{
  "best_metric": 0.4526550769805908,
  "best_model_checkpoint": "NLP-project-training/checkpoint-5250",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 5250,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.014285714285714285,
      "grad_norm": 8.86188793182373,
      "learning_rate": 2.3809523809523808e-06,
      "loss": 0.7214,
      "step": 25
    },
    {
      "epoch": 0.02857142857142857,
      "grad_norm": 24.593616485595703,
      "learning_rate": 4.7619047619047615e-06,
      "loss": 0.6419,
      "step": 50
    },
    {
      "epoch": 0.04285714285714286,
      "grad_norm": 21.314754486083984,
      "learning_rate": 7.142857142857143e-06,
      "loss": 0.6406,
      "step": 75
    },
    {
      "epoch": 0.05714285714285714,
      "grad_norm": 15.490625381469727,
      "learning_rate": 9.523809523809523e-06,
      "loss": 0.5705,
      "step": 100
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 21.33145523071289,
      "learning_rate": 1.1904761904761905e-05,
      "loss": 0.5283,
      "step": 125
    },
    {
      "epoch": 0.08571428571428572,
      "grad_norm": 11.36423397064209,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 0.5416,
      "step": 150
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.304591178894043,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.5374,
      "step": 175
    },
    {
      "epoch": 0.11428571428571428,
      "grad_norm": 16.739395141601562,
      "learning_rate": 1.9047619047619046e-05,
      "loss": 0.6485,
      "step": 200
    },
    {
      "epoch": 0.12857142857142856,
      "grad_norm": 6.840592861175537,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 0.5512,
      "step": 225
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 6.98801326751709,
      "learning_rate": 2.380952380952381e-05,
      "loss": 0.6068,
      "step": 250
    },
    {
      "epoch": 0.15714285714285714,
      "grad_norm": 4.45384407043457,
      "learning_rate": 2.6190476190476192e-05,
      "loss": 0.5741,
      "step": 275
    },
    {
      "epoch": 0.17142857142857143,
      "grad_norm": 27.597862243652344,
      "learning_rate": 2.857142857142857e-05,
      "loss": 0.57,
      "step": 300
    },
    {
      "epoch": 0.18571428571428572,
      "grad_norm": 18.127330780029297,
      "learning_rate": 3.095238095238095e-05,
      "loss": 0.6102,
      "step": 325
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.04953670501709,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.6557,
      "step": 350
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 6.661813259124756,
      "learning_rate": 3.571428571428572e-05,
      "loss": 0.5559,
      "step": 375
    },
    {
      "epoch": 0.22857142857142856,
      "grad_norm": 5.1511735916137695,
      "learning_rate": 3.809523809523809e-05,
      "loss": 0.5711,
      "step": 400
    },
    {
      "epoch": 0.24285714285714285,
      "grad_norm": 6.911036014556885,
      "learning_rate": 4.047619047619048e-05,
      "loss": 0.6384,
      "step": 425
    },
    {
      "epoch": 0.2571428571428571,
      "grad_norm": 12.719388961791992,
      "learning_rate": 4.2857142857142856e-05,
      "loss": 0.5132,
      "step": 450
    },
    {
      "epoch": 0.2714285714285714,
      "grad_norm": 1.8966089487075806,
      "learning_rate": 4.523809523809524e-05,
      "loss": 0.5363,
      "step": 475
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 4.538297653198242,
      "learning_rate": 4.761904761904762e-05,
      "loss": 0.4992,
      "step": 500
    },
    {
      "epoch": 0.3,
      "grad_norm": 15.931159019470215,
      "learning_rate": 5e-05,
      "loss": 0.5498,
      "step": 525
    },
    {
      "epoch": 0.3142857142857143,
      "grad_norm": 3.791177988052368,
      "learning_rate": 4.973544973544973e-05,
      "loss": 0.5859,
      "step": 550
    },
    {
      "epoch": 0.32857142857142857,
      "grad_norm": 5.441797733306885,
      "learning_rate": 4.9470899470899475e-05,
      "loss": 0.5448,
      "step": 575
    },
    {
      "epoch": 0.34285714285714286,
      "grad_norm": 60.92182540893555,
      "learning_rate": 4.9206349206349204e-05,
      "loss": 0.4737,
      "step": 600
    },
    {
      "epoch": 0.35714285714285715,
      "grad_norm": 3.982450485229492,
      "learning_rate": 4.894179894179895e-05,
      "loss": 0.5708,
      "step": 625
    },
    {
      "epoch": 0.37142857142857144,
      "grad_norm": 3.5229907035827637,
      "learning_rate": 4.8677248677248676e-05,
      "loss": 0.5932,
      "step": 650
    },
    {
      "epoch": 0.38571428571428573,
      "grad_norm": 2.1403772830963135,
      "learning_rate": 4.841269841269841e-05,
      "loss": 0.5139,
      "step": 675
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.647314548492432,
      "learning_rate": 4.814814814814815e-05,
      "loss": 0.4768,
      "step": 700
    },
    {
      "epoch": 0.4142857142857143,
      "grad_norm": 3.251055955886841,
      "learning_rate": 4.7883597883597884e-05,
      "loss": 0.5418,
      "step": 725
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 1.5069653987884521,
      "learning_rate": 4.761904761904762e-05,
      "loss": 0.4964,
      "step": 750
    },
    {
      "epoch": 0.44285714285714284,
      "grad_norm": 2.6693944931030273,
      "learning_rate": 4.7354497354497356e-05,
      "loss": 0.5232,
      "step": 775
    },
    {
      "epoch": 0.45714285714285713,
      "grad_norm": 1.6969568729400635,
      "learning_rate": 4.708994708994709e-05,
      "loss": 0.4567,
      "step": 800
    },
    {
      "epoch": 0.4714285714285714,
      "grad_norm": 2.552241563796997,
      "learning_rate": 4.682539682539683e-05,
      "loss": 0.5207,
      "step": 825
    },
    {
      "epoch": 0.4857142857142857,
      "grad_norm": 10.88866901397705,
      "learning_rate": 4.656084656084656e-05,
      "loss": 0.4581,
      "step": 850
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.16725754737854,
      "learning_rate": 4.62962962962963e-05,
      "loss": 0.5731,
      "step": 875
    },
    {
      "epoch": 0.5142857142857142,
      "grad_norm": 1.6408342123031616,
      "learning_rate": 4.603174603174603e-05,
      "loss": 0.5297,
      "step": 900
    },
    {
      "epoch": 0.5285714285714286,
      "grad_norm": 2.1263160705566406,
      "learning_rate": 4.576719576719577e-05,
      "loss": 0.5486,
      "step": 925
    },
    {
      "epoch": 0.5428571428571428,
      "grad_norm": 5.503107070922852,
      "learning_rate": 4.55026455026455e-05,
      "loss": 0.597,
      "step": 950
    },
    {
      "epoch": 0.5571428571428572,
      "grad_norm": 4.534320831298828,
      "learning_rate": 4.523809523809524e-05,
      "loss": 0.5929,
      "step": 975
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 1.2646028995513916,
      "learning_rate": 4.4973544973544974e-05,
      "loss": 0.4732,
      "step": 1000
    },
    {
      "epoch": 0.5857142857142857,
      "grad_norm": 1.70009446144104,
      "learning_rate": 4.470899470899471e-05,
      "loss": 0.5169,
      "step": 1025
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.2281928062438965,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 0.5879,
      "step": 1050
    },
    {
      "epoch": 0.6142857142857143,
      "grad_norm": 1.7101916074752808,
      "learning_rate": 4.417989417989418e-05,
      "loss": 0.5697,
      "step": 1075
    },
    {
      "epoch": 0.6285714285714286,
      "grad_norm": 2.1661391258239746,
      "learning_rate": 4.391534391534391e-05,
      "loss": 0.5531,
      "step": 1100
    },
    {
      "epoch": 0.6428571428571429,
      "grad_norm": 2.6080431938171387,
      "learning_rate": 4.3650793650793655e-05,
      "loss": 0.5896,
      "step": 1125
    },
    {
      "epoch": 0.6571428571428571,
      "grad_norm": 3.2243294715881348,
      "learning_rate": 4.3386243386243384e-05,
      "loss": 0.5745,
      "step": 1150
    },
    {
      "epoch": 0.6714285714285714,
      "grad_norm": 2.348510503768921,
      "learning_rate": 4.312169312169313e-05,
      "loss": 0.4916,
      "step": 1175
    },
    {
      "epoch": 0.6857142857142857,
      "grad_norm": 1.1530951261520386,
      "learning_rate": 4.2857142857142856e-05,
      "loss": 0.5781,
      "step": 1200
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1737709045410156,
      "learning_rate": 4.259259259259259e-05,
      "loss": 0.5338,
      "step": 1225
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 3.5441572666168213,
      "learning_rate": 4.232804232804233e-05,
      "loss": 0.5066,
      "step": 1250
    },
    {
      "epoch": 0.7285714285714285,
      "grad_norm": 4.667012691497803,
      "learning_rate": 4.2063492063492065e-05,
      "loss": 0.541,
      "step": 1275
    },
    {
      "epoch": 0.7428571428571429,
      "grad_norm": 2.229901075363159,
      "learning_rate": 4.17989417989418e-05,
      "loss": 0.5891,
      "step": 1300
    },
    {
      "epoch": 0.7571428571428571,
      "grad_norm": 2.8626716136932373,
      "learning_rate": 4.153439153439154e-05,
      "loss": 0.5776,
      "step": 1325
    },
    {
      "epoch": 0.7714285714285715,
      "grad_norm": 1.1844497919082642,
      "learning_rate": 4.126984126984127e-05,
      "loss": 0.531,
      "step": 1350
    },
    {
      "epoch": 0.7857142857142857,
      "grad_norm": 3.5622355937957764,
      "learning_rate": 4.100529100529101e-05,
      "loss": 0.5501,
      "step": 1375
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.547538995742798,
      "learning_rate": 4.074074074074074e-05,
      "loss": 0.4882,
      "step": 1400
    },
    {
      "epoch": 0.8142857142857143,
      "grad_norm": 3.654449224472046,
      "learning_rate": 4.047619047619048e-05,
      "loss": 0.5808,
      "step": 1425
    },
    {
      "epoch": 0.8285714285714286,
      "grad_norm": 1.6750679016113281,
      "learning_rate": 4.021164021164021e-05,
      "loss": 0.6032,
      "step": 1450
    },
    {
      "epoch": 0.8428571428571429,
      "grad_norm": 5.005067348480225,
      "learning_rate": 3.9947089947089946e-05,
      "loss": 0.618,
      "step": 1475
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 1.0818400382995605,
      "learning_rate": 3.968253968253968e-05,
      "loss": 0.6164,
      "step": 1500
    },
    {
      "epoch": 0.8714285714285714,
      "grad_norm": 1.5100617408752441,
      "learning_rate": 3.941798941798942e-05,
      "loss": 0.5685,
      "step": 1525
    },
    {
      "epoch": 0.8857142857142857,
      "grad_norm": 1.2186906337738037,
      "learning_rate": 3.9153439153439155e-05,
      "loss": 0.5803,
      "step": 1550
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.3955905437469482,
      "learning_rate": 3.888888888888889e-05,
      "loss": 0.6181,
      "step": 1575
    },
    {
      "epoch": 0.9142857142857143,
      "grad_norm": 1.53127121925354,
      "learning_rate": 3.862433862433863e-05,
      "loss": 0.5226,
      "step": 1600
    },
    {
      "epoch": 0.9285714285714286,
      "grad_norm": 5.5355000495910645,
      "learning_rate": 3.835978835978836e-05,
      "loss": 0.6256,
      "step": 1625
    },
    {
      "epoch": 0.9428571428571428,
      "grad_norm": 4.500365257263184,
      "learning_rate": 3.809523809523809e-05,
      "loss": 0.5899,
      "step": 1650
    },
    {
      "epoch": 0.9571428571428572,
      "grad_norm": 1.2033424377441406,
      "learning_rate": 3.7830687830687835e-05,
      "loss": 0.5788,
      "step": 1675
    },
    {
      "epoch": 0.9714285714285714,
      "grad_norm": 4.775984287261963,
      "learning_rate": 3.7566137566137564e-05,
      "loss": 0.5698,
      "step": 1700
    },
    {
      "epoch": 0.9857142857142858,
      "grad_norm": 4.5160627365112305,
      "learning_rate": 3.730158730158731e-05,
      "loss": 0.5724,
      "step": 1725
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.100114107131958,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 0.5819,
      "step": 1750
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7575,
      "eval_auc": 0.41834098873804904,
      "eval_f1": 0.0,
      "eval_loss": 0.553956925868988,
      "eval_precision": 0.0,
      "eval_recall": 0.0,
      "eval_runtime": 2917.9342,
      "eval_samples_per_second": 1.371,
      "eval_steps_per_second": 0.086,
      "step": 1750
    },
    {
      "epoch": 1.0142857142857142,
      "grad_norm": 5.368441581726074,
      "learning_rate": 3.677248677248677e-05,
      "loss": 0.6113,
      "step": 1775
    },
    {
      "epoch": 1.0285714285714285,
      "grad_norm": 1.4688303470611572,
      "learning_rate": 3.650793650793651e-05,
      "loss": 0.5436,
      "step": 1800
    },
    {
      "epoch": 1.042857142857143,
      "grad_norm": 2.101248025894165,
      "learning_rate": 3.6243386243386245e-05,
      "loss": 0.5432,
      "step": 1825
    },
    {
      "epoch": 1.0571428571428572,
      "grad_norm": 1.0195976495742798,
      "learning_rate": 3.597883597883598e-05,
      "loss": 0.4865,
      "step": 1850
    },
    {
      "epoch": 1.0714285714285714,
      "grad_norm": 2.4836387634277344,
      "learning_rate": 3.571428571428572e-05,
      "loss": 0.6032,
      "step": 1875
    },
    {
      "epoch": 1.0857142857142856,
      "grad_norm": 2.7130017280578613,
      "learning_rate": 3.5449735449735446e-05,
      "loss": 0.5999,
      "step": 1900
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4913809299468994,
      "learning_rate": 3.518518518518519e-05,
      "loss": 0.639,
      "step": 1925
    },
    {
      "epoch": 1.1142857142857143,
      "grad_norm": 1.7654762268066406,
      "learning_rate": 3.492063492063492e-05,
      "loss": 0.5062,
      "step": 1950
    },
    {
      "epoch": 1.1285714285714286,
      "grad_norm": 2.056727409362793,
      "learning_rate": 3.465608465608466e-05,
      "loss": 0.5846,
      "step": 1975
    },
    {
      "epoch": 1.1428571428571428,
      "grad_norm": 2.196526288986206,
      "learning_rate": 3.439153439153439e-05,
      "loss": 0.5972,
      "step": 2000
    },
    {
      "epoch": 1.157142857142857,
      "grad_norm": 1.405187964439392,
      "learning_rate": 3.412698412698413e-05,
      "loss": 0.5244,
      "step": 2025
    },
    {
      "epoch": 1.1714285714285715,
      "grad_norm": 1.658987283706665,
      "learning_rate": 3.386243386243386e-05,
      "loss": 0.6218,
      "step": 2050
    },
    {
      "epoch": 1.1857142857142857,
      "grad_norm": 3.4630517959594727,
      "learning_rate": 3.35978835978836e-05,
      "loss": 0.5025,
      "step": 2075
    },
    {
      "epoch": 1.2,
      "grad_norm": 3.2489068508148193,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.5414,
      "step": 2100
    },
    {
      "epoch": 1.2142857142857142,
      "grad_norm": 2.51055645942688,
      "learning_rate": 3.306878306878307e-05,
      "loss": 0.5332,
      "step": 2125
    },
    {
      "epoch": 1.2285714285714286,
      "grad_norm": 6.732278347015381,
      "learning_rate": 3.280423280423281e-05,
      "loss": 0.5602,
      "step": 2150
    },
    {
      "epoch": 1.2428571428571429,
      "grad_norm": 1.0506640672683716,
      "learning_rate": 3.253968253968254e-05,
      "loss": 0.5447,
      "step": 2175
    },
    {
      "epoch": 1.2571428571428571,
      "grad_norm": 1.4512102603912354,
      "learning_rate": 3.227513227513227e-05,
      "loss": 0.5375,
      "step": 2200
    },
    {
      "epoch": 1.2714285714285714,
      "grad_norm": 6.222798824310303,
      "learning_rate": 3.2010582010582015e-05,
      "loss": 0.5756,
      "step": 2225
    },
    {
      "epoch": 1.2857142857142856,
      "grad_norm": 3.3553268909454346,
      "learning_rate": 3.1746031746031745e-05,
      "loss": 0.5199,
      "step": 2250
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.9587100744247437,
      "learning_rate": 3.148148148148148e-05,
      "loss": 0.6125,
      "step": 2275
    },
    {
      "epoch": 1.3142857142857143,
      "grad_norm": 4.260005950927734,
      "learning_rate": 3.121693121693122e-05,
      "loss": 0.5323,
      "step": 2300
    },
    {
      "epoch": 1.3285714285714285,
      "grad_norm": 2.494187593460083,
      "learning_rate": 3.095238095238095e-05,
      "loss": 0.5925,
      "step": 2325
    },
    {
      "epoch": 1.342857142857143,
      "grad_norm": 2.1405022144317627,
      "learning_rate": 3.068783068783069e-05,
      "loss": 0.5944,
      "step": 2350
    },
    {
      "epoch": 1.3571428571428572,
      "grad_norm": 3.7856087684631348,
      "learning_rate": 3.0423280423280425e-05,
      "loss": 0.5534,
      "step": 2375
    },
    {
      "epoch": 1.3714285714285714,
      "grad_norm": 2.555772542953491,
      "learning_rate": 3.0158730158730158e-05,
      "loss": 0.6063,
      "step": 2400
    },
    {
      "epoch": 1.3857142857142857,
      "grad_norm": 1.4090971946716309,
      "learning_rate": 2.9894179894179897e-05,
      "loss": 0.5586,
      "step": 2425
    },
    {
      "epoch": 1.4,
      "grad_norm": 2.6404404640197754,
      "learning_rate": 2.962962962962963e-05,
      "loss": 0.5804,
      "step": 2450
    },
    {
      "epoch": 1.4142857142857144,
      "grad_norm": 4.23564338684082,
      "learning_rate": 2.9365079365079366e-05,
      "loss": 0.549,
      "step": 2475
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 2.0943710803985596,
      "learning_rate": 2.91005291005291e-05,
      "loss": 0.4948,
      "step": 2500
    },
    {
      "epoch": 1.4428571428571428,
      "grad_norm": 2.4703826904296875,
      "learning_rate": 2.8835978835978838e-05,
      "loss": 0.6087,
      "step": 2525
    },
    {
      "epoch": 1.457142857142857,
      "grad_norm": 3.7438924312591553,
      "learning_rate": 2.857142857142857e-05,
      "loss": 0.5292,
      "step": 2550
    },
    {
      "epoch": 1.4714285714285715,
      "grad_norm": 1.258330225944519,
      "learning_rate": 2.830687830687831e-05,
      "loss": 0.6377,
      "step": 2575
    },
    {
      "epoch": 1.4857142857142858,
      "grad_norm": 6.1410040855407715,
      "learning_rate": 2.8042328042328043e-05,
      "loss": 0.5658,
      "step": 2600
    },
    {
      "epoch": 1.5,
      "grad_norm": 3.4298579692840576,
      "learning_rate": 2.777777777777778e-05,
      "loss": 0.5708,
      "step": 2625
    },
    {
      "epoch": 1.5142857142857142,
      "grad_norm": 2.0775437355041504,
      "learning_rate": 2.7513227513227512e-05,
      "loss": 0.5677,
      "step": 2650
    },
    {
      "epoch": 1.5285714285714285,
      "grad_norm": 1.43012535572052,
      "learning_rate": 2.724867724867725e-05,
      "loss": 0.6432,
      "step": 2675
    },
    {
      "epoch": 1.5428571428571427,
      "grad_norm": 1.8601475954055786,
      "learning_rate": 2.6984126984126984e-05,
      "loss": 0.5402,
      "step": 2700
    },
    {
      "epoch": 1.5571428571428572,
      "grad_norm": 1.1550532579421997,
      "learning_rate": 2.6719576719576723e-05,
      "loss": 0.5667,
      "step": 2725
    },
    {
      "epoch": 1.5714285714285714,
      "grad_norm": 1.9412027597427368,
      "learning_rate": 2.6455026455026456e-05,
      "loss": 0.67,
      "step": 2750
    },
    {
      "epoch": 1.5857142857142859,
      "grad_norm": 2.3738794326782227,
      "learning_rate": 2.6190476190476192e-05,
      "loss": 0.5317,
      "step": 2775
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9972871541976929,
      "learning_rate": 2.5925925925925925e-05,
      "loss": 0.5367,
      "step": 2800
    },
    {
      "epoch": 1.6142857142857143,
      "grad_norm": 6.379265785217285,
      "learning_rate": 2.5661375661375664e-05,
      "loss": 0.5625,
      "step": 2825
    },
    {
      "epoch": 1.6285714285714286,
      "grad_norm": 2.9254653453826904,
      "learning_rate": 2.5396825396825397e-05,
      "loss": 0.5421,
      "step": 2850
    },
    {
      "epoch": 1.6428571428571428,
      "grad_norm": 1.8391549587249756,
      "learning_rate": 2.5132275132275137e-05,
      "loss": 0.5418,
      "step": 2875
    },
    {
      "epoch": 1.657142857142857,
      "grad_norm": 1.8526755571365356,
      "learning_rate": 2.4867724867724866e-05,
      "loss": 0.5375,
      "step": 2900
    },
    {
      "epoch": 1.6714285714285713,
      "grad_norm": 2.465268135070801,
      "learning_rate": 2.4603174603174602e-05,
      "loss": 0.6023,
      "step": 2925
    },
    {
      "epoch": 1.6857142857142857,
      "grad_norm": 3.8061232566833496,
      "learning_rate": 2.4338624338624338e-05,
      "loss": 0.5901,
      "step": 2950
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.1175384521484375,
      "learning_rate": 2.4074074074074074e-05,
      "loss": 0.5869,
      "step": 2975
    },
    {
      "epoch": 1.7142857142857144,
      "grad_norm": 3.1548244953155518,
      "learning_rate": 2.380952380952381e-05,
      "loss": 0.5656,
      "step": 3000
    },
    {
      "epoch": 1.7285714285714286,
      "grad_norm": 1.0895445346832275,
      "learning_rate": 2.3544973544973546e-05,
      "loss": 0.5044,
      "step": 3025
    },
    {
      "epoch": 1.7428571428571429,
      "grad_norm": 3.5141279697418213,
      "learning_rate": 2.328042328042328e-05,
      "loss": 0.5736,
      "step": 3050
    },
    {
      "epoch": 1.7571428571428571,
      "grad_norm": 1.3836239576339722,
      "learning_rate": 2.3015873015873015e-05,
      "loss": 0.5222,
      "step": 3075
    },
    {
      "epoch": 1.7714285714285714,
      "grad_norm": 1.3129069805145264,
      "learning_rate": 2.275132275132275e-05,
      "loss": 0.4588,
      "step": 3100
    },
    {
      "epoch": 1.7857142857142856,
      "grad_norm": 1.5240994691848755,
      "learning_rate": 2.2486772486772487e-05,
      "loss": 0.5327,
      "step": 3125
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.968557834625244,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 0.5576,
      "step": 3150
    },
    {
      "epoch": 1.8142857142857143,
      "grad_norm": 1.841829776763916,
      "learning_rate": 2.1957671957671956e-05,
      "loss": 0.6032,
      "step": 3175
    },
    {
      "epoch": 1.8285714285714287,
      "grad_norm": 6.693917751312256,
      "learning_rate": 2.1693121693121692e-05,
      "loss": 0.4459,
      "step": 3200
    },
    {
      "epoch": 1.842857142857143,
      "grad_norm": 1.515498399734497,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 0.505,
      "step": 3225
    },
    {
      "epoch": 1.8571428571428572,
      "grad_norm": 5.72896146774292,
      "learning_rate": 2.1164021164021164e-05,
      "loss": 0.5016,
      "step": 3250
    },
    {
      "epoch": 1.8714285714285714,
      "grad_norm": 3.2085087299346924,
      "learning_rate": 2.08994708994709e-05,
      "loss": 0.5166,
      "step": 3275
    },
    {
      "epoch": 1.8857142857142857,
      "grad_norm": 1.7017937898635864,
      "learning_rate": 2.0634920634920636e-05,
      "loss": 0.5847,
      "step": 3300
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4550480842590332,
      "learning_rate": 2.037037037037037e-05,
      "loss": 0.5001,
      "step": 3325
    },
    {
      "epoch": 1.9142857142857141,
      "grad_norm": 1.6508489847183228,
      "learning_rate": 2.0105820105820105e-05,
      "loss": 0.4411,
      "step": 3350
    },
    {
      "epoch": 1.9285714285714286,
      "grad_norm": 2.124183416366577,
      "learning_rate": 1.984126984126984e-05,
      "loss": 0.5399,
      "step": 3375
    },
    {
      "epoch": 1.9428571428571428,
      "grad_norm": 1.4454411268234253,
      "learning_rate": 1.9576719576719577e-05,
      "loss": 0.4483,
      "step": 3400
    },
    {
      "epoch": 1.9571428571428573,
      "grad_norm": 2.6225087642669678,
      "learning_rate": 1.9312169312169313e-05,
      "loss": 0.5105,
      "step": 3425
    },
    {
      "epoch": 1.9714285714285715,
      "grad_norm": 2.9631993770599365,
      "learning_rate": 1.9047619047619046e-05,
      "loss": 0.4923,
      "step": 3450
    },
    {
      "epoch": 1.9857142857142858,
      "grad_norm": 1.1103156805038452,
      "learning_rate": 1.8783068783068782e-05,
      "loss": 0.4865,
      "step": 3475
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.239471435546875,
      "learning_rate": 1.8518518518518518e-05,
      "loss": 0.5177,
      "step": 3500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.80325,
      "eval_auc": 0.688394746691164,
      "eval_f1": 0.3336155800169348,
      "eval_loss": 0.49802738428115845,
      "eval_precision": 0.933649289099526,
      "eval_recall": 0.20309278350515464,
      "eval_runtime": 2925.0421,
      "eval_samples_per_second": 1.368,
      "eval_steps_per_second": 0.085,
      "step": 3500
    },
    {
      "epoch": 2.0142857142857142,
      "grad_norm": 1.2115052938461304,
      "learning_rate": 1.8253968253968254e-05,
      "loss": 0.4978,
      "step": 3525
    },
    {
      "epoch": 2.0285714285714285,
      "grad_norm": 1.3883342742919922,
      "learning_rate": 1.798941798941799e-05,
      "loss": 0.4737,
      "step": 3550
    },
    {
      "epoch": 2.0428571428571427,
      "grad_norm": 3.445514678955078,
      "learning_rate": 1.7724867724867723e-05,
      "loss": 0.4667,
      "step": 3575
    },
    {
      "epoch": 2.057142857142857,
      "grad_norm": 6.025082588195801,
      "learning_rate": 1.746031746031746e-05,
      "loss": 0.4632,
      "step": 3600
    },
    {
      "epoch": 2.0714285714285716,
      "grad_norm": 2.5322391986846924,
      "learning_rate": 1.7195767195767195e-05,
      "loss": 0.5541,
      "step": 3625
    },
    {
      "epoch": 2.085714285714286,
      "grad_norm": 5.116058349609375,
      "learning_rate": 1.693121693121693e-05,
      "loss": 0.4842,
      "step": 3650
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8601757287979126,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.4979,
      "step": 3675
    },
    {
      "epoch": 2.1142857142857143,
      "grad_norm": 2.9437739849090576,
      "learning_rate": 1.6402116402116404e-05,
      "loss": 0.3775,
      "step": 3700
    },
    {
      "epoch": 2.1285714285714286,
      "grad_norm": 2.2136480808258057,
      "learning_rate": 1.6137566137566136e-05,
      "loss": 0.4222,
      "step": 3725
    },
    {
      "epoch": 2.142857142857143,
      "grad_norm": 1.2892711162567139,
      "learning_rate": 1.5873015873015872e-05,
      "loss": 0.5466,
      "step": 3750
    },
    {
      "epoch": 2.157142857142857,
      "grad_norm": 4.652440547943115,
      "learning_rate": 1.560846560846561e-05,
      "loss": 0.5013,
      "step": 3775
    },
    {
      "epoch": 2.1714285714285713,
      "grad_norm": 1.655254602432251,
      "learning_rate": 1.5343915343915344e-05,
      "loss": 0.5065,
      "step": 3800
    },
    {
      "epoch": 2.185714285714286,
      "grad_norm": 3.821986675262451,
      "learning_rate": 1.5079365079365079e-05,
      "loss": 0.5241,
      "step": 3825
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8981862664222717,
      "learning_rate": 1.4814814814814815e-05,
      "loss": 0.4249,
      "step": 3850
    },
    {
      "epoch": 2.2142857142857144,
      "grad_norm": 1.351117730140686,
      "learning_rate": 1.455026455026455e-05,
      "loss": 0.4362,
      "step": 3875
    },
    {
      "epoch": 2.2285714285714286,
      "grad_norm": 0.9741631746292114,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 0.5778,
      "step": 3900
    },
    {
      "epoch": 2.242857142857143,
      "grad_norm": 1.9646023511886597,
      "learning_rate": 1.4021164021164022e-05,
      "loss": 0.4759,
      "step": 3925
    },
    {
      "epoch": 2.257142857142857,
      "grad_norm": 2.2216269969940186,
      "learning_rate": 1.3756613756613756e-05,
      "loss": 0.4727,
      "step": 3950
    },
    {
      "epoch": 2.2714285714285714,
      "grad_norm": 3.4054479598999023,
      "learning_rate": 1.3492063492063492e-05,
      "loss": 0.5394,
      "step": 3975
    },
    {
      "epoch": 2.2857142857142856,
      "grad_norm": 0.9018587470054626,
      "learning_rate": 1.3227513227513228e-05,
      "loss": 0.5155,
      "step": 4000
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.619492292404175,
      "learning_rate": 1.2962962962962962e-05,
      "loss": 0.4787,
      "step": 4025
    },
    {
      "epoch": 2.314285714285714,
      "grad_norm": 1.7518454790115356,
      "learning_rate": 1.2698412698412699e-05,
      "loss": 0.4495,
      "step": 4050
    },
    {
      "epoch": 2.3285714285714287,
      "grad_norm": 3.5467257499694824,
      "learning_rate": 1.2433862433862433e-05,
      "loss": 0.5298,
      "step": 4075
    },
    {
      "epoch": 2.342857142857143,
      "grad_norm": 2.8706352710723877,
      "learning_rate": 1.2169312169312169e-05,
      "loss": 0.5083,
      "step": 4100
    },
    {
      "epoch": 2.357142857142857,
      "grad_norm": 1.1621893644332886,
      "learning_rate": 1.1904761904761905e-05,
      "loss": 0.4694,
      "step": 4125
    },
    {
      "epoch": 2.3714285714285714,
      "grad_norm": 2.5403761863708496,
      "learning_rate": 1.164021164021164e-05,
      "loss": 0.5139,
      "step": 4150
    },
    {
      "epoch": 2.3857142857142857,
      "grad_norm": 1.2553601264953613,
      "learning_rate": 1.1375661375661376e-05,
      "loss": 0.451,
      "step": 4175
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.0528290271759033,
      "learning_rate": 1.1111111111111112e-05,
      "loss": 0.5064,
      "step": 4200
    },
    {
      "epoch": 2.414285714285714,
      "grad_norm": 3.6069552898406982,
      "learning_rate": 1.0846560846560846e-05,
      "loss": 0.5407,
      "step": 4225
    },
    {
      "epoch": 2.4285714285714284,
      "grad_norm": 2.984353542327881,
      "learning_rate": 1.0582010582010582e-05,
      "loss": 0.4342,
      "step": 4250
    },
    {
      "epoch": 2.442857142857143,
      "grad_norm": 2.755307674407959,
      "learning_rate": 1.0317460317460318e-05,
      "loss": 0.4685,
      "step": 4275
    },
    {
      "epoch": 2.4571428571428573,
      "grad_norm": 5.530125617980957,
      "learning_rate": 1.0052910052910053e-05,
      "loss": 0.5337,
      "step": 4300
    },
    {
      "epoch": 2.4714285714285715,
      "grad_norm": 1.0662144422531128,
      "learning_rate": 9.788359788359789e-06,
      "loss": 0.5805,
      "step": 4325
    },
    {
      "epoch": 2.4857142857142858,
      "grad_norm": 9.740211486816406,
      "learning_rate": 9.523809523809523e-06,
      "loss": 0.4842,
      "step": 4350
    },
    {
      "epoch": 2.5,
      "grad_norm": 6.550967693328857,
      "learning_rate": 9.259259259259259e-06,
      "loss": 0.5052,
      "step": 4375
    },
    {
      "epoch": 2.5142857142857142,
      "grad_norm": 3.393415927886963,
      "learning_rate": 8.994708994708995e-06,
      "loss": 0.5224,
      "step": 4400
    },
    {
      "epoch": 2.5285714285714285,
      "grad_norm": 1.6557607650756836,
      "learning_rate": 8.73015873015873e-06,
      "loss": 0.5622,
      "step": 4425
    },
    {
      "epoch": 2.5428571428571427,
      "grad_norm": 1.1902128458023071,
      "learning_rate": 8.465608465608466e-06,
      "loss": 0.5065,
      "step": 4450
    },
    {
      "epoch": 2.557142857142857,
      "grad_norm": 3.192903518676758,
      "learning_rate": 8.201058201058202e-06,
      "loss": 0.52,
      "step": 4475
    },
    {
      "epoch": 2.571428571428571,
      "grad_norm": 3.2567169666290283,
      "learning_rate": 7.936507936507936e-06,
      "loss": 0.4642,
      "step": 4500
    },
    {
      "epoch": 2.585714285714286,
      "grad_norm": 2.6848509311676025,
      "learning_rate": 7.671957671957672e-06,
      "loss": 0.491,
      "step": 4525
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.967970132827759,
      "learning_rate": 7.4074074074074075e-06,
      "loss": 0.5568,
      "step": 4550
    },
    {
      "epoch": 2.6142857142857143,
      "grad_norm": 1.5977520942687988,
      "learning_rate": 7.142857142857143e-06,
      "loss": 0.5135,
      "step": 4575
    },
    {
      "epoch": 2.6285714285714286,
      "grad_norm": 3.1753079891204834,
      "learning_rate": 6.878306878306878e-06,
      "loss": 0.4614,
      "step": 4600
    },
    {
      "epoch": 2.642857142857143,
      "grad_norm": 4.909998893737793,
      "learning_rate": 6.613756613756614e-06,
      "loss": 0.5481,
      "step": 4625
    },
    {
      "epoch": 2.657142857142857,
      "grad_norm": 1.6713173389434814,
      "learning_rate": 6.349206349206349e-06,
      "loss": 0.5052,
      "step": 4650
    },
    {
      "epoch": 2.6714285714285713,
      "grad_norm": 6.718080997467041,
      "learning_rate": 6.0846560846560845e-06,
      "loss": 0.5024,
      "step": 4675
    },
    {
      "epoch": 2.685714285714286,
      "grad_norm": 1.5787817239761353,
      "learning_rate": 5.82010582010582e-06,
      "loss": 0.6326,
      "step": 4700
    },
    {
      "epoch": 2.7,
      "grad_norm": 216.9809112548828,
      "learning_rate": 5.555555555555556e-06,
      "loss": 0.6655,
      "step": 4725
    },
    {
      "epoch": 2.7142857142857144,
      "grad_norm": 2.957324504852295,
      "learning_rate": 5.291005291005291e-06,
      "loss": 0.7434,
      "step": 4750
    },
    {
      "epoch": 2.7285714285714286,
      "grad_norm": 4.260836601257324,
      "learning_rate": 5.026455026455026e-06,
      "loss": 0.49,
      "step": 4775
    },
    {
      "epoch": 2.742857142857143,
      "grad_norm": 1.6990563869476318,
      "learning_rate": 4.7619047619047615e-06,
      "loss": 0.5026,
      "step": 4800
    },
    {
      "epoch": 2.757142857142857,
      "grad_norm": 3.801848888397217,
      "learning_rate": 4.497354497354498e-06,
      "loss": 0.5138,
      "step": 4825
    },
    {
      "epoch": 2.7714285714285714,
      "grad_norm": 1.9691251516342163,
      "learning_rate": 4.232804232804233e-06,
      "loss": 0.5051,
      "step": 4850
    },
    {
      "epoch": 2.7857142857142856,
      "grad_norm": 2.8060500621795654,
      "learning_rate": 3.968253968253968e-06,
      "loss": 0.5385,
      "step": 4875
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.293766498565674,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 0.4761,
      "step": 4900
    },
    {
      "epoch": 2.814285714285714,
      "grad_norm": 5.048039436340332,
      "learning_rate": 3.439153439153439e-06,
      "loss": 0.495,
      "step": 4925
    },
    {
      "epoch": 2.8285714285714287,
      "grad_norm": 5.631167888641357,
      "learning_rate": 3.1746031746031746e-06,
      "loss": 0.4568,
      "step": 4950
    },
    {
      "epoch": 2.842857142857143,
      "grad_norm": 2.9913084506988525,
      "learning_rate": 2.91005291005291e-06,
      "loss": 0.554,
      "step": 4975
    },
    {
      "epoch": 2.857142857142857,
      "grad_norm": 3.9224941730499268,
      "learning_rate": 2.6455026455026455e-06,
      "loss": 0.4962,
      "step": 5000
    },
    {
      "epoch": 2.8714285714285714,
      "grad_norm": 1.9809192419052124,
      "learning_rate": 2.3809523809523808e-06,
      "loss": 0.3825,
      "step": 5025
    },
    {
      "epoch": 2.8857142857142857,
      "grad_norm": 3.313559055328369,
      "learning_rate": 2.1164021164021164e-06,
      "loss": 0.4645,
      "step": 5050
    },
    {
      "epoch": 2.9,
      "grad_norm": 5.611025333404541,
      "learning_rate": 1.8518518518518519e-06,
      "loss": 0.3717,
      "step": 5075
    },
    {
      "epoch": 2.914285714285714,
      "grad_norm": 6.190692901611328,
      "learning_rate": 1.5873015873015873e-06,
      "loss": 0.4256,
      "step": 5100
    },
    {
      "epoch": 2.928571428571429,
      "grad_norm": 4.822566986083984,
      "learning_rate": 1.3227513227513228e-06,
      "loss": 0.4603,
      "step": 5125
    },
    {
      "epoch": 2.942857142857143,
      "grad_norm": 6.1476616859436035,
      "learning_rate": 1.0582010582010582e-06,
      "loss": 0.5584,
      "step": 5150
    },
    {
      "epoch": 2.9571428571428573,
      "grad_norm": 2.9397058486938477,
      "learning_rate": 7.936507936507937e-07,
      "loss": 0.4819,
      "step": 5175
    },
    {
      "epoch": 2.9714285714285715,
      "grad_norm": 3.3358285427093506,
      "learning_rate": 5.291005291005291e-07,
      "loss": 0.5368,
      "step": 5200
    },
    {
      "epoch": 2.9857142857142858,
      "grad_norm": 1.9532283544540405,
      "learning_rate": 2.6455026455026455e-07,
      "loss": 0.4653,
      "step": 5225
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7440646886825562,
      "learning_rate": 0.0,
      "loss": 0.492,
      "step": 5250
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.812,
      "eval_auc": 0.7815613623217992,
      "eval_f1": 0.37333333333333335,
      "eval_loss": 0.4526550769805908,
      "eval_precision": 0.9739130434782609,
      "eval_recall": 0.2309278350515464,
      "eval_runtime": 2920.2729,
      "eval_samples_per_second": 1.37,
      "eval_steps_per_second": 0.086,
      "step": 5250
    }
  ],
  "logging_steps": 25,
  "max_steps": 5250,
  "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": 1.105066432512e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}