{
  "best_metric": 0.6936118602752686,
  "best_model_checkpoint": "./wav2vec-bert-korean-dialect-recognition/checkpoint-294606",
  "epoch": 10.0,
  "eval_steps": 500,
  "global_step": 327340,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0030549275982159224,
      "grad_norm": 1.5348891019821167,
      "learning_rate": 1e-05,
      "loss": 1.7854,
      "step": 100
    },
    {
      "epoch": 0.006109855196431845,
      "grad_norm": 1.399305820465088,
      "learning_rate": 2e-05,
      "loss": 1.7564,
      "step": 200
    },
    {
      "epoch": 0.009164782794647768,
      "grad_norm": 1.1098747253417969,
      "learning_rate": 3e-05,
      "loss": 1.7385,
      "step": 300
    },
    {
      "epoch": 0.01221971039286369,
      "grad_norm": 0.9315665364265442,
      "learning_rate": 4e-05,
      "loss": 1.6674,
      "step": 400
    },
    {
      "epoch": 0.015274637991079611,
      "grad_norm": 1.0861330032348633,
      "learning_rate": 5e-05,
      "loss": 1.7178,
      "step": 500
    },
    {
      "epoch": 0.018329565589295535,
      "grad_norm": 1.7394012212753296,
      "learning_rate": 4.9984701994859874e-05,
      "loss": 1.6606,
      "step": 600
    },
    {
      "epoch": 0.021384493187511455,
      "grad_norm": 1.3549752235412598,
      "learning_rate": 4.9969403989719745e-05,
      "loss": 1.6554,
      "step": 700
    },
    {
      "epoch": 0.02443942078572738,
      "grad_norm": 1.4596511125564575,
      "learning_rate": 4.9954105984579616e-05,
      "loss": 1.6682,
      "step": 800
    },
    {
      "epoch": 0.0274943483839433,
      "grad_norm": 1.1955530643463135,
      "learning_rate": 4.993880797943949e-05,
      "loss": 1.671,
      "step": 900
    },
    {
      "epoch": 0.030549275982159223,
      "grad_norm": 0.7402735948562622,
      "learning_rate": 4.992350997429935e-05,
      "loss": 1.6168,
      "step": 1000
    },
    {
      "epoch": 0.03360420358037514,
      "grad_norm": 1.1003531217575073,
      "learning_rate": 4.990821196915922e-05,
      "loss": 1.5873,
      "step": 1100
    },
    {
      "epoch": 0.03665913117859107,
      "grad_norm": 2.4431450366973877,
      "learning_rate": 4.9892913964019094e-05,
      "loss": 1.6629,
      "step": 1200
    },
    {
      "epoch": 0.03971405877680699,
      "grad_norm": 1.907747507095337,
      "learning_rate": 4.9877615958878965e-05,
      "loss": 1.6859,
      "step": 1300
    },
    {
      "epoch": 0.04276898637502291,
      "grad_norm": 1.5831812620162964,
      "learning_rate": 4.9862317953738836e-05,
      "loss": 1.6506,
      "step": 1400
    },
    {
      "epoch": 0.04582391397323884,
      "grad_norm": 2.0335052013397217,
      "learning_rate": 4.984701994859871e-05,
      "loss": 1.6158,
      "step": 1500
    },
    {
      "epoch": 0.04887884157145476,
      "grad_norm": 1.364405632019043,
      "learning_rate": 4.983172194345858e-05,
      "loss": 1.5788,
      "step": 1600
    },
    {
      "epoch": 0.05193376916967068,
      "grad_norm": 0.8212212920188904,
      "learning_rate": 4.981642393831845e-05,
      "loss": 1.6158,
      "step": 1700
    },
    {
      "epoch": 0.0549886967678866,
      "grad_norm": 1.3664904832839966,
      "learning_rate": 4.9801125933178314e-05,
      "loss": 1.6436,
      "step": 1800
    },
    {
      "epoch": 0.058043624366102525,
      "grad_norm": 2.1316192150115967,
      "learning_rate": 4.9785827928038185e-05,
      "loss": 1.6091,
      "step": 1900
    },
    {
      "epoch": 0.061098551964318445,
      "grad_norm": 1.4975366592407227,
      "learning_rate": 4.9770529922898056e-05,
      "loss": 1.6547,
      "step": 2000
    },
    {
      "epoch": 0.06415347956253437,
      "grad_norm": 2.3204729557037354,
      "learning_rate": 4.9755384897809324e-05,
      "loss": 1.5825,
      "step": 2100
    },
    {
      "epoch": 0.06720840716075029,
      "grad_norm": 0.7949225306510925,
      "learning_rate": 4.9740086892669195e-05,
      "loss": 1.5964,
      "step": 2200
    },
    {
      "epoch": 0.0702633347589662,
      "grad_norm": 2.1805293560028076,
      "learning_rate": 4.9724788887529066e-05,
      "loss": 1.6642,
      "step": 2300
    },
    {
      "epoch": 0.07331826235718214,
      "grad_norm": 2.7045323848724365,
      "learning_rate": 4.970949088238894e-05,
      "loss": 1.6367,
      "step": 2400
    },
    {
      "epoch": 0.07637318995539806,
      "grad_norm": 1.8262252807617188,
      "learning_rate": 4.969419287724881e-05,
      "loss": 1.591,
      "step": 2500
    },
    {
      "epoch": 0.07942811755361398,
      "grad_norm": 2.541649341583252,
      "learning_rate": 4.967889487210868e-05,
      "loss": 1.5751,
      "step": 2600
    },
    {
      "epoch": 0.0824830451518299,
      "grad_norm": 2.23138427734375,
      "learning_rate": 4.966359686696855e-05,
      "loss": 1.5997,
      "step": 2700
    },
    {
      "epoch": 0.08553797275004582,
      "grad_norm": 1.8299692869186401,
      "learning_rate": 4.9648298861828415e-05,
      "loss": 1.594,
      "step": 2800
    },
    {
      "epoch": 0.08859290034826174,
      "grad_norm": 1.8534067869186401,
      "learning_rate": 4.963300085668829e-05,
      "loss": 1.6269,
      "step": 2900
    },
    {
      "epoch": 0.09164782794647768,
      "grad_norm": 1.8997411727905273,
      "learning_rate": 4.961770285154816e-05,
      "loss": 1.6089,
      "step": 3000
    },
    {
      "epoch": 0.0947027555446936,
      "grad_norm": 2.0725138187408447,
      "learning_rate": 4.960240484640803e-05,
      "loss": 1.5926,
      "step": 3100
    },
    {
      "epoch": 0.09775768314290952,
      "grad_norm": 1.6708780527114868,
      "learning_rate": 4.95871068412679e-05,
      "loss": 1.6021,
      "step": 3200
    },
    {
      "epoch": 0.10081261074112544,
      "grad_norm": 1.2676891088485718,
      "learning_rate": 4.957180883612777e-05,
      "loss": 1.594,
      "step": 3300
    },
    {
      "epoch": 0.10386753833934136,
      "grad_norm": 2.105405807495117,
      "learning_rate": 4.955651083098764e-05,
      "loss": 1.58,
      "step": 3400
    },
    {
      "epoch": 0.10692246593755728,
      "grad_norm": 0.9499572515487671,
      "learning_rate": 4.9541212825847514e-05,
      "loss": 1.5693,
      "step": 3500
    },
    {
      "epoch": 0.1099773935357732,
      "grad_norm": 1.4242860078811646,
      "learning_rate": 4.952591482070738e-05,
      "loss": 1.6005,
      "step": 3600
    },
    {
      "epoch": 0.11303232113398913,
      "grad_norm": 2.2109808921813965,
      "learning_rate": 4.951061681556725e-05,
      "loss": 1.596,
      "step": 3700
    },
    {
      "epoch": 0.11608724873220505,
      "grad_norm": 1.9320544004440308,
      "learning_rate": 4.949531881042712e-05,
      "loss": 1.5385,
      "step": 3800
    },
    {
      "epoch": 0.11914217633042097,
      "grad_norm": 2.3479628562927246,
      "learning_rate": 4.948002080528699e-05,
      "loss": 1.6029,
      "step": 3900
    },
    {
      "epoch": 0.12219710392863689,
      "grad_norm": 2.0376906394958496,
      "learning_rate": 4.946472280014686e-05,
      "loss": 1.5878,
      "step": 4000
    },
    {
      "epoch": 0.12525203152685283,
      "grad_norm": 1.2168834209442139,
      "learning_rate": 4.944957777505814e-05,
      "loss": 1.557,
      "step": 4100
    },
    {
      "epoch": 0.12830695912506873,
      "grad_norm": 1.6018855571746826,
      "learning_rate": 4.943427976991801e-05,
      "loss": 1.5317,
      "step": 4200
    },
    {
      "epoch": 0.13136188672328467,
      "grad_norm": 1.5891162157058716,
      "learning_rate": 4.941898176477788e-05,
      "loss": 1.5615,
      "step": 4300
    },
    {
      "epoch": 0.13441681432150057,
      "grad_norm": 1.4518991708755493,
      "learning_rate": 4.940368375963775e-05,
      "loss": 1.5113,
      "step": 4400
    },
    {
      "epoch": 0.1374717419197165,
      "grad_norm": 1.9849175214767456,
      "learning_rate": 4.9388385754497615e-05,
      "loss": 1.6008,
      "step": 4500
    },
    {
      "epoch": 0.1405266695179324,
      "grad_norm": 1.123494267463684,
      "learning_rate": 4.9373087749357486e-05,
      "loss": 1.541,
      "step": 4600
    },
    {
      "epoch": 0.14358159711614835,
      "grad_norm": 2.3314383029937744,
      "learning_rate": 4.935778974421736e-05,
      "loss": 1.5266,
      "step": 4700
    },
    {
      "epoch": 0.14663652471436428,
      "grad_norm": 0.6723255515098572,
      "learning_rate": 4.934249173907723e-05,
      "loss": 1.5489,
      "step": 4800
    },
    {
      "epoch": 0.1496914523125802,
      "grad_norm": 1.7696168422698975,
      "learning_rate": 4.93271937339371e-05,
      "loss": 1.5782,
      "step": 4900
    },
    {
      "epoch": 0.15274637991079612,
      "grad_norm": 2.4960386753082275,
      "learning_rate": 4.931189572879697e-05,
      "loss": 1.5567,
      "step": 5000
    },
    {
      "epoch": 0.15580130750901203,
      "grad_norm": 0.6923491954803467,
      "learning_rate": 4.929659772365684e-05,
      "loss": 1.5766,
      "step": 5100
    },
    {
      "epoch": 0.15885623510722796,
      "grad_norm": 1.6389570236206055,
      "learning_rate": 4.928129971851671e-05,
      "loss": 1.568,
      "step": 5200
    },
    {
      "epoch": 0.16191116270544387,
      "grad_norm": 1.4275109767913818,
      "learning_rate": 4.926600171337658e-05,
      "loss": 1.5785,
      "step": 5300
    },
    {
      "epoch": 0.1649660903036598,
      "grad_norm": 0.9533630013465881,
      "learning_rate": 4.925070370823645e-05,
      "loss": 1.5425,
      "step": 5400
    },
    {
      "epoch": 0.16802101790187574,
      "grad_norm": 2.8133769035339355,
      "learning_rate": 4.923540570309632e-05,
      "loss": 1.5054,
      "step": 5500
    },
    {
      "epoch": 0.17107594550009164,
      "grad_norm": 1.8214422464370728,
      "learning_rate": 4.922010769795619e-05,
      "loss": 1.5494,
      "step": 5600
    },
    {
      "epoch": 0.17413087309830758,
      "grad_norm": 1.1751452684402466,
      "learning_rate": 4.920480969281606e-05,
      "loss": 1.5616,
      "step": 5700
    },
    {
      "epoch": 0.17718580069652348,
      "grad_norm": 1.7518926858901978,
      "learning_rate": 4.9189511687675933e-05,
      "loss": 1.4986,
      "step": 5800
    },
    {
      "epoch": 0.18024072829473942,
      "grad_norm": 1.4943835735321045,
      "learning_rate": 4.9174213682535805e-05,
      "loss": 1.539,
      "step": 5900
    },
    {
      "epoch": 0.18329565589295535,
      "grad_norm": 1.664491891860962,
      "learning_rate": 4.9158915677395676e-05,
      "loss": 1.5681,
      "step": 6000
    },
    {
      "epoch": 0.18635058349117126,
      "grad_norm": 1.8616182804107666,
      "learning_rate": 4.9143770652306944e-05,
      "loss": 1.4818,
      "step": 6100
    },
    {
      "epoch": 0.1894055110893872,
      "grad_norm": 1.5610833168029785,
      "learning_rate": 4.9128472647166815e-05,
      "loss": 1.5125,
      "step": 6200
    },
    {
      "epoch": 0.1924604386876031,
      "grad_norm": 3.129401445388794,
      "learning_rate": 4.911317464202668e-05,
      "loss": 1.4901,
      "step": 6300
    },
    {
      "epoch": 0.19551536628581903,
      "grad_norm": 1.6080937385559082,
      "learning_rate": 4.909787663688655e-05,
      "loss": 1.4938,
      "step": 6400
    },
    {
      "epoch": 0.19857029388403494,
      "grad_norm": 1.7884577512741089,
      "learning_rate": 4.908257863174642e-05,
      "loss": 1.4891,
      "step": 6500
    },
    {
      "epoch": 0.20162522148225087,
      "grad_norm": 0.6538684964179993,
      "learning_rate": 4.906728062660629e-05,
      "loss": 1.5345,
      "step": 6600
    },
    {
      "epoch": 0.2046801490804668,
      "grad_norm": 0.9260094165802002,
      "learning_rate": 4.9051982621466164e-05,
      "loss": 1.5286,
      "step": 6700
    },
    {
      "epoch": 0.2077350766786827,
      "grad_norm": 1.738603949546814,
      "learning_rate": 4.9036684616326035e-05,
      "loss": 1.458,
      "step": 6800
    },
    {
      "epoch": 0.21079000427689865,
      "grad_norm": 0.5729855895042419,
      "learning_rate": 4.9021386611185906e-05,
      "loss": 1.4975,
      "step": 6900
    },
    {
      "epoch": 0.21384493187511455,
      "grad_norm": 2.177529811859131,
      "learning_rate": 4.900608860604578e-05,
      "loss": 1.4874,
      "step": 7000
    },
    {
      "epoch": 0.2168998594733305,
      "grad_norm": 2.3976924419403076,
      "learning_rate": 4.899079060090564e-05,
      "loss": 1.5065,
      "step": 7100
    },
    {
      "epoch": 0.2199547870715464,
      "grad_norm": 1.6993240118026733,
      "learning_rate": 4.897549259576551e-05,
      "loss": 1.449,
      "step": 7200
    },
    {
      "epoch": 0.22300971466976233,
      "grad_norm": 1.3176172971725464,
      "learning_rate": 4.8960194590625384e-05,
      "loss": 1.4667,
      "step": 7300
    },
    {
      "epoch": 0.22606464226797826,
      "grad_norm": 0.5767053365707397,
      "learning_rate": 4.8944896585485255e-05,
      "loss": 1.5477,
      "step": 7400
    },
    {
      "epoch": 0.22911956986619417,
      "grad_norm": 2.0425851345062256,
      "learning_rate": 4.8929598580345126e-05,
      "loss": 1.4457,
      "step": 7500
    },
    {
      "epoch": 0.2321744974644101,
      "grad_norm": 2.019169807434082,
      "learning_rate": 4.8914300575205e-05,
      "loss": 1.4033,
      "step": 7600
    },
    {
      "epoch": 0.235229425062626,
      "grad_norm": 1.443758487701416,
      "learning_rate": 4.889900257006487e-05,
      "loss": 1.4599,
      "step": 7700
    },
    {
      "epoch": 0.23828435266084194,
      "grad_norm": 2.3572771549224854,
      "learning_rate": 4.888370456492474e-05,
      "loss": 1.4191,
      "step": 7800
    },
    {
      "epoch": 0.24133928025905785,
      "grad_norm": 2.7198643684387207,
      "learning_rate": 4.8868406559784604e-05,
      "loss": 1.3884,
      "step": 7900
    },
    {
      "epoch": 0.24439420785727378,
      "grad_norm": 1.7861402034759521,
      "learning_rate": 4.8853108554644475e-05,
      "loss": 1.5456,
      "step": 8000
    },
    {
      "epoch": 0.24744913545548972,
      "grad_norm": 1.8129621744155884,
      "learning_rate": 4.883796352955575e-05,
      "loss": 1.4116,
      "step": 8100
    },
    {
      "epoch": 0.25050406305370565,
      "grad_norm": 2.0511136054992676,
      "learning_rate": 4.8822665524415615e-05,
      "loss": 1.3446,
      "step": 8200
    },
    {
      "epoch": 0.25355899065192156,
      "grad_norm": 3.2969486713409424,
      "learning_rate": 4.8807367519275486e-05,
      "loss": 1.4883,
      "step": 8300
    },
    {
      "epoch": 0.25661391825013746,
      "grad_norm": 0.9347625374794006,
      "learning_rate": 4.879206951413536e-05,
      "loss": 1.4372,
      "step": 8400
    },
    {
      "epoch": 0.25966884584835337,
      "grad_norm": 4.156298637390137,
      "learning_rate": 4.877677150899523e-05,
      "loss": 1.5377,
      "step": 8500
    },
    {
      "epoch": 0.26272377344656933,
      "grad_norm": 3.0619304180145264,
      "learning_rate": 4.87614735038551e-05,
      "loss": 1.509,
      "step": 8600
    },
    {
      "epoch": 0.26577870104478524,
      "grad_norm": 1.6035257577896118,
      "learning_rate": 4.874617549871497e-05,
      "loss": 1.4709,
      "step": 8700
    },
    {
      "epoch": 0.26883362864300114,
      "grad_norm": 2.9043540954589844,
      "learning_rate": 4.873087749357484e-05,
      "loss": 1.4119,
      "step": 8800
    },
    {
      "epoch": 0.2718885562412171,
      "grad_norm": 2.1844189167022705,
      "learning_rate": 4.871557948843471e-05,
      "loss": 1.4593,
      "step": 8900
    },
    {
      "epoch": 0.274943483839433,
      "grad_norm": 1.494769811630249,
      "learning_rate": 4.870028148329458e-05,
      "loss": 1.3733,
      "step": 9000
    },
    {
      "epoch": 0.2779984114376489,
      "grad_norm": 1.0596073865890503,
      "learning_rate": 4.868498347815445e-05,
      "loss": 1.4862,
      "step": 9100
    },
    {
      "epoch": 0.2810533390358648,
      "grad_norm": 2.1339333057403564,
      "learning_rate": 4.866968547301432e-05,
      "loss": 1.4308,
      "step": 9200
    },
    {
      "epoch": 0.2841082666340808,
      "grad_norm": 2.508408546447754,
      "learning_rate": 4.865438746787419e-05,
      "loss": 1.4522,
      "step": 9300
    },
    {
      "epoch": 0.2871631942322967,
      "grad_norm": 1.4985188245773315,
      "learning_rate": 4.863908946273406e-05,
      "loss": 1.4005,
      "step": 9400
    },
    {
      "epoch": 0.2902181218305126,
      "grad_norm": 1.525770902633667,
      "learning_rate": 4.862379145759393e-05,
      "loss": 1.4459,
      "step": 9500
    },
    {
      "epoch": 0.29327304942872856,
      "grad_norm": 2.6566789150238037,
      "learning_rate": 4.8608493452453804e-05,
      "loss": 1.4367,
      "step": 9600
    },
    {
      "epoch": 0.29632797702694447,
      "grad_norm": 2.781080722808838,
      "learning_rate": 4.8593195447313675e-05,
      "loss": 1.4287,
      "step": 9700
    },
    {
      "epoch": 0.2993829046251604,
      "grad_norm": 0.9105027914047241,
      "learning_rate": 4.857789744217354e-05,
      "loss": 1.4311,
      "step": 9800
    },
    {
      "epoch": 0.3024378322233763,
      "grad_norm": 1.959030032157898,
      "learning_rate": 4.856259943703341e-05,
      "loss": 1.4048,
      "step": 9900
    },
    {
      "epoch": 0.30549275982159224,
      "grad_norm": 3.855910301208496,
      "learning_rate": 4.854730143189328e-05,
      "loss": 1.3622,
      "step": 10000
    },
    {
      "epoch": 0.30854768741980815,
      "grad_norm": 5.930082321166992,
      "learning_rate": 4.8532156406804557e-05,
      "loss": 1.3551,
      "step": 10100
    },
    {
      "epoch": 0.31160261501802405,
      "grad_norm": 2.937746286392212,
      "learning_rate": 4.851685840166443e-05,
      "loss": 1.3583,
      "step": 10200
    },
    {
      "epoch": 0.31465754261624,
      "grad_norm": 3.188389301300049,
      "learning_rate": 4.85015603965243e-05,
      "loss": 1.4423,
      "step": 10300
    },
    {
      "epoch": 0.3177124702144559,
      "grad_norm": 1.9529550075531006,
      "learning_rate": 4.848626239138417e-05,
      "loss": 1.4067,
      "step": 10400
    },
    {
      "epoch": 0.32076739781267183,
      "grad_norm": 4.006279945373535,
      "learning_rate": 4.847096438624404e-05,
      "loss": 1.417,
      "step": 10500
    },
    {
      "epoch": 0.32382232541088773,
      "grad_norm": 1.2116618156433105,
      "learning_rate": 4.845566638110391e-05,
      "loss": 1.4318,
      "step": 10600
    },
    {
      "epoch": 0.3268772530091037,
      "grad_norm": 1.4079272747039795,
      "learning_rate": 4.844036837596378e-05,
      "loss": 1.3835,
      "step": 10700
    },
    {
      "epoch": 0.3299321806073196,
      "grad_norm": 4.599946975708008,
      "learning_rate": 4.842507037082364e-05,
      "loss": 1.4411,
      "step": 10800
    },
    {
      "epoch": 0.3329871082055355,
      "grad_norm": 5.579395294189453,
      "learning_rate": 4.840977236568351e-05,
      "loss": 1.3978,
      "step": 10900
    },
    {
      "epoch": 0.33604203580375147,
      "grad_norm": 1.4071085453033447,
      "learning_rate": 4.839447436054338e-05,
      "loss": 1.3574,
      "step": 11000
    },
    {
      "epoch": 0.3390969634019674,
      "grad_norm": 1.6776551008224487,
      "learning_rate": 4.8379176355403255e-05,
      "loss": 1.4283,
      "step": 11100
    },
    {
      "epoch": 0.3421518910001833,
      "grad_norm": 4.477878093719482,
      "learning_rate": 4.8363878350263126e-05,
      "loss": 1.4258,
      "step": 11200
    },
    {
      "epoch": 0.34520681859839925,
      "grad_norm": 0.5587853789329529,
      "learning_rate": 4.8348580345123e-05,
      "loss": 1.4444,
      "step": 11300
    },
    {
      "epoch": 0.34826174619661515,
      "grad_norm": 4.845232963562012,
      "learning_rate": 4.833328233998287e-05,
      "loss": 1.3144,
      "step": 11400
    },
    {
      "epoch": 0.35131667379483106,
      "grad_norm": 3.292431592941284,
      "learning_rate": 4.831798433484274e-05,
      "loss": 1.3466,
      "step": 11500
    },
    {
      "epoch": 0.35437160139304696,
      "grad_norm": 2.4652178287506104,
      "learning_rate": 4.8302686329702604e-05,
      "loss": 1.3643,
      "step": 11600
    },
    {
      "epoch": 0.3574265289912629,
      "grad_norm": 2.7705917358398438,
      "learning_rate": 4.8287388324562475e-05,
      "loss": 1.4231,
      "step": 11700
    },
    {
      "epoch": 0.36048145658947883,
      "grad_norm": 3.021454334259033,
      "learning_rate": 4.8272090319422346e-05,
      "loss": 1.4229,
      "step": 11800
    },
    {
      "epoch": 0.36353638418769474,
      "grad_norm": 3.1049673557281494,
      "learning_rate": 4.825679231428222e-05,
      "loss": 1.3907,
      "step": 11900
    },
    {
      "epoch": 0.3665913117859107,
      "grad_norm": 2.0627031326293945,
      "learning_rate": 4.824149430914209e-05,
      "loss": 1.3381,
      "step": 12000
    },
    {
      "epoch": 0.3696462393841266,
      "grad_norm": 2.4559926986694336,
      "learning_rate": 4.822619630400196e-05,
      "loss": 1.3848,
      "step": 12100
    },
    {
      "epoch": 0.3727011669823425,
      "grad_norm": 3.559675693511963,
      "learning_rate": 4.8211051278913234e-05,
      "loss": 1.3528,
      "step": 12200
    },
    {
      "epoch": 0.3757560945805584,
      "grad_norm": 2.9225029945373535,
      "learning_rate": 4.8195753273773105e-05,
      "loss": 1.359,
      "step": 12300
    },
    {
      "epoch": 0.3788110221787744,
      "grad_norm": 2.4771335124969482,
      "learning_rate": 4.8180455268632976e-05,
      "loss": 1.3831,
      "step": 12400
    },
    {
      "epoch": 0.3818659497769903,
      "grad_norm": 2.1963953971862793,
      "learning_rate": 4.816515726349284e-05,
      "loss": 1.3522,
      "step": 12500
    },
    {
      "epoch": 0.3849208773752062,
      "grad_norm": 2.0889718532562256,
      "learning_rate": 4.814985925835271e-05,
      "loss": 1.3587,
      "step": 12600
    },
    {
      "epoch": 0.38797580497342216,
      "grad_norm": 2.777860641479492,
      "learning_rate": 4.813456125321258e-05,
      "loss": 1.3519,
      "step": 12700
    },
    {
      "epoch": 0.39103073257163806,
      "grad_norm": 2.518688201904297,
      "learning_rate": 4.8119263248072454e-05,
      "loss": 1.43,
      "step": 12800
    },
    {
      "epoch": 0.39408566016985397,
      "grad_norm": 2.9270553588867188,
      "learning_rate": 4.8103965242932325e-05,
      "loss": 1.3488,
      "step": 12900
    },
    {
      "epoch": 0.3971405877680699,
      "grad_norm": 2.321699857711792,
      "learning_rate": 4.8088667237792197e-05,
      "loss": 1.3405,
      "step": 13000
    },
    {
      "epoch": 0.40019551536628584,
      "grad_norm": 3.4953596591949463,
      "learning_rate": 4.807336923265207e-05,
      "loss": 1.2441,
      "step": 13100
    },
    {
      "epoch": 0.40325044296450174,
      "grad_norm": 2.8772597312927246,
      "learning_rate": 4.805807122751194e-05,
      "loss": 1.3981,
      "step": 13200
    },
    {
      "epoch": 0.40630537056271765,
      "grad_norm": 1.0560011863708496,
      "learning_rate": 4.80427732223718e-05,
      "loss": 1.3345,
      "step": 13300
    },
    {
      "epoch": 0.4093602981609336,
      "grad_norm": 3.9890592098236084,
      "learning_rate": 4.8027475217231674e-05,
      "loss": 1.2633,
      "step": 13400
    },
    {
      "epoch": 0.4124152257591495,
      "grad_norm": 4.9667816162109375,
      "learning_rate": 4.8012177212091546e-05,
      "loss": 1.3852,
      "step": 13500
    },
    {
      "epoch": 0.4154701533573654,
      "grad_norm": 4.832125186920166,
      "learning_rate": 4.799687920695142e-05,
      "loss": 1.3617,
      "step": 13600
    },
    {
      "epoch": 0.41852508095558133,
      "grad_norm": 1.3396093845367432,
      "learning_rate": 4.798158120181129e-05,
      "loss": 1.348,
      "step": 13700
    },
    {
      "epoch": 0.4215800085537973,
      "grad_norm": 2.829025983810425,
      "learning_rate": 4.796628319667116e-05,
      "loss": 1.3474,
      "step": 13800
    },
    {
      "epoch": 0.4246349361520132,
      "grad_norm": 4.8995280265808105,
      "learning_rate": 4.795098519153103e-05,
      "loss": 1.3866,
      "step": 13900
    },
    {
      "epoch": 0.4276898637502291,
      "grad_norm": 1.4323627948760986,
      "learning_rate": 4.79356871863909e-05,
      "loss": 1.2936,
      "step": 14000
    },
    {
      "epoch": 0.43074479134844507,
      "grad_norm": 1.9841852188110352,
      "learning_rate": 4.7920389181250766e-05,
      "loss": 1.4257,
      "step": 14100
    },
    {
      "epoch": 0.433799718946661,
      "grad_norm": 1.8389599323272705,
      "learning_rate": 4.790524415616204e-05,
      "loss": 1.3501,
      "step": 14200
    },
    {
      "epoch": 0.4368546465448769,
      "grad_norm": 2.149446487426758,
      "learning_rate": 4.788994615102191e-05,
      "loss": 1.3285,
      "step": 14300
    },
    {
      "epoch": 0.4399095741430928,
      "grad_norm": 2.1945650577545166,
      "learning_rate": 4.7874648145881776e-05,
      "loss": 1.375,
      "step": 14400
    },
    {
      "epoch": 0.44296450174130875,
      "grad_norm": 3.0235095024108887,
      "learning_rate": 4.785935014074165e-05,
      "loss": 1.3076,
      "step": 14500
    },
    {
      "epoch": 0.44601942933952465,
      "grad_norm": 1.4999414682388306,
      "learning_rate": 4.784405213560152e-05,
      "loss": 1.4213,
      "step": 14600
    },
    {
      "epoch": 0.44907435693774056,
      "grad_norm": 2.6008243560791016,
      "learning_rate": 4.782875413046139e-05,
      "loss": 1.3755,
      "step": 14700
    },
    {
      "epoch": 0.4521292845359565,
      "grad_norm": 3.3716583251953125,
      "learning_rate": 4.781345612532126e-05,
      "loss": 1.3439,
      "step": 14800
    },
    {
      "epoch": 0.45518421213417243,
      "grad_norm": 1.567351222038269,
      "learning_rate": 4.779815812018113e-05,
      "loss": 1.2619,
      "step": 14900
    },
    {
      "epoch": 0.45823913973238833,
      "grad_norm": 4.242433547973633,
      "learning_rate": 4.7782860115041e-05,
      "loss": 1.3884,
      "step": 15000
    },
    {
      "epoch": 0.46129406733060424,
      "grad_norm": 5.024165153503418,
      "learning_rate": 4.7767562109900874e-05,
      "loss": 1.2998,
      "step": 15100
    },
    {
      "epoch": 0.4643489949288202,
      "grad_norm": 1.833509922027588,
      "learning_rate": 4.775226410476074e-05,
      "loss": 1.4013,
      "step": 15200
    },
    {
      "epoch": 0.4674039225270361,
      "grad_norm": 2.599609613418579,
      "learning_rate": 4.773696609962061e-05,
      "loss": 1.3305,
      "step": 15300
    },
    {
      "epoch": 0.470458850125252,
      "grad_norm": 1.2570773363113403,
      "learning_rate": 4.772166809448048e-05,
      "loss": 1.2685,
      "step": 15400
    },
    {
      "epoch": 0.473513777723468,
      "grad_norm": 2.675783634185791,
      "learning_rate": 4.770637008934035e-05,
      "loss": 1.2194,
      "step": 15500
    },
    {
      "epoch": 0.4765687053216839,
      "grad_norm": 5.149052143096924,
      "learning_rate": 4.769107208420022e-05,
      "loss": 1.3477,
      "step": 15600
    },
    {
      "epoch": 0.4796236329198998,
      "grad_norm": 2.4533028602600098,
      "learning_rate": 4.7675774079060094e-05,
      "loss": 1.3299,
      "step": 15700
    },
    {
      "epoch": 0.4826785605181157,
      "grad_norm": 1.2919026613235474,
      "learning_rate": 4.7660476073919965e-05,
      "loss": 1.3504,
      "step": 15800
    },
    {
      "epoch": 0.48573348811633166,
      "grad_norm": 4.265079498291016,
      "learning_rate": 4.7645178068779837e-05,
      "loss": 1.4127,
      "step": 15900
    },
    {
      "epoch": 0.48878841571454756,
      "grad_norm": 2.5100667476654053,
      "learning_rate": 4.76298800636397e-05,
      "loss": 1.4027,
      "step": 16000
    },
    {
      "epoch": 0.49184334331276347,
      "grad_norm": 3.6750094890594482,
      "learning_rate": 4.761458205849957e-05,
      "loss": 1.2504,
      "step": 16100
    },
    {
      "epoch": 0.49489827091097943,
      "grad_norm": 1.9922250509262085,
      "learning_rate": 4.759943703341084e-05,
      "loss": 1.303,
      "step": 16200
    },
    {
      "epoch": 0.49795319850919534,
      "grad_norm": 2.3737528324127197,
      "learning_rate": 4.758413902827071e-05,
      "loss": 1.3723,
      "step": 16300
    },
    {
      "epoch": 0.5010081261074113,
      "grad_norm": 1.4733338356018066,
      "learning_rate": 4.756884102313058e-05,
      "loss": 1.2958,
      "step": 16400
    },
    {
      "epoch": 0.5040630537056272,
      "grad_norm": 5.010000705718994,
      "learning_rate": 4.7553543017990454e-05,
      "loss": 1.1981,
      "step": 16500
    },
    {
      "epoch": 0.5071179813038431,
      "grad_norm": 1.789291501045227,
      "learning_rate": 4.7538245012850325e-05,
      "loss": 1.385,
      "step": 16600
    },
    {
      "epoch": 0.510172908902059,
      "grad_norm": 4.9523234367370605,
      "learning_rate": 4.7522947007710196e-05,
      "loss": 1.2778,
      "step": 16700
    },
    {
      "epoch": 0.5132278365002749,
      "grad_norm": 1.309713363647461,
      "learning_rate": 4.750764900257007e-05,
      "loss": 1.3772,
      "step": 16800
    },
    {
      "epoch": 0.5162827640984908,
      "grad_norm": 2.1764285564422607,
      "learning_rate": 4.749235099742994e-05,
      "loss": 1.2442,
      "step": 16900
    },
    {
      "epoch": 0.5193376916967067,
      "grad_norm": 5.70428991317749,
      "learning_rate": 4.74770529922898e-05,
      "loss": 1.3491,
      "step": 17000
    },
    {
      "epoch": 0.5223926192949228,
      "grad_norm": 5.1388468742370605,
      "learning_rate": 4.7461754987149674e-05,
      "loss": 1.3646,
      "step": 17100
    },
    {
      "epoch": 0.5254475468931387,
      "grad_norm": 3.198745012283325,
      "learning_rate": 4.7446456982009545e-05,
      "loss": 1.124,
      "step": 17200
    },
    {
      "epoch": 0.5285024744913546,
      "grad_norm": 4.073473930358887,
      "learning_rate": 4.7431158976869416e-05,
      "loss": 1.3609,
      "step": 17300
    },
    {
      "epoch": 0.5315574020895705,
      "grad_norm": 3.4154462814331055,
      "learning_rate": 4.741586097172929e-05,
      "loss": 1.2877,
      "step": 17400
    },
    {
      "epoch": 0.5346123296877864,
      "grad_norm": 2.073560953140259,
      "learning_rate": 4.740056296658916e-05,
      "loss": 1.3057,
      "step": 17500
    },
    {
      "epoch": 0.5376672572860023,
      "grad_norm": 2.3424408435821533,
      "learning_rate": 4.738526496144903e-05,
      "loss": 1.3119,
      "step": 17600
    },
    {
      "epoch": 0.5407221848842182,
      "grad_norm": 1.6415555477142334,
      "learning_rate": 4.73699669563089e-05,
      "loss": 1.307,
      "step": 17700
    },
    {
      "epoch": 0.5437771124824342,
      "grad_norm": 3.323765277862549,
      "learning_rate": 4.7354668951168765e-05,
      "loss": 1.3771,
      "step": 17800
    },
    {
      "epoch": 0.5468320400806501,
      "grad_norm": 6.020792007446289,
      "learning_rate": 4.7339370946028636e-05,
      "loss": 1.303,
      "step": 17900
    },
    {
      "epoch": 0.549886967678866,
      "grad_norm": 3.13588547706604,
      "learning_rate": 4.732407294088851e-05,
      "loss": 1.3615,
      "step": 18000
    },
    {
      "epoch": 0.5529418952770819,
      "grad_norm": 0.8763374090194702,
      "learning_rate": 4.730877493574838e-05,
      "loss": 1.2856,
      "step": 18100
    },
    {
      "epoch": 0.5559968228752978,
      "grad_norm": 4.6722025871276855,
      "learning_rate": 4.729362991065965e-05,
      "loss": 1.2759,
      "step": 18200
    },
    {
      "epoch": 0.5590517504735137,
      "grad_norm": 4.301787853240967,
      "learning_rate": 4.7278331905519524e-05,
      "loss": 1.2397,
      "step": 18300
    },
    {
      "epoch": 0.5621066780717296,
      "grad_norm": 3.807133436203003,
      "learning_rate": 4.7263033900379396e-05,
      "loss": 1.3704,
      "step": 18400
    },
    {
      "epoch": 0.5651616056699457,
      "grad_norm": 3.641347885131836,
      "learning_rate": 4.724773589523927e-05,
      "loss": 1.2844,
      "step": 18500
    },
    {
      "epoch": 0.5682165332681616,
      "grad_norm": 2.7137675285339355,
      "learning_rate": 4.723243789009914e-05,
      "loss": 1.2856,
      "step": 18600
    },
    {
      "epoch": 0.5712714608663775,
      "grad_norm": 4.199899673461914,
      "learning_rate": 4.7217139884959e-05,
      "loss": 1.2635,
      "step": 18700
    },
    {
      "epoch": 0.5743263884645934,
      "grad_norm": 3.4955711364746094,
      "learning_rate": 4.7201841879818873e-05,
      "loss": 1.4038,
      "step": 18800
    },
    {
      "epoch": 0.5773813160628093,
      "grad_norm": 4.2103471755981445,
      "learning_rate": 4.7186543874678745e-05,
      "loss": 1.2279,
      "step": 18900
    },
    {
      "epoch": 0.5804362436610252,
      "grad_norm": 1.8225414752960205,
      "learning_rate": 4.7171245869538616e-05,
      "loss": 1.3211,
      "step": 19000
    },
    {
      "epoch": 0.5834911712592411,
      "grad_norm": 4.89621639251709,
      "learning_rate": 4.715594786439849e-05,
      "loss": 1.2343,
      "step": 19100
    },
    {
      "epoch": 0.5865460988574571,
      "grad_norm": 3.157581090927124,
      "learning_rate": 4.714064985925836e-05,
      "loss": 1.3112,
      "step": 19200
    },
    {
      "epoch": 0.589601026455673,
      "grad_norm": 4.4817423820495605,
      "learning_rate": 4.712535185411823e-05,
      "loss": 1.3675,
      "step": 19300
    },
    {
      "epoch": 0.5926559540538889,
      "grad_norm": 3.6259207725524902,
      "learning_rate": 4.71100538489781e-05,
      "loss": 1.2662,
      "step": 19400
    },
    {
      "epoch": 0.5957108816521048,
      "grad_norm": 1.9341295957565308,
      "learning_rate": 4.7094755843837965e-05,
      "loss": 1.264,
      "step": 19500
    },
    {
      "epoch": 0.5987658092503207,
      "grad_norm": 2.3397438526153564,
      "learning_rate": 4.7079457838697836e-05,
      "loss": 1.2692,
      "step": 19600
    },
    {
      "epoch": 0.6018207368485367,
      "grad_norm": 2.147289514541626,
      "learning_rate": 4.706415983355771e-05,
      "loss": 1.1633,
      "step": 19700
    },
    {
      "epoch": 0.6048756644467526,
      "grad_norm": 4.903138160705566,
      "learning_rate": 4.704886182841758e-05,
      "loss": 1.2628,
      "step": 19800
    },
    {
      "epoch": 0.6079305920449686,
      "grad_norm": 4.019737720489502,
      "learning_rate": 4.703356382327745e-05,
      "loss": 1.2283,
      "step": 19900
    },
    {
      "epoch": 0.6109855196431845,
      "grad_norm": 5.8793182373046875,
      "learning_rate": 4.701826581813732e-05,
      "loss": 1.2297,
      "step": 20000
    },
    {
      "epoch": 0.6140404472414004,
      "grad_norm": 2.11370587348938,
      "learning_rate": 4.700296781299719e-05,
      "loss": 1.3137,
      "step": 20100
    },
    {
      "epoch": 0.6170953748396163,
      "grad_norm": 2.1315717697143555,
      "learning_rate": 4.698782278790846e-05,
      "loss": 1.3261,
      "step": 20200
    },
    {
      "epoch": 0.6201503024378322,
      "grad_norm": 3.4208743572235107,
      "learning_rate": 4.697252478276833e-05,
      "loss": 1.3392,
      "step": 20300
    },
    {
      "epoch": 0.6232052300360481,
      "grad_norm": 2.3667633533477783,
      "learning_rate": 4.69572267776282e-05,
      "loss": 1.2584,
      "step": 20400
    },
    {
      "epoch": 0.626260157634264,
      "grad_norm": 2.951887607574463,
      "learning_rate": 4.6941928772488066e-05,
      "loss": 1.2928,
      "step": 20500
    },
    {
      "epoch": 0.62931508523248,
      "grad_norm": 3.0843470096588135,
      "learning_rate": 4.692663076734794e-05,
      "loss": 1.351,
      "step": 20600
    },
    {
      "epoch": 0.6323700128306959,
      "grad_norm": 3.92063045501709,
      "learning_rate": 4.691133276220781e-05,
      "loss": 1.3723,
      "step": 20700
    },
    {
      "epoch": 0.6354249404289118,
      "grad_norm": 6.2461676597595215,
      "learning_rate": 4.689603475706768e-05,
      "loss": 1.3467,
      "step": 20800
    },
    {
      "epoch": 0.6384798680271278,
      "grad_norm": 5.256413459777832,
      "learning_rate": 4.688073675192755e-05,
      "loss": 1.2848,
      "step": 20900
    },
    {
      "epoch": 0.6415347956253437,
      "grad_norm": 2.6756303310394287,
      "learning_rate": 4.686543874678742e-05,
      "loss": 1.2802,
      "step": 21000
    },
    {
      "epoch": 0.6445897232235596,
      "grad_norm": 2.9479217529296875,
      "learning_rate": 4.685014074164729e-05,
      "loss": 1.3156,
      "step": 21100
    },
    {
      "epoch": 0.6476446508217755,
      "grad_norm": 1.6067280769348145,
      "learning_rate": 4.6834842736507164e-05,
      "loss": 1.2648,
      "step": 21200
    },
    {
      "epoch": 0.6506995784199915,
      "grad_norm": 2.9112439155578613,
      "learning_rate": 4.681954473136703e-05,
      "loss": 1.2988,
      "step": 21300
    },
    {
      "epoch": 0.6537545060182074,
      "grad_norm": 3.3154282569885254,
      "learning_rate": 4.68042467262269e-05,
      "loss": 1.2409,
      "step": 21400
    },
    {
      "epoch": 0.6568094336164233,
      "grad_norm": 1.7731764316558838,
      "learning_rate": 4.678894872108677e-05,
      "loss": 1.2897,
      "step": 21500
    },
    {
      "epoch": 0.6598643612146392,
      "grad_norm": 2.621236801147461,
      "learning_rate": 4.677365071594664e-05,
      "loss": 1.349,
      "step": 21600
    },
    {
      "epoch": 0.6629192888128551,
      "grad_norm": 1.563030481338501,
      "learning_rate": 4.6758352710806513e-05,
      "loss": 1.3203,
      "step": 21700
    },
    {
      "epoch": 0.665974216411071,
      "grad_norm": 3.4372918605804443,
      "learning_rate": 4.6743054705666385e-05,
      "loss": 1.3152,
      "step": 21800
    },
    {
      "epoch": 0.6690291440092869,
      "grad_norm": 5.096141815185547,
      "learning_rate": 4.6727756700526256e-05,
      "loss": 1.2419,
      "step": 21900
    },
    {
      "epoch": 0.6720840716075029,
      "grad_norm": 2.1512577533721924,
      "learning_rate": 4.671245869538613e-05,
      "loss": 1.2946,
      "step": 22000
    },
    {
      "epoch": 0.6751389992057188,
      "grad_norm": 11.683186531066895,
      "learning_rate": 4.669716069024599e-05,
      "loss": 1.2553,
      "step": 22100
    },
    {
      "epoch": 0.6781939268039348,
      "grad_norm": 9.039573669433594,
      "learning_rate": 4.6682015665157266e-05,
      "loss": 1.2411,
      "step": 22200
    },
    {
      "epoch": 0.6812488544021507,
      "grad_norm": 3.796337842941284,
      "learning_rate": 4.666671766001714e-05,
      "loss": 1.2908,
      "step": 22300
    },
    {
      "epoch": 0.6843037820003666,
      "grad_norm": 3.969041347503662,
      "learning_rate": 4.6651419654877e-05,
      "loss": 1.2525,
      "step": 22400
    },
    {
      "epoch": 0.6873587095985825,
      "grad_norm": 5.0099077224731445,
      "learning_rate": 4.663612164973687e-05,
      "loss": 1.265,
      "step": 22500
    },
    {
      "epoch": 0.6904136371967985,
      "grad_norm": 7.791167736053467,
      "learning_rate": 4.6620823644596744e-05,
      "loss": 1.3321,
      "step": 22600
    },
    {
      "epoch": 0.6934685647950144,
      "grad_norm": 2.604672431945801,
      "learning_rate": 4.6605525639456615e-05,
      "loss": 1.2495,
      "step": 22700
    },
    {
      "epoch": 0.6965234923932303,
      "grad_norm": 3.6437833309173584,
      "learning_rate": 4.6590227634316486e-05,
      "loss": 1.3683,
      "step": 22800
    },
    {
      "epoch": 0.6995784199914462,
      "grad_norm": 1.2340764999389648,
      "learning_rate": 4.657492962917636e-05,
      "loss": 1.2403,
      "step": 22900
    },
    {
      "epoch": 0.7026333475896621,
      "grad_norm": 7.440982341766357,
      "learning_rate": 4.655963162403623e-05,
      "loss": 1.2417,
      "step": 23000
    },
    {
      "epoch": 0.705688275187878,
      "grad_norm": 1.300933599472046,
      "learning_rate": 4.65443336188961e-05,
      "loss": 1.3279,
      "step": 23100
    },
    {
      "epoch": 0.7087432027860939,
      "grad_norm": 2.8632686138153076,
      "learning_rate": 4.6529035613755964e-05,
      "loss": 1.3214,
      "step": 23200
    },
    {
      "epoch": 0.71179813038431,
      "grad_norm": 10.080193519592285,
      "learning_rate": 4.6513737608615835e-05,
      "loss": 1.318,
      "step": 23300
    },
    {
      "epoch": 0.7148530579825259,
      "grad_norm": 2.173156976699829,
      "learning_rate": 4.6498439603475706e-05,
      "loss": 1.2769,
      "step": 23400
    },
    {
      "epoch": 0.7179079855807418,
      "grad_norm": 4.581544399261475,
      "learning_rate": 4.648329457838698e-05,
      "loss": 1.2981,
      "step": 23500
    },
    {
      "epoch": 0.7209629131789577,
      "grad_norm": 1.1701802015304565,
      "learning_rate": 4.646799657324685e-05,
      "loss": 1.2509,
      "step": 23600
    },
    {
      "epoch": 0.7240178407771736,
      "grad_norm": 4.420626640319824,
      "learning_rate": 4.6452698568106723e-05,
      "loss": 1.2835,
      "step": 23700
    },
    {
      "epoch": 0.7270727683753895,
      "grad_norm": 7.273269176483154,
      "learning_rate": 4.6437400562966595e-05,
      "loss": 1.3267,
      "step": 23800
    },
    {
      "epoch": 0.7301276959736054,
      "grad_norm": 4.186655521392822,
      "learning_rate": 4.6422102557826466e-05,
      "loss": 1.2442,
      "step": 23900
    },
    {
      "epoch": 0.7331826235718214,
      "grad_norm": 3.82904052734375,
      "learning_rate": 4.640680455268634e-05,
      "loss": 1.195,
      "step": 24000
    },
    {
      "epoch": 0.7362375511700373,
      "grad_norm": 2.4320216178894043,
      "learning_rate": 4.63915065475462e-05,
      "loss": 1.29,
      "step": 24100
    },
    {
      "epoch": 0.7392924787682532,
      "grad_norm": 2.309053897857666,
      "learning_rate": 4.637620854240607e-05,
      "loss": 1.3795,
      "step": 24200
    },
    {
      "epoch": 0.7423474063664691,
      "grad_norm": 3.3796703815460205,
      "learning_rate": 4.6360910537265944e-05,
      "loss": 1.2625,
      "step": 24300
    },
    {
      "epoch": 0.745402333964685,
      "grad_norm": 2.1054587364196777,
      "learning_rate": 4.6345612532125815e-05,
      "loss": 1.2281,
      "step": 24400
    },
    {
      "epoch": 0.7484572615629009,
      "grad_norm": 5.734375476837158,
      "learning_rate": 4.6330314526985686e-05,
      "loss": 1.2712,
      "step": 24500
    },
    {
      "epoch": 0.7515121891611168,
      "grad_norm": 6.285786151885986,
      "learning_rate": 4.631501652184556e-05,
      "loss": 1.1773,
      "step": 24600
    },
    {
      "epoch": 0.7545671167593329,
      "grad_norm": 4.265208721160889,
      "learning_rate": 4.629971851670543e-05,
      "loss": 1.1952,
      "step": 24700
    },
    {
      "epoch": 0.7576220443575488,
      "grad_norm": 4.139396667480469,
      "learning_rate": 4.62844205115653e-05,
      "loss": 1.2747,
      "step": 24800
    },
    {
      "epoch": 0.7606769719557647,
      "grad_norm": 3.3605008125305176,
      "learning_rate": 4.6269122506425164e-05,
      "loss": 1.2392,
      "step": 24900
    },
    {
      "epoch": 0.7637318995539806,
      "grad_norm": 3.1468260288238525,
      "learning_rate": 4.6253824501285035e-05,
      "loss": 1.2669,
      "step": 25000
    },
    {
      "epoch": 0.7667868271521965,
      "grad_norm": 1.4195454120635986,
      "learning_rate": 4.6238526496144906e-05,
      "loss": 1.3578,
      "step": 25100
    },
    {
      "epoch": 0.7698417547504124,
      "grad_norm": 3.4323678016662598,
      "learning_rate": 4.622322849100478e-05,
      "loss": 1.3418,
      "step": 25200
    },
    {
      "epoch": 0.7728966823486283,
      "grad_norm": 8.107824325561523,
      "learning_rate": 4.620793048586465e-05,
      "loss": 1.1864,
      "step": 25300
    },
    {
      "epoch": 0.7759516099468443,
      "grad_norm": 1.496487021446228,
      "learning_rate": 4.619263248072452e-05,
      "loss": 1.2452,
      "step": 25400
    },
    {
      "epoch": 0.7790065375450602,
      "grad_norm": 4.837622165679932,
      "learning_rate": 4.617733447558439e-05,
      "loss": 1.2801,
      "step": 25500
    },
    {
      "epoch": 0.7820614651432761,
      "grad_norm": 5.973275184631348,
      "learning_rate": 4.616203647044426e-05,
      "loss": 1.2408,
      "step": 25600
    },
    {
      "epoch": 0.785116392741492,
      "grad_norm": 4.390771865844727,
      "learning_rate": 4.6146738465304126e-05,
      "loss": 1.1513,
      "step": 25700
    },
    {
      "epoch": 0.7881713203397079,
      "grad_norm": 1.7596487998962402,
      "learning_rate": 4.6131440460164e-05,
      "loss": 1.2074,
      "step": 25800
    },
    {
      "epoch": 0.7912262479379238,
      "grad_norm": 1.3779590129852295,
      "learning_rate": 4.611614245502387e-05,
      "loss": 1.1349,
      "step": 25900
    },
    {
      "epoch": 0.7942811755361397,
      "grad_norm": 9.768011093139648,
      "learning_rate": 4.610084444988374e-05,
      "loss": 1.2668,
      "step": 26000
    },
    {
      "epoch": 0.7973361031343558,
      "grad_norm": 2.795086622238159,
      "learning_rate": 4.608554644474361e-05,
      "loss": 1.4519,
      "step": 26100
    },
    {
      "epoch": 0.8003910307325717,
      "grad_norm": 2.03102970123291,
      "learning_rate": 4.607024843960348e-05,
      "loss": 1.3429,
      "step": 26200
    },
    {
      "epoch": 0.8034459583307876,
      "grad_norm": 3.355180501937866,
      "learning_rate": 4.605495043446335e-05,
      "loss": 1.346,
      "step": 26300
    },
    {
      "epoch": 0.8065008859290035,
      "grad_norm": 4.949088096618652,
      "learning_rate": 4.6039652429323224e-05,
      "loss": 1.297,
      "step": 26400
    },
    {
      "epoch": 0.8095558135272194,
      "grad_norm": 3.3454296588897705,
      "learning_rate": 4.602435442418309e-05,
      "loss": 1.3045,
      "step": 26500
    },
    {
      "epoch": 0.8126107411254353,
      "grad_norm": 3.7700753211975098,
      "learning_rate": 4.600905641904295e-05,
      "loss": 1.2386,
      "step": 26600
    },
    {
      "epoch": 0.8156656687236512,
      "grad_norm": 3.9577269554138184,
      "learning_rate": 4.5993758413902824e-05,
      "loss": 1.336,
      "step": 26700
    },
    {
      "epoch": 0.8187205963218672,
      "grad_norm": 5.364854335784912,
      "learning_rate": 4.5978460408762695e-05,
      "loss": 1.3863,
      "step": 26800
    },
    {
      "epoch": 0.8217755239200831,
      "grad_norm": 3.6088366508483887,
      "learning_rate": 4.5963162403622567e-05,
      "loss": 1.2049,
      "step": 26900
    },
    {
      "epoch": 0.824830451518299,
      "grad_norm": 2.9241647720336914,
      "learning_rate": 4.594786439848244e-05,
      "loss": 1.3627,
      "step": 27000
    },
    {
      "epoch": 0.8278853791165149,
      "grad_norm": 4.087007999420166,
      "learning_rate": 4.593256639334231e-05,
      "loss": 1.2335,
      "step": 27100
    },
    {
      "epoch": 0.8309403067147308,
      "grad_norm": 2.52764630317688,
      "learning_rate": 4.591726838820218e-05,
      "loss": 1.288,
      "step": 27200
    },
    {
      "epoch": 0.8339952343129468,
      "grad_norm": 1.945397138595581,
      "learning_rate": 4.590197038306205e-05,
      "loss": 1.2157,
      "step": 27300
    },
    {
      "epoch": 0.8370501619111627,
      "grad_norm": 3.3560779094696045,
      "learning_rate": 4.5886672377921916e-05,
      "loss": 1.2333,
      "step": 27400
    },
    {
      "epoch": 0.8401050895093787,
      "grad_norm": 2.7868645191192627,
      "learning_rate": 4.587152735283319e-05,
      "loss": 1.1488,
      "step": 27500
    },
    {
      "epoch": 0.8431600171075946,
      "grad_norm": 2.6062686443328857,
      "learning_rate": 4.585622934769306e-05,
      "loss": 1.3793,
      "step": 27600
    },
    {
      "epoch": 0.8462149447058105,
      "grad_norm": 5.000110626220703,
      "learning_rate": 4.5841084322604336e-05,
      "loss": 1.3183,
      "step": 27700
    },
    {
      "epoch": 0.8492698723040264,
      "grad_norm": 3.7855002880096436,
      "learning_rate": 4.58257863174642e-05,
      "loss": 1.3456,
      "step": 27800
    },
    {
      "epoch": 0.8523247999022423,
      "grad_norm": 1.3873869180679321,
      "learning_rate": 4.581048831232407e-05,
      "loss": 1.3148,
      "step": 27900
    },
    {
      "epoch": 0.8553797275004582,
      "grad_norm": 2.0780625343322754,
      "learning_rate": 4.579519030718394e-05,
      "loss": 1.3301,
      "step": 28000
    },
    {
      "epoch": 0.8584346550986741,
      "grad_norm": 4.090784072875977,
      "learning_rate": 4.5779892302043814e-05,
      "loss": 1.2458,
      "step": 28100
    },
    {
      "epoch": 0.8614895826968901,
      "grad_norm": 3.2605838775634766,
      "learning_rate": 4.5764594296903685e-05,
      "loss": 1.2539,
      "step": 28200
    },
    {
      "epoch": 0.864544510295106,
      "grad_norm": 2.4564528465270996,
      "learning_rate": 4.5749296291763556e-05,
      "loss": 1.224,
      "step": 28300
    },
    {
      "epoch": 0.867599437893322,
      "grad_norm": 3.5012292861938477,
      "learning_rate": 4.573399828662343e-05,
      "loss": 1.1227,
      "step": 28400
    },
    {
      "epoch": 0.8706543654915379,
      "grad_norm": 2.125563621520996,
      "learning_rate": 4.57187002814833e-05,
      "loss": 1.1865,
      "step": 28500
    },
    {
      "epoch": 0.8737092930897538,
      "grad_norm": 7.136216640472412,
      "learning_rate": 4.570340227634316e-05,
      "loss": 1.157,
      "step": 28600
    },
    {
      "epoch": 0.8767642206879697,
      "grad_norm": 7.11937141418457,
      "learning_rate": 4.5688104271203034e-05,
      "loss": 1.2754,
      "step": 28700
    },
    {
      "epoch": 0.8798191482861856,
      "grad_norm": 0.7430057525634766,
      "learning_rate": 4.5672806266062905e-05,
      "loss": 1.2952,
      "step": 28800
    },
    {
      "epoch": 0.8828740758844016,
      "grad_norm": 4.634474277496338,
      "learning_rate": 4.5657508260922777e-05,
      "loss": 1.323,
      "step": 28900
    },
    {
      "epoch": 0.8859290034826175,
      "grad_norm": 3.6207408905029297,
      "learning_rate": 4.564221025578265e-05,
      "loss": 1.2202,
      "step": 29000
    },
    {
      "epoch": 0.8889839310808334,
      "grad_norm": 6.108366966247559,
      "learning_rate": 4.562691225064252e-05,
      "loss": 1.2362,
      "step": 29100
    },
    {
      "epoch": 0.8920388586790493,
      "grad_norm": 3.439833879470825,
      "learning_rate": 4.561161424550239e-05,
      "loss": 1.2378,
      "step": 29200
    },
    {
      "epoch": 0.8950937862772652,
      "grad_norm": 4.669163227081299,
      "learning_rate": 4.559631624036226e-05,
      "loss": 1.1483,
      "step": 29300
    },
    {
      "epoch": 0.8981487138754811,
      "grad_norm": 3.7762789726257324,
      "learning_rate": 4.5581018235222126e-05,
      "loss": 1.2534,
      "step": 29400
    },
    {
      "epoch": 0.901203641473697,
      "grad_norm": 5.76541805267334,
      "learning_rate": 4.5565720230082e-05,
      "loss": 1.2709,
      "step": 29500
    },
    {
      "epoch": 0.904258569071913,
      "grad_norm": 1.8639878034591675,
      "learning_rate": 4.555042222494187e-05,
      "loss": 1.1896,
      "step": 29600
    },
    {
      "epoch": 0.907313496670129,
      "grad_norm": 10.215450286865234,
      "learning_rate": 4.553512421980174e-05,
      "loss": 1.3192,
      "step": 29700
    },
    {
      "epoch": 0.9103684242683449,
      "grad_norm": 4.441849231719971,
      "learning_rate": 4.5519979194713014e-05,
      "loss": 1.2901,
      "step": 29800
    },
    {
      "epoch": 0.9134233518665608,
      "grad_norm": 3.3630282878875732,
      "learning_rate": 4.5504681189572885e-05,
      "loss": 1.2784,
      "step": 29900
    },
    {
      "epoch": 0.9164782794647767,
      "grad_norm": 5.191607475280762,
      "learning_rate": 4.5489383184432756e-05,
      "loss": 1.1283,
      "step": 30000
    },
    {
      "epoch": 0.9195332070629926,
      "grad_norm": 3.6215574741363525,
      "learning_rate": 4.547408517929263e-05,
      "loss": 1.1619,
      "step": 30100
    },
    {
      "epoch": 0.9225881346612085,
      "grad_norm": 5.8025946617126465,
      "learning_rate": 4.54587871741525e-05,
      "loss": 1.2857,
      "step": 30200
    },
    {
      "epoch": 0.9256430622594245,
      "grad_norm": 5.739350318908691,
      "learning_rate": 4.544348916901236e-05,
      "loss": 1.3402,
      "step": 30300
    },
    {
      "epoch": 0.9286979898576404,
      "grad_norm": 1.923235297203064,
      "learning_rate": 4.5428191163872234e-05,
      "loss": 1.3791,
      "step": 30400
    },
    {
      "epoch": 0.9317529174558563,
      "grad_norm": 2.0984885692596436,
      "learning_rate": 4.5412893158732105e-05,
      "loss": 1.3596,
      "step": 30500
    },
    {
      "epoch": 0.9348078450540722,
      "grad_norm": 11.962567329406738,
      "learning_rate": 4.5397595153591976e-05,
      "loss": 1.2984,
      "step": 30600
    },
    {
      "epoch": 0.9378627726522881,
      "grad_norm": 3.2890284061431885,
      "learning_rate": 4.538229714845185e-05,
      "loss": 1.2089,
      "step": 30700
    },
    {
      "epoch": 0.940917700250504,
      "grad_norm": 3.9245996475219727,
      "learning_rate": 4.536699914331172e-05,
      "loss": 1.3328,
      "step": 30800
    },
    {
      "epoch": 0.9439726278487199,
      "grad_norm": 2.281853675842285,
      "learning_rate": 4.535170113817159e-05,
      "loss": 1.2884,
      "step": 30900
    },
    {
      "epoch": 0.947027555446936,
      "grad_norm": 3.139033079147339,
      "learning_rate": 4.533640313303146e-05,
      "loss": 1.1133,
      "step": 31000
    },
    {
      "epoch": 0.9500824830451519,
      "grad_norm": 1.584032654762268,
      "learning_rate": 4.5321105127891325e-05,
      "loss": 1.1666,
      "step": 31100
    },
    {
      "epoch": 0.9531374106433678,
      "grad_norm": 3.9088025093078613,
      "learning_rate": 4.530580712275119e-05,
      "loss": 1.273,
      "step": 31200
    },
    {
      "epoch": 0.9561923382415837,
      "grad_norm": 4.956596374511719,
      "learning_rate": 4.529050911761106e-05,
      "loss": 1.3739,
      "step": 31300
    },
    {
      "epoch": 0.9592472658397996,
      "grad_norm": 8.017325401306152,
      "learning_rate": 4.527521111247093e-05,
      "loss": 1.2341,
      "step": 31400
    },
    {
      "epoch": 0.9623021934380155,
      "grad_norm": 3.9635438919067383,
      "learning_rate": 4.52599131073308e-05,
      "loss": 1.1666,
      "step": 31500
    },
    {
      "epoch": 0.9653571210362314,
      "grad_norm": 2.0401432514190674,
      "learning_rate": 4.5244615102190674e-05,
      "loss": 1.2702,
      "step": 31600
    },
    {
      "epoch": 0.9684120486344474,
      "grad_norm": 2.1671855449676514,
      "learning_rate": 4.5229317097050545e-05,
      "loss": 1.144,
      "step": 31700
    },
    {
      "epoch": 0.9714669762326633,
      "grad_norm": 5.701295375823975,
      "learning_rate": 4.5214019091910417e-05,
      "loss": 1.1877,
      "step": 31800
    },
    {
      "epoch": 0.9745219038308792,
      "grad_norm": 3.89962100982666,
      "learning_rate": 4.519872108677029e-05,
      "loss": 1.2391,
      "step": 31900
    },
    {
      "epoch": 0.9775768314290951,
      "grad_norm": 13.442463874816895,
      "learning_rate": 4.518342308163015e-05,
      "loss": 1.2846,
      "step": 32000
    },
    {
      "epoch": 0.980631759027311,
      "grad_norm": 6.973484039306641,
      "learning_rate": 4.516827805654143e-05,
      "loss": 1.2133,
      "step": 32100
    },
    {
      "epoch": 0.9836866866255269,
      "grad_norm": 3.8862478733062744,
      "learning_rate": 4.51529800514013e-05,
      "loss": 1.1195,
      "step": 32200
    },
    {
      "epoch": 0.9867416142237428,
      "grad_norm": 5.151338577270508,
      "learning_rate": 4.513768204626117e-05,
      "loss": 1.2547,
      "step": 32300
    },
    {
      "epoch": 0.9897965418219589,
      "grad_norm": 4.248752593994141,
      "learning_rate": 4.512238404112104e-05,
      "loss": 1.2451,
      "step": 32400
    },
    {
      "epoch": 0.9928514694201748,
      "grad_norm": 4.167698383331299,
      "learning_rate": 4.510708603598091e-05,
      "loss": 1.1808,
      "step": 32500
    },
    {
      "epoch": 0.9959063970183907,
      "grad_norm": 1.6347144842147827,
      "learning_rate": 4.509178803084078e-05,
      "loss": 1.3021,
      "step": 32600
    },
    {
      "epoch": 0.9989613246166066,
      "grad_norm": 3.7852065563201904,
      "learning_rate": 4.5076490025700654e-05,
      "loss": 1.1772,
      "step": 32700
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6393352477546282,
      "eval_loss": 0.969184935092926,
      "eval_runtime": 1839.346,
      "eval_samples_per_second": 17.797,
      "eval_steps_per_second": 4.449,
      "step": 32734
    },
    {
      "epoch": 1.0020162522148226,
      "grad_norm": 2.740159749984741,
      "learning_rate": 4.5061192020560525e-05,
      "loss": 1.3645,
      "step": 32800
    },
    {
      "epoch": 1.0050711798130385,
      "grad_norm": 2.5094220638275146,
      "learning_rate": 4.504589401542039e-05,
      "loss": 1.1684,
      "step": 32900
    },
    {
      "epoch": 1.0081261074112544,
      "grad_norm": 7.7797417640686035,
      "learning_rate": 4.503059601028026e-05,
      "loss": 1.2266,
      "step": 33000
    },
    {
      "epoch": 1.0111810350094703,
      "grad_norm": 2.7064406871795654,
      "learning_rate": 4.501529800514013e-05,
      "loss": 1.1893,
      "step": 33100
    },
    {
      "epoch": 1.0142359626076862,
      "grad_norm": 8.84860610961914,
      "learning_rate": 4.5e-05,
      "loss": 1.2385,
      "step": 33200
    },
    {
      "epoch": 1.0172908902059021,
      "grad_norm": 7.67369270324707,
      "learning_rate": 4.4984701994859874e-05,
      "loss": 1.2131,
      "step": 33300
    },
    {
      "epoch": 1.020345817804118,
      "grad_norm": 4.2192769050598145,
      "learning_rate": 4.4969403989719745e-05,
      "loss": 1.2668,
      "step": 33400
    },
    {
      "epoch": 1.023400745402334,
      "grad_norm": 4.694185733795166,
      "learning_rate": 4.4954105984579616e-05,
      "loss": 1.277,
      "step": 33500
    },
    {
      "epoch": 1.0264556730005499,
      "grad_norm": 4.47376823425293,
      "learning_rate": 4.493880797943949e-05,
      "loss": 1.1455,
      "step": 33600
    },
    {
      "epoch": 1.0295106005987658,
      "grad_norm": 1.780508279800415,
      "learning_rate": 4.492350997429935e-05,
      "loss": 1.194,
      "step": 33700
    },
    {
      "epoch": 1.0325655281969817,
      "grad_norm": 2.220668315887451,
      "learning_rate": 4.490821196915922e-05,
      "loss": 1.3191,
      "step": 33800
    },
    {
      "epoch": 1.0356204557951976,
      "grad_norm": 2.0353493690490723,
      "learning_rate": 4.4892913964019094e-05,
      "loss": 1.248,
      "step": 33900
    },
    {
      "epoch": 1.0386753833934135,
      "grad_norm": 3.851609230041504,
      "learning_rate": 4.4877615958878965e-05,
      "loss": 1.2369,
      "step": 34000
    },
    {
      "epoch": 1.0417303109916296,
      "grad_norm": 2.8817436695098877,
      "learning_rate": 4.4862317953738836e-05,
      "loss": 1.2093,
      "step": 34100
    },
    {
      "epoch": 1.0447852385898455,
      "grad_norm": 3.6299502849578857,
      "learning_rate": 4.484701994859871e-05,
      "loss": 1.3564,
      "step": 34200
    },
    {
      "epoch": 1.0478401661880614,
      "grad_norm": 3.872225284576416,
      "learning_rate": 4.483172194345858e-05,
      "loss": 1.2894,
      "step": 34300
    },
    {
      "epoch": 1.0508950937862773,
      "grad_norm": 1.910213828086853,
      "learning_rate": 4.481642393831845e-05,
      "loss": 1.2529,
      "step": 34400
    },
    {
      "epoch": 1.0539500213844932,
      "grad_norm": 7.973636150360107,
      "learning_rate": 4.4801125933178314e-05,
      "loss": 1.1647,
      "step": 34500
    },
    {
      "epoch": 1.0570049489827091,
      "grad_norm": 9.639257431030273,
      "learning_rate": 4.4785827928038185e-05,
      "loss": 1.2896,
      "step": 34600
    },
    {
      "epoch": 1.060059876580925,
      "grad_norm": 4.413392543792725,
      "learning_rate": 4.477052992289806e-05,
      "loss": 1.205,
      "step": 34700
    },
    {
      "epoch": 1.063114804179141,
      "grad_norm": 2.079163074493408,
      "learning_rate": 4.4755384897809325e-05,
      "loss": 1.2079,
      "step": 34800
    },
    {
      "epoch": 1.0661697317773569,
      "grad_norm": 3.3086769580841064,
      "learning_rate": 4.4740086892669196e-05,
      "loss": 1.1257,
      "step": 34900
    },
    {
      "epoch": 1.0692246593755728,
      "grad_norm": 2.8088862895965576,
      "learning_rate": 4.472478888752907e-05,
      "loss": 1.2156,
      "step": 35000
    },
    {
      "epoch": 1.0722795869737887,
      "grad_norm": 2.2325263023376465,
      "learning_rate": 4.470949088238894e-05,
      "loss": 1.2968,
      "step": 35100
    },
    {
      "epoch": 1.0753345145720046,
      "grad_norm": 1.6548900604248047,
      "learning_rate": 4.469419287724881e-05,
      "loss": 1.1761,
      "step": 35200
    },
    {
      "epoch": 1.0783894421702205,
      "grad_norm": 6.921013355255127,
      "learning_rate": 4.467889487210868e-05,
      "loss": 1.1929,
      "step": 35300
    },
    {
      "epoch": 1.0814443697684364,
      "grad_norm": 3.7694363594055176,
      "learning_rate": 4.466359686696855e-05,
      "loss": 1.3127,
      "step": 35400
    },
    {
      "epoch": 1.0844992973666523,
      "grad_norm": 5.201353073120117,
      "learning_rate": 4.4648298861828416e-05,
      "loss": 1.153,
      "step": 35500
    },
    {
      "epoch": 1.0875542249648684,
      "grad_norm": 2.9227755069732666,
      "learning_rate": 4.463300085668829e-05,
      "loss": 1.1579,
      "step": 35600
    },
    {
      "epoch": 1.0906091525630843,
      "grad_norm": 7.101742267608643,
      "learning_rate": 4.461770285154816e-05,
      "loss": 1.2353,
      "step": 35700
    },
    {
      "epoch": 1.0936640801613002,
      "grad_norm": 8.491111755371094,
      "learning_rate": 4.460240484640803e-05,
      "loss": 1.2764,
      "step": 35800
    },
    {
      "epoch": 1.0967190077595161,
      "grad_norm": 6.6550703048706055,
      "learning_rate": 4.45871068412679e-05,
      "loss": 1.1896,
      "step": 35900
    },
    {
      "epoch": 1.099773935357732,
      "grad_norm": 5.08082914352417,
      "learning_rate": 4.457180883612777e-05,
      "loss": 1.3892,
      "step": 36000
    },
    {
      "epoch": 1.102828862955948,
      "grad_norm": 2.673793077468872,
      "learning_rate": 4.455651083098764e-05,
      "loss": 1.2312,
      "step": 36100
    },
    {
      "epoch": 1.1058837905541639,
      "grad_norm": 6.248423099517822,
      "learning_rate": 4.4541212825847514e-05,
      "loss": 1.2901,
      "step": 36200
    },
    {
      "epoch": 1.1089387181523798,
      "grad_norm": 4.006623268127441,
      "learning_rate": 4.452591482070738e-05,
      "loss": 1.3631,
      "step": 36300
    },
    {
      "epoch": 1.1119936457505957,
      "grad_norm": 2.636411666870117,
      "learning_rate": 4.451061681556725e-05,
      "loss": 1.3197,
      "step": 36400
    },
    {
      "epoch": 1.1150485733488116,
      "grad_norm": 2.9009954929351807,
      "learning_rate": 4.449531881042712e-05,
      "loss": 1.3526,
      "step": 36500
    },
    {
      "epoch": 1.1181035009470275,
      "grad_norm": 6.70359992980957,
      "learning_rate": 4.448002080528699e-05,
      "loss": 1.1885,
      "step": 36600
    },
    {
      "epoch": 1.1211584285452434,
      "grad_norm": 12.23004150390625,
      "learning_rate": 4.446472280014686e-05,
      "loss": 1.1935,
      "step": 36700
    },
    {
      "epoch": 1.1242133561434593,
      "grad_norm": 2.4072132110595703,
      "learning_rate": 4.4449424795006734e-05,
      "loss": 1.1927,
      "step": 36800
    },
    {
      "epoch": 1.1272682837416754,
      "grad_norm": 5.668684959411621,
      "learning_rate": 4.4434126789866605e-05,
      "loss": 1.2155,
      "step": 36900
    },
    {
      "epoch": 1.1303232113398913,
      "grad_norm": 8.243475914001465,
      "learning_rate": 4.4418828784726476e-05,
      "loss": 1.3016,
      "step": 37000
    },
    {
      "epoch": 1.1333781389381072,
      "grad_norm": 6.644149303436279,
      "learning_rate": 4.440353077958634e-05,
      "loss": 1.2682,
      "step": 37100
    },
    {
      "epoch": 1.1364330665363231,
      "grad_norm": 1.5577857494354248,
      "learning_rate": 4.438823277444621e-05,
      "loss": 1.225,
      "step": 37200
    },
    {
      "epoch": 1.139487994134539,
      "grad_norm": 6.358497619628906,
      "learning_rate": 4.437293476930608e-05,
      "loss": 1.2043,
      "step": 37300
    },
    {
      "epoch": 1.142542921732755,
      "grad_norm": 4.491862773895264,
      "learning_rate": 4.4357636764165954e-05,
      "loss": 1.2241,
      "step": 37400
    },
    {
      "epoch": 1.1455978493309709,
      "grad_norm": 11.254999160766602,
      "learning_rate": 4.4342338759025826e-05,
      "loss": 1.3232,
      "step": 37500
    },
    {
      "epoch": 1.1486527769291868,
      "grad_norm": 6.52307653427124,
      "learning_rate": 4.43270407538857e-05,
      "loss": 1.1893,
      "step": 37600
    },
    {
      "epoch": 1.1517077045274027,
      "grad_norm": 1.5769741535186768,
      "learning_rate": 4.431174274874557e-05,
      "loss": 1.2663,
      "step": 37700
    },
    {
      "epoch": 1.1547626321256186,
      "grad_norm": 11.211283683776855,
      "learning_rate": 4.429644474360544e-05,
      "loss": 1.1015,
      "step": 37800
    },
    {
      "epoch": 1.1578175597238345,
      "grad_norm": 4.9694671630859375,
      "learning_rate": 4.42811467384653e-05,
      "loss": 1.1474,
      "step": 37900
    },
    {
      "epoch": 1.1608724873220504,
      "grad_norm": 4.092493057250977,
      "learning_rate": 4.426600171337658e-05,
      "loss": 1.4885,
      "step": 38000
    },
    {
      "epoch": 1.1639274149202663,
      "grad_norm": 7.638351917266846,
      "learning_rate": 4.425070370823645e-05,
      "loss": 1.3411,
      "step": 38100
    },
    {
      "epoch": 1.1669823425184824,
      "grad_norm": 1.5930650234222412,
      "learning_rate": 4.4235405703096314e-05,
      "loss": 1.0829,
      "step": 38200
    },
    {
      "epoch": 1.1700372701166981,
      "grad_norm": 2.2669153213500977,
      "learning_rate": 4.4220107697956185e-05,
      "loss": 1.2747,
      "step": 38300
    },
    {
      "epoch": 1.1730921977149142,
      "grad_norm": 9.886653900146484,
      "learning_rate": 4.4204809692816056e-05,
      "loss": 1.1561,
      "step": 38400
    },
    {
      "epoch": 1.1761471253131301,
      "grad_norm": 1.6329950094223022,
      "learning_rate": 4.418951168767593e-05,
      "loss": 1.1889,
      "step": 38500
    },
    {
      "epoch": 1.179202052911346,
      "grad_norm": 9.446663856506348,
      "learning_rate": 4.41742136825358e-05,
      "loss": 1.2176,
      "step": 38600
    },
    {
      "epoch": 1.182256980509562,
      "grad_norm": 3.5122339725494385,
      "learning_rate": 4.415891567739567e-05,
      "loss": 1.3452,
      "step": 38700
    },
    {
      "epoch": 1.1853119081077779,
      "grad_norm": 4.202631950378418,
      "learning_rate": 4.414361767225554e-05,
      "loss": 1.1617,
      "step": 38800
    },
    {
      "epoch": 1.1883668357059938,
      "grad_norm": 4.116245746612549,
      "learning_rate": 4.412831966711541e-05,
      "loss": 1.3976,
      "step": 38900
    },
    {
      "epoch": 1.1914217633042097,
      "grad_norm": 3.6785387992858887,
      "learning_rate": 4.4113021661975276e-05,
      "loss": 1.2608,
      "step": 39000
    },
    {
      "epoch": 1.1944766909024256,
      "grad_norm": 4.720080375671387,
      "learning_rate": 4.409772365683515e-05,
      "loss": 1.2868,
      "step": 39100
    },
    {
      "epoch": 1.1975316185006415,
      "grad_norm": 8.487894058227539,
      "learning_rate": 4.408242565169502e-05,
      "loss": 1.2178,
      "step": 39200
    },
    {
      "epoch": 1.2005865460988574,
      "grad_norm": 12.185461044311523,
      "learning_rate": 4.406712764655489e-05,
      "loss": 1.1529,
      "step": 39300
    },
    {
      "epoch": 1.2036414736970733,
      "grad_norm": 4.981146335601807,
      "learning_rate": 4.405182964141476e-05,
      "loss": 1.2508,
      "step": 39400
    },
    {
      "epoch": 1.2066964012952892,
      "grad_norm": 2.295912027359009,
      "learning_rate": 4.403653163627463e-05,
      "loss": 1.1667,
      "step": 39500
    },
    {
      "epoch": 1.2097513288935051,
      "grad_norm": 4.060471057891846,
      "learning_rate": 4.40212336311345e-05,
      "loss": 1.2019,
      "step": 39600
    },
    {
      "epoch": 1.2128062564917212,
      "grad_norm": 2.7551541328430176,
      "learning_rate": 4.4005935625994374e-05,
      "loss": 1.2756,
      "step": 39700
    },
    {
      "epoch": 1.2158611840899372,
      "grad_norm": 2.792501926422119,
      "learning_rate": 4.399063762085424e-05,
      "loss": 1.3358,
      "step": 39800
    },
    {
      "epoch": 1.218916111688153,
      "grad_norm": 16.728858947753906,
      "learning_rate": 4.397533961571411e-05,
      "loss": 1.2537,
      "step": 39900
    },
    {
      "epoch": 1.221971039286369,
      "grad_norm": 12.136491775512695,
      "learning_rate": 4.3960194590625384e-05,
      "loss": 1.3623,
      "step": 40000
    },
    {
      "epoch": 1.2250259668845849,
      "grad_norm": 3.166687488555908,
      "learning_rate": 4.3944896585485256e-05,
      "loss": 1.2557,
      "step": 40100
    },
    {
      "epoch": 1.2280808944828008,
      "grad_norm": 3.6806106567382812,
      "learning_rate": 4.392959858034513e-05,
      "loss": 1.3083,
      "step": 40200
    },
    {
      "epoch": 1.2311358220810167,
      "grad_norm": 2.7585878372192383,
      "learning_rate": 4.3914300575205e-05,
      "loss": 1.2948,
      "step": 40300
    },
    {
      "epoch": 1.2341907496792326,
      "grad_norm": 8.795299530029297,
      "learning_rate": 4.389900257006487e-05,
      "loss": 1.2998,
      "step": 40400
    },
    {
      "epoch": 1.2372456772774485,
      "grad_norm": 5.048892498016357,
      "learning_rate": 4.388370456492474e-05,
      "loss": 1.2627,
      "step": 40500
    },
    {
      "epoch": 1.2403006048756644,
      "grad_norm": 1.6317594051361084,
      "learning_rate": 4.386840655978461e-05,
      "loss": 1.344,
      "step": 40600
    },
    {
      "epoch": 1.2433555324738803,
      "grad_norm": 4.000895977020264,
      "learning_rate": 4.3853108554644476e-05,
      "loss": 1.1493,
      "step": 40700
    },
    {
      "epoch": 1.2464104600720962,
      "grad_norm": 4.536451816558838,
      "learning_rate": 4.383781054950435e-05,
      "loss": 1.185,
      "step": 40800
    },
    {
      "epoch": 1.2494653876703121,
      "grad_norm": 1.5634673833847046,
      "learning_rate": 4.382251254436422e-05,
      "loss": 1.1928,
      "step": 40900
    },
    {
      "epoch": 1.2525203152685283,
      "grad_norm": 3.4632925987243652,
      "learning_rate": 4.380721453922409e-05,
      "loss": 1.1444,
      "step": 41000
    },
    {
      "epoch": 1.255575242866744,
      "grad_norm": 1.5671478509902954,
      "learning_rate": 4.379191653408396e-05,
      "loss": 1.1633,
      "step": 41100
    },
    {
      "epoch": 1.25863017046496,
      "grad_norm": 4.259028434753418,
      "learning_rate": 4.377661852894383e-05,
      "loss": 1.3727,
      "step": 41200
    },
    {
      "epoch": 1.261685098063176,
      "grad_norm": 4.04819393157959,
      "learning_rate": 4.37613205238037e-05,
      "loss": 1.2986,
      "step": 41300
    },
    {
      "epoch": 1.2647400256613919,
      "grad_norm": 5.768520355224609,
      "learning_rate": 4.3746022518663574e-05,
      "loss": 1.2835,
      "step": 41400
    },
    {
      "epoch": 1.2677949532596078,
      "grad_norm": 2.447608470916748,
      "learning_rate": 4.373072451352344e-05,
      "loss": 1.1759,
      "step": 41500
    },
    {
      "epoch": 1.2708498808578237,
      "grad_norm": 9.211963653564453,
      "learning_rate": 4.371542650838331e-05,
      "loss": 1.3477,
      "step": 41600
    },
    {
      "epoch": 1.2739048084560396,
      "grad_norm": 4.363831043243408,
      "learning_rate": 4.370012850324318e-05,
      "loss": 1.3011,
      "step": 41700
    },
    {
      "epoch": 1.2769597360542555,
      "grad_norm": 5.887541770935059,
      "learning_rate": 4.368483049810305e-05,
      "loss": 1.2085,
      "step": 41800
    },
    {
      "epoch": 1.2800146636524714,
      "grad_norm": 4.435642719268799,
      "learning_rate": 4.366953249296292e-05,
      "loss": 1.2673,
      "step": 41900
    },
    {
      "epoch": 1.2830695912506873,
      "grad_norm": 5.838521480560303,
      "learning_rate": 4.3654234487822794e-05,
      "loss": 1.366,
      "step": 42000
    },
    {
      "epoch": 1.2861245188489032,
      "grad_norm": 3.4145562648773193,
      "learning_rate": 4.3638936482682665e-05,
      "loss": 1.1811,
      "step": 42100
    },
    {
      "epoch": 1.2891794464471191,
      "grad_norm": 6.8241868019104,
      "learning_rate": 4.362379145759393e-05,
      "loss": 1.3298,
      "step": 42200
    },
    {
      "epoch": 1.2922343740453353,
      "grad_norm": 11.387208938598633,
      "learning_rate": 4.3608493452453804e-05,
      "loss": 1.2655,
      "step": 42300
    },
    {
      "epoch": 1.295289301643551,
      "grad_norm": 2.6375372409820557,
      "learning_rate": 4.3593195447313676e-05,
      "loss": 1.2064,
      "step": 42400
    },
    {
      "epoch": 1.298344229241767,
      "grad_norm": 5.372217178344727,
      "learning_rate": 4.357789744217354e-05,
      "loss": 1.3267,
      "step": 42500
    },
    {
      "epoch": 1.301399156839983,
      "grad_norm": 4.916769027709961,
      "learning_rate": 4.356259943703341e-05,
      "loss": 1.1147,
      "step": 42600
    },
    {
      "epoch": 1.3044540844381989,
      "grad_norm": 4.316017150878906,
      "learning_rate": 4.354730143189328e-05,
      "loss": 1.3126,
      "step": 42700
    },
    {
      "epoch": 1.3075090120364148,
      "grad_norm": 2.563248634338379,
      "learning_rate": 4.353200342675315e-05,
      "loss": 1.2966,
      "step": 42800
    },
    {
      "epoch": 1.3105639396346307,
      "grad_norm": 3.107074022293091,
      "learning_rate": 4.3516705421613025e-05,
      "loss": 1.3133,
      "step": 42900
    },
    {
      "epoch": 1.3136188672328466,
      "grad_norm": 6.040319919586182,
      "learning_rate": 4.3501407416472896e-05,
      "loss": 1.2535,
      "step": 43000
    },
    {
      "epoch": 1.3166737948310625,
      "grad_norm": 3.597527503967285,
      "learning_rate": 4.348610941133277e-05,
      "loss": 1.2451,
      "step": 43100
    },
    {
      "epoch": 1.3197287224292784,
      "grad_norm": 4.143153190612793,
      "learning_rate": 4.347081140619264e-05,
      "loss": 1.2788,
      "step": 43200
    },
    {
      "epoch": 1.3227836500274943,
      "grad_norm": 3.319704055786133,
      "learning_rate": 4.34555134010525e-05,
      "loss": 1.1359,
      "step": 43300
    },
    {
      "epoch": 1.3258385776257102,
      "grad_norm": 5.486532211303711,
      "learning_rate": 4.3440215395912374e-05,
      "loss": 1.1321,
      "step": 43400
    },
    {
      "epoch": 1.3288935052239261,
      "grad_norm": 6.724696636199951,
      "learning_rate": 4.3424917390772245e-05,
      "loss": 1.0973,
      "step": 43500
    },
    {
      "epoch": 1.331948432822142,
      "grad_norm": 5.1426591873168945,
      "learning_rate": 4.3409619385632116e-05,
      "loss": 1.2437,
      "step": 43600
    },
    {
      "epoch": 1.335003360420358,
      "grad_norm": 8.841686248779297,
      "learning_rate": 4.339432138049199e-05,
      "loss": 1.3292,
      "step": 43700
    },
    {
      "epoch": 1.338058288018574,
      "grad_norm": 3.7120001316070557,
      "learning_rate": 4.337902337535186e-05,
      "loss": 1.1654,
      "step": 43800
    },
    {
      "epoch": 1.3411132156167898,
      "grad_norm": 1.776436686515808,
      "learning_rate": 4.336372537021173e-05,
      "loss": 1.2399,
      "step": 43900
    },
    {
      "epoch": 1.3441681432150059,
      "grad_norm": 3.7421560287475586,
      "learning_rate": 4.33484273650716e-05,
      "loss": 1.3145,
      "step": 44000
    },
    {
      "epoch": 1.3472230708132218,
      "grad_norm": 14.425456047058105,
      "learning_rate": 4.3333129359931465e-05,
      "loss": 1.2523,
      "step": 44100
    },
    {
      "epoch": 1.3502779984114377,
      "grad_norm": 5.034737586975098,
      "learning_rate": 4.331798433484274e-05,
      "loss": 1.2599,
      "step": 44200
    },
    {
      "epoch": 1.3533329260096536,
      "grad_norm": 9.89389705657959,
      "learning_rate": 4.330268632970261e-05,
      "loss": 1.2971,
      "step": 44300
    },
    {
      "epoch": 1.3563878536078695,
      "grad_norm": 2.4960923194885254,
      "learning_rate": 4.3287388324562475e-05,
      "loss": 1.2746,
      "step": 44400
    },
    {
      "epoch": 1.3594427812060854,
      "grad_norm": 12.708929061889648,
      "learning_rate": 4.3272090319422346e-05,
      "loss": 1.1215,
      "step": 44500
    },
    {
      "epoch": 1.3624977088043013,
      "grad_norm": 5.918141841888428,
      "learning_rate": 4.325679231428222e-05,
      "loss": 1.155,
      "step": 44600
    },
    {
      "epoch": 1.3655526364025172,
      "grad_norm": 1.8433784246444702,
      "learning_rate": 4.324149430914209e-05,
      "loss": 1.4129,
      "step": 44700
    },
    {
      "epoch": 1.3686075640007331,
      "grad_norm": 9.525556564331055,
      "learning_rate": 4.322619630400196e-05,
      "loss": 1.2244,
      "step": 44800
    },
    {
      "epoch": 1.371662491598949,
      "grad_norm": 3.9486777782440186,
      "learning_rate": 4.321089829886183e-05,
      "loss": 1.2334,
      "step": 44900
    },
    {
      "epoch": 1.374717419197165,
      "grad_norm": 2.2689762115478516,
      "learning_rate": 4.31956002937217e-05,
      "loss": 1.0697,
      "step": 45000
    },
    {
      "epoch": 1.377772346795381,
      "grad_norm": 7.874233245849609,
      "learning_rate": 4.318030228858157e-05,
      "loss": 1.2112,
      "step": 45100
    },
    {
      "epoch": 1.3808272743935968,
      "grad_norm": 1.4350064992904663,
      "learning_rate": 4.316500428344144e-05,
      "loss": 1.0985,
      "step": 45200
    },
    {
      "epoch": 1.3838822019918129,
      "grad_norm": 3.0830788612365723,
      "learning_rate": 4.314970627830131e-05,
      "loss": 1.1946,
      "step": 45300
    },
    {
      "epoch": 1.3869371295900288,
      "grad_norm": 2.0088775157928467,
      "learning_rate": 4.313440827316118e-05,
      "loss": 1.3359,
      "step": 45400
    },
    {
      "epoch": 1.3899920571882447,
      "grad_norm": 3.8786025047302246,
      "learning_rate": 4.311911026802105e-05,
      "loss": 1.2748,
      "step": 45500
    },
    {
      "epoch": 1.3930469847864606,
      "grad_norm": 6.242239952087402,
      "learning_rate": 4.310381226288092e-05,
      "loss": 1.4031,
      "step": 45600
    },
    {
      "epoch": 1.3961019123846765,
      "grad_norm": 2.5156424045562744,
      "learning_rate": 4.3088514257740793e-05,
      "loss": 1.289,
      "step": 45700
    },
    {
      "epoch": 1.3991568399828924,
      "grad_norm": 4.84622049331665,
      "learning_rate": 4.3073216252600665e-05,
      "loss": 1.1921,
      "step": 45800
    },
    {
      "epoch": 1.4022117675811083,
      "grad_norm": 4.694828033447266,
      "learning_rate": 4.3057918247460536e-05,
      "loss": 1.1959,
      "step": 45900
    },
    {
      "epoch": 1.4052666951793242,
      "grad_norm": 3.817330837249756,
      "learning_rate": 4.30426202423204e-05,
      "loss": 1.3833,
      "step": 46000
    },
    {
      "epoch": 1.4083216227775401,
      "grad_norm": 4.423964500427246,
      "learning_rate": 4.302732223718027e-05,
      "loss": 1.2469,
      "step": 46100
    },
    {
      "epoch": 1.411376550375756,
      "grad_norm": 6.940568447113037,
      "learning_rate": 4.301202423204014e-05,
      "loss": 1.0547,
      "step": 46200
    },
    {
      "epoch": 1.414431477973972,
      "grad_norm": 12.37020206451416,
      "learning_rate": 4.2996726226900014e-05,
      "loss": 1.2522,
      "step": 46300
    },
    {
      "epoch": 1.4174864055721879,
      "grad_norm": 2.3224236965179443,
      "learning_rate": 4.298158120181129e-05,
      "loss": 1.1729,
      "step": 46400
    },
    {
      "epoch": 1.4205413331704038,
      "grad_norm": 4.079832553863525,
      "learning_rate": 4.296628319667116e-05,
      "loss": 1.1279,
      "step": 46500
    },
    {
      "epoch": 1.42359626076862,
      "grad_norm": 1.1195186376571655,
      "learning_rate": 4.295098519153103e-05,
      "loss": 1.4004,
      "step": 46600
    },
    {
      "epoch": 1.4266511883668356,
      "grad_norm": 4.244978904724121,
      "learning_rate": 4.29356871863909e-05,
      "loss": 1.1016,
      "step": 46700
    },
    {
      "epoch": 1.4297061159650517,
      "grad_norm": 7.724494934082031,
      "learning_rate": 4.292038918125077e-05,
      "loss": 1.2749,
      "step": 46800
    },
    {
      "epoch": 1.4327610435632676,
      "grad_norm": 4.730603218078613,
      "learning_rate": 4.290509117611064e-05,
      "loss": 1.1651,
      "step": 46900
    },
    {
      "epoch": 1.4358159711614835,
      "grad_norm": 2.9875614643096924,
      "learning_rate": 4.28897931709705e-05,
      "loss": 1.2168,
      "step": 47000
    },
    {
      "epoch": 1.4388708987596994,
      "grad_norm": 3.3418843746185303,
      "learning_rate": 4.287449516583037e-05,
      "loss": 1.1436,
      "step": 47100
    },
    {
      "epoch": 1.4419258263579153,
      "grad_norm": 6.813102722167969,
      "learning_rate": 4.2859197160690244e-05,
      "loss": 1.1113,
      "step": 47200
    },
    {
      "epoch": 1.4449807539561312,
      "grad_norm": 6.928761959075928,
      "learning_rate": 4.2843899155550115e-05,
      "loss": 1.2056,
      "step": 47300
    },
    {
      "epoch": 1.4480356815543471,
      "grad_norm": 3.8548502922058105,
      "learning_rate": 4.2828601150409986e-05,
      "loss": 1.2025,
      "step": 47400
    },
    {
      "epoch": 1.451090609152563,
      "grad_norm": 1.8238892555236816,
      "learning_rate": 4.281330314526986e-05,
      "loss": 1.3061,
      "step": 47500
    },
    {
      "epoch": 1.454145536750779,
      "grad_norm": 4.400331974029541,
      "learning_rate": 4.279800514012973e-05,
      "loss": 1.2991,
      "step": 47600
    },
    {
      "epoch": 1.4572004643489949,
      "grad_norm": 6.429203510284424,
      "learning_rate": 4.27827071349896e-05,
      "loss": 1.3095,
      "step": 47700
    },
    {
      "epoch": 1.4602553919472108,
      "grad_norm": 4.6192803382873535,
      "learning_rate": 4.2767409129849464e-05,
      "loss": 1.3917,
      "step": 47800
    },
    {
      "epoch": 1.463310319545427,
      "grad_norm": 1.5252143144607544,
      "learning_rate": 4.2752111124709335e-05,
      "loss": 1.1591,
      "step": 47900
    },
    {
      "epoch": 1.4663652471436426,
      "grad_norm": 6.269423484802246,
      "learning_rate": 4.2736813119569206e-05,
      "loss": 1.1887,
      "step": 48000
    },
    {
      "epoch": 1.4694201747418587,
      "grad_norm": 13.407902717590332,
      "learning_rate": 4.272151511442908e-05,
      "loss": 1.178,
      "step": 48100
    },
    {
      "epoch": 1.4724751023400746,
      "grad_norm": 6.713379859924316,
      "learning_rate": 4.270621710928895e-05,
      "loss": 1.1894,
      "step": 48200
    },
    {
      "epoch": 1.4755300299382905,
      "grad_norm": 9.634526252746582,
      "learning_rate": 4.269091910414882e-05,
      "loss": 1.3624,
      "step": 48300
    },
    {
      "epoch": 1.4785849575365064,
      "grad_norm": 1.1485018730163574,
      "learning_rate": 4.267562109900869e-05,
      "loss": 1.0988,
      "step": 48400
    },
    {
      "epoch": 1.4816398851347223,
      "grad_norm": 6.0236358642578125,
      "learning_rate": 4.266032309386856e-05,
      "loss": 1.16,
      "step": 48500
    },
    {
      "epoch": 1.4846948127329382,
      "grad_norm": 2.204787254333496,
      "learning_rate": 4.264517806877984e-05,
      "loss": 1.22,
      "step": 48600
    },
    {
      "epoch": 1.4877497403311541,
      "grad_norm": 8.660630226135254,
      "learning_rate": 4.26298800636397e-05,
      "loss": 1.3341,
      "step": 48700
    },
    {
      "epoch": 1.49080466792937,
      "grad_norm": 18.680225372314453,
      "learning_rate": 4.261458205849957e-05,
      "loss": 1.1561,
      "step": 48800
    },
    {
      "epoch": 1.493859595527586,
      "grad_norm": 10.239827156066895,
      "learning_rate": 4.2599284053359444e-05,
      "loss": 1.1995,
      "step": 48900
    },
    {
      "epoch": 1.4969145231258019,
      "grad_norm": 12.529239654541016,
      "learning_rate": 4.2583986048219315e-05,
      "loss": 1.3687,
      "step": 49000
    },
    {
      "epoch": 1.4999694507240178,
      "grad_norm": 5.2959465980529785,
      "learning_rate": 4.2568688043079186e-05,
      "loss": 1.2584,
      "step": 49100
    },
    {
      "epoch": 1.503024378322234,
      "grad_norm": 4.77400541305542,
      "learning_rate": 4.255339003793906e-05,
      "loss": 1.2276,
      "step": 49200
    },
    {
      "epoch": 1.5060793059204496,
      "grad_norm": 2.8349411487579346,
      "learning_rate": 4.253809203279893e-05,
      "loss": 1.1254,
      "step": 49300
    },
    {
      "epoch": 1.5091342335186657,
      "grad_norm": 4.453884601593018,
      "learning_rate": 4.25227940276588e-05,
      "loss": 1.2116,
      "step": 49400
    },
    {
      "epoch": 1.5121891611168814,
      "grad_norm": 5.747006416320801,
      "learning_rate": 4.2507496022518664e-05,
      "loss": 1.0758,
      "step": 49500
    },
    {
      "epoch": 1.5152440887150975,
      "grad_norm": 3.7872657775878906,
      "learning_rate": 4.2492198017378535e-05,
      "loss": 1.2889,
      "step": 49600
    },
    {
      "epoch": 1.5182990163133132,
      "grad_norm": 5.094516277313232,
      "learning_rate": 4.2476900012238406e-05,
      "loss": 1.2575,
      "step": 49700
    },
    {
      "epoch": 1.5213539439115293,
      "grad_norm": 8.902609825134277,
      "learning_rate": 4.246160200709828e-05,
      "loss": 1.1852,
      "step": 49800
    },
    {
      "epoch": 1.5244088715097452,
      "grad_norm": 4.5981292724609375,
      "learning_rate": 4.244630400195815e-05,
      "loss": 1.105,
      "step": 49900
    },
    {
      "epoch": 1.5274637991079612,
      "grad_norm": 8.625105857849121,
      "learning_rate": 4.243100599681802e-05,
      "loss": 1.171,
      "step": 50000
    },
    {
      "epoch": 1.530518726706177,
      "grad_norm": 4.452911853790283,
      "learning_rate": 4.241570799167789e-05,
      "loss": 1.205,
      "step": 50100
    },
    {
      "epoch": 1.533573654304393,
      "grad_norm": 6.153209686279297,
      "learning_rate": 4.240040998653776e-05,
      "loss": 1.1071,
      "step": 50200
    },
    {
      "epoch": 1.5366285819026089,
      "grad_norm": 3.6518421173095703,
      "learning_rate": 4.2385111981397626e-05,
      "loss": 1.4151,
      "step": 50300
    },
    {
      "epoch": 1.5396835095008248,
      "grad_norm": 3.2835469245910645,
      "learning_rate": 4.23698139762575e-05,
      "loss": 1.2222,
      "step": 50400
    },
    {
      "epoch": 1.542738437099041,
      "grad_norm": 4.837490558624268,
      "learning_rate": 4.235451597111737e-05,
      "loss": 1.2641,
      "step": 50500
    },
    {
      "epoch": 1.5457933646972566,
      "grad_norm": 3.214068651199341,
      "learning_rate": 4.233937094602864e-05,
      "loss": 1.3104,
      "step": 50600
    },
    {
      "epoch": 1.5488482922954727,
      "grad_norm": 3.558753728866577,
      "learning_rate": 4.232407294088851e-05,
      "loss": 1.3417,
      "step": 50700
    },
    {
      "epoch": 1.5519032198936884,
      "grad_norm": 3.0552566051483154,
      "learning_rate": 4.230877493574838e-05,
      "loss": 1.2345,
      "step": 50800
    },
    {
      "epoch": 1.5549581474919045,
      "grad_norm": 5.351703643798828,
      "learning_rate": 4.229347693060825e-05,
      "loss": 1.0696,
      "step": 50900
    },
    {
      "epoch": 1.5580130750901202,
      "grad_norm": 3.6090290546417236,
      "learning_rate": 4.227817892546812e-05,
      "loss": 1.0948,
      "step": 51000
    },
    {
      "epoch": 1.5610680026883363,
      "grad_norm": 3.92417049407959,
      "learning_rate": 4.226288092032799e-05,
      "loss": 1.1325,
      "step": 51100
    },
    {
      "epoch": 1.5641229302865522,
      "grad_norm": 5.536214351654053,
      "learning_rate": 4.2247582915187864e-05,
      "loss": 1.1196,
      "step": 51200
    },
    {
      "epoch": 1.5671778578847682,
      "grad_norm": 1.7182681560516357,
      "learning_rate": 4.2232284910047735e-05,
      "loss": 1.3105,
      "step": 51300
    },
    {
      "epoch": 1.570232785482984,
      "grad_norm": 2.2255499362945557,
      "learning_rate": 4.22169869049076e-05,
      "loss": 1.0988,
      "step": 51400
    },
    {
      "epoch": 1.5732877130812,
      "grad_norm": 2.182570695877075,
      "learning_rate": 4.220168889976747e-05,
      "loss": 1.0547,
      "step": 51500
    },
    {
      "epoch": 1.5763426406794159,
      "grad_norm": 4.099241733551025,
      "learning_rate": 4.218639089462734e-05,
      "loss": 1.2795,
      "step": 51600
    },
    {
      "epoch": 1.5793975682776318,
      "grad_norm": 2.609814167022705,
      "learning_rate": 4.217109288948721e-05,
      "loss": 1.132,
      "step": 51700
    },
    {
      "epoch": 1.582452495875848,
      "grad_norm": 4.413924217224121,
      "learning_rate": 4.2155794884347084e-05,
      "loss": 1.1425,
      "step": 51800
    },
    {
      "epoch": 1.5855074234740636,
      "grad_norm": 1.853229284286499,
      "learning_rate": 4.2140496879206955e-05,
      "loss": 1.2918,
      "step": 51900
    },
    {
      "epoch": 1.5885623510722797,
      "grad_norm": 1.5005695819854736,
      "learning_rate": 4.2125198874066826e-05,
      "loss": 1.1795,
      "step": 52000
    },
    {
      "epoch": 1.5916172786704954,
      "grad_norm": 2.4563257694244385,
      "learning_rate": 4.21099008689267e-05,
      "loss": 1.2424,
      "step": 52100
    },
    {
      "epoch": 1.5946722062687115,
      "grad_norm": 1.5169684886932373,
      "learning_rate": 4.209460286378656e-05,
      "loss": 1.2374,
      "step": 52200
    },
    {
      "epoch": 1.5977271338669272,
      "grad_norm": 3.0069658756256104,
      "learning_rate": 4.207930485864643e-05,
      "loss": 1.2134,
      "step": 52300
    },
    {
      "epoch": 1.6007820614651433,
      "grad_norm": 1.1836366653442383,
      "learning_rate": 4.2064006853506304e-05,
      "loss": 1.0342,
      "step": 52400
    },
    {
      "epoch": 1.6038369890633593,
      "grad_norm": 5.595178604125977,
      "learning_rate": 4.2048708848366175e-05,
      "loss": 1.1342,
      "step": 52500
    },
    {
      "epoch": 1.6068919166615752,
      "grad_norm": 5.822509288787842,
      "learning_rate": 4.203356382327744e-05,
      "loss": 1.5188,
      "step": 52600
    },
    {
      "epoch": 1.609946844259791,
      "grad_norm": 3.0418248176574707,
      "learning_rate": 4.2018265818137314e-05,
      "loss": 1.0407,
      "step": 52700
    },
    {
      "epoch": 1.613001771858007,
      "grad_norm": 1.6683357954025269,
      "learning_rate": 4.2002967812997185e-05,
      "loss": 1.1343,
      "step": 52800
    },
    {
      "epoch": 1.6160566994562229,
      "grad_norm": 5.703723907470703,
      "learning_rate": 4.1987669807857057e-05,
      "loss": 1.1517,
      "step": 52900
    },
    {
      "epoch": 1.6191116270544388,
      "grad_norm": 3.9760212898254395,
      "learning_rate": 4.197237180271693e-05,
      "loss": 1.2402,
      "step": 53000
    },
    {
      "epoch": 1.6221665546526547,
      "grad_norm": 5.224232196807861,
      "learning_rate": 4.19570737975768e-05,
      "loss": 1.089,
      "step": 53100
    },
    {
      "epoch": 1.6252214822508706,
      "grad_norm": 8.600234985351562,
      "learning_rate": 4.194177579243666e-05,
      "loss": 1.2579,
      "step": 53200
    },
    {
      "epoch": 1.6282764098490867,
      "grad_norm": 2.693863868713379,
      "learning_rate": 4.1926477787296534e-05,
      "loss": 1.2177,
      "step": 53300
    },
    {
      "epoch": 1.6313313374473024,
      "grad_norm": 2.958672046661377,
      "learning_rate": 4.1911179782156406e-05,
      "loss": 1.1448,
      "step": 53400
    },
    {
      "epoch": 1.6343862650455185,
      "grad_norm": 4.920342445373535,
      "learning_rate": 4.189588177701628e-05,
      "loss": 1.2276,
      "step": 53500
    },
    {
      "epoch": 1.6374411926437342,
      "grad_norm": 3.0722827911376953,
      "learning_rate": 4.188058377187615e-05,
      "loss": 1.1841,
      "step": 53600
    },
    {
      "epoch": 1.6404961202419504,
      "grad_norm": 4.592638969421387,
      "learning_rate": 4.186528576673602e-05,
      "loss": 1.3231,
      "step": 53700
    },
    {
      "epoch": 1.643551047840166,
      "grad_norm": 11.629206657409668,
      "learning_rate": 4.184998776159589e-05,
      "loss": 1.2177,
      "step": 53800
    },
    {
      "epoch": 1.6466059754383822,
      "grad_norm": 7.090597152709961,
      "learning_rate": 4.183468975645576e-05,
      "loss": 1.17,
      "step": 53900
    },
    {
      "epoch": 1.649660903036598,
      "grad_norm": 2.535602331161499,
      "learning_rate": 4.1819391751315626e-05,
      "loss": 1.1368,
      "step": 54000
    },
    {
      "epoch": 1.652715830634814,
      "grad_norm": 3.7535698413848877,
      "learning_rate": 4.18040937461755e-05,
      "loss": 1.241,
      "step": 54100
    },
    {
      "epoch": 1.6557707582330299,
      "grad_norm": 2.1686646938323975,
      "learning_rate": 4.178879574103537e-05,
      "loss": 1.2229,
      "step": 54200
    },
    {
      "epoch": 1.6588256858312458,
      "grad_norm": 5.6875433921813965,
      "learning_rate": 4.177349773589524e-05,
      "loss": 1.0793,
      "step": 54300
    },
    {
      "epoch": 1.6618806134294617,
      "grad_norm": 8.435276985168457,
      "learning_rate": 4.175819973075511e-05,
      "loss": 1.317,
      "step": 54400
    },
    {
      "epoch": 1.6649355410276776,
      "grad_norm": 4.2010369300842285,
      "learning_rate": 4.174290172561498e-05,
      "loss": 1.1259,
      "step": 54500
    },
    {
      "epoch": 1.6679904686258937,
      "grad_norm": 3.4158153533935547,
      "learning_rate": 4.172760372047485e-05,
      "loss": 1.2646,
      "step": 54600
    },
    {
      "epoch": 1.6710453962241094,
      "grad_norm": 12.006670951843262,
      "learning_rate": 4.1712305715334724e-05,
      "loss": 1.1726,
      "step": 54700
    },
    {
      "epoch": 1.6741003238223255,
      "grad_norm": 3.64186954498291,
      "learning_rate": 4.1697160690246e-05,
      "loss": 1.513,
      "step": 54800
    },
    {
      "epoch": 1.6771552514205412,
      "grad_norm": 6.152316093444824,
      "learning_rate": 4.168186268510586e-05,
      "loss": 1.1317,
      "step": 54900
    },
    {
      "epoch": 1.6802101790187574,
      "grad_norm": 5.903543472290039,
      "learning_rate": 4.1666564679965734e-05,
      "loss": 1.2468,
      "step": 55000
    },
    {
      "epoch": 1.683265106616973,
      "grad_norm": 8.211106300354004,
      "learning_rate": 4.1651266674825605e-05,
      "loss": 1.261,
      "step": 55100
    },
    {
      "epoch": 1.6863200342151892,
      "grad_norm": 2.6759305000305176,
      "learning_rate": 4.1635968669685476e-05,
      "loss": 1.2852,
      "step": 55200
    },
    {
      "epoch": 1.689374961813405,
      "grad_norm": 11.397104263305664,
      "learning_rate": 4.162067066454535e-05,
      "loss": 1.0922,
      "step": 55300
    },
    {
      "epoch": 1.692429889411621,
      "grad_norm": 7.850461959838867,
      "learning_rate": 4.160537265940522e-05,
      "loss": 1.1855,
      "step": 55400
    },
    {
      "epoch": 1.6954848170098369,
      "grad_norm": 9.271197319030762,
      "learning_rate": 4.159007465426509e-05,
      "loss": 1.1907,
      "step": 55500
    },
    {
      "epoch": 1.6985397446080528,
      "grad_norm": 2.4433703422546387,
      "learning_rate": 4.157477664912496e-05,
      "loss": 1.1701,
      "step": 55600
    },
    {
      "epoch": 1.7015946722062687,
      "grad_norm": 1.9493122100830078,
      "learning_rate": 4.1559478643984825e-05,
      "loss": 1.2341,
      "step": 55700
    },
    {
      "epoch": 1.7046495998044846,
      "grad_norm": 1.9953112602233887,
      "learning_rate": 4.1544180638844697e-05,
      "loss": 1.2181,
      "step": 55800
    },
    {
      "epoch": 1.7077045274027005,
      "grad_norm": 4.831192493438721,
      "learning_rate": 4.152888263370457e-05,
      "loss": 1.0811,
      "step": 55900
    },
    {
      "epoch": 1.7107594550009164,
      "grad_norm": 4.006892204284668,
      "learning_rate": 4.151358462856444e-05,
      "loss": 1.1862,
      "step": 56000
    },
    {
      "epoch": 1.7138143825991325,
      "grad_norm": 3.466545343399048,
      "learning_rate": 4.149828662342431e-05,
      "loss": 1.1184,
      "step": 56100
    },
    {
      "epoch": 1.7168693101973482,
      "grad_norm": 2.2860095500946045,
      "learning_rate": 4.148298861828418e-05,
      "loss": 1.1433,
      "step": 56200
    },
    {
      "epoch": 1.7199242377955644,
      "grad_norm": 2.8950862884521484,
      "learning_rate": 4.146769061314405e-05,
      "loss": 1.1777,
      "step": 56300
    },
    {
      "epoch": 1.72297916539378,
      "grad_norm": 4.594806671142578,
      "learning_rate": 4.1452392608003923e-05,
      "loss": 1.2518,
      "step": 56400
    },
    {
      "epoch": 1.7260340929919962,
      "grad_norm": 9.209367752075195,
      "learning_rate": 4.143709460286379e-05,
      "loss": 1.0943,
      "step": 56500
    },
    {
      "epoch": 1.7290890205902119,
      "grad_norm": 7.0374908447265625,
      "learning_rate": 4.142179659772366e-05,
      "loss": 1.0603,
      "step": 56600
    },
    {
      "epoch": 1.732143948188428,
      "grad_norm": 5.589990615844727,
      "learning_rate": 4.140649859258353e-05,
      "loss": 1.1151,
      "step": 56700
    },
    {
      "epoch": 1.735198875786644,
      "grad_norm": 2.3889482021331787,
      "learning_rate": 4.13912005874434e-05,
      "loss": 1.2266,
      "step": 56800
    },
    {
      "epoch": 1.7382538033848598,
      "grad_norm": 8.789185523986816,
      "learning_rate": 4.137605556235467e-05,
      "loss": 1.1758,
      "step": 56900
    },
    {
      "epoch": 1.7413087309830757,
      "grad_norm": 2.146101474761963,
      "learning_rate": 4.136075755721454e-05,
      "loss": 1.0922,
      "step": 57000
    },
    {
      "epoch": 1.7443636585812916,
      "grad_norm": 5.941165924072266,
      "learning_rate": 4.134545955207441e-05,
      "loss": 1.3349,
      "step": 57100
    },
    {
      "epoch": 1.7474185861795075,
      "grad_norm": 1.3134785890579224,
      "learning_rate": 4.133016154693428e-05,
      "loss": 1.1338,
      "step": 57200
    },
    {
      "epoch": 1.7504735137777234,
      "grad_norm": 3.1560745239257812,
      "learning_rate": 4.1314863541794154e-05,
      "loss": 1.2561,
      "step": 57300
    },
    {
      "epoch": 1.7535284413759396,
      "grad_norm": 1.139880895614624,
      "learning_rate": 4.1299565536654025e-05,
      "loss": 1.0852,
      "step": 57400
    },
    {
      "epoch": 1.7565833689741552,
      "grad_norm": 1.4170360565185547,
      "learning_rate": 4.128426753151389e-05,
      "loss": 1.342,
      "step": 57500
    },
    {
      "epoch": 1.7596382965723714,
      "grad_norm": 7.173251628875732,
      "learning_rate": 4.126896952637376e-05,
      "loss": 1.1708,
      "step": 57600
    },
    {
      "epoch": 1.762693224170587,
      "grad_norm": 3.3711791038513184,
      "learning_rate": 4.125367152123363e-05,
      "loss": 1.3737,
      "step": 57700
    },
    {
      "epoch": 1.7657481517688032,
      "grad_norm": 3.1640279293060303,
      "learning_rate": 4.12383735160935e-05,
      "loss": 1.2542,
      "step": 57800
    },
    {
      "epoch": 1.7688030793670189,
      "grad_norm": 5.858263969421387,
      "learning_rate": 4.1223075510953374e-05,
      "loss": 1.048,
      "step": 57900
    },
    {
      "epoch": 1.771858006965235,
      "grad_norm": 4.2827911376953125,
      "learning_rate": 4.1207777505813245e-05,
      "loss": 1.0881,
      "step": 58000
    },
    {
      "epoch": 1.774912934563451,
      "grad_norm": 1.3887640237808228,
      "learning_rate": 4.1192479500673116e-05,
      "loss": 1.2263,
      "step": 58100
    },
    {
      "epoch": 1.7779678621616668,
      "grad_norm": 3.8297247886657715,
      "learning_rate": 4.117718149553299e-05,
      "loss": 1.1554,
      "step": 58200
    },
    {
      "epoch": 1.7810227897598827,
      "grad_norm": 8.833450317382812,
      "learning_rate": 4.116188349039285e-05,
      "loss": 1.1723,
      "step": 58300
    },
    {
      "epoch": 1.7840777173580986,
      "grad_norm": 6.00846529006958,
      "learning_rate": 4.114658548525272e-05,
      "loss": 1.121,
      "step": 58400
    },
    {
      "epoch": 1.7871326449563145,
      "grad_norm": 7.4083733558654785,
      "learning_rate": 4.1131287480112594e-05,
      "loss": 1.231,
      "step": 58500
    },
    {
      "epoch": 1.7901875725545304,
      "grad_norm": 9.703535079956055,
      "learning_rate": 4.1115989474972465e-05,
      "loss": 1.3855,
      "step": 58600
    },
    {
      "epoch": 1.7932425001527463,
      "grad_norm": 2.700291395187378,
      "learning_rate": 4.1100691469832337e-05,
      "loss": 1.213,
      "step": 58700
    },
    {
      "epoch": 1.7962974277509622,
      "grad_norm": 9.795341491699219,
      "learning_rate": 4.108539346469221e-05,
      "loss": 1.3951,
      "step": 58800
    },
    {
      "epoch": 1.7993523553491784,
      "grad_norm": 4.961639881134033,
      "learning_rate": 4.107009545955208e-05,
      "loss": 1.0949,
      "step": 58900
    },
    {
      "epoch": 1.802407282947394,
      "grad_norm": 14.932546615600586,
      "learning_rate": 4.105495043446335e-05,
      "loss": 1.225,
      "step": 59000
    },
    {
      "epoch": 1.8054622105456102,
      "grad_norm": 0.001140201580710709,
      "learning_rate": 4.103965242932322e-05,
      "loss": 1.2837,
      "step": 59100
    },
    {
      "epoch": 1.8085171381438259,
      "grad_norm": 3.3460593223571777,
      "learning_rate": 4.102435442418309e-05,
      "loss": 1.1232,
      "step": 59200
    },
    {
      "epoch": 1.811572065742042,
      "grad_norm": 9.262133598327637,
      "learning_rate": 4.100905641904296e-05,
      "loss": 1.1341,
      "step": 59300
    },
    {
      "epoch": 1.8146269933402577,
      "grad_norm": 3.290485382080078,
      "learning_rate": 4.0993758413902825e-05,
      "loss": 1.1736,
      "step": 59400
    },
    {
      "epoch": 1.8176819209384738,
      "grad_norm": 8.72140121459961,
      "learning_rate": 4.0978460408762696e-05,
      "loss": 1.305,
      "step": 59500
    },
    {
      "epoch": 1.8207368485366897,
      "grad_norm": 2.2909820079803467,
      "learning_rate": 4.096316240362257e-05,
      "loss": 1.1022,
      "step": 59600
    },
    {
      "epoch": 1.8237917761349056,
      "grad_norm": 3.055588960647583,
      "learning_rate": 4.094786439848244e-05,
      "loss": 1.148,
      "step": 59700
    },
    {
      "epoch": 1.8268467037331215,
      "grad_norm": 3.7513368129730225,
      "learning_rate": 4.093256639334231e-05,
      "loss": 1.4451,
      "step": 59800
    },
    {
      "epoch": 1.8299016313313374,
      "grad_norm": 4.179325580596924,
      "learning_rate": 4.091726838820218e-05,
      "loss": 1.3197,
      "step": 59900
    },
    {
      "epoch": 1.8329565589295533,
      "grad_norm": 5.003783702850342,
      "learning_rate": 4.090197038306205e-05,
      "loss": 1.1448,
      "step": 60000
    },
    {
      "epoch": 1.8360114865277692,
      "grad_norm": 3.986809492111206,
      "learning_rate": 4.088667237792192e-05,
      "loss": 1.2398,
      "step": 60100
    },
    {
      "epoch": 1.8390664141259854,
      "grad_norm": 5.309907913208008,
      "learning_rate": 4.087137437278179e-05,
      "loss": 1.3249,
      "step": 60200
    },
    {
      "epoch": 1.842121341724201,
      "grad_norm": 1.3858393430709839,
      "learning_rate": 4.085607636764166e-05,
      "loss": 1.2089,
      "step": 60300
    },
    {
      "epoch": 1.8451762693224172,
      "grad_norm": 3.5477840900421143,
      "learning_rate": 4.084077836250153e-05,
      "loss": 1.1067,
      "step": 60400
    },
    {
      "epoch": 1.8482311969206329,
      "grad_norm": 3.756406545639038,
      "learning_rate": 4.08254803573614e-05,
      "loss": 1.3526,
      "step": 60500
    },
    {
      "epoch": 1.851286124518849,
      "grad_norm": 7.084662437438965,
      "learning_rate": 4.081018235222127e-05,
      "loss": 1.2392,
      "step": 60600
    },
    {
      "epoch": 1.8543410521170647,
      "grad_norm": 5.864764213562012,
      "learning_rate": 4.079488434708114e-05,
      "loss": 1.4818,
      "step": 60700
    },
    {
      "epoch": 1.8573959797152808,
      "grad_norm": 1.9365818500518799,
      "learning_rate": 4.0779586341941014e-05,
      "loss": 1.1989,
      "step": 60800
    },
    {
      "epoch": 1.8604509073134967,
      "grad_norm": 3.3976101875305176,
      "learning_rate": 4.0764288336800885e-05,
      "loss": 1.1127,
      "step": 60900
    },
    {
      "epoch": 1.8635058349117126,
      "grad_norm": 4.278215408325195,
      "learning_rate": 4.074899033166075e-05,
      "loss": 1.0595,
      "step": 61000
    },
    {
      "epoch": 1.8665607625099285,
      "grad_norm": 5.625969409942627,
      "learning_rate": 4.0733845306572024e-05,
      "loss": 1.3236,
      "step": 61100
    },
    {
      "epoch": 1.8696156901081444,
      "grad_norm": 5.3394365310668945,
      "learning_rate": 4.0718547301431896e-05,
      "loss": 1.2003,
      "step": 61200
    },
    {
      "epoch": 1.8726706177063603,
      "grad_norm": 6.125877380371094,
      "learning_rate": 4.0703402276343164e-05,
      "loss": 1.4561,
      "step": 61300
    },
    {
      "epoch": 1.8757255453045762,
      "grad_norm": 6.432266712188721,
      "learning_rate": 4.0688104271203035e-05,
      "loss": 1.273,
      "step": 61400
    },
    {
      "epoch": 1.8787804729027922,
      "grad_norm": 1.3589688539505005,
      "learning_rate": 4.0672806266062906e-05,
      "loss": 1.1759,
      "step": 61500
    },
    {
      "epoch": 1.881835400501008,
      "grad_norm": 22.972280502319336,
      "learning_rate": 4.065750826092278e-05,
      "loss": 1.2454,
      "step": 61600
    },
    {
      "epoch": 1.8848903280992242,
      "grad_norm": 4.607998371124268,
      "learning_rate": 4.064221025578265e-05,
      "loss": 1.351,
      "step": 61700
    },
    {
      "epoch": 1.8879452556974399,
      "grad_norm": 7.128314018249512,
      "learning_rate": 4.062691225064252e-05,
      "loss": 1.1201,
      "step": 61800
    },
    {
      "epoch": 1.891000183295656,
      "grad_norm": 6.283421993255615,
      "learning_rate": 4.061161424550239e-05,
      "loss": 1.3758,
      "step": 61900
    },
    {
      "epoch": 1.8940551108938717,
      "grad_norm": 2.3519177436828613,
      "learning_rate": 4.059631624036226e-05,
      "loss": 1.1033,
      "step": 62000
    },
    {
      "epoch": 1.8971100384920878,
      "grad_norm": 5.391026973724365,
      "learning_rate": 4.0581018235222126e-05,
      "loss": 1.1464,
      "step": 62100
    },
    {
      "epoch": 1.9001649660903035,
      "grad_norm": 2.404320240020752,
      "learning_rate": 4.0565720230082e-05,
      "loss": 1.1599,
      "step": 62200
    },
    {
      "epoch": 1.9032198936885196,
      "grad_norm": 1.6671347618103027,
      "learning_rate": 4.055042222494187e-05,
      "loss": 1.1476,
      "step": 62300
    },
    {
      "epoch": 1.9062748212867355,
      "grad_norm": 1.6540522575378418,
      "learning_rate": 4.053512421980174e-05,
      "loss": 1.1896,
      "step": 62400
    },
    {
      "epoch": 1.9093297488849514,
      "grad_norm": 17.489120483398438,
      "learning_rate": 4.051982621466161e-05,
      "loss": 1.0109,
      "step": 62500
    },
    {
      "epoch": 1.9123846764831673,
      "grad_norm": 8.429814338684082,
      "learning_rate": 4.050452820952148e-05,
      "loss": 1.0935,
      "step": 62600
    },
    {
      "epoch": 1.9154396040813833,
      "grad_norm": 6.822543144226074,
      "learning_rate": 4.048923020438135e-05,
      "loss": 1.1555,
      "step": 62700
    },
    {
      "epoch": 1.9184945316795992,
      "grad_norm": 3.066483497619629,
      "learning_rate": 4.0473932199241224e-05,
      "loss": 1.255,
      "step": 62800
    },
    {
      "epoch": 1.921549459277815,
      "grad_norm": 3.349703550338745,
      "learning_rate": 4.045863419410109e-05,
      "loss": 1.4048,
      "step": 62900
    },
    {
      "epoch": 1.9246043868760312,
      "grad_norm": 4.069706916809082,
      "learning_rate": 4.044333618896096e-05,
      "loss": 1.1313,
      "step": 63000
    },
    {
      "epoch": 1.9276593144742469,
      "grad_norm": 3.599609613418579,
      "learning_rate": 4.042803818382083e-05,
      "loss": 1.4575,
      "step": 63100
    },
    {
      "epoch": 1.930714242072463,
      "grad_norm": 4.190948486328125,
      "learning_rate": 4.04127401786807e-05,
      "loss": 1.2446,
      "step": 63200
    },
    {
      "epoch": 1.9337691696706787,
      "grad_norm": 4.354323863983154,
      "learning_rate": 4.039744217354057e-05,
      "loss": 1.1491,
      "step": 63300
    },
    {
      "epoch": 1.9368240972688948,
      "grad_norm": 5.348269939422607,
      "learning_rate": 4.0382144168400444e-05,
      "loss": 1.3107,
      "step": 63400
    },
    {
      "epoch": 1.9398790248671105,
      "grad_norm": 3.1174705028533936,
      "learning_rate": 4.0366846163260315e-05,
      "loss": 1.1938,
      "step": 63500
    },
    {
      "epoch": 1.9429339524653266,
      "grad_norm": 12.221755981445312,
      "learning_rate": 4.0351548158120187e-05,
      "loss": 1.1975,
      "step": 63600
    },
    {
      "epoch": 1.9459888800635425,
      "grad_norm": 7.461677551269531,
      "learning_rate": 4.033625015298005e-05,
      "loss": 1.0214,
      "step": 63700
    },
    {
      "epoch": 1.9490438076617584,
      "grad_norm": 4.798915863037109,
      "learning_rate": 4.032095214783992e-05,
      "loss": 1.3375,
      "step": 63800
    },
    {
      "epoch": 1.9520987352599743,
      "grad_norm": 3.232374906539917,
      "learning_rate": 4.030565414269979e-05,
      "loss": 1.1388,
      "step": 63900
    },
    {
      "epoch": 1.9551536628581903,
      "grad_norm": 6.82913875579834,
      "learning_rate": 4.0290356137559664e-05,
      "loss": 1.1365,
      "step": 64000
    },
    {
      "epoch": 1.9582085904564062,
      "grad_norm": 18.124427795410156,
      "learning_rate": 4.0275058132419536e-05,
      "loss": 1.196,
      "step": 64100
    },
    {
      "epoch": 1.961263518054622,
      "grad_norm": 2.9495561122894287,
      "learning_rate": 4.025976012727941e-05,
      "loss": 1.2184,
      "step": 64200
    },
    {
      "epoch": 1.964318445652838,
      "grad_norm": 5.870420455932617,
      "learning_rate": 4.024446212213928e-05,
      "loss": 1.1024,
      "step": 64300
    },
    {
      "epoch": 1.9673733732510539,
      "grad_norm": 36.10947036743164,
      "learning_rate": 4.022916411699915e-05,
      "loss": 1.5031,
      "step": 64400
    },
    {
      "epoch": 1.97042830084927,
      "grad_norm": 6.070558071136475,
      "learning_rate": 4.0213866111859013e-05,
      "loss": 1.3974,
      "step": 64500
    },
    {
      "epoch": 1.9734832284474857,
      "grad_norm": 2.5208094120025635,
      "learning_rate": 4.0198568106718885e-05,
      "loss": 1.1695,
      "step": 64600
    },
    {
      "epoch": 1.9765381560457018,
      "grad_norm": 4.907090187072754,
      "learning_rate": 4.0183270101578756e-05,
      "loss": 1.1204,
      "step": 64700
    },
    {
      "epoch": 1.9795930836439175,
      "grad_norm": 3.0557048320770264,
      "learning_rate": 4.016797209643863e-05,
      "loss": 1.2246,
      "step": 64800
    },
    {
      "epoch": 1.9826480112421336,
      "grad_norm": 9.957263946533203,
      "learning_rate": 4.01526740912985e-05,
      "loss": 1.0912,
      "step": 64900
    },
    {
      "epoch": 1.9857029388403493,
      "grad_norm": 12.20952320098877,
      "learning_rate": 4.013737608615837e-05,
      "loss": 1.1763,
      "step": 65000
    },
    {
      "epoch": 1.9887578664385654,
      "grad_norm": 2.2833447456359863,
      "learning_rate": 4.012207808101824e-05,
      "loss": 1.2259,
      "step": 65100
    },
    {
      "epoch": 1.9918127940367814,
      "grad_norm": 3.358288049697876,
      "learning_rate": 4.010678007587811e-05,
      "loss": 1.2257,
      "step": 65200
    },
    {
      "epoch": 1.9948677216349973,
      "grad_norm": 2.8749561309814453,
      "learning_rate": 4.0091482070737976e-05,
      "loss": 1.2282,
      "step": 65300
    },
    {
      "epoch": 1.9979226492332132,
      "grad_norm": 3.331350803375244,
      "learning_rate": 4.007633704564925e-05,
      "loss": 1.1915,
      "step": 65400
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.676513716624916,
      "eval_loss": 0.857006847858429,
      "eval_runtime": 1782.201,
      "eval_samples_per_second": 18.367,
      "eval_steps_per_second": 4.592,
      "step": 65468
    },
    {
      "epoch": 2.000977576831429,
      "grad_norm": 5.58222770690918,
      "learning_rate": 4.006103904050912e-05,
      "loss": 1.1468,
      "step": 65500
    },
    {
      "epoch": 2.004032504429645,
      "grad_norm": 3.9189741611480713,
      "learning_rate": 4.0045741035368986e-05,
      "loss": 1.044,
      "step": 65600
    },
    {
      "epoch": 2.007087432027861,
      "grad_norm": 5.215442180633545,
      "learning_rate": 4.003044303022886e-05,
      "loss": 1.1866,
      "step": 65700
    },
    {
      "epoch": 2.010142359626077,
      "grad_norm": 5.244242191314697,
      "learning_rate": 4.001514502508873e-05,
      "loss": 1.2026,
      "step": 65800
    },
    {
      "epoch": 2.0131972872242927,
      "grad_norm": 10.0203218460083,
      "learning_rate": 3.99998470199486e-05,
      "loss": 1.2438,
      "step": 65900
    },
    {
      "epoch": 2.016252214822509,
      "grad_norm": 24.34562873840332,
      "learning_rate": 3.998454901480847e-05,
      "loss": 1.2173,
      "step": 66000
    },
    {
      "epoch": 2.0193071424207245,
      "grad_norm": 13.229280471801758,
      "learning_rate": 3.996925100966834e-05,
      "loss": 1.2143,
      "step": 66100
    },
    {
      "epoch": 2.0223620700189406,
      "grad_norm": 5.116329669952393,
      "learning_rate": 3.995395300452821e-05,
      "loss": 1.2472,
      "step": 66200
    },
    {
      "epoch": 2.0254169976171563,
      "grad_norm": 2.4487433433532715,
      "learning_rate": 3.9938654999388084e-05,
      "loss": 1.3071,
      "step": 66300
    },
    {
      "epoch": 2.0284719252153725,
      "grad_norm": 1.7104898691177368,
      "learning_rate": 3.992335699424795e-05,
      "loss": 1.1499,
      "step": 66400
    },
    {
      "epoch": 2.031526852813588,
      "grad_norm": 1.6790919303894043,
      "learning_rate": 3.990805898910782e-05,
      "loss": 1.0116,
      "step": 66500
    },
    {
      "epoch": 2.0345817804118043,
      "grad_norm": 3.6588943004608154,
      "learning_rate": 3.989276098396769e-05,
      "loss": 1.2898,
      "step": 66600
    },
    {
      "epoch": 2.03763670801002,
      "grad_norm": 9.815014839172363,
      "learning_rate": 3.987746297882756e-05,
      "loss": 1.1312,
      "step": 66700
    },
    {
      "epoch": 2.040691635608236,
      "grad_norm": 3.727752208709717,
      "learning_rate": 3.986216497368743e-05,
      "loss": 1.1964,
      "step": 66800
    },
    {
      "epoch": 2.043746563206452,
      "grad_norm": 4.570804595947266,
      "learning_rate": 3.9846866968547304e-05,
      "loss": 1.1733,
      "step": 66900
    },
    {
      "epoch": 2.046801490804668,
      "grad_norm": 5.729971408843994,
      "learning_rate": 3.9831568963407176e-05,
      "loss": 1.1444,
      "step": 67000
    },
    {
      "epoch": 2.049856418402884,
      "grad_norm": 3.439488649368286,
      "learning_rate": 3.981627095826705e-05,
      "loss": 1.1501,
      "step": 67100
    },
    {
      "epoch": 2.0529113460010997,
      "grad_norm": 8.150869369506836,
      "learning_rate": 3.980097295312691e-05,
      "loss": 1.1009,
      "step": 67200
    },
    {
      "epoch": 2.055966273599316,
      "grad_norm": 5.003384590148926,
      "learning_rate": 3.978567494798678e-05,
      "loss": 1.1368,
      "step": 67300
    },
    {
      "epoch": 2.0590212011975315,
      "grad_norm": Infinity,
      "learning_rate": 3.977052992289805e-05,
      "loss": 1.0697,
      "step": 67400
    },
    {
      "epoch": 2.0620761287957476,
      "grad_norm": 2.6935789585113525,
      "learning_rate": 3.975523191775792e-05,
      "loss": 1.2134,
      "step": 67500
    },
    {
      "epoch": 2.0651310563939633,
      "grad_norm": 4.748971462249756,
      "learning_rate": 3.973993391261779e-05,
      "loss": 1.2622,
      "step": 67600
    },
    {
      "epoch": 2.0681859839921795,
      "grad_norm": 8.399434089660645,
      "learning_rate": 3.9724635907477664e-05,
      "loss": 1.1459,
      "step": 67700
    },
    {
      "epoch": 2.071240911590395,
      "grad_norm": 2.4680871963500977,
      "learning_rate": 3.9709337902337535e-05,
      "loss": 1.0392,
      "step": 67800
    },
    {
      "epoch": 2.0742958391886113,
      "grad_norm": 6.203686714172363,
      "learning_rate": 3.9694039897197406e-05,
      "loss": 1.0936,
      "step": 67900
    },
    {
      "epoch": 2.077350766786827,
      "grad_norm": 2.707479476928711,
      "learning_rate": 3.967874189205728e-05,
      "loss": 1.282,
      "step": 68000
    },
    {
      "epoch": 2.080405694385043,
      "grad_norm": 6.8733720779418945,
      "learning_rate": 3.966344388691715e-05,
      "loss": 1.2343,
      "step": 68100
    },
    {
      "epoch": 2.083460621983259,
      "grad_norm": 0.5146437883377075,
      "learning_rate": 3.964814588177701e-05,
      "loss": 1.0634,
      "step": 68200
    },
    {
      "epoch": 2.086515549581475,
      "grad_norm": 2.7069003582000732,
      "learning_rate": 3.9632847876636884e-05,
      "loss": 1.1075,
      "step": 68300
    },
    {
      "epoch": 2.089570477179691,
      "grad_norm": 4.112438678741455,
      "learning_rate": 3.9617549871496755e-05,
      "loss": 1.1765,
      "step": 68400
    },
    {
      "epoch": 2.0926254047779067,
      "grad_norm": 5.514732360839844,
      "learning_rate": 3.9602251866356626e-05,
      "loss": 1.1638,
      "step": 68500
    },
    {
      "epoch": 2.095680332376123,
      "grad_norm": 9.44277286529541,
      "learning_rate": 3.95871068412679e-05,
      "loss": 1.1738,
      "step": 68600
    },
    {
      "epoch": 2.0987352599743385,
      "grad_norm": 4.027563571929932,
      "learning_rate": 3.957180883612777e-05,
      "loss": 1.2366,
      "step": 68700
    },
    {
      "epoch": 2.1017901875725546,
      "grad_norm": 2.5387253761291504,
      "learning_rate": 3.955651083098764e-05,
      "loss": 1.3013,
      "step": 68800
    },
    {
      "epoch": 2.1048451151707703,
      "grad_norm": 5.709287643432617,
      "learning_rate": 3.9541212825847514e-05,
      "loss": 1.2576,
      "step": 68900
    },
    {
      "epoch": 2.1079000427689865,
      "grad_norm": 3.3813297748565674,
      "learning_rate": 3.9525914820707386e-05,
      "loss": 1.2087,
      "step": 69000
    },
    {
      "epoch": 2.110954970367202,
      "grad_norm": 8.29941177368164,
      "learning_rate": 3.951061681556725e-05,
      "loss": 1.1075,
      "step": 69100
    },
    {
      "epoch": 2.1140098979654183,
      "grad_norm": 4.356234073638916,
      "learning_rate": 3.949531881042712e-05,
      "loss": 1.0665,
      "step": 69200
    },
    {
      "epoch": 2.117064825563634,
      "grad_norm": 2.651749849319458,
      "learning_rate": 3.948002080528699e-05,
      "loss": 1.1263,
      "step": 69300
    },
    {
      "epoch": 2.12011975316185,
      "grad_norm": 12.057615280151367,
      "learning_rate": 3.9464722800146863e-05,
      "loss": 1.1807,
      "step": 69400
    },
    {
      "epoch": 2.1231746807600658,
      "grad_norm": 3.4325954914093018,
      "learning_rate": 3.9449424795006735e-05,
      "loss": 1.1849,
      "step": 69500
    },
    {
      "epoch": 2.126229608358282,
      "grad_norm": 1.0763434171676636,
      "learning_rate": 3.9434126789866606e-05,
      "loss": 1.0558,
      "step": 69600
    },
    {
      "epoch": 2.129284535956498,
      "grad_norm": 10.428380012512207,
      "learning_rate": 3.941882878472648e-05,
      "loss": 1.2732,
      "step": 69700
    },
    {
      "epoch": 2.1323394635547137,
      "grad_norm": 1.0638651847839355,
      "learning_rate": 3.940353077958635e-05,
      "loss": 1.1089,
      "step": 69800
    },
    {
      "epoch": 2.13539439115293,
      "grad_norm": 3.8725669384002686,
      "learning_rate": 3.938823277444621e-05,
      "loss": 1.3897,
      "step": 69900
    },
    {
      "epoch": 2.1384493187511455,
      "grad_norm": 2.9482781887054443,
      "learning_rate": 3.9372934769306084e-05,
      "loss": 1.2474,
      "step": 70000
    },
    {
      "epoch": 2.1415042463493617,
      "grad_norm": 2.1782474517822266,
      "learning_rate": 3.9357636764165955e-05,
      "loss": 1.1152,
      "step": 70100
    },
    {
      "epoch": 2.1445591739475773,
      "grad_norm": 2.9231390953063965,
      "learning_rate": 3.9342338759025826e-05,
      "loss": 1.198,
      "step": 70200
    },
    {
      "epoch": 2.1476141015457935,
      "grad_norm": 6.574098587036133,
      "learning_rate": 3.93270407538857e-05,
      "loss": 1.0875,
      "step": 70300
    },
    {
      "epoch": 2.150669029144009,
      "grad_norm": 10.772805213928223,
      "learning_rate": 3.931174274874557e-05,
      "loss": 1.2619,
      "step": 70400
    },
    {
      "epoch": 2.1537239567422253,
      "grad_norm": 2.2910847663879395,
      "learning_rate": 3.929644474360544e-05,
      "loss": 1.217,
      "step": 70500
    },
    {
      "epoch": 2.156778884340441,
      "grad_norm": 3.5544190406799316,
      "learning_rate": 3.928114673846531e-05,
      "loss": 1.1816,
      "step": 70600
    },
    {
      "epoch": 2.159833811938657,
      "grad_norm": 11.924046516418457,
      "learning_rate": 3.9265848733325175e-05,
      "loss": 1.1116,
      "step": 70700
    },
    {
      "epoch": 2.1628887395368728,
      "grad_norm": 4.180816173553467,
      "learning_rate": 3.9250550728185046e-05,
      "loss": 1.0373,
      "step": 70800
    },
    {
      "epoch": 2.165943667135089,
      "grad_norm": 7.700500011444092,
      "learning_rate": 3.923525272304492e-05,
      "loss": 1.2157,
      "step": 70900
    },
    {
      "epoch": 2.1689985947333046,
      "grad_norm": 3.0716445446014404,
      "learning_rate": 3.921995471790479e-05,
      "loss": 1.1892,
      "step": 71000
    },
    {
      "epoch": 2.1720535223315207,
      "grad_norm": 6.259620189666748,
      "learning_rate": 3.920465671276466e-05,
      "loss": 1.1653,
      "step": 71100
    },
    {
      "epoch": 2.175108449929737,
      "grad_norm": 6.640773773193359,
      "learning_rate": 3.918935870762453e-05,
      "loss": 1.1338,
      "step": 71200
    },
    {
      "epoch": 2.1781633775279525,
      "grad_norm": 7.011377811431885,
      "learning_rate": 3.91740607024844e-05,
      "loss": 1.1523,
      "step": 71300
    },
    {
      "epoch": 2.1812183051261687,
      "grad_norm": 2.073110580444336,
      "learning_rate": 3.915876269734427e-05,
      "loss": 1.1036,
      "step": 71400
    },
    {
      "epoch": 2.1842732327243843,
      "grad_norm": 1.5513907670974731,
      "learning_rate": 3.914346469220414e-05,
      "loss": 1.1772,
      "step": 71500
    },
    {
      "epoch": 2.1873281603226005,
      "grad_norm": 4.17803430557251,
      "learning_rate": 3.912816668706401e-05,
      "loss": 1.1437,
      "step": 71600
    },
    {
      "epoch": 2.190383087920816,
      "grad_norm": 2.0113983154296875,
      "learning_rate": 3.911286868192388e-05,
      "loss": 1.0487,
      "step": 71700
    },
    {
      "epoch": 2.1934380155190323,
      "grad_norm": 0.9932565093040466,
      "learning_rate": 3.909757067678375e-05,
      "loss": 1.1172,
      "step": 71800
    },
    {
      "epoch": 2.196492943117248,
      "grad_norm": 1.959645390510559,
      "learning_rate": 3.908227267164362e-05,
      "loss": 1.1405,
      "step": 71900
    },
    {
      "epoch": 2.199547870715464,
      "grad_norm": 4.294275760650635,
      "learning_rate": 3.906697466650349e-05,
      "loss": 1.4798,
      "step": 72000
    },
    {
      "epoch": 2.2026027983136798,
      "grad_norm": 2.311213493347168,
      "learning_rate": 3.9051676661363364e-05,
      "loss": 1.1973,
      "step": 72100
    },
    {
      "epoch": 2.205657725911896,
      "grad_norm": 10.680132865905762,
      "learning_rate": 3.9036378656223236e-05,
      "loss": 1.0047,
      "step": 72200
    },
    {
      "epoch": 2.208712653510112,
      "grad_norm": 5.8984904289245605,
      "learning_rate": 3.90210806510831e-05,
      "loss": 1.3464,
      "step": 72300
    },
    {
      "epoch": 2.2117675811083277,
      "grad_norm": 1.8181579113006592,
      "learning_rate": 3.900578264594297e-05,
      "loss": 1.1493,
      "step": 72400
    },
    {
      "epoch": 2.214822508706544,
      "grad_norm": 2.351229190826416,
      "learning_rate": 3.899048464080284e-05,
      "loss": 1.0957,
      "step": 72500
    },
    {
      "epoch": 2.2178774363047595,
      "grad_norm": 3.8195061683654785,
      "learning_rate": 3.897518663566271e-05,
      "loss": 1.2018,
      "step": 72600
    },
    {
      "epoch": 2.2209323639029757,
      "grad_norm": 3.727231740951538,
      "learning_rate": 3.896004161057398e-05,
      "loss": 1.3295,
      "step": 72700
    },
    {
      "epoch": 2.2239872915011913,
      "grad_norm": 3.8231828212738037,
      "learning_rate": 3.894474360543385e-05,
      "loss": 1.0968,
      "step": 72800
    },
    {
      "epoch": 2.2270422190994075,
      "grad_norm": 6.009609699249268,
      "learning_rate": 3.8929445600293724e-05,
      "loss": 1.0188,
      "step": 72900
    },
    {
      "epoch": 2.230097146697623,
      "grad_norm": 17.05879783630371,
      "learning_rate": 3.8914147595153595e-05,
      "loss": 1.2487,
      "step": 73000
    },
    {
      "epoch": 2.2331520742958393,
      "grad_norm": 5.385801792144775,
      "learning_rate": 3.8898849590013466e-05,
      "loss": 1.1919,
      "step": 73100
    },
    {
      "epoch": 2.236207001894055,
      "grad_norm": 5.083590030670166,
      "learning_rate": 3.888355158487334e-05,
      "loss": 1.048,
      "step": 73200
    },
    {
      "epoch": 2.239261929492271,
      "grad_norm": 5.49558687210083,
      "learning_rate": 3.88682535797332e-05,
      "loss": 1.3023,
      "step": 73300
    },
    {
      "epoch": 2.242316857090487,
      "grad_norm": 5.167341709136963,
      "learning_rate": 3.885295557459307e-05,
      "loss": 1.242,
      "step": 73400
    },
    {
      "epoch": 2.245371784688703,
      "grad_norm": 7.800171852111816,
      "learning_rate": 3.8837657569452944e-05,
      "loss": 0.942,
      "step": 73500
    },
    {
      "epoch": 2.2484267122869186,
      "grad_norm": 4.706507682800293,
      "learning_rate": 3.8822359564312815e-05,
      "loss": 1.0041,
      "step": 73600
    },
    {
      "epoch": 2.2514816398851347,
      "grad_norm": 12.750622749328613,
      "learning_rate": 3.8807061559172686e-05,
      "loss": 1.0336,
      "step": 73700
    },
    {
      "epoch": 2.254536567483351,
      "grad_norm": 8.06450366973877,
      "learning_rate": 3.879176355403256e-05,
      "loss": 1.1878,
      "step": 73800
    },
    {
      "epoch": 2.2575914950815665,
      "grad_norm": 4.1706953048706055,
      "learning_rate": 3.877646554889243e-05,
      "loss": 1.2378,
      "step": 73900
    },
    {
      "epoch": 2.2606464226797827,
      "grad_norm": 9.237957954406738,
      "learning_rate": 3.87611675437523e-05,
      "loss": 1.316,
      "step": 74000
    },
    {
      "epoch": 2.2637013502779983,
      "grad_norm": 15.578306198120117,
      "learning_rate": 3.8745869538612164e-05,
      "loss": 1.1625,
      "step": 74100
    },
    {
      "epoch": 2.2667562778762145,
      "grad_norm": 4.735302925109863,
      "learning_rate": 3.8730571533472035e-05,
      "loss": 1.2111,
      "step": 74200
    },
    {
      "epoch": 2.26981120547443,
      "grad_norm": 3.8716487884521484,
      "learning_rate": 3.8715273528331906e-05,
      "loss": 1.2688,
      "step": 74300
    },
    {
      "epoch": 2.2728661330726463,
      "grad_norm": 1.822843074798584,
      "learning_rate": 3.869997552319178e-05,
      "loss": 1.182,
      "step": 74400
    },
    {
      "epoch": 2.275921060670862,
      "grad_norm": 4.336980819702148,
      "learning_rate": 3.868467751805165e-05,
      "loss": 1.1425,
      "step": 74500
    },
    {
      "epoch": 2.278975988269078,
      "grad_norm": 10.102789878845215,
      "learning_rate": 3.866937951291152e-05,
      "loss": 1.2088,
      "step": 74600
    },
    {
      "epoch": 2.282030915867294,
      "grad_norm": 1.985364317893982,
      "learning_rate": 3.865408150777139e-05,
      "loss": 1.1665,
      "step": 74700
    },
    {
      "epoch": 2.28508584346551,
      "grad_norm": 11.066972732543945,
      "learning_rate": 3.863893648268266e-05,
      "loss": 1.1125,
      "step": 74800
    },
    {
      "epoch": 2.2881407710637256,
      "grad_norm": 3.418748617172241,
      "learning_rate": 3.862363847754253e-05,
      "loss": 1.245,
      "step": 74900
    },
    {
      "epoch": 2.2911956986619417,
      "grad_norm": 3.346759557723999,
      "learning_rate": 3.86083404724024e-05,
      "loss": 1.025,
      "step": 75000
    },
    {
      "epoch": 2.2942506262601574,
      "grad_norm": 5.483870983123779,
      "learning_rate": 3.859304246726227e-05,
      "loss": 1.1594,
      "step": 75100
    },
    {
      "epoch": 2.2973055538583735,
      "grad_norm": 2.363171339035034,
      "learning_rate": 3.857774446212214e-05,
      "loss": 1.2255,
      "step": 75200
    },
    {
      "epoch": 2.3003604814565897,
      "grad_norm": 3.270732879638672,
      "learning_rate": 3.856244645698201e-05,
      "loss": 1.1561,
      "step": 75300
    },
    {
      "epoch": 2.3034154090548054,
      "grad_norm": 6.600897789001465,
      "learning_rate": 3.854714845184188e-05,
      "loss": 1.4451,
      "step": 75400
    },
    {
      "epoch": 2.3064703366530215,
      "grad_norm": 9.933273315429688,
      "learning_rate": 3.853185044670175e-05,
      "loss": 1.1166,
      "step": 75500
    },
    {
      "epoch": 2.309525264251237,
      "grad_norm": 9.13111400604248,
      "learning_rate": 3.851655244156162e-05,
      "loss": 1.2099,
      "step": 75600
    },
    {
      "epoch": 2.3125801918494533,
      "grad_norm": 0.8172520995140076,
      "learning_rate": 3.850125443642149e-05,
      "loss": 1.4097,
      "step": 75700
    },
    {
      "epoch": 2.315635119447669,
      "grad_norm": 2.1834285259246826,
      "learning_rate": 3.8485956431281364e-05,
      "loss": 1.1568,
      "step": 75800
    },
    {
      "epoch": 2.318690047045885,
      "grad_norm": 3.5525593757629395,
      "learning_rate": 3.8470658426141235e-05,
      "loss": 1.3119,
      "step": 75900
    },
    {
      "epoch": 2.321744974644101,
      "grad_norm": 5.114786624908447,
      "learning_rate": 3.84553604210011e-05,
      "loss": 1.0931,
      "step": 76000
    },
    {
      "epoch": 2.324799902242317,
      "grad_norm": 3.612886428833008,
      "learning_rate": 3.844006241586097e-05,
      "loss": 1.1166,
      "step": 76100
    },
    {
      "epoch": 2.3278548298405326,
      "grad_norm": 2.317949056625366,
      "learning_rate": 3.842476441072084e-05,
      "loss": 1.2106,
      "step": 76200
    },
    {
      "epoch": 2.3309097574387487,
      "grad_norm": 7.012603282928467,
      "learning_rate": 3.840946640558071e-05,
      "loss": 1.1442,
      "step": 76300
    },
    {
      "epoch": 2.333964685036965,
      "grad_norm": 4.5827717781066895,
      "learning_rate": 3.8394168400440584e-05,
      "loss": 1.161,
      "step": 76400
    },
    {
      "epoch": 2.3370196126351805,
      "grad_norm": 8.932015419006348,
      "learning_rate": 3.8378870395300455e-05,
      "loss": 0.9779,
      "step": 76500
    },
    {
      "epoch": 2.3400745402333962,
      "grad_norm": 4.946058750152588,
      "learning_rate": 3.8363572390160326e-05,
      "loss": 1.1326,
      "step": 76600
    },
    {
      "epoch": 2.3431294678316124,
      "grad_norm": 3.1590678691864014,
      "learning_rate": 3.83482743850202e-05,
      "loss": 1.328,
      "step": 76700
    },
    {
      "epoch": 2.3461843954298285,
      "grad_norm": 6.242021083831787,
      "learning_rate": 3.833297637988006e-05,
      "loss": 1.0906,
      "step": 76800
    },
    {
      "epoch": 2.349239323028044,
      "grad_norm": 6.75691556930542,
      "learning_rate": 3.8317831354791336e-05,
      "loss": 1.4751,
      "step": 76900
    },
    {
      "epoch": 2.3522942506262603,
      "grad_norm": 8.400306701660156,
      "learning_rate": 3.830253334965121e-05,
      "loss": 1.1472,
      "step": 77000
    },
    {
      "epoch": 2.355349178224476,
      "grad_norm": 14.731451034545898,
      "learning_rate": 3.828723534451108e-05,
      "loss": 1.2924,
      "step": 77100
    },
    {
      "epoch": 2.358404105822692,
      "grad_norm": 5.320265293121338,
      "learning_rate": 3.827193733937095e-05,
      "loss": 1.2347,
      "step": 77200
    },
    {
      "epoch": 2.361459033420908,
      "grad_norm": 2.3398170471191406,
      "learning_rate": 3.825663933423082e-05,
      "loss": 1.354,
      "step": 77300
    },
    {
      "epoch": 2.364513961019124,
      "grad_norm": 4.871092319488525,
      "learning_rate": 3.824134132909069e-05,
      "loss": 1.3481,
      "step": 77400
    },
    {
      "epoch": 2.3675688886173396,
      "grad_norm": 5.175891876220703,
      "learning_rate": 3.8226043323950563e-05,
      "loss": 1.2666,
      "step": 77500
    },
    {
      "epoch": 2.3706238162155557,
      "grad_norm": 4.1197686195373535,
      "learning_rate": 3.8210745318810435e-05,
      "loss": 1.1468,
      "step": 77600
    },
    {
      "epoch": 2.3736787438137714,
      "grad_norm": 7.0223164558410645,
      "learning_rate": 3.81954473136703e-05,
      "loss": 1.2952,
      "step": 77700
    },
    {
      "epoch": 2.3767336714119875,
      "grad_norm": 6.587507724761963,
      "learning_rate": 3.818014930853017e-05,
      "loss": 1.1041,
      "step": 77800
    },
    {
      "epoch": 2.3797885990102037,
      "grad_norm": 3.74049711227417,
      "learning_rate": 3.816485130339004e-05,
      "loss": 1.2098,
      "step": 77900
    },
    {
      "epoch": 2.3828435266084194,
      "grad_norm": 16.616947174072266,
      "learning_rate": 3.814955329824991e-05,
      "loss": 1.2535,
      "step": 78000
    },
    {
      "epoch": 2.3858984542066355,
      "grad_norm": 2.7336437702178955,
      "learning_rate": 3.8134255293109784e-05,
      "loss": 1.1514,
      "step": 78100
    },
    {
      "epoch": 2.388953381804851,
      "grad_norm": 4.99415397644043,
      "learning_rate": 3.8118957287969655e-05,
      "loss": 1.2196,
      "step": 78200
    },
    {
      "epoch": 2.3920083094030673,
      "grad_norm": 3.396183729171753,
      "learning_rate": 3.8103659282829526e-05,
      "loss": 1.1826,
      "step": 78300
    },
    {
      "epoch": 2.395063237001283,
      "grad_norm": 4.325381278991699,
      "learning_rate": 3.80883612776894e-05,
      "loss": 1.1332,
      "step": 78400
    },
    {
      "epoch": 2.398118164599499,
      "grad_norm": 4.070424556732178,
      "learning_rate": 3.807306327254926e-05,
      "loss": 1.3016,
      "step": 78500
    },
    {
      "epoch": 2.401173092197715,
      "grad_norm": 4.667051792144775,
      "learning_rate": 3.8057765267409126e-05,
      "loss": 1.1709,
      "step": 78600
    },
    {
      "epoch": 2.404228019795931,
      "grad_norm": 10.673643112182617,
      "learning_rate": 3.8042467262269e-05,
      "loss": 1.2628,
      "step": 78700
    },
    {
      "epoch": 2.4072829473941466,
      "grad_norm": 5.150026321411133,
      "learning_rate": 3.802716925712887e-05,
      "loss": 1.3025,
      "step": 78800
    },
    {
      "epoch": 2.4103378749923627,
      "grad_norm": 10.943438529968262,
      "learning_rate": 3.801187125198874e-05,
      "loss": 1.228,
      "step": 78900
    },
    {
      "epoch": 2.4133928025905784,
      "grad_norm": 5.951319217681885,
      "learning_rate": 3.7996726226900014e-05,
      "loss": 1.1702,
      "step": 79000
    },
    {
      "epoch": 2.4164477301887946,
      "grad_norm": 8.593432426452637,
      "learning_rate": 3.7981428221759885e-05,
      "loss": 1.3426,
      "step": 79100
    },
    {
      "epoch": 2.4195026577870102,
      "grad_norm": 4.801974773406982,
      "learning_rate": 3.7966130216619756e-05,
      "loss": 1.0271,
      "step": 79200
    },
    {
      "epoch": 2.4225575853852264,
      "grad_norm": 4.933838367462158,
      "learning_rate": 3.795083221147963e-05,
      "loss": 1.0458,
      "step": 79300
    },
    {
      "epoch": 2.4256125129834425,
      "grad_norm": 2.4156911373138428,
      "learning_rate": 3.79355342063395e-05,
      "loss": 1.2557,
      "step": 79400
    },
    {
      "epoch": 2.428667440581658,
      "grad_norm": 18.329809188842773,
      "learning_rate": 3.792023620119936e-05,
      "loss": 1.0828,
      "step": 79500
    },
    {
      "epoch": 2.4317223681798743,
      "grad_norm": 3.4031643867492676,
      "learning_rate": 3.7904938196059234e-05,
      "loss": 1.1404,
      "step": 79600
    },
    {
      "epoch": 2.43477729577809,
      "grad_norm": 29.743398666381836,
      "learning_rate": 3.7889640190919105e-05,
      "loss": 1.1868,
      "step": 79700
    },
    {
      "epoch": 2.437832223376306,
      "grad_norm": 3.8394060134887695,
      "learning_rate": 3.7874342185778976e-05,
      "loss": 1.1176,
      "step": 79800
    },
    {
      "epoch": 2.440887150974522,
      "grad_norm": 3.4328103065490723,
      "learning_rate": 3.785904418063885e-05,
      "loss": 1.3779,
      "step": 79900
    },
    {
      "epoch": 2.443942078572738,
      "grad_norm": 1.7140569686889648,
      "learning_rate": 3.784374617549872e-05,
      "loss": 1.0813,
      "step": 80000
    },
    {
      "epoch": 2.4469970061709536,
      "grad_norm": 11.085058212280273,
      "learning_rate": 3.782844817035859e-05,
      "loss": 1.1175,
      "step": 80100
    },
    {
      "epoch": 2.4500519337691697,
      "grad_norm": 7.249815464019775,
      "learning_rate": 3.781315016521846e-05,
      "loss": 1.1873,
      "step": 80200
    },
    {
      "epoch": 2.4531068613673854,
      "grad_norm": 10.735973358154297,
      "learning_rate": 3.7797852160078325e-05,
      "loss": 1.2008,
      "step": 80300
    },
    {
      "epoch": 2.4561617889656016,
      "grad_norm": 6.401257514953613,
      "learning_rate": 3.77825541549382e-05,
      "loss": 1.2233,
      "step": 80400
    },
    {
      "epoch": 2.4592167165638177,
      "grad_norm": 1.9271202087402344,
      "learning_rate": 3.776725614979807e-05,
      "loss": 1.2601,
      "step": 80500
    },
    {
      "epoch": 2.4622716441620334,
      "grad_norm": 2.7985684871673584,
      "learning_rate": 3.775195814465794e-05,
      "loss": 1.1866,
      "step": 80600
    },
    {
      "epoch": 2.465326571760249,
      "grad_norm": 5.0379509925842285,
      "learning_rate": 3.773666013951781e-05,
      "loss": 1.1215,
      "step": 80700
    },
    {
      "epoch": 2.468381499358465,
      "grad_norm": 19.3048152923584,
      "learning_rate": 3.772136213437768e-05,
      "loss": 1.1121,
      "step": 80800
    },
    {
      "epoch": 2.4714364269566813,
      "grad_norm": 5.526917457580566,
      "learning_rate": 3.770606412923755e-05,
      "loss": 1.023,
      "step": 80900
    },
    {
      "epoch": 2.474491354554897,
      "grad_norm": 3.810134172439575,
      "learning_rate": 3.7690766124097424e-05,
      "loss": 1.1027,
      "step": 81000
    },
    {
      "epoch": 2.477546282153113,
      "grad_norm": 4.790931224822998,
      "learning_rate": 3.767562109900869e-05,
      "loss": 1.229,
      "step": 81100
    },
    {
      "epoch": 2.480601209751329,
      "grad_norm": 4.748075008392334,
      "learning_rate": 3.766032309386856e-05,
      "loss": 1.1788,
      "step": 81200
    },
    {
      "epoch": 2.483656137349545,
      "grad_norm": 7.152939796447754,
      "learning_rate": 3.7645025088728434e-05,
      "loss": 0.9895,
      "step": 81300
    },
    {
      "epoch": 2.4867110649477606,
      "grad_norm": 9.12071418762207,
      "learning_rate": 3.76297270835883e-05,
      "loss": 1.4014,
      "step": 81400
    },
    {
      "epoch": 2.4897659925459767,
      "grad_norm": 2.651780605316162,
      "learning_rate": 3.761442907844817e-05,
      "loss": 1.2332,
      "step": 81500
    },
    {
      "epoch": 2.4928209201441924,
      "grad_norm": 1.2754027843475342,
      "learning_rate": 3.759913107330804e-05,
      "loss": 1.2788,
      "step": 81600
    },
    {
      "epoch": 2.4958758477424086,
      "grad_norm": 4.905016899108887,
      "learning_rate": 3.758383306816791e-05,
      "loss": 1.1859,
      "step": 81700
    },
    {
      "epoch": 2.4989307753406242,
      "grad_norm": 6.369762897491455,
      "learning_rate": 3.756853506302778e-05,
      "loss": 1.2471,
      "step": 81800
    },
    {
      "epoch": 2.5019857029388404,
      "grad_norm": 0.8389211297035217,
      "learning_rate": 3.7553237057887654e-05,
      "loss": 1.128,
      "step": 81900
    },
    {
      "epoch": 2.5050406305370565,
      "grad_norm": 6.334069728851318,
      "learning_rate": 3.7537939052747525e-05,
      "loss": 1.4241,
      "step": 82000
    },
    {
      "epoch": 2.508095558135272,
      "grad_norm": 4.841616153717041,
      "learning_rate": 3.7522641047607396e-05,
      "loss": 1.0423,
      "step": 82100
    },
    {
      "epoch": 2.511150485733488,
      "grad_norm": 6.821234226226807,
      "learning_rate": 3.750734304246726e-05,
      "loss": 1.1206,
      "step": 82200
    },
    {
      "epoch": 2.514205413331704,
      "grad_norm": 3.092060089111328,
      "learning_rate": 3.749204503732713e-05,
      "loss": 1.0781,
      "step": 82300
    },
    {
      "epoch": 2.51726034092992,
      "grad_norm": 4.442360877990723,
      "learning_rate": 3.7476747032187e-05,
      "loss": 1.3774,
      "step": 82400
    },
    {
      "epoch": 2.520315268528136,
      "grad_norm": 5.051080226898193,
      "learning_rate": 3.7461449027046874e-05,
      "loss": 1.2533,
      "step": 82500
    },
    {
      "epoch": 2.523370196126352,
      "grad_norm": 2.790591239929199,
      "learning_rate": 3.7446151021906745e-05,
      "loss": 1.2949,
      "step": 82600
    },
    {
      "epoch": 2.5264251237245676,
      "grad_norm": 4.006244659423828,
      "learning_rate": 3.7430853016766617e-05,
      "loss": 1.1801,
      "step": 82700
    },
    {
      "epoch": 2.5294800513227838,
      "grad_norm": 4.384695053100586,
      "learning_rate": 3.741555501162649e-05,
      "loss": 1.1284,
      "step": 82800
    },
    {
      "epoch": 2.5325349789209994,
      "grad_norm": 6.040104866027832,
      "learning_rate": 3.740025700648636e-05,
      "loss": 1.3011,
      "step": 82900
    },
    {
      "epoch": 2.5355899065192156,
      "grad_norm": 14.280811309814453,
      "learning_rate": 3.738495900134622e-05,
      "loss": 1.1675,
      "step": 83000
    },
    {
      "epoch": 2.5386448341174317,
      "grad_norm": 5.165515899658203,
      "learning_rate": 3.73698139762575e-05,
      "loss": 1.0729,
      "step": 83100
    },
    {
      "epoch": 2.5416997617156474,
      "grad_norm": 16.37873077392578,
      "learning_rate": 3.735451597111736e-05,
      "loss": 1.1082,
      "step": 83200
    },
    {
      "epoch": 2.544754689313863,
      "grad_norm": 3.9798662662506104,
      "learning_rate": 3.7339217965977233e-05,
      "loss": 1.1677,
      "step": 83300
    },
    {
      "epoch": 2.547809616912079,
      "grad_norm": 8.16595458984375,
      "learning_rate": 3.7323919960837105e-05,
      "loss": 1.1792,
      "step": 83400
    },
    {
      "epoch": 2.5508645445102953,
      "grad_norm": 1.22407865524292,
      "learning_rate": 3.7308621955696976e-05,
      "loss": 1.1477,
      "step": 83500
    },
    {
      "epoch": 2.553919472108511,
      "grad_norm": 5.23829984664917,
      "learning_rate": 3.729332395055685e-05,
      "loss": 1.1713,
      "step": 83600
    },
    {
      "epoch": 2.5569743997067267,
      "grad_norm": 2.0325045585632324,
      "learning_rate": 3.727802594541672e-05,
      "loss": 1.2988,
      "step": 83700
    },
    {
      "epoch": 2.560029327304943,
      "grad_norm": 5.054375648498535,
      "learning_rate": 3.726272794027659e-05,
      "loss": 1.2534,
      "step": 83800
    },
    {
      "epoch": 2.563084254903159,
      "grad_norm": 5.4261155128479,
      "learning_rate": 3.724742993513646e-05,
      "loss": 1.1863,
      "step": 83900
    },
    {
      "epoch": 2.5661391825013746,
      "grad_norm": 4.739152908325195,
      "learning_rate": 3.7232131929996325e-05,
      "loss": 1.1728,
      "step": 84000
    },
    {
      "epoch": 2.5691941100995908,
      "grad_norm": 4.027005672454834,
      "learning_rate": 3.7216833924856196e-05,
      "loss": 1.1391,
      "step": 84100
    },
    {
      "epoch": 2.5722490376978064,
      "grad_norm": 2.6760764122009277,
      "learning_rate": 3.720153591971607e-05,
      "loss": 1.208,
      "step": 84200
    },
    {
      "epoch": 2.5753039652960226,
      "grad_norm": 7.8956217765808105,
      "learning_rate": 3.718623791457594e-05,
      "loss": 1.1109,
      "step": 84300
    },
    {
      "epoch": 2.5783588928942383,
      "grad_norm": 5.012131690979004,
      "learning_rate": 3.717093990943581e-05,
      "loss": 1.2318,
      "step": 84400
    },
    {
      "epoch": 2.5814138204924544,
      "grad_norm": 7.069117069244385,
      "learning_rate": 3.715564190429568e-05,
      "loss": 1.44,
      "step": 84500
    },
    {
      "epoch": 2.5844687480906705,
      "grad_norm": 7.637176990509033,
      "learning_rate": 3.714034389915555e-05,
      "loss": 1.18,
      "step": 84600
    },
    {
      "epoch": 2.587523675688886,
      "grad_norm": 5.741633415222168,
      "learning_rate": 3.712504589401542e-05,
      "loss": 1.2804,
      "step": 84700
    },
    {
      "epoch": 2.590578603287102,
      "grad_norm": 7.103206157684326,
      "learning_rate": 3.710974788887529e-05,
      "loss": 1.3692,
      "step": 84800
    },
    {
      "epoch": 2.593633530885318,
      "grad_norm": 4.079885959625244,
      "learning_rate": 3.709444988373516e-05,
      "loss": 1.276,
      "step": 84900
    },
    {
      "epoch": 2.596688458483534,
      "grad_norm": 5.2246479988098145,
      "learning_rate": 3.707915187859503e-05,
      "loss": 1.2933,
      "step": 85000
    },
    {
      "epoch": 2.59974338608175,
      "grad_norm": 4.833429336547852,
      "learning_rate": 3.70638538734549e-05,
      "loss": 1.3188,
      "step": 85100
    },
    {
      "epoch": 2.602798313679966,
      "grad_norm": 11.477587699890137,
      "learning_rate": 3.7048708848366176e-05,
      "loss": 1.1811,
      "step": 85200
    },
    {
      "epoch": 2.6058532412781816,
      "grad_norm": 5.056561470031738,
      "learning_rate": 3.703341084322605e-05,
      "loss": 1.124,
      "step": 85300
    },
    {
      "epoch": 2.6089081688763978,
      "grad_norm": 2.5134527683258057,
      "learning_rate": 3.701811283808592e-05,
      "loss": 1.1202,
      "step": 85400
    },
    {
      "epoch": 2.6119630964746134,
      "grad_norm": 2.793060779571533,
      "learning_rate": 3.700281483294579e-05,
      "loss": 1.1034,
      "step": 85500
    },
    {
      "epoch": 2.6150180240728296,
      "grad_norm": 4.571871757507324,
      "learning_rate": 3.698751682780566e-05,
      "loss": 1.0702,
      "step": 85600
    },
    {
      "epoch": 2.6180729516710453,
      "grad_norm": 3.5763347148895264,
      "learning_rate": 3.6972218822665525e-05,
      "loss": 1.2301,
      "step": 85700
    },
    {
      "epoch": 2.6211278792692614,
      "grad_norm": 11.870597839355469,
      "learning_rate": 3.6956920817525396e-05,
      "loss": 1.3386,
      "step": 85800
    },
    {
      "epoch": 2.624182806867477,
      "grad_norm": 7.040190696716309,
      "learning_rate": 3.694162281238527e-05,
      "loss": 1.1226,
      "step": 85900
    },
    {
      "epoch": 2.627237734465693,
      "grad_norm": 3.1670444011688232,
      "learning_rate": 3.692632480724514e-05,
      "loss": 1.0523,
      "step": 86000
    },
    {
      "epoch": 2.6302926620639093,
      "grad_norm": 0.7648332715034485,
      "learning_rate": 3.691102680210501e-05,
      "loss": 1.1571,
      "step": 86100
    },
    {
      "epoch": 2.633347589662125,
      "grad_norm": 27.462358474731445,
      "learning_rate": 3.689572879696488e-05,
      "loss": 1.3515,
      "step": 86200
    },
    {
      "epoch": 2.6364025172603407,
      "grad_norm": 3.599266529083252,
      "learning_rate": 3.688043079182475e-05,
      "loss": 1.2769,
      "step": 86300
    },
    {
      "epoch": 2.639457444858557,
      "grad_norm": 4.287918567657471,
      "learning_rate": 3.686513278668462e-05,
      "loss": 1.1693,
      "step": 86400
    },
    {
      "epoch": 2.642512372456773,
      "grad_norm": 10.636252403259277,
      "learning_rate": 3.684983478154449e-05,
      "loss": 1.2402,
      "step": 86500
    },
    {
      "epoch": 2.6455673000549886,
      "grad_norm": 6.561535358428955,
      "learning_rate": 3.683453677640436e-05,
      "loss": 1.1243,
      "step": 86600
    },
    {
      "epoch": 2.6486222276532048,
      "grad_norm": 3.99881649017334,
      "learning_rate": 3.681923877126423e-05,
      "loss": 1.1089,
      "step": 86700
    },
    {
      "epoch": 2.6516771552514204,
      "grad_norm": 5.040919780731201,
      "learning_rate": 3.68039407661241e-05,
      "loss": 1.1632,
      "step": 86800
    },
    {
      "epoch": 2.6547320828496366,
      "grad_norm": 1.5201897621154785,
      "learning_rate": 3.678864276098397e-05,
      "loss": 1.1074,
      "step": 86900
    },
    {
      "epoch": 2.6577870104478523,
      "grad_norm": 3.217228412628174,
      "learning_rate": 3.677334475584384e-05,
      "loss": 1.1696,
      "step": 87000
    },
    {
      "epoch": 2.6608419380460684,
      "grad_norm": 2.4591336250305176,
      "learning_rate": 3.6758046750703714e-05,
      "loss": 1.3259,
      "step": 87100
    },
    {
      "epoch": 2.663896865644284,
      "grad_norm": 15.164278030395508,
      "learning_rate": 3.674290172561498e-05,
      "loss": 1.1429,
      "step": 87200
    },
    {
      "epoch": 2.6669517932425,
      "grad_norm": 5.4548845291137695,
      "learning_rate": 3.672760372047485e-05,
      "loss": 1.4199,
      "step": 87300
    },
    {
      "epoch": 2.670006720840716,
      "grad_norm": 7.886258602142334,
      "learning_rate": 3.6712305715334724e-05,
      "loss": 1.1499,
      "step": 87400
    },
    {
      "epoch": 2.673061648438932,
      "grad_norm": 7.226693630218506,
      "learning_rate": 3.669700771019459e-05,
      "loss": 1.1624,
      "step": 87500
    },
    {
      "epoch": 2.676116576037148,
      "grad_norm": 9.446024894714355,
      "learning_rate": 3.668170970505446e-05,
      "loss": 1.1044,
      "step": 87600
    },
    {
      "epoch": 2.679171503635364,
      "grad_norm": 2.4468486309051514,
      "learning_rate": 3.666641169991433e-05,
      "loss": 1.1352,
      "step": 87700
    },
    {
      "epoch": 2.6822264312335795,
      "grad_norm": 3.7844583988189697,
      "learning_rate": 3.66511136947742e-05,
      "loss": 1.3393,
      "step": 87800
    },
    {
      "epoch": 2.6852813588317956,
      "grad_norm": 3.6775496006011963,
      "learning_rate": 3.663581568963407e-05,
      "loss": 1.1504,
      "step": 87900
    },
    {
      "epoch": 2.6883362864300118,
      "grad_norm": 6.129753112792969,
      "learning_rate": 3.6620517684493944e-05,
      "loss": 1.0526,
      "step": 88000
    },
    {
      "epoch": 2.6913912140282275,
      "grad_norm": 11.557202339172363,
      "learning_rate": 3.6605219679353816e-05,
      "loss": 1.3494,
      "step": 88100
    },
    {
      "epoch": 2.6944461416264436,
      "grad_norm": 9.309710502624512,
      "learning_rate": 3.658992167421369e-05,
      "loss": 1.2957,
      "step": 88200
    },
    {
      "epoch": 2.6975010692246593,
      "grad_norm": 4.096008777618408,
      "learning_rate": 3.657462366907355e-05,
      "loss": 1.3352,
      "step": 88300
    },
    {
      "epoch": 2.7005559968228754,
      "grad_norm": 8.638252258300781,
      "learning_rate": 3.655932566393342e-05,
      "loss": 1.1125,
      "step": 88400
    },
    {
      "epoch": 2.703610924421091,
      "grad_norm": 3.886626958847046,
      "learning_rate": 3.654402765879329e-05,
      "loss": 0.9983,
      "step": 88500
    },
    {
      "epoch": 2.706665852019307,
      "grad_norm": 0.3538905382156372,
      "learning_rate": 3.6528729653653165e-05,
      "loss": 1.304,
      "step": 88600
    },
    {
      "epoch": 2.7097207796175233,
      "grad_norm": 5.2617926597595215,
      "learning_rate": 3.6513431648513036e-05,
      "loss": 1.1223,
      "step": 88700
    },
    {
      "epoch": 2.712775707215739,
      "grad_norm": 15.284337997436523,
      "learning_rate": 3.649813364337291e-05,
      "loss": 1.1987,
      "step": 88800
    },
    {
      "epoch": 2.7158306348139547,
      "grad_norm": 5.358856201171875,
      "learning_rate": 3.648283563823278e-05,
      "loss": 1.3171,
      "step": 88900
    },
    {
      "epoch": 2.718885562412171,
      "grad_norm": 3.3485710620880127,
      "learning_rate": 3.646753763309265e-05,
      "loss": 1.19,
      "step": 89000
    },
    {
      "epoch": 2.721940490010387,
      "grad_norm": 3.345658302307129,
      "learning_rate": 3.6452239627952514e-05,
      "loss": 1.2676,
      "step": 89100
    },
    {
      "epoch": 2.7249954176086026,
      "grad_norm": 9.497928619384766,
      "learning_rate": 3.6436941622812385e-05,
      "loss": 1.1102,
      "step": 89200
    },
    {
      "epoch": 2.7280503452068183,
      "grad_norm": 7.673614978790283,
      "learning_rate": 3.6421643617672256e-05,
      "loss": 0.9989,
      "step": 89300
    },
    {
      "epoch": 2.7311052728050345,
      "grad_norm": 3.095411777496338,
      "learning_rate": 3.6406498592583524e-05,
      "loss": 1.1854,
      "step": 89400
    },
    {
      "epoch": 2.7341602004032506,
      "grad_norm": 11.466713905334473,
      "learning_rate": 3.6391200587443395e-05,
      "loss": 1.5457,
      "step": 89500
    },
    {
      "epoch": 2.7372151280014663,
      "grad_norm": 11.617925643920898,
      "learning_rate": 3.6375902582303266e-05,
      "loss": 1.115,
      "step": 89600
    },
    {
      "epoch": 2.7402700555996824,
      "grad_norm": 7.993494987487793,
      "learning_rate": 3.636060457716314e-05,
      "loss": 1.3089,
      "step": 89700
    },
    {
      "epoch": 2.743324983197898,
      "grad_norm": 4.979888916015625,
      "learning_rate": 3.634530657202301e-05,
      "loss": 1.142,
      "step": 89800
    },
    {
      "epoch": 2.746379910796114,
      "grad_norm": 6.180172920227051,
      "learning_rate": 3.633000856688288e-05,
      "loss": 1.311,
      "step": 89900
    },
    {
      "epoch": 2.74943483839433,
      "grad_norm": 2.8136937618255615,
      "learning_rate": 3.631471056174275e-05,
      "loss": 1.1723,
      "step": 90000
    },
    {
      "epoch": 2.752489765992546,
      "grad_norm": 12.260469436645508,
      "learning_rate": 3.629941255660262e-05,
      "loss": 1.4877,
      "step": 90100
    },
    {
      "epoch": 2.755544693590762,
      "grad_norm": 8.104580879211426,
      "learning_rate": 3.6284114551462486e-05,
      "loss": 1.0212,
      "step": 90200
    },
    {
      "epoch": 2.758599621188978,
      "grad_norm": 4.810283660888672,
      "learning_rate": 3.626881654632236e-05,
      "loss": 1.1489,
      "step": 90300
    },
    {
      "epoch": 2.7616545487871935,
      "grad_norm": 4.126216888427734,
      "learning_rate": 3.625351854118223e-05,
      "loss": 1.107,
      "step": 90400
    },
    {
      "epoch": 2.7647094763854096,
      "grad_norm": 4.295004367828369,
      "learning_rate": 3.62382205360421e-05,
      "loss": 1.0719,
      "step": 90500
    },
    {
      "epoch": 2.7677644039836258,
      "grad_norm": 7.011680603027344,
      "learning_rate": 3.622292253090197e-05,
      "loss": 1.2707,
      "step": 90600
    },
    {
      "epoch": 2.7708193315818415,
      "grad_norm": 2.7568280696868896,
      "learning_rate": 3.620762452576184e-05,
      "loss": 1.1977,
      "step": 90700
    },
    {
      "epoch": 2.7738742591800576,
      "grad_norm": 4.980713367462158,
      "learning_rate": 3.619232652062171e-05,
      "loss": 1.1931,
      "step": 90800
    },
    {
      "epoch": 2.7769291867782733,
      "grad_norm": 4.1475067138671875,
      "learning_rate": 3.6177028515481584e-05,
      "loss": 1.4218,
      "step": 90900
    },
    {
      "epoch": 2.7799841143764894,
      "grad_norm": 17.66010856628418,
      "learning_rate": 3.616173051034145e-05,
      "loss": 1.0312,
      "step": 91000
    },
    {
      "epoch": 2.783039041974705,
      "grad_norm": 11.250906944274902,
      "learning_rate": 3.614643250520132e-05,
      "loss": 1.4046,
      "step": 91100
    },
    {
      "epoch": 2.786093969572921,
      "grad_norm": 3.021059513092041,
      "learning_rate": 3.613113450006119e-05,
      "loss": 1.2493,
      "step": 91200
    },
    {
      "epoch": 2.789148897171137,
      "grad_norm": 3.5739548206329346,
      "learning_rate": 3.611583649492106e-05,
      "loss": 1.0393,
      "step": 91300
    },
    {
      "epoch": 2.792203824769353,
      "grad_norm": 3.9998271465301514,
      "learning_rate": 3.6100538489780933e-05,
      "loss": 1.088,
      "step": 91400
    },
    {
      "epoch": 2.7952587523675687,
      "grad_norm": 7.070734977722168,
      "learning_rate": 3.608539346469221e-05,
      "loss": 1.1436,
      "step": 91500
    },
    {
      "epoch": 2.798313679965785,
      "grad_norm": 4.2065839767456055,
      "learning_rate": 3.607009545955208e-05,
      "loss": 1.0456,
      "step": 91600
    },
    {
      "epoch": 2.801368607564001,
      "grad_norm": 8.773880958557129,
      "learning_rate": 3.605479745441195e-05,
      "loss": 1.184,
      "step": 91700
    },
    {
      "epoch": 2.8044235351622167,
      "grad_norm": 6.073965549468994,
      "learning_rate": 3.603949944927182e-05,
      "loss": 1.1699,
      "step": 91800
    },
    {
      "epoch": 2.8074784627604323,
      "grad_norm": 4.126655101776123,
      "learning_rate": 3.6024201444131686e-05,
      "loss": 1.052,
      "step": 91900
    },
    {
      "epoch": 2.8105333903586485,
      "grad_norm": 2.200648784637451,
      "learning_rate": 3.600890343899156e-05,
      "loss": 1.3698,
      "step": 92000
    },
    {
      "epoch": 2.8135883179568646,
      "grad_norm": 5.055072784423828,
      "learning_rate": 3.599360543385143e-05,
      "loss": 1.0363,
      "step": 92100
    },
    {
      "epoch": 2.8166432455550803,
      "grad_norm": 3.9162421226501465,
      "learning_rate": 3.59783074287113e-05,
      "loss": 1.3419,
      "step": 92200
    },
    {
      "epoch": 2.8196981731532964,
      "grad_norm": 4.10654878616333,
      "learning_rate": 3.596300942357117e-05,
      "loss": 1.2493,
      "step": 92300
    },
    {
      "epoch": 2.822753100751512,
      "grad_norm": 5.665062427520752,
      "learning_rate": 3.594771141843104e-05,
      "loss": 0.998,
      "step": 92400
    },
    {
      "epoch": 2.825808028349728,
      "grad_norm": 4.1935577392578125,
      "learning_rate": 3.593241341329091e-05,
      "loss": 1.1017,
      "step": 92500
    },
    {
      "epoch": 2.828862955947944,
      "grad_norm": 11.51850414276123,
      "learning_rate": 3.5917115408150784e-05,
      "loss": 1.2248,
      "step": 92600
    },
    {
      "epoch": 2.83191788354616,
      "grad_norm": 11.625690460205078,
      "learning_rate": 3.590181740301065e-05,
      "loss": 1.1488,
      "step": 92700
    },
    {
      "epoch": 2.8349728111443757,
      "grad_norm": 3.1991724967956543,
      "learning_rate": 3.588651939787052e-05,
      "loss": 1.1318,
      "step": 92800
    },
    {
      "epoch": 2.838027738742592,
      "grad_norm": 1.6752854585647583,
      "learning_rate": 3.587122139273039e-05,
      "loss": 1.0442,
      "step": 92900
    },
    {
      "epoch": 2.8410826663408075,
      "grad_norm": 5.884594917297363,
      "learning_rate": 3.585592338759026e-05,
      "loss": 1.0908,
      "step": 93000
    },
    {
      "epoch": 2.8441375939390237,
      "grad_norm": 10.092119216918945,
      "learning_rate": 3.584062538245013e-05,
      "loss": 1.1826,
      "step": 93100
    },
    {
      "epoch": 2.84719252153724,
      "grad_norm": 2.494101047515869,
      "learning_rate": 3.5825327377310004e-05,
      "loss": 1.2812,
      "step": 93200
    },
    {
      "epoch": 2.8502474491354555,
      "grad_norm": 4.2990641593933105,
      "learning_rate": 3.5810029372169875e-05,
      "loss": 1.1301,
      "step": 93300
    },
    {
      "epoch": 2.853302376733671,
      "grad_norm": 1.3677186965942383,
      "learning_rate": 3.5794731367029747e-05,
      "loss": 1.3642,
      "step": 93400
    },
    {
      "epoch": 2.8563573043318873,
      "grad_norm": 2.856877565383911,
      "learning_rate": 3.5779586341941015e-05,
      "loss": 1.1834,
      "step": 93500
    },
    {
      "epoch": 2.8594122319301034,
      "grad_norm": 5.638931751251221,
      "learning_rate": 3.5764288336800886e-05,
      "loss": 1.1668,
      "step": 93600
    },
    {
      "epoch": 2.862467159528319,
      "grad_norm": 3.2125723361968994,
      "learning_rate": 3.574899033166075e-05,
      "loss": 1.1466,
      "step": 93700
    },
    {
      "epoch": 2.8655220871265352,
      "grad_norm": 4.384485721588135,
      "learning_rate": 3.573369232652062e-05,
      "loss": 1.4209,
      "step": 93800
    },
    {
      "epoch": 2.868577014724751,
      "grad_norm": 4.23681640625,
      "learning_rate": 3.571839432138049e-05,
      "loss": 1.0677,
      "step": 93900
    },
    {
      "epoch": 2.871631942322967,
      "grad_norm": 5.866428375244141,
      "learning_rate": 3.5703096316240364e-05,
      "loss": 1.1571,
      "step": 94000
    },
    {
      "epoch": 2.8746868699211827,
      "grad_norm": 5.092837333679199,
      "learning_rate": 3.5687798311100235e-05,
      "loss": 1.3121,
      "step": 94100
    },
    {
      "epoch": 2.877741797519399,
      "grad_norm": 1.3914189338684082,
      "learning_rate": 3.56726532860115e-05,
      "loss": 1.1406,
      "step": 94200
    },
    {
      "epoch": 2.880796725117615,
      "grad_norm": 1.6459842920303345,
      "learning_rate": 3.5657355280871374e-05,
      "loss": 1.0341,
      "step": 94300
    },
    {
      "epoch": 2.8838516527158307,
      "grad_norm": 6.084038734436035,
      "learning_rate": 3.5642057275731245e-05,
      "loss": 1.0185,
      "step": 94400
    },
    {
      "epoch": 2.8869065803140463,
      "grad_norm": 2.1041250228881836,
      "learning_rate": 3.5626759270591116e-05,
      "loss": 1.13,
      "step": 94500
    },
    {
      "epoch": 2.8899615079122625,
      "grad_norm": 0.017624640837311745,
      "learning_rate": 3.561146126545099e-05,
      "loss": 1.1235,
      "step": 94600
    },
    {
      "epoch": 2.8930164355104786,
      "grad_norm": 2.9495761394500732,
      "learning_rate": 3.559616326031086e-05,
      "loss": 1.2972,
      "step": 94700
    },
    {
      "epoch": 2.8960713631086943,
      "grad_norm": 0.7582499384880066,
      "learning_rate": 3.558086525517072e-05,
      "loss": 1.2099,
      "step": 94800
    },
    {
      "epoch": 2.89912629070691,
      "grad_norm": 5.334167003631592,
      "learning_rate": 3.5565567250030594e-05,
      "loss": 1.2623,
      "step": 94900
    },
    {
      "epoch": 2.902181218305126,
      "grad_norm": 17.52392578125,
      "learning_rate": 3.5550269244890465e-05,
      "loss": 1.2304,
      "step": 95000
    },
    {
      "epoch": 2.9052361459033422,
      "grad_norm": 2.240028142929077,
      "learning_rate": 3.5534971239750336e-05,
      "loss": 1.1974,
      "step": 95100
    },
    {
      "epoch": 2.908291073501558,
      "grad_norm": 3.9233059883117676,
      "learning_rate": 3.551967323461021e-05,
      "loss": 1.0685,
      "step": 95200
    },
    {
      "epoch": 2.911346001099774,
      "grad_norm": 3.115082263946533,
      "learning_rate": 3.550437522947008e-05,
      "loss": 0.9877,
      "step": 95300
    },
    {
      "epoch": 2.9144009286979897,
      "grad_norm": 6.149535655975342,
      "learning_rate": 3.548907722432995e-05,
      "loss": 1.1253,
      "step": 95400
    },
    {
      "epoch": 2.917455856296206,
      "grad_norm": 6.213714122772217,
      "learning_rate": 3.547377921918982e-05,
      "loss": 1.256,
      "step": 95500
    },
    {
      "epoch": 2.9205107838944215,
      "grad_norm": 10.935718536376953,
      "learning_rate": 3.5458481214049685e-05,
      "loss": 1.1948,
      "step": 95600
    },
    {
      "epoch": 2.9235657114926377,
      "grad_norm": 3.0346519947052,
      "learning_rate": 3.5443183208909556e-05,
      "loss": 1.1022,
      "step": 95700
    },
    {
      "epoch": 2.926620639090854,
      "grad_norm": 8.563993453979492,
      "learning_rate": 3.542788520376943e-05,
      "loss": 1.0593,
      "step": 95800
    },
    {
      "epoch": 2.9296755666890695,
      "grad_norm": 1.6278303861618042,
      "learning_rate": 3.54125871986293e-05,
      "loss": 1.2639,
      "step": 95900
    },
    {
      "epoch": 2.932730494287285,
      "grad_norm": 1.3326663970947266,
      "learning_rate": 3.539728919348917e-05,
      "loss": 1.1684,
      "step": 96000
    },
    {
      "epoch": 2.9357854218855013,
      "grad_norm": 8.367751121520996,
      "learning_rate": 3.538199118834904e-05,
      "loss": 1.2524,
      "step": 96100
    },
    {
      "epoch": 2.9388403494837174,
      "grad_norm": 7.878076553344727,
      "learning_rate": 3.536669318320891e-05,
      "loss": 1.0507,
      "step": 96200
    },
    {
      "epoch": 2.941895277081933,
      "grad_norm": 3.120497703552246,
      "learning_rate": 3.5351395178068783e-05,
      "loss": 1.0112,
      "step": 96300
    },
    {
      "epoch": 2.9449502046801492,
      "grad_norm": 9.512770652770996,
      "learning_rate": 3.533609717292865e-05,
      "loss": 1.2574,
      "step": 96400
    },
    {
      "epoch": 2.948005132278365,
      "grad_norm": 10.365994453430176,
      "learning_rate": 3.532079916778852e-05,
      "loss": 1.0112,
      "step": 96500
    },
    {
      "epoch": 2.951060059876581,
      "grad_norm": 13.449652671813965,
      "learning_rate": 3.530550116264839e-05,
      "loss": 1.3926,
      "step": 96600
    },
    {
      "epoch": 2.9541149874747967,
      "grad_norm": 8.135039329528809,
      "learning_rate": 3.529020315750826e-05,
      "loss": 1.1585,
      "step": 96700
    },
    {
      "epoch": 2.957169915073013,
      "grad_norm": 4.279914855957031,
      "learning_rate": 3.527490515236813e-05,
      "loss": 1.1064,
      "step": 96800
    },
    {
      "epoch": 2.9602248426712285,
      "grad_norm": 2.8328933715820312,
      "learning_rate": 3.5259607147228004e-05,
      "loss": 1.0075,
      "step": 96900
    },
    {
      "epoch": 2.9632797702694447,
      "grad_norm": 5.973811149597168,
      "learning_rate": 3.5244309142087875e-05,
      "loss": 1.0137,
      "step": 97000
    },
    {
      "epoch": 2.9663346978676604,
      "grad_norm": 5.440450668334961,
      "learning_rate": 3.5229011136947746e-05,
      "loss": 0.983,
      "step": 97100
    },
    {
      "epoch": 2.9693896254658765,
      "grad_norm": 7.617842197418213,
      "learning_rate": 3.521371313180761e-05,
      "loss": 1.096,
      "step": 97200
    },
    {
      "epoch": 2.9724445530640926,
      "grad_norm": 2.359002113342285,
      "learning_rate": 3.519841512666748e-05,
      "loss": 1.425,
      "step": 97300
    },
    {
      "epoch": 2.9754994806623083,
      "grad_norm": 7.6081156730651855,
      "learning_rate": 3.518311712152735e-05,
      "loss": 1.3298,
      "step": 97400
    },
    {
      "epoch": 2.978554408260524,
      "grad_norm": 14.962251663208008,
      "learning_rate": 3.5167819116387224e-05,
      "loss": 1.3354,
      "step": 97500
    },
    {
      "epoch": 2.98160933585874,
      "grad_norm": 6.850179672241211,
      "learning_rate": 3.5152521111247095e-05,
      "loss": 1.2817,
      "step": 97600
    },
    {
      "epoch": 2.9846642634569562,
      "grad_norm": 4.044615745544434,
      "learning_rate": 3.5137223106106966e-05,
      "loss": 1.1176,
      "step": 97700
    },
    {
      "epoch": 2.987719191055172,
      "grad_norm": 7.540977478027344,
      "learning_rate": 3.512192510096684e-05,
      "loss": 1.2165,
      "step": 97800
    },
    {
      "epoch": 2.990774118653388,
      "grad_norm": 10.70987319946289,
      "learning_rate": 3.510662709582671e-05,
      "loss": 1.0701,
      "step": 97900
    },
    {
      "epoch": 2.9938290462516037,
      "grad_norm": 2.2019271850585938,
      "learning_rate": 3.509148207073798e-05,
      "loss": 1.3646,
      "step": 98000
    },
    {
      "epoch": 2.99688397384982,
      "grad_norm": 5.027231693267822,
      "learning_rate": 3.507618406559785e-05,
      "loss": 1.3256,
      "step": 98100
    },
    {
      "epoch": 2.9999389014480355,
      "grad_norm": 11.271944999694824,
      "learning_rate": 3.506088606045772e-05,
      "loss": 1.198,
      "step": 98200
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7097207796175231,
      "eval_loss": 0.7809557318687439,
      "eval_runtime": 1782.749,
      "eval_samples_per_second": 18.362,
      "eval_steps_per_second": 4.591,
      "step": 98202
    },
    {
      "epoch": 3.0029938290462517,
      "grad_norm": 4.658991813659668,
      "learning_rate": 3.504558805531759e-05,
      "loss": 1.1387,
      "step": 98300
    },
    {
      "epoch": 3.0060487566444674,
      "grad_norm": 9.121160507202148,
      "learning_rate": 3.503029005017746e-05,
      "loss": 1.2615,
      "step": 98400
    },
    {
      "epoch": 3.0091036842426835,
      "grad_norm": 10.340275764465332,
      "learning_rate": 3.501499204503733e-05,
      "loss": 0.9685,
      "step": 98500
    },
    {
      "epoch": 3.012158611840899,
      "grad_norm": 6.134807109832764,
      "learning_rate": 3.49996940398972e-05,
      "loss": 1.3476,
      "step": 98600
    },
    {
      "epoch": 3.0152135394391153,
      "grad_norm": 6.84115743637085,
      "learning_rate": 3.4984396034757074e-05,
      "loss": 1.1311,
      "step": 98700
    },
    {
      "epoch": 3.0182684670373314,
      "grad_norm": 9.961685180664062,
      "learning_rate": 3.4969098029616946e-05,
      "loss": 1.0183,
      "step": 98800
    },
    {
      "epoch": 3.021323394635547,
      "grad_norm": 10.408890724182129,
      "learning_rate": 3.495380002447681e-05,
      "loss": 1.1134,
      "step": 98900
    },
    {
      "epoch": 3.0243783222337632,
      "grad_norm": 5.6719183921813965,
      "learning_rate": 3.4938502019336674e-05,
      "loss": 1.2485,
      "step": 99000
    },
    {
      "epoch": 3.027433249831979,
      "grad_norm": 4.5017523765563965,
      "learning_rate": 3.4923204014196546e-05,
      "loss": 1.0285,
      "step": 99100
    },
    {
      "epoch": 3.030488177430195,
      "grad_norm": 3.855346918106079,
      "learning_rate": 3.490790600905642e-05,
      "loss": 1.2724,
      "step": 99200
    },
    {
      "epoch": 3.0335431050284107,
      "grad_norm": 7.594958782196045,
      "learning_rate": 3.489260800391629e-05,
      "loss": 1.0123,
      "step": 99300
    },
    {
      "epoch": 3.036598032626627,
      "grad_norm": 5.651984214782715,
      "learning_rate": 3.487730999877616e-05,
      "loss": 1.0763,
      "step": 99400
    },
    {
      "epoch": 3.0396529602248425,
      "grad_norm": 3.7878353595733643,
      "learning_rate": 3.486201199363603e-05,
      "loss": 1.1441,
      "step": 99500
    },
    {
      "epoch": 3.0427078878230587,
      "grad_norm": 3.542327404022217,
      "learning_rate": 3.48467139884959e-05,
      "loss": 1.0631,
      "step": 99600
    },
    {
      "epoch": 3.0457628154212744,
      "grad_norm": 2.071479082107544,
      "learning_rate": 3.483141598335577e-05,
      "loss": 1.1079,
      "step": 99700
    },
    {
      "epoch": 3.0488177430194905,
      "grad_norm": 7.825259208679199,
      "learning_rate": 3.481611797821564e-05,
      "loss": 1.1047,
      "step": 99800
    },
    {
      "epoch": 3.051872670617706,
      "grad_norm": 6.0007452964782715,
      "learning_rate": 3.480081997307551e-05,
      "loss": 1.3561,
      "step": 99900
    },
    {
      "epoch": 3.0549275982159223,
      "grad_norm": 3.6153903007507324,
      "learning_rate": 3.478552196793538e-05,
      "loss": 1.1561,
      "step": 100000
    },
    {
      "epoch": 3.057982525814138,
      "grad_norm": 5.3658552169799805,
      "learning_rate": 3.477022396279525e-05,
      "loss": 1.1673,
      "step": 100100
    },
    {
      "epoch": 3.061037453412354,
      "grad_norm": 6.298893928527832,
      "learning_rate": 3.475492595765512e-05,
      "loss": 1.0913,
      "step": 100200
    },
    {
      "epoch": 3.0640923810105702,
      "grad_norm": 3.745986223220825,
      "learning_rate": 3.473962795251499e-05,
      "loss": 1.2041,
      "step": 100300
    },
    {
      "epoch": 3.067147308608786,
      "grad_norm": 2.570671558380127,
      "learning_rate": 3.4724329947374864e-05,
      "loss": 1.2944,
      "step": 100400
    },
    {
      "epoch": 3.070202236207002,
      "grad_norm": 2.7574243545532227,
      "learning_rate": 3.4709031942234735e-05,
      "loss": 1.2773,
      "step": 100500
    },
    {
      "epoch": 3.0732571638052177,
      "grad_norm": 2.7146689891815186,
      "learning_rate": 3.46937339370946e-05,
      "loss": 1.0322,
      "step": 100600
    },
    {
      "epoch": 3.076312091403434,
      "grad_norm": 5.170768737792969,
      "learning_rate": 3.467843593195447e-05,
      "loss": 1.084,
      "step": 100700
    },
    {
      "epoch": 3.0793670190016496,
      "grad_norm": 6.875863552093506,
      "learning_rate": 3.466313792681434e-05,
      "loss": 1.2769,
      "step": 100800
    },
    {
      "epoch": 3.0824219465998657,
      "grad_norm": 10.701215744018555,
      "learning_rate": 3.464783992167421e-05,
      "loss": 1.3305,
      "step": 100900
    },
    {
      "epoch": 3.0854768741980814,
      "grad_norm": 5.790326118469238,
      "learning_rate": 3.4632541916534084e-05,
      "loss": 0.9886,
      "step": 101000
    },
    {
      "epoch": 3.0885318017962975,
      "grad_norm": 10.692790031433105,
      "learning_rate": 3.4617243911393955e-05,
      "loss": 1.0596,
      "step": 101100
    },
    {
      "epoch": 3.091586729394513,
      "grad_norm": 0.11457054316997528,
      "learning_rate": 3.4601945906253826e-05,
      "loss": 1.0932,
      "step": 101200
    },
    {
      "epoch": 3.0946416569927293,
      "grad_norm": 2.845723867416382,
      "learning_rate": 3.45866479011137e-05,
      "loss": 1.1524,
      "step": 101300
    },
    {
      "epoch": 3.097696584590945,
      "grad_norm": 8.70290470123291,
      "learning_rate": 3.457134989597356e-05,
      "loss": 1.2215,
      "step": 101400
    },
    {
      "epoch": 3.100751512189161,
      "grad_norm": 3.0512425899505615,
      "learning_rate": 3.455605189083343e-05,
      "loss": 1.051,
      "step": 101500
    },
    {
      "epoch": 3.1038064397873772,
      "grad_norm": 4.125731945037842,
      "learning_rate": 3.4540753885693304e-05,
      "loss": 1.3042,
      "step": 101600
    },
    {
      "epoch": 3.106861367385593,
      "grad_norm": 10.031002044677734,
      "learning_rate": 3.4525455880553175e-05,
      "loss": 1.1348,
      "step": 101700
    },
    {
      "epoch": 3.109916294983809,
      "grad_norm": 7.4683637619018555,
      "learning_rate": 3.4510157875413046e-05,
      "loss": 1.2125,
      "step": 101800
    },
    {
      "epoch": 3.1129712225820247,
      "grad_norm": 3.547574520111084,
      "learning_rate": 3.449485987027292e-05,
      "loss": 1.1046,
      "step": 101900
    },
    {
      "epoch": 3.116026150180241,
      "grad_norm": 17.192188262939453,
      "learning_rate": 3.447971484518419e-05,
      "loss": 1.2065,
      "step": 102000
    },
    {
      "epoch": 3.1190810777784566,
      "grad_norm": 2.1484580039978027,
      "learning_rate": 3.4464416840044063e-05,
      "loss": 1.235,
      "step": 102100
    },
    {
      "epoch": 3.1221360053766727,
      "grad_norm": 2.382115602493286,
      "learning_rate": 3.4449118834903935e-05,
      "loss": 1.0214,
      "step": 102200
    },
    {
      "epoch": 3.1251909329748884,
      "grad_norm": 9.626750946044922,
      "learning_rate": 3.44338208297638e-05,
      "loss": 1.2341,
      "step": 102300
    },
    {
      "epoch": 3.1282458605731045,
      "grad_norm": 7.635776042938232,
      "learning_rate": 3.441852282462367e-05,
      "loss": 1.3048,
      "step": 102400
    },
    {
      "epoch": 3.13130078817132,
      "grad_norm": 1.5449594259262085,
      "learning_rate": 3.440322481948354e-05,
      "loss": 1.1685,
      "step": 102500
    },
    {
      "epoch": 3.1343557157695363,
      "grad_norm": 2.670468807220459,
      "learning_rate": 3.438792681434341e-05,
      "loss": 1.0141,
      "step": 102600
    },
    {
      "epoch": 3.137410643367752,
      "grad_norm": 7.990098476409912,
      "learning_rate": 3.4372628809203284e-05,
      "loss": 1.1673,
      "step": 102700
    },
    {
      "epoch": 3.140465570965968,
      "grad_norm": 11.402932167053223,
      "learning_rate": 3.435748378411455e-05,
      "loss": 1.2708,
      "step": 102800
    },
    {
      "epoch": 3.1435204985641843,
      "grad_norm": 10.994685173034668,
      "learning_rate": 3.434218577897442e-05,
      "loss": 1.2813,
      "step": 102900
    },
    {
      "epoch": 3.1465754261624,
      "grad_norm": 8.80547046661377,
      "learning_rate": 3.4326887773834294e-05,
      "loss": 1.2177,
      "step": 103000
    },
    {
      "epoch": 3.149630353760616,
      "grad_norm": 8.42204475402832,
      "learning_rate": 3.4311589768694165e-05,
      "loss": 1.0678,
      "step": 103100
    },
    {
      "epoch": 3.1526852813588317,
      "grad_norm": 4.392109394073486,
      "learning_rate": 3.4296291763554036e-05,
      "loss": 1.0808,
      "step": 103200
    },
    {
      "epoch": 3.155740208957048,
      "grad_norm": 9.018641471862793,
      "learning_rate": 3.428099375841391e-05,
      "loss": 1.1999,
      "step": 103300
    },
    {
      "epoch": 3.1587951365552636,
      "grad_norm": 7.8954854011535645,
      "learning_rate": 3.426569575327377e-05,
      "loss": 0.9657,
      "step": 103400
    },
    {
      "epoch": 3.1618500641534797,
      "grad_norm": 1.4429759979248047,
      "learning_rate": 3.425039774813364e-05,
      "loss": 1.0844,
      "step": 103500
    },
    {
      "epoch": 3.1649049917516954,
      "grad_norm": 5.75102424621582,
      "learning_rate": 3.4235099742993514e-05,
      "loss": 1.2737,
      "step": 103600
    },
    {
      "epoch": 3.1679599193499115,
      "grad_norm": 6.8921122550964355,
      "learning_rate": 3.4219801737853385e-05,
      "loss": 1.0959,
      "step": 103700
    },
    {
      "epoch": 3.171014846948127,
      "grad_norm": 3.6340596675872803,
      "learning_rate": 3.4204503732713256e-05,
      "loss": 1.0954,
      "step": 103800
    },
    {
      "epoch": 3.1740697745463433,
      "grad_norm": 19.560989379882812,
      "learning_rate": 3.418920572757313e-05,
      "loss": 1.0977,
      "step": 103900
    },
    {
      "epoch": 3.177124702144559,
      "grad_norm": 3.10819673538208,
      "learning_rate": 3.4173907722433e-05,
      "loss": 1.0771,
      "step": 104000
    },
    {
      "epoch": 3.180179629742775,
      "grad_norm": 21.455175399780273,
      "learning_rate": 3.415860971729287e-05,
      "loss": 1.0025,
      "step": 104100
    },
    {
      "epoch": 3.183234557340991,
      "grad_norm": 2.593312978744507,
      "learning_rate": 3.4143311712152734e-05,
      "loss": 1.0311,
      "step": 104200
    },
    {
      "epoch": 3.186289484939207,
      "grad_norm": 9.986220359802246,
      "learning_rate": 3.4128013707012605e-05,
      "loss": 1.0512,
      "step": 104300
    },
    {
      "epoch": 3.189344412537423,
      "grad_norm": 5.130756378173828,
      "learning_rate": 3.4112715701872477e-05,
      "loss": 0.8577,
      "step": 104400
    },
    {
      "epoch": 3.1923993401356388,
      "grad_norm": 2.575563669204712,
      "learning_rate": 3.409741769673235e-05,
      "loss": 1.2148,
      "step": 104500
    },
    {
      "epoch": 3.195454267733855,
      "grad_norm": 10.687899589538574,
      "learning_rate": 3.408211969159222e-05,
      "loss": 1.1314,
      "step": 104600
    },
    {
      "epoch": 3.1985091953320706,
      "grad_norm": 19.845109939575195,
      "learning_rate": 3.406682168645209e-05,
      "loss": 1.1975,
      "step": 104700
    },
    {
      "epoch": 3.2015641229302867,
      "grad_norm": 5.279234409332275,
      "learning_rate": 3.405167666136336e-05,
      "loss": 1.1001,
      "step": 104800
    },
    {
      "epoch": 3.2046190505285024,
      "grad_norm": 1.6543632745742798,
      "learning_rate": 3.403637865622323e-05,
      "loss": 1.2652,
      "step": 104900
    },
    {
      "epoch": 3.2076739781267185,
      "grad_norm": 3.378157377243042,
      "learning_rate": 3.40210806510831e-05,
      "loss": 1.11,
      "step": 105000
    },
    {
      "epoch": 3.210728905724934,
      "grad_norm": 6.591047763824463,
      "learning_rate": 3.400578264594297e-05,
      "loss": 1.201,
      "step": 105100
    },
    {
      "epoch": 3.2137838333231503,
      "grad_norm": 2.332491636276245,
      "learning_rate": 3.3990484640802836e-05,
      "loss": 1.0388,
      "step": 105200
    },
    {
      "epoch": 3.216838760921366,
      "grad_norm": 1.7545006275177002,
      "learning_rate": 3.397518663566271e-05,
      "loss": 1.1818,
      "step": 105300
    },
    {
      "epoch": 3.219893688519582,
      "grad_norm": 6.403434753417969,
      "learning_rate": 3.395988863052258e-05,
      "loss": 1.3045,
      "step": 105400
    },
    {
      "epoch": 3.222948616117798,
      "grad_norm": 2.699486255645752,
      "learning_rate": 3.394459062538245e-05,
      "loss": 1.0548,
      "step": 105500
    },
    {
      "epoch": 3.226003543716014,
      "grad_norm": 3.450688600540161,
      "learning_rate": 3.392929262024232e-05,
      "loss": 1.2869,
      "step": 105600
    },
    {
      "epoch": 3.2290584713142296,
      "grad_norm": 8.332239151000977,
      "learning_rate": 3.391399461510219e-05,
      "loss": 1.2839,
      "step": 105700
    },
    {
      "epoch": 3.2321133989124458,
      "grad_norm": 5.950933933258057,
      "learning_rate": 3.389869660996206e-05,
      "loss": 1.1488,
      "step": 105800
    },
    {
      "epoch": 3.235168326510662,
      "grad_norm": 4.980724811553955,
      "learning_rate": 3.3883398604821934e-05,
      "loss": 1.2878,
      "step": 105900
    },
    {
      "epoch": 3.2382232541088776,
      "grad_norm": 6.563425540924072,
      "learning_rate": 3.38681005996818e-05,
      "loss": 1.1434,
      "step": 106000
    },
    {
      "epoch": 3.2412781817070937,
      "grad_norm": 6.405630588531494,
      "learning_rate": 3.385280259454167e-05,
      "loss": 1.2905,
      "step": 106100
    },
    {
      "epoch": 3.2443331093053094,
      "grad_norm": 5.564048767089844,
      "learning_rate": 3.383750458940154e-05,
      "loss": 1.3337,
      "step": 106200
    },
    {
      "epoch": 3.2473880369035255,
      "grad_norm": 12.21498966217041,
      "learning_rate": 3.382220658426141e-05,
      "loss": 0.9791,
      "step": 106300
    },
    {
      "epoch": 3.250442964501741,
      "grad_norm": 4.009174823760986,
      "learning_rate": 3.380690857912128e-05,
      "loss": 1.1921,
      "step": 106400
    },
    {
      "epoch": 3.2534978920999573,
      "grad_norm": 6.204054832458496,
      "learning_rate": 3.3791610573981154e-05,
      "loss": 1.061,
      "step": 106500
    },
    {
      "epoch": 3.256552819698173,
      "grad_norm": 1.8835408687591553,
      "learning_rate": 3.3776312568841025e-05,
      "loss": 1.1738,
      "step": 106600
    },
    {
      "epoch": 3.259607747296389,
      "grad_norm": 3.0553841590881348,
      "learning_rate": 3.3761014563700896e-05,
      "loss": 1.1092,
      "step": 106700
    },
    {
      "epoch": 3.262662674894605,
      "grad_norm": 2.027163028717041,
      "learning_rate": 3.374571655856076e-05,
      "loss": 1.1073,
      "step": 106800
    },
    {
      "epoch": 3.265717602492821,
      "grad_norm": 6.837578773498535,
      "learning_rate": 3.373041855342063e-05,
      "loss": 1.0253,
      "step": 106900
    },
    {
      "epoch": 3.268772530091037,
      "grad_norm": 5.914719104766846,
      "learning_rate": 3.371527352833191e-05,
      "loss": 1.3284,
      "step": 107000
    },
    {
      "epoch": 3.2718274576892528,
      "grad_norm": 2.149627447128296,
      "learning_rate": 3.369997552319178e-05,
      "loss": 1.0899,
      "step": 107100
    },
    {
      "epoch": 3.2748823852874684,
      "grad_norm": 5.665188312530518,
      "learning_rate": 3.368467751805165e-05,
      "loss": 1.1932,
      "step": 107200
    },
    {
      "epoch": 3.2779373128856846,
      "grad_norm": 3.7738704681396484,
      "learning_rate": 3.366937951291152e-05,
      "loss": 1.1043,
      "step": 107300
    },
    {
      "epoch": 3.2809922404839007,
      "grad_norm": 8.023479461669922,
      "learning_rate": 3.365408150777139e-05,
      "loss": 1.062,
      "step": 107400
    },
    {
      "epoch": 3.2840471680821164,
      "grad_norm": 13.017199516296387,
      "learning_rate": 3.363878350263126e-05,
      "loss": 1.275,
      "step": 107500
    },
    {
      "epoch": 3.2871020956803325,
      "grad_norm": 9.704121589660645,
      "learning_rate": 3.3623485497491134e-05,
      "loss": 1.2566,
      "step": 107600
    },
    {
      "epoch": 3.290157023278548,
      "grad_norm": 1.2066890001296997,
      "learning_rate": 3.3608187492351e-05,
      "loss": 1.1705,
      "step": 107700
    },
    {
      "epoch": 3.2932119508767643,
      "grad_norm": 7.091336250305176,
      "learning_rate": 3.359288948721087e-05,
      "loss": 1.1722,
      "step": 107800
    },
    {
      "epoch": 3.29626687847498,
      "grad_norm": 5.988069534301758,
      "learning_rate": 3.357759148207074e-05,
      "loss": 1.2878,
      "step": 107900
    },
    {
      "epoch": 3.299321806073196,
      "grad_norm": 8.537280082702637,
      "learning_rate": 3.356229347693061e-05,
      "loss": 1.1086,
      "step": 108000
    },
    {
      "epoch": 3.302376733671412,
      "grad_norm": 10.873692512512207,
      "learning_rate": 3.354699547179048e-05,
      "loss": 1.0681,
      "step": 108100
    },
    {
      "epoch": 3.305431661269628,
      "grad_norm": 6.819619178771973,
      "learning_rate": 3.3531697466650354e-05,
      "loss": 1.1786,
      "step": 108200
    },
    {
      "epoch": 3.3084865888678436,
      "grad_norm": 3.012190103530884,
      "learning_rate": 3.3516399461510225e-05,
      "loss": 1.1915,
      "step": 108300
    },
    {
      "epoch": 3.3115415164660598,
      "grad_norm": 1.4920376539230347,
      "learning_rate": 3.3501101456370096e-05,
      "loss": 1.0751,
      "step": 108400
    },
    {
      "epoch": 3.314596444064276,
      "grad_norm": 0.5798110365867615,
      "learning_rate": 3.348580345122996e-05,
      "loss": 1.0887,
      "step": 108500
    },
    {
      "epoch": 3.3176513716624916,
      "grad_norm": 3.4853806495666504,
      "learning_rate": 3.347050544608983e-05,
      "loss": 1.1365,
      "step": 108600
    },
    {
      "epoch": 3.3207062992607077,
      "grad_norm": 6.759474754333496,
      "learning_rate": 3.34552074409497e-05,
      "loss": 1.2419,
      "step": 108700
    },
    {
      "epoch": 3.3237612268589234,
      "grad_norm": 2.99652099609375,
      "learning_rate": 3.3439909435809574e-05,
      "loss": 1.4198,
      "step": 108800
    },
    {
      "epoch": 3.3268161544571395,
      "grad_norm": 9.81747055053711,
      "learning_rate": 3.3424611430669445e-05,
      "loss": 1.2768,
      "step": 108900
    },
    {
      "epoch": 3.329871082055355,
      "grad_norm": 6.646118640899658,
      "learning_rate": 3.340946640558071e-05,
      "loss": 1.4317,
      "step": 109000
    },
    {
      "epoch": 3.3329260096535713,
      "grad_norm": 2.144011974334717,
      "learning_rate": 3.3394168400440584e-05,
      "loss": 1.2185,
      "step": 109100
    },
    {
      "epoch": 3.335980937251787,
      "grad_norm": 3.1215665340423584,
      "learning_rate": 3.3378870395300455e-05,
      "loss": 1.1337,
      "step": 109200
    },
    {
      "epoch": 3.339035864850003,
      "grad_norm": 8.02299976348877,
      "learning_rate": 3.3363572390160327e-05,
      "loss": 1.1333,
      "step": 109300
    },
    {
      "epoch": 3.342090792448219,
      "grad_norm": 11.480727195739746,
      "learning_rate": 3.33482743850202e-05,
      "loss": 1.0622,
      "step": 109400
    },
    {
      "epoch": 3.345145720046435,
      "grad_norm": 0.7370632290840149,
      "learning_rate": 3.333297637988006e-05,
      "loss": 1.1977,
      "step": 109500
    },
    {
      "epoch": 3.3482006476446506,
      "grad_norm": 8.795260429382324,
      "learning_rate": 3.331767837473993e-05,
      "loss": 1.058,
      "step": 109600
    },
    {
      "epoch": 3.3512555752428668,
      "grad_norm": 3.5496344566345215,
      "learning_rate": 3.3302380369599804e-05,
      "loss": 1.2247,
      "step": 109700
    },
    {
      "epoch": 3.3543105028410825,
      "grad_norm": 2.8768632411956787,
      "learning_rate": 3.3287082364459676e-05,
      "loss": 1.1976,
      "step": 109800
    },
    {
      "epoch": 3.3573654304392986,
      "grad_norm": 0.17894093692302704,
      "learning_rate": 3.327178435931955e-05,
      "loss": 1.0641,
      "step": 109900
    },
    {
      "epoch": 3.3604203580375147,
      "grad_norm": 4.932908535003662,
      "learning_rate": 3.325648635417942e-05,
      "loss": 1.1351,
      "step": 110000
    },
    {
      "epoch": 3.3634752856357304,
      "grad_norm": 4.198153972625732,
      "learning_rate": 3.324118834903929e-05,
      "loss": 1.1921,
      "step": 110100
    },
    {
      "epoch": 3.3665302132339465,
      "grad_norm": 4.184053897857666,
      "learning_rate": 3.322589034389916e-05,
      "loss": 1.1354,
      "step": 110200
    },
    {
      "epoch": 3.369585140832162,
      "grad_norm": 6.549489974975586,
      "learning_rate": 3.3210592338759025e-05,
      "loss": 1.3167,
      "step": 110300
    },
    {
      "epoch": 3.3726400684303783,
      "grad_norm": 10.888463973999023,
      "learning_rate": 3.3195294333618896e-05,
      "loss": 1.542,
      "step": 110400
    },
    {
      "epoch": 3.375694996028594,
      "grad_norm": 3.289114475250244,
      "learning_rate": 3.317999632847877e-05,
      "loss": 1.2994,
      "step": 110500
    },
    {
      "epoch": 3.37874992362681,
      "grad_norm": 21.504901885986328,
      "learning_rate": 3.316469832333864e-05,
      "loss": 1.4213,
      "step": 110600
    },
    {
      "epoch": 3.381804851225026,
      "grad_norm": 15.273396492004395,
      "learning_rate": 3.314940031819851e-05,
      "loss": 1.2337,
      "step": 110700
    },
    {
      "epoch": 3.384859778823242,
      "grad_norm": 9.156777381896973,
      "learning_rate": 3.313410231305838e-05,
      "loss": 1.1234,
      "step": 110800
    },
    {
      "epoch": 3.3879147064214576,
      "grad_norm": 12.19512939453125,
      "learning_rate": 3.311880430791825e-05,
      "loss": 1.275,
      "step": 110900
    },
    {
      "epoch": 3.3909696340196738,
      "grad_norm": 5.635876655578613,
      "learning_rate": 3.310350630277812e-05,
      "loss": 1.1857,
      "step": 111000
    },
    {
      "epoch": 3.39402456161789,
      "grad_norm": 4.696619510650635,
      "learning_rate": 3.308820829763799e-05,
      "loss": 1.1384,
      "step": 111100
    },
    {
      "epoch": 3.3970794892161056,
      "grad_norm": 4.503530979156494,
      "learning_rate": 3.307291029249786e-05,
      "loss": 1.0952,
      "step": 111200
    },
    {
      "epoch": 3.4001344168143213,
      "grad_norm": 3.3427395820617676,
      "learning_rate": 3.305761228735773e-05,
      "loss": 1.1545,
      "step": 111300
    },
    {
      "epoch": 3.4031893444125374,
      "grad_norm": 10.8120698928833,
      "learning_rate": 3.30423142822176e-05,
      "loss": 1.0966,
      "step": 111400
    },
    {
      "epoch": 3.4062442720107535,
      "grad_norm": 3.6469085216522217,
      "learning_rate": 3.302701627707747e-05,
      "loss": 1.1971,
      "step": 111500
    },
    {
      "epoch": 3.409299199608969,
      "grad_norm": 3.417675256729126,
      "learning_rate": 3.301171827193734e-05,
      "loss": 0.9999,
      "step": 111600
    },
    {
      "epoch": 3.4123541272071853,
      "grad_norm": 5.76259183883667,
      "learning_rate": 3.2996420266797214e-05,
      "loss": 1.1023,
      "step": 111700
    },
    {
      "epoch": 3.415409054805401,
      "grad_norm": 3.8383865356445312,
      "learning_rate": 3.2981122261657085e-05,
      "loss": 1.2614,
      "step": 111800
    },
    {
      "epoch": 3.418463982403617,
      "grad_norm": 6.684022903442383,
      "learning_rate": 3.296582425651695e-05,
      "loss": 0.9996,
      "step": 111900
    },
    {
      "epoch": 3.421518910001833,
      "grad_norm": 4.4537129402160645,
      "learning_rate": 3.295052625137682e-05,
      "loss": 1.1022,
      "step": 112000
    },
    {
      "epoch": 3.424573837600049,
      "grad_norm": 4.510301113128662,
      "learning_rate": 3.293522824623669e-05,
      "loss": 1.0628,
      "step": 112100
    },
    {
      "epoch": 3.4276287651982646,
      "grad_norm": 6.627735614776611,
      "learning_rate": 3.291993024109656e-05,
      "loss": 1.288,
      "step": 112200
    },
    {
      "epoch": 3.4306836927964808,
      "grad_norm": 3.7537848949432373,
      "learning_rate": 3.2904632235956434e-05,
      "loss": 1.2107,
      "step": 112300
    },
    {
      "epoch": 3.4337386203946965,
      "grad_norm": 9.15528392791748,
      "learning_rate": 3.2889334230816305e-05,
      "loss": 1.3645,
      "step": 112400
    },
    {
      "epoch": 3.4367935479929126,
      "grad_norm": 3.508101224899292,
      "learning_rate": 3.2874036225676177e-05,
      "loss": 1.1195,
      "step": 112500
    },
    {
      "epoch": 3.4398484755911287,
      "grad_norm": 17.049793243408203,
      "learning_rate": 3.285873822053605e-05,
      "loss": 1.0965,
      "step": 112600
    },
    {
      "epoch": 3.4429034031893444,
      "grad_norm": 13.650372505187988,
      "learning_rate": 3.2843593195447316e-05,
      "loss": 1.1358,
      "step": 112700
    },
    {
      "epoch": 3.44595833078756,
      "grad_norm": 2.2558670043945312,
      "learning_rate": 3.282829519030719e-05,
      "loss": 1.0418,
      "step": 112800
    },
    {
      "epoch": 3.449013258385776,
      "grad_norm": 5.237791538238525,
      "learning_rate": 3.281299718516706e-05,
      "loss": 1.2597,
      "step": 112900
    },
    {
      "epoch": 3.4520681859839923,
      "grad_norm": 14.691550254821777,
      "learning_rate": 3.279769918002692e-05,
      "loss": 1.1266,
      "step": 113000
    },
    {
      "epoch": 3.455123113582208,
      "grad_norm": 12.126757621765137,
      "learning_rate": 3.2782401174886793e-05,
      "loss": 1.1784,
      "step": 113100
    },
    {
      "epoch": 3.458178041180424,
      "grad_norm": 4.572924613952637,
      "learning_rate": 3.2767103169746665e-05,
      "loss": 0.97,
      "step": 113200
    },
    {
      "epoch": 3.46123296877864,
      "grad_norm": 3.525087594985962,
      "learning_rate": 3.2751805164606536e-05,
      "loss": 1.0714,
      "step": 113300
    },
    {
      "epoch": 3.464287896376856,
      "grad_norm": 6.896517276763916,
      "learning_rate": 3.273650715946641e-05,
      "loss": 1.3458,
      "step": 113400
    },
    {
      "epoch": 3.4673428239750717,
      "grad_norm": 4.984236240386963,
      "learning_rate": 3.272120915432628e-05,
      "loss": 1.3075,
      "step": 113500
    },
    {
      "epoch": 3.470397751573288,
      "grad_norm": 8.506633758544922,
      "learning_rate": 3.270591114918615e-05,
      "loss": 1.0258,
      "step": 113600
    },
    {
      "epoch": 3.4734526791715035,
      "grad_norm": 5.038571357727051,
      "learning_rate": 3.269061314404602e-05,
      "loss": 1.1029,
      "step": 113700
    },
    {
      "epoch": 3.4765076067697196,
      "grad_norm": 7.582324028015137,
      "learning_rate": 3.2675315138905885e-05,
      "loss": 1.5252,
      "step": 113800
    },
    {
      "epoch": 3.4795625343679353,
      "grad_norm": 2.692021131515503,
      "learning_rate": 3.2660017133765756e-05,
      "loss": 1.1836,
      "step": 113900
    },
    {
      "epoch": 3.4826174619661514,
      "grad_norm": 1.044533133506775,
      "learning_rate": 3.264471912862563e-05,
      "loss": 1.1172,
      "step": 114000
    },
    {
      "epoch": 3.4856723895643675,
      "grad_norm": 4.991665840148926,
      "learning_rate": 3.26294211234855e-05,
      "loss": 1.2193,
      "step": 114100
    },
    {
      "epoch": 3.488727317162583,
      "grad_norm": 7.2558393478393555,
      "learning_rate": 3.261412311834537e-05,
      "loss": 1.0472,
      "step": 114200
    },
    {
      "epoch": 3.4917822447607993,
      "grad_norm": 16.940465927124023,
      "learning_rate": 3.259882511320524e-05,
      "loss": 1.279,
      "step": 114300
    },
    {
      "epoch": 3.494837172359015,
      "grad_norm": 4.868178844451904,
      "learning_rate": 3.258352710806511e-05,
      "loss": 1.0672,
      "step": 114400
    },
    {
      "epoch": 3.497892099957231,
      "grad_norm": 4.789906978607178,
      "learning_rate": 3.256822910292498e-05,
      "loss": 1.058,
      "step": 114500
    },
    {
      "epoch": 3.500947027555447,
      "grad_norm": 3.3610029220581055,
      "learning_rate": 3.255293109778485e-05,
      "loss": 1.1544,
      "step": 114600
    },
    {
      "epoch": 3.504001955153663,
      "grad_norm": 6.413651466369629,
      "learning_rate": 3.253763309264472e-05,
      "loss": 1.1485,
      "step": 114700
    },
    {
      "epoch": 3.5070568827518787,
      "grad_norm": 3.62406587600708,
      "learning_rate": 3.252233508750459e-05,
      "loss": 1.1035,
      "step": 114800
    },
    {
      "epoch": 3.510111810350095,
      "grad_norm": 2.6523101329803467,
      "learning_rate": 3.250703708236446e-05,
      "loss": 1.0266,
      "step": 114900
    },
    {
      "epoch": 3.5131667379483105,
      "grad_norm": 10.615732192993164,
      "learning_rate": 3.249173907722433e-05,
      "loss": 1.1927,
      "step": 115000
    },
    {
      "epoch": 3.5162216655465266,
      "grad_norm": 4.684859275817871,
      "learning_rate": 3.24764410720842e-05,
      "loss": 1.2108,
      "step": 115100
    },
    {
      "epoch": 3.5192765931447427,
      "grad_norm": 5.885335922241211,
      "learning_rate": 3.2461143066944074e-05,
      "loss": 1.1479,
      "step": 115200
    },
    {
      "epoch": 3.5223315207429584,
      "grad_norm": 16.513399124145508,
      "learning_rate": 3.2445845061803945e-05,
      "loss": 1.2975,
      "step": 115300
    },
    {
      "epoch": 3.525386448341174,
      "grad_norm": 7.12460470199585,
      "learning_rate": 3.243054705666381e-05,
      "loss": 1.134,
      "step": 115400
    },
    {
      "epoch": 3.5284413759393902,
      "grad_norm": 14.912757873535156,
      "learning_rate": 3.241524905152368e-05,
      "loss": 1.1733,
      "step": 115500
    },
    {
      "epoch": 3.5314963035376064,
      "grad_norm": 3.456683397293091,
      "learning_rate": 3.239995104638355e-05,
      "loss": 1.1846,
      "step": 115600
    },
    {
      "epoch": 3.534551231135822,
      "grad_norm": 1.5884753465652466,
      "learning_rate": 3.238465304124342e-05,
      "loss": 1.2762,
      "step": 115700
    },
    {
      "epoch": 3.537606158734038,
      "grad_norm": 4.264137268066406,
      "learning_rate": 3.2369355036103294e-05,
      "loss": 1.2741,
      "step": 115800
    },
    {
      "epoch": 3.540661086332254,
      "grad_norm": 7.918459892272949,
      "learning_rate": 3.2354057030963166e-05,
      "loss": 1.095,
      "step": 115900
    },
    {
      "epoch": 3.54371601393047,
      "grad_norm": 21.39558219909668,
      "learning_rate": 3.233875902582304e-05,
      "loss": 1.2129,
      "step": 116000
    },
    {
      "epoch": 3.5467709415286857,
      "grad_norm": 12.257303237915039,
      "learning_rate": 3.232346102068291e-05,
      "loss": 1.015,
      "step": 116100
    },
    {
      "epoch": 3.549825869126902,
      "grad_norm": 6.592789173126221,
      "learning_rate": 3.230816301554277e-05,
      "loss": 1.2219,
      "step": 116200
    },
    {
      "epoch": 3.5528807967251175,
      "grad_norm": 10.32773494720459,
      "learning_rate": 3.2292865010402643e-05,
      "loss": 1.157,
      "step": 116300
    },
    {
      "epoch": 3.5559357243233336,
      "grad_norm": 2.9441211223602295,
      "learning_rate": 3.2277567005262515e-05,
      "loss": 1.1678,
      "step": 116400
    },
    {
      "epoch": 3.5589906519215493,
      "grad_norm": 4.8874897956848145,
      "learning_rate": 3.2262269000122386e-05,
      "loss": 1.1722,
      "step": 116500
    },
    {
      "epoch": 3.5620455795197654,
      "grad_norm": 17.394983291625977,
      "learning_rate": 3.224697099498226e-05,
      "loss": 1.3023,
      "step": 116600
    },
    {
      "epoch": 3.5651005071179815,
      "grad_norm": 3.721679449081421,
      "learning_rate": 3.2231825969893525e-05,
      "loss": 1.3001,
      "step": 116700
    },
    {
      "epoch": 3.5681554347161972,
      "grad_norm": 3.0622265338897705,
      "learning_rate": 3.2216527964753396e-05,
      "loss": 1.2024,
      "step": 116800
    },
    {
      "epoch": 3.571210362314413,
      "grad_norm": 7.172619819641113,
      "learning_rate": 3.220122995961327e-05,
      "loss": 1.1626,
      "step": 116900
    },
    {
      "epoch": 3.574265289912629,
      "grad_norm": 7.2092790603637695,
      "learning_rate": 3.218593195447314e-05,
      "loss": 1.374,
      "step": 117000
    },
    {
      "epoch": 3.577320217510845,
      "grad_norm": 9.982060432434082,
      "learning_rate": 3.217063394933301e-05,
      "loss": 1.1833,
      "step": 117100
    },
    {
      "epoch": 3.580375145109061,
      "grad_norm": 4.011295318603516,
      "learning_rate": 3.2155335944192874e-05,
      "loss": 1.1522,
      "step": 117200
    },
    {
      "epoch": 3.583430072707277,
      "grad_norm": 12.791787147521973,
      "learning_rate": 3.2140037939052745e-05,
      "loss": 1.0462,
      "step": 117300
    },
    {
      "epoch": 3.5864850003054927,
      "grad_norm": 14.182363510131836,
      "learning_rate": 3.2124739933912616e-05,
      "loss": 1.2998,
      "step": 117400
    },
    {
      "epoch": 3.589539927903709,
      "grad_norm": 4.25027322769165,
      "learning_rate": 3.210944192877249e-05,
      "loss": 1.0125,
      "step": 117500
    },
    {
      "epoch": 3.5925948555019245,
      "grad_norm": 1.9661411046981812,
      "learning_rate": 3.209414392363236e-05,
      "loss": 1.186,
      "step": 117600
    },
    {
      "epoch": 3.5956497831001406,
      "grad_norm": 3.385308027267456,
      "learning_rate": 3.207884591849223e-05,
      "loss": 1.0252,
      "step": 117700
    },
    {
      "epoch": 3.5987047106983563,
      "grad_norm": 9.881473541259766,
      "learning_rate": 3.20635479133521e-05,
      "loss": 1.1523,
      "step": 117800
    },
    {
      "epoch": 3.6017596382965724,
      "grad_norm": 5.886788845062256,
      "learning_rate": 3.204824990821197e-05,
      "loss": 1.2045,
      "step": 117900
    },
    {
      "epoch": 3.604814565894788,
      "grad_norm": 7.499330520629883,
      "learning_rate": 3.2032951903071836e-05,
      "loss": 1.2884,
      "step": 118000
    },
    {
      "epoch": 3.6078694934930042,
      "grad_norm": 11.073080062866211,
      "learning_rate": 3.201765389793171e-05,
      "loss": 1.0986,
      "step": 118100
    },
    {
      "epoch": 3.6109244210912204,
      "grad_norm": 3.1535775661468506,
      "learning_rate": 3.200235589279158e-05,
      "loss": 1.2812,
      "step": 118200
    },
    {
      "epoch": 3.613979348689436,
      "grad_norm": 6.089379787445068,
      "learning_rate": 3.198705788765145e-05,
      "loss": 1.1118,
      "step": 118300
    },
    {
      "epoch": 3.6170342762876517,
      "grad_norm": 2.273271083831787,
      "learning_rate": 3.197175988251132e-05,
      "loss": 1.1566,
      "step": 118400
    },
    {
      "epoch": 3.620089203885868,
      "grad_norm": 5.785590648651123,
      "learning_rate": 3.195646187737119e-05,
      "loss": 1.0192,
      "step": 118500
    },
    {
      "epoch": 3.623144131484084,
      "grad_norm": 4.7622480392456055,
      "learning_rate": 3.194116387223106e-05,
      "loss": 1.0438,
      "step": 118600
    },
    {
      "epoch": 3.6261990590822997,
      "grad_norm": 5.416982173919678,
      "learning_rate": 3.192601884714234e-05,
      "loss": 1.2225,
      "step": 118700
    },
    {
      "epoch": 3.629253986680516,
      "grad_norm": 3.769057512283325,
      "learning_rate": 3.191072084200221e-05,
      "loss": 1.115,
      "step": 118800
    },
    {
      "epoch": 3.6323089142787315,
      "grad_norm": 4.778010845184326,
      "learning_rate": 3.1895422836862074e-05,
      "loss": 1.1272,
      "step": 118900
    },
    {
      "epoch": 3.6353638418769476,
      "grad_norm": 3.8819594383239746,
      "learning_rate": 3.1880124831721945e-05,
      "loss": 1.2657,
      "step": 119000
    },
    {
      "epoch": 3.6384187694751633,
      "grad_norm": 2.2964558601379395,
      "learning_rate": 3.1864826826581816e-05,
      "loss": 1.1895,
      "step": 119100
    },
    {
      "epoch": 3.6414736970733794,
      "grad_norm": 2.7459022998809814,
      "learning_rate": 3.184952882144169e-05,
      "loss": 1.0707,
      "step": 119200
    },
    {
      "epoch": 3.6445286246715956,
      "grad_norm": 10.472131729125977,
      "learning_rate": 3.183423081630156e-05,
      "loss": 1.3939,
      "step": 119300
    },
    {
      "epoch": 3.6475835522698112,
      "grad_norm": 3.7901763916015625,
      "learning_rate": 3.181893281116143e-05,
      "loss": 1.3832,
      "step": 119400
    },
    {
      "epoch": 3.650638479868027,
      "grad_norm": 5.768411636352539,
      "learning_rate": 3.18036348060213e-05,
      "loss": 1.0307,
      "step": 119500
    },
    {
      "epoch": 3.653693407466243,
      "grad_norm": 6.40388298034668,
      "learning_rate": 3.178833680088117e-05,
      "loss": 1.1032,
      "step": 119600
    },
    {
      "epoch": 3.656748335064459,
      "grad_norm": 10.709514617919922,
      "learning_rate": 3.1773038795741036e-05,
      "loss": 1.0204,
      "step": 119700
    },
    {
      "epoch": 3.659803262662675,
      "grad_norm": 7.594704627990723,
      "learning_rate": 3.175774079060091e-05,
      "loss": 1.1728,
      "step": 119800
    },
    {
      "epoch": 3.6628581902608905,
      "grad_norm": 2.95936918258667,
      "learning_rate": 3.174244278546078e-05,
      "loss": 1.1619,
      "step": 119900
    },
    {
      "epoch": 3.6659131178591067,
      "grad_norm": 8.665496826171875,
      "learning_rate": 3.172714478032065e-05,
      "loss": 1.0554,
      "step": 120000
    },
    {
      "epoch": 3.668968045457323,
      "grad_norm": 6.218582630157471,
      "learning_rate": 3.171184677518052e-05,
      "loss": 1.2974,
      "step": 120100
    },
    {
      "epoch": 3.6720229730555385,
      "grad_norm": 1.6152470111846924,
      "learning_rate": 3.169654877004039e-05,
      "loss": 1.2447,
      "step": 120200
    },
    {
      "epoch": 3.6750779006537546,
      "grad_norm": 4.754295349121094,
      "learning_rate": 3.168125076490026e-05,
      "loss": 1.2054,
      "step": 120300
    },
    {
      "epoch": 3.6781328282519703,
      "grad_norm": 6.933786392211914,
      "learning_rate": 3.1665952759760134e-05,
      "loss": 1.1301,
      "step": 120400
    },
    {
      "epoch": 3.6811877558501864,
      "grad_norm": 1.8776979446411133,
      "learning_rate": 3.165065475462e-05,
      "loss": 1.1499,
      "step": 120500
    },
    {
      "epoch": 3.684242683448402,
      "grad_norm": 6.305478096008301,
      "learning_rate": 3.163535674947987e-05,
      "loss": 0.9828,
      "step": 120600
    },
    {
      "epoch": 3.6872976110466182,
      "grad_norm": 11.32885456085205,
      "learning_rate": 3.1620211724391144e-05,
      "loss": 1.386,
      "step": 120700
    },
    {
      "epoch": 3.6903525386448344,
      "grad_norm": 3.6041526794433594,
      "learning_rate": 3.160491371925101e-05,
      "loss": 1.1403,
      "step": 120800
    },
    {
      "epoch": 3.69340746624305,
      "grad_norm": 5.200024127960205,
      "learning_rate": 3.158961571411088e-05,
      "loss": 1.1867,
      "step": 120900
    },
    {
      "epoch": 3.6964623938412657,
      "grad_norm": 4.821595668792725,
      "learning_rate": 3.157431770897075e-05,
      "loss": 1.1326,
      "step": 121000
    },
    {
      "epoch": 3.699517321439482,
      "grad_norm": 6.276610374450684,
      "learning_rate": 3.155901970383062e-05,
      "loss": 1.2586,
      "step": 121100
    },
    {
      "epoch": 3.702572249037698,
      "grad_norm": 2.143345355987549,
      "learning_rate": 3.1543721698690493e-05,
      "loss": 1.2188,
      "step": 121200
    },
    {
      "epoch": 3.7056271766359137,
      "grad_norm": 2.437786340713501,
      "learning_rate": 3.1528423693550365e-05,
      "loss": 1.246,
      "step": 121300
    },
    {
      "epoch": 3.70868210423413,
      "grad_norm": 4.366087913513184,
      "learning_rate": 3.1513125688410236e-05,
      "loss": 0.9416,
      "step": 121400
    },
    {
      "epoch": 3.7117370318323455,
      "grad_norm": 10.98507308959961,
      "learning_rate": 3.149782768327011e-05,
      "loss": 1.2776,
      "step": 121500
    },
    {
      "epoch": 3.7147919594305616,
      "grad_norm": 8.077423095703125,
      "learning_rate": 3.148252967812997e-05,
      "loss": 1.0204,
      "step": 121600
    },
    {
      "epoch": 3.7178468870287773,
      "grad_norm": 11.440945625305176,
      "learning_rate": 3.146723167298984e-05,
      "loss": 1.2277,
      "step": 121700
    },
    {
      "epoch": 3.7209018146269934,
      "grad_norm": 13.08006477355957,
      "learning_rate": 3.1451933667849714e-05,
      "loss": 1.1211,
      "step": 121800
    },
    {
      "epoch": 3.723956742225209,
      "grad_norm": 14.114601135253906,
      "learning_rate": 3.1436635662709585e-05,
      "loss": 1.338,
      "step": 121900
    },
    {
      "epoch": 3.7270116698234252,
      "grad_norm": 4.147833824157715,
      "learning_rate": 3.1421337657569456e-05,
      "loss": 1.1939,
      "step": 122000
    },
    {
      "epoch": 3.730066597421641,
      "grad_norm": 2.173335075378418,
      "learning_rate": 3.140603965242933e-05,
      "loss": 0.9904,
      "step": 122100
    },
    {
      "epoch": 3.733121525019857,
      "grad_norm": 1.0117582082748413,
      "learning_rate": 3.13907416472892e-05,
      "loss": 1.1817,
      "step": 122200
    },
    {
      "epoch": 3.736176452618073,
      "grad_norm": 4.836818695068359,
      "learning_rate": 3.137544364214907e-05,
      "loss": 1.153,
      "step": 122300
    },
    {
      "epoch": 3.739231380216289,
      "grad_norm": 3.414306402206421,
      "learning_rate": 3.1360145637008934e-05,
      "loss": 1.1321,
      "step": 122400
    },
    {
      "epoch": 3.7422863078145046,
      "grad_norm": 5.739386558532715,
      "learning_rate": 3.1344847631868805e-05,
      "loss": 1.1302,
      "step": 122500
    },
    {
      "epoch": 3.7453412354127207,
      "grad_norm": 8.772282600402832,
      "learning_rate": 3.1329549626728676e-05,
      "loss": 1.1282,
      "step": 122600
    },
    {
      "epoch": 3.748396163010937,
      "grad_norm": 7.5907464027404785,
      "learning_rate": 3.1314404601639944e-05,
      "loss": 1.1954,
      "step": 122700
    },
    {
      "epoch": 3.7514510906091525,
      "grad_norm": 2.3881170749664307,
      "learning_rate": 3.1299106596499815e-05,
      "loss": 1.1865,
      "step": 122800
    },
    {
      "epoch": 3.7545060182073686,
      "grad_norm": 5.920051574707031,
      "learning_rate": 3.1283808591359686e-05,
      "loss": 1.0913,
      "step": 122900
    },
    {
      "epoch": 3.7575609458055843,
      "grad_norm": 6.340502738952637,
      "learning_rate": 3.126851058621956e-05,
      "loss": 1.2684,
      "step": 123000
    },
    {
      "epoch": 3.7606158734038004,
      "grad_norm": 13.953529357910156,
      "learning_rate": 3.125321258107943e-05,
      "loss": 1.1196,
      "step": 123100
    },
    {
      "epoch": 3.763670801002016,
      "grad_norm": 2.678821325302124,
      "learning_rate": 3.12379145759393e-05,
      "loss": 1.148,
      "step": 123200
    },
    {
      "epoch": 3.7667257286002322,
      "grad_norm": 4.08374547958374,
      "learning_rate": 3.122261657079917e-05,
      "loss": 0.9931,
      "step": 123300
    },
    {
      "epoch": 3.769780656198448,
      "grad_norm": 4.527195453643799,
      "learning_rate": 3.1207318565659035e-05,
      "loss": 1.2992,
      "step": 123400
    },
    {
      "epoch": 3.772835583796664,
      "grad_norm": 6.90880012512207,
      "learning_rate": 3.1192020560518907e-05,
      "loss": 1.1887,
      "step": 123500
    },
    {
      "epoch": 3.7758905113948797,
      "grad_norm": 11.422993659973145,
      "learning_rate": 3.117672255537878e-05,
      "loss": 1.0616,
      "step": 123600
    },
    {
      "epoch": 3.778945438993096,
      "grad_norm": 3.5831525325775146,
      "learning_rate": 3.116142455023865e-05,
      "loss": 1.3278,
      "step": 123700
    },
    {
      "epoch": 3.782000366591312,
      "grad_norm": 13.411617279052734,
      "learning_rate": 3.114612654509852e-05,
      "loss": 1.2322,
      "step": 123800
    },
    {
      "epoch": 3.7850552941895277,
      "grad_norm": 5.012452125549316,
      "learning_rate": 3.113082853995839e-05,
      "loss": 1.1816,
      "step": 123900
    },
    {
      "epoch": 3.7881102217877434,
      "grad_norm": 5.499082088470459,
      "learning_rate": 3.111553053481826e-05,
      "loss": 1.079,
      "step": 124000
    },
    {
      "epoch": 3.7911651493859595,
      "grad_norm": 18.250707626342773,
      "learning_rate": 3.1100232529678133e-05,
      "loss": 1.1843,
      "step": 124100
    },
    {
      "epoch": 3.7942200769841756,
      "grad_norm": 1.7736492156982422,
      "learning_rate": 3.1084934524538e-05,
      "loss": 1.3134,
      "step": 124200
    },
    {
      "epoch": 3.7972750045823913,
      "grad_norm": 4.700921058654785,
      "learning_rate": 3.106963651939787e-05,
      "loss": 1.2205,
      "step": 124300
    },
    {
      "epoch": 3.8003299321806074,
      "grad_norm": 3.189049482345581,
      "learning_rate": 3.105433851425774e-05,
      "loss": 1.1814,
      "step": 124400
    },
    {
      "epoch": 3.803384859778823,
      "grad_norm": 3.8542063236236572,
      "learning_rate": 3.103904050911761e-05,
      "loss": 1.2317,
      "step": 124500
    },
    {
      "epoch": 3.8064397873770393,
      "grad_norm": 5.694820880889893,
      "learning_rate": 3.102374250397748e-05,
      "loss": 1.1466,
      "step": 124600
    },
    {
      "epoch": 3.809494714975255,
      "grad_norm": 4.0774455070495605,
      "learning_rate": 3.100859747888876e-05,
      "loss": 1.1709,
      "step": 124700
    },
    {
      "epoch": 3.812549642573471,
      "grad_norm": 2.863539457321167,
      "learning_rate": 3.099329947374863e-05,
      "loss": 1.2222,
      "step": 124800
    },
    {
      "epoch": 3.815604570171687,
      "grad_norm": 4.64631986618042,
      "learning_rate": 3.0978154448659896e-05,
      "loss": 1.2713,
      "step": 124900
    },
    {
      "epoch": 3.818659497769903,
      "grad_norm": 7.270814895629883,
      "learning_rate": 3.096285644351977e-05,
      "loss": 0.9987,
      "step": 125000
    },
    {
      "epoch": 3.8217144253681186,
      "grad_norm": 2.774071216583252,
      "learning_rate": 3.094755843837964e-05,
      "loss": 1.3305,
      "step": 125100
    },
    {
      "epoch": 3.8247693529663347,
      "grad_norm": 8.187067031860352,
      "learning_rate": 3.093226043323951e-05,
      "loss": 1.2963,
      "step": 125200
    },
    {
      "epoch": 3.827824280564551,
      "grad_norm": 6.8783369064331055,
      "learning_rate": 3.0916962428099374e-05,
      "loss": 1.2131,
      "step": 125300
    },
    {
      "epoch": 3.8308792081627665,
      "grad_norm": 11.448376655578613,
      "learning_rate": 3.0901664422959245e-05,
      "loss": 1.2464,
      "step": 125400
    },
    {
      "epoch": 3.8339341357609826,
      "grad_norm": 11.253005027770996,
      "learning_rate": 3.0886366417819116e-05,
      "loss": 1.0755,
      "step": 125500
    },
    {
      "epoch": 3.8369890633591983,
      "grad_norm": 5.660057544708252,
      "learning_rate": 3.087106841267899e-05,
      "loss": 1.1623,
      "step": 125600
    },
    {
      "epoch": 3.8400439909574144,
      "grad_norm": 6.888990879058838,
      "learning_rate": 3.085577040753886e-05,
      "loss": 1.2219,
      "step": 125700
    },
    {
      "epoch": 3.84309891855563,
      "grad_norm": 4.959685325622559,
      "learning_rate": 3.084047240239873e-05,
      "loss": 1.1457,
      "step": 125800
    },
    {
      "epoch": 3.8461538461538463,
      "grad_norm": 7.199165344238281,
      "learning_rate": 3.08251743972586e-05,
      "loss": 1.3069,
      "step": 125900
    },
    {
      "epoch": 3.849208773752062,
      "grad_norm": 8.368234634399414,
      "learning_rate": 3.080987639211847e-05,
      "loss": 1.315,
      "step": 126000
    },
    {
      "epoch": 3.852263701350278,
      "grad_norm": 17.217199325561523,
      "learning_rate": 3.079457838697834e-05,
      "loss": 1.0141,
      "step": 126100
    },
    {
      "epoch": 3.8553186289484938,
      "grad_norm": 8.5066556930542,
      "learning_rate": 3.077928038183821e-05,
      "loss": 1.1976,
      "step": 126200
    },
    {
      "epoch": 3.85837355654671,
      "grad_norm": 3.0581226348876953,
      "learning_rate": 3.076398237669808e-05,
      "loss": 1.1097,
      "step": 126300
    },
    {
      "epoch": 3.861428484144926,
      "grad_norm": 4.513498783111572,
      "learning_rate": 3.074868437155795e-05,
      "loss": 1.1935,
      "step": 126400
    },
    {
      "epoch": 3.8644834117431417,
      "grad_norm": 5.329300403594971,
      "learning_rate": 3.073338636641782e-05,
      "loss": 1.3866,
      "step": 126500
    },
    {
      "epoch": 3.8675383393413574,
      "grad_norm": 7.060613632202148,
      "learning_rate": 3.071808836127769e-05,
      "loss": 1.2348,
      "step": 126600
    },
    {
      "epoch": 3.8705932669395735,
      "grad_norm": 4.2163615226745605,
      "learning_rate": 3.0702790356137564e-05,
      "loss": 1.046,
      "step": 126700
    },
    {
      "epoch": 3.8736481945377896,
      "grad_norm": 6.511301040649414,
      "learning_rate": 3.0687492350997435e-05,
      "loss": 1.1079,
      "step": 126800
    },
    {
      "epoch": 3.8767031221360053,
      "grad_norm": 7.438748359680176,
      "learning_rate": 3.06721943458573e-05,
      "loss": 1.4255,
      "step": 126900
    },
    {
      "epoch": 3.8797580497342214,
      "grad_norm": 3.3718574047088623,
      "learning_rate": 3.065689634071717e-05,
      "loss": 1.1054,
      "step": 127000
    },
    {
      "epoch": 3.882812977332437,
      "grad_norm": 7.0842132568359375,
      "learning_rate": 3.064159833557704e-05,
      "loss": 1.0431,
      "step": 127100
    },
    {
      "epoch": 3.8858679049306533,
      "grad_norm": 2.1069562435150146,
      "learning_rate": 3.062630033043691e-05,
      "loss": 1.1558,
      "step": 127200
    },
    {
      "epoch": 3.888922832528869,
      "grad_norm": 2.8387815952301025,
      "learning_rate": 3.0611002325296784e-05,
      "loss": 1.2421,
      "step": 127300
    },
    {
      "epoch": 3.891977760127085,
      "grad_norm": 9.572854995727539,
      "learning_rate": 3.0595704320156655e-05,
      "loss": 1.1307,
      "step": 127400
    },
    {
      "epoch": 3.8950326877253008,
      "grad_norm": 7.052920818328857,
      "learning_rate": 3.0580406315016526e-05,
      "loss": 1.0729,
      "step": 127500
    },
    {
      "epoch": 3.898087615323517,
      "grad_norm": 2.5184378623962402,
      "learning_rate": 3.05651083098764e-05,
      "loss": 1.3155,
      "step": 127600
    },
    {
      "epoch": 3.9011425429217326,
      "grad_norm": 6.567754745483398,
      "learning_rate": 3.054981030473626e-05,
      "loss": 1.1312,
      "step": 127700
    },
    {
      "epoch": 3.9041974705199487,
      "grad_norm": 5.300526142120361,
      "learning_rate": 3.053451229959613e-05,
      "loss": 1.1642,
      "step": 127800
    },
    {
      "epoch": 3.907252398118165,
      "grad_norm": 3.9759981632232666,
      "learning_rate": 3.0519214294456004e-05,
      "loss": 1.047,
      "step": 127900
    },
    {
      "epoch": 3.9103073257163805,
      "grad_norm": 4.425239562988281,
      "learning_rate": 3.0503916289315875e-05,
      "loss": 1.1105,
      "step": 128000
    },
    {
      "epoch": 3.913362253314596,
      "grad_norm": 2.3210442066192627,
      "learning_rate": 3.0488618284175746e-05,
      "loss": 1.1971,
      "step": 128100
    },
    {
      "epoch": 3.9164171809128123,
      "grad_norm": 3.3596367835998535,
      "learning_rate": 3.0473320279035617e-05,
      "loss": 1.2001,
      "step": 128200
    },
    {
      "epoch": 3.9194721085110285,
      "grad_norm": 4.397903919219971,
      "learning_rate": 3.0458022273895485e-05,
      "loss": 1.2177,
      "step": 128300
    },
    {
      "epoch": 3.922527036109244,
      "grad_norm": 5.058310508728027,
      "learning_rate": 3.0442724268755356e-05,
      "loss": 1.2263,
      "step": 128400
    },
    {
      "epoch": 3.9255819637074603,
      "grad_norm": 10.105254173278809,
      "learning_rate": 3.0427426263615227e-05,
      "loss": 1.1451,
      "step": 128500
    },
    {
      "epoch": 3.928636891305676,
      "grad_norm": 6.7497968673706055,
      "learning_rate": 3.04121282584751e-05,
      "loss": 1.0464,
      "step": 128600
    },
    {
      "epoch": 3.931691818903892,
      "grad_norm": 3.9784913063049316,
      "learning_rate": 3.0396830253334966e-05,
      "loss": 1.0723,
      "step": 128700
    },
    {
      "epoch": 3.9347467465021078,
      "grad_norm": 0.002016042824834585,
      "learning_rate": 3.0381532248194838e-05,
      "loss": 0.9551,
      "step": 128800
    },
    {
      "epoch": 3.937801674100324,
      "grad_norm": 4.069670677185059,
      "learning_rate": 3.036623424305471e-05,
      "loss": 1.1617,
      "step": 128900
    },
    {
      "epoch": 3.94085660169854,
      "grad_norm": 13.04504680633545,
      "learning_rate": 3.0351089217965977e-05,
      "loss": 1.1916,
      "step": 129000
    },
    {
      "epoch": 3.9439115292967557,
      "grad_norm": 10.69211483001709,
      "learning_rate": 3.0335791212825848e-05,
      "loss": 1.1484,
      "step": 129100
    },
    {
      "epoch": 3.9469664568949714,
      "grad_norm": 5.319109916687012,
      "learning_rate": 3.032049320768572e-05,
      "loss": 1.2785,
      "step": 129200
    },
    {
      "epoch": 3.9500213844931875,
      "grad_norm": 17.455825805664062,
      "learning_rate": 3.030519520254559e-05,
      "loss": 1.0728,
      "step": 129300
    },
    {
      "epoch": 3.9530763120914036,
      "grad_norm": 10.500404357910156,
      "learning_rate": 3.0289897197405458e-05,
      "loss": 1.6514,
      "step": 129400
    },
    {
      "epoch": 3.9561312396896193,
      "grad_norm": 3.3658556938171387,
      "learning_rate": 3.027459919226533e-05,
      "loss": 1.3522,
      "step": 129500
    },
    {
      "epoch": 3.959186167287835,
      "grad_norm": 6.071293830871582,
      "learning_rate": 3.02593011871252e-05,
      "loss": 1.2511,
      "step": 129600
    },
    {
      "epoch": 3.962241094886051,
      "grad_norm": 18.586950302124023,
      "learning_rate": 3.024400318198507e-05,
      "loss": 1.1664,
      "step": 129700
    },
    {
      "epoch": 3.9652960224842673,
      "grad_norm": 11.053479194641113,
      "learning_rate": 3.022870517684494e-05,
      "loss": 1.2655,
      "step": 129800
    },
    {
      "epoch": 3.968350950082483,
      "grad_norm": 3.921522617340088,
      "learning_rate": 3.021340717170481e-05,
      "loss": 1.1595,
      "step": 129900
    },
    {
      "epoch": 3.971405877680699,
      "grad_norm": 15.36481761932373,
      "learning_rate": 3.019810916656468e-05,
      "loss": 1.1743,
      "step": 130000
    },
    {
      "epoch": 3.9744608052789148,
      "grad_norm": 5.760214328765869,
      "learning_rate": 3.0182811161424553e-05,
      "loss": 0.9441,
      "step": 130100
    },
    {
      "epoch": 3.977515732877131,
      "grad_norm": 6.384634971618652,
      "learning_rate": 3.016751315628442e-05,
      "loss": 1.0591,
      "step": 130200
    },
    {
      "epoch": 3.9805706604753466,
      "grad_norm": 5.667691230773926,
      "learning_rate": 3.015221515114429e-05,
      "loss": 1.3293,
      "step": 130300
    },
    {
      "epoch": 3.9836255880735627,
      "grad_norm": 6.393319606781006,
      "learning_rate": 3.0136917146004163e-05,
      "loss": 1.1806,
      "step": 130400
    },
    {
      "epoch": 3.986680515671779,
      "grad_norm": 4.59672737121582,
      "learning_rate": 3.0121619140864034e-05,
      "loss": 1.287,
      "step": 130500
    },
    {
      "epoch": 3.9897354432699945,
      "grad_norm": 6.654175281524658,
      "learning_rate": 3.01063211357239e-05,
      "loss": 1.1724,
      "step": 130600
    },
    {
      "epoch": 3.99279037086821,
      "grad_norm": 3.8924570083618164,
      "learning_rate": 3.0091023130583773e-05,
      "loss": 1.1715,
      "step": 130700
    },
    {
      "epoch": 3.9958452984664263,
      "grad_norm": 2.988812208175659,
      "learning_rate": 3.0075725125443644e-05,
      "loss": 0.9282,
      "step": 130800
    },
    {
      "epoch": 3.9989002260646425,
      "grad_norm": 10.128042221069336,
      "learning_rate": 3.0060427120303515e-05,
      "loss": 1.2072,
      "step": 130900
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.7121036231441314,
      "eval_loss": 0.7748239040374756,
      "eval_runtime": 1785.0458,
      "eval_samples_per_second": 18.338,
      "eval_steps_per_second": 4.585,
      "step": 130936
    },
    {
      "epoch": 4.001955153662858,
      "grad_norm": 2.5833377838134766,
      "learning_rate": 3.0045282095214783e-05,
      "loss": 1.3073,
      "step": 131000
    },
    {
      "epoch": 4.005010081261074,
      "grad_norm": 5.355751991271973,
      "learning_rate": 3.0029984090074654e-05,
      "loss": 1.0811,
      "step": 131100
    },
    {
      "epoch": 4.00806500885929,
      "grad_norm": 4.671720027923584,
      "learning_rate": 3.0014686084934522e-05,
      "loss": 1.1287,
      "step": 131200
    },
    {
      "epoch": 4.011119936457506,
      "grad_norm": 1.8795313835144043,
      "learning_rate": 2.9999541059845797e-05,
      "loss": 1.0344,
      "step": 131300
    },
    {
      "epoch": 4.014174864055722,
      "grad_norm": 3.870358943939209,
      "learning_rate": 2.9984243054705668e-05,
      "loss": 1.1018,
      "step": 131400
    },
    {
      "epoch": 4.0172297916539375,
      "grad_norm": 1.3954418897628784,
      "learning_rate": 2.996894504956554e-05,
      "loss": 1.0517,
      "step": 131500
    },
    {
      "epoch": 4.020284719252154,
      "grad_norm": 2.2626378536224365,
      "learning_rate": 2.995364704442541e-05,
      "loss": 1.1714,
      "step": 131600
    },
    {
      "epoch": 4.02333964685037,
      "grad_norm": 18.261306762695312,
      "learning_rate": 2.9938349039285278e-05,
      "loss": 0.9777,
      "step": 131700
    },
    {
      "epoch": 4.026394574448585,
      "grad_norm": 3.398371696472168,
      "learning_rate": 2.992305103414515e-05,
      "loss": 1.7022,
      "step": 131800
    },
    {
      "epoch": 4.029449502046801,
      "grad_norm": 10.634683609008789,
      "learning_rate": 2.990775302900502e-05,
      "loss": 1.3408,
      "step": 131900
    },
    {
      "epoch": 4.032504429645018,
      "grad_norm": 3.4038894176483154,
      "learning_rate": 2.989245502386489e-05,
      "loss": 1.0956,
      "step": 132000
    },
    {
      "epoch": 4.035559357243233,
      "grad_norm": 9.867512702941895,
      "learning_rate": 2.987715701872476e-05,
      "loss": 1.1892,
      "step": 132100
    },
    {
      "epoch": 4.038614284841449,
      "grad_norm": 12.254446983337402,
      "learning_rate": 2.986185901358463e-05,
      "loss": 1.3299,
      "step": 132200
    },
    {
      "epoch": 4.041669212439666,
      "grad_norm": 5.026933670043945,
      "learning_rate": 2.98465610084445e-05,
      "loss": 1.115,
      "step": 132300
    },
    {
      "epoch": 4.044724140037881,
      "grad_norm": 7.253444194793701,
      "learning_rate": 2.9831263003304373e-05,
      "loss": 1.225,
      "step": 132400
    },
    {
      "epoch": 4.047779067636097,
      "grad_norm": 1.1668365001678467,
      "learning_rate": 2.981596499816424e-05,
      "loss": 1.1519,
      "step": 132500
    },
    {
      "epoch": 4.050833995234313,
      "grad_norm": 2.0015876293182373,
      "learning_rate": 2.980066699302411e-05,
      "loss": 1.1844,
      "step": 132600
    },
    {
      "epoch": 4.053888922832529,
      "grad_norm": 1.1752444505691528,
      "learning_rate": 2.9785368987883983e-05,
      "loss": 1.1572,
      "step": 132700
    },
    {
      "epoch": 4.056943850430745,
      "grad_norm": 3.1533567905426025,
      "learning_rate": 2.9770070982743854e-05,
      "loss": 1.0884,
      "step": 132800
    },
    {
      "epoch": 4.059998778028961,
      "grad_norm": 8.84151840209961,
      "learning_rate": 2.9754772977603722e-05,
      "loss": 1.5823,
      "step": 132900
    },
    {
      "epoch": 4.063053705627176,
      "grad_norm": 5.8620500564575195,
      "learning_rate": 2.9739474972463593e-05,
      "loss": 1.169,
      "step": 133000
    },
    {
      "epoch": 4.066108633225393,
      "grad_norm": 7.552305221557617,
      "learning_rate": 2.9724176967323464e-05,
      "loss": 1.3024,
      "step": 133100
    },
    {
      "epoch": 4.0691635608236085,
      "grad_norm": 2.8654162883758545,
      "learning_rate": 2.9708878962183335e-05,
      "loss": 1.2343,
      "step": 133200
    },
    {
      "epoch": 4.072218488421824,
      "grad_norm": 4.553361892700195,
      "learning_rate": 2.9693580957043203e-05,
      "loss": 1.1538,
      "step": 133300
    },
    {
      "epoch": 4.07527341602004,
      "grad_norm": 4.676537036895752,
      "learning_rate": 2.9678282951903074e-05,
      "loss": 1.2307,
      "step": 133400
    },
    {
      "epoch": 4.0783283436182565,
      "grad_norm": 8.020150184631348,
      "learning_rate": 2.9662984946762945e-05,
      "loss": 1.1745,
      "step": 133500
    },
    {
      "epoch": 4.081383271216472,
      "grad_norm": 4.467293739318848,
      "learning_rate": 2.9647686941622816e-05,
      "loss": 1.0806,
      "step": 133600
    },
    {
      "epoch": 4.084438198814688,
      "grad_norm": 2.580444574356079,
      "learning_rate": 2.9632388936482684e-05,
      "loss": 1.1719,
      "step": 133700
    },
    {
      "epoch": 4.087493126412904,
      "grad_norm": 3.863105058670044,
      "learning_rate": 2.9617090931342555e-05,
      "loss": 1.02,
      "step": 133800
    },
    {
      "epoch": 4.09054805401112,
      "grad_norm": 3.3739333152770996,
      "learning_rate": 2.9601792926202427e-05,
      "loss": 1.2029,
      "step": 133900
    },
    {
      "epoch": 4.093602981609336,
      "grad_norm": 4.424749374389648,
      "learning_rate": 2.9586494921062298e-05,
      "loss": 1.2085,
      "step": 134000
    },
    {
      "epoch": 4.0966579092075515,
      "grad_norm": 3.976792812347412,
      "learning_rate": 2.9571196915922165e-05,
      "loss": 1.3111,
      "step": 134100
    },
    {
      "epoch": 4.099712836805768,
      "grad_norm": 9.701395034790039,
      "learning_rate": 2.9555898910782037e-05,
      "loss": 1.1155,
      "step": 134200
    },
    {
      "epoch": 4.102767764403984,
      "grad_norm": 0.8991170525550842,
      "learning_rate": 2.9540753885693308e-05,
      "loss": 1.0686,
      "step": 134300
    },
    {
      "epoch": 4.105822692002199,
      "grad_norm": 17.905302047729492,
      "learning_rate": 2.9525455880553176e-05,
      "loss": 1.0627,
      "step": 134400
    },
    {
      "epoch": 4.108877619600415,
      "grad_norm": 5.040673732757568,
      "learning_rate": 2.9510157875413047e-05,
      "loss": 1.0886,
      "step": 134500
    },
    {
      "epoch": 4.111932547198632,
      "grad_norm": 4.204395771026611,
      "learning_rate": 2.9494859870272918e-05,
      "loss": 1.238,
      "step": 134600
    },
    {
      "epoch": 4.114987474796847,
      "grad_norm": 3.686497449874878,
      "learning_rate": 2.947956186513279e-05,
      "loss": 1.2616,
      "step": 134700
    },
    {
      "epoch": 4.118042402395063,
      "grad_norm": 5.15822696685791,
      "learning_rate": 2.9464263859992657e-05,
      "loss": 1.2657,
      "step": 134800
    },
    {
      "epoch": 4.12109732999328,
      "grad_norm": 4.707155227661133,
      "learning_rate": 2.9448965854852528e-05,
      "loss": 1.2191,
      "step": 134900
    },
    {
      "epoch": 4.124152257591495,
      "grad_norm": 3.886612892150879,
      "learning_rate": 2.94336678497124e-05,
      "loss": 1.0955,
      "step": 135000
    },
    {
      "epoch": 4.127207185189711,
      "grad_norm": 2.287773609161377,
      "learning_rate": 2.941836984457227e-05,
      "loss": 1.189,
      "step": 135100
    },
    {
      "epoch": 4.130262112787927,
      "grad_norm": 4.35518741607666,
      "learning_rate": 2.9403071839432138e-05,
      "loss": 1.2268,
      "step": 135200
    },
    {
      "epoch": 4.133317040386143,
      "grad_norm": 8.079886436462402,
      "learning_rate": 2.938777383429201e-05,
      "loss": 1.0111,
      "step": 135300
    },
    {
      "epoch": 4.136371967984359,
      "grad_norm": 7.398697376251221,
      "learning_rate": 2.937247582915188e-05,
      "loss": 1.2892,
      "step": 135400
    },
    {
      "epoch": 4.139426895582575,
      "grad_norm": 2.7290868759155273,
      "learning_rate": 2.935717782401175e-05,
      "loss": 1.1484,
      "step": 135500
    },
    {
      "epoch": 4.14248182318079,
      "grad_norm": 7.36677885055542,
      "learning_rate": 2.934187981887162e-05,
      "loss": 1.0452,
      "step": 135600
    },
    {
      "epoch": 4.145536750779007,
      "grad_norm": 7.556665897369385,
      "learning_rate": 2.932658181373149e-05,
      "loss": 1.1462,
      "step": 135700
    },
    {
      "epoch": 4.1485916783772225,
      "grad_norm": 17.75133514404297,
      "learning_rate": 2.9311283808591362e-05,
      "loss": 1.1841,
      "step": 135800
    },
    {
      "epoch": 4.151646605975438,
      "grad_norm": 8.796733856201172,
      "learning_rate": 2.9295985803451233e-05,
      "loss": 1.2635,
      "step": 135900
    },
    {
      "epoch": 4.154701533573654,
      "grad_norm": 3.0543625354766846,
      "learning_rate": 2.92806877983111e-05,
      "loss": 1.1486,
      "step": 136000
    },
    {
      "epoch": 4.1577564611718705,
      "grad_norm": 8.879738807678223,
      "learning_rate": 2.9265389793170972e-05,
      "loss": 1.0926,
      "step": 136100
    },
    {
      "epoch": 4.160811388770086,
      "grad_norm": 5.291866302490234,
      "learning_rate": 2.9250091788030843e-05,
      "loss": 1.2567,
      "step": 136200
    },
    {
      "epoch": 4.163866316368302,
      "grad_norm": 11.352781295776367,
      "learning_rate": 2.9234793782890714e-05,
      "loss": 1.196,
      "step": 136300
    },
    {
      "epoch": 4.166921243966518,
      "grad_norm": 4.0807785987854,
      "learning_rate": 2.9219495777750582e-05,
      "loss": 1.018,
      "step": 136400
    },
    {
      "epoch": 4.169976171564734,
      "grad_norm": 7.902457237243652,
      "learning_rate": 2.9204197772610453e-05,
      "loss": 1.2478,
      "step": 136500
    },
    {
      "epoch": 4.17303109916295,
      "grad_norm": 6.9937357902526855,
      "learning_rate": 2.9188899767470324e-05,
      "loss": 1.2601,
      "step": 136600
    },
    {
      "epoch": 4.1760860267611655,
      "grad_norm": 19.437175750732422,
      "learning_rate": 2.9173601762330195e-05,
      "loss": 1.2579,
      "step": 136700
    },
    {
      "epoch": 4.179140954359382,
      "grad_norm": 2.6270205974578857,
      "learning_rate": 2.9158303757190063e-05,
      "loss": 1.1452,
      "step": 136800
    },
    {
      "epoch": 4.182195881957598,
      "grad_norm": 11.526654243469238,
      "learning_rate": 2.9143005752049934e-05,
      "loss": 1.4067,
      "step": 136900
    },
    {
      "epoch": 4.185250809555813,
      "grad_norm": 5.607306003570557,
      "learning_rate": 2.9127707746909805e-05,
      "loss": 1.3024,
      "step": 137000
    },
    {
      "epoch": 4.188305737154029,
      "grad_norm": 3.1491425037384033,
      "learning_rate": 2.9112409741769677e-05,
      "loss": 1.1115,
      "step": 137100
    },
    {
      "epoch": 4.191360664752246,
      "grad_norm": 7.519656658172607,
      "learning_rate": 2.9097111736629544e-05,
      "loss": 1.1044,
      "step": 137200
    },
    {
      "epoch": 4.194415592350461,
      "grad_norm": 15.954972267150879,
      "learning_rate": 2.9081813731489416e-05,
      "loss": 1.1498,
      "step": 137300
    },
    {
      "epoch": 4.197470519948677,
      "grad_norm": 1.4020683765411377,
      "learning_rate": 2.9066515726349287e-05,
      "loss": 1.1048,
      "step": 137400
    },
    {
      "epoch": 4.200525447546893,
      "grad_norm": 3.924858331680298,
      "learning_rate": 2.9051217721209158e-05,
      "loss": 1.0014,
      "step": 137500
    },
    {
      "epoch": 4.203580375145109,
      "grad_norm": 13.93403148651123,
      "learning_rate": 2.9035919716069026e-05,
      "loss": 1.1036,
      "step": 137600
    },
    {
      "epoch": 4.206635302743325,
      "grad_norm": 4.891828536987305,
      "learning_rate": 2.9020621710928897e-05,
      "loss": 1.0443,
      "step": 137700
    },
    {
      "epoch": 4.209690230341541,
      "grad_norm": 9.99404525756836,
      "learning_rate": 2.9005323705788768e-05,
      "loss": 1.0636,
      "step": 137800
    },
    {
      "epoch": 4.212745157939757,
      "grad_norm": 3.7061333656311035,
      "learning_rate": 2.899002570064864e-05,
      "loss": 1.105,
      "step": 137900
    },
    {
      "epoch": 4.215800085537973,
      "grad_norm": 9.68040657043457,
      "learning_rate": 2.8974727695508507e-05,
      "loss": 1.2866,
      "step": 138000
    },
    {
      "epoch": 4.218855013136189,
      "grad_norm": 1.8499339818954468,
      "learning_rate": 2.8959429690368378e-05,
      "loss": 0.9657,
      "step": 138100
    },
    {
      "epoch": 4.221909940734404,
      "grad_norm": 2.119901180267334,
      "learning_rate": 2.894413168522825e-05,
      "loss": 1.1226,
      "step": 138200
    },
    {
      "epoch": 4.224964868332621,
      "grad_norm": 7.395858287811279,
      "learning_rate": 2.8928986660139517e-05,
      "loss": 1.007,
      "step": 138300
    },
    {
      "epoch": 4.2280197959308365,
      "grad_norm": 2.304553270339966,
      "learning_rate": 2.8913688654999388e-05,
      "loss": 0.978,
      "step": 138400
    },
    {
      "epoch": 4.231074723529052,
      "grad_norm": 7.8232011795043945,
      "learning_rate": 2.889839064985926e-05,
      "loss": 1.0871,
      "step": 138500
    },
    {
      "epoch": 4.234129651127268,
      "grad_norm": 8.254796981811523,
      "learning_rate": 2.8883092644719127e-05,
      "loss": 1.0689,
      "step": 138600
    },
    {
      "epoch": 4.2371845787254845,
      "grad_norm": 6.827329635620117,
      "learning_rate": 2.8867794639579e-05,
      "loss": 1.2603,
      "step": 138700
    },
    {
      "epoch": 4.2402395063237,
      "grad_norm": 2.7104625701904297,
      "learning_rate": 2.885249663443887e-05,
      "loss": 1.0437,
      "step": 138800
    },
    {
      "epoch": 4.243294433921916,
      "grad_norm": 1.3501704931259155,
      "learning_rate": 2.883719862929874e-05,
      "loss": 0.941,
      "step": 138900
    },
    {
      "epoch": 4.2463493615201315,
      "grad_norm": 10.435741424560547,
      "learning_rate": 2.882190062415861e-05,
      "loss": 1.1538,
      "step": 139000
    },
    {
      "epoch": 4.249404289118348,
      "grad_norm": 6.998001575469971,
      "learning_rate": 2.880660261901848e-05,
      "loss": 1.2373,
      "step": 139100
    },
    {
      "epoch": 4.252459216716564,
      "grad_norm": 3.866121292114258,
      "learning_rate": 2.879130461387835e-05,
      "loss": 1.0814,
      "step": 139200
    },
    {
      "epoch": 4.2555141443147795,
      "grad_norm": 5.841191291809082,
      "learning_rate": 2.8776006608738222e-05,
      "loss": 1.1518,
      "step": 139300
    },
    {
      "epoch": 4.258569071912996,
      "grad_norm": 4.831516265869141,
      "learning_rate": 2.8760861583649497e-05,
      "loss": 1.3793,
      "step": 139400
    },
    {
      "epoch": 4.261623999511212,
      "grad_norm": 3.0584304332733154,
      "learning_rate": 2.8745563578509364e-05,
      "loss": 1.2819,
      "step": 139500
    },
    {
      "epoch": 4.264678927109427,
      "grad_norm": 9.81663703918457,
      "learning_rate": 2.8730265573369236e-05,
      "loss": 1.2213,
      "step": 139600
    },
    {
      "epoch": 4.267733854707643,
      "grad_norm": 3.883660078048706,
      "learning_rate": 2.8714967568229107e-05,
      "loss": 1.163,
      "step": 139700
    },
    {
      "epoch": 4.27078878230586,
      "grad_norm": 3.366514205932617,
      "learning_rate": 2.8699669563088978e-05,
      "loss": 1.0588,
      "step": 139800
    },
    {
      "epoch": 4.273843709904075,
      "grad_norm": 22.876943588256836,
      "learning_rate": 2.8684371557948846e-05,
      "loss": 0.9759,
      "step": 139900
    },
    {
      "epoch": 4.276898637502291,
      "grad_norm": 9.278138160705566,
      "learning_rate": 2.8669073552808717e-05,
      "loss": 1.2093,
      "step": 140000
    },
    {
      "epoch": 4.279953565100507,
      "grad_norm": 12.669641494750977,
      "learning_rate": 2.8653775547668588e-05,
      "loss": 1.2022,
      "step": 140100
    },
    {
      "epoch": 4.283008492698723,
      "grad_norm": 4.376775741577148,
      "learning_rate": 2.863847754252846e-05,
      "loss": 1.2715,
      "step": 140200
    },
    {
      "epoch": 4.286063420296939,
      "grad_norm": 6.5941386222839355,
      "learning_rate": 2.8623179537388327e-05,
      "loss": 1.1646,
      "step": 140300
    },
    {
      "epoch": 4.289118347895155,
      "grad_norm": 10.246788024902344,
      "learning_rate": 2.8607881532248198e-05,
      "loss": 1.3237,
      "step": 140400
    },
    {
      "epoch": 4.29217327549337,
      "grad_norm": 11.5577974319458,
      "learning_rate": 2.859258352710807e-05,
      "loss": 1.1616,
      "step": 140500
    },
    {
      "epoch": 4.295228203091587,
      "grad_norm": 8.360092163085938,
      "learning_rate": 2.857728552196794e-05,
      "loss": 1.0561,
      "step": 140600
    },
    {
      "epoch": 4.298283130689803,
      "grad_norm": 5.822038650512695,
      "learning_rate": 2.8561987516827808e-05,
      "loss": 1.3474,
      "step": 140700
    },
    {
      "epoch": 4.301338058288018,
      "grad_norm": 5.193514823913574,
      "learning_rate": 2.854668951168768e-05,
      "loss": 1.1984,
      "step": 140800
    },
    {
      "epoch": 4.304392985886235,
      "grad_norm": 10.879566192626953,
      "learning_rate": 2.853139150654755e-05,
      "loss": 1.2641,
      "step": 140900
    },
    {
      "epoch": 4.3074479134844506,
      "grad_norm": 0.4382971525192261,
      "learning_rate": 2.851609350140742e-05,
      "loss": 1.1111,
      "step": 141000
    },
    {
      "epoch": 4.310502841082666,
      "grad_norm": 6.622337341308594,
      "learning_rate": 2.850079549626729e-05,
      "loss": 1.0635,
      "step": 141100
    },
    {
      "epoch": 4.313557768680882,
      "grad_norm": 2.3514699935913086,
      "learning_rate": 2.848549749112716e-05,
      "loss": 1.0282,
      "step": 141200
    },
    {
      "epoch": 4.3166126962790985,
      "grad_norm": 2.973541021347046,
      "learning_rate": 2.8470199485987032e-05,
      "loss": 1.1509,
      "step": 141300
    },
    {
      "epoch": 4.319667623877314,
      "grad_norm": 8.881172180175781,
      "learning_rate": 2.8454901480846903e-05,
      "loss": 1.064,
      "step": 141400
    },
    {
      "epoch": 4.32272255147553,
      "grad_norm": 11.173826217651367,
      "learning_rate": 2.843960347570677e-05,
      "loss": 1.1981,
      "step": 141500
    },
    {
      "epoch": 4.3257774790737455,
      "grad_norm": 5.309777736663818,
      "learning_rate": 2.842430547056664e-05,
      "loss": 1.1205,
      "step": 141600
    },
    {
      "epoch": 4.328832406671962,
      "grad_norm": 10.873201370239258,
      "learning_rate": 2.8409007465426506e-05,
      "loss": 1.1992,
      "step": 141700
    },
    {
      "epoch": 4.331887334270178,
      "grad_norm": 5.490105152130127,
      "learning_rate": 2.8393709460286377e-05,
      "loss": 1.1684,
      "step": 141800
    },
    {
      "epoch": 4.3349422618683935,
      "grad_norm": 27.202516555786133,
      "learning_rate": 2.837841145514625e-05,
      "loss": 1.0327,
      "step": 141900
    },
    {
      "epoch": 4.337997189466609,
      "grad_norm": 6.939114093780518,
      "learning_rate": 2.836311345000612e-05,
      "loss": 1.0107,
      "step": 142000
    },
    {
      "epoch": 4.341052117064826,
      "grad_norm": 3.0641424655914307,
      "learning_rate": 2.8347815444865987e-05,
      "loss": 0.9114,
      "step": 142100
    },
    {
      "epoch": 4.344107044663041,
      "grad_norm": 6.63291072845459,
      "learning_rate": 2.833251743972586e-05,
      "loss": 1.1327,
      "step": 142200
    },
    {
      "epoch": 4.347161972261257,
      "grad_norm": 2.9434144496917725,
      "learning_rate": 2.831721943458573e-05,
      "loss": 1.1786,
      "step": 142300
    },
    {
      "epoch": 4.350216899859474,
      "grad_norm": 5.704064846038818,
      "learning_rate": 2.83019214294456e-05,
      "loss": 1.2337,
      "step": 142400
    },
    {
      "epoch": 4.353271827457689,
      "grad_norm": 7.911927223205566,
      "learning_rate": 2.828662342430547e-05,
      "loss": 1.1111,
      "step": 142500
    },
    {
      "epoch": 4.356326755055905,
      "grad_norm": 1.7703015804290771,
      "learning_rate": 2.827132541916534e-05,
      "loss": 1.2244,
      "step": 142600
    },
    {
      "epoch": 4.359381682654121,
      "grad_norm": 6.676158428192139,
      "learning_rate": 2.825602741402521e-05,
      "loss": 1.1667,
      "step": 142700
    },
    {
      "epoch": 4.362436610252337,
      "grad_norm": 21.077234268188477,
      "learning_rate": 2.8240729408885082e-05,
      "loss": 1.0365,
      "step": 142800
    },
    {
      "epoch": 4.365491537850553,
      "grad_norm": 10.691551208496094,
      "learning_rate": 2.822543140374495e-05,
      "loss": 1.1383,
      "step": 142900
    },
    {
      "epoch": 4.368546465448769,
      "grad_norm": 1.3159449100494385,
      "learning_rate": 2.821013339860482e-05,
      "loss": 1.1457,
      "step": 143000
    },
    {
      "epoch": 4.371601393046985,
      "grad_norm": 12.039605140686035,
      "learning_rate": 2.8194835393464692e-05,
      "loss": 1.2284,
      "step": 143100
    },
    {
      "epoch": 4.374656320645201,
      "grad_norm": 4.086942672729492,
      "learning_rate": 2.8179537388324563e-05,
      "loss": 1.1436,
      "step": 143200
    },
    {
      "epoch": 4.377711248243417,
      "grad_norm": 8.750964164733887,
      "learning_rate": 2.816423938318443e-05,
      "loss": 1.0047,
      "step": 143300
    },
    {
      "epoch": 4.380766175841632,
      "grad_norm": 14.825384140014648,
      "learning_rate": 2.8149094358095706e-05,
      "loss": 1.061,
      "step": 143400
    },
    {
      "epoch": 4.383821103439849,
      "grad_norm": 5.547531604766846,
      "learning_rate": 2.8133796352955577e-05,
      "loss": 1.1211,
      "step": 143500
    },
    {
      "epoch": 4.386876031038065,
      "grad_norm": 4.428119659423828,
      "learning_rate": 2.8118498347815448e-05,
      "loss": 1.2147,
      "step": 143600
    },
    {
      "epoch": 4.38993095863628,
      "grad_norm": 4.358971118927002,
      "learning_rate": 2.8103353322726716e-05,
      "loss": 1.0761,
      "step": 143700
    },
    {
      "epoch": 4.392985886234496,
      "grad_norm": 3.263248920440674,
      "learning_rate": 2.8088055317586587e-05,
      "loss": 1.0879,
      "step": 143800
    },
    {
      "epoch": 4.3960408138327125,
      "grad_norm": 6.146893501281738,
      "learning_rate": 2.807275731244646e-05,
      "loss": 1.2855,
      "step": 143900
    },
    {
      "epoch": 4.399095741430928,
      "grad_norm": 6.958042621612549,
      "learning_rate": 2.8057459307306326e-05,
      "loss": 1.2137,
      "step": 144000
    },
    {
      "epoch": 4.402150669029144,
      "grad_norm": 3.671699047088623,
      "learning_rate": 2.8042161302166197e-05,
      "loss": 1.0662,
      "step": 144100
    },
    {
      "epoch": 4.4052055966273596,
      "grad_norm": 6.236356258392334,
      "learning_rate": 2.802686329702607e-05,
      "loss": 1.4093,
      "step": 144200
    },
    {
      "epoch": 4.408260524225576,
      "grad_norm": 7.394711971282959,
      "learning_rate": 2.801156529188594e-05,
      "loss": 0.987,
      "step": 144300
    },
    {
      "epoch": 4.411315451823792,
      "grad_norm": 9.206113815307617,
      "learning_rate": 2.7996267286745807e-05,
      "loss": 0.9402,
      "step": 144400
    },
    {
      "epoch": 4.4143703794220075,
      "grad_norm": 6.848948955535889,
      "learning_rate": 2.798096928160568e-05,
      "loss": 1.343,
      "step": 144500
    },
    {
      "epoch": 4.417425307020224,
      "grad_norm": 6.24761962890625,
      "learning_rate": 2.796567127646555e-05,
      "loss": 1.1268,
      "step": 144600
    },
    {
      "epoch": 4.42048023461844,
      "grad_norm": 4.589752674102783,
      "learning_rate": 2.795037327132542e-05,
      "loss": 1.042,
      "step": 144700
    },
    {
      "epoch": 4.423535162216655,
      "grad_norm": 17.89345359802246,
      "learning_rate": 2.793507526618529e-05,
      "loss": 1.1777,
      "step": 144800
    },
    {
      "epoch": 4.426590089814871,
      "grad_norm": 5.196957111358643,
      "learning_rate": 2.791977726104516e-05,
      "loss": 1.2299,
      "step": 144900
    },
    {
      "epoch": 4.429645017413088,
      "grad_norm": 6.011660099029541,
      "learning_rate": 2.790447925590503e-05,
      "loss": 1.0815,
      "step": 145000
    },
    {
      "epoch": 4.432699945011303,
      "grad_norm": 8.668110847473145,
      "learning_rate": 2.7889181250764902e-05,
      "loss": 1.128,
      "step": 145100
    },
    {
      "epoch": 4.435754872609519,
      "grad_norm": 19.578367233276367,
      "learning_rate": 2.787388324562477e-05,
      "loss": 1.4063,
      "step": 145200
    },
    {
      "epoch": 4.438809800207735,
      "grad_norm": 7.321340084075928,
      "learning_rate": 2.785858524048464e-05,
      "loss": 1.1274,
      "step": 145300
    },
    {
      "epoch": 4.441864727805951,
      "grad_norm": 7.219663143157959,
      "learning_rate": 2.7843287235344512e-05,
      "loss": 1.1132,
      "step": 145400
    },
    {
      "epoch": 4.444919655404167,
      "grad_norm": 3.2994327545166016,
      "learning_rate": 2.7827989230204383e-05,
      "loss": 1.2619,
      "step": 145500
    },
    {
      "epoch": 4.447974583002383,
      "grad_norm": 9.860881805419922,
      "learning_rate": 2.781269122506425e-05,
      "loss": 1.2081,
      "step": 145600
    },
    {
      "epoch": 4.451029510600598,
      "grad_norm": 16.964540481567383,
      "learning_rate": 2.7797393219924122e-05,
      "loss": 1.201,
      "step": 145700
    },
    {
      "epoch": 4.454084438198815,
      "grad_norm": 13.067527770996094,
      "learning_rate": 2.7782095214783994e-05,
      "loss": 1.0872,
      "step": 145800
    },
    {
      "epoch": 4.457139365797031,
      "grad_norm": 2.4433538913726807,
      "learning_rate": 2.7766797209643865e-05,
      "loss": 1.4488,
      "step": 145900
    },
    {
      "epoch": 4.460194293395246,
      "grad_norm": 3.9496967792510986,
      "learning_rate": 2.7751499204503732e-05,
      "loss": 1.3713,
      "step": 146000
    },
    {
      "epoch": 4.463249220993463,
      "grad_norm": 2.2983593940734863,
      "learning_rate": 2.7736201199363604e-05,
      "loss": 1.2593,
      "step": 146100
    },
    {
      "epoch": 4.466304148591679,
      "grad_norm": 6.58331298828125,
      "learning_rate": 2.7720903194223475e-05,
      "loss": 1.032,
      "step": 146200
    },
    {
      "epoch": 4.469359076189894,
      "grad_norm": 9.05478286743164,
      "learning_rate": 2.7705605189083346e-05,
      "loss": 1.091,
      "step": 146300
    },
    {
      "epoch": 4.47241400378811,
      "grad_norm": 4.241959095001221,
      "learning_rate": 2.7690307183943214e-05,
      "loss": 1.0762,
      "step": 146400
    },
    {
      "epoch": 4.4754689313863265,
      "grad_norm": 19.84777069091797,
      "learning_rate": 2.7675009178803085e-05,
      "loss": 1.0127,
      "step": 146500
    },
    {
      "epoch": 4.478523858984542,
      "grad_norm": 3.5767288208007812,
      "learning_rate": 2.7659711173662956e-05,
      "loss": 1.1679,
      "step": 146600
    },
    {
      "epoch": 4.481578786582758,
      "grad_norm": 5.407430171966553,
      "learning_rate": 2.7644413168522827e-05,
      "loss": 1.1976,
      "step": 146700
    },
    {
      "epoch": 4.484633714180974,
      "grad_norm": 5.339792251586914,
      "learning_rate": 2.7629115163382695e-05,
      "loss": 1.1483,
      "step": 146800
    },
    {
      "epoch": 4.48768864177919,
      "grad_norm": 12.349178314208984,
      "learning_rate": 2.7613817158242566e-05,
      "loss": 1.2548,
      "step": 146900
    },
    {
      "epoch": 4.490743569377406,
      "grad_norm": 5.292774200439453,
      "learning_rate": 2.7598519153102437e-05,
      "loss": 1.0486,
      "step": 147000
    },
    {
      "epoch": 4.4937984969756215,
      "grad_norm": 6.7021894454956055,
      "learning_rate": 2.758322114796231e-05,
      "loss": 1.2699,
      "step": 147100
    },
    {
      "epoch": 4.496853424573837,
      "grad_norm": 5.217250823974609,
      "learning_rate": 2.7567923142822176e-05,
      "loss": 1.1054,
      "step": 147200
    },
    {
      "epoch": 4.499908352172054,
      "grad_norm": 5.183603286743164,
      "learning_rate": 2.7552625137682047e-05,
      "loss": 1.0157,
      "step": 147300
    },
    {
      "epoch": 4.5029632797702694,
      "grad_norm": 2.11531400680542,
      "learning_rate": 2.753732713254192e-05,
      "loss": 1.1512,
      "step": 147400
    },
    {
      "epoch": 4.506018207368485,
      "grad_norm": 2.366384983062744,
      "learning_rate": 2.752202912740179e-05,
      "loss": 1.0361,
      "step": 147500
    },
    {
      "epoch": 4.509073134966702,
      "grad_norm": 4.993153095245361,
      "learning_rate": 2.7506731122261657e-05,
      "loss": 1.1872,
      "step": 147600
    },
    {
      "epoch": 4.512128062564917,
      "grad_norm": 3.938058376312256,
      "learning_rate": 2.749158609717293e-05,
      "loss": 1.0318,
      "step": 147700
    },
    {
      "epoch": 4.515182990163133,
      "grad_norm": 5.502648830413818,
      "learning_rate": 2.7476288092032797e-05,
      "loss": 1.2964,
      "step": 147800
    },
    {
      "epoch": 4.518237917761349,
      "grad_norm": 8.001718521118164,
      "learning_rate": 2.746114306694407e-05,
      "loss": 0.9859,
      "step": 147900
    },
    {
      "epoch": 4.521292845359565,
      "grad_norm": 3.583967924118042,
      "learning_rate": 2.7445845061803942e-05,
      "loss": 1.1856,
      "step": 148000
    },
    {
      "epoch": 4.524347772957781,
      "grad_norm": 5.276466369628906,
      "learning_rate": 2.7430547056663814e-05,
      "loss": 1.2215,
      "step": 148100
    },
    {
      "epoch": 4.527402700555997,
      "grad_norm": 4.373233318328857,
      "learning_rate": 2.7415249051523685e-05,
      "loss": 1.1688,
      "step": 148200
    },
    {
      "epoch": 4.530457628154212,
      "grad_norm": 5.781352519989014,
      "learning_rate": 2.7399951046383553e-05,
      "loss": 1.1931,
      "step": 148300
    },
    {
      "epoch": 4.533512555752429,
      "grad_norm": 2.002715826034546,
      "learning_rate": 2.7384653041243424e-05,
      "loss": 1.3794,
      "step": 148400
    },
    {
      "epoch": 4.536567483350645,
      "grad_norm": 11.414073944091797,
      "learning_rate": 2.7369355036103295e-05,
      "loss": 1.4432,
      "step": 148500
    },
    {
      "epoch": 4.53962241094886,
      "grad_norm": 18.380279541015625,
      "learning_rate": 2.7354057030963166e-05,
      "loss": 1.1663,
      "step": 148600
    },
    {
      "epoch": 4.542677338547076,
      "grad_norm": 5.450042724609375,
      "learning_rate": 2.7338759025823034e-05,
      "loss": 1.0466,
      "step": 148700
    },
    {
      "epoch": 4.545732266145293,
      "grad_norm": 9.189093589782715,
      "learning_rate": 2.7323461020682905e-05,
      "loss": 1.0986,
      "step": 148800
    },
    {
      "epoch": 4.548787193743508,
      "grad_norm": 20.606910705566406,
      "learning_rate": 2.7308163015542776e-05,
      "loss": 1.1981,
      "step": 148900
    },
    {
      "epoch": 4.551842121341724,
      "grad_norm": 12.651390075683594,
      "learning_rate": 2.7292865010402647e-05,
      "loss": 1.2555,
      "step": 149000
    },
    {
      "epoch": 4.5548970489399405,
      "grad_norm": 4.8047895431518555,
      "learning_rate": 2.7277567005262515e-05,
      "loss": 0.9693,
      "step": 149100
    },
    {
      "epoch": 4.557951976538156,
      "grad_norm": 3.2293930053710938,
      "learning_rate": 2.7262269000122386e-05,
      "loss": 1.0667,
      "step": 149200
    },
    {
      "epoch": 4.561006904136372,
      "grad_norm": 7.451018810272217,
      "learning_rate": 2.7246970994982257e-05,
      "loss": 1.2841,
      "step": 149300
    },
    {
      "epoch": 4.564061831734588,
      "grad_norm": 3.7198963165283203,
      "learning_rate": 2.723167298984213e-05,
      "loss": 1.3827,
      "step": 149400
    },
    {
      "epoch": 4.567116759332804,
      "grad_norm": 9.99906063079834,
      "learning_rate": 2.7216374984701996e-05,
      "loss": 1.1345,
      "step": 149500
    },
    {
      "epoch": 4.57017168693102,
      "grad_norm": 5.092084884643555,
      "learning_rate": 2.7201076979561867e-05,
      "loss": 0.9375,
      "step": 149600
    },
    {
      "epoch": 4.5732266145292355,
      "grad_norm": 3.065910816192627,
      "learning_rate": 2.718577897442174e-05,
      "loss": 1.0997,
      "step": 149700
    },
    {
      "epoch": 4.576281542127451,
      "grad_norm": 4.881350517272949,
      "learning_rate": 2.717048096928161e-05,
      "loss": 1.0738,
      "step": 149800
    },
    {
      "epoch": 4.579336469725668,
      "grad_norm": 7.635152339935303,
      "learning_rate": 2.7155182964141477e-05,
      "loss": 1.2249,
      "step": 149900
    },
    {
      "epoch": 4.5823913973238835,
      "grad_norm": 4.8124098777771,
      "learning_rate": 2.713988495900135e-05,
      "loss": 1.0719,
      "step": 150000
    },
    {
      "epoch": 4.585446324922099,
      "grad_norm": 3.0546202659606934,
      "learning_rate": 2.712458695386122e-05,
      "loss": 1.2786,
      "step": 150100
    },
    {
      "epoch": 4.588501252520315,
      "grad_norm": 2.5182888507843018,
      "learning_rate": 2.710928894872109e-05,
      "loss": 0.9895,
      "step": 150200
    },
    {
      "epoch": 4.591556180118531,
      "grad_norm": 14.798208236694336,
      "learning_rate": 2.709399094358096e-05,
      "loss": 1.2199,
      "step": 150300
    },
    {
      "epoch": 4.594611107716747,
      "grad_norm": 7.6388983726501465,
      "learning_rate": 2.707869293844083e-05,
      "loss": 1.0752,
      "step": 150400
    },
    {
      "epoch": 4.597666035314963,
      "grad_norm": 11.475516319274902,
      "learning_rate": 2.70633949333007e-05,
      "loss": 1.2698,
      "step": 150500
    },
    {
      "epoch": 4.600720962913179,
      "grad_norm": 19.168842315673828,
      "learning_rate": 2.7048096928160572e-05,
      "loss": 1.3125,
      "step": 150600
    },
    {
      "epoch": 4.603775890511395,
      "grad_norm": 3.9104790687561035,
      "learning_rate": 2.703279892302044e-05,
      "loss": 1.0883,
      "step": 150700
    },
    {
      "epoch": 4.606830818109611,
      "grad_norm": 3.8482766151428223,
      "learning_rate": 2.701750091788031e-05,
      "loss": 1.1854,
      "step": 150800
    },
    {
      "epoch": 4.609885745707826,
      "grad_norm": 12.52638053894043,
      "learning_rate": 2.7002202912740182e-05,
      "loss": 1.0693,
      "step": 150900
    },
    {
      "epoch": 4.612940673306043,
      "grad_norm": 4.986527919769287,
      "learning_rate": 2.6986904907600053e-05,
      "loss": 1.1956,
      "step": 151000
    },
    {
      "epoch": 4.615995600904259,
      "grad_norm": 4.222838878631592,
      "learning_rate": 2.697160690245992e-05,
      "loss": 1.1175,
      "step": 151100
    },
    {
      "epoch": 4.619050528502474,
      "grad_norm": 5.308711051940918,
      "learning_rate": 2.6956308897319792e-05,
      "loss": 1.0547,
      "step": 151200
    },
    {
      "epoch": 4.622105456100691,
      "grad_norm": 4.313049793243408,
      "learning_rate": 2.6941010892179663e-05,
      "loss": 1.1143,
      "step": 151300
    },
    {
      "epoch": 4.625160383698907,
      "grad_norm": 6.646529197692871,
      "learning_rate": 2.6925712887039535e-05,
      "loss": 1.2822,
      "step": 151400
    },
    {
      "epoch": 4.628215311297122,
      "grad_norm": 17.885494232177734,
      "learning_rate": 2.6910414881899402e-05,
      "loss": 1.2093,
      "step": 151500
    },
    {
      "epoch": 4.631270238895338,
      "grad_norm": 4.94324254989624,
      "learning_rate": 2.6895269856810674e-05,
      "loss": 1.4626,
      "step": 151600
    },
    {
      "epoch": 4.634325166493554,
      "grad_norm": 4.033757209777832,
      "learning_rate": 2.6879971851670545e-05,
      "loss": 1.1175,
      "step": 151700
    },
    {
      "epoch": 4.63738009409177,
      "grad_norm": 6.608712673187256,
      "learning_rate": 2.6864673846530413e-05,
      "loss": 1.1387,
      "step": 151800
    },
    {
      "epoch": 4.640435021689986,
      "grad_norm": 9.266666412353516,
      "learning_rate": 2.6849375841390284e-05,
      "loss": 1.219,
      "step": 151900
    },
    {
      "epoch": 4.643489949288202,
      "grad_norm": 3.2637996673583984,
      "learning_rate": 2.6834077836250155e-05,
      "loss": 1.0627,
      "step": 152000
    },
    {
      "epoch": 4.646544876886418,
      "grad_norm": 5.220635414123535,
      "learning_rate": 2.6818779831110026e-05,
      "loss": 1.0148,
      "step": 152100
    },
    {
      "epoch": 4.649599804484634,
      "grad_norm": 4.843519687652588,
      "learning_rate": 2.6803481825969894e-05,
      "loss": 1.1965,
      "step": 152200
    },
    {
      "epoch": 4.6526547320828495,
      "grad_norm": 5.011851787567139,
      "learning_rate": 2.6788183820829765e-05,
      "loss": 1.1357,
      "step": 152300
    },
    {
      "epoch": 4.655709659681065,
      "grad_norm": 5.829235076904297,
      "learning_rate": 2.6772885815689636e-05,
      "loss": 1.0582,
      "step": 152400
    },
    {
      "epoch": 4.658764587279282,
      "grad_norm": 8.131631851196289,
      "learning_rate": 2.6757587810549507e-05,
      "loss": 1.2599,
      "step": 152500
    },
    {
      "epoch": 4.6618195148774975,
      "grad_norm": 3.51049542427063,
      "learning_rate": 2.6742289805409375e-05,
      "loss": 1.1366,
      "step": 152600
    },
    {
      "epoch": 4.664874442475713,
      "grad_norm": 2.632659912109375,
      "learning_rate": 2.6726991800269246e-05,
      "loss": 1.1617,
      "step": 152700
    },
    {
      "epoch": 4.66792937007393,
      "grad_norm": 2.5496699810028076,
      "learning_rate": 2.6711693795129117e-05,
      "loss": 0.9997,
      "step": 152800
    },
    {
      "epoch": 4.670984297672145,
      "grad_norm": 5.918241500854492,
      "learning_rate": 2.669639578998899e-05,
      "loss": 0.9468,
      "step": 152900
    },
    {
      "epoch": 4.674039225270361,
      "grad_norm": 2.7317497730255127,
      "learning_rate": 2.6681097784848856e-05,
      "loss": 1.1179,
      "step": 153000
    },
    {
      "epoch": 4.677094152868577,
      "grad_norm": 25.66161346435547,
      "learning_rate": 2.6665799779708728e-05,
      "loss": 1.2545,
      "step": 153100
    },
    {
      "epoch": 4.6801490804667925,
      "grad_norm": 14.719120025634766,
      "learning_rate": 2.66505017745686e-05,
      "loss": 1.1973,
      "step": 153200
    },
    {
      "epoch": 4.683204008065009,
      "grad_norm": 9.265717506408691,
      "learning_rate": 2.663520376942847e-05,
      "loss": 1.0308,
      "step": 153300
    },
    {
      "epoch": 4.686258935663225,
      "grad_norm": 0.9879122376441956,
      "learning_rate": 2.6619905764288338e-05,
      "loss": 1.3015,
      "step": 153400
    },
    {
      "epoch": 4.68931386326144,
      "grad_norm": 8.03515625,
      "learning_rate": 2.660460775914821e-05,
      "loss": 1.0214,
      "step": 153500
    },
    {
      "epoch": 4.692368790859657,
      "grad_norm": 21.492252349853516,
      "learning_rate": 2.658930975400808e-05,
      "loss": 1.1221,
      "step": 153600
    },
    {
      "epoch": 4.695423718457873,
      "grad_norm": 2.753187894821167,
      "learning_rate": 2.657401174886795e-05,
      "loss": 1.1131,
      "step": 153700
    },
    {
      "epoch": 4.698478646056088,
      "grad_norm": 7.463822364807129,
      "learning_rate": 2.655871374372782e-05,
      "loss": 0.9914,
      "step": 153800
    },
    {
      "epoch": 4.701533573654304,
      "grad_norm": 3.7811901569366455,
      "learning_rate": 2.654341573858769e-05,
      "loss": 1.1621,
      "step": 153900
    },
    {
      "epoch": 4.704588501252521,
      "grad_norm": 5.818233013153076,
      "learning_rate": 2.652811773344756e-05,
      "loss": 1.0053,
      "step": 154000
    },
    {
      "epoch": 4.707643428850736,
      "grad_norm": 11.87837028503418,
      "learning_rate": 2.651297270835883e-05,
      "loss": 1.2525,
      "step": 154100
    },
    {
      "epoch": 4.710698356448952,
      "grad_norm": 6.573636531829834,
      "learning_rate": 2.64976747032187e-05,
      "loss": 0.9842,
      "step": 154200
    },
    {
      "epoch": 4.7137532840471685,
      "grad_norm": 4.809690952301025,
      "learning_rate": 2.648237669807857e-05,
      "loss": 1.0407,
      "step": 154300
    },
    {
      "epoch": 4.716808211645384,
      "grad_norm": 1.3159583806991577,
      "learning_rate": 2.646707869293844e-05,
      "loss": 0.9382,
      "step": 154400
    },
    {
      "epoch": 4.7198631392436,
      "grad_norm": 4.047608375549316,
      "learning_rate": 2.645178068779831e-05,
      "loss": 1.172,
      "step": 154500
    },
    {
      "epoch": 4.722918066841816,
      "grad_norm": 7.231751441955566,
      "learning_rate": 2.643648268265818e-05,
      "loss": 1.1552,
      "step": 154600
    },
    {
      "epoch": 4.725972994440031,
      "grad_norm": 0.9529539942741394,
      "learning_rate": 2.6421184677518053e-05,
      "loss": 1.1272,
      "step": 154700
    },
    {
      "epoch": 4.729027922038248,
      "grad_norm": 2.781122922897339,
      "learning_rate": 2.640588667237792e-05,
      "loss": 1.0022,
      "step": 154800
    },
    {
      "epoch": 4.7320828496364635,
      "grad_norm": 2.895209312438965,
      "learning_rate": 2.639058866723779e-05,
      "loss": 1.3872,
      "step": 154900
    },
    {
      "epoch": 4.735137777234679,
      "grad_norm": 12.057914733886719,
      "learning_rate": 2.6375290662097663e-05,
      "loss": 1.2874,
      "step": 155000
    },
    {
      "epoch": 4.738192704832896,
      "grad_norm": 4.3409881591796875,
      "learning_rate": 2.6359992656957534e-05,
      "loss": 1.4921,
      "step": 155100
    },
    {
      "epoch": 4.7412476324311115,
      "grad_norm": 4.266219139099121,
      "learning_rate": 2.6344694651817402e-05,
      "loss": 1.0867,
      "step": 155200
    },
    {
      "epoch": 4.744302560029327,
      "grad_norm": 4.5013427734375,
      "learning_rate": 2.6329396646677273e-05,
      "loss": 1.1747,
      "step": 155300
    },
    {
      "epoch": 4.747357487627543,
      "grad_norm": 4.732046604156494,
      "learning_rate": 2.6314098641537144e-05,
      "loss": 1.1971,
      "step": 155400
    },
    {
      "epoch": 4.750412415225759,
      "grad_norm": 3.806830883026123,
      "learning_rate": 2.6298800636397015e-05,
      "loss": 1.2849,
      "step": 155500
    },
    {
      "epoch": 4.753467342823975,
      "grad_norm": 14.340802192687988,
      "learning_rate": 2.6283502631256883e-05,
      "loss": 1.0664,
      "step": 155600
    },
    {
      "epoch": 4.756522270422191,
      "grad_norm": 9.971649169921875,
      "learning_rate": 2.6268204626116754e-05,
      "loss": 1.0028,
      "step": 155700
    },
    {
      "epoch": 4.759577198020407,
      "grad_norm": 9.331883430480957,
      "learning_rate": 2.6252906620976625e-05,
      "loss": 1.2983,
      "step": 155800
    },
    {
      "epoch": 4.762632125618623,
      "grad_norm": 7.091375350952148,
      "learning_rate": 2.6237608615836496e-05,
      "loss": 1.1806,
      "step": 155900
    },
    {
      "epoch": 4.765687053216839,
      "grad_norm": 8.6240234375,
      "learning_rate": 2.6222310610696364e-05,
      "loss": 1.1379,
      "step": 156000
    },
    {
      "epoch": 4.768741980815054,
      "grad_norm": 0.5839551091194153,
      "learning_rate": 2.6207012605556235e-05,
      "loss": 1.2997,
      "step": 156100
    },
    {
      "epoch": 4.771796908413271,
      "grad_norm": 1.799262285232544,
      "learning_rate": 2.6191714600416107e-05,
      "loss": 1.3177,
      "step": 156200
    },
    {
      "epoch": 4.774851836011487,
      "grad_norm": 7.29044771194458,
      "learning_rate": 2.6176416595275978e-05,
      "loss": 1.3113,
      "step": 156300
    },
    {
      "epoch": 4.777906763609702,
      "grad_norm": 7.393566131591797,
      "learning_rate": 2.6161118590135845e-05,
      "loss": 1.0873,
      "step": 156400
    },
    {
      "epoch": 4.780961691207918,
      "grad_norm": 5.538482189178467,
      "learning_rate": 2.6145820584995717e-05,
      "loss": 1.0264,
      "step": 156500
    },
    {
      "epoch": 4.784016618806135,
      "grad_norm": 4.523062705993652,
      "learning_rate": 2.6130522579855588e-05,
      "loss": 1.0554,
      "step": 156600
    },
    {
      "epoch": 4.78707154640435,
      "grad_norm": 6.2098894119262695,
      "learning_rate": 2.611522457471546e-05,
      "loss": 1.2072,
      "step": 156700
    },
    {
      "epoch": 4.790126474002566,
      "grad_norm": 5.000922203063965,
      "learning_rate": 2.6099926569575327e-05,
      "loss": 1.0549,
      "step": 156800
    },
    {
      "epoch": 4.793181401600782,
      "grad_norm": 5.6592936515808105,
      "learning_rate": 2.6084628564435198e-05,
      "loss": 1.0814,
      "step": 156900
    },
    {
      "epoch": 4.796236329198998,
      "grad_norm": 11.27538776397705,
      "learning_rate": 2.606933055929507e-05,
      "loss": 1.1692,
      "step": 157000
    },
    {
      "epoch": 4.799291256797214,
      "grad_norm": 9.253680229187012,
      "learning_rate": 2.605403255415494e-05,
      "loss": 1.147,
      "step": 157100
    },
    {
      "epoch": 4.80234618439543,
      "grad_norm": 1.797663927078247,
      "learning_rate": 2.6038734549014808e-05,
      "loss": 1.1446,
      "step": 157200
    },
    {
      "epoch": 4.805401111993646,
      "grad_norm": 4.268359184265137,
      "learning_rate": 2.602343654387468e-05,
      "loss": 1.2718,
      "step": 157300
    },
    {
      "epoch": 4.808456039591862,
      "grad_norm": 12.612486839294434,
      "learning_rate": 2.600813853873455e-05,
      "loss": 1.294,
      "step": 157400
    },
    {
      "epoch": 4.8115109671900775,
      "grad_norm": 3.7770674228668213,
      "learning_rate": 2.599284053359442e-05,
      "loss": 0.9739,
      "step": 157500
    },
    {
      "epoch": 4.814565894788293,
      "grad_norm": 6.431491851806641,
      "learning_rate": 2.597754252845429e-05,
      "loss": 0.9815,
      "step": 157600
    },
    {
      "epoch": 4.81762082238651,
      "grad_norm": 6.823291301727295,
      "learning_rate": 2.596239750336556e-05,
      "loss": 1.265,
      "step": 157700
    },
    {
      "epoch": 4.8206757499847255,
      "grad_norm": 4.733063697814941,
      "learning_rate": 2.5947099498225432e-05,
      "loss": 1.0081,
      "step": 157800
    },
    {
      "epoch": 4.823730677582941,
      "grad_norm": 8.048057556152344,
      "learning_rate": 2.59318014930853e-05,
      "loss": 0.9448,
      "step": 157900
    },
    {
      "epoch": 4.826785605181157,
      "grad_norm": 5.712404251098633,
      "learning_rate": 2.591650348794517e-05,
      "loss": 1.1391,
      "step": 158000
    },
    {
      "epoch": 4.829840532779373,
      "grad_norm": 20.968780517578125,
      "learning_rate": 2.5901205482805042e-05,
      "loss": 1.1777,
      "step": 158100
    },
    {
      "epoch": 4.832895460377589,
      "grad_norm": 27.42873191833496,
      "learning_rate": 2.5885907477664913e-05,
      "loss": 1.2007,
      "step": 158200
    },
    {
      "epoch": 4.835950387975805,
      "grad_norm": 13.017873764038086,
      "learning_rate": 2.587060947252478e-05,
      "loss": 1.0182,
      "step": 158300
    },
    {
      "epoch": 4.8390053155740205,
      "grad_norm": 5.933872222900391,
      "learning_rate": 2.5855311467384652e-05,
      "loss": 0.9818,
      "step": 158400
    },
    {
      "epoch": 4.842060243172237,
      "grad_norm": 6.615920543670654,
      "learning_rate": 2.5840013462244523e-05,
      "loss": 1.0567,
      "step": 158500
    },
    {
      "epoch": 4.845115170770453,
      "grad_norm": 5.392825126647949,
      "learning_rate": 2.5824715457104394e-05,
      "loss": 1.3925,
      "step": 158600
    },
    {
      "epoch": 4.848170098368668,
      "grad_norm": 4.545580863952637,
      "learning_rate": 2.5809417451964262e-05,
      "loss": 1.0607,
      "step": 158700
    },
    {
      "epoch": 4.851225025966885,
      "grad_norm": 3.597367286682129,
      "learning_rate": 2.5794119446824133e-05,
      "loss": 1.1868,
      "step": 158800
    },
    {
      "epoch": 4.854279953565101,
      "grad_norm": 11.348734855651855,
      "learning_rate": 2.5778821441684004e-05,
      "loss": 1.0758,
      "step": 158900
    },
    {
      "epoch": 4.857334881163316,
      "grad_norm": 4.201880931854248,
      "learning_rate": 2.5763523436543875e-05,
      "loss": 1.2604,
      "step": 159000
    },
    {
      "epoch": 4.860389808761532,
      "grad_norm": 2.7427544593811035,
      "learning_rate": 2.5748225431403743e-05,
      "loss": 1.1255,
      "step": 159100
    },
    {
      "epoch": 4.863444736359749,
      "grad_norm": 7.285526275634766,
      "learning_rate": 2.5732927426263614e-05,
      "loss": 0.9783,
      "step": 159200
    },
    {
      "epoch": 4.866499663957964,
      "grad_norm": 3.2996041774749756,
      "learning_rate": 2.5717629421123485e-05,
      "loss": 1.1064,
      "step": 159300
    },
    {
      "epoch": 4.86955459155618,
      "grad_norm": 6.808858394622803,
      "learning_rate": 2.5702331415983357e-05,
      "loss": 1.2622,
      "step": 159400
    },
    {
      "epoch": 4.872609519154396,
      "grad_norm": 10.930204391479492,
      "learning_rate": 2.5687033410843224e-05,
      "loss": 1.091,
      "step": 159500
    },
    {
      "epoch": 4.875664446752612,
      "grad_norm": 3.2368526458740234,
      "learning_rate": 2.5671735405703096e-05,
      "loss": 1.0833,
      "step": 159600
    },
    {
      "epoch": 4.878719374350828,
      "grad_norm": 13.43560791015625,
      "learning_rate": 2.5656437400562967e-05,
      "loss": 1.1661,
      "step": 159700
    },
    {
      "epoch": 4.881774301949044,
      "grad_norm": 3.7159016132354736,
      "learning_rate": 2.5641139395422838e-05,
      "loss": 1.2862,
      "step": 159800
    },
    {
      "epoch": 4.884829229547259,
      "grad_norm": 3.5872607231140137,
      "learning_rate": 2.5625841390282706e-05,
      "loss": 1.105,
      "step": 159900
    },
    {
      "epoch": 4.887884157145476,
      "grad_norm": 5.785370826721191,
      "learning_rate": 2.5610543385142577e-05,
      "loss": 1.0201,
      "step": 160000
    },
    {
      "epoch": 4.8909390847436915,
      "grad_norm": 3.647353172302246,
      "learning_rate": 2.5595245380002448e-05,
      "loss": 1.1966,
      "step": 160100
    },
    {
      "epoch": 4.893994012341907,
      "grad_norm": 4.53983211517334,
      "learning_rate": 2.557994737486232e-05,
      "loss": 1.1269,
      "step": 160200
    },
    {
      "epoch": 4.897048939940124,
      "grad_norm": 0.30907294154167175,
      "learning_rate": 2.5564649369722187e-05,
      "loss": 1.1997,
      "step": 160300
    },
    {
      "epoch": 4.9001038675383395,
      "grad_norm": 3.865354299545288,
      "learning_rate": 2.5549351364582058e-05,
      "loss": 1.1692,
      "step": 160400
    },
    {
      "epoch": 4.903158795136555,
      "grad_norm": 11.909504890441895,
      "learning_rate": 2.553405335944193e-05,
      "loss": 1.1911,
      "step": 160500
    },
    {
      "epoch": 4.906213722734771,
      "grad_norm": 4.162954330444336,
      "learning_rate": 2.55187553543018e-05,
      "loss": 1.2396,
      "step": 160600
    },
    {
      "epoch": 4.909268650332987,
      "grad_norm": 5.563848972320557,
      "learning_rate": 2.5503457349161668e-05,
      "loss": 1.221,
      "step": 160700
    },
    {
      "epoch": 4.912323577931203,
      "grad_norm": 3.500187397003174,
      "learning_rate": 2.548815934402154e-05,
      "loss": 1.1687,
      "step": 160800
    },
    {
      "epoch": 4.915378505529419,
      "grad_norm": 11.680179595947266,
      "learning_rate": 2.547286133888141e-05,
      "loss": 1.4372,
      "step": 160900
    },
    {
      "epoch": 4.918433433127635,
      "grad_norm": 2.8160529136657715,
      "learning_rate": 2.545756333374128e-05,
      "loss": 1.513,
      "step": 161000
    },
    {
      "epoch": 4.921488360725851,
      "grad_norm": 9.372578620910645,
      "learning_rate": 2.544226532860115e-05,
      "loss": 1.1128,
      "step": 161100
    },
    {
      "epoch": 4.924543288324067,
      "grad_norm": 4.864928722381592,
      "learning_rate": 2.542696732346102e-05,
      "loss": 1.397,
      "step": 161200
    },
    {
      "epoch": 4.927598215922282,
      "grad_norm": 4.439268589019775,
      "learning_rate": 2.541166931832089e-05,
      "loss": 1.0441,
      "step": 161300
    },
    {
      "epoch": 4.930653143520498,
      "grad_norm": 29.387102127075195,
      "learning_rate": 2.5396524293232166e-05,
      "loss": 1.4971,
      "step": 161400
    },
    {
      "epoch": 4.933708071118715,
      "grad_norm": 5.6792449951171875,
      "learning_rate": 2.5381226288092038e-05,
      "loss": 1.0976,
      "step": 161500
    },
    {
      "epoch": 4.93676299871693,
      "grad_norm": 4.388849258422852,
      "learning_rate": 2.5365928282951905e-05,
      "loss": 1.2655,
      "step": 161600
    },
    {
      "epoch": 4.939817926315146,
      "grad_norm": 4.719710350036621,
      "learning_rate": 2.5350630277811777e-05,
      "loss": 1.4667,
      "step": 161700
    },
    {
      "epoch": 4.942872853913363,
      "grad_norm": 4.246535301208496,
      "learning_rate": 2.5335332272671648e-05,
      "loss": 1.1536,
      "step": 161800
    },
    {
      "epoch": 4.945927781511578,
      "grad_norm": 8.558767318725586,
      "learning_rate": 2.532003426753152e-05,
      "loss": 1.1652,
      "step": 161900
    },
    {
      "epoch": 4.948982709109794,
      "grad_norm": 5.678195476531982,
      "learning_rate": 2.5304736262391387e-05,
      "loss": 1.2042,
      "step": 162000
    },
    {
      "epoch": 4.95203763670801,
      "grad_norm": 9.030732154846191,
      "learning_rate": 2.5289438257251258e-05,
      "loss": 1.1497,
      "step": 162100
    },
    {
      "epoch": 4.955092564306226,
      "grad_norm": 2.4529170989990234,
      "learning_rate": 2.527414025211113e-05,
      "loss": 1.3533,
      "step": 162200
    },
    {
      "epoch": 4.958147491904442,
      "grad_norm": 6.64496374130249,
      "learning_rate": 2.5258842246971e-05,
      "loss": 1.2636,
      "step": 162300
    },
    {
      "epoch": 4.961202419502658,
      "grad_norm": 10.798649787902832,
      "learning_rate": 2.5243544241830868e-05,
      "loss": 1.2472,
      "step": 162400
    },
    {
      "epoch": 4.964257347100874,
      "grad_norm": 4.542051315307617,
      "learning_rate": 2.522824623669074e-05,
      "loss": 1.2845,
      "step": 162500
    },
    {
      "epoch": 4.96731227469909,
      "grad_norm": 9.684645652770996,
      "learning_rate": 2.521294823155061e-05,
      "loss": 1.1378,
      "step": 162600
    },
    {
      "epoch": 4.9703672022973056,
      "grad_norm": 2.868638753890991,
      "learning_rate": 2.519765022641048e-05,
      "loss": 1.0938,
      "step": 162700
    },
    {
      "epoch": 4.973422129895521,
      "grad_norm": 12.990683555603027,
      "learning_rate": 2.518235222127035e-05,
      "loss": 1.1421,
      "step": 162800
    },
    {
      "epoch": 4.976477057493737,
      "grad_norm": 5.3151140213012695,
      "learning_rate": 2.516705421613022e-05,
      "loss": 1.3563,
      "step": 162900
    },
    {
      "epoch": 4.9795319850919535,
      "grad_norm": 5.537310600280762,
      "learning_rate": 2.515175621099009e-05,
      "loss": 0.9885,
      "step": 163000
    },
    {
      "epoch": 4.982586912690169,
      "grad_norm": 6.696578025817871,
      "learning_rate": 2.5136458205849963e-05,
      "loss": 1.2855,
      "step": 163100
    },
    {
      "epoch": 4.985641840288385,
      "grad_norm": 6.6071648597717285,
      "learning_rate": 2.512116020070983e-05,
      "loss": 1.5194,
      "step": 163200
    },
    {
      "epoch": 4.988696767886601,
      "grad_norm": 3.9273293018341064,
      "learning_rate": 2.51058621955697e-05,
      "loss": 1.1475,
      "step": 163300
    },
    {
      "epoch": 4.991751695484817,
      "grad_norm": 6.573533535003662,
      "learning_rate": 2.5090564190429573e-05,
      "loss": 1.0817,
      "step": 163400
    },
    {
      "epoch": 4.994806623083033,
      "grad_norm": 6.224365234375,
      "learning_rate": 2.5075266185289444e-05,
      "loss": 1.2204,
      "step": 163500
    },
    {
      "epoch": 4.9978615506812485,
      "grad_norm": 8.9005765914917,
      "learning_rate": 2.505996818014931e-05,
      "loss": 1.2897,
      "step": 163600
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.7251787132644957,
      "eval_loss": 0.7394196391105652,
      "eval_runtime": 1778.1354,
      "eval_samples_per_second": 18.409,
      "eval_steps_per_second": 4.603,
      "step": 163670
    },
    {
      "epoch": 5.000916478279465,
      "grad_norm": 9.13719654083252,
      "learning_rate": 2.5044670175009183e-05,
      "loss": 1.0322,
      "step": 163700
    },
    {
      "epoch": 5.003971405877681,
      "grad_norm": 3.0623581409454346,
      "learning_rate": 2.5029372169869047e-05,
      "loss": 1.1183,
      "step": 163800
    },
    {
      "epoch": 5.007026333475896,
      "grad_norm": 9.058491706848145,
      "learning_rate": 2.5014074164728918e-05,
      "loss": 1.2189,
      "step": 163900
    },
    {
      "epoch": 5.010081261074112,
      "grad_norm": 17.310258865356445,
      "learning_rate": 2.499877615958879e-05,
      "loss": 1.1327,
      "step": 164000
    },
    {
      "epoch": 5.013136188672329,
      "grad_norm": 2.7564404010772705,
      "learning_rate": 2.498347815444866e-05,
      "loss": 1.104,
      "step": 164100
    },
    {
      "epoch": 5.016191116270544,
      "grad_norm": 13.609418869018555,
      "learning_rate": 2.4968180149308532e-05,
      "loss": 1.3047,
      "step": 164200
    },
    {
      "epoch": 5.01924604386876,
      "grad_norm": 5.303647041320801,
      "learning_rate": 2.4952882144168403e-05,
      "loss": 1.0716,
      "step": 164300
    },
    {
      "epoch": 5.022300971466977,
      "grad_norm": 3.680124282836914,
      "learning_rate": 2.493758413902827e-05,
      "loss": 1.4862,
      "step": 164400
    },
    {
      "epoch": 5.025355899065192,
      "grad_norm": 5.9867377281188965,
      "learning_rate": 2.4922286133888142e-05,
      "loss": 1.2148,
      "step": 164500
    },
    {
      "epoch": 5.028410826663408,
      "grad_norm": 4.398543834686279,
      "learning_rate": 2.4906988128748013e-05,
      "loss": 1.2117,
      "step": 164600
    },
    {
      "epoch": 5.031465754261624,
      "grad_norm": 4.801534175872803,
      "learning_rate": 2.4891690123607884e-05,
      "loss": 1.1532,
      "step": 164700
    },
    {
      "epoch": 5.03452068185984,
      "grad_norm": 5.785711288452148,
      "learning_rate": 2.4876392118467752e-05,
      "loss": 1.3217,
      "step": 164800
    },
    {
      "epoch": 5.037575609458056,
      "grad_norm": 2.3183858394622803,
      "learning_rate": 2.4861094113327623e-05,
      "loss": 1.3079,
      "step": 164900
    },
    {
      "epoch": 5.040630537056272,
      "grad_norm": 1.9743845462799072,
      "learning_rate": 2.4845796108187494e-05,
      "loss": 1.1051,
      "step": 165000
    },
    {
      "epoch": 5.043685464654487,
      "grad_norm": 4.131516933441162,
      "learning_rate": 2.4830498103047365e-05,
      "loss": 1.0025,
      "step": 165100
    },
    {
      "epoch": 5.046740392252704,
      "grad_norm": 6.193573951721191,
      "learning_rate": 2.4815200097907233e-05,
      "loss": 1.2593,
      "step": 165200
    },
    {
      "epoch": 5.04979531985092,
      "grad_norm": 9.684208869934082,
      "learning_rate": 2.4799902092767104e-05,
      "loss": 1.1306,
      "step": 165300
    },
    {
      "epoch": 5.052850247449135,
      "grad_norm": 7.245005130767822,
      "learning_rate": 2.4784757067678376e-05,
      "loss": 1.0882,
      "step": 165400
    },
    {
      "epoch": 5.055905175047352,
      "grad_norm": 6.9910664558410645,
      "learning_rate": 2.4769459062538247e-05,
      "loss": 0.9976,
      "step": 165500
    },
    {
      "epoch": 5.0589601026455675,
      "grad_norm": 5.503900051116943,
      "learning_rate": 2.4754161057398118e-05,
      "loss": 1.0166,
      "step": 165600
    },
    {
      "epoch": 5.062015030243783,
      "grad_norm": 6.693299770355225,
      "learning_rate": 2.473886305225799e-05,
      "loss": 1.0534,
      "step": 165700
    },
    {
      "epoch": 5.065069957841999,
      "grad_norm": 5.447563171386719,
      "learning_rate": 2.4723565047117857e-05,
      "loss": 1.1812,
      "step": 165800
    },
    {
      "epoch": 5.068124885440215,
      "grad_norm": 6.907769680023193,
      "learning_rate": 2.4708267041977728e-05,
      "loss": 1.0964,
      "step": 165900
    },
    {
      "epoch": 5.071179813038431,
      "grad_norm": 9.287679672241211,
      "learning_rate": 2.46929690368376e-05,
      "loss": 1.1612,
      "step": 166000
    },
    {
      "epoch": 5.074234740636647,
      "grad_norm": 2.0512337684631348,
      "learning_rate": 2.467767103169747e-05,
      "loss": 1.0268,
      "step": 166100
    },
    {
      "epoch": 5.0772896682348625,
      "grad_norm": 2.716292381286621,
      "learning_rate": 2.4662373026557338e-05,
      "loss": 1.2245,
      "step": 166200
    },
    {
      "epoch": 5.080344595833079,
      "grad_norm": 13.274731636047363,
      "learning_rate": 2.464707502141721e-05,
      "loss": 1.3263,
      "step": 166300
    },
    {
      "epoch": 5.083399523431295,
      "grad_norm": 3.94521427154541,
      "learning_rate": 2.463177701627708e-05,
      "loss": 0.9961,
      "step": 166400
    },
    {
      "epoch": 5.08645445102951,
      "grad_norm": 7.5687456130981445,
      "learning_rate": 2.4616479011136948e-05,
      "loss": 1.3137,
      "step": 166500
    },
    {
      "epoch": 5.089509378627726,
      "grad_norm": 6.075013160705566,
      "learning_rate": 2.460118100599682e-05,
      "loss": 1.0343,
      "step": 166600
    },
    {
      "epoch": 5.092564306225943,
      "grad_norm": 13.179591178894043,
      "learning_rate": 2.4585883000856687e-05,
      "loss": 1.2662,
      "step": 166700
    },
    {
      "epoch": 5.095619233824158,
      "grad_norm": 12.164190292358398,
      "learning_rate": 2.4570584995716558e-05,
      "loss": 1.2506,
      "step": 166800
    },
    {
      "epoch": 5.098674161422374,
      "grad_norm": 7.566710472106934,
      "learning_rate": 2.455528699057643e-05,
      "loss": 1.2013,
      "step": 166900
    },
    {
      "epoch": 5.101729089020591,
      "grad_norm": 7.673246383666992,
      "learning_rate": 2.45399889854363e-05,
      "loss": 1.0592,
      "step": 167000
    },
    {
      "epoch": 5.104784016618806,
      "grad_norm": 9.131725311279297,
      "learning_rate": 2.452469098029617e-05,
      "loss": 1.098,
      "step": 167100
    },
    {
      "epoch": 5.107838944217022,
      "grad_norm": 4.740269184112549,
      "learning_rate": 2.450939297515604e-05,
      "loss": 1.0963,
      "step": 167200
    },
    {
      "epoch": 5.110893871815238,
      "grad_norm": 12.692572593688965,
      "learning_rate": 2.449409497001591e-05,
      "loss": 1.2042,
      "step": 167300
    },
    {
      "epoch": 5.113948799413454,
      "grad_norm": 10.311156272888184,
      "learning_rate": 2.4478796964875782e-05,
      "loss": 1.0407,
      "step": 167400
    },
    {
      "epoch": 5.11700372701167,
      "grad_norm": 13.6527681350708,
      "learning_rate": 2.4463804919838453e-05,
      "loss": 1.3487,
      "step": 167500
    },
    {
      "epoch": 5.120058654609886,
      "grad_norm": 6.338456153869629,
      "learning_rate": 2.4448506914698325e-05,
      "loss": 1.1242,
      "step": 167600
    },
    {
      "epoch": 5.123113582208101,
      "grad_norm": 9.074694633483887,
      "learning_rate": 2.4433208909558196e-05,
      "loss": 1.0838,
      "step": 167700
    },
    {
      "epoch": 5.126168509806318,
      "grad_norm": 4.044082164764404,
      "learning_rate": 2.4417910904418063e-05,
      "loss": 1.0163,
      "step": 167800
    },
    {
      "epoch": 5.129223437404534,
      "grad_norm": 6.65750789642334,
      "learning_rate": 2.4402612899277935e-05,
      "loss": 1.1827,
      "step": 167900
    },
    {
      "epoch": 5.132278365002749,
      "grad_norm": 10.863242149353027,
      "learning_rate": 2.4387314894137806e-05,
      "loss": 1.0881,
      "step": 168000
    },
    {
      "epoch": 5.135333292600965,
      "grad_norm": 4.646780967712402,
      "learning_rate": 2.4372016888997677e-05,
      "loss": 0.9714,
      "step": 168100
    },
    {
      "epoch": 5.1383882201991815,
      "grad_norm": 8.3972749710083,
      "learning_rate": 2.4356718883857545e-05,
      "loss": 1.0478,
      "step": 168200
    },
    {
      "epoch": 5.141443147797397,
      "grad_norm": 9.871989250183105,
      "learning_rate": 2.4341420878717416e-05,
      "loss": 1.4897,
      "step": 168300
    },
    {
      "epoch": 5.144498075395613,
      "grad_norm": 9.105977058410645,
      "learning_rate": 2.4326122873577287e-05,
      "loss": 1.1656,
      "step": 168400
    },
    {
      "epoch": 5.1475530029938295,
      "grad_norm": 9.940979957580566,
      "learning_rate": 2.4310824868437158e-05,
      "loss": 1.2031,
      "step": 168500
    },
    {
      "epoch": 5.150607930592045,
      "grad_norm": 4.936947345733643,
      "learning_rate": 2.4295526863297026e-05,
      "loss": 0.9928,
      "step": 168600
    },
    {
      "epoch": 5.153662858190261,
      "grad_norm": 5.739433765411377,
      "learning_rate": 2.4280228858156897e-05,
      "loss": 1.3259,
      "step": 168700
    },
    {
      "epoch": 5.1567177857884765,
      "grad_norm": 3.817854404449463,
      "learning_rate": 2.4264930853016768e-05,
      "loss": 1.2278,
      "step": 168800
    },
    {
      "epoch": 5.159772713386693,
      "grad_norm": 6.3019633293151855,
      "learning_rate": 2.424963284787664e-05,
      "loss": 1.1436,
      "step": 168900
    },
    {
      "epoch": 5.162827640984909,
      "grad_norm": 6.779687404632568,
      "learning_rate": 2.4234334842736507e-05,
      "loss": 1.3622,
      "step": 169000
    },
    {
      "epoch": 5.1658825685831244,
      "grad_norm": 4.4668731689453125,
      "learning_rate": 2.421903683759638e-05,
      "loss": 1.2539,
      "step": 169100
    },
    {
      "epoch": 5.16893749618134,
      "grad_norm": 6.3263068199157715,
      "learning_rate": 2.420373883245625e-05,
      "loss": 1.0856,
      "step": 169200
    },
    {
      "epoch": 5.171992423779557,
      "grad_norm": 13.243605613708496,
      "learning_rate": 2.418844082731612e-05,
      "loss": 1.1165,
      "step": 169300
    },
    {
      "epoch": 5.175047351377772,
      "grad_norm": 3.979931354522705,
      "learning_rate": 2.417314282217599e-05,
      "loss": 0.9953,
      "step": 169400
    },
    {
      "epoch": 5.178102278975988,
      "grad_norm": 5.5036821365356445,
      "learning_rate": 2.415784481703586e-05,
      "loss": 1.0252,
      "step": 169500
    },
    {
      "epoch": 5.181157206574204,
      "grad_norm": 3.225660562515259,
      "learning_rate": 2.414254681189573e-05,
      "loss": 1.1752,
      "step": 169600
    },
    {
      "epoch": 5.18421213417242,
      "grad_norm": 6.257805824279785,
      "learning_rate": 2.4127248806755602e-05,
      "loss": 0.94,
      "step": 169700
    },
    {
      "epoch": 5.187267061770636,
      "grad_norm": 2.3778061866760254,
      "learning_rate": 2.411195080161547e-05,
      "loss": 1.138,
      "step": 169800
    },
    {
      "epoch": 5.190321989368852,
      "grad_norm": 10.44018840789795,
      "learning_rate": 2.409665279647534e-05,
      "loss": 1.2677,
      "step": 169900
    },
    {
      "epoch": 5.193376916967068,
      "grad_norm": 6.572098255157471,
      "learning_rate": 2.4081354791335212e-05,
      "loss": 1.1052,
      "step": 170000
    },
    {
      "epoch": 5.196431844565284,
      "grad_norm": 4.0942463874816895,
      "learning_rate": 2.4066056786195083e-05,
      "loss": 1.229,
      "step": 170100
    },
    {
      "epoch": 5.1994867721635,
      "grad_norm": 3.8638057708740234,
      "learning_rate": 2.405075878105495e-05,
      "loss": 1.1286,
      "step": 170200
    },
    {
      "epoch": 5.202541699761715,
      "grad_norm": 5.156481742858887,
      "learning_rate": 2.4035460775914822e-05,
      "loss": 1.4408,
      "step": 170300
    },
    {
      "epoch": 5.205596627359932,
      "grad_norm": 8.467106819152832,
      "learning_rate": 2.4020162770774693e-05,
      "loss": 1.1772,
      "step": 170400
    },
    {
      "epoch": 5.208651554958148,
      "grad_norm": 7.49545431137085,
      "learning_rate": 2.4004864765634564e-05,
      "loss": 1.1276,
      "step": 170500
    },
    {
      "epoch": 5.211706482556363,
      "grad_norm": 5.997565746307373,
      "learning_rate": 2.3989566760494432e-05,
      "loss": 1.0463,
      "step": 170600
    },
    {
      "epoch": 5.214761410154579,
      "grad_norm": 1.3201504945755005,
      "learning_rate": 2.3974421735405707e-05,
      "loss": 1.1142,
      "step": 170700
    },
    {
      "epoch": 5.2178163377527955,
      "grad_norm": 4.374696731567383,
      "learning_rate": 2.3959123730265575e-05,
      "loss": 1.1381,
      "step": 170800
    },
    {
      "epoch": 5.220871265351011,
      "grad_norm": 9.526435852050781,
      "learning_rate": 2.3943825725125446e-05,
      "loss": 1.3192,
      "step": 170900
    },
    {
      "epoch": 5.223926192949227,
      "grad_norm": 4.740625858306885,
      "learning_rate": 2.3928527719985317e-05,
      "loss": 0.9071,
      "step": 171000
    },
    {
      "epoch": 5.226981120547443,
      "grad_norm": 2.8437039852142334,
      "learning_rate": 2.3913229714845188e-05,
      "loss": 1.136,
      "step": 171100
    },
    {
      "epoch": 5.230036048145659,
      "grad_norm": 8.31369686126709,
      "learning_rate": 2.3897931709705056e-05,
      "loss": 0.9906,
      "step": 171200
    },
    {
      "epoch": 5.233090975743875,
      "grad_norm": 6.5492401123046875,
      "learning_rate": 2.3882633704564924e-05,
      "loss": 1.0642,
      "step": 171300
    },
    {
      "epoch": 5.2361459033420905,
      "grad_norm": 13.388254165649414,
      "learning_rate": 2.3867335699424795e-05,
      "loss": 1.1662,
      "step": 171400
    },
    {
      "epoch": 5.239200830940307,
      "grad_norm": 13.241950035095215,
      "learning_rate": 2.3852037694284666e-05,
      "loss": 1.1376,
      "step": 171500
    },
    {
      "epoch": 5.242255758538523,
      "grad_norm": 15.0129976272583,
      "learning_rate": 2.3836739689144537e-05,
      "loss": 1.0281,
      "step": 171600
    },
    {
      "epoch": 5.2453106861367385,
      "grad_norm": 2.2022969722747803,
      "learning_rate": 2.3821441684004405e-05,
      "loss": 1.2516,
      "step": 171700
    },
    {
      "epoch": 5.248365613734954,
      "grad_norm": 8.645432472229004,
      "learning_rate": 2.3806143678864276e-05,
      "loss": 1.0883,
      "step": 171800
    },
    {
      "epoch": 5.251420541333171,
      "grad_norm": 11.106043815612793,
      "learning_rate": 2.3790845673724147e-05,
      "loss": 0.926,
      "step": 171900
    },
    {
      "epoch": 5.254475468931386,
      "grad_norm": 3.1398587226867676,
      "learning_rate": 2.377554766858402e-05,
      "loss": 1.0641,
      "step": 172000
    },
    {
      "epoch": 5.257530396529602,
      "grad_norm": 9.555487632751465,
      "learning_rate": 2.3760249663443886e-05,
      "loss": 1.0317,
      "step": 172100
    },
    {
      "epoch": 5.260585324127818,
      "grad_norm": 10.592201232910156,
      "learning_rate": 2.3744951658303757e-05,
      "loss": 1.0764,
      "step": 172200
    },
    {
      "epoch": 5.263640251726034,
      "grad_norm": 11.372227668762207,
      "learning_rate": 2.372965365316363e-05,
      "loss": 1.1134,
      "step": 172300
    },
    {
      "epoch": 5.26669517932425,
      "grad_norm": 3.5038912296295166,
      "learning_rate": 2.37143556480235e-05,
      "loss": 1.0342,
      "step": 172400
    },
    {
      "epoch": 5.269750106922466,
      "grad_norm": 9.717251777648926,
      "learning_rate": 2.3699057642883367e-05,
      "loss": 1.1635,
      "step": 172500
    },
    {
      "epoch": 5.272805034520681,
      "grad_norm": 13.201680183410645,
      "learning_rate": 2.368375963774324e-05,
      "loss": 1.2265,
      "step": 172600
    },
    {
      "epoch": 5.275859962118898,
      "grad_norm": 6.547023773193359,
      "learning_rate": 2.366846163260311e-05,
      "loss": 1.0407,
      "step": 172700
    },
    {
      "epoch": 5.278914889717114,
      "grad_norm": 3.413754463195801,
      "learning_rate": 2.365316362746298e-05,
      "loss": 1.2009,
      "step": 172800
    },
    {
      "epoch": 5.281969817315329,
      "grad_norm": 3.3388729095458984,
      "learning_rate": 2.363786562232285e-05,
      "loss": 0.9942,
      "step": 172900
    },
    {
      "epoch": 5.285024744913546,
      "grad_norm": 3.213888645172119,
      "learning_rate": 2.362256761718272e-05,
      "loss": 1.1865,
      "step": 173000
    },
    {
      "epoch": 5.288079672511762,
      "grad_norm": 7.584039688110352,
      "learning_rate": 2.360726961204259e-05,
      "loss": 1.1066,
      "step": 173100
    },
    {
      "epoch": 5.291134600109977,
      "grad_norm": 9.122584342956543,
      "learning_rate": 2.3591971606902462e-05,
      "loss": 1.0166,
      "step": 173200
    },
    {
      "epoch": 5.294189527708193,
      "grad_norm": 14.75819206237793,
      "learning_rate": 2.357667360176233e-05,
      "loss": 1.2447,
      "step": 173300
    },
    {
      "epoch": 5.2972444553064095,
      "grad_norm": 3.378222703933716,
      "learning_rate": 2.35615285766736e-05,
      "loss": 1.1964,
      "step": 173400
    },
    {
      "epoch": 5.300299382904625,
      "grad_norm": 1.2886375188827515,
      "learning_rate": 2.3546230571533472e-05,
      "loss": 1.1858,
      "step": 173500
    },
    {
      "epoch": 5.303354310502841,
      "grad_norm": 14.222728729248047,
      "learning_rate": 2.3530932566393344e-05,
      "loss": 1.3939,
      "step": 173600
    },
    {
      "epoch": 5.3064092381010575,
      "grad_norm": 7.933919906616211,
      "learning_rate": 2.3515634561253215e-05,
      "loss": 1.0642,
      "step": 173700
    },
    {
      "epoch": 5.309464165699273,
      "grad_norm": 7.706429958343506,
      "learning_rate": 2.3500336556113082e-05,
      "loss": 1.0335,
      "step": 173800
    },
    {
      "epoch": 5.312519093297489,
      "grad_norm": 11.539246559143066,
      "learning_rate": 2.3485038550972954e-05,
      "loss": 1.0982,
      "step": 173900
    },
    {
      "epoch": 5.3155740208957045,
      "grad_norm": 8.754623413085938,
      "learning_rate": 2.3469740545832825e-05,
      "loss": 0.9977,
      "step": 174000
    },
    {
      "epoch": 5.318628948493921,
      "grad_norm": 8.987189292907715,
      "learning_rate": 2.3454442540692696e-05,
      "loss": 1.176,
      "step": 174100
    },
    {
      "epoch": 5.321683876092137,
      "grad_norm": 15.71728515625,
      "learning_rate": 2.3439144535552564e-05,
      "loss": 1.0027,
      "step": 174200
    },
    {
      "epoch": 5.3247388036903525,
      "grad_norm": 8.85295295715332,
      "learning_rate": 2.3423846530412435e-05,
      "loss": 1.1556,
      "step": 174300
    },
    {
      "epoch": 5.327793731288568,
      "grad_norm": 20.38521957397461,
      "learning_rate": 2.3408548525272306e-05,
      "loss": 1.2025,
      "step": 174400
    },
    {
      "epoch": 5.330848658886785,
      "grad_norm": 13.312535285949707,
      "learning_rate": 2.3393250520132177e-05,
      "loss": 1.0994,
      "step": 174500
    },
    {
      "epoch": 5.333903586485,
      "grad_norm": 0.015094890259206295,
      "learning_rate": 2.3377952514992045e-05,
      "loss": 1.0021,
      "step": 174600
    },
    {
      "epoch": 5.336958514083216,
      "grad_norm": 8.27639102935791,
      "learning_rate": 2.3362654509851916e-05,
      "loss": 1.0325,
      "step": 174700
    },
    {
      "epoch": 5.340013441681432,
      "grad_norm": 9.89177417755127,
      "learning_rate": 2.3347356504711787e-05,
      "loss": 1.0845,
      "step": 174800
    },
    {
      "epoch": 5.343068369279648,
      "grad_norm": 6.068204879760742,
      "learning_rate": 2.333205849957166e-05,
      "loss": 1.0179,
      "step": 174900
    },
    {
      "epoch": 5.346123296877864,
      "grad_norm": 5.956294536590576,
      "learning_rate": 2.3316760494431526e-05,
      "loss": 1.245,
      "step": 175000
    },
    {
      "epoch": 5.34917822447608,
      "grad_norm": 5.717597007751465,
      "learning_rate": 2.3301462489291397e-05,
      "loss": 1.1191,
      "step": 175100
    },
    {
      "epoch": 5.352233152074296,
      "grad_norm": 17.489805221557617,
      "learning_rate": 2.328616448415127e-05,
      "loss": 1.2599,
      "step": 175200
    },
    {
      "epoch": 5.355288079672512,
      "grad_norm": 4.305554389953613,
      "learning_rate": 2.327086647901114e-05,
      "loss": 1.1257,
      "step": 175300
    },
    {
      "epoch": 5.358343007270728,
      "grad_norm": 16.285791397094727,
      "learning_rate": 2.3255568473871007e-05,
      "loss": 1.2983,
      "step": 175400
    },
    {
      "epoch": 5.361397934868943,
      "grad_norm": 4.100075721740723,
      "learning_rate": 2.324027046873088e-05,
      "loss": 1.0178,
      "step": 175500
    },
    {
      "epoch": 5.36445286246716,
      "grad_norm": 7.133657932281494,
      "learning_rate": 2.322497246359075e-05,
      "loss": 1.1496,
      "step": 175600
    },
    {
      "epoch": 5.367507790065376,
      "grad_norm": 8.204270362854004,
      "learning_rate": 2.320967445845062e-05,
      "loss": 1.0097,
      "step": 175700
    },
    {
      "epoch": 5.370562717663591,
      "grad_norm": 3.40224289894104,
      "learning_rate": 2.319437645331049e-05,
      "loss": 1.4796,
      "step": 175800
    },
    {
      "epoch": 5.373617645261807,
      "grad_norm": 5.546578884124756,
      "learning_rate": 2.317907844817036e-05,
      "loss": 1.4034,
      "step": 175900
    },
    {
      "epoch": 5.3766725728600235,
      "grad_norm": 7.096791744232178,
      "learning_rate": 2.316378044303023e-05,
      "loss": 1.0794,
      "step": 176000
    },
    {
      "epoch": 5.379727500458239,
      "grad_norm": 9.707426071166992,
      "learning_rate": 2.3148482437890102e-05,
      "loss": 0.9545,
      "step": 176100
    },
    {
      "epoch": 5.382782428056455,
      "grad_norm": 13.694910049438477,
      "learning_rate": 2.313318443274997e-05,
      "loss": 0.9155,
      "step": 176200
    },
    {
      "epoch": 5.385837355654671,
      "grad_norm": 8.697030067443848,
      "learning_rate": 2.311788642760984e-05,
      "loss": 1.1714,
      "step": 176300
    },
    {
      "epoch": 5.388892283252887,
      "grad_norm": 8.831588745117188,
      "learning_rate": 2.3102588422469712e-05,
      "loss": 1.022,
      "step": 176400
    },
    {
      "epoch": 5.391947210851103,
      "grad_norm": 9.650738716125488,
      "learning_rate": 2.3087290417329583e-05,
      "loss": 1.1336,
      "step": 176500
    },
    {
      "epoch": 5.3950021384493185,
      "grad_norm": 5.9181227684021,
      "learning_rate": 2.307199241218945e-05,
      "loss": 1.1614,
      "step": 176600
    },
    {
      "epoch": 5.398057066047535,
      "grad_norm": 19.085289001464844,
      "learning_rate": 2.3056694407049322e-05,
      "loss": 1.26,
      "step": 176700
    },
    {
      "epoch": 5.401111993645751,
      "grad_norm": 4.559824466705322,
      "learning_rate": 2.3041396401909193e-05,
      "loss": 1.0525,
      "step": 176800
    },
    {
      "epoch": 5.4041669212439665,
      "grad_norm": 7.130990505218506,
      "learning_rate": 2.3026098396769065e-05,
      "loss": 1.1655,
      "step": 176900
    },
    {
      "epoch": 5.407221848842182,
      "grad_norm": 8.152735710144043,
      "learning_rate": 2.3010800391628932e-05,
      "loss": 1.0988,
      "step": 177000
    },
    {
      "epoch": 5.410276776440399,
      "grad_norm": 0.021126531064510345,
      "learning_rate": 2.2995502386488804e-05,
      "loss": 1.2543,
      "step": 177100
    },
    {
      "epoch": 5.413331704038614,
      "grad_norm": 12.942683219909668,
      "learning_rate": 2.2980204381348675e-05,
      "loss": 0.9339,
      "step": 177200
    },
    {
      "epoch": 5.41638663163683,
      "grad_norm": 5.814689636230469,
      "learning_rate": 2.2964906376208546e-05,
      "loss": 1.0118,
      "step": 177300
    },
    {
      "epoch": 5.419441559235046,
      "grad_norm": 4.256146430969238,
      "learning_rate": 2.2949761351119814e-05,
      "loss": 1.1236,
      "step": 177400
    },
    {
      "epoch": 5.422496486833262,
      "grad_norm": 5.030366897583008,
      "learning_rate": 2.2934463345979685e-05,
      "loss": 1.221,
      "step": 177500
    },
    {
      "epoch": 5.425551414431478,
      "grad_norm": 11.197477340698242,
      "learning_rate": 2.2919165340839556e-05,
      "loss": 1.3183,
      "step": 177600
    },
    {
      "epoch": 5.428606342029694,
      "grad_norm": 4.696968078613281,
      "learning_rate": 2.2903867335699424e-05,
      "loss": 1.2085,
      "step": 177700
    },
    {
      "epoch": 5.431661269627909,
      "grad_norm": 3.7039284706115723,
      "learning_rate": 2.2888569330559295e-05,
      "loss": 1.2063,
      "step": 177800
    },
    {
      "epoch": 5.434716197226126,
      "grad_norm": 4.057086944580078,
      "learning_rate": 2.2873271325419166e-05,
      "loss": 1.1562,
      "step": 177900
    },
    {
      "epoch": 5.437771124824342,
      "grad_norm": 9.291214942932129,
      "learning_rate": 2.2857973320279037e-05,
      "loss": 1.2057,
      "step": 178000
    },
    {
      "epoch": 5.440826052422557,
      "grad_norm": 3.367495536804199,
      "learning_rate": 2.2842675315138905e-05,
      "loss": 1.0456,
      "step": 178100
    },
    {
      "epoch": 5.443880980020774,
      "grad_norm": 10.886100769042969,
      "learning_rate": 2.2827377309998776e-05,
      "loss": 1.4536,
      "step": 178200
    },
    {
      "epoch": 5.44693590761899,
      "grad_norm": 11.006861686706543,
      "learning_rate": 2.2812079304858647e-05,
      "loss": 1.4364,
      "step": 178300
    },
    {
      "epoch": 5.449990835217205,
      "grad_norm": 21.764328002929688,
      "learning_rate": 2.279693427976992e-05,
      "loss": 1.1304,
      "step": 178400
    },
    {
      "epoch": 5.453045762815421,
      "grad_norm": 1.3549553155899048,
      "learning_rate": 2.278163627462979e-05,
      "loss": 1.3561,
      "step": 178500
    },
    {
      "epoch": 5.4561006904136375,
      "grad_norm": 4.178231716156006,
      "learning_rate": 2.2766338269489658e-05,
      "loss": 1.1695,
      "step": 178600
    },
    {
      "epoch": 5.459155618011853,
      "grad_norm": 7.525631427764893,
      "learning_rate": 2.275104026434953e-05,
      "loss": 1.397,
      "step": 178700
    },
    {
      "epoch": 5.462210545610069,
      "grad_norm": 7.3762593269348145,
      "learning_rate": 2.27357422592094e-05,
      "loss": 1.2551,
      "step": 178800
    },
    {
      "epoch": 5.465265473208285,
      "grad_norm": 12.109660148620605,
      "learning_rate": 2.272044425406927e-05,
      "loss": 1.3221,
      "step": 178900
    },
    {
      "epoch": 5.468320400806501,
      "grad_norm": 8.797658920288086,
      "learning_rate": 2.270514624892914e-05,
      "loss": 1.1983,
      "step": 179000
    },
    {
      "epoch": 5.471375328404717,
      "grad_norm": 4.56755256652832,
      "learning_rate": 2.268984824378901e-05,
      "loss": 1.0926,
      "step": 179100
    },
    {
      "epoch": 5.4744302560029325,
      "grad_norm": 4.608061790466309,
      "learning_rate": 2.267455023864888e-05,
      "loss": 1.2058,
      "step": 179200
    },
    {
      "epoch": 5.477485183601148,
      "grad_norm": 8.531103134155273,
      "learning_rate": 2.2659252233508752e-05,
      "loss": 1.3427,
      "step": 179300
    },
    {
      "epoch": 5.480540111199365,
      "grad_norm": 4.44692325592041,
      "learning_rate": 2.264395422836862e-05,
      "loss": 1.063,
      "step": 179400
    },
    {
      "epoch": 5.4835950387975805,
      "grad_norm": 14.634761810302734,
      "learning_rate": 2.262865622322849e-05,
      "loss": 1.4712,
      "step": 179500
    },
    {
      "epoch": 5.486649966395796,
      "grad_norm": 3.928048849105835,
      "learning_rate": 2.2613358218088363e-05,
      "loss": 0.9833,
      "step": 179600
    },
    {
      "epoch": 5.489704893994013,
      "grad_norm": 7.0261030197143555,
      "learning_rate": 2.2598060212948234e-05,
      "loss": 1.0317,
      "step": 179700
    },
    {
      "epoch": 5.492759821592228,
      "grad_norm": 5.174136638641357,
      "learning_rate": 2.25827622078081e-05,
      "loss": 1.1652,
      "step": 179800
    },
    {
      "epoch": 5.495814749190444,
      "grad_norm": 7.133640766143799,
      "learning_rate": 2.2567464202667973e-05,
      "loss": 1.1527,
      "step": 179900
    },
    {
      "epoch": 5.49886967678866,
      "grad_norm": 13.290231704711914,
      "learning_rate": 2.2552166197527844e-05,
      "loss": 1.2461,
      "step": 180000
    },
    {
      "epoch": 5.501924604386876,
      "grad_norm": 2.7865355014801025,
      "learning_rate": 2.2536868192387715e-05,
      "loss": 1.0835,
      "step": 180100
    },
    {
      "epoch": 5.504979531985092,
      "grad_norm": 3.439481496810913,
      "learning_rate": 2.2521570187247583e-05,
      "loss": 1.2283,
      "step": 180200
    },
    {
      "epoch": 5.508034459583308,
      "grad_norm": 16.009737014770508,
      "learning_rate": 2.2506272182107454e-05,
      "loss": 0.9926,
      "step": 180300
    },
    {
      "epoch": 5.511089387181523,
      "grad_norm": 8.959451675415039,
      "learning_rate": 2.2490974176967325e-05,
      "loss": 1.5304,
      "step": 180400
    },
    {
      "epoch": 5.51414431477974,
      "grad_norm": 10.707795143127441,
      "learning_rate": 2.2475676171827196e-05,
      "loss": 1.1731,
      "step": 180500
    },
    {
      "epoch": 5.517199242377956,
      "grad_norm": 8.683422088623047,
      "learning_rate": 2.2460378166687064e-05,
      "loss": 1.1361,
      "step": 180600
    },
    {
      "epoch": 5.520254169976171,
      "grad_norm": 6.5314507484436035,
      "learning_rate": 2.2445080161546935e-05,
      "loss": 1.1169,
      "step": 180700
    },
    {
      "epoch": 5.523309097574387,
      "grad_norm": 5.770356178283691,
      "learning_rate": 2.2429782156406806e-05,
      "loss": 1.1458,
      "step": 180800
    },
    {
      "epoch": 5.526364025172604,
      "grad_norm": 5.172116756439209,
      "learning_rate": 2.2414484151266677e-05,
      "loss": 1.2962,
      "step": 180900
    },
    {
      "epoch": 5.529418952770819,
      "grad_norm": 3.7103822231292725,
      "learning_rate": 2.2399186146126545e-05,
      "loss": 1.1349,
      "step": 181000
    },
    {
      "epoch": 5.532473880369035,
      "grad_norm": 1.0182088613510132,
      "learning_rate": 2.2383888140986416e-05,
      "loss": 1.2462,
      "step": 181100
    },
    {
      "epoch": 5.5355288079672516,
      "grad_norm": 14.949552536010742,
      "learning_rate": 2.2368590135846287e-05,
      "loss": 1.3082,
      "step": 181200
    },
    {
      "epoch": 5.538583735565467,
      "grad_norm": 18.954626083374023,
      "learning_rate": 2.235329213070616e-05,
      "loss": 1.081,
      "step": 181300
    },
    {
      "epoch": 5.541638663163683,
      "grad_norm": 5.126280307769775,
      "learning_rate": 2.2337994125566026e-05,
      "loss": 1.145,
      "step": 181400
    },
    {
      "epoch": 5.544693590761899,
      "grad_norm": 3.3761730194091797,
      "learning_rate": 2.2322696120425898e-05,
      "loss": 1.1507,
      "step": 181500
    },
    {
      "epoch": 5.547748518360115,
      "grad_norm": 9.969486236572266,
      "learning_rate": 2.230739811528577e-05,
      "loss": 1.0478,
      "step": 181600
    },
    {
      "epoch": 5.550803445958331,
      "grad_norm": 17.782325744628906,
      "learning_rate": 2.229210011014564e-05,
      "loss": 1.0812,
      "step": 181700
    },
    {
      "epoch": 5.5538583735565465,
      "grad_norm": 8.68525505065918,
      "learning_rate": 2.2276802105005508e-05,
      "loss": 1.261,
      "step": 181800
    },
    {
      "epoch": 5.556913301154763,
      "grad_norm": 5.081483364105225,
      "learning_rate": 2.2261657079916782e-05,
      "loss": 1.0785,
      "step": 181900
    },
    {
      "epoch": 5.559968228752979,
      "grad_norm": 9.229408264160156,
      "learning_rate": 2.224635907477665e-05,
      "loss": 1.515,
      "step": 182000
    },
    {
      "epoch": 5.5630231563511945,
      "grad_norm": 5.441746234893799,
      "learning_rate": 2.223106106963652e-05,
      "loss": 0.9432,
      "step": 182100
    },
    {
      "epoch": 5.56607808394941,
      "grad_norm": 10.51654052734375,
      "learning_rate": 2.2215763064496392e-05,
      "loss": 1.0654,
      "step": 182200
    },
    {
      "epoch": 5.569133011547626,
      "grad_norm": 4.113022327423096,
      "learning_rate": 2.220046505935626e-05,
      "loss": 1.1169,
      "step": 182300
    },
    {
      "epoch": 5.572187939145842,
      "grad_norm": 5.066014289855957,
      "learning_rate": 2.218516705421613e-05,
      "loss": 1.0663,
      "step": 182400
    },
    {
      "epoch": 5.575242866744058,
      "grad_norm": 4.2851104736328125,
      "learning_rate": 2.2169869049076e-05,
      "loss": 1.4151,
      "step": 182500
    },
    {
      "epoch": 5.578297794342274,
      "grad_norm": 7.80756139755249,
      "learning_rate": 2.215457104393587e-05,
      "loss": 0.9683,
      "step": 182600
    },
    {
      "epoch": 5.58135272194049,
      "grad_norm": 16.59423828125,
      "learning_rate": 2.213927303879574e-05,
      "loss": 1.1639,
      "step": 182700
    },
    {
      "epoch": 5.584407649538706,
      "grad_norm": 12.586332321166992,
      "learning_rate": 2.2123975033655613e-05,
      "loss": 1.1313,
      "step": 182800
    },
    {
      "epoch": 5.587462577136922,
      "grad_norm": 3.1343181133270264,
      "learning_rate": 2.210867702851548e-05,
      "loss": 0.9193,
      "step": 182900
    },
    {
      "epoch": 5.590517504735137,
      "grad_norm": 11.592085838317871,
      "learning_rate": 2.209337902337535e-05,
      "loss": 1.2895,
      "step": 183000
    },
    {
      "epoch": 5.593572432333354,
      "grad_norm": 3.3230690956115723,
      "learning_rate": 2.2078081018235223e-05,
      "loss": 1.092,
      "step": 183100
    },
    {
      "epoch": 5.59662735993157,
      "grad_norm": 7.102207660675049,
      "learning_rate": 2.2062783013095094e-05,
      "loss": 1.1058,
      "step": 183200
    },
    {
      "epoch": 5.599682287529785,
      "grad_norm": 6.228555202484131,
      "learning_rate": 2.204748500795496e-05,
      "loss": 1.0174,
      "step": 183300
    },
    {
      "epoch": 5.602737215128002,
      "grad_norm": 7.223354339599609,
      "learning_rate": 2.2032187002814833e-05,
      "loss": 1.0644,
      "step": 183400
    },
    {
      "epoch": 5.605792142726218,
      "grad_norm": 5.12488317489624,
      "learning_rate": 2.2016888997674704e-05,
      "loss": 1.0768,
      "step": 183500
    },
    {
      "epoch": 5.608847070324433,
      "grad_norm": 2.4839391708374023,
      "learning_rate": 2.2001590992534575e-05,
      "loss": 1.2211,
      "step": 183600
    },
    {
      "epoch": 5.611901997922649,
      "grad_norm": 10.138099670410156,
      "learning_rate": 2.1986292987394443e-05,
      "loss": 1.3718,
      "step": 183700
    },
    {
      "epoch": 5.614956925520865,
      "grad_norm": 5.2676682472229,
      "learning_rate": 2.1970994982254314e-05,
      "loss": 1.0689,
      "step": 183800
    },
    {
      "epoch": 5.618011853119081,
      "grad_norm": 1.3094004392623901,
      "learning_rate": 2.1955696977114185e-05,
      "loss": 1.1548,
      "step": 183900
    },
    {
      "epoch": 5.621066780717297,
      "grad_norm": 4.16325569152832,
      "learning_rate": 2.1940398971974056e-05,
      "loss": 1.1477,
      "step": 184000
    },
    {
      "epoch": 5.624121708315513,
      "grad_norm": 10.696335792541504,
      "learning_rate": 2.1925100966833924e-05,
      "loss": 1.0798,
      "step": 184100
    },
    {
      "epoch": 5.627176635913729,
      "grad_norm": 2.4368295669555664,
      "learning_rate": 2.1909802961693795e-05,
      "loss": 0.9771,
      "step": 184200
    },
    {
      "epoch": 5.630231563511945,
      "grad_norm": 4.798482894897461,
      "learning_rate": 2.1894504956553666e-05,
      "loss": 0.9914,
      "step": 184300
    },
    {
      "epoch": 5.6332864911101606,
      "grad_norm": 5.789841651916504,
      "learning_rate": 2.1879206951413538e-05,
      "loss": 1.2666,
      "step": 184400
    },
    {
      "epoch": 5.636341418708376,
      "grad_norm": 3.598188638687134,
      "learning_rate": 2.1863908946273405e-05,
      "loss": 0.9833,
      "step": 184500
    },
    {
      "epoch": 5.639396346306593,
      "grad_norm": 10.746664047241211,
      "learning_rate": 2.1848610941133277e-05,
      "loss": 1.0929,
      "step": 184600
    },
    {
      "epoch": 5.6424512739048085,
      "grad_norm": 11.472532272338867,
      "learning_rate": 2.1833312935993148e-05,
      "loss": 1.1305,
      "step": 184700
    },
    {
      "epoch": 5.645506201503024,
      "grad_norm": 3.4682164192199707,
      "learning_rate": 2.181801493085302e-05,
      "loss": 0.9999,
      "step": 184800
    },
    {
      "epoch": 5.648561129101241,
      "grad_norm": 5.039991855621338,
      "learning_rate": 2.1802716925712887e-05,
      "loss": 1.0,
      "step": 184900
    },
    {
      "epoch": 5.651616056699456,
      "grad_norm": 22.90473175048828,
      "learning_rate": 2.1787418920572758e-05,
      "loss": 1.366,
      "step": 185000
    },
    {
      "epoch": 5.654670984297672,
      "grad_norm": 4.122875213623047,
      "learning_rate": 2.177212091543263e-05,
      "loss": 1.1675,
      "step": 185100
    },
    {
      "epoch": 5.657725911895888,
      "grad_norm": 3.3970232009887695,
      "learning_rate": 2.17568229102925e-05,
      "loss": 1.1691,
      "step": 185200
    },
    {
      "epoch": 5.660780839494104,
      "grad_norm": 17.90229034423828,
      "learning_rate": 2.1741524905152368e-05,
      "loss": 1.0621,
      "step": 185300
    },
    {
      "epoch": 5.66383576709232,
      "grad_norm": 8.32997989654541,
      "learning_rate": 2.172622690001224e-05,
      "loss": 1.0607,
      "step": 185400
    },
    {
      "epoch": 5.666890694690536,
      "grad_norm": 4.489802837371826,
      "learning_rate": 2.171092889487211e-05,
      "loss": 1.1645,
      "step": 185500
    },
    {
      "epoch": 5.669945622288751,
      "grad_norm": 9.886762619018555,
      "learning_rate": 2.169563088973198e-05,
      "loss": 1.0695,
      "step": 185600
    },
    {
      "epoch": 5.673000549886968,
      "grad_norm": 3.3725993633270264,
      "learning_rate": 2.168033288459185e-05,
      "loss": 0.9897,
      "step": 185700
    },
    {
      "epoch": 5.676055477485184,
      "grad_norm": 3.5722696781158447,
      "learning_rate": 2.166503487945172e-05,
      "loss": 1.1009,
      "step": 185800
    },
    {
      "epoch": 5.679110405083399,
      "grad_norm": 3.939537763595581,
      "learning_rate": 2.164973687431159e-05,
      "loss": 1.1344,
      "step": 185900
    },
    {
      "epoch": 5.682165332681615,
      "grad_norm": 8.358038902282715,
      "learning_rate": 2.1634591849222863e-05,
      "loss": 0.9933,
      "step": 186000
    },
    {
      "epoch": 5.685220260279832,
      "grad_norm": 6.315106391906738,
      "learning_rate": 2.1619293844082734e-05,
      "loss": 1.0294,
      "step": 186100
    },
    {
      "epoch": 5.688275187878047,
      "grad_norm": 7.731125831604004,
      "learning_rate": 2.1603995838942605e-05,
      "loss": 1.1796,
      "step": 186200
    },
    {
      "epoch": 5.691330115476263,
      "grad_norm": 10.588342666625977,
      "learning_rate": 2.1588697833802473e-05,
      "loss": 1.0994,
      "step": 186300
    },
    {
      "epoch": 5.69438504307448,
      "grad_norm": 4.702441215515137,
      "learning_rate": 2.1573552808713744e-05,
      "loss": 1.3255,
      "step": 186400
    },
    {
      "epoch": 5.697439970672695,
      "grad_norm": 0.7584722638130188,
      "learning_rate": 2.1558254803573615e-05,
      "loss": 1.2591,
      "step": 186500
    },
    {
      "epoch": 5.700494898270911,
      "grad_norm": 3.2129814624786377,
      "learning_rate": 2.1542956798433486e-05,
      "loss": 0.942,
      "step": 186600
    },
    {
      "epoch": 5.703549825869127,
      "grad_norm": 6.426685810089111,
      "learning_rate": 2.1527658793293358e-05,
      "loss": 1.2921,
      "step": 186700
    },
    {
      "epoch": 5.706604753467343,
      "grad_norm": 4.396588325500488,
      "learning_rate": 2.1512360788153225e-05,
      "loss": 0.9987,
      "step": 186800
    },
    {
      "epoch": 5.709659681065559,
      "grad_norm": 4.9708428382873535,
      "learning_rate": 2.1497062783013097e-05,
      "loss": 1.2156,
      "step": 186900
    },
    {
      "epoch": 5.712714608663775,
      "grad_norm": 7.02872371673584,
      "learning_rate": 2.1481764777872968e-05,
      "loss": 0.9759,
      "step": 187000
    },
    {
      "epoch": 5.71576953626199,
      "grad_norm": 8.63855266571045,
      "learning_rate": 2.146646677273284e-05,
      "loss": 1.3669,
      "step": 187100
    },
    {
      "epoch": 5.718824463860207,
      "grad_norm": 21.115148544311523,
      "learning_rate": 2.1451168767592707e-05,
      "loss": 1.1757,
      "step": 187200
    },
    {
      "epoch": 5.7218793914584225,
      "grad_norm": 7.059637069702148,
      "learning_rate": 2.1435870762452578e-05,
      "loss": 1.0772,
      "step": 187300
    },
    {
      "epoch": 5.724934319056638,
      "grad_norm": 5.446715831756592,
      "learning_rate": 2.142057275731245e-05,
      "loss": 1.1456,
      "step": 187400
    },
    {
      "epoch": 5.727989246654854,
      "grad_norm": 4.783729076385498,
      "learning_rate": 2.140527475217232e-05,
      "loss": 1.1111,
      "step": 187500
    },
    {
      "epoch": 5.7310441742530704,
      "grad_norm": 7.948205947875977,
      "learning_rate": 2.1389976747032188e-05,
      "loss": 1.2508,
      "step": 187600
    },
    {
      "epoch": 5.734099101851286,
      "grad_norm": 18.996597290039062,
      "learning_rate": 2.137467874189206e-05,
      "loss": 0.9658,
      "step": 187700
    },
    {
      "epoch": 5.737154029449502,
      "grad_norm": 5.666006088256836,
      "learning_rate": 2.1359380736751927e-05,
      "loss": 1.1193,
      "step": 187800
    },
    {
      "epoch": 5.740208957047718,
      "grad_norm": 3.910343885421753,
      "learning_rate": 2.1344082731611798e-05,
      "loss": 1.2442,
      "step": 187900
    },
    {
      "epoch": 5.743263884645934,
      "grad_norm": 1.8572306632995605,
      "learning_rate": 2.132878472647167e-05,
      "loss": 1.0246,
      "step": 188000
    },
    {
      "epoch": 5.74631881224415,
      "grad_norm": 7.690230846405029,
      "learning_rate": 2.1313486721331537e-05,
      "loss": 1.7488,
      "step": 188100
    },
    {
      "epoch": 5.749373739842365,
      "grad_norm": 5.309960842132568,
      "learning_rate": 2.1298188716191408e-05,
      "loss": 1.098,
      "step": 188200
    },
    {
      "epoch": 5.752428667440582,
      "grad_norm": 6.675844192504883,
      "learning_rate": 2.128289071105128e-05,
      "loss": 0.9723,
      "step": 188300
    },
    {
      "epoch": 5.755483595038798,
      "grad_norm": 14.125645637512207,
      "learning_rate": 2.126759270591115e-05,
      "loss": 1.0996,
      "step": 188400
    },
    {
      "epoch": 5.758538522637013,
      "grad_norm": 6.408461093902588,
      "learning_rate": 2.1252294700771018e-05,
      "loss": 1.081,
      "step": 188500
    },
    {
      "epoch": 5.761593450235229,
      "grad_norm": 6.611395835876465,
      "learning_rate": 2.123699669563089e-05,
      "loss": 0.9786,
      "step": 188600
    },
    {
      "epoch": 5.764648377833446,
      "grad_norm": 3.6658895015716553,
      "learning_rate": 2.122169869049076e-05,
      "loss": 1.1722,
      "step": 188700
    },
    {
      "epoch": 5.767703305431661,
      "grad_norm": 7.49069356918335,
      "learning_rate": 2.120640068535063e-05,
      "loss": 1.0996,
      "step": 188800
    },
    {
      "epoch": 5.770758233029877,
      "grad_norm": 13.291098594665527,
      "learning_rate": 2.11911026802105e-05,
      "loss": 0.9886,
      "step": 188900
    },
    {
      "epoch": 5.773813160628093,
      "grad_norm": 6.7149977684021,
      "learning_rate": 2.1175957655121774e-05,
      "loss": 1.1279,
      "step": 189000
    },
    {
      "epoch": 5.776868088226309,
      "grad_norm": 5.4691948890686035,
      "learning_rate": 2.1160659649981642e-05,
      "loss": 1.1921,
      "step": 189100
    },
    {
      "epoch": 5.779923015824525,
      "grad_norm": 7.12191104888916,
      "learning_rate": 2.1145361644841513e-05,
      "loss": 1.1089,
      "step": 189200
    },
    {
      "epoch": 5.782977943422741,
      "grad_norm": 6.525013446807861,
      "learning_rate": 2.1130063639701384e-05,
      "loss": 1.1687,
      "step": 189300
    },
    {
      "epoch": 5.786032871020957,
      "grad_norm": 3.475522518157959,
      "learning_rate": 2.1114765634561255e-05,
      "loss": 1.0949,
      "step": 189400
    },
    {
      "epoch": 5.789087798619173,
      "grad_norm": 4.3520965576171875,
      "learning_rate": 2.1099467629421123e-05,
      "loss": 1.1071,
      "step": 189500
    },
    {
      "epoch": 5.792142726217389,
      "grad_norm": 24.487106323242188,
      "learning_rate": 2.1084169624280994e-05,
      "loss": 1.0645,
      "step": 189600
    },
    {
      "epoch": 5.795197653815604,
      "grad_norm": 11.095251083374023,
      "learning_rate": 2.1068871619140865e-05,
      "loss": 1.1057,
      "step": 189700
    },
    {
      "epoch": 5.798252581413821,
      "grad_norm": 6.842864513397217,
      "learning_rate": 2.1053573614000737e-05,
      "loss": 1.2524,
      "step": 189800
    },
    {
      "epoch": 5.8013075090120365,
      "grad_norm": 7.722935199737549,
      "learning_rate": 2.1038275608860604e-05,
      "loss": 1.0865,
      "step": 189900
    },
    {
      "epoch": 5.804362436610252,
      "grad_norm": 10.01695728302002,
      "learning_rate": 2.1022977603720476e-05,
      "loss": 1.1852,
      "step": 190000
    },
    {
      "epoch": 5.807417364208468,
      "grad_norm": 12.786285400390625,
      "learning_rate": 2.1007679598580347e-05,
      "loss": 1.0876,
      "step": 190100
    },
    {
      "epoch": 5.8104722918066845,
      "grad_norm": 5.113598823547363,
      "learning_rate": 2.0992381593440218e-05,
      "loss": 1.1766,
      "step": 190200
    },
    {
      "epoch": 5.8135272194049,
      "grad_norm": 4.3457865715026855,
      "learning_rate": 2.0977083588300086e-05,
      "loss": 1.4302,
      "step": 190300
    },
    {
      "epoch": 5.816582147003116,
      "grad_norm": 11.015501976013184,
      "learning_rate": 2.0961785583159957e-05,
      "loss": 1.2411,
      "step": 190400
    },
    {
      "epoch": 5.8196370746013315,
      "grad_norm": 12.930902481079102,
      "learning_rate": 2.0946487578019828e-05,
      "loss": 0.9841,
      "step": 190500
    },
    {
      "epoch": 5.822692002199548,
      "grad_norm": 2.641270160675049,
      "learning_rate": 2.09311895728797e-05,
      "loss": 1.273,
      "step": 190600
    },
    {
      "epoch": 5.825746929797764,
      "grad_norm": 13.15196418762207,
      "learning_rate": 2.0915891567739567e-05,
      "loss": 1.4972,
      "step": 190700
    },
    {
      "epoch": 5.8288018573959794,
      "grad_norm": 4.2401814460754395,
      "learning_rate": 2.0900593562599438e-05,
      "loss": 1.2889,
      "step": 190800
    },
    {
      "epoch": 5.831856784994196,
      "grad_norm": 15.105935096740723,
      "learning_rate": 2.088529555745931e-05,
      "loss": 1.0851,
      "step": 190900
    },
    {
      "epoch": 5.834911712592412,
      "grad_norm": 2.798130750656128,
      "learning_rate": 2.086999755231918e-05,
      "loss": 1.1304,
      "step": 191000
    },
    {
      "epoch": 5.837966640190627,
      "grad_norm": 2.9408164024353027,
      "learning_rate": 2.0854699547179048e-05,
      "loss": 1.0078,
      "step": 191100
    },
    {
      "epoch": 5.841021567788843,
      "grad_norm": 18.773927688598633,
      "learning_rate": 2.083940154203892e-05,
      "loss": 1.1115,
      "step": 191200
    },
    {
      "epoch": 5.84407649538706,
      "grad_norm": 3.5359275341033936,
      "learning_rate": 2.082410353689879e-05,
      "loss": 1.2128,
      "step": 191300
    },
    {
      "epoch": 5.847131422985275,
      "grad_norm": 3.876824378967285,
      "learning_rate": 2.080880553175866e-05,
      "loss": 1.1701,
      "step": 191400
    },
    {
      "epoch": 5.850186350583491,
      "grad_norm": 5.664970397949219,
      "learning_rate": 2.079350752661853e-05,
      "loss": 1.3018,
      "step": 191500
    },
    {
      "epoch": 5.853241278181708,
      "grad_norm": 4.4520111083984375,
      "learning_rate": 2.07782095214784e-05,
      "loss": 1.0826,
      "step": 191600
    },
    {
      "epoch": 5.856296205779923,
      "grad_norm": 0.5348922610282898,
      "learning_rate": 2.076291151633827e-05,
      "loss": 1.1042,
      "step": 191700
    },
    {
      "epoch": 5.859351133378139,
      "grad_norm": 5.47227668762207,
      "learning_rate": 2.0747766491249543e-05,
      "loss": 1.2615,
      "step": 191800
    },
    {
      "epoch": 5.862406060976355,
      "grad_norm": 6.208307266235352,
      "learning_rate": 2.0732468486109414e-05,
      "loss": 1.4392,
      "step": 191900
    },
    {
      "epoch": 5.86546098857457,
      "grad_norm": 15.780241966247559,
      "learning_rate": 2.0717170480969282e-05,
      "loss": 1.3018,
      "step": 192000
    },
    {
      "epoch": 5.868515916172787,
      "grad_norm": 5.232784271240234,
      "learning_rate": 2.0701872475829153e-05,
      "loss": 1.1271,
      "step": 192100
    },
    {
      "epoch": 5.871570843771003,
      "grad_norm": 17.728816986083984,
      "learning_rate": 2.0686574470689024e-05,
      "loss": 1.2694,
      "step": 192200
    },
    {
      "epoch": 5.874625771369218,
      "grad_norm": 13.02619743347168,
      "learning_rate": 2.0671276465548895e-05,
      "loss": 1.1319,
      "step": 192300
    },
    {
      "epoch": 5.877680698967435,
      "grad_norm": 3.499558448791504,
      "learning_rate": 2.0655978460408763e-05,
      "loss": 1.1393,
      "step": 192400
    },
    {
      "epoch": 5.8807356265656505,
      "grad_norm": 14.559880256652832,
      "learning_rate": 2.0640680455268634e-05,
      "loss": 1.004,
      "step": 192500
    },
    {
      "epoch": 5.883790554163866,
      "grad_norm": 13.002328872680664,
      "learning_rate": 2.0625382450128505e-05,
      "loss": 1.1475,
      "step": 192600
    },
    {
      "epoch": 5.886845481762082,
      "grad_norm": 7.72279691696167,
      "learning_rate": 2.0610084444988377e-05,
      "loss": 1.2198,
      "step": 192700
    },
    {
      "epoch": 5.8899004093602985,
      "grad_norm": 5.531588554382324,
      "learning_rate": 2.0594786439848244e-05,
      "loss": 1.0233,
      "step": 192800
    },
    {
      "epoch": 5.892955336958514,
      "grad_norm": 9.753616333007812,
      "learning_rate": 2.0579488434708116e-05,
      "loss": 1.304,
      "step": 192900
    },
    {
      "epoch": 5.89601026455673,
      "grad_norm": 7.747321605682373,
      "learning_rate": 2.0564190429567987e-05,
      "loss": 1.2117,
      "step": 193000
    },
    {
      "epoch": 5.899065192154946,
      "grad_norm": 4.747279644012451,
      "learning_rate": 2.0548892424427858e-05,
      "loss": 1.0239,
      "step": 193100
    },
    {
      "epoch": 5.902120119753162,
      "grad_norm": 2.2190704345703125,
      "learning_rate": 2.0533594419287726e-05,
      "loss": 1.4906,
      "step": 193200
    },
    {
      "epoch": 5.905175047351378,
      "grad_norm": 12.153207778930664,
      "learning_rate": 2.0518296414147597e-05,
      "loss": 1.1606,
      "step": 193300
    },
    {
      "epoch": 5.9082299749495935,
      "grad_norm": 2.3085525035858154,
      "learning_rate": 2.0502998409007465e-05,
      "loss": 1.0179,
      "step": 193400
    },
    {
      "epoch": 5.911284902547809,
      "grad_norm": 1.8929448127746582,
      "learning_rate": 2.0487700403867336e-05,
      "loss": 1.1237,
      "step": 193500
    },
    {
      "epoch": 5.914339830146026,
      "grad_norm": 11.247891426086426,
      "learning_rate": 2.0472402398727207e-05,
      "loss": 1.0379,
      "step": 193600
    },
    {
      "epoch": 5.917394757744241,
      "grad_norm": 17.98173713684082,
      "learning_rate": 2.0457104393587075e-05,
      "loss": 1.149,
      "step": 193700
    },
    {
      "epoch": 5.920449685342457,
      "grad_norm": 6.99159049987793,
      "learning_rate": 2.0441806388446946e-05,
      "loss": 1.1825,
      "step": 193800
    },
    {
      "epoch": 5.923504612940674,
      "grad_norm": 6.678034782409668,
      "learning_rate": 2.0426508383306817e-05,
      "loss": 1.1868,
      "step": 193900
    },
    {
      "epoch": 5.926559540538889,
      "grad_norm": 1.5114043951034546,
      "learning_rate": 2.0411210378166688e-05,
      "loss": 1.0141,
      "step": 194000
    },
    {
      "epoch": 5.929614468137105,
      "grad_norm": 8.493728637695312,
      "learning_rate": 2.039606535307796e-05,
      "loss": 1.2834,
      "step": 194100
    },
    {
      "epoch": 5.932669395735321,
      "grad_norm": 4.0135602951049805,
      "learning_rate": 2.038076734793783e-05,
      "loss": 1.4871,
      "step": 194200
    },
    {
      "epoch": 5.935724323333537,
      "grad_norm": 19.089086532592773,
      "learning_rate": 2.03654693427977e-05,
      "loss": 1.229,
      "step": 194300
    },
    {
      "epoch": 5.938779250931753,
      "grad_norm": 3.464934825897217,
      "learning_rate": 2.035017133765757e-05,
      "loss": 1.4185,
      "step": 194400
    },
    {
      "epoch": 5.941834178529969,
      "grad_norm": 20.901973724365234,
      "learning_rate": 2.033487333251744e-05,
      "loss": 1.1335,
      "step": 194500
    },
    {
      "epoch": 5.944889106128185,
      "grad_norm": 2.50457763671875,
      "learning_rate": 2.0319575327377312e-05,
      "loss": 1.1107,
      "step": 194600
    },
    {
      "epoch": 5.947944033726401,
      "grad_norm": 3.334287166595459,
      "learning_rate": 2.030427732223718e-05,
      "loss": 1.0931,
      "step": 194700
    },
    {
      "epoch": 5.950998961324617,
      "grad_norm": 6.1684980392456055,
      "learning_rate": 2.028897931709705e-05,
      "loss": 1.1948,
      "step": 194800
    },
    {
      "epoch": 5.954053888922832,
      "grad_norm": 3.7260806560516357,
      "learning_rate": 2.0273681311956922e-05,
      "loss": 1.0075,
      "step": 194900
    },
    {
      "epoch": 5.957108816521048,
      "grad_norm": 6.643927574157715,
      "learning_rate": 2.0258383306816793e-05,
      "loss": 1.0888,
      "step": 195000
    },
    {
      "epoch": 5.9601637441192645,
      "grad_norm": 6.050364971160889,
      "learning_rate": 2.024308530167666e-05,
      "loss": 1.1477,
      "step": 195100
    },
    {
      "epoch": 5.96321867171748,
      "grad_norm": 2.7299838066101074,
      "learning_rate": 2.0227787296536532e-05,
      "loss": 1.1081,
      "step": 195200
    },
    {
      "epoch": 5.966273599315696,
      "grad_norm": 8.811811447143555,
      "learning_rate": 2.0212489291396403e-05,
      "loss": 1.0636,
      "step": 195300
    },
    {
      "epoch": 5.9693285269139125,
      "grad_norm": 9.84276294708252,
      "learning_rate": 2.0197191286256274e-05,
      "loss": 1.3142,
      "step": 195400
    },
    {
      "epoch": 5.972383454512128,
      "grad_norm": 12.914100646972656,
      "learning_rate": 2.0181893281116142e-05,
      "loss": 1.1995,
      "step": 195500
    },
    {
      "epoch": 5.975438382110344,
      "grad_norm": 3.8939926624298096,
      "learning_rate": 2.0166595275976013e-05,
      "loss": 0.9676,
      "step": 195600
    },
    {
      "epoch": 5.9784933097085595,
      "grad_norm": 9.499131202697754,
      "learning_rate": 2.0151297270835884e-05,
      "loss": 1.036,
      "step": 195700
    },
    {
      "epoch": 5.981548237306776,
      "grad_norm": 8.472640037536621,
      "learning_rate": 2.0135999265695756e-05,
      "loss": 1.2044,
      "step": 195800
    },
    {
      "epoch": 5.984603164904992,
      "grad_norm": 7.215255260467529,
      "learning_rate": 2.0120701260555623e-05,
      "loss": 1.2218,
      "step": 195900
    },
    {
      "epoch": 5.9876580925032075,
      "grad_norm": 18.55949592590332,
      "learning_rate": 2.0105403255415495e-05,
      "loss": 1.0842,
      "step": 196000
    },
    {
      "epoch": 5.990713020101424,
      "grad_norm": 2.0042617321014404,
      "learning_rate": 2.0090105250275366e-05,
      "loss": 1.2315,
      "step": 196100
    },
    {
      "epoch": 5.99376794769964,
      "grad_norm": 3.453805923461914,
      "learning_rate": 2.0074807245135237e-05,
      "loss": 1.0177,
      "step": 196200
    },
    {
      "epoch": 5.996822875297855,
      "grad_norm": 6.27379035949707,
      "learning_rate": 2.0059509239995105e-05,
      "loss": 1.4143,
      "step": 196300
    },
    {
      "epoch": 5.999877802896071,
      "grad_norm": 2.6724729537963867,
      "learning_rate": 2.0044211234854976e-05,
      "loss": 1.206,
      "step": 196400
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.7196187450357426,
      "eval_loss": 0.7456799745559692,
      "eval_runtime": 1781.9466,
      "eval_samples_per_second": 18.37,
      "eval_steps_per_second": 4.593,
      "step": 196404
    },
    {
      "epoch": 6.002932730494288,
      "grad_norm": 8.19288158416748,
      "learning_rate": 2.0028913229714847e-05,
      "loss": 1.233,
      "step": 196500
    },
    {
      "epoch": 6.005987658092503,
      "grad_norm": 5.379980087280273,
      "learning_rate": 2.0013615224574718e-05,
      "loss": 1.0025,
      "step": 196600
    },
    {
      "epoch": 6.009042585690719,
      "grad_norm": 8.692319869995117,
      "learning_rate": 1.9998317219434586e-05,
      "loss": 1.0007,
      "step": 196700
    },
    {
      "epoch": 6.012097513288935,
      "grad_norm": 3.728940963745117,
      "learning_rate": 1.9983019214294457e-05,
      "loss": 1.3789,
      "step": 196800
    },
    {
      "epoch": 6.015152440887151,
      "grad_norm": 14.83003044128418,
      "learning_rate": 1.9967721209154328e-05,
      "loss": 1.2283,
      "step": 196900
    },
    {
      "epoch": 6.018207368485367,
      "grad_norm": 10.604146003723145,
      "learning_rate": 1.99524232040142e-05,
      "loss": 1.1086,
      "step": 197000
    },
    {
      "epoch": 6.021262296083583,
      "grad_norm": 2.400364875793457,
      "learning_rate": 1.993727817892547e-05,
      "loss": 0.9838,
      "step": 197100
    },
    {
      "epoch": 6.024317223681798,
      "grad_norm": 16.308069229125977,
      "learning_rate": 1.992198017378534e-05,
      "loss": 1.3164,
      "step": 197200
    },
    {
      "epoch": 6.027372151280015,
      "grad_norm": 21.39978790283203,
      "learning_rate": 1.990668216864521e-05,
      "loss": 1.1362,
      "step": 197300
    },
    {
      "epoch": 6.030427078878231,
      "grad_norm": 5.836481094360352,
      "learning_rate": 1.989138416350508e-05,
      "loss": 1.2144,
      "step": 197400
    },
    {
      "epoch": 6.033482006476446,
      "grad_norm": 8.957596778869629,
      "learning_rate": 1.9876086158364952e-05,
      "loss": 1.0569,
      "step": 197500
    },
    {
      "epoch": 6.036536934074663,
      "grad_norm": 11.061244010925293,
      "learning_rate": 1.986078815322482e-05,
      "loss": 1.0122,
      "step": 197600
    },
    {
      "epoch": 6.0395918616728785,
      "grad_norm": 16.330211639404297,
      "learning_rate": 1.984549014808469e-05,
      "loss": 1.6281,
      "step": 197700
    },
    {
      "epoch": 6.042646789271094,
      "grad_norm": 5.61469030380249,
      "learning_rate": 1.9830192142944562e-05,
      "loss": 0.9955,
      "step": 197800
    },
    {
      "epoch": 6.04570171686931,
      "grad_norm": 4.7612786293029785,
      "learning_rate": 1.9814894137804433e-05,
      "loss": 0.8676,
      "step": 197900
    },
    {
      "epoch": 6.0487566444675265,
      "grad_norm": 17.290145874023438,
      "learning_rate": 1.97995961326643e-05,
      "loss": 1.308,
      "step": 198000
    },
    {
      "epoch": 6.051811572065742,
      "grad_norm": 2.9975016117095947,
      "learning_rate": 1.9784298127524172e-05,
      "loss": 1.0552,
      "step": 198100
    },
    {
      "epoch": 6.054866499663958,
      "grad_norm": 5.638766288757324,
      "learning_rate": 1.9769000122384043e-05,
      "loss": 1.4797,
      "step": 198200
    },
    {
      "epoch": 6.0579214272621735,
      "grad_norm": 7.252283096313477,
      "learning_rate": 1.9753702117243914e-05,
      "loss": 1.1544,
      "step": 198300
    },
    {
      "epoch": 6.06097635486039,
      "grad_norm": 12.537910461425781,
      "learning_rate": 1.9738404112103782e-05,
      "loss": 1.267,
      "step": 198400
    },
    {
      "epoch": 6.064031282458606,
      "grad_norm": 5.209641456604004,
      "learning_rate": 1.9723106106963653e-05,
      "loss": 1.2551,
      "step": 198500
    },
    {
      "epoch": 6.0670862100568215,
      "grad_norm": 5.152153968811035,
      "learning_rate": 1.9707808101823524e-05,
      "loss": 1.0689,
      "step": 198600
    },
    {
      "epoch": 6.070141137655037,
      "grad_norm": 21.221385955810547,
      "learning_rate": 1.9692510096683396e-05,
      "loss": 1.2884,
      "step": 198700
    },
    {
      "epoch": 6.073196065253254,
      "grad_norm": 14.44279956817627,
      "learning_rate": 1.9677212091543263e-05,
      "loss": 1.0771,
      "step": 198800
    },
    {
      "epoch": 6.076250992851469,
      "grad_norm": 9.928838729858398,
      "learning_rate": 1.966191408640313e-05,
      "loss": 1.0506,
      "step": 198900
    },
    {
      "epoch": 6.079305920449685,
      "grad_norm": 2.891791582107544,
      "learning_rate": 1.9646616081263002e-05,
      "loss": 1.1673,
      "step": 199000
    },
    {
      "epoch": 6.082360848047902,
      "grad_norm": 11.18900203704834,
      "learning_rate": 1.9631318076122873e-05,
      "loss": 1.0406,
      "step": 199100
    },
    {
      "epoch": 6.085415775646117,
      "grad_norm": 6.0363993644714355,
      "learning_rate": 1.9616020070982745e-05,
      "loss": 1.1409,
      "step": 199200
    },
    {
      "epoch": 6.088470703244333,
      "grad_norm": 5.4089508056640625,
      "learning_rate": 1.9600722065842612e-05,
      "loss": 1.3594,
      "step": 199300
    },
    {
      "epoch": 6.091525630842549,
      "grad_norm": 5.459392070770264,
      "learning_rate": 1.9585424060702484e-05,
      "loss": 1.2248,
      "step": 199400
    },
    {
      "epoch": 6.094580558440765,
      "grad_norm": 2.2770681381225586,
      "learning_rate": 1.9570279035613755e-05,
      "loss": 1.1575,
      "step": 199500
    },
    {
      "epoch": 6.097635486038981,
      "grad_norm": 3.4792544841766357,
      "learning_rate": 1.9554981030473626e-05,
      "loss": 1.1068,
      "step": 199600
    },
    {
      "epoch": 6.100690413637197,
      "grad_norm": 0.7737553715705872,
      "learning_rate": 1.9539683025333497e-05,
      "loss": 1.0847,
      "step": 199700
    },
    {
      "epoch": 6.103745341235412,
      "grad_norm": 5.00324821472168,
      "learning_rate": 1.952438502019337e-05,
      "loss": 1.234,
      "step": 199800
    },
    {
      "epoch": 6.106800268833629,
      "grad_norm": 1.3894213438034058,
      "learning_rate": 1.9509087015053236e-05,
      "loss": 1.3646,
      "step": 199900
    },
    {
      "epoch": 6.109855196431845,
      "grad_norm": 6.558831214904785,
      "learning_rate": 1.9493789009913107e-05,
      "loss": 1.2715,
      "step": 200000
    },
    {
      "epoch": 6.11291012403006,
      "grad_norm": 4.3237762451171875,
      "learning_rate": 1.947849100477298e-05,
      "loss": 1.1757,
      "step": 200100
    },
    {
      "epoch": 6.115965051628276,
      "grad_norm": 24.602190017700195,
      "learning_rate": 1.946319299963285e-05,
      "loss": 1.1975,
      "step": 200200
    },
    {
      "epoch": 6.1190199792264925,
      "grad_norm": 9.489306449890137,
      "learning_rate": 1.9447894994492717e-05,
      "loss": 1.2334,
      "step": 200300
    },
    {
      "epoch": 6.122074906824708,
      "grad_norm": 1.5155296325683594,
      "learning_rate": 1.943259698935259e-05,
      "loss": 1.2309,
      "step": 200400
    },
    {
      "epoch": 6.125129834422924,
      "grad_norm": 15.262107849121094,
      "learning_rate": 1.941729898421246e-05,
      "loss": 1.1219,
      "step": 200500
    },
    {
      "epoch": 6.1281847620211405,
      "grad_norm": 7.496032238006592,
      "learning_rate": 1.940200097907233e-05,
      "loss": 1.178,
      "step": 200600
    },
    {
      "epoch": 6.131239689619356,
      "grad_norm": 7.0843305587768555,
      "learning_rate": 1.93867029739322e-05,
      "loss": 1.1752,
      "step": 200700
    },
    {
      "epoch": 6.134294617217572,
      "grad_norm": 6.570627689361572,
      "learning_rate": 1.937140496879207e-05,
      "loss": 1.0997,
      "step": 200800
    },
    {
      "epoch": 6.1373495448157875,
      "grad_norm": 15.903374671936035,
      "learning_rate": 1.935610696365194e-05,
      "loss": 1.1207,
      "step": 200900
    },
    {
      "epoch": 6.140404472414004,
      "grad_norm": 5.491482257843018,
      "learning_rate": 1.9340808958511812e-05,
      "loss": 1.2633,
      "step": 201000
    },
    {
      "epoch": 6.14345940001222,
      "grad_norm": 9.441786766052246,
      "learning_rate": 1.932551095337168e-05,
      "loss": 0.9868,
      "step": 201100
    },
    {
      "epoch": 6.1465143276104355,
      "grad_norm": 10.245373725891113,
      "learning_rate": 1.931021294823155e-05,
      "loss": 1.2599,
      "step": 201200
    },
    {
      "epoch": 6.149569255208651,
      "grad_norm": 20.67123031616211,
      "learning_rate": 1.9294914943091422e-05,
      "loss": 1.1123,
      "step": 201300
    },
    {
      "epoch": 6.152624182806868,
      "grad_norm": 3.138038396835327,
      "learning_rate": 1.9279616937951293e-05,
      "loss": 1.113,
      "step": 201400
    },
    {
      "epoch": 6.155679110405083,
      "grad_norm": 3.0966243743896484,
      "learning_rate": 1.926431893281116e-05,
      "loss": 1.2311,
      "step": 201500
    },
    {
      "epoch": 6.158734038003299,
      "grad_norm": 6.163281440734863,
      "learning_rate": 1.9249020927671032e-05,
      "loss": 1.0435,
      "step": 201600
    },
    {
      "epoch": 6.161788965601515,
      "grad_norm": 5.311619758605957,
      "learning_rate": 1.9233722922530903e-05,
      "loss": 1.1483,
      "step": 201700
    },
    {
      "epoch": 6.164843893199731,
      "grad_norm": 7.495104789733887,
      "learning_rate": 1.9218424917390775e-05,
      "loss": 1.0817,
      "step": 201800
    },
    {
      "epoch": 6.167898820797947,
      "grad_norm": 5.036028861999512,
      "learning_rate": 1.9203126912250642e-05,
      "loss": 1.147,
      "step": 201900
    },
    {
      "epoch": 6.170953748396163,
      "grad_norm": 11.376052856445312,
      "learning_rate": 1.9187828907110514e-05,
      "loss": 1.2904,
      "step": 202000
    },
    {
      "epoch": 6.174008675994379,
      "grad_norm": 2.396482467651367,
      "learning_rate": 1.9172530901970385e-05,
      "loss": 0.9409,
      "step": 202100
    },
    {
      "epoch": 6.177063603592595,
      "grad_norm": 12.605530738830566,
      "learning_rate": 1.9157232896830256e-05,
      "loss": 1.2427,
      "step": 202200
    },
    {
      "epoch": 6.180118531190811,
      "grad_norm": 5.409971714019775,
      "learning_rate": 1.9141934891690124e-05,
      "loss": 1.2063,
      "step": 202300
    },
    {
      "epoch": 6.183173458789026,
      "grad_norm": 11.71444034576416,
      "learning_rate": 1.9126636886549995e-05,
      "loss": 1.0652,
      "step": 202400
    },
    {
      "epoch": 6.186228386387243,
      "grad_norm": 6.157597064971924,
      "learning_rate": 1.9111338881409866e-05,
      "loss": 1.0756,
      "step": 202500
    },
    {
      "epoch": 6.189283313985459,
      "grad_norm": 12.192460060119629,
      "learning_rate": 1.9096040876269737e-05,
      "loss": 0.9874,
      "step": 202600
    },
    {
      "epoch": 6.192338241583674,
      "grad_norm": 10.153812408447266,
      "learning_rate": 1.9080742871129605e-05,
      "loss": 1.2897,
      "step": 202700
    },
    {
      "epoch": 6.19539316918189,
      "grad_norm": 7.526157855987549,
      "learning_rate": 1.9065444865989476e-05,
      "loss": 1.1976,
      "step": 202800
    },
    {
      "epoch": 6.1984480967801066,
      "grad_norm": 6.983723163604736,
      "learning_rate": 1.9050146860849347e-05,
      "loss": 1.3308,
      "step": 202900
    },
    {
      "epoch": 6.201503024378322,
      "grad_norm": 5.48733377456665,
      "learning_rate": 1.9034848855709218e-05,
      "loss": 1.0038,
      "step": 203000
    },
    {
      "epoch": 6.204557951976538,
      "grad_norm": 4.00633430480957,
      "learning_rate": 1.9019550850569086e-05,
      "loss": 0.9692,
      "step": 203100
    },
    {
      "epoch": 6.2076128795747545,
      "grad_norm": 7.790951728820801,
      "learning_rate": 1.9004252845428957e-05,
      "loss": 1.3422,
      "step": 203200
    },
    {
      "epoch": 6.21066780717297,
      "grad_norm": 4.072429656982422,
      "learning_rate": 1.898895484028883e-05,
      "loss": 1.322,
      "step": 203300
    },
    {
      "epoch": 6.213722734771186,
      "grad_norm": 6.779109001159668,
      "learning_rate": 1.89736568351487e-05,
      "loss": 0.9392,
      "step": 203400
    },
    {
      "epoch": 6.2167776623694015,
      "grad_norm": 1.8601343631744385,
      "learning_rate": 1.895851181005997e-05,
      "loss": 1.1972,
      "step": 203500
    },
    {
      "epoch": 6.219832589967618,
      "grad_norm": 4.4953508377075195,
      "learning_rate": 1.894336678497124e-05,
      "loss": 1.2135,
      "step": 203600
    },
    {
      "epoch": 6.222887517565834,
      "grad_norm": 4.9885735511779785,
      "learning_rate": 1.892806877983111e-05,
      "loss": 1.2205,
      "step": 203700
    },
    {
      "epoch": 6.2259424451640495,
      "grad_norm": 2.344512701034546,
      "learning_rate": 1.891277077469098e-05,
      "loss": 1.0766,
      "step": 203800
    },
    {
      "epoch": 6.228997372762265,
      "grad_norm": 26.173669815063477,
      "learning_rate": 1.889747276955085e-05,
      "loss": 1.1281,
      "step": 203900
    },
    {
      "epoch": 6.232052300360482,
      "grad_norm": 7.559381484985352,
      "learning_rate": 1.888217476441072e-05,
      "loss": 1.1236,
      "step": 204000
    },
    {
      "epoch": 6.235107227958697,
      "grad_norm": 9.910603523254395,
      "learning_rate": 1.886687675927059e-05,
      "loss": 1.3141,
      "step": 204100
    },
    {
      "epoch": 6.238162155556913,
      "grad_norm": 2.8729636669158936,
      "learning_rate": 1.8851578754130462e-05,
      "loss": 1.1617,
      "step": 204200
    },
    {
      "epoch": 6.241217083155129,
      "grad_norm": 4.2151055335998535,
      "learning_rate": 1.883628074899033e-05,
      "loss": 1.1621,
      "step": 204300
    },
    {
      "epoch": 6.244272010753345,
      "grad_norm": 5.4340410232543945,
      "learning_rate": 1.88209827438502e-05,
      "loss": 1.1694,
      "step": 204400
    },
    {
      "epoch": 6.247326938351561,
      "grad_norm": 5.660393238067627,
      "learning_rate": 1.8805684738710072e-05,
      "loss": 1.1854,
      "step": 204500
    },
    {
      "epoch": 6.250381865949777,
      "grad_norm": 4.4400177001953125,
      "learning_rate": 1.8790386733569944e-05,
      "loss": 1.2309,
      "step": 204600
    },
    {
      "epoch": 6.253436793547993,
      "grad_norm": 6.119187831878662,
      "learning_rate": 1.877508872842981e-05,
      "loss": 1.4101,
      "step": 204700
    },
    {
      "epoch": 6.256491721146209,
      "grad_norm": 8.3779296875,
      "learning_rate": 1.8759790723289683e-05,
      "loss": 1.4293,
      "step": 204800
    },
    {
      "epoch": 6.259546648744425,
      "grad_norm": 13.137105941772461,
      "learning_rate": 1.8744492718149554e-05,
      "loss": 1.0669,
      "step": 204900
    },
    {
      "epoch": 6.26260157634264,
      "grad_norm": 10.320351600646973,
      "learning_rate": 1.8729194713009425e-05,
      "loss": 1.1582,
      "step": 205000
    },
    {
      "epoch": 6.265656503940857,
      "grad_norm": 5.954596996307373,
      "learning_rate": 1.8713896707869293e-05,
      "loss": 1.1082,
      "step": 205100
    },
    {
      "epoch": 6.268711431539073,
      "grad_norm": 8.020731925964355,
      "learning_rate": 1.8698598702729164e-05,
      "loss": 1.0602,
      "step": 205200
    },
    {
      "epoch": 6.271766359137288,
      "grad_norm": 20.765810012817383,
      "learning_rate": 1.8683300697589035e-05,
      "loss": 0.9351,
      "step": 205300
    },
    {
      "epoch": 6.274821286735504,
      "grad_norm": 5.073605537414551,
      "learning_rate": 1.8668002692448906e-05,
      "loss": 1.0308,
      "step": 205400
    },
    {
      "epoch": 6.277876214333721,
      "grad_norm": 7.651671886444092,
      "learning_rate": 1.8652704687308774e-05,
      "loss": 1.17,
      "step": 205500
    },
    {
      "epoch": 6.280931141931936,
      "grad_norm": 5.585090160369873,
      "learning_rate": 1.8637406682168645e-05,
      "loss": 1.0851,
      "step": 205600
    },
    {
      "epoch": 6.283986069530152,
      "grad_norm": 6.514379024505615,
      "learning_rate": 1.8622108677028516e-05,
      "loss": 1.1694,
      "step": 205700
    },
    {
      "epoch": 6.2870409971283685,
      "grad_norm": 15.221776962280273,
      "learning_rate": 1.8606810671888387e-05,
      "loss": 1.1695,
      "step": 205800
    },
    {
      "epoch": 6.290095924726584,
      "grad_norm": 2.7496187686920166,
      "learning_rate": 1.8591512666748255e-05,
      "loss": 1.2547,
      "step": 205900
    },
    {
      "epoch": 6.2931508523248,
      "grad_norm": 21.13014030456543,
      "learning_rate": 1.8576214661608126e-05,
      "loss": 1.0187,
      "step": 206000
    },
    {
      "epoch": 6.2962057799230156,
      "grad_norm": 4.162718296051025,
      "learning_rate": 1.8560916656467997e-05,
      "loss": 1.0812,
      "step": 206100
    },
    {
      "epoch": 6.299260707521232,
      "grad_norm": 13.786234855651855,
      "learning_rate": 1.854561865132787e-05,
      "loss": 1.5332,
      "step": 206200
    },
    {
      "epoch": 6.302315635119448,
      "grad_norm": 19.818574905395508,
      "learning_rate": 1.8530320646187736e-05,
      "loss": 1.0737,
      "step": 206300
    },
    {
      "epoch": 6.3053705627176635,
      "grad_norm": 10.488554954528809,
      "learning_rate": 1.8515022641047608e-05,
      "loss": 1.069,
      "step": 206400
    },
    {
      "epoch": 6.308425490315879,
      "grad_norm": 13.240741729736328,
      "learning_rate": 1.849972463590748e-05,
      "loss": 0.979,
      "step": 206500
    },
    {
      "epoch": 6.311480417914096,
      "grad_norm": 15.739484786987305,
      "learning_rate": 1.848457961081875e-05,
      "loss": 1.2026,
      "step": 206600
    },
    {
      "epoch": 6.314535345512311,
      "grad_norm": 8.837204933166504,
      "learning_rate": 1.846928160567862e-05,
      "loss": 1.3493,
      "step": 206700
    },
    {
      "epoch": 6.317590273110527,
      "grad_norm": 4.217745304107666,
      "learning_rate": 1.8453983600538492e-05,
      "loss": 1.2004,
      "step": 206800
    },
    {
      "epoch": 6.320645200708743,
      "grad_norm": 13.267827987670898,
      "learning_rate": 1.843868559539836e-05,
      "loss": 1.0603,
      "step": 206900
    },
    {
      "epoch": 6.323700128306959,
      "grad_norm": 1.135472059249878,
      "learning_rate": 1.842338759025823e-05,
      "loss": 1.1188,
      "step": 207000
    },
    {
      "epoch": 6.326755055905175,
      "grad_norm": 5.91626501083374,
      "learning_rate": 1.8408089585118102e-05,
      "loss": 1.2033,
      "step": 207100
    },
    {
      "epoch": 6.329809983503391,
      "grad_norm": 1.7794185876846313,
      "learning_rate": 1.8392791579977974e-05,
      "loss": 0.9587,
      "step": 207200
    },
    {
      "epoch": 6.332864911101607,
      "grad_norm": 16.53619956970215,
      "learning_rate": 1.837749357483784e-05,
      "loss": 1.1301,
      "step": 207300
    },
    {
      "epoch": 6.335919838699823,
      "grad_norm": 7.456695556640625,
      "learning_rate": 1.8362195569697713e-05,
      "loss": 0.9313,
      "step": 207400
    },
    {
      "epoch": 6.338974766298039,
      "grad_norm": 1.256122350692749,
      "learning_rate": 1.8346897564557584e-05,
      "loss": 1.3876,
      "step": 207500
    },
    {
      "epoch": 6.342029693896254,
      "grad_norm": 5.864501476287842,
      "learning_rate": 1.8331599559417455e-05,
      "loss": 0.9638,
      "step": 207600
    },
    {
      "epoch": 6.345084621494471,
      "grad_norm": 5.211721420288086,
      "learning_rate": 1.8316301554277323e-05,
      "loss": 1.2297,
      "step": 207700
    },
    {
      "epoch": 6.348139549092687,
      "grad_norm": 8.266345977783203,
      "learning_rate": 1.8301003549137194e-05,
      "loss": 1.1224,
      "step": 207800
    },
    {
      "epoch": 6.351194476690902,
      "grad_norm": 7.6695556640625,
      "learning_rate": 1.8285705543997065e-05,
      "loss": 1.2214,
      "step": 207900
    },
    {
      "epoch": 6.354249404289118,
      "grad_norm": 4.3584184646606445,
      "learning_rate": 1.8270407538856936e-05,
      "loss": 1.1276,
      "step": 208000
    },
    {
      "epoch": 6.357304331887335,
      "grad_norm": 2.5651164054870605,
      "learning_rate": 1.8255109533716804e-05,
      "loss": 0.9946,
      "step": 208100
    },
    {
      "epoch": 6.36035925948555,
      "grad_norm": 6.8474602699279785,
      "learning_rate": 1.8239811528576675e-05,
      "loss": 1.2859,
      "step": 208200
    },
    {
      "epoch": 6.363414187083766,
      "grad_norm": 5.764113426208496,
      "learning_rate": 1.8224513523436546e-05,
      "loss": 1.1951,
      "step": 208300
    },
    {
      "epoch": 6.366469114681982,
      "grad_norm": 3.486423969268799,
      "learning_rate": 1.8209215518296417e-05,
      "loss": 1.5135,
      "step": 208400
    },
    {
      "epoch": 6.369524042280198,
      "grad_norm": 6.736783027648926,
      "learning_rate": 1.8193917513156285e-05,
      "loss": 0.8972,
      "step": 208500
    },
    {
      "epoch": 6.372578969878414,
      "grad_norm": 7.037610054016113,
      "learning_rate": 1.8178619508016156e-05,
      "loss": 1.3713,
      "step": 208600
    },
    {
      "epoch": 6.37563389747663,
      "grad_norm": 4.605196475982666,
      "learning_rate": 1.8163321502876027e-05,
      "loss": 1.1864,
      "step": 208700
    },
    {
      "epoch": 6.378688825074846,
      "grad_norm": 11.924447059631348,
      "learning_rate": 1.81480234977359e-05,
      "loss": 1.1253,
      "step": 208800
    },
    {
      "epoch": 6.381743752673062,
      "grad_norm": 7.568887710571289,
      "learning_rate": 1.8132725492595766e-05,
      "loss": 1.0449,
      "step": 208900
    },
    {
      "epoch": 6.3847986802712775,
      "grad_norm": 4.9831671714782715,
      "learning_rate": 1.8117427487455637e-05,
      "loss": 1.2972,
      "step": 209000
    },
    {
      "epoch": 6.387853607869493,
      "grad_norm": 6.626806735992432,
      "learning_rate": 1.810212948231551e-05,
      "loss": 1.2032,
      "step": 209100
    },
    {
      "epoch": 6.39090853546771,
      "grad_norm": 4.027082920074463,
      "learning_rate": 1.808683147717538e-05,
      "loss": 1.1557,
      "step": 209200
    },
    {
      "epoch": 6.3939634630659254,
      "grad_norm": 3.5578677654266357,
      "learning_rate": 1.8071533472035248e-05,
      "loss": 1.2836,
      "step": 209300
    },
    {
      "epoch": 6.397018390664141,
      "grad_norm": 6.211548805236816,
      "learning_rate": 1.805623546689512e-05,
      "loss": 1.1505,
      "step": 209400
    },
    {
      "epoch": 6.400073318262357,
      "grad_norm": 21.222150802612305,
      "learning_rate": 1.804093746175499e-05,
      "loss": 1.266,
      "step": 209500
    },
    {
      "epoch": 6.403128245860573,
      "grad_norm": 4.08786153793335,
      "learning_rate": 1.802563945661486e-05,
      "loss": 1.0255,
      "step": 209600
    },
    {
      "epoch": 6.406183173458789,
      "grad_norm": 34.38716125488281,
      "learning_rate": 1.801034145147473e-05,
      "loss": 1.2649,
      "step": 209700
    },
    {
      "epoch": 6.409238101057005,
      "grad_norm": 5.348664283752441,
      "learning_rate": 1.79950434463346e-05,
      "loss": 1.1752,
      "step": 209800
    },
    {
      "epoch": 6.41229302865522,
      "grad_norm": 2.724579334259033,
      "learning_rate": 1.797974544119447e-05,
      "loss": 1.1778,
      "step": 209900
    },
    {
      "epoch": 6.415347956253437,
      "grad_norm": 4.925905227661133,
      "learning_rate": 1.796444743605434e-05,
      "loss": 1.2821,
      "step": 210000
    },
    {
      "epoch": 6.418402883851653,
      "grad_norm": 13.180278778076172,
      "learning_rate": 1.7949149430914207e-05,
      "loss": 1.3623,
      "step": 210100
    },
    {
      "epoch": 6.421457811449868,
      "grad_norm": 25.796751022338867,
      "learning_rate": 1.7933851425774078e-05,
      "loss": 1.2821,
      "step": 210200
    },
    {
      "epoch": 6.424512739048085,
      "grad_norm": 3.784736394882202,
      "learning_rate": 1.791855342063395e-05,
      "loss": 1.0298,
      "step": 210300
    },
    {
      "epoch": 6.427567666646301,
      "grad_norm": 7.122948169708252,
      "learning_rate": 1.790325541549382e-05,
      "loss": 1.1683,
      "step": 210400
    },
    {
      "epoch": 6.430622594244516,
      "grad_norm": 7.788839817047119,
      "learning_rate": 1.7887957410353688e-05,
      "loss": 1.0356,
      "step": 210500
    },
    {
      "epoch": 6.433677521842732,
      "grad_norm": 4.615478992462158,
      "learning_rate": 1.7872812385264963e-05,
      "loss": 1.326,
      "step": 210600
    },
    {
      "epoch": 6.436732449440949,
      "grad_norm": 18.982412338256836,
      "learning_rate": 1.785751438012483e-05,
      "loss": 0.952,
      "step": 210700
    },
    {
      "epoch": 6.439787377039164,
      "grad_norm": 5.564967155456543,
      "learning_rate": 1.78422163749847e-05,
      "loss": 1.3681,
      "step": 210800
    },
    {
      "epoch": 6.44284230463738,
      "grad_norm": 5.302816867828369,
      "learning_rate": 1.7826918369844573e-05,
      "loss": 1.0967,
      "step": 210900
    },
    {
      "epoch": 6.445897232235596,
      "grad_norm": 9.340934753417969,
      "learning_rate": 1.7811773344755844e-05,
      "loss": 1.2459,
      "step": 211000
    },
    {
      "epoch": 6.448952159833812,
      "grad_norm": 5.11415958404541,
      "learning_rate": 1.7796475339615715e-05,
      "loss": 0.9813,
      "step": 211100
    },
    {
      "epoch": 6.452007087432028,
      "grad_norm": 2.0551705360412598,
      "learning_rate": 1.7781177334475586e-05,
      "loss": 1.0639,
      "step": 211200
    },
    {
      "epoch": 6.455062015030244,
      "grad_norm": 8.070032119750977,
      "learning_rate": 1.7765879329335454e-05,
      "loss": 1.0419,
      "step": 211300
    },
    {
      "epoch": 6.458116942628459,
      "grad_norm": 7.958362579345703,
      "learning_rate": 1.7750581324195325e-05,
      "loss": 0.9593,
      "step": 211400
    },
    {
      "epoch": 6.461171870226676,
      "grad_norm": 5.6488800048828125,
      "learning_rate": 1.7735283319055196e-05,
      "loss": 1.063,
      "step": 211500
    },
    {
      "epoch": 6.4642267978248915,
      "grad_norm": 7.8869242668151855,
      "learning_rate": 1.7719985313915068e-05,
      "loss": 1.3276,
      "step": 211600
    },
    {
      "epoch": 6.467281725423107,
      "grad_norm": 5.1599016189575195,
      "learning_rate": 1.7704687308774935e-05,
      "loss": 0.8885,
      "step": 211700
    },
    {
      "epoch": 6.470336653021324,
      "grad_norm": 5.28355598449707,
      "learning_rate": 1.7689389303634807e-05,
      "loss": 1.3002,
      "step": 211800
    },
    {
      "epoch": 6.4733915806195395,
      "grad_norm": 11.972784042358398,
      "learning_rate": 1.7674091298494678e-05,
      "loss": 1.0144,
      "step": 211900
    },
    {
      "epoch": 6.476446508217755,
      "grad_norm": 3.3315412998199463,
      "learning_rate": 1.765879329335455e-05,
      "loss": 0.9171,
      "step": 212000
    },
    {
      "epoch": 6.479501435815971,
      "grad_norm": 6.229414463043213,
      "learning_rate": 1.7643495288214417e-05,
      "loss": 1.2993,
      "step": 212100
    },
    {
      "epoch": 6.482556363414187,
      "grad_norm": 11.299790382385254,
      "learning_rate": 1.7628197283074288e-05,
      "loss": 1.0803,
      "step": 212200
    },
    {
      "epoch": 6.485611291012403,
      "grad_norm": 8.365095138549805,
      "learning_rate": 1.761289927793416e-05,
      "loss": 1.323,
      "step": 212300
    },
    {
      "epoch": 6.488666218610619,
      "grad_norm": 11.41671371459961,
      "learning_rate": 1.759760127279403e-05,
      "loss": 1.1004,
      "step": 212400
    },
    {
      "epoch": 6.4917211462088344,
      "grad_norm": 3.927541971206665,
      "learning_rate": 1.7582303267653898e-05,
      "loss": 1.0298,
      "step": 212500
    },
    {
      "epoch": 6.494776073807051,
      "grad_norm": 5.735844135284424,
      "learning_rate": 1.756700526251377e-05,
      "loss": 1.2621,
      "step": 212600
    },
    {
      "epoch": 6.497831001405267,
      "grad_norm": 5.7011284828186035,
      "learning_rate": 1.755170725737364e-05,
      "loss": 1.1243,
      "step": 212700
    },
    {
      "epoch": 6.500885929003482,
      "grad_norm": 9.130974769592285,
      "learning_rate": 1.753640925223351e-05,
      "loss": 1.0719,
      "step": 212800
    },
    {
      "epoch": 6.503940856601698,
      "grad_norm": 7.565042018890381,
      "learning_rate": 1.752111124709338e-05,
      "loss": 0.9947,
      "step": 212900
    },
    {
      "epoch": 6.506995784199915,
      "grad_norm": 3.490960121154785,
      "learning_rate": 1.750581324195325e-05,
      "loss": 1.0295,
      "step": 213000
    },
    {
      "epoch": 6.51005071179813,
      "grad_norm": 5.046875953674316,
      "learning_rate": 1.749051523681312e-05,
      "loss": 1.0738,
      "step": 213100
    },
    {
      "epoch": 6.513105639396346,
      "grad_norm": 2.79036808013916,
      "learning_rate": 1.7475217231672993e-05,
      "loss": 1.3818,
      "step": 213200
    },
    {
      "epoch": 6.516160566994563,
      "grad_norm": 0.009335226379334927,
      "learning_rate": 1.7460072206584264e-05,
      "loss": 1.1112,
      "step": 213300
    },
    {
      "epoch": 6.519215494592778,
      "grad_norm": 6.8894243240356445,
      "learning_rate": 1.7444774201444135e-05,
      "loss": 1.1441,
      "step": 213400
    },
    {
      "epoch": 6.522270422190994,
      "grad_norm": 4.209587574005127,
      "learning_rate": 1.7429476196304003e-05,
      "loss": 1.0272,
      "step": 213500
    },
    {
      "epoch": 6.52532534978921,
      "grad_norm": 2.185170888900757,
      "learning_rate": 1.7414178191163874e-05,
      "loss": 1.0875,
      "step": 213600
    },
    {
      "epoch": 6.528380277387426,
      "grad_norm": 14.174199104309082,
      "learning_rate": 1.7398880186023745e-05,
      "loss": 1.089,
      "step": 213700
    },
    {
      "epoch": 6.531435204985642,
      "grad_norm": 9.62555980682373,
      "learning_rate": 1.7383582180883616e-05,
      "loss": 1.1095,
      "step": 213800
    },
    {
      "epoch": 6.534490132583858,
      "grad_norm": 5.337947368621826,
      "learning_rate": 1.7368284175743484e-05,
      "loss": 0.9957,
      "step": 213900
    },
    {
      "epoch": 6.537545060182074,
      "grad_norm": 4.797677993774414,
      "learning_rate": 1.7352986170603355e-05,
      "loss": 1.2343,
      "step": 214000
    },
    {
      "epoch": 6.54059998778029,
      "grad_norm": 7.560909271240234,
      "learning_rate": 1.7337688165463226e-05,
      "loss": 1.1702,
      "step": 214100
    },
    {
      "epoch": 6.5436549153785055,
      "grad_norm": 3.0463943481445312,
      "learning_rate": 1.7322390160323098e-05,
      "loss": 1.2553,
      "step": 214200
    },
    {
      "epoch": 6.546709842976721,
      "grad_norm": 6.867596626281738,
      "learning_rate": 1.7307092155182965e-05,
      "loss": 1.0156,
      "step": 214300
    },
    {
      "epoch": 6.549764770574937,
      "grad_norm": 7.988159656524658,
      "learning_rate": 1.7291794150042837e-05,
      "loss": 1.2007,
      "step": 214400
    },
    {
      "epoch": 6.5528196981731535,
      "grad_norm": 4.5299482345581055,
      "learning_rate": 1.7276496144902708e-05,
      "loss": 1.0585,
      "step": 214500
    },
    {
      "epoch": 6.555874625771369,
      "grad_norm": 17.113630294799805,
      "learning_rate": 1.726119813976258e-05,
      "loss": 1.226,
      "step": 214600
    },
    {
      "epoch": 6.558929553369585,
      "grad_norm": 6.469338417053223,
      "learning_rate": 1.7245900134622443e-05,
      "loss": 1.5139,
      "step": 214700
    },
    {
      "epoch": 6.561984480967801,
      "grad_norm": 9.526185989379883,
      "learning_rate": 1.7230602129482314e-05,
      "loss": 0.9568,
      "step": 214800
    },
    {
      "epoch": 6.565039408566017,
      "grad_norm": 4.849943161010742,
      "learning_rate": 1.7215304124342186e-05,
      "loss": 1.1666,
      "step": 214900
    },
    {
      "epoch": 6.568094336164233,
      "grad_norm": 4.016180515289307,
      "learning_rate": 1.7200006119202057e-05,
      "loss": 1.0448,
      "step": 215000
    },
    {
      "epoch": 6.5711492637624485,
      "grad_norm": 3.0566747188568115,
      "learning_rate": 1.7184708114061924e-05,
      "loss": 1.2607,
      "step": 215100
    },
    {
      "epoch": 6.574204191360665,
      "grad_norm": 13.017189025878906,
      "learning_rate": 1.7169410108921796e-05,
      "loss": 1.2375,
      "step": 215200
    },
    {
      "epoch": 6.577259118958881,
      "grad_norm": 3.1500213146209717,
      "learning_rate": 1.7154112103781667e-05,
      "loss": 1.1126,
      "step": 215300
    },
    {
      "epoch": 6.580314046557096,
      "grad_norm": 15.420360565185547,
      "learning_rate": 1.7138814098641538e-05,
      "loss": 1.0861,
      "step": 215400
    },
    {
      "epoch": 6.583368974155313,
      "grad_norm": 15.716479301452637,
      "learning_rate": 1.712366907355281e-05,
      "loss": 1.357,
      "step": 215500
    },
    {
      "epoch": 6.586423901753529,
      "grad_norm": 4.730207443237305,
      "learning_rate": 1.710837106841268e-05,
      "loss": 1.0504,
      "step": 215600
    },
    {
      "epoch": 6.589478829351744,
      "grad_norm": 5.073116779327393,
      "learning_rate": 1.7093073063272548e-05,
      "loss": 1.3092,
      "step": 215700
    },
    {
      "epoch": 6.59253375694996,
      "grad_norm": 7.078702449798584,
      "learning_rate": 1.707777505813242e-05,
      "loss": 1.2368,
      "step": 215800
    },
    {
      "epoch": 6.595588684548177,
      "grad_norm": 6.552230358123779,
      "learning_rate": 1.706247705299229e-05,
      "loss": 1.1605,
      "step": 215900
    },
    {
      "epoch": 6.598643612146392,
      "grad_norm": 12.837909698486328,
      "learning_rate": 1.704717904785216e-05,
      "loss": 1.5417,
      "step": 216000
    },
    {
      "epoch": 6.601698539744608,
      "grad_norm": 3.1361231803894043,
      "learning_rate": 1.703188104271203e-05,
      "loss": 1.187,
      "step": 216100
    },
    {
      "epoch": 6.604753467342824,
      "grad_norm": 3.2314138412475586,
      "learning_rate": 1.70165830375719e-05,
      "loss": 1.1333,
      "step": 216200
    },
    {
      "epoch": 6.60780839494104,
      "grad_norm": 7.735902309417725,
      "learning_rate": 1.7001285032431772e-05,
      "loss": 1.2677,
      "step": 216300
    },
    {
      "epoch": 6.610863322539256,
      "grad_norm": 4.2524189949035645,
      "learning_rate": 1.6985987027291643e-05,
      "loss": 1.195,
      "step": 216400
    },
    {
      "epoch": 6.613918250137472,
      "grad_norm": 5.863406658172607,
      "learning_rate": 1.697068902215151e-05,
      "loss": 1.0911,
      "step": 216500
    },
    {
      "epoch": 6.616973177735687,
      "grad_norm": 6.36102294921875,
      "learning_rate": 1.6955391017011382e-05,
      "loss": 1.036,
      "step": 216600
    },
    {
      "epoch": 6.620028105333904,
      "grad_norm": 3.2900092601776123,
      "learning_rate": 1.6940093011871253e-05,
      "loss": 1.2053,
      "step": 216700
    },
    {
      "epoch": 6.6230830329321195,
      "grad_norm": 0.4474376142024994,
      "learning_rate": 1.6924795006731124e-05,
      "loss": 1.1469,
      "step": 216800
    },
    {
      "epoch": 6.626137960530335,
      "grad_norm": 22.14536476135254,
      "learning_rate": 1.6909497001590992e-05,
      "loss": 1.2501,
      "step": 216900
    },
    {
      "epoch": 6.629192888128552,
      "grad_norm": 3.733264207839966,
      "learning_rate": 1.6894198996450863e-05,
      "loss": 1.0177,
      "step": 217000
    },
    {
      "epoch": 6.6322478157267675,
      "grad_norm": 3.5817043781280518,
      "learning_rate": 1.6878900991310734e-05,
      "loss": 1.3182,
      "step": 217100
    },
    {
      "epoch": 6.635302743324983,
      "grad_norm": 5.059553146362305,
      "learning_rate": 1.6863602986170605e-05,
      "loss": 1.3476,
      "step": 217200
    },
    {
      "epoch": 6.638357670923199,
      "grad_norm": 3.9976773262023926,
      "learning_rate": 1.6848304981030473e-05,
      "loss": 1.0562,
      "step": 217300
    },
    {
      "epoch": 6.641412598521415,
      "grad_norm": 16.386789321899414,
      "learning_rate": 1.6833006975890344e-05,
      "loss": 1.0905,
      "step": 217400
    },
    {
      "epoch": 6.644467526119631,
      "grad_norm": 4.764182090759277,
      "learning_rate": 1.6817708970750215e-05,
      "loss": 1.1609,
      "step": 217500
    },
    {
      "epoch": 6.647522453717847,
      "grad_norm": 3.44966721534729,
      "learning_rate": 1.6802410965610087e-05,
      "loss": 1.0737,
      "step": 217600
    },
    {
      "epoch": 6.6505773813160625,
      "grad_norm": 3.3409371376037598,
      "learning_rate": 1.6787112960469954e-05,
      "loss": 1.084,
      "step": 217700
    },
    {
      "epoch": 6.653632308914279,
      "grad_norm": 9.89992904663086,
      "learning_rate": 1.677196793538123e-05,
      "loss": 1.0378,
      "step": 217800
    },
    {
      "epoch": 6.656687236512495,
      "grad_norm": 6.71497106552124,
      "learning_rate": 1.6756669930241097e-05,
      "loss": 1.1315,
      "step": 217900
    },
    {
      "epoch": 6.65974216411071,
      "grad_norm": 3.8522651195526123,
      "learning_rate": 1.6741371925100968e-05,
      "loss": 1.0395,
      "step": 218000
    },
    {
      "epoch": 6.662797091708926,
      "grad_norm": 7.571756362915039,
      "learning_rate": 1.672607391996084e-05,
      "loss": 1.1587,
      "step": 218100
    },
    {
      "epoch": 6.665852019307143,
      "grad_norm": 8.427057266235352,
      "learning_rate": 1.671077591482071e-05,
      "loss": 1.1139,
      "step": 218200
    },
    {
      "epoch": 6.668906946905358,
      "grad_norm": 4.802609443664551,
      "learning_rate": 1.6695477909680578e-05,
      "loss": 1.0882,
      "step": 218300
    },
    {
      "epoch": 6.671961874503574,
      "grad_norm": 5.4690961837768555,
      "learning_rate": 1.668017990454045e-05,
      "loss": 1.0002,
      "step": 218400
    },
    {
      "epoch": 6.675016802101791,
      "grad_norm": 13.944718360900879,
      "learning_rate": 1.666488189940032e-05,
      "loss": 1.2559,
      "step": 218500
    },
    {
      "epoch": 6.678071729700006,
      "grad_norm": 4.6269989013671875,
      "learning_rate": 1.664958389426019e-05,
      "loss": 1.0556,
      "step": 218600
    },
    {
      "epoch": 6.681126657298222,
      "grad_norm": 10.070877075195312,
      "learning_rate": 1.663428588912006e-05,
      "loss": 1.1925,
      "step": 218700
    },
    {
      "epoch": 6.684181584896438,
      "grad_norm": 8.7732515335083,
      "learning_rate": 1.661898788397993e-05,
      "loss": 0.9621,
      "step": 218800
    },
    {
      "epoch": 6.687236512494654,
      "grad_norm": 8.3976411819458,
      "learning_rate": 1.66036898788398e-05,
      "loss": 0.9891,
      "step": 218900
    },
    {
      "epoch": 6.69029144009287,
      "grad_norm": 5.672944068908691,
      "learning_rate": 1.6588391873699673e-05,
      "loss": 1.0646,
      "step": 219000
    },
    {
      "epoch": 6.693346367691086,
      "grad_norm": 3.278470754623413,
      "learning_rate": 1.657309386855954e-05,
      "loss": 1.0843,
      "step": 219100
    },
    {
      "epoch": 6.696401295289301,
      "grad_norm": 4.4352641105651855,
      "learning_rate": 1.6557795863419412e-05,
      "loss": 1.1092,
      "step": 219200
    },
    {
      "epoch": 6.699456222887518,
      "grad_norm": 6.883086204528809,
      "learning_rate": 1.6542497858279283e-05,
      "loss": 1.1694,
      "step": 219300
    },
    {
      "epoch": 6.7025111504857335,
      "grad_norm": 5.87219762802124,
      "learning_rate": 1.6527199853139154e-05,
      "loss": 1.1636,
      "step": 219400
    },
    {
      "epoch": 6.705566078083949,
      "grad_norm": 9.403863906860352,
      "learning_rate": 1.6511901847999022e-05,
      "loss": 1.0704,
      "step": 219500
    },
    {
      "epoch": 6.708621005682165,
      "grad_norm": 5.002470970153809,
      "learning_rate": 1.6496603842858893e-05,
      "loss": 1.1899,
      "step": 219600
    },
    {
      "epoch": 6.7116759332803815,
      "grad_norm": 0.014414245262742043,
      "learning_rate": 1.6481305837718764e-05,
      "loss": 1.2799,
      "step": 219700
    },
    {
      "epoch": 6.714730860878597,
      "grad_norm": 6.7424187660217285,
      "learning_rate": 1.6466007832578635e-05,
      "loss": 1.0311,
      "step": 219800
    },
    {
      "epoch": 6.717785788476813,
      "grad_norm": 6.960201263427734,
      "learning_rate": 1.6450709827438503e-05,
      "loss": 0.9706,
      "step": 219900
    },
    {
      "epoch": 6.720840716075029,
      "grad_norm": 4.6439290046691895,
      "learning_rate": 1.6435411822298374e-05,
      "loss": 1.1177,
      "step": 220000
    },
    {
      "epoch": 6.723895643673245,
      "grad_norm": 3.3932769298553467,
      "learning_rate": 1.6420113817158245e-05,
      "loss": 1.2257,
      "step": 220100
    },
    {
      "epoch": 6.726950571271461,
      "grad_norm": 4.096286296844482,
      "learning_rate": 1.6404815812018113e-05,
      "loss": 1.0774,
      "step": 220200
    },
    {
      "epoch": 6.7300054988696765,
      "grad_norm": 5.358587741851807,
      "learning_rate": 1.6389670786929385e-05,
      "loss": 0.8585,
      "step": 220300
    },
    {
      "epoch": 6.733060426467893,
      "grad_norm": 4.067366600036621,
      "learning_rate": 1.6374372781789256e-05,
      "loss": 1.0609,
      "step": 220400
    },
    {
      "epoch": 6.736115354066109,
      "grad_norm": 5.973146438598633,
      "learning_rate": 1.6359074776649123e-05,
      "loss": 1.1005,
      "step": 220500
    },
    {
      "epoch": 6.739170281664324,
      "grad_norm": 8.303258895874023,
      "learning_rate": 1.6343776771508995e-05,
      "loss": 1.1691,
      "step": 220600
    },
    {
      "epoch": 6.74222520926254,
      "grad_norm": 12.337789535522461,
      "learning_rate": 1.6328478766368866e-05,
      "loss": 1.146,
      "step": 220700
    },
    {
      "epoch": 6.745280136860757,
      "grad_norm": 12.040741920471191,
      "learning_rate": 1.6313180761228737e-05,
      "loss": 1.5116,
      "step": 220800
    },
    {
      "epoch": 6.748335064458972,
      "grad_norm": 8.398398399353027,
      "learning_rate": 1.6297882756088605e-05,
      "loss": 1.2403,
      "step": 220900
    },
    {
      "epoch": 6.751389992057188,
      "grad_norm": 6.159371852874756,
      "learning_rate": 1.6282584750948476e-05,
      "loss": 1.169,
      "step": 221000
    },
    {
      "epoch": 6.754444919655404,
      "grad_norm": 4.961534023284912,
      "learning_rate": 1.6267286745808347e-05,
      "loss": 1.2808,
      "step": 221100
    },
    {
      "epoch": 6.75749984725362,
      "grad_norm": 2.4521520137786865,
      "learning_rate": 1.6251988740668218e-05,
      "loss": 1.0539,
      "step": 221200
    },
    {
      "epoch": 6.760554774851836,
      "grad_norm": 4.656429290771484,
      "learning_rate": 1.6236690735528086e-05,
      "loss": 0.9846,
      "step": 221300
    },
    {
      "epoch": 6.763609702450052,
      "grad_norm": 4.705031394958496,
      "learning_rate": 1.6221392730387957e-05,
      "loss": 1.0849,
      "step": 221400
    },
    {
      "epoch": 6.766664630048268,
      "grad_norm": 4.43703556060791,
      "learning_rate": 1.6206094725247828e-05,
      "loss": 1.0477,
      "step": 221500
    },
    {
      "epoch": 6.769719557646484,
      "grad_norm": 8.725431442260742,
      "learning_rate": 1.61907967201077e-05,
      "loss": 1.0679,
      "step": 221600
    },
    {
      "epoch": 6.7727744852447,
      "grad_norm": 8.772459030151367,
      "learning_rate": 1.6175498714967567e-05,
      "loss": 1.2964,
      "step": 221700
    },
    {
      "epoch": 6.775829412842915,
      "grad_norm": 5.187252044677734,
      "learning_rate": 1.616020070982744e-05,
      "loss": 1.1037,
      "step": 221800
    },
    {
      "epoch": 6.778884340441132,
      "grad_norm": 6.5184526443481445,
      "learning_rate": 1.614490270468731e-05,
      "loss": 1.0777,
      "step": 221900
    },
    {
      "epoch": 6.7819392680393475,
      "grad_norm": 8.145726203918457,
      "learning_rate": 1.612960469954718e-05,
      "loss": 1.1475,
      "step": 222000
    },
    {
      "epoch": 6.784994195637563,
      "grad_norm": 3.8500397205352783,
      "learning_rate": 1.611430669440705e-05,
      "loss": 1.0347,
      "step": 222100
    },
    {
      "epoch": 6.78804912323578,
      "grad_norm": 5.5146894454956055,
      "learning_rate": 1.609900868926692e-05,
      "loss": 1.2915,
      "step": 222200
    },
    {
      "epoch": 6.7911040508339955,
      "grad_norm": 6.543974876403809,
      "learning_rate": 1.608371068412679e-05,
      "loss": 1.2405,
      "step": 222300
    },
    {
      "epoch": 6.794158978432211,
      "grad_norm": 4.565728664398193,
      "learning_rate": 1.6068412678986662e-05,
      "loss": 1.3069,
      "step": 222400
    },
    {
      "epoch": 6.797213906030427,
      "grad_norm": 2.4047398567199707,
      "learning_rate": 1.605311467384653e-05,
      "loss": 1.1599,
      "step": 222500
    },
    {
      "epoch": 6.8002688336286425,
      "grad_norm": 5.106794357299805,
      "learning_rate": 1.60378166687064e-05,
      "loss": 1.1013,
      "step": 222600
    },
    {
      "epoch": 6.803323761226859,
      "grad_norm": 6.793829917907715,
      "learning_rate": 1.6022518663566272e-05,
      "loss": 1.2487,
      "step": 222700
    },
    {
      "epoch": 6.806378688825075,
      "grad_norm": 7.5256452560424805,
      "learning_rate": 1.6007220658426143e-05,
      "loss": 1.1052,
      "step": 222800
    },
    {
      "epoch": 6.8094336164232905,
      "grad_norm": 27.9412841796875,
      "learning_rate": 1.599192265328601e-05,
      "loss": 1.1104,
      "step": 222900
    },
    {
      "epoch": 6.812488544021507,
      "grad_norm": 9.895366668701172,
      "learning_rate": 1.5976624648145882e-05,
      "loss": 1.3076,
      "step": 223000
    },
    {
      "epoch": 6.815543471619723,
      "grad_norm": 3.188260555267334,
      "learning_rate": 1.5961326643005753e-05,
      "loss": 0.9702,
      "step": 223100
    },
    {
      "epoch": 6.818598399217938,
      "grad_norm": 3.186352252960205,
      "learning_rate": 1.5946028637865624e-05,
      "loss": 1.1281,
      "step": 223200
    },
    {
      "epoch": 6.821653326816154,
      "grad_norm": 1.098803997039795,
      "learning_rate": 1.5930730632725492e-05,
      "loss": 1.1198,
      "step": 223300
    },
    {
      "epoch": 6.824708254414371,
      "grad_norm": 3.8436925411224365,
      "learning_rate": 1.5915432627585363e-05,
      "loss": 1.2168,
      "step": 223400
    },
    {
      "epoch": 6.827763182012586,
      "grad_norm": 8.489556312561035,
      "learning_rate": 1.5900134622445234e-05,
      "loss": 1.2194,
      "step": 223500
    },
    {
      "epoch": 6.830818109610802,
      "grad_norm": 4.756037712097168,
      "learning_rate": 1.5884836617305106e-05,
      "loss": 1.2152,
      "step": 223600
    },
    {
      "epoch": 6.833873037209019,
      "grad_norm": 3.0809168815612793,
      "learning_rate": 1.5869538612164973e-05,
      "loss": 0.9161,
      "step": 223700
    },
    {
      "epoch": 6.836927964807234,
      "grad_norm": 5.461037635803223,
      "learning_rate": 1.5854240607024845e-05,
      "loss": 1.042,
      "step": 223800
    },
    {
      "epoch": 6.83998289240545,
      "grad_norm": 13.365588188171387,
      "learning_rate": 1.5838942601884716e-05,
      "loss": 1.1589,
      "step": 223900
    },
    {
      "epoch": 6.843037820003666,
      "grad_norm": 5.828542709350586,
      "learning_rate": 1.5823644596744587e-05,
      "loss": 1.2168,
      "step": 224000
    },
    {
      "epoch": 6.846092747601881,
      "grad_norm": 7.322714805603027,
      "learning_rate": 1.5808346591604455e-05,
      "loss": 1.035,
      "step": 224100
    },
    {
      "epoch": 6.849147675200098,
      "grad_norm": 2.8676154613494873,
      "learning_rate": 1.579320156651573e-05,
      "loss": 1.1801,
      "step": 224200
    },
    {
      "epoch": 6.852202602798314,
      "grad_norm": 2.7156238555908203,
      "learning_rate": 1.5777903561375597e-05,
      "loss": 1.0304,
      "step": 224300
    },
    {
      "epoch": 6.855257530396529,
      "grad_norm": 11.405207633972168,
      "learning_rate": 1.5762605556235468e-05,
      "loss": 1.239,
      "step": 224400
    },
    {
      "epoch": 6.858312457994746,
      "grad_norm": 12.10234260559082,
      "learning_rate": 1.574730755109534e-05,
      "loss": 1.3059,
      "step": 224500
    },
    {
      "epoch": 6.8613673855929616,
      "grad_norm": 35.208656311035156,
      "learning_rate": 1.573200954595521e-05,
      "loss": 1.1348,
      "step": 224600
    },
    {
      "epoch": 6.864422313191177,
      "grad_norm": 6.893289089202881,
      "learning_rate": 1.571671154081508e-05,
      "loss": 1.1835,
      "step": 224700
    },
    {
      "epoch": 6.867477240789393,
      "grad_norm": 4.095139980316162,
      "learning_rate": 1.570141353567495e-05,
      "loss": 1.0936,
      "step": 224800
    },
    {
      "epoch": 6.8705321683876095,
      "grad_norm": 8.315845489501953,
      "learning_rate": 1.568611553053482e-05,
      "loss": 1.0834,
      "step": 224900
    },
    {
      "epoch": 6.873587095985825,
      "grad_norm": 9.586271286010742,
      "learning_rate": 1.5670817525394692e-05,
      "loss": 1.2971,
      "step": 225000
    },
    {
      "epoch": 6.876642023584041,
      "grad_norm": 3.927130699157715,
      "learning_rate": 1.565551952025456e-05,
      "loss": 1.6149,
      "step": 225100
    },
    {
      "epoch": 6.879696951182257,
      "grad_norm": 4.632936477661133,
      "learning_rate": 1.564022151511443e-05,
      "loss": 1.1813,
      "step": 225200
    },
    {
      "epoch": 6.882751878780473,
      "grad_norm": 6.257320880889893,
      "learning_rate": 1.5624923509974302e-05,
      "loss": 0.9665,
      "step": 225300
    },
    {
      "epoch": 6.885806806378689,
      "grad_norm": 23.577136993408203,
      "learning_rate": 1.5609625504834173e-05,
      "loss": 1.2635,
      "step": 225400
    },
    {
      "epoch": 6.8888617339769045,
      "grad_norm": 13.904678344726562,
      "learning_rate": 1.559432749969404e-05,
      "loss": 1.0694,
      "step": 225500
    },
    {
      "epoch": 6.89191666157512,
      "grad_norm": 6.760549545288086,
      "learning_rate": 1.5579029494553912e-05,
      "loss": 1.1164,
      "step": 225600
    },
    {
      "epoch": 6.894971589173337,
      "grad_norm": 10.959383964538574,
      "learning_rate": 1.5563731489413783e-05,
      "loss": 1.3547,
      "step": 225700
    },
    {
      "epoch": 6.898026516771552,
      "grad_norm": 5.473452091217041,
      "learning_rate": 1.554843348427365e-05,
      "loss": 1.0673,
      "step": 225800
    },
    {
      "epoch": 6.901081444369768,
      "grad_norm": 4.299148082733154,
      "learning_rate": 1.5533135479133522e-05,
      "loss": 1.1235,
      "step": 225900
    },
    {
      "epoch": 6.904136371967985,
      "grad_norm": 5.2035417556762695,
      "learning_rate": 1.551783747399339e-05,
      "loss": 1.2148,
      "step": 226000
    },
    {
      "epoch": 6.9071912995662,
      "grad_norm": 8.355424880981445,
      "learning_rate": 1.550253946885326e-05,
      "loss": 1.0416,
      "step": 226100
    },
    {
      "epoch": 6.910246227164416,
      "grad_norm": 17.467309951782227,
      "learning_rate": 1.5487241463713132e-05,
      "loss": 1.0306,
      "step": 226200
    },
    {
      "epoch": 6.913301154762632,
      "grad_norm": 6.781820297241211,
      "learning_rate": 1.5471943458573003e-05,
      "loss": 0.9252,
      "step": 226300
    },
    {
      "epoch": 6.916356082360848,
      "grad_norm": 6.654858589172363,
      "learning_rate": 1.545664545343287e-05,
      "loss": 1.124,
      "step": 226400
    },
    {
      "epoch": 6.919411009959064,
      "grad_norm": 10.962957382202148,
      "learning_rate": 1.5441347448292742e-05,
      "loss": 1.0413,
      "step": 226500
    },
    {
      "epoch": 6.92246593755728,
      "grad_norm": 12.6961030960083,
      "learning_rate": 1.5426049443152613e-05,
      "loss": 1.265,
      "step": 226600
    },
    {
      "epoch": 6.925520865155496,
      "grad_norm": 5.4705681800842285,
      "learning_rate": 1.5410751438012485e-05,
      "loss": 1.0878,
      "step": 226700
    },
    {
      "epoch": 6.928575792753712,
      "grad_norm": 7.821584701538086,
      "learning_rate": 1.5395453432872352e-05,
      "loss": 1.2651,
      "step": 226800
    },
    {
      "epoch": 6.931630720351928,
      "grad_norm": 1.9477187395095825,
      "learning_rate": 1.5380155427732223e-05,
      "loss": 1.2534,
      "step": 226900
    },
    {
      "epoch": 6.934685647950143,
      "grad_norm": 10.880365371704102,
      "learning_rate": 1.5364857422592095e-05,
      "loss": 1.028,
      "step": 227000
    },
    {
      "epoch": 6.93774057554836,
      "grad_norm": 3.2521772384643555,
      "learning_rate": 1.5349559417451966e-05,
      "loss": 1.0412,
      "step": 227100
    },
    {
      "epoch": 6.940795503146576,
      "grad_norm": 6.3504743576049805,
      "learning_rate": 1.5334261412311834e-05,
      "loss": 1.2927,
      "step": 227200
    },
    {
      "epoch": 6.943850430744791,
      "grad_norm": 11.716618537902832,
      "learning_rate": 1.5319116387223105e-05,
      "loss": 1.0605,
      "step": 227300
    },
    {
      "epoch": 6.946905358343007,
      "grad_norm": 12.864195823669434,
      "learning_rate": 1.5303818382082976e-05,
      "loss": 0.941,
      "step": 227400
    },
    {
      "epoch": 6.9499602859412235,
      "grad_norm": 5.322654724121094,
      "learning_rate": 1.5288520376942847e-05,
      "loss": 1.1716,
      "step": 227500
    },
    {
      "epoch": 6.953015213539439,
      "grad_norm": 7.079869270324707,
      "learning_rate": 1.527322237180272e-05,
      "loss": 1.0776,
      "step": 227600
    },
    {
      "epoch": 6.956070141137655,
      "grad_norm": 4.664093017578125,
      "learning_rate": 1.5257924366662588e-05,
      "loss": 1.1936,
      "step": 227700
    },
    {
      "epoch": 6.959125068735871,
      "grad_norm": 10.354986190795898,
      "learning_rate": 1.5242626361522457e-05,
      "loss": 1.0978,
      "step": 227800
    },
    {
      "epoch": 6.962179996334087,
      "grad_norm": 6.121084213256836,
      "learning_rate": 1.5227328356382328e-05,
      "loss": 1.1082,
      "step": 227900
    },
    {
      "epoch": 6.965234923932303,
      "grad_norm": 6.739871025085449,
      "learning_rate": 1.5212030351242198e-05,
      "loss": 1.2247,
      "step": 228000
    },
    {
      "epoch": 6.9682898515305185,
      "grad_norm": 7.606113910675049,
      "learning_rate": 1.5196732346102069e-05,
      "loss": 1.314,
      "step": 228100
    },
    {
      "epoch": 6.971344779128735,
      "grad_norm": 1.3897545337677002,
      "learning_rate": 1.5181434340961939e-05,
      "loss": 0.996,
      "step": 228200
    },
    {
      "epoch": 6.974399706726951,
      "grad_norm": 8.686017036437988,
      "learning_rate": 1.516613633582181e-05,
      "loss": 1.1612,
      "step": 228300
    },
    {
      "epoch": 6.977454634325166,
      "grad_norm": 11.423168182373047,
      "learning_rate": 1.515083833068168e-05,
      "loss": 1.0611,
      "step": 228400
    },
    {
      "epoch": 6.980509561923382,
      "grad_norm": 11.569713592529297,
      "learning_rate": 1.513554032554155e-05,
      "loss": 0.9411,
      "step": 228500
    },
    {
      "epoch": 6.983564489521599,
      "grad_norm": 7.2182745933532715,
      "learning_rate": 1.512024232040142e-05,
      "loss": 1.1535,
      "step": 228600
    },
    {
      "epoch": 6.986619417119814,
      "grad_norm": 9.209665298461914,
      "learning_rate": 1.5104944315261291e-05,
      "loss": 1.445,
      "step": 228700
    },
    {
      "epoch": 6.98967434471803,
      "grad_norm": 3.6917855739593506,
      "learning_rate": 1.508964631012116e-05,
      "loss": 1.1005,
      "step": 228800
    },
    {
      "epoch": 6.992729272316246,
      "grad_norm": 7.044301986694336,
      "learning_rate": 1.5074348304981032e-05,
      "loss": 1.1654,
      "step": 228900
    },
    {
      "epoch": 6.995784199914462,
      "grad_norm": 5.988764762878418,
      "learning_rate": 1.5059050299840901e-05,
      "loss": 1.1821,
      "step": 229000
    },
    {
      "epoch": 6.998839127512678,
      "grad_norm": 4.651966094970703,
      "learning_rate": 1.5043752294700772e-05,
      "loss": 1.0204,
      "step": 229100
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.7272560640312825,
      "eval_loss": 0.7299212217330933,
      "eval_runtime": 1775.0348,
      "eval_samples_per_second": 18.441,
      "eval_steps_per_second": 4.611,
      "step": 229138
    },
    {
      "epoch": 7.001894055110894,
      "grad_norm": 4.024294853210449,
      "learning_rate": 1.5028454289560642e-05,
      "loss": 1.1047,
      "step": 229200
    },
    {
      "epoch": 7.004948982709109,
      "grad_norm": 4.152279376983643,
      "learning_rate": 1.5013156284420513e-05,
      "loss": 1.1013,
      "step": 229300
    },
    {
      "epoch": 7.008003910307326,
      "grad_norm": 7.752127647399902,
      "learning_rate": 1.4997858279280382e-05,
      "loss": 0.9694,
      "step": 229400
    },
    {
      "epoch": 7.011058837905542,
      "grad_norm": 26.886756896972656,
      "learning_rate": 1.4982560274140253e-05,
      "loss": 1.76,
      "step": 229500
    },
    {
      "epoch": 7.014113765503757,
      "grad_norm": 5.418153285980225,
      "learning_rate": 1.4967262269000123e-05,
      "loss": 1.2325,
      "step": 229600
    },
    {
      "epoch": 7.017168693101974,
      "grad_norm": 3.349722146987915,
      "learning_rate": 1.4951964263859994e-05,
      "loss": 1.0413,
      "step": 229700
    },
    {
      "epoch": 7.02022362070019,
      "grad_norm": 11.646615028381348,
      "learning_rate": 1.4936666258719864e-05,
      "loss": 1.0621,
      "step": 229800
    },
    {
      "epoch": 7.023278548298405,
      "grad_norm": 11.823214530944824,
      "learning_rate": 1.4921368253579735e-05,
      "loss": 1.0923,
      "step": 229900
    },
    {
      "epoch": 7.026333475896621,
      "grad_norm": 16.650747299194336,
      "learning_rate": 1.4906070248439604e-05,
      "loss": 1.0721,
      "step": 230000
    },
    {
      "epoch": 7.0293884034948375,
      "grad_norm": 6.320542335510254,
      "learning_rate": 1.4890772243299475e-05,
      "loss": 1.1855,
      "step": 230100
    },
    {
      "epoch": 7.032443331093053,
      "grad_norm": 8.720442771911621,
      "learning_rate": 1.4875474238159345e-05,
      "loss": 1.0078,
      "step": 230200
    },
    {
      "epoch": 7.035498258691269,
      "grad_norm": 4.015342712402344,
      "learning_rate": 1.4860176233019216e-05,
      "loss": 1.1124,
      "step": 230300
    },
    {
      "epoch": 7.038553186289485,
      "grad_norm": 5.141261100769043,
      "learning_rate": 1.4844878227879085e-05,
      "loss": 0.9912,
      "step": 230400
    },
    {
      "epoch": 7.041608113887701,
      "grad_norm": 6.7251296043396,
      "learning_rate": 1.4829580222738957e-05,
      "loss": 1.1093,
      "step": 230500
    },
    {
      "epoch": 7.044663041485917,
      "grad_norm": 5.447814464569092,
      "learning_rate": 1.4814282217598826e-05,
      "loss": 1.0289,
      "step": 230600
    },
    {
      "epoch": 7.0477179690841325,
      "grad_norm": 10.36055850982666,
      "learning_rate": 1.4798984212458697e-05,
      "loss": 0.9546,
      "step": 230700
    },
    {
      "epoch": 7.050772896682348,
      "grad_norm": 5.748435020446777,
      "learning_rate": 1.4783686207318567e-05,
      "loss": 1.0437,
      "step": 230800
    },
    {
      "epoch": 7.053827824280565,
      "grad_norm": 4.40362548828125,
      "learning_rate": 1.4768388202178438e-05,
      "loss": 1.151,
      "step": 230900
    },
    {
      "epoch": 7.0568827518787804,
      "grad_norm": 5.855711936950684,
      "learning_rate": 1.4753090197038307e-05,
      "loss": 0.9614,
      "step": 231000
    },
    {
      "epoch": 7.059937679476996,
      "grad_norm": 14.346360206604004,
      "learning_rate": 1.4737792191898178e-05,
      "loss": 1.1533,
      "step": 231100
    },
    {
      "epoch": 7.062992607075213,
      "grad_norm": 9.502619743347168,
      "learning_rate": 1.4722494186758048e-05,
      "loss": 0.9982,
      "step": 231200
    },
    {
      "epoch": 7.066047534673428,
      "grad_norm": 10.261883735656738,
      "learning_rate": 1.4707349161669318e-05,
      "loss": 1.2859,
      "step": 231300
    },
    {
      "epoch": 7.069102462271644,
      "grad_norm": 1.5850389003753662,
      "learning_rate": 1.4692051156529189e-05,
      "loss": 1.2657,
      "step": 231400
    },
    {
      "epoch": 7.07215738986986,
      "grad_norm": 9.610984802246094,
      "learning_rate": 1.4676753151389058e-05,
      "loss": 1.2829,
      "step": 231500
    },
    {
      "epoch": 7.075212317468076,
      "grad_norm": 9.58398151397705,
      "learning_rate": 1.466145514624893e-05,
      "loss": 1.1443,
      "step": 231600
    },
    {
      "epoch": 7.078267245066292,
      "grad_norm": 2.9476981163024902,
      "learning_rate": 1.4646157141108799e-05,
      "loss": 0.9903,
      "step": 231700
    },
    {
      "epoch": 7.081322172664508,
      "grad_norm": 7.310122966766357,
      "learning_rate": 1.463085913596867e-05,
      "loss": 0.9488,
      "step": 231800
    },
    {
      "epoch": 7.084377100262723,
      "grad_norm": 5.88616418838501,
      "learning_rate": 1.461556113082854e-05,
      "loss": 1.0228,
      "step": 231900
    },
    {
      "epoch": 7.08743202786094,
      "grad_norm": 6.997511386871338,
      "learning_rate": 1.460026312568841e-05,
      "loss": 1.0473,
      "step": 232000
    },
    {
      "epoch": 7.090486955459156,
      "grad_norm": 4.911482810974121,
      "learning_rate": 1.458496512054828e-05,
      "loss": 1.0401,
      "step": 232100
    },
    {
      "epoch": 7.093541883057371,
      "grad_norm": 0.0005010916502214968,
      "learning_rate": 1.4569667115408151e-05,
      "loss": 1.1492,
      "step": 232200
    },
    {
      "epoch": 7.096596810655588,
      "grad_norm": 3.7000105381011963,
      "learning_rate": 1.455436911026802e-05,
      "loss": 1.0027,
      "step": 232300
    },
    {
      "epoch": 7.099651738253804,
      "grad_norm": 13.737366676330566,
      "learning_rate": 1.4539071105127892e-05,
      "loss": 1.0997,
      "step": 232400
    },
    {
      "epoch": 7.102706665852019,
      "grad_norm": 6.119179725646973,
      "learning_rate": 1.4523773099987761e-05,
      "loss": 0.9705,
      "step": 232500
    },
    {
      "epoch": 7.105761593450235,
      "grad_norm": 1.9026933908462524,
      "learning_rate": 1.4508475094847632e-05,
      "loss": 1.1789,
      "step": 232600
    },
    {
      "epoch": 7.1088165210484515,
      "grad_norm": 20.846769332885742,
      "learning_rate": 1.4493177089707502e-05,
      "loss": 1.1999,
      "step": 232700
    },
    {
      "epoch": 7.111871448646667,
      "grad_norm": 6.88537073135376,
      "learning_rate": 1.4477879084567373e-05,
      "loss": 1.2442,
      "step": 232800
    },
    {
      "epoch": 7.114926376244883,
      "grad_norm": 5.11663293838501,
      "learning_rate": 1.4462581079427242e-05,
      "loss": 1.2855,
      "step": 232900
    },
    {
      "epoch": 7.117981303843099,
      "grad_norm": 9.254779815673828,
      "learning_rate": 1.4447283074287114e-05,
      "loss": 1.2988,
      "step": 233000
    },
    {
      "epoch": 7.121036231441315,
      "grad_norm": 7.493042945861816,
      "learning_rate": 1.4431985069146983e-05,
      "loss": 1.1818,
      "step": 233100
    },
    {
      "epoch": 7.124091159039531,
      "grad_norm": 9.90920352935791,
      "learning_rate": 1.4416687064006854e-05,
      "loss": 1.0613,
      "step": 233200
    },
    {
      "epoch": 7.1271460866377465,
      "grad_norm": 4.25264835357666,
      "learning_rate": 1.4401542038918126e-05,
      "loss": 0.9499,
      "step": 233300
    },
    {
      "epoch": 7.130201014235962,
      "grad_norm": 7.848468780517578,
      "learning_rate": 1.4386397013829397e-05,
      "loss": 1.0826,
      "step": 233400
    },
    {
      "epoch": 7.133255941834179,
      "grad_norm": 14.143811225891113,
      "learning_rate": 1.4371099008689268e-05,
      "loss": 1.0622,
      "step": 233500
    },
    {
      "epoch": 7.1363108694323945,
      "grad_norm": 1.9151153564453125,
      "learning_rate": 1.4355801003549138e-05,
      "loss": 0.9594,
      "step": 233600
    },
    {
      "epoch": 7.13936579703061,
      "grad_norm": 8.173432350158691,
      "learning_rate": 1.4340502998409009e-05,
      "loss": 1.0031,
      "step": 233700
    },
    {
      "epoch": 7.142420724628827,
      "grad_norm": 1.022454857826233,
      "learning_rate": 1.4325204993268878e-05,
      "loss": 1.2063,
      "step": 233800
    },
    {
      "epoch": 7.145475652227042,
      "grad_norm": 9.124458312988281,
      "learning_rate": 1.430990698812875e-05,
      "loss": 1.1762,
      "step": 233900
    },
    {
      "epoch": 7.148530579825258,
      "grad_norm": 12.708720207214355,
      "learning_rate": 1.4294608982988619e-05,
      "loss": 1.1155,
      "step": 234000
    },
    {
      "epoch": 7.151585507423474,
      "grad_norm": 1.8738713264465332,
      "learning_rate": 1.427931097784849e-05,
      "loss": 1.0435,
      "step": 234100
    },
    {
      "epoch": 7.15464043502169,
      "grad_norm": 7.204272747039795,
      "learning_rate": 1.426401297270836e-05,
      "loss": 1.263,
      "step": 234200
    },
    {
      "epoch": 7.157695362619906,
      "grad_norm": 1.4616661071777344,
      "learning_rate": 1.424871496756823e-05,
      "loss": 1.0202,
      "step": 234300
    },
    {
      "epoch": 7.160750290218122,
      "grad_norm": 3.6434624195098877,
      "learning_rate": 1.42334169624281e-05,
      "loss": 0.9481,
      "step": 234400
    },
    {
      "epoch": 7.163805217816337,
      "grad_norm": 5.396409511566162,
      "learning_rate": 1.4218118957287971e-05,
      "loss": 1.1474,
      "step": 234500
    },
    {
      "epoch": 7.166860145414554,
      "grad_norm": 1.5187724828720093,
      "learning_rate": 1.420282095214784e-05,
      "loss": 1.1487,
      "step": 234600
    },
    {
      "epoch": 7.16991507301277,
      "grad_norm": 2.525434732437134,
      "learning_rate": 1.4187522947007712e-05,
      "loss": 1.1669,
      "step": 234700
    },
    {
      "epoch": 7.172970000610985,
      "grad_norm": 2.303208112716675,
      "learning_rate": 1.4172224941867581e-05,
      "loss": 1.1454,
      "step": 234800
    },
    {
      "epoch": 7.176024928209201,
      "grad_norm": 6.004208564758301,
      "learning_rate": 1.4156926936727452e-05,
      "loss": 1.168,
      "step": 234900
    },
    {
      "epoch": 7.179079855807418,
      "grad_norm": 11.640929222106934,
      "learning_rate": 1.4141628931587322e-05,
      "loss": 1.2544,
      "step": 235000
    },
    {
      "epoch": 7.182134783405633,
      "grad_norm": 6.673908233642578,
      "learning_rate": 1.4126330926447193e-05,
      "loss": 1.2008,
      "step": 235100
    },
    {
      "epoch": 7.185189711003849,
      "grad_norm": 14.122396469116211,
      "learning_rate": 1.4111032921307063e-05,
      "loss": 1.0178,
      "step": 235200
    },
    {
      "epoch": 7.1882446386020655,
      "grad_norm": 11.688652992248535,
      "learning_rate": 1.4095734916166934e-05,
      "loss": 1.2486,
      "step": 235300
    },
    {
      "epoch": 7.191299566200281,
      "grad_norm": 4.625455856323242,
      "learning_rate": 1.4080436911026803e-05,
      "loss": 1.1022,
      "step": 235400
    },
    {
      "epoch": 7.194354493798497,
      "grad_norm": 5.053491115570068,
      "learning_rate": 1.4065138905886674e-05,
      "loss": 1.1197,
      "step": 235500
    },
    {
      "epoch": 7.197409421396713,
      "grad_norm": 4.066915035247803,
      "learning_rate": 1.4049840900746544e-05,
      "loss": 1.1365,
      "step": 235600
    },
    {
      "epoch": 7.200464348994929,
      "grad_norm": 11.92159366607666,
      "learning_rate": 1.4034542895606415e-05,
      "loss": 1.0201,
      "step": 235700
    },
    {
      "epoch": 7.203519276593145,
      "grad_norm": 15.129595756530762,
      "learning_rate": 1.4019244890466284e-05,
      "loss": 1.3958,
      "step": 235800
    },
    {
      "epoch": 7.2065742041913605,
      "grad_norm": 5.431704521179199,
      "learning_rate": 1.4003946885326156e-05,
      "loss": 1.5226,
      "step": 235900
    },
    {
      "epoch": 7.209629131789576,
      "grad_norm": 9.33706283569336,
      "learning_rate": 1.3988648880186025e-05,
      "loss": 1.2875,
      "step": 236000
    },
    {
      "epoch": 7.212684059387793,
      "grad_norm": 1.7640498876571655,
      "learning_rate": 1.3973350875045896e-05,
      "loss": 1.2525,
      "step": 236100
    },
    {
      "epoch": 7.2157389869860085,
      "grad_norm": 5.957800388336182,
      "learning_rate": 1.3958052869905766e-05,
      "loss": 1.1711,
      "step": 236200
    },
    {
      "epoch": 7.218793914584224,
      "grad_norm": 1.0705599784851074,
      "learning_rate": 1.3942754864765637e-05,
      "loss": 1.1135,
      "step": 236300
    },
    {
      "epoch": 7.221848842182441,
      "grad_norm": 2.2979612350463867,
      "learning_rate": 1.3927456859625506e-05,
      "loss": 1.3061,
      "step": 236400
    },
    {
      "epoch": 7.224903769780656,
      "grad_norm": 7.749421119689941,
      "learning_rate": 1.3912158854485377e-05,
      "loss": 1.109,
      "step": 236500
    },
    {
      "epoch": 7.227958697378872,
      "grad_norm": 2.989555835723877,
      "learning_rate": 1.3897013829396645e-05,
      "loss": 1.0205,
      "step": 236600
    },
    {
      "epoch": 7.231013624977088,
      "grad_norm": 5.32132625579834,
      "learning_rate": 1.3881715824256517e-05,
      "loss": 1.2055,
      "step": 236700
    },
    {
      "epoch": 7.234068552575304,
      "grad_norm": 5.634767055511475,
      "learning_rate": 1.3866417819116386e-05,
      "loss": 1.0373,
      "step": 236800
    },
    {
      "epoch": 7.23712348017352,
      "grad_norm": 7.112980365753174,
      "learning_rate": 1.3851119813976257e-05,
      "loss": 1.4055,
      "step": 236900
    },
    {
      "epoch": 7.240178407771736,
      "grad_norm": 10.331284523010254,
      "learning_rate": 1.3835821808836127e-05,
      "loss": 1.1856,
      "step": 237000
    },
    {
      "epoch": 7.243233335369951,
      "grad_norm": 10.450326919555664,
      "learning_rate": 1.3820523803695998e-05,
      "loss": 1.229,
      "step": 237100
    },
    {
      "epoch": 7.246288262968168,
      "grad_norm": 4.122806549072266,
      "learning_rate": 1.3805225798555867e-05,
      "loss": 1.1836,
      "step": 237200
    },
    {
      "epoch": 7.249343190566384,
      "grad_norm": 6.546516418457031,
      "learning_rate": 1.3789927793415738e-05,
      "loss": 1.4002,
      "step": 237300
    },
    {
      "epoch": 7.252398118164599,
      "grad_norm": 20.594345092773438,
      "learning_rate": 1.3774629788275608e-05,
      "loss": 1.0855,
      "step": 237400
    },
    {
      "epoch": 7.255453045762815,
      "grad_norm": 4.27672815322876,
      "learning_rate": 1.3759331783135479e-05,
      "loss": 1.1928,
      "step": 237500
    },
    {
      "epoch": 7.258507973361032,
      "grad_norm": 24.660978317260742,
      "learning_rate": 1.3744033777995348e-05,
      "loss": 1.0662,
      "step": 237600
    },
    {
      "epoch": 7.261562900959247,
      "grad_norm": 6.918597221374512,
      "learning_rate": 1.372873577285522e-05,
      "loss": 1.5574,
      "step": 237700
    },
    {
      "epoch": 7.264617828557463,
      "grad_norm": 3.216303586959839,
      "learning_rate": 1.3713437767715089e-05,
      "loss": 1.0191,
      "step": 237800
    },
    {
      "epoch": 7.2676727561556795,
      "grad_norm": 10.302803039550781,
      "learning_rate": 1.369813976257496e-05,
      "loss": 1.0279,
      "step": 237900
    },
    {
      "epoch": 7.270727683753895,
      "grad_norm": 9.102781295776367,
      "learning_rate": 1.368284175743483e-05,
      "loss": 1.2373,
      "step": 238000
    },
    {
      "epoch": 7.273782611352111,
      "grad_norm": 24.65117645263672,
      "learning_rate": 1.3667543752294701e-05,
      "loss": 1.0409,
      "step": 238100
    },
    {
      "epoch": 7.276837538950327,
      "grad_norm": 3.3516035079956055,
      "learning_rate": 1.365224574715457e-05,
      "loss": 1.0934,
      "step": 238200
    },
    {
      "epoch": 7.279892466548543,
      "grad_norm": 8.557634353637695,
      "learning_rate": 1.3636947742014441e-05,
      "loss": 1.0797,
      "step": 238300
    },
    {
      "epoch": 7.282947394146759,
      "grad_norm": 5.520821571350098,
      "learning_rate": 1.3621649736874311e-05,
      "loss": 1.0617,
      "step": 238400
    },
    {
      "epoch": 7.2860023217449745,
      "grad_norm": 5.227816104888916,
      "learning_rate": 1.3606351731734182e-05,
      "loss": 0.977,
      "step": 238500
    },
    {
      "epoch": 7.28905724934319,
      "grad_norm": 4.158600330352783,
      "learning_rate": 1.3591053726594052e-05,
      "loss": 1.057,
      "step": 238600
    },
    {
      "epoch": 7.292112176941407,
      "grad_norm": 19.767812728881836,
      "learning_rate": 1.3575908701505325e-05,
      "loss": 1.484,
      "step": 238700
    },
    {
      "epoch": 7.2951671045396225,
      "grad_norm": 4.106821060180664,
      "learning_rate": 1.3560610696365194e-05,
      "loss": 0.9952,
      "step": 238800
    },
    {
      "epoch": 7.298222032137838,
      "grad_norm": 8.97727108001709,
      "learning_rate": 1.3545312691225065e-05,
      "loss": 1.2,
      "step": 238900
    },
    {
      "epoch": 7.301276959736054,
      "grad_norm": 11.650214195251465,
      "learning_rate": 1.3530014686084935e-05,
      "loss": 1.2169,
      "step": 239000
    },
    {
      "epoch": 7.30433188733427,
      "grad_norm": 5.218963623046875,
      "learning_rate": 1.3514716680944806e-05,
      "loss": 1.1261,
      "step": 239100
    },
    {
      "epoch": 7.307386814932486,
      "grad_norm": 7.4983086585998535,
      "learning_rate": 1.3499418675804675e-05,
      "loss": 0.9903,
      "step": 239200
    },
    {
      "epoch": 7.310441742530702,
      "grad_norm": 4.398214817047119,
      "learning_rate": 1.3484120670664546e-05,
      "loss": 1.12,
      "step": 239300
    },
    {
      "epoch": 7.313496670128918,
      "grad_norm": 4.890592098236084,
      "learning_rate": 1.3468822665524416e-05,
      "loss": 1.1316,
      "step": 239400
    },
    {
      "epoch": 7.316551597727134,
      "grad_norm": 4.648405075073242,
      "learning_rate": 1.3453524660384287e-05,
      "loss": 1.2801,
      "step": 239500
    },
    {
      "epoch": 7.31960652532535,
      "grad_norm": 3.6855125427246094,
      "learning_rate": 1.3438226655244157e-05,
      "loss": 1.1782,
      "step": 239600
    },
    {
      "epoch": 7.322661452923565,
      "grad_norm": 7.176767349243164,
      "learning_rate": 1.3422928650104028e-05,
      "loss": 0.9889,
      "step": 239700
    },
    {
      "epoch": 7.325716380521782,
      "grad_norm": 8.887348175048828,
      "learning_rate": 1.3407630644963897e-05,
      "loss": 1.0531,
      "step": 239800
    },
    {
      "epoch": 7.328771308119998,
      "grad_norm": 3.1371002197265625,
      "learning_rate": 1.3392332639823768e-05,
      "loss": 0.9987,
      "step": 239900
    },
    {
      "epoch": 7.331826235718213,
      "grad_norm": 18.492141723632812,
      "learning_rate": 1.3377034634683638e-05,
      "loss": 1.1184,
      "step": 240000
    },
    {
      "epoch": 7.334881163316429,
      "grad_norm": 19.492719650268555,
      "learning_rate": 1.3361736629543509e-05,
      "loss": 1.266,
      "step": 240100
    },
    {
      "epoch": 7.337936090914646,
      "grad_norm": 5.891000747680664,
      "learning_rate": 1.3346438624403378e-05,
      "loss": 0.932,
      "step": 240200
    },
    {
      "epoch": 7.340991018512861,
      "grad_norm": 3.2475335597991943,
      "learning_rate": 1.333114061926325e-05,
      "loss": 1.3091,
      "step": 240300
    },
    {
      "epoch": 7.344045946111077,
      "grad_norm": 7.269662857055664,
      "learning_rate": 1.3315842614123119e-05,
      "loss": 1.0285,
      "step": 240400
    },
    {
      "epoch": 7.347100873709293,
      "grad_norm": 8.6480073928833,
      "learning_rate": 1.330054460898299e-05,
      "loss": 1.038,
      "step": 240500
    },
    {
      "epoch": 7.350155801307509,
      "grad_norm": 8.871041297912598,
      "learning_rate": 1.328524660384286e-05,
      "loss": 1.051,
      "step": 240600
    },
    {
      "epoch": 7.353210728905725,
      "grad_norm": 1.0303308963775635,
      "learning_rate": 1.326994859870273e-05,
      "loss": 1.0538,
      "step": 240700
    },
    {
      "epoch": 7.356265656503941,
      "grad_norm": 6.683864593505859,
      "learning_rate": 1.32546505935626e-05,
      "loss": 1.135,
      "step": 240800
    },
    {
      "epoch": 7.359320584102157,
      "grad_norm": 7.670544624328613,
      "learning_rate": 1.3239352588422471e-05,
      "loss": 1.0207,
      "step": 240900
    },
    {
      "epoch": 7.362375511700373,
      "grad_norm": 6.8379807472229,
      "learning_rate": 1.3224054583282341e-05,
      "loss": 0.986,
      "step": 241000
    },
    {
      "epoch": 7.3654304392985885,
      "grad_norm": 4.02485990524292,
      "learning_rate": 1.3208756578142212e-05,
      "loss": 1.4324,
      "step": 241100
    },
    {
      "epoch": 7.368485366896804,
      "grad_norm": 3.0033915042877197,
      "learning_rate": 1.3193458573002082e-05,
      "loss": 1.2366,
      "step": 241200
    },
    {
      "epoch": 7.371540294495021,
      "grad_norm": 10.496794700622559,
      "learning_rate": 1.3178160567861953e-05,
      "loss": 1.1109,
      "step": 241300
    },
    {
      "epoch": 7.3745952220932365,
      "grad_norm": 13.235259056091309,
      "learning_rate": 1.3162862562721822e-05,
      "loss": 1.185,
      "step": 241400
    },
    {
      "epoch": 7.377650149691452,
      "grad_norm": 1.6905969381332397,
      "learning_rate": 1.3147564557581693e-05,
      "loss": 1.1872,
      "step": 241500
    },
    {
      "epoch": 7.380705077289668,
      "grad_norm": 6.437650680541992,
      "learning_rate": 1.3132266552441563e-05,
      "loss": 1.0061,
      "step": 241600
    },
    {
      "epoch": 7.383760004887884,
      "grad_norm": 9.396056175231934,
      "learning_rate": 1.3116968547301434e-05,
      "loss": 1.3761,
      "step": 241700
    },
    {
      "epoch": 7.3868149324861,
      "grad_norm": 8.510388374328613,
      "learning_rate": 1.3101670542161303e-05,
      "loss": 1.1561,
      "step": 241800
    },
    {
      "epoch": 7.389869860084316,
      "grad_norm": 3.6753456592559814,
      "learning_rate": 1.3086372537021175e-05,
      "loss": 1.3092,
      "step": 241900
    },
    {
      "epoch": 7.3929247876825315,
      "grad_norm": 5.2257585525512695,
      "learning_rate": 1.3071074531881044e-05,
      "loss": 1.0462,
      "step": 242000
    },
    {
      "epoch": 7.395979715280748,
      "grad_norm": 4.928442001342773,
      "learning_rate": 1.3055776526740915e-05,
      "loss": 1.0274,
      "step": 242100
    },
    {
      "epoch": 7.399034642878964,
      "grad_norm": 14.141927719116211,
      "learning_rate": 1.3040478521600785e-05,
      "loss": 0.9869,
      "step": 242200
    },
    {
      "epoch": 7.402089570477179,
      "grad_norm": 6.100747108459473,
      "learning_rate": 1.3025180516460656e-05,
      "loss": 1.0678,
      "step": 242300
    },
    {
      "epoch": 7.405144498075396,
      "grad_norm": 3.6259148120880127,
      "learning_rate": 1.3009882511320524e-05,
      "loss": 1.1743,
      "step": 242400
    },
    {
      "epoch": 7.408199425673612,
      "grad_norm": 10.17052173614502,
      "learning_rate": 1.2994584506180393e-05,
      "loss": 1.2426,
      "step": 242500
    },
    {
      "epoch": 7.411254353271827,
      "grad_norm": 16.377918243408203,
      "learning_rate": 1.2979286501040264e-05,
      "loss": 1.0444,
      "step": 242600
    },
    {
      "epoch": 7.414309280870043,
      "grad_norm": 9.223760604858398,
      "learning_rate": 1.2964141475951536e-05,
      "loss": 1.1732,
      "step": 242700
    },
    {
      "epoch": 7.41736420846826,
      "grad_norm": 9.646039962768555,
      "learning_rate": 1.2948843470811407e-05,
      "loss": 0.9626,
      "step": 242800
    },
    {
      "epoch": 7.420419136066475,
      "grad_norm": 6.570456504821777,
      "learning_rate": 1.2933545465671276e-05,
      "loss": 1.1104,
      "step": 242900
    },
    {
      "epoch": 7.423474063664691,
      "grad_norm": 6.207412242889404,
      "learning_rate": 1.2918247460531147e-05,
      "loss": 1.2049,
      "step": 243000
    },
    {
      "epoch": 7.426528991262907,
      "grad_norm": 4.209483623504639,
      "learning_rate": 1.2902949455391017e-05,
      "loss": 0.9876,
      "step": 243100
    },
    {
      "epoch": 7.429583918861123,
      "grad_norm": 9.007349967956543,
      "learning_rate": 1.2887651450250888e-05,
      "loss": 1.226,
      "step": 243200
    },
    {
      "epoch": 7.432638846459339,
      "grad_norm": 2.926483154296875,
      "learning_rate": 1.2872353445110757e-05,
      "loss": 1.1661,
      "step": 243300
    },
    {
      "epoch": 7.435693774057555,
      "grad_norm": 12.762822151184082,
      "learning_rate": 1.2857055439970629e-05,
      "loss": 1.211,
      "step": 243400
    },
    {
      "epoch": 7.43874870165577,
      "grad_norm": 6.0010085105896,
      "learning_rate": 1.2841757434830498e-05,
      "loss": 1.15,
      "step": 243500
    },
    {
      "epoch": 7.441803629253987,
      "grad_norm": 6.056092739105225,
      "learning_rate": 1.282645942969037e-05,
      "loss": 1.1764,
      "step": 243600
    },
    {
      "epoch": 7.4448585568522025,
      "grad_norm": 18.341081619262695,
      "learning_rate": 1.2811161424550239e-05,
      "loss": 0.924,
      "step": 243700
    },
    {
      "epoch": 7.447913484450418,
      "grad_norm": 1.1248468160629272,
      "learning_rate": 1.279601639946151e-05,
      "loss": 1.6367,
      "step": 243800
    },
    {
      "epoch": 7.450968412048635,
      "grad_norm": 30.9728946685791,
      "learning_rate": 1.2780718394321381e-05,
      "loss": 1.2798,
      "step": 243900
    },
    {
      "epoch": 7.4540233396468505,
      "grad_norm": 9.996712684631348,
      "learning_rate": 1.276542038918125e-05,
      "loss": 1.017,
      "step": 244000
    },
    {
      "epoch": 7.457078267245066,
      "grad_norm": 11.12072467803955,
      "learning_rate": 1.2750122384041122e-05,
      "loss": 1.148,
      "step": 244100
    },
    {
      "epoch": 7.460133194843282,
      "grad_norm": 17.706918716430664,
      "learning_rate": 1.2734824378900991e-05,
      "loss": 1.2085,
      "step": 244200
    },
    {
      "epoch": 7.463188122441498,
      "grad_norm": 6.079930782318115,
      "learning_rate": 1.2719526373760862e-05,
      "loss": 1.168,
      "step": 244300
    },
    {
      "epoch": 7.466243050039714,
      "grad_norm": 6.82859468460083,
      "learning_rate": 1.2704228368620732e-05,
      "loss": 0.9845,
      "step": 244400
    },
    {
      "epoch": 7.46929797763793,
      "grad_norm": 3.8205902576446533,
      "learning_rate": 1.2688930363480603e-05,
      "loss": 1.064,
      "step": 244500
    },
    {
      "epoch": 7.472352905236146,
      "grad_norm": 4.047750473022461,
      "learning_rate": 1.2673632358340472e-05,
      "loss": 1.2488,
      "step": 244600
    },
    {
      "epoch": 7.475407832834362,
      "grad_norm": 7.491181373596191,
      "learning_rate": 1.2658334353200344e-05,
      "loss": 1.0114,
      "step": 244700
    },
    {
      "epoch": 7.478462760432578,
      "grad_norm": 6.972776889801025,
      "learning_rate": 1.2643036348060213e-05,
      "loss": 0.9721,
      "step": 244800
    },
    {
      "epoch": 7.481517688030793,
      "grad_norm": 6.191954135894775,
      "learning_rate": 1.2627738342920084e-05,
      "loss": 1.131,
      "step": 244900
    },
    {
      "epoch": 7.48457261562901,
      "grad_norm": 0.34253376722335815,
      "learning_rate": 1.2612440337779954e-05,
      "loss": 1.3406,
      "step": 245000
    },
    {
      "epoch": 7.487627543227226,
      "grad_norm": 4.320346832275391,
      "learning_rate": 1.2597142332639825e-05,
      "loss": 1.23,
      "step": 245100
    },
    {
      "epoch": 7.490682470825441,
      "grad_norm": 2.7404799461364746,
      "learning_rate": 1.2581844327499694e-05,
      "loss": 1.0078,
      "step": 245200
    },
    {
      "epoch": 7.493737398423657,
      "grad_norm": 12.498672485351562,
      "learning_rate": 1.2566546322359565e-05,
      "loss": 1.0024,
      "step": 245300
    },
    {
      "epoch": 7.496792326021874,
      "grad_norm": 5.074945449829102,
      "learning_rate": 1.2551248317219435e-05,
      "loss": 1.1319,
      "step": 245400
    },
    {
      "epoch": 7.499847253620089,
      "grad_norm": 6.632750511169434,
      "learning_rate": 1.2535950312079306e-05,
      "loss": 1.0165,
      "step": 245500
    },
    {
      "epoch": 7.502902181218305,
      "grad_norm": 9.370561599731445,
      "learning_rate": 1.2520652306939176e-05,
      "loss": 1.0874,
      "step": 245600
    },
    {
      "epoch": 7.505957108816521,
      "grad_norm": 13.537225723266602,
      "learning_rate": 1.2505354301799047e-05,
      "loss": 1.1978,
      "step": 245700
    },
    {
      "epoch": 7.509012036414737,
      "grad_norm": 6.405066013336182,
      "learning_rate": 1.2490056296658916e-05,
      "loss": 0.9219,
      "step": 245800
    },
    {
      "epoch": 7.512066964012953,
      "grad_norm": 2.5054051876068115,
      "learning_rate": 1.2474758291518787e-05,
      "loss": 1.2472,
      "step": 245900
    },
    {
      "epoch": 7.515121891611169,
      "grad_norm": 4.03517484664917,
      "learning_rate": 1.2459460286378657e-05,
      "loss": 1.0767,
      "step": 246000
    },
    {
      "epoch": 7.518176819209385,
      "grad_norm": 1.487331509590149,
      "learning_rate": 1.2444315261289928e-05,
      "loss": 1.3574,
      "step": 246100
    },
    {
      "epoch": 7.521231746807601,
      "grad_norm": 11.857295989990234,
      "learning_rate": 1.2429017256149798e-05,
      "loss": 1.1426,
      "step": 246200
    },
    {
      "epoch": 7.5242866744058166,
      "grad_norm": 21.534475326538086,
      "learning_rate": 1.2413719251009669e-05,
      "loss": 1.1534,
      "step": 246300
    },
    {
      "epoch": 7.527341602004032,
      "grad_norm": 49.81202697753906,
      "learning_rate": 1.2398421245869538e-05,
      "loss": 1.2241,
      "step": 246400
    },
    {
      "epoch": 7.530396529602249,
      "grad_norm": 0.9042030572891235,
      "learning_rate": 1.238312324072941e-05,
      "loss": 1.0467,
      "step": 246500
    },
    {
      "epoch": 7.5334514572004645,
      "grad_norm": 6.449215412139893,
      "learning_rate": 1.2367825235589279e-05,
      "loss": 1.0632,
      "step": 246600
    },
    {
      "epoch": 7.53650638479868,
      "grad_norm": 5.2182841300964355,
      "learning_rate": 1.235252723044915e-05,
      "loss": 1.0664,
      "step": 246700
    },
    {
      "epoch": 7.539561312396896,
      "grad_norm": 3.7590441703796387,
      "learning_rate": 1.233722922530902e-05,
      "loss": 1.2511,
      "step": 246800
    },
    {
      "epoch": 7.542616239995112,
      "grad_norm": 6.888582229614258,
      "learning_rate": 1.232193122016889e-05,
      "loss": 1.1955,
      "step": 246900
    },
    {
      "epoch": 7.545671167593328,
      "grad_norm": 11.217452049255371,
      "learning_rate": 1.230663321502876e-05,
      "loss": 0.9752,
      "step": 247000
    },
    {
      "epoch": 7.548726095191544,
      "grad_norm": 0.6481238603591919,
      "learning_rate": 1.2291335209888631e-05,
      "loss": 1.1871,
      "step": 247100
    },
    {
      "epoch": 7.5517810227897595,
      "grad_norm": 12.974531173706055,
      "learning_rate": 1.22760372047485e-05,
      "loss": 1.0176,
      "step": 247200
    },
    {
      "epoch": 7.554835950387976,
      "grad_norm": 11.442163467407227,
      "learning_rate": 1.2260739199608372e-05,
      "loss": 1.2426,
      "step": 247300
    },
    {
      "epoch": 7.557890877986192,
      "grad_norm": 6.187389850616455,
      "learning_rate": 1.2245441194468241e-05,
      "loss": 1.1011,
      "step": 247400
    },
    {
      "epoch": 7.560945805584407,
      "grad_norm": 9.375640869140625,
      "learning_rate": 1.2230143189328112e-05,
      "loss": 1.1226,
      "step": 247500
    },
    {
      "epoch": 7.564000733182624,
      "grad_norm": 9.352968215942383,
      "learning_rate": 1.2214845184187982e-05,
      "loss": 1.0503,
      "step": 247600
    },
    {
      "epoch": 7.56705566078084,
      "grad_norm": 1.1412074565887451,
      "learning_rate": 1.2199547179047853e-05,
      "loss": 1.2557,
      "step": 247700
    },
    {
      "epoch": 7.570110588379055,
      "grad_norm": 4.671082019805908,
      "learning_rate": 1.2184249173907723e-05,
      "loss": 1.3057,
      "step": 247800
    },
    {
      "epoch": 7.573165515977271,
      "grad_norm": 3.711911916732788,
      "learning_rate": 1.2168951168767594e-05,
      "loss": 1.1333,
      "step": 247900
    },
    {
      "epoch": 7.576220443575488,
      "grad_norm": 2.88683819770813,
      "learning_rate": 1.2153653163627463e-05,
      "loss": 1.1266,
      "step": 248000
    },
    {
      "epoch": 7.579275371173703,
      "grad_norm": 5.689997673034668,
      "learning_rate": 1.2138355158487334e-05,
      "loss": 1.2176,
      "step": 248100
    },
    {
      "epoch": 7.582330298771919,
      "grad_norm": 19.876480102539062,
      "learning_rate": 1.2123057153347204e-05,
      "loss": 1.0021,
      "step": 248200
    },
    {
      "epoch": 7.585385226370135,
      "grad_norm": 14.01642894744873,
      "learning_rate": 1.2107759148207075e-05,
      "loss": 1.0996,
      "step": 248300
    },
    {
      "epoch": 7.588440153968351,
      "grad_norm": 7.023202896118164,
      "learning_rate": 1.2092461143066944e-05,
      "loss": 1.0367,
      "step": 248400
    },
    {
      "epoch": 7.591495081566567,
      "grad_norm": 2.8547801971435547,
      "learning_rate": 1.2077163137926816e-05,
      "loss": 1.1222,
      "step": 248500
    },
    {
      "epoch": 7.594550009164783,
      "grad_norm": 4.367480754852295,
      "learning_rate": 1.2061865132786685e-05,
      "loss": 1.0549,
      "step": 248600
    },
    {
      "epoch": 7.597604936762998,
      "grad_norm": 13.797048568725586,
      "learning_rate": 1.2046567127646556e-05,
      "loss": 1.085,
      "step": 248700
    },
    {
      "epoch": 7.600659864361215,
      "grad_norm": 4.761076927185059,
      "learning_rate": 1.2031269122506426e-05,
      "loss": 0.9709,
      "step": 248800
    },
    {
      "epoch": 7.603714791959431,
      "grad_norm": 9.693614959716797,
      "learning_rate": 1.2015971117366295e-05,
      "loss": 1.0974,
      "step": 248900
    },
    {
      "epoch": 7.606769719557646,
      "grad_norm": 9.48823356628418,
      "learning_rate": 1.2000673112226166e-05,
      "loss": 1.1598,
      "step": 249000
    },
    {
      "epoch": 7.609824647155863,
      "grad_norm": 2.078655481338501,
      "learning_rate": 1.1985375107086036e-05,
      "loss": 1.1203,
      "step": 249100
    },
    {
      "epoch": 7.6128795747540785,
      "grad_norm": 8.278289794921875,
      "learning_rate": 1.1970230081997307e-05,
      "loss": 1.0562,
      "step": 249200
    },
    {
      "epoch": 7.615934502352294,
      "grad_norm": 20.82103157043457,
      "learning_rate": 1.1954932076857178e-05,
      "loss": 1.0621,
      "step": 249300
    },
    {
      "epoch": 7.61898942995051,
      "grad_norm": 1.3670573234558105,
      "learning_rate": 1.1939634071717048e-05,
      "loss": 1.2731,
      "step": 249400
    },
    {
      "epoch": 7.6220443575487264,
      "grad_norm": 12.100311279296875,
      "learning_rate": 1.1924336066576919e-05,
      "loss": 1.0303,
      "step": 249500
    },
    {
      "epoch": 7.625099285146942,
      "grad_norm": 13.72838020324707,
      "learning_rate": 1.1909038061436788e-05,
      "loss": 1.2989,
      "step": 249600
    },
    {
      "epoch": 7.628154212745158,
      "grad_norm": 8.885602951049805,
      "learning_rate": 1.189374005629666e-05,
      "loss": 0.9904,
      "step": 249700
    },
    {
      "epoch": 7.6312091403433735,
      "grad_norm": 12.419922828674316,
      "learning_rate": 1.1878442051156529e-05,
      "loss": 1.1415,
      "step": 249800
    },
    {
      "epoch": 7.63426406794159,
      "grad_norm": 3.4879326820373535,
      "learning_rate": 1.18631440460164e-05,
      "loss": 1.1963,
      "step": 249900
    },
    {
      "epoch": 7.637318995539806,
      "grad_norm": 10.463134765625,
      "learning_rate": 1.184784604087627e-05,
      "loss": 1.4307,
      "step": 250000
    },
    {
      "epoch": 7.640373923138021,
      "grad_norm": 8.788840293884277,
      "learning_rate": 1.183254803573614e-05,
      "loss": 1.2819,
      "step": 250100
    },
    {
      "epoch": 7.643428850736237,
      "grad_norm": 20.626136779785156,
      "learning_rate": 1.181725003059601e-05,
      "loss": 1.025,
      "step": 250200
    },
    {
      "epoch": 7.646483778334454,
      "grad_norm": 7.884489059448242,
      "learning_rate": 1.1801952025455881e-05,
      "loss": 0.9718,
      "step": 250300
    },
    {
      "epoch": 7.649538705932669,
      "grad_norm": 5.414434432983398,
      "learning_rate": 1.178665402031575e-05,
      "loss": 1.2849,
      "step": 250400
    },
    {
      "epoch": 7.652593633530885,
      "grad_norm": 6.032369613647461,
      "learning_rate": 1.1771356015175622e-05,
      "loss": 1.0759,
      "step": 250500
    },
    {
      "epoch": 7.655648561129102,
      "grad_norm": 4.3075642585754395,
      "learning_rate": 1.1756058010035491e-05,
      "loss": 1.1365,
      "step": 250600
    },
    {
      "epoch": 7.658703488727317,
      "grad_norm": 1.8966740369796753,
      "learning_rate": 1.1740760004895363e-05,
      "loss": 1.0646,
      "step": 250700
    },
    {
      "epoch": 7.661758416325533,
      "grad_norm": 2.841329574584961,
      "learning_rate": 1.1725461999755232e-05,
      "loss": 0.9975,
      "step": 250800
    },
    {
      "epoch": 7.664813343923749,
      "grad_norm": 11.211971282958984,
      "learning_rate": 1.1710163994615103e-05,
      "loss": 1.0907,
      "step": 250900
    },
    {
      "epoch": 7.667868271521965,
      "grad_norm": 6.564040660858154,
      "learning_rate": 1.1694865989474973e-05,
      "loss": 1.2923,
      "step": 251000
    },
    {
      "epoch": 7.670923199120181,
      "grad_norm": 9.287890434265137,
      "learning_rate": 1.1679567984334844e-05,
      "loss": 0.9328,
      "step": 251100
    },
    {
      "epoch": 7.673978126718397,
      "grad_norm": 17.20326805114746,
      "learning_rate": 1.1664269979194713e-05,
      "loss": 1.2543,
      "step": 251200
    },
    {
      "epoch": 7.677033054316612,
      "grad_norm": 8.150099754333496,
      "learning_rate": 1.1648971974054584e-05,
      "loss": 0.9473,
      "step": 251300
    },
    {
      "epoch": 7.680087981914829,
      "grad_norm": 7.258596897125244,
      "learning_rate": 1.1633673968914454e-05,
      "loss": 1.3426,
      "step": 251400
    },
    {
      "epoch": 7.683142909513045,
      "grad_norm": 7.076548099517822,
      "learning_rate": 1.1618375963774325e-05,
      "loss": 1.2426,
      "step": 251500
    },
    {
      "epoch": 7.68619783711126,
      "grad_norm": 6.173647880554199,
      "learning_rate": 1.1603077958634195e-05,
      "loss": 1.1409,
      "step": 251600
    },
    {
      "epoch": 7.689252764709476,
      "grad_norm": 12.131219863891602,
      "learning_rate": 1.1587932933545466e-05,
      "loss": 1.1354,
      "step": 251700
    },
    {
      "epoch": 7.6923076923076925,
      "grad_norm": 6.685665607452393,
      "learning_rate": 1.1572634928405335e-05,
      "loss": 0.8743,
      "step": 251800
    },
    {
      "epoch": 7.695362619905908,
      "grad_norm": 8.49716854095459,
      "learning_rate": 1.1557336923265207e-05,
      "loss": 1.0753,
      "step": 251900
    },
    {
      "epoch": 7.698417547504124,
      "grad_norm": 5.741514682769775,
      "learning_rate": 1.1542038918125076e-05,
      "loss": 0.9022,
      "step": 252000
    },
    {
      "epoch": 7.7014724751023405,
      "grad_norm": 6.333673477172852,
      "learning_rate": 1.1526740912984947e-05,
      "loss": 1.1122,
      "step": 252100
    },
    {
      "epoch": 7.704527402700556,
      "grad_norm": 12.152358055114746,
      "learning_rate": 1.1511442907844817e-05,
      "loss": 1.2149,
      "step": 252200
    },
    {
      "epoch": 7.707582330298772,
      "grad_norm": 6.272040367126465,
      "learning_rate": 1.1496144902704688e-05,
      "loss": 1.0376,
      "step": 252300
    },
    {
      "epoch": 7.7106372578969875,
      "grad_norm": 9.917558670043945,
      "learning_rate": 1.1480846897564557e-05,
      "loss": 1.2105,
      "step": 252400
    },
    {
      "epoch": 7.713692185495204,
      "grad_norm": 5.992608070373535,
      "learning_rate": 1.1465548892424428e-05,
      "loss": 1.2978,
      "step": 252500
    },
    {
      "epoch": 7.71674711309342,
      "grad_norm": 15.975558280944824,
      "learning_rate": 1.1450250887284298e-05,
      "loss": 1.0692,
      "step": 252600
    },
    {
      "epoch": 7.7198020406916354,
      "grad_norm": 4.722947597503662,
      "learning_rate": 1.1434952882144169e-05,
      "loss": 1.2287,
      "step": 252700
    },
    {
      "epoch": 7.722856968289852,
      "grad_norm": 11.227826118469238,
      "learning_rate": 1.1419654877004038e-05,
      "loss": 1.0139,
      "step": 252800
    },
    {
      "epoch": 7.725911895888068,
      "grad_norm": 2.9598705768585205,
      "learning_rate": 1.140435687186391e-05,
      "loss": 1.13,
      "step": 252900
    },
    {
      "epoch": 7.728966823486283,
      "grad_norm": 7.798380374908447,
      "learning_rate": 1.1389058866723779e-05,
      "loss": 1.2424,
      "step": 253000
    },
    {
      "epoch": 7.732021751084499,
      "grad_norm": 12.195430755615234,
      "learning_rate": 1.137376086158365e-05,
      "loss": 1.1667,
      "step": 253100
    },
    {
      "epoch": 7.735076678682715,
      "grad_norm": 1.4271444082260132,
      "learning_rate": 1.135846285644352e-05,
      "loss": 1.0682,
      "step": 253200
    },
    {
      "epoch": 7.738131606280931,
      "grad_norm": 7.447188377380371,
      "learning_rate": 1.1343164851303391e-05,
      "loss": 1.0512,
      "step": 253300
    },
    {
      "epoch": 7.741186533879147,
      "grad_norm": 7.641314506530762,
      "learning_rate": 1.132786684616326e-05,
      "loss": 1.2633,
      "step": 253400
    },
    {
      "epoch": 7.744241461477363,
      "grad_norm": 5.096051216125488,
      "learning_rate": 1.1312568841023131e-05,
      "loss": 0.9755,
      "step": 253500
    },
    {
      "epoch": 7.747296389075579,
      "grad_norm": 5.797981262207031,
      "learning_rate": 1.1297270835883001e-05,
      "loss": 1.0048,
      "step": 253600
    },
    {
      "epoch": 7.750351316673795,
      "grad_norm": 26.29532814025879,
      "learning_rate": 1.1281972830742872e-05,
      "loss": 1.0032,
      "step": 253700
    },
    {
      "epoch": 7.753406244272011,
      "grad_norm": 15.508411407470703,
      "learning_rate": 1.1266674825602742e-05,
      "loss": 1.1657,
      "step": 253800
    },
    {
      "epoch": 7.756461171870226,
      "grad_norm": 5.2701897621154785,
      "learning_rate": 1.1251376820462613e-05,
      "loss": 1.1883,
      "step": 253900
    },
    {
      "epoch": 7.759516099468443,
      "grad_norm": 8.632824897766113,
      "learning_rate": 1.1236078815322484e-05,
      "loss": 1.0014,
      "step": 254000
    },
    {
      "epoch": 7.762571027066659,
      "grad_norm": 7.206718921661377,
      "learning_rate": 1.1220780810182353e-05,
      "loss": 1.0059,
      "step": 254100
    },
    {
      "epoch": 7.765625954664874,
      "grad_norm": 4.950286865234375,
      "learning_rate": 1.1205482805042224e-05,
      "loss": 0.9502,
      "step": 254200
    },
    {
      "epoch": 7.768680882263091,
      "grad_norm": 7.537136077880859,
      "learning_rate": 1.1190184799902094e-05,
      "loss": 0.9475,
      "step": 254300
    },
    {
      "epoch": 7.7717358098613065,
      "grad_norm": 8.627643585205078,
      "learning_rate": 1.1174886794761965e-05,
      "loss": 1.2275,
      "step": 254400
    },
    {
      "epoch": 7.774790737459522,
      "grad_norm": 4.86769962310791,
      "learning_rate": 1.1159588789621833e-05,
      "loss": 1.0626,
      "step": 254500
    },
    {
      "epoch": 7.777845665057738,
      "grad_norm": 2.8688037395477295,
      "learning_rate": 1.1144290784481704e-05,
      "loss": 1.2409,
      "step": 254600
    },
    {
      "epoch": 7.780900592655954,
      "grad_norm": 3.97426700592041,
      "learning_rate": 1.1129145759392975e-05,
      "loss": 1.1648,
      "step": 254700
    },
    {
      "epoch": 7.78395552025417,
      "grad_norm": 12.675214767456055,
      "learning_rate": 1.1113847754252847e-05,
      "loss": 0.9781,
      "step": 254800
    },
    {
      "epoch": 7.787010447852386,
      "grad_norm": 5.294600486755371,
      "learning_rate": 1.1098549749112716e-05,
      "loss": 1.2876,
      "step": 254900
    },
    {
      "epoch": 7.7900653754506015,
      "grad_norm": 18.957189559936523,
      "learning_rate": 1.1083251743972587e-05,
      "loss": 1.2355,
      "step": 255000
    },
    {
      "epoch": 7.793120303048818,
      "grad_norm": 18.206850051879883,
      "learning_rate": 1.1067953738832457e-05,
      "loss": 1.1789,
      "step": 255100
    },
    {
      "epoch": 7.796175230647034,
      "grad_norm": 5.690242290496826,
      "learning_rate": 1.1052655733692328e-05,
      "loss": 1.2387,
      "step": 255200
    },
    {
      "epoch": 7.7992301582452495,
      "grad_norm": 9.694096565246582,
      "learning_rate": 1.1037357728552197e-05,
      "loss": 1.1317,
      "step": 255300
    },
    {
      "epoch": 7.802285085843465,
      "grad_norm": 14.475813865661621,
      "learning_rate": 1.1022059723412068e-05,
      "loss": 1.0649,
      "step": 255400
    },
    {
      "epoch": 7.805340013441682,
      "grad_norm": 3.3793346881866455,
      "learning_rate": 1.1006761718271938e-05,
      "loss": 1.1786,
      "step": 255500
    },
    {
      "epoch": 7.808394941039897,
      "grad_norm": 7.643477916717529,
      "learning_rate": 1.0991463713131809e-05,
      "loss": 1.0217,
      "step": 255600
    },
    {
      "epoch": 7.811449868638113,
      "grad_norm": 11.50840950012207,
      "learning_rate": 1.0976165707991678e-05,
      "loss": 1.2502,
      "step": 255700
    },
    {
      "epoch": 7.81450479623633,
      "grad_norm": 3.7972185611724854,
      "learning_rate": 1.096086770285155e-05,
      "loss": 1.024,
      "step": 255800
    },
    {
      "epoch": 7.817559723834545,
      "grad_norm": 6.405685901641846,
      "learning_rate": 1.0945569697711419e-05,
      "loss": 1.0386,
      "step": 255900
    },
    {
      "epoch": 7.820614651432761,
      "grad_norm": 6.837638854980469,
      "learning_rate": 1.093027169257129e-05,
      "loss": 1.1037,
      "step": 256000
    },
    {
      "epoch": 7.823669579030977,
      "grad_norm": 0.05840953066945076,
      "learning_rate": 1.091497368743116e-05,
      "loss": 1.1804,
      "step": 256100
    },
    {
      "epoch": 7.826724506629192,
      "grad_norm": 3.761370897293091,
      "learning_rate": 1.0899675682291031e-05,
      "loss": 0.9695,
      "step": 256200
    },
    {
      "epoch": 7.829779434227409,
      "grad_norm": 1.4047656059265137,
      "learning_rate": 1.08843776771509e-05,
      "loss": 1.0739,
      "step": 256300
    },
    {
      "epoch": 7.832834361825625,
      "grad_norm": 13.112373352050781,
      "learning_rate": 1.0869079672010772e-05,
      "loss": 1.071,
      "step": 256400
    },
    {
      "epoch": 7.83588928942384,
      "grad_norm": 9.92342472076416,
      "learning_rate": 1.0853781666870641e-05,
      "loss": 1.4552,
      "step": 256500
    },
    {
      "epoch": 7.838944217022057,
      "grad_norm": 3.913094997406006,
      "learning_rate": 1.0838483661730512e-05,
      "loss": 1.2118,
      "step": 256600
    },
    {
      "epoch": 7.841999144620273,
      "grad_norm": 3.993881940841675,
      "learning_rate": 1.0823185656590382e-05,
      "loss": 1.2099,
      "step": 256700
    },
    {
      "epoch": 7.845054072218488,
      "grad_norm": 11.119633674621582,
      "learning_rate": 1.0807887651450253e-05,
      "loss": 1.0744,
      "step": 256800
    },
    {
      "epoch": 7.848108999816704,
      "grad_norm": 1.991754412651062,
      "learning_rate": 1.0792589646310122e-05,
      "loss": 1.0599,
      "step": 256900
    },
    {
      "epoch": 7.8511639274149205,
      "grad_norm": 3.5789756774902344,
      "learning_rate": 1.0777444621221394e-05,
      "loss": 1.2062,
      "step": 257000
    },
    {
      "epoch": 7.854218855013136,
      "grad_norm": 3.4579782485961914,
      "learning_rate": 1.0762146616081263e-05,
      "loss": 1.4146,
      "step": 257100
    },
    {
      "epoch": 7.857273782611352,
      "grad_norm": 6.728309631347656,
      "learning_rate": 1.0746848610941134e-05,
      "loss": 1.0293,
      "step": 257200
    },
    {
      "epoch": 7.8603287102095685,
      "grad_norm": 8.280560493469238,
      "learning_rate": 1.0731550605801004e-05,
      "loss": 0.9475,
      "step": 257300
    },
    {
      "epoch": 7.863383637807784,
      "grad_norm": 7.0217366218566895,
      "learning_rate": 1.0716252600660875e-05,
      "loss": 1.1673,
      "step": 257400
    },
    {
      "epoch": 7.866438565406,
      "grad_norm": 4.446942329406738,
      "learning_rate": 1.0700954595520744e-05,
      "loss": 1.3345,
      "step": 257500
    },
    {
      "epoch": 7.8694934930042155,
      "grad_norm": 2.9662723541259766,
      "learning_rate": 1.0685656590380615e-05,
      "loss": 1.1369,
      "step": 257600
    },
    {
      "epoch": 7.872548420602432,
      "grad_norm": 7.454762935638428,
      "learning_rate": 1.0670358585240485e-05,
      "loss": 0.892,
      "step": 257700
    },
    {
      "epoch": 7.875603348200648,
      "grad_norm": 1.9147623777389526,
      "learning_rate": 1.0655060580100356e-05,
      "loss": 1.2211,
      "step": 257800
    },
    {
      "epoch": 7.8786582757988635,
      "grad_norm": 2.9092390537261963,
      "learning_rate": 1.0639762574960226e-05,
      "loss": 1.0067,
      "step": 257900
    },
    {
      "epoch": 7.881713203397079,
      "grad_norm": 22.353246688842773,
      "learning_rate": 1.0624464569820097e-05,
      "loss": 1.1166,
      "step": 258000
    },
    {
      "epoch": 7.884768130995296,
      "grad_norm": 4.585965156555176,
      "learning_rate": 1.0609166564679966e-05,
      "loss": 1.1157,
      "step": 258100
    },
    {
      "epoch": 7.887823058593511,
      "grad_norm": 10.530981063842773,
      "learning_rate": 1.0593868559539837e-05,
      "loss": 1.1327,
      "step": 258200
    },
    {
      "epoch": 7.890877986191727,
      "grad_norm": 6.027432918548584,
      "learning_rate": 1.0578570554399707e-05,
      "loss": 1.0564,
      "step": 258300
    },
    {
      "epoch": 7.893932913789943,
      "grad_norm": 10.110967636108398,
      "learning_rate": 1.0563272549259578e-05,
      "loss": 1.0226,
      "step": 258400
    },
    {
      "epoch": 7.896987841388159,
      "grad_norm": 6.894031047821045,
      "learning_rate": 1.0547974544119447e-05,
      "loss": 1.0349,
      "step": 258500
    },
    {
      "epoch": 7.900042768986375,
      "grad_norm": 7.420821666717529,
      "learning_rate": 1.0532676538979319e-05,
      "loss": 1.3148,
      "step": 258600
    },
    {
      "epoch": 7.903097696584591,
      "grad_norm": 7.061910629272461,
      "learning_rate": 1.0517378533839188e-05,
      "loss": 1.1127,
      "step": 258700
    },
    {
      "epoch": 7.906152624182807,
      "grad_norm": 5.673000335693359,
      "learning_rate": 1.0502080528699059e-05,
      "loss": 1.0278,
      "step": 258800
    },
    {
      "epoch": 7.909207551781023,
      "grad_norm": 6.4721574783325195,
      "learning_rate": 1.0486782523558929e-05,
      "loss": 1.0613,
      "step": 258900
    },
    {
      "epoch": 7.912262479379239,
      "grad_norm": 2.630610942840576,
      "learning_rate": 1.04716374984702e-05,
      "loss": 1.3414,
      "step": 259000
    },
    {
      "epoch": 7.915317406977454,
      "grad_norm": 3.0736887454986572,
      "learning_rate": 1.045633949333007e-05,
      "loss": 1.0752,
      "step": 259100
    },
    {
      "epoch": 7.918372334575671,
      "grad_norm": 17.174854278564453,
      "learning_rate": 1.044104148818994e-05,
      "loss": 1.0473,
      "step": 259200
    },
    {
      "epoch": 7.921427262173887,
      "grad_norm": 5.889415740966797,
      "learning_rate": 1.042574348304981e-05,
      "loss": 1.0409,
      "step": 259300
    },
    {
      "epoch": 7.924482189772102,
      "grad_norm": 8.806005477905273,
      "learning_rate": 1.0410445477909681e-05,
      "loss": 1.2461,
      "step": 259400
    },
    {
      "epoch": 7.927537117370318,
      "grad_norm": 5.232207775115967,
      "learning_rate": 1.039514747276955e-05,
      "loss": 1.1434,
      "step": 259500
    },
    {
      "epoch": 7.9305920449685345,
      "grad_norm": 6.207265853881836,
      "learning_rate": 1.0379849467629422e-05,
      "loss": 1.1761,
      "step": 259600
    },
    {
      "epoch": 7.93364697256675,
      "grad_norm": 3.6977267265319824,
      "learning_rate": 1.0364551462489291e-05,
      "loss": 1.1148,
      "step": 259700
    },
    {
      "epoch": 7.936701900164966,
      "grad_norm": 9.098258018493652,
      "learning_rate": 1.0349253457349162e-05,
      "loss": 1.0661,
      "step": 259800
    },
    {
      "epoch": 7.939756827763182,
      "grad_norm": 13.900864601135254,
      "learning_rate": 1.0333955452209032e-05,
      "loss": 1.0208,
      "step": 259900
    },
    {
      "epoch": 7.942811755361398,
      "grad_norm": 3.2146785259246826,
      "learning_rate": 1.0318657447068903e-05,
      "loss": 1.31,
      "step": 260000
    },
    {
      "epoch": 7.945866682959614,
      "grad_norm": 5.783294200897217,
      "learning_rate": 1.0303359441928773e-05,
      "loss": 1.357,
      "step": 260100
    },
    {
      "epoch": 7.9489216105578295,
      "grad_norm": 6.471368789672852,
      "learning_rate": 1.0288061436788644e-05,
      "loss": 1.0479,
      "step": 260200
    },
    {
      "epoch": 7.951976538156046,
      "grad_norm": 5.971426486968994,
      "learning_rate": 1.0272763431648513e-05,
      "loss": 1.0926,
      "step": 260300
    },
    {
      "epoch": 7.955031465754262,
      "grad_norm": 2.5386502742767334,
      "learning_rate": 1.0257465426508384e-05,
      "loss": 0.9909,
      "step": 260400
    },
    {
      "epoch": 7.9580863933524775,
      "grad_norm": 13.469215393066406,
      "learning_rate": 1.0242167421368254e-05,
      "loss": 1.0925,
      "step": 260500
    },
    {
      "epoch": 7.961141320950693,
      "grad_norm": 8.153322219848633,
      "learning_rate": 1.0226869416228125e-05,
      "loss": 1.0247,
      "step": 260600
    },
    {
      "epoch": 7.96419624854891,
      "grad_norm": 9.48679256439209,
      "learning_rate": 1.0211571411087994e-05,
      "loss": 1.1772,
      "step": 260700
    },
    {
      "epoch": 7.967251176147125,
      "grad_norm": 3.1017658710479736,
      "learning_rate": 1.0196273405947866e-05,
      "loss": 1.2221,
      "step": 260800
    },
    {
      "epoch": 7.970306103745341,
      "grad_norm": 2.5709776878356934,
      "learning_rate": 1.0180975400807735e-05,
      "loss": 1.5175,
      "step": 260900
    },
    {
      "epoch": 7.973361031343557,
      "grad_norm": 13.61955738067627,
      "learning_rate": 1.0165677395667606e-05,
      "loss": 1.1054,
      "step": 261000
    },
    {
      "epoch": 7.976415958941773,
      "grad_norm": 8.353294372558594,
      "learning_rate": 1.0150379390527476e-05,
      "loss": 1.0566,
      "step": 261100
    },
    {
      "epoch": 7.979470886539989,
      "grad_norm": 8.776516914367676,
      "learning_rate": 1.0135234365438747e-05,
      "loss": 1.3163,
      "step": 261200
    },
    {
      "epoch": 7.982525814138205,
      "grad_norm": 14.955212593078613,
      "learning_rate": 1.0119936360298618e-05,
      "loss": 1.1869,
      "step": 261300
    },
    {
      "epoch": 7.98558074173642,
      "grad_norm": 10.493465423583984,
      "learning_rate": 1.0104638355158488e-05,
      "loss": 1.1279,
      "step": 261400
    },
    {
      "epoch": 7.988635669334637,
      "grad_norm": 25.103595733642578,
      "learning_rate": 1.0089340350018359e-05,
      "loss": 1.26,
      "step": 261500
    },
    {
      "epoch": 7.991690596932853,
      "grad_norm": 6.875821113586426,
      "learning_rate": 1.0074042344878228e-05,
      "loss": 1.021,
      "step": 261600
    },
    {
      "epoch": 7.994745524531068,
      "grad_norm": 1.227331519126892,
      "learning_rate": 1.00587443397381e-05,
      "loss": 1.2149,
      "step": 261700
    },
    {
      "epoch": 7.997800452129285,
      "grad_norm": 6.490436553955078,
      "learning_rate": 1.0043446334597969e-05,
      "loss": 1.1207,
      "step": 261800
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.7329687786399463,
      "eval_loss": 0.7225077748298645,
      "eval_runtime": 1807.7404,
      "eval_samples_per_second": 18.108,
      "eval_steps_per_second": 4.527,
      "step": 261872
    },
    {
      "epoch": 8.0008553797275,
      "grad_norm": 4.015980243682861,
      "learning_rate": 1.0028148329457838e-05,
      "loss": 1.074,
      "step": 261900
    },
    {
      "epoch": 8.003910307325716,
      "grad_norm": 5.590630054473877,
      "learning_rate": 1.001285032431771e-05,
      "loss": 1.3674,
      "step": 262000
    },
    {
      "epoch": 8.006965234923932,
      "grad_norm": 4.984447002410889,
      "learning_rate": 9.997552319177579e-06,
      "loss": 1.1473,
      "step": 262100
    },
    {
      "epoch": 8.010020162522148,
      "grad_norm": 18.93263816833496,
      "learning_rate": 9.98225431403745e-06,
      "loss": 1.5645,
      "step": 262200
    },
    {
      "epoch": 8.013075090120363,
      "grad_norm": 11.363048553466797,
      "learning_rate": 9.96695630889732e-06,
      "loss": 1.2486,
      "step": 262300
    },
    {
      "epoch": 8.01613001771858,
      "grad_norm": 18.34299659729004,
      "learning_rate": 9.95165830375719e-06,
      "loss": 1.0042,
      "step": 262400
    },
    {
      "epoch": 8.019184945316796,
      "grad_norm": 14.327021598815918,
      "learning_rate": 9.93636029861706e-06,
      "loss": 1.0541,
      "step": 262500
    },
    {
      "epoch": 8.022239872915012,
      "grad_norm": 8.86902141571045,
      "learning_rate": 9.921062293476931e-06,
      "loss": 1.3444,
      "step": 262600
    },
    {
      "epoch": 8.025294800513228,
      "grad_norm": 20.985706329345703,
      "learning_rate": 9.9057642883368e-06,
      "loss": 1.2337,
      "step": 262700
    },
    {
      "epoch": 8.028349728111444,
      "grad_norm": 23.785953521728516,
      "learning_rate": 9.890466283196672e-06,
      "loss": 1.2492,
      "step": 262800
    },
    {
      "epoch": 8.03140465570966,
      "grad_norm": 19.114500045776367,
      "learning_rate": 9.875168278056541e-06,
      "loss": 1.3941,
      "step": 262900
    },
    {
      "epoch": 8.034459583307875,
      "grad_norm": 6.7520833015441895,
      "learning_rate": 9.859870272916413e-06,
      "loss": 0.9711,
      "step": 263000
    },
    {
      "epoch": 8.037514510906092,
      "grad_norm": 2.405261516571045,
      "learning_rate": 9.844572267776282e-06,
      "loss": 0.9566,
      "step": 263100
    },
    {
      "epoch": 8.040569438504308,
      "grad_norm": 3.6269869804382324,
      "learning_rate": 9.829274262636153e-06,
      "loss": 1.2396,
      "step": 263200
    },
    {
      "epoch": 8.043624366102524,
      "grad_norm": 4.379708766937256,
      "learning_rate": 9.813976257496023e-06,
      "loss": 1.0622,
      "step": 263300
    },
    {
      "epoch": 8.04667929370074,
      "grad_norm": 4.235758304595947,
      "learning_rate": 9.798678252355894e-06,
      "loss": 0.9627,
      "step": 263400
    },
    {
      "epoch": 8.049734221298955,
      "grad_norm": 5.870180606842041,
      "learning_rate": 9.783380247215763e-06,
      "loss": 0.9076,
      "step": 263500
    },
    {
      "epoch": 8.05278914889717,
      "grad_norm": 5.316646575927734,
      "learning_rate": 9.768082242075634e-06,
      "loss": 1.1855,
      "step": 263600
    },
    {
      "epoch": 8.055844076495386,
      "grad_norm": 5.945902347564697,
      "learning_rate": 9.752784236935504e-06,
      "loss": 1.0746,
      "step": 263700
    },
    {
      "epoch": 8.058899004093602,
      "grad_norm": 5.88398551940918,
      "learning_rate": 9.737486231795375e-06,
      "loss": 1.0877,
      "step": 263800
    },
    {
      "epoch": 8.06195393169182,
      "grad_norm": 5.979194641113281,
      "learning_rate": 9.722188226655244e-06,
      "loss": 1.061,
      "step": 263900
    },
    {
      "epoch": 8.065008859290035,
      "grad_norm": 7.003350734710693,
      "learning_rate": 9.706890221515116e-06,
      "loss": 1.2109,
      "step": 264000
    },
    {
      "epoch": 8.068063786888251,
      "grad_norm": 5.546398162841797,
      "learning_rate": 9.691592216374985e-06,
      "loss": 1.3,
      "step": 264100
    },
    {
      "epoch": 8.071118714486467,
      "grad_norm": 7.2468743324279785,
      "learning_rate": 9.676294211234856e-06,
      "loss": 1.1125,
      "step": 264200
    },
    {
      "epoch": 8.074173642084682,
      "grad_norm": 4.5122575759887695,
      "learning_rate": 9.660996206094726e-06,
      "loss": 0.9328,
      "step": 264300
    },
    {
      "epoch": 8.077228569682898,
      "grad_norm": 5.045060634613037,
      "learning_rate": 9.645698200954597e-06,
      "loss": 1.0791,
      "step": 264400
    },
    {
      "epoch": 8.080283497281114,
      "grad_norm": 6.919245719909668,
      "learning_rate": 9.630400195814466e-06,
      "loss": 1.0238,
      "step": 264500
    },
    {
      "epoch": 8.083338424879331,
      "grad_norm": 2.2628605365753174,
      "learning_rate": 9.615102190674338e-06,
      "loss": 0.9938,
      "step": 264600
    },
    {
      "epoch": 8.086393352477547,
      "grad_norm": 4.537121295928955,
      "learning_rate": 9.599804185534207e-06,
      "loss": 1.0183,
      "step": 264700
    },
    {
      "epoch": 8.089448280075763,
      "grad_norm": 19.846834182739258,
      "learning_rate": 9.584506180394078e-06,
      "loss": 1.1355,
      "step": 264800
    },
    {
      "epoch": 8.092503207673978,
      "grad_norm": 9.19655990600586,
      "learning_rate": 9.569208175253948e-06,
      "loss": 1.3802,
      "step": 264900
    },
    {
      "epoch": 8.095558135272194,
      "grad_norm": 5.800466060638428,
      "learning_rate": 9.553910170113819e-06,
      "loss": 1.1175,
      "step": 265000
    },
    {
      "epoch": 8.09861306287041,
      "grad_norm": 0.6195187568664551,
      "learning_rate": 9.538612164973688e-06,
      "loss": 1.0464,
      "step": 265100
    },
    {
      "epoch": 8.101667990468625,
      "grad_norm": 3.0343387126922607,
      "learning_rate": 9.52346713988496e-06,
      "loss": 0.9321,
      "step": 265200
    },
    {
      "epoch": 8.104722918066841,
      "grad_norm": 19.209545135498047,
      "learning_rate": 9.508169134744829e-06,
      "loss": 1.1553,
      "step": 265300
    },
    {
      "epoch": 8.107777845665058,
      "grad_norm": 3.6193349361419678,
      "learning_rate": 9.4930241096561e-06,
      "loss": 1.1593,
      "step": 265400
    },
    {
      "epoch": 8.110832773263274,
      "grad_norm": 7.876428127288818,
      "learning_rate": 9.477726104515972e-06,
      "loss": 1.1373,
      "step": 265500
    },
    {
      "epoch": 8.11388770086149,
      "grad_norm": 6.429228782653809,
      "learning_rate": 9.462428099375841e-06,
      "loss": 1.0856,
      "step": 265600
    },
    {
      "epoch": 8.116942628459705,
      "grad_norm": 15.394742012023926,
      "learning_rate": 9.447130094235712e-06,
      "loss": 1.1688,
      "step": 265700
    },
    {
      "epoch": 8.119997556057921,
      "grad_norm": 3.984320878982544,
      "learning_rate": 9.431832089095582e-06,
      "loss": 1.2846,
      "step": 265800
    },
    {
      "epoch": 8.123052483656137,
      "grad_norm": 23.03862762451172,
      "learning_rate": 9.416534083955453e-06,
      "loss": 1.3795,
      "step": 265900
    },
    {
      "epoch": 8.126107411254353,
      "grad_norm": 5.847175598144531,
      "learning_rate": 9.401236078815322e-06,
      "loss": 0.988,
      "step": 266000
    },
    {
      "epoch": 8.12916233885257,
      "grad_norm": 4.800594806671143,
      "learning_rate": 9.385938073675193e-06,
      "loss": 1.0304,
      "step": 266100
    },
    {
      "epoch": 8.132217266450786,
      "grad_norm": 13.866425514221191,
      "learning_rate": 9.370640068535063e-06,
      "loss": 1.1209,
      "step": 266200
    },
    {
      "epoch": 8.135272194049001,
      "grad_norm": 4.881516456604004,
      "learning_rate": 9.355342063394934e-06,
      "loss": 1.3845,
      "step": 266300
    },
    {
      "epoch": 8.138327121647217,
      "grad_norm": 8.563227653503418,
      "learning_rate": 9.340044058254803e-06,
      "loss": 1.2194,
      "step": 266400
    },
    {
      "epoch": 8.141382049245433,
      "grad_norm": 6.669600486755371,
      "learning_rate": 9.324746053114675e-06,
      "loss": 0.9601,
      "step": 266500
    },
    {
      "epoch": 8.144436976843648,
      "grad_norm": 6.998952388763428,
      "learning_rate": 9.309448047974546e-06,
      "loss": 1.1143,
      "step": 266600
    },
    {
      "epoch": 8.147491904441864,
      "grad_norm": 6.582401752471924,
      "learning_rate": 9.294150042834415e-06,
      "loss": 0.8779,
      "step": 266700
    },
    {
      "epoch": 8.15054683204008,
      "grad_norm": 7.145859718322754,
      "learning_rate": 9.278852037694286e-06,
      "loss": 1.1478,
      "step": 266800
    },
    {
      "epoch": 8.153601759638297,
      "grad_norm": 4.021243572235107,
      "learning_rate": 9.263554032554156e-06,
      "loss": 1.0431,
      "step": 266900
    },
    {
      "epoch": 8.156656687236513,
      "grad_norm": 8.807994842529297,
      "learning_rate": 9.248256027414027e-06,
      "loss": 1.3477,
      "step": 267000
    },
    {
      "epoch": 8.159711614834729,
      "grad_norm": 14.921202659606934,
      "learning_rate": 9.232958022273896e-06,
      "loss": 1.0948,
      "step": 267100
    },
    {
      "epoch": 8.162766542432944,
      "grad_norm": 12.610527992248535,
      "learning_rate": 9.217660017133768e-06,
      "loss": 1.2238,
      "step": 267200
    },
    {
      "epoch": 8.16582147003116,
      "grad_norm": 6.394305229187012,
      "learning_rate": 9.202362011993637e-06,
      "loss": 1.0813,
      "step": 267300
    },
    {
      "epoch": 8.168876397629376,
      "grad_norm": 3.418588161468506,
      "learning_rate": 9.187064006853508e-06,
      "loss": 1.1551,
      "step": 267400
    },
    {
      "epoch": 8.171931325227591,
      "grad_norm": 10.825216293334961,
      "learning_rate": 9.171766001713376e-06,
      "loss": 1.2308,
      "step": 267500
    },
    {
      "epoch": 8.174986252825809,
      "grad_norm": 14.792623519897461,
      "learning_rate": 9.156467996573247e-06,
      "loss": 0.9937,
      "step": 267600
    },
    {
      "epoch": 8.178041180424025,
      "grad_norm": 4.262557029724121,
      "learning_rate": 9.141169991433117e-06,
      "loss": 1.0925,
      "step": 267700
    },
    {
      "epoch": 8.18109610802224,
      "grad_norm": 7.162014007568359,
      "learning_rate": 9.125871986292988e-06,
      "loss": 1.0426,
      "step": 267800
    },
    {
      "epoch": 8.184151035620456,
      "grad_norm": 4.955055236816406,
      "learning_rate": 9.110573981152857e-06,
      "loss": 1.0415,
      "step": 267900
    },
    {
      "epoch": 8.187205963218672,
      "grad_norm": 8.408870697021484,
      "learning_rate": 9.095275976012728e-06,
      "loss": 1.1049,
      "step": 268000
    },
    {
      "epoch": 8.190260890816887,
      "grad_norm": 3.665238857269287,
      "learning_rate": 9.079977970872598e-06,
      "loss": 1.0314,
      "step": 268100
    },
    {
      "epoch": 8.193315818415103,
      "grad_norm": 11.393309593200684,
      "learning_rate": 9.064679965732469e-06,
      "loss": 1.388,
      "step": 268200
    },
    {
      "epoch": 8.196370746013319,
      "grad_norm": 3.749567747116089,
      "learning_rate": 9.04953494064374e-06,
      "loss": 1.1032,
      "step": 268300
    },
    {
      "epoch": 8.199425673611536,
      "grad_norm": 7.865976333618164,
      "learning_rate": 9.034236935503612e-06,
      "loss": 1.0042,
      "step": 268400
    },
    {
      "epoch": 8.202480601209752,
      "grad_norm": 7.517879962921143,
      "learning_rate": 9.018938930363481e-06,
      "loss": 1.0302,
      "step": 268500
    },
    {
      "epoch": 8.205535528807967,
      "grad_norm": 11.752159118652344,
      "learning_rate": 9.003640925223352e-06,
      "loss": 1.0179,
      "step": 268600
    },
    {
      "epoch": 8.208590456406183,
      "grad_norm": 19.774866104125977,
      "learning_rate": 8.988342920083222e-06,
      "loss": 1.079,
      "step": 268700
    },
    {
      "epoch": 8.211645384004399,
      "grad_norm": 4.879484176635742,
      "learning_rate": 8.973044914943093e-06,
      "loss": 1.2598,
      "step": 268800
    },
    {
      "epoch": 8.214700311602614,
      "grad_norm": 16.126920700073242,
      "learning_rate": 8.957746909802962e-06,
      "loss": 1.0387,
      "step": 268900
    },
    {
      "epoch": 8.21775523920083,
      "grad_norm": 8.293771743774414,
      "learning_rate": 8.942448904662833e-06,
      "loss": 0.9176,
      "step": 269000
    },
    {
      "epoch": 8.220810166799048,
      "grad_norm": 15.987733840942383,
      "learning_rate": 8.927150899522703e-06,
      "loss": 1.2638,
      "step": 269100
    },
    {
      "epoch": 8.223865094397263,
      "grad_norm": 0.636125385761261,
      "learning_rate": 8.911852894382574e-06,
      "loss": 1.1503,
      "step": 269200
    },
    {
      "epoch": 8.226920021995479,
      "grad_norm": 10.927716255187988,
      "learning_rate": 8.896554889242444e-06,
      "loss": 1.2616,
      "step": 269300
    },
    {
      "epoch": 8.229974949593695,
      "grad_norm": 5.758749961853027,
      "learning_rate": 8.881256884102315e-06,
      "loss": 1.031,
      "step": 269400
    },
    {
      "epoch": 8.23302987719191,
      "grad_norm": 3.450516700744629,
      "learning_rate": 8.865958878962184e-06,
      "loss": 1.3088,
      "step": 269500
    },
    {
      "epoch": 8.236084804790126,
      "grad_norm": 3.5513436794281006,
      "learning_rate": 8.850660873822055e-06,
      "loss": 1.1002,
      "step": 269600
    },
    {
      "epoch": 8.239139732388342,
      "grad_norm": 2.624598979949951,
      "learning_rate": 8.835362868681925e-06,
      "loss": 1.2626,
      "step": 269700
    },
    {
      "epoch": 8.24219465998656,
      "grad_norm": 14.29702091217041,
      "learning_rate": 8.820064863541796e-06,
      "loss": 0.9916,
      "step": 269800
    },
    {
      "epoch": 8.245249587584775,
      "grad_norm": 8.308228492736816,
      "learning_rate": 8.804766858401665e-06,
      "loss": 1.0105,
      "step": 269900
    },
    {
      "epoch": 8.24830451518299,
      "grad_norm": 16.899057388305664,
      "learning_rate": 8.789468853261537e-06,
      "loss": 1.0939,
      "step": 270000
    },
    {
      "epoch": 8.251359442781206,
      "grad_norm": 1.783301591873169,
      "learning_rate": 8.774170848121406e-06,
      "loss": 1.0611,
      "step": 270100
    },
    {
      "epoch": 8.254414370379422,
      "grad_norm": 9.797531127929688,
      "learning_rate": 8.758872842981275e-06,
      "loss": 1.3537,
      "step": 270200
    },
    {
      "epoch": 8.257469297977638,
      "grad_norm": 11.86088752746582,
      "learning_rate": 8.743574837841145e-06,
      "loss": 1.1028,
      "step": 270300
    },
    {
      "epoch": 8.260524225575853,
      "grad_norm": 9.216573715209961,
      "learning_rate": 8.728429812752418e-06,
      "loss": 1.0754,
      "step": 270400
    },
    {
      "epoch": 8.263579153174069,
      "grad_norm": 6.141629219055176,
      "learning_rate": 8.713131807612287e-06,
      "loss": 0.928,
      "step": 270500
    },
    {
      "epoch": 8.266634080772286,
      "grad_norm": 2.7297372817993164,
      "learning_rate": 8.697833802472159e-06,
      "loss": 1.2339,
      "step": 270600
    },
    {
      "epoch": 8.269689008370502,
      "grad_norm": 7.393041610717773,
      "learning_rate": 8.682535797332028e-06,
      "loss": 0.9285,
      "step": 270700
    },
    {
      "epoch": 8.272743935968718,
      "grad_norm": 4.140475273132324,
      "learning_rate": 8.6672377921919e-06,
      "loss": 0.9032,
      "step": 270800
    },
    {
      "epoch": 8.275798863566934,
      "grad_norm": 7.7887773513793945,
      "learning_rate": 8.651939787051769e-06,
      "loss": 0.9346,
      "step": 270900
    },
    {
      "epoch": 8.27885379116515,
      "grad_norm": 5.817099571228027,
      "learning_rate": 8.63664178191164e-06,
      "loss": 1.2652,
      "step": 271000
    },
    {
      "epoch": 8.281908718763365,
      "grad_norm": 6.46265983581543,
      "learning_rate": 8.62134377677151e-06,
      "loss": 1.1182,
      "step": 271100
    },
    {
      "epoch": 8.28496364636158,
      "grad_norm": 8.130231857299805,
      "learning_rate": 8.60604577163138e-06,
      "loss": 1.213,
      "step": 271200
    },
    {
      "epoch": 8.288018573959796,
      "grad_norm": 5.918152332305908,
      "learning_rate": 8.59074776649125e-06,
      "loss": 1.1042,
      "step": 271300
    },
    {
      "epoch": 8.291073501558014,
      "grad_norm": 7.691157341003418,
      "learning_rate": 8.575449761351121e-06,
      "loss": 1.1968,
      "step": 271400
    },
    {
      "epoch": 8.29412842915623,
      "grad_norm": 5.404082298278809,
      "learning_rate": 8.56015175621099e-06,
      "loss": 0.9836,
      "step": 271500
    },
    {
      "epoch": 8.297183356754445,
      "grad_norm": 6.673370361328125,
      "learning_rate": 8.544853751070862e-06,
      "loss": 0.969,
      "step": 271600
    },
    {
      "epoch": 8.30023828435266,
      "grad_norm": 1.4831641912460327,
      "learning_rate": 8.529555745930731e-06,
      "loss": 1.2201,
      "step": 271700
    },
    {
      "epoch": 8.303293211950876,
      "grad_norm": 9.45586109161377,
      "learning_rate": 8.514257740790602e-06,
      "loss": 0.9741,
      "step": 271800
    },
    {
      "epoch": 8.306348139549092,
      "grad_norm": 3.7599129676818848,
      "learning_rate": 8.498959735650472e-06,
      "loss": 1.2079,
      "step": 271900
    },
    {
      "epoch": 8.309403067147308,
      "grad_norm": 2.7845051288604736,
      "learning_rate": 8.483661730510343e-06,
      "loss": 1.0476,
      "step": 272000
    },
    {
      "epoch": 8.312457994745525,
      "grad_norm": 9.171859741210938,
      "learning_rate": 8.468363725370212e-06,
      "loss": 1.0662,
      "step": 272100
    },
    {
      "epoch": 8.315512922343741,
      "grad_norm": 3.499755859375,
      "learning_rate": 8.453065720230084e-06,
      "loss": 1.2587,
      "step": 272200
    },
    {
      "epoch": 8.318567849941957,
      "grad_norm": 23.11410140991211,
      "learning_rate": 8.437767715089953e-06,
      "loss": 0.9943,
      "step": 272300
    },
    {
      "epoch": 8.321622777540172,
      "grad_norm": 5.886302471160889,
      "learning_rate": 8.422622690001224e-06,
      "loss": 1.2263,
      "step": 272400
    },
    {
      "epoch": 8.324677705138388,
      "grad_norm": 30.187957763671875,
      "learning_rate": 8.407324684861094e-06,
      "loss": 1.2,
      "step": 272500
    },
    {
      "epoch": 8.327732632736604,
      "grad_norm": 21.372512817382812,
      "learning_rate": 8.392026679720965e-06,
      "loss": 0.9827,
      "step": 272600
    },
    {
      "epoch": 8.33078756033482,
      "grad_norm": 2.0684807300567627,
      "learning_rate": 8.376728674580834e-06,
      "loss": 1.0877,
      "step": 272700
    },
    {
      "epoch": 8.333842487933037,
      "grad_norm": 4.702449798583984,
      "learning_rate": 8.361430669440706e-06,
      "loss": 1.0267,
      "step": 272800
    },
    {
      "epoch": 8.336897415531253,
      "grad_norm": 11.90904712677002,
      "learning_rate": 8.346132664300575e-06,
      "loss": 1.1004,
      "step": 272900
    },
    {
      "epoch": 8.339952343129468,
      "grad_norm": 7.533595561981201,
      "learning_rate": 8.330834659160446e-06,
      "loss": 1.0142,
      "step": 273000
    },
    {
      "epoch": 8.343007270727684,
      "grad_norm": 5.445414066314697,
      "learning_rate": 8.315536654020316e-06,
      "loss": 0.9771,
      "step": 273100
    },
    {
      "epoch": 8.3460621983259,
      "grad_norm": 20.57103729248047,
      "learning_rate": 8.300238648880187e-06,
      "loss": 1.1579,
      "step": 273200
    },
    {
      "epoch": 8.349117125924115,
      "grad_norm": 1.6556389331817627,
      "learning_rate": 8.284940643740056e-06,
      "loss": 1.1873,
      "step": 273300
    },
    {
      "epoch": 8.352172053522331,
      "grad_norm": 8.812055587768555,
      "learning_rate": 8.269642638599927e-06,
      "loss": 1.1352,
      "step": 273400
    },
    {
      "epoch": 8.355226981120547,
      "grad_norm": 2.017615795135498,
      "learning_rate": 8.254344633459797e-06,
      "loss": 1.0793,
      "step": 273500
    },
    {
      "epoch": 8.358281908718764,
      "grad_norm": 15.955278396606445,
      "learning_rate": 8.239046628319668e-06,
      "loss": 1.0959,
      "step": 273600
    },
    {
      "epoch": 8.36133683631698,
      "grad_norm": 6.15826416015625,
      "learning_rate": 8.223748623179538e-06,
      "loss": 0.9902,
      "step": 273700
    },
    {
      "epoch": 8.364391763915195,
      "grad_norm": 3.7233998775482178,
      "learning_rate": 8.208450618039409e-06,
      "loss": 0.9586,
      "step": 273800
    },
    {
      "epoch": 8.367446691513411,
      "grad_norm": 13.683627128601074,
      "learning_rate": 8.193152612899278e-06,
      "loss": 0.9191,
      "step": 273900
    },
    {
      "epoch": 8.370501619111627,
      "grad_norm": 8.237982749938965,
      "learning_rate": 8.17785460775915e-06,
      "loss": 1.2411,
      "step": 274000
    },
    {
      "epoch": 8.373556546709843,
      "grad_norm": 20.673309326171875,
      "learning_rate": 8.162556602619019e-06,
      "loss": 0.9853,
      "step": 274100
    },
    {
      "epoch": 8.376611474308058,
      "grad_norm": 3.522181749343872,
      "learning_rate": 8.14725859747889e-06,
      "loss": 1.0354,
      "step": 274200
    },
    {
      "epoch": 8.379666401906276,
      "grad_norm": 3.1122303009033203,
      "learning_rate": 8.13196059233876e-06,
      "loss": 1.2159,
      "step": 274300
    },
    {
      "epoch": 8.382721329504491,
      "grad_norm": 4.235086917877197,
      "learning_rate": 8.11666258719863e-06,
      "loss": 1.2349,
      "step": 274400
    },
    {
      "epoch": 8.385776257102707,
      "grad_norm": 9.490307807922363,
      "learning_rate": 8.101517562109902e-06,
      "loss": 1.3173,
      "step": 274500
    },
    {
      "epoch": 8.388831184700923,
      "grad_norm": 9.003594398498535,
      "learning_rate": 8.086219556969771e-06,
      "loss": 1.2762,
      "step": 274600
    },
    {
      "epoch": 8.391886112299138,
      "grad_norm": 23.088111877441406,
      "learning_rate": 8.070921551829643e-06,
      "loss": 1.2825,
      "step": 274700
    },
    {
      "epoch": 8.394941039897354,
      "grad_norm": 4.005666732788086,
      "learning_rate": 8.055623546689512e-06,
      "loss": 1.1964,
      "step": 274800
    },
    {
      "epoch": 8.39799596749557,
      "grad_norm": 8.891443252563477,
      "learning_rate": 8.040325541549381e-06,
      "loss": 1.0961,
      "step": 274900
    },
    {
      "epoch": 8.401050895093785,
      "grad_norm": 2.082915782928467,
      "learning_rate": 8.025027536409253e-06,
      "loss": 0.9843,
      "step": 275000
    },
    {
      "epoch": 8.404105822692003,
      "grad_norm": 3.974708080291748,
      "learning_rate": 8.009729531269122e-06,
      "loss": 1.0311,
      "step": 275100
    },
    {
      "epoch": 8.407160750290219,
      "grad_norm": 4.6099982261657715,
      "learning_rate": 7.994431526128993e-06,
      "loss": 1.0286,
      "step": 275200
    },
    {
      "epoch": 8.410215677888434,
      "grad_norm": 8.777955055236816,
      "learning_rate": 7.979133520988863e-06,
      "loss": 1.0127,
      "step": 275300
    },
    {
      "epoch": 8.41327060548665,
      "grad_norm": 9.076045036315918,
      "learning_rate": 7.963835515848734e-06,
      "loss": 1.284,
      "step": 275400
    },
    {
      "epoch": 8.416325533084866,
      "grad_norm": 7.679018974304199,
      "learning_rate": 7.948537510708603e-06,
      "loss": 1.2001,
      "step": 275500
    },
    {
      "epoch": 8.419380460683081,
      "grad_norm": 8.124194145202637,
      "learning_rate": 7.933239505568474e-06,
      "loss": 1.0003,
      "step": 275600
    },
    {
      "epoch": 8.422435388281297,
      "grad_norm": 5.4354729652404785,
      "learning_rate": 7.917941500428344e-06,
      "loss": 1.1908,
      "step": 275700
    },
    {
      "epoch": 8.425490315879514,
      "grad_norm": 8.532306671142578,
      "learning_rate": 7.902643495288215e-06,
      "loss": 0.9961,
      "step": 275800
    },
    {
      "epoch": 8.42854524347773,
      "grad_norm": 9.226902961730957,
      "learning_rate": 7.887345490148085e-06,
      "loss": 1.1599,
      "step": 275900
    },
    {
      "epoch": 8.431600171075946,
      "grad_norm": 16.286670684814453,
      "learning_rate": 7.872047485007956e-06,
      "loss": 1.1251,
      "step": 276000
    },
    {
      "epoch": 8.434655098674162,
      "grad_norm": 11.915355682373047,
      "learning_rate": 7.856749479867825e-06,
      "loss": 1.2525,
      "step": 276100
    },
    {
      "epoch": 8.437710026272377,
      "grad_norm": 2.382817506790161,
      "learning_rate": 7.841451474727696e-06,
      "loss": 1.1368,
      "step": 276200
    },
    {
      "epoch": 8.440764953870593,
      "grad_norm": 6.8670854568481445,
      "learning_rate": 7.826153469587566e-06,
      "loss": 1.1651,
      "step": 276300
    },
    {
      "epoch": 8.443819881468809,
      "grad_norm": 5.160354137420654,
      "learning_rate": 7.810855464447437e-06,
      "loss": 1.1812,
      "step": 276400
    },
    {
      "epoch": 8.446874809067024,
      "grad_norm": 6.945872783660889,
      "learning_rate": 7.795557459307306e-06,
      "loss": 1.1228,
      "step": 276500
    },
    {
      "epoch": 8.449929736665242,
      "grad_norm": 5.746914863586426,
      "learning_rate": 7.780259454167178e-06,
      "loss": 1.4192,
      "step": 276600
    },
    {
      "epoch": 8.452984664263457,
      "grad_norm": 12.10163402557373,
      "learning_rate": 7.764961449027047e-06,
      "loss": 1.0688,
      "step": 276700
    },
    {
      "epoch": 8.456039591861673,
      "grad_norm": 10.079050064086914,
      "learning_rate": 7.749663443886918e-06,
      "loss": 1.3006,
      "step": 276800
    },
    {
      "epoch": 8.459094519459889,
      "grad_norm": 3.420541524887085,
      "learning_rate": 7.734365438746788e-06,
      "loss": 1.1754,
      "step": 276900
    },
    {
      "epoch": 8.462149447058104,
      "grad_norm": 9.938535690307617,
      "learning_rate": 7.719067433606659e-06,
      "loss": 1.1401,
      "step": 277000
    },
    {
      "epoch": 8.46520437465632,
      "grad_norm": 5.8871378898620605,
      "learning_rate": 7.703769428466528e-06,
      "loss": 1.1469,
      "step": 277100
    },
    {
      "epoch": 8.468259302254536,
      "grad_norm": 2.248385190963745,
      "learning_rate": 7.6884714233264e-06,
      "loss": 1.0529,
      "step": 277200
    },
    {
      "epoch": 8.471314229852753,
      "grad_norm": 3.6512601375579834,
      "learning_rate": 7.673173418186269e-06,
      "loss": 0.9395,
      "step": 277300
    },
    {
      "epoch": 8.474369157450969,
      "grad_norm": 17.915828704833984,
      "learning_rate": 7.65787541304614e-06,
      "loss": 0.9789,
      "step": 277400
    },
    {
      "epoch": 8.477424085049185,
      "grad_norm": 11.305119514465332,
      "learning_rate": 7.64257740790601e-06,
      "loss": 1.3882,
      "step": 277500
    },
    {
      "epoch": 8.4804790126474,
      "grad_norm": 21.870290756225586,
      "learning_rate": 7.627432382817282e-06,
      "loss": 1.1529,
      "step": 277600
    },
    {
      "epoch": 8.483533940245616,
      "grad_norm": 1.5844974517822266,
      "learning_rate": 7.61213437767715e-06,
      "loss": 1.2412,
      "step": 277700
    },
    {
      "epoch": 8.486588867843832,
      "grad_norm": 10.673868179321289,
      "learning_rate": 7.596836372537021e-06,
      "loss": 1.2452,
      "step": 277800
    },
    {
      "epoch": 8.489643795442047,
      "grad_norm": 5.95991849899292,
      "learning_rate": 7.581538367396891e-06,
      "loss": 1.2224,
      "step": 277900
    },
    {
      "epoch": 8.492698723040263,
      "grad_norm": 2.6692888736724854,
      "learning_rate": 7.566240362256761e-06,
      "loss": 0.9876,
      "step": 278000
    },
    {
      "epoch": 8.49575365063848,
      "grad_norm": 5.3354363441467285,
      "learning_rate": 7.5509423571166316e-06,
      "loss": 1.0738,
      "step": 278100
    },
    {
      "epoch": 8.498808578236696,
      "grad_norm": 10.510762214660645,
      "learning_rate": 7.535644351976502e-06,
      "loss": 1.056,
      "step": 278200
    },
    {
      "epoch": 8.501863505834912,
      "grad_norm": 3.5407023429870605,
      "learning_rate": 7.520346346836372e-06,
      "loss": 1.1578,
      "step": 278300
    },
    {
      "epoch": 8.504918433433128,
      "grad_norm": 17.953824996948242,
      "learning_rate": 7.5050483416962425e-06,
      "loss": 0.9995,
      "step": 278400
    },
    {
      "epoch": 8.507973361031343,
      "grad_norm": 8.821012496948242,
      "learning_rate": 7.489750336556113e-06,
      "loss": 1.2082,
      "step": 278500
    },
    {
      "epoch": 8.511028288629559,
      "grad_norm": 11.681960105895996,
      "learning_rate": 7.474452331415983e-06,
      "loss": 1.2717,
      "step": 278600
    },
    {
      "epoch": 8.514083216227775,
      "grad_norm": 8.691248893737793,
      "learning_rate": 7.4591543262758534e-06,
      "loss": 1.3788,
      "step": 278700
    },
    {
      "epoch": 8.517138143825992,
      "grad_norm": 7.976315498352051,
      "learning_rate": 7.443856321135724e-06,
      "loss": 1.043,
      "step": 278800
    },
    {
      "epoch": 8.520193071424208,
      "grad_norm": 16.34596061706543,
      "learning_rate": 7.428558315995594e-06,
      "loss": 1.0461,
      "step": 278900
    },
    {
      "epoch": 8.523247999022423,
      "grad_norm": 4.934373378753662,
      "learning_rate": 7.413260310855464e-06,
      "loss": 0.9906,
      "step": 279000
    },
    {
      "epoch": 8.52630292662064,
      "grad_norm": 10.014142990112305,
      "learning_rate": 7.397962305715335e-06,
      "loss": 1.0192,
      "step": 279100
    },
    {
      "epoch": 8.529357854218855,
      "grad_norm": 2.8271279335021973,
      "learning_rate": 7.382664300575205e-06,
      "loss": 0.9847,
      "step": 279200
    },
    {
      "epoch": 8.53241278181707,
      "grad_norm": 3.4308598041534424,
      "learning_rate": 7.367366295435075e-06,
      "loss": 1.1499,
      "step": 279300
    },
    {
      "epoch": 8.535467709415286,
      "grad_norm": 4.892797946929932,
      "learning_rate": 7.352068290294946e-06,
      "loss": 1.1534,
      "step": 279400
    },
    {
      "epoch": 8.538522637013504,
      "grad_norm": 6.5739922523498535,
      "learning_rate": 7.336770285154816e-06,
      "loss": 1.0535,
      "step": 279500
    },
    {
      "epoch": 8.54157756461172,
      "grad_norm": 16.024093627929688,
      "learning_rate": 7.321472280014686e-06,
      "loss": 1.2655,
      "step": 279600
    },
    {
      "epoch": 8.544632492209935,
      "grad_norm": 4.712281703948975,
      "learning_rate": 7.3061742748745565e-06,
      "loss": 1.0642,
      "step": 279700
    },
    {
      "epoch": 8.54768741980815,
      "grad_norm": 2.757760524749756,
      "learning_rate": 7.290876269734427e-06,
      "loss": 1.302,
      "step": 279800
    },
    {
      "epoch": 8.550742347406366,
      "grad_norm": 6.04436731338501,
      "learning_rate": 7.275578264594297e-06,
      "loss": 1.0924,
      "step": 279900
    },
    {
      "epoch": 8.553797275004582,
      "grad_norm": 4.666031837463379,
      "learning_rate": 7.2602802594541675e-06,
      "loss": 1.2949,
      "step": 280000
    },
    {
      "epoch": 8.556852202602798,
      "grad_norm": 4.954720497131348,
      "learning_rate": 7.244982254314038e-06,
      "loss": 1.1814,
      "step": 280100
    },
    {
      "epoch": 8.559907130201013,
      "grad_norm": 11.050684928894043,
      "learning_rate": 7.229684249173908e-06,
      "loss": 1.175,
      "step": 280200
    },
    {
      "epoch": 8.562962057799231,
      "grad_norm": 24.601699829101562,
      "learning_rate": 7.21453922408518e-06,
      "loss": 1.2731,
      "step": 280300
    },
    {
      "epoch": 8.566016985397447,
      "grad_norm": 8.36152172088623,
      "learning_rate": 7.199241218945051e-06,
      "loss": 1.2436,
      "step": 280400
    },
    {
      "epoch": 8.569071912995662,
      "grad_norm": 6.849078178405762,
      "learning_rate": 7.183943213804919e-06,
      "loss": 1.0235,
      "step": 280500
    },
    {
      "epoch": 8.572126840593878,
      "grad_norm": 5.327115058898926,
      "learning_rate": 7.1686452086647895e-06,
      "loss": 0.8993,
      "step": 280600
    },
    {
      "epoch": 8.575181768192094,
      "grad_norm": 5.185020446777344,
      "learning_rate": 7.15334720352466e-06,
      "loss": 0.9513,
      "step": 280700
    },
    {
      "epoch": 8.57823669579031,
      "grad_norm": 5.5773444175720215,
      "learning_rate": 7.13804919838453e-06,
      "loss": 0.9164,
      "step": 280800
    },
    {
      "epoch": 8.581291623388525,
      "grad_norm": 6.858551025390625,
      "learning_rate": 7.1227511932444004e-06,
      "loss": 0.9887,
      "step": 280900
    },
    {
      "epoch": 8.58434655098674,
      "grad_norm": 7.5612616539001465,
      "learning_rate": 7.107453188104271e-06,
      "loss": 1.022,
      "step": 281000
    },
    {
      "epoch": 8.587401478584958,
      "grad_norm": 10.310321807861328,
      "learning_rate": 7.092155182964141e-06,
      "loss": 1.2092,
      "step": 281100
    },
    {
      "epoch": 8.590456406183174,
      "grad_norm": 1.7474365234375,
      "learning_rate": 7.076857177824011e-06,
      "loss": 0.9727,
      "step": 281200
    },
    {
      "epoch": 8.59351133378139,
      "grad_norm": 8.458216667175293,
      "learning_rate": 7.061559172683882e-06,
      "loss": 1.3668,
      "step": 281300
    },
    {
      "epoch": 8.596566261379605,
      "grad_norm": 8.884044647216797,
      "learning_rate": 7.046261167543752e-06,
      "loss": 1.0134,
      "step": 281400
    },
    {
      "epoch": 8.599621188977821,
      "grad_norm": 8.570650100708008,
      "learning_rate": 7.030963162403622e-06,
      "loss": 1.2935,
      "step": 281500
    },
    {
      "epoch": 8.602676116576037,
      "grad_norm": 2.1662871837615967,
      "learning_rate": 7.015665157263493e-06,
      "loss": 1.5576,
      "step": 281600
    },
    {
      "epoch": 8.605731044174252,
      "grad_norm": 2.5884835720062256,
      "learning_rate": 7.000367152123363e-06,
      "loss": 1.1192,
      "step": 281700
    },
    {
      "epoch": 8.60878597177247,
      "grad_norm": 3.6408133506774902,
      "learning_rate": 6.985069146983233e-06,
      "loss": 1.1488,
      "step": 281800
    },
    {
      "epoch": 8.611840899370685,
      "grad_norm": 11.819531440734863,
      "learning_rate": 6.9697711418431035e-06,
      "loss": 1.1942,
      "step": 281900
    },
    {
      "epoch": 8.614895826968901,
      "grad_norm": 3.195770263671875,
      "learning_rate": 6.954473136702974e-06,
      "loss": 0.9802,
      "step": 282000
    },
    {
      "epoch": 8.617950754567117,
      "grad_norm": 1.4171463251113892,
      "learning_rate": 6.939175131562844e-06,
      "loss": 1.0642,
      "step": 282100
    },
    {
      "epoch": 8.621005682165332,
      "grad_norm": 5.6690802574157715,
      "learning_rate": 6.9238771264227145e-06,
      "loss": 1.0642,
      "step": 282200
    },
    {
      "epoch": 8.624060609763548,
      "grad_norm": 13.222604751586914,
      "learning_rate": 6.908579121282585e-06,
      "loss": 1.0442,
      "step": 282300
    },
    {
      "epoch": 8.627115537361764,
      "grad_norm": 9.634571075439453,
      "learning_rate": 6.893281116142455e-06,
      "loss": 1.315,
      "step": 282400
    },
    {
      "epoch": 8.630170464959981,
      "grad_norm": 10.725183486938477,
      "learning_rate": 6.877983111002325e-06,
      "loss": 1.2638,
      "step": 282500
    },
    {
      "epoch": 8.633225392558197,
      "grad_norm": 12.106181144714355,
      "learning_rate": 6.862685105862196e-06,
      "loss": 1.2009,
      "step": 282600
    },
    {
      "epoch": 8.636280320156413,
      "grad_norm": 8.581542015075684,
      "learning_rate": 6.847387100722066e-06,
      "loss": 0.9775,
      "step": 282700
    },
    {
      "epoch": 8.639335247754628,
      "grad_norm": 7.633388996124268,
      "learning_rate": 6.832089095581936e-06,
      "loss": 1.227,
      "step": 282800
    },
    {
      "epoch": 8.642390175352844,
      "grad_norm": 5.2652130126953125,
      "learning_rate": 6.816791090441807e-06,
      "loss": 1.2009,
      "step": 282900
    },
    {
      "epoch": 8.64544510295106,
      "grad_norm": 2.624347448348999,
      "learning_rate": 6.801493085301677e-06,
      "loss": 1.2672,
      "step": 283000
    },
    {
      "epoch": 8.648500030549275,
      "grad_norm": 4.857085227966309,
      "learning_rate": 6.786195080161547e-06,
      "loss": 1.18,
      "step": 283100
    },
    {
      "epoch": 8.651554958147491,
      "grad_norm": 4.37367582321167,
      "learning_rate": 6.7708970750214176e-06,
      "loss": 1.0433,
      "step": 283200
    },
    {
      "epoch": 8.654609885745709,
      "grad_norm": 8.585063934326172,
      "learning_rate": 6.755599069881288e-06,
      "loss": 1.3093,
      "step": 283300
    },
    {
      "epoch": 8.657664813343924,
      "grad_norm": 9.964705467224121,
      "learning_rate": 6.740301064741158e-06,
      "loss": 1.3228,
      "step": 283400
    },
    {
      "epoch": 8.66071974094214,
      "grad_norm": 7.176522731781006,
      "learning_rate": 6.7250030596010285e-06,
      "loss": 1.131,
      "step": 283500
    },
    {
      "epoch": 8.663774668540356,
      "grad_norm": 7.689813613891602,
      "learning_rate": 6.709705054460899e-06,
      "loss": 1.1918,
      "step": 283600
    },
    {
      "epoch": 8.666829596138571,
      "grad_norm": 6.8202714920043945,
      "learning_rate": 6.694407049320769e-06,
      "loss": 0.9765,
      "step": 283700
    },
    {
      "epoch": 8.669884523736787,
      "grad_norm": 6.130142688751221,
      "learning_rate": 6.6791090441806394e-06,
      "loss": 1.1884,
      "step": 283800
    },
    {
      "epoch": 8.672939451335003,
      "grad_norm": 7.813923358917236,
      "learning_rate": 6.66381103904051e-06,
      "loss": 0.9992,
      "step": 283900
    },
    {
      "epoch": 8.675994378933218,
      "grad_norm": 3.493478775024414,
      "learning_rate": 6.64866601395178e-06,
      "loss": 1.1072,
      "step": 284000
    },
    {
      "epoch": 8.679049306531436,
      "grad_norm": 9.310433387756348,
      "learning_rate": 6.6333680088116506e-06,
      "loss": 1.118,
      "step": 284100
    },
    {
      "epoch": 8.682104234129651,
      "grad_norm": 20.439428329467773,
      "learning_rate": 6.618070003671521e-06,
      "loss": 0.9944,
      "step": 284200
    },
    {
      "epoch": 8.685159161727867,
      "grad_norm": 6.449985504150391,
      "learning_rate": 6.602771998531391e-06,
      "loss": 1.0403,
      "step": 284300
    },
    {
      "epoch": 8.688214089326083,
      "grad_norm": 1.44782292842865,
      "learning_rate": 6.5874739933912615e-06,
      "loss": 1.0567,
      "step": 284400
    },
    {
      "epoch": 8.691269016924299,
      "grad_norm": 9.718864440917969,
      "learning_rate": 6.572175988251132e-06,
      "loss": 0.9943,
      "step": 284500
    },
    {
      "epoch": 8.694323944522514,
      "grad_norm": 3.4997923374176025,
      "learning_rate": 6.556877983111002e-06,
      "loss": 1.2147,
      "step": 284600
    },
    {
      "epoch": 8.69737887212073,
      "grad_norm": 3.804351568222046,
      "learning_rate": 6.541579977970872e-06,
      "loss": 0.8873,
      "step": 284700
    },
    {
      "epoch": 8.700433799718947,
      "grad_norm": 8.257146835327148,
      "learning_rate": 6.526281972830743e-06,
      "loss": 1.3033,
      "step": 284800
    },
    {
      "epoch": 8.703488727317163,
      "grad_norm": 5.3684821128845215,
      "learning_rate": 6.510983967690613e-06,
      "loss": 0.9764,
      "step": 284900
    },
    {
      "epoch": 8.706543654915379,
      "grad_norm": 4.762368202209473,
      "learning_rate": 6.495685962550483e-06,
      "loss": 1.0703,
      "step": 285000
    },
    {
      "epoch": 8.709598582513594,
      "grad_norm": 4.3477678298950195,
      "learning_rate": 6.480387957410354e-06,
      "loss": 1.1059,
      "step": 285100
    },
    {
      "epoch": 8.71265351011181,
      "grad_norm": 10.474966049194336,
      "learning_rate": 6.465089952270224e-06,
      "loss": 1.0836,
      "step": 285200
    },
    {
      "epoch": 8.715708437710026,
      "grad_norm": 5.3630194664001465,
      "learning_rate": 6.449791947130094e-06,
      "loss": 1.01,
      "step": 285300
    },
    {
      "epoch": 8.718763365308241,
      "grad_norm": 2.3403940200805664,
      "learning_rate": 6.434493941989965e-06,
      "loss": 1.0378,
      "step": 285400
    },
    {
      "epoch": 8.721818292906459,
      "grad_norm": 5.546512603759766,
      "learning_rate": 6.419195936849835e-06,
      "loss": 1.1563,
      "step": 285500
    },
    {
      "epoch": 8.724873220504675,
      "grad_norm": 39.056983947753906,
      "learning_rate": 6.403897931709705e-06,
      "loss": 1.2946,
      "step": 285600
    },
    {
      "epoch": 8.72792814810289,
      "grad_norm": 10.053462028503418,
      "learning_rate": 6.3885999265695755e-06,
      "loss": 1.1364,
      "step": 285700
    },
    {
      "epoch": 8.730983075701106,
      "grad_norm": 2.9167656898498535,
      "learning_rate": 6.373301921429446e-06,
      "loss": 1.3461,
      "step": 285800
    },
    {
      "epoch": 8.734038003299322,
      "grad_norm": 4.460560321807861,
      "learning_rate": 6.358003916289316e-06,
      "loss": 1.0223,
      "step": 285900
    },
    {
      "epoch": 8.737092930897537,
      "grad_norm": 7.659598350524902,
      "learning_rate": 6.3428588912005875e-06,
      "loss": 1.1359,
      "step": 286000
    },
    {
      "epoch": 8.740147858495753,
      "grad_norm": 8.358567237854004,
      "learning_rate": 6.327560886060458e-06,
      "loss": 1.039,
      "step": 286100
    },
    {
      "epoch": 8.74320278609397,
      "grad_norm": 7.288459300994873,
      "learning_rate": 6.312262880920328e-06,
      "loss": 1.3169,
      "step": 286200
    },
    {
      "epoch": 8.746257713692186,
      "grad_norm": 7.670612812042236,
      "learning_rate": 6.2969648757801984e-06,
      "loss": 1.0615,
      "step": 286300
    },
    {
      "epoch": 8.749312641290402,
      "grad_norm": 4.944478988647461,
      "learning_rate": 6.281666870640069e-06,
      "loss": 0.9396,
      "step": 286400
    },
    {
      "epoch": 8.752367568888618,
      "grad_norm": 14.047966957092285,
      "learning_rate": 6.266368865499939e-06,
      "loss": 1.0738,
      "step": 286500
    },
    {
      "epoch": 8.755422496486833,
      "grad_norm": 2.8102049827575684,
      "learning_rate": 6.251070860359809e-06,
      "loss": 1.0472,
      "step": 286600
    },
    {
      "epoch": 8.758477424085049,
      "grad_norm": 8.91782283782959,
      "learning_rate": 6.23577285521968e-06,
      "loss": 1.2268,
      "step": 286700
    },
    {
      "epoch": 8.761532351683265,
      "grad_norm": 13.246162414550781,
      "learning_rate": 6.22047485007955e-06,
      "loss": 1.2976,
      "step": 286800
    },
    {
      "epoch": 8.76458727928148,
      "grad_norm": 3.5406272411346436,
      "learning_rate": 6.20517684493942e-06,
      "loss": 1.4642,
      "step": 286900
    },
    {
      "epoch": 8.767642206879698,
      "grad_norm": 0.011519878171384335,
      "learning_rate": 6.189878839799291e-06,
      "loss": 1.1179,
      "step": 287000
    },
    {
      "epoch": 8.770697134477913,
      "grad_norm": 3.2644083499908447,
      "learning_rate": 6.174580834659161e-06,
      "loss": 1.1199,
      "step": 287100
    },
    {
      "epoch": 8.77375206207613,
      "grad_norm": 5.073540210723877,
      "learning_rate": 6.159282829519031e-06,
      "loss": 1.1659,
      "step": 287200
    },
    {
      "epoch": 8.776806989674345,
      "grad_norm": 6.0912089347839355,
      "learning_rate": 6.1439848243789015e-06,
      "loss": 1.286,
      "step": 287300
    },
    {
      "epoch": 8.77986191727256,
      "grad_norm": 7.724213600158691,
      "learning_rate": 6.128686819238772e-06,
      "loss": 1.0862,
      "step": 287400
    },
    {
      "epoch": 8.782916844870776,
      "grad_norm": 14.391536712646484,
      "learning_rate": 6.113388814098642e-06,
      "loss": 1.1989,
      "step": 287500
    },
    {
      "epoch": 8.785971772468992,
      "grad_norm": 4.909726619720459,
      "learning_rate": 6.0980908089585124e-06,
      "loss": 0.9811,
      "step": 287600
    },
    {
      "epoch": 8.789026700067208,
      "grad_norm": 3.266512870788574,
      "learning_rate": 6.082792803818383e-06,
      "loss": 1.0923,
      "step": 287700
    },
    {
      "epoch": 8.792081627665425,
      "grad_norm": 17.347148895263672,
      "learning_rate": 6.067494798678253e-06,
      "loss": 1.1496,
      "step": 287800
    },
    {
      "epoch": 8.79513655526364,
      "grad_norm": 13.67123031616211,
      "learning_rate": 6.052196793538123e-06,
      "loss": 1.0582,
      "step": 287900
    },
    {
      "epoch": 8.798191482861856,
      "grad_norm": 13.52348804473877,
      "learning_rate": 6.036898788397994e-06,
      "loss": 1.0022,
      "step": 288000
    },
    {
      "epoch": 8.801246410460072,
      "grad_norm": 6.078239440917969,
      "learning_rate": 6.021600783257864e-06,
      "loss": 1.2463,
      "step": 288100
    },
    {
      "epoch": 8.804301338058288,
      "grad_norm": 2.679036855697632,
      "learning_rate": 6.006302778117734e-06,
      "loss": 1.0169,
      "step": 288200
    },
    {
      "epoch": 8.807356265656503,
      "grad_norm": 4.248081684112549,
      "learning_rate": 5.991004772977605e-06,
      "loss": 0.9784,
      "step": 288300
    },
    {
      "epoch": 8.810411193254719,
      "grad_norm": 0.44658416509628296,
      "learning_rate": 5.975706767837475e-06,
      "loss": 1.0276,
      "step": 288400
    },
    {
      "epoch": 8.813466120852937,
      "grad_norm": 3.2473037242889404,
      "learning_rate": 5.960408762697345e-06,
      "loss": 0.8594,
      "step": 288500
    },
    {
      "epoch": 8.816521048451152,
      "grad_norm": 8.046907424926758,
      "learning_rate": 5.9451107575572155e-06,
      "loss": 1.0015,
      "step": 288600
    },
    {
      "epoch": 8.819575976049368,
      "grad_norm": 5.975189208984375,
      "learning_rate": 5.929812752417085e-06,
      "loss": 1.2663,
      "step": 288700
    },
    {
      "epoch": 8.822630903647584,
      "grad_norm": 6.2016215324401855,
      "learning_rate": 5.914514747276955e-06,
      "loss": 1.1905,
      "step": 288800
    },
    {
      "epoch": 8.8256858312458,
      "grad_norm": 5.386969089508057,
      "learning_rate": 5.899216742136826e-06,
      "loss": 1.1232,
      "step": 288900
    },
    {
      "epoch": 8.828740758844015,
      "grad_norm": 3.4425082206726074,
      "learning_rate": 5.883918736996696e-06,
      "loss": 1.226,
      "step": 289000
    },
    {
      "epoch": 8.83179568644223,
      "grad_norm": 7.528200626373291,
      "learning_rate": 5.868620731856566e-06,
      "loss": 1.1668,
      "step": 289100
    },
    {
      "epoch": 8.834850614040448,
      "grad_norm": 17.77597427368164,
      "learning_rate": 5.8533227267164366e-06,
      "loss": 1.0165,
      "step": 289200
    },
    {
      "epoch": 8.837905541638664,
      "grad_norm": 9.727446556091309,
      "learning_rate": 5.838024721576307e-06,
      "loss": 1.0466,
      "step": 289300
    },
    {
      "epoch": 8.84096046923688,
      "grad_norm": 4.185564994812012,
      "learning_rate": 5.822726716436177e-06,
      "loss": 1.2831,
      "step": 289400
    },
    {
      "epoch": 8.844015396835095,
      "grad_norm": 13.54357624053955,
      "learning_rate": 5.8075816913474485e-06,
      "loss": 1.2904,
      "step": 289500
    },
    {
      "epoch": 8.84707032443331,
      "grad_norm": 3.099177598953247,
      "learning_rate": 5.792283686207319e-06,
      "loss": 1.1769,
      "step": 289600
    },
    {
      "epoch": 8.850125252031527,
      "grad_norm": 18.087329864501953,
      "learning_rate": 5.776985681067189e-06,
      "loss": 1.0588,
      "step": 289700
    },
    {
      "epoch": 8.853180179629742,
      "grad_norm": 4.369546413421631,
      "learning_rate": 5.7616876759270595e-06,
      "loss": 0.9452,
      "step": 289800
    },
    {
      "epoch": 8.856235107227958,
      "grad_norm": 6.6351447105407715,
      "learning_rate": 5.74638967078693e-06,
      "loss": 1.1907,
      "step": 289900
    },
    {
      "epoch": 8.859290034826175,
      "grad_norm": 13.804483413696289,
      "learning_rate": 5.7310916656468e-06,
      "loss": 1.4949,
      "step": 290000
    },
    {
      "epoch": 8.862344962424391,
      "grad_norm": 14.235151290893555,
      "learning_rate": 5.71579366050667e-06,
      "loss": 1.0575,
      "step": 290100
    },
    {
      "epoch": 8.865399890022607,
      "grad_norm": 5.2210235595703125,
      "learning_rate": 5.700495655366541e-06,
      "loss": 1.2209,
      "step": 290200
    },
    {
      "epoch": 8.868454817620822,
      "grad_norm": 4.419956684112549,
      "learning_rate": 5.685197650226411e-06,
      "loss": 1.1516,
      "step": 290300
    },
    {
      "epoch": 8.871509745219038,
      "grad_norm": 11.630651473999023,
      "learning_rate": 5.669899645086281e-06,
      "loss": 1.1808,
      "step": 290400
    },
    {
      "epoch": 8.874564672817254,
      "grad_norm": 4.662689685821533,
      "learning_rate": 5.654601639946152e-06,
      "loss": 1.5103,
      "step": 290500
    },
    {
      "epoch": 8.87761960041547,
      "grad_norm": 5.733534812927246,
      "learning_rate": 5.639303634806022e-06,
      "loss": 1.1499,
      "step": 290600
    },
    {
      "epoch": 8.880674528013685,
      "grad_norm": 4.516679763793945,
      "learning_rate": 5.624005629665892e-06,
      "loss": 1.0164,
      "step": 290700
    },
    {
      "epoch": 8.883729455611903,
      "grad_norm": 9.872199058532715,
      "learning_rate": 5.6087076245257626e-06,
      "loss": 1.0991,
      "step": 290800
    },
    {
      "epoch": 8.886784383210118,
      "grad_norm": 5.59382963180542,
      "learning_rate": 5.593409619385633e-06,
      "loss": 1.1131,
      "step": 290900
    },
    {
      "epoch": 8.889839310808334,
      "grad_norm": 4.885242938995361,
      "learning_rate": 5.578111614245503e-06,
      "loss": 1.1361,
      "step": 291000
    },
    {
      "epoch": 8.89289423840655,
      "grad_norm": 9.7994966506958,
      "learning_rate": 5.5628136091053735e-06,
      "loss": 1.0083,
      "step": 291100
    },
    {
      "epoch": 8.895949166004765,
      "grad_norm": 6.640185356140137,
      "learning_rate": 5.547515603965244e-06,
      "loss": 1.1433,
      "step": 291200
    },
    {
      "epoch": 8.899004093602981,
      "grad_norm": 8.89209270477295,
      "learning_rate": 5.532217598825114e-06,
      "loss": 0.9466,
      "step": 291300
    },
    {
      "epoch": 8.902059021201197,
      "grad_norm": 14.11497974395752,
      "learning_rate": 5.516919593684984e-06,
      "loss": 1.0704,
      "step": 291400
    },
    {
      "epoch": 8.905113948799414,
      "grad_norm": 4.109190940856934,
      "learning_rate": 5.501621588544854e-06,
      "loss": 1.2348,
      "step": 291500
    },
    {
      "epoch": 8.90816887639763,
      "grad_norm": 4.684236526489258,
      "learning_rate": 5.486323583404724e-06,
      "loss": 1.1173,
      "step": 291600
    },
    {
      "epoch": 8.911223803995846,
      "grad_norm": 5.410808563232422,
      "learning_rate": 5.4710255782645945e-06,
      "loss": 1.1324,
      "step": 291700
    },
    {
      "epoch": 8.914278731594061,
      "grad_norm": 7.469236373901367,
      "learning_rate": 5.455727573124465e-06,
      "loss": 1.0794,
      "step": 291800
    },
    {
      "epoch": 8.917333659192277,
      "grad_norm": 7.113474369049072,
      "learning_rate": 5.440429567984335e-06,
      "loss": 1.1584,
      "step": 291900
    },
    {
      "epoch": 8.920388586790493,
      "grad_norm": 7.708322048187256,
      "learning_rate": 5.4251315628442054e-06,
      "loss": 1.0252,
      "step": 292000
    },
    {
      "epoch": 8.923443514388708,
      "grad_norm": 12.872169494628906,
      "learning_rate": 5.409833557704076e-06,
      "loss": 1.3283,
      "step": 292100
    },
    {
      "epoch": 8.926498441986926,
      "grad_norm": 12.395779609680176,
      "learning_rate": 5.394535552563946e-06,
      "loss": 1.3083,
      "step": 292200
    },
    {
      "epoch": 8.929553369585141,
      "grad_norm": 0.006554835010319948,
      "learning_rate": 5.379237547423816e-06,
      "loss": 1.2929,
      "step": 292300
    },
    {
      "epoch": 8.932608297183357,
      "grad_norm": 15.197608947753906,
      "learning_rate": 5.363939542283687e-06,
      "loss": 1.0421,
      "step": 292400
    },
    {
      "epoch": 8.935663224781573,
      "grad_norm": 4.074192047119141,
      "learning_rate": 5.348641537143557e-06,
      "loss": 1.1509,
      "step": 292500
    },
    {
      "epoch": 8.938718152379789,
      "grad_norm": 10.62905502319336,
      "learning_rate": 5.333496512054828e-06,
      "loss": 0.9111,
      "step": 292600
    },
    {
      "epoch": 8.941773079978004,
      "grad_norm": 5.3739118576049805,
      "learning_rate": 5.318198506914699e-06,
      "loss": 1.1221,
      "step": 292700
    },
    {
      "epoch": 8.94482800757622,
      "grad_norm": 7.762158393859863,
      "learning_rate": 5.302900501774569e-06,
      "loss": 1.1094,
      "step": 292800
    },
    {
      "epoch": 8.947882935174436,
      "grad_norm": 2.196645736694336,
      "learning_rate": 5.287602496634439e-06,
      "loss": 1.1351,
      "step": 292900
    },
    {
      "epoch": 8.950937862772653,
      "grad_norm": 2.2527201175689697,
      "learning_rate": 5.2723044914943096e-06,
      "loss": 0.873,
      "step": 293000
    },
    {
      "epoch": 8.953992790370869,
      "grad_norm": 4.806405544281006,
      "learning_rate": 5.25700648635418e-06,
      "loss": 1.2991,
      "step": 293100
    },
    {
      "epoch": 8.957047717969084,
      "grad_norm": 3.205721855163574,
      "learning_rate": 5.24170848121405e-06,
      "loss": 1.3393,
      "step": 293200
    },
    {
      "epoch": 8.9601026455673,
      "grad_norm": 2.415811777114868,
      "learning_rate": 5.2264104760739205e-06,
      "loss": 1.1942,
      "step": 293300
    },
    {
      "epoch": 8.963157573165516,
      "grad_norm": 12.766294479370117,
      "learning_rate": 5.211112470933791e-06,
      "loss": 1.0837,
      "step": 293400
    },
    {
      "epoch": 8.966212500763731,
      "grad_norm": 6.507505416870117,
      "learning_rate": 5.195814465793661e-06,
      "loss": 1.0663,
      "step": 293500
    },
    {
      "epoch": 8.969267428361947,
      "grad_norm": 1.3235666751861572,
      "learning_rate": 5.1805164606535314e-06,
      "loss": 0.9236,
      "step": 293600
    },
    {
      "epoch": 8.972322355960163,
      "grad_norm": 3.646758794784546,
      "learning_rate": 5.165218455513402e-06,
      "loss": 1.2117,
      "step": 293700
    },
    {
      "epoch": 8.97537728355838,
      "grad_norm": 5.10009241104126,
      "learning_rate": 5.149920450373272e-06,
      "loss": 1.0119,
      "step": 293800
    },
    {
      "epoch": 8.978432211156596,
      "grad_norm": 4.952513217926025,
      "learning_rate": 5.134622445233142e-06,
      "loss": 1.0933,
      "step": 293900
    },
    {
      "epoch": 8.981487138754812,
      "grad_norm": 3.1457278728485107,
      "learning_rate": 5.119324440093013e-06,
      "loss": 1.1274,
      "step": 294000
    },
    {
      "epoch": 8.984542066353027,
      "grad_norm": 10.905735969543457,
      "learning_rate": 5.104026434952883e-06,
      "loss": 1.1653,
      "step": 294100
    },
    {
      "epoch": 8.987596993951243,
      "grad_norm": 6.256532669067383,
      "learning_rate": 5.088728429812753e-06,
      "loss": 1.3448,
      "step": 294200
    },
    {
      "epoch": 8.990651921549459,
      "grad_norm": 2.832413673400879,
      "learning_rate": 5.073430424672623e-06,
      "loss": 1.0589,
      "step": 294300
    },
    {
      "epoch": 8.993706849147674,
      "grad_norm": 5.614567279815674,
      "learning_rate": 5.058132419532493e-06,
      "loss": 1.2088,
      "step": 294400
    },
    {
      "epoch": 8.996761776745892,
      "grad_norm": 2.566842555999756,
      "learning_rate": 5.042834414392363e-06,
      "loss": 1.171,
      "step": 294500
    },
    {
      "epoch": 8.999816704344108,
      "grad_norm": 3.9589226245880127,
      "learning_rate": 5.027536409252234e-06,
      "loss": 1.3417,
      "step": 294600
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.7450357426528992,
      "eval_loss": 0.6936118602752686,
      "eval_runtime": 1891.6602,
      "eval_samples_per_second": 17.304,
      "eval_steps_per_second": 4.326,
      "step": 294606
    },
    {
      "epoch": 9.002871631942323,
      "grad_norm": 3.6633384227752686,
      "learning_rate": 5.012238404112104e-06,
      "loss": 1.1321,
      "step": 294700
    },
    {
      "epoch": 9.005926559540539,
      "grad_norm": 3.5542895793914795,
      "learning_rate": 4.996940398971974e-06,
      "loss": 1.0508,
      "step": 294800
    },
    {
      "epoch": 9.008981487138755,
      "grad_norm": 14.556180000305176,
      "learning_rate": 4.981642393831845e-06,
      "loss": 0.9371,
      "step": 294900
    },
    {
      "epoch": 9.01203641473697,
      "grad_norm": 7.045968055725098,
      "learning_rate": 4.966344388691715e-06,
      "loss": 0.9754,
      "step": 295000
    },
    {
      "epoch": 9.015091342335186,
      "grad_norm": 9.45659351348877,
      "learning_rate": 4.951046383551585e-06,
      "loss": 1.1034,
      "step": 295100
    },
    {
      "epoch": 9.018146269933403,
      "grad_norm": 0.298971951007843,
      "learning_rate": 4.9357483784114555e-06,
      "loss": 1.2968,
      "step": 295200
    },
    {
      "epoch": 9.021201197531619,
      "grad_norm": 8.019341468811035,
      "learning_rate": 4.920450373271326e-06,
      "loss": 0.9928,
      "step": 295300
    },
    {
      "epoch": 9.024256125129835,
      "grad_norm": 7.284595966339111,
      "learning_rate": 4.905152368131196e-06,
      "loss": 1.3505,
      "step": 295400
    },
    {
      "epoch": 9.02731105272805,
      "grad_norm": 5.352906227111816,
      "learning_rate": 4.8900073430424675e-06,
      "loss": 0.9553,
      "step": 295500
    },
    {
      "epoch": 9.030365980326266,
      "grad_norm": 1.7522146701812744,
      "learning_rate": 4.874709337902338e-06,
      "loss": 1.1125,
      "step": 295600
    },
    {
      "epoch": 9.033420907924482,
      "grad_norm": 3.4803643226623535,
      "learning_rate": 4.859411332762208e-06,
      "loss": 1.3101,
      "step": 295700
    },
    {
      "epoch": 9.036475835522698,
      "grad_norm": 3.04394793510437,
      "learning_rate": 4.8441133276220784e-06,
      "loss": 1.0202,
      "step": 295800
    },
    {
      "epoch": 9.039530763120913,
      "grad_norm": 2.7401368618011475,
      "learning_rate": 4.828815322481949e-06,
      "loss": 0.9919,
      "step": 295900
    },
    {
      "epoch": 9.04258569071913,
      "grad_norm": 8.470032691955566,
      "learning_rate": 4.813517317341819e-06,
      "loss": 1.2767,
      "step": 296000
    },
    {
      "epoch": 9.045640618317346,
      "grad_norm": 4.937151908874512,
      "learning_rate": 4.798219312201689e-06,
      "loss": 1.078,
      "step": 296100
    },
    {
      "epoch": 9.048695545915562,
      "grad_norm": 5.262059211730957,
      "learning_rate": 4.78292130706156e-06,
      "loss": 1.1927,
      "step": 296200
    },
    {
      "epoch": 9.051750473513778,
      "grad_norm": 5.695133209228516,
      "learning_rate": 4.76762330192143e-06,
      "loss": 0.99,
      "step": 296300
    },
    {
      "epoch": 9.054805401111993,
      "grad_norm": 6.795228004455566,
      "learning_rate": 4.7523252967813e-06,
      "loss": 1.2354,
      "step": 296400
    },
    {
      "epoch": 9.057860328710209,
      "grad_norm": 3.628720760345459,
      "learning_rate": 4.737027291641171e-06,
      "loss": 1.2754,
      "step": 296500
    },
    {
      "epoch": 9.060915256308425,
      "grad_norm": 13.540319442749023,
      "learning_rate": 4.721729286501041e-06,
      "loss": 0.9649,
      "step": 296600
    },
    {
      "epoch": 9.063970183906642,
      "grad_norm": 6.967558860778809,
      "learning_rate": 4.706431281360911e-06,
      "loss": 1.1276,
      "step": 296700
    },
    {
      "epoch": 9.067025111504858,
      "grad_norm": 8.045135498046875,
      "learning_rate": 4.6911332762207815e-06,
      "loss": 1.2847,
      "step": 296800
    },
    {
      "epoch": 9.070080039103074,
      "grad_norm": 11.819404602050781,
      "learning_rate": 4.675835271080652e-06,
      "loss": 1.0835,
      "step": 296900
    },
    {
      "epoch": 9.07313496670129,
      "grad_norm": 4.425651550292969,
      "learning_rate": 4.660537265940522e-06,
      "loss": 1.1954,
      "step": 297000
    },
    {
      "epoch": 9.076189894299505,
      "grad_norm": 5.550767421722412,
      "learning_rate": 4.645239260800392e-06,
      "loss": 1.053,
      "step": 297100
    },
    {
      "epoch": 9.07924482189772,
      "grad_norm": 3.0203421115875244,
      "learning_rate": 4.629941255660262e-06,
      "loss": 1.034,
      "step": 297200
    },
    {
      "epoch": 9.082299749495936,
      "grad_norm": 17.99982261657715,
      "learning_rate": 4.614643250520132e-06,
      "loss": 1.1257,
      "step": 297300
    },
    {
      "epoch": 9.085354677094152,
      "grad_norm": 12.948797225952148,
      "learning_rate": 4.5993452453800026e-06,
      "loss": 1.0862,
      "step": 297400
    },
    {
      "epoch": 9.08840960469237,
      "grad_norm": 4.156719207763672,
      "learning_rate": 4.584047240239873e-06,
      "loss": 1.0115,
      "step": 297500
    },
    {
      "epoch": 9.091464532290585,
      "grad_norm": 7.684045791625977,
      "learning_rate": 4.568749235099743e-06,
      "loss": 1.1726,
      "step": 297600
    },
    {
      "epoch": 9.0945194598888,
      "grad_norm": 3.3296148777008057,
      "learning_rate": 4.5534512299596135e-06,
      "loss": 1.0634,
      "step": 297700
    },
    {
      "epoch": 9.097574387487017,
      "grad_norm": 18.09479331970215,
      "learning_rate": 4.538306204870886e-06,
      "loss": 1.1221,
      "step": 297800
    },
    {
      "epoch": 9.100629315085232,
      "grad_norm": 12.1429443359375,
      "learning_rate": 4.523008199730756e-06,
      "loss": 1.4943,
      "step": 297900
    },
    {
      "epoch": 9.103684242683448,
      "grad_norm": 9.859488487243652,
      "learning_rate": 4.5077101945906255e-06,
      "loss": 0.9249,
      "step": 298000
    },
    {
      "epoch": 9.106739170281664,
      "grad_norm": 3.7430107593536377,
      "learning_rate": 4.492412189450496e-06,
      "loss": 1.0232,
      "step": 298100
    },
    {
      "epoch": 9.109794097879881,
      "grad_norm": 4.004197120666504,
      "learning_rate": 4.477114184310366e-06,
      "loss": 1.4043,
      "step": 298200
    },
    {
      "epoch": 9.112849025478097,
      "grad_norm": 8.506260871887207,
      "learning_rate": 4.461816179170236e-06,
      "loss": 1.0303,
      "step": 298300
    },
    {
      "epoch": 9.115903953076312,
      "grad_norm": 6.506143093109131,
      "learning_rate": 4.446518174030107e-06,
      "loss": 1.2924,
      "step": 298400
    },
    {
      "epoch": 9.118958880674528,
      "grad_norm": 3.8566203117370605,
      "learning_rate": 4.431220168889977e-06,
      "loss": 1.1308,
      "step": 298500
    },
    {
      "epoch": 9.122013808272744,
      "grad_norm": 9.278569221496582,
      "learning_rate": 4.415922163749847e-06,
      "loss": 1.3331,
      "step": 298600
    },
    {
      "epoch": 9.12506873587096,
      "grad_norm": 3.4998819828033447,
      "learning_rate": 4.400624158609718e-06,
      "loss": 1.1288,
      "step": 298700
    },
    {
      "epoch": 9.128123663469175,
      "grad_norm": 5.379426956176758,
      "learning_rate": 4.385326153469588e-06,
      "loss": 1.1651,
      "step": 298800
    },
    {
      "epoch": 9.131178591067393,
      "grad_norm": 4.583761215209961,
      "learning_rate": 4.370028148329458e-06,
      "loss": 1.0494,
      "step": 298900
    },
    {
      "epoch": 9.134233518665608,
      "grad_norm": 6.520390033721924,
      "learning_rate": 4.3547301431893286e-06,
      "loss": 1.0415,
      "step": 299000
    },
    {
      "epoch": 9.137288446263824,
      "grad_norm": 11.750895500183105,
      "learning_rate": 4.339432138049199e-06,
      "loss": 0.9235,
      "step": 299100
    },
    {
      "epoch": 9.14034337386204,
      "grad_norm": 6.232056617736816,
      "learning_rate": 4.324134132909069e-06,
      "loss": 0.9507,
      "step": 299200
    },
    {
      "epoch": 9.143398301460255,
      "grad_norm": 4.731393337249756,
      "learning_rate": 4.3088361277689395e-06,
      "loss": 1.1672,
      "step": 299300
    },
    {
      "epoch": 9.146453229058471,
      "grad_norm": 6.675495624542236,
      "learning_rate": 4.29353812262881e-06,
      "loss": 1.4035,
      "step": 299400
    },
    {
      "epoch": 9.149508156656687,
      "grad_norm": 8.32022476196289,
      "learning_rate": 4.27824011748868e-06,
      "loss": 0.8868,
      "step": 299500
    },
    {
      "epoch": 9.152563084254902,
      "grad_norm": 10.285120964050293,
      "learning_rate": 4.26294211234855e-06,
      "loss": 1.1058,
      "step": 299600
    },
    {
      "epoch": 9.15561801185312,
      "grad_norm": 10.524968147277832,
      "learning_rate": 4.247644107208421e-06,
      "loss": 1.064,
      "step": 299700
    },
    {
      "epoch": 9.158672939451336,
      "grad_norm": 8.351820945739746,
      "learning_rate": 4.23234610206829e-06,
      "loss": 1.0773,
      "step": 299800
    },
    {
      "epoch": 9.161727867049551,
      "grad_norm": 8.429049491882324,
      "learning_rate": 4.2170480969281605e-06,
      "loss": 1.3182,
      "step": 299900
    },
    {
      "epoch": 9.164782794647767,
      "grad_norm": 8.29355525970459,
      "learning_rate": 4.201750091788031e-06,
      "loss": 1.1658,
      "step": 300000
    },
    {
      "epoch": 9.167837722245983,
      "grad_norm": 1.961021900177002,
      "learning_rate": 4.186452086647901e-06,
      "loss": 1.1509,
      "step": 300100
    },
    {
      "epoch": 9.170892649844198,
      "grad_norm": 7.757428169250488,
      "learning_rate": 4.1711540815077714e-06,
      "loss": 1.1293,
      "step": 300200
    },
    {
      "epoch": 9.173947577442414,
      "grad_norm": 3.276634693145752,
      "learning_rate": 4.155856076367642e-06,
      "loss": 1.2559,
      "step": 300300
    },
    {
      "epoch": 9.17700250504063,
      "grad_norm": 3.877615213394165,
      "learning_rate": 4.140558071227512e-06,
      "loss": 1.6137,
      "step": 300400
    },
    {
      "epoch": 9.180057432638847,
      "grad_norm": 17.469383239746094,
      "learning_rate": 4.125413046138784e-06,
      "loss": 1.1948,
      "step": 300500
    },
    {
      "epoch": 9.183112360237063,
      "grad_norm": 8.345824241638184,
      "learning_rate": 4.1101150409986546e-06,
      "loss": 1.2168,
      "step": 300600
    },
    {
      "epoch": 9.186167287835278,
      "grad_norm": 2.92280650138855,
      "learning_rate": 4.094817035858525e-06,
      "loss": 1.0463,
      "step": 300700
    },
    {
      "epoch": 9.189222215433494,
      "grad_norm": 1.5842794179916382,
      "learning_rate": 4.079519030718394e-06,
      "loss": 1.1139,
      "step": 300800
    },
    {
      "epoch": 9.19227714303171,
      "grad_norm": 6.409753322601318,
      "learning_rate": 4.064221025578265e-06,
      "loss": 1.2472,
      "step": 300900
    },
    {
      "epoch": 9.195332070629926,
      "grad_norm": 8.010771751403809,
      "learning_rate": 4.048923020438135e-06,
      "loss": 1.619,
      "step": 301000
    },
    {
      "epoch": 9.198386998228141,
      "grad_norm": 4.597021579742432,
      "learning_rate": 4.033625015298005e-06,
      "loss": 0.9576,
      "step": 301100
    },
    {
      "epoch": 9.201441925826359,
      "grad_norm": 10.441827774047852,
      "learning_rate": 4.0183270101578756e-06,
      "loss": 1.0894,
      "step": 301200
    },
    {
      "epoch": 9.204496853424574,
      "grad_norm": 3.910430431365967,
      "learning_rate": 4.003029005017746e-06,
      "loss": 1.1124,
      "step": 301300
    },
    {
      "epoch": 9.20755178102279,
      "grad_norm": 4.471078395843506,
      "learning_rate": 3.987730999877616e-06,
      "loss": 1.0833,
      "step": 301400
    },
    {
      "epoch": 9.210606708621006,
      "grad_norm": 12.565783500671387,
      "learning_rate": 3.9724329947374865e-06,
      "loss": 1.3408,
      "step": 301500
    },
    {
      "epoch": 9.213661636219221,
      "grad_norm": 4.613448619842529,
      "learning_rate": 3.957134989597357e-06,
      "loss": 1.0302,
      "step": 301600
    },
    {
      "epoch": 9.216716563817437,
      "grad_norm": 12.828263282775879,
      "learning_rate": 3.941836984457227e-06,
      "loss": 1.2364,
      "step": 301700
    },
    {
      "epoch": 9.219771491415653,
      "grad_norm": 17.703357696533203,
      "learning_rate": 3.9265389793170974e-06,
      "loss": 1.1193,
      "step": 301800
    },
    {
      "epoch": 9.22282641901387,
      "grad_norm": 4.079858303070068,
      "learning_rate": 3.911240974176968e-06,
      "loss": 1.165,
      "step": 301900
    },
    {
      "epoch": 9.225881346612086,
      "grad_norm": 5.560415744781494,
      "learning_rate": 3.895942969036838e-06,
      "loss": 1.4166,
      "step": 302000
    },
    {
      "epoch": 9.228936274210302,
      "grad_norm": 4.343051910400391,
      "learning_rate": 3.880644963896708e-06,
      "loss": 1.0804,
      "step": 302100
    },
    {
      "epoch": 9.231991201808517,
      "grad_norm": 11.356123924255371,
      "learning_rate": 3.865346958756579e-06,
      "loss": 0.9958,
      "step": 302200
    },
    {
      "epoch": 9.235046129406733,
      "grad_norm": 8.211952209472656,
      "learning_rate": 3.850048953616449e-06,
      "loss": 0.9514,
      "step": 302300
    },
    {
      "epoch": 9.238101057004949,
      "grad_norm": 6.394819736480713,
      "learning_rate": 3.834750948476319e-06,
      "loss": 0.9585,
      "step": 302400
    },
    {
      "epoch": 9.241155984603164,
      "grad_norm": 7.153532028198242,
      "learning_rate": 3.81945294333619e-06,
      "loss": 1.3609,
      "step": 302500
    },
    {
      "epoch": 9.24421091220138,
      "grad_norm": 3.2434120178222656,
      "learning_rate": 3.804154938196059e-06,
      "loss": 1.1877,
      "step": 302600
    },
    {
      "epoch": 9.247265839799597,
      "grad_norm": 9.158300399780273,
      "learning_rate": 3.7888569330559294e-06,
      "loss": 1.0076,
      "step": 302700
    },
    {
      "epoch": 9.250320767397813,
      "grad_norm": 7.51426362991333,
      "learning_rate": 3.7735589279157997e-06,
      "loss": 1.0217,
      "step": 302800
    },
    {
      "epoch": 9.253375694996029,
      "grad_norm": 3.0722098350524902,
      "learning_rate": 3.75826092277567e-06,
      "loss": 1.121,
      "step": 302900
    },
    {
      "epoch": 9.256430622594245,
      "grad_norm": 2.5164806842803955,
      "learning_rate": 3.7431158976869418e-06,
      "loss": 1.4275,
      "step": 303000
    },
    {
      "epoch": 9.25948555019246,
      "grad_norm": 8.31301212310791,
      "learning_rate": 3.727817892546812e-06,
      "loss": 1.0477,
      "step": 303100
    },
    {
      "epoch": 9.262540477790676,
      "grad_norm": 2.260798215866089,
      "learning_rate": 3.7125198874066824e-06,
      "loss": 1.1562,
      "step": 303200
    },
    {
      "epoch": 9.265595405388892,
      "grad_norm": 3.4062960147857666,
      "learning_rate": 3.6972218822665527e-06,
      "loss": 1.1084,
      "step": 303300
    },
    {
      "epoch": 9.268650332987109,
      "grad_norm": 6.769488334655762,
      "learning_rate": 3.681923877126423e-06,
      "loss": 1.0529,
      "step": 303400
    },
    {
      "epoch": 9.271705260585325,
      "grad_norm": 10.702885627746582,
      "learning_rate": 3.6666258719862933e-06,
      "loss": 1.2395,
      "step": 303500
    },
    {
      "epoch": 9.27476018818354,
      "grad_norm": 5.437123775482178,
      "learning_rate": 3.651327866846163e-06,
      "loss": 1.0823,
      "step": 303600
    },
    {
      "epoch": 9.277815115781756,
      "grad_norm": 8.948333740234375,
      "learning_rate": 3.6360298617060335e-06,
      "loss": 1.1325,
      "step": 303700
    },
    {
      "epoch": 9.280870043379972,
      "grad_norm": 6.607554912567139,
      "learning_rate": 3.620731856565904e-06,
      "loss": 1.3233,
      "step": 303800
    },
    {
      "epoch": 9.283924970978187,
      "grad_norm": 7.056314468383789,
      "learning_rate": 3.605433851425774e-06,
      "loss": 1.1637,
      "step": 303900
    },
    {
      "epoch": 9.286979898576403,
      "grad_norm": 3.6274731159210205,
      "learning_rate": 3.5901358462856444e-06,
      "loss": 1.2798,
      "step": 304000
    },
    {
      "epoch": 9.290034826174619,
      "grad_norm": 3.242490768432617,
      "learning_rate": 3.5748378411455148e-06,
      "loss": 1.1217,
      "step": 304100
    },
    {
      "epoch": 9.293089753772836,
      "grad_norm": 25.473543167114258,
      "learning_rate": 3.559539836005385e-06,
      "loss": 1.0237,
      "step": 304200
    },
    {
      "epoch": 9.296144681371052,
      "grad_norm": 6.082782745361328,
      "learning_rate": 3.5442418308652554e-06,
      "loss": 1.3431,
      "step": 304300
    },
    {
      "epoch": 9.299199608969268,
      "grad_norm": 6.279635429382324,
      "learning_rate": 3.5289438257251257e-06,
      "loss": 1.1487,
      "step": 304400
    },
    {
      "epoch": 9.302254536567483,
      "grad_norm": 6.003155708312988,
      "learning_rate": 3.513645820584996e-06,
      "loss": 1.1343,
      "step": 304500
    },
    {
      "epoch": 9.305309464165699,
      "grad_norm": 4.19450044631958,
      "learning_rate": 3.4983478154448663e-06,
      "loss": 1.1376,
      "step": 304600
    },
    {
      "epoch": 9.308364391763915,
      "grad_norm": 3.5574228763580322,
      "learning_rate": 3.4830498103047366e-06,
      "loss": 1.089,
      "step": 304700
    },
    {
      "epoch": 9.31141931936213,
      "grad_norm": 8.025125503540039,
      "learning_rate": 3.467751805164607e-06,
      "loss": 1.0255,
      "step": 304800
    },
    {
      "epoch": 9.314474246960348,
      "grad_norm": 3.3775525093078613,
      "learning_rate": 3.4524538000244772e-06,
      "loss": 1.0984,
      "step": 304900
    },
    {
      "epoch": 9.317529174558564,
      "grad_norm": 4.80861234664917,
      "learning_rate": 3.4371557948843475e-06,
      "loss": 1.1859,
      "step": 305000
    },
    {
      "epoch": 9.32058410215678,
      "grad_norm": 16.27497673034668,
      "learning_rate": 3.421857789744218e-06,
      "loss": 1.3898,
      "step": 305100
    },
    {
      "epoch": 9.323639029754995,
      "grad_norm": 21.027254104614258,
      "learning_rate": 3.406559784604088e-06,
      "loss": 1.0316,
      "step": 305200
    },
    {
      "epoch": 9.32669395735321,
      "grad_norm": 4.368397235870361,
      "learning_rate": 3.3912617794639585e-06,
      "loss": 1.0431,
      "step": 305300
    },
    {
      "epoch": 9.329748884951426,
      "grad_norm": 3.5449934005737305,
      "learning_rate": 3.375963774323828e-06,
      "loss": 1.0531,
      "step": 305400
    },
    {
      "epoch": 9.332803812549642,
      "grad_norm": 5.919412612915039,
      "learning_rate": 3.3606657691836982e-06,
      "loss": 1.1873,
      "step": 305500
    },
    {
      "epoch": 9.335858740147858,
      "grad_norm": 2.567575693130493,
      "learning_rate": 3.3455207440949705e-06,
      "loss": 1.085,
      "step": 305600
    },
    {
      "epoch": 9.338913667746075,
      "grad_norm": 12.758691787719727,
      "learning_rate": 3.3302227389548408e-06,
      "loss": 1.0725,
      "step": 305700
    },
    {
      "epoch": 9.34196859534429,
      "grad_norm": 3.886571168899536,
      "learning_rate": 3.314924733814711e-06,
      "loss": 1.1712,
      "step": 305800
    },
    {
      "epoch": 9.345023522942506,
      "grad_norm": 8.74039077758789,
      "learning_rate": 3.2996267286745814e-06,
      "loss": 1.5482,
      "step": 305900
    },
    {
      "epoch": 9.348078450540722,
      "grad_norm": 5.288308143615723,
      "learning_rate": 3.2843287235344517e-06,
      "loss": 1.2712,
      "step": 306000
    },
    {
      "epoch": 9.351133378138938,
      "grad_norm": 9.838608741760254,
      "learning_rate": 3.269030718394322e-06,
      "loss": 1.143,
      "step": 306100
    },
    {
      "epoch": 9.354188305737154,
      "grad_norm": 5.224956035614014,
      "learning_rate": 3.2537327132541923e-06,
      "loss": 1.01,
      "step": 306200
    },
    {
      "epoch": 9.35724323333537,
      "grad_norm": 5.820688724517822,
      "learning_rate": 3.2384347081140618e-06,
      "loss": 1.1268,
      "step": 306300
    },
    {
      "epoch": 9.360298160933587,
      "grad_norm": 2.071725368499756,
      "learning_rate": 3.223136702973932e-06,
      "loss": 0.8517,
      "step": 306400
    },
    {
      "epoch": 9.363353088531802,
      "grad_norm": 17.318157196044922,
      "learning_rate": 3.2078386978338024e-06,
      "loss": 1.4184,
      "step": 306500
    },
    {
      "epoch": 9.366408016130018,
      "grad_norm": 8.712560653686523,
      "learning_rate": 3.1925406926936727e-06,
      "loss": 1.0402,
      "step": 306600
    },
    {
      "epoch": 9.369462943728234,
      "grad_norm": 9.98942756652832,
      "learning_rate": 3.177242687553543e-06,
      "loss": 1.0045,
      "step": 306700
    },
    {
      "epoch": 9.37251787132645,
      "grad_norm": 42.62532043457031,
      "learning_rate": 3.1619446824134133e-06,
      "loss": 1.188,
      "step": 306800
    },
    {
      "epoch": 9.375572798924665,
      "grad_norm": 7.075395584106445,
      "learning_rate": 3.1466466772732836e-06,
      "loss": 1.1637,
      "step": 306900
    },
    {
      "epoch": 9.37862772652288,
      "grad_norm": 3.448315382003784,
      "learning_rate": 3.131348672133154e-06,
      "loss": 0.9102,
      "step": 307000
    },
    {
      "epoch": 9.381682654121096,
      "grad_norm": 10.89814281463623,
      "learning_rate": 3.1160506669930243e-06,
      "loss": 1.0922,
      "step": 307100
    },
    {
      "epoch": 9.384737581719314,
      "grad_norm": 25.21414566040039,
      "learning_rate": 3.1007526618528946e-06,
      "loss": 1.1627,
      "step": 307200
    },
    {
      "epoch": 9.38779250931753,
      "grad_norm": 9.04411506652832,
      "learning_rate": 3.085454656712765e-06,
      "loss": 1.1222,
      "step": 307300
    },
    {
      "epoch": 9.390847436915745,
      "grad_norm": 7.841292381286621,
      "learning_rate": 3.070156651572635e-06,
      "loss": 0.9225,
      "step": 307400
    },
    {
      "epoch": 9.393902364513961,
      "grad_norm": 9.054701805114746,
      "learning_rate": 3.0548586464325055e-06,
      "loss": 1.1585,
      "step": 307500
    },
    {
      "epoch": 9.396957292112177,
      "grad_norm": 4.471860408782959,
      "learning_rate": 3.039560641292376e-06,
      "loss": 1.3753,
      "step": 307600
    },
    {
      "epoch": 9.400012219710392,
      "grad_norm": 4.296054363250732,
      "learning_rate": 3.0242626361522457e-06,
      "loss": 1.1125,
      "step": 307700
    },
    {
      "epoch": 9.403067147308608,
      "grad_norm": 4.867402076721191,
      "learning_rate": 3.008964631012116e-06,
      "loss": 0.8846,
      "step": 307800
    },
    {
      "epoch": 9.406122074906826,
      "grad_norm": 5.357305526733398,
      "learning_rate": 2.9938196059233878e-06,
      "loss": 1.1692,
      "step": 307900
    },
    {
      "epoch": 9.409177002505041,
      "grad_norm": 5.116623878479004,
      "learning_rate": 2.978521600783258e-06,
      "loss": 0.8862,
      "step": 308000
    },
    {
      "epoch": 9.412231930103257,
      "grad_norm": 2.731386661529541,
      "learning_rate": 2.9632235956431284e-06,
      "loss": 0.8937,
      "step": 308100
    },
    {
      "epoch": 9.415286857701473,
      "grad_norm": 4.1633501052856445,
      "learning_rate": 2.9479255905029987e-06,
      "loss": 1.3928,
      "step": 308200
    },
    {
      "epoch": 9.418341785299688,
      "grad_norm": 6.098063945770264,
      "learning_rate": 2.932627585362869e-06,
      "loss": 1.1213,
      "step": 308300
    },
    {
      "epoch": 9.421396712897904,
      "grad_norm": 9.61737060546875,
      "learning_rate": 2.9173295802227393e-06,
      "loss": 1.0766,
      "step": 308400
    },
    {
      "epoch": 9.42445164049612,
      "grad_norm": 6.344211578369141,
      "learning_rate": 2.9020315750826096e-06,
      "loss": 1.0052,
      "step": 308500
    },
    {
      "epoch": 9.427506568094337,
      "grad_norm": 5.4332170486450195,
      "learning_rate": 2.8867335699424795e-06,
      "loss": 1.1181,
      "step": 308600
    },
    {
      "epoch": 9.430561495692553,
      "grad_norm": 3.6471481323242188,
      "learning_rate": 2.87143556480235e-06,
      "loss": 1.3035,
      "step": 308700
    },
    {
      "epoch": 9.433616423290768,
      "grad_norm": 7.321690082550049,
      "learning_rate": 2.85613755966222e-06,
      "loss": 1.0789,
      "step": 308800
    },
    {
      "epoch": 9.436671350888984,
      "grad_norm": 6.397017955780029,
      "learning_rate": 2.8408395545220905e-06,
      "loss": 1.3215,
      "step": 308900
    },
    {
      "epoch": 9.4397262784872,
      "grad_norm": 7.486538887023926,
      "learning_rate": 2.8255415493819608e-06,
      "loss": 1.0702,
      "step": 309000
    },
    {
      "epoch": 9.442781206085415,
      "grad_norm": 13.288386344909668,
      "learning_rate": 2.810243544241831e-06,
      "loss": 1.0518,
      "step": 309100
    },
    {
      "epoch": 9.445836133683631,
      "grad_norm": 4.601072788238525,
      "learning_rate": 2.7949455391017014e-06,
      "loss": 1.1671,
      "step": 309200
    },
    {
      "epoch": 9.448891061281847,
      "grad_norm": 10.149903297424316,
      "learning_rate": 2.7796475339615717e-06,
      "loss": 1.16,
      "step": 309300
    },
    {
      "epoch": 9.451945988880064,
      "grad_norm": 6.56465482711792,
      "learning_rate": 2.764349528821442e-06,
      "loss": 0.9742,
      "step": 309400
    },
    {
      "epoch": 9.45500091647828,
      "grad_norm": 11.361266136169434,
      "learning_rate": 2.7490515236813123e-06,
      "loss": 1.1496,
      "step": 309500
    },
    {
      "epoch": 9.458055844076496,
      "grad_norm": 6.141317844390869,
      "learning_rate": 2.733753518541182e-06,
      "loss": 1.4959,
      "step": 309600
    },
    {
      "epoch": 9.461110771674711,
      "grad_norm": 10.651205062866211,
      "learning_rate": 2.7184555134010525e-06,
      "loss": 1.0295,
      "step": 309700
    },
    {
      "epoch": 9.464165699272927,
      "grad_norm": 4.727423191070557,
      "learning_rate": 2.703157508260923e-06,
      "loss": 1.2795,
      "step": 309800
    },
    {
      "epoch": 9.467220626871143,
      "grad_norm": 10.940389633178711,
      "learning_rate": 2.687859503120793e-06,
      "loss": 1.1406,
      "step": 309900
    },
    {
      "epoch": 9.470275554469358,
      "grad_norm": 3.7946255207061768,
      "learning_rate": 2.6725614979806634e-06,
      "loss": 1.1332,
      "step": 310000
    },
    {
      "epoch": 9.473330482067574,
      "grad_norm": 7.23513650894165,
      "learning_rate": 2.6572634928405337e-06,
      "loss": 0.9592,
      "step": 310100
    },
    {
      "epoch": 9.476385409665792,
      "grad_norm": 7.274559020996094,
      "learning_rate": 2.641965487700404e-06,
      "loss": 1.1219,
      "step": 310200
    },
    {
      "epoch": 9.479440337264007,
      "grad_norm": 4.508282661437988,
      "learning_rate": 2.6266674825602744e-06,
      "loss": 1.164,
      "step": 310300
    },
    {
      "epoch": 9.482495264862223,
      "grad_norm": 18.602556228637695,
      "learning_rate": 2.6113694774201447e-06,
      "loss": 1.0263,
      "step": 310400
    },
    {
      "epoch": 9.485550192460439,
      "grad_norm": 3.0879483222961426,
      "learning_rate": 2.5960714722800146e-06,
      "loss": 1.0635,
      "step": 310500
    },
    {
      "epoch": 9.488605120058654,
      "grad_norm": 12.606521606445312,
      "learning_rate": 2.580773467139885e-06,
      "loss": 1.1495,
      "step": 310600
    },
    {
      "epoch": 9.49166004765687,
      "grad_norm": 17.32771110534668,
      "learning_rate": 2.565475461999755e-06,
      "loss": 1.5242,
      "step": 310700
    },
    {
      "epoch": 9.494714975255086,
      "grad_norm": 14.84314250946045,
      "learning_rate": 2.550330436911027e-06,
      "loss": 1.0958,
      "step": 310800
    },
    {
      "epoch": 9.497769902853303,
      "grad_norm": 21.077354431152344,
      "learning_rate": 2.5350324317708973e-06,
      "loss": 1.1022,
      "step": 310900
    },
    {
      "epoch": 9.500824830451519,
      "grad_norm": 6.4609246253967285,
      "learning_rate": 2.5197344266307676e-06,
      "loss": 1.3823,
      "step": 311000
    },
    {
      "epoch": 9.503879758049735,
      "grad_norm": 3.544327974319458,
      "learning_rate": 2.504436421490638e-06,
      "loss": 1.1492,
      "step": 311100
    },
    {
      "epoch": 9.50693468564795,
      "grad_norm": 14.659440994262695,
      "learning_rate": 2.489138416350508e-06,
      "loss": 1.0902,
      "step": 311200
    },
    {
      "epoch": 9.509989613246166,
      "grad_norm": 6.885101318359375,
      "learning_rate": 2.4738404112103785e-06,
      "loss": 1.0379,
      "step": 311300
    },
    {
      "epoch": 9.513044540844382,
      "grad_norm": 19.84307098388672,
      "learning_rate": 2.4585424060702484e-06,
      "loss": 1.0196,
      "step": 311400
    },
    {
      "epoch": 9.516099468442597,
      "grad_norm": 15.895182609558105,
      "learning_rate": 2.4432444009301187e-06,
      "loss": 1.2111,
      "step": 311500
    },
    {
      "epoch": 9.519154396040815,
      "grad_norm": 9.044354438781738,
      "learning_rate": 2.427946395789989e-06,
      "loss": 1.1352,
      "step": 311600
    },
    {
      "epoch": 9.52220932363903,
      "grad_norm": 5.041273593902588,
      "learning_rate": 2.4126483906498593e-06,
      "loss": 1.1897,
      "step": 311700
    },
    {
      "epoch": 9.525264251237246,
      "grad_norm": 6.682356357574463,
      "learning_rate": 2.3973503855097296e-06,
      "loss": 1.2135,
      "step": 311800
    },
    {
      "epoch": 9.528319178835462,
      "grad_norm": 10.315226554870605,
      "learning_rate": 2.3820523803696e-06,
      "loss": 1.0691,
      "step": 311900
    },
    {
      "epoch": 9.531374106433677,
      "grad_norm": 6.702560901641846,
      "learning_rate": 2.3667543752294703e-06,
      "loss": 1.0155,
      "step": 312000
    },
    {
      "epoch": 9.534429034031893,
      "grad_norm": 3.5035879611968994,
      "learning_rate": 2.3514563700893406e-06,
      "loss": 1.126,
      "step": 312100
    },
    {
      "epoch": 9.537483961630109,
      "grad_norm": 9.319344520568848,
      "learning_rate": 2.336158364949211e-06,
      "loss": 1.0382,
      "step": 312200
    },
    {
      "epoch": 9.540538889228324,
      "grad_norm": 8.65913200378418,
      "learning_rate": 2.3208603598090808e-06,
      "loss": 0.9698,
      "step": 312300
    },
    {
      "epoch": 9.543593816826542,
      "grad_norm": 11.004213333129883,
      "learning_rate": 2.305562354668951e-06,
      "loss": 1.2223,
      "step": 312400
    },
    {
      "epoch": 9.546648744424758,
      "grad_norm": 6.043313503265381,
      "learning_rate": 2.2902643495288214e-06,
      "loss": 0.9833,
      "step": 312500
    },
    {
      "epoch": 9.549703672022973,
      "grad_norm": 2.2580814361572266,
      "learning_rate": 2.2749663443886917e-06,
      "loss": 1.1875,
      "step": 312600
    },
    {
      "epoch": 9.552758599621189,
      "grad_norm": 6.7491278648376465,
      "learning_rate": 2.259668339248562e-06,
      "loss": 1.0376,
      "step": 312700
    },
    {
      "epoch": 9.555813527219405,
      "grad_norm": 3.858389377593994,
      "learning_rate": 2.2443703341084323e-06,
      "loss": 1.4318,
      "step": 312800
    },
    {
      "epoch": 9.55886845481762,
      "grad_norm": 4.029982089996338,
      "learning_rate": 2.2290723289683026e-06,
      "loss": 0.9203,
      "step": 312900
    },
    {
      "epoch": 9.561923382415836,
      "grad_norm": 10.59288215637207,
      "learning_rate": 2.213774323828173e-06,
      "loss": 1.0497,
      "step": 313000
    },
    {
      "epoch": 9.564978310014052,
      "grad_norm": 6.483738422393799,
      "learning_rate": 2.1984763186880432e-06,
      "loss": 1.3017,
      "step": 313100
    },
    {
      "epoch": 9.56803323761227,
      "grad_norm": 8.11201286315918,
      "learning_rate": 2.183178313547913e-06,
      "loss": 1.0831,
      "step": 313200
    },
    {
      "epoch": 9.571088165210485,
      "grad_norm": 6.691319465637207,
      "learning_rate": 2.168033288459185e-06,
      "loss": 1.1194,
      "step": 313300
    },
    {
      "epoch": 9.5741430928087,
      "grad_norm": 17.671133041381836,
      "learning_rate": 2.1527352833190552e-06,
      "loss": 1.0706,
      "step": 313400
    },
    {
      "epoch": 9.577198020406916,
      "grad_norm": 4.4608869552612305,
      "learning_rate": 2.1374372781789255e-06,
      "loss": 1.1263,
      "step": 313500
    },
    {
      "epoch": 9.580252948005132,
      "grad_norm": 5.7190446853637695,
      "learning_rate": 2.122139273038796e-06,
      "loss": 1.0572,
      "step": 313600
    },
    {
      "epoch": 9.583307875603348,
      "grad_norm": 13.810622215270996,
      "learning_rate": 2.106841267898666e-06,
      "loss": 1.1233,
      "step": 313700
    },
    {
      "epoch": 9.586362803201563,
      "grad_norm": 3.3503191471099854,
      "learning_rate": 2.0915432627585365e-06,
      "loss": 1.2015,
      "step": 313800
    },
    {
      "epoch": 9.58941773079978,
      "grad_norm": 19.035480499267578,
      "learning_rate": 2.0762452576184068e-06,
      "loss": 1.5642,
      "step": 313900
    },
    {
      "epoch": 9.592472658397996,
      "grad_norm": 3.9842231273651123,
      "learning_rate": 2.060947252478277e-06,
      "loss": 0.9717,
      "step": 314000
    },
    {
      "epoch": 9.595527585996212,
      "grad_norm": 1.611772894859314,
      "learning_rate": 2.0456492473381474e-06,
      "loss": 0.9999,
      "step": 314100
    },
    {
      "epoch": 9.598582513594428,
      "grad_norm": 8.103382110595703,
      "learning_rate": 2.0303512421980173e-06,
      "loss": 1.092,
      "step": 314200
    },
    {
      "epoch": 9.601637441192644,
      "grad_norm": 3.4380149841308594,
      "learning_rate": 2.0150532370578876e-06,
      "loss": 1.1504,
      "step": 314300
    },
    {
      "epoch": 9.60469236879086,
      "grad_norm": 11.057697296142578,
      "learning_rate": 1.999755231917758e-06,
      "loss": 1.1114,
      "step": 314400
    },
    {
      "epoch": 9.607747296389075,
      "grad_norm": 3.9668221473693848,
      "learning_rate": 1.984457226777628e-06,
      "loss": 1.0464,
      "step": 314500
    },
    {
      "epoch": 9.610802223987292,
      "grad_norm": 3.590650796890259,
      "learning_rate": 1.9691592216374985e-06,
      "loss": 1.231,
      "step": 314600
    },
    {
      "epoch": 9.613857151585508,
      "grad_norm": 6.6729817390441895,
      "learning_rate": 1.953861216497369e-06,
      "loss": 1.1776,
      "step": 314700
    },
    {
      "epoch": 9.616912079183724,
      "grad_norm": 13.663302421569824,
      "learning_rate": 1.938563211357239e-06,
      "loss": 1.1555,
      "step": 314800
    },
    {
      "epoch": 9.61996700678194,
      "grad_norm": 16.635059356689453,
      "learning_rate": 1.9232652062171094e-06,
      "loss": 1.199,
      "step": 314900
    },
    {
      "epoch": 9.623021934380155,
      "grad_norm": 5.7954230308532715,
      "learning_rate": 1.9079672010769797e-06,
      "loss": 1.1143,
      "step": 315000
    },
    {
      "epoch": 9.62607686197837,
      "grad_norm": 4.609525203704834,
      "learning_rate": 1.8926691959368498e-06,
      "loss": 1.0792,
      "step": 315100
    },
    {
      "epoch": 9.629131789576586,
      "grad_norm": 3.5899178981781006,
      "learning_rate": 1.8773711907967202e-06,
      "loss": 1.2464,
      "step": 315200
    },
    {
      "epoch": 9.632186717174802,
      "grad_norm": 4.332881927490234,
      "learning_rate": 1.8620731856565905e-06,
      "loss": 1.0524,
      "step": 315300
    },
    {
      "epoch": 9.63524164477302,
      "grad_norm": 11.647631645202637,
      "learning_rate": 1.8467751805164608e-06,
      "loss": 1.1577,
      "step": 315400
    },
    {
      "epoch": 9.638296572371235,
      "grad_norm": 4.526301383972168,
      "learning_rate": 1.831477175376331e-06,
      "loss": 1.3823,
      "step": 315500
    },
    {
      "epoch": 9.641351499969451,
      "grad_norm": 6.170322895050049,
      "learning_rate": 1.8161791702362014e-06,
      "loss": 1.2213,
      "step": 315600
    },
    {
      "epoch": 9.644406427567667,
      "grad_norm": 16.08941650390625,
      "learning_rate": 1.8008811650960717e-06,
      "loss": 1.2398,
      "step": 315700
    },
    {
      "epoch": 9.647461355165882,
      "grad_norm": 13.574227333068848,
      "learning_rate": 1.785583159955942e-06,
      "loss": 1.1134,
      "step": 315800
    },
    {
      "epoch": 9.650516282764098,
      "grad_norm": 17.764299392700195,
      "learning_rate": 1.7702851548158123e-06,
      "loss": 1.2985,
      "step": 315900
    },
    {
      "epoch": 9.653571210362314,
      "grad_norm": 1.291345477104187,
      "learning_rate": 1.7549871496756822e-06,
      "loss": 1.1614,
      "step": 316000
    },
    {
      "epoch": 9.656626137960531,
      "grad_norm": 11.866973876953125,
      "learning_rate": 1.7396891445355525e-06,
      "loss": 0.9506,
      "step": 316100
    },
    {
      "epoch": 9.659681065558747,
      "grad_norm": 5.622418403625488,
      "learning_rate": 1.7243911393954228e-06,
      "loss": 1.0569,
      "step": 316200
    },
    {
      "epoch": 9.662735993156963,
      "grad_norm": 24.80992317199707,
      "learning_rate": 1.7090931342552931e-06,
      "loss": 1.3718,
      "step": 316300
    },
    {
      "epoch": 9.665790920755178,
      "grad_norm": 1.4948173761367798,
      "learning_rate": 1.6937951291151634e-06,
      "loss": 1.2264,
      "step": 316400
    },
    {
      "epoch": 9.668845848353394,
      "grad_norm": 0.5281773805618286,
      "learning_rate": 1.6784971239750338e-06,
      "loss": 1.0053,
      "step": 316500
    },
    {
      "epoch": 9.67190077595161,
      "grad_norm": 0.00047525722766295075,
      "learning_rate": 1.663199118834904e-06,
      "loss": 0.9378,
      "step": 316600
    },
    {
      "epoch": 9.674955703549825,
      "grad_norm": 9.675890922546387,
      "learning_rate": 1.6479011136947744e-06,
      "loss": 1.0745,
      "step": 316700
    },
    {
      "epoch": 9.678010631148041,
      "grad_norm": 9.780930519104004,
      "learning_rate": 1.6326031085546447e-06,
      "loss": 1.0422,
      "step": 316800
    },
    {
      "epoch": 9.681065558746258,
      "grad_norm": 10.785592079162598,
      "learning_rate": 1.6173051034145146e-06,
      "loss": 1.0482,
      "step": 316900
    },
    {
      "epoch": 9.684120486344474,
      "grad_norm": 12.213613510131836,
      "learning_rate": 1.6020070982743849e-06,
      "loss": 1.1848,
      "step": 317000
    },
    {
      "epoch": 9.68717541394269,
      "grad_norm": 3.4983580112457275,
      "learning_rate": 1.5867090931342552e-06,
      "loss": 1.0039,
      "step": 317100
    },
    {
      "epoch": 9.690230341540905,
      "grad_norm": 3.852393865585327,
      "learning_rate": 1.5714110879941255e-06,
      "loss": 1.2497,
      "step": 317200
    },
    {
      "epoch": 9.693285269139121,
      "grad_norm": 6.902023792266846,
      "learning_rate": 1.5562660629053973e-06,
      "loss": 1.1257,
      "step": 317300
    },
    {
      "epoch": 9.696340196737337,
      "grad_norm": 19.849651336669922,
      "learning_rate": 1.5409680577652676e-06,
      "loss": 1.1279,
      "step": 317400
    },
    {
      "epoch": 9.699395124335553,
      "grad_norm": 2.910151720046997,
      "learning_rate": 1.525670052625138e-06,
      "loss": 1.6378,
      "step": 317500
    },
    {
      "epoch": 9.70245005193377,
      "grad_norm": 5.355201244354248,
      "learning_rate": 1.510372047485008e-06,
      "loss": 0.9853,
      "step": 317600
    },
    {
      "epoch": 9.705504979531986,
      "grad_norm": 5.7815775871276855,
      "learning_rate": 1.4950740423448783e-06,
      "loss": 1.1443,
      "step": 317700
    },
    {
      "epoch": 9.708559907130201,
      "grad_norm": 10.988548278808594,
      "learning_rate": 1.4797760372047486e-06,
      "loss": 1.0067,
      "step": 317800
    },
    {
      "epoch": 9.711614834728417,
      "grad_norm": 7.885775089263916,
      "learning_rate": 1.464478032064619e-06,
      "loss": 1.126,
      "step": 317900
    },
    {
      "epoch": 9.714669762326633,
      "grad_norm": 5.713939666748047,
      "learning_rate": 1.4491800269244892e-06,
      "loss": 1.1281,
      "step": 318000
    },
    {
      "epoch": 9.717724689924848,
      "grad_norm": 39.32889175415039,
      "learning_rate": 1.4338820217843593e-06,
      "loss": 1.2703,
      "step": 318100
    },
    {
      "epoch": 9.720779617523064,
      "grad_norm": 10.617399215698242,
      "learning_rate": 1.4185840166442296e-06,
      "loss": 1.0991,
      "step": 318200
    },
    {
      "epoch": 9.723834545121282,
      "grad_norm": 2.650892734527588,
      "learning_rate": 1.4032860115041e-06,
      "loss": 1.1306,
      "step": 318300
    },
    {
      "epoch": 9.726889472719497,
      "grad_norm": 5.076470375061035,
      "learning_rate": 1.3879880063639703e-06,
      "loss": 1.4098,
      "step": 318400
    },
    {
      "epoch": 9.729944400317713,
      "grad_norm": 7.810786724090576,
      "learning_rate": 1.3728429812752418e-06,
      "loss": 1.4563,
      "step": 318500
    },
    {
      "epoch": 9.732999327915929,
      "grad_norm": 30.487869262695312,
      "learning_rate": 1.357544976135112e-06,
      "loss": 1.1686,
      "step": 318600
    },
    {
      "epoch": 9.736054255514144,
      "grad_norm": 15.498873710632324,
      "learning_rate": 1.3422469709949822e-06,
      "loss": 1.0622,
      "step": 318700
    },
    {
      "epoch": 9.73910918311236,
      "grad_norm": 12.014664649963379,
      "learning_rate": 1.3269489658548526e-06,
      "loss": 1.0155,
      "step": 318800
    },
    {
      "epoch": 9.742164110710576,
      "grad_norm": 3.6885623931884766,
      "learning_rate": 1.3116509607147229e-06,
      "loss": 1.0831,
      "step": 318900
    },
    {
      "epoch": 9.745219038308791,
      "grad_norm": 11.527379035949707,
      "learning_rate": 1.2963529555745932e-06,
      "loss": 1.2619,
      "step": 319000
    },
    {
      "epoch": 9.748273965907009,
      "grad_norm": 3.429348945617676,
      "learning_rate": 1.2810549504344633e-06,
      "loss": 1.1128,
      "step": 319100
    },
    {
      "epoch": 9.751328893505224,
      "grad_norm": 5.2909064292907715,
      "learning_rate": 1.2657569452943336e-06,
      "loss": 1.2422,
      "step": 319200
    },
    {
      "epoch": 9.75438382110344,
      "grad_norm": 11.643526077270508,
      "learning_rate": 1.2504589401542039e-06,
      "loss": 1.235,
      "step": 319300
    },
    {
      "epoch": 9.757438748701656,
      "grad_norm": 8.330674171447754,
      "learning_rate": 1.2351609350140742e-06,
      "loss": 1.0515,
      "step": 319400
    },
    {
      "epoch": 9.760493676299872,
      "grad_norm": 6.047513484954834,
      "learning_rate": 1.2198629298739445e-06,
      "loss": 1.1808,
      "step": 319500
    },
    {
      "epoch": 9.763548603898087,
      "grad_norm": 3.790055990219116,
      "learning_rate": 1.2045649247338148e-06,
      "loss": 1.1685,
      "step": 319600
    },
    {
      "epoch": 9.766603531496303,
      "grad_norm": 11.13185977935791,
      "learning_rate": 1.1892669195936851e-06,
      "loss": 1.0524,
      "step": 319700
    },
    {
      "epoch": 9.769658459094519,
      "grad_norm": 3.2141988277435303,
      "learning_rate": 1.1739689144535554e-06,
      "loss": 1.2405,
      "step": 319800
    },
    {
      "epoch": 9.772713386692736,
      "grad_norm": 3.883542060852051,
      "learning_rate": 1.1586709093134257e-06,
      "loss": 1.1421,
      "step": 319900
    },
    {
      "epoch": 9.775768314290952,
      "grad_norm": 3.9310715198516846,
      "learning_rate": 1.1433729041732958e-06,
      "loss": 0.9411,
      "step": 320000
    },
    {
      "epoch": 9.778823241889167,
      "grad_norm": 4.970428943634033,
      "learning_rate": 1.1280748990331662e-06,
      "loss": 1.0685,
      "step": 320100
    },
    {
      "epoch": 9.781878169487383,
      "grad_norm": 9.02895736694336,
      "learning_rate": 1.1127768938930365e-06,
      "loss": 1.5052,
      "step": 320200
    },
    {
      "epoch": 9.784933097085599,
      "grad_norm": 7.479128837585449,
      "learning_rate": 1.0974788887529068e-06,
      "loss": 1.251,
      "step": 320300
    },
    {
      "epoch": 9.787988024683814,
      "grad_norm": 9.539288520812988,
      "learning_rate": 1.0821808836127769e-06,
      "loss": 1.0766,
      "step": 320400
    },
    {
      "epoch": 9.79104295228203,
      "grad_norm": 4.297487258911133,
      "learning_rate": 1.0668828784726472e-06,
      "loss": 1.1509,
      "step": 320500
    },
    {
      "epoch": 9.794097879880248,
      "grad_norm": 8.104453086853027,
      "learning_rate": 1.0515848733325175e-06,
      "loss": 1.1988,
      "step": 320600
    },
    {
      "epoch": 9.797152807478463,
      "grad_norm": 5.961874485015869,
      "learning_rate": 1.0362868681923878e-06,
      "loss": 1.2145,
      "step": 320700
    },
    {
      "epoch": 9.800207735076679,
      "grad_norm": 13.597023010253906,
      "learning_rate": 1.0209888630522581e-06,
      "loss": 1.0281,
      "step": 320800
    },
    {
      "epoch": 9.803262662674895,
      "grad_norm": 5.224933624267578,
      "learning_rate": 1.0056908579121282e-06,
      "loss": 1.0692,
      "step": 320900
    },
    {
      "epoch": 9.80631759027311,
      "grad_norm": 6.424704074859619,
      "learning_rate": 9.905458328233998e-07,
      "loss": 1.1236,
      "step": 321000
    },
    {
      "epoch": 9.809372517871326,
      "grad_norm": 2.5682260990142822,
      "learning_rate": 9.7524782768327e-07,
      "loss": 1.1231,
      "step": 321100
    },
    {
      "epoch": 9.812427445469542,
      "grad_norm": 11.884291648864746,
      "learning_rate": 9.599498225431404e-07,
      "loss": 1.4074,
      "step": 321200
    },
    {
      "epoch": 9.81548237306776,
      "grad_norm": 5.21375846862793,
      "learning_rate": 9.446518174030108e-07,
      "loss": 1.0686,
      "step": 321300
    },
    {
      "epoch": 9.818537300665975,
      "grad_norm": 4.836977481842041,
      "learning_rate": 9.293538122628809e-07,
      "loss": 1.1199,
      "step": 321400
    },
    {
      "epoch": 9.82159222826419,
      "grad_norm": 3.2210793495178223,
      "learning_rate": 9.140558071227512e-07,
      "loss": 1.0152,
      "step": 321500
    },
    {
      "epoch": 9.824647155862406,
      "grad_norm": 2.9414494037628174,
      "learning_rate": 8.987578019826215e-07,
      "loss": 1.08,
      "step": 321600
    },
    {
      "epoch": 9.827702083460622,
      "grad_norm": 3.7541918754577637,
      "learning_rate": 8.834597968424918e-07,
      "loss": 1.1733,
      "step": 321700
    },
    {
      "epoch": 9.830757011058838,
      "grad_norm": 3.0762391090393066,
      "learning_rate": 8.681617917023619e-07,
      "loss": 1.2142,
      "step": 321800
    },
    {
      "epoch": 9.833811938657053,
      "grad_norm": 6.95817232131958,
      "learning_rate": 8.528637865622323e-07,
      "loss": 1.3079,
      "step": 321900
    },
    {
      "epoch": 9.836866866255269,
      "grad_norm": 5.202836990356445,
      "learning_rate": 8.375657814221026e-07,
      "loss": 1.2138,
      "step": 322000
    },
    {
      "epoch": 9.839921793853486,
      "grad_norm": 4.930724620819092,
      "learning_rate": 8.222677762819729e-07,
      "loss": 1.1406,
      "step": 322100
    },
    {
      "epoch": 9.842976721451702,
      "grad_norm": 5.462831974029541,
      "learning_rate": 8.069697711418432e-07,
      "loss": 1.0949,
      "step": 322200
    },
    {
      "epoch": 9.846031649049918,
      "grad_norm": 6.223543167114258,
      "learning_rate": 7.916717660017134e-07,
      "loss": 1.1979,
      "step": 322300
    },
    {
      "epoch": 9.849086576648133,
      "grad_norm": 3.8657338619232178,
      "learning_rate": 7.763737608615837e-07,
      "loss": 0.9746,
      "step": 322400
    },
    {
      "epoch": 9.85214150424635,
      "grad_norm": 4.022277355194092,
      "learning_rate": 7.61075755721454e-07,
      "loss": 1.2531,
      "step": 322500
    },
    {
      "epoch": 9.855196431844565,
      "grad_norm": 12.097512245178223,
      "learning_rate": 7.457777505813242e-07,
      "loss": 1.2117,
      "step": 322600
    },
    {
      "epoch": 9.85825135944278,
      "grad_norm": 20.343971252441406,
      "learning_rate": 7.304797454411945e-07,
      "loss": 1.1413,
      "step": 322700
    },
    {
      "epoch": 9.861306287040996,
      "grad_norm": 20.783205032348633,
      "learning_rate": 7.151817403010648e-07,
      "loss": 1.0459,
      "step": 322800
    },
    {
      "epoch": 9.864361214639214,
      "grad_norm": 18.140941619873047,
      "learning_rate": 6.99883735160935e-07,
      "loss": 1.0668,
      "step": 322900
    },
    {
      "epoch": 9.86741614223743,
      "grad_norm": 3.446868419647217,
      "learning_rate": 6.845857300208053e-07,
      "loss": 1.0923,
      "step": 323000
    },
    {
      "epoch": 9.870471069835645,
      "grad_norm": 4.770709991455078,
      "learning_rate": 6.692877248806755e-07,
      "loss": 1.0852,
      "step": 323100
    },
    {
      "epoch": 9.87352599743386,
      "grad_norm": 12.75534439086914,
      "learning_rate": 6.539897197405459e-07,
      "loss": 1.0167,
      "step": 323200
    },
    {
      "epoch": 9.876580925032076,
      "grad_norm": 5.473841667175293,
      "learning_rate": 6.386917146004162e-07,
      "loss": 1.1701,
      "step": 323300
    },
    {
      "epoch": 9.879635852630292,
      "grad_norm": 2.295653820037842,
      "learning_rate": 6.233937094602865e-07,
      "loss": 1.2056,
      "step": 323400
    },
    {
      "epoch": 9.882690780228508,
      "grad_norm": 3.8608434200286865,
      "learning_rate": 6.080957043201567e-07,
      "loss": 1.1849,
      "step": 323500
    },
    {
      "epoch": 9.885745707826725,
      "grad_norm": 0.010462761856615543,
      "learning_rate": 5.92797699180027e-07,
      "loss": 1.1015,
      "step": 323600
    },
    {
      "epoch": 9.888800635424941,
      "grad_norm": 12.851978302001953,
      "learning_rate": 5.774996940398973e-07,
      "loss": 1.0735,
      "step": 323700
    },
    {
      "epoch": 9.891855563023157,
      "grad_norm": 7.366413116455078,
      "learning_rate": 5.622016888997675e-07,
      "loss": 1.0405,
      "step": 323800
    },
    {
      "epoch": 9.894910490621372,
      "grad_norm": 6.953218936920166,
      "learning_rate": 5.469036837596378e-07,
      "loss": 0.9401,
      "step": 323900
    },
    {
      "epoch": 9.897965418219588,
      "grad_norm": 2.4097061157226562,
      "learning_rate": 5.31605678619508e-07,
      "loss": 1.2281,
      "step": 324000
    },
    {
      "epoch": 9.901020345817804,
      "grad_norm": 6.127590179443359,
      "learning_rate": 5.164606535307796e-07,
      "loss": 1.069,
      "step": 324100
    },
    {
      "epoch": 9.90407527341602,
      "grad_norm": 8.705816268920898,
      "learning_rate": 5.011626483906499e-07,
      "loss": 1.1442,
      "step": 324200
    },
    {
      "epoch": 9.907130201014237,
      "grad_norm": 2.994471311569214,
      "learning_rate": 4.858646432505201e-07,
      "loss": 1.0269,
      "step": 324300
    },
    {
      "epoch": 9.910185128612452,
      "grad_norm": 1.9631928205490112,
      "learning_rate": 4.7056663811039046e-07,
      "loss": 1.3986,
      "step": 324400
    },
    {
      "epoch": 9.913240056210668,
      "grad_norm": 1.6768262386322021,
      "learning_rate": 4.5526863297026066e-07,
      "loss": 1.1931,
      "step": 324500
    },
    {
      "epoch": 9.916294983808884,
      "grad_norm": 5.2185235023498535,
      "learning_rate": 4.39970627830131e-07,
      "loss": 1.1374,
      "step": 324600
    },
    {
      "epoch": 9.9193499114071,
      "grad_norm": 5.210101127624512,
      "learning_rate": 4.2467262269000123e-07,
      "loss": 1.2098,
      "step": 324700
    },
    {
      "epoch": 9.922404839005315,
      "grad_norm": 6.5714545249938965,
      "learning_rate": 4.0937461754987154e-07,
      "loss": 0.9999,
      "step": 324800
    },
    {
      "epoch": 9.925459766603531,
      "grad_norm": 3.536860942840576,
      "learning_rate": 3.9407661240974174e-07,
      "loss": 1.1187,
      "step": 324900
    },
    {
      "epoch": 9.928514694201747,
      "grad_norm": 9.010876655578613,
      "learning_rate": 3.7877860726961205e-07,
      "loss": 1.3612,
      "step": 325000
    },
    {
      "epoch": 9.931569621799964,
      "grad_norm": 18.201263427734375,
      "learning_rate": 3.634806021294823e-07,
      "loss": 1.1017,
      "step": 325100
    },
    {
      "epoch": 9.93462454939818,
      "grad_norm": 10.483688354492188,
      "learning_rate": 3.4818259698935257e-07,
      "loss": 1.0799,
      "step": 325200
    },
    {
      "epoch": 9.937679476996395,
      "grad_norm": 3.2840802669525146,
      "learning_rate": 3.328845918492229e-07,
      "loss": 1.2754,
      "step": 325300
    },
    {
      "epoch": 9.940734404594611,
      "grad_norm": 3.8806185722351074,
      "learning_rate": 3.175865867090932e-07,
      "loss": 1.0955,
      "step": 325400
    },
    {
      "epoch": 9.943789332192827,
      "grad_norm": 10.518257141113281,
      "learning_rate": 3.0228858156896344e-07,
      "loss": 1.3088,
      "step": 325500
    },
    {
      "epoch": 9.946844259791042,
      "grad_norm": 10.4476957321167,
      "learning_rate": 2.869905764288337e-07,
      "loss": 1.0003,
      "step": 325600
    },
    {
      "epoch": 9.949899187389258,
      "grad_norm": 10.631278991699219,
      "learning_rate": 2.7169257128870396e-07,
      "loss": 1.2052,
      "step": 325700
    },
    {
      "epoch": 9.952954114987476,
      "grad_norm": 15.547146797180176,
      "learning_rate": 2.563945661485742e-07,
      "loss": 0.9217,
      "step": 325800
    },
    {
      "epoch": 9.956009042585691,
      "grad_norm": 8.524794578552246,
      "learning_rate": 2.410965610084445e-07,
      "loss": 1.2376,
      "step": 325900
    },
    {
      "epoch": 9.959063970183907,
      "grad_norm": 10.924880981445312,
      "learning_rate": 2.2579855586831478e-07,
      "loss": 0.891,
      "step": 326000
    },
    {
      "epoch": 9.962118897782123,
      "grad_norm": 6.899311542510986,
      "learning_rate": 2.1050055072818504e-07,
      "loss": 1.0654,
      "step": 326100
    },
    {
      "epoch": 9.965173825380338,
      "grad_norm": 9.488914489746094,
      "learning_rate": 1.9520254558805532e-07,
      "loss": 1.3211,
      "step": 326200
    },
    {
      "epoch": 9.968228752978554,
      "grad_norm": 4.252932071685791,
      "learning_rate": 1.799045404479256e-07,
      "loss": 1.0891,
      "step": 326300
    },
    {
      "epoch": 9.97128368057677,
      "grad_norm": 2.436544418334961,
      "learning_rate": 1.6460653530779588e-07,
      "loss": 1.1258,
      "step": 326400
    },
    {
      "epoch": 9.974338608174985,
      "grad_norm": 10.499905586242676,
      "learning_rate": 1.4930853016766614e-07,
      "loss": 1.2108,
      "step": 326500
    },
    {
      "epoch": 9.977393535773203,
      "grad_norm": 6.818589687347412,
      "learning_rate": 1.3401052502753642e-07,
      "loss": 1.131,
      "step": 326600
    },
    {
      "epoch": 9.980448463371419,
      "grad_norm": 12.36090087890625,
      "learning_rate": 1.1871251988740668e-07,
      "loss": 1.1388,
      "step": 326700
    },
    {
      "epoch": 9.983503390969634,
      "grad_norm": 15.297159194946289,
      "learning_rate": 1.0341451474727696e-07,
      "loss": 1.1518,
      "step": 326800
    },
    {
      "epoch": 9.98655831856785,
      "grad_norm": 3.3843600749969482,
      "learning_rate": 8.811650960714722e-08,
      "loss": 0.97,
      "step": 326900
    },
    {
      "epoch": 9.989613246166066,
      "grad_norm": 0.6283736228942871,
      "learning_rate": 7.28185044670175e-08,
      "loss": 1.2331,
      "step": 327000
    },
    {
      "epoch": 9.992668173764281,
      "grad_norm": 12.6235933303833,
      "learning_rate": 5.7520499326887774e-08,
      "loss": 1.2373,
      "step": 327100
    },
    {
      "epoch": 9.995723101362497,
      "grad_norm": 9.48670768737793,
      "learning_rate": 4.222249418675805e-08,
      "loss": 0.9996,
      "step": 327200
    },
    {
      "epoch": 9.998778028960714,
      "grad_norm": 10.663695335388184,
      "learning_rate": 2.707746909802962e-08,
      "loss": 1.1021,
      "step": 327300
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.7415225759149509,
      "eval_loss": 0.7014397382736206,
      "eval_runtime": 1867.5396,
      "eval_samples_per_second": 17.528,
      "eval_steps_per_second": 4.382,
      "step": 327340
    },
    {
      "epoch": 10.0,
      "step": 327340,
      "total_flos": 2.8536683814220615e+20,
      "train_loss": 1.1817914221248462,
      "train_runtime": 92460.3356,
      "train_samples_per_second": 14.161,
      "train_steps_per_second": 3.54
    }
  ],
  "logging_steps": 100,
  "max_steps": 327340,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.8536683814220615e+20,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}