{
  "best_metric": 0.0027391575276851654,
  "best_model_checkpoint": "sj-det-6/checkpoint-9162",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 9162,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.008185985592665358,
      "grad_norm": 14.986274719238281,
      "learning_rate": 1.2540894220283533e-06,
      "loss": 1.1682,
      "step": 25
    },
    {
      "epoch": 0.016371971185330715,
      "grad_norm": 8.189007759094238,
      "learning_rate": 2.6172300981461287e-06,
      "loss": 0.7882,
      "step": 50
    },
    {
      "epoch": 0.02455795677799607,
      "grad_norm": 4.545085906982422,
      "learning_rate": 3.980370774263904e-06,
      "loss": 0.3987,
      "step": 75
    },
    {
      "epoch": 0.03274394237066143,
      "grad_norm": 0.8907670974731445,
      "learning_rate": 5.343511450381679e-06,
      "loss": 0.1198,
      "step": 100
    },
    {
      "epoch": 0.040929927963326784,
      "grad_norm": 0.8682488203048706,
      "learning_rate": 6.7066521264994545e-06,
      "loss": 0.1166,
      "step": 125
    },
    {
      "epoch": 0.04911591355599214,
      "grad_norm": 0.1406942456960678,
      "learning_rate": 8.069792802617231e-06,
      "loss": 0.0797,
      "step": 150
    },
    {
      "epoch": 0.0573018991486575,
      "grad_norm": 0.030166020616889,
      "learning_rate": 9.432933478735007e-06,
      "loss": 0.0518,
      "step": 175
    },
    {
      "epoch": 0.06548788474132286,
      "grad_norm": 0.0494314506649971,
      "learning_rate": 1.0796074154852782e-05,
      "loss": 0.0505,
      "step": 200
    },
    {
      "epoch": 0.07367387033398821,
      "grad_norm": 0.01392674632370472,
      "learning_rate": 1.2159214830970557e-05,
      "loss": 0.0272,
      "step": 225
    },
    {
      "epoch": 0.08185985592665357,
      "grad_norm": 0.0814874917268753,
      "learning_rate": 1.3522355507088334e-05,
      "loss": 0.0365,
      "step": 250
    },
    {
      "epoch": 0.09004584151931892,
      "grad_norm": 0.04325518757104874,
      "learning_rate": 1.4885496183206107e-05,
      "loss": 0.0576,
      "step": 275
    },
    {
      "epoch": 0.09823182711198428,
      "grad_norm": 0.09735051542520523,
      "learning_rate": 1.624863685932388e-05,
      "loss": 0.0329,
      "step": 300
    },
    {
      "epoch": 0.10641781270464964,
      "grad_norm": 0.64588862657547,
      "learning_rate": 1.761177753544166e-05,
      "loss": 0.0802,
      "step": 325
    },
    {
      "epoch": 0.114603798297315,
      "grad_norm": 0.011760449036955833,
      "learning_rate": 1.8974918211559434e-05,
      "loss": 0.0524,
      "step": 350
    },
    {
      "epoch": 0.12278978388998035,
      "grad_norm": 0.1449550837278366,
      "learning_rate": 2.033805888767721e-05,
      "loss": 0.0347,
      "step": 375
    },
    {
      "epoch": 0.13097576948264572,
      "grad_norm": 0.018402298912405968,
      "learning_rate": 2.1701199563794985e-05,
      "loss": 0.0155,
      "step": 400
    },
    {
      "epoch": 0.13916175507531106,
      "grad_norm": 0.016214437782764435,
      "learning_rate": 2.306434023991276e-05,
      "loss": 0.0988,
      "step": 425
    },
    {
      "epoch": 0.14734774066797643,
      "grad_norm": 0.08800540119409561,
      "learning_rate": 2.4427480916030535e-05,
      "loss": 0.0428,
      "step": 450
    },
    {
      "epoch": 0.15553372626064177,
      "grad_norm": 0.005783313885331154,
      "learning_rate": 2.5790621592148313e-05,
      "loss": 0.1005,
      "step": 475
    },
    {
      "epoch": 0.16371971185330714,
      "grad_norm": 0.028863191604614258,
      "learning_rate": 2.7153762268266085e-05,
      "loss": 0.0427,
      "step": 500
    },
    {
      "epoch": 0.1719056974459725,
      "grad_norm": 6.242043495178223,
      "learning_rate": 2.851690294438386e-05,
      "loss": 0.0868,
      "step": 525
    },
    {
      "epoch": 0.18009168303863785,
      "grad_norm": 0.013895904645323753,
      "learning_rate": 2.9880043620501635e-05,
      "loss": 0.0898,
      "step": 550
    },
    {
      "epoch": 0.1882776686313032,
      "grad_norm": 0.019419686868786812,
      "learning_rate": 3.124318429661941e-05,
      "loss": 0.0351,
      "step": 575
    },
    {
      "epoch": 0.19646365422396855,
      "grad_norm": 0.008130255155265331,
      "learning_rate": 3.260632497273719e-05,
      "loss": 0.0164,
      "step": 600
    },
    {
      "epoch": 0.20464963981663392,
      "grad_norm": 6.5059814453125,
      "learning_rate": 3.3969465648854964e-05,
      "loss": 0.0359,
      "step": 625
    },
    {
      "epoch": 0.2128356254092993,
      "grad_norm": 0.010089697316288948,
      "learning_rate": 3.533260632497274e-05,
      "loss": 0.0305,
      "step": 650
    },
    {
      "epoch": 0.22102161100196463,
      "grad_norm": 0.014347407966852188,
      "learning_rate": 3.6695747001090515e-05,
      "loss": 0.0435,
      "step": 675
    },
    {
      "epoch": 0.22920759659463,
      "grad_norm": 0.017506686970591545,
      "learning_rate": 3.805888767720829e-05,
      "loss": 0.0486,
      "step": 700
    },
    {
      "epoch": 0.23739358218729534,
      "grad_norm": 0.004301242996007204,
      "learning_rate": 3.9422028353326065e-05,
      "loss": 0.0002,
      "step": 725
    },
    {
      "epoch": 0.2455795677799607,
      "grad_norm": 0.00440743425861001,
      "learning_rate": 4.078516902944384e-05,
      "loss": 0.0134,
      "step": 750
    },
    {
      "epoch": 0.2537655533726261,
      "grad_norm": 0.037835538387298584,
      "learning_rate": 4.2148309705561615e-05,
      "loss": 0.0334,
      "step": 775
    },
    {
      "epoch": 0.26195153896529144,
      "grad_norm": 0.11478447914123535,
      "learning_rate": 4.351145038167939e-05,
      "loss": 0.0665,
      "step": 800
    },
    {
      "epoch": 0.27013752455795675,
      "grad_norm": 0.006648090668022633,
      "learning_rate": 4.4874591057797165e-05,
      "loss": 0.0033,
      "step": 825
    },
    {
      "epoch": 0.2783235101506221,
      "grad_norm": 8.948714256286621,
      "learning_rate": 4.623773173391494e-05,
      "loss": 0.0996,
      "step": 850
    },
    {
      "epoch": 0.2865094957432875,
      "grad_norm": 0.012912426143884659,
      "learning_rate": 4.7600872410032716e-05,
      "loss": 0.0102,
      "step": 875
    },
    {
      "epoch": 0.29469548133595286,
      "grad_norm": 0.0026352498680353165,
      "learning_rate": 4.89640130861505e-05,
      "loss": 0.0226,
      "step": 900
    },
    {
      "epoch": 0.3028814669286182,
      "grad_norm": 0.15103408694267273,
      "learning_rate": 4.996361431170407e-05,
      "loss": 0.0225,
      "step": 925
    },
    {
      "epoch": 0.31106745252128354,
      "grad_norm": 0.04878251254558563,
      "learning_rate": 4.981200727713766e-05,
      "loss": 0.0005,
      "step": 950
    },
    {
      "epoch": 0.3192534381139489,
      "grad_norm": 0.00841938890516758,
      "learning_rate": 4.966040024257126e-05,
      "loss": 0.0513,
      "step": 975
    },
    {
      "epoch": 0.3274394237066143,
      "grad_norm": 0.04339870065450668,
      "learning_rate": 4.9508793208004854e-05,
      "loss": 0.0328,
      "step": 1000
    },
    {
      "epoch": 0.33562540929927964,
      "grad_norm": 0.00045626997598446906,
      "learning_rate": 4.935718617343845e-05,
      "loss": 0.0124,
      "step": 1025
    },
    {
      "epoch": 0.343811394891945,
      "grad_norm": 0.00013843162741977721,
      "learning_rate": 4.9205579138872046e-05,
      "loss": 0.0,
      "step": 1050
    },
    {
      "epoch": 0.3519973804846103,
      "grad_norm": 1.9331814050674438,
      "learning_rate": 4.905397210430564e-05,
      "loss": 0.0504,
      "step": 1075
    },
    {
      "epoch": 0.3601833660772757,
      "grad_norm": 0.002882858505472541,
      "learning_rate": 4.890236506973924e-05,
      "loss": 0.1157,
      "step": 1100
    },
    {
      "epoch": 0.36836935166994106,
      "grad_norm": 0.022938862442970276,
      "learning_rate": 4.875075803517283e-05,
      "loss": 0.0716,
      "step": 1125
    },
    {
      "epoch": 0.3765553372626064,
      "grad_norm": 0.0010779292788356543,
      "learning_rate": 4.859915100060643e-05,
      "loss": 0.0419,
      "step": 1150
    },
    {
      "epoch": 0.3847413228552718,
      "grad_norm": 0.03177574649453163,
      "learning_rate": 4.8447543966040024e-05,
      "loss": 0.0249,
      "step": 1175
    },
    {
      "epoch": 0.3929273084479371,
      "grad_norm": 0.0006776170921511948,
      "learning_rate": 4.8295936931473626e-05,
      "loss": 0.0415,
      "step": 1200
    },
    {
      "epoch": 0.4011132940406025,
      "grad_norm": 0.00865075271576643,
      "learning_rate": 4.814432989690722e-05,
      "loss": 0.0304,
      "step": 1225
    },
    {
      "epoch": 0.40929927963326784,
      "grad_norm": 0.004142033401876688,
      "learning_rate": 4.799272286234082e-05,
      "loss": 0.066,
      "step": 1250
    },
    {
      "epoch": 0.4174852652259332,
      "grad_norm": 0.006911768112331629,
      "learning_rate": 4.784111582777441e-05,
      "loss": 0.0345,
      "step": 1275
    },
    {
      "epoch": 0.4256712508185986,
      "grad_norm": 0.011829985305666924,
      "learning_rate": 4.768950879320801e-05,
      "loss": 0.0324,
      "step": 1300
    },
    {
      "epoch": 0.4338572364112639,
      "grad_norm": 0.024410191923379898,
      "learning_rate": 4.7537901758641604e-05,
      "loss": 0.0412,
      "step": 1325
    },
    {
      "epoch": 0.44204322200392926,
      "grad_norm": 0.0020343943033367395,
      "learning_rate": 4.73862947240752e-05,
      "loss": 0.0133,
      "step": 1350
    },
    {
      "epoch": 0.4502292075965946,
      "grad_norm": 0.0009943109471350908,
      "learning_rate": 4.7234687689508795e-05,
      "loss": 0.0004,
      "step": 1375
    },
    {
      "epoch": 0.45841519318926,
      "grad_norm": 0.00024950483930297196,
      "learning_rate": 4.708308065494239e-05,
      "loss": 0.0,
      "step": 1400
    },
    {
      "epoch": 0.46660117878192536,
      "grad_norm": 0.0015270832227542996,
      "learning_rate": 4.6931473620375986e-05,
      "loss": 0.0314,
      "step": 1425
    },
    {
      "epoch": 0.4747871643745907,
      "grad_norm": 9.285660743713379,
      "learning_rate": 4.677986658580958e-05,
      "loss": 0.0955,
      "step": 1450
    },
    {
      "epoch": 0.48297314996725604,
      "grad_norm": 0.004155229777097702,
      "learning_rate": 4.6628259551243184e-05,
      "loss": 0.003,
      "step": 1475
    },
    {
      "epoch": 0.4911591355599214,
      "grad_norm": 0.013534590601921082,
      "learning_rate": 4.647665251667678e-05,
      "loss": 0.0269,
      "step": 1500
    },
    {
      "epoch": 0.4993451211525868,
      "grad_norm": 0.004476598929613829,
      "learning_rate": 4.6325045482110375e-05,
      "loss": 0.0011,
      "step": 1525
    },
    {
      "epoch": 0.5075311067452521,
      "grad_norm": 0.0038523597177118063,
      "learning_rate": 4.617343844754397e-05,
      "loss": 0.0015,
      "step": 1550
    },
    {
      "epoch": 0.5157170923379175,
      "grad_norm": 0.03839994966983795,
      "learning_rate": 4.602183141297756e-05,
      "loss": 0.1135,
      "step": 1575
    },
    {
      "epoch": 0.5239030779305829,
      "grad_norm": 0.000390953995520249,
      "learning_rate": 4.587022437841116e-05,
      "loss": 0.0015,
      "step": 1600
    },
    {
      "epoch": 0.5320890635232483,
      "grad_norm": 0.00021860921697225422,
      "learning_rate": 4.571861734384476e-05,
      "loss": 0.0015,
      "step": 1625
    },
    {
      "epoch": 0.5402750491159135,
      "grad_norm": 0.005596529692411423,
      "learning_rate": 4.556701030927835e-05,
      "loss": 0.1111,
      "step": 1650
    },
    {
      "epoch": 0.5484610347085789,
      "grad_norm": 5.192307949066162,
      "learning_rate": 4.541540327471195e-05,
      "loss": 0.056,
      "step": 1675
    },
    {
      "epoch": 0.5566470203012442,
      "grad_norm": 0.004266222473233938,
      "learning_rate": 4.5263796240145544e-05,
      "loss": 0.0078,
      "step": 1700
    },
    {
      "epoch": 0.5648330058939096,
      "grad_norm": 4.940586090087891,
      "learning_rate": 4.511218920557914e-05,
      "loss": 0.0416,
      "step": 1725
    },
    {
      "epoch": 0.573018991486575,
      "grad_norm": 0.056938640773296356,
      "learning_rate": 4.4960582171012735e-05,
      "loss": 0.0299,
      "step": 1750
    },
    {
      "epoch": 0.5812049770792403,
      "grad_norm": 0.025186289101839066,
      "learning_rate": 4.480897513644634e-05,
      "loss": 0.0046,
      "step": 1775
    },
    {
      "epoch": 0.5893909626719057,
      "grad_norm": 0.0009007578482851386,
      "learning_rate": 4.465736810187993e-05,
      "loss": 0.0402,
      "step": 1800
    },
    {
      "epoch": 0.5975769482645711,
      "grad_norm": 0.006263722199946642,
      "learning_rate": 4.450576106731352e-05,
      "loss": 0.0008,
      "step": 1825
    },
    {
      "epoch": 0.6057629338572365,
      "grad_norm": 0.002118312055245042,
      "learning_rate": 4.435415403274712e-05,
      "loss": 0.0003,
      "step": 1850
    },
    {
      "epoch": 0.6139489194499018,
      "grad_norm": 0.011466097086668015,
      "learning_rate": 4.420254699818071e-05,
      "loss": 0.0025,
      "step": 1875
    },
    {
      "epoch": 0.6221349050425671,
      "grad_norm": 0.27655062079429626,
      "learning_rate": 4.4050939963614315e-05,
      "loss": 0.0694,
      "step": 1900
    },
    {
      "epoch": 0.6303208906352324,
      "grad_norm": 0.011102208867669106,
      "learning_rate": 4.389933292904791e-05,
      "loss": 0.0771,
      "step": 1925
    },
    {
      "epoch": 0.6385068762278978,
      "grad_norm": 0.03391822427511215,
      "learning_rate": 4.3747725894481507e-05,
      "loss": 0.0516,
      "step": 1950
    },
    {
      "epoch": 0.6466928618205632,
      "grad_norm": 1.2409979104995728,
      "learning_rate": 4.35961188599151e-05,
      "loss": 0.0052,
      "step": 1975
    },
    {
      "epoch": 0.6548788474132285,
      "grad_norm": 0.005223038140684366,
      "learning_rate": 4.34445118253487e-05,
      "loss": 0.0265,
      "step": 2000
    },
    {
      "epoch": 0.6630648330058939,
      "grad_norm": 0.015766901895403862,
      "learning_rate": 4.329290479078229e-05,
      "loss": 0.0023,
      "step": 2025
    },
    {
      "epoch": 0.6712508185985593,
      "grad_norm": 0.030491165816783905,
      "learning_rate": 4.3141297756215896e-05,
      "loss": 0.0343,
      "step": 2050
    },
    {
      "epoch": 0.6794368041912247,
      "grad_norm": 0.00046129990369081497,
      "learning_rate": 4.298969072164949e-05,
      "loss": 0.0121,
      "step": 2075
    },
    {
      "epoch": 0.68762278978389,
      "grad_norm": 0.0001839818141888827,
      "learning_rate": 4.283808368708308e-05,
      "loss": 0.0012,
      "step": 2100
    },
    {
      "epoch": 0.6958087753765554,
      "grad_norm": 0.0003591451095417142,
      "learning_rate": 4.2686476652516676e-05,
      "loss": 0.0374,
      "step": 2125
    },
    {
      "epoch": 0.7039947609692206,
      "grad_norm": 0.009124216623604298,
      "learning_rate": 4.253486961795027e-05,
      "loss": 0.0082,
      "step": 2150
    },
    {
      "epoch": 0.712180746561886,
      "grad_norm": 0.0014978774124756455,
      "learning_rate": 4.2383262583383874e-05,
      "loss": 0.0306,
      "step": 2175
    },
    {
      "epoch": 0.7203667321545514,
      "grad_norm": 1.0534560680389404,
      "learning_rate": 4.223165554881747e-05,
      "loss": 0.0008,
      "step": 2200
    },
    {
      "epoch": 0.7285527177472167,
      "grad_norm": 0.005840160418301821,
      "learning_rate": 4.2080048514251065e-05,
      "loss": 0.0562,
      "step": 2225
    },
    {
      "epoch": 0.7367387033398821,
      "grad_norm": 4.12369966506958,
      "learning_rate": 4.192844147968466e-05,
      "loss": 0.01,
      "step": 2250
    },
    {
      "epoch": 0.7449246889325475,
      "grad_norm": 0.0006641410291194916,
      "learning_rate": 4.1776834445118256e-05,
      "loss": 0.0001,
      "step": 2275
    },
    {
      "epoch": 0.7531106745252129,
      "grad_norm": 0.0009534868295304477,
      "learning_rate": 4.162522741055185e-05,
      "loss": 0.0854,
      "step": 2300
    },
    {
      "epoch": 0.7612966601178782,
      "grad_norm": 0.02312465012073517,
      "learning_rate": 4.147362037598545e-05,
      "loss": 0.0523,
      "step": 2325
    },
    {
      "epoch": 0.7694826457105436,
      "grad_norm": 0.009137257933616638,
      "learning_rate": 4.132201334141904e-05,
      "loss": 0.0337,
      "step": 2350
    },
    {
      "epoch": 0.777668631303209,
      "grad_norm": 0.01361013948917389,
      "learning_rate": 4.117040630685264e-05,
      "loss": 0.0228,
      "step": 2375
    },
    {
      "epoch": 0.7858546168958742,
      "grad_norm": 0.0018830469343811274,
      "learning_rate": 4.1018799272286234e-05,
      "loss": 0.0129,
      "step": 2400
    },
    {
      "epoch": 0.7940406024885396,
      "grad_norm": 0.6149205565452576,
      "learning_rate": 4.086719223771983e-05,
      "loss": 0.0404,
      "step": 2425
    },
    {
      "epoch": 0.802226588081205,
      "grad_norm": 0.005154294427484274,
      "learning_rate": 4.0715585203153425e-05,
      "loss": 0.0011,
      "step": 2450
    },
    {
      "epoch": 0.8104125736738703,
      "grad_norm": 0.022492127493023872,
      "learning_rate": 4.056397816858703e-05,
      "loss": 0.042,
      "step": 2475
    },
    {
      "epoch": 0.8185985592665357,
      "grad_norm": 0.020115699619054794,
      "learning_rate": 4.041237113402062e-05,
      "loss": 0.0495,
      "step": 2500
    },
    {
      "epoch": 0.826784544859201,
      "grad_norm": 0.008540102280676365,
      "learning_rate": 4.026076409945422e-05,
      "loss": 0.0074,
      "step": 2525
    },
    {
      "epoch": 0.8349705304518664,
      "grad_norm": 0.00022775791876483709,
      "learning_rate": 4.0109157064887814e-05,
      "loss": 0.0001,
      "step": 2550
    },
    {
      "epoch": 0.8431565160445318,
      "grad_norm": 0.0014885278651490808,
      "learning_rate": 3.995755003032141e-05,
      "loss": 0.0207,
      "step": 2575
    },
    {
      "epoch": 0.8513425016371972,
      "grad_norm": 0.0014814437599852681,
      "learning_rate": 3.9805942995755005e-05,
      "loss": 0.0219,
      "step": 2600
    },
    {
      "epoch": 0.8595284872298625,
      "grad_norm": 0.0025843679904937744,
      "learning_rate": 3.96543359611886e-05,
      "loss": 0.0137,
      "step": 2625
    },
    {
      "epoch": 0.8677144728225278,
      "grad_norm": 0.04107142984867096,
      "learning_rate": 3.9502728926622196e-05,
      "loss": 0.0695,
      "step": 2650
    },
    {
      "epoch": 0.8759004584151932,
      "grad_norm": 0.005535195115953684,
      "learning_rate": 3.935112189205579e-05,
      "loss": 0.0157,
      "step": 2675
    },
    {
      "epoch": 0.8840864440078585,
      "grad_norm": 5.580929756164551,
      "learning_rate": 3.919951485748939e-05,
      "loss": 0.0806,
      "step": 2700
    },
    {
      "epoch": 0.8922724296005239,
      "grad_norm": 0.0086448909714818,
      "learning_rate": 3.904790782292298e-05,
      "loss": 0.046,
      "step": 2725
    },
    {
      "epoch": 0.9004584151931893,
      "grad_norm": 0.0019950985442847013,
      "learning_rate": 3.8896300788356585e-05,
      "loss": 0.0004,
      "step": 2750
    },
    {
      "epoch": 0.9086444007858546,
      "grad_norm": 0.04631026089191437,
      "learning_rate": 3.874469375379018e-05,
      "loss": 0.1026,
      "step": 2775
    },
    {
      "epoch": 0.91683038637852,
      "grad_norm": 0.14316339790821075,
      "learning_rate": 3.8593086719223776e-05,
      "loss": 0.0173,
      "step": 2800
    },
    {
      "epoch": 0.9250163719711854,
      "grad_norm": 0.002735959365963936,
      "learning_rate": 3.844147968465737e-05,
      "loss": 0.0102,
      "step": 2825
    },
    {
      "epoch": 0.9332023575638507,
      "grad_norm": 0.007864673621952534,
      "learning_rate": 3.828987265009097e-05,
      "loss": 0.001,
      "step": 2850
    },
    {
      "epoch": 0.9413883431565161,
      "grad_norm": 0.0006219564820639789,
      "learning_rate": 3.813826561552456e-05,
      "loss": 0.0121,
      "step": 2875
    },
    {
      "epoch": 0.9495743287491814,
      "grad_norm": 0.20409245789051056,
      "learning_rate": 3.798665858095816e-05,
      "loss": 0.0383,
      "step": 2900
    },
    {
      "epoch": 0.9577603143418467,
      "grad_norm": 0.0005641476600430906,
      "learning_rate": 3.7835051546391754e-05,
      "loss": 0.0054,
      "step": 2925
    },
    {
      "epoch": 0.9659462999345121,
      "grad_norm": 0.004506452474743128,
      "learning_rate": 3.768344451182535e-05,
      "loss": 0.0167,
      "step": 2950
    },
    {
      "epoch": 0.9741322855271775,
      "grad_norm": 0.0007494801538996398,
      "learning_rate": 3.7531837477258945e-05,
      "loss": 0.0016,
      "step": 2975
    },
    {
      "epoch": 0.9823182711198428,
      "grad_norm": 0.0016925218515098095,
      "learning_rate": 3.738023044269254e-05,
      "loss": 0.0144,
      "step": 3000
    },
    {
      "epoch": 0.9905042567125082,
      "grad_norm": 0.007691397797316313,
      "learning_rate": 3.7228623408126137e-05,
      "loss": 0.0264,
      "step": 3025
    },
    {
      "epoch": 0.9986902423051736,
      "grad_norm": 0.004836967680603266,
      "learning_rate": 3.707701637355974e-05,
      "loss": 0.0003,
      "step": 3050
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9955795677799607,
      "eval_auc": 0.9977481210079818,
      "eval_f1": 0.8187919463087249,
      "eval_loss": 0.014572993852198124,
      "eval_precision": 0.8133333333333334,
      "eval_recall": 0.8243243243243243,
      "eval_runtime": 74.4335,
      "eval_samples_per_second": 82.06,
      "eval_steps_per_second": 5.132,
      "step": 3054
    },
    {
      "epoch": 1.006876227897839,
      "grad_norm": 0.0026801538188010454,
      "learning_rate": 3.6925409338993335e-05,
      "loss": 0.0003,
      "step": 3075
    },
    {
      "epoch": 1.0150622134905043,
      "grad_norm": 0.0013839630410075188,
      "learning_rate": 3.677380230442693e-05,
      "loss": 0.0415,
      "step": 3100
    },
    {
      "epoch": 1.0232481990831697,
      "grad_norm": 0.0052743577398359776,
      "learning_rate": 3.662219526986052e-05,
      "loss": 0.0551,
      "step": 3125
    },
    {
      "epoch": 1.031434184675835,
      "grad_norm": 0.001981929875910282,
      "learning_rate": 3.6470588235294114e-05,
      "loss": 0.0011,
      "step": 3150
    },
    {
      "epoch": 1.0396201702685004,
      "grad_norm": 0.005122967530041933,
      "learning_rate": 3.631898120072772e-05,
      "loss": 0.0013,
      "step": 3175
    },
    {
      "epoch": 1.0478061558611658,
      "grad_norm": 0.0008022825350053608,
      "learning_rate": 3.616737416616131e-05,
      "loss": 0.0001,
      "step": 3200
    },
    {
      "epoch": 1.0559921414538311,
      "grad_norm": 0.002690242137759924,
      "learning_rate": 3.601576713159491e-05,
      "loss": 0.0,
      "step": 3225
    },
    {
      "epoch": 1.0641781270464965,
      "grad_norm": 0.0003424021415412426,
      "learning_rate": 3.5864160097028504e-05,
      "loss": 0.0066,
      "step": 3250
    },
    {
      "epoch": 1.0723641126391619,
      "grad_norm": 0.0005132851074449718,
      "learning_rate": 3.57125530624621e-05,
      "loss": 0.0145,
      "step": 3275
    },
    {
      "epoch": 1.080550098231827,
      "grad_norm": 0.0189173873513937,
      "learning_rate": 3.5560946027895695e-05,
      "loss": 0.0465,
      "step": 3300
    },
    {
      "epoch": 1.0887360838244924,
      "grad_norm": 0.0036357955541461706,
      "learning_rate": 3.54093389933293e-05,
      "loss": 0.0492,
      "step": 3325
    },
    {
      "epoch": 1.0969220694171578,
      "grad_norm": 0.003961147274821997,
      "learning_rate": 3.525773195876289e-05,
      "loss": 0.039,
      "step": 3350
    },
    {
      "epoch": 1.1051080550098231,
      "grad_norm": 0.003041486954316497,
      "learning_rate": 3.510612492419648e-05,
      "loss": 0.0123,
      "step": 3375
    },
    {
      "epoch": 1.1132940406024885,
      "grad_norm": 0.001711357617750764,
      "learning_rate": 3.495451788963008e-05,
      "loss": 0.0003,
      "step": 3400
    },
    {
      "epoch": 1.1214800261951539,
      "grad_norm": 0.0013232220662757754,
      "learning_rate": 3.480291085506367e-05,
      "loss": 0.0096,
      "step": 3425
    },
    {
      "epoch": 1.1296660117878192,
      "grad_norm": 0.000497565430123359,
      "learning_rate": 3.4651303820497275e-05,
      "loss": 0.0233,
      "step": 3450
    },
    {
      "epoch": 1.1378519973804846,
      "grad_norm": 0.005278216674923897,
      "learning_rate": 3.449969678593087e-05,
      "loss": 0.0399,
      "step": 3475
    },
    {
      "epoch": 1.14603798297315,
      "grad_norm": 0.3594399392604828,
      "learning_rate": 3.4348089751364466e-05,
      "loss": 0.068,
      "step": 3500
    },
    {
      "epoch": 1.1542239685658153,
      "grad_norm": 0.013597418554127216,
      "learning_rate": 3.419648271679806e-05,
      "loss": 0.0018,
      "step": 3525
    },
    {
      "epoch": 1.1624099541584807,
      "grad_norm": 0.004102836363017559,
      "learning_rate": 3.404487568223166e-05,
      "loss": 0.0292,
      "step": 3550
    },
    {
      "epoch": 1.170595939751146,
      "grad_norm": 0.0014996347017586231,
      "learning_rate": 3.389326864766525e-05,
      "loss": 0.0114,
      "step": 3575
    },
    {
      "epoch": 1.1787819253438114,
      "grad_norm": 0.018868234008550644,
      "learning_rate": 3.374166161309885e-05,
      "loss": 0.0169,
      "step": 3600
    },
    {
      "epoch": 1.1869679109364768,
      "grad_norm": 0.001259682234376669,
      "learning_rate": 3.3590054578532444e-05,
      "loss": 0.0042,
      "step": 3625
    },
    {
      "epoch": 1.1951538965291422,
      "grad_norm": 0.15819938480854034,
      "learning_rate": 3.343844754396604e-05,
      "loss": 0.0007,
      "step": 3650
    },
    {
      "epoch": 1.2033398821218075,
      "grad_norm": 0.0013165829004719853,
      "learning_rate": 3.3286840509399635e-05,
      "loss": 0.0357,
      "step": 3675
    },
    {
      "epoch": 1.211525867714473,
      "grad_norm": 0.0017084279097616673,
      "learning_rate": 3.313523347483323e-05,
      "loss": 0.0034,
      "step": 3700
    },
    {
      "epoch": 1.2197118533071383,
      "grad_norm": 0.0006795689114369452,
      "learning_rate": 3.2983626440266826e-05,
      "loss": 0.0004,
      "step": 3725
    },
    {
      "epoch": 1.2278978388998034,
      "grad_norm": 0.0009630739223212004,
      "learning_rate": 3.283201940570043e-05,
      "loss": 0.0133,
      "step": 3750
    },
    {
      "epoch": 1.2360838244924688,
      "grad_norm": 0.0106749152764678,
      "learning_rate": 3.2680412371134024e-05,
      "loss": 0.0173,
      "step": 3775
    },
    {
      "epoch": 1.2442698100851342,
      "grad_norm": 0.001351616345345974,
      "learning_rate": 3.252880533656762e-05,
      "loss": 0.0023,
      "step": 3800
    },
    {
      "epoch": 1.2524557956777995,
      "grad_norm": 0.0006208529230207205,
      "learning_rate": 3.2377198302001215e-05,
      "loss": 0.0,
      "step": 3825
    },
    {
      "epoch": 1.260641781270465,
      "grad_norm": 0.0005773603334091604,
      "learning_rate": 3.222559126743481e-05,
      "loss": 0.0373,
      "step": 3850
    },
    {
      "epoch": 1.2688277668631303,
      "grad_norm": 0.0014631077647209167,
      "learning_rate": 3.2073984232868406e-05,
      "loss": 0.0025,
      "step": 3875
    },
    {
      "epoch": 1.2770137524557956,
      "grad_norm": 0.0002054146461887285,
      "learning_rate": 3.1922377198302e-05,
      "loss": 0.0013,
      "step": 3900
    },
    {
      "epoch": 1.285199738048461,
      "grad_norm": 0.0010797285940498114,
      "learning_rate": 3.17707701637356e-05,
      "loss": 0.0427,
      "step": 3925
    },
    {
      "epoch": 1.2933857236411264,
      "grad_norm": 0.022801030427217484,
      "learning_rate": 3.161916312916919e-05,
      "loss": 0.0297,
      "step": 3950
    },
    {
      "epoch": 1.3015717092337917,
      "grad_norm": 0.33411428332328796,
      "learning_rate": 3.146755609460279e-05,
      "loss": 0.0412,
      "step": 3975
    },
    {
      "epoch": 1.309757694826457,
      "grad_norm": 0.014158491976559162,
      "learning_rate": 3.1315949060036384e-05,
      "loss": 0.0185,
      "step": 4000
    },
    {
      "epoch": 1.3179436804191225,
      "grad_norm": 0.00048260248149745166,
      "learning_rate": 3.116434202546999e-05,
      "loss": 0.0,
      "step": 4025
    },
    {
      "epoch": 1.3261296660117878,
      "grad_norm": 0.19578149914741516,
      "learning_rate": 3.101273499090358e-05,
      "loss": 0.0772,
      "step": 4050
    },
    {
      "epoch": 1.3343156516044532,
      "grad_norm": 0.0038196328096091747,
      "learning_rate": 3.086112795633718e-05,
      "loss": 0.0031,
      "step": 4075
    },
    {
      "epoch": 1.3425016371971186,
      "grad_norm": 0.001698371721431613,
      "learning_rate": 3.070952092177077e-05,
      "loss": 0.054,
      "step": 4100
    },
    {
      "epoch": 1.350687622789784,
      "grad_norm": 0.0078005073592066765,
      "learning_rate": 3.055791388720437e-05,
      "loss": 0.0095,
      "step": 4125
    },
    {
      "epoch": 1.3588736083824493,
      "grad_norm": 0.0008572742226533592,
      "learning_rate": 3.040630685263796e-05,
      "loss": 0.0508,
      "step": 4150
    },
    {
      "epoch": 1.3670595939751147,
      "grad_norm": 0.011146394535899162,
      "learning_rate": 3.0254699818071557e-05,
      "loss": 0.0004,
      "step": 4175
    },
    {
      "epoch": 1.37524557956778,
      "grad_norm": 0.003408160526305437,
      "learning_rate": 3.0103092783505156e-05,
      "loss": 0.0431,
      "step": 4200
    },
    {
      "epoch": 1.3834315651604454,
      "grad_norm": 0.01908206194639206,
      "learning_rate": 2.995148574893875e-05,
      "loss": 0.0387,
      "step": 4225
    },
    {
      "epoch": 1.3916175507531108,
      "grad_norm": 0.001946812029927969,
      "learning_rate": 2.9799878714372347e-05,
      "loss": 0.0005,
      "step": 4250
    },
    {
      "epoch": 1.3998035363457761,
      "grad_norm": 0.008685130625963211,
      "learning_rate": 2.9648271679805946e-05,
      "loss": 0.0786,
      "step": 4275
    },
    {
      "epoch": 1.4079895219384415,
      "grad_norm": 0.07280147820711136,
      "learning_rate": 2.949666464523954e-05,
      "loss": 0.0217,
      "step": 4300
    },
    {
      "epoch": 1.4161755075311069,
      "grad_norm": 0.01758684776723385,
      "learning_rate": 2.9345057610673137e-05,
      "loss": 0.0042,
      "step": 4325
    },
    {
      "epoch": 1.424361493123772,
      "grad_norm": 0.003927039448171854,
      "learning_rate": 2.9193450576106736e-05,
      "loss": 0.0033,
      "step": 4350
    },
    {
      "epoch": 1.4325474787164374,
      "grad_norm": 0.003762061009183526,
      "learning_rate": 2.904184354154033e-05,
      "loss": 0.0313,
      "step": 4375
    },
    {
      "epoch": 1.4407334643091028,
      "grad_norm": 0.00642773462459445,
      "learning_rate": 2.8890236506973924e-05,
      "loss": 0.0003,
      "step": 4400
    },
    {
      "epoch": 1.4489194499017681,
      "grad_norm": 0.005841416772454977,
      "learning_rate": 2.873862947240752e-05,
      "loss": 0.0001,
      "step": 4425
    },
    {
      "epoch": 1.4571054354944335,
      "grad_norm": 0.0015138218877837062,
      "learning_rate": 2.8587022437841115e-05,
      "loss": 0.0001,
      "step": 4450
    },
    {
      "epoch": 1.4652914210870989,
      "grad_norm": 0.005790230352431536,
      "learning_rate": 2.8435415403274714e-05,
      "loss": 0.0166,
      "step": 4475
    },
    {
      "epoch": 1.4734774066797642,
      "grad_norm": 0.002549233613535762,
      "learning_rate": 2.828380836870831e-05,
      "loss": 0.0006,
      "step": 4500
    },
    {
      "epoch": 1.4816633922724296,
      "grad_norm": 0.0019311434589326382,
      "learning_rate": 2.8132201334141905e-05,
      "loss": 0.0094,
      "step": 4525
    },
    {
      "epoch": 1.489849377865095,
      "grad_norm": 0.0018175276927649975,
      "learning_rate": 2.7980594299575504e-05,
      "loss": 0.0179,
      "step": 4550
    },
    {
      "epoch": 1.4980353634577603,
      "grad_norm": 0.00037457983125932515,
      "learning_rate": 2.78289872650091e-05,
      "loss": 0.0,
      "step": 4575
    },
    {
      "epoch": 1.5062213490504257,
      "grad_norm": 0.0009425774333067238,
      "learning_rate": 2.7677380230442695e-05,
      "loss": 0.0,
      "step": 4600
    },
    {
      "epoch": 1.514407334643091,
      "grad_norm": 0.00077505485387519,
      "learning_rate": 2.752577319587629e-05,
      "loss": 0.0002,
      "step": 4625
    },
    {
      "epoch": 1.5225933202357562,
      "grad_norm": 0.00036527996417135,
      "learning_rate": 2.737416616130989e-05,
      "loss": 0.0002,
      "step": 4650
    },
    {
      "epoch": 1.5307793058284216,
      "grad_norm": 0.0006952973199076951,
      "learning_rate": 2.7222559126743478e-05,
      "loss": 0.0,
      "step": 4675
    },
    {
      "epoch": 1.538965291421087,
      "grad_norm": 0.0010176642099395394,
      "learning_rate": 2.7070952092177077e-05,
      "loss": 0.0465,
      "step": 4700
    },
    {
      "epoch": 1.5471512770137523,
      "grad_norm": 0.00043089999235235155,
      "learning_rate": 2.6919345057610673e-05,
      "loss": 0.013,
      "step": 4725
    },
    {
      "epoch": 1.5553372626064177,
      "grad_norm": 0.0003544854116626084,
      "learning_rate": 2.676773802304427e-05,
      "loss": 0.0001,
      "step": 4750
    },
    {
      "epoch": 1.563523248199083,
      "grad_norm": 0.0013717171968892217,
      "learning_rate": 2.6616130988477867e-05,
      "loss": 0.0002,
      "step": 4775
    },
    {
      "epoch": 1.5717092337917484,
      "grad_norm": 0.0011255746940150857,
      "learning_rate": 2.6464523953911463e-05,
      "loss": 0.0002,
      "step": 4800
    },
    {
      "epoch": 1.5798952193844138,
      "grad_norm": 0.0009519586456008255,
      "learning_rate": 2.631291691934506e-05,
      "loss": 0.0042,
      "step": 4825
    },
    {
      "epoch": 1.5880812049770792,
      "grad_norm": 0.0012405363377183676,
      "learning_rate": 2.6161309884778657e-05,
      "loss": 0.0,
      "step": 4850
    },
    {
      "epoch": 1.5962671905697445,
      "grad_norm": 0.0004118427459616214,
      "learning_rate": 2.6009702850212253e-05,
      "loss": 0.0515,
      "step": 4875
    },
    {
      "epoch": 1.60445317616241,
      "grad_norm": 0.0020054271444678307,
      "learning_rate": 2.585809581564585e-05,
      "loss": 0.0162,
      "step": 4900
    },
    {
      "epoch": 1.6126391617550753,
      "grad_norm": 0.0009532085387036204,
      "learning_rate": 2.570648878107944e-05,
      "loss": 0.0003,
      "step": 4925
    },
    {
      "epoch": 1.6208251473477406,
      "grad_norm": 0.009459923952817917,
      "learning_rate": 2.5554881746513036e-05,
      "loss": 0.0538,
      "step": 4950
    },
    {
      "epoch": 1.629011132940406,
      "grad_norm": 0.0009569909307174385,
      "learning_rate": 2.5403274711946635e-05,
      "loss": 0.0346,
      "step": 4975
    },
    {
      "epoch": 1.6371971185330714,
      "grad_norm": 0.0007560375961475074,
      "learning_rate": 2.525166767738023e-05,
      "loss": 0.0362,
      "step": 5000
    },
    {
      "epoch": 1.6453831041257367,
      "grad_norm": 0.0004632064083125442,
      "learning_rate": 2.5100060642813826e-05,
      "loss": 0.0005,
      "step": 5025
    },
    {
      "epoch": 1.653569089718402,
      "grad_norm": 0.0006956621073186398,
      "learning_rate": 2.4948453608247425e-05,
      "loss": 0.0028,
      "step": 5050
    },
    {
      "epoch": 1.6617550753110675,
      "grad_norm": 0.0004222158167976886,
      "learning_rate": 2.479684657368102e-05,
      "loss": 0.0002,
      "step": 5075
    },
    {
      "epoch": 1.6699410609037328,
      "grad_norm": 0.0006862828740850091,
      "learning_rate": 2.4645239539114617e-05,
      "loss": 0.0395,
      "step": 5100
    },
    {
      "epoch": 1.6781270464963982,
      "grad_norm": 0.001836837618611753,
      "learning_rate": 2.4493632504548212e-05,
      "loss": 0.0268,
      "step": 5125
    },
    {
      "epoch": 1.6863130320890636,
      "grad_norm": 0.0008472290937788785,
      "learning_rate": 2.4342025469981808e-05,
      "loss": 0.0028,
      "step": 5150
    },
    {
      "epoch": 1.694499017681729,
      "grad_norm": 0.0012132123811170459,
      "learning_rate": 2.4190418435415403e-05,
      "loss": 0.0001,
      "step": 5175
    },
    {
      "epoch": 1.7026850032743943,
      "grad_norm": 0.0005179749568924308,
      "learning_rate": 2.4038811400849002e-05,
      "loss": 0.0397,
      "step": 5200
    },
    {
      "epoch": 1.7108709888670597,
      "grad_norm": 0.0008750595734454691,
      "learning_rate": 2.3887204366282598e-05,
      "loss": 0.0002,
      "step": 5225
    },
    {
      "epoch": 1.719056974459725,
      "grad_norm": 0.02541407383978367,
      "learning_rate": 2.373559733171619e-05,
      "loss": 0.0001,
      "step": 5250
    },
    {
      "epoch": 1.7272429600523904,
      "grad_norm": 0.02840971015393734,
      "learning_rate": 2.358399029714979e-05,
      "loss": 0.0002,
      "step": 5275
    },
    {
      "epoch": 1.7354289456450558,
      "grad_norm": 0.0008489146712236106,
      "learning_rate": 2.3432383262583385e-05,
      "loss": 0.0142,
      "step": 5300
    },
    {
      "epoch": 1.7436149312377212,
      "grad_norm": 0.002642701379954815,
      "learning_rate": 2.328077622801698e-05,
      "loss": 0.0562,
      "step": 5325
    },
    {
      "epoch": 1.7518009168303865,
      "grad_norm": 0.004949661903083324,
      "learning_rate": 2.312916919345058e-05,
      "loss": 0.0147,
      "step": 5350
    },
    {
      "epoch": 1.759986902423052,
      "grad_norm": 0.11022938042879105,
      "learning_rate": 2.297756215888417e-05,
      "loss": 0.028,
      "step": 5375
    },
    {
      "epoch": 1.768172888015717,
      "grad_norm": 0.006674617063254118,
      "learning_rate": 2.282595512431777e-05,
      "loss": 0.0014,
      "step": 5400
    },
    {
      "epoch": 1.7763588736083824,
      "grad_norm": 0.0017073052003979683,
      "learning_rate": 2.2674348089751366e-05,
      "loss": 0.0028,
      "step": 5425
    },
    {
      "epoch": 1.7845448592010478,
      "grad_norm": 3.2796061038970947,
      "learning_rate": 2.252274105518496e-05,
      "loss": 0.0025,
      "step": 5450
    },
    {
      "epoch": 1.7927308447937131,
      "grad_norm": 0.0006569511606357992,
      "learning_rate": 2.237113402061856e-05,
      "loss": 0.0009,
      "step": 5475
    },
    {
      "epoch": 1.8009168303863785,
      "grad_norm": 0.00020327128004282713,
      "learning_rate": 2.2219526986052153e-05,
      "loss": 0.0,
      "step": 5500
    },
    {
      "epoch": 1.8091028159790439,
      "grad_norm": 0.041166484355926514,
      "learning_rate": 2.2067919951485748e-05,
      "loss": 0.0937,
      "step": 5525
    },
    {
      "epoch": 1.8172888015717092,
      "grad_norm": 0.05974828451871872,
      "learning_rate": 2.1916312916919347e-05,
      "loss": 0.0147,
      "step": 5550
    },
    {
      "epoch": 1.8254747871643746,
      "grad_norm": 0.004328828305006027,
      "learning_rate": 2.1764705882352943e-05,
      "loss": 0.0041,
      "step": 5575
    },
    {
      "epoch": 1.83366077275704,
      "grad_norm": 0.004382645711302757,
      "learning_rate": 2.1613098847786538e-05,
      "loss": 0.0006,
      "step": 5600
    },
    {
      "epoch": 1.8418467583497053,
      "grad_norm": 0.005558451171964407,
      "learning_rate": 2.1461491813220134e-05,
      "loss": 0.0002,
      "step": 5625
    },
    {
      "epoch": 1.8500327439423707,
      "grad_norm": 0.02043689414858818,
      "learning_rate": 2.130988477865373e-05,
      "loss": 0.044,
      "step": 5650
    },
    {
      "epoch": 1.8582187295350359,
      "grad_norm": 0.007403884083032608,
      "learning_rate": 2.1158277744087325e-05,
      "loss": 0.0295,
      "step": 5675
    },
    {
      "epoch": 1.8664047151277012,
      "grad_norm": 0.002108983462676406,
      "learning_rate": 2.1006670709520924e-05,
      "loss": 0.0033,
      "step": 5700
    },
    {
      "epoch": 1.8745907007203666,
      "grad_norm": 0.0540454126894474,
      "learning_rate": 2.085506367495452e-05,
      "loss": 0.0247,
      "step": 5725
    },
    {
      "epoch": 1.882776686313032,
      "grad_norm": 0.02874263934791088,
      "learning_rate": 2.0703456640388115e-05,
      "loss": 0.0117,
      "step": 5750
    },
    {
      "epoch": 1.8909626719056973,
      "grad_norm": 0.0010341499000787735,
      "learning_rate": 2.055184960582171e-05,
      "loss": 0.0029,
      "step": 5775
    },
    {
      "epoch": 1.8991486574983627,
      "grad_norm": 0.0022526606917381287,
      "learning_rate": 2.0400242571255306e-05,
      "loss": 0.0404,
      "step": 5800
    },
    {
      "epoch": 1.907334643091028,
      "grad_norm": 0.0009332990157417953,
      "learning_rate": 2.0248635536688905e-05,
      "loss": 0.0001,
      "step": 5825
    },
    {
      "epoch": 1.9155206286836934,
      "grad_norm": 0.01605447567999363,
      "learning_rate": 2.00970285021225e-05,
      "loss": 0.0744,
      "step": 5850
    },
    {
      "epoch": 1.9237066142763588,
      "grad_norm": 0.057828161865472794,
      "learning_rate": 1.9945421467556096e-05,
      "loss": 0.038,
      "step": 5875
    },
    {
      "epoch": 1.9318925998690242,
      "grad_norm": 0.01713876612484455,
      "learning_rate": 1.9793814432989692e-05,
      "loss": 0.0071,
      "step": 5900
    },
    {
      "epoch": 1.9400785854616895,
      "grad_norm": 0.0016962456284090877,
      "learning_rate": 1.9642207398423287e-05,
      "loss": 0.0075,
      "step": 5925
    },
    {
      "epoch": 1.948264571054355,
      "grad_norm": 0.001768580754287541,
      "learning_rate": 1.9490600363856883e-05,
      "loss": 0.0008,
      "step": 5950
    },
    {
      "epoch": 1.9564505566470203,
      "grad_norm": 0.005728750489652157,
      "learning_rate": 1.9338993329290482e-05,
      "loss": 0.0627,
      "step": 5975
    },
    {
      "epoch": 1.9646365422396856,
      "grad_norm": 7.606292724609375,
      "learning_rate": 1.9187386294724078e-05,
      "loss": 0.0418,
      "step": 6000
    },
    {
      "epoch": 1.972822527832351,
      "grad_norm": 0.0018708339193835855,
      "learning_rate": 1.903577926015767e-05,
      "loss": 0.0006,
      "step": 6025
    },
    {
      "epoch": 1.9810085134250164,
      "grad_norm": 7.210286617279053,
      "learning_rate": 1.888417222559127e-05,
      "loss": 0.0324,
      "step": 6050
    },
    {
      "epoch": 1.9891944990176817,
      "grad_norm": 0.021973388269543648,
      "learning_rate": 1.8732565191024864e-05,
      "loss": 0.0005,
      "step": 6075
    },
    {
      "epoch": 1.9973804846103471,
      "grad_norm": 0.2803119421005249,
      "learning_rate": 1.858095815645846e-05,
      "loss": 0.0454,
      "step": 6100
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9975442043222004,
      "eval_auc": 0.9995912800437162,
      "eval_f1": 0.9044585987261147,
      "eval_loss": 0.008845364674925804,
      "eval_precision": 0.8554216867469879,
      "eval_recall": 0.9594594594594594,
      "eval_runtime": 74.792,
      "eval_samples_per_second": 81.666,
      "eval_steps_per_second": 5.107,
      "step": 6108
    },
    {
      "epoch": 2.0055664702030125,
      "grad_norm": 0.022678203880786896,
      "learning_rate": 1.8435415403274715e-05,
      "loss": 0.0121,
      "step": 6125
    },
    {
      "epoch": 2.013752455795678,
      "grad_norm": 0.0033217219170182943,
      "learning_rate": 1.8283808368708307e-05,
      "loss": 0.0446,
      "step": 6150
    },
    {
      "epoch": 2.021938441388343,
      "grad_norm": 0.006960027851164341,
      "learning_rate": 1.8132201334141906e-05,
      "loss": 0.0067,
      "step": 6175
    },
    {
      "epoch": 2.0301244269810086,
      "grad_norm": 1.543547511100769,
      "learning_rate": 1.79805942995755e-05,
      "loss": 0.0201,
      "step": 6200
    },
    {
      "epoch": 2.038310412573674,
      "grad_norm": 0.01694468781352043,
      "learning_rate": 1.7828987265009097e-05,
      "loss": 0.0258,
      "step": 6225
    },
    {
      "epoch": 2.0464963981663393,
      "grad_norm": 0.004578434396535158,
      "learning_rate": 1.7677380230442696e-05,
      "loss": 0.0006,
      "step": 6250
    },
    {
      "epoch": 2.0546823837590047,
      "grad_norm": 1.5190191268920898,
      "learning_rate": 1.7525773195876288e-05,
      "loss": 0.0126,
      "step": 6275
    },
    {
      "epoch": 2.06286836935167,
      "grad_norm": 0.0018938812427222729,
      "learning_rate": 1.7374166161309884e-05,
      "loss": 0.0003,
      "step": 6300
    },
    {
      "epoch": 2.0710543549443354,
      "grad_norm": 0.0011117063695564866,
      "learning_rate": 1.7222559126743483e-05,
      "loss": 0.0002,
      "step": 6325
    },
    {
      "epoch": 2.079240340537001,
      "grad_norm": 0.000687290565110743,
      "learning_rate": 1.7070952092177078e-05,
      "loss": 0.0154,
      "step": 6350
    },
    {
      "epoch": 2.087426326129666,
      "grad_norm": 0.001672151731327176,
      "learning_rate": 1.6919345057610674e-05,
      "loss": 0.0006,
      "step": 6375
    },
    {
      "epoch": 2.0956123117223315,
      "grad_norm": 5.109748363494873,
      "learning_rate": 1.676773802304427e-05,
      "loss": 0.0352,
      "step": 6400
    },
    {
      "epoch": 2.103798297314997,
      "grad_norm": 0.020103197544813156,
      "learning_rate": 1.6616130988477865e-05,
      "loss": 0.0007,
      "step": 6425
    },
    {
      "epoch": 2.1119842829076623,
      "grad_norm": 0.005654872860759497,
      "learning_rate": 1.646452395391146e-05,
      "loss": 0.0004,
      "step": 6450
    },
    {
      "epoch": 2.1201702685003276,
      "grad_norm": 0.009023274295032024,
      "learning_rate": 1.631291691934506e-05,
      "loss": 0.0018,
      "step": 6475
    },
    {
      "epoch": 2.128356254092993,
      "grad_norm": 0.0016646806616336107,
      "learning_rate": 1.6161309884778655e-05,
      "loss": 0.0001,
      "step": 6500
    },
    {
      "epoch": 2.1365422396856584,
      "grad_norm": 0.0016474899603053927,
      "learning_rate": 1.600970285021225e-05,
      "loss": 0.0005,
      "step": 6525
    },
    {
      "epoch": 2.1447282252783237,
      "grad_norm": 0.014943123795092106,
      "learning_rate": 1.5858095815645846e-05,
      "loss": 0.0147,
      "step": 6550
    },
    {
      "epoch": 2.1529142108709887,
      "grad_norm": 0.0021474878303706646,
      "learning_rate": 1.570648878107944e-05,
      "loss": 0.0002,
      "step": 6575
    },
    {
      "epoch": 2.161100196463654,
      "grad_norm": 0.0017727293306961656,
      "learning_rate": 1.555488174651304e-05,
      "loss": 0.0,
      "step": 6600
    },
    {
      "epoch": 2.1692861820563194,
      "grad_norm": 0.0010321360314264894,
      "learning_rate": 1.5403274711946636e-05,
      "loss": 0.0001,
      "step": 6625
    },
    {
      "epoch": 2.1774721676489848,
      "grad_norm": 0.0004758843861054629,
      "learning_rate": 1.5251667677380232e-05,
      "loss": 0.0114,
      "step": 6650
    },
    {
      "epoch": 2.18565815324165,
      "grad_norm": 0.0012892503291368484,
      "learning_rate": 1.5100060642813826e-05,
      "loss": 0.0262,
      "step": 6675
    },
    {
      "epoch": 2.1938441388343155,
      "grad_norm": 0.01857004500925541,
      "learning_rate": 1.4948453608247423e-05,
      "loss": 0.0003,
      "step": 6700
    },
    {
      "epoch": 2.202030124426981,
      "grad_norm": 0.0008667947258800268,
      "learning_rate": 1.479684657368102e-05,
      "loss": 0.0099,
      "step": 6725
    },
    {
      "epoch": 2.2102161100196462,
      "grad_norm": 0.001100408611819148,
      "learning_rate": 1.4645239539114616e-05,
      "loss": 0.0341,
      "step": 6750
    },
    {
      "epoch": 2.2184020956123116,
      "grad_norm": 0.001291294815018773,
      "learning_rate": 1.4493632504548213e-05,
      "loss": 0.0001,
      "step": 6775
    },
    {
      "epoch": 2.226588081204977,
      "grad_norm": 0.011078906245529652,
      "learning_rate": 1.4342025469981807e-05,
      "loss": 0.0479,
      "step": 6800
    },
    {
      "epoch": 2.2347740667976423,
      "grad_norm": 0.0018653717124834657,
      "learning_rate": 1.4190418435415404e-05,
      "loss": 0.0026,
      "step": 6825
    },
    {
      "epoch": 2.2429600523903077,
      "grad_norm": 0.0010668339673429728,
      "learning_rate": 1.4038811400849e-05,
      "loss": 0.0005,
      "step": 6850
    },
    {
      "epoch": 2.251146037982973,
      "grad_norm": 0.002098933095112443,
      "learning_rate": 1.3887204366282597e-05,
      "loss": 0.0005,
      "step": 6875
    },
    {
      "epoch": 2.2593320235756384,
      "grad_norm": 0.0008214248227886856,
      "learning_rate": 1.3735597331716193e-05,
      "loss": 0.0185,
      "step": 6900
    },
    {
      "epoch": 2.267518009168304,
      "grad_norm": 0.001296183792874217,
      "learning_rate": 1.3583990297149786e-05,
      "loss": 0.0354,
      "step": 6925
    },
    {
      "epoch": 2.275703994760969,
      "grad_norm": 0.0009354737703688443,
      "learning_rate": 1.3432383262583384e-05,
      "loss": 0.0127,
      "step": 6950
    },
    {
      "epoch": 2.2838899803536346,
      "grad_norm": 0.0020546901505440474,
      "learning_rate": 1.3280776228016981e-05,
      "loss": 0.0004,
      "step": 6975
    },
    {
      "epoch": 2.2920759659463,
      "grad_norm": 0.02266145683825016,
      "learning_rate": 1.3129169193450577e-05,
      "loss": 0.0364,
      "step": 7000
    },
    {
      "epoch": 2.3002619515389653,
      "grad_norm": 0.010976303368806839,
      "learning_rate": 1.2977562158884174e-05,
      "loss": 0.0018,
      "step": 7025
    },
    {
      "epoch": 2.3084479371316307,
      "grad_norm": 0.006094989832490683,
      "learning_rate": 1.2825955124317768e-05,
      "loss": 0.0091,
      "step": 7050
    },
    {
      "epoch": 2.316633922724296,
      "grad_norm": 0.07841998338699341,
      "learning_rate": 1.2674348089751365e-05,
      "loss": 0.0094,
      "step": 7075
    },
    {
      "epoch": 2.3248199083169614,
      "grad_norm": 2.630479335784912,
      "learning_rate": 1.252274105518496e-05,
      "loss": 0.0029,
      "step": 7100
    },
    {
      "epoch": 2.3330058939096268,
      "grad_norm": 0.0039812372997403145,
      "learning_rate": 1.2371134020618558e-05,
      "loss": 0.0086,
      "step": 7125
    },
    {
      "epoch": 2.341191879502292,
      "grad_norm": 0.0007643400458618999,
      "learning_rate": 1.2219526986052153e-05,
      "loss": 0.0001,
      "step": 7150
    },
    {
      "epoch": 2.3493778650949575,
      "grad_norm": 0.0021422533318400383,
      "learning_rate": 1.2067919951485749e-05,
      "loss": 0.0029,
      "step": 7175
    },
    {
      "epoch": 2.357563850687623,
      "grad_norm": 0.0017942421836778522,
      "learning_rate": 1.1916312916919346e-05,
      "loss": 0.0026,
      "step": 7200
    },
    {
      "epoch": 2.3657498362802882,
      "grad_norm": 0.0007688822224736214,
      "learning_rate": 1.1764705882352942e-05,
      "loss": 0.0018,
      "step": 7225
    },
    {
      "epoch": 2.3739358218729536,
      "grad_norm": 0.0004730868386104703,
      "learning_rate": 1.1613098847786537e-05,
      "loss": 0.0028,
      "step": 7250
    },
    {
      "epoch": 2.382121807465619,
      "grad_norm": 0.0004939533537253737,
      "learning_rate": 1.1461491813220133e-05,
      "loss": 0.0003,
      "step": 7275
    },
    {
      "epoch": 2.3903077930582843,
      "grad_norm": 0.0010573838371783495,
      "learning_rate": 1.130988477865373e-05,
      "loss": 0.0158,
      "step": 7300
    },
    {
      "epoch": 2.3984937786509497,
      "grad_norm": 0.0003521484904922545,
      "learning_rate": 1.1158277744087328e-05,
      "loss": 0.0008,
      "step": 7325
    },
    {
      "epoch": 2.406679764243615,
      "grad_norm": 0.00083108467515558,
      "learning_rate": 1.1006670709520921e-05,
      "loss": 0.0209,
      "step": 7350
    },
    {
      "epoch": 2.4148657498362804,
      "grad_norm": 0.00046702896361239254,
      "learning_rate": 1.0855063674954519e-05,
      "loss": 0.0002,
      "step": 7375
    },
    {
      "epoch": 2.423051735428946,
      "grad_norm": 0.0009466594783589244,
      "learning_rate": 1.0703456640388114e-05,
      "loss": 0.0534,
      "step": 7400
    },
    {
      "epoch": 2.431237721021611,
      "grad_norm": 0.0005027965526096523,
      "learning_rate": 1.055184960582171e-05,
      "loss": 0.0031,
      "step": 7425
    },
    {
      "epoch": 2.4394237066142765,
      "grad_norm": 0.00883590430021286,
      "learning_rate": 1.0400242571255307e-05,
      "loss": 0.0078,
      "step": 7450
    },
    {
      "epoch": 2.4476096922069415,
      "grad_norm": 0.0009790252661332488,
      "learning_rate": 1.0248635536688903e-05,
      "loss": 0.002,
      "step": 7475
    },
    {
      "epoch": 2.455795677799607,
      "grad_norm": 0.017697228118777275,
      "learning_rate": 1.00970285021225e-05,
      "loss": 0.0049,
      "step": 7500
    },
    {
      "epoch": 2.463981663392272,
      "grad_norm": 0.0005443913978524506,
      "learning_rate": 9.945421467556094e-06,
      "loss": 0.0672,
      "step": 7525
    },
    {
      "epoch": 2.4721676489849376,
      "grad_norm": 0.0002733923611231148,
      "learning_rate": 9.793814432989691e-06,
      "loss": 0.0089,
      "step": 7550
    },
    {
      "epoch": 2.480353634577603,
      "grad_norm": 0.002381040947511792,
      "learning_rate": 9.642207398423288e-06,
      "loss": 0.0001,
      "step": 7575
    },
    {
      "epoch": 2.4885396201702683,
      "grad_norm": 0.0006111008697189391,
      "learning_rate": 9.490600363856882e-06,
      "loss": 0.0,
      "step": 7600
    },
    {
      "epoch": 2.4967256057629337,
      "grad_norm": 0.0005736036109738052,
      "learning_rate": 9.33899332929048e-06,
      "loss": 0.0176,
      "step": 7625
    },
    {
      "epoch": 2.504911591355599,
      "grad_norm": 0.00394043792039156,
      "learning_rate": 9.187386294724077e-06,
      "loss": 0.0039,
      "step": 7650
    },
    {
      "epoch": 2.5130975769482644,
      "grad_norm": 0.0005367947742342949,
      "learning_rate": 9.035779260157672e-06,
      "loss": 0.0001,
      "step": 7675
    },
    {
      "epoch": 2.52128356254093,
      "grad_norm": 0.004565802868455648,
      "learning_rate": 8.884172225591268e-06,
      "loss": 0.0255,
      "step": 7700
    },
    {
      "epoch": 2.529469548133595,
      "grad_norm": 0.0027460469864308834,
      "learning_rate": 8.732565191024863e-06,
      "loss": 0.0007,
      "step": 7725
    },
    {
      "epoch": 2.5376555337262605,
      "grad_norm": 0.006453353445976973,
      "learning_rate": 8.58095815645846e-06,
      "loss": 0.0003,
      "step": 7750
    },
    {
      "epoch": 2.545841519318926,
      "grad_norm": 0.05628466606140137,
      "learning_rate": 8.429351121892056e-06,
      "loss": 0.0005,
      "step": 7775
    },
    {
      "epoch": 2.5540275049115913,
      "grad_norm": 0.0020795781165361404,
      "learning_rate": 8.277744087325652e-06,
      "loss": 0.0001,
      "step": 7800
    },
    {
      "epoch": 2.5622134905042566,
      "grad_norm": 6.260252952575684,
      "learning_rate": 8.126137052759249e-06,
      "loss": 0.065,
      "step": 7825
    },
    {
      "epoch": 2.570399476096922,
      "grad_norm": 0.032053545117378235,
      "learning_rate": 7.974530018192845e-06,
      "loss": 0.0005,
      "step": 7850
    },
    {
      "epoch": 2.5785854616895874,
      "grad_norm": 0.0007737306877970695,
      "learning_rate": 7.82292298362644e-06,
      "loss": 0.0005,
      "step": 7875
    },
    {
      "epoch": 2.5867714472822527,
      "grad_norm": 0.0028309288900345564,
      "learning_rate": 7.671315949060038e-06,
      "loss": 0.0004,
      "step": 7900
    },
    {
      "epoch": 2.594957432874918,
      "grad_norm": 0.002330320654436946,
      "learning_rate": 7.519708914493632e-06,
      "loss": 0.0001,
      "step": 7925
    },
    {
      "epoch": 2.6031434184675835,
      "grad_norm": 0.002036983147263527,
      "learning_rate": 7.3681018799272296e-06,
      "loss": 0.0011,
      "step": 7950
    },
    {
      "epoch": 2.611329404060249,
      "grad_norm": 0.0005035591893829405,
      "learning_rate": 7.216494845360824e-06,
      "loss": 0.0012,
      "step": 7975
    },
    {
      "epoch": 2.619515389652914,
      "grad_norm": 0.011760620400309563,
      "learning_rate": 7.064887810794421e-06,
      "loss": 0.0015,
      "step": 8000
    },
    {
      "epoch": 2.6277013752455796,
      "grad_norm": 0.07406982779502869,
      "learning_rate": 6.913280776228018e-06,
      "loss": 0.0001,
      "step": 8025
    },
    {
      "epoch": 2.635887360838245,
      "grad_norm": 0.002441051648929715,
      "learning_rate": 6.761673741661613e-06,
      "loss": 0.0008,
      "step": 8050
    },
    {
      "epoch": 2.6440733464309103,
      "grad_norm": 0.001771932584233582,
      "learning_rate": 6.61006670709521e-06,
      "loss": 0.0261,
      "step": 8075
    },
    {
      "epoch": 2.6522593320235757,
      "grad_norm": 0.003007644321769476,
      "learning_rate": 6.458459672528806e-06,
      "loss": 0.0027,
      "step": 8100
    },
    {
      "epoch": 2.660445317616241,
      "grad_norm": 0.0029620120767503977,
      "learning_rate": 6.306852637962402e-06,
      "loss": 0.0001,
      "step": 8125
    },
    {
      "epoch": 2.6686313032089064,
      "grad_norm": 0.002668931847438216,
      "learning_rate": 6.161309884778654e-06,
      "loss": 0.0218,
      "step": 8150
    },
    {
      "epoch": 2.6768172888015718,
      "grad_norm": 0.005034355912357569,
      "learning_rate": 6.0097028502122506e-06,
      "loss": 0.0002,
      "step": 8175
    },
    {
      "epoch": 2.685003274394237,
      "grad_norm": 0.2363610863685608,
      "learning_rate": 5.858095815645846e-06,
      "loss": 0.0237,
      "step": 8200
    },
    {
      "epoch": 2.6931892599869025,
      "grad_norm": 0.0029050330631434917,
      "learning_rate": 5.7064887810794426e-06,
      "loss": 0.0142,
      "step": 8225
    },
    {
      "epoch": 2.701375245579568,
      "grad_norm": 0.0002962597936857492,
      "learning_rate": 5.554881746513038e-06,
      "loss": 0.0008,
      "step": 8250
    },
    {
      "epoch": 2.7095612311722332,
      "grad_norm": 0.002445927122607827,
      "learning_rate": 5.4032747119466346e-06,
      "loss": 0.0395,
      "step": 8275
    },
    {
      "epoch": 2.7177472167648986,
      "grad_norm": 4.508545875549316,
      "learning_rate": 5.251667677380231e-06,
      "loss": 0.0135,
      "step": 8300
    },
    {
      "epoch": 2.725933202357564,
      "grad_norm": 0.001652977429330349,
      "learning_rate": 5.1000606428138266e-06,
      "loss": 0.0005,
      "step": 8325
    },
    {
      "epoch": 2.7341191879502293,
      "grad_norm": 0.007348712533712387,
      "learning_rate": 4.948453608247423e-06,
      "loss": 0.0001,
      "step": 8350
    },
    {
      "epoch": 2.7423051735428947,
      "grad_norm": 0.0047335317358374596,
      "learning_rate": 4.796846573681019e-06,
      "loss": 0.0002,
      "step": 8375
    },
    {
      "epoch": 2.75049115913556,
      "grad_norm": 0.0006250338046811521,
      "learning_rate": 4.645239539114615e-06,
      "loss": 0.002,
      "step": 8400
    },
    {
      "epoch": 2.7586771447282254,
      "grad_norm": 0.0020159140694886446,
      "learning_rate": 4.493632504548211e-06,
      "loss": 0.0001,
      "step": 8425
    },
    {
      "epoch": 2.766863130320891,
      "grad_norm": 0.005961321294307709,
      "learning_rate": 4.342025469981807e-06,
      "loss": 0.0825,
      "step": 8450
    },
    {
      "epoch": 2.775049115913556,
      "grad_norm": 0.004888585302978754,
      "learning_rate": 4.190418435415403e-06,
      "loss": 0.0003,
      "step": 8475
    },
    {
      "epoch": 2.7832351015062216,
      "grad_norm": 0.0033836057409644127,
      "learning_rate": 4.038811400849e-06,
      "loss": 0.0002,
      "step": 8500
    },
    {
      "epoch": 2.791421087098887,
      "grad_norm": 0.048891954123973846,
      "learning_rate": 3.887204366282595e-06,
      "loss": 0.0002,
      "step": 8525
    },
    {
      "epoch": 2.7996070726915523,
      "grad_norm": 0.0019098659977316856,
      "learning_rate": 3.735597331716192e-06,
      "loss": 0.0014,
      "step": 8550
    },
    {
      "epoch": 2.8077930582842177,
      "grad_norm": 0.005250090733170509,
      "learning_rate": 3.583990297149788e-06,
      "loss": 0.0019,
      "step": 8575
    },
    {
      "epoch": 2.815979043876883,
      "grad_norm": 0.0005940294940955937,
      "learning_rate": 3.432383262583384e-06,
      "loss": 0.0003,
      "step": 8600
    },
    {
      "epoch": 2.8241650294695484,
      "grad_norm": 0.00769090885296464,
      "learning_rate": 3.28077622801698e-06,
      "loss": 0.0003,
      "step": 8625
    },
    {
      "epoch": 2.8323510150622138,
      "grad_norm": 0.004769071005284786,
      "learning_rate": 3.129169193450576e-06,
      "loss": 0.0082,
      "step": 8650
    },
    {
      "epoch": 2.8405370006548787,
      "grad_norm": 0.0006792404456064105,
      "learning_rate": 2.983626440266829e-06,
      "loss": 0.071,
      "step": 8675
    },
    {
      "epoch": 2.848722986247544,
      "grad_norm": 0.0006690117879770696,
      "learning_rate": 2.832019405700425e-06,
      "loss": 0.0303,
      "step": 8700
    },
    {
      "epoch": 2.8569089718402094,
      "grad_norm": 0.003535321680828929,
      "learning_rate": 2.6804123711340204e-06,
      "loss": 0.0041,
      "step": 8725
    },
    {
      "epoch": 2.865094957432875,
      "grad_norm": 0.007222822401672602,
      "learning_rate": 2.528805336567617e-06,
      "loss": 0.0002,
      "step": 8750
    },
    {
      "epoch": 2.87328094302554,
      "grad_norm": 0.001671507372520864,
      "learning_rate": 2.377198302001213e-06,
      "loss": 0.0102,
      "step": 8775
    },
    {
      "epoch": 2.8814669286182055,
      "grad_norm": 0.0008472661720588803,
      "learning_rate": 2.2255912674348092e-06,
      "loss": 0.0004,
      "step": 8800
    },
    {
      "epoch": 2.889652914210871,
      "grad_norm": 0.0037155933678150177,
      "learning_rate": 2.0739842328684052e-06,
      "loss": 0.0007,
      "step": 8825
    },
    {
      "epoch": 2.8978388998035363,
      "grad_norm": 0.0019001095788553357,
      "learning_rate": 1.9223771983020012e-06,
      "loss": 0.0071,
      "step": 8850
    },
    {
      "epoch": 2.9060248853962016,
      "grad_norm": 0.01914367638528347,
      "learning_rate": 1.7707701637355974e-06,
      "loss": 0.0003,
      "step": 8875
    },
    {
      "epoch": 2.914210870988867,
      "grad_norm": 0.0017183177405968308,
      "learning_rate": 1.6191631291691936e-06,
      "loss": 0.0001,
      "step": 8900
    },
    {
      "epoch": 2.9223968565815324,
      "grad_norm": 0.000283712986856699,
      "learning_rate": 1.4675560946027896e-06,
      "loss": 0.007,
      "step": 8925
    },
    {
      "epoch": 2.9305828421741977,
      "grad_norm": 0.0013638654490932822,
      "learning_rate": 1.3159490600363858e-06,
      "loss": 0.031,
      "step": 8950
    },
    {
      "epoch": 2.938768827766863,
      "grad_norm": 0.0024313374888151884,
      "learning_rate": 1.1643420254699818e-06,
      "loss": 0.0001,
      "step": 8975
    },
    {
      "epoch": 2.9469548133595285,
      "grad_norm": 0.0011194221442565322,
      "learning_rate": 1.012734990903578e-06,
      "loss": 0.0001,
      "step": 9000
    },
    {
      "epoch": 2.955140798952194,
      "grad_norm": 0.0018601082265377045,
      "learning_rate": 8.611279563371741e-07,
      "loss": 0.0091,
      "step": 9025
    },
    {
      "epoch": 2.963326784544859,
      "grad_norm": 0.001517856726422906,
      "learning_rate": 7.095209217707701e-07,
      "loss": 0.0003,
      "step": 9050
    },
    {
      "epoch": 2.9715127701375246,
      "grad_norm": 0.00028195424238219857,
      "learning_rate": 5.579138872043664e-07,
      "loss": 0.0225,
      "step": 9075
    },
    {
      "epoch": 2.97969875573019,
      "grad_norm": 0.0005502802086994052,
      "learning_rate": 4.0630685263796246e-07,
      "loss": 0.0001,
      "step": 9100
    },
    {
      "epoch": 2.9878847413228553,
      "grad_norm": 0.0016429908573627472,
      "learning_rate": 2.5469981807155856e-07,
      "loss": 0.0003,
      "step": 9125
    },
    {
      "epoch": 2.9960707269155207,
      "grad_norm": 0.0061945500783622265,
      "learning_rate": 1.0309278350515465e-07,
      "loss": 0.0033,
      "step": 9150
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9988539620170268,
      "eval_auc": 0.9999216153508497,
      "eval_f1": 0.9523809523809523,
      "eval_loss": 0.0027391575276851654,
      "eval_precision": 0.958904109589041,
      "eval_recall": 0.9459459459459459,
      "eval_runtime": 74.4951,
      "eval_samples_per_second": 81.992,
      "eval_steps_per_second": 5.128,
      "step": 9162
    }
  ],
  "logging_steps": 25,
  "max_steps": 9162,
  "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": 2
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.944931470697103e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}