{
  "best_metric": 0.003335413057357073,
  "best_model_checkpoint": "sj-det-2/checkpoint-8991",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 8991,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.008341675008341674,
      "grad_norm": 16.313989639282227,
      "learning_rate": 1.388888888888889e-06,
      "loss": 0.9123,
      "step": 25
    },
    {
      "epoch": 0.01668335001668335,
      "grad_norm": 8.612642288208008,
      "learning_rate": 2.777777777777778e-06,
      "loss": 0.6485,
      "step": 50
    },
    {
      "epoch": 0.025025025025025027,
      "grad_norm": 4.427529335021973,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.3344,
      "step": 75
    },
    {
      "epoch": 0.0333667000333667,
      "grad_norm": 1.323181390762329,
      "learning_rate": 5.555555555555556e-06,
      "loss": 0.1139,
      "step": 100
    },
    {
      "epoch": 0.04170837504170837,
      "grad_norm": 1.4804118871688843,
      "learning_rate": 6.944444444444445e-06,
      "loss": 0.0661,
      "step": 125
    },
    {
      "epoch": 0.05005005005005005,
      "grad_norm": 0.09656789153814316,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.0585,
      "step": 150
    },
    {
      "epoch": 0.05839172505839173,
      "grad_norm": 0.036666255444288254,
      "learning_rate": 9.722222222222223e-06,
      "loss": 0.0217,
      "step": 175
    },
    {
      "epoch": 0.0667334000667334,
      "grad_norm": 0.024522973224520683,
      "learning_rate": 1.1111111111111112e-05,
      "loss": 0.0832,
      "step": 200
    },
    {
      "epoch": 0.07507507507507508,
      "grad_norm": 0.48196515440940857,
      "learning_rate": 1.25e-05,
      "loss": 0.027,
      "step": 225
    },
    {
      "epoch": 0.08341675008341674,
      "grad_norm": 0.08288281410932541,
      "learning_rate": 1.388888888888889e-05,
      "loss": 0.0154,
      "step": 250
    },
    {
      "epoch": 0.09175842509175842,
      "grad_norm": 0.018726350739598274,
      "learning_rate": 1.527777777777778e-05,
      "loss": 0.0565,
      "step": 275
    },
    {
      "epoch": 0.1001001001001001,
      "grad_norm": 2.332498788833618,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.0314,
      "step": 300
    },
    {
      "epoch": 0.10844177510844177,
      "grad_norm": 0.6220697164535522,
      "learning_rate": 1.8055555555555555e-05,
      "loss": 0.0056,
      "step": 325
    },
    {
      "epoch": 0.11678345011678345,
      "grad_norm": 0.03775189071893692,
      "learning_rate": 1.9444444444444445e-05,
      "loss": 0.0006,
      "step": 350
    },
    {
      "epoch": 0.12512512512512514,
      "grad_norm": 0.014640103094279766,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 0.0559,
      "step": 375
    },
    {
      "epoch": 0.1334668001334668,
      "grad_norm": 0.00695949187502265,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 0.0313,
      "step": 400
    },
    {
      "epoch": 0.14180847514180847,
      "grad_norm": 0.01726640574634075,
      "learning_rate": 2.361111111111111e-05,
      "loss": 0.0142,
      "step": 425
    },
    {
      "epoch": 0.15015015015015015,
      "grad_norm": 0.0029850194696336985,
      "learning_rate": 2.5e-05,
      "loss": 0.0008,
      "step": 450
    },
    {
      "epoch": 0.15849182515849183,
      "grad_norm": 0.009473559446632862,
      "learning_rate": 2.6388888888888892e-05,
      "loss": 0.0627,
      "step": 475
    },
    {
      "epoch": 0.1668335001668335,
      "grad_norm": 0.01642836630344391,
      "learning_rate": 2.777777777777778e-05,
      "loss": 0.0229,
      "step": 500
    },
    {
      "epoch": 0.17517517517517517,
      "grad_norm": 0.037022557109594345,
      "learning_rate": 2.916666666666667e-05,
      "loss": 0.0433,
      "step": 525
    },
    {
      "epoch": 0.18351685018351685,
      "grad_norm": 0.0169846098870039,
      "learning_rate": 3.055555555555556e-05,
      "loss": 0.0387,
      "step": 550
    },
    {
      "epoch": 0.19185852519185853,
      "grad_norm": 0.0074409060180187225,
      "learning_rate": 3.194444444444444e-05,
      "loss": 0.0064,
      "step": 575
    },
    {
      "epoch": 0.2002002002002002,
      "grad_norm": 0.022537339478731155,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.0473,
      "step": 600
    },
    {
      "epoch": 0.20854187520854187,
      "grad_norm": 0.00912486296147108,
      "learning_rate": 3.472222222222222e-05,
      "loss": 0.0071,
      "step": 625
    },
    {
      "epoch": 0.21688355021688355,
      "grad_norm": 0.015622120350599289,
      "learning_rate": 3.611111111111111e-05,
      "loss": 0.018,
      "step": 650
    },
    {
      "epoch": 0.22522522522522523,
      "grad_norm": 0.038481712341308594,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.0609,
      "step": 675
    },
    {
      "epoch": 0.2335669002335669,
      "grad_norm": 0.06651923805475235,
      "learning_rate": 3.888888888888889e-05,
      "loss": 0.0588,
      "step": 700
    },
    {
      "epoch": 0.24190857524190856,
      "grad_norm": 0.013173066079616547,
      "learning_rate": 4.027777777777778e-05,
      "loss": 0.0006,
      "step": 725
    },
    {
      "epoch": 0.2502502502502503,
      "grad_norm": 0.002995034446939826,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.0008,
      "step": 750
    },
    {
      "epoch": 0.2585919252585919,
      "grad_norm": 0.001744964043609798,
      "learning_rate": 4.305555555555556e-05,
      "loss": 0.0002,
      "step": 775
    },
    {
      "epoch": 0.2669336002669336,
      "grad_norm": 0.5510344505310059,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 0.0373,
      "step": 800
    },
    {
      "epoch": 0.2752752752752753,
      "grad_norm": 0.0068357582204043865,
      "learning_rate": 4.5833333333333334e-05,
      "loss": 0.0269,
      "step": 825
    },
    {
      "epoch": 0.28361695028361694,
      "grad_norm": 0.014822622761130333,
      "learning_rate": 4.722222222222222e-05,
      "loss": 0.1029,
      "step": 850
    },
    {
      "epoch": 0.29195862529195865,
      "grad_norm": 0.006352724973112345,
      "learning_rate": 4.8611111111111115e-05,
      "loss": 0.0527,
      "step": 875
    },
    {
      "epoch": 0.3003003003003003,
      "grad_norm": 0.006285039242357016,
      "learning_rate": 5e-05,
      "loss": 0.0007,
      "step": 900
    },
    {
      "epoch": 0.30864197530864196,
      "grad_norm": 0.21172092854976654,
      "learning_rate": 4.984550735384996e-05,
      "loss": 0.001,
      "step": 925
    },
    {
      "epoch": 0.31698365031698367,
      "grad_norm": 0.025125497952103615,
      "learning_rate": 4.969101470769992e-05,
      "loss": 0.0001,
      "step": 950
    },
    {
      "epoch": 0.3253253253253253,
      "grad_norm": 0.0006277966895140707,
      "learning_rate": 4.9536522061549875e-05,
      "loss": 0.0,
      "step": 975
    },
    {
      "epoch": 0.333667000333667,
      "grad_norm": 0.0005465027061291039,
      "learning_rate": 4.9382029415399826e-05,
      "loss": 0.0,
      "step": 1000
    },
    {
      "epoch": 0.3420086753420087,
      "grad_norm": 0.0011532187927514315,
      "learning_rate": 4.922753676924979e-05,
      "loss": 0.0303,
      "step": 1025
    },
    {
      "epoch": 0.35035035035035034,
      "grad_norm": 0.0010950276628136635,
      "learning_rate": 4.907304412309974e-05,
      "loss": 0.0001,
      "step": 1050
    },
    {
      "epoch": 0.35869202535869205,
      "grad_norm": 0.0015246294206008315,
      "learning_rate": 4.89185514769497e-05,
      "loss": 0.0007,
      "step": 1075
    },
    {
      "epoch": 0.3670337003670337,
      "grad_norm": 0.0425708144903183,
      "learning_rate": 4.876405883079966e-05,
      "loss": 0.0592,
      "step": 1100
    },
    {
      "epoch": 0.37537537537537535,
      "grad_norm": 0.001601433614268899,
      "learning_rate": 4.8609566184649615e-05,
      "loss": 0.001,
      "step": 1125
    },
    {
      "epoch": 0.38371705038371706,
      "grad_norm": 0.04269692674279213,
      "learning_rate": 4.8455073538499566e-05,
      "loss": 0.052,
      "step": 1150
    },
    {
      "epoch": 0.3920587253920587,
      "grad_norm": 10.334559440612793,
      "learning_rate": 4.830058089234953e-05,
      "loss": 0.0149,
      "step": 1175
    },
    {
      "epoch": 0.4004004004004004,
      "grad_norm": 0.0104717668145895,
      "learning_rate": 4.814608824619948e-05,
      "loss": 0.0524,
      "step": 1200
    },
    {
      "epoch": 0.4087420754087421,
      "grad_norm": 0.004411773756146431,
      "learning_rate": 4.799159560004944e-05,
      "loss": 0.0003,
      "step": 1225
    },
    {
      "epoch": 0.41708375041708373,
      "grad_norm": 0.0019501554779708385,
      "learning_rate": 4.7837102953899396e-05,
      "loss": 0.0001,
      "step": 1250
    },
    {
      "epoch": 0.42542542542542544,
      "grad_norm": 0.001354438834823668,
      "learning_rate": 4.7682610307749354e-05,
      "loss": 0.0011,
      "step": 1275
    },
    {
      "epoch": 0.4337671004337671,
      "grad_norm": 0.0017444408731535077,
      "learning_rate": 4.752811766159931e-05,
      "loss": 0.0003,
      "step": 1300
    },
    {
      "epoch": 0.44210877544210875,
      "grad_norm": 0.00919169932603836,
      "learning_rate": 4.737362501544927e-05,
      "loss": 0.06,
      "step": 1325
    },
    {
      "epoch": 0.45045045045045046,
      "grad_norm": 0.001080819871276617,
      "learning_rate": 4.721913236929923e-05,
      "loss": 0.0031,
      "step": 1350
    },
    {
      "epoch": 0.4587921254587921,
      "grad_norm": 0.03211037814617157,
      "learning_rate": 4.706463972314918e-05,
      "loss": 0.0454,
      "step": 1375
    },
    {
      "epoch": 0.4671338004671338,
      "grad_norm": 0.025529522448778152,
      "learning_rate": 4.691014707699914e-05,
      "loss": 0.0036,
      "step": 1400
    },
    {
      "epoch": 0.4754754754754755,
      "grad_norm": 0.0008930154144763947,
      "learning_rate": 4.675565443084909e-05,
      "loss": 0.0451,
      "step": 1425
    },
    {
      "epoch": 0.4838171504838171,
      "grad_norm": 0.023177633062005043,
      "learning_rate": 4.660116178469905e-05,
      "loss": 0.0028,
      "step": 1450
    },
    {
      "epoch": 0.49215882549215884,
      "grad_norm": 0.0005259920726530254,
      "learning_rate": 4.644666913854901e-05,
      "loss": 0.0461,
      "step": 1475
    },
    {
      "epoch": 0.5005005005005005,
      "grad_norm": 0.007525968365371227,
      "learning_rate": 4.6292176492398966e-05,
      "loss": 0.0403,
      "step": 1500
    },
    {
      "epoch": 0.5088421755088421,
      "grad_norm": 0.005125482566654682,
      "learning_rate": 4.613768384624892e-05,
      "loss": 0.0015,
      "step": 1525
    },
    {
      "epoch": 0.5171838505171839,
      "grad_norm": 0.00863750372081995,
      "learning_rate": 4.598319120009888e-05,
      "loss": 0.0205,
      "step": 1550
    },
    {
      "epoch": 0.5255255255255256,
      "grad_norm": 0.0001800910395104438,
      "learning_rate": 4.582869855394883e-05,
      "loss": 0.0037,
      "step": 1575
    },
    {
      "epoch": 0.5338672005338672,
      "grad_norm": 0.00019970687571913004,
      "learning_rate": 4.567420590779879e-05,
      "loss": 0.0152,
      "step": 1600
    },
    {
      "epoch": 0.5422088755422089,
      "grad_norm": 0.011676608584821224,
      "learning_rate": 4.551971326164875e-05,
      "loss": 0.0635,
      "step": 1625
    },
    {
      "epoch": 0.5505505505505506,
      "grad_norm": 0.0008765487000346184,
      "learning_rate": 4.5365220615498706e-05,
      "loss": 0.0018,
      "step": 1650
    },
    {
      "epoch": 0.5588922255588922,
      "grad_norm": 0.0001437152095604688,
      "learning_rate": 4.5210727969348656e-05,
      "loss": 0.0026,
      "step": 1675
    },
    {
      "epoch": 0.5672339005672339,
      "grad_norm": 0.00043227593414485455,
      "learning_rate": 4.505623532319862e-05,
      "loss": 0.0001,
      "step": 1700
    },
    {
      "epoch": 0.5755755755755756,
      "grad_norm": 0.00048412373871542513,
      "learning_rate": 4.490174267704857e-05,
      "loss": 0.0164,
      "step": 1725
    },
    {
      "epoch": 0.5839172505839173,
      "grad_norm": 7.867912769317627,
      "learning_rate": 4.474725003089853e-05,
      "loss": 0.0548,
      "step": 1750
    },
    {
      "epoch": 0.5922589255922589,
      "grad_norm": 5.28706169128418,
      "learning_rate": 4.459275738474849e-05,
      "loss": 0.0287,
      "step": 1775
    },
    {
      "epoch": 0.6006006006006006,
      "grad_norm": 0.005814212840050459,
      "learning_rate": 4.4438264738598445e-05,
      "loss": 0.0353,
      "step": 1800
    },
    {
      "epoch": 0.6089422756089423,
      "grad_norm": 0.00043659639777615666,
      "learning_rate": 4.42837720924484e-05,
      "loss": 0.0005,
      "step": 1825
    },
    {
      "epoch": 0.6172839506172839,
      "grad_norm": 0.009556015022099018,
      "learning_rate": 4.412927944629836e-05,
      "loss": 0.0004,
      "step": 1850
    },
    {
      "epoch": 0.6256256256256256,
      "grad_norm": 0.0006456194678321481,
      "learning_rate": 4.397478680014832e-05,
      "loss": 0.0003,
      "step": 1875
    },
    {
      "epoch": 0.6339673006339673,
      "grad_norm": 0.0007494451128877699,
      "learning_rate": 4.382029415399827e-05,
      "loss": 0.0001,
      "step": 1900
    },
    {
      "epoch": 0.6423089756423089,
      "grad_norm": 0.00023574185615871102,
      "learning_rate": 4.366580150784823e-05,
      "loss": 0.028,
      "step": 1925
    },
    {
      "epoch": 0.6506506506506506,
      "grad_norm": 0.0008282416383735836,
      "learning_rate": 4.3511308861698184e-05,
      "loss": 0.0181,
      "step": 1950
    },
    {
      "epoch": 0.6589923256589924,
      "grad_norm": 0.0017523517599329352,
      "learning_rate": 4.335681621554814e-05,
      "loss": 0.0761,
      "step": 1975
    },
    {
      "epoch": 0.667334000667334,
      "grad_norm": 0.0005382935050874949,
      "learning_rate": 4.32023235693981e-05,
      "loss": 0.0007,
      "step": 2000
    },
    {
      "epoch": 0.6756756756756757,
      "grad_norm": 0.0023508896119892597,
      "learning_rate": 4.304783092324806e-05,
      "loss": 0.0001,
      "step": 2025
    },
    {
      "epoch": 0.6840173506840174,
      "grad_norm": 0.02000766433775425,
      "learning_rate": 4.289333827709801e-05,
      "loss": 0.0461,
      "step": 2050
    },
    {
      "epoch": 0.6923590256923591,
      "grad_norm": 0.0016236711526289582,
      "learning_rate": 4.273884563094797e-05,
      "loss": 0.0015,
      "step": 2075
    },
    {
      "epoch": 0.7007007007007007,
      "grad_norm": 0.01102663204073906,
      "learning_rate": 4.2584352984797923e-05,
      "loss": 0.0,
      "step": 2100
    },
    {
      "epoch": 0.7090423757090424,
      "grad_norm": 7.1402587890625,
      "learning_rate": 4.242986033864788e-05,
      "loss": 0.0458,
      "step": 2125
    },
    {
      "epoch": 0.7173840507173841,
      "grad_norm": 0.029623601585626602,
      "learning_rate": 4.227536769249784e-05,
      "loss": 0.0265,
      "step": 2150
    },
    {
      "epoch": 0.7257257257257257,
      "grad_norm": 0.04872233048081398,
      "learning_rate": 4.2120875046347796e-05,
      "loss": 0.0418,
      "step": 2175
    },
    {
      "epoch": 0.7340674007340674,
      "grad_norm": 0.01899750903248787,
      "learning_rate": 4.1966382400197754e-05,
      "loss": 0.0797,
      "step": 2200
    },
    {
      "epoch": 0.7424090757424091,
      "grad_norm": 0.014597664587199688,
      "learning_rate": 4.181188975404771e-05,
      "loss": 0.0239,
      "step": 2225
    },
    {
      "epoch": 0.7507507507507507,
      "grad_norm": 0.001034777145832777,
      "learning_rate": 4.165739710789767e-05,
      "loss": 0.0067,
      "step": 2250
    },
    {
      "epoch": 0.7590924257590924,
      "grad_norm": 0.0015550162643194199,
      "learning_rate": 4.150290446174762e-05,
      "loss": 0.0482,
      "step": 2275
    },
    {
      "epoch": 0.7674341007674341,
      "grad_norm": 0.013912219554185867,
      "learning_rate": 4.1348411815597585e-05,
      "loss": 0.0248,
      "step": 2300
    },
    {
      "epoch": 0.7757757757757757,
      "grad_norm": 0.0020062343683093786,
      "learning_rate": 4.1193919169447536e-05,
      "loss": 0.0156,
      "step": 2325
    },
    {
      "epoch": 0.7841174507841174,
      "grad_norm": 6.4558515548706055,
      "learning_rate": 4.1039426523297493e-05,
      "loss": 0.0581,
      "step": 2350
    },
    {
      "epoch": 0.7924591257924591,
      "grad_norm": 0.020143140107393265,
      "learning_rate": 4.088493387714745e-05,
      "loss": 0.0033,
      "step": 2375
    },
    {
      "epoch": 0.8008008008008008,
      "grad_norm": 0.00565760675817728,
      "learning_rate": 4.073044123099741e-05,
      "loss": 0.0003,
      "step": 2400
    },
    {
      "epoch": 0.8091424758091424,
      "grad_norm": 0.0011780583299696445,
      "learning_rate": 4.057594858484736e-05,
      "loss": 0.0003,
      "step": 2425
    },
    {
      "epoch": 0.8174841508174842,
      "grad_norm": 0.001397665124386549,
      "learning_rate": 4.0421455938697324e-05,
      "loss": 0.0033,
      "step": 2450
    },
    {
      "epoch": 0.8258258258258259,
      "grad_norm": 0.11108289659023285,
      "learning_rate": 4.0266963292547275e-05,
      "loss": 0.0693,
      "step": 2475
    },
    {
      "epoch": 0.8341675008341675,
      "grad_norm": 0.0012789678294211626,
      "learning_rate": 4.011247064639723e-05,
      "loss": 0.0012,
      "step": 2500
    },
    {
      "epoch": 0.8425091758425092,
      "grad_norm": 0.0016298526898026466,
      "learning_rate": 3.995797800024719e-05,
      "loss": 0.0001,
      "step": 2525
    },
    {
      "epoch": 0.8508508508508509,
      "grad_norm": 0.0012462561717256904,
      "learning_rate": 3.980348535409715e-05,
      "loss": 0.0,
      "step": 2550
    },
    {
      "epoch": 0.8591925258591925,
      "grad_norm": 0.0011716202134266496,
      "learning_rate": 3.9648992707947106e-05,
      "loss": 0.0008,
      "step": 2575
    },
    {
      "epoch": 0.8675342008675342,
      "grad_norm": 0.0004461667558643967,
      "learning_rate": 3.9494500061797063e-05,
      "loss": 0.0013,
      "step": 2600
    },
    {
      "epoch": 0.8758758758758759,
      "grad_norm": 0.004892207216471434,
      "learning_rate": 3.934000741564702e-05,
      "loss": 0.0444,
      "step": 2625
    },
    {
      "epoch": 0.8842175508842175,
      "grad_norm": 0.0013353752437978983,
      "learning_rate": 3.918551476949697e-05,
      "loss": 0.0014,
      "step": 2650
    },
    {
      "epoch": 0.8925592258925592,
      "grad_norm": 0.05706316977739334,
      "learning_rate": 3.9031022123346936e-05,
      "loss": 0.1006,
      "step": 2675
    },
    {
      "epoch": 0.9009009009009009,
      "grad_norm": 0.0032118011731654406,
      "learning_rate": 3.887652947719689e-05,
      "loss": 0.0004,
      "step": 2700
    },
    {
      "epoch": 0.9092425759092426,
      "grad_norm": 0.002578709740191698,
      "learning_rate": 3.8722036831046845e-05,
      "loss": 0.0049,
      "step": 2725
    },
    {
      "epoch": 0.9175842509175842,
      "grad_norm": 0.007876419462263584,
      "learning_rate": 3.85675441848968e-05,
      "loss": 0.0393,
      "step": 2750
    },
    {
      "epoch": 0.9259259259259259,
      "grad_norm": 0.01825695112347603,
      "learning_rate": 3.841305153874676e-05,
      "loss": 0.0313,
      "step": 2775
    },
    {
      "epoch": 0.9342676009342676,
      "grad_norm": 0.018400903791189194,
      "learning_rate": 3.825855889259671e-05,
      "loss": 0.0362,
      "step": 2800
    },
    {
      "epoch": 0.9426092759426092,
      "grad_norm": 9.522905349731445,
      "learning_rate": 3.8104066246446676e-05,
      "loss": 0.0044,
      "step": 2825
    },
    {
      "epoch": 0.950950950950951,
      "grad_norm": 0.055028580129146576,
      "learning_rate": 3.794957360029663e-05,
      "loss": 0.0046,
      "step": 2850
    },
    {
      "epoch": 0.9592926259592927,
      "grad_norm": 0.0008863616967573762,
      "learning_rate": 3.7795080954146584e-05,
      "loss": 0.0007,
      "step": 2875
    },
    {
      "epoch": 0.9676343009676343,
      "grad_norm": 0.001261973986402154,
      "learning_rate": 3.764058830799654e-05,
      "loss": 0.0001,
      "step": 2900
    },
    {
      "epoch": 0.975975975975976,
      "grad_norm": 0.04143986105918884,
      "learning_rate": 3.74860956618465e-05,
      "loss": 0.0871,
      "step": 2925
    },
    {
      "epoch": 0.9843176509843177,
      "grad_norm": 0.0786428451538086,
      "learning_rate": 3.733160301569645e-05,
      "loss": 0.01,
      "step": 2950
    },
    {
      "epoch": 0.9926593259926594,
      "grad_norm": 0.0012078924337401986,
      "learning_rate": 3.7177110369546415e-05,
      "loss": 0.0392,
      "step": 2975
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9978311644978312,
      "eval_auc": 0.9945649196493791,
      "eval_f1": 0.9078014184397163,
      "eval_loss": 0.009599411860108376,
      "eval_precision": 0.9552238805970149,
      "eval_recall": 0.8648648648648649,
      "eval_runtime": 120.7986,
      "eval_samples_per_second": 49.62,
      "eval_steps_per_second": 3.104,
      "step": 2997
    },
    {
      "epoch": 1.001001001001001,
      "grad_norm": 0.006058184430003166,
      "learning_rate": 3.7022617723396366e-05,
      "loss": 0.041,
      "step": 3000
    },
    {
      "epoch": 1.0093426760093427,
      "grad_norm": 0.003738154424354434,
      "learning_rate": 3.6868125077246324e-05,
      "loss": 0.0005,
      "step": 3025
    },
    {
      "epoch": 1.0176843510176843,
      "grad_norm": 0.00978489127010107,
      "learning_rate": 3.671363243109628e-05,
      "loss": 0.0452,
      "step": 3050
    },
    {
      "epoch": 1.026026026026026,
      "grad_norm": 0.01102358940988779,
      "learning_rate": 3.655913978494624e-05,
      "loss": 0.0005,
      "step": 3075
    },
    {
      "epoch": 1.0343677010343677,
      "grad_norm": 0.008122111670672894,
      "learning_rate": 3.64046471387962e-05,
      "loss": 0.0003,
      "step": 3100
    },
    {
      "epoch": 1.0427093760427093,
      "grad_norm": 0.0037865727208554745,
      "learning_rate": 3.6250154492646154e-05,
      "loss": 0.0046,
      "step": 3125
    },
    {
      "epoch": 1.0510510510510511,
      "grad_norm": 0.0031646236311644316,
      "learning_rate": 3.609566184649611e-05,
      "loss": 0.0005,
      "step": 3150
    },
    {
      "epoch": 1.0593927260593927,
      "grad_norm": 0.0017720437608659267,
      "learning_rate": 3.594116920034606e-05,
      "loss": 0.0001,
      "step": 3175
    },
    {
      "epoch": 1.0677344010677343,
      "grad_norm": 0.011880212463438511,
      "learning_rate": 3.578667655419603e-05,
      "loss": 0.0286,
      "step": 3200
    },
    {
      "epoch": 1.0760760760760761,
      "grad_norm": 0.00772507768124342,
      "learning_rate": 3.563218390804598e-05,
      "loss": 0.0001,
      "step": 3225
    },
    {
      "epoch": 1.0844177510844177,
      "grad_norm": 0.012478929944336414,
      "learning_rate": 3.5477691261895936e-05,
      "loss": 0.0001,
      "step": 3250
    },
    {
      "epoch": 1.0927594260927593,
      "grad_norm": 0.004284675233066082,
      "learning_rate": 3.5323198615745894e-05,
      "loss": 0.0001,
      "step": 3275
    },
    {
      "epoch": 1.1011011011011012,
      "grad_norm": 0.0012670083669945598,
      "learning_rate": 3.516870596959585e-05,
      "loss": 0.0001,
      "step": 3300
    },
    {
      "epoch": 1.1094427761094428,
      "grad_norm": 0.0011124334996566176,
      "learning_rate": 3.50142133234458e-05,
      "loss": 0.0381,
      "step": 3325
    },
    {
      "epoch": 1.1177844511177844,
      "grad_norm": 0.35427987575531006,
      "learning_rate": 3.485972067729576e-05,
      "loss": 0.0306,
      "step": 3350
    },
    {
      "epoch": 1.1261261261261262,
      "grad_norm": 0.0010480162454769015,
      "learning_rate": 3.470522803114572e-05,
      "loss": 0.0002,
      "step": 3375
    },
    {
      "epoch": 1.1344678011344678,
      "grad_norm": 0.01626741513609886,
      "learning_rate": 3.4550735384995675e-05,
      "loss": 0.0049,
      "step": 3400
    },
    {
      "epoch": 1.1428094761428094,
      "grad_norm": 0.0009041897719725966,
      "learning_rate": 3.439624273884563e-05,
      "loss": 0.0033,
      "step": 3425
    },
    {
      "epoch": 1.1511511511511512,
      "grad_norm": 0.005210678093135357,
      "learning_rate": 3.424175009269559e-05,
      "loss": 0.0385,
      "step": 3450
    },
    {
      "epoch": 1.1594928261594928,
      "grad_norm": 0.00467864703387022,
      "learning_rate": 3.408725744654555e-05,
      "loss": 0.0001,
      "step": 3475
    },
    {
      "epoch": 1.1678345011678344,
      "grad_norm": 0.019908057525753975,
      "learning_rate": 3.39327648003955e-05,
      "loss": 0.0001,
      "step": 3500
    },
    {
      "epoch": 1.1761761761761762,
      "grad_norm": 0.0005183572066016495,
      "learning_rate": 3.3778272154245464e-05,
      "loss": 0.0004,
      "step": 3525
    },
    {
      "epoch": 1.1845178511845178,
      "grad_norm": 0.00818272028118372,
      "learning_rate": 3.3623779508095414e-05,
      "loss": 0.0454,
      "step": 3550
    },
    {
      "epoch": 1.1928595261928596,
      "grad_norm": 0.0328977033495903,
      "learning_rate": 3.346928686194537e-05,
      "loss": 0.0003,
      "step": 3575
    },
    {
      "epoch": 1.2012012012012012,
      "grad_norm": 6.452710151672363,
      "learning_rate": 3.331479421579533e-05,
      "loss": 0.0432,
      "step": 3600
    },
    {
      "epoch": 1.2095428762095428,
      "grad_norm": 0.005078097339719534,
      "learning_rate": 3.316030156964529e-05,
      "loss": 0.0322,
      "step": 3625
    },
    {
      "epoch": 1.2178845512178846,
      "grad_norm": 0.0003529058594722301,
      "learning_rate": 3.300580892349524e-05,
      "loss": 0.0003,
      "step": 3650
    },
    {
      "epoch": 1.2262262262262262,
      "grad_norm": 0.0013959240168333054,
      "learning_rate": 3.28513162773452e-05,
      "loss": 0.0849,
      "step": 3675
    },
    {
      "epoch": 1.2345679012345678,
      "grad_norm": 0.016115982085466385,
      "learning_rate": 3.2696823631195154e-05,
      "loss": 0.056,
      "step": 3700
    },
    {
      "epoch": 1.2429095762429097,
      "grad_norm": 0.0022523789666593075,
      "learning_rate": 3.254233098504511e-05,
      "loss": 0.0002,
      "step": 3725
    },
    {
      "epoch": 1.2512512512512513,
      "grad_norm": 0.016599655151367188,
      "learning_rate": 3.238783833889507e-05,
      "loss": 0.0319,
      "step": 3750
    },
    {
      "epoch": 1.2595929262595928,
      "grad_norm": 0.0018683940870687366,
      "learning_rate": 3.223334569274503e-05,
      "loss": 0.0227,
      "step": 3775
    },
    {
      "epoch": 1.2679346012679347,
      "grad_norm": 0.002231100806966424,
      "learning_rate": 3.207885304659498e-05,
      "loss": 0.0001,
      "step": 3800
    },
    {
      "epoch": 1.2762762762762763,
      "grad_norm": 0.005379370413720608,
      "learning_rate": 3.192436040044494e-05,
      "loss": 0.0439,
      "step": 3825
    },
    {
      "epoch": 1.284617951284618,
      "grad_norm": 0.003467306261882186,
      "learning_rate": 3.176986775429489e-05,
      "loss": 0.0025,
      "step": 3850
    },
    {
      "epoch": 1.2929596262929597,
      "grad_norm": 0.0015170868718996644,
      "learning_rate": 3.161537510814485e-05,
      "loss": 0.0037,
      "step": 3875
    },
    {
      "epoch": 1.3013013013013013,
      "grad_norm": 0.00604633055627346,
      "learning_rate": 3.146088246199481e-05,
      "loss": 0.0001,
      "step": 3900
    },
    {
      "epoch": 1.309642976309643,
      "grad_norm": 10.07129955291748,
      "learning_rate": 3.1306389815844766e-05,
      "loss": 0.0232,
      "step": 3925
    },
    {
      "epoch": 1.3179846513179847,
      "grad_norm": 0.0015994218410924077,
      "learning_rate": 3.1151897169694724e-05,
      "loss": 0.0004,
      "step": 3950
    },
    {
      "epoch": 1.3263263263263263,
      "grad_norm": 0.009616601280868053,
      "learning_rate": 3.099740452354468e-05,
      "loss": 0.0165,
      "step": 3975
    },
    {
      "epoch": 1.3346680013346681,
      "grad_norm": 0.009371902793645859,
      "learning_rate": 3.084291187739464e-05,
      "loss": 0.0008,
      "step": 4000
    },
    {
      "epoch": 1.3430096763430097,
      "grad_norm": 0.3239502012729645,
      "learning_rate": 3.068841923124459e-05,
      "loss": 0.0026,
      "step": 4025
    },
    {
      "epoch": 1.3513513513513513,
      "grad_norm": 0.00045479455729946494,
      "learning_rate": 3.0533926585094554e-05,
      "loss": 0.0002,
      "step": 4050
    },
    {
      "epoch": 1.3596930263596931,
      "grad_norm": 6.515894889831543,
      "learning_rate": 3.0379433938944505e-05,
      "loss": 0.0591,
      "step": 4075
    },
    {
      "epoch": 1.3680347013680347,
      "grad_norm": 0.0006543731433339417,
      "learning_rate": 3.0224941292794463e-05,
      "loss": 0.0527,
      "step": 4100
    },
    {
      "epoch": 1.3763763763763763,
      "grad_norm": 0.0006008436321280897,
      "learning_rate": 3.007044864664442e-05,
      "loss": 0.0278,
      "step": 4125
    },
    {
      "epoch": 1.3847180513847182,
      "grad_norm": 0.02133520320057869,
      "learning_rate": 2.991595600049438e-05,
      "loss": 0.0002,
      "step": 4150
    },
    {
      "epoch": 1.3930597263930597,
      "grad_norm": 0.0017505575669929385,
      "learning_rate": 2.9761463354344333e-05,
      "loss": 0.0019,
      "step": 4175
    },
    {
      "epoch": 1.4014014014014013,
      "grad_norm": 0.0024641742929816246,
      "learning_rate": 2.9606970708194294e-05,
      "loss": 0.0001,
      "step": 4200
    },
    {
      "epoch": 1.4097430764097432,
      "grad_norm": 0.0026479989755898714,
      "learning_rate": 2.9452478062044248e-05,
      "loss": 0.0005,
      "step": 4225
    },
    {
      "epoch": 1.4180847514180848,
      "grad_norm": 0.0036438798997551203,
      "learning_rate": 2.9297985415894202e-05,
      "loss": 0.0001,
      "step": 4250
    },
    {
      "epoch": 1.4264264264264264,
      "grad_norm": 0.005783462896943092,
      "learning_rate": 2.9143492769744163e-05,
      "loss": 0.0686,
      "step": 4275
    },
    {
      "epoch": 1.4347681014347682,
      "grad_norm": 0.024165183305740356,
      "learning_rate": 2.8989000123594118e-05,
      "loss": 0.0198,
      "step": 4300
    },
    {
      "epoch": 1.4431097764431098,
      "grad_norm": 0.04456571117043495,
      "learning_rate": 2.8834507477444072e-05,
      "loss": 0.0007,
      "step": 4325
    },
    {
      "epoch": 1.4514514514514514,
      "grad_norm": 0.0014297548914328218,
      "learning_rate": 2.8680014831294033e-05,
      "loss": 0.0009,
      "step": 4350
    },
    {
      "epoch": 1.4597931264597932,
      "grad_norm": 0.002099397825077176,
      "learning_rate": 2.8525522185143987e-05,
      "loss": 0.0116,
      "step": 4375
    },
    {
      "epoch": 1.4681348014681348,
      "grad_norm": 0.0008780402131378651,
      "learning_rate": 2.837102953899394e-05,
      "loss": 0.0001,
      "step": 4400
    },
    {
      "epoch": 1.4764764764764764,
      "grad_norm": 0.00024115787527989596,
      "learning_rate": 2.8216536892843903e-05,
      "loss": 0.0216,
      "step": 4425
    },
    {
      "epoch": 1.4848181514848182,
      "grad_norm": 0.0003840152348857373,
      "learning_rate": 2.8062044246693857e-05,
      "loss": 0.0361,
      "step": 4450
    },
    {
      "epoch": 1.4931598264931598,
      "grad_norm": 0.0005036329384893179,
      "learning_rate": 2.7907551600543815e-05,
      "loss": 0.0001,
      "step": 4475
    },
    {
      "epoch": 1.5015015015015014,
      "grad_norm": 0.0002450717438478023,
      "learning_rate": 2.7753058954393772e-05,
      "loss": 0.0001,
      "step": 4500
    },
    {
      "epoch": 1.5098431765098432,
      "grad_norm": 0.0007805937202647328,
      "learning_rate": 2.759856630824373e-05,
      "loss": 0.0001,
      "step": 4525
    },
    {
      "epoch": 1.5181848515181848,
      "grad_norm": 0.0013042781502008438,
      "learning_rate": 2.7444073662093684e-05,
      "loss": 0.0645,
      "step": 4550
    },
    {
      "epoch": 1.5265265265265264,
      "grad_norm": 0.0023273963015526533,
      "learning_rate": 2.7289581015943645e-05,
      "loss": 0.0016,
      "step": 4575
    },
    {
      "epoch": 1.5348682015348682,
      "grad_norm": 0.0012803663266822696,
      "learning_rate": 2.71350883697936e-05,
      "loss": 0.0001,
      "step": 4600
    },
    {
      "epoch": 1.5432098765432098,
      "grad_norm": 0.0039305612444877625,
      "learning_rate": 2.6980595723643554e-05,
      "loss": 0.0001,
      "step": 4625
    },
    {
      "epoch": 1.5515515515515514,
      "grad_norm": 0.0023296186700463295,
      "learning_rate": 2.6826103077493515e-05,
      "loss": 0.0001,
      "step": 4650
    },
    {
      "epoch": 1.5598932265598933,
      "grad_norm": 0.0008701834594830871,
      "learning_rate": 2.667161043134347e-05,
      "loss": 0.0306,
      "step": 4675
    },
    {
      "epoch": 1.5682349015682349,
      "grad_norm": 0.009118028916418552,
      "learning_rate": 2.6517117785193424e-05,
      "loss": 0.0004,
      "step": 4700
    },
    {
      "epoch": 1.5765765765765765,
      "grad_norm": 0.004671914968639612,
      "learning_rate": 2.6362625139043385e-05,
      "loss": 0.0003,
      "step": 4725
    },
    {
      "epoch": 1.5849182515849183,
      "grad_norm": 0.0014299708418548107,
      "learning_rate": 2.620813249289334e-05,
      "loss": 0.0001,
      "step": 4750
    },
    {
      "epoch": 1.5932599265932599,
      "grad_norm": 0.0011580680729821324,
      "learning_rate": 2.6053639846743293e-05,
      "loss": 0.0001,
      "step": 4775
    },
    {
      "epoch": 1.6016016016016015,
      "grad_norm": 0.0004150031309109181,
      "learning_rate": 2.5899147200593254e-05,
      "loss": 0.0201,
      "step": 4800
    },
    {
      "epoch": 1.6099432766099433,
      "grad_norm": 0.15517137944698334,
      "learning_rate": 2.574465455444321e-05,
      "loss": 0.0133,
      "step": 4825
    },
    {
      "epoch": 1.6182849516182851,
      "grad_norm": 0.0004611278127413243,
      "learning_rate": 2.5590161908293163e-05,
      "loss": 0.0004,
      "step": 4850
    },
    {
      "epoch": 1.6266266266266265,
      "grad_norm": 7.501170635223389,
      "learning_rate": 2.5435669262143124e-05,
      "loss": 0.0056,
      "step": 4875
    },
    {
      "epoch": 1.6349683016349683,
      "grad_norm": 0.0007478469051420689,
      "learning_rate": 2.5281176615993078e-05,
      "loss": 0.0002,
      "step": 4900
    },
    {
      "epoch": 1.6433099766433101,
      "grad_norm": 2.6761367321014404,
      "learning_rate": 2.5126683969843036e-05,
      "loss": 0.0319,
      "step": 4925
    },
    {
      "epoch": 1.6516516516516515,
      "grad_norm": 0.0006212076405063272,
      "learning_rate": 2.4972191323692994e-05,
      "loss": 0.0,
      "step": 4950
    },
    {
      "epoch": 1.6599933266599933,
      "grad_norm": 0.0011125955497846007,
      "learning_rate": 2.481769867754295e-05,
      "loss": 0.0001,
      "step": 4975
    },
    {
      "epoch": 1.6683350016683351,
      "grad_norm": 9.793978097150102e-05,
      "learning_rate": 2.466320603139291e-05,
      "loss": 0.0,
      "step": 5000
    },
    {
      "epoch": 1.6766766766766765,
      "grad_norm": 0.0029056610073894262,
      "learning_rate": 2.4508713385242863e-05,
      "loss": 0.0001,
      "step": 5025
    },
    {
      "epoch": 1.6850183516850183,
      "grad_norm": 0.0007343544857576489,
      "learning_rate": 2.435422073909282e-05,
      "loss": 0.0521,
      "step": 5050
    },
    {
      "epoch": 1.6933600266933602,
      "grad_norm": 0.014967703260481358,
      "learning_rate": 2.419972809294278e-05,
      "loss": 0.0002,
      "step": 5075
    },
    {
      "epoch": 1.7017017017017015,
      "grad_norm": 0.0071108159609138966,
      "learning_rate": 2.4045235446792733e-05,
      "loss": 0.0362,
      "step": 5100
    },
    {
      "epoch": 1.7100433767100434,
      "grad_norm": 0.004571543540805578,
      "learning_rate": 2.389074280064269e-05,
      "loss": 0.0047,
      "step": 5125
    },
    {
      "epoch": 1.7183850517183852,
      "grad_norm": 0.007166516967117786,
      "learning_rate": 2.3736250154492648e-05,
      "loss": 0.0028,
      "step": 5150
    },
    {
      "epoch": 1.7267267267267268,
      "grad_norm": 0.008930371142923832,
      "learning_rate": 2.3581757508342603e-05,
      "loss": 0.0396,
      "step": 5175
    },
    {
      "epoch": 1.7350684017350684,
      "grad_norm": 0.007438777945935726,
      "learning_rate": 2.342726486219256e-05,
      "loss": 0.0004,
      "step": 5200
    },
    {
      "epoch": 1.7434100767434102,
      "grad_norm": 0.0038795331493020058,
      "learning_rate": 2.3272772216042518e-05,
      "loss": 0.0257,
      "step": 5225
    },
    {
      "epoch": 1.7517517517517518,
      "grad_norm": 0.003188981907442212,
      "learning_rate": 2.3118279569892472e-05,
      "loss": 0.0003,
      "step": 5250
    },
    {
      "epoch": 1.7600934267600934,
      "grad_norm": 0.00612386642023921,
      "learning_rate": 2.296378692374243e-05,
      "loss": 0.0321,
      "step": 5275
    },
    {
      "epoch": 1.7684351017684352,
      "grad_norm": 0.0007353666005656123,
      "learning_rate": 2.2809294277592388e-05,
      "loss": 0.0026,
      "step": 5300
    },
    {
      "epoch": 1.7767767767767768,
      "grad_norm": 0.006964649073779583,
      "learning_rate": 2.2654801631442342e-05,
      "loss": 0.0009,
      "step": 5325
    },
    {
      "epoch": 1.7851184517851184,
      "grad_norm": 0.009796655736863613,
      "learning_rate": 2.25003089852923e-05,
      "loss": 0.0542,
      "step": 5350
    },
    {
      "epoch": 1.7934601267934602,
      "grad_norm": 0.00797263439744711,
      "learning_rate": 2.2345816339142257e-05,
      "loss": 0.001,
      "step": 5375
    },
    {
      "epoch": 1.8018018018018018,
      "grad_norm": 0.0027469745837152004,
      "learning_rate": 2.2191323692992215e-05,
      "loss": 0.0002,
      "step": 5400
    },
    {
      "epoch": 1.8101434768101434,
      "grad_norm": 0.0032520880922675133,
      "learning_rate": 2.2036831046842172e-05,
      "loss": 0.0162,
      "step": 5425
    },
    {
      "epoch": 1.8184851518184852,
      "grad_norm": 0.0021738584619015455,
      "learning_rate": 2.188233840069213e-05,
      "loss": 0.0002,
      "step": 5450
    },
    {
      "epoch": 1.8268268268268268,
      "grad_norm": 0.0015207937685772777,
      "learning_rate": 2.1727845754542084e-05,
      "loss": 0.0004,
      "step": 5475
    },
    {
      "epoch": 1.8351685018351684,
      "grad_norm": 0.001575661008246243,
      "learning_rate": 2.1573353108392042e-05,
      "loss": 0.0314,
      "step": 5500
    },
    {
      "epoch": 1.8435101768435103,
      "grad_norm": 0.0009630912682041526,
      "learning_rate": 2.1418860462242e-05,
      "loss": 0.0047,
      "step": 5525
    },
    {
      "epoch": 1.8518518518518519,
      "grad_norm": 0.001678589964285493,
      "learning_rate": 2.1264367816091954e-05,
      "loss": 0.0338,
      "step": 5550
    },
    {
      "epoch": 1.8601935268601935,
      "grad_norm": 0.004351487848907709,
      "learning_rate": 2.1109875169941912e-05,
      "loss": 0.0063,
      "step": 5575
    },
    {
      "epoch": 1.8685352018685353,
      "grad_norm": 0.008160965517163277,
      "learning_rate": 2.095538252379187e-05,
      "loss": 0.0304,
      "step": 5600
    },
    {
      "epoch": 1.8768768768768769,
      "grad_norm": 0.0037402340676635504,
      "learning_rate": 2.0800889877641824e-05,
      "loss": 0.0006,
      "step": 5625
    },
    {
      "epoch": 1.8852185518852185,
      "grad_norm": 0.0016508076805621386,
      "learning_rate": 2.064639723149178e-05,
      "loss": 0.0001,
      "step": 5650
    },
    {
      "epoch": 1.8935602268935603,
      "grad_norm": 0.011413435451686382,
      "learning_rate": 2.049190458534174e-05,
      "loss": 0.0197,
      "step": 5675
    },
    {
      "epoch": 1.901901901901902,
      "grad_norm": 0.0007151237805373967,
      "learning_rate": 2.0337411939191693e-05,
      "loss": 0.0002,
      "step": 5700
    },
    {
      "epoch": 1.9102435769102435,
      "grad_norm": 24.82006072998047,
      "learning_rate": 2.018291929304165e-05,
      "loss": 0.0506,
      "step": 5725
    },
    {
      "epoch": 1.9185852519185853,
      "grad_norm": 0.018485259264707565,
      "learning_rate": 2.002842664689161e-05,
      "loss": 0.0324,
      "step": 5750
    },
    {
      "epoch": 1.926926926926927,
      "grad_norm": 0.008674918673932552,
      "learning_rate": 1.9873934000741566e-05,
      "loss": 0.0006,
      "step": 5775
    },
    {
      "epoch": 1.9352686019352685,
      "grad_norm": 0.0022723872680217028,
      "learning_rate": 1.9719441354591524e-05,
      "loss": 0.0002,
      "step": 5800
    },
    {
      "epoch": 1.9436102769436103,
      "grad_norm": 0.0006509014638140798,
      "learning_rate": 1.9564948708441482e-05,
      "loss": 0.0022,
      "step": 5825
    },
    {
      "epoch": 1.951951951951952,
      "grad_norm": 0.001537534175440669,
      "learning_rate": 1.9410456062291436e-05,
      "loss": 0.0002,
      "step": 5850
    },
    {
      "epoch": 1.9602936269602935,
      "grad_norm": 0.0028848988004028797,
      "learning_rate": 1.9255963416141394e-05,
      "loss": 0.0312,
      "step": 5875
    },
    {
      "epoch": 1.9686353019686353,
      "grad_norm": 0.001275117159821093,
      "learning_rate": 1.910147076999135e-05,
      "loss": 0.0004,
      "step": 5900
    },
    {
      "epoch": 1.976976976976977,
      "grad_norm": 0.003794446587562561,
      "learning_rate": 1.8946978123841306e-05,
      "loss": 0.0303,
      "step": 5925
    },
    {
      "epoch": 1.9853186519853185,
      "grad_norm": 0.0013351526577025652,
      "learning_rate": 1.8792485477691263e-05,
      "loss": 0.0028,
      "step": 5950
    },
    {
      "epoch": 1.9936603269936604,
      "grad_norm": 0.0022188324946910143,
      "learning_rate": 1.863799283154122e-05,
      "loss": 0.0001,
      "step": 5975
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9983316649983317,
      "eval_auc": 0.9996210737764792,
      "eval_f1": 0.9285714285714286,
      "eval_loss": 0.006147409323602915,
      "eval_precision": 0.9848484848484849,
      "eval_recall": 0.8783783783783784,
      "eval_runtime": 118.7551,
      "eval_samples_per_second": 50.474,
      "eval_steps_per_second": 3.158,
      "step": 5994
    },
    {
      "epoch": 2.002002002002002,
      "grad_norm": 0.03426537662744522,
      "learning_rate": 1.8483500185391175e-05,
      "loss": 0.0001,
      "step": 6000
    },
    {
      "epoch": 2.0103436770103436,
      "grad_norm": 0.017995424568653107,
      "learning_rate": 1.8329007539241133e-05,
      "loss": 0.007,
      "step": 6025
    },
    {
      "epoch": 2.0186853520186854,
      "grad_norm": 0.0008380690705962479,
      "learning_rate": 1.817451489309109e-05,
      "loss": 0.0163,
      "step": 6050
    },
    {
      "epoch": 2.027027027027027,
      "grad_norm": 0.0041703651659190655,
      "learning_rate": 1.8020022246941045e-05,
      "loss": 0.0002,
      "step": 6075
    },
    {
      "epoch": 2.0353687020353686,
      "grad_norm": 0.0007581566460430622,
      "learning_rate": 1.7865529600791003e-05,
      "loss": 0.0001,
      "step": 6100
    },
    {
      "epoch": 2.0437103770437104,
      "grad_norm": 0.000699601077940315,
      "learning_rate": 1.771103695464096e-05,
      "loss": 0.0003,
      "step": 6125
    },
    {
      "epoch": 2.052052052052052,
      "grad_norm": 0.021615630015730858,
      "learning_rate": 1.7556544308490915e-05,
      "loss": 0.0195,
      "step": 6150
    },
    {
      "epoch": 2.0603937270603936,
      "grad_norm": 0.0024253269657492638,
      "learning_rate": 1.7402051662340872e-05,
      "loss": 0.0016,
      "step": 6175
    },
    {
      "epoch": 2.0687354020687354,
      "grad_norm": 8.936987433116883e-05,
      "learning_rate": 1.724755901619083e-05,
      "loss": 0.0,
      "step": 6200
    },
    {
      "epoch": 2.0770770770770772,
      "grad_norm": 0.0001456751924706623,
      "learning_rate": 1.7093066370040788e-05,
      "loss": 0.0,
      "step": 6225
    },
    {
      "epoch": 2.0854187520854186,
      "grad_norm": 0.0011276202276349068,
      "learning_rate": 1.6938573723890745e-05,
      "loss": 0.0382,
      "step": 6250
    },
    {
      "epoch": 2.0937604270937604,
      "grad_norm": 0.00011645181803032756,
      "learning_rate": 1.6784081077740703e-05,
      "loss": 0.0001,
      "step": 6275
    },
    {
      "epoch": 2.1021021021021022,
      "grad_norm": 0.00045101341675035655,
      "learning_rate": 1.6629588431590657e-05,
      "loss": 0.0002,
      "step": 6300
    },
    {
      "epoch": 2.1104437771104436,
      "grad_norm": 0.0001981251552933827,
      "learning_rate": 1.6475095785440615e-05,
      "loss": 0.0002,
      "step": 6325
    },
    {
      "epoch": 2.1187854521187854,
      "grad_norm": 0.0004596684011630714,
      "learning_rate": 1.6320603139290573e-05,
      "loss": 0.0001,
      "step": 6350
    },
    {
      "epoch": 2.1271271271271273,
      "grad_norm": 0.0003594656300265342,
      "learning_rate": 1.6166110493140527e-05,
      "loss": 0.0,
      "step": 6375
    },
    {
      "epoch": 2.1354688021354686,
      "grad_norm": 0.0015784379793331027,
      "learning_rate": 1.6011617846990485e-05,
      "loss": 0.0103,
      "step": 6400
    },
    {
      "epoch": 2.1438104771438105,
      "grad_norm": 0.001940034911967814,
      "learning_rate": 1.5857125200840442e-05,
      "loss": 0.0019,
      "step": 6425
    },
    {
      "epoch": 2.1521521521521523,
      "grad_norm": 0.0011929641477763653,
      "learning_rate": 1.5702632554690397e-05,
      "loss": 0.0,
      "step": 6450
    },
    {
      "epoch": 2.1604938271604937,
      "grad_norm": 0.0035334331914782524,
      "learning_rate": 1.5548139908540354e-05,
      "loss": 0.007,
      "step": 6475
    },
    {
      "epoch": 2.1688355021688355,
      "grad_norm": 0.0003299266390968114,
      "learning_rate": 1.5393647262390312e-05,
      "loss": 0.0002,
      "step": 6500
    },
    {
      "epoch": 2.1771771771771773,
      "grad_norm": 0.0005644136690534651,
      "learning_rate": 1.5239154616240266e-05,
      "loss": 0.0,
      "step": 6525
    },
    {
      "epoch": 2.1855188521855187,
      "grad_norm": 0.0003227370616514236,
      "learning_rate": 1.5084661970090224e-05,
      "loss": 0.0,
      "step": 6550
    },
    {
      "epoch": 2.1938605271938605,
      "grad_norm": 0.0011611609952524304,
      "learning_rate": 1.4930169323940182e-05,
      "loss": 0.0,
      "step": 6575
    },
    {
      "epoch": 2.2022022022022023,
      "grad_norm": 0.003065961878746748,
      "learning_rate": 1.4775676677790138e-05,
      "loss": 0.0291,
      "step": 6600
    },
    {
      "epoch": 2.2105438772105437,
      "grad_norm": 0.0016901785274967551,
      "learning_rate": 1.4621184031640095e-05,
      "loss": 0.0,
      "step": 6625
    },
    {
      "epoch": 2.2188855522188855,
      "grad_norm": 0.003457328537479043,
      "learning_rate": 1.4466691385490053e-05,
      "loss": 0.0578,
      "step": 6650
    },
    {
      "epoch": 2.2272272272272273,
      "grad_norm": 0.0015183900250121951,
      "learning_rate": 1.4312198739340007e-05,
      "loss": 0.0004,
      "step": 6675
    },
    {
      "epoch": 2.2355689022355687,
      "grad_norm": 0.01456731278449297,
      "learning_rate": 1.4157706093189965e-05,
      "loss": 0.0001,
      "step": 6700
    },
    {
      "epoch": 2.2439105772439105,
      "grad_norm": 0.006311188917607069,
      "learning_rate": 1.4003213447039923e-05,
      "loss": 0.0001,
      "step": 6725
    },
    {
      "epoch": 2.2522522522522523,
      "grad_norm": 0.00036790382000617683,
      "learning_rate": 1.3848720800889879e-05,
      "loss": 0.0333,
      "step": 6750
    },
    {
      "epoch": 2.2605939272605937,
      "grad_norm": 0.002643864369019866,
      "learning_rate": 1.3694228154739836e-05,
      "loss": 0.0001,
      "step": 6775
    },
    {
      "epoch": 2.2689356022689355,
      "grad_norm": 0.0004877290630247444,
      "learning_rate": 1.3539735508589794e-05,
      "loss": 0.0001,
      "step": 6800
    },
    {
      "epoch": 2.2772772772772774,
      "grad_norm": 0.0006125778309069574,
      "learning_rate": 1.3385242862439748e-05,
      "loss": 0.0001,
      "step": 6825
    },
    {
      "epoch": 2.2856189522856187,
      "grad_norm": 0.006749690044671297,
      "learning_rate": 1.3230750216289706e-05,
      "loss": 0.0295,
      "step": 6850
    },
    {
      "epoch": 2.2939606272939606,
      "grad_norm": 0.00032511603785678744,
      "learning_rate": 1.3076257570139664e-05,
      "loss": 0.0003,
      "step": 6875
    },
    {
      "epoch": 2.3023023023023024,
      "grad_norm": 0.00033115336555056274,
      "learning_rate": 1.2921764923989618e-05,
      "loss": 0.0001,
      "step": 6900
    },
    {
      "epoch": 2.3106439773106437,
      "grad_norm": 0.0015968529041856527,
      "learning_rate": 1.2767272277839576e-05,
      "loss": 0.0001,
      "step": 6925
    },
    {
      "epoch": 2.3189856523189856,
      "grad_norm": 0.0019475755980238318,
      "learning_rate": 1.2612779631689531e-05,
      "loss": 0.0001,
      "step": 6950
    },
    {
      "epoch": 2.3273273273273274,
      "grad_norm": 0.004731306340545416,
      "learning_rate": 1.245828698553949e-05,
      "loss": 0.0001,
      "step": 6975
    },
    {
      "epoch": 2.3356690023356688,
      "grad_norm": 0.0013892946299165487,
      "learning_rate": 1.2303794339389447e-05,
      "loss": 0.0002,
      "step": 7000
    },
    {
      "epoch": 2.3440106773440106,
      "grad_norm": 0.00041954405605793,
      "learning_rate": 1.2149301693239403e-05,
      "loss": 0.0215,
      "step": 7025
    },
    {
      "epoch": 2.3523523523523524,
      "grad_norm": 0.0006229266291484237,
      "learning_rate": 1.1994809047089359e-05,
      "loss": 0.0001,
      "step": 7050
    },
    {
      "epoch": 2.360694027360694,
      "grad_norm": 0.0022454725112766027,
      "learning_rate": 1.1840316400939316e-05,
      "loss": 0.0003,
      "step": 7075
    },
    {
      "epoch": 2.3690357023690356,
      "grad_norm": 0.0003879007126670331,
      "learning_rate": 1.1685823754789272e-05,
      "loss": 0.0002,
      "step": 7100
    },
    {
      "epoch": 2.3773773773773774,
      "grad_norm": 0.0023330808617174625,
      "learning_rate": 1.1531331108639228e-05,
      "loss": 0.0001,
      "step": 7125
    },
    {
      "epoch": 2.3857190523857192,
      "grad_norm": 0.009034736081957817,
      "learning_rate": 1.1376838462489186e-05,
      "loss": 0.0409,
      "step": 7150
    },
    {
      "epoch": 2.3940607273940606,
      "grad_norm": 0.012751072645187378,
      "learning_rate": 1.1222345816339142e-05,
      "loss": 0.0001,
      "step": 7175
    },
    {
      "epoch": 2.4024024024024024,
      "grad_norm": 0.0005218230071477592,
      "learning_rate": 1.10678531701891e-05,
      "loss": 0.0003,
      "step": 7200
    },
    {
      "epoch": 2.4107440774107443,
      "grad_norm": 0.005197747144848108,
      "learning_rate": 1.0913360524039057e-05,
      "loss": 0.0002,
      "step": 7225
    },
    {
      "epoch": 2.4190857524190856,
      "grad_norm": 0.0004819717141799629,
      "learning_rate": 1.0758867877889013e-05,
      "loss": 0.0052,
      "step": 7250
    },
    {
      "epoch": 2.4274274274274275,
      "grad_norm": 0.0005321923526935279,
      "learning_rate": 1.060437523173897e-05,
      "loss": 0.0,
      "step": 7275
    },
    {
      "epoch": 2.4357691024357693,
      "grad_norm": 0.0027098068967461586,
      "learning_rate": 1.0449882585588927e-05,
      "loss": 0.0048,
      "step": 7300
    },
    {
      "epoch": 2.4441107774441106,
      "grad_norm": 15.976853370666504,
      "learning_rate": 1.0295389939438883e-05,
      "loss": 0.0131,
      "step": 7325
    },
    {
      "epoch": 2.4524524524524525,
      "grad_norm": 0.06583759933710098,
      "learning_rate": 1.0140897293288839e-05,
      "loss": 0.0002,
      "step": 7350
    },
    {
      "epoch": 2.4607941274607943,
      "grad_norm": 0.0012560704490169883,
      "learning_rate": 9.986404647138797e-06,
      "loss": 0.0002,
      "step": 7375
    },
    {
      "epoch": 2.4691358024691357,
      "grad_norm": 0.0004110535664949566,
      "learning_rate": 9.831912000988754e-06,
      "loss": 0.0001,
      "step": 7400
    },
    {
      "epoch": 2.4774774774774775,
      "grad_norm": 0.0021631394047290087,
      "learning_rate": 9.67741935483871e-06,
      "loss": 0.0001,
      "step": 7425
    },
    {
      "epoch": 2.4858191524858193,
      "grad_norm": 0.002222039271146059,
      "learning_rate": 9.522926708688668e-06,
      "loss": 0.0169,
      "step": 7450
    },
    {
      "epoch": 2.4941608274941607,
      "grad_norm": 0.0026871326845139265,
      "learning_rate": 9.368434062538624e-06,
      "loss": 0.0243,
      "step": 7475
    },
    {
      "epoch": 2.5025025025025025,
      "grad_norm": 0.00477555999532342,
      "learning_rate": 9.21394141638858e-06,
      "loss": 0.0005,
      "step": 7500
    },
    {
      "epoch": 2.5108441775108443,
      "grad_norm": 0.012109145522117615,
      "learning_rate": 9.059448770238536e-06,
      "loss": 0.0282,
      "step": 7525
    },
    {
      "epoch": 2.5191858525191857,
      "grad_norm": 0.0005930490442551672,
      "learning_rate": 8.904956124088494e-06,
      "loss": 0.0002,
      "step": 7550
    },
    {
      "epoch": 2.5275275275275275,
      "grad_norm": 0.003786754095926881,
      "learning_rate": 8.75046347793845e-06,
      "loss": 0.0023,
      "step": 7575
    },
    {
      "epoch": 2.5358692025358693,
      "grad_norm": 0.0012933415127918124,
      "learning_rate": 8.595970831788407e-06,
      "loss": 0.0119,
      "step": 7600
    },
    {
      "epoch": 2.5442108775442107,
      "grad_norm": 0.0005488534225150943,
      "learning_rate": 8.441478185638365e-06,
      "loss": 0.0135,
      "step": 7625
    },
    {
      "epoch": 2.5525525525525525,
      "grad_norm": 0.0013369097141548991,
      "learning_rate": 8.286985539488321e-06,
      "loss": 0.0172,
      "step": 7650
    },
    {
      "epoch": 2.5608942275608944,
      "grad_norm": 0.0037656030617654324,
      "learning_rate": 8.132492893338277e-06,
      "loss": 0.0163,
      "step": 7675
    },
    {
      "epoch": 2.569235902569236,
      "grad_norm": 0.0011651143431663513,
      "learning_rate": 7.978000247188235e-06,
      "loss": 0.0001,
      "step": 7700
    },
    {
      "epoch": 2.5775775775775776,
      "grad_norm": 0.0026386177632957697,
      "learning_rate": 7.82350760103819e-06,
      "loss": 0.0002,
      "step": 7725
    },
    {
      "epoch": 2.5859192525859194,
      "grad_norm": 0.004785032011568546,
      "learning_rate": 7.669014954888147e-06,
      "loss": 0.0,
      "step": 7750
    },
    {
      "epoch": 2.594260927594261,
      "grad_norm": 0.00011592906957957894,
      "learning_rate": 7.514522308738104e-06,
      "loss": 0.0,
      "step": 7775
    },
    {
      "epoch": 2.6026026026026026,
      "grad_norm": 0.001778353936970234,
      "learning_rate": 7.360029662588061e-06,
      "loss": 0.0001,
      "step": 7800
    },
    {
      "epoch": 2.6109442776109444,
      "grad_norm": 0.001059061149135232,
      "learning_rate": 7.205537016438017e-06,
      "loss": 0.0095,
      "step": 7825
    },
    {
      "epoch": 2.619285952619286,
      "grad_norm": 0.00041067678830586374,
      "learning_rate": 7.051044370287975e-06,
      "loss": 0.0001,
      "step": 7850
    },
    {
      "epoch": 2.6276276276276276,
      "grad_norm": 0.0005333448643796146,
      "learning_rate": 6.896551724137932e-06,
      "loss": 0.0003,
      "step": 7875
    },
    {
      "epoch": 2.6359693026359694,
      "grad_norm": 0.03314538672566414,
      "learning_rate": 6.742059077987888e-06,
      "loss": 0.0019,
      "step": 7900
    },
    {
      "epoch": 2.6443109776443112,
      "grad_norm": 0.0016822376055642962,
      "learning_rate": 6.587566431837845e-06,
      "loss": 0.0063,
      "step": 7925
    },
    {
      "epoch": 2.6526526526526526,
      "grad_norm": 0.0006041646120138466,
      "learning_rate": 6.433073785687801e-06,
      "loss": 0.0001,
      "step": 7950
    },
    {
      "epoch": 2.6609943276609944,
      "grad_norm": 0.00021345775166992098,
      "learning_rate": 6.278581139537758e-06,
      "loss": 0.0001,
      "step": 7975
    },
    {
      "epoch": 2.6693360026693362,
      "grad_norm": 0.00019755800894927233,
      "learning_rate": 6.124088493387715e-06,
      "loss": 0.0,
      "step": 8000
    },
    {
      "epoch": 2.6776776776776776,
      "grad_norm": 0.0004531990271061659,
      "learning_rate": 5.969595847237672e-06,
      "loss": 0.0,
      "step": 8025
    },
    {
      "epoch": 2.6860193526860194,
      "grad_norm": 0.002286186208948493,
      "learning_rate": 5.815103201087629e-06,
      "loss": 0.0,
      "step": 8050
    },
    {
      "epoch": 2.6943610276943613,
      "grad_norm": 9.491475066170096e-05,
      "learning_rate": 5.660610554937585e-06,
      "loss": 0.0002,
      "step": 8075
    },
    {
      "epoch": 2.7027027027027026,
      "grad_norm": 0.00024974957341328263,
      "learning_rate": 5.506117908787542e-06,
      "loss": 0.0044,
      "step": 8100
    },
    {
      "epoch": 2.7110443777110445,
      "grad_norm": 0.0010395270073786378,
      "learning_rate": 5.351625262637499e-06,
      "loss": 0.0001,
      "step": 8125
    },
    {
      "epoch": 2.7193860527193863,
      "grad_norm": 0.0001526384148746729,
      "learning_rate": 5.197132616487455e-06,
      "loss": 0.0,
      "step": 8150
    },
    {
      "epoch": 2.7277277277277276,
      "grad_norm": 0.0006884171743877232,
      "learning_rate": 5.042639970337412e-06,
      "loss": 0.0001,
      "step": 8175
    },
    {
      "epoch": 2.7360694027360695,
      "grad_norm": 9.014210809255019e-05,
      "learning_rate": 4.88814732418737e-06,
      "loss": 0.0001,
      "step": 8200
    },
    {
      "epoch": 2.7444110777444113,
      "grad_norm": 0.0002122920996043831,
      "learning_rate": 4.733654678037326e-06,
      "loss": 0.0001,
      "step": 8225
    },
    {
      "epoch": 2.7527527527527527,
      "grad_norm": 0.0002799911017064005,
      "learning_rate": 4.579162031887282e-06,
      "loss": 0.0001,
      "step": 8250
    },
    {
      "epoch": 2.7610944277610945,
      "grad_norm": 0.00325685553252697,
      "learning_rate": 4.424669385737239e-06,
      "loss": 0.0002,
      "step": 8275
    },
    {
      "epoch": 2.7694361027694363,
      "grad_norm": 0.00023966682783793658,
      "learning_rate": 4.270176739587196e-06,
      "loss": 0.0001,
      "step": 8300
    },
    {
      "epoch": 2.7777777777777777,
      "grad_norm": 0.00024113825929816812,
      "learning_rate": 4.115684093437153e-06,
      "loss": 0.0,
      "step": 8325
    },
    {
      "epoch": 2.7861194527861195,
      "grad_norm": 0.00069383904337883,
      "learning_rate": 3.96119144728711e-06,
      "loss": 0.0282,
      "step": 8350
    },
    {
      "epoch": 2.7944611277944613,
      "grad_norm": 0.00674516661092639,
      "learning_rate": 3.8066988011370657e-06,
      "loss": 0.0,
      "step": 8375
    },
    {
      "epoch": 2.8028028028028027,
      "grad_norm": 0.0006425007013604045,
      "learning_rate": 3.652206154987023e-06,
      "loss": 0.0001,
      "step": 8400
    },
    {
      "epoch": 2.8111444778111445,
      "grad_norm": 0.001739656669087708,
      "learning_rate": 3.49771350883698e-06,
      "loss": 0.0003,
      "step": 8425
    },
    {
      "epoch": 2.8194861528194863,
      "grad_norm": 0.00028822454623878,
      "learning_rate": 3.343220862686936e-06,
      "loss": 0.0022,
      "step": 8450
    },
    {
      "epoch": 2.8278278278278277,
      "grad_norm": 0.0007758860592730343,
      "learning_rate": 3.188728216536893e-06,
      "loss": 0.0002,
      "step": 8475
    },
    {
      "epoch": 2.8361695028361695,
      "grad_norm": 6.443789607146755e-05,
      "learning_rate": 3.03423557038685e-06,
      "loss": 0.0003,
      "step": 8500
    },
    {
      "epoch": 2.8445111778445114,
      "grad_norm": 0.008539211004972458,
      "learning_rate": 2.8797429242368063e-06,
      "loss": 0.0009,
      "step": 8525
    },
    {
      "epoch": 2.8528528528528527,
      "grad_norm": 0.0164475180208683,
      "learning_rate": 2.7252502780867635e-06,
      "loss": 0.0098,
      "step": 8550
    },
    {
      "epoch": 2.8611945278611945,
      "grad_norm": 0.0004140229430049658,
      "learning_rate": 2.57075763193672e-06,
      "loss": 0.0,
      "step": 8575
    },
    {
      "epoch": 2.8695362028695364,
      "grad_norm": 0.0007194245117716491,
      "learning_rate": 2.4162649857866768e-06,
      "loss": 0.0002,
      "step": 8600
    },
    {
      "epoch": 2.8778778778778777,
      "grad_norm": 8.639365114504471e-05,
      "learning_rate": 2.261772339636633e-06,
      "loss": 0.0,
      "step": 8625
    },
    {
      "epoch": 2.8862195528862196,
      "grad_norm": 0.00026035771588794887,
      "learning_rate": 2.1072796934865904e-06,
      "loss": 0.0,
      "step": 8650
    },
    {
      "epoch": 2.8945612278945614,
      "grad_norm": 0.0003119956236332655,
      "learning_rate": 1.952787047336547e-06,
      "loss": 0.0164,
      "step": 8675
    },
    {
      "epoch": 2.9029029029029028,
      "grad_norm": 0.0001721863227430731,
      "learning_rate": 1.7982944011865034e-06,
      "loss": 0.0,
      "step": 8700
    },
    {
      "epoch": 2.9112445779112446,
      "grad_norm": 0.0003014954272657633,
      "learning_rate": 1.6438017550364605e-06,
      "loss": 0.0002,
      "step": 8725
    },
    {
      "epoch": 2.9195862529195864,
      "grad_norm": 0.002242905320599675,
      "learning_rate": 1.489309108886417e-06,
      "loss": 0.0001,
      "step": 8750
    },
    {
      "epoch": 2.9279279279279278,
      "grad_norm": 0.00019342848099768162,
      "learning_rate": 1.334816462736374e-06,
      "loss": 0.0,
      "step": 8775
    },
    {
      "epoch": 2.9362696029362696,
      "grad_norm": 0.00041662275907583535,
      "learning_rate": 1.1803238165863305e-06,
      "loss": 0.0023,
      "step": 8800
    },
    {
      "epoch": 2.9446112779446114,
      "grad_norm": 8.121671999106184e-05,
      "learning_rate": 1.0258311704362874e-06,
      "loss": 0.0,
      "step": 8825
    },
    {
      "epoch": 2.952952952952953,
      "grad_norm": 0.00027466562460176647,
      "learning_rate": 8.71338524286244e-07,
      "loss": 0.0,
      "step": 8850
    },
    {
      "epoch": 2.9612946279612946,
      "grad_norm": 0.00015804750728420913,
      "learning_rate": 7.168458781362007e-07,
      "loss": 0.0,
      "step": 8875
    },
    {
      "epoch": 2.9696363029696364,
      "grad_norm": 5.4930391343077645e-05,
      "learning_rate": 5.623532319861574e-07,
      "loss": 0.0011,
      "step": 8900
    },
    {
      "epoch": 2.977977977977978,
      "grad_norm": 0.0005976063548587263,
      "learning_rate": 4.0786058583611417e-07,
      "loss": 0.0015,
      "step": 8925
    },
    {
      "epoch": 2.9863196529863196,
      "grad_norm": 0.0008383295498788357,
      "learning_rate": 2.5336793968607095e-07,
      "loss": 0.0,
      "step": 8950
    },
    {
      "epoch": 2.9946613279946614,
      "grad_norm": 0.001575443078763783,
      "learning_rate": 9.88752935360277e-08,
      "loss": 0.0102,
      "step": 8975
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9988321654988321,
      "eval_auc": 0.9999543462381301,
      "eval_f1": 0.9523809523809523,
      "eval_loss": 0.003335413057357073,
      "eval_precision": 0.958904109589041,
      "eval_recall": 0.9459459459459459,
      "eval_runtime": 118.1878,
      "eval_samples_per_second": 50.716,
      "eval_steps_per_second": 3.173,
      "step": 8991
    }
  ],
  "logging_steps": 25,
  "max_steps": 8991,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.01
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.6433249389401342e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}