{
  "best_metric": 0.002970785601064563,
  "best_model_checkpoint": "sj-det-3/checkpoint-11988",
  "epoch": 4.0,
  "eval_steps": 500,
  "global_step": 11988,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.008341675008341674,
      "grad_norm": 14.153964042663574,
      "learning_rate": 1.0008340283569642e-06,
      "loss": 0.7821,
      "step": 25
    },
    {
      "epoch": 0.01668335001668335,
      "grad_norm": 11.691067695617676,
      "learning_rate": 2.0433694745621353e-06,
      "loss": 0.6237,
      "step": 50
    },
    {
      "epoch": 0.025025025025025027,
      "grad_norm": 6.019664764404297,
      "learning_rate": 3.0859049207673064e-06,
      "loss": 0.4143,
      "step": 75
    },
    {
      "epoch": 0.0333667000333667,
      "grad_norm": 2.7289156913757324,
      "learning_rate": 4.128440366972477e-06,
      "loss": 0.1794,
      "step": 100
    },
    {
      "epoch": 0.04170837504170837,
      "grad_norm": 0.4015839397907257,
      "learning_rate": 5.170975813177649e-06,
      "loss": 0.0836,
      "step": 125
    },
    {
      "epoch": 0.05005005005005005,
      "grad_norm": 0.41425979137420654,
      "learning_rate": 6.213511259382819e-06,
      "loss": 0.0735,
      "step": 150
    },
    {
      "epoch": 0.05839172505839173,
      "grad_norm": 0.04649379849433899,
      "learning_rate": 7.256046705587991e-06,
      "loss": 0.0196,
      "step": 175
    },
    {
      "epoch": 0.0667334000667334,
      "grad_norm": 0.03922989219427109,
      "learning_rate": 8.298582151793162e-06,
      "loss": 0.0703,
      "step": 200
    },
    {
      "epoch": 0.07507507507507508,
      "grad_norm": 0.09923669695854187,
      "learning_rate": 9.341117597998331e-06,
      "loss": 0.0382,
      "step": 225
    },
    {
      "epoch": 0.08341675008341674,
      "grad_norm": 0.41646531224250793,
      "learning_rate": 1.0383653044203504e-05,
      "loss": 0.0306,
      "step": 250
    },
    {
      "epoch": 0.09175842509175842,
      "grad_norm": 0.030469655990600586,
      "learning_rate": 1.1426188490408674e-05,
      "loss": 0.0672,
      "step": 275
    },
    {
      "epoch": 0.1001001001001001,
      "grad_norm": 6.139395713806152,
      "learning_rate": 1.2468723936613845e-05,
      "loss": 0.0419,
      "step": 300
    },
    {
      "epoch": 0.10844177510844177,
      "grad_norm": 1.2067941427230835,
      "learning_rate": 1.3511259382819017e-05,
      "loss": 0.0047,
      "step": 325
    },
    {
      "epoch": 0.11678345011678345,
      "grad_norm": 0.019348450005054474,
      "learning_rate": 1.4553794829024186e-05,
      "loss": 0.001,
      "step": 350
    },
    {
      "epoch": 0.12512512512512514,
      "grad_norm": 0.024063795804977417,
      "learning_rate": 1.559633027522936e-05,
      "loss": 0.048,
      "step": 375
    },
    {
      "epoch": 0.1334668001334668,
      "grad_norm": 0.01152120716869831,
      "learning_rate": 1.663886572143453e-05,
      "loss": 0.0275,
      "step": 400
    },
    {
      "epoch": 0.14180847514180847,
      "grad_norm": 0.007050159387290478,
      "learning_rate": 1.7681401167639698e-05,
      "loss": 0.0109,
      "step": 425
    },
    {
      "epoch": 0.15015015015015015,
      "grad_norm": 0.007726074196398258,
      "learning_rate": 1.8723936613844872e-05,
      "loss": 0.0007,
      "step": 450
    },
    {
      "epoch": 0.15849182515849183,
      "grad_norm": 0.008293859660625458,
      "learning_rate": 1.9766472060050043e-05,
      "loss": 0.028,
      "step": 475
    },
    {
      "epoch": 0.1668335001668335,
      "grad_norm": 0.0058615542948246,
      "learning_rate": 2.0809007506255214e-05,
      "loss": 0.0025,
      "step": 500
    },
    {
      "epoch": 0.17517517517517517,
      "grad_norm": 0.028854593634605408,
      "learning_rate": 2.1851542952460384e-05,
      "loss": 0.0685,
      "step": 525
    },
    {
      "epoch": 0.18351685018351685,
      "grad_norm": 0.042628563940525055,
      "learning_rate": 2.2894078398665555e-05,
      "loss": 0.0401,
      "step": 550
    },
    {
      "epoch": 0.19185852519185853,
      "grad_norm": 0.378325879573822,
      "learning_rate": 2.3936613844870726e-05,
      "loss": 0.0014,
      "step": 575
    },
    {
      "epoch": 0.2002002002002002,
      "grad_norm": 0.007751138880848885,
      "learning_rate": 2.49791492910759e-05,
      "loss": 0.0445,
      "step": 600
    },
    {
      "epoch": 0.20854187520854187,
      "grad_norm": 0.008357309736311436,
      "learning_rate": 2.602168473728107e-05,
      "loss": 0.0079,
      "step": 625
    },
    {
      "epoch": 0.21688355021688355,
      "grad_norm": 0.17902156710624695,
      "learning_rate": 2.7064220183486238e-05,
      "loss": 0.026,
      "step": 650
    },
    {
      "epoch": 0.22522522522522523,
      "grad_norm": 0.030053455382585526,
      "learning_rate": 2.810675562969141e-05,
      "loss": 0.1212,
      "step": 675
    },
    {
      "epoch": 0.2335669002335669,
      "grad_norm": 0.0116727901622653,
      "learning_rate": 2.9149291075896586e-05,
      "loss": 0.0459,
      "step": 700
    },
    {
      "epoch": 0.24190857524190856,
      "grad_norm": 0.005300463642925024,
      "learning_rate": 3.0191826522101753e-05,
      "loss": 0.0004,
      "step": 725
    },
    {
      "epoch": 0.2502502502502503,
      "grad_norm": 0.005538405384868383,
      "learning_rate": 3.1234361968306924e-05,
      "loss": 0.0205,
      "step": 750
    },
    {
      "epoch": 0.2585919252585919,
      "grad_norm": 0.009918354451656342,
      "learning_rate": 3.2276897414512094e-05,
      "loss": 0.0025,
      "step": 775
    },
    {
      "epoch": 0.2669336002669336,
      "grad_norm": 0.018012989312410355,
      "learning_rate": 3.3319432860717265e-05,
      "loss": 0.082,
      "step": 800
    },
    {
      "epoch": 0.2752752752752753,
      "grad_norm": 0.0018511965172365308,
      "learning_rate": 3.4361968306922436e-05,
      "loss": 0.0071,
      "step": 825
    },
    {
      "epoch": 0.28361695028361694,
      "grad_norm": 0.0468880757689476,
      "learning_rate": 3.540450375312761e-05,
      "loss": 0.0281,
      "step": 850
    },
    {
      "epoch": 0.29195862529195865,
      "grad_norm": 0.004004584159702063,
      "learning_rate": 3.644703919933278e-05,
      "loss": 0.0662,
      "step": 875
    },
    {
      "epoch": 0.3003003003003003,
      "grad_norm": 0.006573744583874941,
      "learning_rate": 3.748957464553795e-05,
      "loss": 0.0017,
      "step": 900
    },
    {
      "epoch": 0.30864197530864196,
      "grad_norm": 0.00465534208342433,
      "learning_rate": 3.8532110091743125e-05,
      "loss": 0.0159,
      "step": 925
    },
    {
      "epoch": 0.31698365031698367,
      "grad_norm": 0.6863465905189514,
      "learning_rate": 3.957464553794829e-05,
      "loss": 0.0037,
      "step": 950
    },
    {
      "epoch": 0.3253253253253253,
      "grad_norm": 0.0018059064168483019,
      "learning_rate": 4.0617180984153466e-05,
      "loss": 0.0001,
      "step": 975
    },
    {
      "epoch": 0.333667000333667,
      "grad_norm": 0.001327801262959838,
      "learning_rate": 4.165971643035864e-05,
      "loss": 0.0001,
      "step": 1000
    },
    {
      "epoch": 0.3420086753420087,
      "grad_norm": 0.0013706604950129986,
      "learning_rate": 4.27022518765638e-05,
      "loss": 0.0194,
      "step": 1025
    },
    {
      "epoch": 0.35035035035035034,
      "grad_norm": 0.0007294805254787207,
      "learning_rate": 4.374478732276898e-05,
      "loss": 0.0001,
      "step": 1050
    },
    {
      "epoch": 0.35869202535869205,
      "grad_norm": 0.018524976447224617,
      "learning_rate": 4.478732276897415e-05,
      "loss": 0.0481,
      "step": 1075
    },
    {
      "epoch": 0.3670337003670337,
      "grad_norm": 0.009606343694031239,
      "learning_rate": 4.582985821517932e-05,
      "loss": 0.0586,
      "step": 1100
    },
    {
      "epoch": 0.37537537537537535,
      "grad_norm": 0.0046248408034443855,
      "learning_rate": 4.687239366138449e-05,
      "loss": 0.0105,
      "step": 1125
    },
    {
      "epoch": 0.38371705038371706,
      "grad_norm": 0.017802894115447998,
      "learning_rate": 4.791492910758966e-05,
      "loss": 0.0019,
      "step": 1150
    },
    {
      "epoch": 0.3920587253920587,
      "grad_norm": 14.872435569763184,
      "learning_rate": 4.895746455379483e-05,
      "loss": 0.0849,
      "step": 1175
    },
    {
      "epoch": 0.4004004004004004,
      "grad_norm": 0.001058954163454473,
      "learning_rate": 5e-05,
      "loss": 0.0395,
      "step": 1200
    },
    {
      "epoch": 0.4087420754087421,
      "grad_norm": 0.0007220833795145154,
      "learning_rate": 4.988414125498193e-05,
      "loss": 0.0001,
      "step": 1225
    },
    {
      "epoch": 0.41708375041708373,
      "grad_norm": 0.0015725018456578255,
      "learning_rate": 4.9768282509963857e-05,
      "loss": 0.0,
      "step": 1250
    },
    {
      "epoch": 0.42542542542542544,
      "grad_norm": 0.00035438829218037426,
      "learning_rate": 4.965242376494578e-05,
      "loss": 0.0001,
      "step": 1275
    },
    {
      "epoch": 0.4337671004337671,
      "grad_norm": 0.004498378373682499,
      "learning_rate": 4.9536565019927704e-05,
      "loss": 0.0617,
      "step": 1300
    },
    {
      "epoch": 0.44210877544210875,
      "grad_norm": 0.0019637339282780886,
      "learning_rate": 4.9420706274909634e-05,
      "loss": 0.0452,
      "step": 1325
    },
    {
      "epoch": 0.45045045045045046,
      "grad_norm": 0.0009583417559042573,
      "learning_rate": 4.930484752989156e-05,
      "loss": 0.0004,
      "step": 1350
    },
    {
      "epoch": 0.4587921254587921,
      "grad_norm": 0.011658573523163795,
      "learning_rate": 4.918898878487349e-05,
      "loss": 0.0626,
      "step": 1375
    },
    {
      "epoch": 0.4671338004671338,
      "grad_norm": 0.002904647495597601,
      "learning_rate": 4.9073130039855406e-05,
      "loss": 0.0067,
      "step": 1400
    },
    {
      "epoch": 0.4754754754754755,
      "grad_norm": 0.004803840536624193,
      "learning_rate": 4.8957271294837336e-05,
      "loss": 0.0453,
      "step": 1425
    },
    {
      "epoch": 0.4838171504838171,
      "grad_norm": 0.0012352573685348034,
      "learning_rate": 4.8841412549819267e-05,
      "loss": 0.0003,
      "step": 1450
    },
    {
      "epoch": 0.49215882549215884,
      "grad_norm": 0.0022556546609848738,
      "learning_rate": 4.872555380480119e-05,
      "loss": 0.0002,
      "step": 1475
    },
    {
      "epoch": 0.5005005005005005,
      "grad_norm": 0.01666790060698986,
      "learning_rate": 4.8609695059783114e-05,
      "loss": 0.0142,
      "step": 1500
    },
    {
      "epoch": 0.5088421755088421,
      "grad_norm": 0.0008659372106194496,
      "learning_rate": 4.849383631476504e-05,
      "loss": 0.0003,
      "step": 1525
    },
    {
      "epoch": 0.5171838505171839,
      "grad_norm": 0.009269647300243378,
      "learning_rate": 4.837797756974697e-05,
      "loss": 0.0001,
      "step": 1550
    },
    {
      "epoch": 0.5255255255255256,
      "grad_norm": 0.00042679469333961606,
      "learning_rate": 4.826211882472889e-05,
      "loss": 0.0,
      "step": 1575
    },
    {
      "epoch": 0.5338672005338672,
      "grad_norm": 0.0003430478391237557,
      "learning_rate": 4.814626007971082e-05,
      "loss": 0.0,
      "step": 1600
    },
    {
      "epoch": 0.5422088755422089,
      "grad_norm": 0.002377784810960293,
      "learning_rate": 4.8030401334692746e-05,
      "loss": 0.0926,
      "step": 1625
    },
    {
      "epoch": 0.5505505505505506,
      "grad_norm": 0.003815292613580823,
      "learning_rate": 4.791454258967467e-05,
      "loss": 0.0001,
      "step": 1650
    },
    {
      "epoch": 0.5588922255588922,
      "grad_norm": 0.0007572752656415105,
      "learning_rate": 4.779868384465659e-05,
      "loss": 0.0001,
      "step": 1675
    },
    {
      "epoch": 0.5672339005672339,
      "grad_norm": 0.003480930346995592,
      "learning_rate": 4.7682825099638524e-05,
      "loss": 0.023,
      "step": 1700
    },
    {
      "epoch": 0.5755755755755756,
      "grad_norm": 0.0046263947151601315,
      "learning_rate": 4.7566966354620454e-05,
      "loss": 0.0219,
      "step": 1725
    },
    {
      "epoch": 0.5839172505839173,
      "grad_norm": 0.3202002942562103,
      "learning_rate": 4.745110760960237e-05,
      "loss": 0.0298,
      "step": 1750
    },
    {
      "epoch": 0.5922589255922589,
      "grad_norm": 6.295952320098877,
      "learning_rate": 4.73352488645843e-05,
      "loss": 0.0546,
      "step": 1775
    },
    {
      "epoch": 0.6006006006006006,
      "grad_norm": 0.08465034514665604,
      "learning_rate": 4.7219390119566225e-05,
      "loss": 0.0376,
      "step": 1800
    },
    {
      "epoch": 0.6089422756089423,
      "grad_norm": 0.003993997350335121,
      "learning_rate": 4.7103531374548156e-05,
      "loss": 0.0541,
      "step": 1825
    },
    {
      "epoch": 0.6172839506172839,
      "grad_norm": 2.8580386638641357,
      "learning_rate": 4.698767262953008e-05,
      "loss": 0.0475,
      "step": 1850
    },
    {
      "epoch": 0.6256256256256256,
      "grad_norm": 0.0002566488110460341,
      "learning_rate": 4.6871813884512e-05,
      "loss": 0.0155,
      "step": 1875
    },
    {
      "epoch": 0.6339673006339673,
      "grad_norm": 0.004632447846233845,
      "learning_rate": 4.675595513949393e-05,
      "loss": 0.0,
      "step": 1900
    },
    {
      "epoch": 0.6423089756423089,
      "grad_norm": 0.0001871311542345211,
      "learning_rate": 4.664009639447586e-05,
      "loss": 0.0,
      "step": 1925
    },
    {
      "epoch": 0.6506506506506506,
      "grad_norm": 0.0010421440238133073,
      "learning_rate": 4.652423764945779e-05,
      "loss": 0.0416,
      "step": 1950
    },
    {
      "epoch": 0.6589923256589924,
      "grad_norm": 0.005437193904072046,
      "learning_rate": 4.640837890443971e-05,
      "loss": 0.0564,
      "step": 1975
    },
    {
      "epoch": 0.667334000667334,
      "grad_norm": 0.0026402724906802177,
      "learning_rate": 4.6292520159421635e-05,
      "loss": 0.0209,
      "step": 2000
    },
    {
      "epoch": 0.6756756756756757,
      "grad_norm": 0.0003550160035956651,
      "learning_rate": 4.617666141440356e-05,
      "loss": 0.0012,
      "step": 2025
    },
    {
      "epoch": 0.6840173506840174,
      "grad_norm": 0.00020284796482883394,
      "learning_rate": 4.606080266938549e-05,
      "loss": 0.0002,
      "step": 2050
    },
    {
      "epoch": 0.6923590256923591,
      "grad_norm": 0.0005185810732655227,
      "learning_rate": 4.594494392436741e-05,
      "loss": 0.0001,
      "step": 2075
    },
    {
      "epoch": 0.7007007007007007,
      "grad_norm": 0.29189226031303406,
      "learning_rate": 4.582908517934934e-05,
      "loss": 0.0002,
      "step": 2100
    },
    {
      "epoch": 0.7090423757090424,
      "grad_norm": 0.006394149269908667,
      "learning_rate": 4.571322643433126e-05,
      "loss": 0.0,
      "step": 2125
    },
    {
      "epoch": 0.7173840507173841,
      "grad_norm": 0.003997980151325464,
      "learning_rate": 4.559736768931319e-05,
      "loss": 0.0431,
      "step": 2150
    },
    {
      "epoch": 0.7257257257257257,
      "grad_norm": 0.007415532600134611,
      "learning_rate": 4.548150894429512e-05,
      "loss": 0.0193,
      "step": 2175
    },
    {
      "epoch": 0.7340674007340674,
      "grad_norm": 0.0035364318173378706,
      "learning_rate": 4.5365650199277045e-05,
      "loss": 0.0631,
      "step": 2200
    },
    {
      "epoch": 0.7424090757424091,
      "grad_norm": 0.002615898149088025,
      "learning_rate": 4.524979145425897e-05,
      "loss": 0.0349,
      "step": 2225
    },
    {
      "epoch": 0.7507507507507507,
      "grad_norm": 0.0017047699075192213,
      "learning_rate": 4.513393270924089e-05,
      "loss": 0.021,
      "step": 2250
    },
    {
      "epoch": 0.7590924257590924,
      "grad_norm": 0.001677798223681748,
      "learning_rate": 4.501807396422282e-05,
      "loss": 0.0209,
      "step": 2275
    },
    {
      "epoch": 0.7674341007674341,
      "grad_norm": 0.0003330695035401732,
      "learning_rate": 4.490221521920475e-05,
      "loss": 0.0005,
      "step": 2300
    },
    {
      "epoch": 0.7757757757757757,
      "grad_norm": 9.45928695728071e-05,
      "learning_rate": 4.478635647418668e-05,
      "loss": 0.026,
      "step": 2325
    },
    {
      "epoch": 0.7841174507841174,
      "grad_norm": 6.650515556335449,
      "learning_rate": 4.46704977291686e-05,
      "loss": 0.018,
      "step": 2350
    },
    {
      "epoch": 0.7924591257924591,
      "grad_norm": 0.0031244370620697737,
      "learning_rate": 4.4554638984150525e-05,
      "loss": 0.0001,
      "step": 2375
    },
    {
      "epoch": 0.8008008008008008,
      "grad_norm": 0.0006237108027562499,
      "learning_rate": 4.443878023913245e-05,
      "loss": 0.0145,
      "step": 2400
    },
    {
      "epoch": 0.8091424758091424,
      "grad_norm": 0.0002782900701276958,
      "learning_rate": 4.432292149411438e-05,
      "loss": 0.0677,
      "step": 2425
    },
    {
      "epoch": 0.8174841508174842,
      "grad_norm": 0.0003246616688556969,
      "learning_rate": 4.420706274909631e-05,
      "loss": 0.0224,
      "step": 2450
    },
    {
      "epoch": 0.8258258258258259,
      "grad_norm": 0.0015066479099914432,
      "learning_rate": 4.4091204004078226e-05,
      "loss": 0.0112,
      "step": 2475
    },
    {
      "epoch": 0.8341675008341675,
      "grad_norm": 0.0002765478566288948,
      "learning_rate": 4.397534525906016e-05,
      "loss": 0.0078,
      "step": 2500
    },
    {
      "epoch": 0.8425091758425092,
      "grad_norm": 0.002668420784175396,
      "learning_rate": 4.385948651404208e-05,
      "loss": 0.0011,
      "step": 2525
    },
    {
      "epoch": 0.8508508508508509,
      "grad_norm": 0.00015687810082454234,
      "learning_rate": 4.374362776902401e-05,
      "loss": 0.0004,
      "step": 2550
    },
    {
      "epoch": 0.8591925258591925,
      "grad_norm": 1.4734420776367188,
      "learning_rate": 4.3627769024005935e-05,
      "loss": 0.0023,
      "step": 2575
    },
    {
      "epoch": 0.8675342008675342,
      "grad_norm": 0.0018520369194447994,
      "learning_rate": 4.351191027898786e-05,
      "loss": 0.0001,
      "step": 2600
    },
    {
      "epoch": 0.8758758758758759,
      "grad_norm": 0.012340963818132877,
      "learning_rate": 4.339605153396978e-05,
      "loss": 0.0488,
      "step": 2625
    },
    {
      "epoch": 0.8842175508842175,
      "grad_norm": 0.00010058472980745137,
      "learning_rate": 4.328019278895171e-05,
      "loss": 0.0123,
      "step": 2650
    },
    {
      "epoch": 0.8925592258925592,
      "grad_norm": 0.022818908095359802,
      "learning_rate": 4.316433404393364e-05,
      "loss": 0.1159,
      "step": 2675
    },
    {
      "epoch": 0.9009009009009009,
      "grad_norm": 0.006666264962404966,
      "learning_rate": 4.304847529891557e-05,
      "loss": 0.0008,
      "step": 2700
    },
    {
      "epoch": 0.9092425759092426,
      "grad_norm": 11.06509780883789,
      "learning_rate": 4.293261655389749e-05,
      "loss": 0.0061,
      "step": 2725
    },
    {
      "epoch": 0.9175842509175842,
      "grad_norm": 0.011369475163519382,
      "learning_rate": 4.2816757808879414e-05,
      "loss": 0.0413,
      "step": 2750
    },
    {
      "epoch": 0.9259259259259259,
      "grad_norm": 7.088267803192139,
      "learning_rate": 4.2700899063861345e-05,
      "loss": 0.0311,
      "step": 2775
    },
    {
      "epoch": 0.9342676009342676,
      "grad_norm": 4.856973171234131,
      "learning_rate": 4.258504031884327e-05,
      "loss": 0.0024,
      "step": 2800
    },
    {
      "epoch": 0.9426092759426092,
      "grad_norm": 0.0006875558174215257,
      "learning_rate": 4.246918157382519e-05,
      "loss": 0.0133,
      "step": 2825
    },
    {
      "epoch": 0.950950950950951,
      "grad_norm": 0.008509138599038124,
      "learning_rate": 4.2353322828807116e-05,
      "loss": 0.0,
      "step": 2850
    },
    {
      "epoch": 0.9592926259592927,
      "grad_norm": 0.0004129385342821479,
      "learning_rate": 4.2237464083789046e-05,
      "loss": 0.0025,
      "step": 2875
    },
    {
      "epoch": 0.9676343009676343,
      "grad_norm": 0.0008459310047328472,
      "learning_rate": 4.212160533877097e-05,
      "loss": 0.0,
      "step": 2900
    },
    {
      "epoch": 0.975975975975976,
      "grad_norm": 0.01643763668835163,
      "learning_rate": 4.20057465937529e-05,
      "loss": 0.0981,
      "step": 2925
    },
    {
      "epoch": 0.9843176509843177,
      "grad_norm": 0.0025746412575244904,
      "learning_rate": 4.1889887848734824e-05,
      "loss": 0.0006,
      "step": 2950
    },
    {
      "epoch": 0.9926593259926594,
      "grad_norm": 0.003589796368032694,
      "learning_rate": 4.177402910371675e-05,
      "loss": 0.0022,
      "step": 2975
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.995995995995996,
      "eval_auc": 0.9962769357195034,
      "eval_f1": 0.8518518518518519,
      "eval_loss": 0.014670311473309994,
      "eval_precision": 0.7840909090909091,
      "eval_recall": 0.9324324324324325,
      "eval_runtime": 78.4692,
      "eval_samples_per_second": 76.387,
      "eval_steps_per_second": 4.779,
      "step": 2997
    },
    {
      "epoch": 1.001001001001001,
      "grad_norm": 0.003225616877898574,
      "learning_rate": 4.165817035869868e-05,
      "loss": 0.0324,
      "step": 3000
    },
    {
      "epoch": 1.0093426760093427,
      "grad_norm": 0.23891094326972961,
      "learning_rate": 4.15423116136806e-05,
      "loss": 0.0233,
      "step": 3025
    },
    {
      "epoch": 1.0176843510176843,
      "grad_norm": 0.004559539258480072,
      "learning_rate": 4.142645286866253e-05,
      "loss": 0.0402,
      "step": 3050
    },
    {
      "epoch": 1.026026026026026,
      "grad_norm": 0.005625482648611069,
      "learning_rate": 4.131059412364445e-05,
      "loss": 0.0003,
      "step": 3075
    },
    {
      "epoch": 1.0343677010343677,
      "grad_norm": 0.004497454967349768,
      "learning_rate": 4.119473537862638e-05,
      "loss": 0.0004,
      "step": 3100
    },
    {
      "epoch": 1.0427093760427093,
      "grad_norm": 0.0791538655757904,
      "learning_rate": 4.1078876633608304e-05,
      "loss": 0.0057,
      "step": 3125
    },
    {
      "epoch": 1.0510510510510511,
      "grad_norm": 0.0016846450744196773,
      "learning_rate": 4.0963017888590234e-05,
      "loss": 0.0003,
      "step": 3150
    },
    {
      "epoch": 1.0593927260593927,
      "grad_norm": 0.0009143862989731133,
      "learning_rate": 4.0847159143572165e-05,
      "loss": 0.0001,
      "step": 3175
    },
    {
      "epoch": 1.0677344010677343,
      "grad_norm": 0.0010049792472273111,
      "learning_rate": 4.073130039855408e-05,
      "loss": 0.017,
      "step": 3200
    },
    {
      "epoch": 1.0760760760760761,
      "grad_norm": 0.003489920636638999,
      "learning_rate": 4.061544165353601e-05,
      "loss": 0.0001,
      "step": 3225
    },
    {
      "epoch": 1.0844177510844177,
      "grad_norm": 1.6741670370101929,
      "learning_rate": 4.0499582908517936e-05,
      "loss": 0.0237,
      "step": 3250
    },
    {
      "epoch": 1.0927594260927593,
      "grad_norm": 0.000581038068048656,
      "learning_rate": 4.0383724163499866e-05,
      "loss": 0.0001,
      "step": 3275
    },
    {
      "epoch": 1.1011011011011012,
      "grad_norm": 0.0003763669228646904,
      "learning_rate": 4.026786541848179e-05,
      "loss": 0.0001,
      "step": 3300
    },
    {
      "epoch": 1.1094427761094428,
      "grad_norm": 0.006402493920177221,
      "learning_rate": 4.0152006673463714e-05,
      "loss": 0.0147,
      "step": 3325
    },
    {
      "epoch": 1.1177844511177844,
      "grad_norm": 0.022542010992765427,
      "learning_rate": 4.003614792844564e-05,
      "loss": 0.0507,
      "step": 3350
    },
    {
      "epoch": 1.1261261261261262,
      "grad_norm": 0.0030998587608337402,
      "learning_rate": 3.992028918342757e-05,
      "loss": 0.0103,
      "step": 3375
    },
    {
      "epoch": 1.1344678011344678,
      "grad_norm": 0.001442295964807272,
      "learning_rate": 3.98044304384095e-05,
      "loss": 0.0003,
      "step": 3400
    },
    {
      "epoch": 1.1428094761428094,
      "grad_norm": 0.007275531068444252,
      "learning_rate": 3.968857169339142e-05,
      "loss": 0.0001,
      "step": 3425
    },
    {
      "epoch": 1.1511511511511512,
      "grad_norm": 0.000654282805044204,
      "learning_rate": 3.9572712948373346e-05,
      "loss": 0.0185,
      "step": 3450
    },
    {
      "epoch": 1.1594928261594928,
      "grad_norm": 0.016119500622153282,
      "learning_rate": 3.945685420335527e-05,
      "loss": 0.0052,
      "step": 3475
    },
    {
      "epoch": 1.1678345011678344,
      "grad_norm": 0.0006950986571609974,
      "learning_rate": 3.93409954583372e-05,
      "loss": 0.0002,
      "step": 3500
    },
    {
      "epoch": 1.1761761761761762,
      "grad_norm": 0.05980084836483002,
      "learning_rate": 3.9225136713319123e-05,
      "loss": 0.0001,
      "step": 3525
    },
    {
      "epoch": 1.1845178511845178,
      "grad_norm": 0.0003822998551186174,
      "learning_rate": 3.910927796830105e-05,
      "loss": 0.0011,
      "step": 3550
    },
    {
      "epoch": 1.1928595261928596,
      "grad_norm": 0.0003051291569136083,
      "learning_rate": 3.899341922328297e-05,
      "loss": 0.0004,
      "step": 3575
    },
    {
      "epoch": 1.2012012012012012,
      "grad_norm": 0.02143264003098011,
      "learning_rate": 3.88775604782649e-05,
      "loss": 0.0151,
      "step": 3600
    },
    {
      "epoch": 1.2095428762095428,
      "grad_norm": 0.012560728006064892,
      "learning_rate": 3.8761701733246825e-05,
      "loss": 0.1055,
      "step": 3625
    },
    {
      "epoch": 1.2178845512178846,
      "grad_norm": 0.06956183165311813,
      "learning_rate": 3.8645842988228756e-05,
      "loss": 0.0288,
      "step": 3650
    },
    {
      "epoch": 1.2262262262262262,
      "grad_norm": 0.019330767914652824,
      "learning_rate": 3.852998424321068e-05,
      "loss": 0.0564,
      "step": 3675
    },
    {
      "epoch": 1.2345679012345678,
      "grad_norm": 0.022493431344628334,
      "learning_rate": 3.84141254981926e-05,
      "loss": 0.0188,
      "step": 3700
    },
    {
      "epoch": 1.2429095762429097,
      "grad_norm": 0.0016129710711538792,
      "learning_rate": 3.8298266753174533e-05,
      "loss": 0.0001,
      "step": 3725
    },
    {
      "epoch": 1.2512512512512513,
      "grad_norm": 0.0006044628680683672,
      "learning_rate": 3.818240800815646e-05,
      "loss": 0.0001,
      "step": 3750
    },
    {
      "epoch": 1.2595929262595928,
      "grad_norm": 0.0004552229365799576,
      "learning_rate": 3.806654926313839e-05,
      "loss": 0.0005,
      "step": 3775
    },
    {
      "epoch": 1.2679346012679347,
      "grad_norm": 0.0007301606819964945,
      "learning_rate": 3.7950690518120305e-05,
      "loss": 0.0,
      "step": 3800
    },
    {
      "epoch": 1.2762762762762763,
      "grad_norm": 0.00021806442236993462,
      "learning_rate": 3.7834831773102235e-05,
      "loss": 0.0001,
      "step": 3825
    },
    {
      "epoch": 1.284617951284618,
      "grad_norm": 0.0004119027580600232,
      "learning_rate": 3.771897302808416e-05,
      "loss": 0.0,
      "step": 3850
    },
    {
      "epoch": 1.2929596262929597,
      "grad_norm": 0.0005425213603302836,
      "learning_rate": 3.760311428306609e-05,
      "loss": 0.0256,
      "step": 3875
    },
    {
      "epoch": 1.3013013013013013,
      "grad_norm": 0.0020909614395350218,
      "learning_rate": 3.748725553804801e-05,
      "loss": 0.0001,
      "step": 3900
    },
    {
      "epoch": 1.309642976309643,
      "grad_norm": 6.709009647369385,
      "learning_rate": 3.7371396793029937e-05,
      "loss": 0.0233,
      "step": 3925
    },
    {
      "epoch": 1.3179846513179847,
      "grad_norm": 0.00022848871594760567,
      "learning_rate": 3.725553804801187e-05,
      "loss": 0.0341,
      "step": 3950
    },
    {
      "epoch": 1.3263263263263263,
      "grad_norm": 0.00047850931878201663,
      "learning_rate": 3.713967930299379e-05,
      "loss": 0.0311,
      "step": 3975
    },
    {
      "epoch": 1.3346680013346681,
      "grad_norm": 0.0011013607727363706,
      "learning_rate": 3.702382055797572e-05,
      "loss": 0.0001,
      "step": 4000
    },
    {
      "epoch": 1.3430096763430097,
      "grad_norm": 0.03528917208313942,
      "learning_rate": 3.6907961812957645e-05,
      "loss": 0.0428,
      "step": 4025
    },
    {
      "epoch": 1.3513513513513513,
      "grad_norm": 0.0020049402955919504,
      "learning_rate": 3.679210306793957e-05,
      "loss": 0.001,
      "step": 4050
    },
    {
      "epoch": 1.3596930263596931,
      "grad_norm": 6.149662017822266,
      "learning_rate": 3.667624432292149e-05,
      "loss": 0.0331,
      "step": 4075
    },
    {
      "epoch": 1.3680347013680347,
      "grad_norm": 0.0008450220921076834,
      "learning_rate": 3.656038557790342e-05,
      "loss": 0.0138,
      "step": 4100
    },
    {
      "epoch": 1.3763763763763763,
      "grad_norm": 0.0007816566503606737,
      "learning_rate": 3.644452683288535e-05,
      "loss": 0.0239,
      "step": 4125
    },
    {
      "epoch": 1.3847180513847182,
      "grad_norm": 0.004917916841804981,
      "learning_rate": 3.632866808786728e-05,
      "loss": 0.0056,
      "step": 4150
    },
    {
      "epoch": 1.3930597263930597,
      "grad_norm": 0.00037450282252393663,
      "learning_rate": 3.62128093428492e-05,
      "loss": 0.0014,
      "step": 4175
    },
    {
      "epoch": 1.4014014014014013,
      "grad_norm": 0.0002487332094460726,
      "learning_rate": 3.6096950597831124e-05,
      "loss": 0.0081,
      "step": 4200
    },
    {
      "epoch": 1.4097430764097432,
      "grad_norm": 0.0002034240314969793,
      "learning_rate": 3.5981091852813055e-05,
      "loss": 0.0,
      "step": 4225
    },
    {
      "epoch": 1.4180847514180848,
      "grad_norm": 0.0002251726109534502,
      "learning_rate": 3.586523310779498e-05,
      "loss": 0.0376,
      "step": 4250
    },
    {
      "epoch": 1.4264264264264264,
      "grad_norm": 0.0068957749754190445,
      "learning_rate": 3.57493743627769e-05,
      "loss": 0.0547,
      "step": 4275
    },
    {
      "epoch": 1.4347681014347682,
      "grad_norm": 0.14848369359970093,
      "learning_rate": 3.5633515617758826e-05,
      "loss": 0.0398,
      "step": 4300
    },
    {
      "epoch": 1.4431097764431098,
      "grad_norm": 0.006616557948291302,
      "learning_rate": 3.5517656872740756e-05,
      "loss": 0.0009,
      "step": 4325
    },
    {
      "epoch": 1.4514514514514514,
      "grad_norm": 0.0264744870364666,
      "learning_rate": 3.540179812772268e-05,
      "loss": 0.0219,
      "step": 4350
    },
    {
      "epoch": 1.4597931264597932,
      "grad_norm": 0.011937894858419895,
      "learning_rate": 3.528593938270461e-05,
      "loss": 0.0284,
      "step": 4375
    },
    {
      "epoch": 1.4681348014681348,
      "grad_norm": 0.00034804438473656774,
      "learning_rate": 3.5170080637686534e-05,
      "loss": 0.0027,
      "step": 4400
    },
    {
      "epoch": 1.4764764764764764,
      "grad_norm": 0.00043386907782405615,
      "learning_rate": 3.505422189266846e-05,
      "loss": 0.0,
      "step": 4425
    },
    {
      "epoch": 1.4848181514848182,
      "grad_norm": 0.0003097046574112028,
      "learning_rate": 3.493836314765039e-05,
      "loss": 0.0005,
      "step": 4450
    },
    {
      "epoch": 1.4931598264931598,
      "grad_norm": 0.0003939344023820013,
      "learning_rate": 3.482250440263231e-05,
      "loss": 0.0002,
      "step": 4475
    },
    {
      "epoch": 1.5015015015015014,
      "grad_norm": 0.00022050646657589823,
      "learning_rate": 3.470664565761424e-05,
      "loss": 0.0001,
      "step": 4500
    },
    {
      "epoch": 1.5098431765098432,
      "grad_norm": 0.0003953372943215072,
      "learning_rate": 3.459078691259616e-05,
      "loss": 0.0001,
      "step": 4525
    },
    {
      "epoch": 1.5181848515181848,
      "grad_norm": 0.00029923071269877255,
      "learning_rate": 3.447492816757809e-05,
      "loss": 0.0408,
      "step": 4550
    },
    {
      "epoch": 1.5265265265265264,
      "grad_norm": 0.0014319141628220677,
      "learning_rate": 3.4359069422560014e-05,
      "loss": 0.0001,
      "step": 4575
    },
    {
      "epoch": 1.5348682015348682,
      "grad_norm": 0.0010346460621804,
      "learning_rate": 3.4243210677541944e-05,
      "loss": 0.016,
      "step": 4600
    },
    {
      "epoch": 1.5432098765432098,
      "grad_norm": 0.0012767055304720998,
      "learning_rate": 3.412735193252387e-05,
      "loss": 0.0001,
      "step": 4625
    },
    {
      "epoch": 1.5515515515515514,
      "grad_norm": 0.0006332869525067508,
      "learning_rate": 3.401149318750579e-05,
      "loss": 0.0001,
      "step": 4650
    },
    {
      "epoch": 1.5598932265598933,
      "grad_norm": 0.013449088670313358,
      "learning_rate": 3.389563444248772e-05,
      "loss": 0.0124,
      "step": 4675
    },
    {
      "epoch": 1.5682349015682349,
      "grad_norm": 0.0019397336291149259,
      "learning_rate": 3.3779775697469646e-05,
      "loss": 0.0488,
      "step": 4700
    },
    {
      "epoch": 1.5765765765765765,
      "grad_norm": 0.001804558327421546,
      "learning_rate": 3.3663916952451576e-05,
      "loss": 0.0001,
      "step": 4725
    },
    {
      "epoch": 1.5849182515849183,
      "grad_norm": 0.0007520993240177631,
      "learning_rate": 3.35480582074335e-05,
      "loss": 0.0005,
      "step": 4750
    },
    {
      "epoch": 1.5932599265932599,
      "grad_norm": 0.0034453764092177153,
      "learning_rate": 3.3432199462415424e-05,
      "loss": 0.0,
      "step": 4775
    },
    {
      "epoch": 1.6016016016016015,
      "grad_norm": 0.0007546308333985507,
      "learning_rate": 3.331634071739735e-05,
      "loss": 0.0284,
      "step": 4800
    },
    {
      "epoch": 1.6099432766099433,
      "grad_norm": 0.018241139128804207,
      "learning_rate": 3.320048197237928e-05,
      "loss": 0.0,
      "step": 4825
    },
    {
      "epoch": 1.6182849516182851,
      "grad_norm": 0.0001632855273783207,
      "learning_rate": 3.308462322736121e-05,
      "loss": 0.0007,
      "step": 4850
    },
    {
      "epoch": 1.6266266266266265,
      "grad_norm": 0.0043875593692064285,
      "learning_rate": 3.2968764482343125e-05,
      "loss": 0.0,
      "step": 4875
    },
    {
      "epoch": 1.6349683016349683,
      "grad_norm": 0.0005138323176652193,
      "learning_rate": 3.2852905737325056e-05,
      "loss": 0.0322,
      "step": 4900
    },
    {
      "epoch": 1.6433099766433101,
      "grad_norm": 0.0033361820969730616,
      "learning_rate": 3.273704699230698e-05,
      "loss": 0.0285,
      "step": 4925
    },
    {
      "epoch": 1.6516516516516515,
      "grad_norm": 0.0008180899312719703,
      "learning_rate": 3.262118824728891e-05,
      "loss": 0.0001,
      "step": 4950
    },
    {
      "epoch": 1.6599933266599933,
      "grad_norm": 0.0002278868923895061,
      "learning_rate": 3.2505329502270834e-05,
      "loss": 0.0008,
      "step": 4975
    },
    {
      "epoch": 1.6683350016683351,
      "grad_norm": 0.0010708553018048406,
      "learning_rate": 3.238947075725276e-05,
      "loss": 0.0001,
      "step": 5000
    },
    {
      "epoch": 1.6766766766766765,
      "grad_norm": 0.0006777086528018117,
      "learning_rate": 3.227361201223468e-05,
      "loss": 0.0,
      "step": 5025
    },
    {
      "epoch": 1.6850183516850183,
      "grad_norm": 0.0007990729645825922,
      "learning_rate": 3.215775326721661e-05,
      "loss": 0.0588,
      "step": 5050
    },
    {
      "epoch": 1.6933600266933602,
      "grad_norm": 0.00486621493473649,
      "learning_rate": 3.2041894522198535e-05,
      "loss": 0.0109,
      "step": 5075
    },
    {
      "epoch": 1.7017017017017015,
      "grad_norm": 0.020536229014396667,
      "learning_rate": 3.1926035777180466e-05,
      "loss": 0.0241,
      "step": 5100
    },
    {
      "epoch": 1.7100433767100434,
      "grad_norm": 0.003294334514066577,
      "learning_rate": 3.181017703216239e-05,
      "loss": 0.023,
      "step": 5125
    },
    {
      "epoch": 1.7183850517183852,
      "grad_norm": 0.0024423557333648205,
      "learning_rate": 3.169431828714431e-05,
      "loss": 0.0002,
      "step": 5150
    },
    {
      "epoch": 1.7267267267267268,
      "grad_norm": 0.010359351523220539,
      "learning_rate": 3.1578459542126244e-05,
      "loss": 0.0208,
      "step": 5175
    },
    {
      "epoch": 1.7350684017350684,
      "grad_norm": 0.008612546138465405,
      "learning_rate": 3.146260079710817e-05,
      "loss": 0.0002,
      "step": 5200
    },
    {
      "epoch": 1.7434100767434102,
      "grad_norm": 0.0005397327477112412,
      "learning_rate": 3.13467420520901e-05,
      "loss": 0.0078,
      "step": 5225
    },
    {
      "epoch": 1.7517517517517518,
      "grad_norm": 0.00102256890386343,
      "learning_rate": 3.1230883307072015e-05,
      "loss": 0.0,
      "step": 5250
    },
    {
      "epoch": 1.7600934267600934,
      "grad_norm": 0.004368093330413103,
      "learning_rate": 3.1115024562053945e-05,
      "loss": 0.0485,
      "step": 5275
    },
    {
      "epoch": 1.7684351017684352,
      "grad_norm": 0.0030503266025334597,
      "learning_rate": 3.099916581703587e-05,
      "loss": 0.0002,
      "step": 5300
    },
    {
      "epoch": 1.7767767767767768,
      "grad_norm": 0.005283331032842398,
      "learning_rate": 3.08833070720178e-05,
      "loss": 0.0003,
      "step": 5325
    },
    {
      "epoch": 1.7851184517851184,
      "grad_norm": 0.004458490759134293,
      "learning_rate": 3.077208267680045e-05,
      "loss": 0.0769,
      "step": 5350
    },
    {
      "epoch": 1.7934601267934602,
      "grad_norm": 0.015544002875685692,
      "learning_rate": 3.065622393178238e-05,
      "loss": 0.0005,
      "step": 5375
    },
    {
      "epoch": 1.8018018018018018,
      "grad_norm": 0.01679283380508423,
      "learning_rate": 3.0540365186764294e-05,
      "loss": 0.0006,
      "step": 5400
    },
    {
      "epoch": 1.8101434768101434,
      "grad_norm": 0.0037997430190443993,
      "learning_rate": 3.0424506441746225e-05,
      "loss": 0.0217,
      "step": 5425
    },
    {
      "epoch": 1.8184851518184852,
      "grad_norm": 0.003071027109399438,
      "learning_rate": 3.030864769672815e-05,
      "loss": 0.0001,
      "step": 5450
    },
    {
      "epoch": 1.8268268268268268,
      "grad_norm": 0.0016744754975661635,
      "learning_rate": 3.0192788951710076e-05,
      "loss": 0.0001,
      "step": 5475
    },
    {
      "epoch": 1.8351685018351684,
      "grad_norm": 0.019519317895174026,
      "learning_rate": 3.0076930206692e-05,
      "loss": 0.0474,
      "step": 5500
    },
    {
      "epoch": 1.8435101768435103,
      "grad_norm": 0.016507979482412338,
      "learning_rate": 2.996107146167393e-05,
      "loss": 0.0004,
      "step": 5525
    },
    {
      "epoch": 1.8518518518518519,
      "grad_norm": 0.002733981003984809,
      "learning_rate": 2.9845212716655857e-05,
      "loss": 0.0049,
      "step": 5550
    },
    {
      "epoch": 1.8601935268601935,
      "grad_norm": 0.002405300736427307,
      "learning_rate": 2.972935397163778e-05,
      "loss": 0.0012,
      "step": 5575
    },
    {
      "epoch": 1.8685352018685353,
      "grad_norm": 0.0028873877599835396,
      "learning_rate": 2.9613495226619708e-05,
      "loss": 0.0001,
      "step": 5600
    },
    {
      "epoch": 1.8768768768768769,
      "grad_norm": 0.0029900181107223034,
      "learning_rate": 2.949763648160163e-05,
      "loss": 0.0204,
      "step": 5625
    },
    {
      "epoch": 1.8852185518852185,
      "grad_norm": 0.0025838094297796488,
      "learning_rate": 2.938177773658356e-05,
      "loss": 0.0001,
      "step": 5650
    },
    {
      "epoch": 1.8935602268935603,
      "grad_norm": 0.0009950456442311406,
      "learning_rate": 2.9265918991565482e-05,
      "loss": 0.0,
      "step": 5675
    },
    {
      "epoch": 1.901901901901902,
      "grad_norm": 0.0015871904324740171,
      "learning_rate": 2.9150060246547413e-05,
      "loss": 0.0002,
      "step": 5700
    },
    {
      "epoch": 1.9102435769102435,
      "grad_norm": 0.0018013713415712118,
      "learning_rate": 2.9034201501529333e-05,
      "loss": 0.0676,
      "step": 5725
    },
    {
      "epoch": 1.9185852519185853,
      "grad_norm": 0.006069496739655733,
      "learning_rate": 2.8918342756511263e-05,
      "loss": 0.0378,
      "step": 5750
    },
    {
      "epoch": 1.926926926926927,
      "grad_norm": 0.0046617318876087666,
      "learning_rate": 2.880248401149319e-05,
      "loss": 0.0037,
      "step": 5775
    },
    {
      "epoch": 1.9352686019352685,
      "grad_norm": 0.0014451640890911222,
      "learning_rate": 2.8686625266475114e-05,
      "loss": 0.0003,
      "step": 5800
    },
    {
      "epoch": 1.9436102769436103,
      "grad_norm": 0.0014788233675062656,
      "learning_rate": 2.8570766521457045e-05,
      "loss": 0.0266,
      "step": 5825
    },
    {
      "epoch": 1.951951951951952,
      "grad_norm": 0.0011505110887810588,
      "learning_rate": 2.8454907776438965e-05,
      "loss": 0.0,
      "step": 5850
    },
    {
      "epoch": 1.9602936269602935,
      "grad_norm": 0.004947907291352749,
      "learning_rate": 2.8339049031420895e-05,
      "loss": 0.0098,
      "step": 5875
    },
    {
      "epoch": 1.9686353019686353,
      "grad_norm": 0.0015439976705238223,
      "learning_rate": 2.8223190286402816e-05,
      "loss": 0.0001,
      "step": 5900
    },
    {
      "epoch": 1.976976976976977,
      "grad_norm": 0.003472746815532446,
      "learning_rate": 2.8107331541384746e-05,
      "loss": 0.0001,
      "step": 5925
    },
    {
      "epoch": 1.9853186519853185,
      "grad_norm": 0.0020330217666924,
      "learning_rate": 2.799147279636667e-05,
      "loss": 0.0001,
      "step": 5950
    },
    {
      "epoch": 1.9936603269936604,
      "grad_norm": 0.000714894209522754,
      "learning_rate": 2.7875614051348597e-05,
      "loss": 0.0005,
      "step": 5975
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.997997997997998,
      "eval_auc": 0.9992763878743608,
      "eval_f1": 0.9142857142857143,
      "eval_loss": 0.009895329363644123,
      "eval_precision": 0.9696969696969697,
      "eval_recall": 0.8648648648648649,
      "eval_runtime": 78.0253,
      "eval_samples_per_second": 76.821,
      "eval_steps_per_second": 4.806,
      "step": 5994
    },
    {
      "epoch": 2.002002002002002,
      "grad_norm": 0.0006246891571208835,
      "learning_rate": 2.775975530633052e-05,
      "loss": 0.0,
      "step": 6000
    },
    {
      "epoch": 2.0103436770103436,
      "grad_norm": 0.009155196137726307,
      "learning_rate": 2.7643896561312448e-05,
      "loss": 0.0,
      "step": 6025
    },
    {
      "epoch": 2.0186853520186854,
      "grad_norm": 0.0010195561917498708,
      "learning_rate": 2.752803781629438e-05,
      "loss": 0.0005,
      "step": 6050
    },
    {
      "epoch": 2.027027027027027,
      "grad_norm": 0.00037840736331418157,
      "learning_rate": 2.7412179071276302e-05,
      "loss": 0.0,
      "step": 6075
    },
    {
      "epoch": 2.0353687020353686,
      "grad_norm": 2.285806655883789,
      "learning_rate": 2.729632032625823e-05,
      "loss": 0.0008,
      "step": 6100
    },
    {
      "epoch": 2.0437103770437104,
      "grad_norm": 0.000318177102599293,
      "learning_rate": 2.7180461581240153e-05,
      "loss": 0.0,
      "step": 6125
    },
    {
      "epoch": 2.052052052052052,
      "grad_norm": 0.00042503979057073593,
      "learning_rate": 2.706460283622208e-05,
      "loss": 0.0001,
      "step": 6150
    },
    {
      "epoch": 2.0603937270603936,
      "grad_norm": 0.0004734281974378973,
      "learning_rate": 2.6948744091204004e-05,
      "loss": 0.0,
      "step": 6175
    },
    {
      "epoch": 2.0687354020687354,
      "grad_norm": 0.00044175496441312134,
      "learning_rate": 2.683288534618593e-05,
      "loss": 0.0001,
      "step": 6200
    },
    {
      "epoch": 2.0770770770770772,
      "grad_norm": 0.0005970322526991367,
      "learning_rate": 2.6717026601167854e-05,
      "loss": 0.0,
      "step": 6225
    },
    {
      "epoch": 2.0854187520854186,
      "grad_norm": 0.00034779220004566014,
      "learning_rate": 2.6601167856149785e-05,
      "loss": 0.0001,
      "step": 6250
    },
    {
      "epoch": 2.0937604270937604,
      "grad_norm": 0.0002642270992510021,
      "learning_rate": 2.6485309111131712e-05,
      "loss": 0.0046,
      "step": 6275
    },
    {
      "epoch": 2.1021021021021022,
      "grad_norm": 0.0005227423971518874,
      "learning_rate": 2.6369450366113636e-05,
      "loss": 0.004,
      "step": 6300
    },
    {
      "epoch": 2.1104437771104436,
      "grad_norm": 0.0027825473807752132,
      "learning_rate": 2.6253591621095563e-05,
      "loss": 0.0441,
      "step": 6325
    },
    {
      "epoch": 2.1187854521187854,
      "grad_norm": 0.0015170240076258779,
      "learning_rate": 2.6137732876077486e-05,
      "loss": 0.0004,
      "step": 6350
    },
    {
      "epoch": 2.1271271271271273,
      "grad_norm": 0.0009673828026279807,
      "learning_rate": 2.6021874131059414e-05,
      "loss": 0.0001,
      "step": 6375
    },
    {
      "epoch": 2.1354688021354686,
      "grad_norm": 0.0016775678377598524,
      "learning_rate": 2.5906015386041337e-05,
      "loss": 0.0033,
      "step": 6400
    },
    {
      "epoch": 2.1438104771438105,
      "grad_norm": 0.4511735737323761,
      "learning_rate": 2.5790156641023268e-05,
      "loss": 0.0226,
      "step": 6425
    },
    {
      "epoch": 2.1521521521521523,
      "grad_norm": 0.00039455911610275507,
      "learning_rate": 2.5674297896005188e-05,
      "loss": 0.0005,
      "step": 6450
    },
    {
      "epoch": 2.1604938271604937,
      "grad_norm": 0.0009889901848509908,
      "learning_rate": 2.555843915098712e-05,
      "loss": 0.0298,
      "step": 6475
    },
    {
      "epoch": 2.1688355021688355,
      "grad_norm": 0.0007957897614687681,
      "learning_rate": 2.5442580405969046e-05,
      "loss": 0.0242,
      "step": 6500
    },
    {
      "epoch": 2.1771771771771773,
      "grad_norm": 0.00022962476941756904,
      "learning_rate": 2.532672166095097e-05,
      "loss": 0.0,
      "step": 6525
    },
    {
      "epoch": 2.1855188521855187,
      "grad_norm": 0.0003863972960971296,
      "learning_rate": 2.5210862915932896e-05,
      "loss": 0.0,
      "step": 6550
    },
    {
      "epoch": 2.1938605271938605,
      "grad_norm": 0.0011119615519419312,
      "learning_rate": 2.509500417091482e-05,
      "loss": 0.0001,
      "step": 6575
    },
    {
      "epoch": 2.2022022022022023,
      "grad_norm": 0.0002698484167922288,
      "learning_rate": 2.4979145425896747e-05,
      "loss": 0.0001,
      "step": 6600
    },
    {
      "epoch": 2.2105438772105437,
      "grad_norm": 0.0003634936292655766,
      "learning_rate": 2.4863286680878674e-05,
      "loss": 0.0,
      "step": 6625
    },
    {
      "epoch": 2.2188855522188855,
      "grad_norm": 0.10544054955244064,
      "learning_rate": 2.47474279358606e-05,
      "loss": 0.0729,
      "step": 6650
    },
    {
      "epoch": 2.2272272272272273,
      "grad_norm": 0.0028164817485958338,
      "learning_rate": 2.4631569190842525e-05,
      "loss": 0.0194,
      "step": 6675
    },
    {
      "epoch": 2.2355689022355687,
      "grad_norm": 0.0028918420430272818,
      "learning_rate": 2.4515710445824452e-05,
      "loss": 0.0004,
      "step": 6700
    },
    {
      "epoch": 2.2439105772439105,
      "grad_norm": 0.003687906078994274,
      "learning_rate": 2.439985170080638e-05,
      "loss": 0.0001,
      "step": 6725
    },
    {
      "epoch": 2.2522522522522523,
      "grad_norm": 0.0016472985735163093,
      "learning_rate": 2.4283992955788303e-05,
      "loss": 0.0016,
      "step": 6750
    },
    {
      "epoch": 2.2605939272605937,
      "grad_norm": 0.0029458694625645876,
      "learning_rate": 2.416813421077023e-05,
      "loss": 0.0001,
      "step": 6775
    },
    {
      "epoch": 2.2689356022689355,
      "grad_norm": 0.0009019913850352168,
      "learning_rate": 2.4052275465752154e-05,
      "loss": 0.0001,
      "step": 6800
    },
    {
      "epoch": 2.2772772772772774,
      "grad_norm": 0.0009054698166437447,
      "learning_rate": 2.393641672073408e-05,
      "loss": 0.0001,
      "step": 6825
    },
    {
      "epoch": 2.2856189522856187,
      "grad_norm": 0.001989714102819562,
      "learning_rate": 2.3820557975716008e-05,
      "loss": 0.0001,
      "step": 6850
    },
    {
      "epoch": 2.2939606272939606,
      "grad_norm": 0.0007018431788310409,
      "learning_rate": 2.3704699230697935e-05,
      "loss": 0.0001,
      "step": 6875
    },
    {
      "epoch": 2.3023023023023024,
      "grad_norm": 0.0006220221403054893,
      "learning_rate": 2.3588840485679862e-05,
      "loss": 0.0001,
      "step": 6900
    },
    {
      "epoch": 2.3106439773106437,
      "grad_norm": 0.001839311677031219,
      "learning_rate": 2.3472981740661786e-05,
      "loss": 0.0,
      "step": 6925
    },
    {
      "epoch": 2.3189856523189856,
      "grad_norm": 0.0005744717200286686,
      "learning_rate": 2.3357122995643713e-05,
      "loss": 0.0,
      "step": 6950
    },
    {
      "epoch": 2.3273273273273274,
      "grad_norm": 0.0004123161779716611,
      "learning_rate": 2.324126425062564e-05,
      "loss": 0.0001,
      "step": 6975
    },
    {
      "epoch": 2.3356690023356688,
      "grad_norm": 0.0013543826062232256,
      "learning_rate": 2.3125405505607564e-05,
      "loss": 0.0,
      "step": 7000
    },
    {
      "epoch": 2.3440106773440106,
      "grad_norm": 0.001142465160228312,
      "learning_rate": 2.300954676058949e-05,
      "loss": 0.0198,
      "step": 7025
    },
    {
      "epoch": 2.3523523523523524,
      "grad_norm": 0.0012254125904291868,
      "learning_rate": 2.2893688015571414e-05,
      "loss": 0.0008,
      "step": 7050
    },
    {
      "epoch": 2.360694027360694,
      "grad_norm": 0.0004832675331272185,
      "learning_rate": 2.277782927055334e-05,
      "loss": 0.0154,
      "step": 7075
    },
    {
      "epoch": 2.3690357023690356,
      "grad_norm": 0.00016073780716396868,
      "learning_rate": 2.266197052553527e-05,
      "loss": 0.0,
      "step": 7100
    },
    {
      "epoch": 2.3773773773773774,
      "grad_norm": 0.002141432836651802,
      "learning_rate": 2.2546111780517196e-05,
      "loss": 0.0001,
      "step": 7125
    },
    {
      "epoch": 2.3857190523857192,
      "grad_norm": 0.0022791263181716204,
      "learning_rate": 2.2430253035499123e-05,
      "loss": 0.0442,
      "step": 7150
    },
    {
      "epoch": 2.3940607273940606,
      "grad_norm": 0.0010367584181949496,
      "learning_rate": 2.2314394290481047e-05,
      "loss": 0.0004,
      "step": 7175
    },
    {
      "epoch": 2.4024024024024024,
      "grad_norm": 0.014609686098992825,
      "learning_rate": 2.2198535545462974e-05,
      "loss": 0.0001,
      "step": 7200
    },
    {
      "epoch": 2.4107440774107443,
      "grad_norm": 0.0009007493499666452,
      "learning_rate": 2.2082676800444897e-05,
      "loss": 0.0001,
      "step": 7225
    },
    {
      "epoch": 2.4190857524190856,
      "grad_norm": 0.005582588724792004,
      "learning_rate": 2.1966818055426824e-05,
      "loss": 0.0416,
      "step": 7250
    },
    {
      "epoch": 2.4274274274274275,
      "grad_norm": 0.030880291014909744,
      "learning_rate": 2.185095931040875e-05,
      "loss": 0.0002,
      "step": 7275
    },
    {
      "epoch": 2.4357691024357693,
      "grad_norm": 0.007195931393653154,
      "learning_rate": 2.1735100565390675e-05,
      "loss": 0.0019,
      "step": 7300
    },
    {
      "epoch": 2.4441107774441106,
      "grad_norm": 0.0043434868566691875,
      "learning_rate": 2.1619241820372602e-05,
      "loss": 0.0003,
      "step": 7325
    },
    {
      "epoch": 2.4524524524524525,
      "grad_norm": 0.00130070885643363,
      "learning_rate": 2.150338307535453e-05,
      "loss": 0.0001,
      "step": 7350
    },
    {
      "epoch": 2.4607941274607943,
      "grad_norm": 0.011537245474755764,
      "learning_rate": 2.1387524330336456e-05,
      "loss": 0.0001,
      "step": 7375
    },
    {
      "epoch": 2.4691358024691357,
      "grad_norm": 0.0012115085264667869,
      "learning_rate": 2.127166558531838e-05,
      "loss": 0.0001,
      "step": 7400
    },
    {
      "epoch": 2.4774774774774775,
      "grad_norm": 0.0006190771819092333,
      "learning_rate": 2.1155806840300307e-05,
      "loss": 0.0001,
      "step": 7425
    },
    {
      "epoch": 2.4858191524858193,
      "grad_norm": 0.0007142933318391442,
      "learning_rate": 2.1039948095282234e-05,
      "loss": 0.0,
      "step": 7450
    },
    {
      "epoch": 2.4941608274941607,
      "grad_norm": 0.0024094288237392902,
      "learning_rate": 2.0924089350264158e-05,
      "loss": 0.0349,
      "step": 7475
    },
    {
      "epoch": 2.5025025025025025,
      "grad_norm": 0.0028155215550214052,
      "learning_rate": 2.0808230605246085e-05,
      "loss": 0.0349,
      "step": 7500
    },
    {
      "epoch": 2.5108441775108443,
      "grad_norm": 0.00238401023671031,
      "learning_rate": 2.069237186022801e-05,
      "loss": 0.0087,
      "step": 7525
    },
    {
      "epoch": 2.5191858525191857,
      "grad_norm": 0.0014537558890879154,
      "learning_rate": 2.0576513115209936e-05,
      "loss": 0.0033,
      "step": 7550
    },
    {
      "epoch": 2.5275275275275275,
      "grad_norm": 0.0015429011546075344,
      "learning_rate": 2.0460654370191863e-05,
      "loss": 0.0143,
      "step": 7575
    },
    {
      "epoch": 2.5358692025358693,
      "grad_norm": 0.0013287997571751475,
      "learning_rate": 2.034479562517379e-05,
      "loss": 0.0014,
      "step": 7600
    },
    {
      "epoch": 2.5442108775442107,
      "grad_norm": 0.0010292051592841744,
      "learning_rate": 2.0228936880155717e-05,
      "loss": 0.0049,
      "step": 7625
    },
    {
      "epoch": 2.5525525525525525,
      "grad_norm": 0.007215111516416073,
      "learning_rate": 2.011307813513764e-05,
      "loss": 0.005,
      "step": 7650
    },
    {
      "epoch": 2.5608942275608944,
      "grad_norm": 0.007656760048121214,
      "learning_rate": 1.9997219390119568e-05,
      "loss": 0.0355,
      "step": 7675
    },
    {
      "epoch": 2.569235902569236,
      "grad_norm": 0.001918874098919332,
      "learning_rate": 1.9881360645101492e-05,
      "loss": 0.0045,
      "step": 7700
    },
    {
      "epoch": 2.5775775775775776,
      "grad_norm": 0.0005019979435019195,
      "learning_rate": 1.976550190008342e-05,
      "loss": 0.0001,
      "step": 7725
    },
    {
      "epoch": 2.5859192525859194,
      "grad_norm": 0.00030773007893003523,
      "learning_rate": 1.9649643155065346e-05,
      "loss": 0.0001,
      "step": 7750
    },
    {
      "epoch": 2.594260927594261,
      "grad_norm": 0.0003611762367654592,
      "learning_rate": 1.953378441004727e-05,
      "loss": 0.0001,
      "step": 7775
    },
    {
      "epoch": 2.6026026026026026,
      "grad_norm": 0.0019309103954583406,
      "learning_rate": 1.9417925665029197e-05,
      "loss": 0.0,
      "step": 7800
    },
    {
      "epoch": 2.6109442776109444,
      "grad_norm": 0.004138192627578974,
      "learning_rate": 1.930206692001112e-05,
      "loss": 0.0533,
      "step": 7825
    },
    {
      "epoch": 2.619285952619286,
      "grad_norm": 0.042996328324079514,
      "learning_rate": 1.918620817499305e-05,
      "loss": 0.0001,
      "step": 7850
    },
    {
      "epoch": 2.6276276276276276,
      "grad_norm": 0.0017391805304214358,
      "learning_rate": 1.9070349429974978e-05,
      "loss": 0.0001,
      "step": 7875
    },
    {
      "epoch": 2.6359693026359694,
      "grad_norm": 0.002806081436574459,
      "learning_rate": 1.89544906849569e-05,
      "loss": 0.0002,
      "step": 7900
    },
    {
      "epoch": 2.6443109776443112,
      "grad_norm": 0.0011362434597685933,
      "learning_rate": 1.883863193993883e-05,
      "loss": 0.0001,
      "step": 7925
    },
    {
      "epoch": 2.6526526526526526,
      "grad_norm": 0.0013199028326198459,
      "learning_rate": 1.8722773194920752e-05,
      "loss": 0.0001,
      "step": 7950
    },
    {
      "epoch": 2.6609943276609944,
      "grad_norm": 0.0009320782264694571,
      "learning_rate": 1.860691444990268e-05,
      "loss": 0.0001,
      "step": 7975
    },
    {
      "epoch": 2.6693360026693362,
      "grad_norm": 0.0012447205372154713,
      "learning_rate": 1.8491055704884607e-05,
      "loss": 0.0001,
      "step": 8000
    },
    {
      "epoch": 2.6776776776776776,
      "grad_norm": 0.0005381664377637208,
      "learning_rate": 1.837519695986653e-05,
      "loss": 0.0,
      "step": 8025
    },
    {
      "epoch": 2.6860193526860194,
      "grad_norm": 0.0014626341871917248,
      "learning_rate": 1.8259338214848457e-05,
      "loss": 0.0002,
      "step": 8050
    },
    {
      "epoch": 2.6943610276943613,
      "grad_norm": 0.0006017221603542566,
      "learning_rate": 1.8143479469830384e-05,
      "loss": 0.0001,
      "step": 8075
    },
    {
      "epoch": 2.7027027027027026,
      "grad_norm": 0.0003165393427480012,
      "learning_rate": 1.802762072481231e-05,
      "loss": 0.0023,
      "step": 8100
    },
    {
      "epoch": 2.7110443777110445,
      "grad_norm": 0.0005794435273855925,
      "learning_rate": 1.7911761979794235e-05,
      "loss": 0.0001,
      "step": 8125
    },
    {
      "epoch": 2.7193860527193863,
      "grad_norm": 0.0025553391315042973,
      "learning_rate": 1.7795903234776162e-05,
      "loss": 0.0003,
      "step": 8150
    },
    {
      "epoch": 2.7277277277277276,
      "grad_norm": 0.0002689988468773663,
      "learning_rate": 1.768004448975809e-05,
      "loss": 0.0,
      "step": 8175
    },
    {
      "epoch": 2.7360694027360695,
      "grad_norm": 0.0001785249332897365,
      "learning_rate": 1.7564185744740013e-05,
      "loss": 0.0,
      "step": 8200
    },
    {
      "epoch": 2.7444110777444113,
      "grad_norm": 0.000306436704704538,
      "learning_rate": 1.744832699972194e-05,
      "loss": 0.0,
      "step": 8225
    },
    {
      "epoch": 2.7527527527527527,
      "grad_norm": 0.00017187983030453324,
      "learning_rate": 1.7332468254703864e-05,
      "loss": 0.0007,
      "step": 8250
    },
    {
      "epoch": 2.7610944277610945,
      "grad_norm": 0.00022748754417989403,
      "learning_rate": 1.721660950968579e-05,
      "loss": 0.0097,
      "step": 8275
    },
    {
      "epoch": 2.7694361027694363,
      "grad_norm": 0.0016567155253142118,
      "learning_rate": 1.7100750764667718e-05,
      "loss": 0.0018,
      "step": 8300
    },
    {
      "epoch": 2.7777777777777777,
      "grad_norm": 0.0004330672090873122,
      "learning_rate": 1.6984892019649645e-05,
      "loss": 0.0001,
      "step": 8325
    },
    {
      "epoch": 2.7861194527861195,
      "grad_norm": 0.006991783156991005,
      "learning_rate": 1.6869033274631572e-05,
      "loss": 0.0174,
      "step": 8350
    },
    {
      "epoch": 2.7944611277944613,
      "grad_norm": 0.0004689243796747178,
      "learning_rate": 1.6753174529613496e-05,
      "loss": 0.0015,
      "step": 8375
    },
    {
      "epoch": 2.8028028028028027,
      "grad_norm": 0.0006888713105581701,
      "learning_rate": 1.6637315784595423e-05,
      "loss": 0.0001,
      "step": 8400
    },
    {
      "epoch": 2.8111444778111445,
      "grad_norm": 0.0015181319322437048,
      "learning_rate": 1.6521457039577347e-05,
      "loss": 0.0013,
      "step": 8425
    },
    {
      "epoch": 2.8194861528194863,
      "grad_norm": 0.00026851208531297743,
      "learning_rate": 1.6405598294559274e-05,
      "loss": 0.0067,
      "step": 8450
    },
    {
      "epoch": 2.8278278278278277,
      "grad_norm": 0.0003072840045206249,
      "learning_rate": 1.62897395495412e-05,
      "loss": 0.0,
      "step": 8475
    },
    {
      "epoch": 2.8361695028361695,
      "grad_norm": 0.000419327465351671,
      "learning_rate": 1.6173880804523125e-05,
      "loss": 0.0,
      "step": 8500
    },
    {
      "epoch": 2.8445111778445114,
      "grad_norm": 0.0008704416686668992,
      "learning_rate": 1.6058022059505052e-05,
      "loss": 0.0,
      "step": 8525
    },
    {
      "epoch": 2.8528528528528527,
      "grad_norm": 0.00026631655055098236,
      "learning_rate": 1.5942163314486975e-05,
      "loss": 0.0,
      "step": 8550
    },
    {
      "epoch": 2.8611945278611945,
      "grad_norm": 0.00015377814997918904,
      "learning_rate": 1.5826304569468906e-05,
      "loss": 0.0,
      "step": 8575
    },
    {
      "epoch": 2.8695362028695364,
      "grad_norm": 0.0001188891037600115,
      "learning_rate": 1.571044582445083e-05,
      "loss": 0.0,
      "step": 8600
    },
    {
      "epoch": 2.8778778778778777,
      "grad_norm": 0.00026092492043972015,
      "learning_rate": 1.5594587079432757e-05,
      "loss": 0.0,
      "step": 8625
    },
    {
      "epoch": 2.8862195528862196,
      "grad_norm": 0.0021843460854142904,
      "learning_rate": 1.5478728334414684e-05,
      "loss": 0.0025,
      "step": 8650
    },
    {
      "epoch": 2.8945612278945614,
      "grad_norm": 0.00022988433192949742,
      "learning_rate": 1.5362869589396608e-05,
      "loss": 0.0094,
      "step": 8675
    },
    {
      "epoch": 2.9029029029029028,
      "grad_norm": 0.00016332197992596775,
      "learning_rate": 1.5247010844378535e-05,
      "loss": 0.006,
      "step": 8700
    },
    {
      "epoch": 2.9112445779112446,
      "grad_norm": 0.00020633054373320192,
      "learning_rate": 1.513115209936046e-05,
      "loss": 0.0002,
      "step": 8725
    },
    {
      "epoch": 2.9195862529195864,
      "grad_norm": 0.00025880796601995826,
      "learning_rate": 1.5015293354342385e-05,
      "loss": 0.0,
      "step": 8750
    },
    {
      "epoch": 2.9279279279279278,
      "grad_norm": 0.0003759913379326463,
      "learning_rate": 1.489943460932431e-05,
      "loss": 0.0003,
      "step": 8775
    },
    {
      "epoch": 2.9362696029362696,
      "grad_norm": 0.00019277114188298583,
      "learning_rate": 1.4783575864306238e-05,
      "loss": 0.0044,
      "step": 8800
    },
    {
      "epoch": 2.9446112779446114,
      "grad_norm": 0.0007216227240860462,
      "learning_rate": 1.4667717119288165e-05,
      "loss": 0.0,
      "step": 8825
    },
    {
      "epoch": 2.952952952952953,
      "grad_norm": 0.0001798528101062402,
      "learning_rate": 1.4551858374270092e-05,
      "loss": 0.0304,
      "step": 8850
    },
    {
      "epoch": 2.9612946279612946,
      "grad_norm": 0.00019520522619131953,
      "learning_rate": 1.4435999629252017e-05,
      "loss": 0.0001,
      "step": 8875
    },
    {
      "epoch": 2.9696363029696364,
      "grad_norm": 0.0001343146723229438,
      "learning_rate": 1.4320140884233943e-05,
      "loss": 0.0489,
      "step": 8900
    },
    {
      "epoch": 2.977977977977978,
      "grad_norm": 12.330880165100098,
      "learning_rate": 1.4204282139215868e-05,
      "loss": 0.0347,
      "step": 8925
    },
    {
      "epoch": 2.9863196529863196,
      "grad_norm": 0.00025471311528235674,
      "learning_rate": 1.4088423394197794e-05,
      "loss": 0.0245,
      "step": 8950
    },
    {
      "epoch": 2.9946613279946614,
      "grad_norm": 0.0009992901468649507,
      "learning_rate": 1.397256464917972e-05,
      "loss": 0.0337,
      "step": 8975
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9983316649983317,
      "eval_auc": 0.9996484660336011,
      "eval_f1": 0.935064935064935,
      "eval_loss": 0.007099705282598734,
      "eval_precision": 0.9,
      "eval_recall": 0.972972972972973,
      "eval_runtime": 77.3547,
      "eval_samples_per_second": 77.487,
      "eval_steps_per_second": 4.848,
      "step": 8991
    },
    {
      "epoch": 3.003003003003003,
      "grad_norm": 0.0015147333033382893,
      "learning_rate": 1.3856705904161646e-05,
      "loss": 0.0003,
      "step": 9000
    },
    {
      "epoch": 3.0113446780113446,
      "grad_norm": 0.002197373891249299,
      "learning_rate": 1.3740847159143572e-05,
      "loss": 0.0002,
      "step": 9025
    },
    {
      "epoch": 3.0196863530196865,
      "grad_norm": 0.0003656800545286387,
      "learning_rate": 1.36249884141255e-05,
      "loss": 0.001,
      "step": 9050
    },
    {
      "epoch": 3.028028028028028,
      "grad_norm": 0.00043537470628507435,
      "learning_rate": 1.3509129669107426e-05,
      "loss": 0.0,
      "step": 9075
    },
    {
      "epoch": 3.0363697030363697,
      "grad_norm": 0.0017388892592862248,
      "learning_rate": 1.3393270924089351e-05,
      "loss": 0.0,
      "step": 9100
    },
    {
      "epoch": 3.0447113780447115,
      "grad_norm": 0.00047031775466166437,
      "learning_rate": 1.3277412179071278e-05,
      "loss": 0.0,
      "step": 9125
    },
    {
      "epoch": 3.053053053053053,
      "grad_norm": 0.0003807338362094015,
      "learning_rate": 1.3161553434053204e-05,
      "loss": 0.0,
      "step": 9150
    },
    {
      "epoch": 3.0613947280613947,
      "grad_norm": 0.00046666429261676967,
      "learning_rate": 1.3045694689035129e-05,
      "loss": 0.0,
      "step": 9175
    },
    {
      "epoch": 3.0697364030697365,
      "grad_norm": 0.000990239786915481,
      "learning_rate": 1.2929835944017054e-05,
      "loss": 0.0001,
      "step": 9200
    },
    {
      "epoch": 3.078078078078078,
      "grad_norm": 0.0009268262074328959,
      "learning_rate": 1.281397719899898e-05,
      "loss": 0.0,
      "step": 9225
    },
    {
      "epoch": 3.0864197530864197,
      "grad_norm": 0.0001752979151206091,
      "learning_rate": 1.2698118453980907e-05,
      "loss": 0.0,
      "step": 9250
    },
    {
      "epoch": 3.0947614280947615,
      "grad_norm": 0.00017092602502088994,
      "learning_rate": 1.2582259708962832e-05,
      "loss": 0.0001,
      "step": 9275
    },
    {
      "epoch": 3.103103103103103,
      "grad_norm": 0.0003801763814408332,
      "learning_rate": 1.246640096394476e-05,
      "loss": 0.0,
      "step": 9300
    },
    {
      "epoch": 3.1114447781114447,
      "grad_norm": 0.0007715587271377444,
      "learning_rate": 1.2350542218926685e-05,
      "loss": 0.0,
      "step": 9325
    },
    {
      "epoch": 3.1197864531197865,
      "grad_norm": 0.0001741141895763576,
      "learning_rate": 1.223468347390861e-05,
      "loss": 0.0,
      "step": 9350
    },
    {
      "epoch": 3.128128128128128,
      "grad_norm": 0.00024119947920553386,
      "learning_rate": 1.2118824728890537e-05,
      "loss": 0.0,
      "step": 9375
    },
    {
      "epoch": 3.1364698031364697,
      "grad_norm": 0.0003123109054286033,
      "learning_rate": 1.2002965983872463e-05,
      "loss": 0.0027,
      "step": 9400
    },
    {
      "epoch": 3.1448114781448115,
      "grad_norm": 0.0002255546278320253,
      "learning_rate": 1.188710723885439e-05,
      "loss": 0.0001,
      "step": 9425
    },
    {
      "epoch": 3.153153153153153,
      "grad_norm": 0.00022687349701300263,
      "learning_rate": 1.1771248493836315e-05,
      "loss": 0.0,
      "step": 9450
    },
    {
      "epoch": 3.1614948281614947,
      "grad_norm": 0.0009094116394408047,
      "learning_rate": 1.1655389748818242e-05,
      "loss": 0.0001,
      "step": 9475
    },
    {
      "epoch": 3.1698365031698366,
      "grad_norm": 0.00019586944836191833,
      "learning_rate": 1.1539531003800168e-05,
      "loss": 0.0001,
      "step": 9500
    },
    {
      "epoch": 3.1781781781781784,
      "grad_norm": 0.0013932987349107862,
      "learning_rate": 1.1423672258782093e-05,
      "loss": 0.0,
      "step": 9525
    },
    {
      "epoch": 3.1865198531865198,
      "grad_norm": 0.00014476373326033354,
      "learning_rate": 1.130781351376402e-05,
      "loss": 0.0,
      "step": 9550
    },
    {
      "epoch": 3.1948615281948616,
      "grad_norm": 0.00020363023213576525,
      "learning_rate": 1.1191954768745945e-05,
      "loss": 0.0,
      "step": 9575
    },
    {
      "epoch": 3.2032032032032034,
      "grad_norm": 0.0005529238260351121,
      "learning_rate": 1.1076096023727873e-05,
      "loss": 0.0005,
      "step": 9600
    },
    {
      "epoch": 3.2115448782115448,
      "grad_norm": 0.00039333957829512656,
      "learning_rate": 1.0960237278709798e-05,
      "loss": 0.0001,
      "step": 9625
    },
    {
      "epoch": 3.2198865532198866,
      "grad_norm": 0.0006521272589452565,
      "learning_rate": 1.0844378533691723e-05,
      "loss": 0.0,
      "step": 9650
    },
    {
      "epoch": 3.2282282282282284,
      "grad_norm": Infinity,
      "learning_rate": 1.0733154138474373e-05,
      "loss": 0.0231,
      "step": 9675
    },
    {
      "epoch": 3.23656990323657,
      "grad_norm": 0.0003595724992919713,
      "learning_rate": 1.0617295393456298e-05,
      "loss": 0.0,
      "step": 9700
    },
    {
      "epoch": 3.2449115782449116,
      "grad_norm": 0.00021334755001589656,
      "learning_rate": 1.0501436648438225e-05,
      "loss": 0.0003,
      "step": 9725
    },
    {
      "epoch": 3.2532532532532534,
      "grad_norm": 0.00028163444949314,
      "learning_rate": 1.038557790342015e-05,
      "loss": 0.0,
      "step": 9750
    },
    {
      "epoch": 3.261594928261595,
      "grad_norm": 0.00030702631920576096,
      "learning_rate": 1.0269719158402076e-05,
      "loss": 0.0,
      "step": 9775
    },
    {
      "epoch": 3.2699366032699366,
      "grad_norm": 0.0002560736029408872,
      "learning_rate": 1.0153860413384003e-05,
      "loss": 0.0,
      "step": 9800
    },
    {
      "epoch": 3.2782782782782784,
      "grad_norm": 7.156394713092595e-05,
      "learning_rate": 1.0038001668365928e-05,
      "loss": 0.0004,
      "step": 9825
    },
    {
      "epoch": 3.28661995328662,
      "grad_norm": 0.00014015445776749402,
      "learning_rate": 9.922142923347855e-06,
      "loss": 0.0211,
      "step": 9850
    },
    {
      "epoch": 3.2949616282949616,
      "grad_norm": 8.722938946448267e-05,
      "learning_rate": 9.806284178329781e-06,
      "loss": 0.0355,
      "step": 9875
    },
    {
      "epoch": 3.3033033033033035,
      "grad_norm": 0.00010665127047104761,
      "learning_rate": 9.690425433311706e-06,
      "loss": 0.0,
      "step": 9900
    },
    {
      "epoch": 3.311644978311645,
      "grad_norm": 0.0003643149684648961,
      "learning_rate": 9.574566688293633e-06,
      "loss": 0.0159,
      "step": 9925
    },
    {
      "epoch": 3.3199866533199867,
      "grad_norm": 0.000472669635200873,
      "learning_rate": 9.458707943275559e-06,
      "loss": 0.0001,
      "step": 9950
    },
    {
      "epoch": 3.3283283283283285,
      "grad_norm": 0.0002747498219832778,
      "learning_rate": 9.342849198257484e-06,
      "loss": 0.0002,
      "step": 9975
    },
    {
      "epoch": 3.33667000333667,
      "grad_norm": 0.000370238529285416,
      "learning_rate": 9.226990453239411e-06,
      "loss": 0.0,
      "step": 10000
    },
    {
      "epoch": 3.3450116783450117,
      "grad_norm": 0.0001508592686150223,
      "learning_rate": 9.111131708221338e-06,
      "loss": 0.0153,
      "step": 10025
    },
    {
      "epoch": 3.3533533533533535,
      "grad_norm": 0.00012634581071324646,
      "learning_rate": 8.995272963203264e-06,
      "loss": 0.0,
      "step": 10050
    },
    {
      "epoch": 3.361695028361695,
      "grad_norm": 7.435754378093407e-05,
      "learning_rate": 8.879414218185189e-06,
      "loss": 0.0,
      "step": 10075
    },
    {
      "epoch": 3.3700367033700367,
      "grad_norm": 0.00024650897830724716,
      "learning_rate": 8.763555473167115e-06,
      "loss": 0.0,
      "step": 10100
    },
    {
      "epoch": 3.3783783783783785,
      "grad_norm": 0.00031506569939665496,
      "learning_rate": 8.64769672814904e-06,
      "loss": 0.0,
      "step": 10125
    },
    {
      "epoch": 3.38672005338672,
      "grad_norm": 0.00038327754009515047,
      "learning_rate": 8.531837983130967e-06,
      "loss": 0.0,
      "step": 10150
    },
    {
      "epoch": 3.3950617283950617,
      "grad_norm": 0.00010452201968291774,
      "learning_rate": 8.415979238112894e-06,
      "loss": 0.0,
      "step": 10175
    },
    {
      "epoch": 3.4034034034034035,
      "grad_norm": 0.003248438471928239,
      "learning_rate": 8.30012049309482e-06,
      "loss": 0.0,
      "step": 10200
    },
    {
      "epoch": 3.411745078411745,
      "grad_norm": 0.0001290349755436182,
      "learning_rate": 8.184261748076745e-06,
      "loss": 0.0,
      "step": 10225
    },
    {
      "epoch": 3.4200867534200867,
      "grad_norm": 0.0004630775947589427,
      "learning_rate": 8.06840300305867e-06,
      "loss": 0.0,
      "step": 10250
    },
    {
      "epoch": 3.4284284284284285,
      "grad_norm": 0.00028067638049833477,
      "learning_rate": 7.952544258040597e-06,
      "loss": 0.0,
      "step": 10275
    },
    {
      "epoch": 3.43677010343677,
      "grad_norm": 0.00016888586105778813,
      "learning_rate": 7.836685513022524e-06,
      "loss": 0.0,
      "step": 10300
    },
    {
      "epoch": 3.4451117784451117,
      "grad_norm": 0.00010634879436111078,
      "learning_rate": 7.72082676800445e-06,
      "loss": 0.0,
      "step": 10325
    },
    {
      "epoch": 3.4534534534534536,
      "grad_norm": 0.004633384756743908,
      "learning_rate": 7.604968022986375e-06,
      "loss": 0.0,
      "step": 10350
    },
    {
      "epoch": 3.461795128461795,
      "grad_norm": 0.00013694832159671932,
      "learning_rate": 7.489109277968301e-06,
      "loss": 0.0,
      "step": 10375
    },
    {
      "epoch": 3.4701368034701368,
      "grad_norm": 0.0003641396469902247,
      "learning_rate": 7.373250532950228e-06,
      "loss": 0.0337,
      "step": 10400
    },
    {
      "epoch": 3.4784784784784786,
      "grad_norm": 0.000520822883117944,
      "learning_rate": 7.257391787932154e-06,
      "loss": 0.0,
      "step": 10425
    },
    {
      "epoch": 3.4868201534868204,
      "grad_norm": 0.000284605281194672,
      "learning_rate": 7.141533042914079e-06,
      "loss": 0.0,
      "step": 10450
    },
    {
      "epoch": 3.4951618284951618,
      "grad_norm": 0.0006933792028576136,
      "learning_rate": 7.025674297896006e-06,
      "loss": 0.0002,
      "step": 10475
    },
    {
      "epoch": 3.5035035035035036,
      "grad_norm": 0.0002910966577474028,
      "learning_rate": 6.914449902678655e-06,
      "loss": 0.0221,
      "step": 10500
    },
    {
      "epoch": 3.5118451785118454,
      "grad_norm": 0.0009816793026402593,
      "learning_rate": 6.79859115766058e-06,
      "loss": 0.0,
      "step": 10525
    },
    {
      "epoch": 3.520186853520187,
      "grad_norm": 0.00022771142539568245,
      "learning_rate": 6.6827324126425065e-06,
      "loss": 0.0,
      "step": 10550
    },
    {
      "epoch": 3.5285285285285286,
      "grad_norm": 0.00018489710055291653,
      "learning_rate": 6.566873667624434e-06,
      "loss": 0.0,
      "step": 10575
    },
    {
      "epoch": 3.5368702035368704,
      "grad_norm": 0.0005421001696959138,
      "learning_rate": 6.451014922606359e-06,
      "loss": 0.0001,
      "step": 10600
    },
    {
      "epoch": 3.545211878545212,
      "grad_norm": 0.00024389017198700458,
      "learning_rate": 6.335156177588284e-06,
      "loss": 0.0,
      "step": 10625
    },
    {
      "epoch": 3.5535535535535536,
      "grad_norm": 0.0002914558572228998,
      "learning_rate": 6.219297432570211e-06,
      "loss": 0.0,
      "step": 10650
    },
    {
      "epoch": 3.5618952285618954,
      "grad_norm": 0.0002373899769736454,
      "learning_rate": 6.103438687552137e-06,
      "loss": 0.0,
      "step": 10675
    },
    {
      "epoch": 3.570236903570237,
      "grad_norm": 0.006195495370775461,
      "learning_rate": 5.987579942534063e-06,
      "loss": 0.0,
      "step": 10700
    },
    {
      "epoch": 3.5785785785785786,
      "grad_norm": 0.00012051707017235458,
      "learning_rate": 5.8717211975159885e-06,
      "loss": 0.0,
      "step": 10725
    },
    {
      "epoch": 3.5869202535869205,
      "grad_norm": 0.002486045239493251,
      "learning_rate": 5.755862452497915e-06,
      "loss": 0.0001,
      "step": 10750
    },
    {
      "epoch": 3.595261928595262,
      "grad_norm": 0.0001380012690788135,
      "learning_rate": 5.640003707479841e-06,
      "loss": 0.0,
      "step": 10775
    },
    {
      "epoch": 3.6036036036036037,
      "grad_norm": 0.00035765368374995887,
      "learning_rate": 5.524144962461767e-06,
      "loss": 0.0,
      "step": 10800
    },
    {
      "epoch": 3.6119452786119455,
      "grad_norm": 0.00022478456958197057,
      "learning_rate": 5.408286217443693e-06,
      "loss": 0.0,
      "step": 10825
    },
    {
      "epoch": 3.620286953620287,
      "grad_norm": 0.00024634675355628133,
      "learning_rate": 5.292427472425619e-06,
      "loss": 0.0,
      "step": 10850
    },
    {
      "epoch": 3.6286286286286287,
      "grad_norm": 9.515749115962535e-05,
      "learning_rate": 5.176568727407545e-06,
      "loss": 0.0,
      "step": 10875
    },
    {
      "epoch": 3.6369703036369705,
      "grad_norm": 0.0015800537075847387,
      "learning_rate": 5.0607099823894705e-06,
      "loss": 0.0,
      "step": 10900
    },
    {
      "epoch": 3.645311978645312,
      "grad_norm": 0.00018024566816166043,
      "learning_rate": 4.944851237371398e-06,
      "loss": 0.0,
      "step": 10925
    },
    {
      "epoch": 3.6536536536536537,
      "grad_norm": 0.00037315645022317767,
      "learning_rate": 4.828992492353323e-06,
      "loss": 0.0,
      "step": 10950
    },
    {
      "epoch": 3.6619953286619955,
      "grad_norm": 0.00012279977090656757,
      "learning_rate": 4.713133747335249e-06,
      "loss": 0.0098,
      "step": 10975
    },
    {
      "epoch": 3.670337003670337,
      "grad_norm": 0.00033114643883891404,
      "learning_rate": 4.5972750023171755e-06,
      "loss": 0.0,
      "step": 11000
    },
    {
      "epoch": 3.6786786786786787,
      "grad_norm": 0.00015320570673793554,
      "learning_rate": 4.481416257299101e-06,
      "loss": 0.0,
      "step": 11025
    },
    {
      "epoch": 3.6870203536870205,
      "grad_norm": 0.00033146454370580614,
      "learning_rate": 4.365557512281027e-06,
      "loss": 0.0131,
      "step": 11050
    },
    {
      "epoch": 3.695362028695362,
      "grad_norm": 0.0005683086346834898,
      "learning_rate": 4.249698767262953e-06,
      "loss": 0.0,
      "step": 11075
    },
    {
      "epoch": 3.7037037037037037,
      "grad_norm": 0.0013795130653306842,
      "learning_rate": 4.13384002224488e-06,
      "loss": 0.0001,
      "step": 11100
    },
    {
      "epoch": 3.7120453787120455,
      "grad_norm": 0.0038050978910177946,
      "learning_rate": 4.017981277226805e-06,
      "loss": 0.0004,
      "step": 11125
    },
    {
      "epoch": 3.720387053720387,
      "grad_norm": 0.00022509918198920786,
      "learning_rate": 3.902122532208731e-06,
      "loss": 0.0001,
      "step": 11150
    },
    {
      "epoch": 3.7287287287287287,
      "grad_norm": 0.0005176962004043162,
      "learning_rate": 3.7862637871906575e-06,
      "loss": 0.0,
      "step": 11175
    },
    {
      "epoch": 3.7370704037370706,
      "grad_norm": 0.0005020920070819557,
      "learning_rate": 3.6704050421725833e-06,
      "loss": 0.0598,
      "step": 11200
    },
    {
      "epoch": 3.745412078745412,
      "grad_norm": 0.0011666618520393968,
      "learning_rate": 3.5545462971545095e-06,
      "loss": 0.0001,
      "step": 11225
    },
    {
      "epoch": 3.7537537537537538,
      "grad_norm": 0.009826689027249813,
      "learning_rate": 3.4386875521364354e-06,
      "loss": 0.0005,
      "step": 11250
    },
    {
      "epoch": 3.7620954287620956,
      "grad_norm": 0.011009103618562222,
      "learning_rate": 3.3228288071183616e-06,
      "loss": 0.0371,
      "step": 11275
    },
    {
      "epoch": 3.770437103770437,
      "grad_norm": 0.01023051142692566,
      "learning_rate": 3.2069700621002874e-06,
      "loss": 0.0001,
      "step": 11300
    },
    {
      "epoch": 3.7787787787787788,
      "grad_norm": 0.0009649486164562404,
      "learning_rate": 3.0911113170822137e-06,
      "loss": 0.0001,
      "step": 11325
    },
    {
      "epoch": 3.7871204537871206,
      "grad_norm": 0.003300480078905821,
      "learning_rate": 2.97525257206414e-06,
      "loss": 0.0073,
      "step": 11350
    },
    {
      "epoch": 3.795462128795462,
      "grad_norm": 0.004097791388630867,
      "learning_rate": 2.8593938270460653e-06,
      "loss": 0.0016,
      "step": 11375
    },
    {
      "epoch": 3.803803803803804,
      "grad_norm": 0.0011136537650600076,
      "learning_rate": 2.7435350820279915e-06,
      "loss": 0.0001,
      "step": 11400
    },
    {
      "epoch": 3.8121454788121456,
      "grad_norm": 0.001229249406605959,
      "learning_rate": 2.6276763370099178e-06,
      "loss": 0.0002,
      "step": 11425
    },
    {
      "epoch": 3.820487153820487,
      "grad_norm": 0.001040545990690589,
      "learning_rate": 2.5118175919918436e-06,
      "loss": 0.0104,
      "step": 11450
    },
    {
      "epoch": 3.828828828828829,
      "grad_norm": 0.0006933126132935286,
      "learning_rate": 2.39595884697377e-06,
      "loss": 0.0006,
      "step": 11475
    },
    {
      "epoch": 3.8371705038371706,
      "grad_norm": 0.0001256452960660681,
      "learning_rate": 2.2801001019556957e-06,
      "loss": 0.0109,
      "step": 11500
    },
    {
      "epoch": 3.845512178845512,
      "grad_norm": 0.00040541382622905076,
      "learning_rate": 2.1642413569376215e-06,
      "loss": 0.0,
      "step": 11525
    },
    {
      "epoch": 3.853853853853854,
      "grad_norm": 0.007394419051706791,
      "learning_rate": 2.0483826119195477e-06,
      "loss": 0.0,
      "step": 11550
    },
    {
      "epoch": 3.8621955288621956,
      "grad_norm": 0.00017597030091565102,
      "learning_rate": 1.932523866901474e-06,
      "loss": 0.0,
      "step": 11575
    },
    {
      "epoch": 3.870537203870537,
      "grad_norm": 0.00032824205118231475,
      "learning_rate": 1.8166651218834e-06,
      "loss": 0.0,
      "step": 11600
    },
    {
      "epoch": 3.878878878878879,
      "grad_norm": 0.00014257676957640797,
      "learning_rate": 1.700806376865326e-06,
      "loss": 0.0,
      "step": 11625
    },
    {
      "epoch": 3.8872205538872207,
      "grad_norm": 0.003545379266142845,
      "learning_rate": 1.5849476318472518e-06,
      "loss": 0.0,
      "step": 11650
    },
    {
      "epoch": 3.895562228895562,
      "grad_norm": 0.00017835904145613313,
      "learning_rate": 1.4690888868291779e-06,
      "loss": 0.0,
      "step": 11675
    },
    {
      "epoch": 3.903903903903904,
      "grad_norm": 0.000521348207257688,
      "learning_rate": 1.353230141811104e-06,
      "loss": 0.0,
      "step": 11700
    },
    {
      "epoch": 3.9122455789122457,
      "grad_norm": 0.0002032867632806301,
      "learning_rate": 1.23737139679303e-06,
      "loss": 0.0,
      "step": 11725
    },
    {
      "epoch": 3.920587253920587,
      "grad_norm": 0.0018738268408924341,
      "learning_rate": 1.121512651774956e-06,
      "loss": 0.0009,
      "step": 11750
    },
    {
      "epoch": 3.928928928928929,
      "grad_norm": 0.0014717292506247759,
      "learning_rate": 1.005653906756882e-06,
      "loss": 0.0,
      "step": 11775
    },
    {
      "epoch": 3.9372706039372707,
      "grad_norm": 0.00021975289564579725,
      "learning_rate": 8.897951617388081e-07,
      "loss": 0.0,
      "step": 11800
    },
    {
      "epoch": 3.945612278945612,
      "grad_norm": 0.0005329960258677602,
      "learning_rate": 7.739364167207342e-07,
      "loss": 0.0001,
      "step": 11825
    },
    {
      "epoch": 3.953953953953954,
      "grad_norm": 0.0001847147213993594,
      "learning_rate": 6.580776717026602e-07,
      "loss": 0.0,
      "step": 11850
    },
    {
      "epoch": 3.9622956289622957,
      "grad_norm": 0.00019793544197455049,
      "learning_rate": 5.422189266845861e-07,
      "loss": 0.0,
      "step": 11875
    },
    {
      "epoch": 3.970637303970637,
      "grad_norm": 0.0004685576423071325,
      "learning_rate": 4.2636018166651216e-07,
      "loss": 0.0,
      "step": 11900
    },
    {
      "epoch": 3.978978978978979,
      "grad_norm": 0.00016259195399470627,
      "learning_rate": 3.1050143664843824e-07,
      "loss": 0.0,
      "step": 11925
    },
    {
      "epoch": 3.9873206539873207,
      "grad_norm": 0.0003845692554023117,
      "learning_rate": 1.9464269163036427e-07,
      "loss": 0.0,
      "step": 11950
    },
    {
      "epoch": 3.995662328995662,
      "grad_norm": 0.0001322088937740773,
      "learning_rate": 7.87839466122903e-08,
      "loss": 0.0225,
      "step": 11975
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9991658324991658,
      "eval_auc": 0.9999155405405404,
      "eval_f1": 0.9659863945578231,
      "eval_loss": 0.002970785601064563,
      "eval_precision": 0.9726027397260274,
      "eval_recall": 0.9594594594594594,
      "eval_runtime": 77.7268,
      "eval_samples_per_second": 77.116,
      "eval_steps_per_second": 4.825,
      "step": 11988
    }
  ],
  "logging_steps": 25,
  "max_steps": 11988,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.01
      },
      "attributes": {
        "early_stopping_patience_counter": 3
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.1910999185868456e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}