{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.8838652482269502,
  "eval_steps": 500,
  "global_step": 21250,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004432624113475178,
      "grad_norm": 0.009513450788341307,
      "learning_rate": 5.066502241635949e-06,
      "loss": 0.0353,
      "step": 50
    },
    {
      "epoch": 0.008865248226950355,
      "grad_norm": 0.12959164414811705,
      "learning_rate": 5.964204463534375e-06,
      "loss": 0.0209,
      "step": 100
    },
    {
      "epoch": 0.013297872340425532,
      "grad_norm": 0.03079899259156006,
      "learning_rate": 6.489326600159014e-06,
      "loss": 0.0114,
      "step": 150
    },
    {
      "epoch": 0.01773049645390071,
      "grad_norm": 1.904334998115993,
      "learning_rate": 6.855414889666249e-06,
      "loss": 0.0426,
      "step": 200
    },
    {
      "epoch": 0.022163120567375887,
      "grad_norm": 0.00029657300723797526,
      "learning_rate": 7.14571142073327e-06,
      "loss": 0.0092,
      "step": 250
    },
    {
      "epoch": 0.026595744680851064,
      "grad_norm": 0.5635451567630495,
      "learning_rate": 7.382704575824005e-06,
      "loss": 0.0207,
      "step": 300
    },
    {
      "epoch": 0.03102836879432624,
      "grad_norm": 17.174480153841266,
      "learning_rate": 7.582965380729319e-06,
      "loss": 0.0128,
      "step": 350
    },
    {
      "epoch": 0.03546099290780142,
      "grad_norm": 0.0029856714585610065,
      "learning_rate": 7.756367076993413e-06,
      "loss": 0.0087,
      "step": 400
    },
    {
      "epoch": 0.0398936170212766,
      "grad_norm": 46.59324697917607,
      "learning_rate": 7.909269732702365e-06,
      "loss": 0.0358,
      "step": 450
    },
    {
      "epoch": 0.044326241134751775,
      "grad_norm": 0.011038539916815544,
      "learning_rate": 8.046011658635775e-06,
      "loss": 0.0099,
      "step": 500
    },
    {
      "epoch": 0.04875886524822695,
      "grad_norm": 0.06122707900561642,
      "learning_rate": 8.169684802472282e-06,
      "loss": 0.0257,
      "step": 550
    },
    {
      "epoch": 0.05319148936170213,
      "grad_norm": 11.17528630173658,
      "learning_rate": 8.28257072561721e-06,
      "loss": 0.0399,
      "step": 600
    },
    {
      "epoch": 0.057624113475177305,
      "grad_norm": 0.004434274436946811,
      "learning_rate": 8.386401186603816e-06,
      "loss": 0.0106,
      "step": 650
    },
    {
      "epoch": 0.06205673758865248,
      "grad_norm": 0.00022486652288303773,
      "learning_rate": 8.482521734002021e-06,
      "loss": 0.0175,
      "step": 700
    },
    {
      "epoch": 0.06648936170212766,
      "grad_norm": 0.08252216933057055,
      "learning_rate": 8.571998643735778e-06,
      "loss": 0.0122,
      "step": 750
    },
    {
      "epoch": 0.07092198581560284,
      "grad_norm": 38.71346721678791,
      "learning_rate": 8.655691228400912e-06,
      "loss": 0.0107,
      "step": 800
    },
    {
      "epoch": 0.07535460992907801,
      "grad_norm": 0.016399532174579123,
      "learning_rate": 8.734302159716567e-06,
      "loss": 0.0272,
      "step": 850
    },
    {
      "epoch": 0.0797872340425532,
      "grad_norm": 13.96210364288869,
      "learning_rate": 8.808413368821617e-06,
      "loss": 0.0162,
      "step": 900
    },
    {
      "epoch": 0.08421985815602837,
      "grad_norm": 20.616657581291367,
      "learning_rate": 8.878512216644468e-06,
      "loss": 0.018,
      "step": 950
    },
    {
      "epoch": 0.08865248226950355,
      "grad_norm": 8.554096191036654,
      "learning_rate": 8.945010936770532e-06,
      "loss": 0.0122,
      "step": 1000
    },
    {
      "epoch": 0.09308510638297872,
      "grad_norm": 9.002995310481221,
      "learning_rate": 9.008261325009621e-06,
      "loss": 0.004,
      "step": 1050
    },
    {
      "epoch": 0.0975177304964539,
      "grad_norm": 0.04332552525052799,
      "learning_rate": 9.068566005374956e-06,
      "loss": 0.0049,
      "step": 1100
    },
    {
      "epoch": 0.10195035460992907,
      "grad_norm": 0.566818534495066,
      "learning_rate": 9.126187187473155e-06,
      "loss": 0.0348,
      "step": 1150
    },
    {
      "epoch": 0.10638297872340426,
      "grad_norm": 2.2102759352121244,
      "learning_rate": 9.181353557126703e-06,
      "loss": 0.0157,
      "step": 1200
    },
    {
      "epoch": 0.11081560283687943,
      "grad_norm": 0.0007883033335850071,
      "learning_rate": 9.233228424991629e-06,
      "loss": 0.0153,
      "step": 1250
    },
    {
      "epoch": 0.11524822695035461,
      "grad_norm": 8.962657174789879,
      "learning_rate": 9.28410340850224e-06,
      "loss": 0.011,
      "step": 1300
    },
    {
      "epoch": 0.1196808510638298,
      "grad_norm": 0.08478258585114734,
      "learning_rate": 9.333055212266742e-06,
      "loss": 0.0193,
      "step": 1350
    },
    {
      "epoch": 0.12411347517730496,
      "grad_norm": 5.126269299634291,
      "learning_rate": 9.380223955900445e-06,
      "loss": 0.0224,
      "step": 1400
    },
    {
      "epoch": 0.12854609929078015,
      "grad_norm": 0.001630326745394445,
      "learning_rate": 9.425734982425631e-06,
      "loss": 0.0198,
      "step": 1450
    },
    {
      "epoch": 0.13297872340425532,
      "grad_norm": 31.3251284414765,
      "learning_rate": 9.469700865634203e-06,
      "loss": 0.0201,
      "step": 1500
    },
    {
      "epoch": 0.1374113475177305,
      "grad_norm": 0.003720201610220148,
      "learning_rate": 9.512223087748166e-06,
      "loss": 0.0386,
      "step": 1550
    },
    {
      "epoch": 0.14184397163120568,
      "grad_norm": 1.4908536371789232,
      "learning_rate": 9.553393450299339e-06,
      "loss": 0.0155,
      "step": 1600
    },
    {
      "epoch": 0.14627659574468085,
      "grad_norm": 0.18557111682138577,
      "learning_rate": 9.593295267571064e-06,
      "loss": 0.0132,
      "step": 1650
    },
    {
      "epoch": 0.15070921985815602,
      "grad_norm": 0.0012351164899243579,
      "learning_rate": 9.632004381614992e-06,
      "loss": 0.0169,
      "step": 1700
    },
    {
      "epoch": 0.1551418439716312,
      "grad_norm": 4.390950581164127,
      "learning_rate": 9.669590029926173e-06,
      "loss": 0.0188,
      "step": 1750
    },
    {
      "epoch": 0.1595744680851064,
      "grad_norm": 1.5823201233659507,
      "learning_rate": 9.706115590720043e-06,
      "loss": 0.0176,
      "step": 1800
    },
    {
      "epoch": 0.16400709219858156,
      "grad_norm": 0.6041784257950227,
      "learning_rate": 9.741639225963296e-06,
      "loss": 0.0076,
      "step": 1850
    },
    {
      "epoch": 0.16843971631205673,
      "grad_norm": 8.096633927567137,
      "learning_rate": 9.776214438542894e-06,
      "loss": 0.0254,
      "step": 1900
    },
    {
      "epoch": 0.17287234042553193,
      "grad_norm": 0.001381384244099819,
      "learning_rate": 9.80989055697396e-06,
      "loss": 0.021,
      "step": 1950
    },
    {
      "epoch": 0.1773049645390071,
      "grad_norm": 0.6173202999510908,
      "learning_rate": 9.842713158668956e-06,
      "loss": 0.0153,
      "step": 2000
    },
    {
      "epoch": 0.18173758865248227,
      "grad_norm": 0.4702304667158159,
      "learning_rate": 9.874724440882673e-06,
      "loss": 0.0322,
      "step": 2050
    },
    {
      "epoch": 0.18617021276595744,
      "grad_norm": 0.007418267836546937,
      "learning_rate": 9.905963546908046e-06,
      "loss": 0.0189,
      "step": 2100
    },
    {
      "epoch": 0.19060283687943264,
      "grad_norm": 0.040611583128721894,
      "learning_rate": 9.936466853848526e-06,
      "loss": 0.004,
      "step": 2150
    },
    {
      "epoch": 0.1950354609929078,
      "grad_norm": 35.80087991837194,
      "learning_rate": 9.96626822727338e-06,
      "loss": 0.022,
      "step": 2200
    },
    {
      "epoch": 0.19946808510638298,
      "grad_norm": 0.44489231698014525,
      "learning_rate": 9.995399247226628e-06,
      "loss": 0.0165,
      "step": 2250
    },
    {
      "epoch": 0.20390070921985815,
      "grad_norm": 0.0002357955980587179,
      "learning_rate": 9.980299448384557e-06,
      "loss": 0.0091,
      "step": 2300
    },
    {
      "epoch": 0.20833333333333334,
      "grad_norm": 0.011156109382490509,
      "learning_rate": 9.955673758865249e-06,
      "loss": 0.0293,
      "step": 2350
    },
    {
      "epoch": 0.2127659574468085,
      "grad_norm": 0.010237311021590336,
      "learning_rate": 9.931048069345942e-06,
      "loss": 0.0267,
      "step": 2400
    },
    {
      "epoch": 0.21719858156028368,
      "grad_norm": 0.1907181340319684,
      "learning_rate": 9.906422379826635e-06,
      "loss": 0.0035,
      "step": 2450
    },
    {
      "epoch": 0.22163120567375885,
      "grad_norm": 7.244482111036572,
      "learning_rate": 9.881796690307329e-06,
      "loss": 0.0129,
      "step": 2500
    },
    {
      "epoch": 0.22606382978723405,
      "grad_norm": 2.614333142067009,
      "learning_rate": 9.857171000788024e-06,
      "loss": 0.0176,
      "step": 2550
    },
    {
      "epoch": 0.23049645390070922,
      "grad_norm": 0.19026758190049964,
      "learning_rate": 9.832545311268717e-06,
      "loss": 0.0118,
      "step": 2600
    },
    {
      "epoch": 0.2349290780141844,
      "grad_norm": 1.926354870277716,
      "learning_rate": 9.80791962174941e-06,
      "loss": 0.0113,
      "step": 2650
    },
    {
      "epoch": 0.2393617021276596,
      "grad_norm": 1.3390029353713434e-05,
      "learning_rate": 9.783293932230104e-06,
      "loss": 0.0095,
      "step": 2700
    },
    {
      "epoch": 0.24379432624113476,
      "grad_norm": 0.25506782394102356,
      "learning_rate": 9.758668242710797e-06,
      "loss": 0.0235,
      "step": 2750
    },
    {
      "epoch": 0.24822695035460993,
      "grad_norm": 0.0011667263533171556,
      "learning_rate": 9.73404255319149e-06,
      "loss": 0.01,
      "step": 2800
    },
    {
      "epoch": 0.2526595744680851,
      "grad_norm": 2.676783820787547,
      "learning_rate": 9.709416863672184e-06,
      "loss": 0.0305,
      "step": 2850
    },
    {
      "epoch": 0.2570921985815603,
      "grad_norm": 0.0019099837220493063,
      "learning_rate": 9.684791174152877e-06,
      "loss": 0.0246,
      "step": 2900
    },
    {
      "epoch": 0.26152482269503546,
      "grad_norm": 0.6898702380926142,
      "learning_rate": 9.66016548463357e-06,
      "loss": 0.0211,
      "step": 2950
    },
    {
      "epoch": 0.26595744680851063,
      "grad_norm": 0.0006438124789484562,
      "learning_rate": 9.635539795114264e-06,
      "loss": 0.008,
      "step": 3000
    },
    {
      "epoch": 0.2703900709219858,
      "grad_norm": 3.819700570925564e-05,
      "learning_rate": 9.610914105594957e-06,
      "loss": 0.0199,
      "step": 3050
    },
    {
      "epoch": 0.274822695035461,
      "grad_norm": 0.019487722143147885,
      "learning_rate": 9.58628841607565e-06,
      "loss": 0.0126,
      "step": 3100
    },
    {
      "epoch": 0.27925531914893614,
      "grad_norm": 0.49075883570668455,
      "learning_rate": 9.561662726556344e-06,
      "loss": 0.0114,
      "step": 3150
    },
    {
      "epoch": 0.28368794326241137,
      "grad_norm": 0.661703227107662,
      "learning_rate": 9.537037037037037e-06,
      "loss": 0.0081,
      "step": 3200
    },
    {
      "epoch": 0.28812056737588654,
      "grad_norm": 0.00012406562558187035,
      "learning_rate": 9.512411347517732e-06,
      "loss": 0.0154,
      "step": 3250
    },
    {
      "epoch": 0.2925531914893617,
      "grad_norm": 10.456271349514031,
      "learning_rate": 9.487785657998426e-06,
      "loss": 0.0056,
      "step": 3300
    },
    {
      "epoch": 0.2969858156028369,
      "grad_norm": 34.88954914653437,
      "learning_rate": 9.463159968479119e-06,
      "loss": 0.0215,
      "step": 3350
    },
    {
      "epoch": 0.30141843971631205,
      "grad_norm": 0.0037401951660654617,
      "learning_rate": 9.43853427895981e-06,
      "loss": 0.0073,
      "step": 3400
    },
    {
      "epoch": 0.3058510638297872,
      "grad_norm": 1.9770145427070573,
      "learning_rate": 9.413908589440504e-06,
      "loss": 0.026,
      "step": 3450
    },
    {
      "epoch": 0.3102836879432624,
      "grad_norm": 0.0005171843885202088,
      "learning_rate": 9.389282899921197e-06,
      "loss": 0.0054,
      "step": 3500
    },
    {
      "epoch": 0.3147163120567376,
      "grad_norm": 6.671945039975605,
      "learning_rate": 9.364657210401892e-06,
      "loss": 0.0193,
      "step": 3550
    },
    {
      "epoch": 0.3191489361702128,
      "grad_norm": 24.199369765973504,
      "learning_rate": 9.340031520882586e-06,
      "loss": 0.0253,
      "step": 3600
    },
    {
      "epoch": 0.32358156028368795,
      "grad_norm": 0.03237450123162466,
      "learning_rate": 9.31540583136328e-06,
      "loss": 0.0108,
      "step": 3650
    },
    {
      "epoch": 0.3280141843971631,
      "grad_norm": 0.06582091922650166,
      "learning_rate": 9.290780141843973e-06,
      "loss": 0.0052,
      "step": 3700
    },
    {
      "epoch": 0.3324468085106383,
      "grad_norm": 1.1046639468259907,
      "learning_rate": 9.266154452324666e-06,
      "loss": 0.0077,
      "step": 3750
    },
    {
      "epoch": 0.33687943262411346,
      "grad_norm": 0.12685978465854558,
      "learning_rate": 9.24152876280536e-06,
      "loss": 0.0055,
      "step": 3800
    },
    {
      "epoch": 0.34131205673758863,
      "grad_norm": 0.001176875548815465,
      "learning_rate": 9.216903073286053e-06,
      "loss": 0.0116,
      "step": 3850
    },
    {
      "epoch": 0.34574468085106386,
      "grad_norm": 19.93923722594081,
      "learning_rate": 9.192277383766746e-06,
      "loss": 0.0339,
      "step": 3900
    },
    {
      "epoch": 0.350177304964539,
      "grad_norm": 0.5136555426144171,
      "learning_rate": 9.16765169424744e-06,
      "loss": 0.0109,
      "step": 3950
    },
    {
      "epoch": 0.3546099290780142,
      "grad_norm": 29.577148937276096,
      "learning_rate": 9.143026004728133e-06,
      "loss": 0.0195,
      "step": 4000
    },
    {
      "epoch": 0.35904255319148937,
      "grad_norm": 5.471652062023114,
      "learning_rate": 9.118400315208826e-06,
      "loss": 0.0075,
      "step": 4050
    },
    {
      "epoch": 0.36347517730496454,
      "grad_norm": 0.0039179745195827426,
      "learning_rate": 9.09377462568952e-06,
      "loss": 0.0203,
      "step": 4100
    },
    {
      "epoch": 0.3679078014184397,
      "grad_norm": 30.214203138409303,
      "learning_rate": 9.069148936170213e-06,
      "loss": 0.0323,
      "step": 4150
    },
    {
      "epoch": 0.3723404255319149,
      "grad_norm": 1.0344365911211837,
      "learning_rate": 9.044523246650908e-06,
      "loss": 0.0252,
      "step": 4200
    },
    {
      "epoch": 0.37677304964539005,
      "grad_norm": 0.6080095692433664,
      "learning_rate": 9.019897557131601e-06,
      "loss": 0.0065,
      "step": 4250
    },
    {
      "epoch": 0.38120567375886527,
      "grad_norm": 1.132001494443736,
      "learning_rate": 8.995271867612294e-06,
      "loss": 0.0137,
      "step": 4300
    },
    {
      "epoch": 0.38563829787234044,
      "grad_norm": 0.6281310332658542,
      "learning_rate": 8.971138691883373e-06,
      "loss": 0.0086,
      "step": 4350
    },
    {
      "epoch": 0.3900709219858156,
      "grad_norm": 2.5248461085388904e-05,
      "learning_rate": 8.946513002364066e-06,
      "loss": 0.0015,
      "step": 4400
    },
    {
      "epoch": 0.3945035460992908,
      "grad_norm": 0.010001162867900915,
      "learning_rate": 8.921887312844761e-06,
      "loss": 0.0174,
      "step": 4450
    },
    {
      "epoch": 0.39893617021276595,
      "grad_norm": 0.3111641106006767,
      "learning_rate": 8.897261623325454e-06,
      "loss": 0.0174,
      "step": 4500
    },
    {
      "epoch": 0.4033687943262411,
      "grad_norm": 0.022375831871911778,
      "learning_rate": 8.872635933806148e-06,
      "loss": 0.0115,
      "step": 4550
    },
    {
      "epoch": 0.4078014184397163,
      "grad_norm": 0.08396303659869327,
      "learning_rate": 8.848010244286841e-06,
      "loss": 0.0136,
      "step": 4600
    },
    {
      "epoch": 0.4122340425531915,
      "grad_norm": 5.854304446661381,
      "learning_rate": 8.823384554767534e-06,
      "loss": 0.0042,
      "step": 4650
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 16.679488984474407,
      "learning_rate": 8.798758865248228e-06,
      "loss": 0.0102,
      "step": 4700
    },
    {
      "epoch": 0.42109929078014185,
      "grad_norm": 0.6379686502927935,
      "learning_rate": 8.774625689519308e-06,
      "loss": 0.019,
      "step": 4750
    },
    {
      "epoch": 0.425531914893617,
      "grad_norm": 0.05658909698246918,
      "learning_rate": 8.750000000000001e-06,
      "loss": 0.0178,
      "step": 4800
    },
    {
      "epoch": 0.4299645390070922,
      "grad_norm": 0.36425987836236606,
      "learning_rate": 8.725374310480694e-06,
      "loss": 0.0212,
      "step": 4850
    },
    {
      "epoch": 0.43439716312056736,
      "grad_norm": 2.0754475949590696,
      "learning_rate": 8.700748620961388e-06,
      "loss": 0.0077,
      "step": 4900
    },
    {
      "epoch": 0.43882978723404253,
      "grad_norm": 0.012181818674097341,
      "learning_rate": 8.676122931442081e-06,
      "loss": 0.0125,
      "step": 4950
    },
    {
      "epoch": 0.4432624113475177,
      "grad_norm": 0.00018403879761046281,
      "learning_rate": 8.651497241922774e-06,
      "loss": 0.0188,
      "step": 5000
    },
    {
      "epoch": 0.44769503546099293,
      "grad_norm": 0.004407047372502831,
      "learning_rate": 8.626871552403468e-06,
      "loss": 0.0073,
      "step": 5050
    },
    {
      "epoch": 0.4521276595744681,
      "grad_norm": 0.012230911143282817,
      "learning_rate": 8.602245862884161e-06,
      "loss": 0.008,
      "step": 5100
    },
    {
      "epoch": 0.45656028368794327,
      "grad_norm": 0.9921871395560984,
      "learning_rate": 8.577620173364854e-06,
      "loss": 0.0168,
      "step": 5150
    },
    {
      "epoch": 0.46099290780141844,
      "grad_norm": 0.1285429154497177,
      "learning_rate": 8.552994483845548e-06,
      "loss": 0.0122,
      "step": 5200
    },
    {
      "epoch": 0.4654255319148936,
      "grad_norm": 3.308588561647349,
      "learning_rate": 8.528861308116628e-06,
      "loss": 0.0158,
      "step": 5250
    },
    {
      "epoch": 0.4698581560283688,
      "grad_norm": 2.8184778898385976,
      "learning_rate": 8.504235618597323e-06,
      "loss": 0.0237,
      "step": 5300
    },
    {
      "epoch": 0.47429078014184395,
      "grad_norm": 0.47692861720548735,
      "learning_rate": 8.479609929078016e-06,
      "loss": 0.0181,
      "step": 5350
    },
    {
      "epoch": 0.4787234042553192,
      "grad_norm": 0.002116990082837995,
      "learning_rate": 8.454984239558708e-06,
      "loss": 0.0235,
      "step": 5400
    },
    {
      "epoch": 0.48315602836879434,
      "grad_norm": 0.016230203235315883,
      "learning_rate": 8.430358550039401e-06,
      "loss": 0.0123,
      "step": 5450
    },
    {
      "epoch": 0.4875886524822695,
      "grad_norm": 0.10798463365752625,
      "learning_rate": 8.405732860520094e-06,
      "loss": 0.0202,
      "step": 5500
    },
    {
      "epoch": 0.4920212765957447,
      "grad_norm": 0.00038212180284718085,
      "learning_rate": 8.381107171000788e-06,
      "loss": 0.0049,
      "step": 5550
    },
    {
      "epoch": 0.49645390070921985,
      "grad_norm": 0.0022870323047176084,
      "learning_rate": 8.356481481481483e-06,
      "loss": 0.0061,
      "step": 5600
    },
    {
      "epoch": 0.500886524822695,
      "grad_norm": 0.4205002106769229,
      "learning_rate": 8.331855791962176e-06,
      "loss": 0.0018,
      "step": 5650
    },
    {
      "epoch": 0.5053191489361702,
      "grad_norm": 0.0003576863161645832,
      "learning_rate": 8.30723010244287e-06,
      "loss": 0.01,
      "step": 5700
    },
    {
      "epoch": 0.5097517730496454,
      "grad_norm": 27.715734272602834,
      "learning_rate": 8.282604412923563e-06,
      "loss": 0.0173,
      "step": 5750
    },
    {
      "epoch": 0.5141843971631206,
      "grad_norm": 0.0029419354875175013,
      "learning_rate": 8.257978723404256e-06,
      "loss": 0.0067,
      "step": 5800
    },
    {
      "epoch": 0.5186170212765957,
      "grad_norm": 27.610931266268892,
      "learning_rate": 8.23335303388495e-06,
      "loss": 0.0116,
      "step": 5850
    },
    {
      "epoch": 0.5230496453900709,
      "grad_norm": 0.004511848611750885,
      "learning_rate": 8.208727344365643e-06,
      "loss": 0.0115,
      "step": 5900
    },
    {
      "epoch": 0.5274822695035462,
      "grad_norm": 0.5244401751220326,
      "learning_rate": 8.184101654846336e-06,
      "loss": 0.0025,
      "step": 5950
    },
    {
      "epoch": 0.5319148936170213,
      "grad_norm": 0.0075670950741103474,
      "learning_rate": 8.15947596532703e-06,
      "loss": 0.0164,
      "step": 6000
    },
    {
      "epoch": 0.5363475177304965,
      "grad_norm": 0.06038816285079991,
      "learning_rate": 8.134850275807723e-06,
      "loss": 0.0079,
      "step": 6050
    },
    {
      "epoch": 0.5407801418439716,
      "grad_norm": 0.0007768875893699433,
      "learning_rate": 8.110224586288416e-06,
      "loss": 0.0052,
      "step": 6100
    },
    {
      "epoch": 0.5452127659574468,
      "grad_norm": 0.2909181418775317,
      "learning_rate": 8.08559889676911e-06,
      "loss": 0.0032,
      "step": 6150
    },
    {
      "epoch": 0.549645390070922,
      "grad_norm": 6.463507993140611,
      "learning_rate": 8.060973207249803e-06,
      "loss": 0.0098,
      "step": 6200
    },
    {
      "epoch": 0.5540780141843972,
      "grad_norm": 0.26740235451618943,
      "learning_rate": 8.036347517730498e-06,
      "loss": 0.0144,
      "step": 6250
    },
    {
      "epoch": 0.5585106382978723,
      "grad_norm": 0.00010577380374890247,
      "learning_rate": 8.011721828211191e-06,
      "loss": 0.0156,
      "step": 6300
    },
    {
      "epoch": 0.5629432624113475,
      "grad_norm": 0.14691208386477084,
      "learning_rate": 7.987096138691885e-06,
      "loss": 0.0152,
      "step": 6350
    },
    {
      "epoch": 0.5673758865248227,
      "grad_norm": 0.07499554653157833,
      "learning_rate": 7.962470449172578e-06,
      "loss": 0.0282,
      "step": 6400
    },
    {
      "epoch": 0.5718085106382979,
      "grad_norm": 0.005398079692845853,
      "learning_rate": 7.93784475965327e-06,
      "loss": 0.0193,
      "step": 6450
    },
    {
      "epoch": 0.5762411347517731,
      "grad_norm": 0.8004218807047324,
      "learning_rate": 7.913219070133963e-06,
      "loss": 0.0059,
      "step": 6500
    },
    {
      "epoch": 0.5806737588652482,
      "grad_norm": 0.0018961263364736887,
      "learning_rate": 7.888593380614658e-06,
      "loss": 0.0192,
      "step": 6550
    },
    {
      "epoch": 0.5851063829787234,
      "grad_norm": 7.167523295933813,
      "learning_rate": 7.863967691095352e-06,
      "loss": 0.0026,
      "step": 6600
    },
    {
      "epoch": 0.5895390070921985,
      "grad_norm": 0.2545664640109103,
      "learning_rate": 7.839342001576045e-06,
      "loss": 0.0178,
      "step": 6650
    },
    {
      "epoch": 0.5939716312056738,
      "grad_norm": 0.0003677116767933612,
      "learning_rate": 7.814716312056738e-06,
      "loss": 0.0127,
      "step": 6700
    },
    {
      "epoch": 0.598404255319149,
      "grad_norm": 1.449028640163906,
      "learning_rate": 7.790090622537432e-06,
      "loss": 0.0353,
      "step": 6750
    },
    {
      "epoch": 0.6028368794326241,
      "grad_norm": 0.018076926628577003,
      "learning_rate": 7.765464933018125e-06,
      "loss": 0.0022,
      "step": 6800
    },
    {
      "epoch": 0.6072695035460993,
      "grad_norm": 1.314844716480283,
      "learning_rate": 7.740839243498818e-06,
      "loss": 0.0125,
      "step": 6850
    },
    {
      "epoch": 0.6117021276595744,
      "grad_norm": 0.05713668183450516,
      "learning_rate": 7.716213553979512e-06,
      "loss": 0.0103,
      "step": 6900
    },
    {
      "epoch": 0.6161347517730497,
      "grad_norm": 0.00012171886896694071,
      "learning_rate": 7.691587864460207e-06,
      "loss": 0.0411,
      "step": 6950
    },
    {
      "epoch": 0.6205673758865248,
      "grad_norm": 0.005050200509222459,
      "learning_rate": 7.666962174940898e-06,
      "loss": 0.0088,
      "step": 7000
    },
    {
      "epoch": 0.625,
      "grad_norm": 0.044231465953618015,
      "learning_rate": 7.642336485421592e-06,
      "loss": 0.0121,
      "step": 7050
    },
    {
      "epoch": 0.6294326241134752,
      "grad_norm": 0.002970509653416846,
      "learning_rate": 7.617710795902286e-06,
      "loss": 0.0115,
      "step": 7100
    },
    {
      "epoch": 0.6338652482269503,
      "grad_norm": 0.061580769377440274,
      "learning_rate": 7.593085106382979e-06,
      "loss": 0.0074,
      "step": 7150
    },
    {
      "epoch": 0.6382978723404256,
      "grad_norm": 0.0030036827619670152,
      "learning_rate": 7.568459416863673e-06,
      "loss": 0.0075,
      "step": 7200
    },
    {
      "epoch": 0.6427304964539007,
      "grad_norm": 0.12485293215547061,
      "learning_rate": 7.543833727344366e-06,
      "loss": 0.0112,
      "step": 7250
    },
    {
      "epoch": 0.6471631205673759,
      "grad_norm": 0.001231009728533709,
      "learning_rate": 7.519208037825059e-06,
      "loss": 0.0068,
      "step": 7300
    },
    {
      "epoch": 0.651595744680851,
      "grad_norm": 1.066825644497329,
      "learning_rate": 7.494582348305754e-06,
      "loss": 0.0219,
      "step": 7350
    },
    {
      "epoch": 0.6560283687943262,
      "grad_norm": 0.18183803085941427,
      "learning_rate": 7.469956658786447e-06,
      "loss": 0.0052,
      "step": 7400
    },
    {
      "epoch": 0.6604609929078015,
      "grad_norm": 7.176261541067922,
      "learning_rate": 7.44533096926714e-06,
      "loss": 0.012,
      "step": 7450
    },
    {
      "epoch": 0.6648936170212766,
      "grad_norm": 0.08470951895549413,
      "learning_rate": 7.420705279747834e-06,
      "loss": 0.005,
      "step": 7500
    },
    {
      "epoch": 0.6693262411347518,
      "grad_norm": 0.43564519071924557,
      "learning_rate": 7.396079590228526e-06,
      "loss": 0.0214,
      "step": 7550
    },
    {
      "epoch": 0.6737588652482269,
      "grad_norm": 0.006827701243280031,
      "learning_rate": 7.37145390070922e-06,
      "loss": 0.0382,
      "step": 7600
    },
    {
      "epoch": 0.6781914893617021,
      "grad_norm": 0.0006360372069413289,
      "learning_rate": 7.346828211189914e-06,
      "loss": 0.0058,
      "step": 7650
    },
    {
      "epoch": 0.6826241134751773,
      "grad_norm": 0.011063982123596255,
      "learning_rate": 7.322202521670607e-06,
      "loss": 0.0074,
      "step": 7700
    },
    {
      "epoch": 0.6870567375886525,
      "grad_norm": 0.8991727722769103,
      "learning_rate": 7.2975768321513005e-06,
      "loss": 0.0108,
      "step": 7750
    },
    {
      "epoch": 0.6914893617021277,
      "grad_norm": 0.012017251234112784,
      "learning_rate": 7.272951142631995e-06,
      "loss": 0.0127,
      "step": 7800
    },
    {
      "epoch": 0.6959219858156028,
      "grad_norm": 0.19374738160794408,
      "learning_rate": 7.248325453112688e-06,
      "loss": 0.0035,
      "step": 7850
    },
    {
      "epoch": 0.700354609929078,
      "grad_norm": 2.341510058377949,
      "learning_rate": 7.223699763593381e-06,
      "loss": 0.0317,
      "step": 7900
    },
    {
      "epoch": 0.7047872340425532,
      "grad_norm": 5.84240259775226e-06,
      "learning_rate": 7.199074074074075e-06,
      "loss": 0.0205,
      "step": 7950
    },
    {
      "epoch": 0.7092198581560284,
      "grad_norm": 0.0022186038540356347,
      "learning_rate": 7.174448384554769e-06,
      "loss": 0.0206,
      "step": 8000
    },
    {
      "epoch": 0.7136524822695035,
      "grad_norm": 2.3575055124131414,
      "learning_rate": 7.149822695035462e-06,
      "loss": 0.0047,
      "step": 8050
    },
    {
      "epoch": 0.7180851063829787,
      "grad_norm": 0.01023308841827913,
      "learning_rate": 7.125197005516155e-06,
      "loss": 0.0293,
      "step": 8100
    },
    {
      "epoch": 0.7225177304964538,
      "grad_norm": 0.006391574218526128,
      "learning_rate": 7.100571315996848e-06,
      "loss": 0.0137,
      "step": 8150
    },
    {
      "epoch": 0.7269503546099291,
      "grad_norm": 1.7478784234914295,
      "learning_rate": 7.0759456264775415e-06,
      "loss": 0.0144,
      "step": 8200
    },
    {
      "epoch": 0.7313829787234043,
      "grad_norm": 0.4480580469017403,
      "learning_rate": 7.051319936958235e-06,
      "loss": 0.0236,
      "step": 8250
    },
    {
      "epoch": 0.7358156028368794,
      "grad_norm": 0.0001810108372971621,
      "learning_rate": 7.0271867612293155e-06,
      "loss": 0.0162,
      "step": 8300
    },
    {
      "epoch": 0.7402482269503546,
      "grad_norm": 0.004944937344540902,
      "learning_rate": 7.002561071710009e-06,
      "loss": 0.0045,
      "step": 8350
    },
    {
      "epoch": 0.7446808510638298,
      "grad_norm": 0.009033810427587684,
      "learning_rate": 6.977935382190701e-06,
      "loss": 0.0021,
      "step": 8400
    },
    {
      "epoch": 0.749113475177305,
      "grad_norm": 0.0007266417865651897,
      "learning_rate": 6.953309692671395e-06,
      "loss": 0.0028,
      "step": 8450
    },
    {
      "epoch": 0.7535460992907801,
      "grad_norm": 0.5472428310965101,
      "learning_rate": 6.928684003152089e-06,
      "loss": 0.0072,
      "step": 8500
    },
    {
      "epoch": 0.7579787234042553,
      "grad_norm": 0.40243503312801704,
      "learning_rate": 6.904058313632782e-06,
      "loss": 0.015,
      "step": 8550
    },
    {
      "epoch": 0.7624113475177305,
      "grad_norm": 0.0047908998463532726,
      "learning_rate": 6.879432624113476e-06,
      "loss": 0.0071,
      "step": 8600
    },
    {
      "epoch": 0.7668439716312057,
      "grad_norm": 0.0009125120021038514,
      "learning_rate": 6.854806934594169e-06,
      "loss": 0.0037,
      "step": 8650
    },
    {
      "epoch": 0.7712765957446809,
      "grad_norm": 0.08131393086136933,
      "learning_rate": 6.830181245074863e-06,
      "loss": 0.0047,
      "step": 8700
    },
    {
      "epoch": 0.775709219858156,
      "grad_norm": 0.0006308892137283422,
      "learning_rate": 6.8055555555555566e-06,
      "loss": 0.014,
      "step": 8750
    },
    {
      "epoch": 0.7801418439716312,
      "grad_norm": 0.0454133683781099,
      "learning_rate": 6.78092986603625e-06,
      "loss": 0.0143,
      "step": 8800
    },
    {
      "epoch": 0.7845744680851063,
      "grad_norm": 24.069321019826646,
      "learning_rate": 6.756304176516943e-06,
      "loss": 0.0069,
      "step": 8850
    },
    {
      "epoch": 0.7890070921985816,
      "grad_norm": 0.018348520956196407,
      "learning_rate": 6.731678486997636e-06,
      "loss": 0.008,
      "step": 8900
    },
    {
      "epoch": 0.7934397163120568,
      "grad_norm": 0.045503745490795604,
      "learning_rate": 6.707052797478329e-06,
      "loss": 0.0038,
      "step": 8950
    },
    {
      "epoch": 0.7978723404255319,
      "grad_norm": 0.2684381266709852,
      "learning_rate": 6.682427107959023e-06,
      "loss": 0.0026,
      "step": 9000
    },
    {
      "epoch": 0.8023049645390071,
      "grad_norm": 0.11829149058897026,
      "learning_rate": 6.657801418439717e-06,
      "loss": 0.0233,
      "step": 9050
    },
    {
      "epoch": 0.8067375886524822,
      "grad_norm": 0.16600459567065376,
      "learning_rate": 6.63317572892041e-06,
      "loss": 0.0149,
      "step": 9100
    },
    {
      "epoch": 0.8111702127659575,
      "grad_norm": 0.0005939769820284411,
      "learning_rate": 6.608550039401103e-06,
      "loss": 0.0232,
      "step": 9150
    },
    {
      "epoch": 0.8156028368794326,
      "grad_norm": 4.742621865403474,
      "learning_rate": 6.583924349881798e-06,
      "loss": 0.003,
      "step": 9200
    },
    {
      "epoch": 0.8200354609929078,
      "grad_norm": 0.0022694228849504507,
      "learning_rate": 6.559298660362491e-06,
      "loss": 0.0088,
      "step": 9250
    },
    {
      "epoch": 0.824468085106383,
      "grad_norm": 0.01267761477851963,
      "learning_rate": 6.534672970843184e-06,
      "loss": 0.0042,
      "step": 9300
    },
    {
      "epoch": 0.8289007092198581,
      "grad_norm": 6.0142319714349085e-05,
      "learning_rate": 6.510047281323878e-06,
      "loss": 0.0036,
      "step": 9350
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 0.029585797478583992,
      "learning_rate": 6.485914105594957e-06,
      "loss": 0.0089,
      "step": 9400
    },
    {
      "epoch": 0.8377659574468085,
      "grad_norm": 0.11418292860246061,
      "learning_rate": 6.461288416075651e-06,
      "loss": 0.0023,
      "step": 9450
    },
    {
      "epoch": 0.8421985815602837,
      "grad_norm": 0.49982011062420273,
      "learning_rate": 6.436662726556344e-06,
      "loss": 0.0044,
      "step": 9500
    },
    {
      "epoch": 0.8466312056737588,
      "grad_norm": 18.307836570245335,
      "learning_rate": 6.4120370370370375e-06,
      "loss": 0.0169,
      "step": 9550
    },
    {
      "epoch": 0.851063829787234,
      "grad_norm": 0.001667923555688216,
      "learning_rate": 6.387411347517731e-06,
      "loss": 0.0072,
      "step": 9600
    },
    {
      "epoch": 0.8554964539007093,
      "grad_norm": 0.07738383812395583,
      "learning_rate": 6.362785657998425e-06,
      "loss": 0.0088,
      "step": 9650
    },
    {
      "epoch": 0.8599290780141844,
      "grad_norm": 4.3471602630804025,
      "learning_rate": 6.3381599684791185e-06,
      "loss": 0.0092,
      "step": 9700
    },
    {
      "epoch": 0.8643617021276596,
      "grad_norm": 0.0011124382494966075,
      "learning_rate": 6.313534278959811e-06,
      "loss": 0.0053,
      "step": 9750
    },
    {
      "epoch": 0.8687943262411347,
      "grad_norm": 0.19437101421378722,
      "learning_rate": 6.288908589440504e-06,
      "loss": 0.0044,
      "step": 9800
    },
    {
      "epoch": 0.87322695035461,
      "grad_norm": 0.00870818269634644,
      "learning_rate": 6.264282899921198e-06,
      "loss": 0.0059,
      "step": 9850
    },
    {
      "epoch": 0.8776595744680851,
      "grad_norm": 0.42404259032680214,
      "learning_rate": 6.239657210401892e-06,
      "loss": 0.0141,
      "step": 9900
    },
    {
      "epoch": 0.8820921985815603,
      "grad_norm": 0.000391160720835517,
      "learning_rate": 6.215031520882585e-06,
      "loss": 0.0144,
      "step": 9950
    },
    {
      "epoch": 0.8865248226950354,
      "grad_norm": 0.07186189200249775,
      "learning_rate": 6.190405831363279e-06,
      "loss": 0.0087,
      "step": 10000
    },
    {
      "epoch": 0.8909574468085106,
      "grad_norm": 9.1428649766071,
      "learning_rate": 6.165780141843972e-06,
      "loss": 0.0141,
      "step": 10050
    },
    {
      "epoch": 0.8953900709219859,
      "grad_norm": 0.28709998082991656,
      "learning_rate": 6.141154452324666e-06,
      "loss": 0.0161,
      "step": 10100
    },
    {
      "epoch": 0.899822695035461,
      "grad_norm": 0.0012613353256104856,
      "learning_rate": 6.1165287628053595e-06,
      "loss": 0.0083,
      "step": 10150
    },
    {
      "epoch": 0.9042553191489362,
      "grad_norm": 0.00043243530861486094,
      "learning_rate": 6.091903073286053e-06,
      "loss": 0.0062,
      "step": 10200
    },
    {
      "epoch": 0.9086879432624113,
      "grad_norm": 0.0012885774516513972,
      "learning_rate": 6.067277383766746e-06,
      "loss": 0.0032,
      "step": 10250
    },
    {
      "epoch": 0.9131205673758865,
      "grad_norm": 0.13567630623602833,
      "learning_rate": 6.042651694247439e-06,
      "loss": 0.0047,
      "step": 10300
    },
    {
      "epoch": 0.9175531914893617,
      "grad_norm": 0.001507828461742134,
      "learning_rate": 6.018026004728132e-06,
      "loss": 0.0111,
      "step": 10350
    },
    {
      "epoch": 0.9219858156028369,
      "grad_norm": 28.02482049835188,
      "learning_rate": 5.993400315208826e-06,
      "loss": 0.013,
      "step": 10400
    },
    {
      "epoch": 0.9264184397163121,
      "grad_norm": 0.0005712245953407916,
      "learning_rate": 5.96877462568952e-06,
      "loss": 0.0074,
      "step": 10450
    },
    {
      "epoch": 0.9308510638297872,
      "grad_norm": 0.6880719016963764,
      "learning_rate": 5.944148936170213e-06,
      "loss": 0.0118,
      "step": 10500
    },
    {
      "epoch": 0.9352836879432624,
      "grad_norm": 0.0004923493172913049,
      "learning_rate": 5.919523246650906e-06,
      "loss": 0.0285,
      "step": 10550
    },
    {
      "epoch": 0.9397163120567376,
      "grad_norm": 0.0005338033588183963,
      "learning_rate": 5.8948975571316006e-06,
      "loss": 0.0171,
      "step": 10600
    },
    {
      "epoch": 0.9441489361702128,
      "grad_norm": 0.01769753140101656,
      "learning_rate": 5.8707643814026795e-06,
      "loss": 0.0077,
      "step": 10650
    },
    {
      "epoch": 0.9485815602836879,
      "grad_norm": 12.31398706121417,
      "learning_rate": 5.846138691883373e-06,
      "loss": 0.0098,
      "step": 10700
    },
    {
      "epoch": 0.9530141843971631,
      "grad_norm": 1.086003198625972,
      "learning_rate": 5.821513002364066e-06,
      "loss": 0.0099,
      "step": 10750
    },
    {
      "epoch": 0.9574468085106383,
      "grad_norm": 2.675868747143798,
      "learning_rate": 5.7968873128447604e-06,
      "loss": 0.0134,
      "step": 10800
    },
    {
      "epoch": 0.9618794326241135,
      "grad_norm": 0.0004962731547164742,
      "learning_rate": 5.772261623325454e-06,
      "loss": 0.0231,
      "step": 10850
    },
    {
      "epoch": 0.9663120567375887,
      "grad_norm": 0.11931707651459274,
      "learning_rate": 5.747635933806147e-06,
      "loss": 0.0031,
      "step": 10900
    },
    {
      "epoch": 0.9707446808510638,
      "grad_norm": 0.027268096455214143,
      "learning_rate": 5.7230102442868405e-06,
      "loss": 0.0114,
      "step": 10950
    },
    {
      "epoch": 0.975177304964539,
      "grad_norm": 0.00014594318076532122,
      "learning_rate": 5.698384554767534e-06,
      "loss": 0.0049,
      "step": 11000
    },
    {
      "epoch": 0.9796099290780141,
      "grad_norm": 20.25133896156815,
      "learning_rate": 5.673758865248228e-06,
      "loss": 0.0114,
      "step": 11050
    },
    {
      "epoch": 0.9840425531914894,
      "grad_norm": 6.232025239063323,
      "learning_rate": 5.649133175728921e-06,
      "loss": 0.0115,
      "step": 11100
    },
    {
      "epoch": 0.9884751773049646,
      "grad_norm": 19.42076091686018,
      "learning_rate": 5.624507486209614e-06,
      "loss": 0.0084,
      "step": 11150
    },
    {
      "epoch": 0.9929078014184397,
      "grad_norm": 1.603272060616664,
      "learning_rate": 5.599881796690307e-06,
      "loss": 0.008,
      "step": 11200
    },
    {
      "epoch": 0.9973404255319149,
      "grad_norm": 1.0847716076156708,
      "learning_rate": 5.575256107171001e-06,
      "loss": 0.0114,
      "step": 11250
    },
    {
      "epoch": 1.00177304964539,
      "grad_norm": 2.3633525863666684e-05,
      "learning_rate": 5.550630417651695e-06,
      "loss": 0.0043,
      "step": 11300
    },
    {
      "epoch": 1.0062056737588652,
      "grad_norm": 0.013210380349044053,
      "learning_rate": 5.526004728132388e-06,
      "loss": 0.0164,
      "step": 11350
    },
    {
      "epoch": 1.0106382978723405,
      "grad_norm": 12.943589109031448,
      "learning_rate": 5.5013790386130815e-06,
      "loss": 0.0028,
      "step": 11400
    },
    {
      "epoch": 1.0150709219858156,
      "grad_norm": 0.24737069186236824,
      "learning_rate": 5.476753349093775e-06,
      "loss": 0.0342,
      "step": 11450
    },
    {
      "epoch": 1.0195035460992907,
      "grad_norm": 8.077551697802006,
      "learning_rate": 5.452127659574469e-06,
      "loss": 0.0069,
      "step": 11500
    },
    {
      "epoch": 1.023936170212766,
      "grad_norm": 21.99730319053338,
      "learning_rate": 5.4275019700551625e-06,
      "loss": 0.0077,
      "step": 11550
    },
    {
      "epoch": 1.0283687943262412,
      "grad_norm": 0.012495400608913515,
      "learning_rate": 5.402876280535856e-06,
      "loss": 0.0082,
      "step": 11600
    },
    {
      "epoch": 1.0328014184397163,
      "grad_norm": 23.172233223798802,
      "learning_rate": 5.378250591016549e-06,
      "loss": 0.0134,
      "step": 11650
    },
    {
      "epoch": 1.0372340425531914,
      "grad_norm": 0.00010344838148590977,
      "learning_rate": 5.354117415287628e-06,
      "loss": 0.0077,
      "step": 11700
    },
    {
      "epoch": 1.0416666666666667,
      "grad_norm": 0.026233233489346697,
      "learning_rate": 5.329491725768322e-06,
      "loss": 0.0064,
      "step": 11750
    },
    {
      "epoch": 1.0460992907801419,
      "grad_norm": 0.2523227021500782,
      "learning_rate": 5.304866036249016e-06,
      "loss": 0.0051,
      "step": 11800
    },
    {
      "epoch": 1.050531914893617,
      "grad_norm": 0.0036641836500692503,
      "learning_rate": 5.280240346729709e-06,
      "loss": 0.0096,
      "step": 11850
    },
    {
      "epoch": 1.0549645390070923,
      "grad_norm": 0.08388736022952638,
      "learning_rate": 5.255614657210402e-06,
      "loss": 0.0146,
      "step": 11900
    },
    {
      "epoch": 1.0593971631205674,
      "grad_norm": 0.7881169460771477,
      "learning_rate": 5.230988967691097e-06,
      "loss": 0.0048,
      "step": 11950
    },
    {
      "epoch": 1.0638297872340425,
      "grad_norm": 0.001005287973612562,
      "learning_rate": 5.206363278171789e-06,
      "loss": 0.0059,
      "step": 12000
    },
    {
      "epoch": 1.0682624113475176,
      "grad_norm": 3.0530614403128533,
      "learning_rate": 5.1817375886524825e-06,
      "loss": 0.0031,
      "step": 12050
    },
    {
      "epoch": 1.072695035460993,
      "grad_norm": 0.005283256557374208,
      "learning_rate": 5.157111899133176e-06,
      "loss": 0.0121,
      "step": 12100
    },
    {
      "epoch": 1.077127659574468,
      "grad_norm": 0.005667942165875843,
      "learning_rate": 5.132486209613869e-06,
      "loss": 0.0058,
      "step": 12150
    },
    {
      "epoch": 1.0815602836879432,
      "grad_norm": 0.012103555924136282,
      "learning_rate": 5.107860520094563e-06,
      "loss": 0.0022,
      "step": 12200
    },
    {
      "epoch": 1.0859929078014185,
      "grad_norm": 0.0015274237471755395,
      "learning_rate": 5.083234830575257e-06,
      "loss": 0.0043,
      "step": 12250
    },
    {
      "epoch": 1.0904255319148937,
      "grad_norm": 0.0017057462918492119,
      "learning_rate": 5.05860914105595e-06,
      "loss": 0.0055,
      "step": 12300
    },
    {
      "epoch": 1.0948581560283688,
      "grad_norm": 0.0678202349342981,
      "learning_rate": 5.0339834515366434e-06,
      "loss": 0.0208,
      "step": 12350
    },
    {
      "epoch": 1.099290780141844,
      "grad_norm": 5.820093588584535,
      "learning_rate": 5.009357762017338e-06,
      "loss": 0.0071,
      "step": 12400
    },
    {
      "epoch": 1.1037234042553192,
      "grad_norm": 0.0002967902068188711,
      "learning_rate": 4.98473207249803e-06,
      "loss": 0.0279,
      "step": 12450
    },
    {
      "epoch": 1.1081560283687943,
      "grad_norm": 0.00016505633598070985,
      "learning_rate": 4.9601063829787235e-06,
      "loss": 0.0079,
      "step": 12500
    },
    {
      "epoch": 1.1125886524822695,
      "grad_norm": 8.265879214198836e-05,
      "learning_rate": 4.935480693459418e-06,
      "loss": 0.0119,
      "step": 12550
    },
    {
      "epoch": 1.1170212765957448,
      "grad_norm": 5.33596270040032,
      "learning_rate": 4.910855003940111e-06,
      "loss": 0.0079,
      "step": 12600
    },
    {
      "epoch": 1.12145390070922,
      "grad_norm": 0.004774018567874852,
      "learning_rate": 4.886229314420804e-06,
      "loss": 0.0127,
      "step": 12650
    },
    {
      "epoch": 1.125886524822695,
      "grad_norm": 0.04021199029597082,
      "learning_rate": 4.861603624901498e-06,
      "loss": 0.0139,
      "step": 12700
    },
    {
      "epoch": 1.1303191489361701,
      "grad_norm": 11.357779746350145,
      "learning_rate": 4.836977935382191e-06,
      "loss": 0.0019,
      "step": 12750
    },
    {
      "epoch": 1.1347517730496455,
      "grad_norm": 0.0002728474474517395,
      "learning_rate": 4.8123522458628845e-06,
      "loss": 0.0054,
      "step": 12800
    },
    {
      "epoch": 1.1391843971631206,
      "grad_norm": 0.0015642870272121851,
      "learning_rate": 4.787726556343578e-06,
      "loss": 0.0021,
      "step": 12850
    },
    {
      "epoch": 1.1436170212765957,
      "grad_norm": 0.19008505520492977,
      "learning_rate": 4.763100866824271e-06,
      "loss": 0.0012,
      "step": 12900
    },
    {
      "epoch": 1.148049645390071,
      "grad_norm": 5.237984668093131,
      "learning_rate": 4.7384751773049646e-06,
      "loss": 0.0028,
      "step": 12950
    },
    {
      "epoch": 1.1524822695035462,
      "grad_norm": 0.004434596646202779,
      "learning_rate": 4.713849487785658e-06,
      "loss": 0.0043,
      "step": 13000
    },
    {
      "epoch": 1.1569148936170213,
      "grad_norm": 0.005018009875278763,
      "learning_rate": 4.689716312056738e-06,
      "loss": 0.0142,
      "step": 13050
    },
    {
      "epoch": 1.1613475177304964,
      "grad_norm": 0.0002677238680507841,
      "learning_rate": 4.665090622537432e-06,
      "loss": 0.0026,
      "step": 13100
    },
    {
      "epoch": 1.1657801418439717,
      "grad_norm": 0.000470140748192606,
      "learning_rate": 4.640464933018125e-06,
      "loss": 0.0092,
      "step": 13150
    },
    {
      "epoch": 1.1702127659574468,
      "grad_norm": 0.0018058005755851082,
      "learning_rate": 4.615839243498818e-06,
      "loss": 0.0023,
      "step": 13200
    },
    {
      "epoch": 1.174645390070922,
      "grad_norm": 0.04680604382760653,
      "learning_rate": 4.591213553979512e-06,
      "loss": 0.0058,
      "step": 13250
    },
    {
      "epoch": 1.1790780141843973,
      "grad_norm": 0.044852503825975776,
      "learning_rate": 4.566587864460205e-06,
      "loss": 0.0039,
      "step": 13300
    },
    {
      "epoch": 1.1835106382978724,
      "grad_norm": 11.036633612390176,
      "learning_rate": 4.541962174940899e-06,
      "loss": 0.0022,
      "step": 13350
    },
    {
      "epoch": 1.1879432624113475,
      "grad_norm": 0.09774370568985054,
      "learning_rate": 4.517336485421592e-06,
      "loss": 0.0024,
      "step": 13400
    },
    {
      "epoch": 1.1923758865248226,
      "grad_norm": 0.0004694848374047449,
      "learning_rate": 4.492710795902286e-06,
      "loss": 0.0088,
      "step": 13450
    },
    {
      "epoch": 1.196808510638298,
      "grad_norm": 0.023902851668442845,
      "learning_rate": 4.468085106382979e-06,
      "loss": 0.0114,
      "step": 13500
    },
    {
      "epoch": 1.201241134751773,
      "grad_norm": 0.05315316107757854,
      "learning_rate": 4.443459416863672e-06,
      "loss": 0.0156,
      "step": 13550
    },
    {
      "epoch": 1.2056737588652482,
      "grad_norm": 1.358682872795046,
      "learning_rate": 4.418833727344366e-06,
      "loss": 0.0035,
      "step": 13600
    },
    {
      "epoch": 1.2101063829787235,
      "grad_norm": 0.004028994195015635,
      "learning_rate": 4.39420803782506e-06,
      "loss": 0.0078,
      "step": 13650
    },
    {
      "epoch": 1.2145390070921986,
      "grad_norm": 2.60025465642063e-05,
      "learning_rate": 4.369582348305753e-06,
      "loss": 0.0031,
      "step": 13700
    },
    {
      "epoch": 1.2189716312056738,
      "grad_norm": 0.15044035027572922,
      "learning_rate": 4.344956658786446e-06,
      "loss": 0.0073,
      "step": 13750
    },
    {
      "epoch": 1.2234042553191489,
      "grad_norm": 0.029395537343344148,
      "learning_rate": 4.32033096926714e-06,
      "loss": 0.0092,
      "step": 13800
    },
    {
      "epoch": 1.2278368794326242,
      "grad_norm": 0.00013265795388455031,
      "learning_rate": 4.295705279747833e-06,
      "loss": 0.002,
      "step": 13850
    },
    {
      "epoch": 1.2322695035460993,
      "grad_norm": 4.134334983257184,
      "learning_rate": 4.2710795902285265e-06,
      "loss": 0.0059,
      "step": 13900
    },
    {
      "epoch": 1.2367021276595744,
      "grad_norm": 2.4040766603197826e-05,
      "learning_rate": 4.246453900709221e-06,
      "loss": 0.009,
      "step": 13950
    },
    {
      "epoch": 1.2411347517730495,
      "grad_norm": 5.73501558270658e-05,
      "learning_rate": 4.2223207249803e-06,
      "loss": 0.0334,
      "step": 14000
    },
    {
      "epoch": 1.2455673758865249,
      "grad_norm": 0.0006514599750282896,
      "learning_rate": 4.197695035460993e-06,
      "loss": 0.004,
      "step": 14050
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.3366246965865256e-05,
      "learning_rate": 4.173069345941686e-06,
      "loss": 0.006,
      "step": 14100
    },
    {
      "epoch": 1.2544326241134751,
      "grad_norm": 0.031237656606814084,
      "learning_rate": 4.1484436564223805e-06,
      "loss": 0.0057,
      "step": 14150
    },
    {
      "epoch": 1.2588652482269502,
      "grad_norm": 0.3490867651221054,
      "learning_rate": 4.123817966903074e-06,
      "loss": 0.004,
      "step": 14200
    },
    {
      "epoch": 1.2632978723404256,
      "grad_norm": 0.003986709752659374,
      "learning_rate": 4.099192277383767e-06,
      "loss": 0.009,
      "step": 14250
    },
    {
      "epoch": 1.2677304964539007,
      "grad_norm": 0.041880938463403676,
      "learning_rate": 4.074566587864461e-06,
      "loss": 0.0018,
      "step": 14300
    },
    {
      "epoch": 1.272163120567376,
      "grad_norm": 0.13043894056028277,
      "learning_rate": 4.049940898345154e-06,
      "loss": 0.0089,
      "step": 14350
    },
    {
      "epoch": 1.2765957446808511,
      "grad_norm": 20.525739931994153,
      "learning_rate": 4.025315208825847e-06,
      "loss": 0.0038,
      "step": 14400
    },
    {
      "epoch": 1.2810283687943262,
      "grad_norm": 0.03957694418665241,
      "learning_rate": 4.000689519306541e-06,
      "loss": 0.0047,
      "step": 14450
    },
    {
      "epoch": 1.2854609929078014,
      "grad_norm": 9.270907908158689e-06,
      "learning_rate": 3.976063829787235e-06,
      "loss": 0.0085,
      "step": 14500
    },
    {
      "epoch": 1.2898936170212765,
      "grad_norm": 0.019093848993790852,
      "learning_rate": 3.951438140267928e-06,
      "loss": 0.0073,
      "step": 14550
    },
    {
      "epoch": 1.2943262411347518,
      "grad_norm": 0.000544650767309452,
      "learning_rate": 3.926812450748621e-06,
      "loss": 0.004,
      "step": 14600
    },
    {
      "epoch": 1.298758865248227,
      "grad_norm": 8.507651438455978e-05,
      "learning_rate": 3.902186761229315e-06,
      "loss": 0.0149,
      "step": 14650
    },
    {
      "epoch": 1.3031914893617023,
      "grad_norm": 0.00015750838935733676,
      "learning_rate": 3.877561071710008e-06,
      "loss": 0.0024,
      "step": 14700
    },
    {
      "epoch": 1.3076241134751774,
      "grad_norm": 2.1216331063148686,
      "learning_rate": 3.852935382190702e-06,
      "loss": 0.0027,
      "step": 14750
    },
    {
      "epoch": 1.3120567375886525,
      "grad_norm": 0.0865583346196146,
      "learning_rate": 3.828309692671395e-06,
      "loss": 0.0073,
      "step": 14800
    },
    {
      "epoch": 1.3164893617021276,
      "grad_norm": 0.010228577051675729,
      "learning_rate": 3.8036840031520884e-06,
      "loss": 0.0127,
      "step": 14850
    },
    {
      "epoch": 1.3209219858156027,
      "grad_norm": 0.0005275434137609081,
      "learning_rate": 3.7790583136327817e-06,
      "loss": 0.0016,
      "step": 14900
    },
    {
      "epoch": 1.325354609929078,
      "grad_norm": 0.0030741880563327027,
      "learning_rate": 3.7544326241134755e-06,
      "loss": 0.0111,
      "step": 14950
    },
    {
      "epoch": 1.3297872340425532,
      "grad_norm": 0.4322702812359092,
      "learning_rate": 3.729806934594169e-06,
      "loss": 0.0108,
      "step": 15000
    },
    {
      "epoch": 1.3342198581560285,
      "grad_norm": 16.680559293669063,
      "learning_rate": 3.7051812450748626e-06,
      "loss": 0.0082,
      "step": 15050
    },
    {
      "epoch": 1.3386524822695036,
      "grad_norm": 0.0004440540435326648,
      "learning_rate": 3.680555555555556e-06,
      "loss": 0.008,
      "step": 15100
    },
    {
      "epoch": 1.3430851063829787,
      "grad_norm": 0.0004015706606442299,
      "learning_rate": 3.655929866036249e-06,
      "loss": 0.003,
      "step": 15150
    },
    {
      "epoch": 1.3475177304964538,
      "grad_norm": 0.00018125937118334472,
      "learning_rate": 3.6313041765169427e-06,
      "loss": 0.0009,
      "step": 15200
    },
    {
      "epoch": 1.351950354609929,
      "grad_norm": 0.07848257548289153,
      "learning_rate": 3.606678486997636e-06,
      "loss": 0.002,
      "step": 15250
    },
    {
      "epoch": 1.3563829787234043,
      "grad_norm": 0.023025301609373643,
      "learning_rate": 3.58205279747833e-06,
      "loss": 0.0031,
      "step": 15300
    },
    {
      "epoch": 1.3608156028368794,
      "grad_norm": 4.517888013582396,
      "learning_rate": 3.557427107959023e-06,
      "loss": 0.004,
      "step": 15350
    },
    {
      "epoch": 1.3652482269503547,
      "grad_norm": 1.2658270686339672,
      "learning_rate": 3.5328014184397165e-06,
      "loss": 0.0079,
      "step": 15400
    },
    {
      "epoch": 1.3696808510638299,
      "grad_norm": 0.004071845591775745,
      "learning_rate": 3.50817572892041e-06,
      "loss": 0.0055,
      "step": 15450
    },
    {
      "epoch": 1.374113475177305,
      "grad_norm": 0.0002310397513515275,
      "learning_rate": 3.4835500394011037e-06,
      "loss": 0.0034,
      "step": 15500
    },
    {
      "epoch": 1.37854609929078,
      "grad_norm": 0.019889125511476966,
      "learning_rate": 3.458924349881797e-06,
      "loss": 0.0089,
      "step": 15550
    },
    {
      "epoch": 1.3829787234042552,
      "grad_norm": 0.12069528532648284,
      "learning_rate": 3.4342986603624904e-06,
      "loss": 0.0103,
      "step": 15600
    },
    {
      "epoch": 1.3874113475177305,
      "grad_norm": 2.221152066469194,
      "learning_rate": 3.409672970843184e-06,
      "loss": 0.0115,
      "step": 15650
    },
    {
      "epoch": 1.3918439716312057,
      "grad_norm": 0.21584041208425908,
      "learning_rate": 3.385047281323877e-06,
      "loss": 0.0024,
      "step": 15700
    },
    {
      "epoch": 1.3962765957446808,
      "grad_norm": 0.07962326798958183,
      "learning_rate": 3.360421591804571e-06,
      "loss": 0.0004,
      "step": 15750
    },
    {
      "epoch": 1.400709219858156,
      "grad_norm": 0.0002935001100762825,
      "learning_rate": 3.3357959022852642e-06,
      "loss": 0.0108,
      "step": 15800
    },
    {
      "epoch": 1.4051418439716312,
      "grad_norm": 9.58127799729887,
      "learning_rate": 3.311170212765958e-06,
      "loss": 0.0013,
      "step": 15850
    },
    {
      "epoch": 1.4095744680851063,
      "grad_norm": 0.013975459324715883,
      "learning_rate": 3.2865445232466514e-06,
      "loss": 0.006,
      "step": 15900
    },
    {
      "epoch": 1.4140070921985815,
      "grad_norm": 0.0003136739495490558,
      "learning_rate": 3.2619188337273443e-06,
      "loss": 0.0067,
      "step": 15950
    },
    {
      "epoch": 1.4184397163120568,
      "grad_norm": 0.23057539945365355,
      "learning_rate": 3.237293144208038e-06,
      "loss": 0.0029,
      "step": 16000
    },
    {
      "epoch": 1.422872340425532,
      "grad_norm": 0.11630128906398461,
      "learning_rate": 3.2126674546887314e-06,
      "loss": 0.0011,
      "step": 16050
    },
    {
      "epoch": 1.427304964539007,
      "grad_norm": 1.9025801682092172,
      "learning_rate": 3.1880417651694252e-06,
      "loss": 0.0008,
      "step": 16100
    },
    {
      "epoch": 1.4317375886524824,
      "grad_norm": 0.9957743413096815,
      "learning_rate": 3.1634160756501186e-06,
      "loss": 0.0006,
      "step": 16150
    },
    {
      "epoch": 1.4361702127659575,
      "grad_norm": 4.523729944492489e-05,
      "learning_rate": 3.1392828999211984e-06,
      "loss": 0.0064,
      "step": 16200
    },
    {
      "epoch": 1.4406028368794326,
      "grad_norm": 0.017807506139140707,
      "learning_rate": 3.1146572104018913e-06,
      "loss": 0.0028,
      "step": 16250
    },
    {
      "epoch": 1.4450354609929077,
      "grad_norm": 0.2936267584110255,
      "learning_rate": 3.0900315208825847e-06,
      "loss": 0.0218,
      "step": 16300
    },
    {
      "epoch": 1.449468085106383,
      "grad_norm": 2.201825023255903e-06,
      "learning_rate": 3.0654058313632784e-06,
      "loss": 0.0089,
      "step": 16350
    },
    {
      "epoch": 1.4539007092198581,
      "grad_norm": 0.0020967040078515953,
      "learning_rate": 3.040780141843972e-06,
      "loss": 0.0011,
      "step": 16400
    },
    {
      "epoch": 1.4583333333333333,
      "grad_norm": 0.0035587730480887963,
      "learning_rate": 3.0161544523246656e-06,
      "loss": 0.0061,
      "step": 16450
    },
    {
      "epoch": 1.4627659574468086,
      "grad_norm": 0.00011908233856549432,
      "learning_rate": 2.9915287628053585e-06,
      "loss": 0.0104,
      "step": 16500
    },
    {
      "epoch": 1.4671985815602837,
      "grad_norm": 1.358266707504535,
      "learning_rate": 2.9669030732860523e-06,
      "loss": 0.0146,
      "step": 16550
    },
    {
      "epoch": 1.4716312056737588,
      "grad_norm": 0.0005173968272110369,
      "learning_rate": 2.9422773837667456e-06,
      "loss": 0.0072,
      "step": 16600
    },
    {
      "epoch": 1.476063829787234,
      "grad_norm": 5.0037899912089685,
      "learning_rate": 2.917651694247439e-06,
      "loss": 0.0077,
      "step": 16650
    },
    {
      "epoch": 1.4804964539007093,
      "grad_norm": 0.00036722887470731584,
      "learning_rate": 2.8930260047281328e-06,
      "loss": 0.022,
      "step": 16700
    },
    {
      "epoch": 1.4849290780141844,
      "grad_norm": 0.002660326684541599,
      "learning_rate": 2.868400315208826e-06,
      "loss": 0.0052,
      "step": 16750
    },
    {
      "epoch": 1.4893617021276595,
      "grad_norm": 0.004956960764106037,
      "learning_rate": 2.8437746256895195e-06,
      "loss": 0.0072,
      "step": 16800
    },
    {
      "epoch": 1.4937943262411348,
      "grad_norm": 1.0478071930407054,
      "learning_rate": 2.819148936170213e-06,
      "loss": 0.0037,
      "step": 16850
    },
    {
      "epoch": 1.49822695035461,
      "grad_norm": 0.032938488132694524,
      "learning_rate": 2.7945232466509066e-06,
      "loss": 0.003,
      "step": 16900
    },
    {
      "epoch": 1.502659574468085,
      "grad_norm": 0.04903975876178212,
      "learning_rate": 2.7698975571316e-06,
      "loss": 0.0061,
      "step": 16950
    },
    {
      "epoch": 1.5070921985815602,
      "grad_norm": 0.3263491645553027,
      "learning_rate": 2.7452718676122938e-06,
      "loss": 0.019,
      "step": 17000
    },
    {
      "epoch": 1.5115248226950353,
      "grad_norm": 2.5675365914148904,
      "learning_rate": 2.7206461780929867e-06,
      "loss": 0.0014,
      "step": 17050
    },
    {
      "epoch": 1.5159574468085106,
      "grad_norm": 0.0036317115050709474,
      "learning_rate": 2.69602048857368e-06,
      "loss": 0.0055,
      "step": 17100
    },
    {
      "epoch": 1.520390070921986,
      "grad_norm": 0.0009379009024431418,
      "learning_rate": 2.671394799054374e-06,
      "loss": 0.0169,
      "step": 17150
    },
    {
      "epoch": 1.524822695035461,
      "grad_norm": 0.0022970151160776,
      "learning_rate": 2.646769109535067e-06,
      "loss": 0.0053,
      "step": 17200
    },
    {
      "epoch": 1.5292553191489362,
      "grad_norm": 0.0019246470303829883,
      "learning_rate": 2.622143420015761e-06,
      "loss": 0.0025,
      "step": 17250
    },
    {
      "epoch": 1.5336879432624113,
      "grad_norm": 0.007413966936557272,
      "learning_rate": 2.5985027580772264e-06,
      "loss": 0.0113,
      "step": 17300
    },
    {
      "epoch": 1.5381205673758864,
      "grad_norm": 0.00024108723854638272,
      "learning_rate": 2.57387706855792e-06,
      "loss": 0.0056,
      "step": 17350
    },
    {
      "epoch": 1.5425531914893615,
      "grad_norm": 0.08798384755826176,
      "learning_rate": 2.549251379038613e-06,
      "loss": 0.0181,
      "step": 17400
    },
    {
      "epoch": 1.5469858156028369,
      "grad_norm": 0.17481763887643156,
      "learning_rate": 2.524625689519307e-06,
      "loss": 0.0056,
      "step": 17450
    },
    {
      "epoch": 1.5514184397163122,
      "grad_norm": 0.00012363471801080713,
      "learning_rate": 2.5e-06,
      "loss": 0.0048,
      "step": 17500
    },
    {
      "epoch": 1.5558510638297873,
      "grad_norm": 0.000281235141606495,
      "learning_rate": 2.4753743104806936e-06,
      "loss": 0.0041,
      "step": 17550
    },
    {
      "epoch": 1.5602836879432624,
      "grad_norm": 0.002074284717507087,
      "learning_rate": 2.450748620961387e-06,
      "loss": 0.0078,
      "step": 17600
    },
    {
      "epoch": 1.5647163120567376,
      "grad_norm": 0.011639759081317758,
      "learning_rate": 2.4261229314420807e-06,
      "loss": 0.001,
      "step": 17650
    },
    {
      "epoch": 1.5691489361702127,
      "grad_norm": 0.0002970237662211247,
      "learning_rate": 2.401497241922774e-06,
      "loss": 0.0107,
      "step": 17700
    },
    {
      "epoch": 1.5735815602836878,
      "grad_norm": 0.0007912621277405812,
      "learning_rate": 2.3768715524034674e-06,
      "loss": 0.0114,
      "step": 17750
    },
    {
      "epoch": 1.5780141843971631,
      "grad_norm": 2.5015768800092233,
      "learning_rate": 2.352245862884161e-06,
      "loss": 0.0131,
      "step": 17800
    },
    {
      "epoch": 1.5824468085106385,
      "grad_norm": 1.1611881483542175,
      "learning_rate": 2.327620173364854e-06,
      "loss": 0.0048,
      "step": 17850
    },
    {
      "epoch": 1.5868794326241136,
      "grad_norm": 8.345867902582887,
      "learning_rate": 2.302994483845548e-06,
      "loss": 0.0062,
      "step": 17900
    },
    {
      "epoch": 1.5913120567375887,
      "grad_norm": 2.226488828276802,
      "learning_rate": 2.2783687943262413e-06,
      "loss": 0.0033,
      "step": 17950
    },
    {
      "epoch": 1.5957446808510638,
      "grad_norm": 0.2250974162797621,
      "learning_rate": 2.2537431048069346e-06,
      "loss": 0.0039,
      "step": 18000
    },
    {
      "epoch": 1.600177304964539,
      "grad_norm": 3.541017240933167,
      "learning_rate": 2.2291174152876284e-06,
      "loss": 0.0025,
      "step": 18050
    },
    {
      "epoch": 1.604609929078014,
      "grad_norm": 0.0010147231973171634,
      "learning_rate": 2.2044917257683217e-06,
      "loss": 0.0014,
      "step": 18100
    },
    {
      "epoch": 1.6090425531914894,
      "grad_norm": 0.005109813507854219,
      "learning_rate": 2.179866036249015e-06,
      "loss": 0.0077,
      "step": 18150
    },
    {
      "epoch": 1.6134751773049647,
      "grad_norm": 0.5285122394100901,
      "learning_rate": 2.1552403467297085e-06,
      "loss": 0.003,
      "step": 18200
    },
    {
      "epoch": 1.6179078014184398,
      "grad_norm": 0.007852944630078929,
      "learning_rate": 2.130614657210402e-06,
      "loss": 0.0063,
      "step": 18250
    },
    {
      "epoch": 1.622340425531915,
      "grad_norm": 5.6094508415245195e-05,
      "learning_rate": 2.1059889676910956e-06,
      "loss": 0.004,
      "step": 18300
    },
    {
      "epoch": 1.62677304964539,
      "grad_norm": 0.5491008148980681,
      "learning_rate": 2.081363278171789e-06,
      "loss": 0.0055,
      "step": 18350
    },
    {
      "epoch": 1.6312056737588652,
      "grad_norm": 1.5525025008668276,
      "learning_rate": 2.0567375886524823e-06,
      "loss": 0.0021,
      "step": 18400
    },
    {
      "epoch": 1.6356382978723403,
      "grad_norm": 0.019464790999853782,
      "learning_rate": 2.032111899133176e-06,
      "loss": 0.0015,
      "step": 18450
    },
    {
      "epoch": 1.6400709219858156,
      "grad_norm": 0.03354862267681748,
      "learning_rate": 2.0074862096138694e-06,
      "loss": 0.0009,
      "step": 18500
    },
    {
      "epoch": 1.6445035460992907,
      "grad_norm": 3.661779589498576e-05,
      "learning_rate": 1.982860520094563e-06,
      "loss": 0.005,
      "step": 18550
    },
    {
      "epoch": 1.648936170212766,
      "grad_norm": 29.226511335493843,
      "learning_rate": 1.958234830575256e-06,
      "loss": 0.018,
      "step": 18600
    },
    {
      "epoch": 1.6533687943262412,
      "grad_norm": 2.6959490480503963,
      "learning_rate": 1.9336091410559495e-06,
      "loss": 0.0057,
      "step": 18650
    },
    {
      "epoch": 1.6578014184397163,
      "grad_norm": 0.3429448060734279,
      "learning_rate": 1.9089834515366433e-06,
      "loss": 0.0019,
      "step": 18700
    },
    {
      "epoch": 1.6622340425531914,
      "grad_norm": 1.3403285429948613e-05,
      "learning_rate": 1.8843577620173366e-06,
      "loss": 0.0025,
      "step": 18750
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.0006843325116012807,
      "learning_rate": 1.85973207249803e-06,
      "loss": 0.0083,
      "step": 18800
    },
    {
      "epoch": 1.6710992907801419,
      "grad_norm": 2.896950341714159e-05,
      "learning_rate": 1.8351063829787236e-06,
      "loss": 0.0047,
      "step": 18850
    },
    {
      "epoch": 1.675531914893617,
      "grad_norm": 0.03579736142023673,
      "learning_rate": 1.8104806934594171e-06,
      "loss": 0.0044,
      "step": 18900
    },
    {
      "epoch": 1.6799645390070923,
      "grad_norm": 0.0007634411592739593,
      "learning_rate": 1.7858550039401105e-06,
      "loss": 0.0074,
      "step": 18950
    },
    {
      "epoch": 1.6843971631205674,
      "grad_norm": 0.00025338095037660184,
      "learning_rate": 1.761229314420804e-06,
      "loss": 0.0028,
      "step": 19000
    },
    {
      "epoch": 1.6888297872340425,
      "grad_norm": 0.8367439016912624,
      "learning_rate": 1.7366036249014972e-06,
      "loss": 0.0123,
      "step": 19050
    },
    {
      "epoch": 1.6932624113475176,
      "grad_norm": 0.0009791356139938734,
      "learning_rate": 1.7119779353821908e-06,
      "loss": 0.0043,
      "step": 19100
    },
    {
      "epoch": 1.6976950354609928,
      "grad_norm": 1.1288708215981906,
      "learning_rate": 1.6873522458628843e-06,
      "loss": 0.0014,
      "step": 19150
    },
    {
      "epoch": 1.702127659574468,
      "grad_norm": 6.775118398159698,
      "learning_rate": 1.6627265563435777e-06,
      "loss": 0.0027,
      "step": 19200
    },
    {
      "epoch": 1.7065602836879432,
      "grad_norm": 0.07111263325664992,
      "learning_rate": 1.6381008668242713e-06,
      "loss": 0.0054,
      "step": 19250
    },
    {
      "epoch": 1.7109929078014185,
      "grad_norm": 5.746382572617145e-05,
      "learning_rate": 1.6134751773049648e-06,
      "loss": 0.0023,
      "step": 19300
    },
    {
      "epoch": 1.7154255319148937,
      "grad_norm": 0.0005896955942330902,
      "learning_rate": 1.588849487785658e-06,
      "loss": 0.0028,
      "step": 19350
    },
    {
      "epoch": 1.7198581560283688,
      "grad_norm": 0.7760810372694008,
      "learning_rate": 1.5642237982663515e-06,
      "loss": 0.0017,
      "step": 19400
    },
    {
      "epoch": 1.724290780141844,
      "grad_norm": 0.006697039214016149,
      "learning_rate": 1.5400906225374313e-06,
      "loss": 0.0187,
      "step": 19450
    },
    {
      "epoch": 1.728723404255319,
      "grad_norm": 0.00022903477785452336,
      "learning_rate": 1.5154649330181245e-06,
      "loss": 0.0004,
      "step": 19500
    },
    {
      "epoch": 1.7331560283687943,
      "grad_norm": 2.7535469066175384e-05,
      "learning_rate": 1.490839243498818e-06,
      "loss": 0.0049,
      "step": 19550
    },
    {
      "epoch": 1.7375886524822695,
      "grad_norm": 0.02985438719947455,
      "learning_rate": 1.4662135539795116e-06,
      "loss": 0.0044,
      "step": 19600
    },
    {
      "epoch": 1.7420212765957448,
      "grad_norm": 0.00019760588466390203,
      "learning_rate": 1.441587864460205e-06,
      "loss": 0.0142,
      "step": 19650
    },
    {
      "epoch": 1.74645390070922,
      "grad_norm": 3.047480283427747,
      "learning_rate": 1.4169621749408985e-06,
      "loss": 0.0071,
      "step": 19700
    },
    {
      "epoch": 1.750886524822695,
      "grad_norm": 0.0037448846335741332,
      "learning_rate": 1.3923364854215921e-06,
      "loss": 0.0031,
      "step": 19750
    },
    {
      "epoch": 1.7553191489361701,
      "grad_norm": 0.00020748441056905848,
      "learning_rate": 1.3677107959022853e-06,
      "loss": 0.0057,
      "step": 19800
    },
    {
      "epoch": 1.7597517730496453,
      "grad_norm": 0.00016023525645660029,
      "learning_rate": 1.3430851063829788e-06,
      "loss": 0.0126,
      "step": 19850
    },
    {
      "epoch": 1.7641843971631206,
      "grad_norm": 7.791162563353672e-06,
      "learning_rate": 1.3184594168636722e-06,
      "loss": 0.012,
      "step": 19900
    },
    {
      "epoch": 1.7686170212765957,
      "grad_norm": 0.9465404763088664,
      "learning_rate": 1.2938337273443658e-06,
      "loss": 0.0052,
      "step": 19950
    },
    {
      "epoch": 1.773049645390071,
      "grad_norm": 0.07414027193365252,
      "learning_rate": 1.2692080378250593e-06,
      "loss": 0.0019,
      "step": 20000
    },
    {
      "epoch": 1.7774822695035462,
      "grad_norm": 0.0055180504122557894,
      "learning_rate": 1.2445823483057527e-06,
      "loss": 0.0006,
      "step": 20050
    },
    {
      "epoch": 1.7819148936170213,
      "grad_norm": 0.0006138651547325419,
      "learning_rate": 1.2199566587864462e-06,
      "loss": 0.0031,
      "step": 20100
    },
    {
      "epoch": 1.7863475177304964,
      "grad_norm": 1.2899272619970438e-05,
      "learning_rate": 1.1953309692671396e-06,
      "loss": 0.0032,
      "step": 20150
    },
    {
      "epoch": 1.7907801418439715,
      "grad_norm": 15.895355390743765,
      "learning_rate": 1.170705279747833e-06,
      "loss": 0.0064,
      "step": 20200
    },
    {
      "epoch": 1.7952127659574468,
      "grad_norm": 0.0004655677253189525,
      "learning_rate": 1.1460795902285265e-06,
      "loss": 0.0045,
      "step": 20250
    },
    {
      "epoch": 1.799645390070922,
      "grad_norm": 4.462022318771812,
      "learning_rate": 1.1214539007092199e-06,
      "loss": 0.0051,
      "step": 20300
    },
    {
      "epoch": 1.8040780141843973,
      "grad_norm": 7.791263127884315,
      "learning_rate": 1.0968282111899134e-06,
      "loss": 0.0024,
      "step": 20350
    },
    {
      "epoch": 1.8085106382978724,
      "grad_norm": 0.001981428634821069,
      "learning_rate": 1.0722025216706068e-06,
      "loss": 0.0047,
      "step": 20400
    },
    {
      "epoch": 1.8129432624113475,
      "grad_norm": 0.9791286474652084,
      "learning_rate": 1.0480693459416864e-06,
      "loss": 0.0067,
      "step": 20450
    },
    {
      "epoch": 1.8173758865248226,
      "grad_norm": 4.49890759411509e-05,
      "learning_rate": 1.02344365642238e-06,
      "loss": 0.003,
      "step": 20500
    },
    {
      "epoch": 1.8218085106382977,
      "grad_norm": 0.0008872317773727661,
      "learning_rate": 9.988179669030735e-07,
      "loss": 0.0198,
      "step": 20550
    },
    {
      "epoch": 1.826241134751773,
      "grad_norm": 0.02413008880272575,
      "learning_rate": 9.741922773837669e-07,
      "loss": 0.0027,
      "step": 20600
    },
    {
      "epoch": 1.8306737588652482,
      "grad_norm": 0.0064929573762096335,
      "learning_rate": 9.495665878644602e-07,
      "loss": 0.0051,
      "step": 20650
    },
    {
      "epoch": 1.8351063829787235,
      "grad_norm": 2.0483772647244787,
      "learning_rate": 9.249408983451537e-07,
      "loss": 0.0024,
      "step": 20700
    },
    {
      "epoch": 1.8395390070921986,
      "grad_norm": 0.008048228817919003,
      "learning_rate": 9.003152088258473e-07,
      "loss": 0.0062,
      "step": 20750
    },
    {
      "epoch": 1.8439716312056738,
      "grad_norm": 0.0011709925053889282,
      "learning_rate": 8.756895193065406e-07,
      "loss": 0.0008,
      "step": 20800
    },
    {
      "epoch": 1.8484042553191489,
      "grad_norm": 2.6180834531581985e-05,
      "learning_rate": 8.510638297872341e-07,
      "loss": 0.0087,
      "step": 20850
    },
    {
      "epoch": 1.852836879432624,
      "grad_norm": 0.00019269222943037962,
      "learning_rate": 8.264381402679275e-07,
      "loss": 0.0045,
      "step": 20900
    },
    {
      "epoch": 1.8572695035460993,
      "grad_norm": 3.881137417498048,
      "learning_rate": 8.018124507486211e-07,
      "loss": 0.0055,
      "step": 20950
    },
    {
      "epoch": 1.8617021276595744,
      "grad_norm": 0.005418221882606532,
      "learning_rate": 7.771867612293145e-07,
      "loss": 0.0018,
      "step": 21000
    },
    {
      "epoch": 1.8661347517730498,
      "grad_norm": 1.07656255093645,
      "learning_rate": 7.525610717100079e-07,
      "loss": 0.0002,
      "step": 21050
    },
    {
      "epoch": 1.8705673758865249,
      "grad_norm": 9.912206549708396,
      "learning_rate": 7.279353821907014e-07,
      "loss": 0.0055,
      "step": 21100
    },
    {
      "epoch": 1.875,
      "grad_norm": 2.1879933703908734e-05,
      "learning_rate": 7.033096926713949e-07,
      "loss": 0.0033,
      "step": 21150
    },
    {
      "epoch": 1.8794326241134751,
      "grad_norm": 0.00016572161736391663,
      "learning_rate": 6.786840031520883e-07,
      "loss": 0.0177,
      "step": 21200
    },
    {
      "epoch": 1.8838652482269502,
      "grad_norm": 0.27069341578831,
      "learning_rate": 6.540583136327818e-07,
      "loss": 0.0121,
      "step": 21250
    }
  ],
  "logging_steps": 50,
  "max_steps": 22560,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 4250,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}