rate": 7.646276595744682e-05,
      "loss": 0.5226,
      "step": 2300
    },
    {
      "epoch": 0.1595877316931261,
      "grad_norm": 0.5893256664276123,
      "learning_rate": 7.978723404255319e-05,
      "loss": 0.5285,
      "step": 2400
    },
    {
      "epoch": 0.166237220513673,
      "grad_norm": 0.9117974042892456,
      "learning_rate": 8.311170212765957e-05,
      "loss": 0.5183,
      "step": 2500
    },
    {
      "epoch": 0.17288670933421993,
      "grad_norm": 1.3572144508361816,
      "learning_rate": 8.643617021276597e-05,
      "loss": 0.5129,
      "step": 2600
    },
    {
      "epoch": 0.17953619815476685,
      "grad_norm": 0.890178382396698,
      "learning_rate": 8.976063829787234e-05,
      "loss": 0.5081,
      "step": 2700
    },
    {
      "epoch": 0.18618568697531376,
      "grad_norm": 2.725203275680542,
      "learning_rate": 9.308510638297873e-05,
      "loss": 0.5042,
      "step": 2800
    },
    {
      "epoch": 0.19283517579586068,
      "grad_norm": 0.962457001209259,
      "learning_rate": 9.640957446808511e-05,
      "loss": 0.5357,
      "step": 2900
    },
    {
      "epoch": 0.19948466461640763,
      "grad_norm": 2.529735803604126,
      "learning_rate": 9.97340425531915e-05,
      "loss": 0.5407,
      "step": 3000
    },
    {
      "epoch": 0.20613415343695454,
      "grad_norm": 1.643441081047058,
      "learning_rate": 9.999714964504067e-05,
      "loss": 0.5244,
      "step": 3100
    },
    {
      "epoch": 0.21278364225750146,
      "grad_norm": 0.9840033054351807,
      "learning_rate": 9.998758599534463e-05,
      "loss": 0.5562,
      "step": 3200
    },
    {
      "epoch": 0.21943313107804838,
      "grad_norm": 1.0612396001815796,
      "learning_rate": 9.997128878619808e-05,
      "loss": 0.5273,
      "step": 3300
    },
    {
      "epoch": 0.2260826198985953,
      "grad_norm": 1.1562029123306274,
      "learning_rate": 9.994826021291056e-05,
      "loss": 0.5412,
      "step": 3400
    },
    {
      "epoch": 0.2327321087191422,
      "grad_norm": 1.5558404922485352,
      "learning_rate": 9.991850337753762e-05,
      "loss": 0.5405,
      "step": 3500
    },
    {
      "epoch": 0.23938159753968913,
      "grad_norm": 2.0897717475891113,
      "learning_rate": 9.988202228846291e-05,
      "loss": 0.5049,
      "step": 3600
    },
    {
      "epoch": 0.24603108636023605,
      "grad_norm": 1.031879186630249,
      "learning_rate": 9.983882185985808e-05,
      "loss": 0.5785,
      "step": 3700
    },
    {
      "epoch": 0.25268057518078296,
      "grad_norm": 1.3407084941864014,
      "learning_rate": 9.978890791102109e-05,
      "loss": 0.5487,
      "step": 3800
    },
    {
      "epoch": 0.2593300640013299,
      "grad_norm": 1.261853814125061,
      "learning_rate": 9.973228716559209e-05,
      "loss": 0.533,
      "step": 3900
    },
    {
      "epoch": 0.2659795528218768,
      "grad_norm": 1.1546040773391724,
      "learning_rate": 9.966896725064786e-05,
      "loss": 0.5166,
      "step": 4000
    },
    {
      "epoch": 0.27262904164242374,
      "grad_norm": 1.1395092010498047,
      "learning_rate": 9.959895669567435e-05,
      "loss": 0.5366,
      "step": 4100
    },
    {
      "epoch": 0.2792785304629707,
      "grad_norm": 1.5521488189697266,
      "learning_rate": 9.952226493141765e-05,
      "loss": 0.5339,
      "step": 4200
    },
    {
      "epoch": 0.2859280192835176,
      "grad_norm": 1.9154760837554932,
      "learning_rate": 9.943890228861383e-05,
      "loss": 0.4976,
      "step": 4300
    },
    {
      "epoch": 0.2925775081040645,
      "grad_norm": 1.273290753364563,
      "learning_rate": 9.934887999659707e-05,
      "loss": 0.5331,
      "step": 4400
    },
    {
      "epoch": 0.2992269969246114,
      "grad_norm": 1.6757972240447998,
      "learning_rate": 9.925221018178728e-05,
      "loss": 0.5212,
      "step": 4500
    },
    {
      "epoch": 0.30587648574515836,
      "grad_norm": 1.3594906330108643,
      "learning_rate": 9.914890586605638e-05,
      "loss": 0.4991,
      "step": 4600
    },
    {
      "epoch": 0.31252597456570524,
      "grad_norm": 1.622048258781433,
      "learning_rate": 9.903898096497441e-05,
      "loss": 0.569,
      "step": 4700
    },
    {
      "epoch": 0.3191754633862522,
      "grad_norm": 1.6205965280532837,
      "learning_rate": 9.892245028593483e-05,
      "loss": 0.5436,
      "step": 4800
    },
    {
      "epoch": 0.3258249522067991,
      "grad_norm": 2.7058024406433105,
      "learning_rate": 9.879932952616009e-05,
      "loss": 0.5028,
      "step": 4900
    },
    {
      "epoch": 0.332474441027346,
      "grad_norm": 1.8897464275360107,
      "learning_rate": 9.8669635270587e-05,
      "loss": 0.7608,
      "step": 5000
    },
    {
      "epoch": 0.33912392984789297,
      "grad_norm": 0.9735029935836792,
      "learning_rate": 9.853338498963272e-05,
      "loss": 0.523,
      "step": 5100
    },
    {
      "epoch": 0.34577341866843986,
      "grad_norm": 0.9950180053710938,
      "learning_rate": 9.839059703684139e-05,
      "loss": 0.5426,
      "step": 5200
    },
    {
      "epoch": 0.3524229074889868,
      "grad_norm": 0.7262745499610901,
      "learning_rate": 9.82412906464119e-05,
      "loss": 0.5387,
      "step": 5300
    },
    {
      "epoch": 0.3590723963095337,
      "grad_norm": 1.133631706237793,
      "learning_rate": 9.808548593060681e-05,
      "loss": 0.5259,
      "step": 5400
    },
    {
      "epoch": 0.36572188513008064,
      "grad_norm": 1.8873803615570068,
      "learning_rate": 9.792320387704328e-05,
      "loss": 0.5363,
      "step": 5500
    },
    {
      "epoch": 0.3723713739506275,
      "grad_norm": 2.6183230876922607,
      "learning_rate": 9.775446634586584e-05,
      "loss": 0.542,
      "step": 5600
    },
    {
      "epoch": 0.37902086277117447,
      "grad_norm": 1.7073196172714233,
      "learning_rate": 9.757929606680181e-05,
      "loss": 0.4947,
      "step": 5700
    },
    {
      "epoch": 0.38567035159172136,
      "grad_norm": 1.0039314031600952,
      "learning_rate": 9.73977166360994e-05,
      "loss": 0.5584,
      "step": 5800
    },
    {
      "epoch": 0.3923198404122683,
      "grad_norm": 1.0683766603469849,
      "learning_rate": 9.720975251334929e-05,
      "loss": 0.5318,
      "step": 5900
    },
    {
      "epoch": 0.39896932923281525,
      "grad_norm": 1.4205585718154907,
      "learning_rate": 9.701542901818974e-05,
      "loss": 0.5352,
      "step": 6000
    },
    {
      "epoch": 0.40561881805336214,
      "grad_norm": 2.596090316772461,
      "learning_rate": 9.681477232689596e-05,
      "loss": 0.5229,
      "step": 6100
    },
    {
      "epoch": 0.4122683068739091,
      "grad_norm": 1.1745169162750244,
      "learning_rate": 9.660780946885397e-05,
      "loss": 0.5419,
      "step": 6200
    },
    {
      "epoch": 0.418917795694456,
      "grad_norm": 2.7921059131622314,
      "learning_rate": 9.639456832291974e-05,
      "loss": 0.5428,
      "step": 6300
    },
    {
      "epoch": 0.4255672845150029,
      "grad_norm": 3.064069986343384,
      "learning_rate": 9.617507761366367e-05,
      "loss": 0.5256,
      "step": 6400
    },
    {
      "epoch": 0.4322167733355498,
      "grad_norm": 1.322499394416809,
      "learning_rate": 9.594936690750129e-05,
      "loss": 0.5633,
      "step": 6500
    },
    {
      "epoch": 0.43886626215609675,
      "grad_norm": 1.6360127925872803,
      "learning_rate": 9.571746660871058e-05,
      "loss": 0.5269,
      "step": 6600
    },
    {
      "epoch": 0.44551575097664364,
      "grad_norm": 1.442179799079895,
      "learning_rate": 9.547940795533627e-05,
      "loss": 0.5105,
      "step": 6700
    },
    {
      "epoch": 0.4521652397971906,
      "grad_norm": 1.1154464483261108,
      "learning_rate": 9.523522301498202e-05,
      "loss": 0.5301,
      "step": 6800
    },
    {
      "epoch": 0.45881472861773753,
      "grad_norm": 1.2995010614395142,
      "learning_rate": 9.498494468049072e-05,
      "loss": 0.5173,
      "step": 6900
    },
    {
      "epoch": 0.4654642174382844,
      "grad_norm": 1.0487501621246338,
      "learning_rate": 9.472860666551369e-05,
      "loss": 0.5284,
      "step": 7000
    },
    {
      "epoch": 0.47211370625883137,
      "grad_norm": 1.8207134008407593,
      "learning_rate": 9.446624349996929e-05,
      "loss": 0.5317,
      "step": 7100
    },
    {
      "epoch": 0.47876319507937826,
      "grad_norm": 1.7722837924957275,
      "learning_rate": 9.419789052539157e-05,
      "loss": 0.5195,
      "step": 7200
    },
    {
      "epoch": 0.4854126838999252,
      "grad_norm": 2.749199390411377,
      "learning_rate": 9.392358389016961e-05,
      "loss": 0.5274,
      "step": 7300
    },
    {
      "epoch": 0.4920621727204721,
      "grad_norm": 1.3200327157974243,
      "learning_rate": 9.364336054467819e-05,
      "loss": 0.523,
      "step": 7400
    },
    {
      "epoch": 0.49871166154101904,
      "grad_norm": 2.9674599170684814,
      "learning_rate": 9.335725823630035e-05,
      "loss": 0.5294,
      "step": 7500
    },
    {
      "epoch": 0.5053611503615659,
      "grad_norm": 2.8731188774108887,
      "learning_rate": 9.306531550434268e-05,
      "loss": 0.5087,
      "step": 7600
    },
    {
      "epoch": 0.5120106391821129,
      "grad_norm": 2.215608596801758,
      "learning_rate": 9.276757167484389e-05,
      "loss": 0.5219,
      "step": 7700
    },
    {
      "epoch": 0.5186601280026598,
      "grad_norm": 1.5501729249954224,
      "learning_rate": 9.246406685527739e-05,
      "loss": 0.5582,
      "step": 7800
    },
    {
      "epoch": 0.5253096168232068,
      "grad_norm": 1.8816163539886475,
      "learning_rate": 9.21548419291487e-05,
      "loss": 0.5394,
      "step": 7900
    },
    {
      "epoch": 0.5319591056437536,
      "grad_norm": 1.338460087776184,
      "learning_rate": 9.183993855048811e-05,
      "loss": 0.5299,
      "step": 8000
    },
    {
      "epoch": 0.5386085944643005,
      "grad_norm": 1.7432674169540405,
      "learning_rate": 9.151939913823988e-05,
      "loss": 0.5307,
      "step": 8100
    },
    {
      "epoch": 0.5452580832848475,
      "grad_norm": 3.2521355152130127,
      "learning_rate": 9.119326687054802e-05,
      "loss": 0.5004,
      "step": 8200
    },
    {
      "epoch": 0.5519075721053944,
      "grad_norm": 0.9576961994171143,
      "learning_rate": 9.086158567894013e-05,
      "loss": 0.5453,
      "step": 8300
    },
    {
      "epoch": 0.5585570609259414,
      "grad_norm": 1.3667104244232178,
      "learning_rate": 9.052440024240956e-05,
      "loss": 0.5003,
      "step": 8400
    },
    {
      "epoch": 0.5652065497464882,
      "grad_norm": 1.9807591438293457,
      "learning_rate": 9.018175598139696e-05,
      "loss": 0.508,
      "step": 8500
    },
    {
      "epoch": 0.5718560385670352,
      "grad_norm": 2.281437397003174,
      "learning_rate": 8.983369905167191e-05,
      "loss": 0.4788,
      "step": 8600
    },
    {
      "epoch": 0.5785055273875821,
      "grad_norm": 2.3274779319763184,
      "learning_rate": 8.948027633811557e-05,
      "loss": 0.4994,
      "step": 8700
    },
    {
      "epoch": 0.585155016208129,
      "grad_norm": 3.7350873947143555,
      "learning_rate": 8.912153544840507e-05,
      "loss": 0.546,
      "step": 8800
    },
    {
      "epoch": 0.5918045050286759,
      "grad_norm": 1.8497984409332275,
      "learning_rate": 8.875752470660043e-05,
      "loss": 0.4939,
      "step": 8900
    },
    {
      "epoch": 0.5984539938492228,
      "grad_norm": 2.2370221614837646,
      "learning_rate": 8.838829314663522e-05,
      "loss": 0.5288,
      "step": 9000
    },
    {
      "epoch": 0.6051034826697698,
      "grad_norm": 1.4384081363677979,
      "learning_rate": 8.80138905057114e-05,
      "loss": 0.5176,
      "step": 9100
    },
    {
      "epoch": 0.6117529714903167,
      "grad_norm": 3.588320255279541,
      "learning_rate": 8.763436721759952e-05,
      "loss": 0.4898,
      "step": 9200
    },
    {
      "epoch": 0.6184024603108637,
      "grad_norm": 2.6353256702423096,
      "learning_rate": 8.724977440584497e-05,
      "loss": 0.5011,
      "step": 9300
    },
    {
      "epoch": 0.6250519491314105,
      "grad_norm": 1.7550867795944214,
      "learning_rate": 8.686016387688153e-05,
      "loss": 0.5101,
      "step": 9400
    },
    {
      "epoch": 0.6317014379519574,
      "grad_norm": 1.9632377624511719,
      "learning_rate": 8.646558811305268e-05,
      "loss": 0.5023,
      "step": 9500
    },
    {
      "epoch": 0.6383509267725044,
      "grad_norm": 2.103327989578247,
      "learning_rate": 8.60661002655421e-05,
      "loss": 0.5055,
      "step": 9600
    },
    {
      "epoch": 0.6450004155930513,
      "grad_norm": 2.518740653991699,
      "learning_rate": 8.566175414721384e-05,
      "loss": 0.4997,
      "step": 9700
    },
    {
      "epoch": 0.6516499044135982,
      "grad_norm": 2.6153292655944824,
      "learning_rate": 8.525260422536358e-05,
      "loss": 0.5104,
      "step": 9800
    },
    {
      "epoch": 0.6582993932341451,
      "grad_norm": 1.0710101127624512,
      "learning_rate": 8.483870561438161e-05,
      "loss": 0.5281,
      "step": 9900
    },
    {
      "epoch": 0.664948882054692,
      "grad_norm": 3.5133893489837646,
      "learning_rate": 8.442011406832859e-05,
      "loss": 0.5147,
      "step": 10000
    },
    {
      "epoch": 0.664948882054692,
      "eval_loss": 0.4313640594482422,
      "eval_runtime": 4340.2869,
      "eval_samples_per_second": 1.54,
      "eval_steps_per_second": 1.54,
      "step": 10000
    },
    {
      "epoch": 0.671598370875239,
      "grad_norm": 1.416164755821228,
      "learning_rate": 8.399688597342535e-05,
      "loss": 0.5229,
      "step": 10100
    },
    {
      "epoch": 0.6782478596957859,
      "grad_norm": 4.202746391296387,
      "learning_rate": 8.356907834045726e-05,
      "loss": 0.5008,
      "step": 10200
    },
    {
      "epoch": 0.6848973485163328,
      "grad_norm": 1.6467583179473877,
      "learning_rate": 8.313674879709475e-05,
      "loss": 0.5084,
      "step": 10300
    },
    {
      "epoch": 0.6915468373368797,
      "grad_norm": 0.9791653752326965,
      "learning_rate": 8.269995558013049e-05,
      "loss": 0.5049,
      "step": 10400
    },
    {
      "epoch": 0.6981963261574267,
      "grad_norm": 1.2445042133331299,
      "learning_rate": 8.225875752763468e-05,
      "loss": 0.4711,
      "step": 10500
    },
    {
      "epoch": 0.7048458149779736,
      "grad_norm": 3.4758522510528564,
      "learning_rate": 8.181321407102929e-05,
      "loss": 0.4835,
      "step": 10600
    },
    {
      "epoch": 0.7114953037985204,
      "grad_norm": 1.5787672996520996,
      "learning_rate": 8.136338522708233e-05,
      "loss": 0.4849,
      "step": 10700
    },
    {
      "epoch": 0.7181447926190674,
      "grad_norm": 2.1171529293060303,
      "learning_rate": 8.090933158982338e-05,
      "loss": 0.4881,
      "step": 10800
    },
    {
      "epoch": 0.7247942814396143,
      "grad_norm": 3.9836864471435547,
      "learning_rate": 8.045111432238121e-05,
      "loss": 0.4699,
      "step": 10900
    },
    {
      "epoch": 0.7314437702601613,
      "grad_norm": 3.3204433917999268,
      "learning_rate": 7.998879514874491e-05,
      "loss": 0.5022,
      "step": 11000
    },
    {
      "epoch": 0.7380932590807082,
      "grad_norm": 1.5986028909683228,
      "learning_rate": 7.952243634544929e-05,
      "loss": 0.4933,
      "step": 11100
    },
    {
      "epoch": 0.744742747901255,
      "grad_norm": 3.2928895950317383,
      "learning_rate": 7.905210073318605e-05,
      "loss": 0.4677,
      "step": 11200
    },
    {
      "epoch": 0.751392236721802,
      "grad_norm": 2.922384738922119,
      "learning_rate": 7.857785166834144e-05,
      "loss": 0.4757,
      "step": 11300
    },
    {
      "epoch": 0.7580417255423489,
      "grad_norm": 1.8758389949798584,
      "learning_rate": 7.809975303446195e-05,
      "loss": 0.4938,
      "step": 11400
    },
    {
      "epoch": 0.7646912143628959,
      "grad_norm": 2.5382237434387207,
      "learning_rate": 7.761786923364878e-05,
      "loss": 0.4733,
      "step": 11500
    },
    {
      "epoch": 0.7713407031834427,
      "grad_norm": 3.0387401580810547,
      "learning_rate": 7.713226517788275e-05,
      "loss": 0.4655,
      "step": 11600
    },
    {
      "epoch": 0.7779901920039897,
      "grad_norm": 3.491082191467285,
      "learning_rate": 7.664300628028017e-05,
      "loss": 0.4933,
      "step": 11700
    },
    {
      "epoch": 0.7846396808245366,
      "grad_norm": 2.891706943511963,
      "learning_rate": 7.615015844628157e-05,
      "loss": 0.4699,
      "step": 11800
    },
    {
      "epoch": 0.7912891696450836,
      "grad_norm": 1.315558671951294,
      "learning_rate": 7.565378806477377e-05,
      "loss": 0.4514,
      "step": 11900
    },
    {
      "epoch": 0.7979386584656305,
      "grad_norm": 5.145135402679443,
      "learning_rate": 7.515396199914708e-05,
      "loss": 0.4532,
      "step": 12000
    },
    {
      "epoch": 0.8045881472861773,
      "grad_norm": 2.2307186126708984,
      "learning_rate": 7.46507475782885e-05,
      "loss": 0.4749,
      "step": 12100
    },
    {
      "epoch": 0.8112376361067243,
      "grad_norm": 1.4436883926391602,
      "learning_rate": 7.414421258751212e-05,
      "loss": 0.4977,
      "step": 12200
    },
    {
      "epoch": 0.8178871249272712,
      "grad_norm": 2.0418126583099365,
      "learning_rate": 7.363442525942826e-05,
      "loss": 0.4544,
      "step": 12300
    },
    {
      "epoch": 0.8245366137478182,
      "grad_norm": 1.2745349407196045,
      "learning_rate": 7.31214542647521e-05,
      "loss": 0.5053,
      "step": 12400
    },
    {
      "epoch": 0.831186102568365,
      "grad_norm": 3.4680609703063965,
      "learning_rate": 7.260536870305347e-05,
      "loss": 0.4648,
      "step": 12500
    },
    {
      "epoch": 0.837835591388912,
      "grad_norm": 0.97030109167099,
      "learning_rate": 7.208623809344879e-05,
      "loss": 0.4313,
      "step": 12600
    },
    {
      "epoch": 0.8444850802094589,
      "grad_norm": 3.7414844036102295,
      "learning_rate": 7.156413236523656e-05,
      "loss": 0.4816,
      "step": 12700
    },
    {
      "epoch": 0.8511345690300058,
      "grad_norm": 1.8031582832336426,
      "learning_rate": 7.103912184847757e-05,
      "loss": 0.4534,
      "step": 12800
    },
    {
      "epoch": 0.8577840578505528,
      "grad_norm": 1.6116315126419067,
      "learning_rate": 7.051127726452102e-05,
      "loss": 0.4304,
      "step": 12900
    },
    {
      "epoch": 0.8644335466710996,
      "grad_norm": 2.230163335800171,
      "learning_rate": 6.998066971647817e-05,
      "loss": 0.4511,
      "step": 13000
    },
    {
      "epoch": 0.8710830354916466,
      "grad_norm": 1.854151964187622,
      "learning_rate": 6.944737067964429e-05,
      "loss": 0.4412,
      "step": 13100
    },
    {
      "epoch": 0.8777325243121935,
      "grad_norm": 2.757798194885254,
      "learning_rate": 6.891145199187065e-05,
      "loss": 0.4455,
      "step": 13200
    },
    {
      "epoch": 0.8843820131327405,
      "grad_norm": 2.430884838104248,
      "learning_rate": 6.837298584388771e-05,
      "loss": 0.4598,
      "step": 13300
    },
    {
      "epoch": 0.8910315019532873,
      "grad_norm": 1.6644638776779175,
      "learning_rate": 6.783204476958058e-05,
      "loss": 0.4561,
      "step": 13400
    },
    {
      "epoch": 0.8976809907738342,
      "grad_norm": 2.7078986167907715,
      "learning_rate": 6.728870163621836e-05,
      "loss": 0.4818,
      "step": 13500
    },
    {
      "epoch": 0.9043304795943812,
      "grad_norm": 1.8221591711044312,
      "learning_rate": 6.674302963463876e-05,
      "loss": 0.4322,
      "step": 13600
    },
    {
      "epoch": 0.9109799684149281,
      "grad_norm": 1.0504103899002075,
      "learning_rate": 6.61951022693887e-05,
      "loss": 0.4601,
      "step": 13700
    },
    {
      "epoch": 0.9176294572354751,
      "grad_norm": 1.788719892501831,
      "learning_rate": 6.564499334882312e-05,
      "loss": 0.4787,
      "step": 13800
    },
    {
      "epoch": 0.9242789460560219,
      "grad_norm": 1.5331158638000488,
      "learning_rate": 6.509277697516255e-05,
      "loss": 0.4766,
      "step": 13900
    },
    {
      "epoch": 0.9309284348765688,
      "grad_norm": 2.4608702659606934,
      "learning_rate": 6.453852753451119e-05,
      "loss": 0.4614,
      "step": 14000
    },
    {
      "epoch": 0.9375779236971158,
      "grad_norm": 1.432626485824585,
      "learning_rate": 6.39823196868368e-05,
      "loss": 0.48,
      "step": 14100
    },
    {
      "epoch": 0.9442274125176627,
      "grad_norm": 3.4951376914978027,
      "learning_rate": 6.342422835591368e-05,
      "loss": 0.4717,
      "step": 14200
    },
    {
      "epoch": 0.9508769013382097,
      "grad_norm": 1.7661138772964478,
      "learning_rate": 6.286432871923e-05,
      "loss": 0.4609,
      "step": 14300
    },
    {
      "epoch": 0.9575263901587565,
      "grad_norm": 2.387974500656128,
      "learning_rate": 6.230269619786111e-05,
      "loss": 0.4233,
      "step": 14400
    },
    {
      "epoch": 0.9641758789793035,
      "grad_norm": 1.5833674669265747,
      "learning_rate": 6.173940644630996e-05,
      "loss": 0.4368,
      "step": 14500
    },
    {
      "epoch": 0.9708253677998504,
      "grad_norm": 2.6246190071105957,
      "learning_rate": 6.117453534231606e-05,
      "loss": 0.4205,
      "step": 14600
    },
    {
      "epoch": 0.9774748566203973,
      "grad_norm": 1.6440967321395874,
      "learning_rate": 6.060815897663447e-05,
      "loss": 0.4247,
      "step": 14700
    },
    {
      "epoch": 0.9841243454409442,
      "grad_norm": 2.0472023487091064,
      "learning_rate": 6.004035364278593e-05,
      "loss": 0.4338,
      "step": 14800
    },
    {
      "epoch": 0.9907738342614911,
      "grad_norm": 3.3487770557403564,
      "learning_rate": 5.9471195826779834e-05,
      "loss": 0.4114,
      "step": 14900
    },
    {
      "epoch": 0.9974233230820381,
      "grad_norm": 2.1973326206207275,
      "learning_rate": 5.8900762196811175e-05,
      "loss": 0.4209,
      "step": 15000
    },
    {
      "epoch": 1.004072811902585,
      "grad_norm": 2.8700783252716064,
      "learning_rate": 5.832912959293304e-05,
      "loss": 0.3562,
      "step": 15100
    },
    {
      "epoch": 1.0107223007231319,
      "grad_norm": 0.8708151578903198,
      "learning_rate": 5.775637501670579e-05,
      "loss": 0.3388,
      "step": 15200
    },
    {
      "epoch": 1.017371789543679,
      "grad_norm": 0.9684904217720032,
      "learning_rate": 5.718257562082471e-05,
      "loss": 0.3572,
      "step": 15300
    },
    {
      "epoch": 1.0240212783642257,
      "grad_norm": 1.5389324426651,
      "learning_rate": 5.660780869872711e-05,
      "loss": 0.3377,
      "step": 15400
    },
    {
      "epoch": 1.0306707671847726,
      "grad_norm": 1.646532416343689,
      "learning_rate": 5.6032151674180575e-05,
      "loss": 0.3561,
      "step": 15500
    },
    {
      "epoch": 1.0373202560053196,
      "grad_norm": 2.062817096710205,
      "learning_rate": 5.5455682090853624e-05,
      "loss": 0.3698,
      "step": 15600
    },
    {
      "epoch": 1.0439697448258665,
      "grad_norm": 2.0568647384643555,
      "learning_rate": 5.4878477601870194e-05,
      "loss": 0.3223,
      "step": 15700
    },
    {
      "epoch": 1.0506192336464135,
      "grad_norm": 2.101365566253662,
      "learning_rate": 5.430061595934941e-05,
      "loss": 0.3483,
      "step": 15800
    },
    {
      "epoch": 1.0572687224669604,
      "grad_norm": 1.6456376314163208,
      "learning_rate": 5.372217500393205e-05,
      "loss": 0.3642,
      "step": 15900
    },
    {
      "epoch": 1.0639182112875072,
      "grad_norm": 2.0329535007476807,
      "learning_rate": 5.314323265429501e-05,
      "loss": 0.3234,
      "step": 16000
    },
    {
      "epoch": 1.0705677001080542,
      "grad_norm": 2.085597276687622,
      "learning_rate": 5.2563866896655275e-05,
      "loss": 0.3448,
      "step": 16100
    },
    {
      "epoch": 1.077217188928601,
      "grad_norm": 2.2339513301849365,
      "learning_rate": 5.198415577426493e-05,
      "loss": 0.3621,
      "step": 16200
    },
    {
      "epoch": 1.0838666777491481,
      "grad_norm": 2.7302517890930176,
      "learning_rate": 5.140417737689822e-05,
      "loss": 0.3532,
      "step": 16300
    },
    {
      "epoch": 1.090516166569695,
      "grad_norm": 2.6171867847442627,
      "learning_rate": 5.0824009830332606e-05,
      "loss": 0.3424,
      "step": 16400
    },
    {
      "epoch": 1.0971656553902418,
      "grad_norm": 1.7419236898422241,
      "learning_rate": 5.02437312858248e-05,
      "loss": 0.3465,
      "step": 16500
    },
    {
      "epoch": 1.1038151442107889,
      "grad_norm": 1.149902582168579,
      "learning_rate": 4.966341990958347e-05,
      "loss": 0.3471,
      "step": 16600
    },
    {
      "epoch": 1.1104646330313357,
      "grad_norm": 2.0462777614593506,
      "learning_rate": 4.908315387223985e-05,
      "loss": 0.3408,
      "step": 16700
    },
    {
      "epoch": 1.1171141218518827,
      "grad_norm": 1.4826865196228027,
      "learning_rate": 4.850301133831786e-05,
      "loss": 0.3351,
      "step": 16800
    },
    {
      "epoch": 1.1237636106724296,
      "grad_norm": 1.297524094581604,
      "learning_rate": 4.792307045570486e-05,
      "loss": 0.3531,
      "step": 16900
    },
    {
      "epoch": 1.1304130994929764,
      "grad_norm": 0.9654291272163391,
      "learning_rate": 4.734340934512492e-05,
      "loss": 0.3171,
      "step": 17000
    },
    {
      "epoch": 1.1370625883135235,
      "grad_norm": 1.5802520513534546,
      "learning_rate": 4.6764106089615454e-05,
      "loss": 0.3388,
      "step": 17100
    },
    {
      "epoch": 1.1437120771340703,
      "grad_norm": 3.1234793663024902,
      "learning_rate": 4.618523872400921e-05,
      "loss": 0.3356,
      "step": 17200
    },
    {
      "epoch": 1.1503615659546171,
      "grad_norm": 4.1535444259643555,
      "learning_rate": 4.56068852244225e-05,
      "loss": 0.3483,
      "step": 17300
    },
    {
      "epoch": 1.1570110547751642,
      "grad_norm": 2.166757583618164,
      "learning_rate": 4.5029123497751514e-05,
      "loss": 0.3223,
      "step": 17400
    },
    {
      "epoch": 1.163660543595711,
      "grad_norm": 1.9507828950881958,
      "learning_rate": 4.445203137117788e-05,
      "loss": 0.3181,
      "step": 17500
    },
    {
      "epoch": 1.170310032416258,
      "grad_norm": 2.319566011428833,
      "learning_rate": 4.3875686581685e-05,
      "loss": 0.3416,
      "step": 17600
    },
    {
      "epoch": 1.176959521236805,
      "grad_norm": 2.692155599594116,
      "learning_rate": 4.330016676558651e-05,
      "loss": 0.3269,
      "step": 17700
    },
    {
      "epoch": 1.1836090100573518,
      "grad_norm": 1.3827095031738281,
      "learning_rate": 4.272554944806831e-05,
      "loss": 0.3166,
      "step": 17800
    },
    {
      "epoch": 1.1902584988778988,
      "grad_norm": 1.681174635887146,
      "learning_rate": 4.2151912032745547e-05,
      "loss": 0.3354,
      "step": 17900
    },
    {
      "epoch": 1.1969079876984456,
      "grad_norm": 1.9821159839630127,
      "learning_rate": 4.157933179123599e-05,
      "loss": 0.3066,
      "step": 18000
    },
    {
      "epoch": 1.2035574765189927,
      "grad_norm": 1.9395629167556763,
      "learning_rate": 4.100788585275125e-05,
      "loss": 0.3427,
      "step": 18100
    },
    {
      "epoch": 1.2102069653395395,
      "grad_norm": 1.5355767011642456,
      "learning_rate": 4.043765119370699e-05,
      "loss": 0.3074,
      "step": 18200
    },
    {
      "epoch": 1.2168564541600864,
      "grad_norm": 2.809375047683716,
      "learning_rate": 3.9868704627354e-05,
      "loss": 0.3172,
      "step": 18300
    },
    {
      "epoch": 1.2235059429806334,
      "grad_norm": 2.0123348236083984,
      "learning_rate": 3.930112279343094e-05,
      "loss": 0.3088,
      "step": 18400
    },
    {
      "epoch": 1.2301554318011803,
      "grad_norm": 3.793424367904663,
      "learning_rate": 3.8734982147840756e-05,
      "loss": 0.3053,
      "step": 18500
    },
    {
      "epoch": 1.236804920621727,
      "grad_norm": 3.0886518955230713,
      "learning_rate": 3.817035895235159e-05,
      "loss": 0.3178,
      "step": 18600
    },
    {
      "epoch": 1.2434544094422741,
      "grad_norm": 3.0276455879211426,
      "learning_rate": 3.760732926432407e-05,
      "loss": 0.2876,
      "step": 18700
    },
    {
      "epoch": 1.250103898262821,
      "grad_norm": 2.8517942428588867,
      "learning_rate": 3.704596892646593e-05,
      "loss": 0.3264,
      "step": 18800
    },
    {
      "epoch": 1.256753387083368,
      "grad_norm": 3.9238364696502686,
      "learning_rate": 3.648635355661577e-05,
      "loss": 0.3151,
      "step": 18900
    },
    {
      "epoch": 1.2634028759039149,
      "grad_norm": 2.554077625274658,
      "learning_rate": 3.5928558537556895e-05,
      "loss": 0.3086,
      "step": 19000
    },
    {
      "epoch": 1.270052364724462,
      "grad_norm": 1.959659457206726,
      "learning_rate": 3.537265900686286e-05,
      "loss": 0.2933,
      "step": 19100
    },
    {
      "epoch": 1.2767018535450088,
      "grad_norm": 2.120922327041626,
      "learning_rate": 3.4818729846776254e-05,
      "loss": 0.2867,
      "step": 19200
    },
    {
      "epoch": 1.2833513423655556,
      "grad_norm": 2.423426389694214,
      "learning_rate": 3.426684567412153e-05,
      "loss": 0.2956,
      "step": 19300
    },
    {
      "epoch": 1.2900008311861026,
      "grad_norm": 2.462615489959717,
      "learning_rate": 3.371708083025392e-05,
      "loss": 0.2944,
      "step": 19400
    },
    {
      "epoch": 1.2966503200066495,
      "grad_norm": 1.9235759973526,
      "learning_rate": 3.316950937104518e-05,
      "loss": 0.2868,
      "step": 19500
    },
    {
      "epoch": 1.3032998088271963,
      "grad_norm": 3.4119069576263428,
      "learning_rate": 3.26242050569081e-05,
      "loss": 0.306,
      "step": 19600
    },
    {
      "epoch": 1.3099492976477434,
      "grad_norm": 5.942444801330566,
      "learning_rate": 3.208124134286038e-05,
      "loss": 0.294,
      "step": 19700
    },
    {
      "epoch": 1.3165987864682902,
      "grad_norm": 1.254133701324463,
      "learning_rate": 3.1540691368630185e-05,
      "loss": 0.3237,
      "step": 19800
    },
    {
      "epoch": 1.323248275288837,
      "grad_norm": 3.514387369155884,
      "learning_rate": 3.100262794880363e-05,
      "loss": 0.3111,
      "step": 19900
    },
    {
      "epoch": 1.329897764109384,
      "grad_norm": 7.018650054931641,
      "learning_rate": 3.0467123563016513e-05,
      "loss": 0.3106,
      "step": 20000
    },
    {
      "epoch": 1.329897764109384,
      "eval_loss": 0.31190410256385803,
      "eval_runtime": 4327.0173,
      "eval_samples_per_second": 1.545,
      "eval_steps_per_second": 1.545,
      "step": 20000
    },
    {
      "epoch": 1.336547252929931,
      "grad_norm": 0.9992861747741699,
      "learning_rate": 2.9934250346190818e-05,
      "loss": 0.2899,
      "step": 20100
    },
    {
      "epoch": 1.343196741750478,
      "grad_norm": 2.0007855892181396,
      "learning_rate": 2.9404080078817924e-05,
      "loss": 0.284,
      "step": 20200
    },
    {
      "epoch": 1.3498462305710248,
      "grad_norm": 1.0622339248657227,
      "learning_rate": 2.8876684177289404e-05,
      "loss": 0.2886,
      "step": 20300
    },
    {
      "epoch": 1.3564957193915719,
      "grad_norm": 1.351544737815857,
      "learning_rate": 2.8352133684276853e-05,
      "loss": 0.2738,
      "step": 20400
    },
    {
      "epoch": 1.3631452082121187,
      "grad_norm": 1.9642298221588135,
      "learning_rate": 2.7830499259162213e-05,
      "loss": 0.2981,
      "step": 20500
    },
    {
      "epoch": 1.3697946970326655,
      "grad_norm": 1.769395351409912,
      "learning_rate": 2.7311851168519496e-05,
      "loss": 0.3159,
      "step": 20600
    },
    {
      "epoch": 1.3764441858532126,
      "grad_norm": 3.744534492492676,
      "learning_rate": 2.6796259276649693e-05,
      "loss": 0.3019,
      "step": 20700
    },
    {
      "epoch": 1.3830936746737594,
      "grad_norm": 0.7485169768333435,
      "learning_rate": 2.6283793036169603e-05,
      "loss": 0.3029,
      "step": 20800
    },
    {
      "epoch": 1.3897431634943063,
      "grad_norm": 1.2193599939346313,
      "learning_rate": 2.5774521478656343e-05,
      "loss": 0.2799,
      "step": 20900
    },
    {
      "epoch": 1.3963926523148533,
      "grad_norm": 3.9265201091766357,
      "learning_rate": 2.52685132053484e-05,
      "loss": 0.2783,
      "step": 21000
    },
    {
      "epoch": 1.4030421411354002,
      "grad_norm": 1.0576444864273071,
      "learning_rate": 2.4765836377904787e-05,
      "loss": 0.3043,
      "step": 21100
    },
    {
      "epoch": 1.4096916299559472,
      "grad_norm": 1.704397201538086,
      "learning_rate": 2.4266558709223293e-05,
      "loss": 0.2823,
      "step": 21200
    },
    {
      "epoch": 1.416341118776494,
      "grad_norm": 5.58804178237915,
      "learning_rate": 2.377074745431931e-05,
      "loss": 0.2899,
      "step": 21300
    },
    {
      "epoch": 1.422990607597041,
      "grad_norm": 2.1619229316711426,
      "learning_rate": 2.3278469401266178e-05,
      "loss": 0.259,
      "step": 21400
    },
    {
      "epoch": 1.429640096417588,
      "grad_norm": 3.6991875171661377,
      "learning_rate": 2.2789790862198628e-05,
      "loss": 0.2508,
      "step": 21500
    },
    {
      "epoch": 1.4362895852381348,
      "grad_norm": 0.9216620922088623,
      "learning_rate": 2.2304777664380176e-05,
      "loss": 0.2701,
      "step": 21600
    },
    {
      "epoch": 1.4429390740586818,
      "grad_norm": 1.0544836521148682,
      "learning_rate": 2.182349514133583e-05,
      "loss": 0.2741,
      "step": 21700
    },
    {
      "epoch": 1.4495885628792287,
      "grad_norm": 1.3279638290405273,
      "learning_rate": 2.134600812405151e-05,
      "loss": 0.2608,
      "step": 21800
    },
    {
      "epoch": 1.4562380516997755,
      "grad_norm": 2.69641375541687,
      "learning_rate": 2.0872380932240832e-05,
      "loss": 0.2527,
      "step": 21900
    },
    {
      "epoch": 1.4628875405203225,
      "grad_norm": 1.3837428092956543,
      "learning_rate": 2.0402677365681112e-05,
      "loss": 0.2739,
      "step": 22000
    },
    {
      "epoch": 1.4695370293408694,
      "grad_norm": 2.859133720397949,
      "learning_rate": 1.99369606956191e-05,
      "loss": 0.2823,
      "step": 22100
    },
    {
      "epoch": 1.4761865181614162,
      "grad_norm": 2.0277671813964844,
      "learning_rate": 1.9475293656248182e-05,
      "loss": 0.2753,
      "step": 22200
    },
    {
      "epoch": 1.4828360069819633,
      "grad_norm": 1.3588882684707642,
      "learning_rate": 1.9017738436257655e-05,
      "loss": 0.2806,
      "step": 22300
    },
    {
      "epoch": 1.48948549580251,
      "grad_norm": 0.9804733991622925,
      "learning_rate": 1.8564356670455767e-05,
      "loss": 0.2749,
      "step": 22400
    },
    {
      "epoch": 1.4961349846230572,
      "grad_norm": 1.8412333726882935,
      "learning_rate": 1.8115209431467074e-05,
      "loss": 0.265,
      "step": 22500
    },
    {
      "epoch": 1.502784473443604,
      "grad_norm": 0.9384289979934692,
      "learning_rate": 1.767035722150582e-05,
      "loss": 0.264,
      "step": 22600
    },
    {
      "epoch": 1.509433962264151,
      "grad_norm": 1.8639055490493774,
      "learning_rate": 1.7229859964225868e-05,
      "loss": 0.2587,
      "step": 22700
    },
    {
      "epoch": 1.5160834510846979,
      "grad_norm": 1.7392961978912354,
      "learning_rate": 1.679377699664884e-05,
      "loss": 0.2638,
      "step": 22800
    },
    {
      "epoch": 1.5227329399052447,
      "grad_norm": 2.0196003913879395,
      "learning_rate": 1.6362167061171063e-05,
      "loss": 0.2718,
      "step": 22900
    },
    {
      "epoch": 1.5293824287257918,
      "grad_norm": 1.6492582559585571,
      "learning_rate": 1.5935088297650674e-05,
      "loss": 0.2553,
      "step": 23000
    },
    {
      "epoch": 1.5360319175463386,
      "grad_norm": 1.1431901454925537,
      "learning_rate": 1.551259823557602e-05,
      "loss": 0.2569,
      "step": 23100
    },
    {
      "epoch": 1.5426814063668854,
      "grad_norm": 1.3804975748062134,
      "learning_rate": 1.509475378631603e-05,
      "loss": 0.2474,
      "step": 23200
    },
    {
      "epoch": 1.5493308951874325,
      "grad_norm": 1.6143451929092407,
      "learning_rate": 1.468161123545413e-05,
      "loss": 0.2618,
      "step": 23300
    },
    {
      "epoch": 1.5559803840079793,
      "grad_norm": 2.3701083660125732,
      "learning_rate": 1.4273226235206178e-05,
      "loss": 0.2412,
      "step": 23400
    },
    {
      "epoch": 1.5626298728285262,
      "grad_norm": 0.5906326174736023,
      "learning_rate": 1.3869653796923993e-05,
      "loss": 0.234,
      "step": 23500
    },
    {
      "epoch": 1.5692793616490732,
      "grad_norm": 0.9088567495346069,
      "learning_rate": 1.3470948283684925e-05,
      "loss": 0.2529,
      "step": 23600
    },
    {
      "epoch": 1.5759288504696203,
      "grad_norm": 2.244845151901245,
      "learning_rate": 1.307716340296904e-05,
      "loss": 0.2444,
      "step": 23700
    },
    {
      "epoch": 1.5825783392901671,
      "grad_norm": 4.36794376373291,
      "learning_rate": 1.268835219942433e-05,
      "loss": 0.2533,
      "step": 23800
    },
    {
      "epoch": 1.589227828110714,
      "grad_norm": 2.707139730453491,
      "learning_rate": 1.23045670477215e-05,
      "loss": 0.2467,
      "step": 23900
    },
    {
      "epoch": 1.595877316931261,
      "grad_norm": 1.5950795412063599,
      "learning_rate": 1.1925859645498722e-05,
      "loss": 0.2349,
      "step": 24000
    },
    {
      "epoch": 1.6025268057518078,
      "grad_norm": 1.1506330966949463,
      "learning_rate": 1.1552281006397819e-05,
      "loss": 0.2518,
      "step": 24100
    },
    {
      "epoch": 1.6091762945723547,
      "grad_norm": 2.356943130493164,
      "learning_rate": 1.1183881453192479e-05,
      "loss": 0.2376,
      "step": 24200
    },
    {
      "epoch": 1.6158257833929017,
      "grad_norm": 1.1734323501586914,
      "learning_rate": 1.082071061100945e-05,
      "loss": 0.2277,
      "step": 24300
    },
    {
      "epoch": 1.6224752722134486,
      "grad_norm": 3.4108047485351562,
      "learning_rate": 1.0462817400643959e-05,
      "loss": 0.2313,
      "step": 24400
    },
    {
      "epoch": 1.6291247610339954,
      "grad_norm": 2.890157699584961,
      "learning_rate": 1.0110250031969709e-05,
      "loss": 0.2364,
      "step": 24500
    },
    {
      "epoch": 1.6357742498545425,
      "grad_norm": 1.4353581666946411,
      "learning_rate": 9.763055997444897e-06,
      "loss": 0.2401,
      "step": 24600
    },
    {
      "epoch": 1.6424237386750895,
      "grad_norm": 0.6841472387313843,
      "learning_rate": 9.421282065714676e-06,
      "loss": 0.2344,
      "step": 24700
    },
    {
      "epoch": 1.6490732274956361,
      "grad_norm": 2.0864779949188232,
      "learning_rate": 9.08497427531128e-06,
      "loss": 0.2186,
      "step": 24800
    },
    {
      "epoch": 1.6557227163161832,
      "grad_norm": 1.5929943323135376,
      "learning_rate": 8.754177928452328e-06,
      "loss": 0.2307,
      "step": 24900
    },
    {
      "epoch": 1.6623722051367302,
      "grad_norm": 0.5336441993713379,
      "learning_rate": 8.428937584938496e-06,
      "loss": 0.2401,
      "step": 25000
    },
    {
      "epoch": 1.669021693957277,
      "grad_norm": 2.829430103302002,
      "learning_rate": 8.109297056151067e-06,
      "loss": 0.2285,
      "step": 25100
    },
    {
      "epoch": 1.675671182777824,
      "grad_norm": 1.822135090827942,
      "learning_rate": 7.79529939915029e-06,
      "loss": 0.238,
      "step": 25200
    },
    {
      "epoch": 1.682320671598371,
      "grad_norm": 3.5365030765533447,
      "learning_rate": 7.486986910875499e-06,
      "loss": 0.2245,
      "step": 25300
    },
    {
      "epoch": 1.6889701604189178,
      "grad_norm": 3.5611343383789062,
      "learning_rate": 7.184401122447398e-06,
      "loss": 0.2344,
      "step": 25400
    },
    {
      "epoch": 1.6956196492394646,
      "grad_norm": 3.247079372406006,
      "learning_rate": 6.887582793573727e-06,
      "loss": 0.2393,
      "step": 25500
    },
    {
      "epoch": 1.7022691380600117,
      "grad_norm": 2.570157766342163,
      "learning_rate": 6.596571907058707e-06,
      "loss": 0.2238,
      "step": 25600
    },
    {
      "epoch": 1.7089186268805585,
      "grad_norm": 1.488516092300415,
      "learning_rate": 6.31140766341713e-06,
      "loss": 0.2326,
      "step": 25700
    },
    {
      "epoch": 1.7155681157011053,
      "grad_norm": 1.449952483177185,
      "learning_rate": 6.032128475593924e-06,
      "loss": 0.2196,
      "step": 25800
    },
    {
      "epoch": 1.7222176045216524,
      "grad_norm": 2.537747621536255,
      "learning_rate": 5.758771963789722e-06,
      "loss": 0.2065,
      "step": 25900
    },
    {
      "epoch": 1.7288670933421995,
      "grad_norm": 1.783013105392456,
      "learning_rate": 5.4913749503932575e-06,
      "loss": 0.2001,
      "step": 26000
    },
    {
      "epoch": 1.7355165821627463,
      "grad_norm": 2.2559077739715576,
      "learning_rate": 5.229973455021231e-06,
      "loss": 0.2183,
      "step": 26100
    },
    {
      "epoch": 1.7421660709832931,
      "grad_norm": 2.048455238342285,
      "learning_rate": 4.974602689666252e-06,
      "loss": 0.2106,
      "step": 26200
    },
    {
      "epoch": 1.7488155598038402,
      "grad_norm": 1.1111422777175903,
      "learning_rate": 4.725297053953692e-06,
      "loss": 0.2397,
      "step": 26300
    },
    {
      "epoch": 1.755465048624387,
      "grad_norm": 1.549456238746643,
      "learning_rate": 4.48209013050781e-06,
      "loss": 0.2209,
      "step": 26400
    },
    {
      "epoch": 1.7621145374449338,
      "grad_norm": 2.2229185104370117,
      "learning_rate": 4.245014680428117e-06,
      "loss": 0.211,
      "step": 26500
    },
    {
      "epoch": 1.768764026265481,
      "grad_norm": 3.065992593765259,
      "learning_rate": 4.014102638876205e-06,
      "loss": 0.2134,
      "step": 26600
    },
    {
      "epoch": 1.7754135150860277,
      "grad_norm": 1.1763139963150024,
      "learning_rate": 3.789385110774013e-06,
      "loss": 0.2095,
      "step": 26700
    },
    {
      "epoch": 1.7820630039065746,
      "grad_norm": 2.860395908355713,
      "learning_rate": 3.5708923666137927e-06,
      "loss": 0.238,
      "step": 26800
    },
    {
      "epoch": 1.7887124927271216,
      "grad_norm": 2.1108574867248535,
      "learning_rate": 3.358653838380571e-06,
      "loss": 0.2087,
      "step": 26900
    },
    {
      "epoch": 1.7953619815476687,
      "grad_norm": 0.8311635255813599,
      "learning_rate": 3.1526981155875156e-06,
      "loss": 0.2164,
      "step": 27000
    },
    {
      "epoch": 1.8020114703682153,
      "grad_norm": 0.8638597130775452,
      "learning_rate": 2.9530529414247608e-06,
      "loss": 0.2143,
      "step": 27100
    },
    {
      "epoch": 1.8086609591887624,
      "grad_norm": 3.8020806312561035,
      "learning_rate": 2.7597452090223354e-06,
      "loss": 0.2037,
      "step": 27200
    },
    {
      "epoch": 1.8153104480093094,
      "grad_norm": 4.387086868286133,
      "learning_rate": 2.572800957827476e-06,
      "loss": 0.2156,
      "step": 27300
    },
    {
      "epoch": 1.8219599368298562,
      "grad_norm": 1.4160668849945068,
      "learning_rate": 2.3922453700970295e-06,
      "loss": 0.2182,
      "step": 27400
    },
    {
      "epoch": 1.828609425650403,
      "grad_norm": 1.636501669883728,
      "learning_rate": 2.2181027675052534e-06,
      "loss": 0.2238,
      "step": 27500
    },
    {
      "epoch": 1.8352589144709501,
      "grad_norm": 2.2405593395233154,
      "learning_rate": 2.0503966078676217e-06,
      "loss": 0.2179,
      "step": 27600
    },
    {
      "epoch": 1.841908403291497,
      "grad_norm": 1.6398348808288574,
      "learning_rate": 1.8891494819808841e-06,
      "loss": 0.2149,
      "step": 27700
    },
    {
      "epoch": 1.8485578921120438,
      "grad_norm": 3.231600522994995,
      "learning_rate": 1.7343831105800511e-06,
      "loss": 0.2071,
      "step": 27800
    },
    {
      "epoch": 1.8552073809325909,
      "grad_norm": 2.70027232170105,
      "learning_rate": 1.5861183414124403e-06,
      "loss": 0.2132,
      "step": 27900
    },
    {
      "epoch": 1.8618568697531377,
      "grad_norm": 2.990867853164673,
      "learning_rate": 1.4443751464294664e-06,
      "loss": 0.2097,
      "step": 28000
    },
    {
      "epoch": 1.8685063585736845,
      "grad_norm": 1.75977623462677,
      "learning_rate": 1.3091726190962329e-06,
      "loss": 0.1923,
      "step": 28100
    },
    {
      "epoch": 1.8751558473942316,
      "grad_norm": 0.49737560749053955,
      "learning_rate": 1.18052897181965e-06,
      "loss": 0.189,
      "step": 28200
    },
    {
      "epoch": 1.8818053362147786,
      "grad_norm": 1.1605116128921509,
      "learning_rate": 1.0584615334950643e-06,
      "loss": 0.2089,
      "step": 28300
    },
    {
      "epoch": 1.8884548250353252,
      "grad_norm": 2.284116268157959,
      "learning_rate": 9.429867471720255e-07,
      "loss": 0.2045,
      "step": 28400
    },
    {
      "epoch": 1.8951043138558723,
      "grad_norm": 2.2018344402313232,
      "learning_rate": 8.341201678392974e-07,
      "loss": 0.2235,
      "step": 28500
    },
    {
      "epoch": 1.9017538026764194,
      "grad_norm": 1.3855736255645752,
      "learning_rate": 7.318764603295447e-07,
      "loss": 0.2037,
      "step": 28600
    },
    {
      "epoch": 1.9084032914969662,
      "grad_norm": 2.4615190029144287,
      "learning_rate": 6.362693973439193e-07,
      "loss": 0.2108,
      "step": 28700
    },
    {
      "epoch": 1.915052780317513,
      "grad_norm": 2.775005578994751,
      "learning_rate": 5.47311857596794e-07,
      "loss": 0.1984,
      "step": 28800
    },
    {
      "epoch": 1.92170226913806,
      "grad_norm": 3.7345826625823975,
      "learning_rate": 4.6501582408096657e-07,
      "loss": 0.2124,
      "step": 28900
    },
    {
      "epoch": 1.928351757958607,
      "grad_norm": 0.46711668372154236,
      "learning_rate": 3.893923824534629e-07,
      "loss": 0.1897,
      "step": 29000
    },
    {
      "epoch": 1.9350012467791537,
      "grad_norm": 2.135247230529785,
      "learning_rate": 3.204517195422696e-07,
      "loss": 0.2267,
      "step": 29100
    },
    {
      "epoch": 1.9416507355997008,
      "grad_norm": 0.9539075493812561,
      "learning_rate": 2.5820312197411543e-07,
      "loss": 0.2104,
      "step": 29200
    },
    {
      "epoch": 1.9483002244202476,
      "grad_norm": 3.765820026397705,
      "learning_rate": 2.0265497492352735e-07,
      "loss": 0.2072,
      "step": 29300
    },
    {
      "epoch": 1.9549497132407945,
      "grad_norm": 2.848856210708618,
      "learning_rate": 1.538147609832896e-07,
      "loss": 0.2018,
      "step": 29400
    },
    {
      "epoch": 1.9615992020613415,
      "grad_norm": 1.4174009561538696,
      "learning_rate": 1.1168905915652228e-07,
      "loss": 0.1963,
      "step": 29500
    },
    {
      "epoch": 1.9682486908818886,
      "grad_norm": 3.067931652069092,
      "learning_rate": 7.628354397045123e-08,
      "loss": 0.2062,
      "step": 29600
    },
    {
      "epoch": 1.9748981797024354,
      "grad_norm": 2.1927988529205322,
      "learning_rate": 4.760298471201963e-08,
      "loss": 0.2007,
      "step": 29700
    },
    {
      "epoch": 1.9815476685229823,
      "grad_norm": 2.1186487674713135,
      "learning_rate": 2.565124478545733e-08,
      "loss": 0.2119,
      "step": 29800
    },
    {
      "epoch": 1.9881971573435293,
      "grad_norm": 1.58423912525177,
      "learning_rate": 1.043128119184167e-08,
      "loss": 0.2192,
      "step": 29900
    },
    {
      "epoch": 1.9948466461640761,
      "grad_norm": 1.9130926132202148,
      "learning_rate": 1.945144130788279e-09,
      "loss": 0.2238,
      "step": 30000
    },
    {
      "epoch": 1.9948466461640761,
      "eval_loss": 0.23713438212871552,
      "eval_runtime": 4319.7371,
      "eval_samples_per_second": 1.547,
      "eval_steps_per_second": 1.547,
      "step": 30000
    },
    {
      "epoch": 1.9999002576676919,
      "step": 30076,
      "total_flos": 1.2039888836550205e+19,
      "train_loss": 0.385877049085006,
      "train_runtime": 246042.9971,
      "train_samples_per_second": 0.489,
      "train_steps_per_second": 0.122
    }
  ],
  "logging_steps": 100,
  "max_steps": 30076,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 10000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.2039888836550205e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}